Space.geocities.jp

Ligand binding and co-activator
assembly of the peroxisome
proliferator-activated receptor-g
Robert T. Nolte*†, G. Bruce Wisely*†, Stefan Westin‡, Jeffery E. Cobb§, Millard H. Lambert*, Riki Kurokawa‡,Michael G. Rosenfeldk, Timothy M. Willson§, Christopher K. Glass‡ & Michael V. Milburn* * Department of Structural Chemistry, and § Department of Medicinal Chemistry, Division of Chemistry, Glaxo Wellcome Research and Development,Research Triangle Park, North Carolina 27709, USA‡ Division of Cellular and Molecular Medicine, and k Howard Hughes Medical Institute, Department of Medicine, University of California, San Diego, La Jolla,California 92093-0651, USA† These authors contributed equally to this work The peroxisome proliferator-activated receptor-g (PPAR-g) is a ligand-dependent transcription factor that is important
in adipocyte differentiation and glucose homeostasis and which depends on interactions with co-activators, including
steroid receptor co-activating factor-1 (SRC-1). Here we present the X-ray crystal structure of the human apo-PPAR-g
ligand-binding domain (LBD), at 2.2 A˚ resolution; this structure reveals a large binding pocket, which may explain the
diversity of ligands for PPAR-g. We also describe the ternary complex containing the PPAR-g LBD, the antidiabetic
ligand rosiglitazone (BRL49653), and 88 amino acids of human SRC-1 at 2.3 A˚ resolution. Glutamate and lysine
residues that are highly conserved in LBDs of nuclear receptors form a ‘charge clamp' that contacts backbone atoms of
the LXXLL helices of SRC-1. These results, together with the observation that two consecutive LXXLL motifs of SRC-1
make identical contacts with both subunits of a PPAR-g homodimer, suggest a general mechanism for the assembly of
nuclear receptors with co-activators.
The peroxisome proliferator-activated receptor belongs to the receptor structures is folded against the LBD and the AF-2 hydro- family of nuclear receptors that includes the oestrogen receptors, phobic residues form part of the ligand-binding pocket. On the thyroid hormone receptors, retinoic-acid receptors (RARs) and basis of these observations, a ‘mouse trap' model of receptor retinoid-X receptors (RXRs)1. Although many of these receptors activation has been proposed, in which the AF-2 helix closes on have well-studied physiological ligands, such as steroids, thyroxine the ligand-binding site in response to ligand and establishes a and retinoids, more than half of the gene family are ‘orphan' transcriptionally active form of the receptor18.
receptors that, in many cases, are important in development and There is evidence that nuclear receptors require the ligand- gene regulation but have not yet been associated with a physiolo- dependent recruitment of co-activator proteins to effectively sti- gically relevant ligand. A bona fide cellular ligand for PPAR has not mulate gene transcription, which is dependent on allosteric altera- yet been found; however, several essential fatty acids, oxidized lipids tions in the AF-2 helical domain20,21. Indeed, all known ligand- and prostaglandin J derivatives can bind and activate the receptor2,3.
dependent nuclear receptors require the AF-2 domain for effective The three known human PPAR subtypes, a, g and d, show interaction with co-activators22–25. Among the best studied of the distinct tissue distributions4,5 and are associated with selective co-activators are the p160 proteins, which were initially identified ligands3,4,6. PPAR-g is expressed at high levels in adipose tissue biochemically as ligand-dependent nuclear-receptor-interacting and macrophages and can strongly induce adipogenesis of preadi- factors and whose genes were cloned and identified as members pocyte cells, indicating that an endogenous PPAR-g ligand may be of a gene family of steroid receptor co-activating factors, referred to an important regulator of adipogenesis5,7–12. At submicromolar as SRC-1/NcoA-1, TIF2/GRIP-1/NcoA-2 and pCIP/ACTR/AIB1 levels, a class of compounds known as thiazolidinediones (TZDs) (refs 26–32). The nuclear-receptor-interaction domain of these activates PPAR-g and are used pharmaceutically for the treatment factors is highly conserved and contains three repeated motifs of of type 2 diabetes because they sensitize tissues to insulin13,14. PPAR- the consensus sequence LXXLL (where X is any amino acid), each of g may be the biochemical target of TZDs6,13.
which is sufficient for ligand-dependent interaction with several PPAR-g contains the general structural features of other nuclear- receptor-family members, including a central DNA-binding Here we report X-ray structure analysis of the apo-PPAR-g LBD, domain and a carboxy-terminal domain that mediates ligand- and of the ternary structure consisting of the PPAR-g LBD, the binding, dimerization and transactivation functions. PPAR-g antidiabetic ligand rosiglitazone35, and a region of the SRC-1 must form a heterodimer with RXR to bind DNA and activate nuclear-receptor-interaction domain containing two LXXLL transcription. Ligand-dependent transcription requires a highly conserved motif, termed activating function-2 (AF-2), located atthe far C terminus of the LBD15. The structures of the LBDs of RXR, Overall structure and refinement
RAR, the thyroid hormone receptor and the oestrogen receptor are The results of the crystallographic refinement are summarized in similar16–19. The LBDs are folded into three layers of a-helices that Table 1. In both the apo and the ternary complex crystal structures, allow the formation of a ligand-binding site, which is almost entirely amino acids 207–476 were modelled into the electron-density map buried within the core of these a-helices. In the structure of the and crystallographically refined. The nomenclature of the second- unliganded RXR, the AF-2 motif adopts an a-helical structure that ary-structure elements (Fig. 1a) is based on the RXR LBD crystal extends away from the LBD. In contrast, the AF-2 helix in the structure16. Most PPAR-g amino acids were well defined in the agonist-bound RAR, thyroid hormone receptor and oestrogen electron-density map, except for the loop between helices H29 and Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 10 SEPTEMBER 1998 H3, which is the most thermally mobile loop. In the ternary interfaces are conserved in both PPAR and RXR, namely, F432, complex, SRC-1 was fitted to the electron density and crystal- A433, L436, E418, E407 and K438 of PPAR-g.
