Genome-guided analysis of the syntrophic acetate oxidizer C . ultunense and comparative genomics reveal different strategies for acetate oxidation and energy conservation in syntrophic acetate-oxidising bacteria

Syntrophic acetate oxidation operates close to the thermodynamic equilibrium and very little is known about the participating organisms and their metabolism. Clostridium ultunense is one of the most abundant syntrophic acetate-oxidising bacteria (SAOB) found in engineered biogas processes operating with high ammonia concentrations. It has been proven to oxidise acetate in cooperation with hydrogenotrophic methanogens. There is evidence that the WoodLjungdahl (WL) pathway plays an important role in acetate oxidation. In this study we analysed the physiological and metabolic capacities of C. ultunense on genome scale and conducted a comparative study of all known characterised SAOB, namely Syntrophaceticus schinkii, Thermacetogenium phaeum, Tepidanaerobacter acetatoxydans and Pseudothermotoga lettingae. The results clearly indicated physiological robustness beneficial for anaerobic digestion environments and revealed unexpected metabolic diversity with respect to acetate oxidation and energy conservation systems., Unlike S. schinkii and Th. phaeum, C. ultunense clearly does not employ the oxidative WL pathway for acetate oxidation, as its genome (and that of P. lettingae) lack important key genes. In both those species, a proton motive force is likely formed by chemical protons involving putative electron-bifurcating [Fe-Fe] hydrogenases rather than proton pumps. No genes encoding a respiratory Ech hydrogenase, as involved in energy conservation in Th. phaeum and S. schinkii, were identified in C. ultunense and P. lettingae. Moreover, two respiratory complexes sharing similarities to the proton-translocating ferredoxin:NAD+ oxidoreductase (Rnf) and the Na+ pumping NADH:quinone hydrogenase (NQR) were predicted. These might form a respiratory chain involved in reduction of electron acceptors other than protons. However, involvement of these complexes in acetate oxidation in C. ultunense and P. lettingae needs further study. This genome-based comparison provides a solid platform for future meta-proteomics and meta-transcriptomics studies and for metabolic engineering, control and monitoring of SAOB.


INTRODUCTION
Syntrophic acetate oxidation is essential for the biomethanisation of organic matter in engineered anaerobic digestion (AD) processes, in particular those characterised by high ammonia levels [1][2][3].Engineered biogas processes are being increasingly used in Europe and worldwide due to its potential to meet environmental and climate-related targets and to secure future energy supplies by reducing the exploitation of finite resources and recycling nutrients and energy.Using organic wastes produced by agriculture, municipalities and the food industry as substrates in the biogas process achieves sustainable and economical recycling of nutrients and energy between urban and rural areas, due the multi-functionality of the process, and reduces competition for land between food and energy production.However, use of highenergy biomass materials such as industrial food waste is still challenging, as degradation of proteinaceous materials releases high levels of ammonia, which has a direct impact on the prevailing methane production pathway, with serious consequences for process stability and efficiency [1][2][3].
Effective processing of organic matter requires the cooperative activity of microbial food chains.Fermentative microorganisms partially degrade polysaccharides, proteins, sugars and amino acids to carbon dioxide, hydrogen and short-chain fatty acids, of which acetate is by far the most abundant.Acetate is cleaved to methane and carbon dioxide by the activity of aceticlastic methanogens, a mechanism well studied and occurring in anoxic environments such as aquatic sediments and terrestrial subsurfaces [4].However, AD processes operating at high ammonia levels rely on so-called syntrophic acetate oxidation bacteria (SAOB) [5][6][7][8][9][10][11], since aceticlastic methanogens become inhibited under high ammonia conditions [2,3].The acetate-converting activity of SAOB strongly depends on the abundant presence of hydrogenotrophic methanogens, which enable the thermodynamically unfavourable acetate oxidation (reaction 1) by immediately consuming the hydrogen or formate produced (reaction 2) [12][13][14].The very small overall free energy change of ΔG 0 `-35 kJ/mol -1 (reaction 3) needs to be shared by two organisms doing the work: 1) CH 3 COO -+H + +2H 2 O → 2CO 2 +4H 2 ΔG 0 ` = +95 kJ per mol rct.
