Metabolic Response of Clostridium ljungdahlii to Oxygen Exposure

Abstract

Clostridium ljungdahlii is an important synthesis gas-fermenting bacterium used in the biofuels industry, and a preliminary investigation showed that it has some tolerance to oxygen when cultured in rich mixotrophic medium. Batch cultures not only continue to grow and consume H₂, CO, and fructose after 8% O₂ exposure, but fermentation product analysis revealed an increase in ethanol concentration and decreased acetate concentration compared to non-oxygen-exposed cultures. In this study, the mechanisms for higher ethanol production and oxygen/reactive oxygen species (ROS) detoxification were identified using a combination of fermentation, transcriptome sequencing (RNA-seq) differential expression, and enzyme activity analyses. The results indicate that the higher ethanol and lower acetate concentrations were due to the carboxylic acid reductase activity of a more highly expressed predicted aldehyde oxidoreductase (CLJU_c24130) and that C. ljungdahlii's primary defense upon oxygen exposure is a predicted rubrerythrin (CLJU_c39340). The metabolic responses of higher ethanol production and oxygen/ROS detoxification were found to be linked by cofactor management and substrate and energy metabolism. This study contributes new insights into the physiology and metabolism of C. ljungdahlii and provides new genetic targets to generate C. ljungdahlii strains that produce more ethanol and are more tolerant to syngas contaminants.

INTRODUCTION

Clostridium ljungdahlii is an anaerobic, motile, endospore-forming, Gram-positive, rod-shaped, acetogenic bacterium isolated from chicken yard waste and has application as a biocatalyst to transform syngas components (CO, CO₂, and H₂) into more valuable chemicals (1–4). It was the first bacterium discovered to metabolize these syngas components and to produce ethanol with acetate as its primary fermentation product (1, 4). To reduce the amount of acetate and increase the amount of ethanol produced by C. ljungdahlii for its use in industrial solvent production, different reactor designs and agitation parameters, varied syngas component concentrations, increased gas flow rates and pressures, reduced nitrogen sources, addition of reducing agents to media, adjustment of growth medium pH, and addition of nanoparticles have all been evaluated (1, 5–15). Some of these efforts have reportedly improved ethanol/acetate product ratios from 1:20 to 2:1 in batch cultures and from 1:1 to 21:1 for cultures grown in continuously stirred reactors (8, 9, 16). More recent sequencing and genetic modification techniques have greatly enhanced our understanding of C. ljungdahlii's metabolism and increased production of ethanol as well as enabled production of other chemicals (e.g., acetone, butyrate, and butanol) (2, 3, 17–21).

Despite these advancements, for C. ljungdahlii to be effectively used for industrial syngas transformation, some of its catalytic limitations related to gaseous headspace composition still require evaluation. Although steel mill waste gas was recently used as the sole carbon and energy source in a C. ljungdahlii fermentation, artificial syngas is commonly used in research experiments to limit the variability of results (18). Artificial syngas does not typically contain contaminants common to industrial syngas streams, such as oxygen (O₂), sulfurous species (COS, SO₂, and H₂S), nitrogenous species (such as HCN, NH₃, and nitrogen oxide [NOx]), methane (CH₄), and tars (22, 23). Several studies have been performed to evaluate the effect of these syngas contaminants on Clostridium spp.; however, inhibitory compounds in syngas have slowed the progress of companies moving into the commercial phase of syngas-based fermentation product development (24–26; http://www.biofuelsdigest.com/bdigest/2014/09/05/on-the-mend-why-ineos-bio-isnt-reporting-much-ethanol-production/).

Researchers have observed that H₂S concentrations up to 2.7% do not significantly affect substrate (H₂ and CO) uptake by C. ljungdahlii, and growth is maintained with concentrations as high as 5.2% (7, 9). In a related bacterium, Clostridium carboxidivorans P7^T, concentrations of NOx above 0.0040% in syngas were found to inhibit growth and noncompetitively inhibit hydrogenase activity (27, 28). Although NOx inhibition resulted in higher ethanol production by C. carboxidivorans P7^T, hydrogenase activity was inhibited, reducing available carbon for product formation since electrons came from CO rather than H₂ (27, 28). NH₃ is a nitrogen source for syngas-fermenting bacteria, but in high concentrations from a continuous syngas feed, it can inhibit cell growth and decrease acetate-to-ethanol conversion of related bacterium Clostridium ragsdalei (P11) (23). In chemostat experiments with C. carboxidivorans P7^T, tars were shown to promote cell dormancy but increase ethanol/acetate production ratios (29). With C. ragsdalei (P11), up to 5% CH₄ in syngas did not affect product formation or cell growth (30). Oxygen is also a common contaminant of syngas that can have a significant impact on microbially catalyzed syngas fermentation; however, an understanding of how the clostridial catalysts manage oxygen and reactive oxygen species (ROS) has not been well examined to date (22).

Several enzymes of the Wood-Ljungdahl pathway, including hydrogenase and carbon monoxide dehydrogenase (CODH), responsible for syngas metabolism are sensitive to oxygen (31–43). Pyruvate:ferredoxin oxidoreductase (PFOR), pyruvate formate lyase (PFL), and its activating enzyme (enzymes involved in sugar metabolism) are also oxygen labile in a variety of microbes (44–50). Therefore, acetogens in general, which possess some or all of these oxygen-labile enzymes, have traditionally been classified as strict anaerobes; nevertheless, they have been isolated from different aerobic or microaerobic environments (51–53). It has been demonstrated that many of these acetogens are equipped with an assortment of oxidative stress enzymes, and some can even reduce oxygen by other mechanisms (e.g., superoxide reductase and peroxidase) (54–60). In addition to tolerating various amounts of oxygen, it has been suggested that exposing acetogens to microaerobic conditions triggers a shift in electron flow toward more reduced products, such as ethanol, lactate, H₂, and/or NH₄⁺ production instead of acetate formation (51, 59, 61). C. ljungdahlii was previously shown to grow well and coferment fructose and artificial syngas in a complex medium (5, 6, 62). Upon exposure to low concentrations of oxygen (<10%) in the headspace of batch cultures at early log phase, C. ljungdahlii continued to grow and coferment fructose and artificial synthesis gas components while showing an increase in ethanol/acetate ratios (63). These initial experiments suggested C. ljungdahlii had the mechanisms to handle oxygen contaminants and positively affect solvent production, but the response was not fully elucidated. Therefore, the objectives of this study were to characterize C. ljungdahlii's transcriptional, metabolic, and physiological responses to oxygen exposure when grown in rich mixotrophic medium, with a particular focus on the effect oxygen has on ethanol and acetate formation.

MATERIALS AND METHODS

Media and growth conditions.

C. ljungdahlii (ATCC 55383) was obtained from the American Type Culture Collection and was cultured in modified reinforced clostridial medium (mRCM) supplemented with 5 g/liter fructose (mRCMf) (5, 6). The artificial syngas composition (20% CO, 20% CO₂, and 10% H₂ with N₂ balance) that was injected into the headspace (110 ml) of batch cultures was selected based on the approximate theoretical composition of a gas stream from biomass gasification with air as the fumigator.

