Bacterial and archaeal communities in the deep-sea sediments of inactive hydrothermal vents in the Southwest India Ridge

Active deep-sea hydrothermal vents harbor abundant thermophilic and hyperthermophilic microorganisms. However, microbial communities in inactive hydrothermal vents have not been well documented. Here, we investigated bacterial and archaeal communities in the two deep-sea sediments (named as TVG4 and TVG11) collected from inactive hydrothermal vents in the Southwest India Ridge using the high-throughput sequencing technology of Illumina MiSeq2500 platform. Based on the V4 region of 16S rRNA gene, sequence analysis showed that bacterial communities in the two samples were dominated by Proteobacteria, followed by Bacteroidetes, Actinobacteria and Firmicutes. Furthermore, archaeal communities in the two samples were dominated by Thaumarchaeota and Euryarchaeota. Comparative analysis showed that (i) TVG4 displayed the higher bacterial richness and lower archaeal richness than TVG11; (ii) the two samples had more divergence in archaeal communities than bacterial communities. Bacteria and archaea that are potentially associated with nitrogen, sulfur metal and methane cycling were detected in the two samples. Overall, we first provided a comparative picture of bacterial and archaeal communities and revealed their potentially ecological roles in the deep-sea environments of inactive hydrothermal vents in the Southwest Indian Ridge, augmenting microbial communities in inactive hydrothermal vents.

Deep-sea hydrothermal vent environments exhibit complex dynamic habitats that are characterized with steep thermal and chemical gradients 1 . Microorganisms thriving in hydrothermal vent environments have been extensively studied by cultivation-dependent and -independent approaches 2 . From these environments, thermophilic archaeal members of the orders Thermococcales, Methanococcales, Archaeoglobales, and Aquificales [3][4][5] , and mesophilic and thermophilic bacterial members of Epsilon-proteobacteria 6,7 have been detected. However, active hydrothermal vents eventually become inactive ones, which may last for thousands of years 8 . Inactive hydrothermal vents provide potential habitats for microbial communities. Many reports have shown that inactive hydrothermal vents harbor a completely different assemblage of microorganisms, compared to active hydrothermal vents. For example, Thaumarchaeota, which are usually not major archaea in active vents, have been found to be widespread not only in several inactive hydrothermal vents [8][9][10] , but also in diffuse flow chimneys 11 . In addition, Alpha-, Beta-, Delta-, and Gamma-proteobacteria are dominant bacteria in several inactive hydrothermal vents 8,10,12 .
The Southwest Indian Ridge is a boundary between the Antarctic and African plates and extends from the Bouvet triple junction in the Atlantic Ocean to the Rodriguez triple junction in the Indian Ocean 13 . The first active hydrothermal vent (E 49°39, S 37°47) in the Southwest Indian Ridge was discovered in 2011, which is characterized with low temperatures and enrichment for chimney sulfides, such as pyrite, marcasite, sphalerite, and chalcopyrite 14 . In previous studies, arsenite-resistant bacteria were isolated from deep-sea sediments in the Southwest Indian Ridge 15 , and microbial sulfur cycle in two hydrothermal chimneys on the Southwest Indian Ridge were also revealed 16 . Furthermore, microbial communities in semi-consolidated carbonate sediments (51.009 o E, 37.6081 o S) of the Southwest Indian Ridge were investigated by traditional 16S rDNA clone library analysis 9 ; however, extensive microbial diversity and abundance in inactive hydrothermal vents in the Southwest Indian Ridge have not been well understood.
The primary goal of this study was to reveal microbial communities of two deep-sea sediments located at different regions in inactive hydrothermal vents in the Southwest Indian Ridge. For this purpose, the V4 region of 16S rRNA gene were sequenced via an Illumina MiSeq 2500 platform, which is a high-throughput sequencing technology that has been frequently used to investigate microbial community structure in various environments [17][18][19][20][21] . In this study, bacteria and archaea covering 15 phyla, 34 classes, 58 orders, 101 families and 141 genera were detected, augmenting the microbial community structure in inactive hydrothermal vents in the Southwest Indian Ridge. Furthermore, our results revealed that bacterial communities were dominated by Proteobacteria and Bacteroidetes and archaeal communities were mostly composed of Thaumarchaeota and Euryarchaeota. To the best of our knowledge, this is the first report on bacterial and archaeal communities of the deep-sea sediments in inactive hydrothermal vents in the Southwest Indian Ridge by Illumina high-throughput sequencing.

