Cross-talk between Janus Kinase-Signal Transducer and Activator of Transcription (JAK-STAT) and Peroxisome Proliferator-activated Receptor-α (PPARα) Signaling Pathways

Hepatic peroxisome proliferation induced by structurally diverse non-genotoxic carcinogens is mediated by the nuclear receptor peroxisome proliferator-activated receptor (PPARα) and can be inhibited by growth hormone (GH). GH-stimulated Janus kinase-signal transducer and activator of transcription 5b (JAK2/STAT5b) signaling and the PPAR activation pathway were reconstituted in COS-1 cells to investigate the mechanism for this GH inhibitory effect. Activation of STAT5b signaling by either GH or prolactin inhibited, by up to 80–85%, ligand-induced, PPARα-dependent reporter gene transcription. GH failed to inhibit 15-deoxy-Δ12,14-prostaglandin-J2-stimulated gene transcription mediated by an endogenous COS-1 PPAR-related receptor. GH inhibition of PPARα activity required GH receptor and STAT5b and was not observed using GH-activated STAT1 in place of STAT5b. GH inhibition was not blocked by the mitogen-activated protein kinase pathway inhibitor PD98059. STAT5b-PPARα protein-protein interactions could not be detected by anti-STAT5b supershift analysis of PPARα-DNA complexes. The GH inhibitory effect required the tyrosine phosphorylation site (Tyr-699) of STAT5b, an intact STAT5b DNA binding domain, and the presence of a COOH-terminaltrans-activation domain. Moreover, GH inhibition was reversed by a COOH-terminal-truncated, dominant-negative STAT5b mutant. STAT5b must thus be nuclear and transcriptionally active to mediate GH inhibition of PPARα activity, suggesting an indirect inhibition mechanism, such as competition for an essential PPARα coactivator or STAT5b-dependent synthesis of a more proximal PPARα inhibitor. The cross-talk between STAT5b and PPARα signaling pathways established by these findings provides new insight into the mechanisms of hormonal and cytokine regulation of hepatic peroxisome proliferation.

PPAR␣ and PPAR␥ play a key role in lipid homeostasis, adipocyte differentiation, and inflammatory responses (3,12). PPAR␣ target genes include liver-expressed enzymes involved in fatty acid ␤-oxidation and microsomal -hydroxylation (13)(14)(15). PPAR␣ thus plays a direct role in liver homeostasis by regulating lipid storage and by modulating the metabolism of important lipid signaling molecules, including prostaglandins and leukotrienes. PPAR␣ gene knock-out mice do not exhibit hepatic (16) or renal (17) peroxisome proliferative responses induced by fibrate drugs and other PPAR activators, demonstrating the key role played by PPAR␣ in these processes. PPAR␣ also mediates the carcinogenicity of foreign peroxisome proliferators (18), which are non-genotoxic hepatocarcinogens when administered chronically to rodents (19).
PPAR␣ gene expression and PPAR␣-stimulated transcriptional activity are tightly controlled by a variety of hormones that act at multiple levels and via different mechanisms. Glucocorticoids induce PPAR␣ protein expression at the transcriptional level (20), which may account for the expression of PPAR␣ diurnally and in a stress-inducible manner (21), whereas insulin treatment decreases PPAR␣ mRNA levels (22). Peroxisome proliferative responses in rodents are suppressed by the thyroid hormone triiodothyronine (23), whose receptor may compete with PPAR for heterodimerization with the retinoid X receptor RXR and for trans-activation of PPAR DNA response elements (PPREs) (24). Hepatic peroxisome proliferation can also be modulated by sex hormones, with female rats being less responsive than males to clofibrate and other peroxisome proliferators (25) and testosterone treatment abolishing this sex difference (26).
The observation that hypophysectomy enhances peroxisome proliferation in female rats (26) suggests that a pituitary factor(s) serves as a negative regulator of peroxisome proliferation. In rats, the continuous plasma growth hormone (GH) profile characteristic of adult females fully suppresses liver expression of the clofibrate-inducible P450 4A2 fatty acid -hydroxylase (27). The same suppressive effect is observed in primary rat hepatocyte cultures, where GH inhibits peroxisomal ␤-oxidation induced by clofibrate (28,29). GH has diverse effects on metabolism and growth, some of which are indirectly mediated by insulin-like growth factor-1 but many of which reflect the direct effects that GH has on gene expression. In particular, GH, like many cytokines and growth factors, directly activates JAK-STAT signaling pathways. GH binds to and thereby dimerizes its plasma membrane receptor (GHR) in a process that leads to JAK2 kinase-catalyzed tyrosine phosphorylation of STAT proteins, which are latent, cytoplasmic transcription factors (30). The tyrosine-phosphorylated STAT proteins dimerize and translocate into the nucleus, where they bind to specific DNA response element and thereby activate target gene transcription (31). Among the seven mammalian STATs, four forms (STATs 1, 3, 5a, and 5b) can be activated by GH (32)(33)(34)(35).