To date, only five SAOB species have been characterised [22][23][24][25][26]. Three of these, Tepidanaerobacter acetatoxydans, Clostridium ultunense and Syntrophaceticus schinkii, were recovered from mesophilic biogas processes, whereas Thermacetogenium phaeum and Pseudothermotoga lettingae were isolated from thermophilic biogas facilities.Pure culture and enzymatic studies affiliate all but P. lettingae to the physiological group of acetogens, producing acetate as the main end product by employing the reductive Wood-Ljungdahl (WL) pathway when growing heterotrophically [14,23,24,[26][27][28].Whole cell extract-based enzyme activities and mRNA expression studies indicate that the SAOB use a reversal of the WL pathway for oxidising acetate to hydrogen and carbon dioxide [14,27,29,30].The thermophilic P. lettingae does not express acetogenic metabolism [25] and must therefore use an alternative pathway.T. acetatoxydans might also use another metabolic route for acetate oxidation, as genome-scale analysis has revealed a lack of F0/F1-ATP synthase, which makes use of the oxidative WL pathway for acetate oxidation energetically unfeasible [31].In contrast, transcriptomic studies of a methane-producing syntrophic co-culture have demonstrated expression of the WL pathway genes in S. schinkii [29].Genome-based analysis, which has been thoroughly performed for T. acetatoxydans, Th. phaeum and S.
schinkii, indicates different strategies for energy conservation.The genome of the mesophilic T. acetatoxydans encodes an electron-bifurcating Rnf complex, as found in other acetogens [32,33], and a potentially electron-bifurcating [FeFe] hydrogenase, but lacks the ATP synthase operon [31].In addition to the energy-conserving complexes found in T.
acetatoxydans, the genome of S. schinkii harbours an Ech hydrogenase and several duplicates of [Fe-Fe] hydrogenase [29].The genome of the thermophilic Th. phaeum codes for a potentially bifurcating [Fe-Fe] hydrogenase and Ech hydrogenase, but lacks the Rnf complex [34].
Here we describe genome-based analysis of the mesophilic SAOB C. ultunense strain Esp and strain BS T , which have recently been sequenced [35,36].We also compare and summarise selected metabolic features and energy-conserving systems of all known SAOB (T.acetatoxydans, S. schinkii, Th. phaeum, P. lettingae and C. ultunense).Genome-guided analyses reported for Th.phaeum [34], T. acetatoxydans [31] and S. schinkii [29] provided the basis for the comparison.In the case of P. lettingae, we analysed a publically available genome draft.
The knowledge gained in this study will form the basis for future metabolic engineering and synthetic biology in order to employ SAOB as microbial cellular factories for hydrogen and methane production on demand in a controlled and monitored manner.In particular, when using complex conditions such as organic waste streams as fuel for microbial anodes, a deep understanding of the physiological attributes of potential electroactive bacteria such as the SAOB is needed, in order to achieve successful deployment under competitive industrial conditions.

Growth conditions
Since its isolation, C. ultunense strain Esp (JCM accession number JCM 16670) has been kept as liquid culture in our laboratory.For DNA isolation, cells were transferred to fresh basal medium.Stock solutions and basal medium were prepared as described by Zehnder et al. [37].
In brief, stock solutions A (15 mL), B (15 mL), I (5 mL), F (1 mL) and 0.2 g/L yeast extract were mixed with 1100 mL distilled water and cooked down to a final volume of 900 mL.The medium was cooled and aliquoted under flushing with N2/CO2 (80/20, v/v) to 180 mL portions in 500 mL bottles, sealed with butyl rubber stoppers, screw-topped and autoclaved for 20 min at 121°C.The cooled bottles were each supplemented with 10 mL of mixture C1 (containing stock solutions E (1mL), G (1 mL), C (12.5 mL) and 34.5 mL distilled water), 10 mL of mixture C2 (containing stock solutions D (49 mL), H (1 mL) and 0.5 g cysteine-HCl) and 2 ml Na-lactate (1M), by injecting the mixtures directly through a sterile filter (0.2 µm) into the bottles.Final pH of the medium was within the neutral range when checked by pH indicator paper.Bottles were inoculated 1:100 with C. ultunese strain Esp and harvested at 5000 x g after four weeks of growth at 37°C without shaking.

Isolation of DNA
DNA was isolated using the Blood&Tissue Kit from Qiagen (Hilden, Germany) following the standard protocol provided by the manufacturer.The quality was visualised by agarose gel electrophoresis and the quantity determined by fluorometric measurements using Qubit (Thermo Fisher Scientific, Waltham, MA, USA).

Genome sequencing and assembly
Clostridium ultunense Esp was sequenced at SciLife Laboratory Uppsala, Sweden.The genome was first sequenced using the Ion Torrent PM system with a de novo assembly approach, which produced 281 scaffolds with a total size of 6,159,766 bp.Sequencing and assembling details are reported in [38].In order to improve the quality of the genome, we used Illumina HiSeq-2500 mate-pair sequencing with a library size of 3 Kb in mapping assembly.
