Eur J Biochem 271, 386–397 (2004) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03937.x Binding analyses between Human PPARc–LBD and ligands Surface plasmon resonance biosensor assay correlating with circular dichroic spectroscopy determination and molecular docking Changying Yu1,2, Lili Chen1, Haibing Luo1, Jing Chen1, Feng Cheng1, Chunshan Gui1, Ruihao Zhang1, Jianhua Shen1, Kaixian Chen1, Hualiang Jiang1 and Xu Shen1 Drug Discovery and Design Center, State Key Laboratory of Drug Research, Shanghai Institute of Materia Medica, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China; 2College of Marine Life Sciences, Ocean University of China, Qingdao, China The binding characteristics of a series of PPARc ligands (GW9662, GI 262570, cis-parinaric acid, 15-deoxy-D12,14prostaglandin J2, LY171883, indomethacin, linoleic acid, palmitic acid and troglitazone) to human PPARc ligand binding domain have been investigated for the first time by using surface plasmon resonance biosensor technology, CD spectroscopy and molecular docking simulation The surface plasmon resonance biosensor determined equilibrium dissociation constants (KD values) are in agreement with the results reported in the literature measured by other methods, indicating that the surface plasmon resonance biosensor can assume a direct assay method in screening new PPARc agonists or antagonists Conformational changes of PPARc caused by the ligand binding were detected by CD determination It is interesting that the thermal stability of the receptor, reflected by the increase of the transition temperature (Tm), was enhanced by the binding of the ligands The increment of the transition temperature (DTm) of PPARc owing to ligand binding correlated well with the binding affinity This finding implies that CD could possibly be a complementary technology with which to determine the binding affinities of ligands to PPARc Molecular docking simulation provided reasonable and reliable binding models of the ligands to PPARc at the atomic level, which gave a good explanation of the structure-binding affinity relationship for the ligands interacting with PPARc Moreover, the predicted binding free energies for the ligands correlated well with the binding constants measured by the surface plasmon resonance biosensor, indicating that the docking paradigm used in this study could possibly be employed in virtual screening to discover new PPARc ligands, although the docking program cannot accurately predict the absolute ligand-PPARc binding affinity The peroxisome proliferator-activated receptor (PPAR) belongs to the nuclear receptor superfamily [1] that plays an important role in the regulation of the storage and catabolism of dietary fats [2] PPAR contains three subtypes, PPARa, PPARb (also termed PPARd) and PPARc PPARc is a ligand-dependent transcription factor influencing the adipocyte differentiation and glucose homeostasis [3] Binding of ligands to PPARc causes conformational change in the receptor Upon binding of an agonist to PPARc, a-helices H12, H3, H4, and H5 of the receptor form a charge clamp and a hydrophobic pocket, which are essential for the recruitment of coactivator– receptor complexing and the transcriptional activation of the PPARc target genes [4,5] It has been demonstrated that PPARc is the receptor of the thiazolidinedione (TZD) class of ligands [6] Among the TZD type of anti-diabetic drugs, rosiglitazone and troglitazone are potent adipocyte differentiating agents, which activate ap2 gene expression in a PPARc-dependent manner [7] As PPARc ligands may regulate the adipogenesis, they can be designed and modified for the treatment of cardiovascular and diabetes diseases [2] Therefore, PPARc is an attractive target for new drug discovery Ligand binding to PPARc is responsible for controlling the biological functions, and discovering new ligands that may modulate PPARc’s function is a major focus in the pharmaceutical industry Accordingly, using new technology to measure ligand–PPARc binding is significant for Correspondence to H Jiang, Drug Discovery and Design Center, State Key Laboratory of Drug Research, Shanghai Institute of Materia Medica, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, 555 Zu Chong Zhi Road, Zhangjiang Hi-Tech Park, Shanghai 201203, China Fax: + 86 21 50806918, Tel.: + 86 21 50807188, E-mail: jiang@iris3.simm.ac.cn and X Shen, address as above Fax: + 86 21 50807088, Tel.: + 86 21 50806600 ext 2112; E-mail: xshen@mail.shcnc.ac.cn Abbreviations: 15-d-PGJ2, 15-deoxy-D12,14-prostaglandin J2; CPA, cis-parinaric acid; GW9662, 2-chloro-5-nitrobenzanilide; indomethacin, 1-(4-chlorobenzoyl)-5-methoxy-2-methyl-1H-indole-3-acetic acid; LBD, ligand binding domain; LBP, ligand binding pocket; LY171883, 1-{2-hydroxy-3-propyl-4-[(1H-tetrazol-5-yl)butoxyl]phenyl} ethanone; PPARc, peroxisome proliferator-activated receptor c; RU, resonance unit; SPR, surface plasmon resonance; TZD, thiazolidinedione (Received 31 July 2003, revised 10 November 2003, accepted 20 November 2003) Keywords: PPARc; receptor binding; surface plasmon resonance biosensor; circular dichroism spectroscopy; molecular docking Ó FEBS 2003 Binding analyses of human PPARc–LBD to ligands (Eur J Biochem 271) 387 both the function study of the receptor and ligand discovery Numerous technologies, such as competition radioreceptor assay [8–10], protease protection assay [11], coactivatordependent receptor ligand assay (CARLA) [12] and scintillation proximity assay (SPA) [13], have been used to measure the binding constants for ligand–PPARc interactions and in screening of ligands By employing these technologies, some important parameters evaluating the binding affinity or activity for many ligands to PPARc, such as Ki, KD, EC50 and IC50, have been obtained However, these technologies either need specific radioligands for labeling or the reporter gene has to be transfected in the cell to be detected, both of which limit the screening speed for finding new ligands, especially at the primary screening step Recently, the surface plasmon resonance (SPR) biosensor technology has been recognized as a powerful tool in monitoring receptor–ligand interactions with advantages of no labeling, real-time and noninvasive