pmid stringlengths 8 8 | pmcid stringlengths 8 11 ⌀ | source stringclasses 2
values | rank int64 1 9.78k | sections unknown | tokens int64 3 46.7k |
|---|---|---|---|---|---|
31328163 | PMC6636988 | pmc | 628 | {
"abstract": "Electron-conducting cable bacteria are found around worm tubes in bioturbated sediments in which they were thought to be rare.",
"conclusion": "CONCLUSIONS These data demonstrate that cable bacteria can be abundant in intensely bioturbated deposits, particularly in aggregations directly associated with physically stable and irrigated biogenic structures, in this case, predominantly Chaetopterus tubes. The presence of short filament sections throughout the bioturbated zone in the sulfidic estuarine muds of GPB, together with the demonstration of metabolic flexibility ( 9 , 10 ), implies that these organisms are opportunistic and facultative and are capable of responding to the presence of stable (about >1 week), oxygenated microenvironments generated by macrofauna and to a range of biogeochemical conditions. These distribution patterns further imply that the present dominant description of cable bacteria likely represents an end-member life habit and morphological expression in a continuum: from classic metabolic adaptations associated with local redox conditions and traditional successional zonation to conducting filaments promoting nonlocal electron transport in physically stable, 2D and 3D electrogenic grids. We conclude that the evolution of bioturbation and biogenic structure in the late Proterozoic and early Phanerozoic (~550 million years ago) ( 29 ) did not substantially inhibit cable bacteria but rather further enhanced their ecological options.",
"introduction": "INTRODUCTION Since their discovery in laboratory-incubated sediment ( 1 ), cable bacteria, all of which belong to the deltaproteobacterial family Desulfobulbaceae, have been the subject of intense study because of their unique adaptations and nontraditional utilization of redox conditions in sedimentary deposits ( 2 ). They are now recognized as being widely distributed in natural systems but under apparently restricted environmental conditions ( 3 – 6 ). The primary conceptual and quantitative model characterization of these organisms is as multicellular filaments extending uninterrupted over centimeter scales from the oxygenated sediment–water interface vertically into underlying sulfidic regions ( 2 , 7 ). In this mode, filaments act as organically insulated electron conductors directly connecting cathodic reduction in the oxic zone with anodic oxidation in the sulfidic zone, bypassing possible intermediate suboxic zones. Conductive sediment pore water completes the circuit. As a distinctive signature of cable bacteria activity, reduction of O 2 elevates pH and the oxidation of sulfide (HS − , Fe-sulfides) lowers pH in spatially separated regions, promoting alkalinity production and carbonate and oxide mineral precipitation in the oxic zone, and acid production and carbonate mineral dissolution in a distal sulfidic zone. Additional biogeochemical consequences attributed to cable bacteria are enhancement of sedimentary capture of P and possible modulation of eutrophication dynamics in coastal environments ( 5 , 8 ). The uniqueness of cable bacteria has stimulated continued research into their physiology and impacts, with recent observations contributing to a rapidly changing view of this microbial group and the conceptual paradigms surrounding them. It is now known that cable bacteria can use a wide range of cathodic potentials, extending to ~170 mV relative to the standard hydrogen electrode, implying, but not proving, that cathodic reactions coupled to NO 3 − , Mn 4+ , and Fe 3+ can all be used in addition to O 2 ( 9 , 10 ). A range of reductants can also be used, including dissolved organic matter (e.g., propionate) ( 11 ). Rather than a standalone species, cable bacteria appear to be one component of complex e-community consortia and are apparently heterotrophs closely associated with sulfur-oxidizing chemoautotrophs ( 11 , 12 ). Their taxonomy may be more diverse than originally recognized, although this is a subject of ongoing research (Supplementary Materials) ( 10 , 13 ). All of these modifications from the original model imply that cable bacteria consist of multiple species and strains that are metabolically and ecologically flexible. While shown to be ubiquitously distributed, they are believed to be largely inhibited in bioturbated deposits ( 3 , 4 , 6 ). Here, we demonstrate that well-developed filaments can be abundant in stable subdomains of the bioturbated zone of sedimentary deposits, and conclude that the assumed model of cable bacteria likely represents an end member in a metabolic and morphologic continuum for a highly facultative group of microorganisms.",
"discussion": "DISCUSSION Br − tracer tracks exchange of pore water with overlying water through the walls of ventilated dwellings and feeding cavities into surrounding sediment. Cable bacteria filaments and free bacterial cells were abundant in subsurface aggregations associated with regions of rapid solute exchange (Br − ) and specifically Chaetopterus tubes, which can be stable for months or more ( Figs. 1 to 3 ). Relatively longer cables were present in the immediate vicinity of tubes, ranging to >650 μm in these examples ( Figs. 1 to 3 ). The appearance times of filament aggregations near the sediment-water interface in laboratory microcosm incubations are ~7 to 10 days ( 11 , 21 , 25 ), suggesting that if transport conditions around subsurface biogenic structures are established for similar time scales, cable bacteria can and do respond. The occurrence of cable bacteria in close association with oxygenated regions of seagrass roots is consistent with this inference ( 26 ). We have never directly observed filaments of centimeter length in natural sediments, perhaps as a result of fragmentation during separation for identification and counting, despite geochemical indicators (pH distributions) that would suggest these scales ( Fig. 4B ). We note that with some exceptions ( 22 , 27 ), most investigators report cable abundances that must also reflect typical lengths less than 1 to 5 mm, or major biases in sampling procedures that prevent accurate length measurements ( 4 , 11 , 22 ). If lengths of 1 to 5 mm or fewer cells are indeed common, electron conduction through networks of filaments and minerals rather than continuous cables may occur, or cells may optionally use local redox reaction pathways in a traditional mode. In any case, we presume that the relative distributions are accurate and imply more optimal growth of cables in association with Chaetopterus tubes. Geochemical evidence of cable bacterial activity associated with Chaetopterus tubes consists of elevated pH at the oxygenated wall and a distal pH minimum, solid-phase S gradients with depletion around tubes, and focused authigenic mineral precipitates within the tube lining (cathodic terminus), in these cases composed predominantly of Mn and Fe oxides ( Fig. 5B ). The scaling of the elevated pH zone relative to the pH minimum presumably reflects the radial symmetry and volumetrically comparable anodic and cathodic zones (e.g., annulus areas defined by radius distance differences, expand as a function of radius) ( Fig. 4B ). Mn, in particular, is strongly enriched at the oxygenated tube boundary relative to surrounding sediment (>130×). Oxic and anoxic incubations of tube material suggest that the Mn-oxide micronodules inhibit aerobic decomposition of the carbohydrate-rich tube linings ( 28 ). The elevation of dissolved Fe 2+ around tubes is also consistent with cable bacteria activity and Fe-oxide enrichment in the tube lining ( Fig. 4D ). Dissolved H 2 S in these deposits is generally below 20 μM and is further depleted near all irrigated structures, including Chaetopterus ( 23 ). Although, with the exception of elevated pH at the inner tube wall, the geochemical patterns need not be uniquely due to cable bacteria, they are consistent with reactions associated with the activity of cable bacteria, which, along with associated consortia of S oxidizers and reducers ( 11 , 12 ), presumably promotes them. The overall relationships imply a mutualistic association between macro- and microfauna, with the former benefiting from enhanced tube longevity, lowered sulfide fluxes, and scavenging of potential toxins by Mn, Fe-oxide precipitates, and the latter by stable physical conditions and a reliably oxidized interface. As in the case of other stable tube and burrows, the abundance of nematodes is generally enhanced near Chaetopterus tubes, and they may be important in grazing the elevated populations of both cable and free bacteria (fig. S5) ( 24 )."
} | 2,146 |
36582302 | PMC9793950 | pmc | 629 | {
"abstract": "The atypical Markov decision processes (MDPs) are decision-making for maximizing the immediate returns in only one state transition. Many complex dynamic problems can be regarded as the atypical MDPs, e.g., football trajectory control, approximations of the compound Poincaré maps, and parameter identification. However, existing deep reinforcement learning (RL) algorithms are designed to maximize long-term returns, causing a waste of computing resources when applied in the atypical MDPs. These existing algorithms are also limited by the estimation error of the value function, leading to a poor policy. To solve such limitations, this paper proposes an immediate-return algorithm for the atypical MDPs with continuous action space by designing an unbiased and low variance target Q-value and a simplified network framework. Then, two examples of atypical MDPs considering the uncertainty are presented to illustrate the performance of the proposed algorithm, i.e., passing the football to a moving player and chipping the football over the human wall. Compared with the existing deep RL algorithms, such as deep deterministic policy gradient and proximal policy optimization, the proposed algorithm shows significant advantages in learning efficiency, the effective rate of control, and computing resource usage.",
"conclusion": "Conclusion The atypical MDPs exist widely in the engineering field, which involves one state transition with continuous action space. The control goal of the atypical MDPs is to maximize the immediate returns. However, the existing RL algorithms are designed for standard MDPs to maximize long-term returns. Thus, they can cause significant estimation errors for the value function and a waste of computing resources when dealing with the atypical MDPs. To solve such problems, this paper analyzes the characteristics of the atypical MDPs systematically and explains the differences between estimating the state-value function and estimating the action-value function. On this basis, the immediate-return RL algorithm was proposed to deal with the atypical MDPs. In the proposed algorithm, the method of average reward is developed to provide the unbiased and low variance target Q-value. Thus, the problems of large estimation errors can be overcome. And a newly designed network framework is designed for the proposed algorithm, which can significantly reduce computing resource usage. Then, two scenarios of the football trajectory control, i.e., passing the football to a moving player, and chipping the football over the human wall, are designed as the benchmark to test the algorithms designed for the atypical MDPs. Numerical results demonstrate that the learning efficiency of the proposed algorithm is 1.5 times that of the DDPG and 5 times that of the PPO. For the controllers based on the proposed algorithm, their effective rates of control are more than 97.7%, and their reliabilities are approaching 100%. Such performance is far superior to DDPG and PPO. As the proposed controller increases the shot's accuracy significantly, it can promote the development of high-level football robots in the Robot world cup. Furthermore, the proposed algorithm can also consume fewer computing resources than existing RL algorithms. Thus, the immediate-return RL algorithm has higher learning efficiency, higher performance, and lower computing resource usage than the existing RL algorithms, such as PPO and DDPG. It should be pointed out that the immediate-return RL algorithm can output only one determined action. This determined value can be seen as the best solution according to the specified rewards function. However, a single best solution based on the specified rewards function is impractical for many complex engineering problems (e.g., strongly non-linear dynamic system with parameter uncertainties). As one focus of the future work, efforts will be made to improve the algorithm to find a proper basin which corresponds to the specified scenario. After that, the action output shall be more practical. In the future, we will devote ourselves to expand the use of the proposed immediate-return RL algorithm and achieve more engineering applications, such as stamping process, directional blasting, approximations of the compound Poincaré maps, etc.",
"introduction": "Introduction Inspired by the learning pattern of humans, i.e., learning by interacting with the external environment, the concepts of reinforcement learning (RL) were first proposed by Minsky ( 1954 ). Subsequently, Bellman ( 1957 ) presented a method to define an RL problem using Markov decision processes (MDPs). As a result, an RL problem can be described clearly in terms of states, actions, and rewards. In recent years, with an in-depth combination of deep learning, traditional RL has evolved into deep RL. Generally speaking, deep RL algorithms can be subdivided into value-based algorithms and policy gradient algorithms. Deep Q Network (DQN) was the first exploration for value-based algorithms (Mnih et al., 2015 ). It solved the dimension explosion problem. Subsequently, various improved DQN algorithms were developed, such as Double DQN (Van Hasselt et al., 2016 ), Dueling DQN (Wang et al., 2016 ), etc. However, value-based algorithms could only be applied in discrete rather than continuous action space. In contrast, policy gradient algorithms could solve the RL problem with continuous action space, as an independent actor was constructed to output actions. Note that policy gradient algorithms were generally divided into stochastic policy algorithms and deterministic policy algorithms. The stochastic policy algorithms could output the probability distribution of the actions, such as the asynchronous advantage actor-critic (A3C) (Mnih et al., 2016 ) and proximal policy optimization (PPO) (Schulman et al., 2017 ). The deterministic policy algorithms could output the deterministic actions, such as deep deterministic policy gradient (DDPG) (Lillicrap et al., 2015 ). Due to the advantages of model-free, great self-learning ability, etc., the RL has shown excellent performance in the application of complex control processes. For example, the RL methods were applied to robot manipulators to solve trajectory planning under complex environments (Chen et al., 2022 ). Tutsoy and Brown studied the RL in problems with Chaotic dynamics and proved that a reasonable discount factor could avoid singular learning problems (Tutsoy and Brown, 2016 ). Pan et al. ( 2023 ) designed a controller for a three-link biped robot using the twin delayed deep deterministic policy gradient algorithm (TD3). Sharbafi et al. designed controllers based on the RL for their football robots and won third place in the 2011 world games (Sharbafi et al., 2011 ). Massi et al. ( 2022 ) increase the learning speed of a navigating robot to improve its performance using the RL method. Even in the financial sector, the RL could be used to learn investment trading policy (Lee et al., 2021 ). Such trading systems based on RL improved trading performance effectively. Indeed, the above application scenarios belong to the standard MDPs, containing a series of state transitions. However, the atypical MDP case, which involves only one state transition in continuous action space, can also arise in the engineering field, such as the stamping process (Wang and Budiansky, 1978 ), directional blasting (Zhu et al., 2008 ), football trajectory control (Myers and Mitchell, 2013 ), approximations of the compound Poincaré maps (Li et al., 2020 ), etc. In such atypical MDPs, the control goal is to maximize the immediate returns rather than the long-term returns. Therefore, compared to the standard MDPs, the atypical MDPs can exhibit many new characteristics. Furthermore, to the best knowledge of the authors, all existing RL algorithms are designed for the standard MDPs to maximize long-term returns. Applying the existing RL algorithms to the atypical MDPs shall lead to the following problems. On the one hand, the existing RL algorithms are also limited by their open problem, i.e., the estimation error of the value function. For example, the sampling errors caused by incomplete samplings will lead to bias for the estimated state-value function (e.g., A3C and PPO) (Mnih et al., 2016 ; Schulman et al., 2017 ). For the estimated action-value function, DQN and DDPG can cause the overestimation due to the max operation in off-policy temporal-difference (TD) learning (Mnih et al., 2015 ; Van Hasselt et al., 2016 ). In comparison, the TD3 and double DQN may lead to underestimation as the minimum output of two independent target critic networks is selected to update the action-value function (Lillicrap et al., 2015 ; Fujimoto et al., 2018 ). Furthermore, the uncertain environment may bring a high variance for the estimated value functions as the uncertainties can lead to entirely different rewards for the same state-action pair. Since the policy gradient formulation is directly related to the value function, the estimation error of the value function can lead to a poor policy and limit the performance of the existing RL algorithms. On the other hand, as the atypical MDPs focus only on immediate returns, the common designs for calculating long-term returns are redundant in the existing RL algorithms. It may result in a waste of computing resources. Moreover, existing algorithms do not notice the difference between estimating the state-value function and the action-value function in atypical MDPs. Such difference determines which approach is more suitable for dealing with atypical MDPs. Thus, regarding the above problems of the existing RL algorithms, this paper aims to propose an immediate-return RL algorithm for atypical MDPs with continuous action space. On this basis, this paper further takes the football trajectory control as the illustration example to present the superior performance of the proposed algorithm. Indeed, the football trajectory control shall be an ideal test case for the proposed algorithm. The reasons are as follows. As the whole process contains only one state transition from take-off to end and its action, i.e., the football's initial velocity, is continuous, football flight is an atypical MDP case with continuous action space. Meanwhile, the aerodynamic model of football is strongly non-linear and has no analytical solutions (Myers and Mitchell, 2013 ; Javorova and Ivanov, 2018 ), which involves many complex physical laws (Horowitz and Williamson, 2010 ; Norman and McKeon, 2011 ; Javorova and Ivanov, 2018 ; Kiratidis and Leinweber, 2018 ). It is difficult for the traditional control method to control football flight (Hou and Wang, 2013 ; Hou et al., 2016 ). Thus, as a challenging task, football trajectory control is an ideal example to test the proposed algorithm. In addition, related researches also have practical application value. The accuracy of the shot is a key of the football robot. Designing a high-performance controller based on the proposed algorithm can promote the development of high-level football robots in the Robot world cup (Sharbafi et al., 2011 ). The main contents and contributions of this paper are summarized as the following aspects. Firstly, the characteristics of the atypical MDPs are analyzed systematically based on the RL theory. The disadvantage of estimating the state-value function in the atypical MDPs is explained qualitatively, i.e., the large samples requirement and the unavoidable sampling error. These studies indicate the way to the development of RL algorithms in the atypical MDPs. That is, the deterministic policy has natural advantages in dealing with the atypical MDPs in continuous action space. Secondly, based on the deterministic policy and estimated action-value function, an immediate-return RL algorithm is proposed for the atypical MDPs. In the proposed algorithm, the average reward method is developed to construct an unbiased and low variance target Q-value. Compared with existing RL algorithms, e.g., DDPG and PPO, the proposed algorithm reduces the estimation error significantly. More details are introduced in following Section Immediate-return RL algorithm for the atypical MDPs. Meanwhile, a simplified network framework is also designed for the proposed algorithm. Thus, the proposed decreases both the space complexity and time complexity. The comparison tests also demonstrate that the computing resource consumed by the proposed algorithm is lower than the DDPG and PPO. Thirdly, two challenging scenarios of the football trajectory control, i.e., passing the football to a moving player, and chipping the football over the human wall (chip kick), are presented to test the feasibility of the proposed algorithm. These scenarios can be used as the benchmark to test the algorithms designed for the atypical MDPs. Meanwhile, the controllers based on the proposed algorithm in this paper can improve the football robot's shot accuracy in competitions, such as the Robot world cup (Sharbafi et al., 2011 ). In the above scenarios, existing RL algorithms (i.e., DDPG, PPO) are also tested as references. Numerical results demonstrate that the immediate-return RL algorithm has higher learning efficiency, a higher effective rate of control, and lower computing resource usage than the reference RL algorithms. The rest of the present work is organized as follows. In Section The atypical MDPs, the analysis of the atypical MDPs is introduced. Then, the immediate-return RL algorithm for the atypical MDPs is proposed in Section Immediate-return RL algorithm for the atypical MDPs. In Section Illustration examples: Football trajectory control for different scenarios, two illustration examples in MDPs, i.e., passing the football to a moving player and chipping the football over the human wall, are designed. In Section Comparison and discussion, the feasibility and high performance of the RL controllers are demonstrated by simulation tests. And the advantages of the immediate-return RL algorithm are discussed by comparison with the existing RL algorithms. Lastly, the conclusion of this paper is drawn in Section Conclusion.",
"discussion": "Comparison and discussion In this section, the advantages of the immediate-return RL algorithm for atypical MDPs will be discussed and demonstrated. Meanwhile, PPO (the representative of the stochastic policy algorithms) and DDPG (the representative of the deterministic policy algorithms) are chosen as the references for the proposed algorithm. All these algorithms will train corresponding controllers for two football flight scenarios. Then, the advantages of the proposed algorithm will be discussed and analyzed from the training process, training results (i.e., the performance of the controllers), and computing resource usage by comparing with these reference algorithms. Training process For the control problems of the football trajectory, the proposed algorithm's detailed network framework is designed in Figure 7 , including an independent actor network and an independent critic network. Here, the proposed algorithm's actor network and critic network have the same hidden layers and node numbers, i.e., the same network architectures. Indeed, each independent network in the three discussed algorithms shares the same network architectures to avoid the influence of the network architectures on the test results. Similarly, all discussed algorithms use the same reward function designed in Section Illustration examples: Football trajectory control for different scenarios. Furthermore, it should be noted that different deep RL algorithms have different sensitivities to hyperparameters (Henderson et al., 2018 ). Based on the trial and error and the experience of Dewey ( 2014 ), Henderson et al. ( 2018 ); and Silver et al. ( 2021 ), the detailed hyperparameters of each algorithm are selected (see Table 4 ). Under the premise of ensuring the algorithm's performance, each algorithm's hyperparameters are set to the same value. Figure 7 Detailed network architectures of the proposed algorithm. Table 4 The hyperparameters of the discussed deep RL algorithms. \n Learning rate for actor \n \n Learning rate for critic \n \n Discount factor \n \n Soft target updates \n The proposed algorithm 1e-4 1e-4 \\ \\ DDPG 1e-4 1e-4 0.9 0.01 PPO 5e-6 1e-5 0.9 \\ Then, all algorithms, i.e., the proposed algorithm, DDPG, and PPO, will train the corresponding controllers for these two scenarios. Here, the learning efficiency of the algorithm can be evaluated by the consumption of the training steps. After 450,000 training steps, all reward curves in these two scenarios are shown in Figure 8 . In both scenarios, the reward curves of the proposed algorithm (red line in Figure 8 ) converge to the high-level reward value after 300,000 training steps. Then, the suitable controllers can be obtained. Although the reward value of the DDPG algorithm also has risen during training (green line in Figure 8 ), DDPG's learning efficiency is worse than the proposed algorithm from the perspective of convergence speed. As shown in Figure 8 , DDPG needs about 450,000 training steps to converge the reward curves. That is, the learning efficiency of the proposed algorithm is 1.5 times that of the DDPG. And the convergency reward value of the DDPG is also less than the proposed algorithm. As a stochastic policy algorithm, PPO shows poor learning ability in football trajectory control. As show in Figure 8 , 450,000 training steps do not allow the PPO to converge. Actually, PPO can also be converged after consuming about 1,500,000 training steps. That is, the learning efficiency of the proposed algorithm is 5 times that of the PPO. Furthermore, the final convergency reward values of the PPO are far less than the proposed algorithm. Note that the more training steps, the more samples are required. Thus, the training process confirms the analysis in Section Atypical MDPs: Definition and characteristic analyses. That is, PPO's learning efficiency is low in the atypical MDPs, as estimating a state-value requires more samples. The above training process demonstrates that the proposed algorithm converges faster and consumes fewer samples compared to DDPG and PPO. That is, the proposed algorithm shows better learning efficiency. Actually, it is a significant advantage for the proposed algorithm, as the samples are difficult to obtain in many atypical MDP cases. Figure 8 The reward curves of different algorithms. (A) The first scenario. (B) The second scenario. Controller's performance In this section, the performance of the controllers will be analyzed from three aspects, i.e., accuracy, unit time cost, and reliability. As described in Section Illustration examples: Football trajectory control for different scenarios, two control targets, i.e., shooting the football to the destination and reducing the time of football flight, are considered for each scenario. Thus, accuracy and unit time cost are the core index for evaluating the control performance of the controllers. Actually, the control performance is closely related to the value function's estimation bias. Besides, the considered aerodynamic model of football is an uncertain environment. That is, the football trajectory may be completely different under the same state-action pair, bringing a high variance for the value function. To evaluate the effect of variance caused by the uncertain environment on the controller, the reliability is set as another index for the controller's performance. Here, Monte Carlo tests are applied to analyze the control performance of the controllers. In each scenario, 1,000 independent state will be chosen randomly and a set of initial velocities will then be generated by the tested controller for each chosen state. Then, only one flight trajectory will be generated for the chosen state and the outputted initial velocities. Here, the effective rate of control Re is defined as follows to evaluate the accuracy of the RL controller.\n (50) R e = N R e / 1000 \nwhere N Re is the number of the flight trajectories successfully controlled in 1000 tests. For the first passing scenario, if the relative error δ is less than 5%, the flight control will be regarded as success. Here, the relative error δ is defined as follows.\n (51) δ = ( x d - x f ) 2 + ( y d - y f ) 2 ( x d - x 0 ) 2 + ( y d - y 0 ) 2 \nAs shown in Figure 9A , the effective rate of control Re of the proposed controller in the first scenario, i.e., passing the football to a moving player, is 98.2%. In particular, there are 36.0% tests with relative error less than 1%, 56.6% tests with relative error from 1 to 3%, and 5.6% tests with relative error between 3 and 5%. Under the same tests, the DDPG controller's Re is 79.3%, and the PPO controller's Re is 80.5%. For the second scenario, scoring goals are regarded as the successful controls. The effective rate of control Re of the proposed controller for chipping the football over the human wall is 97.7% (see Figure 9B ). Meanwhile, the DDPG controller's Re and PPO controller's Re are 91.1 and 24.1%, respectively. Compared to DDPG and PPO, the good accuracy of the proposed controller is verified in both two scenarios. Figure 9 The accuracy tests. (A) The accuracy test's results in the first scenario. (B) The accuracy test's results in the second scenario. Based on 1,000 Monte Carlo tests, the average unit time cost t a of 1,000 tests is used to evaluate the unit time cost, which can be written as.\n (52) t a = ∑ 1 1000 t s / 1000 \nhere, t s is the unit time cost index, which can be found in Equation (39). For the sake of comparison and evaluation, the proposed controllers without the time cost optimization are also trained for two scenarios. In the first scenario, the proposed controller reduces the average unit time cost t a from 0.2080s to 0.0483s, comparing to the proposed controller without the time cost optimization (see Figure 10 ). Meanwhile, the DDPG controller can reduces the unit time cost t a to 0.0484s. And the PPO controller can reduce the unit time cost t a to 0.074. In the second scenario, adding the time optimization has little effect on flight time. However, the unit time cost of the proposed controller is the lowest compared to the DDPG and PPO controllers. Figure 10 The average unit time cost in flights. (A) The first scenario. (B) The second scenario. As analyzed in Section Limitations of existing RL algorithms in the atypical MDPs, the estimated value functions in existing RL algorithm, e.g., DDPG and PPO, is biased due to the TD learning method. Meanwhile, the sampling error err ( s t ) can further increase the estimation bias of the state-value function for the stochastic policy algorithms, as analyzed in Section Atypical MDPs: Definition and characteristic analyses. These estimation biases have adverse effects on the policy update. However, due to the average reward method (see Section The immediate-return RL algorithm), an unbiased target Q -value is provided for the proposed algorithm. Thus, the disadvantages of the estimation bias can be overcome. According to the above test data, the effective rate of control Re of the proposed controller in the first scenario is increased by 18.9% than the DDPG controller and increased by 17.7% than the PPO controller. In the second scenario, the effective rate of control Re of the proposed controller is increased by 6.6% than the DDPG controller and increased by 73.6% than the PPO controller. The proposed algorithm also shows better time cost optimization than DDPG and PPO in both two scenarios. Thus, the high accuracy and low unit time cost of the proposed controllers can be verified. This also means that the immediate-return RL algorithm has better performance than existing RL algorithms in deal with the atypical MDPs. In the reliability tests, several specified states will be chosen for the tested controllers in each scenario (see Figure 11 ). For each chosen state, the only set of definite initial velocities will be outputted by the corresponding controller. Then, in the uncertain environment, 200 different flight trajectories will be generated based on the same chosen states and the same initial velocities. To evaluate the reliability of the controllers, the reliable rate Rr is defined as the effective rate of control of the repeated 200 tests on the same chosen state, which is written as Equation (53)\n (53) R r = N R r / 200 \nwhere N Rr is the number of the flight trajectories controlled successfully in 200 reliability tests. Figure 11 Reliability tests. (A) The first scenario. Blue circle is the allowed landing range. (B) The second scenario. Blue plane is the human wall. Black wireframe is the goal. In the first scenario, a point is selected as the initial position of the moving player. The moving player is assumed to move along the four directions marked by the orange arrows in Figure 11A now. That is, four states are chosen for the tested controllers. According to Figure 12 , the average reliable rate of the proposed controller for the first scenario is 100.00%. The average reliable rates of the DDPG controller and PPO controller are 84.88 and 96.88% respectively. In the second scenario, one point is selected as the initial take-off position of the football ( Figure 11B ). In this initial take-off position, three specified directions where the football flies over the human wall are tested. That is, three states are constructed in the second scenario to test controllers. In this scenario, only 4 trajectories are not control in the total of 600 trajectories under the effect of the proposed controller. The average reliable rate of the proposed controller is 99.33%. The DDPG's average reliable rate in the second scenario is 96.17%. Notice that the PPO controller do not finish the reliability tests due to its terrible control policy. Figure 12 The results of the reliability tests. The reliability in uncertain environments is also an important index to evaluate the controller's performance. In this paper, the aerodynamic model of football with parameter uncertainties is regarded as the uncertain environment. Due to the strong non-linear of the football model, there may be more than one set of initial velocities to meet the requirements of the specified flight purpose. Meanwhile, the same initial velocities may generate different trajectories due to the parameter uncertainties. Thus, high reliability means that the expected reward under the specified state-action pair can be estimated accurately. And the controller can find a good set of initial velocities from multiple possible initial velocities, reducing the effects of the parameter uncertainties on the flight trajectories. According to test data in Figure 12 , the reliabilities of the proposed controllers are approaching or equal to 100% in both two football flight scenarios, which is significantly better than DDPG and PPO controllers. The above results verify that the proposed controllers have great reliability and can find the best initial velocities to resist the adverse effects of uncertain environments. As analyzed in Section The immediate-return RL algorithm, the great reliability of the proposed controllers come from the average operation for reward. For the sake of comparison, two controllers based on the proposed algorithm without using the average reward are also trained. As shown in Figure 12 , the reliable rate of the controller without the average reward is reduced by 6.37% in the first scenario and reduced by 3.83% in the second scenario. Numerical results indicate that the average reward method can improve the reliability of the controller. Computing resource usage As analyzed in Section Complexity analysis, compared to existing RL algorithms, the network framework of the immediate-return RL algorithm is greatly simplified, and its complexity is reduced significantly. That is, when solving the same problem in the atypical MDPs, the immediate-return RL algorithm may consume fewer computing resources than existing RL algorithms. Therefore, taking the first scenario of the football trajectory control as an example, the computational resource requirements of different algorithms, i.e., immediate-return RL algorithm, DDPG, and PPO, are analyzed. In these tests, the hardware is a normal computer with Intel I5 8600k processor and Nvidia GPU RTX2060. And all networks are built by the Tensroflow. For unity, 300,000 training steps are provided for each tested algorithm. Then, the computing resources consumed by three tested algorithms are shown in Table 5 . As can be seen, the immediate-return RL algorithm reduces the CPU utilization by 18.8%, the memory utilization by 26.3%, computing time by 29.4%, and size of the networks by 25.0% than the DDPG. Compared to PPO, the immediate-return RL algorithm also reduces the CPU utilization by 13.3%, the memory utilization by 12.5%, computing time by 2.0%, and size of the networks by 14.2%. It should be noticed that the number of training steps is limited to 300,000 in all tests. However, the computing resource usage of the algorithms also depends on the number of training steps required. Since the convergence speed of both DDPG and PPO is slower than the proposed algorithm, they require much more training steps than the proposed algorithm in actuality (see Figure 8 ). As analyzed in Section Training process, the number of training steps used by the proposed algorithm is 66.7% of the DDPG and 20% of the PPO. That is, the advantage of proposed algorithm in computing time is greater than that shown in the Table 5 . Thus, the test data demonstrates that, when dealing with the same problem in the atypical MDPs, the immediate-return RL algorithm trains faster, occupies less CPU and Memory, and generates fewer networks than existing RL algorithms. Furthermore, it should be noted that the transfer processes of data between CPU and GPU also consumes computing resources. The simulations of the football flight also affect the usage of computing resources. Thus, the differences between the comparison results and the theoretical analysis in Section Complexity analysis are acceptable. Table 5 Computing resources usage tests. \n CPU utilization \n \n Memory utilization (GB) \n \n Computing time (s) \n \n Size of the networks weights (KB) \n The proposed algorithm 26% 1.4 2,359 4,682 DDPG 32% 1.9 3,342 6,243 PPO 30% 1.6 2,408 5,455"
} | 7,638 |
24526260 | null | s2 | 630 | {
"abstract": "Carbohydrate feedstocks are at the root of bioindustrial production and are needed in greater quantities than ever due to increased prioritization of renewable fuels with reduced carbon footprints. Cyanobacteria possess a number of features that make them well suited as an alternative feedstock crop in comparison to traditional terrestrial plant species. Recent advances in genetic engineering, as well as promising preliminary investigations of cyanobacteria in a number of distinct production regimes have illustrated the potential of these aquatic phototrophs as biosynthetic chassis. Further improvements in strain productivities and design, along with enhanced understanding of photosynthetic metabolism in cyanobacteria may pave the way to translate cyanobacterial theoretical potential into realized application."
} | 205 |
24026984 | PMC3920635 | pmc | 631 | {
"abstract": "The ability of unicellular green algal species such as Chlamydomonas reinhardtii to produce hydrogen gas via iron-hydrogenase is well known. However, the oxygen-sensitive hydrogenase is closely linked to the photosynthetic chain in such a way that hydrogen and oxygen production need to be separated temporally for sustained photo-production. Under illumination, sulfur-deprivation has been shown to accommodate the production of hydrogen gas by partially-deactivating O 2 evolution activity, leading to anaerobiosis in a sealed culture. As these facets are coupled, and the system complex, mathematical approaches potentially are of significant value since they may reveal improved or even optimal schemes for maximizing hydrogen production. Here, a mechanistic model of the system is constructed from consideration of the essential pathways and processes. The role of sulfur in photosynthesis (via PSII) and the storage and catabolism of endogenous substrate, and thus growth and decay of culture density, are explicitly modeled in order to describe and explore the complex interactions that lead to H 2 production during sulfur-deprivation. As far as possible, functional forms and parameter values are determined or estimated from experimental data. The model is compared with published experimental studies and, encouragingly, qualitative agreement for trends in hydrogen yield and initiation time are found. It is then employed to probe optimal external sulfur and illumination conditions for hydrogen production, which are found to differ depending on whether a maximum yield of gas or initial production rate is required. The model constitutes a powerful theoretical tool for investigating novel sulfur cycling regimes that may ultimately be used to improve the commercial viability of hydrogen gas production from microorganisms. Biotechnol. Bioeng. 2014;111: 320–335. © 2013 The Authors. Biotechnology and Bioengineering Published by Wiley Periodicals, Inc.",
"introduction": "Introduction Although the ability of the unicellular microorganism Chlamydomonas reinhardtii to photosynthetically produce hydrogen gas from water under illumination has been known for over 60 years (Gaffron and Rubin, 1942 ), until recently it remained largely a biological curiosity as hydrogen producing iron-hydrogenase is inhibited by oxygen co-produced from the photosynthetic pathway under normal illumination and nutrient conditions (Benemann et al., 1973 ; Ghirardi et al., 1997 , 2000 ). Thus photosynthetic growth and hydrogen production are incompatible and need to be spatially or temporally separated in order to achieve significant hydrogen production. Melis et al. ( 2000 ) proposed a groundbreaking two-stage process for temporally separating the hydrogen and oxygen components of the photosynthetic pathway: cells are grown as normal in a sulfur-replete media and then in a second non-growth stage, partial deactivation of the oxygen-evolving photosystem II (PSII) occurs in response to sulfur-deprivation. In essence, during water splitting in PSII, the sulfur-rich reaction-center D1 proteins are damaged and need to be replaced (Mattoo and Edelman, 1987 ). In the absence of sulfur, D1 protein biosynthesis is impeded and the PSII repair cycle is blocked (Wykoff et al., 1998 ), leading to a reduction in oxygen production to a low level (Melis et al., 2000 ). Aerobic respiration and the light-dependent activity of photosystem I (PSI) are not directly affected by sulfur-deprivation (Cao et al., 2001 ; Davies et al., 1994 ; Melis et al., 2000 ; Zhang and Melis, 2002 ). After approximately 24 h under illumination, the rate of oxygen produced from photosynthesis is less than the rate of oxygen consumed by respiration; in a sealed container, the cells consume dissolved oxygen in the medium and the culture becomes anaerobic (Ghirardi et al., 2000 ; Kosourov et al., 2002 ; Melis et al., 2000 ; Zhang et al., 2002 ). In addition, during this time electrons result from the catabolism of endogenous substrates such as protein and starch (e.g., Chochois et al., 2009 ; Fouchard et al., 2005 ; Posewitz et al., 2004 ), both of which have been shown to increase significantly in the initial stages of sulfur-deprivation before hydrogen is produced (Fouchard et al., 2005 ; Kosourov et al., 2002 ; Melis et al., 2000 ; Posewitz et al., 2004 ). These events cause morphological changes in the cells during hydrogen production (Zhang et al., 2002 ). During dark fermentation ethanol acts as an electron sink for any reducing equivalents produced, but ethanol is harmful to the cell (Kennedy et al., 1992 ). In the light, under sulfur-deprivation, the partially active respiratory chain does not suffice as an electron sink and nor does the Calvin cycle since Rubisco, a necessary sulfur-rich enzyme in carbon fixation, is broken down and not synthesized (White and Melis, 2006 ; Zhang et al., 2002 ). The oxygen sensitive iron-hydrogenase enzyme on the thylakoid membrane is activated under these conditions and steps in as a major electron sink, re-oxidizing potentially harmful electrons produced from both the PSII-dependent (via water splitting) and the PSII-independent (fermentation) pathways, yielding H 2 gas for around 100 h in the light (Fouchard et al., 2005 ; Happe et al., 2002 ; Hemschemeier et al., 2008 ; Kosourov et al., 2002 ; Melis et al., 2000 ). The catabolic PSII-independent pathway is thought to contribute 20% of the hydrogen production and the PSII-dependent pathway contributes 80% (Fouchard et al., 2005 ; Volgusheva et al., 2013 ). Substantial hydrogen production ceases after around 120–140 h of sulfur-deprivation, thought to be due to depletion of the endogenous substrate available for catabolism (see Melis, 2002 ). Hence, there is a metabolic transition between an aerobic state with photosynthetic growth and an anaerobic state characterized by fermentation, H 2 production and biomass reduction (Hemschemeier et al., 2008 , see also Fig. 1 ). If sulfur is added to the culture once hydrogen production has ceased, the cells, and particularly PSII, can repair; cycles of oxygen production under S -sufficiency and H 2 production under S -deprivation can result (e.g., Ghirardi et al., 2000 ). Figure 1 Schematics of the intracellular processes and pathways that occur under normal, sulfur-replete conditions ( panel a ) and during sulfur-deprivation ( panel b ). Light gray arrows and text indicate an inactive pathway/process. In panel (a), sufficient sulfur levels allow maximal PSII repair. Electron flow (dashed arrows) from PSII to PSI leads to ATP synthetase and oxygen production that inhibits the activity of the iron-hydrogenase (thick black line), where the Calvin cycle is active. Under sulfur-deprivation (panel b), PSII activity decreases, fermentation begins (releasing minimal quantities of sulfur and electrons) and low Calvin cycle activity, caused by Rubisco depletion, activates the iron-hydrogenase under anaerobic conditions. The above description of the interplay between cellular processes is a simplification of very complex dynamics that whilst gaining general acceptance in the research community is subject to improvement (for recent reviews see Antal et al., 2010 ; Ghysels and Franck, 2010 ). Although the promising sulfur-deprivation protocol allows for significant hydrogen production, the efficiency of the two-stage process and the yields of hydrogen need to be improved to allow for commercial exploitation (see for example Das and Veziroglu, 2008 ; Melis, 2002 ). Scoma et al. ( 2012 ) demonstrated hydrogen production from green algae from solar light for the first time, but found that light conditions and mixing had a large effect on the H 2 yield (see also Giannelli et al., 2009 ) (which is expected since the collective swimming behavior of such species is sensitive to light conditions, which in turn affects photosynthetic efficiency; Bees and Croze, 2010 ; Williams and Bees, 2011b ). Furthermore, a large downtime arises due to sulfur-cycling between anaerobic sulfur-deprived hydrogen production and aerobic, sulfur-replete recovery periods. In order to advance beyond the standard two-stage process it is first necessary to understand the system within the limits of this procedure. Strategies for the optimization of hydrogen gas production via the two-stage process can be designed and tested using dynamical models to represent the main pathways and processes of the system. To this end, we construct a simple mechanical, mathematical model of an algal culture that can describe sulfur-deprived hydrogen production in C. reinhardtii from a careful consideration of the biology and biochemistry, including important feedback pathways. The model is general in the sense that it captures both sulfur-deprived and sulfur-replete conditions. Beyond non-mechanistic approaches (Jo et al., 2006 ; Jorquera et al., 2008 ), there are two mechanistic models of aspects of the algal system under these conditions. Park and Moon ( 2007 ) constructed three separate state models of the biochemical photosynthetic processes involved in hydrogen production and specifically modeled eight primary metabolites. The release of hydrogen gas and the effects of illumination were explicitly modeled, but the role of endogenous substrates was omitted. Furthermore, the model is a discrete, multi-state model rather than a continuous formulation, and parameters values were difficult to identify. Fouchard et al. ( 2009 ) improved upon this approach by formulating a continuous description of the role of sulfur and light limitation in photosynthetic growth and anaerobiosis under general conditions and applied their model to the case of sulfur-deprivation, but the model stopped short of modeling the production of hydrogen gas. Model validation and optimization were considered by Degrenne et al. ( 2011 ). In this study we shall improve upon previous work by modeling the principal mechanisms for the whole hydrogen production system, including feedback between sulfur uptake, photosynthetic growth, endogenous substrate, and the release of H 2 gas. There are elements that are modeled in a similar fashion to Fouchard et al. ( 2009 ). In particular, both intra-and extra-cellular sulfur are considered and we describe the uptake of external sulfur using a modified Monod formulation (Monod, 1949 ) and illumination and photosynthetic activity are dynamically coupled, since it is well known that culture growth has an effect on the light available for photosynthesis. We describe the effects of sulfur-deprivation on the rate of photosynthesis using a similar modified-Droop relationship (Droop, 1968 , 1979 ), and the use of sulfur in PSII repair, and the release of oxygen from PSII and its consumption in respiration are also included. But, significantly, the current approach extends and refines previous work in a number of ways. Firstly, we model explicitly and mechanistically the initial storage and subsequent catabolism of endogenous substrate: protein breakdown in particular is important due to the release of small amounts of sulfur that can permit residual PSII activity, a key source of electrons for H 2 production (e.g., Fouchard et al., 2005 ; Melis et al., 2000 ). We model substrate storage as dependent on the illuminated and S -dependent photosynthetic pathway (since proteins and starch are made via the Calvin cycle) and substrate breakdown (fermentation) as an emergency response to anaerobiosis, which also provides electrons to the hydrogenase. Culture growth can then be modeled as a function of endogenous substrate. These aspects differ from Degrenne et al. ( 2011 ) and Fouchard et al. ( 2009 ), since in these articles, changes in biomass are partitioned into growth and starch accumulation and these processes are not modeled independently (both depend in the same way on photosynthetic rate), protein dynamics and fermentation are not modeled explicitly, and feedback between substrate catabolism and sulfur release for PSII is not incorporated. The model could not adequately capture observed starch accumulation dynamics (Degrenne et al., 2011 ). The new description also deviates from the previous culture growth models in that it allows for both culture growth and biomass reduction under nutrient limitation (as shown experimentally in Zhang et al., 2002 ). Furthermore, it provides feedback pathways between growth, substrate catabolism and S -dependent photosynthesis. And finally, hydrogen production is modeled explicitly as a system output that is dependent on light and electron donation via both the PSII-independent and PSII-dependent pathways and is inhibited by oxygen within the culture (Ghirardi et al., 1997 , 2000 ). The model presented here consists of a set of coupled ordinary differential equations driven by evolving culture conditions (see Williams, 2009 ). In the following sections, the model is constructed from a mechanistic perspective and the solutions explored numerically. As in previous publications in this area, parameter estimation and the determination of functional forms were considerable challenges. Our objective was to produce a robust mechanistic model that exhibits the same qualitative trends as observed in experiments, rather than to refine parameter values arbitrarily to obtain quantitative agreement. Parameter values or estimated ranges were obtained from published experimental studies (see “Supplementary Material”) and the model was then employed to probe the system subject to the constraints of the two-stage process outlined above. Model results are compared with published experimental data, and optimal external sulfur and illumination conditions are determined. In a subsequent paper, novel sulfur-cycling strategies will be explored for optimizing hydrogen production outside the confines of the two-stage process.",
"discussion": "Discussion and Conclusions A simple mechanistic model has been constructed to describe sulfur-deprived hydrogen production in green algae. By modeling mechanistically, we have significantly simplified this complex system to just six variables. Key features of the model, including sulfur-dependent photosynthesis, growth, changes in endogenous substrate, and hydrogen gas release, have been incorporated. Solutions were obtained for the standard values of the parameters and with a range of initial conditions. The experimental studies of Kosourov et al. ( 2002 ), Melis et al. ( 2000 ), and Zhang et al. ( 2002 ) guided the construction of the model and some parameters in the growth and hydrogen functions, in particular, were extrapolated from the experiments therein. For example, the hydrogen production rate constant k 4 was taken from measured data in Kosourov et al. ( 2002 ) for the case of S 0 = 0 μM, thus the simulated hydrogen dynamics match the experimental data for S 0 = 0 μM reasonably well, as expected. However, the model was not fit to the data, and the hydrogen dynamics for different initial external sulfur and illumination conditions can still be compared independently with experimental data in order to test whether the model correctly captures the system dynamics under different conditions. Better independent measurements of the parameters, rather than fitting, should be the focus of future research efforts. Encouragingly, good qualitative and quantitative agreement was obtained between experimental results and model simulations for H 2 yield for different initial external sulfur concentrations, S 0 : after 140 h if S 0 = 25 μM or S 0 = 50 μM, the model predicts yields of h = 168 mL H 2 /L culture and h = 213 mL H 2 /L culture, respectively, in good agreement with Kosourov et al. ( 2002 ) ( h = 127 and h = 159 mL H 2 /L culture, respectively). The optimal S 0 for maximum hydrogen output over a fixed period was found to be a dynamic balance between high culture density, light limitation, and production start time. Hydrogen production onset time also corresponded approximately to experimental results: for S 0 = 25 μM and S 0 = 50 μM, t = 36.4, and t = 45.2 h, respectively, compared to t = 43–49 h in Kosourov et al. ( 2002 ). In simulation results, hydrogen production began almost as soon as the system became anaerobic when S = 0 μM, as in Zhang et al. ( 2002 ), but Kosourov et al. ( 2002 ) found a slight delay between onset of anaerobiosis and hydrogen production. This delay was predicted by our model for S > 0 μM, due to slower sulfur decay causing an extended period of Calvin cycle activity. The initial rate of hydrogen production per cell was also investigated, and we found that it increased slightly then decreased substantially as S 0 increased. The relatively constant production rate per cell is consistent with experimental observations from Degrenne et al. ( 2011 ) and Zhang et al. ( 2002 ) (where rate is per gram of biomass), but inconsistent with Kosourov et al. ( 2002 ), who found an increase in initial H 2 production rate per mole of chlorophyll for S 0 = 25 μM compared to S 0 = 0. We attribute increased hydrogen yield for S 0 ≈ 50 μM to increased cell volume fraction, as found experimentally by Zhang et al. ( 2002 ), rather than increased production rate per cell as proposed by Kosourov et al. ( 2002 ). The decrease in the initial rate for S 0 > 43.5 μM found from our model is also consistent with the trends found by Kosourov et al. ( 2002 ) and Zhang et al. ( 2002 ) for S 0 ≥ 25 μM. Likewise, we attribute corresponding decreases in H 2 yield to increased light limitation counteracting further increases in Λ when S 0 is large. The optimal sulfur concentration for maximizing this H 2 production rate (approximately 0 ≤ S 0 ≤ 29 μM) was found to be different from the optimal sulfur for increasing overall yield ( S 0 = 89.9 μM). Thus methods of optimization of the hydrogen production system depend on whether maximum cell activity or maximum H 2 output per culture is required. Model simulations for changes in illumination are consistent with experimental data from Hahn et al. ( 2004 ) and Kim et al. ( 2006 ): increasing the light intensity I 0 can significantly increase yields up to an optimal value due to earlier onset of production and increased culture density and PSII-dependent electron flow. However, increasing I 0 beyond the optimal value decreases H 2 yields due to increased photo-damage, as in Kim et al. ( 2006 ). Simulation results predict an optimal light intensity for total H 2 output of I 0 = 146.5 μmol m −2 s −1 for S 0 = 0 μM or I 0 = 340 μmol m −2 s −1 for S 0 = 50 μM with illumination from both sides, which are of the same order as those predicted by Park and Moon ( 2007 ) (238 μE m −2 s −1 ) and Kim et al. ( 2006 ) (200 μE m −2 s −1 ). Using the model of Degrenne et al. ( 2011 ), Fouchard et al. ( 2009 ) found qualitatively similar results: they predicted that a high hydrogen yield would require high external sulfur and light irradiance. Experimental data supported this conclusion. However, in those studies H 2 gas production was not explicitly modelled but was extrapolated from biomass and starch concentrations. We find that higher yields of H 2 are found for higher cell volume fraction: for S 0 = 0, h = 247 mL H 2 /L with Λ 0 = 0.0045 and 106 mL H 2 /L culture when Λ 0 = 0.00225, which supports the hypothesis that one may optimize H 2 yield by maximizing biomass. However, we caution that for sufficiently high initial sulfur and light conditions our model predicts diminished H 2 yields due to over-concentrated cultures and photo-damage. Melis ( 2002 ), Melis ( 2009 ), and Polle et al. ( 2002 ) suggested that truncating the chlorophyll antenna to decrease cellular absorbance (modelled as D C ) decreases wasted light and increases photosynthetic activity, which may increase the hydrogen yield. Model results also suggest that decreasing absorbance could optimize H 2 yield, provided that the light intensity I 0 is not too high, or D C is not too low, otherwise yields decrease due to increased photo-damage (as for high light intensities in this model and in Kim et al. ( 2006 ) and Park and Moon ( 2007 ). To our knowledge, this is the first simple mechanistic model of sulfur-deprived hydrogen production to include feedback between sulfur, photosynthetic growth, endogenous substrate, and hydrogen production. Good qualitative agreement is found between model simulations and experimental results. In order to model such a complex system, key assumptions were made. The role of starch was not modelled independently; instead, endogenous substrate is representative of both protein and starch in order to capture the dynamical feedback between sulfur, photosynthetic growth and fermentation. This may be a reasonable approximation, but the two may be better modelled separately, with growth a function of both. However, we do not expect this extension qualitatively to alter results. Additionally, a switch ( H Calvin ( s ; S 1 )) was used to close the system and specify that H 2 -producing hydrogenase requires both anaerobiosis and an inactive Calvin cycle to function as an electron sink (e.g., Happe et al., 2002 ; Hemschemeier et al., 2008 ; White and Melis, 2006 ), so a sealed system with high culture density leads to anaerobiosis due to light limitations but no hydrogen is produced (in accordance with Zhang et al., 2002 ). In this study, the switch had little effect when initial external sulfur was minimal and it allowed the omission of the complex Calvin cycle and the interplay between electron sinks from the model. It may be revealing to explore further the explicit nature of the coupling between the hydrogenase and the Calvin cycle. To describe the suspension, the cultures were assumed to be perfectly mixed and cell swimming behaviour was not described. Biased swimming is known to induce hydrodynamic instabilities, resulting in non-uniform distributions of cells, called bioconvection, in tens of seconds on length scales of centimeters. This significantly affects light transmittance and thus photosynthesis (Bees and Croze, 2010 ; Williams and Bees, 2011a , 2011b ), and could have a substantial impact on H 2 yield. All of these assumptions should be explored in future developments of the current model. As new data emerge, refinements of the parameter values and key mechanisms can be incorporated in the model. Perhaps more importantly, the current description is ideal for examining novel regimes for optimizing the total yield, or rates of production, of hydrogen gas produced under a range of sulfur-deprivation schemes. Such analysis may provide valuable insight into future commercialization of algal H 2 production and will be presented in a future article."
} | 5,718 |
32203116 | PMC7242416 | pmc | 632 | {
"abstract": "Interspecies hydrogen transfer in anoxic ecosystems is essential for the complete microbial breakdown of organic matter to methane. Acetogenic bacteria are key players in anaerobic food webs and have been considered as prime candidates for hydrogen cycling. We have tested this hypothesis by mutational analysis of the hydrogenase in the model acetogen Acetobacterium woodii . Hydrogenase-deletion mutants no longer grew on H 2 + CO 2 or organic substrates such as fructose, lactate, or ethanol. Heterotrophic growth could be restored by addition of molecular hydrogen to the culture, indicating that hydrogen is an intermediate in heterotrophic growth. Indeed, hydrogen production from fructose was detected in a stirred-tank reactor. The mutant grew well on organic substrates plus caffeate, an alternative electron acceptor that does not require molecular hydrogen but NADH as reductant. These data are consistent with the notion that molecular hydrogen is produced from organic substrates and then used as reductant for CO 2 reduction. Surprisingly, hydrogen cycling in A. woodii is different from the known modes of interspecies or intraspecies hydrogen cycling. Our data are consistent with a novel type of hydrogen cycling that connects an oxidative and reductive metabolic module in one bacterial cell, “intracellular syntrophy.”",
"introduction": "Introduction Molecular hydrogen is present only in trace concentrations (550 parts per billion) in the Earth’s atmosphere [ 1 ], but plays an important part in the global carbon cycle and is a major constituent of microbial metabolism. In anoxic ecosystems it is rapidly produced and consumed by microorganisms resulting in a large turnover [ 2 ]. Hydrogen connects different parts of the anaerobic food web and is usually produced by primary fermenters [ 3 ]. Fermentations typically yield between 1 and 4 mol of ATP per mol of sugar, and the maximum is only observed if electrons can be blown away into the environment as molecular hydrogen thus allowing the cells to make acetate according to Eq. ( 1 ) [ 4 ]: 1 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\rm{C}}_{6}{\\rm{H}}_{12}{\\rm{O}}_{6} +\t {\\rm{2H}}_{2}{\\rm{O}} + {\\rm{4ADP}} + {\\rm{4P}}_{\\rm{i}}{\\longrightarrow}{\\rm{2CH}}_{3}{\\rm{COOH}} \\\\ +\t {\\rm{2CO}}_{2} + {\\rm{4H}}_{2} + {\\rm{4ATP}} \\,\\,\\, \\Delta{\\rm{G}}^{0\\prime} = -206.3 \\,\\,{\\rm{{kJ/mol}}}$$\\end{document} C 6 H 12 O 6 + 2H 2 O + 4ADP + 4P i → 2CH 3 COOH + 2CO 2 + 4H 2 + 4ATP Δ G 0 ′ = 206.3 kJ∕mol However, hydrogen formation from reduced pyridine nucleotides or flavins is energetically unfavourable and growth according to Eq. ( 1 ) requires removal of hydrogen by a syntrophic partner such as a sulfate reducing bacterium, a methanogenic archaeon or an acetogenic bacterium [ 5 – 8 ]. The latter produces acetate according to Eq. ( 2 ): 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}$$4{\\rm{H}}_{2} +\t {\\rm{2CO}}_{2} + {\\rm{xADP}} + {\\rm{xP}}_{\\rm{i}}{\\longrightarrow} {\\rm{CH}}_{3}{\\rm{COOH}} \\\\ +\t {\\rm{2H}}_{2}{\\rm{O}} + {\\rm{xATP}}\\,\\,\\,\\Delta{\\rm{G}}^{0\\prime} = -95 \\,\\,{\\rm{{kJ/mol}}}$$\\end{document} 4 H 2 + 2CO 2 + xADP + xP i → CH 3 COOH + 2H 2 O + xATP Δ G 0 ′ = 95 kJ∕mol Since acetogens grow by conversion of H 2 + CO 2 to acetate, the reaction has to be coupled to net synthesis of ATP [ 9 ]. Detailed studies in the acetogenic model organism Acetobacterium woodii estimated the amount of ATP to 0.3 mol per mol of acetate produced [ 9 ]. In contrast to methanogenic archaea, acetogenic bacteria do not only grow lithotrophically according to Eq. ( 2 ) but also by fermentation [ 10 ]. Acetogenesis is a modular metabolism with an oxidative and a reductive branch [ 11 ] (Fig. 1 ). In the oxidative branch, hydrogen (during lithotrophic growth) or an organic carbon and energy source (during heterotrophic growth) are oxidised. Electrons are carried over to the reductive branch (the Wood-Ljungdahl pathway [WLP]) in which 2 mol of CO 2 are reduced to acetate according to Eq. ( 2 ). Overall, fermentation of fructose to three molecules of acetate by a combination of Eqs. ( 1 ) and ( 2 ) gives the highest ATP yield in fermenting bacteria of 4.3 mol ATP/mol of sugar [ 11 ]. Fig. 1 The modularity of acetogenesis in A. woodii . Shown are the oxidation of fructose to acetate in the oxidative branch (left) and the reduction of CO 2 to acetate (right) in the reductive branch (WLP). Redox balancing is achieved by a third module, in which the Rnf complex and the electron-bifurcating hydrogenase produce the reductants required for the the WLP. Fd, ferredoxin; Fd 2− , reduced ferredoxin; THF, tetrahydrofolate; HDCR, hydrogen-dependent CO 2 reductase; CODH/ACS, carbon monoxide dehydrogenase/acetyl-CoA synthetase; Co-FeS-P, corronoid iron-sulfur protein. Electron carriers involved in transferring electrons from the oxidative to the reductive branch in acetogens are NADH, NADPH, or reduced ferredoxin [ 9 ]. A third module, the redox balance module, ensures that the different reduced electron carriers from the oxidative module are fed in and are converted to the specific redox carriers required by the WLP [ 11 ]. In the model acetogen A. woodii , 2 mol of NADH from glycolysis and 2 mol of reduced ferredoxin from conversion of pyruvate to acetyl-CoA are converted to 1 mol of hydrogen, 2 mol of NADH and 1 mol of reduced ferredoxin by the combined action of the Rnf complex and the electron-bifurcating hydrogenase [ 11 , 12 ]. NADH is used as reductant for the methylene-tetrahydrofolate (THF) reductase [ 13 ] and the methylene-THF dehydrogenase [ 14 ], and reduced ferredoxin is used by the CO dehydrogenase/acetyl-CoA synthase as electron donor for CO 2 reduction in the carbonyl branch of the WLP [ 15 , 16 ]. In contrast, the first enzyme used for CO 2 reduction in the methyl branch of A. woodii (Fig. 1 ), the hydrogen-dependent CO 2 reductase (HDCR), uses H 2 as reductant in vitro [ 17 ], indicating the need for electron transfer via hydrogen from the oxidative module (glycolysis) to the reductive module (CO 2 reduction). On the other hand, the purified HDCR can also accept electrons from reduced ferredoxin, albeit with 17-fold lower activities [ 17 ]. To address a potential hydrogen transfer from the oxidative to the reductive branch of the acetogenic metabolism, we have deleted the only hydrogenase in A. woodii and studied the phenotype of the mutant.",
"discussion": "Discussion Sulfate reducing bacteria grow by oxidation of lactate or other organic substrates coupled to energy conservation by the reduction of sulfate [ 29 , 30 ]. They also grow lithotrophically and, indeed, hydrogen is the most effective hydrogen donor for sulfate reduction [ 4 , 29 ]. This led Odom and Peck to postulate hydrogen cycling as a general mechanism for energy coupling in sulfate reducing bacteria [ 31 ] and they postulated this mechanism to be present in methanogens and acetogens as well [ 5 ]. According to their model, hydrogen is produced inside the cell and diffuses across the membrane to the periplasm. The authors speculated that the same organism oxidises hydrogen to 2 H + + 2e − at the periplasmic side of the membrane, thus, producing scalar protons that create a proton motive force (pmf) across the cytoplasmic membrane that drives ATP synthesis. This mechanism requires a soluble, cytoplasmic hydrogenase and a periplasmic, membrane-bound hydrogenase. Biochemical and physiological experiments performed are in line with this hypothesis [ 32 ]. More than 35 years later, hydrogen cycling as a mode of pmf generation was directly demonstrated by mutational analyses for the archaeon Methanosarcina barkeri [ 33 ]. When grown on methanol, 25 % of the substrate is oxidised to CO 2 to generate the electrons needed to reduce the other 75 % to methane. The oxidation of methanol is coupled to the reduction of protons to H 2 , as catalysed by the F 420 -reducing hydrogenase. Hydrogen diffuses out of the cell and is harnessed by the membrane-bound Vht hydrogenase that reduces the membrane-integral electron-carrier methanophenazine, the electron donor for the respiratory enzyme, the heterodisulfide reductase [ 33 ]. The proton potential established by the heterodisulfide reductase drives the synthesis of ATP in this archaeon [ 34 ]. Deletion of the soluble hydrogenase abolished hydrogen formation and deletion of the Vht hydrogenase was lethal, demonstrating nicely that there is hydrogen cycling in one cell, important for energy conservation [ 33 ]. The second type of hydrogen transfer in anoxic ecosystems is observed between different species and is called interspecies hydrogen cycling [ 5 ]. There, one partner oxidises a substrate linked to the production of hydrogen. Since hydrogen formation is thermodynamically unfavourable, the hydrogen producer can only survive if the hydrogen concentration in the environment is kept low by hydrogen oxidising microorganisms such as methanogenic archaea [ 6 , 7 ]. Whether or not hydrogen cycling occurs in the ecophysiologically relevant group of acetogenic bacteria remained to be established. In this report, we show clear evidence that A. woodii releases hydrogen into the atmosphere but only ~8.4 mmol hydrogen per mol of fructose was liberated, which is less than 10 % of the acetic acid generated, but similar to the result of Braun et al. [ 35 ]. This result indicates that the HDCR efficiently captures the hydrogen produced by the electron-bifurcating hydrogenase. If the electron-bifurcating hydrogenase is missing, cells growing on fructose, ethanol or lactate are unable to produce hydrogen needed for CO 2 reduction by the HDRC and therefore are unable to grow, except when they grow mixotrophically on fructose + hydrogen or fructose + formate. This is also clear evidence that hydrogen evolution takes place inside the cell and hydrogen is directly used within the cell before it can diffuse through the cell membrane. Since A. woodii lacks a membrane-bound hydrogenase, hydrogen oxidation is not linked to energy conservation and, therefore, does not fall into the category “intraspecies hydrogen cycling linked to energy conservation”, as postulated by Odom and Peck [ 5 ]. In contrast, A. woodii combines the metabolic features of two syntrophic partners in one bacterial cell. Depending on the environmental conditions A. woodii can play the part of the fermenting partner as in coculture with a methanogen [ 36 ] or the hydrogen consuming partner in syntrophic interactions. When grown together with methanogens on H 2 + CO 2 , acetogens usually would be outcompeted by methanogens since methanogenesis from H 2 + CO 2 delivers much more energy than acetogenesis [ 37 ]. However, under certain conditions acetogens dominate as H 2 + CO 2 -consuming partner in syntrophic interactions. For example, hydrogen-utilising acetogens compete successfully with hydrogen-utilising methanogens in wood-feeding termites [ 38 ]. Further studies showed that acetogens can outcompete methanogens for hydrogen at a pH of 6.2 and also at more acidic pH values [ 39 ] and at low temperature, for example at an in situ temperature of 4 °C in sediments of Lake Constance [ 40 ]. The partial pressure of hydrogen measured in pore water is too low to allow growth of pure cultures and it is speculated that the in situ partial pressure of hydrogen might be higher for acetogens living in close proximity to the hydrogen-producing organism [ 38 ]. Here, we demonstrate that, in addition, A. woodii can also play both parts—the fermenting and the hydrogen consuming part—in one cell. This is the closest proximity one can get. We propose to call this novel type of hydrogen cycling that connects an oxidative and reductive metabolic module in one bacterial cell “intracellular syntrophy.” Acetogenic microorganisms are phylogenetically very divers [ 41 ]. Acetogenesis has been found in different phylogenetic clades of the Bacteria and has been studied there for the last hundred years, but rather recently acetogenesis was also found in different phyla of the Archaea . This is not only based on genomic but also on physiological analyses [ 42 – 44 ]. These archaea are supposed to grow autotrophically on H 2 + CO 2 to produce acetate but also on organic substrates. It is postulated that they ferment the organic substrates such as short chain fatty acids to acetate, alcohols and molecular hydrogen [ 43 ]. At the same time the WLP can act as electron sink to make the fermentation energetically possible, which makes fermentation independent from a syntrophic partner [ 44 ]. Our mutational analyses in A. woodii fully support this model. The coupling of hydrogen-dependent CO 2 reduction to hydrogen-producing fermentations allows acetogens to grow, for example, in the deep biosphere on substrates that are otherwise inaccessible for energetic reasons. Hydrogen production from fermentation is common in aquatic sediments and hydrogen production and consumption is essential for anaerobic food webs [ 45 ]. Hydrogen production from organic substrates also puts a new perspective on the origin of the eukaryotic cell [ 43 ]. Many acetogens have just one soluble hydrogenase like A. woodii [ 46 ], others such as Moorella thermoacetica [ 47 ], Thermoanaerobacter kivui [ 48 ] or Heimdallarchaeota and Odinarchaeota [ 43 ] have membrane-bound, ion-translocating hydrogenase activities as hydrogen consuming respiratory enzymes. Their role in hydrogen transfer is still an open question."
} | 3,487 |
27155452 | PMC5015573 | pmc | 633 | {
"abstract": "Sulfide mineral processing often produces large quantities of wastewaters containing acid-generating inorganic sulfur compounds. If released untreated, these wastewaters can cause catastrophic environmental damage. In this study, microbial fuel cells were inoculated with acidophilic microorganisms to investigate whether inorganic sulfur compound oxidation can generate an electrical current. Cyclic voltammetry suggested that acidophilic microorganisms mediated electron transfer to the anode, and that electricity generation was catalyzed by microorganisms. A cation exchange membrane microbial fuel cell, fed with artificial wastewater containing tetrathionate as electron donor, reached a maximum whole cell voltage of 72 ± 9 mV. Stepwise replacement of the artificial anolyte with real mining process wastewater had no adverse effect on bioelectrochemical performance and generated a maximum voltage of 105 ± 42 mV. 16S rRNA gene sequencing of the microbial consortia resulted in sequences that aligned within the genera Thermoplasma , Ferroplasma , Leptospirillum , Sulfobacillus and Acidithiobacillus . This study opens up possibilities to bioremediate mining wastewater using microbial fuel cell technology.",
"introduction": "1 Introduction The world's demand for metals requires the continued mining and processing of metal-bearing ores. A common treatment process for extracting valuable metals from metal-sulfide minerals involves crushing and subsequent flotation to create a mineral concentrate suitable for further (bio)hydrometallurgical treatment. In addition to the valuable metal-containing phases, the low economic value mineral pyrite (FeS 2 ) is also usually present. Pyrite is oxidized during ore crushing and flotation [1] , [2] , during which time the inorganic sulfur compound (ISC) thiosulfate ( S 2 O 3 2 − ) is produced (Eq. (1) ). (1) 2 FeS 2 + 3 . 5 O 2 + 4 OH − + H 2 O ⇒ 2 Fe ( OH ) 3 + 2 S 2 O 3 2 − Additional sources of ISCs in sulfide mineral processing include hydrogen sulfide from flotation of molybdenum and from ammoniacal thiosulfate leaching for gold recovery. The generated ISCs are contained in the processed waters that carry waste (gangue) minerals to tailings ponds, where the solid waste metal sulfides sink to the bottom of the pond. The tailings are covered in water to impair ingress of oxidation that, in turn, retards the release of acid and metals (reviewed in [3] ). In the tailings pond, acidophilic microorganisms can subsequently oxidize thiosulfate to tetrathionate ( S 4 O 6 2 − ; Eq. (2) ) and ultimately, the sulfur moiety will end up as sulfuric acid (Eq. (3) ), lowering the pH to values typically between 1 and 3. This provides requisite conditions for the growth of extremely acidophilic microorganisms that have an optimum growth pH < 3 [4] . (2) 2 S 2 O 3 2 − + 0 . 5 O 2 + 2 H + ⇒ 2 S 4 O 6 2 − + H 2 O (3) S 4 O 6 2 − + 3 . 5 O 2 + 3 H 2 O ⇒ 4 SO 4 2 − + 6 H + If released to recipient water bodies, these ISCs can pose environmental risks, including microbial oxidation to sulfuric acid and the depletion of dissolved oxygen (reviewed in [5] ). The most common treatment method for ISC-containing mining wastewaters is to raise the pH with ‘lime’ (CaO) before release of the water. A drawback of this technique is the production of a metal-containing sludge that requires safe disposal. Further investigated biotechnological solutions include metal precipitation with biologically produced hydrogen sulfide at near neutral [6] or acidic [7] , [8] pH, as well as biological oxidation of the ISCs to sulfuric acid prior to neutralization with lime and release to downstream water bodies [1] . However, neither of these methods have been extensively implemented at mining sites around the world. Microbial fuel cells (MFCs) are bioelectrochemical systems capable of sustainable microbial oxidation of a substrate in the anode compartment, while reducing electron acceptors in the cathode, with the flow of electrons forming an electrical current (reviewed in [9] ). The general principles of MFCs ( Fig. 1 ) are that the electron donor is oxidized in the anode compartment, often by microorganisms attached as a biofilm to the anode surface and, in the absence of competing electron acceptors, they pass their electrons to the anode. Microorganisms able to donate electrons at the anode can be termed ‘electricigens’ and the most commonly investigated species come from the genera Geobacter and Shewanella \n [10] , [11] . Electrons donated to the anode are chemically or biologically reduced at the cathode. Research into MFCs has intensified in the past decade, and they have been extensively tested for treatment of organic carbon containing wastewaters at neutral pH to simultaneously generate a current and remove the waste product (reviewed in [12] ). Most MFCs use microbial communities from ‘non-extreme’ environments such as municipal wastewater, activated sludge or sediment [13] , [14] , [15] , [16] , while MFCs using extremophilic microorganisms are less common. Several studies have documented the use acidophiles in MFCs fed with organic carbon and energy sources [17] , [18] , [19] , [20] , [21] . However, to treat ISCs in mining wastewaters at acidic pH values, it is required to utilize microorganisms that donate electrons to the anode during growth using carbon dioxide and inorganic compounds as carbon and energy sources, respectively. However, research into the use of ISCs in MFCs is rare. A single study has shown that a mixed culture of acidophilic microorganisms is capable of generating electricity from tetrathionate, with Acidithiobacillus ferrooxidans and Ferroplasma spp. being the dominant populations present in the anolyte and on the bioanode surface [22] . The maximum current and power densities achieved during operation were 79.6 mA m −2 and 13.9 mW m −2 , respectively, but the low coulombic efficiency of 5% suggested that processes other than donation of the electrons to the anode also occurred [22] . In this study, the feasibility of using acidophilic microorganisms in MFCs fed with wastewater from an industrial sulfide mineral flotation process was investigated. If successful, the acid-generating ISCs would be removed before pH neutralization of water and release to recipient water bodies. The specific aims were to select suitable acidophilic microbes to remove the ISC compounds while simultaneously producing an electrical current in MFCs.",
"discussion": "3 Results and discussion 3.1 Enrichment of acidophilic microorganisms in the anode compartment Two Kristineberg enrichment MFCs were operated for 61 days: one with 5 mM tetrathionate and the second with a mixture of 3 mM tetrathionate plus 3 mM thiosulfate ( Supplemental File 2 ). The tetrathionate-fed MFC reached a maximum whole cell voltage of 35 mV on day 61, whereas the tetrathionate-plus thiosulfate-fed MFC took until day 59 to reach a whole cell voltage of approximately zero. The Vaasa acid sulfate soil MFCs fed with 5 mM tetrathionate or 5 mM thiosulfate took a longer time to generate a positive whole cell voltage compared to the Kristineberg tetrathionate-fed MFC and peaked at voltages of 48 and 42 mV, respectively ( Supplemental File 2 ). The time to select for a microbial consortium that generated a positive whole cell potential took longer than 15 days in all cases. As these MFCs were operated for an extended time period it was necessary to add further substrate that resulted in a dip in cell voltage that took approximately 10 days to recover. It is possible that this dip was due to a small concentration of oxygen entering the system that was consumed before electron donor addition to the anode recommenced. The anode potentials with 5 mM tetrathionate as substrate were in the range of 550–600 mV ( Supplemental File 2 ) that was considerably higher than the calculated value of −272 mV (versus Ag/AgCl electrode at pH 2, 20 °C, and 2 g L −1 tetrathionate). Despite the variations in microbial communities present, this result was consistent with previously reported tetrathionate-fed acidic MFCs and was attributed to MFC losses and lower reaction efficiencies [22] . The whole cell voltages produced and coulombic efficiencies were lower than those achieved with acidic anolyte MFCs fed with organic carbon and energy containing wastewaters [21] , [36] . The increased whole cell voltage with organic carbon was likely due to the higher energy present in organic carbon compounds. In addition, the low coulombic efficiency in this study may have been due to the energy required for extremely acidophilic microorganisms to maintain pH homeostasis, such as proton pumping out of the cytoplasm (reviewed in [37] ) and reverse electron transport to generate NADH [31] . The whole cell voltages in this study were also lower than obtained with 2.0–2.5 g L −1 (7.7–9.6 mM) tetrathionate [22] , potentially due to the higher substrate concentration in the previous study. 3.2 Sulfide mineral flotation process water MFC performances The average cation exchange membrane whole cell voltage fed with 5 mM tetrathionate (in the absence of process wastewater) was 76 ± 7 mV ( Table 1 and Fig. 2 ). The coulombic efficiency was 3.0% and a power density of 2.5 mW m −2 , values similar to previous observed tetrathionate-fed MFCs [22] . The whole cell voltage values fluctuated when new substrate was added ( Fig. 2 ), indicating that cell voltage depended on the availability of substrate. Increasing the percentage of process wastewater slightly decreased the average and maximum whole cell voltages until 100% process wastewater plus tetrathionate was added when the whole cell voltage had an average of 105 ± 42 mV and a maximum of 166 mV. The increase in whole cell voltage may have been due to transfer of electrons to the anode from the 1.22 mM thiosulfate present in the process wastewater. The maximum whole cell voltage in the presence of 100% process water and no additional tetrathionate decreased to 29 ± 9 mV, likely due to the decreased availability of tetrathionate. A parallel control MFC with artificial mineral salts medium containing tetrathionate (but no process wastewater) also resulted in substrate utilization coupled to generation of a whole cell potential, although the achieved power densities and coulombic efficiencies were lower than in the presence of process wastewater ( Table 1 ). This suggested that any potential contaminants in the process wastewater (e.g. residues of flotation chemicals [38] ) did not have any negative effects on the MFC performance. Similar whole cell voltage values and anode potentials were achieved using the anion exchange membrane MFCs ( Table 1 ). However, the power densities and coulombic efficiencies were lower than with the cation exchange membrane. A stable whole cell voltage was observed in the first 15 days with 100% process wastewater without added tetrathionate after which additional tetrathionate (1 mM) was amended after 16 and 24 days, resulting in increased whole cell voltage ( Table 1 ). This suggested the MFC was substrate-limited under batch mode, and it would be interesting to investigate whether higher voltage could be achieved in continuous mode. An anion exchange membrane control MFC (with no process wastewater) also generated an electrical voltage with no statistically valid difference between the data sets ( Table 1 ). This also supported the hypothesis that the process wastewater did not inhibit the MFC. 3.3 Microbial community in the anion exchange membrane MFC The microbial community of the tetrathionate-fed Kristineberg-inoculated MFC was dominated by strains with 16S rRNA genes that aligned (in order of greatest to least number of sequence reads) within the genera Thermoplasma , Ferroplasma , Leptospirillum , and Acidithiobacillus ( Supplemental File 3 ). This was similar to the microbial community of the Kristineberg acid mine drainage stream (rather than the stream sediment used in this study) that is dominated by Acidithiobacillus spp., but also contains both Thermoplasma - and Ferroplasma -like species [23] . The presence of strains with 16S rRNA gene sequences similar to Acidithiobacillus was not surprising, as some species grow anaerobically utilizing elemental sulfur [39] and A. ferrooxidans is present in other acidic, tetrathionate-fed MFCs [22] . The low number of Acidithiobacillus -like 16S rRNA gene reads compared to, for instance, Ferroplasma -like species that have not previously been shown to utilize ISCs, was surprising. However, it is believed that the Ferroplasma -like population played a role in the MFCs, as it was retained during successive subculturing and re-inoculation in increasing concentrations of the mining process water ( Fig. 3 ). In addition, the presence of 16S rRNA gene sequences similar to those from the genus Leptospirillum was also surprising, as characterized species from this genus have not previously been shown to grow anaerobically [40] . It is possible that the Leptospirillum -like population was carried over from the acid mine drainage stream sediment inoculum [23] , although they potentially also play an active role in the MFCs, as similar populations were reported in tetrathionate-fed MFCs [22] . The microbial community from the anode compartment of the tetrathionate-fed acid-sulfate-soil-inoculated MFC was also investigated by 16S rRNA gene sequencing. The identified community consisted of 16S rRNA gene sequences that aligned within the Acidithiobacillus and Sulfobacillus genera ( Supplemental File 3 ). Some cultured Sulfobacillus species are facultative anaerobes [41] , although species from this genus have not previously been identified in MFCs. The 16S rRNA gene sequencing data identified some populations that for instance, have not previously been demonstrated to utilize ISCs. These results highlight the difficulty of assigning functional and metabolic traits based solely upon 16S rRNA gene similarity, and this type of data must be interpreted with caution. The microbial community from the tetrathionate-fed, anion exchange membrane MFC originally inoculated from the Kristineberg culture with increasing concentrations of mining process water continued to be dominated by 16S rRNA gene sequences that aligned within the Ferroplasmaceae family and Acidithiobacillus genus present in the inoculum ( Fig. 3 ). In addition, 16S RNA gene sequences that aligned within the Sulfobacillus genus were also selected that were not part of the dominating community in the preliminary MFCs. Finally, a ‘tail’ of low abundance species was present that might also play a role in substrate utilization and electricity generation in the MFC. This mixed microbial community was similar to that identified in tetrathionate-fed MFCs inoculated with mining process waters [22] , but varied from another low pH bioelectrochemical system in which organic carbon-oxidizing Acidiphilium spp. were detected on the anode of a pH 3 sediment/water interface microcosm in the Rio Tinto river [19] . 3.4 Role of microorganisms in electricity generation Cyclic voltammetry was applied to the anodes during electricity generation by the process wastewater test and mineral salts medium controls to investigate the role of microbial catalyzed electricity generation. The results showed that, compared to the abiotic control, cation exchange membrane MFCs inoculated with the Kristineberg acid mine drainage stream sediment produced more current during cyclic voltammetry ( Fig. 4 ). A similar result was obtained with the other MFCs operated at the same time ( Supplemental File 4 ). The current production indicated the acidophilic microorganisms at least partially mediated electron transfer to the anode. Additional abiotic controls were carried out to investigate electrical voltage generation in the absence of acidophilic microorganisms compared to data from the inoculated MFCs with 100% mining process water without additional tetrathionate ( Fig. 5 ). The whole cell voltage values of the inoculated MFCs dropped slowly in a linear manner compared to the rapid decrease in control MFCs ( Fig. 1 & Table 1 ). This suggested the ISC had a certain chemical energy that can be extracted as electricity in a chemical fuel cell, but the microorganisms increased the efficiency of the process by catalyzing substrate degradation and transferring electrons to the anode. This was also supported by the inoculation of acidophilic microorganisms into the anolyte that resulted in increased whole cell voltage ( Fig. 5 ). 3.5 Potential substrate utilization pathway A speculative pathway for substrate utilization in the ISC-fed MFCs is hydrolysis of tetrathionate to thiosulfate, elemental sulfur and sulfate carried out by Acidithiobacillus spp. [31] . Anaerobic growth of A. ferrooxidans with elemental sulfur as substrate has been suggested to occur, at least in part, by disproportionation to form hydrogen sulfide and sulfate [39] . This was supported by the end products from tetrathionate-fed MFCs [22] and accumulation of a yellow precipitate on the anode surface in this study (visual observation), that was confirmed as elemental sulfur by transmission electron microscopy [22] . 16S rRNA gene sequences that aligned with Ferroplasma and Thermoplasma genera were surprising since, although cultured species from this genera can grow anaerobically [42] , [43] , they have not been shown to grow utilizing ISCs [44] . It is possible that these populations were able to utilize organic carbon released from the autotrophic species [45] ; that the Ferroplasma- like species acted as an electron shuttle in a similar manner to synergistic interactions between contaminant-degrading species and electron transfer to the anode by e.g. Geobacter spp. (reviewed in [46] ); or alternatively, a novel species able to utilize ISCs had been selected. These hypotheses are under investigation via metagenomic and metatranscriptomic analyses of the anion exchange membrane MFC population. In conclusion, this study demonstrated that electrical currents can be generated by acidophilic microorganisms from real industrial wastewater, opening up the possibility of using this technology to bioremediate mining process wasters by bioelectrochemistry. A quick response time from tetrathionate addition to increased whole cell voltage indicated that electricity generation was highly dependent on the level of substrate, and it may be possible to increase cell voltage by continuously supplying the MFC with process wastewater. The microorganisms selected in the population included microorganisms with 16S rRNA gene sequences similar to those previously identified in an acidic MFC (e.g. Acidithiobacillus -like and Ferroplasma -like species), as well as a population aligning with the Sulfobacillus genus not previously identified in MFCs."
} | 4,746 |
35208662 | PMC8879049 | pmc | 634 | {
"abstract": "Coral ingestion by crown-of-thorns starfish (COTS) is an important cause of coral reef degradation, although the impacts of COTS feeding on coral-associated microbial communities are not well understood. Therefore, in this study, we analyzed the coral tissue-weight, Symbiodiniaceae density (SD), bacterial community composition, and the predicted functions of bacterial genes associated with Pocillopora corals in healthy portions and feeding scars, following COTS feeding. Coral tissue-weight loss rate in the feeding scars was 71.3–94.95%. The SDs were significantly lower in the feeding scars, and the SD-loss rate was 92.05% ± 2.12%. The relative abundances of bacterial communities associated with Pocillopora corals after COTS feeding changed significantly and were almost completely reorganized at the phylum and genus levels. Analysis of the microbial metagenomic-functional capacities showed that numerous physiological functions of the coral-bacterial holobionts in the feeding scars were different, including amino acid metabolism, xenobiotic biodegradation and metabolism, lipid metabolism, membrane transport, signal transduction, and cell motility, and all these capacities could be corroborated based on metagenomic, transcriptomic or proteomic technologies. Overall, our research suggests that coral holobionts may be destroyed by COTS, and our findings imply that bacterial communities in feeding scars could affect the health of Pocillopora corals.",
"introduction": "1. Introduction Since the 1980s, most coral reefs worldwide have been threatened and have undergone rapid degradation [ 1 , 2 ]. Many factors cause coral reef degradation, mainly including human destructive activities (e.g., overfishing and pollutant discharge), climate change (e.g., abnormally high temperatures), the emergence of coral diseases, and damage from coral predators [ 2 , 3 , 4 , 5 ]. Among these factors, damage from coral predators is a particularly important factor in coral reef degradation. It is well known that many aquatic organisms are coral predators, including crown-of-thorns starfish (COTS, Acanthaster planci ), sea urchins, Drupella spp., members of the family Labridae, etc. [ 6 , 7 , 8 ]. Among these predators, crown-of-thorns starfish (COTS) is a greatly harmful predator to coral reefs, and damages corals far more than other coral predators [ 9 ]. For example, four COTS outbreaks occurred in 1962, 1979, 1993, and 2010 in the Great Barrier Reef (GBR) [ 10 ], and damage from COTS has been extensive and frequent in many areas around the world [ 11 , 12 , 13 ]. The average live coral cover (LCC) of the GBR dropped from 28.0% in 1982 to 13.8% in 2012, and 47.5% of the live coral loss was directly caused by COTS outbreaks [ 10 ]. Similarly, COTS outbreaks have also occurred in Guam [ 13 ], southern Japan [ 11 ], and French Polynesia [ 11 ] in recent decades, with a large number of corals lost and rapid decreases in the LCC. For example, corals dominate a healthy reef with a LCC > 40% in the Moorean outer-reef, but algae colonize dead coral skeletons following severe predation by COTS (~10% LCC) [ 11 ]. In this case, mostly dead and weakened coral skeletons were swept away by a cyclone occurring at the end of the COTS outbreak and colonizing algae once again dominate the devastated reef (~5% LCC) [ 11 ]. COTS is a carnivorous animal whose main food is reef-building corals. It can feed on a large area of coral surface tissue in a short time. Each COTS can consume hundreds of square centimeters of coral tissue per day [ 14 ]. When feeding on corals, COTS wraps its extrudable stomach around the corals [ 15 ], after which gastric secretions digest the coral tissues within 3–5 h and large feeding scars are left on the corals [ 10 , 16 , 17 ]. For some encrusting corals (such as Montipora corals) and massive corals (such as Favia and Favites corals), COTS wraps the entire coral surface, thereby feeding on the entire coral and causing it to die [ 15 ]. In contrast, for some branching corals, COTS can only cover the outside coral branches, and the inner branches often survive, resulting in partial feeding scars [ 15 ]. At present, the direct impact of COTS on corals is generally assessed by performing macro-ecological surveys, that is, the area and numbers of feeding scars are counted through underwater video and photographs [ 12 , 18 , 19 ], but the effects on coral-associated microorganisms after COTS feeding remain unknown. Scleractinian corals rely on the symbiotic relationship between coral hosts and microorganisms, which enables corals to survive and thrive in the oligotrophic tropical ocean [ 20 , 21 ]. Coral holobionts contain a variety of microorganisms, including bacteria, archaea, fungi, viruses, and Symbiodiniaceae [ 22 ]. These microorganisms live in coral mucus, tissues, and skeleton, and cooperate with the host to support its physiological functions (including biogeochemical cycling, material transformation, antibacterial defenses, and nutrient acquisition), thereby helping to maintain the health of the coral reef ecosystem [ 21 , 23 , 24 ]. Among these microorganisms, Symbiodiniaceae play an important role in coral holobionts, providing ~95% of the energy of coral hosts [ 1 , 3 ]. In addition, the associated bacteria are highly diverse and complex microorganisms [ 22 , 25 ]. Some coral-associated bacteria, such as Endozoicomonas , are considered vital to coral health and play important roles in the responses of holobionts to environmental stresses [ 25 , 26 ]. The high abundance and diversification of coral-associated bacterial communities can help maintain the microbial flexibility of the coral holobionts to adapt to the changing environment [ 27 , 28 ]. In the case of coral-bacterial symbionts, some bacteria are beneficial to the coral host [ 22 , 29 , 30 ]. For example, the photosensitive-associated microorganisms Oceanospirillales and Halomonas, which promote dimethyl sulfoniopropionate metabolism and yielding products used in the metabolism of the host, may enable the coral holobionts to more efficiently carry out energy recycling [ 22 , 31 ]. Endozoicomonas is a good indicator candidate of the health state of corals; massive losses of these bacteria would have a serious negative impact on the physiological functions of the holobionts [ 32 , 33 ]. In contrast, certain bacteria are potential pathogens, such as some species in the Vibrio genus (e.g., Vibrio shiloi and Vibrio coralliilyticus ). An increased abundance of these bacteria may lead to poor health of corals [ 34 , 35 , 36 ]. Therefore, exploring changes in coral-associated bacterial communities can provide a basis for assessing the health state of corals. However, the effect of COTS ingestion on the coral-associated bacterial communities, physiological functions, and health states of coral holobionts are currently unknown. The Pocillopora genus is a widespread genus in the coral reef regions of the Indo-Pacific and commonly found in the tropical coral reefs of the South China Sea (SCS). Adult Pocillopora corals are usually >20 cm wide and these colonies are covered with different types of verrucae. Colony branches are thick and compact in habitats exposed to strong waves, whereas they can become open and thinner in protected habitats [ 37 ]. Colonies occur in most shallow water environments, ranging from exposed reef fronts to protected fringing reefs [ 37 ]. The Xisha Islands (also known as the Paracel Islands, 15°40′ N–17°10′ N, 110° E–113° E) are located in the central SCS. More than 20 tropical atolls and islands of various sizes are distributed in the SCS, and they have a high diversity of reef corals and numerous scleractinian coral species [ 38 , 39 ]. Previous macro-ecological surveys showed that Pocillopora is one of the most dominant coral genera of the Xisha Islands [ 38 ]. By conducting an in-situ investigation of the Passu Keah atoll of the Xisha Islands in 2019, we found that many COTS fed upon Pocillopora corals and that the outer-side branches were ingested, whereas the inner branches were in a healthy state. By comparing the associated microorganisms in the feeding scars and the healthy portions of the Pocillopora corals, it is possible to better understand the damage to coral holobionts caused by COTS feeding, as well as the healthy state. In this study, samples of Pocillopora corals with COTS feeding scars and healthy portions were collected from the outer reef slopes of Passu Keah, Xisha Islands. The coral tissue-weight loss rate, Symbiodiniaceae density (SD) associated bacterial compositions, microbial-metagenomic functional capacities of healthy corals, and coral feeding scars were analyzed to explore the damage and impact of COTS feeding on coral holobionts.",
"discussion": "4. Discussion COTS ingestion can dramatically affect the coral host and associated microbes [ 10 , 50 ]. COTS feeding can dramatically destroy coral holobionts within 3–5 h, resulting in the failure to recover and the death of numerous corals. Similar patterns of destruction have been observed with other coral predator animals, such as sea urchins, parrotfish, and members of the Drupella genus [ 6 , 8 , 51 ]. For example, Bessey et al. [ 8 ] investigated the ingestion of Drupella cornus on Acropora spicifera in Ningaloo Marine Park (Western Australia) and found that the aggregation feeding behavior of Drupella cornus directly led to the ingestion of entire coral tissues, resulting in large Symbiodiniaceae losses. Our study showed that Pocillopora feeding by COTS caused corals to lose most of their Symbiodiniaceae due to massive tissue loss. In this study, COTS ingested Pocillopora corals and caused a large loss of coral tissue, thereby changing the microbial diversity and community. Results of the bacterial community structure showed that they had changed significantly no matter the level of bacteria phyla or genus. The relative abundance of Pseudomonas , Thermus , Endozoicomonas , and Halomonas bacteria in healthy portions was significantly higher than that in feeding scars. These bacteria usually live in the coral mucus layer and endoderm layer [ 30 , 52 , 53 ]. In contrast, Cyanobacteria, Rhodobacteraceae, Leptolyngbys , Rhodospirillaceae, Muricauda , Ruegeria , and Vibrio bacteria usually live in the coral mucus, endodermal layer, and skeletal surface [ 53 , 54 ], and the relative abundance of these bacteria in feeding scars is significantly higher than that in healthy portions. Drastic changes in the coral-associated bacterial community may affect the physiological function of the coral holobiont [ 22 ]. Many coral-associated bacteria, such as Endozoicomonas and Roseobacteriales, are considered beneficial to the coral host [ 25 , 26 ]. The potential role of these beneficial bacteria in biology are to transfer, spread, and regulate energy and promote protein supply and carbohydrate circulation to coral hosts [ 30 , 55 ]. Roseobacteriales involved in the sulfur cycle are generally considered to be obligate partners in symbiotic Symbiodiniaceae, which can increase the growth rate of Symbiodiniaceae [ 22 , 29 ]. In our study, COTS ingestion resulted in considerable tissue loss from Pocillopora corals, with a significant decrease in associated bacteria, including Pseudomonas , Thermus , Endozoicomonas , and Halomonas . In turn, some low-abundance bacteria were increased, which eventually resulted in significant changes in the coral-associated bacterial community. Based on other corallivory, it is suggested that the ingestion of many corallivory has a great effect on a coral’s microbial community composition and diversity [ 50 , 51 , 56 ]. For example, Maher et al. [ 50 ] found that the feeding scars alone greatly affected changing microbial community composition and diversity of Pocillopora meandrina fed by corallivory of parrotfishes and puffer fishes. The relative abundance of Desulfovibrionaceae with feeding scars causes the significant change in this microbe from control corals with 0.09% ± 0.09% to 9.92% ± 6.30% ( p < 0.01), while scarred treatments produced a high mean community diversity (0.67 ± 0.11), six times greater than that of the controls [ 50 ]. Moreover, Nicolet et al. [ 56 ] found that corallivorous invertebrate Drupella spp. create deep feeding scars that change microbial community composition and diversity of Acropora muricata and increase pathogens transmission. They found that Drupella spp. transmitted brown band disease to healthy corals in 40% of cases immediately following feeding on infected corals, and even in 12% of cases 12 and 24 h following feeding [ 56 ]. In our study results, it is similar to Drupella spp. that the bacterial community diversity of scars fed by COTS were higher than healthy portions, and feeding by COTS indeed greatly changed the composition and diversity of coral bacterial holobionts. Microbial functional genic predictions suggested that the health state differed between feeding scars and healthy portions in Pocillopora corals. Using PICRUSt, which is highly effective in the functional predictions of metabolic pathways, genetic information processing, environmental information processing, and cellular processes of microbial communities [ 48 ], seems to be essential for the analysis of differences in functional predictions related to digestion in coral-associated bacterial communities between feeding scars and healthy portions in Pocillopora corals. The coral microbiome is enriched in several protein functions that support carbohydrate metabolism in coral holobionts, which reflect different environmental conditions of corals [ 27 ]. Based on the changes in the microbiome metabolic levels, significant differences were found in the functional characteristics of the coral-bacterial communities between feeding scars and healthy portions in Pocillopora corals due to a loss of the tissue, including amino acid metabolism, exogenous biodegradation and metabolism, lipid metabolism, membrane transport, signaling conduction, and cell motility. In this study, numerous feeding scars of corals and significant changes in bacterial communities were observed after ingestion by COTS, which may affect the health state and survival of Pocillopora corals. For example, the COTS outbreak in Guam from 1967–1969 lasted for 30 months, resulting in the death of ~90% of all corals in the shallow water area of the 38 km coastline [ 13 ]. A COTS outbreak in the French Polynesia Islands and directional diffusion towards Moorea Island caused a 96% loss in the coral population [ 12 ]. After a COTS outbreak, numerous feeding scars provide areas for the growth of filamentous algae and pathogens such as Vibrio coralliilyticus , which may cause coral white plague disease [ 12 , 35 ]. Our research suggests that after Pocillopora corals were ingested by COTS, the relative abundances of several bacterial pathogens significantly increased in the feeding scars, which may have caused the healthy portions of the corals to become vulnerable to invasion by filamentous algae."
} | 3,787 |
36539439 | PMC9768204 | pmc | 635 | {
"abstract": "The impact of biomineralization and redox processes on the formation and growth of ferromanganese deposits in the World Ocean remains understudied. This problem is particularly relevant for the Arctic marine environment where sharp seasonal variations of temperature, redox conditions, and organic matter inflow significantly impact the biogenic and abiotic pathways of ferromanganese deposits formation. The microbial communities of the fast-growing Arctic Fe–Mn deposits have not been reported so far. Here, we describe the microbial diversity, structure and chemical composition of nodules, crust and their underlying sediments collected from three different sites of the Kara Sea. Scanning electron microscopy revealed a high abundance of microfossils and biofilm-like structures within the nodules. Phylogenetic profiling together with redundancy and correlation analyses revealed a positive selection for putative metal-reducers ( Thermodesulfobacteriota ), iron oxidizers ( Hyphomicrobiaceae and Scalinduaceae ), and Fe-scavenging Nitrosopumilaceae or Magnetospiraceae in the microenvironments of the Fe–Mn deposits from their surrounding benthic microbial populations. We hypothesize that in the Kara Sea, the nodules provide unique redox-stable microniches for cosmopolitan benthic marine metal-cycling microorganisms in an unsteady environment, thus focusing the overall geochemical activity of nodule-associated microbial communities and accelerating processes of ferromanganese deposits formation to uniquely high rates.",
"introduction": "Introduction In recent years, deep-sea ferromanganese (Fe–Mn) nodules and crusts have attracted considerable attention as a potential source of iron and manganese, as well as other technology metals and rare earth elements (REE) 1 – 3 . Wide and dense ore fields occur mainly in the regions with very low sedimentation rates (pelagic areas and submarine outcrops of the Pacific and Eastern Indian Oceans). Polymetallic deposits slowly grow there at a rate of several mm to several cm per million years, depending on their origin 3 . Several suggestions have been made on the mode of origin of ferromanganese nodules, depending on their metal sources. Those included hydrogenetic precipitation of metals from cold ambient water, oxic/suboxic diagenesis from sediment-pore fluids, or precipitation of Fe–Mn oxyhydroxides directly from hydrothermal solutions 4 – 8 . However, the current view on the genesis of Fe–Mn deposits leaves out the processes leading to rapid formation of the Fe–Mn nodules and crusts in shallow-water regions. These ‘shallow-water’ deposits are mainly located at depths of down to 300 m and grow about 10 3 times faster than their deep-sea counterparts 9 . Along with the generally accepted view on the genesis of the Fe–Mn nodules and crusts as a combination of different geochemical processes, an increasing amount of evidence is now accumulating in favor of microbial activity involvement in ferromanganese deposits formation 10 – 20 . As the major players of diagenesis, microorganisms can influence iron and manganese mineralization. There are prokaryotes able to oxidize Fe 2+ and Mn 2+ for using it as a source of energy, and microorganisms capable of reducing Fe 3+ and Mn 4+ as terminal electron acceptors in electron transfer chains 21 , 22 . The three basic processes of biological control (structural, spatial and chemical) over mineralization 23 were thoroughly described back in the 1980s. However, the information about the exact microbial groups that drive biomineralization has only become available with the development of molecular phylogenetics and ecology. Recent studies based on 16S rRNA gene profiling of microbial communities from the deep-sea (various sites of Pacific Ocean) and shallow-water (Baltic Sea) Fe–Mn nodules, as well as from the Atlantic deep-sea Fe–Mn crusts, provided the first insights into the structure and possible functions of microbial populations inhabiting the ferromanganese deposits of the World Ocean. The communities of the nodules revealed a lower phylogenetic diversity compared to that reported for underlying sediments. At the same time, the reported nodules communities were enriched with one or several taxa represented in both microenvironments 11 , 13 , 14 , 16 , 24 – 28 . Notably, the microbial communities of any Fe–Mn deposits from the Arctic Ocean have not been characterized so far. As a shallow marginal shelf-sea of the Arctic Ocean, Kara Sea is considered an important reference point for the understanding of global biogeochemical cycles. In this work, we describe the phylogenetic composition of microbial communities associated with the Arctic ferromanganese nodules and crusts collected from geographically distant sites of the Kara Sea (Fig. 1 ), characterize their chemical parameters, and discuss the geochemical and ecological aspects of presumed microbe-to-mineral interactions in the view of their involvement in iron and manganese mineralization within the contrasting polar environment. Figure 1 ( a ) The map of sampling sites location and circulation of surface currents ( from Stein 29 ). ( b ) General view of the studied Fe–Mn deposits. ( c ) Fe–Mn nodules on site AMK76-6259 covered by a fluffy layer (sea-floor photo by Sonar ocean bottom surveying Lab, IO RAS).",
"discussion": "Discussion Kara Sea nodules and crust genesis The main factors affecting the Fe–Mn deposits formation in the Kara Sea are relatively high sedimentation rate (2–45 cm/Kyr), enhanced amount of TOC in sediments (av. 1.18%), high input of terrestrial matter, and ice covering around nine months a year 29 , 30 . These environmental characteristics are reflected in the morphology and geochemistry of shallow-water nodules and crusts 31 . Furthermore, the unstable organic matter (OM) supply, as well as bioturbation and physical mixing of the sediments by bottom current activity, contribute to the alternation of redox conditions in the sediments and porewater 32 . These variations can affect the fate of chemical elements involved in the formation of the Fe–Mn deposits. The sediments of the Kara Sea shelf are represented by reduced grayish muds covered by a thin (a few cm thick) oxidized layer. Oxygenation of these sediments can lead to the precipitation of many metals including Fe and Mn, while O 2 depletion in this sedimentary environment can mobilize metals 33 , 34 . Our data revealed similar content of major elements in the nodules and crust collected at sites AMK76-6259, AMK76-6236, and PSh128.35 (Table 1 ). The elemental composition of the nodules collected at these sites appeared to be comparable with that previously described for the nodules from the central and eastern parts of the Kara Sea 35 – 37 thus indicating the common chemical characteristics of Fe–Mn deposits from this shallow-water Arctic region. Manganese content of the studied samples is lower compared to that of deep-sea nodules (Clarion-Clipperton Zone, Peru Basin), while Fe content is generally higher than that reported for abyssal deposits 38 . All our samples of the deposits showed low Co + Ni + Cu content and highly variable Mn/Fe ratio. To classify Fe–Mn deposits, a triangle diagram by Bonatti 39 based on their chemical composition (Fe—Mn—(Cu + Ni + Co) × 10, Fe—Mn–Co × 100) is generally used. However, in case of the shallow-water fast-growing deposits this diagram does not work clearly. To date, the genesis of the Fe–Mn deposits has been characterized using various ratios of REY concentrations according to Bau 40 . Negative Ce anomaly, YSN/HoSN ratio close to 1, intermediate Nd concentrations of approximately 25 ppm suggest a strong impact of diagenetic processes on the formation of the studied Fe–Mn nodules and crust. The diagenetic origin of deposits in the other parts of the Kara Sea also was previously reported 35 – 37 , 41 . The observed depletion of major and trace elements (including REE) in the studied Kara Sea nodules and crust can be explained by their extremely high growth rate (0.4–8 mm/kyr) which results in dilution of authigenic mineral phases by detrital material due to short time of their connection with sediments and pore water 35 , . The intercalation observed between the Fe- and Mn-rich layers can occur due to differences in the normal redox potential of these elements and reflect the abrupt changes in their precipitation conditions. Redox oscillations in sediments of the Kara Sea occur periodically, or episodically, and vary between sampling sites 33 , 34 . Under temporal depletion of oxygen in the sediments, Mn and Fe can be remobilized and diffuse upwards. Mn-rich layers are usually formed in relatively steady bottom conditions (like those, prevailing at sampling site AMK76-6259 located on the inner shelf region with negligible freshwater inflow). The reduced sediments in this case are covered by a thin fluffy layer that acts as the upper oxidized one and prevents Mn outflow to seawater. Washing out of the fluffy layer due to dynamic bottom water circulation (that might take place at sampling site PSh128.35 on the outer shelf) could cause Mn transfer to the water (with its further carrying out by bottom currents) and initiate the formation of Fe-rich phases. This scenario of Fe-rich deposit formation is likely realized at site AMK76-6236 at the flank of the Saint Anna Trough, where the crusts lie on top of reduced sediments with low Mn content. The sea-floor in the trough is swept by intense bottom currents 42 , 43 . The early diagenesis of nodules and crust can also be driven by the oxidation (remineralization) of OM. The input of OM to the Kara Sea bottom sediments, and as a consequence, to nodules and crust is strongly influenced by the river discharge, coastal erosion, sea-ice cover, and bottom circulation 44 . We observe the predominance of terrigenous OM in Mn-rich nodule, which presence is revealed by the unimodal distribution pattern of n-alkanes dominated by the C 27 , C 29 , and C 31 odd homologues (Table S1 ). However, presence of bimodal n-alkane distribution in Fe-rich nodule indicates labile autochthonous OM (in moderate amount) along with terrigenous components. Accumulation of n-alkanes with different group composition of marine, bacterial-derived and terrigenous origin in the studied nodules and crust suggests diagenetic alteration of OM within the ore deposits. Odd–even ratio of n-alkanes (OEP 17-19 < 1) suggests the OM undergoes an intense microbial degradation which could be coupled to biotic redox transformations of Fe and Mn minerals. SEM images of the Kara Sea nodules showed a high abundance of different microfossils, similar to bacterial cocci and rods (Fig. 3 ) along with biofilm-like structures. These microfossils were identified in the interior of the Fe-rich and Mn-rich nodules. The same structures were previously observed in Fe–Mn nodules of the Pacific and Indian Ocean 11 , 20 , 45 – 47 . Well-crystallized Fe- and Mn-oxides around the microfossils indicate a post-accretional diagenetic process within the nodules. The same features were also observed in thin sections (Fig. 3 i), where microbial cell-like structures are partially or completely filled with Fe- and Mn-containing minerals. Uniform thickness of ‘microbial cell coatings’ (0.2 µm) of the nodules indicates that fossilization of all the cells in the colony began simultaneously and was possibly caused by the exhaustion of organic carbon sources or hindered access to mineral electron acceptors (Fe and Mn-oxyhydroxides) due to the formation of surface thin films of their reduced forms upon microbial growth and dissimilatory metal reduction. Microbial communities of shallow-water Fe–Mn nodules and crust An overview of relative abundance of prokaryotic taxa in microbial communities of the Kara Sea Fe–Mn deposits and their underlying sediments reveals a sufficient share of phylotypes common for all the analyzed econiches (Figs. 4 , 5 ). Dominant taxa, in particular, Pirellulaceae , and alphaproteobacterial families Kiloniellaceae , Hyphomicrobiaceae, and Methyloligellaceae , have equally high representation in each of the analyzed samples (Fig. 4 ). This finding highlights general similarity of the microbial communities from the Kara Sea Fe–Mn deposits with previously reported communities associated with deep-sea or shallow-water Fe–Mn nodules. The majority of these diverse communities are dominated by Alpha- (e.g., Rhizobiales , Kilonellales and Ricketsiales orders) and Gammaproteobacteria (e.g., Vibrionales and Xanthomonadales orders), as well as by ammonia-oxidizing thaumarchaea of the Nitrosopumilaceae family and uncultured Marine Group I 11 , 13 , 14 , 16 , 24 , 25 . All these taxa include metabolically versatile organisms, which are capable of chemoorganotrophy or anoxygenic phototrophy 48 . The distribution of unique OTUs between the communities of different Fe–Mn deposits from the Kara Sea indicates a divergence of these microbial populations from each other (Fig. 5 ). Each of the three microbial communities, associated with the Fe–Mn deposits, has its distinctive point on the RDA plot, according to the Fe or Mn content of its microenvironment (Fig. 6 ). The resulting picture indicates that the Fe and Mn abundance exerts selective pressure on the composition of the microbial populations inhabiting the nodules. Previously, a positive selection for Gammaproteobacteria of Shewanella and Colwellia genera was reported for deep-sea nodules from the Clarion-Clipperton Zone and South Pacific Gyre 13 , 14 . Shewanella is the model genus of metal-reducing bacteria 49 , and the Colwellia genus harbors Mn 4+ -reducers isolated from three different habitats 50 . In the Kara Sea nodules, the microbial taxa, for which metal-reducing or -oxidizing activities were shown earlier, are represented by the Geopsychrobacteraceae family whose members are enriched in the nodules and crust from the sites AMK76-6236 and PSh128.35 (Fig. 4 ). These organisms belong to a recently proposed Thermodesulfobacteriota phylum 51 . Interestingly, a positive selection for another taxon of typical metal reducers ( Desulfuromonadaceae family) was only observed in the Fe-rich nodule sample (Fig. 4 ). A more pronounced selection for Fe 3+ reducers in the nodules might reflect a higher content and an amorphous state of Fe 3+ minerals in this type deposits 35 , 37 , which makes them more readily accessible electron acceptors for iron reducing prokaryotes 52 . In addition to increased Fe availability, the Mn/Fe ratio of the Kara Sea nodules appeared to notably impact the abundance of metal-cycling organisms. The communities of the Mn-rich nodule sample (site AMK76-6259) have increased shares of Magnetospiraceae and planctomycetes of ‘ Scalinduaceae ’ family (Fig. 4 ) which were poorly represented in the rest of the analyzed samples (Fig. 7 ). None of these taxa comprise organisms utilizing Mn compounds as electron acceptors or donors for growth. Instead, magnetospirilla are known for their unique capability to produce fine-grained intracellular magnetite crystals, while ‘ Scalinduaceae ’ are capable of Fe 2+ oxidation 53 . Visualization of microcolonies-like structures inside the studied nodules (Fig. 3 i) indicates that these deposits were formed under the influence of microbial biofilm growth and concomitant processes of exopolysaccharides production and acidification. Indeed, sponge symbionts of the Entotheonellaceae family, well-adapted to colonize the extended surfaces, were significantly selected for in Fe-rich nodules (Figs. 4 , 7 ). Besides, the microbial communities of Fe-rich nodules and crust are enriched with actinobacteria of the Microtrichaceae family (Fig. 4 ) harboring filamentous heterotrophic species 54 . Within complex biofilms, fermentative bacteria of the Hyphomicrobiaceae family, abundant in the Kara Sea nodule communities (Fig. 4 ), could induce acidification by the production of fatty acids and thus, influence the mobility of Fe and Mn species and their susceptibility to further microbial redox transformations. Ratio Pr/Ph < 1 suggests reduced, low oxygen conditions in the samples from sites AMK76-6236 and AMK76-6259. Such microenvironments favor the microbial reduction of Fe 3+ and Mn 4+ minerals followed by their mobilization from the nodule matrix (Fig. 2 ). Nodules' organic matter can be degraded within the nodule microbial communities by bacteroidetes of the Flavobacteriaceae and Cyclobacteriaceae families, providing the substrates for predominating fermentative bacteria of the Woeseiaceae and Hyphomicrobiaceae families (Fig. 4 ), and for anaerobic organotrophs of the Geopsychrobacteraceae and Desulfuromonadaceae families capable of dissimilatory metal reduction. In addition to direct involvement in redox cycling of Fe and Mn compounds, microbial populations of ferromanganese deposits could drive the biogeochemical cycle of nitrogen including the processes coupling the transformation of nitrogen and iron compounds. Indeed, the representatives of the Kiloniellaceae family prevailing in all the Kara Sea deposits (Fig. 4 ) were reported to couple nitrate reduction to organic matter oxidation. Representatives of archaeal Nitrosopumilaceae and bacterial Nitrospiraceae families, ubiquitous in almost all the sampled deposits (Fig. 4 ), can be involved in nitrification 17 , 55 . In addition, Nitrospiraceae representatives, which are mostly abundant in Fe-rich deposits (from sites PSh128.35 and AMK76-6236; Figs. 4 , 7 ) can oxidize ammonia via comammox. In the Mn-rich nodule from site AMK76-6259, ammonium can be oxidized by planctomycetes of the ‘Scalinduaceae’ family 56 , 57 , which are enriched in this microenvironment (Fig. 4 ). Ammonium concentration in the Kara Sea is generally low (micromolar range) 58 , 59 , and the selective pressure for ammonium oxidizers in this environment needs further evaluation. Previously, the enrichment of ammonia-oxidizing archaea of the Nitrosopumilaceae family was reported for several deep-sea nodules 11 , 13 , 14 , 16 , 27 . This taxon is globally abundant in marine environments and comprises autotrophs 60 , many of which depend on Fe availability for their growth 61 , 62 . Such physiological features could enhance the preferred colonization of growing ferromanganese nodules with these Fe-scavenging prokaryotes. Thus, the growth and metabolic activity of ammonia oxidizers could impact the Mn/Fe ratio during diagenetic processes occurring within the nodules. In its turn, ammonium oxidation in the oxygenated surface layer of nodules could be fueled by ammonium release from organic matter degradation in the water column or underlying sediments 63 , 64 or within the biofilms of fermentative bacteria inside the Fe–Mn deposits. In addition, ammonium can be formed via dissimilatory nitrate reduction to ammonium (DNRA). This process was shown for Nitrincolaceae , enriched in the Mn-rich nodule of site AMK76-6259 (Figs. 4 , 7 ). In the Fe-rich nodule, nitrate reduction can be also coupled to the oxidation of Fe 2+ (NRFeOx). This process is proposed to be common for mixotrophic denitrifiers 65 , 66 , including those of the Hyphomicrobiaceae family with rather high abundance in Fe-rich sites PSh128.35 and AMK76-6236 (Fig. 4 ). Microbial communities of sediments underlying ferromanganese deposits The microbial communities identified in shallow-water Kara Sea sediments appeared to be depleted with the taxa potentially involved in nitrogen cycling, when compared to the communities of Fe–Mn deposits from the same sites. This effect is clearly evidenced by decreased abundance of potential ammonia-oxidizers (harbored by the Nitrosomonadaceae and Nitrosococcaceae families) in the sampled sediments (Fig. 4 ). The decreased representation of ammonia oxidizers could result from the peculiarities of OM inflow regime, observed at the sampling sites. Unstable inflow of OM, temporal variation of redox gradient and local formation of anoxic niches in the sediments exert selective pressure on sedimentary microbial communities towards the prevalence of anaerobically respiring organisms. Comparatively low abundance of nitrogen cycling microorganisms was previously reported for Baltic Sea sediments, although the diversity of these organisms appeared to be rather high and included the nitrifying archaea of Nitrosopumilales order, various nitrifying, denitrifying and DNRA performing bacteria 67 . Obviously, high anthropogenic activity and local hydrodynamics significantly contribute to nitrogen input and the intensity of its biological transformation in the Baltic Sea region 67 , that is not the case for the Kara Sea, subjected to incomparably lower input of nitrogen compounds. All the benthic microbial communities in our study are dominated by the taxa capable of sulfate respiration ( Desulfosarcinaceae or Desulfocapsaceae families) (Fig. 4 ). The abundance of these taxa positively correlates with the abundance of sulfur in the environment (Fig. 7 ). RDA analysis clearly separated the community of sulfur-enriched sediments from the others (Fig. 6 ), and the 16S rRNA gene sequence abundance supported positive selection for sulfate reducers in this microenvironment (Fig. 4 ). High relative abundance of the orders Desulfobacterales, Desulfovibrionales, Desulfuromonadales , which include a lot of sulfate reducing bacteria, was reported for the sediments of the Yamal Sector of the Kara Sea 68 . Also, Desulfobulbaceae, Desulfuromonadaceae , and Pelobacteraceae were found to be enriched in the sediments of the North Sea and Baltic Sea near the Fe(III)/Fe(II) redox boundary, while higher abundance of Desulfococcus and Desulfobacterium was detected at greater depths of these sediments with decreased Fe-reducing activity 69 . All these taxa harbor a wide variety of both Fe reducing, and sulfate reducing bacteria. In our study, we did not detect the enrichment of any typical metal reducing organisms in benthic microbial communities, except for the sediments underlying Mn-rich nodules, where phylotypes belonging to organotrophic Mn 4+ -reducing Colwelliaceae were enriched. In addition, anaerobically respiring organotrophs of the Halieaceae family and fermentative organotrophs belonging to Anaerolineaceae are enriched in almost all the benthic communities (Fig. 4 ). The increased abundance of Rhodobacteraceae -related phylotypes in sedimentary microbial communities could play a crucial role in sulfur and carbon biogeochemical cycling under fluctuating redox conditions of the Kara Sea sediments, as this family comprises aerobic phototrophs, chemoheterotrophs, and anoxygenic phototrophs which utilize hydrogen sulfide as the electron donor 48 . The benthic microbial community at site AMK76-6236 is highly enriched with Sulfurovaceae -related phylotypes (Fig. 4 ). Relatively low but considerable (ca. 1%) abundance of Sulfurovaceae was previously reported for the sediments of the Yamal Sector of the Kara Sea 68 . Members of this family are chemolithoautotrophs coupling sulfur oxidation with nitrate reduction or oxygen respiration, which makes them crucial primary biomass producers in the community at both oxic and anoxic conditions. The increased representation of aerobic marine oligotrophs of the Rubritaleaceae family in the same benthic community of site AMK76-6236 correlates with its higher oxygenation compared to the other two sampling sites. The oxygenation of the bottom water in this deep part of the St. Anna Trough is driven by the cooling of oxygen-rich surface waters near the Novaya Zemlya archipelago and their further downwelling across the slope 58 . During the temporal decrease of oxygenation at this site, reduced sulfur species can be produced by sulfate-reducing bacteria of the Desulfobulbaceae , Desulfosarcinaceae and Desulfocapsaceae families, some of which were reported to establish single-species filamentous conductive structures (‘cables’), which couple the oxidation of sulfide in deeper sediment layers to the reduction of oxygen or nitrate near the sediment–water interface 70 . The enhanced abundance of microorganisms capable of respiring a variety of electron acceptors in the Kara Sea sediments correlates with unstable redox conditions of this environment. The sharp changes of the redox state due to fluctuating input of organic matter and oxygenated waters seem to be the main selective force making the Kara Sea sediments the reservoir of metabolically versatile cosmopolitan marine prokaryotes involved in the biogeochemical cycles of carbon and sulfur, as well as in redox cycling of metals. The cumulative geochemical activity of such a diverse microbial community is expected to follow the changes in environmental conditions, which periodically favor the predominance of different metabolic groups of microorganisms. We have shown that comparatively stable redox and geochemical settings within the semi-closed ferromanganese deposits directly select for the organisms which reduce, oxidize or intracellularly accumulate metals. The overall metabolic activity of these microorganisms, sheltered in nodules and crust from unfavorable fluctuations of environmental factors, is expected to dramatically accelerate the formation of the Fe–Mn deposits different in their morphology, internal structure and elemental composition."
} | 6,381 |
33233378 | PMC7700301 | pmc | 636 | {
"abstract": "Microorganisms are effective platforms for the production of a variety of chemicals including biofuels, commodity chemicals, polymers and other natural products. However, deep cellular understanding is required for improvement of current biofuel cell factories to truly transform the Bioeconomy. Modifications in microbial metabolic pathways and increased resistance to various types of stress caused by the production of these chemicals are crucial in the generation of robust and efficient production hosts. Recent advances in systems and synthetic biology provide new tools for metabolic engineering to design strategies and construct optimal biocatalysts for the sustainable production of desired chemicals, especially in the case of ethanol and fatty acid production. Yeast is an efficient producer of bioethanol and most of the available synthetic biology tools have been developed for the industrial yeast Saccharomyces cerevisiae . Non-conventional yeast systems have several advantageous characteristics that are not easily engineered such as ethanol tolerance, low pH tolerance, thermotolerance, inhibitor tolerance, genetic diversity and so forth. Currently, synthetic biology is still in its initial steps for studies in non-conventional yeasts such as Yarrowia lipolytica , Kluyveromyces marxianus , Issatchenkia orientalis and Pichia pastoris . Therefore, the development and application of advanced synthetic engineering tools must also focus on these underexploited, non-conventional yeast species. Herein, we review the basic synthetic biology tools that can be applied to the standard S. cerevisiae model strain, as well as those that have been developed for non-conventional yeasts. In addition, we will discuss the recent advances employed to develop non-conventional yeast strains that are efficient for the production of a variety of chemicals through the use of metabolic engineering and synthetic biology.",
"conclusion": "4. Conclusions and Future Directions Synthetic biology has made it possible to control several cellular processes in order to obtain biofuels, the demand for which has increased in recent years. However, the metabolic engineering of cell factories still faces many limitations. The libraries of promoters and terminators, for example, still need to achieve greater amounts of sequence and functional diversity. Computational tools have driven the development of new hosts, facilitating the experimental design of new metabolic pathways and the understanding of genomic data. New strategies and tools are continually progressing for engineering non-microbes, so that they can be optimized for real industrial applications. Such natural microbial genetic obstacles have been overcome with tools such as CRISPR/Cas systems and DNA assembly, improved computer models, automation and design by computer systems that quickly build and characterize the cellular genome. The vast availability of generated biological data has more rapidly developed data science and artificial intelligence strategies to suggest better strategies for genetic and metabolic engineering of cellular systems [ 158 , 159 ]. The simplicity in terms of regulation and the availability of excellent databases of the S. cerevisiae genome, in comparison to other yeasts, make this yeast the model organism even more used in research for biofuels. However, the development of new techniques of systems biology and synthetic biology to generate high cellular performance, has also made possible the genetic engineering of non-conventional yeasts, whose phenotypes of cellular resistance are of extreme interest in the fuel industry.",
"introduction": "1. Introduction The production of biofuels and chemical bioproducts utilizing live cells and thus their enzymatic conversion pathways, requires less energy than other conversion processes. Therefore, microbial factories are increasingly being developed through the engineering of metabolic pathways and the redirection of carbon towards desired products [ 1 ]. The main challenge for the commercialization of biofuels and other chemicals is the gap in the path between the laboratory and the commercial market, mainly due to the fact that often the engineered strains do not fit commercialization needs [ 2 ]. Optimizing native gene expression and expression of non-native genes and pathways enables the engineering of microbes with new catalytic activities or improvements of native phenotypes such as tolerance to stress conditions [ 2 , 3 ]. Currently, metabolic engineering allows for a systematic workflow for designing more robust and efficient cell factories. Such a process is characterized mainly by manipulating the expression of pathway genes, balancing cofactors, eliminating by-products and increasing the supply of precursors, among other possible approaches [ 4 ]. Metabolic engineering integrates biological information underlying graphical and mathematical representations of metabolic flows, mainly in order to search for cellular functions that can improve the production of a desired compound or the development of a desired characteristic [ 5 ]. Recently, advanced sequencing technologies have enabled rapid analysis of genomic variations that lead to desired phenotypes [ 6 ]. In addition the development of new genome editing tools, such as CRISPR/Cas systems, have facilitated and increased the speed of construction of microbial cell factories [ 4 ]. Such efforts have paved the way for the engineering of higher efficiency production of biologically derived chemicals. With more than a thousand unique species already identified, yeasts are one of the most widely studied microorganisms. For thousands of years, these species have been associated with the production of fermented drinks [ 7 ]. The budding yeast S. cerevisiae has long been the favorite organism used as a cell factory in the commercial production of biologically-based chemicals, mainly due to its ancient use in classical industrial applications (wine and beer production) and our vast physiological and genetic knowledge of this organism [ 8 ]. A scheme representing different substrate assimilation routes is illustrated in the Figure 1 . The efficient conversion of carbon substrates such as glucose, different types of hexoses and pentoses is processed through hydrolytic enzymes and transporters in yeast. These metabolic pathways lead to the production of the most important products in the biofuel industry, such as ethanol, butanol, alkanes and fatty acids. Its success as a model organism is mainly due to the high degree of conservation of important cellular processes (autophagy, protein translocation and secretion, cell division, heat shock, protein folding and chaperone functions) and the availability of genetic engineering tools. In addition to its classic application in the production of wine and beer, its use has also extended to the production of bioethanol, aimed at the production of transport fuels. In 2017, the United States produced more than 60 billion liters, followed by Brazil with production of about 30 billion liters, with the rest of the world producing about 15 billion liters [ 5 ]. Despite the ease of engineering, S. cerevisiae has disadvantages, such as the inability to consume more economical and sustainable substrates such as xylose, arabinose and glycerol. Additionally, S. cerevisiae is not able to tolerate high temperatures (>34 °C) and low external pH (<3) [ 9 ], so its application in the production of second-generation biofuels has been limited. Non-conventional yeasts such as Yarrowia lipolytica, Kluyveromyces marxianus, Issatchenkia orientalis and Pichia pastoris have many physiological, metabolic and regulatory advantages that could overcome these limitations [ 3 , 10 , 11 ]. While most reviews in the literature addressing the engineering and characteristics of yeast for biofuel production focus on the model industrial yeast S. cerevisiae [ 12 , 13 , 14 , 15 , 16 ], there are fewer that have highlighted updated studies of non-conventional yeasts [ 3 , 17 , 18 , 19 ]. Here, we focus both on the advanced genome editing tools currently available for S. cerevisiae and their adaptation for use in non-model yeast systems to further improve the range and scale of industries for the biological production of chemicals using yeast platforms."
} | 2,085 |
22015987 | null | s2 | 637 | {
"abstract": "Synthetic Escherichia coli consortia engineered for syntrophy demonstrated enhanced biomass productivity relative to monocultures. Binary consortia were designed to mimic a ubiquitous, naturally occurring ecological template of primary productivity supported by secondary consumption. The synthetic consortia replicated this evolution-proven strategy by combining a glucose positive E. coli strain, which served as the system's primary producer, with a glucose negative E. coli strain which consumed metabolic byproducts from the primary producer. The engineered consortia utilized strategic division of labor to simultaneously optimize multiple tasks enhancing overall culture performance. Consortial interactions resulted in the emergent property of enhanced system biomass productivity which was demonstrated with three distinct culturing systems: batch, chemostat and biofilm growth. Glucose-based biomass productivity increased by ∼15, 20 and 50% compared to appropriate monoculture controls for these three culturing systems, respectively. Interestingly, the consortial interactions also produced biofilms with predictable, self-assembling, laminated microstructures. This study establishes a metabolic engineering paradigm which can be easily adapted to existing E. coli based bioprocesses to improve productivity based on a robust ecological theme."
} | 339 |
29468803 | PMC5947290 | pmc | 638 | {
"abstract": "Summary The sulfate‐dependent, anaerobic oxidation of methane (AOM) is an important sink for methane in marine environments. It is carried out between anaerobic methanotrophic archaea (ANME) and sulfate‐reducing bacteria (SRB) living in syntrophic partnership. In this study, we compared the genomes, gene expression patterns and ultrastructures of three phylogenetically different microbial consortia found in hydrocarbon‐rich environments under different temperature regimes: ANME‐1a/HotSeep‐1 (60°C), ANME‐1a/Seep‐SRB2 (37°C) and ANME‐2c/Seep‐SRB2 (20°C). All three ANME encode a reverse methanogenesis pathway: ANME‐2c encodes all enzymes, while ANME‐1a lacks the gene for N5,N10‐methylene tetrahydromethanopterin reductase ( mer ) and encodes a methylenetetrahydrofolate reductase (Met). The bacterial partners contain the genes encoding the canonical dissimilatory sulfate reduction pathway. During AOM, all three consortia types highly expressed genes encoding for the formation of flagella or type IV pili and/or c‐type cytochromes, some predicted to be extracellular. ANME‐2c expressed potentially extracellular cytochromes with up to 32 hemes, whereas ANME‐1a and SRB expressed less complex cytochromes (≤ 8 and ≤ 12 heme respectively). The intercellular space of all consortia showed nanowire‐like structures and heme‐rich areas. These features are proposed to enable interspecies electron exchange, hence suggesting that direct electron transfer is a common mechanism to sulfate‐dependent AOM, and that both partners synthesize molecules to enable it.",
"introduction": "Introduction In the ocean floor, the sulfate‐dependent, anaerobic oxidation of methane (AOM) consumes a substantial fraction of the potent greenhouse gas methane before it can reach the hydrosphere and atmosphere (Boetius and Wenzhöfer, 2013 ). AOM is mediated by microbial consortia of anaerobic methanotrophic archaea (ANME) and sulfate‐reducing bacteria, which couple the oxidation of methane to the reduction of sulfate via a syntrophic process (Boetius et al ., 2000 ; Orphan et al ., 2001b ). The ANME are relatives of methanogenic archaea and use the methanogenesis pathway in reverse to oxidize methane to CO 2 (Krüger et al ., 2003 ; Hallam et al ., 2004 ; Meyerdierks et al ., 2005 , 2010 ; Wang et al ., 2014 ). They are classified into at least five distinct branches of the Methanomicrobia (Knittel and Boetius, 2009 , 2010 ; Haroon et al ., 2013 ), none of which contains methanogens or cultured representatives of methanotrophs. Most types of marine ANME‐1 and ‐2 archaea form consortia with sulfate‐reducing bacteria of the Seep‐SRB1a or Seep‐SRB2 cluster ( Desulfosarcina / Desulfococcus ; DSS; Orphan et al ., 2001a ; Michaelis et al ., 2002 ; Knittel et al ., 2003 ; Kleindienst et al ., 2012 ; Milucka et al ., 2013 ). However, the thermophilic ANME‐1 archaea associate with sulfate‐reducing members of the deep‐branching HotSeep‐1 cluster (Holler et al ., 2011 ; Wegener et al ., 2015 ), and the psychrophilic archaea of the ANME‐3 branch form consortia with bacteria of the Desulfobulbus cluster (Niemann et al ., 2006 ). Recent studies show that some linages of ANME‐2d (including Candidatus Methanoperedens nitroreducens) can couple methane oxidation to the reduction of nitrate and associate with partner bacteria that consume their toxic reaction product nitrite (Haroon et al ., 2013 ). In ANME‐2/DSS consortia, Milucka and colleagues ( 2012 ) proposed that an internal cycling of zero‐valent sulfur was accomplished by a partial sulfate reduction in ANME‐2 and sulfur disproportionation by the partner DSS. Other types of ANME‐2 were able to directly transfer reducing equivalents from methane oxidation to their partner bacteria via redox‐active extracellular cytochrome c (McGlynn et al ., 2015 ). Observations of thermophilic ANME‐1a/HotSeep‐1 consortia suggested that this transfer was supported by nanowire‐like connections presumably encoded by the pilA gene (Wegener et al ., 2015 ). Here, we used enrichments of three marine AOM consortia growing at different temperature regimes [i.e., ANME‐1a/HotSeep‐1 (60°C), ANME‐1a/Seep‐SRB2 (37°C) and ANME‐2c/Seep‐SRB2 (20°C); Supporting Information Fig. S1] for metagenomic and ‐transcriptomic analyses combined with high‐resolution imaging, to investigate the nature of their interaction and the molecular basis of their methane and sulfate catabolism. Recently, it was found that genomes of the ubiquitous AOM partner bacterium Seep‐SRB1 encode specific, large multiheme cytochromes that may be relevant for the uptake of electrons from ANME (Skennerton et al ., 2017 ). The main focus of this study was to compare the gene expression profiles of different environmental AOM consortia, and to elucidate the key mechanisms for electron transfer and growth, including the transfer of reducing equivalents between the archaeal and bacterial partners. We tested the hypotheses that at AOM conditions a variety of ANME and SRB produce multiheme cytochromes and nanowire structures for the transfer of reducing equivalents.",
"discussion": "Results and discussion Microbial composition of AOM enrichments Our AOM enrichments were retrieved by long‐term (> 4 years) in vitro cultivation at the in situ temperature of their corresponding sampling sites: ANME‐2c/Seep‐SRB2 from Elba seep sediments (20°C, E20; Ruff et al ., 2016 ), ANME‐1a/Seep‐SRB2 (37°C, G37) and ANME‐1a/HotSeep‐1 (60°C, G60) from Guaymas Basin hydrothermal sediment (Holler et al ., 2011 ; Wegener et al ., 2016 ; see Supporting Information Table S1). For Guaymas Basin samples, sediment‐free enrichments were obtained after ∼ 2 years, while samples from Elba were sediment‐free after gravimetric separation of biomass from the sand matrix (see Wegener et al ., 2016 ). Microscopic analyses showed that in these enrichments ANME and their partner bacteria grew in consortia (Fig. 1 A–C, Supporting Information Fig. S1). The consortia were embedded in an extracellular polymeric matrix and contained iron sulfide precipitates as well as carbonate crystals, particularly in the G60 culture. Visible brownish‐red aggregates with sizes of up to 500 µm grew in the G60 culture. Consortia sizes were considerably smaller (< 100 µm, not visible by eye) in our G37 and E20 cultures. The characteristic autofluorescence of the methanogenic cofactor F 420 was present in all ANME cells, indicative of an active methanogenesis pathway in methanogens (Doddema and Vogels, 1978 ) or of the reverse methanogenesis pathway in ANME (Fig. 1 D–F). We noted that large coccoid ANME‐1 cells (∼ 2 µm diameter) dominated the G37 consortia, whereas G60 consortia contained cylindrical ANME‐1 enclosed in an envelope (∼ 0.7 × 1.5 µm) and the ANME‐2c cells in E20 consortia formed dense sarcina‐like cell packages (∼ 1 µm cell diameter; Fig. 1 G–I, Supporting Information Table S1). All partner bacteria could be distinguished visually from the ANME by their rod‐shaped cell shape and cell diameters of < 0.5 µm in E20 consortia and between 0.5 and 0.8 µm in G37 and G60 consortia (Fig. 1 G–I, Supporting Information Table S1 and Supporting Information Figs. S2 and S3). Figure 1 Visualization and composition of long‐term AOM enrichments from Elba (E20) and Guaymas Basin (G37 and G60). A–C. Micrographs of ANME and SRB cells labeled with CARD‐FISH, present in consortia from the different enrichments: (A) ANME‐2c/Seep‐SRB2 from E20, (B) ANME‐1a/Seep‐SRB2 from G37 and (C) ANME‐1a/HotSeep‐1 from G60; scale bar, 10 µm. D–F. Autofluorescence of the methanogenic co‐factor F 420 in ANME‐2c cells from E20 (D) and ANME‐1 cells from G37 (E) or G60 (F); scale bar, 5 µm. G–I . Transmission electron micrographs showing morphology of cells within AOM consortia from E20 (G), G37 (H) and G60 (I). The ‘a’ and ‘b’ annotations indicate ANME and partner bacteria cells, respectively; while a filled arrow points to the matrix enclosing cells of a consortium (H); scale bar, 1 µm. J–L. Taxonomic profile of the E20 (J), G37 (K) and G60 (L) AOM enrichments based on the proportions of 16S rRNA gene and transcript fragments retrieved from metagenomes (MG) and metatranscriptomes (MT‐1 to ‐3), respectively. See legend in Figure for color coding in J–K. Based on the taxonomic classification of metagenomic 16S rRNA gene fragments, we inferred that the enriched community was composed of ∼ 33% (E20) and ∼ 40–56% (G37 and G60) ANME‐2c and ANME‐1a archaea, respectively (Fig. 1 J–L, Supporting Information Fig. S4 and Supporting Information Tables S2 and S3). Similarly, we inferred that 21% (E20) to 26% (G37) of the 16S rRNA gene reads belonged to the partner bacterium Seep‐SRB2; however, HotSeep‐1 accounted for only 14% (G60) of the metagenomic 16S rRNA gene reads. The dominance of ANME‐associated reads over those associated with their partner bacteria was even more pronounced in the proportions of 16S rRNA gene transcripts. Based on the expression of this marker gene, we associated 44% and 61–81% of all 16S rRNA gene transcripts to ANME‐2c and ANME‐1a, respectively. In contrast, only 10%–19% and 7% of the transcriptomic 16S rRNA gene fragments were assigned to the partner bacteria Seep‐SRB2 and HotSeep‐1, respectively (Fig. 1 J–L, Supporting Information Fig. S4 and Supporting Information Tables S2 and S3). Together with the larger size of ANME in the consortia this suggests that they use a greater share of the energy yield from AOM under ambient conditions for growth and sustaining biomass, relative to their partner bacteria. The remaining sequences belonged to other bacteria and archaea such as Bacteroidetes, Chloroflexi, Deferribacteres, Planctomycetes, Firmicutes, Nitrospira or Thaumarchaeota (see Supporting Information Table S2). Genomic features of ANME archaea The reconstructed draft genomes of ANME‐1b (Meyerdierks et al ., 2010 ), ANME‐2d (Haroon et al ., 2013 ) and ANME‐2a (Wang et al ., 2014 ) have greatly improved our understanding of these organisms’ metabolic potential (see also Timmers et al ., 2017 ), yet the gene expression patterns under optimal AOM conditions were not assessed. Furthermore, the substantial phylogenetic and ecological diversity within the ANME groups (Ruff et al ., 2015) calls for greater insight into their underlying genomic variation. Here, we retrieved a draft genome from a meso‐ and thermophilic ANME‐1a of approximately 1.4 and 1.8 Mb (Table 1 ; see also Supporting Information Table S4). This size range represents about half of the assembly size previously described for psychrophilic ANME‐1b (Meyerdierks et al ., 2010 ). In contrast, the retrieved mesophilic ANME‐2c draft genome is considerably larger with ∼ 3.6 Mb and similar in size to the previously described ANME‐2a draft genome (Wang et al ., 2014 ). Based on archaeal and euryarchaeal specific single‐copy genes, the draft genomes are 96%–97% (ANME‐2c, E20), 92%–94% (ANME‐1a, G37) and 91%–92% (ANME‐1a, G60) complete (Table 1 ). Notably, seven of the single‐copy genes that are generally believed to be present in Archaea and Euryarchaea , were not found in the three analyzed ANME‐1 data sets (the two ANME‐1a from this study and ANME‐1b from Meyerdierks and colleagues ( 2010 ); see Supporting Information Table S5). Further, based on single copy gene analysis the degree of contamination by other strains was ≤ 5% in all three draft genomes. In contrast, the previously described ANME‐1b dataset (Meyerdierks et al ., 2010 ) from microbial mats in the Black Sea included a high strain heterogeneity and was considered to be a composite genome of several ANME‐1 strains (Meyerdierks et al ., 2010 ; Parks et al ., 2015 ). This could account for the size discrepancy between the ANME‐1b genome and the meso‐ and thermophilic ANME‐1a genomes reported here. Notably, the genome of mesophilic ANME‐1a contains a higher GC content (52%) relative to the draft genome of thermophilic ANME‐1a (46%) or the low‐temperature adapted ANME‐1b (43%; Meyerdierks et al ., 2010 ) and ANME‐2c (49%), which is at odds with the view that elevated growth temperatures lead to increased GC content in ANME, as suggested in other archaea (Merkel et al ., 2013 ). Table 1 General genome properties and estimated genome completeness of the retrieved draft genomes of ANME and SRB. E20 ANME‐2c G37 ANME‐1a G60 ANME‐1a Size (Mb) 3.61 1.40 1.81 Scaffolds 169 5 38 GC 49 52 46 Genes 3576 1402 1877 rRNAs 5S (2), 16S, 23S 5S, 16S, 23S 5S, 16S, 23S tRNAs 43 44 44 Missing tRNAs Cys, Trp Cys Asp Completeness \n a \n \n 96/97 94/92 92/91 Contamination \n a \n \n 5/4 2/2 4/3 Strain heterogeneity \n a \n \n 25/25 50/50 25/50 E20 Seep‐SRB2 G37 Seep‐SRB2 G60 HotSeep‐1 Size (Mb) 3.57 2.55 2.54 Scaffolds 166 167 1 GC 47 49 37 Genes 3334 2456 2517 rRNAs 5S, 16S, 23S 5S, 16S, 23S 5S, 16S, 23S tRNAs 44 45 47 Missing tRNAs His – – Completeness \n b \n \n 91/94 91/95 95/96 Contamination \n b \n \n 0/1 0/1 1/2 Strain heterogeneity \n b \n \n 0/0 0/0 0/0 \n a. Determined with checkM (Parks et al ., 2015 ) using 207 archaeal‐/189 euryarchaeal‐specific single copy genes. \n b. Determined with checkM (Parks et al ., 2015 ) using 104 bacterial‐/198 deltproteobacterial‐specific single copy genes. ANME archaea highly express genes for reverse methanogenesis In our experiments, we investigated ANME originating from marine enrichments, which used only methane and sulfate as redox couple to deliver energy for growth (Wegener et al ., 2016 ). In the absence of sulfate or methane, neither methane oxidation nor sulfate reduction was detectable. Furthermore, ANME did not mediate methane production or showed growth when provided with other electron donors than methane over 30 days (Wegener et al ., 2016 ). Hence, we attribute the detection of methanogenesis‐related genes and gene transcripts in the ANME draft genomes and transcriptomes, respectively, to their role in methane oxidation. Both, ANME‐1a and ‐2c highly express the genes necessary to produce enzymes that convert methane to CO 2 (Figs. 2 and 3 , Supporting Information Tables S6 and S7). Notably, together with the 16S and 23S rRNA genes, the gene encoding the subunit A of methyl‐coenzyme M reductase ( mcrA ) was among the highest expressed genes in all of the three studied ANME types, approximately 10‐fold higher than the mean expression of all genes of the respective organism (Fig. 3 , Supporting Information Tables S6 and S7). The McrA subunit contains the catalytic center required to activate methane, which is, due to the high activation energy required to break the C—H bond, the rate‐limiting step of the overall pathway. We thus interpret the high expression of mcr as a means for ANME cells to sustain a large amount of Mcr protein to activate sufficient methane for cell catabolism (Krüger et al ., 2003 ). Consequently, the genes encoding the enzymes that catalyze the subsequent steps of methane oxidation are less expressed (Fig. 3 ). Figure 2 Model of the metabolic capacities of the different ANME (red) and SRB (green) relevant for direct electron transfer in syntrophic AOM. Filled circles indicate the gene(s) encoding a feature are detected in the genome; circle color corresponds to clade membership (see legend in Figure); feature color indicates function, for example, methane metabolism (see legend in Figure). Predicted subcellular localization of c‐type cytochromes (CytC) is indicated with lettering as follows: c, cytoplasmic; s, S‐layer incorporated; e, extracellular; p, periplasmic; m, membrane‐associated. Mcr, methyl‐coenzyme M reductase; Mtr, tetrahydromethanopterin S‐methyltransferase; Mer, 5,10‐methylenetetrahydromethanopterin reductase; Met, bifunctional homocysteine S‐methyltransferase/5,10‐methylenetetrahydrofolate reductase; Mtd, F 420 ‐dependent methylenetetrahydromethanopterin dehydrogenase; Mch, methenyltetrahydromethanopterin cyclohydrolase; Ftr, formylmethanofuran‐tetrahydromethanopterin formyltransferase; Fmd, formylmethanofuran dehydrogenase; Fpo, F 420 H 2 :methanophenazine oxidoreductase; Fqo, F 420 H 2 :quinone oxidoreductase; Hdr, CoB‐CoM heterodisulfide reductase; Cdh, acetyl‐CoA decarbonylase/synthase complex; rTCA, reverse tricarboxylic acid cycle; Sat, sulfate adenylyltransferase; Apr, adenylylsulfate reductase; Dsr, sulfite reductase, dissimilatory‐type; Qmo, quinone‐interacting membrane‐bound oxidoreductase complex; Dsr*, dissimilatory sulfite reductase‐associated complex; Qrc, quinone reductase complex; Tmc, transmembrane complex; Aps, adenylyl‐sulfate kinase; Cys, phosphoadenosine phosphosulfate reductase; Asr, sulfite reductase, assimilatory‐type; Nif, nitrogenase; Rnf, electron transport complex; Nuo, NADH‐quinone oxidoreductase complex; Hya, hydrogenase; Mvh/Hdr, methylviologen‐reducing hydrogenase/heterodisulfide reductase complex; Flx/Hdr, flavin oxidoreductase/heterodisulfide reductase complex; Atp, ATP synthase, F‐type; Ntp, ATP synthase, V‐type; MP, methanophenazine; MQ, menaquinone. The F 420 ‐dependent sulfite reductase detected in ANME is not shown. For further details on depicted proteins and the genomic data used in this reconstruction see Supporting Information Table S6. Figure 3 Relative gene expression of ANME and SRB cells during AOM. Expression of the core enzymes of methane oxidation, sulfur metabolism and carbon fixation (colored blue, yellow or orange, respectively, in Fig. 2 ). Circles indicate relative median expression (n = 3 transcriptomic replicates) of a gene in a particular clade. Circle color corresponds to an ANME or SRB clade (see legend in Figure). Note, to account for different sequence counts and compositional effects in the data, the expression of a given gene is shown as a clr using a logarithm base of 2, relative to the expression of all genes of a specific clade. Zero indicates the (geometric) mean expression level, thus positive values indicate greater than mean expression while negative values indicate less than mean expression. For example, a relative expression value of 1 (log 2 2) represents an expression twice as high as the geometric mean expression while a relative expression value of −1 (log 2 0.5) represents an expression half that of the geometric mean expression (i.e., one‐fold change in expression). For enzyme complexes, the expression of genes encoding each subunit is represented by a circle (e.g., for mcr , the expression of subunit A, B, C is shown by individual filled circles). If multiple copies of a gene were detected in one genome, only the most highly expressed one is shown. For gene abbreviations see legend of Fig. 2 and for details on transcriptomic data see Supporting Information Tables S6 and S7. The oxidation of methane by ANME is accomplished by a strict reversal of the canonical methanogenesis pathway (ANME‐2), or by introducing a modification in one step of this pathway (ANME‐1). To initiate the pathway of methane oxidation, a CoM‐bound methyl group is transferred to tetrahydromethanopterin (H 4 MPT). Both ANME‐1a draft genomes do not contain the gene encoding N5,N10‐methylene tetrahydromethanopterin reductase ( mer ). The lack of the mer gene has also been described in genomic data of cold‐adapted ANME‐1b (Meyerdierks et al ., 2010 ; Stokke et al ., 2012 ). The Mer enzyme is required for the F 420 ‐dependent conversion of methyl‐H 4 MPT to methylene‐H 4 MPT and, without it, a bypass for this enzymatic step is needed. Meyerdierks and colleagues ( 2010 ) and Welander and Metcalf (2008) proposed a bypass in which the methyl group is converted to methanol and formaldehyde before re‐entering the methanogenesis pathway via methylene‐H 4 MPT. Genes encoding a fusion protein of a formaldehyde‐activating enzyme (Fae) and a hexulose‐6‐phosphate synthase (Hps) alongside an alcohol dehydrogenase (Adh) are present in ANME‐1 to catalyze such a bypass of the Mer step. However, in both investigated ANME‐1a types transcription of these genes was rather low (i.e., 0.8‐ to 2.2‐fold higher than the geometric mean expression). While absolute expression levels cannot be quantified, such low expression levels could point to other biochemical pathways as bypass. As an alternative to the methanol bypass, Stokke and colleagues ( 2012 ) proposed a substitution of Mer by a methylenetetrahydrofolate reductase (Met). In contrast to Mer, Met is NADH‐dependent and catalyzes the reduction of methylene‐tetrahydrofolate (H 4 F) via a two‐step process in which NADH reduces FAD to FADH 2 and FADH 2 reduces methylene‐H 4 F to methyl‐H 4 F (Shima et al ., 2000 ). Both, the meso‐ and thermophilic ANME‐1a encode the metF and metV genes (Bertsch et al ., 2015 ) and express them at levels comparable to other genes of the reverse methanogenesis pathway (Fig. 3 , Supporting Information Table S6). Moreover, these genes were also present in the ANME‐1b data set of Meyerdierks and colleagues ( 2010 ). Sequence comparison against the NCBI non‐redundant protein database suggests that ANME‐1 may have acquired their met genes by horizontal gene transfer from bacteria. It is, however, unclear why ANME‐1 would employ this enzyme as it involves the transfer to a different cofactor. A biochemical characterization of MetFV including a metabolite analysis is required to elucidate it's functioning in methane turnover in ANME‐1. Autotrophic growth and sulfur metabolism in ANME archaea In line with previous genomic (Meyerdierks et al ., 2010 ; Wang et al ., 2014 ) and immunolabeling (Milucka et al ., 2013 ) observations, we did not identify genes of the canonical sulfate reduction pathway in the ANME draft genomes. However, all three ANME draft genomes encode enzymes of an assimilatory sulfate reduction pathway: adenosine 5′‐phosphosulfate kinase, phosphoadenosine phosphosulfate reductase and assimilatory sulfite reductase (Fig. 2 ). These genes showed low expression levels (Fig. 3 ), indicative of a function in assimilatory rather than dissimilatory sulfur metabolism. Further, the ANME‐2c and ANME‐1a (G37) draft genomes encode a fusion protein containing a F 420 ‐reducing hydrogenase subunit B (FrhB) and a dissimilatory sulfite reductase subunit AB (DsrAB) domain. This F 420 ‐dependent sulfite reductase (Fsr) has a described sulfite detoxifying function in methanogenic archaea (Johnson and Mukhopadhyay, 2005 , 2008 ). Both ANME showed a considerable expression of this protein during AOM (i.e., 3.8‐ and 1.9‐fold more expressed than the geometric mean expression of ANME‐1 (G37) and ANME‐2c, respectively, but substantially less than the mcrA gene; see Supporting Information Table S6). The Fsr enzyme may qualify to catalyze parts of a novel sulfate reduction pathway in ANME, as required in dissimilatory sulfur metabolism in ANME‐2 (Milucka et al ., 2012 ). A function of Fsr in dissimilatory sulfur metabolism has, however, not been demonstrated. Furthermore, our genomic analysis of ANME did not provide evidence for membrane‐bound electron transport complexes known from sulfate reducers to link cytoplasmic sulfite reduction to energy conservation. Based on earlier observations, the three studied ANME types produce their biomass autotrophically by the assimilation of inorganic carbon (Kellermann et al ., 2012 ; Wegener et al ., 2016 ). The presence of genes encoding the acetyl‐CoA decarbonylase/synthase (CODH/ACS) complex in ANME is in agreement with the results of previous studies (Meyerdierks et al ., 2005 , 2010 ; Wang et al ., 2014 ). The CODH/ACS complex combines a carbonyl and methyl group with coenzyme A (CoA) to yield acetyl‐CoA, a central metabolite of the cell. These genes were expressed during AOM, confirming this route of carbon fixation in ANME (Figs. 2 and 3 ). In methanogens utilizing the reductive acetyl‐CoA pathway, the required tetrahydromethanopterin‐bound methyl group is produced in the methanogenesis pathway. Consequently, in the reversal of the methanogenesis pathway, ANME cells should be able to use a methane‐derived tetrahydromethanopterin‐bound methyl group for biomass generation. However, earlier experiments showed that ANME use predominantly inorganic carbon in biomass production (Kellermann et al ., 2012 ; Wegener et al ., 2016 ), hence they seem to use the enzymes of the methanogenesis pathway also in the reductive direction to generate CO 2 ‐derived methyl‐tetrahydromethanopterin. Employing the same enzymes for carbon fixation and energy metabolism could be beneficial for these generally energy‐limited organisms, but the required mechanisms to regulate the reaction direction for the two pathways are not resolved. As autotrophs, ANME use an inorganic nitrogen source. They express the genes for glutamate synthase and glutamine synthetase encoding a two‐step conversion of ammonium to glutamate. ANME‐2c additionally has genes encoding a nitrogenase ( nifDHK ; Fig. 2 , Supporting Information Table S6). This supports previous reports of nitrogen fixation in ANME‐2 archaea as shown in 15 N 2 isotope labeling experiments (Dekas et al ., 2009 , 2015 ). However, the presence of genes encoding N 2 ‐fixation is puzzling as the process is energy intensive (16 ATP per molecule of N 2 fixed), and the ANME typically inhabit very ammonium‐rich and energy‐limited environments. Consequently, the nitrogenase subunits of ANME‐2c show only low expression during growth in the ammonium‐rich medium supplied in cultivation, and it remains to be shown if nitrogen fixation occurs under ammonium limitation. The draft genomes of ANME‐1a contain only the nifH gene, encoding one subunit of the three‐membered nitrogenase, which shows sequence similarity to nifH genes of other ANME‐1. The gene products of these subgroup IV nitrogenases are unlikely functional in nitrogen fixation (Dekas et al ., 2015 ). Also, ANME‐2c contains an additional isolated copy of such a nifH gene (Supporting Information Table S6). In ANME‐2c this nifH gene shows an elevated expression compared to the nifDHK gene cluster. In contrast, in ANME‐1a the nifH gene has only a low expression level (Supporting Information Table S6). Zheng and colleagues ( 2016 ) suggested that ANME‐2d use the nifH gene product in the biosynthesis of cofactor F 430 , an important element in the methyl‐coenzyme M reductase. A similar functionality is likely in ANME‐1a and ANME‐2c, which also require large amounts of this cofactor. Haroon and colleagues ( 2013 ) found that members of the ANME‐2d clade couple methane oxidation to the reduction of nitrate to nitrite. For this ANME‐2d uses a nitrate reductase encoded by narGH , a gene cluster that has been acquired from bacteria. Genes encoding for nitrate‐ or nitrite‐reduction [nitrate reductase ( nar ) and nitrite reductase ( nir )] were not present in the draft genomes of ANME‐1 and ‐2c obtained from fully anoxic, highly sulfidic sediments. Genomic capacities of the partner bacteria in AOM consortia Recent studies provided insights into the draft genome of the AOM partner bacteria HotSeep‐1 (Wegener et al ., 2015 ; Krukenberg et al ., 2016 ) and Seep‐SRB1 (Skennerton et al ., 2017 ). Here we analyzed the draft genome and transcriptome of two Seep‐SRB2 types, representing environmentally relevant partner bacteria from a clade described by Kleindienst and colleagues ( 2012 ). The draft genomes of the two types of Seep‐SRB2 bacteria associated with the mesophilic ANME‐2c and ‐1a had a size of 3.6 Mb (E20) and 2.6 Mb (G37), and an estimated completeness of 91%–94% and 91%–95%, respectively (see Table 1 ). This is in the range of what has been reported for HotSeep‐1 (2.5 Mb, 97% completeness; Krukenberg et al ., 2016 ) and Seep‐SRB1a (∼ 3 Mb; Skennerton et al ., 2017 ). Congruent with our observations from the ANME draft genomes, the GC content of the partner bacteria draft genomes did not show temperature‐dependent variations (47% in E20 Seep‐SRB2, 49% in G37 Seep‐SRB2 and 37% in G60 HotSeep‐1). As previously reported for HotSeep‐1 and Seep‐SRB1 (Krukenberg et al ., 2016 ; Skennerton et al ., 2017 ), both Seep‐SRB2 draft genomes encode the core gene set for dissimilatory sulfate reduction, including membrane‐bound electron transport complexes (Fig. 2 ), but lack genes related to methane metabolism (see Supporting Information Tables S6 and S7). Both, Seep‐SRB2 and HotSeep‐1 contain and express the genes for cytoplasmic sulfate reduction: sulfate adenylyltransferase ( sat ), adenosine phosphosulfate reductase ( aprAB ) and dissimilatory sulfite reductase ( dsrABCD ), as well as the membrane‐bound electron transport complexes: menaquinone‐interacting oxidoreductase complex ( qmoABC ) and dissimilatory sulfite reductase associated complex ( dsrMJKOP ), which are generally found to be encoded in genomes of sulfate reducers (Pereira et al ., 2011 ; Rabus et al ., 2015 ). The previously described draft genomes of HotSeep‐1 (Krukenberg et al ., 2016 ) and Seep‐SRB1a (Skennerton et al ., 2017 ) encode two additional membrane‐bound electron transport complexes: quinone reductase complex (QrcABCD) and transmembrane complex (TmcABCD). All these membrane‐bound multi‐subunit complexes enable the transfer of electrons from the periplasmic cytochrome c pool to the cytoplasmic Apr and Dsr enzymes, allowing for sulfate reduction. The DsrMJKOP and TmcABCD complexes are proposed to directly transfer electrons from the periplasmic cytochrome c pool to the cytoplasmic DsrAB enyzme via the reduction of a disulfide bond in the DsrC protein (Pereira et al ., 2011 ; Venceslau et al ., 2014 ). In contrast, the QmoABC complex is thought to receive electrons from reduced menaquinone for the reduction of the cytoplasmic AprAB enyzme (Pires et al ., 2003 ; Ramos et al ., 2012 ; Grein et al ., 2013 ). The QrcABCD complex is suggested to deliver electrons from periplasmic cytochromes into the membrane menaquinone pool (Venceslau et al ., 2010 ), thus interacting with the Qmo complex via a menaquinone redox loop. Since we did not detect the gene set for a complete Qrc complex in the Seep‐SRB2 draft genomes an alternative route for the reduction of the Qmo complex may exist in these organisms. The Qrc complex is present in most hydrogenotrophic sulfate reducers with a periplasmic hydrogenase and is proposed to be important in coupling periplasmic hydrogen oxidation to sulfate reduction (Venceslau et al ., 2010 ). Our genomic data indicates that only HotSeep‐1 but not Seep‐SRB2 contains a periplasmic hydrogenase. Thus, HotSeep‐1's Qrc complex might be primarily relevant during growth on hydrogen (see below). All three studied SRB also encode other membrane‐bound and cytoplasmic complexes relevant for energy conservation. The Seep‐SRB2 encode genes for the multi‐subunit Nuo complex, which are present in the genomes of many sulfate reducers (Pereira et al ., 2011 ). The genes for this complex were not detected in the HotSeep‐1 genome. Instead, HotSeep‐1 encodes the Rnf complex, which is also found in numerous sulfate reducers including Seep‐SRB1, and other anaerobes (Pereira et al ., 2011 ; Rabus et al ., 2015 , Skennerton et al ., 2017 ). The Rnf complex is proposed to couple the exergonic reduction of NAD + by ferredoxin to the translocation of protons (Na + or H + ) across the membrane (Schmehl et al ., 1993 ). This complex may also work in reverse to catalyze the reverse electron transfer from NADH to ferredoxin driven by a proton gradient (Müller et al ., 2008 ). Like the Seep‐SRB1 draft genome (Skennerton et al ., 2017 ), both Seep‐SRB2 encode the cytoplasmic Flx/Hdr complex (Ramos et al ., 2015 ), which is not encoded in the HotSeep‐1 draft genome. This complex couples the reduction of ferredoxin by NADH to the reduction of a disulfide bond, possibly in the DsrC protein, via electron bifurcation. During hydrogenotrophic growth of HotSeep‐1 a similar reaction may be carried out by the Mvh/Hdr complex. This complex may couple hydrogen oxidation to the endergonic reduction of ferredoxin and the exergonic reduction of a heterodisulfide bond, possibly in DsrC, in a flavin‐based electron bifurcation (Buckel and Thauer, 2013 ). Genes encoding a cytoplasmic hydrogenase (Mvh) are also contained in the Seep‐SRB2 draft genomes, thus an Mvh/Hdr complex may also be present in Seep‐SRB2. However, the functioning of this complex during AOM (i.e., in the absence of hydrogen as energy source) is unclear. Like the ANME genomes, the genomes of the three partner bacteria contain the genes encoding autotrophic carbon fixation. The Seep‐SRB2 genomes encode the reductive acetyl‐CoA pathway, a common feature among autotrophic sulfate reducers including Seep‐SRB1 (Skennerton et al ., 2017 ). Notably, in the E20 and G37 enrichment, the expression of genes encoding the reductive acetyl‐CoA pathway in ANME and their partner bacteria was very similar, documenting the coupled growth patterns of the two consortial members. In contrast, in the thermophile HotSeep‐1, CO 2 fixation likely proceeds via the reductive tricarboxylic acid (rTCA) cycle. This alternative carbon fixation mode is often found in thermophilic microbes inhabiting hydrothermal environments (Campbell and Cary, 2004 ), but so far it is described for only a few sulfate reducers such as Desulfobacter hydrogenophilus (Schauder et al ., 1987 ; Widdel, 1987 ). Compared with the reductive acetyl‐CoA pathway, the rTCA cycle requires approximately twice as much ATP to generate a molecule of pyruvate from CO 2 . Thus, Seep‐SRB1 and Seep‐SRB2 might have an energetic advantage over their thermophilic counterpart HotSeep‐1, but they might not be able to thrive at high temperatures. Based on the genomic data, all three partner bacteria have the capacity to assimilate ammonium via glutamate synthase and glutamine synthetase. As described above for the ANME there is no experimental evidence for N 2 fixation in most AOM partner bacteria (Dekas et al ., 2009 , 2015 ). However, the Seep‐SRB2 draft genome obtained from the E20 enrichment also contains a nitrogenase ( nifDHK ), suggesting at least the capacity of some Seep‐SRB2 to fix molecular nitrogen. Yet the expression of the nif genes is very low (∼ 4‐fold lower than the geometric mean expression), however under cultivation conditions sufficient ammonium is always supplied. Most AOM partner bacteria seem to be obligate syntrophs, hence they do not have the ability to grow alone. HotSeep‐1, however, can grow on hydrogen as an alternative electron donor (Wegener et al ., 2015 ; Krukenberg et al ., 2016 ). Without an ANME partner, HotSeep‐1 couples sulfate reduction to hydrogen oxidation, possibly via its periplasmic and cytoplasmic hydrogenases. Notably, also both Seep‐SRB2 draft genomes encode a cytoplasmic hydrogenase (Fig. 2 ), whereas the Seep‐SRB1 encodes a membrane‐bound Ech hydrogenase (Skennerton et al ., 2017 ). Hydrogenases are common features of sulfate reducers, but all previous attempts to enrich Seep‐SRB2 on hydrogen were unsuccessful (Wegener et al ., 2016 ), hence hydrogen is likely also not metabolized by Seep‐SRB2. Although hydrogen is a well‐known electron carrier in syntrophic interactions (Schink, 1997), it seems that the partner bacteria in AOM consortia never developed or completely lost the capability of hydrogen metabolism. As an exception, the thermophilic partner bacteria may have preserved hydrogen metabolism as additional trait, likely due to the generally higher probability for exposure to hydrogen in vent environments (Martin et al ., 2008). Interaction of ANME with their partner bacteria We investigated the recent hypothesis that a direct exchange of reducing equivalents occurs between the members of AOM consortia (McGlynn et al ., 2015 ; Wegener et al ., 2015 ; Lovley, 2017 ). It was proposed that the transfer of electrons is mediated by large multiheme cytochromes and nanowire‐like structures produced by the consortial partners (McGlynn et al ., 2015 ; Wegener et al . 2015 ; Skennerton et al . 2017 ). This interaction would resemble direct electron transfer between Geobacter species (Summers et al ., 2010 ), or consortia of Geobacter with different methanogens (Rotaru et al ., 2014 ). In case of direct cell‐to‐cell contacts, extracellular c‐type cytochromes might be sufficient to mediate intercellular electron transfer. Here we found that all analyzed ANME and SRB draft genomes contain the genes encoding several c‐type cytochromes (Figs. 2 and 4 , Table 2 , Supporting Information Table S8). We categorized these cytochromes based on their predicted target localization (see ‘Materials and Methods’ section), and considered those with predicted extracellular, unknown or membrane‐/cell wall‐associated localization of potential relevance in interspecies electron transfer. Figure 4 Expression levels and transmission electron microscopic visualization of cytochrome c and potential extracellular structures. Expression levels in ANME ( A–C ) and expression levels in SRB ( D–F ) are shown as relative to the mean expression of all genes of the respective organism (as in Fig. 2 ; n = 3 transcriptomic replicates, for details see Supporting Information Tables S6 and S8). Squares indicate expression level of flagellin ( fla ) genes in ANME (red) and pilin ( pil ) genes in SRB (green). Circles indicate expression level of cytochrome c genes in ANME (red) and SRB (green). Circle size indicates number of heme units in cytochromes (based on the detection of the CXXCH motif). Cytochrome categories are based on predicted subcellular localization: e, extracellular; m, membrane (SRB only); w, cell wall (ANME only); c/p, cytoplasmic or periplasmic (SRB only); c, cytoplasmic (ANME only); u, unknown. Localization predicted using Psortb (see ‘Material and Methods’ section). Filled symbols indicate features with extracellular or potential extracellular localization that are potential participants in interspecies electron transfer. The white star (in row e of panel A) indicates a cytochrome c, potentially incorporated into the S‐layer (detected only in ANME‐2c from E20). For comparison, the median expression level of metabolic key genes for sulfate reduction ( dsrA ; green line) or methane oxidation ( mcrA ; red line) are included in A–F. G–I. Transmission electron microscopy (TEM) micrographs of AOM consortia showing intercellular structures. J–L. Transmission electron microscopy (TEM) micrographs of AOM consortia after DAB staining of heme groups to localize extracellular cytochrome c. Counterstaining was minimized in J–L, thus cell contrast is less pronounced and filaments appear less apparent than in G–I. In images G–L, ‘a’ indicates archaeal cells, ‘b’ indicates bacterial cells and filled arrows point to filaments in the intercellular space (G–I), filamentous connections between cells (G–I) and potential extracellular (J–I) or membrane associated (J,I) heme‐stained cytochrome c. [Colour figure can be viewed at http://wileyonlinelibrary.com ] Table 2 Overview of c‐type cytochromes detected in ANME and SRB, their predicted localization and number of heme binding sites based on the CXXCH motif. E20 G37 G60 Subcellular localization ANME SRB ANME SRB ANME SRB Cytoplasmic or periplasmic (c/p) 6 11 2 16 4 10 Membrane‐ or cell wall‐associated (m/w) – 1 – – – 1 Extracellular (e) 3 2 – 2 1 1 Unknown (u) 3 4 2 7 4 9 Total number of c‐type cytochromes 12 18 4 25 9 21 The three ANME genomes encode for two to five different cytochromes which fulfill these criteria, and for each of the organisms at least one of their cytochromes was highly expressed (∼ 7‐fold higher expression level than the geometric mean expression of all genes of that organism; Fig. 4 A–C). McGlynn and colleagues ( 2015 ) suggested a superior role of multiheme cytochromes detected in the S‐layer of ANME‐2 archaea in enabling electron shuttling to their partner bacteria. Among the three here studied ANME genomes, only ANME‐2c encode a c‐type cytochrome in the genomic context with an S‐layer‐related protein. This cytochrome with 24 heme units is only moderately expressed (0.8‐fold of the geometric mean expression; Fig. 4 , Supporting Information Fig. S5 and Supporting Information Table S8), suggesting a minor role in electron transfer. For comparison, the highest expressed cytochrome in ANME‐2c, an 8‐heme cytochrome, shows an expression sevenfold greater than the geometric mean expression (Fig. 4 , Supporting Information Fig. S5 and Supporting Information Table S8). Both ANME‐1a archaea highly express 4‐heme and 8‐heme cytochromes and do not encode for cytochromes with more than 9 heme groups or for those with an S‐layer domain (Fig. 4 , Supporting Information Fig. S5 and Supporting Information Table S8). The latter is consistent with earlier findings for ANME‐1b (McGlynn et al ., 2015 ). This suggests that large S‐layer associated multiheme cytochromes are not generally required for the archaeal‐bacterial electron transfer. The genomes of the bacterial consortium members encode a larger number (7 to 11) of potentially extracellular cytochromes with up to 16 heme units (Table 2 , Supporting Information Table S8). However, in particular cytochromes with high heme numbers (> 12) were relatively low expressed (less than 2‐fold of the geometric mean; Supporting Information Fig. 5 and Supporting Information Table S8), which is substantially lower than the expression of the metabolic key gene of sulfate reduction ( dsrA ). Based on the gene expression profile of ANME and SRB, it is questionable if large multiheme cytochromes are generally involved in interspecies electron transfer. Instead, this function might be fulfilled by cytochromes with lower heme content (4‐8 or 4‐12 heme units in ANME and SRB, respectively; Fig. 4 and Table 2 , Supporting Information Fig. S5 and Supporting Information Table S8), which are highly expressed by the three studied types of ANME and SRB. Our genomic and transcriptomic data support the hypothesis that extracellular c‐type cytochromes play an important role in archaeal‐bacterial direct electron transfer. However, although cytochromes may efficiently bridge minor distances in the sub‐micrometer scale, electron hopping is not possible on longer distances (Malvankar et al ., 2012 ). Hence, we looked for genes encoding potential extracellular structural connections between the ANME and SRB partners. These include bacterial type IV pili, which potentially form ‘nanowires’ and may play a key role in microbial interspecies electron transfer (Summers et al ., 2010 ; Wegener et al ., 2015 ; Laso‐Pérez et al ., 2016 ; Lovley, 2017 ). Archaea can form cell appendages in form of an archaeal flagellum (i.e., archaellum), which resembles the bacterial type IV pili (Albers et al ., 2015) and which thus might be able to also mediate electron transfer. All three genomes of the partner bacteria contain the genes required for type IV pilus assembly, including the gene encoding the major subunit, pilin ( pilA ). Notably, HotSeep‐1 contains two pilA genes, but only one of those is highly expressed (Fig. 4 ). The pilA gene of the E20 Seep‐SRB2 draft genome was likewise highly expressed whereas the expression of pilA by the G37 Seep‐SRB2 was low (Fig. 4 D–F). The ANME‐1a (G60) and ANME‐2c (E20) genomes encode proteins involved in the formation of archaeal flagella and the high expression of genes encoding the major subunit, flagellin, in these aggregate‐bound ANME may indicate that these proteins are involved in interspecies electron transfer. However, the G37 ANME‐1a genome did not contain such flagellin‐encoding genes and the Seep‐SRB2 partner showed only low pilA expression, suggesting differences between consortia types. Using transmission electron microscopy, we examined all three AOM consortia types for extracellular structures. In all consortia it was possible to distinguish two main cell morphotypes (Fig. 1 , Supporting Information Figs. S2 and S3), corresponding to ANME and bacterial cells. As previously described, the ANME‐1a/HotSeep‐1 consortia (G60 enrichment) form abundant filamentous structures, which connect the two partners under AOM conditions (Wegener et al ., 2015 ). Examining the ANME‐2c/Seep‐SRB2 consortia (E20 enrichment), we also observed a network of intercellular filaments between the densely packed sarcina‐like archaeal cells and the Seep‐SRB2 cells. This is congruent with the high expression levels of pilin and flagellin genes in the studied organisms (Fig. 4 ). Consortia of the G37 enrichment contained some filamentous structures, but coherent with the low expression of pilA and the lack of flagellin encoding genes, these structures were far less abundant than in the other two consortia types. In the G37 consortia the archaea seem to form tight packages, possibly enclosing the partner bacteria (Fig. 1E, H). Hence in this arrangement, the proximity of the two partners may allow electron transfer by cytochromes alone. To locate possible accumulations of cytochromes in the microbial consortia, thin sections of fixed aggregates were stained with the 3,3′‐diaminobenzidine tetrahydrochloride (DAB) assay which allows detection of heme‐rich regions (McGlynn et al ., 2015 ). All three AOM consortia accumulated large amounts of DAB particles around archaeal and bacterial cells and in particular along intercellular filaments (Fig. 4 J–L). This co‐distribution of cytochrome c and filaments in the intercellular space supports an important role of these protein‐derived structures in electron transfer in AOM consortia. However, the conductivity of extracellular structures in AOM consortia potentially formed by pilin and archaeal flagellin is not yet understood. Aromatic ring structures might enable electron transfer in a conductive manner as suggested for the geopilin of Geobacter (Malvankar et al ., 2011 ; Adhikari et al ., 2015 ). Recently Walker and colleagues ( 2018 ) heterologously expressed different pilA genes in Geobacter sulfurreducens and measured the conductivity of the formed pili. In contrast to some pili produced by G. sulfurreducens strains with pilin encoding genes from different organisms, the one formed with the pilin from HotSeep‐1 showed low conductivity (Walker et al ., 2018 ); it is yet unclear if the pili of HotSeep‐1 are correctly assembled by G. sulfurreducens . Direct conductivity measurements on pili and other extracellular structures in AOM consortia would be required to conclusively determine their electrical conductivity."
} | 11,675 |
27502051 | PMC4977534 | pmc | 639 | {
"abstract": "Microbial fuel cells operating with autotrophic microorganisms are known as biophotovoltaic devices. It represents a great opportunity for environmentally-friendly power generation using the energy of the sunlight. The efficiency of electricity generation in this novel system is however low. This is partially reflected by the poor understanding of the bioelectrochemical mechanisms behind the electron transfer from these microorganisms to the electrode surface. In this work, we propose a combination of electrochemical and fluorescence techniques, giving emphasis to the pulse amplitude modulation fluorescence. The combination of these two techniques allow us to obtain information that can assist in understanding the electrical response obtained from the generation of electricity through the intrinsic properties related to the photosynthetic efficiency that can be obtained from the fluorescence emitted. These were achieved quantitatively by means of observed changes in four photosynthetic parameters with the bioanode generating electricity. These are the maximum quantum yield (Fv/Fm), alpha (α), light saturation coefficient (Ek) and maximum rate of electron transfer (rETRm). The relationship between the increases in the current density collected by the bioanode to the decrease of the rETRm values in the photosynthetic pathway for the two microorganisms was also discussed.",
"discussion": "Discussion The F v /F m is typically used to measure the stress conditions that the photosynthetic apparatus or more specifically, the PSII and its electron carriers Q A and Q B , are experiencing. It was observed previously that environmental conditions such as drought 29 , heat 30 , and nutrient limitation resulting from the lack or simply accessibility of nutrients 31 , can affect the values of F v /F m when compared to the same system in a non-stressful and healthy conditions. The experiments performed in this work were executed in the same conditions of biofilm growth, nutrients availability and light exposure. As such, the only parameters that may be causing stress to the cells structure should have electrical origin. The decrease in the values of F v /F m observed is quasi-linear in relation to the decrease in the electrochemical cell voltage for both microorganisms. It is important to point out that the application of the voltage to the electrochemical cell is accompanied not only by the flow of current responsible for the production of electricity, but also by the generation of an electric field. This electric field can interact and modify the proteins and lipids that constitute cellular structures and organelles, and affect photosynthesis directly or indirectly. It is known that application of an electrical field to the cell wall may cause weakening of the lipid-lipid interactions in the lipid bilayer membrane 32 . In extreme cases, the effect of strong electrical fields in the order of kV.cm −1 can exceed the dielectric strength of the cell membrane, and result in the formation of hydrophobic pores through a process called electroporation, increasing the permeability of the lipid bilayer, or in some cases resulting in its disruption 33 . The increase in the flow of electrical current per unitary cell, even with a considerably low value, can also cause the increase of temperature locally, resulting in additional stress that can affect not only the cell itself, but the biofilm as a whole. The stress on the microorganisms resulting from the generation of electricity as a bioanode does not seem to be permanent. Independent to the source of this stress, the original values of F v /F m could be easily recovered by moving the electrochemical cell with the bioelectrodes containing the microorganisms back to the incubator for 15 minutes The value of alpha (α) calculated through the slope in the linear region of the light curve is related to the light requirements for the microorganism to reach its maximum photosynthetic activity. With the increase of voltage applied from 50 to 240 mv, the α values were observed to increase in Chlorella sp. (α increased by 4.8%) and Synechococcus (α increased by 25.2%). The photosynthetic efficiency of the algae may have been enhanced by the increased voltage, as exhibited by the increased α values. The photosynthetic parameter E k represents the value of light intensity in which the photosynthetic rate of the studied material is optimal, and the light absorbed by the active centers equals the rETR m or maximum electron transport rate 34 . It also gives the level of photoadaptation that indicates the threshold of light exposure that can affect its health and growth. Garcia-Mendoza 35 showed that growing the Chlorophyte Chlorella fusca under light conditions with irradiance lower than E k presented a situation where the microorganism can easily cope with their light environment, and adapt itself for an optimum growth while situations under light conditions with irradiance higher than E k would reduce its growth potential 35 . Although changes associated with differential photo acclimation exist, these changes are considered to be developmental, i.e. controlled by gene expression 36 . Such changes involving adaptation by protein and pigment synthesis take more than 30 minutes, typically several hours or even several days 37 . Since the time required to perform each experiment was less than 30 minutes, such changes related to photo-adaptation is not expected to affect the experiments. Although the biofilm was always brought into incubation to recover from any possible stress caused by the experiment itself, the possibility of having modifications in some important structures responsible for the gene expression apparatus resulting from the applied voltage, cannot be disregarded. Since the time scale for the performed experiments is too short to evoke such effects, further experiments are necessary to analyze the effects of long term adaptation to the microorganisms present in the biofilms in terms of E k as well as genomic modifications. For the electrochemical measurements, it is possible to observe that similar to the observations in Fig. 1 , the photosynthetic kinetics involved in Chlorella appears to be less influenced by the circulation of electrical current in its biofilm as a bioanode in comparison with Synechococcus . Under both conditions of OCP, and with circulation of electrical current, the bioelectrode containing Chlorella behaved with fast photosynthetic kinetics. On the other hand, the bioelectrode containing Synechococcus showed a complete change in its photosynthetic behavior, changing from a fast kinetics under conditions of OCP to a low kinetics with circulation of electrical current. The observed effect of having the circulation of electrical current in the electrochemical device is the decrease in the photosynthetic parameter rETR m . This time, the circulation of current in the bioanode provokes a “slowdown” to the relative rate of electron transfer happening in the chain of electrochemical reactions involving the transport of charges from photosystem II to photosystem I. Two possible phenomena may be responsible for the decrease in the rETR m . First, some of the chemicals responsible for the functionality of the photosystems may have been affected by the electrical or electrochemical environment that the biofilm was experiencing, following the changes observed in F v /F m . Second, the electrons that have been collected by the bioanode during the generation of electricity can be linked to one of the redox components acting in this chain of reactions. The electrons therefore can be “leaked” to a secondary chemical (redox) pathway that is electrically connected to the electrode surface. Examination of the shape of the polarization curves suggests that the electrochemical device operating with the bioelectrode containing the biofilm of Chlorella works under conditions of activation polarization voltage drop due to the small increase of net current with the decrease of the potential. This is usually reflected by a poor kinetics of electron transfer from a catalyst in an electrode surface that hinders the collection of electrons from the electrode. On the other hand, the electrochemical device running with the bioanode containing the biofilm of Synechococcus works under conditions of Ohmic polarization. Voltage drop due to the quasi-linear increase of current with the decrease of voltage, means that the kinetics of electron transfer is not a limiting factor for this bioelectrode. In both cases, the voltage drop due to the mass transfer limitation doesn’t appear to play a role within the performance of these bioelectrodes. This is expected for a biocatalyst in the form of biofilm adsorbed in the surface of the electrode containing the active chemicals that may react right away in the surface of the electrode, i.e. not diffusing from the solution to the electrode surface. McCormick et al . 38 recently presented a very interesting discussion regarding the state-of-art on the current development in BPV systems. The authors discussed the possible mechanisms for the electron transfer involving photosynthetic microorganisms adsorbed in the surface of the electrode, oxidation of end-chemicals produced as metabolites, electrochemical cycling of endogenous lipid-soluble and insoluble natural mediators, direct electron transfer through available surface redox proteins and finally direct contact by conductive nanostructures produced by the cells in the biofilm with the electrode surface 38 . Considering the evidences obtained from this work, we may speculate that the possible mechanisms occurring with the two types of microorganisms working as the bioanodes should be direct electron transfer through redox reactions happening between surface proteins present in the structure of these cells, or possibly through conductive electron transfer structures, such as pili or nanowires. Since the last two proposed structures were never observed in these two types of microorganisms, this possibility is disregarded. The fact that the polarization drop effect in the two microorganisms studied is different suggests that either the mechanism responsible for the electrochemical reactions is different, or the nature of the microorganisms play an important role in the bioelectrochemical process. In fact, Chlorella being an eukaryotic algae with a cell wall, presents an organized, and slightly compartmentalized structure, while Synechococcus as a prokaryote presents all the intracellular components scattered in the cytoplasm. This peculiarity gives extra mobility to the redox components present inside the cell, which can experience the electrode potential with higher intensity, leading to an increase in the probability to reach and react with the electrode surface. Examination of the shape of the curves presented in Fig. 5 gives a comprehensive vision of the bioelectrochemical phenomena happening within the biofilm/electrode interface. In both cases, the curves presenting the measurements of current and maximum rate of electron transfer rate seem to be almost mirror imaged. Even though the parameters were obtained from completely different techniques, electrical current measured through the connection with a potentiostat, and the maximum rate of electron transfer rate measured through the PAM fluorescence measurement, there seems to be a direct relationship in the two values, since the increase on the normalized current collected by the bioanode is coupled to the decrease on the normalized maximum rate of electron transfer rate. As mentioned earlier, a correlation coefficient of −0.76 and −0.79 for the values of current density and rETR m for Chlorella and Synechococcus bioelectrodes, respectively. A value of −1 would represent a perfect correlation coefficient, where every single electron “lost” from the photosynthetic pathway would be given to the electrode for the generation of electricity. The fact that the correlation coefficient in both cases is not close to −1 suggests that this process of electron transfer does not happen directly. This is expected considering that the photosynthetic mechanism is happening at the thylakoid membrane, which is located inside the chloroplast for Chlorella and within the cell interior for Synechococcus 39 . These two biological structures isolate the internal mechanism from the outside, and therefore the electrons are not coming directly from the photosynthetic pathway, but possibly from a side mechanism connected to the photosynthesis process. Although the results obtained in this work are suggesting direct correlation between the electrical current generated by the electrochemical device with some of the biochemical redox processes happening on the cells present in the biofilm, the true source of the electrons responsible for the generation of electrical current in both Chlorella and Synechococcus cannot be confirmed. In spite of that, we believe that we have accomplished a remarkable step towards the comprehension of the fundamental bioelectrochemical processes involving the interface between the biochemical reactions happening within the microorganisms in the biofilm and the electrode reactions involving the transfer of electrons with the generation of electricity. We have reported here a novel tool towards understanding the various fundamental biochemical processes that occur at the electrode and biofilm interface. This is crucial to comprehend the transfer of electrons that generates electricity. The comprehension of these fundamentals is an indispensable factor necessary to overcome the low performance in the generation of electricity that BPVs present."
