Escherichia coli TOP10 (Thermo Fisher Scientific, Massachusetts, United States), E. coli EPI300 (Lucigen, Wisconsin, United States), and E. coli HB101 pKR2013 (DSM 5599) were grown at 37°C in Luria Broth (LB) medium containing (per liter): 5 g NaCl; 10 g peptone; and 5 g yeast extract. C. ljungdahlii ATCC13528 was generally cultivated in anaerobic Rich Clostridial Medium (RCM) containing per liter: 5 g fructose; 3 g yeast extract; 10 g meat extract; 10 g peptone; 5 g NaCl; 1 g soluble starch; 3 g sodium acetate; 0.5 g L-cysteine HCl; and 4 mL resazurin-solution (0.025 vol-%). For growth experiments with C. ljungdahlii, standard PETC medium (Klask et al., 2020) was used containing (per liter): 1 g yeast extract; 1.0 g NH₄Cl; 0.1 g KCl; 0.2 g MgSO₄x7 H₂O; 0.8 g NaCl; 0.1 g KH₂PO₄; 0.02 g CaCl₂x2 H₂O; 4 mL resazurin-solution (0.025 vol-%); 10 mL trace element solution (TE, 100x); 10 mL Wolfe’s vitamin solution (100x); 10 mL reducing agent (100x); and 20 mL of fructose/2-(N-morpholino)ethanesulfonic acid (MES) solution (50x). TE was prepared as 100x stock solution containing (per liter): 2 g nitrilotriacetic acid (NTA); 1 g MnSO₄xH₂O; 0.8 g Fe(SO₄)2(NH₄Cl)2 × 6 H₂O; 0.2 g CoCl₂x6 H2O; 0.0002 g ZnSO₄x7 H₂O; 0.2 g CuCl₂x2 H₂O; 0.02 g NiCl₂x6 H₂O; 0.02 g Na₂MoO₄x2 H₂O; 0.02 g Na₂SeO₄; and 0.02 g Na₂WO₄. The pH of the TE was adjusted to 6.0 after adding NTA. The solution was autoclaved and stored at 4°C. Wolfe’s vitamin solution was prepared aerobically containing (per liter): 2 mg biotin; 2 mg folic acid; 10 mg pyridoxine-hydrochloride; 5 mg thiamin-HCl; 5 mg riboflavin; 5 mg nicotinic acid; 5 mg calcium pantothenate; 5 mg p-aminobenzoic acid; 5 mg lipoic acid; and 0.1 mg cobalamin. The vitamin solution was sterilized using a sterile filter (0.2 μm), sparged with N₂ through a sterile filter, and stored at 4°C. The 50x fructose/MES solution contained (per 100 mL): 25 g fructose; and 10 g MES. The pH was adjusted to 6.0 by adding KOH. For autotrophic experiments, fructose was omitted. In nitrate experiments, ammonium chloride was replaced with sodium nitrate (NaNO₃) in the equal molar amount (=18.7 mM). The reducing agent solution contained (per 100 mL): 0.9 g NaCl and 4 g L-cysteine HCl and was prepared with anaerobic water under anaerobic conditions. The reducing agent was stored at room temperature. For solid LB medium, 1.5 weight-% agar was added. For solid RCM or PETC medium 1.0–2.0 weight-% agar was added. For conjugation of C. ljungdahlii cells (see below) a modified PETC medium (PETC + 5gS) was used containing additionally (per liter): 5 g peptone and 5 g meat extract.

Liquid E. coli cultures and autotrophic C. ljungdahlii cultures were agitated at 150 revolutions per minute (rpm) (Lab Companion Incubater Shaker ISS-7100R, Jeio Tech, Daejeon, Republic of Korea). Heterotrophic cultures of C. ljungdahlii and LB plates with E. coli cells were incubated without shaking (Incubator IN260, Memmert, Schwabach, Germany). Anaerobic work was performed in an anaerobic chamber (Glovebox-System UNIlab Pro, MBraun, Garching bei München, Germany) with an N₂ (100 vol-%) atmosphere. However, C. ljungdahlii cultures in bottles were transferred at the bench with sterile syringes and needles. Before each transfer between serum bottles, we flamed the top of the rubber stopper with ethanol (70 vol-%) at a Bunsen burner. All plating work with C. ljungdahlii was performed in the anaerobic chamber with a maximum of 5 parts per million (ppm) oxygen in the atmosphere. All plating work with E. coli was carried out in a laminar flow bench (Hera Safe KS18, Thermo Fischer Scientific, Massachusetts, United States). Antibiotics (see below) were added to maintain plasmid stability in recombinant cultures of E. coli and C. ljungdahlii.

Chloramphenicol (30 μg/mL), ampicillin (100 μg/mL), kanamycin (50 μg/mL), and trimethoprim (10 μg/ml) were applied to maintain plasmids in E. coli strains, while thiamphenicol (5 mg/mL) was used for recombinant strains of C. ljungdahlii. Trimethoprim was dissolved in DMSO (100 vol-%). Thiamphenicol was prepared as aerobic stock solution (25 mg/mL) in DMSO (100 vol-%) and diluted with sterile water (1:10) before use. The diluted thiamphenicol solution (2.5 mg/mL) was transferred into a sterile 1 mL syringe. 100 μL of this solution was used to add to a 50 mL RCM or PETC medium (final concentration of 5 μg/mL). The use of DMSO over ethanol as solvent for thiamphenicol prevented the addition of external ethanol, which is a metabolite, to cultures of C. ljungdahlii. All antibiotic stock solutions were stored at −20°C.

The broad-host shuttle-vector system pMTL80000 (Heap et al., 2009) was used for all cloning steps. All generated plasmids of this study (Supplementary Table 1) were cloned with restriction endonucleases and T4 ligase (New England Biolabs, Frankfurt am Main, Germany) or Gibson assembly (NEBuilder^® HiFi DNA Assembly, New England Biolabs, Frankfurt am Main, Germany). PCR work was carried out with primers provided by IDT (Integrated DNA Technologies, Iowa, United States) (Supplementary Table 2) and with a proof-reading Q5^® High-Fidelity DNA Polymerase (New England Biolabs, Frankfurt am Main, Germany)) according to the manufacturer’s guidelines. Genomic DNA (gDNA) was purified from 2 mL of exponential cultures of C. ljungdahlii with the NucleoSpin Tissue Mini kit (Macherey-Nagel, Düren, Germany) and used as PCR-template. Notably, instead of performing harsh cell disruption according to the manufacturer’s recommendation, we applied a 6 × 10 s vortex interval during the procedure. All PCR steps were performed with Q5^® High-Fidelity DNA Polymerase (New England Biolabs, Frankfurt am Main, Germany) and primers provided by IDT (Integrated DNA Technologies, Iowa, United States) (Supplementary Table 2). PCR products were purified with QIAquick PCR Purification Kit (Qiagen, Hilden, Germany).

