Enhancing CO₂-Valorization Using Clostridium autoethanogenum for Sustainable Fuel and Chemicals Production

Introduction

Gas fermentation has attractive waste carbon valorization properties, for which the need is intensifying (International Panel on Climate Change [IPCC], 2014; Emerson and Stephanopoulos, 2019). Recently, LanzaTech commercialized the first waste gas-to-ethanol process, efficiently incorporating the carbon from steel mill off-gas into fuel quality ethanol via the model acetogen Clostridium autoethanogenum. The key carbon source—carbon monoxide (CO)—accounts for a significant portion of steel mill off-gas and synthesis gas (syngas), which can be generated from multiple high-volume, non-gaseous waste feedstocks (e.g., biomass, municipal solid waste) (Liew et al., 2016). Therefore, LanzaTech's process is significant in that it valorizes waste carbon by fusing two one-carbon gas molecules (C1) into liquid fuel. Furthermore, Handler et al. (2016) found that ethanol produced by LanzaTech's process reduced greenhouse gas emissions by 67 to 98% when compared to petroleum gasoline on an energy content and “cradle-to-grave” basis (feedstock dependent). Carbon dioxide (CO₂) represents a more diverse and plentiful waste stream compared to CO (International Panel on Climate Change [IPCC], 2014), thus embodying a feedstock with greater climate change mitigation and carbon recycling potential.

Increasing acetogenic carbon capture as CO₂ would build on the success of commercial gas fermentation and continue the expansion of the technology as a platform for sustainable chemical production (Redl et al., 2017; Bengelsdorf et al., 2018; Müller, 2019). Compared to other CO₂ valorization methods, acetogens are ideal candidates due to their high metabolic efficiency, ability to handle variable gas compositions, high product specificity, scalability, and low susceptibility to poisoning by sulfur, chlorine, and tars (Liew et al., 2016; Artz et al., 2018). However, metabolism of CO₂ requires an energy source, for which some see an appropriate solution is lacking (Emerson and Stephanopoulos, 2019).

Gas fermenting acetogens harbor the Wood-Ljungdahl pathway (WLP) (Drake et al., 2008), a non-photosynthetic C1-fixation metabolic pathway with the highest-known theoretical thermodynamic efficiency (Fast and Papoutsakis, 2012; Schuchmann and Müller, 2014; Müller, 2019). Various potential energy sources exist for metabolizing CO₂, primarily hydrogen, nitrates, sugars, and arginine. Yet, acetogenic CO₂ valorization, which is actively being developed for industrial implementation (Tizard and Sechrist, 2015), poses challenges along with promise. These include potential adenosine triphosphate (ATP) starvation in autotrophic conditions and carbon catabolite repression in hetero/mixotrophic conditions (Emerson and Stephanopoulos, 2019).

Hydrogen (H₂) is the most recognized energy source for CO₂ utilization—as metabolism of sugars or nitrates cause shifts in metabolism that result in lower CO₂ or H₂ utilization (Liew et al., 2016; Emerson and Stephanopoulos, 2019). H₂ production will also logically transition to renewable sources in the future, whereas production of sugars and nitrates are dependent on less-sustainable methods. Furthermore, levelized cost predictions for solar H₂ indicate a 30% reduction by 2030, potentially becoming competitive with the current levelized cost of fossil fuel derived H₂ by 2035 (Detz et al., 2018; Glenk and Reichelstein, 2019). This is in part due to rapidly decreasing solar electricity costs (IRENA, 2017) and projections of H₂ electrolysis technology development (Detz et al., 2018; Glenk and Reichelstein, 2019). Similarly, atmospheric CO₂ capture via direct air contact showed promising feasibility recently (Keith et al., 2018), which represents an essential development for carbon recycling (Otto et al., 2015). Various power-to-gas technologies are being discussed for mediating fluctuations in renewable power generation (Götz et al., 2016). By extension, gas fermentation to liquid products could couple mediation of renewable power fluctuations to carbon recycling (Redl et al., 2017). This provides an attractive new opportunity for bacterial artificial-photosynthesis, whereby renewable H₂ supplementation facilitates acetogenic CO₂ valorization (Claassens et al., 2016; Haas et al., 2018).

