Jump to comment:

• Type 1 diabetes association with OAS1: Splice site SNP is best functional candidate
  L. Leigh Field
  Published on: 27 April 2016

• Published on: 27 April 2016

  Type 1 diabetes association with OAS1: Splice site SNP is best functional candidate

  • L. Leigh Field, Professor
  • Other Contributors:
    • Vagn Bonnevie-Nielsen, Shao Lu, Min Li, and Dong-Jun Zheng

  Dear Editor,

  We read with interest the recent article in the J Med Genet by Tessier et al.[1] confirming our report[2] of association between type 1 diabetes and the 2´5´-oligoadenylate synthetase OAS1 antiviral gene. However, their conclusion differs from ours concerning which single nucleotide polymorphism (SNP) in OAS1 is most likely to produce the functional effect on diabetes predisposition – the exon 3 non-...

  Show More

  Dear Editor,

  We read with interest the recent article in the J Med Genet by Tessier et al.[1] confirming our report[2] of association between type 1 diabetes and the 2´5´-oligoadenylate synthetase OAS1 antiviral gene. However, their conclusion differs from ours concerning which single nucleotide polymorphism (SNP) in OAS1 is most likely to produce the functional effect on diabetes predisposition – the exon 3 non-synonymous SNP rs3741981 or the intron5/exon7 splice site SNP rs10774671.

  Since these SNPs are in linkage disequilibrium, determining which is more likely to be functional is not a simple exercise. We postulated that the splice site SNP, which as we previously showed[3] creates different isoforms of the enzyme and is highly significantly associated with OAS enzyme activity, is the best functional candidate, rather than the exon 3 SNP which Tessier et al. champion. There were two reasons for our conclusion:

  1) the splice site SNP was more strongly associated with diabetes in our data than the exon 3 SNP;

  2) we clearly demonstrated in our previous study[3] that while controlling for genetic variation at the exon 3 SNP, the splice site SNP was still significantly associated with enzyme activity, but the reverse was not true – there was no association of enzyme activity with the exon 3 SNP while controlling for genetic variation at the splice site SNP.

  Since the strong effect on antiviral enzyme activity is a plausible functional link to diabetes susceptibility, these two observations led us to suggest that the splice site SNP is the most likely determinant of diabetes predisposition. Tessier et al. found similar strength of diabetes association for the exon 3 and splice site SNPs, but further analysis (see below) convinced them that the exon 3 SNP was more important. They suggested that the exon 3 SNP could alter dsRNA binding, which is required for enzyme activation. Hartmann et al.[4] reported that dsRNA binds to OAS1 along an extensive positively-charged groove. There is no evidence that the GGC(Glycine) to AGC(Serine) substitution encoded by the exon 3 SNP is within this groove or alters dsRNA binding.

  Nevertheless, it is theoretically possible that some other effect of OAS1 genetic variation (other than changing enzyme activity) could also predispose to type 1 diabetes. The question of which SNPs create functional effects becomes even more pertinent as further studies reveal new OAS1 associations -- for example, it was recently reported[5] that susceptibility to SARS virus infection is significantly associated with both the exon 3 SNP and exon 7 SNP rs2660 (the latter is in tight linkage disequilibrium with the exon 7 splice site SNP).

  Therefore, to directly address the question of possible effects of the exon 3 SNP on diabetes susceptibility, we have re-analyzed our data in the same manner as Tessier et al. They argued that the critical test was to examine transmission of the haplotype G-A (G allele at the exon 3 SNP and A allele at the splice site SNP) to diabetic children from parents who were heterozygous for haplotypes G-A and A-A. Since the G allele is associated with diabetes at both SNPs, this particular transmission test would be expected to show overtransmission of G-A to diabetic children if the exon 3 SNP is functional (as they suggested), but no overtransmission of G-A if the exon 3 SNP is not functional (as we postulated). [Note -- in Tessier et al., the major and minor alleles at the exon 3 SNP are incorrectly called T and C, rather than A and G as read off the plus strand.]

