Peatland Acidobacteria with a dissimilatory sulfur metabolism

Sulfur-cycling microorganisms impact organic matter decomposition in wetlands and consequently greenhouse gas emissions from these globally relevant environments. However, their identities and physiological properties are largely unknown. By applying a functional metagenomics approach to an acidic peatland, we recovered draft genomes of seven novel Acidobacteria species with the potential for dissimilatory sulfite (dsrAB, dsrC, dsrD, dsrN, dsrT, dsrMKJOP) or sulfate respiration (sat, aprBA, qmoABC plus dsr genes). Surprisingly, the genomes also encoded DsrL, which so far was only found in sulfur-oxidizing microorganisms. Metatranscriptome analysis demonstrated expression of acidobacterial sulfur-metabolism genes in native peat soil and their upregulation in diverse anoxic microcosms. This indicated an active sulfate respiration pathway, which, however, might also operate in reverse for dissimilatory sulfur oxidation or disproportionation as proposed for the sulfur-oxidizing Desulfurivibrio alkaliphilus. Acidobacteria that only harbored genes for sulfite reduction additionally encoded enzymes that liberate sulfite from organosulfonates, which suggested organic sulfur compounds as complementary energy sources. Further metabolic potentials included polysaccharide hydrolysis and sugar utilization, aerobic respiration, several fermentative capabilities, and hydrogen oxidation. Our findings extend both, the known physiological and genetic properties of Acidobacteria and the known taxonomic diversity of microorganisms with a DsrAB-based sulfur metabolism, and highlight new fundamental niches for facultative anaerobic Acidobacteria in wetlands based on exploitation of inorganic and organic sulfur molecules for energy conservation.


Introduction
Specialized microorganisms oxidize, reduce, or disproportionate sulfur compounds of various oxidation states (-II to +VI) to generate energy for cellular activity and growth and thereby drive the global sulfur cycle. The capability for characteristic sulfur redox reactions such as dissimilatory sulfate reduction or sulfide oxidation is not confined to single taxa but distributed across different, often unrelated taxa. The true extent of the taxon-diversity within the different guilds of sulfur microorganisms is unknown [1]. However, ecological studies employing specific sulfur metabolism genes (e.g., dissimilatory adenylyl-sulfate reductase-encoding aprBA, dissimilatory sulfite reductaseencoding dsrAB, or soxB that codes for a part of the thiosulfate-oxidizing Sox enzyme machinery) as phylogenetic and functional markers have repeatedly demonstrated that only a minor fraction of the sulfur metabolism gene diversity in many environments can be linked to recognized taxa [2][3][4]. A systematic review of dsrAB diversity has revealed that the reductive bacterial-type enzyme branch of the DsrAB tree contains at least thirteen family-level lineages without any cultivated representatives. This indicates that major taxa of sulfate-/sulfite-reducing microorganisms have not yet been identified [3].
Wetlands are among those ecosystems that harbor a diverse community of microorganisms with reductive-type DsrAB, most of which cannot be identified because they are distant from taxonomically classified DsrAB sequences [5]. Sulfur-cycling microorganisms provide significant ecosystem services in natural and anthropogenic wetlands, which are major sources of the climate-warming greenhouse gas methane [6,7]. While inorganic sulfur compounds are often detected only at low concentration (lower μM range), fast sulfur cycling nevertheless ensures that oxidized sulfur compounds such as sulfate are rapidly replenished for anaerobic respiration. The activity of sulfate-reducing microorganisms (SRM) fluctuates with time and space, but at peak times can contribute considerably to the anaerobic mineralization of organic carbon in wetlands [5]. Simultaneously, SRM prevent methane production by rerouting carbon flow away from methanogenic archaea. Peat microorganisms that are affiliated to known SRM taxa, such as Desulfosporosinus, Desulfomonile, and Syntrophobacter, are typically found in low abundance [8][9][10][11][12][13][14][15]. In contrast, some microorganisms that belong to novel, environmental dsrAB lineages can be considerably more abundant in wetlands than species-level dsrAB operational taxonomic units of known taxa [13]. However, the taxonomic identity of these novel dsrAB-containing microorganisms and their role in sulfur and carbon cycling has yet to be revealed.
