Unexpected carbon utilization activity of sulfate-reducing microorganisms in temperate and permanently cold marine sediments

Abstract Significant amounts of organic carbon in marine sediments are degraded, coupled with sulfate reduction. However, the actual carbon and energy sources used in situ have not been assigned to each group of diverse sulfate-reducing microorganisms (SRM) owing to the microbial and environmental complexity in sediments. Here, we probed microbial activity in temperate and permanently cold marine sediments by using potential SRM substrates, organic fermentation products at very low concentrations (15–30 μM), with RNA-based stable isotope probing. Unexpectedly, SRM were involved only to a minor degree in organic fermentation product mineralization, whereas metal-reducing microbes were dominant. Contrastingly, distinct SRM strongly assimilated 13C-DIC (dissolved inorganic carbon) with H2 as the electron donor. Our study suggests that canonical SRM prefer autotrophic lifestyle, with hydrogen as the electron donor, while metal-reducing microorganisms are involved in heterotrophic organic matter turnover, and thus regulate carbon fluxes in an unexpected way in marine sediments.


Introduction
Marine sediments are the largest organic matter sink on Earth [1].Mineralization of buried organic matter is driven by the anaerobic microbial food chain, a network of fermenting and anaerobically respiring microorganisms perched below the seaf loor [2,3], which orchestrates the fate of organic compounds as well as the biogeochemical cycling of elements such as carbon, sulfur, nitrogen, iron and manganese [4,5].
In the anoxic, sulfate-laden layers below the surface of the sediment, i.e. the sulfate reduction zone (SRZ), sulfate-reducing microorganisms (SRM) are one of the most important players that mediate a large fraction of organic matter degradation [6,7].Accordingly, SRM are genetically equipped to utilize divergent organic compounds, such as short-chain fatty acids, alcohols, carbohydrates, organohalogens, and aromatics [2,6,8].Among those compounds, organic fermentation products are believed to be the most crucial substrates, the degradation of which is coupled to sulfate reduction as the terminal electron-accepting process in sediments [6,9].Fermentation products originating from organic matter degradation such as acetate, lactate, propionate, butyrate, and ethanol are typically present in micromolar concentrations [9][10][11][12][13][14].However, organic fermentation products are not under thermodynamic control for potential degraders such as SRM and metal-reducing microorganisms [14][15][16], and thus it is not clear which microbes are active for low concentration of fermentation products.It is assumed that SRM is one of the most important microbial groups responsible for the degradation of organic fermentation products.For example, multiple SRM, such as representatives of the Desulfobacteraceae, Desulfocapsaceae, Desulfosarcinaceae, and Desulfovibrionaceae, can degrade short-chain fatty acids and alcohols [17][18][19][20].Iron-reducing microorganisms Desulfuromondales also can use these organic substrates [21].Sulfate reduction and iron reduction can co-occur in the SRZ of marine sediments [9,[22][23][24][25]; however, how fermentation products can be degraded by these microorganisms is not well studied in situ.
SRM have been enriched [15,[26][27][28][29][30][31] and isolated in pure culture [17,18] from many different sediments, but using concentrations of fermentation products in the mM range, which is much higher than typically encountered in situ [9][10][11][12][13][14]. Apparently, enrichments and pure cultures amended with a high level of fermentation products challenge whether these microorganisms are relevant in situ.Detecting the activity modes of SRM in sediments rather than in enrichments or cultures is key to understand their physiological features in the environment as well as their role in elemental cycling.To date, however, much of the work on active SRM has focused on enrichments with known limitations regarding its environmental relevance.Only few studies attempting to capture in situ conditions suggest that the well-known SRM of the family Desulfobacteraceae can use both, acetate or H 2 [32][33][34].On the other hand, significant discrepancies between sulfate reduction rate and acetate oxidation rate have been reported.Such imbalanced ratios of sulfate reduction and acetate oxidation rates between 1:4 and 20:1 [16,[35][36][37] suggest that either SRM might use other fermentation products, or other microorganisms present in sediments participate in fermentation product degradation [16,18,21,25,32,[38][39][40]; however, the active microbes are not well linked to specific processes when rates are measured.
Given the apparent knowledge gap in understanding the role of SRM in fermentation product degradation in anoxic marine sediments, the following questions require investigation: (i) which fermentation products are utilized by SRM in the SRZ of marine sediment and (ii) what is the impact on biogeochemical cycling therein.We hypothesize that SRM and other microorganisms occupy different but complementary niches for fermentation product utilization.Thus, in the SRZ, the oxidation of organic (e.g.acetate, propionate, butyrate, lactate, ethanol) and inorganic fermentation products (e.g.H 2 ) is driven by different guilds of microorganisms and have different trophic categories among these microbes, respectively.However, the low in situ concentrations of fermentation products in sediments are difficult to mimic in incubations because such low concentrations challenge the sensitivity of detecting and identifying active SRM without enrichment.To test the hypothesis and overcome the technical limitation, we used the highly sensitive RNA-based stable isotope probing (RNA-SIP) [41] approach with low, close to in situ concentrations of multiple organic fermentation products (max.30 μM) in sediment incubations.RNA-SIP is an ultrasensitive technique with a threshold below 0.001% of fully 13 Clabeled nucleic acids [41,42].In combination with metagenomic analysis, our findings reveal novel features of fermentation product degradation in the SRZ regarding to carbon cycling in marine sediments.

