(Received for publication, December 31, 1996, and in revised form, February 6, 1997)
From the Department of Cell Biology, Parke-Davis Pharmaceutical Research Division, Warner-Lambert Company, Ann Arbor, Michigan 48105
Adipocyte differentiation is regulated both
positively and negatively by external growth factors such as insulin,
platelet-derived growth factor (PDGF), and epidermal growth factor
(EGF). A key component of the adipocyte differentiation process is
PPAR, peroxisomal proliferator-activated receptor
. To determine
the relationship between PPAR
activation and growth factor
stimulation in adipogenesis, we investigated the effects of PDGF and
EGF on PPAR
1 activity. PDGF treatment decreased ligand-activated
PPAR
1 transcriptional activity in a transient reporter assay.
In vivo [32P]orthophosphate labeling
experiments demonstrated that PPAR
1 is a phosphoprotein that
undergoes EGF-stimulated MEK/mitogen-activated protein (MAP)
kinase-dependent phosphorylation. Purified PPAR
1 protein
was phosphorylated in vitro by recombinant activated MAP kinase. Examination of the PPAR
1 sequence revealed a single MAP kinase consensus recognition site at Ser82. Mutation of
Ser82 to Ala inhibited both in vitro and
in vivo phosphorylation and growth factor-mediated
transcriptional repression. Therefore, phosphorylation of PPAR
1 by
MAP kinase contributes to the reduction of PPAR
1 transcriptional
activity by growth factor treatment.
Peroxisome proliferator-activated receptors
(PPARs)1 are members of the nuclear hormone
receptor superfamily (1). These receptors heterodimerize with retinoic
acid-like receptor, RXR, and become transcriptionally active when bound
to ligand. The three PPAR isoforms (,
, and
) differ in their
C-terminal ligand binding domains, and each appears to bind and respond
to a specific subset of agents including hypolipidemic drugs, long
chain fatty acids, aracadonic acid metabolites, and antidiabetic
thiazolidinediones (2-4). PPAR
is expressed predominantly in mouse
white and brown fat, with lower levels in liver, whereas PPAR
is
present in heart, kidney, and liver (5, 6). PPAR
expression is
ubiquitous (7, 8).
Ectopic expression of either PPAR or PPAR
in NIH-3T3 cells is
sufficient to induce adipocyte differentiation in the presence of
PPAR
activators (9, 10). The rapid induction of PPAR
during
adipocyte differentiation and its enriched expression in adipose
tissues suggest that PPAR
is responsible for the initiation and
maintenance of the adipocyte phenotype in vivo (9).
Previously two isotypes of PPAR
(PPAR
1 and PPAR
2) have been
identified in 3T3-L1 adipocytes (11). Zhu et al. (12) have
demonstrated that these two isotypes are derived from a single PPAR
gene by alternative promoter usage and RNA splicing. However, thus far, no functional difference has been found between the two isotypes.
Adipogenesis is a complex process; multiple hormones and factors
regulate the conversion of progenitor cells to adipocytes. Insulin
and/or insulin-like growth factor enhance the ability of PPAR ligand to
induce differentiation of both 3T3-L1- and PPAR-overexpressing cell
lines (9, 13). In contrast, growth factors such as epidermal growth
factor (EGF), platelet-derived growth factor (PDGF), and fibroblast
growth factor inhibit adipocyte conversion (14-18). In this report, we
find that activation of EGF and PDGF receptors and subsequent
phosphorylation of PPAR
1 by the MAP kinase signaling pathway
decreases PPAR
1 transcriptional activity. This repression is
mediated by MAP kinase phosphorylation of Ser82 on
PPAR
1. These studies identify PPAR
1 as a substrate of MAP kinase
and provide evidence for regulation of PPAR
1 activity by
phosphorylation.
Cell culture reagents were purchased from Life Technologies, Inc. The ECL detection system and carrier-free [32P]orthophosphate were obtained from Amersham Corp. The PDGF was purchased from Intergen, while EGF was from Harlan. PD98059 and BRL49653 were synthesized at Parke-Davis Pharmaceutical Research Division of Warner-Lambert Co.