In the present study, we investigate the mechanism that underlies the inhibitory effect of GH on peroxisome proliferation using COS-1 cells transfected to express both the GHR/ JAK/STAT signaling pathway and the peroxisome proliferatoractivated PPAR pathway. We find that GH inhibits PPAR␣stimulated reporter gene transcription in a process that is mediated by STAT5b but not by STAT1. We further demonstrate that STAT5b tyrosine phosphorylation, DNA binding, and transcriptional activation are each essential for GH to mediate its inhibitory effects on PPAR␣ activity. These findings are discussed in the context of the implications of this cross-talk between STAT5b and PPAR␣ for the regulation of PPAR␣-dependent responses by hormones and cytokines.
Cell Culture and Transfections-COS-1 cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. Transfection of COS-1 cells grown in 12-well tissue culture plates was carried out by calcium phosphate precipitation (46). At 9 h after addition of the DNA-calcium phosphate precipitate, cells were washed and incubated in Dulbecco's modified Eagle's medium without serum for 12 h. Peroxisome proliferators were then added to the culture medium in combination with GH at concentrations specified in each figure. Cells were lysed 24 h later, and firefly luciferase and ␤-galactosidase (pCMV-␤gal; internal control) reporter activities were measured using a Galacto-light chemiluminescent reporter kit (Tropix, Bedford, MA). In some experiments, Renilla luciferase expression plasmid (pRL-TK; Promega, Madison, WI) was used in place of the pCMV-␤gal internal control plasmid, as indicated in the figure legends. Firefly and Renilla luciferase activities were measured using a dual-luciferase assay kit (Promega, Madison, WI). Transfections were performed using the following amounts of plasmid DNA/well (3.8 cm 2 ) of a 12-well tissue culture plate: 0.36 g of reporter construct (pHD(X3)Luc or pLUCA6-880), 14 ng of pCMV-mPPAR␣, 0.2 g of GHR expression plasmid, 0.12 g of JAK2 expression plasmid, and 0.2 g of STAT expression plasmid. pCMV-␤gal (0.16 g) or pRL-TK (30 ng) were included as internal controls in each cell transfection. The total amount of DNA was adjusted to 0.96 g/well using salmon sperm DNA (Stratagene, La Jolla, CA). Data shown in each figure are mean values Ϯ S.D. (for n ϭ 3 replicates) or mean values Ϯ half the range (for duplicates) and are representative of at least three such independent experiments.
EMSA Analysis-Whole cell extracts were prepared by lysing transfected COS-1 cells in lysis buffer (Tropix, Inc.) containing 1 mM dithiothreitol added prior to use. 10 g of cell extract was added to 2 l of 5ϫ gel-shift buffer (20% glycerol, 5 mM MgCl 2 , 2.5 mM EDTA, 2.5 mM dithiothreitol, 50 mM Tris-HCl), plus 1 l of 2 g/l poly(deoxyinosinicdeoxycytidylic) acid (Boehringer Mannhein), with water added to adjust the total volume to 15 l. Samples were incubated for 10 min at room temperature. 32 P-Labeled double-stranded DNA probe (1 l; 10 fmol) was then added, and the incubation was continued for 20 min at room temperature and then 10 min on ice. Loading dye (2 l of 30% glycerol, 0.25% bromphenol blue, 0.25% xylene cyanol) was then added before the mixture was loaded onto an acrylamide gel (5.5% acrylamide, 0.7% bisacrylamide in 0.5ϫ TBE). The gel was electrophoresed at 4°C for 40 min at 100 V before loading. The gel was first electrophoresed for 20 min at 100 V at 4°C at which point the dye entered the gel; electrophoresis was then continued at room temperature for 5 h. This procedure minimizes formation of nonspecific protein-DNA complexes (47). For STAT5b supershift assays, 3 l of anti-STAT5b antibody (Santa Cruz Biotechnology, Santa Cruz, CA; antibody sc-835) was added 10 min after the labeled DNA probe, followed by a 10-min incubation at room temperature and 10-min incubation on ice before samples were loaded on the gel. The STAT5 binding site of the rat ␤-casein promoter (5Ј-GGA-CTT-CTT-GGA-ATT-AAG-GGA-3Ј) was used as gel-shift probe for GH-activated STAT5a and STAT5b, and the sis-inducible element (SIE)-binding site (5Ј-gtc-gaC-ATT-TCC-CGT-AAA-TCg-tcga-3Ј) was used as gel-shift probe for analyzing GH-activated STAT1 and STAT3 (34). 32 P-Labeled DNA probe corresponding to the Z element of the CYP4A6 gene (5Ј-g-CGC-AAA-CAC-TGA-ACT-AGG-GCA-AAG-TTG-AGG-GCA-G-3Ј) was used as probe for PPAR␣ binding (48).

