Perilipins are phosphorylated proteins in adipocytes localized at the surface of the lipid droplet in adipocytes, steroid-producing cells and ruptured atherosclerotic plaques 1234. These proteins are essential in the regulation of triglyceride deposition and mobilization 567. When protein kinase A is activated, perilipin A becomes phosphorylated and translocates away from the lipid droplet, which allows hormone-sensitive lipase to hydrolyse the adipocyte triglyceride core 89. Therefore, perilipin A increases cellular triglyceride storage by decreasing the rate of triglyceride hydrolysis. Mice knockout for the perilipin gene are lean, have increased basal lipolysis and are resistant to diet-induced obesity 1011. However, these mice also develop glucose intolerance and insulin resistance more readily, probably due to the elevated levels of non-esterified fatty acids.

Several studies determined the level of perilipin expression according to obesity status. Two studies found obese subjects displayed lower levels of perilipin than lean individuals 712 whereas another study showed perilipin mRNA and protein were elevated in obese subjects 13. Qi et al. previously showed that polymorphisms in the perilipin locus were associated with obesity-related phenotypes in American and Spanish White women 1415. Moreover, they showed that a particular haplotype was associated with increased obesity risk in Malays and Indians, but not in Chinese 16. In this study, we examined the genetic variability of the perilipin gene and possible associations with obesity, type 2 diabetes and related phenotypes in a French random sample of population.

Table 1 describes the genotyping conditions and frequencies of the 7 tested PLIN SNPs (issued from the NCBI dbSNP database). DNA from 90 individuals issued from the population study was used to estimate the SNP frequencies. The rs8179072 (A386V), rs8179070 (R274W), rs3743373 (E293K) SNPs were not detected at all. Only 1 heterozygote for both the rs6496589 (Pro194Ala) and rs8179071 (Ser348Leu) SNPs could be found (rare allele frequency = 0.6 %). Therefore we did not pursue the genotyping of these SNPs. The rare allele frequencies of the rs4578621 (5'UTR -1234 C>G) and rs894160 (intron 6) SNPs were 7.6 and 30 % respectively. Therefore, we genotyped these two SNPs in the whole population study (n = 1120). There were 958 (85.5%) GG, 153 (13.7 %) GA, 9 (0.8%) AA and 556 (49.6%) GG, 470 (42.0%) GA and 94 (8.4%) AA for the rs4578621 and rs894160 SNPs respectively. These frequencies were not different from the expected frequencies under Hardy-Weinberg equilibrium.

We compared the genotype distribution of the two SNPs according to BMI categories (normal weight: BMI<25 kg/m², overweight : 25≤BMI<30 kg/m² and obese : BMI≥30 kg/m²) and diabetes status in men and women separately (table 2). No significant differences in the SNPs frequency could be detected between BMI or diabetes categories.

Table 3 and 4 show the impact of the PLIN rs4578621 and rs894160 SNPs respectively on anthropometric and biological variables in men and women separately. The means of anthropometric variables (weight, BMI, waist and hip circumferences, or waist-to-hip ratio), plasma lipid (plasma cholesterol, HDL-cholesterol, LDL-cholesterol, triglycerides) or insulin concentrations were not statistically different between genotype groups neither in men nor in women. Men carrying the rare AA genotype of the rs894160 SNP had lower fasting plasma glucose levels than GG subjects (5.27 ± 0.64 vs. 5.56 ± 0.90 mmol/L or 1.65 ± 0.12 vs. 1.71 ± 0.15 log-transformed values for AA vs GG men respectively, p = 0.04 when adjusted for covariates).

Between the two SNPs, the linkage disequilibrium D' was 0.87 (p < 0.0001), the G allele of the rs4578621 SNPS being associated with the A allele of the rs894160 SNP, and the r² was 0.15. Therefore, three haplotypes covered 99 % of the possible haplotypes (haplotype frequencies, CG: 0.70, CA: 0.22, GA: 0.07). No significant association could be found between haplotypes and the phenotypes studied (data not shown).

In the present study, we assessed the impact of the genetic variability of the PLIN gene on obesity, type 2 diabetes and related phenotypes in a sample of men and women issued from a French random sample of population (n = 1120). The genotyping of two common SNPs (rs4578621 and rs894160, frequency = 7.6 and 30 % respectively) were performed. No significant associations could be detected between these SNPs and the anthropometric variables, plasma leptin, lipids, glucose and insulin concentrations, even when using haplotype analyses.

The only significant association we found in our study was that men carrying the rare AA genotype of the rs894160 SNP (also called11482 G>A) had significant lower fasting plasma glucose levels than GG subjects but this difference was statistically borderline (p = 0.04) and clinically marginal (around 5% lower) and therefore this association is not clinically of great consequence and probably obtained by hazard. Moreover, this polymorphism was not associated with type 2 diabetes neither in men, nor in women in our study. Although the 11482 G>A SNP seems to have a functional impact on the perilipin protein activity as Mottagui-Tabar et al. reported that the A allele was associated with enhanced basal and noradrenaline-induced lipolysis in human subcutaneous fat cells 7, it does not seem to have a major impact in human French populations.

Other studies have previously described associations between PLIN SNPs and obesity-related phenotypes, especially the rs894160 (11482 G>A) SNP. Qi L et al. showed that a PLIN specific haplotype was associated with an increased risk of obesity in Malays, Indians and Whites (odds ratio around 1.7) 141516. In Whites, the PLIN 11482 G>A was associated with obesity risk in women but not in men. We were not able to reproduce this association in our French study. Although our study bears on a large number of subjects (573 men, 547 women), its statistical power may still be insufficient to detect small associations. This hypothesis is supported by an a posteriori calculation which indicates that according to the observed allele distribution, the sample size has sufficient statistical power (1-β≥ 80%) to detect an odds ratio above 2.4 and 2.1 for obesity and 2.1 and 1.8 for type 2 diabetes for the rs4578621 and rs894160 SNPs respectively. Therefore, only major associations could be detected. Moreover, we can not exclude that other PLIN SNPs that we did not explore, taken individually or in haplotype combinations, might be associated with obesity phenotypes.

In conclusion, the PLIN rs4578621 and rs894160 polymorphisms do not seem to be major genetic determinants of obesity and type 2 diabetes risk in French men and women. Other larger studies and/or other PLIN SNPs may need to be studied to conclude definitely about the impact of the PLIN gene variability on metabolic diseases.

The author(s) declare that they have no competing interests.

FA, ST genotyped the population samples. AM analyzed the data and wrote the paper. DC, DA, JF and PA enrolled the participants and contributed to writing the paper.