For screening and continuous purity control of our C. ljungdahlii strains (Supplementary Table 4), we performed PCRs from culture samples or from purified DNA with the Phire Plant Direct PCR Master Mix (Thermo Fischer Scientific, Massachusetts, United States). E. coli colonies grown on selective LB plates after receiving plasmid constructs were analyzed for the correctly assembled plasmids using the Phire Plant Direct PCR Master Mix (Thermo Fischer Scientific, Massachusetts, United States). A small amount of recombinant E. coli cell material was directly transferred to the reaction mix. For C. ljungdahlii cells, 0.5–1 mL culture sample was harvested by centrifugation for 3 min at 13,806 rpm (Centrifuge 5,424, FA-45-24-11, Eppendorf, Hamburg, Germany) and resuspended in 100–500 μL 10 mM NaOH depending on the size of the cell pellet. Subsequently, cell suspensions were boiled for 10 min at 98°C. The hot reaction tubes were incubated on ice for 1 min and quickly vortexed before they served as a DNA template. In general, we used 20 μL PCR master mix, which consisted of 10 μL (Thermo Fisher Scientific, Massachusetts, United States) 2x, 0.8 μL of each primer, 1 μL cell lysate sample, and 7.4 μL nuclease-free water. The PCR reaction was carried out according to the manufacturer’s guidelines. We generally used the primers tra60 bp_fwd and repH_401 bp_rv or repH_643 bp_rv for these control PCRs (Supplementary Table 2), because they bind to the plasmid backbone of every pMTL plasmid used in this study. Verification of gene deletion in the genome of C. ljungdahlii was performed with “outside” primers, which bound upstream and downstream of the used homology-directed repair arms (HDR1/2) on the genomic DNA (Supplementary Table 2). In addition, we performed test-digestion of the generated plasmids with restriction enzymes (New England Biolabs, Frankfurt am Main, Germany), and analyzed the fragment pattern via gel electrophoresis. The final plasmid sequence was verified by Sanger sequencing. Plasmid DNA was purified from E. coli with self-made purification buffers (described below). Correct plasmid DNA was then purified with the QIAprep Spin Miniprep kit (Qiagen, Hilden, Germany) prior to further use.

For screening of successfully transformed E. coli cells, we used a time- and money-saving protocol to purify plasmid DNA from multiple samples without using a commercial kit, which is a modified alkaline lysis protocol adapted from Sambrook et al. (1989). All centrifugation steps were performed at 13,806 rpm for 5 min (Centrifuge 5,424, FA-45-24-11, Eppendorf, Hamburg, Germany). Recombinant E. coli cells were grown overnight in 5 mL selective liquid LB at 37°C and 150 rpm. 1.5–3 mL cell suspension were harvested in 1.5 mL reaction tubes. The supernatant was discarded, and the pellet was resuspended by vortexing in 150 μL P1-buffer (50 mM Tris, 10 mM EDTA, 100 μg/mL RNAseA, pH 8.0 with HCl). Cells were lysed in 150 μL P2-buffer (200 mM NaOH, 1 vol-% SDS) and inverted five times. Proteins were precipitated by adding 250 μL P3-buffer (2.55 M Na-acetate, pH 4.8 with acetic acid). The samples were inverted five times and centrifuged. Subsequently, 500 μL of the supernatant were transferred into new 1.5 mL tubes and mixed with 500 μL isopropanol. The samples were quickly vortexed and centrifuged again. Afterward, the supernatant was discarded. At this step, the precipitated and non-visible plasmid-DNA pellet remained on the bottom of the tube. The pellet was washed twice with ice-cold ethanol (70 vol-%) omitting resuspending the DNA. After the second washing, the supernatant was discarded completely and the remaining ethanol was first removed by snapping the tube on a piece of clean paper towel and then through drying at 50–65°C for 10 min. The dried pellet was resuspended in 30 μL elution buffer (Tris/EDTA, pH 7.2) or deionized water. Purified plasmid-DNA with a concentration of 250–500 ng/μL was clean enough for subsequent cloning steps and test-digestion, however, an additional clean-up with the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) was carried out when a Sanger sequencing reaction was necessary. P1-buffer needed to be stored at 4°C to maintain RNAse activity for up to 3 months. P2- and P3-buffer were stored at room temperature.

This protocol was adapted and modified according to Mock et al. (2015). Cells of C. ljungdahlii were grown in RCM overnight to mid exponential growth until an OD₆₀₀ of 0.4–0.8 was reached (NanoPhotometer^® NP80, Implen, München, Germany). E. coli HB101 pKR2013 (DSM 5599) harboring the desired CRISPR-Cas12a-plasmid was grown as pre-culture in selective 5 mL LB medium overnight. The plasmid pKR2013 contains essential genes to mediate conjugation and a kanamycin resistance cassette. 1–2 mL of the E. coli cells were used to inoculate 10 mL selective LB medium in 50 mL baffled flask and cultivated until mid-exponential growth (OD₆₀₀ 0.5–1.0). Subsequently, the E. coli culture was cooled to 4°C and 2 mL were transferred into sterile 2 mL reaction tubes. The C. ljungdahlii culture was kept at room temperature until use. Inside the anaerobic chamber, E. coli cells were centrifuged softly at 2,900 rpm (mySpin™ 12 mini centrifuge, Thermo Fischer Scientific, Massachusetts, United States) to protect pili, and washed once with sterile and anaerobic 0.1 M phosphate buffered saline (PBS) at pH 6.0. Afterward, the washed pellet was resuspended gently in 100–150 μL cell suspension of C. ljungdahlii and directly transferred to well-dried RCM-agar plates (2 vol-% agar). Spot-mating was carried out at 37°C inside the anaerobic chamber overnight. After 8–24 h the spot was resuspended with anaerobic PBS (pH 6.0) and centrifuged at 10,000 rpm (mySpin™ 12 mini centrifuge, Thermo Fischer Scientific, Massachusetts, United States) for 3 min. The supernatant was discarded, and the pellet was resuspended in the remaining volume of the tube. Subsequently, 100 μL of the cell suspension was plated onto selective PETC + 5gS-agar plates, which contained 5 g/L of peptone and 5 g/L meat extract to support growth. Selective agar plates should not be older than 2–3 days. Thiamphenicol was added for plasmid selectivity. Trimethoprim (10 mg/mL) was added to counter-select against E. coli. Growth was obtained after 4–5 days at 37°C inside the anaerobic chamber. C. ljungdahlii colonies were transferred into (Glasgerätebau Ochs, Bovenden, Germany) containing 5 mL RCM with the respective antibiotics. A successful transformation of C. ljungdahlii with the correct plasmid was confirmed as follow: (1) growth in selective RCM with a characteristic pH decrease due to acetogenesis; (2) control PCRs with primers for plasmid specific fragments; and (3) plasmid purification from the culture and re-transformation into E. coli TOP10 cells.

Electroporation of C. ljungdahlii cells was performed as previously reported (Xia et al., 2020) and applied for all non-CRISPR-based plasmids. Single colonies growing on selective plates were verified by PCR analyses and by re-transformation of E. coli with plasmid DNA, which was extracted from C. ljungdahlii.

