The metabolic model of G. sulfurreducens PCA

The in silico model of G. sulfurreducens PCA, a model reconstructed in 2005 and updated in 2008 using the SimPheny (Genomatica, San Diego, CA) platform was applied in the following simulations [17,18]. The compartmentalized model contained 747genes, 698 metabolites, and 649 reactions. All reactions occurred in either the cytoplasm or the extracellular space. In the 649 reactions, 530 metabolic reactions were obtained based on sequence similarity, available biochemical databases (eg KEGG, BioCyc, etc.) and genetic evidences. In terms of available substrates and G. sulfurreducens secretory products, 64 reactions corresponding to transport between the cytoplasm and extracellular space were added. The rest were exchange reactions, which allowed extracellular metabolites to enter into or end products to be secreted from the cell.

The metabolic reconstruction used electron donor acetate as carbon source and energy source simultaneously, with soluble fumarate or insoluble Fe³⁺ as the electron acceptor. Biomass reaction agg_GS13m_2, which represents growth-associated biosynthetic demands, contains biomass components and essential growth metabolites, and the flux through it is equal to the strain growth rate v_gr. G. sulfurreducens could completely oxidize acetate to carbon dioxide under anaerobic conditions, together with electron transfer outside the cell to reduce Fe³⁺. The flux through exchange reaction EX_fe3(e) is equal to the rate of Fe³⁺ reduction, also named as respiration rate v_res, and its value indicates the speed of extracellular electron transfer. Both the anode of MFC systems and Fe³⁺ are insoluble electron acceptors that can be reduced by G. sulfurreducens with a similar mechanism, hence, Fe³⁺ was taken as the electron acceptor to simulate cellular metabolism in MFCs [17].

Flux Balance Analysis and Flux Variability Analysis

Flux Balance Analysis (FBA) is a linear programming approach, an approach assuming the metabolic network at steady state to ensure the rate at which every metabolite is consumed is equal to that of being produced. FBA allows predictive computation of the cellular flux distribution that maximizes objective function with a particular meaning under certain cultivation conditions [19]. The form of an FBA problem concerning a metabolic model with m metabolites and n reactions is given below:

(1)

Parameters description: S_i,j is the stoichiometric coefficient of metabolite i in reactionj, v_j  is the flux through reactionj, lb_j is the minimum allowable flux through reactionj, ub_j is the maximum allowable flux through reactionj, and $v_{}^{uptake}$ is the flux through exchange reaction regarding carbon source and energy source.

Flux Variability Analysis (FVA) [20] has the same principles as FBA except for the growth rate constraint, which can be used to study the entire range of theoretically achievable respiration rate. Here we register the subscript of v_gr asm−1, and the following two formulations was applied to examine alternate respiration states v_res：

(2)

(3)

Simulation of current-generating metabolism

G. sulfurreducens can use acetate as carbon source and energy source. When simulating optimal growth using Flux Balance Analysis (FBA) [19], acetate uptake rate v^uptake was set as a fixed value, and the rate of biomass reaction agg_GS13m_2, v_gr, was defined as objective function with the need to be maximized.

If microbes grow suboptimally, the growth rate is less than that of optimal growth. The equation v_gr =k﹒v_gr^max,k∈(0,1)   could be employed to describe the suboptimal growth state, where v_gr^max is the optimal value for growth. Flux Variability Analysis (FVA) [20] was carried out to identify the maximum and minimum respiratory rate under one same suboptimal growth state. All above problems were solved using COBRA (Constraints Based Reconstruction and Analysis) toolbox accessed via MATLAB modeling environment [21].

Competitive relationship between oxidative respiration and cell growth

In order to examine the impact of growth state on the current production capacity, we simulated optimal and suboptimal growth by constraining acetate uptake rate at 13.630mmol/g.dw/h. We confined our attention to central metabolism (see Figure 1), a coupled characterization of metabolic pathways from extending previous computational and experimental analysis of the metabolic model of G. sulfurreducens [15,22]. Simulation results of optimal and suboptimal growth were placed in Figure 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Central metabolism of G. sulfurreducens and simulation results of optimal and suboptimal growth.

