Key bacteria decomposing animal and plant detritus in deep sea revealed via long-term in situ incubation in different oceanic areas

Abstract

Transport of organic matter (OM) occurs widely in the form of animal and plant detritus in global oceans, playing a crucial role in global carbon cycling. While wood- and whale-falls have been extensively studied, the in situ process of OM remineralization by microorganisms remains poorly understood particularly in pelagic regions on a global scale. Here, enrichment experiments with animal tissue or plant detritus were carried out in three deep seas for 4–12 months using the deep-sea in situ incubators. We then performed community composition analyses as well as metagenomic and metatranscriptomic analyses. The results revealed strikingly similar microbial assemblages responsible for decomposing animal and plant detritus. Genes encoding peptidases and glucoside hydrolases were highly abundant and actively transcribed in OM enrichments, which confirmed the roles of these enriched microbial assemblages in organic decomposition. Marinifilaceae, Desulfocapsaceae, Spirochaetaceae, and o-Peptostreptococcales were found to potentially contribute to nitrogen fixation. These core bacteria, acting as cosmopolitan anaerobes in decomposing fast-sinking particulate OM, may have been underestimated in terms of their role in deep-sea microbial-mediated biogeochemical cycles during conventional sampling and diversity survey.

Introduction

Transport of organic matter (OM) in the form of animal and plant detritus occurs widely in global oceans, playing a crucial role in global carbon cycling [1]. Annually, ~485 000 teragram (Tg) C of net primary production was continuously generated from the oceanic euphotic zone [2] and 380–430 Tg C of terrestrial OM is imported into the oceans [3]. About 1–40% of those OMs in the upper layers sinks into the deep ocean, becoming the main source of energy and material for deep-sea ecosystems [4], among which 10–30% is of terrestrial origin [5]. A. J. West et al. estimated that a single storm event transported up to 3.8–8.4 Tg coarse woody debris into the ocean [6]. The zooplankton and phytoplankton remain, as well as a considerable part of the terrigenous OM originating from animal and plant tissues, are chemically characterized by high-weight proteins or polypeptides, fatty acids, polysaccharides (chitin, cellulose, and hemicellulose), and lignin content [7], respectively. The input of large amounts of OMs, such as plant fragments and animal carcasses, to the sea bottom contributes organic impulses to the deep-sea ecosystem, like the wood-fall and whale-fall [8–11].

Sinking particulate organic matter (POM), which functions as a conduit for exporting organic materials from the euphotic zone to the deep ocean [12], critically contributes to the establishment of habitats for deep-sea benthic microbial communities [13]. Through the water column, the microbial assemblages associated with sinking particulates in the upper ocean layer, such as photoautotrophs of Cyanobacteria, can be largely delivered to the abyss, while deep-sea bacteria respond rapidly to the elevated nutrient delivery of sinking POM [14, 15]. The sinking particulates that transit to the deep sea may contain energy-rich organic compounds or relatively energy-replete POM with fatty acids selectively preserved [12, 16]. High OM influx is usually accompanied by high sedimentation rates to the deep sea [17, 18]. The large and fast sinking particles may escape disaggregation, dissolution, solubilization, and even microbial decomposition en route to the deep sea [14], which determines the organic carbon and energy flux pattern of the deep sea and impacts the microbial communities in the deep sea [19, 20]. Considering the vast volume of the deep sea, processes of POM transformation and remineralization via microorganisms in benthic ecosystems are crucial to the global carbon cycle [21]. However, we only have scratches of information about the microbial processes on fast-sinking particles. Although scientists have conducted more in situ microbial diversity detection, environmental changes and community succession on whale-fall and wood-fall since the last century, it is unknown which microorganisms are involved in the transformation and degradation of macromolecular OMs, such as cellulose, hemicellulose, and protein [8]. Early studies have shown that the microbial community in whale-fall and wood falls is complex, containing microorganisms with different lifestyle and metabolic characteristics, such as sulfate reducing bacteria and sulfide oxidizing bacteria [22, 23].

We have known much about the bacterial diversity in seawater usually sampled with CTD cassettes, filter fractionation techniques, and sediment traps [12]. The bacteria thriving on nutrients during the “hot time” of bulky POM sinking may be different from the background seawater; a similar scenario is true for deep-sea sediments.

