Microbial membrane transporters reveal trace metal niche adaptation in distinct water masses of the Southern Ocean

TL;DR

This study explores how microbes in the Southern Ocean adapt to trace metals in different water layers, revealing patterns in their transporters that suggest niche adaptation.

Contribution

The paper introduces microbial membrane transporters as novel indicators for trace metal niche adaptation in distinct Southern Ocean water masses.

Findings

Normalized gene abundances of transporters for Fe, Mn, Ni, and Cu show distinct spatial and vertical patterns.

Enrichment of efflux and homeostasis genes for Fe, Ni, and Cu is observed in NADW and LCDW water masses.

Alteromonadaceae and Burkholderiaceae are identified as key players in trace metal adaptation in deep ocean water masses.

Abstract

Trace metals are co-factors for enzymes that are essential for microbial metabolism and the cycling of major elements. Membrane transporters allow microbes to sense and react to trace elements in the environment and to balance their uptake and export for the regulation of intracellular metal homeostasis. The acquisition and efflux of trace metals could lead to reciprocal feedbacks between microbes and the surrounding environment. Whether these processes vary among trace metals and across habitats is presently not known. We used membrane transporters into and out of the cell as indicators for the uptake and efflux of trace metals and provide a detailed picture of the distribution of the respective genes in distinct provinces in surface waters and in subsurface water masses across a transect in the Southern Indian Ocean.

We observed marked spatial and vertical patterns in normalized gene abundances of transporters of iron (Fe), manganese (Mn), nickel (Ni) and copper (Cu). Changes in gene abundances were specific to the type of transporter and trace metal, and pronounced differences between surface and specific water masses emerged. We found an enrichment in genes related to efflux and homeostasis of Fe, Ni and Cu in two water masses of the deep ocean that are North Atlantic Deep Water (NADW) and Lower Circumpolar Deep Water (LCDW). This pattern was observed on the community level and for metagenome-assembled genomes (MAGs) affiliated with Alteromonadaceae and Burkholderiaceae that were abundant in these two water masses.

The enrichment in trace metal efflux and resistance genes points to microbially mediated processes, exerted by homeostasis, with potential influence on the trace metal speciation and distribution in specific water masses in the deep ocean. The gene repertoire and distinct distribution pattern of the taxa identified as potential key players could reflect an adaptation to these old water masses with trace metals acting as selective driver.

Video Abstract

Video Abstract

The online version contains supplementary material available at 10.1186/s40168-025-02293-6.

Figures

Normalized abundance of genes related to the transport of Fe (Fe^3+^, Fe^2+^) (A), heme and siderophores (B), Mn (C), Ni (D), Cu (E), and Ni/Cu (F) in free-living communities. Normalized gene abundances are given in genes per kilobase million (GPM). GPM of particle-attached communities are shown in Fig. S1. Note the difference in scale for each plot

Ratios of normalized gene abundance (GPM) of trace metal transporters in water masses to surface waters for the free-living fraction. Genes are ordered according to gene abundance in surface waters (mean of GPM in SASW, PFSW and ASW, left panel). Red colours indicate that gene abundance is higher in a given water mass as compared to the surface and purple colours indicate that gene abundance is lower in a given water mass. Inserted (*) indicates that the gene abundance in a given water mass is 0. Respective ratios for the particle-attached fraction are presented in Fig. S4

Mean coverage of MAGs belonging to the ten most abundant ones in a given water mass (left panel) and corresponding counts of trace metals transport genes (right panel) in the free-living fraction. Water masses are separated by dashed lines. Archaeal MAGs are marked by an asterisk. The MAGs that belong to the top ten in the free-living and particle-attached fractions (Fig. S9) are shown in bold

Full text

Introduction

Bioactive metals are indispensable co-factors for enzymes involved in fundamental cellular processes that support biogeochemical cycling in the ocean. Among these, iron (Fe) is critical for processes such as organic carbon catabolism and oxidative phosphorylation in the respiratory chain [1]. Manganese (Mn) plays a central role in carbon metabolism [2], while nickel (Ni) is essential for nitrogen fixation and the processing of nitrogenous substrates [3]. Copper (Cu) is a catalytic component in key enzymes, including cytochrome c oxidase and nitrite reductase [4]. While essential for metabolism, metals can be toxic if present at high intracellular concentrations. Microbes therefore need to maintain metal homeostasis through tightly regulated import and export mechanisms, facilitated by specific transporters and transcriptional regulators that respond to intracellular and environmental metal concentrations [5].

