Seawater samples were collected from three stations (Sts. 1, 6, and 9) in the western North Pacific (Supplementary Figure S1) between November 3 and 27, 2021, at depths ranging from 200 to 1,500 m (Table 1). Seawater was collected using a Niskin-X sampler with a Teflon-coated inner wall and Kevlar line installed on the ship. Samples were filtered through an AcroPack in-line filter (0.22 μm pore size) to remove particulates. Filtered seawater was stored in 0.1 L and 0.5 L PFA bottles for THg and MeHg analysis, respectively. For THg analysis, 1.0 mL of concentrated ultrapure HCl (Kanto Chemical Co., Tokyo, Japan) and BrCl (Guaranteed Reagent, Kanto Chemical Co.) were added to achieve a final concentration of approximately 0.002 mol L⁻¹ (Marumoto and Imai, 2015). For MeHg analysis, 2.0 mL of 10 M reagent-grade H₂SO₄ (Kanto Chemical Co.) was added (Marumoto et al., 2018). Treated samples were stored at 4 °C in the dark.

Seawater samples for counting prokaryotic cells were preserved with 0.5% glutaraldehyde in 15 mL polypropylene tubes and stored at −80 °C. For metagenomic analysis, 5 L seawater samples were filtered through a 0.22-μm-pore-size Sterivex cartridge filter (Millipore, Burlington, MA, United States) to capture prokaryotic cells. The filters were stored at −80 °C.

Macronutrient samples were collected in acrylic tubes and stored at −30 °C until analysis. Nitrate plus nitrite, phosphate, and silicic acid concentrations were measured using a QuAAtro39 segmented continuous flow analyzer (Bran + Luebbe, Norderstedt, Germany).

Seawater physicochemical parameters, including temperature, salinity, chlorophyll a, and dissolved oxygen concentrations, were recorded using a conductivity, temperature, and depth (CTD) system (SBE9plus CTD system, Sea-Bird Electronics). The depth of the subsurface chlorophyll maximum layer was determined from the CTD profiles.

The dTHg and dMeHg analyses were performed as previously described (Marumoto et al., 2018). dTHg was quantified using EPA Method 1,631 (U.S. Environmental Protection Agency, 2002) via cold vapor atomic fluorescence spectrometry with gold amalgamation (RA-FG+; Nippon Instruments Corporation) after Hg(0) generation with 1 mL of 20% (w/v) SnCl₂ as a reducing agent. Analytical precision was validated through multiple measurements of BCR579 standard reference material (certified range: 1900 ± 500 pg. L⁻¹; 9.5 ± 2.5 pM), yielding values of 1850 ± 60 pg. L⁻¹ (9.3 ± 0.3 pM, n = 6), consistently within the certified range. The method detection limit, calculated from ultrapure water blanks and defined as three times the standard deviation of the blanks, was 7.03 pg. L⁻¹ (0.035 pM, n = 11). The mean blank dTHg concentration was 16 ± 2.3 pg. L⁻¹ (0.08 ± 0.01 pM, n = 11).

The dMeHg determination was performed using solvent extraction with dithizone–toluene and Na₂S solutions (Ministry of the Environment, Japan, Mercury Analysis Manual, 2004). dMeHg concentrations in Na₂S solutions were measured via ethylation with NaB(C₂H₅)₄, preconcentration onto a Tenax trap, thermal desorption, and gas chromatography with atomic fluorescence detection as previously described (Logar et al., 2002). The method detection limit for dMeHg, calculated as for dTHg, was 1.4 pg. L⁻¹ (0.007 pM), with ultrapure water blanks yielding 0.85 ± 0.46 pg. L⁻¹ (0.004 ± 0.002 pM, n = 3). MeHg recovery of dithizone–toluene was 99% ± 3% (n = 7), validated by spiking experiments using a 1 ng mL⁻¹ alkaline-dissolved solution obtained from DORM-2 (certified range: 4.47 ± 0.32 mg kg⁻¹ dry weight), an international reference material for MeHg in dogfish. One DORM-2 solution was analyzed for every seven samples. At St. 1 (1,500 m), St. 6 (967 m), and St. 9 (1,175 m), duplicate samples were collected, and the analytical precision of these duplicates was 9.4, 10.4, and 4.7%, respectively.