In general, all recombinant C. ljungdahlii strains were pre-grown in 50 mL RCM in 100 mL serum bottles for 24–48 h. Subsequently, 2 mL cell suspension were used to inoculate 50 mL PETC medium in 100 mL (Glasgerätebau Ochs, Bovenden, Germany). This PETC pre-culture was cultivated for 40–48 h at 37°C until mid-exponential growth phase at OD₆₀₀ of 0.5–1.0. Afterward, cells were transferred anaerobically into 50 mL reaction tubes, which were equilibrated for 3–5 days inside the anaerobic chamber. Cell harvest was performed outside the anaerobic chamber at 3,700 rpm for 12 min (Centrifuge 5920R, S-4 ×1,000, Eppendorf, Hamburg, Germany) at room temperature. After the centrifugation, the tubes were transferred back immediately into the anaerobic chamber, to keep the time at aerobic conditions at a minimum. Inside the anaerobic chamber, the supernatant was discarded, and the pellet was resuspended in fresh PETC medium to adjust to an OD₆₀₀ of 5–10. The concentrated cell suspension was then transferred into sterile and anaerobic 10 mL (Glasgerätebau Ochs, Bovenden, Germany), sealed carefully, and used to inoculate main cultures outside of the anaerobic chamber. 1 mL of the cell suspension was used to inoculate 100 mL PETC main cultures. For heterotrophic growth experiments, 240 mL (Glasgerätebau Ochs, Bovenden, Germany) were used. Autotrophic growth experiments were performed in 1,000 mL Duran pressure plus bottles (Schott, Mainz, Germany), to provide a high medium-to-headspace ratio. The Duran pressure plus bottles were sealed with butyl stoppers and a GL45 ring cap. Before inoculation of autotrophic cultures, the N₂ headspace was replaced with a sterile gas mixture consisting of H₂/CO₂ (80/20 vol-%). Each bottle contained 0.5 bar overpressure. All cultures were cultivated in biological triplicates as batch cultures. The gas headspace was not refilled during the experiments. However, for the strain C. ljungdahlii pMTL83151_P_nat_rnfCDGEAB and the control strain C. ljungdahlii pMTL83151 we refilled the headspace during this experiment with the same gas mixture to 0.5 bar overpressure at time points 44.5, 73.5, and 148.5 h (section “Plasmid-Based Complementation Relieved the Phenotypes of the Clostridium ljungdahlii ΔRNF and ΔrseC Strains”). We did not measure the headspace gas composition during our experiments. Culture samples of 3 mL were taken at the bench and used for: (1) OD₆₀₀ measurement; (2) pH measurement; (3) HPLC analyses (acetate and ethanol); and (4) FIA analyses (nitrate, nitrite, and ammonium). All culture samples were stored at −20°C until use. OD₆₀₀ samples were diluted with medium or PBS buffer when the absorbance was > 0.5. We applied a two-tailed Student’s t-test for all cultivation data. All p-values (P) below 0.001 indicate high significance and are given as ≤ 0.001.

HPLC analyzes were performed as described before (Klask et al., 2020). In addition, all frozen supernatant samples were thawed at 30°C for 10 min and 300 rpm, vortexed briefly, and centrifuged for 3 min at 13,806 rpm (Centrifuge 5,424, FA-45-24-11, Eppendorf, Hamburg, Germany) before use. All HPLC samples were randomized.

Nitrate and nitrite concentrations were measured in a FIA continuous-flow analyzer system (AA3 HR AutoAnalyzer System, Seal Analytical GmbH, Norderstedt, Germany) as described before (Klueglein et al., 2014). Briefly, nitrate is reduced to nitrite with hydrazine and then reacts with sulfanilamide and NEDD (N-1-Naphthylethylenediamine di-HCl) to form a pink complex, which can be quantified photo-metrically at 550 nm. The protocol follows the DIN 38405/ISO 13395 standard methods. Ammonium concentrations were measured in the same system but with salicylate and dichloroisocyanuric acid forming a blue complex that is measured at 660 nm instead. The protocol was following DIN 38406/ISO 11732 standard methods. Culture samples of C. ljungdahlii were treated as explained above for HPLC preparation. However, we prepared 1:50 dilution in 1 mL with deionized water prior to the FIA analyses. Standards for nitrate, nitrite, and ammonium were measured before and during the analyses for a standard curve and to minimize drift effects. Nitrate concentrations of each sample were calculated by the difference of the amount of nitrite measured with and without the prior reduction by hydrazine.

For the expression analyses, we grew the strains C. ljungdahlii WT, C. ljungdahlii ΔRNF, and C. ljungdahlii ΔrseC under autotrophic and heterotrophic conditions as described above. The cultivation medium was PETC with ammonium as nitrogen source. Pre-cultures were grown in heterotrophic medium for 48 h. Next, the cells were transferred into the anaerobic chamber and harvested for 12 min at 25°C and 3,700 rpm (Centrifuge 5,920 R, S-4 × 1,000, Eppendorf, Hamburg, Germany) outside of the anaerobic chamber. The supernatant was discarded under anaerobic conditions and the pellet was resuspended in fresh medium of the main cultures. The start OD₆₀₀ for autotrophic main cultures was 0.2, while it was 0.15 for heterotrophic conditions. The cultures were cultivated at 37°C. 10 mL culture samples were taken after 3 and 20 h. The samples were immediately cooled on ice and centrifuged for 12 min at 4°C and 3,700 rpm (Centrifuge 5,920 R, S-4 × 1,000, Eppendorf, Hamburg, Germany). The cell pellets were stored at −20°C until RNA extraction.

RNA was purified from C. ljungdahlii with the RNeasy Mini Kit (Qiagen, Hilden, Germany) as described before (Liu et al., 2013). For the RNA extraction, we used 2× 10⁸ cells, which was approximately 10 mL of a C. ljungdahlii culture at OD₆₀₀ 0.2. The cell lysis was performed in the lysis buffer of the kit with 50 mg glass beads (0.1 mm silica spheres, MP Biomedicals, Eschwege, Germany, Eschwege, Germany) in a bead beater (5G FastPrep, MP Biomedicals, Eschwege, Germany) for 2 × 60 s at 9 m/s. RNA samples were eluted in 30 μL nuclease-free water. After the extraction procedure, an additional DNase I digest (RNase free Kit, Thermo Fisher Scientific, Massachusetts, United States) was performed to remove potential DNA contamination. Elimination of genomic DNA was confirmed with PCR analyses and gel electrophoresis. cDNA synthesis was performed with the QuantiTect Reverse Transcriptase Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. We used 500 ng RNA as template for each reaction. cDNA was stored at −20°C until further use.

