Metagenomic insights into the metabolism of microbial communities that mediate iron and methane cycling in Lake Kinneret iron-rich methanic sediments

Complex microbial communities facilitate iron and methane transformations in anoxic methanic sediments of freshwater lakes, such as Lake Kinneret (the Sea of Galilee, Israel). The phylogenetic and functional diversity of these consortia are not fully understood, and it is not clear which lineages perform iron reduction and anaerobic oxidation of methane (AOM). Here, we investigated microbial communities from both natural Lake Kinneret iron-rich methanic sediments (> 20 cm depth) and ironamended slurry incubations from this zone using metagenomics, focusing on functions associated with iron reduction and methane cycling. Analyses of the phylogenetic and functional diversity indicate that consortia of archaea (mainly Bathyarchaeia, Methanomicrobia, Thermoplasmata, and Thermococci) and bacteria (mainly Chloroflexi (Chloroflexota), Nitrospirae (Nitrospirota), and Proteobacteria) perform key metabolic reactions such as amino acid uptake and dissimilation, organic matter fermentation, and methanogenesis. The Deltaproteobacteria, especially Desulfuromondales (Desulfuromonadota), have the potential to transfer electrons extracellularly either to iron mineral particles or to microbial syntrophs, including methanogens. This is likely via transmembrane cytochromes, outer-membrane hexaheme c-type cytochrome (OmcS) in particular, or pilin monomers (PilA), all of which were attributed to this lineage. Bona fide anaerobic oxidizers of methane (ANME) and denitrifying methanotrophs Methylomirabilia (NC10) may mediate AOM in these methanogenic sediments; however we also consider the role of methanogens in active AOM or back flux of methanogenesis. Putative aerobes, such as methaneoxidizing bacteria Methylomonas and their methylotrophic syntrophs Methylotenera, are found among the anaerobic lineages in Lake Kinneret iron-amended slurries and are also involved in the oxidation of methane or its intermediates, as suggested previously. We propose a reaction model for the metabolic interactions in these sediments, linking the potential players that interact via intricate metabolic tradeoffs and direct electron transfer between species. Our results highlight the metabolic complexity of microbial communities in an energy-limited environment, where aerobe and anaerobe communities may co-exist and facilitate AOM as one strategy for survival.

Abstract. Complex microbial communities facilitate iron and methane transformations in anoxic methanic sediments of freshwater lakes, such as Lake Kinneret (the Sea of Galilee, Israel). The phylogenetic and functional diversity of these consortia are not fully understood, and it is not clear which lineages perform iron reduction and anaerobic oxidation of methane (AOM). Here, we investigated microbial communities from both natural Lake Kinneret iron-rich methanic sediments (> 20 cm depth) and ironamended slurry incubations from this zone using metagenomics, focusing on functions associated with iron reduction and methane cycling. Analyses of the phylogenetic and functional diversity indicate that consortia of archaea (mainly Bathyarchaeia, Methanomicrobia, Thermoplasmata, and Thermococci) and bacteria (mainly Chloroflexi (Chloroflexota), Nitrospirae (Nitrospirota), and Proteobacteria) perform key metabolic reactions such as amino acid uptake and dissimilation, organic matter fermentation, and methanogenesis. The Deltaproteobacteria, especially Desulfuromondales (Desulfuromonadota), have the potential to transfer electrons extracellularly either to iron mineral particles or to microbial syntrophs, including methanogens. This is likely via transmembrane cytochromes, outer-membrane hexaheme c-type cytochrome (OmcS) in particular, or pilin monomers (PilA), all of which were attributed to this lineage. Bona fide anaerobic oxidizers of methane (ANME) and denitrifying methanotrophs Methylomirabilia (NC10) may medi-ate AOM in these methanogenic sediments; however we also consider the role of methanogens in active AOM or back flux of methanogenesis. Putative aerobes, such as methaneoxidizing bacteria Methylomonas and their methylotrophic syntrophs Methylotenera, are found among the anaerobic lineages in Lake Kinneret iron-amended slurries and are also involved in the oxidation of methane or its intermediates, as suggested previously. We propose a reaction model for the metabolic interactions in these sediments, linking the potential players that interact via intricate metabolic tradeoffs and direct electron transfer between species. Our results highlight the metabolic complexity of microbial communities in an energy-limited environment, where aerobe and anaerobe communities may co-exist and facilitate AOM as one strategy for survival. (Saunois et al., 2020), almost entirely from freshwater systems (Bastviken et al., 2011;Zhu et al., 2020). In these freshwater environments, anaerobic oxidation of methane (AOM) consumes over 50 % of the produced methane (Segarra et al., 2015). These environments are usually low in sulfate, and thus other terminal electron acceptors such as metals, iron (Fe) in particular; nitrate; nitrite; and humic acids become available for this process, as explored in recent studies (Raghoebarsing et al., 2006;Ettwig et al., 2010;Adler et al., 2011;Haroon et al., 2013;Norði et al., 2013;Scheller et al., 2016;Bai et al., 2019). However, the diversity and metabolic potential of the microbial communities in natural anoxic ferruginous sediments are not fully understood (Vuillemin et al., 2018).
