(Received for publication, July 2, 1996, and in revised form, October 4, 1996)
From the College of Pharmacy and § Program
in Molecular and Cellular Biology, Oregon State University,
Corvallis, Oregon 97331
Structurally diverse peroxisome proliferators and
related compounds that have been demonstrated to induce the
ligand-dependent transcriptional activation function of
mouse peroxisome proliferator-activated receptor (mPPAR
) in
transfection experiments were tested for the ability to induce
conformational changes within mPPAR
in vitro. WY-14,643,
5,8,11,14-eicosatetraynoic acid, LY-171883, and clofibric acid all
directly induced mPPAR
conformational changes as evidenced by a
differential protease sensitivity assay. Carboxyl-terminal truncation
mutagenesis of mPPAR
differentially affected the ability of these
ligands to induce conformational changes suggesting that PPAR ligands
may make distinct contacts with the receptor. Direct interaction of
peroxisome proliferators and related compounds with, and the resulting
conformational alteration(s) in, mPPAR
may facilitate interaction of
the receptor with transcriptional intermediary factors and/or the
general transcription machinery and, thus, may underlie the molecular
basis of ligand-dependent transcriptional activation
mediated by mPPAR
.
Peroxisome proliferator-activated receptors
(PPARs)1 are members of a large family of
ligand-inducible transcription factors that includes receptors for
retinoids, vitamin D, and thyroid and steroid hormones (1-5). The
mammalian PPAR family is composed of at least three genetically and
pharmacologically distinct subtypes, PPAR, -
, and -
(reviewed
in Ref. 6). Murine PPAR
(mPPAR
) was originally isolated from a
mouse liver cDNA library by Issemann and Green (7) who demonstrated
that the receptor was activated in transfection experiments by a group
of compounds known to induce peroxisome proliferation in rodents. A
number of structurally diverse compounds have subsequently been
demonstrated to activate PPAR
in transient transfection experiments.
Particularly noteworthy among these compounds are: 1) lipids such as
arachidonic acid (8-11) and its synthetic analog
5,8,11,14-eicosatetraynoic acid (ETYA, Refs. 9, 11-13),
8-[S]-hydroxyeicosatetraenoic acid (14), a lipoxygenase
metabolite of arachidonic acid, and linoleic acid (8-11, 14, 15); 2)
fibric acid anti-hyperlipidemic drugs (WY-14,643, clofibric acid,
gemfibrozil, ciprofibric acid; Refs. 10, 16, 17) that represent a class
of therapeutic agents useful in the treatment of hypertriglyceridemia
(18); and 3) a leukotriene D4 antagonist, LY-171883 (19). Many of these
compounds, together with phthalate ester plasticizers
(di(-2-ethylhexyl)-phthalate) and herbicides
(2,4,5-trichlorophenoxyacetic acid), are known collectively as
peroxisome proliferators (reviewed in Ref. 20). While chemically
distinct, most of these compounds have been demonstrated to induce
proliferation of peroxisomes leading to hepatic hyperplasia and
hepatocarcinogenesis in many species (20). Peroxisome
proliferator-induced alteration of hepatocyte phenotype is believed to
result from activation of PPAR
and subsequent modulation of gene
expression downstream of this nuclear receptor (reviewed in Refs. 6,
20; see below). The central role of PPAR
in xenobiotic-induced
peroxisomal proliferation was recently demonstrated by the absence of
hepatomegaly and peroxisome proliferation in mice null for expression
of this gene (21).
PPARs modulate expression of target genes by binding to response elements comprised of a degenerate direct repeat of the hexameric nucleotide sequence, TGACCT, separated by one base pair (DR1). PPAR has been shown to bind cognate response elements with high affinity only in the context of a heterodimeric complex with the retinoid X receptor (RXR, Refs. 11, 17, 22-24). PPAR·RXR heterodimeric complexes appear to be responsive to both PPAR activators and 9-cis-retinoic acid, the endogenous ligand for RXR (11, 17, 22-24).
