Metagenomic analysis reveals wide distribution of phototrophic bacteria in hydrothermal vents on the ultraslow-spreading Southwest Indian Ridge

Introduction

Microorganisms have evolved three major mechanisms to exploit sunlight for driving metabolism (Lu et al. 2012; Thiel et al. 2018). The first is referred to as chlorophototrophy because it is based on chlorophyll or bacteriochlorophyll. Absorption of 350–1100 nm sunlight generates electrochemical ion potential and synthesizes highly energetic reducing molecules (Saer and Blankenship 2017; Thiel et al. 2018). These energy sources are subsequently used to drive ATP biosynthesis, molecular transport, flagella movement and photosynthesis. Chlorophototrophic bacteria are affiliated with seven phyla: Cyanobacteria, Chlorobi, Chloroflexi, Gemmatimonadetes, Acidobacteria, Firmicutes, and Proteobacteria (Thiel et al. 2018). Within the Proteobacteria, purple non-sulfur bacteria are represented by members of the Alphaproteobacteria (Acetobacteraceae, Bradyrhizobiaceae, Hyphomicrobiaceae, Rhodobiaceae, Rhodobacteraceae, Rhodospirillaceae, Sphingomonadaceae) and Betaproteobacteria (Rhodocyclaceae and Comamonadaceae). Gammaproteobacteria, encompassing Chromatiaceae and Ectothiorhodospiraceae, forms the group of purple sulfur bacteria, which produce internal and external sulfur granules, respectively, and show differences in their internal membrane structure. The second type of phototrophy relies on retinal photoreceptors and is known as retinalophototrophy. Retinals in rhodopsin and proteorhodopsin (PR) absorb blue or green light to drive transmembrane transport of protons, sodium ions or chloride ions, creating an electrochemical ion potential for energy generation (Gómez-Consarnau et al. 2007, Pinhassi et al. 2016). The phototrophic function of PR in bacteria was first demonstrated in flavobacteria (Gómez-Consarnau et al. 2007), and metagenomic analysis has identified genes encoding the proton-pump PR in Actinobacteria, Alphaproteobacteria, Bacteroidetes, Cyanobacteria, Firmicutes, Gammaproteobacteria and Planctomycetes (Gómez-Consarnau et al. 2007, 2019; Pinhassi et al. 2016; Sabehi et al. 2005). The third mechanism of sunlight exploitation, photosensitization phototrophy, derives energy from mineral photocatalysis by nonphototrophic bacteria such as chemoautotrophic Acidithiobacillus ferrooxidans and heterotrophic Alcaligenes faecalis found in soil (Lu et al. 2012) and the heterotrophic marine bacterium Idiomarina sp. (Ma et al. 2020). Semiconductive minerals absorb visible light and provide photoelectrons for bacteria, which are used as a complementary energy source (Ramprakash and Incharoensakdi 2020; Sakimoto et al. 2016; Wei et al. 2018; Zhang et al. 2018).

Sunlight is absorbed by seawater, making the deep sea a dark, cold, and oligotrophic environment. However, hot fluids are ejected from the seafloor in deep-sea hydrothermal vent fields, and these provide a variety of chemicals generated through water–rock reactions. Chemoautotrophic microorganisms use these chemicals as nutrients for primary production through chemosynthesis to support autotrophy and create hydrothermal vent ecosystems (Dick 2019). Remarkably, black chimneys with temperatures of up to 360 ℃ emit visible and infrared light (Van Dover et al. 1996; White et al. 2002). It has been hypothesized that this geothermal light facilitated evolution of photosynthesis, independent of solar energy, in hydrothermal vent ecosystems (Nisbet et al. 1995). An aerobic, phototrophic alphaproteobacterial strain and a strictly anaerobic, photosynthetic green sulfur bacterial strain were isolated from black smoker plume waters from the Juan de Fuca Ridge and the East Pacific Rise, respectively (Beatty et al. 2005; Yurkov et al. 1999). These results support the hypothesis that, along with chemosynthesis, vent light may also provide metabolic energy for phototrophic growth as seen in ecosystems dependent on solar energy (Beatty et al. 2005; Martin et al. 2008; Yurkov et al. 1999). Recently, Ma et al. (2020) isolated the cadmium-resistant stain Idiomarina sp. OT37-5b from sediments of a deep-sea hydrothermal vent in the Okinawa Trough. This strain is capable of forming CdS semiconductor nanoparticles on the cell surface. These particles might absorb visible light and generate photoelectrons that boost cell growth. It is unknown how broadly distributed the various types of phototrophic bacteria are in hydrothermal vents. Therefore, we carried out a systematic comparative metagenomic analysis of samples collected at a hydrothermal vent in the Longqi field on the ultraslow-spreading Southwest Indian Ridge (SWIR). We found members of the three phototrophic taxonomic groups and identified genes functioning in both chlorophototrophy and retinalophototrophy. These results demonstrate the widespread distribution of phototrophic bacteria in hydrothermal vent fields, implying an underestimated contribution of phototrophic populations to deep-sea ecosystems.

