Mixotrophic chemosynthesis in a deep-sea anemone from hydrothermal vents in the Pescadero Basin, Gulf of California

Numerous deep-sea invertebrates, at both hydrothermal vents and methane seeps, have formed symbiotic associations with internal chemosynthetic bacteria in order to harness inorganic energy sources typically unavailable to animals. Despite success in nearly all marine habitats and their well-known associations with photosynthetic symbionts, Cnidaria remain one of the only phyla present in the deep-sea without a clearly documented example of dependence on chemosynthetic symbionts. A new chemosynthetic symbiosis between the sea anemone Ostiactis pearseae and intracellular bacteria was discovered at ~ 3700 m deep hydrothermal vents in the southern Pescadero Basin, Gulf of California. Unlike most sea anemones observed from chemically reduced habitats, this species was observed in and amongst vigorously venting fluids, side-by-side with the chemosynthetic tubeworm Oasisia aff. alvinae. Individuals of O. pearseae displayed carbon, nitrogen, and sulfur tissue isotope values suggestive of a nutritional strategy distinct from the suspension feeding or prey capture conventionally employed by sea anemones. Molecular and microscopic evidence confirmed the presence of intracellular SUP05-related bacteria housed in the tentacle epidermis of O. pearseae specimens collected from 5 hydrothermally active structures within two vent fields ~ 2 km apart. SUP05 bacteria (Thioglobaceae) dominated the O. pearseae bacterial community, but were not recovered from other nearby anemones, and were generally rare in the surrounding water. Further, the specific Ostiactis-associated SUP05 phylotypes were not detected in the environment, indicating a specific association. Two unusual candidate bacterial phyla (the OD1 and BD1-5 groups) appear to associate exclusively with O. pearseae and may play a role in symbiont sulfur cycling. The Cnidarian Ostiactis pearseae maintains a physical and nutritional alliance with chemosynthetic bacteria. The mixotrophic nature of this symbiosis is consistent with what is known about other cnidarians and the SUP05 bacterial group, in that they both form dynamic relationships to succeed in nature. The advantages gained by appropriating metabolic and structural resources from each other presumably contribute to their striking abundance in the Pescadero Basin, at the deepest known hydrothermal vents in the Pacific Ocean.


Background
Numerous deep-sea annelids, mollusks, and other invertebrates have forged relationships with bacteria in order to harness inorganic sources of energy that are typically unavailable to most animals. Microbial chemosynthesis generates energy through the oxidation of sulfide, as an example, used to fuel the production of organic carbon, which can be shared with a receptive animal host. To date, members of at least six major animal clades, including most recently Trichoplax (Placozoa), have formed symbiotic associations with internal chemosynthetic bacteria [1,2]. Interestingly, Cnidaria, although well-known to host photosynthetic symbionts, remains one of the last prominent animal clades without a documented metabolic dependence on chemosynthetic bacterial symbionts for survival in hydrothermal vents.
Anthozoa, which includes sea anemones and corals, is one of the most successful and diverse groups of Cnidaria. They are found in all marine habitats at most depths and latitudes [3]. Their worldwide ecological success may best be attributed to an ability to form symbiotic relationships with other organisms, including microbial eukaryotes (e.g., dinoflagellates in the family Symbiodiniaceae; [4]), as in the case of shallow-water tropical species. Anthozoa such as sea anemones, octocorals, and zoanthids are also found in deep-sea reducing environments, such as hydrothermal vents, seeps, and whalefalls [5][6][7][8]; however, they have been historically understudied, and most remain undescribed. It would be reasonable, and perhaps even expected, for some of these deep-sea Anthozoa to also host microbial symbionts. In fact, a recent study demonstrated an affiliation between sulfide-oxidizing SUP05 bacteria and certain species of Anthozoa found near deep-sea seeps, and hypothesized a facultative symbiosis based on molecular and isotopic evidence [9].
The recently discovered Pescadero Basin vent field at 3700 m depth in the southern Gulf of California differs markedly from nearby vent localities (e.g., Guaymas Basin and 21°N East Pacific Rise) in physical, chemical, and biological attributes [10][11][12]. In particular, the vents in the Pescadero Basin are uniquely composed of hydrothermal calcite, with venting fluids that contain high levels of aromatic hydrocarbons, hydrogen, methane, and hydrogen sulfide at a pH of~6.5 [11]. The Pescadero Basin vents are also highly unusual in faunal composition with many new species and numerous others that do not occupy nearby regional vents (e.g., Alarcon Rise vents; [11,13,14]). Included in this group of unusual fauna was a very abundant white sea anemone (up to 68 individuals m −2 in some areas) that occurred in and amongst the siboglinid tubeworm Oasisia aff. alvinae, often very near to actively venting fluids (Fig. 1, [11]).