PPAR response elements (PPREs) have been identified in the 5 regions
of several mammalian genes coding for proteins involved in lipid
metabolism such as acyl-CoA oxidase (17, 25), bifunctional enzyme (26,
27), malic enzyme (16), liver fatty acid binding protein (28),
3-hydroxy-3-methylglutaryl-CoA synthase (15), and cytochrome P450 fatty
acid
-hydroxylase (29). Such findings indicate a prominent
regulatory role for the PPAR receptor family in lipid metabolism and
homeostasis. In addition, overexpression of PPAR
and -
in
cultured fibroblasts and subsequent exposure to PPAR ligands has been
shown to confer adipogenicity (30, 31), further illustrating the
central regulatory role of PPAR family members in lipid
homeostasis.
In contrast to many other receptors in the retinoid/thyroid hormone
receptor superfamily, functional domains of PPARs and critical amino
acid residues within such putative domains have not been extensively
characterized. Two previous studies with PPAR have identified: 1) a
Glu282
Gly point mutation in mPPAR
that ameliorates
transcriptional responses to WY-14,643 and ETYA (13), and 2) a
Leu433
Arg point mutation in human PPAR
(hPPAR
)
that abolishes heterodimerization with RXR (32). The present studies
were undertaken to identify mPPAR
carboxyl-terminal receptor regions
that are important for both ligand responsiveness and
heterodimerization with mRXR
and to determine if structurally
diverse PPAR ligands induce similar conformational changes within
mPPAR
. To our knowledge, these studies provide the first
direct biochemical evidence demonstrating that peroxisome proliferators
induce conformational changes within mPPAR
. Ligand-induced
stabilization of particular mPPAR
conformational states likely
underlies the molecular basis for the ability of these compounds to
activate the receptor and to modulate expression of mPPAR
target
genes including those implicated in peroxisome proliferation.
Full-length mPPAR (7)
was kindly provided by Drs. S. Green and J. Tugwood (Macclesfield, UK)
and was used as a template for the polymerase chain reaction during
construction of all PPAR mutants described herein. Full-length mouse
RXR
(mRXR
, Ref. 33) and pGEX-cs (34) were kind gifts from Drs.
Ph. Kastner and P. Chambon (Strasbourg) and Dr. W. Dougherty (Oregon
State University), respectively.
A mPPAR amino-terminal truncation mutant was constructed by
polymerase chain reaction using a 5
primer (ML023) that introduced an
EcoRI site, favorable Kozak sequence, and an initiator
methionine fused to Asp91 of mPPAR
and a 3
primer that
introduced a BamHI site 3
of the mPPAR
natural stop
codon. The resulting fragment was appropriately digested and subcloned
into the eukaryotic expression vector, pTL1 (33), yielding PPAR
AB.
PPAR
AB is transcribed/translated in vitro at least
10-fold more efficiently than full-length receptor and exhibits DNA
binding and heterodimerization activities that are indistinguishable
from full-length receptor (data not shown). The carboxyl-terminal
truncation mutants, PPAR
AB/
448 and PPAR
AB/
425 (Fig.
1A), were prepared by polymerase chain reaction using ML023 as the 5
primer and a 3
primer that introduced stop codons at positions 448 and 425, respectively, preceding a BamHI site.
Both of the resulting fragments were appropriately digested and
subcloned into pTL1 as described above. PPAR
AB, PPAR
AB/
448,
and PPAR
AB/
425 were transcribed/translated in vitro
with equal efficiencies (data not shown).
GST-mRXR was prepared by polymerase chain reaction amplification of
full-length mRXR
using a 5
primer that introduced a HincII site immediately upstream of the natural initiator
methionine and a 3
primer that introduced an EcoRI site 3
of the natural stop codon of mRXR
. The resulting fragment was
appropriately digested and subcloned into a
EheI/EcoRI-digested GST fusion vector (pGEX-cs).
Proteins were prepared by in vitro transcription/translation using rabbit reticulocyte lysate as described previously (3, 33). Translation reactions were carried out in the presence of [35S]methionine for production of radioactively labeled proteins used in DPSAs and GST-pull down experiments, whereas receptor proteins used in electrophoretic mobility shift assays were translated in the presence of unlabeled methionine. Unprogrammed lysates were generated identically using equal amounts of linearized pTL1 in place of receptor-coding templates.