Vector Constructs and Transient TransfectionFor eukaryotic
expression of PPAR1 and RXR
, the entire PPAR
1 or RXR
cDNA was inserted 3
to the cytomegalovirus promoter in pSG5
(Stratagene). Constitutively active MAP kinase kinase (CA-MEK), which
contains mutations at Ser218 to Glu and Ser222
to Glu was obtained from Dr. S. Decker (Parke-Davis). Site-directed mutagenesis of PPAR
1/pSG5 was conducted using the MORPH
site-specific plasmid DNA mutagenesis system (5 Prime
3 Prime,
Inc., Boulder, CO). The oligonucleotide used in mutagenesis was
CAAAGTAGAACCTGCAGCTCCACCTTATTATTCTGAAAAGACCC and changed
Ser82 to Ala. The reporter construct used in the
transfections contained three copies of the PPRE site from the aP2
enhancer (ARE7) inserted upstream of a minimal thymidine kinase (TK)
promoter in the pGL3 basic luciferase vector (a gift from Dr. R. Wyborski). All constructs were sequenced prior to use. For the
transient transfection, NIH 3T3 cells were grown in 10% fetal calf
serum/Dulbecco's modified Eagle's medium and co-transfected with
various expression plasmids and pCMV
-galactosidase plasmid
(Clontech) using Lipofectamine (Life Technologies, Inc.). After
recovery, cells were placed in 0.5% bovine serum albumin/Dulbecco's
modified Eagle's medium for 5 h and then treated with 25 µM BRL49653 and/or 100 ng/ml PDGF for 16 h.
Luciferase and
-galactosidase activities were determined using a
Luciferase assay (Promega) and the Galacto-light system (Tropix,
Inc.).
To express the maltose-binding protein (MBP) fusion
proteins in Escherichia coli, the coding regions of
PPAR1, PPAR
, and RXR
were inserted downstream of the
isopropyl-
-D-thiogalactopyranoside-inducible MalE-lacZ
gene fusion in the pMAL-C2 plasmid (New England
Biolabs). Protein expression was induced with
isopropyl-
-D-thiogalactopyranoside, and the fusion
proteins were partially purified by amylose affinity chromatography
(19). In vitro phosphorylation of MBP, MBP-PPAR
1, and
MBP-PPAR
by MAP kinase was performed as described previously (20)
using a bacterially expressed glutathione S-transferase fusion protein of 44-kDa MAP kinase (GST-MAP kinase) and the 45-kDa MEK
(GST-MEK1). Using a PPAR
-specific polyclonal antibody (produced using the MBP-PPAR
fusion protein),2
in vitro translated PPAR
1 and the mutant PPAR
1 (S82A)
were immunoprecipitated and phosphorylated by active GST-MAP kinase as
described above.
Approximately 0.5 µg of the
partially pure MBP-PPAR1, phosphorylated or unphosphorylated, and
0.5 µg of MBP-RXR
protein were preincubated for 15 min in 1 × mobility shift assay buffer (15 mM Hepes, pH 7.0, 80 mM KCl, 10% glycerol, 1 µg of poly(dI-dC), 0.2 mM EDTA, and 0.4 mM dithiothreitol) to allow
heterodimer formation, or MBP-PPAR
1 was phosphorylated prior to
heterodimerization with MBP-RXR
. Approximately 20 fmol of a
32P-labeled double-stranded ARE7 PPRE-containing
oligonucleotide probe (5
-AATTCAAGGCAGAAAGTGAACTCTGATCCAGTAAGAAG-3
)
was added to the protein mix and incubated at room temperature for 20 min. Protein-DNA complexes were analyzed in 5% 1 × TBE
polyacrylamide gels.
293T cells were
maintained in Dulbecco's modified Eagle's medium containing 10%
fetal calf serum (Life Technologies, Inc.) and transfected using a
calcium phosphate transfection protocol according to the manufacturer
(Stratagene). For in vivo labeling, transfected cells were
serum-starved overnight in 0.5% bovine serum albumin/Dulbecco's
modified Eagle's medium, pretreated with phosphate-free medium for
1 h, and subsequently incubated in 0.8 mCi of
[32P]orthophosphate at 37 °C for 3 h. Cells were
preincubated with either BRL49653 (25 µM) or PD98059 (40 µM) for 15 min followed by the addition of EGF (100 ng/ml). EGF stimulation proceeded for 5 or 15 min prior to removal of
the media and cell lysis. Cells were harvested in radioimmune
precipitation lysis buffer (10% glycerol, 137 mM NaCl, 1%
Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 20 mM Tris, pH
8.0, 2 mM EDTA, complete protease inhibitors, and 20 mM NaVO4). Whole cell extracts were
immunoprecipitated with anti-PPAR antibody and protein A-Sepharose
(Life Technologies, Inc.) for 16 h at 4 °C and resolved in 10%
SDS-PAGE. To detect MAP kinase activity in 293T cells, whole cell
lysates were prepared and subjected to Western blot analysis using the
anti-active MAP kinase antibody (Promega) and ECL system
(Amersham).
