Marine Sponges as Chloroflexi Hot Spots: Genomic Insights and High-Resolution Visualization of an Abundant and Diverse Symbiotic Clade

Chloroflexi abundance in HMA sponges.Recently, members of the phylum Chloroflexi were shown to be present in much higher abundance and diversity in HMA sponges than in LMA sponges, which is why they were termed indicator species for HMA sponges (12). Here, we provide further details for the presence and abundance of Chloroflexi in HMA sponges (12) (Fig. 1; see also Tables S2A and S2B in the supplemental material). The recently compiled Sponge Microbiome Project (6, 31) was used as a reference database. In these 63 investigated sponge species, Chloroflexi abundances ranged from 4.39% ± 3.02% (Chondrilla caribensis) to 31.89% ± 5.27% (Aplysina spp.) (Fig. 1, right panel, and Table S2A). With respect to the Chloroflexi classes, the SAR202 clade was the most abundant, contributing on average to 47.74% ± 22.00% of the phylum total abundance (Fig. 1, top panel, and Table S2B). Members of the classes Caldilineae (22.35% ± 17.93%) and Anaerolineae (11.64% ± 12.30%) were also abundant in some sponges but not others. Unclassified OTUs at the class level represented 14.50% ± 10.77% of Chloroflexi sequences, indicating that there is phylogenetic novelty still to be discovered. Despite some variability (Fig. 1, heatmap), the classes/clades SAR202, Caldilineae, and Anaerolineae as well as diverse hitherto unclassified OTUs dominated the Chloroflexi population in the HMA sponges. The remaining classes/clades amounted to 3.78% of total phylum abundance. At the OTU level, there was a significant effect of geographical location [F (15, 725) = 15.9, P = 0.001] and of sponge taxonomy [F(9, 725) = 18.6, P = 0.001] on beta diversity (Table S2C). For example, a Mediterranean cluster and several Caribbean clusters become visible as shown by Bray-Curtis cluster analysis. In addition, a host taxonomic signature was revealed for example for the sponge genera Aplysina and Agelas. However, there were also exceptions (i.e., Neopetrosia species from the Caribbean did not cluster together, and Stelletta maori from New Zealand fell into the Mediterranean cluster). Altogether, this analysis showed the high abundance and consistency of the three above-mentioned main clades of Chloroflexi in HMA sponges and revealed a geographic and host taxonomic signature for the Chloroflexi community.

• Open in new tab
• Download powerpoint

FIG 1

Heatmap showing the relative abundance of Chloroflexi classes/clades in 63 HMA sponges extracted from Earth Microbiome Project (EMP) data (31). The top panel shows the mean relative abundance of Chloroflexi classes/clades in all sponges (means ± standard deviations [error bars]). The right panel displays the mean relative abundance of the phylum Chloroflexi in predicted (light green names and medium green bars) and classified (dark green names and bars) HMA sponges (means ± standard deviations) determined by machine learning (12). Results of cluster analysis based on Bray-Curtis dissimilarities on mean relative abundances of OTUs within the phylum are presented on the left side. Sponges are marked with stars when all species samples came from one of the three major sample locations.

