Coordinate Regulation of STAT Signaling and c-fosExpression by the Tyrosine Phosphatase SHP-2*

The src homology 2 (SH2) domain-containing protein-tyrosine phosphatase SHP-2 has been implicated as an important positive regulator of several mitogenic signaling pathways. SHP-2 has more recently been shown to be tyrosine phosphorylated and recruited to the gp130 component of the ciliary neurotrophic factor (CNTF) receptor complex upon stimulation with CNTF. CNTF does not, however, have a proliferative effect on responsive cells, but rather enhances the survival and differentiation of sympathetic, motor, and sensory neurons. In this study, expression of an interfering mutant of SHP-2 in the neuroblastoma cell line NBFL increased CNTF induction of a vasoactive intestinal peptide (VIP) reporter gene, and in cultures of sympathetic neurons, it resulted in an up-regulation of endogenous VIP and substance P (SP) gene expression. Members of the CNTF family of cytokines transmit their signal by activating signaling pathways involving both STAT and Fos-Jun transcription factors. In CNTF-stimulated NBFL cells that constitutively express the SHP-2 interfering mutant, there was increased and prolonged formation of STAT/DNA complexes, but decreased AP-1 binding activity, that mirrored a down-regulation of c-fos expression both at the mRNA and protein level. Taken together, these data indicate that SHP-2 has dual and opposing roles in a signaling cascade triggered by the same ligand, as illustrated by its ability to differentially regulate the levels of activity of both STAT and AP-1 transcription factors.

vertebrate nervous system (1,2). In cultured sympathetic neurons derived from superior cervical ganglia (SCG) of newborn rats, CNTF induces a switch from a noradrenergic to a cholinergic phenotype (3). In addition, CNTF increases expression of the genes encoding several peptide neurotransmitters, such as vasoactive intestinal polypeptide (VIP), somatostatin (SOM), and substance P (SP) (4,5). In vivo, a similar noradrenergic to cholinergic switch and increased expression of VIP occur in a subset of sympathetic neurons when they contact the target sweat glands (3). The CNTF-induced neuronal differentiation observed in SCG is mimicked by some neuroblastoma-derived cell lines, such as NBFL, that share a similar embryonic origin with sympathetic neurons (6,7).
The CNTF receptor (CNTFR) consists of a ligand-binding ␣ component, CNTR␣, and two signal transducing ␤ receptor subunits, gp130 and LIFR␤, that lack any obvious catalytic motif in their cytoplasmic domains but are constitutively associated with protein-tyrosine kinases of the JAK (Janus kinase) family. Ligand-induced heterodimerization of gp130 and LIFR␤ leads to tyrosine phosphorylation and activation of the associated JAKs, which in turn phosphorylate the ␤ receptor subunits and downstream signaling molecules (8,9). The phosphotyrosine residues on the ␤ receptor subunits provide binding sites for SH2 domain-containing signaling molecules, and mutational analysis of the cytoplasmic domain of gp130 has identified the SH2 domain docking sites for Stat3 and SHP-2 (10).
Stat3 is a member of a family of transcription factors that, when activated by tyrosine phosphorylation, homo-or heterodimerize and translocate to the nucleus to direct transcriptional responses (see Ref. 11 for review). Many of the genes that are coordinately regulated by CNTF, including choline acetyltransferase (ChAT), VIP, and SP, contain consensus STATbinding sites (12). In particular, within the 5Ј end of the VIP gene is a 180-base pair cytokine response element (CyRE) that binds a complex composed of Stat1␣ and Stat3 and mediates CNTF-stimulated VIP expression (12).
SHP-2 (SHPTP2, PTP1D, PTP2C, or Syp) is a protein-tyrosine phosphatase that contains a C-terminal catalytic domain and a tandem array of two SH2 domains at its N terminus (see Ref. 13 for review). SHP-2 has been implicated as an essential component in a variety of signaling cascades (13,14). In EGF, PDGF, insulin, and FGF pathways, SHP-2 acts as a positive regulator in the activation of mitogen-activated protein kinase (MAPK) and mitogenic signaling (15)(16)(17)(18)(19)(20)(21). Recently, a positive role for SHP-2 has also been shown in JAK/STAT signal transduction pathways triggered by prolactin or IFN-␣ * This work was supported in part by National Institutes of Health Grant RO1 NS27514 -08. 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.
§ Supported by a fellowship from the Associazione Italiana per la Ricerca sul Cancro, Rome, Italy and the Amgen Medical Discovery Fellowship.