Sampling and incubation setup for stable isotope probing
Sediments used in this study were sampled from Helgoland mud area (North Sea; 54  C on board, then the cores were cut in 25 cm sections and stored at 4 • C in 2.6-l jars with anoxic artificial sea water and headspace of N 2 .The information of sediment sampling and geochemical profiles were described in the previous studies [43][44][45].Slurry incubations were set up with sulfate-rich sediments from the top layers of Helgoland mud area (16-41 cm), Cumberland Bay (14-39 cm), and Hornsund fjord (0-15 cm).Sediments were homogenized with artificial water (w: v = 1: 4, 50 ml; 26.4 g l −1 NaCl, 11.2 g l −1 MgCl 2 •6H 2 O, 1.5 g l −1 CaCl 2 •2H 2 O, and 0.7 g l −1 KCl) and filled in sterile 120-ml serum bottles, which were sealed with butyl rubber stoppers.The slurry was vacuumed three times for 3 min in order to remove O 2 introduced during incubation setup, and headspace of culture was f lushed with N 2 at 1.5 atm as described previously [46].For improved isotope labelling, slurries were preincubated for 10 days at 10 • C to deplete organic substrates, O 2 and nitrate remaining in the original slurry [15].Thus, O 2 was not introduced into incubations in order to mimic the anoxic condition of the sediment used for incubations [44,47,48].After preincubation, all the slurries were amended with low concentrations (60 μM carbon) of fully 13 C-labeled (99%) organic fermentation products (acetate: 30 μM; propionate: 20 μM; lactate: 20 μM; butyrate: 15 μM and ethanol: 30 μM; provided by Cambridge Isotope Laboratories, Tewksbury, MA).For the inorganic fermentation products, i.e.H 2 /CO 2 , ∼100 μM H 2 in slurry was transferred (15% of H 2 in the headspace gas given its very low solubility [49]), and 10 mM 13 C dissolved inorganic carbon was added.All incubations were amended with 18 mM sulfate.Since without amendment of substrate during preincubation will not trigger strong bacterial community shift [50], microbial activity will be identified when 13 C-labeled substrates were utilized.In order to prove that SRM were present and had ability to degrade organic fermentation products, five antibiotics with the concentration of 50 mg l −1 (streptomycin, ampicillin, kanamycin, vancomycin and D-Cycloserine) for each were amended to potentially inhibit the activity of metal-reducing bacteria in incubations using one temperate (Helgoland mud) and one permanently cold (Cumberland Bay) sediments.SRMs such as Desulfovibrionaceae and Desulfobacteraceae are able to resist antibiotics [51][52][53][54][55][56].A parallel set of controls containing unlabeled substrates was also conducted.All slurries were incubated at 10 • C in order to have a better comparison among different sediments.After 6-15 days, incubations were stopped based on the development of δ 13 C values of CO 2 in headspace, which was measured as described previously [57] (see Fig. 1 for the details of the incubation time).For incubations with inorganic fermentation product, i.e.H 2 /CO 2 , samples were harvested after 17 days.To identify the ability of glucose degradation by SRM, 10 μM 13 C-glucose (i.e. 60 μM carbon) was amended into the SIP incubations, a same setup with the study for organic fermentation products degradation.In order to reveal metal reduction and avoid effect of sulfate, deep sediment from methanic zone (Helgoland mud area: 95-120 cm [58]) was used for SIP incubation setup using organic fermentation products.