GH Activation of STAT5b Inhibits PPAR␣ Transcriptional
Activity-To investigate the mechanism by which GH inhibits PPAR␣-dependent liver peroxisome proliferative responses, we reconstituted GH signaling and PPAR␣-dependent peroxisome proliferation pathways by cotransfection of key components into COS-1 cells. GH signal transduction was reconstituted by cotransfection of expression plasmids encoding GHR, JAK2 kinase, and STAT5b, the major GH-responsive STAT form in liver (34). The PPAR␣ pathway was reconstituted by cotransfection of a mouse PPAR␣ expression plasmid, together with reporter plasmid containing 880 nucleotides of 5Ј-flanking DNA of the rabbit CYP4A6 gene (13) fused to a firefly luciferase reporter gene. Transfected cells were stimulated with the peroxisome proliferator Wy-14,643 at 20 M for 24 h in the presence or absence of GH (200 ng/ml). Fig. 1A shows that Wy-14,643 activation of the CYP4A6 promoter is significantly decreased in the presence of GH. Wy-14,643 activation of the CYP4A6 promoter is mediated by PPAR␣, which binds to one strong and two weaker PPREs within the 5Ј-flanking 880 nucleotides (13,49). This GH inhibitory effect may be due to GH suppression of PPAR␣-dependent transcription via the CYP4A6 PPREs; alternatively, GH may interfere with other transcription factors required for either basal or Wy-14,643inducible CYP4A6 promoter transcription, independent of PPRE. To distinguish these possibilities, we examined the effect of GH on PPAR␣-activation of the reporter construct pHD(X3)Luc, which contains three tandem repeats of a PPRE from the gene that encodes enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase, cloned upstream of a minimal promoter and luciferase reporter gene (15). Fig. 1B shows that GH inhibits, by up to 80 -85%, PPAR␣ induction mediated by this isolated PPRE. Maximal GH inhibition was achieved at 25 ng/ml, well within the physiological range of GH concentrations and consistent with the reported K d of GHR for GH of ϳ2 ng/ml (50). The inhibitory effects of GH on PPAR␣ activation were also apparent with several other PPAR␣ activators, including both foreign chemicals (nafenopin, a fibrate hypolipidemic drug) and the endogenous PPAR␣ activators (8S)-hydroxyeicosatetraenoic acid and leukotriene B 4 (Fig. 1C).
Requirement of GHR, JAK2, and STAT5b for GH Suppression of PPAR␣ Activity-Since some, but not all GH intracellular events involve JAK/STAT signaling pathways (30), we examined whether GHR, JAK2, and STAT5b are each required for the inhibitory effect of GH on PPAR␣ activity. Fig. 2A shows that inhibition of PPAR␣ activity by 25 ng/ml GH (lanes 3 versus lane 2) was not observed in the absence of cotransfected GHR (lane 4) or STAT5b (lane 5). A similar dependence on GHR and STAT5b was observed at higher GH concentrations (100 -500 ng/ml GH; data not shown). In the absence of transfected JAK2, GH could still activate STAT5b and inhibit PPAR␣ activity using the low levels of JAK2 expressed endogenously in COS-1 cells, although a higher GH concentration (500 ng/ml) was required for maximal PPAR␣ inhibition (data not shown). Interestingly, cotransfection of GHR, JAK2, and STAT5b significantly reduced the moderate PPAR␣-dependent but Wy-14,643-independent basal luciferase reporter activity, even in the absence of GH ( Fig. 2A, lanes 6 versus 1). This Wy-14,643-independent activity has been associated with the presence of endogenous PPAR activators, such as fatty acids (4,5). Inhibition of this basal PPAR activity likely results from the constitutive activation of STAT5b upon overexpression of JAK2, as shown in the gel-shift studies described below (Fig.  2B). GH stimulation further decreased the Wy-14,643-independent PPAR␣ activity up to 4-fold (data not shown).
To confirm the reconstitution of STAT5b activity in the GHstimulated COS-1 cells, EMSA assays were carried out using a ␤-casein promoter probe, which contains a single STAT5bbinding site. No STAT5b DNA binding was detected upon GH stimulation of untransfected COS-1 cells (Fig. 2B, lanes 1 and  2). This indicates that COS-1 cells have at most low levels of GHR and/or STAT5b and are thus suitable for use as recipient cells in these transfection studies. In cells cotransfected with GHR, JAK2, and STAT5b, a low basal STAT5b DNA binding was observed in the absence of GH; this STAT5b activity was not affected by Wy-14,643 treatment (Fig. 2B, lanes 3 versus 5). Thus, overexpression of JAK2 and STAT5b results in some constitutive activation of STAT5b DNA binding activity. Stimulation with GH yielded a much higher level of STAT5b DNA binding activity (lane 7 versus 5). The ␤-casein DNA-binding complexes obtained in these experiments were confirmed to contain STAT5b, as shown by supershift analysis using anti-STAT5b antibody (lanes 4, 6, and 8).