The prokaryotic cells were stained with 4′,6-diamidino-2-phenylindole and filtered with a 0.22 μm pore-size polycarbonate filter (GTTP00250, Millipore). Cells were counted under an epifluorescence microscope (10 fields per sample).

Environmental DNA was extracted from the Sterivex cartridge filter using a PowerWater DNA Isolation Kit (Qiagen, Hilden, Germany). Before extraction, the Sterivex cartridge was opened, and the filter was cut into 16 pieces, which were transferred to PowerWater bead tubes for DNA extraction following the manufacturer’s protocol. DNA solutions were treated with RNase A (final concentration: 0.1 μg μL⁻¹; Promega, Madison, WI, United States). The extracted DNA was used for 16S rRNA deep sequencing and shotgun metagenomic sequencing and was stored at −80 °C. The quality of metagenomic DNA was assessed via 1% agarose gel electrophoresis.

Bacterial and archaeal 16S rRNA (V4 region) gene fragments were amplified using the following primers with adaptor sequences from Illumina (San Diego, CA, USA): 515F, ACACTCTTTCCCTACACGACGCTCTTCCGATCT-GTGCCAGCMGCCGCGGTAA; and 806RB, GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT-GGACTACNVGGGTWTCTAAT (Caporaso et al., 2011; Apprill et al., 2015). The PCR program included an initial denaturation step for 5 min at 94 °C, followed by 25 cycles of denaturation (94 °C, 30 s), annealing (50 °C, 30 s), and extension (72 °C, 30 s); a final extension for 3 min at 72 °C completed the amplification reaction. The amplicons were visualized using electrophoresis on SYBR Gold-stained 1.5% agarose gels. The PCR amplicons were then sequenced considering 2 × 250-bp paired-end sequences on the Illumina MiSeq platform. The raw 16S rRNA sequence data have been deposited in the DNA Data Bank of Japan-Sequence Read Archive (DDBJ-SRA) under the accession number (DRR683897-DRR683908).

Quality filtering for noise and short read sequences removal was completed in the QIIME pipeline.¹ Chimeras were identified and removed with USEARCH (Edgar et al., 2011) using the Greengenes16S rRNA gene dataset (McDonald et al., 2012) as a reference. High-quality sequences were clustered into operational taxonomic units at a 97% similarity threshold.

Metagenomic DNA was barcoded per sample and used for library preparation. Paired-end libraries (~350 bp insert size) were constructed using the TruSeq Nano DNA Library Prep Kit (Illumina) following the manufacturer’s protocol. Sequencing was conducted by Macrogen Japan Corporation (Tokyo, Japan) on the Illumina NovaSeq 6,000 platform using 2 × 250-bp paired-end sequencing. Raw metagenomic sequence data have been deposited in the DDBJ-SRA (accession numbers: DRR683870-DRR683881).

