Alternative Splicing Regulatory Element Analysis
Based on EST Clustering

 Canzhu Yang and W.Richard McCombie
                         

Introduction

Alternative splicing of pre-mRNA is a major cellular process that can affect quantitative control of gene expression and functional diversity of proteins. It was estimated that at least 50% of all human genes are subject to alternative splicing. Approximately 15% of human genetic diseases and developmental defects have been correlated with disruptions of alternative splicing control in particular genes. The regulation of alternative splicing is a complex process involving multiple steps. Splicing signals such donor and acceptor consensus sequences are important in determining the boundaries of introns during the splicing reaction, but they are not always sufficient to identify correctly the intronic and exonic sequences. Recently investigation suggested that the proximity of some cis-elements to splice sites play a crucial role in alternative splicing regulation. These elements are frequent targets of mutations in human genetics disease.

For example, the BRCA1 gene is responsible for approximately half of inherited breast cancer. Adrian and colleagues found that exon 18 of BRCA1 gene can be skipped as a result of disruption of exonic splicing enhancer (ESE) in the exon. The skipping exon 18 results in an abnormal BRCA1 protein lacking a particular segment, and a high incidence of breast cancer in the affected families.

The purpose of our study is to integrate clustered ESTs with several published molecular models of alternative splicing regulation (Fig. 1) to detect splicing regulatory elements at a genomic level of resolution.

Figure 1. Models of splicing silencing.

(a) Silencing by direct competition. A splicing-stimulatory factor, such as an SR protein, binds to an exonic splicing enhancer (ESE) (+; left). A splicing-inhibitory factor, such as a heterogeneous nuclear ribonucleoprotein (hnRNP), binds to an exonic splicing silencer (ESS) (-; right). Because the ESE and ESS overlap, the binding of the positive and negative factors is mutually exclusive. If the positive factor has a higher binding affinity or higher concentration than the negative one, the alternative exon is included; if vice versa, then it is excluded.

(b) Silencing by exon looping. A splicing inhibitory factor, such as an hnRNP protein, binds to duplicate intronic splicing silencer (ISS) elements present in the introns that flank an alternative exon. Dimerization of the bound protein brings the ISS elements into juxtaposition, causing the alternative exon to loop out and to be skipped by the splicing machinery.

(c) Silencing by nucleation and cooperative binding. The alternative exon harbours ESE and ESS elements at some distance from each other. In the absence of an inhibitory factor, the ESS element remains unoccupied, and an SR protein can bind to the ESE and stimulate exon inclusion (+; left). An inhibitory factor (-; right) initially binds to a high-affinity binding site in the ESS, and nucleates cooperative binding of additional inhibitory olecules, which polymerize along the exon and displace the ESE-bound SR protein or prevent its initial binding, resulting in exon skipping.

Methods

Human EST alignment data were downloaded from the UCSC genome web site ( Release: hg12 ). A software tool, DNAFE, was developed to facilitate the analysis of alternative splicing regulatory elements. Genomic sequence-based EST clustering was used to focus attention on regions of the genome that correspond to expressed genes. EST clusters overlapping with masked repeat regions were excluded from our analysis. Remaining clusters were classified into intra- and intergenic categories. To predict splicing regulators, we collected a set of experimentally verified splicing factor binding sites including 16 ESEs, 5 ISSs and 3 ESSs. Genomic DNA fragments corresponding to EST clusters were extracted from the genomic template and were screened to identify splicing factor binding sites. We combined several molecular models of alternative splicing regulation (Fig. 1), with a systemic analysis of EST clusters harboring Alternative Splicing Regulatory Site (ASRS). SNP information was integrated with the clusters to detect potential point mutations in ASRS.

Results

Unannotated Long Terminal Retrotransposon

A total of 444 clusters contained both hnRNP_H2 (ESS) and hnRNP_H1 (ESE) sites. For 90% of the cases, the distance between ESS and ESE was 878 bp.85% of the clusters are located in intergenic regions, and the remaining clusters are located in introns. None of these clusters overlap with any masked repeat element. A BLAST search that none of the clusters match an EPD. A consensus sequence(916 bp) was obtained from Clustalw. Pfam Server results suggested a match with Pfam protein Transposase-22, with 948 bp overlap (including gap) in reverse orientation. Approximately half of the clusters had G+C content lower than 40% (average 44%). The average cluster size is 1803 bp. FGENESH predicted that 120 clusters contain a poly(A) site, 3 of the clusters contain a Transcript Start Site (TSS). We conclude that this cluster is likely to represent an unannotated Long Terminal Retrotransposon (LTR).