To identify these unknown DsrAB-encoding organisms and further investigate their fundamental ecological niches, we recovered thirteen metagenomeassembled genomes (MAGs) encoding DsrAB from a peat soil through a targeted, functional metagenomics approach. We analyzed expression of predicted physiological capabilities of the MAGs by metatranscriptome analyses of anoxic peat soil microcosms that were periodically stimulated by small additions of individual fermentation products with or without supplemented sulfate [9]. Here, we show that facultatively anaerobic members of the diverse Acidobacteria community in wetlands employ one or more types of dissimilatory sulfur metabolism.

Materials and methods
Anoxic microcosm experiments, stable isotope probing, and nucleic acids isolation DNA and RNA samples were retrieved from a previous peat soil microcosm experiment [9]. Briefly, soil from 10-20 cm depth was sampled from an acidic peatland (Schlöppnerbrunnen II, Germany) in September 2010, and stored at 4°C for one week prior to nucleic acids extractions and set-up of soil slurry incubations. Individual soil slurry microcosms were incubated anoxically (100% N 2 atmosphere) in the dark at 14°C, and regularly amended with low amounts (<200 μM) of either formate, acetate, propionate, lactate, butyrate, or without any additional carbon sources (six replicates each). In addition, half of the microcosms for each substrate were periodically supplemented with low amounts of sulfate (initial amendment of 190-387 μM with periodic additions of 79-161 μM final concentrations). DNA and RNA were extracted from the native soil and RNA was additionally extracted from the soil slurries after 8 and 36 days of incubation.
Furthermore, DNA was obtained from the heavy, ¹³Cenriched DNA fractions of a previous DNA-stable isotope probing (DNA-SIP) experiment with soil from the same site [12]. Analogous to the single-substrate incubations, anoxic soil slurries were incubated for two months with lowamounts of sulfate and a ¹³C-labelled mixture of formate, acetate, propionate, and lactate. DNA was extracted, separated on eight replicated density gradients, and DNA from a total of 16 heavy fractions (density 1.715-1.726 g mL −1 ) was pooled for sequencing.
Additional DNA was obtained from soils that were sampled from different depths in the years 2004 and 2007 [13].

Quantitative PCR and metagenome/-transcriptome sequencing
Abundances of Acidobacteria subdivision 1, 2, and 3 in soil samples from different years and depths were determined by newly-developed 16S rRNA gene-targeted real-time quantitative PCR (qPCR) assays (Supplementary Methods). Native soil DNA (two libraries), heavy 13  Binning, phylogeny, and annotation of DsrABencoding genomes The differential coverage binning approach by Albertsen et al. [16] was applied to extract MAGs of interest. The raw FASTQ paired-end reads were imported into the CLC Genomics Workbench 5.5.1 (CLC Bio) and trimmed using a minimum Phred quality score of 20 with no ambiguous nucleotides allowed. TruSeq adapters were removed and a minimum length filter of 50 nt was applied. This resulted in 214, 171, 233, 49, and 294 million reads after quality filtering and trimming for the two native soil and three SIP metagenomes, respectively (84-95% of the raw reads). All reads were co-assembled using CLCs de novo assembly algorithm (kmer size 63, bubble size 50, minimum scaffold size 1000 nt). The reads from all five metagenomes were independently mapped to the assembled scaffolds using CLCs map to reference function (minimum length 0.7, minimum similarity 0.95) to obtained the scaffold coverage. The SIP metagenomes were merged into one mapping. 