Activated G Protein-coupled Receptor Induces Tyrosine Phosphorylation of STAT3 and Agonist-selective Serine Phosphorylation via Sustained Stimulation of Mitogen-activated Protein Kinase

The peptide hormone somatostatin exhibits antiproliferative activity by interacting with the G protein-coupled sst2 or sst5 receptor types. We show here that somatostatin at the human recombinant sst4 receptor induced a concentration-dependent increase in proliferation (EC50 20 nm) with a maximal response 5-fold greater than that produced by its synthetic analog, L-362,855. Analysis of the phosphorylation status of extracellular signal-regulated kinase (ERK)1 and ERK2 showed temporal differences in the changes evoked by the agonists. Phosphorylation induced by somatostatin (100 nm) peaked 10 min after the application and produced a response that continued for at least 4 h. In contrast, L-362,855 (1 μm) showed transient phosphorylation that had declined to basal levels by 1 h. However, both agonists induced rapid and sustained tyrosine phosphorylation of signal transducer and activator of transcription 3 (STAT3) which was pertussis toxin-insensitive. Serine phosphorylation of STAT3 was only apparent after somatostatin treatment and was abolished by pertussis toxin or PD 98059, together with the associated increases in proliferation. Mitogen-activated protein/ERK kinase-1 inhibition also decreased the time interval over which somatostatin-induced ERK phosphorylation was observed (<2 h). We conclude that the difference in the magnitude of the proliferative response evoked by the two agonists at the sst4 receptor can be accounted for by their differential ability to phosphorylate STAT3 on serine residues and supports the concept that selective signaling can be achieved through pharmacological diversity.

The peptide hormone somatostatin exhibits antiproliferative activity by interacting with the G protein-coupled sst 2 or sst 5 receptor types. We show here that somatostatin at the human recombinant sst 4 receptor induced a concentration-dependent increase in proliferation (EC 50 20 nM) with a maximal response 5-fold greater than that produced by its synthetic analog, L-362,855. Analysis of the phosphorylation status of extracellular signal-regulated kinase (ERK)1 and ERK2 showed temporal differences in the changes evoked by the agonists. Phosphorylation induced by somatostatin (100 nM) peaked 10 min after the application and produced a response that continued for at least 4 h. In contrast, L-362,855 (1 M) showed transient phosphorylation that had declined to basal levels by 1 h. However, both agonists induced rapid and sustained tyrosine phosphorylation of signal transducer and activator of transcription 3 (STAT3) which was pertussis toxin-insensitive. Serine phosphorylation of STAT3 was only apparent after somatostatin treatment and was abolished by pertussis toxin or PD 98059, together with the associated increases in proliferation. Mitogen-activated protein/ERK kinase-1 inhibition also decreased the time interval over which somatostatin-induced ERK phosphorylation was observed (<2 h). We conclude that the difference in the magnitude of the proliferative response evoked by the two agonists at the sst 4 receptor can be accounted for by their differential ability to phosphorylate STAT3 on serine residues and supports the concept that selective signaling can be achieved through pharmacological diversity.
Investigations of the growth effects of somatostatin on both normal and neoplastic human tissues suggest that it has a complex mechanism of action, inducing a direct antiproliferative response in a variety of cell types (1)(2)(3) in addition to reducing the circulatory levels of mitogenic hormones and growth factors (4,5). Somatostatin transduces its direct action by stimulating G protein-coupled receptors, named sst [1][2][3][4][5] (6), although little is known as to the identity of the receptor types mediating its antiproliferative functions in tissues, and information has been largely restricted to studies utilizing partially selective receptor analogs (7). Numerous reports have demonstrated the expression of a high density of somatostatin receptors on a variety of human cancer cells (4,8). The antiproliferative action of either somatostatin or its more metabolically stable analog octreotide, however, does not correlate with this expression, having inhibitory actions on pancreatic (9) and breast tumors (1) but eliciting no effect on the growth of small cell lung (10) and colon tumors (11). Growth-promoting effects of somatostatin have also been described in vitro on human pancreatic carcinoid (12) and epidermoid carcinoma cells (13), whereas in rat mesangial cells, somatostatin stimulates proliferation in the absence of serum but inhibits the growth of proliferating cells (14).
As part of a study to resolve some of the apparent conflicting actions of somatostatin on cell proliferation and to identify the receptor types involved, clonal lines have been established expressing the individual human recombinant forms in Chinese hamster ovary (CHO) 1 cells. Using a well characterized in vitro model, changes in cellular proliferation on the application of test substances have been assessed by determining regeneration into denuded areas of a previously confluent monolayer as well as by direct cell counting after a 24-h recovery period (15). Both the activated human recombinant sst 2 (16) and sst 5 receptors (7) in this model have been shown to have no effect on basal proliferation in the absence of exogenously added mitogenic agents, but they can inhibit that induced by a submaximal concentration of basic fibroblast growth factor (bFGF). This is in agreement with earlier studies demonstrating a reduced cell count after the application of the somatostatin analog RC-160 to complete media in cell lines expressing either the mouse sst 2 or the sst 5 receptor type (17). By contrast, preliminary studies employing the regeneration model have shown somatostatin to stimulate basal proliferation in cells expressing the human recombinant sst 4 receptor in the absence of other mitogenic agents (18).