Transcription reporter assays were used to determine the
effect of growth factors on the transcriptional activity of PPAR1. The luciferase reporter constructs used in NIH3T3 cells contained the
TK promoter (TKpGL3) or three copies of ARE7 PPRE elements upstream of
the TK promoter (ARE7-TKpGL3). In the absence of co-transfected PPAR
1 and RXR
expression plasmids, no PPAR
ligand
(BRL49653)-dependent transcription was observed from either
the TkpGL3 or ARE7-TKpGL3 (Fig. 1A). In the
presence of PPAR
1 and RXR
, a 2-fold increase in transcription was
observed from ARE7-TK reporter after 16 h of treatment with
BRL49653. The addition of 100 ng/ml PDGF to these cells decreased both
the basal and BRL49653-activated transcription from the ARE7. This
suggests that at least a fraction of the activity from the ARE7-TKpGL3
plasmid in the absence of exogenously added ligand is due to the
activation of the PPAR
1·RXR
heterodimer by endogenous ligands.
This activity was also reduced by PDGF treatment.
Close examination of the PPAR amino acid sequence revealed that
PPAR
1 contains one serine residue, Ser82, whose
surrounding amino acids correspond to the consensus phosphorylation site for MAP kinase (Fig. 2A) (22). This site
is absolutely conserved between human and mouse PPAR
1. A variation
of the MAP kinase consensus site is also found in mouse PPAR
at a
similar position in the amino acid sequence. PPAR
lacks this site
altogether (Fig. 2A). Since both EGF and PDGF are known to
activate MAP kinase in vivo, a CA-MEK that constitutively
activates MAP kinase was co-transfected with ARE7-TKpGL3, PPAR
1, and
RXR
expression plasmids. As shown in Fig. 1B, CA-MEK
decreased both the basal and the ligand-dependent PPAR
1
transcriptional activity in a dose-dependent manner. No significant effect was seen with the TKpGL3 parental reporter construct. This suggests that the intracellular signaling pathways activated by PDGF or EGF can modulate PPAR
1-dependent
transcriptional activity.
MAP Kinase Phosphorylates PPAR
To determine if PPAR1 can be phosphorylated by MAP
kinase in vitro, partially purified MBP, MBP-PPAR
1, or
MBP-PPAR
fusion proteins were incubated with preactivated GST-MAP
kinase and [
-32P]ATP under conditions that
phosphorylate myelin basic protein, a known MAP kinase substrate. As
shown in Fig. 2B, MAP kinase efficiently phosphorylated
PPAR
1 but not MBP-PPAR
or maltose-binding protein (data not
shown). Coomassie staining verified that nearly equal amounts of intact
proteins were loaded (Fig. 2C). To determine if
Ser82 is the residue phosphorylated in vitro, a
mutation was introduced into PPAR
1 that changed Ser82 to
Ala. Both the wild type PPAR
1 and the mutant PPAR
1 (S82A) were
in vitro translated and immunoprecipitated with a polyclonal anti-PPAR
antibody. The immunoprecipitated products were used as
substrates in the in vitro MAP kinase assay. Mutation at
Ser82 to Ala completely abolished the MAP
kinase-dependent phosphorylation of PPAR
1 (Fig.
2D), indicating that the Ser82 is the only amino
acid in PPAR
1 that is phosphorylated by MAP kinase. Fig.
2E represents the PPAR
antibody Western blot of the
in vitro translated proteins and shows that both proteins were expressed in the reticulocyte extracts.
To determine if
PPAR1 is phosphorylated by growth factor treatment, 293T cells were
transfected with PPAR
1, serum-starved for 24 h, and incubated
with [32P]orthophosphate. To maintain more physiologic
conditions, no phosphatase inhibitors were added to the cells prior to
lysis. Whole cell lysates were prepared after EGF treatment and
immunoprecipitated with a PPAR
-specific antibody. PPAR
1 was
weakly phosphorylated in the absence of sera and growth factors;
however, treatment with 100 ng/ml of EGF for 5 or 15 min increased
PPAR
1 phosphorylation 1.5- and 1.8-fold, respectively (Fig.
