Microbial communities vary throughout the Mariana Trench water column
To determine the vertical distribution of microbial populations in the Challenger Deep, shotgun metagenome sequencing and 16S rRNA gene amplicon analyses were performed on free-living (0.22–3 μm) (FL) and particle-associated (> 3 μm) (PA) communities collected from surface waters and from depths of 4000 m, 9600 m, 10,400 m, and 10,500 m (Additional file 1: Table S1 and Figure S1). Also, environmental analysis was performed on samples collected from 4 to 8320 m, which demonstrated that hadal and abyssal waters shared many comparable physiochemical factors, including temperature, salinity, oxygen, and nutrients (Additional file 1: Table S2).
Quantitative PCR assays showed that the 16S rRNA gene copy numbers decreased significantly from 0 (7.73 × 107 copies/L) to 4000 m (1.21 × 107 copies/L), but then increased with depth to 7.43 × 107, 7.58 × 107, and 7.82 × 107 copies/L at 9600, 10,400, and 10,500 m, respectively (Additional file 1: Figure S2). Similarly, Nunoura et al. showed that the abundance of 16S rRNA gene decreased along with depth before 6000 m but then increased slightly from 6000 m to 10,257 m except for 9000 m [1]. In this study, the increase of 16S rRNA gene copy number in trench bottom water compared to 4000 m may be due to increased Gammaproteobacteria populations (see below), which harbor ~5.7 copies of 16S rRNA gene on average, far higher than other Proteobacteria [9], suggesting that 16S rRNA gene copy number is not always a good proxy of cell number.
Taxonomic profiling of metagenomic data based on the NCBI-nr database revealed that bacteria accounted for the majority of the population at all depths in both FL and PA samples, compared to archaea and eukaryotes (Additional file 1: Figure S3a). As expected, Proteobacteria, particularly Alphaproteobacteria and Gammaproteobacteria, were the predominant organisms at all depths, whereas Cyanobacteria were relatively abundant only in surface water, accounting for 13.9% and 7.3% of the FL and PA samples, respectively (Additional file 1: Figure S3b and 3c). The relative abundance of Gammaproteobacteria increased with depth from 21.3% at 0 m to 31.3% at 4000 m, 38.4% at 9600 m, and 49.2% at both 10,400 m and 10,500 m (Additional file 1: Table S3). Similar to Nunoura et al. [1], the relative abundance of Oceanospirillales increased significantly (P < 0.01) in the two near bottom water (NBW; ~ 300–400 m from the seafloor) samples, comprising 61.4% of the gammaproteobacterial genes (5.6-fold increase compared to 9600 m) (Fig. 1a). This dominance of Oceanospirillales in NBW was confirmed by analyses of 16S rRNA gene recovered from both the metagenomic and amplicon sequencing (Additional file 1: Figure S4), suggesting that a bacterial niche boundary exits at that depth. Several bacterial lineages such as Pelagibacterales differed in relative abundance among different methods used (Additional file 1: Figure S4). Such differences may have resulted from biases in PCR amplification, variation in gene copy number, and genome size [10].
Next, we analyzed the genus-level composition to identify the genera that are significantly enriched in NBW samples. According to the metagenomic data annotated against the NCBI-nr database, seven genera were significantly more abundant in NBW than at other depths (P < 0.05) (Fig. 1b). All belonged to the order Oceanospirillales and included Oleibacter (19.5 ± 6.5%), Thalassolituus (2.5 ± 0.7%), Alcanivorax (1.0 ± 0.4%), Oleiphilus (0.9 ± 0.7%), Oceanobacter (0.6 ± 0.2%), Hahella (0.5 ± 0.4%), and Bermanella (0.3 ± 0.1%). 16S rRNA gene amplicon analysis also confirmed the predominance of Oleibacter in NBW (37.8 ± 8.2%). This is the first environment where Oleibacter has been reported as the most abundant genus.
