The globally widespread genus Sulfurimonas: versatile energy metabolisms and adaptations to redox clines

1The epsilonproteobacterial genus Sulfurimonas was first proposed by. Epsilonproteobacteria were originally considered as human and animal pathogens and known representatives included Campylobacter, Helicobacter, and Arcobacter species. However, in the last decade more and more non-pathogenic Epsilonproteobacteria have been isolated from different types of environments. Of these non-pathogenic Epsilonproteobacteria a great deal are exclusively associated with hydrothermal vent environments, e.g., Caminibacter, Nautilia, Hydrogenimonas, Thioreductor, Nitratiruptor, Nitratifractor, and Lebetimonas (cf., ), but other species like Sulfuricurvum kujiense have so far only been detected in terrestrial environments, e.g., an underground crude-oil storage cavity at Kuji in Iwate, Japan. In contrast, members of Sulfurospirillum, Sulfurovum, and Sulfurimonas have been found in hydrothermal vents, other marine habitats and terrestrial provinces (cf. Supplementary Table S1,; ), demonstrating their widespread occurrence. We here will focus on species of the genus Sulfurimonas, their global distribution, their versatile energy metabolisms and their adaptive abilities which may have represented one of the key features for successfully colonizing various habitats in the course of evolution.

2Among the genus Sulfurimonas five species have been isolated and described: Sulfurimonas denitrificans (DSM 1251), previously named Thiomicrospira denitrificans, and S. hongkongensis strain AST-10T (DSM 22096) were isolated from coastal marine sediments, S. autotrophica strain OK10T (DSM 16294) and S. paralvinellae strain GO25T (DSM 17229) were isolated from sediments and polychaete nests in deep-sea hydrothermal vent fields, and S. gotlandica strain GD1T (DSM 19862) was isolated from the pelagic redox cline in the central Baltic Sea. The genomes of S. denitrificans, S. autotrophica, and S. gotlandica have already been sequenced and genome sequencing of S. hongkongensis is currently in progress. Based on 16S rRNA genes, a further species is phylogenetically placed within the Sulfurimonas group, namely Thiomicrospira sp. CVO (Figure 1). When Thiomicrospira sp. CVO was isolated from an oil field in Canada, the highest 16S rRNA gene similarity to an isolated strain was to that of Thiomicrospira denitrificans (96.1% similarity) and thus it was classified into the Thiomicrospira genus. However, Thiomicrospira denitrificans had been wrongly classified into the Thiomicrospira genus (Gammaproteobacteria) and was soon after reclassified into the Sulfurimonas genus (Epsilonproteobacteria). Consequently, since Thiomicrospira sp. CVO phylogenetically groups among Sulfurimonas species (Figure 1), a reclassification of this strain into the Sulfurimonas genus needs to be discussed and we here incorporate it into our discussion on Sulfurimonas species.

3The top 100 hits of Blast searches with 16S rRNA gene sequences of the cultured Sulfurimonas species demonstrate the widespread occurrence of affiliates of this genus (Figure 1). Sequences have been recovered from various marine environments such as hydrothermal vent habitats, the pelagic redox cline of the Baltic and Black Sea and coastal sediments but also terrestrial habitats like oil caves, contaminated ground water or sediments and sulfidic springs or supra-glacial spring systems (Figure 1,;;;;;;; ). Besides being ubiquitously found across the globe, Sulfurimonas species can also make up large proportions of the bacterial community. For example, in the pelagic redox-cline of the central Baltic Sea they made up 30% of all cells and in plume waters and diffuse fluids of a deep-sea vent they comprised 70 and 20%, respectively, of the bacterial phylotype abundance. In case of the redox-cline of the Baltic Sea, the Black Sea and the plume waters, Sulfurimonas appears to be responsible for much of the chemoautotrophic activity as measured through incorporation of radioactively labeled bicarbonate demonstrating its importance for microbial element cycling in some habitats. Sulfurimonas’ success has been attributed to its ability to grow chemolithoautotrophically, its metabolic versatility (use of various electron donors, electron acceptors, and inorganic carbon sources), its oxygen tolerance and its environmental sensing systems. We discuss possible horizontal gene transfer events for distinct key genes and link the ability of Sulfurimonas species to operate numerous energy metabolisms under various chemical conditions in different types of environments.

4The lineage Sulfurimonas combines a group of sulfur-oxidizing bacteria. Many kinds of reduced sulfur compounds, such as sulfide, elemental sulfur, thiosulfate and sulfite, can serve as an electron donor for the growth of Sulfurimonas species, but not all Sulfurimonas species can use all above mentioned sulfur compounds (Table 1). Besides using sulfur compounds for energy generation, they are also essential for biosynthesis, e.g., amino acids like cysteine, and thus need to be assimilated by the cells.

5Except for S. paralvinellae, all strains can oxidize sulfide and produce sulfate as an end product and elemental sulfur and polysulfide as intermediate products. For chemolithotrophic bacteria two sulfide oxidizing pathways exist: in one pathway sulfide:quinone reductase (SQR, EC 1.8.5.4) and in the other pathway flavocytochrome c (FCC, also known as flavocytochrome c sulfide dehydrogenase) catalyzes the reaction. The released electrons are donated to the electron transport chain either at the level of quinone, if SQR is used, or at the level of cytochrome c, if FCC is used. Since no FCC encoding genes are found in any of the known Sulfurimonas’ genomes, sulfide oxidation is likely catalyzed by SQR. According to the structure-based sequence fingerprints, SQR proteins have been classified into six types, SQR Types I – VI (Figure 2,; ). However, several aspects on SQR functioning remain unresolved, including whether all Types of SQR bind quinones in the same way, and whether the different SQRs generate the same intermediates and same sulfur products during sulfide oxidation.

