|Home | About | Journals | Submit | Contact Us | Français|
We present evidence for a dimorphic life cycle in the vacuolate sulfide-oxidizing bacteria that appears to involve the attachment of a spherical Thiomargarita-like cell to the exteriors of invertebrate integuments and other benthic substrates at methane seeps. The attached cell elongates to produce a stalk-like form before budding off spherical daughter cells resembling free-living Thiomargarita that are abundant in surrounding sulfidic seep sediments. The relationship between the attached parent cell and free-living daughter cell is reminiscent of the dimorphic life modes of the prosthecate Alphaproteobacteria, but on a grand scale, with individual elongate cells reaching nearly a millimeter in length. Abundant growth of attached Thiomargarita-like bacteria on the integuments of gastropods and other seep fauna provides not only a novel ecological niche for these giant bacteria, but also for animals that may benefit from epibiont colonization.
The bacterial domain includes representatives with a variety of cell shapes (Young, 2006) and types of cell division (Angert, 2005). Bacterial cell morphologies can provide selective advantages ranging from increased motility to decreased predation susceptibility (Young, 2006). The enormous size of the sulfide-oxidizing Gammaproteobacterium, Thiomargarita namibiensis, is thought to provide it with the ability to store nitrate for months, persisting in environments in which availability of the primary terminal electron acceptor oxygen, or the alternative acceptor nitrate, can be temporally and spatially variable (Schulz and Jørgensen, 2001; Schulz, 2006). However, the influence of cell morphology on the ecophysiology of these giant sulfur bacteria may not be limited to their large size.
Thiomargarita was first described in 1999 from diatom-rich oozes in the Benguela upwelling system off West Africa (Schulz et al., 1999). The originally described species T. namibiensis occurs primarily as linear chains of spherical cells enclosed in a mucous sheath (Schulz et al., 1999), though other morphotypes have since been described (Schulz, 2006; Gutzke, 2007). In 2005, Thiomargarita-like bacteria that undergo reductive cell division resulting in symmetrical clusters of up to 16 cells were discovered at hydrocarbon seeps from the Gulf of Mexico (Kalanetra et al., 2005; Bailey et al., 2007). Other Thiomargarita-like bacteria have also been obtained from mud volcanoes in the Barents Sea (de Beer et al., 2006) and the Eastern Mediterranean Sea (Girnth et al., 2011). During a 2009 cruise aboard the R/V Atlantis, we observed a population of Thiomargarita-like cells in cores and on carbonate rock samples collected with the DSV Alvin at methane seeps associated with fault scarps, landslides and mud mounds (Bohrmann et al., 2002) at water depths of ~990–1600m along the Pacific margin of Costa Rica. Within the Thiomargarita-like population present at Costa Rica margin seeps are canonical free-living, solitary and reductively dividing forms, as well as cells exhibiting an attached, presumptive dimorphic life cycle that we describe here. Similar populations of free-living and attached Thiomargarita-like cells were observed in Hydrate Ridge (off Oregon, USA) seep settings in 2010. The observation of dimorphic forms at multiple localities suggests that a dynamic morphology has an important role in the ecology of vacuolate sulfur bacteria in seep settings.
Samples were obtained during a research expedition to the Pacific margin of Costa Rica aboard the R/V Atlantis in February–March 2009 (AT15-44) and during a follow-up cruise to the same sites in January 2010 (AT15-59). Push cores, samples of authigenic carbonate and associated macro- and megafaunal animals were collected from methane seeps associated with mud mounds and fault scarps at Mound 11 (1000–1025m, 8° 55.4′N, 84° 18.24′W), Mound 12 (990–1100m, 8° 55.8′N, 84° 18.75′W), Jaco Scarp (1775–1850m, 9° 7.00′N 84° 50.51′W) and Mound Quepos (1030–1179m, 8° 58.09′N, 84° 37.95′W) using the DSV Alvin. Additional collections of living gastropods with attached sulfur bacteria were made during AT 15-68 at Hydrate Ridge North (44° 40′N 125 5.8′W) and Hydrate Ridge South (44° 34.1′N 125° 9.1′W).
During the research expedition, a Leica S8 APO stereo-microscope coupled with a Canon Powershot 9 (Leica, Buffalo Grove, IL, USA) was used to observe and image freshly collected free-living cells, as well as those cells attached to byssal threads, shells and other benthic substrates. Cells that were budding at the time of collection were followed for a 10-day duration of the cruise and imaged on a daily basis in order to document bud detachment. In the laboratory, an Olympus SZX16 stereo-microscope and an Olympus BX51 compound microscope equipped with an Olympus DP72 color camera were used to obtain higher resolution images of these cells and attachment substrates (Olympus, Tokyo, Japan).
