Formate Utilization by the Crenarchaeon Desulfurococcus amylolyticus

Formate is one of the key compounds of the microbial carbon and/or energy metabolism. It owes a significant contribution to various anaerobic syntrophic associations, and may become one of the energy storage compounds of modern energy biotechnology. Microbial growth on formate was demonstrated for different bacteria and archaea, but not yet for species of the archaeal phylum Crenarchaeota. Here, we show that Desulfurococcus amylolyticus DSM 16532, an anaerobic and hyperthermophilic Crenarchaeon, metabolises formate without the production of molecular hydrogen. Growth, substrate uptake, and production kinetics on formate, glucose, and glucose/formate mixtures exhibited similar specific growth rates and similar final cell densities. A whole cell conversion experiment on formate revealed that D. amylolyticus converts formate into carbon dioxide, acetate, citrate, and ethanol. Using bioinformatic analysis, we examined whether one of the currently known and postulated formate utilisation pathways could be operative in D. amylolyticus. This analysis indicated the possibility that D. amylolyticus uses formaldehyde producing enzymes for the assimilation of formate. Therefore, we propose that formate might be assimilated into biomass through formaldehyde dehydrogenase and the oxidative pentose phosphate pathway. These findings shed new light on the metabolic versatility of the archaeal phylum Crenarchaeota.


Introduction
Formic acid, a monocarboxylic acid, is one of the simplest organic compounds. It is a colourless liquid with a pungent odour [1]. The freezing and the boiling points of formate are 8.3 and 100.7 • C, respectively. The carbon of formic acid is a poor electrophile. This characteristic makes formic acid a strong acid (pK a = 3.75), e.g., compared to acetic acid (pK a = 4.75) [2]. The salt of formic acid is formate. Formate can be produced using various routes such as the electrochemical reduction of CO 2 [3][4][5][6][7], photo-reduction of CO 2 [8], hydrogenation of CO 2 [9,10], selective oxidation of biomass [11,12], partial oxidation of natural gas [13], and hydration of syngas (e.g., carbon monoxide) [14]. In 1670, John Wray had already isolated formate from ants [15], and later it was named after "formicidae", the ant family. Formate has numerous functions in biological systems, such as serving as an irritant in the sprayed venom of some ant species [16], as antibacterial substance [17], and it is used for food preservation, as cosmetic additive, or as pesticide [18]. Moreover, it can be used as a non-flammable liquid fuel [19], Until the discovery of the hyperthermophilic archaeon Thermococcus onnurineus NA1 [49,50], it was assumed that the reaction is only thermodynamically possible when a methanogenic or sulphate-reducing partner is present to remove H 2 , which is one of the end products of syntrophic formate oxidation [51][52][53][54]. T. onnurineus is able to oxidise formate into H 2 /CO 2 , and coupling this reaction to chemiosmotic ATP synthesis [25,55]. The metabolic pathway responsible for formate oxidation in T. onnurineus was characterised by using proteomics [56]. It was found that formate oxidation proceeds via a membrane-bound enzyme system, comprised of Fdh, and a membrane-bound hydrogenase (Mfh), a sodium-proton (Na + /H + ) antiporter (Mnh) and a Na + -dependent ATP synthase [57]. Furthermore, it was revealed that several copies of hydrogenase gene clusters (fdh-mfh-mnh) are present in T. onnurineus. The hydrogenase genes in fdh2-mfh2-mnh2 were upregulated more than 2-fold when formate was provided as sole energy source to T. onnurineus, which is an indication of the importance of the hydrogenase genes in the fdh2-mfh2-mnh2 gene cluster for growth coupled to formate oxidation and H 2 production [25].
Recently, a novel synthetic formate-fixing pathway was proposed, which is involved in the acetyl-CoA exchange to formyl-CoA and formate reduction to formaldehyde [58]. Formaldehyde can be integrated into the central carbon metabolism much easier than direct formate utilization, since it is a highly reactive compound [59]. However, the reduction potential of formate to formaldehyde is quite low under standard biochemical conditions [−650 mV≤ E • ≤−450 mV (6 ≤ pH ≤ 8; 0 ≤ I ≤ 0.25 mol L −1 )] [60]. Hence, the activation of formate requires an electron donor such as universal electron carrier NAD(P)H (−370 mV ≤ E ≤ −280 mV) [58]. One of the formate-fixing reactions is catalysed through the formaldehyde dehydrogenase (FoDH) enzyme, which directly converts formate to formaldehyde using NADH as a cofactor [61,62]. Investigating formate assimilation in other organisms could contribute to the identification of new formate-fixation pathways and its enzymes.
Desulfurococcus amylolyticus DSM 16532 [63] is an anaerobic, hyperthermophilic Crenarchaeon able to grow on a broad range of polymers and sugars [63,64]. Recently, physiological variables, such as the specific growth rate (µ), were determined for of D. amylolyticus grown on fructose or glucose in chemically defined medium [65]. A metabolic reconstruction revealed that D. amylolyticus contains all glucose metabolism-related genes and harbours several genes for H 2 production, such as pyruvate ferredoxin oxidoreductase, glyceraldehyde-3-phosphate ferredoxin oxidoreductase, and two membrane-bound hydrogenases [64,65]. As growth on formate coupled to H 2 production by hyperthermophilic Archaea could be an opportunity in biotechnology [66,67], we investigated whether D. amylolyticus could be used for this purpose.
In this work, we examined the substrate uptake, growth, and production kinetics of D. amylolyticus grown on formate, glucose, and a mixture of glucose/formate in closed batch cultivation mode. The intention was to physiologically characterise the organism with respect to µ, cell-specific H 2 and CO 2 productivities, and maximum cell concentration. After this, the mass balance analysis of substrate uptake and product formation kinetics was performed in whole cell conversion experiments. Finally, we investigated and compared the metabolic capacity for formate utilization of D. amylolyticus to other formate metabolising microorganisms on the genomic level with the goal of revealing possible formate assimilation pathways.

