Growth hormone activation of Stat 1, Stat 3, and Stat 5 in rat liver. Differential kinetics of hormone desensitization and growth hormone stimulation of both tyrosine phosphorylation and serine/threonine phosphorylation.

Intermittent plasma growth hormone (GH) pulses, which occur in male but not female rats, activate liver Stat 5 by a mechanism that involves tyrosine phosphorylation and nuclear translocation of this latent cytoplasmic transcription factor (Waxman, D. J., Ram, P. A., Park, S. H., and Choi, H. K.(1995) J. Biol. Chem. 270, 13262-13270). We demonstrate that physiological levels of GH can also activate Stat 1 and Stat 3 in liver tissue, but with a dependence on the dose of GH and its temporal plasma profile that is distinct from Stat 5 and with a striking desensitization following a single hormone pulse that is not observed with liver Stat 5. GH activation of the two groups of Stats leads to their selective binding to DNA response elements upstream of the c-fos gene (c-sis-inducible enhancer element; Stat 1 and Stat 3 binding) and the β-casein gene (mammary gland factor element; liver Stat 5 binding). In addition to tyrosine phosphorylation, GH is shown to stimulate phosphorylation of these Stats on serine or threonine in a manner that either enhances (Stat 1 and Stat 3) or substantially alters (liver Stat 5) the binding of each Stat to its cognate DNA response element. These findings establish the occurrence of multiple, Stat-dependent GH signaling pathways in liver cells that can target distinct genes and thereby contribute to the diverse effects that GH and its sexually dimorphic plasma profile have on liver gene expression.

Growth hormone (GH) 1 regulates a large number of metabolic and other processes in the liver, primary through its effects on gene transcription. GH exerts both stimulatory and inhibitory effects on the expression of a wide range of liver gene products, including cytochrome P450 (1,2), glutathione S-transferase (3), and sulfotransferase enzymes involved in steroid and drug metabolism (4), in addition to several cellsurface receptors (5,6), including GH receptor (7). Hepatic secretory products, such as insulin-like growth factor 1, serine protease inhibitor Spi 2.1 (8), and various urinary proteins (9,10), are also expressed in a GH-dependent manner. Studies carried out in rodents demonstrate that many, although not all, of the effects of GH on liver gene expression are sex-dependent. This sex dependence is a direct consequence of a striking differential response of individual target genes to GH, depending on whether hepatocytes are stimulated by the intermittent plasma pulses of GH that are characteristic of adult male rats or whether the cells are exposed to GH on a more continual basis, as occurs in adult female rats. Prototypic examples of this sex-dependent GH regulation are the cytochrome P450 genes CYP2C11 and CYP2C12, which are transcribed in a male-and female-specific manner, respectively, in adult rat liver in direct dependence on the plasma GH profile (11,12). By contrast, other effects of GH, in particular the acute stimulation of insulin-like growth factor 1 (13), c-fos (14), and serine protease inhibitor (Spi 2.1) gene expression in liver (8), exhibit little or no sex dependence or responsiveness to the temporal pattern of circulating GH.
Tyrosine phosphorylation is an early response to GH that has been documented in several cell types. GH induces the tyrosine phosphorylation of multiple cellular polypeptides (15), including Jak2 kinase, a GH receptor-associated tyrosine kinase that undergoes autophosphorylation following GH stimulation, and GH receptor, which becomes a substrate for Jak2 following GH-induced receptor dimerization (16,17). Secondary events include GH-stimulated tyrosine phosphorylation of a number of intracellular signaling molecules, notably Stat 5 (18 -20), insulin receptor substrate-1 (21), SHC (22), and mitogen-activated protein kinase (23,24). Several of these latter signaling events are likely to involve interaction of the tyrosine-phosphorylated GH receptor-Jak2 kinase complex with these signaling molecules via SH2 domain interactions (25) or perhaps via phosphotyrosine-interacting domains (26,27).
Stat 5, originally cloned from sheep mammary gland (28,29), belongs to the Stat family of transcription factors; these factors serve as signal transducers and activators of transcription for numerous cytokines and growth factors. In response to cytokine treatment of target cells, Stats undergo tyrosine phosphorylation, homo-or heterodimerization, and nuclear translocation, followed by DNA binding and transcriptional activation of target genes (30,31). Studies on the effects of GH on liver Stat 5, carried out in an adult rat model in vivo, have established that this GH-activated Stat is uniquely responsive to the temporal pattern of plasma GH stimulation: intermittent plasma GH pulses, such as those that occur naturally in adult male rats, trigger rapid and repeated tyrosine phosphorylation and nuclear translocation of liver Stat 5, while continuous plasma GH exposure, as occurs in adult female rats, leads to desensitization of this tyrosine phosphorylation pathway and, consequently, a low steady-state level of the active nuclear Stat 5 transcription factor (18). This pattern of response suggests that liver Stat 5 may be an important intracellular mediator of the stimulatory effects of GH pulses on male-specific liver gene transcription.
While liver Stat 5 appears to be the major Stat protein that is tyrosine-phosphorylated following stimulation of hepatocytes by a physiological pulse of GH (18), other liver-expressed Stat proteins may also be activated by GH. This is suggested by the finding that human GH can stimulate tyrosine phosphorylation of Stat 1 and Stat 3, both in 3T3-F442 fibroblasts (32,33) and in livers of hypophysectomized rats (34,35). The relevance of these findings with respect to the somatotropic effects of GH is uncertain, however, given the binding of human GH to both prolactin receptors and GH receptors (36,37) and the use in the hypophysectomized rat studies of a dose of human GH that is supraphysiologic (34,35), being at least 50-fold higher than the dose required to activate liver Stat 5 (18) or to stimulate pulsatile GH-dependent CYP2C11 gene expression in the same rat liver model (38). It is also not known whether Stat 1 and Stat 3 respond to the temporal pattern of GH stimulation in a sex-specific manner, as does liver Stat 5. These questions are addressed in this study, where we characterize the activation of rat liver Stat 1 and Stat 3, in comparison to liver Stat 5, in response to physiological GH pulses given in vivo. In addition to tyrosine phosphorylation, we report a GH-induced Stat phosphorylation event that is distinct from the tyrosine phosphorylation step catalyzed by Jak2 kinase and involves either serine or threonine phosphorylation, leading to a significant enhancement (Stat 3 and Stat 1) or modulation (liver Stat 5) of the specific DNA binding activity of these Stat transcription factors.

MATERIALS AND METHODS
Animals-Male and female Fischer 344 rats, untreated or hypophysectomized at 8 weeks of age, were purchased from Taconic Farms Inc. (Germantown, NY) and were maintained on a 12-h light-dark cycle with free access to food and drinking water. The completeness of hypophysectomy was confirmed by the lack of weight gain over a 2-4-week period following surgery. Hypophysectomized rats were given rat GH (NIDDK, rGH-B-14-SIAFP; 1, 3, or 12.5 g/100 g of body weight) by intraperitoneal injection, as indicated in individual experiments. All experiments used rat GH, unless specifically noted that human GH (NIDDK, hGH-B-1; 12.5 or 150 g/100 g of body weight) was used. The 150 g of human GH/100 g of body weight dose corresponds to the supraphysiologic hormone replacement protocol used by Gronowski and Rotwein (34). GH-treated animals were sacrificed at time intervals ranging from 5 min to 4 h following hormone injection, unless indicated otherwise. Rat prolactin (NIDDK, rPRL-B-8-SIAFP) was given intraperitoneally at 12.5 g/100 g of body weight, and the animals were killed 45 min later. GH and prolactin were obtained from the National Hormone and Pituitary Program, NIDDK. Escherichia coli LPS (serotype 026:B6; Sigma L-2762) was administered to hypophysectomized rats by intraperitoneal injection at 1 mg/150 g of body weight to promote cytokine release, as described (39). Animals were killed 75 min later.