Inhibition of PPAR␣ Activity by GH-activated STAT3 but Not STAT1-In addition to STAT5b, GH can activate three other STAT proteins, STATs 1, 3, and 5a (32-35). Fig. 3 shows, however, that in COS-1 cells transfected with STAT1, GH did not decrease PPAR␣-induced luciferase reporter activity, even at a high GH concentration (500 ng/ml). Partial GH inhibition of PPAR␣ activity was seen in COS-1 cells transfected with STAT3 or with the closely related (41, 51) STAT5a. In control experiments, all four STATs were shown by Western blotting to be highly expressed in the transfected COS-1 cells compared with untransfected controls (Fig. 3B; lanes 2-4 versus lane 1). The lower effectiveness of STAT5a compared with STAT5b is not because of its inefficient activation by GH, as shown by EMSA using 32 P-labeled ␤-casein probe (Fig. 3C, lanes 3 and 4  versus lanes 6 and 7). STAT1 and STAT4 expressed in GHtreated cells were also shown to be functionally active in DNA binding using an SIE (sis-inducible element) EMSA probe (see "Materials and Methods") (data not shown).
Prolactin Inhibition of PPAR␣ Activity-Prolactin is a GHrelated pituitary polypeptide hormone that also signals via its cell-surface receptor through a JAK2/STAT5 pathway (42). STAT5a, originally identified as a prolactin-activated mammary gland factor, and STAT5b can both confer a prolactin response to the mammary ␤-casein gene promoter (42,51). Fig.  4 shows that in the presence of prolactin, STAT5b, but not STAT5a, inhibits PPAR␣ activation in cells cotransfected with prolactin receptor and JAK2. This inhibition by prolactin is less extensive (ϳ40%) than seen for GH-activated STAT5b in the same system (ϳ80%; cf., Fig. 3A) and also required a higher concentration of hormone to achieve its maximum effect (100 ng/ml prolactin versus 25 ng/ml GH). STAT5b activation by prolactin was confirmed by gel-shift assay using a ␤-casein probe (data not shown).
STAT5b Inhibition of PPAR␣ Activity Is Not Mediated by the MAP Kinases ERK1 and ERK2-We next investigated whether MAP kinase plays a role in the GH-induced down-regulation of PPAR␣ activity. GH can activate the MAP kinase kinase MEK1, which phosphorylates and thereby activates the MAP kinases ERK1 and ERK2 (53). MAP kinases are actively involved in growth factor and cytokine receptor signaling (54,55) and catalyze phosphorylation on serine and/or threonine of several transcription factors, including PPAR␥ (56). By analogy to the MAP kinase regulation of PPAR␥, PPAR␣, which is also a phosphoprotein (57), might be regulated by MAP kinasecatalyzed serine phosphorylation. To investigate whether MAP kinase activity is required for GH inhibition of PPAR␣ activity, we used the MEK1 and MEK2 inhibitor PD98059 (58) to block MAP kinase activation. Stimulation of transfected COS-1 cells with Wy-14,643 in combination with PD98059 increased PPREdependent luciferase activity compared with Wy-14,643 treatment alone (Fig. 6). The enhancement by PD98059 of PPRE luciferase activity varied from ϳ1.5to 4-fold in different experiments. However, GH treatment inhibited Wy-14,643-induced reporter activity to a similar extent in the presence and in the absence of PD98059 (Fig. 6). Since PD98059 did not block the GH inhibitory effect, the MAP kinases ERK1 and ERK2 are not likely to mediate GH inhibition of PPAR␣ activity. PD98059 treatment also enhanced Wy-14,643-induced PPAR␣ activity in the absence of GHR, JAK2, and STAT5b (data not shown). This suggests that basal MAP kinase activity in these cells exerts a negative effect on PPAR␣ activity but in a manner that is independent of GHR or STAT5b expression or GH treatment.