How trace elements shape microbial genomic features across environmental gradients and in habitats with distinct trace metal properties remains, however, poorly understood. Previous studies have in particular investigated Fe, as this micro-nutrient is known to be limiting for phototrophic and heterotrophic processes. Genes related to Fe metabolism were shown to vary with in situ Fe concentrations in microbial communities [6] and in genomes of abundant cyanobacteria in surface waters across the global ocean [7, 8]. Transporters of different forms of Fe were proposed as biomarkers for nutrient stress in marine microbes [6–8]. The parallel distribution of the trace metals Fe, Mn, Cu and Ni and prokaryotic taxa in distinct water masses of the Southern Ocean indicated reciprocal links between microbes and their biogeochemical environment [9]. These connections could be driven by the interplay between the acquisition of trace elements and their export to maintain metal homeostasis. Microbes could thereby influence trace metal conditions in the surrounding environment, but the underlying processes and associated genomic signatures remain to be explored [10].

Biogeographic patterns align with distinct characteristics of the habitats that host microbial communities [11]. In the sub surface ocean, the environment is constrained by the presence of different water masses which travel and mix in the ocean interior with time scales ranging from years to centuries [12]. Concomitant biogeochemical processes finally shape these physico-chemical habitats and the microbes inhabiting them [13, 14]. However, the role of bioactive trace elements as drivers for adaptations to these biomes of the deep ocean is not well known. The Southern Ocean is the region of formation of a few water masses (Antarctic Bottom Water, Antarctic Intermediate Water, Mode Waters), but also of transit and transformation of others originating from the three adjacent basins (Atlantic, Indian, Pacific) [12]. This region offers a large panel of different water masses and associated biogeochemical environments, and thus provides a unique context to investigate the potential reciprocal links between microbial trace metal trafficking and the properties of the habitats by which microbes are hosted. Moreover, because the Southern Ocean is a hub of the general ocean circulation, biogeochemically driven changes of water masses leaving this ocean might impact the functioning of marine ecosystems in other basins [15–17].

The aim of this study was to investigate the capacity for uptake and efflux of Fe, Mn, Ni and Cu in habitats with varying trace metal characteristics. We used microbial membrane transporters of these elements as indicators of the respective processes at the community level and in metagenome-assembled genomes (MAGs) in different surface provinces and water masses of the Indian Sector of the Southern Ocean.

Materials and methods

Sample collection

Seawater samples were collected during the South West Indian Ocean GEOTRACES GS02 Section (SWINGS) cruise from 10th January to 8th March 2021 (Fig. 1A). The stations considered here were located in the Subtropical Zone (STZ), Subantarctic Zone (SAZ), the Polar Frontal Zone (PFZ) and the Antarctic Zone (AAZ) and provided access to 4 types of surface water that are Subtropical Surface Water (STSW), Subantarctic Surface Water (SASW), Polar Frontal Surface Water (PFSW) and Antarctic Surface Water (ASW) (Fig. 1 A & B). Below the mixed layer, the seawater was classified into 8 water masses defined as a body of water that is no longer in contact with the atmosphere (Winter Water, WW; Antarctic Intermediate Water, AAIW; Subtropical Mode Water, STMW; South Indian Central Water, SICW; Upper Circumpolar Deep Water, UCDW; Lower Circumpolar Deep Water, LCDW; North Atlantic Deep Water, NADW; Antarctic Bottom Water, AABW) based on the physical and chemical properties such as temperature, salinity, oxygen concentration, depth, and latitude and longitude (Fig. 1B, Table S1, Table S2, Supplementary Material).Fig. 1A Map of stations from the SWINGS cruise in the Indian Sector of the Southern Ocean. Shown are stations that were used to produce the map of water masses as illustrated in panel B, and the stations where samples were collected for the present study. Colour shading represents bathymetry and the grey line contours South Africa (25°–35° E) and Madagascar (45°–50° E). B A cross-section (inserted map) showing the vertical distribution of some water masses at the stations indicated on the upper x-axis. Lines indicate salinity. STSW, Subtropical Surface Water; SASW, Subantarctic Surface Water; ASW, Antarctic Surface Water; WW, Winter Water; AAIW, Antarctic Intermediate Water; UCDW, Upper Circumpolar Deep Water; LCDW, Lower Circumpolar Deep Water; NADW, North Atlantic Deep Water; AABW, Antarctic Bottom Water. The full list of water masses is provided in Fig. 2 and Table S2Fig. 2Normalized abundance of genes related to the transport of Fe (Fe^3+^, Fe^2+^) (A), heme and siderophores (B), Mn (C), Ni (D), Cu (E), and Ni/Cu (F) in free-living communities. Normalized gene abundances are given in genes per kilobase million (GPM). GPM of particle-attached communities are shown in Fig. S1. Note the difference in scale for each plot