} | 3,447 |
27460798 | PMC4981716 | pmc | 640 | {
"abstract": "ABSTRACT Harnessing the metabolic potential of photosynthetic microbes for next-generation biotechnology objectives requires detailed scientific understanding of the physiological constraints and regulatory controls affecting carbon partitioning between biomass, metabolite storage pools, and bioproduct synthesis. We dissected the cellular mechanisms underlying the remarkable physiological robustness of the euryhaline unicellular cyanobacterium Synechococcus sp. strain PCC 7002 ( Synechococcus 7002) and identify key mechanisms that allow cyanobacteria to achieve unprecedented photoautotrophic productivities (~2.5-h doubling time). Ultrafast growth of Synechococcus 7002 was supported by high rates of photosynthetic electron transfer and linked to significantly elevated transcription of precursor biosynthesis and protein translation machinery. Notably, no growth or photosynthesis inhibition signatures were observed under any of the tested experimental conditions. Finally, the ultrafast growth in Synechococcus 7002 was also linked to a 300% expansion of average cell volume. We hypothesize that this cellular adaptation is required at high irradiances to support higher cell division rates and reduce deleterious effects, corresponding to high light, through increased carbon and reductant sequestration.",
"introduction": "INTRODUCTION It has long been established that photoautotrophic growth is dependent on the combined rates of light, carbon, and macronutrient acquisition and the efficiencies by which these resources are directed toward biomass synthesis ( 1 ). The inherent differences between the rates of light and dark processes highlight an important paradigm that defines cyanobacterial productivity constraints, whereby photosynthetic energy acquisition and CO 2 fixation are interdependent but maintain distinct enzymatic mechanisms and kinetics ( 2 , 3 ). Under high irradiances, photosynthesis is adversely affected by damage (photoinhibition) to the light-harvesting machinery ( 1 ) and/or by photorespiration as RuBisCO selectivity shifts toward O 2 ( 4 ). Notably, most of what is known about these effects as a function of irradiance and O 2 tension concerns photosynthesis ( 5 – 7 ) rather than cell growth. The typical range of doubling times for well-characterized unicellular cyanobacteria (e.g., Synechocystis sp. strain PCC 6803 and Synechococcus elongatus PCC 7942) is between 7 and 12 h ( 8 , 9 ). Several hypotheses concerning mechanisms constraining cyanobacterial growth rates have been proposed; these include spatial restrictions within the cell that limit diffusion processes ( 10 ) as well as metabolic costs that determine partitioning of cellular resources and resulting fitness ( 11 , 12 ). The optimization model, developed to simulate the partitioning of material and energy within a photoautotrophic cell ( 11 ), has extended the concept of growth as a function of proteome allocation between adaptation to niche-specific environments and cell division resources. An important implication of this is the ability to minimize the requirement for niche-adaptive responses that may be a key for cyanobacteria to redirect energy and nutrients efficiently toward the biosynthesis of biomass ( 13 ). This also holds significance for understanding regulatory switches governing central metabolism and secondary biosynthetic pathways, which are primary targets for bioengineering of cyanobacteria ( 14 , 15 ). In this study, we systematically dissected the growth and photophysiological performance of Synechococcus sp. strain PCC 7002 (hereafter Synechococcus 7002), a fast-growing euryhaline unicellular cyanobacterium ( 16 ) that has become a promising biotechnological platform ( 17 – 20 ). Although the physiological behavior of Synechococcus 7002 has been investigated under a wide range of irradiance, temperature, and salinity conditions ( 1 , 9 , 21 – 23 ), there is still a paucity of information concerning principles underlying cyanobacterial growth efficiency and robustness. The photosynthetic potential, thresholds of inhibition, and growth are not explicitly equivalent, and significant variation is expected between different environments ( 24 ). Herein, we compared the growth and photosynthetic rates of Synechococcus 7002 across varied incident irradiance and dissolved O 2 with a slower growing unicellular cyanobacterium, Cyanothece sp. strain ATCC 51142 (hereafter Cyanothece 51142) ( 25 ). Cyanothece 51142 was chosen as a comparative organism because it is a well-characterized strain that is amenable to continuous cultivation ( 15 , 17 , 26 ). The growth rates of Cyanothece 51142 are similar, compared to those displayed by other model cyanobacterial strains, such as Synechocystis strain 6803, and Synechococcus elongatus strain 7942 ( 8 , 9 ). The irradiance-dependent response was further investigated through global RNA sequencing analysis to correlate the ultrafast growth with the photophysiological dynamics and gene expression of Synechococcus 7002. Through integration of state-of-the-art cultivation with photophysiological kinetic analyses and transcriptomic measurements, this genome study provides a new level of insight into the mechanisms guiding the energy and resource partitioning in cyanobacteria and sheds light on the phenomenon of ultrafast photoautotrophic growth.",
"discussion": "DISCUSSION This study yields insight into an important biological principle, which allows unicellular cyanobacteria to achieve ultrafast growth by having ultrahigh growth rates and manage cellular resource under different irradiance-controlled growth regimes. Herein, we provide direct evidence that the bimodal transition around the theoretical saturating irradiance ( I k ) extends not only to the adjustment of the photosynthesis and growth rates but is also coupled to regulation of specific metabolic reactions and cellular functions. The mechanisms by which Synechococcus 7002 mitigates the negative effects of high irradiance that typically inhibit the growth and photosynthesis processes of unicellular cyanobacteria are likely to have broader implications for understanding the metabolic and regulatory underpinnings of photosynthetic growth ( 2 , 3 ). Consistent increases in relative transcript abundance with increasing irradiance was observed for genes encoding translational machinery, amino acid and nucleotide biosynthesis, the ATPase complex, and the anaplerotic pathways of central carbon metabolism ( Fig. 4 , cluster I; see Table S2 in the supplemental material). At the same time, the irradiance-driven increase in the relative growth rate of Synechococcus 7002 coincided with the broad decrease in transcripts encoding light acquisition machinery and photosystem I and II reaction centers. These coupled transcriptional responses suggest that the level of resources expended for biomass (i.e., protein synthesis) and energy (ATP) synthesis continue to increase with irradiance, while biomass production and net photosynthesis rates are essentially constant due to increased resource expenditure required. Furthermore, our data provide direct experimental support to earlier calculations ( 11 ) positing that energy demand increases to sustain growth, even when growth is light saturated, across increasing irradiance inputs. Reduced net growth-to-photosynthesis yields (Q lim > Q sat ) during light-replete steady states ( Fig. 2 ) confirm that, while Synechococcus 7002 growth is not inhibited at the high- I i treatments, the ratio of biomass production to energy acquisition decreases once irradiance exceeds an optimum observed near I k . This is most likely due to the decrease in the optical cross section of the photosynthetic apparatus, as Synechococcus 7002 transitions from light-limited to light-saturated growth ( 1 ). This transition has negative effects upon the quantum efficiencies of PS II, and in other oxygenic phototrophs, it is linked to increased photoinhibition. Here, we suggest that the reduction in antenna size also allows the cells to redirect carbon and energy fluxes toward biosynthetic processes which fuel cell division. Notably, this metabolic redistribution in Synechococcus 7002 occurs in conjunction with a tripling in average cell volume ( Fig. 5 ), a physiological phenomenon that is known to also occur in heterotrophic organisms (i.e., Salmonella enterica serotype Typhimurium) in response to increased rate of division ( 37 ). These increases in cell mass and size are thought to accommodate the changes in the number of nuclei/cell and can relieve molecular crowding that limits cell growth ( 11 ). To that end, removing physical constraints may increase the intracellular capacity needed to accommodate the biosynthetic machinery supporting higher growth rates and alleviate photoinhibition through increased reductant sequestration capacities. This mechanistic concept is further corroborated by the absence of observed photoinhibition within the Synechococcus 7002 steady states ( Fig. 3 ), as increases in irradiance correlated with the generally decreased mRNA levels of known light-sensitive photosystem reaction centers ( psbAD and psaAB ). Interestingly, Synechococcus 7002 may be unusual, as the lack of elevated transcription of psb genes under increasing irradiance contrasts with other transcriptional studies of various slower-growing cyanobacteria exposed to differential and/or stress-inducing light regimes ( 38 – 42 ). The signatures of PS II inhibition were also not observed in the Chl a fluorescence analyses performed on Synechococcus 7002 (data not shown). For example, the relative maximum rate of electron transport (rETR max ) increased through the light-limited regime and showed no significant change during light-saturated steady-state growth (transitional saturated states). Furthermore, the regulation of genes mediating reactive oxygen species (ROS) scavenging does not occur uniformly at the transcriptional level. This is consistent with previous gene expression studies ( 23 , 43 , 44 ), which suggested that lack of strong concerted upregulation of stress response machinery under high irradiance levels may reflect other mechanisms employed by Synechococcus 7002 for dealing with excess reductant to avoid photoinhibition. A key observation in support of the above conclusion is based on substantially elevated levels of rCEF displayed by Synechococcus 7002 under saturating high irradiances ( Fig. 2 ), indicating reduction of the plastoquinone (PQ) pool from electron donors not associated with PS II ( 22 ). While the exact mechanisms and role of CEF are not fully understood in cyanobacteria, it is postulated that this process contributes to balancing reductant and ATP pools, especially under low irradiance levels ( 45 ). Interestingly, psaE and ndhOP genes, encoding the PS I subunit and PQ-oxidoreductase, respectively, which were previously implicated in CEF ( 35 , 46 ), displayed maximum relative expression only during the transitional steady state but were downregulated at saturating irradiances (cluster IV in Table S5 in the supplemental material). In contrast, genes encoding the core subunits of the Ndh-1 complex along with the NdhD1 subunit responsible for cyclic electron transfer around PS I ( ndhHIJK and ndhA-ndhB-ndhCG ) showed irradiance-dependent increase in abundance under light-saturating conditions (cluster III in Table S4 ). Thus, our data implicate new genes (i.e., ndhHIJK and ndhA-ndhB-ndhCG ) with potentially important roles in the cyclic electron transport of Synechococcus 7002 that are inherently linked to its ability to effectively partition reductant fluxes and avoid detrimental effects of oxidative stress using multiple strategies. In summary, a coordinated, functionally grouped gene expression was observed which broadly supports inferences about the role that transcriptional regulation may play in processes such as photosynthesis, carbon fixation, electron transport, and stress response. Interestingly, the kinetics of growth and photosynthesis revealed bimodal growth regimes with respect to carbon uptake to chemical energy production yields, providing evidence of increased cellular energy expenditure during growth at high (but not inhibiting) irradiances. This effect was corroborated by observations that the relative expression of genes involved in biomass synthesis and chemical energy production continued to increase even when the growth and photosynthetic rates were essentially constant, showing strong evidence for an increased energy requirement with increasing light energy input. Only a few other studies have ever demonstrated distinct saturating and/or inhibition kinetics of specific growth rate as a function of irradiance ( 47 – 50 ). Finally, this study sets a bench mark for specific growth rate (μ = 0.2 h −1 ) achieved by mesophilic cyanobacteria ( 16 , 43 , 51 ) and provides a foundation for understanding the linkages between photosynthetic performance and growth which can bring fundamental new insights that are broadly applicable to other photosynthetic and nonphotosynthetic biological systems."
} | 3,303 |
37747940 | PMC10553843 | pmc | 641 | {
"abstract": "Sulfate-coupled anaerobic oxidation of methane (AOM) is performed by multicellular consortia of anaerobic methanotrophic archaea (ANME) in obligate syntrophic partnership with sulfate-reducing bacteria (SRB). Diverse ANME and SRB clades co-associate but the physiological basis for their adaptation and diversification is not well understood. In this work, we used comparative metagenomics and phylogenetics to investigate the metabolic adaptation among the 4 main syntrophic SRB clades (HotSeep-1, Seep-SRB2, Seep-SRB1a, and Seep-SRB1g) and identified features associated with their syntrophic lifestyle that distinguish them from their non-syntrophic evolutionary neighbors in the phylum Desulfobacterota. We show that the protein complexes involved in direct interspecies electron transfer (DIET) from ANME to the SRB outer membrane are conserved between the syntrophic lineages. In contrast, the proteins involved in electron transfer within the SRB inner membrane differ between clades, indicative of convergent evolution in the adaptation to a syntrophic lifestyle. Our analysis suggests that in most cases, this adaptation likely occurred after the acquisition of the DIET complexes in an ancestral clade and involve horizontal gene transfers within pathways for electron transfer (CbcBA) and biofilm formation (Pel). We also provide evidence for unique adaptations within syntrophic SRB clades, which vary depending on the archaeal partner. Among the most widespread syntrophic SRB, Seep-SRB1a, subclades that specifically partner ANME-2a are missing the cobalamin synthesis pathway, suggestive of nutritional dependency on its partner, while closely related Seep-SRB1a partners of ANME-2c lack nutritional auxotrophies. Our work provides insight into the features associated with DIET-based syntrophy and the adaptation of SRB towards it.",
"conclusion": "Conclusions This comparative genomic analysis of the major ANME-partnering SRB clades provides a valuable metabolic and evolutionary framework to understand the differences between the various syntrophic sulfate reducing partners of anaerobic methanotrophic archaea and develop insight into their metabolic adaptation. In this work, we show that the electron transport chains of the different syntrophic SRB partners of ANME are adapted to incorporate EET conduits that are needed for DIET. Groups including the Seep-SRB2 appear to have acquired cytoplasmic membrane complexes that can function with the EET conduits, while Seep-SRB1a clades have adapted existing inner-membrane complexes for interaction with the EET conduit. Electron bifurcation also appears to be common across the syntrophic lineages and is often coupled to the cytoplasmic machinery and likely provides an advantage in low-energy environments. We also show that the coevolution between different ANME and SRB partners may have resulted in nutritional interdependencies, with cobalamin auxotrophy observed in at least one of the specific syntrophic SRB subclades. Our genome-based observations provide insight into the various adaptations that are correlated with the formation of different ANME-SRB partnerships. These adaptive traits appear to be related with mechanisms driving other ecological phenomena such as biofilm formation and non-obligate syntrophic interactions. The identification of these traits allowed us to posit important steps in the evolutionary trajectory of these SRB to a syntrophic lifestyle. While the full import of these observations is not yet clear, they offer a roadmap for targeted physiological investigations and phylogenetic studies in the future.",
"introduction": "Introduction Syntrophy is metabolic cooperation between microorganisms for mutual benefit. It is a common adaptation in low-energy environments and enables the utilization of substrates which neither organism could metabolize on its own [ 1 , 2 ]. While the driving force for different microbial syntrophic interactions may vary, both partners benefit from sharing nutrients and electrons in this way, combining their resources and avoiding the need for both partners to expend energy for the synthesis of common nutrients [ 1 , 3 ]. Syntrophic interactions appear to be specific in at least some cases, with the same organisms co-associating across different ecosystems and environments [ 4 ]. However, we do not yet understand the physiological basis driving the specificity of interactions, often because syntrophic associations are difficult to grow in culture. Characterizing the specificity of these interactions is challenging with uncultured syntrophic consortia in the environment [ 2 ]. A classic syntrophic partnership is at the heart of the important biogeochemical process, sulfate-coupled anaerobic oxidation of methane (AOM) [ 5 ]. Anaerobic methanotrophic archaea (ANME) and sulfate-reducing bacteria (SRB) coexist in multicellular consortia, with ANME performing methane oxidation coupled to sulfate reduction by the SRB [ 6 – 8 ]. Direct interspecies electron transfer (DIET) from ANME to SRB is predicted to be the dominant mechanism of syntrophic coupling in many observed cases of sulfate-coupled AOM [ 9 , 10 ] though diazotrophic nitrogen is also shared between these partners [ 11 – 13 ]. There is rich ecological diversity in the observed examples of AOM with taxonomically divergent groups of ANME coexisting with an equally diverse group of SRB in consortia that appear morphologically different [ 13 , 14 ]. ANME-SRB consortia exist in hydrothermal vents [ 15 ], in cold-seeps [ 6 , 7 , 16 ], in mud volcanoes [ 17 ], and a euxinic basin [ 18 ], and can form tight spherical aggregates [ 16 , 19 ], dense microbial mats [ 20 ], or loose associations [ 19 , 21 ]. Past work has suggested that some ANME-SRB associations are more specific than others [ 13 , 22 , 23 ], and ecophysiological studies have demonstrated differences in genes expressed, even pertaining to DIET [ 24 , 25 ]. To investigate whether there is underlying structure to this variety of ANME-SRB interactions in different ecosystems, and to infer the evolutionary trajectories that led to these extant phenomena, it is important to establish a taxonomic, ecological, and physiological framework within which to organize our observations. Investigation of the archaeal and bacterial lineages involved in AOM identified at least 3 divergent taxonomic groups of archaea, by analysis of 16S rRNA gene sequences and fluorescence in situ hybridization (FISH)–ANME-1 (Methanophagales), ANME-2, and ANME-3 (Methanovorans) [ 6 , 17 , 26 ]. All 3 of these groups are clades within the phylum Halobacterota, and ANME-2 is subdivided into the clades ANME-2a ( Methanocomedenaceae ), ANME-2b ( Methanomarinus ), ANME-2c ( Methanogasteraceae ), and ANME-2d ( Methanoperedenaceae ) [ 14 ]. In a recent paper [ 14 ], Chadwick and colleagues established a robust taxonomic framework for ANME, identified key biochemical pathways in the archaea that are important for AOM and demonstrated that the capability for extracellular electron transfer (EET) is a significant metabolic trait that differentiates ANME from its nearest evolutionary neighbors that are typically methanogens or alkane oxidizing microorganisms [ 27 – 29 ]. Within the ANME (ANME-1, ANME-2a, ANME-2b, ANME-2c, and ANME-3) that are known to partner with SRB [ 6 , 16 , 19 , 26 , 30 ], they were also able to show that there were differences between clades with respect to putative EET pathways, electron transport chains, and biosynthetic pathways [ 14 ]. Analysis of ANME genomes and inferences from previous physiological data [ 24 , 25 ] also suggested differences in secretion machinery and cellular adhesion proteins in the archaeal partner that might affect syntrophic interactions with the partner bacteria [ 14 ]. The goal of this work is to establish a similar framework for the well-established clades of sulfate-reducing partner bacteria, to identify important differences in biochemical pathways, and to identify traits that might correlate with differences in ANME-SRB partnership pairing. The key to understanding the evolutionary diversification of SRB is to understand the taxonomic background of each syntrophic SRB involved in this process, since taxonomy is the “expression of evolutionary arrangement” [ 31 ] and the phenotypic traits that differentiate the syntrophs from their non-syntrophic evolutionary neighbors. Previous work demonstrated that there were 4 sulfate-reducing bacterial clades within the Desulfobacterota that partner ANME–HotSeep-1 [ 24 , 32 ], Seep-SRB2 [ 24 ], Seep-SRB1a [ 25 , 33 ], and Seep-SRB1g [ 13 , 33 ]. Other bacteria and archaea (seepDBB within the Desulfobulbaceae [ 34 ], alpha- and beta-proteobacteria [ 35 ] and verrucomicrobia [ 36 ], Anaerolineales and Methanococcoides [ 37 ]) have also been observed to associate with ANME. However, in this work, we investigated only these 4 clades (HotSeep-1, Seep-SRB2, Seep-SRB1a, and Seep-SRB1g) that are consistently and most often found in association with ANME across different ecosystems [ 23 ]. With the use of publicly available datasets and 15 metagenomes, we generated in this study from different marine ecosystems (seeps located off the coast of Costa Rica and off the coast of S. California, as well as hydrothermal vents in the Gulf of California), we curated a database of 46 syntrophic SRB metagenomes with multiple representatives from each clade. With the use of the Genome Taxonomy Database [ 38 ], we created a taxonomic framework to reproducibly classify these syntrophic SRB and proposed scientific names according to the latest guidelines. Significantly, our curated database of representative genomes and 16S rRNA sequences would allow future studies to differentiate the known syntrophic Seep-SRB1a and Seep-SRB1g clades from the non-syntrophic members (Seep-SRB1b, Seep-SRB1c, Seep-SRB1d, Seep-SRB1e, and Seep-SRB1f) of the polyphyletic clade Seep-SRB1 [ 23 , 39 ]. To differentiate the phenotypic traits of syntrophic SRB from non-syntrophic SRB, we synthesized information from prior physiological experiments [ 24 , 25 ] and provide a detailed biochemical description of pathways that are necessary for the formation of ANME-SRB consortia. Our analysis demonstrated that the syntrophic SRB contain all the genomic traits consistent with their participation in DIET (including EET pathways), and with the formation of a multispecies conductive biofilm (cellular adhesion pathways, polysaccharide biosynthesis pathways). Comparative genome analysis between syntrophic genomes and over 550 non-syntrophic bacteria within the phylum Desulfobacterota, showed that these traits are rare in non-syntrophic SRB. We also investigated the importance of partner-pairing as a meaningful ecological factor that differentiates species of syntrophic SRB. We tested this by sequencing single ANME-SRB consortia, isolated by fluorescence-activated cell sorting (FACS) [ 36 ]. We showed that Seep-SRB1a partners of ANME-2c appear to have cobalamin biosynthesis pathways while Seep-SRB1a partners of ANME-2a do not, indicating the latter species of Seep-SRB1a had developed a nutritional dependence on its partner. These results indicate that there might be characteristics that are unique to different ANME-SRB pairings and lay the groundwork for future studies to use a species-level partnership framework to explore the co-diversification of ANME and SRB. Our study highlights the complex evolutionary trajectory of adaptation of these SRB to syntrophy with ANME and provides insight into the defining features of DIET-based syntrophic interactions.",
"discussion": "Results and discussion Taxonomic diversity within syntrophic SRB partners of methanotrophic ANME To investigate the adaptation of SRB to a partnership with ANME, we first placed them into their taxonomic context and assessed the phylogenetic diversity within the SRB clades (Seep-SRB1a, Seep-SRB1g, Seep-SRB2, and HotSeep-1). For this analysis, we compiled a curated dataset of metagenome-assembled genomes (MAGs) from these SRB clades including 34 previously published genomes [ 25 , 32 , 33 , 40 – 44 ] and 12 MAGs assembled for this study. Five of these genomes were reconstructed from seep samples collected off the coast of California, Costa Rica, and within the Gulf of California. We also sequenced single ANME-SRB consortia that were sorted by FACS after they were SYBR-stained as previously described [ 37 ]. With this technique, we could be confident of the assignment of partners that physically co-associate within the sequenced aggregates and begin to identify partnership-specific characteristics. From sequencing of single consortia, we obtained 2 genomes of ANME-2b associated Seep-SRB1g, 1 genome of ANME-2a associated Seep-SRB1a, and 3 genomes of ANME-2c associated Seep-SRB1a ( Table 1 ). We recovered an additional 3 genomes of the nearest evolutionary neighbors of HotSeep-1 within the order Desulfofervidales since this order of bacteria is very poorly represented in public databases. Our dataset for comparative genomics analysis comprised the above mentioned 46 genomes of syntrophic SRB and over 550 other bacteria from Desulfobacterota. Having compiled this dataset of syntrophic SRB, we also designated type material and proposed formal names for 3 of the syntrophic SRB clades, Seep-SRB2 ( Candidatus Desulfomithrium gen. nov.), Seep-SRB1a ( Candidatus Syntrophophila gen. nov.), and Seep-SRB1g ( Candidatus Desulfomellonium gen. nov.). The genomes designated as type material are identified in Figs 1 and S1 . Further details are available in the Supporting information as a proposal for formal nomenclature for Seep-SRB1a, Seep-SRB1g, and Seep-SRB2 ( S1 Text ). 10.1371/journal.pbio.3002292.t001 Table 1 List of genomes from syntrophic SRB labeled by clade, generated in this study and compiled from public databases. Assembly Organism Proposed formal name Genome size (bp) Completeness Contamination GTDB_classification Geographic location Reference JAJSZM000000000 Desulfofervidales_sp._FWG156 * 1452671 90.31 1.42 d__Bacteria;p__Desulfobacterota;c__Desulfofervidia;o__Desulfofervidales; f__DG-60;g__;s__ Pescadero Basin, Gulf of California This study JAJSZL000000000 Candidatus Desulfofervidaceae sp. 1 * 2230686 97.97 3.46 d__Bacteria;p__Desulfobacterota;c__Desulfofervidia;o__Desulfofervidales; f__Desulfofervidaceae;g__;s__ Pescadero Basin, Gulf of California This study JAJSZS000000000 Candidatus Desulfofervidaceae sp. 2 * 2268002 98.24 3.49 d__Bacteria; p__Desulfobacterota; c__Desulfofervidia; o__Desulfofervidales; f__Desulfofervidaceae; g__; s__ Hydrothermal vent Pescadero Basin, Gulf of California This study GCA_001577525.1 HotSeep1 2540211 96.75 4.27 d__Bacteria; p__Desulfobacterota; c__Desulfofervidia; o__Desulfofervidales; f__Desulfofervidaceae; g__Desulfofervidus; s__Desulfofervidus auxilii Hydrothermal vent Guaymas Basin, Gulf of California [ 32 ] HotSeep1_draft_B50 HotSeep1 B50 2215853 95.53 9.35 d__Bacteria; p__Desulfobacterota; c__Desulfofervidia; o__Desulfofervidales; f__Desulfofervidaceae; g__Desulfofervidus; s__Desulfofervidus auxilii Hydrothermal vent Pescadero Basin, Gulf of California This study CAJIMJ000000000 HotSeep1 E37 1863098 87.40 3.46 d__Bacteria; p__Desulfobacterota; c__Desulfofervidia; o__Desulfofervidales; f__Desulfofervidaceae; g__Desulfofervidus; s__Desulfofervidus auxilii Guaymas Basin, Gulf of California [ 40 ] CAJIMK000000000 HotSeep1 E50 1842718 70.82 5.08 d__Bacteria; p__Desulfobacterota; c__Desulfofervidia; o__Desulfofervidales; f__Desulfofervidaceae; g__Desulfofervidus; s__Desulfofervidus auxilii Guaymas Basin, Gulf of California [ 40 ] JAJSZN000000000 HotSeep1 FWG170 1777984 86.7 6.74 d__Bacteria;p__Desulfobacterota;c__Desulfofervidia;o__Desulfofervidales; f__Desulfofervidaceae;g__Desulfofervidus;s__Desulfofervidus auxilii Hydrothermal vent Pescadero Basin, Gulf of California This study JAAXOL000000000 Seep-SRB1a sp. 1 (str. 013792055) Syntrophophila gen. nov. sp. 1 3174343 97.42 4.03 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__B13-G4 sp013792055 Groundwater from Olkiluoto, Finland [ 153 ] JAJSZT000000000 Seep-SRB1a sp. 2 (str. CR9063A) Syntrophophila gen. nov. sp. 2 3859851 99.35 2.15 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Cold seep, Costa Rica Margin This study JAJSZU000000000 Seep-SRB1a sp. 2 (str. CR9063B) Syntrophophila gen. nov. sp. 2 3484925 87.56 1.94 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Cold seep, Costa Rica Margin This study JAJSZV000000000 Seep-SRB1a sp. 2 (str. CR9063C) Syntrophophila gen. nov. sp. 2 2980801 64.05 0.00 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Cold seep, Costa Rica Margin This study JAAXOL000000000 Seep-SRB1a sp. 3 (str. 014237195) Syntrophophila gen. nov. sp. 3 3489866 98.00 1.53 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__B13-G4 sp014237365 Cold seep, Gulf of Cadiz [ 154 ] JABZFQ000000000 Seep-SRB1a sp. 3 (str. 014237365) Syntrophophila gen. nov. sp. 3 3489866 98.00 1.53 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__B13-G4 sp014237365 Cold seep, Gulf of Cadiz [ 154 ] JAJSZO000000000 Seep-SRB1a sp. 4 (str. FWG171) Syntrophophila gen. nov. sp. 4 2505125 77.03 0.65 d__Bacteria;p__Desulfobacterota;c__Desulfobacteria;o__Desulfobacterales; f__ETH-SRB1;g__B13-G4;s__ Microbial mat, cold seep, Santa Monica Basin, California This study JAJSZP000000000 Seep-SRB1a sp. 7 (str. FWG172) Syntrophophila gen. nov. sp. 7 2570901 92.1 4.73 d__Bacteria;p__Desulfobacterota;c__Desulfobacteria;o__Desulfobacterales; f__ETH-SRB1;g__B13-G4;s__ Microbial mat, cold seep, Santa Monica Basin, California This study JAIORM000000000 Seep-SRB1a sp. 5 (str. 20073_SRB) Syntrophophila gen. nov. sp. 5 1938412 78.03 2.60 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Cold seep, Santa Monica Basin, California, USA [ 37 ] JAIORU000000000 Seep-SRB1a sp. 5 (str. 20074_SRB) Syntrophophila gen. nov. sp. 5 1678217 72.61 3.96 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Cold seep, Santa Monica Basin, California, USA [ 37 ] JAABVG000000000 Seep-SRB1a sp. 5 (str. S7142MS3) Syntrophophila gen. nov. sp. 5 2766087 88.86 2.58 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Cold seep, Santa Monica Basin, California, USA [ 25 ] JAJSZW000000000 Seep-SRB1a sp. 5 (str. SM7059A) Syntrophophila gen. nov. sp. 5 3861344 89.65 1.29 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Cold seep, Santa Monica Basin, California, USA This study QMMZ00000000 Seep-SRB1a sp. 6 (str. 003647525) Syntrophophila gen. nov. sp. 6 1690800 56.35 0.97 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__B13-G4 sp003647525 Hydrothermal sediment, Guaymas Basin, Gulf of California [ 41 ] JAFCZG000000000 Seep-SRB1a sp. 8 (str. AB_03_Bin_172) Syntrophophila gen. nov. sp. 8 5464515 52.75 9.74 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDFV000000000 Seep-SRB1a sp. 9 (str. Meg22_02_Bin_90) Syntrophophila gen. nov. sp. 9 3102898 83.52 3.87 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDJY000000000 Seep-SRB1a sp. 9 (str. Meg22_24_Bin_68) Syntrophophila gen. nov. sp. 9 2156769 62.82 5.48 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDKW000000000 Seep-SRB1a sp. 9 (str.Meg22_46_Bin_236) Syntrophophila gen. nov. sp. 9 2114727 57.68 1.68 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__Desulfobacterales; f__ETH-SRB1; g__B13-G4; s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDAE000000000 Desulfobacterales sp. AB_1215_Bin_34 * 3617733 88.46 3.59 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__C00003060; f__C00003106; g__C00003106; s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] MAXM00000000 Seep-SRB1g (str. C00003104) Desulfomellonium gen. nov. sp. 1 2209054 90.01 0.65 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__C00003060; f__C00003106; g__C00003106; s__C00003106 sp001751015 Hydrate Ridge, Oregon, USA [ 33 ] MANV00000000 Seep-SRB1g (str. C00003106) Desulfomellonium gen. nov. sp. 1 2149125 90.08 0.00 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__C00003060; f__C00003106; g__C00003106; s__C00003106 sp001751015 Hydrate Ridge, Oregon, USA [ 33 ] JAJSZX000000000 Seep-SRB1g (str. CR10073A) Desulfomellonium gen. nov. sp. 1 3489866 99.17 0.65 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__C00003060; f__C00003106; g__C00003106; s__C00003106 sp001751015 Cold seep, Costa Rica Margin This study JAJSZY000000000 Seep-SRB1g (str. CR10073B) Desulfomellonium gen. nov. sp. 1 3824408 98.71 2.15 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__C00003060; f__C00003106; g__C00003106; s__C00003106 sp001751015 Cold seep, Costa Rica Margin This study JAABVH000000000 Seep-SRB1g (str. S7142MS4) Desulfomellonium gen. nov. sp. 1 624626 36.28 0.16 d__Bacteria; p__Desulfobacterota; c__Desulfobacteria; o__C00003060; f__C00003106; g__C00003106; s__C00003106 sp001751015 Cold seep, Santa Monica Basin, California, USA [ 25 ] PQXD00000000 Seep-SRB2 sp. 1 (str. G37) Desulfomithrium gen. nov. sp. 1 3571848 93.83 0.00 d__Bacteria; p__Desulfobacterota; c__Dissulfuribacteria; o__Dissulfuribacterales; f__UBA3076; g__UBA3076; s__UBA3076 sp003194485 Hydrothermal sediment, Guaymas Basin, Gulf of California [ 24 ] JAFDGK000000000 Seep-SRB2 sp. 1 (str. Meg22_1012_Bin_255) Desulfomithrium gen. nov. sp. 1 2126229 85.79 0.6 d__Bacteria; p__Desulfobacterota; c__Dissulfuribacteria; o__Dissulfuribacterales; f__UBA3076; g__UBA3076; s__UBA3076 sp003194485 Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDIH000000000 Seep-SRB2 sp. 1 (str. Meg22_1214_Bin_80) Desulfomithrium gen. nov. sp. 1 2269757 91.15 0.6 d__Bacteria; p__Desulfobacterota; c__Dissulfuribacteria; o__Dissulfuribacterales; f__UBA3076; g__UBA3076; s__UBA3076 sp003194485 Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDIM000000000 Seep-SRB2 sp. 1 (str. Meg22_1416_Bin_176) Desulfomithrium gen. nov. sp. 1 1724888 60.53 1.98 d__Bacteria; p__Desulfobacterota; c__Dissulfuribacteria; o__Dissulfuribacterales; f__UBA3076; g__UBA3076; s__UBA3076 sp003194485 Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDIU000000000 Seep-SRB2 sp. 1 (str. Meg22_1618_Bin_165) Desulfomithrium gen. nov. sp. 1 1927125 76.83 2.68 d__Bacteria; p__Desulfobacterota; c__Dissulfuribacteria; o__Dissulfuribacterales; f__UBA3076; g__UBA3076; s__UBA3076 sp003194485 Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAJSZQ000000000 Seep-SRB2 sp. 2 (str. FWG173) Desulfomithrium gen. nov. sp. 2 2393181 87.78 0.6 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Pescadero Basin, Gulf of California This study JAJSZR000000000 Seep-SRB2 sp. 2 (str. FWG174) Desulfomithrium gen. nov. sp. 2 2622393 93.73 0.68 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Pescadero Basin, Gulf of California This study DFBQ00000000 Seep-SRB2 sp. 3 (str. 002367355) Desulfomithrium gen. nov. sp. 3 2606474 94.03 1.98 d__Bacteria; p__Desulfobacterota; c__Dissulfuribacteria; o__Dissulfuribacterales; f__UBA3076; g__UBA3076; s__UBA3076 sp002367355 Coal Oil Point, Santa Barbara, California, USA [ 44 ] PQXE00000000 Seep-SRB2 sp. 4 (str. E20) Desulfomithrium gen. nov. sp. 4 2549842 94.62 1.79 d__Bacteria; p__Desulfobacterota; c__Dissulfuribacteria; o__Dissulfuribacterales; f__UBA3076; g__UBA3076; s__UBA3076 sp003194495 Marine sediment, Elba, Italy [ 24 ] QNAY00000000 Seep-SRB2 sp. 5 (str. 003645605) Desulfomithrium gen. nov. sp. 5 1911994 80.95 2.13 d__Bacteria; p__Desulfobacterota; c__Dissulfuribacteria; o__Dissulfuribacterales; f__UBA3076; g__UBA3076; s__UBA3076 sp003645605 Hydrothermal sediment, Guaymas Basin, Gulf of California [ 41 ] JAFDFS000000000 Seep-SRB2 sp. 6 (str. Meg22_02_Bin_69) Desulfomithrium gen. nov. sp. 6 1600510 70.11 2.7 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] LQBF00000000 Seep-SRB2 sp. 7 (str. ML8_D) Desulfomithrium gen. nov. sp. 7 3527240 99.38 12.95 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Mahoney Lake, Canada, British Columbia [ 42 ] JAFDGC000000000 Seep-SRB2 sp. 8 (str. Meg19_1012_Bin_147) Desulfomithrium gen. nov. sp. 8 2934091 90.85 1.19 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDGV000000000 Seep-SRB2 sp. 8 (str.Meg22_1012_Bin_335) Desulfomithrium gen. nov. sp. 8 2310472 95.49 1.19 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDIE000000000 Seep-SRB2 sp. 8 (str. Meg22_1214_Bin_60) Desulfomithrium gen. nov. sp. 8 2351303 94.32 0.73 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDIO000000000 Seep-SRB2 sp. 8 (str. Meg22_1416_Bin_56) Desulfomithrium gen. nov. sp. 8 2557541 77.56 5.08 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDIR000000000 Seep-SRB2 sp. 8 (str. Meg22_1618_Bin_149) Desulfomithrium gen. nov. sp. 8 2785779 95.51 3.08 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDLI000000000 Seep-SRB2 sp. 8 (str. Meg22_46_Bin_87) Desulfomithrium gen. nov. sp. 8 2368215 91.41 0.15 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] JAFDLJ000000000 Seep-SRB2 sp. 8 (str. Meg22_810_Bin_10) Desulfomithrium gen. nov. sp. 8 2635042 91.64 1.92 d__Bacteria;p__Desulfobacterota;c__Dissulfuribacteria;o__Dissulfuribacterales; f__UBA3076;g__UBA3076;s__ Hydrothermal sediment, Guaymas Basin, Gulf of California [ 42 ] *These species are closely related to the syntrophic partner but, they are not known to be partners of ANME. ANME, anaerobic methanotrophic archaea; SRB, sulfate-reducing bacteria. 10.1371/journal.pbio.3002292.g001 Fig 1 Taxonomic diversity of syntropic SRB. A concatenated gene tree of 71 ribosomal proteins from all the Desulfobacterota genomes within the GTDB database release 95 [ 38 ] was made using Anvi’o [ 45 ]. Genomes from the genus Shewanella were used as outgroup. Within this tree, the 4 most common lineages of the syntrophic partners of ANME—Seep-SRB1a, Seep-SRB1g, Seep-SRB2, and Hot-Seep1 are highlighted. While Seep-SRB1a is a genus within the order Desulfobacterales, Seep-SRB1g and Seep-SRB1c together appear to form a closely related order-level taxonomic clade within the class Desulfobacteria. Seep-SRB2 is a genus within the order Dissulfuribacterales while Hot-Seep1 is its own species within the order Desulfofervidales. The proposed type strains are identified on the tree in white with a white asterisk adjacent to the label. The list of genomes used for the generation of concatenated gene tree is listed in S1 Table and the tree is made available as a newick file in S2 Data . ANME, anaerobic methanotrophic archaea; SRB, sulfate-reducing bacteria. Details for the phylogenetic placement of each of these clades using 16S rRNA phylogeny, concatenated ribosomal protein phylogeny, and the Genome Taxonomy Database are provided in Materials and methods and S1 – S3 Figs. HotSeep-1 is a species within the order Desulfofervidales, an order that is largely associated with thermophilic environments (with 1 exception, Desulfofervidales sp. DG-60 was sequenced from the White Oak Estuary [ 46 ]). Members of HotSeep-1 are the best characterized members of this order and are known to be syntrophic partners to thermophilic clades of methane-oxidizing ANME-1 [ 14 , 24 ] as well as alkane-oxidizing archaeal relatives “ Candidatus Syntrophoarchaeum butanivorans,” “ Candidatus Syntrophoarchaeum caldarius” [ 27 ], and ethane-oxidizing “ Candidatus Ethanoperedens thermophilum” [ 40 ]. Seep-SRB2 is a genus-level clade within the order Dissulfuribacterales [ 47 – 49 ] and class Dissulfuribacteria. Dissulfuribacterales include the genera Dissulfuribacter and Dissulfurirhabdus [ 47 – 49 ], which are chemolithoautotrophs associated with sulfur disproportionation. Seep-SRB1g is a species level clade which groups within a taxonomic order that also includes Seep-SRB1c ( Fig 1 and Table 1 ). This order falls within the class Desulfobacteria along with the sister order Desulfobacterales. Like the Desulfofervidales, the order with Seep-SRB1g is poorly characterized, yet its most well-described members are the Seep-SRB1g that are obligate syntrophic partners of ANME, accepting electrons from the archaeal partner to reduce sulfate [ 13 , 33 ]. Seep-SRB1a is a genus-level clade that along with the genus Eth-SRB1 forms a distinct family within the order Desulfobacterales (Figs 1 and S1 and S2 Table ). Many of the well-characterized members of Desulfobacterales such as Desulfococcus oleovorans , Desulfobacter hydrogenophilus , Desulfosarcina BuS5 are known as hydrogenotrophs and hydrocarbon degraders [ 50 – 52 ]. The nearest evolutionary relative of Seep-SRB1a are the Eth-SRB1 first characterized as a syntrophic partner of ethane-degrading archaea [ 29 ]. Seep-SRB1a and Seep-SRB1g are often described as Seep-SRB1 [ 23 , 39 ], a historical name that refers to a polyphyletic clade including SRB that are not partners of ANME. In order to make our analysis more accurate, and to aid future classification of syntrophic SRB, we have been careful to differentiate the different Seep-SRB1 clades with curated genomes and representative trees of 16S rRNA and ribosomal proteins ( Table 1 and S2 Fig ). Each of the 4 syntrophic SRB clades have evolved from taxonomically divergent ancestors with different metabolic capabilities. While the adaptation to a syntrophic partnership with ANME appears to have been convergently evolved in these clades, their evolutionary trajectories are likely to be different. Species diversity within each of these clades was inferred by calculating the average nucleotide identity (ANI) ( S1 Fig ) and 16S rRNA sequence similarity ( S2 Table ) between different organisms that belong to each clade, using a 95% ANI value and 98.65% similarity in 16S rRNA as cut-offs to delineate different species. Partnership associations, as identified in previous research by our group and others, by FISH [ 23 , 24 , 32 ], magneto-FISH [ 53 ] or FACS sorting [ 36 ], and single-aggregate sequencing [ 37 ] are depicted in Figs 1 and S1 with further details provided in S1 Text . Briefly, HotSeep-1 has been shown to associate with ANME-1 [ 24 ] and other archaea as described above, Seep-SRB2 associate with ANME-2c and ANME-1 [ 24 ], SeepSRB1g appears to specifically partner ANME-2b [ 13 ] while Seep-SRB1a partners ANME-2a and ANME-2c. All the genomes of Seep-SRB1g in our curated database belong to 1 species-level clade and thus far, have been shown to partner only ANME-2b [ 13 ]. In contrast, there is greater species diversity within the clades that are known to partner more than 1 clade of ANME, Seep-SRB2 and Seep-SRB1a. Whether this diversification is driven by adaptation to partnerships with multiple ANME clades remains to be seen. This pattern is also not consistent with HotSeep-1, a species-level clade that partners multiple archaeal species. A better understanding of the physiological basis for syntrophic partnership formation in each of these clades will provide a framework to understand their unique diversification trajectories. Comparative genome analysis of syntrophic SRB To develop insight into the adaptation of SRB to syntrophic partnerships with ANME, we used a comparative genomics analysis approach to (1) identify the unique features of known syntrophic SRB partners relative to their closest non-syntrophic relatives; and (2) compare the physiological traits that define the diversity within a given clade of syntrophic partner bacteria. For our first objective, we placed the metabolic traits of SRB into the phylogenetic context of the Desulfobacterota phylum, correlating the presence or absence of a physiological trait within the context of genus, family, and order level context of each syntrophic SRB clade. As an example, we demonstrate that the multiheme cytochrome conduit [ 33 ] implicated in DIET between ANME and SRB is rare in non-syntrophic Desulfobacterota suggesting that this trait is part of a required adaptation for this syntrophic relationship ( Fig 2 ). 10.1371/journal.pbio.3002292.g002 Fig 2 Gene organization and distribution of the putative cluster implicated in DIET from syntrophic SRB. (a) The syntenic blocks of genes implicated in DIET including the putative EET conduit OetABI and the operon encoding for OmcKL from HotSeep-1 ( Candidatus Desulfofervidus auxilii). (b) A model of the putative EET within syntrophic SRB. ANME electrons are likely to be accepted by 1 of 3 putative nanowires formed by multiheme cytochromes homologous to OmcX, OmcS, and a cytochrome we named Apc2a. The electrons from this nanowire would then be transferred to the porin:multiheme cytochrome c conduits formed by OetABI or OmcKL and ultimately to different periplasmic cytochromes c . (c) The distribution of the putative DIET cluster in the phylum Desulfobacterota is mapped onto a whole genome phylogenetic tree of Desulfobacterota represented in Fig 1 based on the presence of OmcX, OetI, OetB, and OmcS. This cluster is not widely found except in the orders Desulfuromonadales and Geobacterales and the classes Desulfobulbia and Thermodesulfobacteria. ANME, anaerobic methanotrophic archaea; DIET, direct interspecies electron transfer; EET, extracellular electron transfer; SRB, sulfate-reducing bacteria. We also investigated the physiological differences between the species of each syntrophic SRB clade. Two of the syntrophic SRB clades, Seep-SRB1g and HotSeep-1 have low species diversity, while the clades Seep-SRB2 and Seep-SRB1a contain multiple species. To better understand the genomic features underlying this diversity, we performed a comparative analysis of species within the Seep-SRB1a and Seep-SRB2 to identify conserved genes across the clade and species-specific genes. A detailed description of the analysis methods is available in Materials and methods and Supporting information ( S5 and S6 Figs and S3 and S4 Tables). For this comparative analysis, we primarily focused on pathways that are predicted to be important for the syntrophic interactions between ANME and SRB. In the following section, we describe the pathways within the syntrophic SRB in greater detail and their significance for a syntrophic lifestyle—extracellular electron transfer, inner membrane-bound electron transport chain, electron bifurcation, carbon fixation, nutrient sharing, biofilm formation, cell adhesion, and partner identification. Lastly, we explicitly compare the losses and gains of the genes encoding for the above pathways across the syntrophic SRB and infer the evolutionary trajectory of adaptation towards a syntrophic partnership. 1. Respiratory pathways in the 4 syntrophic SRB clades demonstrate significant metabolic flexibility The respiratory pathways in syntrophic SRB are defined by the necessity of ANME to transfer the electrons derived from methane oxidation to SRB. These electrons are then transferred across the outer membrane to periplasmic electron carriers. These periplasmic electron carriers donate electrons to inner membrane complexes and ultimately, to the core sulfate reduction pathway. Some of the electrons are also used for assimilatory pathways such as carbon fixation. Accordingly, our analysis of the respiratory pathways is split into a description of the pathways for interspecies electron transfer, electron transfer across the inner membrane, and carbon fixation pathways. 1.1 Multiple pathways exist for interspecies electron transfer between ANME and syntrophic SRB The dominant mechanism of interspecies electron transfer between ANME and SRB was proposed to be DIET. This hypothesis is supported by the presence of multiheme cytochromes in genomes of ANME-2a, 2b, and 2c [ 10 ], the presence of nanowire-like structures that extend between ANME-1 and its partners Hot-Seep1 [ 9 ] and Seep-SRB2 [ 24 ], and the presence of hemes in the extracellular space between archaeal and bacterial cells in ANME-SRB aggregates [ 10 , 24 ]. This hypothesis was also supported by the presence of a putative large multiheme cytochrome:porin type conduit, analogous to the conduits in Geobacter sp. [ 54 ] and other gram-negative bacteria that have been shown to participate in EET [ 54 ], in Seep-SRB1g [ 33 ], Seep-SRB2 [ 24 ], and Hot-Seep-1 [ 9 ]. Our analysis of a more comprehensive dataset of syntrophic bacterial genomes confirms the presence of this porin:cytochrome c conduit in all the 4 syntrophic bacterial clades studied ( S5 Table ). Henceforth, we refer to this as the as the ( O uter-membrane bound e xtracellular electron t ransfer) or OetI-type conduit. This conduit includes a periplasmic cytochrome c (OetA), an outer-membrane porin (OetI), and extracellular facing cytochrome c lipoprotein (OetB) (Figs 2B and 3C ). The OetI-type conduit was first identified in G . sulfurreducens and is expressed when a Geobacter mutant of omcB is grown on Fe(III) oxide [ 55 ]. The oetABI cassette is found in all 4 syntrophic SRB clades and often includes 2 or 3 other putative extracellular cytochromes c , including homologs of OmcX [ 33 ], OmcS (Supplementary alignment MSA1) and a 6-heme cytochrome that we termed apc2a ( S5 Table ). If they are not found as part of the oet cluster, they could be found elsewhere on the genome, possibly due to genomic rearrangement after acquisition of the cassette ( S6 and S7 Tables). The omcX and omcS-like genes in the oet gene cassette are often found in an analogous position to omcS and omcT in G. sulfurreducens ( Fig 2 ). Based on the homology of one of the cytochromes to OmcS, which polymerizes to form long and highly conductive filaments that facilitate EET in Geobacter [ 56 ], we propose that the extracellular cytochromes c in this gene cassette perform a similar function, forming filaments that accept electrons from ANME. This is consistent with heme staining of the intercellular space between ANME and SRB, and the observation of filaments that connect the partners [ 10 , 24 ]. This is also consistent with the fact that different extracellular cytochromes are among the most highly expressed proteins in the syntrophic SRB: ANME-1/Seep-SRB2 [ 24 ] (OmcX, OmcS-like, and Apc2a), ANME-1/HotSeep-1 [ 24 ] (OmcX and OmcS-like), ANME-2c/Seep-SRB2 [ 24 ] (OmcX) aggregates and ANME-2a/Seep-SRB1a [ 25 ] (OmcX, OmcS-like). The presence of multiple copies of these putative filament-forming proteins in the syntrophic SRB genomes is indicative of their importance to the physiology of syntrophic SRB. The mechanism of electron transfer from extracellular cytochrome filaments to the interior of the cells in Geobacter is not well understood. However, a porin:cytochrome c conduit is always expressed under the same conditions as a cytochrome c containing filament in Geobacter (omcS along with extEFG or omcABC under Fe (III) oxide reducing conditions and omcZ along with extABCD during growth on an electrode [ 57 ]) and in ANME-SRB consortia (OmcS/OmcX with OetABI or OmcKL). These findings suggest that each cytochrome c filament could act in concert with a porin:cytochrome c conduit ( Fig 2 ) to transfer electrons from the extracellular space to the periplasm. 10.1371/journal.pbio.3002292.g003 Fig 3 Summary of the different electron transport chains in syntrophic SRB. The various respiratory proteins essential for the electron transport chain within the syntrophic SRB are identified and marked within their predicted cellular compartments. Filled circles indicate their presence in each of the 4 syntrophic sulfate-reducing bacterial clades, HotSeep-1 (red), Seep-SRB2 (orange), Seep-SRB1g (blue), and Seep-SRB1a (purple). The 2 typical acceptors of electrons transferred across the inner membrane, quinols (QH 2 ) and DsrC, are indicated in shaded circles. These are the 2 nodes which much of the respiratory flexibility of the syntrophic SRB revolves around. SRB, sulfate-reducing bacteria. While oetABI is conserved in all 4 syntrophic SRB clades, there are 2 other putative porin:cytochrome c conduits in syntrophic SRB. A porin (HS1_RS02765) and extracellular cytochrome c (HS1_RS02760) homologous to OmcL and OmcK from G . sulfurreducens is found in HotSeep-1 ( S6 and S7 Tables) and expressed at a 4-fold higher level than the oetABI conduit [ 24 ]. OmcK and OmcL were also up-regulated in G . sulfurreducens when it is grown on hematite and magnetite [ 58 ]. There is no gene encoding a periplasmic cytochrome c adjacent to these genes and this is unusual for previously characterized EET conduits but, given the large number of periplasmic cytochromes in HotSeep-1, it is conceivable that another cytochrome c interacts with the OmcL/K homologs. This conduit is also found in Seep-SRB2 sp. 1, 2, 7, and 8 but does not appear to be expressed as highly as the OetABI in the ANME-1/Seep-SRB2 consortia [ 24 ]. A different putative conduit including the porin, extracellular, and periplasmic cytochromes c is present in the Seep-SRB1g genomes (LWX52_07950- LWX52_07960) ( S6 and S7 Tables). This conduit does not have identifiable homologs in Geobacter . The presence of multiple porin:cytochrome c conduits in the syntrophic partners suggests some flexibility in use of electron donors, possibly from different syntrophic partners. For HotSeep-1, this observation is consistent with its ability to form partnerships with both methane and other alkane-oxidizing archaea [ 28 ]. The role of the second conduit is less clear in Seep-SRB1g which to date has only been shown to partner with ANME-2b [ 13 ]. Future investigation of the multiple syntrophic SRB EET pathways and the potential respiratory flexibility it affords to their partner archaea using transcriptomics, proteomics and possibly heterologous expression methods will further expand our understanding of electron transfer in these diverse consortia. While DIET is believed to be the dominant mechanism of syntrophic coupling between the ANME and SRB partners, the potential to use diffusible intermediates such as formate and hydrogen exists in some genomes of syntrophic SRB. Hydrogenases are present in HotSeep-1, which can grow without ANME using hydrogen as an electron donor [ 32 ]. We also identified periplasmic hydrogenases in Seep-SRB1a sp. 1, 5, and 8 ( S7 Table ) that suggest that these organisms could use hydrogen as an electron donor. However, in Seep-SRB1a, these hydrogenases are expressed at low levels (less than a 20th of the levels of DsrB) in the ANME-2a/Seep-SRB1a consortia [ 25 ]. Further, previous experiments showed that the addition of hydrogen to ANME-2/SRB consortia did not inhibit anaerobic oxidation of methane suggesting that hydrogen is not the predominant agent of electron transfer between ANME and SRB [ 35 , 59 ]. Perhaps, hydrogenases are used by Seep-SRB1a to scavenge small amounts of hydrogen from the environment. While membrane bound and periplasmic hydrogenases are present in non-syntrophic Seep-SRB1c ( S7 Table ), no hydrogenases are found in the syntrophic relative of Seep-SRB1c and ANME partner, Seep-SRB1g. Similarly, periplasmic hydrogenases are present in Dissulfuribacteriales and absent in Seep-SRB2 (one exception in 18 genomes), suggesting that in both these partners, the loss of periplasmic hydrogenases is part of the adaptation to their syntrophic partnership with ANME. We also identified periplasmic formate dehydrogenases in Seep-SRB1g and Seep-SRB1a sp. 2, 3, 8, 9 ( S7 Table ). The periplasmic formate dehydrogenase from Seep-SRB1g is expressed in the environmental proteome at Santa Monica Mounds [ 33 ], but no transcripts from the formate dehydrogenases of Seep-SRB1a were recovered in the ANME-2a/Seep-SRB1a incubations [ 25 ]. It is possible that these syntrophic SRB scavenge formate from the environment. Alternatively, a recent paper found a hybrid of electron transfer by DIET and by diffusible intermediates (mediated interspecies electron transfer or MIET) to be energetically favorable [ 60 ]. In this model, the bulk of electrons would still be transferred by DIET, but up to 10% of electrons could be shared by MIET via formate [ 60 ], an intermediate suggested in earlier studies [ 35 , 59 ]. This might be possible in ANME/SRB consortia with HotSeep-1, some species of Seep-SRB1a and Seep-SRB1g, but not in ANME/Seep-SRB2 consortia. The absence of periplasmic formate dehydrogenases and hydrogenases in Seep-SRB2 as previously observed [ 24 ] is also true in our expanded dataset. If a diffusive intermediate should play a role in mediating electron transfer between ANME-2c or ANME-1 and Seep-SRB2, it is not likely to be formate or hydrogen. 1.2 Pathways for electron transfer across the inner membrane vary in different syntrophic SRB clades Multiheme cytochromes c in SRB are known to mediate diverse modes of electron transfer from different electron donors to a conserved sulfate reduction pathway [ 58 ]. There is significant variety in the number and types of cytochromes c present in SRB from the phylum Desulfobacterota [ 61 ] and an even greater number of large cytochromes is present in syntrophic SRB [ 24 , 33 ]. To explore the potential for different routes of electron transfer, we performed an analysis of all cytochromes c containing 4 hemes or more from the genomes of syntrophic SRB (see Materials and methods ) and identified at least 27 different types of cytochromes c . We split these cytochromes c into those predicted to be involved in EET, those that act as periplasmic electron carriers and those that are components of protein complexes involved in electron transfer across the inner membrane ( S6 Table ). Conserved across the syntrophic SRB partners of ANME were the cytochromes forming the core components of the EET pathway—OetA, OetB, OmcX and OmcS-like and Apc2a extracellular cytochromes, and 2 periplasmic cytochromes of the types, TpI c 3 [ 62 ] and cytochrome c 554 [ 61 , 63 ]. Beyond the conserved periplasmic cytochromes c , TpI c 3 and cytochrome c 554 , there are also cytochromes binding 7–8 hemes that are unique to different SRB clades ( S6 Table ). These include a homolog of ExtKL [ 64 ] from G . sulfurreducens that is highly expressed in Seep-SRB2 spp. 1 and 4 during growth in a syntrophic partnership with ANME [ 24 ] and a homolog of ExtA from G . sulfurreducens [ 54 ] protein expressed in the ANME-2a/Seep-SRB1a consortia [ 25 ]. Previous research has suggested that the tetraheme cytochromes c are not selective as electron carriers and play a role in transferring electrons to multiple different protein complexes [ 65 ]. It is possible that these larger 7–8 heme binding cytochromes c have a more specific binding partner. Both the ExtKL and ExtA-like proteins are very similar (over 45% sequence similarity) to their homologs in Desulfuromonadales. Since the OetI-type conduit is also likely transferred from this order, they might act as binding partners to the OetI-type conduit transferring electrons to the periplasmic cytochromes c . In SRB, the electrons from periplasmic electron donors (reduced by DIET or MIET) are delivered through inner membrane bound complexes to quinones or directly to the heterodisulfide DsrC in the cytoplasm via transmembrane electron transfer [ 61 ] ( Fig 3 ). The electrons from quinones or DsrC are ultimately used for the sulfate reduction pathway (including SatA, AprAB, and DsrAB) [ 24 , 33 , 61 ]. Two conserved protein complexes are always found along with this pathway—the Qmo complexes transfers electrons from reduced quinones to AprAB and the DsrMKJOP complexes transfers electrons from quinones to DsrC and through DsrC to DsrAB. Since both these complexes use electrons from reduced quinones, the source of reduced quinones in the inner membrane is critical to different sulfate respiration pathways. The quinol reducing complexes and complexes that reduce DsrC provide respiratory flexibility to SRB. We also note here that the reduction of AprAB coupled to the oxidation of menaquinone is expected to be endergonic. There is a proposal that QmoABC might function through flavin-based electron confurcation (FBEC), using electrons from reduced quinones and a second electron donor such as ferredoxin to reduced AprAB [ 66 ]. Since, it is not clear what the electron donor is likely to be, we do not explicitly consider this reaction in our analysis. A summary of all the putative complexes that are involved in the electron transport chains of the 4 syntrophic SRB is visualized in Fig 3 to detail how electron transport pathways vary among the clades. A more detailed list of complexes present is found in S6 – S8 Tables. The respiratory pathways in HotSeep-1, Seep-SRB1a, and Seep-SRB1g are broadly similar in structure and are predicted to use the Qrc complex to transfer periplasmic electrons to the quinone pool and both DsrMKJOP and Tmc to reduce cytoplasmic DsrC. Their pathways are analogous to the respiratory pathways in Desulfovibrio alaskensis [ 62 , 67 ]. Curiously, the Tmc complex in Seep-SRB1a and Seep-SRB1g are divergent from Tmc in non-syntrophic SRB ( S8 Fig ) and TmcA is absent in the operons encoding for Tmc. This absence suggests that Tmc has been adapted to use a different electron donor than in Desulfovibrio vulgaris [ 67 ] and is consistent with the fact that the electron donor for Seep-SRB1a and Seep-SRB1g is not hydrogen or formate but, electrons from anaerobic methanotrophic archaea. It is not clear why Tmc is more divergent than DsrMKJOP in syntrophic SRB compared to their evolutionary neighbors since they are both likely important for DsrC reduction and are equally well expressed under methane-oxidizing conditions [ 24 , 25 ]. Qrc is known to be important for energy conservation in this respiratory pathway. Protons are translocated by Qrc from the cytoplasmic side to the periplasmic active side. This movement of charges across the membrane leads to the generation of proton motive force (pmf) that can be utilized by ATP synthase to generate ATP [ 68 ]. In Seep-SRB1a, Seep-SRB1g, and HotSeep-1, Qrc likely acts with the conserved DsrMKJOP and QmoABC to generate pmf. While purified Dsr and Qmo have not yet been shown to be electrogenic, it is expected that DsrMKJOP [ 69 ] and QmoABC [ 69 , 70 ] might be able to generate pmf by charge translocation. In Seep-SRB2, Qrc is absent and we hypothesize that CbcBA, a protein complex that appears to be horizontally transferred from the Desulfuromonadales (Figs 3 and 4 ), mediates electron transfer between periplasmic cytochromes c and quinones [ 71 ]. This is supported by the fact that CbcBA is highly expressed during AOM between ANME-1/Seep-SRB2 and ANME-2c/Seep-SRB2 [ 24 ]. In Geobacter sulfurreducens , which also does not have Qrc this cytoplasmic membrane-bound oxidoreductase is expressed during growth on Fe(III) at low potential and is important for iron reduction and growth on electrodes at redox potentials less than −0.21 mV [ 71 ]. During AOM, the CbcBA protein in Seep-SRB2 is predicted to run in the reverse direction, reducing quinols using electrons from DIET electrons supplied by ANME as opposed to functioning in the metal reducing direction. While the reversibility of this complex has not been biochemically established, the high levels of expression of this complex suggest that this is likely functional in the electron transport chain of Seep-SRB2. It is not clear what the likely site of energetic coupling is within the Seep-SRB2 respiratory chain. In the absence of the Qrc complex, the most likely mechanism for energetic coupling might exist through the action of a Q-loop mechanism [ 69 ]. In this mechanism, energy is conserved by the combined action of 2 protein complexes that reduce and oxidize quinols, leading to the uptake and release of protons on opposite sides of the cytoplasmic membrane. The Q-loop mechanism in Seep-SRB2 would likely involve CbcBA and a quinol oxidizing complex such as Qmo. 10.1371/journal.pbio.3002292.g004 Fig 4 cbcBA as an example of horizontal gene transfer events into Seep-SRB2. We demonstrate one example of an important gene transfer event involving a metabolic gene. The function of cbcBA is essential for the central respiratory pathway in Seep-SRB2 and this gene was acquired by horizontal gene transfer. (A) The presence of CymA, CbcL, CbcAB, and NetBCD, commonly used electron donors to the EET conduits in Shewanella and Geobacter are mapped on to the classes Thermodesulfobacteria and Desulfobulbia. (B) CbcB protein sequences were aligned using MUSCLE [ 72 ] and then a phylogenetic tree was inferred using IQ-Tree2 [ 73 ]. The CbcB sequences from Seep-SRB2 are highlighted in orange. The phylogenetic tree is available in newick format in S2 Data . EET, extracellular electron transfer; SRB, sulfate-reducing bacteria. In addition to the most likely pathways of electron transfer in the syntrophic SRB, as established using transcriptomic data on ANME/SRB partnership [ 24 , 25 ] (Figs 2 and 3 ), other inner membrane complexes exist in these genomes that may provide additional respiratory flexibility. HotSeep-1 genomes contain a complex that involves an HdrA subunit and a protein that also binds hemes c and contains a CCG domain similar to that found in HdrB and TmcB predicted to interact with DsrC [ 70 , 74 ]. This complex a putative cytochrome c oxidoreductase containing a CCG domain, would likely transfer electrons from cytochromes c to the DsrC (AMM42179.1-AMM42180.1) or perhaps to a ferredoxin. The presence of HdrA might indicate a role in electron bifurcation by this complex. It is highly expressed during methane oxidation conditions in the ANME-1/HotSeep-1 consortia to a fifth of the level of the Tmc complex that would play a similar role in electron transfer [ 24 ]. In some Seep-SRB1a and Seep-SRB1g genomes, there is a homolog of Cbc6 (LWX51_14670- LWX51_14685) identified in Geobacter [ 75 ] and implicated in electron transfer from periplasmic cytochromes c to the quinol pool. A NapC/NirT homolog [ 76 ] was conserved in Seep-SRB1g (OEU53943.1-OEU53944.1) and some Seep-SRB2, and another conserved complex that includes a cytochrome c and ruberythrin (AMM39991.1-AMM39993.1) is present in Seep-SRB1g and HotSeep-1. Further research is needed to test whether there are conditions under which these complexes are expressed. Our analysis indicates some degree of respiratory specialization in the syntrophic SRB genomes such as the loss of hydrogenases in Seep-SRB1g and Seep-SRB2 compared to their nearest evolutionary neighbors, suggesting an adaptation towards a partnership with ANME. However, considerable respiratory flexibility still exists within the genomes of these syntrophic partners as is suggested by the presence of the formate dehydrogenases in Seep-SRB1g and Seep-SRB1a, multiple EET conduits in HotSeep-1 and Seep-SRB2 and multiple inner membrane complexes in Seep-SRB1a and HotSeep-1. 1.3 Cytoplasmic redox reactions, electron bifurcation, and carbon fixation pathways The electron transport chain outlined above would transfer electrons from periplasmic cytochromes c to the cytoplasmic electron carrier DsrC or directly to the sulfate reduction pathway. However, the electron donors for carbon or nitrogen fixation are typically NADH, NADPH, or ferredoxin [ 77 ]. The reduction of NADH, NADPH, or ferredoxin could happen through the transfer of electrons from DsrC [ 78 ] or through the interconversion of electrons between cytoplasmic electron carriers through the dissipation of pmf or through the action of electron bifurcating complexes [ 79 ]. The transfer of electrons from DsrC to these reductants likely happens through the action of protein complexes like Flx-Hdr that can oxidize 2 molecules of NADH to reduce 1 molecule of ferredoxin and 1 molecule of DsrC. Electrons from NADH and ferredoxin can also be exchanged with the dissipation or generation of sodium motive force using membrane-bound Rnf and Mrp [ 79 – 81 ] in Seep-SRB1a, Seep-SRB1g, and Hot-Seep1. In marine environments, the naturally occurring sodium gradient can be used to generate ferredoxin from NADH or vice versa using the Rnf complex, while the Na + /H + antiporter, Mrp, can transport Na + or H + in response to the action of Rnf [ 82 ]. The ferredoxin generated from this process can then be used for assimilatory pathways. In Seep-SRB2, which does not contain Rnf or Mrp ( S9 Fig ), the NADH needed for carbon fixation is likely obtained through the oxidation of quinol by complex I and the dissipation of pmf [ 83 ]. In addition to Complex I or Rnf and Mrp, there are additional cytoplasmic protein complexes that can recycle reducing equivalents between DsrC, ferredoxin, and NADH such as the electron bifurcating Flx-Hdr [ 70 , 78 ]. Several putative oxidoreductase complexes in the syntrophic SRB genomes are compiled in S7 Table and S10 and S11 Figs. Syntrophic sulfate-reducing members of the Seep-SRB1a, Seep-SRB1g, and Seep-SRB2 have been shown to fix carbon using the Wood–Ljungdahl pathway, while organisms of the clade HotSeep-1 partnering with ANME-1 are predicted to fix carbon using the reductive tricarboxylic acid cycle (rTCA) [ 24 , 33 , 77 ]. Analysis of gene synteny for a number of Seep-SRB1a, Seep-SRB1g, and Seep-SRB2 MAGs uncovered a number of heterodisulfide (HdrA) subunits and HdrABC adjacent to enzymes involved in the Wood–Ljungdahl pathway ( S10 Fig ). These subunits are typically implicated in flavin-based electron bifurcating reactions utilizing ferredoxins or heterodisulfides and NADH [ 79 ]. Specifically, Seep-SRB1g has an HdrABC adjacent to metF that is predicted to encode for a putative metF-HdrABC, performing the reduction of methylene tetrahydrofolate coupled to the endergonic reduction of ferredoxin to NADH, the same reaction as the bifurcating metFV-HdrABC described below. In Seep-SRB1g, there are also 2 copies of HdrABC next to each other whose function requires further analysis ( S10 Fig ). These complexes are absent in the related group Seep-SRB1c, a lineage which has not yet been found in physical association with ANME ( S10 Fig ). The presence of electron bifurcation machinery in the carbon fixation pathways within several syntrophic SRB lineages, suggests that they are optimized to conserve energy ( S10 Fig ). This is reminiscent of the MetFV-HdrABC in the acetogen Moorella thermoacetica [ 79 ] in which the NADH-dependent methylene tetrahydrofolate reduction within the central metabolic pathway is coupled to the endergonic reduction of ferredoxin by NADH, allowing for the recycling of reducing equivalents. Members of the Seep-SRB1g also have a formate dehydrogenase (fdhF2) subunit adjacent to nfnB, the bifurcating subunit of NfnAB, which performs the NADPH-dependent reduction of ferredoxin ( S11 Fig ). This complex is predicted to function as an additional bifurcating enzyme that would allow for the recycling of NADPH electrons. In addition, HotSeep-1, Seep-SRB2, and Seep-SRB1g appear to have homologs of electron transfer flavoproteins, etfAB, that are expected to be electron bifurcating. These homologs of etfAB cluster with the previously identified bifurcating etfAB and possess the same sequence motif that was previously shown to correlate with the electron bifurcating etfAB [ 84 ] ( S7 Table ). While the capability of electron bifurcation by these enzyme complexes needs to be biochemically confirmed, the possibility of a high number of bifurcating complexes, especially those connected to the carbon fixation pathway, in the genomes of syntrophic SRB partners of ANME is compelling. It could be argued that this is a natural adaptation to growth in very low-energy environments or to low-energy metabolism. In fact, some of these complexes are present in non-syntrophic bacteria of the order Desulfofervidales and genus Eth-SRB1. These adaptations could provide an additional energetic benefit for the syntrophic lifestyle, itself an adaptation to low-energy environments. 2. Cobalamin auxotrophy and nutrient sharing in syntrophic SRB Research on the AOM symbiosis has focused heavily on the nature of the syntrophic intermediates shared between ANME and SRB [ 9 , 10 , 12 , 85 , 86 ]. We currently have an incomplete understanding of the scope of other potential metabolic interdependencies within this long-standing symbiosis. Prior experimental research has demonstrated the potential for nitrogen fixation and exchange in AOM consortia under certain environmental circumstances [ 11 , 13 , 87 , 88 ], and in other energy limited anaerobic syntrophies between bacteria and archaea, amino acid auxotrophies are common [ 89 – 91 ]. Comparative analysis of MAGs from several lineages of ANME [ 14 ] as well as a subset of syntrophic sulfate-reducing bacterial partners [ 33 ] lacked evidence for specific loss of pathways used in amino acid synthesis, and our expanded analysis of SRB here is consistent with these earlier studies. Interestingly, comparative analysis of specific pairings of ANME and their SRB partners revealed the possibility for cobalamin dependency and exchange. Cobamides, also known as the Vitamin B12-type family of cofactors, are critical for many central metabolic pathways [ 92 ]. Mechanisms for complete or partial cobamide uptake and remodeling by microorganisms found in diverse environments are common [ 92 ]. The importance of exchange of cobamide between gut bacteria and between bacteria and eukaryotes has been demonstrated [ 93 , 94 ]. In methanotrophic ANME-SRB partnerships, ANME are dependent on cobalamin as a cofactor in their central metabolic pathway and biosynthetic pathways, while Seep-SRB2, Seep-SRB1a, Seep-SRB1g also have essential cobalamin-dependent enzymes including ribonucleotide reductase, methionine synthase, and acetyl-CoA synthase ( S8 Table ). This is in contrast with the HotSeep-1 clade, which appears to have fewer cobalamin requiring enzymes and may not have an obligate dependence on vitamin B12. However, HotSeep-1 do possess homologs of BtuBCDF and CobT/CobU, genes that are used in cobamide salvage and remodeling [ 95 ] ( S8 Table ). An absence of cobalamin biosynthesis in either ANME or these 3 clades of syntrophic SRB would thus necessarily lead to a metabolic dependence on either the partner or external sources of cobalamin in the environment. We observed such a predicted metabolic dependence for Seep-SRB1a within the species Seep-SRB1a sp. 1 ( n = 1 genomes), Seep-SRB1a sp. 5 ( n = 4 genomes), Seep-SRB1a sp. 3 ( n = 2 genomes), Seep-SRB1a sp. 7 ( n = 1), and Seep-SRB1a sp. 8 ( n = 1). All these genomes are missing the anaerobic corrin ring biosynthesis pathway but, some do retain genes involved in lower ligand synthesis (BzaAB) [ 96 ] ( Fig 5 ). Additionally, recent metatranscriptomic data from an AOM incubation dominated by ANME-2a/Seep-SRB1a associated with Seep-SRB1a sp. 5 (str. SM7059A) that is missing the cobalamin biosynthesis pathways confirmed active expression of cobalamin-dependent pathways in the Seep-SRB1a including ribonucleotide reductase and acetyl-coA synthase AcsD [ 25 ], suggesting that these syntrophs must acquire cobalamin from their ANME partner or the environment. 10.1371/journal.pbio.3002292.g005 Fig 5 The loss of cobalamin biosynthesis genes in the Seep-SRB1a partners of ANME-2a. On the right, a phylogenetic tree of concatenated ribosomal proteins from all the genomes of syntrophic SRB clades—Hot-Seep1, Seep-SRB2, Seep-SRB1g, and Seep-SRB1a and related clades, Seep-SRB1c and Eth-SRB1 was made using Anvi’o [ 45 ] and made available in S2 Data . On the left, a similar concatenated protein tree (available in S2 Data ) was made for ANME genomes highlighting the clades from ANME-1, ANME-2c, ANME-2b, and ANME-2a. Lines in green and light teal are used to depict the partnerships between ANME-2c and verified species of Seep-SRB1a, and ANME-2a and verified species of Seep-SRB1a, respectively. ANME-2c genomes are not separated into those belonging to partners of Seep-SRB2 and Seep-SRB1a. The presence of genes involved in cobalamin biosynthesis and nitrogen fixation are marked in light green and light blue, respectively. The proposed type strains are bolded. ANME, anaerobic methanotrophic archaea; SRB, sulfate-reducing bacteria. Interestingly, the predicted cobalamin auxotrophy is not a uniform trait within the Seep SRB1a lineage, with cobamide biosynthesis genes present in the genomes of species Seep-SRB1a sp. 2 ( n = 3), Seep-SRB1a sp. 4 ( n = 1), and Seep-SRB1a sp. 9 ( n = 3) ( Fig 5 ). Of the 5 species missing cobalamin biosynthesis pathways, 2 are verified ANME-2a partners. Of the 4 species containing cobalamin biosynthesis pathways, one is a verified ANME-2c partner and one was sequenced from a microbial mat that contains ANME-2c and ANME-2a ( Fig 5 ). These patterns suggest that the Seep-SRB1a partners of ANME-2a developed a nutritional auxotrophy that is specific to this partnership. Future experimental work will assist with testing this predicted vitamin dependency among the ANME-2a and Seep-SRB1a and other ANME-SRB partner pairings. The ability to fix nitrogen is found in bacteria and archaea but is relatively rare among them [ 97 ]. Fixed nitrogen availability can impact the productivity of a given ecosystem. Members of the ANME-2 archaea have been demonstrated to fix nitrogen in consortia [ 11 , 12 , 88 ] and may serve as a source of fixed nitrogen for methane-based communities in deep-sea seeps [ 88 ]. We recently demonstrated that within the ANME-2b/Seep-SRB1g partnership, Seep-SRB1g bacteria can also fix nitrogen [ 13 ]. A comparison of the nitrogen fixation ability across ANME and SRB ( Fig 5 ) shows that this function is present in the genome representatives of diverse ANME and also conserved in some syntrophic bacterial partners (Seep-SRB1a and Seep-SRB1g). In the Seep-SRB1a lineage, the nitrogenase operon is retained in both ANME-2a and ANME-2c partners, contrasting the pattern observed with cobalamin synthesis. Interesting, the potential to fix nitrogen occurs in species of Seep-SRB2 that come from psychrophilic deep-sea environments (Seep-SRB2 sp. 4 and Seep-SRB2 sp. 3), while earlier branching clades of Seep-SRB2 adapted to hotter environments (Seep-SRB2 sp. 1 and 2) lack nitrogenases, hinting at potential ecophysiological adaptation to temperature ( Fig 5 ). While the ability to fix nitrogen is retained in several clades of syntrophic SRB, previous stable isotope labeling experiments have shown that ANME is the dominant nitrogen fixing partner [ 11 , 13 , 88 ]. Yet, the potential to fix nitrogen is retained in Seep-SRB1a and Seep-SRB1g members ( Fig 5 ), and in some cases, have been directly linked to N 2 fixation in the case of Seep SRB1g [ 13 ] or indirectly suggested from the recovery of nifH transcripts belonging to Seep-SRB1a and Seep-SRB1g in seep sediments [ 12 ]. These observations indicate that nitrogen sharing dynamics between ANME and SRB is likely more complicated than we have thus far observed and may correspond to differences in environment, or perhaps to specific partnership interactions that require assessment at greater taxonomic resolution. 3. Pathways related to biofilm formation and intercellular communication ANME and SRB form multicellular aggregates in which they are spatially organized in distinct and recognizable ways [ 98 ]. ANME-2a/2b/2c and ANME-3 are known to form tight aggregates with their bacterial partners [ 10 , 25 , 35 , 98 ]. Some members of ANME-1 have been observed in tightly packed consortia with SRB [ 24 ], while others some form more loose associations [ 19 , 26 , 99 , 100 ]. In these consortia, archaeal and bacterial cells are often enmeshed in an extracellular polymeric substance [ 19 , 20 , 101 ]. In large carbonate associated mats of ANME-2c and ANME-1 and SRB from the Black Sea, extractions of exopolymers consisted of 10% neutral sugars, 27% protein, and 2.3% uronic acids [ 20 ]. This composition is consistent with the roles played by mixed protein and extracellular polysaccharide networks shown to be important for the formation of conductive biofilms in Geobacter sulfurreducens [ 102 ], the formation of multicellular fruiting bodies from Myxococcus xanthus [ 103 – 105 ], and the formation of single-species [ 106 ] and polymicrobial biofilms [ 107 ]. Important and conserved features across these biofilms are structural components made up of polysaccharides, cellular extensions such as type IV pili and matrix-binding proteins such as fibronectin-containing domains [ 108 ]. Functional components of the biofilm matrix such as virulence factors in pathogens [ 109 ] and EET components [ 102 ] are variable and depend on the lifestyle of the microorganism. Guided by the molecular understanding of mechanisms and physiological adaptation to microbial growth in biofilms, we examined the genomic evidence for similar adaptations in the syntrophic SRB in consortia with ANME, focusing on structural and functional components of biofilms as well as proteins implicated in partner identification ( Fig 6 ). 10.1371/journal.pbio.3002292.g006 Fig 6 Putative physiological factors involved in ANME/SRB aggregate formation. Extracellular polysaccharides and protein complexes implicated in the formation of the extracellular matrix in ANME-SRB aggregates are visualized as cell-surface embedded or secreted. The capacity for biosynthesis of sulfated polysaccharides is present in 3 of the syntrophic SRB clades—Seep-SRB2, Seep-SRB1a, and Seep-SRB1g. Type VI secretion systems and eCISs are likely important for intercellular communication between ANME and SRB. ANME, anaerobic methanotrophic archaea; eCIS, extracellular contractile injection system; SRB, sulfate-reducing bacteria. 3.1 Multiple polysaccharide biosynthesis pathways are found in syntrophic SRB Our analysis of syntrophic SRB genomes showed the presence of multiple putative polysaccharide biosynthesis pathways in different SRB lineages including secreted extracellular polysaccharide biosynthesis pathways and capsular polysaccharide biosynthesis pathways ( S9 Table ). In particular, homologs of the pel biosynthesis pathway (PelA, PelE, PelF, and PelG), first identified in Pseudomonas aeruginosa [ 110 , 111 ] were present in almost all Seep-SRB1g and Seep-SRB1a genomes ( S12 and S13 Figs). These homologs are part of a conserved operon in these genomes which includes a transmembrane protein that could perform the same function as PelD, which along with PelE, PelF, and PelG forms the synthase component of the biosynthetic pathway and enables transport of the polysaccharide pel across the inner membrane [ 111 ]. Metatranscriptomic data confirms this operon is expressed and was significantly down-regulated when methane-oxidation by ANME-2a was decoupled from its syntrophic Seep SRB1a partner with the addition of AQDS [ 25 ]. This biosynthesis pathway is absent in the nearest evolutionary neighbors of Seep-SRB1a and Seep-SRB1g, Eth-SRB1 and Seep-SRB1c, respectively, suggesting that the presence of the pel operon could serve as a better genomic marker for syntrophic interaction with ANME-2a, ANME-2b, and ANME-2c than the presence of the oetI-type conduit. The pel operon was also detected in one of the Seep-SRB2 genomes but is not conserved across this clade. In Seep-SRB2 clades, multiple capsular polysaccharide biosynthesis pathways are conserved. This includes a neuraminic acid biosynthesis pathway, a sialic acid capsular polysaccharide widely associated with intestinal mucous glycans and used by pathogenic and commensal bacteria to evade the host immune system [ 112 ] ( S14 Fig ). These differences in polysaccharide biosynthesis pathways are likely reflected in the nature of the EPS matrix within each ANME-SRB aggregate. Members of the thermophilic HotSeep-1 syntrophic SRB also encode for multiple putative polysaccharide biosynthesis pathways, including a pathway similar to the xap pathway in G . sulfurreducens ( S15 Fig ). The role of polysaccharides in the formation of conductive extracellular matrices and in intercellular communication is just beginning to be understood but they appear to be essential to its formation. For example, the mutation of the xap polysaccharide biosynthesis pathway in G . sulfurreducens eliminated the ability of this electrogenic bacteria to reduce Fe (III) in the bacterium [ 102 ] and affected the localization of key multiheme cytochromes c OmcS and OmcZ and structure of the biofilm matrix [ 113 ], suggesting that the EPS matrix contributes a structural scaffold for the localization of the multiheme cytochromes. Similarly, the cationic polysaccharide pel in P . aeruginosa biofilms has recently been shown to play a role in binding extracellular DNA or other anionic substrates together forming tight electrostatic networks that provide strength to the extracellular matrix [ 114 ] and may offer a similar role in Seep SRB1a and 1g consortia. Based on the reported chemical composition of EPS from the Black Sea ANME-SRB biofilm [ 20 ], alongside TEM compatible staining of cytochromes c in the extracellular space between ANME and SRB [ 9 , 10 , 24 ], and the genomic evidence provided here of conserved polysaccharide biosynthesis pathways point to the existence of a conductive extracellular matrix within ANME-SRB consortia that has features similar to Geobacter biofilms [ 102 ]. While these conductive biofilms are correlated with the presence of secreted polysaccharides, the highly conserved capsular polysaccharides common in Seep SRB2 likely play a different role. In Myxococcus xanthus , the deletion of capsular polysaccharides leads to a disruption in the formation of multicellular fruiting bodies, suggesting a possible role for capsular polysaccharides in intercellular communication [ 115 ]. This is consistent with the universal role of O-antigen ligated lipopolysaccharides in cell recognition and the Seep SRB2 capsular polysaccharides may serve a similar purpose in consortia with ANME, either influencing within population interactions, or potentially mediating kin recognition. 3.2 Several putative adhesins found in syntrophic SRB are absent in free-living SRB In addition to polysaccharides, there are several conserved adhesion-related proteins present in syntrophic SRB and absent in closely related SRB that are likely important for ANME-SRB biofilm formation. These include cohesin and dockerin domain-containing proteins, similar to those previously identified in ANME [ 14 ], immunoglobulin-like domains, cell-adhesin related domain (CARDB) domains, bacterial S8 protease domains, PEB3 adhesin domains, cadherin, integrin domains, and fibronectin domains ( Fig 6 and S10 Table ). Fibronectin domains are found in the one of the cytochromes c , oetF that is likely part of the EET conduit. This domain might interact with the conductive biofilm matrix itself or serve as a partnership recognition site. PilY1 is another adhesion-related protein that appears to be important in HotSeep-1. This is a subunit of type IV pili that is known to promote surface adhesion in Pseudomonas and intercellular communication in multispecies Pseudomonas biofilms [ 116 ]. Our analysis of the SRB adhesins suggests that some adhesins are conserved across a given syntrophic clade, while others appear to be more species or partnership specific. For example, while PilY1 is conserved across Seep-SRB2, the cohesin/dockerin complexes that are conserved in Hot-Seep1 and Seep-SRB1g are thus far found only in Seep-SRB2 sp. 4 and 8. Analysis of gene expression data suggest that in the Hot-Seep1/ANME-1 partnership, PilY1, an adhesin with an immunoglobulin-like domain and adjacent cohesin/dockerin domains might play a role in the syntrophic lifestyle [ 24 ] ( S10 Table ). In the ANME-2c/Seep-SRB2 partnership, PilY1, cohesin/dockerin complexes and a protein with a CARDB domain are highly expressed [ 24 ] ( S10 Table ). Curiously, in the Seep-SRB2 partnering with ANME-1, we could only identify 1 moderately expressed adhesin with a fibronectin domain [ 24 ] ( S10 Table ). We note the presence and high levels of expression of cohesin/dockerin domains in both ANME-2c and their verified Seep-SRB2 partner [ 24 ], and the presence of fibronectin domains in both ANME-2a and their Seep-SRB1a partner ( S10 Table ) suggesting that perhaps both partners within a partnership express and secrete similar kinds of extracellular proteins. This might serve as a mechanism for partnership sensing. While our analysis and that of earlier research into adhesins present in ANME [ 14 ] identify a number of conserved and expressed adhesins, further work is needed to investigate their potential role in aggregate formation. 3.3 Secretion systems and intercellular communication in syntrophic SRB Extracellular contractile injection systems (eCISs) that resemble phage-like translocation systems (PLTSs) are found in some syntrophic SRB genomes ( S11 Table and S16 Fig ) although they are not as widely distributed as in ANME [ 14 ]. Typically, the eCIS bind to a target microorganism and release effector proteins into its cytoplasm. eCIS have been shown to induce death in worm larvae, induce maturation in marine tubeworm larvae [ 117 ] and found to mediate interactions between the amoeba symbiont and its host [ 118 ]. In ANME-SRB consortia, they might play a similar role with ANME releasing an effector protein into SRB, perhaps an effector molecule to promote the formation of a conductive biofilm or adhesins. Type VI secretion systems (T6SS) are similar to eCIS in facilitating intercellular communication between microorganisms. However, the primary distinction between them is that T6SS are membrane-bound while eCIS appear to be secreted to the extracellular space [ 119 , 120 ]. Interestingly, T6SS appear to be present in the ANME-2a partner Seep-SRB1a but absent in the ANME-2c partner Seep-SRB1a suggesting that they might play a role in mediating partnership specificity. While secretion systems are not uncommon in non-syntrophic bacteria, the high degree of their conservation in ANME and the high levels of expression of secretion systems in the ANME-2/Seep-SRB1a [ 25 ] and ANME-2c/Seep-SRB2 [ 24 ] partnerships suggest an important role for them in ANME-SRB syntrophy. Our analysis identified many conserved mechanisms for biofilm formation and intercellular communication in SRB to complement the pathways previously identified in ANME. Significantly, several polysaccharide biosynthesis pathways and adhesins were absent in the closest evolutionary neighbors of SRB indicating that adaptation to a syntrophic partnership with ANME required not just metabolic specialization but adaptation to a multicellular and syntrophic lifestyle. The adaptation of syntrophic SRB to partnerships with ANME To better understand the evolutionary adaptations acquired by syntrophic SRB to form partnerships with ANME, we mapped the presence and absence of the above-mentioned pathways in central metabolism, nutrient sharing, biofilm formation, cell adhesion, and partner identification across each of the syntrophic SRB clades and their nearest evolutionary neighbors from the same bacterial order ( S12 and S13 Tables). For example, the presence of the EET conduit OetABI in the Seep-SRB1a clade is nearly universal but, this trait is absent in the Desulfobacterales order that Seep-SRB1a belongs to, suggesting strongly that this machinery was horizontally acquired possibly in Seep-SRB1a or a closely related ancestor within the same family that includes Eth-SRB1. In contrast, most genomes in the order that Seep-SRB1g belongs to contain hydrogenases. However, hydrogenases are lacking in the syntrophic clade Seep-SRB1g implying that this trait was lost in the process of specialization to a partnership with ANME-2b. In addition to inferring adaptation based on presence and absence, phylogenetic trees were generated for at least 1 representative gene from each identified characteristic to corroborate the possibility of horizontal gene transfers (trees are available in S1 Data , Github ( https://github.com/ranjani-m/syntrophic-SRB )). These trees provide further insight into the adaptation of various traits, the likely source of the genes received horizontally and in the case of Hot-Seep1 and Seep-SRB2 sp. 1 demonstrate the transfer of OetABI from one syntrophic clade to another. With the trees, we were able to also identify those genes that were vertically acquired but adapted for the respiratory pathways receiving DIET electrons, for example Tmc ( S8 Fig ). A brief summary of the gene gains and losses is provided in Fig 7 and S13 Table . Our analysis suggests that some traits are associated with partnerships with different ANME. The pel operon present in Seep-SRB1g and Seep-SRB1a is more closely associated with aggregates formed with the ANME-2a/b/c species rather than ANME-1. Similarly, the capsular polysaccharide pseudaminic acid is present in those species of Seep-SRB1a that are associated with ANME-2c but absent in those species partnering ANME-2a suggesting that this polysaccharide might play a role in partnership identification and aggregate formation. Curiously, many of the adhesins we identified in the syntrophic SRB genomes have few close homologs in the NCBI NR dataset and almost no homologs in the nearest evolutionary neighbors ( S13 Table ), indicating that these proteins are likely highly divergent from their nearest ancestors. This is consistent with faster adaptive rates observed in extracellular proteins [ 121 ]. 10.1371/journal.pbio.3002292.g007 Fig 7 A summary of important gene loss and gain events in the physiological adaptation of sulfate reducing bacteria that led to a syntrophic partnership with ANME. The presence and absence of genes involved in the electron transport chain, nutrient sharing, biofilm formation, and cellular adhesion are listed in S12 Table . We identified genes that were potentially gained, lost, or biochemically adapted using a comparative analysis of the presence a given gene in a syntrophic clade in its order-level taxonomic background. For example, if a gene is present in a syntrophic SRB clade and is present in fewer than 30% of the remaining species in a given order, this gene is considered a likely horizontally transferred gene. The likelihood of horizontal transfer is then further corroborated with a phylogenetic tree of that gene generated with close homologs from NCBI and our curated dataset. The trees are available in S1 Data . The secondary analysis of the likelihood of gene gains and losses is present in S13 Table . ANME, anaerobic methanotrophic archaea; SRB, sulfate-reducing bacteria. With our analysis, we identified many genes and traits that are correlated with a syntrophic partnership with ANME, but it is less easy to identify whether they are essential. The complete conservation of the OetI-type or other EET cluster (such as OmcKL) suggests these are essential, but not sufficient, for the formation of this partnership since the multiheme cytochrome conduits themselves are present in many organisms not forming a syntrophic partnership with ANME. There is also a strong signature for the presence of a secreted polysaccharide pathway such as the pel operon in Seep-SRB1a and Seep-SRB1g and a xap-like polysaccharide in Hot-Seep1 and Seep-SRB2. With these components, a conductive biofilm matrix can be established, but the means of partnership recognition and communication between the archaea and bacteria are less clear. As suggested previously [ 14 ], the near complete conservation of the eCISs in ANME might play a role in partnership identification. The target receptor of the eCIS is unclear but the presence of conserved capsular polysaccharides in SRB that often are the target of bacteriophages and pathogens is suggestive as a possible site for binding. Likewise, the high levels of expression of cohesin and dockerin complexes by both ANME and SRB in the ANME-2c/Seep-SRB2 partnership are indicative of a role in syntrophic partnership [ 24 ]. In Seep-SRB1a, there are conserved fibronectin domains that likely bind the biofilm matrix and Seep-SRB2 has a conserved cell-surface protein with a PEGA sequence motif ( S12 Table ). We can infer something about the order of evolutionary adaption of syntrophic SRB from what is essential and conserved in syntrophic SRB and what is present in their nearest evolutionary ancestors. The presence of DIET complexes such as OetABI in the nearest evolutionary neighbors of HotSeep-1 (Desulfofervidales), Seep-SRB2 (Dissulfuribacteriales), and Seep-SRB1g (Seep-SRB1c) and the absence of adhesins (cohesins) and polysaccharide biosynthesis (pel) in the related clades ( Fig 7 ) suggests that the acquisition of DIET pathways in an ancestral clade was the first and essential step towards adaptation towards a syntrophic lifestyle. Then, the syntrophic partners likely acquired the pathways needed for aggregate formation (such as adhesins, the pel polysaccharide biosynthesis pathway) after. Seep-SRB2 contains a respiratory trait (CbcBA) that is absent in its nearest evolutionary neighbor ( Fig 4 ). This indicates that more steps were required for the adaptation of this clade to a syntrophic partnership with ANME. The greater diversity within the clades Seep-SRB1a and Seep-SRB2 may be a result of the larger number of partnerships with different ANME compared to a clade such as Seep-SRB1g. However, there is insufficient evidence to rule out the possibility of promiscuous partnership formation with multiple ANME within each SRB species. In these cases, the observed species diversity within Seep-SRB1a and Seep-SRB2 must be driven by other factors. Our analysis shows that the adaptation towards EET and the formation of conductive biofilms was likely driven by a greater selection pressure than the adaptation to a specific ANME partner. Consistent with this, the gain and loss of specific adhesin and matrix-binding proteins is more dynamic. Another aspect of the adaptation of syntrophic SRB is the high number of inter-clade transfers. In addition to the likely transfer of OetABI between HotSeep-1 and Seep-SRB2 sp.1 ( S7 Fig ), we also note a high degree of similarity between the proteins of the following components in different clades of syntrophic SRB—cohesin/dockerin modules, the OmcKL conduit, and enzymes in the pel and xap polysaccharide biosynthesis pathways. These appear to be the result of inter-clade transfers and the high number of transfers might imply that a mechanism promoting the exchange of DNA exists in this environment between ANME and SRB, either through a viral conduit or perhaps with the eCIS carrying DNA as cargo. Further analysis is needed to identify the number of transfer and the sources of transfers. In fact, a thorough accounting of these horizontal gene transfers combined with molecular clock dating might provide insight into the timeline and the relative age of the different ANME/SRB partnerships. Our phylogenomic analysis places the verified ANME-2c partners as ancestral to the ANME-2a partners within the Seep-SRB1a clade (Figs 1 and S1 ). Within the Seep-SRB2 clade, the topology places an ANME-1 partner as basal to the remaining Seep-SRB2 and the only verified ANME-2c partner as one of the later branching members (Figs 1 and S1 ). Earlier research places ANME-1 as the deepest branching lineage of ANME [ 14 ] and this relative ancestry of partners might suggest that Seep-SRB2 is older than Seep-SRB1a. However, it appears that ANME-1 acquired its mcr through horizontal gene transfer [ 14 ], and we have insufficient data to know when this occurred. Thus, we cannot know that ANME-1 was methanotrophic when it diverged from the Methanomicrobiales. These observations suggest that we cannot constrain the emergence of AOM solely through the relative branching patterns of the various ANME and SRB clades. A more thorough reconstruction of the adaptive gene transfers using the framework established for ANME and in this work for syntrophic SRB would provide insight into the evolution of this biogeochemically important syntrophic partnership."
} | 22,537 |
35498331 | PMC9049064 | pmc | 642 | {
"abstract": "The development of stable 3D surfaces for oil/water separation has been of great interest to researchers. Inspired by the lotus leaf, in this study, a superhydrophobic stable and robust surface was generated by the combination of n -octadecyltrichlorosilane, silica, polypyrrole and polyurethane (ODTCS–SiO 2 –PP–PU). The constructed 3D network displayed superhydrophobic and superoleophilic behavior with a high water contact angle of 154.7° ± 0.8°. The superhydrophobic behavior of the porous material was found to be stable for months. Apart from the hydrophobicity analysis of the material, the various forms of the materials were investigated via scanning electron microscopy (SEM), Fourier-transform infrared spectroscopy (FTIR), and energy-dispersive X-ray spectroscopy (EDX). Under the force of gravity, hexane displayed an exceptionally high flux of 102 068 Lm −2 h −1 through ODTCS–SiO 2 –PP–PU. The macroporous network of ODTCS–SiO 2 –PP–PU displayed fewer chances of fouling, which is a common issue with membranes. Moreover, its porous network displayed good absorption capacity for various non-polar organic solvents. The maximum absorption capacity observed for toluene was 34 times its own weight. The separation efficiency of various non-polar organic solvents from water was observed in the range of 99.5 to 99.8%. ODTCS–SiO 2 –PP–PU, due to its superhydrophobicity, 3D porous network, extraordinarily high flux, good absorption capacity, and excellent separation capability, has been established as a good candidate for the separation of organic and oil contaminants from water.",
"conclusion": "4. Conclusions In this work, a cost-effective combination of ODTCS, SiO 2 , polypyrrole, and polyurethane was used to develop a superhydrophobic porous network for the efficient separation of spilled non-polar organic contaminants from water. The surface functionalization of the polyurethane substantially improved the contact angle from 109.6° ± 2.3° to 154.7° ± 0.8°. Its superhydrophobic porous network rapidly passed non-polar organic liquids and displayed an extraordinarily high flux of 102 068 Lm −2 h −1 . The porous three-dimensional hydrophobic network has a great capacity to absorb and keep the non-polar organic solvents. It displayed an absorption capacity of 34 times its weight for toluene under ambient conditions. It can be used multiple times and the absorbed oil is recovered from ODTCS–SiO 2 –PP–PU by squeezing it. ODTCS–SiO 2 –PP–PU displayed good mechanical stability and after squeezing, the porous interconnected network was not destroyed. It displayed excellent separation efficiency for various organic solvents in the range of 99.5 to 99.8%, good recyclability, a facile route of synthesis, high flux, and a good absorption capacity. It displayed an excellent regeneration capability with hexane, dodecane, and gasoline. The RSD was found in the range of ±0.7 to ±2.9 after 12 cycles of use. Due to the high flux, great stability, excellent absorption capacity, and an efficient separation capability of oil from water and water in oil emulsions, ODTCS–SiO 2 –PP–PU is a unique material for the separation of oil from water.",
"introduction": "1. Introduction Fast-growing urbanization and rapidly escalating global energy demands have increased crude oil exploration. Simultaneously, global energy demands also require a large-scale crude oil extraction and rapid offshore movement of oil. Developing new methods to remove oil from the water will potentially replace the conventional technologies. 1–4 Several conventional methods have been applied for the separation of oil from water. 5 The conventional methods include the in situ burning of spilled oil, mechanical skimming, flotation, and oil dispersion using a chemical dispersant. Conventional methods have some limitations, which are not economically favorable and also suffer from low separation efficiencies. 3 In some cases, conventional methods are also a source of secondary pollutants. These demerits of the conventional methods make them unfavorable for the utilization of oil and water separation. This sort of limitation has motivated the scientific community to develop improved methods and materials that are more promising for the separation of oil and organic contaminants from water. These challenges can be addressed by introducing a superhydrophobic and superoleophilic porous surface. 6 A variety of approaches that have been adopted to develop superhydrophobic surfaces include dip coating, 7 temperature-based coating, 8 lithography, 9 chemical etching, sol–gel methods, 10 templating, 11 chemical vapor deposition, casting, 12 electrospinning, 13 and phase inversion methods. Various polymeric materials have been used to develop hydrophobic surfaces. 14 Superhydrophobic surfaces that display a contact angle of greater than 150° have received a great deal of attention. 15 There are two main characteristics that play a crucial role in the generation of superhydrophobic surfaces. The first factor is the surface roughness, which has a significant impact in improving the hydrophobicity. If a flat surface has a contact angle in the range of 100° to 120°, it may appear to be 150° to 170° on a microtextured or rough surface. 16 The second factor that has a dominant role in the material hydrophobicity and hydrophilicity is surface energy. A low surface energy imparts hydrophobic behavior to the material. A material is superhydrophobic through the combination of low surface energy and the appropriate surface topography. 17 In some cases, bumpy rough surfaces were generated by coating polymers on the porous surface and then, the surface was further coated with a low surface energy material to improve the hydrophobicity of the surface. 18 On the basis of the specific wettability, various porous materials that can act as a filter or oil absorber were introduced. 19 A material that displays the capacity to absorb or allow the passage of oil enables the collection of the oil. After the appropriate treatment, the collected organic component or oil can be reused. Although there has been extensive research in the field, there is still a remaining need to develop a material that is environmentally friendly, cost-effective, reusable, possesses a high absorption capacity, displays a high flux and is not a source of secondary pollutants. Herein, a superhydrophobic interconnected porous network was constructed using a combination of polymer-based organic and inorganic materials. On the porous network of polyurethane, pyrrole was catalyzed into polypyrrole and then silica particles were introduced into the polypyrrole-coated polymeric walls of the polyurethane. The silica nanoparticles were furthermore functionalized with n -octadecyltrichlorosilane. As a result, the combination of polypyrene, polyurethane, silica, and n -octadecyltrichlorosilane provided a stable superhydrophobic surface that displayed a high flux and excellent absorption capacity. The developed porous materials displayed an excellent shelf life. The developed superhydrophobic network of ODTCS–SiO 2 –PP–PU displayed good reusability and it was used multiple times for flux studies along with the absorption of various organic solvents. The absorbed oil can be released through a simple process of squeezing. The high flux of 102 068 Lm −2 h −1 and excellent absorption capacity of 34 times its own weight has established a promising future for the superhydrophobic ODTCS–SiO 2 –PP–PU in the separation of oil and other organic contaminants from water.",
"discussion": "3. Results and discussion 3.1 Polyurethane surface functionalization Pure polyurethane lacks specific wettability. By modifying the surface of PU at different steps, different surface wettability was displayed according to the presence of the surface functionalities. During absorption of the liquid, it showed the capacity to absorb both oil and water. Due to this observation, PU cannot be used directly for oil and water separation applications. The surface functionalization is a crucial step to obtain a stable material with a specific surface wettability. In the first step, a polypyrrole network was generated on the polyurethane by pyrrole polymerization. For polymerization, Fe 3+ acts as a catalyst to catalyze pyrrole (C 4 H 4 NH) polymerization into polypyrrole. 20 The ferric catalyst was introduced into the walls of the polyurethane where it facilitated rapid polymerization of the pyrrole and assisted the strong adherence of the polypyrrole to the porous interconnected walls of the polyurethane. After polymerization, the color of the polyurethane turned black due to the presence of polypyrrole. Due to polymerization, the introduced amino group played a vital role in providing a stable surface for further functionalization. The silica was introduced by silicic acid. The silicic acid demonstrated strong intermolecular forces with the polypyrrole on the surface of polyurethane. The silicic acid built a second functionalized network on the polyurethane through polypyrrole. The silicic acid hydroxyl group reacted in the third stage with ODTCS, where the long chains of the octadecyl group could link with polyurethane through the silicon. The overall interaction and the reactions of pyrrole, silica, and ODTCS on the porous walls of polyurethane are illustrated in Scheme 1 . Scheme 1 Schematic illustration of the reactions taking place between pyrrole, silicic acid and ODTCS on the surface of the polyurethane. 3.2 Characterization of the ODTCS–SiO 2 –PP–PU The functionalization of the polyurethane at various stages was established with the help of FTIR. The FTIR spectra of the polyurethane prior to functionalization were collected and the characteristic absorption bands of polyurethane were observed. The polyurethane displayed a carbonyl (–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) stretching absorption band at 1726 cm −1 . The –N–H deforming band appeared at 1537 cm −1 and the stretching band appeared at 3292 cm −1 . The –C–H symmetric and asymmetric stretching vibrations were observed in the range of 2800 to 3000 cm −1 . 21 A sharp absorption band that appeared in the non-functionalized polyurethane at 1094 cm −1 was assigned to –C–O–C– stretching ( Fig. 1A ). The FTIR spectra of the polypyrrole-coated polyurethane was recorded and a broad absorption band was observed at 600 cm −1 . The presence of a broad peak was due to the presence of ferric/ferrous components in the composite. A broad absorption peak after 3000 cm −1 was observed in the spectra of PP–PU. The appearance of a broad peak after 3000 cm −1 was attributed to the NH group of polypyrrole and polyurethane. Most of the absorption peaks in the PP–PU FTIR spectra appeared broad compared to those in the non-functionalized polyurethane spectra. Due to the broad absorption bands, most of the polypyrrole peaks were merged under the signal. The peaks appearing at 1292 cm −1 were assigned to the C–H vibrations 22 ( Fig. 1B ). The FTIR spectrum of SiO 2 –PP–PU revealed that the incorporation of silica into PP–PU was successful. A sharp absorption band appeared at 1081 cm −1 which represents the –Si–O–Si– antisymmetric stretching vibrations. The appearance of Si–O–Si bands in SiO 2 –PP–PU demonstrated the successful incorporation of SiO 2 into PP–PU ( Fig. 1C ). In the case of, ODTCS–SiO 2 –PP–PU along with the –Si–O–Si– peaks, the more prominent sharp peaks of –C–H stretching vibrations were observed at 2849, 2916 and 2955 cm −1 . The appearance of –Si–O–Si– and –C–H prominent sharp peaks confirmed the successful functionalization of SiO 2 –PP–PU with ODTCS ( Fig. 1D ). Fig. 1 The FTIR spectra of (A) polyurethane, (B) PP–PU, (C) SiO 2 –PP–PU, and (D) ODTCS–SiO 2 –PP–PU. Morphological information of various forms of the modified polyurethane was collected by scanning electron microscopy (SEM). The SEM images provided useful information related to the surface changes after each step of modification. The polyurethane consists of a porous network. The walls of the PU appeared planer and this was evident from the SEM images of the unmodified PU ( Fig. 2A(a) and B(b) ). The polymerization of pyrrole on the planar walls of the polyurethane was clearly observed by SEM. The walls of the polyurethane appeared wavier and thicker after the catalytic polymerization of the pyrrole. The polyurethane maintained a porous network after the pyrrole polymerization on the walls of the polyurethane, as shown in Fig. 2A(b) . SEM images have revealed the SiO 2 distribution on the walls of the polypyrrole-modified polyurethane. Apart from the uniform distribution of the SiO 2 , some aggregates of SiO 2 were also observed ( Fig. 2A(c) and B(c) ). In the case of ODTCS–SiO 2 –PP–PU, the SEM images display the sheets which cover the SiO 2 -incorporated pyrrole-polymerized walls of the polyurethane. The sheets appeared as a result of ODTCS and even a lump of the SiO 2 was also covered under the sheet ( Fig. 2A(d) and B(d) ). The SEM images revealed that the surface morphology of the polyurethane walls changed after each step of modification ( Scheme 2 ). Fig. 2 SEM images at two different magnifications (A and B) of (a) PU, (b) PP–PU, (c) SiO 2 –PP–PU and (d) ODCS–SiO 2 –PP–PU. Scheme 2 Schematic showing the generation of functionalized polyurethane by the combination of polypyrrole, silica, and ODTCS on the polyurethane walls. The 3D porous network was further investigated using EDX spectroscopy. EDX spectroscopy is a valuable analytical tool that provides elemental information on the material. The EDX spectrum of ODTCS–SiO 2 –PP–PU displayed additional peaks of Si, Cl, and Fe that were absent in the EDX spectrum of PU ( Fig. 3 ). The sharp peak of Si highlights the presence of SiO 2 and the silane group. Ferric chloride was loaded into the polyurethane to catalyze the polymerization reaction of pyrrole. Due to this, Fe and Cl peaks appear in the ODTCS–SiO 2 –PP–PU EDX spectrum. The appearance of the relevant elements during the EDX analysis support the successful production of the 3D porous composite of polyurethane. Fig. 3 EDX spectra of (A) PU and (B) ODTCS–SiO 2 –PP–PU. 3.3 Specific surface wettability of ODTCS–SiO 2 –PP–PU The surface wettability of PU was evaluated after each step of modification and functionalization. The polyurethane displayed a water contact angle of 109.6° ± 2.3°. Water droplets have also shown a tendency to adhere to the surface. PP–PU wettability was entirely changed, and the water droplet was readily adsorbed by the surface. This change of the surface behavior was due to the presence of numerous NH– groups of polypyrrole that impart the polar behavior to the surface. The same behavior was observed for the SiO 2 –PP–PU. In the case of SiO 2 –PP–PU, the polarity still dominated the surface of polyurethane. The surface of ODTCS–SiO 2 –PP–PU became superhydrophobic after the introduction of the low surface energy of ODTCS functionalization that was stabilized on the walls of the polyurethane with the help of the silica and polypyrrole. The superhydrophobicity of ODTCS–SiO 2 –PP–PU was revealed by the water contact angle of 154.7° ± 0.8°. The adherence of the water droplet that was observed on the unmodified polyurethane disappeared in the case of ODTCS–SiO 2 –PP–PU. The water droplet from the auto-controller micropipette continuously touched the surface of ODTCS–SiO 2 –PP–PU by bringing the auto-controller micropipette down, but the surface did not shown affinity towards the water droplet. The high-water contact angle revealed that after functionalization of the SiO 2 –PP–PU with ODTCS, the surface energy was substantially decreased, which was evident from the three-dimensional macroporous surface behavior towards the water ( Fig. 4 ). Fig. 4 Water contact angle on the surfaces of (A) PU, (B) PP–PU, (C) SiO 2 –PP–PU and (D) ODTCS–SiO 2 –PP–PU. 3.4 Evaluation of the absorption, regeneration and emulsion separation capability The ODTCS–SiO 2 –PP–PU absorption capacity and its regeneration capability were evaluated for oils and various non-polar organic pollutants. The macroporous superhydrophobic network of ODTCS–SiO 2 –PP–PU displayed a good absorption capability for a range of non-polar organic liquids. The absorption capacity was evaluated for the range of non-polar organic solvents and petrol components including hexane, heptane, iso-octane, dodecane, petrol, cyclohexane, o -xylene, and toluene. The weight gain ratio by using eqn (1) was found in the range of 1800 to 3400% ( Fig. 5 ). The absorption capacity for petrol was found at 2162%. It is crucial to note that the absorption capacity might predominantly depend on the density of the various oils and the organic solvents. 23 The developed 3D macroporous superhydrophobic network of ODTCS–SiO 2 –PP–PU displayed an absorption capacity in the range of 18 to 34 times better than that of the previously reported recent materials, including aerogel composites (2–16 times), 24 PDMS-SW (12–27 times), 25 magnetic silicone sponges (7–17 times), 26 PDMS sponges (4–11 times), 23 and nitrogen-rich carbon aerogels (6–11 times). 27 Moreover, toluene absorption was specifically compared with the previous literature and demonstrated better efficiency here than in the previously reported work ( Table 1 ). The increase in weight gain ratio is evidence that ODTCS–SiO 2 –PP–PU has an excellent capability to absorb a large quantity of non-polar organic liquids and spilled oils. Moreover, its facile and robust route of fabrication using cost-effective raw materials is an attractive feature for scale-up and readily deployable for application. Fig. 5 Absorption capacity of ODTCS–SiO 2 –PP–PU for oil and non-polar organic solvents. Comparison of the absorption capacity for toluene by the ODTCS–SiO 2 –PP–PU to the reported materials Sr.# Superhydrophobic material Preparation methods Absorption capacity (toluene) Regeneration Ref. 1 Monolithic superhydrophobic silica aerogel Sol–gel process 9 Distillation and vacuum filtration \n 28 \n 2 Porous BNNS/PVDF composite material Gelation and freeze-drying process 5 Washing in ethanol and drying at 60 °C in air \n 29 \n 3 Magnetic graphene foam Hummer's method 19 Hexane immersion \n 30 \n Gas based reduction Co-precipitation 4 Fluorinated polydopamine/chitosan/reduced graphene oxide composite aerogel Hummers' method 8 Heating and squeezing \n 31 \n Hydrothermal Immersion Fluorination 5 Graphene foam Modified hummers method 20 Washing in ethanol and oven drying \n 32 \n Sol–gel method Hydrolyzed and curing 6 PDMS sponge Sugar templating process 5 Squeezing \n 23 \n 7 Porous BN nanosheets Dynamic templating approach ∼24 Burning and heating in air \n 33 \n 8 Superhydrophobic/superoleophilic cotton fiber Sol–gel process ∼30 Drained under mild suction by a vacuum air pump \n 34 \n Self-assembling 9 Carbon aerogel Hydrothermal and post-pyrolysis process 29 By heating \n 35 \n 10 PDMS-SW Immersion 11.5 Mechanical squeezing \n 25 \n 11 Magnetic silicone sponge Hydrolysis and polymerization 9 By squeezing \n 26 \n 12 ODTCS–SiO 2 –PP–PU Catalytic polymerization 34 By squeezing This work Immersion The regeneration is a crucial factor in deciding the fate of material scalability and practical application. For example, cost-effective materials have lost attention for practical use if they do not display the capability to regenerate after a single use. ODTCS–SiO 2 –PP–PU regeneration was evaluated with hexane, dodecane, and petrol. The same ODTCS–SiO 2 –PP–PU was used for the regeneration study of various organic pollutants. ODTCS–SiO 2 –PP–PU displayed excellent regeneration capabilities for the analyzed hexane, dodecane and petrol with a RSD of ±0.7 ( n = 12), ±1.2 ( n = 12) and ±2.9 ( n = 12), respectively ( Fig. 6 ). The spongy nature of the macroporous material facilitated oil removal by squeezing it out. After multiple uses, the surface maintained its superhydrophobicity and repelled the water strongly from its surface. The regenerated ODTCS–SiO 2 –PP–PU was further investigated by collecting its FTIR spectrum. The FTIR spectrum of the regenerated ODTCS–SiO 2 –PP–PU revealed all the characteristics peaks that were present in the modified PU before its use (Fig. S1 † ). This regeneration behavior has revealed the robustness of the material and that it can be used for a long time. Fig. 6 Regeneration evaluation of ODTCS–SiO 2 –PP–PU for hexane, dodecane, and petrol. The capability of ODTCS–SiO 2 –PP–PU was evaluated for the separation of the surfactant-free and surfactant-stabilized emulsion. The surfactant-free and the surfactant-stabilized water in oil emulsion were prepared by using distilled water and chloroform. The surfactant-free emulsion was stabilized by sonication whereas the surfactant-stabilized emulsion was prepared by adding the surfactant. The development of the emulsions could be observed in their respective vials that appeared as a milky color (Fig. S2 and S3 † ). Emulsions were passed through the ODTCS–SiO 2 –PP–PU that were tightly packed into the nozzle of the apparatus. The ODTCS–SiO 2 –PP–PU permitted the chloroform to pass while preventing the water from passing. It is shown in Fig. S2 and S3. † High flux through a continuous network of absorbent material is an important factor for the porous material in order to absorb large spills of oil or non-organic solvents. The macroporous network of ODTCS–SiO 2 –PP–PU displayed an outstanding capability for the continuous passing of the oil or non-polar organic component. Through the porous network, the non-polar component passed very fast while the water was rejected. The flux for hexane was found to be 102 068 Lm −2 h −1 . This flux value for the non-polar organic component is exceptionally high. As depicted in Fig. 7 , the hexane flux was evaluated 12 times and the RSD was found to be ±2.31. This indicates that after multiple uses, the porous network of ODTCS–SiO 2 –PP–PU was not clogged and continuously allowed a high flux of hexane through it. The flux is high compared to values in previously reported flux studies ( Table 2 ). M. H. Tai studied a SiO 2 -carbon composite membrane with a hexane flux of 2648.8 Lm −2 h −1 under gravity and a contact angle of 144.2°. 36 In the synthesized ODTCS–SiO 2 –PP–PU, the hexane flux (102 068 Lm −2 h −1 ) was greater and displayed more surface hydrophobicity with a contact angle of 154.7° ± 0.8°. The macroporous network provided fewer coagulation chances, which are common issues with the membranes used for oil/water separation. Fig. 7 Gravity-driven passage of hexane flux from ODTCS–SiO 2 –PP–PU under ambient conditions. The comparison of the hydrophobicity and the flux of ODTCS–SiO 2 –PP–PU with other reported materials Sr.# Superhydrophobic material Preparation methods Contact angle Flux (Lm −2 h −1 ) Ref. 1 PVDF–HFP membrane Electrospinning 134.0 ± 3.2° 94 000 (gasoline) \n 37 \n 2 SiO 2 –carbon composite membrane Electrospinning 144.2 ± 1.2° 2648.8 (hexane) \n 36 \n 3 TPU microfiber membrane Force spinner NA 4659 (oil flux) \n 38 \n 4 TPU-PNIPAM membrane Force spinner NA 503 (oil flux) \n 38 \n Free radical polymerization 5 PAA- g -PVDF membrane Salt-induced NA 2320 (hexadecane/H 2 O) \n 39 \n Phase inversion 60 Co γ-ray source irradiation 6 ODTCS–SiO 2 –PP–PU Catalyst based polymerization 154.7° ± 0.8° 102 068 (hexane) This work Apart from the gravity-based separation, the dynamic separation of oil/water was also evaluated by applying pressure with the help of a peristaltic pump. For this purpose, ODTCS–SiO 2 –PP–PU was placed at the interface of the oil and water in such a way that it was more than half dipped into the water (Video S1 † ). This strategy was adopted to observe during oil passage whether the water passed or was prevented from passing by the superhydrophobic nature of the material. In the dynamic separation analysis, it was observed that hexane passed rapidly from the ODTCS–SiO 2 –PP–PU network, while the water was completely prevented from passing. This is evidence that the surface remained superhydrophobic during the separation of the non-polar organic and water mixture. Through the ODTCS–SiO 2 –PP–PU, various systems of hexane/water, heptane/water and octane/water were separated to evaluate the separation efficiency of the porous network of ODTCS–SiO 2 –PP–PU. The separation efficiency for the hexane, heptane and octane were found in the range of 99.5 to 99.8% ( Fig. 8 ). The efficient separation can be explained by the superhydrophobic porous network that spread over all of the polyurethane walls. It rapidly allowed the passage of the non-polar component and completely prevented the passage of the water through it. Fig. 8 Separation efficiency of the ODTCS–SiO 2 –PP–PU for various organic solvents from water."
} | 6,322 |
40053589 | PMC11887809 | pmc | 643 | {
"abstract": "A major challenge in synthesizing strong and tough protein fibers based on spider silk motifs is understanding the coupling between protein sequence and the postspin drawing process. We clarify how drawing-induced elongational force affects ordering, chain extension, interchain contacts, and molecular mobility through mesoscale simulations of silk-based fibers. We show that these emergent features can be used to predict mechanical property enhancements arising from postspin drawing. Simulations recapitulate a purely process-dependent mechanical property envelope in which order enhances fiber strength while preserving toughness. The relationship between chain extension and crystalline domain alignment observed in simulations is validated by Raman spectroscopy of wet-spun fibers. Property enhancements attributed to the progression of anisotropic extension are verified by mechanical tests of drawn silk fibers and justified by theory. These findings elucidate how drawing enhances properties of protein-based fibers and shed light on how to incorporate this effect into predictive models.",
"introduction": "INTRODUCTION Protein-based materials like silk are gaining broad appeal for their exceptional and tunable mechanical properties. Natural spider silk fibers exhibit tensile strength in the same range as steel with toughness achieved through rubber-like extension rivaling that of Nylon and Kevlar fabrics ( 1 , 2 ). Although it cannot yet fully replicate the remarkable mechanical properties of its native counterpart ( 3 ), synthetic silk fiber produced by the recombinant approach is still strong ( 4 ), tough, elastic, biodegradable ( 5 ), biocompatible ( 6 ), and antimicrobial ( 7 ). These properties shared with natural silk have inspired the design of sutures ( 8 ), protein-based adhesives ( 9 ), vascular grafts ( 10 ), metabolite monitoring biosensors ( 11 ), and other tissue engineering applications ( 12 ). Insurmountable challenges prohibit scalable spider silk farming to the extent that complete natural spider silk garments have only ever been made for exhibition at great effort and cost ( 13 ). Furthermore, recombinant synthesis enables researchers to expand the library of sequence-property relationships beyond those fibers found in nature documented in the recently released Spider Silkome Database ( 14 ). Such precise control of primary amino acid sequence has stimulated a research field dedicated to manipulating the properties of wet-spun protein-based fibers ( 15 , 16 ) through consensus sequencing ( 17 , 18 ), domain fusion ( 4 , 19 , 20 ), directed evolution, secondary structure design principles ( 21 , 22 ), and machine learning approaches ( 23 ). However, fiber mechanical properties by no means depend exclusively on sequence. Peptide assembly in the silk gland is a multistage process by which micellar ( 24 ) aggregates are disrupted by pH gradients and shear flow along the tapered duct ( 25 ) before assembling into a matrix of β crystals embedded in an amorphous matrix ( 26 ). Understanding how peptide conformational changes drive assembly during this process is therefore necessary to develop biomimetic spinning approaches yielding fibers with predictable mechanical properties. In the past, computational modeling strategies targeting silk protein dynamics have confirmed theories of shear-induced crystal formation and network percolation ( 27 , 28 ) and the effects of sequence features and hydrodynamic flow on silk assembly ( 29 , 30 ). It has also been shown that terminal domain interactions in shear flow enhance the order of those extended domains tethering aggregates together ( 31 ). These studies all reveal the critical role that anisotropy plays in the mechanics of silk fibers but do not directly correlate molecular descriptors of anisotropy with fiber mechanical properties. In this work, we use a coarse-grained dissipative particle dynamics (DPD) representation of silk as a simple diblock copolymer ( 27 – 29 , 31 ) in solvent to capture mechanical property trends as they relate to protein orientation and extension arising from postspin drawing. At a molecular scale, the degree of anisotropy of individual proteins can be quantified by their shape and orientation relative to the fiber axis. End-to-end length and Hermans order parameter, when considered together, quantify the degree of anisotropy in a bulk protein or polymer matrix. At the same time, the greater end-to-end length of individual proteins promotes formation of stable hydrogen bonds perpendicular to the fiber axis, which can enhance strength and toughness while decreasing fiber extensibility. By extending proteins along the fiber axis directly before uniaxial tensile tests, we demonstrate how order, elongation, and interprotein hydrogen bonding are responsible for the modular toughness, elastic modulus, maximum stress, and maximum strain achieved through postspin drawing. To control the molecular order and alignment of individual proteins in simulations, an equal and opposite pull force is applied to the two ends of each protein during processing. We find substantial variability in the degree to which this pull force increases protein end-to-end length and Hermans order parameter for proteins of differing molecular weight (MW). While higher-MW proteins have greater extensibility at low forces, they require higher forces to achieve the same degree of order as shorter proteins. We also derive correlations between order, hydrogen bond counts, and mean squared displacement (MSD), features that have been emphasized in previous theoretical descriptions of the silk property envelope ( 23 , 32 – 34 ). Last, we link these features to the wide ranging mechanical properties observed through drawing in both simulated and recombinant silk fibers and discuss their role in energy storage for property enhancement. Our work reveals the process dependence of molecular-level features on structure and mechanical properties, which is an important step toward developing hybrid computational, experimental, and theoretical predictive models that consider the process as much as the composition of polymer materials.",
"discussion": "DISCUSSION As the accessibility of data-hungry machine learning methods grows, so does the demand for large libraries linking target properties to critical design features across scientific fields. In the case of spider silk, there are limited data connecting mechanical property data to the primary amino acid sequence ( 14 ), and there are even less data that consider processing and sequence parameters simultaneously. Models to predict physical properties of materials can be enriched through incorporation of simulation-derived relationships between B -factor, order, chain conformation, and mechanical properties. Here, we proposed a method for detecting the dependence of synthetic silk fiber mechanical properties on order and extension arising from postspin drawing using a computationally inexpensive coarse-grained molecular dynamics approach. Previous efforts at modeling spider silk with all-atomistic representations have provided fundamental insights on the mechanical behavior of silk nanostructure ( 60 – 62 ), strain-induced β sheet structural transitions ( 63 ), self-assembly in hydrodynamic flow ( 30 ), and even bulk mechanical properties of spider silk ( 64 ). Our proposed processing method captures critical elements of these models including extension-induced enhancement of overall crystallinity, interprotein hydrogen bonding, and ordering of proteins along the fiber axis. At the same time, the simulated drawing process recreates the order-dependent mechanical property envelope that gives silk-based materials their processing-dependent tunability. The inspiration for this work came after we recreated the data from a paper using this silk DPD model to demonstrate that shear flow substantially increased the order of domains tethering crystal aggregates ( 31 ). We found that those tethering domains only accounted for a small proportion of all the proteins in the system. Most proteins remained in a globular state with short R ee and low s 0 , squandering unrealized potential for extension and ordering. To contextualize our work with theirs, a description of how different combinations of F p and shear flow rate ( γ · ) during processing affect order, end-to-end length, and hydrogen bonding is summarized in fig. S17 (B to E). Figure S17A shows how a single system can access the full extent of the mechanical property envelope through variation of two processing parameters. However, phenomena such as an increase in elastic modulus of shear-processed fibers over fibers that are processed purely with the drawing pull force warrant more detailed analysis of yield mechanisms in tensile tests and validation with experiments that covary shear rate and draw ratio. These results emphasize the synergy between shear flow and application of drawing forces during the fiber-forming process. The seminal work introducing this DPD model of silk, carried out by Lin et al. ( 27 ), inspired Shen et al. to develop a full-atomistic model of a spider silk fiber ( 64 ). Shen et al. used the predicted distribution of shear flow–assembled crystals and the network connecting them to manually build an all-atomistic continuum of silk crystals embedded in an amorphous matrix. They used their full-atomistic model to investigate the impact of prestretching on mechanical properties, but their efforts were limited by a lack of solvent. Breaking strain decreased with increasing breaking strength as a function of prestretched ratio under the dry prestretching condition, but fiber toughness decreased, which contrasts with the simulated and synthesized fibers for this manuscript. This indicates that solvent is a necessary element of silk drawing molecular models for its role in enhancing interprotein hydrogen bonding and facilitating β sheet reorganization. Another notable difference between these two studies was the nonequilibrium molecular dynamics strategy used to simulate drawing. In the full-atomistic model, the entire simulation box was deformed uniformly to prestretch the fiber, whereas here, we apply a homogeneous force to each protein in the system to precisely control their degree of extension. From a modeling perspective, the uniform box deformation strategy markedly narrows the simulation box dimensions, which can cause boundary effects to play a role at lower strains in uniaxial tensile tests. However, it more directly mimics the drawing process than the artificially induced pull force. Both strategies should be considered and compared more directly in future studies while including solvent during prestretching. Although this method currently lacks primary sequence-level resolution, there is potential for the implementation of systematic coarse-graining approaches such as the energy renormalization method ( 54 , 65 , 66 ), which could be used to introduce primary-sequence features to this already process-sensitive workflow. The repulsive DPD parameters of hydrophobic and hydrophilic bead types in this model are largely qualitative, having been borrowed from a generalized DPD model of copolymers in solvent ( 37 , 38 ). This has proven sufficient in linking process-dependent polymer conformation and assembly to mechanical properties. However, a more thorough and conventional approach of deriving coarse-grain force field parameters from linking force field parameters to all-atomistic simulation is necessary to extract sequence-property relationships from simulation. While the coarse-graining process would itself be computationally intensive, the expansion of parameters would not substantially decrease the efficiency of this simulated silk processing workflow. A notable limitation of this model is its omission of the final extrusion step of synthetic silk spinning, which involves solvent exchange in the case of wet spinning ( 16 , 67 ) or a highly viscous dope in the case of dry spinning ( 68 , 69 ). The biggest consequence of leaving out this processing step is the time dependence of s and R ee through the relaxation step after s 0 has been measured. Because the solvent particles remain during the tensile deformation, the stress-strain profiles likely underpredict the true strength and stiffness of silk fibers by maintaining the relative conformational freedom that polymers have in solvent. The presence of solvent during tensile tests may also explain the large deformations up to many times the initial fiber length, which are inconsistent with synthetic fibers that stretch to a maximum of double their initial length. Along with implementation of a systematic coarse-grain framework, this model would benefit from a solvent exchange model by which hydrophilic solvent particles are replaced by hydrophobic particles. This process would induce the rapid protein assembly that occurs during extrusion into nonpolar solvent like hexafluoroisopropanol (HFIP). The drying process could then be modeled by gradually removing solvent particles and relaxing the system between particle deletions. These additions would be an important step toward modeling the aqueous environment and solvents experienced by silk proteins in both the natural and synthetic silk-spinning processes ( 70 ). The brittle behavior of undrawn wet-spun fibers is a common challenge, which severely decreases the maximum strength and toughness of synthetic fibers. This behavior is unique to synthetic silk. Natural silk undergoes a drawing force whether it is forcibly reeled or pulled out of the silk gland by the spider’s own hind legs. Without a drawing step, proteins are immobilized by increased entanglements, which are made permanent upon drying. Undrawn fibers therefore break abruptly when a load is applied since the chains cannot elongate and reorient during the deformation. In simulations, mobility is always conferred by the presence of solvent. Observing the mechanisms of brittle failure in undrawn silk fibers would therefore require a model that accounts for the drying process of the undrawn fibers compared to drawn fibers. This phenomenon will be studied further when a suitable strategy for removing solvent particles is devised. The continuous model of an anisotropic network of silk peptides in solvent also fails to consider fracture mechanisms that are related to impurities and cracks that form at the interface of the silk fiber shell and air. Consistent manufacturing of silk fibers requires that the spun silk is exposed to a consistent ambient environment with minimal particulates, which can disrupt the integrity of the fiber shell during drying. Variation in mechanical properties derived from simulations was also observed, which, in some cases, introduced substantial uncertainty in the measurements. Those variations are summarized by the error bars representing standard deviation across five runs in Fig. 5 . Within simulations that were processed with the same conditions, but with different starting configurations, we could not find any variables that consistently correlated with fluctuations in mechanical properties. Variability within processing conditions must be either a function of many of the parameters quantified in this study or because of some spatial variability that causes weak points in the simulation box. Such spatial variability could be mitigated by either expanding the size of the simulation or improving the homogeneity of the simulation by removing solvent. Nonetheless, our work serves as an augmentation to existing silk modeling strategies and will hopefully inspire further improvements toward developing sequence-process-property prediction methods for silk and other biological materials."
} | 3,952 |
29691409 | PMC5915437 | pmc | 644 | {
"abstract": "Consumption of methane by aerobic and anaerobic microbes governs the atmospheric level of this powerful greenhouse gas. Whereas a biochemical understanding of aerobic methanotrophy is well developed, a mechanistic understanding of anaerobic methanotrophy has been prevented by the unavailability of pure cultures. Here we report a biochemical investigation of Methanosarcina acetivorans , a methane-producing species capable of anaerobic methanotrophic growth dependent on reduction of Fe(III). Our findings support a pathway anchored by Fe(III)-dependent mechanisms for energy conservation driving endergonic reactions that are key to methanotrophic growth. The pathway is remarkably similar to pathways hypothesized for uncultured anaerobic methanotrophic archaea. The results contribute to an improved understanding of the methane cycle that is paramount to understanding human interventions influencing Earth’s climate. Finally, the pathway enables advanced development and optimization of biotechnologies converting methane to value-added products through metabolic engineering of M. acetivorans .",
"introduction": "Introduction The production and consumption of methane is a component of the global carbon cycle. An end product of the anaerobic decomposition of biomass, nearly one billion tons of methane is produced each year in diverse anaerobic habitats of the Earth’s terrestrial biosphere 1 . Methane is also produced in anaerobic marine sediments and released from vast reservoirs of gas hydrates (5.0 × 10 5 –1.0 × 10 7 Tg) 2 . Methane is an important greenhouse gas nearly 20-fold more potent than CO 2 3 . Fortunately, release to the atmosphere is mitigated by the aerobic and anaerobic oxidation of methane or assimilation into cell biomass by microbes that ultimately control the Earth’s climate 4 . The anaerobic oxidation of methane (AOM) requires reduction of electron acceptors (Fe(III), Mn(IV), nitrate or sulfate) to be thermodynamically favorable 5 , 6 . Until recently, it was thought that AOM in marine sediments required a symbiosis of anaerobic methanotrophic archaea (ANME) and sulfate-reducing species for which the latter utilizes reductant produced by the former to make the overall reaction thermodynamically favorable. However, the artificial electron acceptor AQDS (9,10-anthraquinone-2,6-disulfonate), a surrogate for humic substances in the environment, decouples methane oxidation from sulfate reduction in marine sediments presenting the possibility of independent respiratory AOM and growth of usually syntrophic ANME-2 7 , 8 . Indeed, respiratory AOM is in accordance with the occurrence of solitary ANME in microbial mats and sediments from diverse anaerobic environments 9 . Furthermore, incubations of environmental samples with Fe(III)-citrate yielded biosynthetic activity exclusive to uncultured ANME-2c and other ANME-2 cells 7 . These results inspired the hypotheses that symbiotic associations of ANME and sulfate-reducing species evolved from methanogenic species that first acquired the ability to conserve energy by oxidizing methane and reducing metal oxides 7 . Fe(III)-dependent AOM reported for a culture enriched in ‘Candidatus Methanoperedens nitroreducens’ presents the possibility of Fe(III)-dependent respiratory growth although not yet documented 10 . AOM dependent on reduction of Fe(III) is of particular biogeochemical interest. Indeed, it is proposed that if only a small percentage of current global Mn(IV) and Fe(III) influx is used for AOM, it has the potential to consume a large amount of methane 11 . It is also proposed that Fe(III)-dependent AOM was largely responsible for oxidizing the entirety of methane produced on early Earth prior to the advent of oxygen 11 . The prospect that ANME are capable of independent AOM and growth by Fe(III) respiration profoundly changes current views of AOM and iron cycling in Nature, prompting further investigation. It is conjectured that ANME grow by reverse methanogenesis based on environmental metagenomic and metatranscriptomic analyses of sediments 5 , 6 . However, reversal and growth requires mechanisms for energy conservation and overcoming endergonic reactions yet to be investigated biochemically 5 , 6 . Clearly, biochemical approaches with pure cultures are necessary to obtain a rigorous understanding of AOM. Although discovered nearly four decades ago, the unavailability of pure cultures has prevented biochemical investigations of AOM. However, a culture enriched in ‘Candidatus Methanoperedens nitroreducens’ is capable of AOM dependent on reduction of nitrate-, nitrite- or Fe(III) 10 , 12 . Furthermore, Methanosarcina acetivorans strain C2A is capable of trace methane oxidation (TMO) defined as reverse methanogenesis during net methane production from growth substrates in the absence of external electron acceptors 6 , 13 , 14 . TMO contrasts with AOM which is independent of methanogenesis and requires electron acceptors with or without a syntrophic partner. More recently, Fe(III)-dependent AOM was documented for a strain of M. acetivorans engineered to produce methyl-coenzyme M methyl reductase (Mcr) derived from ANME-1 sediments 15 . Thus, M. acetivorans has emerged as a model for advancing a biochemical understanding of AOM. M. acetivorans is also recognized as instrumental in development of biomanufacturing processes converting methane into value-added products 16 – 18 . Here we report a biochemical investigation of wild-type M. acetivorans that supports an AOM pathway anchored by Fe(III)-dependent respiration generating ion gradients that supply the energy to drive endergonic reactions essential for AOM and growth. The results provide a deeper mechanistic understanding of AOM and iron cycling in Nature, and a guide for optimization of methane-based biotechnologies.",
"discussion": "Discussion The proposed biochemical-based pathway elevates AOM by M. acetivorans to a fundamental mechanistic level of understanding. The pathway is essentially a reversal of the biochemically characterized acetoclastic and CO 2 -reduction methanogenic pathways, albeit dependent on reduction of Fe(III). The results reveal the role Fe(III) plays in energy conservation and driving endergonic reactions that support methanotrophic growth 15 . Abundant methane is targeted for development of recently advocated biomanufacturing processes converting it to liquid biofuels and value-added products 16 . Thus, the pathway provides a guide for optimization of existing, and development of advanced, biotechnologies facilitated by the timely announcement of a cas9-mediated genome editing system for facile metabolic engineering of M. acetivorans 16 – 18 , 45 – 47 . The proposed M. acetivorans pathway is remarkably similar to the pathway proposed for ANME-2a based on the single-aggregate genome derived from integrated environmental metagenomic and transcriptomic analyses of a marine environment 21 . The M. acetivorans and ANME-2a pathways both include Rnf, Fpo, MHC, HdrDE and CO dehydrogenase/acetyl-CoA synthase for which the homologs show robust amino acid sequence identities (Supplementary Table 2 ). The results presented, showing generation of Na + and H + ion gradients by Rnf and Fpo of M. acetivorans , validate the same roles for Rnf and Fpo in pathways proposed for uncultured ANME 20 , 21 , 32 . The pathways for M. acetivorans , ANME-1, ANME-2a and ANME-2d include acetate as a product 12 , 21 , 32 . Furthermore, analysis of methane seep sediments show robust 13 C depletion in acetate with δ 13 C values near −90% 48 . These results suggest that the proposed AOM pathway of M. acetivorans is largely representative of ANME, further strengthened by data showing ANME-1, ANME-2d and ANME-2a are phylogenetically related to Methanosarcina species 10 , 20 , 21 , 49 . The one exception is Mcr of wild-type M. acetivorans which could explain why only the engineered strain containing Mcr from ANME-1 is capable of Fe(III)-dependent AOM 15 . No other genetic alterations were imposed on the engineered strain ensuring that mechanisms for Fe(III)-dependent energy conservation and driving endothermic reactions are identical to those investigated in the wild type. Although TMO by wild-type M. acetivorans indicates that Mcr catalyzes the reverse reaction analogous to that of Methanothermobacter marburgensis 50 , bias in the direction of methane formation may be a kinetic block to initiating AOM as opposed to a more favorable rate of methane oxidation for Mcr of ANME. The similarity in pathways for M. acetivorans and proposed for ANME-2a strengthen the prospect of Fe(III)-dependent AOM and respiratory growth by uncultured ANME. Importantly, the roles shown for Rnf, MHC, Fpo and HdrDE in electron transfer to Fe(III) by M. acetivorans are also hypothesized for ANME-2 in transfer of electrons to metal oxides or the syntrophic partner 19 . The MHCs of ANME-2 are proposed to extend beyond the S-layer to facilitate the electron transfer. The results presented here indicate that soluble Fe(III) citrate is reduced by the MHC of M. acetivorans . It is conceivable that the MHC also plays a role in reducing insoluble forms of Fe(III). If so, either multiple MHCs extend to the outer S-layer of M. acetivorans or diffusible electron carriers such as low-molecular-mass humic acids or secreted flavins mediate electron transfer between the MHC and the S-layer 7 , 51 . The latter is consistent with the finding that AQDS, a surrogate for humics, stimulated ferrihydrite reduction in whole cells of M. acetivorans . Extension of MHCs to the outer S-layer is consistent with MHCs from ANME-2a and ANME-2d reported to be fused with a putative S-layer domain homologous to the S-layer protein of M. acetivorans 19 . The pathway proposed for M. acetivorans is also consistent with the hypothesis that symbiotic associations of ANME and sulfate-reducing species evolved from methanogenic species that first acquired the ability to conserve energy by oxidizing methane coupled to reduction of metal oxides 7 . However, it is also possible that the ancestor common to M. acetivorans and ANME first performed AOM with a sulfate reducer and then switched to reduction of metal oxides. Unlike methanogenesis pathways, the proposed AOM pathway of M. acetivorans features multiple roles for heterodisulfide reductases (Fig. 5 ). Two types of heterodisulfide reductases are essential to all methanogenic pathways, the cytoplasmic HdrABC and the membrane-associated HdrDE 23 , 24 . The results presented here support a role for HdrDE coupled to AQDS-mediated reduction of Fe(III) that drives the endergonic methane oxidation and methyl transfer reactions essential to reversal. A role is proposed for the electron bifurcating HdrA2B2C2 in energy conservation when generation of Na + and H + gradients by Rnf and Fpo (Fig. 5 , Rxns. 3 and 9) are limited by the availability of Fe(III). The HdrA2B2C2 plays this role by diverting electron flow from F 420 H 2 to Fdx O promoting acetate production and ATP synthesis by substrate-level phosphorylation (Fig. 5 , Rxns. 5 and 6). Notably, the genome of the ANME ‘ Candidatus Methanoperedens nitroreducens’ encodes a homolog of HdrA2 that is the flavin-containing electron-bifurcating subunit of HdrA2B2C2 43 , 52 . A role for H 2 in electron transport and energy conservation is unlikely for Fe(III)-dependent AOM. M. acetivorans is incapable of metabolizing H 2 which contrasts with other species of Methanosarcina for which ATP synthesis is dependent on the production and consumption of H 2 to generate H + gradients 23 . The thermodynamically unfavorable production of H 2 as an electron transfer agent could impose a barrier to reversing methanogenesis. Metagenomic analyses of the ANME-2a clade, phylogenetically related to M. acetivorans , are devoid of hydrogenase genes 21 . Thus, the ANME-2a clade is unlikely to employ H 2 for transferring electrons to syntrophic partners. It is more likely that electrons are transferred via DIET (direct interspecies electron transfer) as previously hypothesized for uncultured species of the ANME-2 clade 19 . Pathways proposed for ANME include acetate as a product, although disputed as a diffusible electron carrier in syntrophic AOM 12 , 19 , 21 , 32 , 53 . If not involved in syntrophic AOM, acetate may be an essential carbon source for the syntrophic partner 21 . A portion of acetyl-CoA in the pathway of acetate production (Fig. 5 ) may also enter biosynthetic pathways essential for AOM by ANME and M. acetivorans 54 . In addition to AOM , M. acetivorans is capable of TMO producing CO 2 or acetate during net methane production when grown with methanogenic substrates in the absence of exogenous electron acceptors 13 , 14 . Although TMO is likely dependent on reversal of carbon transformation reactions in acetoclastic or CO 2 -reducing methanogenic pathways (Fig. 5 ), endergonic reactions of TMO are driven by energy conservation dependent on methanogenesis rather than reduction of Fe(III)."
} | 3,271 |
25048958 | null | s2 | 647 | {
"abstract": "Butyrate is an important product of anaerobic fermentation; however, it is not directly used by characterized strains of the highly efficient anode respiring bacteria (ARB) Geobacter sulfurreducens in microbial electrochemical cells. By combining a butyrate-oxidizing community with a Geobacter rich culture, we generated a microbial community which outperformed many naturally derived communities found in the literature for current production from butyrate and rivaled the highest performing natural cultures in terms of current density (∼ 11A/m(2)) and Coulombic efficiency (∼ 70%). Microbial community analyses support the shift in the microbial community from one lacking efficient ARB in the marine hydrothermal vent community to a community consisting of ∼ 80% Geobacter in the anode biofilm. This demonstrates the successful production and adaptation of a novel microbial culture for generating electrical current from butyrate with high current density and high Coulombic efficiency, by combining two mixed microbial cultures containing complementing biochemical pathways."
} | 270 |
36733583 | PMC9887407 | pmc | 648 | {
"abstract": "Introduction Biofuel is a kind of sustainable, renewable and environment friendly energy. Lignocellulose from the stems of woody plants is the main raw material for “second generation biofuels”. Lignin content limits fermentation yield and is therefore a major obstacle in biofuel production. Plant laccase plays an important role in the final step of lignin formation, which provides a new strategy for us to obtain ideal biofuels by regulating the expression of laccase genes to directly gain the desired lignin content or change the composition of lignin. Methods Multiple sequence alignment and phylogenetic analysis were used to classify PtrLAC genes; sequence features of PtrLACs were revealed by gene structure and motif composition analysis; gene duplication, interspecific collinearity and Ka/Ks analysis were conducted to identify ancient PtrLACs ; expression levels of PtrLAC genes were measured by RNA-Seq data and qRT-PCR; domain analysis combine with cis-acting elements prediction together showed the potential function of PtrLACs . Furthermore, Alphafold2 was used to simulate laccase 3D structures, proLAC23::LAC23-eGFP transgenic Populus stem transects were applied to fluorescence observation. Results A comprehensive analysis of the P. trichocarpa laccase gene ( PtLAC ) family was performed. Some ancient PtrLAC genes such as PtrLAC25 , PtrLAC19 and PtrLAC41 were identified. Gene structure and distribution of conserved motifs clearly showed sequence characteristics of each PtrLAC . Combining published RNA-Seq data and qRT-PCR analysis, we revealed the expression pattern of PtrLAC gene family. Prediction results of cis-acting elements show that PtrLAC gene regulation was closely related to light. Through above analyses, we selected 5 laccases and used Alphafold2 to simulate protein 3D structures, results showed that PtrLAC23 may be closely related to the lignification. Fluorescence observation of proLAC23::LAC23-eGFP transgenic Populus stem transects and qRT-PCR results confirmed our hypothesis again. Discussion In this study, we fully analyzed the Populus trichocarpa laccase gene family and identified key laccase genes related to lignification. These findings not only provide new insights into the characteristics and functions of Populus laccase, but also give a new understanding of the broad prospects of plant laccase in lignocellulosic biofuel production.",
"conclusion": "5 Conclusions In summary, sequence characteristics, structures and functions of the 53 members of the PtrLAC gene family have been elucidated in detail. They can be divided into 6 subfamilies according to their phylogenetic relationships with Arabidopsis and maize. Gene duplication analysis indicated that some colinear gene pairs separated 200 million years ago. Moreover, in the process of gene family expansion, fragment repetition is the main driving force in the early stage while tandem repetition plays this role in the late stage. Most laccase proteins are likely to be extracellular, but some have been shown to localize to cell membranes or to lysosomes and peroxisomes. Gene structure and motif analysis demonstrated that adjacent members in the phylogenetic tree had similar intro-exon structure and motif composition. Domain analysis, expression pattern analysis and cis-acting elements prediction shed light on the complex function of Populus laccase. Protein structure simulation and molecular docking analysis of PtrLAC12, PtrLAC23, PtrLAC25, PtrLAC41 and PtrLAC51 indicated that PtrLAC23likely to involved in the lignification process. Fluorescence observation of proLAC23:: LAC23-EGFP transgenic Populus stem transects showed that the location of PtrLAC23 highly overlapped with the lignin deposition. This indicates that PtrLAC23 may play an important role in Populus lignin synthesis. Plant laccase is of great significance to improve the utilization efficiency of lignocellulose. Regulating the expression of plant laccase can indirectly influence the content of lignin and cellulose, which can provide the theoretical support for us to obtain ideal biomass materials. These results also provides a broad prospect for us to obtain more ideal biofuels through genetic engineering in the future.",
"introduction": "1 Introduction Biofuels, as a kind of sustainable renewable energy instead of fossil energy, have been widely concerned in recent years. The “second generation” biofuel is made from inedible lignocellulose, through enzymatic hydrolysis of polysaccharides, release of monoses, fermentation, and finally converted into cellulosic ethanol for application ( Chang, 2007 ). Compared with the “first generation” biofuels which produce bioethanol by fermentation of edible parts such as sugar and starches, there is no competition with food supply, and pay more attention to ecological benefits in environmental protection ( Prem et al., 2021 ). According to United States Department of Energy, second-generation biofuels are expected to reduce greenhouse gas emissions by up to 96% ( Clark, 2008 ). The main component of lignocellulosic biomass is plant cell wall, which is composed of cellulose, hemicelluloses and lignin ( Cosgrove, 2015 ). In the process of biofuel conversion, lignin can easily adsorb cellulolytic enzymes and restrict the release of cellulose, thus directly limiting fermentation yield ( Zeng et al., 2014 ). Therefore, changing the content and composition of lignin in plants is an effective measure to boost biofuel production. Laccase (E.C.1.10.3.2), also known as polyphenol oxidase or urushiol oxidase, was first discovered in the SAP of Rhus Venicifera in 1883 ( Zoppellaro et al., 2001 ). As the largest subfamily of multicopper oxidases (LMCOs), laccase has a wide range of substrates and is widely distributed in bacteria, plants, insects and fungi ( Mate et al., 2016 ). Commonly, the molecular size of laccase concentrated in 60-130 KD, composed of 500-600 Aa ( Su et al., 2017 ). Although the amino acid sequence of laccase from different sources is quite different, their catalytic sites are relatively conserved ( Janusz et al., 2020 ). Crystal structure analysis of laccase protein has been found that laccase has three copper ion binding sites (T1, T2, T3) and four copper ions, which can be divided into a type-1 Cu, a type-2 Cu, and two type-3 Cu atoms ( Zoppellaro et al., 2001 ). It is speculated that the catalytic mechanism of laccase is copper ions at the active site of T1 absorb electrons from the reduced substrate, causing the substrate to be oxidized and form free radicals, leading to various non-enzymatic secondary reactions ( Sato et al., 2020 ). The current research on laccase mainly focused on its physicochemical properties and biological activities. Fungal laccase has been widely used in vitro lignin degradation ( Reyes et al., 2021 ), wastewater treatment ( Unuofin and Okoh, 2019 ) and dye decolorization ( Ahmed et al., 2020 ). Laccase CotA in Bacillus subtilis has been reported to play a role in pigment synthesis and protection against ultraviolet damage ( Hullo et al., 2001 ). Insect laccase has been shown to be associated with cuticle formation in insects ( Asano et al., 2019 ). However, the function of plant laccase need to be more determined. Several lines of evidence suggest that laccase can play a key role in plant lignification, which refers to the coupling of a lignin monomer with the terminal of a growing polymer ( Bao et al., 2016 ). Lignin monomers, also called monolignols, are the non-methoxylated p -coumaryl alcohol, the monomethoxylated coniferyl alcohol and the dimethoxylated sinapyl alcohol, which forms p-hydroxyphenyl (H), guaiacyl (G) and syringyl (S) units in lignin polymer, respectively. Laccase can catalyze the oxidative dehydrogenation of lignin monomers to form free radicals, once the radicals are generated, polymerization of lignin monomers will take place in a pure chemical reaction, no longer catalyzed by enzymes or proteins ( Ralph et al., 2019 ). There has been several evidence that laccase is closely associated with lignification. Berthet et al. have found that LAC4 and LAC17 played important roles in the lignification process of Arabidopsis stems, silencing of these two genes significantly reduced the biosynthesis of lignin ( Berthet et al., 2011 ). In addition, simultaneous knockout LAC11 , LAC4 , and LAC17 in Arabidopsis severely affected plant growth, resulting in narrowing of root diameter, closure of anthers, stagnation of vascular tissue development, and a significant reduction in lignification ( Zhao et al., 2016 ). Richard Dixon et al. have reported that GhLAC8, a laccase specifically expressed in the seed coats of Cleome hassleriana , which can facilitate the polymerization of lignin in plants by using caffeyl alcohol as the sole substrate, and determined the lignin component content in plant cell walls ( Wang et al., 2020 ). Except for the role in lignification, laccase has many other functions also worth exploring ( Wang et al., 2015a ). Results show that expression of GhLac1 in transgenic cotton could enhance the defense response of cotton to against pathogens and pests ( Acid et al., 2018 ). Almost all of the GmLAC genes in soybeans are involved in responding to P.sojae infection ( Wang et al., 2019 ). Some researchers inserted Arabidopsis laccase gene AtLAC15 promoter into Brassica napus , and found that AtLAC15 promoter could be used as a seed coat-specific promoter for canola. In addition, this coat-specific promoter can help to overexert or inhibit the expression of some genes, which can alter the metabolism and producing seed with reduced fiber content ( Wu and Saleh, 2011 ). In recent years, Batch anaerobic digestion experiments on 41 energy crops showed that 80% of the sample biofuel yield variation could be explained by lignin content ( Dandikas et al., 2014 ). In addition, inhibition of lignin biosynthesis pathway gene Pt4CL1 in transgenic poplar ( Populus tremuloides Michx. ) showed that there is a compensatory mechanism between lignin and cellulose, that is, inhibition of lignin synthesis can increase the accumulation of cellulose ( Hu et al., 1999 ). From the perspective of energy utilization, treatment of laccase in plants can inhibit lignin deposition and increase cellulose content, which provides a new idea for the improvement of energy plants ( Mai and Kharazipour, 2001 ). As an important biomass energy source ( Dandikas et al., 2014 ), woody plant is of great significance to analyze the function and mechanism of its laccase gene. To date, the Populus trichocarpa laccase gene family members have been identified ( Lu et al., 2013 ) ( Bryan et al., 2016 ), but their sequence characteristics, structures and functions have not been elucidated in detail. Here, we choose Populus trichocarpa as the research object, through collinearity analysis, gene structure analysis, motif and domain analysis, expression pattern analysis, cis-acting elements and protein structure prediction and other detailed analyses of Populus trichocarpa laccase gene family, to find out some PtrLACs which involved in plant lignification, in order to provide more directions for improvement of energy trees.\n\n3.3 Exon-intron structure, motif distribution and domain composition of Populus laccase gene family To better understand the structural features of PtrLACs and gain more insight into the evolution of the laccase family in P. trichocarpa , we analyzed the intron-exons composition of all sequences as shown in \n Figure 4 \n . All the sequences in the group 1, group 3 and group 4 contains 5 introns and 6 exons, most sequences in group 2 also have such structure, except PtrLAC8 , which contains 4 introns and 5 exons. The sequences in group 5 and most of the sequences in group 6 have four introns and five exons. Notably, in group 6, PtrLAC51 contained only 3 exons and 2 introns, while PtrLAC36 contained 7 exons and 6 introns. Overall, the family members with closer genetic relationships have more similar gene structures ( \n Figure 4B \n ). Figure 4 Phylogenetic relationships, gene structure and architecture of conserved protein motifs of P. trichocarpa laccase gene family. (A) Nneighbor-joining tree was constructed based on the full-length sequences of P. trichocarpa laccase proteins using MEGA 10.0 software. (B) Exon-intron structure of P. trichocarpa laccase genes. Blue boxes indicate untranslated 5′- and 3′-regions; yellow boxes indicate exons while black lines indicate introns. (C) The motif composition of Populus trichocarpa laccase proteins. Colored boxes represent the different motifs, indicated in the top right-hand corner. The scales at the bottom of the image indicate the estimated exon/intron length in kb and motif length in numbers of amino acids (aa), respectively. To further clarify the characteristics of the Populus laccase family, we detected 15 conserved motifs in 53 sequences and analyzed their distribution in combination with the evolutionary tree. As expected, most of the laccase protein sequences in Populus trichocarpa had all 15 motifs. However, in group 6, none of the sequences had motif 15. In addition, PtrLAC51 also lost motif 2 and motif 6, PtrLAC36 doesn’t include motif 6, 9 and 11. In group 4, PtrLAC3 doesn’t have motif 15, and PtrLAC16 just contains six motifs (motif 6, 2, 3, 8, 5 and 1). These differences suggest that family members of the two branches may have lost part of their motifs during evolutionary process, resulting in a new function. ( \n Figure 4C \n ). In order to define the function of each member of the Populus laccase gene family, we use the Pfam database ( https://pfam.xfam.org/search ) to analyze the domain of each sequence. Results showed that all the members have Cu-oxidase, Cu-oxidase_2, and Cu-oxidase_3 (PF00394, PF07731, and PF07732) these three conserved domains, which associated with the redox reaction of substrates. Beyond that, we also detected 9 domains and some of them have specific functions: YAF2_RYBP was a C-terminal binding motif, which is first found in YAF2 and RYBP proteins and usually forms the RYBP -/YAF2-PRC1 complex ( Wang et al., 2015b ); ResB domain was related to the synthesis of cytochrome C and is essential for plant growth ( Le Brun et al., 2000 ); NAR2 is a plant protein with a C-terminal transmembrane region, which often works with NRT2 and can transport nitrate at low concentration, this NAR2-NRT2 system plays an important role in the regulation of lateral root growth ( Yong et al., 2010 ); CBiM is a substrate specific component of the cobalt transport complex CBiMNQO, which is related to the synthesis of coenzyme B12 ( Santos et al., 2008 ) ( \n Figure 5 \n ). Laccases with these special domains may enable to perform other specific functions in plants, and should be the focus of future research. Figure 5 Phylogenetic relationships and conservative domain distribution of the PtrLACs family. (A) Nneighbor-joining tree was constructed based on the full-length sequences of P. trichocarpa laccase proteins using MEGA 10.0 software. (B) Domain distribution of P. trichocarpa laccase proteins. Heterochromatic boxes represent different domains, indicated in the top right-hand corner. The length of protein can be estimated using the scale at the bottom.",
"discussion": "4 Discussion Laccases, as the largest subfamily of the polycopper oxidase (LMCOs) protein family, have a wide variety of substrates and are widely distributed in plants, insects and microorganisms. Previous studies have shown that plant laccase plays an important role in the oxidation and polymerization of lignin monomers. Under aerobic conditions, laccase can catalyze the oxidative dehydrogenation of lignin monomers to form free radicals, thus initiating the spontaneous polymerization process between monomers. Therefore, the regulation of plant laccase gene expression can indirectly regulate the composition and content of lignocellulose, which provides an effective basis for the improvement of energy trees. Populus trichocarpa is a model plant in forest research. Although its laccase gene family has been identified, detailed analysis of family members was still relatively scarce. Gene duplication is one of the main causes of gene family formation. There were 22 genes involved in tandem duplication and 22 genes involved in fragment duplication in Populus laccase gene family, indicating that the two gene duplication events played an equally important role in the expansion of PtrLAC gene family. In addition, the Ka/Ks values of almost all PtrLAC duplicate gene pairs were less than 1, illustrating that PtrLACs were underwent a strong purifying selective pressure during the evolution. This conclusion is consistent with the results obtained in Pyrus bretschneideri ( Cheng X et al., 2020 ), Glycine max ( Wang et al., 2019 ) and Setaria viridis ( Simões MS et al., 2020 ), suggesting that this may be a general rule in the plant kingdom. Combined with the molecular evolution rate of Populus , we estimated the replication time of collinear gene pairs and found that fragment duplication and tandem duplication were the main driving force of gene family expansion in the early and late stages, respectively. In addition, most PtrLACs were replicate after the WGD event, this is different from the results in soybean, which laccase genes replication typically occured during WGD events ( Wang et al., 2019 ). This result also shows that the laccase gene family has different expansion modes in different species. Moreover, 20% of the duplicate gene pairs were separated before 150 MYA, indicating that laccase genes have a very long history in Populus , which also allows us to Preliminary lock a series of ancient PtrLACs represented by PtrLAC25 and PtrLAC41 . In the early study, Arabidopsis laccase gene family was divided into 6 subfamilies, and most of the subsequent studies were based on this criterion. In our study, PtrLACs were distributed in all six subfamilies, in which gene duplication was the main cause of the expansion in group 4, group 5, and group 6. The function of acidic or basic isoforms of laccase in lignification has always been controversial, it is generally believed that acidic peroxidases have poor ability to catalyze sinapyl alcohol ( Barceló AR et al., 2007 ). The evolutionary analysis also revealed that acidic PtrLACs mainly clustered within group 3 and group 4, which happened to be the group of ZmLAC1 and ZmLAC3, respectively. Moreover, ZmLAC3 as an acidic laccase has been confirmed to be involved in injury-induced lignification in maize. Therefore, we hypothesized that the acidic laccases in Populus may be functionally specific, and the acidic laccases belonging to the group 3 may also play a role in the process of defense-induced lignification. Although, ZmLAC2, ZmLAC4, and ZmLAC5, which belong to the group 2 have not been confirmed to be involved in the process of injury-induced lignification, their expression patterns were similar to the one expected for genes involved in lignification ( Caparrós-Ruiz et al., 2006 ). Therefore, we speculated that they might play a role in the developmental lignification, and PtrLACs in group 2 may also potentially involved in this process. In conclusion, we believe that PtrLACs in group 3 and 4 may be involved in the defense-induced lignification in Populus , and PtrLACs in Group 2 may be involved in the developmental lignification in Populus . Motif is generally considered to be conserved sequence with biological functions, which may contain specific binding sites or common sequence segments involved in a specific biological process. Adjacent PtrLACs on the evolutionary tree generally have more similar gene structure and motif composition. However, a few of members in group 4 and group 6 have lost part of motifs, indicating that these laccases may have acquired new functions in the evolution. In fact, loss events of genes occur frequently during the evolution. Due to the whole-genome triplication (WGT), genes in Arabidopsis should have three homologs in B. rapa and B. oleracea , but in fact, some AtLACs can not even have homologs in either species (Lu et al., 2019). The genome of Populus is about four times as large as that of Arabidopsis , while the number of laccase genes in Populus is only three times as large as in Arabidopsis . All of these phenomena indicate that gene loss events existed in LAC gene family universally. A large number of studies have shown that laccase exists widely in plants and is of great significance for maintaining plant growth and development. Synteny analysis between Populus and five other species showed that some PtrLACs ( PtrLAC11, PtrLAC25, PtrLAC20, PtrLAC19 ) may have existed before the divergence of monocots and dicots ancestors. Combined with the estimation of the replication time of fragment duplication genes, we found that PtrLAC25 is not only a relatively old gene in Populus , but also has homology with laccase genes in five other species, which deserves more attention. Gene expression pattern can reflect the function of genes to a certain extent. RNA-Seq data showed that most PtrLAC genes were highly expressed in nodes, internodes and catkins, which again indicated the important role of laccase in the polymerization of lignin monomers. Lu et al. have found that 49 laccase genes were all expressed in the 6-month-old Populus . However, we could only detect the expression of a few of laccase genes in the early stage of Populus secondary growth. PtrLACs ( PtrLAC5, PtrLAC7, PtrLAC23, PtrLAC12, PtrLAC18 and PtrLAC22 ) specifically expressed in roots and stems are both members of group 1 and group 2, which is consistent with the findings in Setaria ( Simões MS et al., 2020 ). Among them, PtrLAC5 , PtrLAC7 , PtrLAC23 and PtrLAC12 are both members of group 2. The specific expression of these genes in stem, a lignin-accumulating tissue, again proves the conclusion that members of group 2 may be related to the developmental lignification in Populus . In addition, expression level of some PtrLAC genes were significantly higher in roots. These laccase genes, with the exception of PtrLAC25 , all belong to group 3 and group 4. Cotton laccase GhLAC1 is also preferentially expressed in roots. Previous studies have shown that it was involved in the lignification process induced by defense and thus improves the resistance of cotton to biotic stress ( Acid et al., 2018 ).Therefore, we believe that these PtrLACs , which are mainly expressed in roots at the initial stage of secondary growth, may also play a potential role in plant defense besides participating in root lignification. This also suggests that members of group 3 and group 4 are likely to be involved in defense-induced lignification in Populus . It has been shown that there is a dynamic equilibrium between photosynthesis and cell wall synthesis, and copper ion concentration in plants is the key to regulate this equilibrium. When copper ions are deficient in plants, miRNA ( AtmiR397 , AtmiR398 , AtmiR408 , AtmiR857 in Arabidopsis ; PtmiR397 , PtmiR408 , PtmiR1444 in Populus ) will inhibit the expression of other copper-containing proteins, and preferentially supply copper to the essential copper-containing proteins in the photosynthetic electron transport chain represented by PC. Cis-acting elements prediction showed that PtrLAC genes had a large number of photoresponsiveness cis-acting elements. In addition, the activity of laccase also depends on the copper ion center, which suggests that there is a relationship between photosynthesis, copper ion concentration and the regulation of laccase gene expression. We also found that MYB protein may be involved in this process as a transcription factor. The network mechanism of MYB family members regulating plant secondary cell wall formation has been fully revealed. However, there are still few reports on the interaction between MYB protein and laccase, which needs to be verified by subsequent studies ( Xiao et al., 2021 ). Understanding the structure of a protein is crucial for defining its function. With the development of artificial intelligence (AI), AlphaFold2 made a breakthrough in the protein folding problem ( Marcu et al., 2022 ). In CASP14, AlphaFold2 achieved a remarkable result. Some of its predicted protein structures were almost indistinguishable from the experimental results ( Flower and Hurley, 2021 ). ZmLAC3 in maize is the only plant laccase which structure has been resolved. Moreover, ZmLAC3 has been proved to catalyze the oxidative polymerization of lignin monomers. The structure prediction of five Populus laccase proteins showed that the overall structure of the PtrLAC12, PtrLAC23, PtrLAC25, PtrLAC41 and PtrLAC51 is very close to that of ZmLAC3. These findings indicate that laccase is structurally conserved, and suggest that these proteins may play a potential role in the polymerization of lignin monomers. In addition, PtrLAC23 has the strongest affinity to lignin monomer ConA, which may be a key laccase related to lignification in Populus . At present, there is a lot of evidence that laccase is involved in cell wall lignification in Populus . Bryan et al. found that PdLAC2 plays a role in overcoming plant cell wall recalcitrance in Populus deltoides ( Bryan et al., 2016 ). Liu et al. showed that PtrLAC16 can polymerize lignin monomers in vitro , inhibition the expression of PtrLAC16 led to a significant decrease in lignin content and altered cell wall structure ( Liu et al., 2021 ). Niu et al. proved that overexpression of PeuLAC2 increased the secondary cell wall thickening, fiber cell length and stem tensile strength ( Niu et al., 2021 ). Tissue specificity analysis of these laccases showed that PtrLAC16 is specifically localized in the xylem and phloem of the stem. Expression of PeuLAC2 in P. euphratica tissues mainly occurred in the xylem. In P. deltoides , LAC2 had the highest expression in xylem compared to other tissues. In our study, fluorescence observation of cross sections of proLAC23::LAC23-EGFP line showed that the expression localization of PtrLAC23 was mainly in xylem and a few areas of the phloem, and was highly consistent with the deposition site of lignin. These results are similar to the tissue localization analysis of laccases above, which again proved the correlation between PtrLAC23 and Populus wood formation. In addition, relative expression of PtrLAC23 was significantly increased in the stem of 80-day-old Populus than that of young Populus , suggesting that the role of PtrLAC23 in Populus stem’s secondary growth should not be ignored. The predicted results showed that PtrLAC23 was likely an extracellular protein. Subcellular localization of PtrLAC16 and PeuLAC2 both showed that laccase protein was probably secreted extracellular and transported to the cell wall to play a role ( Niu et al., 2021 ) ( Liu et al., 2021 ). Here, we also observed the fluorescence of proLAC23::LAC23-EGFP transgenic Populus protoplasts ( \n Figure S2 \n ). In our results, only chloroplast spontaneous fluorescence was observed and no EGFP signal was found, which also confirmed that PtrLAC23 should be an extracellular protein that probably be transported to play a role on the cell wall."
} | 6,882 |
29463894 | PMC5955907 | pmc | 649 | {
"abstract": "Ocean warming is resulting in increased occurrence of mass coral bleaching; a response in which the intracellular algal endosymbionts ( Symbiodinium sp.) are expelled from the coral host due to physiological stress. This detrimental process is often attributed to overproduction of reactive oxygen species (ROS) that leak out of the endosymbionts and causes damage to the host cell, though direct evidence validating this link is limited. Here, for the first time, we used confocal microscopy and fluorescent dyes to investigate if endosymbiont ROS production significantly and predictably affects physiological parameters in its host cell. Heat treatment resulted in a 60% reduction in coral symbiont density, a ~70% increase in median endosymbiont ROS and a small reduction in photosystem efficiency ( F V / F M , 11%), indicating absence of severe light stress. Notably, no other physiological parameters were affected in either endosymbionts or host cells, including reduced glutathione and ROS-induced lipid peroxidation. Taken together, the increase in endosymbiont ROS could not be linked to physiological damage in either partner, suggesting that oxidative stress is unlikely to have been the driver for symbiont expulsion in this study.",
"introduction": "Introduction The global success of corals in tropical waters is underpinned by their symbiotic partnership with photosynthetic intracellular dinoflagellate microalgae ( Symbiodinium sp.), which supply the animal host with photosynthetically fixed carbon in exchange for nutrients [ 1 ]. For this mutualistic association to prosper, corals need to balance energy expenditure (nutrient delivery to symbiont) with energy gain (fixed carbon from the symbiont). To achieve this, corals actively regulate their symbiont density both on a seasonal basis and in response to local environmental conditions [ 2 – 4 ]. However, when conditions alter sufficiently to cause physiological stress, it can lead to a phenomenon known as coral “bleaching”; the whitening of corals due to detrimental breakdown of the coral symbiosis and consequent loss-of-symbiotic algae from the host tissue and/or loss-of-pigment [ 5 , 6 ]. To date, numerous environmental triggers of coral bleaching have been identified including, but not limited to, changes in temperature and light [ 7 , 8 ], as well as nutrient imbalance [ 9 , 10 ]. However, complete understanding of the cellular mechanisms responsible for the breakdown of the symbiotic partnership is still lacking [ 11 , 12 ]. In recent decades, increasing ocean temperatures have resulted in extensive loss-of-coral cover around the globe, reflecting post bleaching mortality [ 13 ]. This mass bleaching mortality is considered to be driven by photo-oxidative stress in the coral symbionts following temperature and light-induced damage to the photosystem [ 11 , 14 , 15 ]. According to the oxidative theory of coral bleaching, reactive oxygen species (ROS) produced through activity of the damaged photosynthetic machinery in the symbiont, leak into the host cell where it overwhelms the antioxidant system and causes damage to the host tissue [ 14 , 15 ]. By expelling the symbiont, and thus the source of excess ROS, the host minimises the level of physiological damage under suboptimal environmental conditions [ 11 , 16 ]. In cells, ROS are produced in the mitochondria and chloroplasts [ 17 ], where superoxide dismutase (SOD) works as the first line of defence in an important detoxification process [ 18 ]: the SOD converts superoxide anions (O 2 − ), produced through one-electron reduction of O 2 , to the more stable hydrogen peroxide (H 2 O 2 ). To limit build-up of this cytotoxic compound, H 2 O 2 is quickly reduced to harmless H 2 O and O 2 by a broad arsenal of enzymatic and non-enzymatic antioxidants, including the high-affinity glutathione peroxidase (GPx) in conjunction with the co-factor glutathione, as well as ascorbate peroxidase (APX) and the low affinity catalase enzyme. However, under sub-optimal conditions (i.e., stressful) the local concentration of ROS can increase because of an offset between production and this dissipation mechanism. If the production of ROS is higher than the reductive potential of the antioxidant system, ROS may lead to several forms of cellular damage, including lipid peroxidation (membrane damage), protein oxidation and DNA degeneration [ 17 ]. The proposed importance of ROS in the physiological mechanism leading to mass coral bleaching is buoyed by evidence of temperature and light-induced production of ROS and reactive nitrogen species (RNS), antioxidant upregulation in both host and symbionts [ 15 , 19 – 22 ], as well as a reduction of bleaching in presence of exogenously applied antioxidants [ 23 ]. Additionally, observations of mitochondria and chlorophyll degradation in isolated Symbiodinium [ 24 , 25 ] and host tissue [ 26 ] support the role of severe physiological disorder in the bleaching process. However, despite significant effort over recent decades to verify the implication of ROS in coral bleaching, the extent of the relationship between symbiont ROS production and host physiological stress remains ambiguous. In particular, recent work has found the host to increase its antioxidants in response to heat stress prior to its symbionts and preceding photo-physiological decline [ 27 , 28 ], demonstrating that photosystem stress may in fact be a late stage response in the bleaching process and not the initial driver of expulsion. Irrespective, the initial increase in host antioxidants could suggest a role of ROS in the early bleaching process, which remains to be further investigated. One of the major obstacles to uncovering the mechanisms of coral bleaching is the inherent difficulty of separating the response of the host from that of the intracellular symbiont. Til now, this has been achieved only by physically separating the host and symbionts, destroying the host tissue in the process and consequently averaging analyses across multiple host cell types, most of which are non-symbiotic. However, the progressive loss-of-symbionts from coral tissue during bleaching suggests that it happens on a cell by cell basis, highlighting that coral bleaching is fundamentally a single-celled process occurring on a massive scale. Therefore, averaging measurements across cell types may result in substantial bias or a reduction in the observed responses of the specific cells directly involved in the bleaching process. To adequately separate the stress response of the symbiont from that of the host and to attempt to identify any resulting interactions between the two partners, we propose that it is necessary and advantageous to address the question at the scale of the individual cell. Here we explore, for the first time, the process of coral bleaching at the single-cell level, utilising intact symbiotic cells (host with symbionts) to study heat-induced ROS production and its metabolic consequences in the important reef-building coral Pocillopora damicornis . Measuring key stress indicators in >1700 individual cells using confocal microscopy, we aimed to disentangle the thermal stress responses of these closely entwined symbiotic partners.",
"discussion": "Discussion Reactive oxygen species are an inevitable by-product of electron transport in both mitochondria and chloroplasts. Therefore, all aerobic organisms have developed mechanisms to effectively dissipate ROS, thereby limiting damage to vital cellular components [ 17 , 18 ]. Increased concentration of cellular ROS is an intrinsic consequence of raised metabolic activity and/or reduction in energy dissipation efficiency associated with physiological stress, which explain the observation of heat-induced increases in ROS production and antioxidant expression in corals [ 15 , 20 – 23 ]. As such, the question remains whether the observed correlation of ROS production and antioxidant upregulation with the breakdown of the coral symbiosis is a result of true causality. To this end, evidence confirming a direct detrimental effect of endosymbiont derived ROS on the host metabolism during heat stress is still lacking. In this study, prolonged heat stress of Pocillopora damicornis resulted in significant loss-of-symbionts (i.e., bleaching) (Fig. 2 ). Consistent with previous studies on free-living Symbiodinium sp. exposed to acute heat stress [ 38 , 39 ], elevated temperatures resulted in increased ROS production by the endosymbionts (Fig. 4 ) and a significant but minor decline in the maximum quantum yield of photosystem II (PSII) ( F V / F M , Fig. 2b ). This decline may have been driven by production of singlet oxygen ( 1 O 2 ) in the chloroplasts, which is thought to be the most important ROS responsible for light-induced loss-of-PSII activity [ 40 ], however, the relatively small change in F V / F M indicates that the photosystem was not severely compromised at this stage in the bleaching process. An increase in endosymbiont ROS production to the extent where a significant proportion of non-quenched ROS leaks into the host would be expected to be accompanied by damage to cellular components and changes to the antioxidant capacity of the cell. Yet, we found no additional signs of ROS-related physiological stress (Fig. 4 ). We observed no consistent reduction in the fluorescence of the Monobromobimane dye (mBBr), suggesting that the pool of reduced glutathione (GSH) was maintained. Glutathione plays a crucial role in the cells antioxidant system, where it is oxidised by ROS to glutathione-disulfide (GSSG) directly (chemically) or enzymatically through the action of ascorbate and/or glutathione peroxidases (APX and GPX, respectively) [ 41 ]. Because of its paramount role in this redox cycle, GSH is quickly re-generated by reduction of GSSG with NADPH and glutathione reductase (GR) to maintain the cells redox state [ 41 ]. In Symbiodinium , sub-lethal heat stress has been shown to increase activity of GR, resulting in maintenance of the GSH pool despite increased reactive oxygen production [ 42 ]. However, this protective mechanism collapsed under more severe heat stress [ 42 ], and as such, any significant reduction in the level of GSH can be considered a strong indicator for excessive oxidative stress. In support of the lack of change in the glutathione pool, we found no evidence for ROS-induced damage through lipid peroxidation (Fig. 4 ), which is considered the principle molecular mechanism involved in the oxidative damage to cell structures [ 43 ]. Also, we observed no significant decline in relative electron transport rates of PSII (Fig. 2a ) and no change in the level of chlorophyll fluorescence per cell volume, indicating little or no chlorosis (Fig. S 8 ). Taken together, these data strongly suggest that the symbiont antioxidant system was effective at protecting itself against uncontrolled oxidative damage despite the substantial, heat-induced increase in steady-state ROS content. The presence of “free” symbionts (here termed “ex-symbionts”) in coral tissue has been described previously, where non-host associated symbionts were found to account for up to 30% of the total symbiont count within the tissue of naturally bleached corals [ 24 ]. While it was not possible to determine whether these cells had recently been expelled from a host cell or if they were non-symbiotic cells residing freely within the coral host, the ex-symbiont cells were found to contain more lipid bodies than endosymbiotic cells (2–3×, data not shown), showing that, while a fraction of the cells may have been released from their hosts during the extraction procedure, most of the cells were physiologically different from the remaining endosymbionts. Other studies have found that temperature stress increases lipids and fatty acids in endosymbiont [ 44 ] and expelled symbionts (Petrou K, personal communication), which demonstrates that expulsion of endosymbionts may be correlated with an increase in lipids. This supports the notion that the ex-symbionts observed here may indeed have been expelled from host cells but not yet excluded from the coral tissue. Despite a higher median increase in ROS production (~100% in pooled cells) with heat treatment (Table S4 ), the ex-symbionts did not exhibit significant changes to other ROS-related physiological parameters measured here. Additionally, heat-treated ex-symbionts showed no consistent changes to mitochondrial activity as measured via proton-gradient dependent fluorescence. Given the high content of poly-unsaturated fatty acids and peroxidation catalysts, mitochondria are specific targets for lipid peroxidation damage, which may reduce the electron transfer rate and thus overall activity [ 45 ]. As such, the lack of change with heat treatment substantiates the results of little overall ROS-related physiological stress in these cells, and corroborates the results from the endosymbiont data. Importantly, these data do not preclude that under scenarios in which photosystem damage does occur, ROS could be an important trigger for bleaching due to the breakdown of important cellular and/or symbiotic functions, as suggested by the oxidative theory of coral bleaching. Instead, it highlights that bleaching may occur well before the onset of photosystem damage in the symbionts, perhaps as a result of alterations to the host-regulated nitrogen input to the symbionts as a consequence of physiological stress in the host [ 9 , 46 ]. With the exception of a few individual cells, the heat treatment did not significantly increase the ROS concentration in the host, although, we cannot exclude the possibility of upregulation of ROS that is not detected by the dye employed. For instance, the important cell-signalling molecule NO has previously been shown to increase in corals in response to heat stress [ 47 ], but is not detected by CM-H 2 DCFDA. Similar to the ex-symbionts, we observed no consistent changes to the total mitochondrial fluorescence in the host cells (Fig. 4 ), indicating that the heat treatment employed in this study had little general effect on mitochondrial activity, and this was supported by the esterase activity showing that cell viability was retained [ 48 ]. This contrasts with previous studies that showed loss-of mitochondrial integrity [ 26 ] and caspase-induced apoptosis [ 49 ] in host cells in response to acute heat stress. However, the extreme ramping (6–7 °C over a few hours) and sustained high temperatures (32–33 °C for ≥24 h) employed in these studies likely exacerbated the extent of damage to cellular components by not allowing time for adequate metabolic adjustments, including de novo synthesis of heat shock proteins [ 50 ]. The importance of heating rates and how it might influence the severity of the bleaching response has been discussed previously [ 28 , 51 ]. Given the minimal effect on the physiological state of host cells in the present study, it indicates that little if any ROS-induced damage had occurred despite the prolonged incubation at increased temperature and natural light cycles. The observed correlation between host and endosymbiont ROS in the heat-treated cells (Fig. 5a ) may be explained either by a joint increase in net ROS due to metabolic stress in both partners simultaneously or indeed by diffusive transfer of ROS from one symbiotic partner to the other. However, based on the available data it is not possible to determine which of these mechanisms are more likely. While the pooled control cells only showed a weak correlation between host and symbiont ROS, heat treatment resulted in the generation of a sub-population (~15% of total) of highly fluorescent cells from two of the three coral replicates; indicative of high-ROS production. From this it is tempting to propose that the bleaching process is in fact driven by a small but continuously replenished sub-population of cells reaching a stress-threshold resulting in the expulsion of the symbiont. This would explain the difficulty in obtaining consistent results when investigating coral bleaching using whole-tissue techniques, as the changes in such a relatively small population of cells within a larger pool of less affected cells would invariably reduce the overall observed response. However, the same pattern was not observed for reduced glutathione where the correlation between symbiont and host remained unchanged with no formation of sub-populations (Fig. 5b ). The data presented here are the first of their kind to directly target the link between ROS-related physiological variables of the coral endosymbiont with that of its respective host cell at the cell–cell interface. The results show that the coral endosymbionts are not severely compromised by heat-induced ROS production, and emphasise that, in the absence of severe photosystem damage, ROS leakage from the endosymbiont to the host and subsequent ROS-induced damage is not likely to be the initial trigger for symbiont expulsion during heat-induced bleaching."