Microorganism and Medium Composition
Desulfurococcus amylolyticus DSM 16532 [63,68] was purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ). The medium was prepared as previously described [64,65]. A modified DSMZ medium, No. 395, without yeast extract, powdered sulphur and glucose was prepared when only formic acid containing medium was used for growth of D. amylolyticus. The medium contained (per L): 0.33 g of NH 4 Cl; 0.33 g of KH 2 PO 4 ; 0.33 g of KCl; 0.44 g of CaCl 2 ·2H 2 O; 0.70 g of MgCl 2 ·6H 2 O; 0.50 g of NaCl; 0.80 g of NaHCO 3 ; 0.50 g of Na 2 S·9H 2 O; 1 mL of trace elements SL-10; and 10 mL of vitamin solution as previously described [64,65].

Closed Batch Cultivations
Closed batch cultivations were conducted in two different sets of experiments. The first set of experiments were designed as "end-point experiments" where growth was monitored at each time-point, whereas substrate uptake and production measurements were only performed at end of the experiment. The second set of experiments were designed as "time-point experiments" where substrate uptake, growth and production kinetics were observed and measured at each time-point during the experiment.
Cultures of D. amylolyticus were grown anaerobically at 0.2-0.3·10 5 Pa in a 100 Vol.-% N 2 atmosphere in a closed batch set-up as previously described [64,65]. Concentrations of carbon sources were adjusted to the same carbon concentration (C-mmol L −1 ). For end-point experiments the following carbon sources were individually tested: formic acid and glucose at 116.6 C-mmol L −1 , respectively. Formic acid at a concentration of 50 C-mmol L −1 , and glucose at a concentration of 66.6 C-mmol L −1 (total 116.6 C-mmol L −1 ) were added for the co-substrate experiment. Additionally, glucose and formic acid were tested as carbon sources at 66.6 and 50 C-mmol L −1 , respectively, to assess the difference in substrate uptake, growth and production kinetics between the co-substrate and single substrate experiment.
The pre-culture for inoculation was obtained from a formic acid pre-grown D. amylolyticus culture. All experiments were performed in quadruplicates together with a negative (un-inoculated) control and reproduced twice. Moreover, we performed an additional positive-negative (inoculated on medium without formic acid) control to observe growth kinetics on medium containing just vitamins and trace elements, without formic acid (Supplementary Figure S1). Pressure measurements of the serum bottle headspace were performed with a digital manometer (LEO1-Ei, −1-3bar rel, Keller, Germany) and the measurements were performed before samples were taken for microscope analysis. For time-point experiments, formic acid was used as the carbon source at concentration of 100 C-mmol L −1 . For each time-point, identical sets of serum bottles (quadruplicates) were prepared and inoculated. They were not manipulated until the destructive gas chromatography (GC) measurement (see below in GC section).

Whole Cell Conversion Experiments
D. amylolyticus was pre-grown on formic acid and harvested by centrifugation (Eppendorf Centrifuge 5415R, Eppendorf, Hamburg, Germany) for 20 min and 15,700 g. The supernatant was removed and the resulting pellet was washed with the respective medium. After the washing step, the cells were resuspended in buffer containing (per L): NH 4 Cl 0.33 g; KH 2 PO 4 0.33 g; KCl 0.33 g; CaCl 2 ·2H 2 O 0.44 g; MgCl 2 ·6 H 2 O 0.70 g; NaCl 0.50 g; NaHCO 3 0.80 g. Serum bottles of 120 mL were supplemented with formic acid (concentration of 20, 50, and 100 C-mmol L −1 ). Cultures (1·10 8 cells mL −1 ) were incubated for 5 and 12 h, respectively and GC analyses were immediately conducted afterwards. All experiments were performed in triplicates together with a negative (un-inoculated) control.

Cell Counting
D. amylolyticus cells were counted using a Nikon Eclipse 50i microscope (Nikon, Amsterdam, Netherlands) at each sampling point. The samples for cell count were taken from each individual closed batch run using syringes (Soft-Ject, Henke Sass Wolf, Tuttlingen, Germany) and hypodermic needles (Sterican size 14, B. Braun, Melsungen, Germany). An amount of 10 µL of sample was applied onto a Neubauer improved cell counting chamber (Superior Marienfeld, Lauda-Königshofen, Germany) with a grid depth of 0.1 mm.

Gas Chromatography
Time-point and end-point GC measurements were performed from serum bottles that remained without any manipulation after inoculation, except incubation in air bath until the GC measurement was performed. To analyse the gas composition inside the serum bottles from end-point and time-point experiments, destructive sampling was employed. The gas compositions were analysed using a GC (7890A GC System, Agilent Technologies, Santa Clara, CA, USA) with a 19808 Shin Carbon ST Micropacked Column (Restek GmbH, Bad Homburg, Germany) and provided with a gas injection and control unit (Joint Analytical System GmbH, Moers, Germany) as described earlier [64].