Continuous culture bioprocesses are preferable to batch or fed-batch fermentation bioprocesses (Hoskisson and Hobbs, 2005). Furthermore, systems-level quantification is essential for design-build-test-learn bioprocess optimization by metabolic engineering (Valgepea et al., 2017). Therefore, obtaining quantitative datasets from steady-state chemostat cultures, whose analyses are comparable between experiments, is important for development of these systems (Adamberg et al., 2015). Whilst Bengelsdorf et al. (2018) reviewed autotrophic acetogen growth on CO₂ and H₂ (CO₂/H₂), and Mock et al. (2015) provided notable insight into the CO₂/H₂ metabolism of C. autoethanogenum, the literature lacks a steady-state dataset where carbon flows in a CO₂/H₂ fermentation are quantified. Here we aimed to quantify steady-state CO₂/H₂ fermentation using fully instrumented chemostats and the model acetogen C. autoethanogenum. Subsequently, we showed that CO₂ is a promising feedstock alternative to CO, as more than half of the substrate CO₂ carbon was converted into ethanol. Furthermore, supplementation with CO at low concentrations improved fermentation performance significantly.

Materials and Methods

Bacterial Strain, Growth Medium, and Continuous Culture Conditions

A derivate of Clostridium autoethanogenum DSM 10061 strain—DSM 19630—deposited in the German Collection of Microorganisms and Cell Cultures (DSMZ) was used in all experiments and stored as glycerol stocks at −80°C. This non-commercial strain was grown on CO₂/H₂ (~23% CO₂, ~67% H₂ and ~10% Ar; BOC Australia) and CO/CO₂/H₂ (~2% CO, ~23% CO₂, ~65% H₂, and ~10% Ar; BOC Australia) in chemically defined medium (Valgepea et al., 2017). Cells were grown under strictly anaerobic conditions at 37°C and at a pH of 5 (maintained by 5 M NH₄OH). Chemostat continuous culture achieved steady-states at dilution rates (D) = 0.47 ± 0.01 (CO₂/H₂; specific growth rate (μ) = 0.0196 ± 0.0004 [average ± standard deviation]), 0.5 ± 0.01, and 1 ± 0.01 day⁻¹ (CO/CO₂/H₂; μ = 0.021 ± 0.0004, and 0.042 ± 0.0008 h⁻¹ respectively). See Table 1 for steady-state gas-liquid mass transfer rate data. The steady-state results reported here were collected after optical density (OD), gas uptake and production rates had been stable in chemostat mode for at least three working volumes. See Valgepea et al. (2017) for details on equipment.

Table 1. Summary of low-biomass Clostridium autoethanogenum fermentations.

Experimental Analysis and Quantification

Biomass concentration (gDCW/L) was estimated, and extracellular metabolome analysis carried out as specified in Valgepea et al. (2018).

Bioreactor off-gas was analyzed by an online Hiden HPR-20-QIC mass spectrometer. The Faraday Cup detector monitored the intensities of H₂, CO, ethanol, H₂S, Ar, and CO₂ at 2, 14, 31, 34, 40, and 44 atomic mass units (amu), respectively, in the bioreactor off-gas. These masses were chosen so that each target compound would be represented by a unique signal. This was determined to be essential to achieve the highest confidence in quantification using preliminary experiments, as interferences from other compounds at a shared mass could not be reliably accounted for (e.g., the more intense signal from CO at 28 amu could not be used due to the uncertainty of interference at 28 amu from CO₂ fragmentation). Gas from the cylinder was used as the calibration gas for each MS-cycle (i.e., “online calibration”) to achieve reliable off-gas analysis (Valgepea et al., 2017).

Gas uptake (CO, CO₂, and H₂) and production (ethanol) were determined using “online calibration” of the MS by analyzing the respective feed gas directly from the cylinder after each analysis cycle of the bioreactors. Specific rates (mmol/gDCW/h) were calculated by taking into account the exact composition of the respective gas, bioreactor liquid working volume, feed gas flow rate, off-gas flow rate (based on the fractional difference of the inert gas [Ar] in the feed and off-gas composition), the ideal gas molar volume, and the steady-state biomass concentration.