} | 4,266 |
34723168 | PMC8536778 | pmc | 651 | {
"abstract": "Graphical abstract",
"introduction": "Introduction Biomineralization, the formation of minerals by living organisms, is a well-described process in all kingdoms of life. Although biomineralization covers many elements, most studies focus on calcium-based minerals. Iron is a fascinating biomineralization element that, in contrast to calcium, can create magnetic minerals. MTB are a diverse group of aquatic, magnetic responsive, gram-negative bacteria that can biomineralize iron into magnetic nanoparticles ( Amor et al., 2020 ). MTB organize the magnetic nanoparticles into a chain that creates a sufficient magnetic dipole moment to orient themselves according to the geomagnetic field. It is hypothesized that this ability reduces their random movement to a single dimension, which allows them a better way to navigate toward their optimal habitat around the oxic-anoxic interface, referred as magnetotaxis ( Müller et al., 2020 ). MTB biomineralize iron into the single domain magnetic nanoparticle, magnetite (Fe 3 O 4 ) and/or greigite (Fe 3 S 4 ) ( Staniland and Rawlings, 2016 ). This process takes place in a unique sub-cellular membranous organelle, the magnetosome. The resulting magnetic nanoparticles are not uniform among all MTB and vary in shape and size ( Fig. 1 A). The nanoparticles size is usually in the range of ∼35 − 120 nm, which fits well with the size of a single magnetic domain ( Nudelman and Zarivach, 2014 ). Fig. 1 A) Schematic representation of magnetotactic bacteria (MTB) from different phyla and classes. MTB divide into different types based on their genetics. Additional differences exist in their morphological characteristics, such as cellular shape, magnetosome chain arrangement, number of chains, number of flagella, magnetic crystal composition (magnetite or greigite) and crystal shape (Cuboctahedral, elongated-prismatic, bullet, elongated bullet). B) Four operons from the “magnetosome island” of MSR-1. The genes marked by a single letter represent the genes and their encoded proteins that are discussed in this review. Created with BioRender.com. Magnetosome formation in MTB is genetically controlled and involves a large set of specialized genes. These genes are usually organized in operons, located in a genomic island known as magnetosome island ( Fig. 1 B) ( Lohße et al., 2014 ). These genes encode, primarily, membrane proteins and are named magnetosome membrane-associated (Mam) and magnetic particle membrane-specific (Mms) proteins in Alphaproteobacteria . Most of our knowledge is based on two fresh-water cultivated Alphaproteobacteria , Magnetospirillum magneticum, and Magnetospirillum gryphiswaldense . These bacteria share many conserved genes and a common magnetosome formation mechanism. In both, the magnetosome formation is based on four main steps: 1) Protein sorting and membrane invagination, 2) Magnetosomes alignment into a single or multiple chains 3) Ion transport and magnetosome inner environment control, and 4) Iron nucleation and crystal shape and size control. The last three steps are described separately to simplify the mechanism, while they most likely occur simultaneously and not sequentially. In this brief graphical review, we will discuss the main findings in the field of MTB and highlight the four magnetosome formation steps from the Alphaproteobacteria biomineralization perspective."
} | 841 |
39668362 | PMC11741097 | pmc | 652 | {
"abstract": "The application of synthetic phototrophic microbial consortia\nholds\npromise for sustainable bioenergy production. Nevertheless, strategies\nfor the efficient construction and regulation of such consortia remain\nchallenging. Applying tools of genetic engineering, this study successfully\nconstructed a synthetic community of phototrophs using Rhodopseudomonas palustris ( R. palustris ) and an engineered strain of Synechocystis sp PCC6803\nfor acetate production ( Synechocystis_acs ), enabling\nthe production of biohydrogen and fatty acids during nitrogen and\ncarbon dioxide fixation. Elemental balance confirmed carbon capture\nand nitrogen fixation into the consortium. The strategy of circadian\nillumination effectively limited oxygen levels in the system, ensuring\nthe activity of the nitrogenase in R. palustris , despite oxygenic photosynthesis happening in Synechocystis . When infrared light was introduced into the circadian illumination,\nthe production of H 2 (9.70 μmol mg –1 ) and fatty acids (especially C16 and C18) was significantly enhanced.\nProteomic analysis indicated acetate exchange and light-dependent\nregulation of metabolic activities. Infrared illumination significantly\nstimulated the expression of proteins coding for nitrogen fixation,\ncarbohydrate metabolism, and transporters in R. palustris , while constant white light led to the most upregulation of photosynthesis-related\nproteins in Synechocystis_acs . This study demonstrated\nthe successful construction and light regulation of a phototrophic\ncommunity, enabling H 2 and fatty acid production through\ncarbon and nitrogen fixation.",
"introduction": "1 Introduction Phototrophic microbial\ncommunities are commonly found in light-exposed\nenvironments. 1 Such light-driven consortia\ncontribute substantially to global primary production of organic compounds\nby fixing carbon dioxide and/or nitrogen gas. 2 With humankind facing ever-growing energy demands and environmental\nproblems, 3 such synthetic phototrophic\nconsortia may provide a promising alternative to current energy generation\nmethods. These consortia can efficiently convert CO 2 and\nN 2 gases together with water and solar energy into products\nof bioenergy. 2 , 4 In general, microbial consortia\nare also attracting more attention due to their ability for specialization\nand labor sharing, allowing them to achieve more complex and stable\nphenotypes compared to monocultures. 5 , 6 Nevertheless,\nwhen trying to assemble a synthetic consortium outside the complex\nenvironmental constraints of a given habitat, a single strain in the\nconsortium might become dominating, challenging the consortium stability.\nThus, efficient strategies for maintaining strain balance and controlling\nthe performance of consortia are major challenges. Phototrophic\nmicroorganisms can be differentiated into water-splitting\n(oxygenic) and nonwater-splitting (anoxygenic) phototrophs. The former\ngroup includes cyanobacteria and microalgae, while the latter includes,\nfor instance, purple nonsulfur bacteria (PNSB). Rhodopseudomonas\npalustris ( R. palustris ) is a prominent member of PNSB, which harbors diverse types of nitrogenases\nenabling N 2 fixation under anoxic conditions with obligatory\nproduction of hydrogen gas. 7 Despite being\nregarded as a promising candidate for biohydrogen and lipid production, 8 , 9 the sufficient provision of organic carbon (e.g., acetate), which\nserves as the electron donor, is a current limitation. 10 To address this issue, a coculture with cyanobacteria\ncapable of effectively fixing atmospheric carbon dioxide into organic\ncarbon, 11 offers an alternative. With regard\nto N 2 fixation, two types of cyanobacteria are distinguished:\ndiazotrophic (N 2 fixing) and nondiazotrophic cyanobacteria. 12 The nondiazotrophic cyanobacteria might be advantageous\nin establishing a trophic-dependent coculture with R. palustris . Nevertheless, it remains elusive whether\na coculture of such a consortium could be successfully established\nwhile maintaining sustainable production. The conventional Haber–Bosch\nprocess (HBP), responsible\nfor the annual production of 500 million tons of NH 3 , is\nhighly energy-intensive, highlighting an urgent need for alternative\ntechnologies to improve sustainability. 13 Notably, nitrogenase operates under mild conditions and is thus\nof great interest for addressing the challenge of cost-effective and\nsustainable ammonia synthesis. In addition to nitrogen fixation, nitrogenases\nare also considered as a route for hydrogen production; however, they\nare significantly inhibited by oxygen. 14 In diazotrophic cyanobacteria, nitrogen fixation occurs in heterocyst\ncells under the consumption of organic compounds provided by the vegetative\ncells. Heterocyst cells themselves maintain an anaerobic environment\nby deactivating water splitting. 15 Theoretically,\nwhen using a nondiazotrophic cyanobacterium, such as the model organism Synechocystis sp. PCC 6803 (hereafter Synechocystis ), 16 the role of the heterocyst would\nbe taken over by R. palustris , which\nlacks the photosystem II protein complex, making water splitting impossible.\nTo achieve a stable coculture of Synechocystis and R. palustris , balancing the CO 2 and N 2 fixation rates and the secretion of organic carbon and nitrogen\nwhile controlling the oxygen levels is critical. Unlike strictly anaerobic\nbacteria, R. palustris can switch among\nfour trophic regimes and survive through both oxic and anoxic conditions, 7 but a high oxygen tension inhibits N 2 fixation and can lead to nitrogen starvation of the consortium.\nUnder microaerobic conditions, however, energy can be provided through\noxidative phosphorylation instead of photophosphorylation for nitrogenase\nactivity. 17 While the consortium might\nachieve a stable balance between the organisms through different ratios\nof the cell types themselves, an efficient strategy to regulate and\nlimit the oxygen level in the consortium is crucial to achieving the\ndesired pathway activities. Light conditions play a significant\nrole in regulating the performance\nof phototrophs. Allowing for dark respiration of microbes is an effective\nway to metabolize oxygen, thus ensuring anoxic or oxygen-depleted\nconditions for H 2 production. 18 Complete microbial oxygen reduction would simply undo the water\nsplitting by producing water, and no hydrogen would be generated.\nHowever, lowering oxygen tension around the PNSB enough to enable\nthe activity of nitrogenases or hydrogenases would complement physical\noxygen removal from the reactor, for instance, through gas stripping.\nAs a typical cyanobacterium, Synechocystis contains\na series of pigments, e.g., chlorophylls and phycobiliproteins, enabling\nthe utilization of visible light for photosynthesis. 19 In addition to visible light, R. palustris , however, also harbors a light-harvesting complex (bacteriochlorophylls)\nthat can absorb light in the near-infrared spectrum (NIR, 800–900\nnm). 20 This difference should allow, on\nthe one hand, the possibility to create a coculture that uses a broader\nlight spectrum than cyanobacteria alone, increasing photon efficiency,\nand, on the other hand, should provide a strategy for selectively\nproviding light energy in the coculture to individual strains. This\ncould facilitate the balancing of strain abundance and regulate metabolic\nactivities within the coculture by changing light conditions. In this study, we constructed a coculture consisting of cyanobacteria\nand PNSB through light-dependent regulation. We cocultivated R. palustris with either the wild type of Synechocystis sp. PCC 6803, or an engineered strain, Synechocystis_acs (an acetate overproducing strain). Various\nlight regulation strategies, including constant illumination, circadian\nlight-dark illumination, and circadian light-infrared illumination,\nwere employed. These strategies facilitated trophic dependence through\ncarbon and nitrogen assimilation and allowed for the regulation of\ncoculture growth. The coculture enabled biohydrogen production in\na light-based system feeding on CO 2 and N 2 ,\nhighlighting the potential of controlling a phototrophic community.",
"discussion": "3 Results and Discussion To achieve a\ntrophic dependency and simultaneous growth of each\nstrain in the coculture, acetate, and yeast extract were replaced\nwith bicarbonate, and the major nitrogen source was changed to N 2 compared to the original M27 medium. The provision of bicarbonate\nmeant that the coculture relied on carbon fixation by Synechocystis . Nitrogen was initially provided in the form of urea at a low concentration.\nThis was selected as an initial booster to start the growth of the\ncommunity. The amount of urea provided would become limiting soon\nafter some initial growth, making the coculture reliant on N 2 fixation by R. palustris . 3.1 Growth of the Consortium Under circadian\nillumination of white and infrared light (WI), Synechocystis in the coculture (C-WI) achieved similar growth compared to the\nmonoculture of Synechocystis_acs (Sacs-WI). However,\nmuch slower growth was observed for Synechocystis_acs in the coculture (C-WI, Figure 1 a). Correspondingly, due to the lack of an organic\ncarbon source, the growth of R. palustris in both monoculture (R-WI) and coculture with Synechocystis (C wt -WI) was significantly inhibited ( Figure 1 b). Notably, in the presence\nof Synechocystis_acs , R. palustris (C-WI) achieved significantly higher growth (5.3 and 12.8 times)\ncompared to that of the R-LR and R + S-LR, respectively. This indicates\na potential provision of organic compounds from Synechocystis_acs to R. palustris in the coculture.\nNevertheless, light conditions affected the performance of the cocultures\nstrongly. An illumination with a light and dark cycle promoted the\nsimultaneous growth of both strains (with slow growth rates in C-WD);\nhowever, a constant illumination resulted in a dominant growth of Synechocystis_acs , thus further inhibiting the R. palustris (C-W). Notably, the coculture with Synechocystis_acs under a circadian rhythm of white and\ninfrared light promoted the growth of R. palustris the most. The coculture promoted a constant and longer exponential\ngrowth of R. palustris than the monoculture\nfed with standard organic carbon sources, e.g., acetate and malate, 24 indicating a promising potential to use the\nproposed coculture for generating sustainable growth of R. palustris . Figure 1 Growth (cell number) of (a) Synechocystis. sp\nand (b) R. palustris under different\ntreatments and (c) concentration of acetate in supernatant. R: R. palustris ; S: Synechocystis wild\ntype; Sacs: Synechocystis_acs ; -L: group of constant\nwhite light illumination; -LD: group of circadian illumination of\nlight and dark; -LR: group of circadian illumination of light and\ninfrared. The culture medium used in this study was acetate-depleted.\nAs\na critical compound, acetate could be overproduced and released by\nthe engineered strain ( Synechocystis_acs ) used in\nthe present study, and serve as an organic carbon source for growth\nand/or H 2 production of R. palustris . 8 The results show that the cultures\nwith rapid growth of Synechocystis_acs but limited\ngrowth of R. palustris (or without R. palustris ) accumulated much higher acetate concentrations\nin the supernatant than the other cultures (up to 292.3 and 250.6\nμmol L –1 acetate in C-W and Sacs-WI, respectively, Figure 1 c). In contrast,\nwhen R. palustris exhibited significant\ngrowth, less acetate was detected in the supernatant (up to 34.4 and\n13.8 μmol L –1 acetate in C-WD and C-WI, respectively),\nindicating a potential rapid acetate consumption that supported the\ngrowth of R. palustris . 3.2 Gas Composition and Enriched Products of Renewable\nEnergy Carrier As shown in Figure 2 , the coculture of R. palustris with Synechocystis_acs under light-dark illumination\n(C-WD) started producing hydrogen gas at the earliest and reached\nthe maximum hydrogen yield of 2.27 ± 0.21 μmol mg –1 at day 3 (the hydrogen production yield per milligram of fixed carbon).\nMeanwhile, the coculture under circadian white and infrared illumination\naccumulated hydrogen later (from day 3) but reached the highest hydrogen\nproduction (up to 9.70 ± 2.03 μmol mg –1 , C-WI) among all treatments ( Figure 1 a). Figure 2 Gas composition in headspace with different treatments.\n(a) Produced\nhydrogen gas yield per mg fixed carbon and (b) oxygen gas levels. R. palustris encodes\ndiverse nitrogenases\nresponsible for fixing nitrogen gas into ammonia with the obligatory\nbyproduct of hydrogen. 8 The various enzymatic\noptions of nitrogenases in R. palustris help to mitigate the inhibitions under adverse conditions, such\nas limited nutrients, 25 thus ensuring high\nefficiency of H 2 production in practice. Besides, R. palustris can utilize a wide range of organic\ncarbon sources for H 2 production, 10 which broadens the application scenarios for biohydrogen production.\nHowever, nitrogenases and hydrogenases are sensitive to oxygen. 14 Therefore, limiting the oxygen level in the\nsystem is critical for the regulation of hydrogen gas production.\nIn this study, circadian illumination effectively maintained oxygen\nat depleted levels (C-WI and C-WD, Figure 1 b). The respiration during the dark or infrared\nphase allowed the microbes to partially metabolize oxygen. 18 The reduced oxygen levels were low enough to\nenable hydrogen production. In contrast, even with low oxygen content\nin the monoculture of R. palustris (R-WI),\nno hydrogen gas was produced due to the lack of organic carbon sources\nthat serve as electron donors. Hydrogen production via nitrogenase\nis an energy-intensive process,\nwith nitrogenase competing for ferrodoxin and protons, thereby diverting\nthese resources from other metabolic pathways. This leads to an overall\ncompetition for energy, protons, and electrons, impacting cellular\ngrowth, carbon storage, and various metabolic functions. 24 , 26 However, circadian illumination provides R. palustris with an opportunity to balance resource allocation across different\nmetabolic pathways, potentially improving the overall production efficiency.\nIn R. palustris , light-driven reactions\nfacilitate NADH production and electron transfer. However, electrons\nare ultimately sourced from the oxidation of organic compounds without\nany water-splitting process taking place. The electrons can be directed\nto three primary electron sinks, including nitrogen fixation (with\nH2 production), anabolic pathways, or storage metabolism. 27 During white light exposure, R. palustris consumed organic carbon in the coculture,\nwhile circadian dark phases channeled electrons primarily toward biomass\nand likely to storage metabolism. In contrast, infrared light phases\nunder circadian conditions appear to preferentially allocate electron\nflow toward N 2 fixation with the nitrogenase producing\nH 2 as a byproduct. Consequently, the C-WI treatment achieved\nsubstantially higher H 2 yields compared with the C-WD treatment. Future studies aim at a comprehensive electron balance analysis\nthat could provide deeper insights into the mechanisms by which varying\nlight regimes influence electron allocation across metabolic pathways. In addition to H 2 , fatty acids also serve as critical\nrenewable energy carriers. 28 Results of\nfatty acid profiling suggested a lower fatty acid content (C8–C24)\nin Synechocystis_acs compared to R.\npalustris ( Figure 3 d,e). Thus, the monoculture of Synechocystis_acs generated only 20.6 mg L –1 of fatty acids out\nof the total biomass. Conversely, even though all three cocultures\nproduced significantly higher total fatty acid titers compared to\nthe monocultures, substantial differences in total fatty acid titers\nwere observed among the three cocultures ( Figure 3 f). The coculture of R. palustris with Synechocystis_acs under constant illumination\n(C-W) and circadian white-infrared illumination (C-WI) gained significantly\nhigher total biomass compared to the C-WD group, resulting in higher\ntotal fatty acids titers (68.7 and 59.0 mg L –1 respectively,\nC8–C24) in C-W and C-WI treatments. Although C-WI promoted\ngreater growth of R. palustris (154\nmg L –1 ) than C-W (61 mg L –1 ),\nthe changes in the fatty acid contents led to no significant difference\nin total fatty acid titer between the two treatments. Figure 3 C8–C24 fatty acid\nprofiles of (a) C-WI, (b) C-WD, (c) C-W,\n(d) R-WI, and (e) Sacs-WI. (f) Total fatty acid content (left) and\nyield (right) covering C8–C24 in each treatment. Regarding the fatty acid profile, the coculture\nwith constant white\nlight and circadian white-infrared conditions triggered the highest\naccumulation of C16–C18 fatty acids (mainly C16:1, C16:0, and\nC18:2) in the final biomass. The unsaturated fatty acid palmitoleic\nacid (POA, C16:1) has demonstrated the ability to prevent a series\nof diseases, e.g., stroke and diabetes. 29 In addition, the long-chain unsaturated fatty acids, including oleic\nacid (C18:1) and linoleic acid (C18:2), have been described as attractive\nantibacterial agents, and thus serve as key antimicrobial food additives. 30 Consequently, the cocultures (C-WI and C-W)\ndemonstrate their potential for enhancing the production of both biofuels\nand valuable chemicals through carbon capture. Such fatty acids-derived\nbiofuels (e.g., via transesterification and esterification) even have\na higher energy density and are more compatible with current infrastructure\nwhen compared to other forms of renewable energy. 31 3.3 Carbon and Nitrogen Balancing The\ninitial carbon source in the system was mainly bicarbonate (60.6–72.8%,\ndetected as carbonate), cysteine (13.2–15.8%), urea (5.7–6.8%),\nand the cellular inoculum. Under the chosen cultivation conditions,\nbicarbonate could not be directly metabolized by R.\npalustris , 7 therefore,\nneither significant consumption of bicarbonate, nor significant growth\nof R. palustris was observed in the\nmonoculture of R. palustris (R-WI, Figure 4 a,b). In contrast,\nthe photosynthesis of Synechocystis_acs in monoculture\n(Sacs-WI) rapidly fixed carbon from bicarbonate (carbon content reduced\nfrom 60.6 to 18.2%) and redirected the carbon flow toward biomass\n(73.4%) as well as excreted acetate (1.8% accumulated in the supernatant\nafter consumption by R. palustris )\nat day 4. In comparison, the coculture with constant light (C-W) generated\na similar flow from carbon fixation toward biomass of Synechocystis_acs (67.6%), but with higher excreted acetate (3.1% left after consumption).\nNotably, in both cocultures with circadian illumination, a smaller\nproportion of the fixed carbon was retained within the biomass of Synechocystis_acs (28.4% and 42.2% in C-WI and C-WD, respectively)\ncompared to the monoculture. In these conditions, a larger proportion\nof the fixed carbon was incorporated into the biomass of R. palustris (33.6 and 20.4% in C-WI and C-WD, respectively),\nindicating an organic carbon supply from Synechocystis_acs to R. palustris . Figure 4 Elemental distribution\nfor (a) carbon at day 0, (b) carbon at day\n4, (c) nitrogen at day 0, and (d) nitrogen at day 4. Nitrogen plays a critical role in the synthetic\nconsortium. Urea\nserved as the primary nitrogen source in the system, accounting for\n65.4% of the total nitrogen ( Figure 4 c,d). By day 4, nitrogen from urea was depleted in\nall coculture treatments, while residual nitrogen was still observed\nin both monocultures of R. palustris and Synechocystis_acs . Nitrogen in the form of\nurea can be rapidly transformed into ammonium through urease and further\nmetabolized by Synechocystis_acs and R. palustris . Subsequently, the biomass of Synechocystis_acs and/or R. palustris became the dominant sink of nitrogen, indicating nitrogen uptake\nand supporting further growth of biomass ( Figure 4 d). Notably, in the cocultures, illumination\nconditions significantly regulated the final sink of nitrogen, in\nwhich circadian illumination of white/infrared promoted the highest\naccumulation of nitrogen in the biomass of R. palustris (59.1%, R + Sacs-LR), while constant white light led to the lowest\n(28.5%, C-W). In all three cocultures, the ammonium detected in the\nsupernatant increased initially (from day 0 to day 2) and then decreased\nat day 4 ( Figure S2 ). Meanwhile, in the\nmonocultures, the ammonium levels increased and remained stable until\nday 4. The ammonium in the supernatant could either be hydrolyzed\nfrom urea via urease or be released by R. palustris through nitrogen fixation (nitrogenase). The nitrogen balances closed\nat 117.6, 108.6, and 93.3% on day 4 compared to day 0 in C-WI, C-WD,\nand C-W, respectively, indicating fixation of gaseous nitrogen from\nthe headspace into the cocultures under WI and WD illumination. The\ncoculture presents a promising alternative in sustainable ammonia\nsynthesis, offering a viable substitute for conventional HBP. This\nnitrogen fixation also led to obligatory hydrogen production, especially\nin the coculture of C-WI. Interestingly, thiosulfate was observed\nin the coculture of R. palustris with Synechocystis _acs\nunder both constant illumination (C-W) and circadian illumination\nof light-infrared (C-WI) at day 4 ( Figure S1b ). Thiosulfate can serve as an electron donor for the nitrogenase-catalyzed\nH 2 production by R. palustris . 32 Therefore, in the coculture, both\nacetate and thiosulfate could potentially promote H 2 production\nby donating electrons. 3.4 Response of Proteome to Coculture and Light\nRegulation To better understand the underlying metabolic\nprocesses in the cocultures under varying illumination conditions,\na proteomics analysis was conducted. In general, among all treatments, R. palustris in the coculture during infrared periods\n(infrared C-WI) gained the most enrichment of gene ontology (GO) terms\n( Figure 5 a), meaning\nthat significant changes of metabolic activities (both up- and downregulations)\ntook place when the cocultures were exposed to infrared illumination\nin the coculture compared to the single culture. In the coculture,\ncompared to the white light periods (white C-WI), during infrared\nillumination (infrared C-WI), the proteins encoding the biological\nprocesses of nitrogen fixation, monatomic cation transmembrane transport,\ncarbohydrate metabolic process, translation, and Fe–S cluster\nassembly were significantly enriched and upregulated. Meanwhile, the\nmolecular functions, e.g., nitrogenase activity and 2Fe–2S\ncluster binding also showed an upregulated trend. The carbohydrate\nmetabolic processes, e.g., acetate metabolism, ensure the electron\nbalance for hydrogen production. 8 Besides,\nthe Fe–S cluster is closely related to the formation of hydrogenase\nand nitrogenase. 33 With sufficient electron\ndonors (from acetate metabolism), the upregulated nitrogen fixation\ncould lead to obligatory H 2 production. 8 This indicated a generally stimulated activity of R. palustris in the coculture, particularly toward\nH 2 production, electron/substrate transportation, and biomass\naccumulation during infrared illumination. Unlike C-WI, nitrogenase\nactivity-related proteins in C-WD were exclusively detected during\nwhite illumination, indicating that a different H 2 production\nperiod occurred in C-WD (during white illumination) compared to C-WI\n(during infrared illumination). The presence of white light led to\ncompetitive interactions between R. palustris and Synechocystis_acs , significantly suppressing\nthe carbohydrate metabolism and amino acid biosynthesis of R. palustris in both C–WI and C-WD. Conversely,\nthe circadian cycles of illumination, which included dark or infrared\nperiods, provided R. palustris with\nopportunities for metabolism. Figure 5 Gene ontology (GO) enrichment analysis and pathway\nsignificance\nanalysis of different treatments for (a) R. palustris and (b) Synechocystis_acs. Yellow bulb: white light\nperiod; gray bulb: dark period; red bulb: infrared period; R: R. palustris ; Sacs: Synechocystis_acs ; -W: group of constant white light illumination; -WD: group of circadian\nillumination of light and dark; -WI: group of circadian illumination\nof light and infrared. The fold changes of R. palustris and Synechocystis_acs were compared to the light\nperiod of the single culture of R. palustris (in R-WI treatment) and the single culture of Synechocystis_acs (in Sacs- WI treatment), respectively. Counts refer to protein number\nthat has been detected. Gray bubbles refer to the GO terms with no\nsignificant difference ( p > 0.05). For Synechocystis_acs , the highest\nenrichment\nand upregulation of photosynthesis and photosystem II were achieved\nunder continuous illumination (C-W). This accounts for the highest\ngrowth ( Figure 1 a)\nand carbon fixation ( Figure 4 b) observed in Synechocystis_acs , which also\nled to elevated O 2 levels ( Figure 2 b). Interestingly, the amino acid biosynthesis\nand efflux transmembrane transporter activity 34 were more stimulated in the coculture during infrared illumination\n(C-WI), indicating a potential higher cellular growth of Synechocystis_acs and increased substrate release and supply (e.g., acetate) to R. palustris during infrared periods. In total,\n11 nitrogenase- and nitrogen-fixation-related proteins,\nas well as 5 hydrogenase-related proteins, were detected ( Figure 6 ). Proteomics demonstrated\nthat light conditions significantly regulated the protein expression\nrelated to nitrogen fixation and hydrogen production. R. palustris possesses three different types of nitrogenases\nresponsible for nitrogen fixation. 25 In\nthe group of C-WI, infrared illumination significantly stimulated\nthe expression of molybdenum nitrogenases encoded by the nif genes and the nitrogen regulation proteins, leading to an obligatory\nproduction of hydrogen gas. 35 Among the\nthree classes of nitrogenases, the molybdenum counterpart functions\nmost efficiently. 36 Notably, during the\nwhite light illumination, the expressions of all proteins encoding\nthese functions were downregulated or exhibited no significant difference\ncompared to those in the control (R-WI), resulting in only a significant\nH 2 production in C-WI during infrared phases. Nevertheless,\nwith the upregulation of nitrogenase, a significant upregulation of\nthe nickel-dependent hydrogenase ([NiFe] HydA) was also observed.\nAs an enzyme that catalyzes the reversible oxidation of molecular\nH 2 , the produced H 2 could be uptaken and consumed\nby the hydrogenase to supply reductants for nitrogen fixation or donate\nelectrons for phototrophic growth (with sodium bicarbonate) when lacking\nfavorable organic carbon sources like acetate, 35 leading to a decrease of accumulated H 2 gas\n( Figure 2 a). On day\n4, depleting levels of acetate were observed in C-WI treatment ( Figure 1 c). The [NiFe] HydA\nshowed significant upregulation during infrared illumination and downregulation\nduring white illumination in C-WI, suggesting that hydrogen uptake\nprimarily occurred during infrared illumination. Given the extensive\nupregulation of hydrogenases, including HydA, HypB, and HydB (up to\nover 50,000-fold increase), the balance between hydrogen production\nand consumption appears to be heavily shifted toward consumption,\nthereby reducing net hydrogen accumulation ( Figure 2 a). Figure 6 Heatmaps of relative abundance changes of (a)\nnitrogenase- and\nhydrogenase-related protein levels in R. palustris , (b) electron transport-, urea metabolism-, and ammonium metabolism-related\nprotein levels in R. palustris , (c)\nurea and ammonium metabolism-related protein levels in Synechocystis_acs , and metabolic pathways of (d) nitrogen fixation, urea, ammonium,\nand fatty acid metabolism in R. palustris , and (e) urea, ammonium, and fatty acid metabolism in Synechocystis_acs . Up- and downregulated proteins (calculated by log 2 fold change\ncompared to the control group. All results of R. palustris were compared to the white light period of a monoculture of R. palustris under light-infrared circadian illumination\nconditions. All results of Synechocystis_acs were\ncompared to the white light period of monoculture of Synechocystis_acs under light-infrared circadian illumination conditions, indicated\nin red and blue, respectively. White: white light period; dark: dark\nperiod; infrared: infrared period; R: R. palustris ; S: Synechocystis_acs ; -W: constant white light\nillumination; -WD: circadian illumination of light and dark; -WI:\ncircadian illumination of light and infrared. In contrast, in the group of C-WD, the nitrogenase\nmolybdenum–iron\nprotein (NifD), nitrogen-fixing protein (NifU), nitrogen regulatory\nprotein P(II), and hydrogenase (NiFe) exhibited significant upregulation\nduring white illumination, indicating that H 2 production\nand uptake predominantly occurred during white illumination. This\nresult highlights the superior light-harvesting efficiency of R. palustris under C-WD compared to C-WI. The dark\nphase appears to facilitate the synthesis of nitrogenase-related proteins,\nas evidenced by the significant upregulation of nitrogenase biosynthesis\nprotein NifN and NifW. 37 , 38 Compared to C-WI, the overall\nweaker induction of nitrogen-fixation proteins in C-WD may have resulted\nin a lower production of H 2 than in the C-WI group ( Figure 2 a). Nevertheless,\nwith constant illumination (C-W), most nitrogenase- and hydrogenase-related\nproteins were downregulated ( p < 0.05, Figure 6 ), likely due to\nhigh oxygen levels in the system ( Figure 2 b). Consequently, no H 2 production\nwas detected. In R. palustris , the electrons obtained\nfrom organic acids are first transported to ferredoxin or flavodoxin\n( Figure 6 d), 14 and then, the low-potential electrons from ferredoxin\nor flavodoxin will be further supplied to the nitrogenase for nitrogen\nfixation and hydrogen production. 39 Ferrodoxin\nserves as a critical redox flux regulator among major metabolic pathways. 40 In the treatments of C-WI and C-WD, the protein\nlevels of both ferredoxin and flavodoxin were observed to be significantly\nupregulated ( Figure 6 b). As expected, the conditions of infrared and white-light illumination\nresulted in the highest response of electron transport proteins in\nthe corresponding groups of C-WI and C-WD, respectively. This supported\na highly promoted electron transfer toward the nitrogen fixation process\nduring illumination, especially under the infrared condition ( Figure 6 b). Urea and\nnitrogen gas served as major nitrogen sources for R.\npalustris in this study. On day 4, urea was detected\nin the supernatant of both monocultures (R-WI and Sacs-WI). However,\nit was depleted in all three cocultures ( Figure 4 ), leading to an overall lower protein expression\nof urea transport in both strains ( R. palustris and Synechocystis_acs, Figure 6 b,c) and a reduced urea metabolic cycle in R. palustris than the control group ( Figure 6 b,d). Interestingly, while\nurea was depleted, proteins related to glutamate and glutamine synthase\nwere significantly downregulated in the C-WI group during the light\nperiod. In contrast, infrared illumination significantly upregulated\nthese proteins. This implies that R. palustris may exhibit either lower competitiveness for light harvesting under\nthe visible light spectrum compared to Synechocystis_acs , or that nitrogenase activity was inhibited by the elevated oxygen\nlevels during white light illumination, potentially allowing for an\nenhanced nitrogen fixation during the infrared period. For Synechocystis_acs , urea was the only abiotic\nnitrogen source. Facing urea depletion in the medium, all ureases,\nglutamine synthetase, and glutamate synthase were expected to be downregulated\n( Figure 6 e). Nevertheless,\nin response to infrared illumination (dark C-WI, Figure 6 c), a dramatically upregulated\nammonium transport protein (AMT), glutamine synthetase, and glutamate\nsynthase were observed, indicating a potential ammonium supplier in\nthe coculture. Taking R. palustris into\nconsideration, proteomics results confirmed activated nitrogen fixation\nunder an infrared illumination. This led to obligatory ammonium production\nin the cells of R. palustris . Additionally,\nan upregulation of the ammonium transporter (AMT) was detected in R. palustris during the same period ( Figure 6 b), implying a potential exchange\nof ammonium from R. palustris to Synechocystis_acs in the coculture. Previous studies revealed\nthat the activity of nitrogenase could be inhibited by ammonium accumulation. 10 Thus, the coculture established a symbiotic\nrelationship in which Synechocystis_acs fixes carbon\ndioxide and supplies acetate to R. palustris , while R. palustris fixes nitrogen\ngas and provides ammonium in return. Through timely consumption of\nammonium by Synechocystis_acs , the nitrogenase activity\nof R. palustris can be further promoted. The enzyme acetyl-CoA carboxylase (ACC) is the first step in de\nnovo fatty acid synthesis to catalyze carboxylation of acetyl-CoA\nto produce malonyl-CoA. 41 In the coculture\nunder circadian white and infrared illumination (C-WI), there was\nsubstantial variation of ACC expression, with a significant downregulation\nobserved during the white light phase (White C-WI) and a significant\nupregulation during the infrared phase (infrared C-WI) in R. palustris . Meanwhile, in Synechocystis_acs , coculture with constant white illumination showed the least downregulation\ncompared to the control among all cocultures. In contrast, cocultures\nunder other illumination conditions (C–W and C-WD) demonstrated\nno significant difference in R. palustris compared with the control group. As a subunit of ACC, biotin carboxylase\n(BC) catalyzes the ATP-dependent carboxylation of biotin during fatty\nacid synthesis. 41 The only upregulation\nof BC was observed in C-WI during infrared illumination from R. palustris . In the subsequent step, the produced\nmalonyl-CoA was utilized by the enzyme malonyl-CoA-acyl carrier protein\ntransacylase (MCAT) for fatty acid biosynthesis. 41 In Synechocystis_acs , only C-W demonstrated\nno significant difference compared with the control, while all other\ntreatments exhibited significant downregulation. In R. palustris , both C-W and infrared C-WI demonstrated\nsignificant upregulation. Additionally, the enzyme pyruvate phosphate\ndikinase (PPDK) 42 and glycerol-3 phosphate\ndehydrogenase, 43 which are positively correlated\nwith fatty acid synthesis, also showed upregulation or minimal downregulation\ncompared to the control group in C-W and infrared C-WI treatments.\nThese results indicate that R. palustris contributed the most to fatty acid accumulation during infrared\nillumination in the C-WI group, while both Synechocystis_acs and R. palustris contributed to the\nincreased fatty acid content in the C-W group. The significant difference\nin protein expression related to fatty acid biosynthesis during the\nillumination switch may help in maintaining the redox balance during\nphotoheterotrophic growth. 43 3.5 Environmental Implications Phototrophic\nmicroorganisms hold great promise for capturing solar energy and converting\ngreenhouse gases into sustainable energy carriers, providing an alternative\nsolution to the increasingly fierce energy challenge. Communities\nof such phototrophs could even enable more diverse metabolic activities\nthan single cultures; however, successful construction, control, and\nin-depth characterizations have rarely been reported to date. In this\nstudy, we combined strain engineering and light quality regulation\nto enable the coculture of cyanobacteria with purple nonsulfur bacteria\ntoward clean production of energy carriers, e.g., H 2 and\nfatty acids. The system was successfully constructed with both CO 2 and N 2 fixation. The industrial Haber-Bosch process\nfor nitrogen fixation is one of the most energy-consuming and CO 2 -emitting processes of mankind. 44 Therefore, the biotechnology proposed in this study that relies\non atmospheric N 2 and CO 2 gas fixation will\nenable a more sustainable process and may be an important stepping\nstone toward a net-zero emissions economy. Nevertheless, obstacles\nremain to be overcome before real applications can be developed. Inactivating\nthe uptake hydrogenases in both strains could help prevent the consumption\nof produced H 2 , 45 as observed\nin this study. Additionally, continuous reactors allowing for a constant\nremoval of H 2 and O 2 using for instance selective\nmembranes might be a technical solution to further improve the performance\nand finally long-term stability of N 2 and CO 2 fixation remains to be evaluated."
} | 9,203 |
36178056 | PMC9618323 | pmc | 654 | {
"abstract": "Abstract In recent years, biotechnological conversion of the alternative carbon source acetate has attracted much attention. So far, acetate has been mainly used for microbial production of bioproducts with bulk applications. In this study, we aimed to investigate the potential of acetate as carbon source for heterologous protein production using the acetate‐utilizing platform organism Corynebacterium glutamicum . For this purpose, expression of model protein eYFP with the promoter systems T7 lac and tac was characterized during growth of C. glutamicum on acetate as sole carbon source. The results indicated a 3.3‐fold higher fluorescence level for acetate‐based eYFP production with T7 expression strain MB001(DE3) pMKEx2‐ eyfp compared to MB001 pEKEx2‐ eyfp . Interestingly, comparative eyfp expression studies on acetate or glucose revealed an up to 83% higher biomass‐specific production for T7 RNAP‐dependent eYFP production using acetate as carbon source. Furthermore, high‐level protein accumulation on acetate was demonstrated for the first time in a high cell density cultivation process with pH‐coupled online feeding control, resulting in a final protein titer of 2.7 g/L and product yield of 4 g per 100 g cell dry weight. This study presents a first proof of concept for efficient microbial upgrading of potentially low‐cost acetate into high‐value bioproducts, such as recombinant proteins.",
"conclusion": "CONCLUDING REMARKS Until now, most research efforts have rather focused on microbial conversion of acetate into bio‐based products with bulk applications, such as platform chemicals (succinic acid, itaconic acid) or bioplastics (polyhydroxyalkanoates). In this work, the potential of acetate for biotechnological production of recombinant proteins as high‐value products was demonstrated. Here, the Gram‐positive platform bacterium C. glutamicum has been proven to be a powerful platform for acetate‐based protein production at g/L scale. Compared with previous research works, the finding of our study is quite surprising and unique due to the fact that the presence of acetate is generally reported to strongly inhibit protein production performance in the Gram‐negative cell factory E. coli . To conclude, this study is the first in which a recombinant protein has been produced in a fed‐batch cultivation using (lignocellulosic) acetate as carbon source. Beyond that, the high efficiency of the T7 RNAP‐dependent protein production system in C. glutamicum has been proven for the first time under high cell density conditions in a bioreactor. The novelty of this study presents a proof of concept for efficient microbial conversion of potentially low‐cost acetate into recombinant proteins by acetate‐tolerant platform organisms, such as C. glutamicum . The presented acetate‐based bioprocess could thus be also used to produce (extracellular) target proteins with industrial relevance (e.g. food proteins, industrial enzymes and biopharmaceutical proteins) in order to meet the needs of a future bio‐based industry.",
"introduction": "INTRODUCTION Beyond its traditional use for fermentative amino acid production, the industrial platform organism Corynebacterium glutamicum has gained considerable importance as alternative microbial cell factory for heterologous protein production (Freudl, 2017 ; Lee & Kim, 2018 ; Liu et al., 2015 ). Especially with respect to secretory protein production, it offers beneficial characteristics such as (i) low amounts of endogenous extracellular proteins, (ii) low proteolytical activity in the culture supernatant and (iii) the ability to directly transport proteins across the cytoplasmic membrane with high efficiency. Accordingly, large libraries of signal peptides (Hemmerich et al., 2016 ) and genome‐reduced strains (Hemmerich et al., 2020 ), as well as biosensor‐based monitoring tools (Bakkes et al., 2021 ), have already been established for screening and optimization of protein secretion performance in C. glutamicum . In addition, the Gram‐positive bacterium represents an endotoxin‐free cell factory, thus eliminating the need for endotoxin removal and simplifying downstream processing in case of biopharmaceutical proteins. So far, a first commercial expression system known as CORYNEX® has already been developed by the Company Ajinomoto Co., Inc. As a well‐established and GRAS‐classified workhorse of white biotechnology, C. glutamicum also exhibits easy genetic accessibility (Linder et al., 2021 ), fast growth up to cell densities higher than 80 g/L cell dry weight (Kiefer, Merkel, Lilge, Hausmann et al., 2021 ; Knoll et al., 2007 ), and robustness towards substrate gradients typically found in industrial‐scale bioreactors (Buchholz et al., 2014 ; Limberg et al., 2017 ). Beyond that, the relatively high tolerance to various lignocellulose‐derived inhibitors has enabled this bacterium to produce several value‐added compounds from alternative biorefinery feedstocks (Becker et al., 2018 ; Lange et al., 2017 , 2018 ; Mao et al., 2018 ; Sasaki et al., 2019 ). With regard to a future bio‐based industry, the C2 compound acetate represents a potential next‐generation platform substrate generated by lignocellulose depolymerization and multiple other alternative routes (Kiefer, Merkel, Lilge, Henkel et al., 2021 ). In fact, the microbial production of value‐added products from pure acetate or acetate‐containing biorefinery streams has truly gained increasing interest over the past decade. Starting from compounds directly derived from microbial acetate metabolism (e.g. succinate or microbial lipids), strain and tolerance engineering of acetate‐utilizing microorganisms has largely expanded the portfolio of acetate‐based bioproducts (Kim et al., 2021 ; Kutscha & Pflügl, 2020 ). However, the potential of bioprocess strategies for efficient biotechnological conversion of acetate as potentially inhibitory microbial substrate is still rather underexploited. In a previous study, we demonstrated a novel bioprocess for efficient acetate conversion using C. glutamicum as biocatalyst (Kiefer, Merkel, Lilge, Hausmann et al., 2021 ). By applying an automated pH‐coupled online feeding control, C. glutamicum wild‐type ATCC 13032 was cultivated up to high cell densities using lignocellulosic acetic acid derived from beech wood depolymerization. In addition to the presented biomass accumulation on acetate, this bioprocess was recently adapted for biomass‐decoupled production of the promising C5 platform chemical itaconic acid using engineered C. glutamicum strain ICD R453C (Merkel et al., 2022 ). In the present work, we aimed to evaluate the potential of acetate as carbon source for production of recombinant proteins. Since the commonly used prokaryotic protein production hosts Escherichia coli and Bacillus subtilis naturally exhibit rather poor or even no growth in presence of acetate (Arnold et al., 2019 ), this study focused on C. glutamicum as acetate‐utilizing, alternative protein cell factory. For exemplary intracellular protein expression, the fluorescence model protein eYFP derived from Aequorea victoria GFP (Ormö et al., 1996 ) was chosen as a target protein. In the first part of this work, eYFP expression during growth of C. glutamicum on acetate as carbon source was characterized with the expression vectors pMKEx2 (Kortmann et al., 2015 ) and pEKEx2 (Eikmanns et al., 1991 ) containing the IPTG‐inducible promoter systems T7 lac and tac , respectively. In addition, the T7 RNA polymerase (T7 RNAP)‐dependent production of eYFP on acetate was quantitatively compared with that on glucose as commonly used carbon source for fermentation. Lastly, the heterologous protein accumulation with C. glutamicum T7 expression system was studied under high cell density conditions in a fed‐batch bioreactor process using pure lignocellulosic acetic acid as combined carbon source and pH titrant. The results of this study demonstrate the capability of platform organism C. glutamicum for efficient acetate‐based protein production and further supports the outstanding potential of acetate as next‐generation carbon source in a future biotechnology.",
"discussion": "RESULTS AND DISCUSSION Characterization of IPTG ‐dependent eYFP production on acetate using T7 lac and tac promoter system To investigate the feasibility of acetate‐based protein production, expression of heterologous model protein eYFP in C. glutamicum was first studied using two different promoter systems. For this purpose, the expression vector pMKEx2 (Kortmann et al., 2015 ) based on the strong IPTG‐inducible T7 lac promoter was chosen. Combined with the prophage‐free C. glutamicum strain MB001(DE3) carrying a chromosomally integrated T7 RNA polymerase gene under control of the lacUV5 promoter, this expression system has been already shown to be powerful for glucose‐based protein expression in C. glutamicum (Kortmann et al., 2015 ). As a comparison, the well‐established expression vector pEKEx2 (Eikmanns et al., 1991 ) containing the IPTG‐inducible tac promoter with weaker promoter strength was tested, which is recognized by the corynebacterial RNA polymerase itself. To analyse the respective eYFP production on acetate, C. glutamicum strains MB001(DE3) pMKEx2‐ eyfp , MB001 pEKEx2‐ eyfp and wild‐type strain ATCC 13022 pEKEx2‐ eyfp (see Table 1 ) were grown in shake flasks with modified CGXII a minimal medium containing 10 g/L acetate as sole carbon source. The IPTG‐dependent expression of the target protein was induced at an OD 600nm of 2 (=0.47 g CDW /L) using varying inductor concentrations of 0, 10, 15, 20, 25, 50, 100, 250, 500 and 1000 μM IPTG. For a quantitative comparison of the protein production performances, the eYFP‐specific fluorescence of the bacterial cultures was measured in the late‐exponential growth phase (=6 h of post‐induction growth). Fluorescence signals were correlated to the respective biomass concentrations to consider the slightly differing final cell densities of the different strains. A diagrammatic comparison showing the biomass‐specific eYFP fluorescence of the different C. glutamicum strains is provided in Figure 1 . FIGURE 1 Effect of IPTG concentration on heterologous eYFP production in C. glutamicum with different promoter systems using acetate as sole carbon source . C. glutamicum strains MB001(DE3) pMKEx2‐ eyfp (black circles), MB001 pEKEx2‐ eyfp (grey circles) and ATCC 13032 pEKEx2‐ eyfp (white circles) were grown at 30°C and 130 rpm in baffled shake flasks with 50 ml modified CGXII a minimal medium containing 10 g/L acetate. Heterologous expression of target gene eyfp was induced at an optical density of 2 (=0.47 g CDW /L) using varying IPTG concentrations ranging from 0 to 1000 μM. After 6 h of post‐induction growth, the biomass‐specific eYFP fluorescence levels of the bacterial cell suspensions were measured in the late‐exponential growth phase. Production performance is shown relative to that of MB001(DE3) pMKEx2‐ eyfp , which was set to 100%. Strains MB001(DE3) (black squares), MB001 (grey squares) and ATCC 13032 (white squares) harbouring the respective empty vectors without eyfp gene were grown as negative control strains. Values and error bars represent means ± standard deviations from biological duplicate cultivations. For both promoter systems tested, a characteristic IPTG dose‐dependent saturation curve as previously shown for eYFP production during growth of C. glutamicum on glucose (Kortmann et al., 2015 ) was also observed for acetate as carbon source. When cultivated in the presence of IPTG, maximum biomass‐specific eYFP fluorescence signals could be detected at inductor concentrations above 250 μM for all three tested strains. However, higher IPTG concentrations did not result in significantly enhanced fluorescence values, obviously indicating IPTG‐saturated production levels. Compared with strain MB001 pEKEx2‐ eyfp (30.0 ± 2.2% AU/g CDW ·L at 500 μM IPTG), a 3.3‐fold higher maximum biomass‐specific eYFP fluorescence was found for T7 expression strain MB001(DE3) pMKEx2‐ eyfp (100 ± 11% AU/g CDW ·L at 500 μM IPTG). This is in accordance with the results from Kortmann et al. ( 2015 ), where a 3.5‐fold higher specific fluorescence in glucose‐containing CGXII minimal medium has been described for eYFP production under control of the T7 lac promoter. For wild‐type strain ATCC 13032 pEKEx2‐ eyfp , a 3.9‐fold lower maximum biomass‐specific fluorescence of 25.8 ± 3.2% AU/g CDW ·L (1000 μM IPTG) was observed. When comparing the eYFP production levels between the pEKEx2‐ eyfp harbouring strains, prophage‐free strain MB001 showed a clearly higher fluorescence level (up to 22.4%). This is in agreement with a previous study, where the same was reported for eYFP production on glucose due to presumably higher plasmid copy number in the genome‐reduced strain (Baumgart et al., 2013 ). Accordingly, this also holds true for eYFP production during growth of MB001 on acetate as carbon source. For the negative control strains harbouring the empty vectors pMKEx2 and pEKEx2 without eyfp gene, no background fluorescence could be detected. Interestingly, bacterial growth performances of the different strains were quite unaffected by the respective inductor concentration added and minor growth differences could be rather explained due to biological variations (Figure S1 ). It can thus be concluded that the phenomenon known as ‘metabolic burden’ (Bentley et al., 1990 ) typically described for heterologous protein production is obviously not observed for overexpression of eYFP as target protein. In fact, the same has been also reported for T7 RNAP‐dependent production of fluorescence model protein GFP with E. coli in a previous study, where almost no negative effect on cell growth could be observed (Li & Rinas, 2020 ). To summarize, heterologous production of model protein eYFP in C. glutamicum based on acetate as carbon source was demonstrated with two different, commonly used IPTG‐inducible promoter systems. In this context, no inhibitory effect of acetate on production performances of the C. glutamicum strains could be obviously found, although acetate was still present in concentrations above 8 g/L at the time point of induction. Moreover, the T7 RNAP‐driven expression system was also identified to be powerful for acetate‐based protein production. For this reason, further experiments in this study were only focused on C. glutamicum strain MB001(DE3) pMKEx2‐ eyfp . Comparison of T7 RNAP ‐dependent eYFP production using acetate and glucose as carbon sources Since carbohydrate‐based substrates such as glucose are still the commonly used biotechnological carbon sources, the next aim was to quantitatively compare the T7 RNAP‐dependent eYFP protein production on acetate to that on glucose. For this purpose, C. glutamicum strain MB001(DE3) pMKEx2‐ eyfp was cultivated in shake flasks with modified CGXII a minimal medium containing either 10 g/L acetate or 10 g/L glucose as sole carbon source. Expression of eyfp gene was induced at an OD 600nm of 2 (=0.47 g CDW /L) using an IPTG concentration of 1 mM to guarantee maximum induction. To quantitatively compare the heterologous protein production on the different substrates, the eYFP fluorescence of the bacterial cultures was analysed until reaching the stationary growth phase (=10 h of post‐induction growth). The corresponding graphs showing biomass growth, substrate consumption and eYFP production for the different carbon sources are illustrated in Figure 2 . FIGURE 2 Comparison of T7 RNAP‐dependent eYFP production in C. glutamicum MB001(DE3) using acetate or glucose as carbon source. C. glutamicum strain MB001(DE3) pMKEx2‐ eyfp was grown at 30°C and 130 rpm in baffled shake flasks with 200 ml modified CGXII a minimal medium containing either 10 g/L acetate (white symbols) or 10 g/L glucose (grey symbols) as sole carbon source. Heterologous expression of target gene eyfp was induced at an optical density of 2 (=0.47 g CDW /L) using an IPTG concentration of 1 mM. Values and error bars represent means ± standard deviations from biological duplicate cultivations. (A) Time‐courses of biomass growth (squares) and substrate consumption (triangles) during cultivation on the respective carbon sources. (B) Time‐courses of biomass‐specific fluorescence (circles) and biomass‐specific product yield Y \n PIX (bars) during the post‐induction growth on the respective carbon sources. As shown in Figure 2A , biomass concentrations of up to 6.0 ± 0.1 g CDW /L were achieved for growth of MB001(DE3) pMKEx2‐ eyfp on 10 g/L glucose as carbon source ( Y \n XIS = 0.54 ± 0.01 g CDW /g glucose ). In contrast to this, maximum cell densities reached for 10 g/L acetate were shown to be 1.6‐fold lower (3.7 ± 0.1 g CDW /L) due to the lowered biomass yield of 0.36 ± 0.00 g CDW /g acetate . However, maximum growth rates on acetate (μ max \n acetate = 0.44 ± 0.01 h −1 ) were found to be competitive to that on glucose as carbon source (μ max \n glucose = 0.43 ± 0.01 h −1 ). This is in accordance with our previous study, where μ max values of 0.45 h −1 were found for batch culture studies of wild‐type strain ATCC 13032 on 10 g/L acetate (Kiefer, Merkel, Lilge, Hausmann et al., 2021 ). With respect to heterologous protein production, comparable maximum eYFP fluorescence values could be measured for growth on glucose (207 ± 33 AU·10 4 ) and acetate (225 ± 13 AU∙10 4 ), respectively (see Table S1 ). However, when considering the much lower cell densities achieved on acetate, a drastically higher biomass‐specific fluorescence level was surprisingly found for acetate‐based eYFP production (Figure 2B ). While the specific fluorescence for growth on 10 g/L glucose reached a maximum of 48.4 ± 3.1 AU ∙ 10 4 /g CDW ·L (t = 8 h induction), a 1.6‐fold higher value of 75.5 ± 1.3 AU ∙ 10 4 /g CDW ·L ( t = 10 h induction) could be achieved for 10 g/L acetate as substrate. In order to compare the eYFP production not only on fluorescence activity but also on a level of protein amount, the intracellular target protein contents were additionally determined. As can be seen in the pattern of the respective biomass‐specific product yields ( Y \n PIX ), the same was also true for eYFP production on protein level. In fact, a 1.8‐fold higher Y \n PIX (26.8 ± 1.1 mg eYFP /g CDW ) was reached for growth and eYFP production of MB001(DE3) pMKEx2‐ eyfp on acetate compared with that on glucose (14.6 ± 0.4 mg eYFP /g CDW ). Interestingly, the results thus indicate that T7 RNAP‐dependent protein production in C . glutamicum was more efficient for acetate as carbon source. This finding is rather unexpected, and the reason thereof remains unclear, as common fermentations of C. glutamicum rely on carbohydrate‐based carbon sources with glucose being the most commonly used carbon and energy source of this bacterium. One may speculate that microbial growth on acetate via the glyoxylate cycle as anaplerotic pathway is associated with a higher precursor supply for eYFP biosynthesis or even a higher plasmid copy number of pMKEx2‐ eyfp . Apart from this, the presence of glucose for a long time period after IPTG addition might have affected the induction of the chromosomally encoded T7 RNAP gene itself by high levels of cAMP. Even though the targeted point mutations in the chromosomal lacUV5 promoter are known to reduce its sensitivity towards catabolite repression, it has been reported that it is not fully eliminated, and the highest target protein expression can be observed in absence of glucose (Novy & Morris, 2001 ). Therefore, a stronger expression and activity of T7 RNAP resulting in higher target protein expression at low cAMP levels might be another possible explanation for improved eYFP production in acetate‐containing CGXII minimal medium. However, further investigations such as plasmid copy number determination or intracellular cAMP level measurement might be necessary to clarify the reasons for improved eYFP production during growth on acetate as carbon source. Despite the fact that final cell densities of MB001(DE3) pMKEx2‐ eyfp for growth on glucose were 62% higher, the higher biomass‐specific production performance for growth on acetate was found to yield a quite similar eYFP titer of approximately 86 mg per L of culture (Table S1 ). With regard to a high cell density cultivation process with integrated biomass‐coupled product formation, acetate might thus potentially present a beneficial substrate for target protein accumulation due to the 83% higher biomass‐specific product yield Y \n PIX . In theory, acetate‐based production of model protein eYFP might thus enable higher product titers at the same biomass concentration compared with glucose as carbon source. High cell density cultivation of C. glutamicum \n T7 expression system on acetate for high‐level eYFP accumulation In general, high cell density fed‐batch cultures are commonly the method of choice for accumulation of heterologous target proteins (Choi et al., 2006 ). Accordingly, the next goal of this study was to establish acetate‐based protein production with C. glutamicum T7 expression system under high cell density conditions in a 42 L stirred‐tank bioreactor. For this purpose, our recently described fed‐batch process strategy based on a pH‐coupled online feeding control for acetate supply (Kiefer, Merkel, Lilge, Hausmann et al., 2021 ) was used for growth and eYFP production of C. glutamicum MB001(DE3) pMKEx2‐ eyfp . After an initial batch phase with a starting concentration of 10 g/L acetate, a fed‐batch phase with pH‐coupled supply of pure acetic acid (HAc) was followed. For compensating the pH‐decreasing effect by consumption of ammonia as nitrogen source, the latter was fed in form of urea present in CGXII feed solution (C/N feeding ratio = 10 C‐mole/N‐mole). The production phase for heterologous eYFP accumulation was initiated by addition of 1 mM IPTG upon reaching a cell density of OD 600nm 50 (=11.6 g CDW /L). Graphs showing the results of the bioreactor fed‐batch cultivation are presented in Figure 3 . Obtained cultivation data and calculated parameters are further summarized in Table 2 . FIGURE 3 Fed‐batch cultivation of C. glutamicum MB001(DE3) for T7 RNAP‐dependent eYFP production under high cell density conditions using acetate as carbon source. C. glutamicum strain MB001(DE3) pMKEx2‐ eyfp was grown at 30°C in a stirred‐tank bioreactor with an initial volume of 10 L modified CGXII b high cell density minimal medium. Feeding of pure acetic acid (HAc) was coupled to the pH control of the bioreactor system. Heterologous expression of target gene eyfp was induced at an optical density of 50 (=11.6 g CDW /L) using an IPTG concentration of 1 mM. Values and error bars represent means ± standard deviations from biological duplicate cultivations. (A) Time‐courses of biomass growth (dark grey squares), fed amounts of acetic acid and CGXII feed solution (red line), concentrations of acetate (white triangles) and ammonium (black diamonds), eYFP titer (grey circles), volumetric productivity (grey dashed line), relative target protein content (black squares) and fluorescence level (yellow bars) during the cultivation process. (B) SDS‐PAGE analysis showing the soluble protein fractions at different post‐induction cultivation time points. Equivalent amounts of total protein (6 μg) were loaded onto the gel and protein bands were visualized by Coomassie brilliant blue staining. St, purified His(6)‐eYFP protein standard; M, protein marker with molecular weights indicated in kDa. TABLE 2 Cultivation data and calculated parameters for fed‐batch high cell density cultivation of C. glutamicum MB001(DE3) pMKEx2‐ eyfp using acetate as carbon source. Values and error bars represent means ± standard deviations from biological duplicate cultivations. Cultivation parameter Values Process time [h] 27 c biomass \n max [g CDW /L] 67.8 ± 1.9 Reactor volume [L] 15.9 ± 1.3 m biomass \n max produced [g CDW ] 1074 ± 115 Acetate added/consumed [g] 4188 ± 113/ 3966 ± 131 Ammonia added/consumed [g] 240 ± 4/ 165 ± 7 \n Y \n XIS [g CDW /g acetate ] 0.27 ± 0.02 μ max [h −1 ] 0.42 ± 0.00 μ overall feeding phase [h −1 ] 0.08 ± 0.00 Fluorescence [AU·10 4 ] 9537 ± 98 Biomass‐specific fluorescence [AU·10 4 /g CDW ·L] 208 ± 1 Relative target protein content [% of soluble protein] 17.7 ± 2.5 eYFP titer [g/L] 2.7 ± 0.3 \n Y \n PIX [mg YFP /g CDW ] 40.0 ± 1.0 P V overall [g/L·h] 0.10 ± 0.02 Within the initial batch phase of the bioreactor cultivation, a cell density of 4.3 ± 0.5 g/L cell dry weight was reached after about 6 h (Figure 3A ). In the end of this process phase, residual acetate and ammonium concentrations of 2.3 ± 0.4 and 2.0 ± 0.2 g/L were detected in the culture supernatant. In the subsequent pH‐coupled fed‐batch phase, a total amount of 4188 ± 113 g acetate was fed by the auto‐regulated pH control resulting in a final biomass concentration of up to 67.8 ± 1.9 g CDW /L after 27 h. With respect to the growth performance of C. glutamicum MB001(DE3) pMKEx2‐ eyfp under bioreactor conditions, μ max values of 0.42 ± 0.00 h −1 were determined. However, the specific growth rate was found to decline after a process time of about 10 h, although acetate as carbon source was sufficiently present in the culture medium (1.3 ± 0.1 g/L). Consequently, the μ overall for the feeding phase clearly decreased to a value of 0.08 ± 0.00 h −1 . It should be noted that this finding is consistent with our previous study, where the same trend was observed for fed‐batch cultivations of C. glutamicum ATCC 13032 without product formation (Kiefer, Merkel, Lilge, Hausmann et al., 2021 ). More strikingly, this is also true for the slight accumulations of acetate und ammonium over time, reaching final concentrations of 14.9 ± 1.9 and 6.1 ± 0.3 g/L in the end. The reason for this can be explained by the shifting bacterial C/N consumption ratio, as already described for the previous wild‐type cultivations. When reaching a process time of about 13 h, the C/N consumption ratio started to clearly exceed the constantly applied C/N feeding ratio of 10 C‐mole/N‐mole (Figure S2 ). As cultivation continued, nitrogen oversupply from CGXII feed solution consequently caused an overfeeding of acetate due to the pH‐increasing effect of excess ammonia acting as a base. With regard to acetate‐based protein production, a high eYFP production level was found under high cell density conditions in the bioreactor. In fact, the fluorescence intensity of the bacterial culture steadily increased after IPTG induction ( t = 9 h), reaching a maximum of 9537 ± 98 AU∙10 4 after 27 h of cultivation. A highly correlating trend could be also observed for the intracellular eYFP titer, which increased up to a maximum of 2.7 ± 0.3 g/L until the end of cultivation. In comparison to production performance under batch conditions in shake flasks, maximum values for fluorescence level and eYFP titre could be enhanced by factors of 42 and 35, respectively. Beyond that, the relative target protein content was found to be 2.2‐fold higher compared to shake flask cultivations (7.9 ± 0.1%) and steadily increased up to a maximum of 17.7 ± 2.5% of soluble protein. This is reflected by the representative expression pattern shown in SDS‐PAGE analysis (Figure 3B ), where an eYFP protein band with continuously increasing intensity could be observed at a size slightly above 25 kDa (theoretical mass: 27 kDa). The extended post‐induction cultivation time of 18 h realized under fed‐batch conditions was also associated with a 1.5‐fold higher biomass‐specific product yield Y \n PIX compared to batch cultivations with 10 h of post‐induction growth (40.0 ± 1.0 mg eYFP /g CDW vs. 26.8 ± 1.1 mg eYFP /g CDW ). Interestingly, the overall productivity P V was still found to be maximal (0.10 ± 0.02 g/L∙h) until the end of cultivation. This gives reason to assume that an extended production phase could also lead to even higher eYFP protein titers. However, due to the steadily increasing substrate concentrations which might have caused growth (and product) inhibition as previously shown for wild‐type ATCC 13032 (Kiefer, Merkel, Lilge, Hausmann et al., 2021 ), fed‐batch cultivation experiments were certainly stopped after a total process time of 27 h. For this reason, the presented acetate‐based production process may thus be further optimized to first avoid substrate accumulation over time and second to extend the production phase for reaching even higher product accumulation. This might be either realized by a manual increase of the C/N feeding ratio in the later process phase as already demonstrated (Kiefer, Merkel, Lilge, Hausmann et al., 2021 ) or by applying a model‐based feeding control which continuously adapts the applied C/N feeding ratio to the actual C/N consumption ratio of the bacterial culture. Additionally, an optimization of the induction time point for initiating the production phase might affect the final product titer. However, the presented bioprocess was rather intended to be a first proof of concept demonstrating the feasibility of acetate‐based protein production in platform organism C. glutamicum under industry‐relevant fed‐batch cultivation conditions. To the best of our knowledge, there is only one study in literature so far dealing with recombinant protein production using acetate as carbon source (Leone et al., 2015 ). This is most likely due to the fact that acetate is generally reported to inhibit recombinant protein production, which is why most studies have only focused on minimizing or fully eliminating acetate formation during heterologous protein production (De Anda et al., 2006 ; Jensen & Carlsen, 1990 ; Lozano Terol et al., 2019 ; Waegeman et al., 2013 ; Wong et al., 2008 ). In the only study from Leone and colleagues, a sweet protein (MNEI) was recombinantly produced on acetate with average concentrations of about 180 mg/L using E. coli T7 expression system. However, the described production experiments were only restricted to batch cultures and did not include any process strategies for fed‐batch cultivations. Compared to general research works which study heterologous protein production in C. glutamicum fed‐batch cultivations using common carbohydrate carbon sources, the protein titer and overall productivity achieved in our presented acetate‐based bioprocess can be regarded as highly competitive (Table 3 ). For instance, the recombinant enzymes α ‐amylase, endoxylanase and cutinase have been produced in highest titers of 0.8 g/L (Yim et al., 2015 ), 1.8 g/L (Zhang et al., 2019 ) and 1.5 g/L (Bakkes et al., 2021 ), respectively. In addition, product titers of 1.6, 0.3 and 1.8 g/L have been achieved for production of the biopharmaceutical relevant proteins VHH (Yim et al., 2015 ), SUMO‐NT‐proBNP (Peng et al., 2019 ) and PINP (Sun et al., 2020 ). When comparing the product titers, it must be considered that many target proteins of other studies have been produced in an extracellular manner in contrast to intracellular protein accumulation in this study. So far, only the study from Ravasi et al. ( 2015 ) reported a higher titer of 5.5 g/L for heterologous production of Phospholipase C in a fed‐batch cultivation of C. glutamicum ATCC 13869. In conclusion, the results of this study indicate that acetate might be also a highly promising and competitive alternative platform substrate for production of heterologous proteins in industry‐relevant titers (g/L). TABLE 3 Summary of existing literature dealing with heterologous protein production in fed‐batch cultivations of C. glutamicum strains (2015–2022) Target protein Cellular localization \n C. glutamicum host strain Carbon source (type of medium) Product titer [g/L] Process time [h] Productivity \n a \n [g/L∙h] References eYFP Intracellular MB001(DE3) Acetate (defined) 2.7 27 0.1 This study Cutinase‐GFP11 Extracellular ATCC 13032 K9 Glucose (defined) 1.5 25 0.06 Bakkes et al. ( 2021 ) PINP Extracellular CGMCC1.15647 Glucose (semi‐defined) 1.2 44 0.03 Sun et al. ( 2020 ) Endoxylanase Extracellular CGMCC1.15647 Δ cspB2 Δ clpSInX \n Maltose (semi‐defined) 1.8 44 0.04 Zhang et al. ( 2019 ) SUMO‐NT‐proBNP Intracellular CGMCC1.15647Δ clpC Δ porB Δ mepA \n Glucose (semi‐defined) 0.3 \n b \n \n 28 0.01 Peng et al. ( 2019 ) \n α ‐Amylase Extracellular ATCC 13032 Glucose (semi‐defined) 0.8 21 0.04 Yim et al. ( 2015 ) Endoxylanase Extracellular ATCC 13032 Glucose (semi‐defined) 1.1 32 0.03 Yim et al. ( 2015 ) Camelid antibody fragment (VHH) Extracellular ATCC 13032 Glucose (semi‐defined) 1.6 30 0.05 Yim et al. ( 2015 ) Phospholipase C Extracellular ATCC 13869 Glucose (semi‐defined) 5.5 55 0.1 Ravasi et al. ( 2015 ) \n a \n Calculated from the given maximum titer and respective process time. \n b \n Calculated from the given yield of 29.96 mg/gDCW. Recombinant production and purification of a His(6)‐tagged eYFP variant To demonstrate the feasibility of the established acetate‐based production process for subsequent product recovery up to pure protein, the bioprocess was finally applied to an affinity‐tagged eYFP variant. For this purpose, we synthetically added a hexahistidine sequence to the 5′ end of the eyfp gene construct from Kortmann et al. ( 2015 ) in order to allow a simple one‐step protein purification from the cell lysate. The resulting expression vector pMKEx2‐His(6)‐ eyfp was transformed into C. glutamicum strain MB001(DE3) and the newly constructed expression strain was applied for acetate‐based His(6)‐eYFP production under high cell density conditions as described before. To isolate the expressed target protein, biomass samples from 40 ml culture broth were disrupted using a high‐pressure homogenizer. Thereafter, the soluble protein fractions were used for His(6)‐eYFP purification by immobilized metal affinity chromatography (IMAC). As shown in Figure 4A , a sharp peak of the UV signal in the chromatogram (blue line) visible after elution with a one‐step gradient indicated that the recombinantly produced His(6)‐eYFP was successfully isolated and captured. A dominant protein band with a size of ~27 kDa in SDS‐PAGE analysis (Figure 4B ) confirmed the presence of the overexpressed target protein in the different disruption cycles (lanes 1–5). As can be seen in the eluted peak fraction (lane 8), a highly purified target protein band with a purity of approximately 88 ± 2% was obtained. In total, an amount of 42 ± 2 mg purified His(6)‐eYFP could be recovered from 40 ml culture broth by using this simple capturing purification step. Taken together, the exemplary high‐level recombinant production based on acetate as carbon source followed by subsequent target protein recovery was also successfully demonstrated for the newly constructed His(6)‐tagged eYFP variant. FIGURE 4 Exemplary isolation and purification of His(6)‐tagged eYFP produced in high cell density cultivation of C. glutamicum MB001(DE3) on acetate as carbon source. Biomass pellets (~7 g) from 40 ml of culture broth were disrupted with a high pressure homogenizer. The soluble protein fraction was subsequently used for His(6)‐eYFP purification by IMAC using an automated chromatography system. (A) Chromatogram of a representative purification run showing the courses of UV signal (blue line), conductivity (black line) and used gradient concentration (red line). (B) SDS‐PAGE analysis showing the soluble protein fractions at different disruption cycles (lanes 1–5) as well as flow‐through, wash and elution fractions from protein purification (lanes 6–8). Equivalent amounts of total protein (6 μg for lanes 1–6, 0.6 μg for lane 7, 2 μg for lane 8) were loaded onto the gel and protein bands were visualized by Coomassie brilliant blue staining. St, purified His(6)‐eYFP protein standard; M, protein marker with molecular weights indicated in kDa. (C) Picture taken from glowing, purified His(6)‐eYFP elution fraction."
} | 9,047 |
36929912 | PMC10075063 | pmc | 655 | {
"abstract": "Abstract Gram-positive Firmicutes bacteria and their mobile genetic elements (plasmids and bacteriophages) encode peptide-based quorum-sensing systems (QSSs) that orchestrate behavioral transitions as a function of population densities. In their simplest form, termed “RRNPP”, these QSSs are composed of two adjacent genes: a communication propeptide and its cognate intracellular receptor. RRNPP QSSs notably regulate social/competitive behaviors such as virulence or biofilm formation in bacteria, conjugation in plasmids, or lysogeny in temperate bacteriophages. However, the genetic diversity and the prevalence of these communication systems, together with the breadth of behaviors they control, remain largely underappreciated. To better assess the impact of density dependency on microbial community dynamics and evolution, we developed the RRNPP_detector software, which predicts known and novel RRNPP QSSs in chromosomes, plasmids, and bacteriophages of Firmicutes . Applying RRNPP_detector against available complete genomes of viruses and Firmicutes , we identified a rich repertoire of RRNPP QSSs from 11 already known subfamilies and 21 novel high-confidence candidate subfamilies distributed across a vast diversity of taxa. The analysis of high-confidence RRNPP subfamilies notably revealed 14 subfamilies shared between chromosomes/plasmids/phages, 181 plasmids and 82 phages encoding multiple communication systems, phage-encoded QSSs predicted to dynamically modulate bacterial behaviors, and 196 candidate biosynthetic gene clusters under density-dependent regulation. Overall, our work enhances the field of quorum-sensing research and reveals novel insights into the coevolution of gram-positive bacteria and their mobile genetic elements.",
"introduction": "Introduction Quorum sensing is the mechanism by which microbial entities sense when their population density reaches a threshold level and thereupon typically switch from individual to group behaviors ( Mukherjee and Bassler 2019 ). The population density is reflected by the extracellular concentration of a communication signal, produced and secreted by individual entities. The quorum is met when this signal reaches a threshold concentration, at which it starts to be robustly detected and transduced population-wide by its cognate receptor module. If quorum sensing seems to be used by diverse prokaryotic and unicellular eukaryotic lineages ( Hornby et al. 2001 ; Paggi et al. 2003 ; Sun et al. 2004 ; Sharif et al. 2008 ; Tian et al. 2018 ), most of the knowledge about this communication mechanism comes from the three Pseudomonadota (formerly Proteobacteria ), Actinomycetota (formerly Actinobacteria ), and Bacillota (formerly Firmicutes ) bacterial phyla. In the Pseudomonadota / Proteobacteria and Actinomycetota/Actinobacteria phyla, the communication signals typically are small molecules synthesized by enzymes ( Papenfort and Bassler 2016 ; Polkade et al. 2016 ), whereas in the Bacillota/Firmicutes phylum, these are oligopeptides, matured from genetically encoded propeptides ( Bhatt 2019 ). Peptide-based quorum-sensing systems (QSSs) can be divided into two main categories: those with a receptor module composed of a membrane-bound sensor coupled with an intracellular response regulator (two-component system) like the ComX-ComQ-ComP-ComA of Bacillus subtilis ( Sturme et al. 2002 ) and those in which the receptor is an intracellular transcription factor (or a protein inhibitor) that gets either turned-on or -off upon binding with the imported communication peptide (one-component system) ( Rocha-Estrada et al. 2010 ; Neiditch et al. 2017 ). The latter are generally included under the term RRNPP, named after the five first experimentally characterized subfamilies of such receptors: Rap ( Bacillus genus), Rgg ( Streptococcus genus), NprR ( Bacillus cereus group), PlcR ( B. cereus group), and PrgX (pCF10 plasmid of Enterococcus faecalis ) ( Do and Kumaraswami 2016 ; Perez-Pascual et al. 2016 ; Neiditch et al. 2017 ). The initial members of the RRNPP group of QSSs were reported to trigger key biological pathways when their encoding population reaches high densities: from virulence (Rgg, PlcR) to competence (Rgg, Rap), necrotropism (NprR), sporulation, biofilm formation (Rap, NprR), and inhibition of conjugation (PrgX) ( Do and Kumaraswami 2016 ; Perez-Pascual et al. 2016 ; Neiditch et al. 2017 ). Considering that the virulence of Bacillus and Streptococcus pathogens may cause infectious diseases in humans ( Baldwin 2020 ; Lannes-Costa et al. 2021 ), that the spore is the transmissive form of many Bacillus and Clostridium human pathogens ( Mallozzi et al. 2010 ), that biofilms contribute to infections or food poisoning ( Costerton et al. 1999 ; Høiby et al. 2011 ; Galié et al. 2018 ), and that competence and conjugation are responsible for the spread of antibiotic resistance genes ( von Wintersdorff et al. 2016 ), RRNPP QSSs are directly linked to central health issues. Interestingly, the case of the plasmidic PrgX system illustrates that RRNPP QSSs may be not only present on bacterial chromosomes but also on mobile genetic elements (MGEs). However, conjugative elements are not the only MGEs relying on RRNPP QSSs as a means to assess their population density. Indeed, in 2017, Erez et al. made the groundbreaking discovery of the viral “arbitrium” communication system, an RRNPP QSS encoded by temperate phages of Bacillus and guiding the lysis–lysogeny decision upon Bacillus infection ( Erez et al. 2017 ; Stokar-Avihail et al. 2019 ). In total, to our knowledge, we can count today 11 subfamilies of RRNPP receptors with experimental evidence of interaction with a communication peptide: the five aforementioned initial RRNPP members Rgg, Rap, NprR, PrgX, and PlcR (which can be divided into the PlcR and TprA subfamilies [ Hoover et al. 2015 ]) ( Neiditch et al. 2017 ) and the six following additional members: TraA—plasmids of E. faecalis ( Kohler et al. 2019 ), AimR—temperate phages of Bacillus ( Stokar-Avihail et al. 2019 ), ComR— Streptococcus genus ( Shanker et al. 2016 ), AloR— Paenibacillaceae family ( Voichek et al. 2020 ), Qsr— Clostridium acetobutylicum ( Kotte et al. 2020 ), and QssR— Clostridium saccharoperbutylacetonicum ( Feng et al. 2020 ) ( fig. 1 B ) \n Fig. 1. Characteristics of RRNPP QSSs. ( A ) Canonical molecular mechanism of communication via an RRNPP QSS. . An RRNPP QSS can be encoded by chromosomes, plasmids, phage genomes, or prophages (phage genomes inserted within the bacterial genome). Either way, upon bacterial expression, the propeptide is secreted via the bacterial SEC translocon and is cleaved extracellularly into a short mature communication peptide. As a QSS-encoding element replicates, the communication peptide accumulates in the extracellular environment. At high concentrations of the peptide, reflecting a quorum of bacterial cells, plasmids, and/or (pro)phages, the peptide starts to be frequently imported within bacterial cells. In bacterial cells hosting the QSS-encoding genetic element(s), the communication peptide binds to the TPRs of its cognate cytosolic receptor. Consequently, the receptor gets either turned-on or -off as a protein inhibitor or as a transcription factor, which is at the basis of density-dependent regulations of target proteins or genes. As a result, a behavioral transition is initiated at the scale of the entire QSS-encoding population. ( B ) Common features between experimentally validated RRNPP QSSs. Each genomic context corresponds to the representative QSS of an experimentally validated subfamily of RRNPP QSSs. . A gray gene indicates an adjacent target gene (or set of genes) demonstrated to be regulated by the QSS. The legend on the top-left corresponding to the PlcR-PapR QSS indicates all genomic features being displayed for each QSS. The different QSSs share a computationally testable signature of five criteria: 1) the propeptide is small; 2) the propeptide is secreted by the SEC translocon (computationally testable by SignalP); 3) the receptor is ∼250–500aa long; 4) the receptor harbors TPRs involved in the recognition of the mature communication peptide (computationally testable by HMMs of TPRs); and 5) the receptor and the propeptide genes are direct neighbors. ( C ) RRNPP QSSs involving a secretion of the propeptide via the alternative PptAB translocon. Consistently, SignalP did not predict a SEC-dependent secretion for them (as shown by a SEC-secretion likelihood score colored in gray). Yet, the genetic diversity of RRNPP QSSs may not have been fully explored, as hinted, for instance, by the candidate receptors of E. faecalis reported to harbor local similarities with regions of Rap, PlcR, or Rgg ( Parthasarathy et al. 2020 ). Hence, new communication codes as well as novel density-dependent evolutionary strategies likely await to be discovered. These discoveries not only could transform our views of microbial interaction, adaptation, and evolution, but also could have major practical outcomes as novel communication systems could regulate the production of new antimicrobial compounds ( Hoover et al. 2015 ; Rued et al. 2021 ) or could underlie adaptive mechanisms by which some human pathogens acquire virulence ( Edwards et al. 2016 , 2019 ; Do et al. 2017 , 2019 ). However, expanding this diversity requires overcoming an important challenge: identifying candidate systems beyond close homologs of already known RRNPP subfamilies. Conveniently, we noticed that the members of all the aforementioned experimentally validated RRNPP subfamilies share a common signature of five criteria ( fig. 1 B ): 1) the propeptide is a small protein (10–100aa); 2) the propeptide is secreted via the SEC translocon and further matured by exopeptidases into a communication peptide (with the exception of propeptides of short hydrophobic peptides (SHPs) and PrgQ mature peptides associated with Rgg and PrgX receptors that are translocated via the PptAB export system[ Neiditch et al. 2017 ]); 3) the receptor has a length comprised between 250 and 500aa; 4) the receptor harbors tetratricopeptide repeats (TPRs), which are structural motifs involved in the binding of small peptides (in this case, the cognate communication peptide); and 5) the genes encoding the propeptide and the receptor are directly adjacent to each other. Advantageously, a large amount of reference hidden Markov models (HMMs) from the Cath-Gene3D ( Sillitoe et al. 2021 ), Superfamily ( Wilson et al. 2009 ), SMART ( Letunic et al. 2021 ), Pfam ( Mistry et al. 2021 ), and TIGRFAM ( Haft et al. 2013 ) databases are already available to detect TPRs in protein sequences. Moreover, a tool called SignalP specifically computes the likelihood that proteins harbor a signal sequence for the SEC translocon ( Almagro Armenteros et al. 2019 ) ( fig. 1 ). Consequently, the generic, yet specific signature of RRNPP QSSs could be detectable in silico, without requiring homology searches that would limit the output to representatives of already known QSSs. On this basis, we have developed RRNPP_detector, a Python software dedicated to the detection of the RRNPP signature in chromosomes, plasmids, and bacteriophages of gram-positive bacteria, available at https://github.com/TeamAIRE/RRNPP_detector . The fact that the Rgg and PrgX subfamilies involve a secretion of their cognate SHP and PrgQ propeptides via the PptAB translocon rather than via the SEC translocon ( Neiditch et al. 2017 ) implies that some functional RRNPP QSSs can slightly deviate from the previously described canonical signature. Accordingly, RRNPP_detector was designed to identify putative QSSs with three different strictness levels: 1) the “strict” level outputs all candidate receptor–propeptide pairs with the propeptide being annotated or preceded by a high-confidence ribosomal binding site (RBS) motif and either matching the HMM profile of SHPs or predicted to undergo a SEC/SPI-dependent secretion according to SignalP; 2) the “relaxed” level outputs all remaining receptor–propeptide pairs in which the propeptide harbors any of the SP(Sec/SPI), TAT(Tat/SPI), or LIPO(Sec/SPII) secretion tag according to PrediSi ( Hiller et al. 2004 ) or SignalP ( Almagro Armenteros et al. 2019 ); and 3) the “loose” level outputs remaining TPR-containing putative receptors only if found adjacent to a peptide without a detected secretion tag but with a high-confidence upstream Shine–Dalgarno RBS (SD RBS) motif indicative of a likely translation ( Shine and Dalgarno 1975 ; Omotajo et al. 2015 ), with the cognate peptide being chosen as the most likely translated small protein in the close genomic vicinity of the candidate receptor (fig. 2 ). Of course, the relaxed and loose outputs are associated with a higher risk of false positives but are nonetheless interesting for exploratory purposes. \n Fig. 2. Workflow of RRNPP_detector illustrated with real data from complete genomes/chromosomes of Firmicutes and viruses. RRNPP_detector defines candidate RRNPP-type QSSs with a “strict” detection strictness level as tandems of adjacent genes encoding a candidate receptor (250–500aa protein matching HMMs of peptide-binding TPRs) and a candidate propeptide (10–100aa protein predicted by SignalP to be secreted via the SEC translocon or matching the HMM of SHP propeptides). Each green and red rectangle represents a step toward the final identification of “strict” candidate receptors and candidate propeptides, respectively (details in Materials and Methods). The final “strict” receptors and propeptides subsequently serve as queries in a Blastp search to identify additional homologous QSSs that did not pass the conservative thresholds of RRNPP_detector. Additional pairs are predicted with either a “relaxed” or a “loose” detection strictness level (Materials and Methods). To assess the extent of the impact of RRNPP QSSs on microbial community dynamics and evolution, we applied RRNPP_detector against all complete genomes and chromosomes of Firmicutes and viruses available on the NCBI. We report a wide phylogenetic, genetic, and functional diversity of RRNPP QSSs that enhances our current knowledge of the coevolution of gram-positive Bacillota and their MGEs.",
"discussion": "Discussion We predicted a wide range of novel candidate RRNPP QSSs in chromosomes or in MGEs (e.g., plasmids and phages) of Firmicutes using a computational approach that does not rely on sequence similarity search using known QSS proteins as queries ( supplementary tables S1–S3 , Supplementary Material online and figs. 2 and 3 ). This massive, publicly available library of candidate communication systems shows great promise for the characterization of many density-dependent mechanisms in bacteria, plasmids, and phages, with major fundamental and applied outcomes. In this regard, we gave a few examples of how the prediction of QSSs can be coupled with functional insights, by exploiting the trend that the target regulon of an RRNPP QSS often lies in its genomic vicinity, a trend especially true for MGEs ( Erez et al. 2017 ; Neiditch et al. 2017 ; Kohler et al. 2019 ; Stokar-Avihail et al. 2019 ). This allowed to highlight a putative convergent evolution of the functional association between QS-mediated phage–phage communication and bacterial sporulation manipulation, with nonhomologous viral QSSs from different bacteriophage species found adjacent to a viral homolog of one of the bacterial Rap, AbrB, or Spo0E sporulation regulator ( fig. 6 ). If this association was experimentally validated, the fact that phages and/or prophages could communicate to decipher when it is the most evolutionary advantageous to manipulate host pathways would capture a novel feature of bacteria–phages coevolution, since the experimentally validated phage-encoded QSSs were thus far shown to only regulate viral processes such as the lysis–lysogeny decision ( Erez et al. 2017 ). In addition, this validation could invite to reconsider the sporulation decision-making process as a biological process that may sometimes fall under the scope of a (pro)phage–host collective, rather a strict bacterial process of last resort, with important implications considering that the endospore is the transmissive form of many bacteria, be they commensal or pathogens for humans ( Mallozzi et al. 2010 ; Postollec et al. 2012 ; Swick et al. 2016 ). It is also interesting to mention that we found one of these putative host-hijacking QSS, the Rap–Phr system, in co-occurrence with the arbitrium system within 67 Bacillus phage genomes ( supplementary table S4; Supplementary Material online). Although the arbitrium system regulates the lysis–lysogeny transition upon Bacillus infection, we previously hypothesized that prophage-encoded Rap–Phr systems might confer upon lysogenized hosts selective advantages over nonlysogenized hosts such as the evasion to public good production at low population densities, for the evolutionary benefit of the prophage–host collective ( Bernard et al. 2020 ). In general, owning multiple QSSs regulating distinct biological processes might enable behavioral transitions according to different regimes of densities, reflected by the different quorums associated with each QSS ( Mehta et al. 2009 ). In total, of the 2,078 MGEs within which at least one “strict” candidate QSS has been predicted, 263 were found to encode more than one QSS ( supplementary table S4; Supplementary Material online). The observation that 12.65% of the QSS-encoding detected MGEs encode multiple QSSs generalizes the notion that phages and plasmids may subtly assess changes in their social context and adapt their evolutionary strategy accordingly. In light of the consideration that different QSSs owned by an MGE can be more or less conserved across nonkin MGEs, neighbors, or hosts, such “multilingual” MGEs could theoretically react to the density of multiple heterogeneous subpopulations to which these MGEs nonetheless always belong. Accordingly, encoding several QSSs more or less specific to its kins might enable an MGE to contextualize its own population density with respect to that of other heterogeneous populations. In addition to these fundamental aspects of bacteria–MGE coevolution, a more applied example of the functional investigations conducted in this study was given by the identification of 196 BGCs of specialized metabolism inferred to be regulated by a candidate RRNPP QSS. Importantly, the predicted density-dependent expression of these BGCs hints at important adaptive ecological roles for the metabolites they produce. Thus, functional characterization of these BGCs may not only lead to the discovery of novel molecules of applied interest, such as novel antimicrobial molecules or candidate virulence factors to fight against, but could also be rich in lessons to better understand the lifestyle of their encoding species. Overall, our analyses demonstrate that our methodology can unlock new biological knowledge regarding peptide-based biocommunication and can reveal novel density-dependent decision-making processes in bacteria, plasmids, and bacteriophages, with potential to enhance our understanding of microbial adaptation and bacteria–MGE coevolution. Yet, the communication systems described in this study likely do not represent the entire landscape of RRNPP QSSs. Indeed, we analyzed only complete genomes of Firmicutes , and many candidate RRNPP QSSs likely await to be unearthed in bacterial scaffolds, contigs, and metagenomics-assembled genomes. In this respect, it is important to mention that the Firmicutes phylum represents with Bacteroidetes the most prevalent phylum in human gut microbiomes ( Manor et al. 2020 ). Accordingly, the application of our publicly available RRNPP_detector software against human-associated metagenomics-assembled genomes or MGEs (e.g., from the human MGE database [ Lai et al. 2021 ] or the Gut Phage Database [ Camarillo-Guerrero et al. 2021] ) would be of high relevance to infer density-dependent behaviors that may take place within human intestinal microbiomes, plasmidomes, and viromes."
} | 5,058 |
29034004 | PMC5629779 | pmc | 657 | {
"abstract": "Background Photosynthetic microalgae are emerging as potential biomass feedstock for sustainable production of biofuels and value-added bioproducts. CO 2 biomitigation through these organisms is considered as an eco-friendly and promising alternative to the existing carbon sequestration methods. Nonetheless, the inherent relatively low photosynthetic capacity of microalgae has hampered the practical use of this strategy for CO 2 biomitigation applications. Results Here, we demonstrate the feasibility of improving photosynthetic capacity by the genetic manipulation of the Calvin cycle in the typical green microalga Chlorella vulgaris . Firstly, we fused a plastid transit peptide to upstream of the enhanced green fluorescent protein (EGFP) and confirmed its expression in the chloroplast of C. vulgaris . Then we introduced the cyanobacterial fructose 1,6-bisphosphate aldolase, guided by the plastid transit peptide, into C. vulgaris chloroplast, leading to enhanced photosynthetic capacity (~ 1.2-fold) and cell growth. Molecular and physiochemical analyses suggested a possible role for aldolase overexpression in promoting the regeneration of ribulose 1,5-bisphosphate in the Calvin cycle and energy transfer in photosystems. Conclusions Our work represents a proof-of-concept effort to enhance photosynthetic capacity by the engineering of the Calvin cycle in green microalgae. Our work also provides insights into targeted genetic engineering toward algal trait improvement for CO 2 biomitigation uses. Electronic supplementary material The online version of this article (doi:10.1186/s13068-017-0916-8) contains supplementary material, which is available to authorized users.",
"conclusion": "Conclusions In this study, we first employed EGFP as a reporter to demonstrate that expressing a nucleus-encoded heterologous protein is achievable in chloroplast of the most typical green microalga C. vulgaris . Then we generated transgenic C. vulgaris lines expressing cyanobacterial aldolase in chloroplasts and assessed the impact of increased aldolase activity on physicochemical characteristics. Compared with WT cells, transgenic C. vulgaris cells showed significantly enhanced photosynthetic capacity and cell growth, highlighting its great potential in CO 2 biomitigation. Our work represents a proof-of-concept effort to improve the photosynthetic capacity by the engineering of the Calvin cycle in green microalgae. This study will provide implications into targeted genetic engineering toward algal trait improvement for CO 2 biomitigation uses in future. Besides, the success in subcellular localization of EGFP and visual detection of fluorescence in live green algal cells will to some extent expand the application of fluorescent protein technology in microalgal molecular biology.",
"discussion": "Discussion The inherent relatively low photosynthetic capacity of green microalgae represents one of the most critical issues that remain to be addressed in the practical CO 2 biomitigation applications. Strain improvement through genetic engineering appears to be a feasible strategy to overcome this obstacle [ 4 , 18 ]. Until now, to the best of our knowledge, there has been no information on the feasibility of expressing a nucleus-encoded heterologous protein in C. vulgaris chloroplast, which is of importance to the genetic engineering of the Calvin cycle for improved photosynthetic capacity. In a previous report, we have established a stable genetic system for C. vulgaris using EGFP as an effective reporter [ 24 ]. On this basis, we also employed EGFP as a reporter and demonstrated the correct targeting of EGFP into chloroplast by the rbcS transit peptide (Fig. 2 b). This is, to our knowledge, the first report with regard to plastid localization of a fluorescent reporter in Chlorella species. In addition, our results indicated that expressing a nucleus-encoded heterologous protein is achievable in C. vulgaris chloroplast, providing ample support for the subsequent overexpression experiment. More recent studies have shown that CaMV35S promoter shows poor performance in driving transgene expression in C. reinhardtii [ 26 , 27 ]. In contrast, our results showed that CaMV35S promoter could efficiently drive transgene expression in C. vulgaris , in line with those ever reported in different Chlorella species [ 24 , 28 , 29 ]. Nevertheless, using this promoter in our study, the level of aldolase activity in transgenic cell lines just increased modestly (1.27–1.30-fold, Fig. 3 f), lower than that previously reported in higher plants (1.4–1.9-fold) [ 20 ]. This indicated that this promoter might not be a desirable promoter for highly efficient expression of transgenes in C. vulgaris . Other strong endogenous or exogenous promoters, such as RbcS and Chlorella virus promoters, may therefore need to be tested and developed for this alga in our future studies. It is generally acknowledged that a transgene highly homologous to the native gene usually has a high susceptibility to gene silencing [ 30 ]. Because aldolase genes from higher plants or eukaryotic microalgae usually have a high homology (> 60%, data not shown) with Chlorella aldolase gene, to avoid the possible gene silencing scenario, we chose a cyanobacterial aldolase with a low homology (~ 30%) in the present study. We succeeded in generating transgenic C. vulgaris lines that expressed mature cyanobacterial aldolase in the chloroplast, leading to a modest but significant increase in aldolase activity. This demonstrated that the cyanobacterial aldolase functioned efficiently in the chloroplast of transgenic cells. Transgenic lines also exhibited phenotypes under ambient CO 2 conditions with elevated O 2 evolution rates and CO 2 fixation rates (photosynthetic capacity) over WT cells, resulting in a modest increase in biomass production and chlorophyll concentration. To further verify the positive impact of aldolase overexpression on cell growth, we conducted a ‘semi-continuous-like’ experiment under normal conditions by inoculating from 10-day cultures at 0.5 g L −1 and sampling after 4-day cultivation. Similarly, we also found a modest biomass enhancement in transgenic lines (data not shown). It has been shown in plants that aldolase, previously considered as a non-regulated enzyme, plays an important role in regulating the photosynthetic carbon flux [ 20 , 31 ]. Our results also supported this finding and demonstrated that enhancing aldolase activity was effective to promote CO 2 fixation and biomass productivity in microalgae. In our work, the phenotypes observed under ambient conditions are in agreement with the results observed in transgenic tobacco expressing cyanobacterial FBP/SBPase in plastids, in which photosynthetic activity and growth were both enhanced under ambient CO 2 concentration [ 15 ]. However, in contrast to our results, Uematsu et al. [ 20 ] found that the overexpression of plastid aldolase had no significant effect on photosynthetic activity under ambient CO 2 concentration in transgenic tobacco. These different phenotypes may probably be ascribed to the differences in genetic traits of species and Rubisco activity among three groups of transgenic organisms. Notably, we found that transgenic lines were able to show phenotypes even after about 1-year subcultures (data not shown), suggesting good genetic stability for potential practical applications. We only evaluated the impact of aldolase overexpression under ambient CO 2 conditions in our work. However, in most cases for practical outdoor applications, algal cultures are often aerated with supplemental CO 2 (e.g., 2% CO 2 ). Hence, on the one hand, it may be useful to examine the photosynthetic performance of our transgenic algal strains under high CO 2 concentration. Our preliminary results suggested that transgenic lines could also show enhanced photosynthesis and growth when supplemented with 1.5% CO 2 (data not shown). Further full evaluation of transgenic lines under high CO 2 concentration either indoor or outdoor will provide valuable information for possible practical applications in future. On the other hand, it may likely be important to screen more robust transgenic algal lines, e.g., further enhancing aldolase gene expression with stronger endogenous promoters as discussed above, for the comprehensive indoor/outdoor evaluation in open ponds and/or photobioreactors under high CO 2 concentration. In our study, the carbon content of transgenic lines did not change in response to increased photosynthetic CO 2 fixation (Table 1 ). This may be explained by an inconsiderable change in cellular components. In addition, aldolase overexpression had no noticeable effect on the (total) activities of tested key enzymes in the Calvin cycle (Fig. 6 b), in keeping with the data from Uematsu et al. [ 20 ]. However, the transcript levels of these key enzymes (except PRK) were all significantly increased, especially Rubisco (Fig. 6 a). This discrepancy between transcript levels and enzyme activities may be attributed to sophisticated regulation of intracellular protein expression such as post-translational modification and negative feedback regulation. Interestingly, we found that the initial activity of Rubisco in transgenic lines was significantly higher than that in WT cells (Fig. 6 b). This is in agreement with a previous report that cyanobacterial FBP/SBPase was overexpressed in higher plants [ 15 ], suggesting that increased aldolase would induce more in vivo activated state of Rubisco. It has been observed that the attenuated aldolase activity by antisense knockdown in higher plants inhibited RuBP regeneration and thus impaired photosynthesis and growth [ 31 , 32 ]. In addition, previous reports have shown that the photosynthetic rate is limited by Rubisco capacity to regenerate RuBP at relatively low CO 2 levels [ 33 ]. On this basis, as Rubisco is well known to be activated by Rubisco activase, we may infer that increased chloroplastic aldolase might induce high activity of Rubisco activase by promoting regeneration of RuBP in the Calvin cycle, thereby giving rise to more activated state of Rubisco in transgenic lines. The resultant more in vivo activation of Rubisco might therefore accelerate carbon turnover rate in the Calvin cycle and thus stimulate photosynthesis and growth. Enhancement of RuBP regeneration by overexpressing aldolase or other Calvin cycle enzyme such as FBP/SBPase has been previously investigated in cyanobacteria and higher plants [ 15 , 20 – 22 ]. Notably, in such aldolase-overexpressed cyanobacteria, increased aldolase was demonstrated to raise RuBP level by acceleration of SBPase. In our work, however, it remains unclear if elevated RuBP level was indirectly induced by increased levels of the other Calvin cycle enzymes caused by aldolase enhancement. To elucidate this mechanism more unequivocally, further investigation on the carbon partitioning, e.g., quantification of the dynamic levels of carbohydrates, especially the intermediates involved in the Calvin cycle, is in need, which is currently in progress in our laboratory. Our data have clearly shown that aldolase overexpression did not affect the maximal quantum efficiency of PSII ( F \n v / F \n m ) and the efficiency of open PSII centers ( F \n v ′/ F \n m ′) (Fig. 5 a, b). This is in accordance with the results observed in transgenic tobaccos [ 15 , 20 ]. However, enhancement of aldolase activity in chloroplast gave rise to a noticeable lower NPQ value and higher qP and Φ \n PSII values in transgenic lines (Fig. 5 c–e). The lower NPQ value indicated a decreased thermal dissipation of PSII, while the higher qP and Φ \n PSII values suggested an increased activity of reaction center of PSII and a promoted efficiency of electron transport in PSII. In other words, these results implied that transgenic lines were likely to have a faster energy transfer in photosystems than did WT cells, as could be expected by the significantly increased true rate of O 2 evolution of transgenic lines (Fig. 4 c). Overall, a possible working model to elucidate the role of aldolase overexpression for improved photosynthetic capacity was proposed and is depicted in Fig. 7 . The enhanced chloroplastic aldolase may probably lead to an increase in RuBP levels and thus induce more activated state of Rubisco, resulting in an accelerated carbon turnover rate in the Calvin cycle. The resultant acceleration in this cycle may raise demands on assimilatory power (ATP and NADPH), generating a pulling force and thereby stimulating energy transfer in photosystems to produce higher levels of ATP and NADPH. This will in turn impose accelerated carbon flow in the Calvin cycle and finally boost biomass production in C. vulgaris . Fig. 7 A proposed working model for the role of aldolase in photosynthetic CO 2 fixation. (i) Increased aldolase in chloroplast (ii) may induce more activated state of Rubisco by promoting regeneration of RuBP (iii) and thus stimulate carbon turnover rate in the Calvin cycle. (iv) This may provide a pulling power and raise electron transport in photosystems (v) to produce more NADPH and ATP, (vi) which in turn impose increased carbon turnover rate in the Calvin cycle (vii) and finally lead to increased biomass production in C. vulgaris . A red up arrow indicates upregulation \n \n Chlorella vulgaris has demonstrated its potential for lipid production [ 6 ]. Interestingly, overexpression of aldolase also enhanced total fatty acid content in this alga, though triacylglycerol (TAG) content showed little change (Additional file 2 : Figure S2), indicating a possible increase in membrane polar lipids (especially chloroplast lipids) caused by enhanced photosynthetic capacity. Although the mechanism for this phenomenon remains unclear, this may point to the possible integration of CO 2 fixation with biofuel production. Future genetic approaches to manipulate C. vulgaris for integrated applications may lie in multiple-gene engineering, e.g., overexpression of aldolase gene (pushing carbon flux to fatty acids) together with diacylglycerol acyltransferase (pulling carbon flux to TAG) [ 34 , 35 ] and/or downregulation of TAG lipase genes (protecting TAG from degradation) [ 36 ]."
} | 3,583 |
33255882 | PMC7728330 | pmc | 658 | {
"abstract": "With the rapid development of wearable electronic systems, the need for stretchable nanogenerators becomes increasingly important for autonomous applications such as the Internet-of-Things. Piezoelectric nanogenerators are of interest for their ability to harvest mechanical energy from the environment with its inherent polarization arising from crystal structures or molecular arrangements of the piezoelectric materials. In this work, 3D printing is used to fabricate a stretchable piezoelectric nanogenerator which can serve as a self-powered sensor based on synthesized oxide–polymer composites.",
"conclusion": "4. Conclusions In summary, a stretchable piezoelectric nanogenerator was developed with BaTiO 3 NPs and photocurable elastomers EAA and AUD, that remain to be applied in printable piezoelectric energy applications. Both piezoelectric materials and electrodes are 3D printed, which makes them the first 3D printed stretchable piezoelectric nanogenerator. Excellent stretchability has been achieved by forming the composite by combining BaTiO 3 NPs and the photocurable elastomer. The maximum strain for the printed BaTiO 3 NP/EAA/AUD sample is 434%. This device generates an output voltage of 0.29 V, the current density is 0.20 μA/cm 2 and the calculated power density is 57 nW/cm 2 . Furthermore, this stretchable piezoelectric nanogenerator has a sensitivity of 59.8 mV/N, and it was demonstrated to real-time monitor the foot stepping signal, which indicates its potential to be used as a self-powered body motion or gait sensor for stretchable or wearable electronic systems.",
"introduction": "1. Introduction Stretchable electronics are becoming essential for the next-generation wearable electronic systems due to their deformability. These can be widely applied in health monitoring, sensing, sports, and many life-quality or well-being improvement purposes. The largest challenge in stretchable electronics is the power supply for portable devices [ 1 ]. To realize autonomous electronics such as self-powered sensors, stretchable energy harvesters are required [ 2 , 3 , 4 ]. A piezoelectric nanogenerator is a kind of energy harvester that converts mechanical energy into electricity by the inherent polarization in the piezoelectric materials. The output power is stable and it can also serve as a self-powered sensor due to the linear relationship of the external force and the output signal [ 5 ]. While a 3D printable stretchable triboelectric nanogenerator has been demonstrated [ 6 ], there was no report on the fully 3D printable stretchable piezoelectric nanogenerator, due to the lack of printable stretchable piezoelectric ink, challenges in poling the piezoelectric elastomer system, and the lack of stretchable conductive ink for 3D printing. Nanostructured oxides including nanowires, nanoparticles, and nanoclusters have attracted attention as composite fillers in many research areas such as energy storage, catalysis, sensing, and energy harvesting due to their functional and mechanical reinforcement to the matrix. Barium titanate (BaTiO 3 ) is one of the most popular piezoelectric oxides with high piezoelectricity, high dielectric constant, and ferroelectricity. The potential applications of BaTiO 3 nanocomposites include high energy capacitors which our group published actively [ 7 ], sensors [ 8 ], bone scaffolds [ 9 ], and piezoelectric devices [ 10 , 11 , 12 ]. In this work, BaTiO 3 nanoparticles were combined with oligomers of 3D printable elastomers to form a 3D printable piezoelectric ink. Stretchable electrodes were also 3D printed onto the stretchable piezoelectric nanocomposite to fabricate an all 3D printed stretchable piezoelectric nanogenerator.",
"discussion": "3. Results and Discussion XRD testing was done on the 3D printed 15 wt % BaTiO 3 NP/EAA/AUD film ( Figure 2 a). The film thickness is about 100 µm. The peak at 2θ = 45° splits into two peaks, indicating that the BaTiO 3 is in the piezoelectric tetragonal phase [ 14 ], all other peaks can be assigned to the BaTiO 3 tetragonal phase in accordance with the International Centre for Diffraction Data (ICDD) PDF #01-074-1965. This indicates that the ink formation and printing process do not affect the crystal structure of the BaTiO 3 NPs after the UV curing during the 3D printing. The ferroelectric hysteresis loop was measured for the BaTiO 3 NP/EAA/AUD samples with a BaTiO 3 NP concentration from 5 to 15 wt % as shown in Figure 2 b. The saturation polarization and the remanent polarization both increased with the increasing concentration of the BaTiO 3 NP under the same electric field, indicating the improved piezoelectricity at a higher content of piezoelectric oxides. As the BaTiO 3 NPs could block the UV light, the ink with more than 15 wt % NPs could not be successfully cured during 3D printing with the same photoinitiator content and the same power of the UV light in the current DLP printer. Thus, 15 wt % BaTiO 3 NP was chosen to be the optimum concentration for the printable ink in piezoelectric devices. The piezoelectric coefficient of the printed piezoelectric films with 15 wt % BaTiO 3 NP that is poled under 25 V/μm electric field is measured to be 0.78 pC/N. This value is smaller than the other reported bulk or microscale BaTiO 3 materials [ 15 , 16 , 17 ]. This may be due to the damping effect of the 3D printed elastomer matrix, which reduces the force transferred onto the nanoparticle in the d 33 measurement process. Another possible reason is that the nanoscale BaTiO 3 particles tend to recover from the aligned dipole moment after poling due to the lack of domain wall restrictions from neighbouring grains [ 18 , 19 ]. To examine the morphology, SEM imaging was conducted on the sample and the electrode. Figure 3 a is the surface of the printed BaTiO 3 NP/EAA/AUD composite. There are obvious buckling structures throughout the sample surface, which contributed to the large elasticity that was verified later on. During mixing, the nanoparticles with oligomers and air bubbles are easily formed and difficult to remove, which makes the piezoelectric material susceptible to discharge during poling. With the DLP printing system, this problem can be solved using the slider to distribute a thin layer of ink between the window and the substrate and remove the air bubbles from each printing layer as shown in the schematic ( Figure S1 in the Supplementary Materials ) and the pore-free cross-section image ( Figure 3 b). To fabricate a stretchable nanogenerator, stretchable electrodes are required. A highly stretchable and printable conductive ink formulation was reported by Wang et al. based on ethylene–vinyl acetate (EVA), silver flakes, and eutectic gallium indium (liquid metal) particles [ 20 ]. This ink system provides stable electric conductivity under as train as high as 1000% and excellent cycling stability. The formulation with 50 wt % silver flakes and 50% EVA solution was adapted for the stretchable electrode in this work. The size of silver flakes ranges from 1 to 5 micrometers. Liquid metal was not added because the liquid metal NPs tend to aggregate under repeated compression, which may cause short circuits. Without the addition of liquid metal, the conductor is also highly conductive (R/R 0 = 8) under a 200% strain with conductivity before stretching 11,240 S cm −1 [ 20 ]. The morphology of the stretchable conductor is shown in Figure 3 c. Silver flakes are distributed inside the EVA matrix, forming connections among each other to conduct the electricity. The interface between the electrode and the piezoelectric composite is shown in Figure S2 of the Supplementary Materials , indicating that the two materials are in good contact and without interdiffusion. Tensile testing was done on the printed BaTiO 3 NP/EAA/AUD composite sample with 15% BaTiO 3 NP content as shown in Figure 4 a. The maximum tensile strain for the printed film sample is 434%, and Young’s modulus is 0.83 MPa, in which both the elongation and Young’s modulus increase compared with the previously reported 3D printed EAA/AUD samples without BaTiO 3 nanoparticles [ 13 ], due to the reinforcement of the nanoparticle filler. The maximum strain is higher than all the stretchable piezoelectric materials reported to date [ 2 , 3 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 ]. This higher elasticity could be mainly attributed to the EAA/AUD matrix. According to previous research, the high stretchability of the polymer matrix is due to the presence of hydrogen bonds between hard domains of the AUD. Upon mechanical loading, the breakage of hydrogen bonds dissipates energy and therefore results in the high stretchability of the elastomer system [ 13 ]. A cyclic tensile testing was done on the BaTiO 3 NP/EAA/AUD as shown in Figure S3 of the Supplementary Materials , which further proves the repeatable elastic behaviour of the composite. The energy harvesting behaviour is evaluated by pressing on the printed stretchable BaTiO 3 NP/EAA/AUD sample with a 1 cm 2 effective area. As shown in Figure 4 b,c, the voltage output under 60 N force under 5 Hz frequency is 0.29 V, and the current density is around 0.20 μA/cm 2 . The power density can be calculated by Ohm’s law, which is 57 nW/cm 2 . The values are comparable to other piezoelectric nanogenerators made of lead zirconate titanate (PZT), zinc oxide (ZnO), and polyvinylidene difluoride (PVDF)-based piezoelectric materials [ 31 , 32 , 33 , 34 , 35 , 36 , 37 ], and this power density can be used for powering various types of low energy consumption sensors such as temperature sensors, electrochemical sensors, biosignals sensors, and other mechanical sensors [ 38 , 39 , 40 , 41 , 42 ]. Under 100% strain stretching, the voltage and current output slightly increased to around 0.38 V and 0.23 μA/cm 2 at the same conditions of 60 N 5 Hz, respectively ( Figures S4 and S5 in the Supplementary Materials ). This enhancement may be due to the larger stress subjected onto the BaTiO 3 NPs under the reduced thickness upon stretching. The output voltage will not be affected by reducing the number of BaTiO 3 NPs since the device can be considered as a matrix of parallelly connected units. When it is stretched, the number of units connected in parallel is reduced since the force is exerted in a fixed 1 cm 2 area. In piezoelectric material, the current is induced from the external circuit by the potential generated from the internal polarization. Therefore, considering the area of the sample under compression force, the current density will not decrease with the decrease in the number of BaTiO 3 NPs. A similar increase in power output has been earlier found in a stretchable hydrogel-based triboelectric nanogenerator [ 43 ]. Furthermore, the output voltage of the device under compressive force ranging from 30 to 60 N was measured at an un-strained state ( Figure 5 a,b). The output voltage linearly increases with the input force with a sensitivity of 59.8 mV/N, which is similar to some reported BaTiO 3 and PZT self-powered sensors for body motion sensing [ 3 , 44 ]. After stretching, the sensitivity slightly increases ( Figure 5 b, Figures S6 and S7 in the Supplementary Materials ), which may be due to the less damping from the elastomers to transfer more force onto the BaTiO 3 nanoparticles. The linear increase in output voltage versus input force shows a good sensing behavior, which shows the potential for this stretchable piezoelectric nanogenerator to act as a wearable sensor for body motion sensing and gait sensing. To demonstrate the gait sensing potential, the piezoelectric nanogenerator was mounted on the bottom of the human heel to sense the foot stepping motion and frequency as a self-powered physiological sensor ( Figure S8 in the Supplementary Materials )."
} | 2,957 |
28834420 | PMC5609275 | pmc | 659 | {
"abstract": "Summary Biological production of hydrogen is poised to become a significant player in the future energy mix. This review highlights recent advances and bottlenecks in various approaches to biohydrogen processes, often in concert with management of organic wastes or waste CO \n 2 . Some key bottlenecks are highlighted in terms of the overall energy balance of the process and highlighting the need for economic and environmental life cycle analyses with regard also to socio‐economic and geographical issues.",
"introduction": "Introduction Hydrogen provides a CO 2 ‐free sustainable alternative to fossil fuels. A pioneering global initiative, the ‘Hydrogen Council’, comprising thirteen leading energy, transport and related industries, intends to increase investment in the hydrogen and fuel cell sectors (currently €1.4 Bn year −1 ) to stimulate hydrogen as a key part of the future energy mix via new policies and schemes (Anon, 2017 ). Hydrogen is currently obtained mainly by steam reforming of hydrocarbons, releasing multiple greenhouse gas emissions (DOE, 2013 ). Hence, new H 2 production methods are required such as biological production (bio‐H 2 ; Dincer and Acar, 2015 ). H biotechnologies are maturing towards benchmarking against established clean energy from electrolysis of water, solar photovoltaics and wind farms. Biohydrogen can be made fermentatively from wastes, providing a simultaneous method of organic waste management (Chang et al ., 2011 ). This short review highlights progress and bottlenecks of bio‐H 2 towards a sustainable development goal to ensure access to affordable, reliable, sustainable and modern energy for all. Biohydrogen has been reviewed in comparison with other hydrogen production processes (Nikolaidis and Poullikkas, 2017 ). Biohydrogen embraces any H 2 production involving biological material (Mohan and Pandey, 2013 ). The energy source can be solar or can come from conversion of fixed carbon substrates (or both, in various combinations). An approach to CO 2 ‐end of pipe treatment (e.g. from flue gas from fossil fuel combustion or carbon‐neutral fermentation of biomass) is to grow algae on waste CO 2 . Algal biohydrogen production is well‐described, but O 2 from algal oxygenic photosynthesis inhibits the hydrogenase that makes H 2 . A key study (Kubas et al ., 2017 ) will open the way to developing O 2 ‐resistant hydrogenase. Emerging technology uses cyanobacteria (blue‐green algae) that make H 2 via hydrogenase and also nitrogenase; their O 2 ‐sensitivity is managed by temporal separation of photosynthetic O 2 evolution and nitrogenase action, and by compartmentalization into microanaerobic heterocysts (Tiwari and Pandey, 2012 ). Despite a note that cyanobacterial biohydrogen is probably uneconomic (Singh et al ., 2016 ), an environmental life cycle analysis (LCA) has shown for the first time that cyanobacterial bio‐H 2 has the potential to be a competitor to desulfurized natural gas; the associated environmental impact of producing and extracting each gas, including use in a solid oxide fuel cell, was calculated and simulated respectively using the LCA software simapro (Archer et al ., 2017 ). This research used published data from a raceway growth system (James et al ., 2009 ). However, at latitudes above ~40°N, the generally low incident solar energy makes stand‐alone photobiological H 2 systems seasonal and uneconomic without some form of process intensification. Boosting light delivery (e.g. LEDs, quantum dots) can be effective, but these may risk photopigment saturation and inhibition; this approach may be questionable economically and would be best addressed by a life cycle analysis. In sunny countries, light is plentiful, but in this case, ‘delivering cold’ is needed to extend crop product and food life; cooling is energy‐demanding and a global challenge (Strahan, 2017 ). Another challenge is organic materials from agri‐food and municipal wastes, which must be managed to avoid landfilling which yields methane, a potent greenhouse gas. Current practices use anaerobic digestion (AD) with biogas – methane used for power. We review some options for combining waste treatments with bio‐H 2 technology as possibly the best approach to tackling effectively these dual socio‐economic problems; stand‐alone biohydrogen is possibly uneconomic, but this awaits a life cycle analysis, currently in progress."
} | 1,103 |
34277591 | PMC8280782 | pmc | 660 | {
"abstract": "In context of the global climate change, microalgae processes are gaining momentum as a biotechnological tool for direct fixation and valorization of greenhouse gases. Algae have the metabolic capacity to photosynthetically convert CO 2 into high value products, such as food additives, under economic boundary conditions. High cost, commercial flat panel gas-lift bioreactors for microalgae cultivation at laboratory scale provide either small volumes or no sterile operation, which limits academic research. This brief report presents initial data for a new type of sterile operating flat panel gas-lift bioreactor with a unique asymmetrical U-shape. It utilizes automatable process control technologies that adhere to industrial standards to enhance data reproducibility and aid industrial scale up. The practicability was demonstrated using a Chlorella sorokiniana cultivation, which showed the typical growth behavior. Due to the sophisticated implemented control engineering technology, pivotal parameters as pH and temperature can be determined within a range of ±0.1 units, which was confirmed experimentally. The new flat panel gas-lift photobioreactor presented in this brief report fills the technology gap at laboratory scale with an autoclavable volume of 7.2 L. Moreover, it is easy to rebuild by means of the hereby provided blueprint, while exhibiting a six-fold cost reduction compared to commercially available flat panel photobioreactors.",
"introduction": "Introduction Excessive anthropogenic CO 2 emissions are the main cause of the greenhouse gas effect that results in progressive global warming. It represents a focal challenge for mankind to preserve our planets ecosystem. To alleviate climate change effects there is a demand for innovative technologies that can actively fix and reduce atmospheric CO 2 ( Goli et al., 2016 ; Vo Hoang Nhat et al., 2018 ). Microalgae are a taxonomically diverse group of photosynthetic microorganisms which, due to their superior photosynthetic efficiency, can convert CO 2 up to ten times faster into biomass than any vascular, terrestrial plant ( Goli et al., 2016 ; Adeniyi et al., 2018 ). As microalgae can be cultivated on marginal lands using waste- or salt water, their mass cultivation does not impact agricultural activity or generate land use change associated with negative impacts on biodiversity. Since microalgae biomass contains value adding products, these whole cell biocatalysts represent a yet underexploited toolbox for the efficient CO 2 fixation and value creation under economic and ecologic boundary conditions. Depending on the selected algae strain and its respective cultivations conditions, microalgae biomass contains varying concentrations of multiple value adding products such as sugars, lipids, proteins, pigments, vitamins, or extracellular polymers ( Chisti, 2007 ; Griesbeck and Kirchmayr, 2012 ; Woortman et al., 2020a ). Examples for microalgae biomass valorization encompass diverse applications, such as high folate (vitamin B9) concentrations accumulating in the rapidly growing green algae Picochlorum sp. but also aviation biofuel production using oleaginous microalgae like Microchloropsis salina ( Chew et al., 2017 ; Woortman et al., 2020a , b ). These aforementioned oil producing microorganisms can accumulate up to 60% of their dry biomass as lipids with substantially higher oil yields per hectar than currently favored feedstock ( Chisti, 2007 ). Therefore, oleaginous microalgae may help to counteract the second central challenge of humanity as well, which is to meet the rising energy demands coupled with diminishing fossil fuel supplies ( UN Energy Statistics, 2018 ). Extracted oil from algae biomass can serve as raw material for the production of biodiesel through transesterification. During the conversion process, triacylglycerides are separated to fatty acid methyl esters and glycerol under the presence of a short chain alcohol, most commonly methanol, and a catalyst. Subsequently, the oil free biomass waste stream can be processed into further energy carriers like biobutanol by means of fermentation with Clostridia sp. bacteria or biogas through gasification or pyrolization ( Potts et al., 2012 ). In order to use microalgae cultivation as a technology platform, however, entails identification of suitable microalgae strains and their selection for a respective application, which subsequently has to be integrated into an industrial process by iterative scale up in standardized open- or closed system photobioreactors. While open pond cultivation approaches allow for scale up and production, disadvantages involve contamination and low control over experimental parameters. Enclosed systems on the other hand offer lower cultivation volumes but higher levels of control over experimental conditions instead, which renders them more suitable for process development ( Chisti, 2007 ). Various designs of closed photobioreactors have been established: tubular ( Molina et al., 2001 ; Hulatt and Thomas, 2011 ), flat panel ( Hu et al., 1998 ; Lakaniemi et al., 2012 ), column ( Béchet et al., 2013 ), and biofilm ( Blanken et al., 2014 ) systems. Flat-panel configurations are among the best designs due to a high surface area to volume ratio in addition to a short light path length while consuming less energy than tubular systems ( Zou et al., 2000 ; Jorquera et al., 2010 ). The initial processes of algae identification and laboratory process characterization are commonly conducted by academic groups, which due to limited funds often use different self-constructed cultivation vessels and conditions, which conventionally do not allow sterile operation with medium to high vessel volumes of >5 L. Axenic cultivation is of great importance to eliminate external influences and contaminations, potentially allowing for the development of Food and Drugs Administration agency regulation compliant medical and food products ( Chew et al., 2017 ). Furthermore, sterile operation conditions are vital for research when examining co-cultivation of specific microorganisms or single strain experiments ( Yong et al., 2021 ). Examination of the literature data used for composing Table 1 indicates, that with increasing culture volume photobioreactor assemblies either are not sterilizable or require extensive technical effort to provide a sterile operation mode. Additionally, custom-developed closed photobioreactors documented in literature are mostly square shaped, potentially promoting the emergence of disadvantageous dead zones, where the medium admixture is flawed ( Sierra et al., 2008 ; Tamburic et al., 2011 ; Gifuni et al., 2018 ). In that regard, our photobioreactor concept fills a technology and volume gap, as it can be disassembled easily and each component is sterilizable by itself as well as the entire system. Moreover, the sterilization process is using standard equipment such as a laboratory autoclave. TABLE 1 Comparison of the photobioreactor presented in this brief report with selected flat panel photobioreactors in the laboratory scale as described in literature regarding their volume and sterilizability. Volume [L] Sterilizable Reactor form References 0.3 +* Erected flat rectangle Gifuni et al., 2018 0.4 + Square with rounded edges Hulatt et al., 2017 0.9 + Square with inner comb shape Wang et al., 2019 >1 N.D. V-shaped converging bottom Jiang et al., 2020 3.0 + Square with rounded edges Nyberg et al., 2015 3.4 − Rectangle shaped Barbosa et al., 2005 7.0 − − Lakaniemi et al., 2012 7.0 − Stacked floors Cheah et al., 2020 7.2 + Square with asymmetrical U-shape This brief report 15.5 − Rectangle shaped Yang et al., 2014 18.5 − Cube shaped Choi et al., 2013 30.0 − Square shaped Singh Khichi et al., 2018 48.0 +** Stacked pipes Li et al., 2015 86.0 − Rectangle shaped Mhatre et al., 2018 (*) chemical sterilization, (**) ozone sterilization. Conversely, commercially available photobioreactors at laboratory scale with the desired operation features are cost intensive and are only available in small volume capacities of less than 4 L 12,3 . A summative comparison of the new photobioreactor’s specifications presented in this brief report with established flat panel bioreactors systems operated as gas-lift reactors can be found in Table 2 . TABLE 2 Comparison of the photobioreactor presented in this brief report with selected commercially available photobioreactors operated as gas-lift reactors in the laboratory scale. Feature Photobioreactor of this report Subitec GmbH 4 Infors GmbH 1 Biostream International BV 3 PSI spol. s r.o. 2 Prevention of dead zones + + + + + Autoclavability + − + + + Working volume > 7L + − − − − Cost-effective <10,000 € + + − − − 1 Labfors 5 Phototrophic 1.9 L TV, Infors GmbH, Sulzemoos, Germany. 2 Photobioreactors FMT 150, PSI spol. s r.o., Drásov, Czechia. 3 Biostream International BV, Doetinchem, Netherlands. 4 Flat Panel Airlift Photobioreaktor, Subitec GmbH, Stuttgart, Germany. To circumvent issues of current commercial systems, a new design of flat panel gas-lift photobioreactor was developed in this brief report. Its unique asymmetrical U-shape prevents formation of dead zones, which are induced by microalgae biomass precipitation at high cell densities commonly observed in the late exponential growth phase. Supplementary Figure 1 serves to illustrate, how the unique geometry of the new flat panel air lift photobioreactor allows for enhanced mixing of the cultivation medium and distinguishes itself from commercially available equivalents. In addition, the new flat panel photobioreactor vessel can be sterilized in a conventional laboratory autoclave, therefore enabling axenic microalgae cultivation, allowing for algae process optimization targeted at the food- and pharmaceutical sector. Due to its straightforward architecture, this 7.2 L photobioreactor is easily constructed and exhibits a six-fold cost reduction compared to commercially available systems. Owing to the implemented automated, industrially standardized systems control technology, it is not only possible to manipulate different growth parameters, such as temperature, pH value and light intensity, but also to control light and temperature cycles, thereby enabling complex climate simulations, which are key for iterative process scaling toward outdoor cultivation deployment at industrial scales.",
"discussion": "Discussion The newly designed, autoclavable flat panel gas-lift photobioreactor presented in this brief report offers a low-cost alternative for laboratory applications at the 7.2 L scale. Moreover, the presented photobioreactor utilizes a set of industrially standardized and automatable process control technologies, which will enhance process development and data reproducibility as well as integrity, which provides for accelerated iterative scale up procedures toward industrial deployment. In order to accelerate iterative scale up of microalgae processes for academic research, all of the necessary data to rebuild this photobioreactor was deliberately fully disclosed in this brief report. The unique asymmetrical U-shape design provides enhanced mixing of the cultivation medium, which prevents formation of dead zones at high cell density biomass production scenarios required for an economically viable technology platform. Additionally, the sophisticated control engineering of this bioreactor offers an alternative open platform for microalgae cultivation at the intermediate laboratory scale of 7.2 L. In this flat panel based photobioreactor concept, illumination is performed from both sides, providing a maximum of light energy intake for efficient production of biomass and intracellular products, such as lipids or carotenoid pigments. Due to its modular construction, illumination from one individual side is feasible, allowing the determination of light dependent growth parameters, which are critical for iterative process and reactor scale up ( Pfaffinger et al., 2016 , 2019 ; Koller et al., 2017 ). The former is only applicable in early growth phases, since a relatively high glass panel thickness of 40 mm was chosen. Functionality of the novel photobioreactor was tested and verified by performing batch cultivation experiments with C. sorokiniana , a rapidly growing microalgae that exhibited conventional microbial growth stages in these experiments. In addition, an accuracy of temperature and pH control comparable to other reactors was demonstrated. Through utilization of glucose containing medium, highest standards where applied during the sterilization assay. The sterility of the reactor enables the axenic cultivation of single algae strains including recombinant microalgae, rendering it appropriate for research and development in the field of food and pharmacology. Further sophistication of this photobioreactor system could involve incorporation of circadian light cycles and temperature shifts or the implementation of exhaust gas measurements for industrial relevant process development."
} | 3,263 |
37809477 | PMC10559818 | pmc | 662 | {
"abstract": "We investigate how simple physical interactions can generate remarkable diversity in the life history of social agents using data of social wasps, yielding complex scalable task partitioning. We built and analyzed a computational model to investigate how diverse task allocation patterns found in nature can emerge from the same behavioral blueprint. Self-organizing mechanisms of interwoven behavioral feedback loops, task-dependent time delays and simple material flows between interacting individuals yield an emergent homeostatic self-regulation while keeping the global colony performance scalable. Task allocation mechanisms based on implicitly honest signaling via material flows are not only very robust but are also highly evolvable due to their simplicity and reliability. We find that task partitioning has evolved to be scalable and adaptable to life history traits, such as expected colony size or temporal bottlenecks in the available workforce or materials. By tuning solely the total number of agents and a social connectivity-related parameter in the model, our simulations yield the whole range of emergent patterns in task allocation and task fidelity akin to observed field data. Our model suggests that the material exchange (“common stomach mechanism”) found in many paper wasps provides a common functional “core” across these genera, which not only provides self-regulation of the colony, but also provides a scalable mechanism allowing natural selection to yield complex social integration in larger colonies over the course of their evolutionary trajectory.",
"introduction": "1 Introduction Complex adaptive systems are characterized as such when patterns emerge on upper system levels due to both localized interactions and selection processes acting at lower system levels [ 1 ]. Such systems are characterized by nested hierarchical networks of components which are organized as interconnected modules [ 2 , 3 ] and exhibit nonlinear dynamics with multiple possible outcomes [ 4 ]. Complex adaptive systems are composed of multiple components and can exhibit an increase in global performance with increasing system size [ 5 , 6 ]. However, instead of observing an unbounded performance increase (Gustafson's law), it is more common to observe diminishing returns in performance (Amdahl's law and Gunther's universal scalability law) with increasing system size [ 7 ]. In technical systems, parallelizing the execution of tasks is constrained by limitations and bottlenecks. These limitations can be structural, for example if serial operations are required to perform well. Limitations can also be logistical, for example, when parallel processing is limited by the costs of coordination [ 8 , 9 ]. In biological systems, a similar phenomenon is known as ‘Michener's paradox’ or as the ‘Per Capita Paradox’ [ 10 ], where the per-capita productivity in social insect colonies does not increase with colony size. Rationalizations for Michener's paradox include biological (kin selection, phylogeny) and statistical (sampling error, central limit theorem) explanations [ [10] , [11] , [12] ]. We hypothesize here that both physical constraints and fundamental interaction mechanisms can explain the ‘Per Capita Paradox’ as well as the observed emergence of task fidelity (individual agents repeat a given task instead of switching between tasks frequently) in insect societies. The evolution of emergent biological complexity requires the development of information-processing systems at multiple scales [ 13 ]. This information processing fosters the evolution of short-term response mechanisms, enabling better adaptations in dynamic environments. In turn, the structure and the constraints of the surrounding environment mutually link to the evolving biological system, allowing for co-evolution. Both biological and cultural evolution operate under many constraints of different magnitude [ 14 ]. Some of these constraints stem from the fundamental rules of network size, which in turn strongly limit the repertoire of potential designs [ 15 ]. We adopt the view that the fundamental principles of a system are shaped by natural selection as a result of the interference between the constraints of the system and the demands of the tasks that are required to survive and to reproduce [ 16 ]. We use insect societies to demonstrate how complexification can emerge as a function of constraints and how this process can be beneficial for the evolution of the complex adaptive system as a whole. Insect societies can be described as analogues to ‘liquid brains’, ‘active matter’, or swarms of interactive agents [ 17 , 18 ]. We emphasize two important characteristics of cognitive living networks: first, their agents move in space and thus their interactions are local and dynamic, constraining the system in a non-trivial way; second, the attractors of these systems are not always based on the strength of interactions (connection weights) but are also dependent on population abundances [ 17 ]. The fundamental mechanisms of colony functioning must work over a large scale of population sizes, given that many colonies start with one or few individual(s) and can reach a size of thousands or millions of animals within a few years [ 19 ]. The evolution and function of such simple robust mechanisms have been one of the central questions in understanding complex societies [ 12 , 20 ]. Division of labor is one of the most important and common phenomena in social animal groups [ [21] , [22] , [23] ]. Insect colonies need to perform a set of tasks, such as constructing nests or gathering food and materials, under changing environmental conditions. Thus, each colony must be capable of sensing the internal state of the colony utilizing distributed information processing. Ants, for example, use pheromones to mark trails that lead to food sources. This pheromone network influences the behavior of the individuals, which in turn can reinforce or adapt the network. This broadcast-type signaling amongst individuals via pheromones allows the whole colony to access global system states, ultimately yielding a plethora of emerging collective behaviors [ 17 , 24 ]. In a previous study [ 20 ], we analyzed the task allocation mechanisms of honeybees [ 25 , 26 ], ants [ 27 ], and wasps [ 12 , [28] , [29] , [30] , [31] , [32] , [33] , [34] ]. We discovered a common core mechanism akin to an integral control regulation, which had previously only been identified in biological systems at the physical and cellular-molecular levels [ [35] , [36] , [37] ], but not above the individual organization level [ 20 ]. The common core element of this closed loop regulation, which we call the “common stomach”, ensures that foraging for a crucial substance and the consequent use of it in a colony is regulated by the very substance itself. This is achieved via a network of behavioral feedback loops, which are facilitated by local physical worker-to-worker interactions. This regulatory core exhibits a high level of redundancy, ensuring resilience, reliability, and robustness within the self-regulation. Besides its role as an information center, the common stomach also provides a material buffer against fluctuations in substance flows and also acts as a natural integrator, reducing the impact of system noise [ 37 ]. Our study demonstrates the spontaneous emergence of parallel processing and an increase of task fidelity on the macroscopic system level as a function of increasing colony size. This phenomenon arises exclusively from the microscopic mechanics of self-regulation via the common stomach. An increasing level of task fidelity will increase the degree of system redundancy at the subunit level (parallel processing), which according to the reliability theory, is more efficient than redundancy at the system level [ 38 ]. We propose that a scalable integral control mechanism has evolved for ensuring the operability of the system across different environmental and colony size fluctuations. In addition, we show that although task fidelity increases with colony size, the average per-capita efficiency stays similar across all tested colony sizes, which agrees with Michener's Paradox [ 10 ]. This suggests that several important biological traits may have the same mechanistic explanation at their core, which is captured by our model.",
"discussion": "4 Discussion Previous work [ 20 ] showed that the core of task regulation in wasp, bee, and ant societies can be considered akin to a form of integral control regulation, which has previously been described for cellular and subcellular levels in biology. Here we present an explanation on how such a mechanism can explain both the commonality in the construction behavior of social wasps and the differences in life history aspects of different wasp genera. Mechanisms of interaction and material exchange among workers are shaped by natural selection into emergent complex pulp and water sharing systems that are scalable and robust. We stress that scalability is one of the most important properties of/in a complex adaptive system, especially in a “swarm system” or a “hive mind” as social insect colonies are often called. Our simulation experiments showed not only interesting scalability properties of the system, which are often the effect of emergent strategy changes in the modeled social interaction systems, but we observed also remarkable emergent counterstrategies in our perturbation experiments. The specifically modeled mechanisms of water and pulp sharing showed to be crucial elements that allow the wasps to switch to alternative behaviors in response to alterations of work force sizes or environmental conditions. The model we present here strictly follows the parsimony principle: the explanation should be as simple as possible and as complex as necessary. In this model the mechanisms we modeled are material transfers and linear modulations of work engagement or abandonment. No other mechanism were assumed and implemented into our model, such as learning processes, genetic dispositions, physiological or morphological agent heterogeneity, individual preferences or other specific neuroethological predispositions, as they are often expressed by heterogenous sigmoid stimulus-response curves in agent-based models. One of our key findings here is, that we do not require to hypothesize about colony-size-dependent adaptations of proximate mechanisms, as our model can predict the observed ultimate effects without such a layer of additional complexity. In this paper we emphasize that the common stomach mechanism, due to its scalable nature, will result in different colony level outcomes with differing workforce sizes. Ultimately, a complex task partitioning process is the emergent consequence of more interactions, as the workforce size increases, using the same basic construction mechanism. Thus, our findings show that complexity can arise from simplicity in wasp societies. Our agent-based common-stomach model differs significantly from ODE models on the common stomach found in literature. We avoid here the implicit assumptions of mean-field ODE modeling, such as assuming an optimal mixture of agents or allowing infinite interaction distances. Such assumptions are implicitly made in previous common stomach models on where the task allocation of several species (ants, honeybees, wasps) was described [ [25] , [26] , [27] , 29 , 30 ], or when crosscutting comparisons between these insect societies have been made [ 20 ]. In contrast to these studies, we focus here on investigating task allocation of paper wasps at different levels of eusocial complexity concerning their colony integration mechanisms. Our bottom-up model presented here implements only a small set of basic proximate mechanisms and physical interaction principles, which then yield ultimate colony-level effects as emergent property in simulation runs, Such emergent properties were found to be flexible task division and allocation ( Fig. 2 ) and colony-level resilience against environmental perturbations ( Fig. 5 ). Ultimately, our model allows investigating how simple physical interactions, that are modeled purely locally, can yield different performance metrics of the overall colony, allowing to scale the colony concerning the number or modulating specific aspects of interaction. Our model keeps the modeled workplaces (water collection sites, pulp collection sites, construction site) very abstract and ad-libitum concerning the ability of the wasps to perform their work there. In their natural habitat, there will always be water and wood to be found for the wasps. However, there are intrinsic limitation of work: Water foraging requires empty crop space, pulp foraging requires a full water load and nest building requires pulp loads. These limitations are modeled explicitly, but the resources for construction assumed to be available ad libitum. In addition, environmental fluctuations in the resource accessibility can be modeled via modulating the corresponding length of a given collection duration. For example, in the dry season the water collectors may need to travel farther for water. To account for sudden change of resource availability we implemented additional influx of water or pulp, to analyze the sensitivity of these systems ( Fig. 5 ). Jensen et al. [ 41 ] concluded that the optimization of material flows appears to be a universal feature of biological transport systems. They emphasized that most transport systems are also subject to a set of limiting constraints. In insect societies Michener's (Per-Capita) Paradox [ 10 ] describes a diminishing return of per capita productivity with increasing colony size. The reason for the emergence of this paradox in these societies remains elusive, even though it also seems to emerge from other facets of social life [ 42 ]. This paradox is similar to Amdahl's Law [ 43 ] found in technical computation systems, which shows that technical bottlenecks impede unbounded increases of computation rate with increasing system size. Our study demonstrates how local and simple physical interactions (material exchanges) can give rise to such emerging patterns with similar key properties. Although Michener's Paradox was neglected at first and criticized thereafter [ 44 ], it is a significant biological observation of biological scalability that is highly relevant beyond biology today. Michener discovered, years ahead of Amdahl, a general law of scalability that became crucial in understanding complex systems such as economy and complex computational systems. Michener's views were crucial to recognize that insect societies are complex adaptive systems and they operate under the same physical rules as other similar systems. Our simple agent-based model shows that strategies with no or little interaction between wasps will not scale up well with increasing work force size. Polistes wasps do not share materials and work as jack-of-all-trades individuals in a sequential manner. Task partitioning will consequently not emerge in their colonies. These wasps do not spend much time in social interactions and are working as independent individuals on nest construction even if the colony size increases. Our model shows that this strategy is efficient for construction, but it is costly when foraging. In nature, the cost of this strategy is mostly incurred by the potential loss of the foragers, which is largely due to predation. Polistes colonies evolved to remain small and to grow slowly. This ensures small nests and moderates total foraging traffic, which makes them less detectable by predators. Vespula wasps have a life history very similar to Polistes wasps. For example, their nest is initially built by a single queen in a sequential manner. However, Vespula colonies will develop a rudimentary form of task partitioning (emergence of water foragers) when their colony's size increases. These wasps spend more time with social interaction while transferring water, so their building efficiency is a bit smaller than of the Polistes , but their foraging is more economical. In natural colonies, the smaller number of specialized foragers will find the resources faster and will thus be exposed for less time to predation risks. Characteristically, Vespula colonies grow larger than Polistes colonies, even if their per-capita efficiency, similar to Polistes , does not change with colony size. Metapolybia and Polybia wasps have similar life histories, but Polybia wasps generally build larger colonies. These wasps share pulp and water via worker-to-worker interactions, and they thus spend considerable time at the interaction platform region of the nest. Our model predicts that this strategy would be very detrimental to their efficiency in small colony sizes. In nature, these colonies never reside in small colonies, as they reproduce by swarming rather than via a single queen like Polistes and Vespula . Our model shows that the construction strategy using pulp and water sharing is clearly superior for the building and the colony efficiency except at very small colony sizes. This high efficiency of large colonies is achieved by a greatly reduced foraging activity than is observed in other wasp genera. Task partitioning is well defined in these genera and the workers’ task fidelity is high. These colonies operate with small numbers of specialist foragers that can carry out foraging quickly, therefore decreasing predation risk. This emergent phenomenon of task fidelity does not cause rigidity of the colony-level behavior. Our perturbation analyses show that in fact these societies are very resilient against environmental changes, and that this resilience property is essentially achieved via dynamic rearrangements of the workforce governed by common stomach regulation [ 20 ]. Our model focuses only on the construction behavior of some important wasp genera, this way it emphasizes some key elements and neglects less important factors from their complex behavior. Our model is based on previous empirical and modeling work that we did on these wasp societies [12.28, 29, 30–34], thus this model cannot be directly generalized to other groups of organisms. However, the core mechanism of this model stems from our previous studies [ [25] , [26] , [27] ] where we showed that the task regulation of bees, ants and wasps relies on an integral control mechanism that we named “common stomach”. The common stomach is a functioning colony-level regulatory system, based on explicitly defined local worker-to-worker interactions. This is a significant finding, as our previous ODE models have demonstrated that the common stomach's key principles are generalizable also to other organism groups [ 20 ]. Diverse biological complexity can emerge from simple physical laws and from mechanisms such as transport processes and material flows, paired with physical constraints and restrictions to local interactions [ 45 ]. Natural selection acts upon the involved proximate mechanisms and a rich diversity of ultimately colony-level regulation systems can emerge. For example, task fidelity has been presented in previous literature to be a consequence of a lack of learning abilities, or that it requires specific genetic makeup, and such specialists are able to be “activated” or “deactivated” while staying in their task's role [ 46 ]. In contrast to these hypotheses, we here show that task fidelity can be an emergent property purely resulting from the simple construction behavior of wasp societies, and that it can be explained via physical material-exchange interactions without any specific worker precondition or specific properties required [ 20 , 28 , 32 , 47 , 48 ]. Task fidelity and task partitioning will automatically emerge in larger colonies, a testimony to the power of scalable general mechanisms in explaining the diversity of life histories."
} | 4,997 |
35424607 | PMC8981771 | pmc | 663 | {
"abstract": "Thermal interface materials (TIMs) are one of the efficacious ways to alleviate the heat accumulation problem of microelectronics devices. However, conventional TIMs based on polydimethylsiloxane (PDMS) always suffer from mechanical damage, leading to shortened service life or loss of thermal conductivity. In this work, we fabricated a high-thermal conductivity and fast self-healable Al 2 O 3 @siloxane composite by hydrosilylation reaction. The siloxane matrix consisted of thermosetting silicone rubber matrix (SR) and heat reversibility matrix (SCNR); the SR was synthesized via hydrosilylation between silicon hydrogen bond and vinyl, the SCNR was fabricated by thermal-curing between amino and carboxyl functionalized PDMS. Different sized spherical Al 2 O 3 fillers were introduced into the SR/SCNR matrix system to construct the Al 2 O 3 @SR/SCNR composites. By adjusting the ratio of SR/SCNR, the obtained composites can achieve flexibility, self-healing and high filling simultaneously. It is notable that the self-healing efficiency of the composite is high, up to 95.6% within 3 minutes with 6.7 wt% mass ratio of SCNR/SR; these fast self-healing behaviors benefit from the assistance of thermal diffusion by 3D heat conduction pathways on the rearrangement of the dynamic cross-linked network. The resultant composites also exhibited the optimal thermal conductivity of 5.85 W mK −1 . This work provides a novel approach for constructing longer service life and high thermal conductivity multifunctional TIM based PDMS.",
"conclusion": "Conclusions In this work, we successfully fabricated fast self-healing and high thermally conductive Al 2 O 3 @SR/SCNR composites based on carboxyl-amine dynamic reversible bonds. We used the dynamic and reversible cross-linking between the carboxyl and the amino functional PDMS to synthesize the SCNR with heat reversibility, then the different sizes spherical Al 2 O 3 fillers and SCNR particles were introduced into the SR system to construct the Al 2 O 3 @SR/SCNR composites via in situ polymerization. The 20Al 2 O 3 @SR/SCNR-1/0.075 composites with 6.97 wt% SCNR content of matrix exhibited the thermal conductivity of 5.89 W mK −1 and the self-healing efficiency up to 95.6% simultaneously. The dynamic ionic reaction between the carboxyl and the amino functionalized PDMS was considered to be the source of high-efficiency self-healing performance. The introduction of alumina improved the thermal conductivity of the composites and accelerated the speed of heat transfer. At the same time, the heat conduction pathway provided fast and uniform heat for the self-healing quickly of the damaged area, and 20Al 2 O 3 @SR/SCNR composite behaved excellent heat dissipation performance in the field of microelectronics. The brief strategy in this work provides a practical method that can be used as a reference for other self-healing thermally conductive polymer thermal interface materials.",
"introduction": "Introduction With the increasing refinement of the internal layout of microelectronic devices, heat is further accumulated. The electronics industry urgently desires to develop low-cost thermal interface materials (TIMs) which are suitable for irregular gaps to dissipate heat quickly. 1–3 Polydimethylsiloxane (PDMS) is widely used in the matrix of a new generation of TIMs due to its chemical stability and weather resistance. 4,5 However, the thermal conductivity of PDMS is only about 0.2 W mK −1 , so it is often necessary to fill with thermal conductive fillers to increase the thermal conductivity. Metal materials such as Ag, 6,7 Cu 8 and carbon materials including graphite, 9,10 graphene, 11,12 carbon nanotubes 13,14 with high electrical conductivity have been reported as fillers to increase the thermal conductivity of the polymer-composites. Ceramic materials such as AlN, 15 BN, 16,17 Al 2 O 3 18,19 have not only good thermal conductivity, but also good electrical insulation and excellent thermal stability. Among them, although the thermal conductivity of Al 2 O 3 is not so high (∼30 W mK −1 ), it is widely sourced, inexpensive and suitable for bulk filling. It is applied in currently commercialized TIMs (such as thermally conductive gaskets and thermally conductive gels) to improve the thermal conductivity. However, the conventional thermosetting Al 2 O 3 /PDMS composites are easily damaged to occur cracks during use, and thermal conductivity or other functions of the materials will be reduced after damage. The concept of self-healing comes from the healing process of natural organisms. Materials can maintain their properties after self-healing. If these TIMs are integrated with self-healing ability to automatically repair damages inflicted on them, it would avail long-term usage as well as enhanced reliability and durability. The intrinsic self-healing ability of materials is mainly achieved by dynamic reversible bonds, such as disulfide bonds, 20,21 Diels–Alder reactions, 22,23 amide bonds, 24,25 metal ligand-coordination, 26,27 borate bonds, 28,29 borate ester bond, 30,31 electrostatic interactions 32,33 and so on. Considering the practical application of thermal interface, the introduction of thermally triggered reversible dynamic bond is a suitable scheme. Whereas, among reported self-healing materials with thermally triggered reversible dynamic bond, Cao et al. fabricated a mechanically strong and self-healing rubbers via dynamic Fe 3+ –pyridine coordination bonds, the sample shown self-healing efficiency of 87%. 34 Whereas, such metal coordination bonds are not suitable for occasions where insulation is required. Zhao et al. prepared the composite with thermal conductivity of 0.8 W mK −1 using DA adduct cross-linked silicone polymer as matrix and BN as fillers. 35 Therefore, achieving high thermal conductivity and fast self-healing efficiency is still a challenge. In particular, the materials should remain flexible. It is an ingenious strategy to meet the demand introducing of the thermal-triggered carboxyl-amine dynamic bond into the siloxane matrix. Liu et al. fabricated a self-healing silicone elastomer via thermo-curing to develop dynamic crosslinked ionic bonds between carboxyl and amide functionalized PDMS, which showed a self-healing efficiency of 97% after 2 h, at 100 °C. 33 Sun et al. prepared a supramolecular network by two components assembled by hydrogen and ionic bonds, which a self-healing efficiency of 115% in tensile strength and almost 100% in actuated strain at a given electric field can be achieved after self-healing at 80 °C for 5 h. 32 Inspired by the above works, we developed fast self-healing and high thermally conductive Al 2 O 3 @SR/SCNR composite based on carboxyl-amine dynamic reversible crosslinking. To meet the soft and high filling requirements of TIMs, we constructed a thermosetting silicone rubber matrix (SR) with low cross-linking degree via controlled hydrosilylation. The dynamic cross-linking network of SCNR formed between carboxyl and amide functionalized PDMS ensured self-healing and reprocessing capabilities. The spherical Al 2 O 3 , as a thermal conductive candidate, could make great contribution not only to fast heat pathway but also to the thermal-triggered self-healing properties. In addition, we adjusted the ratio of covalent crosslinks and thermo-reversible crosslinks in matrix to obtain both the fast self-healing performance and high thermal conductivity. It is reasonably expected that these multifunctional Al 2 O 3 @SR/SCNR composites would be used as preferable thermal interface materials which behave the ability to automatically repair damages and avail long-term usage as well as enhanced reliability and durability.",
"discussion": "Results and discussion The method and strategy of fabrication Al 2 O 3 @SR/SCNR composites SCNR elastomer was prepared by blending PDMS-COOH and PDMS-NH 2 , as shown in Fig. S1. † PDMS-COOH was prepared by hydrosilylation of side chain hydrogen polydimethylsiloxane (PDMS-H, H wt% = 1.2%) and MMA. Then, the synthesized PDMS-COOH was mixed with PDMS-NH 2 via solution blending. After removing the solvent and thermal curing, SCNR can be successfully obtained. The ionic bonds can be formed between PDMS-COO − and PDMS-NH 3 + , which are generated by the deprotonation of –COOH on the side chain of PDMS-COOH and the protonation of –NH 2 at the end of PDMS-NH 2 . In addition, hydrogen bonds can also be formed between carbonyl (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) and amino (NH 2 ). Carbonyl as a receptor, amine as a donor hydrogen bond and ion bond can be used as physical crosslinking points to form an elastomer network. Al 2 O 3 @SR/SCNR composite was fabricated with blending and crosslinking SCNR particles, Al 2 O 3 , PDMS-Vi and PDMS-H. It is found that the free amino group destructed the reaction between PDMS-Vi and PDMS-H, so the PDMS-COOH in this system is slightly excessive and SCNR needs to be cured in advance. Under the pression, the added components crosslinked together uniformly to obtain Al 2 O 3 @SR/SCNR composites. It can be seen in Fig. 2(b) that there are no larger particles (such as SCNR) in the Al 2 O 3 @SR composites than the spherical alumina (20–70 μm). The low-crosslinking network we designed in this work can accommodate 10 times more alumina than itself. The amount of alumina added reaches 90.9 wt%, and a large amount of spherical alumina builds thermal conduction paths in all directions. And the successful preparation of Al 2 O 3 @SR/SCNR composites implied that SCNR is not recombined during the preparation. SR and SCNR networks are physically interspersed with each other. Moreover, SCNR particles were uniformly fused and inserted into SR network after continuous friction of Al 2 O 3 . Meanwhile, the permanently crosslinked SR network also acted as a skeleton to accommodate Al 2 O 3 and SCNR networks. Fig. 2 SEM images of spherical alumina (a); Al 2 O 3 @SR/SCNR composites (b); thermal path of Al 2 O 3 @SR/SCNR composites (c). Characterization and dynamic reversibility of the SCNR composites SCNR is obtained via thermal-crosslinking of PDMS-COOH and PDMS-NH 2 (Fig. S1 † ), which the dynamic ionic bond formed between PDMS-COO − and PDMS-NH 3 + . FTIR spectra for PDMS-H, PDMS-MMA, PDMS-COOH were present in Fig. S3(a). † PDMS-H has an obvious Si–H absorption at 2159 cm −1 . For PDMS-MMA, the absorption of Si–H basically disappeared, and a strong absorption of saturated ester carbonyl C O appeared at 1743 cm −1 , indicating that a hydrosilylation reaction had occurred between MMA and PDMS-H. The main change in the FTIR spectrum of the product PDMS-COOH after the hydrolysis reaction was the disappearance of the absorption of the ester carbonyl C O at 1743 cm −1 and the appearance of the absorption of the carboxyl group C O at 1708 cm −1 . This confirmed that the ester functional group grafted onto the siloxane chain is successful hydrolysis to be a carboxyl functional group. \n 1 HNMR results (Fig. S2 † ) of PDMS-MMA 1 HNMR (400 MHz, CDCl 3 ) δ 7.15 (dd, J = 9.3, 5.5 Hz, 1H), 7.09–7.02 (m, 1H), 3.54 (s, 3H), 2.25 (s, 1H), 1.10 (d, J = 7.0) and PDMS-COOH ( 1 H NMR (400 MHz, DMSO-d 6 ) δ 11.91 (s, 1H), 1.74–1.61 (m, 1H), 0.68–0.44 (m, 1H)) also agree well with the FTIR data. To illustrate the self-healing property of SCNR, the samples were cut into two different colour separate pieces with a razor blade ( Fig. 3(a) ), and two half-plates were brought back into contact and then healed by a thermal treatment at 90 °C, 1 h. The healed SCNR could be stretched and bent, indicating that the cured sample had good uniformity, and the chopped SCNR ( Fig. 3(b) ) could be reprocessed after hot pressing at 2 kPa, 90 °C for 15 min. Fig. 3 Photographs of the SCNR: (a) self-healing of damaged SCNR samples; (b) reprocessing of SCNR samples. \n In situ FTIR was performed to study dynamic reversibility and the hydrogen bonds and ionic bonds network of SCNR. The infrared spectrum from 1500 cm −1 to 1800 cm −1 was presented in Fig. 4 . Herein, in SCNR, hydrogen bonds were mainly formed among amide bond (N–H) and carbonyl (C O) groups. As the peak representing stretching vibration of (C O) groups moved from 1738 cm −1 to the high frequency region to 1749 cm −1 , accompanied with the shifting peak of in-plane bending vibration of N–H at 1540 cm −1 gradually to the low frequency region and disappeared. It could be inferred that hydrogen bonds interact with each other and the dynamic change of hydrogen bonds in the SCNR. 36 And the N–H peak disappears and the C O peak weakens, which indicates that with the increase of temperature, the dynamic interaction decreases until dissociation, resulting in an increase in the amount of free carboxyl, thereby promoting the reaction of –COOH with –NH 2 to form carboxyl-amine dynamic bonds. Fig. 4 \n In situ FTIR spectra of SCNR at different temperature. X-ray diffraction (XRD) patterns of SCNR before and after self-healed exhibited in Fig. S3(b). † The SCNR sample had no obvious crystallization peak, while the sample after heat treatment had a significant peak at about 45°, which may be explained by the fact that some hydrogen bonds are converted into ionic bonds after heat treatment, resulting in higher crystallinity, which is accordant with the results from in situ FTIR."
} | 3,457 |
30680121 | PMC6341978 | pmc | 664 | {
"abstract": "Abstract Reef‐building corals may harbor genetically distinct lineages of endosymbiotic dinoflagellates in the genus Symbiodinium , which have been shown to affect important colony properties, including growth rates and resilience against environmental stress. However, the molecular processes underlying these differences are not well understood. In this study, we used whole transcriptome sequencing (RNA‐seq) to assess gene expression differences between 27 samples of the coral Montipora capitata predominantly hosting two different Symbiodinium types in clades C and D. The samples were further characterized by their origin from two field sites on Hawai‘i Island with contrasting environmental conditions. We found that transcriptome‐wide gene expression profiles clearly separated by field site first, and symbiont clade second. With 273 differentially expressed genes (DEGs, 1.3% of all host transcripts), symbiont clade had a measurable effect on host gene expression, but the effect of field site proved almost an order of magnitude higher (1,957 DEGs, 9.6%). According to SNP analysis, we found moderate evidence for host genetic differentiation between field sites ( F \n ST = 0.046) and among corals harboring alternative symbiont clades ( F \n ST = 0.036), suggesting that site‐related gene expression differences are likely due to a combination of local adaptation and acclimatization to environmental factors. The correlation between host gene expression and symbiont clade may be due to several factors, including host genotype or microhabitat selecting for alternative clades, host physiology responding to different symbionts, or direct modulation of host gene expression by Symbiodinium . However, the magnitude of these effects at the level of transcription was unexpectedly small considering the contribution of symbiont type to holobiont phenotype.",
"conclusion": "5 CONCLUSIONS Unraveling the complexity of host–symbiont interactions at the molecular level remains an important goal in the life sciences. In this study, we investigated transcriptome‐wide differences in host gene expression associated with two dominant types of Symbiodinium clade C and D in the coral M. capitata , which is of particular interest considering some clade D symbionts may convey higher resistance to elevated water temperatures and are globally increasing in prevalence. In our study population, we found that association with predominantly clade C or clade D types was correlated with measurable changes in host gene expression profiles, a moderate number of DEGs, and changes in the activity of coexpressed gene networks. However, considering the differences in physiology and performance between coral colonies harboring different types, in particular of clades C and D, the clade‐dependent effects appeared unexpectedly limited in scope. This was especially evident in comparison with the much more pronounced effects of environmental factors, which we explored through the inclusion of different field sites. Notably, we did not uncover large‐scale transcriptional changes pertaining to the immune or stress response, although increased activity of these systems might be expected from hosts confronted with less optimally coadapted symbionts of clade D Symbiodinium . This raises the question whether the types investigated here indeed differ significantly with regards to their role along the mutualism‐parasitism spectrum, as has been hypothesized for clades C and D (Stat & Gates, 2011 ). The clade‐based effects we observed, however, were consistent with previous studies, and involved changes in metabolic processes, as well as more subtle variation within several other systems, including the regulation of transcription and translation, cell signaling, and membrane trafficking. Differences in metabolic pathways and transport may be linked to differences in symbiont–host nutrient transfer (Loram et al., 2007 ), but to what extent transcriptional changes translate into metabolic changes remains unclear until further experiments (compare Matthews et al., 2017 ). Our experimental setup was also not designed to determine to what extent the correlation between host gene expression and symbiont type is caused by the host genomic background, the host responding to symbiont type, or more intriguingly, cross‐kingdom transcriptional modulation of the host by Symbiodinium . A promising avenue of research to this end would be to investigate the role of symbiont regulatory RNAs. Such an approach could further illuminate where the relationships between clade D symbionts and their hosts are located on the spectrum from mutualism to parasitism, which has important long‐term implications for coral reef health and conservation (Stat & Gates, 2011 ). By providing a candidate list of clade‐associated genes in M. capitata , we hope to motivate the further characterization of these genes, which all too often remain only incompletely annotated—after all, identifying and comparing unknown candidate genes that are found consistently in nonmodel organisms has the potential to make unexpected contributions to our understanding of host–symbiont interactions.",
"introduction": "1 INTRODUCTION Symbiosis, the persistent, intimate association between species (Bouchard, 2009 ), shapes all aspects of the biosphere from molecular structure to ecosystem function. Most, if not all, animals engage in complex interactions with microbial symbionts and depend on them for nutrition, defense, and development (Chaston & Goodrich‐Blair, 2010 ). Indeed, the emerging view of animals as holobionts—ecosystems consisting of a multicellular host and associated microbial communities—has sparked new insights into genome evolution, development, metabolism, organismal health, and ecology (McFall‐Ngai et al., 2013 ). Scleractinian corals are one of the most prominent examples of animal–microbe symbiosis and hold particular ecological, economic, and cultural importance (Cesar & van Beukering, 2004 ). In the coral holobiont, dinoflagellate algae of the genus Symbiodinium provide organic nutrients, oxygen, and energy derived from photosynthetic activity to the coral host in return for inorganic nutrients, carbon dioxide, and protection (Stat, Carter, & Hoegh‐Guldberg, 2006 ). A host of genetically and ecologically distinct species (“types”) of Symbiodinium belonging to several divergent lineages (“clades”) have been identified as endosymbionts in scleractinian corals (Baker, 2003 ; LaJeunesse et al., 2018 ; Weber & Medina, 2012 ). While in many coral species colonies associate with a single dominant type of Symbiodinium , others are able to host multiple types simultaneously, or in temporal succession (Baker, 2003 ; Berkelmans & van Oppen, 2006 ). The composition of the Symbiodinium community may influence important host and holobiont properties, including colony growth rates (Jones & Berkelmans, 2010 ; Mieog et al., 2009 ), nutrient transfer (Loram, Trapido‐Rosenthal, & Douglas, 2007 ), local adaptation (e.g., with regard to light exposure and depth, Iglesias‐Prieto, Beltrán, LaJeunesse, Reyes‐Bonilla, & Thomé, 2004 ), fitness, and tolerance of environmental stress (Berkelmans & van Oppen, 2006 ; Iglesias‐Prieto et al., 2004 ; Mieog et al., 2009 ). How Symbiodinium clades impact holobiont resilience has received particular attention. Some types, most notably in clade D, may provide the holobiont with increased tolerance of elevated temperatures and can reduce susceptibility to coral bleaching (Berkelmans & van Oppen, 2006 ; Stat & Gates, 2011 ), the breakdown of the coral– Symbiodinium symbiosis which is emblematic of the current coral health crisis. Genome‐wide analyses of host gene expression hold great promise to reveal the genetic underpinnings of these functional differences. Host transcriptional changes have been studied in nonpathogenic symbioses of bacteria and vertebrates (Rawls, Samuel, & Gordon, 2004 ), insects (Wilson et al., 2006 ), cephalopods (Chun et al., 2008 ; Wier et al., 2010 ), deep‐sea mussels (Boutet et al., 2011 ), and polychaetes (Nyholm, Robidart, & Girguis, 2008 ). In cnidarians, most transcriptomic studies have focused on the establishment of symbiosis in coral larvae (Mohamed et al., 2016 ; Schnitzler & Weis, 2010 ; Voolstra et al., 2009 ), recently settled coral polyps (O'Rourke et al., 2017 ), and adult symbiotic and nonsymbiotic sea anemones (Lehnert et al., 2014 ; Rodriguez‐Lanetty, Phillips, & Weis, 2006 ). In most of these studies, significant changes in host gene expression suggested an active role of the host in the establishment and regulation of symbiosis. However, only few studies have investigated the transcriptional response of cnidarian hosts to different symbiont types. The first to show a correlation between Symbiodinium type and host transcription, DeSalvo et al. ( 2010 ) reported expression changes in genes involved in protein metabolism in the coral Montastraea faveolata . This observation was more recently corroborated in Acropora millepora by Barfield, Aglyamova, Bay, & Matz ( 2018 ), who demonstrated that both the symbiont type and the native reef environment result in host gene expression differences in a common garden experiment. Modulation of host physiology, including metabolism, stress response, and immune system, was also found in the sea anemone Exaiptasia pallida during colonization by an opportunistic Symbiodinium type, with respect to the dominant type (Matthews et al., 2017 ). The objective of the present study was to expand upon this incipient knowledge of how different coral– Symbiodinium partnerships affect gene expression, and to shed more light on how interactions between symbiont partners are regulated at the genetic level in Montipora capitata , one of the most common coral species in Hawai‘i. Using whole transcriptome sequencing (RNA‐seq), we examined the transcriptional profiles of M. capitata colonies in field populations harboring Symbiodinium types representing clades C and D. We compared these profiles to the transcriptomic states corresponding to two other variables, field site and disease status. The two field sites, while both located on Hawai‘i Island, differed markedly with respect to underwater topography and climatic conditions, especially rainfall and temperature. Organisms can respond to environmental changes through local adaptation (allele frequency changes between generations due to natural selection) and acclimatization (nongenetic, usually short‐term phenotypic responses), both of which can manifest in gene expression changes. Environmentally sensitive genes may comprise a significant portion of the transcriptome (5%–15%; Santo et al., 2013 ; Zhou, Campbell, Stone, Mackay, & Anholt, 2012 ). The third variable we included was the presence or absence of Growth Anomaly (GA), a widespread coral disease characterized by abnormally enlarged skeletal growth. In a previous study partially based on the same dataset, we found that GA has a surprisingly limited impact on gene expression (Frazier, Helmkampf, Bellinger, Geib, & Takabayashi, 2017 ). Comparing the effects of field site and disease status to clade‐related gene expression differences in the present study allowed us to evaluate the magnitude of these effects. Further, we measured genetic differentiation between field sites and among corals harboring alternative clades directly from the RNA‐seq data to assess the role of host genotype. We also provided a list of host genes whose expression was correlated with Symbiodinium clade, and employed coexpression network analysis to illuminate the connectivity and function of genes that might be responsive to the presence of different symbiont clades.",
"discussion": "4 DISCUSSION 4.1 Symbiodinium composition Our primary objective was to evaluate the effects of Symbiodinium types from different clades on host gene expression in M. capitata , and identify groups of genes potentially involved in the maintenance and regulation of different symbiotic relationships. We also considered two other variables—samples were collected from sites on two opposite sides of Hawai‘i Island, and from colonies and tissues unequally affected by GA, a widespread coral disease we focused on in a previous publication (Frazier et al., 2017 ). At an average sequencing depth of 25 million paired‐end reads per holobiont, we were able to identify the dominant Symbiodinium type (C31 or D1a) in each colony from ITS2 transcript abundance, with only a minority of samples hosting both types (Table 1 ). At the individual and population level, this Symbiodinium community composition is congruent with previous studies conducted at Wai‘ōpae (Burns, Gregg, & Takabayashi, 2013 ), and in a M. capitata population on nearby O‘ahu Island (Stat et al., 2011 ). Reads not perfectly matching one of the references typically diverged only very slightly, suggesting they did not originate from additional Symbiodinium types, but rather represent sequencing errors, intragenomic variants, or alternative genotypes (compare Cunning, Gates, & Edmunds, 2017 ). However, determining the Symbiodinium community composition directly from RNA‐seq data comes with a few limitations. Short reads (75 bp in the present study) may not always span diagnostic nucleotide positions, and low coverage of symbiont transcripts can mean that rare variants (including of alternative clades) remain undetected. Further limitations stem from the nature of the ITS2 marker itself, which is not sufficiently variable to distinguish between all types, and is prone to intragenomic variation, both of which complicates the assessment of Symbiodinium diversity (Stat et al., 2011 ). However, while we may have underestimated the contribution of alternative types by our approach, it is evident the study population is dominated by symbionts from two different clades (C and D), and types C31 and D1a specifically. The impact of rare and likely transient background symbionts on host functions is contentious (Lee et al., 2016 ; Ziegler, Eguíluz, Duarte, & Voolstra, 2018 ), and differences in holobiont properties have been widely recognized in corals hosting symbionts from divergent clades. Accepting some uncertainty in the symbiont composition at the type level, we therefore focus on transcriptional effects associated with clade C and D symbionts, mainly represented by types C31 and D1a. 4.2 Host gene expression profiles Across 27 RNA‐seq datasets from 18 colonies, we detected distinctly different transcriptional responses to the three variables described above. According to overall transcriptional profiles (Figure 1 a) and DGE analysis (Figure 1 b), the field site had the largest effect on host gene expression, followed by symbiont clade, and then GA disease status. With a clear profile separation according to MDS ordination, and 1,957 differentially expressed genes (9.6% of all 20,461 host transcripts), the effect of field site proved to be almost an order of magnitude larger than that of symbiont clade, which corresponded to 273 DEGs (1.3%) in the default dataset. Because the two sites at Waiʻōpae and Kīholo are separated by more than 200 km along the coast (125 km linear distance), we considered whether the samples represented two different populations with limited gene flow between them, resulting in gene expression differences due to contrasting genomic backgrounds. We calculated a genome‐wide F \n ST of 0.046 (confidence interval 0.040–0.052) from synonymous SNPs, indicating moderate genetic differentiation between Waiʻōpae and Kīholo. F \n ST values in this range have also been reported in other coral species with comparable life histories across even larger distances, using neutral and non‐neutral markers (Ayre & Hughes, 2004 ; Polato, Concepcion, Toonen, & Baums, 2010 ). Gene expression differences between sites might therefore be driven by underlying genetic variation in transcriptional regulators due to local adaptation. They may also represent a phenotypic response (acclimatization), for example, through epigenetic effects or polyphenism. Conceivably, the environment at the two field sites differed sufficiently to induce both adaptation and acclimatization, especially in a sessile organism. For instance, as a typical fringing reef on Hawai‘i Island's arid Western side, Kīholo is characterized by stable conditions with constant sea water circulation and little rainfall. In contrast, Waiʻōpae on the Eastern side received higher rainfall and groundwater influx and was defined by its complex underwater topography with partially submerged tide pools of different size and connectivity to the open ocean. Waiʻōpae was therefore subject to much higher fluctuations in temperature and salinity (Wiegner et al., 2016 ) than Kīholo, necessitating different short‐ and long‐term physiological responses. The magnitude of gene expression differences between sites was comparable to those in a similar study, which reported about 3,000 DEGs between two sites (Barfield et al., 2018 ) in Australian A. millepora . These genes showed enrichment in macromolecule biosynthesis and RNA processing functions, suggesting some overlap with our GO term analysis, which highlighted cyclic compound biosynthesis and transcription as potentially important functions in response to different environments (Table 2 ). Finally, we found no clustering of samples by GA disease status (Figure 1 a), and only 11 DEGs between healthy and diseased colonies (Figure 1 b). This number was likely deflated by superseding effects of site and symbiont type, which may have led to a decrease in the signal‐to‐noise ratio. Indeed, the number of DEGs went up to 78 after removing all samples from Kīholo. This magnitude of DGE, and the pattern of colony‐level response to GA, was consistent with the results of our previous study based on the same samples from Waiʻōpae (Frazier et al., 2017 ). 4.3 Site‐controlled clade effects We found little to no overlap in DEGs obtained from comparisons based on site, clade, and GA disease (Figure 2 ), and in coexpression modules correlated with these variables (Supporting Information Table S2 ). This suggests that gene expression responses to site, clade, and GA disease are largely independent of each other, despite the overrepresentation of clade D in Kīholo and healthy samples, respectively. Because the sampling site was found to be a significant factor and clade C was underrepresented at Kīholo (Figure 1 a), we ran a more focused analysis with only Waiʻōpae samples to obtain a more accurate picture of gene expression changes associated with symbiont clade. This decreased the number of clade‐dependent DEGs to 154, slightly more than half of which were also found in the 273 DEGs observed in the default dataset (Figure 2 ). While moderate in comparison with transcriptome size (0.8%), this number suggests a correlation between the composition of the Symbiodinium community and the transcriptional state of the host. To our knowledge, differences in host gene expression associated with different Symbiodinium types in established symbioses have been reported by only a handful of studies before (Barfield et al., 2018 ; DeSalvo et al., 2010 ; Matthews et al., 2017 ). These observations are intriguing as they open possibilities to examine host–symbiont genetic interactions that maintain and regulate different symbiotic relationships. However, unraveling the nature and cause of correlation between host gene expression and symbiont type is challenging. For instance, host genotype might influence which symbiont types are able to establish themselves in the host, so that both host gene expression and symbiont type are ultimately correlated with host genotype. To test this, we estimated genome‐wide genetic differentiation between hosts harboring clade C and clade D. With F \n ST = 0.036 (CI 0.031–0.041) across all samples and 0.058 (CI 0.049–0.065) for Waiʻōpae only, we found limited but potential evidence for systematic host genotypic differences that might drive the distribution of symbiont types. Studying field samples, we were not able to take advantage of host clones to control for such effects of host genotype, as were DeSalvo et al. ( 2010 \n ) and Matthews et al. ( 2017 ). However, the Symbiodinium community composition has been shown to be uncorrelated with host genotype in a geographically nearby population of M. capitata (Stat et al., 2011 , based on ATP synthase subunit beta intron—but see Quigley, Willis, & Bay, 2017 ) and is likely influenced by microhabitat conditions, in particular light exposure (Innis, Cunning, Ritson‐Williams, & Gates, 2018 ). Considering that symbiotic interactions require a complex dialog between partners, it is also conceivable—even likely—that Symbiodinium symbionts and their coral hosts have evolved the ability to modulate each other's transcriptional activity. For instance, hosts seem to be able to influence physiological responses in their symbionts (Parkinson, Banaszak, Altman, LaJeunesse, & Baums, 2015 ). Conversely, DeSalvo et al. ( 2010 \n ) speculated that host physiology may respond to different symbiont types, or that Symbiodinium may be able to explicitly influence host gene expression and physiology. Gene regulation following transfer of small RNAs between pathogen and host, or between mutualistic partners, has for instance been documented in plant–microbe interactions (Lelandais‐Brière, Moreau, Hartmann, & Crespi, 2016 ; Weiberg et al., 2013 ). Finally, the gene expression patterns observed here may also be a reflection of differences in host physiology associated with symbiont type. Coral colonies predominantly harboring Symbiodinium types of clade C or D can differ markedly with regard to several important properties. Colonies dominated by clade D generally grow more slowly (Jones & Berkelmans, 2010 ; Little, Oppen, & Willis, 2004 ; Mieog et al., 2009 ), which may be linked to differences in nutrient transfer between symbiont and host (Loram et al., 2007 ; Matthews et al., 2017 ). Symbiodinium types in clades C and D may also be adapted to different light regimes, and influence the vertical distribution pattern of their hosts (Iglesias‐Prieto et al., 2004 ). Most notably, some clade D symbionts confer increased resistance to elevated temperatures to the host, and reduce its susceptibility to coral bleaching (Berkelmans & van Oppen, 2006 ; Mieog et al., 2009 ; but see e.g., Hume et al., 2015 ). As the abundance of clade D symbionts has been increasing globally in habitats impacted by bleaching and across a wide range of host species, some are suspected to be opportunists capable of outcompeting and replacing optimally coadapted symbionts (e.g., in clade C) in health‐compromised corals (Stat & Gates, 2011 ). However, our experimental design does not allow us to distinguish between the host merely responding to differences in nutrient transfer, photosynthetic function, and other Symbiodinium ‐associated traits, or direct modulation of gene expression by the symbionts. Parallel sequencing of small RNA and conventional RNA‐seq, together with advances in host–symbiont transcript separation, or experiments inducing symbiont shuffling under controlled conditions, could allow addressing this issue in future studies. 4.4 Key genes and coexpression networks The magnitude of gene expression differences (e.g., low hundreds of genes) associated with variation in symbiont composition was similar to that reported from several other cnidarian host species. For instance, Matthews et al. ( 2017 ) observed roughly 100 DEGs characterizing sea anemones with normal and opportunistic Symbiodinium clades using a comparable RNA‐seq approach. Studies using microarrays reported similar absolute numbers, for example, DeSalvo et al. ( 2010 ) in the coral M. faveolata , and Boutet et al. ( 2011 ) in associations of deep‐sea mussel and chemoautotrophic bacteria. However, because these microarrays represented only a small fraction of the host transcriptome, the effect size is not fully comparable to the present study. In contrast, Barfield et al. ( 2018 ) found more than 3,000 DEGs between corals associated with different symbionts, possibly because of this study's extra level of sensitivity due to its common garden design. Among the DEGs we identified in coral hosts harboring different Symbiodinium clades, genes involved in DNA metabolism and integration, protein phosphorylation, and transport processes were significantly over‐represented (Table 2 ). Individual genes with the most significant expression differences between clades included genes encoding two transcription elongation factors, a putative but fragmentary Toll‐like receptor (TLR), collagen and ankyrin repeat‐containing proteins (the former possibly involved in Wnt signaling), and a pore‐forming cytotoxin homologous to Delta actitoxin Aas1a. We detected few transcriptional differences in genes associated with the immune system and stress response (e.g., in lectins, additional TLRs, antioxidants)—differences which might be expected when comparing an optimally coadapted (clade C) and a more opportunistic symbiont (clade D), as observed by Matthews et al. ( 2017 ) in sea anemones. Similarly, genes involved in translation and protein folding/degradation seemed to play less of a role than reported by DeSalvo et al. ( 2010 ). However, while DGE remains the most straightforward approach to study transcriptional responses, not all biologically relevant genes may vary significantly in expression under different conditions. Most genes are organized into networks governing cellular processes or pathways, and the activity of such a network may change by the accumulation of gene expression differences that are not detectable individually. We therefore complemented our DGE analysis with the more systemic approach of reconstructing gene networks from coexpression data (i.e., correlation in gene expression). In the default and the Waiʻōpae dataset, we identified 17 and 25 modules of coexpressed genes, respectively, which may correspond to biological pathways or shared cellular functions. Modules whose eigengenes (first principal component) were significantly correlated with site (default dataset only) and Symbiodinium clade (both datasets; Supporting Information Table S2 ; Figure 3 ), suggested that these gene networks play a role in biological processes affected by environmental factors and symbiont clade. Consistent with the DEGs between sites, two of the site‐dependent modules appeared to revolve around the production of macromolecules and organic cyclic compounds, as well as RNA metabolism. Focusing on clade effects in the smaller dataset, two of the modules significantly correlated with clade (M6 and M17) did not produce meaningful GO term results likely due to their small size. However, some information could be gleaned from the 10 most highly connected genes in each module, which represent hub gene candidates that may influence the activity of a disproportionate number of genes in the network (Supporting Information Tables S3 and S4 ). In module M6, hub gene candidates included serine proteases, as well as genes potentially involved in DNA repair and cytoskeleton regulation (e.g., homologs of Mediator of DNA damage checkpoint protein 1, spectrins). Potential hub genes in module M17 comprised a homolog of S‐adenosylmethionine synthase, several aminoacyl tRNA synthases, and ABC transporters, suggesting a module function in translation, membrane trafficking, and possibly immune response. The two larger of the significant modules, M2 and M3, showed enrichment in GO terms mostly pertaining to metabolic processes, as well as to a lesser extent, signal transduction and regulation of transcription (Table 3 ). Both modules featured a functionally diverse cast of potential hub genes. For instance, M2 hub gene candidates may be involved in endocytosis (Adaptor protein complex 2), RNA processing (e.g., La protein, Polypyrimidine tract binding protein 1), and signal transduction (e.g., Cyclic nucleotide phosphodiesterase, and a putative G protein), among others. In the case of M3, potential hub genes seemed to mostly play a role in signaling pathways and cytoskeleton regulation, comprising several kinases, putative kinase receptors, and others (e.g., STK, Ras and Zap70 homologs, Ankyrin 2, to focus on the most reliably annotated ones). Notably, module M2 also featured a disproportionate number of highly connected genes, since the majority of these genes in the entire transcriptome (top 100 by kTotal) fell into this group (while kTotal is correlated with the number of genes per module, M1 contained more genes in total but only a minority of highly connected genes). By combining DGE and coexpression network analyses, we discovered that none of the top overall connected genes overlapped with the 154 DEGs between clades, suggesting that clade‐associated genes are not central regulators in the host transcriptome. On the other hand, most DEGs (70%) were found within the modules significantly correlated with clade. Coexpression network analyses therefore pointed to more extensive, subtle changes in biological processes between hosts colonized by different clades than was observable directly through DGE analysis. In summary, these processes appeared to revolve around metabolic changes, transcription and translation, and cell signaling, which are consistent with previous observations in similar systems (DeSalvo et al., 2010 ; Matthews et al., 2017 ). However, it is important to note that the usefulness of homology‐based gene annotation and GO term analysis remains limited in nonmodel organisms, and typically rely on a small number of genes. In the present study, we were only able to annotate a quarter of all transcribed host genes with Swiss‐Prot homologs, and to identify Pfam domains (on which GO annotations were based) in three‐quarters. Sequence homology may not be a reliable predictor of functional homology across greater phylogenetic distances, and genes that cannot be annotated may represent taxonomically restricted or highly divergent genes with important functional roles (Tautz & Domazet‐Lošo, 2011 ; Wissler, Gadau, Simola, Helmkampf, & Bornberg‐Bauer, 2013 ). These concerns appeared to be particularly pronounced in the site‐dependent DEGs. Despite their high number, these genes were characterized by only four main GO terms, three of which were highly similar despite efforts to reduce term redundancy (Table 2 ). In addition, only four out of the 10 most significant DEGs could be functionally annotated based on sequence homology, indicating the rest may be taxonomically restricted. Such genes without homology to known genes have been linked to species‐specific adaptations in response to changing environments (Colbourne et al., 2011 ; Voolstra et al., 2011 ). It is thus conceivable taxonomically restricted genes are over‐represented in DGEs from two sites characterized by different environmental conditions, although this possibility requires further analysis."