The existence of different somatostatin receptors suggests that their functional responses may be mediated by a variety of effector systems. Although this aspect is not yet fully clarified, so far all five human receptors have been shown to be functionally coupled to inhibition of adenylate cyclase (19) and mediate, albeit to different degrees, the stimulation of phospholipase C with subsequent calcium mobilization (20). However, it is difficult to explain the specificity and diversity of the responses to somatostatin through the apparently similar modulation and subsequent actions of a small number of second messengers. The mechanism involved in transmission of the antiproliferative effect of the sst 2 receptor has been attributed to the activation of a protein tyrosine phosphatase, SHP-1 (21), and it has been suggested that such an activity may counteract the growth-promoting properties of receptors containing an intrinsic tyrosine kinase domain (22).
Activation of the mitogen-activated protein (MAP) kinase cascade has been demonstrated after stimulation of sst 4 receptors (23), which is an important transduction mechanism utilized by many growth factor receptors, although increasing evidence suggests that the outcome of the receptor's signal depends both on the duration of extracellular signal-regulated kinase (ERK) activation and cell context (24,25). A critical aspect of cell phenotype will be which ERK-responsive transcription factors are present. A number of transcription factors have been identified as targets for MAP kinase, which enters the nucleus after phosphorylation by MAP/ERK kinase (MEK) (26). The Ras/ERK pathway provides a common route by which signals from different growth-inducing receptors converge at a major regulatory element of the promoters of the c-fos and other coregulated genes, the serum response element. The major targets for the pathway in the c-fos promoter are the ternary complex factor family members, Elk1 and SAP1, whose activity is regulated by phosphorylation of a cluster of Ser/Thr-Pro motifs (Ser 383 and Ser 389 of Elk1) at their COOH termini (27).
Another potential substrate for MAP kinase, belonging to a very different family of transcription factors, is signal transducer and activator of transcription 3 (STAT3). The STAT proteins are involved in the transcriptional attenuation of many cytokine-and growth factor-inducible genes, and, prior to receptor activation, they appear to be cytoplasmic (28). They at least associate transiently with their cognate receptors during activation; after phosphorylation at a conserved tyrosine, which allows dimerization through reciprocal phosphotyrosine-SH2 domain interactions, they translocate to the nucleus and bind the cis-inducible element found in the promoter of the c-fos gene. STAT activation by receptors lacking intrinsic kinase activity, including G protein-coupled receptors (29), involves specific members of the Janus kinase (JAK) family (30). However, STAT3 is also phosphorylated by a serine/threonine kinase thought to be MAP kinase which enhances its transcriptional activity (31).
The purpose of this study was to investigate the transduction mechanism mediating the proliferative action of somatostatin through stimulation of the human sst 4 receptor and to compare the effects with those induced by the somatostatin analog L-362,855, which displays full agonist activity at this receptor type (32). In particular, the kinetics of the phosphorylation of ERK1 and ERK2 were monitored throughout the initial period of proliferation and compared with responses induced by bFGF. In addition, correlation between any late phase MAP kinase activity and induced phosphorylation of Elk1 or STAT3 was also determined.

EXPERIMENTAL PROCEDURES
Materials-Chinese hamster ovary (CHO-K1) cells were obtained from The European Collection of Animal Cell Cultures (ECACC). Lipo-fectAMINE, Geneticin (G418 sulfate, specific activity 500 g/ml), and reagents for culturing cells were obtained from Life Technologies. Thermanox coverslips, manufactured by Nunc, were purchased through Life Technologies, otherwise Costar tissue culture plastic ware was used. bFGF and monoclonal antibodies to ␤-actin were obtained from Sigma. Somatostatin and L-362,855 were synthesized by the GlaxoWellcome Chemistry Unit, Department of Chemistry, University of Cambridge, U. K. PD 98059 (2Ј-amino-3Ј-methoxyflavone) and Bordetella pertussis toxin were from Calbiochem. Antibodies to ERK1 (C-16) and ERK2 (C-14) were obtained from Santa Cruz Biotechnology; that specific to the phosphoforms (at both Thr 202 and Tyr 204 ) was supplied by New England Biolabs. Phosphospecific STAT3 antibodies recognizing either the serine (Ser 727 ) or the tyrosine phosphorylation sites (Tyr 705 ) and that to the phosphorylated form of Elk1 (Ser 383 ) were also obtained from New England Biolabs. Antibodies to these transcription factors but whose specificity is independent of the phosphorylation state were supplied by Santa Cruz Biotechnology.