6Most of the genome sequenced Epsilonproteobacteria have Type II, IV, and VI and none have Type I (see Supplementary Table S1 for details). Interestingly, Sulfurimonas species appear to be the only Epsilonproteobacteria that have Type III (except for Arcobacter sp. AF1581) and Type V, aside from Type II, IV, and VI SQRs. Since Type IV SQR is so conserved among all sequenced Sulfurimonas species, it appears highly likely that it is vital for Sulfurimonas’ basic cell maintenance. One well characterized Type IV SQR is CT0117 (Accession No. WP_010931811) from Chlorobium tepidum. It is expressed constitutively but apparently primarily supports growth at sulfide levels between 2 and 4 mM. C. tepidum also contains a Type VI SQR (CT1087, Accession No. WP_010932765). Its transcript is induced at high sulfide concentrations and it is posited to be important for C. tepidum’s growth at sulfide concentrations above 4 mM. S. denitrificans is the only genome sequenced Sulfurimonas that lacks the Type VI SQR (Figure 2; Supplementary Table S1). Since the Sulfurimonas Type VI SQR phylogeny does not indicate multiple acquisitions from foreign hosts, but suggests that a common ancestor may have had a Type VI SQR, S. denitrificans appears to have lost its version in the course of time. The free sulfide concentration in tidal flats of the Dutch Wadden Sea, where S. denitrificans was originally isolated from, is much lower than 4 mM (on average <2.9 μM to maximal 21 μM; at two individual sites maximum sulfide concentrations of 587 μM and 1.8 mM were measured; ). Other studies from the German Wadden Sea have yielded similar results. Much of the sulfide appears to be buried in form of iron-sulfide (0.16–14 mM; ). Thus, given that in C. tepidum the Type VI SQR is specialized to function at considerably higher sulfide concentrations, i.e., above 4 mM, than the Type IV, we hypothesize that S. denitrificans does not need a Type VI SQR because it would unlikely encounter free sulfide in the range of what the Type VI SQR from C. tepidum appears to be specialized on. S. denitrificans may also simply deal with elevated sulfide concentrations differently to C. tepidum and thus not need a SQR VI.

7No Type I SQR was identified on any of the sequenced Sulfurimonas genomes. However, Type I and Type IV SQRs form a monophyletic clade (Figure 2), where sequences share conserved cysteines required for FAD binding and redox active coupling as well as conserved residues responsible for sulfide access. This suggests that these two types of SQR may function in the same way and thus it is not surprising that Sulfurimonas species and as such other Epsilonproteobacteria only encode one of these two types.

8The Type II SQR appears to have only low (millimolar) affinity to sulfide (cf. ). It is the only SQR that can be found in multiple copies on various epsilonproteobacterial genomes (Figure 2, Supplementary Table S1). Sulfurovum sp. NBC 37-1 even has four copies. S. hongkongensis and S. gotlandica have two copies each, which group together and which, in one case cluster together with the respective SQR from S. denitrificans (Figure 2). In contrast, S. autotrophica’s Type II SQR is phylogenetically distinct from the other Sulfurimonas. Distinct roles have been assigned to the Type II SQR: it is involved in sulfur assimilation in the non-pathogenic bacterium Pseudomonas putida KT2440, in the heavy metal tolerance in the yeast Saccharomyces pombe and in sulfide signaling in mammalian cells. Given that some Sulfurimonas species can considerably influence the mobility of Zn, Cu, and Pb through sulfide-oxidizing denitrification in contaminated marine sediments, some Type II SQRs from Sulfurimonas species may play a role in heavy metal tolerance. Interestingly, all Type II SQRs from Sulfurimonas species and non-Sulfurimonas Epsilonproteobacteria in Supplementary Table S1 have a two-arginine protein translocation (TAT) motif at the N-terminus (indicated in Figure 2). Protein folding in the periplasm may cause problems under specific environmental conditions, such as high salt concentrations. Under conditions as found at highly reduced and acidic hydrothermal vents, toxic sulfidic environments or heavy metal-contaminated fields protein folding may also be impaired in the periplasm. Hence, in such a case it is useful if the protein can be folded and matured under more controlled conditions in the cytoplasm and then transported to the periplasmic side of the membrane via the translocation pathway. The TAT motifs in Type II SQRs might be helpful for the translocation of folded and maturated Type II SQRs from the cytoplasm to the periplasmic side in Sulfurimonas and in other Epsilonproteobacteria.