} | 7,820 |
38241353 | PMC10830016 | pmc | 666 | {
"abstract": "Until recently, most ecological network analyses investigating the effects of species’ declines and extinctions have focused on a single type of interaction (e.g. feeding). In nature, however, diverse interactions co-occur, each of them forming a layer of a ‘multilayer’ network. Data including information on multiple interaction types has recently started to emerge, giving us the opportunity to have a first glance at possible commonalities in the structure of these networks. We studied the structural features of 44 tripartite ecological networks from the literature, each composed of two layers of interactions (e.g. herbivory and pollination), and investigated their robustness to species losses. Considering two interactions simultaneously, we found that the robustness of the whole community is a combination of the robustness of the two ecological networks composing it. The way in which the layers of interactions are connected to each other affects the interdependence of their robustness. In many networks, this interdependence is low, suggesting that restoration efforts would not automatically propagate through the whole community. Our results highlight the importance of considering multiple interactions simultaneously to better gauge the robustness of ecological communities to species loss and to more reliably identify key species that are important for the persistence of ecological communities.",
"introduction": "Introduction The rate of decline of many species populations is accelerating [ 1 ], and species extinctions are seriously threatening the functioning of ecological communities worldwide. Understanding how species interact and how this affects the robustness of ecological communities to species loss is essential to anticipate the consequences of biodiversity losses and extinction cascades as well as to design protection and restoration plans. The study of ecological networks—where species are represented by nodes and the ecological interactions by links between these nodes—have contributed significantly to the understanding of how ecological interactions are structured and have unveiled important relationships between network structure and their robustness to species loss [ 2 – 5 ]. However, while the ecological network literature has long been dominated by studies of networks containing a single interaction type, it has become increasingly clear that species in nature are connected by a myriad of interaction types simultaneously and that considering networks which include this diversity of interaction types could greatly improve our knowledge of the structure and dynamics of ecological communities [ 6 – 13 ]. A number of previous studies have investigated the effect of including multiple interaction types on the functioning of ecological communities, especially on their stability [ 14 – 20 ]. Yet the vast majority of these studies have so far remained theoretical. With the publication of the first multi-interaction empirical networks, we begin to know more about their structure [ 6 , 10 , 13 , 21 – 27 ], and how this structure affects their persistence [ 6 , 10 ] and robustness [ 21 , 24 , 27 , 28 ]. In particular, studies on multi-interaction networks have provided new insights on whether the inclusion of several interactions can significantly alter their robustness to species loss [ 24 ] and how extinctions propagate through such networks [ 21 ]. However, in spite of these pioneering studies, there is currently no consensus about the structure of multi-interaction networks and its consequences for the robustness of ecological communities, in part due to the lack of data sets, whose amount has only recently started to increase. A key question, of relevance given the current biodiversity crisis, is how robustness varies across network types, and what we can learn from including multiple interactions simultaneously. With this in mind, we gathered ecological networks with multiple interaction types currently available in the literature. More specifically, we focused on tripartite networks because they were the most abundant in the literature, allowing us to compare a wide variety of ecological systems. Tripartite ecological networks are composed of two interaction layers (e.g. pollination and herbivory), each of the bipartite kind [ 29 ]. They therefore contain three different species sets (e.g. plant, pollinator and herbivore guilds in a pollination-herbivory network), one of which is shared between the two interaction layers (e.g. plant species can interact with both pollinators and herbivores in a pollination-herbivory network). We call the set of nodes that can have interactions in both interaction layers the shared set , and the subset of nodes in the shared set that have interactions in both interaction layers the connector nodes (see Fig 1A and 1B ). 10.1371/journal.pcbi.1011770.g001 Fig 1 Tripartite networks, robustness and interdependence. A) An Herbivory(h)—Pollination(p) tripartite network, where plants (P) are the shared set of species. B) An Herbivory(h)—Parasitism(pa) tripartite network, where herbivores (H) are the shared set of species. Link colours represent the two interaction layers, and node colours the three sets of species. Connector nodes in the shared set of species are highlighted in black. C) Extinction curve showing the fraction of surviving animal species as a function of plant loss for a given plant extinction sequence in network A. The robustness to plant loss, R , is the area under the curve. Extinction protocol: plants (green nodes) are progressively removed from the community in the prescribed order, their corresponding links are erased (colored in red) and animal species are declared extinct (colored in red) whenever they lose all their feeding links. D) Pairwise correlation in the robustness of the two animal sets—interdependence, I —resulting from 3.000 simulations of random sequential loss of plant taxa in network A. Our data set consists of 44 tripartite networks from 6 different studies, in which the interaction layers include mutualistic (pollination, seed-dispersal and ant-mutualism) and antagonistic (herbivory and parasitism) interactions (see Methods ). To identify possible generalities across interaction types as well as singularities specific to a given interaction type, we divided the networks in three types according to the signs of the interactions involved: mutualism-mutualism (MM) if both interactions were positive, antagonism-antagonism (AA) if both interactions were negative, and mutualism-antagonism (MA) if one interaction was positive and the other negative, given that interaction type can determine network architecture through the underlying biological constraints [ 30 ]. Using this data set, we investigated how the two interaction layers are connected and the consequences for the robustness of these networks to plant loss. Robustness was assessed by sequentially removing plants in a random order and estimating secondary extinctions ( Fig 1C and Methods ). Although this approach lacks realism (since there are no underlying temporal dynamics), it has proven useful in understanding the threat that biodiversity loss poses to ecosystem services and functioning [ 3 , 21 , 31 , 32 ]. Furthermore, it provides a lower bound on the damages that may be caused to an ecological community since it relies on the conservative hypothesis that secondary extinctions happen only when an animal species has lost all its links. We focused on the extinctions of plants because they are the only group of species, whose disappearance can potentially harm all other species groups, and also because plants can be managed more directly [ 21 ]. Note that while plants are not the shared set of species in all networks (see Fig 1 ), it is still possible to quantify robustness to plant loss in all the networks of our data set (see Methods ). Extending the study of robustness to include multiple interactions simultaneously allowed us to study the interdependence of the robustness of animal species sets ( Fig 1D ), which is relevant to know how cascading extinctions will propagate through a multi-interaction network [ 21 ], and to better identify keystone plant species [ 13 , 21 ], of importance when designing protection and restoration interventions. We used four null models with increasing constraints (see Methods ) to study how different structural properties could determine the interdependence and robustness in the tripartite networks. Taken together, our results suggest that considering multiple ecological interactions simultaneously does not have a dramatic impact on the robustness of tripartite networks to plant losses. However, a multi-interaction approach is crucial to better gauge the overall robustness of ecological communities, to know the interdependence of the robustness of the different animal sets, and to correctly determine the relative importance of different plants species at the whole community level, which can be key for biodiversity conservation.",
"discussion": "Discussion We gathered 44 tripartite ecological networks composed by two types of ecological interactions (including herbivory, parasitism, pollination, seed dispersal, and ant-mutualism) to investigate how different interaction types were connected to each other in tripartite ecological networks and to study how considering multiple interactions simultaneously changed our knowledge of their robustness to plant loss. While multi-interaction network data sets have been gradually appearing in the literature in the last years, only a few studies have compared several of them [ 13 ]. Such comparison allows us to reveal possible commonalities of network properties (or particularities) across the different types of tripartite networks, categorized based on the sign of the ecological interactions composing them. The rationale behind this categorization is that previous studies showed that the structure of mutualistic and antagonistic ecological networks was clearly different [ 30 ]. We found fundamental differences in the way the two interaction layers are connected in the different types of tripartite networks ( Fig 2A–2C ), possibly as a consequence of underlying biological constraints. In antagonistic-antagonistic networks, the shared species hubs are almost all connectors (meaning that generalist herbivores tend to have more parasitoids, maybe because they tend to be more abundant too, or maybe due to the sampling procedure in which parasitoids can only be reared out of the sampled herbivores), while in mutualistic-mutualistic networks most shared species hubs are not connectors (meaning that generalist plants tend not to be involved in two types of mutualism simultaneously, which hints at trade-offs in the type of interactions a given species can invest in, making it unlikely that a species can e.g. invest in attracting both pollinators and ant bodyguards [ 38 ]). The more varied behaviour of mutualistic-antagonistic networks may be related to highly complex trade-offs between herbivory and pollination [ 39 ]. Intuitively, we expected these differences in the connection patterns to affect the correlation between the robustness of the animal species sets in the different types of tripartite networks. These correlations (which we named ‘interdependence’) suggest that in antagonistic-antagonistic networks the same plant species are important for both animal sets (in terms of secondary extinctions), whereas this is not the case in mutualistic-mutualistic networks. Our results add to previous evidence showing that the benefits of an intervention are not always expected to propagate throughout the whole network [ 21 ], which has implications for biodiversity conservation. They highlight the relevance of knowing the type of ecological interactions involved in an ecological community before planning restoration efforts, since, in the analysed networks containing mutualistic interactions, positive cascading effects could only be expected if the generalist plants acted as connector nodes and were the focus of the restoration plan. Surprisingly, we found that more interdependent communities are not necessarily less robust to plant losses. Rather, robustness of the overall tripartite network is determined by the particular organization of each network, with degree heterogeneity playing an important role, especially in antagonistic-antagonistic networks. The positive effect of degree heterogeneity on the robustness of food webs and bipartite mutualistic networks was already reported in [ 36 ] and in [ 3 , 40 ] (in mutualistic networks through nestedness, but it was also shown that nestedness is a consequence of degree heterogeneity [ 41 ]). It is worth noting that the robustness of mutualistic-mutualistic and mutualistic-antagonistic tripartite networks was found to be a combination of the robustness of the two bipartite networks composing them, stressing the relevance of knowing the structure of connections in both interaction layers to better quantify the robustness of the whole tripartite network. This is good news for ecologists, because it means that when measuring overall robustness to plant loss it is still possible to use multiple bipartite networks (with only one interaction type) and assume their effects are additive, as long as we know how plants connect them. Interestingly, looking at the two interaction layers simultaneously did not result in a dramatic change in the robustness of the whole community, as already reported for one of the networks in the database [ 24 ]. Nonetheless, considering the two interactions simultaneously improved the quantification of the overall robustness and is crucial to identify the most important plants in a given community. The approach we used to study robustness also allowed us to identify keystone species in the whole community. In most tripartite networks, the ranking of plant importance in the whole community is determined by the importance of plants for both animal sets (with the exception of mutualistic-mutualistic networks, that are mostly driven by one interaction layer, probably because of their disproportionate size and low connection among interaction layers). In a few cases, considering the whole community could even alter the picture considerably, since the ranking of plant importance in the whole community is emergent, i.e. it is not similar to the ranking of importance for neither of the animal sets. This evidences that considering multiple interactions simultaneously can be crucial for correctly identifying keystone species in a community. The results we present here advance our knowledge of how different interactions connect ecological communities, and how that affects the robustness of tripartite networks to plant losses. Taken together, our results suggest that considering multiple ecological interactions simultaneously does not have a dramatic impact on the overall robustness of multi-interaction networks to plant losses. However, a multi-interaction approach is crucial to know the interdependence of the robustness of the different animal sets, to better gauge the overall robustness, and to correctly determine the importance of the plants at the whole community level."
} | 3,825 |
31896758 | PMC6940364 | pmc | 667 | {
"abstract": "Neuromorphic computing based on spikes offers great potential in highly efficient computing paradigms. Recently, several hardware implementations of spiking neural networks based on traditional complementary metal-oxide semiconductor technology or memristors have been developed. However, an interface (called an afferent nerve in biology) with the environment, which converts the analog signal from sensors into spikes in spiking neural networks, is yet to be demonstrated. Here we propose and experimentally demonstrate an artificial spiking afferent nerve based on highly reliable NbO x Mott memristors for the first time. The spiking frequency of the afferent nerve is proportional to the stimuli intensity before encountering noxiously high stimuli, and then starts to reduce the spiking frequency at an inflection point. Using this afferent nerve, we further build a power-free spiking mechanoreceptor system with a passive piezoelectric device as the tactile sensor. The experimental results indicate that our afferent nerve is promising for constructing self-aware neurorobotics in the future.",
"introduction": "Introduction In the era of big data and IoT, a vast amount of sensing data, such as pictures, speeches, and videos, need to be processed in real time with a high energy efficiency 1 , 2 . This poses a significant challenge to the traditional computing architecture due to the von-Neumann bottleneck 3 – 5 . Neuromorphic computing architecture based on spiking neural networks (SNNs) has been recognized as an attractive candidate for its promising energy efficiency and powerful computing capacity 6 – 8 . Recently, various technologies have been explored to build hardware SNNs, such as digital logic circuits 7 , 8 , complementary metal-oxide semiconductor (CMOS) analog circuits 9 , 10 , and emerging memristors 11 – 13 . Given the physical limit of transistors and their lack of desirable dynamics, memristors have attracted special attention owing to their high integration intensity 14 , low power consumption 15 , analog behavior 16 – 18 , and diffusive dynamics 13 , 19 – 21 , etc. Accordingly, memristor-based artificial synapses 22 – 24 , spiking neurons 25 – 30 , have been actively studied to construct hardware implementations of SNNs lately. However, the signals collected from the environment are usually in the continuous and analog domain, and needs to be transformed into spikes first to serve as the inputs to SNNs. Therefore, a special cell analogous to afferent nerves in biology is required to receive signals from receptors and transmit spikes to the central nervous system and brain 31 , 32 . Fortunately, a bio-inspired afferent nerve based on an organic ring oscillator (ORO), whose output frequency matches the action potential of the biological sensory neuron, has been reported to control the biological motor nerves by connecting to a synapse transistor 33 . The spiking frequency of the ORO could be modulated by the input voltage controlled by the pressure sensor. Then the output of the ORO was further used to trigger a synapse transistor that connected with a biological efferent nerve, in which the different output current of the synaptic transistor is converted to voltage signals to stimulate the cockroach’s leg to generate different extension force. In addition, other types of devices, including two-terminal memristors and three-terminal transistors, have also been reported to emulate nociceptors 34 , 35 , mechanoreceptors 36 – 38 , and optical sensorimotor synapses 39 , etc., to construct high-efficient artificial sensory systems. For these systems, a high-compact artificial spiking afferent nerve (ASAN) is needed to further transform the sensed signals into spikes. The NbO x memristor is a two-terminal device with a high integration intensity. It features negative differential resistance (NDR) behavior 40 , 41 , which can serve as the basis of dynamic threshold switching with voltage sweeps 41 and has enabled emulation of biological neurons 26 , 27 , and analog computing 42 . In this work, we report an artificial spiking afferent nerve (ASAN) based on a specifically designed NbO x memristor for the first time. This NbO x device is fabricated in a three-dimensional (3D) structure and with a low thermal conductivity polysilicon (poly-Si) bottom electrode to reduce the threshold current. To construct the ASAN, a compact NbO x oscillator with a NbO x device and a resistor is built first, which can transform analog input signals into correlated spiking frequencies. For such an ASAN, the input stimuli are correlated with the voltage generated by receptor devices, and the oscillation frequency is related to the spiking frequency of the neuron, which in turn depends on the intensity of the stimuli 43 , 44 . This ASAN shows a quasi-linear relationship between input intensity and spiking frequencies under proper stimuli, and tends to reduce firing frequency when noxious stimuli are provided, which faithfully emulates the function of biological neurons 43 – 45 . To further demonstrate the spiking properties of the ASAN, three types of input pulses (square, triangular, and sinusoidal) are applied, respectively. Using this ASAN, a power-free spiking mechanoreceptor system with piezoelectric device is further proposed and demonstrated experimentally. Experimental results demonstrate that our ASAN has a great potential for using in neurorobotics and can be explored to build a general afferent nerve to communicate with higher-order SNNs.",
"discussion": "Discussion Neuromorphic machines consisting of spiking neurons and synapses could provide a more efficient approach to performing complex tasks than traditional hardware. An ASAN combined with sensors is critical for interacting with the environment, which converts the analog signal in the environment into spiking signals that could be further processed by the neuromorphic machine. The ASAN using CMOS ring oscillators shows the analog information could be converted into spiking signals 33 , 37 , in which the spiking frequency is dominated by the inverter delay. Given the physical limitations of transistors and their lack of desirable dynamics, oscillators with memristors become a more promising candidate, owing to their high integration intensity, low power consumption, and inherent dynamics, etc. However, previous works merely focus on the emulation of dynamic cortex neurons or operations for classifications 13 , 26 , 27 , 56 . The interface between the neuromorphic machine and the environment is also a critical component to construct a self-aware machine. In summary, we have proposed and experimentally demonstrated an ASAN based on specially engineered NbO x Mott memristors. The ASAN can transform analog signals into dynamic oscillation frequencies. The frequency has a quasi-linear relationship to the input voltage within a certain range of the input signal intensity and tends to stop spiking under noxious input intensity, closely resembling a biological neuron. The dynamic spiking behavior under various input signals, such as rectangular, triangular, and sinusoidal pulses, was studied systematically. We further integrated the ASAN into a piezoelectric device to construct an ASMS without any external power source. The ASMS can respond to the pressure signal and transform the pressure intensity into a corresponding spiking frequency. The ASAN can be readily extended to process sensory signals from other sensors, such as smell, taste, sight, hearing, temperature, magnetic field, and humidity. In addition to constructing the diversiform sense systems, our nerve cell is also suitable for applications in spiking neurons owning to its leaky integration and fire characteristics, or coupled oscillator neural network owing to its input-intensity-dependent oscillation behavior 56 , 60 . The nerve can thus be further used to construct complex neural networks to process central information and fabricate a highly efficient spiking neurorobotics system."
} | 1,999 |
32863734 | null | s2 | 668 | {
"abstract": "We used metabolic modeling to computationally investigate the potential of bacterial coculture system designs for CO conversion to the platform chemical butyrate. By taking advantage of the native capabilities of wild-type strains, we developed two anaerobic coculture designs by combining "
} | 72 |
37829197 | PMC10565787 | pmc | 669 | {
"abstract": "Summary The first discovered and well-characterized bacterial quorum sensing (QS) system belongs to Vibrio fischeri , which uses N -acyl homo-serine lactones (AHLs) for cell-cell signaling. AHL QS cell-cell communication is often regarded as a cell density–dependent regulatory switch. Since the discovery of QS, it has been known that AHL concentration (which correlates imperfectly with cell density) is not necessarily the only QS trigger. Additionally, not all cells respond to a QS signal. Bacteria could, via QS, exhibit phenotypic heterogeneity, resulting in sub-populations with unique phenotypes. It is time to ascribe greater importance to QS-dependent phenotypic heterogeneity, and its potential purpose in natura , with emphasis on the division of labor, specialization, and “bet-hedging”. We hope that this perspective article will stimulate the awareness that QS can be more than just a cell-density switch. This basic mechanism could result in “bacterial civilizations”, thus forcing us to reconsider the way bacterial communities are envisioned in natura.",
"conclusion": "Conclusion There is a strong need to shift the study of QS to a more appropriate ecological context where the conditions under which we propagate, and study QS resemble more closely those in natura. This perspective aims to re-envision the way we understand QS cell-cell communication in bacteria and revive research in this field by redirecting it toward the exploration of its ecological and socio-microbiological relevance, as already envisioned by Whiteley et al. 60 QS is under considerable underemphasized additional regulation which in their natural habitat, could enable other behaviors, including bacterial phenotypic differentiation and specialization, ultimately allowing a higher degree of societal organization and collective decision-making. QS could be considered as a molecular mechanism that helps determine and establish the “bacterial societal vocation”. By understanding more about signaling mechanisms that do not function as we would expect and their genetic/molecular configurations, we can also assume more about the environmental selective pressures, which shaped them in the first place, because we argue that QS systems in natura have a purpose in overcoming them. A teleological understanding of QS systems will provide us with testable hypotheses, will guide us toward a deeper understanding of cell-cell signaling, and will provide a strong basis for further translational research enabling easier exploitation and, most of all, understanding of bacteria in natura .",
"introduction": "Introduction Canonical bacterial cell-cell signaling systems consist of a “signaling” and “sensing” module, where the first produces a chemical signal, and the second interacts with it. Many species undergo cell-cell communication, each with its own variety of cell-cell signals that regulate several adaptive phenotypes. 1 Cell-cell signaling systems are usually referred to by using the broad and well-established term “Quorum Sensing” (QS). 2 , 3 In this perspective article, we will discuss the common proteobacterial LuxI-family N-acyl homoserine lactone (AHL) synthases and the LuxR-family AHL-dependent transcriptional regulators 2 as a model example of “signaling” and “sensing”, respectively. However, we think that the insights provided here can be generalized and applied to bacteria that possess other types of QS mechanisms. AHLs interact directly with the LuxR-family transcriptional regulators located in the cytoplasm. The LuxR-AHL complexes then bind to specific DNA motifs upstream of target genes and modulate gene transcription. The AHL synthase gene is usually one of the targets, thus generating a positive feedback loop. 1 , 4 The purpose of QS is usually considered to be cell density sensing (which correlates imperfectly with AHL concentration); at “quorum” (threshold of cell density) target gene expression is modulated. A “quorum”, however, is often only one of the many conditions that must be met to trigger QS-regulated phenotypes. Supra-regulation of QS has been already reported at the discovery of the first QS in Vibrio fischeri , 5 where the QS response was shown to be dependent upon catabolite repression and cyclic adenosine monophosphate (cAMP). 6 , 7 , 8 , 9 Supra-regulation is also present in many other AHL QS systems. 3 , 10 , 11 , 12 , 13 , 14 ; for example, the very well-studied AHL QS response of the plant pathogen Agrobacterium tumefaciens is strongly influenced by the plant. 15 Similarly, the two extensively studied hierarchically organized AHL QS systems of Pseudomonas aeruginosa are part of a complex regulatory network involving many other regulators (e.g., RpoS, RpoN, RsaL, …) which affect their expression. 13 , 16 QS systems of other bacteria like Vibrio harveyi, Ralstonia solanacearum , and Rhizobium leguminosarum, have also been evidenced to be under the control of other regulators, small RNAs and environmental cues. 11 , 12 , 17 , 18 , 19 Cyclic-di-guanosine monophosphate (cyclic-di-GMP) has been shown to play a role in triggering the QS response in Sinorhizobium meliloti 20 , 21 and P . aeruginosa . 22 Some AHL QS systems feature an intergenic element between the luxI and luxR genes that most often negatively and stringently affects the quantity of produced AHLs via regulation of the luxI family AHL synthase. 23 The production of signaling molecules is therefore not always turned on by default, as one could assume if considering QS as a mere cell-density switch. Pseudomonas fuscovaginae is one such example, where both of its AHL QS systems, while functional, do not produce AHLs in laboratory conditions. 24 Additionally, even when a QS signal is present, bacteria can often modulate the response. For example, Smith and Schuster 25 described an anti-activator system in P. aeruginosa , which dampens the cellular response to the signal and with it prevents self-activation. Elucidating the biological role of this supra-regulation of AHL QS systems and how it affects the cell-cell communication response of a bacterial community therefore still represents a major future challenge."
} | 1,543 |
38755154 | PMC11519548 | pmc | 670 | {
"abstract": "Atmospheric methane oxidizing bacteria (atmMOB) constitute the sole biological sink for atmospheric methane. Still, the physiological basis allowing atmMOB to grow on air is not well understood. Here we assess the ability and strategies of seven methanotrophic species to grow with air as sole energy, carbon, and nitrogen source. Four species, including three outside the canonical atmMOB group USCα, enduringly oxidized atmospheric methane, carbon monoxide, and hydrogen during 12 months of growth on air. These four species exhibited distinct substrate preferences implying the existence of multiple metabolic strategies to grow on air. The estimated energy yields of the atmMOB were substantially lower than previously assumed necessary for cellular maintenance in atmMOB and other aerobic microorganisms. Moreover, the atmMOB also covered their nitrogen requirements from air. During growth on air, the atmMOB decreased investments in biosynthesis while increasing investments in trace gas oxidation. Furthermore, we confirm that a high apparent specific affinity for methane is a key characteristic of atmMOB. Our work shows that atmMOB grow on the trace concentrations of methane, carbon monoxide, and hydrogen present in air and outlines the metabolic strategies that enable atmMOB to mitigate greenhouse gases.",
"introduction": "Introduction During the first two decades after emission to the atmosphere, methane (CH 4 ) is a greenhouse gas 80 times more potent than carbon dioxide (CO 2 ) 1 , 2 . Since 2007, the atmospheric CH 4 concentration (1905 p.p.b.v. in July 2022 https://gml.noaa.gov/ccgg/trends_ch4/ ), that is responsible for approximately 20% of the direct radiative forcing, has been increasing rapidly 1 . The CH 4 increase further accelerated in 2014 and is linked to several causes: A decline in the atmospheric concentration of hydroxyl radicals (OH) which is the main sink of atmospheric CH 4 , as OH oxidize CH 4 in the atmosphere 3 – 5 ; anthropogenic emissions from fossil fuel, agricultural, and waste sources 6 ; increased microbial CH 4 production in wetlands which suggests that current increases are also driven by feedback responses to global warming 7 . Atmospheric CH 4 oxidizing bacteria (atmMOB), a subgroup of aerobic methanotrophs, that oxidize CH 4 at its atmospheric trace concentration are the only known biological sink of atmospheric CH 4 . Compared to the OH sink (~500 Tg), the biological sink is rather small as it removes approximately 30 Tg (11−49 Tg) CH 4 from the atmosphere every year 4 . However, the biological sink has the potential to grow with increasing CH 4 concentrations. This is of particular importance as the decline of atmospheric OH, caused by reaction with atmospheric hydrogen (H 2 ) and other gases, might accelerate due to increasing H 2 emissions from a hydrogen-based economy 8 , 9 . Additionally, the biological sink is within reach of management practices devised to maximize its natural potential and harness it for CH 4 removal 10 , 11 . Yet, substantial uncertainties concerning the size of the biological sink, and the ecology and metabolic basis for growth on atmospheric CH 4 by atmMOB, caused by a historical lack of atmMOB in pure culture, has impeded our ability to study, manage, and exploit the sink 10 . In this study, by screening seven methanotrophic species, we have outlined the physiological basis that enables atmMOB to grow on atmospheric CH 4 and serve as an atmospheric CH 4 sink. In 1992, ten years after Harriss et al. reported the first indications of atmospheric CH 4 oxidation by microorganisms 12 , Bender and Conrad concluded from biphasic CH 4 oxidation kinetics of soils that an unknown group of methanotrophs might be responsible for atmospheric CH 4 oxidation 13 . Since then, two major questions have remained partially unanswered: Which methanotrophs are responsible for oxidation of atmospheric CH 4 ? How can these organisms survive and grow despite the apparent energetic limitations inherent to the oxidation of the low atmospheric CH 4 concentrations? Isotopic labeling studies revealed that members of Alpha- and Gammaproteobacteria contributed to atmospheric CH 4 oxidation and assigned them to the upland soil clusters alpha and gamma (USCα and USCγ) 14 – 16 . Several environmental and ecological studies have ascribed atmospheric CH 4 oxidation mainly to these two clusters 17 – 20 . However, over the years, studies targeting methanotrophs in upland soils have reported the presence of alphaproteobacterial methanotrophs outside the USCα 14 , 21 – 24 . These observations suggest that also conventional methanotrophs (methanotrophs assumed to grow only at high CH 4 concentrations), from genera like Methylocapsa , Methylosinus and Methylocystis , could contribute to the atmospheric CH 4 sink. Dunfield 25 summarized three potential lifestyles of atmMOB that might enable cellular maintenance and growth at the low CH 4 concentrations in air and the associated energy limitation: (i) Flush feeding on high CH 4 concentrations generated periodically in deeper soil layers, in addition to atmospheric CH 4 oxidation; (ii) An oligotrophic lifestyle based on atmospheric CH 4 as sole carbon and energy source; (iii) A mixotrophic lifestyle to utilize other substrates for energy conservation in addition to CH 4 . Flush feeding (i) is supported by the declining potential of methanotrophs to oxidize atmospheric CH 4 after several months of CH 4 starvation 25 , 26 . A study on conventional methanotrophs in rice paddy soils showed that methanotrophs regained the ability to oxidize atmospheric CH 4 after exposure to high CH 4 concentrations 27 . A high specific affinity ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 ) for CH 4 has been suggested as the key trait of oligotrophic atmMOB (ii) to enable growth with air as their only energy and carbon source 25 , 28 . This assumes that cellular energy requirements for maintenance are 4.5 kJ per carbon mole of biomass per hour (C-mol −1 h −1 ) (at 25 °C) 29 . Thus, to survive, oligotrophic atmMOB presumably need an atmospheric CH 4 oxidation rate high enough to meet these energy requirements. Such a rate can be achieved by the combination of a high affinity for CH 4 , reflected in a low half saturation constant (K m ), and a high maximum CH 4 oxidation rate (V max ), the fraction of V max and K m being referred to as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 30 . This theory is supported by low K m values for CH 4 found in several soils 13 . However, in a later study the cell specific CH 4 oxidation of USCα members was estimated to be 2.9 to 40 times lower than the presumed rate needed for cellular maintenance 31 . Therefore, the authors considered that a mixotrophic lifestyle (iii) could be the basis for atmospheric CH 4 oxidation. In line with this, a recent study reported the simultaneous oxidation of atmospheric H 2 , carbon monoxide (CO), and CH 4 by Methylocapsa gorgona MG08, the first known methanotroph and USCα member in pure culture that can grow on air (with air as sole energy and carbon source) 32 , 33 . Despite its mixotrophic lifestyle, M. gorgona MG08 did not conserve enough energy (0.38 kJ C-mol −1 h −1 ) to cover the 2.8 kJ C-mol −1 h −1 theoretically required to support maintenance at 20 °C (2.8 kJ C-mol −1 h −1 at 20 °C correspond to 4.5 kJ C-mol −1 h −1 at 25 °C) 29 , 32 , questioning whether this maintenance energy value and thus a high \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 is a relevant benchmark for the physiological capabilities of trace gas oxidizing bacteria. While the initial isolation and study of M. gorgona MG08 led to important advancements in our understanding of atmMOB, our knowledge of the metabolic basis allowing atmMOB to grow on atmospheric CH 4 remained limited. Recent energy estimates relied on the untested assumptions that all cells remained active over time and could not grow on air without CH 4 , H 2 and CO. Furthermore, these methodologically limited estimates had only been carried out on one strain, M. gorgona MG08, and had not been combined with other methods to reveal the metabolic strategies associated to the energy yields. Here we use filter cultivation to screen six alphaproteobacterial, methanotrophic strains and one gammaproteobacterial, methanotrophic strain for their ability to grow on air. This selection includes the six alphaproteobacterial strains Methylocapsa gorgona MG08, Methylosinus trichosporium OB3b, Methylocystis rosea SV97, Methylocapsa aurea KYG, Methylocapsa acidiphil a B2, and Methylocapsa palsarum NE2 33 – 38 since members of Methylocapsa , Methylosinus and Methylocystis have been observed in net sinks for atmospheric CH 4 (upland soils). Additionally, we include Methylobacter tundripaludum SV96 as representative methanotroph from the Gammaproteobacteria . To outline the physiological basis for growth on air, we perform trace gas oxidation experiments, estimate energy yields from the oxidation of trace gases in air, investigate proteome allocations, and determine 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}$${a}_{A}^{0}$$\\end{document} a A 0 for CH 4 during growth on air. Based on the lack of knowledge about nitrogen limitation of atmMOB 39 , we also evaluate their ability to grow on the nitrogen sources available in air.",
"discussion": "Results and discussion Colony formation, trace gas oxidation, and cellular energy yield during growth on air To test the ability of the seven selected methanotrophs to grow on air, we incubated each strain on filters floating on carbon free medium with ambient air as the sole carbon and energy source, hereafter referred to as filter culture. Microscopy demonstrated that all strains except Methylobacter tundripaludum SV96, the gammaproteobacterial strain, formed colonies during six months of incubation (Fig. 1A ). To test the metabolic activity of the six months old strains, we performed trace gas oxidation experiments (Fig. 1A , B ). During these experiments, filter cultures were floating on mineral medium in bottles with trace concentrations of CH 4 , H 2, and CO in the headspace. The strains, M. aurea KYG, M. gorgona MG08, M. palsarum NE2, and M. rosea SV97, hereafter referred to as atmMOB, oxidized CH 4 , CO, and H 2 , or CH 4 and CO to sub-atmospheric concentrations showing strain-specific oxidation patterns (Fig. 1A , Supplementary Fig. 1 ). M. aurea KYG oxidized CO at the highest rate, followed by CH 4 and H 2 . M. gorgona MG08 oxidized H 2 at the highest rate, followed by CO and CH 4 . M. rosea SV97 oxidized CO at the highest rate, followed by CH 4 and H 2 . M. palsarum NE2 oxidized CH 4 and CO at similar rates but did not oxidize H 2 . The gas oxidation patterns were similar after 12 months of incubation with air (Supplementary Fig. 1 ). To verify growth on the three trace gases in air, we incubated M. gorgona MG08 cells on filters under two different atmospheres: One of synthetic air without the trace gases CH 4 , CO, and H 2 (Gas composition: 400 p.p.m.v. CO 2 and 20.9% O 2 in N 2 ) and another of ambient air. Growth by colony formation was only observed in the ambient air control, whereas no growth was observed in synthetic air (Supplementary Fig. 2 ). This and the repeated observations of trace gas oxidation confirm that these strains can live and grow with air as the sole energy and carbon source. The observed oxidation of at least one atmospheric trace gas in addition to CH 4 demonstrates that all four strains are mixotrophic, matching the proposition by Dunfield that mixotrophy could be a physiological basis for atmMOB 25 . Additionally, the strain-specific trace gas uptake patterns demonstrate substantial metabolic differences between these four strains, indicating that multiple metabolic strategies can support oxidation of atmospheric CH 4 for growth. These results also demonstrated that the enduring oxidation of atmospheric CH 4 after 12 months of growth on air is not restricted to members of the USCα and USCγ, but also include members from the genera Methylocystis and Methylocapsa : M. aurea KYG, M. palsarum NE2, and M. rosea SV97 (Fig. 2 ). Fig. 1 Colony formation and trace gas oxidation on air. A Images of SYBR green I stained cells and colonies formed by M. acidiphila B2, M. aurea KYG, M. gorgona MG08, M. palsarum NE2, M. trichosporium OB3b, and M. rosea SV97 at incubation start (t0) and after six months of incubation on air as sole energy and carbon source and the related CH 4 , H 2 , and CO oxidation at atmospheric pressure displayed in red, blue, and green, respectively. Colony formation experiments with air as sole energy and carbon source have been repeated independently for at least three times per strain all leading to similar results. Source data are provided in the Source Data file and in the Supplementary Data file (Dataset 1) . B Cellular energy yield on air. Mean estimated energy yield per cell from the oxidation of atmospheric CH 4 , H 2 , and CO displayed in red, blue, and green, respectively, by M. aurea KYG, M. gorgona MG08, M. palsarum NE2, and M. rosea SV97. Dots represent the energy yield of the respective biological replicates ( n = 5) and error bars the standard deviation. Source data are provided in the Source Data file and in the Supplementary Data file (Dataset 3) . Fig. 2 Phylogeny of atmMOB based on 16S rRNA genes. Unrooted maximum-likelihood tree 84 , 85 . Full length 16S rRNA gene based phylogenetic relationship of the respective methanotrophs and the USCα. The species investigated in this study are indicated in bold print. The percentage of trees in which the associated taxa clustered together is shown next to the branches. NCBI accession numbers are given in brackets. Putative members of the USCα are highlighted in green. Source data are provided in the Source Data file. Despite formation of colonies when exposed to air only, M. acidiphila B2 and M. trichosporium OB3b did not oxidize trace gases after six months of incubation (Fig. 1A ). The ability of these two strains to produce polyhydroxyalkanoates (PHA) as storage compounds 35 , 40 might serve as an explanation for the initial colony formation. Possibly, the two strains accumulated PHA during pre-cultivation at 20% CH 4 , and then sustained their growth on air by utilizing PHA and atmospheric CH 4 as energy and carbon sources until the depletion of their storage compounds. A study showing that both strains could not stay active at low CH 4 concentrations 26 supports our observations, but further studies are needed to clarify whether the initial colony formation is based on storage compounds. Next, we asked whether atmMOB can obtain enough energy from growth on air to meet the basic maintenance energy of 2.8 kJ C-mol −1 h −1 at 20 °C postulated previously 28 . To calculate the strain-specific energy yields in C-mol, we first estimated the cellular energy yields based on the measured CH 4 , H 2 , and CO oxidation rates. M. aurea KYG yielded 6.89 x 10 −12 J cell −1 h −1 ( n = 5, SD = 3.22 x 10 −12 ), M. gorgona MG08 2.21 x 10 −12 J cell −1 h −1 ( n = 5, SD = 4.36 x 10 −13 ), M. palsarum NE2 3.09 x 10 −12 J cell −1 h −1 ( n = 5, SD = 2.90 x 10 −13 ), and M. rosea SV97 3.879 x 10 −12 J cell −1 h −1 ( n = 5, SD = 1.57 x 10 −12 ) by the oxidation of either two or three trace gases in air (Fig. 1B ). Due to the higher free energy potential and atmospheric concentration of CH 4 compared to atmospheric H 2 and CO, the energy estimates predict CH 4 as the major energy source for all four strains. Our estimates derive from the Gibbs free energy change of the following reactions at atmospheric conditions: \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{CH}}_{4}+2{O}_{2}\\to 2{H}_{2}O+C{O}_{2}$$\\end{document} C H 4 + 2 O 2 → 2 H 2 O + C O 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}$${2H}_{2}+{O}_{2}\\to 2{H}_{2}O$$\\end{document} 2 H 2 + O 2 → 2 H 2 O , 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}$$2{CO}+{O}_{2}\\to 2C{O}_{2}$$\\end{document} 2 C O + O 2 → 2 C O 2 that amount to −797.4 kJ mol −1 , −236.8 kJ mol −1 , and −199.9 kJ mol −1 , respectively. However, these numbers do not account for the energy required for activation of CH 4 by the pMMO or production cost of the enzymes involved in energy conservation from the gases. While the oxidation of H 2 and CO is catalyzed by one enzyme 41 , the oxidation of CH 4 to CO 2 by atmMOB involves at least seven enzymes (see section: Comparative proteomics: Trace gas oxidation). Thus, due to the larger investments required for energy conservation from CH 4 , the gases CO and H 2 might play more important roles as energy sources for growth on trace gases in air than indicated by the energy calculations alone. Considering the cellular dry masses and carbon content (Supplementary data file, Dataset 3 ) of M. gorgona MG08, M. aurea KYG, M. palsarum NE2, and M. rosea SV97, the energy yields per cell and hour translates to 0.40 kJ C-mol −1 h −1 (SD = 0.08 kJ C-mol −1 h −1 ), 0.71 kJ C-mol −1 h −1 (SD = 0.33 kJ C-mol −1 h −1 ), 0.38 kJ C-mol −1 h −1 (SD = 0.04 kJ C-mol −1 h −1 ), and 0.65 kJ C-mol −1 h −1 (SD = 0.26 kJ C-mol −1 h −1 ), at 20 °C, respectively (Fig. 3 ). These estimated energy yields for M. gorgona MG08, M. aurea KYG, M. palsarum NE2, and M. rosea SV97 are 3.9 – 7.4 times lower than the reported average energy requirements for cellular maintenance in aerobic bacteria (2.8 kJ C-mol −1 h −1 at 20 °C correspond to 4.5 kJ C-mol −1 h −1 at 25 °C) 29 , 32 . Fig. 3 Energy yield from oxidation of trace gases in air. Mean total energy yield of M. aurea KYG, M. gorgona MG08, M. palsarum NE2, M. rosea SV97 from the oxidation of atmospheric CH 4 , H 2 , and CO displayed in red, blue, and green, respectively, in kJ per mol cellular carbon (C-mol) and hour. Colors indicate the contribution of the individual trace gases to the total energy yield. Error bars represent the standard deviation. Black dots indicate the total energy yield from H 2 , CO, and CH 4 of the respective biological replicates ( n = 5). Source data are provided in the Source Data file and in the Supplementary Data file (Dataset 3) . The energy estimation assumes that all cells of the filter cultures actively contributed to the observed oxidation rate. Thus, in case of inactive cells, the cellular oxidation rates and the energy yields might have been underestimated. However, while the applied cell quantification (see Methods section) considers only intact cells without assessing cellular activity, Nanoscale secondary ion mass spectrometry (NanoSIMS) based 15 N incorporation confirmed activity of all measured cells (see section below: Growth on nitrogen from air), supporting our approach for energy yield estimations. However, we acknowledge that differences in activity between individual cells and potential inaccuracies in the quantification of cells contributing to observed activities might have introduced a minor error to our estimations. Thus, our observations contradict the basic energy premise for atmospheric CH 4 oxidizing bacteria 28 . Our energy estimations correspond to the similarly low energy estimates previously reported for M. gorgona MG08 32 and that of acetogenic and methanogenic microorganisms (0.2 kJ C-mol −1 h −1 at 37 °C) 42 . Comparative proteomics We further investigated the cellular adjustments required for life on air by M. gorgona MG08, M. rosea SV97, and M. palsarum NE2 by comparing the proteomes of the three strains when exposed to air (~1.9 p.p.m.v. CH 4 ) and when exposed to high CH 4 concentrations (~1000 p.p.m.v. CH 4 ) in air. Correspondence analyses (CA) of relative protein abundances at the two CH 4 concentrations revealed a clear difference in proteome allocation. This is shown by the separation of samples from the two conditions along the first CA dimension, which accounted for 79.3%, 76.5%, and 69.6% of the total inertia for M. gorgona MG08, palsarum NE2, and M. rosea SV97, respectively (Supplementary Fig. 3 ). Thus, the CA indicate that the largest shifts in the proteomes occurred as responses to changes in CH 4 concentration. To identify which proteins contributed most to these shifts, we identified the proteins (top 10%) with the largest contribution to the first-dimension inertias of the CA and plotted their abundances across the two conditions (Supplementary Fig. 4 ). We found major protein expression shifts within a variety of functional categories (based on hierarchical EggNOG 43 annotations), but for all three strains, a large proportion was related to core metabolisms, including transcription, translation, growth, energy metabolism, and amino acid and carbohydrate transport. A prominent trend observed for all three strains was that proteins associated with the categories Cell cycle control, cell division, chromosome partitioning and Cell wall/membrane/envelope biogenesis shifted towards lower relative abundances, hereafter referred to as downregulation, at atmospheric CH 4 compared to the 1000 p.p.m.v. CH 4 treatment. This pattern suggests that atmMOB lower the allocation of resources for growth when CH 4 availability is low, which is in line with previously reported differences in colony growth over time when comparing incubations on atmospheric and 1000 p.p.m.v. CH 4 concentrations 33 . We also observed major shifts in protein abundances for the categories Energy production and conversion and Carbohydrate transport and metabolism, including proteins involved in trace gas oxidation and carbon assimilation (Supplementary data file, Dataset 12 − 14 ). Based on that and the differences in trace gas oxidation patterns between the strains (Fig. 1 ), the relative abundances of proteins involved in trace gas oxidation, carbon assimilation via the serine cycle (Fig. 4 ), and the electron transport chain (Supplementary Fig. 5 ) were further investigated. Fig. 4 Metabolic adjustments during growth on air. Comparative proteomics of M. gorgona MG08, M. rosea SV97, and M. palsarum NE2 exposed to 1000 p.p.m.v. CH 4 (High) in air and 1.9 p.p.m.v. (Atm) CH 4 in air. Normalized and standardized expression of enzymes involved in the central carbon and energy metabolism is shown. High relative abundance = orange, low relative abundance expression = blue. * indicates significant difference in expression between treatments (two-sided t-test). Horizontal lines in the heatmaps separate operons of enzymes catalyzing the same reaction.? = unknown enzyme. Abbreviations: pMMO—particulate methane monooxygenase, MDH—methanol dehydrogenase (Mox) and (XoxF), Hhy— [NiFe] hydrogenase, CODH—carbon monoxide dehydrogenase, Fae—formaldehyde activating enzyme, MtdB—NAD(P)-dependent methylene-tetrahydromethanopterin dehydrogenase, Mch—methenyl-tetrahydromethanopterin cyclohydrolase, Fhc—formyltransferase/hydrolase, Fdh A—NAD-dependent formate-dehydrogenase, Fdh AB—molybdopterin dependent formate dehydrogenase, Fhs—formate-tetrahydrofolate ligase, FolD—bifunctional methylene-tetrahydrofolate dehydrogenase/methenyl-tetrahydrofolate cyclohydrolase, FchA—methenyl-tetrahydrofolate cyclohydrolase, MtdA—NADP-dependent methylene-tetrahydrofolate dehydrogenase, GcvH—glycine cleavage system H protein, GcvT—aminomethyltransferase, GcvP—glycine dehydrogenase, Lpd—dihydrolipoyl dehydrogenase, GlyA—serine hydroxymethyltransferase, SgaA—serine-glyoxylate aminotransferase, HprA—glycerate dehydrogenase, GarK – 2-glycerate kinase, Eno—enolase, Ppc—phosphoenolpyruvate carboxylase, Mdh—malate dehydrogenase, Mal—malate-CoA ligase, Mcl—L-malyl-CoA lyase, Acc—Acetyl-CoA carboxylase, NAD + —nicotinamide adenine dinucleotide, NADP + —nicotinamide adenine dinucleotide phosphate, ATP—adenosine triphosphate. Source data are provided in the Source Data file and in the Supplementary Data file (Dataset 5) . Comparative proteomics: trace gas oxidation All three strains expressed a particulate methane monooxygenase (pMMO) that catalyzes the hydroxylation of CH 4 to methanol (CH 3 OH). M. gorgona MG08 and M. rosea SV97 contained higher relative abundances, hereafter referred to as upregulation, of pMMO at 1000 p.p.m.v. CH 4 , whereas M . palsarum NE2 upregulated pMMO at atmospheric CH 4 concentrations. Furthermore, both M. gorgona MG08 and M. rosea SV97 upregulated a high affinity [NiFe] hydrogenase (Hhy), and M. gorgona MG08 a molybdenum-dependent carbon monoxide dehydrogenase (CODH), when exposed to air. These differences in enzyme expression patterns demonstrate different metabolic strategies to grow on air: M. rosea SV97 and M. gorgona MG08 compensate for energy limitation by upregulating enzymes for energy conservation from H 2 or H 2 and CO, while M . palsarum NE2 compensates for the limitation by upregulating pMMO. The Hhy increase is similar to the Hhy increase in Mycobacterium smegmatis enabling long-term persistence after carbon limitation 44 . The observed upregulation of pMMO, Hhy, or CODH implies that the strains allocate resources to increase their \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 (specific apparent affinity) for the respective trace gases as adaptations to growth on air. Additionally, the distinct strategies to grow on air by the closely related strains M. gorgona MG08 and M. palsarum NE2 suggest niche differentiation between atmMOB in nature. Furthermore, our observations of the consumption of multiple trace gases by atmMOB and adjusted resource allocation for trace gas uptake driven by changes in CH 4 concentration, are in line with earlier observations of a negative correlation between soil H 2 concentrations and the uptake of atmospheric CH 4 45 . Despite the ability of M. rosea SV97 and M . palsarum NE2 to oxidize CO (Fig. 1A ), we were not able to determine the responsible protein complex(es) or the corresponding gene expression responses. In M. rosea SV97, candidate genes for CODH were identified by blasting the genome sequences against a non-redundant CODH sequence database created by using the Identical Protein Groups resource 46 (Supplementary data file, Dataset 11 ). However, not all the potential CODH subunits necessary to form a functional CODH were detected in the proteomes. The same blast-based approach did not result in potential candidate genes that could encode a functioning CODH in M. palsarum NE2 (Supplementary data file, Dataset 10 ). Thus, our results indicate that distantly related or previously undiscovered enzymes catalyze atmospheric CO oxidation in M. palsarum NE2. The expression of the methanol dehydrogenase (MDH), which catalyzes the second step in CH 4 oxidation, the oxidation of methanol (CH 3 OH) to formaldehyde (CH 2 O), matched the pMMO expression patterns for all three strains. This indicates a close interaction between these two enzymes as previously suggested for Methylococcus capsulatus 47 . MDHs were upregulated at high CH 4 concentrations by M. gorgona MG08 and M. rosea SV97, but not by M . palsarum NE2 (Fig. 4 ). Only the putative lanthanide-dependent methanol dehydrogenase (XoxF) of M . palsarum NE2 did not follow the pMMO pattern. Formaldehyde is a key intermediate of both catabolism and anabolism in methanotrophs. The enzymes involved in the catabolic oxidation of formaldehyde via methylene-tetrahydromethanopterin (-H 4 MPT), methenyl-H 4 MPT, and formyl-H 4 MPT to formate (CHOO − ), were downregulated in the air treatment of the three strains (Fig. 4 ). The toxic and highly reactive formaldehyde condenses spontaneously to methylene-H 4 MPT 48 and methylenetetrahydrofolate (H 4 F) 49 and thus, higher concentrations of formaldehyde activating enzymes (Fae) may not be required for the oxidation of formaldehyde during growth on air. This may explain why enzymes involved in formaldehyde oxidation, despite the upregulation of pMMO and MDH, were downregulated in M. palsarum NE2 at atmospheric CH 4 concentrations. At high CH 4 concentrations, the upregulation of Fae and downstream enzymes for further oxidation to formate could represent a cellular detoxification mechanism to avoid high cellular formaldehyde concentrations and to increase NADH synthesis 50 . Two different formate dehydrogenases (Fdh A and Fdh AB) that catalyze the energy-conserving oxidation of formate to CO 2 were expressed by M. gorgona MG08 and M. palsarum NE2 (Fig. 4 ). The metal-free and potentially irreversible NAD + -dependent Fdh A 33 , 51 was upregulated at atmospheric CH 4 concentrations in M. gorgona MG08 and M. palsarum NE2. Possibly, the upregulation of a one-directional enzyme minimizes back-flow of CO 2 to formate at high CO 2 concentrations and thus prevents energy loss by CO 2 reduction. Fdh AB, which was expressed by all three strains, is molybdopterin-dependent and homologous to the reversible Fdh found in Rhodobacter capsulatus 52 . It catalyzes, in addition to formate oxidation, the reduction of CO 2 to formate 52 , 53 . The Fdh AB was upregulated in all three strains at high CH 4 concentration. Under these conditions, when excess reducing power is available, the Fdh AB might enable the reduction of CO 2 for carbon assimilation via the reductive glycine pathway and serine cycle, but further investigations are needed to test whether CO 2 reduction to formate truly occurs in atmMOB. Comparative proteomics: carbon assimilation Carbon assimilation was downregulated in all three strains at atmospheric CH 4 concentrations. In the three strains, formaldehyde can condense with H 4 F to methylene-H 4 F and then be further assimilated through the glycine cleavage system and serine pathway. Additionally, formaldehyde can be first oxidized via the H 4 MPT-mediated pathway to formate and then enter, instead of being oxidize to CO 2 , the H 4 F-mediated reductive pathway to methylene-H 4 F (Fig. 4 ). The increased expression of enzymes involved in the reduction of formate via formyl-H 4 F and methenyl-H 4 F to methylene-H 4 F in the high CH 4 treatment indicates an enhanced investment into carbon assimilation. Since formaldehyde spontaneously condenses with H 4 F to methylene-H 4 F, the upregulated enzymes for H 4 F-mediated formaldehyde oxidation could also contribute to formate formation. However, experiments with Methylorubrum extorquens demonstrated that formaldehyde oxidation occurred through its H 4 MPT-mediated pathway while the reductive pathway via formate to methylene-H 4 F represented the major assimilatory flux 50 , 54 . Therefore, we consider this as the most likely explanation for the expression patterns in M. gorgona MG08, M. rosea SV97, and M . palsarum NE2. The glycine cleavage/synthase system (GCS) 55 , 56 catalyzes the oxidative cleavage of glycine to NADH, NH 3 , CO 2 , and a methylene group. It also catalyzes the reverse reaction, the synthesis of glycine from NADH, NH 3 , CO 2 , and methylene-H 4 F. The upregulation of the GCS by M. rosea SV97 and M . palsarum NE2 at high CH 4 concentrations suggests an increased synthase activity and thus increased carbon assimilation, corresponding to the overall upregulation of the reductive glycine pathway (Fig. 4 ). However, we lack a plausible explanation for the increased expression of the GCS by M. gorgona MG08 at atmospheric CH 4 concentrations. As the next step in carbon assimilation, the bidirectional serine hydroxymethyltransferase (GlyA) condenses glycine with methylene-H 4 F to serine, representing the first step of the serine cycle (Fig. 4 ). The enzymes involved in the serine cycle were upregulated by the strains when exposed to high CH 4 concentrations, indicating investment into carbon assimilation at this condition. Comparative proteomics: electron transport chain The relative abundances of protein complexes involved in the electron transport chain for ATP synthesis was lower in atmospheric CH 4 compared to the high CH 4 treatment (Supplementary Fig. 5 ). The NADH-quinone oxidoreductase, cytochrome c oxidase, and the ATP synthase were all highly expressed at high CH 4 concentrations indicating increased investment into energy conservation at high substrate supply. We propose that to overcome energy limitations when exposed to air, all three strains upregulate the expression of enzymes for the oxidation of at least one trace gas to maximize uptake rates and energy yield, while investments in energy conservation and energy-intensive carbon assimilation are reduced. This reduced investment in assimilation is in line with the low concentrations of the trace gases, low uptake rates, and overall slow growth of atmMOB when incubated with air as energy and carbon source 33 . Specific affinity 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}$${a}_{A}^{0}$$\\end{document} a A 0 for CH 4 , expressed as the fraction of V max(app) and K m(app) , directly represents the capacity to oxidize CH 4 at low concentrations. To test if the pMMO upregulation in M. palsarum NE2 and downregulation in M. gorgona MG08 at atmospheric CH 4 concentrations translate into a 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}$${a}_{A}^{0}$$\\end{document} a A 0 for CH 4 by M. palsarum NE2 compared to M gorgona MG08, we measured 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}$${a}_{A}^{0}$$\\end{document} a A 0 of the two strains when grown on air. To avoid increases in the apparent half saturation constant (K m(app) ) estimates at high CH 4 concentrations 57 , we pre-incubated the cultures on filters floating on carbon-free medium for five months with air as sole energy and carbon source. By measuring the CH 4 oxidation per cell and hour at different CH 4 concentrations, we estimated a K m(app) of 48.54 nM CH 4 and a V max(app) of 4.91 x 10 −8 nmol cell −1 h −1 resulting in a \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 of 1.01 x 10 −9 L cell −1 h −1 for M. gorgona MG08 (Fig. 5 ). A K m(app) of 48.54 nM CH 4 is within the range of the K m(app) values measured for fresh oxic soils reported by Bender and Conrad (30 – 51 nM) 13 , which led to the theory that atmMOB are oligotrophs with a high affinity for CH 4 . This is the first observation of a methanotroph in pure culture showing a K m(app) that is in the range of the low K m(app) values measured in upland soils 25 , 58 . Our \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 estimate for M. gorgona MG08 (1.01 x 10 −9 L cell −1 h −1 ) is approximately five times higher than 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}$${a}_{A}^{0}$$\\end{document} a A 0 reported by Tveit et al. 1.95 × 10 −10 L cell −1 h −1 33 . However, the K m(app) and V max(app) reported by Tveit et al. amount 4905 nM and 95.4 x 10 −8 nmol cell −1 h −1 , respectively, values approximately 100 and 20 times higher than in the current study. These differences might derive from the use of liquid cultures pre-incubated at 20% CH 4 by Tveit et al. Such a high CH 4 concentration could have influenced the cellular pMMO concentration, as indicated by the pMMO upregulation by M. gorgona MG08 at 1000 p.p.m.v. CH 4 described above (Fig. 4 ), and thus increased the K m(app) and V max(app) estimates. Dunfield and Conrad reported a similar alteration of K m(app) and V max(app) for Methylocystis strain LR1 after comparing starved cells to cells exposed to 10% CH 4 while 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}$${a}_{A}^{0}$$\\end{document} a A 0 was more constant 57 . Fig. 5 CH 4 oxidation kinetics. Michaelis-Menten hyperbolic curve (displayed in red) fitted to the oxidation rate per cell of M. gorgona MG08 and M. palsarum NE2 at CH 4 concentrations ranging between 3 nM and 287 nM. The nM CH 4 display CH 4 concentrations dissolved in water and correlate with the partial pressure of CH 4 above the water. At 20 °C and atmospheric pressure (1.013 bar), 2.98, 50, 100, and 250 nM CH 4 dissolved in water correspond to 1.9, 31.9, 63.7, and 159.3 p.p.m.v. in air. Source data are provided in the Source Data file and in the Supplementary Data file (Dataset 9) . The K m(app) of M. palsarum NE2 was 402.08 nM CH 4 , which is approximately eight times higher than the K m(app) of M. gorgona MG08. Despite this higher K m(app) , 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}$${a}_{A}^{0}$$\\end{document} a A 0 of M. palsarum NE2 was 3.30 x 10 −9 L cell −1 h −1 , three times higher than estimated for M. gorgona MG08. This high \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 derives from its substantially higher V max(app) of 133 x 10 −8 nmol cell −1 h −1 (Fig. 5 ) and aligns with the proteomic data. The upregulation of the pMMO at atmospheric CH 4 concentrations by M. palsarum NE2 seem to translate into higher a \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 for CH 4 compared to M. gorgona MG08 that downregulated its pMMO. This is also reflected in the CH 4 oxidation rate of M. palsarum NE2 at atmospheric concentrations, which surpassed the rate of M. gorgona MG08, despite having a lower apparent affinity for CH 4 (Fig. 5 ). An apparent affinity of 402.08 nM CH 4 is not considered as high affinity 28 , 41 . Thus, as both strains grow on air, a high apparent affinity for CH 4 cannot be considered a prerequisite for this lifestyle. 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}$${a}_{A}^{0}$$\\end{document} a A 0 of M. gorgona MG08 and M. palsarum NE2 are approximately equally high and three times higher, respectively, than the recently reported \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 of Methylotuvimicrobium buryatense 5GB1C 59 and 30 and 100 times higher, respectively, than 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}$${a}_{A}^{0}$$\\end{document} a A 0 of Methylocystis sp. SC2 (3.4 × 10 −11 L cell −1 h −1 ) 26 , the MOB with the fourth highest \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 for CH 4 known so far. However, the different experimental setups and CH 4 concentrations used to determine \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 might render the comparisons invalid 26 , 59 . Nevertheless, our results show that the specific affinity ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 ), rather than the affinity (K m(app) ), is the appropriate model to determine the efficiency of atmospheric CH 4 utilization by atmMOB. This is in line with the work on oligotrophic substrate uptake at low concentrations by Button 60 . Growth on nitrogen from air The four atmMOB, M. gorgona MG08, M. palsarum NE2, M. rosea SV97, and M. aurea KYG, encode all genes required for dinitrogen (N 2 ) fixation 33 and grow in nitrogen-free medium at high CH 4 concentrations 33 , 36 , 37 , 38 . To test the potential for growth on nitrogen from air, we incubated filter cultures of these strains with air as the sole energy, carbon, and nitrogen source. The colony formation of all four strains after three months, and trace gas uptake by M. gorgona MG08 after one year, demonstrates growth in the absence of bioavailable nitrogen sources in the medium (Fig. 6A , B , and Supplementary Fig. 6 ). This suggests that all four strains can either cover their nitrogen requirements by the fixation of N 2 , by the utilization of atmospheric reactive nitrogen during growth on air, or both. While the N 2 concentration in air is approximately 78 %, ammonia concentrations, for example, have been observed to range between 0.2 and 24 p.p.b.v 61 . Fig. 6 Growth with air as nitrogen source. A Images of SYBR green I stained colonies formed by M. gorgona MG08, M. palsarum NE2, M. rosea SV97, and M. aurea KYG after three months of incubation with air as sole energy, carbon, and nitrogen source. Colony formation experiments with air as sole energy, carbon, and nitrogen source have been repeated independently for at least two times per strain all leading to similar results. B Trace gas oxidation with air as nitrogen source. Mean trace gas oxidation of M. \n gorgona MG08 filter cultures after 12 months of incubation with air as sole carbon, energy, and nitrogen source. Error bars indicate standard deviation between the biological replicates ( n = 4). Source data are provided in the Source Data file and in the Supplementary Data file (Dataset 1) . C \n 15 N incorporation on air. NanoSIMS visualization of 15 N incorporation by M. gorgona MG08 after two months of incubation under a 15 N 2 -enriched atmosphere. Fraction values of 15 N/( 14 N + 15 N) are given in at%. Different colors represent the cellular 15 N content as displayed by the color bar. The 15 N incorporation on air experiment has not been repeated. Data are provided in the Supplementary Data file (Dataset 15) . To test for N 2 fixation during growth on air by M. gorgona MG08, we incubated filter cultures in air enriched with 15 N-N 2 (~23 atom % (at%) of the total N 2 ) and CH 4 , H 2 , and CO concentrations fluctuating between 0.03 and 3 p.p.m.v. for two months. Afterwards, we measured cellular 15 N 2 fixation using NanoSIMS (Fig. 6C ). All cells ( n = 379) measured during NanoSIMS incorporated 15 N (Supplementary Dataset 15 and 16 ) indicating that all cells have been active during incubation. This supports the validity of our energy estimations (as mentioned above) and demonstrates growth on trace gases in air. The cellular 15 N ranged from 2.99 at% to 61.89 at% with an average of 30.79 at% (SD = 7.82 at%) (Supplementary data file, Dataset 15 ) while the control without 15 N 2 enrichment averaged at 0.37 at% (SD = 0.04) (Supplementary data file, Dataset 16 ). Since the 15 N 2 in the headspace amounted to approximately 23 at% of the total N 2 during incubation, the 15 N average of the cells should not have amounted to more than 23 at%. The high values might issue from bioavailable 15 N-species in the 15 N 2 gas used for incubation 62 . Purity tests of the 15 N 2 gas prior to the 15 N 2 fixation experiments revealed only a minor NO x contamination of 101.59 (nmol ml −1 ) with a very low 15 N fraction of 0.0014 at% and ammonia levels below detection limit (1 nmol ml −1 ), and thus do not provide indications that contamination can explain our observations. However, even undetectable trace amounts of 15 N ammonia contaminating the 15 N 2 might be sufficient to cause high cellular 15 N-enrichments given the low amount of biomass of the filter cultures and the long incubation times. Thus, despite not being able to detect any 15 N-contaminants, we cannot exclude contamination and thereby cannot confirm N 2 fixation. However, considering the growth on nitrogen-free medium (Supplementary Fig. 6 ) and the NanoSIMS experiment, we can conclude that M. gorgona MG08 either fixes N 2 or trace concentrations of reactive nitrogen species. This demonstrates that atmMOB can cover their nitrogen requirements for growth, in addition to energy and carbon, from air. Additionally, it suggests that atmMOB may not be nitrogen limited under most natural conditions, partially answering the question by Bodelier and Steenberg regarding conditions that can be nitrogen limiting to atmMOB 39 . Implications of the findings Overall, the results demonstrate that growth on air and enduring oxidation of atmospheric CH 4 is not restricted to members of the clades USCα and γ, but more widespread than previously assumed. Our data show that former liquid culture-based cultivation approaches lead to an underestimation of the true potential of methanotrophic pure cultures to oxidize atmospheric CH 4 . The appearance of atmMOB outside the USCα and γ revises our understanding of the biological atmospheric CH 4 sink and should be considered in future studies. The strain specific oxidation patterns and the estimations of the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 for CH 4 demonstrate that both the mixotrophic oxidation of atmospheric trace gases and a high \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 for CH 4 are key to obtain sufficient energy for growth on air. The estimated energy requirements for growth of the four atmMOB are substantially lower than the maintenance energy value used as basic premise for an oligotrophic lifestyle of MOB. Additionally, atmMOB seem to cover not only their energy and carbon but also their nitrogen requirements from the atmosphere. This opens a new perspective on physiological limitations of atmospheric trace gas oxidizers and suggests that atmMOB may carry an ideal set of properties needed for pioneering species to initiate primary succession in unfavorable environments. Additionally, the high \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${a}_{A}^{0}$$\\end{document} a A 0 for CH 4 enables atmMOB to utilize trace concentrations of CH 4 as energy and carbon source for growth while being extremely oligotrophic. This bears the potential for cost-effective and efficient biofiltration of anthropogenic emissions containing CH 4 concentrations far below the lower explosive limit. The common metabolic strategy to grow on air seems to be the downregulation of enzymes involved in energy-intensive processes combined with the upregulation of enzymes for the oxidation of trace gases. However, the differing expression patterns of enzymes for trace gas oxidation and the strain-specific trace gas oxidation patterns indicate that a diverse metabolic repertoire has evolved to enable life on air."
} | 13,053 |
35061541 | PMC8782456 | pmc | 671 | {
"abstract": "Conductive-bridging random access memory (CBRAM) has garnered attention as a building block of non–von Neumann architectures because of scalability and parallel processing on the crossbar array. To integrate CBRAM into the back-end-of-line (BEOL) process, amorphous switching materials have been investigated for practical usage. However, both the inherent randomness of filaments and disorders of amorphous material lead to poor reliability. In this study, a highly reliable nanoporous–defective bottom layer (NP–DBL) structure based on amorphous TiO 2 is demonstrated (Ag/a-TiO 2 /a-TiO x /p-Si). The stoichiometries of DBL and the pore size can be manipulated to achieve the analog conductance updates and multilevel conductance by 300 states with 1.3% variation, and 10 levels, respectively. Compared with nonporous TiO 2 CBRAM, endurance, retention, and uniformity can be improved by 10 6 pulses, 28 days at 85°C, and 6.7 times, respectively. These results suggest even amorphous-based systems, elaborately tuned structural variables, can help design more reliable CBRAMs.",
"introduction": "INTRODUCTION Various resistance-switching memristors have been demonstrated as nonvolatile memory and bioinspired neuromorphic applications ( 1 – 4 ). There are diverse mechanisms that explain a change of device resistance ( 5 – 7 ). A filamentary switching mechanism, which involves the formation and disruption of a nanoscale conductive filament (CF), has been extensively studied. Oxide-based resistive random access memory (RRAM) and conductive-bridging random access memory (CBRAM) are two representative examples operated via filamentary switching; however, varying resistances originate from anion vacancies in RRAM, while they do in active metal cations [e.g., copper (Cu) or silver (Ag)] in CBRAM ( 8 , 9 ). Therefore, they exhibit different characteristics such as the on/off current ratio, endurance, retention, reliability, and switching speed ( 10 ). CBRAM’s strengths compared with RRAM are large on/off ratio, fast switching speed, and low power consumption, which are essential characteristics for neuromorphic applications, owing to the high mobility of metal cations compared with oxygen anions ( 11 , 12 ). Thus, CBRAM is an attractive candidate for a biomimicking artificial synapse with tunable conductance ( 13 ). However, despite these advantages, CBRAM has reliability issues that prevent its integration into a large-scale array ( 14 , 15 ) and the formation of high-density features required to reproduce the high connectivity characteristic of the biological synapse system ( 16 ). For instance, the set/reset voltage is highly variable because of stochastic metal cation transport through an amorphous three-dimensional (3D) switching medium. In addition, the endurance and retention of CBRAM are generally worse than those of oxide-based RRAM because of direct (metal-metal) contact with the active metal electrode [e.g., copper (Cu) or silver (Ag)] and an inert counter electrode. This is prone to irreversible breakdown, known as set stuck effect ( 17 , 18 ). Moreover, continuous metal filament breaks up into a sphere-like shape as time goes on because of the high surface tension between active metal and switching medium known as “Rayleigh instability” ( 19 – 21 ). To solve these problems in CBRAM, several researchers have explored how oxide-based RRAM has made meaningful progress to enhance device variation, endurance, and retention from both material and structural perspectives. In general, oxide-based RRAM can enhance reliability through localized switching, which reduces the uncontrollable 3D random filament growth and rupture process ( 12 , 22 , 23 ). Similarly, recent studies have shown the remarkable stability of CBRAM using confined filaments through various methods ( 4 , 24 – 26 ). Previously, the 1D channel formation by using intrinsic dislocation of epitaxially grown SiGe on Si exhibited an exceptionally low variation, and it can be used as the confined pathway of CF to improve the CBRAM performance, which is known as epitaxial random access memory (epiRAM) ( 4 ). However, epitaxial layer growth requires a high temperature and epitaxial seed layer, which is difficult to apply to the back end of line (BEOL) process to integrate the EpiRAM on the complementary metal-oxide semiconductor (CMOS) circuit vertically for high-bandwidth communication. Therefore, an amorphous switching medium is highly recommended for further development of CBRAM-based non–von Neumann architectures and neuromorphic systems. Another approach is that the reliability including endurance and retention can be enhanced by introducing an additional thin layer that acts as a buffer layer to regulate filament overgrowth ( 27 – 32 ). Although there are positive effects by preventing set stuck effect, most of the devices still have a large variation. In table S1, detailed characteristics are compared. In this study, we propose a highly reliable device structure that has approximately 10-nm-diameter nanopores with a sub–10 nm defective bottom layer (DBL) (see fig. S1 for the detailed process). The pores were formed simultaneously with the oxidation process of Ti and facilitated localized filament growth via 1D pores to achieve highly reliable switching in both dc I - V characteristics and pulse measurement with stable retention. Consequently, the CF is confined in a 1D predefined path and exhibits stable multilevel memory properties with a high on/off ratio (>10 3 ). This pore-assisted switching was identified as having low variation, and it was further verified by observing the retention properties as an aspect ratio of the pore size. Furthermore, the thin and defective layer beneath the pore successfully improved the device endurance by efficiently regulating filament overgrowth with on/off ratio tunability. Thus, this first attempt to accommodate both confined filaments and a thin DBL with BEOL compatibility will pave the way for the optimal amorphous-based CBRAM structure to construct a highly reliable large-scale array.",
"discussion": "DISCUSSION In this study, we first performed a systemic analysis of the pore confinement effect and the stoichiometries of the bottom layer that affect the memristor reliability through anodization. Because of the porous structure in the switching medium, the retention characteristic is enhanced to maintain multilevel conductance, and it can be easily controlled by increasing or decreasing the size of the pore with a low temporal variation. These findings provide compelling evidence for pore-assisted resistance switching, which is the key to reducing unwanted variations. In addition, changing the stoichiometries of the inserted layer below the pore (DBL) facilitates the adjustment of the on/off ratio and endurance despite having a somewhat tradeoff relation with a variation. Moreover, DBL-induced nonlinearity can be used for array operation with the help of self-rectifying behavior. These results suggest a general framework for optimizing device performance by structural modifications depending on diverse applications. Last, this amorphous-based NP-DBL device can be integrated and commercialized using the BEOL process for monolithic 3D integration and constructing non–von Neumann architectures because the entire fabrication process was performed at a low temperature."
} | 1,840 |
27762316 | PMC5071877 | pmc | 673 | {
"abstract": "Most simulations of neuroplasticity in memristors, which are potentially used to develop artificial synapses, are confined to the basic biological Hebbian rules. However, the simplex rules potentially can induce excessive excitation/inhibition, even collapse of neural activities, because they neglect the properties of long-term homeostasis involved in the frameworks of realistic neural networks. Here, we develop organic CuPc-based memristors of which excitatory and inhibitory conductivities can implement both Hebbian rules and homeostatic plasticity, complementary to Hebbian patterns and conductive to the long-term homeostasis. In another adaptive situation for homeostasis, in thicker samples, the overall excitement under periodic moderate stimuli tends to decrease and be recovered under intense inputs. Interestingly, the prototypes can be equipped with bio-inspired habituation and sensitization functions outperforming the conventional simplified algorithms. They mutually regulate each other to obtain the homeostasis. Therefore, we develop a novel versatile memristor with advanced synaptic homeostasis for comprehensive neural functions."
} | 288 |
31728021 | PMC6976610 | pmc | 674 | {
"abstract": "Cable bacteria of the family Desulfobulbaceae couple spatially separated sulfur oxidation and oxygen or nitrate reduction by long-distance electron transfer, which can constitute the dominant sulfur oxidation process in shallow sediments. However, it remains unknown how cells in the anoxic part of the centimeter-long filaments conserve energy. We found 16S rRNA gene sequences similar to groundwater cable bacteria in a 1-methylnaphthalene-degrading culture (1MN). Cultivation with elemental sulfur and thiosulfate with ferrihydrite or nitrate as electron acceptors resulted in a first cable bacteria enrichment culture dominated >90% by 16S rRNA sequences belonging to the Desulfobulbaceae . Desulfobulbaceae -specific fluorescence in situ hybridization (FISH) unveiled single cells and filaments of up to several hundred micrometers length to belong to the same species. The Desulfobulbaceae filaments also showed the distinctive cable bacteria morphology with their continuous ridge pattern as revealed by atomic force microscopy. The cable bacteria grew with nitrate as electron acceptor and elemental sulfur and thiosulfate as electron donor, but also by sulfur disproportionation when Fe(Cl) 2 or Fe(OH) 3 were present as sulfide scavengers. Metabolic reconstruction based on the first nearly complete genome of groundwater cable bacteria revealed the potential for sulfur disproportionation and a chemo-litho-autotrophic metabolism. The presence of different types of hydrogenases in the genome suggests that they can utilize hydrogen as alternative electron donor. Our results imply that cable bacteria not only use sulfide oxidation coupled to oxygen or nitrate reduction by LDET for energy conservation, but sulfur disproportionation might constitute the energy metabolism for cells in large parts of the cable bacterial filaments.",
"introduction": "Introduction Cable bacteria are filamentous multicellular microorganisms belonging to the family Desulfobulbaceae [ 1 ]. They appear in redox gradients where the cells of one end of the filaments seemingly oxidize sulfide to sulfate [ 2 ]. The electrons from sulfide oxidation can be transported over several centimeters by long-distance electron transfer (LDET) to the sediment surface where they are used for oxygen or nitrate reduction [ 3 – 5 ]. The electrons are transported via conductive fibers in the periplasm leading to the distinctive morphology of a continuous ridge pattern over the whole length of cable bacteria [ 6 ]. Since their first discovery in sediments from Aarhus Bay [ 1 ], cable bacteria were found in many other marine sediments all over the world [ 7 ] but also in a freshwater stream in Denmark [ 8 ] as well as in groundwater contaminated with hydrocarbons [ 9 ]. So far, no attempts to cultivate cable bacteria in pure culture or in a stable enrichment culture have been successful. Based on genome sequencing, the cable bacteria known so far belong to a monophyletic sister clade of the genus Desulfobulbus with two proposed genera Candidatus Electrothrix and Candidatus Electronema [ 10 , 11 ]. 16S rRNA gene sequences of groundwater cable bacteria formed a distinct phylogenetic clade with the closest cultivable relative Desulfurivibrio alkaliphilus [ 12 ]; a single-celled, rod-shaped alkaliphilic microorganism capable of sulfur disproportionation [ 12 ] and sulfide oxidation with nitrate as electron acceptor [ 13 ]. Surprisingly, we found 16S rRNA gene sequences of groundwater cable bacteria in the enrichment culture 1MN [ 14 ] that anaerobically degrades 1-methylnaphthalene or naphthalene with ferric iron as electron acceptor. This culture contains two dominant organisms affiliated to Thermoanaerobacteraceae and Desulfobulbaceae (Fig. 1b ). The Thermoanaerobacteraceae were identified as the degraders of naphthalene by stable isotope probing experiments and the detection of putative genes encoding enzymes for naphthalene degradation [ 14 ]. The Desulfobulbaceae shared 16S rRNA gene identity of >98% with previously published sequences of groundwater cable bacteria (Fig. 1 a) [ 9 ]. Since iron reduction and naphthalene oxidation are in stark contrast to the environmental conditions where cable bacteria are usually found, the discovery of groundwater cable bacteria in this chemo-organo-heterotrophic culture raised the question for their metabolic role. Our hypothesis was that sulfur disproportionation plays a major role in energy conservation of cable bacteria. Therefore, we enriched the cable bacteria in the absence of an organic electron source with elemental sulfur and Fe(OH) 3 as sulfide scavenger or terminal electron acceptor. After four consecutive transfers, we performed substrate-turnover experiments with culture 1MN where we simulated the conditions that cells of the cable bacteria filaments might be facing along the geochemical gradients by adding sulfide, elemental sulfur, or thiosulfate as electron sources. In addition, we performed genome-resolved metagenomics of the enrichment culture 1MN and our cable bacteria enrichment culture and generated the first available, near complete genome (MAG Dsb_1MN) (Table S2 , Fig. S3 ) of a groundwater cable bacterium, of which we elucidated the genetic potential. Fig. 1 Microbial composition in the obtained enrichment cultures. a Maximum likelihood phylogenetic tree of full-length 16S rRNA gene sequences of Desulfobulbaceae retrieved from the NCBI database in comparison to the cable bacteria (MAG Dsb_1MN) from culture 1MN (red frame). Partial sequences from amplicon sequencing (OTU 1) and sequences from the metagenomes of culture 1MN (MAG Dsb_1MN) and the cable bacteria enrichment showed 100% similarity. Scale bar represents the number of substitutions per site. Known cable bacteria are represented by full-length 16S gene sequences of Candidatus Electrothrix and Candidatus Electronema. b Changes in microbial community composition of culture 1MN in the presence of different electron donor and acceptor combinations. The relative abundances of the MAG Dsb_1MN population and the Thermoanaerobacteraceae in the culture grown on 1-methylnaphthalene and ferrihydrite (top panel) were deduced from the average read coverage in the metagenome, which confirmed previous results obtained from fingerprinting by T-RFLP (14). The relative abundances in the absence of 1-methylnaphthalene were inferred from fingerprinting by T-RFLP and confirmed by amplicon sequencing (Fig. S6 ). c Fluorescence in situ hybridization (FISH) of the cable bacteria enrichment culture grown with elemental sulfur as electron donor and nitrate as electron acceptor stained with probe FliDSB194 specific for the MAG Dsb_1MN cable bacteria population. d Atomic force micrograph of filaments in culture 1MN grown with elemental sulfur and nitrate as electron acceptor showing the characteristic cell envelope of cable bacteria. The image displays the vertical deflection measured in contact mode",
"discussion": "Discussion In laboratory enrichment cultures as well as in contaminated aquifers, hydrocarbon-degrading organisms are frequently associated with highly abundant bacteria of the family Desulfobulbaceae closely related to groundwater cable bacteria [ 9 , 14 , 59 ]. We enriched groundwater cable bacteria originating from the iron-reducing, naphthalene-degrading culture 1MN to more than 90% in relative abundance, only with elemental sulfur as electron source and ferrihydrite as electron acceptor and sulfide scavenger. This supports the recent proposal that the Desulfobulbaceae might be involved in sulfur cycling during 1MN degradation in culture 1MN [ 14 ]. Specific FISH for groundwater cable bacteria revealed that the Desulfobulbaceae were present as several hundred µm long filaments, but also shorter filaments and single cells. This contrasts with findings for marine cable bacteria where to our knowledge no single-celled state was observed so far. Atomic force microscopy revealed the typical cable bacterial morphology with the continuous ridge pattern for our cable bacteria enrichment, similar to the originally discovered monophyletic cluster of cable bacteria 16S rRNA sequences from marine and freshwater [ 10 ]. Since the discovery of cable bacteria, it has been a major question how the cells in the middle of the filaments conserve energy because there is no visible reaction taking place in the suboxic zone of the geochemical gradient. Obvious reactions are only the sulfide oxidation at the anodic end and oxygen reduction at the cathodic end of the filaments. In our substrate-turnover experiments with the cable bacteria enrichment culture we simulated the conditions that cells in the cable bacteria filament are facing along the geochemical gradients. The results presented here provide clear evidence that cable bacteria can conserve energy by sulfur or thiosulfate disproportionation with FeCl 2 as sulfide scavenger (Fig. 2 , Fig. S1 ). In this case, energy could be conserved in all cells via substrate-level phosphorylation in the last step of a reverse sulfate reduction pathway, when adenosinephosphosulfate is converted to sulfate and ATP by a reverse operating sulfate adenylyltransferase (Fig. 3 , Table S3 ). We thus propose that the cable bacterial cells oxidize sulfide to elemental sulfur in a first step that is coupled by LDET to oxygen reduction or nitrate reduction to ammonium. The sulfur is then disproportionated by a reverse sulfate reduction pathway producing sulfate and sulfide. Hence, the role of LDET might be to provide elemental sulfur for the energy-conserving sulfur disproportionation. LDET thus mainly serves as an electron sink or acceptor for sulfide oxidation by cable bacteria but no energy can be conserved in this step. A similar mechanism has been demonstrated recently for Desulfurivibrio alkaliphilus [ 13 ]. Transcriptomics indicated that D. alkaliphilus oxidizes sulfide to elemental sulfur in a first step, which can then be either disproportionated or oxidized with nitrate as electron acceptor [ 13 ]. In contrast to D. alkaliphilus , our cable bacteria enrichment culture showed no sulfur disproportionation or oxidation of sulfide at concentrations higher than 300 µM indicating a thermodynamic or toxic inhibition of sulfur disproportionation by free hydrogen sulfide. Since this inhibition is complete and inhibiting energy conservation, the cable bacteria can also not slowly oxidize the sulfide to lower concentrations where it could start off with growth. At the slightly acidic pH of 6.4 during our substrate-turnover experiments with nitrate most of the sulfide was present as gaseous H 2 S, which can pass cell membranes [ 60 ] and consequently inhibit sulfur disproportionation. In contrast, at the alkaline pH during cultivation of D. alkaliphilus (>pH 9.5) almost all sulfide is present as HS − or S 2− , which cannot pass the cell membranes. This might be the reason why D. alkaliphilus can grow at higher sulfide concentrations, whereas our cable bacteria cannot [ 13 , 61 ]. Recently, three genomes of marine Ca . Electrothrix and one genome of Ca . Electronema have been published based on single-cell sequencing and metagenomics [ 11 ]. In the following, we provide an overview of the similarities and differences of these genomes to the genome MAG DSB_1MN of our cable bacteria. While the genome size of 3.1 Mbps of MAG DSB_1MN is within the range of 2.7–4.0 Mbps reported for other cable bacteria, MAG DSB_1MN has a clearly higher GC content of 57% compared with ~50% already distinguishing MAG DSB_1MN from other cable bacteria. MAG DSB_1MN has several genes which might have been lost, reduced, or replaced in other cable bacteria such as the glycolytic enzyme enolase, a complete DsrKMJOP complex, and the NADH-quinone oxidoreductase (Nuo) enzyme complex (Table S3 ) [ 11 ]. Like in other cable bacteria and in D. alkaliphilus , an SQR might oxidize sulfide to elemental sulfur and the sulfate reduction pathway might be operated in reverse for energy conservation. No reverse-type dissimilatory sulfite reductase was observed, which is in accordance to other cable bacteria, D. alkaliphilus , and also other sulfur disproportionating Desulfobulbaceae such as D. propionicus . Kieldsen et al. [ 11 ] suggested energy conservation by sulfur disproportionation by a polysulfide reductase when cable bacteria are disconnected from electron acceptors. So far, we were not able to detect genes encoding for this enzyme in MAG DSB_1MN. One of the main questions since the discovery of cable bacteria is about the composition of the electron conductor. Based on metagenomic and proteomic data, Kieldsen et al. hypothesized electrically conductive type IV pili (e-pili) might form conductive superstructures in the periplasm. Our genomic data of MAG DSB_1MN also allow for this possibility, since we also found the gene coding for PilA in the genome. The amino acid sequence shows the same distribution of aromatic amino acids like electrically conductive e-pili (Fig. S10 ) [ 58 ]. Interestingly, genome analysis revealed genes for hydrogenases indicating the potential of MAG DSB_1MN to use hydrogen as alternative electron donor (Fig. 3 , Table S3 ). Hydrogen might be an alternative electron source for cable bacteria in organic-rich habitats dominated by fermentation. However, this is in contrast to the genomes of marine and freshwater cable bacteria where a cytoplasmatic hydrogenase was detected only in Ca. E. aarhusiensis and periplasmatic hydrogenases were absent [ 11 ]. The presence of a complete Wood–Ljungdahl pathway for CO 2 fixation, which is in accordance to previously published genomes [ 11 ], and the absence of an organic C-source in our enrichment culture strongly indicates the capability of MAG DSB_1MN of a chemo-litho-autotrophic metabolism. Our cable bacteria enrichment culture was also capable of nitrate reduction to ammonium, which was confirmed by genes encoding for nitrate and nitrite reductases in genome MAG DSB_1MN. Although we did not test for oxygen as electron acceptor, genes encoding for cytochrome bd oxidase indicate that these organisms can reduce oxygen as terminal electron acceptor (Fig. 3 , Table S3 ). Nevertheless, groundwater cable bacteria showed oxygen reduction in laboratory incubations of aquifer sediments [ 9 ]. Intriguingly, genes encoding for a cytochrome bd oxidase for oxygen as electron acceptor were absent in the genomes of Ca . Electrothrix and Ca . Electronema [ 11 ]. These results allow us to suggest a new model for energy conservation of cable bacteria, which provides an explanation of how each cell within the cable bacterial filament can conserve energy (Fig. 4 ). Near the surface, cable bacteria perform the cathodic reaction, i.e., the reduction of oxygen and nitrate to water and ammonium, respectively. So far, it is unclear if cable bacteria conserve energy from oxygen reduction. For instance, closely related species such as Desulfobulbus propionicus can reduce oxygen but show no growth with oxygen as electron acceptor [ 62 ]. We propose that below the cathodic zone elemental sulfur is disproportionated to sulfate and sulfide, whereas the sulfide is again oxidized to sulfur by LDET. The elemental sulfur can either be produced abiotically by fluctuating redox conditions or by a LDET by the cable bacteria themselves. Hence, the apparent overall reaction at the anodic part of the filaments is a net oxidation of sulfide to sulfate but energy is most likely conserved by sulfur disproportionation only. In natural sediments, chemo-organo-heterotrophic, sulfate-reducing bacteria will be abundant all along the cable bacteria filament and oxidize organic material with concomitant reduction of sulfate to sulfide (Fig. 4 ) [ 63 ]. We propose that all cells of the cable bacteria can oxidize this sulfide to elemental sulfur by LDET and the electrons are channeled through the cable filaments to the oxygen- or nitrate-reducing cathodic end. In fact, this pathway provides an explanation for energy conservation throughout the entire filament. Fig. 4 Conceptual model for energy conservation in groundwater cable bacteria. a , b Filaments span the suboxic zone by a long-distance electron transfer. Within the suboxic zone and the anodic zone, sulfide is oxidized to elemental sulfur, which is then used to conserve energy by sulfur disproportionation via a reverse sulfate reduction pathway. At the cathodic end, oxygen or nitrate reduction take place as electron accepting process for the LDET. Sulfide is provided all along the filament by sulfate-reducing bacteria. The sulfate is recycled by the cable bacteria providing a cryptic sulfur cycle in the suboxic zone. IM inner membrane; OM outer membrane; EC electric conductor The energy-conserving sulfur disproportionation reaction requires low sulfide concentrations [ 64 ]. This suggests that in sediments the anodic oxidation of sulfide is limited to the suboxic zone and a narrow zone at the measurable end of the sulfide gradient (Fig. 4 ), which is characterized by low concentrations but high fluxes of sulfide. Hence, the functioning of cable bacteria relies on a delicate equilibrium between the rate of electron removal by LDET (and consequent oxygen or nitrate reduction rates) and the sulfide reduction rates by sulfate reducers (Fig. 4 ). Either a decrease of LDET, by, e.g., lower oxygen supply, or higher sulfate reduction rates could lead to increased sulfide concentrations along the filaments and immediate inactivation of the cable bacteria function. This might explain the frequently observed sudden disappearance of cable bacteria populations and LDET in marine sediments [ 65 , 66 ]."
} | 4,443 |
39372857 | PMC11449778 | pmc | 675 | {
"abstract": "Introduction Understanding the microbial diversity and potential functional dynamics within the rhizocompartments of Dendrobium huoshanense is crucial for unraveling the plant–microbe interactions that influence its medicinal properties. Methods This study is the first to characterize the microbiome associated with the rhizocompartments of D. huoshanense , including its cultivation medium, rhizosphere, rhizoplane, and root endosphere, using high-throughput sequencing and subsequent bioinformatic analysis. Results Bacterial phylogenetic diversity was significantly higher in the endosphere than in the rhizosphere, while fungal α-diversity significantly decreased from the cultivation medium to the endosphere. Both bacterial and fungal niche widths decreased from the cultivation medium to the endosphere. β-Diversity analysis revealed distinct spatial patterns in both bacterial and fungal communities across the rhizocompartments, with the most pronounced differences between the cultivation medium and the endosphere. Taxonomically, Proteobacteria and Ascomycota were predominant in the endosphere for bacterial and fungal communities, respectively. Functional predictions showed significant enrichment of pathways related to xenobiotics biodegradation, lipid metabolism, and nitrogen fixation in the endosphere, while functions associated with plant pathogens and saprotrophs were significantly reduced. Discussion The results indicate a shift from generalist to specialist microbes from the cultivation medium to the endosphere, suggesting that D. huoshanense exerts strong selective pressure for endophytic fungi. Interestingly, a high proportion of fungi with unknown functions were found in the endosphere, highlighting an area for further research regarding the medicinal efficacy of D. huoshanense . Overall, this study provides foundational data for understanding the adaptive evolution of these microbial communities in response to specific microhabitats.",
"conclusion": "5 Conclusions In summary, our study represents the first comprehensive characterization of the microbiome associated with the rhizocompartments of D. huoshanense using high-throughput sequencing. Our findings revealed several key insights: first, fungal α-diversity exhibits a significant decrease from the cultivation medium to the root endosphere, while bacterial α-diversity peaks in the endosphere. Moreover, both bacterial and fungal community niche widths contract from the cultivation medium to the endosphere, indicating a shift toward more specialized microbial communities. Second, the distinct structures and taxonomic compositions observed in the bacterial and fungal communities within the roots, compared to other rhizocompartments, highlight adaptive evolution toward specific microhabitats. Lastly, we observed a notable reduction in functions associated with plant pathogens and saprotrophs within the root microbiome, indicating strong selective pressures favoring endophytic fungi beneficial to D. huoshanense . Conversely, the endosphere is enriched with fungi of unknown function, underscoring the need for further investigation into their potential roles in influencing the medicinal properties of D. huoshanense . Overall, our study not only provides novel insights into the microbial ecology of D. huoshanense but also underscores the complexity and specificity of its endophytic microbiome.",
"introduction": "1 Introduction \n Dendrobium , a genus within the Orchidaceae family, includes plants that are widely utilized in Chinese traditional medicine for their broad medicinal properties. These benefits encompass gastrointestinal nourishment, yin nourishment, heat clearance, lung moisture enhancement, cough relief, vision improvement, and body fortification ( Shun et al., 2017 ; Gao et al., 2022 ). Consequently, Dendrobium is frequently consumed fresh as a dietary supplement or incorporated into traditional Chinese medicinal preparations. Modern pharmacological research has further unveiled additional benefits, including anti-aging, immune enhancement, anti-tumor effects, anti-atherosclerotic effects, and alleviation of rheumatoid arthritis ( Xie et al., 2019 ; Fan et al., 2020 ; Shang et al., 2021 ; Ye et al., 2023 ). Huoshan County in Anhui Province, China, is noted as the earliest recorded habitat for Dendrobium in ancient herbal texts ( Shun et al., 2017 ). The species Dendrobium huoshanense C.Z.Tang et S.J.Cheng, identified by Tang and Cheng (1984) , is endemic to China, and its inclusion in the Chinese Pharmacopoeia in 2020 underscores its importance. Recognized as one of the superior species within the Dendrobium genus, D. huoshanense is prominent among the “Nine Immortal Herbs of China” and is renowned for its rich array of bioactive compounds, including polysaccharides, flavonoids, alkaloids, amino acids, phenols, and terpenoids, which contribute to its pharmacological efficacy ( Li et al., 2020 ; Liu et al., 2021a ; Gao et al., 2022 ). The growth, development, and defense against pests and diseases of medicinal plants are intricately linked to their habitats and the endophytic microorganisms, particularly those residing in the rhizosphere ( Jain et al., 2020 ; Ling et al., 2022 ). These interactions between plants and their associated microbiota are crucial for plant growth, quality, and overall health ( Morgan et al., 2005 ; Köberl et al., 2013 ; Xu et al., 2023 ). Consequently, the rhizospheric microbiota are often referred to as a plant’s second genome. In addition to rhizospheric microbiota, endophytic fungi are essential symbiotic partners for orchids, significantly affecting the host plants through their life activities and secondary metabolites ( Deng and Cao, 2017 ). Studies have suggested that the microbial community associated with Dendrobium , including both rhizospheric and endophytic components, plays a pivotal role in determining its quality. For example, Sphingomonas paucimobilis ZJSH1 has been shown to significantly promote the growth of Dendrobium catenatum and the accumulation of active polysaccharides ( Yang et al., 2014 ; Li et al., 2023 ). Additionally, endophytic fungi isolated from Dendrobium exhibit antimicrobial and anti-inflammatory activities ( Yue and Yu, 2022 ), with some fungi having the potential to enhance the growth of D. huoshanense or act as antagonists to plant pathogens ( Chen et al., 2019 ). Traditional microbial culture methods are limited by the inability to isolate and purify more than 95% of microorganisms ( Amann et al., 1995 ; Rappé and Giovannoni, 2003 ), resulting in incomplete and often biased microbial profiles. In recent years, high-throughput sequencing technologies for prokaryotic 16S rRNA genes and fungal internal transcribed spacer (ITS) genes have emerged as critical tools for studying plant microbiomes. For instance, high-throughput sequencing has revealed higher diversity in root-associated bacteria and fungi in D. huoshanense compared to stems, with notable differences in bacterial diversity across different growth years ( Chen et al., 2020 ). Liu et al. (2021b) employed high-throughput sequencing to investigate the diversity and composition of mycorrhizal fungi associated with three types of 1-year-old D. huoshanense , uncovering rich fungal diversity and significant differences among the types of D. huoshanense . Plants are known to recruit specific groups of microorganisms to their rhizosphere, with only a subset penetrating the roots via the rhizoplane. These spatially heterogeneous microbial communities are shaped by complex plant–microbe interactions ( Bulgarelli et al., 2013 ; Edwards et al., 2015 ; Attia et al., 2022 ). However, very little is known about the differences in the rhizosphere, rhizoplane, and root endophytic microbiomes of D. huoshanense and their potential functions, which severely limits our understanding of the interaction mechanisms between D. huoshanense and microorganisms. To address these research gaps, this study systematically investigated the root-associated microbiome of D. huoshanense across four distinct rhizocompartments—the cultivation medium, rhizosphere, rhizoplane, and endosphere ( \n Figure 1 \n )—using high-throughput sequencing technology. The primary objectives were to elucidate the characteristics of bacterial and fungal diversity, observe community structure changes across these rhizocompartments, and identify potential functional differences among the bacterial and fungal communities. These findings will contribute to a deeper understanding of interactions between medicinal plants, rhizospheric microorganisms, and endophytic microorganisms, thereby enriching the theoretical foundation for traditional medicinal plant–microbe interactions. Figure 1 Sampling strategy diagram. Samples were taken at four compartments (different spatial categories), i.e., cultivation medium, rhizosphere, rhizoplane, and endosphere.",
"discussion": "4 Discussion 4.1 Bacterial community diversity and composition across rhizocompartments of D. huoshanense \n Our results demonstrated an increasing phylogenetic diversity of bacteria from the rhizosphere to the root endosphere of D. huoshanense ( \n Figure 2 \n ). The bacterial α-diversity pattern across rhizocompartments of D. huoshanense differs from that observed in other plants. For instance, the characterization of the root microbiome of Arabidopsis revealed that bacterial diversity inside the root (the endosphere) was much lower compared to the soil around the root (the rhizosphere) ( Schlaeppi et al., 2014 ). Similarly, bacterial richness in the rhizoplane and rhizosphere of wheat and faba bean was significantly lower than that in the soil ( Attia et al., 2022 ). The richness of bacteria in the endosphere of roots, stems, and leaves of poplar trees was also significantly lower than that in rhizosphere soil ( Beckers et al., 2017 ). These variations in root-associated bacterial diversity patterns may be related to the plant’s growth environment. The artificial cultivation medium of D. huoshanense in this study differs from natural soil, potentially resulting in lower bacterial diversity in the cultivation medium due to lower concentrations of nutrients. In contrast, the relatively abundant nutrients at the root surface and interior may lead to increased bacterial diversity. Additionally, the niche width of bacterial communities within the roots of D. huoshanense was significantly narrower than that in the growth medium, rhizosphere, and rhizoplane ( \n Figure 2C \n ). This indicates a gradual shift from generalists to specialists as bacteria transition from the medium to the root interior, adapting to the increasingly specific microenvironment ( Lambers et al., 2009 ; Beckers et al., 2017 ). Regarding bacterial community composition, the relative abundance of Proteobacteria was significantly enriched in the endophytic environment of D. huoshanense roots ( \n Figure 5A \n ). This finding is consistent with the results of Chen et al. (2020) . At the genus level, Devosia was also significantly enriched in the endosphere of D. huoshanense ( \n Figure 5A \n , \n Supplementary Figure 2A \n ). Devosia consists of rod-shaped, motile, gram-negative soil bacteria capable of symbiotically fixing atmospheric nitrogen ( Hungria and Nogueira, 2023 ), decomposing urea, and producing various volatile organic compounds, which contribute to soil microflora diversity and mediate microbe–microbe interactions ( Schenkel et al., 2015 ). Genomic analysis further revealed Devosia ’s ability to sense environmental signals and exhibit chemotaxis in stressed habitats ( Talwar et al., 2020 ). The enrichment of Devosia in the endosphere of D. huoshanense may positively affect the plant’s growth and adaptation to stressed environments. Another notable feature in the community composition is that the relative abundance of Acidobacteriota within the roots of D. huoshanense is much lower than in other rhizocompartments, especially in the rhizosphere ( \n Figure 4 \n ). Many Acidobacteriota are acidophilic and abundant in soil habitats ( Kielak et al., 2016 ). The high abundance of Acidobacteriota in the cultivation medium, rhizosphere, and rhizoplane observed in this study may be related to the low pH value of the growth substrate for D. huoshanense ( \n Supplementary Table S1 \n ). 4.2 Fungal community diversity and composition across rhizocompartments of D. huoshanense \n The richness and phylogenetic diversity of fungi within the roots of D. huoshanense are significantly lower than those in other rhizocompartments ( \n Figure 2B \n ). Studies have shown that fungal diversity decreases progressively from the bulk soil to the root interior ( Mendes et al., 2013 ; Hardoim et al., 2015 ; Bahram et al., 2018 ). This gradient is shaped by nutrient availability, environmental conditions, and the selective pressure exerted by plant roots. The bulk soil supports a rich and diverse fungal community due to its complexity and variety of niches ( Tedersoo et al., 2014 ). In contrast, the rhizosphere and rhizoplane select fungi that can utilize root exudates and adhere to root surfaces, respectively ( Berendsen et al., 2012 ; Li et al., 2021 ). Finally, the endosphere has the lowest fungal diversity because it is inhabited only by specific fungi capable of overcoming plant defenses and establishing symbiotic or endophytic relationships with their hosts ( Hardoim et al., 2015 ; Qian et al., 2019 ; Sonam and Liu, 2024 ). Additionally, the niche width of fungal communities within the roots of D. huoshanense was significantly narrower than that in the growth medium, rhizosphere, and rhizoplane ( \n Figure 2D \n ). This also reflects the adaptation of specialist fungi to the increasingly specialized microenvironment ( Huang et al., 2022 ; Sonam and Liu, 2024 ). Regarding fungal community composition, the most significant feature is the substantial increase in the relative abundance of Ascomycota from the cultivation medium to the interior of D. huoshanense roots, while the relative abundance of Basidiomycota significantly decreases ( \n Figure 4B \n ). The enrichment of Ascomycota within the endosphere of D. huoshanense is mainly due to the high abundance of an unclassified genus (labeled “a” in \n Figure 5B \n ; \n Supplementary Figure 2B \n ). This fungal genus accounts for up to 62.2% of the endosphere, whereas its proportion in the other three rhizocompartments does not exceed 6.0%. Due to the lack of taxonomic information and culturable strains, the function and ecological role of this fungal taxon remain to be further studied. Although fungal community composition varies across different plant-associated environments, bulk soil and the rhizosphere are dominated by Ascomycota, Basidiomycota, and Zygomycota ( Berendsen et al., 2012 ; Tedersoo et al., 2014 ; Bahram et al., 2018 ). In contrast, the endosphere is primarily dominated by Ascomycota, with fewer Basidiomycota ( Hardoim et al., 2015 ). Fusarium spp. were enriched in the rhizoplane and endosphere ( \n Supplementary Figure 2B \n ). They are filamentous fungi commonly found in soil and plant debris ( Fravel et al., 2003 ). Most of them are harmless saprobes, although some species can be pathogenic and produce mycotoxins ( Ma et al., 2010 ). Ecologically, they play a role in decomposing organic matter and nutrient cycling ( Chen et al., 2014 ). Some species have beneficial interactions, such as promoting plant growth or acting as biocontrol agents against other pathogens ( Nikitin et al., 2023 ). 4.3 Bacterial and fungal community structures across rhizocompartments of D. huoshanense \n Bacterial and fungal community structures (β-diversity) of D. huoshanense in the rhizosphere and rhizoplane showed no significant differences ( \n Figure 3 \n ). This may be attributed to the cultivation medium of D. huoshanense , which mainly consists of pine bark. This environment differs from soil environments, resulting in smaller differences in the microenvironments of the rhizosphere and rhizoplane of D. huoshanense , leading to minimal spatial differentiation. However, the bacterial and fungal community structures of the rhizosphere and rhizoplane of D. huoshanense differed significantly from those in the cultivation medium and the internal root environment of D. huoshanense , indicating that different microhabitats have a substantial impact on microbial community structure. This finding is consistent with the results reported by Sonam and Liu (2024) . Additionally, there were considerable differences in the endophytic community structures within the roots of different D. huoshanense plants ( \n Figure 3 \n ), suggesting that environmental selection by individual plants strongly determines local microbial communities of the plant endosphere ( Attia et al., 2022 ). 4.4 Microbial functional differences across rhizocompartments of D. huoshanense \n Bacterial functional analysis indicates that metabolic processes related to xenobiotics, lipids, and functions associated with nutrient acquisition (e.g., nitrogen fixation) are significantly enriched in the D. huoshanense endosphere ( \n Figure 6 \n ). The enrichment of these functions may be causally related to the significant enrichment of membrane transport, symbionts, and chemoheterotrophy ( \n Figure 6 \n ). This suggests that D. huoshanense may selectively enrich bacterial groups with these functions through interactions with environmental bacteria, thereby promoting its own growth, enhancing specific metabolic pathways, and facilitating the formation of bioactive macromolecules ( Yang et al., 2014 ; Yu and Hochholdinger, 2018 ; Li et al., 2023 ; Ravelo-Ortega et al., 2023 ). For example, the enrichment of terpenoid alkaloid metabolic functions in the endosphere of D. huoshanense ( \n Figure 6 \n ) may related to the high content of medicinally active alkaloid metabolites in D. huoshanense ( Wu et al., 2022 ). FungalTraits analysis indicated that functions associated with plant pathogens and saprotrophs were significantly depleted in the endosphere, while these functions were abundant in the growth medium, rhizosphere, or rhizoplane ( \n Figure 6C \n ). These results suggest that the roots of D. huoshanense serve as a crucial barrier in interactions with environmental fungi ( Li et al., 2021 ). This barrier restricts the entry of pathogens into the roots and prevents hyper-decomposing saprophytic fungi from bypassing the root barrier, effectively protecting D. huoshanense from both pathogenic and saprophytic fungi. For instance, the relative abundance of Alternaria in the external root environment (from the growth medium to the rhizoplane) ranges from 6.3% to 15.3%, whereas in the endosphere, it is only approximately 0.1% ( \n Supplementary Figure 2 \n ). Alternaria , a genus of Deuteromycetes fungi, includes species that are major plant pathogens and common agents of decay and decomposition, as well as producers of mycotoxin ( Thomma, 2003 ; Du et al., 2023 ). In this study, up to 87% of fungal sequences within the roots remain functionally uncharacterized. This greatly limits our understanding of the structure and function of the D. huoshanense holobiome ( Hardoim et al., 2015 ). Future research should employ advanced methods such as metagenomics, metabolomics, and culturomics ( Rinke et al., 2013 ; Mhlongo et al., 2018 ; Matar and Bilen, 2022 ) to further investigate the ecological functions of dominant microbial groups in the rhizosphere and within the roots."
} | 4,903 |
34857934 | PMC8941099 | pmc | 676 | {
"abstract": "Efficient nutrient cycling in the coral-algal symbiosis requires constant but limited nitrogen availability. Coral-associated diazotrophs, i.e., prokaryotes capable of fixing dinitrogen, may thus support productivity in a stable coral-algal symbiosis but could contribute to its breakdown when overstimulated. However, the effects of environmental conditions on diazotroph communities and their interaction with other members of the coral holobiont remain poorly understood. Here we assessed the effects of heat stress on diazotroph diversity and their contribution to holobiont nutrient cycling in the reef-building coral Stylophora pistillata from the central Red Sea. In a stable symbiotic state, we found that nitrogen fixation by coral-associated diazotrophs constitutes a source of nitrogen to the algal symbionts. Heat stress caused an increase in nitrogen fixation concomitant with a change in diazotroph communities. Yet, this additional fixed nitrogen was not assimilated by the coral tissue or the algal symbionts. We conclude that although diazotrophs may support coral holobiont functioning under low nitrogen availability, altered nutrient cycling during heat stress abates the dependence of the coral host and its algal symbionts on diazotroph-derived nitrogen. Consequently, the role of nitrogen fixation in the coral holobiont is strongly dependent on its nutritional status and varies dynamically with environmental conditions.",
"introduction": "Introduction The association with microbial symbionts is central to the ecological success of reef-building corals in the oligotrophic tropical ocean [ 1 , 2 ]. The close metabolic coupling between heterotrophic corals and their phototrophic algal symbionts supports their immense productivity, has given rise to their rapid radiation, and has led to the formation of coral reef ecosystems [ 3 – 7 ]. Consequently, the destabilization of this symbiosis in times of anthropogenic environmental change is posing a direct threat to the functioning of coral holobionts, i.e., the ecological unit of corals and their associated microorganisms, and the reefs they support [ 8 – 10 ]. In recent years, ocean warming has repeatedly caused mass coral bleaching, the breakdown of the coral-algal symbiosis, frequently followed by the death of the coral host and subsequent ecosystem-wide reef degradation [ 11 , 12 ]. An in-depth understanding of the processes underlying the functioning and destabilization of the coral-algal symbiosis is thus required to predict or mitigate the effects of climate change on coral reefs [ 13 – 16 ]. Efficient recycling of organic and inorganic carbon in the coral-algal symbiosis depends on the nutrient-limited state of the algal symbionts [ 17 – 19 ]. In a stable state of the symbiosis, low bioavailable nitrogen availability limits algal growth and results in the accumulation of photosynthates in algal cells [ 20 , 21 ]. The release of these excess nutrients, in turn, fuels the energy metabolism of the coral host. Although low levels of nitrogen assimilation are required to support the holobiont’s net productivity and growth [ 22 ], increases in the coral host’s catabolic activity during heat stress or other environmental stressors may increase nitrogen availability for the algal symbionts. Excess nitrogen availability may destabilize nutrient recycling in the coral-algal symbiosis resulting in a breakdown of coral holobiont functioning and eventually host starvation [ 15 , 23 – 26 ]. The stability and functioning of the coral holobiont under changing environmental conditions thus depend on its ability to maintain a limited nitrogen availability for the algal symbionts [ 16 , 25 ]. Importantly, nitrogen assimilation in the coral holobiont is not limited to the coral host and its algal symbionts. In addition to heterotrophic feeding and the uptake of nutrients from seawater, most coral holobionts also show detectable rates of nitrogen fixation, i.e., the prokaryotic conversion of atmospheric dinitrogen into bioavailable ammonium [ 27 – 29 ]. Diazotrophs, Bacteria, and Archaea, capable of nitrogen fixation, are common coral associates and show indications of host specificity [ 30 – 34 ]. As the abundance and activity of diazotrophs are highly dynamic and influenced by environmental change [ 35 – 39 ], nitrogen fixation may either stabilize or destabilize coral holobiont functioning depending on prevailing environmental conditions [ 25 ]. During periods of low environmental nitrogen availability, nitrogen fixation activity is positively correlated with coral holobiont productivity and could fulfill up to 11% of the algal symbionts’ nitrogen requirement [ 38 ]. In this context, the stimulation of nitrogen fixation rates by temperature stress or eutrophication has been proposed to enhance the productivity and resilience of coral holobionts during stress [ 40 – 42 ]. In addition, stimulated nitrogen fixation and increased nitrogen to phosphorus ratios in the holobiont have been reported during the breakdown of the coral-algal symbiosis under elevated sugar concentrations [ 43 ]. The importance of diazotrophs, thus, seems to depend on the fate of fixed nitrogen in the coral holobiont and its effects on other holobiont members [ 44 ]. However, our understanding of the environmental drivers controlling nitrogen fixation and the assimilation of diazotroph-derived nitrogen in the coral holobiont remains largely speculative at this point. Recent studies showed that increased metabolic energy demands and the enhanced release of catabolic waste products may severely alter coral holobiont nutrient cycling during heat stress [ 15 , 45 ]. We hypothesized that under such conditions, stimulated nitrogen fixation activity would enhance nitrogen availability for algal symbionts and thus contribute further to the destabilization of the coral-algal symbiosis. To test this, here we assessed the effects of heat stress on nitrogen fixation in the coral Stylophora pistillata from the central Red Sea. Combining amplicon sequencing, nutrient flux incubations, isotope labeling, and NanoSIMS imaging, we aimed to assess how heat stress affects diazotroph community structure, nitrogen fixation activity, and the assimilation of diazotroph-derived nitrogen in the coral holobiont before the onset of bleaching.",
"discussion": "Results and discussion Diazotroph community composition varies with temperature To assess the role of diazotrophs in the early response of the coral holobiont to heat stress, we sampled the colonies of S. pistillata on day 10 of the experiment (7 days at a maximum temperature of 32.9 °C). At this time point, corals from heat stress and control treatments maintained a healthy appearance, similar levels of algal symbiont densities, and showed no visual signs of bleaching (for an extended discussion of processes and timeframes of bleaching during the experiment, please refer to Rädecker et al. [ 15 ]). The early heat-stress response did not affect overall bacterial community composition (ANOSIM, p = 0.145; Fig. S1 ). Members of the order Oceanospirillales, which are prevalent in healthy corals and S. pistillata in particular [ 48 , 71 , 72 ], represented the largest component of the coral microbiomes from both treatments (~20% of 16S rRNA sequences). In contrast to the stable overall bacterial microbiome composition, nifH sequencing revealed that heat stress caused a distinct shift in the community composition of diazotrophs (ANOSIM, p = 0.049). This shift was not driven by the loss or the recruitment of novel taxa but rather by variations in the relative abundance of taxa already present in the holobiont (Fig. 1A, B ). The diazotroph community was largely dominated by the orders Alteromonadales and Chroococcales accounting for 65% and 27% of nifH sequences in colonies from the ambient control, respectively. Heat stress, however, caused a significant decline in the relative abundance of Chroococcales (LEfSe, χ 2 = 3.94, p = 0.047) accompanied by the proportional (albeit not significant) increase of Alteromonadales in the diazotroph community (LEfSe, χ 2 = 2.46, p = 0.117) (Fig. 1B ). Importantly, Chroococcales were previously identified as endosymbionts in the epithelium of the Caribbean coral Montastrea cavernosa [ 73 ] and have been proposed to be an important source of fixed nitrogen for algal symbionts in these corals [ 74 ]. Consequently, the observed shifts in diazotroph community composition could directly affect nitrogen availability for other holobiont members during heat stress. Fig. 1 Characterization of coral-associated diazotroph communities after 10 days of heat stress. A Relative diazotroph community composition of individual colonies under control (left) and heat stress (right) conditions based on nifH amplicon sequencing. Notably, no archaeal diazotrophs could be detected in the present study. B Change in the relative abundance of dominant diazotroph orders during heat stress relative to control conditions (mean ± SE). Asterisks indicate a significant change ( p < 0.05) under heat stress compared to control corals. Algal symbionts assimilate diazotroph-derived nitrogen in a stable coral holobiont Using the acetylene reduction technique, we estimated daily nitrogen fixation rates in S. pistillata holobionts. Under ambient control conditions (i.e., 29.1 °C), coral holobionts showed detectable rates of nitrogen fixation around 1.0 ± 0.8 nmol N 2 cm −2 day −1 in line with rates previously reported for corals from this region (Fig. 2A ) [ 31 , 75 , 76 ]. Further, 15 N 2 isotope labeling resulted in enriched δ 15 N values in the soft tissue of coral holobionts (mean ± SE = 16.0 ± 4.7‰), which positively correlated with nitrogen fixation rates across coral colonies (Pearson’s correlation, r = 0.99, p = 0.002, Fig. 2B, C ). Although these nitrogen fixation rates may be relatively low compared to other nutrient sources, they are a non-negligible component of the overall nutrient cycling in the nitrogen-limited S . pistillata holobiont. Fig. 2 Nitrogen fixation activity and assimilation in the coral holobiont after 10 days of heat stress. A Nitrogen fixation activity of coral holobionts ( n = 5 per treatment) was quantified via the acetylene reduction assay. B Net assimilation of diazotroph-derived nitrogen in the coral holobiont ( n = 5 per treatment) quantified by 15 N 2 isotope labeling and bulk stable isotope analysis. C Nitrogen fixation activity and assimilation of diazotroph-derived nitrogen show a positive correlation across colonies (symbols) under control (blue) and heat stress conditions (red) with corresponding confidence intervals (gray). Pearson’s correlation coefficients ( r ) correspond to correlations among each condition of the corresponding color. Samples from both control and heat stress conditions of the colony with the highest enrichment (ellipse) were imaged in detail using the NanoSIMS ion microprobe. Bars and error bars indicate mean ± SE. Asterisks indicate significant differences between treatments ( p < 0.05). To identify which holobiont partner(s) assimilated diazotroph-derived nitrogen in the coral holobiont, we used NanoSIMS imaging to trace the incorporation of the 15 N 2 isotope marker in the coral colony with the highest nitrogen fixation activity (Fig. 3A–D ). Overall, the algal symbionts showed significantly higher levels of 15 N enrichment than the coral host tissue (118.7 ± 9.4‰ (symbionts), 9.5 ± 1.6‰ (host); Tukey’s honestly significant difference (HSD), p < 0.001, Fig. 3F ). These observations are consistent with previous reports for S. pistillata from the Great Barrier Reef and show that a large fraction of diazotroph-derived nitrogen is eventually assimilated by the algal symbionts [ 33 ]. In addition, NanoSIMS images revealed the presence of small 15 N enrichment hotspots in the host epithelium that displayed two distinct morphologies: oval compartments/cells 2–3 µm in length and clusters of smaller circular and rod-shaped compartments/cells with 1–3 µm in length (Fig. 3B, C ). Although the NanoSIMS images do not allow us to elucidate the identity of these hotspots, their shape and location are broadly consistent with previous reports of the presence of endosymbiotic bacteria in the coral holobiont [ 73 , 77 – 79 ]. Specifically, the larger oval hotspots resemble observations of Chroococcales in the epithelium of M. cavernosa , whereas the rod-shaped clusters resemble previously described bacterial aggregates in the coral tissue in their structure and localization [ 73 , 77 – 79 ]. It is thus plausible that the observed epithelial 15 N hotspots represent endosymbiotic diazotrophs in the tissue of S. pistillata . However, if these 15 N hotspots are indeed diazotrophs, they are not situated near the algal symbionts that are the primary sink of diazotroph-derived nitrogen. In many other phototroph–diazotroph symbioses, diazotrophs predominantly release fixed nitrogen via the passive diffusion of ammonium [ 80 ]. Although the ways of nitrogen release by coral-associated diazotrophs are currently unknown, it is plausible that passive ammonium release and/or the catabolic breakdown of diazotrophs during host digestion may contribute to the inorganic nitrogen pool in the holobiont. If this is the case in S. pistillata , the uneven partitioning of diazotroph-derived nitrogen between the coral host and its algal symbionts likely reflects the respective metabolic demands of symbiotic partners for external nitrogen sources. In this scenario, the nitrogen-limited state of algal symbionts might support strong concentration gradients in the host tissue, which would enable efficient assimilation of exogenous nitrogen (derived from either diazotrophs or from the surrounding seawater) [ 81 ]. In a stable holobiont state, diazotrophs thus provide a nitrogen source to support algal symbiont growth and allow for net productivity of the coral holobiont (Fig. 4A ). Fig. 3 NanoSIMS imaging of 15 N 2 assimilation in the coral holobiont after 10 days of heat stress. A Mosaic of hue saturation images of 15 N 12 C − / 14 N 12 C − ratios illustrating the spatial distribution of 15 N enrichment in the coral tissue under control conditions. B , C Enlarged 14 N 12 C − NanoSIMS images of epithelial 15 N 2 hotspots highlighted in A . Epithelial hotspots were primarily characterized by two distinct morphologies: oval compartments/cells with 2–3 µm length (yellow circles) and clusters of smaller rod-shaped compartments/cells with 1–3 µm in length (white circles). D , E Hue saturation images of 15 N 12 C − / 14 N 12 C − ratios illustrating 15 N assimilation in the coral host and algal symbionts under control ( D ) and heat stress ( E ) conditions. F \n 15 N 2 assimilation in the coral tissue and algal symbionts under control (blue) and heat stress (red) conditions based on NanoSIMS images ( n = 15 per treatment). Differing letters above the boxplot indicate significant differences between groups ( p < 0.05). All scale bars are 10 µm. Fig. 4 Conceptual view of the effect of heat stress on nitrogen assimilation in the coral holobiont. A In a stable state, low environmental nitrogen availability limits nitrogen uptake in the coral holobiont. Under these conditions, diazotroph-derived nitrogen provides an important nitrogen source for the coral-algal symbiosis. B During heat stress, the release of excess nitrogen waste from the host metabolism increases nitrogen availability in the coral holobiont. Under these conditions, the coral-algal symbiosis does not depend on diazotroph-derived nitrogen. The color of arrows indicates the proportion of diazotroph-derived nitrogen in the nitrogen flux/pool: continuous scale ranging from “low contribution of diazotroph-derived nitrogen (dark blue)” on one end of the spectrum to “high contribution of diazotroph-derived nitrogen (pink)” on the other end of the spectrum. The line width of arrows in the heat stress ( B ) indicates their proportional increase relative to control conditions ( A ). Reduced assimilation of fixed nitrogen by algal symbionts despite increased nitrogen fixation activity during heat stress On day 10 of the experiment, heat stress caused a 74% increase in nitrogen fixation rates (indirectly quantified as acetylene reduction) compared to coral holobiont rates from ambient control conditions (sign test, z = 2.24, p = 0.025, Fig. 2A ). Similar increases in coral-associated nitrogen fixation during heat stress have been previously reported for other coral species [ 40 , 41 ], suggesting that the here reported processes are unlikely restricted to the S. pistillata holobiont, but may be a common feature of coral-associated diazotroph communities. 15 N 2 isotope labeling in this study revealed that the additional fixed nitrogen was not assimilated in the soft tissue of the holobiont, in which 15 N enrichment decreased by 30% compared to ambient controls (sign test, z = 2.24, p = 0.025, Fig. 2B ). NanoSIMS analysis corroborated this observation, revealing that the 15 N enrichment remained stable in the host tissue (Tukey’s HSD, p = 0.976) but declined by 53% in the algal symbionts compared to ambient controls (Tukey’s HSD, p < 0.001, Fig. 3E, F ). Furthermore, we did not detect any epithelial 15 N enrichment hotspots in corals exposed to heat stress, suggesting lower abundances of diazotrophs and/or lower nitrogen fixation activity. The overall increase in nitrogen fixation activity in the coral holobiont during heat stress (Fig. 2A ), hence, was likely not driven by endosymbiotic diazotrophs in the coral tissue and did not result in increased assimilation of diazotroph-derived nitrogen by the algal symbionts in the present study (Fig. 3F ). Importantly, the dependence of algal symbionts on diazotroph-derived nitrogen is a function of their nutritional status. In this context, Rädecker et al. [ 15 ] recently showed (using the same coral colonies and experimental design) that heat stress shifted these coral holobionts from a state of nitrogen limitation towards a state of carbon limitation. Specifically, energy starvation caused by heat stress promoted the catabolic generation of inorganic nutrients (including ammonium) in the coral tissue, thereby promoting the proliferation of algal symbionts (Fig. S2 ). This ammonium released by the host catabolism was not isotopically enriched and therefore diluted diazotroph-derived nitrogen in the inorganic nutrient pool. Overall, the reduced 15 N enrichment observed here suggests that heat stress reduces the relative contribution of diazotroph-derived nitrogen to the coral holobiont nitrogen pool (Fig. 4B ). In this scenario, altered nutrient cycling may help explain the absence of epithelial 15 N enrichment hotspots during heat stress. Low photosynthate and high ammonium availability in the coral tissue likely suppress the nitrogen fixation activity of endosymbiotic diazotrophs and may give a competitive advantage to other microbes better adapted to exploit the altered nutrient regime during heat stress [ 82 – 84 ]. It is important to consider that the bulk isotope and NanoSIMS analyses only quantify the assimilation of anabolically incorporated 15 N into the coral and symbiont cells. As such, the assimilation of diazotroph-derived nitrogen in other holobiont compartments (i.e., the coral skeleton and the surface mucus layer) is not accounted for in our analyses. Indeed, a study by Moynihan et al. [ 34 ] recently suggested that endolithic microbes in the coral skeleton were the main source and sink of diazotroph-derived nitrogen in the closely related Pocillopora acuta holobiont. Further, El-Khaled et al. [ 85 ] showed that moderate increases in inorganic nutrient concentrations may stimulate nitrogen fixation rates in Red Sea corals. In this context, the observed net release of ammonium and phosphate by the coral host during heat stress would directly affect nutrient availability in other compartments of the coral holobiont such as the skeleton (Fig. S2 [ 15 ]). It is thus plausible that the increase in nitrogen fixation activity described here was predominantly driven by an increase in the activity of endolithic microbes associated with the holobiont. While some studies suggested that endolithic microbial communities may eventually release some of their nutrients to the coral tissue [ 34 , 42 , 86 , 87 ], our results clearly show that diazotroph-derived nitrogen (regardless of its origin within the holobiont) is an insignificant source of nitrogen for the coral-algal symbiosis during heat stress. As such, the increase in stimulated nitrogen fixation may primarily be absorbed within the endolithic community itself, thereby supporting the frequently documented rapid proliferation of endolithic microbes during heat stress [ 86 , 88 , 89 ]. Further, nitrogen fixation rates correlate with denitrification activities in some Red Sea corals [ 75 ]. Although we did not quantify denitrification rates in the present study, increases in the abundance and or activity of denitrifying microbes may have contributed to the removal of diazotroph-derived nitrogen from the coral holobiont during heat stress. Taken together, we conclude that the excess availability and release of ammonium through host catabolic processes, combined with increased microbial utilization of fixed nitrogen, likely reduce the relative contribution of diazotroph-derived nitrogen to holobiont nitrogen cycling during heat stress. The functional importance of diazotrophs in the coral holobiont Nitrogen fixation rates in coral holobionts are highly variable—they can differ between host species, locations, and fluctuate depending on local environmental conditions [ 34 , 37 , 38 , 41 , 42 , 82 , 87 , 90 ]. Indeed, the molecular, physiological, and ultrastructural characterization of nitrogen fixation presented here paints a complex picture of the role of diazotrophs in coral holobiont functioning. Our results suggest that endosymbiotic diazotrophs in the coral epithelium actively fix nitrogen in the coral holobiont. The characterization of tissue-associated diazotroph communities suggests that the observed 15 N hotspots may resemble individual endosymbiotic diazotrophs of the order Chroococcales and bacterial aggregates potentially including diazotrophs of the order Alteromonadales. Further, the disappearance of these 15 N hotspots during heat stress coincided with a decline in Chroococcales nifH relative sequence abundance and the reduced 15 N assimilation in algal symbiont cells. Taken together, our findings indicate that endosymbiotic diazotrophs in the coral tissue, especially Chroococcales, represent an important source of nitrogen for algal symbionts under nitrogen-limited conditions. Hence, the association with diazotrophs may supplement holobiont nutrition and enable net productivity under oligotrophic conditions [ 38 ]. However, this beneficial role of diazotrophs is likely limited to a narrow window of environmental conditions in which algal symbionts are strongly nitrogen-limited. Environmental conditions that reduce the demand for nitrogen in the coral holobiont (e.g., heat stress in the present study), may hence undermine the beneficial role of diazotrophs in holobiont functioning. This dynamic dependence on diazotroph-derived nitrogen in the coral holobiont is directly reflected in the variable diazotroph community assemblage, their activity, and the fate of diazotroph-derived nitrogen in the coral holobiont and may indirectly facilitate holobiont acclimatization and adaptation to changing levels of nutrient availability. In other words, this study suggests that the nutritional status of the coral holobiont determines the structure and activity of associated diazotroph communities and not vice versa. As such, diazotrophs may support coral holobiont functioning under nitrogen-limited conditions. However, anthropogenic impacts such as eutrophication and ocean warming likely reduce the benefits of hosting diazotrophs in the coral holobiont."
} | 6,056 |
25055832 | PMC4108945 | pmc | 679 | {
"abstract": "The mechanisms that underlie fascinating inter-individual interactions among animal groups have attracted increasing attention from biologists, physicists, and system scientists. There are two well-known types of interaction patterns: hierarchical and egalitarian. In the former type, individuals follow their leaders, whereas they follow their neighbors in the latter. Using high-resolution spatiotemporal data derived from the free flights of a flock of pigeons, we show that pigeon flocks actually adopt a mode that switches between the two aforementioned strategies. To determine its flight direction, each pigeon tends to follow the average of its neighbors while moving along a smooth trajectory, whereas it switches to follow its leaders when sudden turns or zigzags occur. By contrast, when deciding how fast to fly, each pigeon synthesizes the average velocity of its neighbors. This switching mechanism is promising for possible industrial applications in multi-robot system coordination, unmanned vehicle formation control, and other areas.",
"discussion": "Discussion Inter-individual interactions in collective biological groups are considered to involve two well-known mechanisms: following leaders or following neighbors. In the former mechanism, leadership can be hierarchical 4 or mono-level 12 27 , whereas the neighbor-hood can be defined by geometry 13 38 or topology 3 in the latter. The controversy between the leadership-dominating and egalitarian strategies has lasted for years, and it can be further generalized to monarchical and democratic ruling systems in socialized animal groups 42 . In the present study, we propose a more sophisticated system where pigeons switch between the two aforementioned strategies to facilitate the fine regulation of their movements in a variable environment. Each pigeon employs a hybrid strategy when deciding the direction of movement: it follows the leaders during sudden turns but synthesizes its neighbors movements when the trajectory is smooth. To determine its speed of flight, each pigeon also synthesizes its instantaneous neighbors to avoid potential lagging. To some extent, this is similar to how group decisions are made in other animal groups 43 . In addition, a pigeon flock is a sophisticated self-organized system, which could inspire the design of multi-agent systems. For example, the switchable hybrid strategy may improve the performance of multi-robot system coordination, particularly by reducing communication costs because HLN generally requires the transmission of much less information compared with FNR/FNN. The present study only considers small pigeon flocks, thus these results still require further verification using real experimental data for other animal groups (e.g., starlings 3 31 , surf scoters 2 , wild geese 29 , insect colonies 10 , and fish schools 9 ). Cross-species comparisons will determine the universality of the conclusions presented in this study, i.e., does this general switching mechanism underlie the behavior of many disparate animal flocks or is it only applicable to bird flocks (possibly even pigeons)? Another open question that requires further investigation is whether this switchable interaction mechanism operates in large-scale pigeon flocks and, if this is the case, how pigeons organize a large-scale hierarchical leadership network and whether the leader-follower relationships depend on the actual distance."
} | 855 |
28499334 | PMC5425657 | pmc | 680 | {
"abstract": "Realization of brain-like computer has always been human’s ultimate dream. Today, the possibility of having this dream come true has been significantly boosted due to the advent of several emerging non-volatile memory devices. Within these innovative technologies, phase-change memory device has been commonly regarded as the most promising candidate to imitate the biological brain, owing to its excellent scalability, fast switching speed, and low energy consumption. In this context, a detailed review concerning the physical principles of the neuromorphic circuit using phase-change materials as well as a comprehensive introduction of the currently available phase-change neuromorphic prototypes becomes imperative for scientists to continuously progress the technology of artificial neural networks. In this paper, we first present the biological mechanism of human brain, followed by a brief discussion about physical properties of phase-change materials that recently receive a widespread application on non-volatile memory field. We then survey recent research on different types of neuromorphic circuits using phase-change materials in terms of their respective geometrical architecture and physical schemes to reproduce the biological events of human brain, in particular for spike-time-dependent plasticity. The relevant virtues and limitations of these devices are also evaluated. Finally, the future prospect of the neuromorphic circuit based on phase-change technologies is envisioned.",
"conclusion": "Conclusions The ability to gradually induce a reversible switch between SET and RESET states, integrated with several superior transition properties such as fast switching speed, low energy consumption, and long retention, has made phase-change-based devices a leading candidate to emulate the biological synaptic events. Moreover, the excellent scalability of phase-change materials down to 2-nm size [ 112 ] also forecasts its potential to reproduce the ultra-high density neurons and synapses inside human brain. In this case, majority of the current work on electronic synapse are devoted to simulate the STDP event of the biological synapse that was reported to govern the learning and memory function by gradually changing the conductance of the phase-change materials, thereby resulting in several novel pulse schemes to adjust the device conductance with respect to the relative time delay between two external stimulus applied to device that represent pre-post spikes. Most importantly, the construction of PCM elements in array level allows for the emulation of large-scale connectivity of human brain with any given neuron having as many as 10,000 inputs from other neurons, which is exemplified by an integrated hardware with 256 × 256 neurons and 64,000 synapses [ 113 ]. In spite of the aforementioned merits, the neurons and synapses based on PCM devices are also facing some serious issues. Although the conductance of PCM device in crystalline phase can be modulated continuously, the device conductance in amorphous phase is found to suffer from a sudden change. This can be alleviated using multiple conductances per synapse and periodic corrections [ 3 ], but still remaining questionable. Besides, the inherent weakness of amorphous phase-change materials including resistance drift or relaxation of the amorphous phase after the melt-quenching also exacerbates the application of PCM on neuromorphic circuit systems. Another issue of PCM-based neuromorphic device arises from its fairly poor performance metrics mainly due to immaturities or inefficiencies in currently developed STDP learning algorithm. Under this circumstance, such inherent imperfections of the PCM devices may pose some indiscernible problems. To boost the classification accuracy, a so-called back propagation method that has received extensive application in computer science field to train artificial neural networks has recently been introduced a three-layer perceptron network with 164,885 synapses based on 2-PCM structure that was trained on a subset of a database of handwritten digits, leading to training and test accuracies of 82–83% [ 114 ], as shown in Fig. 26 . This work simply implies that classification accuracy can be achieved on the condition that either phase-change materials or the training algorithm permits PCM devices to serve more like a bidirectional NVM with a symmetric, linear conductance response of high dynamic range [ 115 ]. Fig. 26 Implementation of dense crossbar of PCM and selector devices for non-Von Neumann computing where neuron activates each other through dense networks of programmable synaptic weights. Reprinted with permission from [ 114 ] \n Despite the aforementioned challenges, the excellent physical properties of phase-change materials in conjunction with the currently mature technologies on PCM devices has provided an opportunity to envision the success of the future artificial neural networks that can perform the similar complex tasks to human brain while without sacrificing the occupied area and energy consumption. Device models suitable for neuromorphic architectures are still needed for application-specific performance evaluations of these systems."
} | 1,302 |
32753606 | PMC7403589 | pmc | 681 | {
"abstract": "Sediment microbial fuel cells (SMFCs) generate electricity through the oxidation of reduced compounds, such as sulfide or organic carbon compounds, buried in anoxic sediments. The ability to remove sulfide suggests their use in the remediation of sediments impacted by point source organic matter loading, such as occurs beneath open pen aquaculture farms. However, for SMFCs to be a viable technology they must remove sulfide at a scale relevant to the environmental contamination and their impact on the sediment geochemistry as a whole must be evaluated. Here we address these issues through a laboratory microcosm experiment. Two SMFCs placed in high organic matter sediments were operated for 96 days and compared to open circuit and sediment only controls. The impact on sediment geochemistry was evaluated with microsensor profiling for oxygen, sulfide, and pH. The SMFCs had no discernable effect on oxygen profiles, however porewater sulfide was significantly lower in the sediment microcosms with functioning SMFCs than those without. Depth integrated sulfide inventories in the SMFCs were only 20% that of the controls. However, the SMFCs also lowered pH in the sediments and the consequences of this acidification on sediment geochemistry should be considered if developing SMFCs for remediation. The data presented here indicate that SMFCs have potential for the remediation of sulfidic sediments around aquaculture operations.",
"introduction": "Introduction Living organisms harvest energy for their growth and metabolism by catalyzing redox reactions. While most organisms carry out both oxidation and reduction entirely within their own cells, a small subset of microbes are able to decouple these reactions, using redox shuttles and mediators, direct connections, or appendages called nanowires to access reactants outside the organism 1 , 2 . In some cases when microbes are provided with an electrically conductive connection across a redox gradient, they will use energy from only a single half reaction and transfer electrons to the electric circuit, capturing energy for their growth and metabolism, and producing modest amounts of electrical power as a consequence 3 . To do this, electrons generated by an oxidation reaction at an anode are transferred to a circuit either extracellularly 4 , 5 or with the aid of mediators 3 , 6 , 7 , and used by another microbial population at the cathode for reduction. This allows microbes to take advantage of more favourable redox pairings than they would otherwise have access to in their immediate vicinity. Since discovery of this phenomena over 100 years ago 8 there has been interest in harnessing it for practical applications 9 .\n Devices that exploit this are termed microbial fuel cells (MFCs) and proposed uses include energy capture from wastewater treatment 9 , environmental sensors 10 , power sources for low energy devices in remote locations 11 , 12 and bioremediation 13 , 14 . However, despite promising research and proof-of-concept protypes, they have yet to be adapted for widespread use in society, primarily due to low power generation, challenges in scaling up laboratory prototype systems, and the inability to demonstrate improvements upon current technologies 10 , 15 . One application that has shown promise is sediment microbial fuel cells (SMFCs) 16 . These are MFCs that take advantage of the naturally occurring redox gradients in organic rich sediments. In sediments, diffusive transport limits oxygen supply, causing a buildup of reducing equivalents and a switch to alternative electron acceptors. These alternatives are preferentially used according to available free energy along a vertically structured redox gradient 17 (i.e., nitrate reduction, Fe and Mn reduction, sulfate reduction, methanogenesis). By placing an electrode (anode) in the reduced layer of sediment and connecting it to a cathode in the overlying oxygenated water, an SMFC can drive current using oxygen in the water column as an electron accepter,essentially bypassing the transport limitation that gives rise to the redox gradient. SMFCs have been tested as a means to provide energy to low power oceanographic sensors in remote settings 11 , 12 , 16 , 18 and for bioremediation 14 , 19 , 20 . Recently, Kubota et al. 13 deployed a set of five SMFCs in the organic rich sediment (12% w/w organic carbon) of Tokyo Bay, and found a reduction in porewater sulfide concentrations, identifying SMFC’s potential as a remediation technology for organic matter contaminated sediments. However, it is unclear how such an SMFC could be scaled up to combat eutrophication over an entire bay (Tokyo Bay for example has a surface area of 1,500 km 2 ). SMFCs may be useful when there are localized point sources of organic matter loading. The sediments beneath aquaculture operations may be one such environment 21 , 22 . These sediments receive elevated inputs of organic matter due to the deposition of unused fish feed and feces, and the spatial scale of fin-fish pens (diameters of ~ 10 m) are of a size that scaling up an SMFC might be reasonable. Gilles 23 , in a review of 65 studies of aquaculture impacts reported organic carbon concentrations beneath aquaculture cages ranging from < 1 to 26% C w/w, compared to < 1–8% C w/w at unimpacted reference sites. Such organic matter loading can alter sediment chemistry, resulting in wide ranging effects to the surrounding ecosystems and biogeochemical cycles 24 . When organic matter builds up in sediments, sediment oxygen demand (SOD) increases and oxygen is depleted at much shallower depths 25 , 26 . This shoaling of the aerobic layer results in a shift to anerobic metabolism and a buildup of by-products, such as ammonium, sulfide, and methane 27 . Sulfate reduction, which produces sulfide (tot-S 2− = H 2 S + HS − + S 2− ), can be responsible for up to 65% of total decomposition in marine sediments and increases with higher rates of organic matter loading 28 . High rates of benthic metabolism and sulfate reduction, up to an order of magnitude greater, have been observed beneath aquaculture farms relative to reference sites 27 , 29 , 30 . Dissolved sulfide is toxic to aerobic organisms living on or in the sediments 31 . Toxicity can be observed at μmolar concentrations and effects include neurotoxicity, modification of oxygen transport proteins, and inhibition of a variety of enzymes 32 . Accumulation of sulfide can also expedite a transition to hypoxic conditions by further increasing SOD as chemoautotrophic microbes re-oxidize sulfide to obtain energy. Brooks and Mahnken 27 reported that as sulfide concentrations increased in sediments, the number of benthic taxa declined, dropping by 50% when sulfide concentrations reached ~ 1,000 μM. The accumulation of sulfide beneath aquaculture cages is well documented 23 and of concern, such that sediment sulfide levels form the basis of environmental monitoring regulatory regimes in many jurisdictions 33 . Evidence that SMFCs use sulfide as an electron donor 34 and may accelerate the degradation of organic matter 35 , suggests they could be used for remediation of open-pen aquaculture sites. However, for this to be successful SMFCs must remove or prevent the accumulation of sulfide at levels comparable to those found beneath fish pens. This provides the motivation of this study, which was to quantify SMFC sulfide removal and compare this to sulfide accumulation at aquaculture sites. Our approach was to measure current generation and sulfide in laboratory SMFC microcosms over a period of 96 days. High organic matter sediments (6% C w/w) 36 were collected from a coastal inlet and used to construct 4 laboratory microcosms in aquarium tanks. Two tanks contained operating sediment microbial fuel cells (SMFC-1 and SMFC-2) while the other two served as controls, an open circuit control (OC), and a sediment only control (SO). The open circuit control contained all the components of a sediment microbiol fuel cell, but without an electrical connection between the anode and cathode. The sediment only tank contained sediment and overlying water without microbial fuel cell components. A photograph and diagram are shown in Fig. 1 a, b. To assess the influence of the SMFCs on sediment geochemistry, micro-electrodes were used to measure depth profiles of oxygen, total dissolved sulfide (tot-S 2− = H 2 S + HS − + S 2− ), and pH at high spatial resolution (100–500 µm). Differences in profiles between the SMFCs and controls were used to determine the influence of SMFCs on sediment oxygen demand (SOD), sulfide, and pH. Figure 1 ( a ) One of the sediment microbial fuel cell microcosms used in the experiment (SMFC-2). The anode depth is indicated by the dashed red line. ( b ) Schematic diagram of the sediment microbial fuel cell showing directions of current, electron, and proton flow and possible oxidation and reduction reactions at the anode and cathode.",
"discussion": "Discussion SMFCs provide an electrochemical connection between anoxic sediments and the oxygenated water column above, essentially functioning as a biogeochemical snorkel 45 . This allows sediment microbes to access more favorable electron acceptors (i.e. oxygen) than would otherwise be available. This suggests SMFCs could be used for the remediation of sediments impacted by point-source organic matter loading, such as the accumulation of aquaculture waste beneath fish-pens. We examined this question with a SMFC microcosm laboratory experiment that addressed two specific questions: (1) do SMFCs lower sediment oxygen demand (SOD) by accelerating organic matter decomposition? and (2) can SMFCs lower porewater sulfide concentrations to levels sufficient for the remediation of sediments beneath aquaculture farms? To address the former, we examined changes in oxygen microsensor profiles. Since bioturbators and irrigators were absent, diffusion was the only process transporting oxygen into the sediments and depth integrated oxygen consumption rates provide a reasonable estimate of carbon remineralization (Eq. 2 ). If SMFCs increased organic carbon degradation, then by the end of the experiment oxygen consumption in the SMFCs microcosms should be lower than the controls. However, though consumption did decrease throughout the experiment, there was no difference between the SMFCs and controls. This suggests SMFCs did not increase the rate of organic matter remineralization. This is in contrast to Ishii et al. 35 who found that an MFC bioreactor fed with primary clarifier effluent from a wastewater treatment plant showed complete removal of organic matter in 8–12 days, compared to 15–20 days without. However, our results are consistent with Kubota et al. 13 , who observed no change in sediment organic matter content over a 142 day deployment of SMFCs in Tokyo Bay. A likely explanation for this discrepancy is differences in organic matter reactivity and the rate limiting step for remineralization. The clarifier effluent feeding the Ishii et al. 35 MFC reactors is likely composed of small labile organic molecules that can be easily metabolized by microbes. In this case the transport of electrons, reaction kinetics, and activity of the microbial population would control the rate of organic matter degradation. However, in sediments organic matter is composed of larger complex molecules which must first be hydrolyzed by extracellular enzymes to small labile compounds that can be more easily metabolized by microbes. It is this hydrolyzation step, determined by the reactivity of organic matter, that is often the rate limiting step 46 – 48 , rather than the metabolic activity of the microbial population. Therefore, unless the biofilm can significantly increase the rate of hydrolyzation of these molecules the rate of carbon remineralization would not be expected to increase. However, although the rate of organic matter remineralization appears unaffected by the presence of the SMFCs, the distribution of remineralization between aerobic and anerobic processes is markedly different. This is indicated by the lower sulfide concentrations in the SMFCs compared to the controls (Fig. 4 a). Reductions in porewater sulfide concentrations have been observed near SMFCs deployed in other marine sediments 11 , 12 , 34 , 49 as well. The linear decease in sulfide at the anode (Fig. 4 b) suggests sulfide is a key source of electrons for current generation. This is similar to Tender et al. 12 , who observed a linear decreases in porewater sulfide above and below an anode during an SMFC deployment in a salt marsh near Tuckerton, NJ. Likewise, Ryckelynck et al. 34 demonstrated with a laboratory experiment, that chamber MFCs can be driven solely with sulfide as an electron donor. They proposed a mechanism for sulfur cycling near the anodes of SMFCs whereby reduced sulfur, produced from organic matter fueled sulfate reduction, is re-oxidized to provide the electrons for current generations. However, while sulfide is clearly a source of electrons to the anode, the question remains as to the end product of oxidation and whether additional electron donors are required to balance current generation. In sediments, sulfide can be oxidized completely to sulfate (SO 4 2− ), or partially to elemental sulfur, S 0 , according to the following half reactions, 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}$$ {\\text{HS}}^{ - } \\to {\\text{S}}^{{0}} + 2e^{ - } + {\\text{H}}^{ + } $$\\end{document} HS - → S 0 + 2 e - + H + \n 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}$$ {\\text{HS}}^{ - } + 4{\\text{H}}_{2} {\\text{O}} \\to {\\text{SO}}_{4}^{2 - } + 8e^{ - } + 9{\\text{H}}^{ + } . $$\\end{document} HS - + 4 H 2 O → SO 4 2 - + 8 e - + 9 H + . \n Whether Eq. ( 5 ) or ( 6 ) is the oxidation reaction is an important question with regards to the remediation of sulfidic sediments. If S 0 is the end product, then its long-term fate after the SMFC is removed needs to be considered. Elemental sulfur in sediments may undergo several reactions 50 , it can be converted back to sulfide either through disproportionation 51 , 52 or direct reduction and could negate some of the benefits of the SMFC. It can be converted to iron sulfides (FeS) and eventually pyrite (FeS 2 ) which would immobilize it in the sediment, or ideally it can be further oxidized to sulfate 53 , decreasing the reducing capacity of the sediment as desired. To ascertain which sulfide oxidation reaction, ( 5 ) or ( 6 ), is occurring and what additional pathways, if any, are needed for current generation, we compared the current, expressed as an electron flux, with the sulfide flux to the anode (Table S2 ). If reaction ( 5 ) is occurring then 2 electrons are transferred per sulfide oxidized, and sulfide reduction generates an electron flux of 3.5 mmol m −2 d −1 (day 96) to the anode, while if reaction ( 6 ) is the oxidation pathway then 14 mmol m −2 d −1 (day 96) were supplied to the anode. This could account for between 18 and 73% of the electrons required for the observed current. This suggests that other oxidation pathways in addition to sulfide oxidation are occurring at the anode, most likely the oxidation of iron sulfide minerals, or the direct oxidation of organic matter. A more detailed study of sulfur speciation and organic matter degradation in the vicinity of the anode would be needed to distinguish between these cases. The SMFC had maximum sulfide concentrations approximately 700–750 μM lower than the controls. However, for SMFCs to be a viable tool for remediation, this difference must be comparable to the level of environmental impact. In aquaculture settings negative impacts on meiofauna communities, such as lower biomass, decreased diversity, or the invasion of opportunistic sulfide tolerant species, are observed to begin when sulfide concentrations beneath fish pens are 350–1,500 μM 31 , 54 , 55 . Recently, Cranford et al. 55 examined a variety of benthic community health indicators and related them to sediment sulfide levels at six aquaculture farms in Canada and New Zealand. They showed that benthic ecosystem health declined dramatically when sulfide exceeded 1,000 μM and developed a scale relating ecological quality status (EQS) to sulfide in the upper sediments whereby; sulfide < 200 μM was considered “good”, above 500 μM “poor”, and > 1,000 μM “bad”, for the health of benthic communities. Based on this, the approximately 750 μM reduction in sulfide observed in our experiment would be a significant improvement to the health of benthic ecosystems impacted by aquaculture. In addition to lower sulfide concentrations, the SMFC microcosms had lower sediment pH than the OC and SO controls. This decrease is due both to the direct generation of protons at the anode and the additional removal of alkalinity associated with the consumption of HS − . The effect of pH on sediment geochemistry is complicated, due to the large number of potential proton-producing and consuming reactions that may occur 56 . However, generally lowering pH will promote the dissolution of mineral phases, such as CaCO 3, or iron sulfides ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{FeS}} \\to {\\text{Fe}}^{2 + } + {\\text{S}}^{2 - }$$\\end{document} FeS → Fe 2 + + S 2 - ). The dissolution of FeS can be a positive feedback on SMFC current generation because the sulfide produced can also be oxidized by the SMFC. The liberated dissolved iron (Fe 2+ ) will diffuse upwards towards the sediment surface where it is oxidized and precipitated as iron oxides. The accumulation of an iron oxide pool in the surface sediments can provide an additional layer of protection against sulfide even after the SMFC has been removed. Iron oxides have the ability to oxidize sulfides, and the Fe 2+ produced through this sulfide oxidation can combine with additional sulfide to form FeS. While dissolved sulfide is highly toxic, FeS minerals are not. A similar iron cycling mechanism plays a role in the recently described cable bacteria “firewall” against euxinia 57 . On the other hand, FeS also binds and detoxifies heavy metals such as Cu and Zn 58 . If an SMFC induced drop in pH promoted FeS dissolution these metals could be mobilized and become bioavailable. Both Cu and Zn are additives to fish feed and Cu is a common biofouling agent applied to nets and farm equipment 59 . Although work has been done to limit their use in aquaculture 60 , the diagenetic fate of these metals due to pH changes should be considered. A consideration of any manipulated laboratory experiment is how representative it is of the natural environment. To evaluate this, we compare current and power densities of our microcosm SMFCs to SMFCs deployed in the field. When comparing SMFC studies it is important to ensure that current and power are normalized similarly in both studies. Normalization can be done relative to the anode, cathode, or sediment footprint, depending upon the goals of the researchers. Here we only compare with studies that normalized to the anode sediment footprint as we have done. Kubota et al. 13 obtained power densities of 11.5 mW m −2 and current densities between 10 and 30 mA m −2 for SMFCs deployed in Tokyo Bay. Tender et al. 12 and Ryckelynck et al. 34 report power densities of 10–30 mW m −2 from SMFCs deployed in a salt marsh 2 , and 30 mA m −2 and 11 mW m −2 for an SMFC in a coastal estuary. The similarity in power and current densities between the field deployed SMFCs and ours gives confidence that our results are transferrable to the natural environment. Interestingly, the electrogenic properties (power density and internal resistance) of the OC improved between day 0 and 46. This suggests the mere presence of the carbon fiber electrode might provide a surface for the development of a biofilm with some potential for electron transfer regardless of whether it had been connected to a circuit or not. Sediments have steep redox gradients and one possible explanation is that a conductive material placed in the sediments could encourage biofilm formation by providing electrical connections across small scale redox gradients. However, the higher internal resistance of the OC (247–347 Ω) compared to the SMFCs (98–128 Ω) suggests that any biofilm that developed on the OC electrode was less adapted to carry out this task. SMFC-1 and 2 had selective pressure, in the form of a voltage gradient, to shift the biofilm community toward those members of the microbial community that could take advantage of the electrical connection to the overlying water while the OC did not. In conclusion, we have demonstrated that SMFCs have the potential to remove sulfide from sediments at a scale that could benefit the benthic environment beneath aquaculture pens. Aquaculture is a growing industry and its expansion is only expected to accelerate as it becomes increasingly difficult for wild-capture fisheries to meet society’s demand for marine protein 61 . Therefore, solutions to environmental problems, such as organic matter loading and sulfide accumulation, need to be developed. SMFCs are a promising solution, provided they can be scaled up. Most SMFCs deployed in the environment generally are on the order of 0.5 m 2 \n 11 – 13 , 34 , 49 , still much smaller than aquaculture fish pens which have diameters of 10 s of meters. Nevertheless, although there are likely challenges to be overcome, this work will inspire future research aimed at investigating the feasibility of SMFCs as a solution to this persistent environmental problem."
} | 5,561 |
34652130 | PMC8609572 | pmc | 682 | {
"abstract": "Biofilms are three-dimensional\n(3D) bacterial communities that\nexhibit a highly self-organized nature in terms of their composition\nand complex architecture. Bacteria in biofilms display emergent biological\nproperties, such as resistance to antimicrobials and disinfectants\nthat the individual planktonic cells lack. Bacterial biofilms possess\nspecialized architectural features including unique extracellular\nmatrix compositions and a distinct spatially patterned arrangement\nof cells and matrix components within the biofilm. It is unclear which\nof these architectural elements of bacterial biofilms lead to the\ndevelopment of their emergent biological properties. Here, we report\na 3D printing-based technique for studying the emergent resistance\nbehaviors of Escherichia coli biofilms\nas a function of their architecture. Cellulose and curli are the major\nextracellular-matrix components in E. coli biofilms. We show that 3D-printed biofilms expressing either curli\nalone or both curli and cellulose in their extracellular matrices\nshow higher resistance to exposure against disinfectants than 3D prints\nexpressing either cellulose alone or no biofilm-matrix components.\nThe 3D-printed biofilms expressing cellulose and/or curli also show\nthicker anaerobic zones than nonbiofilm-forming E.\ncoli 3D prints. Thus, the matrix composition plays\na crucial role in the emergent spatial patterning and biological endurance\nof 3D-printed biofilms. In contrast, initial spatial distribution\nof bacterial density or curli-producing cells does not have an effect\non biofilm resistance phenotypes. Further, these 3D-printed biofilms\ncould be reversibly attached to different surfaces (bacterial cellulose,\nglass, and polystyrene) and display resistance to physical distortions\nby retaining their shape and structure. This physical robustness highlights\ntheir potential in applications including bioremediation, protective\ncoatings against pathogens on medical devices, or wastewater treatment,\namong many others. This new understanding of the emergent behavior\nof bacterial biofilms could aid in the development of novel engineered\nliving materials using synthetic biology and materials science approaches.",
"conclusion": "Conclusions Taken together, our findings indicate that 3D printing can be effectively\nemployed for studying the emergent biological endurance of bacterial\nbiofilms by tuning the design principles and bioink composition (using\nmonoculture or coculture inks). 3D-printed biofilms closely mimic\nnatural biofilms in terms of their spatial heterogeneity and diffusion\nof molecular oxygen. Oxygen penetration in 3D-printed biofilms (expressing\ncellulose and/or curli) is limited such that the top layers are exposed\nto oxygen, whereas the layers toward the bottom exhibit an anaerobic\nstate. The composition of the extracellular matrix is crucial for\ndetermining the resultant emergence of resistance against disinfectants.\n3D-printed biofilms expressing either only curli or both curli and\ncellulose are more resistant to ethanol and Virkon S than the 3D prints\nexpressing cellulose alone, indicating the protective nature of curli\nfibers. 3D-printed biofilms retain an adhesive nature and can be reversibly\nattached to different surfaces such as bacterial cellulose, glass,\nand polystyrene. The physical stability of 3D-printed biofilms on\nmechanically robust bacterial cellulose surfaces and their biological\nendurance to extreme environmental conditions highlights their suitability\nfor such applications as protective probiotic coatings on medical\ndevices that prevent colonization by pathogens, degradation of toxic\nchemicals, bioremediation, or use in wastewater treatment facilities,\namong many others.",
"introduction": "Introduction Bacterial\nbiofilms are three-dimensional (3D) assemblages of bacteria\nin a self-generated matrix (composed of proteins, polysaccharides,\nlipids, and extracellular DNA) that strongly attach to biotic or abiotic\nsurfaces. 1 − 4 Biofilms are widely present in natural, medical, and industrial\nsettings. 5 , 6 Depending on the context, biofilms can be\nregarded as harmful (e.g., causing device-related infections, sepsis,\nfood-borne infections, etc.) or beneficial (e.g., in degradation of\ntoxic chemicals, bioremediation, bioleaching, sustainable material\nproduction, etc.). 7 Bacteria in biofilms\nsubstantially differ from their free-living or planktonic counterparts\nin terms of their resilience and adaptability to extreme conditions,\nincluding the presence of antimicrobials, solvents, detergents, high\ntemperature, and so forth. 8 − 11 Bacterial cells in a biofilm exhibit emergent biological\nproperties (e.g., resistance to antimicrobials/disinfectants) and\nmechanical properties (viscoelastic nature) that individual planktonic\ncells do not possess. The resilient nature of cells in a biofilm is\nthought to arise as a consequence of local physical interactions between\ndifferent or individual extracellular-matrix components within a biofilm. 12 The emergent endurance of biofilms is frequently\nassociated with factors including the structure, composition, architecture,\nspatial organization, or mechanical properties (including cohesiveness,\nviscoelastic nature, resistance to hydrodynamic shear, and stiffness)\nof constituent biofilm molecules. 13 − 15 Recently, there has\nbeen a growing interest in investigating these emergent properties\nof bacteria in biofilms. 13 , 16 While the contribution\nof individual extracellular-matrix components in the biofilm to the\nemergent mechanical properties has been studied, 13 their contribution to the emergent biological endurance\nremains poorly understood. The emergent resistance properties\nof biofilms must be due to the\nunique features of the biofilms that planktonic cells do not possess,\nsuch as the presence of extracellular-matrix components and/or their\nspatial structuring. The spatial structuring of natural biofilms is\ngoverned by parameters including local bacterial density, biofilm\nmatrix composition and density, and so forth. These crucial variables\nhave been hypothesized 10 to have dynamic\nconsequences on the distribution of molecular oxygen and the emergent\nbiological and mechanical endurance of biofilms. 10 , 12 , 17 However, it has been impossible to study\nand control these variables individually, such that the ultimate design\nprinciples of biofilms remain unknown. Tuning these variables and\nstudying their emergent consequences can generate useful information\nabout the structure–function relationships of bacterial biofilms\nover time, leading to better understanding of the underlying biology. 10 Further, since the majority of human infections\nare caused by biofilm-forming bacteria, 18 elucidating the causes of emergent biofilm resistance behavior can\nfacilitate better design of new antibiofilm strategies. Beyond the\nfundamental or medical nature of such studies, these emergent properties\nof biofilms could serve as new platforms for construction of robust\nnext-generation smart materials using synthetic biology and materials\nscience approaches. 17 , 19 , 20 However, the challenge here lies in achieving top-down spatial patterning\nof the biofilm components in order to study their effect on the emergent\nbiological properties. 3D printing is a robust technology that\ncan be used to tackle this\nchallenge. With the development of 3D bioprinting, it is now possible\nto intentionally alter the spatial patterning of individual extracellular-matrix\ncomponents and probe their contributions to the emergent resistance\nphenotypes of biofilms. 3D printing has been increasingly used for\nthe fabrication of living functional materials from nano- to macroscales\nthrough printing algae, bacteria, fungi, yeast, plant, and animal\ncells. 21 − 31 3D printing allows for the spatial patterning of constituents mimicking\nthe complex 3D microenvironments and time-evolving nature of living\nsystems. 32 − 34 The spatial heterogeneity and mechanical robustness\nof natural biofilms can be simulated with a high degree of control\nover freedom of shape, design, and resolution provided by 3D printing.\n3D-printed biofilm models can potentially better mimic the 3D organization\nof natural biofilms than conventionally studied biofilms (grown in\nliquid media or agar) in the laboratory and can be employed for studying\nfundamental topics including emergent biological endurance to antimicrobials. 35 We have previously shown that biofilms of the\nGram-negative bacterium Escherichia coli can be effectively 3D-printed into desirable patterns at the sub-millimeter\nscale resolution using a do-it-yourself home-built 3D printer. 36 , 37 The extracellular matrix of E. coli biofilms is primarily composed of curli fibers, a proteinaceous\ncomponent, and/or cellulose, a polysaccharide component. 5 , 13 , 14 , 38 , 39 This nanocomposite matrix has been suggested\nto confer biological and mechanical endurance to the E. coli cells. 40 Thus,\ntuning the matrix composition and the design principles of E. coli biofilms with 3D printing is a powerful approach\nto analyze the effect of biofilm composition and architecture on their\nemergent endurance. Herein, we report a simple approach for\nspatial patterning of different E. coli strains and studying the emergent biological\nendurance of their biofilms using 3D printing technology ( Figure 1 ). We employed a\ncustomized do-it-yourself 3D printer 23 , 36 for arbitrary\npatterning of biofilm-forming E. coli onto agar substrates. Different strains of E. coli expressing cellulose+/curli+, cellulose–/curli+, and cellulose–/curli–\nin their biofilm matrix were 3D-printed as four-layered constructs\nand tested for their emergent biological endurance to disinfectants.\nWe demonstrate that the bacterial biofilms can be effectively 3D-printed\ninto different shapes, and the 3D-printed biofilms display emergent\nresistance to commonly used laboratory disinfectants such as ethanol\nand Virkon S. Particularly, we show that tuning the biofilm matrix\ncomposition and design principles, such as alteration of bacterial,\ncurli, or cellulose density, through the design of the bioink and\n3D printing process has a major influence on the development of resistance\ntoward these disinfectants. For strains expressing curli and/or cellulose,\nthe diffusion of molecular oxygen into the 3D-printed biofilms is\nlimited, and anaerobic zones exist in the lower layers of the biofilm.\nFurther, these 3D-printed biofilms exhibit striking resistance to\nphysical distortions and stably retain their original shape. The 3D\nprints can be reversibly attached and detached to different surfaces\nsuch as bacterial cellulose, glass, and polystyrene, demonstrating\ntheir utility in beneficial applications including probiotic biofilm\ncoatings on medical devices, bioremediation, and wastewater treatment\nplants, among many others. Figure 1 3D printing of bacterial biofilms for studying\ntheir emergent biological\nendurance. Altering the bioink composition (by mixing one or two different\ntypes of biofilm-forming E. coli together\nwith sodium alginate) and printing them into spatially defined patterns\nresult in the formation of 3D-printed biofilms over time. These biofilms\nclosely mimic the spatial heterogeneity found in natural biofilms\nand could be readily used for understanding the emergent biological\nendurance to disinfectants. This illustration was created with images\nfrom https://smart.servier.com/ .",
"discussion": "Results and Discussion 3D Printing E. coli Biofilms\nwith Different Extracellular Matrix Compositions Since the\nmatrix of E. coli biofilms is predominantly\ncomposed of cellulose and curli, 5 , 40 we employed three different\nstrains of E. coli expressing curli\nand/or cellulose: E. coli Nissle wildtype\n(cellulose+/curli+); E. coli MG1655\nΔ csgA carrying a plasmid for expression of\nconstitutive green fluorescent protein (GFP) and rhamnose-inducible\nCsgA (cellulose–/curli+); and E. coli MG1655 Δ csgA (cellulose–/curli–).for\nstudying the contribution of biofilm-matrix polymers and their spatial\ndistribution to emergent biological endurance. We utilize the inducible\ncurli by preparing media and bioink without an inducer and then depositing\nor printing the bacteria onto surfaces containing the inducer. In\nthis way, biofilm production only begins after deposition, resulting\nin reproducible curli production for both biofilms and biofilm prints. E. coli Nissle (cellulose+/curli+) is a natural biofilm-forming\nstrain that has a different strain background from the two E. coli MG1655 strains (cellulose–/curli+\nand cellulose–/curli−). As such, the Nissle strain is\nincluded in these studies as a positive control representing a natural\n3D-printed biofilm to compare to our engineered 3D-printed biofilms.\nIn E. coli , cellulose and curli are\nbetter expressed under lower temperature conditions (<30 °C)\ndue to the transcriptional regulator CsgD that upregulates the expression\nof both curli and cellulose. 5 Therefore,\nall biofilm formation assays were carried out under room-temperature\nconditions. Biofilm formation by our experimental strains and\nthe presence of cellulose and/or curli were evaluated by three different\nstaining assays: crystal violet, Congo red, and calcofluor ( Figure 2 ). For the first\nassay, the ability of the strains to form biofilms in liquid culture\nwas evaluated by a crystal violet biofilm assay performed in glass\ntest tubes ( Figure 2 A). The formation of biofilms at air–liquid boundaries is\ncharacteristic of E. coli biofilms. 12 E. coli strains\nexpressing no cellulose but only curli (cellulose–/curli+)\nor both cellulose and curli (cellulose+/curli+) were able to form\nbiofilms, as visualized from a crystal violet-stained ring formed\nat the air–liquid interface in the culture tube. In contrast,\nno crystal violet staining was seen in E. coli strains that expressed neither cellulose nor curli (cellulose–/curli−)\nor the no-bacteria samples. Figure 2 Biofilm formation by E. coli strains\nin this study. (A) Crystal violet (top), Congo red (middle), and calcofluor\nassays (bottom) for visualization of total biofilm and to identify\nthe biofilm-matrix components. The crystal violet assay detects the\ntotal biofilm formation in liquid culture, whereas the Congo red assay\ndetects the presence of cellulose and/or curli, and the calcofluor\nassay detects the presence of cellulose in colony biofilms (hydrogel\nculture) and (B) resistance of 3D-printed biofilms (one- or four-layered\nprints) to citrate treatment. Images on the top depict 3D-printed\nbiofilms before citrate treatment, and the images on the bottom depict\n3D-printed biofilms after citrate treatment. All biofilm samples were\ngrown at room temperature for 7 days before these experiments were\ncarried out. E. coli are also known to form biofilms\non gel–air boundaries when grown as colonies on agar plates. 13 Therefore, in the second assay, the presence\nof curli and/or cellulose in colony biofilms was visualized by the\nbinding of the diazo dye Congo red to these biofilms. 14 , 41 Strains expressing cellulose and/or curli (cellulose+/curli+ and\ncellulose–/curli+) appeared red in color due to the binding\nof Congo red to cellulose and/or curli fibers, confirming the presence\nof curli and/or cellulose. In contrast, the strain expressing neither\ncellulose nor curli (cellulose–/curli−) appeared pale\nin color ( Figure 2 A).\nLastly, the presence of cellulose in the biofilm matrices was visualized\nby a calcofluor assay, 14 in which bacteria\nare grown on a supportive growth medium and calcofluor in the growth\nmedium binds to bacterially produced cellulose to produce a fluorescent\nsignal. The cellulose+/curli+ strains exhibited bright fluorescence\nunder UV, indicating the presence of cellulose, while the strains\nnot expressing cellulose (cellulose–/curli– or cellulose–/curli+)\nappeared nonfluorescent ( Figure 2 A) due to the absence of cellulose. Taken together,\nthese results indicate that our bacterial strains expressing curli\nand/or cellulose were able to robustly produce biofilm matrix components\nupon growth in liquid or hydrogel culture under our laboratory growth\nconditions. We 3D-printed each of these three E. coli strains expressing cellulose and/or curli\n(cellulose+/curli+, cellulose–/curli+,\ncellulose–/curli−) biofilm matrix polymers as one-layered\nor four-layered stripes using our customized 3D bioprinter. 23 , 36 The 3D prints consisted of E. coli bacteria immobilized within a calcium alginate hydrogel. The alginate\nhydrogel serves as a cell-compatible physical support during the bacterial\nbiofilm formation process. After 1 week of incubation at room temperature,\nthe formed 3D-printed biofilms were tested for resistance to citrate\ntreatment as a proxy to detect the formation of biofilm-matrix components\n( Figure 2 B). Treatment\nwith sodium citrate dissolves the alginate hydrogel matrix unless\nit is reinforced by biofilm-matrix polymers produced by the encapsulated\nbacteria. 37 Only the 3D prints expressing\ncurli alone (cellulose–/curli+) or both curli and cellulose\n(cellulose+/curli+) were capable of resisting the citrate treatment,\nwhereas the 3D prints containing no bacteria or bacteria expressing\nneither curli nor cellulose (cellulose–/curli−) were\ndissolved by citrate treatment. These results indicated that the E. coli strains expressing either curli alone or\nboth curli and cellulose were capable of robust biofilm matrix production\nafter 3D printing. We examined the 3D printability of these E. coli biofilms by designing arbitrary shapes and\npatterns. A variety of\ndifferent patterns of biofilms could be generated with this 3D printing\napproach ( Figure 3 ).\nThe production of cellulose in the 3D-printed biofilms was determined\nby visualizing the fluorescence of the 3D prints under UV by a calcofluor\nassay. 3D-printed biofilms expressing cellulose (cellulose+/curli+)\nshowed bright fluorescence, whereas 3D-printed biofilms not expressing\ncellulose (cellulose–/curli+) or expressing no matrix components\n(cellulose–/curli−) appeared nonfluorescent ( Figure 3 , bottom). Thus,\nour approach makes it possible to 3D-print E. coli biofilms composed solely of curli or a combination of curli and\ncellulose while providing freedom of design and patterning. Figure 3 3D printability\nof 7 day E. coli biofilms expressing\ncellulose and/or curli. Different types of possible\npatterning of 3D-printed biofilms (top three rows) and their fluorescence\nunder UV in a calcofluor assay (bottom row). Calcofluor fluorescence\nunder UV (wavelength: 312 nm) indicates cellulose production in the\n3D-printed biofilms. 3D-Printed E. coli Biofilms Display\nLimited Penetration of Molecular Oxygen In natural biofilms,\nthe spatial arrangement of cells within the biofilm has implications\nfor their aerobic or anaerobic state. For instance, cells in the top\nlayers of a biofilm are exposed to ambient levels of oxygen and therefore\ncomposed of fast-growing cells. In contrast, the intermediate and\nlower layers of the biofilm comprise an anaerobic zone of slow-growing\ncells due to the limited diffusion of oxygen through the biofilm. 42 , 43 The altered growth and microbial metabolism of biofilm cells and\nthe presence of slow-growing cells due to the anoxic zone has been\nhypothesized to lead to emergent resistance behavior of biofilms. 12 , 44 , 45 For instance, the oxygen-depleted\nstate of Pseudomonas aeruginosa biofilms\nhas been shown to contribute to increased antibiotic tolerance. 46 In order to understand whether specific\nbiofilm-matrix components were able to allow the 3D-printed biofilms\nto reproduce the anaerobic zones of native biofilms, we measured the\noxygen penetration profile of 3D-printed biofilms with three different\nphenotypes (cellulose+/curli+, cellulose–/curli+ and cellulose–/curli−)\nat different depths using an oxygen microelectrode system ( Figure 4 ). The presence of\noxygen within the 3D-printed E. coli varied noticeably in the presence of curli. Oxygen within the 3D-printed E. coli expressing cellulose+/curli+ and cellulose–/curli+\nrapidly disappeared within the upper 200 μm of the biofilm,\nwhile oxygen reached the bottom (400 μm depth) of the 3D-printed E. coli expressing neither cellulose nor curli (cellulose–/curli−). E. coli 3D prints expressing cellulose+/curli+ and\ncellulose–/curli+ in their biofilms showed extended zero-oxygen\nzones of ∼300–400 μm in height. Hence, these 3D-printed\ncurli+ biofilms presented anoxic conditions similar to native biofilms.\nThis rapid decrease in oxygen concentration within the 3D prints expressing\ncurli could be due to diffusion limitations rendered by increased\ndensity of the biofilm matrix (physical) as well as the microbial\nactivity (biological). Figure 4 Oxygen profiles of four-layered 3D-printed biofilms revealing\nthe\npresence of zero-O 2 zones in the bottom layers. An oxygen\nmicroelectrode was used to profile the oxygen concentration at different\ndepths in 3D-printed E. coli . The deepest\nO 2 measurement coincides with the bottommost point of the\n3D print at the interface with the supportive media such that the\nthickness of 3D prints can be compared with this method. In contrast, in the absence of biofilm extracellular-matrix\ncomponents\n(cellulose–/curli−), microbial consumption of oxygen\nlikely became the limiting factor for oxygen availability rather than\nphysical diffusion of oxygen. No zero-oxygen zones were observed for\nthe 3D-printed E. coli expressing neither\ncurli nor cellulose (cellulose–/curli−). In the absence\nof cells and biofilm extracellular-matrix components (no cells), oxygen\nconcentrations remained consistently high throughout the entire 3D-printed\nalginate hydrogel structure (thickness: 600 μm). The thickness\nof the 3D prints studied using this method was determined.\n3D-printed E. coli expressing cellulose\nand/or curli (cellulose+/curli+ and cellulose–/curli+) had\na maximum thickness of 550 μm, whereas the 3D-printed E. coli expressing neither cellulose nor curli (cellulose–/curli−)\nhad a maximum thickness of 400 μm. Hence, biofilms expressing\nthe extracellular-matrix components cellulose and/or curli were thicker\nthan the ones not expressing the extracellular-matrix components. We compared the bacterial viability (cfu/mL) of the 3D prints per\nunit thickness among the 3D-printed biofilms expressing cellulose\nand/or curli ( Figure S1 ). E. coli expressing cellulose and curli (cellulose+/curli+)\ndemonstrated higher viability per unit thickness than E. coli expressing curli and no cellulose (cellulose–/curli+).\nInterestingly, E. coli expressing neither\ncellulose nor curli (cellulose–/curli−) exhibited higher\nviability per unit thickness than E. coli expressing curli and no cellulose (cellulose–/curli+). Thus,\nthe production of curli may reduce the overall viability and/or density\nof the individual cells, and the lower oxygen concentrations in the\ncellulose–/curli+ strains cannot be attributed to an increased\nbacterial concentration in the biofilms as a whole. To fully understand\nthe differences in oxygen profiles observed between different 3D-printed\nbiofilms, further research is needed to determine the microscale distribution\nand activity of cells in the oxic layers of the 3D-printed biofilms,\nas well as the changes in the physical properties of the biofilms\ndue to curli and/or cellulose expression that may affect the diffusion\nof gases within the biofilms. The reduction in oxygen concentration\nat lower biofilm depths for\nengineered 3D-printed curli+ biofilms could potentially have an impact\non the viability or metabolic functioning of bacterial cells at these\ndepths. Since the 3D-printed E. coli expressing curli only (cellulose–/curli+) also express GFP\nconstitutively, we used confocal microscopy to assess whether GFP\nexpression was observed at the lower depths of the biofilm ( Figure S2 ). Confocal microscopy images indicated\nthat cells at the lower depths of the biofilm expressed GFP robustly,\nwhereas the GFP signal was not detected in the negative control cellulose+/curli+\n3D-printed biofilm lacking GFP production. These results indicated\nthat even the deepest biofilm cells still showed viability in the\n3D-printed cellulose–/curli+ biofilm, despite the low oxygen\nconcentration. Overall, we can conclude that 3D-printed biofilms\nthat express\neither curli alone or both cellulose and curli closely resemble natural\nbiofilm systems with respect to limited oxygen penetration through\nthe biofilm matrix and the presence of a thicker anaerobic zone in\nthe bottom layers of the biofilm. Extracellular Matrix Composition\nGoverns the Emergent Biological\nEndurance of 3D-Printed Biofilms Bacteria in biofilms display\nbiological endurance such as resistance to antimicrobials and disinfectants\nthat their planktonic counterparts do not possess due to their marked\n3D nature and the presence of the extracellular matrix that acts as\na physical or chemical diffusion barrier. 11 , 47 , 48 While it is known that the molecular composition\nof the biofilm matrix has a major influence on the emergent viscoelastic mechanical properties of bacteria in biofilms, 13 the influence of the molecular composition of\nthe biofilm matrix on the emergent biological properties\nremains unclear. We aimed to solve this question by 3D printing\ndifferent E. coli strains expressing\ncurli only (cellulose–/curli+), both curli and cellulose (cellulose+/curli+),\nor neither curli nor cellulose (cellulose–/curli−) in\ntheir biofilm matrix and testing their biological endurance against\nexposure to varying concentrations of the widely used disinfectants\nethanol and Virkon S ( Figure 5 ). A schematic of the methodology for performing endurance\nassays is shown in Figure S3 . The colony-forming\nunits (cfus) of 3D-printed E. coli strains\nexpressing neither cellulose nor curli (cellulose–/curli−),\nonly curli (cellulose–/curli+), or both cellulose and curli\n(cellulose+/curli+) reached between 8 and 12 log(cfu/mL) after 7 days\nof incubation at room temperature ( Figure 5 ). Treatment with ethanol at concentrations\nof 30–70% resulted in a dose-dependent reduction in the number\nof viable cells in each of the 3D-printed E. coli biofilms tested ( Figure 5 A). Since the bacterial strains each grew at different rates\nwithin the printed biofilms, we assessed resistance to antibacterial\ntreatments by comparing log reductions in cfu/mL values. At the highest\nconcentrations of 50 and 70% ethanol, 3D-printed E.\ncoli expressing neither cellulose nor curli (cellulose–/curli−)\nwas more sensitive to ethanol than the 3D-printed E.\ncoli expressing cellulose and/or curli, leading to\nan approximate 5–6 log reduction in cfus/mL. 3D-printed E. coli expressing only curli (cellulose–/curli+)\nwas more resistant to ethanol than the cellulose–/curli–\nstrain, as shown by a smaller ∼1.5–3 log reduction in\ncfu/mL upon treatment with 50–70% ethanol. In comparison to\nthe cellulose–/curli+ strain, the 3D-printed E. coli expressing both cellulose and curli (cellulose+/curli+)\nrevealed a lesser resistance to ethanol, with a ∼3–5\nlog reduction in cfu/mL upon treatment with 50–70% ethanol.\nThus, the presence of curli fibers in the 3D-printed biofilms enhanced\nthe resistance of E. coli to ethanol.\nThe cellulose+/curli+ strain, which produced cellulose fibers in addition\nto the curli matrix, demonstrated a slightly reduced resistance to\nethanol. These resistance phenotypes of the cellulose+/curli+ strain\ncould potentially reflect inherent differences in biological resistance\ndue to a different strain background as well as physical differences\nderiving from biofilm matrix composition; future experiments to tease\napart the contributions to resistance from matrix polymers could resolve\nthis question. In E. coli biofilms,\ncurli fibers present in the biofilm extracellular matrix are known\nto dominate the biofilm mechanical behavior, 13 , 49 reflecting their strong internal structure and pronounced viscoelasticity.\nThese properties provided by the curli fibers may be the cause of\nthe emergent resistance to ethanol demonstrated by the resultant 3D-printed\nbiofilms. Figure 5 Emergent disinfectant resistance of 7 day-old 3D-printed biofilms\nto a 10 min exposure to varying concentrations of (A) ethanol or (B)\nVirkon S. Gray bars depict cellulose–/curli–, orange\nbars depict cellulose–/curli+, and blue bars depict cellulose+/curli+\n3D-printed E. coli . The control condition\nindicates treatment with sterile saline (0.9% (w/v) sodium chloride).\nNs, not significant; ** p < 0.01, *** p < 0.001, **** p < 0.0001. Statistical significance\nwas assessed by comparing the disinfectant samples with their respective\ncontrol sample using Student’s t -test ( p < 0.05; statistically significant). A similar trend of resistance to the disinfectant was observed\nupon the treatment of 7 day-old 3D-printed biofilms with Virkon S,\na broad-spectrum disinfectant ( Figure 5 B). Exposure to increasing concentrations of Virkon\nS (0.1–4.0%) resulted in a dose-dependent reduction in the\nnumber of viable bacteria in each of the 3D-printed biofilms. The\nlowest concentration of Virkon S (0.1%) had little effect, leading\nto an approximate 1.6-log reduction in cfu/mL only in 3D-printed biofilms\nexpressing both cellulose and curli (cellulose+/curli+), whereas the\nhighest concentration of Virkon S tested (4.0%) resulted in 100% killing\nof bacteria in all of the 3D-printed biofilms. The protective nature\nof curli and/or cellulose in conferring resistance was observable\nat Virkon S concentrations between 0.2 and 0.5%. At these concentrations,\nthe reduction in cfus in comparison to untreated control samples was\nconsistently 1–2 log lower for the 3D-printed biofilms expressing\neither only curli and no cellulose (cellulose–/curli+) or both\ncellulose and curli (cellulose+/curli+) in comparison to the reductions\nin cfus measured for the 3D-printed biofilms expressing neither cellulose\nnor curli (cellulose–/curli−). The biofilms expressing\nextracellular-matrix polymers were viable at higher concentrations\nof Virkon S; the highest concentration of Virkon S that allowed detectable\nviable cfus for each strain was 1.0% for cellulose+/curli+ biofilms\nand 0.5% for cellulose–/curli+ biofilms, in comparison to 0.3%\nfor cellulose–/curli– biofilms. Moreover, at these concentrations\nthe expression of both cellulose and curli (cellulose+/curli+) in\nthe 3D-printed biofilm conferred a greater resistance to Virkon S\nthan the expression of curli alone without cellulose (cellulose–/curli+). We evaluated the sensitivity of 3D-printed biofilms to either ethanol\n[70% (v/v)] or Virkon S [0.5% (w/v)] at 2 days of growth, an early\ntime point where biofilm-matrix components would be less expressed.\nThe cfus of 3D-printed E. coli strains\nexpressing neither cellulose nor curli (cellulose–/curli−),\nonly curli and no cellulose (cellulose–/curli+), or both cellulose\nand curli (cellulose+/curli+) reached approximately 7–10 log\ncfu/mL after 2 days of incubation at room temperature ( Figure 6 ). These values were approximately\n10–100-fold lower than the values seen after 7 days of growth.\nBoth the 3D-printed biofilms expressing neither cellulose nor curli\n(cellulose–/curli−) or only curli and no cellulose (cellulose–/curli+)\nshowed complete sensitivity to either ethanol or Virkon S treatment,\nindicating the absence of sufficient curli production to be protective\nat this time point. We have previously shown that curli production\nin these 3D-printed E. coli has occurred\nby 3 days of incubation at room temperature at levels sufficient to\nprevent citrate-based dissolving of the alginate matrix. 37 In contrast, 3D-printed biofilms expressing\nboth cellulose and curli (cellulose+/curli+) showed marked resistance\nto both ethanol and Virkon S resulting in a survival of about 10 2 –10 3 cfu/mL. This result indicates that\ncellulose production could happen much earlier in the 3D-printed biofilms\nthan the curli production such that this resistance would be an emergent\nproperty resulting from cellulose in the matrix. Alternatively, the\ncellulose+/curli+ strain could produce more curli at earlier time\npoints. Figure 6 Sensitivity of 2 day-old 3D-printed biofilms to a 10 min exposure\nto (A) ethanol [70% (v/v)] or (B) Virkon S [0.5% (w/v)]. Gray bars\ndepict cellulose–/curli–, orange bars depict cellulose–/curli+,\nand blue bars depict cellulose+/curli+ 3D-printed E.\ncoli . The control condition indicates treatment with\nsterile saline [0.9% (w/v) sodium chloride]. **** p < 0.0001. Statistical significance between the ethanol or Virkon\nS samples versus the control samples was assessed with Student’s t -test ( p < 0.05; statistically significant). Thus, 3D prints of E. coli expressing\nbiofilm matrix polymers (containing cellulose and/or curli) were more\nresistant to disinfectants than the 3D prints containing nonbiofilm-forming E. coli (neither cellulose nor curli). Based on our\ndata, it is evident that the extracellular matrix composition, particularly\nthe presence of curli fibers, plays an important role in development\nof biological endurance against disinfectants in 3D-printed E. coli biofilms. This result is in agreement with\nthe results of previous studies that show that curli rather than cellulose\nexpression is directly related to the emergent resistance of E. coli against sanitizers in 2D models. 16 , 50 Effect of Tuning the Cell- and Matrix-Component Densities on\nEmergent Biofilm Endurance In natural biofilms, the living\n(bacterial cells) and the nonliving components (extracellular-matrix\ncomponents such as cellulose and curli) are spatially patterned, which\nhas been hypothesized to give rise to their emergent behavior under\nextreme conditions. 12 , 15 , 51 However, the exact nature of such spatial patterns and the influence\nof changing the spatial patterns on the emergent behavior is not fully\ncharacterized. We investigated this topic by 3D printing monoculture\nor coculture inks containing different ratios of different biofilm-forming\nbacteria. First, we studied the influence of altering cell density\nusing a step increase/decrease function ( Figure 7 A) on the emergent resistance to ethanol\n[70% (v/v)]. Bacteria expressing only curli and no cellulose (cellulose–/curli+)\nwere 3D-printed as four-layered constructs with higher cell density\nin the bottom two layers and lower cell density in the top two layers\nor vice versa. In both configurations, the final cell density reached\napproximately ∼9 log cfu/mL after 7 days of incubation at room\ntemperature. Treatment with ethanol resulted in ∼2 log reduction\nof cfu/mL in the configuration with higher cell density at the bottom\nand lower cell density at the top and ∼3.5 log reduction of\ncfu/mL in the configuration with lower cell density at the bottom\nand higher cell density at the top. Thus, the biofilm design with\nhigher cell density at the bottom is more resistant to ethanol than\nthe design with lower cell density at the bottom. This design with\nhigher resistance also adopts the pattern of cell density distribution\nseen for most natural biofilms. 10 Figure 7 Effect of tuning\nthe (A) bacterial density, (B) curli density,\nand (C) cellulose density on emergent endurance to ethanol [70% (v/v)].\n(A) Bacterial density was varied by 3D printing cellulose–/curli+\nbioinks as four-layered constructs with higher cell density (shown\nin orange color) in the bottom two layers and lower cell density (shown\nin white color) in the top two layers or vice versa. (B) Curli density\nwas varied in a step function by 3D printing two layers of bioink\ncontaining E. coli expressing neither\ncellulose nor curli (cellulose–/curli–; shown in gray\ncolor) overtop of two layers of 3D-printed bioink containing E. coli expressing only curli and no cellulose (cellulose–/curli+;\nshown in orange color), or vice versa. Curli density was also varied\nin a gradient function by 3D printing two layers of coculture bioinks\ncontaining E. coli expressing neither\ncellulose nor curli (cellulose–/curli−) mixed with E. coli expressing only curli and no cellulose (cellulose–/curli+)\nin a ratio of 3:1 overtop of two layers of 3D-printed coculture bioinks\ncontaining E. coli expressing neither\ncellulose nor curli (cellulose–/curli−) mixed with E. coli expressing only curli and no cellulose [(cellulose–/curli+)\nin a ratio of 1:3] or vice versa. (C) Cellulose density was varied\nby 3D printing four-layered constructs of coculture bioinks containing E. coli expressing neither cellulose nor curli (cellulose–/curli−)\nmixed with G. hansenii (cellulose+/curli−)\nin ratios of 1:1, 1:10, or 1:25 ( E. coli / G. hansenii ). (A–C) The control\nconditions indicate treatment with sterile saline [0.9% (w/v) sodium\nchloride]. Ethanol treatment resulted in statistically significant\nreduction of bacterial cfus in each of the experiments, **** p < 0.0001. Statistical significance was assessed by\ncomparing the disinfectant samples with their respective control samples\nusing Student’s t -test ( p < 0.05; statistically significant). Next, we studied whether tuning the curli density would lead to\ndifferences in the emergent resistance to ethanol. For this, we employed\na step increase/decrease configuration of curli density during biofilm\ndesign, that is, we 3D-printed two layers of bioink containing E. coli expressing neither cellulose nor curli (cellulose–/curli−)\novertop of two layers of 3D-printed bioink containing E. coli expressing only curli and no cellulose (cellulose–/curli+)\nor vice versa ( Figure 7 B). We also employed a gradient increase/decrease configuration of\ncurli density, in which we 3D-printed the top two layers using coculture\nbioinks containing E. coli expressing\nneither cellulose nor curli (cellulose–/curli−) mixed\nwith E. coli expressing only curli\nand no cellulose (cellulose–/curli+) in a ratio of 3:1, overtop\nof two layers of 3D-printed coculture bioinks containing E. coli expressing neither cellulose nor curli (cellulose–/curli−)\nmixed with E. coli expressing only\ncurli and no cellulose [(cellulose–/curli+) in a ratio of 1:3]\nor vice versa. In each configuration, the final cell density reached\napproximately ∼9 log cfu/mL after 7 days of incubation at room\ntemperature. Treatment with ethanol resulted in ∼3 log reduction\nof cfu/mL in each of the four printed configurations. Thus, tuning\nthe curli density by adjusting the proportion of curli-expressing\nbacteria in each layer during the 3D printing had no effect on the\nemergent resistance against ethanol. Data from our earlier experiments\nindicated that cellulose production\nin 3D-printed biofilms could lead to emergence of resistance to ethanol\nand Virkon S ( Figure 6 ). In order to further understand this phenomenon, we evaluated whether\ntuning the cellulose density could lead to differences in the emergent\nresistance to ethanol. To achieve this, we employed 3D printing of\ncoculture bioinks containing E. coli expressing neither cellulose nor curli (cellulose–/curli−)\nmixed with Gluconacetobacter hansenii , a bacterium that produces copious cellulose but not curli fibers. 52 The two strains were mixed in ratios of 1:1,\n1:10, or 1:25 and printed to make four-layered constructs. To study\ntheir emergent resistance to ethanol, the individual survival rates\nof E. coli and G. hansenii were determined by plating the samples onto Luria–Bertani\n(LB) agar supplemented with chloramphenicol (selective for E. coli ) or HS agar supplemented with acetic acid\n(selective for G. hansenii ). Determination\nof cfu values revealed that both the E. coli and G. hansenii strains were able\nto grow in the 3D-printed coculture inks ( Figure 7 C). As the ratio of E. coli to G. hansenii increased from 1:1\nto 1:25 in the bioinks, the cfus of G. hansenii in the fully grown mock-treated 3D prints increased from ∼3\nto 6 log cfu/mL, and the cfus of E. coli decreased from ∼6 to 2 log cfu/mL. Following ethanol treatment,\na survival rate of 0 cfus/mL was measured for both E. coli and G. hansenii in each 3D-printed biofilm, indicating that none of the 3D-printed\nbiofilms displayed emergent resistance to ethanol for either species.\nThus, cellulose alone in the biofilm matrix cannot confer emergent\nendurance against ethanol to 3D-printed biofilms in our tested conditions. G. hansenii has been shown to produce higher-ordered\ncrystalline cellulose, whereas E. coli produces lesser-ordered amorphous forms of cellulose. 52 SEM images of our G. hansenii and E. coli Nissle (cellulose+/curli+)\ncolony biofilms also revealed differences in the biofilm architecture\nof the two strains ( Figure S4 ). Hence,\ndifferences in the network and the microstructure properties of G. hansenii cellulose in comparison with the E. coli cellulose could contribute to the absence\nof protection to E. coli against ethanol\nin our experiments. In the future, a better understanding into this\nphenomenon could be obtained using a strain of E. coli that produces high amounts of cellulose but not curli. Physical Stability\nof 3D-Printed Biofilms 3D-printed\nbiofilms could be used in various beneficial applications including\nbioremediation, wastewater treatment, or probiotic coatings on medical\ndevices and surfaces to prevent colonization by pathogenic bacteria.\nIn order to be employed in such applications, reversible adhesion\nof 3D-printed biofilms to different surfaces and physical stability\nare important aspects. We tested these parameters by removing fully\ngrown (7 days old) 3D-printed E. coli (cellulose+/curli+) biofilms from agar and attaching them to fresh\nsurfaces composed of bacterial cellulose, glass, or polystyrene ( Video S1 ). The 3D-printed biofilms displayed\nreversible attachment to fresh bacterial cellulose as well as glass\nand polystyrene surfaces ( Figure 8 ). Figure 8 Adhesion of 3D-printed E. coli cellulose+/curli+\nbiofilms to bacterial cellulose, glass, and polystyrene surfaces after\ndetachment from agar. Since our 3D-printed\nbiofilms adhered to bacterial cellulose, which\nis sustainably produced and possesses excellent mechanical properties\nincluding remarkable tensile strength (73–194 MPa) and toughness\n(2–25 MJ m –3 ), 19 , 31 we studied\nthe deformation of the 3D-printed biofilms. Bacterial cellulose has\nbeen found to be a flexible substrate to support hydrogel-based living\nmaterials. 31 We further subjected the 3D-printed\nbiofilms on the bacterial cellulose surface to manual distortions\nby folding, twisting, and crushing them ( Figure S5 ). The 3D-printed biofilms resumed their original shapes\nupon unfolding, untwisting, and uncrushing, indicating their high\nphysical stability. Thus, 3D-printed biofilms may be used as physically\nresilient materials for desired applications merely by attaching them\nto mechanically robust surfaces such as bacterial cellulose."
