**4. Discussion**

Aiming to further clarify the reasons behind reported differences in the association of perilipin polymorphisms with obesity and obesity-related parameters, we analyzed the most commonly studied SNPs in *PLIN1* (11482G > A (rs894160), 14995A > T (rs1052700), and 13041A > G (rs2304795)) and the SNP in *PLIN2* Ser251Pro (rs35568725), causing a missense mutation in exon 6, in a cohort of prepubertal children showing significant differences in plasma leptin concentration between boys and girls [30]. In this population, potential confounding factors are fewer than in the pubertal and adult population. In our analysis, sex-dependent differences were observed in the association of the SNP rs1052700 of *PLIN1* with Apo A-I concentrations. Furthermore, an association of the SNP rs35568725 of *PLIN2* was found with NEFA, HDL-cholesterol, and Apo A-I concentrations in boys, which was not observed in girls. No significant associations of the polymorphisms in *PLIN1* and *PLIN2* were observed with body weight, BMI, LDL-cholesterol, or Apo B.

The association of *PLIN* SNPs with anthropometric traits and obesity has been described in studies in adults, including populations of different ethnic groups [10,11,31–33]. Although studies analyzing the association of *PLIN* SNPs with plasma lipid concentrations are scarce, some studies have also reported an association of these SNPs with triglycerides and HDL-cholesterol levels [10,15]. However, other studies have failed to detect association between these SNPs and obesity or obesity-related parameters [34–39]. Interaction of *PLIN* SNPs with nutritional factors may represent a plausible explanation for discrepancies among studies [16]. Another important issue is the sex-dependent association between SNPs at the *PLIN* locus and the obesity risk reported in adult populations [31,40] which may contribute to explaining divergent findings.

Few studies have analyzed the relationship of *PLIN* SNPs with obesity or obesityrelated alterations in children [19,20]. The design of these studies differs from ours, as Deram et al. [19] analyzed the effect of *PLIN* gene variation on weight loss in children with obesity aged 7–14 year, while the study of Tokgöz analyzed their association with obesity in a case-control study including 206 children with obesity and 102 healthy controls [20], which complicates efforts to compare our findings, as we analyzed a general child population and the analysis is not performed in overweight/obese children.

Here, we described a different sex-based association, particularly concerning the SNP rs1052700 of *PLIN1* and the SNP rs35568725 of *PLIN2* in a cohort of children in which a different nutritional status by sex, as reflected by plasma leptin levels, had been previously described [30].

Differences in diet have been associated with variations in the effect of the PLIN polymorphisms on obesity and obesity-related parameters [21–24]. Diet-induced changes in body fat and energy metabolism may be responsible for a different nutritional metabolism status, and these changes affect leptin levels. We hypothesized that the differences found in the effect of the polymorphism between boys and girls would be associated with the fact that the boys and girls in our population had significantly different leptin levels, and that plasma leptin levels could modulate the association of the polymorphisms with NEFA concentrations, which conditions its association with BMI and lipid metabolism. The *PLIN1* rs894160 and *PLIN1* rs1052700 have been associated with changes in abdominal fat and blood NEFA levels that occur in weight loss [41], which may also sugges<sup>t</sup> an influence on these associations exerted by changes in leptin levels associated with changes in body weight.

In our study, when analyzing the effect of the SNPs studied in children grouped according to plasma leptin levels, we observed that the relationship of the polymorphisms with BMI, Apo A-I, and NEFA varied depending on leptin concentrations. The effect of leptin on adipocyte metabolism has been demonstrated, and both direct and indirect effects of leptin on adipocyte metabolism have been suggested [27]. As adipocytes express leptin receptors, leptin may influence adipocyte metabolism directly and, as adipocytes are insulin-responsive, leptin can also modify adipocyte metabolism indirectly [27].

As described previously, perilipin is a protein that coats lipid droplets (LDs) in adipocytes [42] and plays an important role in lipolysis as, upon activation by protein kinase A, phosphorylated perilipin translocates from the lipid droplet and allows HSL to hydrolyze the TG and release NEFA. An important triggering role of leptin has been suggested for this intracellular lipolytic pathway. Indeed, PLIN1, the most abundant protein associated with LDs [43], is highly expressed in white adipocytes [44], and lower PLIN1 expression is related to higher rates of lipolysis [43]. Additionally, *PLIN* genetic variants may affect the protein content and lipolytic rates of adipocytes. Mottagui-Tabar et al. linked the rs894160 polymorphism of *PLIN1* to perilipin content in the adipocyte and basal and noradrenaline-induced lipolysis, with the 11482A allele being associated with a decreased perilipin content and with an increase in lipolysis [4]. The rs35568725 polymorphism of *PLIN2* has also been associated with an alteration of the gene that affects lipolysis and which is related to lower concentrations of TG [17]. Thus, based on the hypothesis that leptin triggers the lipolytic pathway that leads to phosphorylated perilipin and stimulates lipolysis, we assume a different effect of the genetic variants of *PLIN*, which determine the perilipin content in the adipocyte and PLIN functionality, and on BMI and lipid metabolism depending on leptin levels.

We should mention the lack of information regarding body composition as the main limitation of our study, as information on body fat might help us to understand differences on plasma leptin levels between boys and girls. An inherent limitation of all cross-sectional studies is the inability to demonstrate causality. Therefore, further studies are needed to confirm the causal nature of these associations.