Seawater samples (6 L) were collected using 12 L Niskin bottles attached to a frame with conductivity, temperature and depth (CTD) sensors (SeaBird SBE911plus). The seawater was then sequentially filtered through 0.8 μm polycarbonate filters (47 mm diameter, Nuclepore, Whatman, Sigma-Aldrich, St Louis, MO) and 0.22 μm Sterivex filter units (Sterivex, Millipore, EMD, Billerica, MA). The prokaryotes collected on the 0.8 μm filters were classified as particle-attached, while those collected on the 0.22 μm filters were considered as free-living prokaryotes. A total of 42 seawater samples were collected, comprising 23 free-living and 19 particle-attached samples (Table S1). The filters were stored at − 80 °C until returned to the home laboratory for DNA extraction.

DNA extraction and metagenomic sequencing

Total DNA extraction was performed from 0.8 μm polycarbonate filters and 0.22 µm Sterivex filter units using the DNeasy PowerWater Kit (Qiagen) following the manufacturer’s instructions with a few modifications. Sterivex units were opened according to Perrine’s methods [18]. The 0.8 and 0.22 filter membranes were then cut into small pieces using scissors and transferred to the new 2 mL Eppendorf tubes. Subsequently, solution PW1 and a lysozyme solution were added, and the mixture was incubated at 37 °C for 45 min to promote cell lysis, and then a proteinase K solution was added and incubated at 55 °C for 1 h to digest the proteins in the cell lysate; the remaining steps were carried out following the manufacturer’s instructions. DNA concentrations were measured using a Promega Quantus fluorometer with the QuantiFluor® Double stranded DNA (dsDNA) system. Metagenomes were sequenced with an Illumina NovaSeq 6000 system using 2 × 150 bp chemistry at Fasteris SA, Inc. (Switzerland), and the total number of reads is shown in Table S1. Raw sequences in fastq format are deposited at the European Nucleotide Archive (ENA) repository under project ID PRJEB75506.

Metagenome assembly and functional profiling

The raw reads underwent quality assessment using FastQC v0.11.9 and subsequent preprocessing was carried out using Trimmomatic (v 0.32) as described by Bolger et al. [19]. Following quality control, the retained short reads were assembled by sample using MEGAHIT v1.2.9 [20] with specific parameters (-min-contig-len 1000 -presets meta-large). The resulting contigs exhibited N50 lengths ranging from 1.4 k to 3.5 k. The open reading frames (ORFs) were predicted using Prodigal v2.6.3 [21] with the following parameters: -p meta. To create a non-redundant protein database of all 42 samples, CD-HIT v4.8.1 [22] was used for clustering of the concatenated files with the parameters: -c 1 -aS 1 -g 1. This resulted in a non-redundant protein collection of 18,497,675 proteins.

For read recruitment of short reads to the non-redundant gene sequence collection and quantification of each gene in each sample, Salmon v1.4.0 [23] was used. The specific parameters utilized were as follows: -meta -incompatPrior 0.0 -seqBias -gcBias -biasSpeedSamp 5 -validateMappings. In order to make the protein occurrences comparable across samples, the data was normalized as genes per kilobase million (GPM). The GPM calculation is defined as follows: GPM = 1 M × (mapped genes/gene length)/(sum of mapped reads/gene length).

To identify genes associated with intracellular iron cycling, we used FeGenie [24] based on the non-redundant protein sequence set for alignment and annotation (parameters used -meta). FeGenie generated its own database of HMMs based on iron-related genes and was used for the identification of proteins involved in intracellular Fe cycling and transport. To identify genes associated with other trace metal transporters, the non-redundant protein sequence set was searched against the KEGG database using GhostKOALA [25] and KofamKOALA [26]. In this study, two specific Mn-related uptake transporters were detected in all samples: the NRAMP (natural resistance-associated macrophage protein) family type transporter MntH, and the ABC (ATP-binding cassette) type transporter MntABC. Additionally, two Mn-efflux transporters, MntP and MneA, were detected, along with non-specific Mn uptake transporter genes. Most of the genes associated with copper transport identified in the free-living and particle-attached samples were exporter and efflux genes, and only one was an uptake gene (SLC31A1, CTR1). Details of trace metal transporters can be found in Table S3. To investigate relationships between the trace metal transport gene abundance and dissolved or particulate trace metal concentrations among water masses, the Spearman’s correlation analyses were made.

