From yeast to mice to chicks to stem cells – 4 ways RNA-Seq analysis sheds light on the world around us (a four part series) – Part I

Last month, as in my wont, I was taking a look at publications that had very recently used our software for analysis of NGS data. Of these, four were by groups who had used the RNA-Seq workflow from the Avadis NGS suite to run the analysis of their data. Intrigued by the fact that this research had been done in 4 diverse biological systems, I decided to take a closer look at their work.  I was hoping to discover how the RNA-Seq tool had been adapted (if differently) for each of these research problems.

Starting with research from a group elucidating the yeast mitochondrial transcriptome, I moved on to a paper investigating genes involved in tongue morphogenesis in birds vs mice, to work by a group demonstrating how the use of 2 bio-similars altered the transcriptome differently in mice suffering from Gaucher’s disease, and finally to research that showed that the knockdown of the housekeeping gene HPRT in murine embryonic stem cells altered developmental and metabolic pathways during neuronal differentiation. What follows is a 4-part series which offers a brief synopsis of the experimental work done in each system, and the part RNA-Seq data had to play in answering questions raised by the work.

In the first paper on my list, Turk E.M. et. al used bioinformatics and RNA-Seq analysis to map out the mitochondrial transcriptome in Saccharomyces cerevisiae or in yeast reference strain S288C. Since yeast is one of the organisms popularly used to model mitochondrial genetics, the publication of its mitochondrial transcriptome is important for future research in this field. The transcriptome, as per the authors, is basically “a parts list of all RNAs of a system and a description of their boundaries, their physical location on the genome, and their abundance”.   In order to map it, they used a combination of in-silico methods, RNA-Seq analysis, and RT-qPCR. This helped them to correct all promoter, origin of replication (ori) and tRNA annotations; aided in estimating the expression levels of all mitochondrial transcripts; demonstrated the presence of alternate splicing; and helped to determine the identity of a ribonuclease (RNase) that potentially sculpted some of this landscape. Here is my perspective on the role played by the RNA-Seq analysis in context to the bigger mapping efforts by them.

First some methodology: after sequencing RNA using an Illumina HiSeq2000 platform, and mapping reads with Bowtie, Avadis NGS was used to visualize and analyze mapped reads. Differential expression of genes was demonstrated using the DESeq statistical method (this is an R script that can be set up to run within the Avadis NGS suite).

The primary role played by the RNA-Seq data analysis in this paper involved quantifying the abundance of different RNAs in yeast’s mitochondrial transcriptome by analyzing the expression of 35 gene products under conditions of mitochondrial activation. However, by simultaneously measuring all the mRNA, rRNA, tRNA and other ncRNA, the authors could not optimize for the detection of transcription start sites (requires tRNA and rRNA subtraction for deep coverage), RNA 3’ ends or active ori using RNA-Seq analysis.

Interestingly, the authors were still able to use RNA-Seq data to precisely map the 3’ terminal nucleotide of the 24 mitochondrial tRNAs due to the inclusion of a non-encoded tri-nucleotide CCA, post- transcriptionally to all their 3’ ends. Additionally, after deriving a 20-base promoter sequence and a 5-part consensus sequence as the sequence for active ori, the authors were able to use RNA-Seq analysis to show that out of the 8 putative ori in the yeast mitochondrial transcriptome, only 3 were active. The other 5 appeared disabled due to the presence of insertions within the promoter sequence; this could be inferred from far fewer RNA-Seq reads associated with the five ori.

Another interesting result from RNA-Seq experiments was confirmation that for genes present on the same primary transcript, relative expression levels were different. RNA-Seq data was also used to demonstrate alternate splicing for the first time in yeast mitochondrial RNA.  Additionally, it could be used to show that all yeast mitochondrial proteins in strain S288C used AUG as their start codon instead of AUA.  So in spite of the limitations posed by the methodology in this paper, the RNA-Seq tool from Avadis NGS was used in smart ways to get meaningful information about the transcriptome. The authors were also able to establish the role of Dis3p as a mitochondrial RNase whose absence allowed the detection of RNA sequences such as introns and antisense transcripts (or mirror RNAs) in Dis3p mutants, again using RNA-Seq analysis.

The authors of this paper thus coupled basic quantification of RNA reads and some differential expression data from RNA-Seq analysis, with other results, to offer the reader a fairly complete view of the yeast mitochondrial transcriptome.

Next week, I take a look at how RNA-Seq from Avadis NGS was used to show that the alteration in expression of certain genes during development, changed the phenotype of tongues in mice compared to birds- a story of class differences indeed!