Read length of 206 bp, a longest read length of 392 bp and a total of final library reads of 2,631,078 for single end reads were obtained from Ion Torrent sequencing and 82,386,171 from the Illumina mate pair sequencing.General details about the sequencing performed can be found on the SciLifeLab website [39].The FastQC software package [40] was used for read quality assessment.After pre-assembly quality checking, the reads were assembled with MIRA 4.0 [41] and Newbler 2.8 assemblers.The filtered reads were fed into MIRA for both mapping and de novo assembly, and the same read data were also provided to the Newbler de novo assembler.Mapping assembly was undertaken against the available genome of C. ultunense BS (DSMZ 10521) (accession no.AZSU00000000) [42].Four different assemblies were produced: (i) de novo assembly of Ion Torrent reads, (ii) de novo assembly of Illumina reads, (iii) hybrid de novo assembly of Ion Torrent reads and Illumina reads and (iv) hybrid mapping assembly of Ion Torrent reads and Illumina reads.The resulting assemblies varied in size from 3.0 to 3.5 Mbp, with an average GC content of 32%.
Resulting assemblies were compared using the MAUVE genome alignment tool [43].The hybrid mapping assembly produced from Ion Torrent reads and Illumina reads was selected for further analysis.Selection was based on N50 statistics, number of contigs and the length of the largest contig.Scaffolding of the selected assembly was performed using SSPACE [44] and removal of possible gaps present in scaffolds was performed using Gapfiller [45].
Homopolymer errors were corrected manually using Consed [46].Possible miss-assemblies were corrected manually using Tablet, a graphical viewer for visualisation of assemblies and read mappings [47].Based on these analyses, the high-quality permanent genome obtained was only half the size (3,093,245 bp) of that sequenced by the Ion Torrent PM system [38].
The reason for the initially greater genome size was identified as being a low abundance bacterium that was co-cultured together with C. ultunense Esp.The genome project is deposited in the Genomes OnLine Database [48] with GOLD id Gp0161004, and the genome sequence is deposited in the European Nucleotide Archive database with accession number ERS1433212.Genome statistics are summarised in Table S1 in Supporting Information.

Genome annotation and genome comparison
Automated gene modelling was performed using MaGe [49], a bacterial genome annotation system.Genes were identified using Prodigal [50] and AMIGene [51] as part of the MaGe genome annotation pipeline.The predicted CDSs were translated and used to search the NCBI non-redundant database and UniProt, TIGRFam, Pfam, PRIAM, KEGG, COG and InterPro databases using BLASTP.Predicted coding sequences were subjected to manual analysis using the MaGe web-based platform, which also provides functional information about proteins and was used to assess and correct genes predicted through the automated pipeline.
The predicted functions were also further analysed by the MaGe annotation system.Genomic features of C. ultunense strain Esp, C. ultunense strain BS T (accession number AZSU00000000), T. acetatoxydans strain Re1 (accession number HF563609), S. schinkii strain Sp3 (accession number CDRZ00000000), Th. phaeum strain PB (accession number CP003732) and P. lettingae strain TMO (accession number CP000812.1)were compared using the MaGe web-based platform.Prophage sequences were identified using the PHAST web server [52].Tandem duplicated genomic regions of protein coding genes were calculated on the basis of identity (≥ 35% ), minLRap (≥ 0.8) and gene separation (maximum of five consecutive genes) by the MaGe platform [49].

Phylogenetic placement and synteny
The 16S rDNA sequence alignment was carried out using MUSCLE [53] and a phylogenetic tree was constructed using the maximum-likelihood (ML) algorithm [54] with MEGA 6.06 [55,56].Bootstrap analysis [57] with 100 replicates was performed to assess the support of the clusters.All predicted gene loci from C. ultunense Esp were subjected to linear comparisons with C. ultunense BS T , S. schinkii, T. acetatoxydans, Th. phaeum and P.
lettingae, using the built-in tool in the MaGe platform with synton size ≥3 genes.
Synteny-based analysis revealed that C. ultunense strains BS T (and Esp) have approximately 98% of the total genome size in synteny (Figure S1).Synteny-based analysis with all bacterial genomes present in the NCBI database reference sequence (RefSeq) revealed that Alkaliphilus metalliredigens QYMF appeared to be the closest current relative, with approximately 43% of its total genome size in synteny with C. ultunense Esp (Figure S2).In general, synteny analysis and comparison of inferred proteins showed a lower relationship of Clostridium ultunense to other SAOB than to A. metalliredigens (Table 1, Figure 2).

Genome properties
The draft genome sequence for C. ultunense strain Esp comprises one scaffold with a total size of 3,093,245 bp and a calculated GC content of 32.40%.The genome showed a protein coding density of 97.97%, with an average intergenic length of 130.04 bp.The genome contains 66 tRNA, four 5S, one 16S and one 23S rRNA genes (Table S1, Figure S3).No tRNA for selenocysteine incorporation was affiliated automatically by the MaGe pipeline.