137, 112, and 376 million reads could be mapped to the two native soil metagenomes and the SIP metagenome, respectively (64-66% of quality filtered reads). Gene prediction of the complete assembly was performed using prodigal [17]. In addition to the detection and taxonomic classification of 105 essential marker genes [16], dsrA and dsrB homologs were identified using TIGRFAM's hidden Markov model (HMM) profiles TIGR02064 and TIGR02066, respectively, with HMMER 3.1 [18] and the provided trusted cut-offs. Additional dsrAB-containing scaffolds were identified by using tblastn with the published DsrAB database as a query against the assembly [3]. DsrAB sequences were classified by phylogenetic analysis (Supplementary Methods; [3]). Binning and decontamination was finalized utilizing the G +C content and tetramer frequencies of the scaffolds, as well as paired-end information, as described and recommended in Albertsen et al. [16]. Completeness, contamination, and strain heterogeneity was estimated using CheckM 1.0.6 [19] with lineage-specific marker sets selected at phylum rank (or class rank for Proteobacteria). MAGs were taxonomically classified by phylogenomic analysis of concatenated marker sequences and calculation of average nucleic and amino acid identities (ANI, AAI, Supplementary Methods). MAGs were annotated using the MicroScope annotation platform [20] and eggNOG [21]. Table S2) were manually curated using the full range of tools integrated in Micro-Scope annotation platform (Supplementary Methods).

Genome-centric activity analysis: iRep and metatranscriptomics
The index of replication (iRep) was calculated for each MAG with the combined native soil metagenomes. Settings and thresholds were applied as recommended [22] using bowtie2 [23] and the iRep script with default settings. Quality-filtered metatranscriptome reads were mapped to all genomes using bowtie2 and counted with featureCounts [24]. To determine gene expression changes, we applied the DESeq2 pipeline with recommended settings [25] (Supplementary Methods).

Data availability
Metagenomic and -transcriptomic data were deposited under the BioProject accession numbers PRJNA412436 and PRJNA412438, respectively, and can also be obtained via the JGI's genome portal (JGI Proposal ID 605). MAGs are available at MicroScope (https://www.genoscope.cns.fr/a gc/microscope/) and were deposited at the European Nucleotide Archive (PRJEB24926). DsrAB sequences were deposited at NCBI GenBank under the accession numbers MG182080-MG182141.

Results
Functional metagenomics: recovery of dsrABcontaining genomes from native soil and ¹³C-DNA fraction metagenomes This study was conducted with soil samples from the Schlöppnerbrunnen II peatland in Germany, which is a long-term study site with active sulfur cycling and harbors a large diversity of unknown microorganisms with divergent dsrAB genes [5,13]. We initially generated co-assembled metagenomes from native peat soil DNA (53 Gb) and a pool of DNA extracts from the heavy fractions of a previous DNA-stable isotope probing (DNA-SIP) experiment with soil from the same peat (101 Gb). The heavy fractions, which were obtained from anoxic peat incubations with periodically supplemented sulfate and a mixture of ¹³Clabelled formate, acetate, propionate, and lactate at low concentrations, were enriched in DNA from Desulfosporosinus and also harbored DNA from yet unidentified dsrABcontaining microorganisms [12]. Based on the metagenome data, the native peat was dominated by Acidobacteria (61%), but also had Actinobacteria, Alphaproteobacteria, and Deltaproteobacteria as abundant (>5%) phyla/classes (Fig. 1). Dominance of Acidobacteria, Alpha-and Deltaproteobacteria is typical for peatlands [26]. Quantitative PCR confirmed that Acidobacteria subdivisions 1, 2, and 3 persistently dominated the Schlöppnerbrunnen II peat microbiota in oxic and anoxic soil layers (Supplementary Methods, Fig. 1), as observed in other peatlands [27][28][29].