The carbon balances were determined at 116 ± 11%, 103 ± 12%, and 108 ± 11% for CO₂/H₂, and CO/CO₂/H₂ at D = 0.5 and 1 day⁻¹ respectively (total C-mol products/total C-mol substrates), as specified in Valgepea et al. (2017).

Genome-Scale Metabolic Modeling With GEM iCLAU786

Model simulations were performed using genome scale model (GEM) iCLAU786 of C. autoethanogenum and flux balance analysis (FBA) (Orth and Palsson, 2011) as specified in Valgepea et al. (2018). Briefly, we used FBA to estimate intracellular fluxes (SIM1–26) by incorporating experimentally measured constraints (specific growth rate, and specific acetate, ethanol, cysteine, CO, CO₂, and H₂ consumption or production rates; Table S3) and employing maximization of ATP dissipation as the objective function. These values were validated by prediction of “optimal” growth phenotypes for experimental conditions (SIM27–48), which incorporated experimental constraints (specific cysteine, CO, CO₂, and H₂ consumption rates), ATP dissipation flux calculated above, and maximization of biomass yield as the objective function. Complete simulation results identified as SIMx (e.g., SIM1) in the text are in Table S11. SIM1-19, 27-41, and 49-55 are from Valgepea et al. (2018). Additional constraints were also used to improve the accuracy of predictive simulations were justifiable [SIM49-55; details described in Valgepea et al. (2018)]. Here, CO₂ reduction to formate was forced from the formate dehydrogenase (FdhA) reaction scheme (Reaction ID: rxn00103_c0) in SIM42-48, to the correct reaction scheme in SIM56-62, as identified by Mock et al. (2015)—FdhA/Hydrogenase ABCDE complex (HytABCDE; Reaction ID: rxn08518_c0). Additionally, since extracellular metabolome analysis did not detect excretion of pyruvate by the cells, we manually blocked pyruvate export in the model (rxn05469_c0) in SIM56-62. Reaction IDs comprising letters and numbers are labels corresponding to reactions in Table S11 (follows reference to reaction/enzyme).

Results

Clostridium autoethanogenum Steady-State Fermentation of Carbon Dioxide and Hydrogen

Clostridium autoethanogenum cells reached steady-state when growing on CO₂/H₂ in chemostats at dilution rate (D) ~0.5 day⁻¹ [specific growth rate (μ) ~0.02 h⁻¹] with a biomass concentration of 0.18 ± 0.02 g dry cell weight (gDCW)/L (Figure 1A). It is important to note that attempts to reach a steady-state at D = 1 day⁻¹ were unsuccessful. Unlike the chemostat cultures of C. autoethanogenum with CO (Valgepea et al., 2017, 2018) and CO₂/H₂ retentostat cultures (Mock et al., 2015), the CO₂/H₂ cultures could not reach stable biomass concentrations before the culture began oscillation cycles; previously observed above ~1.6 gDCW/L (Valgepea et al., 2017). The physiological reason and mechanism for such oscillatory culture behavior are under investigation, but we assumed that cell recycling is a requirement for CO₂/H₂ culture stability. For example, Molitor et al. (2019) showed consistent, high-biomass concentration and high-acetate CO₂/H₂ fermentation with Clostridium ljungdahlii in a retentostat with complete recycling.

Figure 1. Important fermentation characteristics of Clostridium autoethanogenum in autotrophic chemostats. Results from Valgepea et al. (2018) are also displayed (B–D), the conditions of all fermentations are summarized in Table 1. Growth curves of novel fermentations with standard deviation at steady-state (A). Specific rates of uptake (B) and production (C) for important metabolites. Product carbon balances (D). Values represent the average ± standard deviation between biological replicates. Number of biological replicates, and detailed gas composition for each fermentation are available in Table 1. Patterned bars indicate a D of 1 day⁻¹, full bars indicate a D of 0.5 day⁻¹ (B–D). q, specific rate; DCW, dry cell weight.