Stable Expression of Human Somatostatin sst 4 Receptors in CHO Cells-The cDNA encoding the human sst 4 receptor was subcloned into the mammalian expression vector pCIN4 harboring a neomycin-resistant gene as a selection marker. CHO-K1 cells were cultured in Dulbecco's modified Eagle's medium/Ham's F-12 medium (1:1) containing 10% fetal calf serum and 1 mM Glutamax I and transfected in the absence of serum with 10 g of sst 4 -pCIN4 using a cationic liposome formulation-mediated transfer (LipofectAMINE). Selection was performed in the presence of complete medium containing 1 mg/ml G418 sulfate, and clonal cell lines expressing the cDNA were isolated by single cell cloning. Receptor expression was assessed by binding of 125 I-Tyr 11 -somatostatin. The transfected cell line expressing the human recombinant sst 4 receptor (CHOsst 4 ) used in the current study gave IC 50 values for somatostatin binding of 0.33 Ϯ 0.06 nM with the corresponding B max value being 2.07 Ϯ 0.43 pmol/mg of membrane protein (n ϭ 3). All cultures were routinely maintained in their appropriate growth medium at 37°C in humidified air containing 5% carbon dioxide and passaged when 95% confluence was reached.
Partial Denudation of Confluent Cell Monolayers and Assessment of Proliferation-Cells were grown to confluence in complete media on Thermanox coverslips. Proliferation was assessed by determining the regeneration of multiple denuded areas (400 m wide), created by dragging a Perspex comb across the surface of the coverslip, according to the method described previously (7). The Perspex comb was designed so that 50% of the confluent monolayer was removed by the partial denudation process, leaving parallel strips of cells. Importantly, for CHO-K1 cells, it has been demonstrated previously that repopulation of the denuded areas in the presence of bFGF for 24 h is associated with a concomitant increase in cell number, abolished by protein synthesis inhibitors and independent of motorgenic processes (7,15).
After partial denudation, coverslips were washed twice in phosphatebuffered saline and placed in a fresh well containing drug or vehicle in appropriate media but in the absence of fetal calf serum. Regeneration of the cell monolayer was terminated after 24 h by rapid washing in ice-cold phosphate-buffered saline followed by fixing in absolute ethanol. Regeneration of the denuded areas was quantified using a Leica Q500 MC image analysis system. Image analysis has revealed that regeneration occurs by an outgrowth of CHO-K1 cells along the perimeter of the denuded areas with an insignificant number of cells becoming detached from the coverslip at any time point during the investigative period (15). For each coverslip, four fields of view selected at random were analyzed, and data are expressed as a percentage of the denuded area recovered. All results were calculated from a minimum of three experiments with four replicates per test group; values are expressed as the arithmetic mean Ϯ S.E. Statistical analysis was by analysis of variance followed by Tukey's test, taking p Ͻ 0.05 as the level of significance (SigmaStat ® version 2.0). The EC 50 values were calculated by interpolation from the concentration-effect curves plotted and are defined as the concentration of drug which produced half its maximal possible response.
Determination of Change in the Phosphorylation Status of ERK1, ERK2, STAT3, and Elk1-To analyze changes in the phosphorylation status of ERK1 and ERK2 and the transcription factors STAT3 and Elk1 at various stages during the regenerative processes after partial denudation, whole cell protein extract was combined from eight coverslips for each treatment group. Immediately before partial denudation, cells forming the confluent monolayers of CHOsst 4 cells were in either the G 0 or early G 1 phase of the cell cycle. Producing multiple denuded areas on a single coverslip dramatically increases the number of cells recruited into the regenerative process and amplifies the resultant biochemical signals. For endothelial cells, it has been shown that after denudation, at least 10 cells adjacent to the denuded edge respond to this process. Subsequent analysis of changes induced in the phosphorylation state of effector proteins involved in basal or growth factorstimulated regeneration for CHO-K1 cells will thus reflect those of a large synchronized, proliferating cell population as well as from a small, contact-inhibited subpopulation localized in the central regions of the confluent strips. Application of bFGF to nondenuded, confluent CHO-K1 monolayers has very little effect on the phosphorylation status of ERK1 or ERK2, a weak transient increase in phosphorylation between 10 and 20 min after application. 2 Regeneration was terminated by washing the CHOsst 4 cell monolayers in ice-cold phosphate-buffered saline before applying SDS-polyacrylamide gel sample buffer (50 l of 3 ϫ strength) to each test well (1 ϫ sample buffer: 4% SDS, 5% glycerol, 60 mM Tris, pH 6.8, and 0.01% bromphenol blue) under reducing conditions (50 mM 2-mercaptoethanol). After solubilization of cellular protein by rapid mixing, the well contents were transferred to a separate tube and combined with two further washings of the well with deionized water (50 l). Samples were vortexed, centrifuged at 10,000 ϫ g for 2 min, and heated at 95°C for 5 min. Total cell protein for each of the extracts was measured by mi-croBCA (Pierce), and equivalent amounts of protein were resolved electrophoretically on 10% polyacrylamide gels.
