Analysis of Metagenomic raw reads
We have generated metagenomic datasets from a Black Sea depth profile. Samples were collected along the Bulgarian coast at two stations (Fig. S1A). Sampling depth was guided by the physicochemical measurements (Fig. S1B) to cover representative temperature, oxygen, and chlorophyll-a values (Additional File 1). Thus, for St. 307, with the maximum depth of 1100 m, samples came from the near-surface at 5 m depth, the deep chlorophyll maximum (DCM) at 30 m, a sample from the redoxcline/pycnocline at 150 m and finally a sample at 750 m depth, corresponding to the euxinic water layer. Additionally, we collected a single near-surface sample (5 m) closer to the shore at station 301 with a maximum depth of 22.5 m. For each depth, we performed a first unassembled read analysis to obtain a rough taxonomic profile based on metagenomic 16S rRNA gene fragments against the SILVA database [11] (Fig. 1A) and the main predicted metabolic functions assessed by the SEED subsystems [12] (Fig. 1B).
The oxic strata, surface and DCM, presented, at this rough level, very similar taxonomic composition (Fig. 1A). Alphaproteobacteria (orders SAR11, SAR116, Rhodobacterales and Rhodospirillales), Gammaproteobacteria (mostly orders SAR86 and Pseudomonadales), and picocyanobacteria (order Synechococcales) were the most abundant groups, representing > 70% of total microbial biomass (assessed by total 16S rRNA classification). It must be highlighted the complete absence of the genus Prochlorococcus in all our Black Sea samples. The predominant subsystems of the oxic layer (Fig. 1B) were, as expected, associated with phototrophic lifestyles such as those from Synechococcales (photosystems/phycobilisomes) or photoheterotrophy with type-1 rhodopsin pumps (typical of SAR11, SAR86 or Flavobacteriales). In addition, ammonia was the preferred N source.
The taxonomic composition changed dramatically as we reached oxygen extinction in the pycnocline (150 m), where various taxa and microbial lifestyles coexisted, with a prevalence of anaerobic N and S related subsystems (Fig. 1B). Marinimicrobia (ca. 30% of 16S rRNA assigned reads) and Gammaproteobacteria (ca. 20%) were the dominant taxa of the redoxcline (Fig. 1A). Chemolithotrophs and anaerobes, such as SUP05 (Ca. Thioglobus spp.), Nitrosopumilaceae (aerobic archaeal ammonia oxidizers), Campylobacterota (dissimilatory nitrate reducer), Marinimicrobia (fermenters and hydrogen metabolizers), Nitrospirota (nitrite oxidizers) (N fixers), Chlorobi (anoxygenic photosynthesizers), Desulfobacterota (sulfate-reducers) and various associated streamlined microbes such as Patescibacteria and Nanoarchaeota appeared here.
The euxinic waters at 750 m showed an increase in fermentation, hydrogen metabolism, anaerobic respiratory reductases or methanogenesis pathways (Fig. 1B). Overall, we observed an increase in sulfate reducers (Desulfobacterota), Dehalococcoidia/Anaerolineae Chloroflexota and a huge diversity of accompanying microbiota providing hydrogen and fermentation by-products that conformed a syntrophic network fueling the sulfate reducers at the redox end. There were representatives from Omnitrophota and Kiritimatiellae (both classified inside Verrucomicrobiota according to SILVA standards [11], although Omnitrophota is classified as a single phylum according to GTDB [13]), Phycisphaerae Planctomycetota, Marinimicrobia, Nanoarchaeota, Patescibacteria, andCloacimonadota. Finally, Halobacterota (Syntrophoarchaeia) and Crenarchaeota (Bathyarchaeia) minor representation (< 2% of total microbial biomass assessed by 16S rRNA) showed that methanogenesis coexisted with sulfate reduction in these euxinic waters if in much more reduce fraction.
MAGs recovered from the different samples
Automated binning followed by manual curation of generated bins allowed the recovery of 359 MAGs with > 50% completeness and < 5% contamination. Detailed stats of these MAGs are described in Table 1 (MAGs from oxic samples) and Table 2 (redoxcline/anoxic MAGs) and in individual MAG detail in Additional File 2. Genomes are also showed in an estimated genome size versus GC content plot in Fig. S2. The taxonomic nomenclature used in this work was based on GTDB (ref). To estimate the binning efficiency, we mapped the reads of each metagenome against the MAGs obtained for each sample at the thresholds of > 95% of identity and > 50 bp of alignment lengths. The percentages of reads mapped to the MAGs varied between samples, being maximum in the redoxcline (50%) and minimum in the euxinic sample (33%). With regard to the oxic samples, the MAG recovery efficiency was ca. 50% of the total reads mapped with the MAGs from the coastal epipelagic sample (BS301-5 m), 42% for the off-shore epipelagic sample (BS307-5 m) and 34% for the DCM sample (BS307-30 m). These recovery values are in the range of what was previously obtained for other aquatic environments [14].