Phylogenomic analysis showed that Acidobacteria MAGs SbA1, SbA5, and SbA7 are affiliated with subdivision 1, while SbA3, SbA4, and SbA6 are affiliated with subdivision 3 (Supplementary Figure S3). The partial MAG SbA2 lacked the marker genes used for phylogenomic treeing, but was unambiguously assigned to Acidobacteria using an extended marker gene set [16] and DsrAB phylogeny. The two near complete (96%) MAGs SbA1 and SbA5 have a size of 5.4 and 5.3 Mb, respectively. The G+C content of all acidobacterial MAGs ranges from 58 to 63% (Supplementary Table S1). This in accordance with genome characteristics of acidobacterial isolates, which have genome sizes of 4.1-10.0 Mb and G+C contents of 57-62% [31,32]. SbA1 and SbA7 form a monophyletic clade in the Acidobacteria subdivision 1 with an AAI [33] of 63% (Supplementary Figure S3) and DsrAB identity of 80% as was calculated with T-Coffee 11 [34] using the unfiltered reference alignment without the intergenic region [3]. They have 56% AAI to their closest relative, Ca. Koribacter versatilis, which is lower than AAIs among members of known acidobacterial genera (60-71%). The third MAG from subdivision 1, SbA5, is affiliated with Terracidiphilus gabretensis with an AAI of 61%. DsrAB identity of SbA5 to SbA1 and SbA7 is 79%. The three subdivision 3 MAGs form a monophyletic clade with Ca. Solibacter usitatus (Supplementary Figure S3). SbA3, SbA4, and SbA6 have AAIs of 59-73% amongst them and 61-62% to Ca. S. usitatus. DsrAB identity amongst the three MAGs is 80-94% and 74-79% to the subdivisions 1 MAGs.
The DsrAB sequences encoded on all seven MAGs belong to the uncultured DsrAB family-level lineage 8 (Supplementary Figure S1), which so far only consisted of environmental dsrAB sequences of unknown taxonomic identity [3]. Based on these MAGs and metatranscriptome analyses of anoxic peat soil microcosms, we describe the putative metabolic capabilities of these novel DsrABencoding Acidobacteria. Details on the other MAGs will be described elsewhere (Hausmann et al., unpublished; Anantharaman et al., unpublished). Functional interpretations of the recovered MAGs are made under the premise that the genomes are not closed, and thus it is unknown if genes are absent in these organisms or are missing due to incomplete sequencing, assembly, or binning.

Dissimilatory sulfur metabolism
Although Acidobacteria are abundant in diverse environments with active sulfur cycling [28,29,35,36], this is the Fig. 1 Microbial community composition in Schlöppnerbrunnen II peatland in samples from different years and soil depths. a Abundance of phyla and proteobacterial classes in the native soil (relative to all classified reads/amplicons). Taxa less abundant than 1% are grouped in grey. Coverage abundance is based on metagenomic reads mapped to classified scaffolds. Amplicon abundance is based on rrn operoncopy number-corrected abundance of 16S rRNA gene operational taxonomic units [9]. b Relative abundance of acidobacterial subdivisions (SD) in the native soil samples as determined by 16S rRNA gene qPCR assays. In addition, all subdivisions more abundant than 1% in a 16S rRNA gene amplicon dataset are shown [9] first discovery of members of this phylum with a putative dissimilatory sulfur metabolism. SbA2, SbA3, and SbA7 encode the complete canonical pathway for dissimilatory sulfate reduction, including homologs for sulfate transport (sulP and/or dass, not in SbA7) and activation (sat, ppa, hppA), adenosine 5′-phosphosulfate (APS) reduction (aprBA, qmoABC), and sulfite reduction (dsrAB, dsrC, dsrMKJOP) (Fig. 2, Supplementary Table S2a) [37]. In the AprBA tree, the acidobacterial sequences are part of a large cluster of yet uncultured organisms and Deltaproteobacteria and Firmicutes that respire sulfate, sulfite, or thiosulfate (Supplementary