Tyr-699 Phosphorylation of STAT5b Is Required for GH In-hibitionofPPAR␣Activity-STAT5bisactivatedbyJAK2kinasedependent phosphorylation on tyrosine 699. Mutation of this tyrosine to phenylalanine (STAT5b-Y699F) results in a loss of STAT5b dimerization, DNA binding, and transcriptional activation (44). Fig. 7 shows that, when activated by GH, wild-type human STAT5b inhibited PPAR␣ activity to a similar extent as did mouse STAT5b. By contrast, the Y699F substitution abrogated GH inhibition of PPAR␣ activity. Accordingly, STAT5b and GH (100 ng/ml) for 24 h, as indicated, followed by preparation of whole cell extracts. EMSA assays were carried out with a 32 P-labeled ␤-casein promoter DNA probe, which contains a STAT5b-binding site. The specific STAT5b-DNA complex is marked 5b. Anti-STAT5b-specific antibody was used to supershift STAT5b-containing DNA-protein complexes (STAT5b SShift), as indicated. Nonspecific protein binding to the ␤-casein probe is labeled ns. tyrosine phosphorylation and/or its downstream activities (STAT5b dimerization, nuclear translocation, DNA binding, or transcriptional activation) are obligatory for STAT5b to mediate GH inhibition of PPAR␣. When equal amounts of human STAT5a expression plasmid were transfected, it failed to inhibit PPAR␣ activity (Fig. 7). Partial inhibition (ϳ40%) was observed, however, at 4-fold higher human STAT5a plasmid levels (data not shown).

Requirement of STAT5b COOH-terminal trans-Activation Domain and Effects of Dominant-negative STAT5b Mutant-
Deletions of the COOH-terminal trans-activation domain of mouse STAT5a and STAT5b (constructs STAT5a⌬749 and STAT5b⌬754, respectively) result in a loss of transcriptional activity and yield truncated STAT5 proteins that exert dominant-negative effects on wild-type STAT5-induced gene transcription (43). These COOH-terminal truncated STAT5 proteins undergo hormone-induced tyrosine phosphorylation and retain DNA binding activity but show delayed tyrosine dephos-phorylation (43). These mutants were used to examine whether the transcriptional activity of STAT5b is necessary for GH inhibition of PPAR␣ activity. Fig. 8A shows that, in contrast to wild-type STAT5b, STAT5b⌬754 cannot mediate GH-induced PPAR␣ inhibition, suggesting that the trans-activation domain of STAT5b is required for GH inhibitory effects. In contrast, STAT5a⌬749 inhibited PPAR␣ activity by ϳ40% following GH stimulation.
To test for dominant-negative activity, COS-1 cells were transfected with wild-type STAT5b in the presence of increasing amounts of STAT5a⌬749 or STAT5b⌬754. Fig. 8B shows that STAT5b⌬754 could fully block the STAT5b-dependent GH inhibition of PPAR␣ activity in a dose-dependent manner (lane 2 versus lanes 3 and 4). Interestingly, while STAT5a⌬749 has dominant-negative activity toward wild-type STAT5b and can block its transcriptional activity (43), this mutant had only a modest effect on STAT5b-dependent PPAR␣ inhibition (lane 2 versus lanes 5 and 6). However, interpretation of this result is  (20 M) and GH at the indicated concentrations. Whole cell extracts were prepared, and luciferase activities relative to a ␤-galactosidase transfection control were determined. B, expression of STAT proteins in transiently transfected COS-1 cells. The same cell lysates assayed in A were analyzed for STAT protein expression by Western blot using each of the indicated STAT form-specific antibodies. Small amounts of STAT1 and STAT5a are seen to be present in untransfected COS-1 cell extracts (lane 1). C, GH activates STAT5a and STAT5b DNA binding activity in transfected COS-1 cells. COS-1 cells were transfected with expression plasmids encoding GHR, STAT5a or STAT5b, and PPAR␣. Cells were treated with GH (500 ng/ml) for 30 min, and cell extracts were assayed for EMSA activity using the ␤-casein probe. Supershift analysis was carried out using anti-STAT5a (Santa Cruz, sc-1081x) or anti-STAT5b antibody (Santa Cruz sc-835), as indicated. GH-activated STAT5a (lane 3) migrates more slowly than STAT5b (lane 6). ns, nonspecific protein-DNA complex. complicated by the fact that STAT5a⌬749 itself confers partial inhibition of PPAR␣ in response to GH stimulation (Fig. 8A). Fig. 8C confirms the expression of STAT5a⌬749, STAT5b⌬754, and wild-type STAT5 proteins in the transfected COS-1 cells, detected with an antibody against the STAT5 SH2 and SH3 domains, which are upstream of the deleted sequences. As expected, the STAT5b⌬754 and STAT5a⌬749 bands seen on this blot were of lower molecular weight (lanes 4 and 7) and could not be detected using an antibody that specifically recognizes the extreme COOH-terminal region of STAT5b (Santa Cruz antibody sc-835; data not shown). When equal amounts of expression plasmid were transfected, STAT5a⌬749 protein was expressed at a level similar to wild-type STAT5b (lane 7 versus  lanes 2 and 3). The level of STAT5b⌬754 protein (lane 4) was much lower (Fig. 8C), highlighting the potency of its dominantnegative effects (Fig. 8B). EMSA analysis showed that STAT5a⌬749 and STAT5b⌬754 both bind to a ␤-casein DNA probe much more efficiently than wild-type STAT5b (Fig. 8D; 3,8,9,11, and 12 versus lanes 5 and 6), in agreement with earlier studies (43). This suggests that these mutants achieve their dominant-negative effect, at least in part, by interfering with wild-type STAT5b DNA binding activity. As expected, antibody specific for the STAT5b COOH-terminal peptide can completely supershift wild-type STAT5b but not the much more intense STAT5b⌬754 binding to the ␤-casein probe (Fig. 8D, lanes 7 versus 13).