9Among the genome sequenced Sulfurimonas species only S. autotrophica contains a Type V SQR. This sqr lineage is predominated by SQRs from thermoacidophilic bacteria and archaea (Figure 2). While some of these thermoacidophilic microorganisms have been isolated from deep-sea hydrothermal vents (Sulfobacillus acidophilus; ), the genome sequences of other thermoacidophilic microorganisms with an SQR Type V closely related to that of S. autotrophica (Figure 2), originate from organisms that were isolated from acid mines (; Thermoplasmatales archaeon I-plasma, Bioproject: PRJNA209804). These lineages, however, include species that have also been detected in deep-sea hydrothermal vents but for which currently no genome sequences are available. S. autotrophica was isolated from a hydrothermal vent and thus is likely the only genome sequenced Sulfurimonas species that directly comes into contact with high abundances of thermoacidophiles. This may explain why it is the only Sulfurimonas we are aware of that appears to have acquired the Type V SQR. However, if acquiring the Type V SQR is of such an advantage, it does not explain why none of the other hydrothermal vent Epsilonproteobacteria carries a Type V SQR on their genome. For example, S. sp. NBC 37-1 has a very similar metabolism to S. autotrophica (both are mesophiles that can oxidize various reduced sulfur compounds, can use oxygen as electron acceptor and fix CO2 autotrophically), but S. sp. NBC 37-1 lacks a Type V SQR. However, there is also a metabolic difference between these two species: S. sp. NBC-37 can additionally use hydrogen as an electron donor and nitrate as an electron acceptor, whereas S. autotrophica cannot. Therefore, oxidation of reduced sulfur compounds coupled with oxygen as an electron acceptor is very crucial for S. autotrophica. Compared to Type I SQR from Aquifex aeolicus, the Type V SQR has a different ‘capping loop,’ which guarantees sulfide access to the oxidation site and it is expected that sulfide reaches the active site of the enzyme through a differently structured channel. We posit that such a structure in Type V SQR may play an important role for sulfide oxidation to enhance energy generation in S. autotrophica. In the thermoacidophilic archaeon Acidianus ambivalens this type of SQR exhibits the highest activity at 70°C and pH 6.5 and only 3% activity was detected at 25°C and pH 6.5. This indicates that Type V may have a tolerance to elevated temperatures. Since S. autotrophica’s gene clusters with archaeal sqr genes, it has likely acquired its Type V sqr gene from a member of the domain archaea through horizontal gene transfer.

10S. denitrificans, S. gotlandica, and S. hongkongensis all contain a Type III SQR, which form a monophyletic group together with Arcobacter sp. AF1581 (Figure 2). Although, many Type III SQRs are associated with thermophilic archaea, the Sulfurimonas–Arcobacter clade group together with β- and γ-proteobacterial SQRs (Figure 2). Hence, it is unlikely that these SQRs are specialized to function at elevated temperatures, since several bacteria colonizing moderate temperature regimes have this SQR as well. The Type III SQRs are phylogenetically clustered with Type II SQRs, indicating that they may have similar functions. However, hardly anything is known about the bacterial Type III SQRs. Although a Type III sqr-gene (CT1025, accession No. NP_661917) has been identified in C. tepidum, no SQR activity was detected in the respective membrane proteins when both Type IV and VI sqr genes (CT0117 and CT1087) were deleted. Consequently, it can be assumed that under the provided conditions the Type III SQR is not catalyzing sulfide oxidation. It may, however, have a specific function under distinct environmental conditions. Since most Epsilonproteobacteria have one or more copies of Type II SQRs, Type III SQRs might not be crucial for Epsilonproteobacteria, which could explain why Type III SQRs are present in only very few epsilonproteobacterial species.

11Oxidation of reduced sulfur compounds can also be catalyzed by the Sox pathway, which is a thiosulfate-oxidizing multi-enzyme system and a widespread mechanism within sulfur-oxidizing bacteria. SoxXA, SoxYZ, and SoxB make up the Sox multi-enzyme complex and can oxidize the reduced sulfur compounds stepwise, finally producing sulfate excreted out of cells. In the Sox-mediated pathway SoxB encodes the key enzyme sulfate thiohydrolase. Except for Thiomicrospira sp. CVO, all Sulfurimonas species have been shown to utilize thiosulfate (Table 1). More so, all sequenced Sulfurimonas genomes encode all sox genes (soxXYZAB) required for a fully functional Sox multi-enzyme system.. According to the presence of SoxCD, the Sox pathway can be grouped into two different versions: (1) SoxCD acts as a sulfur dehydrogenase, and no sulfur globule is accumulated or (2) sulfur is formed as an intermediate because of lacking SoxCD. The soxCD genes were found on all of the sequenced Sulfurimonas genomes, indicating that these species perform the first Sox pathway with the presence of SoxCD as sulfur dehydrogenase, which oxidizes sulfane bound to the cysteine residue of SoxYZ to a sulfite anion. The key enzyme SoxB then catalyzes this sulfite anion to sulfate and releases a sulfate group from SoxYZ. The phylogenetic tree of SoxB showed that the SoxB proteins from Sulfurimonas species are closely related and form a monophyletic group (Supplementary Figure S1A), consistent with previous studies, and suggesting that it is an essential part of the Sulfurimonas metabolism. Sox enzyme systems are also found in several other Epsilonproteobacteria inhabiting different environments. They include Sulfurovum, Sulfuricurvum, Nitratiruptor, Acrobacter, and Nitratifractor (see Supplementary Table S1), although no growth of strain Nitratifractor salsuginis on thiosulfate could be shown.