Gastropods covered with attached Thiomargarita-like sulfur bacteria were preserved in Karnovsky's fixative (Karnovsky, 1965) before undergoing ethanol series dehydration (25%, 50%, 75%, 90%, 99%) in artificial seawater/ethanol mixtures for 30min each. The bacteria-covered gastropod shells were then dried in a Tousimis Model 780A Critical-Point Dryer (Tousimis, Rockville, MD USA) before being imaged uncoated with a Hitachi TM-1000 tabletop scanning electron microscope (Hitachi, Tokyo, Japan).
For transmission electron microscopy imaging, samples were fixed with Karnosky's fixative solution and after three times washing with artificial seawater, the samples were post fixed with 1% OsO4 for 1h. Following three times rinsing with H2O, the samples were dehydrated in a graded series of ethanol (see above). The samples were embedded in Epon 812 resin (Electron Microscopy Sciences, Fort Washington, PA, USA). Ultrathin sections (65nm) were produced using a Reichert UltraCut S Ultramicrotome (Reichert Technologies, Depew, NY, USA) and stained with uranyl acetate and lead citrate before being examined with a JEOL 1200EX electron microscope (JEOL Inc., Tokyo, Japan).
Free-living Thiomargarita-like cells from Costa Rica and Namibia were removed from the sediment (Salman et al., 2011). The novel attached morphotype from Costa Rica were carefully scraped off byssal threads with two needles and the detached cells were removed from the sample as described above. Individual cells, that is, free-living cells and formerly attached cells sometimes showing distal budding, were washed in artificial seawater, photographed under the microscope and subsequently used for performing single-cell 16S rRNA PCR methods described elsewhere (Salman et al., 2011). The specific primer for Thiomargarita-related organisms, VSO233F, 5′-CCTATGCCGGATTAGCTTG-3′ (Escherichia coli 16S rRNA position 233–251, Salman et al., 2011), was combined with the universal primer ITSReub, 5′-GCCAAGGCATCCACC-3′ (E. coli 23S rRNA position 23–38; Cardinale et al., 2004). Some attached cells did not reveal any PCR product with these primers, thus another specific forward primer was developed based on Thiomargarita sequences from the public databases: VSO360F, 5′-GGAATATTGGACAATGGGC-3′ (E. coli 16S rRNA positions 361–379, developed in this study), which was used in combination with primer VSO1300R, 5′-ATCCGGACTACGAGTAG-3′ (E. coli 16S rRNA position 1291–1307; Salman et al., 2011).
The amplification products were analyzed by agarose gel electrophoresis, bands of ~1750bp length (and ~940bp length with the other primer pair) were excised and DNA was purified with a gel extraction kit (Qiagen, Hilden, Germany). PCR products were sequenced directly using the Big Dye Cycle Sequencing Kit and analyzed on the ABI Prism Genetic Analyzer 3130x (Applied Biosystems, Carlsbad, CA, USA). The seven 16S RNA Thiomargarita-affiliated sequences have been submitted to the DDBJ/EMBL/GenBank databases. The accession numbers are FN811658–FN811664. The internal transcribed spacer region (spacer region between 16S rRNA and 23S rRNA as part of the PCR product when using primer ITSReub) was removed before further analysis of sequences.
The phylogenetic affiliation of the retrieved sequences was inferred with the ARB software package (Ludwig et al., 2004) based on release 95 of the SILVA SSURef database (Pruesse et al., 2007). The alignment of all sequences was automatically performed with the integrated aligner and checked manually afterwards, taking into account the secondary structure of the 16S rRNA molecule. For tree reconstruction, a total of 90 nearly full-length sequences, including representatives of the Gamma- and Deltaproteobacteria, were used. Trees were calculated including nucleotide positions between 279 and 1463 (according to E. coli numbering) and performing neighbor joining, maximum parsimony and maximum likelihood analyses, each with the application of 0%, 30% and 50% positional conservatory filters excluding highly variable regions. Partial sequences (AY632420, AF129012, FN811664) were inserted into each of the reconstructed trees by parsimony criteria without allowing changes in the overall tree topology. A consensus tree was manually produced based on the maximum likelihood tree calculated with a 30% filter using an approach described in (Peplies et al., 2008). The branching patterns of all trees were compared and multifurcations were introduced at nodes where the exact tree topology could not be clearly resolved (Peplies et al., 2008). The tree presented in Figure 5 shows a subset of the sequences originally used for calculation. Neighbor-joining and maximum likelihood trees are included in the supplement for comparison (Supplementary Figure S1 and S2).