Nuclear Protein Extraction-Rat liver nuclear proteins were prepared from intact, hypophysectomized, and GH-treated rats by the method described in Ref. 40 as modified (18). Nuclear extracts were aliquoted and immediately frozen in liquid nitrogen for storage. Cytosolic and microsomal fractions were prepared from the same liver samples using methods already described (18). In some cases, the final dialysis step was carried out in the absence of phosphatase inhibitors to facilitate an analysis of the effects of phosphatase treatment (see Figs. 6B, 7, and 8).
Immunoprecipitation and Western Blotting-Immunoprecipitation with anti-phosphotyrosine monoclonal antibody 4G10 (Upstate Biotechnology, Inc., Lake Placid, NY) or with monoclonal antibody PY20agarose conjugate (Transduction Labs, Lexington, KY) and Western immunoblotting were carried out essentially as described (18). Immunoblots were probed with anti-phosphotyrosine monoclonal antibody 4G10 or with mouse monoclonal anti-Stat antibodies (Transduction Labs). Anti-Stat 1 (S21120) was raised to amino acids 592-731 of human Stat 1, anti-Stat 3 (S21320) to amino acids 1-178 of rat Stat 3, anti-Stat 5 (S21520, lot 3) to amino acids 451-649 of sheep Stat 5, and anti-Stat 6 (S25420) to amino acids 1-272 of human IL-4-activated Stat. The anti-Stat 5 antibody was shown to react with mouse Stat 5a and to cross-react partially with mouse Stat 5b when the latter two cDNAs, kindly provided by Dr. Alice Mui (41), were transiently expressed in COS-1 cells. By contrast, anti-Stat 5 antibody C-17 (Santa Cruz sc-835x), used for the gel mobility supershift analysis described below, was shown to be specific for mouse Stat 5b. 2 Protein samples were fractionated through 10% polyacrylamide gels (10 -25 g of nuclear extract protein/lane), and the proteins were electrophoretically transferred to nitrocellulose overnight at 4°C. Blots were preincubated for 1 h at 37°C with TBS blocking buffer (0.1% Tween 20, 3% bovine serum albumin, 0.9% NaCl) and then incubated for an additional 1 h at room temperature in TBS containing monoclonal anti-phosphotyrosine antibody (1:3000) or the specific anti-Stat antibodies (1:2000) described above. After washing 3 ϫ 5 min in TST buffer (10 mM Tris-Cl (pH 7.5), 0.1 M NaCl, 0.1% Tween 20), blots were incubated at room temperature for 1 h with anti-mouse IgG conjugated to horseradish peroxidase (1:3000; Amersham Corp.) and then washed in a high Tween buffer (0.3% Tween 20, 10 mM potassium P i (pH 7.4), 0.9% NaCl; 1 ϫ 15 min) followed by a TST wash (1 ϫ 15 min). Antibody binding was detected on x-ray film by enhanced chemiluminescence using the ECL kit from Amersham Corp. To reprobe with another primary antibody, nitrocellulose blots were stripped by incubation in 2% sodium dodecyl sulfate, 100 mM 2-mercaptoethanol, and 62.5 mM Tris-HCl (pH 6.7), for 30 min at 50°C and then rinsed 3 ϫ 10 min in TST before reblocking in TBS and reprobing. Results presented in the individual figures are based on grayscale scans of portions of the x-ray films of each blot. Scans were obtained using a Cannon IX-4015 scanner and Ofoto scanning software.
Gel Mobility Shift Analysis-Liver nuclear protein, 5 g dissolved in 5 l of NED (nuclear extract dialysis) buffer (25 mM Hepes (pH 7.6 at 4°C), 40 mM KCl, 0.5 mM phenylmethanesulfonyl fluoride, 0.1 mM EDTA, 1 mM dithiothreitol, 0.5 mM sodium fluoride, 1 mM sodium orthovanadate, 10% glycerol), was preincubated for 10 min at room temperature with 9 l of gel mobility shift buffer (10 mM Tris-HCl (pH 7.5), 2 g of poly(dI-dC), 4% glycerol, 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl). Double-stranded oligonucleotide probe (1 l, 10 fmol), 32 P-end-labeled on one strand using T4 polynucleotide kinase, was then added and further incubated for 30 min at room temperature. An additional 10-min incubation was carried out in the presence of anti-Stat antibody prior to the addition of the 32 P-labeled DNA probe for the gel mobility supershift analysis. Mouse monoclonal anti-ISGF3 antibody (Transduction Labs G16920), rabbit polyclonal anti-Stat 3 antibody C-20 (Santa Cruz sc-482x), and anti-Stat 5 antibody C-17 were used for Stat 1, Stat 3, and Stat 5 supershift analysis, respectively. Where indicated, anti-Stat 3C antibody, kindly provided by Drs. Z. Zhong and J. Darnell, Jr. (Rockefeller University) (42), was used for supershift analysis. This latter antibody supershifted SIE complexes formed by both Stat 3-Stat 3 homodimers and Stat 1-Stat 3 heterodimers, whereas anti-Stat 3 antibody C-20 preferentially supershifted the Stat 3-Stat 3 homodimers (see Fig. 3D). Samples were electrophoresed at room temperature through nondenaturing polyacrylamide gels (5.5% acrylamide, 0.07% bisacrylamide for the SIE probe; 4% acrylamide, 0.05% bisacrylamide for the ␤-casein probe; see below) in 0.5 ϫ TBE buffer (44.5 mM Tris, 44.5 mM boric acid, 5 mM EDTA (pH 8.0)) for 2.5 h at 100 V using a standard method after a 30-min preelectrophoresis step under identical conditions. In some cases, the electrophoresis time was increased to 3-3.5 h, resulting in the elution of unbound probe from the bottom of the gel, in order to increase the resolution of the gel shift complexes.
Phosphatase Treatment-Treatment of cytosolic and liver nuclear extracts with phosphoprotein phosphatases was carried out as follows. Protein samples (25 g in a total volume of 70 l) were incubated for 1 h at 37°C with the phosphoserine/phosphothreonine-specific phosphoprotein phosphatase 2A (PP2A) (0.8 units; Upstate Biotechnology 14-111) or 1 (PP1) (0.8 units; Upstate Biotechnology 14-110), with phosphotyrosine phosphatase 1B (PTP-1B) (10 l of enzyme containing 1 g of fusion protein-agarose bead conjugate; Upstate Biotechnology 14-109), or with the general phosphatase CIP (20 units;Promega M1821) in the presence of 25 mM Tris-Cl (pH 7.0), 50 M CaCl 2 containing 25 g of bovine serum albumin and 0.1 mM ZnCl 2 . Control samples included the specific phosphatase inhibitor okadaic acid (100 nM for PP2A and 1 M for PP1), sodium orthovanadate (0.1 mM for PTP-1B), or EDTA (0.1 mM for CIP). Following incubation with the phosphatases, phosphatase inhibitors were added to block further dephosphorylation during sample analysis. Western immunoblotting using anti-phosphotyrosine and anti-Stat antibodies and gel mobility assay for Stat-DNA complex activity using the SIE probe (Stat 1 and Stat 3) or a rat ␤-casein promoter probe (liver Stat 5) were then carried out.