lographically refined for amino acids 628–640 and 684–703. Forboth the apo and the ternary complex crystals, the same non- crystallographic dimer configuration of PPAR-g is observed as in The apo-PPAR-g ligand-binding site is a large T-shaped cavity that the RXR-a and oestrogen receptor-a crystal structures16,19.
extends from the C-terminal helix to the b-sheet lying between The crystal structure of the PPAR-g LBD reveals a bundle of 13 a- helices H3 and H6. This domain is mainly hydrophobic and is helices and a small four-stranded b-sheet (Fig. 1a). The PPAR-g buried within the bottom half of the LBD (Fig. 1b). One region of LBD is very similar to the overall fold of other nuclear-receptor the domain is 20 A˚ in length, lies between H3 and the b-sheet, and is structures from helix 3 to the C terminus. However, PPAR- parallel to H3. Another cavity, which extends from the unique in its overall tertiary structure and contains an extra helix, the C-terminal AF-2 helix and is 16 A˚ in length, is orthogonal to the designated H29, between the first b-strand and H3. The tertiary T-shaped cavity, behind H3. The total volume of this domain in the placement of H2 is different in PPAR-g compared with other apo-PPAR-g crystal structure is ,1,300 A˚3 (Fig. 1b). A proposed nuclear-receptor LBD tertiary structures, and provides easier ligand-entry site in this cavity is apparent from the crystal structure access for ligands. Helices H4, H5 and H8 are tightly packed and lies between H3 and the b-sheet. The surface around the entry between helices H1, H3, H7 and H10 at the top half of the LBD site comprises several hydrophilic sidechain amino acids, namely, (Fig. 1). This structural arrangement seems to be important in D243, E290, R288 and E295. PPAR-g is probably unique among forming the ligand-binding site: helices H3, H7 and H10 are other nuclear-receptor LBDs in having this ligand-accessible anchored to allow them to jut downwards and form the scaffolding for the large cavity of the ligand-binding site in PPAR-g.
Rosiglitazone occupies roughly 40% of the ligand-binding site in The crystal structures of the apo and ternary complexes of PPAR- the ternary complex. In general the ligand is in a U-shaped g contain a homodimer of the protein that involves a total surface conformation, wrapping around H3 with its central benzene ringarea of ,1,600 A˚2. Helix 10 of one monomer forms a coiled-coil directly behind H3 and the TZD head group and pyridine ring interaction with helix 10 of the other monomer, with amino acids wrapping around H3 (Fig. 2). The TZD head group makes several F432, A433, L436 and T440 of each monomer contributing to the specific interactions with amino acids in H3, H4, H10 and the AF-2 interaction. Helices 9 and 10 and a loop before helix 9 contribute helix (Fig. 2). The carbonyl groups of the TZD form hydrogen several salt bridges to the dimer interface; these bridges are formed bonds with two histidine residues, H323 and H449. Y473 in the AF- between residues E418 and K422, E407 and K434, and K438 and 2 helix forms a secondary hydrogen bond with H323. The partly D396. There is no evidence that PPAR-g functions as a homodimer negatively charged nitrogen of the TZD head group is within in vivo; however, both oestrogen receptor-a and RXR-a use the hydrogen-bonding distance of the Y473 sidechain hydroxyl group.
same homodimer interface as that seen in PPAR-g. The similarity of A buried lysine residue, K367, forms another secondary hydrogen the homodimer interfaces in RXR and PPAR indicates that the bond with the ligand, at residue H449. All of these primary and PPAR and RXR homodimer interface may constitute the hetero- secondary hydrogen bonds result in a fixed conformation of the dimer PPAR–RXR interface for the LBDs. In fact, PPAR and RXR TZD head group and of the participating amino acids.
share only 27% sequence identity within their LBDs, but several Next to the head group, the sulphur atom of the TZD ring is amino acids that form the strongest interactions at the homodimer positioned in a hydrophobic region of the PPAR-g ligand-binding Table 1 PPAR-g LBD structure determinations Unique (N) SRC-1, rosiglitazone, PPAR-g .
Space group a ¼ 91:87 ˚A, b ¼ 62:0 ˚A, c ¼ 118:3 ˚A a ¼ 52:21 ˚A, b ¼ 69:06 ˚A, c ¼ 177:97 ˚A Unique (N) 85% (60.0 A˚–2.2 A˚) 89.7% (20.0 A˚–2.3 A˚) 68% (2.3 A˚–2.2 A˚) 53.5% (2.4 A˚–2.3 A˚) Rfree (%)(1 j) r.m.s.d. bond lengths (A˚) r.m.s.d. bond angles (degrees) sym ¼ SjIavg † Phasing power ¼ SjF scale ¼ SjFP factor ¼ SjFP Pcalc j=SFp p and FPcalc are observed and calculated structure factors, Rfree is calculated for a randomly chosen 5% of reflections and Rfactor is calculated for the remaining 95% of reflections.