Free-living Thiomargarita-like cells were observed in sulfidic sediments at several mud mounds and associated methane seeps (Figure 1a). As in sediments from off Namibia (Schulz et al., 1999), and in the Gulf of Mexico (Kalanetra et al., 2005), free-living sulfur bacteria off Costa Rica resembling Thiomargarita were most abundant in white and gray mat associated sediments in the uppermost 3cm beneath the sediment/water interface. At some localities, cell populations consisted primarily of single Thiomargarita-like cells mixed with large Beggiatoa-like filaments (Figure 1a), whereas at other locations, symmetrical cell clusters (not shown) predominated in gray flocculant material at the seabed. These cell clusters contained 2, 4, 8 and 16 cells, similar to those produced by reductive division in Thiomargarita sp. from the Gulf of Mexico (Kalanetra et al., 2005). In the vicinity of these sediments, rich in free-living spherical Thiomargarita-like cells, thin films of microbial biomass were found attached to methane-derived authigenic carbonates (Figure 1b), byssal threads of the mussel Bathymodiolus spp. (Figure 1c), fragments of organic detritus, and most commonly, the integuments of seep-endemic invertebrates (Figures 2a–c). Microscopic examination of this benthic biofilm material revealed densely packed attached sulfur-storing vacuolated cells that closely resemble Thiomargarita, but with most cells elongated into a stalk-like form. Individual cells occasionally reached nearly 700μm in length, but average elongate cells were 338±169μm long (Supplementary Figure S3). These elongate cells were observed to bud spherical Thiomargarita-like cells from their unattached pole (Figures 3 and and4).4). The attached poles of the elongate cells ranged from rounded (Figures 3 and and4)4) to a heavily tapered, and sometimes extruded shape, forming a short appendage (Figure 4c). Both the attached elongate cells and the buds consisted of a single thin layer of sulfur globule-containing cytoplasm surrounding a single large vacuole (Figure 3d and transmission electron microscopy observations not shown)—a basic cell structure that they share with all previously described Thiomargarita ecotypes (Schulz et al., 1999; Kalanetra et al., 2005; Schulz, 2006). Elongate morphotypes were not observed in the unattached free-living state. We hypothesize that the attached, budding types, and at least some of the free-living cells, represent not different species, but different stages of a reproductive developmental cycle. We evaluated this hypothesis by directly observing the budding forms after collection. Furthermore, we analyzed the 16S rRNA gene sequences of the new, attached morphotypes, as well as of the free-living canonical morphotypes. Evidence for a developmental relationship between the attached cells and free-living spherical cells was provided by direct observation of buds formed at the time of collection completely detaching from budding parent cells at a rate of approximately one detachment in 300 budding cells over a 24h period (12 in approximately 3600 individuals observed), demonstrating that the budded daughter cells are indeed being released into the surrounding environment (Figures 3b, c, f). Fully detached, budded daughter cells were morphologically identical to free-living, spherical sulfur bacteria in surrounding sediments and both have similar cell diameters; 71±22μm, n=40 versus 88±14μm, n=37. The morphological similarity of the released buds to the free-living Thiomargarita-like cells in the surrounding sediments suggests the possibility that at least some of the latter population results from the released spherical buds. However, morphology can be misleading, particularly with respect to the phylogenetic identification of vacuolated sulfur bacteria. For example, marine filaments with a Beggiatoa-like morphology, that have similar filament diameters and were collected from the same locality, can be quite divergent with respect to their phylogenetic positions (Teske et al., 1999; Mußmann et al., 2007). The observations that the attached elongate cells are found on hard substrates that are physically separated by sediments that contain spherical, free-living cells, and that elongate cells are never observed in the free-living populations, strongly suggests that at least some of the free-living population includes the spherical offspring of the attached morphotypes. In addition, the assertion that the life cycle is initiated by the attachment of spherical cells is supported by the observation of attached spherical cells that are just beginning elongation (Supplementary Figure S4), along with the spectrum of elongation morphologies from sphere, to prolate spheroid, to increasingly longer cylindrical cells occurring in close proximity (Figure 4d). However, the percentage of free-living cells that represent the progeny of attached budding forms remains unclear.