GH-induced Stat 3 Phosphorylation-Intermittent plasma
GH pulses, which occur in adult male rats, stimulate liver Stat 5 tyrosine phosphorylation and nuclear translocation to high levels, while the more continuous plasma GH profile of adult female rats desensitizes hepatocytes, resulting in low Stat 5 tyrosine phosphorylation and little or no liver Stat 5 protein in the nucleus (18). To ascertain whether other Stat proteins are also activated by physiological pulses of GH, we examined the effects of GH treatment on nuclear levels of two other liverexpressed Stat proteins, Stat 1 and Stat 3. Western blot analysis of liver nuclear extracts revealed that a pulse of GH given to hypophysectomized rats stimulates an increase in nuclear Stat 1␣ (M r ϳ91,000) and Stat 1␤ (M r ϳ84,000) as well as Stat 3 (M r ϳ89,000) within 15 min following GH injection (Fig. 1). In some experiments, the increase in nuclear Stat 1␣ and Stat 1␤ (Fig. 1A) preceded the increases in nuclear Stat 3 (Fig. 1B) and liver Stat 5 (M r ϳ93,000) (Fig. 1C). In the case of Stat 3, a distinct up-shift in the protein's electrophoretic mobility was apparent by 15 min after GH treatment (lanes 8 and 9 versus lanes 6 and 7; note the appearance of the band marked 3(P)), suggesting that GH stimulates Stat 3 phosphorylation. A corresponding up-shift of Stat 1␣ could also be detected at 15 and 45 min; however, the mobility difference was too small to be seen in the blot shown in Fig. 1A. Stat 1 and Stat 3 were both present at significant basal levels in hypophysectomized rat liver nuclei, in contrast to liver Stat 5, which was not detectable in the nucleus in the absence of GH treatment (lane 6).
To determine whether GH induces tyrosine phosphorylation of Stat 1 or Stat 3, liver nuclear extracts were immunoprecipitated with anti-phosphotyrosine monoclonal antibody 4G10, and the immunoprecipitates were then analyzed on Western blots probed sequentially with anti-Stat 1, anti-Stat 3, and anti-Stat 5 antibodies. Fig. 1 (lanes 3 and 4) shows that GH strongly stimulates tyrosine phosphorylation of Stat 3 (panel B), in addition to liver Stat 5 tyrosine phosphorylation (panel C), as reported previously (18). By contrast, only a very low level of tyrosine-phosphorylated Stat 1 was detectable in liver nuclear extracts by anti-phosphotyrosine antibody 4G10 immunoprecipitation ( Fig. 1A; also see below). GH-induced Stat 3 tyrosine phosphorylation was apparent in both female (Fig. 1B) and male (Fig. 1D, lane 11) hypophysectomized rats. Stat 3 tyrosine phosphorylation was not stimulated by prolactin (lane 12), but was stimulated by bacterial LPS (lane 13), which may act by inducing the release of one or more Stat 3-activating cytokines (39). LPS did not induce tyrosine phosphorylation of liver Stat 5 (Fig. 1E, lane 13), despite the fact that mammary Stat 5 can be activated by multiple cytokines, including FIG. 1. GH-induced accumulation of tyrosine-phosphorylated Stats in rat liver nuclei. A-C, Western blot of liver nuclear extracts prepared from individual hypophysectomized (Hx) female rats treated with GH (3 g/100 g of body weight, intraperitoneally) and then sacrificed 0, 5, 15, or 45 min later, as indicated. The blot shown was probed sequentially with anti-Stat 1, anti-Stat 3, and anti-Stat 5 antibodies (A-C, respectively) followed by ECL detection. Lanes 5-9 correspond to nuclear extracts (NE), and lanes 1-4 are the corresponding immunoprecipitates (IP) obtained using anti-phosphotyrosine antibody 4G10. Stat 1␣ and Stat 1␤ were as marked (A). Anti-Stat 3 detected tyrosinephosphorylated and nonphosphorylated Stat 3 (bands marked 3(P) and 3, respectively) as well as an additional unidentified band (*) in the intact nuclear extracts, which may correspond to the recently described Stat 3␤ (71), whereas only band 3(P) was detected following immunoprecipitation (B). Several nonspecific bands seen on the anti-Stat 5 blot were removed by immunoprecipitation (C, lanes 5-9 versus lanes 1-4). D and E, Western blot of nuclear extracts prepared from hypophysectomized male (M) rats given the treatments indicated, analyzed either with (lanes 10 -13) or without (lanes 14 and 15) antibody 4G10 immunoprecipitation, and probed sequentially with anti-Stat 3 and anti-Stat 5, as indicated. GH treatment was for 45 min at 3 g/100 g of body weight. Treatments with prolactin (PRL) and LPS were as described under "Materials and Methods." Heavily stained bands seen for the immunoprecipitated samples at the bottom of each panel are Ig-derived. Lanes 5 and 14, intact female and male liver nuclear extracts, respectively. IL-3, IL-5, and GM-CSF, in addition to prolactin and GH (20,41,44).
GH Activates Liver Stat 1 at High Doses of Hormone-As noted above, only low levels of tyrosine-phosphorylated Stat 1 could be detected in our studies of hypophysectomized rats treated with a biologically effective replacement dose of rat GH (3 g/100 g of body weight). By contrast, GH-induced liver Stat 1 tyrosine phosphorylation was readily apparent in an earlier study when hypophysectomized rats were treated with human GH at very high levels (150 g GH/100 g of body weight) (34). To test whether the differences in hormone dose or species of hormone origin account for the apparent differences in the extent of Stat 1 activation, we examined the dose dependence of Stat 1 tyrosine phosphorylation induced by rat GH and human GH. Liver nuclear extracts prepared from GH-treated hypophysectomized rats were analyzed by Western blotting after immunoprecipitation with anti-phosphotyrosine antibody PY20, which in preliminary experiments was found to be more efficient for Stat 1 immunoprecipitation than antibody 4G10. Human GH induced an increase in tyrosine-phosphorylated Stat 1 that was substantially enhanced at high hormone doses as compared with rat GH given at a low or an intermediate dose (Fig. 2, lanes 5 and 6 versus lanes 3 and 4). By contrast, Stat 3 showed a more modest GH dose-dependent increase in tyrosine phosphorylation, and liver Stat 5 no increase at all over the 50-fold range of GH doses used in these experiments. Moreover, significant liver Stat 5 tyrosine phosphorylation was observed within 10 min in rats treated with a dose of rat GH as low as 1 g/100 g of body weight (data not shown). Conceivably, the differential dose-dependent and species-dependent effects of GH on Stat 1 versus Stat 3 and liver Stat 5 phosphorylation seen in these experiments may reflect differences between the Stats with respect to their kinetics of phosphorylation/dephosphorylation as influenced by the persistence of GH in circula-tion under the different hormone treatment protocols.

