| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Function through Mutation of a Conserved Salt Bridge (R425C) in Familial Partial LipodystrophyDepartment of Metabolic and Endocrine Diseases (E.H.J., O.v.B., N.H., B.I.H.-S., R.B., E.K.), University Medical Center Utrecht, 3584 EA Utrecht, The Netherlands; and Bijvoet Center for Biomolecular Research (A.D.J.v.D., A.M.J.J.B.), Utrecht University, 3584 CH Utrecht, The Netherlands
Address all correspondence and requests for reprints to: Eric Kalkhoven, Department of Metabolic and Endocrine Diseases, University Medical Center Utrecht, Room KE.03.139.2, Lundlaan 6, 3584 EA Utrecht, The Netherlands. E-mail: e.kalkhoven{at}umcutrecht.nl.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
plays a key role in the regulation of glucose and lipid metabolism in adipocytes by regulating their differentiation, maintenance, and function. A heterozygous mutation in the PPARG gene, which changes an arginine residue at position 425 into a cysteine (R425C), has been reported in a patient with familial partial lipodystrophy subtype 3 (FPLD3). The strong conservation of arginine 425 among nuclear receptors that heterodimerize with retinoic acid X receptor prompted us to investigate the functional consequences of the R425C mutation on PPAR function. Here we show that this mutant displayed strongly reduced transcriptional activity compared with wild-type PPAR, irrespective of cell type, promoter context, or ligand, whereas transrepression of nuclear factor-
B activity remained largely intact. Our data indicate that the reduced transcriptional activity of PPAR R425C is not caused by impaired corepressor release, but due to reduced dimerization with retinoic acid X receptor 
in combination with reduced ligand binding and subsequent coactivator binding. As a consequence of these molecular defects, the R425C mutant was less effective in inducing adipocyte differentiation. PPAR R425C did not inhibit its wild-type counterpart in a dominant-negative manner, suggesting a haploinsufficiency mechanism in at least some FPLD3 patients. Using molecular dynamics simulations, substitution of R425 with cysteine is predicted to cause the formation of an alternative salt bridge. This structural change provides a likely explanation of how mutation of a single conserved residue in a patient with FPLD3 can disrupt the function of the adipogenic transcription factor PPAR on multiple levels. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Lipodystrophies represent a heterogeneous group of diseases characterized by altered amounts and/or distribution of body fat and major metabolic complications. The main forms can be classified according to their origin, either genetic or acquired, and subclassified according to the clinical pattern (3, 4). Whereas acquired lipodystrophies, like the metabolic syndrome and lipodystrophy due to highly active antiretroviral therapy (5) are fairly common, inherited forms of lipodystrophy are rare. Familial partial lipodystrophy (FPLD) is an autosomal dominantly inherited disorder, characterized by gradual loss of sc fat from the extremities and an accumulation of excess fat in the intraabdominal regions (3, 4). FPLD subtype 3 (FPLD3, MIM 604367) is associated with mutations in the peroxisome proliferator-activated receptor- (PPARG) gene (MIM 601487) (6, 7). The majority of FPLD3 patients have profound metabolic disorders, in particular severe insulin resistance and early onset diabetes mellitus, with polycystic ovarium syndrome seen in female subjects as a direct consequence of insulin resistance and marked or extreme dyslipidemia, characterized by high triglycerides and low high-density lipoprotein cholesterol. Some subjects also suffer from hypertension and hepatic steatosis (6, 7).
The PPARG gene encodes the PPAR protein, a member of the nuclear receptor superfamily of ligand inducible transcription factors (8, 9, 10). Differential promoter usage and alternative splicing of the PPAR gene generates four mRNA species (PPAR 1–4), but just two different receptor proteins, PPAR1 and PPAR2. The PPAR2 isoform contains 28 additional amino acids at its N terminus and its expression is restricted to adipose tissue, whereas PPAR1 is more widely expressed (e.g. in adipose tissue, lower intestine, and macrophages). PPAR plays a key role in glucose and lipid metabolism in adipocytes (8, 9, 10). Studies in murine cell lines have established that liganded PPAR is both essential and sufficient for adipogenesis (11), whereas PPAR knockout mice fail to develop adipose tissue (12, 13, 14). In addition, PPAR directly regulates the expression of a number of genes involved in net lipid partitioning into mature adipocytes. Besides PPAR, two closely related receptors have been identified, named PPAR and PPARß/
, which are encoded by different genes (10, 15). These three related receptors all bind to the PPAR-responsive elements (PPREs) in the promoter regions of target genes as obligate heterodimers with retinoic acid X receptors (RXRs) but exhibit different physiological roles due to their distinct expression patterns and specific activation by different ligands. Like other nuclear receptors, PPAR consists of distinct functional domains including a constitutively active transactivation domain (AF-1), at the N terminus, a centrally located highly conserved DNA binding domain (DBD) composed of two zinc finger motifs, and a C-terminal ligand binding domain (LBD) that contains a powerful ligand-dependent transactivation function (AF-2). Ligand binding stabilizes the active conformation of the PPAR LBD, thereby serving as a molecular switch between activation and repression functions of the receptor (16, 17). On some promoters, unliganded PPAR recruits corepressors like nuclear receptor corepressor and SMRT (silencing mediator for retinoid and thyroid-hormone receptors), which are part of multiprotein complexes containing histone deacetylase activity that repress gene transcription (18, 19, 20). Upon ligand binding, these corepressor complexes are released and replaced by coactivator complexes, including the steroid receptor coactivator (SRC) 1/cAMP response element binding protein-binding protein (CBP) and TRAP (thyroid hormone receptor-associated proteins)/DRIP (vitamin D receptor interacting proteins)/ARC (activator-recruited cofactor) complexes, that are involved in transcriptional activation (21, 22). The endogenous ligands for PPAR are not firmly established although some natural compounds, like polyunsaturated fatty acids and prostaglandin J2 derivatives (15-deoxy-
12,14-PGJ2) have been shown to be able to activate PPAR (23) (24). Synthetic PPAR agonists include the thiazolidinediones (TZDs) (25) and tyrosine-based agonists (26), which ameliorate insulin resistance and lower blood glucose levels in patients with type 2 diabetes (8). At least part of this response is thought to occur through indirect regulation of gene expression by PPAR, e.g. by transrepression of TNF-induced nuclear factor-B (NF-B) activity (27, 28).