This study investigates the microbial communities in the sulfate-depleted, iron-rich methanic sediments of Lake Kinneret, a monomictic lake located in the north of Israel at mid-latitudes. The lake is thermally stratified from March until December. During the stratified period, due to the spring phytoplankton bloom decline, the hypolimnion of the lake becomes enriched with particulate organic carbon, and oxygen is gradually depleted . At the top 5 cm of the sediment, sulfate reduction is the main process throughout the year . At 5 cm below the sediment surface, methanogenesis becomes dominant . Methane production peaks at 7-12 cm in the sediment while sulfate is depleted (below 10 µM). At 20 cm depth, methane decrease is observed, and ferrous iron concentrations increase to about 200 µM Bar-Or et al., 2015). Iron-dependent AOM (Fe-AOM) appears to play a role in methane removal from this deep methanogenic zone of the lake sediments, based on porewater depth profiles, rate modeling, microbial profiles, and incubation experiments on top cores and slurries Sivan et al., 2011;Bar-Or et al., 2015.
Complex microbial consortia mediate the biogeochemical transformations in anoxic lake sediments (Vuillemin et al., 2018). These include methylotrophic and hydrogenotrophic methanogens from the Thermoplasmata, Methanomicrobia, Methanobacteria, and Bathyarchaeota clades, as well as potential sulfate or iron reducers, such as Deltaproteobacteria, Firmicutes, and Nitrospirae lineages (Vuillemin et al., 2018). Alongside the abovementioned lineages, Bar-Or et al. (2015) identified acetoclastic methanogens such as Methanothrix (formerly Methanosaeta, Methanosarcinales, recently reclassified as Methanothrichales by the Genome Taxonomy Database, GTDB;Parks et al., 2018), as well as the H 2 and CO 2 -using methanogenic Methanomicrobiales genera Methanolinea and Methanoregula in Lake Kinneret sediments. Very few bona fide anaerobic oxidizers of methane (ANME) were found, and the identity of microbes that perform Fe-AOM in these sediments was unknown. Desulfuromondales (Deltaproteobacteria/Desulfobacterota) and Thermodesulfovibrio (Nitrospirae/Nitrospirota) were suggested to couple catabolism of organic substances to the reduction of metals (Bar-Or et al., 2015). We note that deltaproteobacterial lineages have been recently reclassified based on the GTDB (e.g., Desulfuromondales were reclassified as Desulfuromonadota phylum; Parks et al., 2018). Hereafter, we use the "Deltaproteobacteria" terminology to describe the taxonomy of these lineages, as implemented in the Silva 132 database, as the new Genome Taxonomy Database classification has not been peer-reviewed and widely accepted at the time of preparation of this paper.