12Sulfite can be oxidized by the sulfite dehydrogenase (SorAB, EC 1.8.2.1), which is a soluble SOR and contains molybdenum pterin as a cofactor. SorAB directly oxidizes sulfite to sulfate with cytochrome c as an electron acceptor. Only few of the Epsilonproteobacteria genera have the SOR enzymes encoded on their genomes: besides Sulfurimonas this includes species of the Sulfurovum and Sulfuricurvum (Supplementary Table S1), which have been isolated from sulfide-rich environments. S. denitrificans is the only genome sequenced Sulfurimonas species that does not encode for the sorAB-homologous genes, which coincides with its inability to grow on sulfite (Supplementary Figure S1B; Table 1,; ). Growth experiments showed that sulfite can indeed serve as an electron donor for the growth of S. gotlandica (all growth experiments are summarized in Table 1,; ). However, no sulfite-based growth could be reported for S. autotrophica, despite SOR activity was detected in its cell extracts. Likewise, S. paralvinellae’s cell extract also exhibited SOR activity, but this species does not appear to use sulfite as an electron donor either. This, however, can be explained: unlike other sulfite oxidases (EC 1.8.3.1), pathways depending on SorAB are unable to use oxygen as an electron acceptor and its cofactor molybdenum pterin is labile in the presence of oxygen. Since the only electron acceptor that S. autotrophica can use is oxygen, it has always been cultivated in the presence of oxygen. Under this condition, however, SorAB from S. autotrophica might not be active because of its oxygen-sensitive cofactor. This is in line with its inability to use sulfite as an electron donor (Table 1). The SOR activities that were measured for S. autotrophica and S. paralvinellae may in fact reflect the activity of SoxCD in the cell extracts. It is homologous to SOR and has been demonstrated to act as a periplasmic SOR in the Alphaproteobacterium Paracoccus pantotrophus. In summary, we conclude that S. gotlandica expresses an active SOR, S. autotrophica contains an inactive SOR, likely because of its aerobic growth, and sor genes are completely lost in S. denitrificans. Since sulfide concentrations can be very low in the Dutch Wadden Sea sediments, reactive and instable sulfite is likely very low as well. S. denitrificans may have lost these genes due to very low sulfite concentrations in the Dutch Wadden Sea habitat. Should sulfite accumulate intracellularly, the sox enzyme system could step in for detoxification of sulfite. The ability to use sulfite has not been examined in S. hongkongensis. Yet, given that both SorAs form a monophyletic cluster and appear to be similar in structure (Supplementary Figure S1B) we argue that S. hongkongensis may express an active SOR like S. gotlandica.

13Besides the above mentioned sulfur metabolisms, another two sulfur metabolizing pathways for either assimilatory sulfate reduction or dissimilatory sulfide oxidation exist that involve the intermediates adenylylsulfate (APS, also called adenosine-5′-phosphosulfate) or phosphoadenosine-5′-phosphosulfate (PAPS) and different types of APS reductase (APSR, EC 1.8.4.8 in assimilatory pathway and EC 1.8.99.2 in dissimilatory pathway) and ATP sulfurylase (ATPS, ATP: sulfate adenylyltransferase, EC 2.7.7.4) enzymes: in the assimilatory sulfate reduction pathway, sulfate is reduced to APS via the ATPS (two subunits, also called CysDN-type ATPS) and then to sulfite via the assimilatory APSR (one subunit, encoded by cysH gene). Alternatively, APS can also be phosphorylated to PAPS via the APS kinase (CysC) and then to sulfite via the PAPS reductase (also encoded by cysH gene;; ). The sulfite is then reduced to sulfide via the sulfite reductase (encoded by cysI). The homologs of the CysDN-type ATPS were found in S. denitrificans, S. gotlandica, and S. autotrophica (cf. Supplementary Figure S1C) but appear to be missing in S. hongkongensis. The homologs of the CysH-type APSR (or PAPS reductase) and of CysI were found only in S. denitrificans and S. gotlandica (cf. Supplementary Figures S1D,E). Additionally, CysC was found in S. gotlandica and S. autotrophica (cf. Supplementary Figure S1F). Hence, S. denitrificans is likely performing assimilatory sulfate reduction without PAPS as an intermediate, which was also proposed by; and S. gotlandica is probably able to perform assimilatory sulfate reduction without and with PAPS as an intermediate. In contrast, S. autotrophica and S. hongkongensis are missing genes encoding key enzymes of this pathway and are thus likely either taking up the required sulfur molecules directly from the environment or have other means to produce the necessary sulfur compounds. In the dissimilatory sulfide oxidation pathway, sulfite is oxidized through the indirect and AMP-dependent oxidation by the dissimilatory APSR (two subunits, encoded by aprBA) to APS and then to sulfate via the ATPS (one subunit, also called as Sat-type ATPS). Interestingly enough, S. denitrificans, S. gotlandica, S. autotrophica, and S. hongkongensis all have a Sat-type ATPS (cf. Supplementary Figure S1C). However, since none of these species have an AprBA-type dissimilatory APSR, it appears highly unlikely that this Sat-type ATPS is involved in dissimilatory sulfide oxidation. Also, respective AprBA activities were not detected in cell extracts from S. autotrophica and S. paralvinellae and in cell suspensions of S. denitrificans. Although S. gotlandica was described to be able to operate a dissimilatory sulfide oxidation pathway with APSR (cf. ), sequence analyses show that its APSR (Accession No. WP_008339777) contains only one subunit as it is typically found for the assimilatory sulfate reduction pathway (cf. Supplementary Figure S1D, see also above). The absence of the AprBA-type dissimilatory APSR indicates that S. denitrificans does not oxidize sulfite through this indirect pathway and thus ATPS likely only plays a role in assimilatory sulfate reduction. Moreso, despite the Sat-type ATPS being typically associated with dissimilatory sulfide oxidation, we posit that this Sat-type ATPS is also involved in assimilatory sulfate reduction as also shown in Fungi and yeast.