Formic Acid Analysis
The Formic Acid Assay Kit (Megazyme Inc., Bray, Ireland) was used for measurements of formic acid concentrations in samples which were previously diluted to the linear range of the assay kit to yield a formic acid concentration of 0.004-0.200 g L −1 . The microplate (Crystal Clear, Greiner Bio One) assay was applied according to manufacturer's instructions: a 255 µL reaction volume.

HPLC
The determination of sugars, volatile fatty acids (VFAs), organic acids, and alcohols were performed with high performance liquid chromatography (HPLC) system (Agilent 1100), consisting of a G1310A isocratic pump, a G1313A ALS autosampler, a Transgenomic ICSep ICE-ION-300 column, a G1316A column thermostat set at 45 • C, a G1362A RID refractive index detector, measuring at 45 • C (all modules were from Agilent 1100 (Agilent Technologies, CA, USA)). The measurement was performed with 0.005 mol L −1 H 2 SO 4 as solvent, with a flow rate of 0.325 mL min −1 and a pressure of 48-49 bar. The injection volume was 40 µL.

Data Analysis
For the quantitative analysis, the maximum specific growth rate (µ max [h −1 ]) and mean specific growth rate (µ mean [h −1 ]) were calculated as follows: N = N 0 ·e µt with N, cell number [cells mL −1 ]; N 0 , initial cell number [cells mL −1 ]; t, time [h] and e, Euler number. According to the delta cell counts in between sample points, µ was assessed. The CO 2 evolution rate (CER [mmol L −1 h −1 ] (C-molar)), the cell specific CO 2 productivity (qCO 2 , cell [mmol cell −1 h −1 ] (C-molar)) [67] was calculated from the end point gas composition of the non-manipulated serum bottles. The ash content and elementary composition of D. amylolyticus were presumed to relate to published results [69]. The elementary composition was used for the calculation of the mean molar weight, Carbon balance (C-balance) and the degree of reduction (DoR) balance of the corresponding biomass. DoR denotes the number of electrons an atom can donate, summed up for all the atoms of a molecule, or biomass elementary composition, divided by the number of carbon atoms in the molecule/biomass elementary composition. Yields of by-products were determined after HPLC measurement. The values were normalized according to zero (time-point zero) control and negative values (if there were any) were assumed as zero.

Gibbs Free Energy Calculations
Standard Gibbs free energy change (∆G 0 ) is used to describe whether a certain chemical reaction can be utilized for microbial energy conservation. However, the underlying thermodynamic calculations are usually standardized to 25 • C and 1 bar pressure. For thermophilic bioprocesses the physiological conditions differ significantly and, consequently, values have to be adapted, as especially temperature has a huge impact on thermodynamics [70]. In the present study, D. amylolyticus was cultivated at 80 • C. The recalculation method applied in the current study is based on previously published results [70], which provide standard state thermodynamic properties at temperatures up to 200 • C for a wide variety of anaerobic metabolic reactions. Moreover, they discuss the thermodynamic framework in detail and the application of the revised Helgeson Kirkham Flowers (HKF) equation of state to obtain paper. Unfortunately, the named study does not include data for formaldehyde, which were therefore obtained from another publication [71], and recalculated as specified. Finally, the Gibbs values for the overall reaction at standard concentrations (1 mol L −1 ) and pH of 7 (∆G 0 ) were calculated as previously described [70].

Genome Analysis
To investigate the formate metabolism of D. amylolyticus, the protein sequences from the whole genome of organisms where formate related pathways were previously described like T. onnurineus (Ton_GCF_000018365.1_ASM1836v1), Pyrococcus furious (Pfu_GCF_000007305.1_ASM730v1), E. coli K-12 (Eco_GCF_000750555.1_ASM75055v1), Methanosarcina barkeri (Mba_GCF_000195895.1_ASM19589v1), Thermoanaerobacter kivui (Tki_GCF_000763575.1_ASM76357v1) and A. woodii (Awo_GCF_000247605.1_ ASM24760v1) were obtained from the NCBI RefSeq [72] database. Homologous proteins involved in formate-related metabolism of D. amylolyticus were identified by using Basic Local Alignment Search Tool (BLAST) [73] against the manually sorted proteins from characterised enzymes (Supplementary  Table S1) with E-values and local identity cutoffs of <10 −10 and >25%, respectively. Orthologous genes (genome level) were obtained by pair-wise all versus all BLAST of the aforementioned organisms using the "OrthoFinder" tool [74]. Furthermore, the orthologous gene (gene in a different species that evolved from a common ancestral gene by speciation) groups (ortho-groups) containing the characterised enzymes related to formate-related metabolism were retrieved. After sorting the proteins of interest, enzyme complexes were identified by using bidirectional BLAST. Results for query coverage, E-value, and identity can be found in Supplementary Table S2 and Supplementary Table S3. In addition, the Pfam domains [75] of all the proteins in D. amylolyticus were also predicted for a further look into the domains of enzymes related to the formate metabolism.