Previously, several unidentified Pescadero Basin Actiniaria (sea anemones) were reported to be quite depleted in tissue δ 13 C values (− 33 to − 38‰ [11,15]), compare to anemones that feed via traditional suspension feeding or prey capture. This evidence, along with their unusual life position and abundance in zones of active fluid venting, hinted at their possible nutritional reliance on chemoautotrophic carbon production, as opposed to traditional suspension feeding or prey capture via cnidae; however, the specific details were not explored further. Here, by combining microbial community profiling, ultrastructural analysis via microscopy, and stable isotope measurements, we document the only known species of chemosynthetic sea anemone at vents deep in the Gulf of California, identified as Ostiactis pearseae (previously known only from whalefalls [5]). This dense population of O. pearseae likely receives nutritional supplementation of carbon, nitrogen, and sulfur via intracellular bacteria within the SUP05 clade, housed in their epidermis, albeit direct evidence of nutrient transfer remains outstanding.

Results
Actiniaria of various morphotypes were observed to be one of only a handful of dominant animal species in both the Pescadero Basin Auka vent field [11,12] and the newly discovered JaichMaa 'ja'ag vent field, both within~2 km of each other in the Gulf of California (Fig. 1). A conspicuous white actiniarian species represented a significant fraction of the animal community and was collected from 5 vent edifices in zones of active venting, in the immediate vicinity of the obligate vent tubeworm, Oasisia aff. alvinae (Fig. 1). Several other sea anemones (by morphotype) were observed and collected near these same sites, usually in areas of less active fluid flow ( Table 1). The white actiniarian morphotype was identified as Ostiactis pearseae [5], based on anatomical, cnidae, and DNA sequencing of preserved polyps. The Pescadero Basin populations of O. pearseae showed slight differences in morphology and cnidae to the description of specimens from the type locality and, thus, an amendment to the species diagnosis is provided. Diagnosis: (amended after [5,7], modifications in italics). Ostiactinidae with basilar muscles and mesogleal marginal sphincter. Body with well-developed base. Column not clearly divisible into scapus and scapulus; scapus without cuticle, maybe with scattered demarcated suckers distally; column without cinclides or with a distal row of round papillae with inconspicuous cinclides. Tentacles regularly arranged, not thickened on the aboral side. Six pairs of perfect and fertile mesenteries, hexamerously arranged, not divisible into macro-and micro-cnemes. Same number of mesenteries proximally and distally. Retractor muscles weak but restricted. No acontia. Some populations with chemosynthetic bacteria in tentacles. Cnidom: Robust spirocysts, basitrichs, holotrichs, and p-mastigophores A and B1. Ostiactis pearseae had been previously collected only in deep-sea waters (2800 m depth) of the Eastern Pacific, at a whalefall habitat in Monterey Bay [5]. Newly collected specimens are from deep-sea waters (3655-3692 m depth) associated with Southern Pescadero Basin hydrothermal vents (Diane's vent) in the Gulf of California (Pacific Ocean).

Ostiactis pearseae species diagnosis -Amendment
Intrapopulation variability in morphology was observed in the Pescadero Basin Ostiactis pearseae specimens (Fig. 2). Differences were observed mainly in the abundance and categories of cnidae amongst specimens (particularly holotrichs in the column), but also in the presence of a distal row of papillae (with only basitrichs) associated with inconspicuous cinclides in two specimens (SO197-S2 and SO200-R2; Fig. 2). Because of the small size of the papillae, the relatively small sizes and state of contraction and preservation condition of the specimens, it is not definitive that this row of distal papillae is only present in these two individuals. Nevertheless, the rest of the morphological and molecular characters, as well as the cnidae data, from the specimens with distal papillae agree with those of the other specimens studied, suggesting that differences should be treated as intrapopulation variation. Morphologically, specimens of O. pearseae from the Pescadero Basin pos-sess~70 tentacles, compared to specimens of similar sizes from the type locality, described as having~100 tentacles [5], with some having poorly demarked suckers in the column (which were not observed in Pescadero Basin specimens). Additionally, the first and second cycles of mesenteries are fertile in the type specimens, with males observed brooding larvae internally in the tentacles [5]. Although the fertility of the first cycle could not be confirmed for the Pescadero Basin specimens, the second and third cycles were confirmed to be fertile, but no brooding individuals were identified. The previous