All qRT-PCR analyses were performed in a Quantstudio 3 Thermocycler (Applied Biosystems, Thermo Fisher Scientific, Massachusetts, United States). The PCR reaction mix contained 10 μL SYBR Green Master mix (Thermo Fisher Scientific, Massachusetts, United States), 1 μL of a fwd and rv qRT-PCR primer (final concentration 500 nM) (Supplementary Table 2), and 1 μL (∼5 ng) cDNA template. We used the rho gene as reference gene, which was described before as suitable candidate for qRT-PCR experiments with C. ljungdahlii (Liu et al., 2013). We added RNA controls to further exclude gDNA contamination in our samples. All qRT-PCR reactions were performed in technical triplicate according to the manufacturer’s instructions. We set the Ct threshold to 0.1. The fold change in gene expression between the samples was determined with the 2^–ΔΔCt method as described before (Livak and Schmittgen, 2001). We examined the PCR efficiency of our qPCR master mix by using plasmid DNA containing the sequence of each tested gene in a series of dilutions (10^–1, 10^–2, 5 × 10^–3, 10^–3, 5 × 10^–4, 10^–4). The slopes were ranging from 0.04 to 0.09 for the RNF-gene cluster genes and 0.17 for rseC, and were thus, close to zero, which proofs that the efficiencies are similar and the 2^–ΔΔCt can be used for interpretation of the qRT-PCR data (Livak and Schmittgen, 2001). We applied a two-tailed Student’s t-test based on our ΔCt values for each gene to analyze the significance of our samples in comparison to the wild type.

Cultures of C. ljungdahlii were stored at −80°C. For this, cultures were grown in RCM until late exponential growth phase (OD₆₀₀ 0.8–1.2) at 37°C for 36–48 h. The cells were transferred into anaerobic 50 mL reaction tubes inside the anaerobic chamber and harvested outside of the anaerobic chamber for 12 min at 4°C and 3,700 rpm (Centrifuge 5,920 R, S-4 × 1,000, Eppendorf, Hamburg, Germany). The supernatant was discarded inside the anaerobic chamber and the pellet was resuspended in fresh RCM medium to an OD₆₀₀ of 5–10. Two milliliter of the cell suspension was transferred into 10 mL serum bottles, which were previously filled with 2 mL of 25–50 vol-% anaerobic and autoclaved glycerol. The serum bottles were briefly vortexed outside the anaerobic chamber, incubated on ice for 10–15 min and subsequently frozen at −80°C. For inoculation of a new RCM culture, a single serum bottle was quickly thawed up under rinsing water and 1–2 mL of the cell suspension was immediately transferred with a syringe into the medium bottle. Cultures of E. coli were stored at −80°C in sterile screw-cap tubes filled with 25–50 vol-% glycerol.

First, we achieved a full deletion of the RNF-gene cluster in C. ljungdahlii with a clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein Cas12a (CRISPR-Cas12a) system, which we implemented and used to generate all deletion strains in this study (Figure 1A and Supplementary Text 1A). We confirmed the identity of the C. ljungdahlii ΔRNF strain (Figure 1B and Supplementary Text 1B), and compared the growth of C. ljungdahlii wild-type (WT) to the growth of C. ljungdahlii ΔRNF. We performed growth experiments with carbon dioxide and hydrogen (autotrophy) and with fructose (heterotrophy), while we added equimolar amounts of either ammonium or nitrate as nitrogen source to the medium for both autotrophy and heterotrophy (section “Materials and Methods,” Figure 2 and Supplementary Figure 1). As expected, we observed growth for C. ljungdahlii WT in all growth experiments (Figure 2A and Supplementary Figure 1A). However, the nitrogen source had a distinct influence on the growth rate, final OD₆₀₀, fermentation product spectrum, and pH (Table 1, Figure 2B, Supplementary Text 1C, and Supplementary Figure 1B). We found that nitrate reduction occurred rapidly in our growth experiments (Figure 2F and Supplementary Figure 1F). C. ljungdahlii WT utilized all provided nitrate within 53 h of cultivation with carbon dioxide and hydrogen (Figure 2F) and within 47 h of cultivation with fructose (Supplementary Figure 1F). The ammonium concentrations increased concomitant with decreasing nitrate concentrations when nitrate was provided in the medium (Figure 2E and Supplementary Figure 1E). Noteworthy, we also observed an increase in the ammonium concentration when ammonium was provided as the nitrogen source during autotrophy (Figure 2F). We had added a small amount of yeast extract (0.1 weight-%) in all cultivation conditions (section “Materials and Methods”). We found this increase in the ammonium concentration in all our cultivation experiments, and we argued that the additional ammonium was a by-product of the fermentation of the added yeast extract in our cultivation medium (section “Materials and Methods,” Supplementary Text 1C). We did not measure any nitrite as an intermediate of the nitrate reduction pathway (discussed below).

In contrast, the C. ljungdahlii ΔRNF strain was unable to grow with carbon dioxide and hydrogen regardless of the nitrogen source (Figure 2A). This was expected and confirmed previous findings in which rnf genes were deleted or disrupted (Tremblay et al., 2012; Westphal et al., 2018). We did not observe a pH decrease, and also not an accumulation of ethanol as a metabolic activity of non-growing cultures of C. ljungdahlii ΔRNF. However, some minor amounts of acetate were detected, which arguably resulted from the fermentation of the small amount of yeast extract (Figures 2B–D and Table 1). Furthermore, nitrate reduction as a metabolic activity of non-growing cultures was not detectable in C. ljungdahlii ΔRNF with carbon dioxide and hydrogen (Figure 2F).

For the C. ljungdahlii ΔRNF strain, heterotrophic growth with fructose was still possible but notably reduced (Figure 1C and Supplementary Figure 1A). Compared to the wild type, we observed significant reduction for C. ljungdahlii ΔRNF with ammonium and nitrate, respectively, of the growth rates (−34%, −42%), maximum OD₆₀₀ values (−53%, −56%), and maximum acetate concentrations (−32%, −42%) (Table 1 and Supplementary Table 5). The maximum ethanol concentration was significantly reduced with ammonium (−41%), and ethanol was not produced at all with nitrate (Table 1 and Supplementary Figures 1C,D). During heterotrophy, C. ljungdahlii ΔRNF was able to utilize nitrate but considerably slower than the wild type (Supplementary Figure 1F). At the end of the cultivation, cultures of C. ljungdahlii ΔRNF had only consumed 49% of the provided nitrate (Supplementary Figure 1F). Overall, we observed a halt in growth and metabolic activity for cultures of C. ljungdahlii ΔRNF with fructose after 47 h of cultivation in nitrate-containing medium and after 56 h of cultivation in ammonium-containing medium (Supplementary Figure 1). Fructose concentrations at the end of the cultivation remained at a concentration of 8.0–9.7 mM, which is still 30–35% of the initially provided concentration (Supplementary Figure 1G). The pH did not increase during heterotrophy with nitrate in C. ljungdahlii ΔRNF, but instead slowly decreased until the end of the cultivation (Supplementary Figure 1B). Notably, the final pH for heterotrophic cultures of C. ljungdahlii ΔRNF with nitrate was still higher compared to cultures with ammonium (Supplementary Figure 1B).

Next, we investigated the role of the small putative regulator gene rseC (CLJU_c11350). We applied our CRISPR-Cas12a system to delete the rseC gene from the genome (Supplementary Figure 2A). We performed growth experiments with the generated C. ljungdahlii ΔrseC strain under the same conditions as for the C. ljungdahlii WT and ΔRNF strains. Cultures of C. ljungdahlii ΔrseC did not grow with carbon dioxide and hydrogen, neither with ammonium nor with nitrate, during a total cultivation time of 189 h (Figure 2). Non-growing cultures for this strain did not accumulate notable concentrations of acetate or ethanol during the cultivation time (Figures 2C,D). Furthermore, we did not observe nitrate reduction or a remarkable change in pH as a metabolic activity of non-growing cultures for this strain during autotrophy (Figures 2B,E,F).

