Identification of a distal STAT5-binding DNA region that may mediate growth hormone regulation of insulin-like growth factor-I gene expression.

Growth hormone (GH) regulates insulin-like growth factor-I (IGF-I) gene expression through signal transducer and activator of transcription 5b (STAT5b) and STAT5a. The objective of this study was to identify the cis-regulatory DNA region involved in this process. By cotransfection analyses of shotgun DNA fragments of a bacterial artificial chromosome sequence containing the entire human IGF-I gene and a large 5'-flanking region, a approximately 700-bp DNA region approximately 75 kb 5' to the IGF-I gene was found to have the ability to enhance gene expression from both heterologous and homologous promoters in the presence of constitutively active STAT5a or STAT5b. This 700-bp DNA region contains two closely located consensus STAT5-binding sites, and its sequence appears to be evolutionarily conserved. Electrophoretic mobility shift assays verified the ability of the two putative STAT5-binding sites to bind to STAT5a and STAT5b. Cotransfection analyses confirmed that both STAT5-binding sites were necessary for the 700-bp DNA region to mediate STAT5a or STAT5b activation of gene transcription. Chromatin immunoprecipitation assays demonstrated that the chromosomal region containing these two STAT5-binding sites was bound by constitutively active STAT5b protein in HepG2 cells and that the binding was accompanied by increased expression of IGF-I mRNA. In reconstituted GH-responsive cells, this 700-bp DNA region was able to mediate GH-induced STAT5a or STAT5b activation of gene expression. These results together suggest that this STAT5-binding site-containing distal 5'-flanking region of IGF-I gene may be an enhancer mediating GH-induced STAT5 activation of IGF-I gene transcription.


Growth hormone (GH) regulates insulin-like growth factor-I (IGF-I) gene expression through signal transducer and activator of transcription 5b (STAT5b) and
STAT5a. The objective of this study was to identify the cis-regulatory DNA region involved in this process. By cotransfection analyses of shotgun DNA fragments of a bacterial artificial chromosome sequence containing the entire human IGF-I gene and a large 5-flanking region, a ϳ700-bp DNA region ϳ75 kb 5 to the IGF-I gene was found to have the ability to enhance gene expression from both heterologous and homologous promoters in the presence of constitutively active STAT5a or STAT5b. This 700-bp DNA region contains two closely located consensus STAT5-binding sites, and its sequence appears to be evolutionarily conserved. Electrophoretic mobility shift assays verified the ability of the two putative STAT5-binding sites to bind to STAT5a and STAT5b. Cotransfection analyses confirmed that both STAT5-binding sites were necessary for the 700-bp DNA region to mediate STAT5a or STAT5b activation of gene transcription. Chromatin immunoprecipitation assays demonstrated that the chromosomal region containing these two STAT5-binding sites was bound by constitutively active STAT5b protein in HepG2 cells and that the binding was accompanied by increased expression of IGF-I mRNA. In reconstituted GH-responsive cells, this 700-bp DNA region was able to mediate GH-induced STAT5a or STAT5b activation of gene expression. These results together suggest that this STAT5-binding sitecontaining distal 5-flanking region of IGF-I gene may be an enhancer mediating GH-induced STAT5 activation of IGF-I gene transcription.
Insulin-like growth factor-I (IGF-I) 1 is an important endocrine and paracrine regulator of cell proliferation and metabolism (1). Most of the circulating IGF-I is produced from the liver (2), and IGF-I production in this tissue is primarily controlled by pituitary growth hormone (GH) at the transcriptional level (3,4). GH regulation of IGF-I gene expression has been known for decades, but the underlying mechanism is not completely understood. Because the effect of GH on IGF-I gene expression is rapid (4,5) and independent of protein synthesis (6), GHinduced IGF-I gene expression is believed to result from GHinduced direct interaction of transcription factors with regulatory DNA regions in the IGF-I gene. A lot of work has been attempted to map such regulatory DNA regions (7)(8)(9)(10)(11)(12), but a convincing GH-responsive regulatory region has not been identified. The slow progress in identifying the cis-regulatory DNA regions mediating GH induction of IGF-I gene transcription is perhaps due to the lack of convenient GH-responsive cell lines and the complexity of IGF-I gene structure.