After electrophoretic transfer onto nitrocellulose (0.22 m) using a semidry blotter, the membrane was washed briefly in Tris-buffered saline (TBS: 50 mM Tris, pH 7.5, 250 mM NaCl) and saturated overnight in TBS supplemented with 0.1% Tween 20 and 5% dried milk. For detection with the antibodies to ERK1 and ERK2, the membranes were incubated with a 1:2,000 dilution (1:1 mix of ERK1 and 2) or 1:1,000 dilution of the anti-phospho-ERK antibody. Antibodies to the phosphospecific forms of the transcription factors were used at a 1:500 dilution, whereas those to STAT3 and Elk1 were at a 1:1,000 dilution. When used, the antibody to ␤-actin was at a 1:5,000 dilution. All primary incubations were for 1 h at 22°C in TBS containing 0.1% Tween 20 (TBST) followed by washing five times for 10 min each in TBST. Membranes were incubated for 1 h at 22°C with a 1:5,000 dilution of the appropriate horseradish peroxidase-conjugated secondary antibody in TBST containing 5% dried milk. Excess antibody was removed by washing as above, and immunocomplexes were visualized using enhanced chemiluminescence (ECL) detection, according to the manufacturer's instructions (Amersham Pharmacia Biotech). The Western blots shown are representative of at least three separate experiments, and each panel is taken from a single immunoblot. 4 Receptors-Using the in vitro model to determine changes in proliferation of CHOsst 4 cells, the basal rate of regeneration in incomplete media, 24 h after partial denudation of a previously confluent monolayer, was 7.1 Ϯ 0.4%. Application of somatostatin caused a concentration-dependent increase in the proliferation of these cells ( Fig. 1) with its potency, expressed as an EC 50 value, estimated at 19.9 Ϯ 0.3 nM. The synthetic peptide, L-362,855, showed some agonist activity, producing concentration-dependent increases in the regeneration of CHOsst 4 cells with an EC 50 of 63.1 Ϯ 0.8 nM, but the maximal response (10.0 Ϯ 0.1%) was greatly suppressed compared with that induced by somatostatin (24.1 Ϯ 0.2%) and was only just significantly different from basal values (p ϭ 0.045) (Fig. 1). The concentration-effect curve for bFGF reached a maximal response similar to that produced by somatostatin ( Fig. 1) but was at least 300-fold more potent (EC 50 0.05 Ϯ 0.004 nM). In all further experiments, submaximal concentrations of bFGF and somatostatin were used at 10 ng/ml (0.6 nM) and 100 nM, respectively, which gave approximately equivalent amounts of regeneration (80% of the maxima). Two concentrations of L-362,855 were used: L-362,855 1 at 100 nM and L-362,855 2 at 1 M.

Effect of Somatostatin and L-362,855 on the Proliferation of Chinese Hamster Ovary Cells Recombinantly Expressing Human sst
Effect of Pertussis Toxin Pretreatment or MEK Inhibition on Proliferation-Pretreatment for 20 h with pertussis toxin (100 ng/ml) had no significant effect on either basal proliferation of CHOsst 4 cells (6.9 Ϯ 0.2%) or that induced by bFGF (10 ng/ml) treatment (22.5 Ϯ 0.2% and 22.7 Ϯ 0.4% with or without pertussis toxin, respectively). However, the increased proliferation evoked by application of somatostatin (100 nM) was abolished after pretreatment with the toxin to values not significantly different from basal (7.6 Ϯ 0.5% and 21.6 Ϯ 0.5% with or without toxin treatment, respectively). The increased proliferation of CHOsst 4 cells in the presence of L-362,855 2 (1 M) also decreased after pertussis toxin pretreatment, again to values not significantly different from basal (6.8 Ϯ 0.6% and 9.3 Ϯ 0.2% with and without pertussis toxin, respectively).
Changes in the Phosphorylation Status of ERK1 and ERK2-Activation of MAP kinase during the initial processes involved in the proliferative response of CHOsst 4 cells was assessed by monitoring changes in the phosphorylation status of ERK1 and ERK2. A time course of the immunoreactivity detected with the anti-phospho-ERK1 and ERK2 antibody by Western analysis of whole cell extracts of CHOsst 4 cells over the initial 4 h of regeneration is shown in Fig. 2. During this period, there was no detectable change in the expression of ERK1 or ERK2 protein in CHOsst 4 cells ( Fig. 2A), and this level was not affected by any drug treatment (data not shown).
Proliferation under basal conditions produced little apparent change in the phosphorylation of ERK1 and ERK2 during the initial 10 min compared with levels detected in confluent monolayers prior to denudation (Fig. 2B). However, this apparent "steady state" of phosphorylation decreased over the next 20 min to almost undetectable levels by 30 min (Fig. 2B). Proliferation in the presence of somatostatin (100 nM), L-362,855 1 (100 nM), or bFGF (10 ng/ml) produced a marked increase in the phosphorylated forms of both ERK1 and ERK2 compared with predenudation levels (Fig. 2B). The increase in the phosphorylation induced by L-362,855 1 reached a peak after recovery for 20 min but diminished to undetectable levels by 1 h. In contrast, somatostatin treatment evoked a maximal response at 10 min, and although this level subsequently declined, that observed at 4 h postdenudation was substantially increased over basal (Fig. 2B). The phosphorylation of ERK1 and ERK2 in the presence of bFGF appeared biphasic in that a transient increase occurred at 10 min with a second peak following at 4 h postdenudation, although a persistent increase in the immunoreactivity over basal levels was observed throughout the entire time course (Fig. 2, B and C).
