Introduction of glycine synthase enables uptake of exogenous formate and strongly impacts the metabolism in Clostridium pasteurianum

Autotrophic or mixotrophic use of one‐carbon (C1) compounds is gaining importance for sustainable bioproduction. In an effort to integrate the reductive glycine pathway (rGP) as a highly promising pathway for the assimilation of CO2 and formate, genes coding for glycine synthase system from Gottschalkia acidurici were successfully introduced into Clostridium pasteurianum, a non‐model host microorganism with industrial interests. The mutant harboring glycine synthase exhibited assimilation of exogenous formate and reduced CO2 formation. Further metabolic data clearly showed large impacts of expression of glycine synthase on the product metabolism of C. pasteurianum. In particular, 2‐oxobutyrate (2‐OB) was observed for the first time as a metabolic intermediate of C. pasteurianum and its secretion was solely triggered by the expression of glycine synthase. The perturbation of C1 metabolism is discussed regarding its interactions with pathways of the central metabolism, acidogenesis, solventogenesis, and amino acid metabolism. The secretion of 2‐OB is considered as a consequence of metabolic and redox instabilities due to the activity of glycine synthase and may represent a common metabolic response of Clostridia in enhanced use of C1 compounds.

Desulfovibrio desulfuricans (Sánchez-Andrea et al., 2020), Cupriavidus necator , and Corynebacterium glutamicum (Hennig et al., 2020;Tuyishime et al., 2018;Witthoff et al., 2015). For the most studied Clostridia, namely Clostridium acetobutylicum, foundations for the utilization of the reductive acetyl-CoA pathway (rAP) were laid by studying the expression and activity of acetyl-CoA synthase (Carlson & Papoutsakis, 2017;Fast & Papoutsakis, 2018). In general, however, non-model microorganisms have received less attention in the effort of integration of C1-utilization pathways mostly due to the lack of genetic tools, and probably also the less competitiveness compared with using native C1-fixating microorganisms.
Our long-term goal is to construct an anaerobic heterotrophic strain capable of synthesizing valuable chemicals with carbons derived from fixated C1 compounds and nitrogen from molecular nitrogen, in combination with energy and electrons derived from inexpensive waste materials, such as raw glycerol, even partially from electric currents. In this regard, Clostridium pasteurianum, a gram-positive, anaerobic, endospore-forming, and molecular nitrogen fixating bacterium with excellent solvent producing properties, was chosen to be engineered into a C1-utilizing model strain. For metabolic engineering of C. pasteurianum, one of the largest challenges lies in its low transformation efficiency. Although an in vivo methylation and electro-transformation protocol based on Pyne et al. (2013) was established, and C. pasteurianum R525 with improved electro-competence was isolated in our laboratory (Schmitz et al., 2019), transformation of several variations of pMTL-based plasmids failed in follow-up studies (data not shown). While the reason for the transformation failure remains unclear, we observed that smaller plasmids, such as pMTL85141 as a non-expression plasmid (Heap et al., 2009), were able to yield higher transformation efficiencies.
Thus, it was reasonable to assume that chances of being rejected by the host strain due to restriction-modification system or other sequence-specific incompatibilities might be reduced by minimizing the plasmid size and complexity. Therefore, we approached meeting the criteria of simplicity and compliance with the native cellular machinery, using a design based on the principle of orthogonality (Pandit et al., 2017).
Comparing natural and synthetic C1-assimilation pathways, the reductive glycine pathway schemed in Figure 1 represents a very promising one (Bar-Even, 2016;Bar-Even et al., 2013). Genomic analysis of genes required by rGP revealed that glycine synthase (or glycine cleavage system) is the only missing reaction step in C. pasteurianum. Comprised of four proteins (T, H, P, and L) the glycine synthase converts 5,10-methylene-THF, CO 2 , NH 3, and NADH to glycine (Kikuchi et al., 2008). Interestingly, many species among the well-known Clostridia lack the glycine synthase/glycine cleavage system (Table S1). This makes them suitable candidates to investigate the effects in a neutral and orthogonal manner without F I G U R E 1 Reductive glycine pathway and sources of formate and CO 2 for Clostridium pasteurianum. Formate is assimilated through the partial folate cycle and enters the glycine synthase in a reduced state. Tetrahydrofolate (THF) acts as a carrier of C1-unit and provides two of the three C1 units for the reductive glycine pathway, while the third carbon is derived from CO 2 . Glycine synthase as a reverse operating glycine cleavage system (gcvT: aminomethyltransferase, T-protein; gcvP A P B : glycine dehydrogenase, P-protein; lpd: dihydrolipoyl dehydrogenase; L-protein, gcvH: H-protein) converts 5,10-methylene-THF, NH 3 , and CO 2 to glycine, which is further converted to serine via serine hydroxymethyltransferase. After deamination of serine, one molecule of pyruvate is generated, which is further channeled into the central metabolic pathways or into biomass [Color figure can be viewed at wileyonlinelibrary.com] interference through a parallel reaction system. As an effort to integrate rGP in C. pasteurianum, genes of the glycine synthase from Gottschalkia acidurici (f. Clostridium acidurici; Poehlein et al., 2017) were chosen, because it is one of the few organisms operating the glycine synthase natively in the direction of glycine synthesis (Gariboldi & Drake, 1984;Waber & Wood, 1979). In this study, genes of glycine synthase from G. acidurici were cloned and introduced into C. pasteurianum, which was then characterized using glycerol, glucose, and formate as feed compounds. By comparing the mutant with the wild-type strain under variation of cultivation conditions, the metabolic impact of introducing glycine synthase into C. pasteurianum was addressed.