The medium was prepared and dispensed into 160-ml serum bottles (final volume including reducing agents and inoculum of 50 ml/bottle). Reducing agents (2.5% [wt/vol] cysteine-HCl and 2.5% [wt/vol] sodium sulfide) were added to better ensure anaerobicity of growth media. However, in some experiments, reducing agents were omitted in an effort to observe their effect on oxygen removal, protein production (as observed on SDS-PAGE gel), and benzyl viologen reduction by C. ljungdahlii cell-free cell extract. Bottles were capped with butyl rubber stoppers (Bellco, Vineland, NJ) and aluminum seals, connected to a vacuum manifold using a needle and a 0.22-μm-pore filter, and made anaerobic by cycling three times (30-s cycles) between (i) vacuum headspace evacuation and (ii) sparging with artificial syngas filtered through heated copper pellets. The bottles were then autoclaved for 30 min (121°C, 15 lb/in² gauge [psig]). After the bottles cooled to room temperature, excess pressure was released from the bottles by inserting a needle connected to a 0.22-μm-pore filter and a hose with its end immersed in a water trap to achieve a condition of atmospheric pressure. Culture bottles were initially inoculated with an aliquot from freezer stocks (10% [vol/vol]) and incubated with shaking at 100 rpm at 37°C. After 24 h, culture aliquots (5% [vol/vol]) were transferred to fresh medium and incubated (100 rpm, 37°C). After 12 h of growth, experiments were initiated with a second transfer of actively growing cells (5% [vol/vol]) into fresh medium and incubated (100 rpm, 37°C). For the oxygen exposure experiments, after 12 h of culture incubation, 8% (by headspace volume) O₂ was injected by syringe.

Gas and liquid product analysis.

Liquid samples for ethanol and acetate were taken at the 12-, 14-, 24-, 36-, 48-, and 72-h time points from the C. ljungdahlii cultures and stored (−20°C) prior to analysis. Gas samples for CO, H₂, and CO₂ and liquid samples for fructose were taken at 0, 24, 36, 42, 54, and 66 h. Fructose samples were also stored (−20°C) prior to analysis, but gas was analyzed within 1 h of sampling. Gas and liquid product analysis results are reported as an average of six replicates completed as triplicates of two biological repeats.

Liquid samples (500 μl) for ethanol and acetate analysis were prepared by adding 125 μl 25% m-phosphoric acid and centrifuging the mixture at 14,000 × g for 10 min at room temperature. The sample supernatants were analyzed by gas chromatography using an Agilent 7890A gas chromatograph containing a 0.25-μm J & W DB-FFAP column (30 m by 0.32 mm inside diameter [i.d.]) and a flame ionization detector. Argon was used as the carrier gas at a flow rate of 30 ml/min. The injector and detector temperatures were 250°C, and the oven temperature was 160°C.

Headspace gas samples were collected using a 5 ml gas-tight syringe with sample lock (SGE, Ringwood, Australia). The samples were analyzed by gas chromatography using an Agilent 7890A gas chromatograph containing a 13823 molsieve 5-Å zeolite molecular sieve packed stainless steel column (6 ft by 1/8 in. i.d. [Supelco]) and a thermal conductivity detector. Argon was used as the carrier gas at a flow rate of 30 ml/min. The injector and detector temperatures were 150°C and 250°C, respectively, and the oven temperature was 160°C.

The amount of fructose present in the culture medium was determined by high-performance liquid chromatography. Liquid samples (600 μl) were centrifuged (10 min, 14,000 × g, room temperature). Supernatant was filtered through a 0.22-μm-pore syringe filter and analyzed using a Shimadzu LC20 liquid chromatograph with a HPX-87H column (65°C) and refractive index detector. Sulfuric acid (5 mM) was used as the eluent at a rate of 0.6 ml/min.

Dissolved oxygen (DO) and dissolved CO₂ species (including bicarbonate in the liquid phase) were also quantified during batch culture fermentations. DO in the cultures exposed to 0 and 8% O₂ was measured with a NeoFox system (Ocean Optics, Dunedin, FL) as per the manufacturer's instructions at 14, 30, and 36 h to determine if oxygen in the medium was being reduced over time. To take the measurements, bottles were opened in a fume hood, the NeoFox probe was immediately lowered to the bottom of the bottles, and readings were recorded once they stabilized (less than 1 min). The DO was also measured in uninoculated bottles containing media with and without reducing agents to measure the amount of abiotic oxygen reduction; uninoculated medium was exposed to either 0 or 8% O₂. Total carbon dioxide in the medium was measured using a Bioprofile 400 analyzer (Nova Biomedical, Waltham, MA) per the manufacturer's instructions. Total CO₂ (tCO₂) was calculated using the equation tCO₂ = [HCO₃⁻] + α × pCO₂, where the solubility coefficient of CO₂ (α) is estimated to be 0.0307.

mRNA isolation, cDNA library preparation, and sequencing.

C. ljungdahlii was grown as described above in mRCMf with a syngas headspace and exposed to 0 or 8% O₂. Three culture bottles per experimental condition (0 and 8% O₂ exposed) were transferred into the anaerobic chamber at 14 h (2 h after oxygen exposure [early response]) and 36 h (12 h after oxygen exposure [late response]) time points (12 samples total). Serum bottles were opened, and 0.5 ml of culture was added to 1 ml of Qiagen RNA Protect bacterial reagent (Qiagen, Venlo, Limburg, Netherlands). These samples were centrifuged (10 min, 21,460 × g, 4°C), the supernatant was removed, and pellets were stored at −65°C. Batch cultures used for RNA sequencing were not sampled for fermentation analysis to save as much culture for total RNA isolation as possible. To ensure that the liquid fermentation products for the cultures used for RNA sequencing were similar to the other fermentations in this study, liquid analysis (as described above) was performed on other cultures that were simultaneously inoculated.

The Qiagen RNeasy minikit (Qiagen, Venlo, Limburg, Netherlands) with on-column RNase-free DNase (Qiagen, Venlo, Limburg, Netherlands) treatment was used in conjunction with the enzymatic lysis and proteinase K treatment (New England BioLabs, Ipswich, MA) per the supplier's protocol to isolate total RNA. RNA quality was checked by gel electrophoresis. The Epicentre ScriptSeq Complete kit (Bacteria) (Illumina, San Diego, CA) was used to isolate the mRNA as well as synthesize and amplify the cDNA with a different index for each sample (12 total). The Qiagen RNeasy MinElute kit (Qiagen, Venlo, Limburg, Netherlands) was used to purify the mRNA. mRNA and cDNA were analyzed for quality and concentration using an Agilent 2500 Bioanalyzer on an Agilent RNA 6000 Pico chip and DNA high-sensitivity chip, respectively, at the North Carolina State Genomic Sciences Lab (Raleigh, NC). Indexed cDNA was pooled and subjected to bioanalysis prior to sequencing at the BGI and Children's Hospital of Philadelphia collaborative genome facility using the 100-bp single-read protocol for one lane on the Illumina HiSeq2000. Base quality calls and demultiplexing were performed with the CASAVA 1.8.2 pipeline (Illumina, San Diego, CA).

Differential expression analysis.

Differential analysis was performed with DEGseq as described by Tan et al., with the exception of raw reads being counted with eXpress prior to importing into DEGseq (64, 65). Unless otherwise stated, genes mentioned in this paper were significantly differentially expressed based on a false discovery rate (FDR) (66) of ≤0.001 and a normalized log₂ fold change of ≥2 or ≤−2. Tablet software was used to visualize differentially regulated gene clusters (67, 68).

Nucleotide sequence alignment and database search.

Reannotation of differentially expressed genes was performed with the blastx program (National Centre of Biotechnology Information [http://www.ncbi.nlm.nih.gov/BLAST]) using the default parameters. Multiple sequence alignments were performed with Vector NTI using the standard parameters (69).

Enzyme assays.