Results
Mineralogical and geochemical characteristics of the samples. Mineralogical and geochemical characteristics of the two samples were determined using X-ray diffraction phase, scanning electron microscope, carbon and oxygen isotope, and trace element analyses 22 . Generally, the two samples had similar mineralogical and geochemical characteristics summarized in such as Sr and Ba, but depleted in siderophile elements. Major calcite (> 95%) and lower content of clay minerals, aragonite and quartz were present in the two samples. Furthermore, the two samples had the higher content of δ 18 O PDB than δ 13 C PDB . However, the TVG4 sample was white sticky mud while the TVG11 sample was black loose mud and sand, which may lead to different microbial community structure.
Sequencing data. Sequencing information, diversity index, and estimators of richness were summarized in Table 2. The Illumina-based analysis of the hypervariable V4 region of 16S rRNA gene produced 78,461 and 358,589 total tags for bacteria and archaea, respectively. After filtering and removing potential erroneous sequences, a total of 74,980 and 311,035 effective tags were obtained for bacteria and archaea, respectively. Based on 97% similarity (Table 2), a total of 366 and 280 OTUs (operational taxonomic units) for bacterial diversity were obtained in TVG4 and TVG11 respectively, and a total of common 19 OTUs for archaeal diversity were obtained in TVG4 and TVG11. The Venn diagram for bacterial diversity showed that TVG4 and TVG11 shared 267 OTUs, 99 and 13 OTUs were unique in TVG4 and TVG11 respectively (Fig. 1). Interestingly, no unique archaeal OTU was observed in TVG4 and TVG11, besides the shared 19 OTUs.
In alpha diversity analysis, rarefaction curves, Chao1 index and Shannon index were generated based on 97% similarity. Rarefaction curve analysis of OTUs in the two samples indicated that TVG4 had more bacterial species than TVG11 ( Fig. 2A), and the two samples had the same archaeal species (Fig. 2B). This was confirmed by Chao1 index (Table 2). Furthermore, both rarefaction curves for bacterial and archaeal species approached an asymptote ( Fig. 2), suggesting that the sampling depths were sufficient to capture overall microbial diversities in the two samples. In addition, Shannon's diversity index showed that the order of the diversity, ranging from high to low, was TVG4 for bacterial diversity (Shannon = 5.76), TVG11 for bacterial diversity (Shannon = 3.99), TVG11 for archaeal diversity (Shannon = 3.05), and TVG4 for archaeal diversity (Shannon = 2.05) ( Table 2).

Microbial community analysis.
A total of 386,015 effective tags were obtained, including 74,980 bacterial tags and 311,035 archaeal tags (Table 2). Bacterial tags covered 13 phyla, 32 classes, 56 orders, 97 families, 138 genera, and archaeal tags covered 2 phyla, 2 classes, 2 orders, 2 families, and 3 genera. Bacterial and archaeal sequences from the two samples were further analyzed at the phylum, class and genus levels.

Figure 1. Venn diagrams of the OTUs for bacterial diversity.
Unique and shared OTUs between the two samples were based on 97% similarity. The numbers inside the diagram are the numbers of OTUs. than TVG11, TVG4 had higher abundance of Bacteroidetes, Actinobacteria and Firmicutes than TVG11. Thus, the results suggest that Proteobacteria and Bacteroidetes are dominant phyla in TVG4 and TVG11. At the phylum level, 99.57% of sequences belonged to Euryarchaeota phylum while only 0.43% of sequences were members of Thaumarchaeota phylum in TVG4 (Fig. 3B), suggesting that Euryarchaeota are major phyla in TVG4. In contrast, 88.22% of sequences belonged to Thaumarchaeota phylum while 11.78% of sequences were members of Euryarchaeota in TVG11 (Fig. 3B), indicating that Thaumarchaeota are predominant phyla in TVG11. Thus, the results suggest the significant divergence in the relative abundance of archaeal phyla between TVG11 and TVG4, although no unique OTU was observed for them.
Halobacteriales of Euryarchaeota phylum and Nitrosopumilales of Thaumarchaeota phylum were the main classes of archaea in the two samples (Fig. 4B). While TVG11 had a low relative abundance of Halobacteriales (9.3%) and a high relative abundance of Nitrosopumilales (90.7%), TVG4 had a high relative abundance of Halobacteriales (99.3%) and a low relative abundance of Nitrosopumilales (0.7%). Overall, the two samples had same archaeal species, but different relative abundance.