Isopycnic centrifugation, gradient fractionation
For RNA-SIP analysis, RNA was extracted from slurries as described previously [59,60].Isopycnic centrifugation and gradient fractionation were employed to separate 13 C-labeled from unlabeled RNA.Brief ly, in order to obtain enough RNA for SIP, we combined RNA from biological replicates (n = 3).About 500-1000 ng of RNA was loaded with formamide (240 μl), cesium trif luoroacetate solution (6 ml, CsTFA, GE Healthcare, Buckinghamshire, UK), and gradient buffer solution.RNA was density separated by an Optima L-90 XP ultracentrifuge (Beckman Coulter, Brea, CA).At the same time, a mixture of fully 13 C-labeled and unlabeled RNA from Escherichia coli was used as standard during density separation for defining heavy and light gradient Biosystems, Foster City, CA).SIP fractions including 13 C-labeled RNA were defined by standardization with RNA of fully labeled and unlabeled RNA standards from E. coli.

16S rRNA gene sequencing
Polymerase chain reaction (PCR) was performed with barcoded bacterial primer pair (Bac515F: 5 -GTGYCAGCMGCCGCGGTAA-3 ; Bac805R: 5 -GACTACHVGGGTATCTAATCC-3 ) [61] using KAPA HiFi HotStart PCR kit (KAPA Biosystems, Cape Town, South Africa).Thermocycling was set as follows: 95 • C for 3 min; 35 cycles at 98 • C for 20 s, 61 • C for 15 s, and 72 • C for 15 s; 72 • C for 1 min.PCR products were then purified and quantified for library preparation [62].Amplicons were sequenced through NovaSeq 6000 platform (Illumina, San Diego, USA; 2× 250 bp) at Novogene (Cambridge, UK).The raw reads were analysed according to Hassenrück 2022 [63].Brief ly, barcodes were extracted followed by de-multiplexing and primer clipping using cutadapt (version 2.1).The de-multiplexed reads were then analysed using dada2 (version 1.16.0).In detail, the quality of sequencing reads was checked and then the reads were trimmed, followed by the correction of error estimates and error learning in order to retrieve the final clean reads.The clean reads were then dereplicated and denoised, which were further merged for both forward and reverse reads to obtain the long sequences.The chimera reads were then filtered and the unusual reads below 248 bp or above 256 bp were removed.Taxonomy was assigned using the final reads based on the database SILVA 138 database [64] For each sample, 8000 to 60 000 reads were retrieved for abundance analysis.

Phylogenetic analyses
The concatenated set of 71 ribosomal protein genes based on a previously published hidden Markov Model profile [77] were used for phylogenetic analyses in Anvi'o (6.1) [78].Maximum-likelihood trees were built using IQ-TREE (1.6.12)[79] with the best-fit model (LG + F + R7) and 1000 times ultrafast bootstrapping.The tree files were edited using the online tool iTOL [80].
Because of the different names among GTDB and Silva databases for Sva1033 (Desulfuromonadales), 16S rRNA genes in the MAGs for all analysed Desulfuromonadales were extracted by Barrnap (version 0.3, http://www.vicbioinformatics.com/software.barrnap.shtml).The 16S rRNA genes together with references were aligned with SINA Aligner [81].Maximum-likelihood tree was inferred using RAxML (8.2.11) with rapid bootstrapping and the GTRGAMMA model [82].In our previous work, we have not found evidence for sulfate-reducing archaeal taxa [83], and thus archaeal analysis was not included in this study.
Amino acid sequences of reductive dehalogenase were used for orthology analysis.Reference sequences with 100 hits were retrieved from NCBI nonredundant protein database by blasting sequences of reductive dehalogenase of SRM obtained from this study.The combined sequences of each protein were filtered and clustered using cd-hit (Version 4.6.8)[84] with cut-off of 70%, which was followed by MAFFT-LINSI (Version 7.455) alignment with default parameters [85] and trimming by BMGE with f lags "-t AA -m BLOSUM30" [86].Un-rooted phylogenetic trees for protein sequence were built with 1000 times ultrafast bootstrapping using IQ-TREE with the best-fit models (LG + R8).