Phylogeny of Chloroflexi metagenome bins and single amplified genomes.In total, 260 single amplified genomes (SAGs) were screened for the presence of the phylogenetic marker. A total of 125 16S rRNA genes were identified in 112 SAGs, of which some were duplicates per well. Sequencing revealed that 39 genes (31.2%) were affiliated with the phylum Chloroflexi (33). We randomly chose 13 SAGs for genome sequencing. Phylogenetic analysis of 16S rRNA genes revealed that one SAG (3D), four SAGs (1B, 1G, 1H, and 4H), and eight SAGs (2D, 3B, 3H, 4A, 5H, 6B, 6C, and 6F) belonged to the classes/clades SAR202, Anaerolineae, and Caldilineae, respectively, forming a well-supported sequence cluster (bootstrap, 100) with other sponge-derived sequences (Fig. S2). The binning of metagenomic sequence data (21) resulted in additional five high-quality bins (Table 1). The only metagenome bin containing a 16S rRNA gene (S156) belonged to SAR202. The SAR202 sequences formed a well-supported cluster (bootstrap value, 98) with other sponge-derived 16S rRNA gene sequences (Fig. S1). To elucidate the clade affiliation of the remaining four metagenome bins lacking 16S rRNA genes of appropriate length, a concatenated genome tree based on nine ribosomal genes was calculated (Fig. 2). One bin (A154) was affiliated with the class Anaerolineae, two metagenome bins were associated with the Caldilineae (C141 and C174), and two bins were associated with the SAR202 clade (S152 and S156) within the phylum Chloroflexi. The phylogenetic affiliation of metagenome bin S156 was congruent with the 16S rRNA gene analysis. SAGs were included in the protein-based phylogenetic analysis when they encoded at least three of the nine ribosomal genes. Due to the lack of more-complete reference genomes from SAR202 microorganisms, the most complete one (ca. 25%, SAR202 cluster bacterium sp. strain SCGC AAA240-N13, IMG Gold Study ID Gs0017605 [22]) was included in this analysis, although only one ribosomal protein could be used for tree construction. Both analyses showed a stable phylogeny of all SAGs and metagenome bins to above-described classes or clades within the phylum Chloroflexi. All three classes/clades were visualized in the A. aerophoba sponge mesohyl matrix by fluorescence in situ cohybridization (FISH) on ultrathin tissue sections using class/clade-specific probes. The Chloroflexi cell signal was abundant as judged by the stained versus unstained bacterial signal, and SAR202 (green) seemed to dominate over either Anaerolineae (red) or Caldilineae (orange). The cells were metabolically active as judged by the brightness of the FISH probe (Fig. 3). These visual observations are consistent with the relative abundance of the phylum Chloroflexi within sponge microbiomes (24% ± 6% within A. aerophoba sponge microbiome; Fig. 1 and Table S2A) and the representation of Chloroflexi cells among SAGs (31%).

• Open in new tab
• Download powerpoint

FIG 2

Concatenated protein tree. Maximum likelihood phylogenetic analysis of Chloroflexi metagenome bins and SAGs (in red) from 1,914 positions of 60 sequences using ribosomal proteins. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (100 replicates) are shown. The initial tree for the heuristic search was obtained automatically by applying neighbor-joining and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model. Multiple sequences are included in the collapsed branches representing Chloroflexi classes/clades (bold). To root the tree, three representative genomes from the phylum Actinobacteria were used. Reference genomes with accession numbers can be found in Table S1 in the supplemental material.

• Open in new tab
• Download powerpoint

FIG 3

Distribution of Chloroflexi clades in Aplysina aerophoba mesohyl using fluorescence in situ hybridization (FISH). The image shows the overlay of all probes. SAR202 cells are displayed in green, Caldilineae cells in orange, and Anaerolineae cells in red. The nucleotide stain DAPI (white/gray) served as a reference for the localization of unstained cells. Bar, 10 µm.

General description of genomes.The final genome assembly sizes for the sponge-associated Chloroflexi single cells (SAGs) ranged from 0.16 to 4.25 Mbp, representing up to 66.85% of genome completeness derived from IMG-based estimations (Table 1). The guanine-cytosine (GC) content ranged from 58.08 to 59.32%, 58.36 to 62.98%, and 56.93 to 65.59% for Anaerolineae, Caldilineae, and SAR202, respectively. The numbers of identified genes were highly variable, ranging from 3,358 genes for Anaerolineae genome bin A154 to 5,448 for the SAR202 bin S152 and 5,662 genes for the Caldilineae bin C174 (Table 1). The five metagenome bins (two for Caldilineae [C141 and C174], two for SAR202 [S152 and S156], and one for Anaerolineae [A154]) which had >90% coverage were chosen for detailed metabolic analysis and inner-phylum comparison. The letters of the bins were chosen to reflect their phylogenetic identity (A for Anaerolineae, C for Caldilineae, and S for SAR202). Additionally, the most complete Anaerolineae SAG 1B (55.76% genome completeness estimation) was included in the analysis. Due to the lack of complete genomes of marine representatives of the Anaerolineae, Caldilineae, and SAR202 available at the time of analysis, sponge-derived Chloroflexi genomes could not be fully assessed. Taking the pitfalls inherent to metagenome sequencing into account (i.e., fragmented assemblies, unresolved ambiguities), we have opted for a presence/absence approach: a gene or enzyme was considered present when it was identified in both bins of the corresponding clade. For Anaerolineae, we consider an enzyme or gene present when identified in bin A154, and the SAG 1B was taken as additional support.