Signal transducer and activator of transcription 5b (STAT5b) is a well established component of the GH signaling pathway (13). The STAT5b-null mice had 50% less liver IGF-I mRNA and 30% lower serum IGF-I concentration than the wild-type mice (14,15) and did not increase liver IGF-I mRNA abundance or blood IGF-I concentration in response to GH treatment (14). Overexpression of a dominant-negative STAT5b mutant completely prevented GH-induced IGF-I gene expression in the liver, whereas that of a constitutively active STAT5b mutant led to robust, GH-independent IGF-I gene expression in the hypophysectomized rats (16). A homozygous STAT5b gene missense mutation was recently discovered to cause a series of characteristics of growth hormone insensitivity, including reduced serum IGF-I concentration in a human patient (17). These findings together indicate that STAT5b is a key transcription factor mediating GH regulation of IGF-I gene transcription. In addition to STAT5b, STAT5a, a protein that is 96% identical in sequence to STAT5b (18) and recognizes the same DNA sequence as STAT5b (19), may be another transcription factor mediating GH stimulation of IGF-I gene transcription. Supporting this role of STAT5a is the observations that STAT5a and STAT5b double knock-out mice had lower serum IGF-I levels and showed severer growth retardation than STAT5b knock-out mice (20).
Because GH stimulation of IGF-I gene expression may be directly mediated by STAT5, the cis-regulatory DNA regions involved might be the regions that contain STAT5-binding sites. A region within the rat IGF-I intron 2 has recently been identified to contain two STAT5-binding sites and to mediate GH-stimulated IGF-I gene expression in the liver of rats (21). However, we have found that the corresponding human or bovine IGF-I intron 2 region bears limited sequence homology to the rat IGF-I intron 2 region and contains only one putative STAT5-binding site. We therefore hypothesized that GH-induced STAT5 activation of IGF-I gene expression in humans might be mediated by DNA regions other than or in addition to the IGF-I intron 2 region. By cotransfection analysis of a large chromosomal region containing the human IGF-I gene with constitutively active STAT5 proteins, we have identified an evolutionarily conserved distal 5Ј-flanking region of IGF-I gene that contains two consensus STAT5-binding sites and can mediate GH-induced STAT5 activation of gene expression.

EXPERIMENTAL PROCEDURES
Shotgun Library Construction-A bacterial artificial chromosome (BAC) clone (RP11-210L7) containing the entire human IGF-I gene and the ϳ84-kb 5Ј-flanking region was purchased from BAC-PAC Resources (Children's Hospital Oakland Research Institute, Oakland, CA). The BAC plasmid DNA was purified using a Plasmid Maxiprep kit (Qiagen) and sheared by sonication. The sonicated DNA was size-fractionated by gel electrophoresis, and fragments between 2 and 4 kb were selected and purified using a QIAquick gel extraction kit (Qiagen). Ends of DNA fragments were blunted with T4 and Klenow DNA polymerases. The resultant DNA fragments were ligated into SmaI-digested and dephosphorylated pGL2-Promoter vector (Promega, Madison, WI) and transformed into Escherichia coli strain DH10B cells by electroporation. Approximately 300 clones were randomly picked to verify the inclusion of inserts in the plasmids.
Plasmid Construction-A 940-bp promoter 1 of human IGF-I gene, from which class 1 IGF-I mRNA is transcribed, was amplified by PCR (Table I) and subsequently cloned into pGL2-Basic vector (Promega) to generate plasmid IGFP1-pGL2B. A 2871-bp human IGF-I intron 2 region was amplified by PCR (Table I) and cloned into pGL2-Promoter vector to generate plasmid IGFIntron2-pGL2P. The 700-bp distal 5Јflanking region of IGF-I gene, identified as a STAT5-binding enhancer in this study, was amplified by PCR (Table I) and inserted upstream of the SV40 promoter in pGL2-Promoter vector, the human IGF-I promoter in IGFP1-pGL2B, and a minimal thymidine kinase (TK) promoter in pGL2-TK vector to generate plasmids 700bp-pGL2P, 700bp-IGFP1-pGL2B, and 700bp-pGL2-TK, respectively. The pGL2-TK vector was prepared by inserting the TK promoter from pSPI-LUC plasmid (provided by Dr. Tim Wood from Karolinska Institute, Novum, Huddinge, Sweden) (22) into pGL2-Basic vector. Six copies of a growth hormone response element (GHRE), each containing two adjacent STAT5-binding sites (22), were removed from pSPI-LUC and inserted upstream in pGL2-Promoter vector to generate pGL2P-GHRE. The inserts in all new plasmids were verified by sequencing.
