P465L‐PPARγ mutation confers partial resistance to the hypolipidaemic action of fibrates

Aims Familial partial lipodystrophic syndrome 3 (FPLD3) is associated with mutations in the transcription factor PPARγ. One of these mutations, the P467L, confers a dominant negative effect. We and others have previously investigated the pathophysiology associated with this mutation using a humanized mouse model that recapitulates most of the clinical symptoms observed in patients who have been phenotyped under different experimental conditions. One of the key clinical manifestations observed, both in humans and mouse models, is the ectopic accumulation of fat in the liver. With this study we aim to dissect the molecular mechanisms that contribute to the excessive accumulation of lipids in the liver and characterize the negative effect of this PPARγ mutation on the activity of PPARα in vivo when activated by fibrates. Material and Methods P465L‐PPAR mutant and wild‐type mice were divided into 8 experimental groups, 4 different conditions per genotype. Briefly, mice were fed a chow diet or a high‐fat diet (HFD 45% Kcal from fat) for a period of 28 days and treated with WY14643 or vehicle for five days before culling. At the end of the experiment, tissues and plasma were collected. We performed extensive gene expression, fatty acid composition and histological analysis in the livers. The serum collected was used to measure several metabolites and to perform basic lipoprotein profile. Results P465L mice showed increased levels of insulin and free fatty acids (FFA) as well as increased liver steatosis. They also exhibit decreased levels of very low density lipoproteins (VLDL) when fed an HFD. We also provide evidence of impaired expression of a number of well‐established PPARα target genes in the P465L mutant livers. Conclusion Our data demonstrate that P465L confers partial resistance to the hypolipidemic action of fibrates. These results show that the fatty liver phenotype observed in P465L mutant mice is not only the consequence of dysfunctional adipose tissue, but also involves defective liver metabolism. All in all, the deleterious effects of P465L‐PPARγ mutation may be magnified by their collateral negative effect on PPARα function.


| INTRODUCTION
Peroxisome proliferator-activated receptors (PPARs) are nuclear receptors that regulate energy homeostasis and coordinate biochemical processes that are involved in anabolic and catabolic processes.
This family of transcription factors comprises 3 members: PPARγ (with 2 isoforms: PPARγ1 and PPARγ2), PPARα and PPARδ. 1 PPARγ regulates adipose tissue development and expansion, and harmonizes the functional balance between lipogenic and lipolytic programmes. 2 Genetic defects in PPARγ cause severe metabolic lipodystrophy phenotypes, 3 known as familial partial lipodystrophy syndrome type III (FPLD3). 4 Among patients suffering from this syndrome, those carrying the P467L mutation (rs121909244) exhibit a lipodystrophic phenotype, hypertension, hyperglycaemia, hepatic steatosis, and severe dyslipidaemia, a complex phenotype that is partially recapitulated in the humanized P465L-PPARγ mutant mouse under different nutritional and genetically induced challenges, such as high-fat diet (HFD) feeding, and when backcrossed into ob/ob, apoliporotein E knockout (APOEKO) and Akita murine genetic backgrounds. [5][6][7] The isoform PPARγ2 is, under physiological conditions, preferentially expressed in WAT and BAT. Other organs, such as the liver, predominantly express the PPARγ1 isoform, but can, under pathological conditions such as overnutrition and obesity, induce de novo the expression of the PPARγ2 isoform. 8 This indicates that the types and relative amounts of PPARs coexisting in the same cell/tissue under specific physiological and pathophysiological conditions vary according to specific nutritional status and metabolic adaptations.
PPARα is another important member of the PPAR family, which plays a fundamental role in lipid oxidation and biosynthesis, gluconeogenesis, cholesterol catabolism and ketogenesis. 9 PPARα is detected in tissues characterized by high rates of β-oxidation, such as heart, skeletal muscle and liver. 10 Conversely, genetic ablation of PPARα in mice has confirmed its preferential involvement in fatty acid oxidation and has confirmed that, when PPARα is dysfunctional, it causes hepatic steatosis 11 and severe fasting hypoglycaemia. 12 According to the role of PPARα in controlling fatty acid metabolism, fibrates, a class of synthetic PPARα ligands, exert beneficial metabolic effects in patients with metabolic syndrome and in rodent models of obesity, insulin resistance and diabetes. For instance, a well-established fibrate, WY14643 (pirinixic acid), decreases plasma triglycerides, reduces adiposity and improves hepatic steatosis and insulin sensitivity in lipoatrophic mice. 13 Given that PPARs share common co-activators, co-repressors and partners, as well as DNA responsive elements (PPRE), we hypothesized that the changes in expression patterns or activity of PPARs may affect the transactivation capacity of the members of the family.
In vitro evidence indicates that PPARs exhibit promiscuity in their binding to specific coactivators/corepressors, known to be involved in complex functional crosstalks. This has been partially addressed in vitro by showing that mutants for PPARα, such as a dominant negative 14 or a ligand-binding domain-lacking mutant, 15 exert crossinhibition of the wild type (WT) form of PPARα, and also of PPARγ and PPARδ, by competition for coactivators. 14 Similarly, several PPARγ dominant negative mutants have been shown to repress the activity of PPARα in vitro. 14,16 The P467L-PPARγ mutation exerts a dominant negative effect on WT PPARγ in vitro by reducing the promoter turnover rate, by outcompeting the WT receptor for promoter binding sites, and by attenuating the release of a corepressor and recruitment coactivator. [17][18][19][20] The pathophysiological relevance of this crosstalk between PPARs in vivo has not been studied.
Our laboratory has previously shown that the humanized murine model P465L-PPARγ developed hepatic steatosis after 4 months on an HFD 5 in the same way that P467L human carriers do. 21 Here, we dissect the mechanisms leading to this phenotype in vivo and propose that the P465L-PPARγ mutation increases susceptibility to fatty liver through a mechanism that involves a partial deficiency of the transactivation capacity of PPARα. We found that P465L mice have increased levels of insulin and FFA, both risk factors associated with fatty liver.
P465L mice also display decreased levels of VLDL when fed an HFD and a partially impaired response to the hypolipidaemic action of WY14643. Moreover, gene expression profiling revealed that P465L-PPARγ negatively impacts on the transcriptional activation and repression mediated by PPARα agonists in a variety of metabolic pathways.