Figure S4) [4]. SbA1, SbA4, SbA5, and SBA6 have an incomplete sulfate reduction gene set but contain all dsr genes for sulfite reduction. Several other dsr genes were present on some of the MAGs. The dsrD and dsrN genes occurred in pairs. The acidobacterial DsrD sequences have the same conserved, hydrophobic residues as Desulfovibrio vulgaris DsrD (Supplementary Figure S5) [38]. Ubiquity of DsrD among SRM suggests an essential function in sulfate reduction, but the physiological role of this small protein is unresolved [39]. DsrN is a homolog of cobyrinate a,c-diamide synthase in cobalamin biosynthesis and may be involved in amidation of the siroheme prosthetic group of DsrAB [40]. DsrV, a homolog of precorrin-2 dehydrogenase, and DsrWa, a homolog of uroporphyrin-III C-methyltransferase, may also be involved in siroheme biosynthesis [41]. DsrT is required for sulfide oxidation in Chlorobaculum tepidum, but also found in SRM [41]. The presence of dsrMK-paralogs (dsrM2, dsrK2) upstream of dsrAB is not uncommon in SRM [42]. DsrMK are present in all dsrAB-containing microorganisms and are a transmembrane module involved in reduction of cytoplasmic DsrC-trisulfide in SRM, the final step in sulfate reduction [37]. DsrC encoded on the MAGs have the two essential cysteine residues at the C-terminal end for full functionality [43]. Interestingly, dsrC forms a gene duo with dsrL downstream of dsrAB in all seven MAGs. This is surprising, because dsrL is not found in SRM but in sulfur oxidizers. DsrL is highly expressed and essential for sulfur oxidation by the purple sulfur bacterium Allochromatium vinosum [40,44]. DsrL is a cytoplasmic iron-sulfur flavoprotein with proposed NAD(P)H: acceptor oxidoreductase activity and was copurified with DsrAB from the soluble fraction of A. vinosum [45]. The acidobacterial DsrL sequences are shorter than their homolog in A. vinosum (Supplementary Table S2a), but have the same functional domains (Supplementary Figure S6). Given the possible role of DsrL in sulfur oxidation, we sought to detect additional genes indicative of oxidative sulfur metabolism in the acidobacterial MAGs. However, genes for Sox enzyme machinery (soxABXYZ), thiosulfate dehydrogenase (tsdA), sulfide:quinone reductase (sqr), adenylyl-sulfate reductase membrane anchor subunit (aprM), flavocytochrome c sulfide dehydrogenase (fccAB), sulfur reductase (sreABC), thiosulfate reductase (phsABC), polysulfide reductase (psrABC), membrane-bound sulfite oxidizing enzyme (soeABC), cytoplasmic sulfur trafficking enzymes (tusA, dsrE2, dsrEFH), or DsrQ/DsrU (unknown functions) were absent [1,41,46,47]. SbA1, SbA3, SbA4, and SbA6 contain genes that have only low homology to soxCD/ sorAB, periplasmic sulfite-oxidizing enzymes (Supplementary Results) and, thus, might have another function [48].

DsrAB-encoding Acidobacteria are metabolically active under anoxic conditions
We calculated the iRep based on the native peat soil metagenomes to assess whether DsrAB-encoding Acidobacteria were active in situ [22]. SbA1 and SbA5, which were sufficiently complete (≥75%) for a reliable estimate, had iRep values of 1.21 and 1.19, respectively. This shows that a fraction of each population was metabolically active, i.e., on average 21% of SbA1 and 19% of SbA5 cells were actively replicating at the time of sampling. Concordantly, SbA1-7 were also transcriptionally active in the same native soil samples. 35-46% of the SbA1-7 genes were expressed in at least one replicate. SbA1 and SbA5 contributed a considerable fraction (0.4% and 1.8%, respectively, Supplementary Table S1) of the total mRNA reads in the native soil metatranscriptome. These data likely underestimate the metabolic activity of SbA1-7 in situ because freshly sampled soil was stored at 4°C for one week prior to nucleic acids extraction.