However, heterotrophic growth of C. ljungdahlii ΔrseC was possible, and in contrast to C. ljungdahlii ΔRNF, the impact was less pronounced for growth with ammonium but limited to some extent with nitrate (Supplementary Figure 1A). In comparison to the wild type with ammonium, only the maximum ethanol concentration (−29%) was significantly reduced (Supplementary Table 5 and Supplementary Figure 1C). In contrast, with nitrate, the heterotrophic growth rates (−34%), maximum OD₆₀₀ (−30%), and maximum ethanol concentrations (−42%) were all significantly reduced, while the maximum acetate concentration was not (Supplementary Table 5 and Supplementary Figure 1D). Nitrate reduction was not restricted during heterotrophy in C. ljungdahlii ΔrseC (Supplementary Figures 1E,F). Indeed, we observed a rapid utilization of all supplied nitrate within 60 h of cultivation, which is similar to the observations that we had made for the wild type (Supplementary Figure 1F). Thus, rseC seems to be involved in regulating the expression of the RNF-gene cluster during autotrophy, but not during heterotrophy. However, the exact impact on gene expression of the RNF-gene cluster cannot be deduced from these findings.

We questioned whether the wild-type phenotype, particularly with respect to autotrophy, can be restored by plasmid-based gene complementation in the C. ljungdahlii ΔRNF and C. ljungdahlii ΔrseC strains. Therefore, we generated the plasmid-carrying strains C. ljungdahlii ΔRNF pMTL83151_P_nat_rnfCDGEAB and C. ljungdahlii ΔrseC pMTL83152_rseC. The plasmids encode the RNF-gene cluster under the control of the native promoter region upstream of the rnfC gene from the genome (P_nat) in pMTL83151_P_nat_rnfCDGEAB and the rseC gene under the control of the constitutive thiolase promoter (P_thl) in pMTL83152_rseC, respectively. We investigated the complementation strains in ammonium-containing medium with carbon dioxide and hydrogen for growth (Supplementary Figure 3). Indeed, the plasmid-based expression of the deleted genes relieved the phenotype and enabled autotrophy with carbon dioxide and hydrogen for both strains (Table 2 and Supplementary Figure 3). The control strains that carried an empty plasmid failed to grow autotrophically, as we had already observed for the non-complemented deletion strains.

For C. ljungdahlii ΔRNF pMTL83151_P_nat_rnfCDGEAB we found a significantly reduced maximum OD₆₀₀ (−71%), and maximum acetate concentration (−22%) (Table 1 and Supplementary Figures 3A–C), while the maximum ethanol concentration was similar to the wild type (Supplementary Figure 3D). Furthermore, the complemented strain had a prolonged lag phase of 71 h (Supplementary Figure 3A). Notably, the medium for the complementation experiments always contained antibiotics, which generally caused a slightly negative impact on growth of plasmid-carrying C. ljungdahlii strains such as just described for the C. ljungdahlii ΔRNF pMTL83151_P_nat_rnfCDGEAB strain. Surprisingly, this was not the case for the C. ljungdahlii ΔrseC pMTL83152_rseC strain. Instead of a prolonged lag phase, we observed a slightly shortened lag phase for this strain (Figures 1A, 3A). Compared to the wild type, this strain reached a slightly but significantly increased maximum OD₆₀₀ (+17%) and maximum acetate concentration (+6%) (Table 2 and Supplementary Figure 3A). However, this strain did not produce any detectable ethanol during the cultivation (Table 2 and Supplementary Figure 3D). Furthermore, the pH value did not show any notable change (Supplementary Figure 3B). The differences in the performance of our complementation strains might be explained by the plasmid copy number and the differences in the used promoter sequences for the complementation plasmids, while this hypothesis awaits further experimentation (Supplementary Text 1D).

We further investigated the activating or repressing function on the gene expression of the RNF-gene cluster by RseC. For this, we performed qRT-PCR analyses to investigate the individual expression profiles of the genes rnfC, rnfD, rnfG, rnfE, rnfA, rnfB, and rseC in the C. ljungdahlii ΔrseC strain. We included the C. ljungdahlii ΔRNF and C. ljungdahlii WT strains as controls (Supplementary Figure 5, section “Materials and Methods”). We analyzed samples after 3 and 20 h to investigate the transcriptomic response after inoculating the autotrophic and heterotrophic main cultures from heterotrophic pre-cultures. During the cultivation of the six main cultures (three strains, two conditions), C. ljungdahlii WT grew during autotrophy and heterotrophy, while C. ljungdahlii ΔRNF and C. ljungdahlii ΔrseC only grew during heterotrophy (Supplementary Figures 5C,D). Thus, we exposed non-growing cells of C. ljungdahlii ΔRNF and C. ljungdahlii ΔrseC to an autotrophic environment and collected samples after 3 and 20 h. We argued that this would result in a transcriptomic response, even though the cultures did not show a difference in OD₆₀₀ over the 20-h incubation time of this experiment (C. ljungdahlii ΔRNF, 0.20 ± 0.02 and C. ljungdahlii ΔrseC, 0.22 ± 0.02, respectively), but remained at the inoculation OD₆₀₀. In comparison, for the wild type, 1–2 generations can be expected within 20 h of cultivation during autotrophy, which resulted in an increase of the OD₆₀₀ from 0.28 to 0.39 (Supplementary Figure 5C).

The qRT-PCR results in this paragraph are given as log₂ (fold change in gene expression), where we discuss a value of ≤ −2 (0.25-fold) as a biologically relevant downregulation, and a value of ≥ + 2 (fourfold) as a biologically relevant upregulation (Figure 3). We did not measure any expression signals for any of the deleted RNF genes in the C. ljungdahlii ΔRNF strain and for the deleted rseC gene in the C. ljungdahlii ΔrseC strain. We found that all RNF-gene cluster genes (except for rnfE, −1.8 ± 0.2) were downregulated [ranging from −2.2 ± 0.1 (rnfD) to −4.7 ± 0.1 (rnfC)] in the C. ljungdahlii ΔrseC strain, when exposing non-growing cells of this strain to hydrogen and carbon dioxide (Figure 3). We observed a similar pattern of downregulation for the 3 and 20-h samples of the C. ljungdahlii ΔrseC strain (Figures 3A, 4C). In the heterotrophic samples, the RNF-gene cluster genes were not downregulated in the 3-h samples [ranging from −1.0 ± 0.1 (rnfC) to −1.8 ± 0.1 (rnfE)]. After 20 h of cultivation time during heterotrophy, we even observed a pattern in which none of the genes had a log2 (fold change in gene expression) of ≤ −1 or ≥ + 1. In the C. ljungdahlii ΔRNF strain as a control, we found that rseC expression was upregulated in the 3-h samples during autotrophy (+2.4 ± 0.04) but not during heterotrophy (+1.3 ± 0.1) (Figure 3B). In the 20-h samples both conditions were not different (Figure 3D). For the wild type, all genes (except for rnfD and rnfE in the 3-h sample) were upregulated during autotrophy when compared to heterotrophy for the 3-h samples [ranging from + 2.3 ± 0.03 (rnfB) to + 5.4 ± 0.1 (rseC)], and for the 20-h samples [ranging from + 2.8 ± 0.04 (rnfB) to + 3.8 ± 0.1 (rnfE)], respectively (Supplementary Figure 5). Thus, from our qPCR results we concluded that RseC likely has a positive regulatory impact on the RNF-gene cluster during autotrophy, but not during heterotrophy, while we cannot rule out that this regulation was an indirect effect, which involves further regulatory elements, from these experiments.