| Chemicals
Agarose was purchased from Biozyme Scientific GmbH and molecular nitrogen from Westfalen AG. 13 C labeled sodium formate (99%) from Sigma-Aldrich/Merck was used for 13 C-labeling experiments.
All the other chemicals of analytical grade were purchased either from Carl Roth or from Sigma-Aldrich/Merck.

| Strains and plasmids
Strains and plasmids used in this study are listed in Table 1. Primers and their purpose are described in Table 2 2.3 | Media, growth conditions, and storage of bacteria E. coli strains were stored in Roti-Store cryo vials (Carl Roth) and cultivations were performed in LB medium (tryptone, 10 g L −1 ; yeast extract, 5 g L −1 ; NaCl, 5 g L −1 ) at 37°C in 50 ml conical tubes or Erlenmeyer flasks with baffles. If required, additional 1.5% (w/v) agar was added. For selection, chloramphenicol and kanamycin were supplemented to final concentrations of 25-34 μg ml −1 and 35-50 μg ml −1 , respectively.
C. pasteurianum strains were stored at −80°C as 20% (v/v) glycerol stocks in 1.8 ml cryovials (VWR). For cultivation, glycerol stocks were first inoculated to serum bottles containing 2 × YTG medium (tryptone, 16 g L −1 ; yeast extract, 10 g L −1 ; glucose, 5 g L −1 ; Pyne et al., 2013)  formate was used. For the labeling experiment, concentrations of K 2 HPO 4 and KH 2 PO 4 were increased to 3 g L −1 , CaCO 3 was not included, yeast extract, 13 C sodium formate, and glucose were supplemented at 1 g L −1 , 6 g L −1 , and 6 g L −1 , respectively. All liquid medium for serum bottles were adjusted to pH 6.5 by titration with HCl and flushed with O 2 -free N 2 for 20 min at 90°C before sterilization at 121°C. Agar plates with 2 × YTG medium were prepared with an additional 1.5% (w/v) agar, which was cultivated in anaerobic jars containing Oxoid AnaeroGen (Thermo Fisher Scientific). For the mutant strain harboring pMTL-GCSY1, thiamphenicol (Tm) was supplemented to a final concentration of 7-10 μg ml −1 .  Verification PCR for uptake of pMTL-GCSY1 via re-extraction from C. pasteurianum HONG ET AL.

| Electro-transformation of C. pasteurianum
For the electro-transformation of C. pasteurianum R525 strain, previously described protocols by Pyne et al. (2013) and Schmitz et al.
(2019) were adapted. Briefly, an overnight culture was grown in 2 × YTG medium from a glycerol stock, and used to inoculate a prewarmed fresh 2 × YTG medium. When an OD 600 of approx. 0.45 was reached, the C. pasteurianum R525 culture was added with glycine days, colonies were visible, which were re-streaked onto new 2 × YTG agar plates with Tm. After further incubation for 5 days, freshly grown colonies were picked and inoculated into 2 × YTG medium supplemented with Tm. Glycerol stocks were prepared from 50 h grown cultures. The successful transformation was verified by PCR using re-extracted plasmids from the selected C. pasteurianum strain.