14Besides functioning as electron donors, sulfur compounds can also be electron acceptors. Although the solubility of elemental sulfur is quite low, polysulfide is formed by dissolving sulfur flower in an aqueous sulfide solution. One of Sulfurimonas species strain NS25-1, which was isolated from a deep-sea hydrothermal vent, can apparently grow with hydrogen as an electron donor and with elemental sulfur as an electron acceptor. However, there is no experimental evidence of sulfur reduction taking place in the cultivation. In another Epsilonproteobacteria, namely Wolinella succinogenes, polysulfide is a possible intermediate of sulfur reduction with hydrogen where electrons are transferred from a hydrogenase to the polysulfide reductase (PSR; ). A similar mechanism was also proposed in the epsilonproteobacterial Nautilia profundicola, isolated from a hydrothermal vent.

15All genome sequenced Sulfurimonas species harbor PSR-homologs on their genomes (Supplementary Table S1, Supplementary Figure S1G,;;; ) and we posit that the reduction of elemental sulfur coupled to hydrogen oxidation may also be prevalent in distinct Sulfurimonas species. Phylogenetic analysis shows that PSR catalytic subunits (PsrA) from non-vent Sulfurimonas species form a monophyletic group and are close to those from Sulfuricurvum species, which are terrestrial strains (Supplementary Figure S1G). The PsrA from the hydrothermal-vent isolate S. autotrophica is clustered with that of another vent isolate, i.e., Nitratiruptor sp SB155-2 (Supplementary Figure S1G). Hence, PSRs from the same environment but different epsilonproteobacterial genera appear to be more closely related than those from the same genera but different habitats.

16Most Sulfurimonas species are capable of growing with hydrogen as an energy source: S. denitrificans, S. paralvinellae, S. gotlandica, and S. hongkongensis (Table 1;;;;; ). S. denitrificans and S. hongkongensis can grow with hydrogen and nitrate. S. paralvinellae uses hydrogen as an electron donor in the presence of reduced sulfur compounds. Hydrogen utilization was also detected in S. gotlandica when it grew in medium with nitrate and thiosulfate. In contrast, S. autotrophica and Thiomicrospira sp. CVO have not been shown to grow with hydrogen under aerobic (6%) and anaerobic conditions, respectively (Table 1;;; ).

17Epsilonproteobacteria mostly use the [NiFe]-hydrogenase to catalyze the reaction H2 ↔ 2H+ + 2e-. The [NiFe]-hydrogenase is a heterodimer with a small subunit containing three iron–sulfur clusters and a large subunit containing the active site. They are classified in four groups, namely Groups I to IV. Except for some of the human and animal pathogens and Thiovolum, most Epsilonproteobacteria have a broad array of different [NiFe]-hydrogenases affiliated with Group I, II, and IV (Supplementary Table S1; Figure 3). Group I hydrogenases are membrane-bound hydrogenases performing respiratory hydrogen oxidation linked to the reduction of electron acceptors. They are found in all five Sulfurimonas species (Figure 3, Supplementary Table S1 and references therein), indicating that this group of hydrogenases might be essential for growth. The periplasmic Group I [NiFe]-hydrogenase of S. denitrificans contributes to hydrogen uptake activity in the membrane fractions of the cell extracts. In S. paralvinella hydrogen uptake activity was also detected. However, for S. gotlandica and S. hongkongensis these experiments have not been performed to date. In line with lacking growth on hydrogen, no hydrogen uptake activity was detected in the cell extracts of S. autotrophica. However, since its genome harbors hydrogenases, under specific environmental conditions S. autotrophica may well be able to express hydrogenases and consume hydrogen. S. hongkongensis and S. gotlandica even have two and three Group I [NiFe]-hydrogenases, respectively, which may reflect the enzymes’ abilities to function at different hydrogen concentrations, as has been suggested in the Epsilonproteobacteria N. profundicola and S. sp. NBC 37-1. The Group I Sulfurimonas hydrogenases group into different clusters with phylogenetically diverse Epsilonproteobacteria (Figure 3). This may suggest multiple acquisitions or possibly repeated loss of these hydrogenases among the Sulfurimonas species.

18Group II hydrogenases were not found in the hydrothermal deep-sea vent Sulfurimonas species, but were present in the isolates from marine water and sediments, i.e., S. denitrificans, S. hongkongensis and S. gotlandica (Figure 3). In these habitats hydrogen is less abundant than in hydrothermal vents, suggesting that this group of hydrogenases may be important under low hydrogen concentrations. Intriguingly, no hydrogen uptake activity in the soluble protein fractions of S. denitrificans was detected, even though the transcripts of the large subunit of the cytoplasmic Group II [NiFe]-hydrogenase was recovered by reverse transcription PCR. Its role thus remains speculative: in other Epsilonproteobacteria it has been proposed to function as a hydrogen-sensing hydrogenase. In contrast, posited that the cytoplasmic hydrogenase of S. denitrificans most likely functions as a catalytically active hydrogenase because of the sequence similarity of both subunits to the enzyme from A. aeolicus, where it is involved in the reverse electron transport whereby increasing efficiency of growth. Alternatively, it has been suggested that the Group II hydrogenase is important during phases of hydrogen and polysulfide starvation, where internal carbon storages are used up due to running the oxidative tricarboxylic acid cycle. Reduced ferredoxin could be reoxidized by the hydrogenase whereby the cycle could continue and hydrogen could be produced favorably. Further experiments will be necessary to confirm the actual role of the cytoplasmic hydrogenase for S. denitrificans and the other Sulfurimonas species that encode one.