Turnover of fermentation products in sediment incubations
RNA-SIP incubations with 13 C labeled substrates and sulfate were set up in order to investigate the turnover of different fermentation products and the activity of the associated microorganisms.Three different sediments were compared including the temperate site Helgoland mud area (North Sea) and two permanently cold sites from Cumberland Bay (South Georgia, sub-Antarctic) and Hornsund fjord (Svalbard, Arctic).The degradation of 13 C labeled substrate was monitored by the formation of 13 CO 2 in the headspace over time to determine incubation stopping time by avoiding cross-feeding when samples were incubated for too long time (Fig. 1).Delta 13 C values of CO 2 increased rapidly to ∼1030-4470 within 15 days for sediments from all three sites.In Helgoland mud sediment incubations, the addition of antibiotics (to suppress SRM competitors [52][53][54][55][56]) to incubations slowed down the formation of 13 CO 2 (∼1250-2760 ) compared to those without antibiotics (∼2470-4470 ) within 15 days (Fig. 1A).In contrast, antibiotics had a smaller inhibitory effect on the turnover of fermentation products in incubations using Cumberland Bay sediment (Fig. 1B).The chosen incubation times were similar for incubations with and without antibiotics (Fig. 1).

Differential activities of sulfate-reducing microorganisms and other active microorganisms
Acetate, lactate, propionate, butyrate, and ethanol were used at a concentration of 60 μM carbon, similar in situ (15-30 μM dissolved organics carbon), in RNA-SIP incubations to identify those microorganisms actively assimilating labeled substrate during fermentation product degradation.In contrast to the isotopically "light" RNA and unlabeled controls, we recovered several bacterial groups, which were highly abundant in the isotopically "heavy" RNA fractions in their respective incubations (Fig. 2, Supplementary Figs S1 and S2).Most notably, Desulfuromonadales members (∼15%-50%) were highly active in Helgoland mud and Cumberland Bay sediments (Fig. 2A and B).In addition, members of Arcobacteraceae (∼18%-65%) were also active when amended with acetate, lactate, and propionate (Helgoland mud) or propionate only (Cumberland Bay) (Fig. 2A and B).We found that those Desulfuromonadales groups were likely using iron oxides (lepidocrocite) as electron acceptors in methanic, sulfate-depleted sediments from Helgoland (Supplementary Fig. S3).Arcobacteraceae (>50%) also dominated in Hornsund fjord sediment incubations with 13 Clabeled acetate, lactate, and propionate, whereas active Desulfuromonadales (∼25%) were found in 13 C-butyrate incubations of Cumberland Bay sediments.Other notable findings include large active populations of Sedimenticolaceae (up to 20%), Ferrimonas (up to 50%), and Amphritea (up to 74%), which are known as metal reducers as well [40,[87][88][89][90], respectively, in specific incubations (Fig. 2A and B).A rather uniform picture emerged from ethanol-amended incubations, in which Desulfuromonadales dominated fermentation product degradation in incubations from all three sites.However, canonical SRM were not very active in the incubations with organic fermentation products (acetate, lactate, propionate, butyrate, and ethanol), only showing minor activity in a few incubations (with propionate [Helgoland mud] and lactate[Cumberland Bay]).Instead, Desulfobacteraceae, Desulfocapsaceae, and Desulfovibrionaceae were strongly stimulated (∼37-80% in the heavy fraction) when H 2 and 13 CO 2 were amended in the incubations for all the three sites (Fig. 2).

