Steerable isobutyric and butyric acid production from CO₂ and H₂ by Clostridium luticellarii

Abstract

Clostridium luticellarii is a recently discovered acetogen that is uniquely capable of producing butyric and isobutyric acid from various substrates (e.g. methanol), but it is unclear which factors influence its (iso)butyric acid production from H₂ and CO₂. We aimed to investigate the autotrophic metabolism of C. luticellarii by identifying the necessary growth conditions and examining the effects of pH and metabolite levels on product titers and selectivity. Results show that autotrophic growth of C. luticellarii requires the addition of complex nutrient sources and the absence of shaking conditions. Further experiments combined with thermodynamic calculations identified pH as a key parameter governing the direction of metabolic fluxes. At circumneutral pH (~6.5), acetic acid is the sole metabolic end product but C. luticellarii possesses the unique ability to co‐oxidize organic acids such as valeric acid under high H₂ partial pressures (>1 bar). Conversely, mildly acidic pH (≤5.5) stimulates the production of butyric and isobutyric acid while partly halting the oxidation of organic acids. Additionally, elevated acetic acid concentrations stimulated butyric and isobutyric acid production up to a combined selectivity of 53 ± 3%. Finally, our results suggest that isobutyric acid is produced by a reversible isomerization of butyric acid, but valeric and caproic acid are not isomerized. These combined insights can inform future efforts to optimize and scale‐up the production of valuable chemicals from CO₂ using C. luticellarii.

Clostridium luticellarii is a highly flexible microorganism, able to reverse its (reverse) beta‐oxidation pathway depending on the pH. At circumneutral pH, C. luticellarii produces acetic acid from H₂ and CO₂ but is able to co‐oxidize organic acids and isomerize butyric acid. At mildly acidic pH (~5.5), C. luticellarii is able to produce acetic acid but also the chain elongation products butyric and isobutyric acid.

BACKGROUND

Since the onset of the industrial revolution, atmospheric carbon dioxide (CO₂) concentrations have been increasing at a rapid pace due to anthropogenic activities (Song, 2006). Global emissions of greenhouse gases amounted to 36 Gtons of CO₂ equivalents per year in 2020 (Ritchie 2021), of which approximately 29% resulted from a limited number of energy‐intensive industries (the most notorious being the steel and iron, cement, (petro)chemical and paper and pulp industries) (Koytsoumpa et al., 2018; European Commission, Joint Research Centre, 2021). These industries form interesting targets for the carbon capture and utilization (CCU) concept, in which CO₂ is captured at the point of emission or after emission and used as a feedstock for the production of added‐value compounds such as platform chemicals (e.g. methanol or organic acids), plastics or biofuels (Boot‐Handford et al., 2014; Martens et al., 2017). CCU thereby decreases both direct CO₂ emissions and the dependence of the chemical industry on fossil fuels as carbon source (Garcia‐Garcia et al., 2021).

One interesting technology for CCU, gas fermentation, harnesses acetogens, a group of strict anaerobic bacteria. Acetogens are highly flexible microorganisms: they can grow on a wide variety of organic compounds such as carbohydrates and methylated compounds such as methylamines, methanol or formic acid (Drake, 1995; Schuchmann & Müller, 2014). The most fascinating part of their metabolism, however, lies in their facultative autotrophic behaviour, meaning that they can grow on inorganic carbon sources such as CO₂ using inorganic electron sources such as H₂ and CO (Schuchmann & Müller, 2014). Their carbon‐fixating pathway, the Wood‐Ljungdahl Pathway (WLP), is considered one of the most energy‐efficient carbon fixating pathways in nature, achieving energetic efficiencies of 55%–95% depending on the C1 carbon source, rendering acetogens interesting biocatalysts for CCU (Claassens et al., 2019).

In the WLP, two CO₂ molecules are reduced with electrons coming from, for example, H₂ oxidation to, respectively, a methyl and carbonyl group, which are condensed together with coenzyme A (CoA) to the central intermediate acetyl‐CoA (Schuchmann & Müller, 2014). While most acetogens rely on converting that acetyl‐CoA to acetic acid as the sole possible end‐product of their metabolism, several can produce more valuable end‐metabolites (Bengelsdorf et al., 2018). For example, the acetogen Clostridium autoethanogenum is currently used at a commercial scale by Lanzatech to convert carbonaceous off‐gases from steel mills (mainly CO) to ethanol (Liew et al., 2016). However, the production of longer‐chain products is gaining increased interest due to their substantially higher market value (e.g. butyric acid has 3–5 times higher value than acetic acid) (Calvo et al., 2022; Kim et al., 2019). Several acetogens have been shown to convert CO₂, CO and/or organic C1 compounds to these longer acids via a process called chain elongation. This process uses the reverse β‐oxidation pathway where two acetyl‐CoA molecules (either produced via the WLP or assimilated from acetic acid) are condensed to acetoacetyl‐CoA which is subsequently transformed and reduced (with electrons coming from H₂ oxidation) to butyryl‐CoA and finally converted to butyric acid. As such, the carbon chain of acetyl‐CoA (or acetic acid) is elongated with two carbon atoms to butyric acid, which in some acetogens can be elongated in a second chain elongation cycle to caproic acid (Bengelsdorf et al., 2018). Examples of chain elongating acetogens include: Eubacterium limosum, able to produce acetic and butyric acid from C1 compounds (Litty & Müller, 2021); Butyribacterium methylotrophicum, capable of producing the same products and also their respective alcohols, ethanol and butanol (Worden et al., 1991); and C. carboxidivorans, able to produce acetic acid, butyric acid, caproic acid and their respective alcohols (Ramió‐Pujol et al., 2015).

Nevertheless, while many acetogens have these interesting metabolic properties and end‐products, they do not produce significant amounts of products beyond acetic acid autotrophically at optimal growth conditions (Litty & Müller, 2021; Richter et al., 2016). However, suboptimal growth conditions such as low pH have been reported to stimulate shifts in product spectrum from organic acids to alcohol production (Abubackar et al., 2012; Arslan et al., 2019; Fernández‐Naveira et al., 2016; Martin et al., 2016), and from acetic acid and ethanol to butyric acid and butanol (Worden et al., 1991). Similarly, high concentrations of intermediate metabolic products (i.e. acetic acid) have been reported to stimulate chain elongation from methanol and CO₂ in methylotrophic acetogens (Lynd & Zeikus, 1983; Pacaud et al., 1986; Petrognani et al., 2020), although this effect seems to be less studied under autotrophic growth conditions.

The recently isolated acetogen Clostridium luticellarii was found to produce butyric and isobutyric acid from methanol and CO₂, hypothetically via the WLP and consecutive reverse β‐oxidation to butyric acid and isobutyric acid (Petrognani et al., 2020; Wang et al., 2015). Both butyric and isobutyric acids are interesting platform molecules. The former has applications in food additives, fragrances, antibacterial agents and bioplastics, while its branched isomer has a high market volume as a precursor for poly‐methyl methacrylate (acrylic glass) with a market size of 2.7 million tons per year (Lang et al., 2014; Moscoviz et al., 2018). Since both the compounds are currently produced from petrochemically derived propylene (Zhang et al., 2011), establishing a direct route to biologically produce butyric and isobutyric acid from CO₂ and H₂ could greatly improve their sustainability as high‐volume platform chemicals while simultaneously capturing carbon. However, it currently remains unclear whether C. luticellarii is able to grow autotrophically and which factors influence its (iso)butyric acid production from H₂ and CO₂. Here, we elucidate the autotrophic growth capacity of C. luticellarii and how its butyric and isobutyric acid production is affected by the presence of increased concentrations of intermediate and end‐products. Next, the impact of pH as a factor governing the production of longer‐chain products was investigated. Finally, a thermodynamic approach was used to understand the metabolic flexibility of C. luticellarii under diverse environmental conditions.