Here, we provide metagenomic and metatranscriptomic evidence that the newly discovered species represent at least three novel genera in Acidobacteria subdivision 1 and 3 (Supplementary Figure S3) and possess a dissimilatory sulfur metabolism. The seven acidobacterial MAGs from the Schlöppnerbrunnen II peatland encode the complete canonical pathway for dissimilatory sulfite or sulfate reduction. The sulfate reduction pathway, however, could also operate in reverse as proposed for a sulfur-oxidizing deltaproteobacterium [81]. The phylogenetic separation into two subdivisions as based on the concatenated marker gene tree is also apparent in the DsrAB phylogeny (Supplementary Figure S1). The acidobacterial DsrAB sequences are distributed among two monophyletic clades within the uncultured family-level lineage 8, which is part of the reductive, bacterial-type DsrAB branch [3]. The phylogenetic breadth of the acidobacterial DsrAB sequences is representative for the intra-lineage sequence divergence within uncultured DsrAB lineage 8, which suggests that this entire lineage represents yet uncultivated bacteria of the phylum Acidobacteria. Members of this uncultured DsrAB lineage are widespread in freshwater wetlands (Supplementary Figure S1) [5]. In particular, they represented an abundant fraction of the DsrAB diversity and were permanent autochthonous inhabitants of oxic and anoxic soil layers in the analyzed Schlöppnerbrunnen II peatland [12,13].
Presence of the complete gene set for canonical dissimilatory sulfate reduction suggests that the pathway is functional, as the genetic capability for sulfate reduction can be rapidly lost by adaptive evolution if unused [82]. Except for a truncated aprB on SbA6, we found no indications of pseudogenes, i.e., unexpected internal stop codons or reading frame shifts, for any of the sulfate/sulfite reduction genes on the acidobacterial MAGs [3]. In addition, sulfur metabolism genes of each MAG were expressed in the native soil and the anoxic microcosms (Supplementary Table S3a). Many sulfur metabolism genes were even significantly upregulated in the anoxic microcosms, with dsrC and aprBA among the top 10 most expressed genes in SbA7 (Supplementary Table S3a). These findings further support full functionality of the acidobacterial dissimilatory sulfur pathways under anoxic condition.
Known SRM typically couple sulfate respiration to oxidation of fermentation products such as volatile fatty acids, alcohols, or hydrogen [83]. While other microorganisms in the Schlöppnerbrunnen II soil, such as Desulfosporosinus, showed sulfate-specific and substrate-specific responses in our microcosms, hundreds of acidobacterial 16S rRNA phylotypes did not (with the exception of two) [9]. Gene expression patterns of DsrAB-encoding Acidobacteria in the individual anoxic microcosms as analyzed in the present study were ambiguous. Genes for putative oxidation of the supplemented substrates (formate, acetate, propionate, lactate, butyrate) were not specifically upregulated, neither without nor with supplemental sulfate. However, sulfur metabolism genes were upregulated in several incubations as compared to no-substrate-controls, suggesting indirect stimulation of a sulfur-based metabolism (Supplementary Results, Supplementary Table S3a). Indirect changes in microbial activity after the addition of fresh organic matter is often observed in soils (priming effects, [84]). One explanation for this priming effect is the co-metabolism theory stating that easily available substrates provide the energy for microorganisms to produce extracellular enzymes to make immobile carbon accessible, which is then also available to other microorganisms. The DsrABencoding Acidobacteria have a large genetic repertoire to utilize carbohydrates and monomeric sugars (Fig. 3). This is in accordance with the carbohydrate utilization potential previously described for subdivision 1 and 3 Acidobacteria [31,32]. Yet utilization of monomeric sugars is a rare feature of known SRM [85,86] and utilization of polysaccharides or oligosaccharides by sulfate-reducing bacteria was not yet reported. While the studied Acidobacteria expressed many of their glycoside hydrolase genes in our anoxic peat soil microcosms, further experiments are required to confirm if DsrAB-encoding Acidobacteria couple degradation of carbohydrate polymers or monomers to sulfate reduction.