The detailed metagenomic sequence analysis of Hg-related genes is presented in Tada et al. (2023). Raw metagenomic sequences were processed using the fastq preprocessor “fastp” (Chen et al., 2018) to remove Illumina adapters, low-quality sequences (Q-value > 30; sequence length > 25 bp), and polyG tails. After quality filtering, contigs were assembled using MEGAHIT (Li et al., 2016) with default parameters. Protein-coding genes were identified from the metagenomic contigs using Prodigal 2.6.3 (Hyatt et al., 2010). To detect hgcA and hgcB sequences, a Hidden Markov Model (HMM) profile was generated using HMMER v3.2.1 (Eddy, 2009; Finn et al., 2011) with e-value thresholds of 10⁻⁵ for hgcAB and merB and 10⁻¹⁰ for merA. The HMM profile was constructed using representative sequences for hgcAB (Gionfriddo et al., 2019), merA, and merB (Christakis et al., 2021). The specificity of the HMMs for Hg-related gene sequences was validated through a local search using hmmsearch (HMMER v3.2.1) with reference sequences. Sequences lacking conserved functional motifs were excluded: hgcA-like sequences without the conserved cysteine C93 motif (Gionfriddo et al., 2016; Parks et al., 2013; Smith et al., 2015), hgcB lacking two strictly conserved CX₂CX₂CX₃C motifs (Gionfriddo et al., 2016; Smith et al., 2015), merA without a conserved cysteine pair at positions 207 and 212 (CX₄C motif), tyrosine at position 264, tyrosine or phenylalanine (for Bacteria and Archaeota, respectively) at position 605, and vicinal cysteine pair at position 628 and 629 in merA from Bacillus sp. RC607 (Boyd and Barkay, 2012; Christakis et al., 2021) (Supplementary Figure S2A), and merB without Cys-96, Asp-99, Cys-159, and Cys-117 (Christakis et al., 2021; Lafrance-Vanasse et al., 2009; Pitts and Summers, 2002) (Supplementary Figure S2B). hgcAB, merA, and merB abundances in metagenomic sequences were normalized to the abundance of rpoB (TIGR02013), a conserved single-copy bacterial gene.

After redundant sequences were removed with CD-HIT (Li and Godzik, 2006) at a 90% identity cutoff, amino acid sequences were aligned using MAFFT with the UPGMA clustering method for phylogenetic analysis (Katoh and Standley, 2013). Phylogenetic trees were constructed using FastTree (Price et al., 2009) and visualized with iTOL (Letunic and Bork, 2019).

MAGs were constructed using MetaBat2 (Kang et al., 2019) and SemiBin2 (Pan et al., 2023) with default parameters and refined with Binning Refiner (Song and Thomas, 2017). MAG completeness and contamination scores were assessed using CheckM v1.0.7–13 (Parks et al., 2015). Taxonomic classification of MAGs was assigned using GTDB-tk v2 (Chaumeil et al., 2022). Metabolic pathways of MAGs with Hg-related genes were predicted using GhostKOALA.² The completeness of each pathway was estimated and visualized using KEGG-decoder.³

Data were analyzed with R software (v.3.4.3; Pinheiro et al., 2017) using Spearman’s rank correlation analysis with the function cor.test() in the stats package. A heatmap of functional pathways of MAGs was constructed using the ggplot2 (Villanueva and Chen, 2019) in R after standardization.

The seawater temperature and dissolved oxygen concentration decreased from the surface to the mesopelagic layer (Figure 2). The depth of the oxygen minimum zone at each station was 1,000 m at St. 1, 967 m at St. 6, and 1,175 m at St. 9, respectively. In contrast, the nitrate, phosphate, and silicate concentrations increased with increasing depth. dTHg concentrations ranged from 0.42 to 0.66 at St. 1, from 0.46 to 0.63 at St. 6, and from 0.40 to 0.62 pM. The dMeHg concentrations and their proportion of dTHg increased with depth at all stations. The dMeHg concentration at 1,500 m deep was 0.033–0.041 pM, representing 5.2–9.8% of the dTHg. dMeHg concentration was significantly positively correlated with NO₃, PO₄, SiO₂ concentrations, and AOU (p < 0.01, n = 12). In contrast, no correlation was observed between dTHg concentrations and environmental parameters.

The prokaryotic abundance decreased with increasing depths and ranged from 0.52 × 10⁵ to 3.56 × 10⁵ cells mL⁻¹ at St. 1, 0.29 × 10⁵ to 3.95 × 10⁵ cells mL⁻¹at St. 6, and from 0.35 × 10⁵ to 4.19 × 10⁵ cells mL⁻¹ at St. 9 (Rho = −0.91, p < 0.001, n = 12) (Figure 3A). The number of predicted genes decreased below 500 m depth (Rho = −0.85, p < 0.005, n = 9) (Figure 3B). The abundance of the rpoB gene also tended to decline below 500 m, but this trend was not statistically significant (Figure 3C). The number of hgcA, merB, and merA genes tended to increase in the mesopelagic layers (from oxygen minimum zone to 1,500 m depths), whereas this trend was not observed for hgcB (Figure 3D). The relative abundance of hgcA, merB, and merA in the mesopelagic layers ranged from 0.10 to 0.25%, from 0.02 to 0.11, and from 0.02 to 0.10% for rpoB. Detailed information regarding metagenome sequences, contigs, predicted genes, and Hg-related genes is shown in Supplementary Table S1.