Central metabolism of sponge-associated Chloroflexi.Metabolic reconstruction suggests that Chloroflexi are aerobic and heterotrophic bacteria including glycolysis, tricarboxylic acid (TCA) cycle, pentose phosphate pathway (PPP) (see Text S1 and Fig. S3A to C for details), and the respiratory chain as energy-producing pathways in all three clades.

With respect to autotrophic carbon fixation, the reductive citrate acid cycle (Arnon-Buchanan cycle), which is largely present, and the Wood-Ljungdahl pathway, which was partially identified (Fig. S2D) might be functional in times when sponges are not pumping and the mesohyl turns anoxic (34). Genes encoding components involved in ammonia import and assimilation are carried on all investigated genomes, but SAR202 and Caldilineae have additional genes for glutamate synthesis from glutamine and directly from ammonia. The transport of nitrite (and possibly also nitrate) is encoded on all investigated genomes, while the reduction to ammonia is encoded only by SAR202 (Fig. S3E). The incorporation of sulfur (with, e.g., thiosulfate as donor) into S-containing amino acids might be possible in all clades, whereas the assimilatory reduction of sulfate is restricted to Anaerolineae and Caldilineae genomes (Fig. S3F).

These processes additionally provide precursors for further metabolic pathways such as biosynthesis of purines and pyrimidines, amino acids, and cofactors, or structural compounds. Machinery for transcription, translation, and purine and pyrimidine metabolism are largely present. Fatty acid (FA) biosynthesis and degradation pathways were detected in all six genomes. All genomes encode a high number of different ABC transporters compared to genomes of free-living bacteria (22, 35) to supplement for nutrition and cell growth-related compounds [including oligopeptides, phosphate, l- and branched-chain amino acids, minerals such as iron(III) and molybdate, metal ions such as zinc, manganese, and iron(II)]. Additionally, all six genomes largely encode enzymes needed for biosynthesis of most amino acids (Text S1). We could not identify any of the typical phosphotransferase systems, as was the case for Ca. Poribacteria described previously (19).

We found genomic potential for aromatic degradation in Chloroflexi genomes, but pathways remain incomplete (Text S1). Several genes encoding components involved in phenylpropionate and cinnamate degradation, terephthalate degradation, catechol degradation, and xylene degradation were identified in Chloroflexi genomes. Also, genes encoding enzymes involved in ring cleavage by Baeyer-Villinger oxidation and beta oxidation as well as ring-hydroxylating dioxygenases and isomerases were identified which could be involved in degradation of aromatic compounds possibly synthesized by the sponge host or other microbes. This finding is interesting in the context that many sponge species contain secondary metabolites that often contain aromatic ring structures that serve as a defense strategy against predators and biofouling (36). Sponge symbionts may be able to degrade such substances, enabling them to a life within sponge hosts. These findings fit with the potential degradation of organic compounds which was suggested for Chloroflexi bacteria of the class SAR202 by (meta)genomic studies (22, 23, 35). The highest (20 to 30% relative to the total microbiome) and most consistent presence of Chloroflexi within a sponge genus was found in the sponge genera Plakortis, Agelas (with the exception of Agelas dispar), and Aplysina and sister taxon Aiolochroia. Interestingly, all of them contain characteristic natural products with aromatic ring structures. It is therefore tempting to speculate that the presence and abundance of Chloroflexi and especially SAR202 are shaped, at least to some extent, by the natural products present in the corresponding host sponges.