Sequence comparisons of individual cells indicated that Costa Rica margin seeps harbor not just one, but at least three distinct phylogenetic clusters. The relatedness of the organisms in this study to Thiomargarita namibiensis and Thiomargarita sp. from the Gulf of Mexico is apparent from the phylogenetic tree produced by sequencing individual representatives from Costa Rica and additional specimens from Namibia (Figure 5). The sequences published before this study constitute the closest relatives to at least five sequences obtained in this study. Although some sequences of free-living and attached morphotypes are closely related to previously published Thiomargarita spp. (Kalanetra et al., 2005, Schulz et al., 1999), two sequences (COS001 and COS016) each formed a separate branch that did not cluster within the Thiomargarita clade. The paraphyletic relationship of the retrieved 16S rRNA gene sequences surely complicates identification of cells based on morphology. Nomenclature for Thiomargarita-like cells and other vacuolate sulfur bacteria can be confusing, as some canonical genera include paraphyletic taxa. However, the taxonomy of the colorless sulfur bacteria, including the novel morphotypes identified in this study, was recently revisited by Salman et al. (2011).
Sequences in the three clusters represent either attached, free-living or paired free-living/attached morphotypes (Figure 5). Some sequences showed >99% similarity between attached and free-living morphotypes (Table 1) that may represent different developmental stages of the same organism, as the <1% difference is within the range of experimental error. Alternatively, these cells could represent closely related phylotypes that superficially resemble parts of the putative dimorphic life cycle, but do not represent the direct progeny of the attached cells. The disparity could be caused by natural variability within the seep population coupled with the relatively low number of sequenced specimens, and the inherent limitations of sampling a deepwater site via submersible, thereby possibly missing directly corresponding mother and daughter cells. It is also unclear whether the reductively dividing free-living sulfur bacteria present in nearby sediments exhibit attachment or dimorphism as part of their life cycle(s).
Reproductive dimorphism is known in the Alphaproteobacteria, such as Caulobacter crescentus, whose parent cells are stalked, and the daughter cells are vibroid—but in those organisms the daughter cells are actively motile (Brun and Janakiraman, 2000; Poindexter, 2006). In the sulfur bacteria detected here, the daughter cells are not yet known to be actively motile—though an inconspicuous sporadic slow rolling motility has been observed in different, spherical sulfur bacteria from Namibia (Salman et al., 2011). In addition, free-living daughter cells can be readily transported by bottom currents. Broken, transported and re-deposited clasts of methane-derived authigenic carbonate edifices observed on the collection cruises, as well as evidence for local sediment winnowing, record the presence of such currents at the sites where the attached Thiomargarita-like morphotypes occur. On detachment, buds may also attach to substrates directly adjacent to the mother cell. This process is known to occur in some prosthecate Alphaproteobacteria and would explain the observation that mats of attached Thiomargarita-like sulfur bacteria are often densely packed and frequently exhibit more than one developmental stage on the same substrate (Figure 3b). Alternatively, certain cells in the mat may simply have slower elongation and budding rates than others.
The factors responsible for initiating attachment, elongation and budding in these bacteria are not fully understood because Thiomargarita and its vacuolated relatives have yet to be cultivated in the laboratory. However, comparisons with other organisms that exhibit similar behaviors may provide clues. Attachment to solid substrates is also known in other Gammaproteobacteria, including the closely-related marine, vacuolated Thiothrix-like bacteria (Kalanetra et al., 2004, Kalanetra and Nelson 2010). Indeed, attached filamentous Thiothrix-like filaments sometimes co-occur with the attached Thiomargarita-like cells described here. Attachment provides stabilization of a parent cell under conditions in which it can produce daughter cells for dispersal to other potential habitats via bottom currents. Presumably, stabilization of reproductively active sulfide-oxidizing cells over zones of sulfide enrichment would be more important in advective seep habitats where sulfide generated by the anaerobic oxidation of methane or vented directly from subsurface fluids is more localized than the sulfide generated by sulfate reduction in upwelling regions (Linke et al., 2005).