Seven PPAR missense mutations have been described in patients suffering from FPLD3, which are located either in the DBD [C142R, C159Y, C190W (PPAR2 nomenclature)] or in the LBD [V318M, F388L, R425C, P495L (PPAR2 nomenclature)] (29, 30, 31, 32). In addition, two nonsense mutations (R385X, Y355X) and two frameshift/premature stop mutations have been reported (32, 33, 34). Except for R425C, all mutations have been shown to result in receptors with reduced transcriptional activity and some, but not all mutants inhibited their wild-type counterpart in a dominant-negative manner (see also Discussion). The functional consequences of the R425C FPLD3-associated missense mutation, which changes arginine 425 into cysteine, are currently unknown. Interestingly, a structure-based sequence alignment of nuclear receptors revealed that this arginine residue is part of a structural signature, which defines nuclear receptors that form heterodimers with RXR, including the thyroid hormone receptor ß (TRß) (35). Within this subclass of proteins, named class II receptors, R425 is involved in the formation of a conserved salt bridge with a negatively charged residue in helix 4/5 (E352 in PPAR2). In contrast, the homodimeric class I receptors lack this conserved arginine residue and the internal salt bridge (35).
The strong conservation of arginine 425 in PPAR2 among class II nuclear receptors together with the lack of functional data on the natural R425C mutation motivated us to investigate the functional consequences of this FPLD3-associated mutant. Here we combine detailed molecular analysis with structural modeling, using molecular dynamics simulations, to show that the R425C mutation found in a patient with FPLD3 causes aberrant salt bridge formation and thereby abrogates PPAR transcriptional activity leading to an inhibition of adipocyte differentiation.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
2 R425C Displays Reduced Transcriptional Activity2 to activate transcription was tested in reporter assays. For this, the human osteosarcoma cell line U2OS was used, which lacks endogenous PPAR expression and displayed the most significant reporter activation by liganded PPAR among several human cell lines tested. Cells were transfected with expression vectors encoding wild-type PPAR2 or R425C mutant together with a reporter construct containing three copies of the PPRE found in the rat acyl coenzyme A promoter (36). Whereas wild-type PPAR2 readily activated this 3xPPRE-tk-Luc reporter approximately 120-fold upon addition of the synthetic ligand rosiglitazone (1 µM), PPAR2 R425C only activated this reporter 20-fold (Fig. 1A
). Figure 1B shows that PPAR2 R425C was functionally defective at all concentrations of rosiglitazone tested, including saturating amounts. Cotransfection with RXR did not increase the transcriptional activity of either wild-type or mutant PPAR, indicating that RXR is not limiting under these experimental conditions (data not shown). Similar results were obtained in 293T cells, indicating that the reduced activity of the R425C mutant is not cell type specific (data not shown).
|
can be activated by various ligands (23, 24, 25, 26), reporter assays were also performed using several other synthetic and one natural PPAR ligand. All the TZDs tested (rosiglitazone, pioglitazone, troglitazone, and ciglitazone) and the natural ligand 15-deoxy-12,14-PGJ2 showed similar differences in transcriptional activity between wild-type and mutant PPAR2 (Fig. 1C
). Interestingly, the impaired activity of the R425C mutant was partially restored by the tyrosine-based agonist GW1929, a strong agonist that is also capable of activating the FPLD3-associated V318M and P495L mutants of PPAR2 (37).
The reporter construct used above contains three copies of a functional PPRE. To test the effect of PPAR2 R425C on a more natural promoter, reporter assays were performed using a luciferase reporter construct containing the promoter of the human adipose aquaporin gene (AQP7), which contains a single functional PPRE (38). As shown in Fig. 1D, activation of this reporter by wild-type PPAR2 was less pronounced (6- and 12-fold in the absence and presence of rosiglitazone, respectively), compared with the 3xPPRE reporter (Fig. 1A). However, mutation of R425 into cysteine reduced the ability of PPAR2 to activate this natural promoter in a similar fashion (Fig. 1D).
Amino acid sequence alignment of PPARs revealed that the arginine at position 425 in PPAR2 lies within a region that is highly conserved between species and different PPAR family members (Fig. 2A). To investigate the effect of the R425C mutation on other PPAR isoforms, analogous mutations were generated in PPAR1 (R397C), PPAR (R388C), and PPARß/ (R361C) and the proteins were subsequently tested for their ability to activate a 3xPPRE reporter construct in U2OS cells. Wild-type PPAR1 activated the 3xPPRE-tk-Luc reporter approximately 70-fold upon addition of rosiglitazone, whereas the R397C mutant only activated this reporter 15-fold (Fig. 2B, left panel). As reported earlier (39), wild-type PPAR displayed significant activity in the absence of ligand, which was further increased by addition of the synthetic PPAR-specific ligand WY14643. Mutation of PPAR R388 into cysteine almost completely inhibited transcriptional activity, both in the absence and presence of ligand (Fig. 2B, middle panel). Finally, assays performed with PPARß/ showed that this receptor also activated the reporter upon addition of a specific PPARß/ ligand (GW501516), albeit significantly less than the PPAR isoforms or PPAR. Nevertheless, the weak transcriptional activity of PPARß/ was clearly inhibited by mutating R361 into cysteine (Fig. 2B, right panel).
|
2, independent of ligand type, ligand concentration and promoter used. Furthermore, the reduced transcriptional activity of R425C is not limited to the PPAR2 protein because similar effects were observed when analogous mutations were generated in the PPAR1 isoform as well as in the related PPAR and PPARß/ proteins.