measurements [14] This advanced technology will become a potential secondary screening tool in drug screening It has been successfully used to measure the binding interactions of small molecules to the ligand-binding domain (LBD) of human estrogen receptor [15] To the best of our knowledge, there is to date no report concerning the ligand–PPAR binding assay by using SPR biosensor technology Promoted by the discovery of new PPAR agonists with the eventual aim of developing new drugs for the treatment of type II diabetes, we are trying to construct screening modes and corresponding assay methods SPR biosensor technology was used to determine the binding affinities of PPARc ligand binding domain (PPARc–LBD) with nine typical ligands, viz GI262570 [16], troglitazone [16], linoleic acid [17,18], GW9662 [19], cis-parinaric acid [20], 15-d-PGJ2 [17], indomethacin [21], Palmitic acid, and LY171883 [22] (Fig 1) It can be demonstrated that SPR biosensor technology can quantitatively detect the binding affinities of the tested ligands to PPARc, and the dissociation constants (KDs) measured by SPR biosensor are in agreement with data reported in the literature Upon binding of ligands, great conformational changes take place for PPARc [23] To address the ligand binding effect to thermal stability of PPARc, circular dichroism (CD) spectroscopic technology was used to investigate the conformational changes of PPARc–LBD resulted from the ligand binding In addition, the thermally induced unfolding process of both apo-PPARc–LBD and its ligand-bound complexes were also studied using CD spectroscopy, and the transition temperature (Tm) for each complex was estimated from the CD responses To our knowledge, this is the first report of use of CD to detect the conformational change and to monitor the Tm of PPARc unfolding The result indicated that the thermal stability of PPARc–LBD enhanced by the ligand binding, the transition temperature increments (DTm) of PPARc– LBD caused by ligand binding have a good correlation with the binding affinities This finding suggests that CD can also be used in studying ligand–PPARc binding and in screening new ligands To address the structure–binding affinity relationship, molecular docking method was used to construct the binding models of the tested ligands with PPARc–LBD Fig Structures of the PPARc ligands used in this study The 3D models provided a good explanation for the differences of the binding affinities from a structural viewpoint The predicted binding free energies of the ligands to PPARc correlate well with the binding affinities derived from the SPR biosensor determination, indicating that the docking paradigm used in this study may be involved in the cycle of discovering new PPARc agonists or antagonists as virtual screening tool Experimental procedures Preparation of ligand samples The structures of the ligands used in this study are shown in Fig Indomethacin, cis-parinaric acid and palmitic acid were purchased from Calbiochem, 15-deoxy-D12,14-protaglandin J2 (15-d-PGJ2) and linoleic acid were from Biomol and GW9662, LY171883 and troglitazone were from CAYMAN Chem Co (Ann Arbor, MI, USA) All the other reagents were purchased from Sigma in AR grade GI262570 was synthesized in our laboratory by using methods modified from Henke et al [16] and Collins et al [24] 1H NMR (400 MHz, CDCl3): d(p.p.m.) 8.82 (s, 1H), 8.03 (m, 2H), 7.60–7.37 (m, 10H), 7.22 (d, 2H, J ¼ 8.60Hz), 4.37 (m, 1H), 4.16 (t, 2H, J ¼ 6.22 Hz), 3.20 (m, 2H), 3.00 (t, 2H, J ¼ 6.22Hz), 2.36 (s, 3H); LRESI-MS: m/e 546(MH)-; anal C30H30N2O5, found: C, 74.79; H, 5.51; N, 5.07; required: C, 74.71; H, 5.53; N, 5.12 Ó FEBS 2003 388 C Yu et al (Eur J Biochem 271) All the test compounds were dissolved in DMSO as 20 mM stock solutions for the Biacore and CD experiments Expression and purification of human PPARc ligand binding domain (PPARc–LBD) protein pET15b-hPPARc–LBD plasmid was kindly provided by J Uppenberg (Department of Structural Chemistry, Pharmacia and Upjohn, Stockholm, Sweden) The expression and purification of the recombinant human PPARc–LBD in Escherichia coli were carried out by using a method slightly modified from Uppenberg et al [3] E coli BL21(DE3) cells transformed with the plasmid were grown in LB medium containing 50 lgỈmL)1 of ampicillin at 37 °C The expression of PPARc–LBD was induced by the addition of 0.2 mM of isopropyl b-D-thiogalactoside (IPTG) After induction for h at 20 °C, the cells were harvested and disrupted by sonication against NaCl/Pi buffer The supernatant was applied to a Ni-nitrilotriacetic acid column (1 mL resin), and the column was washed with 30 column volumes of loading buffer A (NaCl/Pi containing 10 mM imidazole, pH 8.8) followed by 10 column volumes of loading buffer B (NaCl/Pi containing 25 mM imidazole, pH 8.8) The PPARc–LBD protein was then eluted with elution buffer C (NaCl/Pi containing 500 mM imidazole, pH 8.8) For the Biacore experiments, imidazole in PPARc–LBD protein was removed by dialysis against HBS-EP buffer (10 mM Hepes, 150 mM NaCl, 3.4 mM EDTA, 0.005% (v/v) surfactant P20, pH 7.4), while for the CD experiment, imidazole in PPARc–LBD protein was eradicated by dialysis against CD buffer (20 mM sodium phosphate, pH 7.4) The PPARc–LBD protein sample was concentrated by using Centriprep and Centricon concentrators Any insoluble materials in the protein were removed by filtration The concentration of protein was determined from its molar extinction coefficient of e280 ẳ 12 045 M)1ặcm)1 Purication of PPARcLBD/ligand complexes To purify the PPARc–LBD/ligand complex, 20 lM PPARc–LBD in 1.5 mL of CD buffer was incubated with 15 lL of the ligand stock solution [20 mM in dimethyl sulfoxide (DMSO)] at °C for 12 h, the excessive DMSO and the ligand compound were then removed by use of a HiTrapTM Desalting column (Amersham Pharmacia Biotech AB) with CD buffer The PPARc–LBD/ligand complex with desired concentration was concentrated through a Centricon concentrator on demand Surface plasmon resonance (SPR) analyses The interaction analyses between immobilized PPARc– LBD and its ligands were performed using the dual flow cell Biacore 3000 instrument (Biacore AB, Uppsala, Sweden) Immobilization of the protein to the hydrophilic