Either or both putative STAT5-binding sites in the 700-bp distal 5Ј-flanking region of IGF-I gene in plasmid 700bp-pGL2P were replaced with one or two NotI sites to generate STAT5-binding site-mutated constructs 700bpm1-pGL2P (in which the first STAT5-binding site was mutated), 700bpm2-pGL2P (in which the second STAT5-binding site was mutated), and 700bpm3-pGL2P (in which both STAT5-binding sites were mutated). The mutations were made by PCR-based sitedirected mutagenesis ( Table I). The mutated inserts from these pGL2Pbased constructs were inserted into pGL2TK vector at the SmaI and KpnI sites to generate the pGL2TK-based mutation constructs 700bpm1-pGL2TK, 700bpm2-pGL2TK, and 700bpm3-pGL2TK, respectively. All of the mutations were confirmed by sequencing.
A standard reverse transcription-PCR was used to clone the complete coding region of the bovine GHR mRNA. Briefly, liver poly(A) mRNA from a Holstein cow was reverse transcribed into cDNA in the presence of an oligo(dT) primer. The cDNA was then amplified by PCR with primers (Table I) based on the bovine GHR mRNA sequence (23). The PCR products were cloned into pcDNA3.1 vector (Invitrogen), generating bovine GHR expression plasmid pcDNA3-bGHR. The GHR cDNA sequence was confirmed by sequencing. The ability of the pcDNA3-bGHR plasmid to express GHR protein was verified by in vitro transcription and translation in rabbit reticulocyte lysates (Promega) in the presence of [ 35 S]methionine.
The expression plasmids encoding wild-type mouse STAT5a, constitutively active STAT5a mutant STAT5a-N642H, and wild-type mouse STAT5b were provided by Dr. Kouichi Ariyoshi (The University of Tokyo, Tokyo, Japan) (24). The plasmid encoding constitutively active STAT5b mutant STAT5b CA was provided by Dr. Peter Rotwein (Oregon Health and Science University, Portland, OR) (16).
Cell Culture and Cotransfection Analysis-MAC-T cells, a bovine mammary epithelial cell line (25), were cultured in Dulbecco's modified Eagle's medium with L-glutamine and 10% fetal bovine serum. CHO cells (ATCC, Manassas, VA) were grown in minimum essential medium containing sodium pyruvate, L-glutamine, and 10% fetal bovine serum. All of the reagents used in cell culture were from Sigma. In the transfection analyses to identify STAT5-responsive IGF-I DNA regions, 0.5 g of reporter gene construct and 40 ng of STAT5 expression plasmid were cotransfected with 1 ng of pRL-CMV (transfection efficiency control plasmid from Promega) into 50% confluent MAC-T cells in 24-well plates using FuGENE 6 (Roche Applied Science). Forty-eight hours after transfection, the cells were lysed, and the luciferase activities were measured using a dual luciferase reporter assay system (Promega). In the transfection analyses that were used to determine GH response, cells in each well were transfected with 0.5 g of the respective reporter gene construct, 200 ng of bovine GHR expression plasmid, 200 ng of wild-type STAT5a or STAT5b expression plasmid, and 1 ng of pRL-CMV. Twenty-four hours after transfection, the medium was replaced with serum-free medium, and the cells were grown for a further 16 h. Following that, the medium was replaced with serum-free medium containing 500 ng/ml recombinant bovine GH (provided by the National Hormone and Peptide Program), and the cells were grown for another 8 h before luciferase assay.
Electrophoretic Mobility Shift Assay-Double-stranded oligonucleotides (Table I) corresponding to the two putative STAT5-binding sites in the distal 5Ј-flanking region of human IGF-I gene were end-labeled with 32 P using T4 polynucleotide kinase and [␥-32 P]ATP. Nuclear proteins were prepared from CHO cells transfected with STAT5a-N642H or STAT5b CA expression plasmid, according to a standard protocol (26). Ten g of nuclear proteins were incubated with 2 g of anti-STAT5a (Upstate Biotechnology, Inc., Lake Placid, NY), anti-STAT5b antibody (Abcam Inc., Cambridge, MA), or 2 g of the corresponding preimmune serum in reaction buffer containing 20% glycerol, 20 mM Tris-HCl, pH 7.5, 100 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, and 2 g of poly(dI-dC) for 1 h at 4°C. The binding reactions were subsequently added with 1 ϫ 10 5 cpm of 32 P-labeled oligonucleotide probe, and the incubations were continued for 1 h at 4°C. The reactions were resolved on native 6% polyacrylamide gels. After electrophoresis, the gels were dried, exposed to phosphor screens, and scanned on a Molecular Imager FX System (Bio-Rad).