Results

Characteristics of the sampling sites

During the TS10-2 expedition, between the 4th and 30th of January 2019, the manned submersible ShenHaiYongShi was used to survey the active vent DFF1 (also called Jabberwocky) located at 37.78°S and 49.65°E and at a depth of 2736 m. Hydrothermal fluids diffused out from active beehive structures or were ejected from orifices at various places. Thermal irradiances of 12–18 μW/cm², compared with a background level of about 1 µW/cm² (see Materials and Methods), were detected at 880 nm, 930 nm, and 1060 nm at approximately 3–4 m away from the edifice. This observation is consistent with the detection of thermal radiation at hydrothermal vents on the East Pacific Rise (EPR) and the Mid-Atlantic Ridge (MAR) (Van Dover et al. 1996; White et al. 2002).

We measured temperatures at various positions on the edifices. The highest temperature of 170–240 ℃ was detected inside an orifice (Fig. 1) with obvious emission of black smoke. Notably, the temperature decreased to 8 ℃ within approximately 10 cm away from the orifice, indicating a sharp temperature gradient in this niche. Next to the orifice was an active beehive structure continuously diffusing black smoke; here the surface temperature was approximately 30 ℃. Alvinocaridid shrimp swarm fossa were seen beneath the beehive structure where the temperature was around 6 ℃. The shrimp as Mirocaris indica was identified based on comparison of their reconstructed mitochondrial genome with NCBI Nucleotide Sequence Database (NT) data. Samples were taken from these three representative niches with different temperatures: the shrimp swarm fossa (SSF, site 1 in Fig. 1) is a low-temperature symbiotic niche; the active beehive diffuser structure (ABD, site 2 in Fig. 1) is a warm niche without animal disturbance; and the black smoker orifice (BSO, site 3 in Fig. 1) has high temperatures and a sharp temperature gradient.

Figure  1.  Sampling sites at the DFF1 (Jabberwocky) edifice. White spots indicate the locations where the temperature was measured. Temperatures in parentheses are those measured inside the orifice and are indicated by an arrow. Green triangles and diamonds mark grey and black shrimp, respectively. Three sampling niches are indicated: (1) shrimp swarm fossa (SSF), (2) an active beehive diffuser (ABD), and (3) an area within 30 cm of the black smoker orifice (BSO). Bar, distance of 10 cm as shown between the two red dots above the temperature sensor

DownLoad: Full-Size Img PowerPoint

Metagenomic analysis of community structures

Over 15 Gb of clean data were obtained for each of the three samples and a total of 1, 820, 925 genes were identified from the three metagenome libraries (Supplementary Table S1). Sequences were affiliated with Bacteria (98.21% relative abundance), Archaea (1.64%), Eukaryota (0.10%), and viruses (0.05%) (Supplementary Table S2). Twice as many genes were detected in the BSO compared with the ABD or SSF. As expected, more eukaryotic genes were found in samples from the SSF niche (5992 genes) than in those from the ABD (5255 genes) or BSO (1944 genes). In all three niches, genes affiliated with Bacteria were dominant (ranging from 97.66 to 98.47% relative abundance), while archaeal genes represented 1.33 to 2.20%. At the class level, Epsilonproteobacteria was the most abundant class in all samples (71.6% in SSF, 77.39% in ABD, 60.58% in BSO). Next most abundant was the Gammaproteobacteria (7.60% and 9.81%, respectively) in warm ABD and hot–cold BSO samples, and Zetaproteobacteria (8.00%) in the cold and symbiotic SSF sample (Fig. 2A). Alphaproteobacteria was the third most abundant population in the BSO sample, but was present at much lower abundances in SSF and ABD samples. Similarly, Deltaproteobacteria was more abundant in the BSO niche than in the SSF or ABD. Aquificae was the fourth most abundant class found in the relatively lowtemperature (6 ℃) SSF niche and was clearly enriched in this environment compared with the ABD (30 ℃) and BSO (8–240 ℃) niches.