Genomic Location Preference of ASRS Clusters

Our results indicated that clusters containing ASRS are widely distributed in the genome. A total of 9219 clusters was obtained. Excluding single exon cases, 3854 out of 8970 clusters (43%) are located within an intron or overlap the last exons of UCSC annotated RefGene. 66% of ESS associated clusters are located in an intron. These percentages imply that intronic and 3' terminal clusters preferentially contain splicing regulators. Similar conclusions can be drawn from a subset of the clusters with multiple ASRS (cooperative occurence). This result is consistent with previous reports that the intron and 3' end of genes are the major locations for regulatory elements.

Splicing Pattern Preference of ASRS Clusters

Based on their alignment to exons of annotated RefGene, these results also suggested that ASRS clusters are preferably aligned with splicing patterns. The clusters were classified into 11 different splicing categories. We found 37% of the clusters are in the category that overlaps with the whole exon with extension boundaries. 27% of the clusters overlap with exons at the 3' end and with the extension boundary.

Estimated Fraction of Genes Involving in Alternative Splicing

To estimate how many genes involve in alternative splicing event. A total of 15893 genes ( transcripts from RefGene ) were used to align with ASRS clusters. Our results indicated that 46% (3635/7923) and 44% (3506/7970) of genes from plus and minus strand respectively are associated with ASRS clusters. Based on above molecular models, each ASRS cluster may represent one alternative or regulatory exon. Thus it is reasonable to infer around 45% of genes subject to alternative splicing. This percentage is consist with some previous report

Web Server

We have developed a web server, ASREB ( Alternative Splicing Regulatory Element Browser), which allows users to access EST clusters associated with ASRS and analyze results. The users can query a cluster by chromosome region, or accession number of Contig, EST or gene. The clusters can be viewed by their genomic locations or splicing categories. Query by accession number of one gene, ASREB will return all of clusters associated with the gene. The assembly of these clusters and their genomic locations will provide a preview of alternative splicing variants for the gene. Each cluster is linked with the UCSC genome browser and the available comprehensive information includes EST distribution, alignment of clusters and genes EPD and Pfam. ASREB also provides a graphic display of this information. Our results will facilitate investigation of alternative splicing and human disease.

Selected References

[1]. Lopez AJ. Aternative splicing of pre-mRNA: Developmental consequences and mechanisms of regulation. Annu. Rev. Genet. 1998, 32: p279-30
[2]. Modrek B., Lee C. A genomic view of alternative splicing. Nature Genet. 2002, Vol.30, No.1, p13-19
[3]. Brett D., Pospisil H. et al. Alternational splicing and genome complexity. Nature Genetics. 2001, Vol.30, No.1, p29-30.
[4]. The international Genome Sequencing Consortium. Initial assembly and analysis of human genome. Nature 2001, 409, p860-921
[5]. Crag V. etal. Human genome sequence. Science. 2001, Vol.16, p1304-1351
[6]. Ceres JF. & Kornblihtt AR. Alternative: multiple control mechanisms and involvement in human disease. Trends in Genetics. 2002, Vol.18, No.4, p186-193
[7]. Blencowe BJ. Exonic splicing enhancers: mechanism of action, diversity and role in human genetics disease. Trends in Bio. Sci. 2000, Vol.25, No.3, p106-110
[8]. Cartegni L. and Krainer AR. 2002. Disruption of an SF2/ASF-dependent exonic splicing enhancer in SMN2 causes spinal muscular atrophy in the absence of SMN1. Nature Genet. 30: p377-384. 
[9]. HX.Liu, Cartegni L., Zhang MQ., and Krainer AR. 2001. A mechanism for exon skipping caused by nonsense or missense mutations in BRCA1 and other genes. Nature Genet. 27: 55-58.
[10]. Cartegni L., Chew SL., Krainer AR. Listening to silence and understanding nonsense: exonic mutations that affect splicing . Nature Rev. Genet., 2002, Vol.3, No.4, p285-298
[11]. Fairbrother WG., Yeh RF., Sharp PA., Burge CB. et al. 2002. Predictive Identification of Exonic Splicing Enhancers in Human Genes. Science, Vol.297, No.5583, p1007-1013
[12]. http://www.lecb.ncifcrf.gov/~toms/paper/rfs/latex/index.html
[13]. http://www.cgen.com/

Acknowledge

Citing this work:
Canzhu Yang and W. Richard McCombie, Computational Analysis of Alternative Splicing,
http://nucleus.cshl.org/genseq/ 

Maintained by Canzhu Yang. Questions/Comments please email to:
yangc@cshl.org Last update: May 10, 2003