Electrophoretic Mobility Shift Assays (EMSA)Two
probes were used in these studies as follows. DR1,
5-cgag
c
ctaccctcga-3
;
ACO-PPRE,
5
-cttcgcgaacg
t
ccccttttgctcgatc-3
. One strand of each probe is shown for clarity, and the directly repeated motifs are indicated in uppercase letters and by underlining. Both DR1 (35, 36) and ACO-PPRE (25) probes have been described previously.
Receptor proteins (10 and 20 fmol of PPARAB and mRXR
,
respectively) were preincubated on ice for 15 min prior to addition of
a mix containing ~50,000 cpm of Klenow end-filled DR1 or ACO-PPRE probes. Components of the probe mix were (in mM)
HEPES-NaOH, pH 7.5, 10; EDTA, 1; dithiothreitol, 1; and NaCl, 150. The
mix was supplemented with 10% glycerol, 1 µg/µl bovine serum
albumin, and poly[d(I·C)] (2 µg/tube). The amount of lysate in
each binding reaction was held constant by addition of unprogrammed
reticulocyte lysate. Samples were loaded on a 5% polyacrylamide gel,
electrophoresed, and gels were dried and subjected to autoradiography
as described previously (33, 36).
GST-mRXR expression in the DH5
F
strain of
Escherichia coli was induced by addition of isopropyl
-D-thiogalactopyranoside (1 mM final
concentration) to the growth media and cultured for an additional
2 h. Bacterial extracts were prepared using standard methods. The
fusion protein was purified on a glutathione-Sepharose 4B column as per
the manufacturer's (Pharmacia Biotech Inc.) recommendations.
Glutathione-Sepharose 4B
(Pharmacia) was washed extensively in phosphate-buffered saline (PBS)
and resuspended in a volume of PBS sufficient to generate a 50%
slurry. This slurry (1 ml) was mixed with 2 volumes of PBS (2 ml)
containing either no protein, purified GST, or purified GST-mRXR
(proteins were at a concentration of ~1 mg/ml) and incubated with
rotation at 4 °C overnight. The resin slurry was gently centrifuged,
washed 5 times in 2 volumes of PBS (2 ml) to remove all unbound
protein, and finally resuspended in 1 volume of binding buffer (same as
EMSA buffer but without poly[d(I·C)]). GST pull-down experiments
were conducted using 20 µl (~100 fmol) of in vitro
translated 35S-PPAR
AB, 35S-PPAR
AB/
448,
or 35S-PPAR
AB/
425, and 40 µl of GST bound,
GST-mRXR
bound, or unbound glutathione-Sepharose slurries. After an
overnight incubation at 4 °C with continuous rotation, samples were
gently centrifuged and washed 10 times using 250 µl of binding
buffer. After the final wash the resin was resuspended in 30 µl of
2 × loading buffer (125 mM Tris-HCl, pH 6.8; 4%
(w/v) SDS; 1.4 M
-mercaptoethanol; 25% (v/v) glycerol;
0.1% (w/v) bromphenol blue) of which one-half was electrophoresed on
12.5% denaturing gels and processed as described previously (33, 36).
Some GST pull-down experiments were conducted as described above but
with the addition of unlabeled, annealed oligonucleotides corresponding
to ACO and DR1 PPREs to all incubations and wash buffers at a final
concentration of 5 fmol/µl.
Two µl (~10
fmol) of in vitro translated 35S-PPARAB,
35S-PPAR
AB/
448, or 35S-PPAR
AB/
425
were preincubated in 7 µl of binding buffer (as described under
"GST Pull-down Experiments") containing either WY-14,643
(pirinixic acid), ETYA, LY-171883
(5-[4
-(4"-acetyl-3"-hydroxy-2"-propylphenoxy)butyl]tetrazole), clofibrate (2-(4-chlorophenoxy)-2-methylpropanoic acid ethyl ester), clofibric acid (2-(4-chlorophenoxy)-2-methylpropanoic acid), or an
equal volume of vehicle for 30 min at 22 °C. The final concentration of vehicle did not exceed 0.15% (v/v) in any experiment conducted. Stock solutions of all ligands were prepared on the day of the experiment in dimethyl sulfoxide (WY-14,643, and LY-171883) or ethanol
(ETYA, clofibric acid, and clofibrate). DPSAs were initiated by
addition of 1 µl of 10 × stock solution of chymotrypsin in water and were allowed to proceed for 20 min at 22 °C. Reactions were terminated by addition of 1 volume of 2 × loading buffer (as
described under "GST Pull-Down Experiments").