GH-induced Stat 3 and Stat 1 Phosphorylation Activates Factor Binding to c-fos Enhancer, but Not to ␤-Casein Promoter
Element-We next examined whether these effects of GH on Stat protein tyrosine phosphorylation are associated with enhanced binding to DNA response elements associated with GH-activated genes. Liver nuclear protein binding to a synthetic, double-stranded oligonucleotide containing a DNA response element derived from SIE of the c-fos gene (sequence m67) (45) was examined by a gel mobility shift assay. c-fos is an early response gene that is activated by various cytokines and growth factors, including GH (46,47). Tyrosine-phosphorylated Stat proteins bind to SIE and form three distinct gel mobility shift complexes; these have been identified as Stat 3-Stat 3 homodimers, Stat 1-Stat 3 heterodimers, and Stat 1-Stat 1 homodimers (SIE complexes A, B, and C, respectively) (42,48). Fig. 3A shows that rat GH rapidly activates liver nuclear protein binding to an SIE probe to give significant amounts of all three gel shift complexes. GH dose-response analysis (Fig.  3B) revealed an enhanced amount of SIE complex C, as compared with SIE complex A, in liver nuclear extracts from hypophysectomized rats treated with high levels of human GH (lane 6), which have comparatively high levels of tyrosinephosphorylated Stat 1 (cf. Fig. 2). By contrast, LPS, which activates Stat 3, but not Stat 1 ( Fig. 1D and data not shown), preferentially induced formation of SIE complex A (Fig. 3B,  lane 7).
The specificity of these protein-DNA complexes was verified by the inhibition of complex formation by a 50-fold molar excess of unlabeled SIE probe (Fig. 3, A, lanes 6 and 12; and C, lane 2), but not by a GAS/ISRE sequence (49) or by probes containing mutated SIE or GAS/ISRE core binding sequences (Fig. 3C, lanes [3][4][5]. Partial inhibition of SIE complex C was observed with a ␤-casein promoter probe (lane 6), which, in addition to binding Stat 5-related proteins activated by GH and prolactin (18,20), can also bind Stat 1 homodimers in samples containing high levels of activated Stat 1 (50,51). Partial inhibition of SIE complex C was also effected by a Stat 5-binding (19) Spi 2.1 GH response element probe (lane 7), suggesting that Stat 1 homodimers may also bind to the Spi 2.1 gene Stat response element. The same pattern of SIE complexes and competitor probe specificity seen with these GH-activated hypophysectomized rat liver nuclear extracts was also observed when using untreated adult male and adult female rat liver extracts, although the SIE complex signal intensities were much weaker in the latter two cases (Fig. 3A, lanes 1 and 7; also see Fig. 4D).
The presence of Stat 3 in the GH-activated SIE complex A gel shift band was verified by the ability of polyclonal antibody raised to a COOH-terminal peptide derived from mouse Stat 3 to form a gel "supershift" of SIE complex A (Fig. 3D, lane 3).  2) or from hypophysectomized male rats treated with rat GH (rGH) (lanes 3 and 4) or human GH (hGH) (lanes 5 and 6) at 3-150 g/100 g of body weight or with LPS (lane 7), as indicated, were immunoprecipitated (IP) with anti-phosphotyrosine antibody PY20. Samples were analyzed on a Western blot probed with the indicated anti-Stat antibodies. Lane 8, untreated male rat liver nuclear extract without immunoprecipitation, shown as a control. either the SIE probe or the ␤-casein probe. 3 Similar Levels of Liver Nuclear Stat 1 and Stat 3 in Males and Females-Since the tyrosine phosphorylation and nuclear accumulation of liver Stat 5 are sex-dependent and occur in intact adult male but not female rats in response to the inter- 3 Although Stat 1-Stat 1 homodimers have recently been shown to bind to the ␤-casein promoter probe in extracts containing high levels of activated Stat 1 (Stat 1-containing ␤-casein complex migrating distinctly more rapidly than the corresponding Stat 5-containing complex I) (50,51), no such complex could be detected in our studies using GH-activated liver nuclear extracts (cf. absence of anti-Stat 1 supershift (Fig. 3D, lane 12)). The partial inhibition of the Stat 1-containing SIE complex C by a high concentration of competitor ␤-casein probe (33 M) in these same nuclear extracts (Fig. 3C, lane 6) suggests that the ␤-casein Stat site binds Stat 1 homodimers with comparatively low affinity.
FIG. 3. Gel mobility shift analysis of GH-activated Stat 1 and Stat 3 using SIE probe. Liver nuclear extracts prepared from hypophysectomized (Hx) male (M) or female (F) rats treated with rat GH (rGH) or human GH (hGH), as indicated, were analyzed for DNA binding activity by gel shift using the SIE probe as described under "Materials and Methods." Gel shift complexes designated A, B, and C correspond to Stat 3 homodimers, Stat 1-Stat 3 heterodimers, and Stat 1 homodimers, respectively (see D). A, time course for GH induction of DNA binding activity in hypophysectomized male and female rats (3 g of GH/100 g of body weight, intraperitoneally). Lanes 6 and 12, samples as in lanes 5 and 11, respectively, plus a 50-fold molar excess of unlabeled SIE probe (self-competition (sc)). Hypophysectomized female samples are the same as shown in Fig. 1 (42), which gave a prominent supershift complex near the top of the gel (band ss3Ј). Lanes 9 -12, the ␤-casein probe yielded a major gel shift complex (I) and a minor one (II). Of the three antibodies tested, only anti-Stat 5 supershifted complex I (band ss5 (lane 10); also note the minor band above ss5, which may correspond to a supershift of ␤-casein complex II) (also see Fig. 8D). mittent plasma GH pulsation that is characteristic of the male (18), we examined whether there is a corresponding sex difference with respect to the levels of liver nuclear Stat 1 or Stat 3. Fig. 4 (A and B) shows that Stat 1 and Stat 3 are present at only slightly higher levels in male as compared with female rat liver nuclei. This is in striking contrast to liver nuclear Stat 5, which is strongly male-dominant (Fig. 4C). Correspondingly, intact male rat liver nuclear extracts exhibited Stat 1-and Stat 3-dependent SIE complex activity at a level that was very similar to that of intact female rat liver nuclear extracts (Fig. 4D, lane 1  versus lane 4). This further suggests that in contrast to liver Stat 5, the liver nuclear DNA binding activity of Stat 1 and Stat 3 is not strikingly dependent on the sex-specific temporal pattern of circulating GH.
