Transcriptional Regulation of Insulin-like Growth Factor-I by Interferon-γ Requires STAT-5b*

Insulin-like growth factor (IGF)-I, a growth factor important for cell proliferation, cellular differentiation, and multiple metabolic functions, is regulated in vivo by growth factors and cytokines, but the mechanism(s) of regulation at the cellular level is not well understood. We now demonstrate, employing primary human dermal fibroblasts (CF), that the multipotent cytokine, interferon-γ (IFN-γ), can up-regulate IGF-I mRNA expression and that this regulation occurs via activation of the signal transducer and activator of transcription-5b (STAT-5b) pathway. IFN-γ (100 units/ml) treatment of CF cells resulted in a preferential, time-dependent activation of STAT-5b, although both STAT-5a and STAT-5b isoforms are present. The activated STAT-5b translocated to the nucleus within 30 min of treatment and induced an increase in IGF-I mRNA of 6 ± 1.0-fold, 3 h post-treatment, with a further increase to 8 ± 1.7-fold at 5 h. In contrast, IFN-γ treatment of primary human dermal fibroblasts with a nonfunctional STAT-5b (PF cells) resulted in activation of only STAT-5a and an increase of the IGF-I mRNA level of 1.7 ± 0.6-fold, 5 h post-treatment. The IFN-γ-induced regulation of the interferon regulatory factor-1 gene, whose expression is dependent on activated STAT-1, was the same between CF and PF cells. In summary, our results provide evidence of the following in human primary dermal fibroblasts: (a) IFN-γ preferentially activates STAT-5b, but, in the absence of a functional STAT-5b, STAT-5a is activated; (b) IFN-γ time-dependently up-regulates IGF-I mRNA expression; (c) the regulation of IGF-I requires an active STAT-5b, and activated STAT-5a cannot substitute for an inactive STAT-5b; and (d) STAT-5b has an essential role in the transcriptional up-regulation of IGF-I.

Interferon-␥ (IFN-␥) 1 is a cytokine with pleiotropic effects on immune (e.g. macrophages) and nonimmune (e.g. fibroblasts and epithelial) cells. The biological effects, which include regulation of cell proliferation, immune surveillance, and tumor suppression (1,2), are mediated through the regulation of gene expression, via the Janus kinase (JAK)-signal transducers and activators of transcription (STAT) signaling pathways. The JAK-STAT-1 pathway has been shown to induce expression of gene products that are responsible for the majority of the pleiotropic effects of IFN-␥ (3). Signal transduction is initiated when IFN-␥ binds to a heterodimeric complex of IFN-␥ receptors consisting of two IFN-␥R1 and two IFN-␥R2 subunits per complex. Association of JAK1 and JAK2 subsequently phosphorylates Tyr 440 (in humans) on the IFN-␥R1 subunits, creating docking sites for STAT-1 (4,5). The recruited STAT-1, which binds through its v-src homology-2 domain, is tyrosine-phosphorylated by JAK-2, dimerizes, and translocates to the nucleus, where the activated dimer functions as a transcription factor (6). In addition to STAT-1, STAT-3 and STAT-5a/b can also be activated by IFN-␥ (7,8). The biological consequences of IFN-␥-induced activation of STAT-3 and -5, however, are not well understood.
We recently demonstrated that normal, human, primary dermal fibroblasts responded to growth hormone (GH) as well as IFN-␥ (9). GH treatment induced the up-regulation of insulinlike growth factor-I (IGF-I), a secreted peptide important for somatic growth (10), and of IGF-binding protein-3 (IGFBP-3) mRNA through activation of the JAK-STAT signaling pathways (9). In particular, STAT-5b was implicated as a critical factor for the transcriptional regulation process, based on analysis of dermal fibroblasts from a patient with severe GH insensitivity resulting from an autosomal recessive mutation in the STAT-5b gene. As a consequence, the GH signaling pathway was disrupted, and IGF-I and IGFBP-3 expression was dysregulated, thereby demonstrating the importance of STAT-5b for GH-induced regulation of IGF-I and IGFBP-3 (9). These results raised the question of whether other cytokines capable of activating STAT-5b can also regulate IGF-I mRNA.