Corepressor Binding and Release of PPAR2 R425C Is Impaired
Mutations in the thyroid hormone receptor ß (TRß) are found in patients with resistance to thyroid hormone (RTH) (40). One RTH-associated mutation changes an arginine at position 383 into a histidine (R383H), resulting in impaired release of the corepressor SMRT (41). Because R383 in TRß is analogous to R425 in PPAR2 (Fig. 3A), the effect of the R425C mutation in PPAR2 on binding and release of SMRT was investigated. For this, bacterially expressed and purified glutathione-S-transferase (GST) or GST-SMRT fusion proteins were incubated with in vitro-translated [35S]methionine-labeled PPAR2 proteins (wild type or R425C) in the absence or presence of rosiglitazone. As shown in Fig. 3B, wild-type PPAR2 interacted with SMRT in the absence of ligand, and this interaction was reduced upon addition of rosiglitazone (Fig. 3B). The PPAR2 R425C mutant interacted with SMRT less efficiently, both in the absence and presence of ligand and did not show an obvious impairment of SMRT release in response to ligand (Fig. 3B).
|
LBD with the activation domain of viral protein 16 of herpes simplex virus (VP16). In the absence of ligand, wild-type PPARLBD clearly interacted with the SMRT-ID1, as demonstrated by the activation of the 5xGal4-TK-Luc reporter (Fig. 3C). Upon addition of rosiglitazone, this interaction decreased in a dose-dependent manner. In contrast, PPARLBD R425C showed limited interaction with SMRT-ID1, either in the absence or presence of ligand. In contrast, the FPLD3-associated P495L mutant of PPAR2, which was previously shown to exhibit impaired corepressor release (37), showed a comparable level of transcriptional activity at all concentrations of rosiglitazone. The negative controls, in which VP16 protein was coexpressed with either Gal4DBD or Gal4DBD-SMRT-ID1, displayed negligible luciferase activity.
Thus, our data indicate that the R425C mutation in PPAR2 does not impair corepressor release, but rather reduces the affinity for the corepressor. We conclude, therefore, that in contrast to the natural R383H mutant of TRß, the reduction in PPAR-mediated transcription caused by the R425C mutation is unlikely to be caused by impaired corepressor release.
The PPAR2 R425C Mutation Results in Reduced Heterodimerization with RXR
The crystal structure of PPAR-LBD reveals 13 -helices and a small four-stranded ß-sheet (16, 17). The arginine at position 425 lies within a highly conserved sequence that is located in a loop between helix 8 and 9, and some of the residues surrounding arginine 425 (G423, D424, and P426) are involved in heterodimerization with RXR. This prompted us to investigate whether the PPAR R425C mutation affected dimerization with RXR. First, we studied the effect of this mutation and mutations of the other highly conserved loop residues on in vitro heterodimerization by performing GST-pull down assays. For this, in vitro-translated [35S]methionine-labeled PPAR2 proteins (wild type or mutants) were incubated with bacterially expressed and purified GST-RXR. Whereas wild-type PPAR2 efficiently interacted with RXR, the R425C mutant protein displayed reduced binding (Fig. 4A). Mutation of arginine 425 into alanine (R425A) or lysine (R425K) also decreased the PPAR2-RXR interaction. PPAR2 D424A showed a slightly reduced binding compared with wild-type PPAR2, whereas mutation of glycine 423 or proline 426 into alanine (G423A and P426A, respectively) did not have an appreciable effect on in vitro dimerization with RXR. As a control, a PPAR2 mutant was generated (L464R), analogous to the homodimerization-defective mouse ER mutant (L511R) (42). This L464R mutant completely failed to dimerize with RXR in these in vitro assays (Fig. 4A).
|
2 were studied. Like the R425C mutant (Fig. 1), the R425A and R425K mutants displayed strongly reduced ligand-dependent activation of a 3xPPRE-tk-Luc reporter in U2OS cells (Fig. 4B). The D424A mutant showed slightly reduced transcriptional activity, whereas the activity of the G423A and P426A mutants was similar to wild-type PPAR2. Transcriptional activity of the heterodimerization-defective mutant L464R was completely abolished. The expression of all PPAR proteins was confirmed by Western blot analysis (Fig. 4B, lower panel). In general, the observed differences in transcriptional levels between the PPAR2 mutants reflected the differential effects on heterodimerization of these mutants (Fig. 4A), confirming that heterodimerization is a prerequisite for the transcriptional activity of the PPAR2 protein. In addition, mutation of the positively charged arginine at position 425 into a cysteine (R425C), into the uncharged residue alanine (R425A) or into a lysine residue (R425K), which is also positively charged, all affected the in vitro RXR heterodimerization capacity as well as the transcriptional activity of PPAR2. These findings point to the specific requirement of the arginine residue at position 425 for optimal function of PPAR2 as a transcription factor.