Here, we use metagenomic analyses to explore the phylogenetic diversity and the metabolic potential of the microbial communities in the deep (27-41 cm) ironrich methanic part of Lake Kinneret sediments (LK-2017) and the slurry incubations from the Bar-Or et al. (2017) study. These incubations -including (a) 13 CH 4 , (b) 13 CH 4 + hematite, or (c) 13 CH 4 + amorphous iron + molybdate (A.Fe(III) + MoO 4 ) -produced substantial amounts of 13 C-labeled dissolved inorganic carbon (DIC) over 470 d Fig. S1 in the Supplement). In all the treatments, the diversity of bacteria and archaea was similar to that of the natural sediments. We supplemented these data with analyses of bacterial and archaeal diversity at the 16S rRNA gene level for a wider range of treatments analyzed by Bar-Or et al. (2017). We sought genetic evidence for the ability of the microorganisms to perform iron reduction and produce or oxidize methane and aimed to identify and classify genes that are necessary to catalyze reactions of the respective pathways. As in most natural sediments iron and manganese are present mainly as low-solubility oxide minerals at circumneutral pH (Norði et al., 2013;He et al., 2018) and microbial cells are impermeable to these solid minerals (Shi et al., 2007), we aimed to identify the strategies that may allow microorganisms to cope with this constraint and potentially perform metal-dependent AOM, including (1) direct transfer of electrons to the mineral at the cell surface via an electron carrier, such as cytochrome c (Shi et al., 2007); (2) bridging electrons to a mineral via a cellular appendage, such as conductive nanowires (Gralnick and Newman, 2007;Schwarz et al., 2007;Shi et al., 2016;Lovley and Walker, 2019); (3) indirect electron transfer via metal chelate (Paquete et al., 2014); and (4) indirect electron transfer by electron shuttling compounds, such as quinones or methanophenazines (Newman and Kolter, 2000;Wang and Newman, 2008;He et al., 2019). porewater extracted from the parallel geochemical zone to create homogenized 1 : 5 sediment-to-porewater ratio slurries. The homogenized slurry was transferred under continuous N 2 flushing in 40 mL portions into 60 mL bottles, crimped, and sealed. The sediment slurries were amended with 13 C-labeled methane and, except the natural sample, treated with different iron oxide minerals. Each sample set was kept (I) without inhibitors, (II) with inhibition of sulfate reduction and sulfur disproportionation by sodium molybdate (Na 2 MoO 4 ) addition, and (III) inhibition of methanogenesis by 2-bromoethanesulfonate (BES) addition. One of the sample sets was killed (autoclaved) as a control. The various treatments are summarized in the Supplement, Table S1. DNA was extracted from each incubation at the beginning and end of the experiment (after 470 d), and analysis of DNA 16S rRNA genes was performed for all of these incubations and the untreated sediments (Sect. S1 in the Supplement). Four metagenomic libraries (untreated sediments -at the initial time when the sediment arrived at the laboratory and before it was incubated (referred to as t0 hereafter) in 2013; incubations -+ 13 CH 4 , 13 CH 4 + hematite, and 13 CH 4 + amorphous (A.) Fe(III) + MoO 4 ) were prepared at the sequencing core facility at the University of Illinois at Chicago using the Nextera XT DNA library preparation kit (Illumina, USA). Our preliminary analyses of the microbial diversity (16S rRNA amplicons and metagenomics) at t0 in 2013 revealed contamination with common laboratory bacteria, such as Firmicutes and Bacilli (Figs. S2a and S4). To avoid the discovery of contaminant functions, we prepared a DNA library for an additional sample, collected from the same water and sediment depth in 2017 (LK-2017). A total of 12-28 million 2 × 150 bp paired-end reads per library were sequenced using Illumina NextSeq500.

Geochemical evidence for iron-coupled AOM in Lake Kinneret iron-rich methanic sediments
We explore here slurries amended with Lake Kinneret sediments from the deep methanic zone (26-41 cm). In this potentially ferruginous zone, sedimentary profiles show that the concentration of methane decreased from its maximum values of above 2 mM at around 10 cm depth to 500 µM at 40 cm depth, and that of dissolved ferrous iron increased (from 1-6 µM at the first 10 cm below the sediment surface to ∼ 60-100 µM, depending on sampling season). This, combined with an increase of δ 13 C of methane (from about −65 ‰ at 7 cm to −53 ‰ at 24 cm below the sediment surface) and a decrease of δ 13 C of total lipid compounds (from 27 ‰ at 23 cm to −31 ‰ at 27 cm below the sediment surface), suggests AOM in the deep sediment coupled to iron reduction Sivan et al., 2011). The iron-coupled AOM process was supported by rate modeling and by microbial profiles Sivan et al., 2011;Bar-Or et al., 2015. We note that alternative common electron acceptors were scarce: dissolved manganese oxides concentrations were ∼ 0.04 %, and nitrate and sulfate were below the detection limit . We amended the slurries with isotopically labeled 13 CH 4 , 13 CH 4 + hematite, and 13 CH 4 + amorphous iron + molybdate for 470 d. Methane concentration increased with time (by up to ∼ 650 µM in the treatment with hematite addition, up to ∼ 1000 µM enrichment in the natural treatment, and up to ∼ 300 µM in the treatment with amorphous iron + molybdate), reflecting net methanogenesis. In these incubations, we also observed a marked enrichment of labeled carbon in the DIC pool after 10 months of incubation (up to 250 ‰ enrichment in the treatment with hematite addition, up to 80 ‰ enrichment in the natural treatment, and up to 450 ‰ enrichment in the treatment with amorphous iron + molybdate, Fig. S1), indicating conversion of methane to DIC. Ferrous iron concentrations increased by ∼ 20-50 µM following iron oxide amendments (with and without molybdate addition), indicating that iron was reduced. The BES amendments resulted in the highest increase in ferrous iron concentrations (∼ 50-110 µM), most likely due to the abiotic reaction of BES with iron minerals. The evidence for iron reduction, together with the fact that δ 13 C DIC values increased by 250-450 ‰ in the different iron-amended treatments but not in methane-only additions (only up to 80 ‰, Fig. S1), suggests iron-coupled AOM. Sulfate did not seem to play a role in the AOM, as the addition of molybdate, sulfate reduction, and disproportionation antagonist did not inhibit methane turnover (Fig. S1). The addition of BES to specific slurries inhibited the production of δ 13 C DIC , indicating the essential role of methanogens in the AOM activity (Fig. S1).