The intensity of the immunoreactivity detected with the anti-phospho-ERK1 and ERK2 antibody was strongest after somatostatin (100 nM) treatment for 10 min or bFGF (10 ng/ml) treatment for 4 h (Fig. 2C). The level of phosphorylation after regeneration in somatostatin (100 nM) or a high concentration of L-362,855 2 (1 M) for 20 min was comparable (Fig. 2C). The level of phosphorylation observed after a 2-or 4-h incubation period with L-362,855 2 was considerably lower for both ERK1 and ERK2 compared with that induced by somatostatin and comparable to that detected for samples allowed to recover for these same times under basal conditions (Fig. 2C).
Changes Induced in the Serine Phosphorylation Levels of STAT3 and Elk1-Activation of the transcription factors STAT3 and Elk1 is associated with phosphorylation on serine residues Ser 727 (33) and Ser 383 (27), respectively. A time course of the immunoreactivity detected with anti-STAT3 or anti-Elk1 antibodies showed that the expression of these transcription factors remained unchanged during the early basal regenerative processes of partially denuded CHOsst 4 cell monolayers (Fig. 3A). The expression of STAT3 and Elk1 was also unaffected by incubation with any of the drug treatments during this initial 4-h period (data not shown).
Under basal conditions, regeneration was associated with a small increase in the serine phosphorylation of both STAT3 and Elk1 as demonstrated by the enhanced immunoreactivity compared with samples prepared immediately after denudation, together with a concomitant electrophoretic mobility shift of the proteins (Fig. 3B). Recovery in the presence of somatostatin (100 nM) increased the phosphorylation on Ser 727 of STAT3, producing two peaks: one at 20 min and a second at 4 h (Fig.  3B). Somatostatin treatment, however, appeared to have no effect on the basal level of phosphorylation of Elk1. Recovery in the presence of L-362,855 1 (100 nM) produced little change in the intensity of the immunoreactivity detected by the antiphosphospecific antibodies to either transcription factor, and the shift in mobility was comparable to that observed under basal conditions over the same time interval (Fig. 3B). In contrast, bFGF (10 ng/ml) induced a steady increase in the phosphorylation of both STAT3 and Elk1 throughout the first 4 h of proliferative processes, producing maximal immunoreactivity and electrophoretic mobility shifts at the longest recovery time examined (Fig. 3B).
Increasing the concentration of L-362,855 2 10-fold to 1 M produced no further change in the level of immunoreactivity detected with the anti-STAT3 phosphospecific antibody compared with that obtained after recovery for 20 min or 2 h with the lower dose (Fig. 4, A and B). After the longer recovery time, the antibody appeared to detect a doublet for all treatment groups, with a similar intensity of staining for the band with the retarded mobility (Fig. 4B). The phosphorylation induced by somatostatin (100 nM) or bFGF (10 ng/ml) of the product with the greater mobility was similar at 2 h postdenudation and considerably increased over basal levels (Fig. 4B). However, the intensity in staining of this latter product appeared less than that of the control after recovery with either concentration of L-362,855 (Fig. 4B). The increased phosphorylation at Ser 727 of STAT3 induced by somatostatin (100 nM) or bFGF (10 ng/ml) after recovery for 4 h was abolished by coapplication of PD 98059 (2 M) (Fig. 4C). PD 98059 had no observable effect on the basal level of serine phosphorylation of STAT3 (data not shown).
Changes Induced in the Tyrosine Phosphorylation of STAT3-Tyrosine phosphorylation on residue Tyr 705 of STAT3 is required for dimerization of the transcription factor and subsequent binding to DNA (34). During basal recovery over the initial period of regeneration of CHOsst 4 cells there was no change in the expression of STAT3 (Fig. 5A), but a small and transient increase in the immunoreactivity detected by the anti-phosphospecific antibody was observed, producing a peak 5 min after denudation and falling to levels below those detected in confluent monolayers by 30 min (Fig. 5B). Associated with this increased phosphorylation was a concomitant splitting of the immunoreactivity into discrete bands.
Regeneration in the presence of somatostatin (100 nM) resulted in a marked and rapid increase in the tyrosine phosphorylation of STAT3, which was apparent after 5 min of regener- ative processes and appeared to be sustained throughout the first 4 h (Fig. 5B). After approximately 20 min of regeneration, multiple products with discrete electrophoretic mobilities were detected by the anti-phosphospecific antibody, and the intensity of the staining shifted to those with retarded mobilities. L-362,855 1 (100 nM) also induced rapid tyrosine phosphorylation of STAT3 which was sustained throughout the initial 20 min of recovery, but in contrast to somatostatin-treated samples, the immunoreactivity subsequently diminished to predenudation levels by 1 h (Fig. 5B). The appearance of multiple products with distinct electrophoretic mobilities, as observed for somatostatin-treated samples, was also less evident during the longer recovery period in the presence of L-362,855 1 (100 nM). Treatment with bFGF (10 ng/ml) produced a pattern of staining with the phosphospecific antibody which was comparable to that obtained with somatostatin, whereby sustained tyrosine phosphorylation of STAT3 was apparent with multiple products being detected after regeneration for 30 min (Fig. 5B).