Electrophoresis, autoradiography, and densitometric quantification were
carried out as described previously (33, 36). DPSAs conducted to
determine the effect of heterodimerization with mRXR
were carried
out essentially as described above except that
[35S]methionine-labeled PPAR preparations were incubated
with a 2-fold molar excess of in vitro translated mRXR
or
unprogrammed lysate for 10 min at 22 °C prior to addition of
ligand.
WY-14,643 was purchased from Chemsyn Science Labs (Lenexa, KS). LY-171883 and ETYA were obtained from BIOMOL (Plymouth Meeting, PA). Clofibric acid, clofibrate, and chymotrypsin were purchased from Sigma.
Carboxyl-terminal truncation mutants of PPARAB were constructed
to define regions of the receptor required for interaction with RXR and
to determine if diverse ligands require distinct mPPAR
structural
features. Based on the crystal structures of RXR
(37, 38) and
retinoic acid receptor
(RAR
, Ref. 38) LBDs and the predicted
structural similarity of these receptors to mPPAR
(39 and data not
shown) two PPAR
AB carboxyl-terminal truncation mutants were prepared
as follows: 1) PPAR
AB/
448 that lacks a portion of putative helix
H11 and all of helix H12, and 2) PPAR
AB/
425 that lacks putative
helices H10-H12 (see Fig. 1A). Because both
carboxyl-terminal truncation mutants lack the core of the putative
ligand-dependent transcriptional activation function (AF-2,
Ref. 39), neither would be expected to activate transcription in a
ligand-dependent manner.
EMSAs were conducted to compare the
ability of PPARAB, PPAR
AB/
448, and PPAR
AB/
425 to bind
two degenerate DR1 probes: a DR1 retinoid responsive element described
previously (35) and a peroxisome proliferator-activated response
element (PPRE) identified in the promoter region of the rat acyl-CoA
oxidase gene (ACO-PPRE) that confers peroxisome proliferator
inducibility on this gene (25). While none of the PPAR
receptors bound either probe alone (Fig. 2,
lanes 3, 5, 7, 11, 13, and 15), addition of
in vitro translated mRXR
resulted in
mRXR
·PPAR
AB (Fig. 2, lanes 4 and 12) and
mRXR
·PPAR
AB/
448 (Fig. 2, lanes 6 and
14) heterodimeric complex formation on both probes.
PPAR
AB/
425 did not interact with mRXR
on either probe (Fig. 2,
lanes 8 and 16). RXR homodimeric complexes have
previously been demonstrated to bind DR1 response elements (36,
40-43), and indeed such complexes are observed in our binding assays
on the DR1 but not the ACO-PPRE probe (Fig. 2, compare lanes
2 and 10). The efficiency of mRXR
·PPAR
AB/
448 complex formation on both probes was reduced approximately 2-fold relative to that of mRXR
·PPAR
AB, suggesting that mPPAR
residues 448-468 contribute to the stability of the heterodimeric
complex but are not absolutely required for complex formation and DNA binding. However, truncation of an additional 23 mPPAR
carboxyl-terminal residues (amino acids 425-468; PPAR
AB/
425)
abolished the ability of the receptor to interact with mRXR
on
either probe (Fig. 2, lanes 8 and 16).
Protein-protein interaction experiments were carried out to investigate
the ability of PPAR carboxyl-terminal truncation mutants to interact
with RXR independently of DNA binding. GST-mRXR
fusion protein,
immobilized on glutathione-Sepharose, was used in standard GST
pull-down experiments for this purpose. In vitro translated
35S-PPAR
AB and 35S-PPAR
AB/
448 both
interacted with GST-mRXR
(Fig. 3, lanes 4 and 5) while an interaction between
35S-PPAR
AB/
425 and GST-mRXR
was not detected (Fig.