D. amylolyticus Grows on Formate
Recently, we investigated growth characteristics of D. amylolyticus on cellulose, fructose, arabinose, glucose, lactose, maltose, starch, and sucrose. Moreover, we partially re-annotated the genome and metabolically reconstructed the central carbon metabolism [65]. According to the metabolic reconstruction of D. amylolyticus, it seemed to be possible that the organism could grow through metabolisation of formate. To elucidate the growth kinetics of D. amylolyticus on formate and to be able to compare them to glucose and formate-glucose mixtures, end-point experiments were designed to analyse growth kinetics in defined medium with each of the substrates at the same concentration (166.6 C-mmol L −1 ). Additionally, formate and glucose were tested at concentrations of 50 and 66.6 C-mmol L −1 , respectively, to be able to examine the growth kinetics at lower substrate concentrations. The results of the growth characteristics are shown in Figure 1. Cleary, D. amylolyticus did not grow on a medium where formic acid was omitted (Supplementary Figure S1). The lag time lasted approximately 125 h, until growth of D. amylolyticus on each of the substrates commenced.
The key physiological variables are presented in Table 1. The organism grew almost equally well on all substrates tested and at all concentrations. Astonishingly, the organism grew on formate as the sole energy source with a µ max of 0.032 h −1 . A slightly higher µ max of 0.035 and 0.036 h −1 was only obtained on glucose and formate/glucose, respectively. The differences in µ max might be explained by slightly different concentrations and hyperbolic relationship between µ and the substrate concentration [76,77]. In a previous study, we showed that D. amylolyticus comprises a µ max of 0.059 h −1 at a concentration of 166.5 C-mmol L −1 glucose [65]. In our previous studies [64,65], growth of D. amylolyticus resulted in low cell densities only, a characteristic shared with many other hyperthermophilic microorganisms [78]. The growth of D. amylolyticus on glucose was accompanied by lactate, acetate, and formate production, whereas growth on formate resulted in production of acetate, citrate, and ethanol (Table 2). All the substrate concentrations are given as C-mmol L −1 . A negative (un-inoculated) control and positive-negative (inoculated into medium where formic acid was omitted) control were performed in each set and no growth was observed.  Figure 1. Growth curves of D. amylolyticus on formate, glucose, and glucose/formate at different concentrations. A slightly higher µ could be obtained when glucose/formate was used as substrate.
All the substrate concentrations are given as C-mmol L −1 . A negative (un-inoculated) control and positive-negative (inoculated into medium where formic acid was omitted) control were performed in each set and no growth was observed.  Table 2. Productivity and yields of D. amylolyticus from glucose or formic acid metabolisation during closed batch end-point experiments. Following this, we examined the gaseous product formation spectrum from all quadruplicate closed batch experiments on formate, glucose, and formate/glucose. CO 2 was the only gas detectable in all growth experiments. The CO 2 concentrations are shown in Figure 2, and an overview of CO 2 productivities are presented in Table 2. In cultures which were grown on formate, the cumulative CO 2 levels were in the ppm range. However, during growth on glucose or formate/glucose, the cumulative CO 2 values were more than one order of magnitude higher. H 2 could not be detected in any growth experiment, which is in contradiction to experiments in bioreactors [64], but in agreement to our previous closed batch experiments [65]. This indicates that D. amylolyticus did not to use the electrons from formate or glucose to balance homeostasis by producing H 2 . However, instead it could be possible that the electrons were used to balance anaplerotic reactions, which indicates that during formate metabolization, CO 2 was assimilated into biomass through some of the several annotated CO 2 -fixing enzymes [65]. However, up to now it is not clear if the ATP for the CO 2 fixation is retrieved from substrate level phosphorylation via glycolysis or from chemiosmotic ATP production or the pseudo-TCA cycle. Following this, we examined the gaseous product formation spectrum from all quadruplicate closed batch experiments on formate, glucose, and formate/glucose. CO2 was the only gas detectable in all growth experiments. The CO2 concentrations are shown in Figure 2, and an overview of CO2 productivities are presented in Table 2. In cultures which were grown on formate, the cumulative CO2 levels were in the ppm range. However, during growth on glucose or formate/glucose, the cumulative CO2 values were more than one order of magnitude higher. H2 could not be detected in any growth experiment, which is in contradiction to experiments in bioreactors [64], but in agreement to our previous closed batch experiments [65]. This indicates that D. amylolyticus did not to use the electrons from formate or glucose to balance homeostasis by producing H2. However, instead it could be possible that the electrons were used to balance anaplerotic reactions, which indicates that during formate metabolization, CO2 was assimilated into biomass through some of the several annotated CO2-fixing enzymes [65]. However, up to now it is not clear if the ATP for the CO2 fixation is retrieved from substrate level phosphorylation via glycolysis or from chemiosmotic ATP production or the pseudo-TCA cycle. The results indicate that the CO2 production is very low in the cultures grown on formate compared to cultures grown on glucose or glucose/formate. All the substrate concentrations are given as C-mmol L −1 .

