J Biol Chem, Vol. 275, Issue 19, 14563-14572, May 12, 2000

Activation of Downstream Signals by the Long Form of the Leptin Receptor^*

Alexander S. Banks^§¶, Sarah M. Davis^§, Sarah H. Bates, and Martin G. Myers Jr.

From the  Research Division, Joslin Diabetes Center and Department of Medicine, Harvard Medical School, Boston, Massachusetts 02215

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The adipocyte-derived hormone leptin signals the status of body energy stores by activating the long form of the leptin receptor (LRb). Activation of LRb results in the activation of the associated Jak2 tyrosine kinase and the transmission of downstream phosphotyrosine-dependent signals. We have investigated the signaling function of mutant LRb intracellular domains under the control of the extracellular erythropoietin (Epo) receptor. By using this system, we confirm that two tyrosine residues in the intracellular domain of murine LRb become phosphorylated to mediate LRb signaling; Tyr⁹⁸⁵ controls the tyrosine phosphorylation of SHP-2, and Tyr¹¹³⁸ controls STAT3 activation. We furthermore investigated the mechanisms by which LRb controls downstream ERK activation and c-fos and SOCS3 message accumulation. Tyr⁹⁸⁵-mediated recruitment of SHP-2 does not alter tyrosine phosphorylation of Jak2 or STAT3 but results in GRB-2 binding to tyrosine-phosphorylated SHP-2 and is required for the majority of ERK activation during LRb signaling. Tyr⁹⁸⁵ and ERK activation similarly mediate c-fos mRNA accumulation. In contrast, SOCS3 mRNA accumulation requires Tyr¹¹³⁸-mediated STAT3 activation. Thus, the two LRb tyrosine residues that are phosphorylated during receptor activation mediate distinct signaling pathways as follows: SHP-2 binding to Tyr⁹⁸⁵ positively regulates the ERK  c-fos pathway, and STAT3 binding to Tyr¹¹³⁸ mediates the inhibitory SOCS3 pathway.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In the homozygous state, both the obese (ob) and diabetes (db) mutations result in morbid obesity, endocrine dysfunction, and predisposition to type 2 diabetes in mice (1, 2). Positional cloning revealed that the ob gene product, leptin, is a hormone secreted by adipocytes as a signal of energy stores; leptin levels are tied to nutritional status (3-5). The subsequent cloning of the leptin receptor (LR)¹ in turn demonstrated that the db mutations (as well as the rat fatty mutation) lie within the LR gene (6-10). Humans homozygous for rare mutations of leptin or the LR, like ob/ob and db/db mice, display obesity, hyperphagia, and endocrine dysfunction (11).

Alternate splicing of the LR results in the production of multiple isoforms (LRa-f) that contain a common extracellular domain (8, 12, 13). LRe lacks a transmembrane domain, LRa, -c, -d, and -f have short (approximately 30-40 amino acids) cytoplasmic extensions, and LRb has an approximately 300 amino acid intracellular tail (8, 10, 12, 13). The original db mutation (which yields the full db phenotype) alters splicing of the LRb and results in the truncation of LRb to a short form similar to LRa (8, 9, 14). Thus, signals unique to the intracellular tail of LRb are critical to the regulation of energy balance, endocrine function, and prevention of diabetes.

LRb is a member of the interleukin (IL)-6 receptor (IL6R) family of class I cytokine receptors (10, 15-18). Ligand binding to the LRb results in the activation of Jak2 by transphosphorylation and the subsequent tyrosine phosphorylation of tyrosine residues on the intracellular LRb (19-22). Importantly, LRb is the only LR isoform that contains intracellular tyrosine residues (10, 15).

In general, tyrosine phosphorylation of cytokine and growth factor receptors activates intracellular signals by recruiting specific signaling proteins with specialized phosphotyrosine-binding domains called SH2 domains (23, 24). Alternate SH2 domain isoforms in different signaling proteins bind phosphotyrosine in differing amino acid motifs; thus each tyrosine phosphorylation site recruits specific downstream signaling proteins based on its surrounding amino acids (24, 25). On LRb, Tyr⁹⁸⁵ recruits the tyrosine phosphatase SHP-2 (26, 27), and Tyr¹¹³⁸ recruits the STAT3 transcription factor (26-28). The utilization of these SH2 domain-containing proteins by LRb in the hypothalamus and pancreatic  cells implies that they may mediate physiologically important signals (27, 29, 30).