Because PPAR can only bind to PPRE sequences in the DNA upon heterodimerization with RXR, we investigated the effect of the R425C mutant on DNA binding. EMSAs were performed in which in vitro-translated RXR and PPAR2 proteins (either wild type or mutant) were incubated with a [32P]-labeled PPRE sequence. A specific PPAR2-RXR heterodimeric complex was formed on the PPRE because formation of this protein-DNA complex could be diminished by addition of an excess of unlabeled wild-type PPRE, but not by an excess of mutant PPRE (Fig. 4C). The R425C mutant protein also displayed specific DNA binding in the presence of RXR, albeit 2.3-fold less than wild type, as determined by densitometric analysis. A similar difference in binding capacity was observed with increasing amounts of PPAR2 protein (wild type or R425C) (see supplemental Figure S1, which is published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org). Specific antibodies against PPAR and RXR supershifted both the wild-type and mutant protein-DNA complexes, confirming the heterodimeric composition of both complexes (see supplemental data Figure S2). The dimerization-defective L464R mutant failed to bind the PPRE in the presence of RXR (Fig. 4C). Taken together, these results indicate that heterodimerization with RXR and subsequent DNA binding of PPAR2 is significantly, but not completely impaired by the R425C mutation.
Ligand Binding and Subsequent Coactivator Binding of PPAR2 R425C Is Reduced
Because heterodimerization with RXR and subsequent DNA binding of PPAR2 R425C is not completely impaired, this suggests that additional molecular defects contribute to the impaired function of this FPLD3-associated mutant. To investigate this, we performed reporter assays in which PPAR driven transcription is independent of heterodimerization with RXR. For this, the mutations described above were introduced into a chimeric Gal4DBD-PPAR LBD receptor and expressed together with a Gal4 reporter gene in U2OS cells. Both the wild-type Gal4-PPARLBD receptor and the dimerization defective L464R mutant stimulated transcription from the reporter up to 350-fold in the presence of rosiglitazone, showing that transcriptional activation is indeed independent of heterodimerization with RXR in this assay (Fig. 5A). In contrast, the activity of the R425C mutant, as well as the R425A and R425K mutants, was strongly reduced. The G423A, D4242A and P426A mutants displayed levels of transcription similar to wild-type PPAR. As a control, Western blot analysis was performed and showed comparable amounts of GAL4DBD fusion proteins (Fig. 5A, lower panel). These results indicate that the transcriptional defect in the PPAR2 R425C protein is not only due to reduced heterodimerization with RXR.
|
2 fusion proteins (wild type or R425C) were incubated with either 10 nM or 100 nM of [3H]-labeled rosiglitazone. Whereas wild-type PPAR2 protein displayed specific binding of rosiglitazone, the mutant protein showed reduced binding capacity for the ligand at both concentrations tested (Fig. 5B). The amounts of the bacterially produced receptors were similar (Fig. 5B, inset), indicating that the lower binding capacity for rosiglitazone displayed by the mutant protein was not caused by differences in protein levels. Because a coactivator binding surface is generated in the LBD of PPAR upon ligand binding, we next examined whether the reduced affinity for rosiglitazone displayed by the R425C protein also affected interaction with coactivators. For this, bacterially expressed and purified GST-fusion proteins of the coactivators CBP and SRC1 were incubated together with in vitro-translated [35S]methionine-labeled PPAR2 wild-type or PPAR2 R425C proteins in the absence or presence of rosiglitazone. As observed before (43, 44, 45), wild-type PPAR2 bound to SRC1 in the presence of rosiglitazone in a dose-dependent manner (Fig. 5C), whereas the interaction between PPAR2 and CBP was less dependent on the presence of ligand (Fig. 5D). Although PPAR2 R425C also displayed specific binding to SRC1 and CBP under these conditions, these protein-protein interactions were significantly weaker compared with wild-type PPAR2.
In summary, these experiments indicate that, in addition to impaired heterodimerization (Fig. 4), the R425C mutation also abrogates ligand binding compared with the wild-type protein and as a consequence diminishes ligand-dependent interaction with transcriptional coactivators.
Transrepression of NF-B by PPAR R425C Remains Largely Intact and Is Independent of Heterodimerization with RXR
To examine the effect of the R425C mutation on PPAR-mediated repression of NF-B activity, we performed reporter assays in which U2OS cells were transfected with expression vectors encoding wild-type PPAR2 or mutant PPAR (R425C or L464R) or an empty vector together with a 2B-luciferase reporter. Upon addition of TNF, this reporter was activated up to 10-fold and cotransfection of increasing amounts of either wild-type PPAR2 or R425C both reduced this activation in a dose-dependent manner. Addition of rosiglitazone enhanced this inhibitory effect to some extent (Fig. 6A). Because the R425C mutant displayed reduced dimerization with RXR (Fig. 4), these results suggest that monomeric forms of the receptor may be sufficient for transrepression of NF-B activity, as was also shown for other nuclear receptors (46). To investigate this, we employed the PPAR2 L464R mutant, which displays a complete loss of dimerization activity (Fig. 4). A dose-dependent decrease in reporter activation by TNF was observed upon addition of increasing amounts of L464R mutant, corroborating the hypothesis that PPAR-mediated transrepression of NF-B is independent of heterodimerization with RXR.
|
and NF-B occurs through the RelA (p65) subunit of NF-B (47), we assessed this protein-protein interaction by incubating bacterially expressed GST-RelA fusion proteins with in vitro-translated PPAR2 proteins (either wild type, R425C, or L464R). As shown in Fig. 6B, all three PPAR2 proteins were able to bind to GST-RelA and this interaction was largely rosiglitazone independent.