Budding is known in Thiothrix-like filaments (Kalanetra et al., 2004), in certain low-GC gram-positive bacteria, cyanobacteria, planctomycetes and prosthecate Alphaproteobacteria (Angert, 2005) among others. In prosthecate bacteria, it is the stalked parent cell that undergoes budding, suggesting that perhaps elongation enhances dispersal of daughter cells. However, elongation is apparently not a prerequisite to budding in attached Thiomargarita-like cells, as division of attached but not elongated cells, or slightly elongated cells, was observed on a handful of substrates (Figure 1b). Access to scarce metabolites may instead drive elongation in attached sulfur bacteria. Cell elongation in Caulobacter crescentus is thought to enhance the uptake of phosphorous from the environment by increasing the surface area-to-volume ratio of a cell (Poindexter, 1984). Thiomargarita namibiensis, which lives in sediments beneath upwelling zones, sequesters phosphorous as intracellular polyphosphate granules (Schulz and Schulz, 2005). In seep settings, phosphorous is less available than in an upwelling environment (Joye et al., 2004; Linke et al., 2005). Thus, cell elongation in seep-dwelling sulfur bacteria may provide for enhanced phosphorous uptake as it does in the Alphaproteobacteria. Nutrient uptake and adenosine triphosphate synthesis scales with cell surface area, and Thiomargarita functions at a relatively low level of metabolic power per gene (Lane and Martin, 2010). An elongate cell has a higher surface area-to-volume ratio than a spherical cell of the same volume (Young, 2006). For example, if a cell was assumed to be a cylinder capped with hemispherical ends, 300μm in length and 80μm in diameter, the cell would have a 38.5% greater surface area-to-volume ratio (0.054) over a sphere with the same volume (0.039). The larger a Thiomargarita cell grows, the larger its potential store of nitrate, but at the cost of decreasing its surface area-to-volume ratio. Growth of a cell into an elongate form, as opposed to simply expanding the sphere volume, minimizes the decrease in surface area-to-volume ratio that accompanies growth. Reductive cell division, which is observed in free-living Thiomargarita-like cells at this site, also effectively increases surface area while maintaining a constant nitrate-storage volume, although by producing smaller cells.
The elongation of budding cells may also allow attached cells greater access to the terminal electron acceptors, oxygen and nitrate. It is not known whether attached and free-living reductively dividing Thiomargarita-like bacteria represent different stages within the same life cycle. However, one possibility is that elongation increases surface area in attached cells that have access to nitrate and can readily enlarge, whereas reductive division apportions stored nitrate into smaller, easily dispersible cells with higher surface area–volume ratios when nitrate availability is limited. Closely related vacuolate sulfur bacteria resembling Beggiatoa and Thioploca can actively move along gradients between sulfide and terminal electron acceptors (Nelson et al., 1986; Huettel et al., 1996). In seep settings, alternating conditions of access to electron donor and acceptor can be introduced by brine flows (Girnth et al., 2011), changing currents, and heterogeneous pore fluid advection, exposing populations of sulfur bacteria to alternating conditions of high sulfide, nitrate and oxygen fluxes. Elongation in sessile bacteria may provide cells with a larger surface area for more efficient uptake of spatially segregated metabolites, under the fluctuating and heterogeneous fluxes of O2, NO3− and H2S measured in Costa Rica margin sediments (Linke et al., 2005). In some freshly collected samples, elongate cells possess a visibly marked increase in sulfur globules at the distal end of the cell. The apical, sulfur-enriched region is the site of bud formation. These sulfur resources might be incorporated into the budding offspring, providing an initial energy reserve for the free-living state. Alternatively, the irregular distribution of sulfur globules may be driven by the local geochemical environment, where the observed sulfur accumulation at the apical end of the cell is linked to close contact with either oxygenated, or sulfidic waters, depending on the orientation of the cell. However, this explanation seems less likely, as cells attached to opposite sides of byssal threads also showed apical sulfur enrichment in cells oriented in opposite directions.