| Analytical methods
Cell concentrations were measured turbidometrically at 600 nm and a previously determined conversion factor of 0.336 g L −1 cell dry weight per unit cell density (Groeger et al., 2017) was used to calculate cell dry weight. Soluble extracellular metabolites were measured via HPLC (KNAUER) on an Aminex HPX-87H column (300 × 7.8 mm; Bio-Rad) at 60°C, using 5 mM H 2 SO 4 as mobile phase at a flow rate of 0.6 ml min −1 , and a refractive index detector and an ultraviolet detector. For fermentations, the off-gas flow rate was measured by an EL-FLOW flowmeter (Bronkhorst) and gas composition was determined using a Balzers Omnistar GSD 300 Mass Spectrometer (Pfeiffer Vacuum GmbH). For 13 C-labeling experiments, approx. 0.5 mg of dry biomass was used to prepare proteinogenic amino acids as described by Zamboni et al. (2009), including the hydrolysis of protein with 6 M HCl, and the derivatization of proteinogenic amino acids with N-terbutyldimethylsilyl-Nmethyltrifluoroacetamide/terbutyldimethyl-chlorosilane. The measurement of isotopomer abundances of proteinogenic amino acids was performed on a gas chromatograph coupled to a mass spectrometer (GC/MS) (Agilent Technologies) using an Agilent HP5 ms column (30 m × 0.25 mm, 0.25 µm), with helium as carrier gas at a flow rate of 1 ml min −1 . The temperature profile was as following: 80°C held for 1 min, increase at 20°C min −1 to 120°C, increase at 4°C min −1 to 270°C, increase at 20°C min −1 to 290°C, and held for 2 min. Correction of the natural abundance of isotopes was performed using IsoCor v2 (Millard et al., 2019).
The maximum growth rate was calculated with linear fit tool via Origin 2020 (OriginLab). Further calculations were performed via Excel (Microsoft). For batch fermentations, averages of five sampling points from end-exponential phase were used for calculation. Carbon and electron recoveries as measure for analytical completeness were calculated according to Equations (1) and (2) (3) and (4), respectively.  2.6 | Reverse transcription-polymerase chain reaction C. pasteurianum R525 strain and pMTL-GCSY1 harboring mutant strain were grown in serum bottle (25 ml of modified Biebl medium supplemented with 15 g L −1 glucose and 1 g L −1 sodium formate) at 35°C and harvested after 22 and 25 h, respectively. Total RNA was extracted using Aurum Total RNA Mini Kit (Bio-Rad) with lysozyme purchased from Carl Roth (≥45,000 FIP U mg −1 ). Reverse transcrip-tion of RNA to cDNA was performed by using the reaction setup with gene-specific primers of iScript Select cDNA Synthesis Kit (Bio-Rad) including DNase I treatment. The manufacturer's instructions were followed for both kits. The obtained cDNA was qualitatively verified by CloneAmp HiFi PCR Premix and by using cDNA as a template. pMTL-GCSY1 was used as a positive control. As a negative control, water sample, cDNA of reverse-transcribed RNA from C. pasteurianum R525 strain with corresponding primers, and total RNA of the mutant strain were used as a template. Primers for reverse transcription PCR and the subsequent PCR for qualitative analysis are listed in Table S2, which were designed using Primer-BLAST (Ye et al., 2012).

| Genomic analysis methods
To find the presence of glycine cleavage system/glycine synthase proteins UniProtKB database (www.uniprot.org; The UniProt Consortium, 2019) was utilized for the search of specific conserved proteins or domains (Pfam; El-Gebali et al., 2019; Table S1).