STAT5b DNA Binding Activity Is Required for Inhibition of PPAR␣ Activity-STAT5a interacts with glucocorticoid receptor through formation of a protein-protein complex that inhibits glucocorticoid-induced gene expression (59,60). STAT5b inhibition of the stimulatory effects of prolactin at the IRF-1 promoter is also proposed to involve direct protein-protein interactions (45). We therefore considered whether a direct interaction between GH-activated STAT5b and PPAR␣ can be detected using an EMSA assay for PPAR␣-PPRE protein-DNA complexes. Fig. 9 shows that anti-STAT5b antibody does not alter the mobility of a PPAR␣-containing DNA complex formed on a PPRE probe from the CYP4A6 gene (lanes 3 versus 2) under conditions where the antibody fully supershifts STAT5bcontaining DNA complexes (Figs. 2B and 3C). These data indicate that STAT5b does not bind directly to a PPAR␣-DNA complex. In control samples, the PPAR-PPRE complex seen in lane 2 was either disrupted or supershifted by various anti-PPAR␣ antibodies (lanes 4 and 5), competed by an excess of unlabeled PPRE probe (lane 6) and was fully dependent on transfected PPAR␣ for its formation (lane 7).
We next investigated whether STAT5b DNA binding activity is required to inhibit PPAR␣ activity. The effects of two rat STAT5b mutants, STAT5b-EE and STAT5b-VVVI (45), were compared with that of wild-type rat STAT5b. In these mutants, amino acid residues EE (437-438) and VVVI (466 -469) within the VTEE and SLPVVVI sequences of the DNA binding domain of STAT5b are replaced by alanine. The STAT5b-VVVI mutant is devoid of DNA binding activity and is transcriptionally inactive following prolactin stimulation, whereas the STAT5b-EE mutation does not abolish STAT5b DNA binding activity or prolactin-induced transcription from the ␤-casein promoter (45). STAT5b-EE and STAT5b-VVVI were therefore tested for their ability to inhibit PPAR␣ activity in GH-treated COS-1 cells. Fig. 10 shows that wild-type rat STAT5b and rat STAT5b-EE both inhibit PPAR␣ activity in GH-treated cells in a manner similar to the effects of mouse and human STAT5b shown above. By contrast, much less inhibition is seen with STAT5b-VVVI. Thus, the VVVI residues in the DNA binding domain of STAT5b are critical for GH-activated STAT5b to efficiently inhibit PPAR␣ activity, strongly suggesting that this inhibition requires STAT5b DNA binding and transcriptional activation activity. DISCUSSION GH and several other hormones, including thyroid hormone (triiodothyronine) and glucocorticoids, modulate the pleiotropic responses of rodent liver to structurally diverse peroxisome proliferators mediated by the nuclear receptor PPAR␣. The present studies demonstrate that the GH-activated transcription factor STAT5b, and to a lesser extent STAT3, can mediate the inhibitory effects of GH on PPAR␣ transcriptional activity and that these effects occur at a physiological GH level. This specific inhibitory effect on PPAR␣ activity was seen both with rodent and human STAT5b proteins and occurred when STAT5b was activated by either GH or prolactin. Given that STAT5b can also be activated by a large number of cytokines and growth factors, and is widely expressed in mammalian tissues (41,51), the potential for inhibitory interactions between PPAR␣ and JAK-STAT5b signaling pathways is widespread. Although STAT5a (mammary gland factor) and STAT5b show ϳ90% amino acid sequence identity, and both Cell extracts were then prepared and assayed for firefly luciferase activity relative to a Renilla luciferase transfection control. ⌬STAT5a, but not ⌬STAT5b, partially inhibited Wy-14,643-induced reporter gene activity in GH-treated cells. B, STAT5b⌬754 blocks wild-type STAT5bdependent inhibition of PPAR␣ activity. COS-1 cells were transfected with pHD(X3)Luc reporter and PPAR␣, GHR, JAK2, and mouse STAT5b expression plasmids together with increasing amounts of ⌬STAT5b or ⌬STAT5a expression plasmids (0-, 0.5-, 1-, and 5-fold relative to the amount of cotransfected wild-type STAT5b, calculated on a per microgram plasmid DNA basis). Cells were stimulated with Wy-14,643 (20 M) and GH (100 ng/ml) for 24 h. Luciferase activities relative to a ␤-galactosidase transfection control were then assayed in whole cell extracts. C, expression of STAT5a, STAT5b, and their COOH-terminal truncated mutants. The same cell lysates analyzed in B were analyzed for STAT5 protein expression by anti-STAT5 Western blotting (Transduction Laboratories, antibody S21520). The COOHterminal truncated ⌬STAT5b and ⌬STAT5a bands were not detectable bind to and trans-activate target genes via STAT5 response elements such as that found in the ␤-casein promoter, they have distinct tissue-specific functions that cannot be compensated by each other in vivo, as shown in the case of mouse STAT5 gene knock-out studies (61,62). The present study provides further evidence for differences between the two STAT5 forms, insofar as STAT5a was less effective at inhibiting PPAR␣ activity, particularly when activated by prolactin (Fig. 4).