carboxymethylated dextran matrix of the sensor chip CM5 (Biacore) was carried out by the standard primary amine coupling reaction The protein to be covalently bound to the matrix was diluted in 10 mM sodium acetate buffer (pH 4.3) to a final concentration of 0.35 mgỈmL)1 Equilibration of the baseline was completed by a continuous flow of HBS-EP buffer through the chip for 1–2 h All the Biacore data were collected at 25 °C with HBS-EP as running buffer at a constant flow of 20 lLỈmin)1 All the sensorgrams were processed by using automatic correction for nonspecific bulk refractive index effects All the equilibrium constants (KDs) evaluating the protein–ligand binding affinity were determined by the steady state affinity fitting analysis of the results from Biacore data As the binding process for 15-dPGJ2 is slow, its kinetic analysis of the binding to PPARc– LBD regarding the association (kon) and dissociation (koff) rate constants were investigated based on the : (Langmuir) binding fitting mode CD spectral analyses CD spectra of PPARc–LBD and its complexes at different temperatures were obtained by use of a JASCO 715 spectropolarimeter equipped with a Neslab water bath The CD spectra scans of the molar ellipticity were recorded using an optical cell with a 0.1 cm path-length for the farUV region Averages of six scans were collated The mean residue ellipticity of the protein was calculated using molar concentration multiplied by the number of residues The ellipticities at 222 nm for PPARc–LBD and its complexes were accumulated for analysis by ORIGIN 7.0 (http:// www.OriginLab.com), a program that combines numerical integration and nonlinear global fitting routines Molecular modeling The 3D structures of the ligands were constructed using standard geometric parameters of molecular modeling software package SYBYL 6.8 (http://www.tripos.com) The geometries of the ligands were subsequently optimized by using the Power method encoded in SYBYL 6.8 to a root˚ mean-squared (rms) energy gradient of 0.05 kcalỈmol)1ỈA)1 Tripos force field [25] with GasteigerHuckel charges [26,27] ă was employed during the ligand minimization The protein models were constructed according to the crystal structure of PPARc–LBD–thiazolidinedione (TZD) complex retrieved from the Brookhaven Protein Data Bank (PDB) [28,29], entry 2PRG [5] The ligand-binding pocket (LBP) of the receptor was defined as the collection of the amino acids ˚ enclosed within a sphere of 6.5 A radius around the bound ligand (TZD) The binding models of the ligands to the receptor were constructed by docking the ligands into the LBP of PPARc–LBD employing the flexible docking program FLEXX [30] During the docking simulations, standard parameters of the FLEXX implemented in SYBYL 6.8 were used The global lowest-energy binding configuration of a ligand to the protein was identified by optimizing the rotation and translation of the ligand within the binding pocket Normally FLEXX provides more than 10 candidate configurations; configuration corresponding to the lowest interaction energy was selected as the final structure for further analysis The binding free energies of the ligands with the receptor were predicted by using the scoring function of AUTODOCK 3.0 [31] The scoring function of AUTODOCK was empirically calibrated at the level of binding free energy based on the traditional molecular force field terms, in which not only the restriction of internal rotors depending on the number of torsion angles of the ligand, but Ó FEBS 2003 Binding analyses of human PPARc–LBD to ligands (Eur J Biochem 271) 389 also on the desolvation upon binding and the hydrophobic effect (solvent entropy changes at solute–solvent interfaces) were calculated Thus, this scoring function can reflect the ligand–protein binding free energies more accurately All molecular modeling and docking simulations were performed on a Silicon Graphics Origin3200 workstation (with four CPUs) Results SPR determination of binding affinity Immobilization of PPARc–LBD typically resulted in a resonance signal at about 2000–2100 resonance units (RUs) The binding responses in RUs were continuously recorded and presented graphically as a function of time The association could be described in a simple equilibrium (A, analyte; B, ligand; AB, complex) A þ B Ð AB To determine the equilibrium dissociation constant for the interaction, the equilibrium response (Req) data were fit to an independent-binding-site model [32]: Req ¼ X Rmax;i  C Kon;i i ỵ C Kon;i 1ị where, Rmax stands for the maximal response, C is the concentration of a ligand, and Kon is the equilibrium association constant For a single-site interaction, i ¼ 1, for a two-site binding, i ¼ 2, and so on The Biacore biosensor determination results for the binding of the ligands with immobilized PPARc–LBD in the CM5 chip are shown in Fig The response data indicate that, in reaching the equilibrium, both the association and dissociation of 15-d-PGJ2 towards the immobilized PPARc–LBD are slow (Fig 2A) However, the association and dissociation phases of the other compounds were transitory, the responses reach equilibrium towards PPARc–LBD quickly, within s, and the compounds dissociated from the protein chip surface completely after s as shown in Fig 2B Two fitting methods are generally used in the data analyses for slow and fast response modes, respectively The first fitting method is the : (Langmuir) binding fitting model, in which the association rate constant (kon) and dissociation rate constant (koff) are fitted simultaneously by rate Equation 2, Fig Specificity of ligands binding to PPARc–LBD measured by SPR (Biacore 3000) Representative sensorgrams obtained from injections for 15-d-PGJ2 at concentrations of 0.156, 0.312, 0.625, 1.25, 2.5, 5.0, 10.0, and 20.0 lM (A); for troglitazone at concentrations of 0.00977, 0.0195, 0.0391, 0.0781, 0.156, 0.625, 5.0, and 20.0 lM (B); for LY171883 at concentrations of 0.625, 1.25, 2.5, 5.0, 10.0, and 20.0 lM (C) and for GW9662 at concentrations of 0.00977, 0.039, 0.156, 0.625, 2.5, 5.0, and 20.0 lM (D); over PPARc–LBD immobilized on the CM5 chip The ligands were injected for 120 s, and dissociation was monitored for more than 150 s Ó FEBS 2003 390 C Yu et al (Eur J Biochem 