The phosphotyrosine-dependent association SHP-2 displays with the gp130 component of the CNTF receptor complex suggests an important role for SHP-2 in this signaling pathway. We investigated this possibility by introducing an SHP-2 interfering mutant (N-P2) into NBFL cells and then examining CNTF-induced gene expression. We found that overexpression of the N-P2 mutant resulted in an up-regulation of CNTFinduced transcriptional activation of a VIP gene reporter plasmid, but a down-regulation of CNTF-induced c-fos expression. In NBFL cells that stably expressed the N-P2 mutant, gel shift analysis demonstrated that the N-P2 mutant prolongs STAT/ DNA binding activity while decreasing AP-1/DNA binding activity in response to CNTF. In sympathetic neurons, overexpression of the N-P2 mutant resulted in an increase in the CNTF-induced expression of endogenous VIP and SP genes. Together, these data indicate that in NBFL cells SHP-2 regulates CNTF-induced gene expression by controlling the signaling pathways that activate both STAT and AP-1 transcription factors. Hence, it is likely that SHP-2 controls the neuronal gene expression induced by CNTF in sympathetic neurons through a similar molecular mechanism.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-The various SHP-2 and SHP-1 interfering mutants were constructed using the polymerase chain reaction (PCR) and primers that resulted in PCR amplification of either the N-terminal SH2 domain or the colinear N-and C-terminal SH2 domains. For all interfering mutants, the 5Ј primer included sequences encoding the Flag epitope (DYKDDDDK). All mutants were cloned into the pRK5 mammalian expression vector (kindly provided by R. Derynck, University of California San Francisco, CA) under the control of the cytomeglovirus promoter, and pHSV-1 amplicon vector under the control of the HSV-1 IE 4/5 promoter (25) (kindly provided by R. Neve, McLean Hospital, Belmont, MA). All PCR-amplified DNAs were sequenced and found not to contain nonspecific Taq polymerase-induced mutations. The full-length SHP-2 CS mutant (kindly provided by B. Neel, Beth Israel Hospital, Boston, MA) was also cloned as an EcoRI fragment into pRK5.
Cell Culture and Transfection-NBFL cells were maintained as described previously (6). Cells were seeded at 1 ϫ 10 6 /10-cm, and 24 h later were cotransfected using the calcium phosphate precipitation technique (12). For VIP promoter activation studies, cells were cotransfected with 10 g of VIP-luc (luciferase) plasmid, 10 g of pRK5 expression vector encoding one of the SHP-2 or SHP-1 mutants, and 3 g of pRSV-CAT to control for transfection efficiency. For c-fos promoter activation studies, cells were cotransfected with 10 g of c-fos-CAT plasmid, 10 g of pRK5 expression vector encoding the N-P2 mutant, and 1 g of pRSV-luc to control for transfection efficiency. Transfected cells were placed in serum-free medium for 12-24 h before treatment with CNTF (25 ng/ml). Twelve hours later (VIP-luc) or 36 h later (c-fos-CAT), the cells were harvested, and luciferase (26) and CAT (27) activities were assayed. For stable transfectants, NBFL cells were cotransfected with 12 g of a pRK5 expression vector encoding the N-P2 mutant and 2 g of puromycin resistance plasmid (pJ⍀). Puromycinresistant colonies were selected in 1 g of drug/ml for 3 weeks. Single colonies expressing high levels of the N-P2 mutant were grown in medium supplemented with 1 g of puromycin.
Sympathetic neurons were prepared and cultured as described previously (28,29) with minor modifications. SCG were dissected from neonatal rats (Sprague-Dawley) and enzymatically dissociated (30). Cells were seeded in 24-well plates at a density of 1.5 ganglia/well and cultured in Opti-Mem (Life Technologies, Inc.) medium containing 100 ng/ml of 2.5 S nerve growth factor, 3% rat serum, and 10 M cytosine-1-␤-D-arabinoside (ara-C) (Sigma) to eliminate nonneuronal cells. After 2 days, cultures were switched to serum-free medium lacking ara-C. CNTF (10 ng/ml) was added on the 4th day for 12 h. All animal studies were done in accordance with guidelines for animal care by the Massachusetts General Hospital Subcommittee on Animal Care.
Immunoprecipitation and Western Blotting-NBFL cells were lysed for 20 min on ice in 50 mM Tris (pH 7.4), 150 mM NaCl, 1% Triton X-100, 50 mM NaF, 2 mM Na 3 VO 4 , aprotinin (5 g/ml), leupeptin (5 g/ml), and phenylmethylsulfonyl fluoride (50 g/ml). For immunoprecipitation, cell lysates were incubated with 4 g of a C-terminal SHP-2 polyclonal antibody (Santa Cruz Biotechnology) for 1 h at 4°C, and then with protein A-agarose for 1 h at 4°C. Immunoprecipitation of the Flag epitope-tagged N-P2 mutant from lysates of NBFL transfected cells was performed with anti-Flag M1/M2 Affinity gel (Kodak) for 2 h at 4°C. Immunoprecipitates were washed four times in lysis buffer and boiled for 5 min in 2ϫ SDS sample loading buffer. Immunoprecipitates were then subjected to SDS-polyacrylamide gel electrophoresis (8% gel), transferred to nitrocellulose membrane, and probed with the indicated antisera. All the immunoblots were visualized using the enhanced chemiluminescence (ECL) Western blotting Detection System Kit (Amersham Pharmacia Biotech). For reprobing, the membranes were incubated for 30 min at 50°C in stripping buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, 100 mM 2-mercaptoethanol).