19Sulfurimonas differ from several other hydrogen metabolizing Epsilonproteobacteria in that they have very few Group IV hydrogenases (Supplementary Table S1), which are hydrogen-evolving, energy-conserving, membrane-associated multimeric enzymes. S. autotrophica and S. gotlandica have one Group IV [NiFe]-hydrogenase each, Hyc and Ech, respectively (Figure 3; Supplementary Table S1). Hyc from S. autotrophica is clustered with Hycs from other Epsilonproteobacteria, which are all isolated from hydrothermal deep-sea vents (Figure 3; Supplementary Table S1). Since hydrothermal vents are extremely acidic (pH 2-4), Hyc from vent isolates might be related to anaerobic acid resistance through proton conversion as proposed in anaerobic growing cultures of Escherichia coli. Here, Hyc catalyzes hydrogen generation with protons at low external pH. Another Group IV hydrogenase, Ech, has been well characterized in the archaeon Methanosarcina barkeri. It catalyzes hydrogen formation through reduced ferredoxin and plays a role in carbon fixation from CO2 and acetate. Likewise, it may function in the same way in S. gotlandica.

20S. denitrificans was the first isolated Sulfurimonas species and was characterized as a denitrifier using thiosulfate as an electron donor and nitrate as an electron acceptor. Numerous 16S rRNA gene sequences have been identified since from coastal marine sediments and the Baltic Sea that are closely related to S. denitrificans and appear to account for much of the autotrophic denitrification. This indicates that S. denitrificans-based denitrification plays an important role in nitrogen and carbon cycling in these and likely similar environments. Except for S. autotrophica, all strains including Thiomicrospira sp. CVO are known to grow with nitrate as an electron acceptor. However, nitrate turnover in the distinct Sulfurimonas differs considerably: S. denitrificans uses nitrate up rapidly, namely 20 mM nitrate in three (with hydrogen) to six (without hydrogen) days during growth when coupled to thiosulfate-oxidation. S. gotlandica uses roughly 1 mM nitrate within 9 days in the presence of thiosulfate and S. paralvinellae utilizes ∼10 mM nitrate within 90 h in the presence of hydrogen and sulfur. S. denitrificans, S. gotlandica, and Thiomicrospira sp. CVO can additionally grow with nitrite as an electron acceptor (see Table 1 for references). All genes required for the complete reduction of nitrate to nitrogen gas, i.e., nitrate reductases, nitrite reductases, nitric oxide reductases and nitrous oxide reductases, were found on the genomes of S. denitrificans, S. hongkongensis, S. autotrophica, and S. gotlandica. However, even though the genome of S. autotrophica contains the napAGHBFLD operon, S. autotrophica is incapable of growing at a concentration of 5 mM sodium nitrate under the tested conditions.

21All so far isolated and tested Sulfurimonas species have the periplasmic nitrate reductase (Nap) catalytic subunit (NapA) (Figure 4) and all so far genome sequenced Sulfurimonas species have the napAGHBFLD operon. This organization of the nap operon is distinct from other proteobacterial nap operons, but is conserved in deep-sea hydrothermal vent Epsilonproteobacteria. Since such vent NapAs have a high affinity for nitrate, they may represent the adaption of Sulfurimonas species to the low nitrate concentrations in vent systems and other nitrate low environments.

22Interestingly, epsilonproteobacterial NapAs of isolates from hydrothermal vents clustered, while those from non-vent but marine habitats grouped and were closely related to Arcobacter species (Figure 4). Thus, the NapA proteins from the same habitats appear to be more similar to each other than to those from the same genus.

23Besides nitrate and nitrite, oxygen can also serve as an electron acceptor in most Sulfurimonas species and Thiomicrospira sp. CVO (Table 1). All four sequenced genomes of Sulfurimonas species contain a ccoNOQP operon (also called fixNOQP) encoding a cbb3-type cytochrome c oxidase (cbb3-COX) (Table 2,;;; ), which is a proton-pumping respiratory heme-copper oxidase involved in transferring electrons from cytochrome c to oxygen and reduction of oxygen to water. Because of its high affinity for oxygen, it plays an important role in supporting bacterial growth under microaerobic conditions. Phylogenetic analysis of the cbb3-COX catalytic subunit (CcoN/FixN) shows that these subunits from Sulfurimonas species form a monophyletic cluster (Clade I in Supplementary Figure S1H), indicating that this type of cbb3-COXs is important for oxygen reduction in Sulfurimonas species. These type Sulfurimonas’ cbb3-COXs are clustered within CcoNs from non-Sulfurimonas Epsilonproteobacteria, which can also use oxygen as electron acceptors (Clade I in Supplementary Figure S1H), indicating that these cbb3-COXs are conserved enzymes throughout the group. Among Epsilonproteobacteria only S. denitrificans, S. autotrophica, and S. sp. NBC 37-1 contain an additional CcoN, which is located on different parts of the genome than the ccoNOQP operon. These second Sulfurimonas’ CcoNs are clustered with those from Aquificales and Actinobacteria (Clade II in Supplementary Figure S1H). The oxygen reductase activity of this type CcoN was demonstrated in Aquificales: Sulfurihydrogenibium azorense. The growth experiments illustrated that S. denitrificans and S. autotrophica can indeed use oxygen as an electron acceptor, coinciding with the presence of a ccoNOQP operon on their genomes. In contrast, no growth of S. gotlandica with oxygen as a sole electron acceptor was detected, despite the ccoNOQP operon’s presence on its genome. Nevertheless, the authors claimed that S. gotlandica might be able to use oxygen as an electron acceptor but that the growth rate is too low for detection with their experimental design. In the presence of oxygen, S. gotlandica can still grow using nitrate as an electron acceptor. When the oxygen concentration is below 10%, S. gotlandica’s growth is not affected in the medium (with nitrate and thiosulfate), but growth is considerably impaired when the oxygen concentration exceeds 20%. This phenomenon can be explained by the induction of ccoN expression with low oxygen tension and its possible function as an oxygen scavenger to protect oxygen sensitive enzymes. Under low oxygen conditions, the expression of ccoN is induced and the activity of cbb3-COX is higher than under high oxygen conditions, diminishing the amount of oxygen and thereby protecting the oxygen-labile enzymes. Growth of S. gotlandica with nitrate and thiosulfate can then proceed. In contrast, under high oxygen concentrations the activity of cbb3-COX is lower, the activities of oxygen-sensitive enzymes for the growth with nitrate and thiosulfate are reduced, resulting in less growth. S. hongkongensis was defined as a strict anaerobe and there is no more information about its usage of oxygen. However, apparently S. hongkongensis survived after being exposed to air at 4°C for up to 8 weeks (personal communication with M. Shao and T. Zhang), which does suggest some oxygen tolerance. Nevertheless, since it has the same type of ccoNOQP operon and it lacks the same additional ccoN gene as S. gotlandica, we expect it to behave in a similar way to oxygen as S. gotlandica does.