Influence of Sequential Plasma GH Pulses on Stat 1 and Stat 3 Tyrosine Phosphorylation: GH Desensitization-While Stat 1 and Stat 3 proteins are present at substantial levels in liver nuclei, anti-phosphotyrosine immunoprecipitation analysis indicated the presence of only a low level of the corresponding tyrosine-phosphorylated Stat proteins in these untreated rats (cf. Fig. 2, A and B, lane 1 versus lane 8). By contrast, tyrosinephosphorylated liver Stat 5 was present at a high steady-state level in the male samples of these same nuclear samples (cf. Fig. 2C, lane 1). To determine why the basal level of tyrosinephosphorylated Stat 1 and Stat 3 in liver nuclei is low compared with that of liver Stat 5, we investigated the effect of GH pulsation on Stat protein tyrosine phosphorylation. These experiments were carried out in hypophysectomized rats treated with two successive pulses of GH, spaced 4 h apart, to mimic the physiological plasma GH pulsation that occurs in intact adult male rats. As shown in Fig. 5, the GH-induced increase in tyrosine phosphorylation of nuclear Stat 1, Stat 3, and liver Stat 5, readily evident 45 min after hormone injection (lanes 2 and 3 versus lane 1), decays substantially 4 h after hormone treatment (lanes 4 and 5). Moreover, although a second injection of GH at 4 h leads to strong rephosphorylation of liver Stat 5 (Fig. 5C, lanes 6 and 7), only a weak rephosphorylation of Stat 1 and Stat 3 was observed (Fig. 5, A and B, lanes 6 and 7). Gel mobility shift analysis confirmed the relative ineffectiveness of a second pulse of GH with respect to induction of SIE complex formation (Fig. 5D, lanes 6 and 7 versus lanes 4 and 5). Interestingly, whereas SIE complex B (Stat 1-Stat 3 heterodimer) was the most abundant of the three SIE complexes after the first GH injection (Fig. 5D, lanes 2 and 3), SIE complex A (Stat 3 heterodimer) dominated after the second GH injection ( lanes  6 and 7). Thus, Stat 1 and Stat 3, but not liver Stat 5, become  1 and 2 (B and C), including one that migrates below Stat 5 and is present in both male (M) and female (F) samples (C).

FIG. 5. Activation of Stats by sequential pulses of GH. Shown in
A-C is a Western blot of anti-phosphotyrosine PY20 immunoprecipitates (IP) of liver nuclear extracts probed sequentially with the indicated anti-Stat antibodies. D shows a gel shift analysis of the corresponding nuclear extracts, without immunoprecipitation, using the SIE probe. Hypophysectomized (Hx) male (M) rats were treated with a single intraperitoneal injection of GH at 12.5 g/100 g of body weight and then sacrificed either 45 min later (lanes 2 and 3) or 240 min later (lanes 4 and 5). Other hypophysectomized male rats were given two intraperitoneal GH injections, each at 12.5 g/100 g of body weight and spaced 4 h apart, and then sacrificed 45 min after the second injection to test for the responsiveness of the Stats to a second GH pulse (lanes 6 and 7). Lane  desensitized with respect to tyrosine phosphorylation and DNA binding in response to repeated cycles of GH pulsation. This, in turn, may account for the low steady-state level of tyrosinephosphorylated liver nuclear Stat 1 and Stat 3 that is seen in intact male rats, which are repeatedly exposed to plasma GH pulses. The somewhat lower level of liver nuclear Stat 1 and Stat 3 proteins seen in intact females as compared with males (Fig. 4, A and B) and the correspondingly lower SIE binding activity of the female liver samples (Fig. 4D) may result from a more complete desensitization of Stat 1 and Stat 3 tyrosine phosphorylation in response to the continuous presence of circulating GH that is characteristic of the female rats.
GH Activation of Stat 3 Involves Serine/Threonine Phosphorylation in Addition to Tyrosine Phosphorylation-GH-induced liver Stat 5 tyrosine phosphorylation can be detected in the cytosol by a distinct mobility decrease, or "up-shift," in the SDS gel migration of the cytosolic Stat 5 protein (18). To ascertain whether a corresponding GH-induced mobility shift can be detected for cytosolic Stat 3 or Stat 1, we performed Western blot analysis of liver cytosols using anti-Stat antibodies. Fig. 6A shows that GH induces a small decrease in the electrophoretic mobility of Stat 3 that is detectable within 5 min of GH treatment (lanes 4 -6 versus lane 3). LPS, which, like GH, also activates Stat 3 by tyrosine phosphorylation (Fig. 1D), caused a similar decrease in Stat 3 mobility (Fig. 6A, lane 7). A very small mobility decrease for cytosolic Stat 1␣ appears to follow GH treatment; however, in this case, the resolution of the two protein bands is too small for any definitive conclusions to be drawn (data not shown).
Direct comparison of the GH-activated cytosolic Stat 3 band (upper band of closely spaced doublet marked by arrows in Fig.  6A, lanes 4 -6) with the tyrosine-phosphorylated Stat 3 that accumulates in the nucleus following GH treatment (lane 8 and upper band of doublet in lane 1) revealed a distinct difference in the electrophoretic mobility of the GH-activated cytosolic versus nuclear Stat 3 proteins. This mobility difference is comparable to the difference in mobility between tyrosine-phosphorylated nuclear Stat 3 and the constitutive nuclear Stat 3 protein found in untreated male or female rats (e.g.  (52), we employed PP2A, a phosphoprotein phosphatase that specifically cleaves serine and threonine phosphates, as well as PTP-1B, a phosphatase specific for tyrosine phosphates, to probe for these two distinct classes of protein phosphorylation. Treatment of GH-activated liver nuclear extract with PP2A followed by immunoblotting with anti-Stat 3 antibody revealed a small increase in Stat 3 electrophoretic mobility, consistent with a specific dephosphorylation of either phosphoserine or phosphothreonine residues (Fig. 6B, lane 2 versus lane 1). The new Stat 3 band that resulted (fuzzy band migrating just above the main Stat 3 band, marked by an arrow in Fig. 6B, lane 2) was indistinguishable from the GH-induced cytosolic Stat 3 protein seen in Fig. 6A (lanes 4 -6, arrows). A similar change in Stat 3 mobility was observed following treatment with PP1, a distinct phosphoserine/phosphothreonine phosphatase (data not shown). An increase in protein mobility was also observed following treatment with the phosphotyrosine-specific phosphatase PTP-1B (Fig. 6B, lane 3). These Stat 3 protein mobility changes were blocked by okadaic acid (100 nM for PP2A) and vanadate (0.1 mM for PTP-1B), specific inhibitors of the respective phosphatases (data not shown). Treatment with the broad specificity phosphatase CIP, which cleaves both serine/threonine phosphates and tyrosine phosphates, effected a larger increase in Stat 3 mobility and was accompanied by an intensification of the band marked 3 (lane 4). This mobility change is consistent with the removal by CIP of two Stat 3 phosphate groups, yielding dephosphorylated Stat 3 protein with a mobility indistinguishable from that of unactivated Stat 3 (i.e. lower band of doublet seen in lane 1). Together, these experiments demonstrate that Stat 3 undergoes both tyrosine phosphorylation and serine or threonine phosphorylation following GH treatment. The initial phosphorylation event, which appears to be tyrosine phosphorylation, occurs in the cytosol (Fig. 6A). The secondary phosphorylation event (serine/threonine phosphorylation) is closely linked to the nuclear translocation of Stat 3 (note the absence of a GH-activated cytosolic Stat 3 band at the migration position of the doubly phosphorylated nuclear protein designated 3(P) (Fig. 6A, lanes 4 -6)); however, we cannot determine from our data whether the secondary phosphorylation occurs in the cytosol or in the nucleus.  1-3). Band 3(P) thus corresponds to Stat 3 that is both tyrosine-phosphorylated and serine-or threonine-phosphorylated.