271) Table The kinetic constants of 15-deoxy-D12,14-protaglandin J2 (15-d-PGJ2) binding to PPARc–LBD Rmax, maximum analyte binding capacity; kon, association rate constant; koff: dissociation rate constant; KD, equilibrium dissociation constant KD ¼ koff/kon; v2 statistical value in Biacore Rmax (RU) kon (M)1Ỉs)1) koff (s)1) KD (M) v2 36.7 ± 3.06 257 ± 9.86 3.90 ± 0.074 · 10)3 1.51 ± 0.105 · 10)5 0.386 dR ¼ kon  C  ðRmax À RÞ À koff  R dt ð2Þ where, R represents the response unit, C is the concentration of the ligand This fitting model is normally used in the determination of slow binding For the fast binding ligands, steady state affinity fitting model has to be employed in calculating the binding constants Accordingly, the binding kinetic constants of 15-d-PGJ2 to PPARc–LBD were calculated by using Equation The results are shown in Table The binding constants, in terms of KD, of other compounds to PPARc–LBD were obtained employing steady state fitting methods; the steady state plots against the concentrations of troglitazone are shown in Fig 3A For ligand LY171883, up to 20 lM, the response only reached two units, as shown in Fig 2C, and its biosensor RU was independent of the analyte concentration Therefore, it can be tentatively concluded that LY171883 did not bind or showed very weak affinity to PPARc–LBD, at least in the present experimental Fig Equilibrium data analysis of ligands binding to PPARc–LBD The data for the SPR sensorgrams (Fig 2) were fitted to a single-site interaction model The plots of steady state RU vs the concentrations of troglitazone (A) and GW9662 (B), respectively, were obtained by using a steady-state fitting model conditions For ligand GW9662, at concentrations ranging from 9.77 nM to 20 lM, the responses at equilibrium increased from approximate 0.3–17RUs (Fig 2D) Estimated from the steady state plot against the concentration (Fig 3B), the KD value of GW9662 binding to PPARc– LBD is about 1.59 lM Similar to GW9662, the KD values of the remaining ligands binding to PPARc–LBD were evaluated employing the steady state-fitting model, which are listed in Table CD determination Large conformational change occurs for the PPARc–LBD when binding with ligands, especially for helix 12 (H12) [23] To investigate the thermal properties associated with the conformational changes caused by ligand binding and to identify the relationship between the binding affinity and the thermal parameter, CD spectroscopic analyses were performed to both the apo-PPARc–LBD and its ligand complexes The CD spectroscopic data were collected at the temperatures ranging from to 90 °C Because all the ligands not exhibit CD spectroscopic reflection within far-ultraviolet wavelength (data are not shown), the CD responses may assign to conformational change of the protein As an example, the CD spectra of PPARc–LBD in the absence and presence of Troglitazone and GI262570 at 4, 20, 40, 60, 90 °C, and °C again (cooled down to °C from 90 °C) are shown in Fig Similar profiles were observed for the remaining ligands (data are not shown) Comparing the CD features of the apo- and ligand bound PPARc– LBDs, we can see that ligand binding indeed induced a secondary structure change for PPARc–LBD This is in agreement with the X-ray crystallographic results [33,34], which clearly demonstrated apo- and ligand bound PPARc– LBDs adopted different conformational arrangements When comparing the CD spectra at °C with those at °C cooled down from 90 °C, a major difference of the CD features is observed, suggesting that the unfolding processes for either PPARc–LBD or its ligand complexes are irreversible (Fig 4) Corresponding to the thermally induced unfolding processes, transition temperatures exist between 40 and 60 °C (Fig 4) Thermal unfolding profiles of apo-PPARc–LBD and its complexes with the tested ligands were obtained by monitoring the 222-nm ellipticities (h) as functions of temperature Dh is defined as the ellipticity determined at a given temperature subtracting that determined at the lowest experimental temperature (4 °C in this study); and Dhmax is defined as the Dh at the highest experimental temperature (90 °C in this study) The profiles of Dh/Dhmax for apo-PPARc–LBD and its ligandbound complexes plotted against temperature are shown in Fig The transition temperature (Tm) values were obtained by fitting Dh/Dhmax data in ORIGIN 7.0 The result Ó FEBS 2003 Binding analyses of human PPARc–LBD to ligands (Eur J Biochem 271) 391 Table The equilibrium constant and Tm for the PPARc–LBD and the compounds complex The equilibrium constants (KDs) and Tm values were obtained by Biacore and CD measurements, respectively K¢D values are the equilibrium constants from the references (numbers in the parentheses) Number Analyte KD (lM) PPARc–LBD GI262570 Troglitazone Linoleic acid GW9662 cis-Parinaric acid 15-d-PGJ2 Indomethacin Palmitic acid LY171883 – 0.0034 ± 0.274 ± 1.3 ± 1.59 ± 7.80 ± 15.1 ± 38.0 ± 156 ± >1000 a Ligand bound to GST–PPARc–LBD b K¢D (lM) 0.00023 0.0142 0.084 0.187 0.24 1.05 0.88 4.72 Tm (°C) – 0.0011 [16] 0.30 [16] 4.9 [17]a 0.0033b [19] 0.669 [20] 11.6 [17] 42 [17] – – 46.14 53.06 50.39 49.31 48.31 48.94 49.49 48.15 47.79 46.91 ± ± ± ± ± ± ± ± ± ± 0.31 0.25 0.27 0.19 0.14 0.22 0.04 0.33 0.25 0.38 IC50 value is listed in Table The Tm value of apo-PPARc–LBD is % 46.14 °C, while for the ligand-bound complexes, the Tm temperatures increased with the values of 46.91–53.06 °C Binding models For the tested ligands, only the co-crystal structure of GI262570 with PPARc-LBP was reported [35], PDB entry 1FM9 Therefore, we obtained the binding models of the tested ligands with PPARc-LBP employing the docking program, FLEXX [30] The binding conformations of the ligands to PPARc-LBP derived by docking are schematically presented in Fig The corresponding hydrogen bonds and hydrophobic interactions were, respectively, calculated by using HBPLUS [36] and LIGPLOT [37] program, which are shown in Fig The binding fashions of these ligands with PPARc-LBP are in general analogous to that of TZD class agonists: the polar head interacts with the hydrophilic portion of the LBD, and the hydrophobic tail stretches down into the large hydrophobic pocket of PPARc forming strong hydrophobic