With respect to cell wall structure, the Anaerolineae and Caldilineae genomes carry the gene repertoire for peptidoglycan biosynthesis. The noticeable lack of peptidoglycan biosynthesis genes in the SAR202 genomes (Text S1) is consistent with previous analyses of three Chloroflexi genomes derived from uranium-contaminated aquifers (28) and hyperoxic zones from the Gulf of Mexico (37). Additionally, consistent with previous observations (38), none of the six genomes contained flagellar or chemotaxis genes.

Metabolic specialization: extensive carbohydrate uptake and degradation in Anaerolineae and Caldilineae.The following features are metabolic specialties of Anaerolineae and Caldilineae and appear to be missing in SAR202, unless otherwise mentioned (Fig. 4). We found a number of ABC transporters for the import of diverse monosaccharides (ribose/xylose, inositol, glycerol-3-phosphate [glycerol-3P], and rhamnose) and oligosaccharides (sorbitol, raffinose/stachyose/melibiose, maltose, N-acetylglucosamine, and arabinosaccharide) into Anaerolineae and Caldilineae cells. Xylose can be processed to xylulose which may enter the pentose phosphate cycle finally leading into glycolysis. Ribose can be converted to 5-phosphoribosyl 1-pyrophosphate (PRPP), which is a precursor for the biosynthesis of the amino acid histidine or it may fuel purine and pyrimidine synthesis (Text S1). Both groups may also be able to import glycerol-3P, which is a phosphoric ester of glycerol (a component of glycerophospholipids) which can be converted to fatty acids. Additionally, we found evidence for arabinose and rhamnose import and degradation; however, annotation was incomplete (Text S1).

• Open in new tab
• Download powerpoint

FIG 4

Summarized metabolic features which were found only in Anaerolineae and Caldilineae (left side, blue and green arrows, respectively) or in SAR202 genomes (right side, red arrows). The central metabolic pathways (glycolysis, TCA cycle, purine, pyrimidine histidine biosynthesis) located in the middle of the figure are general features found in all genomes. Lines are dashed when pathways or transporter could not be annotated completely (single enzymes of the pathway or single genes from the transporter were missing) or could not be annotated in both genomes of one clade. Gray dashed arrows indicate that those transporters were not identified.

Arabinooligosaccharides (such as α-l-arabinofuranosides, α-l-arabinans, arabinoxylans, and arabinogalactans) result from degradation of plant-like cell material entering the sponge by filtration. These substances may be imported by the almost completely annotated AraNPQ and MsmX transporters and be utilized to l-arabian and l-arabinose by the enzyme α-N-arabinofuranosidase (EC 3.2.1.55, GH3). The enzyme l-arabinose isomerase (EC 5.3.1.4, AraA) is present in all four genomes of Anaerolineae and Caldilineae and converts l-arabinose to l-ribulose, which can further be converted by reactions of PPP to glucose-6P suitable for entering glycolysis. Additionally, other oligosaccharides such as stachyose, raffinose, melibiose, and galactose can be imported and used in central metabolism (Fig. 4 and Text S1).

The utilization of myo-inositol as a carbon source and possibly as a regulatory agent was hypothesized previously for sponge-associated Ca. Poribacteria (19). Similarly, sponge-associated Anaerolineae and Caldilineae genes encode almost all the components involved in the inositol degradation pathway (Text S1). myo-Inositol is likely degraded to glyceraldehyde-3-phosphate and acetyl-CoA, which are further used in the central metabolism. Inositol phosphates are found as part of eukaryotic and archaeal cell wall components (39). Phosphorylated inositol is a precursor for several lipid molecules, including sphingolipids, ceramides, and glycosylphosphatidylinositol anchors (40), as well as many stress-protective solutes of eukaryotes (39), and it might be part of the signal transduction in sponges (41). Therefore, the sponge itself or eukaryotic microorganisms can probably provide inositol as a carbon source or regulatory agent for the microbial symbionts.