Limited data exist on the effects of cytokines on IGF-I expression. In murine macrophages, IFN-␥ was shown to transcriptionally down-regulate IGF-I (11) through a STAT-1-dependent mechanism (12), and IFN-␣/␤ down-regulated IGF-I mRNA expression in rat glioma cells (13). More recently, tumor necrosis factor-␣ (TNF-␣) was demonstrated to similarly downregulate IGF-I mRNA in rat myoblasts (C2C12) (14). Interestingly, the down-regulation was suggested to be mediated via a Jun N-terminal kinase pathway, but independent of STAT-5 phosphorylation (14). We now demonstrate that, in contrast to the murine cell models, IFN-␥, similar to GH, up-regulates IGF-I mRNA in human dermal fibroblasts and that this upregulation is STAT-5b-dependent.

EXPERIMENTAL PROCEDURES
Antibodies-Antibodies directed against tyrosine-phosphorylated STAT-1 and tyrosine-phosphorylated STAT-5 were purchased from Cell Signaling Technology (Beverly, MA). Antibodies against STAT-5a (L-* This work was supported by National Institutes of Health Grant CA58110 (to R. G. R.). 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 U.S.C. Section 1734 solely to indicate this fact.
Cell Culture-Primary fibroblast cultures, established from skin biopsies taken from a GH-insensitive patient (PF cells) and a normal 30-year-old female (CF cells), have been previously described (9). PF cells were derived from a patient homozygous for a missense mutation that resulted in a Pro to Ala substitution at amino acid 630 (A630P) in the STAT-5b gene (9). Additional normal human dermal fibroblasts from a 31-year-old female were purchased from BioWhittaker (Walkersville, MD) (NHDF 8560). The fibroblasts were cultured as previously described (9). COS-7 cells were from ATCC and were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen). For all experiments, cells were serum-starved 24 h prior to treatment with either IFN-␥ (Roche Applied Science) or recombinant human GH (a generous gift from Genentech, Inc., South San Francisco, CA).
Transfection Experiments-COS-7 cells were grown to ϳ50% confluence in 100-mm plates. Cells were transiently transfected with vector, pcDNA3.1, or vector carrying F-STAT-5b or F-STAT-5b(A630P), using TransIT-LT1 (Mirus, Madison, WI). After a 24-h transfection, cells were washed and serum-starved (Dulbecco's modified Eagle's medium with 0.1% bovine serum albumin) for a further 24 h prior to treatment with IFN-␥ (100 units/ml). Total RNA and cell lysates were collected 1 h post-treatment. In some cases, total cell lysates (ϳ320 g) from the transfected COS-7 cells were immunoprecipitated with anti-FLAG-M2agarose gel by standard methods prior to Western immunoblot analysis.
Real Time Quantitative (RTQ)-PCR-RTQ-PCR was performed as previously described (9). Briefly, total RNA samples were extracted from primary fibroblast cell cultures exposed to 100 units/ml IFN-␥ for the times indicated. The experiment was performed four independent times in duplicate. RTQ-PCR for each sample was done in triplicate, with 5-l aliquots of the diluted (1:1) RT reaction. RTQ-PCR was carried out on a Prism 7700 sequence detector (PerkinElmer Life Sciences) using Taqman  Western Immunoblot Analysis-Cells were solubilized in radioimmune precipitation assay lysis buffer (1ϫ phosphate-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 10 mg/ml phenylmethylsulfonyl fluoride, 100 mM sodium orthovanadate, and protease inhibitor mixture (Roche Applied Science)). Equal quantities of protein (Bio-Rad protein assay) were size-fractionated on reducing 7 or 13% SDS-polyacrylamide gels, and electroblotted onto nitrocellulose membranes. Western blots were processed with the appropriate primary and secondary antibodies, following the manufacturers' protocols, and visualized by enhanced chemiluminescence (PerkinElmer Life Sciences).
In some cases, single-comb preparative wells (Bio-Rad) were employed for protein fractionation by SDS-PAGE. A total of 250 g of protein lysate was loaded into the single well protein gel and processed as for immunoblot analysis. The resultant membrane was marked prior to cutting of the membrane into strips for immunoblotting with the panel of anti-STAT-5 antibodies. The strips were subsequently reassembled for visualization by enhanced chemiluminescence.