Binning

The individual metagenome assemblies (contigs ≥ 2500 bp) underwent a comprehensive binning process to yield higher quality metagenome-assembled genomes (MAGs). This binning process used MetaWRAP (v1.3) [27] with two binning tools: MaxBin2 (v2.2.5) [28] and MetaBAT2 (v2.12.1) [29]. Subsequent refinement was carried out by the Bin_refinement module in MetaWRAP (v1.3). First, mixed bin sets were created by combining bin sets generated by MaxBin2 and MetaBAT2, and then the selection of the best MAGs was based on the completeness and contamination for each MAG in these three sets as evaluated by CheckM [30]. To enhance the quality of the MAGs, the sequence reads from the metagenome were mapped to each MAG. The MAGs were then reassembled using the reassemble_bins module in MetaWRAP, contributing to an improvement in their overall quality. Following this reassembly step, the MAGs were reevaluated by CheckM. A total of 1103 MAGs were retained based on the thresholds of completeness greater than or equal to 50% and contamination less than or equal to 10%. To reduce redundancy, all reassembled MAGs were dereplicated using dRep [31] with the following parameters: -comp 50 -con 10 -nc 0.25 -pa 0.9 -sa 0.95. Subsequently, 556 MAGs were retained based on the 95% average nucleotide identities (ANI). Within this set, the completeness of 157 MAGs was greater than or equal to 90%, the completeness of 270 MAGs was greater than or equal to 80%, and the completeness of 361 MAGs was greater than or equal to 70%.

To calculate the abundance of each MAG across samples, we first employed Bowtie2 v2.4.4 [32] with default parameters to map short reads from 42 metagenomes. Subsequently, samtools [33] were utilized to convert the resulting SAM files into sorted and indexed BAM files. Anvi’o v7.1 [34] was then used to profile short metagenomic reads aligned to MAGs, enabling the estimation of coverage statistics for each metagenome, and the details are as described by Zhang [35]. Open reading frames (ORFs) were predicted for all MAGs using Prodigal. Functional annotation was conducted using GhostKOALA [25] and KofamKOALA [26], and metabolic pathways were determined with METABOLIC (version 4.0), applying the threshold of 75% of metabolic steps/genes’ presence within a given KEGG module [36]. Taxonomic classification of 556 MAGs was assigned using GTDB-Tk v2.3.0 [37], referencing the GTDB database release 214, and the classify_wf function with default parameters was used for this purpose. Maximum Likelihood (ML) phylogenetic trees of MAGs were constructed using IQ-Tree (v2.0.3) [38] with the following parameters: -m TESTMERGE -bb 1000 -bnni. The resulting trees were then visualized using iTOL [39].

Results and discussion

We observed marked differences in the distribution of several trace metal transporters in the free-living fraction (< 0.8 µm) (Fig. 2). In surface waters, normalized gene abundances (GPM) of transporters of Fe (Fig. 2 A, B) and Mn (Fig. 2C) were higher in SASW, PFSW and ASW as compared to STSW, a pattern that was not observed for transporters of Cu and Ni (Fig. 2 D, E, F). GPM revealed overall minor variability among SASW, PFSW and ASW for any type of transporter, but distinct patterns between surface and depth emerged. GPM of siderophore transporters and heme were higher in surface waters (SASW, PFSW, ASW) as compared to water masses (WW, STMW, AAIW, SICW, UCDW, LCDW, NADW, AABW) (Fig. 2B), with the exception of LCDW and NADW. Transporters related to Mn and Ni revealed a similar pattern (Fig. 2C, D), while the opposite was observed for Cu (Fig. 2 E). The highest GPM for Ni and Cu transporters were again observed in LCDW and NADW (Fig. 2 D, E, F). In the particle-attached fraction, trace metal transporters had a patchier distribution as compared to the free-living fraction; still, the highest GPM for siderophore, Ni and Cu transporters were again observed in LCDW and NADW (Fig. S1).

Acquisition of Fe was dominated by transporters of ferric Fe (Fe^3+^; fbp, fbp-futB, fut A1/A2) across stations and depths and transporters of ferrous Fe (Fe^2+^; efeUOB, feoAB) had substantially lower GPM (Fig. 2A, Fig. S2), a pattern that was less pronounced in the particle-attached communities (Fig. S1, S3). GPM of the transporter yfe (Fe/Mn) were higher in surface waters (STSW, SASW, PFSW, ASW) as compared to water masses, while an inverse pattern was observed for fbp (Fig. S1–S3). The most abundant siderophore transporter genes were fpv (pyoverdine), pirA (ferric enterobactin), hatD (aerobactin) and viuB (vibriobactin) across surface waters and at depth (Fig. 2B and Fig. S1–S3). The sitABCD transport system (Mn/Fe) dominated south of the polar front in SASW, PFSW and ASW, while the putative Mn exporter mne was abundant in STSW (Fig. 2C and Fig. S1–S3). In deeper waters, the transport system tro, mnt, znu (Mn/Zn/Fe) accounted for most of the Mn-related transporters. The two putative Mn exporters, mntP and mneA had increased GPM in LCDW and NADW, both in the free-living and particle-attached communities (Fig. 2C, Fig. S1–S3). In the case of Ni, we observed a clear spatial differentiation in the types of transporter genes. While cbiO (Co/Ni), nikM/cbiM (Co/Ni), hup/ure/cooJ (Ni) and cbik (Ni) had a roughly equal contribution in STSW, cbik dominated in SASW, PFSW and ASW, and cbik and hup/ure/cooJ had a similar share at depth (Fig. 2D and Fig. S2). Most of the genes associated with Cu transport identified were Cu efflux or resistance genes (copC/pcoC, pcoB/copB, cusABF), with the exception of CTR1 (Fig. 2E), and these genes had overall similar contributions across surface and depth.