3, lane 6). The efficiency of
35S-PPAR
AB/
448 interaction with GST-mRXR
was
reduced approximately 2-fold relative to that of
35S-PPAR
AB with GST-mRXR
in agreement with DNA
binding experiments described above. No interactions between any of the
PPAR receptor proteins and an immobilized GST protein (Fig. 3,
lanes 7-9) or glutathione-Sepharose alone were observed
(data not shown). Additionally, results from experiments conducted in
the presence of unlabeled response elements, identical to those used in
DNA binding assays (see above), were indistinguishable from those
described above (data not shown).
The Sensitivity of mPPAR
We have
adapted a differential protease sensitivity assay (DPSA, Ref. 36) for
use with 35S-PPARAB to address the possibility that
peroxisome proliferators and related compounds (see Fig. 1B)
interact directly with and alter the protease sensitivity of the
receptor. Digestion of 35S-PPAR
AB with increasing
concentrations of chymotrypsin in the presence of 100 µM
LY-171883, ETYA, or WY-14,643 (Fig. 4A,
lanes 11-13, 14-16, and 17-19,
respectively) resulted in the appearance of protease-resistant
fragments of approximately 33, 31, and 27 kDa, referred to hereafter as
PF33, PF31, and PF27, respectively. Clofibric acid and clofibrate, when
examined at concentrations of 100 µM, resulted in very
weak signals (data not shown); therefore, these PPAR ligands were
examined at concentrations of 1 mM. While clofibric acid
clearly induced formation of PF33, PF31, and PF27 (Fig. 4A,
lanes 5-7), clofibrate only weakly induced formation of
these proteolytic fragments (Fig. 4A, lanes
8-10). The glucocorticoid receptor ligand, dexamethasone, had no
effect on the proteolytic sensitivity of 35S-PPAR
AB at
concentrations up to 1 mM (data not shown). Moreover, none
of the mPPAR
activators examined affected the protease sensitivity of other nuclear receptors such as mRXR
(data not shown), indicative of the specificity of these observations. These results suggest that
mPPAR
undergoes a ligand-induced conformational change upon interaction with compounds previously demonstrated to activate the
receptor in transient transfection experiments (7, 13, 19, 44).
Chymotrypsin-resistant fragments induced by clofibric acid, clofibrate,
LY-171883, ETYA, and WY-14,643 appear to be indistinguishable
suggesting that a similar change within mPPAR
may be induced by all
five PPAR activators. The following rank order of efficacy of the five
PPAR activators for induction of PFs within PPAR
AB, at
concentrations of 100 µM, was determined using
quantitative densitometric scanning of autoradiographs from DPSAs (as
described previously in Ref. 36; data not shown): WY-14,643
ETYA > LY-171883
clofibric acid > clofibrate.
DPSAs were carried out using truncation mutants
35S-PPARAB/
448 and 35S-PPAR
AB/
425
(see Fig. 1A) toward the goal of determining if distinct
receptor regions are required for responsiveness to structurally diverse PPAR ligands. A clear differential proteolytic pattern was
observed with 35S-PPAR
AB/
448 in the presence of 1 mM clofibric acid, 100 µM LY-171883, and 100 µM WY-14,643 (Fig. 4B, lanes 5-7,
11-13, and 17-19, respectively) and a weaker but
detectable differential proteolytic pattern was observed with 1 mM clofibrate (Fig. 4B, lanes 8-10).
Ligand-induced alterations in the protease sensitivity of
35S-PPAR
AB and 35S-PPAR
AB/
448 appeared
to be qualitatively indistinguishable for all ligands examined except
that the proteolytic fragments derived from
35S-PPAR
AB/
448 were of smaller mass reflecting the
truncation of 21 carboxyl-terminal amino acids (see arrows
in Fig. 4B; termed PF33
448, PF31
448, and PF27
448).