Despite the attempt to reach a steady-state at D = 1 day⁻¹, cells reached steady-state at dilution rate = 0.5 day⁻¹, with a specific rate of carbon incorporation (i.e., qCO₂) of 480 ± 80 mmol/gDCW/day (Figure 1B). Furthermore, the specific production rates of ethanol and acetate were 140 ± 10 and 113 ± 9 mmol/gDCW/day, respectively (Figure 1C). Strikingly, this meant around half of the captured carbon was converted to ethanol (54 ± 3 C mol%) (Figure 1D and Table S4). Fermentation conditions and titers are available in Table 1, showing an impressive ethanol concentration compared to previous fermentations where CO was the main carbon and energy source.

Despite the different dilution rate, the CO₂/H₂ results generated were compared to previously published chemostat cultures of C. autoethanogenum grown on CO, syngas, and CO/H₂ (Valgepea et al., 2018) at similar biomass concentrations (~0.5 gDCW/L) (Figures 1B–D). Specific rates of acetate and ethanol production achieved here for CO₂/H₂ cultures fell between those for syngas () and CO/H₂ () cultures (Figures 1B,D). However, the specific rate of carbon incorporation was higher for CO₂/H₂ (Figure 1C). We found that more than half of the captured CO₂ was converted into ethanol (Figure 1D). These results were encouraging, especially as ethanol production has unfavorable stoichiometry compared to acetate (Mock et al., 2015). Furthermore, the H₂ specific uptake rate (1,130 ± 160 mmol/gDCW/day) showed that higher H₂ uptake rates are achievable (compared to old datasets). These results show that higher carbon yields are possible (Valgepea et al., 2018). To investigate further the metabolic demand and the feasibility of CO₂/H₂ fermentation, we utilized the measured specific consumption and production rates and specific growth rate from the steady-state dataset as constraints for the genome-scale metabolic model (GEM) to find candidate mechanisms for improving CO₂/H₂ fermentation using the metabolic model iCLAU786.

Metabolic Model of Carbon Dioxide and Hydrogen Fermentation

Estimation of intracellular processes constrained by in vivo datasets represents an important developmental step for progressing acetogenic CO₂ valorization. Here, we used the GEM to compare intracellular metabolic flux distributions on CO₂/H₂ and CO-containing gases (Figure 2). See Figures S2, S3 and Tables S8–S10 for further flux comparison summaries.

Figure 2. Intracellular metabolic fluxes in Clostridium autoethanogenum growing on various gas mixes, estimated using the metabolic model iCLAU786 and flux balance analysis. Bar charts show specific flux rates (mmol/gDCW/h) from Tables S10, S11 and represent the average ± standard deviation between biological Replicates from SIM: 1–4 (CO), 9–10 (Syngas), 13–15 (CO/H₂), 20–21 (CO/CO₂/${\text{H}}_{2^1}$), 22–24 (CO₂/H₂), and 25–26 (CO/CO₂/${\text{H}}_{2^{0.5}}$). Results for CO, syngas, and CO/H₂ are low biomass condition data from Valgepea et al. (2018), the conditions of these fermentations are summarized in Table 1. Number of biological replicates, and detailed gas composition for each fermentation are available in Table 1. Arrows show the direction of calculated fluxes; red arrows denote uptake or secretion, dashed arrows denote a series of reactions. Brackets denote metabolites bound by an enzyme. Refer to Figures S1, S2 for enzyme involvement, metabolite abbreviations, and complete flux balance analysis datasets.

Intracellular metabolic fluxes estimated using the model iCLAU786 and flux balance analysis (FBA) (Figure 2) showed remarkable similarity to the combined theoretical stoichiometry of acetate and ethanol production (Mock et al., 2015) and indicated energetic cofactor circuits with mapping close to 1:1 (experimental:theoretical stoichiometry; Figure S4). Ethanol production likely occurred via acetaldehyde:ferredoxin oxidoreductase (AOR; ReactionID: leq000004; see Materials and Methods) under autotrophic conditions, with the HytABCDE (leq000001) and Nfn complex (leq000002) likely facilitating cofactor production via electron bifurcation (Figure 2) (Valgepea et al., 2018). This is a mechanism for minimization of free energy loss employed by C. autoethanogenum and may play a key role in sustaining proton motive force by balancing acetate, ethanol, and ATP production (Mock et al., 2015; Valgepea et al., 2018). Engineering acetogens to redirect this energy toward cellular growth, sacrificing some ethanol production, could be beneficial for CO₂ fermentation (Emerson and Stephanopoulos, 2019).