The attachment of newly detected sulfur bacteria to benthic substrates also has potentially important implications for metazoan ecology at seeps. Elongate sulfur bacteria were found attached to byssal threads of Bathymodiolus spp. (Mytilidae, Bivalvia), the carapace of a Yeti crab specimen (Kiwa sp.), the shells of several species of gastropod including Provanna laevis, Hyalogyrina sp. and Pyropelta spp., and the limpet Lepetodrilus sp. These gastropod shells were commonly covered with thick mats of attached elongate sulfur bacteria (Figure 2). Interestingly, bacteria-covered Provanna collected from Hydrate Ridge in 2010 exhibited an unusual behavior that may benefit the attached sulfide-oxidizing bacteria. When placed in a container with seawater overlying sulfidic sediments, they were observed to either climb to the uppermost portion of the jar and maintain a position adjacent to the small air headspace (n=4), or actively dig a burrow into the sulfidic sediments (n=8) and subsequently inverting themselves, so that their shell and attached bacteria were positioned within the freshly dug burrow (Figure 6). Gastropods removed from the burrows with forceps repeatedly re-established their burrows and inverted positions (n=6). Epibiont-bearing invertebrates are known to migrate between oxygenated and sulfidic sediment pore waters as a means of supplying symbionts with the metabolic substrates required for sulfide oxidation (for example, Ott et al., 1991), and a similar behavior may be on display here. However, this behavior and association with seep invertebrates is apparently not a requirement for growth of the attached Thiomargarita-like cells, as these bacteria are sometimes found on stabilized mineral and detrital substrates such as carbonate rocks. The impetus for gastropods to exhibit this behavior that potentially exposes them to predation while lying inverted with their soft parts exposed remains unclear. The bacteria may reduce the animal's exposure to hydrogen sulfide as is thought to occur in other invertebrates (Alain et al., 2002). These species of gastropods are not known to contain sulfide-oxidizing endobionts. Attached bacteria may also serve as food source for grazing gastropods, as occurs in some other animal/epibiont symbioses (for example, Ott et al., 1991; Zbinden and Cambon-Bonavita, 2003; Gillian and Dubilier, 2004; Dattagupta et al., 2009), although these gastropods are seemingly incapable of consuming the bacteria attached to their own shells.
At Costa Rica seep settings, attached sulfur bacteria are rare on rocks and shells with large surface areas. The substrates most well colonized by the cells are those with low surface areas such as small shells and mussel byssal threads. When they do occur on larger substrates, they are found in small isolated patches or within small crevices. The apparent absence of attached bacteria on Bathymodiolus mussel shells, Lamellibrachia tubes, and most carbonate rocks suggests that they may be limited to surfaces that are not subject to intense shearing forces or grazing by gastropods. The shells and carapaces of the grazers themselves and cryptic habitats, such as mussel byssal threads, may offer a refuge from consumers. Alternatively (or additionally), the availability of sulfide, nitrate and oxygen may make certain microenvironments more conducive to colonization by attached sulfur bacteria. High oxygen and nitrate fluxes, primarily driven by bacterial oxidation of sulfide (781 and 700μmol cm−2 per year, respectively), have been reported at the collection sites, but these fluxes vary by up to three orders of magnitude over distances of tens of centimeters as a result of heterogeneities in seep fluid flow (Linke et al., 2005). Whether these attached bacteria show preferential attachment to specific substrates that guarantee access to sulfide or other metabolites, or if their distribution is influenced by the very abundant microbial grazers in the system remain open questions.
Examination of photographs and specimens of seep fauna collected during 2005 and 2006 cruises to the Eel River Basin and Hydrate Ridge (off the coast of California and Oregon, USA), and observations of gastropods collected from Hydrate Ridge in 2010 both show inhabitation by morphologically similar elongate Thiomargarita-resembling cells (Figure 2b), suggesting that certain ecotypes of attached sulfur bacteria are geographically widespread in seep settings. The difficulty in collecting these specimens and maintaining them under enrichment conditions that allow for observation of their development renders our understanding of these organisms incomplete. However, the dimorphic life cycle in seep-associated cells closely related to Thiomargarita spp. described here reveals a complex reproductive ecology for these giants of the bacterial world that further confounds the generalization that bacteria are small and simple.
This study was funded by the Agouron Institute, the University of Minnesota Department of Geology and Geophysics, the Max Planck Society, and the National Science Foundation (OCE-0826254 and OCE 09-39557). Parts of this work were carried out in the Characterization Facility, University of Minnesota, a member of the NSF-funded Materials Research Facilities Network (http://www.mrfn.org) via the MRSEC program with assistance from Chris Frethem and Fang Zhou. Scanning electron microscopy was performed in the University of Minnesota, Twin Cities, Limnological Research Center with the support of Amy Myrbo. Furthermore, we thank Tony Rathburn, Shana Goffredi, Anders Waren, Danwei Huang, Anne Dekas, Olivia Mason, Ben Harrison, the crew of the R/V Atlantis and pilots of the DSV Alvin (15–44 and 15–59) for assistance at sea, and Ben Grupe for providing live gastropods from Hydrate Ridge. We thank the reviewers of this manuscript for valuable comments that helped improve the manuscript.
The authors declare no conflict of interest.
Supplementary Information accompanies the paper on The ISME Journal website (http://www.nature.com/ismej)