EXPERIMENTAL PROCEDURES

Inoculum and media preparation

Clostridium luticellarii DSM 29923 was obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). The strain was activated at pH 7.2 and 37°C on Reinforced Clostridial medium (RCM, as described by Petrognani et al. (2020)) and sub‐cultured at pH 6.5 on basal medium (10% v/v inoculum ratio) with 200 mM methanol and 40 mM NaHCO₃ as carbon sources before transferring to autotrophic growth conditions on basal medium. An inoculum ratio of 10% (v/v%) was used in all experiments unless stated otherwise. Non‐inoculated controls were used alongside each subculture and experimental run to check for contaminations and the purity of the culture was regularly checked via microscopic examination and Sanger sequencing of the 16S rRNA gene.

The basal medium contained (per 1 L): 0.36 g NH₄H₂PO₄; 0.133 g CaCl₂; 0.325 g MgCl₂.6H₂O; 0.197 g MgSO₄.7H₂O; 0.149 KCl; 14.643 g 2‐(N‐morpholino) ethanesulfonic acid (MES) buffer; and 0.001 g resazurine. The basal medium was boiled, cooled down under continuous sparging of N₂ and dispensed in 1 L Duran pressure plus Schott bottles under an N₂ atmosphere. Subsequently, the headspace of the bottles was flushed with N₂ for at least 15 min. After autoclaving, the headspace of the bottles was flushed with filter sterilized H₂/CO₂ (80/20 v/v%) for at least 15 min and pressurized to 150 kPa absolute. Before inoculation, vitamins and trace elements (Tables S1–S3), 30 mM NaHCO₃, 0.4 mM titanium (III) citrate (reducing agent) and 0.5 g L⁻¹ tryptone (only added in the experiments detailed in Impact of organic acid addition and pH on the autotrophic metabolism) were added to the basal medium from anaerobic filter sterilized concentrated stock solutions.

Batch experiments

Impact of environmental and operational conditions enabling autotrophic growth

Experiments were conducted using various media compositions to pinpoint the conditions enabling the autotrophic growth of C. luticellarii. First, four conditions were tested in triplicate in 1050 mL bottles (Duran Pressure Plus) with 75 mL working volume: (i) growth on basal medium without the addition of any complex nutrient sources, (ii) growth on basal medium with twice concentrated vitamins and trace elements, (iii) addition of 1 g L⁻¹ of yeast extract to (i), and (iv) addition of 0.5 g L⁻¹ tryptone to (i). To eliminate any traces of methanol for the subsequent autotrophic experiments from the basal medium, the methylotrophic culture was sub‐cultured once on a new basal medium containing no organic carbon sources under 150 kPa of an 80:20 H₂:CO₂ atmosphere and used as inoculum for the subsequent experiments. After the addition of the respective nutrient sources and adjustment of the pH to 6.5, the bottles were inoculated and incubated statically and horizontally (to maximize gas–liquid mass transfer area) at 37°C. Right after inoculation, and at regular time intervals, the headspace pressure was measured with a tensiometer (GMH 3111 equipped with a 603310 MSD 2.5 BAE sensor, Greisinger) and needle, gas samples (≈5.0 mL) were taken for immediate GC analysis and liquid samples (1.0 mL) were taken for immediate pH and OD measurements. The remaining liquid samples were filtered (0.20 μm) and stored at −18°C until HPLC analysis.

The impact of shaking incubation was investigated by comparing the growth of C. luticellarii incubated statically and horizontally, with incubation horizontally on a shaker at 150 rpm. Bottles were incubated at 37°C. The same medium as case (iv) described above was used. The pressure decrease was monitored daily by using a tensiometer with a needle as a proxy for gas consumption and growth. Experiments were conducted in triplicate in 250 mL penicillin bottles with 40 mL working volume to be able to rapidly and accurately detect the pressure decrease.

Impact of organic acid addition and pH on the autotrophic metabolism

The autotrophic product spectrum was characterized in further experiments using the medium described in Inoculum and media preparation with 0.5 g L⁻¹ tryptone and by incubating bottles statically and horizontally (Table 1). To obtain a reference for comparison, C. luticellarii was grown in the first experiment under 150 kPa absolute of 80:20 H₂:CO₂ atmosphere and at pH 6.5. Simultaneously, the effect of organic acids addition at initial pH 6.5 was investigated by growing C. luticellarii in the presence of acetic acid (50 mM), butyric acid (50 mM), isobutyric acid (50 mM), valeric acid (25 mM) and caproic acid (25 mM). The next experiment further investigated the effect of valeric acid on autotrophic metabolism at initial pH 6.5 by growing C. luticellarii at different valeric acid concentrations of, respectively, 10 mM, 50 mM and 75 mM. A third experiment investigated the effect of pH by growing C. luticellarii at initial pH 5.5 and pH 6.5 while manually controlling the pH between 5.0–5.5 and 6.0–6.5, respectively. To that end, pH was measured on the taken samples, and 2 M NaOH was added to the bottles to correct the pH to 5.5 or 6.5 depending on the condition. This was done in three different scenarios: without organic acid addition, with acetic acid addition (50 mM) and with valeric acid addition (25 mM). All the conditions were tested in triplicate. A final fourth experiment was conducted in triplicate by incubating C. luticellarii with 25 mM valeric acid at pH 6.5 in the absence of H₂. The bottles were flushed with an N₂ and CO₂ gas mixture (N₂:CO₂ 90:10) and pressurized to 150 kPa absolute.

TABLE 1.

Summary of experimental conditions.

Experiment	Condition	Initial pH	Addition	Concentration (mM)	Inoculum ratio (% v/v)
1	A	6.5	–	–	10
	B		Acetic acid	50
	C		Butyric acid	50
	D		Isobutyric acid	50
	E		Valeric acid	25
	F		Caproic acid	25
	G		Acetic acid	50
			Butyric acid	50
2	A	6.5	Valeric acid	10	10
	B			50
	C			75
3	A	6.5 (controlled)	–	–	25
	B	5.5 (controlled)	–	–
	C	5.5 (controlled)	Acetic acid	50
	D	6.5 (controlled)	Valeric acid	25
	E	5.5 (controlled)	Valeric acid	25
4	–	6.5	Valeric acid without H2	25	10

After the addition of organic acids (except for conditions 1A, 3A and 3B), and adjustment of the pH, the bottles were inoculated with a 10% inoculum ratio (except for experiment 3 in which a 25% inoculum ratio was used for all conditions to minimize the lag phase in the conditions at pH 5.5) of C. luticellarii in end‐exponential growth phase after at least sub‐culturing two times on autotrophic growth medium. These experiments were carried out in Schott bottles of a total volume of 1050 mL with a 75 mL working volume, incubated statically, horizontally, and at 37°C. Sampling was carried out as described in Impact of environmental and operational conditions enabling autotrophic growth. Experiments were sampled until no net decrease of headspace pressure was observed for two consecutive sampling moments.