IFN-␥ Specifically Activates STAT-5b in Normal Dermal
Fibroblasts-We recently demonstrated that IFN-␥ robustly activates the JAK/STAT signaling pathway in normal human dermal fibroblasts (CF cells), leading to phosphorylation of not only STAT-1 but also STAT-3 (9). The effect of IFN-␥ on the STAT-5 signaling pathway in these cells, however, was not known. CF cells were therefore treated with IFN-␥ (100 units/ ml), and the activation of STAT-5 was examined. The activation of STAT-1 was confirmed, with pSTAT-1 readily detected within 15 min of IFN-␥ treatment (Fig. 1A). STAT-5, like STAT-1 (and STAT-3; data not shown), was rapidly tyrosinephosphorylated, and phosphorylation was sustained for at least 120 min (Fig. 1B). Interestingly, compared with growth hormone-treated CF cells (9), the activation of STAT-5 by IFN-␥ was considerably more robust (Fig. 1C). Increasing concentrations of GH did not enhance phosphorylation of STAT-5.
Although tyrosine phospho-STAT-5 was detected upon IFN-␥ treatment, it was unclear initially which isoform (STAT-5a or STAT-5b) was phosphorylated. Both are present in CF cells (9) and have similar molecular mass (STAT-5a, 91 kDa; STAT-5b, 84 kDa). Cell lysates were further analyzed by immunoblotting with a panel of anti-STAT-5 antibodies (Fig. 1D). As shown, STAT-5a and STAT-5b were successfully separated, and the specific isoform of STAT-5 that was phosphorylated could be determined. Under uninduced conditions, neither STAT-5a nor STAT-5b was activated (top panel). Upon IFN-␥ treatment (bottom panel), it was apparent that only STAT-5b was phosphorylated (Fig. 1D).
IFN-␥-induced Nuclear Localization of pSTAT-5-Activated STATs are known to translocate from the cytoplasm to the nucleus. Immunofluorescent analysis of treated and untreated CF fibroblasts indicated that within 30 min of IFN-␥ treatment, pSTAT-5 accumulated in the nucleus, as did pSTAT-1 ( Fig. 2A). Unlike pSTAT-1, which remained nuclear for at least 90 min, depletion of nuclear pSTAT-5 was observed at 90 min. It is noted that for growth hormone-treated cells, the immunofluorescent analysis was not sufficiently sensitive to detect nuclear pSTAT-5 (Fig. 2B). The results correlated with the robust activation of STAT-5b by IFN-␥, compared with that by growth hormone. Altogether, these results suggest that in dermal fibroblasts, of the two STAT-5 isoforms, IFN-␥ preferentially activates STAT-5b, which subsequently translocates to the nucleus.
IFN-␥ Up-regulates IGF-I mRNA Expression-Recently, we demonstrated that the transcriptional regulation of IGF-I by growth hormone in CF cells was STAT-5b-dependent (9). We therefore hypothesized that since IFN-␥ preferentially activates STAT-5b over STAT-5a, IGF-I may be a target gene for IFN-␥-activated pSTAT-5b. Preliminary analysis by semiquantitative RT-PCR amplification supported this hypothesis, with IGF-I mRNA observed to be up-regulated 3 and 5 h post-IFN-␥ treatment (Fig. 3A). IRF-1, a STAT-1-dependent transcription factor known to be up-regulated by IFN-␥ (17), was up-regulated within 1 h. In contrast, expression of IGFBP-3 mRNA was not altered by IFN-␥ treatment (Fig. 3A).