| Blood biochemistry
Enzymatic assay kits were used for determination of plasma-free fatty acids (Roche), triglycerides (Siemens), glucose and insulin (Meso Scale Diagnostics, Rockville, Maryland) according to manufacturers' instructions.

| Liver triglycerides content
Hepatic lipid content was measured using the Folch method as described previously. 22

| Lipoprotein separation by fast protein liquid chromatography (FPLC)
Pooled plasma of each experimental group was used for the isolation of lipoproteins using FPLC according to the protocol of the Diabetic Complications Consortium (https://www.diacomp.org/shared/document. aspx?id=14&docType=Protocol).

| Glycogen determination
Hepatic glycogen content was measured as described previously. 23

| Histology
Livers were dissected and fixed in 10% formalin; they were then cryoprotected (20% sucrose) and frozen in chilled isopentane prior to sectioning using a cryostat. Cryosectioned tissue was stained for lipid with Oil Red O (Sigma-Aldrich, St. Louis, Missouri).

| Lipid analysis
Liver tissue was homogenized with a tissue blender. Lipids were extracted using n-hexane/2-propanol. Tissue extracts were centrifuged at 1000 g to pellet debris. The lipid-containing organic phase was decanted and stored under nitrogen at −80 C until analysis. 24,25 Total lipids were subjected to base catalyzed transesterification, converting the acyl chains to fatty acid methyl esters (FAME). 26   nih.gov/tools/primer-blast/). (Sequences are available at http://tvp. mrl.ims.cam.ac.uk/primer-database-pagemax.) The geometrical average of 4 different genes (β2 microglobulin, β-actin, 18S and 36B4) was used as an internal control, following an already described normalization method. 27 Heat Maps were generated with the free software Multiple Experiment Viewer-MeV-(http://mev.tm4.org/).

| Statistical analysis
Two-way or three-way ANOVA was used for analysis of the interaction between genotype (G), diet (D) and treatment (T) for the fibrates intervention and between genotype (G), fasting (F) and/or diet (D) for the fasting cohort. IBM SPSS14 was used as statistical software.

| RESULTS
3.1 | P465L-PPARγ mutant mice are hyperlipidaemic and hyperinsulinaemic P465L mutant mice showed higher levels of FFAs and TGs, and a tendency to increased cholesterol in serum, compared to WT mice ( Figure 1A), independent of diet, recapitulating the hyperlipidaemia observed in human carriers of the P467L mutation. Both on a chow diet or after a short challenge with an HFD, P465L mice were normoglycaemic in the presence of hyperinsulinaemia ( Figure 1A), suggesting that their insulin secretion was able to compensate for their peripheral insulin resistance; moreover, when challenged with an HFD for a period of 12 weeks, P465L mice became hyperglycaemic ( Figure S1). The P465L mice also had elevated hepatic glycogen levels in the chow-fed state ( Figure S2A).
3.2 | P465L-PPARγ mutant mice are resistant to the pro-lipolytic effect of WY14643 in adipose tissue WY14643 treatment increased plasma cholesterol levels in both genotypes, as previously reported in other studies, 28 and also increased levels of β-hydroxybutyrate (BHB) in both genotypes ( Figure 1A). The increase in BHB is compatible with fibrate-mediated induction of hepatic β-oxidation, a response that was slightly increased (ns) in P465L mice, particularly those on an HFD for 4 weeks. Interestingly, when fasted o/n, P465L mice showed significantly higher levels of BHB than WT mice ( Figure S1), reflecting an increase in hepatic FFA delivery and β-oxidation in P465L livers. At the organismal level, acute treatment with WY14643 induced hepatomegaly in both genotypes, coupled with a specific decrease in the fat mass of WT mice, especially in gWAT, but not in P465L mice on an HFD ( Figure 1B and Figure S2). This difference in fat mass suggested   that the P465L mutation may confer some degree of resistance against the pro-lipolytic effect of fibrates in adipose tissue. 29