Analytical methods

Carboxylic acids (C1, C2, C3, C4, C5, C6 including C4, C5 and C6 isoforms) and alcohols (ethanol, isobutanol, 1‐butanol) were analysed by HPLC (LC‐2030C Plus Prominence‐© series, Shimadzu) equipped with an Aminex HPX‐87H column (300 × 7.8 mm, BioRad) coupled with a Micro‐Guard cartridge. Carboxylic acids and alcohols were quantified using a UV detector with wavelength 210 nm (UV/Vis SPD‐40, Shimadzu) and a refractive index detector (RID‐20A, Shimadzu), respectively. 20 μL of the sample was eluted with 5 mM H₂SO₄ and 1% acetonitrile at a flow rate of 0.45 mL min⁻¹. The column oven temperature was set at 30°C and the UV and RID cell temperatures were set at 40°C. The composition of the headspace was analysed using a Compact Gas Chromatograph (Global Analyser Solutions, Breda, The Netherlands), equipped with a Molsieve 5A pre‐column and Porabond column (CH₄, O₂, H₂ and N₂) and an Rt‐Q‐bond pre‐column and column (CO₂). Concentrations of gases were determined by means of a thermal conductivity detector. OD was measured on 250 μL aliquots at 620 nm in 96 well plates using a plate reader (Infinite M200 PRO, Tecan).

Calculations

Gas consumption

The absolute amount of gas i in the bottles was calculated from the measured headspace pressure and headspace composition of that gas using the ideal gas law according to equation 1:

$$n_i=\frac{\mathit{PV}}{\mathit{RT}}{\chi }_i$$ (1)

where n _i is the absolute amount of gas i (mol), R is the molar gas constant (8.31 J mol⁻¹ K⁻¹), T is the temperature at which P was measured (310.15 K), P is the headspace pressure (Pa), χ _i is the molar fraction of gas i in the headspace and V is the headspace volume in the bottle (m³).

The absolute amount of gas was calculated before and after every sampling point, using the change of headspace pressure in the control to determine the change in pressure due to sampling (~1 kPa per sample). The change between the absolute amount of gas after sampling on a certain timepoint and the gas amount before sampling on the next timepoint represented the gas consumption between two timepoints. These gas consumptions were summed over all timepoints to determine the total gas consumption (equation 2).

$$\begin{array}{c}\text{total}\mathrm{gas}\text{consumption of}\mathrm{gas}i=\sum _{j=t_1}^{t_{\mathrm{end}}}{n}_{i,j-1\left(\text{after sampling}\right)}-n_{i,j\left(\text{before sampling}\right)}\end{array}$$ (2)

Metabolite production

The volumes taken during sampling were taken into account when calculating electron balances. The absolute amount of products was calculated before and after every sampling point by multiplication of the concentration with the volume in the bottles taking into account the change in liquid volume due to sampling. The change between the absolute amount of product after sampling on a certain timepoint and the absolute amount of product before sampling on the next timepoint represented the production between two timepoints. These productions were summed over all timepoints to determine the total production (equation 3).

$$\begin{array}{c}\text{total production of compound}i=\sum _{j=t_1}^{t_{\mathrm{end}}}{m}_{i,j\left(\text{before sampling}\right)}-m_{i,j‐1\left(\text{after sampling}\right)}\end{array}$$ (3)

where m _i,j is the absolute amount of electron equivalents contained in compound i at time point j.

Carboxylic acid selectivity

The selectivity with which specific carboxylic acids were produced was calculated as the net absolute amount of electron equivalents (mol eeq.) of that specific carboxylic acid relative to the net electron equivalents of all produced carboxylic acids (equation 4).

$$\begin{array}{c}\text{Carboxylic acid selectivity}\left(\%\mathrm{el}.\mathrm{eq}.\right)=\frac{\text{carboxylic acid}\left(\mathrm{mol}\mathrm{eeq}.\right)}{\sum \text{carboxylic acids}\left(\mathrm{mol}\mathrm{eeq}.\right)}100\%\end{array}$$ (4)

Thermodynamic calculations

The thermodynamics of the valeric acid oxidation pathway to propionic and acetic acid under an H₂‐containing atmosphere were calculated in two cases: (i) oxidation of valeric acid assuming that protons are used as a final electron sink to produce H₂ (equation 5) and (ii) oxidation of valeric acid with using intracellular electron carriers as electron acceptors (equation 6), which are regenerated in other metabolic process such as the WLP.

$$\begin{array}{c}{\mathrm{C}}_5{\mathrm{H}}_9{{\mathrm{O}}_2}^{-}+2{\mathrm{H}}_2\mathrm{O}\to {\mathrm{C}}_3{\mathrm{H}}_5{{\mathrm{O}}_2}^{-}+{\mathrm{C}}_2{\mathrm{H}}_3{{\mathrm{O}}_2}^{-}+{\mathrm{H}}^{+}+2{\mathrm{H}}_2\end{array}$$ (5)

$$\begin{array}{c}\begin{array}{c}{\mathrm{C}}_5{\mathrm{H}}_9{{\mathrm{O}}_2}^{-}+2{\mathrm{Fd}}_{\mathrm{red}}+3{\mathrm{NAD}}^{+}+2{\mathrm{H}}_2\mathrm{O}\to {\mathrm{C}}_3{\mathrm{H}}_5{{\mathrm{O}}_2}^{-}\\ +{\mathrm{C}}_2{\mathrm{H}}_3{{\mathrm{O}}_2}^{-}+2{\mathrm{H}}^{+}+2{\mathrm{Fd}}_{\mathrm{ox}}+3\text{NADH}\hfill \end{array}\hfill \end{array}$$ (6)

Electron confurcation (i.e. the reverse of electron bifurcation as occurring in valeric and butyric acid formation via reverse β‐oxidation) was assumed in case (ii) for the oxidation of valeryl‐CoA to pent‐2‐enoyl‐CoA with NAD⁺ where the oxidation needs an energy investment (provided by the exergonic NAD⁺ reduction with ferredoxin) (González‐Cabaleiro et al., 2013; Kleerebezem & Stams, 2000; Li et al., 2008).

Standard Gibbs free energy and enthalpy of formation of NADH were taken from Alberty (2003) and, in the case of Ferredoxin, the Gibbs free energy was calculated from the E°′ values (at standard conditions corrected for pH 7) described in the literature (~−450 mV (Battistuzzi et al., 2000; Bellei et al., 2010)) using the Nernst equation, while all other values were taken from Kleerebezem and Van Loosdrecht (2010). All used values are listed in Table S4. These values were used to calculate the standard Gibbs energy change of the reaction and the standard enthalpy change of reaction at 25°C and used to calculate the Gibbs energy change of reaction at 37°C for both cases as described by Kleerebezem and Van Loosdrecht (2010). The resulting Gibbs energy values (∆G°′) were converted to account for realistic intracellular metabolite concentrations (i.e. 10 mM valeric acid and 1 mM propionic and acetic acid (Bennett et al., 2009; González‐Cabaleiro et al., 2013)) and H₂ partial pressures of 120 kPa absolute using equation 7. Calculations were done assuming a reduced:oxidized ferredoxin ratio of 9 and NADH:NAD⁺ ratio of 0.1 (Buckel & Thauer, 2013).

$$\begin{array}{c}∆G^{\prime }=∆G^{0^{\prime }}+\mathit{\text{RTln}}\left(Q\right)\end{array}$$ (7)

where ∆G′ and ∆G°′ are the Gibbs free energy change of reaction (J mol⁻¹) at intracellular metabolite concentrations and at standard concentrations, respectively, R is the molar gas constant (8.31 J mol⁻¹ K⁻¹), T is the temperature of reaction (310.15 K) and Q is the reaction quotient.

