Solution Structures of PPARγ2/RXRα Complexes

PPARγ is a key regulator of glucose homeostasis and insulin sensitization. PPARγ must heterodimerize with its dimeric partner, the retinoid X receptor (RXR), to bind DNA and associated coactivators such as p160 family members or PGC-1α to regulate gene networks. To understand how coactivators are recognized by the functional heterodimer PPARγ/RXRα and to determine the topological organization of the complexes, we performed a structural study using small angle X-ray scattering of PPARγ/RXRα in complex with DNA from regulated gene and the TIF2 receptor interacting domain (RID). The solution structures reveal an asymmetry of the overall structure due to the crucial role of the DNA in positioning the heterodimer and indicate asymmetrical binding of TIF2 to the heterodimer.


Introduction
PPARγ, a member of the nuclear receptor family, is a key regulator of adipocyte differentiation and is involved in glucose homeostasis and insulin sensitization (reviewed in [1,2]). PPARγ together with the CCAAT/enhancer-binding proteins had been identified as key transcription factors of driving fat cell differentiation (reviewed in [3]). PPARγ is absolutely required for both white and brown fat cell development. Several PPARγ coregulators have also been shown to affect positively or negatively this differentiation (reviewed in [4]).
PPARγ is activated through the binding of diverse ligands including natural fatty acid derivatives and nonsteroidal drugs and is the target of therapeutically active antidiabetics such as rosiglitazone (reviewed in [5]). Furthermore, cdk-5 phosphorylation of PPARγ leads to deregulation of some genes involved in metabolism [6].
The actions of PPARγ are mediated by 2 isoforms that result from alternative splicing. PPARγ2 is 28 amino acids longer at the N-terminal end (Figure 1(a)) and is mainly expressed in adipocyte cells, while PPARγ1 is ubiquitously expressed. Interestingly, PPARγ2 is ten times more active in ligand-independent transcriptional activation than PPARγ1 [7,8].
PPARγ as the other nuclear receptors is a modular protein with a DNA binding domain (DBD) and a ligand binding domain (LBD). It is active as a heterodimer with the retinoid X nuclear receptor (RXR). The PPAR/RXR heterodimer recognizes specific DNA sequences called PPAR response elements (PPRE), composed of imperfect direct repeats separated by one nucleotide (DR1) and an extended 5 half-site hexanucleotide [8]. Both the nonsymmetrical DR1 and the 5 flanking sequence contribute to the selective binding of PPAR. PPAR binds the 5 hexanucleotide and the 5 flanking region, while RXR binds the 3 half-site. To be fully active, PPARg2/RXR must be associated with coactivators which include p300/CBP, p160 members, PGC-1α, as well as Med1 (reviewed in [9,10]). These coactivators are themselves highly regulated at the transcriptional and posttranscriptional levels.
Structurally, most of our current knowledge on the molecular basis of the mechanism of action of NRs is based on the X-ray and NMR structures of isolated DNA and ligand binding domains. with coactivator peptides have revealed a conserved binding mode of the coactivators to NRs. Agonist ligands trigger a conformational change of the NR LBDs allowing a leucinerich interacting motif (LXXLL motif) present in the coactivator sequence that forms an α-helix to interact through a charge clamp and hydrophobic interactions at the LBD surface [11][12][13]. Coactivators such as p160 or PGC-1α contain several LXXLL motifs (NR boxes) involved in specific binding to NRs and other transcription factors. For TIF2 (NCoA2, p160 member), 3 NR boxes are found in a central domain that interacts with NRs and is called the receptor interacting domain (RID) (Figure 1(a)). The third motif has been shown to preferentially bind to PPARγ [14].
Extensive studies on structure-function relationships have been done on PPARγ LBD monomers or dimers and have provided information on conformational change induced by various ligands. The structures of full-length PPARγ/RXR complexes have also been studied by X-ray crystallography [15] and by small-angle X-ray scattering (SAXS) [16,17], although the N-terminal domain (NTD) was not visible in the electron density map. Remaining questions concern the mode of recognition of coactivator domains by PPARγ/RXRα complexes. To understand how TIF2 RID is recognized by PPARγ/RXRα and the topological organization of the complex, we performed a structural analysis in solution by SAXS of PPARγ/RXRα LBD in complexes with TIF2 RID and of PPARγ2/RXRα full-length or depleted of their N-terminal domain (ΔNTD) bound to PPRE and TIF2 RID protein.