Compound and Concentration
During the time-point experiments (Figure 3), the observed growth kinetics showed a similar trend as in the end-point experiments (Figure 1). Production and consumption of citrate and CO2 are shown in Figure 3 and an overview of CO2 productivities and mass balances are presented in Table  3. The results of the closed batch time-point experiment indicate that CO2-fixation occurs. Hence, During the time-point experiments (Figure 3), the observed growth kinetics showed a similar trend as in the end-point experiments (Figure 1). Production and consumption of citrate and CO 2 are shown in Figure 3 and an overview of CO 2 productivities and mass balances are presented in Table 3. The results of the closed batch time-point experiment indicate that CO 2 -fixation occurs. Hence, during the first 300 h, CO 2 was produced and subsequently consumed within the next 200 h of the experiment. We have previously shown that D. amylolyticus harbours several genes, which might be involved in CO 2 fixation [65], and the results presented in this study suggest the possibility that CO 2 might be fixed through enzymes of the reductive citric acid cycle, as citric acid is one of the produced metabolites and consumed compounds.
during the first 300 h, CO2 was produced and subsequently consumed within the next 200 h of the experiment. We have previously shown that D. amylolyticus harbours several genes, which might be involved in CO2 fixation [65], and the results presented in this study suggest the possibility that CO2 might be fixed through enzymes of the reductive citric acid cycle, as citric acid is one of the produced metabolites and consumed compounds. Figure 3. Growth, substrate uptake, and production kinetics of D. amylolyticus on 100 C-mmol L −1 formate. The results indicate that CO2 and citric acid were produced and consumed completely during the cultivation and only after the consumption of CO2 and citric acid, acetic acid was produced.

D. amylolyticus Converts Formate to CO2, Acetate, Citrate, and Ethanol
To investigate the metabolism on formate in more detail, whole cell conversion experiments were performed. A whole cell conversion experiment has unique advantages, such as minimising side reactions and avoiding biomass production [79][80][81]. Therefore, we performed experiments with high cell densities (1·10 8 cells mL −1 ) using D. amylolyticus at formate concentrations of 20, 50, and 100 C-mmol L −1 in buffer. This experiment revealed astonishing findings. A summary of the obtained physiological variables is shown in Table 4. The highest formate uptake rate of 20.7 C-mmol L −1 was observed when D. amylolyticus was incubated at a concentration of 100 C-mmol L −1 formate. After 5 h of incubation, the production of small amounts of butyrate was detected in 50 and 100 C-mmol L −1 formate, and citrate, as well as ethanol, was only detected in one of the other. After 12 h of incubation, ethanol and low amounts of citrate were detected at all tested formate concentrations, but acetate was only detected at 100 C-mmol L −1 formate. The detection of citrate production during growing conditions and during the whole cell conversion experiment (compare Table 2 and Table 4) might indicate that the citric acid cycle was involved during formate assimilation. However, in our previous study, a canonical citric acid cycle or a reverse citric acid cycle could not be observed in the genome of D. amylolyticus [65]. The results indicate that CO 2 and citric acid were produced and consumed completely during the cultivation and only after the consumption of CO 2 and citric acid, acetic acid was produced.

