Growth Hormone Preferentially Induces the Rapid, Transient Expression of SOCS-3, a Novel Inhibitor of Cytokine Receptor Signaling*

Four members (SOCS-1, SOCS-2, SOCS-3, and CIS) of a family of cytokine-inducible, negative regulators of cytokine receptor signaling have recently been identified. To address whether any of these genes are induced in response to growth hormone (GH), serum-starved 3T3-F442A fibroblasts were incubated with GH for various time points, and the expression of the SOCS gene family was analyzed by Northern blotting. GH stimulated the rapid, transient induction ofSOCS-3 mRNA, peaking 30 min after the initiation of GH exposure and declining to basal levels by 2 h. Expression of the other SOCS genes (SOCS-1, SOCS-2,CIS) was also up-regulated by GH, although to a lesser extent than SOCS-3 and with differing kinetics.SOCS-3 expression was also strongly induced in 3T3-F442A cells treated with leukemia-inhibitory factor (LIF), with weaker induction of SOCS-1 and CIS being observed. The preferential induction of SOCS-3 mRNA was also observed in hepatic RNA isolated from the livers of mice that had received a single supraphysiological dose of GH intraperitoneally. Co-transfection studies revealed that constitutive expression of SOCS-1 and SOCS-3, but not SOCS-2 or CIS, blocked GH-induced transactivation of the GH-responsive serine protease inhibitor 2.1 gene promoter.


Four members (SOCS-1, SOCS-2, SOCS-3, and CIS) of a family of cytokine-inducible, negative regulators of cytokine receptor signaling have recently been identified.
To address whether any of these genes are induced in response to growth hormone (GH), serum-starved 3T3-F442A fibroblasts were incubated with GH for various time points, and the expression of the SOCS gene family was analyzed by Northern blotting. GH stimulated the rapid, transient induction of SOCS-3 mRNA, peaking 30 min after the initiation of GH exposure and declining to basal levels by 2 h. Expression of the other SOCS genes (SOCS-1, SOCS-2, CIS) was also up-regulated by GH, although to a lesser extent than SOCS-3 and with differing kinetics. SOCS-3 expression was also strongly induced in 3T3-F442A cells treated with leukemia-inhibitory factor (LIF), with weaker induction of SOCS-1 and CIS being observed. The preferential induction of SOCS-3 mRNA was also observed in hepatic RNA isolated from the livers of mice that had received a single supraphysiological dose of GH intraperitoneally. Co-transfection studies revealed that constitutive expression of SOCS-1 and SOCS-3, but not SOCS-2 or CIS, blocked GH-induced transactivation of the GH-responsive serine protease inhibitor 2.1 gene promoter.
The elucidation of the pivotal role played by the tyrosine kinase Jak2 in initiating signal transduction from the GH 1 receptor has led to the identification of a number of intracellular pathways that mediate the cellular response to GH (1). However, the mechanism(s) by which signaling from GH receptor-activated Jak2 is attenuated is unclear. Ligand-induced tyrosine phosphorylation/activation of Jak2 by the erythropoietin (EPO) receptor, a member of the cytokine receptor superfamily that includes the GH receptor, is followed by the binding of the protein-tyrosine phosphatase SHP-1 to the cytoplasmic domain of the receptor (2). The recruitment of SHP-1 is accompanied by the dephosphorylation/inactivation of Jak2 and subsequent termination of EPO-induced cellular proliferation.
A similar role for SHP-1 in mediating the down-regulation of Jak2 following stimulation of cells with GH has been proposed (3), although whether SHP-1 can directly associate with the GH receptor remains to be established. Recently, a novel family of cytokine-inducible genes has been identified that appear to function as negative regulators of the JAK signaling pathway (4 -7). Constitutive expression of one member, SOCS-1 (also referred to as SSI-1 and JAB) in the murine myeloid leukemia M1 cell line blocked growth factor-induced differentiation and apoptosis and inhibited interleukin-6 (IL-6)-mediated tyrosine phosphorylation of the cell-surface receptor component, gp130, and the transcription factor, Stat3 (5, 6). SOCS-1 can interact with all four members of the JAK family of tyrosine kinases (6,7), suppressing kinase activity and the subsequent tyrosine phosphorylation/activation of STAT factors (7). As a diverse array of cytokines appears to be able to induce expression of one or more members of this gene family (5), we sought to establish which, if any, of these genes might be a target for transcriptional activation by GH in vivo and in the classically GH-responsive cell line, 3T3-F442A fibroblasts (8). Cell Culture-3T3-F442A fibroblasts were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, 2 mM glutamine, 50 IU/ml penicillin, and 50 mg/ml streptomycin in a humidified atmosphere of 5% CO 2 at 37°C. For hormone induction experiments, cells were grown to confluence and serum-starved overnight in DMEM containing 1% (w/v) bovine serum albumin. Human GH (500 ng/ml), LIF (10 4 units/ml), or insulin (56 ng/ml) were added, and cell culture was maintained for different time intervals.
Animals and GH Administration-Female C57BL/6 mice (7 weeks old) were injected intraperitoneally with GH (10 g/10 g of body weight) and were subsequently sacrificed at different time points. Livers were removed, snap frozen in liquid nitrogen, and stored at Ϫ75°C.
RNA Extraction and Northern Blotting-Cell cultures were rinsed twice with ice-cold phosphate-buffered saline; total RNA was extracted from cell cultures and liver samples using guanidinium thiocyanateacid phenol (9). Total RNA was fractionated by electrophoresis through agarose-formaldehyde gels and transferred and subsequently fixed onto Hybond-N ϩ membranes as described (10). Membranes were hybridized at 63°C overnight with random primed, radiolabeled probes derived from full-length cDNA inserts encoding SOCS-1, SOCS-2, SOCS-3, or CIS (5). Membranes were then washed at high stringency (0.1 ϫ SSC, 0.1% SDS at 60°C) and exposed to Kodak BioMax MS film with an intensifying screen at Ϫ70°C. Where blots were reprobed, membranes were first stripped in boiling 0.1% SDS prior to prehybridization.
Cell Transfection and Reporter Gene Assays-Chinese hamster ovary (CHO) cells were maintained in Ham's F-12 medium supplemented with 10% fetal calf serum, 100 IU/ml penicillin, and 100 mg/ml streptomycin. The cells were transfected as described previously (11). Briefly, serum-starved cells were transfected using the calcium phosphate procedure with 1.5 g of rat GH receptor-encoding plasmid, 3 g of ␤-galactosidase internal control reporter plasmid, 1.5 g of the GHresponsive, serine protease inhibitor (Spi) 2.1 promoter-chloramphenicol acetyltransferase (CAT) construct, and 1 g of Flag epitope-tagged SOCS-1, SOCS-2, SOCS-3, or CIS expression vectors. After overnight culture in the absence or presence of GH (25 nM), cell extracts were prepared, and CAT and ␤-galactosidase activities were measured. Equivalent expression of the various SOCS and CIS poteins was confirmed by Western blotting portions of extracts from transfected cells with the Flag-specific monoclonal antibody, M2.