Electrophoretic Mobility Shift Assay (EMSA)-Complementary oligonucleotides with GGG or GATC overhangs were annealed and labeled with [␣-32 P]dCTP using MMLV reverse transcriptase (RT) (Promega) or Klenow fragment (Promega). Complementary AP-1 oligonucleotides (Santa Cruz Biotechnology) had no overhang and were labeled with [␥-32 P]ATP using T4 polynucleotide kinase. Nuclear extracts were prepared, and binding reactions were carried out as described previously (12). For supershift experiments, nuclear extracts were preincubated with 2 g of poly(dI-dC) and anti-Stat1␣ (Transduction Laboratories), or anti-Stat3 (C-20; Santa Cruz Biotechnology), or control antibodies for 15 min on ice. After the addition of the probe (ϳ20,000 cpm), the binding reactions were continued for 20 min at room temperature. Samples were then resolved on a 5% non-denaturing polyacrylamide gel in 0.5 ϫ Tris borate/EDTA buffer at 200 V. The sequences of the complementary nucleotides used (SIE (SIEm67), G3, and noncanonical AP-1 (m3CyB)) have been reported (12,31).

Packaging of HSV-1 Amplicon Vectors and Infection of Neuronal Cultures-pHSV-N-P2
, pHSV-N-P1, and pHSV-LacZ amplicon DNAs were packaged into HSV-1 particles as described previously (25) with the following modifications. Plasmid DNAs were transfected into permissive Vero 2-2 cells using LipofectAMINE (Life Technologies, Inc.). One day later, cells were infected with dl5.2 helper virus (32). Virus were passaged and amplified three times to optimize helper:amplicon ratios. Amplified virus were purified and concentrated on sucrose gradients. pHSV-N-P2 and pHSV-N-P1 amplicon titers were determined by immunocytochemistry using M5 anti-Flag antibody (Kodak) on BHK cells. Helper virus titers were determined by plaque formation on Vero 2-2 cells. Titers of the viral stocks were 2 ϫ 10 4 infectious amplicon particles/l and 1 ϫ 10 5 dl5.2 helper particles/l, with an amplicon to helper ratio of 2:10. The pHSV-LacZ virus was used at the same helper virus titer as the helper virus titer of the other two stocks to control for viral cytotoxicity. Sympathetic neurons were infected on the 3rd day of culture with 0.5 l of each viral stock for 3 h. Twenty-four h after infection, the neurons were stimulated with CNTF (10 ng/ml) for 12 h.
Quantitative Reverse Transcriptase-PCR-To obtain a quantitative estimate of SP and VIP gene transcripts, we performed a quantitative RT-PCR assay (33), using neurofilament M (NF-M) as an internal standard. Total RNA was isolated from sympathetic neurons cultured in each well using the TRIzol reagent (Life Technologies, Inc.). Contaminating genomic DNA was removed by incubation with 10 units of DNase for 30 min at 37°C in the presence of 40 units of RNasin (Promega). RNA was phenol/chloroform extracted and ethanol-precipitated. An average of 1 g of RNA was obtained from each cell sample; half was used as control for genomic DNA contamination during PCR reactions and half was reverse transcribed as follows. Total RNA was mixed with 0.5 g of oligo(dT) (Life Technologies, Inc.), 0.5 mM dNTP, and 1ϫ RT buffer (Promega) in a final volume of 19 l and incubated at 65°C for 5 min. After 10 min at 37°C, 200 units of MMLV reverse transcriptase were added. The solution was incubated for 50 min at 37°C, and the RT was then inactivated by heating to 95°C for 5 min. The consistency of reverse transcription was validated by comparing the results of PCR analyses of 0.2, 0.5, and 1 g of total RNA; reverse transcription was found to be quantitative because the PCR amplification products of SP, VIP, and NF-M were directly proportional to the amounts of input RNA.