3A, lanes 3 and 4). In
293T cells, EGF treatment stimulated MAP kinase activity as determined by Western blot analysis with the anti-active MAP kinase antibody (Fig.
3B, lanes 1 and 2). To determine if
the MAP kinase signaling pathway is involved in the phosphorylation of
PPAR
1, the transfected cells were pretreated with 40 µM PD98059, a specific MEK inhibitor (20) for 15 min
prior to EGF treatment. PD98059 prevented EGF-stimulated phosphorylation of PPAR
1 (Fig. 3A, lane 6),
suggesting that MAP kinase activation is involved in the
phosphorylation of PPAR
1. At this concentration, PD98059 inhibited
MAP kinase activation by EGF (Fig. 3B, lane 3).
Interestingly, pretreatment of the cells with 25 µM
BRL49653 for 15 min also reduced the EGF-dependent phosphorylation of PPAR
1 (Fig. 3A, lane 5)
without affecting the ability of EGF to stimulate MAP kinase activity
(Fig. 3B, lane 4). This implies that occupation
of the ligand binding domain may inhibit the ability of MAP kinase to
recognize and/or phosphorylate PPAR
1.
Mutation of Ser82 to Ala in PPAR
To determine if Ser82 is the
residue-phosphorylated in vivo in response to EGF
treatment, the Ser82 Ala PPAR
1 mutant was introduced
into 293T cells, and in vivo labeling was performed in the
presence and absence of 100 ng/ml EGF (Fig.
4A). Although phosphorylation of the wild
type PPAR
1 was enhanced by EGF treatment as before, phosphorylation
of the mutant was unaffected. Similar amounts of both mutant and wild type protein were expressed in the transfected cells, as shown by
Western blot analysis (Fig. 4B). Since all PPAR
1
phosphorylation was abolished by this mutation, this result
demonstrates that the MAP kinase site at Ser82 is the only
phosphorylation site on PPAR
1.
To verify that the negative regulation of PPAR1 by growth factors
was dependent upon PPAR
1 phosphorylation, NIH 3T3 cells were
co-transfected with either the wild type PPAR
1 or Ser82
Ala PPAR
1 mutant and ARE7-TKpGL3. Transfected cells were then treated with BRL49653 in the presence or absence of PDGF. Neither basal
nor BRL49653-stimulated activity was affected by the Ser82
Ala mutant. In contrast, the activity of the Ser82
Ala mutant PPAR
1 was resistant to PDGF-mediated repression (Fig.
5).
Phosphorylation of PPAR
To determine if phosphorylation affects PPAR1 DNA
binding, a mobility shift assay was performed on a labeled
double-stranded oligonucleotide containing the ARE7 PPRE with both
in vitro phosphorylated and unphosphorylated MBP-PPAR
1.
As previously reported, PPAR
1 (Fig. 6, lane
2) alone did not bind to the ARE7 element (6). However, in the
presence of RXR
, both the phosphorylated and unphosphorylated forms
of PPAR
1 bound equally well to the ARE7 probe (Fig. 6, lanes
3 and 5). In addition, phosphorylation of preformed
PPAR
1·RXR
heterodimer prior to mobility shift assay did not
alter PPAR
1 DNA binding.
The complexity of gene expression requires the utilization of
multiple regulatory mechanisms to control both the quantity and
activity of all components of the transcription machinery including
upstream enhancer proteins. In this study, we have shown that
activation of the MAP kinase signaling pathway by EGF and PDGF induces
the phosphorylation of PPAR1 on Ser82 and that this
event decreases the ability of PPAR
1 to activate transcription.
Mutation of the phosphorylated residue (Ser82) prevents
PPAR
1 phosphorylation as well as the growth factor-mediated repression of PPAR
1-dependent transcription. This
phosphorylation-mediated transcriptional repression is not due to a
reduced capacity of the PPAR
1·RXR
complex to heterodimerize or
recognize its DNA binding site but is due to its ability to become
transcriptionally activated by ligand.
The activity of several nuclear hormone receptors is regulated by
phosphorylation. Okadaic acid-induced phosphorylation of the human 1
thyroid receptor enhances the DNA binding capacity of the protein and
increases the ligand-mediated transcription (23). Phosphorylation of
retinoic acid receptor
and RXR
modulates heterodimerization of
the receptors and consequently increases DNA binding activity (24). In
addition, the MAP kinase-dependent phosphorylation of
Ser118 on the estrogen receptor causes a 1.8-2.3-fold
increase in transcriptional activation by the AF1 domain (25). Taken
together, these data suggest that in general phosphorylation of nuclear
receptors enhances their transcriptional activity. In contrast, our
data suggest that MAP kinase phosphorylation of PPAR
1 negatively
regulates its function.