It was notable that, unlike CO fermentations, the intracellular pyruvate:ferredoxin oxidoreductase (PFOR; rxn05938_c0; acetyl-CoA ↔ pyruvate) flux was not significantly in the direction of pyruvate (Figure 2) (Valgepea et al., 2018). Under autotrophic conditions, PFOR links the WLP to anabolic pathways associated with biomass (Furdui and Ragsdale, 2000), and therefore this indicated high cell-specific energetic limitation. From this observation, we hypothesized that CO supplementation could provide a potential solution, as CO oxidation would generate additional Fd_red. Furthermore, an ATP/H₂ flux ratio of ~0.15 was estimated using metabolic modeling here compared to an ATP/CO ratio of ~0.28 in CO only fermentations (Valgepea et al., 2018). Considering CO/H₂ and CO₂/H₂ fermentations had equal carbon-flux through the WLP (~10 mmol/gDCW/h; Figure 2), supplementation with renewable CO from CO₂ electrolysis could aid biomass formation and culture stability. A similar process (but CO fermentation) was detailed by Haas et al. (2018).

Clostridium autoethanogenum Steady-State Fermentation of Carbon Dioxide and Hydrogen Supplemented With Carbon Monoxide

To validate our modeling hypothesis, Clostridium autoethanogenum was cultured with a low concentration of carbon monoxide in addition to CO₂ and H₂ (CO/CO₂/H₂) in chemostats. A steady-state was reached at D = 0.5 day⁻¹ (μ ~0.02 h⁻¹), and at D =1 day⁻¹ (μ ~0.04 h⁻¹; Figure 1A; biomass concentrations of 0.54 ± 0.01 and 0.34 ± 0.02 gDCW/L respectively). CO/CO₂/H₂ fermentations at a D = 1 day⁻¹ (CO/CO₂/${\text{H}}_{2^1}$; superscript 1 denotes D of 1 day⁻¹) and at a D = 0.5 day⁻¹ (CO/CO₂/${\text{H}}_{2^{0.5}}$; superscript 0.5 denotes D of 0.5 day⁻¹) showed simultaneous uptake of CO (89 ± 2 and 36 ± 4 mmol/gDCW/day, respectively) and CO₂ (940 ± 20 and 540 ± 20 mmol/gDCW/day, respectively) (Figure 1B). The co-utilization of both C1 gases is, to the best of our knowledge, an unquantified phenomenon. This led to a specific carbon incorporation (CO/CO₂/${\text{H}}_{2^1}-$1030 ± 30 mmol/gDCW/day) larger than any other gas type (maximum of ~450 mmol/gDCW/day for fermentations with CO in Valgepea et al. (2018) or CO₂/H₂ in this work). This also resulted in significant improvements to culture performance compared to CO₂/H₂ fermentations.

Compared to CO₂/H₂, CO/CO₂/${\text{H}}_{2^{0.5}}$ showed higher acetate and ethanol titers (Table 1) and specific productivities (Figure 1C), and a higher ethanol/acetate ratio (2.15 vs. 1.24 mol/mol respectively; Tables S1, S2). While at a similar biomass concentration (CO/CO₂/${\text{H}}_{2^1}$ best comparison due to similarity in dilution rate), acetate and ethanol titers (Table 1), and specific productivities (Figure 1C) are greater than during fermentation of other CO-containing gases. When comparing to high biomass (~1.4 gDCW/L) CO cultures, CO-supplementation still performs impressively—CO/H₂ fermentation achieved a higher ethanol titer (11.6 ± 0.4 g/L), while CO and syngas fermentations were similar (3.9 ± 0.2 and 5.4 ± 0.3 g/L respectively; Table S5). Otherwise, all specific productivities were higher for CO/CO₂/${\text{H}}_{2^1}$ (Figure S3). Furthermore, the distribution of carbon to ethanol was still >50% (53.8 ± 0.4% and 66 ± 2% for CO/CO₂/${\text{H}}_{2^1}$ and CO/CO₂/${\text{H}}_{2^{0.5}}$, respectively; Figure 1D and Table S4).

