Decoupling of strain- and intrastrain-level interactions of microbiomes in a sponge holobiont

Abstract

Holobionts are highly organized assemblages of eukaryotic hosts, cellular microbial symbionts, and viruses, whose interactions and evolution involve complex biological processes. It is largely unknown which specific determinants drive similarity or individuality in genetic diversity between holobionts. Here, we combine short- and long-read sequencing and DNA-proximity-linkage technologies to investigate intraspecific diversity of the microbiomes, including host-resolved viruses, in individuals of a model marine sponge. We find strong impacts of the sponge host and the cellular hosts of viruses on strain-level organization of the holobiont, whereas substantial overlap in nucleotide diversity between holobionts suggests frequent exchanges of microbial cells and viruses at intrastrain level in the local sponge population. Immune-evasive arms races likely restricted virus-host co-evolution at the intrastrain level, generated holobiont-specific genome variations, and linked virus-host genetics through recombination. Our work shows that a decoupling of strain- and intrastrain-level interactions is a key factor in the genetic diversification of holobionts.

Fig. 1. Strain-level diversity and intrastrain-level genetic variations in individuals of the sponge C. concentrica.

a Phylogeny and abundance of archaeal and bacterial MAGs. Gray boxes show zero values. The maximum likelihood tree was constructed based on 40 single-copy marker proteins. The archaeal tree is rooted when the bacteria are treated as the outgroup, and the bacterial tree is rooted when the archaea are treated as the outgroup. The tree scale is shown. Branch supporting values ≥ 95% are shown as dots based on 1000 iterations of ultrafast bootstrapping of IQ-TREE. b Nucleotide diversity of the archaeal, bacterial, diatom, and sponge host MAGs. c Abundance of viral families. Gray boxes show zero values. d Nucleotide diversity of the viral families. For boxplots, data points of independent samples and the sample sizes are shown. The center lines represent median values. Box limits represent upper and lower quartiles. Whiskers represent 1.5 × interquartile range. Outliers are detected based on chi-squared scores using the R package outliers and are shown as dark red points. Source data are provided as a Source Data file.

Fig. 2. Virus-host interactive relationships and genome recombination events in the C. concentrica holobiont.

Virus-host links based on CRISPR, provirus, and HiC evidence are shown as lines in different colors connecting MAGs and vOTUs. HiC-links are retrieved from the metagenomes of the tissue of individual CC1 and the microbial cell fractions of individuals CC1-CC3. The thickness of a HiC link reflects the number of HiC read pairs supporting the link. Genome recombination events between MAGs and vOTUs are shown as gray lines. Two MAGs with ANI > 95% and < 99% are connected by spiral lines.

Fig. 3. Phylosymbiosis-like pattern and nucleotide diversity overlap of holobiont members between individuals of C. concentrica.

Illumina metagenomes of microbial cell fractions of individuals CC1-CC4 are used for analysis. a The analysis for phylosymbiosis-like patterns between sponge host genetics and community composition of the sponge host viromes (HiC-linked), the cellular microbiomes, and the viromes of cellular microbial symbionts (HiC-linked). The maximum likelihood tree of the sponge host is inferred from the consensus sequences of sponge ribosomal protein S6e in sponge individuals. The tree scale is shown. The tree is rooted to CC4 as its sequence shows the largest deviation from the others. Branch support value is based on 1000 iterations of ultrafast bootstrapping in IQ-TREE. The community compositions are clustered based on Bray-Curtis dissimilarity using the average method. The information on virus-host links is based on the results of Fig. 2. RF, Robinson-Foulds test (one-sided test). b The analysis for phylosymbiosis-like patterns between the popANI relationships of bacterial symbionts and the community compositions of their viromes. The virome compositions are clustered based on Bray-Curtis dissimilarity using the average method. RF, Robinson-Foulds test (one-sided test). c and d popANIs (c) and SGCs (d) of the sponge host viromes, cellular microbial symbionts, and their viromes between sponge individuals in comparison with the phylogenetic distance of the sponge host. For boxplots, data points of independent samples and the sample sizes are shown. The center lines represent median values. Box limits represent upper and lower quartiles. Whiskers represent 1.5 × interquartile range. Outliers are detected based on chi-squared scores using the R package outliers and are shown as dark red points. e and f Consensus trees of sample clusters based on popANIs (e) and SGCs (f) of cellular microbial symbionts (n = 33). Source data are provided as a Source Data file.

Fig. 4. Immune-related proteins encoded in C. concentrica-associated bacterial and archaeal symbionts and their viruses.

a Abundance of defense systems against mobile genetic elements in archaeal and bacterial symbionts. Gray boxes show zero values. Cbass, cyclic-oligonucleotide-based anti-phage signaling system. The phylogenetic trees are the same as the ones in Fig. 1a. b Presence of proteins encoding immune- and transposase-related functions in host-linked bacterial and archaeal viruses. Detailed information is in Supplementary Data 8.

Fig. 5. Intrastrain genetic variations driven by immune interactions and genome recombination between cellular microbial symbionts and their viruses.

a Mutation proportion and degree at restriction recognition sites in genomes of prokaryotic host-linked viruses in the metagenomes of microbial cell fractions of individuals CC1-CC4. Mutated proportion is calculated by dividing the number of mutated loci by the number of all recognition loci found in the viruses. The mutation degree is calculated by summing up the proportion of mutated bases in all recognition sites found in viruses. Gray boxes show viruses with recognition sites having average mapped bases of less than 100, which are excluded from the analysis. b Read coverage of bacterial contigs containing CRISPR arrays in the metagenomes of microbial cell fractions of individuals CC1-CC4. The array regions are shown with a shaded background. Sequencing depth is the number of bases mapped to each site. c The Spearman correlation (two-sided test) between intrastrain nucleotide diversity of viruses and their bacterial/archaeal hosts in their entire genomes in the metagenomes of microbial cell fractions of individuals CC1-CC4. d The Spearman correlation (two-sided test) between nucleotide diversity of viruses and their bacterial/archaeal hosts in their recombination regions in the metagenomes of microbial cell fractions of individuals CC1-CC4. A sequence identity threshold of 95% is used in the mapping step of inStrain to ensure read coverage. e Genomic map of the transposable virus V957 infecting the bacterium M10 Rhizobiaceae and M11238 Alphaproteobacteria. f Genome recombination regions (shaded sections) between V63657 and M6 Arenicellales, V36967 and M6398 Opitutales, and V4473 and M8024 Alphaproteobacteria. Source data are provided as a Source Data file.

Fig. 6. A model describing the microbial ecology and evolution at strain and intrastrain levels in individuals of the sponge C. concentrica holobiont.

At the strain level (ANI < 99% MAGs and vOTUs as defined in this study), the community composition of the microbial symbionts and sponge viruses is shaped by the genetic heterogeneity of the sponge hosts. At the intrastrain level (ANI > 99% MAGs and vOTUs down to single cells or viruses), this host impact generally disappears, and microbial cells and viruses are frequently exchanged between sponge hosts. The specificity between viruses and their cellular microbial hosts in sponge holobionts is preserved at the strain to species levels, and interspecific HGTs are generally blocked because of strong immunity. At the intrastrain level, interactions and coevolution between virus-host pairs are frequent resulting in holobiont-specific genetic diversity.