Ultracentrifugation Equilibrium Sedimentation.
For sedimentation equilibrium experiments, samples were spun at 12,000 rpm and systems were first allowed to equilibrate for 12 hours before absorbance profiles were compared at different times to ensure that system had reached equilibrium. Using nonlinear least-squares analysis, these datasets were fitted using single component model and several equilibrium models with Sedphat program.

Electrospray Ionization Mass Spectrometry.
Prior to ESI-MS analysis, samples were desalted on Zeba Spin desalting columns (Pierce) in 150 mM ammonium acetate (pH 8.0). ESI-MS measurements were performed on an electrospray time-of-flight mass spectrometer (MicrOTOF, Bruker Daltonic, Germany). Purity and homogeneity of the proteins were verified by mass spectrometry in denaturing conditions (samples were diluted at 2 pmol/μL in a 1 : 1 water-acetonitrile mixture (v/v) acidified with 1% formic acid). The mass measurements of the noncovalent complexes were performed in ammonium acetate (200 mM; pH 8.0). Samples were diluted to 8 pmol/mL in the previous buffer and continuously infused into the ESI ion source at a flow rate of 3 mL/min through a Harvard syringe pump (Harvard Apparatus model 11). A careful optimization of the interface parameters was performed to obtain the best sensitivity and spectrum quality without affecting the noncovalent complexes stability. In particular, the capillary exit (CE) ranged from 60 to 150 V with a vacuum interface pressure of 2.3 mbar and was set to 80 V.

SAXS Experiments and Data
Processing. Synchrotron X-ray solution scattering data were collected at the X33 beamline (DESY, Hamburg) [19] using a PILATUS detector at a sample-detector distance of 2.7 m, covering the range of momentum transfer 0.01 < q < 0.6Å −1 (q = 4π sin(θ)/λ, where 2θ is the scattering angle and λ = 0.15 nm is the Xray wavelength) in eight frames (15 seconds each) to check for possible radiation damage. All scattering measurements were carried out at 10 • C with automated filling, and samples were measured at several concentrations.
SAXS experiments were also conducted on the SWING beamline at SOLEIL Synchrotron (Gif-sur-Yvette, France), using a 17×17 cm 2 low-noise Aviex CCD detector positioned at a distance of 2.107 m from the sample. Sample solutions were circulated in a thermostated Quartz capillary with a diameter of 1.5 mm and 10 μm wall thickness, positioned within a vacuum chamber. Fifty frames of 2 s each were collected, normalized to the transmitted intensity, and subsequently averaged using the image analysis software Foxtrot (SWING beamline at SOLEIL Synchrotron).
The SAXS data were averaged and processed by standard procedures using PRIMUS [20]. The forward scattering I(0) and the radii of gyration R g were evaluated using the Guinier approximation assuming that at very small angles (s < 1.3/R g ) the intensity is represented as I(s) = I(0) exp(−(sR g ) 2 /3). These parameters were also computed from the entire scattering pattern using the indirect transform package GNOM [21], which also provides the maximum dimension of the particle D max and the distance distribution function P(r). Low resolution shape analysis of the solutes was performed using the ab initio program DAMMIF [22]. The scattering from the atomic models was calculated using the program CRYSOL [23] which either predicts theoretical scattering patterns or fits the experimental data by adjusting the excluded volume and the contrast of the hydration layer. The program SASREF [24] was employed for molecular rigid body modeling of the PPARγΔNTD/RXRαΔNTD/DNA complex, based on the crystal structure of PPARγ/RXR (PDB ID: 3DZY). For both ab initio and rigid body analysis, multiple runs were performed to verify the stability of the solution, and the most typical reconstructions were selected using the programs DAMAVER [25] and SUPCOMB [26].

Characterization of PPARγ/RXRα
Complexes. PPARγ/ RXRα LBDs or full-length proteins were copurified in two steps as described [16,18]. DNA from Cyp4A1 regulated gene was added in a 1.2-fold excess to the full-length heterodimer and the complex further purified by analytical gel-filtration. In order to investigate the binding of TIF2 RID on PPARγ/RXRα, a 15 kDa fragment of TIF2 (632-772) containing the three LXXLL motifs and interacting with NRs ( Figure 1(a)) was added to the heterodimer and further purified by gel-filtration. After gel-filtration, the complexes were analyzed by SDS-PAGE and native electrophoresis confirming the formation and homogeneity of the complexes (see Supplementary Figure 1 available online at doi:10.1155/2012/701412). The stoichiometry of the TIF2 RID/PPARγ/RXRα complexes was determined using analytical ultracentrifugation (AUC) and electrospray ionization mass spectrometry (ESI-MS), a method that has proven its efficiency for the study of noncovalent complexes [27,28]. ESI-MS experiments were performed under nodenaturing conditions on PPARγ/RXRα LBDs in the presence of TIF2 RID. The molecular mass obtained (Figure 1(b)) is consistent with a complex composed of one TIF2 RID molecule per heterodimer. The presence of mass peaks corresponding to free PPARγ/RXRα LBDs is due to partial dissociation of the complex in the ESI ion source. Equilibrium sedimentation AUC experiments were also carried out to characterize the association state of TIF2 RID with PPARγΔNTD/ RXRαΔNTD/DNA in solution as well as its molecular mass (Figure 1(c)). The molecular mass of Mw of 114 kDa is in excellent agreement with the expected value of 114.3 kDa.
Both measurements indicate that only one TIF2 RID molecule binds to PPARγ/RXRα heterodimer.