Uronic acids are sugar acids that can be found in biopolymers of plants, animals, and bacteria (42, 43) and are known to occur in glycosaminoglycans (GAGs). GAGs in sponges are mainly composed of fucose, glucuronic acid (glucoronate), mannose, galactose, N-acetylglucosamine, and sulfate (44–46). Genes encoding enzymes involved in degradation of uronic acids were found in Anaerolineae and Caldilineae genomes. The possibility of galacturonate and glucuronate catabolism is supported by the conversion of 2-dehydro-3-deoxy-d-gluconate by the enzymes glucoronate isomerase (EC 5.3.1.12), tagaturonate reductase (EC 1.1.1.58), and altronate hydrolase (EC 4.2.1.7). Furthermore, the presence of genes encoding oligogalacturonide lyase (EC 4.2.2.6), 2-deoxy-d-gluconate 3-dehydrogenase (EC 1.1.1.125), and 2-dehydro-3-desoxy-d-glucokinase (EC 2.7.1.45) supports possible 4(4-α-d-gluc-4-enuronosyl)-d-galacturonate degradation activity. The products could then enter the Entner-Doudoroff (ED) pathway via 2-dehydro-3-desoxyphophogluconate aldolase (EC 4.1.2.14). Uronic acid degradation could principally be connected to the inositol degradation pathway via d-galacturonate even though additional genome evidence, such as genes encoding the enzyme inositol oxidase (EC 1.13.99.1), remain wanting (Fig. 4 and Text S1). A number of transporters for N-acetylglucosamine, digalacturonate, mannose, and galactose (Fig. 4 and Text S1) were identified in Anaerolineae and Caldilineae genomes. Digalacturonate can be utilized by the uronic acid degradation pathway (Text S1), and N-acetylglucosamine can be used directly in amino sugar and nucleotide sugar synthesis. The presence of uronic acid degradation pathways provides strong support that Anaerolineae and Caldilineae, similar to the previously described Ca. Poribacteria, degrade glycosaminoglycan chains of proteoglycans, which are important components of the sponge host matrix (19). In that line, Anaerolineae and Caldilineae genomes were enriched in arylsulfatases A (Fig. S4A), which are thought to be involved in metabolization of sulfated polysaccharides from the sponge extracellular matrix (19, 21) and in the heterotrophic ability of symbionts to use sponge components for nutritional purposes.

Expanded carbohydrate-active enzyme (CAZymes) repertoire in Caldilineae and Anaerolineae.In order to search for CAZymes, we screened the Chloroflexi genome data against dbCAN (47) and classified the enzymes according to the CAZy database (48). Most Chloroflexi hits were against glycosyl hydrolases (GH), glycosyltransferases (GT), and carbohydrate-binding modules (CBM). Consistent with the above-described metabolic specializations, these enzyme classes were present in larger amounts in Caldilineae and Anaerolineae than in SAR202 (Fig. S4B and Table S3). Altogether, 40 GH families were identified in all Chloroflexi genomes (Table S3). Glycosyl hydrolase family 109 was the most abundant family of GHs and was identified in all six genomes. GH109 family proteins are predicted as α-N-acetylgalactosaminidases (EC 3.2.1.49) with putative substrates such as glycolipids, glycopeptides, and glycoproteins, all of which are common constituents of sponge mesohyl as well as dissolved organic matter from seawater. The GH74 family is the second most abundant and is also present in all six genomes. These appear to be xyloglucan-hydrolyzing enzymes, that act on β-1,4 linkages and might help degrade various oligo- and polysaccharides. The previously reported glycosylhydrolases GH33 and GH32 (19, 20) were the third most abundant, but they were restricted to Caldilineae bin C174. This enzyme family is annotated as sialidase (EC 3.2.1.18), capable of hydrolyzing glycosidic linkages of terminal sialic acid residues, which are present in sponge mesohyl (49). Altogether, 17 glycosyltransferases were identified on Chloroflexi genomes with families GT2, GT4, and GT83 being the most abundant. Among the 11 CBM families identified on Chloroflexi genomes, CBM50 was the most abundant, but it was restricted to Caldilineae and Anaerolineae. CMB50 modules, also known as LysM domains, attach to various GH enzymes which are involved in the cleavage of chitin or peptidoglycan. The numbers of carbohydrate-active enzymes on Chloroflexi symbiont genomes reflect their extensive potential to degrade complex carbohydrates as reported previously for Ca. Poribacteria and the sponge-associated unidentified lineage SAUL (19, 20).

