Peroxisome proliferator‐activated receptor gamma‐ligand‐binding domain mutations associated with familial partial lipodystrophy type 3 disrupt human trophoblast fusion and fibroblast migration

Abstract The transcription factor peroxisome proliferator‐activated receptor gamma (PPARG) is essential for placental development, and alterations in its expression and/or activity are associated with human placental pathologies such as pre‐eclampsia or IUGR. However, the molecular regulation of PPARG in cytotrophoblast differentiation and in the underlying mesenchyme remains poorly understood. Our main goal was to study the impact of mutations in the ligand‐binding domain (LBD) of the PPARG gene on cytotrophoblast fusion (PPARGE352Q) and on fibroblast cell migration (PPARGR262G/PPARGL319X). Our results showed that, compared to cells with reconstituted PPARGWT, transfection with PPARGE352Q led to significantly lower PPARG activity and lower restoration of trophoblast fusion. Likewise, compared to PPARGWT fibroblasts, PPARGR262G/PPARGL319X fibroblasts demonstrated significantly inhibited cell migration. In conclusion, we report that single missense or nonsense mutations in the LBD of PPARG significantly inhibit cell fusion and migration processes.


| INTRODUC TI ON
Peroxisome proliferator-activated receptor gamma (PPARG) is a nuclear receptor involved in lipid metabolism, 1,2 insulin resistance 3 and inflammation. 3 Furthermore, experiments with knockout mice have demonstrated the importance of PPARG for placental formation and gestational outcome. 4,5 In humans, the placenta is composed of a mesenchymal core covered by mononucleated villous cytotropho- functions and provides immunological support to the fetus. In the human placenta, PPARG is highly expressed in trophoblasts and is directly involved in the differentiation of both villous and extravillous cytotrophoblasts. 6,7 Thus, abnormal PPARG expression and/ or activity is likely to result in placental dysfunction and pregnancy complications such as pre-eclampsia and intra-uterine growth retardation. [8][9][10] PPARG comprises a DNA-binding domain, an agonist-independent activation domain (AF-1) and an agonist-dependent activation domain (AF-2), which contains the ligand-binding domain (LBD).
PPARs heterodimerize with the retinoid X receptor (RXR)-α and activate the transcription of target genes by binding to the PPAR response element (PPRE). The transcriptional activity of PPARG is principally modulated by agonists, which recruit either coactivators or corepressors. In general, ligand-bound PPARG recruits coactivators, whereas ligand-free PPARG is bound to corepressors.
Six types of familial partial lipodystrophy (FPLD) have been recorded in the literature. [11][12][13][14][15] Of these, FPLD type 3 (FPLD3) is caused by rare autosomal dominant mutations in the PPARG gene. 16 18 In this study, we were interested in cell fusion and migration, two processes that are essential for placental development. First, we examined a mutation in PPARG, PPARG E352Q , which is known to be  Barroso et al. 1999 associated with FPLD3. This mutation was first described in a report of placental abnormalities that led to premature delivery of a baby who developed hydrops foetalis and died 24 hour after delivery. The infant's karyotype was normal (46 XX) but placental DNA analysis revealed the presence of the same heterozygous mutation in the PPARG-LBD as in the mother. 19 Here, we specifically investigated by a combination of knockdown and reconstitution approach, the effect of this mutation on in vitro differentiation of villous cytotrophoblast, that is formation of ST by a cell-cell fusion process. We also evaluated the impact on the migration of mesenchymal cell (fibroblast) isolated from two patients carrying mutations in the LBD of PPARG, the novel missense mutation PPARG R262G and the previously described nonsense mutation PPARG L339X , which has been linked with a defect in transrepression of cellular RAS and consequent cellular dysfunction. 20

| Ethical statement
The study was performed according to the principles of the All subjects gave their written informed consent for these studies, which were approved by an institutional review committee.

| siRNA and mammalian expression vectors
siRNA transfections were performed as described in 21 and NotI-mCherry_R 5'-GA(GCGGCCGC)TTACTTGTACAGCTCGT CCATGCC-3' and subcloned into a pCMV vector using the restriction sites KpnI and NotI. The fusion ss.mCherry protein was used as a transfection control for the nanoluciferase assay, when cells were transfected, ss.mCherry was secreted into the cell culture media and quantified in tandem with luminescence by an EnSpire Multimode plate reader (Perkin Elmer).

| Structural analysis
Visual inspection of the PPARG structure (PDB code 5YCP) was done using COOT 28 and Figures were prepared with PyMOL (http:// pymol.org/).

| Statistical analyses
All measurements were performed at least in three independent experiments. The data are expressed as the mean + SEM for nanoluciferase and migration assays, or mean + SD of the indicated number for Western blots and assays of the fusion index and PPRE-H2B-eGFP. Statistical comparison (paired t test) of each treated group versus control was performed using GraphPad Prism 6 software.