Oceanospirillales species encode genes involved in alkane degradation
To explore the genetic potential in Oceanospirillales that dominate the NBW environment, we compared the metagenomic data against the COG and KEGG databases. The NBW harbored a significantly higher abundance of genes encoding cell motility proteins compared to other depths (P < 0.05) (Fig. 2 and Additional file 1: Figure S5). Predominant among these were genes encoding methyl-accepting chemotaxis proteins (MCP) and type IV pilus assembly genes that function in cell adhesion and twitching motility [11] (Additional file 1: Figure S6), which were mainly attributed to Oleibacter (Additional file 1: Figure S7), a genus which includes species known to degrade hydrocarbons [12, 13]. MCPs have been found to play an important role in sensing various substrates including alkanes [14] and thus may facilitate hydrocarbon degradation by Oleibacter species. MCP and pilus genes were highly transcribed in microbial populations following the Deepwater Horizon oil spill [15, 16], further supporting a key role for these genes in microbial hydrocarbon degradation. NBW harbored a lower abundance of genes related to carbohydrate and protein metabolism (COG E and G; Fig. 2), as well as genes involved in the cycling of carbon and sulfur (Additional file 1: Figure S8), suggesting the microbial population in NBW might utilize alternative carbon sources.
To predict hydrocarbon degradation pathways in the abundant NBW Oceanospirillales bacteria described above, we constructed 43 high-quality genomic bins, hereafter referred to as metagenome assembled genomes (MAGs), from the combined contigs of the four NBW samples with a contaminant threshold of 10% and a completeness threshold of 80%. These include 33 Proteobacteria and three Actinobacteria MAGs; the former contain 19 Alpha-, nine Gamma-, and five Deltaproteobacteria MAGs (Additional file 1: Table S4). Five of the Gammaproteobacteria MAGs, termed MAG 01 (completeness 90.8%), 78 (completeness 87.7%), 5.60 (completeness 83.2%), 12 (completeness 93.2%), and 156 (completeness 97.8%), are related to Oceanospirillales. The first three MAGs (01, 78, and 5.60) are phylogenetically similar to Oleibacter, while the latter two are similar to Alcanivorax and Oleiphilus, respectively. The three Oleibacter MAGs (01, 78, and 5.60) showed an average nucleotide identity of 74.7% to each other, and 98.6%, 77.2%, and 73.6% to Oleibacter marinus DSM 24913, respectively. Strain DSM 24913 is a hydrocarbonoclastic bacterium isolated from a crude oil enrichment of Indonesian seawater [12] and the only cultured species available in this genus. Our analyses suggest that MAG 01, 78, and 5.60 represent different Oleibacter species, and MAG 01 is affiliated to O. marinus. This was further verified by the high 16S rRNA gene sequence similarity (99.0%) between MAG 01 and O. marinus DSM 24913. Mapping raw reads to the MAGs showed that MAG 01 was the dominant Oleibacter clade in NBW; its relative abundance was the highest among all MAGs and accounted for 11.6 ± 2.4% of all NBW reads (Additional file 1: Figure S9). These five Oceanospirillales MAGs possess predicted pathways for both medium-chain alkane 1-monooxygenase (AlkB) and long-chain alkane monooxygenase (AlmA)-mediated alkane degradation, for cyclohexane degradation, MCP, and most other chemotactic genes (Fig. 3). Thus, metagenomic predictions suggest that Challenger Deep NBW Oceanospirillales bacteria can metabolize medium-chain, long-chain, and some cyclic hydrocarbons and that such molecules may be important sources of carbon and/or energy in such environments.
To the best of our knowledge, the proportion of Oceanospirillales-derived alkane degraders in NBW (~ 20% in the metagenome and ~ 38% in the 16S rRNA gene amplicon; Additional file 1: Figure S4a and c) is higher than in any other natural environments observed on Earth. However, the predominance of Oceanospirillales, along with MCP, alkB, and type IV pilus assembly genes in NBW, resembles the situation (dominant microbial communities and functional genes) in the Deepwater Horizon spill (DWH) [15,16,17,18,19], suggesting that the microorganisms in these two distinct environments most likely share similar hydrocarbon-degrading pathways. However, the predominant DWH Oceanospirillales showed high sequence similarity to Candidatus Bermanella macondoprimitus and/or Candidatus Oceanospirillales desum [20, 21]. Although O. marinus was detected in n-hexadecane enrichments using water collected from oil plume at 1000 m after the DWH, it only represented 3% of the total sequences [22].