To understand the metabolic effects of supplementing CO, FBA was performed using the same conditions and alterations as for CO₂/H₂ (Figure 2). Notably, the WLP specific flux throughput for CO/CO₂/${\text{H}}_{2^1}$ was ~2-fold greater than for any other gas type (including high-biomass Valgepea et al., 2018). Furthermore, for CO₂ fermentations, Nfn complex flux direction was opposite that of CO and syngas fermentations. CO/CO₂/${\text{H}}_{2^{0.5}}$ also showed significantly greater flux through the AOR, whilst specific WLP productivity was insignificantly different compared to CO₂/H₂.

Discussion

Achieving steady-state continuous cultures using CO₂/H₂ mixtures, without cell recycling here, was challenging. Yet, compared to other organisms fermenting CO₂/H₂ with continuous medium exchange, Clostridium autoethanogenum performs well (Table 2). No direct comparisons can be made to other experiments due to variations in conditions, but C. autoethanogenum clearly achieves the highest ethanol production, with comparable quantities of carbonous products also. Acetobacterium woodii, along with Sporomusa ovata, were shown to perform well when compared to a wide range of acetogens under batch CO₂/H₂ conditions (Groher and Weuster-Botz, 2016). Yet, as evidenced by omission of S. ovata from Table 2, few continuous culture characterizations of acetogens are available—an essential step for validation of industrial robustness in gas fermentation. As discussed by Molitor et al. (2019), the lack of yeast extract or C_≥2 substrates is also distinguishing between fermentations.

Table 2. Summary of quantitative and continuous CO₂/H₂ fermentations.

Notably, CO₂/H₂ cultures displayed higher variability between biological replicates compared to those of CO-containing gas mixtures (Figure 1) (Valgepea et al., 2017). This may indicate variable organism fitness, a trait previously discussed for C. autoethanogenum by Liew et al. (2016), who extensively covered numerous techniques used for enhancing gas fermentation including—coupling to other processes, adaptive laboratory evolution, and metabolic engineering of acetogens using genetic tools. CO-supplementation could be a valuable option for enhancement as it overcomes inherent problems linked to engineering acetogens. Supplementation of low quantities of CO here stabilized the culture, enabled culturing at D = 1 day⁻¹, and achieved higher biomass concentration with a carbon incorporation larger than any other gas type—all without compromising by-product distribution.

While, Valgepea et al. (2018) found that syngas fermentation lead to CO-only fermentation at steady-state, we observed co-utilization of CO and CO₂. Tizard and Sechrist (2015) have also shown co-utilization for C. autoethanogenum continuous cultures, and it seems that co-uptake may also occur for some points of syngas batch fermentation (Infantes et al., 2020). Co-utilization of sugars was found for E. coli in chemostats—where inhibition of consumption, but no change in induction time was observed (Standing et al., 1972). The WLP is most likely no different, in that metabolism of CO is preferential, yet the pathway can co-consume CO₂ under certain conditions.

Various efforts have been made toward enhancing CO₂(+H₂) fermentation to C_≥2 products (Table 2) (Emerson and Stephanopoulos, 2019). Braun and Gottschalk (1981) first discovered the potential for enhancement when Acetobacterium woodii simultaneously consumed fructose and a headspace of CO₂/H₂ during batch cultivation. Growth and acetate production was high but no characterization of the headspace was performed. More recently, continuous glucose-supplemented CO₂/H₂ fermentation of Moorella thermoacetica by Park et al. (2019) did not lead to net uptake of CO₂. Furthermore, Jones et al. (2016) did not show net CO₂ uptake for a wide range of acetogens (not A. woodii) fermenting syngas and fructose. A. woodii generates a sodium ion (Na⁺) gradient (Hess et al., 2013) rather than a proton (H⁺) gradient for membranous ATP generation (Pierce et al., 2008; Poehlein et al., 2015; Bengelsdorf et al., 2018). This may highlight an important metabolic difference from other model acetogens—decoupling the resources of the WLP and membranous ATP generation pathways could facilitate fermentation of sugar and CO₂/H₂ simultaneously.