24Growth experiments revealed that Sulfurimonas species fix carbon dioxide autotrophically via the reductive tricarboxylic acid (rTCA) cycle. This is supported through genetic and enzymatic studies. Genes encoding key enzymes, e.g., 2-oxoglutarate:ferredoxin oxidoreductase (OOR), pyruvate:ferredoxin oxidoreductase (POR), and ATP-dependent citrate lyase (ACL), for the reductive citric acid cycle are found on the genomes of S. denitrificans, S. autotrophica, S. gotlandica, and S. paralvinellae either by genome sequencing or PCR amplification (Supplementary Figure S1I; Table 2,;;; ). The activities of OORs, PORs, and ACLs were detected in the cell extracts of S. denitrificans, S. autotrophica and S. paralvinellae. It is known that the rTCA cycle genes are very abundant in the sequences from deep-sea hydrothermal vents. Phylogenetic analysis shows that AclBs from Sulfurimonas species form a monophyletic group (Supplementary Figure S1I), suggesting this pathway might be conserved in Sulfurimonas species. AclBs from hydrothermal vent isolates S. autotrophica and S. paralvinellae are clustered together and AclBs from non-vent-isolated Sulfurimonas species are clustered together, indicating that different environments influence the evolution of AclB proteins.

25Although Sulfurimonas species are characterized as chemolithoautotrophs, some organic compounds were found to contribute to their growth (Table 1). For example, S. paralvinellae can use yeast extract besides reduced sulfur species as sulfur source. Thiomicrospira sp. CVO and S. gotlandica can use acetate in addition to carbon dioxide and bicarbonate as a carbon source (Table 1,;; ). Aside from inorganic compounds, S. denitrificans can grow with formate, fumarate, yeast extract, alcohol mix as electron donors and S. gotlandica can grow with formate, acetate, yeast extract, pyruvate, and amino acid mix as electron donors. The ability of Sulfurimonas species to additionally exploit organic compounds suggests that those versatile metabolic strategies max contribute to the successful colonization of a broad type of environments.

26Of the investigated structural proteins from Sulfurimonas species the SoxBs, CcoNs (Clade I) and AclBs form monophyletic groups within the Epsilonproteobacteria (Figure 2; Supplementary Figures S1A,H,I) suggesting that these genes are vital for the cells’ survival and likely originate from a common ancestor. However, other genes of key pathways are phylogenetically positioned with non-Sulfurimonas Epsilonproteobacteria or non-Epsilonproteobacteria: e.g., Type III and Type V SQRs are clustered with SQRs from mesophilic bacteria (including a group of pathogens), thermophilic bacteria (e.g., Sulfobacillus) and thermophilic or/and acidophilic archaea, e.g., Archaeoglobus, Pyrobaculum, and some Sulfolobales and Thermoplasmatales (Figure 2); and CcoNs (Clade II) from two Sulfurimonas species cluster with CcoNs from the hydrothermal vent isolates S. sp. NB C37-1, Aquificales and Actinobacteria (Supplementary Figure S1H). The distribution of the structural protein NapAs and PsrAs from Sulfurimonas species is somehow different: NapAs from the hydrothermal vent isolated Sulfurimonas species and the ones from non-vent isolates form a clade each (Figure 4). NapA proteins from the two hydrothermal vent isolates S. paralvinellae and S. autotrophica are clustered with other deep-sea vent Epsilonproteobacteria (e.g., Sulfurovum and Nitratifractor) and NapA proteins from Sulfurimonas species isolated from marine non-vent sediments and the water column are clustered with pathogenic Epsilonproteobacteria Arcobacter species (Figure 4). Hence, NapA proteins from the hydrothermal vent organisms Sulfurovum, Nitratifractor, and Sulfurimonas are more similar to each other than NapAs from other marine Sulfurimonas species. The same pattern is also observed for PsrA proteins (Supplementary Figure S1G). In summary, these phylogenies strongly indicate the large role that horizontal gene acquisitions (from archaea, Aquificales, Actinobacteria, Gammaproteobacteria, and other Epsilonproteobacteria) appear to have had for shaping Sulfurimonas’ genomes and physiological properties in successfully surviving and flourishing in chemically distinct environments. This conclusion is consistent with the finding that besides conserved core genes, Epsilonproteobacteria generally also contain niche-specific genes, which are important to their metabolic diversity and versatility.