A significantly positive correlation was observed between hgcA relative abundance and dMeHg concentration, as well as the dMeHg / dTHg ratio (p < 0.01, n = 12) (Figure 4). In addition, a significantly positive correlation was noted between dMeHg and merB (p < 0.05, n = 12). A significant correlation was also observed between hgcA and merB abundance and NO₃, PO₄, and SiO₂, and AOU (p < 0.01, n = 12). These results suggest that the presence of Hg-related genes in prokaryotes influences the distribution of MeHg in seawater, which is in turn affected by various environmental factors. The hgcA abundance was higher than that of merB and merA, with merA being slightly more abundant than merB (Supplementary Figure S3).

Phylogenetic analysis of hgcA (methyl carrier), merB, and merA indicated that a distinct phylogenetic lineage was dominant for each Hg-related gene (Figure 5A and Supplementary Figures S4–S7). Nitrospina dominated the total hgcA-like sequences and accounted for 70.8% of hgcA genes. This lineage was predominant among the hgcB genes (546 out of 550 sequences) (Supplementary Figure S5 and Supplementary Table S2). Additional hgcA genes were affiliated with Uncultured Deltaproteobacteria and Alphaproteobacteria, representing 28.7 and 0.5% of rpoB genes, respectively. In merB, various lineages were identified, including Alphaproteobacteria (59.8%), Betaproteobacteria (14.9%), Deltaproteobacteria (10.3%), Euryarchaeota (9.2%), Actinobacteria (2.3%), Nitrospirae (1.1%), and unclassified bacteria (2.3%). Alphaproteobacteria and Betaproteobacteria were also detected in merA and accounted for 16.9 and 15.5% of rpoB genes, respectively. However, Gammaproteobacteria (42.3%) was the dominant lineage in total merA sequences. Additionally, Acidithiobacillia, Cyanobacteria, and unclassified bacteria merA were also observed.

The vertical distribution of each phylogenetic lineage associated with hgcA, merB, and merA is shown in Figure 5B and Supplementary Figures S3–S7. Nitrospina and uncultured Deltaproteobacteria hgcA sequences were abundant in the mesopelagic layers at all stations. Alphaproteobacteria merB were observed below 500 m and were abundant at a depth of 1,500 m at all stations. Additionally, Euryarchaeota merB sequences were detected between 500 m and 1,500 m. In contrast, merB sequences affiliated with Desulfobacterota and Actinobacteria were detected in the upper 500 m. Notably, Betaproteobacteria merB was widely distributed from the surface to mesopelagic layers. For the merA genes, Betaproteobacteria and Gammaproteobacteria sequences were distributed from the surface to mesopelagic layers. The abundance of Gammaproteobacteria merB increased at 1,500 m deep at Sts. 6 and 9. MerA sequences affiliated with Actinobacteria and Cyanobacteria were detected below 500 m. Similar to the distribution of merB, Alphaproteobacteria merA were abundant in the mesopelagic layers across all stations.