Table 1
Summary statistics and features of Black Sea MAGs retrieved from 5 and 30 m samples.
Phylum/Division | Taxonomic affiliation of MAGs (GTDB, Referenced groups) | nº of MAGs | Range Estimated Genome size (Mp) | Range GC (%) | Range median intergenic spacer (bp) | Av. Compl. (%) | Av. Cont. (%) |
α-Proteobacteria | g_Planktomarina (3), g_Puniceispirillum (6), g_Reyranella (1), f_Rhodobacteraceae (7), o_Pelagibacterales (8), o__Parvibaculales (5), f_Puniceispirillaceae (6), f_Nisaeaceae (1), o_Rickettsiales (3), o_Rhizobiales (2), o_Rhodospirillales_A (1), c_Alphaproteobacteria (3) | 52 | 1-7.9 | 28–66 | 2–65 | 75.55 | 1.19 |
ɣ-Proteobacteria | g_Luminiphilus (11), g_Litoricola (3), g_Nevskia (1), f_Methylophilaceae (5), f_Porticoccaceae (1), f_Pseudohongiellaceae (5), f_Shewanellaceae (1), o_Burkholderiales (2), o_SAR86 (11) o_Pseudomonadales (5), c_Gammaproteobacteria (3) | 51 | 1-4.1 | 31–69 | 1–86 | 70.34 | 1.29 |
Bacteroidota | f_Cryomorphaceae (4), f_Flavobacteriaceae (23), o_Flavobacteriales (8), f_Balneolaceae (2), f_Crocinitomicaceae (2), c_Bacteroidia (1) | 44 | 1.19–2.57 | 28–58 | 4–43 | 73.91 | 0.94 |
Thermoplasmatota | f_Poseidoniaceae (6), f_Thalassoarchaeaceae (1), g_Poseidonia (6) | 13 | 1.82–2.35 | 37–58 | 25–38 | 80.67 | 0.30 |
Actinobacteriota | o_Nanopelagicales (1), g_Aquiluna (1), o_Actinomarinales (5), f_Ilumatobacteraceae (4), c_Thermoleophilia (1) | 13 | 1.23–2.22 | 32–71 | 2–23 | 76.79 | 1.45 |
Cyanobacteria | g_Synechococcus_C (10) | 10 | 1.8–2.23 | 55–63 | 25–36 | 79.59 | 2.37 |
Planctomycetota | f_Planctomycetaceae (1), o_Pirellulales; (4), p_Planctomycetota (3), g_Rubripirellula (1) | 9 | 3-6.4 | 49–72 | 45–132 | 87.14 | 0.95 |
Verrucomicrobiota | o_Pedosphaerales (2), o_Opitutales (1), f_Puniceicoccaceae (4), f__Akkermansiaceae (1) | 8 | 2-4.8 | 42–60 | 23–74 | 83.51 | 4.21 |
Marinisomatota | g_Marinisoma (3) | 3 | 0.8–0.93 | 31–32 | 2–3 | 55.31 | 0.36 |
Margulisbacteria | c_ZB3 (1) | 1 | 1.71 | 42.5 | 10 | 67.53 | 0 |
Parenthesis () indicate the average value of each field in case of range values and number of MAGs in bold for each taxonomic affiliation. Taxonomic classification follows GTDB criteria. Marinisomatota includes former Marinimicrobia. Thermoplasmatota includes former Euryarchaeota group. d_:Domain, p_:Phylum, c_:Class, o_:Order, f_:family, g_:Genus, s_: Species. Av. (Average), Compl. (Completeness), Cont. (Contamination). |
Table 2
Summary statistics and features of Black Sea 150 and 750 m retrieved MAGs.