27Epsilonproteobacteria can colonize a broad range of habitats ranging from animal and human bodies, to terrestrial and marine environments, which include shallow waters and sediments but also deep sea hydrothermal vents (cf. Supplementary Table S1 and references therein). However, in the chemically distinct types of habitats that are inhabited by members of this lineage, the genomic inventory can differ considerably: particularly among the genes encoding SoxB (thiosulfate oxidation), SorA (sulfite oxidation), SQR (sulfide oxidation) and hydrogenases (both hydrogen uptake and evolution) (see Supplementary Table S1). Noticeably, the genomes of strains like the heterotrophic Helicobacter, Campylobacter, and Wolinella, which are limited to one type of habitat, namely animals and humans, appear relatively depleted in certain energy metabolisms. have recently suggested that members of these groups lost hydrogen sensing and evolving hydrogenases as well as the ability to fix CO2 autotrophically because they have adapted themselves to their lifestyle, in case of Helicobacter and Campylobacter a pathogenic lifestyle, and do not depend on a diverse suit of metabolisms any longer.

28Besides these heterotrophs colonizing only a comparable narrow habitat range, other epsilonproteobacterial heterotrophs exist, which have a broader global distribution: heterotrophic Arcobacter and Sulfurospirillum have been isolated from globally widespread and chemically diverging biotopes including terrestrial and marine provinces (Supplementary Table S1). While the Arcobacter and Sulfurospirillum listed in Supplementary Table S1 were isolated from humans, marshland, freshwater marsh and mud, forest ponds, their genomes still appear relatively limited with respect to key enzymes of different sulfur metabolizing pathways. This is likely related to the environments they were isolated from, either rich in organic compounds or rich in elemental sulfur. For example, some Arcobacter species were isolated from organic rich crude-oil contaminated sea water, petroleum reservoir, animals and plants. Therefore, they are adapted to using different organic compounds from their habitats as energy sources (e.g., dimethyl sulfoxide as electron acceptor) and carbon sources (e.g., organic and amino acids). It will be interesting to sequence the genome of the only so far isolated autotrophic Arcobacter and compare the genetic inventory with that from the heterotrophic Arcobacter. Sulfurospirillum deleyianum was isolated from an anaerobic enrichment culture targeting Desulfuromonas where a medium that was rich in elemental sulfur was used. Thus, in the latter case, sulfur reduction appears to be more important for growth than oxidation of reduced sulfur compounds. In contrast, these genera can harbor a relatively broad suit of hydrogenases (Supplementary Table S1), which may catalyze hydrogen oxidation coupled to the reduction of various electron acceptors, including nitrate, oxygen, elemental sulfur, and organic compounds, e.g., fumarate. Likewise, the anaerobic hydrogen oxidizers, which have so far only been isolated from hydrothermal vents, namely Caminibacter, Nautilia, and Lebetimonas, exhibit an exceptional high number of distinct hydrogenases (Supplementary Table S1). However, they appear to be relatively limited in their inventory for key enzymes of sulfur oxidizing metabolisms, which coincides with their physiology.

29The broadest suit of genes encoding for enzymes capable of oxidizing thiosulfate, sulfite and sulfide as well as hydrogen uptake and evolution can be found among autotrophic, aerobic sulfur oxidizing species, primarily among Sulfurimonas, Sulfurovum, and Sulfuricurvum. Sulfurimonas, Sulfuricurvum, and Sulfurovum have been isolated from marine sediments and hydrothermal vents, whereas Sulfurimonas and Sulfuricurvum have additionally been found in terrestrial systems (NCBI database,;;;;;;;;;; ). In these type of habitats, sulfide concentrations range from being undetectable up to several milimolar. All these three genera have Type IV and VI SQRs (Supplementary Table S1), likely to cope with different concentrations of sulfide, such as free sulfide in hydrothermal vents and metal sulfide in marine sediments or contaminated landmasses. Additionally, they have one or more copies of Type II SQRs (Supplementary Table S1), which may compensate for the function of other SQRs under environmental conditions, where folding of SQRs in the periplasm is becoming increasingly difficult and thus no functional SQR enzyme would otherwise be available. In case of Sulfurimonas, some species harbor Type III and V SQRs besides Type II, IV, and VI SQRs (Supplementary Table S1), which may either enhance the energy generation with sulfide or have different functions, such as detoxification and sulfide signaling. With respect to hydrogen metabolism, these three broadly distributed genera harbor multiple copies of Group I hydrogenases for hydrogen uptake (Supplementary Table S1), likely exhibiting different hydrogen affinities, e.g., the Gotland Deep in the central Baltic Sea, where S. gotlandica was isolated from, is known to be a highly dynamic environment. They also encode Group II hydrogenases, possibly involved in hydrogen sensing or energy conversion at low hydrogen concentrations. Additionally they harbor Group IV hydrogenases for hydrogen evolution, probably coupling with proton conversion in hydrothermal vents or with energy conversion through carbon fixation. However, unfortunately, there is no experimental evidence to provide a precise understanding of the distinct SQR and hydrogenase functions and their roles remain enigmatic in these groups. Hence, it will be essential to investigate the roles of the different types of SQRs and hydrogenases in these (and other epsilonproteobacterial) genera to better understand how they deal with distinct environmental conditions. This knowledge would also shed light on how the environmental conditions can shape species’ physiology.

30The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.