Other enhancements have also struggled to achieve net CO₂ uptake. Co-culture of C. acetobutylicum and C. ljungdahlii showed syntrophic metabolic coupling when fermenting glucose, fructose, and CO₂/H₂, but no net CO₂ uptake (Charubin and Papoutsakis, 2019). Addition of nitrate to batch CO₂/H₂ fermentation by C. ljungdahlii, increased biomass concentration and subsequently volumetric productivity of acetate (Emerson et al., 2019). However, the specific WLP productivity decreased, meaning lower utilization of CO₂. Other organisms not recognized as gas fermenters can also use mixotrophy to minimize carbon loss, such as Clostridium beijerinckii but have not displayed net CO₂ uptake either (Sandoval-Espinola et al., 2017). To the best of our knowledge, this is the first report where supplementation of a substrate other than H₂, increased productivities of continuous acetogenic CO₂ fermentation while maintaining net CO₂ utilization. Furthermore, the effect of CO supplementation on CO₂ utilization was superlinear, indicating a synergistic mechanism (Park et al., 2019). This is encouraging for development of bioprocesses valorizing CO₂.

Comparisons between fermentation datasets enables us to speculate about the positive effect of CO-supplementation on CO₂/H₂ fermentation. Although, addition of CO led to minimal metabolic shifts (Figure 2—CO₂/H₂ vs. CO/CO₂/${\text{H}}_{2^{0.5}}$ and Figure S2), FBA showed that CO supplementation caused significant increases to the reduced ferredoxin consumption by AOR and Rnf complex (leq000004 and M002, respectively) compared to CO₂/H₂ (Figure 2). The overflow model proposed by Richter et al. (2016) suggests that high NADH production via Rnf and Nfn complexes (leq000002) is also important for reducing AOR product inhibition. In this way, NADH facilitates fast metabolism of acetaldehyde to ethanol via alcohol dehydrogenase (Adh(E); rxn00543_c0). Decreasing the acetate concentration reduces acidification and the ATP cost for excreting acetate (Valgepea et al., 2018). Including acetaldehyde conversion to ethanol and association to acetic acid, this also leads to consumption of 2 H⁺ (4 here vs. 2 produced via CODH). Therefore, CO consumption decreases the intracellular H⁺ pool, and following Le Chatelier's principle, drives HytABCDE activity. Indeed, the change in specific H₂ uptake relative to specific CO₂ uptake is greater than that of CO (for CO₂/H₂ vs. CO/CO₂/${\text{H}}_{2^{0.5}}$, Table S7). Subsequently, the relative gain in free energy from H₂ is ~ 4-fold greater than CO. We speculate this is ultimately responsible for the improved fitness of CO-supplemented CO₂/H₂ fermentation by C. autoethanogenum. We propose the following five critical factors to this enhanced metabolism: [1] metabolism of CO increases the intracellular pool of reduced ferredoxin; [2] this stimulates oxidation of ferredoxin, which if consumed by the AOR; [3] reduces ATP costs; and [4] decreases the H⁺ pool/acidification; which therefore [5] drives H₂ uptake for further reduction of ferredoxin. Evidently, additional understanding of acetogenic redox metabolism, from a thermodynamic perspective, is important for developing acetogenic CO₂-valorization as a platform industrial bioprocess (Cueto-Rojas et al., 2015).

Physicochemical properties could also play a key role in CO-supplementation enabling to achieve a stable CO₂/H₂ chemostat culture at D = 1 day⁻¹. Generation of a stable and large non-equilibrium is what drives microbial growth (Qian and Beard, 2005; Igamberdiev and Kleczkowski, 2009; Quéméner and Bouchez, 2014) and gas-liquid mass transfer (Ma et al., 2005). For continuous culture of gas fermenting microbes, an inherent relationship between substrate mass transfer and culture growth exists (Supplementary Materials Note 2.1). An important parameter for these systems is the Gibb's free energy of a system (Cueto-Rojas et al., 2015). This describes the thermodynamic favorability of the reaction system—termed spontaneity. Here, analysis of experimental flux and Gibbs free energy suggests that CO₂/H₂ fermentation is infeasible ($\Delta {\dot{G}}_{O{R}^0}=5.4kJ/mol/day$), whereas CO-supplemented CO₂/H₂ fermentation is feasible ($\Delta {\dot{G}}_{O{R}^0}=-12.3kJ/mol/day$; Table S6). Though these calculations use standard conditions, they do indicate how close CO₂/H₂ fermentation is to the thermodynamic limit of metabolism. Theoretically, minute and unobservable changes to chemostat CO₂/H₂ fermentation can disrupt the culture (Henry and Martin, 2016). Thus, increasing the free energy of central metabolism with CO-supplementation appears to keep metabolism in a spontaneous and stable state by increasing reduced ferredoxin production.