However, in contrast to 35S-PPAR
AB, the protease
sensitivity of 35S-PPAR
AB/
448 was only weakly
affected by ETYA (Fig. 4B, lanes 14-16). The
following rank order of the five PPAR activators for induction of PFs
within PPAR
AB/
448, at concentrations of 100 µM, was
determined using quantitative densitometric scanning (data not shown):
WY-14,643
LY-171883
clofibric acid > ETYA = clofibrate.
These results suggest that the most carboxyl-terminal 21 mPPAR
residues (448-468 corresponding to all of putative helix 12 and a
portion of helix 11; see Fig. 1A) are important for mPPAR
responsiveness to ETYA. With the possible exception of WY-14,643, 35S-PPAR
AB/
425 did not exhibit a differential
proteolytic pattern in the presence of any PPAR ligands examined (Fig.
4C) suggesting that the extreme carboxyl-terminal mPPAR
amino acids may be required for responsiveness to many PPAR ligands
(see below).
DPSAs were conducted using
35S-PPARAB at a constant chymotrypsin concentration and
increasing concentrations of PPAR ligands (WY-14, 643, ETYA, LY-171883,
CFA; see Fig. 1B) to determine the dependence of PF33, PF31,
and PF27 on ligand concentration. Induction of all proteolytic
fragments from 35S-PPAR
AB was clearly
ligand-dependent in all cases (Fig.
5A-D), and the relative potencies with which
these compounds induced 35S-PPAR
AB conformational change
in vitro was generally consistent with previously reported
transcriptional activation studies (Refs. 7, 13, 19; see
"Discussion").
PPAR Activator-induced Alteration in Chymotryptic Sensitivity of mPPAR
Because PPAR has been demonstrated to heterodimerize
with RXR (10, 11, 22, 23; see Figs. 2 and 3), DPSAs were conducted to examine the effects of heterodimerization with mRXR
on the induction of 35S-PPAR
AB proteolytic fragments by WY-14,643. DPSAs,
in the presence of unprogrammed lysate or in vitro
translated mRXR
, were carried out at a constant protease
concentration and increasing concentrations of WY-14,643. Interaction
with mRXR
did not alter the protease sensitivity of unliganded
35S-PPAR
AB (Fig. 6A, compare
lanes 2 and 6) or
35S-PPAR
AB/
448 (Fig. 6B, compare
lanes 2 and 6). In addition, the concentration
dependence of WY-14,643 on the induction of proteolytic fragments
derived from either receptor did not differ noticeably in the presence
of mRXR
(compare lanes 2-5 with lanes 6-9 of
Fig. 6A and B, respectively). Similar results
were observed for both 35S-PPAR
AB and
35S-PPAR
AB/
448 when using the PPAR ligands clofibric
acid, clofibrate, LY-171883, and ETYA (data not shown). Moreover, the
rank order of efficacy of the five compounds tested for induction of
PFs within PPAR
AB and PPAR
AB/
448 did not differ from that
stated above (data not shown). Therefore, heterodimerization with
mRXR
does not appear to influence, positively or negatively, the
capacity of mPPAR
to bind PPAR activators and undergo ligand-induced
conformational changes. 35S-PPAR
AB/
425 was not
examined in these experiments due the inability of this receptor mutant
to interact with mRXR
(Figs. 2 and 3) or bind ligand (Fig.
4C).
Our results suggest that the extreme carboxyl-terminal amino acids
of mPPAR are required for formation of PPAR·RXR heterodimeric complexes both in solution and bound to DR1 and ACO-PPRE probes. This
finding is in agreement with a previous study that characterized a
hPPAR
point mutation (Leu433
Arg corresponding to
the same residue in mPPAR
) which abolished heterodimerization with
RXR (32) thus illustrating a critical role for this region (which is
deleted in PPAR
AB/
425). A putative leucine zipper-like heptad
repeat, located between residues 426-433 of mouse, human, and rat
PPAR
, has been postulated to mediate heterodimerization with RXR
(32). Truncation of mPPAR
to amino acid 447 (PPAR
AB/
448) gave
rise to a receptor protein that was capable of interacting with RXR and
binding to DR1 and ACO-PPRE probes, albeit at a 2-fold decreased
efficiency as compared with a receptor protein with an intact carboxyl
terminus (PPAR
AB). Considered together, these results suggest that
the mPPAR
dimerization interface contains at least
Leu433, which is 100% conserved across all PPAR subtypes
(32, data not shown), and extends through at least Ile447.