Phylum/Division | Taxonomic affiliation of MAGs (GTDB, Referenced groups) | nº of MAGs | Range Estimated Genome size (Mp) | Range GC (%) | Range median intergenic spacer (bp) | Av. Compl. (%) | Av. Cont. (%) |
Patescibacteria | c_Microgenomatia (3), c_Paceibacteria, c_ABY1 (4), o_Portnoybacterales (2), o_Shapirobacterales (1), o_Paceibacterales (3), o_Paceibacteria (2) | 15 | 0.6–1.69 | 30–41 | 17–62 | 58.09 | 1.24 |
Omnitrophota | p_Omnitrophota;c_koll11 (13), f_Omnitrophaceae_A (1) | 15 | 0.84–3.78 | 35–47 | 7–84 | 70.45 | 1.69 |
Planctomycetota | c_Brocadiae (1), c_Phycisphaerae (7), o_Pirellulales (2), o_Phycisphaerales (1), f_Pirellulaceae (1), p_Planctomycetota (2) | 14 | 1.92–14.18 | 43–71 | 22–110 | 72.33 | 1.78 |
Desulfobacterota | o_Desulfatiglandales (9), o_Desulfobacterales (4), g_Desulfobacula (1) | 14 | 1.98–8.3 | 40–50 | 53–114 | 65.95 | 1.69 |
Marinisomatota | o_Marinisomatales (5), c_Marinisomatia (2), p_Marinisomatota (6) | 13 | 1.6–4.4 | 34–44 | 10–62 | 81.18 | 1.20 |
Chloroflexota | c_Anaerolineae (3), c_Dehalococcoidia (3), o_Anaerolineales (4), o_Dehalococcoidales (3) | 13 | 1.45–5.78 | 48–63 | 32–89 | 71.59 | 2.14 |
α-Proteobacteria | o_Rhodospirillales_A (6), f_Magnetospiraceae (1), c_Alphaproteobacteria (4) | 11 | 2.59–4.72 | 53–64 | 24–62 | 73.51 | 0.98 |
Nanoarchaeota | o_Woesearchaeales (5), o_Pacearchaeales (1), c_Nanoarchaeia (3) | 9 | 0.8–1.6 | 28–36 | 23–70 | 64.72 | 1.50 |
Bacteroidota | s_Chlorobium_A phaeobacteroides (1), c_Ignavibacteria (1), o_Bacteroidales (5), c_Bacteroidia (1) | 8 | 2.46-5 (3.5) | 33–49 | 40–81 | 85.09 | 2.28 |
ɣ-Proteobacteria | g_Thioglobus_A (2), g_Methylobacter_A (1),g_Acidovorax_D (1), c_Gammaproteobacteria (3) | 7 | 1.18–5.59 | 37–64 | 14–67 | 75.25 | 2.11 |
Crenarchaeota | c_Bathyarchaeia (3), g_Nitrosopumilus (3) | 6 | 1.39–2.92 | 32–58 | 37–85 | 65.18 | 0.88 |
Actinobacteriota | o_Microtrichales;f_MedAcidi-G1 (2), p_Actinobateriota;c_UBA1414 (2) | 4 | 1.42–2.33 | 31–64 | 27–68 | 76.65 | 1.09 |
Verrucomicrobiota | c_Kiritimatiellae (2), o_Kiritimatiellales (2) | 4 | 2-3.6 | 53–60 | 45–56 | 74.04 | 2.53 |
Aenigmarchaeota | c_Aenigmarchaeia (3), o_Aenigmarchaeales (1) | 4 | 0.71–1.29 | 36–45 | 37–59 | 55.9 | 1.86 |
Cloacimonadota | c_Cloacimonadia (2), o_Cloacimonadales (1) | 3 | 1.2–2.91 | 31–36 | 15–50 | 74.34 | 0.11 |
Campylobacterota | f_Arcobacteraceae (1), g_Sulfurimonas (1) | 2 | 1.14–1.81 | 30–35 | 9–15 | 60.01 | 1.04 |
Nitrospirota | c_Thermodesulfovibrionia (2) | 2 | 1.85–3.18 | 44–46 | 53–68 | 68.81 | 0.94 |
KSB1 | p_AABM5-125-24 (1), p_KSB1 (1) | 2 | 4.05–5.13 | 37–46 | 94–144 | 73.31 | 2.2 |
Myxococcota | p_Myxococcota (1) | 1 | 4.46 | 63.7 | 37 | 60.65 | 0.84 |
Bdellovibrionota | f_Bacteriovoracaceae (1) | 1 | 4.78 | 37.2 | 40 | 92.41 | 3.63 |
Spirochaetota | c_Spirochaetia (1) | 1 | 2.92 | 40.9 | 52 | 67.32 | 1.89 |
Halobacterota | d_Archaea;p_Halobacterota;c_Syntrophoarchaeia;o_ANME-1 (1) | 1 | 1.81 | 43.1 | 52 | 55.52 | 0.65 |
Nitrospinota | f_Nitrospinaceae (1) | 1 | 3.08 | 46 | 69 | 93.96 | 2.56 |
Delongbacteria | p_Delongbacteria (1) | 1 | 3.002 | 54 | 45 | 64.03 | 1.1 |
SAR324 | c_SAR324;o_SAR324 (1) | 1 | 2.96 | 41.7 | 53 | 82.86 | 0 |
SM23 | d_Bacteria;p_AABM5-125-24 (1) | 1 | 3.58 | 45.1 | 131 | 73.63 | 3.3 |
Acidobacteriota | f_Aminicenantaceae (1) | 1 | 3.0955127 | 40.5 | 68 | 89.52 | 4.27 |
Parenthesis () indicate the average value of each field in case of range values and number of MAGs in bold for each taxonomic affiliation. Taxonomic classification follows GTDB criteria. Marinisomatota includes former Marinimicrobia. Halobacterota includes former Euryarchaeota methanogens group. Campylobacterota includes former Epsilonproteobacteria. d_:Domain, p_:Phylum, c_:Class, o_:Order, f_:family, g_:Genus, s_: Species. Av. (Average), Compl. (Completeness), Cont. (Contamination). |