Figure  2.  Community structure and KEGG pathway enrichment analysis of three niches at the DFF1 hydrothermal vent. A Class composition distribution in SSF (shrimp swarm fossa), ABD (active beehive diffuser) and BSO (black smoker orifice) niches. B and C Community structures at family and genus levels, respectively. D Abundance of epsilonproteobacterial genera in the DFF1 vent. Horizontal axis indicates the relatedness of taxa based on the abundance in corresponding genera; vertical axis clusters the samples according to the total abundance. Blue and red colors represent low and high abundance, respectively. E KEGG pathway enrichment analysis for four epsilonproteobacterial genera. Horizontal axis indicates different pathways; vertical axis indicates different genera and samples. Blue and red colors represent low and high gene number, respectively

DownLoad: Full-Size Img PowerPoint

At the family level, more than 38.5% of gene sequences belonged to unclassified bacterial families (Fig. 2B). Epsilonproteobacterial Helicobacteraceae were the dominant populations and epsilonproteobacterial Campylobacteraceae were among the most abundant taxa in all three niches. Epsilonproteobacterial Nautiliaceae was rich in the SSF sample. Members of the phototrophic Flavobacteriaceae (Bacteroidetes) and Rhodobacteraceae (Alphaproteobacteria) families were detected in all samples, but Rhodobacteraceae were much more abundant in the high temperature BSO niche than in the SSF and ABD niches. The two most abundant genera in all samples were Sulfurimonas and Sulfurovum (Epsilonproteobacteria) (Fig. 2C). Zetaproteobacterial Ghiorsea and epsilonproteobacterial Nitratifractor were the third most abundant taxa in the SSF and BSO niches, respectively. The high percentages of unclassified (18.19–27.75%) genera indicate that a large proportion of genera are yet to be identified in the DFF1 vent. The abundance of the nitrate-reducing Nitratifractor genus appeared to increase with temperature (Fig. 2D). The composition of proteins encoded by members of the class Epsilonproteobacteria and seven genera based on phylogenetic classification was similar in the three niches (Supplementary Fig. S1). However, fewer KEGG pathway categories and associated genes in the genera Lebetimonas, Caminibacter, Cetia, and Campylobacter were identified in the SSF sample compared with those from the ABD and BSO (Fig. 2E). Most members of these four genera have an optimal growth temperature between 30 and 70 ℃. The temperatures at the ABD (30 ℃) and BSO (8–240 ℃) seem more favorable for their growth than that at the SSF (6 ℃). An exception was observed for genes involved in signal transduction, motility, carbohydrate metabolism and lipid metabolism, which showed no reduction in abundance in the SSF sample for some of the genera.

Potential phototrophic populations

The presence of potential phototrophic bacteria was surveyed in the three niches using metagenomic data. Genes encoding most components of photosynthesis were detected in the metagenomes (Supplementary Fig. S2). Potential phototrophic microorganisms were detected in all three niches with differing abundances. Most of these potential phototrophic microorganisms were bacteria, but some were archaea, mainly Haloarculaceae, Halobacteriaceae, Haloferacaceae, and Halorubraceae. The samples from BSO with a sharp temperature gradient displayed slightly higher abundances of most categories of phototrophs than samples from the SSF and ABD (Fig. 3, Supplementary Table S3). The putative phototrophic microorganisms with highest abundances belonged to the Flavobacteriaceae and Rhodobacteraceae (Supplementary Table S4).

Figure  3.  Distribution of potential phototrophic microorganisms in the DFF1 vent. Families with asterisks belong to the archaea while the others are bacteria. Blue and red colors represent low and high abundance, respectively

DownLoad: Full-Size Img PowerPoint

Capacity for chlorophyll biosynthesis is one of the main criteria for assigning chlorophototrophic bacteria. Complete pathways for chlorophyll a biosynthesis were found in all samples, whereas only the BSO sample contained the complete set of enzymes for chlorophyll b biosynthesis (Fig. 4A). Five genes are involved in the biosynthesis of bacteriochlorophyll a or b from chlorophyllide a. The enzymes required for the first two steps were missing from all samples. The enzyme for the third step of the reaction was found in the ABD and BSO samples, and those for the last two steps were present in all samples. These observations suggest that either different components are involved in bacteriochlorophyll a and b biosynthesis or that chlorophyll a and b, but not bacteriochlorophyll, underpin the chlorophyll-sustained phototrophy at DFF1. Moreover, as shown in Fig. 4, BSO and ABD samples had higher diversity than the SSF sample for certain chlorophyll metabolic steps (Fig. 4B). Detection of the complete chlorophyll biosynthesis pathway confirmed that the microbial communities in the three niches of the DFF1 vent have the capacity for chlorophototrophy.