Selenomethionyl, a selenomethionine-incorporated protein; r.m.s.d., root mean square deviation from ideal geometry; % N, percentage of observed reflections used in the crystallographicrefinement relative to the total theoretically possible. Apo native1 and native2 are two apo native data sets that were used for MIR structure detemrination and crystallographic refinement,respectively.
Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 10 SEPTEMBER 1998

pocket formed by F363, Q286, F282 and L469. The central benzene zone. The initial purification and crystallization of the co-expressed ring of the ligand occupies a very narrow pocket between C285 and PPAR-g and SRC-1 protein resulted in a molar ratio of 2:1 which M364. The bridging oxygen atom between the benzene ring and the was consistent with the observed solution molecular weight (see pyridine ring provides vital geometry for the pyridine ring, which Methods). The crystal structure shows that the 88 amino acids of occupies the pocket between H3 and the b-sheet. Rosiglitazone has SRC-1 bind to a homodimer of PPAR-g in the presence of ligand. In one chiral atom in the head group that is in the S configuration, even the ternary complex, one LXXLL motif binds to one molecule of though a racemic mixture was used for crystallization.
PPAR-g and the second LXXLL motif binds to the other molecule of a-Substituted carboxylic acids can act as bio-isosteric replace- PPAR-g, with the connecting protein spanning the homodimer ments for the TZD head group, maintaining high affinity binding and receptor activation36. We predict that these carboxylic acids Nuclear-receptor AF-2 domains often contain a highly conserve the key interactions with H323 and H449, and that this glutamate residue which is important for transactivation but is not molecular recognition may be important for the binding of other required for ligand binding38,39. For example, mutation of this acidic ligands. Several naturally occurring carboxylic acids, includ- residue in RAR generates a dominant-negative receptor that is ing essential fatty acids, oxidized lipids and prostaglandin J2 transcriptionally silent and can no longer bind to co-activator metabolites, can bind and activate PPAR-g at micromolar proteins39. Here, in the ternary complex, the side chain of this concentrations2. We propose that all of these hydrophobic car- residue, E471, forms hydrogen bonds with the backbone amides of boxylic acids bind to PPAR, and that the key interactions involving K632 and L633 in helical domain HD1 of the LXXLL motif and with the histidine residues required for TZD binding are conserved with the backbone amides of K688, I689 and L690 in helix HD2. In both the acid groups. The other hydrophobic parts of the ligands would HD1 and HD2, there is a lysine residue at the amino terminus of the be free to interact in a relatively nonspecific manner with the large amphiphatic sequence that is solvent-exposed in the crystal ligand-binding pocket, resulting in the observed promiscuity of structure and does not interact directly with the nuclear receptor.
PPAR. In effect, the extra helix, H29, and the altered tertiary The C-terminal ends of HD motifs interact with the PPAR-g LBD by placement of the H2 helix define PPAR as a nuclear receptor with a similar hydrogen-bonding network to that seen at the N-terminal a new hormone specificity; it has evolved to bind to multiple natural ends. A highly conserved lysine, K301, is solvent-exposed in the ligands with moderate affinity.
apo-PPAR-g structure and seems not to be involved in the folding ofthe LBD. In the ternary complex, the lysine side chain forms Determinants of SRC-1 LXXLL interactions
hydrogen bonds with two backbone carbonyls of L636 and T639 Biochemical studies indicated that two LXXLL motifs were in HD1 and L693 and L694 in HD2. There is also biochemical required for cooperative interaction of SRC-1 with nuclear- evidence that this lysine residue is important in transcriptional receptor heterodimers37, and we chose to co-crystallize the PPAR- activity and co-activator binding40. We propose that E471 and K301 g LBD with a region of SRC-1 containing LXXLL motifs 1 and 2 define a ‘charge clamp' that allows the orientation and placement of(SRC-1623–710) in the presence of the high-affinity ligand rosiglita- the HD motif into the coactivator-binding site (Fig. 3).
Figure 1 Crystal structure of the apo-PPAR-g. a, Ribbons drawing of the apo- between PPARs a, g and d are in yellow; the secondary structure that the PPAR-g LBD, amino acids 207–476. The nomenclature of the helices is based on sequence adopts in PPAR-g is shown in red boxes for a-helices and blue the RXR-a crystal structure16. a-Helices are light blue; b-strands are green; loops arrows for b-strands. Residues involved in rosiglitazone binding are underlined.
are brown. b, A worm backbone tracing of PPAR-g, with a surface representation The sequence alignment predicts that several residues involved in direct inter- of the unoccupied van der Waals space in the ligand-binding site. We determined actions of PPAR-g with ligand are not conserved in the other PPAR subtypes, and the ‘unoccupied' volume by fitting in virtual atoms that did not occupy the van der explains the specificity of TZDs for PPAR-g: residue H323 is not found in PPAR-a, Waals surface of the protein. The total unoccupied volume is ,1,300 A˚3. c, and Q286 is not found in PPAR-a or PPAR-d.
Sequence alignment of the human PPAR LBDs. Amino acids that are conserved Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 10 SEPTEMBER 1998

The hydrophobic face of the LXXLL helix of SRC-1 packs into a and the threonine in HD1 or the glutamine in HD2. Although the hydrophobic pocket formed between E471 and K301 of PPAR-g.
