Ribosomal RNA-Depletion Provides an Efficient Method for Successful Dual RNA-Seq Expression Profiling of a Marine Sponge Holobiont

Abstract

Investigations of host-symbiont interactions can benefit enormously from a complete and reliable holobiont gene expression profiling. The most efficient way to acquire holobiont transcriptomes is to perform RNA-Seq on both host and symbionts simultaneously. However, optimal methods for capturing both host and symbiont mRNAs are still under development, particularly when the host is a eukaryote and the symbionts are bacteria or archaea. Traditionally, poly(A)-enriched libraries have been used to capture eukaryotic mRNA, but the ability of this method to adequately capture bacterial mRNAs is unclear because of the short half-life of the bacterial transcripts. Here, we address this gap in knowledge with the aim of helping others to choose an appropriate RNA-Seq approach for analysis of animal host-bacterial symbiont transcriptomes. Specifically, we compared transcriptome bias, depth and coverage achieved by two different mRNA capture and sequencing strategies applied to the marine demosponge Amphimedon queenslandica holobiont. Annotated genomes of the sponge host and the three most abundant bacterial symbionts, which can comprise up to 95% of the adult microbiome, are available. Importantly, this allows for transcriptomes to be accurately mapped to these genomes, and thus quantitatively assessed and compared. The two strategies that we compare here are (i) poly(A) captured mRNA-Seq (Poly(A)-RNA-Seq) and (ii) ribosomal RNA depleted RNA-Seq (rRNA-depleted-RNA-Seq). For the host sponge, we find no significant difference in transcriptomes generated by the two different mRNA capture methods. However, for the symbiont transcriptomes, we confirm the expectation that the rRNA-depleted-RNA-Seq performs much better than the Poly(A)-RNA-Seq. This comparison demonstrates that RNA-Seq by ribosomal RNA depletion is an effective and reliable method to simultaneously capture gene expression in host and symbionts and thus to analyse holobiont transcriptomes.

Keywords: Demosponge; Holobiont; Hologenome; Porifera; Transcriptomics.

Fig. 1

Comparison between two different RNA-Seq methods in capturing transcriptomes of the A. queenslandica holobiont. A and B Taxonomic distribution of reads, expressed as percent of reads aligned to A. queenslandica (Aqu) and its proteobacterial symbionts AqS1, AqS2 and AqS3 for rRNA-depleted-RNA-Seq (A) and Poly(A)-RNA-Seq (B) libraries. C and D Correlation of expressed genes between three biological replicates for rRNA-depleted-RNA-Seq (C, replicates a–c) and Poly(A)-RNA-Seq (D, replicates d–f) libraries

Fig. 2

Gene coverage and depth of rRNA-depleted-RNA-Seq, Poly(A)-RNA-Seq and unenriched Poly(A)-RNA-Seq transcriptomes of the A. queenslandica holobiont. A Gene coverage, expressed as percent of genes to which reads aligned to A. queenslandica (Aqu) and its proteobacterial symbionts AqS1, AqS2 and AqS3. Shown are results for 1 (dots) and 5 (triangles) reads mapped to CDS thresholds for the three biological replicate libraries (rRNA-depleted-RNA-Seq and Poly(A)-RNA-Seq) and single bacterial-unenriched Poly(A)-RNA-Seq library. B Gene depth, expressed as boxplot of the number of reads mapped to each expressed gene

Fig. 3

Gene ontology analyses of transcriptomes generated by the RNA-Seq data sets of the partners in the A. queenslandica holobiont. A Percent of GO terms identified from expressed genes present in rRNA-depleted-RNA-Seq, Poly(A)-RNA-Seq and unenriched Poly(A)-RNA-Seq datasets with at least 1 read pair per CDS compared to genome GO terms. B Heatmaps presenting the average percent of genes represented by each GO term; adjacent dot plots show total gene number attributed to each GO term with the x-axis on a log2 scale. C Average number of expressed genes for the 30 GO terms that are most differentially represented between the different RNA-Seq datasets. Blue lettering, biological processes; green lettering, cellular components; black lettering, molecular functions. See Supplementary Table S1 for the number and proportion of expressed A. queenslandica, AqS1, AqS2 and AqS3 genes in each GO category