D. amylolyticus Converts Formate to CO 2 , Acetate, Citrate, and Ethanol
To investigate the metabolism on formate in more detail, whole cell conversion experiments were performed. A whole cell conversion experiment has unique advantages, such as minimising side reactions and avoiding biomass production [79][80][81]. Therefore, we performed experiments with high cell densities (1·10 8 cells mL −1 ) using D. amylolyticus at formate concentrations of 20, 50, and 100 C-mmol L −1 in buffer. This experiment revealed astonishing findings. A summary of the obtained physiological variables is shown in Table 4. The highest formate uptake rate of 20.7 C-mmol L −1 was observed when D. amylolyticus was incubated at a concentration of 100 C-mmol L −1 formate. After 5 h of incubation, the production of small amounts of butyrate was detected in 50 and 100 C-mmol L −1 formate, and citrate, as well as ethanol, was only detected in one of the other. After 12 h of incubation, ethanol and low amounts of citrate were detected at all tested formate concentrations, but acetate was only detected at 100 C-mmol L −1 formate. The detection of citrate production during growing conditions and during the whole cell conversion experiment (compare Tables 2 and 4) might indicate that the citric acid cycle was involved during formate assimilation. However, in our previous study, a canonical citric acid cycle or a reverse citric acid cycle could not be observed in the genome of D. amylolyticus [65].   Gas production analyses indicated CO 2 production, but again no H 2 was detected. Interestingly, the qCO 2 values detected at any formate concentration were much lower compared to the growth experiments (compare Tables 2 and 4). However, this finding is in strong contrast to the growth experiments, as the final CO 2 concentrations during formate metabolisation were much higher (compare Figure 2 to Figure 4). These observations suggest that the metabolism of D. amylolyticus was retarded during the whole cell conversion experiment and that the released CO 2 could not be assimilated into biomass. The results of the whole cell conversion experiments revealed that the CO 2 , citrate, acetate, and ethanol play key roles in the functioning of the formate metabolism of D. amylolyticus.
Based on the obtained metabolite excretion profile, we examined the bioenergetics of formate conversion. The results are shown in Supplementary Table S4. ∆G 0 and ∆G 0 '/formate at 25 • C and 80 • C for the anaerobic production of formaldehyde + CO 2 , H 2 + CO 2 , ethanol + CO 2 , acetate + CO 2 and acetate + ethanol + CO 2 from formate. ∆G 0' /formate is given to compare the calculated values based on 1 mol of used formate. Negative values of ∆G 0 ' show that a reaction is thermodynamically favourable under the given conditions. The formation of formaldehyde + CO 2 from formate is thermodynamically not favourable. Nevertheless, this does not exclude the proposed formaldehyde pathway, as formaldehyde is not an end product. When formaldehyde is converted into other substances after formation, for example for the assimilation into biomass, the complete reaction has to be considered. Ethanol and acetate were found to be the main end products in whole cell conversion with formate as only carbon source. The results show, that these reactions are thermodynamically favourable under the given conditions. The formation of acetate and ethanol out of formate provides small amounts of energy, consistent with the slow growth of D. amylolyticus. As the results show, temperature has little influence on ∆G 0 ' for these reactions in the range of 25 to 80 • C. Gas production analyses indicated CO2 production, but again no H2 was detected. Interestingly, the qCO2 values detected at any formate concentration were much lower compared to the growth experiments (compare Table 2 and Table 4). However, this finding is in strong contrast to the growth experiments, as the final CO2 concentrations during formate metabolisation were much higher (compare Figure 2 to Figure 4). These observations suggest that the metabolism of D. amylolyticus was retarded during the whole cell conversion experiment and that the released CO2 could not be assimilated into biomass. The results of the whole cell conversion experiments revealed that the CO2, citrate, acetate, and ethanol play key roles in the functioning of the formate metabolism of D. amylolyticus.
Based on the obtained metabolite excretion profile, we examined the bioenergetics of formate conversion. The results are shown in Supplementary Table S4. ∆G 0′ and ∆G 0 '/formate at 25°C and 80°C for the anaerobic production of formaldehyde + CO2, H2 + CO2, ethanol + CO2, acetate + CO2 and acetate + ethanol + CO2 from formate. ∆G 0' /formate is given to compare the calculated values based on 1 mol of used formate. Negative values of ∆G 0 ' show that a reaction is thermodynamically favourable under the given conditions. The formation of formaldehyde + CO2 from formate is thermodynamically not favourable. Nevertheless, this does not exclude the proposed formaldehyde pathway, as formaldehyde is not an end product. When formaldehyde is converted into other substances after formation, for example for the assimilation into biomass, the complete reaction has to be considered. Ethanol and acetate were found to be the main end products in whole cell conversion with formate as only carbon source. The results show, that these reactions are thermodynamically favourable under the given conditions. The formation of acetate and ethanol out of formate provides small amounts of energy, consistent with the slow growth of D. amylolyticus. As the results show, temperature has little influence on ∆G 0 ' for these reactions in the range of 25 to 80°C.  To understand the formate metabolism of D. amylolyticus, we retrieved the amino acid sequences of characterized enzyme complexes, which are known to be involved in different formate utilization pathways: PFL, HDCR, Fdh, and FoDH. We then used the protein sequences of the formate utilization pathway-related enzyme complexes from selected microorganisms (T. onnurineus, P. furious, E. coli, M. barkeri, T. kivui, and A. woodii), to identify their homologous proteins in the genomes of D. amylolyticus. Based on these analyses, formate-related pathways were predicted in D. amylolyticus. Furthermore, we examined if the genetic arrangement of these sequences resembled the one for D. amylolyticus. The results of this analysis are shown in Figure 5. To understand the formate metabolism of D. amylolyticus, we retrieved the amino acid sequences of characterized enzyme complexes, which are known to be involved in different formate utilization pathways: PFL, HDCR, Fdh, and FoDH. We then used the protein sequences of the formate utilization pathway-related enzyme complexes from selected microorganisms (T. onnurineus, P. furious, E. coli, M. barkeri, T. kivui, and A. woodii), to identify their homologous proteins in the genomes of D. amylolyticus. Based on these analyses, formate-related pathways were predicted in D. amylolyticus. Furthermore, we examined if the genetic arrangement of these sequences resembled the one for D. amylolyticus. The results of this analysis are shown in Figure 5. The protein subunits of the HDCR complex of A. woodii are one of the microbial formate assimilation mechanisms. According to our ortho-group analysis, only the electron transfer subunits (Awo_c08200, Awo_c08230, Awo_c08250) of A. woodii belong to the same ortho-group as D. amylolyticus (Desfe_1134), which indicates that both may have the same function. On the other hand, FdhF1/2 and HydA proteins were located in different ortho-groups and were not identified in the genome of D. amylolyticus.