} | 10,746 |
39095478 | PMC11297162 | pmc | 683 | {
"abstract": "Methanogenic hydrocarbon degradation can be carried out by archaea that couple alkane oxidation directly to methanogenesis, or by syntrophic associations of bacteria with methanogenic archaea. However, metagenomic analyses of methanogenic environments have revealed other archaea with potential for alkane degradation but apparent inability to form methane, suggesting the existence of other modes of syntrophic hydrocarbon degradation. Here, we provide experimental evidence supporting the existence of a third mode of methanogenic degradation of hydrocarbons, mediated by syntrophic cooperation between archaeal partners. We collected sediment samples from a hot spring sediment in Tengchong, China, and enriched Hadarchaeota under methanogenic conditions at 60 °C, using hexadecane as substrate. We named the enriched archaeon Candidatus Melinoarchaeum fermentans DL9YTT1. We used 13 C-substrate incubations, metagenomic, metatranscriptomic and metabolomic analyses to show that Ca . Melinoarchaeum uses alkyl-coenzyme M reductases (ACRs) to activate hexadecane via alkyl-CoM formation. Ca . Melinoarchaeum likely degrades alkanes to carbon dioxide, hydrogen and acetate, which can be used as substrates by hydrogenotrophic and acetoclastic methanogens such as Methanothermobacter and Methanothrix .",
"introduction": "Introduction Microbial transformation of hydrocarbons to methane is an environmentally significant phenomenon, which occurs in a wide variety of electron acceptor-limited habitats, including oil reservoirs, coal deposits, oil contaminated groundwater, and marinedeep sediments 1 – 6 . This process occurs in a series of steps and requires close syntrophic associations between fermentative bacteria and methanogenic archaea 2 – 6 . The well-known initial degradation of multi-carbon alkanes is carried out by syntrophic bacteria via addition to fumarate yielding alkyl-substituted succinates by alkylsuccinate synthases (ASS) 1 – 6 . The alkyl-succinates are further transferred to intermediates, such as acetate, hydrogen or carbon dioxide, which are then consumed by hydrogenotrophic and/or acetoclastic methanogenic archaea for methane production. These anaerobic alkane-degrading bacteria reported to be involved in methanogenic hydrocarbon degradation include Desulfatibacillum , Desulfatiferula , Desulfosporosinus , Smithellla , and their methanogenic partners including Methanobacterium , Methanococcus , Methanoculleus , Methanosarcina , Methanothermobacter , Methanothrix 1 – 6 . Later, an alternative mode initially activating short- to long-chain alkanes into corresponding alkyl-CoMs by ACRs has been confirmed in cultured anaerobic alkane-oxidizing archaea, which belong to lineages within the phylum Halobacteriota 7 – 12 . The long-chain alkane-degrading archaea, Ca . Methanoliparia, encodes tetrahydromethanopterin S-methyltransferase (MTR) and an additional canonical methyl-coenzyme M (CoM) reductase (MCR), which allows to couple alkane degradation and methanogenesis in the single archaeal cell 12 , 13 . This is the other mechanism contributing to hydrocarbon degradation under methanogenic conditions. Other acr -containing archaea, such as Ca . Alkanophaga 14 , Ca . Argoarchaeum 7 , Ca . Cerberiarchaeon 11 , Ca . Ethanoperedens 8 and Ca . Syntrophoarchaeum 10 lack mcr and mtr genes, which oxidizes alkanes to carbon dioxide and transfer the reducing equivalents to partner bacteria that perform sulfate reduction. In addition, it has recently been revealed that Ca . Cerberiarchaeon from the Hadarchaeota phylum can thrive on long-chain alkanes (hexadecane) under sulfate-reducing conditions, transferring electrons produced during alkane degradation to sulfate-reducing bacteria 11 . Furthermore, metagenomic studies found metagenome-assembled genomes (MAGs) with acr but not mcr genes distributed widely in methanogenic environments 15 – 19 . For example, several MAGs in the phylum Hadarchaeota (WYZ-LMO4-6, JZ-1 bin_103, and JZ-1 bin_103) were constructed from terrestrial hot springs 18 , 19 ; two MAGs (BA1 and BA2) in the class Bathyarchaeia (phylum Thermoproteota ) were constructed from a coalbed well 12 , 17 , 19 – 21 . However, their eco-physiological role on the process of hydrocarbon degradation has not been reported. Here we enriched an archaeon from the phylum Hadarchaeota , which grew with long-chain alkanes to generate methane in cooperation with methanogenic archaeon.",
"discussion": "Results and discussion Methanogenesis incubation with long-chain alkanes We collected sediments from Tengchong hot spring (Supplementary Fig. 1 ), which releases heated (65.7 °C), sulfate-depleted fluids containing various alkanes (Supplementary Data 1 ). We initially incubated these sediments in sulfate-free artificial freshwater medium (containing NaHCO 3 ) with tetradecane as substrate at 60 °C, and observed a consecutive methane accumulation (data not shown). Then the culture was supplemented with hexadecane and sulfate-free artificial freshwater medium repeatedly during ~2 years of incubation (details see Materials and Methods, Fig. 1a ). Continuous production of methane and carbon dioxide was observed during the incubation with hexadecane (Supplementary Fig. 2 , Supplementary Data 2 ), while the non-alkane groups showed much lower methane and carbon dioxide production (Supplementary Fig. 3 ). when addition with 1,2- 13 C-labeled hexadecane during incubation at day 750, similar increasing of methane and carbon dioxide was observed, with 195.9–272.7 µmol methane and 62.8-76.6 µmol carbon dioxide production at day 810 (Fig. 1b, c ). The ratio of methane to carbon dioxide produced was approximately 3:1, which was close to the theoretical value for methanogenic hexadecane degradation as depicted here: 1 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$4\\,{{{{\\rm{C}}}}}_{16}{{{{\\rm{H}}}}}_{34}+{30} \\, {{{{\\rm{H}}}}}_{2}{{{\\rm{O}}}}\\to 49 \\, {{{{\\rm{CH}}}}}_{4}+15 \\, {{{{\\rm{CO}}}}}_{2}(\\Delta {{{\\rm{G}}}}^{ {\\circ} \\;\\,{\\prime}}=-438.5\\,{{{\\rm{kJ}}}}/{{{\\rm{mol}}}}\\,{{{\\rm{hexadecane}}}})$$\\end{document} 4 C 16 H 34 + 30 H 2 O → 49 CH 4 + 15 CO 2 ( Δ G ∘ ′ = − 438.5 kJ / mol hexadecane ) Fig. 1 Scheme of incubation experiments and formation of methane and carbon dioxide from long-chain alkanes. Scheme of incubation experiments ( a ), concentration of methane ( b ) and carbon dioxide ( c ), carbon isotope of methane ( d ) and carbon dioxide ( e ), and concentration of 13 C-labled methane ( f ) and carbon dioxide ( g ) in the cultures. r1 and r2 represent two replicates. The samples were incubated at 60 °C in Dec. 2021. Alkane groups: two bottles of slurry samples with the addition of 1% (volume/volume) tetradecane and fresh medium (with NaHCO 3 ) during the initial cultivation; after 228 days culturing, 1% (volume/volume) tetradecane and 20% (v/v) fresh medium (with NaHCO 3 ) were supplemented; after 386 days culturing, 0.1% (v/v) hexadecane and 20% (v/v) fresh medium (with NaHCO 3 ) were supplemented, and the headspace in bottles was replaced with nitrogen; at day 614, after mixing the two samples thoroughly, divide the mixture into two equal parts, 0.1% (v/v) hexadecane and 100% (v/v) fresh medium (without NaHCO 3 ) were supplemented, the headspace in the bottles was replaced with nitrogen; after 676 days culturing, 0.1% (v/v) hexadecane and 10% (v/v) fresh medium (without NaHCO 3 ) were supplemented, and the headspace in the bottles was replaced with nitrogen; after 750 days culturing, 0.1% (v/v) 1,2- 13 C-labeled hexadecane and 10% (v/v) fresh medium (without NaHCO 3 ) were supplemented, and the headspace in the bottles was replaced with nitrogen. The methane production rate was ~0.04–0.05 µmol methane per ml of slurry per day. The δ 13 C values of methane and carbon dioxide also increased substantially (Fig. 1d, e ). The amount of 13 C-labeled methane produced were 3.9–6.2 µmol, and the amount of 13 C-labeled carbon dioxide produced was 1.3–1.8 µmol (Fig. 1f, g ). The ratio of 13 C-labeled methane to 13 C-labeled carbon dioxide produced was also approximately 3:1. Ca . Melinoarchaeum spp. archaea dominate in the alkane cultures Genomic DNA extracted from the original sample and incubation experiments were subjected to the metagenomic analysis. In alkane incubations, Hadarchaeota stood out as the most dominant microbial group, with relative abundances of 42.8% and 56.5%; Thermoproteota (12.3–10.2%), and Methanobacteriota (7.6–5.1%) were also dominant; Chloroflexota (3.2–2.7%), Bipolaricaulota (1.7–1.4%), Patescibacteria (1.2–0.9%), Halobacteriota (1.2–0.8%), Asgardarchaeota (1.4–0.5%) and Nitrospirota (0.4–0.6%) showed lower abundance (Fig. 2a and Supplementary Data 3 ). A total of 68 high-quality MAGs (completeness > 80%, contamination < 5%) were obtained, belonging to the phyla Hadarchaeota , Halobacteriota , Methanobacteriota , Thermoplasmatota , Thermoproteota , Acidobacteriota , Bipolaricaulota , Caldisericota , Chloroflexota , Cyanobacteria , Desulfobacterota , Elusimicrobiota , Firmicutes , Nitrospirota , Verrucomicrobiota and WOR-3. The MAGs TC202112_7_001 and TC202112_8_001 of the Hadarchaeota group exhibited the highest relative abundances at 41.6% and 55.5%, respectively (Fig. 2b ). Both were classified within the Hadarchaeia class, Hadarchaeales order, WYZ-LMO6 family, and WYZ-LMO genus based on the Genome Taxonomy Database (GTDB) taxonomy 22 (Supplementary Data 3 ). Both MAGs carried the acr genes (Supplementary Data 4 ). According to the metabolic discussion below we named the dominant organism represented by these two MAGs as Ca . Melinoarchaeum fermentans DL9YTT1. In contrast, the abundance of Hadarchaeota MAGs was less than 0.6% in groups without the addition of alkanes, and we did not detect any Hadarchaeota MAGs in the metagenomic sequences of the original slurry (Supplementary Fig. 4 ). Fig. 2 Ca . Melinoarchaeum fermentans sp. was abundant in alkane cultures. a taxonomies of the MAGs in phylum level; Other, including the unclassified MAGs and MAGs with relative abundance <0.5%; Unmapped, including the relative abundance of reads which did not map to MAGs. b the relative abundance of Ca . Melinoarchaeum fermentans DL9YTT1, Methanothermobacter and Methanothrix . The samples for metagenomic and metatranscriptomic sequencing were collected after culturing for 791 days. c the abundance of Ca . Melinoarchaeum determined by Q-PCR of their 16S rRNA genes; the error bars are obtained from triplicate Q-PCR reaction. r1 and r2 represent two replicates. Quantitative PCR (Q-PCR) revealed that following alkane incubations, the abundance of Ca . Melinoarchaeum spp. increased by 26–136 folds to 1.7–9.1 × 10 7 copies/ml compared to the sample collected on day 0 (Fig. 2c ). These indicates that the growth of this archaeon depends on the availability of long-chain alkanes. However, when hexadecane was added under sulfate conditions, the growth of Ca .Melinoarchaeum spp. was not stimulated (Supplementary Fig. 5 ), indicating that their growth cannot be coupled with sulfate reduction. Fluorescence in situ hybridization with specific probes showed that Ca . Melinoarchaeum spp. grows in mono-species filaments, and does not exhibit tight association with other microbial cells (Supplementary Fig. 6 ). The genome of Ca . Melinoarchaeum fermentans DL9YTT1 is highly similar to the previously published MAG JZ-2_bin_199 18 , which was also binned from Tengchong hot spring sediment metagenome with average nucleotide identity (ANI) of 90.0%. We combined the MAGs obtained in this study with data from public databases, resulting in a total of 48 MAGs of the phylum Hadarchaeota (Fig. 3a ). These MAGs group into one class, two orders and six families based on GTDB taxonomy (Supplementary Data 5 ). Together with 12 other MAGs, Ca . Melinoarchaeum spp. forms the genus Ca . Melinoarchaeum. The genus Ca . Melinoarchaeaum and their sister genus Ca . Cerberiarchaeum form the family Ca . Cerberiarchaeaceae 11 . Fig. 3 Phylogenomic affiliation of the MAGs, AcrAs and McrAs. a phylogenomic affiliation of the Hadarchaeota MAGs based on 37 conserved protein sequences and using 233 representative archaeal genomes. b phylogenomic affiliation of the AcrAs and McrAs based on the alignments with 461 aligned positions. The tree branches were classified into four distinct groups: the ACR clade; TACK clade; Class I&II clade and Class III clade, the last three are canonical McrA clades. ACR clade contains the AcrAs of Ca . Melinoarchaeum spp.; Class I&II clade contains the McrAs of Methanothermobacter and Methanothrix . Alignments were generated using MAFFT 43 and then filtered with trimAl 44 , and the trees were built by the IQ-Tree 45 method with the model LG + C60 + F + G with 1000 bootstrap replicates. Bootstrap values > 90% shown in gray dots. Ca . Melinoarchaeum spp. activate alkanes with highly transcribed ACRs In the entire metagenome, the ACR genes only present in two MAGs of Ca . Melinoarchaeum spp. (Supplementary Data 3 ). Both MAGs contain three copies of the acr A gene. Two of these acr A genes form a continuous operon with the acr B and acr G genes, while the third does not form an operon (Supplementary Data 4 ). Phylogenetic analysis of the AcrA-subunits showed that Hadarchaeota form a distinct cluster next to those of Ca . Bathyarchaeia and Ca . Methanoliparia (Fig. 3b ). We also analyzed the gene expression level of Ca . Melinoarchaeum spp. during alkanes degradation. Ca . Melinoarchaeum spp. exhibited the highest transcriptional activity among all MAGs, with a relative abundance of 36.4 and 43.4% (Fig. 2b , Supplementary Data 3 ). The acr genes of Ca . Melinoarchaeum spp. were highly expressed (with fragments per kilobase of transcript per million mapped reads, FPKM = 179–852) (Fig. 4a , Supplementary Data 4 ). Fig. 4 Metabolic model of Ca . Melinoarchaeum fermentans sp., Methanothermobacter and Methanothrix. a genome-inferred model and corresponding gene expression patterns for hexadecane degradation in Ca . Melinoarchaeum fermentans sp. The steps involving substrate activation (red arrows), β-oxidation (blue arrows), Wood-Ljungdahl pathway (yellow arrows) and energy conservation (black arrows). The non-confirmed steps for transform hexadecyl-CoM to hexadecanoate are indicated in dot line. b genome-inferred model and corresponding gene expression patterns for hydrogenotrophic methanogenesis of genus Methanothermobacter ; c genome-inferred model and corresponding gene expression patterns for acetoclastic methanogenesis of Methanothrix . The names of genes not found in the genome were shown in gray font. Fd red , reduced ferredoxins; Fd ox , oxidized ferredoxins. Squares indicate different gene transcription levels. List of related genes and gene transcription levels was showed in Supplementary Data 4 . We analyzed the hexadecane cultures for the formation of products of the ACR-based alkane activation using Q-Exactive Plus Orbitrap mass spectrometry (Fig. 5 ). We found peaks of hexadecyl-CoM ( m/z = 365.21884) and its fragmentations products, including hexadecyl-thiol ( m/z = 257.23082), ethenesulfonate ( m/z = 106.98079) and sulfonate ( m/z = 80.96517) (Fig. 5a, b ). Moreover, cultures supplied with 1,2- 13 C-hexadecane contained a mass peak of 1,2- 13 C-hexadecyl-CoM ( m/z = 367.22548) and the fragment of 1,2- 13 C-hexadecyl-thiol ( m/z = 259.23739), their mass is shifted by 2 units compared to the unlabeled group (Fig. 5c, d ). The retention times of the extracted metabolites of hexadecyl-CoM, as determined by HPLC-MS/MS, closely matched those of synthesized authentic standards (Supplementary Fig. 7 ). In contrast, the respective alkylsuccinates, the activation products formed by ASS were not detected in alkane cultures. This is in agreement with the lack of ass genes in Ca . Melinoarchaeum spp. and the entire assemblies within the whole metagenome of alkane cultures 23 – 25 . These findings excluded the activation of alkanes through fumarate addition pathway in our cultures. Fig. 5 Identification of the initial intermediate for alkane degradation. a MS analysis of extracts of culture with unlabeled hexadecane showed a peak at m / z = 365.21884 (red), which matches the authentic hexadecyl-CoM standard (blue). b Fragmentation of the isolated m / z range of 365.0–365.4 yields hexadecyl-thiol (C 16 H 33 S − , m / z = 257.23082), ethenesulfonate (C 2 H 3 SO 3 − , m / z = 106.98079) and sulfonate (HSO 3 − , m / z = 80.96517). These peaks are also produced by hexadecyl-CoM standards. c MS analysis of extracts of culture with 1,2- 13 C-labeled hexadecane yielded a peak at m / z = 367.22548 (red). d Fragmentation of the isolated m / z range of 367.0–367.4 yielded 1,2- 13 C-labeled hexadecyl-thiol (C 16 H 33 S − , m / z = 259.23739), unlabeled ethenesulfonate (C 2 H 3 SO 3 − , m / z = 106.98084) and sulfonate (HSO 3 − , m / z = 80.96515). The mass errors for all mass peaks are <5 ppm. Methanogenic alkane degradation in cooperation with multiple archaea The degradation of the alkyl-CoMs generated by the ACRs involves their conversion into fatty acids. The pathways and enzymes involved in this reaction are yet unknown 9 , 10 , 20 , 26 (Fig. 4a ). The Ca . Melinoarchaeum spp. encodes long-chain acyl-CoA synthetase (ACSL/FadD) and a complete β-oxidation pathway. This allows the activation of long-chain fatty acids as CoA-bound acyl units, and splitting of those compounds into acetyl-CoA units. The acetyl-CoA decarbonylase/synthase (ACS/CODH) complex could work in the cleaving of acetyl-CoAs into carbon dioxide and methyl-H 4 MPT, then the H 4 MPT-bound methyl groups are fully oxidized to carbon dioxide with the downstream part of the Wood-Ljungdahl (WL) pathway. Ca . Melinoarchaeum fermentans sp. and all other ACR-containing Hadarchaeota do not encode MTR and canonical MCR, the crucial elements of methanogenesis (Supplementary Data 5 ). In contrast, Ca . Melinoarchaeum spp. might use the reduced ferredoxin, F 420 H 2 and NADH produced from β-oxidation and WL pathway, to form hydrogen (Fig. 4a ). Enzymes for this process are ferredoxin hydrogenase (Mvh), the beta subunit of F 420 -reducing NiFe-hydrogenase (FrhB) and sulphohydrogenase (HydAD). The acetyl-CoA synthetase (ACSS) might be used to convert acetyl-CoAs, the product of β-oxidation, into acetate. Ca . Melinoarchaeum spp. encodes related genes and all of them were highly expressed (FPKM > 100) (Fig. 4a , Supplementary Data 4 ). Based on the metabolic pathway predication and gene expression in our cultures, two groups of methanogens capable of acetate fermentation and carbon dioxide reduction were most likely associated with the alkane degradation process (Fig. 4b, c and Supplementary Data 3 ). The MAGs of the genus Methanothermobacter (TC202112_7_002 and TC202112_8_003) in the phylum Methanobacteriota and the genus Methanothrix (TC202112_7_009 and TC202112_8_014) in phylum Halobacteriota were detected in our long-chain alkane cultures, with relative abundance of 4.8–6.9% and 0.6–0.9% respectively. The Methanothermobacter and Methanothrix MAGs contained the complete pathway of hydrogenotrophic methanogenesis and acetoclastic methanogenesis respectively. The mcr genes of Methanothermobacter exhibited high expression levels in alkane cultures, with average FPKM values of 1254–1466; the expression of mcr genes in Methanothrix was relatively lower, with average FPKM values of 39–47 (Fig. 4b, c and Supplementary Data 4 ). Therefore, we speculate that the degradation proceeds in a syntrophic interactions of Ca . Melinoarchaeum spp. and methanogens. Ca . Melinoarchaeum spp. might degrade alkanes to hydrogen, acetate and carbon dioxide. Then, the hydrogenotrophic methanogens ( Methanothermobacter ) might use hydrogen and carbon dioxide for methane production; and acetoclastic methanogens ( Methanothrix ) might use the acetate for methane and carbon dioxide production. These steps might degrade the model compound hexadecane according to: 2 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\begin{array}{c}4 \\, {{{{\\rm{C}}}}}_{16}{{{{\\rm{H}}}}}_{34}+74 \\, {{{{\\rm{H}}}}}_{2}{{{\\rm{O}}}}\\to 27 \\, {{{{\\rm{CH}}}}}_{3}{{{{\\rm{COO}}}}}^{-}+27 \\, {{{{\\rm{H}}}}}^{+}+88 \\, {{{{\\rm{H}}}}}_{2}+10 \\, {{{{\\rm{CO}}}}}_{2}\\\\ (\\Delta {{{\\rm{G}}}}^ {{\\circ} \\;\\,{\\prime}}=486.3\\,{{{\\rm{kJ}}}}/{{{\\rm{mol}}}}\\,{{{\\rm{hexadecane}}}})\\end{array}$$\\end{document} 4 C 16 H 34 + 74 H 2 O → 27 CH 3 COO − + 27 H + + 88 H 2 + 10 CO 2 ( Δ G ∘ ′ = 486.3 kJ / mol hexadecane ) 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}$${{{{\\rm{CH}}}}}_{3}{{{\\rm{COOH}}}}\\to {{{{\\rm{CH}}}}}_{4}+{{{{\\rm{CO}}}}}_{2}(\\Delta {{{\\rm{G}}}}^{{\\circ} \\;\\, {\\prime}}=-42.2\\,{{{\\rm{kJ}}}}/{{{\\rm{mol}}}}\\,{{{\\rm{acetate}}}})$$\\end{document} CH 3 COOH → CH 4 + CO 2 ( Δ G ∘ ′ = − 42.2 kJ / mol acetate ) 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}$$4 \\, {{{{\\rm{H}}}}}_{2}+{{{{\\rm{CO}}}}}_{2}\\to {{{{\\rm{CH}}}}}_{4}+2{{{{\\rm{H}}}}}_{2}{{{\\rm{O}}}}(\\Delta {{{\\rm{G}}}}^{{\\circ}\\;\\, {\\prime}}=-29.1\\,{{{\\rm{kJ}}}}/{{{\\rm{mol}}}}\\,{{{\\rm{hydrogen}}}})$$\\end{document} 4 H 2 + CO 2 → CH 4 + 2 H 2 O ( Δ G ∘ ′ = − 29.1 kJ / mol hydrogen ) At standard conditions, the breakdown of alkanes in to hydrogen, acetate, and carbon dioxide is highly endergonic, hence very low intermediate concentrations would need to be maintained to allow growth on these organisms 27 . The hexadecane cultures contained low concentration of acetate (8.8–15.2 μM) and hydrogen (8.8–25.6 pa) (Supplementary Data 6 ). Such low of intermediate concentrations were also described for the syntrophic associations of alkane-degrading bacteria and methanogenic archaea 6 , 28 . Most previously cultured ACR-containing archaea were considered to transfer electrons from alkane oxidation to sulfate reducers 7 , 8 , 10 , 14 . In contrast, our cultures do not contain sulfate, but produces methane as electron sink. The interaction between alkane degrading archaea and methanogens likely differs from that suggested for consortia of alkane degrading archaea and their sulfate-reducing partners 7 , 8 , 10 , 14 . Both Ca . Melinoarchaeum spp. and the methanogens ( Methanothermobacter and Methanothrix ) do not encode type IV pilin (PilA) and multi-heme c-type cytochromes (MHCs). These proteins were suggested to form the filamentous nanowires that mediate direct interspecies electron transfer 10 , 14 . The lack of enzymes to establish direct interspecies electron transfer agrees with the members of Ca . Melinoarchaeum spp. appear as single cells rather than form consortia with other cells (Supplementary Fig. 6 ). Potential habitats of the members in Ca . Melinoarchaeum Although culturable Ca . Melinoarchaeum and Ca . Cerberiarchaeon, both exhibiting activity in the degradation of long-chain alkanes, belong to the same family ( Ca . Cerberiarchaeaceae) within the Hadarchaeota phylum, Ca . Cerberiarchaeon was enriched from marine hydrothermal vent sediments 11 , while here Ca . Melinoarchaeum was sourced from terrestrial hot springs. Additionally, they exhibit distinct metabolic pathways; Ca . Cerberiarchaeon interacts with sulfate-reducing bacteria under sulfate conditions, whereas Ca . Melinoarchaeum cooperate with methane-producing archaea under methane-producing conditions. All 13 MAGs of genus Ca . Melinoarchaeum derived from continental hot spring sediments, including Tengchong hot spring in China, Yellowstone National Park and little hot spring in United States 18 , 19 (Supplementary Data 5 ). Furthermore, samples of Yellowstone National Park contain acr A genes and gene transcripts clustering with Ca . Melinoarchaeaum 29 . Thus, the members of genus Ca . Melinoarchaeum are widely distributed among hot spring environments, where they likely grow as alkane degraders. All MAGs in genus Ca . Melinoarchaeum have the genes for alkane degradation, which include the acr operon, a complete β-oxidation and WL pathway for further degradation of alkanes to carbon dioxide (Supplementary Fig. 8 ). Only the less complete MAG GEM56 (58.46%), lacks most of the WL pathway. All Ca . Melinoarchaeum MAGs have very similar electron cycling complexes, such as Mvh, FrhB and HydAD, but lack of respiratory pathways. This indicates incomplete degradation of alkanes into acetate and/or hydrogen would be a common trait for the cluster of Ca . Melinoarchaeum. Hot springs are dynamic geobiologically active environments and often rich in alkanes, including short-, mid-, and long-chain alkanes and aromatic hydrocarbons 15 , 30 – 33 . These alkanes are considered as abiogenic and formed in the deep subsurface, from where they migrate with rising fluids to reach the surface 15 , 30 – 33 . In addition to hot spring environments, the 16S rRNA genes of the genus Ca . Melinoarchaeum were also detected in mud volcanoes and subsurface aquifers (Supplementary Data 7 ). These environments also communicate with the deep Earth. Thus, we hypothesize that the members of genus Ca . Melinoarchaeum, are more abundant in the deep subsurface, and are occasionally transported and emplaced within the surface hot springs. Future studies will reveal if Ca . Melinoarchaeum or related Hadarchaeota have a larger role in reservoir biochemistry. Our study revealed that the archaeon Ca . Melinoarchaeum spp., belonging to the phylum Hadarchaeota , grows as an ACR-based alkane degrader together with methanogens, thereby establishing an archaea-archaea syntrophy for methanogenic hydrocarbon degradation (Fig. 6 ). This finding explains the frequent observation of MAGs of ACR-containing archaea without canonical methanogenesis pathways in methanogenic environments 15 – 19 . Our study expands our understanding of archaeal alkane degradation and significantly extends present knowledge of syntrophy in methanogenic alkane degradation. Fig. 6 Pattern diagram of methanogenic hydrocarbon degradation modes. Three metabolic modes have been identified for the execution of methanogenic hydrocarbon degradation: (I) a symbiotic relationship between syntrophic bacteria and methanogens, (II) a standalone archaeon, and (III) a archaea-archaea syntrophy."
} | 6,826 |
22347163 | PMC3272652 | pmc | 684 | {
"abstract": "Many sounds of ecological importance, such as communication calls, are characterized by time-varying spectra. However, most neuromorphic auditory models to date have focused on distinguishing mainly static patterns, under the assumption that dynamic patterns can be learned as sequences of static ones. In contrast, the emergence of dynamic feature sensitivity through exposure to formative stimuli has been recently modeled in a network of spiking neurons based on the thalamo-cortical architecture. The proposed network models the effect of lateral and recurrent connections between cortical layers, distance-dependent axonal transmission delays, and learning in the form of Spike Timing Dependent Plasticity (STDP), which effects stimulus-driven changes in the pattern of network connectivity. In this paper we demonstrate how these principles can be efficiently implemented in neuromorphic hardware. In doing so we address two principle problems in the design of neuromorphic systems: real-time event-based asynchronous communication in multi-chip systems, and the realization in hybrid analog/digital VLSI technology of neural computational principles that we propose underlie plasticity in neural processing of dynamic stimuli. The result is a hardware neural network that learns in real-time and shows preferential responses, after exposure, to stimuli exhibiting particular spectro-temporal patterns. The availability of hardware on which the model can be implemented, makes this a significant step toward the development of adaptive, neurobiologically plausible, spike-based, artificial sensory systems.",
"conclusion": "4 Conclusion The results presented in this work demonstrate, in hardware, how an implementation of a recurrently connected spiking network is able to learn and selectively respond to the dynamic spectro-temporal features of stimuli. The model relies on delays, which might arise from a number of processes including axonal propagation and spike interaction via intermediate neurons. In the case of the hardware substrate used to implement the model there was no provision for the implementation of delays in the design. We were able to overcome this problem by exploiting the variation and mismatch of the components in analog VLSI devices used in the setup. The stimulus-driven modifications of the network connectivity result from the interaction between the stimulus itself and the spike-based plasticity (STDP) rule adopted for the delayed feedback connections. After learning, the firing patterns of the neurons reflect the emergent connectivity, that is more of the neurons fire during a presentation, and the network is tuned to the stimulus properties. As a result, differences in the response of the network induced by learning can be used to distinguish similar stimuli that have parametrically different dynamic properties, for example differences in the direction or speed of an FM sweep. Our experiments show, in spiking hardware and in a sensory context, the emergence of feature sensitivity, and the limitations imposed by demonstrating these effects in hardware have necessitated the choice of relatively simple stimulus patterns. However comparable results have also shown that the principles generalize to slightly more complex stimuli (Figure 5 B) and, in software modeling work using a similar architecture, to dynamic ripples (Coath et al., 2011 ). We have also demonstrated how the network with a random profile of transmission delays can exhibit similar behavior. This is important because random delay profiles imply that there is no structural necessity for the input neurons to be strictly tonotopically arranged. Therefore, without loss of generality, the input stimulus need not necessarily look well arranged as in Figure 5 and could be shuffled to appear as random patterns. By the same argument the network could also learn correlations in stimuli that are not so clearly ordered. 4.1 Exploiting device mismatch for implementing temporal delays In order to implement propagation delays we made use of extra neurons whose time constants were proportional to the desired time delay. We were able to select delay neurons with different time constants by exploiting the mismatch effect in their analog circuit implementations. The effect of this mismatch is shown in Figure 3 , where we plot the range of delays exhibited by the neurons. Had there been no mismatch, this would not have been possible to achieve on a single chip with global/shared bias voltages. Variability can also be found on other model parameters (e.g., time constants, injection currents). This opens up the possibility to explore network dynamics of populations with a distribution of parameter values, rather than using globally common values. While mismatch induced noise is often minimized in conventional analog VLSI design, we try to use it to introduce sources of variability in the populations of silicon neurons that can model or reproduce the inhomogeneities present in real neural systems. 4.2 Long vs short time scales Sensory stimuli, and in particular auditory stimuli, contain both short and long range temporal correlations. The techniques described in this paper primarily address correlations only over relatively short time scales, i.e., those in a range from the order of synaptic or membrane time constants, up to those represented by the propagation of excitation to adjacent regions. The range of propagation delays implemented in the network defines the range of stimulus velocities that could be learnt. Temporal correlations over a longer time scale could be addressed using many levels of recurrence between widely separated layers, as is observed in the mammalian auditory system. Alternatively it could be tackled with working memory and neuromorphic implementations of state machine based approaches (Neftci et al., 2010 ). 4.3 Limitations The network model described in the paper, in its current form, assumes the stimulus to be dynamic and does not consider static stimuli where frequency channels are persistently active. The learning rule used dictates high probability of LTP in the presence of highly active inputs. As a result, if such persistent stimuli were to be presented to this network a large number of synapses would potentiate to reflect the temporal correlation across different Δ t values. However transient responses, in particular onset responses, are found at all levels of the auditory system and this could be modeled with strong synaptic depression at the input synapses of B 1 and B 2 or as post-synaptic adaptation of neurons in A, thus eliminating the problems caused by persistent input stimuli.",
"introduction": "1 Introduction Hardware implementations of neuromorphic systems (Indiveri and Horiuchi, 2011 ) have been successfully used in the past to implement and characterize biophysically realistic models of the early sensory processing stages both for the visual domain (Koch and Mathur, 1996 ; Liu, 1999 ; Barbaro et al., 2002 ; Kramer, 2002 ; Culurciello et al., 2003 ; Lichtsteiner et al., 2006 ; Zaghloul and Boahen, 2006 ; Leñero-Bardallo et al., 2010 ), and the auditory domain (Watts et al., 1992 ; Toumazou et al., 1994 ; van Schaik et al., 1996 ; Fragnière et al., 1997 ; van Schaik and Meddis, 1999 ; Wen and Boahen, 2006 ; Chan et al., 2007 ; Abdalla and Horiuchi, 2008 ). These systems were typically implemented as single Very Large Scale Integration (VLSI) devices (e.g., as silicon retinas, or silicon cochleas), comprising hybrid analog/digital circuits that faithfully reproduced in real-time the bio-physics of the neural sensory systems they modeled. The definition of an event (spike) based communication protocol based on the Address Event Representation (AER; Deiss et al., 1999 ; Boahen, 2000 ) has led to the development of a new generation of more complex multi-chip neuromorphic sensory systems (Liu et al., 2001 ; Choi et al., 2005 ; Chicca et al., 2007 ; Folowosele et al., 2008 ; Serrano-Gotarredona et al., 2009 ). However, while it is now possible to build hardware sensory systems that can react to auditory and visual stimuli, and eventually drive robotic actuators in real-time (see for example the exciting projects being developed at the Telluride 1 and Capo Caccia Neuromorphic Engineering workshops 2 , it still remains a challenge to build real-time sensory processing systems that can learn about the nature of their input stimuli and perform cognitive tasks, using spikes and spiking neural networks. There are two main bottlenecks that have been hindering progress in this area: one has to do with the practical difficulties of linking real-time asynchronous neuromorphic devices to each-other to build complex multi-chip systems; and the other is related to the theoretical and computational challenges in integrating and extracting information from the time-varying signals representing inputs, outputs, and internal state variables, in these types of systems. The first problem, that of developing real-time interfaces between the different AER chips to create complex multi-chip systems, is currently being addressed by developing both custom real-time hardware solutions (Gomez-Rodriguez et al., 2006 ; Fasnacht et al., 2008 ; Hofstaetter et al., 2010 ; Jin et al., 2010 ; Fasnacht and Indiveri, 2011 ; Scholze et al., 2011 ), as well as software solutions, and principled systematic methods for configuring network structures and system parameters (Davison et al., 2008 ; Neftci et al., 2011 , 2012 ; Sheik et al., 2011 ). The second problem is more fundamental. There is, in general, agreement that cells in primary sensory areas are typically characterized in terms of tuning to particular spectro-temporal features 3 . It is not, however, clear what features are encoded, or what neural mechanisms underlie those feature selectivities, or what developmental processes lead to the formation of those mechanisms. Although pursuing these questions has led to remarkable advances in understanding visual processing in the brain, a corresponding understanding of auditory processing is still lacking. Here we present a multi-chip neuromorphic system in which silicon neurons (Indiveri et al., 2011 ) dynamically adapt and learn, forming feature tuning properties that are derived from spectro-temporal correlations in their input spike trains. The auditory domain was chosen as the focus for the experiments, where the importance of dynamic spectro-temporal patterns in the communication calls of mammals and birds has motivated the study of cortical sensitivity to frequency sweeps (Godey et al., 2005 ; Atencio et al., 2007 ; Ye et al., 2010 ) and dynamic ripple noises (Kowalski et al., 1996 ; Calhoun and Schreiner, 1998 ; Depireux et al., 2001 ; Atencio and Schreiner, 2010 ) as candidates for constituent features which are sufficiently simple to be parametrized. The multi-chip neuromorphic system implements a neural network model of the auditory thalamo-cortical system similar to that described in Coath et al. ( 2011 ). We demonstrate that this system can learn, when repeatedly presented with a specific stimulus, to exhibit a preferential response to such a stimulus. We argue that the functional principles of this neural network, and of the thalamo-cortical model (Coath et al., 2011 ) it is derived from, can be used to produce spike-based feature extractors and that these could form the basis of artificial sensory systems using real-time analog neuromorphic VLSI.",
"discussion": "5 Discussion It is widely believed to be central to the performance of sensory systems in vivo that they respond preferentially to those aspects of the environment in which they operate that are salient. A simple biological example would be to differentiate a conspecific vocalization from the background of noise and other non-salient communication calls. Results suggest that this type of pattern specificity is, at least in some cases, present in primary cortical areas, e.g., Machens et al. ( 2004 ). This same property is also an important goal of artificial sensory systems. In both cases this clearly requires some strategy for encoding salient features of the stimulus within the neural, or silicon substrate. We propose that recurrent connectivity, observed throughout the auditory system, could mediate dynamic feature sensitivity by exploiting propagation delays. We have developed a computational model and verified that the emergent properties arise through the interaction of propagation delays with the spectro-temporal properties of stimuli and spike-time dependent plasticity. While the representation of input stimuli in this network is realized through the pairing of activity at different times and at different positions on the tonotopic axis, it is not limited to a single pair. Each neuron in B 2 , after learning, is connected to more than one B 1 neuron (see Figures 7 and 8 B) in most cases. This leads to the B 2 neurons accumulating evidence from the spikes arriving from more than one B 1 neuron. If only a small fraction of the correlations present in the stimulus match those that have been learnt by the network, and if the threshold of the B 2 neurons was appropriately tuned (the threshold being the minimum number of coincident feedback spikes from B 1 for B 2 neurons to fire), then these neurons would not fire. Consequently, we argue that given a large population and dense connectivity, a reasonable number of learned correlations from ES can be set as a threshold for the B 2 neurons to fire. The higher the number of such pairings present in a PS, the higher probability there is of the stimulus being identified as the current one (i.e., as ES). The feature sensitivity represented in the model is not a property revealed in the firing rate of an individual neuron, or indeed in any property of an individual neuron. The model has widespread recurrent connections and as a consequence, the ensemble response of the network is modified during exposure. Thus the representation of, and sensitivity to, patterns of activity evolving over time is a network property of this model. This is especially interesting because clear examples of such dynamic response properties are routinely absent for single neuron measurements in vivo . The neuromorphic system we describe here could be used with real-world stimuli and validated against noise in the stimulus, and limitations in the system’s state variables (e.g., due to device mismatch, limited resolution, and signal-to-noise ratio limitations, bounded synaptic weights, power constraints, etc.). In further work, we plan to go beyond the demonstration of emergent sensitivity to a stimulus parameter, by quantifying the increase in acuity in information-theoretic terms in the presence of stimulus variation. This will provide a basis for the quantitative comparison of networks, connectivity patterns, and learning strategies."
} | 3,743 |
35495028 | PMC9047717 | pmc | 685 | {
"abstract": "The spiking neural network (SNN) is a possible pathway for low-power and energy-efficient processing and computing exploiting spiking-driven and sparsity features of biological systems. This article proposes a sparsity-driven SNN learning algorithm, namely backpropagation with sparsity regularization (BPSR), aiming to achieve improved spiking and synaptic sparsity. Backpropagation incorporating spiking regularization is utilized to minimize the spiking firing rate with guaranteed accuracy. Backpropagation realizes the temporal information capture and extends to the spiking recurrent layer to support brain-like structure learning. The rewiring mechanism with synaptic regularization is suggested to further mitigate the redundancy of the network structure. Rewiring based on weight and gradient regulates the pruning and growth of synapses. Experimental results demonstrate that the network learned by BPSR has synaptic sparsity and is highly similar to the biological system. It not only balances the accuracy and firing rate, but also facilitates SNN learning by suppressing the information redundancy. We evaluate the proposed BPSR on the visual dataset MNIST, N-MNIST, and CIFAR10, and further test it on the sensor dataset MIT-BIH and gas sensor. Results bespeak that our algorithm achieves comparable or superior accuracy compared to related works, with sparse spikes and synapses.",
"introduction": "1. Introduction Artificial intelligence (AI) has shown impressive abilities in various tasks such as computer vision, natural language processing, and decision making. For example, AlphaGo Zero defeated the world champion of the game of Go (Silver et al., 2017 ). However, the power consumption of AlphaGo Zero is about 1kW (Frenkel et al., 2021 ), which is 50× higher than the 20W power budget of the human brain (Roy et al., 2019 ). The brain-inspired spiking neural network (SNN) plays an important role in addressing the issue of AI energy efficiency. SNN exchanges information through binary spikes between synapses and performs intensive calculation only when spikes are received. Dedicated SNN hardware such as TrueNorth (Akopyan et al., 2015 ), Loihi (Davies et al., 2018 ), Tianjic (Pei et al., 2019 ), and MindWare (Ding et al., 2021 ) can reduce energy consumption from sparse spikes and synapses through spike-driven computing architecture. Despite the merits of improving energy efficiency, there remain a lot of challenges ahead of the SNN in sparsity learning algorithms and efficient network exploration. The commonly adopted SNN learning algorithms can be summarized into three different types as follows. (1) Conversion-based learning . It uses the same SNN structure as an artificial neural network (ANN) and converts the parameters of the learned ANN to SNN. One conversion idea is to use the spiking firing rate (FR) of SNN to quantify the floating value of ANN and establish an approximate mapping between the parameters of two networks (Sengupta et al., 2019 ; Kim et al., 2020 ). This kind of conversion uses rate coding, resulting in dense spikes. Another idea is to use spike timing to represent the floating value in ANN. Methods like time-to-first-spike (TTFS) conversion (Rueckauer and Liu, 2018 ) and few spikes conversion (FS-conversion) (Stöckl and Maass, 2021 ) use temporal coding to protect spiking sparsity. However, the time domain is used for coding so that temporal processing structure such as recurrent neural network (RNN) cannot be converted. (2) Plasticity-based learning . It is a kind of biologically inspired algorithm. The most famous spike-timing-dependent plasticity (STDP) adjusts synaptic weight according to the spike order between the pre- and post-synaptic neurons. The role of STDP is feature clustering. Combined with lateral inhibition structure, STDP can realize unsupervised classification (Diehl and Cook, 2015 ; Białas and Mańdziuk, 2021 ). Reward-modulated STDP draws on the eligibility trace of reinforcement learning to realize supervised learning to further improve performance (Mozafari et al., 2018 ). The plasticity-based learning algorithm is skilled in computation overhead and weak in network accuracy. (3) Gradient-based learning . Like the learning of ANN, it updates the parameters of SNN according to the gradient information from backpropagation. A recent study by Lillicrap et al. ( 2020 ) suggests that a similar propagation mechanism may exist in the brain. Spatio-temporal backpropagation (STBP) (Wu et al., 2018 , 2019 ) provides advanced accuracy by calculating gradient in the spatio-temporal domain. Deep continuous local learning (DECOLLE) (Kaiser et al., 2020 ) reduces the memory overhead through the local error function. Spike-train level recurrent SNN backpropagation (ST-RSBP) (Zhang and Li, 2019 ) further supports the recurrent layer, to deal with temporal information by mimicking import feedback structure in the brain (Luo, 2021 ). The above algorithms focus on the accuracy improvement and lack consideration in the sparsity issue. Compared with local learning based on plasticity, gradient-based learning requires global information. It improves accuracy and brings additional calculation burdens. However, in the offline learning scenario, the computational overhead of SNN is mainly contributed by inference rather than learning. Therefore, reducing the computational overhead in inference through sparsity optimization and ensuring accuracy by gradient-based learning, become the major motivation of this work. Another kind of SNN algorithm aims to improve synaptic sparsity by pruning. Existing studies explore different pruning standards. Liang et al. ( 2021 ) prune synapses through random patterns and quantify synaptic weight to reduce storage overhead. Rathi et al. ( 2018 ) utilize the synaptic weight threshold to prune and optimize storage through weight quantization and sharing. Cho et al. ( 2019 ) prune long-range synaptic connections based on the small world theory of the nervous system. Nguyen et al. ( 2021 ) combine pruning with STDP and use the weight adjustment record as the pruning standard. Shi et al. ( 2019 ) use spiking count as the pruning threshold and propose a soft pruning method to reduce the computation overhead in learning. Moreover, Guo et al. ( 2020 ) prune the neurons rather than synapses according to spiking count, providing a new perspective of sparsity exploration. SNN can perform sparse computing due to the event-driven feature. At the same time, the synaptic operation uses membrane potential accumulation instead of matrix multiplication and addition in traditional ANN, which reduces the amount of calculation. In recent years, similar methods have been proposed in the field of ANN to reduce the number of operations. Binarized neural network (BNN) (Hubara et al., 2016 ) and XNOR-Net (Rastegari et al., 2016 ) introduce binarized weights and activations and replace most arithmetic operations on synapses with bit-wise operations. AdderNet (Chen et al., 2020 ) builds ANN only through addition to avoid the expensive multiplication operation and achieves acceleration with low energy consumption. Beyond that, Bartol et al. ( 2015 ) believe each synapse stores about 4.7 bits of information. Quantization of synaptic weights can also be an idea to further optimize computational speed and compress storage overhead. This work proposes an SNN learning algorithm, namely backpropagation with sparsity regularization (BPSR) to facilitate sparsity. As shown in Figure 1 , the sparse spikes reduce the amount of information that subsequent neurons need to process, meanwhile the sparse synapses prevent each spike from causing intensive calculations. The proposed BPSR enables SNN to improve sparsity during learning and achieve satisfactory energy efficiency in inference. The backpropagation takes advantage of temporal information and adapts the brain-like recurrent structure. BPSR balances the accuracy and FR by combining backpropagation with spiking regularization. Inspired by the fact that the brain learns through synaptic rearrangement (Dempsey et al., 2022 ), rewiring mechanism is proposed to explore efficient SNN structures, which uses the weight and gradient to regulate synaptic pruning and growth. The experimental result is consistent with the concept that the proposed BPSR can achieve low FR with high accuracy. Spiking sparsity is proved to be beneficial to SNN learning (Tang et al., 2017 ), because of the suppression of information redundancy. BPSR not only improves the synaptic sparsity but also generates a bionic structure similar to the nervous system of Caenorhabditis elegans ( C. elegans ). The result on the visual MNIST dataset (LeCun et al., 1998 ) with rank order coding (Thorpe and Gautrais, 1998 ), neuromorphic-MNIST (N-MNIST) (Orchard et al., 2015 ), and CIFAR10 (Krizhevsky et al., 2009 ) reach the accuracy of 98.33, 99.21, and 90.74%, respectively. The evaluation on MNIST also shows 30× the inference overhead advantage compared to other SNN works. With post-training quantization (PTQ), SNN can achieves 15× efficiency compared to BNN with 0.22% accuracy drop. BPSR is further tested on sensor datasets like MIT-BIH arrhythmia (Moody and Mark, 2001 ) and gas senor (Vergara et al., 2013 ), which achieves 98.41 and 98.30% accuracy. Figure 1 Spiking sparsity and synaptic sparsity facilitate the efficiency of SNN by reducing the number of synaptic operations. The remainder of this article is organized as follows. In Section 2, the backpropagation with sparsity regularization is introduced. The suggested heterogeneous neuron dynamic model, the loss function with regularization, and the backpropagation algorithm on the flat and recurrent SNN layers are detailed. In Section 3, the rewiring based on weight and gradient and the corresponding implementation process is introduced. In Section 4, the effect of the proposed BPSR algorithm is tested by experiments, and comparisons with related works on various datasets are reported. In Section 5, we summarize this work and make a discussion.",
"discussion": "5. Discussion SNN promises to realize efficient AI through its brain-inspired mechanism and spike-driven computing architecture. However, the efficiency advantage of the SNN cannot be fully exploited because of the lack of sparsity exploration. This work provides a learning algorithm, namely Backpropagation with Sparsity Regularization (BPSR), to improve efficiency through advanced spiking sparsity and synaptic sparsity. Firstly, a backpropagation algorithm with sparsity regularization is proposed to update parameters and improve sparsity. A heterogeneous LIF neuron dynamics model and a classification loss function with spiking and synaptic regularization are defined. The backpropagation algorithm of the flat and recurrent layer is detailed to calculate the gradient of each parameter. Secondly, the rewiring mechanism based on weight and gradient is proposed to improve synaptic sparsity through pruning and growth. Then, the experimental results show that the proposed BPSR has the advantages of runtime and graphic memory overhead compared with other gradient-based learning algorithms. The improved spiking sparsity can balance the accuracy and FR, and promotes the network performance by simplifying the information representation. Through the BPSR, SNN acquires a structure similar to the nervous system of C. elegans , proving its effectiveness. The proposed BPSR reaches the accuracy of 98.33% on the MNIST dataset while achieving 30× inference overhead than other SNN work and 15× energy efficiency compared to BNN after PTQ (with 0.22% accuracy drop). Finally, BPSR is also evaluated on two visual datasets (N-MNIST and CIFAR10) and two sensor datasets (MIT-BIH and gas sensor). The experimental results show comparable or superior accuracy (99.21, 90.74, 98.41, and 98.30%, respectively), with spiking sparsity and synaptic sparsity."
} | 2,991 |
35009994 | PMC8746732 | pmc | 686 | {
"abstract": "Inspired by the superhydrophobic properties of some plants and animals with special structures, such as self-cleaning, water repellent, and drag reduction, the research on the basic theory and practical applications of superhydrophobic surfaces is increasing. In this paper, the characteristics of superhydrophobic surfaces and the preparation methods of superhydrophobic surfaces are briefly reviewed. The mechanisms of drag reduction on superhydrophobic surfaces and the effects of parameters such as flow rate, fluid viscosity, wettability, and surface morphology on drag reduction are discussed, as well as the applications of superhydrophobic surfaces in boiling heat transfer and condensation heat transfer. Finally, the limitations of adapting superhydrophobic surfaces to industrial applications are discussed. The possibility of applying superhydrophobic surfaces to highly viscous fluids for heat transfer to reduce flow resistance and improve heat transfer efficiency is introduced as a topic for further research in the future.",
"conclusion": "6. Conclusions In this study, a comprehensive review of the research progress of superhydrophobic surfaces and their applications in industry and life in the last decade is presented. Super-hydrophobicity, surface wetting models, and methods for the preparation of superhydrophobicity on metal and alloy substrates are described. The excellent performance of superhydrophobic surfaces for drag reduction, boiling heat transfer, and condensation heat transfer is analyzed in detail in different parts of the paper. Superhydrophobic surfaces have been extensively applied in industry, for example, to delay the formation of ice and frost on airplanes, wind turbines, and heat exchanger surfaces, enhance the corrosion resistance of metal and alloy surfaces, improve the heat transfer properties of boiling and condensation, and so on. These applications have positive implications for energy savings and improved surface performance. Although the application of superhydrophobic surfaces has been deeply researched, more advanced investigations are necessary in fundamental theory; for example, the fabrication of superhydrophobic surfaces still suffers from high cost, poor durability, and technical complexity. When superhydrophobic surfaces are exposed to acidic and alkaline environments, high temperatures, mechanical wear, and cyclic impacts, they tend to lose the superhydrophobic properties, so research on both durability and robustness is indispensable. The boiling and condensation heat transfer characteristics and mechanisms of superhydrophobic surfaces are different from those of normal and hydrophilic surfaces, and further theoretical and experimental studies in this aspect are necessary. In the majority of the papers investigated, the working medium for boiling and condensing heat transfer is pure water or aqueous solutions with very low viscosity. However, in industrial applications, it is often encountered that the working fluid has a high viscosity and tends to adhere to the surface, resulting in degradation of the heat transfer coefficient. Moreover, the research on heat transfer of viscous fluids on superhydrophobic surfaces is rather little, so the possibility of superhydrophobic surfaces to reduce the flow resistance of high-viscosity fluids and improve the heat transfer efficiency is a worthwhile work in the future.",
"introduction": "1. Introduction The wettability of surfaces can be characterized by the contact angle (CA). Generally, the hydrophilic surface has a contact angle below 90°, and the contact angles below 10° are defined as superhydrophilic surfaces. The surfaces with contact angles above 90° are called hydrophobic surfaces, and the contact angle above 150° and the sliding angle below 5° are defined as superhydrophobic surfaces [ 1 , 2 ], as shown in Figure 1 . The inspiration of the superhydrophobic surface comes from natural plants and insects such as lotus leaves, rice leaves, mosquito feet, butterfly wings, and water striders all show superhydrophobic properties like self-cleaning, water-repelling, easily rolling off the surface, as shown in Figure 2 a, the water can completely roll off the lotus leaves without footprint remains. In Figure 2 b, the water striders that are described as “skaters in a pond” can walk on water easily and fastly [ 3 , 4 , 5 ]. These particular characteristics are ascribed to the special binary micro/nanostructures and low surface energy on the plants and insects’ surfaces [ 5 , 6 ]. For instance, they are randomly distributed with micron and nanoscale level papillary on the lotus leaf (shown in Figure 3 ), and a layer of low surface energy cuticle wax is covered on the surface [ 7 , 8 ]. If the superhydrophobic features can be functionalized on various metal surfaces, it will be significant and beneficial in many industrial applications for saving energy and energy storage [ 9 ]. For example, it can drag reduction, anti-fouling, and enhance heat transfer performance. The contact angle is usually measured by a contact angle meter when a 2 µL or 5 µL water droplet rests on a surface. The CA is one of the most important parameters in characterizing the wettability of a surface. In 1805, British scientist Thomas Young first proposed the correlation between the CA and the surface tension on an ideally smooth surface. Since Young’s theory is strictly valid for ideally smooth surfaces, the Wenzel [ 10 ] model and Cassie–Baxter [ 11 ] model were established to introduce the surface wetting state of partial rough surfaces. Based on the Wenzel model, the water droplet completely penetrates the micro/nanostructures of the rough surface, as shown in Figure 4 a, in which the increase of the surface roughness of a hydrophobic surface can enhance the static CA and make the surface more hydrophobic. Therefore, the superhydrophobic surface can be prepared by creating a rough structure on the hydrophobic surface. Wenzel’s conclusion was further studied by Cassie and Baxter; they considered the situation that the water droplet’s adhesion force of a particularly rough surface was not enough to make the surface completely wetting, so in their theory, the water droplet is suspended on the micro/nanostructures, and a discontinuous air cavity is formed between the droplet and the rough surface, as shown in Figure 4 b. The discontinuous air cavities remarkably reduce the contact area of a water droplet and solid surface, which is resulted in a larger corresponding CA of Cassie–Baxter model than the Wenzel model, and the water droplet can easily roll off the surface. Cassie and Baxter studied the effect of the porous composite surface with different chemical components on the droplet contact angle. Cassie–Baxter equation is as follow: (1) cos θ C = f 1 cos θ 1 + f 2 cos θ 2 \nwhere f 1 and f 2 are the area fraction of the liquid contacting components 1 and 2, and the corresponding intrinsic contact angles are θ 1 and θ 2 , and f 1 + f 2 = 1. Swain et al. [ 12 ] investigated the wetting law on geometrically rough and chemically heterogeneous surfaces. A new complex equation is proposed for the structure of substrates with both geometric and chemical structures: (2) cos θ e = ∑ i r i ( cos θ i − λ i C i σ ) + Δ ρ g Z 2 ¯ \nwhere θ e is the average contact angle taken up by the drop on a heterogeneous substrate, r i is the ratio of the non-planar area covered by the material to the total planar area, σ is the interfacial tensions, λ is the line tension, and Z 2 ¯ is the mean square height of the substrate. Both the micro-nano roughness structures and modification with low surface energy on the surface are the important factors influencing the superhydrophobicity of a surface, while superhydrophobic surface cannot be obtained only modified by low surface energy materials [ 13 ]. Therefore, both the surface roughness and the low surface energy are indispensable in order to obtain the superhydrophobic surface [ 14 ]. Based on this theory, scholars have developed a series of methods to fabricate superhydrophobic surfaces on various metal substrates, such as laser etching method [ 15 ], chemical etching [ 16 , 17 ], sol-gel method [ 18 ], chemical deposition method [ 19 ], vapor deposition method [ 20 ], template method [ 21 ], anodic oxidation [ 22 ], electrospinning method [ 23 ], and layer-by-layer assembly method [ 24 ]. With the development of manufacturing superhydrophobic surfaces on various metal substrates, these functionalized metal superhydrophobic surfaces with interesting characteristics are widely applied in multifarious industries and everyday life. For instance, the surface of metal pipelines is usually rough surface and hydrophilic; the fluids inside the pipes undergo enormous pressure drops, which causes tremendous energy consumption. Therefore, the superhydrophobic surface with drag reduction property is able to overcome this problem effectively [ 25 ]. Surface wettability plays a key role in boiling heat transfer, and the heat transfer performance of superhydrophobic surfaces increases significantly at small superheats due to the rapid removal of nucleated bubbles from the surface, improving the heat transfer coefficient. When a superhydrophobic surface is applied to a condensing heat exchanger tube, not only the pressure drop can be reduced, but the convective heat transfer performance is also improved to some extent [ 26 ]. The wind turbines and aircraft are usually working in hostile environments because of ice and snow accumulation and adhere to the surfaces, which causes catastrophic risks; thus, the superhydrophobic surfaces with anti-icing properties can avoid ice and snow accumulation and adhesion and enhance the efficiency [ 27 ]. Superhydrophobic meshes can be practically used in petroleum industries to separate oil and water [ 28 ]. Moreover, superhydrophobic surfaces have many other applications such as waterproof [ 29 , 30 ], self-cleaning [ 31 , 32 , 33 , 34 ], anti-frosting [ 35 , 36 , 37 , 38 ], and anti-corrosion [ 39 , 40 , 41 , 42 ]. The above content demonstrates that superhydrophobic surfaces have many useful potential applications in industries and everyday life, especially in the energy-saving field. This paper illustrates the fabrications and characteristics of superhydrophobic surfaces on the metal surface. The characteristics of drag reduction, boiling heat transfer, and condensation heat transfer on the superhydrophobic surface are illustrated. Finally, we discuss the feasibility of using the superhydrophobic surface to reduce the flow resistance and improve the heat transfer efficiency in the heat transfer process of the viscous fluid. Figure 1 Droplet state on the surface. Figure 2 Superhydrophobic surface. ( a ) Lotus leaf. Reprinted with permission from Ref [ 43 ]. Copyright 2015 Elsevier. ( b ) Water strider. Reprinted with permission from Ref [ 3 ]. Copyright 2007 American Chemical Society. Figure 3 Superhydrophobicity and the multiscale structural surface of Lotus leaf. Reprinted with permission from Ref [ 44 ]. Copyright 2011 Elsevier. Figure 4 The states of droplets on roughness surface. ( a ) The Wenzel model. ( b ) The Cassie–Baxter model. Reprinted from Ref [ 45 ]."
} | 2,816 |
22854073 | null | s2 | 687 | {
"abstract": "Microbes produce many molecules that are important for their growth and development, and the exploitation of these secretions by nonproducers has recently become an important paradigm in microbial social evolution. Although the production of these public-goods molecules has been studied intensely, little is known of how the benefits accrued and the costs incurred depend on the quantity of public-goods molecules produced. We focus here on the relationship between the shape of the benefit curve and cellular density, using a model assuming three types of benefit functions: diminishing, accelerating, and sigmoidal (accelerating and then diminishing). We classify the latter two as being synergistic and argue that sigmoidal curves are common in microbial systems. Synergistic benefit curves interact with group sizes to give very different expected evolutionary dynamics. In particular, we show that whether and to what extent microbes evolve to produce public goods depends strongly on group size. We show that synergy can create an \"evolutionary trap\" that can stymie the establishment and maintenance of cooperation. By allowing density-dependent regulation of production (quorum sensing), we show how this trap may be avoided. We discuss the implications of our results on experimental design."
} | 325 |
28948227 | PMC5609849 | pmc | 688 | {
"abstract": "Here, we show how to predictably impart extremely low ice adhesion to a range of different plastic materials.",
"conclusion": "CONCLUSIONS Here, we have built a framework that predicts the reduction in the ice adhesion strength of both thermoplastics and thermosets, based on the addition of an oil/plasticizer. The addition of a plasticizer lowers the ice adhesion in two ways: by lowering the effective cross-link density or modulus of the polymer and by making only a fraction of the polymer’s surface available to adhere to ice. An excellent agreement was observed between our framework, Eq. 9 , and a wide range of polymer/plasticizer combinations. Moreover, the use of Eq. 11 allows lubricated systems to be fabricated with full knowledge of their expected ice adhesion strength once the lubricant becomes depleted. Overall, our proposed framework will allow for a wide range of icephobic materials to be fabricated from essentially any chemistry and base components. Many of the formulations exemplified in this work may find immediate usage for solving the ice accretion problem in a host of different sectors worldwide.",
"introduction": "INTRODUCTION The removal of accreted ice remains a costly, hazardous concern, both residentially and industrially across the globe ( 1 , 2 ). Coatings that significantly reduce the adhesion between ice and a given surface can be called icephobic ( 3 ). However, very few coatings have been fabricated that facilitate the complete removal of ice without the input of additional energy, that is, surfaces where ice may be removed by its own weight ( 4 – 11 ). Typically, these extremely icephobic surfaces have relied on sacrificial lubricants, where the icephobic properties may diminish as the lubricant becomes depleted. There are currently no methods for predicting the resultant ice adhesion once the lubricant has been removed. This inability to predict the ice adhesion is not limited to initially lubricated surfaces. Although it would be highly desirable, there is not a single reported method for modifying the icephobicity of a given surface in a predictable manner. Materials like glass, steel, and aluminum display ice adhesion strengths of τ ice > 1000 kPa ( 7 , 12 ). The ice adhesion strength of polymeric systems typically ranges 150 kPa ≤ τ ice ≤ 500 kPa ( 12 , 13 ). However, for ice to be removed by wind shear or solely under its own weight, significantly lower τ ice values are necessary. For example, ice accumulation notoriously lowers the efficiency of wind turbines during the winter months ( 14 ). A typical offshore wind turbine blade is 85 m long and rotates at a minimum speed of 10 rpm ( 15 ). At the blade tip, the centripetal force experienced by a square meter slab of ice that is 1 cm thick is around 12 kN. For this ice to detach, τ ice ≤ 12 kPa is necessary. This order-of-magnitude disparity highlights the inherent difficulty in fabricating practically useful icephobic surfaces. Recently, we showed that low-modulus (or low cross-link density, ρ CL ) elastomers are intrinsically icephobic ( 7 ). When ice adhered to a low-ρ CL elastomer experiences a shear stress, the ice-elastomer interface can cavitate, causing the ice to detach at low applied loads ( 16 ). Ice adhesion strengths as low as τ ice ≈ 10 kPa were reported for dry, low-modulus elastomers. However, these extremely soft surfaces are not mechanically robust. Therefore, low ρ CL alone offers limited improvement, in terms of durability toward repeated icing/deicing, over lubricated systems that initially display equally low ice adhesion strengths ( 4 , 9 , 10 ). Fortunately, engendering slip at the ice interface can further reduce the adhesion of ice to elastomers. The no-slip condition is a ubiquitous boundary condition found in the areas of fluid mechanics ( 17 ), polymer rheology ( 18 ), and contact mechanics ( 19 ). Here, the no-slip condition refers to a solid coating remaining rigidly adhered to ice during the application of an external load (but before fracture). In contrast, if the interface is sufficiently mobile, slip between the ice and the coating can occur in a process referred to as interfacial slippage ( 20 ). In our previous work, we showed that elastomers exhibiting interfacial slippage displayed a significantly reduced adhesion to ice ( 7 ). Interfacial slippage was achieved for some of the fabricated coatings by filling the different elastomers with various oils. Although we observed interfacial slippage to be most effective for low-ρ CL elastomers, the extent to which an oil caused interfacial slippage remained largely unexplored. Here, we develop a framework that predicts the reduction in the ice adhesion of surfaces using interfacial slippage, as well as τ ice for lubricated surfaces that have lost their lubricating oil layer. This framework allows one to quantitatively predict the reduction in τ ice and, therefore, enables the design of extremely icephobic surfaces fabricated from essentially any polymeric system, including non–cross-linked thermoplastics, knowing only the properties of the initial polymer and oil. Although some work has previously been done on rendering cross-linked networks icephobic ( 6 , 7 , 9 – 11 , 13 , 21 , 22 ), there are no reported methods for fabricating icephobic systems from common engineering plastics. Framework The purpose of this work is to understand how the ice adhesion of polymers changes when they are plasticized. We will refer to the miscible liquid component within the polymer interchangeably as an oil or a plasticizer, although we recognize that only a subset of polymers may be plasticized with oils and that not all plasticizers are oily ( 23 ). To predict the reduction in ice adhesion of any plasticized polymer, we will show that only two properties need be known: τ ice no − oil , the ice adhesion strength of the unfilled (dry) polymer, and φ oil max , the maximum plasticizer content of the polymer before complete phase separation occurs (that is, the miscibility limit of the plasticizer in the polymer). On the basis of Chernyak and Leonov’s framework for the friction of rubber ( 19 ), we previously showed that when interfacial slippage was enabled τ ice oil τ ice no − oil ∝ ρ oil CL ρ no − oil CL (1) where we use “oil” and “no-oil” to denote either the ice adhesion strength, or the cross-link density, of the swollen (with oil) and dry (without oil) polymer, respectively. The reduction in ice adhesion of an oil-filled network was shown to be directly proportional to the reduction in the cross-link density of the network. It is well known that filling a polymer with a liquid lowers its effective cross-link density ( 24 ). To predict the reduction in ice adhesion caused by filling a polymer with oil, a relation between ρ CL and bulk oil content, φ oil , is needed. Below, we derive such a relation. If interfacial slippage occurs, then the reduction in ice adhesion was shown to strongly depend on the ρ CL . For example, a 24-fold reduction in τ ice was observed for a soft polyurethane rubber, and essentially, no change in τ ice was observed for rigid plastics like polystyrene (PS) ( 7 ). However, modification of the base material was necessary to achieve identical modulus values for comparing surfaces that did, and did not, exhibit interfacial slippage. Here, we discuss the more general case, where both the cross-link density and interfacial slippage are allowed to vary for each polymer/oil combination. Essentially, we answer the question: For surfaces exhibiting interfacial slippage, how does τ ice depend on φ oil ? In our proposed framework for interfacial slippage, we assume that the ice adhesion of pure oil is negligible. Because liquids may be defined by their inability to support shear, this is a reasonable assumption ( 25 ). Highly lubricated systems have approached this idealization, initially displaying τ ice = 1.7 kPa ( 10 ), τ ice = 1.4 kPa ( 9 ), and τ ice = 0.15 kPa ( 7 ). At the ice-polymer interface, we let φ s be the fraction of solid polymer in contact with ice. The remaining 1 − φ s of the surface is composed of oil, and a lubricated surface may be defined such that φ s = 0. We propose that the ice adhesion strength of an oil-filled polymer is given by τ ice oil = τ ice no − oil ρ oil CL ρ no − oil CL φ s (2) Equation 2 combines the reduction in cross-link density through the addition of oil, or relation (1), with the reduction in adhesion sites at the surface of the filled polymer. The linear scaling with φ s has the advantage of bounding the ice adhesion of the oil-filled polymer between τ ice = τ ice no − oil , when φ oil = 0, and τ ice = 0 when the surface is fully lubricated. To measure φ s , we use the theory of hemi-wicking first proposed by Bico et al. for rough, partially wetted surfaces ( 26 ).",
"discussion": "DISCUSSION Designing icephobic surfaces Designing icephobic polymers requires the knowledge of only the inherent ice adhesion strength of the dry polymer ( τ ice no − oil ) and the maximum oil content that the polymer can uptake ( φ oil max ). In Fig. 5A , we construct a phase diagram for icephobic polymers. An upper bound to the ice adhesion strength reduction occurs when no plasticizer is present on the surface of the polymer. In this case, Eq. 10 describes the minimal possible reduction in τ ice for the plasticized polymer, arising solely from the reduction in the modulus. The red region of our phase diagram ( Fig. 5A ) encompasses ice adhesion strengths higher than our framework allows. Because the oil can only exist within the bulk of the polymer, or on its surface, ice adhesion strengths within this regime should never be observed, unless the adhesion of ice to the oil is nonzero. In our previous work ( 7 ), we fabricated one such exceptional system: PS filled with liquid, low–molecular weight ( M w = 200 g/mol) PS. Upon evaluation of a pure film of this liquid (φ oil = 1.0), we observed τ ice ≫ 0. In all of our other systems, we have never observed values of φ oil in the red region of our icephobicity phase diagram. Fig. 5 Designing icephobic surfaces. ( A ) Phase diagram for icephobic polymers. The regime of possible durability, the green region containing surfaces using interfacial slippage, is bounded by Eqs. 10 and 11 . ( B ) When VF40 was lubricated with four oils of differing solubility, the initial τ ice values (open symbols) fell within the lubrication regime. Equation 11 precisely predicted the reduction in the values for the ice adhesion strength upon wiping away the free oil layer (closed symbols). ( C ) For 11 different polymer/plasticizer combinations, our measured reductions in τ ice for surfaces that exhibited interfacial slippage were always bounded by Eqs. 10 and 11 . Note that, for the FPU/HL SFO system, the solid surface energy was lower than the surface tension of the plasticizer. ( D ) Lubricated elastomers from previous studies can initially achieve ultralow reductions in τ ice , and the reported τ ice values correctly lay in the lubrication region predicted by Eq. 11 . The data from Zhu et al. ( 11 ), recast using the literature τ ice value for PDMS, is denoted by an asterisk. For the data of Wang et al. ( 10 ), only one φ oil value was reported, so all the surfaces have been placed at this value. If φ oil > φ oil max , then the polymer’s oil capacity has been exceeded, and the surface is necessarily lubricated. Any lubricated system should exhibit an ice adhesion strength less than τ ice oil τ ice no − oil = ( 1 − φ oil max ) 5 / 3 ( 1 − α ) (11) This lower bound separates the interfacial slippage regime (green region) from the lubrication regime (blue region) in our phase diagram ( Fig. 5A ). However, recall that the ice adhesion strength of lubricated surfaces increases over time ( 7 , 10 ), making Eq. 11 a predictor for the ice adhesion strength that lubricated polymers will exhibit once their lubricating layer has been removed. Lubricated polymers may therefore prove advantageous over lubricant-infused surfaces ( 8 , 32 ). Lubricant-infused surfaces are based on textured solids for which φ oil max ≈ 0, and therefore, τ ice no − oil is very high ( 33 , 34 ). To illustrate this point, we fabricated lubricated rubbers by overfilling ( φ oil ≫ φ oil max ) the VF40 rubber with four oils of varying miscibility [ φ oil max = 0.07, 0.17, 0.29, or 0.51, for squalane oil, high-oleic SFO (HO SFO), HL SFO, and MCT oil, respectively]. All the initial τ ice values fell within the predicted lubrication regime (open symbols, Fig. 5B ). Upon wiping the free oil layer from the surface of the lubricated rubber, simulating a complete loss of lubricant, we observed an exceptional agreement with Eq. 11 and the subsequently measured τ ice values of these four elastomer/oil combinations (closed symbols, Fig. 5B ). The ice adhesion strength of polymeric systems will always be bounded by Eq. 10 (plasticizer completely in the bulk) and Eq. 11 (maximum amount of miscible plasticizer on the surface). A good agreement was found between these two limiting equations and the 11 different systems presented in this study ( Fig. 5C and Materials and Methods). The ice adhesion strengths for 10 different icephobic formulations—ranging in base material, φ oil , and fabrication technique (Materials and Methods)—are presented in Table 1 . As an example highlighting the usefulness of Fig. 5A , consider some extremely durable polymer with τ ice = 300 kPa. This material is not icephobic. For some applications, τ ice ≤ 30 kPa is necessary ( 6 , 7 , 33 ). What oil, and how much oil, should be incorporated for the icephobic coating to achieve this desired ice adhesion strength? According to Fig. 5A , a reduction in the ice adhesion strength ratio to 0.1 necessitates that the coating is at least 47% oil. Any less than 47% oil within the coating will not result in the required ice adhesion strength unless the surface is fully lubricated. No more than 75% oil is necessary, because Eq. 10 dictates that an even lower ice adhesion strength will be achieved for φ oil ≥ 0.75. Thus, an oil must be selected where φ oil max > 0.47, and the amount of oil added should be 0.47 ≤ φ oil ≤ 0.75. Should this material instead be lubricated, for example, with an oil for which φ oil max = 0.2, the ice adhesion strength will increase to τ ice ≈ 60 kPa once the oil is depleted, and the coating will become ineffective. PVC, a durable engineering plastic, displayed a nonplasticized ice adhesion strength of τ ice no − oil = 312 ± 70 kPa. When plasticized with 60 wt % MCT oil, for which φ oil max ≈ 0.9, the ice adhesion strength reduced to τ ice = 24 ± 4 kPa and did not increase over 10 icing/deicing cycles. In contrast, CF50, a polyurethane ( τ ice no − oil = 319 ± 44 kPa), lubricated with MCT oil (for which φ oil max ≈ 0.07) initially displayed τ ice ≈ 14 kPa. When the MCT oil was wiped away, the observed ice adhesion strength increased to τ ice ≈ 150 kPa. The above examples illustrate why engineering icephobic surfaces that display τ ice ≤ 30 kPa without lubrication has been challenging in the past. Almost all materials exhibit τ ice no − oil > 200 kPa ( 12 ), making the precise choice of oil and oil content necessary. The above examples also illustrate why several lubricated, icephobic systems have been reported ( 4 – 6 , 8 – 11 ), whereas we have only recently observed ultralow ice adhesion strengths without the use of lubrication ( 7 ). In Fig. 5D , we compile six previous studies on lubricated, icephobic rubbers. Almost all the τ ice values for these lubricated systems fall into the region of lubrication predicted by our phase diagram. Once the lubricants are removed, the ice adhesion strengths should increase to values given by Eq. 11 . Note that the τ ice values for the surfaces created by Zhu et al. ( 11 ) fall outside the allowable range predicted by our framework, although the surface was a mixture of PDMS and silicone oil, similar to surfaces in this work ( Fig. 3C ) and in the studies of Urata et al. ( 9 ) and Aizenberg et al. ( 4 ). The ice adhesion of the unfilled PDMS in the work of Zhu et al. was reported to be τ ice = 55 kPa, much lower than the accepted literature value of τ ice = 250 to 300 kPa ( 7 , 12 , 22 ). It is likely that unreacted siloxane chains enabled interfacial slippage in their unfilled PDMS, a previously reported phenomenon ( 7 , 30 ). Recasting their data using the literature value for the ice adhesion strength of PDMS placed their measured τ ice values correctly near the lubrication regime."
} | 4,167 |
39728517 | PMC11677674 | pmc | 689 | {
"abstract": "A self-healing superhydrophobic coating was successfully prepared in the present work. The coating comprised PEG (polyethylene glycol) and Fe 3 O 4 nanoparticles modified with stearic acid (SA) via hydrogen bonds, using polyamide resin and epoxy as binders. The chemically damaged surface could restore its original superhydrophobic structure and chemical composition after 4 h at room temperature or 10 min of heating in an oven with a self-healing efficiency of 95.5% and 96.1%, respectively. The hydrogen bonds between SA-OH and Fe 3 O 4 -OH nanoparticles enabled the repeatable and efficient self-healing properties of the superhydrophobic coating. The coating exhibited remarkable chemical resistance, maintaining superhydrophobicity even after 48 h of immersion in strong acidic and alkaline solutions. Additionally, the prepared fabric showed excellent mechanical stability after 2400 mm of abrasion and 125 cycles of tape peeling with a WCA above 150°. Furthermore, the coated fabric was effective for oil/water separation and anti-icing. With these powerful functions, the proposed superhydrophobic coating holds promising applications in both daily life and industry.",
"conclusion": "4. Conclusions A self-healing superhydrophobic coating was successfully prepared. The fabric was coated with SA-Fe 3 O 4 nanoparticles by immersion, followed by further modification with SA, achieving remarkable superhydrophobic properties with a WCA of 154.8°. The coating demonstrated excellent chemical resistance even after 48 h of corrosion in strong acidic, alkaline, and saline solutions. The self-healing performance was attributed to the breakage and reintegration of hydrogen bonds between SA-OH and Fe 3 O 4 -OH nanoparticles. Only 10 min of heating was needed for the self-healing process, restoring the damaged surface to the original superhydrophobic structure. Moreover, the superhydrophobic fabric exhibited outstanding mechanical durability even after 175 tape peeling cycles or 2800 mm of abrasion. The self-healing superhydrophobic fabric proved to be a good candidate for oil/water separation and anti-icing. Consequently, a stable, durable, and efficient self-healing superhydrophobic coating was designed, significantly enhancing reliability for real-life applications and extending lifespan.",
"introduction": "1. Introduction Nature-inspired superhydrophobicity refers to surfaces with water contact angles (WCAs) greater than 150° and sliding angles (SAs) lower than 10° [ 1 , 2 ]. Superhydrophobic surfaces have garnered considerable attention for their significant potential in oil/water separation [ 3 ], self-cleaning [ 4 ], anti-biofouling [ 5 ], and anti-icing [ 6 ]. It has been reported that superhydrophobic coatings can be fabricated by creating hierarchical roughness and modifying the surface with low-energy reagents [ 7 ]. Various fabrication methods, including electrospinning [ 8 ], etching [ 9 ], spraying [ 10 ], electrodeposition [ 11 ], chemical vapor deposition [ 12 ], and hydrothermal treatment [ 13 ], have been explored. However, these coatings often suffer from structural fragility under harsh conditions, such as exposure to light and corrosive liquids [ 14 ]. Therefore, there is an urgent need to develop facile approaches to prepare robust and highly stable superhydrophobic coatings [ 15 ]. To enhance the durability of superhydrophobic coatings, researchers have explored the use of thermosetting polymers to improve mechanical stability and adhesion [ 16 ]. Others have focused on incorporating self-healing functions into superhydrophobic materials, which has proven more efficient [ 17 ]. External stimuli, such as light [ 18 ], heat [ 19 ], pH changes [ 20 ], solvents [ 21 ], applied voltage [ 22 ], or high humidity [ 23 ], are often required to trigger the self-healing capability of coatings formed by covalent networks, such as fluoroalkylsilane (FAS). Qiu et al. [ 14 ] developed a self-healing superhydrophobic surface composed of polydimethylsiloxane (PDMS), Polyfluo-150Wax, and palygorskite, which healed scratches under high-temperature treatment. Zhao et al. [ 24 ] fabricated a self-repairing superhydrophobic coating with 1H, 1H, 2H, 2H-perfluorodecanethiol (PFDT) modified SiO 2 using a spraying approach, where the wettability could be restored at 80 °C after surface damage. Zhang et al. [ 25 ] created a self-healing superhydrophobic coating on an aluminum substrate using spraying and hydrothermal reaction methods. The chemical damage healing process required 40 min at room temperature. The migration of hydrophobic chains in the low-surface-energy material FAS@PDA provided self-repairing capability, which could be enhanced by heating. Sun et al. [ 26 ] designed a self-healing superhydrophobic coating using silica particles modified with polycaprolactone, epoxy resin, EASTMAN Kristalex 3085 resin, and 1H, 1H, 2H, 2H-perfluorodecyltriethoxysilane (PFDTES) via a spray-coating method, significantly extending the lifespan of the coatings. However, commonly used self-healing agents, such as fluorinated compounds, can be harmful to humans and the environment and are costly. Additionally, achieving rapid self-repairing ability for superhydrophobic coatings without external stimuli remains a challenge. Hydrophobic bonds, being dynamic and reversible non-covalent bonds, are easily broken and reformed due to their lower energy compared to covalent bonds [ 27 , 28 ]. Hydrogen bonds can form when there are abundant O-H groups present [ 29 ]. Self-healing superhydrophobic coatings based on hydrogen bonds can restore both chemical and physical damage [ 30 ]. In this study, Fe 3 O 4 nanoparticles were modified with PEG and stearic acid (SA) via hydrogen bonds. The PEG/Fe 3 O 4 -SA nanoparticles were applied to cotton fabric using epoxy and polyamide resins. Following further modification with SA, a superhydrophobic coating was successfully prepared. Although the superhydrophobic properties could be lost after 48 h of immersion in corrosive solutions, they could be restored by simple heat treatment or storage at room temperature. The mechanical stability of the prepared coating was investigated, and its application in reusable oil/water separation was demonstrated. This method provides a simple route to design durable self-healing superhydrophobic coatings with great potential for industrial applications.",
"discussion": "3. Results and Discussion 3.1. Surface Morphology of Samples Surface morphology is crucial for fabricating wettability. Thus, analyzing the surface morphology of superhydrophobic materials is essential. Figure 2 shows the surface topography of pristine fabric, superhydrophobic fabric, Fe 3 O 4 nanoparticles, and PEG/Fe 3 O 4 -SA nanoparticles. Figure 2 a reveals a clear and smooth surface for the pristine fabric. In contrast, a rough structure with micro-protrusions appears on the surface of the superhydrophobic fabric ( Figure 2 b). The pristine Fe 3 O 4 nanoparticles ( Figure 2 c) range from 0.1 to 0.2 μm. The surface morphology of the PEG/Fe 3 O 4 -SA nanoparticles ( Figure 2 d) remains as a layer of fine nanoparticles. With the assistance of EPPA binders, PEG/Fe 3 O 4 -SA nanoparticles are modified on the fabric surface, forming a porous microstructure after ethanol evaporation. The superhydrophobic coating was applied to different substrates besides fabric, including cotton, sponge, copper, and aluminum. The superhydrophobic copper and aluminum were prepared by the spraying method. The superhydrophobic samples based on a variety of substrates are shown in Figure 3 . The WCAs of superhydrophobic pristine fabric, cotton, sponge, copper, and aluminum were 0°, 0°, 113.7°, 70.3°, and 84.9°, respectively. However, the WCAs of superhydrophobic samples were 154.8°, 153.2°, 161.2°, 151.1°, and 151.6°, proving great application potential of the superhydrophobic coating for varieties of substrates. The fabric with an EPPA solution followed by immersion in a stearic acid solution was used as a blank. The WCA of the blank fabric sample was 127.7°, proving the contribution of the PEG/Fe 3 O 4 -SA nanoparticles to superhydrophobic properties. Figure 4 a shows the FTIR spectra of the samples. In the spectra of raw and superhydrophobic fabric, the peak appearing between 3200 and 3500 cm −1 corresponds to the stretching vibrations of the –OH group in fabric. The superhydrophobic fabric exhibits new peaks compared to raw fabric. For the superhydrophobic fabric, 2960 cm −1 and 2870 cm −1 correspond to the symmetrical and asymmetrical stretching vibrations of –CH 3 and –CH 2 groups in stearic acid. The peak at 1000–1100 cm −1 corresponds to the stretching vibration of the C-O-C group, which is the characteristic absorption peak of PEG. Peaks at 1510–1570 cm −1 and 1640–1690 cm −1 correspond to the –N–H and O=C– bending vibrations of the O=C–NH– group in the EPPA binder. Based on this analysis, stearic acid, EPPA, and PEG are characterized and confirmed, demonstrating the successful grafting of these agents onto the fabric surface. The chemical nature of the superhydrophobic fabric was investigated using XPS. The spectra in Figure 4 b–d depict the presence of C, O, and Fe elements in the superhydrophobic fabric. In Figure 3 b, the C1s spectra are decomposed into four peaks: C-C (284.5 eV), C-O (286.4 eV), C=O (288.2 eV), and C-N/C-OH (285.5 eV). The C-C, C-O, and C=O bonds are related to PEG, EPPA, and SA, while the C-N bond originates from EPPA. From the Fe 2p orbitals in Figure 3 d, peaks at 714.3 and 727.3 eV are attributed to Fe 3+ , and those at 711 and 724.4 eV are attributed to Fe 2+ , confirming the existence of Fe 3 O 4 . 3.2. Self-Healing Performaces The coating’s self-healing capability was evaluated by immersing it in strong acid (0.1 M HCl), alkali (0.1 M NaOH), and saline (0.1 M NaCl) solutions for 48 h. Variations in WCAs of etched and self-healed coatings are shown in Figure 5 a. When the fabric was immersed in 0.1 M HCl solution for 48 h, the WCA of the etched coating dropped to 136.4°. After storing the etched fabric at room temperature for 4 h, the WCA stabilized at 150.7°. Heating the etched fabric in an oven at 80 °C for 10 min increased the WCA to 151.6°. The self-healing efficiency, calculated as the ratio of the WCA of the original to the healed fabric, was 95.5% and 96.1% for room temperature and oven heating, respectively. Similar self-healing efficiencies were obtained for coatings etched by NaCl and NaOH solutions. After immersion in NaCl and NaOH solutions, WCAs decreased to 138.0° and 136.8°, respectively. However, heating increased the WCAs of the healing surfaces to 151.8° and 152.6°. SEM images of surfaces etched by immersion in 0.1 M HCl, 0.1 M NaOH, and 0.1 M NaCl for 48 h are shown in Figure 5 b–d. Although some micro/nano-bulges were lost during immersion, the micro/nanostructure was largely retained. Therefore, the superhydrophobic coating demonstrated excellent anti-corrosion and self-healing capabilities. The stability of the superhydrophobic coating was further characterized by repeating the etching/self-healing processes ( Figure 6 ). After immersion in 0.1 M HCl, 0.1 M NaOH, and 0.1 M NaCl for 4 h, the WCAs of the coating were measured, followed by heating for recovery. The initial superhydrophobic coating switched to a hydrophobic state after immersion but returned to its superhydrophobic state after heating. The etching/self-healing processes were conducted for 15 cycles with no significant decrease in superhydrophobicity, demonstrating excellent cycling stability. To explain the self-healing mechanism of the prepared superhydrophobic coating, FTIR and XPS spectra of superhydrophobic fabric, etched fabric after 48 h of immersion in 0.1 M NaOH solution, and healed fabric after heating treatment were analyzed. Figure 7 a shows the FTIR analysis, comparing the chemical composition variations before and after self-healing. The C-H peaks of the etched fabric decreased compared to the superhydrophobic fabric. The O-H peaks of the etched fabric blue-shifted from 3376 cm −1 to 3429 cm −1 , indicating the loss of hydrogen bonds between Fe 3 O 4 -OH nanoparticles and SA-OH. After healing, the C-H peaks increased, and the O-H peaks red-shifted from 3429 cm −1 to 3376 cm −1 , proving the out-migration of SA and its reintegration with Fe 3 O 4 -OH groups through hydrogen bonds. XPS spectra ( Figure 7 b) showed that the C, O, and Fe contents of the superhydrophobic fabric were 84.19%, 14.32%, and 1.50%, respectively. For the etched fabric, these contents changed to 87.62%, 11.86%, and 0.52%. After healing, the contents were 86.39%, 12.11%, and 1.49%, respectively, similar to those of the superhydrophobic fabric. Water droplets on the etched fabric flattened with a WCA of 133.5°, whereas those on the healed fabric increased to 152.0°, restoring superhydrophobic properties. A possible self-healing mechanism is illustrated in Figure 7 c,d. Figure 7 c shows the pristine structure of the superhydrophobic coating. After immersion in a corrosive solution, the outer layer was damaged ( Figure 7 d), partially losing SA from the coating, and exposing −OH groups. The damage of hydrogen bonds and removal of SA increased surface free energy. To minimize surface free energy, SA migrated to the outer damaged surface, reforming hydrogen bonds between SA-OH and PEG/Fe 3 O 4 nanoparticles, thereby recovering superhydrophobic properties through the rearrangement of low-surface-energy chains in SA. A significant number of SA molecules connected through hydrogen bonds in the micro/nanostructure is crucial for the coating’s self-healing properties. 3.3. Mechanical Durability Mechanical durability is usually a challenge for superhydrophobic materials in practical applications. In this study, two durability tests were conducted to assess the mechanical robustness and self-healing properties of the fabricated coating. The tape peeling test was applied to test the degradation of the superhydrophobic surface ( Figure 8 a). A 500 g weight was rolled over the tape surface repeatedly, and the WCAs of the coated surface were measured every 25 cycles, followed by heating for self-healing. As shown in Figure 8 b, the coating maintained excellent superhydrophobicity (WCA > 151.6°) even after 125 cycles of tape peeling. The WCA decreased to 148.4° after 150 cycles but increased to 152.3° after self-healing, indicating exceptional mechanical resistance. Additionally, a sandpaper abrasion test was conducted to evaluate the coating’s abrasion resistance ( Figure 8 c). A 500 g weight was applied to the superhydrophobic coating against 1000-grit sandpaper and moved back and forth for 100 mm. Figure 7 d shows that the WCAs of the superhydrophobic coating remained above 150° after 2400 mm of abrasion. After 2800 mm of abrasion, the coating lost superhydrophobicity. After self-healing, the WCA recovered to 151.2°. These results demonstrated excellent mechanical abrasion resistance. Tang et al. [ 31 ] used methyltrimethoxysilane (MTMS) to prepare superhydrophobic wood by the immersion-spraying method. The WCA decreased to 150° after 1200 mm of sandpaper abrasion. Hou et al. [ 32 ] studied the mechanical durability of the superhydrophobic PVDF- HFP/SiO 2 /CNTs coating on the Al substrate. The tape peeling cycles were repeated 30 times. The WCA decreased from 165.7° to 155.1°. Zhao et al. [ 33 ] fabricated a robust superhydrophobic anti-icing/de-icing composite coating. The surface lost its superhydrophobic properties with a WCA below 150° after 2500 mm of sandpaper abrasion. The surfaces prepared in this study maintained excellent superhydrophobic properties and their rough structure after 175 tape peeling cycles or 2800 mm of abrasion ( Figure 8 e,f), illustrating strong adhesion between the coating and fabric. 3.4. Oil/Water Separation To evaluate the application of the prepared coating, the superhydrophobic fabric was used for oil/water separation ( Figure 9 a). Oil penetrated the superhydrophobic fabric into the beaker, retaining water outside during separation. The process was repeated to assess performance. After 10 cycles, the separation efficiency remained nearly unchanged at 98.7%, proving stable performance ( Figure 9 b). The separation efficiency decreased to 95% after 16 cycles. The used superhydrophobic fabric was washed with AE and heated at 80 °C for 10 min for self-healing. The healed fabric was reused for repeated separation processes, maintaining high efficiency even after 25 cycles, indicating its significant development potential and wide application in oil/water separation. 3.5. The Anti-Icing Ability of the Coating Water droplets on bare copper and superhydrophobic copper were observed to freeze in the low-temperature condition. The time of ice forming was monitored. Figure 10 shows the photos of the freezing process of a 25 μL water droplet on different surfaces. In the freezing process, water droplets changed from transparent to hemispherical and to a peach shape. For bare copper, the droplet showed a peach shape at 73 s at a temperature of −15 °C, illustrating the total freezing of the droplet. The water droplets on superhydrophobic copper showed a peach shape at 238 s at a temperature of −15 °C, exhibiting a longer delay during the freezing process. Superhydrophobic properties could enhance the removal of droplets and minimize the contact area on surfaces, as well as reduce nucleation sites. This ultimately contributes to a delay in freezing time."
} | 4,386 |
37464396 | PMC10355004 | pmc | 690 | {
"abstract": "Background Fatty acid-derived products such as fatty alcohols (FAL) find growing application in cosmetic products, lubricants, or biofuels. So far, FAL are primarily produced petrochemically or through chemical conversion of bio-based feedstock. Besides the well-known negative environmental impact of using fossil resources, utilization of bio-based first-generation feedstock such as palm oil is known to contribute to the loss of habitat and biodiversity. Thus, the microbial production of industrially relevant chemicals such as FAL from second-generation feedstock is desirable. Results To engineer Corynebacterium glutamicum for FAL production, we deregulated fatty acid biosynthesis by deleting the transcriptional regulator gene fasR , overexpressing a fatty acyl-CoA reductase (FAR) gene of Marinobacter hydrocarbonoclasticus VT8 and attenuating the native thioesterase expression by exchange of the ATG to a weaker TTG start codon. C. glutamicum ∆ fasR cg2692 TTG (pEKEx2- maqu2220 ) produced in shaking flasks 0.54 ± 0.02 g FAL L −1 from 20 g glucose L −1 with a product yield of 0.054 ± 0.001 Cmol Cmol −1 . To enable xylose utilization, we integrated xylA encoding the xylose isomerase from Xanthomonas campestris and xylB encoding the native xylulose kinase into the locus of actA . This approach enabled growth on xylose. However, adaptive laboratory evolution (ALE) was required to improve the growth rate threefold to 0.11 ± 0.00 h −1 . The genome of the evolved strain C. glutamicum gX was re-sequenced, and the evolved genetic module was introduced into C. glutamicum ∆ fasR cg2692 TTG (pEKEx2- maqu2220 ) which allowed efficient growth and FAL production on wheat straw hydrolysate. FAL biosynthesis was further optimized by overexpression of the pntAB genes encoding the membrane-bound transhydrogenase of E. coli . The best-performing strain C. glutamicum ∆ fasR cg2692 TTG CgLP12::(P tac - pntAB -T rrnB ) gX (pEKEx2- maqu2220 ) produced 2.45 ± 0.09 g FAL L −1 with a product yield of 0.054 ± 0.005 Cmol Cmol −1 and a volumetric productivity of 0.109 ± 0.005 g FAL L −1 h −1 in a pulsed fed-batch cultivation using wheat straw hydrolysate. Conclusion The combination of targeted metabolic engineering and ALE enabled efficient FAL production in C. glutamicum from wheat straw hydrolysate for the first time. Therefore, this study provides useful metabolic engineering principles to tailor this bacterium for other products from this second-generation feedstock. Supplementary Information The online version contains supplementary material available at 10.1186/s13068-023-02367-3.",
"conclusion": "Conclusion This study established for the first time de novo production of FAL by C. glutamicum , both from a first and a second-generation feedstock. To achieve this, we systematically engineered the FA metabolism and its regulation and optimized the culture conditions. Additional implementation of a transferrable plasmid-free xylose utilization module and further strain optimization enabled sustainable and efficient FAL production from wheat straw hydrolysate. Eventually, a fed-batch process on this second-generation feedstock was established. Therefore, the applied engineering approach and the established bioprocess provide useful principles to optimize FAL production further and utilize C. glutamicum as a robust platform for producing other products from wheat straw hydrolysate.",
"introduction": "Introduction Corynebacterium glutamicum is an established workhorse for the large-scale production of several amino acids, such as l -lysine and l- glutamate, in millions of tons per year [ 83 ]. This facultative anaerobic Gram-positive bacterium is generally recognized as safe (GRAS), robust, and grows on several sugars, organic acids, and phenolic compounds as single or combined carbon and energy sources [ 6 , 22 , 47 , 53 , 70 , 73 ]. The naturally accessible carbon source spectrum was expanded by the introduction of heterologous pathways [ 89 ], among others, to access xylose, which was first attained by heterologous overexpression of xylose isomerase encoding gene xylA from E. coli in combination with native xylulokinase activity and thereupon improved by co-overexpression of xylA and xylB genes from different origins [ 36 , 53 ]. Corynebacterium glutamicum shows a comparable high tolerance to inhibitors such as aromatic compounds typically found in lignocellulosic hydrolysates [ 6 , 20 , 69 , 70 ]. Therefore, in several studies, this bacterium was utilized for the production of chemicals and fuels such as lactic, succinic, cis , cis -muconic, itaconic, 5-aminovaleric acid or 1,2-propanediol and isobutanol from non-food biomass hydrolysates [ 11 , 39 , 40 , 46 , 48 , 62 , 63 ]. The non-oleaginous C. glutamicum lacks essential genes for the β-oxidation of fatty acids [ 4 ] and has already been engineered for the production of lipids and fatty acids (FA) [ 33 , 54 , 72 , 74 ]. However, the production of fatty alcohols (FAL) and derivates thereof with C. glutamicum has not been described so far. Because of their broad application as detergents, lubricants, and additives in cosmetic products [ 1 , 21 ], a continuously increasing demand for FAL is expected [ 51 ]. Commercially, FAL are either produced petrochemically by the oligomerization and subsequent oxidation of ethylene [ 38 ], or by hydrogenation of bio-based fatty acids and fatty acid methyl esters [ 86 ]. Besides the required usage of expensive metal catalysts and the high energy demand [ 58 ], the use of fossil resources and extensive farming of oil plants in monocultures propagate global warming, deforestation, and subsequently, loss of biodiversity [ 26 , 79 ]. Thus, microbial production is an alternative approach for the sustainable production of FAL, especially when agricultural waste or side streams are utilized as substrate. For the biosynthesis of FAL, two well-described pathways are commonly applied: (A) reduction of a free FA or of acyl-ACP/CoA to fatty aldehydes by a carboxylic acid reductase (CAR) or, respectively, by an aldehyde-forming fatty acyl-ACP/CoA reductase (AH-FAR) with a subsequent second reduction to a long-chain alcohol by an alcohol dehydrogenase (ADH) or aldehyde reductase (AHR) [ 1 , 90 ] or (B) two-step reduction of an activated FA by an alcohol-forming fatty acyl-CoA reductase (FAR) [ 16 , 43 ] with an aldehyde intermediate that is formed during the four-electron transfer but which is not released from the enzyme [ 32 , 85 ]. Especially the latter pathway has been extensively exploited to heterologously produce FAL in various organisms such as Escherichia coli [ 43 ], Saccharomyces cerevisiae [ 18 ], and Yarrowia lipolytica [ 16 ] as it solely requires the expression of one gene and does not form free cytotoxic aldehyde intermediates. The commonly used, NADPH-dependent FAR enzymes Maqu_2220 and Maqu_2507 of the marine bacterium Marinobacter hydrocarbonoclasticus VT8 were shown to accept both acyl-ACP and acyl-CoA as substrate, with a higher affinity for C16-C18 acyl-CoAs [ 32 , 85 ]. This renders both reductases promising candidates to be expressed in C. glutamicum , as its native FA biosynthesis primarily produces palmityl-, stearyl- and oleoyl-CoA as intermediates [ 35 , 67 ]. The pathway’s precursor malonyl-CoA is supplied by the carboxylation of acetyl-CoA, catalyzed by the acetyl-CoA carboxylase (ACC) [ 27 ]. In contrast to many other prokaryotes, the successive condensation and elongation reactions of FA biosynthesis in C. glutamicum are catalyzed by two type I fatty acid synthases (FAS-I) [ 57 ]. Thus, in contrast to ACP-bound thioesters found in FAS-II-utilizing microbes, not only substrates but also products of the two multienzymes Fas-IA and Fas-IB are CoA-bound [ 57 ]. Hydrolysis of the respective thioesters catalyzed by an acyl-CoA thioesterase (Tes) results in the formation of the free FAs palmitic acid (hexadecenoic acid), stearic acid (octadecanoic acid) and the monounsaturated oleic acid ( cis -9-octadecenoic acid) [ 33 ]. Those FA serve further as precursors for membrane lipid and mycolic acid biosynthesis. In the presence of acyl-CoAs, FA biosynthesis is tightly regulated by FasR. The TetR-type transcriptional regulator inhibits transcription of both FAS-I-encoding genes fasA and fasB , and of the ACC catalytic subunit-encoding genes accD1 and accBC [ 34 , 52 ]. To tailor C. glutamicum for FAL production, we systematically engineered the FA metabolism and its regulation, optimized the culture conditions, and enabled ALE-supported xylose utilization (Fig. 1 ). Finally, we validated the performance of the newly constructed strain in a bioreactor setup using wheat straw hydrolysate. Fig. 1 Metabolic pathways for fatty acid biosynthesis and heterologous FAL production in C. glutamicum . The heterologous modules for FAL synthesis, xylose utilization, and NADPH regeneration are highlighted in green, purple, and grey, respectively. ACC acetyl-CoA carboxylase, FadD acyl-CoA synthetase, FAR fatty acyl-CoA reductase, Fas-IA type I fatty acid synthase A, Fas-IB type I fatty acid synthase B, GltA citrate synthase, MA mycolic acid, PDHC pyruvate dehydrogenase complex, PL phospholipid, PntAB transhydrogenase, Tkt transketolase, XylA xylose isomerase, XylB xylulokinase. The red arrow and the black X represent decreased expression and deletion of the corresponding gene, respectively",
"discussion": "Discussion Previously, Takeno et al. [ 74 ] demonstrated that deregulation of the FA biosynthetic pathway by inactivating the transcriptional regulator FasR [ 52 ] is an effective approach to increase FA synthesis. In contrast to other published studies [ 33 , 74 ], FA overproduction and efflux relied on nitrogen-limiting cultivation conditions. Accordingly, N-limitation is a common practice for lipid production with several oleaginous organisms [ 2 , 3 ]. While nitrogen limitation was essential for FA production, FAL synthesis followed a growth-coupled production kinetic improving yield, titer, and volumetric productivity. Similarly, Fillet et al. [ 25 ] reported that a high C:N ratio was beneficial for FA production with Rhodosporidium toruloides , while a low C:N ratio influenced FAL production positively. However, the underlying molecular mechanisms in C. glutamicum still need to be elucidated. As chassis with increased flux through the FA biosynthesis pathway, C. glutamicum Δ fasR was a suitable basis strain to be engineered for subsequent FAL production. Screening of the two FARs Maqu_2220 and Maqu_2570 of M. hydrocarbonoclasticus VT8 without codon optimization revealed that both enzymes were active in C. glutamicum . Maqu_2220 was more suited for FAL production by C. glutamicum as 35% higher titers were reached with this reductase within 24 h. The observed efficient conversion of long-chain acyl-CoAs into their corresponding alcohols fits with enzymatic studies reporting a high affinity to long-chain acyl-CoA thioesters [ 32 , 85 ]. Additionally, the obtained FAL distribution agrees with the previously reported FA distribution of C. glutamicum [ 15 ] and indicates that neither of the reductases had a bias toward one of the three available acyl-CoA substrates. As reported previously, C. glutamicum natively possesses a high thioesterase activity [ 33 , 74 ]. While reducing this competing side reaction led to an increased FAL production in plasmid-free strains by up to 750%, the introduced start codon exchange had an adverse effect on plasmid-harboring strains. It can be speculated whether the extensive pressure on the acyl-CoA node by a reduced flux towards mycolic acids and additionally high FAR expression and, thus, depletion of acyl-CoAs led to some metabolic burden in those strains. Nevertheless, extracellular FAL titers increased by 120%, and the relative amount of extracellular FAL increased from 12 to 30% in plasmid-harboring TTG start codon mutants. Portevin et al. [ 55 ] and Takeno et al. [ 72 ] reported an increased permeability of C glutamicum upon deleting or downregulating genes directly involved in the mycolic acid biosynthesis downstream of the acyl-CoA node. Attenuating the thioesterase expression thus had a similar effect. To enable plasmid-free utilization of xylose, we introduced the xylAB genes into the locus of the natively highly expressed actA gene [ 62 ], which was dispensable without notable effects on growth with glucose [ 78 ]. The quantitative mRNA analysis displayed higher transcription of xylAB in C. glutamicum gX, although the promoter sequence and transcriptional start site of replaced actA gene [ 62 ] were not directly affected by the genomic rearrangement. Therefore, the increased transcription level is obviously the result of an unknown regulatory effect. However, the duplicated region in gX includes the RBS of the originally encoded gene actA and resulted in the occurrence of two RBS at a distance of 53 bp, suggesting boosted translation [ 88 ]. Xylose utilization by integration of xylAB genes into the genome of C. glutamicum has been accomplished before. The insertion of one xylAB copy under the control of the constitutive trc promoter into C. glutamicum R, enabled growth with xylose as the sole carbon and energy source [ 61 ] but was limited due to low expression levels of both genes, similar to C. glutamicum Δ actA :: xylAB . Instead of ALE, Sasaki et al. [ 61 ] gradually improved growth by the insertion of up to four additional copies of the xylAB genes and elevated the growth rate by 50% from 0.13 to 0.20 h −1 [ 61 ]. In this study, ALE improved the growth rate by 200% from 0.03 to 0.11 h −1 . However, further optimization of xylAB expression might be beneficial for even faster growth of C. glutamicum gX. This might be achieved by the application of fully automated and miniaturized ALE, which was developed and applied to improve the growth rate on xylose 2.6-fold of a C. glutamicum strain equipped with the Weimberg pathway for xylose utilization [ 56 ]. The genetic changes led to the functional inactivation of the GntR-type transcriptional repressor IolR [ 56 ], affecting improved xylose uptake by derepression of the glucose and myo -inositol permease IolT1 [ 9 ]. Recently, ALE yielded an excellent xylose-utilizing C. glutamicum strain, featuring rapid growth (µ max = 0.34 ± 0.00 h −1 ) with xylose [ 71 ]. One identified beneficial mutation occurred in an endogenous LacI-type transcriptional regulator involved in inositol metabolism ( ipsA , cg2910) [ 5 ]. Taken together, these findings indicate a close connection between xylose and inositol metabolism and suggest the potential to accelerate xylose utilization of C. glutamicum gX by, e.g., derepression of iolT 1. Moreover, similar to the genomic rearrangement in C. glutamicum gX, Sun et al. [ 71 ] identified a beneficial point mutation in the P sod promoter of the integrated genes xylAB . Additionally, they found and verified a beneficial 21 bp deletion in the 5`UTR promoter region of the genomically integrated xylose importer gene araE [ 71 ]. Xylose uptake is known as a limiting factor for the efficient utilization of this sugar, as exemplified by various reports on engineering of heterologous importer genes [ 14 , 46 , 61 , 71 , 87 ]. Therefore, improving xylose uptake in C. glutamicum gX seems to be a straightforward approach to advance the growth properties further. C. glutamicum gX turned out to be an excellent glutamate producer and surpassed the reported product yield of 0.13 Cmol Cmol −1 (0.13 g g −1 ) from glucose [ 45 ] by 40% when xylose served as the sole carbon source. The transfer of the evolved Δ actA :: xylAB module into lysine, DPA, and NMePhe production strains enabled plasmid-free biosynthesis from xylose as sole or with glucose as a combined carbon source. In particular, the product yields for NMePhe and l -lysine of the corresponding strains carrying the gX module with xylose as substrate met or even surpassed previously published yields of plasmid-harboring C. glutamicum mutants [ 41 , 53 ]. Moreover, the successful production of the four compounds mentioned has shown that the implementation of the gX module is generally a promising and transferable approach to make production from xylose-containing second-generation feedstocks, such as lignocellulosic hydrolysates, more sustainable. This was subsequently also demonstrated for FAL, which we produced from refined glucose and wheat straw hydrolysate. Despite the successfully introduced xylose-utilization module, the majority of carbon provided by xylose and acetate seemed not to end up in FAL. When grown on xylose [ 12 ] or acetate [ 84 ], C. glutamicum exhibits a strongly increased TCA activity, resulting in an increased carbon flux from acetyl-CoA into the TCA. There, a substantial amount of the carbon is oxidized to CO 2 [ 12 , 84 ]. For both substrates alike, just a small fraction of carbon is channeled through gluconeogenesis and can thus enter the PPP via the NADPH-regenerating reactions catalyzed by the glucose-6-phosphate and 6-phosphogluconate dehydrogenase [ 12 , 84 ]. As a consequence, NADPH might become limiting in the NADPH-expensive FAL production. When expressing the NADPH-regenerating E. coli transhydrogenase PntAB titers, yields, and volumetric productivities were improved in xylose-utilizing strains. Nevertheless, HPLC data still suggested that product formation was primarily based on carbon provided by glucose, indicating that the precursor supply for FAL synthesis might be limiting when growing on xylose. Reducing the citrate synthase expression as an attempt to increase the acetyl-CoA pool impacted FAL production negatively. Those findings contrast a study by Milke et al. [ 49 ], who reportedly improved the malonyl-CoA supply for plant polyphenol production by reducing the citrate synthase expression in a C. glutamicum mutant. The same authors simultaneously overexpressed the native acetyl-CoA carboxylase (ACC) [ 49 ], which may also be a suitable approach for the here-described FAL producers. Despite deregulated transcription of the ACC’s subunit-encoding genes accD1 and accBC by deletion of fasR , the increased flux towards malonyl-CoA might still not be sufficient to fully de-bottleneck the respective pathway. Even though numerous studies about microbial FAL production have been published using primarily different yeasts [ 16 , 17 , 59 ] or E. coli [ 13 , 24 , 44 ], C. glutamicum has not been engineered yet for the production thereof. Reported titers and volumetric productivities obtained by oleaginous yeasts range from 98 mg FAL L −1 and 1 mg FAL L −1 h −1 [ 59 ] to 5.8 g FAL L −1 and 25 mg FAL L −1 h −1 [ 16 ] using S. cerevisiae and Y. lipolytica , respectively. For FAL production with E. coli titers and volumetric productivities of up to 12.5 g FAL L −1 and 174 mg FAL L −1 h −1 are reported [ 24 ]. However, respective cultivations with E. coli are often conducted in minimal salts media which contain significant concentrations of complex substrates such as yeast extract or tryptone besides the main carbon source [ 13 , 24 , 44 ]. Thus, data about the actual yields, including metabolization of the complex substrates, were rarely provided or solely calculated based on the main substrate’s consumption. Additionally, studies using real second-generation feedstocks like lignocellulosic hydrolysates for FAL production are rare [ 17 ], and no such attempts, to our knowledge, have been published with E. coli yet. Here, we report a yield, volumetric productivity, and titer of 0.054 ± 0.005 Cmol Cmol −1 , 0.109 ± 0.005 g FAL L −1 h −1 and 2.45 ± 0.09 g FAL L −1 , respectively, which were obtained in a pulsed-fed batch process in a 1-L scale using wheat straw hydrolysate as carbon source. General strain performance can be further improved through targeted and non-targeted approaches focusing on the aforementioned xylose import to improve substrate utilization and the ACC as the bottleneck of the FA biosynthetic pathway. In that context, the constructed final strain C. glutamicum Δ fasR cg2692 TTG CgLP12::(P tac -pntAB -T rrnB ) gX (pEKEx2- maqu2220 ) provides a competitive base for future strain engineering strategies. Therefore, this study does not just represent the first report of FAL production with C. glutamicum , but also provides valuable insights into the production thereof from lignocellulosic hydrolysate and the substrate’s metabolization. Furthermore, the organism’s apparent resilience towards high hydrolysate concentrations provides a solid base for sustainable FAL production."
} | 5,208 |
30241736 | PMC6370000 | pmc | 692 | {
"abstract": "Research on mycorrhizal interactions has traditionally developed into separate disciplines addressing different organizational levels. This separation has led to an incomplete understanding of mycorrhizal functioning. Integration of mycorrhiza research at different scales is needed to understand the mechanisms underlying the context dependency of mycorrhizal associations, and to use mycorrhizae for solving environmental issues. Here, we provide a road map for the integration of mycorrhiza research into a unique framework that spans genes to ecosystems. Using two key topics, we identify parallels in mycorrhiza research at different organizational levels. Based on two current projects, we show how scientific integration creates synergies, and discuss future directions. Only by overcoming disciplinary boundaries, we will achieve a more comprehensive understanding of the functioning of mycorrhizal associations.",
"conclusion": "Concluding Remarks and Future Perspectives Key to achieving the integration of mycorrhiza research across different organizational levels is to design and execute common projects where research questions are addressed at different levels at the same time (see Outstanding Questions ). A fruitful approach is to link laboratory and field experiments through common questions, which build on each other. In the laboratory, important processes can be studied by manipulating specific biotic and abiotic parameters precisely. These findings generate hypotheses on their ecological role, which should be tested in field settings. Here, the test organisms are subject to different treatments within the context of a multitude of biotic and abiotic interactions. In turn, based on patterns that are found in real-world ecosystems, single processes and mechanisms can be disentangled using targeted laboratory experiments. Given that real-world ecosystems comprise a multitude of interacting and interwoven processes, it remains challenging to select single processes to be tested independently in the laboratory. Overall, both approaches can create positive feedback loops that fuel each other, promoting hypothesis-driven research at different levels. Another course is to incorporate laboratory approaches and methods in field experimental settings for the integration of different organizational levels. However, the main challenge is the application of highly sensitive analytical techniques designed for in vitro laboratory environments. Furthermore, it requires the collaboration of researchers from both fields to overcome the limitations of individual disciplines. To achieve this goal, it is essential, although challenging, to foster collaborations in common projects at all levels; from problem detection and hypothesis development to experimental (or observational) setup, analysis, and interpretation of findings [ 46 ]. The realization of joint workshops and conferences may lay foundations for such enterprises. Similarly, integrative MS and PhD courses will train the next generation of mycorrhizal scientists to apply both ecological and molecular approaches in their research projects. Initial attempts to make mycorrhiza research crossdisciplinary are underway in the PhytOakmeter project. Here, clonal oaks ( Quercus robur L.), which were originally developed for highly standardized molecular measurements, are now transplanted into field experiments over wide environmental gradients ( Box 1 ). Another example is the tree diversity experiment, MyDiv, where plots consist of tree communities that are characterized by different types of mycorrhization (namely ECM and AM). MyDiv integrates part of the clonal oaks from the PhytOakmeter project, and further implements omics’ approaches (i.e., metabolomics) in the field ( Box 2 ). These are only two example projects among several that are in the process of overcoming the specific technical and scientific boundaries of the single disciplines. Such synergistic projects can ultimately advance the comprehensive understanding of mycorrhizal associations and their functioning that spans from genes to ecosystems and how this can contribute to solving fundamental issues, such as sustainable food production while preserving our planet’s biodiversity."
} | 1,059 |
35221533 | null | s2 | 694 | {
"abstract": "One of the long-standing problems for the nanoparticle-based liquid-repellent coatings is their poor adhesion to substrates. For polymers of low glass transition temperature, it is highly desirable to have low temperature coating strategy to fabricate robust superhydrophobic films. Here, we report a facile method for fabricating robust, transparent, superhydrophobic films on polymer substrates. A mixture of silica particles and silica-based oligomers was spin coated on polymer substrates, followed by oxygen plasma treatment and vapor deposition of 1H,1H,2H,2H-Perfluorodecyltriethoxysilane (FDTS). The resulting superhydrophobic surface has a static contact angle at 160° and contact angle hysteresis lower than 5°. This study provides a practical solution to improve the adhesion of superhydrophobic films on polymer substrates in ambient conditions."
} | 214 |
34430203 | PMC8367835 | pmc | 695 | {
"abstract": "Utilization of lignin, an abundant renewable resource, is limited by its heterogenous composition and complex structure. Biological valorization of lignin provides advantages over traditional chemical processing as it occurs at ambient temperature and pressure and does not use harsh chemicals. Furthermore, the ability to biologically funnel heterogenous substrates to products eliminates the need for costly downstream processing and separation of feedstocks. However, lack of relevant metabolic networks and low tolerance to degradation products of lignin limits the application of traditional engineered model organisms. To circumvent this obstacle, we employed Acinetobacter baylyi ADP1, which natively catabolizes lignin-derived aromatic substrates through the β-ketoadipate pathway, to produce mevalonate from lignin-derived compounds. We enabled expression of the mevalonate pathway in ADP1 and validated activity in the presence of multiple lignin-derived aromatic substrates. Furthermore, by knocking out wax ester synthesis and utilizing fed-batch cultivation, we improved mevalonate titers 7.5-fold to 1014 mg/L (6.8 mM). This work establishes a foundation and provides groundwork for future efforts to engineer improved production of mevalonate and derivatives from lignin-derived aromatics using ADP1.",
"conclusion": "4 Conclusions By utilizing the diverse metabolism and genetic maleability of ADP1, we engineered a strain capable of producing titers of mevalonate up to 1014 mg/L (6.8 mM) from mixed glucose and lignin-derived aromatic carbon sources. The strain showed functional expression of a heterologous pathway in the presence of multiple lignin-derived aromatic compounds, and production capability was significantly improved by a genetic knock-out targeting wax ester synthesis. In the future, our work will address the long-term stability of the mevalonate pathway. Genetic instability that gradually eliminates mevalonate production poses an obstacle to industrial applications of this strain. Chromosomal integration may enable expression while limiting plasmid-based metabolic burden, but it is unlikely to completely alleviate pathway burden. Based on our data, culture medium containing a mixture of aromatic and sugar-based carbon sources may reduce mutational frequency, indicating promise for future lignin-based medium optimization. In all, this work lays a foundation for lignin-based metabolic engineering to produce mevalonate pathway products in Acinetobacter baylyi ADP1 and provides a case study for exploring the impact of the expression of a burdensome heterologous pathway on native ADP1 metabolism.",
"introduction": "1 Introduction Lignin is the second most abundant biomass-derived carbon source on Earth and represents a renewable reservoir of energy-dense substrate to perform green chemistry ( Ragauskas et al., 2014 ). Annual worldwide production is approximately fifty-million tons, and this production is projected to increase significantly as it is a byproduct of biofuel production technology ( Norgren and Edlund, 2014 ). Due to the variability in both composition and bonding structure of lignin, processing is challenging ( Linger et al., 2014 ; Schutyser et al., 2018 ). Furthermore, utilization of processed lignin is difficult due to the inherent compositional heterogeneity of lignin degradation products. Currently, most lignin is treated either as a waste stream or is burned as a solid fuel in biorefinement processes ( Schutyser et al., 2018 ). However, biological upgrading, or valorization of lignin provides advantages over traditional chemical processing or combustion by enabling conversion of diverse lignin-derived substrates to high value products ( Linger et al., 2014 ). Aromatics-degrading microorganisms can catabolize the broad range of substrates found in processed lignin ( Salvachúa et al., 2015 ). The substrate funneling characteristic of microbial lignin-derived aromatics metabolism allows for the utilization of diverse lignin-derived compounds without prior separation. Several microorganisms, including Rhodococcus and Pseudomonas species, can synthesize a range of products like triacylglycerols (TAGs) and polyhydroxyalkanoate (PHA) from lignin derivatives ( Linger et al., 2014 ; MacEachran and Sinskey, 2013 ). Acinetobacter baylyi ADP1 represents a promising candidate for biological valorization of lignin-derived compounds ( Beckham et al., 2016 ). In addition to its versatile metabolism, it possesses natural competence and native homologous recombination machinery that enable rapid and targeted genomic manipulations ( Barbe et al., 2004 ; Elliott and Neidle, 2011 ). Emerging tools for ADP1 have further broadened the feasible scope of engineering and enabled rapid iteration through design-build-test-learn cycles ( Biggs et al., 2020 ; Suárez et al., 2019 ). Leveraging the advantageous characteristics of ADP1, we engineered a strain capable of expressing the mevalonate pathway during growth on lignin-related aromatic substrates. Mevalonate is a small molecule with applications in cosmetics and as a monomer precursor to some classes of polyesters ( Wang, 2017 ). It is also a precursor to terpenoids that have applications in industries ranging from biofuels production to flavorings and fragrances ( Ajikumar et al., 2008 ; Belcher et al., 2020 ; Zhang et al., 2017 ). The dedicated pathway for mevalonate synthesis requires three acetyl-CoA and two NADPH molecules, which at high flux can strain native metabolism. Thus, expression of this pathway in ADP1 provides an opportunity to study the impact of acetyl-CoA and NADPH siphoning in ADP1 as well as to identify potential obstacles to be overcome towards utilization of ADP1 as a production host. We show mevalonate pathway activity in the presence of various lignin-derived aromatic compounds and improved productivity by eliminating the resource competitive wax ester pathway. In addition, we conducted fed-batch cultures to evaluate productivity over time. This work demonstrates ADP1 as a host for biological valorization of lignin-derived substrates to mevalonate and adds to the body of previous efforts ( Ishige et al., 2002 ; Luo et al., 2019 ; Santala et al., 2011 , 2014 ) to synthesize industrially important products using ADP1's potential as an metabolic engineering host and indicates targets for future engineering.",
"discussion": "3 Results and discussion 3.1 Mevalonate pathway is functionally expressed in ADP1, but not genetically stable in protocatechuate-only medium Initial validation of pathway functionality was performed by culturing ADP1 pMev-LacI-trc (ECA10) in minimal M9 medium under three conditions. One condition was supplied with 1.54 g/L (10 mM) PCA, a metabolic intermediate in the catabolism of lignin-derived aromatic compounds, as the sole carbon source. A second condition was supplied with 0.31 g/L (2 mM) PCA and 0.2% (w/v) (11.1 mM) glucose. The last condition was supplied with 1.54 g/L (10 mM) PCA and 0.2% (w/v) (11.1 mM) glucose. Mevalonate and cell density was measured after 24 hours of cultivation in all conditions [ Fig. 1 B]. These data indicate that the mevalonate pathway can be expressed functionally in ADP1, and based on the PCA-only condition, can produce mevalonate solely from an aromatic, lignin-related carbon source (p < 0.01 relative to no mevalonate control). Notably, cultures containing PCA as the sole carbon source accumulated very little mevalonate and grew to a lower OD in comparison to cultures supplemented with glucose ( Fig. 1 B). pMev-LacI-trc (pECA03) plasmids were isolated at 24 hours and sequenced for each culture. Sequencing revealed partial deletions of the mevalonate pathway present in PCA-only cultures [ Fig. 1 C]. These deletions occurred at various loci and impacted multiple genes in the mevalonate pathway. PCA is known as a poor substrate due to its tendency to chelate iron, which may have caused significant cellular stress ( Garner et al., 2004 ). Glucose appears to obviate this stress and stabilize the plasmid. Fig. 1 ADP1 produces mevalonate, but plasmid instability reduces production in PCA-only condition. ADP1 was cultured in batch mode for 24 hours with expression of the mevalonate pathway. (A) The mevalonate pathway comprising atoB, ERG13, and HMG1 , catalyzes the conversion of three molecules of acetyl-CoA into mevalonate. (B) Mevalonate was detected in all conditions, including in medium containing solely 10 mM PCA (p = 0.01, t -test, 2-tailed). The primary vertical axis shows mevalonate production. The secondary vertical axis shows OD 600 . Data is mean and error bars represent S.E.M. (n = 2). (C) Plasmid instability in the 10 mM PCA condition occurred for glucose-free medium. The top band represents the full pathway, with the second, third, and fourth bands being distinct mutants found in the 10 mM PCA-only condition cultures. Fig. 1 3.2 Mevalonate production is affected by the aromatic acid species present After seeing mevalonate titers were very sensitive to carbon source, we examined the productivity in the presence of various lignin-derived aromatic acids ( Salvachúa et al., 2015 ). Aromatic acid substrates were chosen to represent both branches of the β-ketoadipate pathway, funneling either to PCA (POB and ferulate) or catechol (benzoate and anthranilate) ( Salvachúa et al., 2015 ). To reduce growth-inhibition effects, ferulate, benzoate, and anthranilate were supplied at 5 mM concentration, while POB was supplied at 20 mM concentration ( Fischer et al., 2008 ) [ Suppl. Fig. 2 ]. All culture medium contained either POB, ferulate, benzoate, or anthranilate and 0.18% (w/v) (10 mM) glucose as a supplemental carbon source. The glucose was added to ensure plasmid stability. At 48 hours, cultures were harvested and analyzed for mevalonate production [ Fig. 2 ]. Ferulate/glucose cultures resulted in the lowest average titer of mevalonate, which may be due to lower tolerance of ferulate by ADP1 relative to other monomers ( Luo et al., 2019 ). Cultures grown with POB/glucose had 4.6-fold higher titer relative to the other carbon source cultures tested [ Fig. 2 A]. To interrogate whether this was an artifact of the abundance of POB relative to other carbon sources, the carbon utilization, as c-mmoles of mevalonate per c-mole of substrate (for both POB and glucose), was calculated. Analysis of carbon utilization revealed a statistically significant increase in carbon yield for POB cultures over ferulate, benzoate, and anthranilate [ Fig. 2 B]. Due to its higher production per OD and low toxicity, POB/glucose was chosen as the carbon source mixture for future cultivations. Fig. 2 Aromatic substrates affect growth, titer, and yield. As PCA only condition resulted in low titers and mutations, we screened four aromatics to determine which was best tolerated. Glucose was included to repress mutations found in aromatic only experiments. (A) Mevalonate titers (left axis) and optical density (right axis) after 48 h. (B) Carbon-molar yield of mevalonate on substrate. Both glucose and aromatic carbon were included as substrate. Data reflect mean, and error bars are S.E.M. (n = 4). Fig. 2 3.3 Mevalonate is efficiently produced from p-hydroxybenzoate as sole carbon source Because POB performed well when glucose was supplemented, we conducted batch cultivation of ADP1 pMev-LacI-trc (ECA10) with M9 minimal medium containing 20 mM POB as the sole carbon source and observed increased mevalonate production relative to the PCA-containing cultures [ Suppl. Fig. 1 ]. For cultures containing solely 20 mM POB, solely 10 mM glucose, or a mixture of 20 mM POB and 10 mM glucose as carbon sources, no mutations to the mevalonate plasmid were observed. Mixed carbon substrates led to the highest titers and are more representative of the native growth conditions of ADP1 ( Linger et al., 2014 ). Accordingly, glucose was included as a secondary carbon source in subsequent experiments. 3.4 Removing wax ester synthesis and culturing in fed-batch enables higher production of mevalonate from aromatic carbon Next, we studied the impact of competition for acetyl-CoA on mevalonate production. The wax ester pathway competes for acetyl-CoA through fatty acid synthesis and is known to be highly active in ADP1 under carbon-rich and nutrient-limited, particularly nitrogen-limited, conditions ( Alvarez and Steinbüchel, 2002 ). Fed-batch cultivation was utilized to enable conditions that were carbon-rich and nutrient-limited. ADP1Δ acr1 pMev-LacI-trc (ECA15) was evaluated alongside ADP1 pMev-LacI-trc (ECA10) for long-term mevalonate production in 168 hour fed-batch cultivations [ Fig. 3 A]. Three conditions were evaluated for mevalonate production -- one with addition of POB boli, one with glucose boli, and a non-fed culture with no additional carbon subsequently added. The data show a significant 7.5-fold increase in mevalonate production in the wax ester knock-out strain compared to WT in the POB fed-batch. For the wax ester knock-out strain expressing the mevalonate pathway, cultures reached a titer of 1014 ± 379 mg/L (6.8 mM ± 2.6 mM) with a yield of 41.4 ± 17.6 c-mmol/c-mol of all substrate consumed at 168 h. Glucose fed-batch also compared favorably in the wax ester knock-out compared to WT, although to a lesser extent than POB. Cultures not supplied with additional non-carbon nutrients exhibited slow or insignificant growth rate after initial exponential phase growth [ Suppl. Fig. 3 ]. Fig. 3 Mevalonate production is improved by wax ester knock-out and fed-batch cultivation. A. baylyi ADP1 pMev-LacI-trc (ECA10) and the wax ester knock-out, ADP1Δ acr1 pMev-LacI-trc (ECA15) were cultivated in fed-batch. Cultures initially contained 20 mM POB and 10 mM glucose. Cultures were fed every 24 hours one bolus of 20 mM substrate, either p -hydroxybenzoate (POB) or glucose, or not fed, as indicated. (A) Mevalonate titers after 168 h of POB fed-batch. (B) Mevalonate titers over culture time for in ADP1Δ acr1 pMev-LacI-trc (ECA15) with POB feeding. Data reflect mean and error bars represent S.E.M. (n = 6 for ADP1 pMev-LacI-trc POB, glucose, not fed, and for ADP1Δ acr1 pMev-LacI-trc POB, and n = 2 for ADP1Δ acr1 pMev-LacI-trc glucose and not fed). Fig. 3 A time-course analysis of mevalonate concentration revealed that mevalonate production continued via POB catabolism even after the initial glucose was depleted [ Fig. 3 B, Suppl. Fig. 4 A and D ]. No other C 6 compound was detected by HPLC. These data indicate that mevalonate was being produced directly from lignin-derived carbon sources over an extended timeframe. Active synthesis of mevalonate after early stationary phase and improved mevalonate synthesis in the wax ester knockout strain supports the hypothesis that the wax ester pathway competes with the mevalonate pathway for acetyl-CoA flux under carbon-rich, nutrient depleted conditions ( Alvarez and Steinbüchel, 2002 ; Salmela et al., 2019 ). Amplification of the mevalonate pathway in the 168-h cell pellets revealed truncation of the mevalonate pathway for four of twelve POB-fed cultures and for six of eight glucose-fed cultures [ Suppl. Table 4 ; Suppl. Fig. 5 ]. This is consistent with the large spread in mevalonate titers late in the POB culture [ Fig. 3 B]. Glucose-fed cultures exhibited significant accumulation of glucose-derived gluconate that contributed to decreased pH [ Suppl. Table 5 ] and may have led to mutations to the mevalonate pathway, both of which may have reduced or eliminated mevalonate production [ Suppl. Fig. 4 B and E ; Suppl. Fig. 5 ]. Therefore, we postulate that mutations are likely driven by poor cell fitness due to non-optimal pH or high substrate accumulation [ Suppl. Fig. 4 B and E ; Suppl. Fig. 5 ]. Taken together, these data indicate that mevalonate production was significantly improved by eliminating the wax ester pathway and that implementing fed-batch cultivation led to continued production of mevalonate from POB after glucose was depleted, although long-term stability remains a challenge."
} | 4,021 |
25585693 | PMC4293592 | pmc | 696 | {
"abstract": "The phenomenon of resistive switching (RS), which was initially linked to non-volatile resistive memory applications, has recently also been associated with the concept of memristors, whose adjustable multilevel resistance characteristics open up unforeseen perspectives in cognitive computing. Herein, we demonstrate that the resistance states of Li x CoO 2 thin film-based metal-insulator-metal (MIM) solid-state cells can be tuned by sequential programming voltage pulses, and that these resistance states are dramatically dependent on the pulses input rate, hence emulating biological synapse plasticity. In addition, we identify the underlying electrochemical processes of RS in our MIM cells, which also reveal a nanobattery-like behavior, leading to the generation of electrical signals that bring an unprecedented new dimension to the connection between memristors and neuromorphic systems. Therefore, these Li x CoO 2 -based MIM devices allow for a combination of possibilities, offering new perspectives of usage in nanoelectronics and bio-inspired neuromorphic circuits.",
"discussion": "Discussion We have observed memristive behaviors in Li x CoO 2 films, combining the possibilities of multilevel information stacking and biological synapse plasticity emulation. In addition, we have identified the underlying electrochemical reactions inducing RS. Specifically, in the MIM cells, the cobalt redox reaction, which is connected to an insulator-to-metal transition, is coupled to the lithium intercalation/deintercalation in the conducting silicon electrode, whereas in the (CP) AFM probe-film case, it is coupled to the water redox reaction. Furthermore, we have observed a progressive occurrence of an EMF –more specific of a neuron- in MIM devices, which follows the switching from R OFF to R ON state, and disappears before the metal-to-insulator reverse transition. This latter transition is not spontaneous, but is induced by applying either voltage or heating. This specific behavior, observed for the first time in Li x CoO 2 thin film-based MIM solid-state cells, brings together a new combination of possibilities with a potential easy integration on Si-based devices, hence offering exciting new perspectives in nanoelectronics and neuromorphic implementation."
} | 568 |
36711650 | PMC9882120 | pmc | 698 | {
"abstract": "In recent years, the concern from the global climate change has driven an urgent need to develop clean energy technologies that do not involve combustion process that emit carbon into the atmosphere. A promising concept is microbial fuel cells that utilize bacteria as electron donors in a bio-electrochemical cell performing a direct electron transfer via conductive protein complexes or by secretion of redox active metabolites such as quinone or phenazine derivatives. In the case of photosynthetic bacteria (cyanobacteria) electrons can also be extracted from the photosynthetic pathway mediated mostly by NADH and NADPH. In this work, we show for the first time that the intact non-photosynthetic bacteria Escherichia coli can produce photocurrent that is enhanced upon addition of an exogenous electron mediator. Furthermore, we apply 2D-fluorescence measurement to show that NADH is released from the bacterial cells, which may apply as a native electron mediator in microbial fuel cells.",
"conclusion": "Conclusions In this work, we showed that non-photosynthetic bacteria can produce photocurrent. We applied 2D-fluorescence spectra of the external solution and cyclic voltammetry of the bacterial cells to show for the first time that some of the electrical current generation originate from NADH. These discoveries may help to improve the current production in MFCs.",
"introduction": "Introduction In recent years, energy innovations are directed toward the development of clean technologies that do not involve combustion processes and in this way can reduce carbon emission. A promising energy solution that is extensively developed is microbial fuel cells (MFCs) 1 . The main approach of MFCs is to exploit the ability of bacterial cells to donate electrons at the anode 2 , 3 of a bio- electrochemical cell. Alternatively, several species can also apply as electron acceptors at the cathode 2 , 4 , 5 . The bacterial cells perform external electron transport by 2 kinds of mechanisms: direct and mediated electron transfer. Direct electron transfer (DET) is conducted by metal respiratory (MTR) complexes in the cellular membrane 6 - 9 Mediated electron transfer is performed by native secretion of redox active molecules. Among these molecules are various derivatives of quinones and phenazines 10 - 15 . The electrical current production can be enhanced by addition of artificial exogenous electron mediators such as cystine, neutral red, thionin, sulphides, ferric chelated complexes, quinones, phenazines, and humic acids 16 - 21 . Among the most efficient bacterial species in MFCs are Shewanella oneidensis and Geobacter sulfurreducens which consist of a relatively high amount of pili and cytochrome types that are capable of charge transfer 6 - 9 , 22 , 23 . MFCs are not limited to bacteria and can also utilize yeasts 24 . A unique class of MFCs is the bio-photoelectrochemical cells (BPECs) 25 - 30 . This approach utilize the ability of photosynthetic organisms to perform external electron transfer, while the source of the electrons originate from both the respiratory and photosynthetic pathways 31 . The major electron mediators in BPECs are NADH and NADPH that can cycle electrons from photosystem I inside the cells and the external anode of the BPEC 25 . Enhancement of the photocurrent can be achieved by the exogenously adding natural electron mediators such as NADH, NADPH and vitamin b1 or the non-natural mediator potassium ferricyanide 30 . In this work, we show for the first time that it is possible to produce photocurrent from non-photosynthetic bacteria in a BPEC while the electron transfer is being mediated by NADH.",
"discussion": "Results and Discussion E.coli releases NADH and FAD to the external cellular medium Recent works have reported the release of the redox active molecules Nicotinamide adenine dinucleotide (NADH) and Nicotinamide adenine dinucleotide phosphate (NADPH) from various organisms such as cyanobacteria 25 , 30 , microalgae 26 , seaweeds 32 , plant’s leaves 29 , 34 , 36 , roots 38 and sea anemones 39 . These molecules can apply as electron mediators in bio-electrochemical cells catalysing electrons transport between the respiration and photosynthetic pathways of the cells and the external anode. The biological reason for the release of NADH and NADPH is not fully understood. It was suggested that it may derive from a minor leak of cytoplasmatic content, be involved in quorum sensing or apply to reduce iron to internalise it 25 . Mediated electron transport in non-photosynthetic bacteria in microbial fuel cells (MFCs) was previously explained by secretion of quinone and phenazines derivatives 16 - 21 . Nevertheless, the possibility of a release of NADH and NADPH from non-photosynthetic bacteria was not studied yet. To investigate this, we wished to analyse the external cellular solution of E.coli cells by 2D – fluorescence measurements. E.coli cells were were centrifuged (20 min, 6000 rpm) and resuspended in phosphate buffer saline (PBS) After 1 h of incubation, the cells were filtered by 0.22μm filter. 2D-fluorescence of the filtrate was measured (λ ex = 260 – 500 nm, λ em = 280 – 580 nm) ( Fig. 1 ). The fluorescence maps showed peaks with maxima around (λ ex = 280 nm, λ ex = 310 nm), (λ ex = 280 nm, λ ex = 350 nm), (λ ex = 300 nm, λ ex = 450 nm), (λ ex = 360 nm, λ ex = 450 nm), and (λ ex = 440 nm, λ ex = 520 nm) that are correlated with the spectral fingerprints of tyrosine, tryptophane, NAD + or NADP + , NADH or NADPH, and FAD 25 , 33 , 35 . As E.coli not a photoautotroph and cannot produce NADPH in a photosynthetic pathway like cyanobacteria, we suggest that the peaks with maxima at (λ ex = 300 nm, λ ex = 450 nm) and (λ ex = 360 nm, λ ex = 450 nm) originate from the fluorescence of NAD + and NADH respectively. Based on the obtained results, we suggest that the identified redox active species NADH and FAD may apply as native electron mediators in MFCs in addition to quinones and phenazines. Light irradiation of E.coli enhances the release of electron donors. Previous works have reported the ability of photosynthetic bacteria (cyanobacteria) to produce photocurrent in bio-photo electrochemical cells. The major source of electrons in such systems is the photosynthetic pathway to produce NADPH molecules. A small portion of NADPH is also released from the cyanobacterial cells into the external solution. This release is enhanced upon association of the cells with the anode of an electrochemical cell. Another molecule that significantly contributes to the current production in cyanobacteria is NADH that is formed in the tricarboxylic acid cycle (TCA cycle). Recent studies have shown that light irradiation on E.coli cells can enhance the reduction of NAD + to NADH in the TCA cycle 40 . Based on this and the identification of NADH in the external solution ( Fig. 1 ), we sought irradiation of E.coli that may enhance the current production by E.coli when applied as electron donor in MFCs. To explore this, we have designed a bio-electrochemical system that consist of screen-printed electrodes with a graphite working electrode (WE), graphite counter electrode (CE), and a Silver coated with silver chloride reference electrode (RE). A drop of 100 μL of E.coli suspension (10 6 CFU / mL) was placed on the screen-printed electrode. Light irradiation was done from top using a white LED. ( Fig. 2a , b )The light intensity at the height of the electrode surface was 1 Sun (1000W / m 2 ). Cyclic voltammetry of was measured in dark and light at potential varied from 0 to 1 V, with a scan rate of 0.1 V/s. The voltammogram showed 2 peaks at maximal potentials of 0.6 and 0.9 V that were enhanced under illumination ( Fig. 2c ). The bigger peak around 0.9 V correlates with the voltammogramic fingerprint of NADH 41 , strengthening our hypothesis that this molecules plays a key role in the electron mediation between the bacterial cells and the anode. E.coli produces photocurrent in a bio-electrochemical cell. Next, we wished to explore whether E.coli can produce photocurrent in a bio-electrochemical cell. Chronoamperometry of E.coli was measured with dark/light irradiation intervals of 100 s. A potential bias of 0.9 V was applied to the WE ( Fig. 3 ). This potential was chosen based on the bigger peak that was obtained in the cyclic voltammetry measurements ( Fig. 2 ). A current of about 0.07 μA /cm 2 was obtained in dark. Upon irradiation, the current was enhanced by ~ 0.05 μA /cm 2 . PBS applied as a control, showing no significant dark current and a photocurrent of ~ 0.005 μA /cm 2 that was obtained because of a direct light absorption in the WE. We postulated that this effect was significantly smaller for the bacterial suspension as their turbidity do not allow the same transparency as the clear PBS solution. Previous studies about photocurrent production from various intact cyanobacterial species, showed a photocurrent of about 0.3 μA /cm 2 (without being normalised to their chlorophyll content), while about 25 % of the current (~ 0.075 μA /cm 2 ) was reported to originate from NADH. A similar photocurrent production was obtained for E.coli ( Fig. 3 ). Based on this, we suggest that in cyanobacterial based BPECs part of the photocurrent does not originate from photosynthesis but from and enhanced formation of NADH catalysed by light irradiation. The electron transfer mechanism in photo microbial fuel cells. Based on the results obtained in this study, and previous studies we suggest mechanisms for the external transfer between the bacterial cells and the anode. In dark and upon association of the bacteria with the active electrochemical cells. NADH molecules are released from the cells donating electrons at the anode to produce electrical current. The oxidized form NAD + is internalized into the cells cytoplasm were it may be re-reduced to NADH in the Glycolysis pathway. Upon irradiation, the reduction of NAD + to NADH is being enhanced, increasing the number of molecules in the NADH cytoplasmatic pools, and the release of these molecules from the bacterial cells to produce electrical current. FADH2 may also be released from the cells to be oxidized at the anode into FAD, that can be reduced again by MTR complexes. These MTR complexes can also conduct a DET to donate electrons at the anode. DET may also be performed by conductive pili . Some of the non-light dependant electrical current may also derive from secretion of quinone and Phenazine molecules."
} | 2,634 |
36309573 | PMC9617901 | pmc | 699 | {
"abstract": "Redox-based memristive devices have shown great potential for application in neuromorphic computing systems. However, the demands on the device characteristics depend on the implemented computational scheme and unifying the desired properties in one stable device is still challenging. Understanding how and to what extend the device characteristics can be tuned and stabilized is crucial for developing application specific designs. Here, we present memristive devices with a functional trilayer of HfO x /Al 2 O 3 /TiO 2 tailored by the stoichiometry of HfO x ( x = 1.8, 2) and the operating conditions. The device properties are experimentally analyzed, and a physics-based device model is developed to provide a microscopic interpretation and explain the role of the Al 2 O 3 layer for a stable performance. Our results demonstrate that the resistive switching mechanism can be tuned from area type to filament type in the same device, which is well explained by the model: the Al 2 O 3 layer stabilizes the area-type switching mechanism by controlling the formation of oxygen vacancies at the Al 2 O 3 /HfO x interface with an estimated formation energy of ≈ 1.65 ± 0.05 eV. Such stabilized area-type devices combine multi-level analog switching, linear resistance change, and long retention times (≈ 10 7 –10 8 s) without external current compliance and initial electroforming cycles. This combination is a significant improvement compared to previous bilayer devices and makes the devices potentially interesting for future integration into memristive circuits for neuromorphic applications.",
"introduction": "Introduction Memristive devices are typically two-terminal resistive elements with a hysteretic current–voltage characteristic, which enables switching between different resistance states 1 – 3 . As analog resistive switching memory, memristive devices have emerged as one of the critical technologies for implementing neuromorphic computing systems 4 , 5 , i.e., systems that mimic the structure or working mechanisms of the brain 6 . For such computing systems, memristive devices offer a variety of potential advantages compared to traditional memory devices such as flash or SRAM 4 . The simple two-terminal device structure of memristive elements permits a high integration density in crossbar arrays 7 , 8 , and thereby, highly energy-efficient and parallelized in-memory computing. Additionally, the possibility of in-memory computing results in a low latency because it is not restricted by the data exchange between the memory and a central processing unit (von Neumann bottleneck) 4 , 5 . A major challenge for memristive devices in neuromorphic systems is a controllable and reproducible setting of resistance states for storing synaptic weights in artificial neuronal networks (ANNs). This requires high endurance and reproducibility of the device properties over many switching cycles 4 , 9 . In general, the requirements of the device parameters depend on the kind of weight update of the respective neuromorphic system. For example, the spike-timing dependent plasticity (STDP) rule is mainly relevant for spiking neural networks (SNN) and requires a time dependent and nonlinear resistance change 10 . In contrast, classical ANNs (such as deep neural networks or convolutional neural networks) do not require time dependency, but instead a linear resistance change and as many stable resistance values as possible 11 . Although the requirements imposed on the memory duration are lower in many areas of neuromorphic computing than in nonvolatile memories 12 , 13 , too short state retention times preclude the devices from broad application in neuromorphic computing. The properties of memristive devices depend on the underlying mechanisms of the resistive switching process, which can be broadly classified into filament-type switching and area-type switching 14 , 15 . Filamentary switching is characterized by the presence of (at least) one conductive filament, which provides an electrical pathway through the memristive material in the low resistive state. It typically must be induced with an initial forming procedure before operating the device 3 , 5 . The growth and disruption of the filament are stochastic processes and cause an abrupt change in the resistance between a high resistive state (HRS) and a low resistive state (LRS), usually without accessible intermediate resistance states (digital switching). Due to the high nonlinearity of the switching kinetics, additional electronics is typically required to stabilize the current–voltage characteristic by limiting the current through the device (current compliance). Limiting the current prevents uncontrolled filament formation and growth, leading to irreversible changes in the device characteristics and rendering the device useless 16 . From an application point of view, the highly nonlinear switching kinetics have enabled filamentary devices capable of long retention times > 10 years 17 , 18 , high switching speeds < 1 ns 19 , 20 , high endurance > \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${10}^{12}$$\\end{document} 10 12 cycles 7 , 21 , and large on–off ratios (> 1000) 5 , 22 . These properties have made them attractive as a potential technology for nonvolatile memory approaches 23 . Major challenges are the low reproducibility of device parameters and endurance deterioration because of the underlying stochastic process 22 , but also the required initial forming step and additional electronics for the current compliance, which can be disadvantageous for the integration of the devices into large arrays 4 . On the other hand, area-type switching is a less mature technology 14 . Depending on the specific device, various potentially involved microscopic processes are under debate 24 , 25 . Generally, the current through the device is assumed to be dominated by interface-limited conduction processes, such as the tunneling or emission through/over interface potential barriers. Switching has been explained as a modulation of the barrier via the drift and accumulation of mobile charge carriers under the application of an external voltage. In this context, electromigration of mobile vacancies has been discussed 26 – 30 , as also the trapping dynamics of electrons 31 , 32 and the uniform modification of bulk conduction mechanisms 14 . Because of the spatially uniform contribution of conduction mechanisms to the total current, the conductance scales with the device area and the stochastic components of the microscopic processes average and yield more reproducible macroscopic device properties than filamentary devices. Area-type devices can be fabricated with a low variation in the device characteristics and a correspondingly high yield 33 . Additional advantages are many intermediate and continuously accessible resistance states (analog switching) and often the possibility of operating without initial forming cycles or external current compliance 34 – 36 . However, the main challenges of area-type devices are a typically short retention time (seconds to days) and large switching times (> 1 ms) compared to filamentary devices 12 , 37 . A combination of the beneficial properties of both switching types is generally desirable. Among various types of memristive devices investigated for neuromorphic computing 1 , 5 , 9 , 38 – 40 , memristive bilayer and multilayer oxide structures based on hafnium oxide and titanium oxide gained attention for the application as artificial synapses 35 , 41 – 43 . Such oxide structures can show filamentary- or area-type switching 23 , 42 , depending on the details of the respective layer system. Compared to hafnium oxide single-layer devices, HfO x /TiO x bilayer structures demonstrated a variety of improvements 43 , such as a smaller variability 44 and improved resistance modulation linearity 45 , and a larger number of resistance states 46 . These bilayer structures are mainly of the filamentary type and often still require current compliance and an initial electroforming step. Previous studies 47 , 48 suggest that the filament formation in such systems is connected with the concentration of oxygen vacancies in the HfO x layer and oxygen ions in the TiO x layer. According to this work, the ions and vacancies are injected preferably at the HfO x /TiO x interface by a voltage-induced creation of Frenkel pairs, owing to much higher activation energy in the bulk material 47 , 48 . Consequently, altering the layer structure and interfaces could represent a way of modifying the switching type and further improve the device characteristics for the applications pursued. This work presents memristive trilayer devices with a functional layer stack of HfO x /Al 2 O 3 /TiO 2 sandwiched between TiN and Au electrodes. We discuss the influence of the additional aluminum oxide layer, operating conditions, and the composition of the hafnium oxide layer on the resistive switching characteristics. A model is devised to provide a microscopic interpretation of the experimental results and derive further implications for the design of memristive multilayer elements. The paper is structured as follows: after presenting the details of the device materials and geometries in “ Memristive devices and materials ”, we analyze the current–voltage characteristics under different voltage operating conditions and for two different hafnium oxide compositions in the section “ Current–voltage characteristics ”. In “ Discussion of the switching mechanisms ”, the indications found for the underlying switching mechanisms are interpreted before they are quantified with a physics-based device model presented in “ Physics-based compact model ”. Afterwards, we discuss application relevant device characteristics in “ Characterization for neuromorphic applications ”, namely the state retention time and the linearity in the conductance update. Finally, the results are summarized, and a conclusion is drawn.",
"discussion": "Discussion of the switching mechanisms The current–voltage characteristics and the observed change from an area to a filament-type switching behavior are explained in this section with a qualitative model considering the presence and dynamics of charged point defects. Based on the assumptions and measurements, we then estimate the formation energy of these defects. Following ab initio calculations, double positively charged oxygen vacancies \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{2+}$$\\end{document} V O 2 + and negatively charged interstitials \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${O}_{\\mathrm{i}}^{2-}$$\\end{document} O i 2 - are mobile in HfO 2 with activation energies of diffusion of 0.69 eV ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{2+}$$\\end{document} V O 2 + ) 54 , 55 and 0.6 eV ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${O}_{\\mathrm{i}}^{-2}$$\\end{document} O i - 2 ) 56 . However, their formation via the Frenkel mechanism \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${O}_{\\mathrm{O}}^{0}\\to {V}_{\\mathrm{O}}^{2+}+{O}_{\\mathrm{i}}^{2-}$$\\end{document} O O 0 → V O 2 + + O i 2 - in the bulk of HfO 2 can be omitted at room temperature owing to large activation energies of 5–9 eV 57 , 58 , also in the presence of high electric fields 58 . Instead, oxygen vacancies could be created at the HfO x /Al 2 O 3 interface by incorporating oxygen into the aluminum oxide layer, functioning as an oxygen reservoir/scavenger, which increases the formation energy of oxygen interstitials and reduces the formation energy of oxygen vacancies 59 . This mechanism was shown to be energetically favored over the formation of Frenkel pairs in the bulk of hafnium/hafnium oxide layers 60 , and could similarly apply here. Oxygen vacancies are also expected to be introduced during the sputtering process with densities depending on the stoichiometries of the deposited films. Therefore, the switching processes in our devices are likely dominated by the dynamics and formation of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{2+}$$\\end{document} V O 2 + as illustrated in Fig. 4 , while oxygen interstitials can be omitted. Figure 4 Illustration of the role of defect dynamics for the two switching types and the formation of oxygen vacancies \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{2+}$$\\end{document} V O 2 + at the interface. ( a ) Filamentary switching is based on the formation, growth, and disruption of an oxygen vacancy filament in defect rich HfO x . ( b ) Area-type switching in highly stoichiometric hafnium oxide, where the drift and diffusion of mobile vacancies change the potential landscape. ( c ) Density [ \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{2+}$$\\end{document} V O 2 + ] of double positively charged oxygen vacancies \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{2+}$$\\end{document} V O 2 + in the HfO 2 layer, calculated as a function of the applied voltage U and plotted together with the stoichiometric vacancy density \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\left[{V}_{\\mathrm{O},\\mathrm{s}}^{2+}\\right]$$\\end{document} V O , s 2 + (maximum indicated by black, dashed line) in the HfO 2 layer, which is still consistent with the XPS measurements. The electric field-induced vacancy density \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$[{V}_{\\mathrm{O},\\mathrm{i}}^{2+}]$$\\end{document} [ V O , i 2 + ] is plotted for three different formation energies E v . The stoichiometry of the HfO 1.8 layer is close to the computationally estimated ideal stoichiometry (HfO 1.5 –HfO 1.75 ) 61 for the nucleation of a filament of oxygen vacancies (Fig. 4 a). This process might be further fostered by the voltage-induced formation of vacancies during the large-range voltage sweep (Fig. 2 b). Following the results of ab initio calculations 55 , 62 , the double-positive charge state of oxygen vacancies is only energetically favored over single-charged vacancies if the vacancies are sufficiently apart from each other. As \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{2+}$$\\end{document} V O 2 + vacancies reach a filamentary cluster, they preferably trap charges to occupy a neutral or single-charged state. The single-charged vacancies \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{+}$$\\end{document} V O + are immobile in the bulk with activation energies of diffusion 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}$$\\approx 2\\; \\mathrm{eV}$$\\end{document} ≈ 2 eV 55 but comparatively mobile within and alongside the filament with much smaller activation energies of 1.05 eV and 0.8 eV, respectively 55 . Vacancies in these charge states can return to the mobile double-charged state (cohesion-isolation transition) by releasing electrons. The transition energy barrier (≈ 1 eV) corresponds to electric field magnitudes, which can be achieved at small applied voltages of a few volts 55 , 62 . Further, the large current densities expected in the filament cause significant Joule heating and a corresponding increase in the defect mobilities 63 . Via these mechanisms, the filament can grow, and an applied voltage can modulate the electrode-filament gap and the interface potential. From our measurements, the filamentary switching process in the HfO 1.8 device is also connected with a reversed direction of the hysteresis loop. This reversal could indicate a local change from n-type to p-type conduction 23 because a reduced interface barrier for electrons corresponds to an increased barrier for holes. Consistent with experimental results, a change from n-type to p-type conduction can be induced in HfO x by increasing the density of oxygen vacancies 64 . Area-type resistive switching is assumed to occur at small vacancy densities, where filament formation is not favored, by the homogeneous drift and formation of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O}}^{2+}$$\\end{document} V O 2 + as illustrated in Fig. 4 b. Within this model, the charged point defects alter the interface potential as they drift in the applied electric field, which changes the current through the device. The transition of the measured I-U characteristics (Fig. 2 ) to a larger hysteresis with a smaller overall resistance can be explained by the voltage-induced formation of additional vacancies at the HfO x /AlO x interface. The range of the required formation energy of this process can be estimated from the expected density of oxygen vacancies and the voltage range in which it is expected to increase significantly, i.e., the voltage range of the large-range voltage sweep in Fig. 2 a. We use a rate equation to describe the formation of electric field-induced vacancies, derived from the Butler-Volmer equation 47 . Further, considering the uncertainty of ± 1% in the stoichiometry (“ Memristive devices and materials ” section) and the density of oxygen sites in ideal monoclinic HfO 2 , we estimate that the density \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$[{V}_{\\mathrm{O},\\mathrm{s}}^{2+}]$$\\end{document} [ V O , s 2 + ] of stoichiometric vacancies \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{\\mathrm{O},\\mathrm{s}}^{2+}$$\\end{document} V O , s 2 + in our HfO 2 layers must be \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$[{V}_{\\mathrm{O},\\mathrm{s}}^{2+}]\\le 3.2\\cdot {10}^{23} \\;{\\mathrm{m}}^{-3}$$\\end{document} [ V O , s 2 + ] ≤ 3.2 · 10 23 m - 3 to be consistent with the XPS measurements. Example results of the electric field-induced vacancy density \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$[{V}_{\\mathrm{O},\\mathrm{i}}^{2+}]$$\\end{document} [ V O , i 2 + ] as a function of the applied voltage \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$U$$\\end{document} U are plotted in Fig. 4 c. Results are shown for three different formation energies \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${E}_{\\mathrm{V}}$$\\end{document} E V together with the maximum of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$[{V}_{\\mathrm{O},\\mathrm{s}}^{2+}]$$\\end{document} [ V O , s 2 + ] indicated by a black dashed line. From the current–voltage measurements in Fig. 2 a, a significant increase in the vacancy density is expected for applied voltages between 3 and 5 V, which corresponds to a formation energy of approximately \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${E}_{\\mathrm{V}}>1.4 \\;\\mathrm{eV}$$\\end{document} E V > 1.4 eV in Fig. 4 c if we assume that the maximum 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}$$\\left[{V}_{\\mathrm{O},\\mathrm{s}}^{2+}\\right]$$\\end{document} V O , s 2 + is present in the HfO 2 layer. For smaller formation energies \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$[{V}_{\\mathrm{O}}^{2+}]$$\\end{document} [ V O 2 + ] reaches unrealistically high values at \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$U = 5 \\mathrm{V}$$\\end{document} U = 5 V , e.g., \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\left[{V}_{\\mathrm{O}}^{2+}\\right]>{10}^{28}1/\\mathrm{m}^{2}$$\\end{document} V O 2 + > 10 28 1 / m 2 for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${E}_{\\mathrm{V}}\\approx 1.3 \\; \\mathrm{eV}$$\\end{document} E V ≈ 1.3 eV . While such high vacancy densities have been used for filamentary regions 65 , we consider it too high for area-type switching because it by far exceeds the density of oxygen lattice sites 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}$$\\approx 3.2\\cdot {10}^{25}\\;{\\mathrm{m}}^{-3}$$\\end{document} ≈ 3.2 · 10 25 m - 3 in ideal monoclinic HfO 2 . A more precise estimation 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}$${E}_{\\mathrm{V}}$$\\end{document} E V , including an upper limit, requires knowing the range of the vacancy density in the HfO 2 layer. For such an estimation, a model is presented in the next section.\n\nDiscussion and conclusions We presented trilayer HfO x /Al 2 O 3 /TiO 2 memristive devices with two hafnium oxide stoichiometries ( x = 1.8, x = 2) and analyzed their electrical characteristics for neuromorphic applications. We demonstrated the possibility of tuning the switching mechanism by altering the hafnium oxide stoichiometry and showed how the Al 2 O 3 interlayer could extend the range in which area-type switching can be tuned. The Al 2 O 3 interlayer leads to stable area-type switching at small voltages for both hafnium oxide stoichiometries. A change from an area-type switching mechanism to a filament-type is induced at a larger voltage amplitude in the HfO 1.8 device, with strong indications for a transition from an n-type semiconductor to a p-type semiconductor. The origin of this transition and the switching mechanisms is discussed in detail and used for setting up a physics-based model capable of reproducing distinct features of the measured current–voltage characteristics. The model is based on an equivalent-circuit formulation and can be easily integrated into large-scale simulations for the design of memristive circuits. Quantitatively consistent with our model, we estimated an activation energy 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}$$\\approx 1.65 \\; \\mathrm{eV}\\pm 0.05\\;\\mathrm{ eV}$$\\end{document} ≈ 1.65 eV ± 0.05 eV for Frenkel defects at the HfO x /Al 2 O 3 interface. This is larger than the values of HfO x /TiO x interfaces 47 and can explain the stable area-type switching behavior we observe in the trilayer devices. The stabilization leads to various additional advantages in performance compared to previously presented HfO x /TiO 2 bilayer devices 67 , such as the possibility of operating without current compliance and initial electroforming cycles in devices of both hafnium oxide stoichiometries. Further, the state retention time and linearity in the conductance update were characterized. Depending on the stoichiometry and voltage range applied, we observed retention times from several days to months in the area-type devices and several years for the filamentary device with on–off ratios of 10 2 –10 3 Consistent with the switching type, a binary conductance update is observed in the filamentary device, and an analog-type conductance update with many continuously accessible resistance states in the area-type devices. Compared to the other trilayer device configurations, the area-type HfO 1.8 devices show much-improved linearity in the conductance update while keeping a large on–off ratio and a long retention time. These properties could make the devices promising for classical artificial neural networks such as deep neural networks and convolutional neural networks. Area-type switching in devices of this stoichiometry, and thereby, such a combination of properties, was essentially made possible by the stabilizing aluminum oxide layer. The nonlinear, time dependent resistance change in the HfO 2 devices could be on the other hand suitable for implementing the STDP rule in spiking neural networks. In conclusion, we demonstrated a technological methodology for tuning multilayer redox-based memristive devices. The insights gained on stabilizing the area-type switching mechanism are an essential step towards tailored devices unifying the beneficial characteristics of filamentary and area-type devices. We achieved promising properties with trilayer memristive elements and provided a model, which can be integrated into large-scale circuit simulations to support the design of memristive circuits for neuromorphic learning schemes in the future."
} | 7,668 |
39965779 | PMC11898061 | pmc | 700 | {
"abstract": "This Perspective explores the use of biomacromolecules\nin natural\nmaterials synthesized by living organisms, such as spider silk, in\nthe development of sustainable synthetic materials. Currently employed\nsynthetic polymers lack the hierarchical complexity and unique properties\nof natural materials composed of biomacromolecules. By understanding\nthe composition of these natural materials, it may be able to reproduce\ntheir properties synthetically. Additionally, research directions\ninvolving the use of renewable resources such as nitrogen and carbon\ndioxide from the air and seawater to develop biomacromolecules such\nas spider silk and biopolyester via photosynthetic organisms are reviewed.\nNext-generation biomacromolecule research will aid in the creation\nof a sustainable global society, advancing fields such as biomanufacturing,\nagriculture, aquaculture, and other industries.",
"conclusion": "V Conclusions and Future Perspectives While silkworm silk has been extensively studied in the context\nof nylon development, no polymer that combines the toughness, lightness,\nand interfacial affinity of spider silk has been developed. In addition,\nalthough natural rubber is widely used and indispensable to modern\nsociety, elucidation of the mechanisms underlying its physical properties,\nsuch as strain-induced crystallization and artificial reproduction\nof its mechanical properties, has not been achieved. To the best of\nthe author’s knowledge, these scientific limitations may be\nbecause only the chemical structure has been elucidated and not the\nhierarchical or network structure. In the future, I would like to\ncontinue to investigate biomacromolecules, including their hierarchical\nstructure, and clarify the relationship between structure and function\nto provide novel insights for the development of artificial biological\nmaterials. Using spider silk as an example in this Perspective,\nit is important\nto understand biological materials in nature in terms of their biomacromolecules\nand other constituent units and to elucidate the relationships between\ntheir properties and molecular composition, which leads to sustainable\nmaterial innovation. The creation of polymeric materials with both\noptimized physical properties and environmental degradability will\nfundamentally solve the problems caused by polymeric materials, such\nas microplastics in the ocean and carbon dioxide emissions from petroleum\nraw materials and processes, and contribute to the construction of\na sustainable and circular society. On a slightly larger scale, it\nis necessary to move away from the current biomimetic research and\nestablish an academic domain where biological materials can be created\nand used in a practical manner and establish a process where materials\nused by humans do not put the global environment and the natural environment\nat risk and do not compete with food and energy production. I am convinced\nthat such an academic field is essential for a sustainable global\nsociety and can be important for low-resource countries and regions,\nincluding Japan. Although not mentioned in this Perspective, efforts\nare underway in the fish industry to develop zero-carbon fishery feed\nby generating optimized proteins via photosynthetic bacteria and in\nthe agricultural sector to develop zero-carbon fertilizers such as\namino acid-based nitrogen fertilizer. 113 I hope to contribute to the development of a sustainable global\nsociety not only through chemical research but also through biology\nstudies and through tight linkages with diverse fields such as next-generation\naquaculture and agriculture.",
"introduction": "I Introduction The various physical and\nbiological properties and functions of\nnaturally occurring polymers and composite materials have not yet\nbeen able to be replicated in synthetic polymers and artificial materials. 1 − 3 Many of the unique properties of these natural materials depend\non the hierarchical structure, including the primary structure (chemical\nstructure), of the biopolymers produced by a particular species. If\nwe can elucidate these hierarchical structures and clarify the mechanisms\nthat contribute to the physical properties of natural materials, it\nwill be possible for humans to artificially mimic the unique physical\nproperties of natural polymers and biopolymers. While we have learned\nfrom silkworm silk and developed nylon, no polymer that combines the\ntoughness, lightness, and interfacial affinity of spider silk has\nbeen developed. Thus, it is necessary to move away from the current\nbiomimetic science research and establish an academic field where\nbiological materials can be artificially created and used in a practical\nmanner and to establish a process where materials used by humans do\nnot put the global environment and the natural environment at risk\nand do not compete with food and energy production. In nature,\nthere are many attractive biomacromolecules, such as\nprotein, 4 polyhydroxyalkanoate, 5 , 6 natural rubber, 7 , 8 cellulose, 9 − 12 lignin, 13 , 14 chitin, 15 − 17 and nucleic acid. 18 − 21 Among them, I have been continuously\nstudying biomacromolecules synthesized by living organisms, such as\nproteins, biopolyesters, and natural rubber. 1 , 3 , 22 − 25 In recent years, I have attempted\nto develop new processes for the artificial production of biomacromolecules\nsuch as cellulose, polyhydroxyalkanoate (PHA), and those found in\nspider silk and natural rubber by using inexhaustible resources (nitrogen\nand carbon dioxide) from air and seawater and photosynthetic organisms\n( Figure 1 ). 26 , 27 If this series of studies reveals promising results, then it will\nbe possible to prevent resource depletion and reduce greenhouse gas\nemissions while simultaneously establishing a material cycle that\nis symbiotic with nature. While building a production process based\non the fixation of nitrogen and carbon dioxide via photosynthetic\norganisms, we aim to design molecules and materials that are stable\nduring use and return to the natural environment after use. This process\nof recycling materials is important for low-resource countries and\nregions. Moreover, this process does not place a burden on the global\nenvironment by utilizing carbon and nitrogen fixation via photosynthetic\norganisms. I suggest that the creation of next-generation biomanufacturing\nindustries, as well as chemical industries, will lead to the realization\nof a harmonious global society through organic linkages with next-generation\nagriculture, animal husbandry, fisheries, and forestry industries.\nIn\nthis Perspective, structural proteins are introduced as representative\nexamples of biopolymers, and the decomposition of carbon dioxide and\nbiomacromolecule biosynthesis from carbon dioxide are outlined. Figure 1 Overview of\nthe design and sustainable biosynthesis of spider silk-based\nmaterials. Using carbon dioxide and nitrogen, artificial silk was\nsynthesized from marine red photosynthetic bacteria (left figure). 28 − 30 The spinning mechanism of spiders has been elucidated at the molecular\nlevel, and a hierarchical structure similar to spider silk was successfully\nreproduced for the first time in the world (lower right figure). 31 − 35 A database of spider silk proteins was constructed, and the use\nof material informatics enabled the development and commercialization\nof artificial silk with excellent water resistance (upper right figure). 36 , 37 Copyright 2020 AAAS."