To define the key microbial taxa potentially participating in the transformation and degradation of POM sunk to the deep sea, in this study, we carried out in situ enrichment experiments using deep-sea in situ microbial incubator (DIMI), which is a deep-sea microorganism in situ incubation system [24]. We deployed DIMIs with plant detritus and animal tissue in the pelagic deep ocean of the Indian Ocean, the Pacifica Ocean, and the South China Sea. The OM input-spiked microbial assemblages were subjected to diversity analysis to identify the key OM decomposers active in situ and to meta-analysis of the genomic and transcriptomic data to reveal metabolic potentials. This study will gain insights into the microbial decomposition processes of large pieces of POM, for further understanding the key players in elemental biogeochemical cycles in global deep oceans.

Materials and methods

Deep-sea in situ incubation experiments

Deep-sea in situ incubation experiments were carried out from 2016 to 2018, on the seafloor in the South China Sea (SCS, the largest marginal sea of the West Pacific Ocean; 18.49° N, 116.27° E), the deep-sea basin beside the Southwest Indian Ridge in the Indian Ocean (IO, open sea; 33.28° S, 50.75° E), and a flat-topped seamount in the West Pacific Ocean (PO, open sea; 20.41° N, 160.77° E) at 3758 m, 4434 m, and 1622 m water depths, respectively (Fig. 1A, B).

In this study, various natural organic materials in form of animal tissues, plant detritus and oils were selected for in situ incubation (Table S1). The animal tissue substrates included fish muscle, fish scales, and shrimp muscle and the plant detritus included wood chips, wheat bran, wheat straw, and various seaweeds. In addition, fish oil (EPA and DHA) and vegetable oil as fatty acids were used here. Treatment incubation chambers (TICs) with organic materials were mounted on a DIMI device (Fig. 1C). Specific sample setup and handling were provided in Table S1 and the Supplementary Information.

The DIMI is a self-return deep-sea microorganism in situ enrichment system that is deployed at the interface of deep-sea seawater and sediment, and incubations are not in contact with the sediment [24]. In these situations, the DIMI free falls from the sea surface to the deep-sea bottom, and when it reaches the seafloor, in situ enrichment in the deep sea is triggered. The lid of each incubation chamber was kept open to ensure that substrates were accessible to the surroundings during enrichment. Four or twelve months later in this study, the DIMI was released by an acoustic signal, and the lid of each incubation chamber was immediately tightly closed under the action of a spring tension. The temperature and salinity of the surrounding seawater and the depths were recorded using a SBE 37-SM microCAT CTD recorder (Sea-Bird Scientific, Bellevue, WA, USA) mounted on the DIMI during the in situ enrichment.

Sampling and DNA and RNA extraction

After the DIMIs were recovered, we immediately processed the samples on board to obtain DNA and RNA. In brief, the solid substrates, as well as the silicate balls, were ground to powder in liquid nitrogen immersion for DNA extraction. Total DNA was extracted and purified with a DNeasy PowerWater Kit (Qiagen, Germany) according to the manufacturer’s protocol. Total RNA was extracted using the RNeasy minikit (Qiagen, Germany) according to the manufacturer’s instructions. Detailed sampling and DNA and RNA extraction are shown in the Supplementary Information.

16S rRNA gene sequencing and analysis

The V3-V4 hypervariable region of the bacterial 16S rRNA gene (~454 bp) was amplified and sequenced using a PE300 strategy on an Illumina MiSeq platform at Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China). The raw data was quality filtered with fastp v0.19.3 with default parameters [25] and merged with FLASH v1.2.11 [26]. Then the high-quality sequences were de-noised using DADA2 plugin [27] in the Qiime2 pipeline (version 2020.2) [28] with recommended parameters, and amplicon sequence variants (ASVs) were obtained. The taxonomic annotation of ASVs was based on the SILVA SSU 138 rRNA database. A detailed description of the analysis of bacterial community structures procedures based on the 16S rRNA gene sequencing is described in the Supplementary Information.