Sulfate-reducing microorganisms were only utilizing organic fermentation products under inhibition of other microbes
Since SRM unexpectedly had a minor contribution to organic fermentation product degradation, transcripts of dsrA (alpha subunit of the dissimilatory sulfite reductase, a marker gene for sulfate reduction [91]) were quantified within SIP fractions in order to understand the participation of sulfate reduction in incubations with representative cold (Cumberland Bay) and temperate (Helgoland) sediments (Fig. 3).We found that transcripts of dsrA from the light fraction were most abundant in any incubation, regardless if 13 C-labeled or unlabeled substrates were used, while the heaviest fractions always had the lowest amount of dsrA transcripts.In contrast to the organic fermentation products, dsrA abundances in the heavy fractions from inorganic H 2 /DIC were comparatively high, in line with the activity of H 2 /DIC utilization by SRM (Figs 2 and 3).

Versatility of carbon metabolic pathways for active fermentation product utilizers and noncanonical sulfate-reducing microorganisms
In SIP incubations, we identified the activity of fermentation product degraders and some SRM.In order to have a deeper insight into the genomes of SRM and other fermentation product degraders in sediments, we screened 36 metagenome-assembled genomes (MAGs) with middle to high quality from the original sediments and sediment enrichments (Fig. 5, Supplementary Table S1).These MAGs were affiliated to the active fermentation product degraders including Sedimenticolaceae, Halarcobacter, Sva1033, Desulfuromonadaceae Desulfovibrionaceae, Desulfocapsaceae, and Desulfobacteraceae, and noncanonical SRM such as BSN033 (class level of Desulfobacterota), Syntrophorhabdia, Desulfurivibrionaceae, Desulfatiglandales, C00003060 (order level of Desulfobacteria), and other Desulfobacterales (Fig. 5, Supplementary Fig. S4, Supplementary Table S1).
Based on metagenomic analysis, pathways involved in the degradation of fermentation products were widely identified in most MAGs of active microbes.In detail, MAGs of Desulfuromonadales such as Sva1033 and Desulfuromonas included the pathways for acetate, lactate, propionate, and butyrate dissimilation to the corresponding acyl-CoA, which can be oxidized to CO 2 via the acetyl-CoA pathway or citric acid cycle coupled to iron reduction, but they did not feature genes for H 2 oxidation (Fig. 6, Supplementary Tables S2, S3).Halarcobacter were equipped with pathways for acetate, lactate, and propionate degradation similarly to Desulfuromonadales (Fig. 6, Supplementary Table S2).In addition, Pelobacteraceae had gene sets for ethanol oxidation to acetate, a known feature of Pelobacter spp.(now partly known as Syntrophotalea spp.; [92]) such as Pelobacter acetylenicus (now Syntrophotalea acetylenica [92]) (Fig. 6, Supplementary Table S2).For active SRM, the metagenomic analysis indeed ref lected that H 2 /CO 2 utilization via Wood-Ljungdahl (WL) is a common feature for SRM indicated by the presence of complete WL pathway and hydrogenases including group 1a, 1b, or 1c (hydrogenotrophic respiration using sulfate [93,94]) in the MAGs of Desulfobacterales and Desulfobulbales (Supplementary Table S3).
We further checked the metagenomic pathways of noncanonical SRM to predict their potentials for alternative organic carbon degradation pathways.We found that SRM groups including Desulfatiglandales and other Desulfobacterales harbored the pathways for reductive dehalogenesis and potentials involving in aromatic compound degradation (Supplementary Figs S5 and S6, Supplementary Table S2).Furthermore, multiple MAGs have pathways for glycolysis and fatty acid degradation in SRM including Syntrophorhabdia, Desulfobulbales, C00003060, and other Desulfobacterales (Supplementary Fig. S4).Such sugar utilization was also indicated from the SIP incubations in which the RNA of Desulfobacterales (Desulfocapsaceae) was specifically labeled by 13 C-glucose in the presence of antibiotics (Supplementary Figs S7 and S8).

