PPARG F388L, a Transactivation-Deficient Mutant, in Familial Partial Lipodystrophy

Autosomal dominant familial partial lipodystrophy (FPLD) is characterized by adipose tissue repartitioning, typically with peripheral fat loss and central fat accumulation (1,2). In Dunnigan-type partial lipodystrophy (FPLD; MIM 151660), affected subjects have total loss of subcutaneous adipose tissue from extremities, with sparing of head, neck, and abdominal fat depots, and multiple metabolic disturbances including insulin resistance, diabetes, and dyslipidemia (1,2). Early atherosclerosis is seen in some FPLD kindreds (3). Mutations in LMNA (MIM 150330) encoding nuclear lamin A/C cause FPLD (4–6). However, there may be other genetic etiologies for partial lipodystrophy. For instance, a single patient with partial lipodystrophy and normal LMNA was heterozygous for a missense mutation, R425C, in PPARG (MIM 601487) encoding peroxisome proliferator-activated receptor (PPAR)-γ (7). Mutation of this gene would not be an unexpected cause of partial lipodystrophy, given the role of the gene product in adipocyte biology (8). However, there was no functional assessment of PPARG R425C and no demonstration of germline transmission (7). Heterozygotes for dysfunctional PPARG mutations had insulin resistance (9). We now report a family with three generations of subjects with partial lipodystrophy in whom the LMNA gene was normal and a dysfunctional mutation in PPARG cosegregated with the lipodystrophy phenotype.

The study kindred was ascertained by a clinical diagnosis of partial lipodystrophy through the Endocrinology Clinic, which had earlier diagnosed 22 Canadian FPLD patients with mutant LMNA (3,4). A clinical diagnosis of partial lipodystrophy was made in the proband (Fig. 1; III-2) and her daughter (Fig. 1; IV-1). The proband (III-2) was a 46-year-old woman of Anglo-Saxon descent who had been diagnosed with partial lipodystrophy based upon prominent muscularity of her calves and lower arms, marked atrophy of gluteal fat, and marked centripetal distribution of adipose tissue, with accumulation of subcutaneous facial, neck, suprascapular, and abdominal fat. The abnormal fat distribution had been present since adolescence. Clinically, the phenotype was similar to FPLD seen in the Canadian index probands with missense mutations in LMNA (3,4). However, differentiating features included the presence of some subcutaneous upper arm fat to the level of the surgical neck of the humerus, no phlebectasia, and less prominent muscularity of the arms and calves. She had type 2 diabetes since age 38 years. On metformin 1,500 mg daily and rosiglitazone 8 mg daily, she had good glycemic control, with glycosylated hemoglobin always <7.6%. She had been treated for hypertension for 8 years, recently with fosinopril 20 mg daily. She had a history of irregular menses and had bilateral polycystic ovarian disease, which was treated at age 44 years with bilateral salpingo-oophorectomy. She had hyperinsulinemia and type IV hyperlipoproteinemia and hypoalphalipoproteinemia, which was treated by diet alone.

The proband’s daughter (IV-1) was 22 years old when clinically diagnosed with FPLD based on prominent muscularity of her calves and lower arms due to paucity of subcutaneous fat on her arms and legs. Similar to her mother, she had atrophy of gluteal fat, with subcutaneous fat on her upper arms to the level of the surgical neck of the humerus, in addition to excess subcutaneous facial, neck, suprascapular, and abdominal fat since her teenage years. She had regular menses with no evidence of polycystic ovarian disease clinically. She had no medical problems and took no medications. She had hyperinsulinemia and mild type IV hyperlipoproteinemia, which was treated by diet alone.

The proband’s father (II-1) was 71 years old. He had prominent muscularity of calves and lower arms due to paucity of subcutaneous fat, atrophy of gluteal fat, with some subcutaneous fat on his upper arms to the level of the surgical neck of the humerus, in addition to excess subcutaneous facial, neck, suprascapular, and abdominal fat. He had treated hypertension of 20 years’ duration and type 2 diabetes of 18 years’ duration, treated with metformin 1,500 mg daily. Despite no smoking history, at age 56 he suffered an anterior wall myocardial infarction. He had hyperinsulinemia and hypoalphalipoproteinemia, and took atorvastatin for secondary prevention of coronary disease.

The proband’s youngest brother (III-4) was 39 years old, with prominent muscularity of calves and lower arms and accumulation of subcutaneous facial, neck, suprascapular, and abdominal fat. He was physically active, with no medical problems and took no medications. He had mild type IV hyperlipoproteinemia and hypoalphalipoproteinemia.

Clinically, there was no hepatosplenomegaly, phlebectasia, or acanthosis nigricans in any subject. No female subjects had hirsutism. The proband’s husband, aunt, son, and brother (Fig. 1 III-1, II-4, IV-1, and III-4, respectively) had normal anthropometry, with neither diabetes nor hypertension, and normal serum biochemistry. Abdominal ultrasound for subjects II-1 and III-2 showed no fatty liver. No subject had an elevated aspartate transaminase (Fig. 1). All subjects had resting blood pressure below 130/80 mmHg on duplicate determinations, but only subjects II-1 and III-2 were receiving antihypertensive medications.