EGF, PDGF, and fibroblast growth factor inhibit the conversion of
3T3-L1 preadipocytes to adipocytes (15, 17, 18). Moreover, primary rat
adipogenic precursor cells are also inhibited from becoming adipocytes
in the presence of EGF (14), and EGF-treated animals show retardation
of the development of adipose tissue (16). Although the precise
mechanism of this inhibition is unknown, growth arrest is required for
adipogenesis. It is presumed that activation of the intracellular
signaling cascades by growth factors must interfere with the activity
of the factors involved in differentiation. We suggest that this
interference occurs with the activation of MAP kinase. The activation
of MAP kinase by EGF or PDGF induces the phosphorylation of PPAR1,
which negatively regulates its activity, thereby preventing the
progression of adipocyte differentiation.
The one piece still missing in this puzzle is how insulin promotes
adipocyte differentiation. Insulin, like other growth factors, induces
MAP kinase activity in 3T3-L1 adipocytes. In fact, two recent
publications suggest that insulin stimulation does induce the PPAR1,
PPAR
2, and PPAR
phosphorylation (26, 27). However, in contrast to
our data, both groups present data suggesting that the insulin induced
phosphorylation enhances the transcriptional activity of the PPARs. The
use of different growth factors and different cell lines may explain
this discrepancy. Yet, Zhang et al. (27) reported that
mutation of the phosphorylated serine does not prevent the activation
of PPAR
by insulin. In addition, expression of dominant negative MEK
blocks the activity of PPAR
that is not phosphorylated by MAP
kinase. This suggests that the activation of transcription by insulin
in their system occurs through a mechanism independent of the MAP
kinase-induced phosphorylation of PPAR
. Tontonoz et al.
(9) have shown that deletion of the N-terminal portion of PPAR
2,
which lacks Ser82, enhances the ability of PPAR
to
induce adipocyte differentiation. Moreover, recently Hu et
al. (28) demonstrated that the ectopic expression of a mutant
PPAR
2 (a serine to alanine mutation at position 112 in PPAR
2,
which is equivalent to Ser82 of PPAR
1) enhanced
sensitivity to ligand-induced adipogenesis. These results strongly
support the conclusion of the present paper.
Additional studies on adipocyte function show that although insulin
activates MAP kinase in 3T3-L1 adipocytes,
insulin-dependent metabolic responses such as glucose
uptake, glycogen synthesis, and lipogenesis are unaffected by the
inhibition of MAP kinase with the MEK inhibitor PD98059 (29, 30). In
addition, the MEK inhibitor does not prevent or delay 3T3-L1 adipocyte
differentiation (data not shown). Since many of the effects of insulin
in adipocytes do not utilize the MAP kinase signaling cascade, we
suggest that other signaling events induced by insulin during
adipogenesis more strongly regulate PPAR activity than direct
phosphorylation by MAP kinase.
The molecular mechanism of inhibition of PPAR via phosphorylation is
yet to be determined. Data presented here show that under equilibrium
conditions DNA binding of recombinant PPAR
1·RXR
complexes is
unaffected by phosphorylation, implying that heterodimerization of the
complex is also unaffected. This suggests that transcriptional activation by PPAR
1 is regulated by phosphorylation. Transcriptional activation by nuclear receptors is modulated upon the association of
the receptors with co-activators (31, 32) and co-repressors (21,
33-34). Because of allosteric changes in the receptor, ligand-bound receptor has a greater affinity for the co-activator than the co-repressor and thus enhances transcription (21, 33). Since pretreatment with BRL49653 decreased receptor phosphorylation in cell
culture, we speculate that phosphorylation, possibly by hindering
ligand binding or preventing changes in receptor conformation, plays a
role in the selectivity and/or affinity of PPAR
for the cofactors.
We thank Drs. S. Decker, R. Herrera, T. Leff, K. Pumiglia, and A. Saltiel for review of the manuscript and for helpful discussions. We thank Dr. D. Dudley for providing GST-MEK and GST-MAP kinase fusion proteins and Dr. S. Decker for CA-MEK and helpful suggestions.