RTQ-PCR confirmed that IGF-I mRNA was up-regulated upon IFN-␥ treatment. As shown in Fig. 3B, IGF-I mRNA progressively increased in a time-dependent manner, with 6 Ϯ 1.0-fold (S.E.) and 8 Ϯ 1.7-fold induction 3 and 5 h post-treatment, respectively. Similar results were obtained from a second (commercially purchased) normal dermal fibroblast cell line (data not shown), demonstrating that it was not a phenomenon unique to our CF cells. In cells treated with IFN-␥ for 24 or 48 h, IGF-I mRNA levels were similar to those of untreated cells (data not shown), suggesting that the up-regulation is transient. The up-regulation of IGF-I mRNA, furthermore, was not dependent on de novo protein synthesis, since the induction was not inhibited by cycloheximide treatment (data not shown). Altogether, the results demonstrate that IGF-I is a FIG. 1. IFN-␥ activates tyrosine phosphorylation of STAT-5. Primary dermal fibroblasts were treated with 100 units/ml IFN-␥ or 500 ng/ml GH, and cell lysates were collected at the times indicated. Western immunoblot analysis was performed to determine the phosphorylation status of STAT-1 and STAT-5. A, IFN-␥-treated cells were immunoblotted for tyrosine phospho-STAT-1 (upper panel); both the 91-and 84-kDa pSTAT-1 forms were detected. The blot was then stripped and reprobed with antibody against total STAT-1 (bottom panel). B, immunoblot of IFN-␥-treated cells; phospho-STAT-5 (upper panel) was stripped and reprobed using antibody specific for STAT-5b. The tyrosine phospho-STAT-5 antibody cannot distinguish between pSTAT-5a and pSTAT-5b. C, cells were treated with increasing concentrations of GH or with IFN-␥ for 30 min, and cell lysates were immunoblotted for pSTAT-5. The nonspecific immunoreactive band (arrow) indicates that protein loading was equivalent. D, immunoblot analysis for the STAT-5 isoform that is activated by IFN-␥ in cell lysates from CF cells. 250 g of cell lysates (untreated or IFN-␥treated, 60 min) were loaded into single preparative wells, size-fractionated by 7% SDS-PAGE, and electroblotted to nitrocellulose membranes, and membranes were cut into strips. Each strip was immunoblotted with the indicated antibody and reassembled prior to enhanced chemiluminescence treatment. Correct alignment of the reassembled blot was achieved by marking the blot with reference lines drawn prior to cutting of the membranes for immunoblotting. STAT-5a and STAT-5b bands are indicated by arrows.
target gene of IFN-␥ signal transduction and that the increase in IGF-I mRNA is likely to be the result of IFN-␥ activation of STAT-5b (see below). of IGF-I mRNA, we compared the response of CF cells with that of a human primary dermal fibroblast cell line carrying a naturally occurring, nonfunctional STAT-5b (9). We recently characterized these novel fibroblasts, designated PF cells, and demonstrated that they carry a homozygous missense mutation in the STAT-5b gene, which results in expression of an aberrant STAT-5b protein (schematically presented, Fig. 4A) (9). The mutated protein, designated STAT-5b(A630P), contains an Ala 630 3 Pro substitution within the v-src homology-2 domain. Unlike wild-type STAT-5b, the STAT-5b(A630P) is poorly detected by immunoblotting techniques employing STAT-5b-specific antibody (Fig. 4B, top panel) and could not be activated by the growth hormone receptor signaling pathway (9). The PF cells, however, had concentrations of STAT-5a comparable with that in CF cells (Fig. 4B, bottom panel). In addition, IFN-␥ induced a phosphorylation pattern (STAT-1, STAT-3, and extracellular signal-regulated kinase 1/2) in PF cells identical to that observed in CF fibroblasts (9). The PF cells were, therefore, ideal for determining whether the upregulation of IGF-I mRNA by IFN-␥, like growth hormone, is STAT-5b-dependent.

IFN-␥ Activates
Treatment of PF cells with IFN-␥ unexpectedly resulted in a detectable phospho-STAT-5 time-dependent pattern similar to that observed in CF fibroblasts (Fig. 4C, top panel). This intriguing result suggested that either STAT-5a was now phosphorylated (unlike in CF cells) or that the mutant STAT-5b(A630P) could be activated by select cytokines such as IFN-␥ but not GH (Fig. 4C, bottom panel) (9). Detailed analysis of the phospho-STAT-5 product in PF cells (Fig. 4D) indicated the former hypothesis was correct (i.e. STAT-5a was preferentially phosphorylated in PF cells).