Diverse microbial consortia mediate biogeochemical cycles in Lake Kinneret sediments
Diverse microbial consortia inhabit Lake Kinneret iron-rich methanic sediments (> 20 cm depth, Fig. 1). In these sediments, Bacteria outnumber archaea based on mapping of the metagenomic reads either to the Silva (V132) database of the 16S rRNA gene sequences (73-76 % and 24-27 % reads mapped to bacterial and archaeal sequences, respectively; Sect. S4, S.DB.1) or to the MAR (MARine) database of prokaryotic genomes (81-85 % and 15-19 % reads mapped to bacterial and archaeal sequences, respectively; Sect. S4, S.DB.2). We explored here mainly the microbial community in the deep sediment incubations (> 20 cm) amended with 13 CH 4 alone, 13 CH 4 with hematite, or 13 CH 4 with amorphous iron oxides plus molybdate using metagenome analyses (Fig. 1). This diversity of microbes resembled that described previously for Lake Kinneret sedimentary profiles with amplicon sequencing of the 16S rRNA gene (Bar-Or et al., 2015), as well as that determined with either amplicon sequencing or metagenomics in ferruginous lakes across the globe (e.g., Vuillemin et al., 2018;Kadnikov et al., 2019). According to the metagenome analyses, the variation in the diversity of microbial communities in the natural samples and the incubations was small, possibly because iron is not a limiting electron acceptor throughout the sampling interval in the sediments . Amplicon sequencing of the bacterial and archaeal 16S rRNA genes in a wider range of treatments (Figs. S2 and S3), which included additions of various iron minerals as well as amendments of molybdate (inhibitor of sulfate reduction) and BES (inhibitor of methanogenesis), also revealed minor changes in the phylogenetic diversity of the microbial consortia. Anaerolineales (Chloroflexi), Thermodesulfobrionia (Nitrospirae), and the deltaproteobacterial Sva0485 clade were among the most dominant bacterial lineages in these samples (3-6 % read abundance, Fig. 1). While all of these lineages may carry out dissimilatory sulfate reduction (Vuillemin et al., 2018), genetic evidence suggests that bacteria from the Sva0485 clade, which was recently named Candidatus Acidulodesulfobacterales, have the potential to reduce iron (Tan et al., 2019). Sva0485 was suggested recently to be involved in iron reduction also in the methanic zone of marine sediments (Vigderovich et al., 2019). Both these studies revealed a correlation between the distribution of this lineage and ferrous iron concentrations in sediment porewater. Interestingly, Ca. Acidulodesulfobacterales appear to thrive mainly under acidic conditions, while Thermodesulfobrionia was suggested to be either neutrophilic or alkaliphilic (Sekiguchi et al., 2008;Frank et al., 2016). Thus, the cooccurrence of these taxa hints that microenvironments with distinct physicochemical conditions may be present in Lake Kinneret sediments.