We then hypothesised whether D. amylolyticus possesses the genes of PFL and PFL-AE [30]. While, the Desfe_1164 sequence of D. amylolyticus showed similarities with pflA of E. coli [32], Desfe_0583 resembled TON_0415 of T. onnurineus and Awo_c27600 sequence of A. woodii, which are annotated as PFL-AE [28]. A comparison of sequences of PFL and PFL-AE proteins from A. woodii with D. amylolyticus revealed that the alignment is significant concerning E-value and identity ( Figure  5, Supplementary Table S1). The PFL (or formate C-acetyltransferase) (EC 2.3.1.54) present in E. coli and A. woodii could not be detected in D. amylolyticus. Even though the similar PFL systems were not detected in D. amylolyticus, our analysis revealed that the genome harbours a high number of radical SAM proteins (Desfe_0007, Desfe_0149, Desfe_0201, Desfe_0288, Desfe_0298, Desfe_0313, Desfe_0363, Desfe_0369, Desfe_0376, Desfe_0576, Desfe_0583, Desfe_0693, Desfe_0860, Desfe_1164, Desfe_0130, Desfe_0177, Desfe_1197, Desfe_1234) [31]. This might indicate that the PFL function is supported by another radical SAM protein, which is not similar to the PFL of E. coli or A. woodii. This finding is also not surprising considering that very few archaea possess PFL [82,83]. However, D. amylolyticus possesses PFL-AE genes (Desfe_0583, Desfe_1164, and Desfe_1234), and it was recently The protein subunits of the HDCR complex of A. woodii are one of the microbial formate assimilation mechanisms. According to our ortho-group analysis, only the electron transfer subunits (Awo_c08200, Awo_c08230, Awo_c08250) of A. woodii belong to the same ortho-group as D. amylolyticus (Desfe_1134), which indicates that both may have the same function. On the other hand, FdhF1/2 and HydA proteins were located in different ortho-groups and were not identified in the genome of D. amylolyticus.
We then hypothesised whether D. amylolyticus possesses the genes of PFL and PFL-AE [30]. While, the Desfe_1164 sequence of D. amylolyticus showed similarities with pflA of E. coli [32], Desfe_0583 resembled TON_0415 of T. onnurineus and Awo_c27600 sequence of A. woodii, which are annotated as PFL-AE [28]. A comparison of sequences of PFL and PFL-AE proteins from A. woodii with D. amylolyticus revealed that the alignment is significant concerning E-value and identity ( Figure 5, Supplementary Table S1). The PFL (or formate C-acetyltransferase) (EC 2.3.1.54) present in E. coli and A. woodii could not be detected in D. amylolyticus. Even though the similar PFL systems were not detected in D. amylolyticus, our analysis revealed that the genome harbours a high number of radical SAM proteins (Desfe_0007, Desfe_0149, Desfe_0201, Desfe_0288, Desfe_0298, Desfe_0313, Desfe_0363, Desfe_0369, Desfe_0376, Desfe_0576, Desfe_0583, Desfe_0693, Desfe_0860, Desfe_1164, Desfe_0130, Desfe_0177, Desfe_1197, Desfe_1234) [31]. This might indicate that the PFL function is supported by another radical SAM protein, which is not similar to the PFL of E. coli or A. woodii. This finding is also not surprising considering that very few archaea possess PFL [82,83]. However, D. amylolyticus possesses PFL-AE genes (Desfe_0583, Desfe_1164, and Desfe_1234), and it was recently shown that the PFL-AE homolog in T. onnurineus NA1 is strongly upregulated during growth on formate [56].
We then investigated the D. amylolyticus genome with respect to the hydrogenase gene clusters of T. onnurineus to identify possible orthologous proteins. Our sequence alignment showed that D. amylolyticus possesses one multimeric membrane bound hydrogenase subcluster (mfh) Desfe_1135-1141), and two H + /Na + antiporters (mnh) (Desfe_0344-0350 and Desfe_1085-1091) that are similar to subcluster mfh2 and mnh1-mnh2 of T. onnurineus. Regarding the fdh subcluster, which contains fdh and electron transfer genes, we were able to identify only the electron transfer gene (Desfe_1134) in D. amylolyticus. Additionally, all protein sequences of fdh subcluster belonging to molybdopterin Pfam family of proteins were downloaded for all species from Pfam, including the aforementioned strains, and compared them with the D. amylolyticus genome. Unfortunately, we couldn't detect any fdh genes in D. amylolyticus ( Figure 5, Table S1). However, the auxiliary proteins involved in hydrogenase maturation (Desfe_0501, Desfe_0337, Desfe_0339), which were found to be homologous to hydrogenase maturation proteins of T. onnurineus (Hyc I; TON_0263, Hyp F; TON_0287, Hyp E; TON_0286), were identified in the genome of D. amylolyticus. Several studies conducted with E. coli and T. onnurineus resulted in the identification of known auxiliary proteins involved in hydrogenase maturation [28,56,84]. These studies showed that the expression of the hyc operon, which contains hydrogenase maturation genes, was upregulated in formate grown cells. This could indicate that the auxiliary proteins of D. amylolyticus (Desfe_0501, Desfe_0337, Desfe_0339) might also have an important role in formate metabolism in D. amylolyticus.
Furthermore, we examined whether the necessary genes to generate ATP in T. onnurineus can be identified in the genome of D. amylolyticus. In T. onnurineus, a hydrogenase is coupled to an H + antiporter involved in the formation of a Na + gradient, which can be used for ATP generation [25,57]. Despite the fact that D. amylolyticus might possess an orthologous membrane bound hydrogenase, which is coupled to a H + antiporter in T. onnurineus, the fdh subcluster genes could not be detected in the genome. This analysis also supports the experimental observations that D. amylolyticus did not produce any H 2 from formate. However, is must be noted that D. amylolyticus produced ppm amounts of H 2 from cellulose and glucose during batch fermentation in bioreactors [64] and in previously published closed batch cultivations [63]. On the other hand, H 2 was not detectable during our recent closed batch experiments [65]. Hence, such an ATP synthesis system in D. amylolyticus remains to be detected.
The genome of D. amylolyticus encodes for several NADH generating genes, however, according to the results of this study, formate oxidation is not coupled to H 2 evolution and PFL encoding genes are missing in the genome of D. amylolyticus.