Metagenomic and metatranscriptomic sequencing and analyses

Five enrichments with wood chips, wheat bran, fish muscle, fish scales, and fish oil from the West Pacific Ocean were subjected to metagenomic and metatranscriptomic sequencing. Metagenomic and metatranscriptomic library preparation and sequencing using Illumina NovaSeq 6000 platform (paired-end 150-bp mode) were conducted at the Majorbio Company (Shanghai, China). Raw reads were quality-controlled by (i) clipping off primers and adapters and (ii) filtering out artifacts and low-quality reads by using the fastp v0.19.3 with default parameters [25]. Possible animal or plant reads from substrates (wood chips, wheat bran and fish tissue) were removed by mapping to the genomes of Pinus, Triticum aestivum, and Larimichthys crocea using Bowtie2 (identity >0.9) [29]. Filtered reads were individually assembled de novo by metaWRAP with the setting “-metaSPAdes” [30]. The binning process was performed by using the metaWRAP pipeline with three methods, metabat2, maxbin2, and concoct [30]. The Bin_refinement module in the metaWRAP pipeline was then implemented with the parameters “-c 50 and -x 10” [30]. Five sample’s binning results were combined and dereplicated using dRep v2.3.2 with the parameters “-comp 50 -con 10 -sa 0.99 -g --run_tax” [31]. After dereplication, a total of 110 metagenome-assembled genomes (MAGs) were obtained. The completeness and contamination of each MAG were estimated by CheckM v1.0.12 [32]. Taxonomic assessment was performed using GTDBTk v2.2.3 with the database release 207 [33]. Full-length 16S rRNA genes were reconstructed from metagenomic and metatranscriptomic reads using phyloFlash version 3.4 [34] against SILVA SSU 138 rRNA database.

To calculate the abundance of each gene, high-quality reads from metagenomics datasets were mapped to the assembled scaffolds using Bowtie2 with default parameters [29]. Then, fragments (PE reads) assigned to each gene were counted using the FeatureCounts program with the parameters “-p -F GTF -g ID -t CDS” [35]. For metatranscriptomic data, rRNA reads were removed by using RiboDetector v0.2.7 [36] and clean reads without rRNA were mapped to the metagenomic scaffold dataset and each MAGs using Bowtie2 with default parameters [29]. Afterwards, fragments (PE reads) assigned to each gene were counted using the FeatureCounts program with the parameters “-p -F GTF -g ID -t CDS -s 1 -M --fraction” [35]. The transcriptional expression of each gene was normalized with the transcripts per million (TPM) method in each metatranscriptomic dataset [37].

Functional annotations of protein-coding genes of the MAGs

Protein-coding genes of each MAG were predicted using Prodigal v2.6.3 [38]. We used KEGG [39], eggNOG [40], Pfam [41], MEROPS [42], and CAZy [43] as databases to functionally annotate the protein sequences. Online software KAAS v2.1 [39] (https://www.genome.jp/kegg/kaas/) was applied for homology searching against the KEGG database with the GHOSTZ program. Eggnog-mapper v1.0.3 software was used to annotate with the Diamond Blastp (v0.8.36.98) method [40]. To identify carbohydrate degradation-related enzymes, the online dbCAN2 meta-server (http://bcb.unl.edu/dbCAN2/) was used by HMMER against the CAZy database [43]. Pfam 31.0 [41] was used as the reference database for the annotation of peptidases, aminotransferases, and transporters of oligopeptides and amino acids by HMMER v3.1b2 (cutoffs: e value: 1e-10, best hits reserved). Additionally, the MEROPS database (release 12.1) [42] was used for peptidase annotation by Diamond Blastp v0.8.36.98 (cut-off: e value: 1e-10, best hits reserved) [44]. When a protein sequence was annotated to the same peptidase family with the above two databases, its annotation result was accepted and then used for subsequent analyses. The prediction of signal peptides was carried out by using the online SignalP-5.0 server (http://www.cbs.dtu.dk/services/SignalP/). Metabolic pathway analyses were determined by using the METABOLIC v2.2.3 [45]. METABOLIC relies on matches to the HMM databases (KEGG KOfam, Pfam, TIGRfam, and custom HMMs) using hmmsearch implemented within HMMER to infer the presence of specific metabolic pathways in microbial genomes [45]. Individual KEGG annotations are inferred in the context of KEGG modules for a better interpretation of metabolic pathways.

Phylogenetic analyses for 16S rRNA genes and the MAGs

Multiple sequence alignment of 16S rRNA genes was performed using the MAFFT program with default parameters, and a phylogenetic tree was reconstructed with the neighbor-joining method [46]. The reference sequences were retrieved from NCBI (last accessed on May 5, 2020). The phylogenomic tree was built by the up-to-date bacterial core gene (UBCG) method [47].