In addition to heterodimerizing with RXR, PPARs have been reported to
interact with thyroid hormone receptor (Ref. 45), and more recently,
Miyata et al. (46) reported that mPPAR
interacts with a
third member of the nuclear receptor superfamily, the orphan receptor
LXR
. Therefore, it appears that there may be physiologically relevant cross-talk between PPARs and signaling pathways mediated by
other nuclear receptors. It will be of interest to determine if other
nuclear receptors interact with PPARs through distinct or common
heterodimeric protein interfaces and if these protein-protein interactions and/or the functional capacities of the involved receptors
are allosterically regulated by DNA binding as previously demonstrated
for other nuclear receptors (47, 48).
PPARs, like other receptor proteins within the nuclear receptor
superfamily, exhibit a conserved subdivision of receptor regions referred to as A/B, C, D, and E/F (49, reviewed in Refs. 2-5). Experiments conducted with various chimeric receptor proteins composed
of putative PPAR ligand binding domains (LBDs) fused to heterologous
DNA binding domains from estrogen (7, 50) and glucocorticoid (8)
receptors, bacterial tetracycline repressor (14), and GAL4 (44, 51, 52)
have demonstrated the requirement for a large portion of the carboxyl
terminus of PPARs (D and E/F regions as defined in Ref. 7) for
ligand-responsive transcriptional activation.
PPAR activating ligands constitute a chemically diverse group of
compounds in which the most obvious common structural elements are an
acidic group (free carboxyl group, a metabolically labile derivative
thereof, or a bioisostere such as a tetrazole or sulfonamide moiety)
and a electron-rich region (aromatic ring or series of alkenes or
alkynes) (53). When considering the structural diversity exhibited by
these compounds, it seems possible that the molecular determinants of
mPPAR
interaction with each ligand or class of ligands may be
distinct. Indeed, our results indicate that distinct mPPAR
regions
are required for responsiveness to different PPAR activators. While
PPAR
AB is responsive to WY-14,643, ETYA, LY-171883, clofibric acid,
and clofibrate, as detected by DPSAs, deletion of mPPAR
residues
448-468 (PPAR
AB/
448) severely compromises responsiveness to ETYA
but not other PPAR ligands. The distal carboxyl-terminal amino acids of
mPPAR
that are deleted in PPAR
AB/
448 correspond to part (H12)
of the region that has been proposed to stabilize ligand-receptor
interactions with hRAR
by functioning as a "lid" on the ligand
binding cavity (38, 39). The greatly reduced efficacy with which ETYA
induced PPAR
AB/
448 conformational change relative to that of
PPAR
AB suggests that the hydrophobicity of putative H12 may play a
critical role in the stabilization of ETYA binding, perhaps by
stabilizing an extended conformation of this compound. Truncation of
mPPAR
residues 425-468 (PPAR
AB/
425) gave rise to a receptor
protein which was slightly responsive to WY-14,643 but unresponsive to
all other PPAR ligands examined. In addition to deletion of putative
helix H12, PPAR
AB/
425 also lacks putative helices H10 and H11,
encompassing a region that has been proposed to form one side of the
nuclear receptor ligand binding pocket (39), which may explain the
inactivity of this mutant in DPSAs. However, we cannot presently rule
out the possibility that the inactivity of PPAR
AB/
425 is due to improper protein folding and/or detrimental structural distortions outside the deleted region. Nonetheless, it is clear that
mPPAR
residues 448-468 are important for ligand binding
and/or conformational change induced by ETYA while being dispensable
for responsiveness to other PPAR ligands supporting the hypothesis that
distinct mPPAR
receptor regions may be required for interaction with
structurally dissimilar PPAR activators.