Red values for optimal and blue values for suboptimal growth.

https://doi.org/10.1371/journal.pone.0073907.g001

From the conlums of v_gr and v_res of Table 1, with the increasing of v_gr, v_res would decreased. The result indicated that there may be a competitive relationship between oxidative respiration and fuel assimilation into biomass. So microbes growing suboptimally could make good use of fuel for current generation in MFC systems.

After being transported into the cell, acetate was activated to acetyl-CoA via acetate kinase followed by acetyl-CoA transferase [15]. Over 90% of acetyl-CoA was then directed to the TCA cycle for carbon dioxide, NADH, NADPH and reduced ferredoxin generation. The rest was used for fatty acid metabolism, amino acid metabolism and pyruvate synthesis. Pyruvate had a dual role in phosphoenolpyruvate (PEP) synthesis for gluconeogenesis and anapleurotic reaction, a reaction used for converting pyruvate to oxaloacetate via pyruvate carboxylase. Anapleurotic reaction was the process of replenishing the TCA cycle intermediate oxaloacetate - a small part of which was provided to synthesize PEP and the biomass precursor aspartate, which ensured that the TCA cycle ran smoothly and continuously [22].

ATP production of G. sulfurreducens was completely dependent on electrogenic electron transport. Electrons carried in reducing equivalents were transferred into the inner membrane via NADH or NADPH dehydrogenase, and protons were pumped out of the cytoplasm for ATP synthesis via ATP synthase. Although it was well known that c-type cytochromes were important for extracellular electron transfer, a specific electron transfer chain for electron transfer out of the inner membrane to Fe³⁺ or electrodes had not been determined for G. sulfurreducens [23]. In the model of G. sulfurreducens, a simplified extracellular electron transfer process was developed by NADH or NADPH dehydrogenase, cytochrome-c reductase and then Fe³⁺ reductase.

Simulation results presented here indicated that when microbes grew suboptimally, more fuel was directed to the TCA cycle to generate electron and energy, resulting in less being used for biomass synthesis compared with that optimal growth. Therefore microbes growing suboptimally may make good use of fuel for current generation in MFC systems.

Effect of alternative flux distributions on the respiration of G. sulfurreducens

Computational results presented in the above section indicated that suboptimal growth was good for the improvement of respiration rate. However, alternative respiratory states may exist under the same growth state as a result of the inherent redundancies built into metabolic networks [20]. FVA was used to characterize the entire range of respiratory states (Table 1).

The genome-scale flux distributions were examined to illustrate the respiratory rate variance between the maximum and minimum respiration corresponding to all simulated growth states (from 0% to 90%). Excluding the reactions whose flux variances were less than 1% of their original flux values, those reactions whose flux distributions existed remarkable variance (larger than 1% of their original flux values) included five metabolic subsystems: central metabolism, transport, energy metabolism, amino acid metabolism, and exchange. Because of the same reason leading to the respiratory rate variance, we only paid attention to the suboptimal growth rate of 0.006h^-1, a value that was 10% of that of optimal growth, and we thought this may be the best circumstance for electron extraction from the substrate. Figure 2 has illustrated alternative flux distributions through central metabolism of G. sulfurreducens, while Table S1 in the supplemental materials provided details on all the reactions whose fluxes exist variance (The supplementary information of reference [18] have provided a look-up table for the whole names of all the reactions and metabolites).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Alternative flux distributions through central metabolism of G. sulfurreducens during in silico growth at suboptimal growth of 0.006h^-1.

Predicted flux distributions through central metabolism in G. sulfurreducens during in silico suboptimal growth. Red values indicated the flux distribution corresponding to in silico minimum respiration at the rate of 62.177mmol/g.dw/h, and blue values corresponded to in silico maximum respiration at the rate of 107.991 mmol/g.dw/h. Alternative respiration was simulated by constraining the growth rate at 0.006h^-1and maximizing/minimizing the respiration rate.