| Introduction of glycine synthase into C. pasteurianum
Genes of the glycine synthase (glycine cleavage system) from the gcv cluster in G. acidurici (gcvT, gcvH, gcvP A , and gcvP B ) and dihydrolipoamide dehydrogenase (lpd) were cloned into a previously constructed pMTL-based expression vector of pMTL85141-Cas9n (Schmitz et al., 2019). The resulting pMTL-GCSY1 plasmid was transformed into the C. pasteurianum R525 strain (termed WT), resulting in a C. pasteurianum mutant strain (termed as GCSY1) harboring the introduced glycine synthase. The successful plasmid uptake was verified by PCR after re-extraction from the GCSY1 strain. In addition, the heterologous expression of the genes was qualitatively confirmed by reverse transcription PCR (see - Figure S1). Overall, a transformation efficiency of (4.9 ± 0.4) × 10 −1 that unknown barriers other than the restriction-modification system hinder foreign DNA transfer and the elevated competency as it was reported for R525, H1 and H3 strains was rather due to spontaneous and less controllable mutations.
3.2 | Metabolic and redox burdens led to secretion of 2-oxobutyrate and 3-hydroxypropionic aldehyde; formate as glycine synthase trigger The GCSY1 strain showed deviations in growth pattern, secreted metabolites in serum bottle studies. When glycerol stocks of this strain were inoculated into 2 × YTG medium, both the lag-phase and the growth duration were prolonged in comparison with the WT strain from approx. 8-20 h and 16-40 h, respectively. Since the complex 2 × YTG medium is rich in nutrients the reason for the observed growth-inhibition was believed to be due to metabolic burden or instability caused by the introduction of glycine synthase, not the unavailability of certain nutrients in the medium. To further study the impact, pH-uncontrolled cultivations on glycerol in serum bottles with Biebl medium were performed with or without the supplementation of sodium formate and compared to the WT strain (see aceticum, also led to the secretion of 2-OB in trace amounts, when CO 2 was fixated to acetate over the reductive acetyl-CoA pathway (rAP).
Interestingly, these two Clostridia strains naturally harbor a glycine cleavage system (see Table S1). Assuming that a common causality for 2-OB secretion exists between the work of Nevin et al. (2011) and our observation, consumed electric current for acetogenesis in C. ljungdhalii and C. aceticum may pose a similar triggering effect as the redox shift of GCSY1 leading to 2-OB secretion. Supporting this hypothesis is the observation that the consumption of glycerol as a more reduced substrate instead of glucose by GCSY1 led to elevated 2-OB secretion in our study (see Figure 2b).
In addition to 2-OB, 3-hydroxypropionic aldehyde (3-HPA) was also found to be secreted into the medium in the culture of GCSY1 grown on glycerol but only in the presence of additionally supplemented formate. As shown in Figure S2, an asymmetrical HPLC-UV peak typical of 3-HPA in a dynamic equilibrium between its monomer, hydrate, and dimmer in aqueous solution was observed, as described in the HPLC analysis of 3-HPA by Burgé et al. (2015).
Based on the phenomenon for several natural 1,3-PDO producers that 3-HPA accumulation is caused by an insufficient supply of reducing equivalents (NADH; Barbirato et al., 1998;Hao et al., 2008;Maervoet et al., 2016;Sauvageot et al., 2000;Wang et al., 2003) and absence of a direct metabolic link between 3-HPA and formate, we concluded a redox imbalance based on glycine synthase, which is triggered by formate addition.
Natively, the solventogenic pathway (see Figure 3b) of glycerol reduction over 3-HPA to 1,3-PDO is expected to be the major NADH oxidation route in glycerol fermentation of C. pasteurinaum which is crucial for redox homeostasis (Dabrock et al., 1992