Beta diversity index analysis of the samples. Based on the weighted UniFrac distance and unweighted UniFrac distance cluster analysis, a dissimilarity coefficient for the two samples was measured to estimate the divergence of microbial species between them. The lower dissimilarity coefficients suggest the less divergence of microbial species. In this study, the dissimilarity coefficients for the two samples were measured to be 0.2 and  0.568 for bacterial and archaeal diversities, respectively (Fig. 6), suggesting that more divergence in the two samples was observed for archaeal species than for bacterial species.

Discussion
High-throughput sequencing techniques provide powerful tools for studying microbial communities in extreme environments, including deep-sea environments. For example, MiSeq sequencing platform PE250 or PE300 is capable of reading long DNA fragments and is widely used for microbial 16S rDNA sequencing 23 , allowing us to detect rare microorganisms with low relative abundance (< 0.01%) that would been masked by dominant populations if techniques with lower resolution had been applied 24 . Thus, the accuracy of species abundance will be greatly enhanced by MiSeq sequencing platform. Currently, Miseq sequencing platform has been a prior method to study microbial diversities in various environments. To discern more fully bacterial and archaeal communities in deep-sea sediments from inactive hydrothermal vents at the Southwest Indian Ridge, the V4 region of 16S rRNA gene from the two sample DNA were amplified and phylogenetically analyzed by a barcoded IIIumina high-throughput sequencing.
The phylum Proteobacteria was dominant in the taxonomic groups of the two samples in this study. Within this phylum, Alpha-proteobacteria, Gamma-proteobacteria, and Sphingobacteriia, were the three main classes, which are similar to the reports on the deep-sea sediments from Iheya North and Iheya Ridge 19 and metal-rich vent deposits from Pacific Ocean hydrothermal fields 25,26 . In the Southwest Indian ridge, similar observations were also made in semi-consolidated carbonate sediment samples 9 . Previous studies have shown that Epsilon-proteobacteria are dominant bacteria identified from active hydrothermal vents, and play a role in carbon and sulfur cycles 6,12,27,28 . However, Epsilon-proteobacteria was not detected in the two samples in this study. Compared to bacterial communities in active vents, which are dominated by Epsilon-proteobacteria, bacterial communities in inactive vents are mostly composed of Alpha-, Beta-, Delta-, and Gamma-proteobacteria 8,10,12 . Our results augment the hypothesis that while Epsilon-proteobacteria prefers active hydrothermal vents, Alpha-, Beta-, Delta-, and Gamma-proteobacteria prefer inactive hydrothermal vents.
Bacteroidetes was the second most abundant phylum in the two samples examined in this study and contained three classes, i.e. Sphingobacteria, Flavobacteria, and Bacteroidia. Previous studies have shown that Bacteroidetes are widely distributed in many inactive and active deep-sea hydrothermal vents 7,12 . It was also detected as one of the most abundant phylum in methane seep sediments in the Nankai Trough 29 . In this study, the abundance of Bacteroidetes in TVG4 was higher than that in TVG11, likely as a result of environmental difference between these two samples. Other dominant groups of bacteria identified in this study include members of Gemmatimonadetes, Verrucomicrobia, Acidobacteria, Chloroflexi, TM7, etc., which are known to be widely distributed in crustal fluids, deep-sea sediments, and inactive hydrothermal chimneys, but rare in active hydrothermal regions 8,12,25,30 . Euryarchaeota, Crenarchaeota, and Thaumarchaeota are known to be three major archaeal phyla in global environments. In this study, only Euryarchaeota and Thaumarchaeota were identified. Specifically, Thaumarchaeota were the major phylum in TVG11 while Euryarchaeota were the predominant archaea in TVG4. Previous studies have shown that Thaumarchaeota are widely found not only in several inactive hydrothermal vents from Southwest Indian Ridge, Western Pacific Ocean and Central Indian Ridge 8,9 as well as in diffuse flow chimneys from Arctic Mid-Ocean Ridge 31 . Besides, members of Thaumarchaeota are found in marine, freshwater, soils, and anoxic subsurface sediments 32 , all which possess the dissimilatory ammonia oxidation pathway and gain energy by ammonia oxidation 33 . However, few Thaumarchaeota has been found in active hydrothermal vents, which harbor a great number of typical (hyper)thermophilic archaea, such as Desulfurococcales, Thermococcales and Archaeoglobales 4,27,34-36 . Thaumarchaeota also dominated the archaeal communities in the semi-consolidated carbonate deep-sea sediments in the Southwest Indian ridge 9 , which is consistent with our observations. Taken together, these results suggest a wide presence of Thaumarchaeota in inactive hydrothermal vents.