A total of 1,308 MAGs was constructed from metagenome sequences (Table 2). Pathway analysis revealed that MAGs carrying hgcAB, merB, and merA contained several distinctive metabolic functions (Figure 6). hgcAB-carrying Nitrospina possessed the cytochrome c oxidase and nitrite reduction pathways; however, their genome completeness values were relatively low (30–50%). MAGs affiliated with Myxococcota (Deltaproteobacteria) harboring hgcAB included genes for dissimilatory sulfate in the adenosine phosphosulfate pathway. Myxococcota and other Pseudomonadota MAGs with hgcAB also exhibited genes involved in beta-glucosidase, dissimilatory sulfate, and sulfur assimilation pathways. merB-carrying MAGs affiliated with Alphaproteobacteria contained genes involved in methanogenesis via trimethylamine. Additionally, genes associated with dissimilatory sulfite to adenosine phosphosulfate and sulfide oxidation pathways were observed in an alphaproteobacterial MAG with merB. Gammaproteobacteria MAGs with merA exhibited versatile metabolic pathways such as oxidative phosphorylation, carbohydrate metabolism, nitrogen metabolism, sulfur metabolism, methanogenesis, and secretion systems. In contrast, Gemmatimonadota MAGs showed comparatively simple metabolic profiles, comprising carbohydrate metabolism, oxidative phosphorylation, sulfur metabolism, cofactor and vitamin metabolism, and metal-related transporters. Metabolic pathways related to methane and sulfur cycles were observed in a SAR324 MAG.

In the upper mesopelagic layers in the WNPSG, a significant positive relationship was observed between dMeHg (dMeHg/dTHg ratio) and AOU. Generally, the AOU serves as a proxy for microbial activity, suggesting that prokaryotes within the seawater column possibly contribute to MeHg production in the WNPSG. Similar positive correlations between MeHg concentration and AOU in the seawater column were observed in the upper mesopelagic layers (<1,500 m depth) in the Atlantic Ocean (Mason and Sullivan, 1999), the Mediterranean Sea (Cossa et al., 2009; Cossa et al., 2011), and the Oyashio and Kuroshio regions of the western North Pacific (Tada et al., 2021). Notably, the WNPSG is a subtropical gyre characterized by relatively low nutrient supply and reduced vertical organic matter flux (Karl and Church, 2017). The persistence of a clear correlation between dMeHg (or dMeHg/dTHg) and AOU even under these low-productivity conditions suggests that MeHg production driven by local remineralization can be significant even in areas with diffident organic matter input. Furthermore, the linkage between oxygen consumption and Hg methylation is a robust feature across contrasting oceanic environmental conditions. Additionally, the relative abundance of hgcA increased in the oxygen-depleted mesopelagic layers with high dMeHg concentrations (hgcB was also detected). Taken together, WNPSG data and observations from the Atlantic, Mediterranean, North Pacific marginal seas, and Black Sea (Villar et al., 2020; Tada et al., 2021; Cabrol et al., 2023) suggest that hgcA-possessing prokaryotes (strictly speaking, both hgcAB-possessing) contribute to the formation of relatively high MeHg-containing water masses in oxygen-depleted mid-water layers across a range of environments, from highly productive to oligotrophic.

The depth profile of hgcB relative abundance differed from that of hgcA, and no positive correlation was observed between hgcB and MeHg concentration. These discrepancies in the distribution of hgcAB have been observed in the Arctic Ocean metagenomes (Bowman et al., 2020) and in previous studies conducted in the East China Sea and the Oyashio region (Tada et al., 2021). Detection efficiencies for hgcB may be limited by protocol and assembly methods optimized primarily for hgcA detection. Similarly, differences in the evolutionary background and conservation between hgcA and hgcB could lead to discrepancies in their detection rates in metagenomic analyses. These findings highlight the need to consider both methodological and evolutionary factors when interpreting the prevalence of Hg-methylating genes in environmental samples.

In a metagenomic survey of Hg-related genes, merB was detected in the mesopelagic layers of both the Arctic Ocean and the equatorial Pacific (Bowman et al., 2020). Furthermore, incubation experiments using enriched Hg isotopes demonstrated that MeHg demethylation can occur within the central Pacific water column (Munson et al., 2018), indicating that microbial MeHg demethylation may co-occur with Hg methylation in mesopelagic zones. In the WNPSG, an increased relative abundance of merB in high-dMeHg mesopelagic layers further supports the widespread distribution of merB-carrying prokaryotes and their potential involvement in MeHg degradation. A positive correlation between merB abundance and dMeHg concentration suggests that in environments rich in methylmercury, microorganisms possessing MeHg degradation capabilities may exist as a defense mechanism against it. Notably, the higher relative abundance of hgcA than that of merB in the WNPSG implies that Hg methylation is more prominent than MeHg demethylation in these layers. Although Bowman et al. (2020) reported that merB abundance exceeded that of hgcA in the equatorial North Pacific (8°N, 156°W), these samples were collected from depths of 800 m or less, complicating direct comparisons with deeper profiles.