Pseudothermotoga lettingae instead can use a much broader substrate spectrum [25] and T.
acetatoxydans also proved somewhat more versatile [24], both showing higher percentages in this COG category (Table 2).
Both genomes were found to harbour several CRISPR loci.CRISPRs (Clustered Regularly Interspaced Short Palindromic Repeats) have been found in almost half of all bacterial genomes and almost all archaeal genomes sequenced [61].The cas/CRISPR system is a prokaryotic defence mechanism, enabling the organism to respond to and eliminate invading genetic material such as bacteriophages [62].The CRISPR-associated sequence (cas) genes are often directly adjacent to the CRISPR loci.The number of CRISPR loci ranges from one to 21, but up to three loci occur most frequently [34].The genomes of strain Esp and BS T each harbour one operon for encoding cas proteins (CUESP_2343 -2350, AZSU01_30868-30875) and six to nine CRISPR loci were predicted.In T. acetatoxydans, Th. phaeum and S. schinkii, two operons each have been identified and more than eight CRISPR loci each [29,31,34].The P. lettingae genome contains one operon encoding cas proteins (Tlet_0203-0210) and two CRISPR loci were predicted.
Strain Esp and strain BS T are both lysogens, as their genomes harbour one prophage each.
A smaller number of phage remnants were found in strain Esp.In the case of P. lettingae, only phage remnants were predicted (Table S2).This is within the range observed for the other SAOB (Table S2) [29,31,34].Tepidanaerobacter acetatoxydans and Th.phaeum are lysogens.The prophage of the latter has been shown to be inducible [34].All together, the genomes of the SAOBs contain between 1.1% and 3.3% prophage-related genes (Table S2).
The genome encodes both flagellum-related and chemotaxis-related genes (CUESP_1643-1678; AZSU01_30193-30220), which is in accordance with previous observations that C.
Clostridium ultunense Esp harbours about 33 ABC transport systems, of which 11 are predicted to transport oligopeptides and amino acids, which is agreement with the observed requirement for yeast supplement for growth [22,23].Five ABC transport systems likely shuffle ribose and guanosine over the membrane, while two transport Fe and Co respectively (Table S3).Likewise, the majority of the ABC transport systems found encoded by the genomes of P. lettingae (Table S4) and T. acetatoxydans [31] are predicted to transport amino acids and oligopeptides (Table S4).In Th. phaeum and S. schinkii, ABC transport systems are mainly predicted to transport trace elements needed as cofactors [29] (Table S5) rather than organic compounds.
There are at least four phosphoenolpyruvate (PEP):carbohydrate phosphotransferase systems (PTS) encoded in the C. ultunense genome.These are predicted to take up bacterial cell wall components such as N-acetylglucosamine and N-acetylmuramic acids and oligocarbohydrates potentially originating from lichenan or related polymers (Table S6).
One potential TRAP transporter (CUESP_2539-2540) using the ion gradient over the membrane to energise unidirectional transportation of solutes was found encoded in the C. ultunense genome.TRAP transporters are only predicted for the genome of T. acetatoxydans [31] and P. lettingae (Tlet_0052-0054) and not for Th.phaeum and S. schinkii [29].Putative acetate uptake transporters are only predicted for S. schinkii and Th.phaeum (Table S6) [29,34].

Wood-Ljungdahl pathway and acetyl-CoA conversion
One of the central activities of acetogenic metabolism is mediated by phosphotransacetylase

Respiratory complexes
Two potential ion motive electron transport complexes were identified in the genome of C.
ultunense (Table 3).An operon encodes a protein complex (CUESP_0375-0379; AZSU01_10374-10378) which shows similarity to the Na+ pumping NADH:quinone hydrogenase (Na + -NQR), a respiratory complex widely distributed in marine bacteria, that couples electron transport to the generation of a sodium gradient [68].It is composed of six subunits (NQRA-F) pumping out Na+/H+ while electrons travel from NADH to quinone via a minimum of five redox carriers [68,69]: a FAD and a 2Fe-2S cluster in the peripheral NQRF subunit, an FMN in the hydrophilic NQRC subunit, and a FMN and riboflavin in the membrane integral NQRB subunit.NQRF is the site for NADH oxidation and NQRA has been identified as the quinol formation site.The complex has also been identified in the genome of the SAOB P. lettingae (Tlet_0865-0869, Table 3) and its close relative Thermotoga thermarum [70].The genes are arranged in the order NqrBCDEF and share between 36-55% identity (Figure 3).However, the genome lacks a NQRA subunit gene which contains the quinone binding site, as has also been reported for P. lettingae and T. thermarum [70].The genomes of T. acetatoxydans, Th. phaeum and S. schinkii do not encode any Nqr genes [31,34,64].As found in T. anaerobacter [31], and S. schinkii [64] the genomes of C. ultunese (AZSU01_50145-50150; CUESP_2612-2617) and P. lettingae (Tlet_0286-0291), contain a gene cluster each coding for the respiratory Rnf complex.The Rnf complex is a ferredoxin:NAD(+) oxidoreductase that uses the redox span between ferredoxin (E`= -500 mV) and NADH (E`= -320 mV) to establish an ion gradient over the membrane [71].This is equivalent to one-third to one-half of an ATP.Conversely, the generated Δμ˜Na + /H + , can drive the endergonic electron flow from NADH to ferredoxin.The genes are organised in the order rnfCDGEAB, as found in many Clostridia [71].The genome of Th. phaeum lacks Rnf genes [34] (Table 3).