+6 residue may be important as a capping residue for helix This pocket is formed from helix-packing interfaces of H3, H4, H5 stabilization, the side chain does not contact the surface of the and the AF-2 helix (Fig. 3). Amino acids L468, L318, T297, Q314, LBD and is unlikely to convey specificity to the HD. The +7 to +10 L311, V315, K301 and E471 of PPAR-g make up most of the binding residues form no specific sequence, but these residues might interact site for the interacting motifs of SRC-1. The total surface area of the with other LBD domains and contribute to nuclear-receptor region of interaction between LXXLL and PPAR-g is ,800 A˚2 and involves specific salt bridges between the charge clamp on theoutside of the binding site and a buried hydrophobic core of the Structure and function of AF-2 domains
LXXLL motif; L633, L636 and L690, I693 of the two SRC-1 motifs g molecules in the ternary complex have nearly interact hydrophobically with L468 and L318 of the PPAR-g LBD.
identical Ca conformations, with a root mean square (r.m.s.) The ternary complex reveals some of the specificity determinants deviation of 0.76 A˚. Although the two molecules in the apo- that target the individual HD motifs of SRC-1 to specific nuclear PPAR-g homodimer are very similar from the N terminus to receptors. The charge clamp is presumably not selective in placing amino acid 455, they exhibit entirely different conformations of the HD motif into this site, although a particular nuclear-receptor the AF-2 helix and the preceding loop. In one of the apo-PPAR-g dimer may be selective for different co-activators. Selectivity is more molecules, the AF-2 helix is folded against the LBD in a conforma- likely to reside in the amino acids located N-terminally and C- tion that is very similar to that in the ternary ligand-bound terminally to the HD motif that are not conserved between co- complex. In contrast, in the other molecule of the apo-PPAR-g activators but which do interact with the PPAR-g LBD. A study of homodimer the AF-2 helix adopts an extended conformation, the individual contacts that the HD motifs make with the LBD projecting slightly away from the LBD. Perhaps, in the apo receptor, reveals the role that each residue may play in the recognition the AF-2 helix can assume both an ‘active' (AF-2 helix in the process. The residues at positions −3 and −2 of the HD motifs liganded state) and an ‘inactive' conformation, with the ligand (Table 2) do not appear to make any significant interactions with acting to lock the receptor into the active conformation. This the LBD; these residues rise up over the charge-clamp residue scenario would be analogous to the proposed ‘mouse trap' model E471 and away from the surface of the LBD. The amino acid at of receptor activation18. However, a different explanation was position −1 fits in a shallow pocket created by P467 and L468 of suggested by analysis of the interaction of the SRC-1 HD with the the AF-2 helix and the +4 residue of the HD motif. There is a ligand-bound receptor. Closer examination of the apo-PPAR-g preference for large hydrophobic residues at position −1 of the structure revealed that the extended AF-2 helix contacted the HD motif to allow optimal interaction with PPAR-g and, in the charge clamp of a crystallographically related PPAR-g molecule in crystal structure, a lysine residue interacts hydrophobically at this a different homodimer (Fig. 3f). This crystal packing interaction places the AF-2 helix of one PPAR-g molecule in an ‘inactive' The residues at positions +2 and +3 of the HD motif face out into conformation that allows it to bind to the AF-2 helix of another apo- solution, making no sidechain contacts with the LBD, consistent PPAR-g molecule in a crystallographically related homodimer that with the lack of sequence conservation for these positions. The +4 is in a conformation similar to that observed for the liganded residue lies in a hydrophobic pocket at the surface of the LBD. This receptor. E471 and K301 of the ‘active' AF-2 helix form hydrogen leaves the top of the pocket exposed to solvent, perhaps allowing it bonds with the crystallographically related ‘inactive' AF-2 helix. The to easily accommodate residues other than a leucine at this position.
potential generality of this arrangement of AF-2 domains is The PPAR-g pocket that encloses the residue at position +5 of the strengthened by analysis of the apo-RXR-a LBD crystal structure, HD motif is built mainly from residues T297, A300, K301, F306, in which the extended RXR-a AF-2 domain binds the LXXLL- L311 and L318. Large hydrophobic residues, such as methionine or binding pocket of a crystallographically related RXR molecule16.
phenylalanine, could fit into this pocket but would require a modestrearrangement of one of the surrounding groups or a shift of the +5 Mechanism of ligand activation
Ca position. The residues at the +6 position seem to form a Ligand-dependent transcriptional activation by nuclear receptors hydrogen bond between the main-chain carbonyl of the +2 residue probably requires the recruitment of co-activator proteins such as Figure 2 Rosiglitazone binding with the PPAR-g LBD and SRC-1 in the ternary network of polar interactions that include the AF-2 domain. b, The secondary- complex. a, Ribbons drawing showing the ternary complex of PPAR-g LBD, structure elements are shown as a ribbon drawing, with amino acids involved in BRL49653, and the LXXLL helix domain of SRC-1. Residues around K301 and E471 ligand binding labelled. Each atom is coloured separately with the same colour that form the ‘charged clamp' are red, and the LXXLL SRC-1 helix is green.
scheme: light green, carbon; red, oxygen; blue, nitrogen; yellow, sulphur.