Results

OM-decomposing microbial assemblages obtained via long-term deep-sea in situ incubation

After incubation at the South China Sea (SCS), the Indian Ocean (IO), and the Pacific Ocean (PO) for 375, 117, and 348 days, respectively, all incubation tubes, even those with wood chips, exhibited obvious microbial growth indicated by turbidity, color variation, and microscope observation (Table S1 and Fig. S1). Many organic substrates remained in each centrifuge tube, even for liquid fish oil and vegetable oil, which could be seen as oily residue on the surface of sintered silicate balls. The in situ temperature and salinity of seawater at those sites were 2.39–2.40°C and 34.62–34.65‰, respectively (Table S1).

To profile the bacterial communities colonizing different types of OM, all 39 enrichments were subjected to community composition analysis via Illumina high-throughput sequencing and compared with the in situ seawater collected by control incubation chambers (CICs). The α-diversity (Shannon indexes) analysis showed that the bacterial diversity in the enrichments was significantly lower than that in the in situ seawater in the CICs (Fig. 2A, B; P < 0.001), which indicated that bacterial communities were significantly enriched in situ by the OM supplements. From the same site, no matter what the organic substrate was, the bacterial community structures were more similar, they therefore tended to cluster together on the PCoA (except the SCS site was divided into two clusters) (Fig. 2C and Fig. S2). Additionally, adonis analyses also suggested that geographic location (R² = 17%, P = 0.001) had a greater impact on microbial community structure than the type of OM (R² = 8%, P = 0.01), as well as other known factors such as temperature, water depth, incubation time and salinity (Table S2).

Bacterial composition and the prevalent members of deep-sea in situ enrichments

Across the deep-sea in situ OM enrichments, 2093, 967, and 673 ASVs were retrieved from the Indian Ocean, the SCS, the Pacific Ocean enrichment sites, respectively (Fig. S3A). The dominant bacteria were affiliated with the classes Bacteroidia (average abundance of 29.3%), Gammaproteobacteria (28.4%), Spirochaetia (10.5%), Clostridia (7.2%), Desulfobulbia (6.8%), Campylobacteria (6.2%), and Fusobacteriia (6.1%) (Fig. S3B). However, some other classes were detected almost exclusively in the in situ seawater without additional OMs in the CICs, such as Alphaprotebacteria (12%), Dehalococcoidia (11.4%), NB1-j (6.8%), Subgroup_21 (3.8%), Acidimicrobiia (3.2%), Methylomirabilia (3.1%), Phycisphaerae (2.6%), and Gemmatimonadetes (2.1%) (Fig. S3B). These results indicated that the newly input OMs had a significant impact on indigenous microorganisms in the deep sea.

Although these sites are far apart on the oceanic scale, the dominant members of these deep-sea enrichments are relatively similar at the family level (Fig. 3 and Fig. S4). These bacteria were affiliated with the families Marinifilaceae, Spirochaetaceae, Psychromonadaceae, Vibrionaceae, Desulfocapsaceae, Fusobacteriaceae, Arcobacteraceae, Moritellaceae, Flavobacteriaceae, Sulfurospirillaceae, Sulfurimonadaceae, and Sulfurovaceae (Fig. 3 and Fig. S4). In addition, they were dominant in these five representative metagenomic and metatranscriptomic data (five enrichments with wood chips, wheat bran, fish muscle, fish scales, and fish oil from the Pacific Ocean site) based on the small subunit ribosomal RNA (SSU rRNA) marker genes (Fig. S5). Previous studies have shown that most families are anaerobic or facultative anaerobic taxa, such as Marinifilaceae [48], Spirochaetaceae [49], Psychromonadaceae [50], Desulfocapsaceae [51], Arcobacteraceae [52], and Sulfurovaceae [53]. In phylogeny, some dominant taxa were diverged from those in other conventional habitats (Fig. S6). For example, Spirochaetaceae are typical intestinal anaerobic bacteria [49], while most their ASVs detected in this study were obviously separated from those known to be from the animal intestine in the phylogenetic tree (Fig. S6A). On the contrary, they formed several taxonomic branches with those from wood-fall ecosystems (Fig. S6A). The obvious ecotypic differentiation also occurred in Desulfobulbaceae (Fig. S6B), Arcobacteraceae (Fig. S6C), Sulfurovaceae (Fig. S6D), and Desulfobacteraceae (Fig. S6E). They may be sustained in sporadic whale- and wood-fall ecosystems as well as in more frequent sinking particles or benthic debris. As one of the four chemosynthetic deep-sea ecosystems [54], whale- and wood-fall ecosystems may function as the reservoir of these microbial seeds, connected and spread across the ocean via horizontal water current and sinking debris.