Abundance and expression of alkane degradation genes increase in near bottom waters
The data above suggest that microbial populations in NBW have the potential to degrade hydrocarbons. To test this, we first investigated the abundance of putative hydrocarbon-degrading genes at different water depths (Fig. 4). Interestingly, the relative abundance of genes encoding AlkB, AlmA (Fig. 4a), and several enzymes for sensing and transporting alkanes (Additional file 1: Figure S10) increased significantly in NBW compared to other water depths. The absolute abundance of alkB and almA and the proportion of microbial cells possessing them exhibited similar trends (Additional file 1: Figure S11). These numbers displayed positive relationships with the relative abundance of Oceanospirillales (r > 0.85, P < 0.01), notably Oleibacter (Fig. 4b and Additional file 1: Table S5). The increase of alkB and almA in NBW was similar to that of MCP (Additional file 1: Figure S6). The abundance of the other known aliphatic hydrocarbon degradation genes was not statistically different among samples from different depths (Fig. 4a). In addition, many aromatic hydrocarbon degradation genes (e.g., for catechol, protocatechuate and gentisate metabolism) were lower in NBW than in other water depths (Additional file 1: Figure S12). Overall, this suggests that alkanes are the main hydrocarbon source utilized by Oceanospirillales species.
To determine whether increases in the transcription of alkB and almA corelate with their gene abundance, we performed RT-qPCR on 0, 4000, 9600, 10,400, and 10,500 m samples. AlkB and almA genes were highly transcribed in NBW compared to 0, 4000, and 9600 m (up to 42, 61, 26 and 86, 109, 34-fold increased, respectively) (Fig. 4c), and 81.5–100% of transcripts were from Oleibacter. This further supports the hypothesis that Oceanospirillales species are degrading both medium- and long-chain alkanes in this environment.
Near bottom water isolates degrade alkanes under physiologically relevant conditions
To complement genetic predictions that NBW bacteria degrade hydrocarbons, we isolated 38 axenic alkane-degrading strains from NBW on alkane-dependent medium. Most isolates were Oceanospirillales and mainly affiliated with Alcanivorax, including A. jadensis ZYF844, A. venustensis ZYF848, and A. dieselolei ZYF854 (Additional file 1: Figure S13 and Table S6). The three Alcanivorax isolates showed 99.6–100% 16S rRNA gene similarity to sequences derived from high-throughput amplicon sequencing, indicating that they represent indigenous NBW species. These three isolates degraded a wide range (C18–C26) of n-alkanes at 4 °C and 0.1 MPa (atmospheric pressure) (Fig. 5a). Notably, they could utilize n-eicosane (C20) as the sole carbon source at physiologically relevant temperature (4 °C) and pressure (60 MPa) with the strain ZYF844 displaying the highest degradation rate (Fig. 5b).
O. marinus DSM 24913 was tested for hydrocarbon degradation activity, since no Oleibacter species was isolated from NBW (despite repeated attempts with alkane-dependent media and Marine Broth under 4/16 °C and 0.1/60 MPa) and O. marinus is the only cultured species available in this genus. Oleibacter MAG 01, the most abundant clade in NBW, demonstrated 99.0% 16S rRNA gene sequence identity to O. marinus DSM 24913. Moreover, MAG 01 contained a putative alkB and two almA genes, which encoded enzymes with > 99.5% amino acid identity to those of O. marinus DSM 24913 [12], and up to 34.0% and 49.7% identity to functionally characterized AlkB and AlmA from A. dieselolei, respectively [14, 23]. MAG 01 also harbored 11 putative MCP genes (O. marinus DSM 24913 possessed 14 MCP genes), with up to 43.9% amino acid identity to the functionally confirmed A. dieselolei MCP involved in sensing alkanes [14].
Unlike NBW Alcanivorax species, O. marinus DSM 24913 could not degrade hydrocarbons at 4 °C and 60 MPa, but it displayed a high rate of hydrocarbon degradation in cultures fed on n-eicosane at 16 °C and atmospheric pressure (Fig. 5b). This implies that NBW Oleibacter species, such as MAG 01, have evolved to withstand high pressure and low temperature typical of hadal conditions. Thus, we propose that the NBW Oleibacter species may represent novel and hadal-specific ecotypes despite their high 16S rRNA gene (i.e., 99.0% identity) and genome sequence (98.6% average nucleotide identity) identity to O. marinus DSM 24913. A similar phenomenon was shown for Alteromonas macleodii species, which are phylogenetically divided into “surface” or “deep” ecotypes (recently a new species A. mediterranea was separated [24]), according to seawater depth, despite sharing > 99% 16S rRNA gene similarity [25]. In support of bioinformatic predictions, n-alkanes can be utilized as a carbon and/or energy source by all tested NBW Oceanospirillales isolates under high pressure and low temperature, and O. marinus, representative of the dominant Oleibacter genus in NBW under atmospheric pressure at 16 °C.