Methanomicrobiales (5-7 % read abundance) and Bathyarchaeia (4-7 % read abundance) were the most domi- Figure 1. Relative abundance of bacteria (black) and archaea (green) at the order level based on mapping of metagenomic reads to the Silva 132 database of the small subunit rRNA sequences. Lineages < 1 %, which account together for 28-32 % of the microbial community, were removed from the display. LK 2017 refers to the natural sediments at depth 26 cm at t0. nant archaeal lineages in the sediment (Fig. 1). The vast majority of the Methanomicrobiales sequences belonged to Methanoregulaceae genera Methanoregula and Methanolinea, which are known to obtain energy by CO 2 reduction to methane, with H 2 or formate as electron donors (Bräuer et al., 2015;Imachi and Sakai, 2016). Bathyarchaeia may occur in large numbers in lake sediments (e.g., Vuillemin et al., 2018;Kadnikov et al., 2019;Zhang et al., 2019). The role of these generalists -which are capable of using various carbon and energy sources, often fueling their metabolism through acetogenesis and methanogenesis -is not well understood (Evans et al., 2015;He et al., 2016;Yu et al., 2018;Zhou et al., 2018). Other notable archaeal lineages included the acetoclastic Methanothrix (1-3 % read abundance), which are often found in anoxic lake sediments (Smith and Ingram-Smith, 2007;Schwarz et al., 2007;Carr et al., 2018) as well as in oxygenated soil, as recently discovered for Methanothrix paradoxum (Angle et al., 2017). Their role in the anaerobic degradation of alkanes as syntrophs of Anaerolineaceae has been recently proposed based on enrichment culturing (Liang et al., 2015), as well as their ability to oxidize methane (Valenzuela et al., 2017(Valenzuela et al., , 2019. We also identified the putative obligate H 2 -dependent methylotrophic methanogen lineages Methanofastidiosales (Thermococci, 1-3 % read abundance) and Methanomassiliicoccales (Thermoplasmata, 0.4-1 % read abundance), as well as the putative degraders of detrital proteins Marine Benthic Group D (Thermoplasmata, 1-2 % read abundance) (Lazar et al., 2017;Evans et al., 2019). Less than 1 % of the total reads were mapped to sequences of the anaerobic methanotrophs, such as ANME-1 (0.3-0.8 %), as well as those of the nitrite-reducing methane oxidizers Methylomirabilales (NC10, 0.3-0.6 %, Sect. S4, S.DB.1).
Some type I Methylococcales methanotrophs were found (0.4-1.8 %) in Lake Kinneret sediments. This finding is supported by the quantitative polymerase chain reaction analyses of the pmoA gene ( Bar-Or et al., 2017); our analyses of bacteria diversity at the 16S rRNA gene level (Fig. S2); and the 13 C-labeled methane carbon incorporation in phospholipidderived fatty acids that are typical of type I methanotrophs (Bar-Or et al., 2017), suggesting that methane metabolism was active in these bacteria. Methylotrophic Methylotenera (recently reclassified as Burkholderiales, Betaproteobacteriales in Silva 132), which were shown to co-occur with type I methanotrophs under nearly hypoxic conditions (Beck et al., 2013;Cao et al., 2019), were also found (0-1 %, Sect. S4, S.DB.1). The mechanisms behind the increase and elevated activity of the presumably aerobic methanotrophs in the anoxic sediments have not been fully understood, although this phenomenon appears to be widespread (Bar-Or et al., 2017;Martinez-Cruz et al., 2018).

General metabolic potential
Based on the mapping of the metagenomic reads to ORFs to which KEGG orthology was assigned, several metabolic pathways were found to be dominant in Lake Kinneret ironrich methanic sediments (Fig. 2). Overall, all four samples had a similar metabolic repertoire. The vast majority of mapped reads were attributed to amino acid utilization and biosynthesis, suggesting that the turnover of organic nitrogen plays an important role in fueling these microbial communities. Indeed, the ORFs that encode the five components of the branched-chain amino acid transport system (KEGG IDs KO1995-KO1999) were among the top 30 most abundant KEGG functions (out of a total of 9058 KEGG IDs that were assigned to the metagenomic ORFs, Sect. S4, S.DB.3).
In agreement with previous metagenomic assessments of metabolic pathways in microbes from anoxic lake sediments (Vuillemin et al., 2018), fermentation and methanogenesis account for a substantial part of the metabolic repertoire (Fig. 2a). KEGG IDs that are associated with the fermentative metabolism -such as formate dehydrogenase (K00123), 2-oxoglutarate ferredoxin oxidoreductase (K00174-5), and acetolactate synthase (K01652) -were among the functions with the highest metagenomic coverage (Sect. S4, S.DB.3). Functional analysis of the metagenome suggests that carbon dioxide, formate, acetate, and methylated compounds can fuel methanogenesis (Fig. 2b). These findings are in line with the fact that archaeal lineages known to be capable of using the respective pathways were present (Fig. 1).