Increasing the concentration of L-362,855 2 to 1 M produced a pattern and intensity of immunoreactivity similar to those observed after somatostatin (100 nM) or bFGF (10 ng/ml) treatment for 5-and 20-min regeneration (Fig. 5C). At 30 min, the split into discrete products with a concomitant shift in the immunoreactivity to those with retarded mobilities was more apparent after bFGF or somatostatin treatment than for L-362,855 2 . However, the 10-fold higher concentration of L-362,855 2 resulted in an intensity of staining comparable to that produced by somatostatin (100 nM) after 30-min and 4-h regeneration (Fig. 5C). The separation of the immunoreactivity detected with the anti-phosphotyrosine-specific STAT3 antibody into multiple products is most probably the result of electrophoretic mobility shifts caused by the increased phosphorylation at discrete sites within the protein. Although STAT3 is phosphorylated on a single tyrosine residue (as detected by the antibody in Fig. 5, B and C) there are alternative sites for the addition of phosphate at serine/threonine residues. The separation into multiple bands observed particularly after regeneration for 30 min and longer in the presence of somatostatin or bFGF as shown in Fig. 5B is consistent with the concomitant increase in the phosphorylation of the protein at Ser 727 as shown in Fig. 3B.
Pretreatment of CHOsst 4 cells for 20 h with pertussis toxin (100 ng/ml) had no apparent effect on the intensity of immunoreactivity or banding pattern after detection with the antiphosphotyrosine-specific STAT3 antibody, after recovery of partially denuded monolayers in the presence of somatostatin (100 nM), L-362,855 2 (1 M), or bFGF (10 ng/ml) for 10 min (Fig.  6).
Effect of Inhibiting MEK1 Activity on the Phosphorylation of ERK1 and ERK2 Induced by Somatostatin in Regenerating CHOsst 4 Cells-Coapplication of the MEK1 inhibitor PD 98059 (2 M) (35) with somatostatin (100 nM) to CHOsst 4 cells immediately after denudation produced a pattern and intensity of immunoreactivity with the anti-phosphospecific ERK1 and ERK2 antibody similar to those produced by somatostatin treatment alone over the initial 30 min of regenerative processes (Fig. 7). However, after 2 and 4 h of regeneration, the level of ERK phosphorylation in samples treated with PD 98059 and somatostatin was almost undetectable, similar to that observed after incubation with L-362,855 1 (100 nM) for 4 h and markedly reduced compared with samples treated with somatostatin alone (Fig. 7). PD 98059 had no detectable effect on basal levels of phosphorylation of ERK1 and ERK2 throughout the time course examined, and the immunoreactivity detected after regeneration with and without PD 98059 present for 10 min is shown in Fig. 7.

DISCUSSION
Despite numerous reports describing the ability of somatostatin to regulate the growth of a variety of cell lines (1,7,12), little information is available as to the identity of the receptor types involved. An antiproliferative activity has been demonstrated in recombinant systems for the sst 2 and sst 5 receptors (16,17), and stimulation of cell growth in the absence of exogenously added mitogenic factors has been shown after sst 4 receptor activation (18). Thus, the biphasic and sometimes contrasting nature of the growth responses observed on application of somatostatin to cells expressing multiple somatostatin receptors may be caused by the resultant effect of the activation of those types with opposing proliferative functions. The aim of the current investigation was to dissect the transduction mechanism through which the human sst 4 receptor mediates its proliferative action so that a greater understanding of the functional interplay between the individual somatostatin receptors can be made.
Alterations in the proliferative rate of transfected CHO-K1 cells after 24 h in the presence of test agents were assessed by determining the regeneration of denuded areas of a previously confluent monolayer (15). At various times throughout the initial stages of recovery, analysis was made of any change induced in the phosphorylation status of ERK1 and ERK2, in an attempt to correlate persistent or transient activation of MAP kinase with the resultant proliferative effects observed 24 h later. Somatostatin, its analog L-362,855, and bFGF, all produced a rapid and marked increase in the phosphorylation state of ERK1 and ERK2, which reached a peak between 10 and 20 min after application. However, this rapid and potent phosphorylation did not appear to be responsible for the induced proliferative effect of somatostatin in CHOsst 4 cells. Although somatostatin and bFGF both increased cell growth, L-362,855 had little effect; and perhaps more germane, coapplication of PD 98059 with somatostatin abolished the somatostatin-induced proliferation but had no effect on the phosphorylation changes evoked in ERK1 and ERK2 during the initial hour of regeneration. The persistent increase in phosphorylation observed after regeneration in the presence of somatosta-FIG. 6. Effect of pretreatment for 20 h with pertussis toxin (100 ng/ml) on the phosphorylation change induced at Tyr 705 of STAT3 during initial events in the induced proliferation of CHOsst 4 cells as determined by Western analysis. Partially denuded monolayers were allowed to recover for 10 min in the presence of incomplete media (CON), somatostatin (SRIF; 100 nM), L-362,855 1 (L36 1 ; 100 nM), or bFGF (10 ng/ml) which had been pretreated with (PTX) or without pertussis toxin.