Serine/Threonine Phosphorylation of Liver Stat 5-We next examined whether a comparable, GH-induced serine/threonine phosphorylation can also be detected for liver Stat 5. Cytosolic and nuclear extracts prepared from GH-treated hypophysectomized rats were treated with PP2A, PTP-1B, or CIP and then were analyzed by anti-Stat 5 immunoblotting. Fig. 7A shows that following GH treatment, liver Stat 5 is converted to two slower moving proteins that are detectable in the cytosol (upper two bands in lanes 3 and 4) and that accumulate in the nucleus at much higher concentrations (lane NE; also see Fig. 7B, lanes  3 and 4). 4 Phosphotyrosine Western blotting verified that both nuclear Stat 5-immunoreactive protein bands are tyrosinephosphorylated (Fig. 7C, lanes 3 and 4). The broad specificity phosphatase CIP cleaved both tyrosine-phosphorylated proteins and fully reversed the GH-induced Stat 5 mobility changes (Fig. 7, B and C, lanes 7 and 8). Smaller (and unique) liver Stat 5 mobility changes were observed following treatment of the GH-activated samples with either PTP-1B or PP2A. Thus, the tyrosine-phosphorylated liver Stat 5 bands designated bands 2 and 1 were converted to the non-tyrosinephosphorylated bands 1a and 0, respectively, following PTP-1B treatment (lanes 5 and 6). By contrast, bands 2 and 1 were converted to a mixture of bands 1 and 0 following PP2A treatment (lanes 9 and 10). Liver Stat 5 bands 1 and 1a were distinguished from each other by the presence of phosphotyrosine in band 1, but not in band 1a (Fig. 7C, lanes 9 and 10  versus lanes 5 and 6), and by the slightly lower mobility of band 1a as compared with band 1. The specificity of each of these phosphatase-induced Stat 5 mobility changes was verified by the absence of a shift upon inclusion of specific phosphatase inhibitors (see "Materials and Methods"; data not shown).
Together, these experiments are consistent with the identification of liver Stat 5 band 2 as containing tyrosine phosphate and serine/threonine phosphate (upper band of doublet in Fig.  7B, lanes 3 and 4), Stat 5 band 1 as being tyrosine-phosphorylated (lower band of doublet in lanes 3 and 4 and upper band of doublet in lanes 9 and 10), and Stat 5 band 0 as the parent, nonphosphorylated liver Stat 5 (lower band present in Fig. 7, A  and B, lanes 5-10). Liver Stat 5 band 1a is a unique serine/ threonine-phosphorylated Stat 5 species that is formed from band 2 upon treatment with PTP-1B. The presence in these samples of a tyrosine-phosphorylated Stat 5 that is not also serine/threonine-phosphorylated (i.e. band 1) suggests that tyrosine phosphorylation may precede serine/threonine phosphorylation. Indeed, examination of the time course for GH-induced liver Stat 5 tyrosine phosphorylation revealed that at 10 min, Stat 5 band 1 is somewhat more abundant than band 2, while at 15 min, band 2 is more abundant (Fig. 7C, cf. ratio of two bands in lane 3 versus lane 4). Furthermore, phosphataseinduced mobility changes comparable to those seen with nuclear Stat 5 (Fig. 7B) were also observed following phosphatase treatment of cytosolic Stat 5, implying that both phosphorylation events occur in the cytosol (Fig. 7A). As was the case for Stat 3 (Fig. 6B), the fact that neither PTP-1B nor PP2A is sufficient to effect a full reversal of the Stat 5 mobility changes seen following GH treatment demonstrates that individual Stat 5 molecules undergo both types of phosphorylation in response to GH treatment.
Serine/Threonine Phosphorylation Modulates Stat DNA Binding Activity-Previous studies have established that tyrosine phosphorylation is required for Stat proteins to bind to their cognate DNA response elements (30). To probe for the possible importance of serine/threonine phosphorylation for the GH-activated DNA binding activities of Stat 3 and Stat 5, we examined the effect of PP2A treatment, in the absence and presence of the PP2A inhibitor okadaic acid, on the gel shift DNA binding activity of each Stat protein. Fig. 8A demonstrates that PP2A treatment dramatically decreases Stat 3 binding to the SIE probe, as revealed by the substantial loss of all three SIE complexes (lane 7 versus lane 6). In the case of some GH-activated nuclear samples, PP2A treatment preferentially decreased SIE complex A activity (Stat 3 homodimer) as compared with SIE complex B activity (Stat 1-Stat 3 heterodimer) and SIE complex C activity (Stat 1-Stat 1 homodimer), which were also decreased, but were more resistant to the effects of PP2A (Fig. 8C, lane 2 versus lane 1). A similar effect was observed following PP1 treatment (lanes 3 and 4). SIE complex activity was nearly abolished after treatment with either PTP-1B or CIP (Fig. 8A, lanes 3 and 5). Thus, serine/ threonine phosphorylation and tyrosine phosphorylation are both required for maximal Stat 3 and Stat 1 DNA binding activity. ␤-Casein probe gel shift analysis of these same phosphatase-treated nuclear samples revealed that whereas liver Stat 5 DNA binding activity was abolished by PTP-1B or CIP 4 The presence of two tyrosine-phosphorylated nuclear Stat 5 bands in Fig. 7 (B and C, lanes 3 and 4), but not in the earlier experiments (e.g. Fig. 1C), may in part be due to partial dephosphorylation during the dialysis to remove phosphatase inhibitors normally present in the nuclear extract preparation buffers (see Fig. 7 legend).  A and B). C, reprobing of blot shown in B with anti-phosphotyrosine antibody 4G10. Stat 5-immunoreactive bands are marked 2, 1, and 0 according to whether there are two phosphate groups (band 2, phosphotyrosine ϩ phosphoserine/phosphothreonine), one phosphate group (band 1, phosphotyrosine alone), or no phosphate groups (band 0), as determined by these analyses. Stat 5 band 2 comigrated with a non-tyrosine-phosphorylated band that was detectable in both the cytosol and nuclear samples; this latter protein band remained unchanged in its electrophoretic mobility following CIP treatment (cf. A, upper band in lanes 7 and 8). The faint band marked 1a (B, lanes 5 and 6) corresponds to a phosphoserine-or phosphothreonine-containing Stat 5 that is not tyrosine-phosphorylated (cf. absence of a corresponding band in lanes 5 and 6 in C) and was generated from band 2 by PTP-1B treatment. Band 1 in lanes 3 and 4 was electrophoretically indistinguishable from band 1 in lanes 9 and 10 in B and C, as demonstrated by a mixing experiment (data not shown). The higher ratio of band 2 to band 1 seen lane 4 compared with lane 3 in C suggests that tyrosine phosphorylation precedes serine/threonine phosphorylation. Shown in the first lane in A is nuclear extract (NE; 8 g of protein) corresponding to the cytosolic sample shown in lane 4. Samples included in this experiment were dialyzed against nuclear extract buffer in the absence of phosphatase inhibitors (see Footnote 4). treatment (Fig. 8B, lanes 3 and 5), PP2A treatment resulted in the appearance of an additional protein-DNA complex characterized by a distinctly lower gel mobility (complex II shown in Fig. 8, B, lane 7; and D, lanes 2, 4, and 6). This new protein-DNA complex was confirmed to contain Stat 5, as demonstrated by the supershift of both ␤-casein complexes I and II with anti-Stat 5 antibody (Fig. 8D, lane 7), but not with anti-Stat 1 or anti-Stat 3 antibody (data not shown). ␤-Casein complex II thus contains liver Stat 5 that is tyrosine-phosphorylated, but not serine/threonine-phosphorylated. Furthermore, whereas tyrosine-phosphorylated liver Stat 5 can bind to the ␤-casein probe to give two distinct protein-DNA complexes, liver Stat 5 serine/threonine phosphorylation inhibits the formation of complex II. DISCUSSION This study establishes that pulses of GH, administered in vivo in a hypophysectomized rat liver model, can activate three distinct Stat proteins, Stat 1, Stat 3, and liver Stat 5, by a mechanism that involves both tyrosine phosphorylation and either serine or threonine phosphorylation. Tyrosine phosphorylation associated with nuclear localization was shown to proceed with a distinct dependence on GH dose (Fig. 2) and with a distinct kinetics of desensitization for each of the Stats (Fig. 5). These effects of GH on these three Stat proteins are specific insofar as Stat 6, which is readily detected in rat liver cytosol by Western blot analysis, does not translocate to the nucleus following GH stimulation (data not shown). These findings, together with the present demonstration that GH-activated Stat 1 and Stat 3, but not liver Stat 5, interact with the high affinity SIE of the c-fos gene, while liver Stat 5, but not Stat 1 3 or Stat 3, interacts with a ␤-casein promoter Stat-binding site, lend strong support to the hypothesis that the activation of multiple Stat proteins by GH contributes to the widely diverse effects that GH can have on liver gene expression. Moreover, the striking dependence of liver Stat 5 activation (18), but not that of Stat 1 and Stat 3, on the temporal pattern of circulating GH supports the hypothesis that the latter two Stats may preferentially contribute to the regulation of GH-inducible genes, such as insulin-like growth factor 1, which are expressed in the liver at similar levels in male and female rats and whose transcription does not exhibit a strong plasma GH pattern dependence (13,53,54).