| Native uptake of isotopically labeled formate
To characterize the formate-related metabolism of C. pasteurianum, 13 C-labeled formate was supplemented isotopomer distributions of the proteinogenic amino acids were profiled (Figure 4). For the WT strain, glycine should be synthesized via the native biosynthesis pathway from pyruvate either over serine or threonine due to the absence of glycine synthase (Figure 3a), and therefore, we expected only the appearance of M + 0 isotopomer for glycine (because the natural abundances of isotopes were subtracted). To our surprise, in the WT strain, glycine was found to have over 8% of M + 1 isotopomer ( Figure 4b). This indicates that 13 C from labeled formate was incorporated into either the carboxyl or the aminomethyl group of glycine.
Based on this result and the study of Dainty and Peel (1970), we propose that a circular amino acid interconversion pathway exits in the native metabolism of C. pasteurianum as follows: C 3 -central carbon metabolite (pyruvate and derivatives) ↔ oxaloacetate → aspartate → threonine → glycine ↔ serine ↔ C 3 -central carbon metabolite (similar to the so-called serine-threonine-cycle; see carboxyl or carbonyl group in pyruvate). We hypothesize a reversed pyruvate formate lyase reaction which yields in 13 C-labeling of the carboxyl group in pyruvate (see Figure 4c) as previously demonstrated in vitro for other Clostridia (Thauer et al., 1972) and applied in vivo in E. coli (Zelcbuch et al., 2016). Alternatively, fixation of CO 2 from oxidized 13 C-formate via pyruvate carboxylase as anaplerosis and interconversion within symmetrical TCA intermediates yields likewise in labeled pyruvate. However, the incompleteness of genes for the citrate cycle and unclear synthesis routes of TCA intermediates  render a detailed investigation difficult.
Nevertheless, both labeling patterns result in labeled central C 3 metabolite(s), which transfer labeled carbon to form glycine either over aspartate and threonine (with over 24% M + 1) or over serine.
The unusually over 89% high M + 1 abundance and over 7% C-formate fixation routes (from left to right): reverse pyruvate formate lyase based formate uptake (double-lined; blue), pyruvate carboxylase with interconversions to symmetrical TCA intermediates (dashed; pink), and serine hydroxymethyltransferase (dash-dotted; orange). Gray and dotted arrow indicates 13 C-labeled CO 2 from the partial oxidation of labeled formate via formate dehydrogenase [Color figure can be viewed at wileyonlinelibrary.com] abundance of M + 2 in serine are attributed to formate uptake via serine hydroxymethyltransferase from glycine (see Figure 4c) in addition to the canonical biosynthesis from labeled C 3 metabolite.
Furthermore, alanine and glutamate synthesized from pyruvate were found to have approx. 15% and 7% M + 1 isotopomer, respectively.
Cultivating the GCSY1 mutant with exogenously added 13 Cformate, elevated acidogenesis in GCSY1 led to a drastic decrease of pH, despite increased phosphate buffer concentration in the medium was used. This induced early growth inhibition and partial sporulation of the cells. The resulting only 1.9-fold increase of the total biomass disqualified qualitative analysis of the functionality of glycine synthase in the GCSY1 mutant by using 13 C-labeled formate.
Hence, an alternative approach was chosen to quantify the effect of glycine synthase on C. pasteurianum: since the costly usage of 13 Clabeled compounds can be only conducted in small mL-scales, we proceeded to scale-up the cultivation volumes enabling measurements of absolute concentrations of C1 compounds. Thus, batch fermentations in 1.5 L and continuous fermentations in 200 ml scales were performed to enhance the resolution of metabolic characterization. In addition, off-gas analysis was conducted to examine gaseous C1 production or consumption for GCSY1.