The genomes of C. ultunese BS T and Esp do not harbour genes related to Ech (energyconverting hydrogenase) hydrogenase that have been found in S. schinkii and Th.phaeum [34,64].Likewise, the genomes of T. acetatoxydans [31] and P. lettingae do not encode Ech hydrogenase (Table 3).The Ech hydrogenases belong to the group of [Ni-Fe] hydrogenases, which catalyse the reversible exergonic oxidation of Fdred to the reduction of protons coupled to vectoral ion transport across the membrane [72].[29,31,34].

ATP synthase and PPases
Next to the ATP synthase complex (CUESP_1228-1235; AZSU_20626-20633), the genome of C. ultunense encodes a putative membrane-bound ion translocating pyrophosphatase (CUESP_0504; AZSU01__10499).ATP synthase and putative PPase genes are also predicted for the genomes of the other SAOB (Table 3), except for T. acetatoxydans, the genome of which does not encode an ATP synthase complex [31].
However, certain competitive abilities can be accredited to S. schinkii, as this SAOB has been found at high abundance under both low-and high-ammonia conditions [83][84][85]87].
Potential competiveness for the substrate acetate has been suggested as one explanation, as S. schinkii expresses a putative acetate uptake system [29], which is otherwise only encoded by the genome of the thermophilic Th. phaeum.
Another factor affecting the abundance of SAOB might be the proportion of proteinaceous material provided in the feedstock, as indicated by the genomic equipment of SAOB.A large proportion of the transport systems in all SAOB are predicted to shuffle oligopeptides and amino acids over the membrane, rather than carbohydrates.Furthermore, a number of genes needed for de novo synthesis of amino acids are lacking, which explains the observed dependency on yeast extract when cultivating SAOB under laboratory conditions.Thus, higher availability of proteins in the feedstock might be beneficial, as it provides both precursors for biosynthesis and energy sources.Apart from Th. phaeum, none of the SAOB harbours genes for ammonium transporters and the mesophilic SAOB also lack the high ammonium affinity glutamine synthetase (GS), both further emphasising an adaptation to amino acid-rich environments.At this time, there are no indications of habitats other than engineered AD processes when searching for 16S rRNA gene traces in the nucleotide (nr/nt) database.C. ultunense has been identified in a fermenting woad vat, another artificial environment [88].However, all SAOB are spore-forming, which enables (re)-entry and (re)-population of biogas processes depleted in SAOB and is beneficial when conditions become lethal, e.g. through exposure to oxygen.Moreover, none of the known SAOBs appears to be a strictly obligate anaerobe, as indicated by genes conferring protection against super oxide radicals.This attribute facilitates transfer between cultivation bottles and bio-augmentation of laboratory-scale reactors, as well as phenotypic characterisation and metabolic engineering.
One more interesting attribute of the SAOB is their apparent robustness against phages.
In nutrient-rich and cell-dense environments such as AD processes, viruses can be expected to be high in numbers and diversity [89,90].In WWTP, virus-like particles have been found at frequencies as high as 10 7 -10 9 per millilitre [90].Despite the high virus abundance in AD environments, the genomes of C. ultunense and other SAOB are only very weakly infected.
This indicates a certain robustness against phage attacks, as bacterial genomes can contain up to 20% prophage-related genes [91,92].The significantly higher number of CRISPR loci acquired by the SAOB points to strong phage-microbial interactions.The adaptive nature of CRISPR triggered by invading phages confers robustness and the ability to encounter new phages.Thus, the strong impact of bacteriophages on microbial diversity and community composition in AD processes, as recently shown by Zhang et al. [89], might be less pronounced in SAOB populations.Moreover, SAOB are considered to belong to the smaller and more slow-growing bacterial populations of the biogas microbiota [28] and might therefore be less intensively hunted by bacteriophages.There are indications that phages selectively attack the larger populations within a microbial community [93,94].Moreover, the CRISPR/cas loci identified in SAOB might potentially enable strain-typing applications for controlling naturally and metabolically engineered SAOB strains.