RESULTS

Clostridium luticellarii requires strict conditions for autotrophic growth

Initially, several attempts were made to transfer C. luticellarii from a methylotrophic growth medium as in Petrognani et al. (2020) to the same growth medium without methanol but with H₂ and CO₂ as electron and carbon source. However, the strain repeatedly lost its capacity to grow after approx. three subcultures. Hence, a systematic experiment was set up to determine the conditions enabling autotrophic growth of C. luticellarii. C. luticellarii was transferred into pressurized bottles (150 kPa of an 80:20 H₂:CO₂ atmosphere) containing four basal media compositions, all in triplicate. Results showed that autotrophic growth of C. luticellarii without a complex nitrogen source (as in media composition (i) and (ii)) was possible to some extent with observable gas consumption, but the culture grew slowly to a low final OD (0.14 ± 0.02 for case (i) and 0.07 ± 0.00 for case (ii)) and lost its ability to grow after approx. 3 subcultures regardless of the concentration of vitamins and trace elements, indicating that the culture was lacking growth factors (Figure 1A). C. luticellarii showed faster and more stable growth (across sub cultures) with either 1 g L⁻¹ yeast extract or 0.5 g L⁻¹ tryptone to a max OD of 0.21 ± 0.00 and 0.18 ± 0.02, respectively (Figure 1A). However, since yeast extract also contains carbohydrates (Tomé, 2021), which makes it unclear whether observed growth was actually autotrophic and might distort the product spectrum towards more reduced products, we decided to use tryptone as complex nutrient source in subsequent experiments. To verify that tryptone did not interfere with the desired autotrophic growth of C. luticellarii, we also investigated whether it could support growth without addition of H₂ and CO₂ or other carbon sources and observed no growth (Figure S1). Next, the impact of shaking conditions (to maximize mass transfer from the gas to liquid phase) was assessed by comparing the static incubation of C. luticellarii to its incubation on a shaking incubator at 150 rpm. The results of these experiments showed a decreased rate of gas consumption and decreased amount of total gas consumed under shaking conditions compared to static incubations (Figure 1B). Hence, static incubations were used for all subsequent experiments.

FIGURE 1.

(A) Growth curves of the four different conditions tested to determine optimal growth conditions. TE, trace elements. (B) Headspace pressures used as proxy for gas consumption and growth during static incubations and dynamic incubations on a shaker at 150 rpm. (C) Timeseries of product concentrations during autotrophic growth at initial pH 6.5, with 0.5 g L⁻¹ tryptone and statically incubated. Error bars represent standard error of three biological replicates.

Once the growth conditions were established, C. luticellarii was first grown under 150 kPa of an 80:20 H₂:CO₂ atmosphere and at initial pH 6.5 to obtain a reference dataset for comparison with other growth conditions. Under these conditions, C. luticellarii grew to a maximum OD₆₂₀ of 0.18 ± 0.01 and produced almost exclusively acetic acid to a concentration of 60.46 ± 1.41 mM within 35 days. Beyond acetic acid, only traces of butyric (0.77 ± 0.05 mM) and isobutyric (0.57 ± 0.01 mM) acids were detected (Figure 1C).

The effect of acetic and butyric acid on autotrophic isobutyric acid production

In many methylotrophic acetogens capable of producing butyric acid, the concentrations of butyric acid produced from methanol and CO₂ are heavily influenced by the concentrations of metabolic products such as acetic acid (Kremp & Müller, 2020; Pacaud et al., 1986; Petrognani et al., 2020), but their effect on autotrophic growth remains unclear. C. luticellarii was grown autotrophically in the presence of 50 mM acetic acid, 50 mM butyric acid or 50 mM isobutyric acid, to assess their individual impact (Figure 2). Supplementation of acetic acid enabled the production of butyric acid and isobutyric acid to near‐equimolar concentrations of respectively 8.98 ± 0.91 mM and 9.27 ± 1.32 mM. In terms of selectivity, butyric and isobutyric acid accounted for, respectively, 26 ± 1% and 27 ± 2% of the product spectrum based on electron equivalents contained in the products, with the remaining fraction being acetic acid at a final concentration of 100.61 ± 2.35 mM. When 50 mM of butyric acid was added to the medium, a net decrease in the butyric acid concentration by 16.40 ± 1.80 mM was observed with the concurrent production of 18.06 ± 2.41 mM isobutyric acid, which accounted for 59 ± 6% of the electrons in the product spectrum, alongside acetic acid production to a final concentration of 78.75 ± 4.68 mM. The reversed reaction was also investigated by adding 50 mM isobutyric acid, which resulted in isobutyric acid consumption of 15.84 ± 1.43 mM with a concomitant production of butyric acid up to 13.33 ± 0.82 mM at a selectivity of 57 ± 5% based on electrons in the products. The remaining product spectrum consisted of acetic acid at a final concentration of 70.93 ± 3.32 mM. Interestingly, the lag phase when isobutyric acid was added was much longer than with any of the other organic acids (6 days compared with 1–2 days for other organic acids, Figures S2 and S3). In an attempt to further increase isobutyric acid concentrations, both acetic and butyric acid were added at 50 mM each. This resulted again in butyric acid consumption, albeit lower than previously observed at a net consumption of 9.62 ± 0.98 mM, with isobutyric acid production up to similar concentrations as the condition with only butyric acid (net production of 18.71 ± 3.15 mM). The discrepancy between the butyric acid consumed and isobutyric acid produced shows that isobutyric acid might not only be produced from the added exogeneous butyric acid, but part of its production must also come from consumed H₂ and CO₂ and/or the initially supplied acetic acid. Nonetheless, the main product was still acetic acid, with a net production of 59.16 ± 1.46 mM. Alcohols were not detected in any of the tested conditions.

FIGURE 2.

Electron balance of growth of C. luticellarii at initial pH 6.5 with different organic acids (all supplied at 50 mM) in presence of H₂ and CO₂. Net production is shown on the positive y axis and net consumption is on the negative y axis. Values are calculated as absolute amounts of net consumed/produced electron equivalents (eeq.) in mol eeq. C2 = acetic acid, C4 = butyric acid, iC4 = isobutyric acid.

The effect of pH on autotrophic growth of C. luticellarii

In some acetogens, mildly acidic pH stimulates the production of alcohols and longer chain carboxylic acids (Vees et al., 2020; Worden et al., 1991). Therefore, the effect of pH on the growth of C. luticellarii was assessed first without any carboxylic acid additions and with the addition of 50 mM acetic acid The pH values tested were 5.5 and 6.5 and were controlled throughout the experiments between 5.0–5.5 and 6.0–6.5, respectively. These experiments were conducted using a higher inoculum ratio (25% instead of 10%) to minimize the duration of the lag phase due to a potentially suboptimal pH for growth (Petrognani et al., 2020). However, this affected the initial OD and concentrations of products so caution is advised in comparing this data to those of the other experiments.