Partitioning of fermentation product degradation among canonical fermentation product degraders
Using ultra-high sensitivity RNA-SIP in combination with relevant 13 C-labeled fermentation products, we found a consistent partitioning pattern in temperate and permanently cold sediments: organic fermentation products were mostly used by known and novel metal-reducing bacteria, whereas SRM were strongly active in using H 2 autotrophically.
In general, the concentration of organic fermentation product in coastal sediments is low, ranging from nanomolar to fewhundred micromolar [9][10][11][12][13][14].In incubations, we used only 60 μM carbon (15-30 μM fermentation products) for SIP and thus, matched closely in situ concentrations of organic fermentation products.This in turn is likely avoiding enrichment artifacts originating from irrelevant carbon compound concentrations in incubations and thus, better ref lects microbial activities in the studied sediments.We did not detect a strong enrichment of SRM on H 2 /CO 2 since the abundance of unlabeled dsrA was still higher in the light fractions than that of the heavy fractions.Copies of dsrA transcript in different fractions from the RNA-SIP samples; number of transcripts from the heavy (H: 1.815-1.830g/ml), middle (M: 1.803-1.819g/ml), light (L: 1.792-1.808g/ml), and ultra-light (UL: 1.781-1.794g/ml) fractions of RNA-SIP samples from incubations using Helgoland mud (A) and Cumberland Bay (B) sediment; note that values below 100 copies might be not accurate because the detection threshold was above 100 (see Method).
Based on this, we found that Desulfuromonadales (Desulfuromusa, Geothermobacter, Syntrophotalea, and Sva1033), Arcobacteraceae, Ferrimonas, Sedimenticolaceae, Amphritea, and Syntrophotalea were the main organic fermentation product utilizers while SRM were incorporating 13 CO 2 with H 2 as electron donor.The former microorganisms are not known as SRM [25,38,92], which is corroborated by the absence of genes encoding the dissimilatory sulfate reduction pathway in their MAGs (Supplementary Table S2).As revealed in our studies, Desulfuromonadales were iron-reducing bacteria (Supplementary Fig. S3, Fig. 6).Certainly, dissolved Fe(II) and Mn(II) were not detectable due to low concentration of amended substrates in SIP incubations, and abiotic reactions resulting in the formation of insoluble minerals (e.g.siderite) [95].However, the identified active non-SRM (Desulfuromonadales -Desulfuromusa, Geothermobacter, and Sva1033) here were identified previously in Helgoland mud and Cumberland Bay sediment incubations as iron-reducing microbes (Supplementary Fig. S3) [25,58].It is also feasible that Arcobacteraceae reduce Fe(III) or Mn(IV) as electron acceptor linked to the oxidation of organic carbon compounds [25,40,96,97].The other active 13 C-labeled bacteria, such as Sedimenticolaceae, Ferrimonas, and Amphritea, can utilize metal oxides such as manganese, iron, and selenium oxides or have been identified in incubations amended with metal oxides in several studies [25,40,[87][88][89][90]98].Gas exchange in headspace and preincubation of sediments (see Method) ensured that alternative electron acceptors (e.g.traces of oxygen, nitrate) were depleted.In our study, we have used marine sediment from one temperate and two permanently cold sites.The temperate sediment from the Helgoland mud area is characterized by high sedimentation rate resulting in deeply buried iron oxides, which can fuel microbial activity in anoxic sediment layers [58,99].In cold sediment from Cumberland Bay (South Georgia, sub-Antarctic) and Hornsund fjord (Svalbard, Arctic), metal oxides are originating from glacier-associated erosion and meltwater, and thus metal oxides such as iron and manganese oxides are present [100,101].Thus, the identification of active degraders of fermentation product as known metal reducing microorganisms suggests that sediments contained sufficient amounts of metal oxides as electron acceptors to support their metal-reducing activity, even in deep sediment from the methanic zone (Supplementary Fig. S3) [25,40].
In our SIP incubations, SRM were not actively degrading and incorporating label from organic fermentation products; this is surprising as sulfate is present at high concentration (15 to 28 mM in situ; 18 mM in our incubations) in the upper sediment layer of the studied sediments, and SRM have been identified in this and previous studies at high abundance (Supplementary Fig. S9) [25,43,99,102].Albeit their limited activity, SRM were still active in incubations but rather using other substrates since dsrA transcripts were much more abundant in the light fraction than those from heavy fraction, and thus, were not labeled from 13 C-organic fermentation products (Figs 2 and 3).Only in the presence of antibiotics, SRM were found to incorporate 13 C-label from added organic substrates (Fig. 4), ruling out the possibility that amended low concentrations of organic fermentation products were limiting the activity of SRM, and corroborating that the turnover of organic fermentation products is apparently not under thermodynamic control in marine sediments [14][15][16].The low in situ temperature might trigger the observed partitioning of organic fermentation product degradation among SRM and other microbes: low activity of SRM at temperatures below 10 • C; iron reduction in temperate marine sediment was apparently favored at low temperatures (4-10 • C) [62], while the optimal temperature for sulfate reduction was found to be above 15 • C for cold sediments [15], hinting to a potentially better adapted metabolism of ironreducing microorganisms at lower temperature.With antibiotics, most metal-reducing bacteria were inhibited and therefore SRM were identified, which further indicates that there was no O 2 contamination that might inhibit the activity of SRM.In this study, we used temperate and permanently cold sediment and found Figure 4. Identification of the active fermentation product degraders using RNA-SIP in incubations amended with 13 C-labeled substrates, sulfate, and antibiotics; relative abundance of 16S rRNA gene sequences of active bacterial fermentation product degraders in total bacteria from RNA-SIP gradient fractions in the Helgoland mud (A) and Cumberland Bay (B) sediment incubations; Helgoland mud and Cumberland Bay sediments were used as representatives for temperate and permanently cold sediment, respectively; density was indicated as the average density of combined fractions for RNA-SIP samples.that these two types of sediments have very similar features for the activity of SRM at relatively low temperate condition.It is still very interesting for the future study to test the ability of SRM for organic fermentation product utilization in high temperature sediments.
Unlike organic fermentation products, the H 2 concentration is typically under thermodynamic control in marine sediments [15,49,103,104], thus, the hydrogen partial pressure is determined by the free energy available of the energetically most favorable electron accepting process.In sulfate reduction-dominated coastal sediment, hydrogen partial pressures were on similar levels regardless whether metal oxides were added or sulfate was present [103], suggesting that out-competition of SRM by metal reducing microorganisms based on terminal electron acceptor thermodynamics [105] was not operative.Albeit high quality of MAGs, we found a lack of respiratory H 2 -uptake  S1-S3), suggesting that these iron-reducing microorganisms cannot oxidize hydrogen and fix CO 2 .In contrast, SRM were strongly stimulated in SIP incubations with H 2 / 13 CO 2 .Many SRM species in our marine sediment incubations are actually autotrophs capable of fixing inorganic carbon and using H 2 as electron donor (Fig. 6).Besides primary fermentation, secondary, syntrophic oxidations of organic fermentation products are important sources of H 2 in marine sediments [2], and H 2 can contribute up to 75% in electron f low [106].Based on our study, SRM can indirectly participate in organic matter degradation by interspecies hydrogen transfer interactions during fermentation of macromolecules such as protein, carbohydrates, and cell biomass [33,37].Thermodynamically, hydrogen is also a sufficiently strong reductant for CO 2 fixation in the relevant reactions (oxidation of Figure 5. Maximum likelihood tree of 71 concatenated bacterial genes; Uncl.: unclassified; the tree was built using IQ-TREE (1.6.12) with the best-fit model (LG + F + R7) and 1000 times ultrafast bootstrapping; see Supplementary Table S1 for the details of MAG information.).In fact, some genera affiliated with the family Desulfobacteraceae remain lithoautotrophic in the presence of H 2 when acetate is amended [108], indicating H 2 -based preference for lithoautotrophy in some SRM.In addition, other than H 2 , direct interspecies electron transfer might be another mechanism supporting autotrophy of SRM in environments.For example, Desulfosarcina/Desulfococcus utilize electrons transferred from their methanotrophic ANME partners for autotrophic growth [109], a syntrophic consortium mediating anaerobic methane oxidation [110].Although ∼100 μM H 2 in slurry was amended, given that SRM were not active for almost all the common organic fermentation products (acetate, lactate, propionate, butyrate, ethanol) and the substantial overpressure in sediments [111], utilization of H 2 /CO 2 or interspecies electron most likely ref lected the activity of SRM in situ.Overall, the SIP results suggested that SRM were of minor importance during organic fermentation product degradation (Fig. 6).