Functional interactions involving the binding of STAT factors and another nuclear receptor family member, glucocorticoid receptor, have been described (59,60,63). STAT5a and glucocorticoid receptor form a molecular complex that enhances STAT5a-induced transcription from the ␤-casein promoter and inhibits glucocorticoid-stimulated transcription from a glucocorticoid response element. Specific DNA binding by STAT5a is required for cooperation with glucocorticoid receptor on the ␤-casein promoter, and although the receptor does not bind the STAT5a DNA-binding element directly, it associates with the STAT5a-DNA complex. A potential glucocorticoid receptor-binding site is present within the ␤-casein promoter, and this site is required for the synergism between STAT5a and glucocorticoid receptor (64). These studies establish a model in which STAT proteins cross-talk with nuclear receptors by direct protein-protein interactions that modulate gene expression. By contrast, we were unable to detect molecular complexes involving both STAT5b and PPAR␣ in the present study, as judged by EMSA supershift analysis. Transcriptional inhibitory effects of prolactin-activated STAT5b have also been observed in studies of the full-length IRF-1 (interferon regulatory factor-1) promoter but not with an isolated IRF-1 STAT response element (45). In contrast to our findings with PPAR␣, the prolactin inhibitory effects seen on the IRF-1 promoter are conferred equally well by STAT5a or STAT5b. Moreover, unlike our findings in the present study, the effects of STAT5b on the IRF-1 promoter do not require the DNA binding activity of STAT and are proposed to involve STAT5b in protein-protein interactions with other transcription factors (45).
GH activates the MAP kinase pathway in many cell types, including liver cells (32,53,65). MAP kinase can, in turn, phosphorylate and thereby modulate the activity of a variety of transcription factors, including STAT proteins, which may undergo MAP kinase-catalyzed serine phosphorylation required for full transcriptional activity (66,67). Growth factor-activated MAP kinase phosphorylates PPAR␥, resulting in an inhibition of the transcriptional activity of that nuclear receptor (56,68). In our experiments, the MAP kinase kinase inhibitor PD98059 increased the responsiveness of COS-1 cells to PPAR␣ activators, suggesting that MAP kinase phosphorylates and thereby inhibits PPAR␣ by a mechanism similar to that described for PPAR␥ (56,68). However, PD98059 did not alter the extent to which GH-activated STAT5b inhibited PPAR␣ activity. Several mechanisms may account for the inhibition of PPAR␣ by GH-activated STAT5b described in the present study. First, activated STAT5b could inhibit PPAR␣ protein expression and thereby block PPAR activation of PPRE. However, this is considered unlikely, since GH-activated STAT5b had no effect on expression of the internal control ␤-galactosidase expression plasmid, which utilizes the same cytomegalovirus promoter as does the PPAR␣ expression plasmid used in our experiments. Second, STAT5b might compete with PPAR␣ for binding to the PPAR reporter PPRE elements of plasmid. This is also unlikely, since the pHD(X3)Luc reporter used in this study does not contain STAT5b-binding sites. Moreover, gel-shift assays revealed that STAT5b does not bind to an isolated PPRE (Fig. 9, lane 7). Third, by analogy to the interaction of STAT5a with glucocorticoid receptor discussed above, STAT5b may form a protein-protein complex with PPAR␣ and thereby prevent PPAR␣ from trans-activating its target genes. However, the inability of a STAT5b-specific antibody to supershift activated PPAR␣ when bound on a PPRE element (Fig. 9) argues against a direct association of STAT5b with PPAR␣, indicating a mechanism distinct from the previously described STAT5a-glucocorticoid receptor association. We cannot rule out the possibility, however, that STAT5b might form a complex with PPAR␣ that is independent of its binding to PPRE and thus not detected in our experiments.