In addition to the direct binding of SHP-2 and STAT3, LRb also mediates a number of downstream actions in a variety of cells and tissues. LRb activates the extracellular factor-regulated kinases (ERKs), a set of serine/threonine kinases involved in the regulation of cellular physiology and gene transcription (19, 31-33). LRb activation also increases the levels of potentially important messages, including c-fos and SOCS3 (28, 29, 34-38). Expression of c-fos may correlate with the activation of a subset of arcuate nucleus neurons by LRb (38), whereas SOCS3 is an SH2 domain-containing Jak2 inhibitor that may block LRb signaling (37, 39, 40).

The mechanisms by which LRb mediates the activation of ERKs and the increased expression of c-fos and SOCS3 have been unclear. We present data suggesting that SOCS3 expression is regulated by STAT3, that SHP-2 controls the majority of ERK activation, and that c-fos accumulation is controlled by ERK activation. These results suggest that SHP-2 generates important positive signals during leptin action, and at least one function of STAT3 activation is to inhibit signaling by increasing levels of SOCS3.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Antibodies, Growth Factors, and Reagents-- Polyclonal Jak2 and Grb-2 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal PY (4G10), polyclonal Shc, and polyclonal phospho-Stat5A/B(pY) were obtained from Upstate Biotechnology, Inc. (Lake Placid, NY). Monoclonal SHP-2 for immunoblotting was from Transduction Laboratories (Lexington, KY). Immunoprecipitating SHP-2 was raised against a bacterial GST fusion protein containing the full-length SHP-2 (41). The MEK1/2 inhibitor PD98059 and antibodies directed against the phosphorylated (activated) forms of ERK (ERK(PT/PY)) and STAT3 (STAT3(PY)) were purchased from New England Biolabs (Beverly, MA). Rabbit LRb antisera were raised against a bacterial GST fusion protein containing the 100 COOH-terminal amino acids of murine LRb (amino acids 1075-1174). Bovine serum albumin (BSA) fraction V and affinity purified horseradish peroxidase-conjugated goat anti-rabbit or rabbit anti-mouse antibodies were purchased from Calbiochem. Chemiluminescence reagents were purchased from Pierce. Protein A-Sepharose 6MB and GSH-Sepharose were purchased from Amersham Pharmacia Biotech. Recombinant mouse erythropoietin (Epo) was purchased from PharMingen (San Diego, CA). ¹²⁵I-Epo was from Amersham Pharmacia Biotech, and [-³²P]dCTP was from NEN Life Science Products. ¹²⁵I-Protein A was from ICN (Los Angeles, CA).

ELR Chimeras-- The cDNA for the murine Epo receptor (EPOR) (the generous gift of Dr. Mark Goldsmith (UCSF)) was used as the template for PCR of the extracellular and transmembrane domains of the EPOR. This PCR generated a fragment containing a silent mutation resulting in the inclusion of a novel AflII site and a downstream NotI site in the 3'-end of the sequence encoding the EPOR transmembrane domain. The resulting fragment was subcloned into pBluescript (pBS, Stratagene) using Kpn and NotI (pBSEPORAfl). A DNA fragment containing the first 126 base pairs of the coding sequence for the intracellular domain of murine LRb (beginning with the novel AflII site and terminating with an endogenous SphI site and a novel NotI site) was generated by template-free PCR of two synthetic 80 oligonucleotides containing overlapping opposite-stranded 5'- and 3'-sequences. This fragment was subcloned into pBSEPORAfl using AflII and NotI, yielding pBSELRSph. The remainder of the LRb-coding sequence was obtained by PCR of murine genomic DNA using complementary oligonucleotides surrounding the missing LRb sequence. After initial PCR, the product was re-amplified using the same 5'-oligonucleotides and a 3'-oligonucleotide with a novel NotI site; this product was subcloned into pBSELRSph using SphI and NotI to generate pBSELR, which was sequenced in its entirety to ensure the correctness of the construct. Mutant intracellular domains were generated by PCR using mutagenic oligonucleotides that substituted Leu for Tyr⁹⁸⁵ (generating a novel HindIII site), Phe for Tyr¹⁰⁷⁷ (generating a novel Avr II site), and Ser for Tyr¹¹³⁸ (generating a novel SphI site). Mutant intracellular domains were subcloned using AflII and NotI, confirmed by digestion with the appropriate diagnostic enzyme for the novel restriction site(s), and sequenced to confirm the presence of the desired mutation and the absence of adventitious mutations. ELR inserts were subcloned into the expression vector pcDNA3 using KpnI and NotI.