Based on our previous findings, we investigated whether rseC genes are present in the genomes of selected other microbes, including the most prominent model acetogens. We searched for putative rseC genes in the RNF-complex gene-containing genomes of the model acetogens C. ljungdahlii, Clostridium autoethanogenum, A. woodii, Eubacterium limosum, Clostridium carboxidovorans, and the model non-acetogen Clostridium kluyveri (Figure 4 and Table 3). Indeed, we found putative rseC genes for all these candidates. We looked at the genomic location and distance of the rseC gene to the RNF-gene cluster (Figure 4 and Table 3). We noticed that the rseC gene was located directly upstream of the RNF complex gene cluster in C. ljungdahlii (CLJU_c11350), C. autoethanogenum (CAETHG_3225), C. carboxidovorans (Ccar_25725), and C. kluyveri (CKL_1263). However, the rseC gene in A. woodii (Awo_C21740) and E. limosum (B2M23_08890) was not in direct genetic vicinity of the RNF-gene cluster (Figure 4 and Table 3). In addition, we identified a second gene with homologies to rseC in C. carboxidovorans (Cca_07835) and C. kluyveri (CKL_2767), but neither RNF-complex genes nor other genes that are involved in the autotrophic metabolism, such as the genes for the Wood-Ljungdahl pathway, are located in the direct vicinity of these second rseC homologs (Figure 4 and Table 3). Furthermore, the RseC protein sequence seems to be highly conserved in the microbes that contain an RNF-gene cluster (Supplementary Figure 6 and Supplementary Text 1F).

In contrast, we did not find a putative rseC gene when we searched the genomes of two further model acetogens Moorella thermoacetica and Thermoanaerobacter kivui, which possess an energy-converting hydrogenase (Ech) complex instead of an RNF complex (Hess et al., 2014). Notably, we also identified a putative rseC gene in the non-acetogenic bacteria R. capsulatus, which is the microbe in which the RNF complex was first described (Schmehl et al., 1993), and in Thermotoga maritima (Schmehl et al., 1993). The rseC gene in R. capsulatus is located upstream of rnfF instead of rnfC, which is separated by five genes (Figure 4). In T. maritima, RNF genes were not found next to the putative rseC gene (THEMA_1487) (Figure 4). Also E. coli possesses one rseC gene that is organized in the rseABC gene cluster, which plays an important role in the SoxR-mediated oxidative stress response as described elsewhere (Figure 4 and Supplementary Text 1E; Koo et al., 2003).

These findings led to the question, whether we can find an acetogen that contains an RNF-complex but no rseC gene. To answer this question, we performed a more extensive and automated genome search. We based this search on a collection of 47 out of 61 acetogens with a fully sequenced genome from Bengelsdorf et al. (2018), which are stored in the German Collection of Microorganisms and Cell Cultures (Deutsche Sammlung von Mikroorganismen und Zellkulturen, DSMZ) and American Type Culture Collection (ATCC). In this automated genome search, we confirmed our initial findings, and in total identified 30 acetogens that contained potential RNF-complex genes (Supplementary Table 7). Seven of these acetogens (Acetitomaculum ruminis DSM 5522, Blautia hydrogenotrophica DSM 10507, Blautia schinkii, Marvinbryantia formatexigens DSM 14469, Oxobacter pfennigii, Terrisporobacter mayombei, and Treponema primitia ZAS-2) did not show evidence for an rseC gene. We identified several acetogens that contained more than one potential rseC gene (Supplementary Table 7). Interestingly, we found three acetogens (Acetobacterium fimetarium, Clostridium magnum DSM 2767, and Oxobacter pfennigii), which have both potential RNF-complex genes and Ech-complex genes in their genomes, while of those Oxobacter pfennigii was not found to contain an rseC gene as mentioned above (Supplementary Table 7).

Overall, we found with our genome search that not all investigated acetogens that contain RNF genes have putative rseC genes in their genomes. Some acetogens have several putative rseC homologs, and some of these putative rseC genes are associated with the RNF genes, while others are not. We did not identify an acetogen that contained an rseC gene and exclusively an Ech-complex but no RNF-complex (Table 3 and Supplementary Table 7). The reasons for these observations and the regulatory effects in acetogens with and without an rseC gene will be important future research questions.

We had found that nitrate reduction during heterotrophy is impacted for the C. ljungdahlii ΔRNF strain but not the C. ljungdahlii ΔrseC strain. Thus, we aimed to explore nitrate metabolism and the interplay with the RNF complex further. For C. ljungdahlii, it was postulated that nitrate is reduced by nitrate reductase to nitrite and, subsequently, converted via nitrite reductase and hydroxylamine reductase into ammonium, and the involved genes were predicted in the genome (Köpke et al., 2010; Nagarajan et al., 2013). Emerson et al. (2019) had found that in the presence of nitrate the expression level of the genes that encode the putative nitrate reductase (CLJU_c23710-30) were significantly increased. The three genes are annotated as nitrate reductase NADH oxidase subunit (CLJU_c23710), nitrate reductase electron transfer subunit (CLJU_c23720), and nitrate reductase catalytic subunit (CLJU_c23730) (Köpke et al., 2010). We refer to these three genes (CLJU_c23710-30) as the nar gene cluster. We verified the absence of the nar gene cluster from the genome of the C. ljungdahlii Δnar strain, after mediating the deletion with our CRISPR-Cas12a system (Supplementary Figure 2B). This strain was able to grow during autotrophy and heterotrophy, but had completely lost the ability to reduce nitrate under both conditions (Figure 5F and Supplementary Figure 7F). Consequently, we observed similar growth and pH behavior for cultures of C. ljungdahlii Δnar during autotrophy with either ammonium or nitrate (Figures 5A, 6B and Supplementary Figures 7A,B). Enhanced autotrophic growth in nitrate-containing medium when compared to ammonium-containing medium, such as with the wild-type strain, was not detected (Figure 5A).

However, we still observed differences in the growth when compared to the wild type. For ammonium and nitrate conditions, respectively, growth rates (−24%, −76%), maximum OD₆₀₀ (−21%, −55%), and maximum acetate concentrations (−25%, −16%) during autotrophy of C. ljungdahlii Δnar were significantly reduced (Table 1 and Figure 5A). Instead, the maximum ethanol concentrations were significantly increased (+79%) for ammonium, but significantly decreased (−64%) for nitrate conditions, respectively (Figure 5D and Table 1). A pH increase as a consequence of ammonium production from nitrate reduction, such as observed for the wild type, was not observed in cultures of C. ljungdahlii Δnar (Figure 5B).