FIG. 7. Effect of MEK1 inhibition on the phosphorylation change induced in MAP kinase by somatostatin during initial proliferative processes of CHOsst 4 cells as determined by Western analysis.
Whole cell extracts were prepared from confluent monolayers immediately after denudation (z) and after recovery in incomplete media (CON) or incomplete media containing PD 98059 (PD; 2 M) for 10 min as well as somatostatin (SRIF; 100 nM) or somatostatin in the presence of PD 98059 for the times indicated (in minutes). For comparison, the level of immunoreactivity detected after recovery in the presence of incomplete media or L-362,855 1 (L36 1 ; 100 nM) for 4 h is also shown.
tin, however, was abolished by PD 98059 treatment. It would thus seem that in CHO-K1 cells recombinantly expressing the sst 4 receptor, it is the prolonged activation of MAP kinase which is necessary for somatostatin to promote cell growth. The functional role of the transient activity of MAP kinase induced by activated sst 4 and bFGF receptors in CHO-K1 cells remains to be determined.
Recent reports using several cell types have suggested a role for chronic MAP kinase activation in causing exit from the cell cycle followed by reduced DNA synthesis and cellular differentiation (36,37). These studies in particular have focused on the role of nerve growth factor-induced growth arrest and inhibition of cell cycle-dependent kinases. In both PC12 cells and primary cultures of hepatocytes, acute phasic activation (ϳ30 min) of the MAP kinase cascade was found to stimulate DNA synthesis, whereas chronic activation (ϳ5 h) in both cell types inhibited G 1 progression/S phase entry. These data would appear to be in conflict with the effects on proliferation observed in this study and the associated duration of MAP kinase activity. However, several publications have documented that both oncogenic Ras and Raf can stimulate proliferation and suppress differentiation in SV40 virus immortalized hepatocytes (38) in which potent and prolonged activation of MAP kinase has been demonstrated. In addition, it has also been shown that in fibroblasts and other cell types, sustained activation of MAP kinase is associated with proliferation and not differentiation (36); and depending on where it is expressed, the same receptor can induce either response (39). Such observations thus indicate the critical importance of cell phenotype in understanding signaling events and predicting the functional outcome of a sustained or transient activation of a particular transduction pathway.
One aspect that is fundamental to the resultant cellular effect induced by the activation of the MAP kinase cascade will be which transcription factors are present and if they are restricted in localization to the nucleus. In every case examined thus far, sustained ERK activation is associated with translocation of the kinases to the nucleus (39,40), whereas transient activation does not induce cytoplasmic nuclear migration. Transient activation will therefore have very different consequences for gene expression compared with that of sustained MAP kinase activity. Nuclear accumulation of active ERK has been shown to phosphorylate Elk1 in response to growth factors whose receptors contain an intrinsic tyrosine kinase domain (27). Although MAP kinase activation induced by bFGF in this study appeared biphasic, prolonged phosphorylation of ERK1 and ERK2 was evident. Concomitant with the marked increase in phosphorylation of MAP kinase after recovery in the presence of the growth factor for 4 h was the enhanced serine phosphorylation of both Elk1 and STAT3. In contrast, somatostatin had no apparent effect on the phosphorylation at Ser 383 of the nuclear located Elk1 but increased that of STAT3 at Ser 727 . The differential phosphorylation of Elk1 induced by somatostatin and bFGF, despite both drugs producing a persistent activation of MAP kinase, suggests that additional regulation of ERK-responsive transcription factors must be in place other than those enforced through compartmentalization restrictions. Even though Elk1 has been shown to be a good substrate for ERK1 and ERK2 in vitro, there appears to be some selectivity for its requirement in the transduction of mitogenic responses induced through the different receptor types used in this study. The apparent lack of serine phosphorylation of either transcription factor on treatment with L-362,855 suggests that sustained activation of MAP kinase is critical in CHO-K1 cells for the serine phosphorylation of STAT3 although it has not yet been determined if this phos-phorylation event takes place within the nucleus.
STAT proteins are present in a latent form in the cytoplasm and become phosphorylated on a single tyrosine, which is obligatory for STAT activation (28,34). Reports have shown this phosphorylation to occur within minutes after ligand binding (28); this is consistent with the rapid change in the immunoreactivity observed in this study using the anti-phosphotyrosine-specific antibody to STAT3 after bFGF or sst 4 receptor activation. Rapid tyrosine phosphorylation of STAT3 was observed in the regeneration model under basal conditions, although the increase was only transient in nature. A small but detectable increase in serine phosphorylation with associated electrophoretic mobility shifts was also observed for both STAT3 and Elk1 under basal conditions, suggesting that these transcription factors may be involved in the basal regenerative processes of areas denuded in confluent monolayers of CHO-K1 cells. It may be that mitogenic factors are released from damaged cells at the denuded edge, and/or signal transduction mechanisms are triggered by the disruption of integrin clusters or intracellular zonular adheren sites. Regulation of the cytoskeletal architecture is a requirement not only for cell motility but also for growth, and it is possible that effectors involved in the regulation of the actin cytoskeleton, such as Src, FAK, and RhoGAP (41), can activate early response gene transcription in CHO-K1 cells to promote cell proliferation under basal conditions.