The cDNAs obtained from the reverse transcription reaction were used as template for PCR amplification. The sequence of each pair of primers has been reported (5, 34). Each cDNA was amplified in the presence of [␣-32 P]dCTP to quantify incorporation into the specific cDNA products after increasing numbers of cycles (Molecular Dynamics PhosphorImager®). A semi-logarithmic plot of the radioactivity incorporated versus the number of cycles was obtained for each cDNA. For SP and VIP, semi-logarithmic plots were obtained with cDNAs from both unstimulated and CNTF-stimulated sympathetic neurons. The number of cycles used to amplify each cDNA was chosen to allow the PCR amplification to proceed in a linear range. The efficiency of amplification for each cDNA was deduced from the slope of the semi-logarithmic plot (33) and was constant in different runs. In our hands, the method resolves 2-fold differences in input cDNA over a range of 2 orders of magnitude. PCR reactions were carried out in a final volume of 10 l consisting of cDNA, 1ϫ PCR buffer (Promega), 1 unit of Taq polymerase (Promega), 0.2 mM dNTP, 1.5 mM MgCl 2 , 0.5 M each primer, 1.5 ϫ 10 5 cpm [␣-32 P]dCTP. The number of cycles for each cDNA was as follows: SP, 24; VIP, 26; and NF-M, 27. The amount of cDNA (l of the reverse transcription reaction) was as follows: VIP and NF, 0.5 l; and SP, 2 l. The typical amplification profile was 95°C for 1 min, annealing temperature for 1 min, and 72°C for 90 s. The annealing temperature for SP and VIP was 60°C and for NF-M was 65°C. PCR amplified cDNAs were separated by agarose gel electrophoresis on 2% agarose gels, visualized by ethidium bromide staining and quantified by measuring the incorporated radioactivity using a PhosphoImager®. Amplified PCR products of VIP and SP in each sample were normalized with the corresponding NF-M PCR products. The identity of the amplified cDNAs was confirmed by subcloning and sequencing.

SHP-2 Is a Negative Signal Regulator in CNTF-induced VIP
Gene Transcription in NBFL Cells-To determine whether SHP-2 may play a role in CNTF-induced gene expression, we first used the human neuroblastoma cell line, NBFL, as a model system. NBFL cells respond to CNTF by coordinately inducing multiple neuropeptide genes, in a similar way to sympathetic neurons (6). To remove endogenous SHP-2 activity from the CNTF signaling cascade in NBFL cells, we used SHP-2 interfering mutants that lack the catalytic domain ( Fig.  1A). SHP-2 mutants should compete with endogenous SHP-2 for binding to the SHP-2 phosphotyrosine containing binding site on gp130. NBFL cells were cotransfected with SHP-2 mutants (N-P2, NC-P2, or CS-P2) and a CNTF-responsive luciferase reporter (VIP-luc) (6), which contains 1929 base pairs upstream of the transcription start site of the VIP gene. As a control for specificity of the SHP-2 mutants, we used the corresponding mutant forms of the related tyrosine phosphatase SHP-1 (N-P1 and NC-P1; Fig. 1A). While SHP-2 is broadly expressed, SHP-1 is mainly expressed in hematopoietic cells (see Ref. 13 for review). CNTF-stimulated VIP-luc reporter activity was ϳ11-fold in NBFL cells transfected with the parental expression vector alone (Fig. 1B, CT). Transfection of either the colinear N-and C-terminal SH2 domains (NC-P2) or the N-terminal SH2 domain (N-P2) of SHP-2 enhanced CNTFstimulated VIP-luc activity ϳ2-fold over the CT value, without affecting the basal level. The effect of the SHP-2 mutants was specific, because overexpression of the corresponding mutant forms (NC-P1 and N-P1) of SHP-1 did not significantly alter basal or CNTF-induced VIP-luc transcription over the CT value. Transfection of the full-length but catalytically inactive dominant negative SHP-2 mutant (CS-P2), in which the essential catalytic cysteine residue has been replaced with serine (21), enhanced CNTF-stimulated VIP-luc activity ϳ3-fold over the CT value. The stronger effect of CS-P2 compared with the other two SHP-2 mutants could reflect the ability of the CS mutant to tightly bind and "trap" substrates in addition to competing for binding to gp130 (21,35). These transfection studies using catalytically inactive SHP-2 mutants indicate that endogenous SHP-2 plays a negative role in the transmission of the CNTF signal in NBFL cells.
The N-P2 Interfering Mutant Specifically Associates with the CNTF Receptor Complex-Further experiments were performed using the N-P2 interfering mutant because its effect on CNTF-induced VIP expression in NBFL cells is substantial and comparable to the effect of the other mutants. To confirm that the N-P2 mutant interacts specifically with the CNTF receptor complex, we compared the pattern of phosphotyrosine-containing proteins that coimmunoprecipitate with endogenous SHP-2 with those that coimmunoprecipitate with the N-P2 interfering mutant following CNTF stimulation. We found that tyrosinephosphorylated bands of ϳ190 and ϳ130 kDa, which correspond to LIFR␤ and gp130 (36), coimmunoprecipitated with endogenous SHP-2 in CNTF-treated NBFL cells (Fig. 1C). Two additional tyrosine-phosphorylated proteins of ϳ90 and ϳ68 kDa also coimmunoprecipitated with the CNTF receptor complex. The ϳ68-kDa phosphoprotein comigrated with SHP-2 (Fig. 1C); however, the ϳ90 kDa phosphoprotein is of unknown identity (the protein was not recognized by antibodies to Stat1␣ or Stat3; data not shown). Interestingly, a 90-kDa tyrosinephosphorylated protein named SIRP␣1 has recently been characterized, which associates with SHP-2 upon EGF, PDGF, or insulin stimulation (37). Whether the 90-kDa protein that coimmunoprecipitates with SHP-2 upon CNTF stimulation might be SIRP␣1 still remains to be determined. Similar to that observed with endogenous SHP-2, tyrosine-phosphorylated gp130 and LIFR␤ coimmunoprecipitated with the Flagtagged N-P2 interfering mutant after CNTF stimulation ( Fig.  1C; ␣Flag lanes). The 90-kDa phosphoprotein, however, was not detected in the immunoprecipitates, indicating that the 90-kDa phosphoprotein does not associate with SHP-2 through the N-terminal SH2 domain.