As with cultures for transcriptome sequencing (RNA-seq) analysis, C. ljungdahlii cultures for enzyme assays were treated with either 0 or 8% O₂ at 12 h, and samples for assays were taken at 14 h and 36 h for early and late responses to oxygen exposure. At these time points, cultures were transferred to an anaerobic chamber and dispensed into centrifuge bottles (six 50-ml cultures per bottle). The bottles were sealed and centrifuged (25 min, 11,000 × g, 4°C). The cell pellets were washed with 40 ml anaerobic potassium phosphate buffer (50 mM, pH 7.0) and centrifuged as described above. Cell pellets were resuspended in 3 ml of anaerobic potassium phosphate buffer (50 mM, pH 7.0) per g of wet cell pellet. Five hundred microliters of the cell suspension was transferred to 2-ml screw-cap tubes. The cell suspensions were either processed immediately for assays or stored at −65°C.

For cell suspension processing, 500 μl of 0.1-mm-diameter disruption glass beads (RPI, Mt. Prospect, IL) was added to the tubes. Samples were vortexed horizontally (10 min, maximum speed) using a MoBio vortex adapter 13000-V1 for Vortex-Genie 2 (MoBio, Carlsbad, CA) and centrifuged (10 min, 21,000 × g, 4°C). Supernatants were collected and kept on ice for assays. All assays were carried out in a 2-ml total volume (once all reagents were added) at 25°C in anaerobic cuvettes, except peroxidase and oxidase, which had a total volume of 3 ml, and the superoxide dismutase (SOD) and xanthine oxidase assays, which were performed aerobically in 96-well plates. All enzyme assay results are reported as an average from six replicates.

NAD(P)H oxidase was assayed as described by Stanton and Jensen (70). For a negative control, the assay was performed in anaerobic cuvettes with anoxic buffer and headspace. Reactions were initiated by adding cell extract (10 to 100 μg protein). Catalase was assayed by measuring the disappearance of H₂O₂ at 240 nm (71). The catalase reactions were initiated by adding cell extract (10 to 100 μg protein). Peroxidase was assayed as described by Poole and Ellis, except that 200 μM NADH or NADPH (final concentration) was used (72). Superoxide dismutase (SOD) activity was determined using a commercially available water-soluble tetrazolium (WST) salt SOD assay kit (Dojindo Laboratories, Kumamoto, Japan) per the manufacturer's instructions (73). Superoxide dismutase from bovine erythrocytes (MP Biomedicals, Santa Ana, CA) was used as a positive control. Xanthine oxidase activity was assayed with the Amplex Red xanthine/xanthine oxidase assay kit (ThermoFisher Scientific, Waltham, MA) per the manufacturer's instructions. Xanthine was omitted from the reaction mixture as a negative control.

Alcohol dehydrogenase assay mixtures contained 1.5 ml 100 mM Tris-HCl (pH 8.5), 0.5 ml 2 M ethanol, and 1 ml of 0.025 M NAD⁺ or NADP⁺. Reactions were initiated by adding cell extract (10 to 100 μg protein), and the appearance of NADH or NADPH was measured at a wavelength of 340 nm. Acetaldehyde dehydrogenase assays were performed as described in reference 74. The reverse reaction of aldehyde oxidoreductase (AOR), carboxylic acid reduction (CAR), was performed as described in reference 75 to measure acetate reduction, except that the assay was performed at 25°C in 500 mM potassium phosphate (pH 6.0) and the electron donor was methyl viologen (140 μM), which was completely reduced by dithionite (150 μM). The forward reaction, acetaldehyde-dependent reduction of benzyl viologen (also known as acetaldehyde oxidase), was also measured. Sodium dithionite was not used for the forward reaction. Unreduced cultures were also used as a control to avoid reduction of benzyl viologen by the reducing agents rather than by the cell extracts.

For all assays, activities (units per milligram) are defined as micromoles of substrate [e.g., methyl viologen, benzyl viologen, NAD(P)⁺, or NAD(P)H] oxidized or reduced per minute. Methyl viologen oxidation and benzyl viologen reduction were quantified by measuring their absorbance at 600 nm and using a molar extinction coefficient of 12,000 M⁻¹ cm⁻¹ or 7,400 M⁻¹ cm⁻¹, respectively. NAD(P)H oxidation and NAD(P)⁺ reduction were quantified by measuring their absorbance at 340 nm and using a molar extinction coefficient of 6,220 M⁻¹ cm⁻¹.

Electrophoresis and protein identification by MALDI-TOF/TOF and Mascot analysis.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed on 12% acrylamide gels by the procedure developed by Laemmli (76). Fourteen- and 36-h cell-free cell extract samples from 0- and 8%-O₂-exposed cultures, including cultures grown with reduced and unreduced media, were boiled for 10 min with loading dye and run on SDS-polyacrylamide gels for 45 min at 200 V. Protein bands were detected by staining the gel with Coomassie brilliant blue G-250. The gel was submitted to the UNC Michael Hooker Proteomics Center (University of North Carolina, Chapel Hill); the band of interest was excised, trypsin digested, and analyzed by matrix-assisted laser desorption ionization–tandem time of flight mass spectrometry (MALDI-TOF/TOF MS). Tandem MS (MS/MS) spectra were searched for proteins against the NCBI database for C. ljungdahlii DSM13528 (accession no. CP001666) using Mascot software (Matrix Science, Inc., Boston, MA).

Growth and cell protein quantification.

Growth was measured as optical density at 600 nm. Protein quantification in cell extracts was performed using Bio-Rad's protein assay dye reagent (Bio-Rad, Hercules, CA), according to the manufacturer's instructions.

To determine the relationship between optical density and dry cell density, 11 ml of culture was sampled from growing cells at various time points. One milliliter was used for readings of the optical density at 600 nm (OD₆₀₀), while 10 ml of culture was filtered through a disposable preweighed and predried 0.22-μm-pore syringe filter. Flowthrough was discarded. The filter and cells were washed with 10 ml 50 mM phosphate buffer, pH 7.0. The filters were then incubated at 80°C in a dry air oven and were weighed every 24 h until constant weights were obtained. The optical densities were plotted against their corresponding dry cell weights, yielding a linear relationship between the measured OD₆₀₀ and culture densities (milligrams of dry cells per liter). The relationship between optical density and dry cell weight was found to be 312 mg dry cells/liter per OD₆₀₀ unit for C. ljungdahlii.

Nucleotide sequence accession number.

RNA sequencing data for each condition have been submitted to the NCBI Sequence Read Archive (http://www.ncbi.nlm.nih.gov/sra) under accession no. PRJNA296707.

RESULTS

Effects of oxygen on growth and fermentation product formation.

C. ljungdahlii was previously shown to be consistently resistant to concentrations of oxygen as high as 8% (volume added per volume of headspace) injected into the headspace of batch cultures with mixotrophic medium (63). This finding was confirmed in this study (Fig. 1A). Although cultures exposed to 8% O₂ continued to grow, the rate of growth was reduced compared to that of anaerobic cultures (generation time of 5.8 h versus 4.6 h) (Fig. 1A). Similar to previous findings, more ethanol and less acetate were produced by cultures exposed to 8% O₂ as well (1:2 versus 1:6 mol ethanol/mol acetate) (Fig. 1B and C) (63). Substrate utilization of CO, H₂, and fructose continued in 8%-O₂-exposed cultures but the rate was lower than in anaerobic cultures, and 8%-O₂-exposed cultures consumed about 13% less hydrogen than anaerobic cultures (Fig. 2A). By 36 h (24 h after oxygen exposure), the amounts of CO, H₂, and fructose consumed per liter of culture for oxygen-exposed cells were 7%, 19%, and 25% lower than those in the anaerobic cultures, respectively.