An enhanced increase in the tyrosine phosphorylation of STAT3 over basal levels was detected at all time points examined on application of somatostatin, the higher concentration of L-362,855 as well as bFGF. The rapidity of the induced phosphorylation suggests that this event is placed upstream of the activation of MAP kinase and the serine phosphorylation of STAT3. The lack of effect of pertussis toxin treatment on the tyrosine phosphorylation also suggests that for the sst 4 receptor, activation of this pathway is independent of the G i /G omediated stimulation of MAP kinase. However, because the somatostatin-induced proliferation after activation of sst 4 receptors can be inhibited by pertussis toxin or PD 98059, the mechanism involved is presumably critically dependent on serine phosphorylation of STAT3 because this was also abolished by PD 98059 and pertussis toxin. The mechanism causing the small proliferative response induced by L-362,855 is unlikely to be a consequence of the activation of a receptor other than the sst 4 because L-362,855 is devoid of agonistic activity in wildtype CHO-K1 cells (32). 2 It is possible that L-362,855 does cause a small degree of serine phosphorylation of STAT3 but below the level of detection by Western analysis using the anti-phosphospecific antibody. Such a finding would be consistent with both the partial agonist effect observed with L-362,855 in the proliferation assay and the sensitivity of this effect to pertussis toxin and PD 98059.
Tyrosine phosphorylation of STAT proteins has been demonstrated after ligand activation of receptors with intrinsic tyrosine kinase activity as well as the cytokine receptors that lack catalytic activity but to which JAKs are noncovalently associated (42). Ligand-mediated dimerization of either type of receptor results in reciprocal tyrosine phosphorylation and consequent activation of the intrinsic or attached kinase (43). The mechanism by which G protein-coupled receptors activate the JAK/STAT pathway is less clear. In the case of the angiotensin II receptor AT 1 , binding of JAK2 to the receptor itself has been demonstrated (44). It has not been determined here whether JAK activity was required for the tyrosine phosphorylation of STAT3 either by activated bFGF or sst 4 receptor types, although STAT3 can be a substrate for other non-receptor tyrosine kinases such as Src (45). However, what is clear is that tyrosine phosphorylation is necessary for dimerization of the STAT3 proteins and that serine phosphorylation can modulate binding to DNA and transcriptional activity. The kinase responsible for the serine phosphorylation of STATs is not known, but the phosphorylated residue identified in the COOH terminus of STAT3 lies within a consensus site for MAP kinase and is phosphorylated by this kinase in vitro (31). The abolition of the serine phosphorylation induced by somatostatin in STAT3 on application of the MEK1 inhibitor is consistent with the concept that STAT3 is an ERK-responsive transcription factor. Thus, although both phosphorylation events are necessary for full transcriptional activity of STAT3, they are controlled by distinct signaling pathways.
An unexpected and intriguing finding of this study came from the comparison of the functional and transductional responses induced by activation of the sst 4 receptor type using different agonists. Because somatostatin and L-362,855 both caused a transient phosphorylation of ERK1 and ERK2 as well as a rapid and sustained tyrosine phosphorylation of STAT3, the lack of effect observed with L-362,855 on serine phosphorylation of STAT3 together with its poor proliferative action demonstrates agonist-selective receptor transduction. In view of the partial agonist effects observed with L-362,855, these findings suggest that the receptor's mechanism mediating tyrosine phosphorylation of STAT3 is highly efficiently coupled in contrast to the more poorly coupled activation of MAP kinase which results in serine phosphorylation of the transcription factor. The phosphorylation changes in STAT3 are the result of somatostatin stimulating distinct pathways with bifurcation at the level of its receptor as suggested by the differential sensitivity of the phosphorylation events to pertussis toxin.
There is an increasing number of examples in the literature where the functional outcome in response to a mitogenic agent is determined not only by the strength but also the duration of the stimulus, and small differences in signal input can generate large differences in transcriptional response. Even if two ligands activate the same STAT protein, they may do so quantitatively differently for different periods of time, thus contributing to different transcriptional outcomes. Such quantitative variations can control physiological decisions. In this study we have shown that CHO-K1 cells expressing the recombinant sst 4 receptor can be stimulated to proliferate by the application of bFGF or somatostatin. This response is mediated in the case of the sst 4 receptor through a prolonged activation of ERK1 and ERK2 which is dependent on the activity of G i /G o and is a common type of transduction mechanism for many mitogenic G protein-coupled receptors (46). Subsequent phosphorylation of Ser 727 of STAT3 appears dependent on this prolonged activation of MAP kinase and can be activated differentially by agonist-selective stimulation of the recombinant sst 4 receptor. Agonist activation can, however, promote tyrosine phosphorylation of the same transcription factor possibly through a G protein-independent mechanism, in a functionally cooperative but distinct signaling pathway.