contacts with several lipophilic residues such as Cys285, Leu330, Ile341, Met348 and Met364 (Fig 6) The polar heads of the ligands can be divided into three sorts: TZD, carboxylic acid, o-hydroxylacetophenone Ligands with a TZD polar head (2: troglitazone) form five hydrogen bonds with Gln286, His449, Tyr473, His323 and Ser289 (Fig 7B); ligands with a carboxylic acid polar head (5: cis-parinaric acid) form four hydrogen bonds with His449, Tyr473, His323 and Ser289 (Fig 7C); the polar head of LY171883 (9) forms only three hydrogen bonds with Tyr327, Ser289 and His323 (Fig 7D) As far as the hydrophobic interactions are concerned, the a-substituted groups of carboxyl group of GI262570 (1) form several hydrophobic contacts (Fig 7A) with PPARc, besides the four highly conserved hydrogen bonds Based on the binding models derived by FLEXX, the binding free energies of the ligands with PPARc-LBP were predicted by using AUTODOCK program [31] The predicted data are listed in Table As will be discussed later, the AUTODOCK predicted binding free energies are in well agreement with the KD values of Biacore (Table 3), indicating again the reasonability of the binding models for these ligands to PPARc-LBP Discussion Binding affinity derived from the SPR assay In the present study, for the first time, SPR biosensor technology was used to directly measure the binding interactions of small ligands to PPARc–LBD The KD values of the tested ligands to PPARc–LBD derived from the SPR determinations are in general agreement with those measured by other methods (Table 2) Upon 15-d-PGJ2 binding to PPARc–LBD, the association rate constant (kon) and dissociation rate constant (koff) were estimated to be 257 ± 9.86 M)1Ỉs)1 and 3.90 ± 0.074 · 10)3Ỉs)1 (Table 1); these two rate constants have not been reported elsewhere From the rate constants, the KD of 15-d-PGJ2 binding to the receptor was measured as 15.1 ± 1.05 lM, which is close to the value of 11.6 lM produced from the radioligand competition-binding assay [17] (Table 2) Also, the SPR measured KD values of GI262570, troglitazone, linoleic acid, and indomethacin are in agreement with those determined by other methods [16,17] However, disagreement is observed between the Biacoredetermined KD values and the data reported in the literature for GW9662 and cis-parinaric acid (CPA; Table 2) CPA is a naturally existing polyunsaturated fatty acid, it is fluorescent in a hydrophobic environment The binding affinity of CPA to PPARc–LBD produced from Biacore assay (7.80 lM) is % 10-fold larger than that (0.669 lM) obtained from fluorescent assay by Palmer and Wolf [20] This inconsistency may result from the fact that CPA is easily photochemically dimerized During the Biacore assay, the CPA solution could barely escape from the light and air, allowing the monitored concentration of CPA to be lower than expected Therefore, the higher KD value was measured GW9662 has been reported as an irreversible ligand of PPARc–LBD with a very high binding affinity (IC50 ¼ 3.3 nM) [19] GW9662 may react with Cys285 of PPARc–LBD establishing as the site of covalent modification by releasing HCl molecule (Cl atom is from the structure of GW9662) [19] However, in the Biacore assay, such an irreversible binding was not observed Upon the response of GW9662 in Biacore measurement, after equilibrium phase for 120 s, the response returned to the 392 C Yu et al (Eur J Biochem 271) Ó FEBS 2003 Fig Temperature dependence of ellipticity of apo-PPARc–LBD and its complexes at 222 nm Plots were obtained by fitting Dh/Dhmax data with the temperature for apo-PPARc–LBD ( ), GI262570 (—), ¤ linoleic acid (j), cis-parinaric acid (m) 15-d-PGJ2 (·), troglitazone (…) Fig Circular dichroism spectra of PPARc–LBD (A), troglitazone/ PPARc–LBD (B) and GI262570/PPARc–LBD (C) complexes Plots were obtained at °C (—), 20 °C (j), 40 °C (m), 60 C (Ã), 90 C ( ), Ô and C again cooled down from 90 °C ( .) baseline rapidly, followed by another binding in the next cycle, suggesting that the binding of GW9662 to PPARc– LBD is reversible rather than irreversible The binding affinity produced from Biacore assay is only 1.59 lM (Table 2) The reversible nature of GW9662 binding to PPARc–LBD may be attributed to the fact that in the Biacore experiment, the incubation time of GW9662 with PPARc–LBD is not long enough for the ligand to react with Cys285 In addition, reaction conditions such as pH value and temperature might also affect the covalent modification of PPARc by GW9662 Palmitic acid was reported as a natural ligand of PPARa [38], but there is no quantitative binding affinity for this ligand to PPARc as yet For the first time, we found that palmitic acid was also a weak ligand of PPARc Biacore SPR biosensor determination revealed that the binding constant of this ligand to PPARc–LBD is % 156 lM (Table 2) LY171883 is an LTD4 receptor antagonist, which was reported to be capable of activating PPARc by transactivation assay at micromolar concentrations [22] However, SPR determination did not detect the binding of LY171883 to PPARc, even at millimolar concentrations (Fig 2C) SPR biosensor experiments require immobilization of a receptor or ligand on a surface and monitoring its binding to a second component in solution [14] Without an appropriate method for immobilizing one reactant onto the detecting chip, SPR Biacore technology cannot be applied in binding assay and drug screening Omitting the ligands with uncertain KD values (GW9662 and cis-parinaric acid), the SPR Biacore values of KD have a good correlation with those from reported binding affinities (K¢D in Table 2), the correlation relationship between these two data sets is KD ẳ 1.062KÂD, the correlation coefcient R is as higher as 0.985 This demonstrates that SPR Biacore technology and the protein immobilizing method can be used to monitor the ligand–PPARc binding With the advantages of SPR Biacore technology in binding assay such as label-free and real time detections [14], the measurement methods established in this study can also be extended to drug screening for discovering new agonists or antagonists of PPARc Thermal stability correlates with the binding affinity X-ray crystal structures indicated that ligand-bound PPARc adopts