Figure  4.  Chlorophyll biosynthesis genes identified in DFF1 vent samples. A Pathways and gene distribution. Green, blue and red spots indicate detection of corresponding gene homologs in SSF, ABD and BSO samples, respectively. ChlG/BchG chlorophyll synthase, CAO chlorophyllide a oxygenase, NOL chlorophyll(ide) b reductase, BchF chlorophyllide a 31-hydratase, BchXYZ chlorophyllide a reductase, BchC bacteriochlorophyllide a dehydrogenase, ChlP/BchP geranylgeranyl-bacteriochlorophyllide a reductase. B Abundance of chlorophyll biosynthesis genes in the DFF1 vent. Blue and red colors represent low and high abundance, respectively

DownLoad: Full-Size Img PowerPoint

In accordance with the presence of complete chlorophyll biosynthesis pathways, all enzymes of the reductive pentose phosphate cycle (Calvin–Benson–Bassham (CBB) cycle, KEGG pathway M00165, Supplementary Fig. S3) for carbon fixation in photosynthetic organisms were detected from the three niches. In addition, samples from the three sites contained genes involved in crassulacean acid metabolism that are responsible for carbon fixation in photosynthetic organisms under both dark (KEGG M00168) and light (KEGG M00169) conditions (Supplementary Fig. S3). As expected for hydrothermal autotrophic ecosystems, genes associated with prokaryotic carbon fixation pathways were also present in these samples (Fig. 5A). The complete reductive citrate cycle (rTCA cycle, Arnon–Buchanan cycle, KEGG M00173), reductive acetyl-CoA pathway (Wood–Ljungdahl pathway, KEGG M00377) and phosphate acetyltransferase–acetate kinase pathway (KEGG M00579) were identified in all niches. Moreover, the complete dicarboxylate–hydroxybutyrate cycle (DC/4HB, KEGG M00374) was present only in the BSO niche. Two genes encoding succinate semialdehyde reductase (NADPH) [EC: 1.1.1.-] (here referred to as [EC: 1.1.1.-^b]) and succinyl-CoA reductase [EC: 1.2.1.76] of this pathway were absent in the SSF and ABD samples (Fig. 5A). The gene encoding malonyl-CoA reductase/3-hydroxypropionate dehydrogenase (NADP +) [EC: 1.2.1.75] of the 3-hydroxypropionate bi-cycle (3HP, KEGG M00376) and 3-hydroxypropionate/4-hydroxybutylate cycle (3HP/4HB, KEGG M00375) was not found in any of the three niches (Fig. 5A). Genes contributing to the three complete CO₂ fixation pathways, i.e., reductive citrate cycle (blue triangles in Fig. 5B), reductive acetyl-CoA pathway (yellow triangles) and phosphate acetyltransferase–acetate kinase pathway (grey triangles), were among the most abundant, implying a considerable capacity for autotrophy.

Figure  5.  Prokaryotic carbon fixation pathways identified in DFF1 vent samples. A Pathways and gene distribution. Green, blue and red spots indicate detection of corresponding gene homologs in SSF, ABD and BSO samples, respectively. B Abundance of carbon fixation genes in the DFF1 vent. Blue and red colors represent low and high abundance, respectively. Two enzymes have the same [EC: 1.1.1.-] nomenclatures. To distinguish them, they are referred to here as [1.1.1.-^a] for malonic semialdehyde reductase of hydroxypropionaterelated cycles and [1.1.1.-^b] for succinate semialdehyde reductase (NADPH) of the dicarboxylate–hydroxybutyrate cycle. The seven pathways of carbon fixation are indicated by lines with different colors in (A) and different symbols in (B): blue lines and solid triangle represent reductive citrate cycle (Arnon–Buchanan cycle, M00173); orange lines and hollow triangle, 3-hydreoxypropionate bi-cycle (M00376); green lines and solid circle, hydroxypropionate– hydroxybutylate cycle (M00375); purple lines and hollow circle, dicarboxylate– hydroxybutyrate cycle (M00374); yellow lines and solid square, reductive acetylCoA pathway (Wood–Ljungdahl pathway, M00377); grey lines and hollow square, phosphate acetyltransferase–acetate kinase pathway (M00579); dark yellow lines and solid star, incomplete reductive citrate cycle (M00620)

DownLoad: Full-Size Img PowerPoint

We found 106 genera belonging to the Flavobacteriaceae in our metagenomic library. Searching the NCBI database using Flavobacteriaceae and proteorhodopsin or bacteriorhodopsin as key words, identified 33 genera satisfying these search criteria. Furthermore, when one of the functional PRs from Polaribacter sp. MED 152 (Flavobacteriaceae) was used as a query to search against the metagenomic library ten putative PRs with an e-value lower than 1.00E-05 were found, all of which were only present in the ABD sample. Four candidates consisting of more than 200 residues and containing the seven predicated transmembrane helixes were selected for further analysis. ABD gene id 73, 573 (GenBank acc. no. MW133229) was 73.4% identical (3.2E-135) to a PR of Candidatus Pelagibacter sp., ABD gene id 90, 269 (MW133230) shared 64% identity (8.9E-100) with a PR of unclassified Gammaproteobacteria, ABD gene id 145, 638 (MW133231) exhibited 76.9% identity (5.4E- 134) to PR of SAR86 cluster, and ABD gene id 214, 696 (MW133232) shared 65.4% (1.5E-87) identity with a PR from unclassified Planctomycetes. Together, these results indicate the presence of four PRs from taxonomic groups other than flavobacteria in the ABD sample. Phylogenetic analysis of 70 rhodopsins clustered three PRs from the ABD niche with PRs of Proteobacteria and the fourth with a PR of Planctomycetes (Fig. 6).