Rosiglitazone (stick diagram) binds in a deep cavity of the protein and provides a Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 10 SEPTEMBER 1998

SRC-1 (refs 26–28, 33). Most co-activator proteins contain one or and in AF-1 domains, indicating that this motif may play a general more copies of the LXXLL motif. The ternary structure of PPAR-g role in mediating interactions between components of cellular shows that each member of the receptor dimer interacts with a single LXXLL motif of SRC-1. The length and orientation of this The observation that two HD motifs of SRC-1 make simulta- helical motif is vital for proper backbone interactions with E471 in neous contacts with both LBDs of the PPAR-g homodimer indicates the receptor AF-2 domain and with K301 in helix H3 of the LBD.
that the multiple LXXLL motifs within co-activators may mediate These residues, which are highly conserved among members of the cooperative binding to nuclear-receptor dimers. In SRC-1, 52 nuclear-receptor family, define a ligand-dependent charge clamp amino acids separate HD1 and HD2 and 54 amino acids separate that positions the HD motif of LXXLL to allow the leucine residues HD2 and HD3. There is a similar number of intervening amino to pack tightly into a hydrophobic pocket in the receptor LBD. The acids between the HDs in pCIP and TIF-2. The distance residues that define this hydrophobic pocket are also highly con- HD motifs may add more specificity to interactions with nuclear- served among other members of the nuclear-receptor family, receptor dimers. Although only two HD motifs would be necessary suggesting a common mechanism for co-activator binding. The for cooperative binding of SRC-1 to a nuclear-receptor dimer, LXXLL motif is also present in several other transcription factors perhaps the extra motif (HD3) may allow combinatorial Figure 3 SRC-1 interactions with PPAR-g. a, A sigma-weighted 2Fo–Fc omit backbone of the PPAR-g LBD is in green. PPAR-g amino acids binding to the electron-density map is shown contoured at 1.0 j for the area surrounding the LXXLL helix are also shown in green. d, Residues H631–T640 of SRC-1 are rosiglitazone ligand. b, A ribbons drawing of the PPAR-g LBD dimer and SRC-1, coloured as in c, with an electrostatic surface of PPAR-g showing the coactivator- including the ligand rosiglitazone. The two PPAR-g monomers are blue and green binding site. E471 and K301 side chains result in the red (negative) and blue and the two SRC-1 interacting helices are yellow. The structure of SRC-1 was (positive) charges on the surface of the coactivator-binding site at the N and C determined from amino acids 628–640 and 684–703 and was crystallographically termini of the SRC-1 helix, respectively. e, Residues H687–E696 of SRC-1 are refined. Very weak electron density from residues 670 to 684 was visible but was coloured as in c, with an electrostatic surface of PPAR-g showing the coactivator- not crystallographically refined and is shown as a dashed line. SRC-1 amino acids binding site. f, Amino acids L465–K474 of the PPAR-g AF-2 helix of one monomer 642–669 were disordered and not structurally determined. The diagram shows in the apo structure are shown in: green, carbon; blue, nitrogen; red, oxygen, with how one SRC-1 molecule, with two interacting domains, forms a complex with a an electrostatic surface of PPAR-g showing the coactivator-binding site. E471 and PPAR-g homodimer. The dashed line connecting the two structurally determined K301 side chains result in the red (negative) and blue (positive) charges on the domains of SRC-1 is the proposed connection between these two domains. c, surface at the N and C terminus of the other PPAR-g monomer. This figure shows The binding of SRC-1 (amino acids 628–642) to the LXXLL-binding site of PPAR-g.
how one monomer in the apo crystal structure orientates its AF-2 helix into the SRC-1 is coloured: yellow, carbon; blue, nitrogen; red, oxygen. The ribbon coactivator-binding site of another crystallographically related monomer.
Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 10 SEPTEMBER 1998 Table 2 AF-2 and HD helix alignments Subcloning and protein purification. Amino acids 176–477 were expressed by subcloning into the T7 Escherichia coli expression vector, pRSETa (Invitrogen). The His-tagged plasmid was transformed into BL21(DE3) cells, which were grown at 25 8C, induced with 1 mM isopropyl-b-D-thiogalactoside (IPTG) at A ¼ 1:0 for 1 h, and then collected. The His-tagged PPAR-g LBD protein was purified with an affinity column of Ni2+-coupled Sepharose (50 mM Tris-HCl, pH 8.0, 250 mM NaCl, 25 mM imidiazole, pH 7.5). The protein was then subjected to limited proteolysis with Arg-C at 4 8C overnight 8 and then purified with a mono-Q–Sepharose column using a salt gradient. The protein was N-terminal sequenced to ensure purity of .95% of the digested Thyroid receptor-a PPAR-g LBD amino acids 183–477. The selenomethionyl protein was prepared in the same way as the wild-type protein.
Oestrogen receptor-a The SRC1–PPARg complex was formed by expression in E. coli BL21(DE3) Glucocorticoid receptor of PPAR-g amino acids 206–477 (plasmid PRSETa; N-terminal His-tag Progesterone receptor MKKGHHHHHHG) and SRC-1 amino acids 623–710 (plasmid pACYC184; Nuclear-receptor (NR) AF-2 helix motifs and co-activator helix motifs are aligned on the basis N-terminal MKK). The transformed cells were grown at 22 8C, induced with of the predicted helical structure overlap. The core box of hydrophobic residues that coulddirectly bind to the LXXLL-binding site of the LBD is underlined. There is a high degree of 1 mM IPTG at A ¼ 1:0 for 1 h, and then collected. The SRC1–PPARg conservation in all of the HD1, HD2 and HD3 boxes, indicating a common function for the complex was purified with an affinity column of Ni2+-coupled Sepharose serial structure of these domains. The numbering for LXXLL motifs used in the text is shownat the top.