The autotrophic growth of C. luticellarii at pH 5.5 without any additions showed a similar maximum achieved OD (0.24 ± 0.02) compared to pH 6.5 (0.24 ± 0.03), but a lower total H₂ and CO₂ consumption (8.74 ± 0.69 mmol H₂ and 3.10 ± 0.29 mmol CO₂ at pH 5.5 compared with 15.03 ± 1.47 mmol H₂ and 5.95 ± 0.37 mmol CO₂ at pH 6.5) (Figure 3). Additionally, butyric and isobutyric acid production started earlier at pH 5.5 and reached a higher concentration of, respectively, 2.36 ± 0.51 mM and 2.79 ± 0.70 mM (compared to 2.22 ± 0.55 mM butyric acid and 1.98 ± 0.40 mM isobutyric acid at pH 6.5) (Figure S5). The final global C4 (C4 + iC4) selectivity was 27 ± 9% at pH 5.5 compared to 13 ± 2% at pH 6.5, indicating low pH favoured chain elongation. Growth at pH 5.5 with 50 mM initial acetic acid concentrations resulted in a final OD of 0.22 ± 0.02. The total amount of gas consumed (10.21 ± 0.86 mmol H₂ and 3.74 ± 0.34 mmol CO₂) was similar to growth at pH 5.5 without initial acetic acid, but the selectivity of C4 products (butyric and isobutyric acid combined) was higher at 42 ± 4% indicating the effect of pH and elevated acetic acid concentrations is cumulative (Figure 3). The selectivity for C4 compounds at pH 5.5 with 50 mM acetic acid was, however, slightly lower than at pH 6.5 with 50 mM acetic acid (53 ± 3%) as described in The effect of acetic and butyric acid on autotrophic isobutyric acid production although those experiments were done at a lower inoculum ratio and uncontrolled pH which may have affected the outcome.

FIGURE 3.

Electron balance of growth of C. luticellarii on H₂ and CO₂ at pH 6.5 and pH 5.5 with and without acetic acid (C2, 50 mM). Net production is shown on the positive y axis and net consumption is on the negative y axis. Values are calculated as absolute amounts of net consumed/produced electron equivalents (eeq.) in mol eeq. *This condition was run without pH control and at inoculum ratio 10% as depicted in Figure 2. The other conditions presented were done at controlled pH and inoculum ratio 25%. C₂ = acetic acid.

Impact of the additions of valeric and caproic acid on the metabolism of C. luticellarii

The production of isobutyric acid observed during growth at elevated butyric acid concentrations strongly suggests the occurrence of bio‐isomerisation. To assess whether C. luticellarii is also capable of forming isomers of other linear carboxylic acids such as valeric and caproic acid, C. luticellarii was grown in presence of either 25 mM of valeric acid or 25 mM of caproic acid (Figure S4). Neither additions resulted in production of their respective isomers. However, C. luticellarii did consume 6.85 ± 1.31 mM of valeric acid with concurrent production of 5.83 ± 0.51 mM of propionic acid. In the valeric acid experiment, its consumption accounted for 25 ± 3% of the total electrons consumed (H₂ and valeric acid combined) and H₂ and valeric acid were consumed simultaneously, showing that C. luticellarii grew mixotrophically. The main product was acetic acid at a final concentration of 66.91 ± 7.44 mM (Figure 4, 25 mM). Caproic acid addition at 25 mM led to a significantly lower consumption of H₂ and CO₂ which was only used for the production of 50.16 ± 1.31 mM acetic acid (Figure S4).

FIGURE 4.

Electron balance of growth of C. luticellarii at different initial valeric acid concentrations in presence of H₂ and CO₂. Net production is shown on the positive y axis and net consumption is on the negative y axis. Values are calculated as absolute amounts of net consumed/produced electron equivalents (eeq.) in mol eeq.

The near‐equimolar nature of the conversion of valeric acid to propionic acid points at valeric acid being oxidized via the β‐oxidation pathway even though hydrogen partial pressures here (> 100 kPa) exceed the limits for organic acid oxidation commonly reported in literature related to chain elongation and/or anaerobic digestion by several orders of magnitude (approx. 1–300 Pa (Ge et al., 2015; Stams, 1994)). To gain further insight into the oxidation of valeric acid under high H₂ partial pressures, C. luticellarii was grown under increasing concentrations of valeric acid (Figure 4, Figure S6). At initial concentrations of 10, 50 and 75 mM valeric acid, respectively, 4.95 ± 0.02 mM, 5.52 ± 0.28 mM and 4.84 ± 1.22 mM of valeric acid was consumed and 4.28 ± 0.08 mM, 5.82 ± 0.12 mM and 4.68 ± 0.34 of propionic acid was produced (Figure 4). The initial concentration of valeric acid thus did not have a large effect on the amount of valeric acid consumed, but consumption of valeric acid and production of propionic acid again occurred in nearly equimolar amounts in all cases. Additionally, valeric acid exerted a toxic effect on C. luticellarii from 50 mM onwards since the gas uptake decreased from 20.25 ± 1.28 mmol H₂ and 8.28 ± 0.69 mmol CO₂ at 10 mM valeric acid to 12.62 ± 1.23 mmol H₂ and 5.28 ± 0.56 mmol at 75 mM valeric acid, and the maximum OD decreased from 0.17 ± 0.01 at 10 mM valeric acid to 0.08 ± 0.00 at 75 mM valeric acid.

A thermodynamic analysis of organic acid oxidation was performed to assess why and how C. luticellarii is able to oxidize valeric acid to propionic acid and acetic acid under an atmosphere with elevated H₂ partial pressure. Two cases were calculated: (1) Oxidation of valeric acid assuming that protons are used as final electron sink to produce H₂ and (2) Oxidation of valeric acid with using intracellular electron carriers as electron acceptors, whose electrons are internally reused to, for example, fuel the WLP. For both cases, the energy conservation mechanism considered is the production of one ATP through substrate‐level phosphorylation per molecule of valeric acid oxidized, thus a global Gibbs free energy (ΔG′) of the process of −50 kJ/mol valeric acid was used as threshold for process thermodynamic feasibility (Buckel & Thauer, 2013; González‐Cabaleiro et al., 2015). In Case 1, the oxidation with co‐production of H₂ at pH 6.5 and 120 kPa of H₂ has a ΔG′ of +25 kJ/mol valeric acid deeming it thermodynamically unfeasible. The reaction only becomes feasible at pH 6.5 at extremely low H₂ partial pressures (<6 × 10⁻² Pa). The pathway does become more favourable with increasing pH, but never surpasses the threshold of thermodynamic feasibility within the range of pH 5–7 (Figure 5A). However, assuming case 2, the reaction is feasible at pH 6.5 and under 120 kPa of H₂ with a ΔG′ of −59.5 kJ/mol valeric acid and is thus also exergonic enough to allow for ATP production. The thermodynamic feasibility of Case 2 is also affected by pH, decreasing its feasibility with decreasing pH, and becomes unfavourable at pH values below approximately 5.7 (Figure 5A). Since the calculations that led to this value hinge on several assumptions (e.g. intracellular metabolite concentrations, reduced:oxidized electron carrier ratios, etc.), the value should be taken as an indication and not an absolute cut‐off.