Beyond fermentation products: non-canonical organic carbon utilization by sulfate-reducing microorganisms
Beyond active SRM, marine sediments inhabit diverse uncultivated SRM (Supplementary Fig. S9), and thereby this leads to the question of what their potential role in carbon degradation in these sediments is.Our SIP incubations amended with glucose suggested that SRM were able to degrade glucose in the presence of antibiotics (Supplementary Fig. S7), and thus they had the ability for glucose uptake into cells and also harbor the complete pathway for glucose degradation (Supplementary Fig. S5).Although the SRM were outcompeted by sugar fermenters when antibiotics were not present (Supplementary Fig. S8), it is still notable that some other carbohydrates might be the substrates for SRM in marine sediments, which is in line with the observation of carbohydrate degradation by SRM in a few studies [112][113][114][115]. Apart from glucose utilization, our enrichment incubations and metagenomic analysis also indicated a wider spectrum of substrate utilization by noncanonical SRM than expected, such as halogens (see supplemental discussion), and thus SIP experiments should focus on the carbon utilization versatility of SRM in the future study.
Our study has new implications for the role of SRM on biogeochemical cycling in marine sediment: (i) canonical SRM have limited contribution on the degradation of organic fermentation products at low concentrations in marine sediments, (ii) canonical SRM appear to prefer autotrophic lifestyle using H 2 oxidation instead of heterotrophy, (iii) many SRM have potentials for utilizing noncanonical carbon compounds.Canonical SRM may actually have an autotrophic lifestyle in environments.In the SRZ, CO 2 assimilation has been identified in some archaeal groups such as Lokiarchaeota, Bathyarchaeota, and ANMEs [116][117][118], while their activities are quite low.It has been recognized that autotrophy is an important lifestyle for sulfur oxidizers with a relatively high activity of bacteria [119].Based on our findings, we propose that SRM are additional CO 2 assimilators that have to be considered to regulate carbon f luxes in marine sediments.

