PPARγ agonist pioglitazone induces PNPLA3 expression in vivo in white adipose tissues of metabolically healthy obese patients.
PPARγ is the master regulator of adipogenesis through the modulation of adipogenic and lipogenic genes like ATGL/PNPLA2, the closest homologue of PNPLA3. Therefore, we investigated whether PPARγ regulates PNPLA3 transcriptionally in adipose tissue in vivo and in vitro. To this aim, we analysed white adipose tissue (WAT) as secondary analysis from the primary outcome of “The Apple & Pear trial (“Cellular Dynamics of Subcutaneous Fat Distribution in Obese Women”; ClinicalTrials.gov ID- NCT01748994) previously reported20, where forty-one women with obesity randomized to receive 30mg/day of pioglitazone, a PPARγ agonist, for 16 weeks or placebo20. WAT gene expression analysis showed that the pioglitazone treatment significantly increased mRNA of PNPLA3 (+ 25%) and FABP4 (+ 30%), an established PPARγ target gene, in the subcutaneous femoral fat (Fig. 1b) but not in the subcutaneous abdominal fat depot (Fig. 1a). Therefore, it is reasonable to suggest that PPARγ may regulate PNPLA3 in human adipocytes in a fat depot-specific manner.
Pioglitazone induces PNPLA3 in white adipose tissues in STAM model of MASH.
To explore whether PPARγ would also transcriptionally regulate the Pnpla3 gene in mice, we next investigated the STAM mouse model21 fed with the PPARγ agonist pioglitazone for two weeks. Pioglitazone treatment did not affect body weight, gonadal WAT, or liver weight (Fig. 2a, b, c, and d) but significantly increased the gene expressions of Pnpla3, Pparg2, Fabp4, and Pgc1a (Fig. 1e) in the adipose tissue, but not in the liver (Fig. 1f). Therefore, a PPARγ selective agonist upregulated PNPLA3 gene expression in the STAM model in a tissue-specific manner.
PPARγ agonist induces PNPLA3 expression in SGBS cells while an antagonist blocks it.
Differentiation of Simpson-Golabi-Behmel Syndrome (SGBS) preadipocytes into mature adipocytes in vitro causes the activation of genes similar to adipose tissue in vivo22. To ascertain whether PNPLA3 is induced in this system, we performed RNA analysis of PNPLA3 during SGBS preadipocyte differentiation. Figure 3a - d shows the gene expression levels of PNPLA3, PPARG2, FABP4, and C/EBPA during SGBS maturation (Fig. 3e). The PNPLA3 mRNA was less detectable in the undifferentiated cells (days 0 to 3) but was induced significantly on day 5 onwards. The expression pattern of the PNPLA3 mRNA mirrors the induction of PPARG2 and the PPARγ target genes, FABP4 and C/EBPA (Fig. 3a, b, c, d). The upregulation of these adipogenic and lipogenic genes is in line with the prominent lipid droplet accumulation observed on day 13 (Fig. 3f). In addition, there was a positive correlation between PNPLA3 and PPARG2 gene expression (r = 0.96; Fig. 3e), suggesting that PPARγ2 transcriptionally regulates PNPLA3 gene expression.
To establish whether PNPLA3 is under the transcriptional regulation of PPARγ, we next treated SGBS cells with the most potent PPARγ agonist, rosiglitazone (2 µM) or the PPARγ antagonist GW9662 (10 µM). Rosiglitazone treatment significantly induced PNPLA3, PPARG2, FABP4, and PCG1A on days 7 and 10. In contrast, the rosiglitazone-mediated induction of PNPLA3 was repressed markedly in cells that received the PPARγ antagonist GW9662 (Fig. 4a). Furthermore, gene expression analysis revealed a similar expression pattern between PPARG2, FABP4, and PGC1A (Fig. 4b - d) showing that PPARγ regulates PNPLA3.
PPARγ regulates PNPLA3 expression via a direct transcriptional mechanism.
The regulation of the PNPLA3 gene in response to the agonists and antagonists (rosiglitazone, pioglitazone, and GW9662) treatment suggests the involvement of a direct and active transcriptional mechanism. To confirm this, on day 3 of the differentiation process, SGBS cells were incubated in the presence or absence of 5 µg/ml actinomycin D, a transcriptional inhibitor of RNA polymerase23, for 1 hour, followed by the incubation with or without 2 µM rosiglitazone for the duration of the differentiation process. The actinomycin D blocks polymerase 2 that is induced by rosiglitazone. Therefore, other genes are not affected except PPARγ and its target genes. Actinomycin D treatment markedly reduced the rosiglitazone-mediated increase of PNPLA3 and PPARG2 mRNA transcripts (Fig. 5a and b). Transcriptional activation can also occur via a non-direct way requiring the neo-synthesis of an intermediate protein. SGBS cells were then incubated in the presence or absence of 10 µg/ml cycloheximide, a potent inhibitor of protein synthesis24, for 1 hour, followed by the incubation with or without 2 µM rosiglitazone for the duration of the differentiation process. Cycloheximide treatment did not reduce the rosiglitazone-mediated increase of PNPLA3 and PPARG2 mRNA transcripts. Interestingly, cycloheximide and rosiglitazone combination treatment increased the expressions of PNPLA3 and PPARG2 genes (Fig. 5c and d). Our results demonstrate that PNPLA3 mRNA regulation by PPARγ is a direct transcriptional mechanism that does not require de novo protein synthesis.