To illustrate the enrichment or depletion of a given trace metal transporter at depth, we calculated the ratio of the normalized gene abundance between the surface and each water mass (Fig. 3, Fig. S4). Given the overall low variability in GPM among SASW, PFSW and ASW, we used the mean GPM of these surface waters as our reference. Most transporters of Fe^3+^ (yfeAB, fbp-futB, fut A1/A2, futC), heme (hmu) and siderophores were equal to or slightly depleted in water masses as compared to the surface, while Fe^2+^ transporters (efeUOB, feoAB) and Fe efflux genes (feoE) were enriched at depth. For comparison, the gene coding for bacterioferritin (ftn) remained stable. The most abundant transporters for Mn (sit) and Ni (cbi, hup) were depleted at depth, while Mn efflux genes (mntP, mneA) and Ni homeostasis genes (rcn) were enriched. The Cu transporter CTR1 was also depleted in several water masses, while Cu export and homeostasis genes (csz, cus, cnr, cop) were enriched at depth. The enrichment in efflux genes of all trace metals was particularly pronounced for LCDW and NADW, a trend that was also observed for the particle-attached microbial communities (Fig. S4).Fig. 3. Ratios of normalized gene abundance (GPM) of trace metal transporters in water masses to surface waters for the free-living fraction. Genes are ordered according to gene abundance in surface waters (mean of GPM in SASW, PFSW and ASW, left panel). Red colours indicate that gene abundance is higher in a given water mass as compared to the surface and purple colours indicate that gene abundance is lower in a given water mass. Inserted (*) indicates that the gene abundance in a given water mass is 0. Respective ratios for the particle-attached fraction are presented in Fig. S4

The distinct distribution patterns of Fe, Mn, Cu and Ni transporters could reflect an adaptation of the prokaryotic communities to trace-metal conditions of given water masses. The concentration, supply rate, chemical speciation and bioavailability of trace metals as well as the requirements of specific elements for prokaryotic metabolism or the need to export metals to regulate the intracellular trace metal concentrations are all intrinsically linked features that could lead to our observations [40]. The concentration of trace metals is among the few parameters that can be determined for a large number of environmental samples and it could potentially serve as an indicator for the micro-nutritional state. We examined whether transporter abundances were linked to the concentration of trace metals (Fig. S5) and observed only a few significant correlations (Spearman correlation analyses; Fig. S6). Dissolved Fe concentrations were significantly positively correlated with the transporters fbp (Fe^3+^) (p < 0.01) and feo (Fe^2+^) (p < 0.01), and the lucA iron reductase (p < 0.01). These correlations were absent in the particle-attached fraction where only pvuB (vibrioferrin) was significantly positively correlated with particulate Fe concentrations (Fig. S6). Concentrations of dissolved Cu and Ni were significantly positively correlated with the Cu exporter copB (*p *< 0.01) and the heavy metal efflux system csz, cus, cnr (p < 0.01). While the concentration of trace metals is likely to have an influence on the fine-tuned membrane transporter sensory regulation, the concentrations determined in a given water sample might not be reflecting those to which a microbial community has been exposed over longer time periods.

Our results only partly align with those obtained from metagenomes from the Global Ocean Sampling (GOS), revealing strong correlations between several Fe-related pathways, including transport of Fe^3+^ and Fe ^2+^ and siderophores, with Fe concentrations, the latter obtained by a modelling approach [6]. The GOS expedition provided surface samples from all major ocean basins and numerous coastal sites, likely covering a larger range of environmental conditions, including trace-metal-related properties. These authors further showed that Fe-related pathways were strongly linked to community composition [6], suggesting that Fe-uptake mechanisms are to some extent taxon-specific, an idea that was confirmed at the genome [41–43] and MAG level for transporters of Fe and other trace metals [35, 44, 45]. These taxon-specific traits could contribute to explain our observations of the spatial distribution of diverse trace metal transporters (Fig. 2). We have previously shown that the water masses considered here are inhabited by distinct microbial communities and that individual taxa (ASVs) have strong regressions with distinct trace elements [9]. To further explore this link, we examined the trace metal transporter repertoire of metagenomes assembled genomes (MAGs).