Chromatin Immunoprecipitation (ChIP) Assay-This assay was performed as described (26) with minor modifications. Approximately 1 ϫ 10 6 human hepatoma-derived HepG2 cells (ATCC) were transiently transfected with 20 g of STAT5b CA expression plasmid. Forty-eight hours after transfection, the cells were cross-linked with 1% formaldehyde for 15 min at 25°C, followed by termination of the reaction with 125 mM glycine. The nuclei were subsequently isolated and sonicated with 10 pulses of 10 s each at power setting 35% using the sonic dismembrator model 300 (Fisher). The following steps were performed using a ChIP kit, following the manufacturer's directions (Upstate Biotechnology, Inc.) with minor modifications. Briefly, ϳ33% of the sonicated chromatin was saved as "input," and the rest was cleared twice to reduce nonspecific immunoprecipitation with salmon sperm DNA/protein A-agarose slurry. One half of the precleared chromatin was incubated with 3 g of anti-STAT5b antibody (Abcam Inc.), and the other half was incubated with 3 g of preimmune rabbit serum at 4°C overnight. The immunocomplexes were collected with protein A-agarose and washed stringently to remove nonspecific antibody binding. The DNA-protein complexes were eluted, and the cross-linking was reversed by incubation at 65°C for 6 h. The DNA from input-, antibody-, or preimmune serum-treated chromatin samples was isolated and purified by standard proteinase K digestion, phenol-chloroform extraction, and ethanol precipitation. The purified DNA from each sample was resuspended in 50 l of water. From the purified DNA, the relative levels of the STAT5-binding site-containing distal 5Ј-flanking region of IGF-I gene or that of IGF-I exon 4 were quantified by PCR with specific primers (Table I)  Statistical Analysis-The luciferase activity data from cotransfection and real time PCR analyses were analyzed by one-factor analysis of variance, followed by t test to compare two means or the Tukey test to compare multiple means. All of the statistical analyses were performed using the respective programs of SAS (SAS Institute, Cary, NC). All of the data are expressed as the means Ϯ S.E. Differences at p Ͻ 0.05 were considered significant.

Identification of STAT5-binding Enhancers in a 170-kb Human Chromosomal Region
Containing IGF-I Gene-Most of the previous studies attempted to identify GH-responsive DNA regions focused on the IGF-I promoter or IGF-I intronic regions proximal to the transcription start site (7)(8)(9)(10)(11)(12). To identify potentially distal GH-responsive DNA regions, we mapped a 170-kb BAC sequence containing the entire human IGF-I gene and the ϳ84-kb 5Ј-flanking region for STAT5-binding enhancers. A shotgun library of this 170-kb BAC insert was constructed in the enhancer-less pGL2-promoter vector pGL2P. From this library, 115 plasmids, which contained inserts 2-4 kb and were estimated to give a 2-fold coverage of the 170-kb BAC sequence, were each cotransfected with a plasmid expressing the constitutively active STAT5 mutant STAT5a-N642H (24) or the corresponding empty plasmid into MAC-T cells. As shown in Fig. 1, the activity of the positive control construct pGL2P-GHRE that contains 12 copies of STAT5binding site from the Spi 2.1 gene (22) was increased 6.5-fold on average by STAT5a-N642H, validating the effectiveness of the assay in identifying STAT5-responsive enhancers. The increases (or decreases) in reporter gene expression from most of the IGF-I DNA-containing pGL2P plasmids were within two times of standard deviation (0.267) of that from the vector plasmid (Fig. 1), and these changes were considered not significant. The increases in reporter gene expression from five plasmids, plasmids 11, 43, 73, 74, and 94, were greater than two times the standard deviation of that from pGL2P (Fig. 1), and the DNA inserts in these five plasmids were believed to have the ability to mediate STAT5 activation of gene expression and were selected to be sequenced. By aligning the sequences of these inserts with the sequence (GenBank TM accession number AC010202) of the 170-kb BAC insert, the inserts in plasmids 11, 43, 73, and 94 were found to belong to the 5Ј-flanking region of IGF-I gene and the insert in plasmid 74 corresponded to IGF-I exon 3 and its downstream region. The inserts in plasmids 73 and 94 were further found to overlap over a 1.2-kb DNA fragment that is located ϳ75 kb 5Ј from IGF-I exon 1, corresponding to the region between nucleotide 6804 and nucleotide 8046 in GenBank TM accession number AC010202 ( Fig. 2A).