Figure  6.  Phylogenetic tree of selected rhodopsins. The bootstrap scores over 60 are presented as numbers on branches. Seven background colors represent the seven bacterial phyla, respectively. Light pink: Bacteroidetes; dark pink: Proteobacteria; royal blue: Cyanobacteria; rosy brown: Actinobacteria; purple: Planctomycetes; dark green: Euryarchaeota; light green: Firmicutes. The stars in yellow indicates sequences obtained in this study

DownLoad: Full-Size Img PowerPoint

Multiple alignment of protein sequences encoded by the four PR genes found in the ABD sample and 17 representative PR sequences extracted from the NCBI databank showed that both the retinal binding site K230 and other functionally important residues such as the primary proton acceptor D97 and the Schiff base counterions R94 and D226 were perfectly conserved (Fig. 7 and Supplementary Fig. S4). The primary proton donor at position 108 is Glu (E) in flavobacteria but might be Lys (K) or Leu (L) in other species (Fig. 7). The four DFF1 PR sequences exhibited species specificity at this position; ABD gene id 145, 638 (closely related to SAR86 cluster) and ABD gene id 73, 573 (Candidatus Pelagibacter sp.) had Glu, ABD gene id 214, 696 (unclassified Planctomycetes) had Lys and ABD gene id 90, 269 (unclassified Gammaproteobacteria) had Ile (I) (Fig. 7). At position 105 three of ABD PR had a hydrophilic blue-light absorption residue Gln (Q) and the fourth had a hydrophobic green-light absorption residue Met (M).

Figure  7.  Alignment of the helix C region of proteorhodopsins. Conserved functional residues involved in retinal binding or as proton donors/receptors are indicated by blue boxes, and residues affecting absorption light wavelength are marked by the red box. Residue positions in Polaribacter sp. MED152 were used as reference

DownLoad: Full-Size Img PowerPoint

There is no well-defined integral co-factor or specific enzyme for semiconductor-mediated phototrophy. Only one marine Idiomarina sp. isolate is reported to use this kind of phototrophic mechanism. Members of the genus Idiomarina were present in all samples in this study. Given that Cd²⁺, Zn²⁺ and Ti²⁺ ions and H₂S are generally present in black smokers and that cystathionine beta-lyase (cbl), methionine gamma-lyase (mgl) and full sulfur metabolic pathways leading to the production of H₂S were found in all samples from the DFF1 vent field (Supplementary Fig. S5), it is plausible that photocatalytic phototrophic bacteria dwell in hydrothermal vent ecosystems.

Discussion

To adapt to the biogeochemical properties of hydrothermal vents, microbial communities vary from vent to vent and among different local niches within a given vent (Dick 2019). Similarly, metabolic and physiological diversity in these bacterial communities changes dynamically as hydrothermal fields develop (Wang et al. 2009). Despite this variation, Epsilonproteobacteria represent the most abundant group of bacteria in deep-sea hydrothermal environments (Nakagawa et al. 2005). In hydrothermal plumes, Epsilonproteobacteria, the SUP05 group of sulfur-oxidizing Gammaproteobacteria, ammonia-oxidizing Betaproteobacteria, methanotrophs and Marine Group I (MGI) archaea are often identified as dominant taxonomic groups (Lesniewski et al. 2012). The Guaymas Basin (GB) hydrothermal plume is dominated by Gammaproteobacteria, similar to those in the background of deep-sea communities (Dick and Tebo 2010; Lesniewski et al. 2012). Sylvan et al. (2012) assessed the variability in bacterial community structure associated with plume particles at two active hydrothermal vents on the EPR. Epsilonproteobacteria clones and hyperthermophilic Aquificae were dominant at the beginning, but this shifted to Gammaproteobacteria and Alphaproteobacteria being more abundant toward the end of their experiment (Sylvan et al. 2012). Vent sediments of the three most-visited and wellknown deep-sea hydrothermal vent fields, the Menez Gwen, Lucky Strike and Rainbow, located in the MAR, harbor different microbial communities (Cerqueira et al. 2017). Menez Gwen samples are mainly composed of members of Sulfurovum, Sulfurimonas and Campylobacter genera belonging to Epsilonproteobacteria. Rainbow samples are rich in organisms belonging to the OCS155 Marine Group, Gemmatimonadaceae family, Nitrospiraceae family, and Nitrosococcus and Acidiferrobacter genera. By contrast, Lucky Strike samples consist of representatives of the Rhodospirillaceae family (purple non-sulfur phototroph) and Hyphomicrobiaceae family (Alphaproteobacteria), Nitrosospira (Betaproteobacteria) and JTB255 Marine Benthic Group (Gammaproteobacteria). Microbial mats from the Lucky Strike hydrothermal vent field on the MAR consist mainly of Gammaproteobacteria including hydrothermal fauna symbionts Thiotrichales and Methylococcales (Crépeau et al. 2011). Recently, Hou et al. (2021) compared microbial communities from an active and a recently extinct (~7 years) sulfide chimney from the hydrothermal vent field on the EPR. The community in the active chimney mainly consisted of sulfide- and/or hydrogen-oxidizing campylobacteria and Aquificae whereas the recently extinct chimney largely harbored Gammaproteobacteria and Nitrospirae with Alphaproteobacteria and Deltaproteobacteria less abundant.