It is intriguing to propose that MAGs SbA2, SbA3, and SbA7 derive from acidobacterial SRM as they lack known sulfur oxidation genes, except dsrL ( Supplementary Figure S6), and express the complete dissimilatory sulfate reduction pathway (Supplementary Table S2a), including reductive, bacterial-type dsrAB, and dsrD that may be exclusive to SRM [39,48,87,88]. However, the functions of DsrL and DsrD are yet unresolved, which prevents functional predictions based only on these genes. The proposal of an alternative hypothesis that these novel Acidobacteria reverse the sulfate reduction pathway for dissimilatory sulfur oxidation or sulfur disproportionation, bases on findings with the deltaproteobacterium Desulfurivibrio alkaliphilus [81]. D. alkaliphilus also lacks known sulfur oxidation genes (including dsrL), except for sqr, and is proposed to gain energy by coupling sulfide oxidation via a reversed sulfate reduction pathway (with a reductive-type DsrAB) to the dissimilatory reduction of nitrate/nitrite to ammonium. Sulfide oxidation in acidobacterial MAGs SbA2, SbA3, and SbA7 could proceed analogous to the pathway models proposed by Thorup et al. [81] and Christiane Dahl [89]. Briefly, hydrogen sulfide might react with DsrC either spontaneously [90] or via an unknown sulfur transfer mechanism to form persulfated DsrC.
Persulfated DsrC is then oxidized by DsrMKJOP, thereby transferring electrons into the membrane quinone pool, and releasing a DsrC-trisulfide, which is the substrate for DsrAB [37,89]. It was hypothesized that electrons released during DsrC-trisulfide oxidation to sulfite and DsrC are transferred to DsrL [89]. Further sulfite oxidation to sulfate would be catalyzed by AprBA-QmoABC and Sat.
The acidobacterial MAGs have the genomic potential to use oxygen as terminal electron acceptor and might thus couple sulfide oxidation to aerobic respiration. Alternative electron acceptors for biological sulfur oxidation in wetlands could include nitrate/nitrite and metals such as Fe(III) [50]. However, known genes for dissimilatory nitrate reduction and metal reduction [91] were absent from these acidobacterial MAGs. Only SbA2 encodes a putative metal reduction complex that was recently characterized in Desulfotomaculum reducens [92]. At this time, it is unclear whether DsrAB-encoding Acidobacteria are capable of Fe (III) respiration, as seen in Geothrix fermentans [74] and certain isolates in subdivision 1 [56,93].

Conclusion
Sulfur cycling exerts important control on organic carbon degradation and greenhouse gas production in wetlands, but knowledge about sulfur microorganisms in these globally important ecosystems is scarce [5]. Here, we show by genome-centric metagenomics and metatranscriptomics that members of the phylum Acidobacteria have a putative role in peatland sulfur cycling. The genomic repertoire of these novel Acidobacteria species encompassed recognized acidobacterial physiologies, such as a facultative anaerobic metabolism, oxygen respiration, fermentation, carbohydrate degradation, and hydrogen metabolism, but was additionally augmented with a DsrAB-based dissimilatory sulfur metabolism (Fig. 5). Based on their genetic repertoire and previous findings on reversibility of the dissimilatory sulfate reduction pathway [81,[95][96][97], it is intriguing to speculate that the described peatland Acidobacteria could use the same pathway for both sulfate reduction and sulfide oxidation. The described DsrAB-carrying Acidobacteria that only encoded the pathway for dissimilatory sulfite reduction had additional genes for sulfite-producing enzymes, which suggests that organosulfonates might be their primary substrate for sulfur respiration. Our results not only extend the current understanding of the genetic versatility and distribution of dissimilatory sulfur metabolism among recognized microbial phyla, but also underpin the challenge to unambiguously differentiate between reductive or oxidative sulfur metabolism solely based on (meta-)genome/transcriptome data [81].

Compliance with Ethical Standards
Conflict of interest The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.