Acetate oxidation in mesophilic and thermophilic SAOB Cultivation experiments, enzyme activity studies and mRNA-based investigations previously led to the assumption that C. ultunense belongs to the physiological group of acetogens, indicating that the WL pathway plays a central metabolic role, as has also been reported for the SAOB Th. phaeum, S. schinkii and T. acetatoxydans [14, 22-24, 30, 31, 34].
Consequently, in AD processes these SAOB were thought to be generally able to perform two metabolic functions: acetate production from smaller organic molecules employing the reductive direction of the WL pathway and syntrophic acetate oxidation using the oxidative direction.However, our study revealed that at least two pathways for mesophilic acetate oxidation must exist, as the genome of C. ultunense lacks key enzymes of the WL pathway (Figure 4, 5).In the case of the thermophilic P. lettingae, the lack of WL pathway genes was expected, as the metabolism has been described as non-acetogenic [25] (Figure 4, 5).Both Th. phaeum and S. schinkii harbour the complete set of WL pathway genes (Figure 4, 5) and transcriptomic profiling of an acetate-oxidising co-culture has clearly confirmed involvement of the oxidative WL pathway in acetate oxidation in S. schinkii [29,34].For Th. phaeum, no such expression data are available, but enzyme activity studies have confirmed expression of key functions of the WL pathway, indicating that this pathway is a potential acetate oxidation route [27].Moreover, Th. phaeum and S. schinkii are closely related (92.1% 16s RNA gene identity), with 50% of the genomes in synteny.Gene organisation of the WL pathway genes in Th. phaeum is identical to that in S. schinkii (Figure 5).Genome-guided analysis of T.
acetatoxydans has revealed a truncated WL pathway [31] (Figure 4).However, a role of the WL pathway in acetate oxidation has been questioned, as net ATP synthesis is strongly impaired due to the lack of F0/F1 ATP synthase [31] and substrate level phosphorylation via the oxidative WL pathway does not gain any net ATP (Figure 4).There has been speculation about acetate oxidation in T. acetatoxydans, including a metabolic route via an oxidative tricarboxylic acid cycle [31].However, no evidence on gene expression or activity level has been supplied so far.In the case of C. ultunense and P. lettingae, this proposed metabolic route cannot be applied, as both genomes lack key enzymes such as fumarate reductase and succinyl-CoA:acetate CoA transferase.However, the free energy released (ΔG 0 `-35 kJ/mol -1 ) in SAO indicates that there is no such reaction which can be coupled to ATP synthesis (ΔG 0 ` 60 to 70 kJ/mol -1 ) directly, no matter which metabolic reactions contribute to SAO.All SAOB are potentially able to activate acetate to acetyl phosphate via acetate kinase (Figure 4).In the case of S. schinkii and Th.phaeum, phosphotransacetylase activity enables the acetyl moiety to enter the WL pathway (Figure 4), as strongly supported by transcriptomic data obtained for S. schinkii [29].For C. ultunense and P. lettingae, the metabolic fade of the acetyl phosphate still needs to be explained.However, mRNA expression and enzyme activity studies clearly indicate involvement of formyltetrahydrofolate synthetase in the acetate oxidation pathway of S. schinkii, T. phaeum, C. ultunense and T. acetatoxydans [27,29,30].For P. lettingae, no such expression or activity data have been published so far.

Energy conservation associated with SAO
The SAOB differ markedly in their energy conserving systems, indicating different strategies.
No core set could be identified, except for [Fe-Fe] hydrogenases, which are known to be involved in H2 evolution and which are encoded by all SAOB genomes (Figure 6).As observed for many anaerobic bacteria, ferredoxin and rubredoxin seem to be the main electron carriers in energy metabolism.Ferredoxin contains iron-sulphur clusters as an electron-mediating cofactor.Flavodoxins contain a tightly bound flavin mononucleotide (FMN) as the electron-receiving component.The redox potential is strongly affected by the apoprotein [95].Rubredoxin contains redox-active iron, but lacks inorganic sulphide.In the case of C. ultunense, Th. phaeum and P. lettingae, flavodoxins can potentially substitute for ferredoxin, e.g.under iron shortage.Both ferredoxin and flavodoxin can serve as electron donors in reactions with standard redox potential as low as -500 mV and can therefore serve as either electron donor or electron acceptor in hydrogen production or consumption [96].The respiratory Ech hydrogenase and Rnf complex can both couple ferredoxin oxidation to proton translocation.Ech hydrogenase transfers electrons from ferredoxin to protons and Rnf complex mediates electron transfer from ferredoxin to NAD + , covering the redox span from E 0 ′−500 to -413 mV and E 0 ′−500 to −320 mV, respectively [71,97].The proton gradient established by either complex can be used for ATP production and for providing energy for endergonic reactions, such as oxidation of methyl THF to methylene THF by methylene THF reductase, as needed in the oxidative WL pathway.The Rnf complex is the most widespread electron transport complex in anaerobic microorganisms [71].However, it appears unlikely that the complex supports energy conservation in the oxidative WL pathway, as the standard potential of the NADH produced might not sufficiently negative to reduce protons.This is in agreement with the observed downregulation of the Rnf genes in S. schinkii when oxidising acetate and the lack of Rnf genes in the genome of Th. phaeum [29,34] (Figure 6).