https://doi.org/10.1371/journal.pone.0073907.g002

Interconversion from acetyl-CoA to pyruvate, the first step of assimilation in G. sulfurreducens, had three pathways (pyruvate dehydrogenase, formate c-acetyltransferase, pyruvate synthase). With regard to maximum respiration, only two pathways were activated and no formate was produced. In contrast, regarding minimum respiration, three pathways were all activated and formate was produced at a rate of 22.907mmol/g.dw/h. Formate was chosen to be secreted as an end product outside the cell totally via proton symport instead of oxidation via formate dehydrogenase, resulting in the loss of electrons. At the same time, less acetate (13.544 vs 14.309 mmol/g.dw/h) was directed toward the TCA cycle, from which the majority of electrons were derived for G. sulfurreducens [24]. The conclusion could be arrived that under minimum respiration, less acetate towards TCA cycle and the total secretion of formate caused electron loss and subsequent large respiration rate difference from maximum respiration.

Cell need for energy was crucial to metabolism regulation. Increased cellular energy consumption inspired reducing equivalent-producing reactions to increase their rates to meet more energy demand [25]. We extracted reducing equivalent-producing reactions in central metabolism or amino acid metabolism in Table S2 for Table S1 in the supplemental materials. Of all 8 reducing equivalent-producing reactions, 5 reactions were involved in the TCA cycle and the redundant pathway for conversion of pyruvate to acetyl-CoA. The eight reactions had higher metabolic fluxes under maximum respiration. Especially, conversion of pyruvate to acetyl-CoA had the greatest difference (999.180 vs 977.038 mmol/g.dw/h.). Subsequently fluxes through the simplified extracellular electron transfer chain (NADH or NADPH dehydrogenase, cytochrome-c reductase and Fe³⁺ reductase) and ATP synthetic reaction (6.333 vs 0.990 mmol/g.dw/h) also increased. Unexpectedly the flux through NADPH dehydrogenase reaction was predicted to decrease (9.619 vs 13.438 mmol/g.dw/h). Conversion of pyruvate to acetyl-CoA, one of the redundancies in central metabolism of G. sulfurreducens [22], contributed largely to the cell for reducing equivalent production discrepancy and consequent respiration discrepancy. The result also suggested that pyruvate dehydrogenase cannot be substituted by formate c-acetyltransferase or pyruvate synthase and different enzymes presented for different tasks [17].

The discrepancy between maximum and minimum respiration increased along with the decrease of growth rate (Table 1). We have investigated the suboptimal growth rate of 0.054h^-1, a value that was 90% of that of optimal growth. The flux difference through pyruvate dehydrogenase reaction was 2.389 mmol/g.dw/h. (999.406 vs 997.017 mmol/g.dw/h) between maximum and minimum respiration. The value increased to 22.142 mmol/g.dw/h. (999.180 vs 977.038 mmol/g.dw/h) at a growth rate of 0.006 h^-1. It was clear that the discrepancy between maximum and minimum respiration depended on the flux difference through reducing equivalent-producing reactions.

Reconstruction of a relatively ideal model of variant of G. sulfurreducens

In the above section, an ideal current-generating metabolism state was given. The ideal state consisted of simultaneous suboptimal growth and maximum respiration. Under this ideal condition, flux towards extracellular electron transfer rather than for biosynthesis is beneficial for the conversion of organic matter to electricity without large accumulations of biomass that might otherwise plug specific parts of an aquifer. And at the same time, electricigens may maximize utilization of limited fuel.

The in silico metabolic model predicted that when G. sulfurreducens grew suboptimally at a rate of 0.006h^-1, minimum respiratory rate was 62.177mmol/g.dw/h and caused by formate secretion. We deleted formate enzyme GSU0234 by constraining the flux through the reaction catalyzed by it at 0 mmol/g.dw/h, and after FVA, minimum respiratory rate increased to 75.267 mmol/g.dw/h, which was caused by pyruvate secretion. Then enzymes for pyruvate, L-valine, L-leucine, citrate, L-isoleucine, L-proline, L-lysine, L-cysteine, h2 and n2 were determined through iterative FVA and were deleted one by one. A relatively ideal model of variant for G. sulfurreducens strain was reconstructed finally by modifying the wild metabolic model in this way (Table 2). The modified model predicted that under suboptimal growth of 0.006h^-1 condition, minimum respiratory rate increased to 107.991 mmol/g.dw/h, a value that was equal to that of maximum respiration. Of all the simulated variants, formate-deficient variant has the biggest discrepancy between minimum and maximum respiration. The relatively ideal variant model was also validated with different suboptimal growth.