In the in situ seawater without additional OMs, uncultivated SAR202, JG30-KF-CM66, and S085, and Woeseiaceae were the dominant taxa, but were almost absent in our incubations (Fig. 3). Therefore, we speculate that they may not be involved in the OM mineralization process, or have no ability to compete with other fast-growing taxa, despite their widespread presence in the deep pelagic ocean [55–57].

Microbial co-occurrence networks within the bacterial assemblages of different OMs

To understand the co-occurrence relationships among bacteria in different OM enrichments, we constructed three co-occurrence networks for plant detritus-, animal tissue-, and fatty acid-enriched communities (Fig. S7). Overall, the relationships among bacteria within the communities were negative in 93% of cases in the animal tissue incubations (Fig. S7A and Table S3). However, more than 84% and 77% positive edges were observed in the plant detritus- (Fig. S7B and Table S3) and fatty acid-networks (Fig. S7C and Table S3), respectively, indicating bacterial groups in these two communities were mainly cooperative. Nutrients in animal tissue are replete and full of not only nitrogen, carbon, and phosphate sources but also other nutrients, such as vitamins. In contrast, many kinds of nutrients are quite limited in fatty acids and plant detritus. Therefore, bacteria in animal detritus enrichments showed competitive relationships and were more independent from each other. However, the fatty acid- and plant detritus-enriched consortia displayed more cooperative relationships to maintain nutrient balance and sustain the whole community. The above results suggested the co-occurrence relationships of the bacteria were closely related to the organic substrates.

Carbon, nitrogen, and sulfur metabolism

We obtained a total of 110 dereplicated MAGs and there were 87 and 69 MAGs with genomic completeness higher than 70% and 90%, respectively (Fig. 4 and Table S4). They belonged to 10 phyla and 46 families. Among, the MAGs B6 and B9 belonging to Marinifilaceae, MAG S1 belonging to Spirochaetaceae, and MAG D3 belonging to Desulfocapsaceae exhibited high relative abundance in the five metagenomic and metatranscriptomic data (Fig. 4).

To understand the metabolism features of each consortia, we analyzed the abundance and expression of the functional genes in the five representative substrate enrichments of wood chips, wheat bran, fish muscle, fish scales, and fish oil from the Pacific Ocean site. In terms of carbon metabolism, we analyzed the genes related to polysaccharide hydrolysis, polypeptide hydrolysis, and fatty acid oxidation. We found a large number of glycoside hydrolase (GH) genes and glycosyltransferase (GT) genes in the samples enriched with OMs, and they had high abundance in metagenomic data and actively transcribed (Fig. 5 and Table S5). In addition, we also found a large number of genes related to peptidase, oligopeptide and amino acid transporter, amino acid oxidation, and aminotransferase in these enrichments (Fig. 5 and Table S5). The metagenomic and metatranscriptomic data showed that the genes associated with polysaccharide hydrolysis and polypeptide hydrolysis had high abundance and high transcriptional activity (Fig. 5 and Table S5). However, obvious differences in the abundance and transcriptional expression of the corresponding genes among different enrichments (Fig. S8A, B). The polysaccharide hydrolyzation-related genes were significantly expressed in plant detrital enrichments, whereas proteolysis-related genes were significantly expressed in animal tissue enrichments (Fig. S8B).

Further analysis on the MAG level showed that the genes related to polysaccharide and/or protein hydrolysis were from Marinifilaceae, Spirochaetaceae, F082 (belonging to Bacteroidales), unclassified groups belonging to Peptostreptococcales, Planococcaceae, Desulfocapsaceae, Desulfocapsaceae, Sulfurovaceae, Arcobacteraceae, and Psychromonadaceae (Fig. 6A, B and Fig. S9A, B). Among, Marinifilaceae and Spirochaetaceae were the major polysaccharide and protein hydrolyzers, which together contributed more than 90% of GH gene abundance (metagenome) and 70% of GH gene transcripts (metatranscriptome) (Fig. 6A and Fig. S9A), and more than 70% of peptidase genes abundance (metagenome) and 60% of peptidase transcripts (metatranscriptome) (Fig. 6B and Fig. S9B).These results suggested that Marinifilaceae and Spirochaetaceae mainly played the role of hydrolyzing polysaccharides and proteins and were the direct depolymerizers of macromolecular OMs (Fig. 7).