Figure 1 .
Figure 1.Turnover of fermentation products in SIP incubations; development of δ 13 C-values of headspace CO 2 in incubations amended with low concentration of fermentation products using Helgoland mud (A), Cumberland Bay (B), and Hornsund fjord sediment (C) (n = 3, error bar = SD); VPDB: The Vienna Peedee belemnite; samples for SIP analysis were harvested after the last time point.

Figure 2 .
Figure 2. Identification of the active fermentation products degraders using RNA-SIP in incubations amended with 13 C-labeled substrates and sulfate; relative abundance of 16S rRNA gene sequences of active bacteria fermentation product degraders in total bacteria from RNA-SIP gradient fractions in the Helgoland mud (A), Cumberland Bay (B), and Hornsund fjord sediment (C) incubations; × indicates that cDNA synthesis failed because of insufficient amount of RNA in these fractions; density was indicated as the average density of combined fractions for RNA-SIP samples; for sampling time points, see Figure 1.

Figure 3 .
Figure 3.Copies of dsrA transcript in different fractions from the RNA-SIP samples; number of transcripts from the heavy (H: 1.815-1.830g/ml), middle (M: 1.803-1.819g/ml), light (L: 1.792-1.808g/ml), and ultra-light (UL: 1.781-1.794g/ml) fractions of RNA-SIP samples from incubations using Helgoland mud (A) and Cumberland Bay (B) sediment; note that values below 100 copies might be not accurate because the detection threshold was above 100 (see Method).

[
NiFe]-hydrogenase and incomplete WL pathways in MAGs of Sva1033 and Desulfuromonas spp.(Supplementary Tables

Figure 6 .
Figure 6.Patterns of fermentation products utilization in SRZ of temperate and permanently cold marine sediments based on SIP and metagenomics; solid and dashed brown lines indicate the strong and weak activity for fermentation product utilization, respectively; see TableS2for the details of annotated genes; TCA: Citric acid cycle.