Recombinant FLAG-STAT-5b(A630P) Is Not Activated by IFN-␥-To further support the observation that mutant STAT-5b(A630P) cannot be activated by IFN-␥ (9), the STAT-5b cDNAs, cloned from CF and PF cells (9), were N-terminally FLAG-tagged for expression in the COS-7 cell system. COS-7 cells have undetectable STAT-5b, and our initial characterization of COS-7 cells indicated that IFN-␥, but not growth hormone, induced the JAK/STAT signaling cascade (9) (data not shown), confirming the presence of sufficient functional endogenous IFN-␥ receptors but not growth hormone receptors. COS-7 cells were transfected with vector (pcDNA3.1) or vector carrying either wild-type FLAG-STAT-5b or FLAG-STAT-5b(A630P) and treated with IFN-␥. The efficiency of transfection and mRNA expression was equivalent, as determined by limiting RT-PCR analysis (Fig. 5A). As shown in Fig. 5B (lanes  1 and 2), IFN-␥ induced phosphorylation of endogenous STAT-5a in nontransfected cells. An equivalent pSTAT-5 pattern was detected in COS-7 cells transfected with pcDNA3.1 (Fig. 5B, lanes 3 and 4). Overexpression of wild-type FLAG-STAT-5b resulted in low, basal, constitutive phosphorylation of the recombinant STAT-5b (Fig. 5B, lane 5). Treatment with IFN-␥ (1 h) significantly increased detectable phospho-FLAG-STAT-5b (Fig. 5B, lane 6). In contrast, cells transfected with FLAG-STAT-5b(A630P) cDNA generated a phospho-STAT-5 pattern (Fig. 5B, lanes 7 and 8) similar to untransfected and vector-transfected cells. Interestingly, the recombinant human FLAG-STAT-5b proteins (both wild-type and mutant) were readily distinguishable from endogenous STAT-5a, since the recombinant proteins ran at a higher molecular weight on SDS-PAGE than does the endogenous protein. Immunoblotting with anti-STAT-5a indicated that total protein loading was the same and that endogenous STAT-5a was unaffected by the transfection process. Sequential stripping and reprobing of the pSTAT-5 blot with anti-FLAG antibody and anti-STAT-5b-specific antibody (Fig.  5B) indicated that the mutant STAT-5b protein was consistently less well immunodetected than wild-type (an ϳ4-fold difference in detection). This difference in detection may account for the inability to detect phospho-FLAG-STAT-5b(A630P). Therefore, increased loading of cell lysates from COS-7 cells overexpressing FLAG-STAT-5b(A630P) (immunologically equivalent to cell lysates of FLAG-STAT-5b protein) were analyzed, and as shown in Fig. 5C, phosphorylated mutant STAT-5b(A630P) was still not detectable. Finally, the overexpressed recombinant proteins were concentrated by immunoprecipitation with anti-FLAG antibody, but, again, only phosphorylated wild-type STAT-5b was detectable (Fig. 5D). Altogether, these results demonstrate that mutant STAT-5b(A630P) is not activated by IFN-␥ treatment.
IGF-I mRNA Is Not Up-regulated by IFN-␥ in PF Cells-Since IFN-␥, like GH, cannot activate mutant STAT-5b(A630P), we determined whether the IFN-␥-induced transcriptional up-regulation of IGF-I expression was affected in PF cells. Compared with CF cells, where induction of IGF-I mRNA was 8 Ϯ 1.7-fold (see also Fig. 3B), up-regulation of IGF-I mRNA was not detected in PF cells treated with IFN-␥ (Fig. 6A). Over the 5-h time course, induction of IGF-I mRNA levels remained less than 2-fold in PF cells. Similar results were obtained with CF and PF cells treated with other cytokines such as IFN-␣/␤ (data not shown). In contrast, IFN-␥ induced similar phospho-STAT-1 profiles (Fig. 6B) and equivalent up-regulation of IRF-I, at the mRNA (Fig. 6C) and protein levels (Fig. 6D) in CF and PF cells. Altogether, the results are Nonspecific immunoreactive bands below the specific pSTAT-5 band indicate that protein loading was equivalent. D, immunoblot analysis for the STAT-5 isoform that is activated by IFN-␥ (60-min treatment) in cell lysates from PF cells, performed as described in the legend to Fig. 1D. consistent with the requirement for an active STAT-5b for the transcriptional regulation of IGF-I by IFN-␥. DISCUSSION The mechanism of regulating IGF-I expression by cytokines in humans has been difficult to study, in part because of the lack of an appropriate cell model. Only two human cell models have been reported. In human umbilical vein endothelial cells, interleukin-1 (IL-1) was shown to induce a transient transcriptional 4.5-fold up-regulation of IGF-I (18), but the mechanism(s) was not demonstrated. In another