Potential alkane accumulation at deeper depths and production in sediments in the Mariana Trench
To study the hydrocarbon sources that may support this population, we investigated alkane concentrations in sinking particle (2000, 4000, and 6000 m; Fig. 6a and Additional file 1: Table S7) and surface sediment samples (Fig. 6b). Total concentrations of n-alkanes (in the range of n-C15 to n-C32) in sinking particles increased from 2000 (7.2 μg/gdw) to 4000 (39.0 μg/gdw), but then decreased slightly at 6000 m (24.2 μg/gdw). These values are comparable or slightly higher than hydrocarbon concentrations (8.3-37 μg/g; dominated by n-alkanes) reported for sinking particles from abyssal waters of the northeastern Pacific Ocean [26]. The unique funnel-shaped geomorphology of the trenches (Additional file 1: Figure S1) promotes the accumulation of particulate organic matter sinking from the surface [27], including possible terrestrial organic matter sources [27], facilitated by lateral transportation from trench rims and slopes that is often triggered by earthquakes [28]. Considering this, we hypothesize that n-alkane hydrocarbons maintain high concentrations, or even increase, at depths below 6000 m. Unfortunately, we were unable to analyze alkane concentrations in sinking particle samples at deeper depths, so further work is required to test this hypothesis.
To determine the sources of n-alkanes in hadal surface sediments at depths of 10,908, 10,909, and 10,911 m (Fig. 6b and Additional file 1: Table S7), we investigated their distribution and concentrations (total n-alkane concentration averaged 2.3 μg/gdw). Interestingly, higher concentrations of n-C18 and n-C20 (and n-C16 in one sample) compared to n-C17/19/21 alkanes (Fig. 6b) were observed. The predominance of even-chain alkanes has been reported previously by other studies, e.g., in sediments from the Black and Mediterranean Seas, Arabian Gulf, and Cariaco Trench [29], and from the Mariana Trench sediments at depths of 4900–7068 m [30]. These were markedly different from those profiles seen in the sinking particle samples, suggesting that even-chain alkanes in hadal surface sediments have a very different source and are likely synthesized in situ and/or released from subsurface sediments. To provide further evidence for the origin of these compounds, we investigated the carbon and hydrogen isotope compositions of n-C16 and n-C18 alkanes (Fig. 7). The δ13C values of n-C16 and n-C18 alkanes ranged from − 30.2 to − 28.7‰, similar to those reported by Guan et al. [30], which are not very source specific as these values fall within the range of isotopic compositions of hydrocarbons derived both from petroleum as well as from autotrophic and heterotrophic organisms [31]. However, more crucially, the δ2H values of n-C16 and n-C18 alkanes are much more informative. They varied from − 79 to − 93‰, which are higher than those typically assigned to petroleum sources (− 90 to − 150‰ [32]) and fall within the range of lipid/water 2H/1H fractionations reported for heterotrophic organisms (− 150‰ and higher [33]) assuming the δ2H of water is ca. 0‰. These δ2H data suggest that n-alkanes in the hadal sediments most likely derive from a distinct biological source (e.g., from heterotrophic communities [33]) that results in a smaller hydrogen isotope fractionation. This source differs from that produced by known hydrocarbon-producing marine organisms, predominantly photoautotrophs, which synthesize only odd-C-numbered hydrocarbons [34,35,36], suggesting that the microbial producers might utilize a different, and still poorly understood biosynthetic pathway.
Given the NBW Alcanivorax isolates were more efficient at degrading C18–C20 compared to > C22 alkanes (Fig. 5a), it is possible that these bacteria have evolved to take advantage of these specific hydrocarbons. Thus, this unknown biological source of n-C16, n-C18, and n-C20 hydrocarbons in the surface sediments at the bottom of the trench may contribute, with sinking particle hydrocarbons, to support the observed NBW hydrocarbon-degrading microbial population. Further work, involving turnover rate measurements, is likely required to establish this hypothesis. A similar hydrocarbon cycle is likely present in surface waters, where hydrocarbon-degrading bacteria are supported by alkanes synthesized by cyanobacteria and algae [34]. Currently, it is uncertain if these hydrocarbons are produced by microbes only found in the sediments or whether they are widespread in NBW, since we were unable to study their prevalence in NBW samples and the enzymes synthesizing even-chain alkanes have not been identified.