Similar to those involved in carbon metabolism, the genes associated with sulfur metabolism were also detected and actively transcribed in these enrichments, including oxidation genes of hydrogen sulfide (sqr), thiosulfate (Sox complex system), sulfite (soeABC and suoX), and various dissimilatory reductions of sulfite, tetrathionate, thiosulfate, and dimethyl sulfoxide (Fig. 5 and Table S6). In particular, the genes involved in dissimilatory sulfite reduction in fish muscle and fish oil enrichments had higher transcriptional levels than those in other enrichments, and the genes encoding sqr showed high transcriptional activity in enrichments with wood chips and fish scales (Fig. 5 and Table S6). Desulfocapsaceae, Desulfobacteraceae, F082, Desulfuromonadaceae, and UBA1556 harbored the capacity of dissimilatory sulfate reduction (sat, aprAB, and dprAB) (Fig. S10 and Table S7), with Desulfocapsaceae playing the most significant role in this process (Fig. 6C, Fig. S9C, and Fig. 7). Comparatively, there were more taxa involved in the process of sulfur oxidation (SOX system, phsA, and sqr), including Desulfocapsaceae, Desulfobacteraceae, Sulfurovaceae, Sulfurimonadaceae, o-Peptostreptococcales, and Arcobacteraceae (Fig. S10 and Table S7), among which Arcobacteraceae, Desulfocapsaceae, and Sulfurovaceae were the most important contributors to this process (Fig. 6C, Fig. S9C, and Fig. 7).

With respect to nitrogen metabolism, we detected highly transcriptionally expressed genes involved in both nitrogen fixation (nifDKH) and dentrification, including dissimilatory nitrate reduction (napAB), nitrite reduction (nirA and narB), nitric oxide reduction (norBC), and nitrous oxide reduction (nosZ) in these enrichments (Fig. 5 and Table S6). However, the abundance and transcriptional expression of nitrogen-fixing genes were significantly enriched in plant detritus and fish oil enrichments compared with protein-rich animal tissue enrichments (Fig. S8A, B). Spirochaetaceae, Desulfocapsaceae, unclassified groups belonging to Peptostreptococcales, and Desulfocapsaceae harbored the capacity of nitrogen fixation (Fig. S11 and Table S8), with Spirochaetaceae and Desulfocapsaceae playing the dominant role in this process (Fig. 6D and Fig. S9D). Arcobacteraceae was the most dominant group in the process of denitrification, and Desulfocapsaceae was the most dominant group in the process of nitrite reduction to ammonia (Fig. 6D and Fig. S9D).

Discussion

OM decomposition and mineralization in global oceans is a key process in global carbon cycling [21]. Characterizing the microorganisms involved in this process in the deep sea is important for understanding how element cycling works in this vast ecosystem. In this report, in situ incubations were conducted to observe bacteria involved in the transformation and mineralization of OMs by mimicking the sinking “compact” biomasses in the deep sea of open areas (the Pacific Ocean and the Indian Ocean) and a marginal sea (the SCS). The results showed that the mimicking organic pulses enriched unique bacterial assemblages according to substrate types and geographic locations, which are different from those usually found in deep-sea water or deep-sea sediment [58]. However, the bacterial community structure was influenced more by geographic location than by the type of organic substrate (Fig. 2B, Fig. S2, and Table S2). We hypothesized that some dominant ASVs had multiple functions, including the hydrolysis of polysaccharides and proteins or peptides and the oxidation of fatty acids, and were therefore simultaneously enriched in different types of OMs. It has been reported that some species of Vibrionaceae [59], Psychromonadaceae [50], Marinifilaceae [60], and Spirochaetaceae [61, 62] all have these abilities simultaneously. In addition, regardless of the organic substrate type, the local environment will become anaerobic after settling on the seabed, like whale falls [10] and wood falls [8], therefore the same ASVs belonging to sulfate-reducing bacteria (SRBs) and sulfur-oxidizing bacteria (SOBs) will be recruited in the same site. However, in terms of geographical location, the three selected sea areas (the marginal sea, the flat-topped seamount in the West Pacific Ocean, and the deep-sea basin beside the Southwest Indian Ridge) have very obvious differences, thus the dominant families enriched with OMs in the in situ environment may have different variants in different sea areas due to geographical isolation [63].