| Shift in solventogenesis and acidogenesis, reduction of carbon loss as C1 units without exogenous formate
Since formate and CO 2 are native by-products of C. pasteurianum metabolism, 1.5 L batch fermentations with glycerol as sole substrate (80 g L −1 ) were conducted to analyze the impact of glycine synthase on natively synthesized C1 compounds (formate and CO 2 ; see Figure S3).
Carbon recovery of approx. 95.3 ± 5.5% to 99.3 ± 0.9% and electron recovery from approx. 92.0 ± 5.4% to 103.0 ± 2.4% confirm that major metabolites were included in the analysis (Figure 5b). Since GCSY1 consumed less substrate, the following discussion is based on the distribution of carbon per unit of consumed glycerol. (a) F I G U R E 5 Analysis of batch fermentation data of Clostridium pasteurianum R525 (WT) and C. pasteurianum R525 glycine synthase mutant (GCSY1; n = 2). pH-controlled batch fermentations with glycerol (initial concentration 80 g L −1 ) as the main carbon source were conducted in duplicates for both strains. The specific maximum growth rates were calculated from the growth curves of the early-exponential phase (a). Carbon and electron recoveries and proportion of carbon loss as C1 compounds (formate and CO 2 ) (b), substrate-specific yields (c), and carbon distributions (d) were calculated from an end-exponential phase of the batch fermentation Shifts in the carbon fluxes were observed for GCSY1, which are attributed to glycine synthase introduction (Figure 5d): decreased carbon fluxes toward butanol (from 16.2 ± 5.4% to 3.5 ± 1.3%), carbon dioxide (from 14.5 ± 0.8% to 6.4 ± 2.3%) and biomass (from 8.3 ± 0.2% to 3.2 ± 0.3%), but increased carbon fluxes to acid production (from 10.7 ± 1.9% to 24.6 ± 8.1% for all detected acids), which is in agreement with substrate-specific yields as shown in Figure 5c. Metabolic burden via glycine synthase is observable by the decrease in the maximum growth rate from 0.309 ± 0.018 h −1 to 0.079 ± 0.002 h −1 (Figure 5a) and the decreased substrate-specific biomass yield from 63.15 ± 1.44 mg g −1 to 25.30 ± 2.37 mg g −1 as shown in Figure 5c. Similar to previously seen in serum bottle experiments, 2-OB secretion was observed ( Figure S3), which accounted for up to 3.1 ± 0.4% of the carbon flux.
As the major impact of glycine synthase integration, carbon loss by CO 2 production from total consumed carbons was more than halved, despite the native formate uptake route. which poses an interesting strategy for bioprocesses for organic acid production (e.g., butyrate).