| P465L-PPARγ mutation promotes fatty liver and alterations in lipoprotein metabolism
We have reported previously that P465L mice showed increased liver mass and hepatic accumulation of TG content when fed an HFD for 16 weeks or when backcrossed with the ob/ob mouse. 5 In this new cohort of mice fed an HFD for only 28 days, the levels of hepatic TGs in P465L were already marginally higher (ns) than those in WT mice ( Figure 2A). Unexpectedly, when mice on an HFD were treated with fibrates, we observed an increased hepatic fat content in P465L, but not in WT mice (Figure 2A)  Figure 2C); however, the FADS1 index was downregulated in P465L livers. Interestingly, the FADS2 index was significantly increased in WT mice treated with WY14643, but to a lesser extent than in P465L mice ( Figure S2B). Interestingly, the expression of fads3, a putative desaturase associated with changes in PUFA levels, 30 was downregulated in P465L livers ( Figure 2C), suggesting that  We also observed a genotype-dependent decrease in levels of apob and minor differences in mttp (microsomal triglyceride transfer protein).
We subsequently analysed the effect of P465L on HDL lipoproteins. Unlike in humans, where fibrates increase HDL-C levels and apoa1 expression, 32 fibrates have been reported to have no effect on, and do not even decrease levels of HDL and apoa1 expression in nontransgenic wild-type mice. 33,34 Surprisingly, in P465L mice fed both a chow diet and an HFD, fibrates increased HDL levels in comparison to WT mice. This finding indicates that the P465L-PPARγ mutation interferes, directly or indirectly, with the normal response to fibrates in HDL metabolism.
Moreover, we observed an increased apoa2/apoa1 mRNA ratio in P465L mice ( Figure 3B). Of note, increased levels of apoa2 have been associated with increased pro-oxidative and pro-inflammatory responses, with alterations in the rate of HDL metabolization and with increased atherogenic risk. 35 FIGURE 4 Hepatic gene expression of candidate genes relevant in A, glucose metabolism, B, fatty acid uptake and mitochondrial/peroxisomal fatty acid oxidation programmes, and C, nuclear transcription factors is shown as log2 conversions of average gene expression data relative to controls (log2 100 = 6.6). Magnitude >6.6 and <6.6 denotes up-and downregulation, respectively, compared with WT controls on a chow diet, analysed by ANOVA (P < .05). Different coloured circles denote genotype effect (blue), treatment (red), diet (green), interactive effect of genotype × treatment (black), genotype × diet (white), diet × treatment (grey) and genotype × treatment × diet (orange) in patients with severe dyslipidaemia. 37 We observed that, in mice on an HFD, treatment with WY14643 reduced the triglyceride-enriched IDL/LDL-TG in WT mice, but not in P465L mice ( Figure 2B). This is in agreement with a decrease in the metabolization of TG from the IDL/LDL fraction, probably mediated by hepatic or peripheral lipases in When assessing the expression of FAO genes in mice fasted overnight we did not observe strong changes associated with the P465L mutation apart from abcd1, scp2 (GxF interactive effect) ( Figure S4).

G G T T D D G x T G x T G x D G x D G x T x D G x T x D T x D T x D
Finally, we also evaluated the expression of transcription factors other than pparα and pparγ that are involved in the transcriptional activation of multiple lipid-related genes ( Figure 4C) and whose expression could also be dysregulated by the activation of pparα (fibrates) and the P465L mutation. In this regard, fibrates and P465L affected the expression of srebp1, hnf4, rxr ( Figure 4C). Interestingly, the expression of pparγ at gene and protein levels ( Figure 4C and Figure S4A) was upregulated in P465L mutant livers, increasing the pool of the WT and also the mutant P465L-PPARγ, and thus competing with pparα in response to fibrates when both transcription factors are upregulated ( Figure 4C and Figure S4A).

