Variation in the expression of integrin α2β1 leads to differences in collagen-induced platelet responses in vitro, reflecting the contribution of this receptor to hemostasis in vivo. In mice, there are two ITGA2 haplotypes, distinguished by several nucleotide substitutions in the 5′-regulatory/promoter region in linkage disequilibrium (LD) with a variable number of CA repeats in intron 1. Haplotype 1, present in seven strains (C3H/HeJ, C57L/J, DBA/2J, BALB/cJ, C57BL/6J, 129X1/SvJ, SM/J) contains 21 CA repeats at one site in intron 1. Haplotype 2, represented in four strains (SJL/J, SWR/J, A/J and FVB/NJ), contains only 7 CA repeats in that position. The strain AKR/J carries a unique haplotype, arising from a single crossover between the 5′-regulatory region of haplotype 1 and intron 1 of haplotype 2 downstream from the CA repeat site. No difference was observed in transcription rate initiated by the two 5′-regulatory/promoter haplotypes. Nonetheless, in every strain expressing haplotype 2 (7 CA repeats), a two-fold decrease in platelet surface α2β1 is found. The difference in CA repeat length affects the binding of heterogeneous nuclear ribonucleotide binding protein L (hnRNP L), which influences the rate and fidelity of mRNA splicing. This is the sole determinant of the two-fold difference in platelet α2β1 level between mouse strains. In humans, control of ITGA2 transcription and α2 expression is more complex. The two oldest human haplotypes, as determine by HapMap analyses (
Di Paola J et al., J. Thromb. Haemost., 2005; 3:1511–1521
), are distinguished by the single nucleotide polymorphism (SNP) C-52T, and we have reported that -52T is associated with at least a five-fold decrease in rate of ITGA2 transcription in vitro. In this study, we have identified a difference in CA repeat length at two sites in human ITGA2. One site is located in the 5′ regulatory region (−512 to −492) upstream from C-52T, where haplotype −52C bears 12 CA repeats, while haplotype −52T bears 10 or 11 CA repeats. A comparable CA repeat site is not present in the murine ITGA2 5′-regulatory region. The presence of 12 repeats increases transcription rate in vitro by two- to three-fold independently of the nucleotide substitution at −52. An additional difference in CA repeat length is seen at the second site, located downstream and within the 5′ region of intron 1. However, unlike the findings in mice, the intron 1 CA repeat length in human ITGA2 does not significantly alter the rate or fidelity of mRNA splicing. Increased CA repeat length was shown to enhance transcription of the human matrix metalloproteinase 9 gene (MMP9), and nuclear CA repeat-binding proteins were detected, but none has since been identified (Shimajiri S. et al., 1999; 455:70–74). Such proteins are likely involved in the enhancement of ITGA2 transcription observed in our study and are distinct from hnRNP L or other mRNA binding proteins that regulate mRNA splicing. Thus, CA repeat length polymorphisms play a role in post-transcriptional regulation of mRNA splicing in mice and transcriptional regulation in humans. The human ITGA2 promoter SNP C-52T and the polymorphic CA repeat site at −512 to −492 define the two oldest human haplotypes and synergistically modulate the expression of human integrin α2β1.