Discussion
Based on the above analysis, we hypothesise that the organism might operate the central metabolism with formaldehyde rather than formate. Therefore, we propose that two formatemetabolising reactions might occur in D. amylolyticus. First, the reduction of formate with coenzyme A (CoA) to formyl-CoA, which might be catalysed by acetyl-CoA synthetase, and furthermore, the conversion of formaldehyde with the support of an acetylating acetaldehyde dehydrogenase [58]. Second, the direct conversion of formate to formaldehyde through NADH via FoDH [61].
Our analysis showed that D. amylolyticus harbours the following proteins: Desfe_0278, Desfe_0067, and Desfe_0019-Desfe_1240 for the enzymes formyl/acetyl transferase (F/AT) [ Table S1). The generated formaldehyde can be assimilated with the oxidative PP pathway (OPPP) which is an efficient route for the assimilation of one-carbon compounds into the central carbon metabolism [85][86][87]. OPPP enzymes catalyse the oxidation of glucose-6-phosphate (G6P) to ribulose-5-phosphate (Ru5P), which was recently shown in halophilic archaea [88]. However, the genes encoding some of the OPPP enzymes, glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase are missing in the genome of D. amylolyticus. In several Archaea, it has been shown that the conventional PP pathway is incomplete [89]. Moreover, it has been confirmed through biochemical and genome analyses of Archaea that ribulose monophosphate pathway (RuMP) substitutes for the incomplete PP pathway [90]. The generated formaldehyde can be assimilated with inclusion of the synthesis of Ru5P from fructose 6-phosphate (F6P) through the reverse reaction of formaldehyde fixation by HPS/PHI via the RuMP ( Figure 6).  Table S1). The generated formaldehyde can be assimilated with the oxidative PP pathway (OPPP) which is an efficient route for the assimilation of one-carbon compounds into the central carbon metabolism [85][86][87]. OPPP enzymes catalyse the oxidation of glucose-6-phosphate (G6P) to ribulose-5-phosphate (Ru5P), which was recently shown in halophilic archaea [88]. However, the genes encoding some of the OPPP enzymes, glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase are missing in the genome of D. amylolyticus. In several Archaea, it has been shown that the conventional PP pathway is incomplete [89]. Moreover, it has been confirmed through biochemical and genome analyses of Archaea that rib Figure 6. Schematic illustration of the proposed route for formate assimilation in D. amylolyticus. The first part of the cycle is formaldehyde production from formate, which might be catalysed by formaldehyde dehydrogenase (FoDH), or formyl/acetyl transferase (F/AT) and aldehyde dehydrogenase (ADH). The second part of the cycle represents formaldehyde assimilation and ribulose 5-phosphate (Ru5P) regeneration via ribulose monophosphate pathway (RuMP) and oxidative pentose phosphate pathway (OPPP). Formaldehyde could be fixed by Ru5P to form Darabino-3-hexulose-6-phosphate (A3H6P) by 3-hexulose-6-phosphate synthase (HPS) (1) and then isomerized to fructose 6-phosphate (F6P) by 6-phospho-3-hexuloisomerase (PHI) (2). In the genome of D. amylolyticus, only gene was found for an HPS-PHI-fused bifunctional enzyme (1-2). F6P is further isomerized to glucose-6-phosphate (G6P) by glucose-6-phosphate isomerase (3). Later, G6P is oxidized to Ru5P by glucose-6-phosphate dehydrogenase (4) and 6-phosphogluconate dehydrogenase (5).
The RuMP provided metabolic precursors for the anabolism. The key enzymes are HPS (Desfe_0079), catalysing the reaction from formaldehyde to arabino-3-hexulose-6-phosphate and PHI (Desfe_0297), which catalyses the isomerization of arabino-3-hexulose-6-phosphate to fructose 6-phosphate (F6P). Further, F6P can be metabolised and generate Ru5P by the bifunctional activity of HPS/PHI (Supplementary Table S1). The required energy can be substituted by the assimilation of CO 2 together with ribulose 1,5 bis-phosphate to 3-phosphoglycerate via the activity of RuBisCO [91]. The produced 3-phosphoglycerate could be used for ATP production via glycolysis, while ATP and CO 2 production can occur via incomplete/pseudo TCA cycle [56]. However, our hypothesis would need to be validated through the combined approach of transcriptomics and proteomics-an endeavour of importance and of high dignity-considering the fastidious growth characteristics of this fascinating organism.

Conclusions
Through a combined approach of in silico analyses and physiological experiments, we could show that D. amylolyticus has the ability to metabolise formate as carbon and energy substrate. D. amylolyticus grew at similar µ on formate and glucose, which suggests that this organism faces inherent growth limitations, independent of the supplied carbon and energy substrate concentration. Supported by our experiments and analyses, we propose that the identified homologs of formaldehyde dehydrogenase genes are the only currently-known possibility allowing the metabolisation of formate. Therefore, we would like to raise the possibility that D. amylolyticus uses FoDH as a formate assimilation mechanism to produce formaldehyde, and that formaldehyde is subsequently assimilated into biomass through the RuMP. We consequently demonstrate that the CO 2 released during growth on formate is efficiently assimilated into biomass. Our findings shed new light on the metabolic versatility of the archaeal phylum Crenarchaeota and offers insight into a putative new C1 assimilation pathway in prokaryotes.