We constructed a total of 556 MAGs (≥ 50% completeness and ≤ 10% potential contamination) from the 42 metagenomes (Table S4). Most MAGs (484) were taxonomically affiliated to Bacteria and 72 MAGs belonged to Archaea (Fig. S7, Fig. S8). The bacterial MAGs mainly belonged to Pseudomonadota (199 MAGs), Bacteroidota (82 MAGs) and Planctomycetota (47 MAGs). MAGs belonging to Pseudomonadota, SAR324, Nitrospinota, Marinisomatota, Gemmatimonadota, Actinomycetota and Chloroflexota had higher coverages in water masses as compared to surface waters, while the opposite trend was observed for Bacteroidota MAGs. The archaeal MAGs belonged to Poseidoniaceae (31 MAGs),* Thalassarchaeaceae* (29 MAGs),* Nitrosopumilaceae* (5 MAGs),* CG-Epi 1* (3 MAGs),* UBA57* (2 MAGs),* ARS21* (1 MAG) and UBA12382 (1 MAG). The diverse MAGs belonging to Poseidoniaceae had higher coverages in surface waters than at depth while MAGs belonging to Thalassarchaeaceae and Nitrosopumilaceae had an inverse distribution (Fig. S7, Fig. S8).

We further focused our analysis on abundant MAGs, defined as those with the highest mean coverages in a given water mass. The top ten abundant MAGs in WW, AAIW, SICW, UCDW, LCDW and NADW resulted in a total of 36 MAGs in the free-living fraction (Fig. 4, Table S5) and in 40 MAGs in the particle-attached fraction (Fig. S9, Table S6). In surface waters (STSW, SASW, PFSW, ASW) a total of 55 and 42 MAGs accounted for the ten most abundant MAGs in the free-living and particle-attached fractions, respectively (Fig. S10, Table S7 and S8). The top ten MAGs accounted for 55–84% of the total mean coverage of all MAGs in a given sample and these MAGs had a completeness of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document} 70% (Table S5-S8). WW and SICW had each a specific assemblage of abundant MAGs, while several MAGs were shared between AAIW and UCDW, and between LCDW and NADW.Fig. 4. Mean coverage of MAGs belonging to the ten most abundant ones in a given water mass (left panel) and corresponding counts of trace metals transport genes (right panel) in the free-living fraction. Water masses are separated by dashed lines. Archaeal MAGs are marked by an asterisk. The MAGs that belong to the top ten in the free-living and particle-attached fractions (Fig. S9) are shown in bold

A few MAGs had high coverages in all water masses, the most pronounced being MAG330_NAC60-12. Many of the MAGs were affiliated with the families Thalassarchaeaceae, Nitrosopumilaceae, Poseidoniaceae, NAC60-12 and Moraxellaceae (Fig. 4) known to be abundant in the deep ocean. We also observed a few water-mass specific MAGs belonging to families other than the above-mentioned ones. Several MAGs belonging to Porticoccaceae had high coverage in WW and in one sample from UCDW located above the shallow Kerguelen plateau (Station 68, overall depth 527 m). MAGs belonging to MedAcidi-G1 (Acidimicrobiales) accounted for the most abundant ones in SICW. Sphingomonadaceae, Pseudomonadaceae, Moraxellaceae,* Alteromonadaceae* and Burkholderiaceae MAGs were particularly abundant in LCDW and NADW.

The trace metal transporter repertoires of the top ten MAGs of distinct water masses did not reveal specific features, except for the MAGs abundant in LCDW and NADW (Fig. 4). These latter MAGs contained a large repertoire and high gene counts of Fe, Cu and Ni transporters. The identified genes included transport of Ni (cbiK, nik) into the cell as well as resistance and efflux genes of Cu (cop, pco, czn, cus, cnr) and Ni (rcnB). These MAGs further harboured a high number of diverse siderophore transporters (fpv, pir, piu, hat, ltb). The high gene counts of transporters into and out of the cell associated with Fe, Cu and Ni in LCDW and NADW are consistent with our findings on the community level and suggest that trace-metal related characteristics of these water masses could contribute to shaping the niche for these abundant MAGs with potential feedbacks on the concentration and chemical speciation of these elements.