Metabolic specialization: cofactor biosynthesis in SAR202 genomes.There is mounting evidence that vitamins and cofactors produced by diverse symbiont lineages could be beneficial to the sponge host (50–53). Parallel transcriptional activity profiling of the symbionts and the sponge showed that the symbionts had the capacity for vitamin B biosynthesis, whereas the host transcripts displayed the capacity for vitamin catabolism (54). It is thus tempting to speculate that the sponges’ nutrition is augmented by symbiont-derived vitamins and cofactors. In the present study, at least two biosynthetic pathways for cofactor biosynthesis were identified on SAR202 genomes, which were absent in Anaerolineae and Caldilineae (Fig. 5). Thiamine is an essential cofactor which is involved in central metabolism. The biosynthesis of the biologically active form thiamine diphosphate (TPP) from l-cysteine, glycine, pyruvate, and glyceraldehyde-3P is encoded on the SAR202 genomes (Fig. 4 and 5A). Although the pathway is incomplete, the data strongly suggest that the synthesis of TPP is restricted to SAR202 bacteria. Instead, Anaerolineae and Caldilineae appear to import thiamine via an ABC transporter (TbpA, ThiPQ) and convert it to TPP by using thiamine pyrophosphokinase (EC 2.7.6.2). Additionally, thiamine synthesis might be a symbiosis-related feature, since free-living SAR202 members lack the synthesis ability (22, 37).

• Open in new tab
• Download powerpoint

FIG 5

Pathway for synthesis of thiamine (A) and riboflavin (B) in both SAR202 genomes. (A) Members of classes Anaerolineae and Caldilineae carry genes encoding components involved in the import of thiamine (see also Fig. 4). (B) The conversion of riboflavin into the biologically active forms (flavin mononucleotide [FMN] and flavin adenine dinucleotide [FAD]) was encoded by genes in the genomes of all three classes (filled pies compared to empty pies). The colors represent the genomes. The numbers are KEGG identifiers. PPP, pentose phosphate pathway. Gray arrows and numbers indicate unidentified enzymes.

Second, riboflavin (vitamin B2) is required by enzymes and proteins to perform certain physiological functions. Specifically, the active forms, flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD), serve as cofactors for a variety of flavoprotein enzyme reactions. Most genomes carry genes encoding the enzymes FMN adenylyltransferase (EC 2.7.7.2) and FAD riboflavin kinase (EC 2.7.1.26) which activate riboflavin into FMN. However, only the SAR202 genomes contain genes encoding the riboflavin biosynthesis enzymes which rely on GTP and ribulose-5P (Fig. 4 and 5B). Both substrates can be provided by pathways of the central metabolism (purine metabolism and PPP). Only one free-living SAR202 bacterium (unclassified Chloroflexi bin 43) (37) has genes encoding components involved in riboflavin synthesis.

