Mechanisms of Growth Hormone (GH) Action

Many of the actions of growth hormone (GH) on somatic growth and tissue maintenance are mediated by insulin-like growth factor-I (IGF-I), a secreted protein whose gene expression is rapidly and potently induced by GH by unknown mechanisms. Recent studies implicating Stat5a and Stat5b in the growth response to GH in mice and observations linking Stat5b to control of IGF-I gene transcription in rats have prompted the current investigations into the molecular determinants of a putative regulatory network extending from GH through Stat5b to IGF-I. Here we characterize as critical components of this hormone-activated transcriptional pathway two adjacent Stat5 binding sites in the second intron of the rat IGF-I gene located within a conserved region previously found to undergo acute and reversible changes in chromatin structure after in vivo GH treatment. As assessed by chromatin immunoprecipitation assays, GH rapidly induced binding of Stat5 to this DNA segment in the liver of GH-deficient rats, just prior to the onset of transcription from both major and minor IGF-I gene promoters. Biochemical reconstitution experiments showed that the two intronic Stat5 DNA elements were able to bind Stat5b in vitro after GH treatment could transmit GH- and Stat5b-dependent transcriptional responsiveness to the major IGF-I promoter and to a minimal neutral gene promoter and were required for full stimulation of reporter gene activity by GH. Taken together, these results identify an intronic enhancer as a key mediator of GH-induced IGF-I gene transcription working through Stat5b and provide new insight into the molecular architecture of this transcriptional pathway.

Many of the actions of growth hormone (GH) on somatic growth and tissue maintenance are mediated by insulin-like growth factor-I (IGF-I), a secreted protein whose gene expression is rapidly and potently induced by GH by unknown mechanisms. Recent studies implicating Stat5a and Stat5b in the growth response to GH in mice and observations linking Stat5b to control of IGF-I gene transcription in rats have prompted the current investigations into the molecular determinants of a putative regulatory network extending from GH through Stat5b to IGF-I. Here we characterize as critical components of this hormone-activated transcriptional pathway two adjacent Stat5 binding sites in the second intron of the rat IGF-I gene located within a conserved region previously found to undergo acute and reversible changes in chromatin structure after in vivo GH treatment. As assessed by chromatin immunoprecipitation assays, GH rapidly induced binding of Stat5 to this DNA segment in the liver of GH-deficient rats, just prior to the onset of transcription from both major and minor IGF-I gene promoters. Biochemical reconstitution experiments showed that the two intronic Stat5 DNA elements were able to bind Stat5b in vitro after GH treatment could transmit GH-and Stat5b-dependent transcriptional responsiveness to the major IGF-I promoter and to a minimal neutral gene promoter and were required for full stimulation of reporter gene activity by GH. Taken together, these results identify an intronic enhancer as a key mediator of GH-induced IGF-I gene transcription working through Stat5b and provide new insight into the molecular architecture of this transcriptional pathway.
The biological effects of growth hormone (GH) 1 on somatic growth and tissue regeneration have been inextricably linked with the actions of insulin-like growth factor-I (IGF-I) ever since the somatomedin hypothesis was first formulated over 47 years ago (1). Much is now known regarding the molecular physiology and biochemistry of both proteins (2,3), and although several of the actions of GH do not require IGF-I (4) and IGF-I is not exclusively regulated by GH (3,5,6), their interdependent roles in controlling normal growth during childhood and maintaining tissue integrity during aging have been both confirmed and amplified in the intervening years (7)(8)(9)(10).
GH initiates its physiological effects after binding to the transmembrane GH receptor and activating JAK2, a receptorassociated intracellular tyrosine protein kinase (2), thereby setting into motion a series of protein phosphorylation cascades that lead to the activation of several transcription factors including AP-1 and Stats1, 3, 5a, and 5b among others (11). It has been known for over a decade that GH rapidly and potently induces IGF-I gene transcription in vivo (12), leading to the sustained production of IGF-I mRNAs and protein (12,13), yet the signaling pathways connecting the GH receptor and cytoplasmic Jak2 to the nuclear IGF-I gene have remained uncharacterized. Contributing to this challenge is the fact that the IGF-I gene is more complicated that its simple protein structure would have predicted. In mammals, the gene is transcribed from two promoters, each with unique leader exons, and its initial transcripts undergo both alternative splicing and differential polyadenylation to yield multiple mature mRNA species (14).
