Characterising alternate splicing and tissue specific expression in the chicken from ESTs

Abstract

Alternate splicing is believed to produce the greatest diversity in transcriptional complexity and function in eukaryotic species. In this study, we present an analysis of alternative splicing events that occur in the chicken, using the recently sequenced genomic sequence and over 580,000 EST sequences mapped back to the genome. A carefully controlled EST-to-genome mapping pipeline is presented, based around the EXONERATE program using the est2genome model, which also considers several quality control steps to filter out erroneous matches. The data is then used to estimate the level of alternate splicing events with respect to Ensembl predicted transcripts. The EST-genome mappings are characterised at the exon level, in order to classify individual splicing events and provide estimates of novel transcripts not currently annotated by the Ensembl genome database. This is the first large scale analysis of this kind in an avian species, and suggests that chicken displays a similar level of alternate splicing as that found in other higher vertebrates such as human and mouse, both in terms of the number of genes that undergo alternate splicing events, and the average number of transcripts produced per gene. The EST data suggests alternate splicing may occur in some 50-60% of the chicken gene set and with an average of around 2.3 transcripts per gene which undergo this process. The EST data is also used to look at gene and transcript usage in the tissues sequenced in embryonic and adult libraries. Genes which display notable biases were analysed in more detail, including twinfilin-2 and embryonic heavy chain myosin. This also highlights several as yet functionally un-annotated genes which appear to be important in embryonic tissues and also undergo alternate splicing events. The analysis also demonstrates some of the difficulties involved in using EST-based data to annotate transcriptional activity in eukaryotic genes, where a broad spectrum of tissues and a large number of sequenced transcripts are required in order to fully characterise alternate splicing and differential expression.

Figure 1. Schematic of the EST to genome mapping pipeline

Processing steps in the mapping and analysis of ESTs, their exons, and mappings to Ensembl predicted genes are shown in schematic form, with appropriate statistics on the numbers of ESTs/exons/genes passing different stages shown in the figure. Further details are given in the Supplementary Materials.

Figure 2. Exemplar EST mappings to predicted Ensemble transcript structure for a single gene

The figures shows the nomenclature discussed in the Methods and Table 1 which shows how different EST exons once mapped to the gene transcripts are classified by the status which their 5' and 3' ends correspond to predicted transcript structure. The naming convention in the schema are: capital letters indicate a perfect match, normally followed by 0, which means the coordinate difference between the EST exon and the Ensembl exon is 0; the small letter indicates a non-perfect match followed by an “x”. A “L” or “l” refers to the 5' end and “R” or ‘r’ refers to the 3' end of the EST exon. The 5p and 3p only apply to the terminal EST exon, such as 5pR0, which means the 5' terminal EST exon has a perfect matching at the 3' end.

Figure 3. Gene Transcripts and ESTs from the Retinoic Acid Receptor gene

The gene transcripts which are a result of alternative splicing are shown at the top, the purple boxes are the exons and the thin lines are the introns.The ESTs that have more than 1 intron are shown underneath matching to their specific sequence. Their tissue specificity and developmental stage are also shown for each EST. The scale at the top shows the size of the transcripts and of the ESTs in Kilo bases (Kb)