In the SWIR hydrothermal vent fields, Ren et al. (2012) reported that the two predominant taxa in hydrothermal plumes from the Longqi vent field were Gammaproteobacteria and Alphaproteobacteria. Djurhuus et al. (2017) compared microbial communities in the Longqi vent field plumes at the SWIR and E2 sites on the slow-spreading East Scotia Ridge (ESR). Members of the genera Arcobacter, Caminibacter (affiliated with Nautiliales (Nakagawa and Takai 2014)) and Sulfurimonas from the Epsilonproteobacteria and the SUP05 group from the Gammaproteobacteria were common deep-sea bacterial representatives in the two fields. However, Epsilonproteobacteria exhibit niche-specificity, with a three orders of magnitude decrease in abundance from the vent orifice to the background water at the ESR. Ding et al. (2017) reported that members of the Epsilonproteobacteria were the most abundant taxa in the two vents, among the dominant taxa but not the most abundant in one vent, and present in low numbers in another vent on the SWIR. Moreover, Epsilonproteobacteria were absent from the sediments of two inactive SWIR vents where Alphaproteobacteria and Gammaproteobacteria were dominant (Zhang et al. 2016). In this study of a Longqi vent, it was found that Epsilonproteobacteria were the most abundant and Gammaproteobacteria were among the most abundant populations in the three niches (Fig. 2). Interestingly, Zetaproteobacteria were more abundant in the SSF and Alphaproteobacteria were more abundant in the area close to the orifice than in other samples (Fig. 2). These results are consistent with the dominance of Epsilonproteobacteria in active venting fields, but also highlight the niche-dependent variation of other populations. We focused our analysis on phototrophic populations that might be influenced by light radiation in these niches. The in situ temperature could indirectly reflect irradiation intensity, but this is technically difficult to determine with any precision.

The metagenomic analysis showed a wide manifestation of phototrophic bacteria in the DFF1 vent. Although chlorophyll biosynthesis gene sequences were found in all samples, members of the Rhodobacteraceae were more abundant in the area close to the black smoker orifice than at the other two sampling sites. Moreover, the PR-coding gene was only present in the ABD sample. Because the ABD sample produced about fourfold fewer clean-reads than the other two samples, the unique occurrence of PR in the ABD niche is significant. The PRs of Polaribacter sp. MED 152 and other near-surface bacteria use the hydrophobic sidechain Met or Leu at position 105 to absorb, respectively, 535 nm or 530 nm green light, whereas PRs from bacteria in deeper water use the hydrophilic amino acid glutamine at this position to absorb 490 nm blue light because blue light penetrates deeper than green light (Gómez-Consarnau et al. 2007). Three of the four PRs in the ABD sample had a blue-light-absorption glutamine whilst the fourth had a green-light-absorption methionine at this position (Fig. 7). Because blue light has higher energy than green light, using glutamine to absorb blue light might confer a competitive advantage in hydrothermal vents with very weak visible light. Photosensitization phototrophy is based on semiconductors such as CdS, ZnS or TiO. Hydrothermal venting fluids are rich in Cd²⁺, Zn²⁺, Ti²⁺, Pb²⁺ and H₂S. Zinc sulfide (wurtzite and sphalerite) is one of the major minerals in venting chimneys (Ding et al. 2017; Wang et al. 2009). Therefore, the black smoker chimney could provide the semiconductors necessary for photosensitization phototrophy. The putative semiconductor-mediated phototrophic genus Idiomarina sp. was found in all three niches. It is tempting to hypothesize that this kind of phototrophism functions in all three niches of DFF1.

Carbon fixation plays a crucial role in photosynthetic and autotrophic metabolisms. Our metagenomic analysis found complete pathways for the CBB and rTCA cycles (Fig. 5) that contribute to photosynthesis. As anticipated for chemosynthetic vent ecosystems, metagenomic analysis also revealed the presence of a complete rTCA cycle and Wood–Ljungdahl pathways at all sites and the DC/4HB pathway in ABD samples. Therefore, vent microorganisms may contribute significantly to carbon cycling in hydrothermal vent ecosystems. Moreover, PR and the complete DC/4HB pathway were found only in ABD samples, implying that the active beehive diffusers might have more active phototrophic activity.