GH Preferentially Induces SOCS-3 mRNA Expression in
3T3-F442A Fibroblasts-In the absence of GH, Northern blotting of total RNA isolated from serum-starved 3T3-F442A cells detected low levels of SOCS-3 and SOCS-1 mRNA (Fig. 1). Transcripts corresponding to SOCS-2 and CIS were only observed upon prolonged exposure. Upon the addition of GH, induction of SOCS-3 expression was apparent at the first time point examined (15 min), peaking at 30 min before declining rapidly over subsequent time points (Fig. 1). This pattern of induction has been confirmed independently using both total and poly(A) ϩ mRNA (data not shown). The induction of SOCS-1, by comparison, was much less pronounced although with similar kinetics (Fig. 1). Both CIS and SOCS-2 mRNA were up-regulated in response to GH treatment (Fig. 1); with respect to CIS, induction peaked 60 min after hormonal exposure, and although the abundance of transcript had declined at time points assayed thereafter, expression remained significantly above basal levels. Similarly, while the induction of SOCS-2 mRNA exhibited delayed kinetics compared with other SOCS genes, elevated levels of the SOCS-2 message persisted to the final time point assayed (4 h).
Because 3T3-F442A fibroblasts express functional receptors for LIF and insulin (12), the ability of these ligands to influence SOCS gene expression was examined. LIF treatment strongly induced SOCS-3 mRNA expression, with the response appearing maximal after a 15-min stimulation with growth factor (Fig. 2). Interestingly, the subsequent down-regulation of SOCS-3 transcripts was significantly delayed by comparison with GH-treated cultures, with elevated levels of transcripts persisting at the final time point (4 h) assayed. Both SOCS-1 and CIS mRNAs were induced in response to LIF, albeit to a much lesser extent than SOCS-3, and returned to basal levels by 4 h (Fig. 2). No induction of SOCS-2 was observed (data not shown), while no member of the SOCS gene family was induced in cell cultures treated with insulin (data not shown).
SOCS-3 Is Induced by GH in Vivo-Strikingly, analysis of hepatic RNA isolated from the livers of mice injected with a single dose of GH confirmed the preferential activation of SOCS-3 mRNA expression by GH (Fig. 3). A marked elevation of SOCS-3 expression was observed around 60 min following GH injection, declining thereafter. Of the other SOCS family genes, clear induction of CIS expression was found 30 min after hormone injection, peaking at 60 min, and remaining elevated even after 24 h (Fig. 3). In contrast, SOCS-1 and SOCS-2 were only weakly induced by GH.