implementation of a different cnidae terminology suggests conspicuous differences in cnidae types and sizes between specimens at the Monterey Canyon whalefall and Pescadero Basin ( Table 2), but a new more precise combined terminology used here allows for distinction within p-mastigophore capsules (i.e., p-mastigophores A and p-mastigophores B1). The types and size ranges of the original description and the newly collected specimens of O. pearseae mostly agree (with only slight variability in some size   Table 2). All molecular phylogenetic analyses, based on the concatenated mitochondrial 12S rDNA, 16S rDNA, COIII genes, and partial nuclear 18S rDNA gene, were congruent and revealed a well-supported clade comprised of specimens of Ostiactis pearseae from Pescadero Basin and the type locality of Monterey Canyon (Fig. 3). DNA sequences from the Pescadero Basin specimens were identical to the Monterey Canyon population for 3 of the genes analyzed and only differed from the type locality by 1-bp for the 16S rDNA gene. Ostiactis was recovered within Metridioidea, as sister to a weakly supported clade formed by deep-sea Actiniaria and those associated with chemosynthetic environments (e.g., clades Deepsina + Chemosynthina, as part of the family Kadosactinidae, sensu [7]), a relationship consistent over different studies (e.g., [7,[16][17][18]). Most representatives from these two clades are characterized by the loss of acontia (filament-like structures packed with nematocysts), the presence of which is a major synapomorphy for Metridioidea. The other actiniarian morphotype included here, identified as Kadosactinidae 'sp.B,' was recovered (only mitochondrial sequence data available for this specimen) sister to Alvinactis chessi, a sea anemone inhabiting hydrothermal vents in the southwestern Pacific ( Fig. 3; [6]).   Fig. 4a) collected at the same locality. By comparison, O. pearseae was significantly more negative in both tissue δ 13 C and δ 15 N than the seep-associated octocoral species Paramuricea sp. type B3 recently reported [9] (ANOVA p = 0.0013 and p < 0.00001, respectively, n = 21; Fig. 4a). Finally, O. pearseae had significantly lower δ 34 S tissue values of − 1.1 ± 6.4‰ (− 11.0 to 8.7‰, n = 7), compared to other Pescadero Basin sea anemones (8.9 ± 5.2‰; range 1.0 to 14.2‰, n = 7; ANOVA p = 0.008), with comparable total tissue sulfur~0.9-1.9% by weight (Fig. 4b).
Bacterial community analysis of Ostiactis pearseae from the Pescadero Basin In Ostiactis pearseae, the tentacles are smooth, tapering, and relatively long when extended, with features similar to most other anemones, including cnidocytes (or cnidoblasts), cnidocysts, and glandular cells, all common cellular components of sea anemone tentacles [21]. The presence of bacteria in the type specimens of O. pearseae from Monterey Canyon was not detected [5]; however, the unusual isotope signatures of the Pescadero Basin specimens [11,15] prompted a more careful examination of this possibility. Indeed, bacterial community analysis via 16S rRNA Illumina barcoding revealed a dominance of the gammaproteobacteria SUP05 clade (64-96% of the bacterial community), comprising 6 putative sulfide-oxidizing bacterial OTUs (= phylotypes, clustered at 99% similarity) associated with the tentacles of Pescadero Basin O. pearseae (n = 8 specimens; Fig. 5, Additional file 2: Table S2). This was in contrast to the sulfide-oxidizing gammaprotebacteria recovered from the nearby obligate vent tubeworms Riftia pachyptila and Oasisia aff. alvinae (100% identical to Candidatus Endoriftia persephone; Fig. 5a; [22]). The SUP05 clade was not detected in association with 6 other individual sea anemones (determined by morphotype or molecular sequencing), and the specific Ostiactis-associated SUP05 phylotypes were not detected in surrounding water samples (n = 3; Fig. 5). Three SUP05 OTUs comprised < 7% of the bacterial community in the surrounding water column (Fig. 5a), based on the semi-quantitative 16S rRNA barcoding, but were distinct, based on the 250-bp 16S rRNA Illumina barcode sequences (Additional file 3: Fig. S1 inset).
To further characterize the SUP05 in association with Ostiactis pearseae, a longer region of the 16S rRNA gene was amplified via direct PCR and sequenced. A 1334-bp long 16S rRNA sequence, only 1-bp different from barcode OTU21762, was 96.5% similar to a free-living bacterium from a mud volcano in the Eastern Mediterranean Sea (AY592908; [24]) and 95% similar to the thiotrophic symbiont of Bathymodiolus aff. brevior from Central Indian Ridge vents (DQ077891; [25]; Additional file 3: Fig. S1). There are no Bathymodiolus mussels at the Pescadero Basin vents, and the SUP05-related sequences recovered from O. pearseae are distinct from those recovered from Bathymodiolus from Costa Rica seeps, some of the closest known mussel populations (only~96% similar for 250-bp barcode sequence; Additional file 3: Fig. S1 inset; [26,27]).