In summary, although the FPLD3-associated R425C mutation results in reduced transcriptional activity of PPAR (Fig. 1), PPAR-mediated transrepression of NF-B activity remains largely intact, especially in the presence of ligand. Furthermore, our data suggest that the PPAR-mediated transrepression is largely independent of heterodimerization with RXR.
Molecular Dynamics Simulation Predicts an Alternative Salt Bridge in PPAR2 R425C
Because the arginine residue on position 425 in PPAR2 is involved in the formation of an internal salt bridge (16), the effects of substituting this amino acid by cysteine was examined using several 5-nsec molecular dynamics simulations of wild-type protein and R425A and R425C mutants starting from various crystal structures of the LBD domain of PPAR (see Materials and Methods). Positional root mean square deviation from the starting structures as well as the various energy terms reached equilibrium in a few nsec for all the simulations (see supplemental data Figures S3 and S4). When analyzing the trajectories resulting from the different simulations, there was a clear and consistent difference between wild-type PPAR2 and PPAR2 R425C with respect to salt bridges that are present. In the wild-type protein, R425 (located in the loop between helix 8 and 9) forms a salt bridge with E352 (in helix 4/5), and R471 (in helix 10) forms a salt bridge with D424 (Fig. 7A, left panel). In the R425C mutant, however, the salt bridge between R425 and E352 obviously cannot be formed, and this resulted, unexpectedly, in a switch from the R471-D424 salt bridge to a salt bridge between R471 and E352 (Fig. 7A, right panel, and supplemental data Figure S5). A similar change was observed when R425 was mutated into alanine (data not shown). We also compared various structural parameters (root mean square deviations, angles between helices, solvent accessibility of surface residues). Although in some simulations helix 10 seemed to be deformed in the R425C mutant, we could not find consistent differences between wild-type and mutant simulations. The difference in salt bridge pattern was consistently found between on the one hand the three wild-type simulations and on the other hand four mutant simulations (both R425A and R425C). The fact that the same rearrangement is observed in independent simulations starting from various crystal structures, make the observed structural difference between wild-type and mutant protein highly significant.
|
2 (Fig. 7B). Western blot analysis was performed as a control and showed comparable expression levels for all PPAR proteins (Fig. 7B, lower panel). These findings suggest that the internal salt bridges between R425 and E352 and between D424 and R471 play an important role in PPAR2-mediated transcription by maintaining the structural integrity of the LBD of this receptor.
PPAR2 R425C and PPAR1 R397C Do Not Display Dominant-Negative Behavior
All of the FPLD3-associated PPAR mutations are heterozygous and some of the resulting PPAR mutant proteins have been shown to inhibit their wild-type counterpart in a dominant-negative manner (29, 37). We therefore examined whether the R425C mutant of PPAR2 and the same mutant in the context of the PPAR1 protein (R397C) would inhibit the activity of the respective wild-type receptors. For this, reporter assays were performed, using a 1:1 ratio between wild-type and mutant receptors and different concentrations of ligand. As shown in Fig. 8A, PPAR1 R397C failed to display dominant-negative behavior independent of ligand concentration, whereas the FPLD3-associated PPAR1 P467L inhibited wild-type PPAR1 at low concentrations of ligand, as reported earlier (37). In addition, the artificial PPAR1 L468A/E471A mutant displayed a strong dominant-negative activity against its wild-type protein independent of ligand concentration, as shown earlier (48).
|
2 protein, the R-to-C mutant also lacked dominant-negative activity, whereas PPAR2 L496A/E499A was able to inhibit wild-type PPAR2 at all concentrations of rosiglitazone (Fig. 8B). Interestingly, whereas PPAR1 P467L displayed a dominant-negative effect over its wild-type counterpart (Fig. 8A), the same mutant in the context of PPAR2 (P495L) was not able to inhibit wild-type receptor in a dominant-negative manner (Fig. 8B). In addition, the PPAR1 P467L mutant was dominant-negative over wild-type PPAR2, but PPAR2 P495L did not display dominant-negative activity over wild-type PPAR1 (data not shown). Similar results were obtained in 293T cells, indicating that the observed effects were not restricted to U2OS cells (data not shown).
In summary, these data show that unlike the artificial PPAR double mutant, the FPLD3-associated PPAR R-to-C mutant does not exert dominant-negative activity over the wild-type receptor, whereas dominant-negative activity of the FPLD3-associated PPAR P-to-L mutant appears to be limited to the PPAR1 isoform under our experimental conditions.
The R425C Mutation Reduces the Ability of PPAR2 to Induce Adipocyte Differentiation
Having established by in vitro and cell-based assays that PPAR function is affected on multiple levels we investigated the effect of this mutation in vivo on adipocyte differentiation. NIH-3T3 fibroblasts were transduced with virus obtained from cells transfected with the parental retroviral pMSCV vector, pMSCV-PPAR2 or pMSCV-PPAR2 R425C and stable cell lines were selected. Approximately 20% of the NIH-PPAR2 cells differentiated into adipocytes, as assessed by staining of triglycerides with Oil-red-O (Fig. 9A). Virtually no adipocyte differentiation could be detected in NIH-vector or NIH-PPAR2 R425C fibroblasts under our experimental conditions. To independently confirm adipocyte differentiation, the mRNA expression of three PPAR target genes, fatty acid binding protein 4 (Fabp4), adipose aquaporin (Aqp7) and glycerol kinase (Gyk), was determined on the same cell lines by quantitative RT-PCR assays. Whereas a 34-fold increase in Fabp4 mRNA expression was observed for NIH-PPAR2 cells, mRNA expression of this gene in NIH-PPAR2 R425C cells was similar to NIH-vector cells (Fig. 9B). The PPAR2 R425C protein also failed to induce mRNA expression of Aqp7 and Gyk, both of which were significantly up-regulated by the wild-type receptor (Fig. 9B). As shown by Western blot analysis, the PPAR protein expression levels for wild-type and R425C proteins were similar (Fig. 9C).