FIG 1.

C. ljungdahlii cell growth (A) and ethanol (B) and acetate (C) production when cultures are exposed to 0 (circles) and 8% (squares) O₂. One-hundred-sixty-milliliter batch reactors have a 110-ml headspace volume. Eight percent O₂ was added to the headspace at 12 h (arrows). The data represent an average from three biological replicates per condition. Error bars indicate standard deviations.

FIG 2.

C. ljungdahlii CO (squares), H₂ (circles), and fructose (triangles) utilization in response to 0% O₂ (anaerobically grown [solid symbols]) and 8% O₂ (open symbols) supplementation of the headspace gas. The data represent an average from six biological replicates per condition (A). Total CO₂ concentration (grams per liter) in 14-h 0%-O₂-exposed, 14-h 8%-O₂-exposed, 36-h 0%-O₂-exposed, and 36-h 8%-O₂-exposed C. ljungdahlii batch cultures. tCO₂ = pCO₂ headspace + [HCO₃⁻] + α × pCO₂ medium, where α (estimated solubility coefficient) = 1.3511 mg/liter/mm Hg. Uninoculated bottles were used as controls. The data represent an average from three biological replicates per condition (B). Concentration of dissolved oxygen (parts per million) at the bottom of 8%-O₂-exposed C. ljungdahlii batch cultures. Zero-percent-O₂-exposed cultures were used as blanks, and uninoculated bottles were used as controls. Reducing agents (Cys-HCl and Na₂S) were omitted in unreduced medium. The data represent an average of three biological replicates per condition (C). Eight percent O₂ was added to the headspace at 12 h (arrow). One-hundred-sixty-milliliter batch reactors have a 110-ml headspace volume. Error bars indicate standard deviations.

CO₂ in both 8%-O₂-exposed and anaerobic cultures increased above initial concentrations by 14 h and continued to increase throughout the fermentation to 72 h (Fig. 2B). However, there was a substantial difference in the amount of CO₂ per liter of culture in 8%-O₂-exposed batch cultures and anaerobic batch cultures: 0.577 g by 72 h (Fig. 2B).

Oxygen injected into the headspace was confirmed to be removed from the inoculated cultures completely by 36 h (Fig. 2C). The total amount of oxygen added to cultures and reduced by the cells was calculated to be ∼6.9 mmol/liter of culture (data not shown). However, based on dissolved oxygen (DO) concentrations of unreduced uninoculated and (reduced) uninoculated medium controls, a portion of this was reduced by cysteine-HCl and sodium sulfide (reducing agents) added to the medium (Fig. 2C). Reducing agents kept the amount of DO in the uninoculated medium close to 0 ppm for at least 2 h after oxygen was injected into the headspace (Fig. 2C). Sometime between 14 and 30 h, the reducing agents were used up, and the concentration of DO in the uninoculated medium increased to ∼2.4 ppm (Fig. 2C). The DO in the unreduced uninoculated medium also increased between the 14- and 30-h time points, indicating that it took longer than 2 h for the DO in the liquid and gas phases to reach equilibrium (Fig. 2C). In this study, reducing agents were added to the medium because ethanol production was higher than when they were excluded, but reducing agents are not required for continued growth of C. ljungdahlii in rich mixotrophic medium when exposed to 8% O₂ (63). Based on a comparison of DO in uninoculated medium containing reducing agents and uninoculated medium without reducing agents, reducing agents removed at most 0.9 ppm (39%) of the oxygen by 36 h (Fig. 2C). This means at least 4.2 mmol/liter of oxygen was removed by the C. ljungdahlii cells (data not shown). At the same time, ∼13.1 mmol/liter (0.577 g/liter) less CO₂ was reduced by 8%-O₂-exposed cells.

Effects of oxygen on gene expression.

RNA sequencing was employed in an effort to identify genes responsible for the oxygen tolerance and higher ethanol/acetate ratio of batch cultures exposed to 8% O₂. Samples for sequencing were taken at 14 h (2 h after oxygen exposure, early response) and 36 h (24 h after oxygen exposure, late response) for 8%-O₂-exposed and anaerobic cultures. In total, 139,256,223 reads with an average length of 100 bp were generated (see Table S1 in the supplemental material), with each treatment condition representing greater than 28.9 million reads. Differential expression analysis was conducted using DEGseq as in a previous RNA-seq study with C. ljungdahlii (77). All 150 differentially expressed genes are displayed in Table S2 in the supplemental material. The expression profile of C. ljungdahlii genes known and predicted to be involved in ethanol and acetate production is presented in Table 1. Although AdhE1 (encoded by CLJU_c16510) was previously identified to be the primary ethanol dehydrogenase in C. ljungdahlii (3, 17), its gene was not differentially expressed at 14 or 36 h: neither were the genes for phosphotransacetylase (CLJU_c12770) or acetate kinase (CLJU_c12780). Genes for predicted acetaldehyde dehydrogenases (CLJU_c11960, CLJU_c39730, and CLJU_c39840), which may be involved in production or consumption of acetaldehyde, the precursor of ethanol, were also not differentially expressed. However, a predicted tungsten-containing aldehyde ferredoxin oxidoreductase (CLJU_c20210), one of two previously hypothesized to catalyze acetate reduction to acetaldehyde with reduced ferredoxin (2), was expressed significantly less by 8%-O₂-exposed cells than by anaerobic cells at 14 h. The other predicted tungsten-containing aldehyde ferredoxin oxidoreductase (CLJU_c20110) was not differentially expressed at 14 or 36 h. The only differentially expressed genes that showed potential for explaining higher ethanol production were a predicted aldehyde oxidoreductase (CLJU_c24130) with homology to molybdenum-containing AORs and a putative xanthine dehydrogenase (CLJU_c23910) with 50% identity to CLJU_c24130 (see Table S3 in the supplemental material). Both were expressed significantly higher in 8%-O₂-exposed cells than anaerobic cells at 36 h; however, expression of CLJU_c24130 was also higher (although not significantly), while expression of CLJU_c23910 was lower (although not significantly) in 8%-O₂-exposed cells at 14 h, and the transcript count of CLJU_c24130 was about 9 times higher than CLJU_c23910 (Table 1). A predicted aldehyde oxidoreductase (CLJU_c24050) with 80% identity to CLJU_c24130 was not differentially expressed at 14 or 36 h (Table 1; see Table S3).

TABLE 1.

Expression profile of C. ljungdahlii genes known and predicted to be involved in ethanol and acetate production

Locus tag	Gene product annotation	Result ata:
		14 h			36 h
		Count		Log2 FC	Count		Log2 FC
		0% O2	8% O2		0% O2	8% O2
CLJU_c11960	Predicted acetaldehyde dehydrogenase	666	1,148	−0.355	32	60	−0.530
CLJU_c12770	Phosphotransacetylase	37,052	38,122	0.388	51,989	49,972	0.468
CLJU_c12780	Acetate kinase	45,467	43,101	0.506	45,696	48,908	0.313
CLJU_c16510	Bifunctional aldehyde/alcohol dehydrogenase (AdhE1)	136,600	184,497	−0.004	4,954	2,901	1.183
CLJU_c20110	Predicted tungsten-containing aldehyde ferredoxin oxidoreductase	64	52	0.708	549	645	0.181
CLJU_c20210	Predicted tungsten-containing aldehyde ferredoxin oxidoreductase	6,547	1,509	2.546	2,055	1,820	0.587
CLJU_c23910	Putative xanthine dehydrogenase, molybdopterin-binding subunit B	4,519	3,648	0.738	1,798	11,958	−2.327
CLJU_c24050	Predicted aldehyde oxidoreductase	56	93	−0.309	1,058	896	0.650
CLJU_c24130	Predicted aldehyde oxidoreductase	9,500	33,544	−1.391	9,697	102,536	−2.991
CLJU_c39730	Predicted acetaldehyde dehydrogenase	6	6	0.635	2	3	0.070
CLJU_c39840	Predicted acetaldehyde dehydrogenase	601	784	0.045	686	623	0.550

^aComparisons of gene expression were made between anaerobic (0%-O₂-exposed) and 8%-O₂-exposed cells. A positive normalized log₂ fold change (FC) value reflects higher expression by anaerobic cells, and a negative normalized log₂ FC value reflects higher expression by 8%-O₂-exposed cells. The threshold for comparison was set at a normalized log₂ FC value of ±2.