Several genes were additionally amplified and directly sequenced in order to inform the possible metabolic capabilities of the SUP05-related bacteria in O. pearseae. The napA gene, encoding a catalytic subunit of the periplasmic nitrate reductase alpha subunit (E.C. 1.7.99.4; [28]), amplified from O. pearseae tentacles, was most closely related (82-85% similarity based on amino acid translation; 77% based on nucleotides), to the napA gene from known SUP05 bacteria, including those from an estuary (ACX30474; [29]) and the endosymbiont from Bathymodiolus mussel gill tissues (SMN16186). The aprA gene, encoding the adenosine phosphosulfate (or APS) reductase alpha subunit (E.C.1.8.99.2), recovered from O. pearseae, was most closely related to the aprA gene from the bacterial symbiont of a nematode (96% similarity based on amino acid translation, ACF93728) and the endosymbiont of Bathymodiolus septemdierum (81% similarity based on nucleotide sequence, AP013042; [30]).
Abundant SUP05 bacteria were observed embedded within the tentacle epidermis of Ostiactis pearseae (Fig. 6). Fluorescent in situ signal amplification via hybridization chain reaction-FISH (HCR-FISH) was necessary to overcome the very highly autofluorescent cnidae produced by the epidermis (Fig. 6g). HCR-FISH and TEM microscopy revealed intracellular cocci-shaped cells (~0.5 μm diameter), positioned just above or  15 nitrogen, and 34 sulfur isotope signatures for Ostiactis pearseae and comparison actiniarians. a δ 13 C and δ 15 N values for the tentacles of Ostiactis pearseae from the Pescadero Basin vents, compared to neighboring anthozoans, including unidentified zoanthids ("zoan") and sea anemone (Kadosactinidae 'sp.B'). Data for Pescadero Basin actiniarians collected in 2015 (from [11]; red checkered triangles) as well as seep-associated corals from the Gulf of Mexico ( [9]; purple circles) and unidentified anemones from Gorda Ridge hydrothermal vents ( [19]; purple squares) are also included. Data from [9] was extracted from their Fig. 7 using an online Web plot digitizer (http://arohatgi.info/WebPlotDigitizer/). b δ 34 S and tissue sulfide content (%, dry weight) values for the tentacles of O. pearseae from the Pescadero Basin vents, compared to neighboring Anthozoa, including an unidentified zoanthid ("zoan") and sea anemone (Kadosactinidae 'sp.B'). Bathymodiolus aduloides from muddy sediments off of Kakaijima Island taken from [20] immediately adjacent to cnidae capsules and nuclei (Fig. 6g, i, j). These cells were putatively identified as members of the SUP05 group, given the consistent overlap between cells hybridized using a general bacterial probe set (Eub338-I-III) and a probe designed specifically to target the O. pearseae SUP05 (Additional File 4: Fig. S2). Although poor tissue fixation somewhat compromised high-resolution electron microscopy (ex. many host cells were visibly ruptured with jumbled mitochondria), TEM provided additional evidence of symbiont integration within O. pearseae. Bacteria within the tentacles appeared concentrated in the periphery of cells within the mono-layered epidermis (Fig. 6i). Additionally, glands with large electron dense vesicles were observed occasionally between the very elongated bacteriacontaining cells (Fig. 6g) and a layer of mucous was observed overlying the epidermis in some instances. Bacteria on the tentacle surface occasionally appeared to be in clathrin-coated pits in various stages of endocytosis (Fig. 6l). For both microscopy methods, bacteria were not observed in either the gastrodermis or mesoglea of O. pearseae (Fig. 6g).

Discussion
A conspicuous actiniarian species, identified as Ostiactis pearseae [5], was dominant at two neighboring hydrothermal vent fields in the Pescadero Basin, Gulf of California. Unlike most vent anemones, which are almost always observed in the vent periphery, this species was found very near to vigorous venting fluids on and amongst the obligate vent tubeworms Oasisia aff. alvinae and Riftia pachyptila (Fig. 1), known to rely  exclusively on sulfide-based chemosynthesis for energy [31,32]. Ostiactis pearseae, formerly named Anthosactis pearseae [7], had been originally described as the first and only endemic Actiniaria from a whalefall community [5]; however, this discovery at hydrothermal vents makes them one of the only sea anemones described from multiple chemosynthetic environments [6,7]. The assumption, until now, was that most sea anemones at hydrothermal vents and methane seeps acquire nutrients via suspension feeding. There was no previous evidence that Ostiactis pearseae harbored chemosynthetic bacteria [5], suggesting that they fed upon dissolved and particulate organic matter and plankton. However, the significantly negative δ 13 C and δ 15 N tissue isotopic values of O. pearseae (at the time labeled as an unidentified species in [11,15]) suggested an entirely different strategy dependent upon bacteria chemosynthesis.