Recent studies from our laboratory have implicated the transcription factor Stat5b as a key component in acute GH-stimulated IGF-I gene activation in rats (15), thereby extending previous observations pointing to roles for Stat5b and to a lesser extent Stat5a in controlling normal somatic growth in mice (16,17) but not elucidating molecular mechanisms. Here we use a combination of in vivo observations in GH-deficient rats and biochemical reconstitution experiments in tissue culture cells to identify a region within the IGF-I gene that binds Stat5 in a GH-dependent manner and is capable of mediating GH-stimulated IGF-I gene activation via this transcription factor.

EXPERIMENTAL PROCEDURES
Materials-Antibodies to Stat5 (C17X) were from Santa Cruz Biotechnology (Santa Cruz, CA), and antibodies to anti-FLAG (M2) were from Sigma. Recombinant rat GH was from the National Hormone and Pituitary Program, NIDDK, National Institutes of Health. Oligonucleotides were synthesized at the Oregon Health & Sciences University Core Facility. Transit-LT1 was purchased from Mirus (Madision, WI), protein A-agarose beads were from Sigma, and the Qiaquick PCR DNA purification kit was from Qiagen (Valencia, CA). All of the other chemicals were reagent grade and were obtained from commercial suppliers.
Recombinant Plasmids-A plasmid in pGL2 has been described encoding 1711 bp of rat IGF-I promoter 1 and the first 328 bp of exon 1 (18). An 825-bp fragment of rat IGF-I intron 2 was added 5Ј to the promoter by standard molecular cloning methods. The mouse GH receptor in pcDNA3 was a gift from Dr. F. Talamantes (University of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: GH, growth hormone; IGF, insulin-like growth factor; Stat, signal transducers and activators of transcription; RT, reverse transcription; ChIP, chromatin immunoprecipitation assays; HS7, hypersensitive site 7; GHRE, GH response element. California, Santa Cruz, CA), and rat Stat5b in pcDNA3 was from Dr. Christin Carter-Su (University of Michigan, Ann Arbor, MI). The latter was modified by site-directed mutagenesis to add an NH 2 -terminal FLAG epitope tag, and point mutations were introduced to create constitutively active Stat5b (Stat5b CA , N699H) as described previously (15). Thymidine kinase luciferase was from Dr. Susan Berry (University of Minnesota, Minneapolis, MN). It was modified by the addition of portions of rat IGF-I intron 2 as indicated in Fig. 5.
Transient Transfections and Reporter Gene Assays-COS-7 cells (from ATCC) were incubated in media under conditions described previously (15) and were co-transfected with expression plasmids for the mouse GH receptor and wild type Stat5b or Stat5b CA using Transit-LT1 and a protocol from the supplier. After incubation for 24 h, serum-free medium was added containing 1% bovine serum albumin for 8 h followed by the addition of recombinant rat GH (40 nM final concentration) or vehicle for 30 min. Cells were then harvested, and nuclear proteins were isolated (19). For reporter gene assays, cells on 6-well tissue culture dishes were co-transfected with mouse GH receptor (750 ng) and wild type Stat5b (750 ng) and the promoter-reporter plasmids indicated in Figs. 4 and 5 (750 ng). Cells were incubated for 24 h followed by the addition of serum-free medium containing 1% bovine serum albumin and either rat GH or vehicle for 18 h. Cells were then harvested, and cell lysates were used for luciferase assays. All of the results were normalized for total cellular protein values.