Stat 1 and Stat 3, but not liver Stat 5, were shown to be desensitized with respect to GH-induced tyrosine phosphorylation following a single GH pulse. The mechanism(s) underlying this striking desensitization of Stat 1 and Stat 3 are not known, but may involve a selective activation by the initial GH pulse of phosphoprotein phosphatase(s) that deactivate and thereby desensitize the Jak/Stat pathway. This could involve direct dephosphorylation by a GH-activated phosphotyrosine phosphatase of Stat 1 and Stat 3, but not liver Stat 5, or perhaps dephosphorylation of the tyrosine phosphate-docking site(s) for Stat 1 and Stat 3 that presumably are present on the GH-(GH receptor-Jak2 kinase) 2 complex and are required for Jak2-catalyzed Stat phosphorylation. GH-induced desensitization of Jak2 kinase has been observed in cultured IM-9 cells (17), and growth factor activation of the phosphotyrosine phosphatases SH-PTP-1 and SH-PTP-2 has been demonstrated (55,56). In some, but not all cases, tyrosine phosphatase activation leads to desensitization of the initial signaling event (56,57). Desensitization of GH-inducible liver Stat 5 activation has been observed, but is complete only following prolonged exposure (1-3 days) to the continuous plasma GH pattern that characterizes adult female rats (18). This difference between the Stats with respect to the kinetics of GH-induced desensitization and its dependence on the temporal pattern of GH stimulation suggests that liver Stat 5 may bind to the GH-(GH receptor-Jak2 kinase) 2 complex at phosphotyrosine-binding site(s) that are distinct from those utilized by Stat 1 and Stat 3, as has been suggested to occur for Stat 1 and Stat 5 with respect to the prolactin receptor-Jak2 kinase complex (58). Investigation of the dependence of Stat activation on the presence of particular tyrosine residues on GH receptor or Jak2 kinase by the use of COOH-terminal truncations of GH receptor (59 -61) or GH receptor-Jak2 kinase fusion proteins (62) may help resolve these questions.
The steady-state level of tyrosine-phosphorylated nuclear Stat 5 appears to be much higher than that of Stat 1 or Stat 3 For the nuclear extract sample included in this experiment, both phosphatases decreased SIE complex A activity to a greater extent than SIE complex B or C activity. D, effects of PP-2A on appearance of ␤-casein complex II, seen in three independent nuclear extracts prepared from GH-treated hypophysectomized rats (lanes 1-6). In lane 7, anti-Stat 5 antibody supershifted both ␤-casein complexes displayed in lane 6 (supershifted complexes designated ss I and ss II). A similar appearance of Stat 5 ␤-casein complex II occurred upon treatment of male nuclear extract with PP2A (data not shown).
in intact male rat liver. This is indicated by our inability to detect in male liver nuclear extracts substantial amounts of tyrosine-phosphorylated Stat 1 or Stat 3 by anti-phosphotyrosine immunoprecipitation analysis and by the dominance of liver Stat 5 on anti-phosphotyrosine Western blots of these same extracts. This same dominance is apparent with each of three anti-phosphotyrosine antibodies (4G10, PY20, and Shafer anti-PT) (data not shown) and is consistent with our finding that the pathway leading to GH-induced Stat 1 and Stat 3 activation can be strongly down-regulated by a single physiological pulse of GH. Accordingly, liver Stat 5 is the most responsive of the three Stat proteins to the repeated stimulation of hepatocytes by plasma GH pulses that occurs in vivo, raising questions regarding the importance and the precise roles of Stat 1 and Stat 3 with respect to GH signaling in the liver in intact animals. Further characterization of Stat 1-specific or Stat 3-specific DNA-binding sites in other GH-responsive genes may be useful in this regard.
In addition to the tyrosine-phosphorylated Stats, high constitutive levels of non-tyrosine-phosphorylated Stat 1 and Stat 3 were also found in the nucleus. The absence of phosphotyrosine was suggested by our inability to immunoprecipitate these proteins with anti-phosphotyrosine antibody (e.g. Fig. 1B,  lanes 1 and 2 versus lanes 5 and 6) and was confirmed by the lack of an effect of several phosphatases on the electrophoretic mobility of these proteins under conditions in which the corresponding GH-activated, slower migrating phosphorylated forms are readily dephosphorylated. This situation is in striking contrast to that of liver Stat 5, which is not detectable in the nucleus in the absence of GH stimulation, and could result from the action of a nuclear phosphotyrosine phosphatase that contributes to the GH-induced desensitization events discussed above. If tyrosine phosphorylation serves as a signal for nuclear localization or for nuclear retention, then the basis for the continued nuclear retention of the non-tyrosine-phosphorylated Stat 3 and Stat 1 proteins seen in our studies and their physiological function remain an enigma. Conceivably, heterodimerization involving an interaction between the SH2 domain of the non-tyrosine-phosphorylated Stat and phosphotyrosine residues present on the corresponding tyrosinephosphorylated Stat proteins, or perhaps on other tyrosinephosphorylated receptors, kinases, or other signaling molecules, may contribute to the continued nuclear localization of the non-tyrosine-phosphorylated Stats.
In recent studies carried out in cultured cell models, Stat 5 tyrosine phosphorylation was shown to be induced not only by prolactin and GH (19,20), but by multiple cytokines and growth factors, including IL-3, erythropoietin, and GM-CSF (20,41,44,63). This multiplicity of Stat 5 activation pathways can occur within a single cell type, raising the question as to how the unique specificity of each hormone and growth factor is preserved. This study, carried out in an intact animal model, indicates, however, that the functional redundancy and apparent overlap of Stat 5-activating hormones and pathways observed in cell culture need not occur in vivo under physiological conditions. Thus, while transfection and heterologous expression studies demonstrate that Stat 5 can be activated by both prolactin and GH (20), GH, but not prolactin, can activate Stat 5 in rat liver, despite the presence of prolactin receptors in liver tissue (64). Similarly, although a large number of cytokines can activate Stat 5 (as well as Stat 3) in cultured cells, treatment of rats with LPS, which stimulates the release of multiple cytokines in vivo, is presently shown to lead to the activation of Stat 3, but not Stat 5, in hepatocytes.