Indeed, distinct bacterial phylotypes related to the SUP05-group were associated with Ostiactis pearseae, compared to other nearby sea anemones and water column bacterial communities. This association was pervasive and dominant, in that SUP05 bacteria were found in all 8 O. pearseae specimens analyzed, comprising up to 96% of the recovered microbial 16S rRNA genes. Sulfuroxidizing bacteria within the SUP05 clade, named after the discovery in the Suiyo seamount plume [33], have been found worldwide in marine oxygen-deficient marine environments, deep-sea hydrothermal systems, and productive upwelling regions [34][35][36]. They exist both as free-living cells [29] and in association with animal hosts (ex. Bathymodiolus mussels and some sponges [37]), where they participate centrally in the provisioning of fixed carbon to the animal. Intracellular SUP05 were observed exclusively in the epidermis of Ostiactis pearseae, which was unexpected given that most Cnidaria house symbionts, mainly photosynthetic, in the gastrodermis [38,39]. Epidermal bacteriocyte-like structures containing Vibrio have been observed in Exaiptasia pallida [40] and bacterial "aggregates" containing Endozoicimonas have been observed in epidermal "caverns" in the sea anemone Metridium senile [41]. In both cases, however, the epidermal bacteria are pathogens commonly associated with animals [42,43]. Octocorals harbored a coral-specific SUP05 clade that dominated their bacterial communities by up to 87%, albeit the specific location of the bacteria within the coral mucous or tissue was not determined [9]. Interestingly, bacteria on the tentacle surface of O. pearseae appeared to be in clathrin-coated pits in various stages of endocytosis. Further examination of this receptor-mediated process is necessary to establish whether bacteria are actively transported inside of host cells and if so, what influences the recognition and selectivity of this process.
Hosting sulfide-oxidizing SUP05 in the outer epidermis may allow Ostiactis pearseae to avoid sulfide toxicity or the costly evolution of unique biochemistry to take up and transport sulfide [11]. Additionally, the epidermis in Anthozoa can function in nutrition [21], even more so than the gastrodermis, through direct uptake of dissolved organic compounds [44,45], thus the positioning of nutritional bacteria in the epidermis may increase effective exchange of small molecules. Like other SUP05 cells, those associated with the tentacles of O. pearseae were small (~500 nm in diameter [46]). Presumably, these symbionts require both oxygen and sulfide near simultaneously, for example "Candidatus Thioglobus autotrophicus," a member of the SUP05 group, has an aerobic phenotype and uses sulfide while respiring oxygen [46,47]. In this regard, it would be reasonable to house bacteria as close to the tissue surface as possible in order to accommodate gas exchange and meet symbiont metabolic demands.
The assumption that the SUP05 group may perform a nutritional role for the Pescadero Basin Ostiactis pearseae is evidenced by the comparatively light tissue δ 13 C values (average − 29.1‰). The contribution of chemosynthesis-derived carbon to O. pearseae biomass appears to exceed that reported for other deep-sea anthozoan species from the Gulf of Mexico (ex. Paramuricea [9]). The facultative, or rather mixotrophic, nature of the SUP05-Anthozoa symbioses proposed by [9] is also suggested for O. pearseae given the large range in negative δ 13 C values observed. Additionally, the SUP05 were found to not only have the gene repertoire necessary for chemosynthesis, but to be transcriptionally active, further supporting the possibility of SUP05 contributions to host nutrition [9]. Like all sea anemones (even those with photosynthetic symbionts), O. pearseae retains an arsenal of nematocysts by which to capture prey; thus, this tentacular team-up with SUP05 symbionts likely provides only a portion of their diet, indicating a mixotrophic strategy.
The SUP05 clade is involved not only in mediating dark carbon fixation, but also in the cycling of nitrogen, whether by denitrification, as has been shown in freeliving SUP05 populations [29,36] or assimilatory nitrate reduction, as in the case of symbiotic SUP05 [9,48]. A significant contribution to tissue nitrogen by microbial nitrate utilization may be possible for the SUP05 symbionts given the considerably low δ 15 N values in O. pearseae (average 1.6‰) and the successful amplification of the SUP05-related periplasmic nitrate reductase alpha subunit (napA) gene. The actual abundance of SUP05 symbionts per individual anemone is not known, nor is the regulation of carbon or nitrogen nutrient exchange, and thus, the overall nutritional influence of the SUP05 bacteria is not yet quantifiable.
Finally, Ostiactis pearseae tissue δ 34 S values (~− 1‰) represented a large offset from typical marine biomass (16-21‰ [49]), where biogenic sulfur is sourced from seawater sulfate with minimal isotopic fractionation (21‰ [50]). The average δ 34 S observed in O. pearseae tissues is consistent with typical hydrothermal vent fauna (− 5 to + 5‰ [51]), which are known to incorporate a local source of sulfur (e.g., volcanic, thermally altered sulfur at~0‰ [52,53]) via internal symbioses or direct consumption of sulfide-oxidizing bacteria. However, several individuals of O. pearseae revealed even lower δ 34 S values (down to − 11‰), which would likely require the additional incorporation of substantial sulfide produced via microbial sulfate reduction, which is expected to have a δ 34 S signature of − 20‰ or lighter [54,55]. The incorporation of sulfide sourced from dissimilatory sulfate reduction, rather than hydrothermal sulfide, has been similarly proposed for SUP05-hosting Bathymodiolus mussels from both Kakaijima Island and the Kaikata Caldera, which had tissue δ 34 S values of − 12‰ and − 25‰, respectively, with a comparable tissue sulfur content of~0.8% (n = only 1 specimen each [56,57]). The wide range of δ 34 S values for O. pearseae tissues (~− 11 to 9‰), compared to other thiosymbiont-hosting animal species, could be due to a combination of traditional feeding by the host, variable sulfide oxidation by the SUP05 symbionts (e.g., utilization of H 2 S, HS − , or other reduced S species, including endogenous elemental sulfur), or variation in the sulfur sourced from the petroleum-rich sediments of the Pescadero Basin.