|
2 protein to activate transiently transfected reporter genes (Fig. 1) but also reduces its potential to activate endogenous target genes (Fabp4, Aqp7, Gyk) and promote adipogenesis (Fig. 9). |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
plays a key role in the regulation of glucose and lipid metabolism in adipocytes by regulating their differentiation, maintenance, and function (9, 10). Compelling genetic evidence for this view comes from human FPLD3 patients, harboring heterozygous mutations in the PPARG gene, because they are characterized by aberrant fat distribution and metabolic disturbances, including insulin resistance and dyslipidemia (6, 7). In this report, we demonstrate that the FPLD3-associated PPAR2 R425C mutant displays multiple molecular defects. These defects probably do not involve impaired corepressor release, but comprise a combination of reduced heterodimerization with RXR, lower ligand binding affinity and subsequent reduced binding of coactivators, resulting in reduced transcriptional activity and eventually in an impaired ability of the R425C mutant to induce adipocyte differentiation. Interestingly, like the dimerization-defective L464R mutant, the R425C mutant could still transrepress TNF-induced NF-B activity. These results are reminiscent of the transrepression of NF-B activity by monomeric forms of other nuclear receptors, like the glucocorticoid receptor (46), and indicate that activation and transrepression critically depend on different regions of the receptor (49). Recently, a structure based sequence alignment revealed two sets of differentially conserved residues that divided the entire nuclear receptor superfamily in two classes related to their oligomeric behavior (35). Nuclear receptors that form homodimers belong to class I, whereas nuclear receptors that form heterodimers with RXR, like PPAR, belong to class II. A differentially conserved arginine (R425 in PPAR2) defines the signature of these class II receptors and is involved in the formation of a salt bridge with a negatively charged residue in H4/H5 (E352 in PPAR2) (16, 17). Moreover, the crystal structure of PPAR LBD revealed an additional internal salt bridge between two different conserved class II residues, D424 (L8/9) and R471 (H10). Our molecular dynamics simulations predicted that substitution of PPAR2 R425 with a cysteine residue would disrupt both internal salt bridges leading to the formation of an alternative salt bridge between R471 and E352. The conserved salt bridge residues are structurally important in defining the conformation of the LBD of PPAR2: both D424 and R471 are involved in dimerization with RXR (16), whereas R425 and E352 form hydrogen bonds with Y505 at the C terminus of helix 12 leading to the stabilization of H12, which is important for cofactor binding (39). In addition, destabilization of H12 could in turn affect heterodimerization because an interdimer salt bridge is formed between Y505 (H12) of PPAR2 and K431 of RXR (16). Finally, crystallographic studies on PPAR, together with a corepressor peptide, indicate that the two hydrophobic residues immediately preceding E352 in PPAR2 are probably part of the corepressor interaction surface (50). Therefore, conformational changes resulting from the simultaneous disruption of both internal salt bridges through the R425C mutation provide a likely explanation for the impaired interactions with cofactors and RXR, as well as the reduction in ligand binding, despite the fact that R425 is not in the direct vicinity of the ligand binding pocket. The partial restoration of PPAR2 R425C activity by the tyrosine-based agonist GW1929 might be explained by the benzophenone group in this type of ligand, which is not present in TZDs. This group can make additional hydrophobic interactions with residues in H3, H7, and H10 (16), which could contribute to the active conformation of the LBD structure and thereby partially restore the activity of PPAR2 R425C, as was observed specifically with this ligand.
Studies on four other mammalian class II nuclear receptors, vitamin D receptor (VDR), TRß, RXR and retinoic acid receptor (RAR), have pointed to the importance of the conserved salt bridge residues for their function. In keeping with our results on the E352A mutation in PPAR2, the analogous mutant in VDR (E269A) displayed reduced ligand binding and transcriptional activity (51). Unlike the PPAR2 R425C mutant, however, mutation of the analogous residue in TRß, as exemplified by the RTH-associated R383H mutant, did not affect activation of target genes, but resulted in impaired corepressor release (41). Furthermore, the R339A mutation reduced the transcriptional activity of RAR, but heterodimerization was unaffected (35), which is in contrast to the analogous PPAR2 mutant (R425A). Finally, the D384A mutant of RXR (D424A in PPAR2) displayed reduced ability to homodimerize, but heterodimerization with RAR was unaffected (52). These findings therefore clearly underscore the important role of the internal salt bridges in maintaining the structural integrity of the LBDs of class II receptors, but also show that the molecular defect(s) caused by their disruption can differ between individual nuclear receptors.