Instead, energy conservation by the oxidative WL pathway most likely involves Ech hydrogenase, which can couple oxidation of ferredoxin directly to formation of a proton gradient (Figure 6).However, the oxidative WL pathway produces NADH as redox equivalent from three enzymatic reactions (reaction 6, 7, 10; Figure 5) and only one reaction reduces ferredoxin (reaction 4, Figure 5) [98].The genomes of S. schinkii and Th.phaeum are predicted to encode potential electron-bifurcating ferredoxin-and NAD+-dependent [Fe-Fe] hydrogenases [29,34], which might couple the endergonic formation of hydrogen from NADH to the exergonic formation of hydrogen from reduced ferredoxin (Figure 6).In that case, a proton motive force would be generated due to the cytosolic consumption of protons.
Transcriptomic studies performed on S. schinkii support involvement of Ech hydrogenase and bifurcating [Fe-Fe] hydrogenase in acetate oxidation [29].For Th. phaeum, a potential link between formate oxidation and bifurcating hydrogen formation has also been suggested than by respiratory complexes, as no such complex has been predicted expect for a putative Rnf complex (Figure 6).The same may apply for C. ultunense and P. lettingae: The proton motive force is very likely formed by chemical or scalar protons rather than vectoral protons, as next to Rnf complex both genomes harbour only a putative bifurcating [Fe-Fe] hydrogenase (Figure 6).
However, both genomes encode a putative NQR complex, which indicates that another electron acceptor rather than protons can potentially be utilised.This respiratory complex transfers electrons from NADH to quinone covering a redox span between -320 and +90 mV.The net redox reaction is similar to that carried out by H+ pumping NADH-quinone oxidoreductase complex found in mitochondria and many respiring bacteria, but the complexes are not homologous [68].
In C. vibrio, the NQR complex generates first an electrochemical Na gradient, followed by formation of a proton motive force by a respiratory complex connected via quinol.
Interestingly, in C. ultunense and in P. lettingae the genome lacks the quinone-binding subunit NqrA, indicating that the final electron acceptor could potentially have a more negative redox potential than quinone.The last redox carrier is very likely FMN, the redox potential of which strongly depends on the apoprotein [95].The NQR complex might follow downstream of the RNF complex, both together forming a potential respiratory chain (Figure 6).However, involvement of the NQR complex in acetate oxidation and the final inorganic or organic electron acceptor(s) still need to be uncovered.Interestingly, A.
metalliredigens, the current closest sequenced relative of C. ultunense, also harbours a putative Nqr complex (Figure 3). A. metalliredigens has been described as a strictly anaerobic metal-reducing bacterium, only distantly related to other commonly studied ironreducing microorganisms [99].It can utilise lactate or acetate (in the presence of yeast extract) when any of the electron acceptors Fe(III)-citrate, Fe(III)-EDTA, Co(III)-EDTA or Cr(VI) is present [99].For C. ultunense, acetate oxidation using Fe(III) as an electron acceptor has been excluded [22,23].In P. lettingae, acetate oxidation has been observed in the presence of thiosulphate [25].In order to provide reduced ferredoxin for anabolic and catabolic pathways, both Rnf complex and Ech hydrogenases can reverse the electron flow using the electrochemical gradient and drive the thermodynamically unfavourable reduction of ferredoxin with NADH or hydrogen.In T. acetatoxydans, C. ultunense and P.
lettingae, the Rnf complex might connect the NADH pool fed by reactions of acetate oxidation to hydrogen evolution by hydrogenases, which requires ferredoxin.ETF complex seems not play a role in energy conservation connected to SAO.This complex is mainly involved in oxidation of specific substrate by means of dehydrogenase reactions [100].In the case of C. ultunense and P. lettingae, the activity is likely associated to betaoxidation of fatty acids, as it is encoded together with an acyl-dehydrogenase subunit.The ETF found encoded in the genome of T. acetatoxydans likely receives electrons from oxidation of lactate, as it is part of an operon encoding a lactate uptake permease and a lactate dehydrogenase subunit.