The mechanisms for achieving the 2-fold higher specific WLP flux throughput for CO/CO₂/${\text{H}}_{2^1}$ compared to others is less clear but appears to be linked to the difference in primary substrate. CO/CO₂/${\text{H}}_{2^1}$ and CO/H₂ are the most similar CO₂ and CO fermentations, respectively (D ~1 day⁻¹ and carbon to hydrogen feed ratio (~1:3); Table 1), and the maximum carbon incorporation per cell for CO/H₂ was roughly half of that of CO/CO₂/${\text{H}}_{2^1}$ (~450 vs. ~1,000 mmol/gDCW). Theoretically, cells will maximize carbon-to-redox metabolism by minimizing thermodynamic losses. CO supplementation to a CO₂/H₂ culture seems to facilitate this as (H₂/carbon)_feed–(H₂/carbon)_flux was ~0 mol/mol for CO/CO₂/H₂ fermentations only (Table S2)—an indication of the relative magnitude of carbon and redox metabolism. This suggests that high specific fluxes for CO/CO₂/${\text{H}}_{2^1}$ may be a result of (close to) optimal co-factor recycling by C. autoethanogenum's WLP and redox pathway. Thus, the lower energy associated with CO₂ fermentation may, counterintuitively, stimulate specific WLP activity when in the presence of appropriate energy-containing substrates. Further quantifications of CO₂ metabolism and characterizations of enzyme activities are required to confirm these hypotheses (Supplementary Materials Note 2.2), as they assist our ability to engineer the links between redox and carbon metabolisms.

We established a dataset quantifying steady-state of the model acetogen C. autoethanogenum during autotrophic-CO₂/H₂ growth in chemostat cultures. This enabled analysis via FBA, and highlighted CO as a potential supplement. CO supplementation successfully improved metabolic stability and CO₂ utilization. This was the first time that intracellular fluxes for net uptake of CO₂ (with enhancement) where characterized. Industry is actively developing gas fermentation to valorize CO₂ (Tizard and Sechrist, 2015; Haas et al., 2018). Previously, genetic and process engineering of gas fermentation successfully developed the technology for industrial CO valorization (Liew et al., 2016). Therefore, progression to industrial CO₂ valorization is foreseeable, and CO supplementation may play a role in the continuing diversification of industrial gas fermentation.

Data Availability Statement

All datasets generated for this study are included in the article/Supplementary Material.

Author Contributions

All authors viewed and approved the manuscript and contributed significantly to the work. KV, EM, and LN conceived the project. JH, KV, and EM designed the experiments and analyzed the results. JH and KV performed experiments, supported by RS, IC, MP, and EM. JH wrote the manuscript with the help of KV, EM, RT, SS, MK, and LN.

Funding

This study was funded by a Grant from the Australian Research Council, partly funded by LanzaTech (ARC LP140100213).

Conflict of Interest

LanzaTech has interest in commercializing gas fermentation with C. autoethanogenum. RT, SS, and MK are employees of LanzaTech.

Acknowledgments

Elements of this research utilized equipment and support provided by the Queensland node of Metabolomics Australia, an initiative of the Australian Government being conducted as part of the NCRIS National Research Infrastructure for Australia. IC would like to acknowledge support from the German Academic Exchange Service (DAAD) through the DAAD Kurzstipendien für Doktoranden. We thank the following investors in LanzaTech's technology: Sir Stephen Tindall, Khosla Ventures, Qiming Venture Partners, Softbank China, the Malaysian Life Sciences Capital Fund, Mitsui, Primetals, CICC Growth Capital Fund I, L.P. and the New Zealand Superannuation Fund. There was no funding support from the European Union for the experimental part of the study. However, KV acknowledges support also from the European Union's Horizon 2020 research and innovation programme under grant agreement N810755.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe.2020.00204/full#supplementary-material

References