A Distance-based redundancy analysis (dbRDA) was conducted to statistically assess the main differences between different Black Sea strata (Fig. 2). To make such analysis we used the physicochemical measurements (Additional File 1), the metabolic abundance of each SEED subsystem (Fig. 1D) and the relative abundance of each microbial species retrieved as MAG and assessed with reads per Kb of Genome per Gb of metagenome (RPKGs), showed in Additional File 3.
MAGs from the epipelagic and DCM oxic strata
As expected, the statistical analysis conducted with the dbRDA (Fig. 2) grouped together the environmental variables of Temperature (T), dissolved oxygen (DO) or ammonia with photo(hetero)trophic lifestyles from well-known marine and brackish groups such as SAR11, SAR116 and Rhodospirillales (Alphaproteobacteria), SAR86 (Gammaproteobacteria), Thermoplasmatota (former marine group II Euryarchaeota), Synechococcales (Synechococcus) and Actinomarinales (Actinobacteria). As noted above, we must highlight a complete absence of Prochlorococcus spp., contrasted with a high abundance of various Synechococcus MAGs that affiliated with the marine clades I, III, IV, VI and WPC1 including isolates (KORDI-49, BL107, CC9902, WH 8016, WH 7805/7803, WH 8103/8102) [15]. The main Actinobacteria MAGs retrieved presented relatively small genome sizes (1.2–2.2 Mb), among which we must highlight the presence of 5 novel Actinomarinales (BS301-5m-G7, BS307-5m-G2, BS30m-G2/G3/G4) and a group of Ilumatobacteraceae genomes related to Caspian MAGs (Casp-actino5) [10]. The major SAR11 Alphaproteobacterial MAGs were eight novel Pelagibacterales that affiliated with the recently described groups Ia.1, IIaB/1and IIIa [16]. Remarkably, we obtained three novel MAGs from the order Rickettsiales. Another relevant Alphaproteobacteria clade from which we obtained MAGs was SAR116, with six MAGs affiliated to Puniceispirillum genus and five more were only classified as representatives of the family Puniceispirillaceae. A remarkable family that has shown a high abundance in Black Sea oxic waters is Flavobacteriaceae (23 MAGs), a group that was commonly detected in the Mediterranean [9] and the Baltic Seas [17]. In fact, various MAGs were related at GTDB genus level with MED-G11, MED-G14 MAGs and at the species level (ANI > 95%) with MED-G20 Mediterranean Sea MAGs. Two MAGs also showed their closest relatives at the GTDB family level with Baltic Sea MAGs BACL11 and at the species level with BACL21. We also found five representatives from the clade OM43 (family Methylophilaceae) affiliating at the genus level to BACL14 Baltic Sea MAGs. Eleven MAGs belonged to the cosmopolitan Gammaproteobacteria SAR86, so far only classified at this order level. Other Gammaproteobacteria that co-occurred in these samples were MAGs with similarity to Luminiphilus (11 MAGs) and Litoricola (3 MAGs) genera. Another relevant taxon from marine systems was the former marine group-II Euryarchaeota (Thermoplasmatota according to GTDB taxonomy). We retrieved six genomes affiliating to the family Poseidoniaceae and other six to the genus Poseidonia. Only one genome was obtained affiliating to Thalassoarchaeaceae. Finally, three ultra-small (1 Mb of estimated genome size) Marinimicrobia MAGs were obtained from oxic metagenomes, which so far are classified by the GTDB as genus Marinisoma.