FIGURE 5.

(A) Thermodynamic calculations of valeric acid oxidation comparing the transfer of electrons to carriers that can be regenerated in the WLP, with the regeneration of electron carriers via H₂ production. (B) The effect of pH on valeric acid oxidation. Net production is shown on the positive y axis and net consumption on the negative y axis. Values are calculated as absolute amounts of net consumed/produced electron equivalents (eeq.) in mol eeq.

To validate the conclusions of the thermodynamic calculations on the effect of pH on valeric acid oxidation, C. luticellarii was grown with 25 mM valeric acid at controlled pH 6.5 and 5.5 (Figure 5B and Figure S7). The pH had a substantial effect on the fraction of valeric acid oxidized. At pH 5.5, 9 ± 3% of the initially supplied valeric acid was consumed, while at pH 6.5 this fraction was 3 times higher at 31 ± 3%. Taking into account the total substrate consumption, valeric acid consumption represented 25 ± 4% of the electrons consumed (H₂ and valeric acid combined) at pH 6.5, whereas this was 14 ± 4% at pH 5.5. Similar trends were observed for propionic acid production, which was higher at pH 6.5. This suggests oxidation still occurred to some extent, but was less favourable at low pH as predicted by the thermodynamic analysis. To gain further insight into the oxidation, C. luticellarii was incubated in presence of 25 mM valeric, but absence of H₂ by flushing with a 90:10 N₂:CO₂ gas mixture. No growth was observed under these conditions and valeric acid was not consumed (Figure S8).

DISCUSSION

Clostridium luticellarii requires complex nutrients and a low‐shear environment to grow on H₂ and CO₂

While C. luticellarii was capable of growing autotrophically, it required tryptone or yeast extract and static conditions to grow reproducibly across sub cultures. The necessity of complex nutrients (e.g. yeast extract and/or tryptone) to enable autotrophic growth of acetogens has been reported earlier. For example, Martin et al. (2016) tested the dependence of six different autotrophic strains of acetogens on yeast extract and only managed to grow three of them without a complex nutrient source. This can be connected to the fact that the metabolism of acetogens often runs close to the thermodynamic limit of life due to the low availability of energy under autotrophic conditions, potentially also impacting their capacity to synthesize all cell constituents de novo (Schuchmann & Müller, 2014). Therefore, many studies focusing on medium development for acetogens and studies characterizing the autotrophic growth of acetogens have used yeast extract as a complex nutrient source providing vitamins, trace elements and amino acids (Arslan et al., 2019; Chang et al., 2007; Groher & Weuster‐Botz, 2016a, 2016b; Litty & Müller, 2021). However, regardless of the source of yeast extract, at least 7%–13% of it is composed of readily degradable carbon sources such as carbohydrates (Tomé, 2021), which can be converted to organic acids. This may distort the results of some studies where, comparatively, the electron equivalents potentially stemming from the carbohydrates in yeast extract are higher than the electron equivalents coming from the actual substrates under study. We advocate the use of moderate concentrations of tryptone over yeast extract whenever possible in studies characterizing production capacities of microorganisms since tryptone is a lean source of amino acids that does not usually support growth in absence of other substrates as shown in Clostridium luticellarii requires strict conditions for autotrophic growth. Additionally, where the use of yeast extract is inevitable, a close eye should be kept on whether electron and carbon balances close (i.e. equal amounts of carbon and electrons consumed as produced) to minimize the interference of yeast extract on the studied metabolism. Here, we chose to supplement growth media with 0.5 g L⁻¹ tryptone instead of 1 g L⁻¹ yeast extract, and managed to reproducibly grow C. luticellarii without any addition of carbohydrates (Figure 1).

Results also indicated that C. luticellarii consumes gas much slower and to a lesser extent under shaking conditions than when statically incubated. This result is counterintuitive, since shaking increases the gas–liquid mass transfer rate and thus the substrate availability for the bacteria but it grows better under passive diffusion. One potential hypothesis may be that C. luticellarii is sensitive to the shear stress induced by shaking conditions. Although this is not uncommon in anaerobic bacteria, the underlying mechanism remains unclear (Jonczyk et al., 2013). Scaling up gas fermentation with shear sensitive C. luticellarii may prove a challenge since classic approaches such as CSTRs may introduce too much shear due to mixing, and bubble column or gas lift reactors may induce too much shear due to bubble flow. Attractive alternatives are low‐shear reactor designs such as the bubble‐less hollow fibre membrane biofilm reactors or the exclusive use of CO₂‐derived liquid molecules (e.g. methanol) (Elisiário et al., 2021; Petrognani et al., 2020).

Autotrophic isobutyric and butyric acid production is stimulated by elevated acetic acid concentrations

Autotrophic growth of C. luticellarii at pH 6.5 was mainly characterized by acetic acid production, which is the dominant product in the majority of acetogens (Bengelsdorf et al., 2018), accompanied by traces of butyric and isobutyric acid. These results are similar to those reported for other acetogens capable of producing traces of butyric acid when grown on H₂ and CO₂, such as B. methylotrophicum (Lynd & Zeikus, 1983), E. limosum KIST612 (Litty & Müller, 2021), B. hydrogenotrophica DSM 10507, C. magnum DSM 2767 and E. aggregans DSM 12183 (Groher & Weuster‐Botz, 2016a). It also aligns well with the findings of González‐Cabaleiro et al. (2013), where chain elongation to butyric acid with H₂ as electron donor was found to be kinetically bottlenecked in the first step of the reverse β‐oxidation pathway due to the unfeasibly low internal acetoacetyl‐CoA levels required to drive the condensation reaction of two acetyl‐CoA molecules.

Increasing the initial acetic acid levels from 0 to 50 mM increased the selectivity of C4 compounds (sum of butyric and isobutyric acid) from 5 ± 1% to 53 ± 3%. Similar selectivity shifts towards butyric acid at elevated levels of acetic acid have been reported for methanol‐based studies with C. luticellarii, but also with E. limosum and B. methylotrophicum (Lynd & Zeikus, 1983; Pacaud et al., 1986; Petrognani et al., 2020). It has been hypothesized that high acetic acid concentrations exert a feedback inhibition on its production, thereby redirecting carbon flux from acetyl‐CoA to reverse β‐oxidation instead of acetic acid (Kremp & Müller, 2020), but it remains unclear how this alleviates the kinetic bottleneck proposed by González‐Cabaleiro et al. (2013).