(50 mM Tris-HCl, pH 8.0, 250 mM NaCl, 25 mM imidiazole, pH 7.5). Twopeaks were isolated from the superdex-75 column; these peaks containedPPAR-g alone and SRC1–PPARg at a 2:1 molar ratio as quantified using N-terminal sequencing and amino-acid-composition analysis. The 2:1 molar ratio interactions of the LXXLL motifs that are optimal for different was also consistent with the observed solution molecular weight from gel- receptor dimers.
filtration analysis and light-scattering measurements.
The observation that the AF-2 helix of an apo receptor can Crystallization and data collection. The Arg-C-digested PPAR-g LBD was interact with the LBD of a second receptor raises several questions concentrated to 15 mg ml−1 (20 mM Tris, pH 8.0, 5 mM dithiothreitol (DTT), regarding allosteric interactions between nuclear receptors and the 100 mM NaCl, 2 mM EDTA) and crystallized using the vapour phase diffusion mechanisms of ligand-dependent activation. Although the crystal method by adding equal volume amounts of concentrated protein and a structures of the apo-PPAR-g and apo-RXR-a LBDs indicate that crystallization buffer of 1.4 M sodium citrate and 100 mM HEPES (pH 7.5). We the AF-2 domains contact the LXXLL-binding pockets of adjacent added rosiglitazone to the SRC1–PPARg complex at a molar ratio of 5:1, dimers, molecular modelling indicates that rotation around one concentrated the complex to 10 mg ml−1 (25 mM Tris, pH 8.0, 5 mM DTT, amino acid of RXR-a and extension of two amino acids at this 500 mM ammonium acetate) and crystallized it using the vapour phase position would allow the RXR-a AF-2 domain to bind to the diffusion method by adding equal volumes of concentrated protein and a LXXLL-binding pocket of a dimeric partner. Specific partners of crystallization buffer of 17% PEG 4000, 200 mM ammonium acetate and RXR in heterodimers, including RAR and the thyroid hormone 20 mM HEPES (pH 7.5). Complete diffraction data were measured at −170 8C receptor, allosterically inhibit the binding of ligands to RXR41,42.
with an R-axis II image plate system and an R-axis IV image plate system for the Allosteric inhibition could be due, in part, to the interaction of the PPAR-g crystals and the PPARg–SRC1–rosiglitazone crystals, respectively.
RXR AF-2 helix with the LXXLL-binding pocket of its dimeric Both detectors are mounted on a Rigaku RU-200 rotating anode generator.
partner. There is evidence for this allosteric mechanism and for Data reduction, merging and scaling were done with the programs DENZO and competition of AF-2 binding with SRC-1 (ref. 37).
Together with the structures of the apo-PPAR-g LBD and of the Phase determination and structure refinement. For the PPAR-g multiple ternary complex, these results indicate that ligand-dependent acti- isomorphous replacement (MIR) structure determination, single isomorphous vation may involve displacement of the AF-2 helix of a dimeric replacement (SIR) phases from the phenyl mercury acetate (PMA) heavy-atom partner from the LXXLL-binding pocket. Because the AF-2 domains data were used to calculate a difference map of the selenomethionyl-PPAR-g of many nuclear receptors are strikingly similar, in both length and data. We identified 12 selenium sites in total, with 6 sites for each individual sequence, to LXXLL motifs, this type of interaction may be relevant monomer of the PPAR-g homodimer. With this information, the non- to the mechanisms of ligand-dependent activation for many of these crystallographic two-fold axis was uniquely determined and used in proteins (Table 2).
combination with the MIR phases for electron-density averaging. MIR There are several subtle differences between the configurations of phases were calculated using PHASES44 and improved by solvent flattening the ‘active' AF-2 domain in the apo-PPAR-g LBD structure and the and electron-density averaging in program DM45,46. We built the apo-PPAR-g AF-2 domains of the ternary complex that may explain why the model using the program Quanta (Molecular Simulations) and refined the binding of ligand would result in a relative preference for binding model in X-PLOR47.
the LXXLL co-activator motif. Thiazolidinedione binding generates The PPARg–SRC1–rosiglitazone ternary complex structure was solved by two key hydrogen bonds between H323 and H449 that essentially conventional molecular replacement methods using the unliganded PPAR-g lock the AF-2 domain into the charged-clamp configuration and structure, with X-PLOR. Molecular-replacement phases were calculated in the alter the sidechain positions in the AF-2 domain that forms the package CCP4 (ref. 46) using SFALL and SIGMAA, and were improved before LXXLL-binding pocket. Because the differences between the con- model building by solvent flattening and density averaging across the PPAR-g formations of the apo ‘active' AF-2 domain and the ternary complex dimer axis with the program DM. Once the SRC1-binding site was determined, AF-2 domain are quite small, it is likely that such changes could be residues within 5 A˚ of the site were excluded from the averaging masks. The achieved by structurally distinct ligands, indicating a potential basis resulting electron density was used to identify and build the structure of the for the mechanism of action of nuclear-receptor antagonists and SRC-1 protein, rosiglitazone and residues of PPAR-g around the LXXLL- partial agonists.
binding pocket. The unique density for each HD1 and HD2 motif relative to the These results suggest that LXXLL motifs are vital in mediating PPAR homodimer is due to three crystal packing interactions involving SRC-1.
cooperative assembly of co-activator complexes on homodimeric We have obtained other holo complexes of thiazolidinediones with PPAR-g and heterodimeric nuclear receptors.
without SRC-1 at a lower resolution than the ternary complex presented, and Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 10 SEPTEMBER 1998 they are not significantly different from the ternary structure. The SRC-1 model 26. Voegel, J. J., Heine, M. J. S., Zechel, C., Chambon, P. & Gronemeyer, H. TIF2, a 160 kDa transcriptional mediator for the ligand-dependent activation function AF-2 of nuclear receptors. EMBO J. 15, 3667–
was built using the software package O (ref. 48), and refined using X-PLOR47.