Animal Studies-Male Sprague-Dawley rats, hypophysectomized by a transauricular route at age 7 weeks, were purchased from Harlan Sprague-Dawley (Indianapolis, IN). Animals were housed at the OHSU Animal Care Facility on a 12-h light/dark schedule with free access to food and water and received care according to National Institutes of Health guidelines. All of the procedures were approved by the OHSU Animal Care and Use Committee. Glucocorticoids (cortisol phosphate, 400 g/kg/day) and thyroxine (10 g/kg/day) were replaced by daily subcutaneous injection, and GH deficiency was confirmed by failure to grow during a 2-week observation period. Following this interval, rats were injected intraperitoneally with either vehicle (saline) or 1.5 g/g recombinant rat GH. At intervals, rats were anesthetized with pentobarbital (50 mg/kg intraperitoneally) and sacrificed. Liver proteins were isolated as described previously (19) as well as RNA and DNA outlined below.
RNA Analysis-Hepatic nuclear RNA was isolated as described previously (12). RNA concentration was determined spectrophotometrically at 260 nm, and its quality was assessed by agarose gel electrophoresis. Nuclear RNA (5 g) was reverse-transcribed with random hexamers in a final volume of 20 l using a RT-PCR kit (Invitrogen). PCR reactions were then performed with 0.5 l of cDNA (15). Primer sequences are listed in Table I. The linear range of product amplification was established in pilot studies for each primer pair, and the cycle number that reflected the approximate midpoint was used in final experiments. This varied from 20 to 28 cycles. Results were quantified by densitometry after electrophoresis through 1.5% agarose gels.
Chromatin Immunoprecipitation Assays (ChIP)-Initial steps were modified from published protocols (20). For each time point, a 200-mg fragment of rat liver was minced and incubated at 20°C in 10 ml of Dulbecco's modified Eagle's medium plus 1% formaldehyde on a rotating platform for 15 min followed by the addition of 1.5 ml of 1 M glycine and incubation for an additional 5 min. After centrifugation at 200 ϫ g for 5 min at 20°C, the pellet was suspended in 1 ml of phosphatebuffered saline, transferred to a 2-ml Dounce apparatus, and homogenized with 10 strokes of a tight fitting pestle. After brief centrifugation in a microcentrifuge, the pellet was suspended in 400 l of lysis buffer (50 mM Tris-HCl, 5 mM EDTA, 1% SDS, pH 8.1, plus protease inhibitors) and incubated for 15 min at 4°C. Each sample was sonicated at 4°C using a total of 5 pulses for 15 s each of a Branson Micro-tip sonicator at setting 5 interspersed with 30-s incubations on ice. After centrifugation at 14,000 ϫ g for 10 min at 4°C in a microcentrifuge, the supernatant was diluted 10-fold with immunoprecipitation buffer (20 mM Tris-HCl, 150 mM NaCl, 2 mM EDTA, pH 8.1, plus protease inhibitors) and 1-ml aliquots were used for immunoprecipitation as described

5Ј-CCAATGTGCCCTCCTCGGCGACC-3Ј
previously (20) with 3 l of the Stat5 C17X antibody and 45 l of a 50% slurry of protein A-agarose beads. DNA was extracted from the immunoprecipitates using the Qiaquick PCR DNA purification kit and a protocol from the supplier and was suspended in 50 l of 10 mM Tris-HCl, 1 mM EDTA, pH 8.0. PCR reactions were performed with 1 l of DNA using the primer pairs listed in Table II. The linear range of product amplification was established for each primer pair in pilot studies, and the cycle number that reflected the approximate midpoint was used in final experiments. This varied from 28 to 30 cycles. Results were visualized after electrophoresis through 1.5% agarose gels.

Identification of a GH-regulated Stat5
Binding Site within the Rat IGF-I Gene-In past studies, we mapped a GH-stimulated DNase I-hypersensitive site termed HS7 to the second intron of the rat IGF-I gene and demonstrated that alterations in this chromosomal region coincided with GH-stimulated induction of IGF-I gene transcription in vivo (12). More recently, we found that Stat5b was required for acute activation of IGF-I gene transcription by GH (15). The latter observations prompted the analysis of the IGF-I gene including HS7 for potential Stat5 binding sequences. A map of the first three exons of the rat IGF-I gene is illustrated in Fig. 1A and includes tandem promoters P1 and P2, their leader exons 1 and 2, respectively, and exon 3, which encodes the NH 2 -terminal half of the mature IGF-I peptide (14). HS7 is located approximately midway between exons 2 and 3 as indicated. The DNA sequence illustrated below the map represents part of the 3Ј end of HS7. It contains two putative Stat5 binding sites labeled GHRE-1 and GHRE-2, respectively, which both conform to the consensus of 5Ј-TTCNNNGAA-3Ј (21). Fig. 1B demonstrates that this 84-bp region of the rat IGF-I gene is fairly well conserved in two other mammalian species and that GHRE-1 is preserved as a potential Stat binding site among rat, mouse, and human IGF-I genes. There is additionally an ϳ85% identity over the 900 bp flanking the 84-bp HS7 core DNA sequence between rats and mice and an ϳ64% identity between rat and human IGF-I genes (data not shown).