Isobutyric acid is likely formed by a selective isomerization of butyric acid

In our experiments, isobutyric and butyric acids were co‐produced during the autotrophic growth of C. luticellarii at pH 5.5 and at elevated concentrations of acetic acid (Figure 2). Additionally, autotrophic growth at elevated butyric acid concentrations resulted in its consumption and the concomitant production of isobutyric acid at nearly equimolar amounts. This aligns with earlier work by Petrognani and co‐workers, who showed that isobutyric acid production from methanol and CO₂ is stimulated by elevated butyric acid concentrations, suggesting an interconversion of both compounds (Petrognani et al., 2020). The most plausible hypothesis is that C. luticellarii isomerizes butyric acid to isobutyric acid. This mechanism has been suggested before for C. luticellarii growing on methanol and CO₂ by Petrognani et al. (2020) and by Liu et al. (2020), who isolated a closely related Clostridium sp. BL3 producing butyric and isobutyric acid from lactic acid. However, neither study reported net butyric acid consumption alongside isobutyric acid production to fully support this hypothesis. The suggested pathway in these studies consists of isomerization of n‐butyryl‐CoA (coming from n‐butyric acid activation or from in situ production through reverse β‐oxidation) to isobutyryl‐CoA by a butyryl‐CoA:isobutyryl‐CoA mutase (BM) and final conversion of isobutyryl‐CoA to isobutyric acid by an isobutyryl‐CoA:acetate CoA transferase (Liu et al., 2020; Petrognani et al., 2020). Homologues for all enzymes involved are present in the C. luticellarii genome (Liu et al., 2020). All steps of the isomerisation pathway seem to be reversible, as supported by the observed consumption of isobutyric acid alongside production of butyric acid at nearly equimolar amounts (Figure 2 and Figure S3). However, C. luticellarii showed a much longer lag phase when grown on isobutyric acid compared to butyric acid, which is counterintuitive assuming the same enzymes are used in both directions. The mechanism behind the long lag phase thus remains unclear.

To check whether C. luticellarii could also convert valeric acid or caproic acid to their respective isomers, autotrophic growth at elevated concentrations of valeric acid and caproic acid was tested. Neither of the added organic acids resulted in occurrence of their respective isomers. This aligns with the myriad of chain elongation studies that reported significant iso‐butyric acid production (but not iso‐caproic acid production) in mixed and pure culture systems engineered for butyric and caproic acid production (Huang et al., 2020; Liu et al., 2020; Mariën et al., 2022), and the lack of studies reporting significant iso‐valeric acid production in similar studies aiming for valeric acid production (Allegue et al., 2022; de Smit et al., 2019; Ganigué et al., 2019). This could be either due to the (iso‐)acids not being transported through the cell membrane, the acids not being activated to their respective CoA form, and/or the BM of C. luticellarii being specific for (iso)butyric acid. The transport into the cell and activation of the acid is likely not preventing production of isovaleric acid (or isomers of other carboxylic acids), since oxidation of valeric acid to propionic and acetic acid was observed, for which valeric acid needs to enter the cell and requires activation to valeryl‐CoA. The likely bottleneck is the specificity of the BM enzyme since several studies report acyl‐CoA mutases to have a strict substrate specificity (Cracan & Banerjee, 2012). Purification and characterization of the C. luticellarii BM could further prove this hypothesis.

The ecological reason behind the interconversion of butyric acid to isobutyric acid (and vice versa) remains obscure since no free energy is released during the isomerization. Liu et al. (2020) and Allison (1978) argued that bacteria may convert n‐butyric acid to isobutyric acid to maintain a high isobutyric acid pool for the synthesis of valine in media poor in amino acids. Here, the medium C. luticellarii was grown on contains 0.5 g L⁻¹ tryptone, providing a readily accessible source of amino acids (albeit at a low concentration). This suggests that either the mechanism may not be directly regulated by amino acid availability or another ecological reason is behind the mechanism. Others have suggested that isomerization is a way to deal with high concentrations of toxic n‐butyric acid by converting it to less toxic isobutyric acid, and thus allowing microorganisms to continue generating energy through butyric acid production (Chen et al., 2017). The production of isobutyric acid was not only observed here at high butyric acid concentrations but was also observed at low butyric acid concentrations (<10 mM) concomitantly with butyric acid production, casting doubt on the toxicity hypothesis. Additionally, neither of the proposed hypotheses explains the benefit of reverse isomerisation (i.e. isobutyric acid to butyric acid). Exploration of the proteome and metabolome of C. luticellarii under different environmental conditions may provide further answers as to why this isomerisation occurs.

pH drives metabolic flexibility between the reductive and oxidative direction of the autotrophic reverse β‐oxidation pathway

In our work, pH was identified to be a controlling parameter between reductive and oxidative processes in C. luticellarii: (1) circumneutral pH (~6.5) increased the fraction of valeric acid oxidized, and (2) mildly acidic pH (≤5.5) increased the selectivity for butyric and isobutyric acid production by stimulating reverse β‐oxidation. While both processes are seemingly different, they are in fact the same pathway running in opposite directions.

Thermodynamic calculations and the results presented in this article show that C. luticellarii is unequivocally able to oxidize valeric acid to acetic and propionic acid at circumneutral pH under a hydrogen partial pressure of 120 kPa, contradicting at first glance established literature reporting the blockage of organic acid oxidation at H₂ partial pressures as low as 1 Pa (Ge et al., 2015; Stams, 1994). Organic acid oxidation is usually considered to be coupled with H₂ formation. This implies that the NAD⁺ reduced to NADH (NAD⁺/NADH E′ = −280 mV (Buckel & Thauer, 2013)) during oxidation must be reoxidised with protons as electron sink and hence the production of H₂, which is thermodynamically unfeasible at high H₂ partial pressures (H⁺/H₂ E′ = −340 mV at 250 Pa H₂ which is even considerably lower than H₂ pressures used in this study (Schuchmann & Müller, 2014; Stams, 1994)). A similar case of oxidation in seemingly unfavourable conditions has been reported in the model chain elongator C. kluyveri. For years, scientists had hypothesized on how C. kluyverii is able to oxidize ethanol (which yields NADH) during reverse β‐oxidation given that the strain internally accumulates H₂ up to a partial pressure of 1 bar (Li et al., 2008; Seedorf et al., 2008). The answer to this question was that NADH can be reoxidised in the reverse β‐oxidation pathway without generation of H₂ (Li et al., 2008; Seedorf et al., 2008). We argue that C. luticellarii may be able to carry out organic acid oxidations in a similar way where NAD⁺ is likely regenerated by NADH oxidation in one of the reducing steps of the methyl branch in the WLP (i.e. reduction of CO₂ to formate, or the sequential reductions of methenyl‐THF to methylene‐THF and methyl‐THF) and not by H₂ production. However, currently very little is known about the electron carriers used in the Wood–Ljungdahl Pathway of C. luticellarii, making it difficult to prove this hypothesis at this point. Further characterization of the cofactors used in the individual reduction steps would allow performing a thermodynamic analysis over the entire metabolism (i.e. combination of the Wood–Ljungdahl pathway and organic acid oxidation) and elucidating the mechanism. Oxidation of organic acids simultaneously with autotrophic metabolism may also offer another advantage to C. luticellarii by providing an efficient alternative for ATP generation from reduced ferredoxin, which is produced during H₂ oxidation. During autotrophic metabolism, acetogens usually generate ATP from the excess reduced ferredoxin produced during H₂ oxidation by oxidizing it in the Ferredoxin:NAD⁺ oxidoreductase complex (RnF), which uses the released energy to generate an ion motive force that drives ATP synthesis via an ATPase (Schuchmann & Müller, 2014). The oxidation of valeric acid offers an alternative to this because the enzyme complex catalysing the oxidation of valeryl‐CoA needs reduced ferredoxin in an electron confurcation process, where electrons from reduced ferredoxin and valeryl‐CoA reduce NAD⁺ to NADH. Hence, the investment of two reduced ferredoxin (assumed monovalent here) in this case yields 3 NADH and one full ATP (Equation 5), while the classic RnF–ATPase route only produces one NADH and ~0.50–0.67 ATP from the same ferredoxin investment (Schuchmann & Müller, 2014). In this sense, we argue that those acetogens with substrate flexibility (i.e. the ability to (co‐)oxidize organic acids even at high H₂ partial pressures) may have a competitive advantage against those who cannot, since they may use more efficiently the available reduced ferredoxin and conserve more energy in the form of ATP. The use of valeric acid oxidation as an alternative way to regenerate ferredoxin is also supported by the fact that valeric acid is not oxidized in absence of H₂, since in that case there is no direct source of reduced ferredoxin. This suggests that valeric acid oxidation is a secondary metabolism, depending on other substrates (here H₂ that drives the WLP) to support it. However, reduced ferredoxin could theoretically be indirectly generated through investing ~0.5–0.67 of the ATP formed in valeric acid oxidation via the reversal of the RnF–ATPase route, thereby consuming one NADH for the production of two reduced ferredoxin. In this way, valeric acid could be oxidized with the remaining two NADH being reoxidized to NAD⁺ through H₂ production, maintaining a final ATP yield of 0.33–0.5 ATP per valeric acid oxidized. It is unclear why C. luticellarii is unable to do so to sustain growth in absence of H₂.