Many uncultivated candidate bacterial phyla have been discovered in recent years within a variety of environments [20,58,59]. They usually have small genomes (< 1 Mb) with dramatically reduced biosynthetic capabilities, and yet exist globally in both marine and terrestrial habitats [60]. Several of these candidate phyla, known as the OD1 and BD1-5 groups (also referred to as Parcubacteria and Gracilibacteria, respectively), comprised up to 16-27% of the Ostiactis pearseae bacterial community and are known to play an important role in sulfur cycling [60]. An ectosymbiotic or parasitic lifestyle for the OD1 has been proposed [61], given their inability to synthesize vitamins, amino acids, nucleotides, and fatty acids. Additionally, while most candidate phyla are found in anoxic habitats, some OD1 genomes contained genes suggestive of O 2 use as a terminal electron acceptor [61,62]. Although not previously associated with the sulfide-oxidizing SUP05 group, or any specific proteobacterial group, the role of OD1 in sulfur reduction [60] and their diverse repertoire for attachment and adhesion [61] forecasts a possible direct association with either the SUP05 bacteria or O. pearseae mucous, for example.

Conclusion
Despite 40+ years of appreciation for chemosynthetic symbioses and the continued search for their occurrence in the most well-known habitats, Cnidaria have only very recently been shown to associate with chemoautorophic bacteria. Here, we identify a hydrothermal vent sea anemone, Ostiactis pearseae, at 3700 m depth in the Pescadero Basin, Gulf of California, that appears to be nutritionally supported by internal chemoautotrophic bacteria. This species, one of only 2 dominant sessile animals observed on the vent chimneys, has an unusual life position, often located in and amongst vent-obligate siboglinid tubeworms, very near to actively venting fluids. Ostiactis pearseae houses putative sulfideoxidizing SUP05 bacteria (Thioglobaceae) in its epidermis, with which it appears to have established a mixotrophic nutritional strategy, based on a broad range of carbon, nitrogen, and sulfur isotopes. Facultative, or rather mixotrophic, nutritional symbioses are often more difficult to recognize, compared to obligate alliances, but they are surely more common in nature [63], particularly in Cnidaria which experience symbiont gain and loss readily and often retain an ability to capture prey [9,64,65]. So, too, is the difficulty in uncovering nested symbioses, often involving microbe-microbe synergies. In this study, an unusual abundance of two candidate phyla, Parcubacteria and Gracilibacteria (a.k.a. OD1 and BD1-5, respectively) within O. pearseae tentacles, hints at the roles they may play in the cycling of nutrients within and on animal hosts. Cnidaria symbioses are considered foundational for coral reefs, and perhaps they also play an important role at hydrothermal vents. It would be worth investigating additional Anthozoa species observed to inhabit venting fluids at other sites worldwide [66][67][68], to see whether they have also forged nutritional relationships with chemosynthetic bacteria, such as the versatile SUP05 group.

Carbon, nitrogen, and sulfur isotope analysis
Tissue samples were dissected at sea, rinsed in milli-Q water, and frozen at − 20°C until thawed, washed with milli-Q water, and dried for 48 h at 60°C. Carbon and nitrogen isotope determinations of anemone tissues were made via isotope ratio mass spectrometry. Samples (0.2-0.8 mg dry weight) were loaded in tin boats and analyzed for total organic carbon (TOC) and total nitrogen (TN) abundances and δ 13 C org and δ 15 N using a Flash 2000 Elemental Analyzer (Thermo Fisher Scientific) interfaced to a Delta V Plus IRMS (Thermo Fisher Scientific) at Washington University, MO, USA. Samples were interspersed with several replicates of both in-house standards and international reference materials, including IAEA-CH-6, IAEA-CH-3, IAEA-NO3, USGS-40, and USGS-41. TOC and TN abundances were quantified by integrating peak areas against those produced by inhouse standards across a range of masses. The isotopic values are expressed in permil (‰) relative to international standards V-PDB (Vienna Pee Dee Belemnite) and Air for carbon and nitrogen, respectively. The longterm standard deviation is 0.2‰ for δ 13 C org and 0.3‰ for δ 15 N. There was no significant difference in δ 13 C between a subset of samples processed with and without acid (n = 7 each; ANOVA p = 0.974). Sulfur isotope analyses were performed by combusting~2-5 mg (dry weight) of tissue using a Costech ECS 4010 elemental analyzer coupled to a Thermo Fisher Scientific Delta V Plus mass spectrometer. Sulfur isotope values are expressed in standard delta notation (δ 34 S) in permil (‰) as a deviation from the Vienna Canyon Diablo Troilite (VCDT) standard. The long-term standard deviation is 0.3‰ for δ 34 S is 0.3‰ based on in-house and international standards, including NBS-127 and IAEA-S1.