The similar pattern of phosphotyrosine-containing proteins that coimmunoprecipitate with either the N-P2 interfering mutant or the endogenous SHP-2 after CNTF stimulation confirms that the N-terminal SH2 domain mutant of SHP-2 retains a high degree of specificity for tyrosine phosphorylated gp130. In addition, the substantial amounts of LIFR␤ and gp130 that coimmunoprecipitate with the N-P2 mutant in the transient transfections ( Fig. 1C; see ␣Flag lanes) indicate that the overexpressed mutant binds with high affinity to tyrosine phosphorylated gp130.
N-P2 Mutant Prolongs DNA Binding of STAT Complexes in Response to CNTF-The VIP gene contains a 180 base pairs CyRE within its 5Ј-flanking sequence that binds activated STAT after CNTF stimulation and which mediates induction of the VIP gene (12,31). To investigate how SHP-2 may regulate CNTF-dependent VIP transcriptional induction, we examined whether SHP-2 affects the ability of activated STAT to bind to elements within the CyRE using NBFL clonal cell lines that stably express the N-P2 mutant. NBFL cells were transfected with an expression vector containing the N-P2 interfering mutant and puromycin-resistant clones were established and characterized for the level of expression of the mutant protein. Several clones were found that expressed the N-P2 mutant at levels comparable with that of the endogenous SHP-2 ( Fig. 2A). In the following experiments, similar results were obtained with different clones; therefore, only data from one representative clone, NBFL-N2, are reported. Parental NBFL and NBFL-N2 cells were serum-starved for 12 h and then stimulated with CNTF for increasing times. Nuclear extracts were prepared before and after CNTF treatment, and the degree of STAT/DNA binding activity was assessed by EMSA using either the high-affinity variant of the c-fos sis-inducible element (SIE) sequence (38) or the acute phase response element (APRE)-like site of the CyRE (G3) (12) as a probe. In parental NBFL cells, CNTF treatment resulted in the stimulation of SIE binding activity of the STATs, which at 30 min was primarily represented by complex A (Fig. 2B). Supershift analysis confirmed this complex as being composed of Stat3 homodimers (Fig. 2B), as described previously (38,39). The two faster migrating complexes B and C were much less abundant. Complex C reacted almost quantitatively with Stat1␣ antibody, while complex B reacted with both anti-Stat1␣ and anti-Stat3 antibodies (Fig. 2B). None of the complexes were altered in the presence of irrelevant antibody. A small amount of complex A was still present 4 h after CNTF treatment, while complexes B and C were undetectable after 1 h. EMSA performed with NBFL-N2 nuclear extracts revealed an increase in the amount and duration of CNTF-induced binding of STAT proteins to the SIE probe (Fig. 2B); most notably in the Stat3-containing complex A, as shown by supershift analysis (Fig. 2B). When the G3

FIG. 1. Expression of interfering SHP-2 mutants specifically increases CNTF stimulation of VIP transcription.