Months to a year round in situ incubation in deep sea revealed bacteria belonging to the families Marinifilaceae, Spirochaetaceae, Psychromonadaceae, Vibrionaceae, and Moritellaceae potentially contributed greatly to the decomposing the various organic substrates (Fig. 3 and Figs. S4, S5), which are rare species in the surrounding seawater (Fig. 3). They are also not the predominant members among bathypelagic bacteria in the global ocean water column and deep-sea sediments [55, 58]. However, as a rare species of famine waiting for the feast of sinking bulky POM on the barren seabed, they can propagate quickly on newly input POM, as suggested previously by Jorgensen [13]. Routinely, Gammaproteobacteria are the most prevalent among deep-sea pelagic prokaryotes, including the genera Alteromonas, Halomonas, Psychrobacter, and so on [12, 20, 55, 64]. The phyla Chloroflexi, Planctomycetes, and Actinobacteria are also usually predominant in deep-sea sediments [58, 65]. However, they are not the components thriving on the newly input POM in this study.

Intriguingly, our observation was partially consistent with those associated with sinking particles sampled by sediment traps in the oceanic interior. In abyssal trapped sinking particles, Arcobacteraceae, Moritellacea, and Vibrionaceae are the most dominant families [12, 66]. Congruently, they were also among the top 5 taxa in part of our enrichments (Fig. 3). Parallelly in coastal sediment, our in situ incubations also revealed Desulfobulbaceae, Desulfobacteraceae, Arcobacteraceae, Vibrionaceae, Spirochaetaceae, and Fusobacteriaeae dominated in the OM-enriched bacterial communities [67]. However, Marinilifaceae, Moritellacea, and Psychromonadaceae were barely detected in the coastal enrichments [67]. In wood and whale falls, the bacteria of Marinilifaceae and Spirochaetaceae also occurred in bacterial assemblages with high abundance [8]. In contrast, the bacteria of Moritellaceae and Psychromonadaceae were rarely enriched in whale and wood-falls [8, 10, 68].

Metagenomic and metatranscriptomic analyses showed that the metabolism of proteins and polysaccharides, oxidation of fatty acids, and fermentation were distinctly enriched in the enrichment assemblages (Figs. 5, 6, Tables S5, S6), highlighting the key bacterial taxa playing important roles in the POM mineralization process. In particular, Marinifilaceae and Spirochaetaceae act as direct decomposers of macromolecular OM derived from animal and plant detritus (Fig. 7). The Marinifilaceae family contains several facultative anaerobic species capable of degrading various macromolecule polysaccharides (e.g., starch, cellulose, and alginate) and proteins (e.g., casein and gelatin), while simultaneously producing various small molecular organic acids via fermentation [48, 60, 69–71]. In case of Spirochaetaceae, species like Pleomorphochaeta multiformis [62] and Spirochaeta perfilievii [61] also could utilize mono-, di- and poly-saccharides (xylan, trehalose, and pectin) and protein to generate formate, acetate, ethanol, pyruvate, and hydrogen via fermentation. Therefore, certain Spirochaetaceae species in the termite gut play an important role in the digestion of breakdown products from cellulose and hemicellulose [49]. In this study, more than two thirds of the transcripts assigned to GH genes and peptidase genes in consortia were generated from Marinifilaceae and Spirochaetaceae (Fig. 6A, B), indicating they played important roles in the degradation of polysaccharides and proteins in plant detritus and animal tissue (Fig. 7).

Previous studies also reported that Vibrionaceae and Psychromonadaceae could degrade various macromolecular substrates, e.g., starch, xylan, alginate, mannan, gelatin, casein, chitin, and lecithin [50, 59]. Recently, we confirmed that the Vibrio isolates dominating in plant detritus enrichment can oxidize lignin and hydrolyze xylan under both aerobic and anaerobic conditions [72]. Therefore, they also contributed to decomposing plant detritus and animal tissues (Fig. 7).

In addition the above taxa, the key bacterial taxa involved in sulfur and nitrogen cycling also dominated the assemblages, and they may be coupled with OM decomposition. Arcobacteraceae, Sulfurimonadaceae, Sulfurospirillaceae, and Sulfurovaceae are typical marine chemoautotrophic SOBs [73–75]. In this study, they were involved in the oxidation of reducing sulfur compounds generated by SRBs like Desulfocapsaceae, Desulfobulbaceae, and Desulfobacteraceae (Fig. 6C and Fig. 7). These SRBs may use the most common microbial fermentation products, e.g., acetate, propionate, butyrate, lactate, and hydrogen as an energy source, coupled with sulfate reduction [76, 77]. Interestingly, Desulfocapsaceae and Desulfobacteraceae also performed the oxidation process of sulfur compounds, in addition to the reduction of sulfate (Fig. 6C and Fig. 7). The predominance of SOBs and SRBs may highlight the significance of chemolithotrophs for energy conservation (producing new organic carbon) [78].

