Sequencing overview, P. lutea and I. palifera skeletal microbiome
We sequenced a total of 1.6383 billion read pairs for P. lutea (3.39%-63.66% host) and 1.2952 billion read pairs for I. palifera samples (Supplementary data file). We obtained 250 high-quality MAGs from P. lutea (average completeness (± standard deviation): 92.89±5.53% and contamination: 2.22±1.72%) and 143 from I. palifera (avg. completeness: 93.89±5.46% and contamination: 1.91±1.61%) (Supplementary Figures S1 and S2). Of the 250 P. lutea MAGs, 235 were bacterial and 15 archaeal and 141 of I. palifera 143 MAGs were bacterial and only 2 archaeal, based on GTDB-tk classification (Figure 1 a and b). A total of 113 MAGs (69 P. lutea and 44 I. palifera) had at least one copy of the 16S rRNA gene predicted in them (Figure 1 a and b; Supplementary data file).
These MAGs spanned the vast majority of microbial lineages (34 phyla and 57 classes) in the coral skeleton (Figure 1, Supplementary data file), including bacteria from phyla Proteobacteria (147 MAGs), Bacteroidota (75), Planctomycetota (42), Desulfobacterota (12, including lineages B and F), Firmicutes (12, including lineages A, F, G and H), Cyanobacteria (11), Verrucomicrobiota (11), Chloroflexota (10), Myxococcota (9), Gemmatimonadota (5), Bdellovibrionota (5), Actinobacteriota (4), Chlamydiota (4), Patescibacteria (4), SAR324 (4), Acidobacteriota (3), Spirochaetota (3), Bipolaricaulota (2), Calditrichota (2), AABM5-125-24 (1), DSWW01 (1), Elusimicrobiota (1), Fibrobacterota (1), Marinisomatota (1), Nitrospinota (1), Omnitrophota (1), SM23-31 (1), Sumerlaeota (1), Zixibacteria (1) and archeal phyla Nanoarchaeota (7), Thermoplasmatota (4), Thermoproteota (3) Aenigmatarchaeota (2, including lineage A) and Iainarchaeota (1).
P. lutea and I. palifera harbour different skeletal microbiome
Comparing microbial communities recovered from MAGs which meet completeness (≥90% and contamination (≤10%) thresholds, and basing our results on the presence and absence of MAGs from the two coral species, we identified some MAGs (Actinobacteriota, Calditrichota, Sumerlaeota and Zixibacter) to be unique to I. palifera, and some other MAGs (AABM5-125-24, Bipolaricaulota, Desulfobacterota, DSWW01, Elusimicrobiota, Fibrobacterota, Firmicutes, Marinisomatota, Nitrospinota, Omnitrophota, Patesibacteria, SAR324 and SM23-31) to be unique to P. lutea. We recovered one archeal MAG each of Thermoproteota and Nanoarchaeota from I. palifera metagenomes, whereas Iainarchaeota, Aenigmatarchaeota and Thermoplasmatota MAGs were recovered from P. lutea metagenomes.
MAGs recovered from P. lutea were differentially abundant among colonies, whereas the relative abundance of MAGs appeared stable among the colonies of I. palifera (Supplementary Figure S3). P. lutea skeletal samples were dominated by MAGs from bacterial classes Alphaproteobacteria, Vampirovibrionia and Planctomycetes and one sample (PL25b) was also dominated by archaeal phyla Thermoproteota (Supplementary Figure S3). In contrast, I. palifera skeletal samples were dominated by MAGs from bacterial classes Bacteroidia, Cyanobacteria, Anaerolineae and Polygania, with one colony (IP29b) harbouring a relatively high abundance (45.76%) of Cyanobacteria MAG (IP29b_bin.176) (Supplementary Figure S3).
Skeletal bacteria show the potential to engage in symbiosis with eukaryotes
Recovered MAGs on an average encoded 0.56±0.31% (P. lutea), 0.61±0.38 % (I. palifera) Eukaryotic Like Proteins (ELPs) per genome. MAGs had a broader range of ELPs including, WD40 (P. lutea: 2.65±4.06; I. palifera:4.68±6.32) and HEAT repeats (P. lutea: 2.65±5.04; I. palifera: 3.28±4.77) (Figure 2 a and b). The most abundant group of ELP in MAGs from P. lutea and I. palifera were TPRs (P. lutea: TPR_16, Pfam: PF13432, avg. proteins: 3.88±3.75; I. palifera: TPR_12: Pfam: PF13424, avg. proteins:4.94±8.27). MAGs harboured relatively low numbers of WD40 and HEAT repeat proteins, with the highest count in a MAG from Candidate phylum SM23-31 (37 WD40 repeat proteins), in P. lutea (PL23a_bin.125) and a MAG from class Bacteroidia (39 WD40 repeat proteins) in I. palifera (IP29b_bin.15). A MAG belonging to class UBA1135 (phylum: Planctomycetes) harboured 47 and 25 HEAT repeat proteins in P. lutea (PL25a_bin.29) and I. palifera (IP29b_bin.26, respectively (Supplementary data file). ARPs were the least abundant ELPs in the MAGs (P. lutea: 2.09±2.6; I. palifera: 2.58±2.25). Out of 235 bacterial MAGs from P. lutea, no ARPs were identified in 72 MAGs, 63 MAGs had only 1 ARP and there were 9 MAGs encoding more than 10 copies of ARP. In contrast, out of 141 bacterial MAGs from I. palifera, 23 had none, 36 MAGs had only 1 copy and 2 MAGs had 10 ARPs (Supplementary data file). Microbes are considered host-associated if they devote more than 0.2% of their total gene repertoire to ARPs [63]. Keeping this conservative threshold as identified earlier, we identified only 10 MAGs belonging to 6 bacterial phyla from P. lutea and 5 MAGs from 3 phyla from I. palifera, meeting this criterion (Figure 2a and b). Further, all 3 Chlamydia MAGs from P. lutea and 2 Bdellovibrionia MAGs from I. palifera encoded >0.2% ARPs.