Chromatin immunoprecipitation experiments were performed to determine whether the IGF-I HS7 region was capable of binding Stat5 in vivo in a GH-dependent way. As shown in Fig. 2A Fig. 1. Sequences for all of the primers may be found in Table II The two promoters, P 1 and P 2 , are shown as well as the location between exons 2 and 3 of DNase I-HS7, found previously to be altered in chromatin in rat hepatic nuclei after acute GH treatment (12). Also illustrated are approximate locations of oligonucleotide primers used for ChIP (black) or RT-PCR experiments (light gray). A 1-kb scale bar is shown. The DNA sequence below the map lists the 84-bp segment containing two potential GH response elements, GHRE-1 and GHRE-2 (underlined). B, comparative anatomy of the chromosomal region spanning GHRE-1 and GHRE-2 in human, rat, and mouse IGF-I genes. Identical nucleotides between two species are shaded in gray and among all three species are shaded in black. GHRE-1 and GHRE-2 are boxed.
60-min duration of the experiments. Nearly identical binding kinetics were observed at the proximal promoter of the GHactivated Sp12.1 gene, which contains tandem Stat5 sites (22), although no binding of Stat5 was observed at IGF-I intron 3 or proximal promoter 1 or to portions of the c-fos or ␤-actin genes, indicating that Stat5 was not being non-specifically crosslinked to DNA in liver chromatin.
Hormone-dependent gene transcription was next assessed to evaluate the functional consequences of GH-stimulated Stat5 binding to the HS7 region of the rat IGF-I gene. Hepatic nuclear RNA was isolated from the same livers used for ChIP assays, and the accumulation of nascent transcripts was measured by semi-quantitative RT-PCR. As seen in Fig. 2B, transcripts directed by IGF-I promoters 1 and 2 were both strongly induced within 30 min after systemic GH injection and were increased in abundance by 60 min. Similar results were observed for Spi 2.1 gene activation. In this experimental model, GH also transiently induced c-fos gene expression as shown previously (23) but had little effect on ␤-actin. Thus, in vivo GH treatment rapidly and potently stimulates binding of Stat5 to the HS7 site in the second intron of the rat IGF-I gene in the liver followed by activation of transcription from both IGF-I promoters.
DNA Binding Properties of GHRE-1 and GHRE-2-We used gel-mobility shift experiments to determine whether Stat5b could bind directly to GHRE-1 or GHRE-2. Fig. 3A shows results using labeled double-stranded oligonucleotides for each element. Nuclear protein extracts were prepared from COS-7 cells transfected with expression plasmids for the mouse GH receptor and either wild type or constitutively active rat Stat5b and treated with GH or vehicle for 30 min. As seen in Fig. 3A, GH induced the binding of wild type Stat5b to both DNA A thin arrow denotes the location of free probe (FP). B, inducible binding after in vivo GH treatment of rat hepatic nuclear protein extracts to a double-stranded oligonucleotide for GHRE-1 as assessed by gel-mobility shift experiment (lanes 1-4).  show that binding of the same nuclear protein extracts to a double-stranded oligonucleotide for Sp-1 is relatively constant after GH. The locations of probes, whereas constitutively active Stat5b was able to bind in the absence of hormone. Antibody supershift experiments confirmed that Stat5b interacted with each oligonucleotide.