Metabolic potential for AOM
In the incubations, methane oxidation was indicated based on the transformation of 13 C-labeled methane to 13 C DIC and was suggested to involve methanogens due to the inhibition of this transformation by BES (inhibitor of the methyl coenzyme M reductase (McrA)). Based on the taxonomic assignment of genes that encode the enzymes in anaerobic methane metabolism pathways, archaeal lineages from Lake Kinneret iron-rich methanic sediments -such as Methanomicrobiales,Methanosarcinales,and Methanomassiliicoccales (66,26, and 4 % of reads mapped to K00399 McrA-encoding ORFs, respectively) -are capable of performing the methane transformations (Fig. 3). As described above, all of these lineages are methanogens that convert inorganic carbon, acetate, and methylated compounds to methane under most environmental conditions. No mcrA sequences were assigned to Bathyarchaeia (Fig. 3), in agreement with previous observations of the rare occurrence of the mcrA gene in this lineage (Evans et al., 2015;Lazar et al., 2015;He et al., 2016;Maus et al., 2018). Thus, Bathyarchaeia are most likely not involved directly in methanogenesis or methanotrophy.
AOM by reversal of methanogenesis is unlikely in obligate hydrogenotrophic methanogens Methanomicrobiales (Rotaru et al., 2014), and it has not been shown also in methylotrophic methanogens Candidatus Methanofastidiosa and Methanomassiliicoccales (Lang et al., 2015;. More likely candidates to perform this AOM in Lake Kinneret sediments are ANME-1, which has been shown to perform this process and exist in our deep sediment, and Methanothrix. Methanothrix is closely related to Methanosarcinales (now classified as order Methanothrichales in class Methanosarcinia) and was suggested to perform AOM (Valenzuela et al., 2017(Valenzuela et al., , 2019. Several studies suggest that reverse methanogenesis may occur in Methanosarcinales species, based on (1) observation of trace methane oxidation (Zehnder and Brock, 1979;Moran et al., 2005), (2) catalyzation of reverse methanogenesis upon insertion of an mcr gene clone of ANME-1 (Soo et al., 2016), and (3) phylogenetic affiliation of bona fide ANME such as Methanoperedenaceae (ANME-2d) and ANME-2a with this clade (Cai et al., 2018;. Methanothrix was shown to receive electrons from Geobacter or other organotrophs such as Sphaerochaeta during acetoclastic growth (Rotaru et al., 2014;Zhao et al., 2020), suggesting that its cell surface may be conductive. Moreover, the electron transfer enables this strict acetoclastic methanogen to use these electrons and reduce CO 2 to methane, a metabolic capability that was unknown in these organisms previously. In line with the abovementioned study, our metagenomic analyses revealed that all the seven enzymes needed to perform both forward and reverse methanogenesis (Fmd/Fwd, Ftr, Mch, Mtd, Mer, Mtr, and Mcr) were assigned to Methanosarcinales (Fig. 3). We note that, since we measured net methanogenesis, the observed 13 C-DIC enrichment can represent the back flux of the close-to-equilibrium methanogenesis. In principle, any methanogen can produce DIC via the back-flux reactions, yet we cannot estimate the isotopic fingerprint and the magnitude of this process in this set of incubations. Thermodynamically, both scenarios (back flux or active Fe-AOM) are plausible, and the bona fide Fe-AOM is feasible and competitive with organoclastic iron reduction ( G of about −200 kJ mol −1 for both processes (Table S2).

Genomic evidence for the microbial iron reduction in Lake Kinneret sediment
We asked whether direct extracellular electron transfer, either to a mineral or interspecies, could potentially play a role in Lake Kinneret methane oxidation, in case it is active Fe-AOM, and which taxa may be involved in this process. Direct interspecies electron transfer (DIET) between electrogenic microbes, such as Desulfuromondales, and their partners, such as Methanosarcinales methanogens, requires the presence of electrically conductive pili (e-pili) on the deltaproteobacterial partner, while in both DIET and Fe (metal)-AOM the archaeal partner/methane oxidizer needs to produce either secreted or membranal conductive entities (Rotaru et al., 2014;McGlynn et al., 2015;Holmes et al., 2018;Walker et al., 2020;Leu et al., 2020). It was previously suggested that thermophilic AOM coupled to sulfate reduction is conducted via DIET between ANME-1 and sulfate-reducing bacteria (Wegener et al., 2015); thus we ex- amined if DIET might also contribute to Fe-AOM in Lake Kinneret.
Our metagenomic results suggest that Lake Kinneret microbiota may transfer electrons directly to extracellular minerals, either through multiheme c-type cytochromes or microbial nanowires, according to mechanisms described in previous reviews (Lovley, 2011;Shi et al., 2016). We investigated the trans-membranal MHCs that are anchored in either the bacterial membrane or archaeal S layer, as well as secreted MHCs. The putative transmembrane and secreted MHCs were encoded by 66 and 592 ORFs per metagenome, respectively. In both MHC types, most of the ORFs were classified as deltaproteobacterial (Fig. 4, 36-52 % in transmembrane MHCs and 29-35 % in secreted MHCs). This is not surprising, as deltaproteobacterial lineages are known to reduce particulate metals through extracellular electron transfer (EET), and to conduct DIET (Leang et al., 2003;Reguera et al., 2005;Lovley, 2011;Adhikari et al., 2016). Other MHC sequences were associated with Nitrospirae, Chloroflexi, and Acidobacteria (both secreted and trans-membranal) and with Actinobacteria (secreted MHCs, Fig. 4). Very few or no archaeal sequences were detected in both types of MHC (0-1.5 %).