The genes that are predicted to be specifically involved in oxygen or reactive oxygen species (ROS) detoxification are provided in Table 2 and include genes coding for two predicted rubrerythrins (CLJU_c28910 and CLJU_c39340). Rubrerythrin is a common name used for NAD(P)H hydrogen peroxidases found in clostridia and other anaerobic bacteria. The predicted rubrerythrin encoded by CLJU_c39340 has 80% identity to Clostridium acetobutylicum's highly active NAD(P)H hydrogen peroxidase, a “reverse rubrerythrin,” also called “rubperoxin” (encoded by homologs CA_C3597 and CA_C3598) (see Table S4 in the supplemental material) (78–80). C. acetobutylicum's rubrerythrin is indirectly reduced by NAD(P)H (preferentially by NADH) through a NADH:rubredoxin oxidoreductase (NROR) and a rubredoxin (57). An alignment of all of the genes encoding differentially expressed rubrerythrin(-like) enzymes from C. ljungdahlii, CLJU_c09090 (differentially expressed gene of the FAD/NAD-dependent oxidoreductase subunit of the anaerobic-type CODH), and C. acetobutylicum NROR (CA_C2448) and rubredoxin (CA_C2778) showed less than 40% identity for all combinations (see Table S4).

TABLE 2.

Expression profile of genes differentially expressed by C. ljungdahlii predicted to be involved in oxygen and reactive oxygen species detoxification

Locus tag	Gene product annotation	Result ata:
		14 h			36 h
		Count		Log2 FC	Count		Log2 FC
		0% O2	8% O2		0% O2	8% O2
CLJU_c21940	Putative flavoprotein	10,478	204,461	−3.857	6,207	36,739	−2.154
CLJU_c28910	Predicted rubrerythrin	1,491	27,416	−3.771	885	16,893	−3.844
CLJU_c36090	Hemerythrin-related protein	2,353	59,405	−4.229	1,263	9,066	−2.433
CLJU_c39340	Predicted rubrerythrin	19,381	195,485	−2.905	39,167	159,979	−1.619

^aComparisons of gene expression were made between anaerobic (0%-O₂-exposed) and 8%-O₂-exposed cells. A positive normalized log₂ fold change (FC) value reflects higher expression by anaerobic cells, and a negative normalized log₂ FC value reflects higher expression by 8%-O₂-exposed cells. The threshold for comparison was set at a normalized log₂ FC value of ±2.

A putative flavoprotein (CLJU_c21940) and a hemerythrin-related protein (CLJU_c36090) are also listed in Table 2. The CLJU_c21940 flavoprotein has 60% identity to a C. acetobutylicum flavoprotein (CA_C2449) that has demonstrated oxidase activity and is reduced by NROR (see Table S4 in the supplemental material) (57). Hemerythrins are proteins that carry oxygen and/or convert oxygen to hydrogen peroxide (81, 82). C. ljungdahlii also has an annotated superoxide dismutase (CLJU_c29780), but expression of the gene coding for it was low for both 8%-O₂-exposed and anaerobically cultured cells (data not shown).

Besides genes directly involved in ethanol/acetate production and oxygen/ROS detoxification, several genes related to substrate, energy, and cofactor metabolism were identified as being differentially expressed (Table 3). These include genes encoding the anaerobic-type CODH complex (CLJU_c09090-110) which are thought to be responsible for CO oxidation to CO₂ in C. ljungdahlii (83), the G-subunit of the RNF complex (CLJU_c11380), which is predicted be a flavin mononucleotide (FMN)-binding redox-active subunit responsible for reduction of NAD (2, 84), a glyceraldehyde 3-phosphate dehydrogenase (CLJU_c13400), which is a glycolysis enzyme, a hydrogenase expression/formation protein (CLJU_c23080) predicted to be involved in nickel insertion (2), the CODH of the Wood-Ljungdahl Pathway gene cluster (CLJU_c37670), which is predicted to bind CO for acetyl-CoA synthesis (2, 18, 83), a predicted citrate:cation symporter, and a NAD-dependent malic enzyme potentially involved in the branched tricarboxylic acid (TCA) “cycle” in C. ljungdahlii (2). Of these genes, the genes coding for the anaerobic CODH complex were significantly more highly expressed by 8%-O₂-exposed cells at both 14 and 36 h, the glyceraldehyde 3-phosphate dehydrogenase, predicted citrate:cation symporter, and NAD-dependent malic enzyme genes were significantly more highly expressed by anaerobic cells at 14 h, and the G-subunit of the RNF complex, the predicted hydrogenase expression/formation protein, and the CODH genes were significantly more highly expressed by 8%-O₂-exposed cells at 36 h (Table 3). Several other genes related to peptide uptake and amino acid metabolism were also more highly expressed in 8%-O₂-exposed cells at 14 h (CLJU_c19310-20, CLJU_c37340-50) and 36 h (CLJU_c28700, CLJU_c28730-60) (see Table S2 in the supplemental material).

TABLE 3.

Expression profile of genes differentially expressed by C. ljungdahlii known and predicted to be involved in substrate, energy, and cofactor metabolism

Locus tag	Gene product annotation	Result ata:
		14 h			36 h
		Count		Log2 FC	Count		Log2 FC
		0% O2	8% O2		0% O2	8% O2
CLJU_c09090	Anaerobic-type CODH, FAD/NAD-dependent oxidoreductase subunit	15,022	184,044	−3.186	7,043	68,954	−2.880
CLJU_c09100	Anaerobic-type CODH, electron transfer subunit	4,988	18,436	−1.457	2,725	27,886	−2.944
CLJU_c09110	Anaerobic-type CODH	16,453	274,126	−3.629	8,432	92,640	−3.047
CLJU_c11380	RnfG	12,694	13,522	0.338	670	10,725	−3.548
CLJU_c13400	Glyceraldehyde 3-phosphate dehydrogenase	15,196	3,997	2.356	21,082	49,392	−0.817
CLJU_c17950	Putative molybdopterin biosynthesis protein	1,625	2,806	−0.358	1,944	21,178	−3.034
CLJU_c19400	Predicted transcriptional regulator	2,438	71,339	−4.442	5,622	30,966	−2.050
CLJU_c19410	Ferric uptake regulation protein	1,336	43,688	−4.602	4,144	23,877	−2.115
CLJU_c19420	Putative epimerase	2,040	38,862	−3.822	2,688	17,482	−2.290
CLJU_c19430	Putative membrane protein	2,187	38,238	−3.699	1,918	14,008	−2.457
CLJU_c19440	Conserved hypothetical protein	3,082	54,143	−3.706	3,467	23,993	−2.380
CLJU_c19450	Predicted two-component response regulator	469	1,549	−1.294	316	1,091	−1.375
CLJU_c19460	Predicted signal transduction histidine kinase	1,134	4,242	−1.474	800	2,404	−1.176
CLJU_c19470	Hypothetical protein	1,415	4,567	−1.261	1,728	3,695	−0.686
CLJU_c19480	Putative flavin reductase-like protein with rubredoxin domain	1,035	37,947	−4.767	796	5,290	−2.322
CLJU_c19490	Ferritin	1,607	68,688	−4.988	1,004	8,844	−2.727
CLJU_c20050	Predicted MoaD/ThiS domain protein	370	579	−0.219	219	4,871	−4.063
CLJU_c20060	Predicted dinucleotide-utilizing enzymes involved in molybdopterin/thiamine biosynthesis	235	321	−0.022	223	2,703	−3.186
CLJU_c23080	Predicted hydrogenase expression/formation protein	5,433	7,215	0.020	3,775	25,025	−2.319
CLJU_c35550	Predicted transcriptional regulator, Fur family	25,077	201,453	−2.577	27,102	104,436	−1.535
CLJU_c37670	CODH	119,541	101,616	0.664	23,751	183,015	−2.535