While the experiments presented here focused on valeric acid oxidation for which the degradation product propionic acid is easily detectable, a similar case can be made for the oxidation of other organic acids such as butyric acid. However, we were unable to prove butyric acid oxidation here due to (i) its degradation product being acetic acid which is also the main metabolic end product from the WLP, and (ii) butyric acid is also consumed for the production of isobutyric acid. Both processes occurring simultaneously with butyric acid oxidation make it difficult to accurately follow the carbon flux, hence, follow‐up studies using isotope‐labelled butyric acid may shed further light on this.

The thermodynamic analysis identified pH as key parameter in Impact of the additions of valeric and caproic acid on the metabolism of C. luticellarii : at mildly acidic pH (<~5.5), C. luticellarii shifts its metabolism towards producing more of the chain elongation products butyric and isobutyric acid. Similar shifts towards chain elongation products have been reported by Worden et al. (1991) who observed a four‐fold increase in butyric acid production from CO by Butyribacterium methylotrophicum when shifting pH from 6.8 to 6.0. To explain the effect of pH on chain elongation and oxidation, we suggest the following hypothesis (Figure 6): at circumneutral pH (~6.5), C. luticellarii can leverage its substrate flexibility to oxidize H₂ and organic acids (e.g. butyric or valeric acid) simultaneously, thereby generating reduced electron carriers and allowing a higher ATP yield than when oxidizing H₂ alone due to the mechanism explained above. As long as pH remains high enough (>~5.5), C. luticellarii can run the Wood–Ljungdahl Pathway and organic acid oxidation simultaneously. At mildly acidic pH (≤~5.5), the oxidation of organic acids becomes less favourable and upon surpassing a thermodynamic threshold, the oxidation reverses and operates in the reductive direction (i.e. chain elongation) via reverse β‐oxidation (i.e. condensing and reducing two acetyl‐CoA molecules to produce butyric and isobutyric acid). This process represents a net consumption of electron equivalents coming from the oxidation of H₂ and is driven by the production of reduced ferredoxin (via electron bifurcation in the crotonyl‐CoA reduction) which can be used for additional ATP generation through its oxidation in the RnF complex. This remarkable metabolic flexibility shows how adapted C. luticellarii is to operate at the limit of thermodynamic viability since it is able to reverse the direction of carbon flux in its metabolism to generate additional ATP from the products of its autotrophic metabolism depending on the prevailing environmental conditions. Additionally, to the best of our knowledge this represents the first report of an acetogen with the capacity to use the (reverse) β‐oxidation in opposite directions according to a change of pH while conserving energy regardless of the direction.

FIGURE 6.

The hypothetical direction of electron flow in the metabolism of C. luticellarii driven by changes in pH. (A) C. luticellarii oxidizes hydrogen to generate reducing equivalents that contribute to the reduction of the electron carrier pool. (B) Reduced electron carriers are used in the WLP and for energy generation. The WLP serves as the central carbon fixation pathway regardless of pH. (C) Whenever pH is high enough (>~5.5) and butyric acid (n = 4) or valeric acid (n = 5) is present, they may be oxidized in parallel with the WLP to generate additional electron equivalents that contribute to the reduction of the electron carrier pool, as well as 1 ATP per organic acid oxidized. (D) When pH is low (≤~5.5) oxidation of organic acids is thermodynamically unfavourable and the pathway reverses to the direction of reverse β‐oxidation, now becoming an electron‐consuming pathway instead of an electron‐generating pathway. Fd, ferredoxin; Hdr, hydrogenase.

CONCLUSIONS

This study investigated the autotrophic growth of C. luticellarii by identifying the necessary growth conditions and examining the effects of metabolite levels and pH on butyric and isobutyric acid titers and selectivity. Our experiments concluded that C. luticellarii requires the addition of complex nutrient sources and the absence of shaking conditions for autotrophic growth. To steer product selectivity, pH was identified as a key parameter governing the direction of metabolic fluxes. At circumneutral pH (>5.5), acetic acid is the sole metabolic end product and, additionally, C. luticellarii shows the particular ability to co‐oxidize organic acids such as valeric acid under high H₂ partial pressures (>1 bar). Conversely, mildly acidic pH (<5.5) stimulates the production of butyric and isobutyric acid. Additionally, elevated acetic acid concentrations stimulated butyric and isobutyric acid production up to a maximum combined selectivity of 53 ± 3%. Finally, our results suggest that isobutyric acid is produced by a reversible isomerization of butyric acid, but valeric and caproic acid are not isomerized by C. luticellarii. While this study positions C. luticellarii and its uniquely flexible metabolism producing butyric and isobutyric acid from H₂ and CO₂ as a promising candidate for industrial biotechnological applications, further research should focus on improving yields and production rates on reactor‐scale. Overall, the combined insights reported here can inform the future efforts to optimize and scale‐up the production of valuable chemicals from CO₂ using C. luticellarii.

AUTHOR CONTRIBUTIONS

Quinten Mariën: Conceptualization (lead); data curation (lead); formal analysis (lead); funding acquisition (lead); investigation (lead); methodology (lead); validation (lead); visualization (lead); writing – original draft (lead). Alberte Regueira: Conceptualization (supporting); investigation (supporting); methodology (supporting); supervision (supporting); writing – review and editing (equal). Ramon Ganigué: Conceptualization (supporting); funding acquisition (supporting); methodology (supporting); resources (lead); supervision (lead); writing – review and editing (equal).

CONFLICT OF INTEREST STATEMENT

The authors declare to have no conflict of interest.

Supporting information

Data S1.

Mariën, Q. , Regueira, A. & Ganigué, R. (2024) Steerable isobutyric and butyric acid production from CO₂ and H₂ by Clostridium luticellarii . Microbial Biotechnology, 17, e14321. Available from: 10.1111/1751-7915.14321

DATA AVAILABILITY STATEMENT

All data can be made available upon request to the corresponding author.