Interestingly, assemblages on different OM types hold different relationships (Fig. S7). We suspect that this phenomenon may be related to available nutrients, such as organic nitrogen sources. Organic nitrogen sources are abundant in animal tissues, but relatively scarce or absent in fatty acid substrates and plant tissues. Consistently, nitrogen fixation-related genes were detected in the enrichments, especially with plant detritus or fatty acids (Fig. S8). This result implied that nitrogen fixation could be accompanied by OM remineralization in nitrogen-nutrient-depleted substrates. Desulfobacterales and Desulfuromonadales are the most abundant diazotrophs in deep-sea sediments [79]. In this study, the dominant Desulfocapsaceae (the sister family of Desulfobacterales and Desulfuromonadales) possessed nitrogen fixation-related genes and showed 3–70% transcriptional activity in enriched assemblages (Fig. 6C). In addition, this study also reported for the first time that members of the Spirochaetaceae and o-Peptostreptococcales harbored the nitrogen fixation potential (Fig. 6C and Fig. S11). Across the ocean, nitrogen fixation accompanies chemoheterotrophic processes, as we observed, which are stimulated by OMs inside organic aggregates lacking nitrogen sources in the deep water column.

Summary and conclusion

Through long-term in situ incubation, we observed a cosmopolitan anaerobic bacterial assemblage that plays a significant role in the transformation and mineralization of OM in biomass in the deep sea of the Pacific Ocean, the Indian Ocean, and the SCS. As depicted in Fig. 7, key taxa belonging to the families Marinifilaceae, Spirochaetaceae, Psychromonadaceae, Vibrionaceae, and Moritellaceae, are crucial in decomposing natural polymers in plant detritus or animal tissue. This decomposition process leads to the formation of low-molecular weight compounds, including fermentation products, and sequentially recruits SRBs and SOBs, ultimately establishing unique microbial assemblages. The bacteria in these assemblages form cooperative relationships or not based on the nutrient abundance of the polymers, which may be facilitated by the nitrogen fixation process conducted by diazotrophes from Desulfocapsaceae, Spirochaetaceae, o-Peptostreptococcales, and Marinifilaceae. Consequently, these uniquely assembled bacteria participate in the mineralization of organic carbon while coupling nitrogen and sulfur cycling. They create niches for different microorganisms to thrive across oxygen gradients at the center of their habitat (Fig. 7). These bacteria distinguish themselves from their counterparts dominating the water column and sediment in both coastal and pelagic areas. Further investigations are needed to fully understand their roles in biogeochemical cycling in the global oceans.

Acknowledgements

The authors give thanks to all the scientists, engineers, and technicians and all crew members of several oceanic cruises of COMRA. We thank Prof. Marcus Elvert (Organic Geochemistry Group, MARUM—Center for Marine Environmental Sciences, University of Bremen, Bremen, Germany) for comments on an earlier version of the manuscript.

Author contributions

Jianyang Li (Formal analysis, Investigation, Methodology, Visualization, Writing—original draft), Chunming Dong (Investigation, Methodology, Writing—original draft), Shizheng Xiang (Methodology), Huiyang Wei (Formal analysis, Visualization), Qiliang Lai (Investigation), Guangshan Wei (Writing—review & editing), Linfeng Gong (Data curation), Zhaobin Huang (Investigation), Donghui Zhou (Investigation), Guangyi Wang (Writing—review & editing), Zongze Shao (Conceptualization, Funding acquisition, Investigation, Methodology, Writing—review & editing).

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding

This work was financially supported by the National Natural Science Foundation of China (No. 42030412), the Scientific Research Foundation of Third Institute of Oceanography, MNR (No. 2024003), Natural Science Foundation of Xiamen, China (No. 3502Z202472042), the Natural Science Fund of Fujian Province of China (No. 2021J02015), the Scientific Research Foundation of Third Institute of Oceanography, MNR (No. 2023021), the National Key Research and Development Program of China (No. 2023YFC2811402), the China Ocean Mineral Resources R&D Association (COMRA) program (No. DY135-B2-01), the High-Tech Research and Development Program of China (No. 2012AA092102), and the Scientific Research Foundation of Third Institute of Oceanography, MNR (No. 2019021).

Data availability

The raw dataset is available in NODE (https://www.biosino.org/node/) under project ID OEP001506.

References

Author notes