To investigate further these possible interactions in LCDW and NADW we refer to results obtained by a recent study that demonstrated reciprocal links between prokaryotic taxa (ASVs) and trace metals in the same water masses as investigated here [9]. Partial Least Square Regression (PLSR) analysis was used to describe water masses by latent vectors, that are a combination of the spatial distribution of dissolved trace metals, apparent oxygen utilization (AOU) and ASVs, named ‘biogeo’-gradients [9]. We searched in the present data set for the MAGs that were taxonomically close (family or genus level) to the 22 ASVs with strong contributions to the ‘biogeo’-gradients as determined by PLSR (Table S9, Fig. S11). Among others, we identified MAGs belonging to Alteromonadaceae, genus Pseudoalteromonas (MAG64, MAG67, MAG312) and to Burkholderiaceae, genus Aquabacterium (MAG263) and genus Cupriavidus (MAG328) that were taxonomically affiliated to the same genera as ASVs that had strong positive regression coefficients with Cu [9]. Alteromonadaceae MAG67 and MAG312 and Burkholderiaceae MAG263 accounted for the top ten MAGs in NADW and LCDW in the free-living and particle-attached communities (Fig. 4 and Fig. S9). One of the Pseudoalteromonas ASVs was additionally a significant contributor to the microbial ‘biogeo’- gradient identified as a good marker of NADW and LCDW and two Pseudoalteromonas ASVs and one Burkholderiaceae ASV were indicator species for NADW (Table S9) [9]. Taken together, the use of independent approaches allowed us to identify a few taxa prevalent in NADW and LCDW that could result from trace metal niche differentiation with Cu as one possible driver.

These observations raise the question of whether the above-identified taxa have higher metal requirements due to metabolic pathways with distinct trace elements as co-factors [46], which could further affect their needs to regulate intracellular metal concentrations. We therefore explored other functional aspects by screening the top ten MAGs of all water masses against the metabolic database (Metabolic_v4.0). Our observations do not reveal any distinct metabolic profiles for Alteromonadaceae MAG67 and MAG312 and Burkholderiaceae MAG263 for central carbon metabolism, including complex carbon degradation and C1 metabolism, as well as pathways involved in nitrogen and sulphur cycling (Table S10). These observations could suggest that the functional differences present in the trace metal transporter repertoire (Fig. 4, Fig. S11) were not reflected in other metabolic traits or that distinct metabolic features might not have been accessible through the analysis of MAGs. Genomic diversity even among closely related strains is well documented, and has also been reported for members of the Alteromonadaceae. A pangenome of Pseudoalteromonas using marine representatives revealed remarkable genomic diversity with horizontal gene transfer being one key process in the acquisition of genes, and illustrated the presence of a high number of efflux pumps [47]. Alteromonas contains a large number of TonB-dependent transporters, but these were shown to have unequal distribution among sequenced representatives [43] and among strains of a single Alteromonas species [48]. These genomic traits could result from positive selective pressure of transporter genes to support niche specialization in the utilization of different substrates or trace metals.

The results of the present and a previous study [9] highlight the potential reciprocal link between Cu and prokaryotic taxa in NADW and LCDW. Both are old water masses favouring adaptations of microbial communities to the environmental conditions within these physical boundaries. Concentrations of dissolved Cu were higher in NADW and LCDW (~2 nM) as compared to other water masses (< 2 nM) while this was not the case for dissolved Fe, Ni and Mn [9]. This distinct distribution of dissolved Cu paralleled the abundance of the Cu exporter (copB) and the heavy metal efflux system (csz, cus, cnr) (Fig. S6). According to the Irving-Williams series, Cu has among the highest binding affinity to metalloproteins [49, 50]. Control of the intracellular Cu concentration is therefore essential to allow less competitive metals, such as Mn, Fe and Ni, to bind to the proteins for which they are required as co-factors. The inadvertent uptake of Cu requires a defence mechanism that could consist of a high number of Cu efflux and resistance genes [4]. An experimental study revealed a narrow range in the Cu quota of marine bacterial strains grown under highly variable Cu concentrations, indicating a strong control of intracellular concentrations maintained by Cu homeostasis [51].

Our observations from microbial membrane transporters point to adaptations to the trace metal environment in specific water masses of the deep ocean. The diverse repertoire and high abundance of genes for the export and resistance of certain trace metals, in particular Cu, in microbial communities and abundant taxa inhabiting NADW and LCDW point to the need for strong regulation of cellular metal concentrations exerted by homeostasis. The respective processes could influence the trace metal properties in the surrounding environment, a feedback mechanism that has thus far not adequately been considered. Prokaryotes that harbour efficient control mechanisms by balancing the uptake and export of metals could be favoured in their metabolic activity that requires metals as co-factors. Some of the abundant taxa identified in the present study were shown to be active members of the microbial communities in the deep ocean [52, 53]. The trace metal adaptation could render these microbial taxa key players in biogeochemical cycles of major elements and in turn influence the properties of micronutrients in distinct water masses of the ocean.