Epsilonproteobacterial genera Caminibacter, Nautilia, and Nitratifractor are considered to be the endemic taxa of hydrothermal vents (Campbell et al. 2006). Caminibacter and Nautilia belong to the order Nautiliales and have been found in close proximity to deep-sea hydrothermal vents (Nakagawa and Takai 2014). Three genera were found in all samples from the DFF1 vent (Fig. 2A). In addition, the abundance of Nitratifractor was associated with increased temperature (Fig. 2D). Interestingly, Gonnella et al. (2016) found these species in global open ocean water samples. In addition, they identified the cyanobacteria Prochlorococcus and Synechococcus in both global open ocean samples and those from 16 hydrothermal vents on the MAR. These authors hypothesized that the open ocean plays an important role as a seed bank to ensure the widespread distribution of microorganisms. Our metagenomic analysis revealed also the occurrences of genes coding for bacteriorhodopsin and key enzyme for chlorophyll biosynthesis in deep sea waters. Therefore, we propose that the phototrophic bacteria in the seed bank can reach deep sea and the vent fields. If their abundance exceeds a certain threshold, these phototrophs might proliferate and evolve in the vent fields because geothermal light provides an advantage when competing in this chemosynthetic ecosystem.

Since 1998, only two chlorophototrophs have been isolated from black smoker plume waters in hydrothermal vent fields of the Pacific Ocean (Beatty et al. 2005; Yurkov et al. 1999). Our metagenomic study identified a complete chlorophyll biosynthesis pathway and PRs in the DFF1 vent samples on the ultraslow-spreading SWIR. However, our attempt to cultivate phototrophs was unsuccessful despite diverse bacteria being isolated from these samples. The reasons for our inability to cultivate phototrophs include their low abundance, lack of adequate growth and illumination conditions, low growth rates and the presence of dormant cells, which has been observed for marine microorganisms in general (Wang et al. 2021). Mu et al. (2021) proposed culturomics and enrichment culture methods for cultivating rare active bacteria and resuscitation culture for isolating dormant bacteria (Mu et al. 2021). These successful guidelines will be applied to cultivation of hydrothermal vent phototrophs in the future.

Materials and methods

Sample collection

Samples were collected at the hydrothermal vent DFF1, using the hydraulic arm of the manned submersible ShenHaiYongShi, and deposited into bio-boxes sterilized with 70% ethanol and filled with autoclaved seawater. Boxes were kept tightly closed during the descent and ascent. Once on board, samples were immediately transferred into sterile sampling bags and preserved at - 80 ℃.

Measuring in situ temperature and hydrothermal vent irradiance

Temperatures at sampling sites were measured using a titanium-sheathed K-type thermocouple with a precision of ± 1 ℃ (Fig. 1) (Wang et al. 2020). Temperature values were calculated and displayed on a computer inside the manned submersible in real-time mode.

To measure vent irradiation, a photodiode power sensor S120VC (Thorlabs SAS, France) was fixed behind the pilot's porthole, facing the hydrothermal edifice of DFF1 at a distance of 3–4 m. All windows were covered with tissues and lights of the submersible were switched off during measurements. The detector was connected to a Powermeter PM100D (Thorlabs SAS, France) with wavelengths set to 880 nm, 930 nm and 1060 nm and a readout unit of μW/cm².

Taxonomic annotation of shrimp

Muscle DNA of shrimp collected at DFF1 was extracted using the CTAB method. After library construction and sequencing using the BGISEQ-500 platform, 25 million paired-end clean reads were used to assemble a mitochondrial genome of 15, 926 bp and GC content of 33.05% using NOVOPlasty (v4.2) (Dierckxsens et al. 2017) with Alvinocarididae mitochondrial genomes as reference sequences. Subsequently, the mitochondrial genome was aligned to the NT (NCBI Nucleotide Sequence) database using Blast + (v2.9.0), which revealed 98.95% identity to Mirocaris indica sequence MT879755.1 with 100% coverage.

Metagenome assembly and taxonomic assignment

Total genomic DNA was extracted directly from each 300 mg homogenized frozen sample using the protocol reported by Costea et al. (2017). To obtain sufficient quantities of DNA for meta whole-genome sequence library construction, each DNA extraction was subjected to three parallel MDA (Multiple displacement amplification) amplifications (Riva et al. 2019). To mitigate bias produced by this amplification, three sets of independently constructed libraries were mixed well for the following steps of the analysis.