Black Sea pycnocline MAGS
The redoxcline of the Black Sea presented the most metabolically diverse set of pathways among all analyzed samples (Fig. 2A). The main environmental variables that statistically grouped with the pycnocline were total nitrogen (TN) and nitrate, which were clearly associated with the different N cycle pathways that completed its biogeochemical cycle in this layer. The highest abundance of N pathways corresponded with denitrification (nitrogen gas as the final product), nitrate/nitrite ammonification and dissimilatory nitrate reduction (with ammonium as the final product), but the N cycle was also completed with ammonia oxidation and N fixation pathways detected both in total reads and MAGs (see below). Nonetheless, various other metabolisms coexisted in this thin layer where oxygen is extinguished. We noticed the presence of anoxygenic photosynthesis, exemplified by MAG BS150m-G13 showing > 99% of ANI with Chlorobium phaeobacteroides, a green sulfur bacterium (GSB) originally isolated from the Black Sea [18] (GCA_000020545.1), that was undergoing a nearly monoclonal bloom (Fig. S3).
Chemoautotrophy was observed in Thioglobus sp. BS150m-G29 and G33 MAGs, both of which are novel representatives of the SUP05 clade which performs a wide variety of metabolisms including S oxidation and C fixation and with only 80% of ANI with its closest relative (Ca. Thioglobus autotrophicus EF1) [19]. It appears that this is a case of a single species (recruiting at > 95% of nucleotide identity) abundant (> 70 RPKG, Fig. S4) in the Black Sea redoxcline. Methane oxidation (Methylobacter sp. BS150m-G31) and ammonia oxidation were also key metabolisms observed in this layer (Nitrosopumilus spp. BS150m-G38/39/40). Nitrite oxidation was detected in Nitrospinaceae BS150m-G45.
Denitrification was frequently detected among pycnocline MAGs, although complete denitrification including the last step involving conversion of nitrous oxide into nitrogen gas (nosZ gene) was seen only in five MAGs (Marinimicrobia BS150m-G46/G47/G71, unclassified Alphaproteobacteria BS150m-G7/G9, Rhodospirillales BS150m-G4/G10, Sulfurimonas sp. BS150m-G26 and unclassified Gammaproteobacteria BS150m-G28/30/32). Dissimilatory nitrate reduction to ammonium (nrfAH genes) was far more restricted and found in Marinimicrobia BS150m-G46, Campylobacterota (Sulfurimonas sp. BS150m-G26) or Bacteroidales BS150m-G15. Nitrate reduction through nirB gene was much more widely detected including in all Alphaproteobacteria MAGs and various Gammaproteobacteria members. N fixation (nifDK dinitrogenase subunits) was detected in only two MAGs (Chlorobium phaeobacteroides and Nitrospirota, BS150m-G55/G56 respectively).
Dissimilatory sulfate reduction and oxidation (dsrAB genes) showed up already in this sample in various genomes such as Desulfobacterota, Nitrospirota MAGs, Planctomycetota (Pirellulaceae BS150m-G36) (already mentioned above), Chloroflexota (Anaerolineales BS150m-G18), Alphaproteobacteria (Rhodospirillales BS150m-G3/G4/10/G11), Chlorobium phaeobacteroides MAG and Gammaproteobacteria (Ca. Thioglobus and Gammaproteobacteria BS150m-G28 MAGs). It must be noted that, among the main features of this habitat, there was the simultaneous activity of sulfate-reducing and sulfide-oxidizing microbes forming part of the same ecological niche, a process known as cryptic sulfur cycle [20]. However, low O2 concentrations (0.87 mg/L) and low ratio (0.16) of peroxidase/recA genes (1.5 in oxic datasets) clearly demonstrate the microaerophilic/anoxic nature of this habitat.