We examined the full range of flux through every reaction in the ideal modified model corresponding to the suboptimal growth of 0.006h^-1 with acetate uptake rate in 13.630 mmol/g.dw/h (Table S3 has provided the detail). Of all 705 reactions, 412 reactions were non-functional, that is, deletion of corresponding reactions had no impact on current-generating metabolism of the ideal variant of G. sulfurreducens. These reactions were involved in almost every subsystem. 101 reactions occurred at a constant rate.

As shown in Figure 3, the variant had two pathways for conversion of a-ketoglutarate to succinyl-CoA: a-ketoglutarate dehydrogenase and a-ketoglutarate synthase. The latter could produce reduced ferredoxin. All fluxes through the other reactions in TCA cycle were variable [22]. Both malate and oxaloacetate could be produced at a predicted rate of 1000 mmol/g.dw/h. Conversion of succinyl-CoA to succinate via acetate-CoA transferase may not be the sole pathway, since the model predicted that maximum flux through the reaction catalyzed by succinyl-CoA synthetase for conversion of succinyl-CoA to succinate was 986.462 mmol/g.dw/h instead of 0 mmol/g.dw/h. Pyruvate was able to be produced via malic enzyme (NAD) or malic enzyme (NADP) at a rate of 5.727 mmol/g.dw/h.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Flux distributions through central metabolism of a relatively ideal variant model of G. sulfurreducens during in silico growth at suboptimal growth of 0.006h^-1.

Predicted flux distributions through central metabolism in the ideal variant of G. sulfurreducens during in silico growth. Red values indicated the flux distribution corresponding to in silico minimum energy synthesis at the rate of 0.990 mmol/g.dw/h, and blue values corresponded to in silico maximum energy synthesis at the rate of 8.640 mmol/g.dw/h. Alternative energy synthesis was simulated by constraining the growth rate at 0.006 h^-1 and maximizing/minimizing the energy synthesis rate.

https://doi.org/10.1371/journal.pone.0073907.g003

Examination of the energy metabolism subsystem in the modified model found that owing to deletion of formate transport enzyme, formate produced via formate c-acetyltransferase was only consumed via formate dehydrogenase at a maximum rate of 40.455 mmol/g.dw/h. Then menaquinol reduced via formate dehydrogenase or NADH or NADPH dehydrogenase was oxidized via cytochrome-c reductase at a constant rate of 53.996 mmol/g.dw/h and subsequent Fe³⁺ reduced at a constant rate of 107.991 mmol/g.dw/h. The first step of the simplified extracellular electron transfer chain of the ideal variant was expanded to involve NADH dehydrogenase, NADPH dehydrogenase and formate dehydrogenase.

The flux through ATP synthetic reaction was variable. Maximum and minimum fluxes were 8.640 and 0.990 mmol/g.dw/h respectively. We further investigated the full flux range through ATP synthetic reaction corresponding to different growth (Table 3). Increasing cellular need for energy was neither the sufficient condition nor the necessary condition for respiration improvement [25]. When the variant grew at a rate of 0.018h^-1 and respired at a rate of 105.894 mmol/g.dw/h, energy could be produced at a minimum rate of 2.070 mmol/g.dw/h and its rate may increase owing to the increasing rate of futile cycle for consumption of energy. When the variant grew at a rate of 0.006 h^-1 and respired at a rate of 107.991 mmol/g.dw/h, energy could be produced at a minimum rate of 0.990 mmol/g.dw/h with no futile cycle for energy consumption.

We have searched all the literatures concerning G. sulfurreducens with paying special attention to the strain modification, but did not find some modification strategies which were similar with what we have put forward (Table 2), so we think our strategy for G. sulfurreducens modification is new. In the next stage, we will do experiments of strain modification for the final variant of G. sulfurreducens, i.e. deleting “FORt2 + PYRt2 + VALt6 + LEUabc + CITt6 + ILEabc + PROt5 + LYSt3 + CYSabc + N2t + H2td”, to verify our in silico ideal mutant for G. sulfurreducens strain.