Some methanogens, such as Methanosarcina barkeri and Methanothrix sp., do not possess outer-surface MHCs, yet they are capable of DIET-based syntrophy (Rotaru et al., 2014;Holmes et al., 2018;Yee and Rotaru, 2020). Tubular sheaths that are made of major sheath protein (MspA) may enable some related archaea, including Methanothrix thermophila and Methanospirillum hungatei, to conduct electron transfer (Dueholm et al., 2015;Christensen et al., 2018;Liu et al., 2019a). The protein sheet of Methanothrix shoeghenii was described as concentrating metal ions like iron, copper, nickel, and zinc (Patel et al., 1986). It was thus proposed to give them an advantage in retrieving electrons from extracellular mineral-rich environments (Yee and Rotaru, 2020), such as our study site. BLAST searches using Methanothrix thermophila and Methanospirillum hungatei MspA queries (ABK14853.1 and WP_011449234.1) in Lake Kinneret metagenomes resulted only in poor (11.9-27.5 %) hits of the WP_011449234.1 query, all of which were annotated as a hypothetical Methanoregulaceae protein. Thus, no known Methanosarcinales MspA proteins were detected, and it is still unclear if and how these archaea conduct DIET.
Of the ANME playing a role in AOM, ANME-2d was recently proposed to perform MHC-mediated metal-dependent AOM (McGlynn et al., 2015;Ettwig et al., 2016;Fu et al., 2016;Scheller et al., 2016;Cai et al., 2018). Thus, we used the previously published ANME-2d MHC sequences (Sect. S4, S.DB.4) as a query for performing a BLAST search against the metagenomes. This analysis resulted in zero MHC BLAST hits, corresponding to the fact that only ANME-1 were found among ANME (as 16S rRNA gene sequences and low-quality bins), while ANME-2d were absent. Other Methanosarcinales lineages such as Methanosarcina acetivorans, were also suggested to conduct metal-dependent AOM (Cai et al., 2018;Leu et al., 2020). In Methanosarcinales, the dimeric membrane-bound HdrDE complex catalyzes the oxidation of CoM and CoB to CoMS-SCoB, and electron shuttling within the membranes is mediated by methanophenazines, which appear to be important in the reduction of extracellular iron by this lineage (Bond and Lovley, 2002;Sivan et al., 2016;Bar-Or et al., 2017;Holmes et al., 2019). While the taxonomic assignments of ORFs that encoded the HdrD subunit (K08264) were diverse (34-39 % Deltaproteobacteria, 9-14 % Chloroflexi, 2-4 % Bathyarchaeia, and 1.5-3 % Methanosarcinales; Fig. S5), 77-97 % of all the ORFs that encoded the HdrE subunit (K08265) belonged to Methanosarcinales genus Methanothrix (Fig. S5). However, these sequences were scarce (3 : 1000 hdrE to hdrA based on read mapping). We also detected all the possible genes involved in the formation of the membrane-bound coenzyme F 420 : methanophenazine dehydrogenase complex Fpo (fpoABCDHIJKLMNO) which couple reduction of F 420 H 2 with methanophenazine oxidation and proton translocation (Welte and Deppenmeier, 2014;Evans et al., 2019;Holmes et al., 2019), and the majority of Fpo-encoding ORFs were classified as Methanothrix (Sect. S4, S.DB.5). As we were unable to identify Methanosarcinales MHCs or proteins that encode other conductive features such as the tubular sheaths, the question regarding their involvement in Lake Kinneret metal-dependent AOM remains open.

Summary
Metagenomic analyses of natural deep iron-rich sediments of Lake Kinneret and slurry incubations from this zone suggest that, similarly to other freshwater systems (e.g., Vuillemin et al., 2018), a community of bacteria and archaea drives the mineralization of organic matter in the deep part of Lake Kinneret sediments, through degradation of amino and fatty acids, as well as hydrogenotrophic, acetoclastic, and methylotrophic methanogenesis (Fig. 5).