Recently, several synthetic antidiabetic thiazolidinediones (44, 51,
52) and 15-deoxy-12,14-prostaglandin J2 (44,
52) have been shown to bind directly to the ligand binding domain of
the mouse PPAR
. Direct binding of any compounds to PPAR
subtypes,
however, has not been demonstrated. Issemann et al. (7)
specifically report the lack of [3H]nafenopin binding by
mPPAR
which may be due to both a low affinity of nafenopin for
mPPAR
and a large amount of endogenous binding activity in the cell
lines tested (7). In the current study, DPSA methodology was adapted to
facilitate detection of ligand interactions with mPPAR
. DPSAs have
been demonstrated to be a useful method for detection of ligand-induced
conformational change within the nuclear receptor superfamily (36,
54-64). The differential sensitivity of liganded and unliganded
receptors is likely related to ligand-induced stabilization of one or
more receptor conformations that exhibit sensitivity to proteolytic
digestion different from that of unliganded receptor. Although DPSAs
yield data that are somewhat less amenable to quantitative analyses
than radioligand binding experiments, an advantage of the former is the
ability to detect low affinity ligand-receptor interactions. All
mPPAR
activators tested induced mPPAR
conformational change
in vitro, albeit it with varying potencies. However, the
relative potencies of these compounds with regard to induction of
PPAR
AB conformational change in vitro is generally
consistent with previously reported transcriptional activation studies
utilizing mPPAR
: WY-14,643
ETYA > LY-171883
clofibric acid (7, 13, 19). ETYA, a potent activator of
Xenopus PPAR
(11), was a substantially weaker ligand than
WY-14,643 in the in vitro studies employing PPAR
AB
described herein (Fig. 5). Hsu and co-workers (13) also reported that
ETYA was approximately 10-fold weaker than WY-14,643 as an activator of
mPPAR
in transient transfection experiments, suggesting that the
enhanced potency of ETYA reported by Keller and co-workers (11) may be
conferred by species-specific receptor activity, cell-specific factors,
and/or mechanism(s) other than direct binding of this arachidonic acid
analog to the receptor.
It has been hypothesized that Hsp72, which has been demonstrated to
interact directly with rat PPAR, may bind PPAR activators and, in
turn, allosterically activate the associated receptor protein (65).
Presently, we cannot exclude this possibility; however, results from
DPSAs in which mPPAR
carboxyl-terminal truncation mutants were used
suggest an alternative molecular mechanism. For example, deletion of 21 mPPAR
carboxyl-terminal amino acids (PPAR
AB/
448) compromises
the responsiveness of the receptor to ETYA but not other PPAR ligands.
If Hsp72 binds these ligands directly and allosterically transduces a
signal that alters the conformation of mPPAR
, it seems unlikely that
this process would be attenuated by carboxyl-terminal truncation of
mPPAR
unless the deleted amino acids are required for mPPAR
-Hsp72
interaction. In such a case, one would expect this truncation to
abolish responsiveness to all ligands. Of course, it is possible that
some ligands selectively interact with mPPAR
, Hsp72, or both
proteins in the context of an mPPAR
·Hsp72 complex. The latter
possibility would be reminiscent of ecdysone receptor in which ecdysone
binding activity is associated with a complex of ecdysone receptor and
ultraspiracle (61). In any event, ligand-induced
conformational change likely underlies the molecular basis of ligand
activation of the putative mPPAR
transcriptional activation
function, AF-2, and identification of ligand-induced mPPAR
proteolytic fragments will be of critical importance to our
understanding of the dynamic process of PPAR activation by ligands and
interaction of liganded receptor with putative transcriptional
intermediary factors.
We thank Ph. Kastner, P. Chambon, T. Lufkin, J. D. Tugwood, S. Green, and W. Dougherty for plasmid constructs; R. McParland and B. Robbins for oligonucleotide synthesis; B. Hettinger-Smith and J. E. Ishmael for useful discussions and critical review of the manuscript; and the Communications Media Center for figure preparation.
Consistent with some of the data
presented herein, Devchand et al. recently published results
demonstrating direct binding of [3H]leukotriene B4 to
a bacterially expressed and purified, GST-Xenopus PPAR
fusion protein and inhibition of this binding by unlabeled WY-14,643
(Devchand, P. R., Keller, H., Peters, J. M., Vazquez, M., Gonzalez, F. J., and Wahli, W. (1996) Nature 384, 39-43).