study, primary dermal fibroblasts were used to investigate the effects of IFN-␥ on IGF-I mRNA expression (19). No regulation was detected, but the study was complicated by the presence of excess fetal bovine serum (10%) in the experimental conditions. The majority of such studies, however, have employed rodent cell models, where cytokines appear to have inhibitory effects on IGF-I mRNA expression (11, 13, 14, 20 -22). For example, IFN-␥ significantly inhibited IGF-I mRNA transcription in a timeand dose-dependent manner in mouse macrophages (11,12), and in rat glioma cells, IFN-␣/␤ also down-regulated IGF-I mRNA expression (13). In contrast, no alterations in IGF-I mRNA were observed in IFN-␥-treated rat aortic vascular smooth muscle cells (21) or C2C12 cells (14,22). Similarly, the cytokine TNF-␣ also inhibited IGF-I expression, as was demonstrated in vascular smooth muscle cells (85% reduction at 24 h) (21) as well as C2C12 cells (14,22). Interestingly, the inhibitory effects of IFN-␣/␤, IFN-␥, and TNF-␣ were sensitive to cycloheximide, suggesting that de novo protein synthesis is required. Other cytokines tested (IL-6 and IL-1␤) in these same rodent cells did not alter IGF-I mRNA expression (21,22). The mechanism(s) of these inhibitions are not known, although recently, the Jun N-terminal kinase pathway was implicated in the TNF-␣-induced down-regulation of IGF-I expression in C2C12 cells (14), and the IFN-␥-induced down-regulation of IGF-I expression in primary mouse macrophages appeared to require STAT-1 (12). In the same bone marrow-derived mouse macrophages, Th2 cytokines such as IL-4 and IL-13 up-regulated IGF-I expression in a STAT-6-dependent manner (12).
An exception to the transcriptional down-regulation of IGF-I mRNA by cytokines is the effects of GH. GH, a class I cytokine, induced the up-regulation of IGF-I mRNA in primary rat hepatocytes (20), C2C12 mouse skeletal myoblasts (14,23,24), and primary human dermal fibroblasts (9), which is consistent with the known in vivo effects of GH in both rodents and humans.
In this report, we demonstrate that IGF-I mRNA expression in primary human dermal fibroblasts is up-regulated in response to IFN-␥. These fibroblasts have been shown to respond to both class I (e.g. IL-4 and GH (9,25,26)) and class II (e.g. IFN␣/␤ and IFN-␥ (9)) cytokines. With IFN-␥ treatment, the increase in IGF-I mRNA was time-dependent but transient, with a 6 Ϯ 1.0-fold induction observed by 3 h post-treatment and a further increase to 8 Ϯ 1.7-fold 5 h after treatment. This is contrary to that observed in rodent macrophages (11), most likely reflecting cell type and/or species differences. The -fold induction was significantly higher than that observed when the dermal fibroblasts were treated with GH (1.8 Ϯ 0.6-fold, 3 h post-treatment (9)). Further, de novo protein synthesis was not required (data not shown), suggesting that positive and negative regulations of IGF-I expression are via different mechanisms. IFN␣/␤ also up-regulated IGF-I mRNA expression to FIG. 5. Human recombinant mutant STAT-5b(A630P) is not activated by IFN-␥ when overexpressed in COS-7 cells. Expression vector, pcDNA3.1, carrying N-terminally FLAG-tagged wildtype STAT-5b cDNA (WT) or mutant STAT-5b(A630P) cDNA (A630), was transfected into COS-7 cells, and total RNA, as well as cell lysates, were collected after IFN-␥ treatment (100 units/ ml, 60 min). A, ethidium-stained agarose gel of PCR products. RT reactions of total RNA collected were employed for limiting PCR amplification. RT reactions were diluted 1:9, 1:49, and 1:99 prior to PCR amplification of FLAG-STAT-5b and 18 S. B, Western immunoblot analysis of cell lysates (30 g/lane) for pSTAT-5. The immunoblot was stripped and reprobed with, consecutively, anti-FLAG, anti-STAT-5b, and anti-STAT-5a. C, immunoblot analysis for phosphorylation of mutant FLAG-STAT-5b(A630P). Immunologically equivalent concentrations of FLAG-STAT-5b (30 g) and FLAG-STAT-5b(A630P) (ϳ120 g) were immunoblotted for pSTAT-5. D, total cell lysates (320 g) were immunoprecipitated (IP) with anti-FLAG antibody and immunoblotted for phospho-STAT-5, stripped, and reprobed with anti-FLAG antibody. Note that the ␣-pSTAT-5 was not completely stripped in lane 4. levels similar to that observed for IFN-␥ (data not shown). Thus, IGF-I mRNA in human dermal fibroblasts not only is readily detectable, but the signaling pathways leading to its regulation are apparently intact.