DNA extraction
Specimens for molecular analysis (Table 1) were preserved immediately upon collection in~90% ethanol and stored at 4°C. Total genomic DNA was extracted from tissues using the Qiagen DNeasy kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. Two-liter water samples were filtered onto a 0.22-μm Sterivex-GP polyethersulfone filter (Millipore-Sigma, St. Louis, MO, USA) and frozen at − 80°C until DNA analysis. DNA extraction from Sterivex PES filters was also performed using the Qiagen DNeasy kit, according to the manufacturer's instructions, with the exception of the first step where 2 ml of ATL lysis buffer was added to the Sterivex filter, via Luer lock and syringe, and rotated at 56°C for 12 h. This solution was recovered from the filter, also via Luer lock and syringe, and processed as usual.

Molecular analysis of the microbial community
A 1000-bp region of the gene coding for napA (periplasmic nitrate reductase) was amplified directly from Ostiactis tissues using the primers V16F (5′-GCNCCNTGYMGNTTYTGYGG-3′) and V17R (5′-RTGYTGRTTRAANCCCATNGTCCA-3′ [28]), while a 408-bp fragment of the aprA gene (subunit of particulate methane monooxygenase enzyme) was generated using primers, aps1F (5-TGGCAGATCATGATYMAYGG-3) and aps4R (5-GCGCCAACYGGRCCRTA-3, described in [69]). A 1465-bp fragment of the 16S rRNA gene was amplified using the primers 27F and 1492R. Annealing conditions of 50°C, 50°C, and 54°C were used for napA, aprA, and 16SrRNA, respectively. Otherwise, all thermal protocols included the following steps: an initial 5 min denaturation at 94°C, then 1 min at 94°C, 1 min annealing step, and 1 min at 72°C, for 25 cycles, and a final 5 min extension at 72°C. Amplification products were sequenced directly using Sanger sequencing, via Laragen Inc., and submitted to GenBank (accession numbers MW148238 and MW158470). Close environmental and cultured relatives were chosen using top hits based on BLAST (www.ncbi.nlm.nih.gov).
The V4-V5 region of the 16S rRNA gene was amplified using bacterial primers with Illumina (San Diego, CA, USA) adapters on the 5′ end 515F (5′-TCGTCGGCAGCGTCAG ATGTGTATAAGAGACAGGTGCCAGCMGCCGCG GTAA-3′) and 806R (5′-GTCTCGTGGGCTCGGA GATGTGTATAAGAGACAGGGACTACHVGGGT WTCTAAT-3′; [70]). The PCR reaction mix was set up in duplicate for each sample with Q5 Hot Start High-Fidelity MiSeq 16S rRNA sequence data was processed in Quantitative Insights Into Microbial Ecology (v1.8.0). Raw sequence pairs were joined and quality-trimmed using the default parameters in QIIME. Sequences were clustered into de novo operational taxonomic units (OTUs) with 99% similarity using UCLUST open reference clustering protocol, and then, the most abundant sequence was chosen as representative for each de novo OTU. Taxonomic identification for each representative sequence was assigned using the Silva-119 database, clustered at 99% similarity. A threshold filter was used to remove any OTU that occurred below 0.01% in the combined samples dataset. Analyses are based on Bray-Curtis distances of fourthroot transformed data, which minimizes errors in the ordination due to PCR bias, while not sacrificing genuine differences between samples. Quantification and statistical analyses are described in the "Results" sections and figure legends. Comparisons were performed using ANOVA and statistical significance was declared at p < 0.05. Statistical analyses of beta diversity (e.g., ANOSIM) were performed with Primer E. The raw Illumina 16S rRNA barcode sequences and metadata collected in this study are available from the NCBI Small Read Archive (BioProject # PRJNA671766). The processed sequence data, as well as representative sequences, are available on the Dryad Digital Repository URL (https://doi.org/10.5061/dryad. mkkwh70wt, along with representative SUP05 OTUs in Additional File 6: DataFile S1).