C/EBPα is required for the PPARγ agonist-mediated repression of PNPLA3 in mature adipocytes.
Pioglitazone treatment induced PPARγ target genes differently in the subcutaneous abdominal and femoral adipose tissues in healthy obese women (Fig. 1a and b). In the subcutaneous femoral adipose tissue, pioglitazone increased the expressions of PNPLA3 and FABP4 mRNA transcripts, whereas, in the subcutaneous abdominal adipose tissue, these genes were unaffected, suggesting that the action of PPARγ is adipose tissue depot-specific. To explore this further, we differentiated SGBS preadipocytes into mature adipocytes, followed by treatment with 5 µM rosiglitazone for 24 hours. We identified a set of genes repressed by rosiglitazone, including, surprisingly, PPARG2 itself, SREBP1c, and FASN, whereas PNPLA3 was not significantly reduced (p = 0.052) (Fig. 6a). Similarly, GLUT4 and ATGL, which are also direct targets of PPARγ2, were unaffected by rosiglitazone (Fig. 6a). However, rosiglitazone induced other genes, including FABP4 and LEPR (Fig. 6b). Our data set of genes is consistent with the observation that troglitazone treatment of rodents represses a set of adipogenic genes and induces other adipogenic genes25. C/EBPα and its corepressors, c - terminal binding proteins 1 and 2 (CtBP1 and CtBP2), were implicated in the PPARγ-agonist-mediated repression of selected visceral white adipose tissue genes25,26.
To establish whether PNPLA3 gene expression is also affected by C/EBPα in matured adipocytes, differentiated SGBS cells were transfected with three target-specific 19–25 nt siRNA targeting C/EBPα and then treated with rosiglitazone for 24 hours. Compared to wildtype, the C/EBPα expression levels in siC/EBPα cells were significantly reduced at both the mRNA (Fig. 6c) and protein level (Fig. 6d). Although the lipid droplet accumulation was comparable (Fig. 6e). Furthermore, the repression of some adipogenic genes by C/EBPα relies on the corepressors CtBP1 and CtBP225. CtBP2 but not CtBP1 was significantly downregulated in the siC/EBPα cells, whereas PGC1A and PGC1B were upregulated in the cells lacking C/EBPα (Fig. 6f). Interestingly, the selected fat genes repressed or unaffected by rosiglitazone like PPARG2, FABP4, FASN, and PNPLA3, were all significantly upregulated in rosiglitazone-treated adipocytes lacking C/EBPα as shown by the increase in PNPLA3, PPARG2, CD36, SREBP1c, FABP4, GLUT4, ATGL, and ADIPOQ mRNA transcripts (Fig. 6g and h). Our data thus suggest that C/EBPα presence is responsible for the rosiglitazone-mediated downregulation of PPARγ and some target genes in mature adipocytes.
Promoter analysis of PNPLA3 and EMSA identified two PPARγ binding sites.
As already demonstrated above, PNPLA3 was induced in response to the PPARγ agonists rosiglitazone and pioglitazone, albeit the direct role of PNPLA3 and PPARγ has not yet been shown. We performed a luciferase transient transfection assay to examine whether PPARγ induces reporter activity of the proximal promoter region of the PNPLA3 gene (Fig. 7a and b). To do so, a 1 kb DNA fragment of the human PNPLA3 promoter, including the transcription start site, was placed before a luciferase reporter gene in the presence or absence of PPARγ2 and rosiglitazone. Rosiglitazone significantly induces the reporter activity in the PPARγ and PNPLA3 promoter-luciferase co-transfected cells, similar to the positive control DR1, demonstrating PPARγ-mediated induction of PNPLA3.
Next, we explored which site(s) on the promoter region of the human PNPLA3 gene is responsible for the PPARγ-mediated induction of PNPLA3. To define the responsible site(s), we mapped four potential peroxisome proliferator response element (PPRE)-predicted sequences, PPRE 71, -103 to -93, -713 to -707, -790 to -780 (Fig. 7c) using online tools NUBIscan and Genomatix software. A gel mobility assay was performed on double-stranded oligonucleotides containing these PPRE-predicted sequences and a DR1 as a positive control and incubated with cytosolic and nuclear protein from SGBS preadipocytes and differentiated adipocytes. PPARγ nuclear protein bound strongly to the DR1 in differentiated adipocytes but not in the preadipocytes (Fig. 7d), confirming that PPARγ is crucial in the adipocyte differentiation process and is part of the complex bound to the DR1. Interestingly, among the four predicted PPRE sequences analysed, only two PPREs at positions − 713 − 707 and − 790 − 780 appear to bound to PPARγ in the differentiated adipocytes (Fig. 7e). The introduction of a mutation in the PPRE − 713 to -707mut, -790 to -780mutER4, and − 790 to -780mutDR1 (Fig. 7e and 8a) abolished binding activity. Next, we cloned the newly identified PPARγ binding sites in the PNPLA3 promoter, -790 to -780, -790 to -780mutER4, and − 790 to -780mutDR1 in front of a luciferase reporter gene, and transfected Cos-7 cells with PPARγ in the presence or absence of rosiglitazone. PPARγ and its agonist did not increase the reporter activity in the mutant sites, therefore demonstrating the inability of PPARγ to bind the mutated PPREs (Fig. 8b), which was in line with our gel shift data. Collectively, these results revealed, for the first time, the specific binding of PPARγ to two possible sites in the proximal promoter of the PNPLA3 gene, regulating PNPLA3 expression in adipocytes.