We present here a reaction model of possible methane and iron cycling routes (Fig. 5). Metagenomics suggests that the fermenters of amino acids and other products of necromass degradation are the abundant Anaerolineaceae, Thermodesulfobrionia, SVA0485, and Bathyarchaeia. One of the major end products of fermentation is acetate, which can be used as a substrate for the acetoclastic methanogens. Both the acetoclastic methanogenesis and CO 2 reduction to methane in these species are likely driven by syntrophy with Desulfuromondales spp., through DIET (Rotaru et al., 2014;Inaba et al., 2019;Wang et al., 2020). The syntrophy between Methanosarcinales with Desulfuromondales is likely in Lake Kinneret deep sediments, based on the fact that the vast majority of ORFs that are needed for the extracellular transfer of electrons were assigned to Desulfuromondales.
Our geochemical isotope labeling experiments suggest that AOM is plausible in the deep methanogenic sediments. AOM may be mediated by ANME, such as ANME-1 that were scarce may be coupled to iron reduction and may involve methanogens which have not been considered as bona fide ANME (Fig. 5). However, we cannot rule out the role of methanogenesis back-flux reactions, which could have produced the isotopically labeled DIC. Both modes of methane oxidation, via the Fe-AOM by an unknown methanogen and the methanogenesis back flux, are justified from the thermodynamic and kinetic perspectives.
Our results, as well as the previous analyses of fatty acids ( Bar-Or et al., 2017), suggest that the aerobic methaneoxidizing Methylomonas and its aerobic methylotrophic partner Methylotenera also have a role in methane oxidation in the anaerobic environment. However, the mechanism behind this process is unclear. One possibility is that a slow release of oxygen from particulate matter  could have fueled methane oxidation by Methylococcales, given that oxidation of methane in the absence of oxygen is unlikely. The alternative is that these lineages may be able to incorporate methanol derived from the incomplete process of reverse methanogenesis (Fig. 5), as was shown for ANME (Xin et al., 2004;Wegener et al., 2016). Another possible explanation for the methylated compound leakage is the re- Numbers of total reads per million reads mapped to a gene according to the treatments are listed on the right side of each bar plot. A.Fe(III) + MoO 4 : amorphous iron and molybdate. We note that the taxonomy of several lineages here is based on the NCBI classification and differs from that of the Genome Taxonomy Database (GTDB). In particular, Deltaproteobacteria have been reclassified in the GTDB; thus the Desulfuromondales lineages described here belong to the phylum Desulfobacterota and class Desulfuromonadia, and Myxococcales to the phylum Myxococcota. The phylum Nitrospirae is reassigned as Nitrospirota, and the order Betaproteobacteriales as Burkholderiales in Gammaproteobacteria. Bacteria and archaea lineages are represented by ellipses. Functions associated with iron reduction (multiheme cytochromes (MHCs), PilA, OmcS, and the complex Fpo) or methanogenesis (McrA: methyl coenzyme M reductase) are represented by different symbols, and these symbols are shown next to the name of the lineages which possess them. Aerobic (orange), anaerobic (green), and unconfirmed anaerobic (dashed green) routes are displayed by arrows. These routes of methane oxidation may occur in parallel. We note that the CO 2 can be produced by back-flux reactions in methanogenesis. The highest assigned taxonomic levels are shown: class (C), order (O), family (F), and genus (G). versibility of the enzymes involved in AOM (Thauer and Shima, 2008;Holler et al., 2011), which may lead to the formation of trace amounts of methylated substrates (Wegener et al., 2016). This option of methanol as an intermediate produced by the methanogenic archaea fits with the reported inhibition of methane oxidation upon the addition of BES (Bar-Or et al., 2017). This, as well as the functions of numerous other lineages that comprise the diverse consortia of Lake Kinneret deep iron-rich sediments, remains to be elucidated through further sequencing efforts, cultivation, and experimental studies.
Data availability. The metagenome and short reads are available as NCBI BioProject accession number PRJNA637457.
The datasets mentioned in the text have been deposited at Figshare in an open-access respiratory and can by accessed at https://doi.org/10.6084/m9.figshare.c.5245157.v1 (Elul et al., 2020; separate links to each dataset can be found in the Supplement).
Author contributions. OS and ZR designed the project. ME prepared the samples, extracted DNA, analyzed the data, designed and created the figures, and took the lead in writing the manuscript. IBO provided the samples and analyzed the 16S rRNA amplicons. MRB carried out the bioinformatics analyses and contributed considerably to the interpretation of the results and writing of the manuscript. All co-authors provided critical feedback and helped to shape and write the manuscript.