Moreover, additional energy might be conserved by the ion-translocating pyrophosphatase found in all SAOB (Figure 6).These PPases are integral membrane proteins that couple the hydrolysis of pyrophosphate (PPi) to the transport of monovalent cations (Na + /H + ) against the electrochemical potential gradient, generating an ion motive force [101].Thus, PPi formed as a by-product of biosynthetic reactions in which nucleoside triphosphates are converted to nucleoside monophosphates could be hydrolysed by the membrane-bound PPase and contribute to the formation of a proton motive force and subsequently to the synthesis of ATP.

Conclusions
Likewise, comparison of all inferred proteins of C. ultunense Esp with all proteins collected in the NCBI RefSeq database revealed the highest number of orthologues with C. ultunense BS T (3,106: 91.87%) and the next highest with A. metalliredigens QYMF (1,920: 56.79%).

Figure 2 .
Figure 2. Synteny comparison of Clostridium ultunense strain Esp and the SAOB Syntrophaceticus schinkii, Tepidanaerobacter acetatoxydans, Thermacetogenium phaeum, and Pseudothermotoga lettingae.The lines indicate syntons between two genomes.Red lines show inversions around the origin of replication.Vertical bars on the border line indicate different elements in genomes, where pink = transposases or insertion sequences, blue = rRNA and green = tRNA.
Very little is known about the metabolic capacity and flexibility of SAOB, mainly due to strict cultivation requirements and difficulties in reconstituting the thermodynamically unfavourable acetate oxidation process under laboratory conditions.Therefore, genomebased studies and comparisons are important, as they provide insights into questions regarding metabolic and physiological potential and restrictions and provide the basis for metabolic engineering, control and monitoring.The results of the present study clearly indicate unexpected metabolic diversity between the known SAOB with regard to acetate oxidation, energy conservation and metabolic flexibility, and a certain environmental robustness beneficial for AD processes.The metabolic diversity also implies differences in regulation of the SAO pathways with respect to potential inducers, repressors and regulators, which has potential consequences regarding the abundance and activity of SAOB in anaerobic digestion processes.was supported by EU-COST action BM1006-SeqAhead.EB-R was also partly supported by EU FP7 ALLBIO project, grant number 289452.AS and BM were supported by the Swedish Energy Agency (project no P36651-1).BM was further supported by FORMAS, grant number 942-2015-1008.The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.SM, BM, EB and AS contributed to the conception and design of this project.SM was involved in acquisition and bioinformatics analysis of the data.BM und SM analysed and interpreted the data.SM prepared the figures and tables.BM wrote the manuscript.AS and EB provided financial support.All authors were involved in critical revision of the manuscript, have given final approval of the version to be published and agree to be accountable for all aspects of the work.
This is within the range observed for the other SAOB except for carbohydrate metabolism (Table 2).Cultivation experiments indicated that C. ultunense, S. schinkii and Th.phaeum are very confined in heterotrophic substrate utilisation, in the case of C. ultunense Esp encompassing only ethanol, betaine, lactate, cysteine and raffinose AckA) to ADP, producing ATP and the end product acetate.The C. ultunense genome harbours an ackA gene (CUESP_1616, AZSU01_30167), but lacks a pta gene.Thus, substrate-level phosphorylation through Pta/AckA is strongly impaired.Moreover, only genes encoding proteins of the methyl branch of the WL pathway are present.These include formyl [66,67]which replaces the CoA moiety from acetyl-CoA by an inorganic phosphate molecule to form acetylphosphate.The phosphate moiety is further transferred by the activity of acetate kinase (AZSU01_30669-72).According to synteny maps obtained with genomes of acetogens, neither of those resemble the CODH subunit of the acetyl-CoA/CODH complex as part of the WL pathway.Moreover, the essential CODH-Ni-insertion protein (CooC) is lacking.Genes encoding acetyl CoA synthase subunit, methyltransferase, methylene reductase and the corrinoid enzyme are absent.Despite this, acetogenic metabolism and acetyl-CoA synthase activities have been reported for C. ultunense strain BS T[14,22].To achieve this, Pta/AckA activity could potentially be replaced by a predicted ADP-forming Acetyl-CoA synthetase (CUESP_2312, AZSU01_30838).Alternatively, site activities of other phosphate acyltransferases (e.g.CUESP_0095; AZSU_10093; CUESP_1619, AZSU01_30169; CUESP_1223, AZSU01_20621) might substitute for the function of PTA.Enzyme activity measurements performed previously[14]are not specific for PTA, because the release of free CoASH from acetyl-CoA is determined[66,67].Thus, free CoASH can be produced by any acyl-CoA active enzyme, which might explain the observed activity of the cell-free extract.