3675 (1996).
27. Torchia, J. et al. The transcriptional co-activator p/CIP binds CBP and mediates nuclear-receptor Received 8 May; accepted 29 July 1998.
function. Nature 387, 677–684 (1997).
1. Mangelsdorf, D. J. et al. The nuclear receptor superfamily: the second decade. Cell 83, 835–839 (1995).
28. Zhu, Y., Qi, C., Calandra, C., Rao, M. S. & Reddy, J. K. Cloning and identification of mouse steroid 2. Kliewer, S. A. et al. Fatty acids and eicosanoids regulate gene expression through direct interactions receptor coactivator-1 (mSRC-1), as a coactivator of peroxisome proliferator-activated receptor g.
with peroxisome proliferator-activated receptors a and g. Proc. Natl Acad. Sci. USA 94, 4318–4323
Gene Expr. 6, 185–195 (1996).
29. Hong, H., Kohli, K., Garabedian, M. J. & Stallcup, M. R. GRIP1, a transcriptional coactivator for the 3. Forman, B. M., Chen, J. & Evans, R. M. Hypolipidemic drugs, polyunsaturated fatty acids, and AF-2 transactivation domain of steroid, thyroid, retinoid, and vitamin D receptors. Mol. Cell. Biol. 17,
eicosanoids are ligands for peroxisome proliferator-activated receptors a and d. Proc. Natl Acad. Sci. 2735–2744 (1997).
USA 94, 4312–4317 (1997).
30. Anzick, S. L. et al. AIB1, a steroid receptor coactivator amplified in breast and ovarian cancer. Science 4. Kliewer, S. A. et al. Differential expression and activation of a family of murine peroxisome 277, 965–968 (1997).
proliferator-activated receptors. Proc. Natl Acad. Sci. USA 91, 7355–7359 (1994).
31. Chen, H. et al. Nuclear receptor coactivator ACTR is a novel histone acetyltransferase and forms a 5. Chawla, A., Schwartz, E. J., Dimaculangan, D. D. & Lazar, M. A. Peroxisome proliferator-activated multimeric activation complex with P/CAF and CBP/p300. Cell 90, 569–580 (1997).
receptor g: adipose-predominant expression and induction early in adipocyte differentiation.
32. Kamei, Y. et al. A CBP integrator complex mediates transcriptional activation and AP-1 inhibition by Endocrinol. 135, 798-800 (1994).
nuclear receptors. Cell 85, 403–414 (1996).
6. Willson, T. M. et al. The structure-activity relationship between peroxisome proliferator-activated 33. Voegel, J. J. et al. The coactivator TIF2 contains three nuclear receptor-binding motifs and mediates receptor g agonism and the antihyperglycemic activity of thiazolidinediones. J. Med. Chem. 39, 665–
transactivation through CBP binding-dependent and -independent pathways. EMBO J. 17, 507–519
7. Wu, Z., Bucher, N. L. R. & Farmer, S. R. Induction of peroxisome proliferator-activated receptor g 34. Heery, D. M., Kalkhoven, E., Hoare, S. & Parker, M. G. A signature motif in transcriptional co- during the conversion of 3T3 fibroblasts into adipocytes is mediated by C/EBPb, C/EBPd, and activators mediates binding to nuclear receptors. Nature 387, 733–736 (1997).
glucocorticoids. Mol. Cell. Biol. 16, 4128–4136 (1996).
35. Keen, H. Insulin resistance and the prevention of diabetes mellitus. New. Eng. J. Med. 331, 1226–1227
8. Tontonoz, P., Hu, E., Graves, R., Budavari, A. & Spiegelman, B. mPPARg2: tissue-specific regulator of an adipocyte enhancer. Genes Dev. 8, 1224–1234 (1994).
36. Buckle, D. R. et al. Non-thazolidinedione antihyperglycemic agents. 2: a-Carbon substituted b- 9. Tontonoz, P., Hu, E. & Spiegelman, B. M. Stimulation of adipogenesis in fibroblasts by PPARg2, a lipid phenylpropanoic acids. Bioorg. Med. Chem. Lett. 6, 2127–2130 (1996).
activated transcription factor. Cell 79, 1147–1156 (1994).
37. Westin, S. et al. Interactions controlling complexes of nuclear-receptor heterodimers and co- 10. Ricote, M., Li, A. C., Willson, T. M., Kelly, C. J. & Glass, C. K. The peroxisome proliferator-activated activators. Nature 395, 199–202 (1998).
receptor g is a negative regulator of macrophage activation. Nature 391, 79–82 (1998).
38. Feng, X. et al. Suprabasal expression of a dominant-negative RXR alpha mutant in transgenic mouse 11. Tontonoz, P., Nagy, L., Alvarez, J. G. A., Thomazy, V. A. & Evans, R. M. PPARg promotes monocyte/ epidermis impairs regulation of gene transcription and basal keratinocyte proliferation by RAR- macrophage differentiation and uptake of oxidized LDL. Cell 93, 241–252 (1998).
selective retinoids. Genes Dev. 11, 59–71 (1997).