Topology of PPARγ/RXRα LBDs and TIF2 RID/PPARγ/ RXRα LBDs Complexes.
The observed stoichiometry of one TIF2 RID per heterodimer contrasts with the crystallographic structures of NRs bound to LXXLL peptides [11][12][13] which have revealed 2 peptides bound by each monomer of the dimer. This raises the question of the binding mode of TIF2 RID to PPARγ/RXRα either asymmetrically to one monomer or to the two subunits involving two NR boxes.
To address this question we used SAXS, a structural method to characterize multidomain proteins and complexes. We analyzed by SAXS the recruitment of the TIF2 RID by the PPARγ/RXRα LBDs. As a reference, scattering profiles of PPARγ/RXRα LBDs alone were collected (Figure 2(a)). Monodisperse concentrated solutions of PPARγ/RXRα LBDs were measured, and the structural parameters including the radius of gyration (R g ) and the maximum particle dimension (D max ) were computed from the experimental scattering patterns from the heterodimer ( Table 1). A symmetrical pair-distribution function is observed (Figure 2(b)) indicating a globular complex. The experimental SAXS data is well fitted by the scattering profile calculated from the crystal structure (PDB ID: 3H0A) using CRYSOL [23] and is in agreement with SAXS parameters measured for heterodimer LBDs [17,18].
We next analyzed the TIF2 RID/PPARγ/RXRα LBDs complex. The values of R g and D max calculated from the distance pair-distribution function P(r) for TIF2 RID are   (Figure 2(c)) shows a continuously increasing curve that indicates an unfolded protein. For the TIF2 RID/ PPARγ/RXRα LBDs complex, it corresponds to a partially folded protein with a bell-shaped curve in agreement with a folded macromolecule and thus suggesting an induced folding of the molecule upon complex formation. Binding of TIF2 RID to the heterodimer led to an increase of the R g (by 8Å) and of the D max (by 65Å) ( Table 1). The large difference in R g and D max compared to those of PPARγ/RXRα LBDs suggests that the TIF2 RID is extended and partially disordered. This is also observed in the asymmetric profile of the pair-distribution function with a tail at high r values (Figure 2(b)) which corresponds to elongated molecules in contrast to the symmetric pair-distribution function of the globular PPAR/RXR LBDs. The molecular envelopes as bead models, derived from the SAXS data, were computed for PPARγ/RXRα LBDs and TIF2 RID/PPARγ/RXRα LBDs (Figure 3). It corresponds to a symmetrical globular shape for the LBD dimer (Figure 3(c)). In contrast, for the TIF2 RID complex, a marked asymmetry as compared to the LBD dimer is observed (Figure 3(d)). In analogy with the TIF2 RID and SRC-1 RID complexes with RAR/RXR and RAR homodimer [16,29], we can speculate that the globular region of the molecular envelope corresponds to the LBD dimer and the asymmetric extended tail to TIF2 RID. The large difference in the structural parameters of the TIF2 RID complex compared to the LBD dimer and the molecular envelope of the complex suggests that TIF2 RID is flexibly attached asymmetrically to the PPARγ/RXRα through only one LXXLL motif. Importantly, the data preclude a model in which two LXXLL motifs bind to each subunit of the PPARγ/RXR heterodimer. A preliminary SAXS study of PPARγ/RXRα bound to PGC-1α RID provides similar results with similar R g and D max values suggesting that in both cases the architecture of the two different complexes is similar with the CoA RID asymmetrically bound to PPAR subunit through only one LXXLL motif. Interestingly, it has been shown that NR2 motif of PGC-1α is sufficient to have a full transcriptional response by PPARγ/RXRα [14]. While for TIF2, it has been shown that the third LXXLL motif binds preferentially to PPARγ [14] and a combination of LXXLL motifs is required for a full transcriptional response [30]. However the SAXS data are in agreement with the crystal structure of PPARγ/RXRα LBDs (PDB ID: 3H0A) that reveals one coactivator peptide tightly bound to PPARγ and one coactivator peptide loosely bound to RXRα as indicated by the poor electron density and high B factor for the peptide bound to RXR (averaged B factor for the peptide bound to RXR is 116.2 compared to the averaged B factor for the CoA peptide bound to PPARγ, 31.6). These observations demonstrate that TIF2 RID binds asymmetrically to PPARγ subunit within the heterodimer through one LXXLL.