  Tessier et al. observed 50 transmissions and 27 non-transmissions of the G-A haplotype from G-A/A-A parents, with a corrected p value of 0.03. In our families, there were 64 parents who were heterozygous G-A/A-A. The G-A haplotype was transmitted 46 times and not transmitted 52 times to their 98 diabetic children (difference not significant). These results do not replicate those of Tessier et al. In fact, when data from the two studies are combined, 96 transmissions and 79 non-transmissions no longer constitutes significant overtransmission of the G-A haplotype (chi-squared = 1.65, uncorrected p = 0.20).

  Since this critical test for a functional effect at the exon 3 SNP fails, we again conclude that the splice site SNP is the best candidate for the effect on predisposition to type 1 diabetes. Ultimately, biological rather than statistical tests will be needed to conclusively establish the effects of these OAS1 SNPs on enzyme activity and other processes such as apoptosis; we are currently performing such experiments.

  In their discussion, Tessier et al. stated they were "puzzled" by the fact that we found the diabetes relative risk of the OAS1 gene to be comparable to that of the insulin (INS) gene region. However, we clearly stated that our relative risks were obtained by comparisons with unaffected siblings, not with unrelated persons in the general population, which could alter the relative risks.[2] We suggested that our OAS1 data supports not only a predisposing effect of splice site allele G (as also found by Tessier et. al.) but a protective effect of splice site allele A, particularly in the context of other diabetes-predisposing genes, since the A allele was found significantly more often in unaffected siblings than in parental non-transmitted alleles.[2]

  Finally, in their discussion Tessier et al. suggested that there may be parental-sex-specific OAS1 effects, since the overtransmission of OAS1 susceptibility alleles was statistically significant from mothers but not statistically significant from fathers for all three SNPs examined (note that due to linkage disequilibrium, these are not independent tests).

  However, they did not report whether these differences between maternal and paternal transmission rates were statistically significant. Data in their Table 4 shows that there is also overtransmission of predisposing alleles from fathers but the effect is not as strong as from mothers; our calculations indicate that the differences between their maternal and paternal transmission rates are not significant (data not shown). They also suggested that mothers of diabetic children may have lower frequencies of diabetes-predisposing genotypes (GG + GA) than fathers (p = 0.045 for one of the three SNPs). In our families, we do not see any significant differences between mothers and fathers in frequencies of alleles transmitted/non-transmitted to diabetic or unaffected children, nor any significant differences between mothers and fathers in frequencies of predisposing genotypes or alleles (data not shown). Thus, given the available data, we conclude that there is currently little evidence for OAS1 parental-sex-specific effects on diabetes susceptibility.

  Conflicting interests: none declared

  References

  1. Tessier MC, Qu HQ, Frechette R, Bacot F, Grabs R, Taback SP, Lawson ML, Kirsch SE, Hudson TJ, Polychronakos C. Type 1 diabetes and the OAS gene cluster: association with splicing polymorphism or haplotype? J Med Genet Published Online First 13 Jul 2005; doi:10.1136/jmg.2005.035212 [Epub ahead of print]

  2. Field LL, Bonnevie-Nielsen V, Pociot F, Lu S, Nielsen TB, Beck-Nielsen H. OAS1 splice site polymorphism controlling antiviral enzyme activity influences susceptibility to type 1 diabetes. Diabetes 2005; 54(5):1588-1591.

  3. Bonnevie-Nielsen V, Field LL, Lu S, Zheng DJ, Li M, Martensen PM, Nielsen TB, Beck-Nielsen H, Lau YL, Pociot F. Variation in antiviral 2',5'-oligoadenylate synthetase (2'5'AS) enzyme activity is controlled by a single-nucleotide polymorphism at a splice-acceptor site in the OAS1 gene. Am J Hum Genet 2005; 76(4):623-633.

  4. Hartmann R, Justesen J, Sarkar SN, Sen GC, Yee VC. Crystal structure of the 2-specific and double-stranded RNA-activated interferon-induced antiviral protein 2-5-oligoadenylate synthetase. Mol Cell 2003; 12(5):1173-1185.

  5. Hamano E, Hijikata M, Itoyama S, Quy T, Phi NC, Long HT, Ha le D, Ban VV, Matsushita I, Yanai H, Kirikae F, Kirikae T, Kuratsuji T, Sasazuki T, Keicho N. Polymorphisms of interferon-inducible genes OAS-1 and MxA associated with SARS in the Vietnamese population. Biochem Biophys Res Commun 2005; 22;329(4):1234-1239.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top