For heterotrophic cultures with ammonium, we did not observe significant differences in growth behavior and acetate production to the wild type. Only the maximum ethanol concentration was significantly increased (+45%) (Supplementary Table 5, Supplementary Figure 7). With fructose and nitrate, the maximum observed OD₆₀₀ (−32%) and maximum acetate concentration (−34%) were significantly reduced, while the maximum ethanol concentration (+234%) was significantly increased (Supplementary Table 5 and Supplementary Figure 7D). The provided fructose was only consumed completely by C. ljungdahlii Δnar in ammonium-containing but not in nitrate-containing medium (Supplementary Figure 7G).

Finally, we confirmed that the complementation of C. ljungdahlii Δnar with the plasmid pMTL83152_nar, which encodes the nar gene cluster under the expression control of the constitutive P_thl promoter, enabled the C. ljungdahlii Δnar pMTL83152_nar strain to utilize nitrate under autotrophic conditions again, while this was not possible in an empty plasmid control strain (Supplementary Figure 8). The nitrate cultures of C. ljungdahlii Δnar pMTL83152_nar had significantly different growth rates (−26%), maximum OD₆₀₀ (+54%), as well as maximum acetate (−17%) and ethanol (−57%) concentrations in comparison to the wild type when growing with nitrate (Table 2 and Supplementary Figure 8). In summary, we revealed that the expression of the nar gene cluster led to the only functional nitrate reductase in C. ljungdahlii under the tested conditions.

Here, we provided further insight into the autotrophy of C. ljungdahlii and the connection to nitrate metabolism. With the strain C. ljungdahlii ΔRNF, we confirmed previous work by Tremblay et al. (2012) that the absence of the RNF complex leads to a complete loss of autotrophy in C. ljungdahlii. Unlike in the previous study by Tremblay et al. (2012), our strain provides a stable genotype that cannot revert back to the wild-type genotype, which can be used to further study the energy conservation principles in this acetogen (Figures 1B, 2). Heterotrophic growth in our deletion strain was still possible, but considerably reduced when compared to the wild type (Figure 1C and Supplementary Figure 1), which was expected from the work by Tremblay et al. (2012). While we did not measure the difference in the headspace gas composition during heterotrophy for C. ljungdahlii ΔRNF and wild type, we argue that C. ljungdahlii ΔRNF lost the ability to fixate the carbon dioxide that is released during glycolysis, which is the defining feature of acetogens (Drake et al., 2008; Schuchmann and Müller, 2014). Thus, even though the Wood-Ljungdahl pathway was still present, this strain was not able to balance the electrons in the metabolism sufficiently to drive the Wood-Ljungdahl pathway and/or in other metabolic pathways, which resulted in a remaining 30% of fructose that was not consumed until the end of the batch cultivation (Supplementary Figure 1). However, further research is required to confirm this hypothesis. The RNF deletion in A. woodii did also lead to reduced acetate production during heterotrophy, but the strain reached similar OD₆₀₀ values compared to the A. woodii wild type (Westphal et al., 2018). In comparison to C. ljungdahlii, the RNF complex of A. woodii uses sodium ions instead of protons to generate the chemiosmotic gradient, which is then consumed by a sodium-dependent F₁F_O ATPase to generate ATP (Biegel and Müller, 2010; Hess et al., 2013). Overall, this further confirms the meticulous differences in the energy conservation and redox balancing in different acetogens (Katsyv and Müller, 2020), which have to be considered to apply acetogens for biotechnological purposes.

We further investigated the regulation of the RNF-gene cluster by the putative regulator RseC. The rseC gene is known to encode a transcriptional regulator in other microbes such as E. coli and S. Typhimurium (Supplementary Text 1E; Beck et al., 1997; De Las Peñas et al., 1997; Yura and Nakahigashi, 1999; Koo et al., 2003). Our results demonstrated that RseC played a critical role for the formation of a functional RNF complex in C. ljungdahlii (Table 2 and Figure 2). A deletion of the rseC gene led to the complete loss of autotrophy (Figure 2). With our qRT-PCR analyses, we confirmed that RseC, indeed, had a positive regulatory effect on the expression of the RNF-gene cluster during autotrophy. Our results indicate that RseC is essential for the activation of RNF-gene cluster expression during autotrophy, but not during heterotrophy, while we cannot rule out that this activation is mediated by other modulating activities such as secondary regulators (Figures 3A,C, 6, Supplementary Figure 5, and Supplementary Text 1G). Al-Bassam et al. (2018) identified a transcription start site upstream of rseC and found that rseC was poorly translated under heterotrophic conditions similar to rnfC in C. ljungdahlii. Two transcription start sites were identified within the RNF-complex gene cluster rnfCDGEAB, which could be preceded by potential promoter-binding sites for RseC (Al-Bassam et al., 2018). We can speculate that RseC most likely binds to the transcription start site, which is located upstream of the rnfC gene, because the entire RNF-gene cluster was downregulated during autotrophy and heterotrophy after 3 h of cultivation (Figure 3A). Based on our qRT-PCR results from the heterotrophic samples after 20 h of cultivation, it can be further speculated that the second transcription start site, which is located upstream of rnfE (Al-Bassam et al., 2018), might be a second binding site for RseC. We see a slight upregulation of the three genes rnfE, rnfA, and rnfB in these samples for the rseC deletion compared to the wild-type strain. Thus, we argue that RseC could act as a negative regulator to modulate RNF-gene cluster expression during heterotrophy by repressing the genes rnfE, rnfA, and rnfB (Figure 3C). The plasmid-based complementation of rseC in the C. ljungdahlii ΔrseC strain (and overexpression in the wild-type background) re-enabled growth with carbon dioxide and hydrogen, and reduced the lag phase during the transition from heterotrophy to autotrophy (Table 2 and Supplementary Figures 3, 4). This further argues for a function of RseC as a positive regulator of the RNF-gene cluster. The RseC protein in E. coli seems to contain two transmembrane domains with the C-terminal end being located in the cytoplasm (Daley et al., 2005). Thus, another possible function of RseC could be the modulation of protein-protein interactions with the RNF complex, because for all the RseC homologs that we investigated here, two transmembrane helices were predicted (Supplementary Table 6). For instance, RseC could stabilize the RNF complex in the membrane, which is required for the electron translocation mechanism or the interaction with other cytoplasmic proteins during autotrophy, but not during heterotrophy. This could explain why a lack of the RseC protein is not leading to the same reduced heterotrophy as observed for the C. ljungdahliiΔRNF strain. The upregulation of the rseC gene in the C. ljungdahliiΔRNF strain indicates that the rseC gene itself is also under further transcriptional control (Figures 3B,D). An upregulation of the rseC gene in all samples might indicate the effort of the cells to induce RNF-gene cluster expression further, which is apparently not possible in the C. ljungdahliiΔRNF strain. Thus, potentially additional direct or indirect regulatory effects are mediated by RseC. This could be, for example, the regulation of further genes as a positive or negative regulator, which in turn could have an effect on the functionality of the RNF complex. Importantly, others had identified a TetR-family protein that is involved as an alternative sigma-factor in the regulation of autotrophy vs. heterotrophy in C. autoethanogenum (Lemgruber et al., 2019). Lemgruber et al. (2019) had identified a promoter sequence upstream of the rseC gene in C. autoethanogenum (CAETHG_3226) that is likely recognized by this alternative sigma-factor. The same sequence motif is found upstream of the rseC gene (CJLU_c11350) in C. ljungdahlii. However, the function of the TetR-family protein has not yet been investigated in C. ljungdahlii. Therefore, protein-DNA binding experiments should be performed to investigate the binding ability of RseC to putative promoter regions, as well as the regulation of the rseC gene itself, in future experiments. In addition, the study of the subcellular localization of RseC will be required to unravel the regulatory functions of RseC in C. ljungdahlii and other acetogens with an RNF complex in more detail.