Phylogenetic analysis of Hg-related genes supports previous findings that Nitrospina bacteria may be the dominant Hg methylators in the mesopelagic layers of the open ocean (Bowman et al., 2020; Villar et al., 2020; Tada et al., 2020, 2021). Metatranscriptomic data from the Tara Oceans project confirmed hgcA expression in this lineage (Villar et al., 2020), indicating a widespread distribution and potential contribution of Nitrospina bacteria to Hg methylation on a basin scale. However, the actual Hg methylation capacity of the Nitrospina lineage remains unconfirmed, as no isolates harboring hgcA have been obtained. For example, the genome of Nitrospina gracilis is associated with nitrite oxidation through the presence of genes encoding ferredoxin-nitrite reductase, copper-containing nitrite reductase, and nitrite oxidoreductase (Lücker et al., 2013). Additionally, Nitrospina lineages have been detected in oxygen minimum zones (Spieck et al., 2014; Pachiadaki et al., 2017; Sun et al., 2019), and their MAGs include cytochrome c oxidase, suggesting adaptation to low-oxygen environments (Sun et al., 2019). Functional pathway analysis further revealed that Nitrospina MAGs harboring hgcAB also possess nitrite-reducing pathways and cytochrome c oxidase, supporting their potential role in Hg methylation under low-oxygen conditions. Additionally, studies employing enriched Hg isotopes have identified a positive correlation between Hg methylation and nitrification rates (Starr et al., 2022; Despins et al., 2023), suggesting a close relationship between microbial Hg methylation by Nitrospina bacteria and the oceanic nitrogen cycle.

Members of Deltaproteobacteria, the second dominant lineage among hgcA-carrying microbes in the metagenome, have been identified in oxygen-deficient seawater in coastal seawater (Capo et al., 2020; Lin et al., 2021; Tada et al., 2023) as well as in the open ocean (Villar et al., 2020). In the WNPSG, hgcA affiliated with uncultured Deltaproteobacteria were particularly abundant. Distinct deltaproteobacterial hgcA sequences associated with Desulfovibrionales, Desulfobacterales, and Syntrophobacterales were observed in the Tara Oceans metagenomes (Villar et al., 2020), suggesting that diverse deltaproteobacterial lineages are involved in Hg methylation in the open ocean environments. Moreover, Deltaproteobacteria-related MAGs closely related to Myxococcota contained genes for dissimilatory sulfate reduction, indicating a potential link between Hg methylation and the sulfur cycle in the mesopelagic layers of the WNPSG.

Nitrospina hgcA and Deltaproteobacteria hgcA were frequently detected in the mesopelagic layer, but 16S rRNA gene analysis showed no significant increase in the relative abundance of Nitrospina or Deltaproteobacteria in these layers (Supplementary Figures S8, S9). Notably, the relative abundance of Nitrospina tended to be higher in the upper mesopelagic layer (200–500 m). These findings suggest that specific Nitrospina and Deltaproteobacteria, which cannot be identified solely from 16S rRNA gene analysis, contribute to Hg methylation within the WNPSG.

To date, Hg methylation by aerobic microorganisms under aerobic conditions also remains unconfirmed. However, one hgcA gene affiliated with Alphaproteobacteria and closely related to Defluviimonas indica—a strictly aerobic and chemoheterotrophic marine bacterium (Jiang et al., 2014)—was reported in the mesopelagic layer, suggesting that aerobic microbes may be involved in Hg methylation in the open ocean. Although metagenomic approaches can identify the potential for Hg methylation, culture-based research is necessary to confirm methylation activity. Therefore, combining culture-dependent and independent methods is essential to fully elucidate the processes underlying microbial MeHg production in marine environments.