Although the PPAR2 R425C mutant displays some similarities with other FPLD3-associated PPAR mutants, the molecular defects are clearly not identical. Based on functional characterizations in combination with studies on dominant-negative activity of these different PPAR mutants, PPARG mutations can be divided in at least three different subgroups (29, 30, 31, 32, 33, 34, 37). The first group comprises the missense DNA binding mutants and LBD truncation mutants (C142R, C159Y, C190W, FS343X, R385X). These mutants have no DNA binding capacity but are still able to recruit coactivators and can therefore compete for common cofactors (squelching), which can explain their dominant-negative activity. The second group consists of missense mutations located in the LBD (V318M, P495L), which inhibit PPAR activity through an alternative mechanism. These mutants are transcriptionally inactive due to impaired ligand and coactivator recruitment but are able to compete for DNA binding with wild-type PPAR. As a consequence the activation of PPAR target genes is reduced, which could explain the dominant-negative activity of these mutants. The third group, represented by the PPAR R425C mutant, displays both impaired heterodimerization with RXR and subsequent DNA binding, as well as impaired ligand and cofactor binding, the combination of which could account for the absence of dominant-negative activity. In line with this, addition of an artificial mutation in TRß, which attenuated heterodimerization with RXR and subsequent DNA binding, abrogated the dominant-negative activity of three different natural LBD mutants (53, 54). This suggests that the combination of either intact DNA binding, together with impaired cofactor binding, or intact cofactor binding in combination with impaired DNA binding is a prerequisite for dominant-negative activity of TRß and PPAR mutants. Like the R425C mutant analyzed here, the FPLD3-associated PPAR2 F388L mutant (30) and the recently published nonsense mutation, which truncates PPAR2 after residue 355 (Y355X) (33), also lack dominant-negative behavior. Together with the FPLD3-associated PPAR4 promoter (-14A>G) mutation (55), which only reduces the expression of the PPAR protein, these findings indicate that in some FPLD3 patients PPARG mutations act through a haploinsufficiency mechanism rather than dominant-negative activity of the affected gene product.
When the FPLD3-associated PPAR P-to-L mutant was tested in the context of the PPAR1 protein (P467L), this protein inhibited wild-type PPAR1 and -2 in a dominant-negative fashion at low concentrations of rosiglitazone, as described earlier (37). Interestingly, the same amino acid substitution in the context of the PPAR2 protein (P495L) resulted in a protein that lacked this capacity, indicating that at least under our experimental conditions the dominant-negative activity of the PPAR P-to-L mutant is isoform specific. These findings suggest that the effect of dominant-negative PPAR mutants might be more pronounced in tissues with a high PPAR1/-2 ratio, but it is currently unknown whether this ratio differs between body fat depots. An important challenge for the future will be to establish how heterozygous PPARG mutations can lead to depot-specific effects on adipose tissue, as exemplified by the characteristic aberrant fat distribution in patients suffering from FPLD3 and mice expressing PPAR mutants (56, 57).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
12,14-PGJ2 were from Cayman Chemical Co. (Ann Arbor, MI). GW1929 and TNF were from Sigma-Aldrich (St. Louis, MO). [3H]-rosiglitazone was purchased from American Radiolabeled Chemicals, Inc. (St. Louis, MO). Fugene 6 transfection reagent was from Roche Applied Biosciences (Indianapolis, IN). Anti-PPAR (sc-7273), anti-RXR (sc-553), and anti-Gal4 (sc-510) antibodies were from Santa Cruz Biotechnologies (Santa Cruz, CA). Anti-tropomyosin antibody (T2780) was purchased from Sigma-Aldrich and Roche Applied Biosciences, respectively. Antirabbit-horseradish peroxidase (111035144) and antimouse-horseradish peroxidase (115035146) were from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA). Oil-Red-O was purchased from Sigma-Aldrich (O-0625). Crystal violet was from Chroma-Gesellschaft Schmid & Co. (Köngen, Germany).
Plasmids
All recombinant DNA work was performed according to standard procedures (58). The pCDNA3-PPAR1, pCDNA3-PPAR2 and pCDNA3-PPARß/ constructs were kind gifts from Dr. V. K. K. Chatterjee (Department of Medicine, University of Cambridge, Cambridge, UK) (48). pCDNA3.1-PPAR expression plasmid was a kind gift from Dr. S. Ali (Imperial College London, London, UK). The reporter containing the human aquaporin promoter (–681/+11) was a kind gift from Dr. N. Maeda (Osaka University, Osaka, Japan) (38). The reporter plasmid 2B-luciferase was a kind gift from Dr. E. Burstein (University of Michigan, Ann Arbor, MI) (59). The retroviral pMSCV-vector (CLONTECH, Palo Alto, CA) containing wild-type murine PPAR2 was a kind gift from Dr. B. M. Spiegelman (Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA) and Dr. R. G. Roeder (Rockefeller University, New York, NY) (60). The bacterial expression vector for GST-SMRT (61) and the 3xPPRE-tk-Luc reporter (23) were kind gifts from Dr. R. M. Evans (Howard Hughes Medical Institute and Gene Expression Laboratory, Salk Institute, La Jolla, CA). The bacterial expression vector for GST-SRC1 (amino acids 570–780) (62), GST-RelA (63) and the 5xGal4-AdMLTATA-Luc (64), 5xGal4-TK-Luc (64), and 3xPPRE-tk-Luc (23) reporters have been described earlier.