The skeletal microbiome harbours an array of oxidative stress alleviators
Approximately half of the P. lutea bacterial MAGs (114) had at least one copy of the dsyB (PF00891, PF16864) gene, conferring the ability to synthesise DMSP and 13 MAGs had at least one copy of DMSP_lyase (PF16867) gene able to metabolise DMSP to other potent antioxidants (Figure 1a). Though the ability to synthesise DMSP was identified in 48.5% of MAGs, only 8 MAGs have at least one copy of both dsyB and DMSP_lyase genes (Figure 1a), with 7 of these belonging to the class Alphaproteobacteria and 1 to Gammaproteobacteria (Supplementary data file). The catalase gene (PF00199) was identified in 13 MAGs. At least one copy of the superoxide dismutase, SOD gene (including SODC (PF00080) and SOD_Fe_N (PF00081)) was identified in 94 bacterial MAGs. In contrast, out of 141 bacterial I. palifera MAGs, 58 had at least one copy of the dsyB gene and 16 MAGs had a copy of DMSP_lyase (Figure 1a). Further, only 13 MAGs belonging to class Alphaproteobacteria (11 MAGs), Anaerolineae (1 MAGs) and Acidimicrobia had at least a copy of dsyB and DMSP_lyase genes (Supplementary data file). SOD genes were annotated in 74 MAGs and catalase genes were identified in 5 MAGs only.
Skeletal archaea and bacteria engage in nitrogen and sulphur metabolism
We identified that 87 P. lutea and 45 I. palifera MAGs harbour the potential to fix nitrogen with at least one copy of the nifH gene (PF00142) (Figure 1a). Ammonia oxidation, AmoA gene, was identified in 2 and 1 Archeal MAGs from P. lutea and I. palifera, respectively (Figure 1b).
We analysed the processes involved in nitrogen cycling, including nitrification, denitrification, nitrogen fixation, and assimilatory and dissimilatory nitrate reduction to obtain comprehensive insights and understanding of nitrogen metabolism by the members of the coral skeleton microbiome. The nitrogen fixation module (M00175) was identified as complete in several MAGs, with 10 MAGs of Chlorobia and 3 of Clostridia encoding complete nitrogen fixation modules in P. lutea (Figure 3a) and 2 Cyanobacteria MAGs, 4 Alphaproteobacteria MAGs and 1 Planctomycetes MAG harbouring the potential to fix nitrogen in I. palifera (Figure 3a). Interestingly, the oxygen-dependent regulatory nitrogen fixation module (M00524) mediated by FixL-FixJ genes was also complete in several MAGs belonging to Alphaproteobacteria, Gammaproteobacteria, Phycisphaerae and Planctomycetes (Figure 3b). The dissimilatory nitrate reduction module (M00530), producing ammonia from nitrate was complete in MAGs spanning different bacterial classes in the two coral species (Figure 3a and 3b). However, assimilatory nitrate reduction (M00531) ability was poorly represented, with only MAGs from Cyanobacteria and Alphaproteobacteria encoding the complete module. The denitrification module was complete in 1 MAG each belonging to Gammaproteobacteria in P. lutea (Figure 3a) and Alphaproteobacteria, Anaerolineae and Bacteroidia in I. palifera (Figure 3b).
In oceans, sulphur is available as inorganic sulphate that can be assimilated by microbes into organic compounds. We searched for the ability of coral skeletal microbes to assimilate inorganic sulphur and use it to produce organic compounds as well as for energy-yielding purposes. We identified MAGs of sulphate reducing bacteria (SRB), including members of Desulfobacteria, Desulfarculia and SAR324 encoding the complete dissimilatory sulphate reduction module (M00596), along with a few MAGs belonging to Gammaproteobacteria, Chlorobia and Alphaproteobacteria in P. lutea (Figure 3a), whereas in I. palifera only 2 Gammaproteobacteria MAGs had the complete module (Figure 3b). We found complete assimilatory sulphate reduction modules in MAGs from several lineages, including Alphaproteobacteria, Bacterodia, Binatia, Gammaproteobacteria, Phycisphaerae, Planktomycetes and Verrucomicrobiae from the two coral species (Figure 3a and 3b). Further, complete Anoxygenic photosystem II module (M00597) was identified in several MAGs belonging to purple sulphur and purple non-sulphur bacteria from different classes, including Alphaproteobacteria, Anaerolineae, Gammaproteobacteria, Gemmatimonadetes and others in both coral species (Figure 3a and 3b). Bacteria harbouring this module have the potential to use H2S produced by assimilatory and dissimilatory reduction of sulphate as primary electron donor.