The phylogenetic distribution of merB is more limited than that of merA (Christakis et al., 2021). However, diverse lineages of merB were detected in the mesopelagic layers of the WNPSG, similar to merA. Notably, phylogenetic lineages with both merB and merA are likely crucial in removing MeHg from the mesopelagic zone. In the present study, members of Novosphingobium (Alphaproteobacteria) were detected in both merB and merA gene datasets, suggesting their potential to transform MeHg to Hg(0). Unfortunately, no MAGs containing both merB and merA genes were detected, likely because of limited sequence depth. Thus, additional metagenomic reads are necessary to clarify the metabolic function of these lineages.

Notably, some alphaproteobacterial MAGs harboring either merB or merA also possessed pathways for methanogenesis via trimethylamine (TMA), aligning with previous observations of elevated methane concentrations in the anoxic layers in the eastern North Pacific (Thamdrup et al., 2019). This finding suggests that microbial processes involved in both MeHg degradation and Hg(II) reduction may be coupled with methane production in the mesopelagic zone.

Burkholderia (Betaproteobacteria), a lineage carrying both merB and merA sequences, was detected throughout the water column, from the surface to mesopelagic layers. Although typically rare in marine environments (Morris et al., 2006), Burkholderia accounted for 0.02–2.1% (average 0.39%) of total sequences (Supplementary Figure S9), suggesting that this low-abundance lineage may nonetheless play a crucial role in MeHg degradation and Hg reduction in the mesopelagic layers of the WNPSG.

Contrary to earlier culture-dependent studies, which failed to identify bacteria possessing merB without merA (Boyd and Barkay, 2012), the findings of this study show that certain lineages, including Desulfobacterota (Deltaproteobacteria) and Nitrospirae possess only merB. These organisms may use alternative pathways for Hg(II) reduction, as some Alphaproteobacteria (e.g., Magnetospirillum gryphiswaldense MSR-1 and M. magnetotacticum MS-1) lack merA reduce Hg(II) to Hg(0) (Liu and Wiatrowski, 2018). A genomic survey further revealed that nearly half of the prokaryotic genomes with merB lack merA (Christakis et al., 2021). Although the Hg(II) reduction capacity of the merB-only lineages detected in this study remains unconfirmed, they may contribute to MeHg degradation in the WNPSG.

A survey of Hg-related genes in ocean genome databases suggests the widespread marine distribution of merA-carrying microorganisms (Sanz-Sáez et al., 2022). In this study, Gammaproteobacteria dominated merA-containing lineages from the surface to mesopelagic layers, with Marinobacter (Alteromonadales) representing the most abundant group (14 out of 38 sequences). Marinobacter merA was consistently detected across depths, and 16S rRNA gene sequencing revealed increased relative Marinobacteraceae abundance in mesopelagic layers (Supplementary Figure S9). Given that 89.1% of tested Marinobacter isolates from surface to deep-sea environments harbor merA genes (Sanz-Sáez et al., 2022), these findings suggest that Marinobacter plays a key role in Hg(II) reduction in mesopelagic waters globally.

Metagenomic and metatranscriptomic surveys across the Atlantic, Pacific, and Indian Oceans have identified diverse merA-harboring lineages, including Corynebacteriales (Actinomycetota), Rhodobacterales (Alphaproteobacteria), Alteromonadales, Oceanospirillales, Moraxellales (Gammaproteobacteria), and Flavobacteriales (Bacteroidota) (Sanz-Sáez et al., 2024). This study additionally detected Burkholderia (Betaproteobacteria), Sphingobium (Alphaproteobacteria), Halothece (Cyanobacteria), and Methanosarcinales (Euryarchaeota), highlighting the broad taxonomic diversity of merA-containing microorganisms in marine environments. Furthermore, merA sequences related to Betaproteobacteria have also been reported in the coastal waters of Japan (Tada et al., 2023), suggesting their potential role in Hg(II) reduction across both coastal and open ocean ecosystems.