Solution Structure of PPARγ2/RXRα/Cyp4A1 DR1 Complexes.
PPARγ/RXRα without DNA has been shown to be elongated with no interdomain interactions [17]. In complex with idealized DNA, the first atomic resolution structure of integral PPARγ/RXRα obtained by X-ray crystallography [15] confirmed the structural information concerning isolated domains [11,31,32] and also revealed an interdomain contact between the LBD of PPARγ and the DBD of RXR [13], although the functional correlation is limited to a single point mutation. The N-terminal domain, unfolded, was not visible in the electron density map. We previously measured the SAXS profile of PPARγ2/RXRα bound to an idealized DR1 [16] and have shown that the conformation in solution is different to that observed in the crystal structure with the heterodimer exhibiting an extended asymmetric shape without additional interdomain contacts between the DBDs and LBDs beyond the connection through the hinge regions. We have now characterized the solution structure of PPARγ2/RXRα full length or truncated of their NTDs and bound to a natural PPRE from CYP4A1. The structural parameters calculated from the experimental scattering patterns (Figures 4(a) and 4(b)) given in Table 1 reveal an extended shape of the complex. With the full length PPARγ complex, the structural parameters are even larger (R g = 44Å and D max = 160Å), suggesting a distinct dissociated NTD. Low resolution models were reconstructed ab initio from the corresponding experimental scattering patterns. The most typical ab initio model of PPARγ2ΔNTD/ RXRαΔNTD/CYP4A1 PPRE presented in Figure 4(c) clearly displays separate DBD and LDB domains similar to the architecture observed for the complex with the idealized DR1 and those observed for other NR heterodimers. For PPARγ2ΔNTD/RXRαΔNTD/PPRE, its atomic structure was refined against SAXS experimental data (Figure 4(a)). The position of the domains was adjusted by rigid body modeling using the available high resolution crystal structure of the complex (PDB ID: 3DZY). The model obtained by rigid body refinement agrees well with the ab initio models as seen from the superposition in Figure 4(c). The refined structure reveals an asymmetric shape with the LBD dimer positioned at the 5 end of the DNA (Figure 4(d)), an asymmetry already observed for other heterodimers studied by SAXS [16]. The scattering patterns of the full-length PPARγ2/RXRα/ PPRE in complex with TIF2 RID were also monitored. The increase of the R g and of the D max (Table 1) upon TIF2 RID interaction to the full-length heterodimer is similar to that observed for the complex with the LBD dimer, suggesting a similar mode of binding as the LBD dimer with no additional interactions of the TIF2 RID with the full-length complex. The solution studies of DNA complexes with integral receptors provided also evidence for a stoichiometry of 1 coactivator per receptor dimer. Each heterodimer binds only one coactivator protein via PPAR, this preferential binding being controlled by affinity, rather than steric exclusion. However, other domains of TIF2 may be implicated in additional interactions to the receptor as TIF2 has been shown to bridge the N-terminal and C-terminal receptor domains of estrogen receptor [33].

Conclusion
The solution structures of the functional heterodimer PPARγ2/RXRα bound to natural DR1 from regulated gene and TIF2 RID reveal the asymmetry induced by the nonsymmetric DNA target of the NR dimer with the position of the LBDs at the 5 end of the target DNA and the asymmetric mode of recruitment of TIF2. The extended conformation in solution of integral PPARγ/RXRα bound to DNA is recognized and maintained during coactivator binding. The TIF2 RID, which is intrinsically disordered, partially folds on binding to PPARγ but retains an extended conformation in the complex. The functional consequence imposed by the DNA is the asymmetric binding of the cofactor with the complex that orients the coactivator on one side of the DNA, allowing interactions with other regulatory proteins and thus leading to specific and controlled NR-mediated gene transcription. As other domains outside the TIF2 RID may interact with the heterodimer, further studies should be carried out to characterize the topology of such large complexes. A better understanding at the molecular level of the regulation of PPARγ/RXRα by their coactivators would benefit new therapeutic approaches for the treatment of metabolic diseases.