^aComparisons of gene expression were made between anaerobic (0%-O₂-exposed) and 8%-O₂-exposed cells. A positive normalized log₂ fold change (FC) value reflects higher expression by anaerobic cells, and a negative log₂ FC value reflects higher expression by 8%-O₂-exposed cells. The threshold for comparison was set at a normalized log₂ FC value of ±2.

Table 3 also includes a Fur family transcriptional regulator (CLJU_c35550) expected to be involved in iron metabolism and a 10-member gene cluster (CLJU_c19400-90) predicted to be involved in metal cofactor repair and management (named crm). A KEGG image and visual representation of the differentially expressed crm gene cluster are provided in Fig. 3A and B, respectively. The crm gene cluster includes genes coding for a predicted transcriptional regulator (CLJU_c19400), a ferric uptake regulation protein (CLJU_c19410), a predicted two-component response regulator (CLJU_c19450), and a predicted signal transduction histidine kinase (CLJU_c19460). In addition to these, the hypothetical protein (CLJU_c19470) contains a polo-box domain of a polo-like kinase. Polo-like kinases have been known to be involved in cell cycle regulation, DNA damage, and oxygen stress in other organisms (85).

FIG 3.

(A) KEGG image of the C. ljungdahlii cofactor repair and management (crm) gene cluster. (B) Visual comparison of read alignment to a locus on the C. ljungdahlii genome containing the crm gene cluster of FastQ reads derived from RNA sequencing of 0%-O₂-exposed (anaerobically grown) and 8%-O₂-exposed cultures. Images were generated using the sequence assembly visualization software Tablet. Sample 1 is one of three 14-h-time point anaerobic cell transcriptomes, and sample 7 is one of three 14-h-time point 8%-O₂-exposed cell transcriptomes. Arrows 1, 2, 3, 4, and 5 point to positions 2107698 (beginning of the CLJU_c19400 ORF), 2111421 (end of the CLJU_c19440 ORF), 2111708 (beginning of the CLJU_c19450 ORF), 2115295 (end of the CLJU_c19470 ORF), and 2116834 on the genome (beginning of the CLJU_c19490 ORF).

Besides these regulatory proteins, the gene cluster also contains a gene encoding ferritin (CLJU_c19490), a protein known to sequester iron, and a gene encoding a flavin reductase-like protein with a rubredoxin domain similar to that of proteins shown to release iron from ferritin using reduced FMN (CLJU_c19480) (86–89). Flavins used by this protein may be transported into the cell by the putative membrane protein encoded by CLJU_c19430, which contains two EamA domains like the characterized riboflavin-transporting transmembrane permeases (RibN) from Ochrobactrum anthropi and Vibrio cholerae (90). A blastx query of the putative epimerase gene (CLJU_c19420) in this oxygen-induced gene cluster revealed that it had higher homology to a phenazine biosynthesis-like protein than its annotated function as an epimerase. Phenazines serve as electron shuttles for various terminal electron acceptors and have been shown to reduce iron (hydr)oxides in Pseudomonas chlororaphis and other bacteria (91). Also, a blastx query of the gene sequence for the conserved hypothetical protein (CLJU_c19440) revealed a DUF1848 domain. The C terminus of DUF1848 domains contain a cluster of cysteines similar to the Fe-S cluster of a radical SAM domain, which are known to catalyze diverse reactions, including cofactor biosynthesis and maturation, posttranscriptional and posttranslational modification, enzyme activation, substrate anchoring, electron transfer, and sulfur donation (92).

Enzyme activities associated with ethanol production.

Figure 4A shows a modified version of the predicted C. ljungdahlii anabolic pathways leading to ethanol formation previously described by Köpke et al. (2). Either ethanol can be produced from acetyl-CoA by acetaldehyde dehydrogenase activity followed by ethanol dehydrogenase activity or from acetate by acetate reductase (reverse reaction of aldehyde oxidation) followed by ethanol dehydrogenase (Fig. 4A).

FIG 4.

(A) Predicted C. ljungdahlii anabolic pathways leading to ethanol. Aldehyde oxidoreductase (AOR), acetate, acetaldehyde, oxidized and reduced ferredoxin (Fd_ox/FD_red), anaerobic-type CODH complex, CO, CO₂, and ethanol are in boldface to indicate metabolic flux proposed to account for higher ethanol and lower acetate production by 8%-O₂-exposed C. ljungdahlii batch cultures. Pta, phosphotransacetylase; Ack, acetate kinase; AdhE, bifunctional aldehyde/alcohol dehydrogenase; 2 [H], reduced form of NAD [phosphate]; PFOR, pyruvate:ferredoxin oxidoreductase; CODH, carbon monoxide dehydrogenase. (B) AOR (methyl viologen acetate oxidoreductase activity) enzyme activity assay results for cell extract from 14-h 0%-O₂-exposed (anaerobically grown), 14-h 8%-O₂-exposed, 36-h 0%-O₂-exposed, and 36-h 8%-O₂-exposed C. ljungdahlii batch cultures. Activity (units per milligram) is defined as micromoles of methyl viologen oxidized per minute. The data represent an average from six biological replicates per condition. Error bars indicate standard deviations.

Acetate reductase activity was at least 6-fold higher at 14 h and 4-fold higher at 36 h on average for cell-free cell extracts from 8%-O₂-exposed cultures compared to cell extracts from anaerobic cultures (Fig. 4B). Acetaldehyde oxidoreductase activity as measured by benzyl viologen reduction did not seem to be present in anaerobic cells or 8%-O₂-exposed cells at 14 or 36 h since acetaldehyde did not increase benzyl viologen reduction by cell extracts (data not shown). However, acetaldehyde-independent benzyl viologen reduction was mediated by cell extracts from cultures of 14-h 8%-O₂-exposed cells, 36-h anaerobic cultures, and 36-h 8%-O₂-exposed cells, but not 14-h anaerobic cultures (see Fig. S2A in the supplemental material). Aldehyde-independent benzyl viologen reduction by cell extracts indicates that cell extracts (except those derived from 14-h anaerobic cultures) have a lower redox potential than benzyl viologen. Also, based on a general trend of higher acetaldehyde-independent benzyl viologen activity for cell extracts from cells grown normally in media with reducing agents versus unreduced cultures, electrons from reducing agents can be used by cells to reduce benzyl viologen (see Fig. S2A). Ethanol and acetaldehyde dehydrogenase activities of cell extracts from 8%-O₂-exposed and anaerobic cultures had less than a 2-fold difference at the 14- and 36-h time points (see Fig. S2B and S2C).