Furthermore, we investigated the nitrate metabolism in C. ljungdahlii. We confirmed that the genes CLJU_c23710-30 encode the functional subunits of the only nitrate reductase under the tested conditions for C. ljungdahlii (Figure 5 and Supplementary Figures 7, 8). In the presence of nitrate, the wild type quickly utilized all nitrate even though we had found in all our growth experiments that a sufficient amount of nitrogen-source was covered by the added yeast extract (Figure 2F and Supplementary Figure 1). Thus, nitrate reduction in C. ljungdahlii is mainly used for energy conversion, and therefore must be of a dissimilatory function (Emerson et al., 2019). However, C. ljungdahlii neither possesses genes for cytochromes nor for the biosynthesis of ubiquinone, which limits the generation of a chemiosmotic gradient to the RNF complex (Köpke et al., 2010). The nitrate reductase in C. ljungdahlii is most likely located in the cytosol rather than associated with the membrane, as one would expect from dissimilatory nitrate reductases in bacteria (Zumft, 1997; Köpke et al., 2010; Nagarajan et al., 2013). This type of nitrate reduction was described as fermentative nitrate reduction and was already observed for other microbes, but is less understood than the assimilatory and dissimilatory nitrate reduction in the microbial world (Hall, 1973; Hasan and Hall, 1975; Seifritz et al., 1993; Emerson et al., 2019). Contrarily, nitrate reduction does not function as an independent energy-conserving pathway, because autotrophic growth is not possible even with the addition of nitrate (Figure 2). The stoichiometry for nitrate reduction in C. ljungdahlii is proposed as follows: 4 H₂ + 2 H⁺ + NO₃^– + 1.5 ADP + 1.5 P_i ⇌ 4 H₂O + NH₄⁺ + 1.5 ATP with Δ_rG’₀ = −150 kJ/mol H₂ (Thauer et al., 1977; Emerson et al., 2019). This mechanism would require electron bifurcation from the hydrogenases and the activity of the RNF complex, but would then provide ATP completely independent of the Wood-Ljungdahl pathway (or more general, independent of the carbon metabolism) (Buckel and Thauer, 2018; Emerson et al., 2019). Thus, we hypothesized that nitrate reduction in C. ljungdahlii requires a functional RNF complex for a correct electron balance. Indeed, non-growing cells of both C. ljungdahlii ΔRNF and C. ljungdahlii ΔrseC were not able to reduce nitrate during autotrophy (Figure 2F). Further biochemical investigations on the enzyme activity under these conditions will have to provide more detailed insight on this hypothesis. However, nitrate reduction still proceeded in both deletion strains during heterotrophy (Supplementary Figure 1F). In C. ljungdahlii ΔRNF a functional RNF complex was not present during heterotrophy because the RNF-complex encoding genes were deleted, but the required reducing equivalents for nitrate reduction were likely provided by glycolysis (Figure 6). In contrast, in C. ljungdahlii ΔrseC, nitrate reduction was not impacted during heterotrophy, because the RNF complex genes were deregulated (expressed) under these conditions and a functional RNF complex was formed (Figures 3, 6). It remains to be answered whether there is a direct interplay between the nitrate reductase and the RNF complex, and whether this interplay is different during heterotrophy and autotrophy.

In general, the reduced growth indicated that C. ljungdahlii ΔRNF was not able to balance the electrons from glycolysis efficiently during heterotrophy. This led to the reduction in biomass and acetate production, while ethanol production was completely absent in heterotrophic cultures of C. ljungdahlii ΔRNF. This indicates that reducing power for a further reduction of acetate was not available (Supplementary Table 5 and Supplementary Figure 1D). In the batch experiments of Emerson et al. (2019), C. ljungdahlii WT did not produce considerable amounts of ethanol when growing with nitrate (and carbon dioxide and hydrogen). When C. ljungdahlii WT was cultivated in pH-controlled bioreactors under continuous conditions, enhanced biomass and increased ethanol production rates were observed (Klask et al., 2020). This observation could not be fully explained yet, but it was assumed that electrons are used concomitantly for the reduction of nitrate and for the reduction of acetate. This distribution of electrons changed in the absence of the nitrate reductase in the C. ljungdahlii Δnar strain and higher maximum ethanol concentrations were observed already in batch conditions. Thus, the overall electron balance between fermentation products, besides the loss of nitrate reduction activity, is impacted by the deletion of the nar gene cluster. A possible explanation for increased ethanol production of the C. ljungdahlii Δnar strain during heterotrophy could be that more reducing equivalents (e.g., NADH, NADPH) are available to the alcohol dehydrogenases (ADHs), which would be used to reduce nitrate in the wild type, and which predominantly catalyze the reduction of acetyl-CoA to ethanol during heterotrophy but not during autotrophy in C. ljungdahlii (Schuchmann and Müller, 2014; Richter et al., 2016; Liew et al., 2017). Consequently, less acetate should be produced by the C. ljungdahlii Δnar strain. Indeed, the C. ljungdahlii Δnar strain produced only 66% (P ≤ 0.001) of the acetate concentration that was measured for the wild type, when growing under heterotrophic conditions with nitrate (Supplementary Table 1 and Supplementary Figure 1C). Furthermore, this would argue for NADH or NADPH as the electron donor of the nitrate reductase as proposed by Emerson et al. (2019). On the contrary, we observed increased ethanol concentrations also in the presence of ammonium instead of nitrate as nitrogen source. This might be due to an involvement of the nitrate reductase in other processes, but could also be due to genetic polar effects in the nar deletion strain. In addition, nitrate reduction could be regulated and controlled differently during autotrophy and heterotrophy. It remains elusive, how the change in the distribution of electrons affects other NADH-dependent metabolic pathways in more detail. While further research, such as the investigation of intracellular NAD(H)/NADP(H) levels, is required to understand the regulatory mechanisms during autotrophy and the mechanism of energy conservation during nitrate reduction, with this work, we provide a deeper insight into the autotrophic metabolism and nitrate reduction in C. ljungdahlii.