Potential for degradation of recalcitrant DOM in SAR202.The uptake of the amino acid l-Asp was shown before in subtropical Atlantic waters (35), and the possible participation of deep sea SAR202 bacteria in degradation of recalcitrant or refractory DOM was recently postulated by Landry et al. (22). Even though the exact composition of DOM in the world’s oceans remains to be elucidated, refractory DOM is an important component of the global carbon budget in terms of sheer mass. Landry et al. (22) argue that SAR202 genomes have an expanded repertoire of oxidative enzymes that may help in the oxidation of recalcitrant compounds. Interestingly, some of the described enzymes were also found to be enriched in SAR202 symbionts of sponges (Table S4). Among them are genes encoding proteins from the CaiB/BaiF family as well as related family III transferases. While the enrichment of CaiB was previously interpreted as carnitine being a carbon and nitrogen source for sponge symbionts (21), an alternative explanation may be that it serves to funnel substrates into degradation pathways without consumption of energy by shuffling CoA, thus generating free electrons (Text S1). Even though the precise function of CaiB/BaiF family proteins cannot be elucidated at this time, the enrichment in SAR202 genomes is noteworthy. Further, a total of 53 flavin-dependent, class C oxidoreductases of the luciferase family (flavin mononucleotide monooxygenases [FMNOs], COG2141) which include alkanesulfonate monooxygenase SsuD and methylene tetrahydromethanopterin reductase, were present and enriched in SAR202 genomes compared to the genomes of Anaerolineae and Caldilineae bacteria (Table S4). These enzymes are proposed to participate in the oxidation of (long-chain) aldehydes to carboxylic acids and/or cleavage of carbon-sulfur bonds in a variety of sulfonated alkanes (55). The SAR202 genomes contained 22 genes encoding short-chain alcohol dehydrogenases (COG0300) which might be involved in canalization of ketone body derivate release. Some of these genes from bin S152 showed homologies to cyclopentanol and 3-α (or 20-β)-hydroxysteroid dehydrogenases which convert alicyclic-bound alcohol groups to ketones (22, 56). The combination of the enzymes described above could allow sponge-associated Chloroflexi to convert recalcitrant alicyclic ring structures to more labile carboxylic acid, as proposed recently for SAR202 bacteria from deep sea (22).

Additionally, a number of genes encoding oxidative enzymes were identified on the Chloroflexi genomes but were not enriched in SAR202 (Table S4). These enzymes include a 2-oxoglutarate:ferrodoxin oxidoreductase (EC 1.2.7.11) which oxidizes acetyl-CoA, carbon monoxide dehydrogenase (EC 1.2.99.2) which might allow the bacteria to oxidize CO as described for some members of the Ktedonobacteria (57), CO- or xanthine dehydrogenases (COG1529) which are possibly involved in oxidation of a broad range of complex substrates (22), choline dehydrogenase (EC 1.1.99.1) being possibly involved in the oxidation of alcohols to aldehydes, sarcosine oxidase (EC 1.5.3.1), the serine hydroxymethyltransferase (2.1.2.1) with predicted function in choline degradation, formaldehyde dehydrogenase (EC 1.2.1.46), and subunits of formate dehydrogenase (EC 1.2.1.2/43) which oxidize formaldehyde and formate and might be involved in demethylation of various compounds. The overall presence and frequent enrichment of enzymes with oxidative capacity in SAR202 would be consistent with gene functions in degradation of recalcitrant DOM. However, owing to the sponges’ existence in shallow-water sun-lit benthic environments, it remains unclear whether the sponge symbionts encounter recalcitrant DOM derived from seawater sources. Interestingly, Colatriano et al. recently proposed the degradation of terrestrial DOM (tDOM) by members of the SAR202 clade (23). Shallow-water sponges and associated symbionts could be faced by tDOM via freshwater inflow in ocean waters they inhabit. Alternatively, and similar to other high-diversity microbiota, for example, of ant, ruminant, and human guts, the resident microbes were likely to specialize in certain substrates, thus promoting maximum nutrient exploitation and also securing their individual niche in the holobiont ecosystem.