Enzyme activities associated with oxygen tolerance.

Based on differential expression of genes for multiple rubrerythrin and rubrerythrin-like enzymes (hemerythrin) (Table 2), it was hypothesized that the 8%-O₂-exposed C. ljungdahlii cultures would have higher NAD(P)H hydrogen peroxidase activity than anaerobic cultures. Figure 5A shows that this prediction was correct at both 14 and 36 h.

FIG 5.

(A) NAD(P)H hydrogen peroxidase enzyme activity assay results for 14-h 0%-O₂-exposed (anaerobically grown), 14-h 8%-O₂-exposed, 36-h 0%-O₂-exposed, and 36-h 8%-O₂-exposed samples. Activity (units per milligram) is defined as micromoles of NAD(P)H oxidized per minute. The data represent an average of six biological replicates per condition. Error bars indicate standard deviations. (B) SDS-PAGE analysis of C. ljungdahlii cell-free cell extract (5 μg) from 14-h 0%-O₂-exposed (lane 2), 14-h 8%-O₂-exposed (lane 3), 36-h 0%-O₂-exposed (lane 4), and 36-h 8%-O₂-exposed (lane 5) cultures. Five microliters of PageRuler prestained protein ladder was loaded into lanes 1 and 6. The image was modified to remove irrelevant lanes. (C) Protein identification by MALDI-TOF/TOF and Mascot analysis of O₂-induced band (denoted by an arrow in panel B). The significant score identification threshold (P < 0.05) for the C. ljungdahlii protein database is 72. (MS/MS and peptide sequenced ion scores higher than 72 are considered a significant identification.) The MS/MS score and peptide sequenced ion score for CLJU_c39340 (the locus tag for a predicted rubrerythrin) are 726 and 697, respectively.

C. ljungdahlii was also assayed for catalase, oxidase, superoxide dismutase, and xanthine oxidase activities, which are other common enzyme activities for ROS and oxygen detoxification. No detectable activity was observed for these enzymes in cell extracts of 8%-O₂-exposed or anaerobically grown C. ljungdahlii cells (data not shown).

Identification of an oxygen-inducible protein.

To determine if there were any observable highly translated oxygen-induced proteins, cell extracts from anaerobic and 8%-O₂-exposed cultures (14 and 36 h) were run on an SDS-PAGE gel (Fig. 5B). A distinct band between the ∼15- and ∼25-kDa ladder markers was present only in lanes 3 and 4 corresponding to the 14-h and 36-h 8%-O₂-exposed cultures (Fig. 5B). Cell extracts from anaerobic and 8%-O₂-exposed cultures (14 and 36 h) and grown in media without reducing agents were also run on an SDS-PAGE gel (see Fig. S1 in the supplemental material); no apparent difference was observed between these and cell extracts from cells grown in medium containing reducing agents. The distinct band was identified as the differentially expressed predicted rubrerythrin (CLJU_c39340) (Fig. 5C and Table 1).

DISCUSSION

In this study, exposure of C. ljungdahlii cells to 8% O₂ when grown in mixotrophic medium was confirmed to result in higher ethanol production with correspondingly lower acetate production (Fig. 1B and C) (63). Differential expression analysis and enzyme activity assays revealed two genes (CLJU_c24130 and CLJU_c39340) that are likely to play major roles in higher ethanol production and oxygen detoxification, respectively (Tables 1 and 2 and Fig. 4B and 5A). Expression analysis and fermentation products also suggest that carbon, energy, and cofactor metabolism link the AOR and rubrerythrin activities (Table 3).

Higher ethanol production is due to a carbon flux from reduction of acetate to acetaldehyde by AOR followed by reduction of acetaldehyde to ethanol by alcohol dehydrogenase (likely AdhE1) (Fig. 4A). The primary enzyme found to be responsible for higher ethanol production in 8%-O₂-exposed cultures compared to anaerobic cultures appears to be an AOR (encoded by CLJU_c24130). In addition to significantly higher expression of this gene in cells exposed to 8% O₂, it was also shown that AOR activity (acetate reductase activity) was 6 and 4 times higher in 8%-O₂-exposed cells than in anaerobic cells at 14 and 36 h, respectively (Table 1 and Fig. 4B). However, a recent study by Xie et al. showed that higher expression of another AOR (encoded by CLJU_c20110) correlated with higher gene expression and higher ethanol production by C. ljungdahlii (93). Prior to this study, Kopke et al. predicted that under surplus reducing equivalents, acetate could be converted to acetaldehyde, the precursor of ethanol, by the AORs encoded by CLJU_c20110 and CLJU_c20210 (2). It was also shown that CLJU_c20110 was significantly more highly expressed when C. ljungdahlii was grown autotrophically (80% CO, 20% CO₂) than when it was grown heterotrophically with fructose (93, 94). In contrast, we observed that both CLJU_c20110 and CLJU_c20210 had relatively low transcript counts, which may be explained by the fact that the growth medium used in this study had fructose as the primary carbon source (Table 1). In fact, our results indicate that CLJU_c20210 was significantly more highly expressed in anaerobic cultures than in 8%-O₂-exposed cultures at 14 h (Table 1). To our knowledge, this is the first study to demonstrate corroborating fermentation, expression, and activity data associated with a particular AOR (encoded by CLJU_c24310), which is a different one than the other AORs reported (Table 1 and Fig. 1B and 4B).

As with the AORs encoded by CLJU_c20110 and CLJU_c20210, the AOR encoded by CLJU_c24130 is expected to utilize ferredoxin as a coenzyme for activity (95). However, this AOR is predicted to use molybdenum as a cofactor, whereas the other two are predicted to be tungsten-containing AORs (2). Of the characterized carboxylic acid-reducing AORs of C. formicaceticum, a molybdenum-containing AOR is much less oxygen labile than the tungsten-containing AORs, and it is true in general that tungsten is more oxygen labile than molybdenum (95–99). More specifically, C. formicaceticum's purified tungsten-containing AOR lost 20% of its activity in 5 min versus 4% for the molybdenum-containing AOR (95). This may explain why the gene for the tungsten-containing AOR was significantly less expressed, while the gene for the AOR encoded by CLJU_c24310 was significantly more highly expressed when cells were exposed to 8% O₂ (Table 1). A drawback of molybdenum-containing AORs is that they are less active than tungsten-containing AORs; in C. formicaceticum, the purified molybdenum-containing AOR was found to have 10 times less activity than the tungsten-containing AOR (95). While it is not a perfect comparison, here we found that even the higher carboxylic acid reductase activity of cell-free extracts of 8%-O₂-exposed cells (acetate reductase) was about 4 times less than the (propionate reductase) activity of C. ragsdalei cell-free extracts (∼0.03 U/mg versus ∼0.12 U/mg) (Fig. 4B). The activities of the cell extracts from anaerobic cultures were 24 times less than those from C. ragsdalei (∼0.005 U/mg versus ∼0.12 U/mg) (Fig. 4B). The difference in activities for anaerobic cultures of these closely related species can be explained by autotrophic growth for C. ragsdalei versus mixotrophic growth, with fructose as the predominant carbon substrate for C. ljungdahlii. As previously mentioned, expression of C. ljungdahlii's main tungsten-containing AOR (CLJU_c20110) was significantly less for fructose-grown cells than for 80%-CO-grown cells (93, 94). Despite the lower activity of molybdenum-containing AORs, their benefit of a higher tolerance to oxidized compounds like O₂, NOx, and SOx may make this predicted molybdenum-containing AOR a more interesting genetic target for overexpression in commercial applications where less syngas purification means higher margins. Further study of the substrate specificity of this molybdenum-containing AOR would also benefit those interested in producing biochemicals with the carboxylate platform.