| Exogenous formate uptake with increasing formate supply
To analyze the ability of exogenous formate uptake and better comparability with the WT, glycerol was replaced by glucose (10 g L −1 ), and substrate-limited continuous fermentation was performed at a dilution rate of 0.1 h −1 with 0, 1, 2, and 4 g L −1 sodium formate (see Figure S4). Interestingly, 2-OB secretion was not observable anymore as soon as the continuous operation mode was initiated. Thus, it appears that usage of more oxidized substrate in limited availability diminishes this metabolic imbalance as observed in batch fermentations with glycerol as the sole substrate. The continuous supply of 1 g L −1 yeast extract in the feed resulted in carbon and electron recoveries over 100% for all analyzed steadystate sampling points of WT and GCSY1 (see Figure 6a). The glucoselimited condition resulted in similar glucose consumption rates between WT and GCSY1 (between 1.050 ± 0.001 mmol h −1 and 1.094 ± 0.000 mmol h −1 ) as shown in Figure 6c. While the biomass production rate for WT decreased by approx. 35% from 36.21 ± 1.28 mg h −1 with increasing sodium formate concentration, the GCSY1 strain showed higher tolerance with only a decrease of approx. 8% from 30.71 ± 0.29 mg h −1 (Figure 6d).
In regard to acidogenesis a tendency was observed: with increasing concentrations of sodium formate in the feed, the substratespecific production rates of acids for WT were increased stepwise from 3.291 ± 0.137 mmol g −1 h −1 by approx. 40%, 98%, and 107%, at 1, 2, and 4 g L −1 sodium formate, respectively. For GCSY1, an increase of 22% from 4.032 ± 0.038 mmol g −1 h −1 was observed, when 1 g L −1 sodium formate was supplemented. However, a further increase in sodium formate did not result in an elevation of acidogenesis (see Figure 6e). As reported for C. acetobutylicum and C.
beijerinckii, sodium formate supply in low concentration creates oxidative stress and leads to the so-called "acid crash" (Cho et al., 2012;Wang et al., 2011). The comparison of the specific acid production rate in regard to the biomass production rates (see Figure 6d) suggests elevated stress response in the WT strain due to increased supply of exogenous formate leading to increased ATP demand, which is compensated by elevated acidogenesis. In GCSY1, however, the increase of acidogenesis is confined despite the integration of glycine synthase. It appears in this context that the expected increase of ATP demand is negligible in comparison to the stress response related ATP demand as seen in WT. Solventogenesis was low for both strains and show no clear trends (see Figure 6f). This is conceivable, since glucose is less reduced than glycerol and solventogenesis should not be as dominant as that in glycerol fermentation for both strains.
The triggering effect of formate was most noticeable by production rates of C1 compounds: increased sodium formate supply increased gradually CO 2 production in WT from 3.894 ± 0.137 mmol g −1 h −1 to 7.223 ± 0.093 mmol g −1 h −1 and in GCSY1 from 3.976 ± 0.038 mmol g −1 h −1 to 5.757 ± 0.065 mmol g −1 h −1 (see Figure 7a). However, at approx. 24% and 20% lower CO 2 production was observed for GCSY1 at 2 and 4 g L −1 sodium formate concentration. Further, the elevation of exogenous formate led to its consumption by GCSY1. Specific consumption rates of 0.085 ± 0.068 mmol g −1 h −1 and 0.290 ± 0.016 mmol g −1 h −1 at 2 and 4 g L −1 sodium formate concentration, respectively, were observed as shown in Figure 7b, in contrast to increasing formate production rates in WT. In the discussed context, a consumption rate in the order of magnitude of 10 2 µmol g −1 h −1 may appear almost irrelevant, the resulting consumption was comprised of the consumption of exogenous formate added to the consumption of synthesized formate by its native metabolism. As shown by the specific C1 unit production rates depicted in Figure 7c, a reduction of up to 30% in total C1 production (at 2 and 4 g L −1 sodium formate in the feed) was observed. Furthermore, oxidation of formate to CO 2 and glycine cleavage can be excluded, since exogenous formate consumption did not lead toward elevated CO 2 production verifying the uptake and fixation of exogenous C1 compounds. It also implies the contrast of (a) F I G U R E 6 Analysis of continuous fermentation data of Clostridium pasteurianum R525 (WT) and C. pasteurianum R525 glycine synthase mutant (GCSY1; n = 3). Steady-state samples of pH-controlled continuous fermentation with glucose (10 g L −1 ) and varying sodium formate concentrations are analyzed and shown as carbon (a), electron recoveries (b), glucose consumption (c), biomass production rates (d), specific acid (e), and solvent production rates (f) (a) (b) (c) F I G U R E 7 Specific C1 compound production and consumption rates from continuous fermentation of Clostridium pasteurianum R525 (WT) and C. pasteurianum R525 glycine synthase mutant (GCSY1; n = 3). Steady-state measurements of pH-controlled continuous fermentation with glucose (10 g L −1 ) and varying sodium formate concentrations for WT and GCSY1 are compared for specific CO 2 (a), formate (b), and C1 (CO 2 and formate) production rates (c) the presented approach to the engineering of aerobic microorganisms, where oxidation of formate and oxidative phosphorylation provide energy and reducing power (Bang & Lee, 2018;Yishai et al., 2018). It appears the additional burden for energy and reducing power was only compensated by lowered biomass production. This may pose a general constraint of this approach for C1 fixation toward valorized chemicals. However, it simultaneously demonstrates the ability to improve the overall carbon yield for the biosynthesis of chemicals through lowered production of biomass and CO 2 . To pursue in-depth tracing of fixated C1 compounds, the growth inhibitory effect of glycine synthase needs to be addressed first by further engineering. Still, closed carbon and electron balances and uptake of C1 compounds implies the integration of fixated carbons into detected metabolites or biomass.

| CONCLUSION
For the first time, an artificial formate assimilation pathway was realized in a Clostridia bacterium by introducing glycine synthase of G. acidurici into C. pasteurianum. The impact of this pathway engineering for the cellular metabolism and native C1 biosynthesis was analyzed under the variation of substrate and cultivation conditions.
The growth inhibitory effect of glycine synthase integration leading to early growth cessation disabled to definitively confirm glycine biosynthesis from 13 C-formate. However, up to 46% reduced native C1 compounds production and uptake of exogenous formate coupled with lowered CO 2 production in cultivation experiments verified the fixation of C1 compounds-solely as an outcome of glycine synthase integration. Unexpected 2-OB secretion and 3-HPA accumulation as metabolic responses toward integration of glycine synthase represent alterations of cellular, energy, and redox metabolism. The discussed cellular response provides insight into central metabolism and pinpoints to serious metabolic imbalances in C. pasteurianum, which may be also prevalent in other Clostridia lacking glycine synthase/cleavage system and represent essential engineering targets for further improvement on C1 fixation.

ACKNOWLEDGMENT
This study did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Open access funding enabled and organized by Projekt DEAL.

CONFLICT OF INTERESTS
The authors declare that there are no conflict of interests.