} | 1,852 |
28553363 | PMC5444522 | pmc | 701 | {
"abstract": "Engineering surfaces that promote rapid drop detachment 1 , 2 is of importance to a wide range of applications including anti-icing 3 – 5 , dropwise condensation 6 , and self-cleaning 7 – 9 . Here we show how superhydrophobic surfaces patterned with lattices of submillimetre-scale posts decorated with nano-textures can generate a counter-intuitive bouncing regime: drops spread on impact and then leave the surface in a flattened, pancake shape without retracting. This allows for a four-fold reduction in contact time compared to conventional complete rebound 1 , 10 – 13 . We demonstrate that the pancake bouncing results from the rectification of capillary energy stored in the penetrated liquid into upward motion adequate to lift the drop. Moreover, the timescales for lateral drop spreading over the surface and for vertical motion must be comparable. In particular, by designing surfaces with tapered micro/nanotextures which behave as harmonic springs, the timescales become independent of the impact velocity, allowing the occurrence of pancake bouncing and rapid drop detachment over a wide range of impact velocities."
} | 282 |
19494577 | null | s2 | 702 | {
"abstract": "Cell-cell communication in bacteria, called quorum sensing, relies on production, release, and detection of signaling molecules, termed autoinducers. Communication enables populations of cells to synchronize gene expression and therefore behave as a group in a manner akin to cells in multicellular organisms. Most quorum-sensing systems allow communication within an individual species of bacteria. However, one autoinducer, called AI-2, is produced and recognized by many different bacterial species, indicating that some bacteria communicate across species boundaries. Current studies are aimed at discovering the role that AI-2 plays in gene regulation. Differential gene expression in response to AI-2 may cause bacterial behavioral changes, such as biofilm formation or transition to a pathogenic state. Interestingly, multiple mechanisms to detect AI-2 exist. These differences likely reflect variations in the role that AI-2 plays for different bacteria. Additionally, structural analyses of the AI-2 receptor in V. harveyi have provided insight into bacterial trans-membrane signal transduction. A further understanding of bacterial quorum-sensing processes may facilitate development of new technologies aimed at interfering with bacterial communication and virulence."
} | 319 |
35665141 | PMC9158544 | pmc | 703 | {
"abstract": "Plant and root fungal interactions are among the most important belowground ecological interactions, however, the mechanisms underlying pairwise interactions and network patterns of rhizosphere fungi and host plants remain unknown. We tested whether neutral process or spatial constraints individually or jointly best explained quantitative plant–ectomycorrhizal fungal network assembly in a subtropical forest in southern China. Results showed that the observed plant–ectomycorrhizal fungal network had low connectivity, high interaction evenness, and an intermediate level of specialization, with nestedness and modularity both greater than random expectation. Incorporating information on the relative abundance and spatial overlap of plants and fungi well predicted network nestedness and connectance, but not necessarily explained other network metrics such as specificity. Spatial overlap better predicted pairwise species interactions of plants and ectomycorrhizal fungi than species abundance or a combination of species abundance and spatial overlap. There was a significant phylogenetic signal on species degree and interaction strength for ectomycorrhizal fungal but not for plant species. Our study suggests that neutral processes (species abundance matching) and niche/dispersal-related processes (implied by spatial overlap and phylogeny) jointly drive the shaping of a plant-ectomycorrhizal fungal network.",
"conclusion": "Conclusion Taken together, our study for the first time illustrates that spatial overlap, species abundance and phylogeny can be important in structuring a complex plant-EM fungal network. In particular, we found nestedness was accurately predicted by a combination of species abundance and spatial overlap. However, even the best predictor model used here could not accurately reproduce the observed values of some network metrics (notably network modularity and specificity). This implies that complex plant-fungal networks may be difficult to be synthesized sufficiently while only considering spatial distribution and species abundance. A weak phylogenetic constraint was detected in species interactions of plants and root fungi. This suggests phylogeny and functional traits have the potential to contribute to the species associations of plants and EM fungi. More studies will be required to incorporate a variety of root physical and chemical traits to reveal the underlying processes of plant and fungus symbiotic networks. Thus, improvements of these models would be very helpful to illustrate the plant-fungal network patterns in future. It is also important to acknowledge that species patterns and network estimates may be affected by limitations of DNA sequence clustering algorithms and root microbial communities may be associated with the taxonomic resolution of ITS sequences. Thus, it would be advisable to investigate microbial communities and construct high-resolution plant–root microbial networks using metagenomics technology in future.",
"introduction": "Introduction The plant–ectomycorrhizal (EM) fungal network is one of the most common mutualistic symbiotic networks known. Both plants and their EM fungal partners are obligate symbionts, as they cannot complete their life cycle without forming a symbiosis ( Taylor, 2008 ). The selection and adaptation of ectomycorrhizal fungi (among the most diverse mycorrhizal fungi) to their host plant roots are products of long-term ecological and evolutionary processes. This is expected to lead to specificity in the interaction between these symbiotic partners. Despite this, EM fungi are reported to have a relatively low degree of host specialization ( Roy-Bolduc et al., 2016 ). EM fungi provide crucial benefits for plants by enhancing soil nutrient acquisition, plant seedling establishment, and disease resistance ( van der Heijden et al., 2008 ; Liang et al., 2020 ; Cahanovitc and Angel, 2022 ) and can promote monodominance in forest communities ( McGuire, 2007 ). In turn, host plants support their EM fungi through provision of photosynthetically fixed carbon ( Högberg and Högberg, 2002 ). Thus, elucidating the symbiotic network patterns and factors that regulate EM fungi and host plants is important for understanding community assembly and ecosystem functioning. Identifying the factors that drive the organization of plants and EM fungi into complex networks remains poorly explored, largely due to historical limitations in sampling and taxonomic identification of root fungi. Increasingly, ecologists are constructing high taxonomic resolution networks of plant-fungal communities using high throughput sequencing technology ( Bahram et al., 2014 ; Chen et al., 2019 ; Wang et al., 2019 ), which have started to illustrate the network structures of EM fungi, arbuscular mycorrhizal fungi and their host plants ( Dickie et al., 2017 ; Põlme et al., 2018 ; Arraiano-Castilho et al., 2020 ). Two common network structures of interest include nestedness and modularity. Nestedness describes the tendency of specialized species to interact with a subset of the interaction partners of more generalized species ( Bascompte et al., 2003 ). Modularity describes how a complex network can be organized into distinct modules, where interactions are more common within modules than between modules ( Olesen et al., 2007 ). However, observed structures of plant-EM fungal networks vary widely, exhibiting either non-nested or anti-nested patterns ( Bahram et al., 2014 ; Arraiano-Castilho et al., 2020 ) and non-modular ( Bahram et al., 2014 ) or modular structures ( Dickie et al., 2017 ; Arraiano-Castilho et al., 2020 ). The reasons for such inconsistent network structural properties remain unclear, but it is notable that most prior work used to construct plant–EM fungal networks have typically used less than a dozen species ( Bahram et al., 2014 ; Dickie et al., 2017 ; Arraiano-Castilho et al., 2020 ). This low number of host plant species might have contributed to the failure to detect network structural properties such as nestedness or modularity. A comprehensive understanding of the community assembly rules that structure plant–EM fungal networks might then require networks that incorporate increased sampling effort for host plant species. Species interaction patterns and structural properties of plant–EM fungal networks are also affected by ecological and evolutionary factors. A neutral view of assembly considers interactions among individuals are essentially random and it is relative species abundances that determine network patterns ( Canard et al., 2012 ). However, deterministic factors such as the mismatch of environmental tolerances of plants and fungi ( Arraiano-Castilho et al., 2021 ) and human effects ( Wall et al., 2020 ) may regulate the establishment of interaction relationships by imposing spatial constraints on encounter probabilities. For instance, structures of aboveground networks are best predicted by a combination of species abundance and other deterministic factors such as spatial overlap ( Vázquez et al., 2009 ; Sáyago et al., 2013 ), phylogeny ( Cagnolo et al., 2011 ) and trait matching ( Olito and Fox, 2015 ). In belowground networks, spatial overlap of plant and fungus communities is also expected to regulate the structure of plant-root fungal networks. However, a lack of substantial structure in the co-occurrence network of plants and arbuscular mycorrhizal fungi ( Encinas-Viso et al., 2016 ) suggests that spatial information might have limited influence on plant-fungal network assembly. Evolutionary history must also be considered, for example, being associated with nonrandom patterns in woody plant–arbuscular mycorrhizal fungal networks ( Chen et al., 2017 ). In general, while species abundances, spatial proximity and phylogeny all contribute to interaction patterns of plant–EM fungal and network structures, their interaction and relative contributions under different environmental conditions remain poorly understood. To fill the gaps in understanding the processes that assemble EM fungi and host plants into complex networks, we analyzed a large interaction database constructed by next-generation sequencing involving 43 plants and 862 EM fungi in a 50 ha subtropical forest plot in southern China ( Wang et al., 2019 ). The database is notable for the high number of host plant species, which we hypothesized might afford a more detailed understanding of the rules determining network assembly. In analyzing the database, we posed the following three questions: What are the structural properties of the plant-EM fungal association network? How do species abundance and spatial proximity contribute to the pairwise interaction patterns and structural properties of the plant-EM fungal network? Are symbiotic patterns of plants and EM fungi conserved in plant and EM fungal phylogenies? We tested whether observed network structures were associated with neural processes, niche/dispersal related processes or the combined effects of both. We anticipated that neutral processes might dominate network structure because environmental and spatial patterns explain little of the variance in EM fungal diversity in the study site ( Wang et al., 2019 ). We also expected to find a strong phylogenetic signal in the network due to the dominant role of host phylogeny in explaining the diversity of EM fungi in the forest ( Wang et al., 2019 ).",
"discussion": "Discussion Network Structure The local plant–EM fungal network in this subtropical forest was characterized by higher species number and more interactions than comparable studies ( Jacquemyn et al., 2011 ; Montesinos-Navarro et al., 2012 ; Toju et al., 2014 ). Consistent with other mutualistic plant-fungal networks ( Jacquemyn et al., 2011 ; Montesinos-Navarro et al., 2012 ; Toju et al., 2014 ), our network had low connectivity ( Montesinos-Navarro et al., 2012 ). Compared with low levels of specialization ( Roy-Bolduc et al., 2016 ), more intermediate levels of specialization as observed here (H2′= 0.42) in the plant-EM fungal network may be advantageous for host trees by increasing their chance to form a symbiosis with suitable EM fungal partners. Nonrandom associations of plants with symbiotic partners are common in biotrophic fungi, which may enhance modular structure in our plant-EM fungal network ( Supplementary Figure S5 ). This is an important mechanism that leads to niche partitioning (reflecting ecological specialization; Jacquemyn et al., 2015 ). This implies that modularity and specialization are possible mechanisms allowing many EM fungal species to coexist locally on the same set of host plants. We found that nested structures in the plant-EM fungi network where host plants of specialized EM fungi were subsets of the host plants of generalized Cenococcum spp. and Russula spp. fungi ( Supplementary Figure S4 ). In contrast, anti-nested structures, in which co-occurring plant species share fewer fungal symbionts than expected by chance, have been reported in many smaller plant-EM fungal networks ( Bahram et al., 2014 ; Roy-Bolduc et al., 2016 ; Arraiano-Castilho et al., 2020 ). This suggests that increasing numbers of species might favor greater network nestedness, as suggested by Bascompte et al. (2003) . As the sampling effort invested in our study over a 50-ha forest extent exceeds many other studies in plant–root fungal association networks ( Jacquemyn et al., 2011 ; Montesinos-Navarro et al., 2012 ; Toju et al., 2014 ), we believe the sampling effort is likely to have weak if any, influence on our estimates of network metrics (see Supplementary Figure S6 ). Determinants of Pairwise Interactions in a Network Spatial co-occurrence patterns of plant and fungal species determined species interaction frequencies while species abundance or the interaction of the two had no support in our study ( Table 1 ). Some evidence suggests that tropical mycorrhizal populations show significant spatial heterogeneity and nonrandom associations with different hosts ( Husband et al., 2002 ). This clarifies how species interactions may be mainly determined by plant partner choice for EM fungi, which is also controlled by soil environment heterogeneity or spatial proximity ( Cavender-Bares et al., 2009 ; Wang et al., 2019 ). Moreover, locally unmeasured microsite effects may confound the influence of the spatial distribution on species interaction patterns. In summary, our findings show the role of niche/dispersal-related processes (as implied by spatial overlap) in organizing symbiotic patterns of plant-EM fungal communities. This in turn points to the importance of co-occurrence as regulated by the effects of belowground environmental filtering in generating the observed links between plants and EM fungi. Prediction of Network-Aggregated Statistics Though species abundance makes only a weak contribution to the species interaction frequency of plants and EM fungi, we found that species abundance matching and spatial constraints made the highest contribution to most network metrics (nestedness included, Figures 2A – C,F ). The mechanisms underlying the effect of species abundance on network structure (e.g., nestedness) could be explained by the regulation of right-skewed frequency distributions of relative species abundance on co-occurrence patterns ( Canard et al., 2014 ). Moreover, spatial autocorrelation over short distances (in a range of 25 m, see Supplementary Figure S7 ) and limited dispersal capacity ( Tedersoo et al., 2011 ; Bahram et al., 2013 ) may lead to aggregated distributions of EM fungi. In addition, the nonrandom distribution of host plants is also dependent to some extent on the occurrence of suitable mycorrhizal fungi ( Jacquemyn et al., 2013 ). This may explain how spatial overlap of plants and fungi can predict network assembly ( Jacquemyn et al., 2015 ). However, co-occurrences of plant-EM fungi in a microsite were only a prerequisite to establishing symbiotic relationships. For example, the most dominant EM species that co-occur at a given study site are associated with distinct plants with scattered distributions ( Tedersoo et al., 2010 ). This suggests that other biological constraints such as trait-matching of plant-fungal networks might contribute to the observed network structure (i.e., network specificity and modularity, Figures 2D , E ). For example, some modules were dominated by Scleroderma spp. fungi with long-distance exploration while other modules were dominated by Russula spp. fungi with contact hyphal exploration ( Supplementary Figure S5 and Zhu et al., unpublished 2022). Plant trait-based host selection behavior in EM fungi or sampling effects associated with species abundance and richness of the plant taxa may contribute to observed network specificity and modularity. Besides this, other deterministic factors such as phenology ( Donatti et al., 2011 ; Morente-López et al., 2018 ) and a combination of local adaptation and competition ( Valverde et al., 2020 ) can affect network modularity. Observed network specialization is poorly predicted by abundance, spatial overlap and plant traits ( Sáyago et al., 2013 ), consistent with our study ( Figure 2D ). Our results suggest that species abundance and spatial distribution can better predict the network structures that are regulated by both neutral processes (e.g., relative species abundance) and niche/dispersal related processes (as implied by spatial overlap), while failing in other network structural properties, including modularity, that may be mostly driven by trait-based niche processes ( Donatti et al., 2011 ; Morente-López et al., 2018 ). Phylogenetic Signal of Species Associations It has been shown that species interaction patterns of plants and root fungi are mainly constrained by the phylogeny of the host plant, including from the study plot ( Toju et al., 2016 ; Wang et al., 2019 ) and weakly constrained by fungal phylogeny ( Munoz et al., 2012 ). Similarly, we found that plant-fungal interactions were strongly associated with plant phylogeny (one-tailed Mantel test r = −0.086, p = 0.025) but weakly associated with fungal phylogeny (one-tailed Mantel test r = −0.023, p = 0.001). A lack of phylogenetic signal in species degree and interaction strength for fungal hosts ( Table 2 ) may be related to the high vulnerability of fungal hosts and only distant relatedness between the multiple EM fungal partners supported. In contrast, phylogenetic constraints are evident for EM fungi on network characteristics, such as network nestedness ( Chen et al., 2017 ) and compartmentalization ( Vacher et al., 2008 ). However, the phylogenetic conservatism for EM fungi in species degree and interaction strength was only supported by λ statistics but not K statistics ( Table 2 ). This suggests K statistics might underestimate the phylogenetic signal because it is sensitive to the number of missing plant taxa ( Jardim et al., 2021 ). Overall, a significant but weak phylogenetic signal on plant-EM fungal symbiotic associations suggests that the contribution of conservative traits on the network structure thus may be limited, while traits unconstrained by phylogenetic history may play important roles in regulating plant-fungal interaction patterns. Additionally, evolutionary history may influence network structure at other scales and analyzing a phylogenetically more diverse root fungal community or a larger network might lead to the consistent detection of a phylogenetic signal."
} | 4,387 |
39747070 | PMC11695594 | pmc | 704 | {
"abstract": "Biological neural circuits demonstrate exceptional adaptability to diverse tasks by dynamically adjusting neural connections to efficiently process information. However, current two-dimension materials-based neuromorphic hardware mainly focuses on specific devices to individually mimic artificial synapse or heterosynapse or soma and encoding the inner neural states to realize corresponding mock object function. Recent advancements suggest that integrating multiple two-dimension material devices to realize brain-like functions including the inter-mutual connecting assembly engineering has become a new research trend. In this work, we demonstrate a two-dimension MoS 2 -based reconfigurable analog hardware that emulate synaptic, heterosynaptic, and somatic functionalities. The inner-states and inter-connections of all modules co-encode versatile functions such as analog-to-digital/digital-to-analog conversion, and linear/nonlinear computations including integration, vector-matrix multiplication, convolution, to name a few. By assembling the functions to fit with different environment-interactive demanding tasks, this hardware experimentally achieves the reconstruction and image sharpening of medical images for diagnosis as well as circuit-level imitation of attention-switching and visual residual mechanisms for smart perception. This innovative hardware promotes the development of future general-purpose computing machines with high adaptability and flexibility to multiple tasks.",
"introduction": "Introduction General-purpose hardware is engineered to mirror human-like advanced problem-solving abilities and cognitive thinking. The fundamental challenge in developing general-purpose hardware lies in replicating the adaptability and flexibility of human brain intelligence 1 . The construction of general-purpose hardware can be decomposed into developing machine learning modules that simulate various units of the brain, such as neuron aggregates and then integrated based on a brain-like cognitive architecture, enabling the system to emulate the operational principles and structural intricacies of the human brain intelligence 2 . In particular, the highly reconfigurable neural network of brain neural circuits is one of the important sources of brain intelligence, giving the brain the exceptional capacity for flexible adaptation to dynamic environments. The brain neural circuits can maintain flexible reconfigurable synapse/heterosynapse/soma inter-connections while frequently adjusting neuronal states 3 . Such distinctive mode orchestrates the connections of neural circuits across various interactive environmental tasks to ensure an appropriate alignment between task complexity and computational resources 4 . For neural circuits, neural signals are transmitted to the receiving neuron via synapse, subsequently relayed through the control of heterosynaptic connection to various soma, where the integration and processing of signals are performed (Fig. 1a ) 5 – 7 . Therefore, neuromorphic computing hardware equipped with this capability needs to be developed to imitate the principles and structures of the brain neural circuits. Such hardware should be able to simultaneously implement precise engineering of individual devices’ inner states and inter-modules’ connections, thus boosting adaptability and plasticity and providing strategies for the implementation of adaptability and flexibility in general-purpose hardware 8 – 10 . Fig. 1 Concept of RAH and reconfigurable functionalities. a Schematic of the complex structures and functions of biological neural circuits formed by synapses, heterosynapses, soma, and their inter-connections. In the synaptic component, the release and reception of neurotransmitters accomplish signal transmission, and their dynamic modulation (concentration and action time) simulates the reconfigurable weights, reflecting the plasticity inherent in biological synapses. Heterosynapses are particularly notable for their ability to facilitate inter-neural circuit communication through state transitions and to transmit signals across various types of neural cells. The soma component underscores the integrative capacity of neural networks in processing complex signals. The multifunctionality of biological neural circuits is demonstrated in the capability of these circuits to perform a multitude of tasks, including perception and learning. b Schematic of the biology-inspired reconfigurable hardware (I) that encompasses three principal modules: the synaptic module (cascaded MoS 2 FET arrays), the heterosynapse module (MoS 2 FET-based OPA and MoS 2 FET connections), and the soma module (MoS 2 FET-based OPA and feedback component integration) (II). These modules employ adjustable states, such as conductance encoding, switchability, and reconfigurable feedback mechanisms, to mimic the components of biological neural circuits, such as synapses, heterosynapses, and soma. This configuration facilitates multitask processing, including signal conversion, artificial visual simulation, and neural network computing (III). c Demonstration of 8-bit DAC. Square-wave input signals ( V in1 – V in8 ) with an input amplitude of 0.1 V and an input frequency of f n = 2 n−1 × f 1 , where f 1 = 5 Hz is the input frequency of signal V in1 (I). Conductance encoding in the synapse module following j 1 ( G n = 2 n−1 × G 1 , G 1 = 6.06 μS) (II), a sustained on-state T 1 as p 1 in the heterosynapse module (III), and a voltage follower as k 1 in the soma module (IV). The inset in the synapse module shows the output curves of the eight FETs. d Output characteristics of the 8-bit DAC. The inset (black box) provides a magnified view of the output signal between 170 and 180 milliseconds. 2D materials, which possess superior physical properties 11 , can support neuromorphic computing hardware 12 – 14 . Neuromorphic hardware based on 2D materials respectively adopts transistors, logic gates, and memory to construct synapses, heterosynapse, and soma components, which is optimized for a specific individual function 15 – 17 . Recent advancements have demonstrated the integration of multiple 2D material devices and multiple sensor modalities to achieve brain-like functionalities 18 – 22 . However, despite these advancements, limitations remain in fully emulating the computational flexibility of brain neural circuits, especially when efficient multitasking is required in dynamic environments 23 , 24 . This limitation results in resource waste in lightweight settings, specifically, excessive driving signals and device redundancy, because computational tasks of varying information content require different device quantities 25 – 27 . Moreover, this limited adaptability may yield suboptimal outcomes because different linear and nonlinear computational processes are required to match different tasks by circuit assembly 28 – 31 . Therefore, the development of 2D material-based reconfigurable analog hardware is the key to truly emulate the computational flexibility of brain neural circuits for multitasking demands in dynamic environments. In this work, motivated by biological principles, we developed an 2D MoS 2 -based reconfigurable analog hardware (RAH) that included synapse, heterosynapse, and soma modules (Fig. 1b, (I) ) and demonstrated its reconfigurable multiple functions and potential as a solution for general-purpose machines with rich dynamics. At the device level, synapse, heterosynapse, and soma modules were fabricated with MoS 2 FETs (including cascaded MoS 2 FETs and MoS 2 FET-based operational amplifier (OPA) units 32 ). At the circuit level, the synapse, heterosynapse, and soma module wiring assembly was adjusted based on the task requirements to process the signal transmission. By co-encoding the inner-states and inter-connections of all modules, the high adaptability and plasticity of RAH allowed the realization of diverse linear and nonlinear computing functions and effective handling of varying task requirements. The proposed RAH realized the functions of analog-to-digital converter (ADC) and digital-to-analog converter (DAC), with reconfigurable resolutions to match different tasks, a bandwidth of 50 kHz, and a maximum power consumption (8-bit ADC and DAC) of ~750 μW in a converting period. Given its adaptability and flexibility, RAH can also support multiple computing functions. For instance, it was employed in sparse coding and a convolutional computing, which was then used to reconstruct and image sharping from pathological regions in computed tomography (CT) images to facilitate easy identification and assessment by medical professionals. In addition, it imitated attention switching and visual persistence in visual systems through designed circuit configurations, which enabled the detection of distance and velocity in autonomous driving applications. This 2D MoS 2 -based RAH implemented the key degrees-of-freedom of the inner states and inter-connections of devices and modules. It can pave the way for the development of future brain-like general-purpose machines with high adaptability and plasticity for multitasking. Biology-inspired reconfigurable hardware Figure 1a illustrates the complex structures and functions of biological neural circuits formed by synapse, heterosynapse, soma, and their inter-connections. Motivated by biological principles, we developed an RAH by employing 2D MoS 2 FETs to translate the biological concepts in Fig. 1a to a concrete electronic hardware design (Fig. 1b ). RAH has synapse, heterosynapse, and soma modules, for which diverse computing functions can be built through connection adaption within or between these modules. As shown in Fig. 1b, (II) , the synapse module constructed with multiple cascaded MoS 2 FETs achieves synaptic-like plasticity by modulating the gate voltage to perform the FET conductance state ( G n ). Each transistor M n follows a customizable discrete equation marked as subfunction j x . The heterosynapse module fabricated with MoS 2 FET-based OPA units and six MoS 2 FETs (T 1 -T 6 ) controls heterosynaptic-like inter-connection among different neural circuits by encoding the on/off state of the MoS 2 FETs and is known as subfunction p x . The connection encoding table of on/off status for 6 MoS 2 FETs programs the heterosynaptic-like inter-connection under different environment-interactive requirements, as shown in Table S1 . The soma module, composed of MoS 2 FET-based OPA and diverse feedback loops, integrates the front-end transmitted signals and generates a soma-like response, thus forming subfunction k x 33 – 35 . The manufacturing details are given in the Methods section and Note S 1 . The optical images, SEM image, STEM image, and schematic of the hardware are shown in Fig. S1 and S2 , which incorporates over 600 MoS 2 -based FETs, with an impressive yield rate exceeding 95%. The performance of the MoS 2 FET arrays in the synapse module was measured and is presented in Figs. S3 and S4 . The MoS 2 FET-based OPA’s design details are shown in Note S 3 , and the corresponding feedback circuits formed with diverse feedback loops are shown in Figs. S5 – S12 . In consideration of the wiring–assembling degree of freedom commonly existing in the brain, the flexible combination of j x , p x , and k x realizes diverse functions for the hardware (Table 1 ). Table 1 Multiple functions with reconfigurable hardware configuration Hardware function Synapse module (subfunction j x ) Heterosynapse module (subfunction p x ) Soma module (subfunction k x ) Function 1: DAC Geometric sequence equation j 1 : \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${G}_{{\\mbox{n}}}={G}_{1}/{2}^{{\\mbox{n}}-1}$$\\end{document} G n = G 1 / 2 n − 1 ( n = 1–8) Sustained on-state p 1 : on state T 1 Operational state k 1 : Voltage follower Function 1: ADC Sustained on-state p 2 : on state T 3 Operational state k 2 : Voltage comparator Function 2: activation function (step function) Non-linear sequence equation j 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}$${G}_{1}={100*G}_{{\\mbox{j}}}$$\\end{document} G 1 = 100 * G j ( j = 2–8) Sustained on-state p 2 : on state T 3 Operational state k 2 : Voltage comparator Function 3: Convolution kernel Sharpness convolutional kernel j 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}$${G}_{5}={-9*G}_{{\\mbox{j}}}$$\\end{document} G 5 = − 9 * G j ( j = 1–4, 6–9) Sustained on-state p 1 : on state T 1 Operational state k 1 : Voltage follower Function 4: attention switching Gaussian distribution j 4 (activation): \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${G}_{{\\mathrm{1,8}}}={G}_{{\\mathrm{2,7}}}/2={G}_{{\\mathrm{3,6}}}/4={G}_{{\\mathrm{4,5}}}/8$$\\end{document} G 1,8 = G 2,7 / 2 = G 3,6 / 4 = G 4,5 / 8 Universal off states j 5 (inactivation): \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${G}_{{\\mbox{n}}}={G}_{{\\mbox{off}}}$$\\end{document} G n = G off Sustained on-state p 1 : on state T 1 Operational state k 1 : Voltage follower Function 5: visual persistence Sustained on-state p 2 : on state T 3 Operational state k 2 : Voltage comparator To demonstrate the functionalities experimentally, we encoded eight square-wave signals in eight input channels, with an amplitude of 0.1 V and frequencies 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}$${f}_{n}={2}^{n-1}\\times {f}_{1}(n=1,\\,2,\\,\\ldots,8)$$\\end{document} f n = 2 n − 1 × f 1 ( n = 1 , 2 , … , 8 ) , where f 1 = 5 Hz and n represents different input channels (Fig. 1c, (I) ). The gate voltages applied to each device in the synapse module could fix the conductance of M 1 –M 8 to the desired values to satisfy geometric sequence equation j 1 : \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${G}_{n}={G}_{1}/{2}^{n-1}$$\\end{document} G n = G 1 / 2 n − 1 ( G 1 is the conductance state of the first transistor, M 1 ). Here, the conductance of M 1 was G 1 = 6.06μS at V g = 3.62 V (Fig. 1c, (II) ). Subfunction p 1 was realized by a transimpedance amplifier with a feedback resistor ( R f = 300 kΩ) and an on-state MoS 2 FET and was connected to the voltage follower of subfunction k 1 (Fig. 1c, (III) and (IV) ). Figure 1d shows the performance of the 8-bit DAC, with the adjacent transformed analog voltage levels discerned distinctly. Versatile functions were achieved in RAH by programming different combinations of subfunctions j x , p x , and k x , which were then utilized to imitate the synapse, heterosynapse, and soma modules, respectively, leading to RAH’s ability to mimic the adaptive behavior of organisms in different environments. Eight-bit DAC/ADC RAH implemented the functionalities of 8-bit ADC and DAC to show its signal conversion capabilities. The related simplified circuit is presented in Fig. 2a . The detailed circuit design and operation mechanism of ADC and DAC are given in Notes S 3 and S 4 . In a differential mode configuration, low-frequency sine waves (1 V peak to peak) were inputted to the MoS 2 FET-constructed OPA in the soma module to characterize the frequency responses. The gain and phase response plots related to the input frequency are depicted in Fig. 2 b, c. The maximum gain achieved was 6 dB because the reference resistor was set to twice the input resistor, leading to a 0 dB gain bandwidth cutoff frequency ( f T ) at 50 kHz. The high-frequency operation can be improved by reducing the MoS 2 FET sizes 36 – 40 . Fig. 2 RAH adapted as 8-bit DAC and 8-bit ADC. a Diagram of an 8-bit DAC and ADC simplified circuit structure. Frequency responses of OPA, including gain bandwidth ( b ) and phase bandwidth ( c ). d Grayscale image of the digit 5 with 256 levels, with 6 selected pixel points as the input signals for the 8-bit DAC. e Entire data conversion process, including digital signal input V in1 – V in8 (1), DAC output signal V out, DAC (2), reference threshold voltage V th for ADC (3), and ADC output signal V out, ADC (4). f | V out, DAC | of the 8-bit DAC corresponding to the signal input in ( d ) and the noise level of the circuit. The inset compares the unit quantization voltage (UQV) with noise. g \n V out, ADC of the 8-bit ADC corresponding to the signal input in ( f ) and the scanning V th . h Error analysis by comparing | V out, DAC | and V th when ADC finishes conversion. Furthermore, to characterize the performance of RAH working as an 8-bit ADC and DAC, we chose six specific pixels from a grayscale image of the digit 5 containing 256 different levels as inputs (Fig. 2d ). Digital pixels were converted into analog signals by DAC and converted back to digital outputs by ADC (the procedure is in Fig. 2e , and the conversion details are in Note S 6 ). The analog voltage output values of DAC were measured (Fig. 2f ), and the inset magnified the signal and noise curves within the dashed red box. The unit quantization voltage (UQV) reached 14.15 mV, and the noise level remained at 24 μV. The UQV-to-noise ratio of the 8-bit DAC was 55.4 dB, indicating that the converted results could be distinguished clearly. Subsequently, the six analog voltage signals were sent to the 8-bit ADC, and the converted digital number was determined by the corresponding converting time width t , as shown in Fig. 2g . Simultaneously, we recorded the V load value when the output voltage switched to a low voltage level. Given that the load resistor of ADC and the reference resistor of DAC shared the same value, V load and | V DAC_OUT | had nearly similar values, with the difference being below 1 mV, which is much lower than UQV. This result confirms that the hardware has excellent fidelity in signal conversion (Fig. 2h ), and the DAC/ADC resolution is reconfigurable. The 4- and 6-bit ADC/DAC functions are presented in Notes S 4 and S 5 and Figs. S13 – S18 , which indicate that RAH can fit the resolution requirements to avoid resource wastage. Reconstruction and feature extraction of medical images The adaptability of RAH makes it applicable to various practical situations. Here, it was implemented to perform the reconstruction and feature extraction of medical computed tomography (CT) images, with the aim of improving the identification of pathological regions and enhancing the accuracy of medical diagnosis. To achieve the image restoration process, the hardware was used to construct Function 1 (DAC/ADC) and Function 3 (convolutional kernels) for signal conversion and convolution calculation, as shown in Fig. 3a . A 256 × 256 pixel original CT image containing bleeding spots in the brain was segmented into 8 × 8 pixel patches for image processing, as shown in the red dotted box in Fig. 3a (I) . First, the hardware working under ADC and DAC functions was employed for a sparse coding algorithm, in which the original image was reconstructed to reduce the noise via data transfer and conversion within ADC and DAC (Fig. 3a, (II) ). Second, the hardware adapted as convolutional kernels was used to sharpen the reconstructed image (Fig. 3a, (III) ). The sparse coding process that involves forward and backward DA/AD conversion (the details are in Note S 7 ) is shown in Fig. 3b . Signal 1 was mapped to 0.1 V, and Signal 0 was mapped to 0 V. Last, 256 random coefficients were constructed initially, and the forward and backward calculation results of the first iteration are shown in Fig. 3c . In the sparse coding experiment, the choice of regularization parameter λ was pivotal for the sparse coding outcomes to balance sparsity and image quality. Various λ values were applied experimentally to determine their effects. Figure S20 shows the corresponding activation level at λ = 2.2, in which only a few dictionary vectors contributed considerably to image reconstruction and captured the signal’s sparsity characteristics effectively. After 100 iterations, the optimized sparse coefficients were obtained and are shown in Fig. 3d . The small λ values could result in low sparsity but could cause overfitting in image reconstruction (Fig. S21 for λ = 0.1), and the large λ values allowed for sparse encoding but resulted in underfitting in image reconstruction (Fig. S22 for λ = 3.5). In the experiment, λ = 2.2 achieved excellent sparsity and image quality balance. Fig. 3 RAH configured for reconstruction and feature extraction of medical images. a Image reconstruction by sparse coding and feature extraction by convolution realized with the ADC/DAC function and nonlinear convolution function of RAH, respectively. b Process of sparse coding. c Output data of forward and backward DA/AD conversion for the first iteration in the experiments. d Sparse coding coefficients for all patches in the original image. e Image sharpening (Function 3) in RAH, in which the synapse, heterosynapse, and soma modules were configured as subfunction j 3 (convolutional kernel), subfunction p 1 (sustained on-state), and subfunction k 1 (voltage follower), respectively. f Electronic performance to evaluate subfunctions j 3 and k 1 . g Output current of a 6 × 6 patch from the convolution operation. The 2D grayscale image of the 6 × 6 patch is given in the inset. After image reconstruction, a convolution operation for image sharpening was applied by configuring RAH as convolutional kernels (Function 3, Fig. 3a (III) ). The configuration of a hardware as convolutional kernels avoids additional accelerators, which supports efficient large-scale convolution computations 41 . Such a configuration enhances the speed of image sharpening, particularly when dealing with extensive datasets. Initially, a 3 × 3 sharpening convolution kernel was designed, and the weight of the central element was nine times the weight of the surrounding elements. Accordingly, the weights of the convolution kernel were represented by the conductivity of the transistors in the synapse module to encode subfunction j 3 (Fig. 3e , left). The conductance of M 5 was 9 G 0 (3.4 μS) under V g = −0.7 V, and the conductance of the remaining components was G 0 (~0.37 μS) under V g = −2.4 V (Fig. 3f , left). The right panel of Fig. 3e shows a sustained on-state T 1 in the heterosynapse module (subfunction p 1 ). The soma module worked as a voltage follower (subfunction k 1 ), and its output characteristics are shown in the right panel of Fig. 3f . The image patch from the reconstructed image was encoded as 8-bit binary digital signals, which were then converted into analog voltage signals through the hardware working as an 8-bit DAC. These analog voltage signals were fed into the reconfigured hardware (Function 3) working for convolution operations. The convolution operation involved sliding a convolution kernel over each patch of the same size and calculating the convolution result in the corresponding region. Afterward, the output voltages were obtained to represent the extracted features from the respective regions. The experimental results are shown in Fig. 3g , and the features from the 256 × 256 grayscale CT image were extracted and are presented in Fig. 3a (III) . The blood vessels and tissue structures were displayed more clearly than those in the original image. Thus, RAH can provide robust support for future medical research and clinical applications. Additionally, image reconstruction with additional noise was implemented using RAH (Figs. S23 and S24 ). The peak signal-to-noise ratios of the noisy and reconstructed images are 20.1026 and 27.1774 dB, respectively, in Fig. S25a (20.2940 and 28.5906 dB in Fig. S25b ), confirming the high-quality reconstruction under additional noise. Notably, the number and conductivity states of cascaded transistors can be configured to form multifunctional convolution cores for flexible convolution operations. Visual processing in bionic receptive fields RAH also allows for brain-like multitasking to realize attention switching and visual persistence in biological visual systems, so it has potential for applications in autonomous driving (Fig. 4a ) 42 – 45 . Attention switching allows visual systems to focus on information only within the active receptive field (RF). The output signal changes immediately when RF shifts. RF selection enhances data sparsity, reduces redundancy, and effectively captures critical features 45 – 47 . Meanwhile, visual persistence can utilize previous data acquired outside the present RF to predict future scenarios and infer the actions of objects. The output signal that responded to the previous RF persists, and its disappearance is delayed after RF shifts. The biological mechanism is discussed in Note S 7 . Fig. 4 RAH configured for visual processing in bionic receptive (RF) fields. a Diagram of RF attention. Without visual residual, only signals within the active RF (left or right RF) are received. With visual residual, signals from the previously active RF persist for a short time and can be processed after the active RF is switched. b The left panel shows the simplified circuit connection for both RFs. The right panel presents the basic functional characteristics. The MoS 2 FET conductance of the active RF satisfies the Gaussian distribution with G 1 = G 8 = G 2 /2 = G 7 /2 = G 23 /4 = G 6 /4 = G 4 /8 = G 5 /8 = 3.76 μS ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${j}_{m1}^{l}$$\\end{document} j m 1 l ), and the output characteristics of each FET are shown in an orange background. The FET conductance of the inactive RF is in the off state with G off of G 1 /14 = 0.271 μS ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${j}_{m1}^{r}$$\\end{document} j m 1 r ), and the output characteristics of each FET are shown in a cyan background. The subfunction \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${k}_{m}^{p}$$\\end{document} k m p of the first-stage OPA constructed from MoS 2 acts as a voltage follower, and its output characteristics are shown in a gray background. The subfunction \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${k}_{m}^{{att}}$$\\end{document} k m a t t of the second-stage OPA constructed from MoS 2 acts as a comparator, and its output characteristics are shown in a pink background. The abovementioned modules were connected through subfunctions p 1 and p 2 . c The left panel shows the input signals 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}$${L}_{3}^{l}\\,$$\\end{document} L 3 l 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}$${L}_{3}^{r}\\,$$\\end{document} L 3 r (red) under left RF activation and 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}$${L}_{6}^{l}$$\\end{document} L 6 l 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}$${L}_{6}^{r}\\,$$\\end{document} L 6 r (blue) under right RF activation. The right panel presents the output signals of the first-stage OPA, where the red curve represents left RF, and the blue curve represents right RF. d Schematic of object detection in autonomous driving via RAH. Within the active left RF, the conductance distribution is programmed to compute distance and speed information. e , g For distance detection, the upper panel shows the output signals V out,l of the first-stage OPA obtained from each programmed location. The lower panel shows the output signals V out,att of the second-stage OPA obtained from each programmed location, whose width determines the distance ( g ). f , h For speed detection, the left panel presents the input signals with different time intervals representing different speed scenarios, and the right panel presents the output signals of the second-stage OPA obtained from each programmed location. The output time interval indicates the speed ( h ). The functionality (Functions 4 and 5) of RAH was adjusted to realize the abovementioned biological mechanism, as shown in Fig. 4b . The simplified hardware system used two synapse modules to represent the left and right RFs (marked with orange and blue backgrounds, respectively). A soma module with three OPAs (The electrical properties of individual FET are shown in Figs. S4 , S26 , and S 27 ) was employed to achieve RF switching and visual persistence, and a heterosynapse module was used to realize the inter-connection of synapse/soma modules by encoding the on/off switching of MoS 2 FETs. The active RF (synapse module 1) encoded a Gaussian distribution with \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${G}_{{\\mathrm{1,8}}}={G}_{{\\mathrm{2,7}}}/2={G}_{{\\mathrm{3,6}}}/4={G}_{{\\mathrm{4,5}}}/8=3.76\\mu S$$\\end{document} G 1,8 = G 2,7 / 2 = G 3,6 / 4 = G 4,5 / 8 = 3.76 μ S marked as subfunction \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${j}_{m}^{l},$$\\end{document} j m l , and the inactive RF (synapse module 2) represented universal off-states with \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${G}_{{off}}={G}_{1}/14=0.271\\mu S$$\\end{document} G o f f = G 1 / 14 = 0.271 μ S marked as subfunction \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${j}_{m}^{r}$$\\end{document} j m r . Signal persistence was realized with a voltage follower circuit by connecting the first-stage OPA (soma modules 1 and 2) circuit with a load capacitor ( C load , 33 μf; subfunction \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${k}_{m}^{p}$$\\end{document} k m p ). The second-stage OPA (soma module 3) served as a voltage comparator to generate the output (subfunction \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${k}_{m}^{{att}}$$\\end{document} k m a t t ). After circuit configuration, the left RF was activated in 0–50 ms, and the right RF was activated in 50–100 ms. A 20 ms, 0.1 V voltage pulse was sent to the input terminal’s \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{\\mbox{L}}}_{3}^{{\\mbox{l}}}$$\\end{document} L 3 l 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}$${{\\mbox{L}}}_{3}^{{\\mbox{r}}}$$\\end{document} L 3 r ports at the time of 25 ms and inputted to 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}$${{\\mbox{L}}}_{6}^{{\\mbox{l}}}$$\\end{document} L 6 l 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}$${{\\mbox{L}}}_{6}^{{\\mbox{r}}}$$\\end{document} L 6 r ports at the time of 50 ms (Fig. 4c, (I) ). When the attention was on the left RF, the left RF’s soma module 1 generated a voltage output ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{out},l}$$\\end{document} V o u t , l ) that was higher than that of the right RF’s soma module 2 ( V out,r ), leading to positive output voltage \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{out},{att}}$$\\end{document} V o u t , a t t for soma module 3 (Fig. 4c, (II) ). Similarly, negative output voltage V out,att was generated for soma module 3, representing the attention on the right RF (Fig. 4c, (II) ). The output signal at soma module 3 persisted for a longer time than the input pulse width, indicating the visual persistence effect. The distance and speed between the driving and target vehicles in autonomous driving can be determined using the abovementioned working principles. To simplify the testing, we focused on the detection of a single RF. The diagram is shown in Fig. 4d . M 1 –M 8 represent eight activation state encoders to program eight different locations (marked #1–#8) centered around the driving vehicle. These locations include four different distances to the driving vehicle that are distributed symmetrically at the front and back sides of the driving vehicle, and the corresponding conductance distribution of M 1 –M 8 follows \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${j}_{{\\mbox{m}}}^{{\\mbox{l}}}$$\\end{document} j m l . Notably, the programming of conductance can be configured based on actual road conditions to expand the detection range and improve accuracy, also highlighting the advantages of the adaptive hardware. Each encoder can receive an activation voltage pulse with a width of 4 ms and amplitude of 0.1 V when a target vehicle is detected at its corresponding programmed location. By processing the output signals of the hardware, we can obtain the distance and speed of the driving vehicle relative to the target vehicle. During distance detection and processing, the current accumulation in soma module 1 is determined by the encoded transistor conductance, with high conductance resulting in a high output voltage, as shown in Fig. 4e (I) . The high output voltage of soma module 1 can lead to a long pulse persistence time width for the output signal of soma module 3, which also indicates a close distance to the driving vehicle, as shown in Fig. 4e (II) . Under the assumption that the target vehicle is passing from locations #1 to #8 sequentially in the experiment setup, the measured time widths of the output signals at #1–#8 are 44, 59, 73, 87, 87, 73, 59, and 44 ms, respectively (Fig. 4g ), indicating that the target vehicle approaches the driving vehicle first then leaves the driving vehicle afterward. The distance information of the driving vehicle relative to the target vehicle can be distinguished directly. In speed detection and processing, the different speeds of the detection vehicle result in different time intervals when the vehicle moves from one programmed location to an adjacent one. Thus, the time intervals of soma module 3’s output voltage pulses can be used to determine the speed, in which a short time interval indicates high speed. As shown in Fig. 4f , to simplify the testing, we defined three scenarios with high, medium, and low speeds. Pulse signals with three different frequencies representing the three speeds were applied 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}$${{{\\rm{L}}}}_{1}^{{{\\rm{l}}}}{{\\rm{\\hbox{--}}}}{{{\\rm{L}}}}_{8}^{{{\\rm{l}}}}$$\\end{document} L 1 l – L 8 l in sequence, as shown in Fig. 4f (I) . The speed information could be distinguished directly by analyzing the time interval of the output signals (Fig. 4f, (II) ). The output frequencies of high, medium, and low speeds were ~12.04, ~6.02, and ~4.02 Hz, respectively (Fig. 4h ).",
"discussion": "Discussion In summary, the 2D MoS 2 -based RAH implemented the connection-adaptable degree of freedom innovated from neural circuits in the brain. The devices’ inner states and inter-connections were co-encoded to enhance adaptability, thus endowing RAH with multiple functions and potential as a solution for general-purpose machines. Notably, the MoS 2 FETs used in this work could be replaced with floating gate transistors, charge-trapping transistors, or memristors to achieve a non-volatile hardware structure and reducing reliance on gate control strategies 41 , 48 . RAH working under ADC and DAC functions exhibited reconfigurable resolutions (maximum of 8 bits), which can match different tasks to avoid resource wastage. ADC and DAC also realized a bandwidth of 50 kHz and a maximum power consumption (8 bits ADC and DAC) of ~750 μW during a converting period. In addition, wide-ranging computing applications, such as AI-assisted diagnoses and autonomous driving, were discussed to directly prove that the high-level adaptability and flexibility of RAH can appropriately meet the linear and nonlinear computing requirements of different tasks. As one potential scheme for future brain-like general-purpose machines, this hardware is expected to be used in many other practical applications, such as speech recognition and decision making. This design strategy opens an avenue to create other intelligent, efficient, flexible general-purpose hardware solutions for complex problems."
} | 10,497 |
38070192 | PMC10832552 | pmc | 705 | {
"abstract": "Abstract Improving methane production through electrical current application to anaerobic digesters has garnered interest in optimizing such microbial electrochemical technologies, with claims suggesting direct interspecies electron transfer (DIET) at the cathode enhances methane yield. However, previous studies with mixed microbial communities only reported interspecies interactions based on species co‐occurrence at the cathode, lacking insight into how a poised cathode influences well‐defined DIET‐based partnerships. To address this, we investigated the impact of continuous and discontinuous exposure to a poised cathode (−0.7 V vs. standard hydrogen electrode) on a defined consortium of Geobacter metallireducens and Methanosarcina barkeri , known for their DIET capabilities. The physiology of DIET consortia exposed to electrical current was compared to that of unexposed consortia. In current‐exposed incubations, overall metabolic activity and cell numbers for both partners declined. The consortium, receiving electrons from the poised cathode, accumulated acetate and hydrogen, with only 32% of the recovered electrons allocated to methane production. Discontinuous exposure intensified these detrimental effects. Conversely, unexposed control reactors efficiently converted ethanol to methane, transiently accumulating acetate and recovering 88% of electrons in methane. Our results demonstrate the overall detrimental effect of electrochemical stimulation on a DIET consortium. Besides, the data indicate that the presence of an alternative electron donor (cathode) hinders efficient electron retrieval by the methanogen from Geobacter , and induces catabolic repression of oxidative metabolism in Geobacter . This study emphasizes understanding specific DIET‐based interactions to enhance methane production during electrical stimulation, providing insights for optimizing tailored interspecies partnerships in microbial electrochemical technologies.",
"conclusion": "CONCLUSION In conclusion, our study challenges the prevailing belief that cathodic current promotes methanogenesis by DIET consortia. A consortium of Geobacter metallireducens and Methanosarcina barkeri exposed to a poised −0.7 V electrode exhibited reduced metabolic activity and lower cell numbers. The presence of a cathode as an alternative electron donor limited efficient electron retrieval by Methanosarcina from Geobacter , leading to the catabolic repression of oxidative metabolism in Geobacter . Despite Methanosarcina retaining methane production at −0.7 V, a significant fraction of the recovered electrons was diverted into acetate. Combining these results indicates that cathodic potentials promote the decoupling of the partnership between Geobacter and Methanosarcina . This underscores the importance of understanding the specific and combined effects of reactor parameters (electrical exposure and gas atmosphere) on key species in MEC‐AD. Such insights are essential for optimizing methane production and improving tailored interspecies partnerships in future microbial electrochemical technologies.",
"introduction": "INTRODUCTION Future applications aim to enhance sustainable biogas upgrading by integrating anaerobic digestion (AD) with microbial electrocatalysis (MEC). This integration combines the breakdown of organic matter through AD to produce biogas (almost equimolar gas mixture of CO 2 :CH 4 ) with MEC where renewable energy and microorganisms from AD are used as catalysts to convert CO 2 into high‐purity methane – a clean fuel. The potential benefits of this approach include improved efficiency, increased sustainability, and enhanced energy recovery from organic waste (Horváth‐Gönczi et al., 2023 ; Wang et al., 2022 ). However, practical implementation requires understanding microbial interactions at the electrode surface and their impact on methane formation. Application of electrical current to anaerobic digesters or MEC‐AD resulted in a significant increase in methane production (Wang et al., 2022 ). Because electroactive bacteria and electroactive methanogens were coexisting on cathodes in MEC‐AD, researchers often speculate that direct interspecies electron transfer (DIET) between electroactive bacteria and archaea at the cathode is a possible reason for improved methanogenesis (Bretschger et al., 2015 ; Jiang et al., 2022 ; Liu et al., 2016 ; Lu et al., 2023 ; Wang et al., 2020 , 2022 ; Wei et al., 2023 ; Zhao et al., 2015 , 2016 , 2022 ). To investigate whether DIET partnerships are indeed the key drivers behind enhanced methanogenesis in MEC‐AD, fundamental insights into how electrochemical exposure impacts defined DIET interspecies associations are required. To address this goal, a defined DIET consortium of Geobacter metallireducens and Methanosarcina barkeri serves as an ideal model system (Rotaru et al., 2014 ). In this consortium, the Geobacter oxidizes organic substrate and donates electrons to Methanosarcina . Yet, without an electron sink, Geobacter 's oxidative metabolism becomes unfavourable. In DIET, the methanogen acts as the sink, relying on Geobacter 's electrons for reductive metabolism, converting CO 2 to methane. Without Geobacter , the methanogen could not grow on ethanol alone (Rotaru et al., 2014 ). Methanosarcina species are prevalent in anaerobic digesters (De Vrieze et al., 2012 ) exhibiting a versatile metabolism, utilizing methylated compounds, acetate, hydrogen (Costa & Leigh, 2014 ), and even electrons directly when in partnership with syntrophic bacteria such as Geobacter (Wang et al., 2021 ). Previously, DIET interactions were established between Geobacter metallireducens and multiple Methanosarcina species commonly found in anaerobic digesters, including M. barkeri and M. mazei (Yee & Rotaru, 2020 ). Additionally, these Methanosarcina species demonstrated electromethanogenesis by directly receiving electrons from a cathode (Rowe et al., 2019 ; Yee et al., 2019 ; Yee & Rotaru, 2020 ). One of the critical parameters influencing electromethanogenesis in MEC‐AD is the cathode potential. A more negative cathode potential has been linked to increased methane production (Kim et al., 2022 ; Zhang et al., 2022 ; Zhen et al., 2016 ). Although methanogens generally thrive under negative electrode potentials, subjecting bacteria to negative potentials is often detrimental especially to biofilm formation. This is a widely adopted strategy in various industries to control biofouling (Sultana et al., 2015 ). Particularly, when a Geobacter was exposed to a negative electrode potential, it adversely affected its metabolism, leading to a buildup of reduced electron carriers, such as NAD(P)H, which can impede extracellular electron transfer and disrupt redox homeostasis (Korth & Harnisch, 2019 ; Song et al., 2016 ). Understanding the physiological and dynamic response of the DIET partners when electrical current is applied could provide valuable insights for designing tailor‐made consortia for MEC‐AD or other upcoming biogas upgrading technologies relying on microbial electrosynthesis. Contrary to prevailing beliefs in MEC‐AD studies, our hypothesis is that exposure to a cathodic potential leads to a metabolic decoupling of the DIET partners by providing a competitive electron donor for the methanogen (the cathode) and inhibiting the oxidative metabolism of the syntroph. To explore this hypothesis, our study aimed to address the following key research questions: What is the impact of high‐potential electrons at −0.7 V delivered by a cathode on partners of a syntrophic consortia of Geobacter metallireducens and Methanosarcina barkeri ? Can we observe a potential decoupling of these two partner organisms, and if so, how does it affect their respective metabolisms and cell distributions? Does the methanogen transition to utilizing cathodic electrons rather than relying on DIET electrons provided by Geobacter ? What are the potential negative effects, if any, due to exposure to a negative cathodic potential? The findings presented in this work enhance our understanding of the impact of cathodic exposure on syntrophic partners and the perspectives of employing syntrophic consortia for sustainable and efficient biogas upgrading in MEC‐AD systems.",
"discussion": "RESULTS AND DISCUSSION Our hypothesis was that exposing DIET consortia to a poised cathode at −0.7 V would disrupt the oxidative metabolism of the syntrophic partner, challenging prevailing beliefs in MEC‐AD studies. We aimed to investigate whether a −0.7 V exposure would decouple the oxidative metabolism of Geobacter from the reductive metabolism of Methanosarcina , with the expectation of a decrease in the former and an increase in the latter. However, our findings revealed a surprising decline in both oxidative metabolism and reductive metabolism at −0.7 V (Table 1 ), suggesting general metabolic constraints for both partners. Consequently, it is unlikely that DIET consortia exposed to electrical current can account for enhanced methanogenesis in MEC‐AD. TABLE 1 Electron recoveries from ethanol and cathode into products, such as acetate, hydrogen gas, and methane, by DIET consortia without or after exposure to electrical current (−0.7 V). Condition Cathodic electrons (mM eq.) Electrons from oxidized ethanol (mM eq.) Decrease below control ( p ‐value) Electron recovery in acetate (mM eq.) Increase above control ( p ‐value) Electron recovery in hydrogen (mM eq.) Increase above control ( p ‐value) Electron recovery in methane (mM eq.) Decrease below control ( p ‐value) % Electron recovery of electrons from ethanol as methane (as a fraction of products) Total electron recovery in products (%) DIET control None 61.8 ± 4.5 4.3 ± 2.5 0.0 ± 0.0 32.7 ± 2.4 ~53 (88%) ~60 DIET at −0.7 V cont. 15.9 ± 3.1 50.1 ± 4.4 1.2‐fold (0.047) 32.6 ± 0.8 8‐fold (0.015) 1.8 ± 0.8 121‐fold (0.009) 15.9 ± 6.1 2‐fold (0.004) ~24 (32%) ~76 DIET at −0.7 V int. 6.4 ± 5.6 44.3 ± 4.6 1.4‐fold (0.020) 32.2 ± 1.7 8‐fold (0.005) 2.2 ± 1.5 146‐fold (0.065) 6.9 ± 3.6 5‐fold (0.001) ~14 (17%) ~82 DIET or −0.7 V int. 7.0 ± 2.0 31.7 ± 4.1 1.9‐fold (0.000) 20.6 ± 1.3 5‐fold (0.021) 1.6 ± 0.6 103‐fold (0.006) 12.2 ± 4.1 3‐fold (0.001) ~32 (36%) ~89 \n Note : Representative chronoamperometry under all conditions, alongside abiotic controls, can be found in Figure S3A–E . Exposure to –0.7 V disrupts oxidative metabolism in DIET consortia Exposure to a negative potential of −0.7 V resulted in a minor reduction in ethanol oxidation compared to control DIET consortia ( p = 0.3, Figure 2A , Table 1 ). However, interruption of −0.7 V application led to a noticeable halt in ethanol oxidation ( p < 0.05, Tables 1 , S1 , S2 ). This was surprising, because a gap in exposure was expected to give Geobacter time to regenerate its oxidized electron carriers, such as NAD(P) + . Ethanol consumption remained detectable in all incubations with cells but was close to the detection limit in abiotic controls (Figure 2B ). FIGURE 2 Ethanol and acetate metabolism of a DIET consortium of Geobacter metallireducens and Methanosarcina barkeri with or without exposure to electrical current (−0.7 V). (A) Ethanol oxidation profiles in reactors provided with cells. (B) Ethanol oxidation profiles in abiotic reactors without cells. (C) Accumulation of acetate in reactors with cells. All conditions were carried out at least in triplicate ( n = 3); the exceptions were with one control DIET reactor lost during gas transfer at day 10 ( n = 2, from day 12) and the abiotic at −0.7 V intermittent ( n = 2). The most significant drop in ethanol oxidation occurred when shifting from an 80:20 N 2 :CO 2 to a 50:50 CH 4 :CO 2 biogas atmosphere, where ethanol consumption for −0.7 V‐exposed consortia was half that of control DIET consortia (p < 0.05, Tables 1 , S1 , S2 ). Thermodynamically, ethanol conversion to methane is less favourable under the biogas atmosphere (Table S3 ). The decline in ethanol oxidation may be attributed to a thermodynamic shift due to abundant reaction products, a pH decrease, or a negative impact of elevated CO 2 on Geobacter's metabolism. There is minimal pH variation (0.16 ± 0.08 pH units) when transitioning to a high CO 2 (50%) atmosphere, with media pH remaining above 6.7, well within the optimal range for both Geobacter and Methanosarcina, suggesting that pH plays an insignificant role. On the other hand, Geobacter's oxidative metabolism could be negatively affected by elevated CO 2 in line with previous reports on syntrophic bacteria (Ceron‐Chafla et al., 2020 ; Jin & Kirk, 2016 ; Kato et al., 2014 ) likely due to thermodynamic limitations (Table S3 ). These findings indicate that Geobacter is sensitive to cumulative changes in their environment – such as negative cathodic potentials and elevated CO 2 from biogas. The negative impact of electrical current on the oxidative metabolism of the consortia was further substantiated by the significant inhibition of acetate utilization (Figure 2C , Table S4 ). Continuous exposure to −0.7 V increased acetate accumulation by 1.8‐fold (under 80:20 N 2 :CO 2 ) or a staggering 9.8‐fold (under 50:50 CH 4 :CO 2 ) compared to control DIET consortia (p < 0.05, Figure 2C , Table S4 ). Interrupting the electrical current did not restore acetate utilization, regardless of the gas phase (dotted lines in Figure 2C ). Only intermittent ethanol addition showed lower acetate accumulation due to a lower ethanol supply (Figure 2A ). Unlike DIET consortia where the acetate to ethanol ratio goes quickly from 1 to zero (Figure S1 ), cultures exposed to −0.7 V maintained a stable acetate to ethanol ratio of 1, denoting impaired acetate metabolism during the 20‐day incubation (Figure S1 ). In these consortia, acetate could serve two purposes: (i) methanogenic substrate for Methanosarcina or (ii) electron donor for G. metallireducens when it has an effective DIET partner (Rotaru et al., 2018 ; Wang et al., 2016 ). Thus, there are two possible reasons for acetate accumulating when cultures are exposed to −0.7 V: Consumption by acetoclastic methanogenesis becomes unfavourable when cathodic H 2 becomes available, or syntrophic acetate oxidation becomes unfavourable when cathodic H 2 becomes available. Under standard conditions, hydrogenotrophic methanogenesis has a 55 kJ/mol favourability (Table S3 ) over acetoclastic methanogenesis. However, in our reactor conditions, regardless of the gaseous atmosphere, acetoclastic methanogenesis becomes rapidly more favourable than hydrogenotrophic methanogenesis. In a typical co‐culture atmosphere, acetoclastic methanogenesis surpasses hydrogenotrophic methanogenesis when acetate exceeds 0.5 mM (Figure 3A ). In a biogas atmosphere, it becomes more favourable once acetate exceeds 2.5 mM (Figure 3B ). Since acetate exceeds these thresholds quickly or right at the incubation onset in both conditions, exposure to −0.7 V does not appear to hinder acetoclastic methanogenesis. And yet, acetate is not consumed by consortia exposed to −0.7 V. FIGURE 3 Gibbs free energy for acetoclastic methanogenesis (black lines) and hydrogenotrophic methanogenesis (red lines) at varied acetate concentrations under (A) a normal co‐culture atmosphere containing 20% CO 2 and 80% N 2 , and (B) a biogas atmosphere containing 50% CO 2 and 50% CH 4 (see detailed parameter information in Table S3 ). Possibly, acetate accumulates because −0.7 V obstructs Geobacter 's ability to perform syntrophic acetate oxidation due to the presence of an alternative electron donor (cathodic H 2 ) for the methanogenic partner. This aligns with the literature indicating such inhibition of syntrophic metabolism (Wu et al., 1996 ). Moreover, Geobacter exhibits a higher acetate affinity (half‐saturation constant/K M = 0.01–1 mM) (Esteve‐Nunez et al., 2005 ) compared to Methanosarcina barkeri (K M = 4 to 5 mM) (Smith & Mah, 1978 ; Westermann et al., 1989 ). Given that acetate accumulation remains below the K M of M. barkeri , Geobacter is the likely primary contributor to acetate utilization. Consequently, the severe acetate accumulation observed at −0.7 V likely results from severe alterations in Geobacter 's acetate metabolism. Exposure to −0.7 V Induces H 2 \n accumulation Hydrogen buildup occurred strictly when cathodes were poised. Abiotic controls exposed to −0.7 V accumulated more H 2 when exposed continuously to −0.7 V rather than intermittently (6.2‐fold, p = 0.008), consistent with electrochemical H 2 production, independent of the overlying atmosphere (80:20 N 2 :CO 2 or biogas 50:50 CH 4 :CO 2 ). However, the H 2 buildup doubled in the abiotic reactors under the biogas atmosphere ( p = 0.01 at −0.7 V continuous). This is most likely due to the elevated CO 2 forming carbonic acid, which can then dissociate into protons (H + ) and carbonate ions (CO 3 \n 2− ) (Bajracharya et al., 2015 ). The marginal pH decrease (−0.16 ± 0.08 pH units) upon switching to a biogas atmosphere may possibly boost proton (H + ) availability at the cathodic surface, leading to increased H 2 evolution at −0.7 V. Contrariwise, DIET control experiments without poised cathodes showed no H 2 accumulation, confirming their H 2 ‐independent nature (Figure 4A ). FIGURE 4 Hydrogen accumulation and methane formation by a DIET consortium with or without exposure to electrical current (−0.7 V). (A) Hydrogen accumulation profiles in reactors with cells. (B) Hydrogen accumulation profiles in abiotic reactors without cells. (C) Methane formation in reactors with cells. Data represent at least triplicate reactors for each condition ( n ≥ 3), with one control DIET reactor lost during gas transfer at day 10 ( n = 2 from day 12) and the abiotic at −0.7 V intermittent ( n = 2). (B) Subfigure shows calculated values of expected methane from cathodic H 2 assuming a 4:1 stoichiometry of H 2 to methane according to the reaction: 4H 2 + CO 2 → CH 4 + 2H 2 O. Initially, it was expected that abiotic H 2 formed at −0.7 V would be consumed by M. barkeri . Consequently, this would decouple M. barkeri from G. metallireducens in consortia exposed to −0.7 V. So, M. barkeri would favour cathodic H 2 /electrons (Rowe et al., 2019 ) over electrons from DIET with G. metallireducens . Contrary to our expectations, when first exposed to −0.7 V, the DIET consortia showed significant H 2 accumulation compared to abiotic media controls (4‐fold, p = 0.007, Figure 4A,B ). This observation indicates that M. barkeri does not have an immediate preference for utilizing cathodic H 2 , suggesting limitations in M. barkeri 's metabolic capability to use H 2 as their electron donor while engaged in electron uptake from Geobacter . Furthermore, the 4‐fold H 2 accumulation above abiotic controls indicates that at −0.7 V, M. barkeri has abundant reduced ferredoxin (Fd red ) and reduced F 420 (F 420 H 2 ) to produce excess extracellular H 2 (Lovley, 2018 ) via reactions catalysed by two hydrogenases – the Ech hydrogenase and the Frh hydrogenase: \n (1) \n Ech − hydrogenase membrane bound : Fd red + 2H + → Fd ox + 2H 2 pumps 2 H + \n \n \n (2) \n Frh − hydrogenase intracellular : F 420 H 2 → F 420 + H 2 \n \n Interestingly, when −0.7 V‐treated DIET consortia were subjected to a 50:50 CH 4 :CO 2 biogas atmosphere they accumulated less H 2 than abiotic incubations without cells (Figure 4A,B ). This strongly supports the notion that elevated CO 2 levels promoted the consumption of H 2 by Methanosarcina , aligning with the findings of other researchers regarding the impact of elevated CO 2 on hydrogenotrophic methanogenesis (Garcia‐Robledo et al., 2016 ). In conclusion, it appears that electron uptake at −0.7 V or H 2 metabolism is not favoured by M. barkeri from DIET consortia, unless the atmosphere has elevated CO 2 , which promotes hydrogenotrophic methanogenesis (Garcia‐Robledo et al., 2016 ). Influence of exposure to −0.7 V on methane formation In our experiments, the provision of additional extracellular electrons at −0.7 V to a DIET consortium resulted in a notable decline in methane buildup compared to control consortia, particularly evident under a biogas atmosphere (Figure 4C , p < 0.005 for all −0.7 V treatments). An increase in methane buildup above that of control DIET consortia was expected, assuming that cathodic H 2 is unlimited and the primary driver for methane production, surpassing electron uptake by DIET (Figure 4B – inset). Strikingly, DIET consortia exhibited a three‐fold increase in methane production under elevated CO 2 levels, regardless of cathodic exposure (Figure 4C , p ≤ 0.05). This is remarkable given that hydrogenotrophic methanogenesis is 90.5 kJ/mol less favourable under the 50:50 CO 2 :CH 4 than under the N 2 :CO 2 atmosphere. However, the higher diffusivity of CO 2 in water compared to methane likely results in elevated CO 2 concentrations promoting CO 2 ‐reductive methanogenesis. This is likely through stimulation of hydrogenotrophic pathways (Garcia‐Robledo et al., 2016 ) or potentially by enhancing direct electron uptake. This connection is supported by increased methane buildup and minimal H 2 detected under high CO 2 . Even in control experiments with DIET consortia unexposed to cathodic current, with H 2 at the detection limit, elevated CO 2 appears to promote electron uptake and increased methanogenesis. Contrary to our expectations, cells exposed to −0.7 V did not exhibit the highest electron recovery in methane. The control DIET consortia directed electrons towards methane as its major sink for electrons (88% of the electrons recovered in products, Table 1 ). In contrast, all −0.7 V incubations directed a significant proportion of electrons towards acetate production (60%–78%), with methane (16%–33%) and even hydrogen gas (3%–5%) as limited electron sinks. These findings indicate a shift in metabolic pathways in response to −0.7 V exposure, diverting electrons towards acetate production rather than methane as originally anticipated. In conclusion, our results challenge the prevailing notion that DIET associations are the primary drivers of enhanced methanogenesis in MEC‐AD. Further investigations are essential to fully comprehend the intricate interactions of microorganisms on electrodes in MEC‐AD, particularly under varying electrochemical conditions. The astonishing increase in methane accumulation under elevated CO 2 levels regardless of cathodic exposure raises intriguing questions about the dual roles of CO 2 in promoting hydrogenotrophic and DIET methanogenesis, warranting further exploration. These insights pave the way for a deeper understanding of the underlying mechanisms governing methane production in MEC‐AD systems, offering valuable implications for optimizing and controlling these bioelectrochemical processes. Impact of −0.7 V exposure on Geobacter and Methanosarcina Negative cathodic potentials were expected to reduce the Geobacter population, given the adverse effects on oxidative metabolism. Our results confirmed this, as Geobacter's estimated abundance significantly decreased at −0.7 V compared to DIET controls (N 2 :CO 2 atmosphere, p ≤ 0.05). Transitioning to a biogas atmosphere led to an even more dramatic decline in Geobacter's population at −0.7 V (13‐ to 37‐fold lower than DIET controls, p ≤ 0.0001, Figure 5A ). FIGURE 5 Cell abundance estimates of Geobacter and Methanosarcina in DIET consortia with and without exposure to −0.7 V as determined through analyses of 16S rRNA copy numbers. Cell abundance estimates were conducted at three critical steps: for the inoculum, after exposure to a typical co‐culture atmosphere (80:20 N 2 :CO 2 ) and after exposure to a biogas atmosphere (50:50 CH 4 :CO 2 ). The estimation was done by dividing the quantified copy numbers of the 16S rRNA gene by the specific number of gene copies in the genome of each organism (two copies for G. metallireducens and three copies for M. barkeri ). Abundance estimates of both (a) Geobacter and (b) Methanosarcina were derived from absolute 16S rRNA gene copy numbers detected using real‐time PCR. The resulting box plots represent the interquartile range, with whiskers representing data variability. Outliers are displayed as small circles outside the box and whisker area. The line inside the box indicates the median, while the cross represents the mean. The data are derived from biological replicates ( n > 3 for all conditions except the DIET control at 50% CO 2 , n = 2). Additional technical duplicates were performed for each biological replicate. In accordance with our hypothesis, exposing a DIET consortium to −0.7 V was expected to decouple the partners, with a negative impact on Geobacter and a neutral or positive impact on Methanosarcina. Indeed, under an N 2: CO 2 atmosphere, Methanosarcina abundance at −0.7 V resembled DIET controls ( p ≥ 0.1 for all conditions). Conversely, under a biogas atmosphere, Methanosarcina abundance at −0.7 V decreased significantly (2‐ to 6‐fold below DIET controls, p ≤ 0.05, Figure 5B ). This decline is inconsistent with the observed increase in methanogenic activity under a biogas atmosphere. This discrepancy could be due to (i) increased metabolic rates per cell or (ii) electrode attachment. Our current quantification method does not account for cells attached to electrodes. However, physiological and thermodynamic considerations suggest that Methanosarcina primarily relies on cathodic H 2 /electrons and CO 2 for methanogenesis in a biogas atmosphere. Greater reliance on cathodic electrons may facilitate cell attachment, potentially reducing planktonic Methanosarcina abundance while sustaining high methanogenesis rates by attached cells. This remains to be validated in future work. To investigate the impact of −0.7 V on the partnership between Geobacter and Methanosarcina, we analysed changes in cell ratios. If partners are decoupled by a competitive electron donor for the methanogen (−0.7 V), the ratio of Geobacter to Methanosarcina is expected to drop. Indeed, we observed a significant drop in the ratio of Geobacter to Methanosarcina at −0.7 V compared to DIET controls, which became more pronounced under a biogas atmosphere ( p < 0.05, Table S5 , Figure S2 ). These findings suggest that the Geobacter–Methanosarcina partnership is decoupled by −0.7 V. These results challenge the prevailing assumption that DIET partnerships are favoured on a cathode in MEC‐AD (Zhao et al., 2015 ; Zhao et al., 2016 ; Liu et al., 2016 ; Wang et al., 2020 ; Zhao et al., 2022 ; Wang et al., 2022 ; Jiang et al., 2022 ; Lu et al., 2023 ; Wei et al., 2023 ). Our results indicate that a Geobacter capable of DIET does not find the cathodic conditions favourable at all. These insights highlight the complexity of microbial interactions in bioelectrochemical systems and emphasize the need for further research to fully understand the underlying mechanisms in MEC‐AD, including the intricate dynamics of Methanosarcina in response to electrochemical conditions. Implications of microbial syntrophy to MEC‐AD In AD, interspecies interactions play a pivotal role in promoting efficient methanogenesis from complex substrates. However, in the context of microbial electrochemical anaerobic digestion (MEC‐AD), these interactions are subject to the influence of specific parameters, including applied voltage and gaseous atmosphere. The coexistence of diverse microbial species on the electrodes during the application of electrical current may not solely result from syntrophic partnerships but rather from the preference of specific microorganisms for distinct respiratory niches available in MEC‐AD. For instance, electrogenic bacteria such as Geobacter may thrive at the anode surface due to their ability to oxidize organic matter and generate electricity. Network analyses conducted by Zakaria and Ranjan Dhar ( 2022 ) confirmed potential interspecies interactions at the anode between Geobacter , Methanosaeta , and Methanobacterium . This association could be established by Geobacter oxidizing organic compounds and releasing acetate and hydrogen, which are subsequently utilized by Methanosaeta and Methanobacterium for their respective methanogenesis pathways (Zakaria & Ranjan Dhar, 2022 ). Similarly, at the cathode surface, network analyses demonstrated a linkage between Methanosarcina and acetogens. However, the predominant interaction between these two groups is most likely competition for cathodic electrons (Zakaria & Ranjan Dhar, 2022 ). Nevertheless, while many studies have noted Geobacter 's abundance at the cathode surface, the specific interplay between Geobacter and coexisting methanogens remains unexplored (Liu et al., 2016 ; Lu et al., 2023 ; Wang et al., 2020 ; Wei et al., 2023 ; Zhao et al., 2016 , 2022 ). It is plausible that cathodic Geobacter may consume cathodic electrons (or H 2 ) while respiring fermentation products from MEC‐AD, such as fumarate – a common metabolism in numerous Geobacter species (Lovley et al., 2011 ). In such cases, Geobacter would likely engage in competition with methanogens for cathodic electrons. Further dedicated investigations are warranted to determine the actual interplay between electroactive microorganisms and methanogens in MEC‐AD. Such studies hold significant promise for advancing our understanding of microbial interactions and optimizing MEC‐AD processes to facilitate the development of sustainable and efficient bioenergy solutions."
} | 7,429 |
28676725 | PMC5496863 | pmc | 706 | {
"abstract": "Workers of social insects, such as bees, ants and wasps, show some degree of inter-individual variability in decision-making, learning and memory. Whether these natural cognitive differences translate into distinct adaptive behavioural strategies is virtually unknown. Here we examined variability in the movement patterns of bumblebee foragers establishing routes between artificial flowers. We recorded all flower visitation sequences performed by 29 bees tested for 20 consecutive foraging bouts in three experimental arrays, each characterised by a unique spatial configuration of artificial flowers and three-dimensional landmarks. All bees started to develop efficient routes as they accumulated foraging experience in each array, and showed consistent inter-individual differences in their levels of route fidelity and foraging performance, as measured by travel speed and the frequency of revisits to flowers. While the tendency of bees to repeat the same route was influenced by their colony origin, foraging performance was correlated to body size. The largest foragers travelled faster and made less revisits to empty flowers. We discuss the possible adaptive value of such inter-individual variability within the forager caste for optimisation of colony-level foraging performances in social pollinators.",
"introduction": "Introduction In recent years, behavioural ecologists have become increasingly interested by the fact that animals often exhibit consistent behavioural traits that vary between individuals from the same group, population or species, irrespective of time or context 1 – 3 . Inter-individual behavioural variability has been described in a wide range of taxa, from invertebrates (nematodes 4 , cnidarians 5 , molluscs 6 , insects 7 , 8 ) to mammals 9 , including humans 10 . The existence of such individualistic behavioural traits may have different adaptive values depending on the ecology of the species 11 – 13 . Social insects, such as ants, some bees and wasps, show extreme cases of inter-individual behavioural variability 14 . In these animals, division of labour typically implies that specific individuals reproduce (the queens and the males), whereas others work to support their reproductive outputs (the workers) 15 . Among the workers different individuals specialise on different roles. Some take care of the brood (the nurses), while others defend the colony entrance (the guards and the soldiers) or collect food (the foragers). These behavioural specialists exhibit specific behavioural repertoires that can be associated with differences in morphology (e.g. bumblebees 16 ), age (e.g. honey bees 17 ), physiology and genetics (e.g. honey bees 18 , 19 ), or experience (e.g. ants 20 ), together defining the caste phenotype. Growing evidence indicates that some level of behavioural variability also exists between individuals of the same caste 21 – 23 . For instance in bumblebees, foragers show consistent inter-individual differences in decision speed and accuracy in flower discrimination tasks 24 , 25 . When having to choose between a rewarding flower and an empty flower in a laboratory decision chamber, some foragers always make slow but accurate decisions, while others are consistently fast and inaccurate 24 . Foragers also show inter-individual variability in learning performance 22 , 26 and colonies containing foragers with high visual learning speeds have a higher foraging efficiency 27 . These differences are independent of body size or any other measurable morphological attributes 27 . Whether such cognitive variability translates into distinct foraging strategies in the more complex and ecologically relevant task of exploiting patchily distributed floral resources remains virtually unexplored. In nature, bees often develop stable foraging routes (sometimes called traplines in analogy to trappers checking their traps along fixed routes 28 ) to exploit multiple feeding locations from their central nest 29 , 30 . Manipulative experiments on bumblebees 31 , 32 and honey bees 33 foraging for sucrose solution in simple arrays of artificial flowers (equivalent to natural flower patches) show how foragers often find the shortest possible route to visit all flowers once and return to the nest using an iterative improvement strategy based on learning and memory that is different from just linking nearest neighbour locations 31 , 34 . Thus far empirical research on trapline foraging has been aimed at describing this behaviour at the species level, using relatively small sample sizes (four to seven individuals per experiment), without characterising variation among individuals 31 – 33 , 35 – 38 . In principle however, some level of variation in the foraging behaviour of the workers of a colony could improve the colony foraging efficiency 39 . Regular trapliners that accurately follow the same route across multiple hours or days may perform better in stable environments when resources are highly predictable, while irregular trapliners that sample new locations at each foraging bout may be advantaged in more variable environments. Consequently, colonies containing foragers of different behavioural profiles may differ in performance in similar environmental conditions. Understanding how natural behavioural variability affects the foraging performances of colonies may help evaluate the adaptability of bees in the face of environmental changes, such as natural climatic events, human-induced habitat degradations or the introduction of predators and parasites 40 . Ultimately, this approach may also help refine predictions of current pollination models based on bee movement patterns 34 , 38 , 39 , 41 , 42 . Here we explored the level of inter-individual variability in the foraging behaviour of bumblebees ( Bombus terrestris ) by comparing the movement patterns of foragers from two colonies collecting sucrose solution in three different arrays of artificial flowers and landmarks in a controlled flight room.",
"discussion": "Discussion Understanding inter-individual behavioural variability in complex societies, such as colonies of social insects, may offer unique insights into how and why relatively high levels of inter-individual behavioural variability are observed in animal groups and populations 22 , 45 . Here we compared the movement patterns of all foragers from two bumblebee colonies exploiting arrays of stable feeder locations, and report consistent inter-individual differences in their spatial foraging behaviour. Rather than defining distinct behavioural profiles of foragers, this natural variability follows a continuum along two behavioural dimensions. Some bees were always more faithful to a route and/or faster and more accurate in their spatial foraging decisions than others. Bees showed consistent inter-individual variability in their tendency to follow stable routes between flowers. This variability was neither explained by the characteristics of our experimental arrays of flowers and landmarks, nor the body size or the age of bees. Interestingly, degrees of route fidelity differed between our two colonies, meaning that foragers from one colony were more regular in following a route than those from the other colony. These results are not due to differences in the average body size or age between the foragers of each colony. Behavioural variability between individuals of different groups or colonies is a widespread phenomenon in social animals 45 , including insects 21 , 46 – 48 . Inter-colonial behavioural variability has been reported previously in bees, (e.g. aggression in honey bees 49 or for both vision- and olfaction-related cognitive tasks in bumblebees 27 ) and suggested to be correlated with the foraging success of colonies 26 , 27 . In bumblebees, high genetic relatedness between colony members, due to female monandry (single mating) and haplo-diploidy (haploid males, diploid females), may favour strong inter-colony variability 26 , 50 . Other non-genetic factors may also contribute to phenotypic variability between colonies, such as changes in the pre-imaginal environment. For instance variation in nest temperature 51 and nutrition 52 during the larval stage can lead to differences in olfactory learning in adult honey bees. Further studies using more colonies with known genetic relatedness are needed to test the existence of a genetically determined inter-colony variability for traplining. In the present spatial task, bees also showed some level of inter-individual variability in their ability to make fast and accurate spatial decisions, so that fast travelling bees made fewer revisits to empty flowers. This result is consistent with the observation that goal-directed flights in experienced bees, for instance between the nest and familiar flowers, are faster than exploration flights, in which naïve bees scan the environment to search for flowers and acquire spatial memories 38 , 43 . Thus potentially bees showed inter-individual variability in their tendency to make exploitation and exploration flights. Interestingly, differences in foraging performance among bumblebee foragers were partly explained by differences in their body size, so that larger foragers tended to travel faster and make fewer revisits than smaller foragers. Bumblebees show a continuous variation in body size that is primarily determined by the frequency of feeding so that larvae raised in the middle of the nest area (where workers are more active) tend to become the largest adults 53 . Size polymorphism is considered a main factor of caste determinism in bumblebees, such that only the largest individuals tend to undertake foraging the tasks 54 . Our novel results suggest that natural size variations also influence within caste behavioural variance among foragers. This observation is consistent with previous studies showing that the largest bumblebees make more foraging trips 55 , take less time 16 and collect more nectar in natural conditions 16 . Large bumblebees also tend to learn faster in visual discrimination tasks 56 . These inter-individual behavioural and cognitive differences may be explained by differences in the sensory equipment of small and large bees. For instance, larger bees have bigger compound eyes and may thus be more accurate at finding small objects 57 . Size polymorphism in bumblebees is primarily determined by the frequency of feeding so that larvae raised in the middle of the nest area (where workers are more active) tend to become the largest adults 53 . Therefore it is very likely that the diversity of body sizes and their associated behavioural traits between and within castes of bumblebee colonies is a self-organised process, regulated by population densities and structural constraints within the nest at a given time during the colony cycle. Our description of inter-individual variability in the spatial foraging behaviour of bumblebees is in line with recent observations that foragers of social bees show high variability to their contribution to the global colony foraging effort 55 , 58 , suggesting that some behavioural traits may support higher foraging success. It has been suggested that behavioural diversity in a social group or population can be an advantageous trait at the collective level 7 , 8 . Honey bee colonies showing higher genetic variability (and thus inter-individual behavioural variability) perform better in group tasks such as nest thermoregulation 59 . Colonies of Thermothorax ants showing high variability in the aggressiveness of workers are more productive 13 . In the social spider Anelosimus studiosus , mixed colonies composed of aggressive (asocial) and docile (social) individuals capture more prey than colonies with high proportion of only one type of individuals 60 . Accordingly, maintaining a diversity of behavioural profiles among foragers of a colony may allow the colony to locate and exploit a larger diversity of resources in fast changing environments 1 , 24 , 61 , 62 . For instance, artificial bumblebee colonies containing individuals with different foraging profiles along a speed-accuracy trade-off have a more constant nectar collection rate than homogenous colonies 24 . Further investigation of the correlates of inter-individual behavioural and cognitive differences among members of a social group, such as bees, holds considerable promise for better assessing plastic collective responses and the adaptability of groups to stressful environmental conditions."
} | 3,133 |
39008501 | PMC11249216 | pmc | 708 | {
"abstract": "Neuroevolution is a promising approach for designing artificial neural networks using an evolutionary algorithm. Unlike recent trending methods that rely on gradient-based algorithms, neuroevolution can simultaneously evolve the topology and weights of neural networks. In neuroevolution with topological evolution, handling crossover is challenging because of the competing conventions problem. Mutation-based evolving artificial neural network is an alternative topology and weights neuroevolution approach that omits crossover and uses only mutations for genetic variation. This study enhances the performance of mutation-based evolving artificial neural network in two ways. First, the mutation step size controlling the magnitude of the parameter perturbation is automatically adjusted by a self-adaptive mutation mechanism, enabling a balance between exploration and exploitation during the evolution process. Second, the structural mutation probabilities are automatically adjusted depending on the network size, preventing excessive expansion of the topology. The proposed methods are compared with conventional neuroevolution algorithms using locomotion tasks provided in the OpenAI Gym benchmarks. The results demonstrate that the proposed methods with the self-adaptive mutation mechanism can achieve better performance. In addition, the adjustment of structural mutation probabilities can mitigate topological bloat while maintaining performance.",
"conclusion": "Conclusion In conclusion, this study has demonstrated the effectiveness of two improvements to the MBEANN algorithm. First, the self-adaptive mutation mechanism was integrated into MBEANN to automatically adjust the mutation step size, which was used to control the magnitude of perturbation in the weights and biases of the neural networks. Second, the structural mutation probabilities were normalized depending on the size of the neural networks to avoid overgrowth of the network topology. The results of this study showed that MBEANN with the self-adaptive mutation mechanism outperformed conventional algorithms by dynamically balancing exploration and exploitation. In addition, the proposed method with self-adaptive mutation and normalized structural mutation probabilities achieved better performance than the standard MBEANN with smaller network structures. In tasks that require complex structured neural networks, the normalized structural mutation probabilities are assumed to suppress topological evolution, which may degrade performance. Further research will explore novel approaches for topological optimization in MBEANN, with an emphasis on devising mechanisms to identify optimal network structures and guide topological evolution toward them.",
"introduction": "Introduction Neuroevolution is an approach within the field of machine learning that utilizes evolutionary algorithms to design artificial neural networks [ 1 – 3 ]. While recent trending methods commonly used in deep learning [ 4 , 5 ] and deep reinforcement learning [ 6 , 7 ] rely on gradient-based algorithms, neuroevolution optimizes neural networks through an evolutionary process involving natural selection and genetic variation. The main advantage of neuroevolution is that it utilizes population-based search methods, which enable broader exploration of the solution space and avoid being trapped in local optima. In addition, neuroevolution uses gradient-free optimization, making it applicable to problems in which the derivative information of the objective function is either unavailable or unreliable. Consequently, it can optimize or learn network structures, hyperparameters for controlling learning, or features of the algorithm itself, which are often challenging to address using gradient-based approaches [ 3 , 8 ]. The most straightforward approaches to neuroevolution encode only the weight values as genotypes. These approaches have been successfully applied to various applications, including the design of controllers for robots [ 9 , 10 ] and game-playing agents [ 11 , 12 ]. However, the parameters that determine the network structure, such as the number of layers, the number of nodes in each layer, and the arrangement of recurrent connections, are regarded as hyperparameters. Designers must specify the network structure before evolution begins, which requires domain knowledge, intuition, and experimentation. On the other hand, Topology and Weight Evolving Artificial Neural Networks (TWEANNs) [ 13 ] evolve both the neural network structure and the weights simultaneously. Examples of TWEANNs are GeNeralized Acquisition of Recurrent Links (GNARL) [ 14 ], Evolutionary Programming Network (EPNet) [ 15 ], Cellular Encoding (CE) [ 16 ], and Evolutionary Acquisition of Neural Topologies (EANT) [ 17 , 18 ]. GNARL is an algorithm based on evolutionary programming that evolves the topology and weights of recurrent neural networks [ 14 ]. Similarly, EPNet utilizes an evolutionary programming algorithm that employs mutation operators to evolve the topology; however, the weights are modified only by a hybrid training algorithm based on backpropagation and simulated annealing [ 15 ]. CE employs indirect encoding, which can evolve using a genetic programming algorithm [ 16 ]. The encoding process uses a graph-based representation, referred to as a grammar tree, to define instructions for constructing neural networks. EANT evolves neural networks from a minimal structure using two optimization loops: structural exploration, which develops a new neural network structure, and structural exploitation, which adjusts the weights of the neural networks [ 17 , 18 ]. NeuroEvolution of Augmenting Topologies (NEAT) [ 13 ] is the most popular and widely used TWEANN algorithm. In this algorithm, the evolution starts from a population of individuals with a minimal network structure and incrementally grows their topology. The NEAT algorithm uses a historical marker known as the innovation number, which tracks the ancestor of each gene. This number is assigned to a new gene whenever a node or connection is added through structural mutations. These innovation numbers are used to handle the speciation of the population and crossover between individuals with different topologies. To date, many successor algorithms, including state-of-the-art TWEANNs, have been developed based on NEAT [ 8 , 19 – 23 ]. In most cases, managing crossover in neuroevolution is challenging due to the competing conventions problem [ 1 , 13 , 24 , 25 ]. This problem arises when applying crossover between genotypes that are encoded differently but represent neural networks with similar behaviors. For example, when two parents have very different genotypes but exhibit high fitness, crossover between them may produce offspring with lower fitness. Moreover, in TWEANN algorithms, genome length varies depending on the network topology, which makes it challenging to define a consistent crossover between individuals. The NEAT algorithm mitigates the competing conventions problem using innovation numbers; however, crossover can still have disruptive effects [ 26 ]. As an alternative TWEANN approach, Mutation-Based Evolving Artificial Neural Network (MBEANN) [ 26 ] uses only mutations for genetic variation, omitting the use of crossover. In addition, structural mutations in MBEANN are designed not to affect the behavior of the phenotype to avoid fitness degradation. Moreover, the MBEANN algorithm introduces a novel encoding technique to define subnetworks within an individual as operons, and each operon grows independently throughout the evolution. Similar to NEAT, the individuals in MBEANN are initialized with a minimal structure. MBEANN outperformed NEAT in tasks such as the double-pole balancing problem [ 26 , 27 ] and the automatic design of controllers for robotic swarms [ 28 , 29 ]. Despite outperforming NEAT in various applications, MBEANN has limitations that must be addressed. One of the limitations is the difficulty in setting the hyperparameter known as the mutation step size, which controls the magnitude of changes in the weights and biases of neural networks. In MBEANN, structural mutations are designed to have small or no changes in the behavior of the neural networks. Consequently, exploration depends on the parameter mutation, which perturbs the weights and biases. The mutation step size in MBEANN is set to a fixed value [ 26 – 28 ]; therefore, a method for balancing the exploration-exploitation tradeoff is required. Another limitation is that the network structure tends to bloat in MBEANN. Individuals in MBEANN are expected to generate modularized subnetworks, each of which corresponds to an operon in the genotype. Each operon is designed to evolve independently without generating connections between two different operons. Moreover, each operon is structurally mutated with a constant probability; therefore, individuals with many operons have a higher expectation of applying structural mutations, leading to an exponential expansion in the network topology relative to the number of operons [ 28 , 30 ]. This rapid growth is beneficial when neural networks require many nodes and connections to perform a task. However, overgrown neural networks typically incur high computational costs. This study proposes two novel improvements to the MBEANN algorithm to overcome the existing limitations. Self-adaptation of the mutation step size. A self-adaptive mutation mechanism, which is often used in evolution strategies [ 31 , 32 ], is employed to automatically adjust the mutation step size during the evolution process. The mutation step size is coevolved within each individual, enabling automatic balancing between exploration and exploitation. Adjustment of structural mutation probabilities. The structural mutation probabilities are normalized depending on the size of the neural networks, which can expand the network topology at a speed similar to that of the traditional MBEANN in the early stages and gradually decrease the growth rate as the structure matures. This improvement aims to evolve high-performance neural networks with topologies smaller than those generated by the conventional MBEANN algorithm. In this study, these improvements are compared and discussed with the traditional MBEANN and NEAT using the HalfCheetah-v4 and Ant-v4 locomotion tasks provided in OpenAI Gym [ 33 , 34 ]. The remainder of this paper is organized as follows. The section “Mutation-based evolving artificial neural network” provides an introduction to MBEANN with details on the genetic encoding method and mutation operators. The section “Self-adaptive mutations for MBEANN” describes the proposed methods. The section “Experimental setup” explains the locomotion tasks provided in Open AI Gym and describes the algorithm settings. The section “Results and discussion” provides the results of the experiments with a discussion. Finally, the last section concludes the paper.",
"discussion": "Results and discussion Fifteen evolutionary trials were conducted for each algorithm. Figs 5 and 6 show the fitness transitions of the best individual within the population in HalfCheetah-v4 and Ant-v4, respectively. In HalfCheetah-v4, all MBEANN algorithms (MBEANN, SA-MBEANN, and SANP-MBEANN) exhibited relatively similar fitness transitions, with SA-MBEANN stagnating at slightly higher values. For each pair of algorithms, a statistical test was performed using data from the last generation to compare the fitness values to which each algorithm converged. The two-sided Mann-Whitney U test with Bonferroni correction was performed at a significance level of 0.05. Significant differences were found between NEAT and each MBEANN algorithm (Bonferroni-corrected p < 0.05; henceforth, all p -values are Bonferroni-corrected). No significant differences were observed between the MBEANN algorithms ( p > 0.05). In Ant-v4, SA-MBEANN and SANP-MBEANN obtained fitness values higher than those of the standard MBEANN and NEAT ( p < 0.05). No significant differences were observed between NEAT and MBEANN or between SA-MBEANN and SANP-MBEANN ( p > 0.05). When comparing SA-MBEANN and SANP-MBEANN, SANP-MBEANN obtained slightly lower fitness values than SA-MBEANN in HalfCheetah-v4; however, in both environments, they showed similar fitness transitions, as shown in Figs 5 and 6 . 10.1371/journal.pone.0307084.g005 Fig 5 Transitions of the fitness value of the best individual in HalfCheetah-v4. Each line represents the mean of the best fitness values over 15 trials, and the shaded regions around them indicate the standard deviations. 10.1371/journal.pone.0307084.g006 Fig 6 Transitions of the fitness value of the best individual in Ant-v4. Each line represents the mean of the best fitness values over 15 trials, and the shaded regions around them indicate the standard deviations. The algorithms used in this study also evolve the structure of neural networks. Figs 7 and 8 show the transitions of the number of nodes and connections of the best individuals in HalfCheetah-v4 and Ant-v4, respectively. The two-sided Mann-Whitney U test with Bonferroni correction was performed for each pair of algorithms using data from the last generation. The MBEANN and SA-MBEANN algorithms showed a similar tendency in the network structure transitions because they have the same structural mutation probabilities. No significant differences were observed between MBEANN and SA-MBEANN in either HalfCheetah-v4 or Ant-v4 ( p > 0.05). Notably, in Ant-v4, MBEANN and SA-MBEANN showed exponential growth in the network structure, as shown in Fig 8 . The growth of the network structures in the NEAT algorithm was very slow compared to the MBEANN family in both HalfCheetah-v4 and Ant-v4. Significant differences were found between NEAT and each MBEANN algorithm ( p < 0.05), except for the number of nodes between NEAT and SANP-MBEANN in HalfCheetah-v4 ( p > 0.05). This exception is because SANP-MBEANN automatically adjusts the structural mutation probabilities, which results in a smaller network topology. Indeed, the network structure of SANP-MBEANN grew more linearly than those of MBEANN and SA-MBEANN, as can be seen in Figs 7 and 8 . 10.1371/journal.pone.0307084.g007 Fig 7 Transitions of the network structure of the best individual in HalfCheetah-v4. (A) Transitions of the number of nodes in the individual, including 17 input and 6 output nodes. (B) Transitions of the number of connections. Each line represents the mean over 15 trials, and the shaded regions around them indicate the standard deviations. 10.1371/journal.pone.0307084.g008 Fig 8 Transitions of the network structure of the best individual in Ant-v4. (A) Transitions of the number of nodes in the individual, including 27 input and 8 output nodes. (B) Transitions of the number of connections. Each line represents the mean over 15 trials, and the shaded regions around them indicate the standard deviations. SANP-MBEANN is expected to grow similarly to the standard MBEANN in the early evolutionary generations, and the mutation probabilities gradually decrease as the topology expands. Therefore, in HalfCheetah-4, which has only 100 generations, no significant differences were observed between MBEANN and SANP-MBEANN, or between SA-MBEANN and SANP-MBEANN ( p > 0.05). However, in Ant-v4, significant differences were observed between MBEANN and SANP-MBEANN, and between SA-MBEANN and SANP-MBEANN ( p < 0.05). Because SA-MBEANN and SANP-MBEANN showed similar fitness transitions, as shown in Figs 5 and 6 , these results imply that SANP-MBEANN can mitigate bloats in network structures while maintaining performance. For further discussion, the individual that obtained the highest fitness value throughout the evolution was used for re-evaluation. The best-evolved individual from each algorithm was re-evaluated for 100 trials. Fig 9 shows the results of the re-evaluation. The two-sided Mann-Whitney U test with Bonferroni correction was performed for each pair of algorithms. In HalfCheetah-v4, the proposed methods SA-MBEANN and SANP-MBEANN obtained significantly higher performance than NEAT and MBEANN ( p < 0.05). There was no significant difference between SA-MBEANN and SANP-MBEANN ( p > 0.05). In Ant-v4, statistically significant differences were observed between all pairs of algorithms ( p < 0.05). Moreover, SA-MBEANN significantly outperformed the other algorithms in Ant-v4. Both SA-MBEANN and SANP-MBEANN significantly outperformed the standard MBEANN, as shown in Fig 9 . 10.1371/journal.pone.0307084.g009 Fig 9 Results of the re-evaluation for 100 trials using the best-evolved individuals. The MBEANN algorithms with self-adaptation of the mutation step size (SA-MBEANN and SANP-MBEANN) achieved higher performance. Figs 10 and 11 show the step size transitions of the best individuals in HalfCheetah-v4 and Ant-v4, respectively. As shown in Figs 10 and 11 , the mutation step size gradually decreased during evolution. The self-adaptive mutation mechanism facilitates a balance between exploration and exploitation. In particular, if larger changes in weights and biases lead to better solutions, individuals with larger mutation step sizes are more likely to survive. Conversely, individuals with smaller step sizes improve their performance based on the current solutions. As the evolutionary process shifts from exploration to exploitation, the mutation step size asymptotically decreases, enabling the algorithm to converge to an optimal solution. Therefore, it can be assumed that the self-adaptation mechanism automatically adjusts the mutation step size to refine the solutions, resulting in better performance. 10.1371/journal.pone.0307084.g010 Fig 10 Transitions of the step size of the best individual in HalfCheetah-v4. Each line represents the mean over 15 trials, while the shaded regions around them show the standard deviations. 10.1371/journal.pone.0307084.g011 Fig 11 Transitions of the step size of the best individual in Ant-v4. Each line represents the mean over 15 trials, while the shaded regions around them show the standard deviations. Overall, the MBEANN algorithms with the self-adaptive mutation mechanism (SA-MBEANN and SANP-MBEANN) achieved better performance than the standard MBEANN. In addition, the proposed approach with normalized mutation probabilities (SANP-MBEANN) showed better performance than the standard MBEANN with smaller neural network structures. Therefore, the proposed approach with normalized mutation probabilities (SANP-MBEANN) is the best choice when considering the computational cost. However, when comparing SA-MBEANN and SANP-MBEANN, SA-MBEANN achieved better performance in Ant-v4, as shown in Fig 9 . Considering the task difficulty, Ant-v4 might require more complex structured neural networks, which were not sufficiently evolved in SANP-MBEANN. The limitation of the proposed approach is the lack of optimization of the neural network structures. In the proposed method, the structure of neural networks continues to grow as evolution progresses. Therefore, if the evolutionary run is executed for many generations, the structure of the neural networks continuously grows over time, eventually reaching the limit of the computational cost. However, suppressing topological evolution may lead to low performance by constraining the behavior of neural networks. Finding an optimal structure is a challenging problem, not only for the proposed approach but also for other TWEANN algorithms, as well as for the recently emerged research topic known as neural architecture search [ 40 , 41 ]. Approaches for determining the optimal structure of neural networks and guiding topological evolution toward this optimal structure are left for future research."
} | 4,958 |
30876404 | PMC6420764 | pmc | 709 | {
"abstract": "Background The coordination of group behaviors in bacteria is achieved by a cell-cell signaling process called quorum sensing (QS). QS is an intercellular communication system, which synchronously controls expression of a vast range of genes in response to changes in cell density and is mediated by autoinducers that act as extracellular signals. Aliivibrio salmonicida, the causative agent of cold-water vibrosis in marine aquacultures, uses QS to regulate several activities such as motility, biofilm formation, adhesion and rugose colony morphology. However, little is known about either genes or detailed mechanisms involved in the regulation of these phenotypes. Results Differential expression profiling allowed us to define the genes involved in controlling phenotypes related to QS in A. salmonicida LFI1238. RNA sequencing data revealed that the number of expressed genes in A. salmonicida, ΔlitR and ΔrpoQ mutants were significantly altered due to changes in cell density. These included genes that were distributed among the 21 functional groups, mainly presented in cell envelope, cell processes, extrachromosomal/foreign DNA and transport-binding proteins functional groups. The comparative transcriptome of A. salmonicida wild-type at high cell density relative to low cell density revealed 1013 genes to be either up- or downregulated. Thirty-six downregulated genes were gene clusters encoding biosynthesis of the flagellar and chemotaxis genes. Additionally we identified significant expression for genes involved in acyl homoserine lactone (AHL) synthesis, adhesion and early colonization. The transcriptome profile of ΔrpoQ compared to the wild-type revealed 384 differensially expressed genes (DEGs) that allowed us to assign genes involved in regulating motility, adhesion and colony rugosity. Indicating the importance of RpoQ in controlling several QS related activities. Furthermore, the comparison of the transcriptome profiles of ΔlitR and ΔrpoQ mutants, exposed numerous overlapping DEGs that were essential for motility, exopolysaccharide production via syp operon and genes associated with tad operon. Conclusion Our findings indicate previously unexplained functional roles for LitR and RpoQ in regulation of different phenotypes related to QS. Our transcriptome data provide a better understanding of the regulation cascade of motility, wrinkling colony morphology and biofilm formation and will offer a major source for further research and analysis on this important field. Electronic supplementary material The online version of this article (10.1186/s12864-019-5594-4) contains supplementary material, which is available to authorized users.",
"conclusion": "Conclusion In this work we have shown that the master regulator LitR and the alternative sigma factor RpoQ regulate genes involved in motility, rugose colony morphology and biofilm formation in A. salmonicida . Our results indicate that RpoQ is an activator of flaA gene either directly or indirectly. Moreover, the positive activation of LitR on rpoQ results in reduced motility, repression of genes involved in adhesion (e.g., tad genes) and exopolysaccharide production via syp operon at HCD in A. salmonicida wild-type. These findings confirm that LitR and RpoQ regulate phenotypic traits related to QS together (dependent) and also independent of each other, where other environmental factors and genes are probably also involved. However further studies are needed to map the elements and factors affecting gene expression and influencing the observed phenotypes during different life cycles.",
"discussion": "Discussion Whole-transcriptome RNA sequencing analysis provides a powerful understanding of the gene expression patterns underlying the basic biology of the organism. In this work we studied the comparative transcriptome of A. salmonicida LFI1238, ΔlitR and ΔrpoQ mutants at low (OD 600 = 0.3) and high (OD 600 = 1.2) cell densities in SWT medium at 8°C. The SWT medium (2.5% salt concentration) and low temperature (8°C) were chosen as appropriated physiological conditions (similar to ocean environment) for A. salmonicida which is responsible for developing of cold-water vibriosis in Atlantic salmon at low seawater temperatures [ 39 – 41 ]. These conditions also favoured the development of several phenotypes (as motility, morphology and biofilm) related to QS in our ΔlitR and ΔrpoQ mutants in vitro [ 19 , 33 ]. The differentially expressed genes identified in this work provide a new insight to explain mechanisms related to QS such as motility, bioluminescence, wrinkled colony morphology, adhesiveness and biofilm formation. Changes in cell density impacts genes related to quorum sensing in A. salmonicida LFI1238 QS is known to be a cell density dependent mechanism allowing communication between bacteria and is regulated through master regulators, as VanT, HapR and LitR [ 28 , 42 , 43 ]. LitR was shown previously to regulate cryptic bioluminescence in A. salmonicida , where its inactivation resulted in less light production [ 44 ]. This led us to propose that cryptic bioluminescence is a high cell density dependen phenotype, where LitR is involved in its regulation. Herein, the transcriptome of A. salmonicida at HCD showed a significant upregulation of lux operon (Additional file 2 : Table S2), confirming that the alteration in gene experession of this operon is affected by changes in population. RpoS sigma factor aids in adaptation to environmental stress, mainly required for virulence, stress resistance and biofilm formation, additionally it has been shown to be required for full motility in some vibrios [ 45 ]. In this study rpoQ (RpoS-like sigma factor) was found to be upregulated in A. salmonicida at HCD compared to LCD. Moreover, the transcriptome of A. salmonicida demonstrated a downregulation in genes associated with motility and chemotaxis. This explains our previously obtained results, where the overexpression of RpoQ in the wild-type resulted in non-motile strains [ 33 ]. Hence, the expression of rpoQ leads to reduced motility in A. salmonicida at HCD. So why do A. salmonicida reduce their motility at HCD? It is believed that bacteria have different expression profiles during the different stages of life cycle. However, a complete life cycle of A. salmonicida is still unknown. But we assume that A. salmonicida similar to V. cholerae , is able to change from planktonic to biofilm life cycle which results in changes in genes expression required for motility and other functions [ 46 , 47 ]. The high cell density transcriptome presented in this study exhibits the activities of the late exponential phase (OD 600 = 1.2). During this phase nutrition accessibility is limited which favors the bacterial cells to enter the stationary phase and QS. Thus, at this time period the accumulation of autoinducers results in the expression of LitR, which in turn activates the rpoQ expression leading to activate regulators responsible for motility reduction, hence protecting the bacteria from excessive energy loss required to manage the motility apparatus. Additionally, it has been shown that A. salmonicida suppresses motility under the late stages of the host colonization (i.e., HCD) [ 48 , 49 ]. In contrast to HCD, at LCD we believe that the expression of motility genes in A. salmonicida are upregulated resulting in motile strains able to swim and colonize new host or environment. However the mechanism by which flagellar biosynthesis is controlled in A. salmonicida seems to be complex and will require further studies. LitR and RpoQ regulate genes vital for motility A. salmonicida is motile by nine polar flagella [ 50 ], where genes required for flagellum biosynthesis and flagellar motility are organized in different loci (Fig. 2 ) in a similar manner to A. fischeri [ 49 ]. The expression of genes involved in the synthesis of flagella in vibrios is tightly regulated through a complex hierarchy requiring the presence of regulatory proteins and the production of the flagellin monomer the basic component of bacterial flagellum, such as, FlaA [ 10 , 51 ]. RpoQ was shown to be a positive regulator of motility in A. salmonicida under our experimental conditions [ 33 ], and here we determine that the deletion of rpoQ resulted in a downregulation of several flagellar and chemotaxis genes, mainly flaA at both cell densities. Although A. salmonicida flagellar filament is composed of six flagellins (Fig. 2 ), it appears that the FlaA protein is mainly essential for motility and most likley regulated by RpoQ. The importance of FlaA for motility was reported in V. cholerae, where its deletion affected motility and thereby virulence [ 44 ]. Similarly, in A. fischeri the inactivation of flaA resulted in strains with reduced motility and symbiotic competence [ 52 ]. Likewise, a considerable importance of FlaA for motility was recently documented in A. salmonicida LFI1238, where the complete deletion of flaA resulted in 62% reduced motility at 8°C [ 53 ]. A similar reduction in motility was observed for the ΔrpoQ using the same temperature and salt concentration [ 33 ]. RpoQ is similar to other sigma factors that functions as a gene activator, and most probably activates a regulator of flaA gene . In V. cholerae it was show that flaA transcription is regulated by sigma factor 54 which depends on and requires an additional regulator, FlrC [ 54 , 55 ]. Thus, it is reasonable to speculate that RpoQ may work in the similar manner as V. cholerae by activating regulators responsible for motility, where in the ΔrpoQ mutant, flaA regulator is not activated resulting in decreased motility. The quorum sensing master regulator LitR, has been shown to be associated with motility in A. salmonicida similar to other bacteria [ 27 , 56 ]. The deletion of litR ( ΔlitR ) resulted in more motile strain than the wild-type [ 27 ]. This led us to conclude that LitR is a repressor of motility at HCD, where its deletion ( ΔlitR ) mimics the low cell density phenotype [ 27 ]. A similar conclusion was also applied to the role of RpoQ in motility [ 33 ]. However, ΔrpoQ transcriptome exhibited downregulation in motility genes regardless of growth phases. This proposes either that QS does not seem to be implicated in the RpoQ-dependent induction of motility and chemotaxis, or that rpoQ is critical for flagellar gene expression, where its deletion does not completely mimic the low cell density phenotype. In summary, these results indicate the importance of RpoQ in controlling the flaA gene which has a direct impact on the motility. Additionally RpoQ seems to tightly regulates several genes essential for flagellar assembly of A. salmonicida . Furthermore, RpoQ is believed to be a stress regulator in A. salmonicida similar to RpoS which may have the ability to switch between motile and non-motile states in response to physical or chemical changes in the environment. LitR and RpoQ repress genes associated with virulence Among the differentially expressed transcripts of ΔrpoQ and ΔlitR we were able to identify a number of significantly upregulated genes that may play an important role in virulence. These included the genes encoding adhesion and fimbrial attachment proteins also known as tad genes or tad operon. Tad loci is a widespread colonization island that is found in numerous pathogenic and non pathogenic bacteria including vibrios such as V. cholerae , A. fischeri, V.vulnificus and Vibrio parahaemolyticus ( V. parahaemolyticus ) [ 36 , 57 ]. The A. salmonicida genome encode a number of potential virulence factors. Among them is the Flp-type pilus (fimbrial –low molecular weight protein), which has high similarity to the Tad macromolecular transport system of Actinobacillus actinomycetemcomitans ( A. actinomycetemcomitans ) [ 36 ]. Tad operon is known to facilitate adhesion and, to play an important role in motility and biofilm formation [ 57 ]. Although the function of the tad operon was not investigated in detail in A. salmonicida and the inactivation of two tad genes ( VSAL_II0367 and VSAL_II0368 ) did not affect the architecture or amounts of biofilm formed [ 19 ], it is reasonable to assume that this widespread colonization island provides important functions for pathogenic bacteria (e.g., A. salmonicida ) in the form of colonization and adhesion. Our previous microarray analyzes on the ΔlitR mutant did not reveal any tad genes to be differentially expressed [ 19 ], although the adhesion of the ΔlitR mutants to the agar plates was observed [ 27 ]. In the study presented here, DEGs related to Tad locus in ΔrpoQ and ΔlitR yielded highly similar findings, where a number of tad genes were significantly upregulated. Whereas, the transcriptome of A. salmonicida wild-type at HCD revealed opposite results, where tad genes were downregulated. Thus, the increased expression level of LitR and RpoQ at HCD, leads to a repression of tad genes in A. salmonicida wild-type. This, proposes the importance of this colonization island at early stages of life cycle (i.e., LCD). Although evidence for the physiological role of this colonization island in Vibrionaceae is scant, recently a correlation between tad genes and phenotypes in V. vulnificus was found to be associated with biofilm formation, auto-aggregation and initial surface attachment to the host [ 58 ]. tad genes were also found to mediate adherence, colonization and micro-colony formation in other bacteria [ 59 – 61 ]. Hypothetically, these findings also can be considered in A. salmonicida, where the tad operon is mainly required for the initial surface attachment of the cells to the biotic surface and formation of micro-colonies and less necessary in the later stages of biofilm or infection. However, further investigations are needed to confirm this hypothesis. Biofilm formation and colony rugosity are low cell density phenotypes involving expression of syp The ability to form rugose colonies and biofilm are often correlated features in vibrios, which is generally associated with enhanced production of exopolysaccharides [ 21 , 25 , 62 ]. Similarly, in A. salmonicida colony rugosity and biofilm formation requires the expression of syp genes responsible for the production of EPS [ 19 , 33 ]. Our previous microarray analysis showed that the expression of 14 out of 18 syp genes was negatively regulated by LitR, where the majority, were genes significantly upregulated in the biofilm compared to the suspension [ 19 ]. However, the data obtained from the current work did not show significant upregulation of the 14 syp genes previously identified [ 19 ], except sypA and sypC genes, that showed to be differentially expressed at HCD. We know from our previous results that changes in medium composition affects the biofilm morphology [ 19 ], and here we assume that changes in some compounds of the SWT medium have affected the transcriptome of ΔlitR and resulted in less differentially expressed syp genes. In contrast to the ΔlitR transcriptome, the ΔrpoQ presented an upregulation among 13 out of 18 syp genes at HCD. We have previously observed what we refere to as a “late and weak” wrinkling colony morphology exhibited by ΔlitR compared to ΔrpoQ, which demonstrated an earlier and stronger rugosity in addition to a heavy and slimy extracellular matrix substance in the biofilm [ 33 ]. This led us to propose that LitR performs its activity on syp through RpoQ, where its expression leads to a strong syp repression. Moreover, the mature biofilm formation exhibited by ΔlitR was proposed to be a result of two independent processes where the first results in repression of syp via RpoQ while the second is independent of rpoQ and represses other biofilm matrix components. When three syp genes were inactivated separately in the rpoQ mutant no biofilm and no wrinckeled colonies were formed, and the ΔrpoQsyp double mutants behaved similar to the wild-type (Additional file 10 : Figure S1). However, the inactivation of the same syp genes in ΔlitR , resulted in some biofilm production using the same conditions [ 19 ]. Hence, the inactivation of syp genes in ΔrpoQ mutant inhibited colony rugosity and biofilm formation completely, which was not the case for the ΔlitR . Consequently, our results provide a clear evidence that the negative regulatory cascade from LitR to syp genes is operated through RpoQ in a cell density dependent manner. Why is RpoQ involved in regulating exopolysaccharide production \n via \n syp? The bacteria, whether it is in the host or in the aquatic environment, employes survival strategies, where sigma factors (e.g., RpoS or RpoQ) are believed to aid in adaptation to environmental stress such as osmotic shock and starvation [ 63 ]. Hence, for RpoQ to be involved in regulating this EPS locus ( syp operon) may suggest that this sigma factor may play an important role in environmental persistence protecting the bacteria under starvation and during infection of the host. We therefore believe that in addition to the negative regulatory cascade operated from LitR to syp genes (via RpoQ), rpoQ is also influenced by other genes and environmental factors leading to repression of syp in a pathway that remains unknown (Fig. 7 ). Fig. 7 Proposed model of QS and the possible LitR and RpoQ interaction in A. salmonicida. The expression of LitR signaling at high cell density represses motility, biofilm and activates transcription of RpoQ [ 19 , 27 ]. The increased level of RpoQ activity leads to strong repression on biofilm formation, rugose colony morphology, motility and adhesion, through a negative regulatory cascade on EPS producing genes (i.e., syp ), flagellar and tad genes, respectively. At low cell density the LitR is not activated, thereby RpoQ levels are low and not sufficient to repress either tad or syp genes, resulting in an upregulation leading to a strong adhesion to surface and thereby biofilm formation. However, the deletion of rpoQ results in reduced motility, where the regulation of flagellar genes maybe affected by other genes and environmental factor either dependent or independent of QS mechanism. Arrows and lines with bar end indicate positive and negative regulation respectively. Lines may also indirect direct or indicate pathways with several steps Even though the relationship between RpoQ and LitR is not well-studied in A. salmonicida, our current transcriptome and previous microarray data showed a positive regulation of LitR on rpoQ , confirming that RpoQ operates downstream of LitR in the QS regulatory hierarchy [ 19 ]. Furthermore, the overexpression of rpoQ in the ΔlitR mutant influenced phenotypes related to QS [ 33 ]. Consistent with the results demonstrated in A. fischeri , where the overexpression of RpoQ in ΔlitR mutant resulted in decreased motility [ 32 ]. Taken together, our data suggest a working model (Fig. 7 ) for how LitR and RpoQ work together in A. salmonicida , proposing that expression of genes in A. salmonicida is not always regulated by QS, and possibly involve other regulatory elements that act independently of the QS regulatory mechanism. Hence, the interaction between RpoQ and LitR and their roles in controlling motility, biofilm formation and rugose colony morphology, may be directly or indirectly regulated by RpoQ independent of LitR and vice versa. Additionally, we assume that RpoQ is regulated by other gene(s) and stress factors rather than LitR alone."
} | 4,934 |
27334443 | PMC4917839 | pmc | 710 | {
"abstract": "Memristive devices are promising candidates for the next generation non-volatile memory and neuromorphic computing. It has been widely accepted that the motion of oxygen anions leads to the resistance changes for valence-change-memory (VCM) type of materials. Only very recently it was speculated that metal cations could also play an important role, but no direct physical characterizations have been reported yet. Here we report a Ta/HfO 2 /Pt memristor with fast switching speed, record high endurance (120 billion cycles) and reliable retention. We programmed the device to 24 discrete resistance levels, and also demonstrated over a million (2 20 ) epochs of potentiation and depression, suggesting that our devices can be used for both multi-level non-volatile memory and neuromorphic computing applications. More importantly, we directly observed a sub-10 nm Ta-rich and O-deficient conduction channel within the HfO 2 layer that is responsible for the switching. This work deepens our understanding of the resistance switching mechanism behind oxide-based memristive devices and paves the way for further device performance optimization for a broad spectrum of applications.",
"discussion": "Discussion We have developed a Ta/HfO 2 /Pt memristive device with fast switching speed (≤5 ns), record high endurance (1.2 × 10 11 cycles), and reliable retention (extrapolated to be ≫10 years at 85 °C). We also achieved 24 non-volatile long-retention resistance states by controlling compliance current during DC programming, and over a million (2 20 ) potentiation and depression epochs using electrical pulse trains. The device performance suggests that our devices can be used for both memory and computing applications. More importantly, we studied the switching mechanism and attributed the switching to the composition modulation of a sub-10 nm Ta-rich O-deficient conduction channel by both anion and cation migration under electric field and thermal effect. The results strongly suggest that the reactive metal electrode in the oxide based memristors plays a much more important role than previously expected in determining the switching mechanism and device performance. We further built a model that successfully explains the excellent behavior of our device. This work broadens our understanding of the resistance switching mechanism behind oxide-based memristive devices and paves the way for further device performance optimization and applications."
} | 611 |
33790492 | null | s2 | 711 | {
"abstract": "Microbial community behavior is coupled to a set of genetically-regulated chemical signals that correlate with cell density - the quorum sensing (QS) system - and there is growing appreciation that the QS-regulated behavior of bacteria is chemically, spatially, and temporally complex. In addition, while it has been known for some time that different species use different QS networks, we are beginning to appreciate that different strains of the same bacterial species also differ in their QS networks. Here we combine mass spectrometric imaging (MSI) and confocal Raman microscopy (CRM) approaches to investigate co-cultures involving different strains (FRD1 and PAO1C) of the same species ("
} | 173 |
29986719 | PMC6036681 | pmc | 712 | {
"abstract": "Background The overreliance on dwindling fossil fuel reserves and the negative climatic effects of using such fuels are driving the development of new clean energy sources. One such alternative source is hydrogen (H 2 ), which can be generated from renewable sources. Parageobacillus thermoglucosidasius is a facultative anaerobic thermophilic bacterium which is frequently isolated from high temperature environments including hot springs and compost. Results Comparative genomics performed in the present study showed that P. thermoglucosidasius encodes two evolutionary distinct H 2 -uptake [Ni-Fe]-hydrogenases and one H 2 -evolving hydrogenases. In addition, genes encoding an anaerobic CO dehydrogenase (CODH) are co-localized with genes encoding a putative H 2 -evolving hydrogenase. The co-localized of CODH and uptake hydrogenase form an enzyme complex that might potentially be involved in catalyzing the water-gas shift reaction (CO + H 2 O → CO 2 + H 2 ) in P. thermoglucosidasius . Cultivation of P. thermoglucosidasius DSM 2542 T with an initial gas atmosphere of 50% CO and 50% air showed it to be capable of growth at elevated CO concentrations (50%). Furthermore, GC analyses showed that it was capable of producing hydrogen at an equimolar conversion with a final yield of 1.08 H 2 /CO. Conclusions This study highlights the potential of the facultative anaerobic P. thermoglucosidasius DSM 2542 T for developing new strategies for the biohydrogen production. Electronic supplementary material The online version of this article (10.1186/s12934-018-0954-3) contains supplementary material, which is available to authorized users.",
"discussion": "Discussion The redox potential and diffusion coefficient of molecular H 2 make it a key component of metabolism and a potent energy source for many microbial taxa [ 25 ]. The ability to utilize this energy source relies on the production of various hydrogenase enzymes, which power both the consumption and production of H 2 and inextricably couple H 2 to energy-yielding pathways such as acetogenesis, methanogenesis and respiration [ 26 , 43 ]. Our comparative genomic analysis revealed that P. thermoglucosidasius contains a unique hydrogenase compliment comprised of two uptake hydrogenases (group 1d and 2a) and one H 2 -evolving hydrogenase (group 4a). Evolutionary analysis showed that these hydrogenases are derived through three independent evolutionary events. This indicates that H 2 is likely to play a pivotal role in P. thermoglucosidasius metabolism and bioenergetics in the ecological niches it occupies. By contrast, members of the sister genus Geobacillus lack orthologous hydrogenase loci and, aside from P. thermoglucosidasius , only the group 1d and 2a uptake hydrogenases share orthology in one and three Parageobacillus spp., respectively, even though they are frequently isolated from the same environments. The group 4a H 2 -evolving hydrogenase of P. thermoglucosidasius is not found in any other members of the class Bacilli and is most closely related to those found in members belonging to the class Clostridia, particularly the family Thermoanaerobacteraceae . Furthermore, it forms an association with a CODH, which is found in common with a more restricted subclade of strict anaerobes within the family Thermoanaerobacteraceae . Our fermentation studies with P. thermoglucosidasius in the presence of CO showed that P. thermoglucosidasius grows efficiently when exposed to high concentrations of CO and that the CODH-group 4a hydrogenase complex can effectively couple CO oxidation to H 2 evolution, P. thermoglucosidasius can do so at a near-equimolar conversion. Furthermore, unlike other CO oxidizing hydrogenogenic bacteria, which are strict anaerobes, P. thermoglucosidasius is a facultative anaerobe capable of first removing residual oxygen from CO gas sources prior to producing H 2 via the water-gas shift reaction. The combination of these features makes P. thermoglucosidasius an attractive target for potential incorporation in industrial-scale production strategies of biohydrogen."
} | 1,028 |
36285622 | PMC10092175 | pmc | 713 | {
"abstract": "Abstract Scleractinian coral populations are increasingly exposed to conditions above their upper thermal limits due to marine heatwaves, contributing to global declines of coral reef ecosystem health. However, historic mass bleaching events indicate there is considerable inter‐ and intra‐specific variation in thermal tolerance whereby species, individual coral colonies and populations show differential susceptibility to exposure to elevated temperatures. Despite this, we lack a clear understanding of how heat tolerance varies across large contemporary and historical environmental gradients, or the selective pressures that underpin this variation. Here we conducted standardised acute heat stress experiments to identify variation in heat tolerance among species and isolated reefs spanning a large environmental gradient across the Coral Sea Marine Park. We quantified the photochemical yield ( F \n \n v \n \n /F \n \n m \n ) of coral samples in three coral species, Acropora cf humilis , Pocillopora meandrina , and Pocillopora verrucosa , following exposure to four temperature treatments (local ambient temperatures, and + 3°C, +6°C and + 9°C above local maximum monthly mean). We quantified the temperature at which F \n \n v \n \n /F \n \n m \n decreased by 50% (termed ED50) and used derived values to directly compare acute heat tolerance across reefs and species. The ED50 for Acropora was 0.4–0.7°C lower than either Pocillopora species, with a 0.3°C difference between the two Pocillopora species. We also recorded 0.9°C to 1.9°C phenotypic variation in heat tolerance among reefs within species, indicating spatial heterogeneity in heat tolerance across broad environmental gradients. Acute heat tolerance had a strong positive relationship to mild heatwave exposure over the past 35 years (since 1986) but was negatively related to recent severe heatwaves (2016–2020). Phenotypic variation associated with mild thermal history in local environments provides supportive evidence that marine heatwaves are selecting for tolerant individuals and populations; however, this adaptive potential may be compromised by the exposure to recent severe heatwaves.",
"conclusion": "5 CONCLUSIONS Coral populations in this study demonstrate extensive phenotypic variation in heat tolerance between distinct populations and across environmental gradients. We identified that thermal regimes are a clear driving force in heat tolerance, explaining spatial variation in heat tolerance among coral reef populations. The strong link between acute heat tolerance and the occurrence of marine heatwaves is evidence that coral populations are likely adapting or acclimatizing to both recent and long‐term thermal history in their local environment. However, decreased coral heat tolerance in response to recent severe heatwaves warrants concern for the potential limits to adaptation and acclimation of coral populations within ecologically relevant timeframes.",
"introduction": "1 INTRODUCTION Marine heatwaves have emerged as the principal threat to coral reef ecosystems (Oliver et al., 2018 ; Smale et al., 2019 ), driving mass coral bleaching events and resulting in extensive coral mortality throughout tropical oceans (Hughes, Kerry, et al., 2018 ; Lough et al., 2018 ). Recent bleaching events have demonstrated a clear link between heat accumulation and coral bleaching (Hughes et al., 2017 ), whereby photosynthetic symbionts (Symbiodiniaceae) disassociate from the coral host during heat stress (either prolonged or acute), threatening the health and survival of corals (Baker, 2003 ; Glynn, 1984 ). The increasing persistence of marine heatwaves exposes corals to temperatures near, or above, their upper thermal limits (Heron et al., 2016 ) and will continue to threaten coral reefs globally (van Hooidonk et al., 2016 ). Despite the growing concerns of coral bleaching, there remains limited understanding of how different species and individuals respond to heat stress or the ability for corals to adapt or acclimate to changing environmental conditions. Therefore, investigating the phenotypic and genotypic diversity that underpins heat tolerance in coral populations is critical to predict the capacity for corals to acclimate and/or adapt to marine heatwaves. Variation in bleaching susceptibility among coral species indicates there is considerable phenotypic variation in heat tolerance. This variability is largely driven by physiological trade‐offs associated with colony morphology and growth rates (Loya et al., 2001 ; van Woesik et al., 2011 ), heterotrophic feeding rates (Grottoli et al., 2006 ), or energy reserves (Grottoli et al., 2014 ). However, even within species, individual genotypes can exhibit variation in heat tolerance within the same environmental conditions (Barshis et al., 2013 ; Bay & Palumbi, 2014 ; Morikawa & Palumbi, 2019 ; Schoepf et al., 2015 ). Differences among genotypes are attributed to phenotypic plasticity (Oliver & Palumbi, 2011 ), underlying standing genetic variation of the coral host (Dixon et al., 2015 ; Drury, 2020 ; Fuller et al., 2020 ; Torda et al., 2017 ), and/or intraspecific variation in the symbiont community composition associated with individual colonies (Berkelmans & van Oppen, 2006 ; LaJeunesse et al., 2009 ). However, there is a paucity of data concerning the mechanisms or drivers of phenotypic variation in heat tolerance derived from standardised experimental approaches (Grottoli et al., 2020 ; McLachlan et al., 2020 ), in particular, those examining spatial variation in heat tolerance (Evensen et al., 2022 ). Marine heatwaves on coral reefs are not evenly distributed in time and space and are key drivers of local‐ and regional‐scale differences in coral community composition (Dietzel et al., 2021 ; Hughes, Anderson, et al., 2018 ; Oliver et al., 2018 ; Smale et al., 2019 ). Coral mortality associated with these events can result in strong selection for individuals with greater tolerance to heat stress (Palumbi et al., 2014 ; Sully et al., 2019 ). Therefore, heat tolerance in corals is expected to vary in relation to thermal exposure, influencing phenotypic diversity at the level of individual genotypes (Lundgren et al., 2013 ), fine‐scale microhabitats (Cornwell et al., 2021 ; Hoogenboom et al., 2017 ; Schoepf et al., 2015 ), and populations (Berkelmans & Willis, 1999 ; Coles et al., 1976 ; Dixon et al., 2015 ; Guest et al., 2012 ; Howells et al., 2016 ). Meanwhile, temporal variability in thermal gradients, such as annual temperature ranges, the rate of summer warming, the frequency of warming events, and prior exposure to heat stress mediate the thermal optimum and thermal range of corals across days, seasons, and years (Ainsworth et al., 2016 ; Jurriaans & Hoogenboom, 2020 ; Middlebrook et al., 2008 ). Overall, a complex interplay of spatial and temporal variation in environmental conditions are important determinants of upper thermal limits in corals and may lead to spatial variation in heat tolerance. Early studies of heat tolerance in corals used long‐term experiments (weeks to months) to simulate the accumulation of heat stress during natural bleaching events, establishing the conditions that trigger bleaching and identifying their thermal maxima (Coles et al., 1976 ; Humanes et al., 2022 ; Jokiel & Coles, 1990 ). More recently, acute heat stress assays have demonstrated the capacity to effectively establish relative thermal tolerance of corals over much shorter periods (Barshis et al., 2013 ; Palumbi et al., 2014 ). While acute heat stress assays do not mimic natural bleaching events, proof‐of‐principle experiments have identified that short‐term acute heat stress assays (7 h) are comparable with longer‐term (21‐day) heat stress assays in bleaching responses using dark‐adapted maximum quantum yield ( F \n \n v \n \n /F \n \n m \n ) as a physiological metric, but not chlorophyll a or Symbiodiniaceae densities (Evensen et al., 2021 ; Voolstra et al., 2020 ). Additional ground‐truthing has shown that estimates of absolute heat tolerance vary according to season and should be considered when comparing across studies. However, relative estimates of heat tolerance rankings among coral genotypes remain consistent regardless of seasonality (Cunning et al., 2021 ). Hence, short‐term acute heat stress assays provide a flexible and rapid approach to estimate heat tolerance for many individuals, populations and species, over much greater temporal and spatial scales than previously possible. To understand the drivers of heat tolerance and improve forecasting for how coral assemblages will respond to future marine heatwaves, we quantified the spatial patterns of heat tolerance in three scleractinian coral species ( Acropora cf humilis , Pocillopora verrucosa , and Pocillopora meandrina ) across nine widely separated populations in the Coral Sea Marine Park (CSMP), Australia. Coral populations spanned 7.7 degrees in latitude (860 km) along a 1.6°C gradient in maximum monthly mean (MMM) sea surface temperatures, providing a range of environmental conditions to investigate possible drivers of heat tolerance. The isolated nature of reefs in the CSMP makes it an ideal system to investigate the possibility of local adaptation in heat tolerance, where the distance between reefs is likely to limit gene‐flow between populations and where reefs are removed from other anthropogenic stressors (e.g., poor water quality). To investigate possible drivers of phenotypic variation in heat tolerance, we compared spatial patterns of relative heat tolerance against trends in sea surface temperatures and the occurrence of marine heatwaves, consistent with local adaptation mediated by changing environmental conditions.",
"discussion": "4 DISCUSSION 4.1 Acute heat stress experiments identify phenotypic variation in heat tolerance Identifying spatial mosaics of heat tolerance across climatic and disturbance gradients is key to understanding the adaptive potential of corals to the increasing frequency of marine heatwaves. To date, smaller reciprocal transplant experiments have identified genetic mechanisms of the coral host (Drury & Lirman, 2021 ; Kenkel et al., 2013 ) and symbiont community structure (Marhoefer et al., 2021 ) that influence thermotolerance and signify local adaptation to thermal regimes, but also indicate limits for corals to respond to temperatures outside of their local conditions (Howells et al., 2013 ). Building on these principles, standardised acute heat stress experiments have qualified as high‐throughput scans for phenotypic variation, successfully demonstrating that heat tolerance variation exists across coral nursery gardens in the Florida Keys (Cunning et al., 2021 ), thermally variable patch reefs across the Palau archipelago (Cornwell et al., 2021 ), and among microhabitats (Voolstra et al., 2020 ) and contrasting reef populations in the Red Sea (Evensen et al., 2022 ; Voolstra et al., 2021 ). The portability and automation of the acute heat stress experimental aquaria system (National Sea Simulator, AIMS) allowed us to quantify heat tolerance across a large spatial scale comprised of variable thermal history. Our findings provide further evidence that inter‐reef differences in thermal tolerance broadly correspond with localised differences in thermal exposure. Thus, providing evidence that coral populations may be locally adapted to climate history, as well as the frequency and severity of marine heatwaves they have been exposed to. 4.2 Phenotypic variation in heat tolerance among species within reefs While knowledge of the mechanisms that confer heat tolerance in reef‐building corals remains limited, experimental studies demonstrate the capacity for short‐term acclimation (DeCarlo et al., 2019 ; Howells et al., 2013 ) and long‐term adaptation in response to heat stress (Bay & Palumbi, 2014 ; Dixon et al., 2015 ; Drury, 2020 ; Drury et al., 2017 ; Kenkel & Matz, 2017 ). Such variability between species, particularly within the same environment, are typically associated with gene‐based adaptation (Fuller et al., 2020 ; Morikawa & Palumbi, 2019 ) and/or variation in symbiont community structure (Oliver & Palumbi, 2011 ). In our experiments, three coral species were exposed to the same local environmental and experimental conditions yet, exhibited variable ED50 thresholds ranging up to 0.7°C. In the case of two closely related Pocillopora species (Johnston et al., 2017 ), the differences in heat tolerance may be attributed to variation in heat tolerance among symbionts, as P. verrucosa and P. meandrina are highly specific in symbiont community selection (Turnham et al., 2021 ), attributed to vertical transmission of symbionts to offspring (Hirose et al., 2000 ). Heat tolerance differences in these symbiont species can influence the ability of the host to respond to heat stress changes (Manzello et al., 2019 ), depending on the heat tolerance potential of the symbiont itself. Phenotypic variation within A . cf humilis may also be attributed to variation in symbiont species, although we could not exclude cryptic host speciation in the CSMP, as these samples have not been genetically confirmed as one species. The question of species identification may also lend itself to the broad range of ED50 values for A . cf humilis compared to the relatively narrow range for both species of Pocillopora , for which species identification has been confirmed. These questions require additional genetic studies to fully disentangle species level patterns of heat tolerance for Acropora . 4.3 Spatial variation in heat tolerance Oceanic islands have served as model systems to evaluate the drivers of species richness, assembly rules of ecological communities and adaptive speciation, and provide insights into ecological and evolutionary processes (Borregaard et al., 2017 ; Santos et al., 2016 ). The geographic separation of reefs in the CSMP and distinct thermal histories may promote phenotypic variation within species and adaptation to local thermal regimes, where limited gene flow can reinforce processes of genetic drift and natural selection in spatially heterogeneous environments (Kawecki & Ebert, 2004 ; Savolainen et al., 2013 ). Of the 24 environmental variables measured, three thermal history metrics were identified as possible drivers of heat tolerance of reefs, driving responses more than latitude, sea surface temperature, depth and the 2020 marine heatwave. Notably, the frequency of mild heatwaves in a local environment was a key driver of increased relative heat tolerance in A . cf humilis . Populations harboring the most heat tolerant corals (e.g., Wreck Reef) experienced historically higher frequency of mild heatwaves over the past 35 years. Conversely, reefs which have evaded a high frequency of mild heatwaves (e.g., Herald Reef) tended to harbor assemblages of less tolerant individuals. For corals, a critical tipping point for bleaching‐induced mortality occurs when accumulated heat exceeds 3–4°C‐weeks, indicating that DHW values above this threshold can influence population dynamics and the relative frequency of traits associated with heat tolerance (Hughes, Kerry, et al., 2018 ). A . cf humilis displayed a strong relationship to exposure to mild heatwaves, which may be linked to this species' higher sensitivity to heat stress. However, it is worth noting that the relationship between ED50 and mild heatwaves was predominantly driven by A . cf humilis at Wreck Reef. In addition to mild heatwaves, a longer return time between severe heatwaves above or equal to 6°C‐weeks increased acute heat tolerance, likely allowing sufficient time for populations to recover from lasting effects of severe heatwaves. The beneficial selection of mild heatwaves, as well as a longer return time between heating events, may be hampered by recent severe heatwaves over the past 5 years, as indicated by the strong effect of recent maximum DHW on acute heat tolerance (i.e., average maximum DHW between 2016 and 2020). The effect of recent severe marine heatwaves over this period is an indication that corals may not be able to keep up with the pace of rapidly reoccurring marine heatwaves. Rapid environmental change, such as three mass bleaching events in 5 years, does not support rates of phenotypic plasticity for most individuals and species (Lindsey et al., 2013 ). Further, the lack of correlation between severe heatwaves (i.e., number of DHW events exceeding 6 or 9°C‐weeks) and higher heat tolerance, suggests significant limits to adaptation potential in corals above a threshold where bleaching‐induced mortality occurs (Ainsworth et al., 2016 ). The lack of improved prediction may be due to severe heatwaves causing increased coral mortality of all genotypes, rather than acting as a selective pressure. A similar phenomenon was observed during the back‐to‐back bleaching events of 2016 and 2017 on the GBR and Coral Sea, where a reduction in the incidence of bleaching in 2017 was attributed to extensive bleaching‐induced mortality of corals in 2016, leaving few corals left to bleach in severely affected reefs (Harrison et al., 2019 ; Hughes, Kerry, et al., 2018 ). Thus, mild heatwaves and local conditioning to MMM provide environmental pressure that is strong enough to select for heat tolerance but not too strong to decimate entire populations. A case can be made to use absolute ED50 values to consider the effects of MMM on ED50 and its interaction with other environmental drivers. To do so, we would suggest having consistent temperature treatments across all sampled populations rather than standardised to local MMM (but see SOM for details). 4.4 Global comparisons of ED50 thresholds The relative and absolute ED50 values allow for direct comparisons within and between studies, overcoming a major challenge in comparing heat stress experiments (Grottoli et al., 2020 ; McLachlan et al., 2020 ). There are several applications for coral acute heat stress data. A few examples include the ability to rank heat tolerance among individuals, populations, and species; to investigate genotype–phenotype associations to identify molecular signatures of heat tolerance; and to explore cross‐study comparisons of heat tolerance thresholds of corals. Coral populations experiencing historically higher temperature regimes are generally less susceptible to bleaching than conspecifics in other regions (Howells et al., 2016 ). However, the absolute ED50 thresholds for P. verrucosa in the CSMP were very similar to conspecifics in the Red Sea (Absolute ED50/ED50: CSMP = 36.1°C; Red Sea = 36.0°C) (Evensen et al., 2022 ), despite the hotter conditions in the Red Sea, 1.3°C above those in the CSMP. Interestingly, P. verrucosa in the CSMP maintained overall higher relative ED50s (i.e.,°C above local MMM temperatures) than P. verrucosa in the Red Sea by 1.2°C (Relative ED50/ED50: CSMP = 7.7°C; Red Sea = 6.3°C) when comparing averages across each reef to characterise a region. The relative tolerance of corals in the CSMP compared with corals in other regions may indicate that corals in the CSMP are not living as close to their thermal limits as predicted. Potentially, the high disturbance history of the past three decades, layered with episodic heatwaves experienced in the last 5 years in the CSMP (Harrison et al., 2019 ) has selected for more heat tolerant individuals. Across a latitudinal gradient, P. verrucosa in both this study and Evensen et al. ( 2022 ) maintained higher relative thermal thresholds in high latitude reefs compared to low‐latitude reefs, supporting previous evidence that high latitude reefs may harbor higher heat tolerance and therefore serve as spatial refugia from bleaching events (Osman et al., 2018 ). These comparisons provide valuable insight to identify reefs and regions of high or low tolerance, albeit the comparisons across variable aquaria systems (e.g., lights, flow, and sampling time‐points) may confound these interpretations and should also be considered."
} | 5,038 |
34770810 | PMC8588144 | pmc | 715 | {
"abstract": "Recently, superhydrophobic surfaces with self-cleaning ability have attracted broad research interest due to their huge potential in daily lives and industrial applications, but the use of fluorinate, toxic organic compounds, and expensive feedstocks make superhydrophobic materials a great challenge in practical application. In this study, we present a facile dip-coating strategy to prepare superhydrophobic coatings with self-cleaning properties based on a non-fluorine and non-toxic system by using eco-friendly corn straw as raw material. During this process, aromatic carbon particles with rough hierarchical structures were prepared firstly via a simple fast pyrolysis process, followed by modification with polydimethylsiloxane (PDMS) in absolute ethanol solvent to decrease the surface free energy. Research shows these natural straw-derived carbons display a microstructure of several protrusions which is similar to the lotus leave’s and the resulted coatings exhibit an outstanding superhydrophobic property with a static water contact angle (WCA) of 151.67 ± 1.36 degrees. In addition, the as-prepared coatings possessed excellent self-cleaning performance: no contaminations were observed on the surfaces after examining with sludge, calcimine, water, and common liquids such as tea, milk, soybean milk as well as ink, which have a broad range of potential application in the field of antifouling, waterproofing, and anticorrosive.",
"conclusion": "4. Conclusions In this work, biochar-based fluorine-free and non-toxic superhydrophobic coatings with self-cleaning ability were fabricated on various substrates via facile pyrolysis carbonization and dip-coating strategies from low-cost, renewable, and widely distributed corn straw. The reported superhydrophobic coatings exhibited a high static water contact angle of 151° in air and excellent anti-fouling and nonwetting performance in self-cleaning experiments. In addition, we studied the effect of pyrolysis temperature on the superhydrophobic performance of PDMS/CPs coatings and the temperature of 600 °C is determined to be the optimal value. In summary, the facile, low-cost, and environmentally friendly fabrication strategy of the PDMS/CPs coatings with an outstanding self-cleaning property is expected to promote the efficient utilization of agricultural resources and have great potential applications in the self-cleaning field.",
"introduction": "1. Introduction In recent years, with environmental pollution intensifies, there are increasingly suspended particles and dust in the air, which leads to the increasing possibility for the surfaces exposed to the ambient atmosphere such as ships, automobile windshields, external wall and glass of high-rise buildings, windmill’s sails, etc. to be contaminated, which results in not only increasing cleaning cost but the difficulty to clean [ 1 , 2 , 3 , 4 ]. Therefore, it is very imperative to explore a contamination-free surface to improve dirt tolerance. Superhydrophobic surfaces with self-cleaning properties have been proposed as a terrific way to turn contamination-free surfaces into reality. Superhydrophobic materials surfaces are defined as a surface that displays a water contact angle greater than 150° along with roll-off angles less than 10° [ 5 ]. The performance of superhydrophobic surfaces is covered by two critical parameters: low surface energy and rough micro-/nanostructures. Correspondingly, there are two main routes to fabricate superhydrophobic surfaces at present: (a) creating rough micro-/nanostructures on hydrophobic surfaces (water contact angle larger than 90 degrees), and (b) chemical modification of a micro-/nanostructured substrate with hydrophobic modifier [ 6 , 7 ]. For the realization of low surface energy, a large number of chemicals have been reported, such as fluorine-containing compounds, long-chain organosilanes, and fatty acids. With tremendous efforts devoted to the superwettability mechanism of organisms in nature including water striders legs [ 8 ], butterfly wings [ 9 ], rose and sunflower petals [ 10 ], lotus plant leaves [ 11 ], gecko foot [ 12 ], etc., a wide range of methods and techniques have been developed to texture hierarchical micro-/nanosized structures which can be divided into two design strategies: one is fabricated from larger-scale materials and the another is manufactured from atomic or molecular scale materials [ 13 ]. In more detail, these methods and techniques involve lithography [ 14 ], diamond cutting [ 15 ], etching [ 16 ], 3D printing [ 17 ], layer-by-layer assembly [ 18 ], self-assembly [ 19 ], sol-gel [ 20 ], and others. Physical nanoparticles are an effective strategy to provide hierarchical micro-/nanostructure. Currently, various kinds of synthesized nanoparticles are used in superhydrophobic performance. For instance, Zhong et al. reported a self-cleaning superhydrophobic surface with excellent self-cleaning properties and water repellency by extracting lignocellulose nanofibrils from wheat straw followed by modification of 1H, 1H, 2H, 2H-perfluorooctyltriethoxysilane [ 21 ]. Chen et al. fabricated self-cleaning and superhydrophobic surfaces on various substrates through coating with calcium carbonate nanoparticles modified with perfluorooctyltriethoxysilane [ 22 ]. Additionally, Nine et al. prepared graphene-based superhydrophobic coatings with favorable self-cleaning and anti-corrosion ability via mixing diatomaceous earth, synthesized graphene oxide, and TiO 2 in THF solution followed by modifying with PDMS [ 23 ]. Lv et al. created a robust and nonfluorinated superhydrophobic coating by using silicon resin as a low surface energy modifier and reinforced fillers, carbon nanotubes/graphene as a rough structure which displayed excellent self-cleaning and anti-wear performance [ 24 ]. However, it could be concluded from the mentioned works that raw materials and the processes used for conferring superhydrophobicity to natural polymer involve some shortcomings, such as, high feedstock costs, complicated preparation process, time-consuming reaction, fluorinated compounds, and toxic organic solvents, causing negative effects to human health and natural environments [ 25 , 26 ]. Thus, the fabrication of new environmentally friendly and biodegradable superhydrophobic coatings is extremely crucial to the sustainable development of humanity and nature. As a kind of widely distributed, low-cost, biodegradable, and environmentally friendly resource, biomass materials such as straws, falls, trees, etc., have attracted extensive research interests in an effort to fabricate superhydrophobic materials. Carbon nanoparticle is a carbonaceous material that can be easily fabricated through the pyrolysis process of biomass waste. Therefore, in this article, we present a straightforward dip-coating method to prepare superhydrophobic coatings on different substrates based on easily generated carbon particles by utilizing low-cost materials and non-fluorinated chemicals without any toxic reagents and expensive equipment, exhibiting great potential for large-scale production. The obtained coatings not only displayed high water repelling properties but also showed excellent self-cleaning performance. Above all, this work provides an economical, facile, and eco-friendly approach to prepare superhydrophobic self-cleaning surfaces which have great potential application in water collection, antifouling, oil-water separation, anticorrosion, and anti-icing. On the other hand, for the resource shortage reason, the present strategy provides a novel insight into the further utilization of biomass waste.",
"discussion": "2. Results and Discussion The preparation process of superhydrophobic coatings is illustrated in Figure 1 . This strategy possesses a simple two-step efficient process. First, amorphous carbon nanoparticles were prepared via fast pyrolysis of corn straw at different temperatures to form a rough structure. Subsequently, superhydrophobic surfaces were fabricated by dip-coating the ethanol-based suspension of carbon particles/PDMS composites onto various substrates. It is well known that the wettability of surfaces is governed by the chemical composition and micro/nanostructure [ 27 ]. To further understand the superhydrophobicity mechanism of the as-prepared samples, a scanning electron microscope (SEM) was used to characterize the surface microstructure of CPs carbonized at different temperatures, as shown in Figure 2 . Clearly, the micromorphology of CPs is shown as rod-like structures with several protrusions. It can be concluded that the specific area and the roughness of CPs samples are noticeably enlarged. As Figure 2 a–c shows, with increasing pyrolysis temperature, CPs display an increase in the number of rod-like structures. Furthermore, it can be seen that there is a larger number of punctiform protrusion in Figure 2 e which is a structure similar to the lotus leaf [ 28 ] on the surface of CPs-600 compared to CPs pyrolyzed at 400 ℃ and 800 ℃. Figure 3 displays the surface morphologies and 3D contour images of PDMS/CPs-400, PDMS/CPs-600, and PDMS/CPs-800 coatings. As exhibited in Figure 3 a–f, porous coatings were formed after curing owing to the stacking arrangement of nanoparticles. It can be seen from Figure 3 a–c that there are more porous coatings on the surface of PDMS/CPs-400 coating, which might be caused by the incomplete pyrolysis process at the lower temperature of 400 °C. Compared with SEM images, the surface roughness can be seen more intuitively from the 3D contour images. It is obvious from Figure 3 g–i that non-uniform protrusions are visible on the surface of PDMS/CPs coatings. For PDMS/CPs-400 coating, there are almost no apparent protrusions observed, indicating a relatively smooth surface morphology. The morphologies of the protrusive microstructure are improved on the surface of the PDMS/CPs coating with the treatment temperature increased to 600 °C. Figure 3 i clearly shows that there are more red areas and the blue area is almost invisible. Simultaneously, fewer protrusions can be observed on the surface of the PDMS/CPs-800 coating. This suggests that a flatter surface was formed under the pyrosis temperature of 800 °C. The roughness average (Ra), a most frequently used surface topography parameter, was employed to probe the impact of pyrosis temperature and the results of measurements for the resulting coatings are shown in Table 1 . In Table 1 , the roughness values, the mean values, as well as the standard deviations are given for each coating. It is found that the Ra mean value of PDMS/CPs-400 coating is 2.87 μm. With the pyrosis temperature increased to 600 °C, Ra reached the largest value of 3.34 μm. However, when the pyrosis temperature increased to 800 °C, the value of Ra decreased to 2.29 μm. The results of the surface morphologies analysis reveal that as pyrolysis temperature increased, the surface roughness exhibit first an increasing trend and then decline, which corresponds to the analysis results of the CPs microstructure. It can be concluded that the changes of surface morphologies consistent with the microstructure of CPs and the PDMS/CPs-600 coating could better prevent the water from contacting the substrate. To investigate the evolution of the chemical structure of raw corn straw fibers, carbonized particles, and coatings modified with PDMS, the FT-IR spectroscopy was taken over the wavenumber range of 400 to 4000 cm −1 and the results are shown in Figure 4 . As indicated, the surface components were changed after carbonization and modification. Besides, it is obvious to find that the IR spectra of straw carbon-based materials fabricated at different temperatures show similar absorption peak features, suggesting the biochars have already formed at the low temperature of 400 °C. The broad band at approximately 3413 cm −1 corresponds to the O–H stretching vibrations of the carboxyl and phenolic hydroxyl groups—it is consistent with the literature report [ 29 ]. However, the peak intensity of the PDMS/CPs-600 product at around 3413 cm −1 was weak which indicating a decrease in the number of O−H groups and the formation of hydrophobicity. The adsorption peaks appearing at 2996 cm −1 and 2908 cm −1 belong to the stretch vibration of the -CH 3 group. The sharp adsorption band around 1612 cm −1 is vibration from C=O and C=C stretching, implying the existence of an aromatic group [ 30 ]. Moreover, the adsorption peak of the raw straw at 1066 cm −1 assigned to the C–O group disappeared after pyrolysis which indicates that the crystallized region of the corn fibers was disrupted and the polysaccharides in the straw had completely degraded at the temperature of 400 °C [ 31 ]. In addition, the FT-IR spectra of modified carbon particles (PDMS/CPs-600) show the new adsorption peaks near 1253 cm −1 , 1066 cm −1 , and 794 cm −1 , which are assigned to the C−H group in Si−CH 3 , Si−O−Si asymmetric, and symmetric stretching vibration [ 32 , 33 ], implying that PDMS was successfully grafted onto the CPs. As we can see from the result of the SEM analysis, the structure of straw carbon particles formed at different temperatures differs from each other. To understand explicitly these changes, XRD analysis was employed to study the crystal structures of CPs pyrolyzed at different temperatures. As depicted in Figure 5 a, two characteristic broad peaks at 2θ = 23.7° and 2θ = 43.4° occur in all cases which correspond to the (002) plane and the (100) plane of disordered graphitized carbon, respectively [ 34 ]. The peak exhibits at the low angle of 23.7° suggesting the existence of a large number of amorphous carbon particles [ 35 ] which might be due to the direct carbonization without any other chemical treatment and this is in accordance with published work. In addition, the formation of amorphous carbon in the process of pyrolysis indicates damage to the cellulose crystal structure. Moreover, the presence of a diffraction peak at around 23.7° (002) of each sample verifies the presence of different micropore-wall structural units [ 36 ]. A close examination of the diagram shows that the peaks near 23.7° shift slightly to the lower angle with the increasing pyrolysis temperature, indicating that the interplanar spacing of CPs increases. As concluded from these results, the change of pyrolysis temperature has a significant influence on the crystal structure of CPs. For further structural characterization, CPs-400, CPs-600, and CPs-800 were characterized by a Raman microspectrometer. As exhibited in Figure 5 b, there are two relative peaks of CPs at around 1324 cm −1 and 1598 cm −1 corresponding to the D bond of amorphous carbon and G bond of Graphite carbon, respectively [ 37 , 38 , 39 ]. It is found that the intensity ratios (I D /I G ) for CPs-400, CPs-600, and CPs-800 are 0.646, 0.897, and 0.988 which shows a low degree of graphitization. It also can be seen that the value of I D /I G increased gradually with increasing heat treatment temperature. This was mainly caused by the increasing surface defects of CPs. Meanwhile, the CPs-800 sample possessed the highest I D /I G value against other samples which indicates a higher extent of disorder in the CPs [ 40 ]. The water contact angle (WCA) test is important to determine the wettability of the as-prepared coatings. The results are shown in Figure 6 a. During this process, the glass samples prepared from different coating solutions were taken as the test object. As can be seen in Figure 6 a, the static water contact angles vary from 146.56 ± 2.14° to 151.67 ± 1.36° and the PDMS/CPs-600 coating exhibited the biggest contact angle value of 151.67 ± 1.36°, indicating its outstanding superhydrophobicity. Moreover, the WCA of PDMS/CPs-400 and PDMS/CPs-800 coatings were measured to be 147.96 ± 1.36° and 146.56 ± 2.14°, respectively, which shows good hydrophobicity of those two surfaces. All these results demonstrated that the performance of non-wettability increased with an increase in carbonization temperature, and the carbonization temperature of 600 °C formed the best hydrophobic property. In order to better understand the superhydrophobic performance of PDMS/CPs-600, the measurements of sliding angle (SA) and contact angle hysteresis were carried out at ambient temperature with a 2 μL water droplet, as shown in Figure 6 b–e. The obtained SA value is 8° lower than 10°and hysteresis angle is 4.1 ± 2°, indicating an outstanding superhydrophobicity of PDMS/CPs-600 coating. The wettability behavior of the original glass slide, glass coated with PDMS, and CPs-600 coated onto a glass surface were also evaluated. As shown in Figure 6 a, the original glass exhibited hydrophilic with a WCA value lower than 90 [ 27 ]. After introduction of PDMS, the hydrophilic original glass came to be hydrophobic (WCA = 124.44 ± 1.26°), indicating the excellent inherent hydrophobicity of PDMS. The water contact angle of PDMS/CPs-600 is shapely higher than PDMS and CPs-600, which suggests that PDMS was successfully grafted onto the surface of CPs. This result is consistent with the FT-IR analysis. Meanwhile, as seen in Figure 7 , the dropped water was absorbed immediately by raw straw fibers, indicating hydrophilicity of raw material, which is due to the existence of numerous hydrophilic groups. By contrast, the water CA of CPs without modification is approximately 132°, which might be due to the decline in the number of polar functional groups of corn straw after pyrolysis at high temperature, as shown in the infrared spectrum. A very important result that can be drawn from the above discussion and SEM analysis is that the micro/nanostructure is one of the key factors of superhydrophobic performance. In order to demonstrate the self-cleaning properties of the superhydrophobic coatings, various self-cleaning tests were carried out on the surface of PDMS/CPs-600 under room temperature. In these experiments, sludge and white wall ash were treated as contaminants, meanwhile, the milk, ink, tea, and soybean milk were used as common household liquids. When the as-prepared coating material was taken out after soaking in the sludge, as expected, there were no visible contaminants on the superhydrophobic surface ( Figure 8 a, Supplementary Video S1 ), which might be caused by the small adhesion force of dirt on the coated surface [ 41 ]. However, the uncoated sample was contaminated obviously during a similar process—shown in Supplementary Video S1 . Coated and uncoated substrates comparison showed an excellent antifouling property of the as-prepared coating. As shown in Figure 8 b, the water was dropped on the superhydrophobic surface that was contaminated by white wall ash and the droplets rolled off rapidly with the dust without wetting the coated slide, illustrating an excellent dirt-removal property of the as-prepared coatings. Figure 8 c shows that a silver mirror phenomenon on the superhydrophobic coating surface was observed when it was immersed in water, and then it floated on the surface of the water once it lost the external force, furthermore, the coated slide retained its non-wetting property after being moved out from the water. As shown in Figure 8 d, different liquid droplets, simulated as real dirt conditions, are spherical in shape on the surface of the superhydrophobic coating, thus, the modified surfaces possess excellent antifouling performance. Besides that, Figure 8 e shows an optical photograph of a jet of water bounced off and slipped off without any water spread out on the coated surface (also shown in Supplementary Video S2 ), conversely, the water droplets remain on the uncoated substrate and leave the surface in the form of water flow ( Supplementary Video S2 ), indicating its outstanding waterproof ability. All these experimental results reveal that the corn straw carbon-based superhydrophobic surfaces modified with PDMS have great potential application in the fields of self-cleaning, anti-fouling, anti-corrosion, water collection, and anti-icing. On the basis of these results, the superhydrophobic mechanism of the as-prepared coatings with self-cleaning ability was proposed ( Figure 9 ). The microstructure of CPs pyrolyzed at the temperature of 600 °C is a kind of rod-like structure with several protrusions. Moreover, the results of the WCA measurement showed that superhydrophobic surfaces fabricated by CPs pyrolyzed at 600 °C have the highest value. Therefore, we can conclude that the structure’s material plays an important role in hydrophobic performance. Besides, the PDMS used in the superhydrophobic coatings preparation procedure is a hydrophobic polymer with advantages of outstanding chemical stability, transparency flexibility bio-compatible [ 42 , 43 , 44 ], etc. When PDMS was mixed with CPs in an ethanol solution, the crosslinked PDMS film formed on the surface of carbon particles which effectively decreased the surface energy of PDMS/CPs coatings [ 45 ]."
} | 5,261 |
39264745 | PMC11459164 | pmc | 717 | {
"abstract": "Modern molecular microbiology elucidates the organizational principles of bacterial biofilms via detailed examination of the interplay between signaling and gene regulation. A complementary biophysical approach studies the mesoscopic dependencies at the cellular and multicellular levels with a distinct focus on intercellular forces and mechanical properties of whole biofilms. Here, motivated by recent advances in biofilm research and in other, seemingly unrelated fields of biology and physics, we propose a perspective that links the biofilm, a dynamic multicellular organism, with the physical processes occurring in the extracellular milieu. Using Bacillus subtilis as an illustrative model organism, we specifically demonstrate how such a rationale explains biofilm architecture, differentiation, communication, and stress responses such as desiccation tolerance, metabolism, and physiology across multiple scales—from matrix proteins and polysaccharides to macroscopic wrinkles and water-filled channels."
} | 253 |
30195437 | null | s2 | 718 | {
"abstract": "The diversity and number of species present within microbial communities create the potential for a multitude of interspecies metabolic interactions. Here, we develop, apply, and experimentally test a framework for inferring metabolic mechanisms associated with interspecies interactions. We perform pairwise growth and metabolome profiling of co-cultures of strains from a model mouse microbiota. We then apply our framework to dissect emergent metabolic behaviors that occur in co-culture. Based on one of the inferences from this framework, we identify and interrogate an amino acid cross-feeding interaction and validate that the proposed interaction leads to a growth benefit in vitro. Our results reveal the type and extent of emergent metabolic behavior in microbial communities composed of gut microbes. We focus on growth-modulating interactions, but the framework can be applied to interspecies interactions that modulate any phenotype of interest within microbial communities."
} | 246 |
25257018 | PMC4263508 | pmc | 719 | {
"abstract": "Quorum signaling (QS) describes how bacteria can use small signaling molecules (autoinducers) to coordinate group-level behaviors. In Vibrio fischeri, QS is achieved through a complex regulatory network that ultimately controls bioluminescence, motility, and host colonization. We conducted a genetic screen focused on qrr1 , which encodes a small regulatory RNA that is necessary for the core quorum-signaling cascade to transduce autoinducer information into cellular responses. We isolated unique mutants with a transposon inserted into one of two genes within the syp locus, which is involved in biofilm formation. We found that overexpression of sypK , which encodes a putative oligosaccharide translocase, is sufficient to activate qrr1 , and, in addition, this effect appears to depend on the kinase activity of the sensor LuxQ. Consistent with the established model for QS in V. fischeri , enhanced expression of qrr1 by the overexpression of sypK resulted in reduced bioluminescence and increased motility. Finally, we found that induction of the syp locus by overexpression of sypG was sufficient to activate qrr1 levels. Together, our results show how conditions that promote biofilm formation impact the quorum-signaling network in V. fischeri , and further highlight the integrated nature of the regulatory circuits involved in complex bacterial behaviors.",
"introduction": "Introduction Quorum signaling (QS) describes the process that enables a bacterium to sense and respond to other bacteria (Fuqua et al. 2001 ; Ng and Bassler 2009 ). The cell-signaling systems associated with QS depend on the synthesis and detection of signaling molecules, called autoinducers. For many bacterial species, these QS systems enable the coordination of population-level responses through gene regulation. Because autoinducer concentrations are often proportional to cell density, the responses to QS are also traditionally characterized according to cell density. However, this correlation can be disrupted by additional signaling components that occur downstream of the autoinducer receptor(s) within the regulatory network. Therefore, studies aimed to identify such inputs are critical for understanding how QS systems function in nature. Vibrio fischeri is a marine bacterium that uses QS to regulate a multitude of cellular processes, including bioluminescence, motility, and colonization of its natural host, the Hawaiian bobtail squid, Euprymna scolopes (Nyholm and McFall-Ngai 2004 ; Miyashiro and Ruby 2012 ; Stabb and Visick 2013 ; Verma and Miyashiro 2013 ). The LuxR-LuxI QS system directly regulates the lux genes, which encode the light-producing enzyme luciferase and several proteins involved in light production and other activities. LuxR is a transcription factor activated by the autoinducer N- 3-oxohexanoyl-homoserine lactone (3-oxo-C6), which is produced by the synthase LuxI. V. fischeri possesses additional QS systems that converge on a signaling cascade that, unlike the LuxR-LuxI system, is conserved among all Vibrionaceae members (Milton 2006 ). At its core is a phosphorelay composed of the histidine phosphotransfer protein LuxU and the response regulator LuxO (Fig. 1 ). Based primarily on the studies of the analogous phosphorelay in Vibrio harveyi , LuxU is predicted to become phosphorylated on a conserved histidine residue by the kinases AinR and LuxQ under conditions of low autoinducer concentrations, for example, low cell density (Freeman and Bassler 1999a , 1999b ; Ray and Visick 2012 ). Whereas AinR appears to serve as the receptor for the AinS-derived autoinducer N- octonoyl-homoserine lactone (C8) (Gilson et al. 1995 ; Kimbrough and Stabb 2013 ), the periplasmic protein LuxP is thought, based on work in V. harveyi , to bind to the furanosyl borate diester, autoinducer-2 (AI-2), which modulates the kinase activity of LuxQ toward LuxU (Neiditch et al. 2005 , 2006 ). Upon phosphorylation, LuxU is predicted to donate the phosphoryl group to a conserved aspartic acid residue of LuxO, which can then activate transcription of qrr1 (Miyashiro et al. 2010 ). The RNA chaperone Hfq assists the small regulatory RNA (sRNA) Qrr1 in the posttranscriptional repression of LitR, a global transcription factor that regulates motility, host colonization factors, and bioluminescence (Fidopiastis et al. 2002 ; Miyashiro et al. 2010 ; Cao et al. 2012 ). The net effect of the integrated QS systems is that under high cell density (i.e ., in the presence of autoinducers) LuxO becomes de-phosphorylated, which leads to low qrr1 expression and the ability of V. fischeri to fully activate the lux genes. Figure 1 Model of the core quorum-signaling (QS) system in Vibrio fischeri . The outputs of the QS systems AinS/AinR and LuxS/LuxP/LuxQ converge on the LuxU/LuxO phosphorelay. Phosphorylated LuxO activates transcription of the small regulatory RNA Qrr1 that posttranscriptionally represses litR , which encodes the transcription factor LitR. In this study, we show that SypK modulates QS by affecting the kinase activity of LuxQ (indicated by the yellow arrow). Within the past decade, V. fischeri has also become a useful model organism to explore the genetic determinants for developing biofilms, which are elaborate structures that bacterial populations or communities can produce to associate with surfaces and each other (Visick 2009 ; Yildiz and Visick 2009 ). By synthesizing and exporting various exopolysaccharides and other molecules (Flemming et al. 2007 ), bacteria can remain attached to a surface and sheltered from unpredictable and potentially stressful environments. Wild-type V. fischeri does not produce a substantial biofilm under standard laboratory conditions. However, activation of a cluster of 18 genes that comprise the syp locus (Yip et al. 2005 ) confers phenotypes associated with biofilms, such as the ability to form wrinkled colonies on solid agar surfaces (Yip et al. 2006 ; Hussa et al. 2008 ). QS was recently shown to impact the dynamics of syp -mediated biofilm development as mutants containing an insertion in luxQ were delayed in wrinkled colony formation (Ray and Visick 2012 ). Further investigation revealed that deletion of luxU but not luxO leads to a similar delay, highlighting a branch within the signaling network that impacts biofilm development but not bioluminescence. In this current study, we report our discovery of another connection between the QS system and the syp locus, which expands our knowledge of the regulatory networks that V. fischeri has evolved to interact with its environment.",
"discussion": "Discussion We initiated this study to identify novel regulators of qrr1 , which encodes a sRNA conserved among members of the Vibrionaceae family (Lenz et al. 2004 ; Miyashiro et al. 2010 ; Weber et al. 2011 ; Shao and Bassler 2012 ). Our results show that the putative oligosaccharide translocase SypK can interact with the QS network to control qrr1 levels in V. fischeri . In particular, our data are consistent with the conclusion that SypK exerts its impact at or above the LuxP/Q complex in a manner that depends on the kinase activity of LuxQ (Figs. 6 , 7 ). Because SypL, which is another predicted inner membrane protein, does not activate qrr1 expression (Fig. 4 ), we posit that the increase in LuxQ signaling by SypK is not due to a general perturbation of the protein composition within the inner membrane. The activation of the LuxU-LuxO phosphorelay resulting from SypK overexpression, either as an individual protein or in the context of activation of the syp locus (Fig. 8 ), leads to the expression of qrr1 . Finally, our results show that the interaction between SypK and the QS pathway is sufficient to affect cellular behaviors associated with quorum sensing in V. fischeri , for example, bioluminescence and motility (Figs. 2 C, 5 ). From these results, we have generated the model presented in Figure 9 . Under conditions that result in transcription of the syp locus, the sypK gene will be expressed, leading to the formation of this putative oligosaccharide translocase within the inner membrane. While contributing to biofilm formation, SypK can also interact with the QS pathway via LuxQ. Our results in Figure 7 B suggest that the levels of the K+/P− variant of LuxQ may be influenced by SypK. One interpretation of these data consistent with our other results is that SypK stabilizes the kinase form of LuxQ; however, a clear understanding of this pathway awaits a more complete study. Regardless, the net result of sypK induction is activation of the LuxU-LuxO phosphorelay and, consequently, qrr1 expression. Whether the resulting increase in LuxU phosphorylation can further enhance syp transcription via SypG (Ray and Visick 2012 ) remains unknown; further characterization of the signaling pathway directly controlling the syp locus is required to determine whether a potential positive feedback loop is present. Interestingly, this regulatory link between SypK and LuxQ suggests a mechanism that enables V. fischeri cells densely packed within a biofilm to activate the LuxU-LuxO phosphorelay even in the presence of AI-2, which is an autoinducer broadly used for bacterial QS. Figure 9 Model of integrated biofilm and QS regulatory networks in Vibrio fischeri . Environmental cues, including conditions associated with the host during initial colonization of the squid light organ, activate the syp locus. In addition to participating in biofilm formation, SypK activates the LuxU-LuxO phosphorelay via the LuxP/Q complex. The resulting expression of qrr1 leads to enhanced flagellar-based motility, which may contribute to host colonization (see Discussion). The threefold increase in qrr1 expression observed in the transposon insertion mutants resulted in a 55-fold decrease in luminescence (Fig. 2 A and C). Such dramatic effects on cellular luminescence from small changes in the expression of genes involved in QS have been previously reported. For instance, deletion of arcA , which encodes a response regulator that responds to redox conditions, results in a 500-fold effect on luminescence despite only changing luxI transcription by 10-fold (Bose et al. 2007 ; Septer and Stabb 2012 ). Similarly, deletion of litR results in less than a twofold effect on luxR expression, but decreases luminescence by 10-fold (Miyashiro et al. 2010 ). The results presented here provide another example of the exquisite sensitivity of bioluminescence to QS. We also found that the sypI mutant is 21-fold dimmer than a Δ litR mutant (Fig. 2 C). Because Qrr1 represses litR mRNA levels, we anticipated that the luminescence levels of strains overexpressing qrr1 would be higher than the Δ litR mutant. Our surprising result with the sypI mutant may be due, in part, to an indirect effect from overexpressed Qrr1 titrating free Hfq, a condition that occurs in E. coli when an sRNA is overexpressed (Moon and Gottesman 2011 ). In V. fischeri , lower levels of free Hfq may result in inefficient posttranscriptional regulation by other sRNAs involved in regulating luminescence. We found that overexpression of SypG also enhances qrr1 expression in a SypK-independent manner (Fig. 8 ). A recent study has revealed that the SypG regulon extends beyond the syp locus (Ray et al. 2013 ), and several of these genes appear to impact bioluminescence (V. A. Ray and K. L. Visick, unpubl. results). Future studies will determine how these SypG-regulated genes affect bioluminescence and whether the mechanism involves qrr1 . All sequenced Vibrionaceae members possess a luxQ homologue within their genomes; however, the presence of sypK and the remaining syp genes appears to be species dependent. In addition to V. fischeri , the pathogens Vibrio parahaemolyticus and Vibrio vulnificus have been reported to each contain the syp locus, including a sypK homologue (Yip et al. 2005 ). Homologues of sypK are also present within the genomes of Aliivibrio salmonicida ( VSAL_II0302 ), V. harveyi ( VIBHAR_02224 ), Photobacterium profundum ( PBPRA1735 ), and Vibrio splendidus ( VS_2150 ). Notably, the syp locus is absent from the genomes of V. cholerae , Vibrio anguillarum , and Vibrio furnissii . As future studies uncover the mechanism underlying the interaction between SypK and LuxQ, they will also provide opportunities to determine how sypK and, more generally, the syp locus have coevolved with the core QS network of the Vibrionaceae , and whether the regulatory link is conserved. Our finding that a protein involved in biofilm formation can also function in signaling through the QS pathway inverts the traditional view of the role of bacterial QS during biofilm development. Generally, many genes that are involved in forming biofilms are regulated by QS systems. For example, in V. cholerae , expression of the vps exopolysaccharide gene cluster is downregulated by HapR, the LitR homologue in V. cholerae , (Hammer and Bassler 2003 ; Zhu and Mekalanos 2003 ). In V. vulnificus , the LitR homologue SmcR was recently shown to control transcription of the capsular polysaccharide (CPS) gene cluster (Lee et al. 2013 ). From this latter work, it has been proposed that QS within mature biofilms results in the production of cell-associated CPS, which decreases the hydrophobicity of the cell surface. As a result, V. vulnificus cells are released from biofilms with high cell density, thereby providing a mechanism to control the overall size of a biofilm. QS control of biofilm formation has also been observed in non- Vibrionaceae bacteria, including the pathogen Pseudomonas aeruginosa , which uses hierarchically arranged LasR/LasI and RhlR/RhlI QS systems to control biofilm development, in addition to virulence, motility, and antibiotic resistance (Williams and Camara 2009 ). The study we report here demonstrates that induction of a gene involved in biofilm development is able to influence QS. Whether this regulatory link represents a general phenomenon in bacteria or is instead specific to V. fischeri remains unknown. Is the regulatory link between SypK and the QS network relevant to the known biology of V. fischeri ? Because biofilm formation is often correlated with a sessile, community lifestyle, it seems somewhat counterintuitive for microbes to activate a signaling cascade associated with the planktonic, that is, low cell density, state while actively developing a biofilm. However, the general developmental cycle of a biofilm includes dispersal, which describes the stage when a subset of cells leaves the matrix to initiate biofilm formation on another surface (McDougald et al. 2012 ). In V. cholerae , mutation of luxO represses biofilm formation, reduces motility, and promotes cellular detachment from biofilms (Zhu et al. 2002 ; Zhu and Mekalanos 2003 ). Consequently, inactivation of the LuxU-LuxO phosphorelay via QS can prime cells within mature, densely packed biofilms for dispersal. In V. fischeri , the effect on motility by the LuxU-LuxO phosphorelay is similar to that observed in V. cholerae : mutation of either luxO or qrr1 results in attenuated motility (Lupp and Ruby 2005 ) (Fig. 5 ). Therefore, the increased motility from SypK-dependent qrr1 activation may enhance V. fischeri dispersal from syp -mediated biofilms. Activation of qrr1 expression by SypK may also play a role during the initiation of the squid- Vibrio symbiosis. The current model of initial host colonization is a two-step process, in which V. fischeri cells first attach individually to host cilia and then aggregate in a syp -dependent manner outside the light-organ pores (Altura et al. 2013 ). The syp genes, including sypK , are required for V. fischeri to efficiently colonize juvenile squid (Yip et al. 2005 ; Shibata et al. 2012 ). Our model predicts that activation of the syp locus will result in high qrr1 expression (Fig. 9 ), which has been shown to repress litR mRNA levels (Miyashiro et al. 2010 ). V. fischeri cells containing a litR deletion allele are able to outcompete wild-type cells in host colonization (Fidopiastis et al. 2002 ; Miyashiro et al. 2010 ). We hypothesize that by linking SypK and the QS network, V. fischeri symbionts can escape via flagellar-based motility from the exopolysaccharide matrix secreted during the aggregation stage. The residual biofilm may hinder other cells from entering the light organ, thereby contributing to the winnowing process during the initial establishment of symbiosis (Nyholm and McFall-Ngai 2004 ). Future molecular-based studies will help resolve the interconnectivity of SypK in biofilm formation, host colonization, and quorum sensing in V. fischeri ."
} | 4,226 |
32527924 | null | s2 | 720 | {
"abstract": "Bacterial biofilms represent a basic form of multicellular organization that confers survival advantages to constituent cells. The sequential stages of cell ordering during biofilm development have been studied in the pathogen and model biofilm-former "
} | 63 |
36569514 | null | s2 | 721 | {
"abstract": "Slippery surfaces are sought after due to their wide range of applications in self-cleaning, drag reduction, fouling-resistance, enhanced condensation, biomedical implants etc. Recently, non-textured, all-solid, slippery surfaces have gained significant attention because of their advantages over super-repellent surfaces and lubricant-infused surfaces. Currently, almost all non-textured, all-solid, slippery surfaces are hydrophobic. In this work, we elucidate the systematic design of non-textured, all-solid, slippery hydrophilic (SLIC) surfaces by covalently grafting polyethylene glycol (PEG) brushes to smooth substrates. Furthermore, we postulate a plateau in slipperiness above a critical grafting density, which occurs when the tethered brush size is equal to the inter-tether distance. Our SLIC surfaces demonstrate exceptional performance in condensation and fouling-resistance compared to non-slippery hydrophilic surfaces and slippery hydrophobic surfaces. Based on these results, SLIC surfaces constitute an emerging class of surfaces with the potential to benefit multiple technological landscapes ranging from thermofluidics to biofluidics."