Supplementary Information

Supplementary Material 1. Fig. S1. Normalized abundance of genes related to the transport of Fe (Fe3+, Fe2+) (A), heme and siderophores (B), Mn (C), Ni (D), Cu (E), and Ni/Cu (F) in particle-attached communities. Normalized gene abundances are given in genes per kilobase million (GPM). Note the difference in scale for each plot. Fig. S2. Normalized gene abundances of trace metal transporters in the free-living fraction. Normalized gene abundances are given in genes per kilobase million (GPM). Fig. S3. Normalized gene abundances of trace metal transporters in the particle-attached fraction. Normalized gene abundances are given in genes per kilobase million (GPM). Fig. S4. Ratios of normalized gene abundance (GPM) of trace metal transporters in water masses to surface waters for the particle-attached fraction. Genes are ordered according to gene abundance in surface waters (mean of all surface samples except for STSW, left panel). Red colors indicate that gene abundance is higher in a given water mass as compared to the surface and purple colors indicate that gene abundance is lower in a given water mass. Inserted (*) indicates that the gene abundance in a given water mass is 0. Fig. S5. A. Concentration of dissolved Fe (in nM), dissolved Cu (in nM), dissolved Mn (in nM) and dissolved Ni (in nM) in different surface waters and water masses. B. Concentrations of particulate Fe (in nM), particulate Cu (in nM), particulate Mn (in nM) and particulate Ni (in nM) in different surface waters and water masses. Fig. S6. Correlation matrix showing Spearman's correlation coefficients of trace metal transport gene abundance in the free-living fraction and dissolved trace metal concentrations (A) and trace metal transport gene abundance in the particle-attached fraction and particulate trace metal concentrations (B). p value significance level: * <0.05, ** <0.01. Fig. S7. Phylogenetic tree of all archaeal species-level MAGs. The heatmap shows the mean coverage of each MAG in each free-living (A) and particle-attached (B) sample. Color intensity is determined by log transformed mean coverage of each MAG. Fig. S8. Phylogenetic tree of all bacterial species-level MAGs. The heatmap shows the mean coverage of each MAG in each free-living (A) and particle-attached (B) sample. Color intensity is determined by log transformed mean coverage of each MAG. Fig. S9. Mean coverage of MAGs belonging to the ten most abundant ones in a given water mass in the particle-attached fraction (left panel) and corresponding counts of trace metals transport genes (right panel). Water masses are separated by dashed lines. Archaeal MAGs are marked by an asterisk. The MAGs that belong to the top ten in the free-living and particle-attached fractions are shown in bold. Fig. S10. Mean coverage of MAGs belonging to the ten most abundant ones across surface waters in the free-living (A) and particle-attached fractions (B) and corresponding counts of trace metals transport genes. Archaeal MAGs are marked by an asterisk. The MAGs that belong to the top ten in the free-living and particle-attached fractions fraction are shown in bold. Fig. S11. Mean coverage in the free-living fraction and corresponding trace metal transport gene counts of MAGs belonging to the same families as the 22 ASVs identified in Zhang et al [1]. Table S1. Basic sample information.

Supplementary Material 2. Table S2. Physical properties of water masses considered in the present study.

Supplementary Material 3. Table S3. List of genes associated with trace metal transport detected in the present study.

Supplementary Material 4. Table S4. Statistics and taxonomy assignment of the 556 metagenome-assembled genomes (MAGs) (completeness ≥50%, contamination ≤10%).

Supplementary Material 5. Table S5. Taxa classification of the top 10 MAGs in water masses in the free-living fraction

Supplementary Material 6. Table S6. Taxa classification of the top 10 MAGs in water masses in the particle-attached fraction.

Supplementary Material 7. Table S7. Taxa classification of the top 10 MAGs in surface water in the free-living fraction.

Supplementary Material 8. Table S8. Taxa classification of the top 10 MAGs in surface water in the particle-attached fraction.

Supplementary Material 9. Table S9. Taxa classification of the MAGs belonging to the same family as 22 ASVs shown by Zhang et al [1]

Supplementary Material 10. Table S10. Metabolic profiles of the top ten MAGs of all water masses in free living fraction after screening against the metabolic database (Metabolic_v4.0). Alteromonadaceae MAG67 and MAG312 and Burkholderiaceae MAG263 are highlighted in yellow.