different conformations with respect to the apoPPARc [5,33–35] The CD spectra indeed reflect the conformational changes induced by the bound ligands (Fig 4) However, ligands studied in this paper with similar function (agonists) bind to a similar conformation of Ó FEBS 2003 Binding analyses of human PPARc–LBD to ligands (Eur J Biochem 271) 393 Fig The binding conformations of the test PPARc ligands The first image is the conformational superposition within the binding pocket of PPARc, showing that these ligands adopt a similar fashion to PPARc The yellow structure in the first image is the binding conformation of GI262570 retrieved from the crystal structure of the PPARc-GI262570 complex (PDB entry 1FM9) PPARc-LDB because different ligand-bound PPARc produced a similar CD spectral feature (Fig 4) This is also in agreement with the crystal structures of ligand–PPARc– LDB complexes [5,33–35] Nevertheless, the CD determination indicated that ligand binding increased the thermal stability of PPARc To quantitatively analyze the relationship between transition temperature and binding affinity, we defined the transition temperature increment (DTm) as the Tm of a ligand complex subtracting that of the apo-PPARc– LBD The DTm data might reflect thermal stability of PPARc–LBD caused by the ligand binding It is interesting that the DTm values correlate linearly with the binding affinities of the ligands except GW9662 (Fig 8) The departure of GW9662 from the linear relationship is also derived from the experimental condition (see Discussion in the above section) Regression analysis without GW9662 resulted following the relationship between DTm values and binding affinities of the ligands to PPARc–LBD: À log KD ¼ 2:52 ỵ 0:90 DTm n ẳ 8; SD ẳ 0:293; R2 ẳ 0:952 3ị where, n is the number of tested ligands, SD is the standard error, R2 is the correlation coefficient This correlation implies the direct relationship between the ligand binding affinity and the thermal stability Apparently, strong binding of a ligand increases the thermal stability of PPARc–LBD, which thereby increases the Tm of thermally induced unfolding of PPARc–LBD This finding implies that CD spectroscopic method can also be used in detecting the binding affinity of ligands to PPARc and in screening new PPARc binders Those compounds exhibiting larger Tm values using this paradigm would therefore be expected to have potent binding affinity Structure–affinity relationship To explore the binding characteristics of the ligands to PPARc at the molecular level, molecular docking method was applied to construct the ligand–PPARc binding models and to predict the binding affinities Due to the uncertain binding affinity, GW9662 was not included in the docking analysis AUTODOCK predicted binding free energies of the eight tested ligands to PPARc to have a good correlation with the binding constants (Table and Fig 9) The regression equation for SPR Biacore measured binding affinity (–logKD), which was obtained by using the predicted binding free energy (Table 3) as a unique descriptor By means of a simple linear regression analysis, the statistical results are presented in Eqn 4: À log KD ¼ 2:93 À 0:34  DGbinding n ¼ 8; SD ¼ 0:726; R2 ¼ 0:846 ð4Þ where, n is the number of tested ligands, SD is the standard error, and R2 is the correlation coefficient This correlation between the predicted binding free energies and the Biacoremeasured binding affinity demonstrates again that the binding models of the ligands to PPARc derived from docking simulation are, in a way, reliable However, Fig 394 C Yu et al (Eur J Biochem 271) Ó FEBS 2003 Fig Schematic representations of hydrogen bonds and hydrophobic interactions of PPARc with GI262570 (A), troglitazone (B), cis-parinaric acid (C), and LY171883 (D) The corresponding hydrogen bonds and hydrophobic interactions were, respectively, calculated by using HBPLUS [36] and LIGPLOT [37] programs Dashed lines represent hydrogen bonds and spiked residues form hydrophobic contacts with the ligands shows several dots, especially those corresponding to linoleic acid and LY171883, that depart from the regression line This indicates that docking parameters would be improved if Eqn was used in predicting ligand-PPARc binding affinity accurately AUTODOCK predicted that binding free energy (DGbinding) contains three terms: intermolecular electrostatic interaction (DGes), intermolecular atomic affinity (DGnes) and intra- molecular torsional free energy (DGtor), which, respectively, represent the contributions of the receptor–ligand electrostatic interactions, non-electrostatic interactions (including hydrogen bonding and hydrophobic interaction), and the entropy effect from the loss of torsion degrees of freedom upon ligand binding (Table 3) The separated terms of the predicted binding free energies indicate that non-electrostatic interactions dominate the binding of the ligands and Ó FEBS 2003 Binding analyses of human PPARc–LBD to ligands (Eur J Biochem 271) 395 Table The binding free energies of the ligands binding to PPARc The binding free energies (kcalỈmol)1) of the protein–ligand complex were estimated by the scoring function of AUTODOCK 3.0 Number Ligand – log (KD) DGbinding DGnes DGes DGtor GI262570 Troglitazone Linoleic acid cis-Parinaric acid 15-d-PGJ2 Indomethacin Palmitic acid LY171883 8.46852 6.56225 5.88606 5.10791 4.82102 4.42022 3.80688 carboxylic acid group > o-hydroxylacetophenone (Fig 7) By considering the fact that the tails of the ligands are located in the same hydrophobic pocket of PPARc (Fig 6), the above order explains adequately why LY171883 is the weakest PPARc binder and troglitazone is much more active than the ligands containing a carboxylic acid polar head such as cis-parinaric acid, linoleic acid, 15-d-PGJ2 and palmitic acid (Table 2) In comparison with other ligands, GI262570 forms several additional hydrophobic contacts with PPARc (Fig 7A), which enhances the binding affinity of GI262570 to PPARc On the contrary, the hydrophobic tail of indomethacin is shorter than those of other ligands, which decreases the hydrophobic interactions with PPARc–LBD The flexible palmitic acid contains a bond with more rotational potential than cis-parinaric acid; binding with