} | 289 |
37653089 | PMC10471604 | pmc | 722 | {
"abstract": "The dominant benthic primary producers in coral reef ecosystems are complex holobionts with diverse microbiomes and metabolomes. In this study, we characterize the tissue metabolomes and microbiomes of corals, macroalgae, and crustose coralline algae via an intensive, replicated synoptic survey of a single coral reef system (Waimea Bay, Oʻahu, Hawaii) and use these results to define associations between microbial taxa and metabolites specific to different hosts. Our results quantify and constrain the degree of host specificity of tissue metabolomes and microbiomes at both phylum and genus level. Both microbiome and metabolomes were distinct between calcifiers (corals and CCA) and erect macroalgae. Moreover, our multi-omics investigations highlight common lipid-based immune response pathways across host organisms. In addition, we observed strong covariation among several specific microbial taxa and metabolite classes, suggesting new metabolic roles of symbiosis to further explore.",
"introduction": "Introduction Coral reef benthic communities in shallow tropical habitats are dominated by three main types of primary producers: hermatypic reef corals (Cnidaria in the order Scleractinia), crustose coralline algae (CCA; Rhodophyta in the order Corallinales), and various types of macroalgae. In addition to serving as the sources for the majority of primary productivity on reefs 1 , 2 , these primary producers together control reef accretion through calcification and dissolution, determining habitat and reef architecture crucial for biodiversity 3 – 5 . The chemical ecology of benthic assemblages has been widely studied for decades, with notable ongoing advances in areas such as allelopathic interactions between corals and algae 6 , 7 , composition and bioavailability of dissolved organic matter exudates 8 – 13 , chemical communications required for symbioses within the coral holobiont 12 , 14 , 15 , and signaling compounds produced by CCA and/or their microbial consortia that act as key larval settlement cues for corals and other invertebrates 16 . Together, they act as hosts for a diverse community of microbial taxa on coral reefs, which includes both important primary producers like cyanobacteria and consumers such as heterotrophic bacteria that are capable of recycling dissolved organic matter. Current methods in untargeted metabolomics 17 have facilitated the rapid analysis of thousands of known and unknown compounds from hundreds to thousands of samples 18 , which allows comparative metabolomics to investigate how the chemical ecology of organisms vary among species 19 . Identifying compound classes that are either shared or distinct among species is the first step to understanding the evolution and function of these compounds within and across ecosystems. Moreover, constraining and contextualizing the chemical milieu of a “healthy” organism is important for developing a baseline against which to chemically interrogate an organism for signs of stress or disease 20 – 22 . Comparing the metabolomes of organisms that are critical to ecosystem functioning (i.e., ecosystem engineers) is poised to become a crucial component of Ecosystem- and Resilience-Based Management 23 , 24 . Comparative metabolomics will help define and characterize the chemical crosstalk that controls the biogeochemistry and function of critical marine habitats like coral reefs. The greatest diversity of functional genes in macroorganisms is found in their microbiomes: the collection of symbiotic unicellular eukarya, bacteria, archaea, and viruses that inhabit the tissues and surfaces of all plants and animals 25 . Linking microbiome structure to the metabolite composition of distinct interacting organisms can reveal the sources and dynamics of metabolites in ecosystems 26 . Whether a given compound or metabolite is associated with a particular clade of microorganisms or a particular host will help us understand symbioses in complex systems where microbial processes are critical. In the case of holobionts, metabolites are co-produced by the intertwined biochemical processes of both host and microbes. As untargeted metabolomics seeks to further annotate and understand the diverse compounds that comprise biological metabolism, every defined association of uncharacterized compounds within a microbe or host advances our understanding of biochemical ecology. In many cases, the evolutionary history of host organisms can predict microbiome composition 27 . However, on coral reefs, the evolutionary relatedness of benthic primary producers is complicated by multiple endosymbiosis events and convergent evolution. In evolutionary terms, CCA is closely related to non-crustose red algae (e.g., Jania sp.) and more distantly related to green algae (e.g., Halimeda sp.). Corals, in contrast, are metazoan mixotrophs harboring dinoflagellates (family: Symbiodiniaceae), which are themselves eukaryotes with photosynthetic organelles ultimately derived from brown algae. As a result, corals are able to both consume particulate organic matter and produce fixed carbon through photosynthesis. Metazoans and dinoflagellates have evolved distinct metabolic pathways relative to algae, though some important and ancient lipid biosynthesis pathways are shared 28 . Variation in how benthic primary producers have evolved to receive carbon, synthesize metabolites, and interact with microbes will all contribute to distinct microbiomes and metabolomes. In addition to the phylogenetic relatedness of host organisms, it is important to consider the functional traits of benthic primary producers that may structure their microbial communities. Traits related to the physical structure, production of microbial food sources, and host immune response are all potential determinants of microbial community composition. Physical structure, including anatomical microhabitats 29 , 30 , can influence the settlement and persistence of both macro- and microorganisms. Host exudates provide a microbial food source, selecting microbial taxa that are capable of breaking down these compounds 31 . Host immune response, and the corresponding release of antimicrobial compounds, can be activated through multiple signaling pathways. Analysis of tissue samples using untargeted metabolomics can help coral reef biologists understand how host organisms respond to microbial colonization. In this study, we sought to synoptically sample specimens of coral, macroalgae, and CCA from a representative tropical reef ecosystem to analyze their metabolomes and microbiomes (workflow in Fig. 1 ). Previous studies have documented metabolites and microbes across benthic primary producers, but they focused on particular benthic groups (e.g., algae 32 – 35 and coral 36 – 38 ), zones of interaction 6 , 37 , 39 , or exuded metabolites 31 . The goal of this study was to characterize the microbes and tissue metabolites of all three types of dominant benthic primary producers within a common framework. Tissues were sampled in a broad sense, with each sample representing a homogenate of multiple distinct physiological compartments. Homogenized samples contained surface biofilms, mucus layers, subsurface tissues, and skeletal components (e.g., calcium carbonate substructures), representing a relatively holistic snapshot of the holobiont. We hypothesized that the three types of primary producers would have distinct microbiomes and metabolomes. We predicted that corals would be enriched in known coral bacterial symbionts, such as members of the class Endozoicomonadaceae 40 and the diazotrophic order Rhizobiales 6 , 41 , while macroalgae would harbor known copiotrophic microbes, such as Flavobacteriales and Rhodobacterales 42 . We expected that geographic location and evolutionary relatedness of the host would predict microbiome and metabolome composition. Within each producer type, we predicted there would be differentiation among host genera following patterns of phylogenetic relatedness. We focused on collecting discrete host individuals, which led us to exclude from our analysis interwoven species assemblages such as turf algae. Finally, we expected to identify groups of immune signaling compounds across all three primary producer types that could potentially show response to microbial colonization. Fig. 1 Metabolomic and microbiome workflows followed in this study. The samples were initially collected, prepared, and submitted for the metabolomics and genomics workflow. a The metabolomics workflow consisted of data-dependent acquisition (DDA) mode in an LC-ESI-HR-MS/MS instrument in the positive ionization mode, followed by data processing in MZmine2 (feature finding step). The resulting metabolomics feature table was submitted to statistical analysis in R and to the following workflows in the GNPS environment, including Feature-Based Molecular Networking, library searches against annotated spectral databases (available in GNPS) and against datasets in public repositories (MASST), and the Qemistree workflow (in silico annotations and chemical hierarchy analysis). b The genomics workflow consisted of 16 S gene sequencing, in which the data were submitted to the Metaflowmics pipeline. The genomic feature table generated was then subjected to statistical analysis in R. c Multi-omics analyses (mmvec and biclustering analysis) were employed for data integration. Logos in the figure were obtained from the GNPS, R, Qiime2, and MZmine official websites. We further hypothesized that selected microbial taxa would covary with some proportion of metabolites within and among host taxa, allowing us to identify putative microbe-metabolite interactions. Previous metabolomic investigations revealed that lipids play an important role in holobionts and can act as signaling chemicals in inflammatory responses 37 ; therefore, we expected to find microbial groups associated with host immune response metabolite pathways. We sought to minimize bias and standardize our analyses by rapidly sampling triplicate biological specimens of multiple species within each category across a wide area of reef over a 2-day period. Our sampling consisted of five benthic sites (Fig. 2 ) roughly 100 m 2 designed to span distinct coral reef types flanking a shallow embayment (Waimea Bay, Oʻahu) representative of the high wave-energy north-facing shores of the Hawaiian Archipelago. We analyzed 112 tissue samples from three types of primary producers (41, 45, and 26 samples for CCA, coral, and macroalgae, respectively) to resolve the paired metabolomes (Fig. 1a ) and microbiomes (Fig. 1b ) of each type, as well as the microbiomes and metabolomes of key genera sampled within each type. We characterized differential abundance of metabolites and microbes, showing clear associations between particular microbial taxa and the primary producer host type. Finally, we uncovered a number of microbe-metabolite covariation patterns (Fig. 1c ) that may indicate the source or use of key metabolites within these holobiont systems. Fig. 2 Sampling locations and taxa. a Map of sampling sites showing depth and elevation gradients within the study area. Bathymetry data courtesy of Hawaiʻi Mapping Research Group, SOEST, UH Mānoa and elevation data courtesy of the U.S.G.S. National Elevation Dataset. Site locations were collected by handheld GPS. The map was created using the R language for statistical programming (code available in the public GitHub repository associated with this manuscript). b – j Representative photos of benthic primary producers collected from Waimea Bay, Oʻahu. Scleractinian reef corals ( b ) Porites lobata , ( c ) Pocillopora meandrina , and ( d ) Montipora capitata ; erect rhodophytes ( e ) Jania and ( f ) Galaxaura , and a calcifying chlorophyte ( g ) Halimeda ; and crustose coralline algae ( h ) Hydrolithon , ( i ) Lithophyllum , and ( j ) Hydrolithon (PC: Keoki and Yuko Stender). For standardization in all of the analyses performed in this work, coral, macroalgae, and CCA samples were defined as orange, green, and blue, respectively. Collection permit: Hawaiʻi State Department of Land and Natural Resources Division of Aquatic Resources Special Activity Permit No. 2020-23. Representative photographs of benthic primary producers were obtained from https://www.marinelifephotography.com with authorization from Keoki Stender.",
"discussion": "Discussion Coral reef benthic primary producers represent chemically and microbially distinct holobionts. The composition of these organisms in a reef ecosystem affects the chemical compounds and microbial communities that occur on that reef. As benthic communities change, there are corresponding shifts in the biochemical processes occurring on the reef, which are a product of both the host genomes and bacterial metagenomes 62 . Understanding these microbially mediated chemical processes may help to explain the persistence of coral dominated reefs and global trends in coral reef degradation and shifts towards algal dominated systems. A necessary first step is to compare microbes and metabolites that are associated with dominant reef primary producers, including corals, CCA, and macroalgae. Our multi-omics analyses identified patterns in microbe-metabolite co-occurrence, which pointed towards potential microbe-host interactions in benthic holobionts. Our results largely corroborate our hypothesis that the three dominant benthic primary producers in coral reefs (corals, CCA, and macroalgae) harbor statistically distinct microbiomes and metabolomes. High-throughput chemical and microbial annotations allowed us to compare known immune signaling compounds across benthic holobionts. Correlations between microbial families and organism-associated immune signaling compounds may provide key insights into how host-microbe associations are either maintained or perturbed in reef ecosystems. This approach could be used to characterize multifaceted shifts in coral reef health, for example monitoring levels of copiotrophic microbes associated with macroalgae along with changes in the immune response pathways of both algae and coral. While targeted experiments are necessary to elucidate specific mechanisms of host-microbe interaction, untargeted metabolomic and metagenomic approaches can provide a snapshot of the chemical and microbial landscape of an ecosystem. The metabolomic analyses revealed that the chemical compounds produced by corals, CCA, and macroalgae were statistically different (Fig. 3 , Supplementary Data 1 ). CCA presented the most diverse chemical profile, emphasized by the high percentage of metabolites detected solely in this sample type. In natural systems, CCA colonizes and binds together coral reef substrates. In turn, CCA and its microbial biofilms provide a living and chemically-attractive habitat 63 , that recruits 16 other reef invertebrates, potentially making their metabolomes and microbiomes complex composites derived from multiple organisms. As expected, the greatest overlap in ion features was between the two algal functional groups we analyzed: CCA and macroalgae. Within coral and macroalgae, chemical profiles varied at the genus level (Fig. 3b, c, e, f ), in accordance with previous reports in the literature 37 . Our results emphasize the potential for metabolomics to be used as a tool for profiling the health of benthic organisms in coral reefs, a priority for the management of these sensitive ecosystems. Concepts that are already familiar to ecologists, like species richness and community dissimilarity, can be helpful when comparing metabolomic profiles across host organisms and geographic locations. Straightforward comparisons of feature richness and sample dissimilarity can reveal broad trends (Fig. 3 ). In our dataset, it was clear that chemical profiles varied spatially, varying significantly more than the microbiomes did among the same samples. Coral and CCA samples collected immediately adjacent to the Bay (sample code Waimea Bay, Fig. 2 ) grouped separately from samples collected elsewhere (Supplementary Fig. 3 ). It is important to note that samples from this site were extracted on a separate 96-well plate, so batch effects may explain some of the observed variation (Supplemental Methods). In terms of habitat, however, the Waimea Bay site was distinct from the other collection sites and we believe this pattern likely represents real biological variation. This site is located near the mouth of the Waimea River, which substantially alters environmental conditions. Within the bay, marine organisms are exposed to periodic runoff, freshwater, and sediment. Corals exposed to these conditions must expend energy to clear colony surfaces, which alters many metabolic processes 64 . In the future, metabolomics may be a sensitive tool for detecting environmental impacts on benthic communities before stressors lead to tissue loss or death. Of the ion features that were annotated based on spectral matches, most belonged to lipid classes. Lipids are ubiquitous metabolites that perform a variety of cellular functions and are present in both corals and algae 65 , 66 . In corals, they are an important energy reserve that can represent up to 40% of the coral’s dry mass and are important in supporting physiological resilience and post-bleaching recovery 66 . In algae, lipids play numerous roles in energy storage, membrane formation, and stress response, for which they are considered biomarkers 67 . While lipids are perhaps best known for storing chemical energy and forming bilayer membranes, their emerging role in complex intercellular signaling has become a focus of host-microbe interactions. In heavily studied mammalian systems, short-chain fatty acids are thought to be a key component of crosstalk between the gut microbiome and host organisms 68 . A separate interaction pathway proceeds through the oxidation of membrane phospholipids, which can tip off a host organism to the presence of pathogenic microbes, igniting a signaling cascade of damage control through a process known as ‘innate immunity 69 . Given the importance of oxylipins, in host immune response, it is perhaps unsurprising that microbial symbionts and pathogens are capable of upregulating, modifying, and mimicking them. In corals, it has been established that oxylipins produced by the metazoan host can be modulated by their dinoflagellate endosymbionts, with endosymbionts effectively dampening host immune response 70 . Numerous fungal and bacterial pathogens are capable of dampening host immune response through modification of host oxylipins or through the production of bioactive oxylipins 52 . Within our dataset, a number of potentially relevant immune signaling compounds were recovered. Glycerophospholipids, including lysophosphatidylcholines (LysoPCs), derivatives such as platelet-activating factor (PAF) C:16, and phosphatidylcholines (PCs) with varied chain extension, were among the most commonly detected lipids in our samples. LysoPCs are widespread in many organisms, including mammals, where they act as proinflammatory signals during immune response. Previous studies have proposed that corals may have an immune system with properties analogous to mammals 37 , 71 . PAF concentration in Porites sp. was observed to increase under stress and inflammatory conditions 37 , 59 and has been suggested as a molecular indicator of coral bleaching 36 . In algae, LysoPCs were previously reported in lipidomic analyses of brown algae 72 , and PCs, in general, are common metabolites found in eukaryotic algae and plants. Other lipid classes were also widely detected in our dataset. Fatty acids, which were recovered in abundance, are precursors for the biosynthesis of other lipids 73 , present antibacterial activity against specific pathogenic microorganisms 74 , and are implicated in microbe-host signaling 68 . Several glycerolipids of marine origin have been reported 75 , and acyl carnitines are widely found in corals 31 , being involved in transport across cell membranes 76 and considered biomarkers of cell toxicity 77 . More specifically, eicosanoids and linoleic acids are reportedly related to immune response in animals, and their biosynthesis was previously related to bacterial infection in corals 37 , 78 and insects 79 . Lastly, linoleic acids and their epoxyoctadecamonoenoic acid (EpOMEs) derivatives have been implicated in mouse and lepidopteran immune response to bacteria 80 , 81 . The potential avenues for lipid mediated host-microbe are both broad and numerous. Through the application of untargeted metabolomics, we were able to identify positive co-occurrence between these putative signaling compounds and microbial taxa with established roles in coral reef ecosystems. In addition to lipids, our untargeted approach to metabolomics recovered a number of non-lipid compounds. Pigments, such as loliolide and fucoxanthin, were annotated. Fucoxanthin plays an essential role in harvesting light for photosynthesis and photoprotection 82 , while loliolide is an apocarotenoid considered as a photo-oxidative or thermal degradation product of carotenoids 83 . Pheophorbide A, a product of the chlorophyll catabolism, was also observed in macroalgae and CCA samples. Lastly, we recovered a number of purine nucleosides, which are universal molecules with a wide variety of vital biological functions in many organisms, forming building blocks for DNA and RNA. Both macro- and microorganisms in marine environments can produce structurally unusual nucleosides with unique biological properties 84 . There are abundant opportunities to apply untargeted metabolomics to investigate other types of compounds that may be enriched in particular organisms or involved in host-microbe interactions. A vast majority of molecular features that were statistically associated with different sample types could not be annotated. In fact, the annotation of metabolites in the metabolomics workflow remains a bottleneck in the field. The in silico tools that were employed unquestionably boosted the annotation rates in this dataset, but the annotation of certain metabolites remained intractable. While close to 18% of the metabolites detected in this dataset originated from CCA samples, many features that were primarily detected in CCA, comprising several molecular networks, did not result in any in silico annotations (Supplementary Fig. 5 ) or library matches. This is consistent with the low annotation rates of exuded metabolites of CCA recently reported 31 . Therefore, CCA represents a source of potentially new compounds to be investigated in future studies. A repository-scale analysis allowed us to determine whether the major networks composed of unannotated CCA MS/MS spectra had been previously reported in other public datasets. Despite comparison to numerous datasets, several CCA associated networks remained unmatched, indicating that CCA might produce distinct molecular families compared to the other reef primary producers, and further emphasizing how little chemical information related to CCA is publicly available. A handful of features from the major CCA molecular family matched with public datasets related to coral reefs, indicating that these compounds are, indeed, found in this environment and are not artifacts. Intriguingly, a molecular family with features matching datasets derived from both coral reefs and cultivated bacteria suggests that some of these compounds may be produced by the microbes associated with CCA. The enigmatic chemical diversity of CCA is noteworthy because of its role in the recruitment of coral larvae. CCA is known to foster coral settlement 16 , 85 and suppress the growth of macroalgae 86 , a necessary step for coral reef regeneration. However, the chemical mechanisms by which CCA signals to coral larvae and promotes their growth have not been characterized definitively. There remains much to be discovered about the chemical landscape of these organisms. Mirroring the metabolomic results, microbial amplicon sequencing revealed that different reef primary producers harbor distinct microbial communities (Fig. 6 , Supplementary Data 1 ). Among coral genera, Monitipora and Porites presented more similar microbiomes compared with Pocillopora , which is in accordance with their closer phylogenetic relatedness 87 . Similarly, the red algae genera Galaxaura and Jania (Rhodophyta) harbored more similar microbial communities compared to the green alga Halimeda (Chlorophyta), further demonstrating that microbial communities tracked the phylogenetic relatedness of the host organisms 88 . Meanwhile, CCA (Rhodophyta, Corallinales) had strikingly different microbiomes compared to the other algal taxa. The microbiomes of CCA were much more similar to those of coral than they were to erect macroalgae, suggesting that the encrusting and calcifying lifestyle of CCA and coral may be an important determinant of microbiome structure. Unlike the metabolomes, which differed by site, the microbial communities were not predictably site-specific (Supplementary Fig. 7 ). Sampling was conducted during the dry summer months, during which the river mouth was blocked by a sand berm, making it unlikely that substantial inputs of microbes from freshwater were occurring. A number of microbial taxa were significantly enriched in a particular host (or hosts) and matched previous reports of host-microbe associations 89 . Coral associated microbes included taxa that are thought to be involved in nitrogen recycling in and around the host, which may help corals to persist in oligotrophic waters 90 . Our data indicated several bacterial families associated with nitrogen cycling (Nitrosopumilaceae, Rhizobiales, Nitrosococcaceae, Nitrospiraceae) were associated with coral samples. CCA was associated with many of the same families as corals, including several families involved in nitrogen cycling. Nitrospiraceae are notable for their functional role as nitrite oxidizers while Nitrosopumilaceae are known for their role as ammonia oxidizers 53 , 54 . Both are thought to contribute to nitrogen cycling in coral holobionts 91 . A prior study reported that Rhizobiales were a core member of CCA microbiomes and that Nitrospiraceae were associated with both CCA and a calcium carbonate substrate control 92 . While Rhizobiales include diazotrophic taxa, their role in nitrogen cycling within the coral microbiome is still unclear 93 , 94 . Both nitrogen cycling and the presence of calcium carbonate substrates may be key to understanding these host-microbe associations. The family Kiloniellaceae was also highly associated with both CCA and corals; these microbes are putative denitrifiers 95 and have been reported in association with healthy corals 96 . There is growing evidence that the association between nitrogen cycling microbes and coral hosts is potentially an important mutualism mediated by chemical exchange of both organic and inorganic forms of nitrogen. Microbial nitrogen transformation, in particular dissimilatory nitrate reduction to ammonium, is thought to be common, though highly variable, in tropical scleractinian corals 97 . Prior work on coral reef exometabolites 31 showed that coral exudates were enriched in organic nitrogen containing compounds, which were distinct from those exuded by macroalgae, and could potentially drive shifts in water column microbial communities. Many Cyanobacteria are capable of nitrogen fixation and have been identified as important constituents of both coral and macroalgae microbiomes 98 , 99 . The presence of intercellular cyanobacteria in some coral species indicates that at least some nitrogen-fixing Cyanobacteria are capable of eluding or dampening coral host immune response 99 . In our analysis, macroalgae were associated with several families of Cyanobacteria (Supplementary Fig. 8 ). Of the prevalent Cyanobacteria in our dataset, most occurred at a higher relative abundance on macroalgae compared to coral and CCA. This trend towards an increased relative abundance of Cyanobacteria on macroalgae merits further investigation. Macroalgae-associated microbes included copiotrophic taxa (Rhodobacteraceae, Flavobacteriaceae, Vibrionaceae, and Altermonadaceae), which specialize in breaking down large organic molecules and have been reported to increase in response to algal exudates 8 . Previous work has shown that macroalgal exudates are carbon rich, while coral exudates contain higher concentrations of nitrogen and phosphorus 31 . Experimental removal of erect macroalgae from coral in French Polynesia altered the relative abundance of these bacterial families in coral tissues. Both Flavobacteriaceae and Rhodobacteraceae decreased in relative abundance following removal of epiphytic algae from coral 100 . These reports point to a compelling link between increased macroalgae abundance, the release of carbon rich exudates, and increased an abundance of copiotrophic bacteria on both the surface of reef organisms and in the water column. It remains untested whether macroalgae in some way benefit from this association with copiotrophic bacteria, perhaps through increased resource acquisition or competitive advantage. The observed difference between microbial communities associated with CCA and macroalgae has several possible explanations. CCA harbored a large number of exclusive compounds, any of which could be shaping microbial community structure. Microbiomes could also be responding to the structural differences in surface topology between upright macroalgae and the crustose carbonate substrates characteristic of CCA. Studies testing the effects of surface topology on bacterial settlement have yielded inconsistent findings, but there is evidence that differences in the physical structure of a substrate can impact microbial colonization 101 . The overlap in taxa may have been influenced by our collection method. For CCA and coral, whole cuttings were homogenized and included both surfaces and calcified skeletal components. Skeletal compartments of coral can harbor high microbial diversity and it is possible these microbes contributed to the similarities between these primary producer types 102 . It should be noted that CCA typically harbors a rich community of boring invertebrates, which could have contributed to the metabolic and microbial diversity that we observed in these samples 103 . The large overlap in microbial taxa between CCA and coral is not easily explained, but is notable given the putative role of CCA in coral larval recruitment; both metabolites and microbes derived from CCA are thought to induce the settlement of coral larvae 63 . Microbial families associated with macroalgae were consistent across genera, while microbial families associated with coral varied between the genera Pocillopora and Montipora / Porites . CCA was not identified to the genus level, but the individual samples did not display any clear patterns in microbial associations. Paired microbe-metabolite datasets are well suited to machine learning techniques, which can identify co-occurrence patterns across a large number of variables. We applied the neural networking tool mmvec 55 to calculate conditional probabilities of co-occurrence for microbes and metabolites 104 – 106 . With this multi-omic tool, it was possible to infer possible positive correlations between specific metabolites and microbes, in which a particular microbe was perhaps producing a metabolite, or, vice versa, the presence of a metabolite was inducing the proliferation of a microbe. Beyond these putative direct relationships, co-occurrence patterns could indicate a variety of indirect association between various microbes and metabolites. Microbes associated with algae co-occurred with a variety of long-chain fatty acids (Fig. 8a ). The algal associated microbial families Flavobacteriaceae (Bacteroidia), Saprospiraceae (Bacteroidia), and Rhodobacteraceae (Alphaproteobacteria) all co-occurred with oxidized fatty acid 8-HETE, a compound involved in algal immune response. In contrast to terrestrial plants, which have developed additional advanced anti-microbial response pathways, algae rely heavily on lipid signaling cascades and reactive oxygen species 107 . In two separate studies, 8-HETE was shown to be upregulated after wounding in the red algal species Gracilaria vermiculophylla 34 . In a related species, Gracilaria chilensis , 8-HETE was upregulated following wounding and, when added directly to G. chilensis , decreased settlement by a competing algal epiphyte 108 . In addition to driving microbial community composition through chemical exudates functioning as a food source 31 , perhaps macroalgae are selecting for microbial associates that are tolerant of the oxylipins and reactive oxygen species that are produced during algal immune response. In the subclass glycerophosphocholines (Fig. 8b ) microbe-metabolite co-occurrence was broadly divided between coral associated derivatives of Lyso-PAF and algal associated derivatives of Lyso-PC. Of the 14 compounds, 8 were differentially abundant in one or more primary producer types based on LM analysis. Previous metabolomics studies have identified lyso-PAF as an important compound in coral stress response that can be used to indicate coral health 36 , 37 . Both coral and algal associated microbes co-occurred with various Lyso-PCs, though algae appeared to harbor a greater number of Lyso-PC related compounds. These compounds are part of lipid synthesis pathways that have multiple connections with symbiosis. Lyso-PC itself has been implicated as a bioactive compound in plants. In plants that form arbuscular mycorrhizal symbioses, the addition of lyso-PC caused rapid alkalinization of plant roots and upregulated transporter genes characteristic of mycorrhizal symbiosis 109 . The base substrate of lyso-PC, phosphatidylcholine (PC), is a common membrane protein in both plants and animals. However, it is uncommon in bacteria and tends to be found in bacteria that have close associations with eukaryotes, including both pathogenic and symbiotic bacteria 110 . When PC synthesis is disrupted in these bacteria, it can result in a loss of virulence in the case of pathogens, and inefficient symbiosis in the case of mutualists 111 . The established role of these glycerophosphocholines in multiple terrestrial symbioses makes them a promising target for further investigation in marine holobiont systems. Multi-omics approaches to address ecological questions are relatively new, representing an opportunity to explore the impact of microbe-metabolite interactions on ecological systems. Coral reefs are highly complex ecosystems, occurring in a narrow latitudinal margin (~30° N/S) distributed across every continent on the globe apart from Antarctica. Multi-omics tools can shed light both on the commonalities and distinguishing features of these systems. Mmvec is a new tool designed for this type of study. In the Caribbean, a mmvec-based analysis of the coral Orbicella faveolata revealed that lipids presented high co-occurrence values with the phylum Firmicutes 106 . In our study, lipids were also key metabolites of reef primary producers, but co-occurred primarily with the phyla Proteobacteria, Bacteroidetes, and Cyanobacteria. Additional comparisons of marine organisms across the globe are likely to reveal distinct and shared co-occurrences patterns. In conclusion, the present study revealed that, while some ubiquitous features are shared across organisms, different types of coral reef primary producers harbor distinct microbial communities and small molecules. A major pattern in the data was significant site-to-site variation in the metabolomes of each benthic primary producer; in contrast, their microbiomes did not vary significantly between sites. This indicates that benthic primary producers maintain broadly consistent relationships with microbial taxa while their composition of tissue metabolites can be susceptible to change as a result of environmental conditions. Further studies are necessary to identify whether the observed patterns in microbe-metabolite association are driven by interactions that occur intracellularly, on the mucus layers and surface biofilms of tissues, or even within the non-living skeletal components of calcifying taxa. The metabolomics analyses revealed the presence of several biologically relevant lipids, which were primarily detected in coral and macroalgae samples. The in silico annotations allowed us to derive useful information about otherwise unknown metabolites based on structural databases, while multi-omics tools enabled us to investigate biological processes within the reef system. Our results suggest that CCA remains chemically underexplored, representing a relevant target for future studies. The microbiomes of CCA and coral were diverse and overlapped significantly, while the microbiome of macroalgae comprised distinct microbial families. These results provide a comparative view of the complex phylogenetic and chemical contexts in which coral reef symbioses occur. They represent a starting point for future studies to further investigate the complex roles of metabolites, which serve as a nutrient source, signaling language, and physical interface for hosts and microbiomes in holobiont systems."
} | 9,372 |
38741379 | PMC11267296 | pmc | 723 | {
"abstract": "Abstract Triboelectric nanogenerators (TENGs), a promising strategy for harvesting distributed low‐quality power sources, face inevitable bottlenecks regarding long‐term abrasion and poor durability. Herein, both issues are addressed by selecting an earthworm‐inspired self‐replenishing bionic film (ERB) as the tribo‐material of sliding‐freestanding TENGs (SF‐TENGs), it consists of an interconnected 3D porous network structure capable of storing and releasing lubricant under cyclic mechanical stimuli. Thanks to the superiority of self‐replenishing property, there is no need for periodic replenishment and accurate content control of lubricant over the interfacial‐lubricating SF‐TENGs based on dense tribo‐layers. Additionally, an SF‐TENG based on ERB film (ERB‐TENG) demonstrates remarkable output stability with only a slight attenuation of 1% after continuous operation for 100 000 cycles. Moreover, the ERB‐TENG displays a distinguished anti‐wear property, exhibiting no distinct abrasion with an ultra‐low coefficient of friction (0.077) and maintaining output stability over a prolonged period of 35 days. Furthermore, integration with an energy management circuit enables the ERB‐TENG to achieve a 39‐fold boost in charging speed. This work proposes a creative approach to enhance the durability and extend the lifespan of TENG devices, which is also successfully applied to wind energy harvesting and intelligent sports monitoring.",
"conclusion": "3 Conclusion In conclusion, we have successfully developed an ultra‐durable and stable SF‐TENG utilizing an ERB film as a tribo‐material. This bionic film is constructed by incorporating lubricant within a 3D porous structure, demonstrating remarkable output stability and outstanding wear resistance. It can continually lubricate the friction interface when subjected to mechanical stimulation, thereby addressing the issue of precise control and periodic replenishment of lubricant content in DT films. Additionally, the ERB film with a COF of 0.077 demonstrates outstanding performance with no discernible surface wear after 100 000 sliding friction cycles. Furthermore, it only experiences a minimal 1% decrease in electrical output. This real‐time replenishment mechanism ensures a consistent output and extends the service life of the SF‐TENGs. Thus, this work presents a promising approach to enhance the durability and output stability of SF‐TENG devices, which have been applied to energy harvesting and intelligent sports monitoring.",
"introduction": "1 Introduction With the escalating issues of global climate change and the unsustainability of fossil energy, achieving carbon neutrality has become a consensus among the international community. [ \n \n 1 \n , \n 2 \n \n ] Conventional power supply approaches face significant challenges in meeting the increasing demand for distributed energy supply, particularly for ubiquitous electronic devices. Consequently, there is an urgent need to develop innovative energy supply systems to facilitate the development of smart, pervasive, and energy‐efficient electronics. As a novel type of energy harvesting and conversion system, triboelectric nanogenerators (TENGs) based on Maxwell displacement current have shown great promise in converting a variety of low‐quality mechanical energy into electricity. [ \n \n 3 \n , \n 4 \n , \n 5 \n \n ] Among them, sliding‐freestanding TENGs (SF‐TENGs) have gained considerable attention from researchers, as they provide high energy conversion efficiency. [ \n \n 6 \n \n ] Nevertheless, the matter of material abrasion caused by in‐plane intense friction severely hinders the further development of SF‐TENGs. [ \n \n 7 \n , \n 8 \n , \n 9 \n \n ] \n Hence, large efforts have been dedicated to prolonging the service life and heightening the robustness of SF‐TENGs utilizing structural optimization, [ \n \n 10 \n , \n 11 \n , \n 12 \n , \n 13 \n \n ] introducing fur brushes, [ \n \n 14 \n , \n 15 \n , \n 16 \n \n ] interfacial lubrication, [ \n \n 17 \n , \n 18 \n , \n 19 \n , \n 20 \n , \n 21 \n , \n 22 \n \n ] and so on. From the perspective of structural optimization, this strategy is not well suitable for further enhancing the output performance of SF‐TENGs due to their inherent device structural design that sacrifices contact area. To improve the durability and long‐term stability of SF‐TENGs on the premise of maintaining high output performance, several studies have focused on material optimization by introducing fur brushes to lessen frictional resistance without compromising the contact area. [ \n \n 15 \n \n ] Nonetheless, fur brushes encounter obstacles such as limited availability of fur sources as well as uncontrollable fur thickness and density. Currently, interfacial liquid lubrication has emerged as the mainstream approach to decrease friction and abrasion among the tribo‐pairs while ensuring sufficient contact intimacy between triboelectric layers. [ \n \n 23 \n \n ] Wu et al. introduced a liquid lubricant on the surface of PI‐dense film to enhance the wear resistance of SF‐TENGs and regulated its output performance by screening the type of lubricant. [ \n \n 24 \n \n ] Zhou et al. used squalane as the lubricant to obtain a high output and durable SF‐TENG by optimizing its content. [ \n \n 25 \n \n ] However, a notable concern arises from the fact that the lubricant on the surface of the triboelectric dense films involved in the above reports is gradually consumed during sliding in all probability, thus necessitating periodic replenishment to ensure optimal electric performance and lubricating property. In addition, the content of lubricant is supposed to be controlled with great precision. Once the lubricant is excessive, it will significantly restrain the output performance of SF‐TENGs, thereby hindering their practical applications. Besides, the long‐term placement stability of liquid lubricant on the triboelectric film is not yet known. These matters highlight the need for further optimizing tribo‐materials of SF‐TENGs to address the aforementioned limitations for enhancing their longevity. As is well known, earthworms possess an extraordinary ability to navigate through dry soil without causing injury. This can be put down to their sophisticated epidermal glands that can continually secrete mucus from dorsal pores and columnar epithelial cells in response to external mechanical stimuli. Moreover, their rough skin, consisting of macroscopic annuli and micro ripples, stabilizes the secreted mucus, thereby forming a thick lubricative layer for drag reduction and anti‐wear. [ \n \n 26 \n \n ] Inspired by earthworms, we propose designing a hierarchical porous structure that can store lubricant using capillary effect and discharge it from holes with external mechanical stimulation, aiming to achieve an ultra‐durable triboelectric material and address the challenges of lubricant loss, poor stability, and the need for precise control of content. Herein, we construct an earthworm‐inspired self‐replenishing bionic film (ERB) that forms an interconnected 3D porous network structure. The film possesses the characteristic of self‐replenishing lubricating when subjected to mechanical stimuli, enabling the creation of an SF‐TENG based on ERB film (ERB‐TENG) with remarkable performance, which features prominent output stability and extraordinary anti‐wear performance after continuous operation for 100 000 cycles. Importantly, precise control of the lubricant content is not required. Additionally, even if the adsorbed oil reaches the saturation point of the ERB film, it will not have a negative impact on the output performance. This is in contrast to ordinary dense films, conferring a significant enhancement in its practicality. Such outstanding merits of ERB hold considerable potential for long‐term energy harvesting and smart sports monitoring sensing domains.",
"discussion": "2 Results and Discussion 2.1 Bionic Structure Design of ERB The lubricating mechanism of earthworms is schematically depicted in Figure \n \n 1 a . Earthworms possess sophisticated epidermal glands that consecutively secrete mucus in response to external mechanical stimuli. The rough skin of the earthworms enables the mucus to be stabilized, creating a slippery layer that reduces drag. This fascinating lubricating mechanism has gained significant attention for the development of low‐friction thin layers with anti‐wear and durable properties. Drawing inspiration from this lubricating mechanism, a bioinspired design is proposed to create a 3D porous structure through phase separation for self‐replenishing lubricating (Figure 1b ), which is fabricated by negative tribo‐materials referred to as THV, a terpolymer comprising tetrafluoroethylene (TFE), hexafluoropropylene (HFP), and vinylidene fluoride (VDF). The resulting porous film, named PT/SiO 2 , is a porous THV (PT) film that has been modified with 0.5 wt.% oleophilic silicon dioxide nanoparticles (SiO 2 ). The typical surface and cross‐section morphology of PT/SiO 2 are shown in Figure 1c,d , respectively. The PT/SiO 2 surface displays lunar‐like crater structures, characterized by micro‐pores in the middle. Moreover, it exhibits a 3D interconnected porous structure on the cross‐section. These features contrast sharply with the dense THV (DT) film depicted in Figure 1e,f . The ERB film with self‐replenishing lubricating is obtained by submerging the PT/SiO 2 in squalane until it becomes saturated with oil. The ultra‐depth of field microscope images of PT/SiO 2 (Figure 1g ) and ERB (Figure 1h ) film reveal a noticeable contrast. The lubricant disperses into small droplets and is uniformly distributed in the ERB film, effectively demonstrating successful lubricant storage within the film. Figure 1 Bionic structure design and morphology characterization of ERB. a) The secretion behavior under extra stimulation: i) graphical illustration of an earthworm passing through arid soil; ii) image of earthworm's skin after mechano‐stimulus; iii) schematic for earthworm surface texture and secretion mechanism. b) Illustration of the stimuli‐responsive release of the ERB film: i) 3D structure and morphology diagram of PT/SiO 2 film; ii) enlarged diagram of the cross‐section of PT/SiO 2 film; iii) the PT/SiO 2 film was immersed in a lubricant until it reached the saturation point. The lubricant was stored within the porous structure of the PT/SiO 2 film; iv) the lubricant is extruded from the porous structure of ERB to the friction interface; v) with periodic reciprocating sliding, ERB film can achieve self‐replenishing lubrication. The scanning electron microscopy (SEM) images of c) the surface and d) cross‐section of PT/SiO 2 , respectively. SEM images show the structure of DT film from different perspectives: e) surface and f) cross‐section. Comparison of ultra depth of field images of PT/SiO 2 g) before and h) after adding lubricant. 2.2 Fabrication of ERB Film and Mechanism of ERB‐TENG \n Figure \n \n 2 a schematically depicts the fabrication process of the ERB film. The film is typically produced using a homogeneous solution consisting of THV particles dissolved in a mixture of N,N‐Dimethylformamide (DMF) and acetone doped with SiO 2 (Figure S1 , Supporting Information). A detailed description of this process is outlined in the Experimental Section. The formation of the internal and external porous structures in the ERB film is due to the synergetic effect of vapor‐induced phase separation (VIPS) and thermally induced phase separation (EIPS), [ \n \n 27 \n , \n 28 \n , \n 29 \n , \n 30 \n , \n 31 \n \n ] as illustrated in Figure 2b and Note S1 (Supporting Information). According to the formation mechanism of porous structure, two strategies are employed to achieve PT films with high porosity (Note S2 , Supporting Information): altering the ratio of DMF/acetone solvent (ranging from 1:4 to 1:0.3) and adjusting the relative humidity (ranging from 33% to 88%). These strategies are advantageous for storing lubricant, as plotted in Figures S2 and S3 (Supporting Information). The experiment results revealed that a decrease in acetone content leads to the reduction of the lunar‐like crater structures and less portion of pores on the surface, but a gradual increase of pore diameter in cross‐section. A maximum porosity of 69.4% was achieved when the DMF/acetone ratio was 1:2, as provided in Figure S3 (Supporting Information). This phenomenon is ascribed to the higher acetone content which results in rapid accumulation and condensation of water vapor on the surface. This accumulation forms larger droplets that leave larger sizes of lunar‐like crater structures on the surface of films after the volatilization of water droplets. Additionally, less water vapor entered the interior of the solution, resulting in the formation of smaller‐sized internal pores. Therefore, it can be inferred that humidity will also affect its surface morphology and porosity. Figure S4 (Supporting Information) shows the surface morphology and the pore structure of the PT films fabricated in different humidity with the DMF/acetone ratio of 1:2 (w/w). It can be observed that with the increase in humidity, the size of the lunar‐like crater structures on the surface noticeably increases, while fewer pore structures are formed (Figure S4a–j , Supporting Information). However, the interior pores which showed in the cross‐section image (Figure S4k–o , Supporting Information) exhibited the opposite trend, highlighting the essential role of water vapor in pore formation. In summary, under the dual regulation of DMF/acetone (1:2) and humidity (40%), the maximum porosity obtained for the PT film is 74.5%. This provides rich pore structures for the storage of lubricants. Figure 2 Schematic diagram of fabricating ERB films and working principle of the ERB‐TENG. a) Diagram of the synthetic strategy employed to obtain the ERB film: vi) Ball‐and‐stick models of the three monomer molecules that consist of THV material. b) Diagrammatic illustration of the methodology for investigating the dynamic formation mechanism of ERB film. c) Architecture of the ERB‐TENG. d) Charge distribution and e) simulated potential distribution of i–iv) the four states existing during a single cycle of electricity generation. The device structure of the ERB‐TENG is illustrated in Figure 2c , where the ERB film and a copper foil (Cu foil, 30 × 30 mm 2 ) serve as the tribo‐materials, while two conductive fabrics with the same dimensions act as symmetric electrodes. The detailed fabrication process is described in the Experimental Section. The operating principle of the ERB‐TENG is schematically presented in Figure 2d . Accompanied by the relative sliding motion between the Cu foil and the ERB layer, the electricity generation process can be universally divided into four states. The Cu foil slides between two adjacent conductive fabric electrodes: electrode A (left electrode) and electrode B (right electrode), which adhere to the ERB film. In the initial state, the Cu foil completely overlaps with electrode A (Figure 2d–i ), producing positive charges on the Cu foil due to triboelectrification, equal amounts of negative charges are generated on the surface of the ERB film. As the Cu foil slides toward electrode B, positive charges flow from electrode B to A, generating a current via the external circuit to balance the change in potential difference (Figure 2d‐ii ). Once the Cu foil is fully in contact with the ERB film directly above electrode B, all the positive charges are transferred to electrode A (Figure 2d‐iii ). As the sliding motion continues, the positive charges flow back to electrode B (Figure 2d‐iv ), inducing an inverse current. The ideal potential distributions of the two electrodes under the aforementioned four states were visualized through a finite element simulation using COMSOL software, as plotted in Figure 2e . 2.3 Output Performance and Durability of the ERB‐TENG In the following electrical performance tests, the normal load was set to 5 N at a frequency of 3 Hz, unless otherwise specified. The electric output tests of SF‐TENG were carried out by the measurement platform, shown in Figure S5 (Supporting Information). The research indicated that the electrical output of fabricated PT films diminished, compared to that of DT layers. This decline could be attributed to the porous structure and a significant decrease in effective contact area during the sliding process. Specifically, on the one hand, a decrease in the dielectric constant resulting from the porous structure may lead to a reduction in the electrical output, as presented in Figures \n \n 3 a–c and S6 (Supporting Information). [ \n \n 32 \n , \n 33 \n , \n 34 \n \n ] On the other hand, the porous configuration of the PT films exhibits relatively large pores and features a highly excellent open‐cell structure, thus resulting in high charge dissipation and poor charge storage ability, which causes a further decline in electrical output. Additionally, the surface structure of the PT films displays concave lunar‐like crater structures, which means that during the sliding process, the copper foil cannot come into contact with the sunken part, resulting in a significant reduction in the effective contact area, as well as the electrical output. To address this issue, different ratios of SiO 2 electret with outstanding charge trapping effect were introduced into PT films for further boosting the dielectric constant and enhancing the electric output. [ \n \n 32 \n \n ] The results revealed that the addition of SiO 2 led to an increase in both output and dielectric constant. However, when the weight ratio of SiO 2 in THV exceeded 1 wt.%, the output performance slightly decreased due to the decrease in the effective electrification area of THV (Figure S7 , Supporting Information). This was attributed to SiO 2 nanoparticles tending to appear on the surface, as illustrated in Figure S8 (Supporting Information). [ \n \n 35 \n , \n 36 \n \n ] As a result, the competition between the increase in charge storage sites and the decrease in the effective THV contact area at high SiO 2 concentrations led to the final decline in the output of TENGs. [ \n \n 35 \n \n ] As an auxiliary proof, the dielectric constant of pristine PT film and PT/ SiO 2 films with various weight ratios of SiO 2 exhibited a similar trend as observed in the above‐mentioned output results. The optimal mass ratio of SiO 2 should be 0.5 wt.%. More importantly, SiO 2 could be utilized as an oil carriers, lifting the oil absorption of PT/SiO 2 film by 70.34% in comparison with PT film without SiO 2 as a result of the large specific surface area of SiO 2 as well as the presence of numerous active sites on the surface of SiO 2 that can trap oil (Figure 3d ; Note S3 , Supporting Information). [ \n \n 37 \n \n ] \n Figure 3 Output performance and superior durability of ERB‐TENG. a) Transferred charges, b) short‐circuit currents, and c) output voltages of DT, PT film, and PT/SiO 2 film in contact with Cu foil, respectively. d) Influence of SiO 2 introduced into PT film on adsorption amount of lubricant. e) The effect of different lubricant dosages on the adsorption amount of lubricant for PT/SiO 2 . f) Variation curve of lubricant adsorption amount of PT/SiO 2 with time. Influence of lubricant volumes on g) short‐circuit currents, h) output voltages of PT/SiO 2 films. i) Influence of lubricant content on short‐circuit currents of DT–Cu contact. Comparisons of j) transferred charge, k) short‐circuit current, and l) output voltage for 100 and 400 µm thick PT/SiO 2 films of varying squalane contents. m) Superior long‐term output stability of ERB‐TENG within 35 days. n) Comparison of the output voltage stability between DT‐based interfacial‐lubricating SF‐TENG (DL‐TENG) with 200 µL lubricant and ERB‐TENG under continuous sliding friction for 100 000 cycles. o) Comparison of the device stability of this work with that reported by others. Experiments were conducted to investigate the maximum oil absorption of PT/SiO 2 films by varying the amount of lubricant. The results manifested that adding 200 µL of squalane reached saturation (Figure 3e ). Upon excessive lubricant was added, an excess of lubricant was visibly observed on the surface of the PT/SiO 2 film (Figure S9 , Supporting Information). Furthermore, it was observed that the PT/SiO 2 film immersed in squalane could reach saturation within 3 min (Figure 3f ). Subsequently, the effect of different amounts of lubricant on the electrical output of PT/SiO 2 films was examined. Results revealed that the addition of low‐content lubricant slightly increased the electrical output while adding 150 µL squalane resulted in a significant boost in output. This improvement can be ascribed to the ability of liquid lubricants to avoid the formation of the transfer film, increase the effective solid–solid contact area, and inhibit interfacial electrostatic breakdown. [ \n \n 24 \n , \n 25 \n \n ] Once it reached saturation, there was almost no distinct variation in the output (Figure 3g,h ). This phenomenon indicates that there is no need to deliberately control the quantity of oil in PT/SiO 2 films. In contrast, dense tribo‐layers only showed improved output when 1 µL of lubricant was added, and the electrical output plummeted to a lower level when excessive lubricant was added owing to a reduction in the efficient contact area between tribo‐pairs (Figure 3i ; Figure S10 , Supporting Information). It can be inferred that it was necessary for dense tribo‐layers to accurately control the lubricant content, and the instability of the lubricant caused by the operational process required periodic replenishment to maintain the output stability (Figure S11 , Supporting Information). It is anticipated that the thickness of PT/SiO 2 film will impact its maximum oil absorption. Hence, it is imperative to investigate whether an increased film thickness necessitates a higher consumption of lubricating oil. Herein, PT/SiO 2 film with a thickness of 400 µm is fabricated and examined to evaluate the impact of lubricant content on its output performance. Results exhibit that the output performance is significantly lower than that of the 100 µm thick PT/SiO 2 film in the absence of lubricant, as illustrated in Figure 3j–l . This discrepancy can be attributed to the weakening of the electrostatic induction to the bottom electrode as the film thickness increases. [ \n \n 38 \n , \n 39 \n \n ] The performance of devices with varying film thicknesses provides additional confirmation of this phenomenon (Figure S12a–c , Supporting Information). Interestingly, when adding lubricating oil with the same content as the 100 µm PT/SiO 2 film (i.e., 200 µL), there is no significant change in output performance for 400 µm PT/SiO 2 film, but an obvious increase for 100 µm PT/SiO 2 film. However, it was not until 1000 µL lubricant was added that the output performance of the 400 µm PT/SiO 2 film significantly improved and tended to stabilize (Figure 3j–l ). And more importantly, with only 200 µL of lubricant added, the output voltage of the 400 µm thick PT/SiO 2 film decreased by 31.9% after 100 000 cycles of durability test (Figure S12d , Supporting Information) and its surface showed noticeable wear (Figure S12e,f , Supporting Information). These results indicate that a positive correlation between film thickness and oil storage capacity. Moreover, a higher volume of lubricating oil is required as the film thickness increases to enhance the lubrication effect, ideally reaching a state of oil absorption saturation. Since the lubricant self‐replenishing is activated under mechanical stimuli, so it is vital to investigate the influence of the magnitude of the mechanical stimuli on the release of the lubricant and the final output. Specifically, we replaced the Cu foil in ERB‐TENG with cotton sheets of the same size. Alterations were made to the mechanical pressure and following a several‐minute sliding period at 3 Hz, the mass difference of the cotton sheets before and after the test was measured. The volume of lubricant released from the ERB films under different mechanical pressures can be calculated via utilizing Equation 1 \n \n (1) \n V = m ρ \n where V is the volume of lubricant released from the ERB under mechanical pressures; m is the mass of the lubricant released from the ERB under mechanical stimuli, which is equivalent to the mass difference of the cotton sheets before and after the test; ρ is the density of squalane, ρ = 0.81 g ml −1 . The experimental results indicate that the greater the mechanical stimulation, the more lubricant is released, as illustrated in Figure S13a (Supporting Information). Digital images of the cotton sheets with the fluorescently labeled lubricant absorbed from the ERB film under different mechanical stimulations are shown in Figure S13b,c (Supporting Information). They are illuminated by visible light (Figure S13b , Supporting Information), and an ultraviolet lamp with a wavelength of 365 nm (Figure S13c , Supporting Information), respectively. It is clearly seen that the color intensity of cotton sheets deepens as mechanical stimulation increases, suggesting a higher release of lubricant. To explore the influence of the magnitude of the mechanical stimuli on the final output, we first investigated the variation in electrical output of PT/SiO 2 film under various pressures, results revealed that the output initially increased with rising pressure, plateauing at 5 N, then maintaining stability with further pressure increments, as outlined in Figure S13d,e (Supporting Information). For ERB‐TENG, at lower pressures, the electrical output of ERB‐TENG increased with increasing pressures; however, at 7 N pressure, the output declined (Figure S13f,g , Supporting Information), which contrasts sharply with PT/SiO 2 film without lubricant. This decline could be due to the excessive release of lubricant under high pressures, leading to a decrease in the effective contact area between the two friction materials. This indicates that achieving a controlled release of lubricating oil by regulating external pressure is crucial for ERB‐TENG. Stability is a vital parameter for TENG devices as it ensures their reliable operation over prolonged durations. Figure S14 (Supporting Information) exhibits the output stability of PT/SiO 2 over 100 000 cycles. Results indicated the device with the PT/SiO 2 film showed a 67% drop in output after 9500 cycles. This drop can be ascribed to the destruction of the porous structure of PT/SiO 2 , leading to the formation of the transfer film. Consequently, the transfer film reduced the effective contact area, ultimately causing a decrease in output. Figure 3m showcases that the durability of the ERB‐TENG was retained for up to 35 days, with no significant decline in output performance. Besides, the ERB‐TENG displayed distinguished long‐term output stability, with a decay of only 1% electric output after 100 000 operation cycles, as depicted in Figure 3n . Given the notable impact of lubricant content on the electrical output of DT films, durability tests were performed for 100 000 cycles using both higher lubricant content (200 µL) as well as lower lubricant content (1 µL). The results revealed a sharp decline of 77% and 32% in the electric output of the corresponding SF‐TENG lubricated with 200 µL and 1 µL squalane after 100 000 cycles, respectively (Figure 3n ; Figure S15 , Supporting Information). In comparison to the state‐of‐the‐art work, [ \n \n 15 \n , \n 25 \n , \n 40 \n , \n 41 \n , \n 42 \n , \n 43 \n , \n 44 \n \n ] the stability/decay value reported in this work exceeded the values documented in previous research on ultra‐durable SF‐TENGs (Figure 3o ). To further demonstrate the superior performance of the ERB film compared to other materials, we select commonly used electronegative materials like PTFE, FEP, and PI, to compare the electrical output and abrasion resistance of them with the ERB film. The results indicate a significant superiority of the ERB‐TENG in terms of the electrical output (277 V, 2.78 µA, 227 nC), surpassing that of devices utilizing PTFE (107 V, 1.15 µA, 97 nC), FEP (84 V, 1.04 µA, 82 nC), and PI (96 V, 0.80 µA, 70 nC), as illustrated in Figure S16a–c (Supporting Information). Furthermore, the electrical output of the PTFE, FEP, and PI‐based TENG decreased by 45%, 56%, and 77% after 100 000 cycles, respectively, accompanied by noticeable scratches on their surface morphologies, as depicted in Figure S16d and S17 (Supporting Information). These experiments have well verified that ERB has excellent performance compared with other materials. The improved performance of the ERB‐TENG presented in this work provides a potential approach to enhance the output stability and durability of SF‐TENG. 2.4 Lubricity and Anti‐Wear Performance of the ERB‐TENG and DL‐TENG In order to distinctly visualize the surface friction behaviors, a dynamic sliding friction test device was designed (Figure S18 , Note S4 , Supporting Information). The friction force and corresponding coefficient of sliding friction (COF) were tested for each material pair mentioned above, following the GB 10006‐88 standard. Three measurements were taken to obtain experimental results. The addition of squalane to PT/SiO 2 ‐Cu tribo‐pair resulted in a significant reduction of 69% for the friction force and COF, as depicted in Figure \n \n 4 a,b . In contrast, when oil was added to the contact interface of DT‐Cu tribo‐pair, the friction force and COF decreased by only 13% (Figure S19 , Supporting Information; Figure 4b ). This phenomenon can be explained by two factors. On the one hand, the ERB film with the self‐replenishment property releases the lubricant, distributing it evenly into small droplets to achieve interfacial lubrication in the ERB‐Cu tribo‐pair. On the other hand, when a lubricant is applied to the contact interface of the smooth DT film and the Cu foil, a continuous oil film with a specific thickness is formed, expelling air from the interface. Subsequently, the atmospheric pressure acts, causing an increase in the friction force between the two interfaces. This hinders the lubricating effect of the lubricant, resulting in only a slight decrease in both the friction force and COF. For further evaluating the wear resistance of the self‐replenishing lubricating ERB film, the surface roughness of Cu foils and corresponding SEM images of various tribo‐layers were analyzed before and after 100 000 cycles of sliding. The results showed that the initial dense Cu foil features a smooth surface (Figure S20a , Supporting Information) with a surface roughness of only 0.38 µm (Figure 4c,d‐i )). Nevertheless, in the case of the DT‐Cu tribo‐pair, the addition of lubricant (1 or 200 µL) could not prevent substantial wear on the Cu surface (Figures S20 and S21 , Supporting Information). Even when 200 µL of lubricant was applied, the surface roughness of the Cu demonstrated a substantial increase, more than 4.7 times compared to that of the pristine Cu sample (Figure 4c,d‐iii )). It is noteworthy that for the PT/SiO 2 ‐Cu tribo‐pair, same as the DT‐Cu tribo‐pair, significant wear was observed on the Cu surface in the absence of any lubricant (Figure 4d‐iv ; Figure S22a , Supporting Information). However, upon the addition of 200 µL of lubricant, the Cu surface exhibited virtually no discernible wear (Figure S22b , Supporting Information), with the surface roughness increasing by a mere 42% as compared to the untreated Cu sample (Figure 4c, d‐v )). Figure 4 Comparison of the lubricity and long‐term wear resistance between the ERB‐TENG and DL‐TENG with 200 µL lubricant. a) The friction force between PT/SiO 2 and Cu without oil and with oil. b) The corresponding coefficient of sliding friction of the SF‐TENG with different tribo‐pairs. c) Comparison of the surface roughness of initial Cu foil, as well as Cu foils rubbing with DT and PT/SiO 2 without oil and in oil after 100 000 sliding friction cycles. d) 3D images were used to depict the surface profile of different Cu foils under varying conditions. These conditions included: i) initial Cu foil, ii) Cu foil in contact with DT without oil, iii) Cu foil in contact with DT with 200 µL oil, iv) Cu foil in contact with PT/SiO 2 without oil, and v) Cu foil in contact with PT/SiO 2 with 200 µL oil. 3D images of Cu foils mentioned in ii–v) were tested after 100 000 sliding friction cycles; SEM images of the surface of e) initial DT film, as well as DT films in contact with Cu f) without oil, and g) with 200 µL squalane after continuous 100 000 cycles. h–j) SEM images of the surface of h) initial PT/SiO 2 , and PT/SiO 2 in contact with Cu i) without oil and j) with 200 µL squalane after 100 000 cycles. The wear characteristics of the DT and PT/SiO 2 tribo‐layers which were in contact with Cu foils demonstrated wear patterns that paralleled the trends observed in the wear of the Cu surface, as previously mentioned. Specifically, the original DT film exhibited a highly smooth surface (Figure 4e ). However, significant scratching was evident after 100 000 sliding friction cycles (Figure 4f ), with substantial wear still observable despite the application of either 1 or 200 µL of lubricant (Figure S23 , Supporting Information; Figure 4g ). This finding is consistent with its high friction resistance and COF. For the PT/SiO 2 film, the surface experienced severe degradation in the absence of lubricant, and even its porous structure was damaged (Figure 4h,i ). Yet, post‐treatment with 200 µL of lubricant (ERB), the surface morphology remained nearly unchanged, indicating remarkable wear resistance (Figure 4j ). This enhanced durability can be ascribed to the lubricant being harbored within the porous structure, endowing the film with self‐lubricating properties that significantly reduce the COF and consequently play a role in friction reduction and wear resistance enhancement. 2.5 Application Demonstrations of ERB‐TENG As a kind of device that converts low‐quality environmental energy into electricity, TENGs have been extensively applied in many scenarios. The output power of a TENG under practical conditions depends on its load resistance. Figure \n \n 5 a,b systematically investigated the output voltage, current, and power density of the ERB‐TENG dependent on external load resistances from 10 4 to 10 10 Ω. The instantaneous power density on the load reaches a maximum value of 0.45 W m −2 at the matched resistance of 304 MΩ. An arrayed ERB‐TENG consisting of two sections is developed for driving commercial electronics and demonstrating the charging performance: an ERB film attached to 16 staggered electrodes as a slider, and 8 separate Cu foils adhered to the substrate as a stator. Figure 5c exhibits the arrayed ERB‐TENG directly illuminating 512 LEDs (Figure S24 and Movie S1 , Supporting Information). An energy management circuit (EMC) is designed that is aimed at addressing the issue of low charging efficiency resulting from circuit loss during the practical application of SF‐TENGs, depicted in Figure 5d . The specific working mechanism is briefly described below. Here, a matching capacitance C in is introduced to store the unstable energy generated by the SF‐TENG via full‐wave rectification, and a gas discharge tube (GDT) is adopted as a switch to efficiently release accumulated energy. The GDT is off till the voltage of C in exceeds the threshold voltage of the discharge tube, allowing the instantaneous release of energy from C in to the inductor via gas breakdown discharge. Subsequently, the energy in the inductor will continuously power the capacitor C out through a closed‐loop circuit, significantly improving the charging speed. Specifically, the arrayed ERB‐TENG is combined with the EMC to charge a 47 µF capacitor, it is able to charge the voltage to 3 V in 0.8 s. However, without the EMC, it takes 32.7 s to charge the capacitor to the voltage of 3 V. This result indicated that the introduction of EMC can remarkably boost the charging efficiency by 39 times, as illustrated in Figure 5e . The charging performance of the device with EMC at various capacitances (47, 100, 220, 470 µF, 1, 2.2 mF) is also examined, as presented in Figure 5f . The voltage on the capacitor of 220 µF can be charged to 4 V within 8 s, highlighting its excellent practical feasibility for driving multiple commercial electronics. Figure 5g shows the charge‐discharge curve of the 20‐inch LCD blackboard recorded during the whole operation, enabling handwriting and erasure functions (Movie S2 , Supporting Information). The ERB‐TENG can be further developed into a rotary‐freestanding TENG (ER‐TENG), which is desirable for more application scenarios, such as efficiently harnessing wind energy. The basic structure of the ER‐TENG is illustrated in Figure 5h , which consists of a rotator, a stator, a wind cup, and a shell. The breeze from the blower enables the output voltage, current, and transferred charge of the ER‐TENG to reach 90 V, 1.25 µA, and 11 nC at a rotation speed of 270 rpm, respectively, as outlined in Figure 5i ; Figure S25 and Movie S3 (Supporting Information). The ER‐TENG mentioned above offers significant advantages for wind energy harvesting, presenting a promising development prospect. Figure 5 Charging performance characterizations and application demonstrations of the ERB‐TENG. a) Instantaneous peak V oc and I sc of ERB‐TENG concerning load resistance. b) Power density of the ERB‐TENG concerning a load resistance of 10 4 –10 10 Ω. c) Photograph of 512 LEDs lighted instantaneously by arrayed ERB‐TENG. d) Equivalent circuit diagram of the integrated energy management circuit (EMC) for the arrayed ERB‐TENG. e) Comparison diagram of the charging speed of 47 µF capacitor by the arrayed ERB‐TENG with and without EMC. f) Charging curve of different capacitors by the arrayed ERB‐TENG with EMC. g) Charging curve of a 470 µF commercial capacitor upon powering a 20‐inch color LCD blackboard driven by arrayed ERB‐TENG. Inset: enlarged view of the discharge curve when erasing the handwriting on the LCD blackboard. h) Schematic view of the structure and materials for the electrode stator and the ERB rotator. Photographs of the electrode stator: i) and the ERB rotator ii). i) The V oc of the ER‐TENG was obtained at the rotating speed of 270 rpm by harvesting wind energy. j) Signal flowchart of the “sit and reach” intelligence test. k) Demonstrations of the “sit and reach” test system with ERB‐TENG for intelligent sports monitoring. Photographs of initial state (i and ii) and status after measurement (iii and iv). Nowadays, smart sports facilities and wearable equipment are gaining popularity in the realm of sports, as it undergoes a digital and intellectual transformation. [ \n \n 45 \n \n ] The “sit and reach” test is a mandatory assessment in the “National Student Physical Fitness and Health Standards” to evaluate the flexibility of adolescents' bodies. Here we designed a self‐powered “sit and reach” system, which can provide real‐time vocal announcements of the test results. This system introduces a single‐electrode arrayed ERB‐TENG as a sensor which consists of two components: a slider and a stator. For the stator, the ERB is securely attached to 14 conductive fabrics. As for the slider, the electropositive material is affixed to the right index finger of the tester. As the finger slides over the stator, continuous pulse electrical signals are generated, consistent with the number of conductive fabric electrodes contacted by the finger, serving as analog signals (AS). AS are subsequently transmitted to a microcontroller unit (MCU) to be converted into digital signals (DS) and undergo further processing. The automatic multi‐scale peak detection algorithm is employed to automatically detect the number of digital signal peaks (DSP), which is utilized to calculate the finger sliding distance ( d ) by the following Equation 2 :\n \n (2) \n d = n ∗ d 0 + n − 1 ∗ d 1 \n where d is the grade of a conner in sit and reach too, n represents the number of signal peaks, d \n 0 is the width of each conductive fabric, and d \n 1 is the width of the intermediate gap between adjacent conductive fabrics. Subsequently, MCU transmits d to the universal asynchronous receiver‐transmitter (UART), thus a screen display terminal (SDT) receives the data via specific protocols in UART and presents it on the SDT. Furthermore, the voice synthesis module in SDT synthesizes the prompt message and broadcasts it via the speaker. The corresponding signal flowchart as well as application illustration are illustrated in Figure 5j,k (Movie S4 , Supporting Information). The system possessing a great merit of durability can realize smarter and more convenient detection for the “sit and reach” test."
} | 10,275 |
30073740 | null | s2 | 724 | {
"abstract": "Silk proteins are biopolymers produced by spinning organisms that have been studied extensively for applications in materials engineering, regenerative medicine, and devices due to their high tensile strength and extensibility. This remarkable combination of mechanical properties arises from their unique semi-crystalline secondary structure and block copolymer features. The secondary structure of silks is highly sensitive to processing, and can be manipulated to achieve a wide array of material profiles. Studying the secondary structure of silks is therefore critical to understanding the relationship between structure and function, the strength and stability of silk-based materials, and the natural fiber synthesis process employed by spinning organisms. However, silks present unique challenges to structural characterization due to high-molecular-weight protein chains, repetitive sequences, and heterogeneity in intra- and interchain domain sizes. Here, experimental techniques used to study the secondary structure of silks, the information attainable from these techniques, and the limitations associated with them are reviewed. Ultimately, the appropriate utilization of a suite of techniques discussed here will enable detailed characterization of silk-based materials, from studying fundamental processing-structure-function relationships to developing commercially useful quality control assessments."
} | 354 |
21589931 | PMC3092765 | pmc | 725 | {
"abstract": "Recent research on ecological networks suggests that mutualistic networks are\nmore nested than antagonistic ones and, as a result, they are more robust\nagainst chains of extinctions caused by disturbances. We evaluate whether\nmutualistic networks are more nested than comensalistic and antagonistic\nnetworks, and whether highly nested, host-epiphyte comensalistic networks fit\nthe prediction of high robustness against disturbance. A review of 59 networks\nincluding mutualistic, antagonistic and comensalistic relationships showed that\ncomensalistic networks are significantly more nested than antagonistic and\nmutualistic networks, which did not differ between themselves. Epiphyte-host\nnetworks from old-growth forests differed from those from disturbed forest in\nseveral topological parameters based on both qualitative and quantitative\nmatrices. Network robustness increased with network size, but the slope of this\nrelationship varied with nestedness and connectance. Our results indicate that\ninteraction networks show complex responses to disturbances, which influence\ntheir topology and indirectly affect their robustness against species\nextinctions.",
"introduction": "Introduction Recent research on the architecture of mutualistic networks (e.g.\nplant–pollinator and plant–seed disperser [1] but also anemone-fish\ninteractions [2] \nand marine cleaning symbiosis [3] ) suggests that their nested structure reflects a\nfundamental difference from antagonistic networks, arising from how specialisation\nis distributed among interacting species [1] , [4] , [5] . In contrast to mutualistic\nnetworks, antagonistic networks (e.g., predator–prey, herbivore–plant)\ntend to be more compartmentalised, i.e., characterised by cohesive groups of\ninteracting species with relatively few interactions among groups [6] , [7] . Several\nauthors have suggested that nested patterns of asymmetrical specialisation may be\nmore likely to develop in mutualistic interactions because natural selection\nspecifically favours the convergence and complementarity of traits in interacting\nspecies [3] ,\n [8] . In\ncontrast, antagonistic interactions may favour greater compartmentalisation through\nthe continual coevolution of defences and counterdefences (i.e., evolutionary arm\nraces involving exploitation barriers), which generates greater specificity [3] . While\ntheoretical studies have shown that the topological properties of one type of\nmutualistic networks (plant-pollinator) are more consistent with a mixture of\ncomplementarity and defence-counterdefence than with a predominance of\ncomplementarity [9] , we are not aware of any study that has addressed the\nhypothesis that mutualistic networks should be more nested than non-mutualistic\nones. A first step in this direction was recently made by Thébault and\nFontaine [10] ,\nwho showed that the nested and compartmentalised structures of mutualistic and\nantagonistic plant-animal interaction networks respectively maximise their\npersistence. However, a later commentary of their work [11] emphasizes that it does not\nevaluate whether differences in persistence are causing or resulting from the\ncontrasting network architectures (i.e. “a correlation does not imply\ncausality”). Furthermore, Gómez et al. [12] showed that phylogenetic\nconservatism of interaction patterns was equally likely to occur in mutualistic and\nantagonistic interactions, suggesting no different mechanism for both type of\ninteractions. In ecological networks, a nested structure indicates that reciprocal specialization\nis rare and, instead, specialists interact predominantly with generalists. It has\nbeen proposed that the robustness of interaction networks to anthropogenic\ndisturbances increases with their level of nestedness, since the loss of\nextinction-prone specialists is less likely to trigger the extinction of other\nspecialists in nested networks [7] , [13] . To illustrate this point, Fortuna & Bascompte [14] showed that,\nwhen simulating extinctions, real-world plant-animal networks start to decay sooner\nbut persist longer than simulated, random networks in response to habitat loss.\nHowever, no study has examined to date this hypothesis using real-world networks\nunder different disturbance regimes. Even more, the handful of studies that have\nexamined how mutualistic interactions respond to habitat loss or disturbance (e.g.\neffect of cattle ranching on pollinator networks [15] – [17] ; effect of fragmentation and\nhabitat loss on seed dispersal networks [18] – [19] ) show\ninconclusive results. While some species proved to be very sensitive [18] , others\nwere unaffected or even benefited from disturbances [19] . Comensalistic interactions, in which one organism benefits while the other is neither\nhelped nor harmed, provide an unexplored testing arena to understand the causes and\nconsequences of interaction-network topology. Because neither complementarity nor\ndefence-counterdefence traits are expected to arise in such interactions, they may\nprovide an evolutionary model against which to evaluate mutualistic and antagonistic\nnetwork properties. In particular, if the nested structure of mutualistic networks\nreflects the ecological effects of co-evolutionary complementarity, we would expect\nweaker degrees of nestedness in comensalistic networks. Moreover, should\ncomensalistic networks prove to be nested, an evaluation of their robustness against\ndisturbances would provide an independent test of the direct effect of network\nnestedness (i.e. teasing apart the potential indirect effects of trait\ncomplementarity) on its response to disturbances. In this study, we review the existing literature on mutualistic, comensalistic and\nantagonistic interactions (complemented with our own data on comensalistic networks)\nto evaluate whether they differ in their topological properties – and, in\nparticular, in their nestedness. We first show that comensalistic interactions are\nhighly nested, and then use both qualitative and quantitative network analyses to\nevaluate their response to disturbance. For this purpose, we identify topological\nchanges that precede rare-species extinctions (contrary to the stable network\nstructure generally assumed by cascading-extinction simulations) and evaluate\nwhether these changes result from neutral responses to species abundances\n( sensu Vázquez [20] , i.e. “network patterns\nresult from the fact that individuals interact randomly, so that abundant species\ninteract more frequently and with more species than rare species”) or do also\ninvolve changes in species-specific interactions (e.g. host selectivity by\nepiphytes). In particular, under the hypothesis of a higher sensitivity of rare\nspecies and interactions, we expect decreased network connectedness and nestedness,\nand lower levels of species specialization under disturbance. Throughout the paper, we use epiphyte-tree interactions and habitat\nmodification/fragmentation (resulting from the logging of host trees) as model\nsystem of comensalistic networks under disturbance. Epiphyte-tree interactions can\nbe regarded as comensalistic, since trees provide epiphytes with support for growth,\nreleasing them from the cost of building a resistant structure, while suffering no\neffect from epiphyte presence [21] . We chose this model system owing to its global\nimportance (an estimated 20,000–25,000 vascular species, representing approx.\n10% of all vascular plant species, are at least occasionally epiphytic; their\nabundances may reach up to 50% of the local flora, and they are involved in\ncritical ecosystem processes such as primary production, nutrient cycling, and\nhydrology [22] – [24] ) and measurement reliability (owing to their lasting\ncharacter, plant-host epiphyte networks are less vulnerable to sampling size biases\nintroduced by the dynamic nature of most mutualistic and antagonistic networks [25] – [27] ). Habitat\nmodification and fragmentation due to logging was chosen as model disturbance owing\nto its global importance (it is considered as a major threat to global biodiversity\n [28] , [29] , as well as a\ncommon cause of local extinctions and even cascade co-extinctions [30] – [32] ) and the\nwell-established sensitivity of the plant-epiphyte interactions to it (since the\npopulation turnover is generally comparable for epiphytes and host trees, patch\ndestruction and changes in host-tree dynamics caused by logging can be expected to\nresult in direct changes in epiphyte-tree interactions; [33] ).",
"discussion": "Discussion Our results show that (plant-epiphyte) comensalistic interactions are highly nested,\nparticularly in comparison to the set of mutualistic and antagonistic networks\nreviewed from the literature (which did not differ significantly between them). The\nhigh levels of nestedness observed in comensalistic networks were, however, largely\ndue to an abundance effect, as confirmed by the significance of the observed N and\nNODF values (only half to one-quarter of cases) and by Burns' null-model\nanalysis [21] . As\nfor the effect of disturbance on these highly-nested networks, it resulted in\nseveral topological changes that preceded rare-species extinctions and, therefore,\npotential extinction cascades. Connectedness, NODF and epiphyte generalization,\nwhich tended to increase with network size in old-growth forests, remained constant\nor decreased with size in disturbed forests. Quantitative-matrix (Procrustes)\nanalysis confirmed both the discordance between old-growth and disturbed-forest, and\nthe combined effect of both abundance-dependent and -independent effects thereupon.\nThese topological changes did not have, however, a straightforward effect on network\nrobustness, as estimated from species-extinction simulations. Robustness did not\ndiffer significantly between old-growth and disturbed forest, though it varied\nsignificantly with network size and NODF – a combination of factors shown\nbefore to vary differently in old-growth and disturbed forest. A first, unexpected result of our analysis was that antagonistic networks did not\nshow significantly lower nestedness than mutualistic networks. This result departed\nfrom our expectations, based on previous works (mainly Thébault &\nFontaine [1] ,\nBascompte et al. [1] , and other papers that elaborated on their conclusions),\nof decreasing nestedness from mutualistic to commensalitic to antagonistic networks.\nRather than differing from Bascompte et al. 's results [1] , however, those presented\nhere contradict their interpretation and generalizations. Bascompte et al. [1] showed that\npollination and seed-dispersal networks were more nested than food-web networks,\nparticularly after correcting for network size; their interpretation (followed by\nother authors, such as Guimãraes et al [3] or Ollerton et al [2] ) was that this\npattern can be extrapolated to mutualistic and antagonistic networks, and may be\nexplained by their evolutionary background (development of complementary versus\ndefence-counterdefence traits). Following this idea, Thébault & Fontaine\n [10] \ndeveloped a population dynamic model and compared pollination and herbivore\nnetworks, and concluded that the type of interaction (mutualistic vs. antagonistic)\nconstrains ecological networks towards different architectures. Our review focus on\nthat interpretation and, building on recently available papers and databases,\nreviews a broader spectrum of networks – including anemone-fish, ant-plant and\nhost-parasite networks. These data clearly show that antagonistic and mutualistic\nnetworks do not differ in their nestedness. It is therefore unlikely that the\nexplanation for the nested structure of many of these networks originates in a\nfundamental (ecological or evolutionary) difference between mutualistic and\nantagonistic interactions. A second, unexpected result was the highly nested nature of comensalistic,\nepiphyte-tree networks – particularly when considering their small network\nsize. Though we cannot rule out that, given the small amount of comensalistic\nnetworks studied to date, they may prove to have comparable nestedness to\nmutualistic and antagonistic networks in the near future, it seems reasonable to\nassume that they will not be any less nested. At any rate, the high values of\nnestedness shown by the networks included in this study made them a perfect\ncandidate to evaluate the effects of disturbance on network topology – thus\nevaluating whether the putative robustness of nested networks originates in\ncomplementary traits, supposedly characteristic of mutualistic interactions. Our comparison of old-growth and disturbed forest networks indeed showed that, though\nthese highly-nested networks were very robust to the strong disturbances imposed\nupon them (i.e. they showed small changes in species composition, despite large\nchanges in host-tree turnover rates), they showed considerable changes in network\nstructure and topology, which are taking place before any significant loss of\nepiphyte or tree species due to local extinctions. In particular, while network\nnestedness and connectedness increased with species richness in old-growth forests,\nit did the opposite in disturbed ones. This variation was largely manifested within\nforest patches (i.e. among transects), suggesting that while disturbed-forest\ncommunities show larger spatial variation in species richness, to the point of\nbecoming more diverse at localized spots, they also show an impoverishment in terms\nof the architecture of their interactions. Because epiphyte-tree network nestedness was caused by a combination of\nabundance-dependent and -independent effects, we used quantitative network\n(Procrustes) analysis to evaluate the relative contribution of both types of effects\nto the changes in network structure associated to disturbance. These analyses\nconfirmed that the aforementioned changes were largely caused by\nabundance-independent effects – abundance effects having, actually, a\nhomogenising effect in old-growth forests. The various mechanisms proposed to\nexplain host preferences (e.g. bark peeling rate [48] , water retention\ncapacity [49] , [50] , host size [36] , [51] or\nallochemical reactions [52] ) are certainly worth exploring in search for more\ndetailed causal effects behind these differences. These findings have important bearings for all published simulation works which,\nassuming fixed or stable network structure, estimate the consequences of extinction\nchains triggered by disturbance. If network structure changes in response to\ndisturbance, these changes must be understood and incorporated to such simulations.\nTo evaluate the potential influence of the observed changes in network structure on\nrobustness estimates, we performed a simple extinction-chain analysis based on the\nnetworks observed in old-growth and disturbed forest. The results indicate that,\nthough the direct effects of disturbance on robustness (in terms of differences\nbetween old-growth and disturbed forests) are of limited importance, it may have\nsignificant indirect effects mediated by changes in network topology (since network\nrobustness increased with both nestedness and connectance). Owing to the complex interactions between disturbance, network size, NODF and C,\nestimating the outcome of forest disturbance of plant-epiphyte networks will require\nmore extensive surveys and simulations. However, a first estimate indicates that, in\ncomparison with disturbed forests, old-growth forests will be particularly sensitive\nto spatial or inter-patch variation in network size. In these forests, local\nincreases in network size will result in increasing nestedness and connectance,\nwhich will in turn result in increased robustness. In contrast, disturbed forest\nwill show the opposite effect: increased network size results in decreased NODF and\nconnectance, which in turn result in decreased robustness. The net result is\ntherefore that old-growth patches (or sites within patches) with few species will be\nless robust to extinctions than disturbed patches (or sites), while those with many\nspecies will be more robust than disturbed patches (or sites). Old-growth forests can therefore be predicted to depend on the preservation of\nspecies-rich patches for the maintenance of the architecture of their interactions;\nwhile, in disturbed forests, all sites or patches will be roughly equivalent. Our\nanalysis thus stresses the importance of spatial heterogeneity to understand key\naspects of community structure and dynamics even in cases, such as network analysis,\nwhere spatial relationships tend to be explicitly ignored."
} | 4,144 |
37753526 | PMC10518492 | pmc | 726 | {
"abstract": "Many biological surfaces are capable of transporting liquids in a directional manner without energy consumption. Inspired by nature, constructing asymmetric gradient surfaces to achieve desired droplet transport, such as a liquid diode, brings an incredibly valuable and promising area of research with a wide range of applications. Enabled by advances in nanotechnology and manufacturing techniques, biomimetics has emerged as a promising avenue for engineering various types of anisotropic material system. Over the past few decades, this approach has yielded significant progress in both fundamental understanding and practical applications. Theoretical studies revealed that the heterogeneous composition and topography mainly govern the wetting mechanisms and dynamics behavior of droplets, including the interdisciplinary aspects of materials, chemistry, and physics. In this review, we provide a concise overview of various biological surfaces that exhibit anisotropic droplet transport. We discussed the theoretical foundations and mechanisms of droplet motion on designed surfaces and reviewed recent research advances in droplet directional transport on designed plane surfaces and Janus membranes. Such liquid-diode materials yield diverse promising applications, involving droplet collection, liquid separation and delivery, functional textiles, and biomedical applications. We also discuss the recent challenges and ongoing approaches to enhance the functionality and application performance of anisotropic materials.",
"conclusion": "Conclusion Natural creatures have evolved excellent capability in liquid transport and control, which inspired us through chemical, physical, and geometrical engineering and modulation in designing new liquid-diode materials. Smart fluid unidirectional control materials have made significant improvements in the last two decades, including theoretical foundation, progressive underlying mechanisms, and widespread application scenarios. It facilitates the development of the discipline, digital microfluids, and functional materials. More importantly, this energy-free model contributes more in alleviating the current energy crisis. Benefitting from its unique unidirectional liquid transport behavior, liquid-diode materials show broad application prospects in liquid manipulation and separation, cargo delivery and microreactors, bioanalysis and fast response platforms, solar water purification, smart fabrics, etc. Currently, strategies and technologies need to be developed for easy fabrication, faster response speed, and long-distance liquid transport control. This development may involve robust covalent bonding, self-assembly techniques, and use of novel materials such as self-healing mechanisms, as well as hierarchical micro/nano-engineering composites. Additionally, computational simulations and predictions can be included, which will effectively mitigate risks and expedite the screening of optimum preparation conditions. Anticipated advancements in nanotechnology, bionic manufacturing, and intelligence are expected to improve the droplet control precision, materials performance, and durability. Moreover, these are likely to increase cost-effectiveness and enable large-scale preparation, thereby facilitating commercial applications. Furthermore, the role of liquid-diode materials needs to be coupled to the application scenarios for synergistic application functions. For example, as a microreactor, it requires precise fluid control, fast response, and repeatability. As a wound dressing, it also needs good breathability, sterilization and anti-inflammation properties, and to be skin friendly. Most importantly, the guiding theoretical research of chemical composition, structure, and material performances requires refined development. This includes in-depth discussions and designs that encompass chemical gradients, roughness gradients, and curvature gradients to enhance the overall performance. The fostering of synergy between theoretical innovation, fabrication technology, and application-driven approaches creates a supportive and mutually reinforcing environment. There are both opportunities and challenges associated with liquid-diode materials. The field of bionic liquid-diode materials represents a multidisciplinary and innovative intersection of research. Through collaborative efforts, it is expected that liquid-diode materials will witness promising advancements and applications.",
"introduction": "Introduction Non-energy-consuming controlled passive transport of liquids over asymmetric surfaces has gained significant attention and become an important research hotspot. 1 The material systems that allow the spontaneous flow/penetration of liquid in only one direction are described as liquid diodes, which attract scientific and application interests in diverse fields such as liquid separation, water collection, digital microfluids 2 (enable fluidic functions at microscale for merging, splitting, transporting, mixing, and incubating, which makes them ideal for numerous biological and chemical platforms), energy, interface catalysis, and smart fabrics. 3 , 4 , 5 , 6 , 7 , 8 The anisotropic motion of droplets on a solid substrate was first observed by Greenspan in 1978. 9 The author pointed out that droplets tend to creep in a direction of greater adherence (lower contact angle) while retracting from weaker attachment regions (higher contact angle) because of the forces at the fluid/solid contact line. This work demonstrated that the surface energy gradient (chemical gradient) governs the typical behavior of droplets moving toward more wettable regions. Since then, droplet movement governed by anisotropic wettable surfaces has been extensively studied by various methods, such as electrochemistry, chemical vapor deposition, photolithography, and three-dimensional printing. 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 However, precise regulation of surface energy and designing of slopes are strictly required to overcome the effect of contact line pinning and hysteresis, which usually limits the movement of droplets. 18 , 19 In addition, the current available wettability gradient is difficult to implement over long distances due to the limited tunability of surface chemical composition and stability. Many natural creature surfaces possess the ability of directional water transport, which provided inspiration to humans, such as desert beetles, 20 , 21 spider silks, 22 cactus thorns, 23 , 24 \n Nepenthes peristome, 25 , 26 and Sarracenia trichome. 27 The evolution of these specific surfaces was accomplished through meticulous design of hierarchical micro/nanoscale structures and chemical compositions. The topological features hold significant potential for advancing the study of liquid-diode materials through structural gradient construction. This work summarizes that self-propelled droplet motions can be considered to rely primarily on three driving forces: asymmetric chemical gradient, roughness gradient, and curvature gradient. The generation of driving forces is mainly attributed to materials morphology of scales, orientation, periodic arrangement, shape, curvatures, as well as composition properties, which affect the droplet wetting preferences and motion tendencies. 28 In 2008, Yang and co-authors reported that droplet contact angle gradually decreased with an increasing pattern density of the microstructure. 29 They suggested that the droplet movement can be attributed to the difference between the higher and lower wetting state. This finding was further supported by Quérér et al. in 2009, who demonstrated droplet transfer toward regions of high texture density on hydrophobic pillar substrates, suggesting the role of roughness gradients. 30 To actively generate roughness gradient surfaces, researchers have designed arrayed asymmetric micro/nanopillars or ratchets with different densities and bending angles. Droplets on such surfaces preferentially move from areas of lower roughness to areas of higher roughness. The curvature gradient, which arises on asymmetrically shaped substrates, such as conical fibers found on cactus spines, and conical tubes such as shorebird beaks, provided examples of curvature gradients with both convex and concave types. In 2004, Quérér et al. studied the dynamics of a droplet on a horizontal conical fiber. They pointed out that the wettable drop spontaneously moves toward the region of lower curvature. 31 The driving force for droplets on surfaces with curvature gradients arises from the Laplace pressure gradient of the droplet. The Laplace pressure is the differential pressure between the inside and outside of a curved surface, which propels the movement of droplets with asymmetric shapes. Currently, the field of liquid-diode materials is very active due to novel preparation methods and advanced technologies. Driven by innovative applications, scientists in this field are working on the transition from fundamental to applied research. The combination of active chemical, physical, and geometrical gradients has sparked innovative advances in droplet transport, increasing flexibility, controllability, and speed for a wide range of applications. 32 , 33 , 34 , 35 , 36 This article describes recent achievements in controllable liquid transport on asymmetric two-dimensional (2D) substrates, including two models for droplet sliding and spreading, as well as anisotropic liquid penetration through three-dimensional (3D) Janus membranes. This study also highlights significant advances in the practical applications of liquid-diode materials in several fields, including liquid collection and separation, cargo delivery, bioanalysis, solar-driven water purification, and functional textiles. These materials offer a convenient, eco-friendly, and energy-efficient approach to interdisciplinary research and development."
} | 2,474 |
19284579 | PMC2669807 | pmc | 728 | {
"abstract": "Background The anaerobic degradation of organic matter in natural environments, and the biotechnical use of anaerobes in energy production and remediation of subsurface environments, both require the cooperative activity of a diversity of microorganisms in different metabolic niches. The Geobacteraceae family contains members with three important anaerobic metabolisms: fermentation, syntrophic degradation of fermentation intermediates, and anaerobic respiration. Results In order to learn more about the evolution of anaerobic microbial communities, the genome sequences of six Geobacteraceae species were analyzed. The results indicate that the last common Geobacteraceae ancestor contained sufficient genes for anaerobic respiration, completely oxidizing organic compounds with the reduction of external electron acceptors, features that are still retained in modern Geobacter and Desulfuromonas species. Evolution of specialization for fermentative growth arose twice, via distinct lateral gene transfer events, in Pelobacter carbinolicus and Pelobacter propionicus . Furthermore, P. carbinolicus gained hydrogenase genes and genes for ferredoxin reduction that appear to permit syntrophic growth via hydrogen production. The gain of new physiological capabilities in the Pelobacter species were accompanied by the loss of several key genes necessary for the complete oxidation of organic compounds and the genes for the c -type cytochromes required for extracellular electron transfer. Conclusion The results suggest that Pelobacter species evolved parallel strategies to enhance their ability to compete in environments in which electron acceptors for anaerobic respiration were limiting. More generally, these results demonstrate how relatively few gene changes can dramatically transform metabolic capabilities and expand the range of environments in which microorganisms can compete.",
"conclusion": "Conclusion These results provide insights into the evolution of Geobacteraceae species into different environmental niches and biotechnological applications. The results suggest that the last common ancestor of the Geobacteraceae was an acetate-oxidizing, respiratory species capable of extracellular electron transfer, and that specialization for fermentative/syntrophic growth evolved at least twice, allowing some Geobacteraceae to fill additional niches. The primacy of the respiratory mode is evident from the conservation of genes for all steps in this process including acetate uptake, central metabolism, and electron transfer across the cell membranes in both of the clades of the family. The fermentative/syntrophic Pelobacter species also contain many of these genes. However, they have lost several key enzymes that leave the pathways incomplete, including several necessary for the oxidation of acetate and most of the cytochromes predicted to provide the electrical connection between the inner membrane and the outside of the cell. Instead, the Pelobacter species have appropriated genes via lateral gene transfer for fermentative/syntrophic growth. It is clear that this has happened on two separate occasions. Although both P. carbinolicus and P. propionicus have closely related dehydrogenase genes for the initial metabolism of their unique substrates, acetoin and 2,3-butanediol, the genes for the further fermentation of these substrates are unrelated in the two organisms, reflecting the separate evolution of distinct metabolic pathways. The fact that P. carbinolicus also fills a syntrophic niche, participating in interspecies hydrogen transfer with hydrogen-consuming methanogens, whereas P. propionicus does not, may be explained by the genes associated with reverse electron transfer that only P. carbinolicus has appropriated. The fact that both Pelobacter species have phylogenetically distinct hydrogenase genes that are different from each other as well as those of the Geobacter species, may also reflect the difference in syntrophic capabilities of these species. The selective pressure to specialize in syntrophic/fermentative growth may have initially been found at the interface of redox boundaries in sedimentary environments. As respiratory Geobacteraceae deplete the supply of the terminal electron acceptor Fe(III) oxide, their capacity for growth is greatly diminished and organisms with other respiratory processes, such as sulfate reduction or methane production, become predominant [ 2 ]. Some Geobacter species can oxidize acetate to carbon dioxide and hydrogen when they lack external electron acceptors [ 44 ], but the slow rate of this metabolism and the requirement for very low hydrogen partial pressures means that they are not competitive with acetate-utilizing sulfate reducers or methanogens. Acquiring the ability to ferment novel substrates and/or to grow syntrophically could have facilitated expansion into Fe(III) oxide depleted environments. Under such conditions investing energy in the production of respiratory enzymes such as the c -type cytochromes Geobacteraceae require to grow under Fe(III)-reducing conditions would be maladaptive. These findings also provide insight into the types of metabolic changes that might take place as these organisms are being adapted for modern biotechnical applications. In applications such as the in situ bioremediation of uranium-contaminated groundwater and the conversion of organic compounds to electricity Geobacter species must deal with a scarcity of electron acceptor because electron donor is generally provided well in excess of electron acceptor availability. During in situ bioremediation Fe(III) oxides are rapidly depleted near the source of subsurface acetate amendments, limiting the growth and effectiveness of Geobacter -catalyzed U(VI) reduction [ 8 ]. Geobacter species form thick biofilms on the electrodes of microbial fuel cells, forcing many of the cells to metabolize acetate at a significant distance from this artificial electron acceptor [ 45 , 46 ]. Preliminary results suggest that, like the Pelobacter species described here, Geobacteraceae that predominate during in situ uranium reduction or on the anodes of microbial fuel cells have fewer c -type cytochromes (DRL, unpublished data). Enhanced ability to release excess electrons as hydrogen, in a manner similar to that of P. carbinolicus , could also be beneficial under conditions in which electron acceptor availability is limiting. Thus, these relatively few changes appear to have allowed a respiratory ancestor to radiate out into fermentative and syntrophic niches in addition to their respiratory roles in anaerobic environments. This information serves as a guide to the history of these organisms and provides information that could aid in optimizing their biotechnological applications.",
"discussion": "Results and discussion Identification of protein families in the six genomes The general features of each of the six genomes are presented in Table 1 . Orthologous proteins, those predicted to have similar functions in the different organisms, were identified by Markov clustering of sets of reciprocal best BLAST matches [ 17 ]. Using all 21,716 protein coding genes in the six genomes (see Additional file 1 ), 3,696 clusters with members from at least two genomes were defined (see Additional file 2 ). 15,207 proteins (70.0%) were classified into clusters. A functional role was predicted for each cluster using the G. sulfurreducens in silico model annotation [ 18 ] and COG categorization [ 19 ]. Table 1 General features of the Geobacteraceae genomes Geobacter sulfurreducens Geobacter metallireducens Geobacter uraniireducens Pelobacter propionicus Pelobacter carbinolicus Desulfuromonas acetoxidans NCBI ID NC_002939 NC_007517 NC_009483 NC_008609 NC_007498 NZ_AAEW00000000 Contigs 1 1 1 1 1 51 Length (nt) 3,814,139 3,997,420 5,136,364 4,008,000 3,665,893 3,828,328 GC Content (%) 60 59 54 59 55 51 Protein coding 3446 3519 4357 3576 3352 3234 rRNA operons 2 2 2 4 2 1 Plasmids none 13.8 kb none 30.7 kb and 202.4 kb none n/a The pattern of conservation of the proteins families across all species (the phyletic pattern) was determined (see Additional files 2 and 3 ). By far the most common pattern was conservation across all species, 5,345 proteins were members of clusters that included at least one protein from each genome (see Additional file 3 ). The second most common pattern was conservation in the Geobacter species and P. propionicus only (1,158 proteins). Proteins unique to the respiratory or fermentative species were relatively rare. Only 370 proteins were found in the respiratory species but not the fermentative species (see Additional file 4 ), and only 260 proteins were found in the fermentative species but not the respiratory species (see Additional file 5 ). Whole-genome phylogeny A phylogeny of the family was constructed using the 481 protein families that had a single protein from each of the six genomes and the outgroup species Anaeromyxobacter dehalogenans (see Additional file 6 ). These proteins had diverse functions, including information storage, metabolism, cell signaling, and unknown. The proteins from each genome were concatenated then aligned, and this alignment was used to create a Bayseian model of the phylogeny (Figure 1 ). This analysis confirmed that, as housekeeping-gene phylogeny suggests [ 20 ], the Geobacteraceae can be divided into two clades, and that the two fermentative Pelobacter species are not most closely related to each other, instead there is one Pelobacter species in each clade of the respiratory species (Figure 1 ). Figure 1 Genome-based Geobacteraceae phylogeny . Bayesian inference of the phylogenetic tree of the six Geobacteraceae species discussed in the text, using another Deltaproteobacterial species, Anaeromyxobacter dehalogenans , as the outgroup. The tree was based on a concatenation of the proteins in the 481 families that had exactly one ortholog conserved in each of the seven genomes (see Additional file 6 ). Values at branch points are posterior probabilities. Thus, based on phylogeny, there are at least three possibilities for how modern Geobacteraceae evolved to fill multiple metabolic niches. The ancestor may have been adapted to a fermentative/syntrophic lifestyle with no reliance on external electron acceptors. In this scenario, the Geobacter and Desulfuromonas species would have independently evolved strategies for the complete oxidation of organic compounds coupled to Fe(III) reduction. Alternatively, the ancestor may have been a respiratory microorganism, and the Pelobacter species independently evolved strategies for fermentation and syntrophy and lost the ability to transfer electrons out of the cell. Another option was that the ancestor may have had both types of metabolism, and the various branches of the family lost certain abilities. We investigated these possibilities by analyzing the conservation and evolutionary history of genes for energy metabolism. Conservation of pathways for acetate oxidation In the respiratory species G. sulfurreducens , G. metallireducens , G. uraniireducens , and D . acetoxidans , acetate is the primary electron donor and it is oxidized via the TCA cycle, generating NADH, NADPH, and reduced ferredoxin (Figure 2 ) [ 21 - 23 ]. In contrast, Pelobacter species are incapable of acetate oxidation [ 12 ]. One protein family predicted to be acetate transporters (GSU0518) [ 24 ] was conserved in all of the acetate-oxidizing species (see Additional file 7 ). However, neither this protein nor any of the other acetate transporters found in G. sulfurreducens (GSU1068, GSU1070, and GSU2352) were conserved in either of the Pelobacter species (see Additional file 7 ). Twenty one genes encode the enzymes of the TCA cycle used by G. sulfurreducens [ 18 ] (Figure 2 , see Additional file 7 ). There was full conservation in all of the respiratory species of at least one copy of each of these enzymes (see Additional file 7 ). The one exception was the malate dehydrogenase in D. acetoxidans . There was no gene predicted to encode a malate dehydrogenase in the incomplete version of the genome, but this enzyme activity is found in D. acetoxidans [ 23 ]. As with the acetate transporters, not all of the genetic redundancy seen in G. sulfurreducens in the enzymes of the TCA cycle was conserved across the family. Only one copy of the aconitase (GSU1660) and the keto/oxoacid ferredoxin oxidoreductase (GSU1859–GSU1862) were conserved in all the respiratory species (see Additional file 7 ). Figure 2 The pathways of energy metabolism conserved in the respiratory Geobacteraceae species . Shown are the pathways, based those characterized in G. sulfurreducens , for acetate activation and oxidation via the TCA cycle in the cytoplasm, for inner membrane oxidation of TCA cycle products coupled with electron and proton transport, and for ATP generation. The genes encoding the enzymes of these pathways and their conservation pattern in all of the Geobacteraceae genomes are given in see Additional file 7 . The enzymes are colored black if there were orthologs for every subunit in all of the respiratory species. The asterisk indicates that malate dehydrogenase activity has been shown in D. acetoxidans , though genes for the enzyme were not found in the draft sequence of the genome. The pathway for electron transfer through the periplasm and out to the external electron acceptor is not well characterized in these species, and is represented here by the three cytochromes known to be required in vivo in G . sulfurreducens that are also conserved in all of the respiratory species. The Pelobacter species contained orthologs to many of the TCA cycle enzymes (see Additional file 7 ), but there were two notable exceptions. NADPH metabolism appears to catalyzed by different enzymes in the Pelobacter species. A monomeric-type isocitrate dehydrogenase (GSU1465) was found in all of the respiratory species (see Additional file 8 ), but was not conserved in either Pelobacter species (see Additional file 7 ). In acetate oxidizers, this reaction is the primary source of NADPH in the cell [ 25 ]. Instead, the Pelobacter species contained non-orthologous isocitrate dehydrogenase genes (Ppro_0452 and Pcar_1038 in cluster 3107) of the more common, homodimeric type [ 25 ]. In addition, both Pelobacter species lacked the dehydrogenase (GSU0509–GSU0510) believed to transfer the electrons from NADPH into the electron transport chain [ 26 ]. This enzyme was conserved in all four of the respiratory species (see Additional file 7 ). These changes suggest that the NADPH produced by the Pelobacter species may not be used primarily as a source of electrons for energy metabolism, as it is for the respiratory species. Furthermore, P. carbinolicus contained a non-orthologous fumarase enzyme (Pcar_0324) predicted to be of the class II, Fe(III)-free type [ 27 ], rather than the class I, Fe(III) type found in the other Geobacteraceae (GSU0994). Conservation of pathways for extracellular electron transfer In the respiratory Geobacteraceae species, the reducing equivalents from the TCA cycle are passed to inner membrane quinones via NADH dehydrogenase (Figure 2 ) [ 21 - 23 ]. This is predicted to be the only step in the electron transport chain where protons are pumped across the inner membrane for ATP generation [ 18 ]. G. sulfurreducens encodes two NADH dehydrogenase operons, one with 12 subunits and one with 14 (see Additional file 7 ). The 14-subunit enzyme is conserved in all of the respiratory species and P. propionicus (see Additional file 7 ). In addition, P. propionicus appears to have recently duplicated this enzyme, there are three virtually identical copies in the genome (Ppro_0628–Ppro_0641, Ppro_1623–Ppro_1636, Ppro_3180–Ppro_3193). In contrast, P. carbinolicus lacks the 14-subunit enzyme, but encodes an ortholog to the 12-subunit enzyme (see Additional file 7 ). This conservation pattern indicates that the 14-subunit enzyme may be the more important for inner membrane proton and electron transport; it is conserved in the four respiratory species, and in the only fermentative species predicted to use an NADH dehydrogenase (as described below and in Figure 3 ). Figure 3 The pathways of energy metabolism in Pelobacter propionicus and the differences from respiratory Geobacteraceae species . Shown are the pathways of butanediol and acetoin fermentation used by P. propionicus . Vertically-inherited enzymes with orthologs in the other Geobacteraceae are shown in black. Both the vertically-inherited NADH dehydrogenase and ATP synthase have been duplicated in P. propionicus and these are shown in dark grey. Enzymes acquired by lateral gene transfer are shown in red: 1) butanediol dehydrogenase, 2) acetoin dehydrogenase, 3) pyruvate:ferredoxin oxidoreductase, 4) transcarboxylase, and 5) methylmalonyl-CoA mutase, and an ATP synthase. Enzymes conserved in the respiratory Geobacteraceae but lost in P. propionicus are shown in white. Reactions with multiple enzymes encoded in the genome for which there are diverse conservation patterns are shown as dashed arrows. The final part of the electron transport chain in the respiratory species is transfer of electrons from the inner membrane to cytochromes for transport to the cell surface. The differences between the respiratory and fermentative species are substantial in this part of the pathway. The genomes of all four of the respiratory species contained at least one copy of a putative cytochrome bc complex (GSU2932–GSU2934), which is predicted to transfer electrons from membrane-bound quinones to periplasmic cytochromes [ 28 ] (Figure 2 , see Additional file 7 ). Neither of the Pelobacter species had any orthologs to this complex. C -type cytochromes span the periplasm and outer membrane in the respiratory species [ 1 ]. All of the Geobacter and Desulfuromonas genomes contained more than 100 ORFs that have the heme-binding motif [ 29 ] characteristic of c -type cytochromes (see Additional file 1 ). Both Pelobacter genomes had far fewer cytochrome genes, ca. 40 per genome (see Additional file 1 ). Many of the proteins found only in the respiratory species were multi-heme, c -type cytochromes (see Additional file 4 ), whereas none of the proteins specific to the fermentative species were multi-heme (see Additional file 5 ). In addition, both Pelobacter species lacked most of the specific cytochromes required for electron transfer in G. sulfurreducens in vivo . These include those conserved in all of the respiratory species: MacA, a c -type cytochrome associated with the inner membrane [ 30 ] and OmcB, an outer-membrane cytochrome [ 31 ]. Lateral transfer of fermentation genes Thus, while the Pelobacter species contained many vertically-inherited genes for respiratory metabolism, several key enzymes are different or missing in both species. To investigate evolution of the fermentative and syntrophic metabolism, Pelobacter -specific genes were analyzed, and genes originating from lateral gene transfer were identified using a combination of phylogenetic and BLAST-based analysis (see Additional file 9 ). Both Pelobacter species catabolize butanediol and acetoin [ 12 ]. The butanediol dehydrogenase (Bdh, Pcar_0330) and acetoin dehydrogenase (AcoABCL, Pcar_0343–Pcar0346), which catalyze the initial steps in the metabolism of these compounds, have been characterized in P. carbinolicus [ 32 ]. P. propionicus has genes with high sequence similarity to these, but the operon structure of the putative acetoin dehydrogenase included a duplication of the A and B subunits (Bdh, Ppro_1043 and AcoABCABL, Ppro_1024–Ppro1029, see Additional file 1 ). Phylogenetic analysis showed that the most closely related enzymes are not from Geobacter or Desulfuromonas , nor any other delta - Proteobacteria species (see Additional file 10 ). Instead the Pelobacter genes were most closely related to the butanediol catabolic genes from Pseudomonas [ 33 ] and Gram-positive species, indicating that these genes originated from lateral gene transfer (see Additional file 10 ). The subsequent steps in the catabolism of the acetyl-CoA and acetaldehyde from acetoin are markedly different in the two Pelobacter species. The pathway in P. propionicus is cyclic and requires membrane-bound electron transport enzymes (Figure 3 ) [ 34 ]. Several of the key enzymes were predicted to have been acquired by lateral gene transfer. Pyruvate:ferredoxin oxidoreductase is required to convert acetyl-CoA to pyruvate (Figure 3 ), and there were two heterotetrameric pyruvate:ferredoxin oxidoreductases from lateral gene transfer. One (Ppro_0322–Ppro_0325, see Additional file 1 ) is most closely related to the enzymes from Clostridia tetani and Thermotoga species, and the other (Ppro_0469–Ppro_0472, see Additional file 1 ) is most similar to the enzyme in Syntrophobacter and several archaeal species. Pyruvate is then converted to succinyl-CoA to regenerate NAD + without the need for a cooperating hydrogen-oxidizing species (Figure 3 ). P. propionicus has orthologs of four TCA cycle enzymes and the NADH dehydrogenase from the respiratory Geobacteraceae that could carry out these reactions (Figure 3 , see Additional file 7 ). Then, propionate is generated via a methylmalonyl-CoA mutase and transcarboxylase (Figure 3 ). Both of these enzyme were predicted to have been acquired from lateral gene transfer. The mutase (Ppro_1284–Ppro_1285, see Additional file 1 ) is related to enzymes from Chlorflexus and Geobacillus species, and the transcarboxylase (Ppro_0033–Ppro_0034, see Additional file 1 ) is related to that crystallized from Propionibacterium freudenreichii [ 35 ]. Thus, it appears that P. propionicus uses a mosaic of vertically inherited and laterally acquired genes for fermentation. Interestingly, the pathway uses a complete, vertically inherited, respiratory electron transport chain – the succinate dehydrogenase operates in reverse as a fumarate reductase, accepting reducing equivalents from the NADH dehydrogenase [ 34 ]. This is equivalent to that used by Geobacter species growing by fumarate respiration [ 18 , 36 ]. Differences between Pelobacter species associated with syntrophy P. carbinolicus ferments via much simpler pathway of cytoplasmic enzymes (Figure 4 ). As described above, the butanediol dehydrogenase and acetoin dehydrogenase appear to have been acquired by lateral gene transfer. NAD + is regenerated either by ethanol production or by proton reduction to hydrogen, which requires a syntrophic partner and ATP is made by substrate level phosphorylation by an acetate kinase [ 12 ]. Figure 4 The pathways of energy metabolism in Pelobacter carbinolicus and the differences from respiratory Geobacteraceae species . Shown are the pathways of butanediol and acetoin ethanol fermentation and syntrophy used by P. carbinolicus . Vertically-inherited enzymes with orthologs in the other Geobacteraceae are shown in black. The vertically-inherited ATP synthase has been duplicated in P. carbinolicus and is shown in dark grey. Enzymes acquired by lateral gene transfer are shown in red: 1) butanediol dehydrogenase, 2) acetoin dehydrogenase, 3) hydrogenases, the RNF electron transfer complex, and possibly an ATP synthase (see text). Enzymes conserved in the respiratory Geobacteraceae but lost in P. carbinolicus are shown in white. Reactions with multiple enzymes encoded in the genome for which there are diverse conservation patterns are shown as dashed arrows. Some Pelobacter species live in syntrophic associations with methanogens by partially oxidizing organic compounds to hydrogen and acetate which the methanogens consume [ 12 , 13 ]. Both Pelobacter species lacked orthologs to any of the Geobacter -type [ 37 ] hydrogenases (GSU0121–GSU0123, GSU0782–GSU0785, GSU2417–GSU2420 or GSU2718–GSU2722, see Additional file 1 ). Furthermore, the hydrogenases in the two Pelobacter species are dissimilar. This may reflect the fact that P. carbinolicus is capable of growing syntrophically with a hydrogen-consuming partner, most typically a methanogen [ 12 ]. Reducing equivalents can be disposed with the production of hydrogen, as long as the methanogen maintains hydrogen concentrations low enough for hydrogen production to be thermodynamically favorable. P. propionicus does not grow in this manner [ 12 ]. P. carbinolicus has two highly-similar, four-subunit hydrogenases (Pcar_1602–Pcar_1605 and Pcar_1633–Pcar_1636) related to the cytoplasmic, hydrogen-producing HndABCD hydrogenase in Desulfovibrio fructosovorans [ 38 ]. P. propionicus , encodes two highly-similar, 12-subunit enzymes (Ppro_0587–Ppro_0598 and Ppro_3521–Ppro_3532) that are related to the membrane-bound, hydrogen-producing, NiFe hydrogenase in Pyrococcus furiosus [ 39 ]. Approximately -45 kJ/mol is predicted to be available to P. carbinolicus and its methanogenic partner when they grow by ethanol oxidation under syntrophic conditions [ 13 ]. This is less than what is predicted to be required for generation of an ATP by substrate-level phosphorylation, and it is assumed that some of the ATP from acetyl-CoA conversion must be reinvested to drive the overall reaction, possibly by reverse electron transport [ 13 ]. It has been suggested that a Rnf-type ion-translocating electron transfer complex may play a role in reverse electron transport in syntrophs [ 40 ], by generating low potential ferredoxin from NADH using the membrane potential [ 41 ]. P. carbinolicus contains an Rnf gene cluster (Pcar_0260–Pcar_0265, Figure 4 ) that is similar to the enzymes from Rhodobacter and Clostridium species, both predicted to be involved in reduction of ferredoxin by NADH [ 42 ]. P. propionicus , which does not grow syntrophically, contains no Rnf-like genes. Both Pelobacter genomes contain unusual numbers and types of ATP synthases, which have been suggested to be important for maintenance of the membrane potential [ 40 ]. Both contain the ATP synthase that is conserved in all of the Geobacteraceae , but these genes appear to have recently been duplicated in both organisms (Pcar_0015–Pcar_0016/Pcar_3130–Pcar_3136, Pcar_0944–Pcar_0952, and Ppro_0599–Ppro_0607, Ppro_1500–Ppro_1508, Figures 3 and 4 ). In addition, both Pelobacter species also have a third ATP synthase encoded in operons organized like that of Methanosarcina barkeri [ 43 ], though they were predicted to have been acquired from different sources. The genes in P. propionicus (Ppro_844–Ppro851) are most closely related to enzymes from species of gamma Proteobacteria : Hahella , Legionella , and Methylococcus . The P. carbinolicus genes (Pcar_2989–Pcar2997) are only distantly related to the Geobacter and Desulfuromonas operons, but because they are similar to genes from a variety of species including D. acetoxidans , it is difficult to be confident about the lateral transfer source."
} | 6,818 |
21211826 | null | s2 | 729 | {
"abstract": "Mine tailings in arid and semi-arid environments are barren of vegetation and subject to eolian dispersion and water erosion. Revegetation is a cost-effective strategy to reduce erosion processes and has wide public acceptance. A major cost of revegetation is the addition of amendments, such as compost, to allow plant establishment. In this paper we explore whether arbuscular mycorrhizal fungi (AMF) can help support plant growth in tailings at a reduced compost concentration. A greenhouse experiment was performed to determine the effects of three AMF inocula on biomass, shoot accumulation of heavy metals, and changes in the rhizosphere microbial community structure of the native plant Prosopis juliflora (mesquite). Plants were grown in an acidic lead/zinc mine tailings amended with 10% (w/w) compost amendment, which is slightly sub-optimal for plant growth in these tailings. After two months, AMF-inoculated plants showed increased dry biomass and root length (p<0.05) and effective AMF colonization compared to controls grown in uninoculated compost-amended tailings. Mesquite shoot tissue lead and zinc concentrations did not exceed domestic animal toxicity limits regardless of whether AMF inoculation was used. The rhizosphere microbial community structure was assessed using denaturing gradient gel electrophoresis (DGGE) profiles of the small subunit RNA gene for bacteria and fungi. Canonical correspondence analysis (CCA) of DGGE profiles showed that the rhizosphere fungal community structure at the end of the experiment was significantly different from the community structure in the tailings, compost, and AMF inocula prior to planting. Further, CCA showed that AMF inoculation significantly influenced the development of both the fungal and bacterial rhizosphere community structures after two months. The changes observed in the rhizosphere microbial community structure may be either a direct effect of the AMF inocula, caused by changes in plant physiology induced by AMF, or a combination of both mechanisms."
} | 509 |
36505707 | PMC9682621 | pmc | 731 | {
"abstract": "The advent of memristors and the continuing research and development in the field of brain-inspired computing could allow realization of a veritable “thinking machine”. In this study, ZnO-based memristors were fabricated using a radio frequency magnetron sputtering method. The ZnO oxide layer was prepared by incorporating silver nanocrystals (NCs). Several synaptic functions, i.e. nonlinear transmission characteristics, short-term potentiation, long-term potentiation/depression, and pair-pulse facilitation, were imitated in the memristor successfully. Furthermore, the transition from synaptic behaviors to bipolar resistive switching behaviors of the device was also observed under repeated stimulus. It is speculated that the switching mechanism is due to the formation and rupture of the conductive Ag filaments and the corresponding electrochemical metallization. The experimental results demonstrate that the Ag/Ag:ZnO/Pt memristor with resistive switching and several synaptic behaviors has a potential application in neuromorphic computing and data storage systems.",
"conclusion": "4. Conclusions In summary, we have verified that synaptic simulation and nonvolatile resistive switching properties were achieved in one device of Ag/Ag:ZnO/Pt through changing compliance current. Several functions of bio-synapse, including nonlinear transmission characteristics, STP, LTP, LTD, PPF were imitated. After increasing the compliance current, the nonvolatile bipolar resistive switching was observed. We can infer that both volatile and nonvolatile switching mechanisms are due to Ag electrochemical metallization and the corresponding formation and rupture of the conductive filaments. The results indicate that the as-fabricated devices possess potential application in the neuromorphic and data storage systems.",
"introduction": "1. Introduction After the resistor, capacitor, and inductor, as the fourth type of passive fundamental electronic device, the memristor can realize 0 and 1 information storage functions, and has the advantages of non-volatility, high speed, low energy consumption, high endurance, simple structure, and compatibility with CMOS technology. 1 It has wide application prospects. It can not only meet the performance requirements of the next generation of high-density information storage and high-performance computing for general-purpose electronic memory, but also realize the functions of non-volatile state logic operation and brain-like morphology calculation. As for electronic memory, resistive random access memory (RRAM) has been deeply investigated and is predicted to become the future general memory device. 2 Transition-metal oxides (ZnO, HfO 2 , TiO 2 , Ta 2 O 5 , Ga 2 O 3 , CuO, NiO, and others) are promising information storage medium for RRAM. 3,4 They are usually fabricated using solution methods 5 and vacuum methods such as sputter, 6 pulse laser deposition, 7 atomic layer deposition, 8 and so on. In recent years, the memristor based on transition-metal-oxide has been put forward as a competitive candidate to imitate synaptic functions in bionic neuromorphic systems, because of their structural similarity with sandwiches, gradually or suddenly changing resistance, low energy consumption and convenience for intensive 3D integration. 9,10 In biological synapse, several ions ( e.g. , Ca 2+ , Na + , K + ) decide the release of neurotransmitters from presynapse to postsynapse. 11 Similar to biological synapses, the conductance of oxide-based memristors can be changed by the movement of cations ( e.g. , Ti 4+ , Ag + ) in the oxide layer, 12–16 which can vividly imitate the dynamic mechanism of neural synapses. Up to now, several synaptic functions, for example, nonlinear transmission characteristics, short-term potentiation (STP), long-term potentiation (LTP), long-term depression (LTD), and pair-pulse facilitation (PPF), 16–18 have been realized with the oxide-based memristors. Moreover, many oxide-based memristor can imitate bio-synapse and reveal nonvolatile resistive switching characteristics at a single device, which is very important for the construction of bionic neuromorphic system. 3,12,16 ZnO-based memristor is one of the earliest studied. 5,6,19–21 Because ZnO is a wide band gap (3.37 eV) semiconductor, which has stable chemical stability, no pollution and low price. Utilizing ZnO as a resistive switching layer may still present some interesting performance. In this study, we manufactured ZnO-based memristor using a full sputter method. And the ZnO layer was prepared by incorporating the Ag NCs. Which can play a charge trapping role in the memristor structure or it can be responsible for the localization and improvement of the stability of the conductive filament or it can play a part in the formation of the conductive filament under applied bias. 22 Multiple features including synaptic characteristics and nonvolatile bipolar resistive switching were achieved successively in one memristor device. These results prove the characteristics of ZnO-based memristor for both neuromorphic computing and data storage systems.",
"discussion": "3. Results and discussion \n Fig. 1(a) and (b) show the schematics illustration of Ag/Ag:ZnO/Pt memristor device and an essential biological neural system, respectively. Interestingly, the Ag top electrode can imitate the presynaptic neuron, the ZnO layer imitates the presynaptic cleft and the Pt bottom electrode imitates the postsynaptic neuron, respectively. As can be seen from Fig. 1(b) , a presynaptic neuron (PRE) links a postsynaptic neuron (POST) through a synapse. While a PRE sends a spike, the action potential is communicated by its axon to the biological synapse, and lots of neurotransmitter governs the synaptic weight. 23,24 The biological synapse can be imitated by a memristor device, where the resistance between the top electrode (TE) and the bottom electrode (BE) is dominated by the defect or dopant distribution in a switching layer, such as TiO 2 , HfO 2 , ZnO etc. 2,6,15 Fig. 1(c) presents the cross-sectional scanning electron microscopy (SEM) image of the ZnO film. As can be seen, the total thickness is about 100 nm. Fig. 1(d) presents SEM-energy dispersive X-ray (SEM-EDX) elemental mapping of Zn, Ag, Pt, Ti and Si, respectively. And Fig. 1(e) shows the distribution of the Ag element. Which reveals that the Zn and Ag as main elements are homogeneously distributed in the entire region ( Fig. 1(d) and (e) ). Fig. 1(f) shows the EDX spectrum taken from Fig. 1(c) . The result illustrates the chemical element composition of the sample with silver, platinum, titanium, oxygen and zinc present in the spectra. In fact, synapses can be regarded as two-terminal devices, which have unique nonlinear transmission characteristics. The strength of the connection between neurons determines the transmission efficiency, which can dynamically change with the training of stimulation signals or inhibition signals, and keep a continuous changing state. Memristor has the characteristic that its resistance can change continuously with the electric quantity flowing through it. This nonlinear electrical characteristic is highly similar to the nonlinear transmission characteristic of synapse. The applied direct current (DC) voltage was warily used to prevent sudden transformation in resistance state from high to low. As shown in Fig. 2(a) and (b) , when 10 continuous scanning positive voltage (0 to 0.8 V) and negative voltage (0 to −0.4 V) are applied to the device, the current will continuously increase or decrease. It can be seen that with the increase of bias voltage, the resistance of the device decreases gradually. And the resistance of the device in the next scan is lower than that in the previous scan. When the reverse voltage is applied, the resistance of the device gradually increases. The change of current shows the characteristic of nonlinear transmission with the scanning of voltage, and shows the trend of continuous change with multiple scans. In order to clearly illustrate the changing trend, the curves of current and voltage versus time are plotted again in Fig. 2(c) . In addition, the change of resistance value after each scan also clearly shows that it gradually decreases with the positive voltage, and vice versa ( Fig. 2(d) ). If we regard the conductance of the device as the synaptic weight, the above results are similar to the nonlinear transmission characteristics of biological synapses. Fig. 2 Memristor properties and its simulation as a synapse. Measured I – V properties of the device (a) positive part and (b) negative part. (c) The current and voltage data versus time for the device in (a) and (b) highlighting the change in current in sequential voltage sweeps. (d) Changes of the device conductance during consecutive sweeps. The conductance can be gradually modulated by applying a series of programmed pulses. As can be seen in Fig. 3 , by applying 50 positive pulses (+1 V, 1 ms) and 50 negative pulses (−1 V, 1 ms) to the memristor, it is found that the device conductance continuously increases and decreases. The modulation of conductance can be regarded as a consequence of the migration of Ag ions caused by the electric field. 17,25 Based on the memristor, the LTP and LTD behavior of synapses is simulated. Fig. 3 The device conductance can be gradually increased or decreased by continuous potentiating or depressing pulses. The inset is diagram of 100 pulses. P, 1 V, 1 ms; D, −1 V, 1 ms; R, 0.1 V, 1 ms. The function of STP is very important to the execution of computational actions in the biological neural system. Which can be imitated through the volatile properties of memristor. 16,17 Fig. 4 exhibits the property of volatile threshold switching (TS) in the Ag/Ag:ZnO/Pt device under DC voltage sweeping with very small compliance current (500 nA). When the positive voltage applied on TE of the device, the current increases quickly at the threshold voltage (0.4 V), and then the device switches to the low resistance state (LRS). With the voltage falling below a certain value (0.2 V), the LRS reverses back to original high resistance state (HRS) spontaneously. It can be inferred that the interfacial energy minimization leads to spontaneous fracture of the slim Ag filament, which contributes to the volatile TS property. 16 Furthermore, conductive filaments have been found directly by in situ transmission electron microscopy. 15,16,26 Fig. 4 The threshold switching (TS) I – V loops with low compliance current (500 nA). \n Fig. 5(a) describes the PPF function in a biological synapse. That is, under the stimulation of two consecutive spikes (orange line), the second postsynaptic response current becomes larger than the first (green line), and the interval between the two spikes is less than the recovery time. 17 Similarly, as shown in Fig. 5(b) , in our device, the PPF function can be simulated by two continuous pulses (1 V, 1.2 ms) with transient time separation (1 ms). The second response current of stimulus pulse is obvious greater than the first (green line). This is because the relaxation time of Ag atoms in RS layer is longer than the interval time between two sequential pulse stimulations. So the conductance of the device increases accordingly. 17 Fig. 5 (a) The diagram of PPF phenomenon in a biological synapse. (b) The PPF property of the device. While the device is tested under a higher compliance current (CC) of 10 μA, a transition from synaptic simulation to nonvolatile resistive switching behavior can be obtained. It can be seen from Fig. 6(a) that an obvious bipolar resistive switching behavior is observed in the Ag/Ag:ZnO/Pt device. The device current reaches the CC at a voltage of about 0.3 V, and the device switches from the HRS to the LRS. Different from the LRS obtained under a lower CC (500 nA) shown in Fig. 4(a) , the conductive LRS obtained under the CC of 10 μA can be maintained even after the imposed positive voltage disappears, which indicates the nonvolatile resistive switching behavior of the Ag/Ag:ZnO/Pt device. Afterwards, by sweeping the voltage from zero to a negative voltage, a sudden drop of the device current can be observed at a reset voltage of about −0.2 V, and the device can be switched back to the HRS. In order to further investigate the effect of the CC on the resistive switching behavior, higher CCs of 25 μA, 50 μA, 100 μA are applied in order. Fig. 6(b)–(d) show the similar typical I – V characteristics of the Ag/Ag:ZnO/Pt device under the next three CCs. Interestingly, different from Fig. 6(a) and (b) , a gradual drop of the device current can be observed at a reset voltage of about −0.3 V in Fig. 6(c) and (d) , which suggests that the sudden and gradual reset can be converted in a single Ag/Ag:ZnO/Pt device by controlling different CC. These results confirm that Ag + ions can migrate into ZnO film to form metallic conductive filaments across the oxide film, and thus the device switches between the HRS and the LRS. In addition, the Ag NCs inserted in the ZnO film can enhance the electric field intensity, which facilitates the formation and rupture of the Ag conductive filament and keeps the performance of the device stable, as reported in many previous work. 2,6,27 Fig. 6 (a)–(d) I – V hysteresis curves with different I CC values for Ag/Ag:ZnO/Pt RRAM device, showing the sharp reset converted to the gradual reset while I CC reached 50 μA. To investigate the RS parameters of the device under four CCs, the switching voltage, current, and resistance of the HRS and the LRS were analyzed statistically. Fig. 7(a) and (b) show the data of statistic of the R HRS and R LRS , respectively. The data of maximum and minimum fluctuation is obvious. The R HRS value fluctuated as (4170 ± 640 kΩ), (4470 ± 360 kΩ), (1120 ± 90 kΩ), and (540 ± 9 kΩ), and the R LRS value fluctuated as (19.35 ± 2.5 kΩ), (7.75 ± 0.7 kΩ), (6.45 ± 0.5 kΩ), and (2.69 ± 0.1 kΩ) under the four CCs, respectively. As can be seen, generally, the higher the CC imposed on the device, the lower the resistance value in HRS or LRS. And the on/off ratio always kept ∼10 2 . Obviously, the CC plays a significant role in RS characteristic. As CC increases, the conductive filament becomes thicker, and the resistance of LRS decreases. Maybe the conductive filaments can't completely disappear during the reset process, which will also greatly reduce the resistance of HRS. Fig. 7(c) and (d) show the statistic data of V set / I set and V reset / I reset under the four CCs, respectively. Both I set and I reset exhibit an increasing trend when the CC increases, while the V set and V reset are only fluctuated in a small range, V set between 0.3 and 0.4 V and V reset between −0.2 and −0.4 V, respectively. The measurement results show that the Ag/Ag:ZnO/Pt device can be applied into both synaptic emulation and bipolar resistive switching memory, making us more possible to fabricate a true “thinking machine”. Fig. 7 While I CC increased, (a) R HRS and (b) R LRS decreased; (c) I set and V set increased; (d) I reset increased. In addition, under 100 μA CC, the endurance and retention ability of the device were measured through applying pulse stimuli, as shown in Fig. 8 . A series of write/erase cycles stimulated by short pulses, namely, 5 V, 1 ms bias for writing and −5 V, 1 ms bias for erasing, respectively. The measured result of the device is very stable. Which confirms the stability of switching, and showing this switching is feasible for future practical memory application. Fig. 8 (a) Endurance and (b) retention ability of the device using short pulses (±5 V, 1 ms)."
} | 3,932 |
35519853 | PMC9064577 | pmc | 734 | {
"abstract": "Fabricating amphiphobic surfaces is often complex, difficult to control, and time-consuming, making the fabrication process very difficult. Herein, a facile and time-saving modification method using a mixed modified solution of stearic acid and perfluorooctanoic acid was initially proposed, which played a key role in the achievement of the superhydrophobicity and highly oleophobicity. The effects of reaction time on surface morphology and wettability as well as the content of perfluorooctanoic acid in the mixed modified solution on wettability were investigated to determine the optimal experiment parameters that maximized the amphiphobicity of the surfaces. The as-fabricated amphiphobic surfaces displayed high oil contact angles of 133.5°, higher water contact angles of 156.8°, ultra-low water sliding angles of less than 5° and excellent self-cleaning properties. The facile, easy to control, and efficient method can provide new insights into fabricating amphiphobic surfaces and can open up a new way for the basic research and practical application of amphiphobic surfaces.",
"conclusion": "Conclusions We proposed a facile and time-saving modification method using a mixed modified solution of stearic acid and perfluorooctanoic acid of fabricating amphiphobic surfaces on copper substrates. The effects of reaction time on surface morphology and wettability as well as the content of PFOA in the mixed modified solution on wettability were investigated. The results revealed that when the reaction time was 15 min and the PFOA content in the mixed modified solution was 60%, the as-fabricated surfaces displayed optimal amphiphobicity with high oil contact angles of 133.5°, higher water contact angles of 156.8° and ultra-low water sliding angles of 3°, which suggested superhydrophobicity and highly oleophobicity. Furthermore, the amphiphobic surface possessed an outstanding self-cleaning property, which enabled the rolling water droplets to remove chalk powder without any traces being left behind. The facile, easy to control, and efficient method can provide new insights into fabricating amphiphobic surfaces and can open up a new way for the basic research and practical application of amphiphobic surfaces.",
"introduction": "Introduction The hydrophilicity and oleophilicity of most metals, such as copper, limit their application in many special environments. 1 Superhydrophobic surfaces have been fabricated on copper substrates by etching, 2–5 electrodeposition, 6–8 femtosecond laser, 9,10 etc. , which can repel water but lack oleophobicity at the same time, resulting in the surfaces being readily polluted by low surface energy oils. 11,12 Transforming the amphiphilicity of copper substrates to amphiphobicity can avoid oil and water contamination. 13,14 Amphiphobic surfaces with unique wettability of being both hydrophobic and oleophobic 15 are widely used in the fields of self-cleaning, 16 anti-fouling, 17,18 anti-corrosion, 19,20 etc. , which have become a new research focus. Recently, researchers have developed several methods on how to fabricate amphiphobic surfaces. Nevertheless, most methods require special experimental conditions and equipment. For example, Yin et al. 21 prepared superhydrophobic and oleophobic copper meshes by immersing samples into a mixed aqueous solution of NaOH and (NH 4 ) 2 S 2 O 8 for 30 min, modifying the etched copper in a 1 H ,1 H ,2 H ,2 H -perfluorooctyltrichlorosilane solution for 2 h, then setting the experimental conditions to 100 °C for 1 h, and annealing the samples in an oven. Only through these complicated procedures and special experimental conditions, the low surface energy groups could be successfully assembled on the rough surfaces to reduce the surface energy. Guo et al. 22 created copper mesh with superhydrophobic and oleophobic properties via etching samples in a mixed acid solution of HCl and CH 3 COOH for 24 h, modification with a 1 H ,1 H ,2 H ,2 H -perfluorooctyltriethoxysilane solution for up to 12 h, a self-assembled monolayer can be formed under special acid–base reaction conditions, and then the high-temperature experimental conditions were set to heat the samples in an oven for 2 h to achieve the purpose of reducing the surface energy. Lee et al. 23 fabricated non-wettable surface through etching copper in an ammonium ambient solution of NaOH and (NH 4 ) 2 S 2 O 8 , heating at 180 °C for 60 min, and modifying with 1 H ,1 H ,2 H ,2 H -perfluorooctyl under vacuum for 2 h, and only by using special vacuum experimental conditions, the low surface energy can be achieved. These fabricating methods which required special experimental conditions and equipment added the difficulty of fabrication and control of experimental conditions. Moreover, modifying the surface with a low-surface energy material was time consuming, which limited the large-scale production of amphiphobic surfaces. To overcome these problems, it has become an urgent need to propose a facile, easy to control, and efficient method for fabricating amphiphobic surfaces on copper substrates with a mixed modified solution. The lowest known surface energy materials can be arranged in descending order as CH > CH 2 > CH 3 > CF 2 > CF 3 . 24,25 Stearic acid (STA, CH 3 (CH 2 ) 16 COOH) with 16 –CH 2 groups and 1 –CH 3 group is a low surface energy material, but it only possesses hydrophobicity and not oleophobicity. Perfluorooctanoic acid (PFOA, CF 3 (CF 2 ) 6 COOH) has 6 –CF 2 groups and 1 –CF 3 group, wherein a –CF 3 group has the lowest surface energy of 6.7 mN m −1 . 26 Hence, PFOA is hydrophobic and oleophobic with a quite lower surface energy. In addition, the reactive group –COOH in both STA and PFOA can react with metals and their compounds to bond low surface energy groups to the surface to achieve the purpose of reducing the surface energy. 27,28 It can be confirmed that a mixed modified solution composed of STA and PFOA can improve the hydrophobicity due to the STA and achieve the oleophobicity due to the PFOA. Thereby, mixing STA and PFOA at a reasonable ratio to form a mixed modified solution can shorten the modification time to obtain amphiphobic surfaces. To the best of our knowledge, the method of modifying the surface using a mixture of two low surface energy materials to fabricate amphiphobic surfaces on copper substrates has not been reported. In this study, we created micro–nano hierarchical dendritic structures by immersing copper substrates into a silver nitrate aqueous solution and proposed a facile and time-saving modification method using a mixed modified solution of STA and PFOA for reducing surface energy to fabricate amphiphobic surfaces on copper substrates. The effects of reaction time on surface morphology and wettability as well as the content of PFOA in the mixed modified solution on wettability were investigated to determine the optimal experiment parameters that maximized the amphiphobicity of the surfaces. Furthermore, the self-cleaning property of the as-fabricated amphiphobic surfaces was also studied. The method designed here was facile, easy to control, and efficient, which can provide new insights into fabricating amphiphobic surfaces with a mixture of two low surface energy materials and can open up a new way for the basic research and practical application of amphiphobic surfaces.",
"discussion": "Results and discussion Surface morphology The surface morphologies under different magnifications obtained when the polished Cu substrate was immersed into an AgNO 3 aqueous solution for different times are exhibited in Fig. 1 . The morphology of the bare copper substrate prior to AgNO 3 immersion is shown in Fig. S2, † from which only scratches on the surface can be observed, which is different from those after being immersed into an AgNO 3 aqueous solution. As shown in Fig. 1a–d , the surface morphologies were mainly composed of dendritic structures. Fig. 1e–h exhibit that under high magnification the feature structures of the surface morphologies mainly included micro–nano hierarchical dendritic structures with the width in the order of hundreds of nanometers and lengths in the order of microns of the main branches, and the width and length in the order of hundreds of nanometers of the small dendrites, growing from the main branches. Besides, there were a small number of irregular multi-sided cubic crystal structures ( Fig. 1e–h ). Fig. 1 SEM images of polished Cu substrate at different reaction times in an aqueous solution of AgNO 3 under different magnifications: (a) and (e) 5 min, (b) and (f) 10 min, (c) and (g) 15 min, (d) and (h) 20 min. When the immersion time was short (5 min), as shown in Fig. 1a and e , the size of the dendritic structures was small. As the immersion time increased, the dendritic structures grew continuously and the size became larger. Extending the time to 15 min ( Fig. 1c and g ), micro–nano hierarchical dendritic structures were clearly visible, and the size significantly increased; therefore, the surface roughness greatly improved. When the time was increased to 20 min ( Fig. 1d and h ), the micro–nano hierarchical dendritic structures had changed. Although the size of the micro-scale structures still increased, the size of the nano-scale structures gradually decreased. The micro–nano hierarchical dendritic structures became insignificant hierarchical dendritic structures dominated by the micro-scale, which made the surface roughness decrease. It was concluded that when the immersion time was 15 min, micro–nano hierarchical dendritic structures were significant and the surface morphology was the roughest. The large water–air interface area can effectively prevent water or oil droplets from penetrating the surface and form stable Cassie state. 29 Surface chemistry The chemical composition of the surfaces under different conditions was characterized by XRD, EDS and FT-IR. Fig. 2 presents the XRD patterns of the polished Cu substrate and Ag@Cu surface (after the polished Cu substrate was immersed into the AgNO 3 aqueous solution). As shown in Fig. 2a , the polished Cu substrate exhibited diffraction peaks of Cu (111), Cu (200) and Cu (220) at 43.4°, 50.53° and 74.16°, 30 respectively. Except for the three diffraction peaks of the polished Cu substrate, four diffraction peaks of Ag (111), Ag (200), Ag (220) and Ag (311) appeared at 38.18°, 44.38°, 64.53° and 77.48° in the XRD pattern of the Ag@Cu surface ( Fig. 2b ), 30,31 respectively. This implied that the polished Cu substrate was successfully covered with a silver film after being immersed into the AgNO 3 aqueous solution. In addition, the diffraction peaks at 29.48°, 36.34°, 42.39°, and 61.43° were attributed to Cu 2 O (110), Cu 2 O (111), Cu 2 O (200) and Cu 2 O (220), 30,32 which were the reason for the irregular multi-sided cubic crystal structures of the surface morphology ( Fig. 1 ). We concluded that both Ag and Cu 2 O were formed on the polished Cu substrate after being immersed into the AgNO 3 aqueous solution. The related reaction between Cu substrate and the AgNO 3 aqueous solution can be further explained by eqn (1) and (2) . 33 1 Cu(sol) + Ag + (aq) → Cu 2+ (aq) + Ag(sol) 2 Cu(sol) +Cu 2+ (aq) + H 2 O(liq) → Cu 2 O(sol) + 2H + (aq) Fig. 2 XRD patterns of (a) polished Cu substrate and (b) Ag@Cu surface. EDS analysis was carried out to study the chemical composition of the polished Cu substrate, Ag@Cu surface, STA@Ag@Cu surface (after the Ag@Cu surface was modified by STA), PFOA@Ag@Cu surface (after the Ag@Cu surface was modified by PFOA) and STA&PFOA@Ag@Cu surface (after the Ag@Cu surface was modified by a mixture of STA and PFOA). As shown in Fig. S3a, † the EDS spectrum of the polished Cu substrate contained only elemental Cu. The elements Ag and O were observed in the EDS spectra of the Ag@Cu surface (Fig. S3b † ). Combined with Fig. 2b , it was confirmed that the silver film covered the polished Cu substrate and the elemental O was from Cu 2 O. Related to Fig. S3b, † elemental C appeared in the EDS spectrum of the STA@Ag@Cu surface (Fig. S3c † ), which demonstrated that the Ag@Cu surface may be successfully modified by STA. The elements C and F in Fig. S3d † were from PFOA, which indicated that the Ag@Cu surface may be successfully modified by PFOA. Fig. S3e † exhibits the elements C and F, which showed that the mixture of STA and PFOA may successfully modify the Ag@Cu surface. This conclusion was further verified by the following FT-IR spectra test. To further confirm whether the surface was successfully modified, FT-IR spectra were obtained, and are shown in Fig. S4. † After being modified by STA, the adsorption peaks at approximately 2918 cm −1 and 2848 cm −1 were assigned to –CH 3 and –CH 2 stretching vibrations, 34,35 respectively (Fig. S4a † ). In combination with Fig. S3c, † it was believed that the STA@Ag@Cu surface had STA. After being modified by PFOA, the bands at approximately 1206 cm −1 and 1149 cm −1 were attributed to the stretching vibrations of the –CF 2 and –CF 3 groups (Fig. S4b † ). 36 In combination with Fig. S3d, † PFOA was shown to be successfully bound to Ag@Cu surface. After being modified by the mixture of STA and PFOA, not only –CH 2 and –CH 3 groups, but also –CF 2 and –CF 3 groups were observed in Fig. S4c. † In combination with Fig. S3e, † we can confirm that the STA&PFOA@Ag@Cu surface had STA and PFOA. Furthermore, the peaks at approximately 1744 cm −1 and 1640 cm −1 arose from the stretching vibration of the coordinated COO– groups. 37 In addition, a broad band at approximately 3500 cm −1 arose from an –OH stretching vibration. 38 Surface wettability The wettability of droplets on a solid surface depends mainly on two factors: surface morphology and surface chemistry. 39 The effects of reaction time and the PFOA content on the wettability of different surfaces were investigated. The molar ratio of PFOA in the mixed modified solution was expressed as M F . The wettability of the amphiphobic surface was evaluated by water and blend oil. Besides, the repellency to daily droplets such as tea and milk was also explored. Effect of reaction time on amphiphobicity The polished Cu substrates were immersed into an AgNO 3 aqueous solution for different times (5, 10, 15 and 20 min), and further modified for 1 h to determine the effect of the reaction time on the CA and SA for STA@Ag@Cu ( M F = 0%), PFOA@Ag@Cu ( M F = 100%) and 4 intermediate mixtures surfaces ( M F = 20%, 40%, 60%, 80%). The effect of the reaction time on the CA for STA@Ag@Cu, PFOA@Ag@Cu and 4 intermediate mixtures surfaces is shown in Fig. 3 and 4 . It was observed that the WCA were all greater than 150° ( Fig. 3 ), while the OCA varied in the range of about 50° to 140° at different reaction times ( Fig. 4 ). Both WCA and OCA showed an upward trend with the increase of reaction time from 5 to 15 min under the same PFOA content. The WCA increased from 159.6° to 161.4° ( M F = 0%), 158.9° to 160.5° ( M F = 20%), 154.5° to 159.6° ( M F = 40%), 152.9° to 156.8° ( M F = 60%), 152.4° to 155.8° ( M F = 80%), 150.9° to 152.9° ( M F = 100%), respectively, and the range was within 5°. The OCA increased from 49.8° to 52.8° ( M F = 0%), from 53.1° to 110.6° ( M F = 20%), from 117.4° to 124.7° ( M F = 40%), from 118.2° to 133.5° ( M F = 60%), from 134.4° to 137.2° ( M F = 80%), and from 141.9° to 145.5° ( M F = 100%), respectively, and the range was around 5–70° (see Table S1 † ). This indicated that the reaction time had a greater impact on the OCA than the WCA. Besides, the WCA was larger than the OCA at each time point, because the surface tension of oil is much smaller than that of water. 1 However, there was a certain downward trend at 20 min. This indicated that the CA achieved the maximum when the reaction time was 15 min, at which time a large amount of air was trapped in the gaps generated in the hierarchical dendritic structures. This was in keeping with the analysis of the surface morphology presented in Section 3.1. Fig. 3 Effects of reaction time and M F on WCA for STA@Ag@Cu ( M F = 0%), PFOA@Ag@Cu ( M F = 100%) and 4 intermediate mixtures surfaces ( M F = 20%, 40%, 60%, 80%). Fig. 4 Effects of reaction time and M F on OCA for STA@Ag@Cu ( M F = 0%), PFOA@Ag@Cu ( M F = 100%) and 4 intermediate mixtures surfaces ( M F = 20%, 40%, 60%, 80%). The effect of the reaction time on WSA for STA@Ag@Cu, PFOA@Ag@Cu and 4 intermediate mixtures surfaces can be observed from Fig. 5 . Except 20 min, the WSA were all less than 5° when the M F was less than or equal to 60%, in addition, WCA were all larger than 150° ( Fig. 3 ), indicating that the surfaces reached superhydrophobicity. Nevertheless, when M F was more than 60%, WSA suddenly increased. Once M F reached 100%, water droplets adhered to the surfaces, and WSA was recorded as 180°. For 20 min, WSA was less than 10° only when M F was 0%, while water droplets adhered to the surfaces when M F was more than 0%. The greater adhesion attributed to proper surface morphologies was not present ( Fig. 1d and h ). Fig. 5 Effects of reaction time and M F on WSA for STA@Ag@Cu ( M F = 0%), PFOA@Ag@Cu ( M F = 100%) and 4 intermediate mixtures surfaces ( M F = 20%, 40%, 60%, 80%). Effect of PFOA content on the CA and SA The effect of PFOA content on the CA and SA for STA@Ag@Cu, PFOA@Ag@Cu and 4 intermediate mixtures surfaces as explored through modified the Ag@Cu surface with different mixed modified solutions for 1 h. \n Fig. 3 and 4 also exhibit the effect of PFOA content on the CA. The WCA were all larger than 150° under the same reaction time with the increase of M F , presenting a downward trend, whereas the OCA showed an upward trend, making the surfaces change from superoleophilic to oleophobic ( Fig. 4 ). Except 5 min, the OCA was larger than 90° when the M F was more than or equal to 20%, showing that the surfaces reached oleophobicity and when the M F was more than or equal to 60%, the OCA exceeded 130°, displaying that the oleophobicity improved. When the reaction time was the same and the PFOA content increased from 0 to 100%, the WCA decreased from 159.6° to 150.9° (5 min), 159.9° to 151.5° (10 min), 161.4° to 152.9° (15 min), 158.6° to 152.3° (20 min), respectively, and the range was within 10°. The OCA increased from 49.8° to 141.9° (5 min), 50.4° to 141.7° (10 min), and 52.8° to 145.5° (15 min), 55.2° to 148.7° (20 min), respectively, the range of variation reached around 90° (see Table S2 † ). This revealed that the PFOA content made greater influence on OCA than WCA. The reason was that with the M F increased, under the same reaction times more oleophobic groups –CF 3 and –CF 2 can be bonded to the surface, so the oleophobicity improved continuously. In addition, when the content of PFOA was the same and the reaction time increased from 5 to 15 min, the range of the OCA was around 5–70°. While when the reaction time was the same and the content of PFOA ranged from 0 to 100%, the range of the OCA reached around 90°. It can be inferred that the influence of PFOA content on the OCA was greater than the reaction time. Therefore, the amphiphobicity mainly depended on the PFOA content in the mixed modified solution. Wettability of the amphiphobic surface From the analysis of Sections 3.3.1 and 3.3.2, under the same PFOA content, when the reaction time was 15 min, the WCA reached the maximum, the WSA was smaller, and the OCA was larger. Hence 15 min can be determined as the optimal reaction time parameter. The optimal parameter of the PFOA content at 15 min was further investigated. Fig. 6 demonstrates the relationship between the CA and PFOA content as well as the corresponding optical images of the CA at 15 min. With the increase of M F , the WCA decreased slowly but were all larger than 150°, while the OCA increased continuously. When the M F was 60%, the OCA exceeded 130° and the WSA was about 3° ( Fig. 7 and Video S1 † ). While once the M F was greater than 60%, WSA suddenly increased to be greater than 10° ( Fig. 5 ). As a result, the M F of 60% was determined as the optimal parameter. Fig. 6 Variation of CA with M F at 15 min reaction time for STA@Ag@Cu ( M F = 0%), PFOA@Ag@Cu ( M F = 100%) and 4 intermediate mixtures surfaces ( M F = 20%, 40%, 60%, 80%). Fig. 7 Photographs of measuring the WSA of the amphiphobic surface. Consequently, amphiphobic surfaces with a high OCA of 133.5° ( Fig. 6 ), a higher WCA of 156.8° ( Fig. 6 ) and a small WSA of 3° ( Fig. 7 ), possessed superhydrophobicity and oleophobicity were fabricated when the reaction time was 15 min and the PFOA content was 60%, which were the optimal experimental parameters. This indicated that they were super repellent to water and highly repellent to blend oil, exhibiting the typical characteristics of amphiphobicity. 36 Images of the droplets on substrates treated with different conditions are displayed in Fig. 8 . Water droplets and blend oil droplets were dyed with methyl blue and Sudan red, respectively. The blend oil droplets spread over the substrate when droplets were dropped onto the polished Cu substrate, which indicated that the polished Cu substrate was hydrophilic and oleophilic ( Fig. 8a ). The water droplets appeared spherical on the STA@Ag@Cu surface, while the blend oil droplets spread on the surface, which indicated that the STA@Ag@Cu surface was hydrophobic and oleophilic ( Fig. 8b ). The PFOA@Ag@Cu surface could support the water droplets and the blend oil droplets, which proved that the PFOA@Ag@Cu surface was hydrophobic and oleophobic ( Fig. 8c ). However, the greater adhesion of the water droplets on the surface made the PFOA@Ag@Cu surface incapable of achieving superhydrophobicity (Video S2 † ). The water droplets and blend oil droplets could stand on the STA&PFOA@Ag@Cu surface ( Fig. 8d ). When compared with the PFOA@Ag@Cu surface, the WSA of the STA&PFOA@Ag@Cu surface was quite small, which further demonstrated that the surface was an amphiphobic surface with superhydrophobicity and oleophobicity. Furthermore, the STA&PFOA@Ag@Cu surface also exhibited excellent repellency to tea droplets and milk droplets that present certain lyophobic properties. Fig. 8 Different liquid droplets on (a) polished Cu substrate, (b) STA@Ag@Cu surface, (c) PFOA@Ag@Cu surface, (d) STA&PFOA@Ag@Cu surface. Self-cleaning property of the amphiphobic surface The outstanding self-cleaning property of the amphiphobic surface can protect the substrate from contamination and serve to protect the material or surface from corrosive reagents. 6 Based on this, we tested the self-cleaning property of the amphiphobic surface. Fig. 9 shows the process of self-cleaning. First, the amphiphobic surface with a small angle of inclination was evenly smeared by a layer of white chalk power ( Fig. 9a ). After that, once the water droplets were dropped onto the surface, the white chalk power was removed immediately as the water droplets rolled off ( Fig. 9b and c ), owing to the small WSA of the amphiphobic surface. A clear cleaned trajectory was formed on the surface without leaving any trace of the chalk residue ( Fig. 9d ), which testified that the surface had an excellent self-cleaning property. The entire process was recorded in Video S3. † Fig. 9 Self-cleaning process of the amphiphobic surface (a) before, (b and c) during and (d) after water droplets cleaning."
} | 5,879 |
37859795 | PMC10582064 | pmc | 735 | {
"abstract": "In a microbial electrolysis cell (MEC), the oxidization of organic compounds is facilitated by an electrogenic biofilm on the anode surface. The biofilm community composition determines the function of the system. Both deterministic and stochastic factors affect the community, but the relative importance of different factors is poorly understood. Anode material is a deterministic factor as materials with different properties may select for different microorganisms. Ecological drift is a stochastic factor, which is amplified by dispersal limitation between communities. Here, we compared the effects of three anode materials (graphene, carbon cloth, and nickel) with the effect of dispersal limitation on the function and biofilm community assembly. Twelve MECs were operated for 56 days in four hydraulically connected loops and shotgun metagenomic sequencing was used to analyse the microbial community composition on the anode surfaces at the end of the experiment. The anode material was the most important factor affecting the performance of the MECs, explaining 54–80 % of the variance observed in peak current density, total electric charge generation, and start-up lag time, while dispersal limitation explained 10–16 % of the variance. Carbon cloth anodes had the highest current generation and shortest lag time. However, dispersal limitation was the most important factor affecting microbial community structure, explaining 61–98 % of the variance in community diversity, evenness, and the relative abundance of the most abundant taxa, while anode material explained 0–20 % of the variance. The biofilms contained nine Desulfobacterota metagenome-assembled genomes (MAGs), which made up 64–89 % of the communities and were likely responsible for electricity generation in the MECs. Different MAGs dominated in different MECs. Particularly two different genotypes related to Geobacter benzoatilyticus competed for dominance on the anodes and reached relative abundances up to 83 %. The winning genotype was the same in all MECs that were hydraulically connected irrespective of anode material used.",
"conclusion": "5 Conclusions Four replicate hydraulic loops consisting of three MECs, each with a different anode material, were operated for 56 days. The anode materials consisted of carbon cloth, graphene, and nickel. MECs with carbon cloth had the lowest lag time as well as the highest current generation and total charge generated of the three anode materials. The microbial community analysis showed that Geobacter spp. dominated all anode biofilms; however, the species that dominated and the diversity and evenness of the communities differed between MECs. MECs placed within the same hydraulic loop had more similar community composition that MECs in different loops, which showed that stochastic factors such as initial colonization and drift affected community structure.",
"introduction": "1 Introduction Microbial electrolysis cells (MECs) have been explored for various applications such as biosensing, recovery of resources from waste streams, and production of chemicals and energy carriers [ 1 , 2 ]. Electrogenic microorganisms, such as those from the Desulfobacterota phylum (formerly known as Deltaproteobacteria ) and more specifically Geobacter species, are typically involved in the breakdown of organic material and the following transportation of electrons [ [3] , [4] , [5] ]. These microorganisms can be utilized in systems such MECs and other microbial electrochemical technologies (METs) to generate current through the transportation of electrons to the anode surface. The electrons are then transferred to the cathode, where they can be used by the cathode for reduction reactions. These reduction reactions typically lead to the recovery of different resources, e.g., hydrogen or methane gas [ [6] , [7] , [8] , [9] ]. Various factors affect the microbial community composition and the function of METs. Deterministic factors such as substrate composition, electrode materials, and system design will select for specific species, which are fit to survive in the system [ 5 , 10 , 11 ]. Stochastic factors such as ecological drift and diversification will cause differences in microbial community composition and function between systems that operate under the same environmental conditions. For example, Zhou et al. [ 12 ] showed that stochastic initial colonization caused differences in both community structure and function in MECs operated under identical conditions. Dispersal limitation increases stochasticity and leads to differences in microbial communities exposed to identical selection pressures whereas high dispersal rates reduces community differences [ 13 ]. The relative importance of deterministic and stochastic factors in shaping the microbial communities and how this relates with the function of the METs under various conditions is still not known. The anode material is one factor that affects the microbial community and electrical current generation in METs. Conventionally, METs use carbon-based materials since metals tend to corrode in aquatic environments [ 14 , 15 ]. Some commonly used materials are carbon paper, carbon felt and carbon cloth [ 16 , 17 ]. Graphene is a highly conductive two-dimensional carbon-based material, which has been shown to have a large surface area and that could be produced at low cost [ [18] , [19] , [20] ]. Some previous studies have used graphene-based electrodes in METs with promising results. For example, they were shown to increase the power density in microbial fuel cells and the hydrogen production in MECs in comparison to conventional materials [ [21] , [22] , [23] ]. In other systems, graphene covered surface has been shown to have an antimicrobial effect. In forward osmosis membranes for water treatment, graphene coating has reduced biofouling [ [24] , [25] , [26] ]. The physiochemical and antimicrobial properties of some forms of graphene have also been highlighted within biomedical applications, such as in preparation of substrates for tissue engineering as well as drug delivery [ 27 , 28 ]. In METs, it is unknown how graphene-based anodes affect the microbial community composition in comparison to other materials and if the deterministic effect of the anode material is large or small in comparison to stochastic factors. The aim of this study was to determine if differences in anode materials affect current generation and microbial community composition in single-chamber MECs, and if this effect is larger or smaller than stochastic factors. We compared three anode materials: carbon cloth, graphene, and nickel. We examined stochastic factors by enabling homogenizing dispersal between some MECs and completely blocking dispersal between others. We found that the anode material was the dominating factor affecting the lag phase and current generation in the MECs, but stochastic factors were more important for microbial community assembly of biofilms on the anode surfaces.",
"discussion": "4 Discussion 4.1 Current generation There were differences observed in the initial current generation between the different materials. Carbon cloth had a much shorter start-up time in comparison to both graphene and nickel. All four replicates for carbon cloth also had similar start-up times while the other materials showed a large variation. Carbon cloth was the most hydrophobic material ( Fig. S2 , Supplementary material), which may have facilitated the initial attachment of microbial cells [ 51 ]. The slow start-up time and variation of graphene could be attributed to the structure of the graphene. The graphene used in this setup consists of sharp vertical flakes. When the initial colonization of the surface occurs, the bacteria may be pierced by the graphene flakes leading to cell death [ 52 ]. The build-up of organic material by the previous bacteria that tried to colonize the surface allows for subsequent bacteria to attach without being pierced by the sharp graphene flakes [ 52 , 53 ]. The variation can also be explained by the stochastic process involved in the initial colonization of the surface [ 54 ]. Some studies have shown the importance of stochastic processes in the community assembly within MECs [ 12 ] and in other systems such as anammox biofilms [ 55 ]. Current density and charge generation varied between the different anode materials. The carbon cloth MECs had the highest current density in all hydraulic loops. The graphene MECs were second best, except in hydraulic loop 2 ( Fig. 2 ). Typically, the current generation in the MEC decreased somewhat after it had reached its peak and was then relatively stable for the remainder of the experimental run. In the early stages of microbial assembly on the anode surface there is a thin layer of biofilm. This allows bacteria to have direct physical contact with the anode surface as well as easy access to the organic substrates in the liquid, resulting in high current generation. Once the biofilm becomes thicker, as both electrogenic and non-electrogenic bacteria attach onto the biofilm, the access to organic compounds become diffusion-limited [ 56 ]. Competition for the organic substrates and nutrients between the electrogenic and non-electrogenic bacteria also limits the access to the substrates. There are, however, methods that some of the electrogenic bacteria could potentially employ to overcome the limited access to organic substrates. Geobacter sulfurreducens , a bacterium commonly found in METs, has been shown to use structures such as nanowires, c-type cytochromes, as well as mediators to aide its transfer of electrons to the anode [ 57 , 58 ]. This would allow electrogens not directly connected to the anode surface to utilize its ability to transfer electrons and contribute to the overall current generation. There is a number of studies that have highlighted the improved performance of METs with graphene electrodes [ 21 , 59 ], as well as some studies that have highlighted the antimicrobial abilities of graphene in other applications [ 26 , 27 , 60 ]. It seems that the biocompatibility of graphene is dependent on the structure of the graphene sheets used as well as the graphene manufacturing process [ 15 , 61 , 62 ]. A reason for the reduced performance of the graphene MECs in this study may be due to a layer of dead cells closest to the electrode surface. This layer could result in the lower ability of the graphene to produce equally high levels of current as the carbon cloth. A potential explanation for this could be that not only are the dead cells preventing the direct contact of the viable electrogens with the electrode surface, but they are also limiting the use of other methods to facilitate electron transfer to the electrode surface. The greater performance of the carbon cloth MECs could be attributed to the structure of the carbon cloth, which has been shown to have a good biocompatibility [ 63 , 64 ]. Nickel is known to have a good electrical conductivity and has been shown to improve performance in MFC when used to reinforce graphite electrodes [ 65 ]. However, it was observed in this study that the nickel MECs, by itself did not perform as well as the carbon cloth MECs. 4.2 Microbial community diversity and composition on the anodes The diversity indices showed that although 40–107 MAGs could be detected on the bioanode, most had low relative abundance. The 1 D and 2 D indices can be interpreted as the number of common and abundant taxa, respectively [ 66 ]. The low values, 1.9–7.4 for 1 D and 1.3–3.2 for 2 D, as well as the low evenness ( Fig. 4 ) show that the biofilm communities are dominated by a few MAGs with high relative abundance. The microbial community was dominated by Desulfobacterota, which are known as electrogens often present in bioelectrochemical systems [ 67 ]. The two most abundant MAGs were most closely related to Geobacter benzoatilyticus , which was recently isolated from petroleum-contaminated soil. G. benzoatilyticus was shown to reduce ferrihydrite, which suggests it can use solid electron acceptors. It was also shown to oxidize acetate, but not propionate [ 49 ]. The two dominating MAGs may, thus, use acetate as electron donor for generating electrical current on the anode surfaces. Trichloromonas species, which were identified to be similar to Desulfuromonas acetexigens , were also present. Trichloromonas have been shown to be capable of electron transfer, indicating they are electrogenic bacteria [ 68 ]. For instance, Desulfuromonas acetexigens have been shown to be involved in the current generation on graphite electrodes through its ability to use acetate as an electron donor [ 69 ] . Propionate and butyrate in the NM are likely used by fermentative bacteria. Several fermentative bacteria were found in the biofilms; for instance, Succinispira mobilis [ 70 ] and MAGs within Oscillospiraceae and Lachnospiraceae . 4.3 Ecological mechanisms All anodes were dominated by Desulfobacterota species, which is consistent with previous studies showing that bioelectrochemical systems having an acetate-containing feed with near-neutral pH and low to moderate salinity and temperature are highly selective for members of this phylum, particularly Geobacter [ 9 ]. However, the community structure and the species that dominate in each anode biofilm appears to be governed by stochastic factors such as initial colonization and drift. In the PCoA, the samples separated more based on hydraulic loop than based on anode material, especially when more emphasis is placed on taxa with high relative abundance ( Fig. 5 f–g). Typically, it might be expected that different microorganisms would randomly attach and colonize the different anode materials within the same hydraulic loop. Instead, it was observed that the anode communities within the same loop had a more similar microbial community structure than those with the same anode material in different loops. This indicates a potential influence of the MEC that was first colonized in each hydraulic loop. The carbon cloth anodes had the shortest lag times and generally the highest DNA concentrations ( Fig. 2 ; Table S1 ), suggesting that they were colonized early in each loop and developed the thickest biofilms. Based on this, it is likely that dispersal of microorganism from the carbon cloth anode resulted in the colonization of the other MECs in the same hydraulic loop. The dominance analysis showed that variation in the most abundant electrogene as well as diversity and evenness metrics were mainly due to hydraulic loop and not material, underscoring that stochasticity was involved in the microbial assembly and development of the microbial communities. Anode biofilms that were present in the same system, i.e., the same hydraulic loop, were subjected to homogenizing dispersal and had higher similarity in community composition than anode biofilms in separate hydraulic loops."
} | 3,745 |
28680387 | PMC5478701 | pmc | 737 | {
"abstract": "An ongoing challenge in neuromorphic computing is to devise general and computationally efficient models of inference and learning which are compatible with the spatial and temporal constraints of the brain. One increasingly popular and successful approach is to take inspiration from inference and learning algorithms used in deep neural networks. However, the workhorse of deep learning, the gradient descent Gradient Back Propagation (BP) rule, often relies on the immediate availability of network-wide information stored with high-precision memory during learning, and precise operations that are difficult to realize in neuromorphic hardware. Remarkably, recent work showed that exact backpropagated gradients are not essential for learning deep representations. Building on these results, we demonstrate an event-driven random BP (eRBP) rule that uses an error-modulated synaptic plasticity for learning deep representations. Using a two-compartment Leaky Integrate & Fire (I&F) neuron, the rule requires only one addition and two comparisons for each synaptic weight, making it very suitable for implementation in digital or mixed-signal neuromorphic hardware. Our results show that using eRBP, deep representations are rapidly learned, achieving classification accuracies on permutation invariant datasets comparable to those obtained in artificial neural network simulations on GPUs, while being robust to neural and synaptic state quantizations during learning.",
"conclusion": "4. Conclusions and future directions This article demonstrates a local, event-based synaptic plasticity rule for deep, feed-forward neural networks achieving classification accuracies on par with those obtained using equivalent machine learning algorithms. The learning rule combines two features: (1) Algorithmic simplicity: one addition and two comparisons per synaptic update provided one auxiliary state per neuron and (2) Locality: all the information for the weight update is available at each neuron and the synapse. The combination of these two features enables synaptic plasticity dynamics for neuromorphic deep learning machines. Our results lay out a key component for the building blocks of spike-based deep learning using neural and synaptic operations largely demonstrated in existing neuromorphic technology (Chicca et al., 2013 ; Park et al., 2014 ; Merolla et al., 2014 ). Together with the near SynOp-MAC parity observed in the learning experiments compared to GPUs (Figure 5 ), we can reasonably expect real-time deep learning machines that operate on at least 100x to 1,000x smaller energy budget compared to current GPU technologies. One limitation eRBP is related to the “loop duration,” i.e., the duration necessary from the input onset to a stable response in the error neurons. This duration scales with the number of layers, raising the question whether eRBP can generalize for very deep networks without impractical delays. Future work currently in investigation is to augment eRBP using recently proposed synthetic gradients (Jaderberg et al., 2016 ; Czarnecki et al., 2017 ), whereby gradients are estimated before the output neurons respond. This technique has been successfully tested with feedback alignment and direct feedback alignment, and thus has high chances of success using eRBP. It can be reasonably expected that the deep learning community will uncover many variants of random BP, including in recurrent neural networks for sequence learning and memory augmented neural networks. In tandem with these developments, we envision that such RBP techniques will enable the embedded learning of pattern recognition, attention, working memory, and action selection mechanisms which promise transformative hardware architectures for embedded computing. This work has focused on unstructured, feed-forward neural networks and a single benchmark task across multiple implementations for ease of comparison. Limitations in deep learning algorithms are often invisible on “toy” datasets like MNIST (Liao et al., 2015 ). Existing literature suggests that that random BP could also work for unsupervised learning (e.g., using autoencoders, Lee et al., 2014 ) in deeper and convolutional networks, as well as more difficult datasets such as CIFAR10. RLandom BP was demonstrated to be effective in a variety of tasks and network structures (Liao et al., 2015 ; Baldi et al., 2016 ), including convolutional neural networks (Baldi et al., 2016 ). In principle, we do not see major roadblocks in applying eRBP to spike-based convolutional neural networks, provided that the neuromorphic architecture can support weight sharing at the level of the feature.",
"introduction": "1. Introduction Biological neurons and synapses can provide the blueprint for inference and learning machines that are potentially 1,000-fold more energy efficient than mainstream computers. However, the breadth of application and scale of present-day neuromorphic hardware remains limited, mainly by a lack of general and efficient inference and learning algorithms compliant with the spatial and temporal constraints of the brain. Thanks to their general-purpose, modular, and fault-tolerant nature, deep neural networks and machine learning has become a popular and effective means for executing a broad set of practical vision, audition and control tasks in neuromorphic hardware (Esser et al., 2016 ; Lee et al., 2016 ; Neftci E. et al., 2016 ). One outstanding question is whether the learning phase in deep neural networks can be efficiently carried out in neuromorphic hardware as well. Performing such learning on-the-fly is appealing in less controlled environments where no prior, representative dataset exists, and can confer more fine-grained context awareness to behaving cognitive agents. However, the workhorse of deep learning, the gradient descent BP rule, commonly relies on the immediate availability of network-wide information stored with high-precision memory. In digital computers, the access to this information funnels through the von Neumann bottleneck, which dictates the fundamental limits of the computing substrate. Distributing computations along multiple cores in GPUs is an effective solution to mitigate this problem, but even there the scalability of gradient backpropagation in neural networks can sometimes be limited by its data and memory-intensive operations (Seide et al., 2014 ; Zhu et al., 2016 ), and more so in the case of fully connected networks (Seide et al., 2014 ). The implementation of Gradient Back Propagation (hereafter BP for short) on a neural substrate is even more challenging (Grossberg, 1987 ; Baldi et al., 2016 ; Lee et al., 2016 ) because it requires (1) using synaptic weights that are identical with forward passes (symmetric weights requirements, also known as the weight transport problem), (2) carrying out the operations involved in BP including multiplications with derivatives and activation functions, (3) propagating error signals with high, floating-point precision, (4) alternating between forward and backward passes, (5) changing the sign of synaptic weights, and (6) availability of targets (labels). While some recent work in neural networks shows that error signals can be propagated using low precision (sometimes down to 1 bit; Courbariaux and Bengio, 2016 ; Rastegari et al., 2016 ), the essence of these challenges is that BP often requires precise linear and non-linear transformations and information that is not local to the computational building blocks in a neural substrate, meaning that special communication channels must be provisioned (Baldi and Sadowski, 2015 ). Whether a given operation is local or not depends on the physical implementation that carries out the computations. For example, while symmetric weights in neural networks are compatible with von Neumann architectures (and even desirable since weights in both directions are shared), the same is not true in a distributed system such as the brain: elementary computing units do not have bidirectional connections with the same weight in each direction. Since neuromorphic implementations generally assume dynamics closely related to the those in the brain, requirements (1–4) above also hinder efficient implementations of BP in neuromorphic hardware. Although, previous work (Lee et al., 2016 ; Lillicrap et al., 2016 ; O'Connor and Welling, 2016 ) overcomes some of the fundamental difficulties of gradient BP listed above in spiking networks, here we tackle all of the key difficulties using event-driven random BP (eRBP), a synaptic plasticity rule for deep spiking neural networks achieving classification accuracies that are similar to those obtained in artificial neural networks, potentially running on a fraction of the energy budget with dedicated neuromorphic hardware. eRBP builds on the recent advances in approximate forms of the gradient BP rule (Lee et al., 2014 ; Liao et al., 2015 ; Baldi et al., 2016 ; Lillicrap et al., 2016 ) for training spiking neurons of the type used in neuromorphic hardware to perform supervised learning. These approximations solve the non-locality problem by replacing weights in the backpropagation phase with random ones, leading to remarkably little loss in classification performance on benchmark tasks (Baldi et al., 2016 ; Lillicrap et al., 2016 ) (requirement 1 above). Although, a general theoretical understanding of random BP (RBP) is still a subject of intense research, extended simulations and analyses of linear networks show that, during learning, the network adjusts its feed-forward weights such that they align with the (random) feedback weights, which is arguably equally good in communicating gradients. eRBP is an asynchronous (event-driven) adaptation of random BP that can be tightly embedded with the dynamics of dual compartment I&F neurons that costs one addition and two comparisons per synaptic weight update. Extended experimentations show that the spiking nature of neuromorphic hardware and the lack of general linear and non-linear computations at the neuron does not prevent accurate learning on classification tasks (requirement 2, 3), and operates continuously and asynchronously without alternation of forward or backward passes (requirement 4). Additional experimental evidence shows that eRBP is robust to fixed width representations of the synaptic weights, making it suitable for dedicated neuromorphic hardware. The focus of eRBP is to achieve real-time, online learning at higher power efficiency compared to deep learning on standard hardware, rather than achieving the highest accuracy on a given task. The success of eRBP on these measures lays out the foundations of neuromorphic deep learning machines, and paves the way for learning with streaming spike-event data in neuromorphic platforms at proficiencies close to those of artificial neural networks. This article is organized as follows: key theoretical and simulation results are provided in the results sections, followed by a general discussion and conclusion. Technical details of eRBP and its implementation are provided as the final section.",
"discussion": "3. Discussion The gradient descent BP rule is a powerful algorithm that is ubiquitous in deep learning, but when implemented in a neuromorphic substrate, it relies on the immediate availability of network-wide information stored with high-precision memory. More specifically, (Baldi et al., 2016 ) and (Lee et al., 2016 ) list several reasons why the following requirements of gradient BP make them biologically implausible. The essence of these difficulties is that gradient BP is non-local in space and in time when implemented on a neural substrate, and requires precise linear and non-linear computations. The feedback alignment work demonstrated that symmetric weights were not necessary for communicating error signals across layers (Lillicrap et al., 2016 ). Here we demonstrated a learning rule inspired by feedback alignment, and membrane voltage-gated plasticity rules, and three-factor synaptic plasticity rules proposed in the computational neuroscience literature. With an adequate network architecture, we find that the spike-based computations and the lack of general linear and non-linear computations and alternating forward and backward steps does not prevent accurate learning. Although, previous work overcome some of the non-locality problems of gradient BP (Lee et al., 2016 ; Lillicrap et al., 2016 ; O'Connor and Welling, 2016 ), eRBP overcomes all of the key difficulties using a simple rule that incurs one addition and two comparisons per synaptic weight update. Taken together, our results suggest that general-purpose deep learning using streaming spike-event data in neuromorphic platforms at artificial neural network proficiencies is realizable. Our experiments target neuromorphic implementations of spiking neural networks with embedded plasticity. Membrane-voltage based learning rules implemented in mixed-signal neuromorphic hardware (Qiao et al., 2015 ; Huayaney et al., 2016 ) are compatible with eRBP provided that synaptic weight updates can be modulated by an external signal on a neuron-to-neuron basis. Following this route, and combined with the recent advances in neuromorphic engineering and emerging nanotechnologies, eRBP can become key to ultra low-power processing in space and power constrained platforms. 3.1. Why neuromorphic learning machines? Spiking neural networks, especially those based on the I&F neuron types severely restrict computations during learning and inference. With the wide availability of graphical processing units and future dedicated machine learning accelerators, the neuromorphic spike-based approach to learning machines is often heavily criticized as being misguided. While this may be true for some hardware designs and on metrics based on absolute accuracy at most standardized benchmark tasks, neuromorphic hardware dedicated for embedded learning can have distinctive advantages thanks to: (1) asynchronous, event-based communication, which considerably reduces the communication between distributed processes, (2) natural exploitation of “rate” codes and “spike” codes where single spikes are meaningful, leading to fast and thus power-efficient and gradual responses (Figure 4 , see also O'Connor and Welling, 2016 ), (3) on-line learning, which can naturally support continual (life-long) learning. In addition, the premise of neuromorphic engineering, i.e., that electronic and biological share similar constraints on communication, power, and reliability (Mead, 1990 ), also extends to the algorithmic domain. That is, accommodating machine learning algorithms within the constraints of ultra-low power hardware for adaptive behavior (i.e., embedded learning) is likely to result in solutions for communication, computations and reliability that are in close resemblance with the brain. The convergence between the two approaches (neuromorphic vs. artificial) will not only improve the design of neuromorphic learning machines, but can also widen the breadth of knowledge transfer between computational neuroscience and deep learning. Many examples that led to the unprecedented success in machine learning have substantial overlap with equivalent neural mechanisms, such as normalization (Ioffe and Szegedy, 2015 ; Ren et al., 2016 ), attention, short-term memory for learning complex tasks (Graves et al., 2014 ), and memory consolidation through fast replays for reinforcement learning (Mnih et al., 2015 ; Kumaran et al., 2016 ). One example relevant to the presented work is the Binarized Neural Network (BNN) (Courbariaux and Bengio, 2016 ). The BNN is trained such that weights and activities are −1 or 1, which considerably reduces the energetic footprint of inference, because multiplications are not necessary and the memory requirements for inference are much smaller. The discrete, quantized dynamics of I&F neurons shares many similarities with the BNN, such as binary activations (spikes), low-precision variables, and straight-through gradient estimators. Our neurally inspired approach has important and potentially advantageous differences with regard to binarized networks: network activity is sparse and data-driven (asynchronous), random variables for stochasticity are generated only when neurons spike, errors are backpropagated only for misclassified examples, and learning is ongoing leading to accurate, early, single-spike classification. 3.2. Relation to prior work in random backpropagation Our learning rule builds on the feedback alignment learning rule demonstrating that random feedback can deliver useful teaching signals by aligning the feed-forward weights with the feed-back weights (Lillicrap et al., 2016 ). The authors also demonstrated a spiking neural network implementing feedback alignment, demonstrating that feedback alignment is able to implicitly adapt to random feedback when the forward and backward pathways both operate continuously. However, their learning rule is not event-based as in eRBP, but operates in a continuous-time fashion that is not directly compatible with spike-driven plasticity, and a direct neuromorphic implementation thereof would be inadequate due to the high bandwidth communication required between neurons. Furthermore, their model is a spike response model that does not emulate the physical dynamics of spiking neurons such as I&F neurons. Another difference between eRBP and the network presented in Lillicrap et al. ( 2016 ) is that eRBP contains only one error-coding layer, whereas feedback alignment contains one error-coding layer per hidden layer. Such direct feedback alignment was recently proposed in Nø kland ( 2016 ) and Baldi et al. ( 2016 ), and theoretical analyses demonstrate that gradients computed in this fashion are within 90 degrees of the backpropagated gradient. Baldi et al. ( 2016 ) studied feedback alignment in the framework of local learning and the learning channel, and derived several other flavors of random BP such as adaptive, sparse, and indirect RBP, along with their combinations. In related work, Lee et al. ( 2014 ) showed how feedback weights can be learned to improve the classification accuracy by training the feedback weights to learn the inverse of the feedforward mapping. After initial submission of this article, Samadi et al. ( 2017 ) demonstrated a related learning rule using integrate and fire neurons. The application focus of our work is different from that of Samadi et al. ( 2017 ), and the learning algorithm has important differences: Samadi et al. ( 2017 ) uses trigonometric functions and updates at every timestep. In contrast, our work demonstrates learning in an end-to-end spike-driven fashion and realizable using only additions and comparisons. Also, we demonstrated a stochastic version of eRBP, which lead to significantly better accuracies on MNIST (2.02% vs. 2.95% for 1,000 hidden neurons total). 3.3. Relation to prior work in spiking deep neural networks Several approaches successfully realized the mapping of pre-trained artificial neural networks onto spiking neural networks using a firing rate code (O'Connor et al., 2013 ; Cao et al., 2014 ; Neftci et al., 2014 ; Das et al., 2015 ; Diehl et al., 2015 ; Hunsberger and Eliasmith, 2015 ; Marti et al., 2015 ; Esser et al., 2016 ; O'Connor and Welling, 2016 ) Such mapping techniques have the advantage that they can leverage the capabilities of existing machine learning frameworks such as Caffe (Jia et al., 2014 ) or Theano (Goodfellow et al., 2013 ) for brain-inspired computers. More recently, (Mostafa, 2016 ) used a temporal coding scheme where information is encoded in spike times instead of spike rates and the dynamics are cast in a differentiable form. As a result, the network can be trained using standard gradient descent to achieve very accurate, sparse and power-efficient classification. Although eRBP achieves comparable results, their approach naturally leads to sparse activity in the hidden layer which can be more advantageous in large and deep networks. An intermediate approach is to learn online with standard BP using spike-based quantization of network states (O'Connor and Welling, 2016 ) and the instantaneous firing rate of the neurons (Lee et al., 2016 ). O'Connor and Welling ( 2016 ) eschews neural dynamics and instead operates directly on event-based (spiking) quantizations of vectors. Using this representation, common neural network operations including online gradient BP are mapped on to basic addition, comparison, and indexing operations applied to streams of signed spikes. As in eRBP, their learning rule achieves better results when weight updates are made in an event-based fashion, as this allows the network to update its parameters many times during the processing of a single data sample. Lee et al. ( 2016 ) propose a method for training spiking neural networks via a formulation of the instantaneous firing rate of the neuron obtained by low-pass filtering the spikes. There, quantities that can be related to the postsynaptic potential (rather than mean rates) are used to compute the derivative of the activity of the neuron, which can provide a useful gradient for backpropagation. Esser et al. ( 2016 ) use multiple spiking convolutional networks trained offline to achieve near state-of-the-art classification in standard benchmark tasks. Their approach maps onto the all-digital spiking neural network architecture using trinary weights. For the above approaches, the eRBP learning rule presented here can be used as a drop-in replacement and can reduce the computational footprint of learning by simplifying the backpropagated chain path and by operating directly with locally available variables i.e., membrane potentials and spikes. 3.4. Relation to prior work in spike-driven plasticity rules STDP has been shown to be very powerful in a number of different models and tasks related to machine learning (Thorpe et al., 2001 ; Nessler et al., 2013 ; Neftci et al., 2014 ). Although, the implementation of acausal updates (triggered by presynaptic firing) is typically straightforward in cases where presynaptic lookup tables are used, the implementation of causal updates (triggered by postsynaptic firing) can be challenging due to the requirement of storing a reverse look-up table. Several approximations of STDP exist to solve this problem (Galluppi et al., 2014 ; Pedroni et al., 2016 ), but require dedicated circuits. Thus, there is considerable benefit in hardware implementations of synaptic plasticity rules that forego the causal updates. Such rules, which we referred to as spike-driven plasticity, can be consistent with STDP (Brader et al., 2007 ; Clopath et al., 2010 ; Qiao et al., 2015 ; Sheik et al., 2016a ), especially when using dynamical variables that are representative of the pre- and postsynaptic firing rates (such as calcium or average membrane voltage). A common feature among spike-driven learning rules is a modulation or gating with a variable that reflects the average firing rate of the neuron, for example through calcium concentration (Graupner and Brunel, 2012 ; Huayaney et al., 2016 ) or the membrane potential (Clopath et al., 2010 ; Sheik et al., 2016a ) or both Brader et al. ( 2007 ). Sheik et al. ( 2016a ) recently proposed a membrane-gated rule inspired by calcium and voltage-based rules with homeostasis for learning unsupervised spike pattern detection. Their rule statistically emulates pairwise STDP using presynaptic spike timing only and using additions and multiplications. Except for homeostasis, eRBP follows similar dynamics but potentiation and depression magnitudes are dynamic and determined by external modulation, and comparisons are made on total synaptic currents. The two compartment neuron model used in this work is motivated by conductance-based dynamics in Urbanczik and Senn ( 2014 ) and previous neuromorphic realizations of two compartment mixed signal spiking neurons Park et al. ( 2014 ). Although, the spiking network used in this work is current-based rather than conductance-based, eRBP shares strong similarities to the three-factor learning rule employed in Urbanczik and Senn ( 2014 ). The latter is composed of three factors: an approximation of the prediction error, the derivative of the membrane potential with respect to the synaptic weight, and a positive weighting function that stabilizes learning in certain scenarios. The first factor corresponds to the error modulation, while the second and third factors roughly correspond to the presynaptic activity and the derivative of the activation function. The differences between eRBP and (Urbanczik and Senn, 2014 ) (besides from the random BP which was considered in Lillicrap et al., 2016 ) stems mainly from two facts: (1) the firing rate description used here for simplicity and for easier comparisons between artificial neural networks and spiking neural networks and (2) eRBP is fully event-based in the sense that weights are updated only when the presynaptic neurons spike."