A, schematic representation of wildtype and mutant SHP-2 and SHP-1. All mutants have been tagged at their N terminus with the Flag epitope. CS-P2 contains a cysteine to serine point mutation at amino acid 459 (21). B, NBFL cells were cotransfected with the VIP promoter-luciferase reporter construct, VIP-luc, and an empty expression vector (CT) or an expression vector encoding the indicated SHP-2 mutants (N-P2, NC-P2, and CS-P2) or SHP-1 mutants (N-P1 and NC-P1). The constitutively expressed pRSV-CAT (chloramphenicol acetyltransferase) reporter construct was cotransfected as an internal control for transfection efficiencies. Twenty-four h after transfection, cells were stimulated with CNTF (25 ng/ ml). Twelve h later, cell extracts were prepared and luciferase and CAT activities were measured. Luciferase activities were normalized to CAT activities. Results are the mean (Ϯ S.D.) of three independent experiments performed in duplicate. C, CNTF induces association of SHP-2 or the N-P2 interfering mutant with the tyrosine phosphorylated LIFR␤ and gp130 complex. NBFL cells were untransfected (Total and ␣SHP-2 lanes) or transfected with an expression vector encoding the Flag epitope-tagged N-P2 mutant (␣Flag lanes). Twenty-four h later, cells were stimulated with CNTF (25 ng/ml) for 10 min and lysed. The lysates were immunoprecipitated (IP) with antibody either to SHP-2 or to the Flag epitope and then Western blotted and probed with antiphosphotyrosine antibody, RC20H. After stripping, the blots were probed with antibody to SHP-2 or to the Flag epitope. Total lysates were run as a control. site within the CyRE was used as a probe in EMSA with nuclear extracts from NBFL or NBFL-N2 cells, complexes A, B, and C were visible after 30 min of CNTF stimulation (Fig. 2C). Supershift assays confirmed the identity of the most retarded complex A as a Stat3 homodimer, while the two faster migrating complexes B and C were Stat1␣/Stat3 heterodimers and Stat1␣ homodimers, respectively (Fig. 2C). While there was no detectable difference in the amount of CNTF-induced G3/STAT complexes in nuclear extracts from parental NBFL and NBFL-N2 cells at 30 min, there was a prolonged duration of binding of the Stat3-containing complex A in NBFL-N2 cells, which at 4 h was still visible as a band of similar intensity of that observed at 1 h.
The increased and/or prolonged formation of DNA/STAT complexes in response to CNTF stimulation observed in NBFL-N2 cells in comparison with parental cells indicates that SHP-2 may control CNTF-induced gene expression by directly affecting the level of STAT activation.

N-P2 Mutant Decreases AP-1 Binding Activity and c-fos Expression in Response to CNTF-Previous studies have
shown that in addition to activating the STATs, CNTF and related cytokines activate members of the MAPK family as well as the Fos-Jun transcription factors (40 -45). Furthermore, an AP-1 complex (referred to as Complex IV) containing c-Fos, JunB, and JunD has been described previously as binding a noncanonical AP-1 site within the VIP CyRE in a CNTF-dependent manner (31). This AP-1 partially overlaps with a C/EBP-like motif, responsible for binding to a complex (Complex III) pres-ent in nuclear extracts from untreated NBFL cells (31). Because the AP-1-like binding site appears to be important for full CNTF-dependent transcriptional activation mediated by the CyRE, we asked whether AP-1 binding activity was altered in NBFL-N2 cells. Nuclear extracts from parental and NBFL-N2 cells were incubated with either a consensus AP-1 binding site or the noncanonical AP-1 site of the CyRE as a probe. As expected, an increase in binding of nuclear proteins to both probes was observed after CNTF stimulation in parental NBFL cells, reaching a maximal level of binding after ϳ1 h (Fig. 3, A  and B). The addition of c-Fos antibody to the EMSA performed with the two AP-1 binding site probes inhibited complex binding, confirming that c-Fos is a major component of the AP-1 complex and complex IV (Fig. 3, A and B). CNTF-induced binding of AP-1 complexes to both probes was substantially reduced when nuclear extract from NBFL-N2 was used (Fig. 3,  A and B). No detectable differences in the formation of the uninducible complex III were observed between NBFL and NBFL-N2 cells.
To address whether the decrease in AP-1 binding activity observed in NBFL-N2 cells was due to a general decrease in CNTF-induced c-Fos levels, exerted by the N-P2 mutant, we compared the levels of c-Fos protein present in nuclear extracts from parental NBFL and NBFL-N2 cells. Western blotting analysis revealed a CNTF-dependent induction of c-Fos at 1 h (Fig. 3C) that paralleled the formation of AP-1 complexes on the consensus and noncanonical AP-1 site probes in parental cells (Fig. 3, A and B). In contrast, CNTF induction of c-Fos was and NBFL-N9) were separated on a 15% SDS gel, Western blotted, and then probed with a monoclonal antibody raised against the N-terminal region of SHP-2. The purified N-P2 fusion protein corresponds to the N-terminal SH2 domain of SHP-2. Endogenous SHP-2 and ectopically expressed N-P2 are indicated. B and C, nuclear extracts were prepared from NBFL or NBFL-N2 cells before and after CNTF stimulation (25 ng/ml) for the indicated times. EMSAs were performed with equal amounts of nuclear extract and labeled probe corresponding to the c-fos sisinducible element (SIE) (B) or the APRElike site of the CyRE (G3) (C). For supershift analysis (right panels), nuclear extracts from NBFL or NBFL-N2 cells treated for 30 min with CNTF (25 ng/ml) were incubated with SIE or G3 probes in the presence or absence of control antibody (Ab), anti-Stat1␣, or anti-Stat3 and then subjected to EMSA. Complexes A, B, and C refer to Stat3 homodimers, Stat1␣/ Stat3 heterodimers, and Stat1␣ homodimers, respectively. S3-SS, Stat3 supershift; S1-SS, Stat1␣ supershift. almost undetectable in NBFL-N2 cells.