Recent studies have demonstrated the occurrence in the mouse of two closely related Stat 5 genes, designated Stat 5a and 5b, which encode proteins that are 96% identical (41,44,65). Both Stat 5 forms are expressed at a similar mRNA level in many mouse tissues, although in some studies, the levels in liver appear to be low (41,44). At present, we do not know whether the GH-activated rat liver Stat 5 characterized in our experiments corresponds to a Stat 5a or a Stat 5b form. Accordingly, we have used the term liver Stat 5 to refer to this protein.
Using cDNA-expressed mouse Stat 5a and Stat 5b, we have recently shown that the anti-Stat 5 monoclonal antibodies used for liver Stat 5 immunoblotting in this study and in our previous experiments (18) are cross-reactive with mouse Stat 5a and Stat 5b. 2 However, an identical banding pattern of GH-induced tyrosine phosphorylation and serine/threonine phosphorylation was obtained when an anti-Stat 5 antibody that is specific for mouse Stat 5b was used in these analyses (data not shown), suggesting that our analyses primarily, if not exclusively, detect Stat 5b, which may be the dominant Stat 5 form activated by GH in rat liver. Stat 5a and Stat 5b can be activated to a similar extent by prolactin, IL-3, IL-5, and GM-CSF (41,65), but the functional significance of these closely related Stat 5 forms with respect to GH signaling, including their transcriptional activation potential and their potential for heterodimerization with each other or with other GH-activated Stats or other factors, is presently unknown.
Although the Jak/Stat model for GH signaling in its simplest form involves tyrosine phosphorylation alone, the results presented here establish that GH induces a second post-translational modification of Stat proteins, namely, phosphorylation on either serine or threonine. This conclusion is supported by our finding that GH induces two distinct up-shifts in the electrophoretic mobility of both Stat 3 and liver Stat 5. Thus, in the case of Stat 3, GH induced a small but reproducible shift in the electrophoretic mobility of cytoplasmic Stat 3, yielding a protein that has a lower apparent M r than that of the tyrosinephosphorylated nuclear form (Fig. 6A). This mobility shift of cytoplasmic Stat 3 was observed as early as 5 min after GH injection, i.e. prior to the time when accumulation of tyrosinephosphorylated nuclear Stat 3 was first detected, and is consistent with tyrosine phosphorylation and serine/threonine phosphorylation occurring as two distinct steps. Similarly, in the case of liver Stat 5, two distinct nuclear Stat 5 proteins, both tyrosine-phosphorylated, were shown to accumulate following GH treatment. This heterogeneity is a consequence of the phosphorylation of liver Stat 5 at two distinct sites, rather than the detection on our immunoblots of a mixture of Stat 5a and Stat 5b (41,44), since (a) the same two bands are also observed using Stat 5b-specific antibodies (data not shown), and (b) the liver Stat 5 band heterogeneity is abolished by treatment with CIP. More direct support for the occurrence of Stat serine/threonine phosphorylation was obtained by treatment of GH-activated nuclear samples with the phosphoserine/ phosphothreonine-specific phosphoprotein phosphatase PP2A (66), which only partially reversed the GH-induced shift in Stat 3 electrophoretic mobility, as did treatment with the phosphotyrosine-specific phosphatase PTP-1B. Moreover, in the case of liver Stat 5, phosphatase treatment in conjunction with phosphotyrosine Western blotting provided conclusive evidence for the presence in the nucleus of Stat 5 species that are tyrosinephosphorylated and (tyrosine ϩ serine/threonine)-phosphorylated. The GH-induced serine/threonine phosphorylation event appears to be closely linked to nuclear translocation, particularly in the case of Stat 3; however, the precise subcellular localization of this secondary phosphorylation event (nucleus versus cytosol) cannot be determined at this time.
While these studies were in progress, several reports appeared demonstrating that cytokine-induced Stat 3 tyrosine phosphorylation is followed by serine or threonine phosphorylation in response to IL-6 and other cytokines that signal via gp130-linked receptors (67)(68)(69). Serine phosphorylation was also recently demonstrated for Stat 1 and Stat 3 following stimulation of cultured cells with interferon-␥ and epidermal growth factor, respectively, and the site of phosphorylation was localized to a Pro-Met-Ser 727 -Pro sequence that is conserved in the COOH-terminal region of several Stats, including Stat 5 (70). These Stat serine phosphorylations appear to be required for maximal DNA binding activity, at least toward some DNA response elements (68), and for maximal Stat-dependent transcriptional activation (67,70). This, in turn, is consistent with our finding that GH-induced Stat 3 DNA binding activity is abrogated by phosphoserine/phosphothreonine dephosphorylation. Stat 1 DNA binding activity was also inhibited by this same phosphatase treatment (e.g. loss of SIE complex C activity (Fig. 8A)), demonstrating that GH also induces a serine or threonine phosphorylation of Stat 1, which enhances its DNA binding potential. In contrast, the DNA binding activity of liver Stat 5 is altered by serine/threonine phosphorylation in a unique fashion, such that formation of the more slowly migrating ␤-casein complex II is inhibited. GH-induced formation of ␤-casein complex I occurs, however, in a manner that is independent of the serine/threonine phosphorylation status of liver Stat 5, while tyrosine phosphorylation is absolutely required for both ␤-casein-liver Stat 5 complexes to form. These findings suggest that GH-induced serine/threonine phosphorylation alters the stoichiometry of liver Stat 5 binding to its DNA response element or perhaps modulates heteromeric interactions of liver Stat 5 with novel nuclear factors that have yet to be identified.
GH-induced Stat 3 phosphorylation on serine or threonine was reversed by phosphatase PP2A as well as by the phosphoserine/phosphothreonine phosphatase PP1, which has a somewhat different specificity than PP2A toward some substrates (66). This contrasts to some extent with recent studies on IL-6-induced Stat 3 serine phosphorylation, where PP2A, but not PP1, was found to reverse serine phosphorylation (67,68). This finding, together with the loss of Stat 3 DNA binding activity toward a high affinity SIE following PP2A-induced dephosphorylation in the present study, but not in experiments using IL-6-activated Stat 3 (68), suggests that GH and IL-6 may perhaps activate Stat 3 by pathways that involve distinct subsets of serine or threonine residues, in addition to the presumed common site of tyrosine phosphorylation. This could provide an important mechanism for retention of target gene specificity for GH as compared with cytokines and other growth factors, despite their activation of the same Jak2 kinase and a common subset of Stat proteins. Further studies are required to identify and localize the kinase(s) involved in these serine/ threonine phosphorylations and to delineate the consequences of GH-induced serine/threonine phosphorylation for other functional properties of these Stats, including nuclear localization and retention, heterodimerization potential, target gene specificity, and transcriptional activation, and their regulation by specific phosphatases or other factors required for deactivation and desensitization of the initial hormone-induced signaling event.