Metabolic pathway analysis revealed that Alphaproteobacterial MAGs containing either merB or merA possess a TMA-linked methane production pathway. Although TMA concentrations in seawater are typically low (nanomolar levels), elevated levels near the thermocline and oxycline have been reported (Gibb et al., 1999; Cree et al., 2018). In coastal sediments, benthic animals and phytoplankton serve as major TMA sources through direct release or decomposition (Wang and Lee, 1994). These findings suggest that TMA-driven methane production may support concurrent MeHg demethylation and Hg(II) reduction in the mesopelagic layers of the WNPSG.

Gemmatimonadota, a rare lineage in marine environments (<1% of 16S rRNA sequences) (Aldeguer-Riquelme et al., 2023), demonstrates metabolic versatility, including organic carbon degradation, denitrification, sulfate reduction, and sulfide oxidation (Gong et al., 2024). In this study, 11 out of 36 merA-carrying MAGs were affiliated with Gemmatimonadota. Additionally, the SAR324 lineage, which has diverse metabolic potential spanning heterotrophic and autotrophic pathways (Malfertheiner et al., 2022), also contained merA, with pathways for carbohydrate, sulfur, and metal metabolism. Although Hg(II) resistance in these groups remains unconfirmed, their metabolic traits suggest a role in Hg speciation associated with carbon and sulfur cycling in marine ecosystems.

In conclusion, the study investigated Hg methylation and MeHg demethylation processes mediated by microorganisms in the WNPSG water column. A significant positive correlation between MeHg concentration and AOU suggests that microbial activity (presumably microbial remineralization of organic matter) plays a crucial role in Hg methylation. The abundance of hgcA, a key gene for Hg methylation, increased in oxygen-depleted mesopelagic layers with high MeHg concentrations. merB, involved in MeHg demethylation, was also prevalent in the WNPSG, indicating simultaneous MeHg degradation. However, the relatively higher abundance of hgcA compared to merB suggests that Hg methylation processes may dominate over demethylation. Phylogenetic analysis of hgcA sequences identified Nitrospina bacteria with nitrite reductase pathway as dominant Hg methylators, linking Hg methylation to the nitrogen cycle. Additionally, Myxococcota (Deltaproteobacteria) was associated with sulfur cycling. Diverse microbial lineages carried merB and merA, suggesting co-occurring MeHg demethylation and Hg(II) reduction in the mesopelagic zone. Gammaproteobacteria, particularly Marinobacter, emerged as key contributors to Hg reduction. These results highlight the complex microbial interactions driving Hg transformations in the mesopelagic layers in the WNPSG. Functional pathway analysis of MAGs with Hg-related genes provides insight into microbial Hg transformations in the ocean. However, future studies integrating comprehensive datasets of hgcA, merB, and MeHg concentrations are essential to clarify the depth-related distribution of MeHg and associated microbial processes.

In this study, we adopted a conservative residue-based curation strategy to identify merA and merB homologs in our metagenomes. As detailed in the ‘Materials and Methods’ section, we excluded merA sequences lacking any of the following conserved residues present in the Bacillus sp. RC607 MerA reference: the catalytic cysteine pair Cys-207 and 212; Tyr-or Phe-605; and the C-terminal vicinal Cys-628 and 629. Similarly, merB sequences were discarded if they lacked Cys-96, Asp-99, Cys-117 and Cys-159, which together form a catalytic and structurally important cysteine cluster in the R831b MerB reference. By restricting our analyses to sequences that retain the full set of experimentally validated catalytic and metal-binding residues, we focused on gene variants most likely to encode functional MerA and MerB enzymes, reducing the impact of spurious annotations, partial open reading frames and distant paralogues. This stringent curation unavoidably leads to conservative estimates of merA and merB prevalence because atypical variants lacking one or more of these residues were not considered. While some of these variants may represent nonfunctional remnants, others could encode enzymes with altered catalytic properties or substrate spectra. Therefore, our counts should be regarded as minimum estimates of canonical, biochemically supported merA and merB genes. Future work combining targeted biochemical characterization with metagenomic surveys will be required to clarify the ecological roles of these atypical variants.