A distal 5Ј-flanking region of IGF-I gene, we suspected that this overlapped region might be the reason why these two plasmids could mediate STAT5 activation of gene expression. We analyzed this overlapped 1.2-kb DNA sequence for putative STAT5-binding sites using the MATCH program (27). Two potential STAT5-binding sites were identified, being 100% identical to the consensus STAT5-binding sequence TTCC/tT/ cagG/aGAA (19,28) and located ϳ200 bp apart (Fig. 2B). Containing two closely located consensus STAT5-binding sites suggests that the inserts in plasmids 73 and 94 may mediate STAT5 activation of reporter gene expression through direct interactions with STAT5 at these sites. We also searched the inserts in plasmids 11, 43, and 74 for potential STAT5-binding sites. Plasmids 11 and 74 each contained one consensus STAT5-binding site (sequence not shown), suggesting that these plasmids may also mediate STAT5 transactivation via direct binding of STAT5. Plasmid 43 did not contain a consensus STAT5-binding site; the DNA insert in this plasmid may contain an unconventional STAT5-binding site, or it may mediate transactivation through a second transcription factor that is activated by STAT5.
A computer search of the 170-kb BAC sequence revealed 29 putative STAT5-binding sites (TTCC/tT/cagG/aGAA) in addition to the five STAT5-binding sites identified by the transfection analysis. One of these 29 putative STAT5-binding sequences was located in the IGF-I proximal promoter region (at 83472 in GenBank TM accession number AC010202), but the IGF-I promoter containing this putative STAT5-binding sequence did not respond to constitutively active STAT5 protein in the transient transfection analysis (Fig. 3). This result together with the fact that 115 shotgun plasmids of the 170-kb BAC sequence were analyzed in the transfection analysis suggest that most of these 29 putative STAT5-binding sites probably do not have the ability to mediate STAT5 activation of gene expression, perhaps because they are not located in the right sequence context (29). However, it is also possible that some of these 29 putative STAT5-binding sites were missed in the transfection analysis and that they may be functional STAT5-binding sites.
A 700-bp Distal 5Ј-Flanking Region of IGF-I Gene Including the Two Putative STAT5-binding Sites Shared by Plasmids 73 and 94 Is Evolutionarily Conserved-The sequences of functionally important regulatory DNA regions are often conserved during evolution (30 -35). We next determined whether the sequence of the DNA region shared by plasmids 73 and 94 and the sequences of the two putative STAT5-binding sites are conserved among the genomes of different species. By aligning the sequence of this overlapped DNA region with that of the mouse and rat genomes in GenBank TM , a ϳ700-bp DNA region shared by plasmids 73 and 94 was found to have a highly homologous region (Ͼ80% identity) at 71 kb from IGF-I exon 1 in the mouse genome and at ϳ39 kb from IGF-I exon 1 in the rat genome (Fig. 2B). The GenBank TM contains little information on the 5Ј-flanking regions of IGF-I genes of other species. However, using PCR, from a BAC clone containing the bovine IGF-I gene (RP42-161J14), which was isolated by screening a bovine BAC library with a bovine IGF-I promoter-specific probe, we were able to amplify a bovine DNA region that is more than 90% identical to the 700-bp distal 5Ј-flanking region of human IGF-I gene (Fig. 2B), suggesting that the 700-bp 5Ј-flanking region of IGF-I gene is also conserved in the bovine genome. A similar sequence alignment, however, revealed that none of the putative STAT5-binding sites in plasmids 11, 43, and 74, which were identified in the transfection analysis (Fig.  1), and the 29 additional putative STAT5-binding sites in the 170-kb BAC sequence, which were identified based on their sequences, were located in homologous regions between the human genome and the mouse genome or the rat genome.
The 700-bp Distal 5Ј-Flanking Region of IGF-I Gene Can Function as a STAT5-responsive Enhancer-To determine whether the evolutionarily conserved, STAT5-binding site-con- Plasmid pGL2P is an enhancer-less plasmid containing SV40 promoter; 700bp-pGL2P contains the 700-bp distal 5Јflanking region of IGF-I gene compared with pGL2P. Compared with pGL2P, IGF-Intron2-pGL2P contains human IGF-I intron 2 region, whose corresponding rat region has recently been found to contain GH-responsive STAT5-binding sites (21). IGFP1-pGL2B is an enhancer-less plasmid containing IGF-I promoter 1; 700bp-IGFP1-pGL2B contains the 700-bp region, compared with IGFP1-pGL2B. taining 700-bp distal 5Ј-flanking region of IGF-I gene can mediate transactivation by STAT5, the 700-bp DNA region was inserted 5Ј to the SV40 promoter in the enhancer-less pGL2P plasmid to generate plasmid 700bpIGF-pGL2P. As shown in Fig. 3, the reporter gene expression from the SV40 promoter in this plasmid was increased (p Ͻ 0.01) by the constitutively active STAT5a mutant STAT5a-N642H (Fig. 3A) or the constitutively active STAT5b mutant STAT5b CA (Fig. 3B). In the same cotransfection analyses, a human IGF-I intron 2 region corresponding to the recently identified GH-responsive rat IGF-I intron 2 region (21) was unable to mediate STAT5a (Fig.  3A) or STAT5b (Fig. 3B) activation of reporter gene expression.