We next evaluated the binding of endogenous hepatic Stat5 to the putative GHREs in HS7. Shown in Fig. 3B are results using labeled GHRE-1 and nuclear protein extracts from GHdeficient rats acutely treated with hormone. Inducible binding was observed beginning within 15 min after systemic GH injection and reached a maximum by 30 -60 min. Under the conditions used in these experiments, no binding to a GHRE-2 oligonucleotide could be detected (data not shown), implying that it is a lower affinity site compared with GHRE-1. As a control, gel-mobility shift assays were performed with a doublestranded oligonucleotide that recognizes members of the Sp1 family of transcription factors (24) that are not regulated by GH. As expected, constant protein-DNA interactions were observed over the same time course, indicating that both the quantity and quality of nuclear protein extracts were similar at each time point after hormone treatment.
The relative affinity of GHRE-1 for hepatic Stat5 was assessed using liver nuclear proteins isolated at 60 min after injection of GH and a variety of unlabeled competitor oligonucleotides. As shown in Fig. 3C, GHRE-1 competed avidly with itself for nuclear protein binding, whereas GHRE-2 competed poorly, even at a ϫ100 molar concentration. The contiguous Stat5 binding sites from the Spi 2.1 gene promoter competed with GHRE-1 but only at ϫ100 concentration, and the unrelated binding site for the Oct-1 transcription factor (24) was ineffective. Taken together, the results in Fig. 3 show that GHRE-1 and GHRE-2 are bona fide Stat5 binding elements and that the affinity of Stat5 for GHRE-1 is greater than that for GHRE-2.
GHRE-1 and GHRE-2 Mediate GH-activated and Stat5b-dependent Gene Transcription-We used reconstitution experiments in COS-7 cells to evaluate the transcriptional properties of HS7 and GHRE-1 and GHRE-2. Cells were transiently trans-fected with expression plasmids for the mouse GH receptor and rat Stat5b and luciferase fusion genes containing rat IGF-I promoter 1 alone or with an 825-bp fragment of rat HS7 including GHRE-1 and GHRE-2. As shown in Fig. 4, GH treatment had no effect on the activity of the native IGF-I promoter but induced a 3.5-fold increase in luciferase activity of fusion genes containing the HS7 segment.
To extend this observation of GH-stimulated gene transcription through HS7, additional reconstitution experiments were performed with luciferase reporter genes containing a minimal thymidine kinase promoter. As pictured in Fig. 5A, in the presence of GHRE-1 and GHRE-2 and Stat5b, GH induced a 6 -7-fold increase in luciferase activity. The importance of the two GHREs for hormonal regulation was next evaluated after engineering point mutations into each site. As illustrated in Fig. 5B, a ϳ8-fold increase in luciferase was detected after GH when both GHREs were present. The hormonal response dropped to ϳ4-fold when only one GHRE was intact and declined to Ͻ1.4-fold when both were mutated. In addition, four copies of GHRE-1 were able to transfer an ϳ7-fold response to GH onto the minimal thymidine kinase promoter (data not shown). Thus, each GHRE is capable of mediating GH-induced gene transcription in the presence of Stat5b.

DISCUSSION
A long-standing challenge in the field of growth biology has been to understand the biochemical mechanisms by which GH activates IGF-I gene expression. Here we identify binding sites for the transcription factor Stat5 in a region within the second intron of the rat IGF-I gene previously shown to undergo a GH-stimulated chromatin transition coincident with hormoneinduced IGF-I transcription (12). We show that this segment of the IGF-I gene binds Stat5 in vivo in a GH-dependent way just prior to the onset of IGF-I gene activation, can bind Stat5b in vitro, and is sufficient to mediate hormone-regulated and Stat5bdependent reporter gene expression in the context of both the FIG. 5. GHRE-1 and GHRE-2 mediate GH-stimulated and Stat5b-regulated gene transcription. Results of luciferase assays in COS-7 cells transiently transfected with expression plasmids encoding the mouse GH receptor and wild-type rat Stab5b and incubated with rat GH (40 nM) or vehicle for 18 h. A, luciferase reporter genes were included that contained the minimal thymidine kinase (TK) promoter alone or with either 932 or 84 bp of DNA from the HS7 region of the rat IGF-I gene (thin lines). GHRE-1 and GHRE-2 are indicated by shaded circles. B, luciferase reporter genes were used that contained the minimal TK promoter alone or with 84 bp of DNA from HS7 encoding either wild type or mutant versions of GHRE-1 and GHRE-2. Mutant sequences are indicated by an X (for GHRE-1, TTCCTGGAA was changed to GTCCTGGTA, and for GHRE-2, TTCTTAGAA was changed to GTCTTAGTA). For both A and B, the graphs summarize the results of 3-5 independent experiments, each performed in duplicate (*, p Ͻ 0.01; **, p Ͻ 0.02 versus no GH or no Stat5b). Luciferase values for the TK promoter ranged from 4,000 to 8,000 relative light units/10 s. major IGF-I gene promoter and a minimal thymidine kinase promoter. Taken together, these results show that Stat5b is a key protein responsible for the induction of IGF-I gene transcription by GH and define the molecular architecture of this transcriptional response.