the receptor the former ligand lost more entropy than the later (Table 3) This is one of the reasons that cis-parinaric acid binds to PPARc more tightly than does palmitic acid (Table 2) In conclusion, we demonstrated that SPR biosensor technology can quantitatively measure the binding affinity for ligand–PPARc interaction, and thereby can be potentially extended in the compounds screening for discovering the new agonists or antagonists of PPARc CD spectroscopy detected the conformational changes of PPARc induced by ligand binding Ligand binding enhances the thermal stability of PPARc, which is reflected in the increase of the transition temperature (Tm), and correlates well with the ligand binding affinity The binding models constructed by using docking modeling for the ligands to PPARc provided a good explanation for the structure-binding affinity relationship, and provided an attractive way for predicting the overall binding affinity, although its separate components cannot be as accurately predicted This result indicated that the binding models, docking paradigm and scoring function might be extended to virtual screening for finding new hits of PPARc ligands from the available databases Accordingly, combining above three methods is Ó FEBS 2003 396 C Yu et al (Eur J Biochem 271) possibly an appropriate strategy for identifying novel ligands that may bind to PPARc, i.e (a) search potentially active compounds from the molecular databases by using molecular docking; (b) perform a primary screening by means of SPR biosensor technology and (c) confirm the binding affinities of candidate compounds employing CD spectroscopic technology [39] It has to be emphasized that the paradigm described above can just provide primary hits for PPARc binders Structural optimization by using either traditional medicinal chemistry or combinatorial chemistry should be performed based on the active hits for finding more potent PPARc ligands Acknowledgements We would like to thank Jonas Uppenberg for providing us the pET15bhPPARc-LBD plasmid The research was supported by grants from National Natural Science Foundation of China (grants 29725203, 20372069 and 20072042), the State Key Program of Basic Research of China (grants 1998051115, 2002CB512807 and 2002CB512802), Life Science Foundation for Young Scientists of CAS (grant STZ-00–06), Shanghai Basic Research Project (grant 02DJ14070), and Qi Ming Xing Foundation of Shanghai Ministry of Science and Technology (grant 00QB14034) The technical support from Biacore AB Co is acknowledged 10 11 12 13 14 15 16 References Mangelsdorf, D.J & Evans, R.M (1995) The RXR heterodimers and orphan receptors Cell 83, 841–850 Willson, T.M., Brown, P.J., Sternbach, D.D & Henke, B.R (2000) The PPARs: from orphan receptors to drug discovery J Med Chem 43, 527–550 Uppenberg, J., Svensson, C., Jaki, M., Bertilsson, G., Jendeberg, L & Berkenstam, A (1998) Crystal structure of the ligand binding domain of the human nuclear receptor PPARc J Biol Chem 273, 31108–31112 Lee, G., Elwood, F., McNally, J., Weiszmann, J., Lindstrom, M., Amaral, K., Nakamura, M., Miao, S.-C., Cao, P., Learned, R.M., Chen, J.-L & Li, Y (2002) T0070907, a selective ligand for peroxisome proliferator-activated receptor c, functions as an antagonist of biochemical and cellular activities J Biol Chem 277, 19649–19657 Nolte, R.T., Wisely, G.B., Westin, S., Cobb, J.E., Lambert, M.H., Kurokawa, R., Rosenfeld, M.G., Willson, T.M., Glass, C.K & Milburn, M.V (1998) Ligand binding and co-activator assembly of the peroxisome proliferator-activated receptor-c Nature 395, 137–143 Reginato, M.J & Lazar, M.A (1999) Mechanisms by which thiazolidinediones enhance insulin action Trends Endocrinol Metab 10, 9–13 Hulin, B., McCarthy, P.A & Gibbs, E.M (1996) The glitazone family of antidiabetic agents Curr Pharm Des 2, 85–102 Cobb, J.E., Blanchard, S.G., Boswell, E.G., Brown, K.K., Charifson, P.S., Cooper, J.P., Collins, J.L., Dezube, M., Henke, B.R., Hull-Ryde, E.A., Lake, D.H., Lenhard, J.M., Oliver, W Jr, Oplinger, J., Pentti, M., Parks, D.J., Plunket, K.D & Tong, W.-Q (1998) N-(2-Benzoylphenyl)-L-tyrosine PPARc agonists Structure–activity relationship and optimization of the N-aryl substituent J Med Chem 41, 5055–5069 Young, P.W., Buckle, D.R., Cantello, B.B.C., Chapman, H., Clapham, J.C., Coyle, P.J., Haigh, D., Hindley, R.M., Holder, J.C., Kallender, H., Latter, A.J., Lawrie, K.W.M., Mossakowska, D., Murphy, G.J., Cox, L.R & Smith, S.A (1998) Identification of high-affinity binding sites for the insulin sensitizer rosiglitazone 17 18 19 20 21 22 23 (BRL-49653) in rodent and human adipocytes using a radioiodinated ligand for peroxisomal proliferator-activated receptor c J Pharmacol Exp Ther 284, 751–759 Rhoads, R.P., Greenwood, P.L., Bell, A.W & Boisclair, Y.R (2000) Organization and regulation of the gene encoding the sheep acid-labile subunit of the 150-kilodalton insulin-like growth factor-binding protein complex Endocrinology 141, 1425–1433 Dowell, P., Peter, V.J., Zabriskie, T.M & Leid, M (1997) Ligandinduced peroxisome proliferator-activated receptor a conformational change J Med Chem 272, 2013–2020 Krey, G., Braissant, O., Horset, F.L., Perroud, M., Kalkhoven, E., Parker, M & Wahli, W (1997) Fatty acids, eicosanoids, and hypolipidemic agents identified as ligands of peroxisome proliferator-activated receptors by coactivator- dependent receptor ligand assay Mol Endocrinol 11, 779–791 Nicholsa, J.S., Parksa, D.J., Conslerb, T.G & Blancharda, S.G (1998) Development of a scintillation proximity assay for peroxisome proliferator-activated receptor c ligand binding domain Anal Biochem 257, 112–119 Cooper, M.A (2003) Optical biosensors in drug discovery Nat Rev Drug Discovery 1, 515–528 Rich, R.L., Hoth, L.R., Geoghegan, K.F., Brown, T.A., LeMotte, P.K., Simons, S.P., Hensley, P & Myszka, D.G (2002) Kinetic analysis of estrogen receptor/ligand interactions Proc Natl Acad Sci USA 99, 8562–8567 Henke, B.R., Blanchard, S.G., Brackeen, M.F., Brown, K.K., Cobb, J.E., Collins, J.L., Harrington, W.W., Hashim, M.A., HullRyde, E.A., Kaldor, I., Kliewer, S.A., Lake, D.H., Leesnitzer, L.M., Lehmann, J.M., Lenhard, J.M., Orband-Miller, L.A., Miller, J.F., Mook, R.A Jr, Noble, S.A., Oliver, W