} | 6,290 |
35417235 | PMC9007498 | pmc | 738 | {
"abstract": "Venus flytrap and bladderwort, capable of rapid predation through a snapping transition, have inspired various designs of soft actuators and robots with fast actions. These designs, in contrast to their natural counterparts, often require a direct force or pressurization. Here, we report a bistable domal hydrogel structure capable of spontaneous and reversible snapping under an electric field. Unlike a mechanical force, the electric field does not drive the gel directly. Instead, it redistributes mobile ions that direct the migration of water molecules and bends the polyelectrolyte hydrogel. Subject to constraint from surrounding neutral gel, the elastic energy accumulates until suddenly released by snapping, just like the process in natural organisms. Several proof-of-concept examples, including an optical switch, a speedy catcher, and a pulse pump, are designed to demonstrate the versatile functionalities of this unit capable of articulate motion. This work should bring opportunities to devise soft robotics, biomedical devices, etc.",
"introduction": "INTRODUCTION Nature provides elegant paradigms and inspirations to devise versatile soft actuators and robots ( 1 – 6 ). Especially, some plants capable of rapid and articulate actions have attracted extensive interest ( 7 – 11 ). For example, the leaves of Venus flytrap undergo a fast snapping upon stimuli, owing to the snap-through instability rooted in the bistable structural geometry ( 8 , 9 ). Similarly, aquatic bladderworts take advantage of the fast release of elastic energy to capture small preys ( 10 , 11 ); when stimulated, the initially concave bladder wall snaps and rapidly recoils to a convex configuration, resulting in intake of water. This fast snapping is crucial for soft actuators/robots, whose natural rates of deformation/motion are usually low. By incorporating structures with snap-through instability, elastic energy can be efficiently stored, quickly released, and converted to kinetic energy ( 12 – 24 ). For example, Bende et al. ( 13 ) investigated the force-induced snap transition of a creased thin shell, in which the inherent geometry creates an energy barrier between isometric stable states. The unit capable of fast snapping has also been used to fabricate versatile devices, and a direct force is usually required to be in physical contact with the device and thereby trigger the snap transition or to reset the snapping structures ( 25 – 28 ). For example, a soft inflatable jumper has been fabricated, which is actuated by snapping of spherical elastomer caps when pressurized ( 28 ), similar to the rubber “popper toys” ( 27 ). The energy landscape for the isometric deformation is revealed by relating the applied force/pressure to the displacement of the snapping unit ( 13 , 28 ). An autonomous earthworm-like walker is developed with a soft valve equipped with bistable elastomeric dome that is switchable by differential pressure ( 29 , 30 ). Unlike natural organisms, however, these artificial systems are nonresponsive to external stimuli and thus often require external forces to flip between the mechanically stable states. Different from elastomers, hydrogels are ideal materials for soft actuators/robots due to their diverse forms of responses, similarity to soft biological tissues, and versatile applications under aqueous and bio-relevant conditions ( 31 – 34 ). Nevertheless, the solvent migration–enabled deformation is usually slow, while mechanical instability such as snapping may significantly expedite the actuation/motion. For example, a bilayer composite comprising thermo- and pH-responsive hydrogels, which is made into a dome constrained by a prestretched hydrogel ring, snaps to opposite directions by pH or temperature stimulation ( 31 ), but the complex fabrication and different conditions for reversion may hinder the applications. Alternatively, photolithography is a powerful strategy to fabricate hydrogels with in-plane gradients; the swelling mismatch between different regions leads to out-of-plane buckling with intrinsic bistability ( 35 – 37 ). If the bistable hydrogel can be actuated by contactless stimuli such as light or an electric/magnetic field ( 38 – 41 ), then spontaneous and fast snapping should be expected. Around 30 years ago, Shiga and Kurauchi ( 42 ) found that a polyelectrolyte hydrogel strip showed bending deformation under an electric field. This phenomenon has been theoretically analyzed by Doi and other scientists ( 43 – 46 ). Although the general physics of differential osmotic pressure has been revealed in terms of ion migration inside the gel under an electric field, most existing models have not considered the coupled dynamics of ion and water migration and gel swelling and deformation. It should be interesting to realize snap transition of polyelectrolyte hydrogel under an electric field, in which the electro-response of the gel couples with the mechanical instability. Another substantial yet challenging task is to formulate a numerical model to reveal the underlying mechanism, which is more complicated than the electro-bending of polyelectrolyte gels. Here, we present a bistable hydrogel that exhibits electric field–steered snapping transition. A composite hydrogel sheet is fabricated via photolithographic polymerization, in which a high-swelling polyelectrolyte gel disc is embedded in a nonswelling neutral gel. When immersed in a saline solution, due to swelling mismatch, the high-swelling region buckles into a domal configuration, which is naturally bistable. Under an appropriate electric field, the domal polyelectrolyte gel tends to bend in the opposite direction. Such an action, however, is prohibited by the surrounding constraint, and the gel is trapped in a local energy minimum. The elastic energy gradually accumulates and ultimately leads to a fast snap-through upon reaching a threshold. This process is reversible and controllable through the applied electric field, which redistributes the ions, causing water migration and bending. Because of the bistable nature of the structure, the flipped state can be maintained without the need of sustained energy input. Furthermore, a transient model with fully coupled electromigration of ions and water and the deformation of polyelectrolyte gel is successfully developed and validated by quantitative analysis of the electro-actuated snap transition. The actuation is imposed through a pair of electrodes without physical contact and triggered spontaneously from internal of the hydrogel, just like the natural organisms. Using the reversible snapping, an autosensing catcher and a self-sustained pulse pump are developed. This work heralds a new strategy to devise hydrogel-based soft actuators and robots with fast responses.",
"discussion": "DISCUSSION We have demonstrated the fast and reversible action of hydrogel actuators by using an electric field to direct ion and water migration in the polyelectrolyte gel and thereby trigger the snapping transition of the bistable configuration. A gel sheet of in-plane pattern, with polyelectrolyte gel surrounded by neutral gel, is fabricated through photolithographic polymerization. The swelling mismatch and geometrical confinement lead to out-of-plane buckling of the high-swelling polyelectrolyte region to form a domal configuration with mechanical bistability. Under the electric field, the polyelectrolyte gel intends to bend to the opposite direction yet is initially hindered by the geometric constrain. After the accumulated elastic energy is above a critical value, a fast snapping takes place. Systematic experiments are conducted to investigate the influences of gel composition, electric field, and salt concentration of the bath on the snapping transition. The snapping mechanism is further analyzed by numerical simulations, and a transient model with coupled electromigration of ions and deformation of polyelectrolyte gel is successfully developed and validated. This electric field–steered snapping has promising applications as optical switch, automatic catcher, and pulse pump, as demonstrated by experiments. Another noteworthy point is that the electro-actuated snap-through transition is also feasible when the domal gel is placed in phosphate-buffered saline (PBS; fig. S16), suggesting the potential biomedical applications of this snap-through-based actuator. Although a polyanion gel is used as an example in this work, polycation gels with opposite electro-responsiveness can be integrated to form complex systems (fig. S17 and movie S10). Sequential snap-through of multiple domes to certain directions at distinct times should afford sophisticated actuation of integrated devices in a spatiotemporal way. Since the electro-snapping and hysteresis between successive actuations are related to the migration of mobile ions and water, these processes can be sped up to some extent by reducing the thickness of the hydrogel (fig. S18), which, together with the feedback control loop, should merit the design of versatile soft actuators and robots. The overall efficiency for electro-actuated bending or snapping is relatively low, because most electrical energy has been wasted by molecule/ion migration in the solution and electrochemical reaction around the electrodes. To weaken or avoid the electrochemical reaction, the applied voltage needs be reduced to an appropriate level by miniaturizing the hydrogel system to shallow the energy barrier, which should be an interesting subject for further studies."
} | 2,380 |
31512709 | PMC8751625 | pmc | 740 | {
"abstract": "Here, we investigate the complete drying of hydrophobic cavities in order to elucidate the dependence of drying on the size, the geometry, and the degree of hydrophobicity of the confinement. Two complementary theoretical approaches are adopted: a macroscopic one based on classical capillarity and a microscopic classical density functional theory. This combination allows us to pinpoint unique drying mechanisms at the nanoscale and to clearly differentiate them from the mechanisms operational at the macroscale. Nanoscale hydrophobic cavities allow the thermodynamic destabilization of the confined liquid phase over an unexpectedly broad range of conditions, including pressures as large as 10 MPa and contact angles close to 90°. On the other hand, for cavities on the micron scale, such destabilization occurs only for much larger contact angles and close to liquid–vapor coexistence. These scale-dependent drying mechanisms are used to propose design criteria for hierarchical superhydrophobic surfaces capable of spontaneous self-recovery over a broad range of operating conditions. In particular, we detail the requirements under which it is possible to realize perpetual superhydrophobicity at positive pressures on surfaces with micron-sized textures by exploiting drying, facilitated by nanoscale coatings. Concerning the issue of superhydrophobicity, these findings indicate a promising direction both for surface fabrication and for the experimental characterization of perpetual surperhydrophobicity. From a more basic perspective, the present results have an echo on a wealth of biological problems in which hydrophobic confinement induces drying, such as in protein folding, molecular recognition, and hydrophobic gating.",
"conclusion": "5 Conclusions In this study we have elucidated the mechanism of spontaneous drying for liquids confined in hydrophobic cavities of varying size, geometry, and surface chemistry. Comparing the classical capillarity model with microscopic density functional theory has provided a route to quantify nanoscale effects, which facilitate drying in hydrophobic confinement ( eqn (4) ), by allowing drying in much broader ranges of contact angles and pressures. The general mechanism for spontaneous drying of hydrophobic nano-cavities is related to the one operative at the macroscale. However, nanoscale effects lead to substantial qualitative and quantitative modifications. One such effect is the shift of the drying contact angles ( i.e. , contact angles above which drying occurs) in nano-cavities to much lower values as compared to the macroscopic expectations. The macroscopic drying contact angles depend on the geometry but not on the size of the surface features; macroscopic values for right wedges or corners, θ wd = 135° and θ cd = 125.3°, are higher than the maximum contact angles typically achievable for water on flat surfaces. Nanoscale effects, however, shift the drying contact angles to much lower values. These shifts depend not only on the geometry but also on the size of the surface features. The shifts become more pronounced for smaller feature sizes. Moreover, the pressures up to which spontaneous drying occurs are increased very much due to nanoscale effects. These nanoscale effects, inter alia , originate from the long range of the fluid–wall interaction and from the oscillations of the liquid density induced by the repulsive part of the fluid–wall interaction together with the strong repulsion between fluid particles at short distances. The present calculations have also revealed that the geometry of the confinement has a crucial impact on drying: the presence of extended hydrophobic walls and acute corners further facilitates drying, e.g. , by favoring it for cylindrical pores as compared to grooves. In general, these findings suggest that, in order to obtain perpetual superhydrophobicity, it is convenient to realize nanoscale hydrophobic cavities, which are capable of remaining permanently dry over a broad range of pressures, with levels of hydrophobicity attainable (for water) by using standard wall coatings ( θ Y < 110°). On the other hand, macroscopic cavities with sizes larger than a micrometer are advantageous in order to enhance certain superhydrophobic properties, such as self-cleaning and drag reduction; however, self-recovery is impossible for surfaces with average contact angles ( i.e. , θ Y ≈ 110°). With this understanding, hierarchical surfaces, consisting of micron-sized textures covered by a rough nanoscale coating, emerge as a practical route to achieve perpetual superhydrophobicity on texture sizes of technological relevance. In order to be effective, this strategy requires that the nanoscale cavities are capable of self-recovery and result in an effective contact angle larger than the macroscopic drying angle. The coherent picture of drying across the scales provided by the present analysis not only offers useful design guidelines for perpetual superhydrophobicity, but also helps to understand the complex interplay of size, geometry, and degree of hydrophobicity which plays a crucial role in other technological (lyophobic porous materials) and biological (interactions of hydrophobic proteins) contexts. Finally, the present results are expected to encourage new experiments capable of realizing controlled nanoscale cavities and probing their drying behavior. This largely unexplored phenomenology is expected to be rich.",
"introduction": "1 Introduction Several interfacial phenomena of technological interest run under the name of superhydrophobicity: self-cleaning, 1 anti-fouling, 2 drag reduction, 3 etc. Such behaviors are tied to the occurrence of a “suspended”, so-called Cassie–Baxter, state in which a liquid body is in contact not only with a solid surface but also with vapor or gas pockets which occupy surface asperities, sustained by capillary forces. However, this superhydrophobic state is fragile, in the sense that external factors, such as pressure variations, contaminants, or mechanical vibrations, can cause its collapse into the Wenzel state in which all surface asperities are wet. If present, perpetual superhydrophobicity promises to heal the fragility of a suspended superhydrophobic state, thus opening novel scenarios for applications in materials science. It has been shown 4 that the Wenzel state can become thermodynamically unstable in nanoscale hydrophobic cavities, making the superhydrophobic state capable of self-recovery when the pressure is lowered to ambient conditions. However, direct experimental evidence of perpetual superhydrophobicity is still lacking for surfaces close to application interests, due to conceptual, experimental, and fabrication difficulties. Here, we clarify the concept of perpetual superhydrophobicity in its various connotations at the macro and nanoscale, with the goal of proposing new design criteria for superhydrophobic surfaces and stimulating accurate experimental studies of this drying phenomenon. Drying is an interfacial phase transition in which, upon changing pressure or temperature, the solid–liquid interface is replaced by a solid–vapor and vapor–liquid interface. 5 Understanding drying in hydrophobic confinement is crucial not only for superhydrophobicity but also for a number of important biological phenomena. For proteins (or large molecules) in solution the proximity of hydrophobic residues can give rise to local evaporation and, as a consequence, to hydrophobic attraction. 6,7 This mechanism is known to play a role in protein folding, 8,9 molecular recognition, 10 self-assembly, 11 and hydrophobic gating 12,13 – to name a few systems. The results discussed here, with a particular focus on superhydrophobic surfaces, are of relevance for this panoply of phenomena, in the sense that they contribute to disentangle the complex dependence of drying on the size, the geometry, and the degree of hydrophobicity of the confinement. In particular, we investigate four textures: two-dimensional grooves, cylindrical, parallelepipedic, and ink-bottle-shaped pits which are representative of different degrees of confinement and of relevant geometrical characteristics, such as corners, tapered walls, etc. In principle, currently available techniques can be used for producing nanotextured surfaces with similar geometries: quasi-2D grooves, cylindrical pits, and tapered cylinders. 14 Superhydrophobicity originates from capillary forces which are capable of sustaining a liquid–gas interface atop rough surfaces – the so-called Cassie–Baxter state. However, in a typical situation, pressure variations or other changes of the environmental conditions may overcome the capillary forces and induce complete wetting of the rough surface (Wenzel state), with the concomitant loss of superhydrophobic properties. The reverse transition from the Wenzel state to the Cassie–Baxter state is usually characterized by large free-energy barriers 15–17 which prevent the recovery of superhydrophobicity under ambient conditions without an external supply of free energy. Various strategies for active recovery have been proposed including the application of electric 18–22 or magnetic fields, 23 illumination by ultraviolet light, 24,25 or heating of the surface. 26,27 On the other hand, self-recovery aims at destabilizing the Wenzel state by purely passive means in order to completely and economically realize perpetual superhydrophobicity, at least over a range of technologically relevant pressures. There have been several attempts to realize “reversible” 28 or “monostable” 29 superhydrophobicity; for highly non-wetting liquids such as mercury, perpetual superhydrophobicity was reported for surfaces decorated with micron-scale textures. 29 For water, instead, due to its higher affinity for numerous substrates (lower hydrophobicity), this simple approach is not viable and the breakdown of superhydrophobicity proves irreversible even when the textures are as small as 20 nm. 30 Therefore, surfaces with hierarchical roughness, one on the micron and one on the nanometer scale, were proposed. 28,29 This promising strategy relies on the intuition that the smaller tier of roughness remains always dry due to nanoconfinement-enhanced hydrophobicity and promotes the drying of the larger tier. However, ensuring that nanoscale roughness is perpetually dry, and that the larger textures inherit this property, is highly nontrivial 31 and is experimentally arduous to verify. The present study aims at providing a coherent picture of drying from the nano- to the macroscale and formulating the conditions under which these mechanisms can be combined to realize self-recovering hierarchical surfaces, providing the hitherto missing theoretical tools to understand and design perpetual superhydrophobicity. In the following, we consider that there is no gas dissolved in water but only its vapor phase is formed. The degassed liquid represents the most unfavorable condition for the survival of superhydrophobicity. 32,33 To the best of our knowledge, spontaneous recovery of superhydrophobicity has not been demonstrated unequivocally in experiments with water. The main experimental difficulties are that (i) it is challenging to realize and to characterize regular surface textures on the nanometer scale, and (ii) the pressures required in order to press the liquid into such textures – in order to successively observe its spontaneous retraction – are exceptionally high. However, for porous materials high-pressure intrusion and extrusion experiments have shown that spontaneous drying can indeed be achieved in hydrophobic nanopores. 34,35 Although these systems, composed of nanoporous granules immersed in water, are quite distinct from superhydrophobic surfaces concerning shape and technological applicability, they positively demonstrate the physical mechanism of spontaneous drying at the nanoscale (based on which perpetual superhydrophobicity can be realized): after liquid intrusion into the pores at high pressures, it is possible to trigger their drying simply by decreasing the pressure below a certain threshold value, which we call the drying pressure P dry . Thus, for the smallest pores, around 2 nm wide or below, drying was observed even at pressures as high as tens of megapascals. 35–37 This phenomenon is related to the thermodynamic destabilization of the liquid phase due to nanoconfinement, which may occur even under “normal” conditions; this fact could also be exploited for designing materials endowed with perpetual superhydrophobicity. Importantly, the drying pressure is controlled by the wetting and the geometric characteristics of the pores, a topic which is rationalized in the present study and which is crucial for designing surfaces featuring perpetual superhydrophobicity. On the theoretical side, early microscopic simulations have shown 38–40 that free-energy barriers for drying decrease with increasing confinement; more recent simulation work has demonstrated barrierless drying in nanostructured hydrophobic surfaces with various geometries. 41,42 Microscopic (classical) density functional theory (DFT) calculations have also confirmed barrierless drying in nanoscale grooves 4 with moderately hydrophobic surfaces ( i.e. , contact angles θ Y slightly above 90°). Finally, macroscopic theories predict that it is possible to realize drying at pressures which are much higher than the liquid–vapor coexistence pressure P coex ( T ) (at a given temperature T ) even for large-scale textures, provided that the surfaces are sufficiently hydrophobic, with the contact angle θ Y being typically much larger than in the nanoscale case. 32,42 Viewed together, these studies underscore an important conclusion: drying is very sensitive to geometry, hydrophobicity, and roughness size, which are the main design parameters for perpetual superhydrophobicity. Another important aspect, which has not yet been fully appreciated, is that the drying mechanism at the macroscopic scale, which is well-known in the special case of a wedge, 43–46 apparently differs from and seems to be unrelated to the nanoscale one which favors drying in a much broader range of thermodynamic conditions and material parameters. 4 In view of the paramount technological interest, the scatter of theoretical results, and the difficulty in realizing and interpreting experiments, it is urgent to clarify the physical mechanisms of drying as well as to uncover their dependence on material parameters with particular attention to size-dependence. Here, we address this issue by devising a generalized macroscopic theory of drying in confinement (Section 2), by studying via microscopic DFT drying in a variety of geometries (Section 3), and by discussing drying on surfaces with multiple roughness scales (Section 4). The coherent picture of drying, which emerges from this multiscale analysis, provides crucial guidelines for designing perpetual superhydrophobicity and for clearly interpreting corresponding experiments on (perpetual) superhydrophobicity."
} | 3,777 |
38206115 | PMC10832564 | pmc | 742 | {
"abstract": "Abstract Microbial synthetic consortia are a promising alternative to classical monoculture for biotechnological applications and fermentative processes. Their versatile use offers advantages in the degradation of complex substrates, the allocation of the metabolic burden between individual partners, or the division of labour in energy utilisation, substrate supply or product formation. Here, stable synthetic consortia between the two industrially relevant production hosts, Pseudomonas putida KT2440 and Corynebacterium glutamicum ATCC13032, were established for the first time. By applying arginine auxotrophy/overproduction and/or formamidase‐based utilisation of the rare nitrogen source formamide, different types of interaction were realised, such as commensal relationships (+/0 and 0/+) and mutualistic cross‐feeding (+/+). These consortia did not only show stable growth but could also be used for fermentative production of the γ‐glutamylated amines theanine and γ‐glutamyl‐isopropylamide (GIPA). The consortia produced up to 2.8 g L −1 of GIPA and up to 2.6 g L −1 of theanine, a taste‐enhancing constituent of green tea leaves. Thus, the advantageous approach of using synthetic microbial consortia for fermentative production of value‐added compounds was successfully demonstrated.",
"conclusion": "CONCLUSION This study reports on the first binary microbial consortia between the industrially relevant microorganisms P. putida and C. glutamicum . Based on the complementation of arginine‐auxotrophy and on formamidase‐based utilisation of formamide as a rare xenobiotic source of nitrogen, consortia with no interaction (0/0), commensal relationships (+/0 and 0/+) and mutualistic cross‐feeding (+/+) were designed and tested, and a stable composition was achieved. As proof‐of‐concept, these synthetic consortia were applied to produce two γ‐glutamylated amines; the food and beverage additive theanine and the related GIPA. Future work on optimising the substrate and precursor supply, the replacement of refined glucose for alternative feedstocks and a deepened understanding of co‐culture dynamics may enable more sustainable, even further improved application of these consortia on larger scales.",
"introduction": "INTRODUCTION The conversion of complex substrates in nature relies on the cooperation of mixed‐species microbial consortia with often unknown functions and compositions (Jiang et al., 2017 ), for example in fermented foods, waste treatment and agriculture (McCarty & Ledesma‐Amaro, 2019 ). While stable communities may form for bioconversion of nutrition and energy (Thuan et al., 2022 ), the composition is often not known and the consortia can only be used for black‐box applications (Sgobba & Wendisch, 2020 ). There are bioinformatic approaches to reveal the compositions of natural consortia for use in synthetic consortia with the desired function (Kumar et al., 2021 ). Consortia with non‐engineered strains can be used for the production of natural products, for example due to the induction of silent gene clusters (Thuan et al., 2022 ), but these consortia often convert the substrates not only into the desired product but also by‐products are produced, which lowers the overall yield of the process (Ben Said & Or, 2017 ). To circumvent this disadvantage, synthetic microbial consortia with defined compositions can be designed. In these consortia, engineered interdependencies are typically used to reduce heterogeneities (Sgobba & Wendisch, 2020 ; Zhang et al., 2018 ). For the design of these synthetic consortia, the interactions, mechanisms and structures of natural inter‐ and intrakingdom consortia need to be studied and understood (Zhang et al., 2018 ). Synthetic microbial consortia have several advantages over monocultures, for example due to the division of labour; reduction of metabolic loads; avoidance of the influence of different functions; optimal catalytic environments for enzymes derived from different sources; balance of cofactor and energy levels; higher adaptability and stability to environmental fluctuations (Harcombe et al., 2018 ; Pan et al., 2023 ; Shong et al., 2012 ). Most biotechnological processes are based on feedstocks, which are homogenous and continuously available, and thus, specialised monocultures can be used. For feedstocks with varying supply and composition, adjusted microbial consortia can be applied that match the respective process steps of substrate conversion and/or product formation (Liu et al., 2019 ). Microbial consortia can often use less defined substrates than monocultures, which lowers cost since purification is not required (Ben Said & Or, 2017 ; Zhang et al., 2018 ). One type of application uses a synthetic consortium that consists of a substrate converter and a producer strain (Sgobba & Wendisch, 2020 ; Thuan et al., 2022 ). For example, in a synthetic Escherichia coli and Corynebacterium glutamicum consortium, starch‐based production of lysine by substrate‐converting E. coli and lysine production by starch‐negative C. glutamicum was established (Sgobba et al., 2018 ). A consortium of Pseudomonas putida and Synechococcus elongatus was designed in which the cyanobacterium fixes CO 2 and secretes carbohydrates, which P. putida in turn can use for growth and the degradation of environmental pollutants. Also, a synthetic consortium for butanol production by Thermoanaerobacterium thermosaccharolyticum M5 with Clostridium acetobutylicum NJ4 by use of lignocellulosic biomass as substrate was established (Jiang et al., 2020 ). It has to be noted that division of labour in consortia requires the transport of compounds over the cellular membrane, for example the uptake or export of substrates, intermediates and products via transport proteins (Lu et al., 2019 ; Sgobba & Wendisch, 2020 ). During the design of a synthetic consortium, different technical hurdles need to be solved, like improving the production parameters or developing process parameters for storage and maintaining the stability of the composition (Sgobba & Wendisch, 2020 ). For complex biosynthesis pathways, consortia can be developed to partition individual biosynthetic steps among two or more microbial strains and/or species. This lowers the metabolic burden of the single microorganism (Brenner et al., 2008 ; Lu et al., 2019 ; Thuan et al., 2022 ; Tsoi et al., 2018 ) and can be used to circumvent the accumulation of inhibitory intermediates (Ben Said & Or, 2017 ). This division of labour can lead to higher production as genes or modules are shared (Bittihn et al., 2018 ). The sum of implemented genetic modifications, which is limited to a single strain, is higher, and the desired features of different strains or species can be combined in a coordinated manner (Cui et al., 2019 ; Wang et al., 2018 ). Also, complicated metabolic and genetic manipulation can be reduced by the use of specialised members in the consortia (Re & Mazzoli, 2023 ). A number of design rules exist for this subdivision, comprising availability of transport proteins, inhibition due to substrates, intermediates or products and distinct biochemical requirements of subpathways regarding NADPH, ATP or oxygen (Sgobba & Wendisch, 2020 ). Consortia provide a kind of compartmentalisation such that undesired cross‐reactions and side products can be reduced through active or passive transport of substrates and intermediates (Shong et al., 2012 ). By dividing metabolic pathways, incompatibilities between gene elements and host cells can be bypassed, for example in the E. coli and Saccharomyces cerevisiae co‐culture used to produce oxygenated taxanes (Zhou et al., 2015 ). Production of bioactive compounds by consortia may support plug‐and‐play biosynthesis: strains used in co‐cultures are genetically designed to adapt to each biosynthesis pathway module more precisely than monocultures to the entire pathway, and the consortium is optimised in host selection, construction of module pathways, initial ratio of strains and culture conditions (Thuan et al., 2022 ). For best production, the selection of the intermediate metabolic node is necessary and important for balancing cofactor and energy supply (Li et al., 2019 ; Pan et al., 2023 ). To establish robust processes of synthetic consortia, different levels of interdependency through cross‐feeding or auxotrophies can be used: no interaction (0/0), competition (−/−), commensalism (0/+) or mutual cooperation (+/+) (Sgobba & Wendisch, 2020 ). Polycultures consisting of more than two host organisms can be designed to express genes of a biosynthetic route of a complex compound and maintain the metabolic flux balance, precursor and energy supply (Thuan et al., 2022 ). In this study, synthetic microbial consortia of P. putida and C. glutamicum strains were constructed for the production of γ‐glutamylated amines such as theanine. For the design of the consortia, different prerequisites, such as growth conditions, growth medium, co‐existence and dependencies, should be noticed. Therefore, two different types of dependent consortia were established by using an arginine‐dependent consortium or an ammonium‐dependent consortium as a commensal consortium (+/0 or 0/+) and an arginine‐ and ammonium‐dependent consortium as mutually cooperative consortium (+/+) for the growth and production of the γ‐glutamylated amines theanine and γ‐glutamyl‐isopropylamide (GIPA). Theanine is the most abundant free amino acid in green tea leaves and can be used in the food and pharmaceutical industries as taste enhancer or for relaxation, respectively (Mu et al., 2015 ). GIPA is naturally involved in the l ‐alaninol pathway of Pseudomonas sp. strain KIE171 (de Azevedo Wäsch et al., 2002 ; Pan et al., 2020 ). For the production of both amines, monocultures of P. putida Thea1 have been described (Benninghaus et al., 2021 , 2023 ) and the strain was developed further for use in consortia with C. glutamicum . The enzyme γ‐glutamyl‐methylamide synthetase (EC 6.3.4.12, GMAS), encoded by gmaS , from Methylorubrum extorquens , was used and monoethylamine (MEA) or isopropylamine (IPA) was added as amine donor for production of theanine or GIPA, respectively. The enzyme GMAS is part of the N ‐methylglutamate pathway for monomethylamine assimilation in methylotrophs and catalyses the ATP‐dependent methylamidation of glutamate to γ‐glutamyl‐methylamide (Gruffaz et al., 2014 ; Nayak & Marx, 2014 ). By addition of MEA or IPA as amine donor, the enzyme can catalyse the conversion of glutamate and MEA or IPA to theanine or GIPA, respectively. The enzyme was used for fermentative production of theanine in C. glutamicum , E. coli and P. putida (Benninghaus et al., 2021 ; Hagihara et al., 2021 ; Ma et al., 2020 ) and also tested with lyophilised E. coli cells for the substrate spectrum using several different amines, including MEA and IPA (Pan et al., 2020 ).",
"discussion": "RESULTS AND DISCUSSION Design of synthetic microbial consortia with P. putida and C. glutamicum \n Culture media have been developed for growth of bacteria in monocultures, but media for consortia may need different compositions, also depending on their type of interaction (Figure 1A ). In this study, interactions between P. putida and C. glutamicum ( P. putida/C. glutamicum) are indicated with ‘0’ for neutral (the strain does not benefit from the co‐culture) and ‘+’ for positive interaction (the strain benefits from the co‐culture). In a consortium without any interaction (interaction type 0/0), both strains, for example the wild‐type (WT) strains P. putida KT2440 and C. glutamicum ATCC13032, can grow independently from each other (Figure 1B ). In a consortium with commensalism (interaction type +/0 or 0/+), the former or the latter strain benefits from the second strain. In this study, an arginine‐auxotrophic strain P. putida KT2440 Δ argH lacking the argininosuccinate lyase gene was constructed in order to establish a dependency on the arginine‐producing strain C. glutamicum ARG2 (Figure 1C ). The latter strain is neither benefiting nor suffering from the former strain, thus being neutral with respect to the presence of the second strain. In reverse direction, an ammonium dependency (0/+) was used, in which the P. putida strain was constructed for plasmid‐based expression of formamidase gene amiF from Helicobacter pylori . This strain can grow with formamide as sole nitrogen source, and it liberates ammonium to the medium. The C. glutamicum wild‐type strain cannot utilise formamide and is therefore dependent on ammonium release by P. putida FORM (Figure 1D ). To establish a cooperative or mutually dependent consortium (interaction type +/+), the P. putida strain KT2440 Δ argH was constructed for plasmid‐based expression of amiF . The formamidase‐negative strain C. glutamicum ARG2 cannot utilise formamide, thus, is dependent on ammonium released by P. putida strain FORM. Therefore, in mineral salts media with formamide as sole nitrogen source, the arginine‐auxotrophic strain P. putida Δ argH FORM requires arginine from the arginine‐overproducing C. glutamicum ARG2 that in turn requires ammonium from P. putida strain FORM (Figure 1E ). FIGURE 1 Design of synthetic microbial consortia with P. putida and C. glutamicum . (A) Monoculture of P. putida KT2440 (WT) and C. glutamicum ATCC13032 (WT) with the fluorescence proteins Crimson and Gfp UV indicated with the colours red and green, respectively. (B) Co‐culture with P. putida WT and C. glutamicum WT without any interaction (0/0) growing in the same medium. (C) Co‐culture of P. putida KT2440 Δ argH as arginine‐auxotrophic strain with the arginine‐overproducer C. glutamicum ARG2 in a commensal consortium (+/0) via arginine dependency of P. putida . (D) Co‐culture of P. putida FORM as formamide‐utilising strain with formamidase‐negative C. glutamicum WT in a commensal consortium (0/+) via ammonium dependency of C. glutamicum . (E) Co‐culture of the arginine‐auxotrophic, formamide‐utilising strain P. putida KT2440 Δ argH FORM with the arginine‐overproducer strain C. glutamicum ARG2, which cannot utilise formamide, in a cooperative consortium (+/+) via arginine and ammonium interdependency. Here, binary microbial consortia of the industrially relevant bacteria P. putida and C. glutamicum were developed for the first time. Four interaction types (0/0, +/0, 0/+ and +/+) were designed and tested for production of the food and beverage additive theanine and the related GIPA. In microbial consortia, intra‐ and interspecies interactions exist, such as mutualism or competition for nutrients, which affect metabolism and product yields (Jiang et al., 2017 ). Natural consortia are rarely binary but mostly consist of many species. The complexity of designing ternary synthetic consortia (or of even higher valency) exceeds that of binary consortia considerably. In principle, a ternary consortium may be based on the here established mutualistic consortium with arginine‐auxotrophic, formamidase‐positive P. putida and arginine‐overproducing, formamidase‐negative C. glutamicum with the addition of an α‐amylase‐positive E. coli . This will require a dependency of E. coli on one of the other partners, for example by means of lysine auxotrophy that could be supplemented by the low lysine concentrations (1 mM) excreted from a C. glutamicum strain (Sgobba et al., 2018 ). Ternary consortia have already been designed, constructed and used, for example for the production of rosmarinic acid with three E. coli strains designed for mutually exclusive access to the carbon sources glucose and xylose and subdivision of the biosynthetic pathway into three modules (Li et al., 2019 ). Future work will guide the construction of higher order interspecies consortia, but we will focus on a binary consortium for the production of theanine and GIPA. Medium for co‐culturing P. putida and C. glutamicum without interaction (0/0) To enable the growth of a synthetic consortium, media conditions are needed in which each partner can grow. Therefore, the minimal media M12 (described for P. putida ) and CgXII (described for C. glutamicum ) were prepared without the addition of the respective trace element solutions. For growth, both trace element solutions and 20 g L −1 glucose as carbon source for the growth of wild‐type strains P. putida KT2440 and C. glutamicum ATCC13032 were used, respectively. As is regular for CgXII, biotin and PKS were also added to the M12 minimal medium. The recorded growth data are listed in Table 3 . C. glutamicum ATCC13032 reached a higher biomass in CgXII minimal medium independent of the used trace elements, while the growth rates of C. glutamicum ATCC13032 and P. putida KT2440 were comparable. In M12 minimal medium with PKS and biotin, both strains reached a similar growth rate and biomass concentration independently of the used trace elements. Since the choice of either trace element solution did not have significant influence, adapted trace elements have been developed. For this purpose, both compositions were compared (Table S2 ). For the same component, the lower concentration was used, only one component per element was used and possible carbon sources were avoided. For the microbial consortia, a medium called M12‐C was found and was based on M12 minimal medium with 20 g L −1 glucose as a carbon source, ammonium sulphate as the sole nitrogen source (if not mentioned otherwise), PKS and biotin and adapted trace elements. All further experiments were performed with this medium, as it enabled the growth of both partners. TABLE 3 Growth data of P. putida KT2440 and C. glutamicum ATCC13032 in CgXII or M12 minimal media with both trace elements. CgXII trace elements M12 trace elements Minimal medium μ [h −1 ] Biomass [g L −1 ] Y X/S [g g −1 ] μ [h −1 ] Biomass [g L −1 ] Y X/S [g g −1 ] \n C. glutamicum ATCC13032 0.32 ± 0.01 13.3 ± 0.2 0.67 0.32 ± 0.01 6.7 ± 0.3 0.34 CgXII—trace elements + 20 g L −1 glucose \n P. putida KT2440 0.29 ± 0.06 4.4 ± 0.1 0.22 0.30 ± 0.07 3.3 ± 0.6 0.17 \n C. glutamicum ATCC13032 0.40 ± 0.01 3.8 ± 0.3 0.19 0.40 ± 0.02 4.0 ± 0.2 0.20 M12—trace elements + biotin + PKS + 20 g L −1 glucose \n P. putida KT2440 0.42 ± 0.01 4.0 ± 0.7 0.20 0.46 ± 0.06 3.9 ± 0.6 0.20 Interaction type 0/0 co‐culture With the developed medium, growth in co‐cultures without interaction (type 0/0) was observed. Preliminary cell counting by flow cytometry showed that the cell numbers of P. putida and C. glutamicum strains correlate with respective proportions of shared total biomass in co‐cultivation. Differentiation of strains was realised by fluorescence proteins through equipment of P. putida and C. glutamicum with plasmids encoding fluorescence reporter proteins Crimson and Gfp UV , respectively. Cells were analysed by FACS measurement for the determination of strain ratios (Figure 2A ). In monoculture, both wild‐type strains grew to similar biomass, and only one fluorescence protein was detected. In the co‐culture of P. putida WT and C. glutamicum WT, the number of cells detected for Crimson fluorescence was higher than that of those detected for Gfp UV , indicating a higher ratio of P. putida in the co‐culture without any interaction. In conclusion, a co‐culture without any interactions (0/0) between wild‐type strains of P. putida and C. glutamicum was established as both strains grew; however, the initial 50:50 ratio between the strains was not maintained. FIGURE 2 Growth of monocultures and inter‐species consortia of P. putida and C. glutamicum strains without any interaction (0/0) and with obligate arginine cross‐feeding (+/0) between arginine‐overproducing C. glutamicum ARG2 and arginine‐auxotrophic P. putida Δ argH with the respective wild‐type strain of each organism as negative controls. Strains of P. putida and C. glutamicum were differentiated by the fluorescence of Crimson and Gfp UV , respectively. Mono‐ and co‐cultivations were inoculated to the indicated ratio and cultivated in M12‐C for 24 h. Culture compositions were determined by flow cytometry (A) according to scatter plots (B) and fluorescence microscopy overlays (C) as exemplarily shown for the co‐cultivation of P. putida Δ argH and C. glutamicum ARG2, inoculated to equal (50:50) proportions. The growth of both bacteria was confirmed without dependencies in an adapted medium, which is an important step for a potential consortium for the production of value‐added compounds (Zhang et al., 2018 ). Interestingly, consortia of type (0/0) showed a tendency to shift towards P. putida , and the biomass formation and growth rate were about the same as P. putida showed in the monoculture. Here, either unknown promotion of P. putida by C. glutamicum or inhibition of C. glutamicum by P. putida through exported metabolites or proteins could have favoured propagation of P. putida (Zhang et al., 2018 ). In‐depth metabolome or proteome analysis may help to learn more about the exchange of metabolites or proteins potentially affecting the P. putida–C. glutamicum consortia. A commensal consortium (interaction type +/0) with dependency of arginine‐auxotrophic P. putida Δ argH on arginine‐producing C. glutamicum \n ARG2 \n Since the synthetic consortium of interaction type 0/0 was not balanced ( P. putida dominated over C. glutamicum , s. above), it was investigated if arginine can serve to impose unilateral obligate dependency of arginine‐auxotrophic P. putida Δ argH on arginine‐overproducing C. glutamicum ARG2 in a commensal co‐culture (+/0). Therefore, both strains were grown in M12‐C medium as monocultures or co‐cultures with different cell ratios at inoculum (Figure 2A ). As expected, the monoculture of the arginine‐auxotrophic strain P. putida Δ argH did not grow, while the monoculture of C. glutamicum ARG2 grew. Due to arginine overproduction, strain ARG2 grew to a lower biomass concentration compared to C. glutamicum WT. In the control co‐culture of arginine prototrophic P. putida KT2440 with arginine‐overproducing C. glutamicum ARG2, only Crimson fluorescence was detected after 24 h, indicating that the P. putida KT2440 strain took over the cultivation completely. In the control co‐culture of the arginine‐requiring P. putida Δ argH with C. glutamicum ATCC13032, which does not overproduce arginine, the C. glutamicum strain dominated the co‐culture. Notably, in the commensal co‐culture of the arginine‐requiring P. putida Δ argH with the arginine‐producing C. glutamicum ARG2, both strains co‐existed. When inoculated with ratios of 70/30, 50/50, 30/70 and 10/90 of P. putida Δ argH and C. glutamicum ARG2, the consortia contained 69%, 62%, 68% and 81%, of P. putida Δ argH cells, respectively, after growth for 24 h. As shown exemplarily, the flow cytometry determinations (Figure 2B ) were corroborated by fluorescence microscopy (Figure 2C ). In conclusion, a commensal consortium (+/0) between the arginine‐requiring P. putida Δ argH and the arginine‐producing C. glutamicum ARG2 via arginine dependency was established. Also, the commensal arginine‐dependent (+/0) consortia showed a tendency to shift towards P. putida . This was also discussed in the consortium without dependency. Even in the commensal arginine‐dependent (+/0) consortia, in which P. putida required arginine from C. glutamicum for growth, P. putida was dominant. This may be explained by the fact that already low concentrations of arginine are sufficient as a supplement for the growth of the arginine‐auxotrophic P. putida strain in arginine‐dependent (+/0) consortia; thus, few C. glutamicum cells provide enough arginine. In a consortium of lysine‐auxotrophic E. coli with lysine‐overproducing C. glutamicum , low amounts of C. glutamicum were needed, indicating enough production for cross‐feeding (Sgobba et al., 2018 ). A mutualistic consortium (interaction type +/+) with formamidase and arginine interdependencies For mutual interdependency by cross‐feeding, a second constraint was introduced in addition to arginine dependency. Formamidase‐dependent formamide hydrolysis to enable the use of formamide as the sole source of nitrogen was established by plasmid‐based expression of amiF in P. putida KT2440 Δ argH , yielding strain P. putida Δ argH FORM. Based on our previous work on formamidase‐positive C. glutamicum strains, which released sufficient ammonium to allow growth of co‐cultivated formamidase‐deficient C. glutamicum strains (Schwardmann, Wu, et al., 2023 ), it was tested if strain P. putida Δ argH FORM can grow in media with formamide as the sole nitrogen source and if it releases enough ammonium to support growth of co‐cultivated formamidase‐negative C. glutamicum strain ARG2. Since the crimson gene was replaced by the amiF gene in strain P. putida KT2440 Δ argH FORM, Gfp UV fluorescence was used to identify C. glutamicum cells and all non‐fluorescent cells were assigned to P. putida . The strains were grown in nitrogen‐free M12‐C medium with 68 mM formamide added as the sole nitrogen source (Figure 3 ). In monoculture, P. putida Δ argH FORM did not grow with formamide as a nitrogen source. C. glutamicum ARG2 could not utilise formamide for growth. Co‐cultures of formamidase‐positive P. putida Δ argH FORM with formamidase‐negative C. glutamicum strain ARG2 did grow under these conditions, and a ratio of 59:41 was observed after 24 h of cultivation. Thus, co‐cultivation of P. putida and C. glutamicum with obligate interdependent mutual cross‐feeding (+/+) is possible, and nutritional constraints can be exploited to steer culture dynamics. FIGURE 3 Biomass formation of arginine‐auxotrophic, formamidase‐positive P. putida strain Δ argH FORM and arginine‐overproducing, formamidase‐deficient C. glutamicum strain ARG2, grown as monocultures or as mutualistic inter‐species consortia (+/+). Strains of C. glutamicum and P. putida were differentiated by the Gfp UV fluorescence of C. glutamicum , while P. putida cells were non‐fluorescent. Mono‐ and co‐cultivations were inoculated to the indicated ratio and cultivated in a modified nitrogen‐free version of M12‐C with 68 mM formamide as a nitrogen source for 24 h. Culture compositions were determined by flow cytometry. The introduction of one dependency each (0/+ or +/0) was used as preparation for a consortium with mutualistic interaction. Mutual interdependencies of both species (+/+) are engineered to ensure the survival of both partners, which can also be used for the production of certain compounds (Zhang et al., 2018 ). Consortia of the (+/+)‐type were based on the presence/absence of formamidase for utilisation of the rare nitrogen source formamide on the one hand and on arginine production/auxotrophy on the other hand, and their composition was shown to be stable. A more balanced ratio between the strains could only be reached within 24 h in the mutualistic (+/+) consortium, in which P. putida releases ammonium as a nitrogen source for C. glutamicum , which in turn secretes arginine to supplement growth by the arginine‐auxotrophic P. putida . In a consortium with formamidase‐positive and formamidase‐negative strains of C. glutamicum , only lower ratios of formamidase‐positive strains were necessary to support growth of formamidase‐negative C. glutamicum strain (Schwardmann, Rieks, et al., 2023 ). Theanine production by synthetic consortia As an application of the designed synthetic microbial consortium, the production of theanine was chosen as example. For production, the concentration of glucose as carbon source and ammonium sulphate as nitrogen source in the medium was adjusted. As glutamate‐derived theanine, and also arginine are nitrogen‐containing substances, the nitrogen concentration in the M12‐C medium is likely to be increased. Our previously developed theanine producing strain of P. putida was used (Benninghaus et al., 2021 ) and 50 mM MEA were added as ethylamine donor for alkylation of endogenously synthesised glutamate by GMAS to theanine. Different concentrations of glucose (10, 20, 40 g L −1 ) were tested. Higher glucose concentrations did not show growth benefits, but production of theanine was increased with 20 g L −1 glucose. Since no further increase with 40 g L −1 was determined (Figure S1A,B ), 20 g L −1 glucose was used for following production experiments. Also, different concentrations of ammonium sulphate (17, 34, 51 mM) were tested. No significant differences were observed in growth and only slight differences were found in the production of theanine and glutamate by consortia. In the monocultures of C. glutamicum ATCC13032 and C. glutamicum ARG2 production of glutamate and arginine was detected only when higher concentrations of the nitrogen source were provided (Figure S2A,B ). Therefore, in the production experiments 34 mM ammonium sulphate as sole nitrogen source was used in M12‐C medium. For theanine production, the producer strain P. putida Thea1 (Benninghaus et al., 2021 ) was used. In a consortium without any dependencies (0/0), the growth rate and biomass formation were higher compared the P. putida Thea1 monoculture, but lower compared the C. glutamicum ATCC13032 (Table 4 ). The co‐culture Thea1+ATCC13032 produced 1.5‐fold more theanine (2.40 ± 0.08 g L −1 ) compared to the P. putida Thea1 monoculture (1.54 ± 0.07 g L −1 ) (Figures 4A and S3A ). The glutamate concentration in the C. glutamicum ATCC13032 monoculture (0.45 ± 0.01 g L −1 ) was higher compared to the co‐culture Thea1+ATCC13032 (0.11 ± 0.01 l ‐L −1 ). TABLE 4 Growth and production data for theanine production in consortia without interaction (0/0), with arginine‐ (+/0) or ammonium‐dependency (0/+) or in a cooperative consortium with arginine‐ and ammonium‐dependency (+/+). Biomass [g L −1 ] μ [h −1 ] Glutamate [g L −1 ] Arginine [g L −1 ] Theanine [g L −1 ] Y P/X [g g −1 ] Y P/S [g g −1 ] No interaction (0/0) \n P. putida Thea1 1.0 ± 0.1 0.09 ± 0.01 0.04 ± 0.01 n.d. 1.54 ± 0.07 1.6 0.17 \n C. glutamicum ATCC13032 7.9 ± 0.3 0.43 ± 0.01 0.45 ± 0.01 n.d. n.d. \n P. putida Thea1 + C. glutamicum ATCC13032 4.6 ± 0.1 0.24 ± 0.02 0.11 ± 0.01 n.d. 2.40 ± 0.08 0.5 0.13 Arginine dependency (+/0) \n P. putida Thea1 Δ argH \n 0.2 ± 0.1 0.01 ± 0.01 0.09 ± 0.01 n.d. 1.16 ± 0.04 6 0.75 \n C. glutamicum ARG2 4.7 ± 0.3 0.31 ± 0.01 0.13 ± 0.06 1.27 ± 0.03 n.d. \n P. putida Thea1 Δ argH + C. glutamicum ARG2 3.4 ± 0.1 0.21 ± 0.01 0.14 ± 0.04 n.d. 2.57 ± 0.14 0.8 0.17 Ammonium dependency (0/+) \n P. putida Thea1 FORM 2.1 ± 0.1 0.12 ± 0.01 n.d. n.d. 4.03 ± 0.08 1.9 0.20 \n C. glutamicum ATCC13032 0.3 ± 0.1 0.07 ± 0.01 n.d. n.d. n.d. \n P. putida Thea1 FORM + C. glutamicum ATCC13032 6.5 ± 0.5 0.34 ± 0.01 n.d. n.d. 0.11 ± 0.01 0.02 0.01 Mutual dependency (+/+) \n P. putida Thea1 Δ argH FORM 0.1 ± 0.1 0.00 ± 0.00 n.d. n.d. 0.10 ± 0.02 1 0.03 \n C. glutamicum ARG2 0.2 ± 0.1 0.04 ± 0.01 n.d. n.d. n.d. \n P. putida Thea1 Δ argH FORM + C. glutamicum ARG2 3.3 ± 0.4 0.20 ± 0.01 n.d. n.d. 1.26 ± 0.51 0.4 0.07 FIGURE 4 Production of theanine by various P. putida and C. glutamicum strains grown as monocultures or as various consortia. Concentrations of theanine (green), glutamate (blue) and arginine (yellow) in mono‐ and co‐cultures of P. putida Thea1 and C. glutamicum ATCC13032 without interdependency (0/0) (A), P. putida Thea1 Δ argH and C. glutamicum ARG2 with arginine‐dependency (+/0) (B), P. putida Thea1 FORM and C. glutamicum ATCC13032 with ammonium‐dependency (0/+) (C) and P. putida Thea1 Δ argH FORM and C. glutamicum ARG2 with arginine‐ and ammonium‐dependency (+/+) (D). An arginine‐dependent co‐culture was established by using P. putida Thea1 Δ argH and C. glutamicum ARG2 for theanine production. The monoculture of P. putida Thea1 Δ argH did not grow, while the monoculture of C. glutamicum ARG2 grew to higher biomass formation and a faster growth rate compared to the arginine‐dependent co‐culture (Table 4 ). The production of theanine in the co‐culture was 2.2‐fold higher (2.57 ± 0.14 g L −1 ) than the P. putida monoculture (1.16 ± 0.04 g L −1 ) (Figures 4B and S3B ). Glutamate concentrations were comparable in all three cultivations (around 0.1 g L −1 ) and arginine was only detected in the monoculture of C. glutamicum ARG2 (1.27 ± 0.03 g L −1 ). After production of theanine was shown to be higher in both (0/0 and +/0) co‐cultivations compared to the respective monocultures, a ammonium‐dependent consortium (0/+) using formamidase‐negative strain C. glutamicum ATCC13032 with formamide‐utilising strain P. putida Thea1 FORM (Figures 4C and S3C , Table 4 ) and a cooperative consortium (+/+) with the arginine‐producing, formamidase‐negative strain C. glutamicum ARG2 and the arginine‐auxotrophic, but formamide‐utilising strain P. putida Thea1 Δ argH FORM were used (Figures 4D and S3D , Table 4 ). For all production samples, carbohydrates, especially glucose, gluconate and ketogluconate, were measured and the yield Y P/S was calculated based on the utilised glucose. In all measurements, gluconate and ketogluconate concentrations were below 0.02 g l \n −1 . Glucose was measured using RID, while gluconate and ketogluconate were measured using DAD, thus a differentiation of the components was possible (Figure S4 ). To initiate growth, very low concentrations (1/100) of ammonium sulphate (0.34 mM for the formamidase‐negative C. glutamicum ARG2) and arginine (0.04 mM for the arginine‐auxotrophic P. putida Thea1 Δ argH FORM) were added to the medium. Formamide was used as a nitrogen source at a concentration of 68 mM and 20 g L −1 glucose were used as a carbon source. In monoculture, P. putida Thea1 FORM produced 4.03 ± 0.08 g L −1 theanine with formamide as nitrogen source. The consortium with C. glutamicum ATCC13032 produced much less theanine (0.11 ± 0.01 g L −1 ) (Figure 4C ). The monoculture of P. putida Thea1 Δ argH FORM produced 0.10 ± 0.02 g L −1 theanine, while the consortium with C. glutamicum ARG2 produced much more theanine (1.26 ± 0.51 g L −1 ) (Figure 4D ). As a conclusion, a cooperative consortium with arginine and nitrogen dependency was established for the production of theanine. As shown here for the synthetic consortia producing theanine, the addition of arginine and ammonium helped to kickstart the growth of the mutualistic consortia (compare Table 4 ). Thus, arginine as a supplement and ammonium as a more accessible nitrogen source are critical compounds that can be used to fine‐tune the composition of the synthetic consortia. Bacteria can utilise different nitrogen sources, ranging from simple inorganic compounds like nitrate to complex nitrogen‐containing compounds like amino acids or nucleosides (Merrick & Edwards, 1995 ). The most preferred nitrogen source is ammonia, with supporting high growth rates. The use of other nitrogen sources is strictly regulated by the synthesis of the necessary transporters (Merrick & Edwards, 1995 ). Formamide is a reduced C1 nitrogen compound, such as methylamine, ammonium carbamate and carbamoyl phosphate. The latter two decompose to yield ammonia, while the first two need enzyme‐catalysed reactions for nitrogen liberation (Pols et al., 2021 ; Schwardmann, Wu, et al., 2023 ). Formamide is a rare nitrogen source in biotechnology (Schwardmann, Wu, et al., 2023 ) and was used in the here developed consortia with ammonium dependency (+/0 and +/+). Cross‐feeding of ammonium was shown between formamidase‐positive and formamide‐negative C. glutamicum strains (Schwardmann, Rieks, et al., 2023 ) and here for interspecies consortia with P. putida and C. glutamicum . Production of glutamate with formamide as a nitrogen source was also established by engineering C. glutamicum for expression of amiF from H. pylori . Comparing the product titre, there was a 2.5‐fold increase in glutamate production with formamide instead of ammonium sulphate and urea (Schwardmann, Rieks, et al., 2023 ). This is in line with our observations for the highest production of theanine with strain Thea1 FORM (Figure 4C ) with formamide as a nitrogen source. Since formamide is also a carbon source accessible to some methylotrophs, formamide may be relevant to fine‐tune and impose nitrogen and/or carbon dependencies in synthetic interspecies microbial consortia that grow with one‐carbon C sources such as methanol or formate (Schwardmann, Benninghaus, et al., 2023 ). GIPA production by synthetic consortia Production of GIPA is similar to theanine production as GMAS is able to use isopropylamide (IPA) for alkylation of glutamate to yield GIPA. Therefore, the co‐culture design of theanine production could easily be transferred to GIPA production and the same strains were used. Instead of MEA, 50 mM IPA was added as alkylamine. For the co‐culture without dependencies (0/0; P. putida Thea1 and C. glutamicum ATCC13032), the growth during the production of GIPA was similar to the growth during theanine production (Table S3 ). The production in monoculture of GIPA was 0.82 ± 0.05 g L −1 , while the production in co‐culture was reduced to about half with 0.46 ± 0.03 g L −1 (Figure S5A ). In the co‐culture, glutamate was not detected as a by‐product, while both monocultures accumulated glutamate (Thea1: 0.04 ± 0.01 g L −1 , ATCC13032: 0.50 ± 0.01 g L −1 ). In the arginine‐dependent co‐culture (+/0), the growth was also comparable with the theanine‐producing co‐culture (Table S3 ). The production of GIPA in mono‐ and co‐culture was similar (Thea1 Δ argH : 0.57 ± 0.001 g L −1 ; Thea1 Δ argH +ARG2: 0.59 ± 0.05 g L −1 ) (Figure S5B ). In the co‐culture, neither glutamate nor arginine were detected, while 1.31 ± 0.01 g L −1 arginine accumulated in the C. glutamicum ARG2 monoculture. In conclusion, GIPA production in co‐cultures was realised and the production in the arginine‐dependent co‐culture was comparable with monoculture production. During the production of GIPA in both, mono‐ and co‐culture, a decreased pH value of 4 to 5 was measured in all tested interaction types. To prevent this decrease, GIPA production was also tested with 100 mM potassium phosphate buffer in minimal medium (Table 5 ), as there is no buffer component in the used M12‐C medium. The production of GIPA was increased in monoculture of P. putida Thea1 (2.22 ± 0.23 g L −1 ) and co‐culture of P. putida Thea1 with the C. glutamicum ATCC13032 (1.47 ± 0.13 g L −1 ) (Figures 5A and S6A ). No glutamate was measured in C. glutamicum ATCC13032 monoculture, unlike the production without buffer. In the arginine‐dependent co‐culture, 2.33 ± 0.24 g L −1 GIPA was produced, while the P. putida Thea1 Δ argH monoculture produced 2.33 ± 0.49 g L −1 GIPA (Figure 5B and S6B ). In all cultivations, neither glutamate nor arginine could be detected. In conclusion, a balanced producing system for commensal (+/0) consortia with arginine dependency was found. Lastly, an ammonium‐dependent consortium (0/+) using the formamidase‐negative strain C. glutamicum ATCC13032 with the formamide‐utilising strain P. putida Thea1 FORM (Figures 5C and S6C , Table 5 ) and a cooperative consortium (+/+) for GIPA production with the arginine‐producer strain C. glutamicum ARG2 and the arginine‐auxotrophic formamide utiliser strain of P. putida Thea1 Δ argH FORM (Figures 5D and S6D , Table 5 ) were tested. As a nitrogen source, 68 mM formamide was added. To help kickstart growth, 1% ammonium sulphate (0.34 mM) and 0.04 mM arginine were added to the medium. Also, potassium phosphate buffer was added. P. putida Thea1 FORM produced 4.52 ± 0.23 g L −1 GIPA in monoculture. The consortium with C. glutamicum ATCC13032 produced less GIPA (0.25 ± 0.02 g L −1 ) (Figure 5C ). In the cooperative consortium, 2.80 ± 0.44 g L −1 GIPA was produced, while the monoculture of P. putida Thea1 Δ argH FORM produced 0.29 ± 0.03 g L −1 GIPA (Figure 5D ). TABLE 5 Growth and production data for gipa production in consortium without interaction (0/0), with arginine‐ (+/0) or ammonium dependency (0/+) or in cooperative consortium with arginine‐ and ammonium‐dependency (+/+) with 100 mM potassium phosphate buffer. Biomass [g L −1 ] μ [h −1 ] Glutamate [g L −1 ] Arginine [g L −1 ] \n gipa [g L −1 ] Y P/X [g g −1 ] Y P/S [g g −1 ] No interaction (0/0) \n P. putida Thea1 1.7 ± 0.1 0.15 ± 0.00 0.12 ± 0.01 n.d. 2.22 ± 0.23 1.3 0.11 \n C. glutamicum ATCC13032 6.1 ± 0.2 0.35 ± 0.01 n d. n.d. n.d. \n P. putida Thea1 + C. glutamicum ATCC13032 2.6 ± 0.2 0.20 ± 0.00 n.d. n.d. 1.47 ± 0.13 0.6 0.08 Arginine dependency (+/0) \n P. putida Thea1 Δ argH \n 0.1 ± 0.0 0.01 ± 0.00 n.d. n.d. 2.33 ± 0.49 23.0 0.39 \n C. glutamicum ARG2 3.4 ± 0.1 0.24 ± 0.01 n.d. n.d. n.d. \n P. putida Thea1 Δ argH + C. glutamicum ARG2 2.5 ± 0.1 0.17 ± 0.00 0.10 ± 0.02 n.d. 2.33 ± 0.24 0.9 0.12 Ammonium dependency (0/+) \n P. putida Thea1 FORM 2.0 ± 0.1 0.08 ± 0.01 n.d. n.d. 4.52 ± 0.23 2.3 0.23 \n C. glutamicum ATCC13032 0.2 ± 0.1 0.09 ± 0.01 n.d. n.d. n.d. \n P. putida Thea1 FORM + C. glutamicum ATCC13032 6.3 ± 0.3 0.34 ± 0.01 n.d. n.d. 0.25 ± 0.02 0.03 0.01 Mutual dependency (+/+) \n P. putida Thea1 Δ argH FORM 0.1 ± 0.1 0.00 ± 0.00 n.d. n.d. 0.29 ± 0.03 3 0.11 \n C. glutamicum ARG2 0.1 ± 0.1 0.06 ± 0.01 n.d. n.d. n.d. \n P. putida Thea1 Δ argH FORM + C. glutamicum ARG2 2.9 ± 0.4 0.16 ± 0.02 n.d. n.d. 2.80 ± 0.44 1.0 0.14 FIGURE 5 Production of GIPA by various P. putida and C. glutamicum strains grown as monocultures or as various consortia. Concentration of gipa (purple), glutamate (blue) and arginine (yellow) in mono‐ and co‐cultures of P. putida Thea1 and C. glutamicum ATCC13032 without interdependency (0/0) (A), P. putida Thea1 Δ argH and C. glutamicum ARG2 with arginine‐dependency (+/0) (B), P. putida Thea1 FORM and C. glutamicum ATCC13032 with ammonium‐dependency (0/+) (C) and P. putida Thea1 Δ argH FORM and C. glutamicum ARG2 with arginine‐ and ammonium‐dependency (+/+) (D) with 100 mM potassium phosphate buffer. For all production samples, carbohydrates, especially glucose, gluconate and ketogluconate, were measured and the yield Y P/S was calculated based on the utilised glucose. In all measurements, gluconate and ketogluconate concentrations were below 0.02 g l \n −1 . The production of GIPA in the different consortia follows the same trend as the production of theanine. The application of different dependencies and the usage of formamide as alternative nitrogen source is already discussed for theanine production. The use of buffer showed a production benefit, and the pH value was neutral at the end of production. The control of pH value is crucial for the structure and function of proteins and for the production of ATP since it relies on the electrochemical potential of the cell. Therefore, it is important to maintain a neutral pH in the cell. By consumption of glucose the pH dropped in E. coli and P. putida cultures (Sánchez‐Clemente et al., 2018 ). This can be prevented by addition of buffer components in the medium. In the original composition of CgXII medium, MOPS buffer is used for maintaining pH value at pH 7 (Eggeling & Bott, 2005 ). In M12 minimal medium, there is no dedicated buffer component (Mückschel et al., 2012 ). In this study, we showed that growth and production benefit from a stable, neutral pH maintained with potassium phosphate buffer. Further optimisation strategies Cooperation between the partners of a mutualistic consortium for the production of a valuable product such as theanine may be fine‐tuned genetically. Instead of overexpressing the genes required for the biosynthesis of the valuable compound, their expression may be controlled by the physiological state of the cell in order to maximise production. This concept is often used for production in monocultures by expression of the biosynthesis genes from inducible promoters. In so‐called autoinduction media, a carbon source mixture of glucose and an inducing sugar (e.g. lactose for promoter P lac or arabinose for promoter P BAD ) is used. Since E. coli preferentially grows with glucose, induction occurs later during growth upon glucose depletion (Studier, 2014 ). In the consortia described here, ammonium and formamide were used as nitrogen sources. Recently, nitrogen deprivation‐inducible promoters were used for dynamic growth‐decoupled gene expression or CRISPRi‐mediated gene knockdown in C. glutamicum for the production of the growth‐inhibitory sugar alcohol xylitol from xylose and were shown to be superior to standard overexpression (Schwardmann, Wu, et al., 2023 ). The use of arginine auxotrophy and overproduction in the consortia described here may also be developed further by using arginine‐dependent gene expression, for example based on the promoter P arg and the transcriptional repressor ArgR from C. glutamicum (Yim et al., 2011 ). Fine‐tuning of consortium composition needs to be performed depending on the desired product to optimise the production and/or utilising of substrates. An equal distribution of partners is not desirable in every production consortium (Zhang et al., 2018 ). This is especially the case when one partner is needed to provide secreted enzymes, for example amylase, or small amounts of a nutrient, for example, lysine or arginine (Sgobba et al., 2018 ). Different ratios of each partner lead to different substrate utilisation and transport, secretion and uptake of intermediates (Bittihn et al., 2018 ; Zhang et al., 2018 ), like arginine and ammonium used in the here designed consortia. Providing an additional carbon source that is only accessible to one partner of the consortium may also be used for fine‐tuning (Pan et al., 2023 ). Since alternative carbon sources are generally interesting in order to prevent competition with human and animal nutrition or to decrease the cost of the process, a number of bacteria, including P. putida and C. glutamicum , have been engineered for the utilisation of non‐native carbon sources (Wendisch et al., 2016 ). Also, complex and less refined substrates can be used (Ben Said & Or, 2017 ; Thuan et al., 2022 ). As examples of consortia utilising carbon sources other than glucose, a consortium of E. coli and Pichia pastoris produced the valuable alkaloid stylopine from glycerol (Urui et al., 2021 ) and a synthetic consortium of E. coli and C. glutamicum was used to produce lysine from sucrose and starch (Sgobba et al., 2018 ). While the provision of an additional carbon source may help to skew the composition of the consortium, this may negatively or positively affect product formation by the consortium as well. As shown for the P. putida monoculture production process (Benninghaus et al., 2021 ), the combined use of xylose and glycerol improved theanine production as compared to glucose‐based production. Sarcosine production by C. glutamicum was better with xylose than with glucose (Mindt et al., 2019 ). Besides being valuable to fine‐tune the composition of mutualistic consortia, synthetic microbial consortia may be key for better utilisation of the individual compounds present in complex substrate mixtures by individual partners of a consortium. These can be used, for example by C. glutamicum in rice straw hydrolysate (Sasikumar et al., 2021 ) or corn stover hydrolysate (Wen & Bao, 2019 ). Instead of producing a monoculture of a strain able to co‐utilise glucose, arabinose and xylose, three different strains that could use only one of these carbon sources were inoculated in various ratios and produced riboflavin (Pérez‐García et al., 2021 ). However, this process was dynamic, with concentrations of substrates and population sizes changing over time since the three strains were not mutually dependent. Thus, this concept may be further developed by using mutualistic instead of commensal consortia."
} | 12,085 |
25642220 | PMC4294203 | pmc | 743 | {
"abstract": "Little is known about the importance and/or mechanisms of biological mineral oxidation in sediments, partially due to the difficulties associated with culturing mineral-oxidizing microbes. We demonstrate that electrochemical enrichment is a feasible approach for isolation of microbes capable of gaining electrons from insoluble minerals. To this end we constructed sediment microcosms and incubated electrodes at various controlled redox potentials. Negative current production was observed in incubations and increased as redox potential decreased (tested −50 to −400 mV vs. Ag/AgCl). Electrode-associated biomass responded to the addition of nitrate and ferric iron as terminal electron acceptors in secondary sediment-free enrichments. Elemental sulfur, elemental iron and amorphous iron sulfide enrichments derived from electrode biomass demonstrated products indicative of sulfur or iron oxidation. The microbes isolated from these enrichments belong to the genera Halomonas, Idiomarina, Marinobacter , and Pseudomonas of the Gammaproteobacteria , and Thalassospira and Thioclava from the Alphaproteobacteria . Chronoamperometry data demonstrates sustained electrode oxidation from these isolates in the absence of alternate electron sources. Cyclic voltammetry demonstrated the variability in dominant electron transfer modes or interactions with electrodes (i.e., biofilm, planktonic or mediator facilitated) and the wide range of midpoint potentials observed for each microbe (from 8 to −295 mV vs. Ag/AgCl). The diversity of extracellular electron transfer mechanisms observed in one sediment and one redox condition, illustrates the potential importance and abundance of these interactions. This approach has promise for increasing our understanding the extent and diversity of microbe mineral interactions, as well as increasing the repository of microbes available for electrochemical applications.",
"introduction": "Introduction Marine sediments are complex environments that can be difficult to characterize physically and biologically as they are highly heterogeneous. They can support a range of temperatures, pHs, Ehs, pressures, and concentrations of organic and inorganic metabolites over a variety of scales generating a wide range of metabolic niches (Jakobsen, 2007 ; Stockdale et al., 2009 ). This environmental variability results in an impressive diversity of microorganisms and microbial metabolisms, including metabolisms specific to a given resource niche, or requiring a specific redox interface (Nealson, 1997 ; Tankéré et al., 2002 ; Jourabchi et al., 2009 ). Though we can assess phylogenetic diversity, our understanding of metabolic diversity and activity is often limited in these systems to the dominant and/or well-studied processes, or often, to those few organisms that can be cultivated. However, observations of previously undetected metabolisms, such as those hidden by cycling of redox species (between element reducing and element oxidizing microbes) are amassing in many marine systems (Canfield et al., 2010 ; Gault et al., 2011 ; Holmkvist et al., 2011 ; Stewart et al., 2012 ). This coupled with the fact that many lithotrophs utilize insoluble substrates that are difficult to monitor has led to these processes being largely overlooked in environmental settings. The importance of these lithotrophic and often autotrophic reactions in marine systems and how they have been grossly under estimated is slowly coming to light (Swan et al., 2011 ). The main goal of this work is to utilize electrochemical methods to culture microbes capable of mineral oxidation from marine sediment where lithotrophic reactions are potentially occurring. The unknowns surrounding lithrophic mineral oxidizers in environmental systems stem from poor representation of these microbes in culture. Characterization of their physiology is limited to a few pathways and a few systems, giving us a limited set of genetic biomarkers for these processes (Ghosh and Dam, 2009 ; Emerson et al., 2010 ; Ilbert and Bonnefoy, 2013 ). This, in turn, has limited our understanding of the extent of microbes (how many families, genera, organizational taxonomic units [OTUs]) performing lithotrophic reactions. Approaches to culturing microbes that can better mimic the resource niche and/or redox gradient a microbe is adapted to will help improve our understanding of these microbial metabolisms and give us a means to detect their presence in these environments. In some cases, electrode cultivation may also offer a means of applying molecular, culturing or analytical methods that are inhibited by mineral byproducts, precipitates or Fenton reactions (Rabaey et al., 2007 ). Electrode microbe interactions have played an important role in understanding extracellular electron transport in mineral reducing microbes. Mechanisms of extracellular electron transport (EET), the transport of electrons to substrates external to the cell, have only been well characterized in members of the genera Shewanella and Geobacter . These include direct transfer via multiheme cytochromes (Myers and Myers, 1992 ), transfer along conductive nanowires (Reguera et al., 2005 ; Gorby et al., 2006 ), and transfer through soluble mediators or shuttles (Lovley et al., 1996 ; Newman and Kolter, 2000 ; Marsili et al., 2008 ; Von Canstein et al., 2008 ). To date, several of these cultures have been shown to be able to reduce electrodes as a surrogate for minerals, and depending on the potential of the electrode, can serve as a source or a sink for electrons (Gregory et al., 2004 ; Rosenbaum et al., 2011 ). It is known and has been demonstrated in many environmental systems (Rabaey and Verstraete, 2005 ; Lovley, 2006 ), and in particular marine environments (Holmes et al., 2004 ; Ryckelynck et al., 2005 ; Reimers et al., 2006 ; White et al., 2009 ), that microbes can utilize electrodes as terminal electron acceptor (TEA). What is less studied and is recently gaining attention is the potential for iron oxidizing microbes to obtain electrons from electrodes as well—using the electrode as a primary electron donor (Rosenbaum et al., 2011 ; Summers et al., 2013 ). Two recent discoveries in iron oxidizing bacteria are beginning to highlight this phenomenon. In the neutraphilic iron oxidizer Sideryoxidans lithotrophicus ES-1 and the phototrophic iron oxidizer Rhodopseudomonas palustris TIE-1, homologs to the Mtr genes in Shewanella oneidensis MR-1 have been shown to oxidize iron in vitro (Jiao and Newman, 2007 ; Liu et al., 2012 ). In S. lithotrophicus , there is also evidence that this gene can complement the iron reducing capacity in MR-1, suggesting a similar location, interaction with iron and potentially a similar ability to utilize an electrode (though this has not been tested) (Liu et al., 2012 ). Direct evidence of an iron oxidizer using an electrode as an electron source has been demonstrated in the marine iron oxidizer Mariprofundus ferroxidans (Summers et al., 2013 ) and the phototrophic iron oxidizer Rhodopseudomonas palustris TIE-1 (Bose et al., 2014 ). Notably, M. ferroxidans does not contain an MtrAB homolog or any other known outer membrane cytochrome protein known to be involved in EET (Ilbert and Bonnefoy, 2013 ). Both of these observations illustrate the potential for lithotrophic microbes to be cultivated and characterized electrochemically. Additionally, the second observation supports the high potential for novel mechanisms of extracellular electron transport in mineral oxidizing microbes. In this work, we assess electrodes poised at electron-donating (cathodic) redox potentials as a means of enriching lithotrophic mineral oxidizing microbes from an environmental system. We successfully isolated several novel electrochemically active bacteria from marine sediments that were further electrochemically characterized giving insight into both the extent and diversity of mechanisms behind metabolisms that oxidize insoluble substrates.",
"discussion": "Discussion Catalina Harbor sediment, like most sediment, contains a gradient of redox conditions. Work from Bertics et al., analyzed sediment cores taken from Catalina Harbor and showed that invertebrate burrowing activities generated a diversity of redox conditions by distributing oxygen (resulting in the production of oxidized ferric iron) and potentially nitrate and sulfate to lower sediment depths (Bertics and Ziebis, 2009 ). This activity creates a wide suite of redox niches with gradients of multiple electron donor and electron acceptor (oxygen, nitrate, ferric iron, etc.) couples and is likely responsible for the extensive microbial diversity and heterogeneity of OTU distributions observed (Bertics and Ziebis, 2009 ). The lower sediment depths were dominated by ferrous iron compared to the upper sediments, ranging from 400 to 1600 μmol per cm 3 sediment (Bertics and Ziebis, 2009 ). The high iron content of these sediments allows for the possible presence of marine neutraphilic iron-oxidizing bacteria and also results in scavenging of sulfide generated from sulfate reduction. Notably, neither previous work (Bertics and Ziebis, 2009 ), nor our work demonstrated detectable hydrogen sulfide. However, over time, the production of black precipitates in sediment microcosms was observed suggesting active sulfate reduction (observations by A. Rowe and B. Lam). This work also suggests the potential importance/abundance of elemental sulfur produced from tetrathionate and sulfide in this system in addition to reduced FeS. Though elemental sulfur is rarely quantified in marine sediments, it has been shown to be the most abundant sulfur intermediate present in both Black Sea and North Sea sediments (1.0–0.1 μmole/cm 3 ) (Zopfi et al., 2004 ). This could have important implications on sediment nutrient cycling. In order to enrich for microbes capable of oxidizing the insoluble inorganic substrates present in Catalina Harbor, we used an electrode to impose a constant redox state and mimic a solid substrate or mineral surface, predicting that mineral-oxidizing microbes would be capable of taking electrons from an electrode. We tested a range of electron donating redox potentials to demonstrate the utility of this approach for enriching a notoriously difficult to culture group of microbes. Cathodic currents were sustained consistently at redox potentials lower than −300 mV vs. Ag/AgCl, which is significant in that these potentials are energetically capable of donating electrons to many oxidized redox couples (oxygen, nitrate, etc.) but do not cause significant hydrogen formation (empirical observations Figure S2 , and Gregory et al., 2004 ). The ability of microbes to utilize cathodes has been demonstrated previously (Clauwaert et al., 2007a , b ; Jia et al., 2008 ; Butler et al., 2010 ; Cournet et al., 2010 ; Desloover et al., 2011 ; Hsu et al., 2012 ; Su et al., 2012a ). This capability has been identified in predominantly mineral reducing microbes (Rabaey et al., 2007 ; Rosenbaum et al., 2011 ). While other works have shown the feasibility of using electrodes at electron accepting redox potentials for enriching metal or sulfur reducing bacteria from marine sediments (Holmes et al., 2004 ; Reimers et al., 2006 ; White et al., 2009 ), only a few studies have looked at microbial communities on cathodes and predominantly from applied systems (wastewater, fuel cells etc.) (Rabaey et al., 2008 ; Faimali et al., 2010 ; Wrighton et al., 2010 ; Strycharz-Glaven et al., 2013 ). Isolation of microbes from cathode enrichments capable of oxygen reduction using electrons obtained from an electrode yielded members of the Betaproteobacteria, Gammaproteobacteria , and Bacteriodetes (Rabaey et al., 2008 ) or Actinobacteria and Flavobacteria (Erable et al., 2010 ). In neither group of isolates was the ability to use an insoluble substrate shown as these strains were grown heterotrophically and/or using hydrogen as the primary electron donor (Rabaey et al., 2008 ; Erable et al., 2010 ). In sediment-free secondary enrichments electrode oxidation was coupled to the reduction of the anaerobic TEA amendments (nitrate and ferric iron). However, we observed no evidence of coupled sulfate reduction and electrode oxidation. This could be due to the unfavorable difference in redox potential between sulfate (E 0 = −590 mV vs. SHE at pH 8.0, see Supplementary Information) and oxidation at the poised working electrode (−203 mV vs. SHE). Though it is difficult to predict the exact energetic gain under the proposed conditions, it seems likely that the energy gained from this reaction would be insufficient to effectively compete with other electrode oxidizing processes. This will be tested in future work by testing lower redox potentials. Alternatively, electrode oxidizing sulfate reducers may be more sensitive to oxygen then the other microbes tested in this work and so were inhibited in sediment-free reactors due to counter electrode processes (potential for oxidation of water to oxygen). Tertiary enrichments focused on nitrate as a TEA couple for mineral oxidizing metabolisms. Multiple solid substrate donors, including elemental iron, elemental sulfur and reduced (amorphous) FeS, were shown to enrich for a variety of different phylogenies. A distinct shift in microbial groups being enriched was observed for the different iron substrates (elemental iron vs. reduced FeS). Interestingly microbial communities enriched on elemental iron, shared a similar phylogenetic composition to the elemental sulfur enrichments; these enrichments were dominated by Gammaproteobacteria . The similarity between the elemental iron and sulfur enrichments could be due to charge similarity (0) between these compounds compared with ferrous iron (+2) enrichments which were dominated by members of the Firmicutes . Nonetheless many of the same genera were isolated from these different enrichments. Interestingly, members of the Marinobacter were only isolated from enrichments containing Fe 2+ or FeS. Though FeS is an amorphous and insoluble substrate it is known to release small amounts of Fe 2+ a substrate potentially amenable to different oxidation processes and thus targeting different microbial groups. Alternatively these differential enrichment patterns could also be a result of a stronger selection for organisms that utilize insoluble rather than soluble substrates. The selection of only certain groups with these particular electron donor/electron acceptor couples suggest that there are even more physiologies enriched by the electrodes than accounted for in isolation methods. Future work will address looking at a greater range of solid substrate enrichments and/or electrochemical isolation techniques to target other electrochemically active populations. Given the preliminary enrichment of microbes based on electrochemical treatment, it was not surprising that all of the microbes isolated from solid substrate enrichments demonstrated electrochemical activity. Though variation was noted across strains in terms of current densities observed, this variation could be a function of the ecology of the organisms. In essence, the rate of electron uptake per cell is likely related to the max reaction rates (V max ) for the overall reaction which could vary across strains. Notably the current densities observed for the strains isolated fall within the range of current densities observed in other organism isolated from cathodes. Specifically, Erable et al., isolated strains with current densities of 0.02–1μA/cm 2 (Erable et al., 2010 ) and Rabaey et al., observed maximum current densities of 27–69 μA/cm 2 for various isolates (Rabaey et al., 2008 ). In this system current densities were increased with the length of time the organisms were incubated in the electrochemical cells (Figure S6 ). This observation suggests the ability of at least some of the strains isolated to grow autotrophically using an electrode as the sole electron donor. Additionally that lack of carbon substrates available could limit the current production in strains that were unable to grow using only the electrode. Further studies are required to better address the optimization of electrochemical activity in these organisms for applied purposes. To our knowledge this is the first instance of an electrochemically active sulfur oxidizing microbe, though several sulfate reducing microbes have been shown to be capable of cathode oxidation. Likewise, members of the Pseudomonas genera have previously been shown to be electrochemically active and some Pseudomonas strains are capable of sulfur oxidation. The electrochemical activity of sulfur oxidizers may be linked to the ability of microbes to use extracellular elemental sulfur. However, the majority of biochemical work on sulfur oxidizers demonstrates the use of elemental sulfur intracellularly (Friedrich et al., 2005 )—most proteins characterized in sulfur oxidation are cytoplasmic with the exception of the phototrophic sulfur oxidizer Allochromatium vinosum . However, it is likely that the organisms isolated on elemental sulfur do not use the traditional pathways of sulfur oxidation and/or all of the intermediates. This is further supported by the observation that many sulfur oxidizing Gammaproteobacteria often do not contain genes from the canonical sulfur oxidizing pathway (Sox), supporting a potentially novel biochemistry in addition to potentially novel EET mechanisms. The majority of isolates obtained from elemental sulfur enrichments are likely to oxidize thiosulfate to tetrathionate given the pH increase observed on thiosulfate plates (thiosulfate being the sole electron donor) in contrast to acid production with oxidation to sulfate. Notable this metabolism has been observed previously in member of the Halomonas and Pseudomonas (Sorokin, 2003 ). Metabolically this product results in one less electron obtained for sulfur compounds compared with oxidation to sulfate. Additionally, tetrathionate is a “dead end” product in that it cannot be further microbially metabolized (Sorokin, 2003 ). However, tetrathionate in the presence of sulfide can abiotically be re-reduced, regenerating thiosulfate and generating elemental sulfur from sulfide. As both of these are substrates readily oxidized by tetrathionate producers this particular sulfur oxidation pathway could have ecological advantages to traditional sulfur oxidation. It has been postulated that this pathway could allow for energy storage in the form of elemental sulfur (Sorokin, 2003 ). This may be especially important in unstable or low nutrient environments. Alternatively, elemental sulfur may be the preferred substrate utilized by these sulfur oxidizers. Differences in electrochemical activity across isolated strains, even from the same genera, were observed in this work. This extends not only to the portion of electrons removed from the electrode but also to the predicted mode of electrode interaction (i.e., biofilm vs. planktonic) and the dominant redox potential of the proteins interacting with the electrode (i.e., midpoint potential). In the Halomonas isolates, we observed organisms that generated thickly populated and sparsely populated biofilms (Figure S5 ). We also observed the formation of suspended aggregated cells that only loosely attached to the electrode and may not have survived SEM processing (e.g., Marinobacter sp . Figure S5C ). This may speak to an intermediate mode of attachment; a mode that is more transient or more porous than a traditional biofilm. We observed no conclusive evidence of the presence of soluble electrochemically active compounds (potential mediators) in the spent media (where cells were removed) though our preliminary CV evidence in some Pseudomonas strains display patterns that suggest mediator binding. However, more sensitive methods may be required to detect such interactions. The observations that the microbes isolated in this work contain multiple metabolisms (i.e., organisms that can oxidize both iron and sulfur), and perform the same metabolism using different physiological mechanisms (predicted from the variation in midpoint potentials of dominant redox active proteins), suggest that the single redox condition applied can enrich a diversity of microbes. Interestingly, the diversity of midpoint potentials suggests that variable energetic recoveries and efficiencies may be colonizing electrodes under a given redox condition. There is the potential for a tradeoff between the amount of energy recovered and the efficiency of a given redox interaction between a soluble substrate and a microbial EET pathway. Which microbes do better under the instituted redox conditions, and or how altering the redox conditions alters the community will be the focus of future work. Given the data obtained from this work, electrochemical enrichment of microbes from environmental systems is a promising approach to: (1) expand the range of different microbes in culture collections, (2) better understand the importance of redox conditions and microbe mineral interactions in the microbial ecology of a system, and (3) expand the microbes available for electrochemical applications (e.g., biocathodes in microbial fuel cells). Many of these isolates are commonly observed in marine sediments in 16S rRNA gene surveys and in the subsurface (Ivanova et al., 2000 ; Inagaki et al., 2003 ; Kaye, 2004 ; Kaye et al., 2011 ; Smith et al., 2011 ; Dong et al., 2013 ; Kato et al., 2013 ) suggesting that that these oxidative metabolisms may be widespread though not generally identified. Identification of genes involved in these processes or extending these enrichment techniques to other environments could help inform the extent of these processes in nature. It should be noted that electromicrobiology, though a young field, has been demonstrating its utility in application including alternative energy generation (microbial fuel cells), removal of water contaminants such as heavy metal, chlorinated compounds, or nitrate from water (bioremediation), energy storage and the generation of small organic molecules (electrosynthesis) (He and Angenent, 2006 ; Rabaey et al., 2007 ). Electrochemical cultivation methods will add to the repository of microbes capable of electrode interactions, and may help maximize the efficiency or productivity of these applications. Conflict of interest statement The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest."
} | 5,687 |
37729205 | PMC10523532 | pmc | 744 | {
"abstract": "Significance That wild-type methanogens are capable of methanotrophic growth contributes to understanding their role in methane cycling that impacts global warming. The remarkable reversibility of catabolic reactions in one organism foreshadows undiscovered metabolic potentials of diverse organisms. Indeed, methanotrophic growth by reversal of methanogenic pathways validates pathways proposed for uncultured anaerobic methanotrophic archaea (ANME) and presents the possibility of methanogenesis by ANME not yet reported. Finally, the results establish M. acetivorans C2A and M. orientis sp. nov. as platforms for further biochemical understanding of anaerobic methanotrophy.",
"conclusion": "Conclusions The results presented definitively show that canonical methanogens are capable of growth as methanotrophs, which supports previous assertions that growth of ANME proceeds by reversal of methanogenic pathways. The results also establish M. acetivorans C2A and M. orientis sp. nov. as unprecedented platforms for universal biochemical understanding of AOM. The work documents methanotrophic growth of wild-type methanogens with potential for previously unrecognized contributions to AOM impacting the methane cycling and redox cycling of iron or humic acids. Finally, the remarkable similarity of proposed pathways to those proposed for ANME predicts that ANME are capable of reversing AOM for methanogenic growth.",
"discussion": "Discussion The present findings end a decades-long absence of any axenic wild-type anaerobic methanotroph beginning with the discovery of AOM ( 23 , 24 ). The discovery of ferrihydrite-dependent methanotrophic growth of methanogens advances the biochemistry and ecology of AOM. Although Fe(III) is less abundant than sulfate in marine environments, iron oxide-dependent AOM plays a significant role in which methanogens are potential actors ( 8 – 13 , 25 ). It is also possible that methanotrophic growth of marine methanogens is symbiotic with electron transfer to sulfate-reducing species by unknown mechanisms. Finally, the biochemical understanding of AOM is advanced by validating the long-held proposal that methanotrophy is supported by the reversal of methanogenic pathways. Methanotrophic growth of M. acetivorans C2A and M. orientis sp. nov. was dependent on the reduction of ferrihydrite or humic acids showing a respiratory mode of energy conservation. Figs. 5 A and B show proposed energy-conserving pathways. All carbon-carrying reactions are the reversal of core reactions in methanogenic pathways. Except reactions producing formate, the enzymes are well characterized and known reversible ( 12 , 26 – 28 ). Notably, endothermic transfer of the methyl group from CH 3 -SCoM to H 4 SPt by the Mtr of M. acetivorans C2A ( Fig. 5 A ) is driven with the sodium gradient generated from electron transfer to Fe(III) by the Rnf complex ( 12 ). Methanotrophic growth of wild-type M. acetivorans C2A shows that wild-type methyl-coenzyme M methylreductase (Mcr) is reversible in contrast to the previously reported requirement to engineer a strain that contains the gene encoding Mcr from ANME-1, although activity of the recombinant ANME-1 Mcr was not reported ( 15 ). Reversibility of the wild-type M. acetivorans C2A Mcr is supported by heterologous expression of ANME Mcr in Methanococcus maripaludis which showed that the recombinant ANME-2 Mcr assembled with the host Mcr in hybrid complexes while ANME-1 Mcr only formed homogenous complexes which suggests that Mcr from ANME-2 and methanogens are structurally similar and that reversibility is regulated by thermodynamics as opposed to structural differences ( 29 ). Reversibility of the Mcr from methanogens is further supported by the AOM for Mcr purified from Methanothermobacter marburgensis ( 28 ). Fig. 5. Energy-conserving reactions proposed for ferrihydrite-dependent methanotrophic growth of M. acetivorans ( A ) and M. orientis ( B ). Atp, adenosine triphosphate synthase; Rnf, homolog of the Rhodobacter nitrogen fixation complex; Fdx, ferredoxin; Cyt c , multiheme cytochrome c ; MP, methanophenazine; HdrDE, membrane-bound heterodisulfide reductase; Mtr, CH 3 -H 4 SPT:HSCoM methyltransferase; Mrp, multisubunit sodium/proton antiporter; Fpo, membrane-bound F 420 dehydrogenase; H 4 SPT, tetrahydrosarcinapterin; HSCoA, coenzyme A; HSCoM, coenzyme M; HSCoB, coenzyme B; MF, methanofuran; HdrA2B2C2, cytoplasmic heterodisulfide reductase; F 420 , coenzyme F 420 ;. Mcr, methyl-coenzyme M reductase; Mer, F 420 -dependent methylene-H 4 SPT reductase; Mtd, F 420 -dependent methylene-H 4 SPT dehydrogenase; Mch, methylene-H 4 SPT cyclohydrolase; Ftr, formylmethanofuran-H 4 SPT formyltransferase; Fmd, formylmethanofuran dehydrogenase; CODH/ACS, CO dehydrogenase/acetyl-CoA synthase complex; Pta, phosphotransacetylase; Ack, acetate kinase; Fdh, formate dehydrogenase. Reactions in red font are an alternative to HdrA2B2C2 for oxidation of F 420 H 2 . Formate was a product for M. orientis sp. nov. and M. acetivorans C2A. The genome of M. orientis sp. nov. is annotated for a formate dehydrogenase gene (FdhA) with 41% sequence identity to the catalytic FdhA subunit of formate dehydrogenase from Methanococcus vannielii and 35% identity to the FdhA of Wolinella succinogenes ( 30 , 31 ). Thus, it is possible that formate is a product of the reduction of CO 2 catalyzed by the putative formate dehydrogenase. No homologs of known formate dehydrogenases are encoded in the genome of M. acetivorans C2A, although formate is a product of growth with carbon monoxide ( 32 ). Here ( Fig. 5 A ) it is postulated that formate is a product of fortuitous formate dehydrogenase activity of tungsten-dependent formylmethanofuran dehydrogenase (Fwd) shown earlier to be involved in the formation of formate in M. acetivorans C2A by an unknown mechanism ( 33 ). Furthermore, proteomic and transcriptomic analyses reveal multiple homologs of formyl-methanofuran dehydrogenase in M. acetivorans C2A for which there is no known function ( 34 ). The finding that acetate was a product of methanotrophic growth of M. acetivorans C2A but not M. orientis sp. nov. reflects physiological differences. Acetate is a substrate for methanogenic growth of M. acetivorans C2A but not M. orientis sp. nov. although both grow with methylotrophic substrates such as methanol or methylamines. In M. acetivorans C2A, phosphotransacetylase (Pta) and acetate kinase (Ack) produce acetyl-CoA. Acetyl-CoA is the substrate for CO dehydrogenase/acetyl-CoA synthase (CODH/ACS) producing CO 2 and CH 3 −H 4 SPT for which the methyl group is reduced to methane. The reversal of reactions by CODH/ACS, Pta, and Ack is a plausible mechanism of acetate formation and substrate level ATP synthesis during methanotrophic growth ( Fig. 5 A ). The production of 13 CH 3 12 COOH and 13 CH 3 13 COOH from 13 CH 4 ( SI Appendix , Fig. S6 C ) supports a role for CODH/ACS wherein the carboxyl groups are derived from 12 CO 2 of the bicarbonate-buffered culture media and 13 CO 2 from complete oxidation of 13 CH 4 . Acetyl-CoA is also an expected precursor for cell biosynthesis. The genome of M. orientis sp. nov. is annotated for genes encoding CODH/ACS although most likely to produce acetyl-CoA for cell biosynthesis ( https://www.ncbi.nlm.nih.gov/genome ). Both pathways propose electron transport and energy conservation roles for Rnf, HdrDE, and HdrA2B2C2 complexes. The complexes of M. acetivorans C2A have been biochemically characterized with Rnf shown to translocate 2 Na + per electron transferred to Fe(III) that drives the endergonic methyl transfer from CH 3 −SCoM to H 4 SPT by Mtr ( Fig. 5 A ) ( 12 ). It is also established that the multiheme cytochrome c (MmcA) of M. acetivorans C2A is essential for transfer of electrons from Rnf to exogenous electron acceptors ( 19 , 35 , 36 ). The genome of M. orientis sp. nov. contains homologs encoding MmcA, Rnf, HdrDE, and HdrA2B2C2 that are included in the pathway ( Fig. 5 B ). The pathways show an alternate to HdrA2B2C2 for the transfer of electrons from F 420 H 2 to Fe(III) involving the Fpo complex, HdrDE, and the electron carrier methanophenazine (MP) that contributes a proton gradient driving ATP synthesis. Although the alternative diminishes electrons available to Rnf, the sodium and proton gradients are unchanged that drive methyl transfer by Mtr and ATP synthesis. M. barkeri reverses methanogenesis dependent on the reduction of Fe(III), although growth was not reported, and it lacks genes encoding the Rnf complex ( 18 ). Instead of Rnf, methanogenic growth of M. barkeri depends on H 2 cycling by the Ech complex to generate a proton gradient for ATP synthesis. However, genes encoding Ech are absent in marine ANME, whereas genes encoding the Rnf complex are repeatedly detected in ANME-2 and ANME-3 which supports an essential role for Rnf in methanotrophic growth ( 14 ). The energy-conserving pathways ( Fig. 5 ) predict greater ATP/methane oxidized for M. orientis sp. nov. vs. M. acetivorans C2A consistent with carbon and electron balances (Eqs. 1 and 2 ) that reveal greater cell biomass yield for M. orientis sp. nov. Assuming that the translocation of 4.0 Na + /H + yields 1.0 ATP ( 37 ), the Y ATP (g dry weight/mole) is 15.7 for M. orientis sp. nov. and 14.4 for M. acetivorans C2A which are within the range ( 10 – 24 ) determined for metabolically diverse microbes ( 38 ). The Y DX (C-mol substrate/C-mol biomass) for M. orientis sp. nov. (0.78) and M. acetivorans C2A (0.68) compares favorably with aerobic growth of Pseudomonas oxalaticus with ethanol ( Y DX = 0.56), considering that methane is 1.33-fold more reduced than ethanol ( 39 ). These results further support the proposed energy-conserving pathways ( Fig. 5 ). The energy-conserving pathways in Figs. 5 A and B are remarkably like pathways proposed for uncultured ANME based on extensive metagenomic analyses ( 3 , 4 , 14 ). Furthermore, M. acetivorans C2A and M. orientis sp. nov. belong to the family Methanosarcinaceae that also includes ANME-2 and ANME-3 ( 14 , 40 ). Thus, methanotrophic metabolisms of M. acetivorans C2A and M. orientis sp. nov. are reasonable representatives of these diverse ANME clades that predict capability of methanogenesis by ANME consistent with the recently reported reversibility of recombinant CODH/ACS from uncultured ANME-2a and ANME-2d ( 41 )."
} | 2,625 |
23029512 | PMC3459915 | pmc | 745 | {
"abstract": "The unique mutualism between corals and their photosynthetic zooxanthellae ( Symbiodinium spp.) is the driving force behind functional assemblages of coral reefs. However, the respective roles of hosts and Symbiodinium in this endosymbiotic association, particularly in response to environmental challenges (e.g., high sea surface temperatures), remain unsettled. One of the key obstacles is to produce and maintain aposymbiotic coral hosts for experimental purposes. In this study, a simple and gentle protocol to generate aposymbiotic coral hosts ( Isopora palifera and Stylophora pistillata ) was developed using repeated incubation in menthol/artificial seawater (ASW) medium under light and in ASW in darkness, which depleted more than 99% of Symbiodinium from the host within 4∼8 days. As indicated by the respiration rate, energy metabolism (by malate dehydrogenase activity), and nitrogen metabolism (by glutamate dehydrogenase activity and profiles of free amino acids), the physiological and biochemical performances of the menthol-induced aposymbiotic corals were comparable to their symbiotic counterparts without nutrient supplementation (e.g., for Stylophora ) or with a nutrient supplement containing glycerol, vitamins, and a host mimic of free amino acid mixture (e.g., for Isopora ). Differences in biochemical responses to menthol-induced bleaching between Stylophora and Isopora were attributed to the former digesting Symbiodinium rather than expelling the algae live as found in the latter species. Our studies showed that menthol could successfully bleach corals and provided aposymbiotic corals for further exploration of coral-alga symbioses.",
"introduction": "Introduction The unique mutualism between corals and their photosynthetic zooxanthellae ( Symbiodinium spp.) underpins ecological success of corals in shallow and oligotrophic seawater. However, this association is highly vulnerable to rising seawater temperatures. A rise of only 1∼2°C above the summer average under moderate to high irradiance will likely be enough to disrupt the symbiotic relationships by causing the symbionts to be expelled from the host, precipitating so-called ‘coral bleaching’ [1] , [2] . Coral bleaching events are known to further cause a breakdown [1] – [4] or phase shift [5] – [7] in coral reefs. These situations are predicted to worsen with time if the increase in seawater surface temperatures cannot be slowed [8] , [9] . In order to understand if corals can survive the coming stressful environments, the mechanisms underlying coral bleaching have been intensively studied (reviewed in Weis [10] ). It is widely accepted that reactive oxygen species (ROS) generated by Symbiodinium photoinhibition and/or mitochondrial dysfunction in the host can cause breakdown of the symbiotic association [10] – [12] . However, the comparative susceptibility of coral hosts and Symbiodinium to thermal stresses is not completely understood. In studies of symbionts, cultured and freshly isolated Symbiodinium (FIS) was widely used to explore the symbiont physiology. Different physiological performances, such as the photosynthesis capability under thermal stress, of FIS or cultured Symbiodinium were also revealed at the clade or subclade levels [13] – [16] . In contrast, studies on physiological responses of aposymbiotic coral hosts are limi'ted due to a lack of suitable protocols. Several methods were used to deplete Symbiodinium from cnidarian hosts, including cold shock (e.g., 4°C) [17] – [19] , a high seawater temperature (e.g., 33°C) [20] , and 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) treatment [21] , but few of them generated healthy aposymbiotic coral hosts which could be used for further studies. Aposymbiotic corals induced by high seawater temperatures either take a long time and need antibiotics treatment [20] or result in high coral mortality [22] . High-temperature treatment might also implant a heat experience in corals which might influence the performance of bleached corals in thermal-tolerance studies. On the other hand, bleaching corals with DCMU requires high light intensities (e.g., 70% of ambient insolation) and large volumes of seawater (ca. 1000 L) to maintain the animals, which prevents laboratories without ample seawater supplies and outdoor facilities from conducting coral-bleaching experiments. Consequently, physiological and biochemical studies on aposymbiotic hosts in Symbiodinium -cnidarian symbioses are mostly confined to sea anemones [19] , [23] – [26] and aposymbiotic larvae from limited coral species [27] , [28] . Nevertheless there are still gaps in applying the knowledge obtained from sea anemones to corals when the coral skeleton, calcification, and surface and endoskeletal microbes should be taken into account [29] . Therefore, a general method needs to be developed to prepare as many species of aposymbiotic corals from adult individuals to conduct comparative analyses among coral species. Menthol is a cyclic terpene alcohol which is usually used to anesthetize cnidarians in marine biological studies [30] . This compound was occasionally found to bleach Symbiodinium -associated corals and sea anemones during anesthetization (unpublished data). Despite menthol's toxicity to corals being unclear, menthol was found to be less toxic to an aquatic invertebrate ( Daphnia magna ) for which the 24-h 50% lethal concentration (LC 50 ) is 37.7∼71.0 mg L −1 \n [31] . In this study, therefore, extant corals from two major lineages, respectively of robust and complex clades, were used to explore a workable procedure to prepare aposymbiotic corals from adult colonies. Furthermore, the physiological and biochemical performances of the aposymbiotic coral hosts were examined, and their comparability to their symbiotic counterparts was evaluated. Feeding an artificial diet was also used to examine the effect of exogenous nutrients on maintaining physiological and biochemical performances of aposymbiotic coral compared to their symbiotic counterparts.",
"discussion": "Discussion In this study, we applied menthol to develop a simple and gentle protocol to prepare aposymbiotic corals which retained comparable physiological and biochemical performances to their symbiotic counterparts by incubation in seawater only (for Stylophora ) or with additional feeding of a nutrient cocktail containing glycerol, vitamins, and a host mimic FAA mixture (for Isopora ). Bleaching coral by menthol, as indicated in Fig. 1B , occurred in a significant dose-dependent manner. However, because continuous incubation always caused high mortality, a repeated 8: 16-h menthol (treatment): ASW (resting) treatment cycle was essential for the success of the protocol ( Fig. 2 ). Menthol is a compound known to act on a variety of different membrane receptors, including the transient receptor potential (TRP)M8, TRPA1, and other ionotrophic receptors [39] . The binding of menthol to TRPM8 results in an increase in intracellular Ca 2+ concentrations and causes a cold sensation in vertebrates [40] – [43] . Menthol was also found to cause antinociceptive and local anesthetic effects in neuronal and skeletal muscles via blocking voltage-operated sodium channels [44] . Menthol is also known to cause many adverse effects to plants, including photoinhibition [45] . In Symbiodinium -associated corals, the mechanism underpinning menthol-induced coral bleaching is not clear. However, based on two different Symbiodinium -releasing modes (ejecting the alga in a cloudy suspension by Isopora and releasing digested alga by Stylophora ), the bleaching mechanism might be attributed to Ca 2+ -triggered exocytosis as described by Pang and Südhof [46] and/or photoinhibition in Symbiodinium . We have no information about Ca 2+ movements in the coral host during menthol treatment, but a preliminary study indicated that menthol might inhibit Symbiodinium photosynthesis II activity in the millimolar range (4-h IC 50 of 0.72∼1.96 mM) which was at a similar level that caused coral bleaching (unpublished data). Further studies are needed to clarify the mechanism of menthol-induced coral bleaching. When depleting Symbiodinium from a cnidarian host, a cessation in the supply of photosynthate released from the algal symbiont would greatly upset the host physiology and metabolism. Although respiration rates of some corals ( Montastraea annularis , Agarwia lamarcki , Porites compressa , and Montipora capitata ) decreased when Symbiodinium algae were depleted [47] , [48] , those of freshly bleached Isopora and Stylophora did not significantly differ from the symbiotic counterparts ( Fig. 4 ). No differences in respiration rates between symbiotic and aposymbiotic corals were found in the temperate coral Astrangia danae , which was interpreted as holozoic feeding in the aposymbiotic coral possibly compensating for the energy loss from the deprivation of photosynthate release by Symbiodinium \n [49] . Because no food sources are available in ASW, energy sources for the bleached Stylophora and Isopora to balance the loss from lack of photosynthate release by Symbiodinium might be derived from consuming previous reserves or digestion of impaired Symbiodinium . Depletion of symbiotic algae would also result in significant changes in nitrogen metabolism of the host [25] , [34] , [50] . For example, GDH, a key enzyme for assimilating (or releasing) ammonium into (or from) amino acids, increases in alga-depleted corals and sea anemones [34] . FAAs, especially the so-called essential amino acids, in the host homogenates were also found to have decreased by nearly half after depletion of symbiotic algae [25] , [50] . In this study, the responses of coral nitrogen metabolism to algal depletion differed between Stylophora and Isopora . Algal depletion caused significant decreases in Isopora GDH activity and FAA contents but not in Stylophora . However, supplementation of the aposymbiotic Isopora with nutrients containing glycerol, a host mimic FAA mixture, and vitamins reverted the nitrogen metabolic indices back to a level and composition comparable to the symbiotic counterpart ( Table 2 , Fig. 5A ). This result is similar to previous findings in Aiptasia \n [25] , [50] . Therefore, aposymbiotic coral generated by expelling Symbiodinium alive during bleaching would need to be fed a proper nutrient supplement before being subjected to physiological studies. With the nutrient A supplement, we successfully maintained Isopora for the test of reinfection with heterogenic Symbiodinium (unpublished data). In summary, comparisons of physiological performances and gene expression profiles between different species of coral hosts per se will be available by preparing freshly bleached aposymbiotic coral with the menthol protocol combined with nutrient supplementation if necessary. This technique will also potentially benefit the search for a generalist coral to re-establish symbiosis with different heterogenic Symbiodinium , which will make the contributions of different Symbiodinium subclades to coral symbiosis more straightforward."
} | 2,802 |
37467329 | PMC10355823 | pmc | 747 | {
"abstract": "Neuromorphic computing (NC) architecture inspired by biological nervous systems has been actively studied to overcome the limitations of conventional von Neumann architectures. In this work, we propose a reconfigurable NC block using a flash-type synapse array, emerging positive feedback (PF) neuron devices, and CMOS peripheral circuits, and integrate them on the same substrate to experimentally demonstrate the operations of the proposed NC block. Conductance modulation in the flash memory enables the NC block to be easily calibrated for output signals. In addition, the proposed NC block uses a reduced number of devices for analog-to-digital conversions due to the super-steep switching characteristics of the PF neuron device, substantially reducing the area overhead of NC block. Our NC block shows high energy efficiency (37.9 TOPS/W) with high accuracy for CIFAR-10 image classification (91.80%), outperforming prior works. This work shows the high engineering potential of integrating synapses and neurons in terms of system efficiency and high performance.",
"introduction": "INTRODUCTION Deep neural networks (DNNs) have shown over-human performance in various fields, such as image classification or natural language processing ( 1 – 6 ). In particular, convolutional neural networks (CNNs) mimicking biological vision systems have become fundamental techniques for high performance in vision fields ( 7 – 9 ). However, the increasing area and power overhead of von Neumann computing architectures, mainly induced by the data communication between processing units and memory, have also become very critical issues in DNNs ( 10 – 14 ). In this regard, interest in neuromorphic computing (NC) architectures, which are inspired by biological nervous systems, has been rapidly increasing ( 15 – 18 ). NC architectures using analog nonvolatile memory cells reduce data communication by computing in the memory domain in an analog manner, enhancing the system efficiency ( Fig. 1A ). Moreover, according to Kirchhoff’s and Ohm’s laws, NC architectures perform massively parallel operations with high energy efficiency and reduced area overhead ( 19 – 23 ). Fig. 1. Proposed NC block. ( A ) Schematic diagram of neuromorphic computing using electronic synaptic devices. The post-synaptic neurons process the signals from the synapses. ( B ) Proposal of the reconfigurable NC block with flash-type synapse arrays, PF neuron devices, and CMOS peripheral circuits. However, conventional NC architectures using emerging memory cells have faced several challenges at the device and circuit levels. The emerging memory cells used for synapse devices, such as ReRAM (resistive random-access memory) ( 24 – 28 ) or PCRAM (phase-change RAM) ( 29 ), have suffered from reliability issues at both the unit device and array levels, including a limited on/off ratio ( 30 – 32 ), device-to-device variation ( 17 , 33 ), and sneak path problem ( 34 – 37 ). High power consumption and a large area of analog-to-digital converters (ADCs) induce ADCs to be shared in NC blocks, lowering system efficiency ( 38 , 39 ). In addition, to perform reliable and accurate vector-matrix multiplication (VMM) operations, precise calibration of ADCs is required, which leads to additional modules or external controllers for the calibration. Therefore, it is necessary to develop a novel NC architecture that can efficiently address these limitations while exhibiting high performance. In this work, we propose a reconfigurable NC block based on a flash-type synapse array and positive feedback (PF) neuron device ( Fig. 1B ) and implement the blocks by integrating not only the synapse array and neurons but also complementary metal-oxide semiconductor (CMOS) peripheral circuits on the same substrate. The operations of proposed NC blocks were experimentally demonstrated in the fabricated blocks. Flash-type synapse arrays with technically mature high reliability can store multi-bit weights and perform VMM accurately. In addition, because of the super-steep switching characteristics of PF neuron devices ( 40 ), the proposed NC block substantially reduces the number of devices to process the synapse signals, addressing the ADC sharing problem very efficiently. Moreover, the proposed NC block can generate a reconfigurable output code by modulating the conductance of flash synapse cells for reference signals, which enables easy calibration as well as the implementation of various activation functions. The successful demonstration of the implemented NC block in this work proves that the cointegration of synapse arrays and neurons mimicking highly efficient biological nervous systems also has promising potential from an engineering perspective, including system efficiency and accuracy.",
"discussion": "DISCUSSION In summary, we have proposed a reconfigurable NC block with an AND-type flash synapse array, super-steep PF neuron devices, and CMOS peripheral circuits. Each component was integrated and connected on the same substrate to implement the NC block efficiently. The proposed NC block exhibits high energy efficiency (37.9 TOPS/W) with a reduced area overhead for analog-to-digital conversions because of the super-steep characteristics of the PF device and flash memory technology. In addition, the fabricated NC block reconfigurably produces output codes, enabling not only easy calibration in the NC blocks but also the implementation of various activation functions. These advantages result in the NC blocks exhibiting a high classification accuracy (91.80%) in CIFAR-10 image classification. Our work provides important technological advancement to effectively overcome the limitations of existing NC architectures and verifies the high potential of integrating synapses and neurons in terms of system efficiency."
} | 1,446 |
36017490 | PMC9396041 | pmc | 748 | {
"abstract": "Lignin is a ubiquitously available and sustainable feedstock that is underused as its depolymerization yields a range of aromatic monomers that are challenging substrates for microbes. In this study, we investigated the growth of Pseudomonas taiwanensis VLB120 on biomass-derived aromatics, namely, 4-coumarate, ferulate, 4-hydroxybenzoate, and vanillate. The wild type strain was not able to grow on 4-coumarate and ferulate. After integration of catabolic genes for breakdown of 4-coumarate and ferulate, the metabolically engineered strain was able to grow on these aromatics. Further, the specific growth rate of the strain was enhanced up to 3-fold using adaptive laboratory evolution, resulting in increased tolerance towards 4-coumarate and ferulate. Whole-genome sequencing highlighted several different mutations mainly in two genes. The first gene was actP , coding for a cation/acetate symporter, and the other gene was paaA coding for a phenyl acetyl-CoA oxygenase. The evolved strain was further engineered for rhamnolipid production. Among the biomass-derived aromatics investigated, 4-coumarate and ferulate were promising substrates for product synthesis. With 4-coumarate as the sole carbon source, a yield of 0.27 (Cmol rhl /Cmol 4-coumarate ) was achieved, corresponding to 28% of the theoretical yield. Ferulate enabled a yield of about 0.22 (Cmol rhl /Cmol ferulate ), representing 42% of the theoretical yield. Overall, this study demonstrates the use of biomass-derived aromatics as novel carbon sources for rhamnolipid biosynthesis.",
"conclusion": "4 Conclusion and outlook This study has reported the successful use of biomass-derived aromatics for producing rhamnolipids using P. taiwanensis VLB120 . This was done by using metabolic engineering approaches for aromatic substrate utilization. In this work, we have accomplished the utilization of ferulate and 4-coumarate by heterologous gene expression of ferulic operon genes ech , fcs , and vdh . This paved the way for using these compounds as substrates, which cannot be consumed by the wild type P. taiwanensis VLB120. Also, the growth on aromatics was improved using adaptive evolution. This was a promising non-rational strain engineering approach that resulted in enormous tolerance towards 4-coumarate and ferulate. With whole-genome sequencing, the genomic alterations during the ALE experiment could be investigated. The evolved strain P. taiwanensis VLB VS2 was successfully engineered to produce mono-rhamnolipids that predominantly consisted of the C10–C10 moiety. 4-coumarate and ferulate were preferred substrates with a titre of about 0.43 g l -1 . The titres we achieved were comparable to that of other unconventional substrates used for rhamnolipid production. Also, this proof of concept study can be applied for industrial applications that involve usage of lignocellulosic biomass. By employing an appropriate method of lignin depolymerization one can yield desired types of selective aromatics. For example, in the mild base-catalyzed depolymerization of corn stover at a temperature of about 120 ◦ C and 2% NaOH, 4-coumarate was the most predominant aromatic compound of about 6 g l -1 followed by ferulate of about 180 mg l -1 . Other compounds such as 4-hydroxybenzoate and vanillate were also present at minimal levels ( Rodriguez et al., 2017 ). So, this study can be extended to valorize low molecular weight lignin compounds by bioconversion of these compounds to value-added products. Thus the use of aromatics as substrates for the production of valuable chemicals contributes to the vision of an integrated biorefinery that uses not only the sugars derived from biomass but also utilizes the untapped lignin.",
"introduction": "1 Introduction Lignin accounts for a substantial amount (15–30%) of lignocellulosic biomass that is abundantly found in our environment. However, lignin is under-utilized due to its high recalcitrance and heterogeneity. Currently, in biorefineries, lignin valorization is just limited to produce heat and electricity. Alternatively, lignin could be depolymerized to generate a range of aromatic monomers, which can be utilized as a substrate for manufacturing bio-based products. Different types of biomass-derived aromatics are generated during hydrolysis depending on the lignin used ( Raj et al., 2007 ; Zhu et al., 2017 ). These aromatics can act as inhibitors for the growth of the commonly used industrial workhorses Saccharomyces cerevisiae and Escherichia coli ( Adeboye et al., 2014 ; Palmqvist, 2000 ). There is a lack of metabolic pathways in some organisms to utilize aromatics. Therefore, there is an increasing need for research on the development of microbial hosts that can tolerate and utilize the aromatics as substrates and produce valuable products ( Becker and Wittmann, 2019 ). Various studies have reported deploying aromatics as substrates for different microbial hosts. Acinetobacter baylyi ADP1 was used to produce alkanes and wax esters ( Salmela et al., 2019 ). With S. cerevisiae , protocatechuic acid was produced ( Zhang et al., 2021 ), and also different Pseudomonas spp. were used for the formation of various industrially essential products such as itaconic acid ( Elmore et al., 2021 ), muconic acid ( Barton et al., 2018 ), adipic acid ( van Duuren et al., 2020 ), polyhydroxyalkanoates ( Borrero-de Acuña et al., 2021 ), 2-pyrone 4,6-dicarboxylic acid ( Qian, 2016 ), and pyruvate and lactate ( Johnson and Beckham, 2015 ). Also, lipid production was achieved in Rhodococci ( Kosa and Ragauskas, 2012 ). To develop a microbial chassis for coupling the utilization of biomass-derived aromatics with product synthesis, Pseudomonas spp. is an attractive candidate due to its versatile metabolism of aromatics ( Weimer et al., 2020 ). These ubiquitously found Gram-negative bacteria have a broad substrate spectrum, including sugars, aromatics, hydrocarbons, and xenobiotics ( Cao et al., 2009 ). The ability of high organic solvent tolerance makes some strains suitable hosts for specific catalytic tasks in industrial applications ( Heipieper and Bont, 1994 ). Pseudomonads have been employed in academia and increasingly in the industry to produce a plenitude of valuable chemicals, including aromatics, polyketides, terpenoids, and glycolipids. Among these products, biosurfactants such as rhamnolipids are industrialized using P. putida as a host due to their immense range of applications in areas such as bioremediation, oil recovery, food processing, cosmetics, medicine, and agriculture ( Borah et al., 2016 ; Husain, 2008 ; Magalhães and Nitschke, 2013 ; Rodrigues et al., 2006 ; Roy et al., 2015 ). High-yield rhamnolipid production is currently reported in plenty of studies with the natural producer Pseudomonas aeruginosa ( Müller et al., 2011 ), but the major drawback of this strain is that it is an opportunistic human pathogen. Therefore, there is a need for heterologous gene expression in non-pathogenic strains. Various studies involved expressing rhlAB genes from P. aeruginosa in non-pathogenic strains to produce rhamnolipids. For instance, in Burkholderia kururiensis , 5.6 g l -1 rhamnolipids were produced from glycerol ( Tavares et al., 2013 ). With glucose as substrate, E. coli produced 55–65 mg l -1 of rhamnolipids ( Haba et al., 2000 ). Additionally, including the rmlBDAC operon to increase rhamnose availability led to 120 mg l -1 of rhamnolipids ( Cabrera-Valladares et al., 2006 ). Germer et al. (2020) reported the production of 200 mg l -1 HAAs, the lipid precursor for rhamnolipids. In the first recombinant study on rhamnolipid production using P. fluorescens and P. putida , 0.25 g l -1 and 0.6 g l -1 of products were reported, respectively ( Ochsner et al., 1995 ). Rhamnolipid composition was altered using the rhlAB operon from Burkholderia glumae in P. putida KT2440 , resulting in long-chain rhamnolipids ( Wittgens et al., 2018 ). Other than conventional substrates such as glucose, glycerol, and plant oils, many reports use renewable resources and industrial waste to produce rhamnolipids. Mostly different strains of P. aeruginosa have been used so far. Rhamnolipid titres of 200 mg l -1 was achieved from olive oil mill waste ( Moya Ramírez et al., 2015 ). Also using waste canola oil as substrate, 3 g l -1 of rhamnolipids was reported ( Pérez-Armendáriz et al., 2019 ). Furthermore, when grown on diesel and kerosene, 1 g l -1 and 700 mg l -1 of rhamnolipids were produced, respectively, using P. aeruginosa J4 ( Wei et al., 2005 ). When cultivated on waste cooking oil, P. aeruginosa M4 produced 120 mg l -1 of rhamnolipids ( Shi et al., 2021 ) . Recently, rhamnolipid production using plastic monomers as a carbon source was reported ( Tiso et al., 2021 ; Utomo et al., 2020 ). To the best of our knowledge, there are no reports on rhamnolipid biosynthesis from biomass-derived aromatics as a novel carbon source. In order to develop a microbial chassis for the production of rhamnolipids, Pseudomonas taiwanensis VLB120 is a promising candidate. Generally, hydrolysate of lignocellulosic biomass would contain pentoses from hemicellulose. For xylose being the most predominant building block of hemicellulose, there have been reports for engineering P. putida KT2440 for its utilization by heterologous expression of xylAB genes from E. Coli ( Wang, et al., 2019 ). But the major advantage of using VLB120 is that it can natively use xylose as a substrate without the need for engineering ( Köhler et al., 2015 ). Also, VLB120 has an important trait of high solvent tolerance that is desirable for industrial applications ( Gross et al., 2010 ; Volmer et al., 2014 ). Additionally, this organism has the ability to do bioconversion using renewable substrates like glycerol ( Lenzen et al., 2019 ). These attributes make them a suitable host for metabolic engineering. Various studies report the use of this strain as a production host for the biosynthesis of value-added products such as 4-hydroxybenzoate ( Lenzen et al., 2019 ), n-octanol ( Gross et al., 2013 ), phenol ( Wynands, 2018 ), and isobutyric acid and isobutanol ( Lang et al., 2014 ). In this study, a metabolically engineered P. taiwanensis VLB120 was developed to grow efficiently on chosen biomass-derived aromatics 4-coumarate, ferulate, 4-hydroxy benzoate and vanillate. Out of all these aromatics, the wild type strain was able to grow natively on 4-hydroxy benzoate and vanillate. For growth on 4-coumarate and ferulate, the pathway genes necessary for the degradation of these aromatics were integrated into the genome. Further, to enhance growth on non-native substrates, adaptive laboratory evolution was performed to achieve better tolerance. Whole genome sequencing was done to understand the mechanism at the genetic level that resulted in the enhanced phenotype. Finally, the evolved strain with increased tolerance was equipped with the required pathway genes for the production of rhamnolipids. Overall, this study demonstrated the utilization of biomass-derived aromatics for the production of value-added chemicals.",
"discussion": "3 Results and discussion 3.1 Establishing growth of P. taiwanensis VLB120 on biomass-derived aromatics Pseudomonads have in-built machinery of enzymes and pathways to metabolize various biomass-derived aromatics ( Paliwal et al., 2014 ). In order to develop a microbial chassis of P. taiwanensis VLB120 to utilize the aromatics, the presence of respective pathways for the catabolism of aromatics was revisited. P. taiwanensis VLB120 possesses pob A (PVLB_11545) that encodes a 4-hydroxybenzoate monooxygenase and van AB (PVLB_10605, PVLB_10610) encoding a vanillate monooxygenase. Both these genes are responsible for converting 4-hydroxybenzoate and vanillate to protocatechuate. However, P. taiwanensis VLB120 does not possess the metabolic pathway genes coding for the feruloyl-CoA synthetase ( fcs ), enoyl-CoA hydratase ( ech ), and vanillin dehydrogenase ( vdh ), required for the degradation of ferulate and 4-coumarate. Therefore, wild type P. taiwanensis VLB120 grew on 4-hydroxybenzoate and vanillate but not on 4-coumarate and ferulate ( Supplementary Fig. 1 ). In order to utilize ferulate as well as 4-coumarate, heterologous genes fcs, ech, and vdh from the ferulic operon of P. putida S12 were genomically integrated into wild type P. taiwanensis VLB120 using the Tn5 mini-transposon system ( Nikel and de Lorenzo, 2013 ; Lenzen et al., 2019 ). Both ferulate and 4-coumarate are degraded using the protocatechuate branch of β-ketoadipate pathway with vanillate as intermediate for ferulate and 4-hydroxybenzoate for 4-coumarate, respectively ( Ravi et al., 2017 ). After conjugation, three clones were randomly picked from selective plates. Since all the clones grew in LB medium with the same performance (data not shown), it was assumed that neither of the integration sites had a negative impact on overall cell fitness. By the same token, we also excluded a direct relation between the disrupted genes and the aromatic catabolic pathway. Therefore, one of the clones was taken for all further experiments, referred to as P. taiwanensis VLB120 VS1. After introducing the genes for ferulate and 4-coumarate degradation, the growth of the strain P. taiwanensis VLB120 VS1 was investigated in MSM containing 10 mM glucose equivalent to individual substrates, as shown in Fig. 1 . Since ferulate and 4-coumarate are the upstream compound in the metabolic pathway, the preculture was grown with 10 mM ferulate and 4-coumarate as a sole carbon source, respectively. Growth rate was chosen as a key indicator for growth performance and substrate affinity. As the equivalent of 10 mM of glucose was applied for all aromatics, final G Values shown in Fig. 1 correspond to a final biomass of ∼1.0 OD. Unlike the wild type strain, P. taiwanensis VLB120 VS1 was found to utilize and grew on all the aromatics. The growth on 4-hydroxybenzoate was highest with a growth rate of 0.36 h -1 , which was faster than on glucose (0.3 h -1 ), while cells grew slower on vanillate (0.17 h -1 ). Growth rates in the presence of 4-coumarate (0.07 h -1 ) and ferulate (0.05 h -1 ) were low, with extended lag phases (7 h for 4-coumarate and 9 h for ferulate). Nonetheless, integration of the peripheral genes fcs, ech, and vdh rendered strain VLB120 VS1 capable of using major aromatics as the sole carbon source and generated a platform for their valorization. Fig. 1 Growth profiles of engineered Pseudomonas: P. taiwanensis VLB120 VS1 were cultivated in 24-well System Duetz plates in a Growth Profiler in MSM containing 10 mM glucose equivalent of each aromatic as the sole carbon source. The working volume of each well was 1.5 ml. Cells were maintained at 30 ⁰C and 300 rpm. The values shown here are from biological triplicates, and error bars indicate the standard error of the mean. Fig. 1 3.2 P. taiwanensis VLB120 VS1 shows enhanced growth on ferulate and 4-coumarate after adaptive evolution Generally, different biomass-derived aromatics exhibit different levels of toxicity towards the microorganisms. For instance, in S. cerevisiae , ferulate was found to have a toxicity limit of 1.8 mM, but 4-coumarate had a higher toxicity limit of 9.7 mM ( Adeboye et al., 2014 ). Improvement of microbial traits can be accomplished by adaptive laboratory evolution ( Sandberg et al., 2020 ). Using this non-rational and straightforward method, host properties linked to growth can be enhanced rapidly without knowing the exact genetic or metabolic basis beforehand. In order to improve the ability of P. taiwanensis VLB120 VS1 to assimilate ferulate and 4-coumarate, the strain was cultivated in the presence of these aromatics as mixed carbon sources. The experiment was carried out in three parallel shake flask cultivations. ALE was performed with MSM containing 4-coumarate and ferulate mixture for about 220 generations for 42 days ( Fig. S2 ). After screening 50 mutants from the evolution approach, a strain with the highest growth rate on ferulate and 4-coumarate was selected and referred to as P. taiwanensis VLB120 VS2 (data not shown). The growth of the evolved VLB120 VS2 strain was investigated at varying concentrations compared to the non-evolved strain VLB120 VS1. The evaluation was done with individual substrates at different concentrations ranging from 5 mM to 30 mM. While cultivating cells in MSM, the lag phase in 4-coumarate was shortened by 89% and by 80% in ferulate ( Fig. S3 - A, B, C, D). An 80% reduction in lag phase was also observed when VLB120 VS2 was grown on a mixture of substrates ( Fig. S3 - E, F). The growth rates improved up to 78% in 4-coumarate, 82% in ferulate, and 67% in the mixture ( Fig. S4 ). Growth of the evolved VLB120 VS2 strain was also investigated under 4-hydroxybenzoate and vanillate conditions. However, no considerable improvement in the lag phase or growth rate could be observed in these aromatics. 3.3 Analysis by whole genome sequencing To investigate the molecular basis of the improved growth of P. taiwanensis VLB120 VS2 on ferulate and 4-coumarate as sole carbon sources, the genome of this strain and the operon consisting of degradative ferulic genes fcs, ech , and vdh , were isolated and sequenced. No mutations were found in the ferulic promoter region and the respective genes. Instead, strong accumulation of mutations was found within two loci of the host genome, including single nucleotide exchanges and insertions ( Table 2 ). One locus was found to be coding for the ActP transporter (PVLB_12535). This sodium:solute symporter is involved in the specific transport of acetate and other small aliphatic monocarboxylates ( Gimenez et al., 2003 ). Its expression is controlled by the CrbS/R regulon, and wherein actP is co-expressed with acs. The acs gene codes for acetyl-CoA synthase that uses acetate as a substrate ( Jacob et al., 2017 ; Sepulveda and Lupas, 2017 ). We identified nine single-nucleotide exchanges resulting in amino acid substitutions and a frameshift caused by the insertion of seven base pairs within the open reading frame of actP. The second mutational hotspot was found to be paaA . This gene is a part of the paaABCDE operon that codes for phenylacetyl-CoA oxygenase of the phenylalanine and phenylacetate catabolic pathway that uses phenylacetyl-CoA as a substrate ( Erb et al., 2008 ; Fernández et al., 2006 ; Teufel et al., 2010 ). There were five nucleotide exchanges identified in this gene. In addition to these two genes, there were further mutations, including SNPs and INDELs in the intergenic regions and the region coding for hypothetical proteins in the genome ( Table S2 ). In a similar study, an evolution experiment was performed in 4-coumarate and ferulate using P. putida KT2440. This study revealed frequent mutations in a gene coding for a hypothetical protein PP_3350 and a gene ttg B that codes for an efflux pump membrane transporter along with mutations in genes involved in flagellar movement and transcriptional regulation ( Mohamed et al., 2020 ). The gene targets with similar identity of that in KT2440 can also be targeted for reverse engineering in order to achieve efficient utilization of biomass-derived aromatics. Table 2 Key mutations found in whole-genome sequencing of P. taiwanensis VLB120 VS2. Table 2 Position in the genome Type of mutation No. of mutations Gene/gene product 2769737 - 2771299 SNP, INDEL 10 actP/acetate permease (cation/acetate symporter) 2774408 - 2775397 SNP 5 paaA /phenylacetate - CoA oxygenase subunit 3.4 Reverse engineering of VLB120 VS1 Using the above information from the genome sequencing, we targeted the genes actP and paaA for reverse engineering of the P. taiwanensis VLB120 VS1 strain. The mutations in actP show that the gene had lost its function in the evolved strain P. taiwanensis VLB120 VS2. Therefore, we knocked out the actP gene in the unevolved strain P. taiwanensis VLB120 VS1. Cultivation of VLB120 VS1 Δ actP in MSM containing 4-coumarate ( Fig. 2 A) or ferulate ( Fig. 2 B) showed an intermediate phenotype between the non-mutated strain VLB120 VS1 and evolved strain VLB120 VS2. However, the reason for the improved growth of VLB120 VS1 Δ actP on 4-coumarate and ferulate is not that obvious. As already mentioned, there is evidence of a coupled expression of the ActP transporter and the acetyl-CoA synthase Acs ( Jacob et al., 2017 ; Sepulveda and Lupas, 2017 ). The absence of the transporter in strain VLB120 Δ actP may result in an increased CoA-SH pool due to decreased acetate intake and activity of Acs, which in turn can be exploited by feruloyl-CoA synthase for faster processing of 4-coumarate and ferulate in the peripheral pathway. Fig. 2 Cultivation experiments were carried out in the growth profiler. Growth profiles were monitored for the strains VLB120 VS1, VLB120 VS2, and VLB120 VS1 Δ actP in 4-coumarate (A) and ferulate (B) as the sole carbon sources. Also, the growth of strains VLB120 VS1 Δ paaA and VLB120 VS1 paaA overexpression in 4-coumarate (C) and ferulate (D) was monitored. All the experiments were performed with biological duplicates, and error bars represent the standard error of the mean. Fig. 2 Upon targeting paaA for reverse engineering, it was not evident whether mutagenesis of paaA in the evolved strain caused a loss or gain of gene function since there was no frameshift mutation in the gene. Therefore, for evaluating loss of function, we knocked out paaA , thereby generating mutant strain VLB120 VS1 Δ paaA . For evaluating the gain of function, we genomically integrated a copy of the evolved paaA into the attTn7 site of VLB120 VS1 for overexpression from a medium-strength constitutive promoter. The generated strain, VLB120 VS1 BG14d paaA , and its deficient counterpart, VLB120 Δ paaA , were grown in MSM with 4-coumarate ( Fig. 2 C) or ferulate ( Fig. 2 D). In comparison to the non-mutated strain VLB120 VS1, the overexpression strain VLB120 VS1 BG14d paaA exhibited biomass formation with an increased growth rate as well as a shorter lag phase, but not to the extent of the evolved strain VLB120 VS2. Conversely, knockout of paaA in strain VLB120 VS1 had a negative effect on the growth on aromatics, in that mutant strain VLB120 VS1 Δ paaA grew even slower with a more prolonged lag phase than the non-evolved strain. We concluded from these observations that paaA is involved in the catabolic breakdown of 4-coumarate and ferulate. It encodes the PaaA subunit of an enoyl-CoA hydratase complex using phenylacetyl-CoA as a substrate. Due to its structural similarities to the activated forms of ferulate and 4-coumarate, feruloyl-CoA and 4-coumaroyl-CoA, and the extremely slow growth of the paaA deficient strain, it can be assumed that PaaA acts in parallel to Ech as being part of the peripheral catabolic pathway. Overexpression of paaA , in turn, leads to higher enzyme activity and thus to faster processing, whereas it remains to be demonstrated that mutation in the evolved paaA version, in particular, has the highest effect. Also the role of paaA in the catabolism of 4-coumarate and ferulate can be understood by knocking out ech gene in the evolved strain and check if paaA can replace its function. In summary, Fig. 3 shows the growth on all the strains. Both actP knockout and paaA overexpression results in enhanced catabolic processing of ferulate and 4-coumarate. Both reverse engineering modifications show an intermediate phenotype between non-evolved (VLB120 VS1) and evolved strain (VLB120 VS2); a synergistic effect of these two key components can be anticipated. Therefore further metabolic engineering was therefore attempted with the VLB120 VS2 strain, which had the best phenotypic characteristics compared to the wild type (VLB120 VS1) and the reverse-engineered strains. Fig. 3 Comparison of specific growth rates of the reverse-engineered strains along with non-evolved and evolved strains. All the experiments were performed with biological duplicates, and error bars represent the standard error of the mean. Fig. 3 3.5 Rhamnolipid production Rhamnolipids are a promising alternative to chemically derived surfactants due to their biodegradability, low toxicity, and their ability for metal ion complexation. Several organisms have been utilized as hosts for microbial rhamnolipid production, among which Pseudomonas species are predominant ( Tiso et al., 2017 ). Mainly sugars or plant oils were used as the substrate, with plastic monomers as an exception ( Utomo et al., 2020 ). This study uses biomass-derived aromatics as substrates since P. taiwanensis VLB120 was successfully used as a production host for aromatics from sugars ( Lenzen et al., 2019 ; Wynands, 2018 ; Wynands et al., 2019 ). The primary objective of this study was to explore the utilization of aromatics as carbon sources for rhamnolipid production. We, therefore, established de novo mono-rhamnolipid production in our evolved strain P. taiwanensis VLB120 VS2. Here, the two genes rhlA and rhlB from P. aeruginosa were introduced. rhlA encodes an acyltansferase responsible for generating 3-(hydroxyalkanoyloxy)alkanoic acid (HAA) derived from the fatty acid de novo synthesis. This activated β-hydroxy fatty acid is then linked to rhamnose via dTDP-rhamnose with the help of a rhamnsosyltransferase encoded by rhlB . For aromatic-based rhamnolipid production, the evolved strain P.taiwanensis VLB120 VS2 was equipped with plasmid pSK02 harboring rhlAB genes from P. aeruginosa under expression control of a constitutive stacked promoter ( Bator et al., 2020 ). The best producing strain was visually identified after conjugation by formation of the largest halos-producing strain on the cetrimide blood agar plates due to hemolytic activity ( Fig. S5 ). The newly-engineered strain for rhamnolipid production was labelled as VLB120 VS3. The overall pathway for the production of rhamnolipid biosynthesis from biomass-derived aromatics is represented in Fig. S6 . All the aromatics tested and their corresponding rhamnolipids production is listed in Table 3 . Among the tested aromatic substrates, growth on ferulate and 4-coumarate generated the highest rhamnolipid titer (∼ 0.43 g l -1 ). Approximately half of this amount was obtained with vanillate as substrate, whereas 4-hydroxybenzoate yielded the least. Wittgens et al. (2012) reported a growth-independent rhamnolipid production of an engineered P. putida strain during growth on LB medium supplemented with glucose, wherein components of the LB medium are assimilated for biomass formation, and glucose was used by the cell for rhamnolipid biosynthesis. In contrast to this system, the aromatics used here act as a carbon source for cell growth and are channelled into product formation. Catabolic breakdown of ferulate as well as 4-coumarate is initiated by their activation to acyl-CoAs catalyzed by feruloyl-CoA synthetase encoded by the fcs gene. Subsequent enzymatic breakdown by enoyl-CoA hydratase, the gene product of ech , then releases one molecule acetyl-CoA per molecule ferulate and 4-coumarate, respectively. The additional gain of acetyl-CoA during this reaction may be beneficial for rhamnolipid precursor production within de novo fatty acid biosynthesis and therefore explain higher rhamnolipid titers reached during growth on ferulate and 4-coumarate. While dealing with a mixture of all these aromatics, a maximum titer of about 0.21 g l -1 was achieved. This was comparable to that of rhamnolipid production with individual aromatics such as coumarate or ferulate. This might be due to the effect of catabolite repression that was observed when dealing with multiple substrates ( Fig. S7 ). Since there is the sequential utilization of hydroxybenzoate and vanillate in the initial phase, followed by coumarate and ferulate, it is most likely that the culture growing on a mixture of these substrates utilizes hydroxybenzoate and vanillate more for biomass production while coumarate and ferulate are utilized more for rhamnolipid production. One can perhaps achieve simultaneous utilization of all the substrates and better productivity by eliminating the crc gene, a global regulator of carbon catabolite repression ( Johnson et al., 2017 ). Table 3 Rhamnolipid production of P. taiwanensis VLB120 VS3 . Production of rhamnolipid was carried out in MSM containing 10 mM of the respective biomass-derived aromatics as the sole carbon source in shake flask cultivations. Table 3 LDM Biomass [g l -1 ] Rhamnolipid titer [g l -1 ] Yield (Cmol rhl /Cmol subsrate ) Ferulate 1.76 ± 0.2 0.432 ± 0.022 0.22 4-Coumarate 2.1 ± 0.3 0.436 ± 0.006 0.27 Vanillate 1.8 ± 0.15 0.236 ± 0.004 0.14 4-Hydroxybenzoate 1.7 ± 0.16 0.131 ± 0.005 0.09 Mixture 1.8 ± 0.14 0.210 ± 0.015 0.06 The highest Cmol yield was achieved during growth on 4-coumarate, corresponding to 0.22 C mol /C mol . One production strain engineered by ( Wittgens et al., 2012 ) produced 0.22 g l -1 mono-rhamnolipids in LB medium supplemented with 10 g l -1 glucose. Further improvement of production performance was achieved by using a PHA-negative mutant of P. putida deficient for the PhaC1 poly(3-hydroxy alkanoic acid) synthase 1. During de novo fatty acid biosynthesis, this enzyme catalyzes the formation of PHA from the precursor β-hydroxyaryl-CoA under specific environmental circumstances, e.g., nitrogen limitation. Competition of PHA accumulation and storage with rhamnolipid biosynthesis was prevented by knocking out the PHA synthase gene, leading to an almost seven-fold increase in mono-rhamnolipid titer. PHA synthesis is widespread among bacteria, and the phaC1 gene can be found in the genome of P. taiwanensis VLB120 (PVLB_02155). Hence rendering strain VLB120 VS3 deficient for PHA accumulation will likely boost mono-rhamnolipid production from aromatics. However, with the setup established here, mono-rhamnolipid production in MSM from biomass-derived aromatics was shown to be feasible and even competitive with the use of glucose."
} | 7,575 |
28361923 | PMC5374439 | pmc | 750 | {
"abstract": "Coral reefs are subject to coral bleaching manifested by the loss of endosymbiotic algae from coral host tissue. Besides algae, corals associate with bacteria. In particular, bacteria residing in the surface mucus layer are thought to mediate coral health, but their role in coral bleaching is unknown. We collected mucus from bleached and healthy Porites lobata colonies in the Persian/Arabian Gulf (PAG) and the Red Sea (RS) to investigate bacterial microbiome composition using 16S rRNA gene amplicon sequencing. We found that bacterial community structure was notably similar in bleached and healthy corals, and the most abundant bacterial taxa were identical. However, fine-scale differences in bacterial community composition between the PAG and RS were present and aligned with predicted differences in sulfur- and nitrogen-cycling processes. Based on our data, we argue that bleached corals benefit from the stable composition of mucus bacteria that resemble their healthy coral counterparts and presumably provide a conserved suite of protective functions, but monitoring of post-bleaching survival is needed to further confirm this assumption. Conversely, fine-scale site-specific differences highlight flexibility of the bacterial microbiome that may underlie adjustment to local environmental conditions and contribute to the widespread success of Porites lobata .",
"discussion": "Discussion In this study, we compared bacterial community composition of the SML from bleached and healthy coral colonies of P. lobata from the PAG and RS in order to determine their structure, stability, and putative functional profiles. Our results show that bacterial community composition of coral mucus is highly similar between bleached and healthy corals. Importantly, core bacterial microbiome members are comprised of abundant and rare bacterial associates, whereas site-specific and health state differences exist for less abundant bacteria. Further, algal symbiont association shows a similar pattern to bacterial community patterns, as we did not find differences between health states, but regional differences. For instance, we could confirm the presence of Symbiodinium type C3 in P. lobata from the PAG 43 44 45 , whereas P. lobata from the RS were associated with Symbiodinium types C15 (and variants thereof) as well as types from clade D 43 46 . However, we did not find apparent differences between bleached and healthy colonies. Hence, it remains to be seen whether a causal relationship between Symbiodinium and bacterial community patterns exists. Given that corals from both the PAG and RS are able to survive seasonal temperature maxima exceeding those form other regions, at least in part due to harboring algal thermal tolerant symbionts that are commonly associated with high temperature environments 44 45 47 , it would be intriguing to find bacterial associates that co-occur with symbiont types. In a recent study with the coral model Aiptasia , microbial community patterns were distinct between symbiotic and aposymbiotic anemones, arguing for a connection between bacterial community composition and the cnidarian-algal symbiosis 48 . At large, the presence of photosymbionts distinguishes the microbiomes of hosts from those without photosymbionts 49 , but microbiomes of juvenile corals hosting different Symbiodinium clades were indistinguishable 50 . To our knowledge, this is the first study that compares mucus-associated bacteria from bleached and healthy P. lobata colonies. Porites spp. have high production rates of mucus that cover coral colonies in the form of mucus sheets that exhibit a distinct ageing cycle making it an ideal model system to study dynamics of the mucus-associated microbiota 20 23 . Commonly, a new fluid mucus layer is produced about every four weeks 20 23 . The microbiome of coral surface mucus has an important role in mediating holobiont health 20 , and hence, the periodical release of mucus supposedly supports maintaining a beneficial bacterial microbiome via the removal of undesirable bacteria from the coral colony surface. Notably, mucus surfaces in many domains of multicellular life provide a protective barrier function and are assumed to constitute a common organismal feature 51 . This is supported by our data, as we find that mucus-associated bacteria in bleached P. lobata are similar to those in healthy P. lobata colonies, and long-term monitoring of their post-bleaching survival rates could further corroborate this assumption. In contrast, Bourne, et al . 32 showed a shift of bacteria associated with the tissue of bleached corals. This indicates that mucus-associated bacterial communities may be less dynamic than those that are tissue-associated. Accordingly, the stably associated bacteria in the SML may provide a protective function, even and especially when coral health is compromised as during coral bleaching. In this regard, Lee, et al . 52 showed that under heat stress the chemical composition of mucus in the coral Acropora muricata changed, which might either influence the associated bacterial community or be a consequence of it. Based on the presence of stable bacterial associates in our study, we infer that the mucus chemical composition did not change, although coral bleaching likely affected the availability of carbohydrates for mucus production 23 . It is interesting to note that all mucus samples were dominated by few OTUs that were previously reported from saline environments, arguing that the high salinity of the PAG and RS might indeed comprise a structural determinant of bacteria associated with corals in these regions. Further, the consistent presence of these taxa in all coral mucus samples irrespective of site or bleaching state implies that they play an important role in the coral holobiont. Pseudomonas veronii has previously been found in fungiid corals experimentally exposed to high salinities (49 PSU) 41 . Dietzia sp. is found in the marine environment and in soil, human skin, and the intestinal tract of a carp, and plays a role in biodegradation, bioremediation, industrial fermentation, and carotenoid pigmentation 53 . The presence of Brachybacterium sp. has previously been reported in oil-contaminated coastal sand 54 and salt-fermented seafood 55 . In contrast, site-specific bacterial taxa displayed lower abundance on average. Nevertheless, these bacteria suggest that Porites spp. have the ability to harbor flexible, and presumably locally adjusted microbiomes, which might at least in part contribute to the resilience of this coral genus 56 57 . Predictive bacterial functional profiling between the PAG and the RS revealed differences in the abundance of bacteria associated with sulfur and nitrogen cycling. Differences in sulfur cycling included a scarcity of ‘Sulfur oxidizer’ and ‘Sulfate reducer’ and an increased abundance of ‘Sulfide oxidizer’ in samples from the PAG. Corals and especially their endosymbiotic algae are major producers of dimethylsulfoniopropionate (DMSP) 58 . Its breakdown products, such as dimethyl sulfoxide (DMSO), result mainly from bacterial metabolism and play a significant role in the scavenging of harmful reactive oxygen species (ROS) 59 . Importantly, Symbiodinium produce elevated levels of ROS during thermal stress, which may result in coral bleaching 60 61 . Consequently, the high temperatures in the PAG likely triggers increased ROS production demanding increased availability of ROS scavengers such as DMSP and DMSO, which could explain the functional differences in sulfur cycling observed in SML-associated bacteria between the PAG and the RS. Differences in nitrogen cycling included increased abundance of ‘Dinitrogen-fixing’, ‘Ammonia oxidizer’, and ‘Nitrite reducer’ in the RS. Efficient nitrogen fixation and nitrogen recycling is essential for corals to thrive in nutrient-limited environments 62 . The RS constitutes a highly oligotrophic environment where nitrogen is presumably not readily available for corals. Compared to the RS, nitrogen is not a limiting nutrient in the PAG 63 , allowing for comparably high uptake for nitrogen sources (increased abundance of ‘Nitrogen fixation’) 64 and lacking the need for efficient nitrogen recycling (decreased abundance of ‘Ammonia oxidizer’). From our analyses, functional profiling supports that environmental conditions strongly influence bacterial nitrogen fixation in corals 65 . Taken together, in this study we found stable bacterial communities in the SML of bleached and healthy coral colonies of P. lobata from the PAG and the RS. This underscores the barrier function of coral mucus and we argue that bleached coral colonies benefit from the stable composition and distribution of SML-associated bacteria that presumably provide protective functions. In line with this, we found several abundant and ubiquitous bacterial taxa that we identified as core bacterial microbiome members of coral mucus. Further, regional differences in the mucus bacterial microbiome between PAG and RS were represented by less abundant bacteria that could be associated with a shift in predicted bacterial functional profiles. The specific regional bacterial taxa may thus contribute to the success of P. lobata colonies across a range of environmental conditions."
} | 2,333 |
29158944 | PMC5660801 | pmc | 752 | {
"abstract": "Bioelectrochemical systems such as microbial fuel cells (MFCs) are promising new technologies for efficient removal of organic compounds from industrial wastewaters, including that generated from swine farming. We inoculated two pairs of laboratory-scale MFCs with sludge granules from a beer wastewater-treating anaerobic digester (IGBS) or from sludge taken from the bottom of a tank receiving swine wastewater (SS). The SS-inoculated MFC outperformed the IGBS-inoculated MFC with regard to COD and VFA removal and electricity production. Using a metagenomic approach, we describe the microbial diversity of the MFC planktonic and anodic communities derived from the different inocula. Proteobacteria (mostly Deltaproteobacteria) became the predominant phylum in both MFC anodic communities with amplification of the electrogenic genus Geobacter being the most pronounced. Eight dominant and three minor species of Geobacter were found in both MFC anodic communities. The anodic communities of the SS-inoculated MFCs had a higher proportion of Clostridium and Bacteroides relative to those of the IGBS-inoculated MFCs, which were enriched with Pelobacter . The archaeal populations of the SS- and IGBS-inoculated MFCs were dominated by Methanosarcina barkeri and Methanothermobacter thermautotrophicus , respectively. Our results show a long-term influence of inoculum type on the performance and microbial community composition of swine wastewater-treating MFCs.",
"conclusion": "5. Conclusion This research demonstrates the importance of inoculum source on the electrogenic and degradative activities and ultimate microbial community composition of SW-treating MFCs. MFC treatment of SW is a potentially more environmentally friendly alternative to energetically costly aerobic treatment or odorous space-demanding anaerobic lagoons. Particularly, the comprehensive analysis of SS- and IGBS-MFCs treated SW revealed that electricity production by MFC pairs remained relatively stable; however, the current density of the SS-MFCs (56.6 ± 2.4 mA m −2 169) was higher. Both MFC pairs displayed the ammonia removal. Among all VFAs propionic and acetic acids were found as dominated. Negligible concentrations indoles were detected. Aromatic compounds as p-cresol and phenol were not found. Analysis of microbial communities of both MFCs showed that MFC anodic communities form their own distinct cluster which contains Geobacter spp., represented by eight predominant and three minor species in both MFC anodic communities. Clustering of microbial communities based on dominant bacterial genera indicates that the electrogenic communities in the MFC developed from their inocula. Spectrum of dominated bacteria is significantly enriched by genera Pelobacter, Pseudomonas, Arcobacter, Syntrophus, Syntrophobacter, Bacteroides, and Clostridium and two acetoclastic methanogens ( Methanosarcina and Methanothermobacter ).",
"introduction": "1. Introduction Livestock farming constitutes an important agricultural sector of many countries but produces considerable amounts of organic wastes that require proper treatment and disposal. The rapidly growing pig farming industry generates high-strength wastewater containing organic compounds, ammonia, phosphates, odorous gases, suspended solids, and pathogens [ 1 ]. Treating swine wastewater is especially difficult where land is limited and pig farming facilities occur in close proximity to population centers, such as in Okinawa, Japan. The lack of available land for application of swine wastewater (SW) as a fertilizer and potential for contamination of surface and ground water sources underscore the need to employ thorough treatment of SW. Common methods of treating SW include aerobic oxidation ponds, lagoons, anaerobic digestion, and constructed wetlands [ 2 ]. Bioelectrochemical systems such as microbial fuel cells (MFCs) are promising new technologies for efficient removal of organic compounds in wastewaters. Inside the confined anaerobic chamber of an MFC, a consortium of bacteria catalyze oxidation reactions, depositing electrons on the anode by a variety of means, such as directly via outer membrane proteins or conductive pili or indirectly via secretion and recycling of redox-active molecules [ 3 ]. A primary target of SW treatment is a set of volatile fatty acids (VFAs) largely responsible for its noxious odor [ 4 ]. The presence of VFAs in an MFC substrate can increase the electrogenic performance of its anodic microbial biofilm [ 5 ]. Laboratory-scale single batch-loaded MFCs have been shown to dramatically lower malodorous compounds (primarily VFAs) as well as other constituents present in SW [ 4 ]. One important determinant of MFC reactor performance is the composition of the microbial community in the anodic chamber [ 6 ]. For obtaining maximal initial power production, the anodic biofilm of an existing MFC has been shown to serve as a better inoculum than anaerobic sludge, but we know of no study that assesses inoculum performance relative to pollutant removal criteria [ 7 ]. To this end, we sought to determine whether a microbial community already familiar with a SW substrate would perform better in an MFC than a distinct beer waste-digesting anaerobic sludge, assessing treatment performance and microbial community composition. Previous studies have assessed microbial community composition in SW-fed MFCs utilizing denaturing gel gradient electrophoresis, while a more recent study has utilized high-throughput amplicon sequencing to examine influences of external resistance and hydrodynamics on the MFC microbiome [ 8 , 9 ]. Using a metagenomic approach here we describe the microbial diversity of the MFC planktonic and anodic communities derived from the different inocula. Clustering of microbial communities based on dominant bacterial genera indicates that the nature of the inoculum is an important influence on the ultimate composition of microbial communities and performance of MFCs.",
"discussion": "4. Discussion This study demonstrated the compositions and phylogenetic distributions of SW, inocula, anodic, and planktonic microbial communities in SS- and IGBS-inoculated MFCs. The results showed insignificant differences in bacterial richness and diversity between microbial communities of both MFCs, while SW differed significantly. 4.1. MFCs Treatment Efficiency of SW Treatment of SW using MFCs inoculated with two different inoculums achieved substantial COD removal rates. A previous study found that a single-chambered MFC with a working volume 28 mL removed only 27% of the COD in SW having a high initial COD of 8,320 mg L −1 after 44 h (Min et al., 2005), whereas we found 76.4% and 65.7% removal of COD from SW by the SS-MFCs and IGBS-MFCs, respectively, after 48 h (data not shown). The average current density of the MFCs ( Figure 1 , Table 1 ) was within the range reported for other wastewater-fed MFC systems [ 5 ]. Consistent with other reports [ 13 , 14 ], differences in external resistance within the range we tested (10–1000 Ω ) did not notably alter the performance of the MFCs (data not shown). Swine wastewater is characterized by high content of VFAs although their initial concentration in raw swine wastewater across different farms can vary substantially. Our results are consistent with others demonstrating that MFC treatment of SW largely eliminates VFAs, which are largely responsible for the SW odor [ 4 ]. Importantly, the SW feed in our experiments was approximately 5-fold higher strength and the HRT is less than five times that utilized by Jung et al. [ 4 ] and yet the MFCs still performed well at removing the VFAs. In summary, use of SS as an anodic inoculum resulted in superior treatment performance of the MFCs over the 67 d course of the experiment compared to IGBS inoculum. This may indicate a more general tendency of preadapted inocula to perform better at degrading the substrate [ 15 ]. 4.2. The Microbiome of Electrogenic of Anodic Biofilms and Planktonic Populations of MFCs We used metagenomic analysis to explore the whole taxonomic diversity of the SS and IGBS inoculums, SW, and anodic and planktonic microbial communities of MFCs. Proteobacteria (mostly Deltaproteobacteria) became the predominant phylum in both MFCs anodic communities, while Firmicutes and Bacteroidetes decreased. The planktonic community of the IGBS-MFCs showed notable variation in relative abundance and became more similar to their anodic communities. In contrast, the planktonic communities of the SS-MFCs were intermediate between the SW and anodic communities. A previous study of a distillery wastewater-treating pilot-scale MFC inoculated with IGBS showed that the dominant anodic phyla (Proteobacteria, Bacteroidetes, and Firmicutes) were similar to that of the IGBS inoculum [ 16 ]. Previous studies have shown that SW could be used as a suitable inoculum for electricity production using MFCs, distinguished by the chamber and cathode types [ 4 , 17 ]. Analysis of the anodic microbial communities in the SS-MFCs mainly showed that dominant species belonged to three major phyla Proteobacteria, Bacteroidetes, and Firmicutes [ 9 , 12 ]. Results of metagenomics analysis in our study are in good agreement with results in the literature [ 9 , 12 ]. Detailed analysis of the dominant anodic bacterial species in SW-treating MFCs showed high diversity in members of the Deltaproteobacteria, Gammaproteobacteria, Firmicutes , Bacteroides, and Archaea. Among all Deltaproteobacteria , Geobacter metallireducens , Pelobacter propionicus , Desulfovibrio vulgaris, Syntrophobacter fumaroxidans, and Syntrophus aciditrophicus were found to be the most abundant in the anodic microbial communities of both MFCs. The well-known electrogenic Geobacter sulfurreducens , dominant in the MFC microbial biofilms, generates a current via membrane c-type cytochromes (omcZ) and secretion of pili encoded by the pilA gene [ 3 , 18 , 19 ]. In contrast, anoditrophilic Fe(III)-reducing Pelobacter carbinolicus was characterized as a nonelectrogenic symbiotic bacterium responsible only for converting of substrates to acetate and hydrogen for use by G. sulfurreducens [ 3 , 20 ]. Cytochrome c localized on the outer cell membrane of Desulfovibrio desulfuricans contributed to the electron transfer in an electricity-generating MFC [ 21 ]. Members of Deltaproteobacteria might contribute to VFA degradation. Our data demonstrate a relative abundance of Syntrophobacter fumaroxidans and Syntrophus aciditrophicus on the MFC anodes, which may aid in metabolism of propionic and butyric acid in the SW. Pure culture experiments with Geobacter species isolated from swine wastes examined the ability to biodegrade individual and mixtures of VFAs [ 22 ]. It was shown that G. metallireducens, G. humireducens, and G. grbiciae consume VFAs and stimulate VFAs oxidation depending on availability of Fe(III). This study demonstrates that Acinetobacter baumannii and Pseudomonas fluorescens belonging to Gammaproteobacteria were prevalent members in the anodic community of both MFCs. The Gammaproteobacteria possess diverse metabolic capabilities involved in a breakdown of different substrates and production of soluble redox-active compounds, resulting in current generation in MFCs [ 3 , 23 , 24 ]. Acinetobacter species dominating in the microbial community of MFCs fed with fermentable substrates were able to produce electricity [ 25 ]. Production of pili -like structures encoded by csuC and csuE genes in A. baumannii influences the colonization of different abiotic surfaces [ 26 ]. We found a considerable number of Pseudomonas species in both MFCs types. The ability of Pseudomonas to consume various carbon sources is known. Moreover, excretion of soluble electrochemically redox mediators participating in the electricity production in MFCs has been observed [ 23 , 27 ]. Thus, dominant Pseudomonas fluorescens might be responsible for COD removal from the SW and the excretion of redox mediators contributes to the observed electricity generation of the MFCs. The relatively high abundance of Shewanella baltica in the anodic microbial communities provides evidence of their importance in the conversion of COD into electricity. Previous studies have showed that electrogenic Shewanella species might transfer electrons to the anodes of MFCs either through nanowires or excretion of redox-active second metabolites [ 28 , 29 ]. Our study demonstrates a relative abundance of bacteria related phyla Firmicutes and Bacteroidetes. It is well known that Clostridium species participate in fermentation processes and conversion of organic substrates to VFAs and hydrogen and that they are indigenous microbiota of the swine gastrointestinal tract and manure [ 12 ]. Bacteroidetes are widely recognized as the intestinal microflora associated with fermentation of carbohydrates and utilization of nitrogenous compounds, as well as odor production [ 30 ]. We found that the remaining dominant bacteria, Flavobacterium johnsoniae and Bacteroides fragilis, became even more abundant in the planktonic populations of both MFCs. 16S rRNA sequence analysis of a SW-treating MFC microbial community showed that two members of Firmicutes, a Gram-positive Turicibacter sp. and Sedimentibacter spp., were the dominant genera on the anodes of a MFC having a maximum power point tracking system [ 9 ]. Earlier studies demonstrated reduction of VFAs level depending on a seasonal shift of Bacteroidetes members in an anaerobic lagoon used for swine waste treatment [ 12 ]. In our study, two Archaea species M. barkeri and M. thermautotrophicus increased in the bacterial communities of SS-MFCs and IGBS-MFCs, respectively. Rotaru et al. established that the acetoclastic methanogen M. barkeri in association with electrogenic bacterium G. metallireducens participates in direct interspecies electron transfer (DIET) [ 31 ]. We found a potential for a DIET-type bacterial association between M. barkeri and G. metallireducens in the anodic microbial community of the SS-MFCs; possible association between M. thermautotrophicus and G. metallireducens was found in the anodic microbial community of IGBS-MFCs. Taken together, the profiling of microbial community diversity based on similarity and phylogeny supports a model for development of electrogenic biofilm in MFCs from their inocula."
} | 3,615 |
36537073 | PMC9948228 | pmc | 753 | {
"abstract": "Abstract Purple phototrophic bacteria are one of the main actors in chemolithotrophic carbon fixation and, therefore, fundamental in the biogeochemical cycle. These microbes are capable of using insoluble electron donors such as ferrous minerals or even carbon‐based electrodes. Carbon fixation through extracellular electron uptake places purple phototrophic bacteria in the field of microbial electrosynthesis as key carbon capturing microorganisms. In this work we demonstrate biomass production dominated by purple phototrophic bacteria with a cathode (−0.6 V vs. Ag/AgCl) as electron donor. In addition, we compared the growth and microbial population structure with ferrous iron as the electron donor. We detect interaction between the cathode and the consortium showing a midpoint potential of 0.05 V (vs. Ag/AgCl). Microbial community analyses revealed different microbial communities depending on the electron donor, indicating different metabolic interactions. Electrochemical measurements together with population analyses point to Rhodopseudomonas genus as the key genus in the extracellular electron uptake. Furthermore, the genera Azospira and Azospirillum could play a role in the photoelectrotrophic consortium.",
"conclusion": "CONCLUSIONS This work explored for the first time the cultivation of a PPB‐dominated consortium under photoelectrotrophic conditions. Using a cathode as a sole electron donor, we have produced biomass dominated by purple phototrophic bacteria without any prior acclimatization period. Electrochemical measurements together with population analyses point to Rhodopseudomonas genus as the key bacterial genus in the extracellular electron uptake. Furthermore, the genera Azospira and Azospirillum could play a role in the photoelectrotrophic consortium. The cultivation of a mixed culture allows to operate under non‐sterile conditions, drastically improving the applicability as a method of bacterial biomass production. Additionally, the biomass composition dominated by purple phototrophic bacteria gives rise to a new exploratory field in the production of protein‐rich biomass and value‐added products.",
"introduction": "INTRODUCTION Microbial anoxygenic photosynthesis is one of the most important processes in the biogeochemical cycle, since it allows light‐driven carbon fixation (Bryce et al., 2018 ; McKinlay & Harwood, 2010 ; Ozaki et al., 2019 ). Chemolithotrophic carbon fixation that occurs in subsurface dictates carbon fluxes in environments such as soils and groundwater (Taubert et al., 2021 ). One of the main actors in this process are purple phototrophic bacteria (PPB), which fix CO 2 using a wide range of electron donors such as H 2 , S 2 O 3 \n 2− or S 2− . Surprisingly, some microorganisms of this type are also capable of accepting electrons from solid phases such as ferrous minerals (Widdel et al., 1993 ) or even graphite electrodes (Xing et al., 2008 ). In this context, the mechanisms behind extracellular electron uptake to accept electrons from a cathode have been revealed (Bose et al., 2014 ). Furthermore, carbon fixation (via the Calvin–Benson cycle) has been recently proved to be strongly linked to phototrophic electron uptake using an electrode (cathode) as electron donor (Guzman et al., 2019 ). Furthermore, other metabolic pathways such as nitrogen fixation or hydrogen production could be also affected (McKinlay & Harwood, 2010 ) under electrode control. Carbon fixation through extracellular electron uptake places purple phototrophic bacteria in the field of microbial electrosynthesis as key carbon capturing microorganisms. Microbial electrosynthesis seeks to use the electrons provided by a biocathode to synthesize value‐added compounds, generally volatile fatty acids and alcohol like acetate, propanol or butyrate (Ganigué et al., 2015 ; Izadi et al., 2021 ; Nevin et al., 2010 ), but also other compounds such as bioplastics (Chen et al., 2018 ; Nishio et al., 2013 ) or microbial biomass as protein source (Xu et al., 2021 ). After a decade of effort and scientific advances (Nevin et al., 2010 ), microbial electrosynthesis is still far from the technical and economic competitiveness of current industrial processes (Prévoteau et al., 2020 ). Microbial electrosynthesis requires engineering improvements such as electrode material (Jourdin et al., 2014 ; Nie et al., 2013 ; Zhang et al., 2012 ) and reactor design (Jourdin et al., 2018 ; Kantzow et al., 2015 ) but also in the selection of the microbial community (Prévoteau et al., 2020 ). The resilience and versatility of the microbial community will determine the range and rate of products obtained as well as the robustness of the synthesis process. Purple phototrophic bacteria could bridge the gap to these limitations. Despite discoveries based on pure cultures (Bose et al., 2014 ), the use of consortium dominated by purple phototrophic bacteria under photoelectroautotrophy still in its infancy. Therefore, understanding the dynamics of a microbial consortium of purple phototrophic bacteria would allow operation under non‐sterile conditions. Despite its limitations such as product purification or possible impediments in the food industry, to explore PPB may reduce costs and facilitates the operation (Hülsen et al., 2014 ), while opening the range of potential products. With the aim of bringing electrode‐mediated biomass production closer to reality, in this work we have explored the photoelectroautotrophic cultivation of a mixed culture dominated by purple phototrophic bacteria. In addition, we have compared electrode‐dependent biomass production with the most ubiquitous inorganic electron donor, ferrous iron.",
"discussion": "RESULTS AND DISCUSSION In contrast with previous experiences with pure culture of purple phototrophic bacteria (Guzman et al., 2019 ), we have explored for the first time the cultivation of a microbial mixed consortium dominated by PPB with a cathode as the sole electron source. Photoferrotrophic growth: Ferrous iron as electron donor Our first approach to understand autotrophic growth extracellular electron uptake led us to cultivate a mixed community dominated by purple phototrophic bacteria with ferrous iron (Fe 2+ ). For this purpose, we used an enriched culture dominated by anodic purple phototrophic bacteria to inoculate batch reactors with ferrous iron a sole electron donor. We observed an increase in optical density (OD 590 ) over time, accompanied by a decrease in Fe 2+ concentration indicating that the consortium was able to fix inorganic carbon into biomass by using Fe 2+ as electron donor (Figure 2A ). Biomass production was also verified at values concentration of 22.08 ± 2.71 g L −1 for total organic carbon (TOC) and 0.1334 ± 0.0171 g L −1 VSS at the end of the experiment. FIGURE 2 Purple phototrophic bacteria dominated consortium under photoferrotrophic conditions. (A) Growth curve. Red circles correspond to optical density (OD 590 ) and white diamonds corresponds to ferrous iron concentration (Fe 2+ ). (B) Absorbance spectrum at time 0 (dashed line) and stationary phase (straight line). The increase in optical density showed a certain delay with respect to the oxidation of ferrous iron (Figure 2A ). This result could indicate that other microorganisms could be playing a role in the process, that is, electroactive acetogens that produce organic compounds that purple phototrophic bacteria can use to grow heterotrophically. We observed two maximum absorption peaks (804 and 870 nm) characteristic peaks of PPB bacteriochlorophylls. In addition to several peaks between the wavelengths 500 and 590 nm, corresponding to carotenoids. These absorption peaks match to those described in purple phototrophic bacteria dominated consortia (Vasiliadou et al., 2018 ). Therefore, the population of purple phototrophic bacteria remains predominant under photoferrotrophic conditions (Figure 2B ). At the end of the experiment, we studied photoferrotrophically cultured consortium using transmission electron microscopy (TEM) and scanning electron microscopy (SEM; Figure 3 ). The micrographs showed microbial cells associated with different mineral particles classically observed in processes of biomineralization of iron (Oggerin et al., 2013 ). The solubility of ferric iron (Fe 3+ ) is considerably less than that of ferrous iron (Fe 2+ ). Therefore, the oxidation of ferrous iron to ferric iron (Fe 3+ ) led to the formation of insoluble ferric iron (Fe 2+ ) minerals (Figure 3B,D,E ). Notably, the images revealed lamellar type internal membrane structures arranged parallel to the cytoplasmic membrane, structures typical of purple phototrophic bacteria (Figure 3A,C ; LaSarre et al., 2018 ; Ramana et al., 2010 ). In addition, despite not having analysed in depth the presence of bioplastics produced, we observed granules in the cytoplasm very similar to the polyhydroxybutyrate (PHB) and polyhydroxyalkanoate (PHA) granules described by other authors (Higuchi‐Takeuchi et al., 2016 ; Figure 3C ). FIGURE 3 Purple phototrophic bacteria dominated consortium under photoferrotrophic conditions. (A) TEM micrograph with type internal membrane structures. (B) TEM micrograph with insoluble iron oxides (Fe 3+ ). (C) TEM micrograph with cell inclusions. (D and E) SEM micrograph with insoluble iron oxides (Fe 3+ ) and bacteria. Furthermore, SEM micrographs showed angular crystal structures of iron oxides with bacteria forming rosette‐like clusters previously described by other researchers in the model electroactive purple phototrophic bacteria, Rhodopseudomonas palustris (Hougardy et al., 2000 ; Figure 3D,E ). Photoelectrotrophic growth: Cathode as electron donor After exploring the photoferrotrophic growth of PPB we explored the capacity of such culture to use an electrode (cathode) as the sole electron donor. Our goal was to study both biomass production and bioelectrochemical response. Under these conditions, we carried out two independent assays using: (i) PPB‐planktonic cells, and (ii) PPB‐biofilm cells. As previously described, cathodic polarization hinders irreversible cell adhesion, due to the electrostatic repulsion between the electrode and the cell wall charges (Bayer & Sloyer, 1990 ; Busalmen & de Sánchez, 2001 ). This phenomenon, despite being a problem to study cathodic biofilms, allowed us to accurately quantify biomass production. To assess biomass production, we inoculated a two‐chamber reactor with our bacterial consortium (Figure S1 ). In this reactor, planktonic growth outcompeted biofilm‐based growth by using a carbon rod as working electrode (−0.6 V vs. Ag/AgCl). We monitored electron uptake (current density) and planktonic microbial growth (OD 590 ) (Figure 4A ). We observed an increase in biomass growth (OD 590 ) 24 h after inoculation; actually, absorption peaks coincided with those typical of PPB (Figure S2 ; Vasiliadou et al., 2018 ). This microbial growth was accompanied by electron uptake, visible by the increase in current intensity from values close to 0 μA to almost reaching −100 μA. Even at the end of the experiment, we did not observe biomass attached to the cathode but planktonic growth (Figure S1 ). Some species of purple phototrophic bacteria have been previously shown to uptake electrons from cathodes through redox mediators (Borghese et al., 2020 ; Hasan et al., 2013 ). In order to clarify the electron transfer mechanism, we studied the reactor supernatant by cyclic voltammetry and no redox species were detected (Figure S3 ), which suggests that direct extracellular electron transfer was the main electron uptake mechanism. We hypothesize that extracellular electron transfer could occur through random contacts between bacterial cells and the polarized electrode, in a similar way than previously reported bacteria interact with fluid electrodes (Tejedor‐Sanz et al., 2017 ). Indeed, the electrostatic repulsion caused by cathodic polarization could promote planktonic interaction over biofilm interaction. FIGURE 4 Purple phototrophic bacteria dominated consortium under photoelectrotrophic conditions. (A) Growth curve. Up: Optical density (OD 590 ). Down: Current intensity. (B) TEM micrograph with type internal membrane structures. At the end of the experiment, we took TEM micrographs of the planktonic biomass to study the cell structure (Figure 4C ). The bacteria had markedly different cell structures than those observed in the previous experiment (Figure 3A–C ). We identified smaller PHB inclusions (Figure 4C ) compared to the iron‐oxidizing consortium (Figure 3A–C ). These results unequivocally indicate that the PPB‐dominated consortium was able to grow with a cathode as sole electron donor. Unfortunately, planktonic cells did not showed electroactivity after cyclic voltammetry analysis (Figure S4 ). An alternative strategy, previously explored (Manchon et al., 2023 ), to acquire electrochemical data from PPB consist of using a biofilm‐based approach. Non‐polarized autotrophic conditions allowed biofilm formation on the carbon felt surface. After ca. 200 h of inoculation, a fragment of the inoculated felt was analysed by TEM, revealing the formation of the biofilm on the surface, with clustered cells and exopolysaccharide (Figure 5A,B ). Some of these cells formed rosette‐like clusters like those described in Figure 3D,E , characteristic of Rhodopseudomonas palustris , the model electroactive purple phototrophic bacteria (Hougardy et al., 2000 ). We observed visible reddish filamentous structures on the surface of the carbon felt (Figure 5C ). FIGURE 5 Purple phototrophic bacteria dominated consortium biofilm under photoferrotrophic conditions. (A, B) SEM micrograph. (C) Carbon felt photograph after inoculation. (D) Cyclic voltammograms (5 mV s −1 ). Grey voltammogram corresponds to abiotic conditions carbon felt. Black voltammogram corresponds to inoculated carbon felt. Vertical line corresponds to thermodynamic potential of hydrogen evolution (−0.63 V vs. Ag/AgCl). Mid‐point potential (E mp ) is indicated by a black point. Another fragment of the colonized carbon felt was connected through a gold wire in a single‐chamber electrochemical cell to assess the electroactivity of the consortium (Figure S5 ). After 10 h of polarization (−0.6 V vs. Ag/AgCl) cyclic voltammetry analysis revealed a redox couple (E mp = 0.05 V vs. Ag/AgCl) and a reductive process below −0.6 V (vs. Ag/AgCl; Figure 5D ). The midpoint redox potential was similar to the redox potential of pure cultures of Rhodosepudomonas palustris TIE‐1 (Bose et al., 2014 ). The identified redox pair indicates the presence of redox active components that could participate in the extracellular electron uptake. Electrochemical analyses of the supernatant, both in the planktonic‐based approach and in the biofilm‐based one, did not reveal the presence of redox mediators. Therefore, the electrochemical characterization of the consortium suggests direct extracellular electron transfer in which the redox site (E mp = 0.05 V) could be involved. In addition, the similarity of our signal/fingerprint to previous signals reported in Rhodopseudomonas sp suggest the presence of this bacterial genus in our PPB consortium. The reduction process observed at redox potential lower than −0.6 V versus Ag/AgCl (Figure 5D , black line) was not observed in the abiotic voltammogram (Figure 5D , grey line), indicating that it was catalysed by the microbial consortium. Based on our knowledge of purple phototrophic bacteria, the reduction process could correspond to hydrogen bioproduction, as we have previously reported (Vasiliadou et al., 2018 ). These results pointed to the biohydrogen production potential by PPB‐dominated cathodes. To clarify, during all experiments, at the potential used (−0.6 V vs. Ag/AgCl) hydrogen production is not thermodynamically possible (−0.63 V vs. Ag/AgCl); thus, the H 2 ‐mediated pathway was ruled out. Iron and cathodes show similar but different pressure on PPB consortium The microbial community analysis was performed using 16S Illumina in order to understand how different microbial communities were established under autotrophic conditions and to evaluate the impact of the insoluble electron donor (i) ferrous iron; and (ii) carbon electrode. The predominant genera in common between the two different electron donors are Rhodopseudomonas , Acinetobacter , Pseudomonas and Acidovorax which are considered core genera in electroactive communities (Figure 6 ; Xiao et al., 2015 ). Rhodopseudomonas , the most abundant genus in all the samples, is considered the electroactive PPB model (Bose et al., 2014 ; Guzman et al., 2019 ). This genus is capable of fixing carbon dioxide into biomass using an extracellular electron donor, for example, ferrous iron (Widdel et al., 1993 ) or cathodes (Bose et al., 2014 ; Guzman et al., 2019 ). Acinetobacter , Pseudomonas and Acidovorax have been previously found in cathodes (Liu et al., 2014 ; Rabaey et al., 2008 ; Rowe et al., 2015 ) but their electroactivity has not been unequivocally demonstrated. The usual presence of these genera in electroactive consortia indicates a fundamental role. However, the lack of reports demonstrating its electroactivity suggests that Rhodopseudomonas is the key genus that performs extracellular electron uptake in our systems, connecting the rest of the community with the extracellular electron donor. Additionally, we observed a strong increase in abundance of Ralstonia genus in both ferrotrophy (Fe 2+ ; 2.7‐fold) and electrotrophy (cathode; 3.3‐fold) compared to the original community used as inoculum. Ralstonia was reported in the autotrophic electrode‐assisted production of valuable compounds such as PHB (Chen et al., 2018 ; Nishio et al., 2013 ) or alcohols (Li et al., 2012 ). Despite its low abundance in this study, our results indicate that Ralstonia could become important in a long‐term photoelectrotrophic system. FIGURE 6 Microbial community analysis and schematic of the genus that might coexist depending on the electron donor. Stacked bars correspond to relative abundance at genus level of inoculum, ferrous iron as electron donor (Fe 2+ ) and cathode as electron donor (cathode). Left to the bars (Fe 2+ and cathode) the main genus and their possible interaction for each electron donor are represented. The structure of the microbial community also revealed differences in terms of the dissimilar abundance and presence of species depending on the extracellular electron donor. First, consortium with Fe 2+ as electron donor showed lower level diversity (Shannon index: 1.84) respect to the inoculum (Shannon index: 2.266). This result was consistent with the proportion of Rhodopseudomonas and Acinetobacter genus species that represent more than 75% of the total genera identified in the samples with ferrous iron as electron donor (Figure 6 ). Both the inoculum and the sample with Fe 2+ showed a notable presence of the genus Proteiniphilum , described as electroactive bacteria (Logan et al., 2019 ); however, it was not found in cultures with the cathode as electron donor. In the samples with cathode as electron donor we observed a similar diversity (Shannon index: 2.24) compared with the inoculum. However, the analyses revealed a decrease in the relative abundance of Rhodopseudomonas in addition to a drastic decrease in the abundance of Proteiniphilum (Figure 6 ). We discovered that cathode promoted the presence of two bacterial genera: Azospirillum and Azospira , both considered core genera in electroactive microbial communities (Xiao et al., 2015 ). Azospirillum is a nitrogen‐fixing bacterium that has rarely been found in electroactive communities (Pisciotta et al., 2012 ). Azospirillum sp. together with other genera such as Rhodopseudomonas and Ralstonia , also present in the cathode, have been described as one of the main carbon sequestrants in soils (Yuan et al., 2012 ). Furthermore, some researchers have reported that Azospirillum strains isolated from electrodes are capable of extracellular respiration of anthraquione‐2,6‐disulfonate (AQDS) (Zhou et al., 2013 ), a redox mediator classically respirable by electroactive bacteria (Dantas et al., 2018 ; Lovley et al., 1999 ). Azospira , another genus primarily described as nitrogen‐fixing bacteria, has also been detected in electroactive communities (Sun et al., 2011 ). Furthermore, some authors have described Azospira sp . electron uptake from cathodes using AQDS as a redox mediator (Thrash et al., 2007 ). Therefore, the decrease in the abundance of Rhodopseudomonas together with the appearance of Azospirillum and Azospira points to the possibility of a syntrophy under photoelectroautotrophic conditions. The PPB‐dominated consortium used as inoculum was previously adapted to perform anodic respiration (Manchon et al., 2023 ), so the electrode was used as an electron acceptor. In contrast, such microbial consortium successfully adapted to conduct the inverse reaction, using the electrode (now as cathode) as extracellular electron donor. So, our PPB‐dominated consortium appeared to have bidirectional electron transfer, being able to both accept and donate electrons from an insoluble material. Some authors have already reported that PPB could carry out transfer in both directions (Bose et al., 2014 ; Xing et al., 2008 ). Furthermore, we have previously reported the same behaviour using a PPB‐dominated consortia growing on fluid‐like electrodes under heterotrophic conditions (Manchon et al., 2023 ). In summary, our results confirm the Rhodopseudomonas genus played the main role in the extracellular electron uptake using Fe 2+ and with cathode as electron donor. The connection between the cathode and PPB‐dominated microbial community still remains unclear. The absence of redox mediators (Figure S3 ) and organic compounds (Table S1 ) suggests a direct extracellular electron uptake, as previously reported in pure cultures of purple phototrophic bacteria (Bose et al., 2014 ). Further studies have shown interspecies electron transfer for methanogenesis (Huang et al., 2022 ) and carbon fixation (Liu et al., 2021 ) between purple phototrophic bacteria and other genera. Thus, we hypothesize that Rhodopseudomonas strains in our reactors are the link between the electrode and the microbial community."
} | 5,627 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.