Another major finding of this study is the identification of a rubrerythrin (encoded by CLJU_c39340) that is likely to play a key role in C. ljungdahlii's ability to tolerate low oxygen exposure by reducing hydrogen peroxide. Unlike several other genes encoding rubrerythrin and rubrerythrin-like (hemerythrin) enzymes that were also differentially expressed at 14 and 36 h, the protein encoded by CLJU_c39340 was also highly translated (Table 2 and Fig. 5B). In a previous study, SDS-PAGE analysis of air-exposed C. acetobutylicum also showed one distinct band at 22 kDa, which was later revealed to be two rubrerythrins with one amino acid difference (encoded by CA_C3597 and CA_C3598) (100, 101). These rubrerythrins, which were shown to be highly active hydrogen peroxidases, have sequence and structural similarities to CLJU_c39340 (see Table S4 in the supplemental material) (57). Actually, the hydrogen peroxidase activity was found to be about 40 times higher in C. ljungdahlii oxygen-exposed cell extracts than in C. acetobutylicum oxygen-exposed cell extracts (∼0.93 U/mg versus ∼0.023 U/mg) (Fig. 5A) (58). However, it is unclear what (co)enzymes are acting as rubredoxin and NADH:rubredoxin oxidoreductase in C. ljungdahlii since no differentially expressed genes predicted to be involved in oxygen/ROS detoxification had higher than 40% homology with the respective [co]enzymes in C. acetobutylicum (see Table S4). Despite the induction of protective oxygen-responsive genes and higher hydrogen peroxidase activity, C. ljungdahlii is much more sensitive to oxygen than is C. acetobutylicum (Table 2 and Fig. 5A). Mid-log-phase batch cultures of C. acetobutylicum (300 mg dry cells/liter) were previously found to rapidly reduce 0.64 to 0.80 ppm DO in the growth media, and cells were able to survive after flushing with air for 30 min (∼20% O₂) and even grow during flushing with 5% O₂ (100, 102). In contrast, C. ljungdahlii cultures required about 24 h to reduce 0.18 ppm DO and were previously shown to die after exposure to greater than 10% O₂ (Fig. 2C) (63). The lower oxygen tolerance of C. ljungdahlii compared to C. acetobutylicum is likely the result of C. ljungdahlii's lack of oxidase and superoxide dismutase/reductase activities, which are present in C. acetobutylicum (data not shown) (78).

AOR and rubrerythrin activities (and thus the higher ethanol production and oxygen detoxification) are connected through carbon, energy, and cofactor metabolism. Both AOR and rubrerythrin are redox proteins that require the coenzymes reduced ferredoxin and NAD(P)H for their respective activities. In C. ljungdahlii, ferredoxin can be reduced by CO oxidation activity, hydrogenase activity, and PFOR (fructose metabolism) activity; NAD(P)+ can be reduced by hydrogenase, transhydrogenase NfnAB (energy conservation), and RNF complex activity (energy conservation) (103). Of these, expression analysis revealed that (at 14 and 36 h) the anaerobic-type CODH and (at 36 h) the redox-active G-subunit of the RNF complex were significantly more highly expressed in 8%-O₂-exposed cells than in anaerobic cells (Table 3). Furthermore, CO consumption per gram of cells was also higher, while hydrogen and fructose consumption were lower in 8%-O₂-exposed cells (Fig. 2A). Therefore, it appears that CO is preferentially used as a source of electrons for oxygen detoxification. The ability of 14-h cell extracts from 8%-O₂-exposed cultures to reduce benzyl viologen (without the addition of any other substrate, i.e., acetaldehyde) and the inability of 14-h cell extracts from anaerobic cultures to do the same indicate that cells exposed to 8% O₂ generate surplus reduced coenzymes with higher redox potentials than benzyl viologen, such as reduced ferredoxin (see Fig. S2A in the supplemental material). Expression analysis indicates that ferredoxin is reduced by the anaerobic-type CODH (Table 3). This is followed by NAD⁺ reduction by ferredoxin:NAD⁺ oxidoreductase (Rnf) activity providing NADH for hydrogen peroxidase activity by rubrerythrin (Table 2, Table 3, and Fig. 4A). Turnover of surplus reduced coenzymes is achieved by ferredoxin:acetate reductase activity by AOR followed by NADH:acetaldehyde dehydrogenase activity by AdhE1 (Table 1, Fig. 4A; see Fig. S2C). Therefore, despite less total liquid products because some electrons were used to reduce O₂ instead of CO₂, these activities resulted in a higher ethanol/acetate ratio in 8%-O₂-exposed cultures (Fig. 1B and C and 2B and C).

Of these redox (co)enzymes mentioned, ferredoxin, the anaerobic-type CODH, rubrerythrin, and AOR require Fe-S clusters. When C. ljungdahlii is exposed to oxygen, the crm gene cluster is significantly upregulated (Table 3 and Fig. 3). Based on blastx query results, it is predicted to have a similar function to the suf operon in E. coli, which has been shown to be involved in the regeneration or replacement of cofactors such as iron-sulfur clusters and molybdopterin of molybdenum and tungsten-containing proteins and limits Fenton reactions during oxygen detoxification (104). Also, several other genes annotated as coding for molybdopterin biosynthesis and molybdopterin-containing proteins were significantly more highly expressed in 8%-O₂-exposed cells (CLJU_c17950, CLJU_c20050-60, CLJU_c23910, CLJU_c24130) (see Table S2 in the supplemental material). Molybdopterin may contain either tungsten or molybdenum metals, but as previously mentioned, molybdenum is less oxygen labile (96–99, 105). Nickel may also be a more preferential cofactor during oxygen exposure based on significantly higher expression of the anaerobic-type (nickel-dependent) CODH complex (CLJU_c09090-110) at 14 and 36 h and predicted hydrogenase expression/formation protein (CLJU_c23080) at 36 h, which is thought to be involved in nickel insertion (Table 3) (2). In general, nickel-containing CODHs and hydrogenases are oxygen labile, but some have long inactivation half-lives, can be protected by the presence of substrate, and can be regenerated by reducing agents (106–108). The shift in cofactors and associated enzymes allows for continued activity and less oxidative damage during oxygen detoxification.

The ability of C. ljungdahlii to tolerate low concentrations of oxygen and other oxidized contaminants (i.e., NOx and SOx) is important for its application in the biofuels industry because it is likely that such contaminants will be contained in biomass-derived syngas (22, 23, 28, 109). Although it was observed that exposure of C. ljungdahlii to low levels of oxygen (8% O₂) increased ethanol and decreased acetate concentrations, the purpose of this work was not to suggest that oxygen should be added to syngas fermentations. Slower metabolism of syngas components H₂ and CO and less carbon from CO going to product formation (through acetyl-CoA synthesis) are undesirable; therefore, oxygen should generally be removed from C. ljungdahlii syngas fermentations. Rather, this work contributes the identification of genetic targets to potentially improve ethanol production in the presence of syngas contaminants by improving the robustness of syngas-consuming microbial catalysts like C. ljungdahlii and thus reduce the need for costly syngas cleaning. This work also contributes new knowledge about the metabolism and physiology of this model organism.