To determine whether the N-P2 mutant down-regulates CNTF-induced c-fos expression (46,47), NBFL cells were cotransfected with the N-P2 mutant and a mouse c-fos-CAT reporter, containing the Ϫ356 to ϩ109 region of the c-fos promoter (48,49). This sequence includes all elements known to be important for regulation, including the SIE and the serum response element (SRE). Consistent with the decrease in c-Fos levels observed in NBFL-N2 cells, the N-P2 mutant abolished the CNTF-induction of the c-fos-CAT reporter (Fig. 3D). This effect, although small because of the high basal activity of the c-fos-CAT reporter, was reproducible and statistically significant (comparison of bars 2 and 4; p Ͻ 0.02). Together, these data indicate that endogenous SHP-2 is a positive regulator of CNTF-induced c-fos expression.

SHP-2 Coordinately Down-regulates CNTF Induction of Endogenous VIP and SP Genes in Sympathetic Neurons-
The VIP gene is one of several genes that are coregulated by CNTF in NBFL cells. The neurotransmitter gene expression induced by CNTF in NBFL cells mimics that which occurs in sympathetic neurons, suggesting that a common molecular mechanism mediates the response to CNTF in both NBFL cells and sympathetic neurons. Therefore, we examined whether SHP-2 may also play a role in CNTF-induced expression of the endogenous VIP gene in cultured sympathetic neurons. In addition, to address whether SHP-2 may control in a coordinate manner the transcriptional activation that occurs in CNTF-induced neuronal differentiation of sympathetic neurons, we also examined the role of SHP-2 in the expression of the SP gene.
Expression of SHP-2 in sympathetic neurons was confirmed by means of immunofluorescent staining and RT-PCR (data not shown). In time course (up to 72 h) and dose-response experiments (10 -25 ng/ml CNTF), we found that 10 ng/ml CNTF induced a substantial and quantifiable increase of VIP and SP mRNA levels in as little as 12 h. VIP and SP gene expression was determined by a quantitative RT-PCR assay, using NF-M as an internal standard (Fig. 4, A and B). Similar results were obtained when VIP and SP mRNA levels were normalized to constitutively expressed ␤-actin (data not shown).
To introduce the N-P2 mutant into sympathetic neurons, we

FIG. 4. CNTF induction of VIP and SP gene expression in sympathetic neurons as assayed by quantitative RT-PCR analysis.
A, sympathetic neurons were stimulated with CNTF (10 ng/ml) for 12 h on the 4th day of culture. Total RNA was extracted and reverse-transcribed with reverse transcriptase. VIP, SP, and NF-M cDNAs were subjected to PCR amplification in the presence of 32 PdCTP, separated on agarose gels, and quantified using a PhosphoImager®. The radioactivity in bands representing VIP and SP was normalized to the radioactivity present in the NF-M band from corresponding samples. The used an HSV-1 amplicon virus (HSV-N-P2). We also used an N-P1-containing virus (HSV-N-P1) to control for SH2 domain specificity, and a LacZ-containing virus (HSV-LacZ) to control for possible toxicity caused by the helper virus. Approximately 70% of the sympathetic neurons were infected with either HSV-N-P2 or HSV-N-P1 virus, demonstrating the efficiency at which the HSV-1 virus can transfer genes to postmitotic neurons (Fig. 5, B and C). The expression of the N-P1 and N-P2 interfering mutants increased up to 24 h and remained stable at least up to 72 h postinfection (data not show). A similar infection frequency and stability of expression was observed in cultures infected with HSV-LacZ and stained with X-gal (data not shown).
The effect of the SHP-2 mutant on VIP and SP gene expression was examined after 12 h of exposure to CNTF using the quantitative RT-PCR assay. In uninfected control cultures, CNTF stimulation resulted in VIP and SP mRNA inductions of 1.7-and 5.0-fold, respectively (Fig. 5D). In cultures infected with the HSV-N-P2 virus, however, CNTF induction of VIP and SP mRNAs was 4.3-and 11.2-fold, respectively. The effect was due to overexpression of the SHP-2 interfering mutant in HSV-N-P2 virus-infected cultures because CNTF induction of VIP and SP mRNAs in neurons infected with HSV-N-P1 or HSV-LacZ virus was similar to that observed in uninfected neurons. Normalization of VIP and SP mRNAs to ␤-actin as an internal standard gave similar results (data not shown). The downregulation of VIP and SP expression in sympathetic neurons elicited by SHP-2 suggests that SHP-2 controls CNTF-induced expression of these genes in a coordinated manner and likely through a common molecular mechanism. DISCUSSION Our studies show that SHP-2 negatively regulates CNTFinduction of VIP and SP gene expression while at the same time positively regulating c-fos expression. These data indicate that SHP-2 has dual and opposing roles in a signaling cascade triggered by the same ligand and clearly demonstrate that SHP-2 is a multifunctional protein in the regulation of the pathways activated by cytokines or growth factors.