To construct sequencing libraries, extracted microbial DNA was fragmented to give a size of 500 ~ 800 bp using a Covaris E220. Fragments between 150 and 250 bp were selected using AMPure XP beads (Agencourt, Beverly, MA, USA). DNA fragments were repaired using T4 DNA polymerase (ENZYMATICS, Beverly, MA, USA) to obtain blunt ends, and 3′ ends were further modified to obtain a dATP sticky end. These DNA fragments were ligated at both ends to T-tailed adapters and amplified for eight cycles. A singlestrand circularization process was subsequently performed to generate a single-strand circular DNA library (Guo et al. 2018). Libraries were sequenced on the BGISEQ-500 platform using a pair-end 100 bp sequencing strategy at BGIQingdao (China).

All raw reads were subjected to quality control using SOAPnuke (v 1.5.2) (Chen et al. 2018) to remove lowquality, adapter-contaminated and duplicated reads. More than 15 Gb of clean data for each sample was obtained and used for metagenomic analysis. Filtered reads were assembled using both IDBA-UD (v1.1.3) (Peng et al. 2012), with the parameters –mink 23 –maxk 83 –step 20 –pre_correction, and MEGAHIT (v1.1-beta) (Li et al. 2016), with the parameters –min-count 2/3 –k-min 33 –k-max 53 –k-step 10 –no-mercy. Superior N50 assembly results were chosen from either of the two software packages. MetaGeneMarK (v3.38) (Zhu et al. 2010) was used to search for coding sequences in assembled contigs longer than 100 bp.

To identify microorganisms, gene sequences from all samples were aligned against those in the Non-Redundant Protein Sequence Database (NCBI Nr database) (v20200619). Taxonomic assignment of genes was performed using the lowest common ancestor (LCA) approach with the parameters match percentage 40% and identity cutoff 40%.

Bowtie2 (v 2.2.5) (Langmead and Salzberg 2012) software was used to index the nucleotide sequences of all samples for quantitative analysis. Clean reads from each sample were then compared to the index file to produce a sam file; the specific parameter settings were –phred33 –sensitive –dpad 0 –gbar 99, 999, 999 –mp 1, 1 –np 1 –score-min L, 0, - 0.1 -I 1 -X 1000 -p 4 -k 200 -q. Gene abundance profiling (transcripts per million reads (TPM) value) was performed using Salmon-v0.9.1 (Patro et al. 2017). Species abundance was determined from the sum of assigned gene counts (Forslund et al. 2015; Jie et al. 2017). Relative abundance was the percentage TPM value of the taxonomic group in the corresponding sample. The abundance of each taxonomic rank (Superkingdom, Phylum, Class, Order, Family, Genus, Species) in the different samples was summarized in a profiling table or histogram, and histograms were drawn using R. Species with abundance of less than 0.5% in all samples were classified into 'others'.

Gene function and pathway annotations

The Blastp program based on diamond (v0.8.23.85) (Buchfink et al. 2015) was used to compare sequences against the Kyoto Encyclopedia of Genes and Genomes (KEGG) [87.0] (Kanehisa et al. 2016) and the Nr database [v20200619], and the best-hit for each gene (highest score, lowest e-value) was selected.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s42995-021-00121-y.

Acknowledgements

This work was supported by the National Key Research and Development Program of China (No. 2018YFC0309904), the National Natural Science Foundation of China (Nos. 91751202, 41806174, 91751108), the Key Research and Development Program of Hainan Province (No. ZDKJ2019011), Grant Y9719105 from the Institute of Deep-sea Technology Innovation, Chinese Academy of Sciences (IDSTI-CAS), Grant 2019YD16 from Sanya City and Grant INSB-DBM2021 and support to LIA-MagMC from Centre National de la Recherche Scientifique. We thank cruise members of expedition TS10 of the R/V TSYH, especially the pilots of the manned submersible ShenHaiYongShi and Y Lu, YW Pan and VH Pellizari for assistance in sampling.

Author contributions

HC, performed research, analyzed data, prepared figures; DHL, analyzed data, wrote the paper; AJJ, performed research; XGL, analyzed data; SJW, performed research; JWC, analyzed data, wrote the paper; MJQ, performed research; XQQ, analyzed data; JD, analyzed data; RZ, analyzed data; WJZ, analyzed data, wrote the paper; SSL, analyzed data; LFW, conceived of and designed study, performed research, analyzed data, wrote the paper.

Data availability

The data that support the findings of this study have been deposited into CNGB Sequence Archive (CNSA)(Guo et al. 2020) of China National GeneBank DataBase (CNGBdb)(Chen et al. 2020) with accession number CNP0002004. All data are available. The nucleotide sequence data reported are available in the GenBank database under accession numbers MW133229, MW133230, MW133231, MW133232, and MZ351491.

Declaration

Conflict of interest

There are no conflicts of interest or competing interests.

Animal and human rights statement

No human or animal materials were used in this study.