Diagnostic Challenge in PLIN1 -Associated Familial Partial Lipodystrophy

frameshift variants in PLIN1 encoding perilipin-1, a key protein for lipid droplet formation and triglyceride metabolism, have been implicated in familial partial lipodystrophy type 4 (FPLD4), a rare entity with only six families reported worldwide. The pathogenicity of other PLIN1 null variants identified in patients with diabetes and/or hyperinsulinemia was recently questioned because of the absence of lipodystrophy in these individuals and the elevated frequency of PLIN1 null variants in the general population. Objectives: To reevaluate the pathogenicity of PLIN1 frameshift variants owing to new data obtained in the largest series of patients with FPLD4. Methods: We performed histological and molecular studies for patients referred to our French National Reference Center for Rare Diseases of Insulin Secretion and Insulin Sensitivity for lipodystrophy and/or insulin resistance and carrying PLIN1 frameshift variants. Results: We identified two heterozygous PLIN1 frameshift variants segregating with the phenotype in nine patients from four unrelated families. The FPLD4 stereotypical signs included postpubertal partial lipoatrophy of variable severity, muscular hypertrophy, acromegaloid features, polycystic ovary syndrome and/or hirsutism, metaboliccomplications ( e.g., hypertriglyceridemia, liversteatosis, insulin resistance, diabetes), and disorganized subcutaneous fat lobules with fibrosis and macrophage infiltration. Conclusions: These data suggest that some FPLD4-associated PLIN1 variants are deleterious. Thus, the evidence for the pathogenicity of each variant ought to be carefully considered before genetic counseling, especially given the importance of an early diagnosis for optimal disease management.

Perilipin-1 is a structural lipid droplet protein that facilitates triglyceride storage or initiates lipolysis, depending on the hormonal stimuli. We have shown that several FPLD4-associated PLIN1 frameshift variants disrupt the ability of perilipin-1 to inhibit basal lipolysis in adipocytes (2,3,5). However, the pathogenicity of other heterozygous PLIN1 null variants, identified in patients referred for evaluation of maturity onset diabetes of the young, hyperinsulinemic hypoglycemia, or type 2 diabetes, has recently been questioned by Laver et al. (6).
Since proper interpretation of PLIN1 variants is crucial for genetic counseling, the aim of the present study was to determine the pathogenicity of PLIN1 null variants in the light of the new genotype-phenotype data obtained in the largest series of patients with FPLD4 reported to date.

Subjects and Methods
Sequencing of a panel of lipodystrophy genes (AGPAT2, AKT2,  BSCL2, CAV1, CIDEC, LIPE, LMNA, PLIN1, POLD1, PPARG, PTRF, and ZMPSTE24) was performed in 237 independent index cases, investigated in our French National Reference Network for Rare Diseases of Insulin Secretion and Insulin Sensitivity, for manifestations evocative of a lipodystrophic syndrome. Capture (SeqCap EZ enrichment protocol, Roche NimbleGen, Roche Sequencing, Pleasanton, CA) was followed by massively parallel sequencing on a MiSeq platform (Illumina Inc., San Diego, CA). The data were analyzed using the Sophia Genetics DDM pipeline ® . PLIN1 frameshift variants were confirmed by Sanger sequencing. Affected relatives were identified after familial investigations. Abdominal subcutaneous adipose tissue biopsy specimens were obtained from two patients. Western blot and histological analyses were performed as previously described (2). All the subjects provided written informed consent in accordance with the legal procedures for molecular and histological investigations and publication of photographs. The Comité de Protection des Personnes Ile-de-France 5 (Paris, France) approved the present study.  (2). Western blot of whole cell extracts was performed using antibodies directed against the N-terminal and C-terminal parts of wild-type perilipin-1, as previously described (2). The mutant isoform was recognized by the N-terminal antibody as an additional band (arrow) just above the 62-kD molecular weight (MW) marker, which was not detected by the C-terminal antibody. Tubulin (antibody T5168, Sigma-Aldrich, St. Quentin-Fallavier, France) was used as a loading control. (B) Consequences on perilipin-1 protein expression of FPLD4-associated PLIN1 frameshift variants studied in (A). (C) Histological and immunohistological analyses of abdominal subcutaneous adipose tissue from patients E-I1 and E-II-2 compared with controls. Adipose tissue samples from patients displayed disorganized fat lobules of heterogeneous size, with increased fibrosis [Sirius red (SR) staining, 16% to 35% of the total sample surface vs 0.2% to 3% in controls], increased vascularization (density of CD34 staining, 0.13% to 0.44% in adipose lobules from patients vs 0.001% to 0.004% in controls), and increased macrophage infiltration with crown-like structures (assessed by CD163 and CD68 staining, density per 10

Molecular diagnosis
A heterozygous PLIN1 frameshift variant was identified in four index cases and five affected relatives (Fig. 1A). Patients from family D and L carried the c.1191_ 1192del deletion, previously shown to be expressed as an abnormal p.(Val398Glyfs*166) elongated form of perilipin-1, leading to constitutive activation of basal lipolysis (2,5). In family E and C, we identified a 4bpduplication in exon 8 (c.1202_1205dup), leading to the synthesis of a p.(Pro403Argfs*164) mutant protein, whose expression in adipose tissue was confirmed by Western blot (Fig. 2). These variants were not found in public databases (Exome Aggregation Consortium, Genome Aggregation Database) and cosegregated with the disease within each family (Fig. 1A). In all patients, we did not find any other molecular defect in known lipodystrophy genes.