Symbiosis-related features.Eukaryotic-like proteins (ELPs) seem to be a general genomic feature of sponge symbionts (20, 38, 50, 51, 58–60). Ankyrin (ANK), tetratricopeptide (TPR), and leucine-rich (LRR) repeat proteins are postulated to be involved in mediating host-microbe interactions (61, 62). Ankyrin and ankyrin repeat-containing proteins were detected in all six genomes of sponge symbionts (Fig. S4C and Table S5A) in higher numbers than in free-living bacteria (22, 37). It was recently proposed that the expression of sponge symbiont-derived ankyrin protein prevents phagocytosis by amoeba (63), and it is tempting to speculate that they protect the symbionts from digestion by the sponge archaeocytes in vivo. TPRs, possibly functioning as a module for protein-protein interaction involved in a variety of cellular functions, including those that participate in bacterial pathogenesis (64) were found in all six genomes. However, LRR genes were identified only in Caldilineae and SAR202 genomes (Fig. S4C and Table S5A). Many LRR proteins are involved in protein-ligand interactions; these interactions include plant immune response and the mammalian innate immune response (for a review, see reference 65), such as the detection of pathogen-associated molecular patterns by recognition receptors (66). Our findings are in good agreement with general patterns previously found in the metagenomes of sponge symbionts (38, 51), in enriched (mini)metagenomes of cyanobacterial sponge symbionts (59) and single amplified genomes from members of SAUL (20) and Ca. Poribacteria (58).

Another example of sponge-symbiont enriched features are the clustered, regularly interspaced, short, palindromic repeats (CRISPRs) and their associated proteins (Cas) that have recently been reported from the genomes of sponge symbionts (20, 38, 50, 51, 59). Here, the investigated Caldilineae and SAR202 genomes contained CRISPR-Cas systems, while the two most complete Anaerolineae genomes did not (Table 2). The presence of CRISPRs can be explained by the extensive filter-feeding activity of sponge hosts that result in high exposure of sponge symbionts to phages and other sources of free DNA from ambient seawater.

The synthesis of secondary metabolites is an important defense mechanism of sessile organisms such as sponges to protect themselves against predators or biofouling (36). Many of these compounds are in fact produced by the sponge microbiome (10, 36). In particular, genes for polyketide synthases (PKS), nonribosomal peptide synthetases (NRPS), and halogenases are regularly enriched in sponge symbionts, often with new structures and putatively novel activities (51, 67–72). Here, we assessed the genomic repertoire of sponge-associated Chloroflexi for secondary metabolism using antiSMASH (73). In both SAR202 genomes and in Caldilineae C141, we found up to three polyketide synthase (PKS) gene clusters, all of which showed homologies to the previously reported type I PKS gene cluster from other sponge symbionts. Additional gene clusters for the production of terpenes and other yet to be identified substances were identified in the two SAR202 genomes and in Caldilineae C174 (Table 2 and Table S5B). Both Anaerolineae genomes did not contain any gene clusters for the biosynthesis of secondary metabolites. While the exact functions of these gene clusters putatively involved in defense remain unknown, it appears that at least SAR202 bacteria and Caldilineae have the genomic repertoire for chemical defense within the sponge holobiont.

Ultrastructural identification of sponge-specific Chloroflexi.The correlation of probe-specific fluorescence with scanning electron microscopy (SEM) images allowed taxon-specific identification of Chloroflexi cells at ultrastructural resolution. Overall, distributions of all three Chloroflexi clades in Aplysina aerophoba indicate that SAR202 cells were more abundant than the other two Chloroflexi classes (Fig. 6). This is consistent with the relative abundances of Chloroflexi classes in HMA sponges extracted from EMP data (Fig. 1). Cells belonging to the SAR202 clade (green signals in Fig. 6A and B) are generally rod shaped (0.8 µm by 1 to 2 µm) with a regular distribution of cell cytosol content. The Anaerolineae-specific probe targeted rod-shaped cells (0.8 by 2.0 µm) (red signal in Fig. 6A). The characteristic feature of Caldilineae-positive cells (ca. 1  by 2 µm) was the presence of electron-dense capsules or mucus-like structures located at the cell poles (orange signal in Fig. 6B). All cells that were stained positive with a corresponding FISH probe showed a consistent morphology, which was taken as a measure of probe specificity.

• Open in new tab
• Download powerpoint

FIG 6

Visualization of sponge-associated Chloroflexi in Aplysina aerophoba mesohyl using FISH-CLEM. (A and B) SAR202 cells are displayed in green, Anaerolineae in red (A), and Caldilineae in orange (B). The nucleotide stain DAPI (blue) served as reference for the localization of unstained cells. In both panels, the picture on the right is the overlay of all probes and DAPI. Bars, 1 µm.