The enhancement of CNTF-induced STAT binding activity observed in NBFL cells that overexpress the SHP-2 interfering mutant suggests that the negative function of SHP-2 could be at the level of the STATs or the STAT activating JAK kinases. Possible targets for SHP-2 phosphatase activity may be Jak1 or Jak2 since these kinases are activated by tyrosine phosphorylation upon CNTF stimulation in NBFL cells (data not shown) and a direct association of SHP-2 with the JAKs has recently been shown (50 -52). Inhibition of JAK-STAT signaling by SHP-2 through dephosphorylation of JAKs has been proposed in the interleukin-6 (IL-6)-triggered signaling pathway (51). In addition, a similar mechanism has been suggested for the related tyrosine phosphatase SHP-1, which plays a major role in terminating the erythropoietin-and ␣/␤ interferon-stimulated JAK-STAT pathway (23,53). Alternatively, because SHP-2 associates directly with gp130, it may dephosphorylate phosphotyrosine docking sites on gp130. One obvious possibility is the Stat3 binding site on gp130 (10). Lastly, because maximal activation of Stat1␣ and Stat3 appears to be dependent on serine phosphorylation (54 -56), it is possible that SHP-2 modulates STAT activation by controlling kinases of the MAPK family. Cytokines of the CNTF family induce both tyrosine phosphorylation and H7 (a serine/threonine kinase inhibitor)sensitive phosphorylation of Stat3 (57). Indeed, in NBFL cells, induction of VIP gene expression by CNTF is abolished by H7 (12,42). A link between SHP-2 and MAPK activation has been shown in mitogenic signaling pathways, with SHP-2 being generally a positive, but sometimes a negative regulator of MAPK activity (58). MAPKs can be activated by cytokines in a cell-dependent manner (59,60), and several lines of evidence suggest that JAKs may be required for activation of both STATs and ERK/MAPKs by cytokines, providing mechanisms for the optimization of gene induction (61)(62)(63)(64).
Both the AP-1 and STAT binding elements present in the CyRE of the VIP gene cooperate to achieve maximal rates of transcription upon CNTF stimulation (12,31). What is puzzling, however, is that while SHP-2 negatively regulates STAT binding activity, it in parallel positively regulates AP-1 binding activity by controlling levels of c-Fos. This observation raises the possibility that SHP-2 controls other regions, as yet unidentified, within the VIP CyRE, resulting in its down-regulation.
In the context of CNTF signaling, the down-regulation that SHP-2 exerts on VIP expression is in striking contrast with the up-regulation SHP-2 has on c-fos expression in NBFL cells. Multiple regulatory elements are present in the c-fos promoter, including the SIE and SRE. In our experiments, although overexpression of the N-P2 interfering mutant significantly increased STAT binding to the SIE, the actual effect on c-fos was to decrease expression. Therefore, it appears that the positive effect SHP-2 has on CNTF induction of c-fos expression is not through STATs, but more likely through regulation of transcription factors that bind other regulatory elements in the c-fos promoter, such as the SRE. Indeed, a number of studies have indicated only a minor contribution of the SIE to induction of c-fos expression (see Ref. 63, and references within) but a major role for the SRE in a variety of physiological settings (38,63,65).
A requirement of SHP-2 in the activation of c-fos transcription has been reported previously in insulin (15), PDGF (66), and ␣-thrombin (20) triggered pathways. Expression of c-fos is rapidly increased in many cell types in response to several mitogens (67), and therefore the up-regulation of c-fos that is dependent on SHP-2 might be related to the mitogenic role that SHP-2 has on these cell types. However, because CNTF is not a mitogenic factor for either NBFL cells or progenitor sympathetic neurons (68,69), but rather a factor that can induce differentiation, the induction of c-fos is likely to be needed in the context of neuronal differentiation.
The negative regulation that SHP-2 exerts on CNTF-induced expression of the SP and VIP genes is likely to be mediated through a common mechanism since STAT consensus binding sites are present in the promoter region of both genes (70,71). It is conceivable then that SHP-2 may have a similar effect on the expression of other genes that contain STAT binding sites and that are coordinately regulated by CNTF. Indeed, STAT binding sites have also been identified within the promoter regions of the SOM, cholecystokinin, enkephalin, and ChAT genes. Together with VIP and SP, these neuropeptide and neurotransmitter synthetic enzyme genes define the cholinergic phenotype in sympathetic neurons (72). Therefore it is possible that SHP-2 is present in these neurons to monitor neurocytokine-induced cholinergic differentiation.