SOCS-1 and -3 Block Transactivation of a GH-responsive
Promoter Element-To explore the functional consequences of GH-induced expression of the different SOCS family genes, we determined the ability of GH to transactivate the GH-responsive Spi 2.1 promoter, linked to a CAT reporter gene, in CHO cells co-transfected with the various SOCS-encoding plasmids. In cells that were not transfected with SOCS expression constructs, GH induced a 3.4-fold increase in CAT activity from the Spi 2.1 promoter (Fig. 4). While constitutive expression of CIS had no effect on the hormonal responsiveness of the Spi 2.1 promoter, expression of either SOCS-1 or SOCS-3 ablated the ability of GH to transactivate reporter gene activity from the same promoter. Paradoxically, co-expression of SOCS-2 re-  2. LIF induction of SOCS-1, SOCS-3, and CIS mRNA expression in 3T3-F442A cells. Replicate blots, containing 15 g of total RNA per lane isolated from serum-starved 3T3-F442A cells prior to or following LIF treatment, were probed with the indicated cDNA inserts. The integrity of RNA samples and efficiency of transfer onto nylon membranes was monitored by ethidium bromide staining. The autoradiograph exposure time is 24 h.

FIG. 3. GH induction of hepatic expression of SOCS gene family mRNA in mice.
Replicate blots, containing 10 g of total RNA per lane isolated from the livers of C57BL/6 mice prior to or following GH treatment, were hybridized with the indicated probes. The integrity/ transfer of RNA was monitored as in Fig. 2. The autoradiograph exposure time is 24 h. sulted in the superinduction of CAT activity in response to GH, giving a 7-9-fold increase above basal levels of activity.

DISCUSSION
The identification of SOCS-3 as a major transcriptional target of GH action raises a number of questions, in particular how might SOCS-3 act to regulate signaling by the GH receptor. SOCS-1 can bind directly to Jak2 and Tyk2 (6,7) and is capable of inhibiting the activity of all four members of the JAK kinase family (6,7). As Jak2 is the principal effector of GH action at a cellular level, it would be anticipated that enforced expression of SOCS-1 in transfected CHO cells would block GH-mediated transactivation of the Spi 2.1 promoted CAT reporter construct, as observed in the present study (Fig. 4). To what extent SOCS-1 actually participates to down-regulate signaling from the ligandactivated GH receptor-Jak2 complex in vivo is not known, particularly given the very low level of SOCS-1 mRNA expressed in hepatic tissue following a GH pulse. In a contrasting mechanism of action, CIS, presumably via its SH2 domain, binds directly to the tyrosine-phosphorylated cytoplasmic domains of the EPO receptor and ␤-chain of the interleukin-3 receptor (4). While CIS has little effect on the activity of Jak2 (6), the principal effector molecule activated by EPO, it appears to act by preventing the docking of Stat5 (and other potential substrates?) onto the tyrosine-phosphorylated cytoplasmic domain of the EPO receptor (13). The absence of a discernible effect of constitutively expressed CIS on the ability of GH to transactivate the Spi 2.1 promoter is an interesting result. Three tyrosines in the cytoplasmic carboxyl terminus of the GH receptor (at positions 534, 566, and 627) can independently mediate GH-induced transcription from the Spi 2.1 promoter (11). Deletion/mutation of all three blocks activation of this promoter by GH. Recently, we have established that, in their phosphorylated state, each of these tyrosines can directly interact with Stat5 (14), thus representing a critical first step leading toward the transactivation of the Spi 2.1 promoter by GH. The inability of CIS overexpression to block transactivation would suggest that CIS does not bind to any of the three phosphotyrosines in the cytoplasmic tail of the GH receptor that serve as Stat5 binding sites, implying there may be specificity in cytokine receptor-CIS interaction. What mechanism is employed by SOCS-3 in down-regulating signaling from the ligand-activated GH receptor, or any other cytokine receptor, is not known at this point in time. Similarly, how co-expression of SOCS-2 facilitates the superinduction of reporter gene activity in response to GH has not been elucidated. It is conceivable that, in transfected cells, overexpression of SOCS-2 might somehow result in a dominant-negative phenotype by interfering with the capacity of GH to activate endogenous regulators (i.e. SOCS-3) of hormone-receptor signal transduction.
GH activates Jak2 in 3T3-F442A cells, while LIF activates both Jak1 and -2; both hormones induce the tyrosyl phosphorylation of Stats1, -3, and -5 in these cells (12). Clearly, one or more of these factors may be involved in the transcriptional induction of SOCS-3 and other gene family members following hormonal exposure. This is supported by the observation that expression of a dominant-negative Stat5 blocks IL-3 induction of CIS in Ba/F3 cells (13) while a dominant-negative Stat3 blocks IL-6/LIF induction of SOCS-1 mRNA in M1 cells (7). Interestingly, only Stat5 is activated in the livers of normal rats in response to a GH pulse (15). If STAT transcription factors assume a dominant role in the transcriptional activation of the SOCS gene family, then this last observation would imply that Stat5 is a critical element in the activation of SOCS-3 and CIS transcription by GH in vivo.
The preferential induction of SOCS-3 mRNA by GH, both in vivo and in vitro, contrasts with results obtained using IL-6 (5). All four SOCS gene transcripts are elevated in mouse liver following the injection of IL-6, although the kinetics of induction/persistence of individual mRNAs differs. However, only the induction of SOCS-1 and CIS mRNA is found in M1 cells treated with IL-6 (5). Whether the association between GH and SOCS-3 occurs in other tissues, or other GH-responsive cell lines, remains to be established.