The pCDNA-Gal4DBD-PPAR LBD (amino acids 173–475) expression construct was generated by cloning a BamH1/Xba1 fragment from pSG424-PPARLBD (gift from Dr. V. K. K. Chatterjee (48) into the respective sites of pCDNA3-Gal4DBD (64). The expression vector containing the complete coding region of human RXR was a kind gift from Dr. J. D. Baxter (65). pCDNA3.1-RXR was generated by PCR amplification of RXR from this vector using primers containing XbaI and HindIII-sites and cloned into the respective sites of pCDNA3.1(-). PGEX2TK-GST-CBP was generated by cloning the BamHI/EcoRI fragment (amino acids 1–452) from pCDNA3.1-CBP (64) into the pGEX2TK vector. To generate a pCDNA3-Gal4DBD-SMRT-ID1 construct, the SMRT-ID1 fragment (amino acids 2302–2352) was amplified from the GST-SMRT construct (61) by PCR using primers containing BamH1and XbaI-sites, respectively, and cloned into the pCDNA3-Gal4DBD construct. PCDNA3-VP16 was constructed by amplification of the VP16 from a psG5-VP16 expression construct (66) by PCR using a forward primer containing a HindIII-site and a reverse primer containing a stopcodon and BamH1-site. The VP16 cDNA was inserted in the corresponding sites of pCDNA3. PCDNA3-VP16-PPARLBD was generated by PCR amplification of VP16 activation domain using primers containing HindIII and BamH1-sites, respectively, and insertion into the respective sites of pCDNA-Gal4DBD-PPARLBD, after removal of the Gal4DBD sequence. All mutations were generated by QuikChange mutagenesis (Stratagene, La Jolla, CA) and verified by sequencing.
Cell Culture, Transient Transfections, and Reporter Assays
The human osteosarcoma cell line U2OS and the human embryonic kidney 293T cell line (HEK293T) were maintained in DMEM Glutamax (Dulbecco) containing 10% fetal calf serum (Invitrogen, Carlsbad, CA), 100 µg of penicillin/ml and 100 µg streptomycin/ml (Invitrogen). U2OS cells used for NF-B-luciferase reporter assays were maintained in DMEM Glutamax (Dulbecco) containing 5% dextran-coated charcoal stripped fetal calf serum (67) (Invitrogen), 100 µg of penicillin/ml and 100 µg streptomycin/ml (Invitrogen). The murine NIH-3T3 cell line was cultured in the same media but now with 10% bovine serum (Invitrogen), 100 µg of penicillin/ml, and 100 µg streptomycin/ml (Invitrogen). For luciferase reporter assays, cells were seeded in 24-well plates and transiently transfected using the calcium-phosphate precipitation method. Each well was cotransfected with a reporter construct, PPAR expression constructs and 2 ng pCMV-Renilla (Promega, Madison, WI) or 12.5 ng TK-Renilla (Promega) as indicated in the figure legends. The next day, cells were washed twice with HBS (pH 7.05) and subsequently maintained in medium in absence or presence of PPAR ligands for 24 h. After incubation, cells were washed twice with PBS and harvested in passive lysisbuffer (Promega) and assayed for luciferase activity according to the manufacturer’s protocol (Promega Dual-Luciferase Reporter Assay System) and for Renilla luciferase activity to correct for transfection efficiency. The relative light units were measured by a Centro LB 960 luminometer (Berthold Technologies, Bad Wildbad, Germany).
For Western blot analyses of the different PPAR proteins, U2OS cells were transiently transfected with 2 µg PPAR expression construct using Fugene 6 transfection reagent and treated as described as above. Cells were lysed in Laemmli sample buffer and subjected to SDS-PAGE and transferred to Immobilon membranes (Millipore, Eugene, OR). -PPAR antibody was used to probe for PPAR protein and ECL (Amersham Biosciences, Arlington Heights, IL) was used for detection.
GST Pull-Downs
Recombinant PPAR2 cDNAs (wild type or mutants) in the pCDNA3 expression vector were transcribed and translated in vitro in reticulolysate in the presence of [35S]methionine according to manufacturer’s protocol (TNT T7 Quick Coupled Transcription/Translation Kit; Promega). Rosetta pLysS competent bacteria (Novagen, Madison, WI) were transformed with GST expression plasmids. GST fusion proteins were induced and purified as described earlier (62). [35S]-labeled proteins were incubated with GST fusion proteins in NETN buffer [20 mM Tris (pH 8.0), 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40) containing protease inhibitors (Complete; Roche Applied Biosciences). Samples were subsequently washed and subjected to SDS-PAGE. Gels were enhanced with Amplify (Amersham) and dried and the [35S]-labeled proteins were visualized by fluorography. For each assay, at least three independent pull downs were performed.
EMSA
Double-stranded DNA oligomers, containing the PPRE from the rat acyltransferase-coenzyme A oxidase promoter (36), were labeled with [-32P]deoxy-ATP using Klenow enzyme and purified using Probe Quant G-50 Micro Colums (Amersham Biosciences). In vitro-translated PPAR (wild type or mutants) and/or in vitro-translated RXR proteins were preincubated for 15 min at room temperature in absence or presence of 10-fold cold probe (wild type or mutant) in a buffer containing 10 µg BSA, 1 µg poly deoyinosine:deoxycytosine, 40 mM HEPES (pH 7.4), 100 mM KCl, 2 mM dithiothreitol and 20% glycerol. For supershift experiments 1 µg of -RXR or -PPAR or -Gal4 antibodies were added to the preincubation mix. After preincubation, purified [32P]-labeled probe was added and incubated for 30 min at room temperature, followed by 30 min at 4 C. Receptor-DNA complexes were separated from unbound DNA on nondenaturing polyacrylamide gels and visualized by autoradiography. At least three independent experiments were performed.
The complete probe sequences used for binding and competition analysis were as follows: PPRE-wild-type, 5'-CCGGGGACCAGGACAAAGGTCACGAAGCT-3' and PPRE-mutant, 5'-CCGGGGGACCAGCACAAAGCACACGAAGCT-3'.
Ligand-Binding Assay
Filter binding assays were performed as described by Adams et al. (68) with minor modifications. Bacterially expressed and purified GST-PPAR2 fusion proteins were incubated with 10 or 100 nM [3H]-rosiglitazone in binding buffer [50 mM HEPES (pH 7.9), 100
