A bioinformatics tool for hnRNP-A1 binding site prediction

Heterogeneous nuclear ribonucleoprotein A1 (hnRNP-A1) has been shown to bind directly to pre-mRNA and regulate pre-mRNA splicing. hnRNP-A1 is the most comprehensively studied splicing factor of its class, and as such, provides the best opportunity to design and evaluate a bioinformatics tool for predicting splicing factor binding sites. We designed a program to predict the binding sites of hnRNP-A1 based on three mechanisms that are known to stabilize RNP-RNA interactions.;The process works in two steps. The first is to annotate known intron/exon boundaries for the query sequence using the Ensembl genome database as reference. The second step annotates RNA binding sites in pre-mRNA for the query sequence and then predicts potential hnRNP-A1 binding sites. Predicted hnRNP-A1 binding sites are identified based on the proximity of hnRNP-Al binding site sequences to binding site sequences of either alternative splicing SR proteins, intronic/exonic splicing enhancers, branch point or multiple hnRNP-A1 sites that may allow the formation of an intron-exonintron loop. Binding site sequence tables and proximity parameters are adjustable in the program code. We have also developed a web-interface for the program that will allow any scientist to predict potential hnRNP-A1 binding sites in a gene of interest which may help identify genes regulated by hnRNP-A1 or to elucidate the mechanisms of pre-mRNA splicing. The software program is written in Java.;To assess and optimize the program, we performed a predicted binding site analysis using a reference test set of genes that have previously been confirmed experimentally as bona fide hnRNP-A1 targets. The reference set includes genes coding for Mus musculus Src, Hras1, Heterogeneous nuclear ribonucleoprotein A1 (Hnrpa1), Fibroblast Growth Factor Receptor-2 (Fgfr2), Adenylyl Cyclase Stimulatory G-protein Galphas (Gnas1), and the Homo sapiens genes SRC, HRAS1, Heterogeneous nuclear ribonucleoprotein (HNRPA1), Fibroblast Growth Factor Receptor-2 (FGFR2 ), Adenylyl Cyclase Stimulatory G-protein Galphas (GNAS1 ) and the tat gene of HIV (HIV-1 tat). Our results show that all hnRNP-A1 binding sites experimentally confirmed in these genes are correctly predicted using our software analysis. In addition, other possible novel hnRNP-A1 binding sites were identified and can be subsequently analyzed for secondary structure and experimental confirmation.;The software can be used genome wide to predict new A1 binding sites and to predict A1 regulated genes. Furthermore, the bioinformatics model established can be adapted to predict the splicing binding sites and target genes of other RNP proteins, including that of the proto-oncogene TLS.;The program can be accessed world wide at http://140.193.242.11/ .