The differences in induction of IGF-I mRNA expression between IFN-␥ and GH correlated with the degree the JAK-STAT signaling pathway was activated. We previously demonstrated that the phosphorylation of extracellular signal-regulated kinase 1/2 by GH and IFN-␥ was similar, whereas the activation of STAT-1 and STAT-3 by IFN-␥ was considerably more robust than that observed with GH (9). The same trend for the activation of STAT-5 was observed in this report. It is not clear at present whether this is due to a higher concentration of functional IFN-␥ receptors compared with GH receptors. Not only was pSTAT-5 more readily detected when cells were treated with IFN-␥, but immunocytochemical analysis of treated fibroblasts indicated that pSTAT-5 and pSTAT-1 localized to the nucleus after IFN-␥, but not GH, treatment. The lack of detectable nuclear pSTAT-5 (and pSTAT-1) under GH-induced conditions indicates that our immunofluorescent technique was not sufficiently sensitive to detect low concentrations of nuclear phosphorylated STAT proteins. Indeed, nuclear extracts of GHtreated fibroblasts suggest that pSTAT-5 is present in the nucleus (data not shown).
IFN-␥, like GH, specifically activated STAT-5b. This is contrary to suggestions that, of the two isoforms, STAT-5a is preferentially activated by IFN-␥ (7). In human dermal fibroblasts, both STAT-5a and STAT-5b are present, although the relative amount of each is not known. In our human dermal fibroblast cell line that lacked wild-type STAT-5b (PF cells), it was STAT-5a that was equivalently activated by IFN-␥. Since activation of STAT-5a/b involves initial docking of the relevant STAT-5 to Tyr 440 in the STAT recruitment site (SRS) on IF-NGR1 (8), our results suggest an absolute preference for recruitment of STAT-5b over STAT-5a by the IFN-␥ receptors. This preference may be due to intrinsic properties in the structure(s) of the receptor and/or the STAT-5 that dictate docking specificities. Another possibility is that the ratio of STAT-5b to STAT-5a is high in the fibroblasts, and STAT-5b has, therefore, a correspondingly higher probability of being recruited. Investigations are currently under way to evaluate the various possibilities. Significantly, in the PF cells, even as IFN-␥ activates FIG. 6. IFN-␥ induced IRF-I mRNA expression but not IGF-I mRNA expression in PF cells. IFN-␥-induced expression of IGF-I mRNA in PF cells was compared with that in CF cells. A, semiquantitative RT-PCR and RTQ-PCR amplification of IGF-I and 18 S as previously described (Fig. 3B). Each time point in the RTQ-PCR is the mean Ϯ S.E. from four independent experiments done in duplicate. The results for CF cells are as shown in Fig. 3B. B, immunoblot for pSTAT-1, reprobed for total STAT-1. C, fibroblasts were treated for 1 h with IFN-␥, and the expression of IRF-I mRNA, determined by RT-PCR, was comparable between CF and PF cells. D, immunoblot analysis of IRF-I protein induced by IFN-␥ in CF and PF cell lysates.
