Complex microbiome underlying secondary and primary metabolism in the tunicate-Prochloron symbiosis

Abstract

The relationship between tunicates and the uncultivated cyanobacterium Prochloron didemni has long provided a model symbiosis. P. didemni is required for survival of animals such as Lissoclinum patella and also makes secondary metabolites of pharmaceutical interest. Here, we present the metagenomes, chemistry, and microbiomes of four related L. patella tunicate samples from a wide geographical range of the tropical Pacific. The remarkably similar P. didemni genomes are the most complex so far assembled from uncultivated organisms. Although P. didemni has not been stably cultivated and comprises a single strain in each sample, a complete set of metabolic genes indicates that the bacteria are likely capable of reproducing outside the host. The sequences reveal notable peculiarities of the photosynthetic apparatus and explain the basis of nutrient exchange underlying the symbiosis. P. didemni likely profoundly influences the lipid composition of the animals by synthesizing sterols and an unusual lipid with biofuel potential. In addition, L. patella also harbors a great variety of other bacterial groups that contribute nutritional and secondary metabolic products to the symbiosis. These bacteria possess an enormous genetic potential to synthesize new secondary metabolites. For example, an antitumor candidate molecule, patellazole, is not encoded in the genome of Prochloron and was linked to other bacteria from the microbiome. This study unveils the complex L. patella microbiome and its impact on primary and secondary metabolism, revealing a remarkable versatility in creating and exchanging small molecules.

Fig. 1.

Prochloron-bearing ascidians. (A) L. patella is shown at center, flanked by two other Prochloron-bearing species. (B) D. molle. (C) Light micrographs showing approximately 10-μm-diameter P. didemni cells examined in this study.

Fig. 2.

The P. didemni genome. The P1 genome, shown as the outer red circle, is compared with genomes from different cyanobacteria (blue and pink circles). The staggering red circle fragments represent the 32 scaffolds of the P1 genome; contigs are separated by black lines. Continuous blue circles indicate regions of greater than 85% DNA sequence identity between P1 and the P4, P2, and P3 genomes (from outside to inside, respectively). Green bars show the location of insecticidal toxin at bottom and secondary metabolite gene clusters at upper right that are present in P1 and P4, but not P2 or P3. Scattered blue bars indicate regions with greater than 85% DNA sequence identity between P. didemni and the cyanobacteria Cyanothece sp. ATCC 51142 (GenBank accession no. CP000806), Nostoc punctiforme PCC 73102 (GenBank accession no. CP001037), and Anabaena variabilis ATCC 29413 (GenBank accession no. CP000117). The pink bars indicate regions of greater than 60% translated protein sequence identity between P1 and Cyanothece sp. ATCC51142, showing that these two cyanobacterial species share a large set of core proteins despite a lack of nucleotide sequence homology. The innermost black line shows trinucleotide composition (χ²) of the P1 draft genome, indicating regions of possible lateral gene transfer. This analysis is biased as a result of the presence of sequence gaps.

Fig. 3.

Metabolic interactions in the ascidian metagenome. L. patella (Lower Right, schematic of tunic containing zooid and other bacteria) obtains nutrients and secondary metabolites from P. didemni and from other bacteria.

Fig. 4.

Hydrocarbons and sterols in P. didemni. (A) GC-MS of an ascidian extract from a tissue slice containing the animal and bacteria (Top) and a standard of heptadec-1-ene (Bottom). (B) Fragmentation pattern of peak at 9.87 min in extract (Top) and control (Bottom), showing that these compounds are identical. The additional peak at 9.91 is an internal C₁₇ olefin almost certainly resulting from the decarboxylative pathway to fatty acids. (C) GC-MS showing lanosterol-based lipids found in whole ascidian extracts. (D) Fragmentation pattern of C, showing observed spectra (above the lines in red) against expected spectra (below the lines in blue). These spectra indicate that lanosterol and an oxidized derivative are present in the ascidian sample.

Fig. 5.

Phylogenetic analysis of PCR-amplified KS domains. All unique (<95% sequence identical) KS clones obtained as described in Materials and Methods were used to generate a phylogenetic tree. Red indicates sequences derived from L2; blue indicates sequences derived from L3; green indicates sequences derived from L5; and black indicates previously described sequence relatives. Maximum parsimony analysis (Mega 5) was used to generate this phylogenetic tree with a bootstrap test of 1,000 replicates. Bootstrap values greater than 50% are shown. Clade names are provided based on the origin of previously identified sequences. For example, the sponge-like clade contains sequences that are most closely related to amplicons previously identified from sponges such as Discodermia (, 96) (However, these are not the sponge-specific sup genes from Poribacteria). trans-AT PKS genes form their own ancient clade regardless of sequence origin (76).

Fig. 6.

Correlation of PKS genes with patellazoles. (A) Patellazoles were found only in L2 and L3 among our extensive ascidian collection examined between 2002 and 2007, which includes many L. patella samples and many other species of Prochloron-bearing didemnids. (B) As an example of the genetic approach, PKS_11 was found in ascidians containing patellazoles but not in ascidians lacking patellazoles, including adjacent colonies on the reef. Further work is required to definitively connect these genes to patellazoles biosynthesis.

Fig. 7.

16S rRNA gene analysis. Visualization of the 16S analysis results by CloVR-16S: (A) Percentages of 16S sequences assigned to major taxonomic classes, with individual classes represented by different colors. All samples are dominated by Cyanobacteria, which includes P. didemni. (B) Complete-linkage (furthest neighbor) clustering of taxonomic classes based on log-normalized counts in each sample, as indicated by color in the scale bar. (C) Rarefaction curves showing that samples L1 and L2 contain more diverse and incompletely sampled microbial communities compared with L3.

Fig. 8.

Taxonomical classification of the non-Prochloron microbiome. P. didemni reads were subtracted from the total metagenome by BLASTn, and the remaining sequences were taxonomically assigned by using CloVR-Metagenomics. The results of complete-linkage (furthest neighbor) clustering of taxonomic classes based on relative counts are shown. The colors indicate the relative ratios of the reads in each sample on a logarithmic scale, as shown in the scale bar.

Fig. 9.

Microbiological approaches. By light microscopy, other bacteria beyond P. didemni such as filamentous cyanobacteria (A) and diatoms (B) were found in L2 and L3, but not L1 or other ascidians. Cultivation analysis led to isolation of many natural product synthesizing strains, such as Salinispora (C) and Verrucosispora. Figs. S3 and S4 show more details.

Fig. P1.

Primary and secondary metabolism in the tunicate microbiome. L. patella (Lower Right) is a colony of individual animals known as zooids that filter bacteria from seawater and consume them. The resulting nitrogen is recycled by the symbiotic cyanobacterium P. didemni (Upper Left). P. didemni also provides organic carbon to the host via photosynthesis. Both P. didemni and diverse other bacteria synthesize natural products (i.e., secondary metabolites) that are present in the holobiont and that have relevance to pharmaceutical development and biofuels.