Besides Thaumarchaeota, Natronomonas and Halolamina were dominant Euryarchaeota in TVG4 in this work. To the best of our knowledge, this is the first report on the presence of Natronomonas and Halolamina in inactive hydrothermal vents. Recently, the analysis of microbial diversity in the new deep-sea hypersaline Lake Thetis located in the Western part of the Mediterranean Ridge indicated that neither Natronomonas nor Halolamina was not detected 37,38 . The presence of Natronomonas and Halolamina in the Southwest Indian ridge may be relevant to the geochemical environments.
The genera Nitrospira of the Nitrospinae phylum are mainly widespread in marine environments including metal-rich sediments from Green Bay and western Pacific Ocean 25,39 , and participate in nitrogen cycling by converting nitrites to nitrates 40,41 . These two bacteria were also detected in the two samples. In this study, Nitrospira accounted for 0.01% of total bacteria in the two samples in Table 2. Furthermore, Rhizobiales, which participates in nitrogen-fixing 42 , accounted for 2.2% of total bacteria in the two samples (Table 3). Besides bacteria, archaea also play an important role in the oxidation of ammonia to nitrite and in N 2 fixation 32 . For example, Thaumarchaeota harbors the ammoniamonooxygenase subunit A (amoA) gene, and the encoded AmoA can oxidize ammonia to nitrite 43 . The genera Nitrosopumilus of the phylum Thaumarchaeota, an ammonia-oxidizing archaea, was detected with a high relative abundance (27.9%) in TVG11 but a low relative abundance (20%) in TVG4 ( Table 3). The presence of these bacterial and archaeal communities that are associated with nitrogen metabolism in this study indicates a likely involvement of these microbes in the process of nitrogen cycle.
Increasing evidences suggest that bacteria play a key role in the formation of deep-sea metallic mineral 44 . For example, Rhodococcus, Caulobacter, Hyphomicrobium and Acinetobacter are known to be involved in metal metabolism [45][46][47][48] . In this study, these four genera accounted for 0.59% of the bacteria and had a higher abundance in TVG11 (0.39%) than in TVG4 (0.20%) ( Table 3). The genus Hyphomicrobium of the Alpha-proteobacteria class, which is known to associate with metal metabolism 45 , accounted for 0.32% of the bacteria in this study. Rhodococcus, one genus of Mn-oxidizing bacteria, was detected with a relative abundance of 0.023%. Rhodococcus, which was first isolated from submarine basalts, can absorb metals such as Co, Cu, Ni and Mn, thus facilitating metal enrichment 46 . The genera Acinetobacter of the Gamma-proteobacteria class, which is involved in nutrient metal acquisition and metabolism 47 , were detected in the two samples with a 0.13% of relative abundance. Furthermore, Caulobacter within the Alpha-proteobacteria class can absorb and metabolize a variety of metals including Mn, Co, and Fe 48 . Caulobacter was detected in the two samples with a 0.11% of relative abundance. Overall, the presence of metal-utilizing bacteria of the samples in this study suggests a mechanism of environment induced microbial adaptation.
Alpha-, Delta-, Gamma-and Epsilon-proteobacteria are known to mediate sulfide reduction and oxidation 49 , which is one of the most important microbial chemosynthetic pathways in deep-sea hydrothermal ecosystems 50 . In this study, potential sulfide oxidation/reduction microbes comprised over 10.977% of the bacteria, including Alpha-and Delta-proteobacteria (Table 3). The order Rhodobacteraceae of Alpha-proteobacteria class and the order Desulfobacterales of Delta-proteobacteria class, which accounted for 10.97% and 0.007% of the bacteria in this study (Table 3), respectively, have been detected in many hydrothermal chimneys with functions of sulfide oxidation or reduction 5,25,51 .
Besides the bacteria and archaea associated with nitrogen and sulfur cycling and metal metabolism, the genus Planctomycetes was detected with a total abundance of 0.08% in this study (Table 3). Chistoserdova et al. suggested a possible role for Planctomycetes in the evolution of the methane cycle on Earth 52 . Thus, the presence of Planctomycetes in the two samples of our study is thought to participate in CH 4 oxidation.
Alpha diversity analysis showed that bacterial richness in TVG4 were higher than that in TVG11 while archaeal richness in TVG4 were lower than that in TVG11 as revealed by Shannon's diversity index, suggesting that environmental parameters probably have a significant impact on microbial community. Beta diversity analysis showed that the two samples had more difference in archaeal abundance than bacterial abundance. These results suggest that the distance to active hydrothermal vents and ambient environmental parameters probably have an important influence on the formation of microbial communities, even in the same oceanic ridge.