| E352Q mutation decreases PPARG activity
To evaluate the effect of the E352Q mutation on PPARG activity, we used the NIH/3T3 cell line, which does not express PPARG. We first validated by Western blots (Figure 2A) and immunofluorescence ( Figure 2B) the expression of all our PPARG WT and PPARG E352Q plasmids in transiently transfected cells. As expected, we observed in the Western blots, a single ~55 kD corresponding to either (pcDNA3) PPARG WT or E352Q , and a single ~89 kD band corresponding to either GFP-PPARG WT or E352Q , and no endogenous PPARG (~55 kD) in (pcDNA3) EMPTY and not transfected controls (Figure 2A). In the immunofluorescence images, a co-localization of GFP-PPARG WT or E352Q with the anti-PPARG antibody in the nuclei of the transfected cells is observed (yellow arrows; Figure 2B). For the PPARG activity assays, we next cotransfected cells with PPARG WT

| PPARG E352Q mutation decreases villous cytotrophoblast fusion
To investigate the effect of PPARG E352Q on the in vitro differentiation of human cytotrophoblast into STs, we performed cell fusion assays on trophoblasts that were depleted of endogenous PPARG by siRNA transfection. To rescue PPARG knockdown, mammalian expression vectors were introduced that encoded PPARG WT or PPARG E352Q fused to GFP and that were insensitive to functional siRNA (GFP-PPARG* WT, GFP-PPARG* E352Q ; Figure 3). After Similar results were obtained from cytotrophoblasts that overexpressed GFP-PPARG* E352Q compared to those that overexpressed PPARG* WT (20.1% vs 49.1%; P < .0001).

| PPARG R262G or PPARG L339X mutation decreases fibroblast migration
We next investigated the effect of two other PPARG-LBD mutations on the cell migration process. We performed wound-healing assays with skin fibroblasts obtained from three control patients (PPARG WT ) and two patients with FPLD3 (PPARG R262G or PPARG L319X mutations; Figure 4A-C and PPARG L319X fibroblasts spread into flattened shapes ( Figure 4D

| D ISCUSS I ON
The effects of mutations in PPARG on cell fusion and migration processes have not been widely studied in the literature. We focused our study on three mutations located in the PPARG LBD that presents dimerization defects and/or impaired ligand-and cofactor binding ( Figure 5). Here, we first investigated the effect of the E352Q mutation that was initially reported from a case of FPLD3 that resulted in foetal death. 19  E352 is located in helix H5, which is one of the structural elements constitutive of the PPARG ligand-binding pocket (LBP, Figure 5A).

E352 is involved in a network of hydrogen bonds and salt bridges
with R471 from helix H10 as well as D424 and R425 from loop L8-9 which are part of the dimerization surface with RXR ( Figure 5B).
The replacement of a negatively charged glutamic acid by a glutamine in a positively charged environment is likely to perturb the local molecular organization and alter the ligand-binding and heterodimerization functions of PPARG. Interestingly, treatment with 1 µmol/L of GW1929 appeared sufficient to overcome the effect of the mutation and partially restore PPARG activity ( Figure 2D).
Another effect of the E352Q mutation studied here was a reduction in VCT fusion into ST, which could be evidence of the involvement of PPARG in the human placental abnormalities that were Values are presented as mean + SD of the fluorescence intensity (n = 3, >100 nuclei). E, Schema of the co-transfection of secreted reporters (secNluc and ss.mCherry). F, PPARG activity was assayed using the Nano-Glo ® Luciferase system (Promega). For each condition (triplicate), 100 µL of culture media was analysed after 48 h of treatment and luminescence signal was normalized with the corresponding mCherry signal. Values are represented as mean + SEM (n = 3 in triplicate). Statistical analysis was performed using paired t test to compare to vehicle control; *P < .05, **P < .01, **P < .001, ****P < .0001 In conclusion, the evidence presented here clearly shows that a single missense or nonsense mutation in the LBD of PPARG significantly inhibits cell fusion and migration processes. F I G U R E 4 R262G and L339X mutations induce morphological changes that reduce fibroblast migration. Wild-type and PPARG-mutated fibroblasts were seeded at equal densities into a 96-well plate (n = 3-5 wells/condition), cultured to confluency, mechanically wounded by scratching and then monitored for 24 h using IncuCyte Zoom. A, Representative images of fibroblast migration at 24 h. The blue region denotes the area of the initial wound (light blue line) covered by advancing cells. B, Time course of wound closure expressed as RWD (%). Values are represented as mean ± SEM of the percentage. C, Comparison between wild-type and PPARG-mutated fibroblasts is shown by AUC analysis of the replicates. Values are represented as mean + SEM. Statistical analysis was performed with paired t test to compare wild-type controls. ****P < .0001. D, Representative images showing immunofluorescent staining for F-actin (red) in upper panels, and vimentin (VIM, green) and vinculin (VINC, red) in lower panels. Cell nuclei were also stained with DAPI (blue). Scale bar: 50 µm. Insets depict a cytoskeleton architecture model of wild-type (WT, Green) and PPARG-mutated (red) fibroblasts

CO N FLI C T O F I NTE R E S T
The authors report no conflict of interest.

AUTH O R CO NTR I B UTI O N S
HS acquired data, analysed and interpreted the results, and wrote the draft. SC, CG and WB acquired data and interpreted the results.
CVi, CVa and CP provided materials and contributed to discussion.
TF provided administrative and technical support, and revised the manuscript. SAD conceived and designed the work, analysed and interpreted the results, and revised and edited the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

R E FE R E N C E S
F I G U R E 5 Structural analysis of three PPARG-LBD mutations. A, L339X, B, E352Q and C, R262G. Overall structure of the LBD of PPARG bound to rosiglitazone (Rosi in magenta) and a short peptide derived from the transcriptional coactivator SRC-1 (CoA, yellow helix). α-helices are labelled from H1 at the N-terminus to H12 at the C-terminus of the domain. The mutated residues (R262, L339, E352) are displayed as red sticks and labelled. The portion shown in grey (H4 to H12) is missing in the L339X mutant