Abstract
Background Proteins arise from splicing of pre-mRNA using either the U2 spliceosome (most proteins); or the U12 spliceosome. The latter is believed to be less efficient, and it has been reported that autoantigens are much more likely than control proteins to use the U12 spliceosome. We set out to extend this work, and using new databases and a better contemporary understanding of splicing, to address the splicing mechanisms used for commonly encountered autoantigens.
Materials and methods We compared splicing characteristics of the UniProt autoantigen database with total genomic proteins, using the approach of Parada et al. (Nucleic Acids Res. 42:10564, 2014). Using this method, splice sites are given a “fit” score reflecting their fit to canonical scoring matrices for the two types of introns. We also determined the number of introns per gene and the average intron length.
Results We confirmed that autoantigens had more U12 spliceosome usage, although the difference was only twofold, much less than in the one previous study (Ng et. al., J All Clin Immunol 114:1463, 2004). Autoantigens had increased average number of introns per gene (24 vs 12) and an increase in noncanonical dinucleotides at the splice site. When they were scored for “problematic” splices, autoantigens were three fold more likely to have problematic introns (39% vs 13%).
Conclusions Genes encoding autoantigens have more introns and more “problematic” introns than control proteins. This may result in greater numbers of splicing errors, giving rise to proteins toward which tolerance has not been established. This model predicts autoantigenic epitopes to be near splice sites and should encourage studies of more extensive databases of autoantigens to extend analysis.
Acknowledgements Supported by grants from NIAMS and the Alliance for Lupus Research (PLC)