In conclusion, we revealed for the first time the comparative analysis of the microbial diversity in different sediments of the Southwest Indian Ridge using the high-throughput sequencing technology of Illumina MiSeq 2500 platform. Although the taxonomies of microbial communities in the two samples are largely similar, the abundances of most of the taxa differ between the two samples. Furthermore, our results revealed that bacterial communities were dominant by Proteobacteria and Bacteroidetes and archaeal communities were mostly composed of Thaumarchaeota and Euryarchaeota. These bacterial and archaeal communities would be potentially involved in nitrogen and sulfur cycling, and metal metabolism, suggesting that they play the important roles in ecological function in inactive deep-sea hydrothermal vents.  TVG11 (E 50.9643°, S 37.6174°, ~1985 m). After sample collection, the TVG4 and TVG11 samples were frozen at − 20 °C immediately before being processed.

Methods
Mineralogical and geochemical analysis. Mineralogical and geochemical characteristics of the two samples were performed by X-ray diffraction phase, scanning electron microscope, carbon and oxygen isotope, and trace element analyses at China University of Geosciences at Beijing in 2013 22 . Thermal cycling consisted of initial denaturation at 98 °C for 1 min, followed by 30 cycles of denaturation at 98 °C for 10 s, annealing at 50 °C for 30 s, and elongation at 72 °C for 60 s, finally 72 °C for 5 min. The PCR products were analyzed on 2% agarose gel, and the DNA (~300 bp) was purified with GeneJET Gel Extrac-tion Kit (Thermo Scientific).

DNA extraction and PCR amplification.
16S rDNA library preparation and sequencing. Sequencing libraries were generated using NEB Next ® Ultra ™ DNA Library Prep Kit for Illumina (New England Biolabs) following manufacturer's recommendations and different multipex indexing barcodes were added. The library quality was assessed on the Qubit@ 2.0 Fluorometer (Thermo Scientific) and Agilent Bioanalyzer 2100 system. The libraries were sequenced on an Illumina MiSeq platform 2500 and 250 bp paired-end reads were generated at Novogene (Beijing, China). Complete data were submitted to the NCBI Short Read Archive database under accession no. SRP062532 for bacterial sequences and no. SRP063077 for archaeal sequences.
Quality filtering, OTUs picking and annotation. Raw data obtained from the Illumina MiSeq sequencing platform must be processed because they contain some low quality data that will interfere with analysis results. Based on the sequences of barcodes, raw data were demerged and then used to merge paired-end reads of each sample by using FLASH software 53 , when at least some of the reads overlap the read generated from the opposite end of the same DNA fragment. Paired-end reads was assigned to each sample according to the unique barcode. Barcode sequences and PCR primer sequences were also removed. The resulting sequences were defined as raw tags. The merged raw tag was strictly filtered and developed into clean tags with high quality. By means of QIMME (quantitative insights into microbial ecology) 54 , these clean tags were further processed. The site of first base from consecutive data with low quality (the default threshold value is < = 3) will be truncated if the base number is up to the defined length (the default length is 3). The truncated tags were developed into tags group. Among tags, some tags that contain base length with consecutively high quality is shorter 75% of tags length were further filtered. After processing mentioned as above, the obtained tags were blasted with Unite Database to detect chimera sequence by UCHIME Algorithm 55 . The chimera sequence was finally removed, and the effective tags were generated 56 .
During OTUs construction, the data of effective tags and tags with low quality, and tags annotated were analyzed. Effective tags with ≥ 97% similarity were clustered by Uparse 57 and were classified into one OTU. For species analysis, sequences with ≥ 97% similarity were assigned to the same OTUs using Uparse, and similarity hits below 97% were not considered for classification purpose. A representative sequence of each OTU was picked out and the taxonomic information was annotated using RDP classifier 58 and GreenGene database 59 . Diversity analysis. QIIME software package was used to analyze alpha (within samples) and beta (among samples) diversity. In alpha diversity analysis, the OTU tables were rarified and three metrics were calculated: Chao1 estimates the richness of microbial species; Observed Species estimates the amount of OTUs in each sample, and Shannon index estimates the diversity of microbial species. Based on these three metrics, rarefaction curves were generated. In beta diversity analysis, QIIME software package was also used to calculate both weighted and unweighted unifrac distance, which generates the beta diversity index.