Sequencing of genomic DNA showed that the proband was heterozygous for a novel T→A mutation at PPARG nucleotide 1164 in exon 5 that predicted substitution of the phenylalanine at codon 388 by leucine (F388L). The four affected family members were each heterozygous for the F388L mutation, and the four unaffected family members did not have the mutation (Fig. 1). No other mutations or single nucleotide polymorphisms (SNPs) in PPARG were detected. Younger subjects with the mutation had only the characteristic adipose repartitioning and hyperinsulinemia, while older subjects also had diabetes and hypertension. The maximal logarithm of odds (LOD) score for cosegregation of the mutation with affected status was 2.11 at 0% recombination. The F388L allele was absent from 520 normal Caucasian alleles (P < 3 × 10⁻⁷).

In the absence of ligand, the transcriptional activity of the mutant receptor was 3.1-fold lower than the wild-type receptor (Fig. 2A, light bars). However, both wild-type and mutant receptors displayed the same degree of transcriptional activity in the presence of a saturating amount of the synthetic PPAR-γ ligand rosiglitazone (Fig. 2A, dark bars). Rosiglitazone dose-response curves showed that F388L caused a fourfold rightward shift in the EC50 for PPAR-γ-mediated transcriptional activation (Fig. 2B), suggesting a reduced affinity for the ligand. Similar findings were observed when the transfections were carried out in the murine C2C12 myoblast cell line (data not shown). Western analysis showed no reduction in F388L expression (Fig. 2C). To determine if the F388L receptor had dominant-negative activity against wild-type PPAR-γ, a preliminary mixing experiment was performed in which wild-type and mutant receptors were cotransfected into NIH 3T3 cells. In the absence of ligand, the transcriptional activity of the combination of receptors (15 ng each) was slightly higher then 15 ng of wild-type receptor alone (Fig. 2D). These results suggest that the F388L mutation does not have dominant-negative activity against the wild-type receptor.

The mutation was within helix 8 of the predicted ligand-binding pocket of PPAR-γ, and the specific phenylalanine residue was conserved across all members of the PPAR gene family and also across all species (10). The mutation was absent from 520 normal alleles. In vitro studies showed that the PPAR-γ F388L mutation caused a reduction in the in vitro affinity of the receptor for its ligand rosiglitazone and reduced transcriptional activity in the absence of exogenous rosiglitazone. The low basal activity of the mutant receptor could have been caused either by reduced affinity for a subsaturating amount of the endogenous ligand or by a disrupted interaction with transcriptional cofactors that mediate basal transcriptional activity. It was perhaps of some interest that the proband had good glycemic control with pharmacologic doses of rosiglitazone, although this treatment was used in combination with metformin.

While the partial lipodystrophy phenotype in the PPARG F388L family was similar to the clinical phenotype of FPLD subjects with mutations in LMNA (3,4), there were some distinguishing features. For instance, subcutaneous fat was present on the upper arms to the surgical neck of the humerus, compared with absence of such fat depots in FPLD probands (3–5). Also, no affected subject had phlebectasia of forearm or leg veins, and the muscularity of arms and legs was less prominent than that seen in FPLD due to mutant LMNA (3–5). Acanthosis nigricans and hepatic steatosis were absent, but the proband’s father had early atherosclerosis and the proband had polycystic ovarian disease. Also, the biochemical disturbances, particularly dyslipidemia, were less severe than those reported in FPLD due to mutant LMNA (3). However, the increase in mean C-peptide concentration in PPARG F388L subjects compared with family members without the mutation was 2.8-fold (95% CI 1.7- to 4.1-fold), but the increase in subjects with LMNA mutations was only 1.6-fold (95% CI 1.2- to 2.1-fold) (3). This suggests a milder clinical and biochemical phenotype in partial lipodystrophy due to mutant PPARG compared with LMNA. The seemingly worse insulin resistance seems out of proportion to the degree of lipodystrophy, perhaps reflecting another effect of F388L. While unlikely, F388L could be in linkage disequilibrium with an unmeasured functional variant.

Earlier studies indicated that subjects with lipodystrophy had no germline PPARG mutations (11,12). A few rare mutations in PPARG have been reported. For instance, heterozygosity for PPARG R115Q was reported in nonlipodystrophic subjects ascertained based upon obesity and diabetes (13). Simple heterozygosity for either PPARG V290M or P467 l was also reported in nonlipodystrophic subjects ascertained based upon severe insulin resistance (9). However, recent magnetic resonance imaging found attenuation of some adipose tissue depots (14), and clinical reevaluation suggested paucity of subcutaneous limb fat and atrophy of buttock fat (15). Heterozygosity for PPARG R425C was found in another patient ascertained on partial lipodystrophy, but there was no demonstration of vertical germline transmission or in vitro functional impairment (7). Taken together, these findings indicate that mutation in PPARG can lead to partial lipodystrophy. More refined biochemical, physiological, and noninvasive imaging studies may help to further clarify phenotypic distinctions between partial lipodystrophy due to mutant PPARG and mutant LMNA.

How might heterozygosity for mutant PPARG lead to lipodystrophy? Studies of PPAR-γ null mice showed that embryonic stem cells lacking one copy of the PPARG gene exhibited impaired adipogenesis (16). Also, dominant-negative mutations in PPARG inhibit adipogenesis (17), although it appears that F388L may not be a dominant-negative mutation. Heterozygosity for the common PPARG P12A variant was associated with both protection (18) and susceptibility (19) to type 2 diabetes, through uncertain mechanisms. In addition, PPAR-γ null heterozygous mice had increased insulin sensitivity (20). Perhaps different missense mutations in PPARG could have specific functional consequences. The findings from the PPARG F388L kindred indicate that inherited partial lipodystrophy is clinically and genetically heterogeneous, extending the spectrum of metabolic phenotypes caused by PPARG mutations.