| P465L-PPARγ interferes with the transrepression capacity of PPARα
Fibrates also exert anti-fibrotic and anti-inflammatory effects through transrepressive mechanisms. 43 We observed that, in HFD WT mice, administration of WY14643 reduces the expression of target genes such as serum amyloid A (saa1 and saa2) and fibrinogen ( Figure 5).
Interestingly, this effect was not observed in P465L mice, suggesting that this mutation also prevented the transrepression activity of PPARα.

| DISCUSSION
This work follows previous research from our group and others concerning the P465L mouse, a humanized model for the dominant negative mutant P467L-PPARγ that resembles the phenotype observed in patients and is characterized by a partial lipodystrophy, insulin resistance, hypertension and fatty liver.
Here, we provide evidence that the fatty liver observed in the P465L-PPARγ knock-in mouse involves a selective impairment in the transcriptional activation of PPARα. This is supported by in vivo data showing that P465L mice developed fatty liver on an HFD and were resistant to treatment with WY14643, a fibrate that acts as a PPARα activator, determining a phenotype highly reminiscent of the resistance to fibrates previously observed in PPARα KO mice. 44  the use of fibrates to treat fatty liver in several rodent models, 49,50 particularly when one of these two programmes may prevail over the other.
It is assumed that the association between lipodystrophy and NAFLD is a direct consequence of the failure of adipose tissue to expand and the consequent spillage of excessive lipids into the liver.
Here, we provide evidence that the fatty liver of the P465L lipodystrophic mice exhibits, not only quantitative, but also qualitative changes (increased MUFA/PUFA ratio) in the accumulated lipids.
These data are consistent with changes in hepatic lipid biosynthetic pathways that are characteristic of "classical models" of NAFLD and indicate that these qualitative changes in the fatty acid pool of P465L livers may reflect a common pathogenic signature of NAFLD, rather than a specific fingerprint driven by the presence of the PPARγ mutation.
Our expression profiling revealed selective impairment of PPARα preferentially regulated genes in P465L mice on a chow diet and treated only with vehicle. This indicates that, even when PPARα ligands are present at low physiological levels, the presence of the P465L mutation is enough to dysregulate the expression of those PPARα genes at a transcriptional level. The P465L mice on a chow diet also showed impaired expression of genes controlling mFAO and pFAO, well known targets of PPARα in mice. These effects were partially masked after treatment with WY14643 and/or an HFD. Additionally, we identified genes in the P465L livers that remained pathologically unresponsive to treatment with fibrates in comparison to WT livers, providing compelling evidence of resistance to the ligand-dependent activation of PPARα when P465L-PPARγ is present. Interestingly, genotype-driven differences were reduced when mice were fasted overnight (GxF interactive effect), indicating that the effect of the P465L mutation on PPARα function may become pathophysiologically relevant only at basal levels, when the expression/activity of PPARα is below a particular threshold or, alternatively, that it may be overcome by the hormonal adaptation taking place in response to starvation.
Despite the fact that our experimental model of treatment with HFD/WY14643 was not specifically designed to induce severe liver damage, we found that the expression of genes involved in fibrosis and known to be downregulated by PPARα activation were altered in P465L livers, suggesting that transrepression activity of PPARα was impaired also by the presence of the P465L-PPARγ mutant.
Globally considered, our data indicate that the pathogenesis of the fatty liver observed in P465L mice is more complex than being simply the result of the failure of the adipose tissue, and that it involves the combination of several pathogenic factors at the hepatic level, including the uncoupling of lipid storage in lipid droplets from the assembly, and transport and secretion of VLDL, associated with impaired hepatic FAO in a pathological context where peripheral uptake of lipids is likely to be compromised.
In summary, we have shown that P465L mice develop hepatic steatosis that is associated with increased lipid trapping and impaired VLDL secretion with HFD, resulting in qualitative changes in the hepatic fatty acids that recapitulate the fingerprint of common NAFLD models. We also provide biochemical data, hepatic lipid content and evidence of impaired expression of a number of well established PPARα target genes in P465L livers that supports the conclusion that P465L confers partial resistance to the hypolipidaemic action of fibrates. Moreover, our results show that the fatty liver phenotype observed in P465L mutant mice is not only the consequence of dysfunctional adipose tissue, but also involves defective liver metabolism. Despite the current lack of trials addressing the efficacy of fibrates in patients with partial lipodystrophy, our results raise concerns regarding the potential value of fibrates for the management of hypertriglyceridaemia/NAFLD in carriers of the dominant negative P467L-PPARγ mutation. Whether our findings are transferable to the overall FPLD3 spectrum of diseases is currently unknown. However, there are reports of FPLD3 patients, characterised by recurrent hypertriglyceridaemia despite treatment with fibrates. [20][21][22][23] Finally, our data also indicate that the specific repertoire of PPARs present under specific metabolic conditions is important, as it may modulate the fine balance between transcription factors with metabolically opposed functions.