To determine whether the 700-bp DNA region can also mediate STAT5 activation of gene transcription from IGF-I promoter, cotransfection analyses were performed on plasmids IGFP1-pGL2B and 700bp-IGFP1-pGL2B, each containing a human IGF-I promoter. As can be seen from Fig. 3, STAT5a-N642H (Fig. 3A) or STAT5b CA (Fig. 3B) activated reporter gene expression from these constructs, indicating that the 700-bp DNA region can mediate STAT5 activation of gene transcription from IGF-I promoter. However, the extent to which 700bp-IGFP1-pGL2B was activated by STAT5a-N642H or STAT5b CA was less than the extent to which 700bpIGF-pGL2P was activated (Fig. 3). The reason for this difference was perhaps because the 940-bp IGF-I promoter lacked sufficient DNA elements to fully cooperate with the 700-bp enhancer or because the MAC-T cells used in this transfection analysis did not contain all of the protein factors to cooperate with the STAT5 protein.
We also determined whether the 700-bp DNA region can function as a promoter. Insertion of the 700-bp region or a 1.6-kb DNA fragment containing this 700-bp region from plasmid 73 into promoter-less pGL2B did not increase reporter gene expression from this plasmid in HepG2, CHO, and MAC-T cells (data not shown), indicating that the 700-bp DNA region is unlikely a promoter or part of a promoter.
The Two Putative STAT5-binding Sites in the Distal 5Ј-Flanking Region of IGF-I Gene Can Bind to STAT5 Proteins in Vitro-To determine whether the two putative STAT5-binding sites within the 700-bp distal 5Ј-flanking region of IGF-I gene can bind to STAT5a or STAT5b protein, electrophoretic mobility shift assay were performed using an oligonucleotide probe corresponding to either STAT5-binding site and nuclear protein extracts from cells transfected with either STAT5a-N642H or STAT5b CA or from untransfected cells. As shown in Fig. 4A, incubation of the oligonucleotide probe corresponding to either putative STAT5-binding site in the 700-bp DNA region with nuclear proteins from cells transfected with STAT5a-N642H resulted in a new DNA-protein complex (denoted A in Fig. 4A), compared with the incubation with nuclear proteins from the untransfected cells (Fig. 4A). This new DNA-protein complex was partially disrupted by preincubation of the nuclear proteins with an antibody against STAT5a (Fig. 4A), indicating the presence of STAT5a protein in this DNA-protein complex. When incubated with nuclear proteins from cells overexpressing STAT5b CA , the oligonucleotide probe corresponding to either STAT5-binding site also formed a new DNA-protein complex, compared with the incubation of the same probe with nuclear proteins from untransfected cells (Fig. 4B). This new DNA-protein complex (denoted B in Fig. 4B) was supershifted (the supershift was denoted S in Fig. 4B) when an antibody against STAT5b was added to the incubation, indicating the presence of STAT5b protein in this complex. The results of these gel shift assays demonstrate that the two putative STAT5-binding sites in the 700-bp distal 5Ј-flanking region of IGF-I gene can bind to STAT5a and STAT5b proteins at least in vitro.
The Two STAT5-binding Sites Are Necessary for the 700-bp Distal 5Ј-Flanking Region of IGF-I Gene to Mediate STAT5 Activation of Reporter Gene Expression-To determine whether the two STAT5-binding sites are necessary for the 700-bp DNA region to mediate STAT5 activation of gene expression, cotransfection analyses were performed on reporter gene plasmids containing mutation in either or both STAT5-binding sites. As shown in Fig. 5, the ability of the 700-bp distal 5Јflanking region of IGF-I gene to mediate STAT5a-N642H (Fig.  5A) or STAT5b CA (Fig. 5B) activation of reporter gene expression was inhibited (p Ͻ 0.01) when either or both STAT5binding sites were mutated in the region, indicating that both STAT5-binding sites are essential for the 700-bp DNA region to mediate STAT5 transactivation and that they probably do so in a cooperative manner.