Previous gene knock-out studies have established a role for Stat5b and to a lesser extent Stat5a in the physiological pathways responsible for normal somatic growth in rodents (16,17). In the absence of these proteins, mice developed up to a 30% deficit in postnatal linear growth that was associated with an ϳ50% decline in serum levels of IGF-I (16,17). More recently, we found that forced expression of a dominant-negative version of Stat5b in GH-deficient rats blocked the acute activation of IGF-I transcription by GH, whereas constitutively active Stat5b stimulated IGF-I gene expression in the absence of hormone (15). Both groups of observations implicated Stat5b in IGF-I gene regulation but did not identify specific biochemical mechanisms, a question that we now have addressed.
Members of the Stat family of transcription factors bind as dimers to DNA response elements consisting of variants of the palindromic nucleotide sequence 5Ј-TTCN 2-4 GAA-3Ј where N signifies any deoxyribonucleotide (21). Ehret et al. (25) recently showed that Stats 1 and 5 preferentially bind to sites where n ϭ 3 (25). This is the precise spacing of IGF-I GHRE-1 and GHRE-2. Based on their results, we find additionally that GHRE-1 is identical in its core DNA sequence to Stat5 binding sites in the Cis and T cell receptor ␥ genes and that GHRE-2 is identical to the Stat5 response element in the ␣s1 casein gene (25), thus providing further evidence supporting our experimental data that GHRE-1 and GHRE-2 are bona fide Stat5 binding sequences.
The exact biochemical steps by which Stat5 activates target gene transcription remain incompletely understood. Previous studies established that the COOH-terminal transcription activation region of the protein could interact directly with the transcriptional co-activators CREB-binding protein and p300 (26,27), a process that led to chromatin modifications including histone acetylation (26,27), and that transcription was facilitated by another Stat5-interacting protein, Nmi (28). Others have identified the widely expressed protein, centrosomal P4.1associated protein, as a second co-activator of Stat5 that also physically interacted with the COOH terminus of the protein (29).
More recent studies have focused on the surprising observation that upon binding of activated Stat5 to regulatory sites on target genes, an associated histone deacetylase activity was enhanced (30,31). In one series of experiments, the deacetylase inhibitor trichostatin A globally blocked cytokine-induced activation of Stat5-dependent genes (30). In another study, recruitment of the histone deacetylase, HDAC-1, to the Id-1 promoter by active Stat5 led to deacetylation of the transcription factor CCAAT/enhancer-binding protein ␤, events that were required for induction of Id-1 transcription (31). Our observations do not shed light on whether co-regulatory proteins with either acetyltransferase or deacetylase activity associate with Stat5 on the IGF-I gene but do provide a robust model system for answering this question.
GH and IGF-I play central roles in human physiology. They are essential for normal growth during fetal and postnatal development and are critical for tissue maintenance and repair during aging (2-5, 7, 8). In contrast, normal levels of both proteins have been found to have potential pathogenic consequences, being linked by epidemiological studies to several cancers (32). Thus, a full understanding of the molecular mechanisms by which GH regulates IGF-I gene expression will be necessary for developing effective therapeutic strategies to use these potent agents safely (33).