We also compared our pycnocline dataset with previously available metagenomes from the redoxcline from Cariaco Basin (Venezuela) [21] (Fig. S5). Overall, it seems that Marinisomatota/Marinimicrobia and Gammaproteobacteria chemolithotrophic groups are the most abundant key players of these two marine redoxclines, accounting for more than 50% of total microbial biomass (Fig. S5A). However, it must be noted that only a few species retrieved as MAGs from the Black Sea were detected in such a similar habitat (Fig. S5C). Among them, two chemolithotrophic Gammaproteobacterial representatives (Ca. Thioglobus and a novel species BS150m-G30 classified only at the order level as o__GCA-2400775 by GTDB), sulfate reducers (Desulfatiglandales), denitrifying and hydrogen-producing Marinimicrobia (three species) and one Actinobacteria (a novel species from the marine MedAcidi-G1 group). Apart from their metabolic potential fitting with microbial lifestyles from pycnocline layers, these species could play key roles in other marine redoxclines and oxygen minimum zones (OMZs), as their detection in two largely separated biomes with different salinities (ca. 2% in the Black Sea and 3.5% in the Cariaco Basin) indicate a widespread distribution in oxygen-depleted marine niches.
Euxinic Black Sea MAGs and the “microbial dark matter”
The main environmental variables grouping with the mesopelagic sample of 750 m were PO4 and Si, both of which are solubilized in anoxic layers and diffuse from the sediment layer. Salinity also increased up to 2.2% in these euxinic waters. There is the expected predominance of sulfate reduction pathways, as carried out by Desulfobacterota MAG representatives (Desulfatiglandales BS750m-G47-G51 and BS750m-G54/G56, Desulfobacterales BS750m-G52/G53/G55), which perform the dissimilatory sulfate reduction pathway (dsr genes).
Methanogenesis was very diluted in these waters but still detectable, albeit we found the complete pathway in MAG Ca. Syntrophoarchaeum BS750m-G82 and most of the genes except for the key enzyme, Methyl-coenzyme M reductase (mcr) in Bathyarchaeota BS750m-G27/G28 MAGs as well. The latter showed various mixed-acid fermentation pathways including the formation of H2 and CO2 (via the formate hydrogenlyase), formate (pyruvate-formate lyase), alcohol (alcohol dehydrogenase) or lactate (lactate dehydrogenase). It must be noted the potential capability of performing reverse methanogenesis, or anaerobic methane oxidation (ANME) by the abovementioned archaeon (MAG Ca. Syntrophoarchaeum BS750-G82). Heterodisulfide reductase genes (HdrABC), which are involved in the last step of methanogenesis by reducing CoB-CoM heterodisulfide, were detected in Bathyarchaeota and Syntrophoarchaeum MAGs. However, these genes were also found in Cloacimonadota, candidate division KSB1, Ca. Aminicenantes, Omnitrophica, Desulfobacterota, Planctomycetes, andChloroflexi MAGs as well as in the unassembled reads (being completely absent from oxic datasets), suggesting that these electron transfer complexes are not exclusive of methanogens. As seen by the dbRDA, we also noted a global predominance of mixed-acid fermentation pathways (with ethanol, lactate, acetate, formate or CO2/H2 as products) and hydrogen uptake hydrogenases that couple with sulfate, fumarate, CO2 or nitrate reduction, thus conforming a complex syntrophic network of microbes. This networking of syntrophic microbes (considered here as interspecies H transfer) includes the abovementioned uncultured taxa plus accompanying streamlined members of the “microbial dark matter” such as Omnitrophota, Patescibacteria (Ca. Microgenomates, Portnoybacteria, Paceibacteria) or Nanoarchaeota (Ca. Aenigmarchaeota, Woesearchaeota, Pacearchaeota), groups from which we also obtained MAGs (see Table 2). Various types of hydrogenases and hydrogen metabolism pathways grouped with the 750 m mesopelagic sample in the dbRDA plot (Fig. 2) and were found in the vast majority of microbes inhabiting this sulfide enriched waters, including NAD-reducing bidirectional (hox genes) and uptake hydrogenases (hup genes), NiFe (hyp genes) and FeFe (hym genes) hydrogenases, Coenzyme F420-reducing hydrogenases or carbon monoxide induced hydrogenases (CooHL genes), all of which showed the highest gene/ recA ratios (from 0.2 in hym genes to 1–2 for hyp and hoxF) in euxinic waters.