The Two STAT5-binding Sites in the Context of Chromatin Can Be Bound by STAT5, and the Binding Is Associated with Increased Expression of IGF-I mRNA-To further determine whether the two STAT5-binding sites in the distal 5Ј-flanking region of IGF-I gene can bind to STAT5 protein when they are in the context of chromatin, human hepatoma-derived HepG2 cells were transfected with constitutively active STAT5b CA expression plasmid, and binding of STAT5b CA to the 700-bp distal 5Ј-flanking region of IGF-I gene in these cells was analyzed with ChIP assay. As shown in Fig. 6A, from the STAT5b CAtransfected HepG2 chromatin, anti-STAT5b antibody precipitated more of the STAT5-binding site-containing distal 5Јflanking region of IGF-I gene than preimmune serum, whereas it did not recover IGF-I exon 4 DNA that does not contain a STAT5-binding site (the recovery of IGF-I exon 4 DNA by preimmune serum was probably nonspecific). These results indicate that STAT5b can bind to the two STAT5-binding sitecontaining distal 5Ј-flanking region of IGF-I gene in HepG2 cells, perhaps at the two STAT5-binding sites. A real time reverse transcription-PCR analysis revealed that HepG2 cells transfected with STAT5b CA expressed approximately three times more IGF-I mRNA than untransfected HepG2 cells (Fig.  6B), whereas transfected and untransfected HepG2 cells had the same levels of GAPDH mRNA (data not shown). Increased expression of IGF-I gene may be the result of binding of active STAT5b to the STAT5-binding site-containing distal 5Ј-flanking region of IGF-I gene.
The Distal 5Ј-Flanking Region of IGF-I Gene Can Mediate GH Stimulation of Reporter Gene Expression in a STAT5-binding Site-dependent Manner-The above experiments demonstrated that the two STAT5-binding sites in the 700-bp distal 5Ј-flanking region of IGF-I gene can bind to STAT5 protein and that binding of STAT5 protein to them increases IGF-I gene expression. We next asked whether the 700-bp DNA region containing these two STAT5-binding sites can mediate GHinduced STAT5 activation of gene expression. Because a natural GH-responsive cell system is not available, we tested the ability of the 700-bp DNA region to mediate GH-induced gene transcription by cotransfection of the 700-bp DNA region-containing reporter gene construct with a GHR expression plasmid into cells. As shown in Fig. 7, GH caused a moderate but statistically significant (p Ͻ 0.05) increase in reporter gene expression from the TK promoter linked to the 700-bp DNA FIG. 5. The two STAT5-binding sites are required for the distal 5-flanking region of IGF-I gene to mediate STAT5 activation of gene expression. In reporter gene construct 700bpIGF-pGL2P, the two STAT5-binding sites are intact; in 700bpm1-pGL2P, the first STAT5-binding site is mutated; in 700bpm2-pGL2P, the second STAT5binding site is mutated; and in 700bpm3-pGL2P, both STAT5-binding sites are mutated. These plasmids were each cotransfected with STAT5a-N642H (A) or STAT5b CA (B) expression plasmid as described in Fig. 3. region in the presence of either wild-type STAT5a or wild-type STAT5b, whereas GH had no effect on reporter gene expression when the TK promoter was not linked to the 700-bp DNA region. Furthermore, mutation of either or both STAT5-binding sites completely abolished the GH response (Fig. 7), indicating that the 700-bp distal 5Ј-flanking region of IGF-I gene mediated GH-induced STAT5 activation of gene expression through the two STAT5-binding sites. DISCUSSION Because of lack of convenient GH-responsive cell systems, it has been difficult to directly identify the cis-regulatory DNA regions that mediate GH regulation of IGF-I gene expression. In this study we took an indirect approach to identify these cis-regulatory DNA regions. Based on recent findings that GH regulation of IGF-I gene expression in the liver depends on STAT5b (14,15,20) and STAT5a (20), we screened the entire human IGF-I gene and an extensive 5Ј-flanking region of this gene for sequences that can function as STAT5-binding enhancers through cotransfection analysis of shotgun DNA fragments of an IGF-I DNA-containing BAC clone in the presence of constitutively active STAT5a (24) or STAT5b mutant (16). A 700-bp DNA region 75 kb upstream of IGF-I exon 1 was found to have the ability to mediate STAT5 activation of gene expression. This region contains two closely located STAT5-binding sites that were demonstrated to be able to bind to STAT5 proteins. Binding of STAT5 to this region increased gene expression from both heterologous and homologous promoters. Binding of STAT5 to the region in chromatin was associated with increased IGF-I mRNA expression. These results indicate that this two STAT5-binding site-containing distal 5Ј-flanking region of IGF-I gene may be a cis-regulatory region that mediates GH-induced STAT5 activation of IGF-I gene expression.