STAT-5a when wild-type STAT-5b is absent, the GH signaling system, in the same PF cells, does not activate STAT-5a. Hence, in our cell system, STAT-5a cannot substitute for STAT-5b. Furthermore the results suggest that the recruitment of specificity in the STAT-5a/b to receptors. Overall, tyrosine phosphorylation of STAT-5 in response to IFN-␥ is clearly dependent on the cell type, since it has been shown that pSTAT-5 is not detected in HeLa cells, despite the expression of both STAT-5 isoforms (7). In C2C12 cells, GH apparently activates both STAT-5a and STAT-5b (24).
The up-regulation of IGF-I mRNA induced by GH and IFN-␥ in CF cells correlated with the level of activated STAT-5b induced by these cytokines. This, together with a mutant STAT-5b(A630P) that is not activated by IFN-␥ (both in dermal fibroblasts and when overexpressed in COS-7 cells (present work) (9)) and the lack of IFN-␥-induced regulation of IGF-I mRNA in PF cells, confirmed that phosphorylated STAT-5b is critical for the up-regulation of IGF-I mRNA. In addition, the lack of detectable pSTAT-5a in CF cells and the preferential activation of STAT-5a by IFN-␥ in PF cells suggest that pSTAT-5a is unlikely to participate in the regulation of IGF-I mRNA, at least in human dermal fibroblasts. Our results affirm the importance of STAT-5b in up-regulating IGF-I expression, based on cumulative data from rodents (27)(28)(29)(30) and humans (9). Other STATs, such as STAT-6, could also be involved (12), although IFN-␥ is not known to activate STAT-6. In cases where IFN-␥ does not appear to regulate IGF-I expression (C2C12 (22) and vascular smooth muscle cells (21)) or downregulates IGF-I (11), the phosphorylation status of STAT-5a/b was not determined. For the TNF-␣-induced down-regulation of IGF-I expression observed in C2C12 cells, it was shown that no pSTAT-5 was detected, and, when co-treated with GH, TNF-␣ was still able to down-regulate IGF-I, even in the presence of pSTAT-5 (14). It was unclear, however, which isoform of STAT-5 was phosphorylated.
It is apparent that the functions of STAT-5a and STAT-5b are not interchangeable. This is most clearly demonstrated in our use of the PF cells. The PF cells were derived from a patient whom we recently identified as carrying a recessive homozygous mutation in the STAT-5b gene (9). The patient was insensitive to GH and displayed a clinical phenotype consistent with Laron dwarfism. Serum level of IGF-I in the patient was less than 10% of normal (9), unlike the STAT-5b Ϫ/Ϫ murine models, where serum IGF-I was 50 -70% of normal (27). At the molecular level, we have now demonstrated that STAT-5a is not only present in the PF cells and can be activated, but its robust phosphorylation cannot compensate for loss of STAT-5b activity, as demonstrated by the dysregulation of IGF-I expression.
Finally, results implicate a subset of human IFN-␥-regulated, STAT-5b-dependent genes that may be important for the biological actions of IFN-␥. We show in this report that IGF-I is one such gene. The over 500 genes (in rodents) estimated to be regulated by IFN-␥ have been divided into those that are STAT-1dependent and STAT-1-independent (3,31). Although STAT-1 has been regarded as the main mediator of IFN-␥ biological actions, evidence is emerging for the biological consequences of STAT-1-independently regulated genes, such as involvement in antiviral responses (32), and in cell cycle progression (33). The specific IFN-␥-induced, STAT-1-independent pathway(s) are, presumably, the consequence of activated STAT-3, STAT-5, and/or Raf-1 pathways, as was shown for c-myc (33). Further studies employing the PF cells will aid in elucidating the bio-logical importance of STAT-1-independent STAT-5b-dependent, IFN-␥-regulated genes.
In summary, we have demonstrated the following in human primary dermal fibroblasts: (a) IFN-␥, like GH, specifically activates STAT-5b, but, unlike GH, STAT-5a can be recruited and activated by IFN-␥ receptor complexes in the absence of a functional STAT-5b; (b) IFN-␥, like GH, time-dependently upregulates IGF-I mRNA expression; (c) the regulation of IGF-I mRNA requires an active STAT-5b, and pSTAT-5a cannot substitute for an inactive STAT-5b; and (d) STAT-5b has an essential role in the transcriptional up-regulation of IGF-I.