Generation of full-length mRNA transcriptome and miRNA target prediction resources utilized in the study of miRNAs associated with response to Moritella viscosa infection in Atlantic salmon

Abstract

Atlantic salmon (salmo salar) is an important aquaculture species around the world. Infectious diseases, including the bacterial pathogen Moritella viscosa, which causes Winter Ulcer disease, remains a significant issue for salmon aquaculture. As part of the efforts to counter these problems, researchers have studied the genetic component of the Atlantic salmon’s disease response to viral and bacterial pathogens. MicroRNAs (miRNAs) are regulatory non-coding RNAs that play important roles across many species in fine-tuning the expression levels of proteins through targeted binding to mRNA transcripts. Included among the transcripts they regulate are many which encode immune-pathway proteins. The study of which genes are post-transcriptionally regulated by miRNAs in Atlantic salmon has, however, been hampered by the lack of access to 3’UTRs from full-length (FL) sequenced mRNA transcripts. An evolutionarily recent whole-genome duplication event in salmonids makes the short-read sequencing methods less reliable at producing such information, as these methods are less capable at differentiating between transcripts that are splice variants of each other or are from highly similar paralogs. We developed a generalized bioinformatics pipeline for the generation and functional annotation of full-length transcriptomes based on hybrid-corrected PacBio long-read SMRT sequencing to address the lack of full-length resources. This pipeline was employed to produce a de novo transcriptome for Atlantic salmon consisting of 71 461 FL sequenced mRNAs from 23 071 loci. Comparisons to the existing transcriptome resources revealed that the FL-transcriptome verified many of the predicted transcripts in the refseq database (25% of reference). However, the majority of the transcripts (70% of the FL-transcriptome) were novel splice variants or from uncharacterized paralogs. All transcripts in the FL-transcriptome unambiguously tied the UTR sequences to the CDS of the specific isoform or paralog variants. Capitalizing on this, all the unique 3’UTR sequences from the mRNA transcriptome were extracted in silico and used in miRNA target gene prediction analysis. The target prediction analysis utilized a weighted hybrid verification approach where each predicted target transcript had to be supported by the RNAHybrid algorithm as well as at least two of three other common target prediction tools. The complete set of target gene predictions, published as the MicroSalmon database, allowed users to search for the predicted targets of each of the Atlantic salmon miRNA along with their functional annotations. Known and putative novel cis-regulatory motifs present in the 3’UTR of each of the FL-transcripts were also characterized and may be searched for in the database. Differential expression studies of miRNAs (small RNA sequencing) and mRNAs (microarrays) identified 52 guide miRNAs and ~4500 protein coding transcripts associated with M. viscosa infection. The observation of concurrent changes in expression levels (both miRNAs and mRNAs) helped identify input lists for MicroSalmon, and predicted target genes of the disease responding guide miRNAs were identified among the differentially expressed protein coding transcripts. These functionally annotated target genes were subsequently used in enrichment analyses. These revealed that target transcripts of the disease-associated miRNAs were significantly overrepresented in gene networks associated with immune response, response to bleeding, wound healing, cell cycle control and response to stress. The distribution of enriched biological functions seemed to be partially tissue-specific, with headkidney materials expressing genes related to haemostasis and general immune responses, while the lesion site materials showed enrichment of genes related to death and breakdown of infected cells and the growth of new cells. Several orthologs of the conserved disease responding miRNAs displayed similar disease-related responses in other species, supporting them having the same roles in Atlantic salmon. These miRNAs also included teleost-specific immune miRNAs like miR-2188 and miR-7132. In conclusion, this study provided important new knowledge suggesting that certain miRNAs have important functions in response to bacterial infections. The findings may be utilized in future studies investigating whether the disease responding miRNAs have potential as infection biomarkers, as therapeutic tools, or as biomarkers in marker assisted breeding.