Jr, Parks, D.J., Plunket, K.D., Szewczyk, J.R & Willson, T.M (1998) N-(2Benzoylphenyl)-L-tyrosine PPARc agonists Discovery of a novel series of potent antihyperglycemic and antihyperlipidemic agents J Med Chem 41, 5020–5036 Ferry, G., Bruneau, V., Beauverger, P., Goussard, M., Rodriguez, M., Lamany, V., Dromaint, S., Caner, E., Galizzi, J.P & Boutin, J.A (2001) Binding of prostaglandins to human PPARc: tool assessment and new natural ligands Eur J Pharmacol 417, 77–89 Kliewer, S.A., Sundseth, S.S., Jones, S.A., Brown, P.J., Wisely, G.B., Koble, C.S., Devchand, P., Wahli, W., Willson, T.M., Lenhard, J.M & Lehmann, J.M (1997) Fatty acids and eicosanoids regulate gene expression through direct interactions with peroxisome proliferator-activated receptors a and c Proc Natl Acad Sci USA 94, 4318–4323 Leesnitzer, L.M., Parks, D.J., Bledsoe, R.K., Cobb, J.E., Collins, J.L., Consler, T.G., Davis, R.G., Hull-Ryde, E.A., Lenhard, J.M., Patel, L., Plunket, K.D., Shenk, J.L., Stimmel, J.B., Therapontos, C., Willson, T.M & Blanchard, S.G (2002) Function consequences of cyseine modification in the ligand binding sites of peroxisome proliferator activated receptor by GW9662 Biochemistry 41, 6640–6650 Palmer, C.N.A & Wolf, C.R (1998) cis-Parinaric acid is a ligand for the human peroxisome proliferator activated receptor c: development of a novel spectrophotometric assay for the discovery of PPARc ligands FEBS Lett 431, 476–480 Lehmann, J.M., Lenhard, J.M., Oliver, B.B., Ringold, G.M & Kliewer, S.A (1997) Peroxisome proliferator-activated receptors a and c are activated by indomethacin and other non-steroidal antiinflammatory drug J Biol Chem 272, 3406–3410 Kliewer, S.A., Forman, B.M., Blumberg, B., Ong, E.S., Borgmeyer, U., Mangelsdorf, D.J., Umesono, K & Evans, R.M (1994) Differential expression and activation of a family of murine peroxisome proliferator-activated receptors Proc Natl Acad Sci USA 91, 7355–7359 Kallenberger, B.C., Love, J.D., Chatterjee, V.K.K & Schwabe, J.W.R (2003) A dynamic mechanism of nuclear receptor activa- Ó FEBS 2003 24 25 26 27 28 29 30 31 Binding analyses of human PPARc–LBD to ligands (Eur J Biochem 271) 397 tion and its perturbation in a human disease Nat Struct Biol 10, 136–140 Collins, J.L., Blanchard, S.G., Boswell, G.E., Charifson, P.S., Cobb, J.E., Henke, B.R., Hull-Ryde, E.A., Kazmierski, W.M., Lake, D.H., Leesnitzer, L.M., Lehmann, J., Lenhard, J.M., Orband-Miller, L.A., Gray-Nunez, Y., Parks, D.J., Plunkett, K.D & Tong, W.-Q (1998) N-(2-Benzoylphenyl)-L-tryosine PPARc agonists Structure–activity relationship and optimization of the pheny alkyl ether moiety J Med Chem 41, 5037–5054 Vinter, J.G., Davis, A & Saunders, M.R (1997) Strategic approaches to drug design I An intergrated software framework for molecular modeling J Comput Aided Mol Des 1, 31–51 Marsili, M & Gasteiger, J (1980) Pi charge distribution from molecular topology and pi orbital electronegativity Croat Chem Acta 53, 601–614 Gasteiger, J & Marsili, M (1980) Iterative partial equalization of orbital eletronegativity – a rapid access to atomic charges Tetrahedron 36, 3219–3228 Bernstein, F.C., Koetzle, T.F., Williams, G.J., Meyer, E.F Jr, Brice, M.D., Rogers, J.R., Kennard, O., Shimanouchi, T & Tasumi, M (1977) The protein data bank: a computer-based archival file for macromolecular structures J Mol Biol., 112, 535–542 Berman, H.M., Westbrood, J., Feng, Z., Gilliland, G., Bhat, T.N., Weissig, H., Shindyalov, I.N., & Bourne, P.E (2000) The Protein Data Bank Nucl Acids Res., 28, 235–242 Rarey, M., Kramer, B., Lengauer, T & Klebe, G (1996) A fast flexible docking method using an incremental construction algorithm J Mol Biol 261, 470–489 Morris, G.M., Goodsell, D.S., Halliday, R.S., Huey, R., Hart, W.E., Belew, R.K & Olson, A.J (1998) Automated docking using a lamarckian genetic algorithm and empirical binding free energy function J Comp Chem 19, 1639–1662 32 Rich, R.L., Dayas, Y.S.N., Morton, T.A & Myszka, D.G (2001) High-resolution and high-throughput protocols for measuring drug/human serum albumin interactions using BIAcore Anal Biochem 296, 197–207 33 Oberfteld, J.L., Collins, J.L., Holmes, C.P., Goreham, D.M., Cooper, J.P., Cobb, J.E., Lenhard, J.M., Hull-Ryde, E.A., Mohr, C.P., Blanchard, S.G., Parks, D.J., Moore, L.B., Lehmann, J.M., Plunket, K., Miller, A.B., Milburn, M.V., Kliewer, S.A & Willson, T.M (1999) A peroxisome proliferator-activated receptor c ligand inhibits adipocyte differentiation Proc Natl Acad Sci USA 96, 6102–6106 34 Cronet, P., Petersen, J.F.W., Folmer, R., Blomberg, N., Sjoblom, K., Karlsson, U., Lindstedt, E & Bamberg, K (2001) Structure of the PPARa and -c ligand binding domain in complex with AZ242; ligand selectivity and agonist activation in the PPAR family Structure 9, 699–706 35 Gampe, R.T., Montana, V.G., Lambert, M.H., Miller, A.B., Bledsoe, R.K., Milburn, M.V., Kliewer, S.A., Willson, T.M & Xu, H.E (2000) Asymmetry in the PPARc/RXRa crystal structure reveals the molecular basis of heterodimerization among nuclear receptors Mol Cell 5, 545–555 36 McDonald, I.K & Thornton, J.M (1994) Satisfying hydrogen bonding potential in proteins J Mol Biol 238, 777–793 37 Wallace, A.C., Laskowski, R.A & Thornton, J.M (1995) LIGPLOT: a program to generate schematic diagrams of protein– ligand interactions Protein Eng 8, 127–134 38 Willson, T.M & Wahli, W (1997) Peroxisome proliferatoractivated receptor agonists Curr Opin Chem Biol 1, 235– 241 39 Shen, J., Xu, X., Cheng, F., Liu, H., Luo, X., Shen, J., Chen, K., Zhao, W., Shen, X & Jiang, H (2003) Virtual screening on natural products for discovering active compounds and target information Curr Medical Chem 10, 2327–2342  ... the bound ligands (Fig 4) However, ligands studied in this paper with similar function (agonists) bind to a similar conformation of Ó FEBS 2003 Binding analyses of human PPARc–LBD to ligands (Eur... the structure? ?binding affinity relationship, molecular docking method was used to construct the binding models of the tested ligands with PPARc–LBD Fig Structures of the PPARc ligands used in this... through a Centricon concentrator on demand Surface plasmon resonance (SPR) analyses The interaction analyses between immobilized PPARc– LBD and its ligands were performed using the dual flow cell