It was remarkable the presence of two Actinobacteria MAGs (BS750m-G1/G2) in these sulfide-rich waters. These yet unclassified members have their highest resemblance with MAGs retrieved from groundwater aquifers (Actinobacteria bacterium CG08_land_8_20_14_0_20_35_9, classified as UBA1414 by GTDB) and have very small GC content (31–34%) and predicted genome sizes (ca. 1.4–1.6 Mb). Their genomes presented various mixed-acid fermentative pathways associated with the production of ethanol (alcohol dehydrogenase), lactate (lactate dehydrogenase), formate (pyruvate-formate lyase) and H/CO2 (formate hydrogen lyase). They also showed an active hydrogen metabolism with various NiFe hydrogenases including Coenzyme F420-reducing hydrogenase, hyp genes and HyaA COG1740 355 Ni-Fe-hydrogenase I. Another remarkable group of microbes was Omnitrophota, from which we obtained 15 MAGs with variable estimated genomes sizes (from 1 to 3 Mb). For instance, the most abundant MAG retrieved from our samples (BS750m-G77) presented a small predicted genome size (ca. 1.2 Mb) and was an obligate fermenter (mainly producing ethanol, H2/CO2 and lactate). Another group of streamlined members of the microbial dark matter were Aenigmarchaeota (BS750m-G24/36/81/83/) and Nanoarchaeota (BS750m-G11/13/70) MAGs, which had estimated genome sizes of 1-1.5 Mb. Among their metabolic potential, they were also mixed-acid fermenters, including lactate or H2/CO2 as fermentation by-products, which would fuel the sulfate reducers, conforming a syntrophic network with the rest of mixed-acid fermenters. Finally, another set of microbes of small genome sizes (0.6–1.6 Mb) were Patescibacteria (former Candidate Phyla Radiation). We must highlight the presence of the protein VirB4, associated with type IV secretion systems that work as injectors into host cells [22], in Ca. Microgenomates BS750m-G73/74, Ca. Paceibacteria BS750m-G71/75 and Ca. Portnoybacteria bacterium BS750m-G76. These proteins were unique for these microbes in the entire euxinic waters, which suggests a parasitic lifestyle from which these Patescibacteria could translocate nutrients, proteins, and DNA from or to a putative host [22].
Similarities between Black Sea datasets assessed by read and recruitment analysis
To assess the representativity of our samples we also compared the reads between our Black Sea datasets and those from a former sampling campaign [5, 8], deposited into the NCBI under bioproject PRJNA649215, observing a clear 16S rRNA taxonomy and read clusterization between samples (Fig. S6). Among all of the MAGs retrieved from this work, we selected the 30 most abundant MAGs (> 10 RPKGs in any of the recruited samples) from the oxic, redoxcline and anoxic waters and recruited them at > 95% of identity (species level) on all metagenomes (Fig. 3). The rest of our MAGs abundance among all datasets is shown in detail in Additional File 3. As expected, emblematic key players of the oxic waters harboring a phototrophic/photoheterotrophic lifestyle such as Ca. Pelagibacter, Ca. Actinomarina, SAR86, Synechococcus or Flavobacteriaceae were detected at high numbers in all oxic metagenomic datasets. Next, we also showed the main ecological drivers of the redoxcline layer, which included chemolithotrophic S oxidizers and C fixers such as Ca. Thioglobus and dissimilatory nitrate reducers such as Sulfurimonas, both of which were recently analyzed members by previous publication [8]. The redoxcline also showed some other ecologically relevant nitrate reducers such as Bacteroidales MAGs, ammonia oxidizers such as Nitrosopumilus spp, sulfate reducers Desulfatiglandales and Desulfococcales novel species and several Marinimicrobia representatives which are specialized in the H metabolism, denitrification and mixed-acid fermentation. Finally, another set of MAGs were detected among all euxinic strata. These included various other sulfate-reducers, their associated microbiota performing mixed-acid fermentations and H metabolism in syntrophism (Cloacimonadetes, Woesearchaeales, Ca. Aminicenantes) and the only MAG able to perform methanogenesis and ANME, a Syntrophoarchaeum that showed a remarkable abundance in 1000 and 2000 m metagenomic datasets, suggesting that methane metabolisms indeed coexist with the sulfate reduction and all associated microbial fermenters and H2 scavengers in a complex syntrophic network.