An alternative possibility is that STAT5b inhibits PPAR␣ activity by an indirect mechanism. For example, when present in the nucleus in a transcriptionally active state, STAT5b may compete for an essential coactivator of PPAR, such as SRC-1, CBP/p300, or PBP (69 -71), or perhaps modulate the binding of PPAR to other interacting proteins, such as RXR␣ or RIP140 (72), leading to inhibition of transcriptional activity of PPAR␣. Indeed, CBP and/or p300 have been implicated in the antagonism between STAT and AP-1 signaling pathways (73). Alternatively, given the requirement for intact, functional STAT5b DNA binding and transcriptional activation domains, STAT5b may activate transcription of a second gene, leading to production of a distinct protein factor that serves as a more proximal inhibitor of PPAR␣ activity. Both of these possibilities are consistent with our observation that a dominant-negative STAT5b mutant (STAT5b⌬754 (43)) blocks wild-type PPAR␣ inhibitory activity of STAT5b (Fig. 8). Of note, some caution about the interpretation of the effects of the STAT5b mutants employed in this study is required. For example, if mutation of the STAT5b DNA binding domain residues VVVI (45) were to interfere with STAT5b tyrosine phosphorylation and/or nuclear translocation, then a failure to translocate into the nucleus rather than the loss of DNA binding activity per se could account for the observed lack of PPAR␣ inhibition (Fig. 10). The precise mechanism underlying the dominant-negative effect of the COOH-terminal truncated STAT5b also needs to be elucidated. Further study will be required to test and evaluate these FIG. 10. DNA-binding activity is required for rat STAT5b to mediate GH inhibition of PPAR␣ activity. COS-1 cells were transiently transfected with pHD(X3)Luc reporter and expression plasmids encoding PPAR␣, GHR, JAK2 and rat STAT5b or either of two rat STAT5b site-specific mutants localized to STAT5b's DNA binding domain. STAT5b-EE retains DNA-binding activity, while STAT5b-VVVI is devoid of DNA-binding activity (45). Transfected cells were treated with Wy-14,643 (20 M) and GH (25 and 100 ng/ml) for 24 h. Cell extracts were assayed for firefly luciferase activity relative to renilla luciferase activity. and other possible inhibitory mechanisms.
The inhibition of PPAR␣ activity by activated STAT5b described in this study may have diverse physiological consequences. PPAR␣ is an important intracellular messenger that transmits pharmacological as well as nutritional and immunological stimuli to cells. PPAR␣ target genes are involved in fatty acid oxidation, transport, and synthesis in liver and other tissues, and PPAR␣ activators include certain steroids, fatty acids, and metabolites of arachidonic acid (12). STAT5b can be activated by many cytokines and growth factors in addition to GH and prolactin, including erythropoietin and interleukins 2, 3, and 5 (41,44). If, as seems likely, STAT5b inhibits PPAR␣ activity when activated by these other cytokines and growth factors, additional cytokine and hormonal effects on the regulation of lipid metabolism and the degradation of lipid signaling molecules are also possible.
Leukotriene B 4 , a mediator of certain inflammatory and immunological reactions, has been identified as an endogenous ligand for PPAR␣ (7). Peroxisome proliferators and PPAR␣ activators, such as clofibrate, are reported to have both inductive (74,75) and suppressive effects (76) on the enzymes involved in leukotriene B 4 -hydroxylation, a reaction that deactivates this potent chemotactic agent and thereby shortens the duration of an inflammatory response. The inhibition of PPAR␣ activity by STAT5b activators may therefore provide a mechanism through which STAT5b-activating cytokines modulate leukotriene B 4 -induced inflammatory responses. The pituitary hormones GH and prolactin can have a marked influence on immune cell types (77,78). Leukocytes express receptors for GH and prolactin (77) and have an intact JAK/ STAT pathway (79), indicating that these polypeptide hormones have the potential to modulate inflammatory responses by suppression of PPAR␣-regulated leukotriene B 4 degradation. Inflammatory cells also have the capacity to synthesize and secrete GH (80), which could provide for fine adjustment of the leukotriene B 4 response in immune cells.
The present demonstration that GH-activated STAT5b can inhibit PPAR␣-dependent transcriptional responses provides a mechanistic basis for the observed inhibitory effects of GH on peroxisomal enzyme induction (28,29). Additional mechanisms may also be operative, however, as suggested by the longer term down-regulation of PPAR␣ mRNA levels seen in GHtreated liver cells (81). Foreign chemical peroxisome proliferators, including chlorinated hydrocarbons and their metabolites, induce hepatocarcinogenesis via PPAR␣-dependent non-genotoxic mechanisms (18,19). Increased oxidative stress and proto-oncogene induction (82) are both postulated to contribute to the carcinogenicity of these peroxisome proliferators; however, the precise mechanism remains unclear. PPAR␣ plays a critical role in tumor development in mice in response to the foreign peroxisome proliferator Wy-14,643, and targeted disruption of the PPAR␣ gene results in loss of this carcinogenic response (18). GH suppression of PPAR␣ activity may therefore inhibit tumor development, suggesting a novel mechanism whereby endogenous hormones beneficially modulate cellular responses to non-genotoxic chemical carcinogens.