In this study, we functionally characterized two upstream visfatin promoter SNPs, c.-3187G > A (rs11977021) and c.-1537C > T (rs61330082), that were previously detected in the upstream promoter region of visfatin gene in our cohort of severely obese children [12]. Through the use of the transcription factor binding site prediction program PROMO, the c.-3187G > A and c.-1537C > T SNPs in the visfatin promoter were predicted to alter the ER-alpha and AP-2alphaA transcription factor binding sites respectively (Fig. 1). Wang et al. reported that the c.-1537C > T SNP altered promoter binding to AP-1 [9] but the study did not examine the promoter binding affinity to AP-2alpha2 which can also be a likely transcription factor that is affected by the c.-1537C > T SNP due to the homology between the promoter sequence and AP-2alphaA as predicted by PROMO.
The 3.7kbp visfatin variant promoter sequence containing both c.-3187G > A and c.-1537C > T SNPs was shown to have a significantly higher luciferase activity than that of the wild-type 3.7kbp promoter (Fig. 2). This indicated that the 3.7kbp variant promoter has an increased transcriptional activity and this is in accordance with our previous finding that the two SNPs were associated with elevated serum visfatin (6.17 ± 0.76 ng/ml vs. 3.92 ± 0.44 ng/ml) in our cohort of severely obese children [12].
In addition, we examined the transcriptional activity of the 1.6kbp variant visfatin promoter containing only c.-1537C > T SNP and found that there is no significant difference in transcriptional activity as compared to wild-type 1.6kbp promoter (Fig. 2). Our finding is similar to that reported by Tokunaga et al. who also found no difference in transcriptional activity between variant 1.6kbp visfatin promoter containing c.-1537C > T SNP and wild-type 1.6kbp promoter [11]. However, this differs from the reports by Wang et al. and Ye et al. who showed that c.-1537C > T SNP was associated with lower transcriptional activity [9, 10]. This difference may be attributed to the length of promoter sequence included in the luciferase reporter assay. While Wang et al. and Ye et al. have included ~216 bp and ~147 bp of the visfatin promoter sequence respectively [9, 10], our study has included a ~1.6kbp promoter sequence upstream of the ATG start codon similar to that in the paper by Tokunaga et al. [11]. The putative upstream promoter region of visfatin gene has been shown to contain multiple transcription factor binding sites [4]. Hence, a full promoter sequence from the ATG start site may account for the effect of other transcription factors on the transcriptional activity of the promoter.
The longer 3.7kbp wild-type visfatin promoter was found to have a significantly higher transcriptional activity as compared to the shorter 1.6kbp wild-type promoter. Moreover, the longer 3.7kbp variant promoter was also found to exhibit a higher transcriptional activity as compared to the shorter 1.6kbp variant promoter (Fig. 2). Though the in-vivo significance of this finding is unclear, we speculate that it might imply that the longer 3.7kbp region is more representative of the putative promoter. The higher transcriptional activity of the wild-type 3.7kbp promoter may also contribute in part to the accentuated in-vitro activity of 3.7kbp variant promoter. In addition, the two SNPs, c.-3187G > A and c.-1537C > T, at perfect linkage disequilibrium may have combinatorial effect on the transcriptional activity of the promoter as it has been shown that variants at linkage disequilibrium located at multiple enhancer sites may cooperatively dictate transcript expression [15]. Therefore, unlike previous studies which only examine the effect of c.-1537C > T SNP, our study takes into account distal promoter variants that may have combinatorial effect on the transcriptional activity of the promoter.
Both c.-3187G > A and c.-1537C > T SNPs have been reported as expression quantitative trait loci (eQTLs) in left ventricular heart tissue with trends of decreasing expression as the genotype changes from homozygous wild-type to heterozygous to homozygous variant [16]. However, our results are contrary to that of the reported eQTLs and this may be possibly due to these SNPs/reported eQTLs having different effects on different cell types, resulting in the difference in expression level. Hence, the choice of cell model should be taken into consideration when examining the effect of SNPs on expression level.
Data from our EMSA experiments indicated an increased binding of nuclear protein to the variant A and T alleles of c.-3187G > A and c.-1537C > T SNPs respectively (Fig. 3). This supports the increased transcriptional activity observed in the 3.7kbp visfatin variant promoter sequence (Fig. 2). Although c.-1537C > T SNP also showed an increase in nuclear protein binding, it did not result in an increase in transcriptional activity. This may be due to other upstream SNPs at different regions of the promoter sequence which may cooperatively affect enhancer or silencer function to regulate transcriptional activity.
Since c.-3187G > A does not exist in solitary without c.-1537C > T in our obese children cohort, we did not examine the effect of c.-3187G > A SNP alone on the transcriptional activity. Hence, we could not determine if both SNPs are needed to work cooperatively in order to exert an effect on transcriptional activity. In addition, we did not examine the specific transcription factor that is affected by the c.-3187G > A and c.-1537C > T SNPs respectively. Hence, we are not able to determine if the SNPs have affected binding with ER-alpha and AP-2alphaA transcription factors as predicted by PROMO.
In conclusion, we have demonstrated for the first time that visfatin variant promoter with both c.-3187G > A and c.-1537C > T SNPs result in increased in-vitro transcriptional activity. This supports our previous finding and postulation that these SNPs contribute to elevated visfatin levels which mediate higher triglyceride levels, severe systolic blood pressure and severe hypertension in obese children. This study supports the role of these upstream visfatin SNPs which could potentially affect phenotypic outcome in obese children through alteration of circulating visfatin level.