Newly generated DNA sequences for Ostiactis pearseae (and those for morphotype identified as Kadosactinidae sp.B. in this contribution) were combined and analyzed with the dataset by [74] for each of the four markers (Additional File 1: Table S1). Sequences for each marker were separately aligned in MAFFT v.7 [75,76], using the following settings: strategy, L-INS-I; scoring matrix for nucleotide sequences, 200PAM/k = 2; gap open penalty, 1.53; offset value, 0.05. Alignments for each marker were analyzed separately and as a concatenated dataset (alignments available on the Dryad Digital Repository URL (https://doi.org/10.5061/dryad. mkkwh70wt)). For each gene region, the best model of nucleotide substitution was chosen using the Akaike information criterion (AIC) on jModeltest2 [77,78] implemented on the CIPRES Portal [79]. Maximum likelihood (ML) analyses were performed using RAxML-NG v0.6.0 [80], using the appropriate model of nucleotide substitution for each gene partition (12S: GTR+I+G; 16S: TVM+G; COIII: TPM3uf+I+G; 18S: TIM2+I+G; 28S: GTR+I+G) in the combined alignment. The Majority Rule Criterion was used to assess clade support allowing bootstrapping to halt automatically (-autoMRE). All analyses were run with gaps treated as missing data.
Morphology and cnidae analysis of the anemone host Ostiactis pearseae Specimens were examined whole and dissected. Histological sections 5-10 μm thick were made from different body regions of two specimens using standard paraffin techniques and stained with Heidenhain Azan stain [81]. The distribution and size ranges of cnidae in the tissues were analyzed from six specimens using light DIC microscopy (× 1000 magnification, oil immersion). Twenty non-fired capsules of each cnida type (when possible) were photographed and measured at random. Cnidae size distribution offers information on the variability in capsule size for each type of nematocyst. We follow a nematocyst terminology that combines the classification of [82], modified by [83], thus differentiating "basitrichs" from "b-mastigophores" with that of [84,85] which captures the underlying variation seen in "rhabdoids" (see [74] for more details). We include photographs of each type of nematocyst for reliable comparison across terminologies and taxa (see [86]). Higher-level classification for Actiniaria follows [16].
Excess probe was removed by sequentially washing the slides for 15 min at 37°C in probe wash buffer (Molecular Technologies; 30% formamide, 5× SSC, 9 mM citric acid (pH 6.0), 0.1% Tween 20, 50 μg/mL heparin) to which 5× SSCT (750 mM NaCl, 75 mM sodium citrate, 0.1% Tween 20, pH 7) had been added to final concentrations (vol/vol) of 25%, 50%, and then 75%. This wash sequence was followed by two 15-min washes in 100% 5× SSCT at 37°C. Before amplification, 6 pmol of each hairpin, per reaction, was "snap-cooled" by heating to 95°C for 90 s, followed by 25°C for 30 min, in a thermocycler in separate PCR tubes. During this time, sections were equilibrated with amplification buffer (Molecular Technologies; 5× SSC, 0.1% Tween 20, 10% dextran sulfate) at room temperature for 30 min. For amplification, each "snapcooled" hairpin in a pair was added to 100 μl amplification buffer (for a final hairpin concentration of 60 nM for each amplifier hairpin), and then sections were incubated overnight (~18 h) at room temperature on a rocking platform protected from light. To remove unbound hairpin sequences, sections were washed twice in 5× SSCT for 15 min at room temperature, followed by two 30-min washes in 5× SSCT. Sections were rinsed with distilled water and counterstained with 4′6′-diamidino-2-phenylindole (DAPI, 5 mg/mL) for 1 min, rinsed again, and mounted in Citifluor. Tissues were examined by epifluorescence microscopy using either a Nikon E80i epifluorescence microscope with a Nikon DS-Qi1Mc high-sensitivity monochrome digital camera or a Zeiss Elyra microscope with an ANDOR-iXon EMCCD camera.

Transmission electron microscopy
Specimens for TEM and semi-thin sectioning were fixed in PFA and preserved in 50% EtOH. Before embedding, specimens were rehydrated, post-fixed with 1% OsO 4 , and subsequently dehydrated again in an ascending acetone series and embedded in Araldite. One-micrometer semi-thin sections were sectioned using a "Diatome Histo Jumbo" diamond knife on a Leica Ultracut S ultramicrotome and stained with toluidine blue (1% toluidine,1% sodium-tetraborate, and 20% saccharose). Coverslips were mounted with Araldite and sections were imaged with an Olympus microscope (BX-51) equipped with the dot.slide system (2.2 Olympus, Hamburg). Silver interferencecolored sections (70-75 nm) were prepared using a "Diatome Ultra 45°" diamond knife. The sections were placed on Formvar-covered, single-slot copper grids and stained with 2% uranyl acetate and lead citrate in an automated TEM stainer (QG-3100, Boeckeler Instruments). Ultra-thin sections were examined using a Zeiss EM10 transmission electron microscope with digital imaging plates (DITABIS Digital Biomedical Imaging Systems, Germany).