Recently, an intron 2 region of the rat IGF-I gene was found to contain two STAT5-binding sites that mediate GH-induced STAT5 activation of IGF-I gene expression in the rat liver (21). However, the corresponding human IGF-I intron 2 region appears to contain only one STAT5-binding site. In the cotransfection analysis, the human IGF-I intron 2 region was unable to enhance gene expression in the presence of constitutively active STAT5 protein. These observations suggest that, unlike the rat IGF-I intron 2 region, the human IGF-I intron 2 region probably does not play an important role, if it plays a role, in mediating GH-induced STAT5 activation of IGF-I gene expression. Compared with the intron 2 region, the distal 5Ј-flanking region of human IGF-I gene identified in this study contains two perfect STAT5-binding sites that can bind to STAT5 and mediate STAT5 activation of gene expression. Therefore, GHinduced STAT5 activation of IGF-I gene expression in humans is more likely mediated by this distal 5Ј-flanking region than by the intron 2 region.
The sequences of functionally important regulatory DNA regions are often conserved during evolution (30 -35). The corresponding STAT5-binding distal 5Ј-flanking regions of IGF-I genes from rat, mouse, human, and bovine are highly similar in sequence, being 85% identical over a 700-bp region. In contrast, the sequences of the corresponding IGF-I intron 2 regions are only moderately homologous among these species, being 64% identical over a 160-bp region (data not shown). Thus, we speculate that the STAT5-binding distal 5Ј-flanking region of IGF-I gene identified in this study may be an enhancer mediating GH-induced STAT5 activation of IGF-I gene expression in a wide variety of species. If this distal 5Ј-flanking region of IGF-I gene indeed mediates GH stimulation of IGF-I gene expression in mice and rats, it may explain why previous searches of IGF-I DNA regions proximal to the transcription start site failed to identify GH-responsive regulatory regions (7,8), why reporter gene constructs containing the IGF-I promoter and proximal introns were not GH-responsive in the liver of transgenic mice (10), and why screening a 30-kb rat 5Ј-flanking region of IGF-I gene did not detect any GH-responsive DNAprotein interactions in the liver (4).
In summary, the results from this study strongly suggest that the two STAT5-binding site-containing distal 5Ј-flanking region of IGF-I gene is a cis-regulatory DNA region that mediates GH-induced STAT5 activation of IGF-I gene expression in humans and perhaps in other species as well. It remains to be determined whether this distal STAT5-binding enhancer is FIG. 7. The distal 5-flanking region of IGF-I gene can mediate GH-induced STAT5 activation of gene expression. Plasmid pGL2TK is an enhancer-less plasmid containing TK promoter; 700bp-pGL2TK contains the 700-bp distal 5Ј-flanking region of IGF-I gene compared with pGL2TK. In constructs 700bpm1-pGL2TK, 700bpm2-pGL2TK, and 700bpm3-pGL2TK, STAT5-binding site 1, site 2, or both STAT5-binding sites are mutated. MAC-T cells were transfected with each reporter gene plasmid, a plasmid expressing bovine GHR, and a plasmid expressing the wild-type STAT5a (A) or STAT5b (B). Twentyfour h after transfection, the cells were serum-starved for 16 h, followed by GH or vehicle treatment for 8 h before dual luciferase assay. The values are expressed as the means Ϯ S.E. (n ϭ 3). The means with different letters are statistically different (p Ͻ 0.05). sufficient for GH stimulation of IGF-I gene expression and how the interaction between this distal enhancer and STAT5 protein activates IGF-I gene transcription. Given its distance from the IGF-I gene, this STAT5-binding enhancer can be categorized as a long range enhancer. An increasing number of such long range enhancers have been identified (36 -41). Several mechanisms have been proposed for how long range enhancers affect transcription (40 -43), including DNA looping and direct interaction with RNA polymerase II and indirect interaction with RNA polymerase II through mediators, chromatin remodeling complexes, or transcription factors binding to the proximal promoter. Which of these mechanisms is used by the distal enhancer and STAT5 protein in mediating GH stimulation of IGF-I transcription remains to be tested.