12. Brun, R. P. et al. Differential activation of adipogenesis by multiple PPAR isoforms. Genes Dev. 10,
39. Durand, B. et al. Activation function 2 (AF-2) of retinoic acid receptor and 9-cis retinoic acid receptor: 974–984 (1996).
presence of a conserved autonomous constitutive activating domain and influence of the nature of the 13. Lehmann, J. M. et al. An antidiabetic thiazolidinedione is a high affinity ligand for peroxisome response element on AF-2 activity. EMBO J. 13, 5370–5382 (1994).
proliferator-activated receptor g (PPARg). J. Biol. Chem. 270, 12953–12956 (1995).
40. Henttu, P. M. A., Kalkhoven, E. & Parker, M. G. AF-2 activity and recruitment of steroid receptor 14. Hulin, B., McCarthy, P. A. & Gibbs, E. M. The glitazone family of antidiabetic agents. Curr. Pharm. coactivator 1 to the estrogen receptor depend on a lysine residue conserved in nuclear receptors. Mol. Des. 2, 85–102 (1996).
Cell. Biol. 17, 1832–1839 (1997).
15. Evans, R. M. The steroid and thyroid hormone receptor superfamily. Science 240, 889–895 (1988).
41. Kurokawa, R. et al. Regulation of retinoid signalling by receptor polarity and allosteric control of 16. Bourguet, W., Ruff, M., Chambon, P., Gronemeyer, H. & Moras, D. Crystal structure of the ligand- ligand binding. Nature 371, 528–531 (1994).
binding domain of the human nuclear receptor RXR-a. Nature 375, 377–382 (1995).
42. Forman, B. M., Umesono, K., Chen, J. & Evans, R. M. Unique response pathways are established by 17. Wagner, R. L. et al. A structural role for hormone in the thyroid hormone receptor. Nature 378, 690–
allosteric interactions among nuclear hormone receptors. Cell 81, 541–550 (1995).
43. Otwinowski, Z., Isaacs, N. & Burley, S. in Proc. CCP4 Study Weekend (ed. Sawyer, L.) 56–62 (SERC 18. Renaud, J.-P. et al. Crystal structure of the RAR-g ligand-binding domain bound to all-trans retinoic Caresbury Lab., Daresbury, 1993).
acid. Nature 378, 681–689 (1995).
44. Furey, W. S. & Swaminathin, S. in Methods in Enzymology: Macromolecular Crystallography Part B Vol. 19. Brzozowski, A. M. et al. Molecular basis of agonism and antagonism in the oestrogen receptor. Nature 277 (eds Carter, C. & Sweet, R.) 590–620 (Academic, Orlando, 1997).
389, 753–758 (1997).
45. Cowtan, K. ‘‘DM'': an automated procedure for phase improvement by density modification. Joint 20. Horwitz, K. B. et al. Nuclear receptor coactivators and corepressors. Mol. Endocrinol. 10, 1167–1177
CCP4 ESF-EACBM Newsletter Protein Crystallogr. 31, 34–38 (1994).
46. Collaborative Computational Project Number 4. The CCP4 suite: programs for protein 21. Glass, C. K., Rose, D. W. & Rosenfeld, M. G. Nuclear receptor coactivators. Curr. Opin. Cell Biol. 9,
crystallography. Acta Cryst. D 50, 760–776 (1994).
222–232 (1997).
47. Brunger, A. X-PLOR Version 3.0: A System for Crystallography and NMR (Yale Univ. Press, New Haven, 22. Barettino, D., Vivanco Ruiz, M. M. & Stunnenberg, H. G. Characterization of the ligand-dependent transactivation domain of thyroid hormone receptor. EMBO J. 13, 3039–3049 (1994).
48. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. Improved methods for building protein models 23. Danielian, P. S., White, R., Lees, J. A. & Parker, M. G. Identification of a conserved region required for in electron density maps and the location of errors in these models. Acta Cryst. A 47, 110–119 (1991).
hormone-dependent transcriptional activation by steroid hormone receptors. EMBO J. 11, 1025–
1033 (1992).
Acknowledgements. We thank W. Burkhart, M. Moyer and K. Blackburn for protein sequencing and
mass spectrometry; S. Blanchard, P. Charifson, T. Consler, S. Jordan, S. Kliewer, J. Lehmann, C. Mohr and
24. Durand, B. et al. Activation function 2 (AF-2) of retinoic acid receptor and 9-cis retinoic acid receptor; E. Xu for discussions; A. Miller for artwork; and K. Milburn for computational assistance.
presence of a conserved autonomous constitutive activating domain and influence of the nature of the
response element on AF-2 activity. EMBO J. 13, 5370–5382 (1994).
Correspondence and requests for materials should be addressed to M.V.M. (e-mail: mvm25452@ 25. Tone, Y., Collingwood, T. N., Adams, M. & Chatterjee, V. K. Functional analysis of a transactivation glaxowellcome.com). Coordinates for the apo-PPAR-g (1PRG) and the PPARg–rosiglitazone–SRC1 domain in the thyroid hormone beta receptor. J. Biol. Chem. 269, 31157–31161 (1994).
ternary complex (2PRG) have been deposited with the Brookhaven Protein Database.
Nature Macmillan Publishers Ltd 1998
NATURE VOL 395 10 SEPTEMBER 1998

Source: http://space.geocities.jp/idensikouzou/book/Nature-article.395.137-143.1998.pdf