Disease phenotype
The clinical and biological features of the nine investigated family members carrying a PLIN1 variant recapitulated the FPLD4 cardinal signs (2) (i.e., lipoatrophy, muscular hypertrophy, facial acromegaloid features, insulin resistance-related ovarian dysfunction, and metabolic complications (e.g., hyperinsulinemia or insulin-resistant diabetes, hypertriglyceridemia, liver steatosis) ( Table 1). The index cases had initially been referred for ovarian hyperandrogenism with lipodystrophy (patient D-II3), suspicion of acromegaly (patients L and E-II2), or early-onset nonautoimmune diabetes (patient C-II1). In several individuals (patients D-I2, C-I2, C-II2), the disease manifestations were only detected after family studies.
Lipoatrophy had mainly affected the trunk, limbs, and femorogluteal regions and was associated with muscular hypertrophy predominantly in the calves (Fig. 1B). Lipoatrophy could be mild, especially in young patients (patients E-III1 and C-II2). Cervicofacial fat accumulation was observed in three patients (patients E-I1, C-I2, and C-II1). Low serum leptin levels and a decrease in the total fat mass, as assessed by dual energy X-ray absorptiometry, were consistent with lipoatrophy. Abdominal subcutaneous adipose tissue, studied in patients E-I1 and E-II2, displayed disorganized fat lobules of heterogeneous size, with macrophage infiltrates, increased fibrosis, and increased vascularization, consistent with previous findings (2) (Fig. 2). Seven patients showed a facial acromegaloid appearance with enlarged hands and feet. All investigated patients had presented with hyperinsulinemia or diabetes, frequently accompanied by acanthosis nigricans. Five of the seven women had polycystic ovary syndrome and/or hirsutism and oligomenorrhea. Hypertriglyceridemia was present in all investigated adult patients. No history of acute pancreatitis was reported. All examined patients had liver steatosis. Patient L-II1 had experienced major complications, including neuropathy, hypertension, and myocardial infarction with rhythm disturbances, which required the implantation of a cardioverter defibrillator.

Discussion
The increasing use of next generation sequencing in clinical practice highlights the need for accurate interpretation of variants. When large population exome data became available, the pathogenicity of several genes involved in Mendelian disorders was questioned (7). This issue was recently raised for PLIN1 by Laver et al. (6). Because of the importance of early genetic counseling for appropriate disease management (1), the clues arguing for and against a pathogenic effect of PLIN1 null variants should be considered carefully (Table 2).
First, the allele frequency of PLIN1 null variants in the general population, estimated at ;4.10 24 in public databases, is higher than the prevalence of all forms of FPLD, estimated at ;3.10 26 (8). Even if FPLD remains  The pathogenicity of PLIN1 null variants was also questioned because the patients reported by Laver et al. (6) did not have overt lipoatrophy. As acknowledged by the authors, it can be difficult to exclude the presence of lipodystrophy in young patients, and the lack of adipose tissue-focused examinations in large cohort studies hamper the recognition of subtle lipodystrophic morphotypes. In this regard, one study underlined that FPLD remains underdiagnosed and could affect .3% of patients investigated for metabolic syndrome (9). Accordingly, lipodystrophy was diagnosed a posteriori in several patients in the present study, owing to the familial investigations.  (5). The p.(Pro439Valfs*125) mutant, which is located in the vicinity of this binding domain, fails to inhibit basal lipolysis by an alternate mechanism (3). It would be interesting to determine the functional consequences of PLIN1 null variants identified in the general population or in those with maturity onset diabetes of the young.
At present, the classification of genetic variants follows the guidelines of the American College of Medicals Genetics and uses a five-class score (10). According to these criteria, FPLD4-associated PLIN1 frameshift variants that disrupt protein function should be classified as "pathogenic." Laver et al. (6) suggested that PLIN1 null variants should not be reported as causative of lipodystrophy. Our study provides a set of arguments supporting the pathogenicity of several PLIN1 frameshift variants. It is thus important to evaluate each of these variants carefully for genetic counseling. We would recommend detailed familial investigations, adipose tissue-focused examination, and careful follow-up of metabolic evolution in patients carrying such variants.