Cell Lines-- All cells were transfected and maintained in a humidified atmosphere containing 5% CO₂ and 95% air at 37 °C. 293 cells were grown in DMEM supplemented with 10% FBS. EPOR and ELR constructs in pcDNA3 were transfected into 293 cells using LipofectAMINE (Life Technologies, Inc.) and selected in DMEM supplemented with 10% FBS and 750 µg/ml G418. 32D mouse myeloid progenitor cells were grown in RPMI 1640 supplemented with 10% FBS, 5% WEHI conditioned media (a source of IL-3). The EPOR and ELR constructs in pcDNA3 were transfected by electroporation as described (42) and selected in RPMI supplemented with 10% FBS, 5% WEHI, and 750 µg/ml G418 (Sigma). 32D and 293 clones expressing EPOR or ELR on the cell surface were identified for further experiments by ¹²⁵I-erythropoietin binding.

Preparation of Cell Lysates for Immunoprecipitation-- Prior to each experiment, subconfluent cells were made quiescent by incubation in serum-free media. 293 cells were made quiescent by overnight incubation in DMEM containing 0.5% BSA. 32D cells were collected by centrifugation and starved for 3 h in unsupplemented DMEM before treatment with 100 µM Na₃VO₄ for 30 min. All cells were stimulated with 0-50 ng/ml Epo for 0-60 min at 37 °C. After stimulation, 32D cells were immediately suspended in ice-cold PBS and collected by centrifugation; for 293 cells, the medium was immediately aspirated. Both cell lines were lysed in 20 mM Tris, pH 7.4, containing 137 mM NaCl, 2 mM EDTA, 10% glycerol, 50 mM -glycerophosphate, 50 mM NaF, 1% Nonidet P-40, 2 mM phenylmethylsulfonyl fluoride, and 2 mM sodium orthovanadate (Lysis Buffer). Insoluble material was removed by centrifugation at 16,000 × g at 4 °C for 5 min.

Protein concentrations of the resulting lysates were determined, and equivalent amounts of protein were added to the appropriate antibodies for immunoprecipitation or denatured in 2× Laemmli buffer for direct resolution by SDS-PAGE. For immunoprecipitates, lysates were incubated with antisera at 4 °C for 1 h to overnight followed by incubation with protein A-Sepharose for 30 min. Immune complexes were collected by centrifugation and washed three times in Lysis Buffer before denaturation in Laemmli buffer and separation by SDS-PAGE.

Immunoblotting-- SDS-PAGE gels were transferred to nitrocellulose membranes (Schleicher & Schuell) in Towbin buffer containing 0.02% SDS and 20% methanol. Membranes were blocked for 1 h at room temperature or overnight at 4 °C in buffer containing 20 mM Tris, pH 7.4, 150 mM NaCl, 0.01% Tween 20 (Wash Buffer) supplemented with 3% BSA (Block Buffer). Membranes were incubated with primary antibody in Block Buffer for 1 h, rinsed three times with Wash Buffer, incubated for 30 min in Block Buffer, and incubated for 1 h with either ¹²⁵I-protein A or the appropriate horseradish peroxidase-conjugated secondary antibodies in Block Buffer. Blots were rinsed four times in Wash Buffer before development with chemiluminescent reagents (SuperSignal, Pierce) or overnight exposure on Kodak XAR film. Some membranes were also exposed on a Molecular Dynamics PhosphorImager. For reblotting, membranes were rewet and/or washed in Wash Buffer twice for 15 min before being blocked overnight in Block Buffer and reprobed as above.

Preparation of RNA and Northern Blot Analysis-- Confluent 32D or 293 cells were prepared and stimulated with 0-50 ng/ml Epo for 0-60 min exactly as for immunoprecipitation. Cells were lysed with ice-cold Ultraspec RNA reagent (Biotecx, Houston, TX). Chloroform was added to the lysate, and following centrifugation, the aqueous layer was reserved for isopropyl alcohol precipitation. The resulting pellet of total cellular RNA was then washed twice with 75% cold ethanol, resuspended in 100 µl of DEPC-treated water, and quantified by spectrophotometry. Equivalent amounts of total RNA from each sample were resolved on a 1.5% agarose gel containing 2.2 M formaldehyde, 1× gel-running buffer (0.1 M MOPS, pH 7.0, 40 mM sodium acetate, 5 mM EDTA, pH 8.0) before being transferred to nylon membranes (Hybond N, Amersham Pharmacia Biotech) in 20× SSC overnight. RNA was cross-linked to the membrane by exposure to UV radiation before being washed in 6× SSC. Membranes were then stored frozen at 70 °C until being probed.

Probe cDNAs were excised from their vectors by restriction digestion, resolved by agarose gel electrophoresis, and purified using the Geneclean II system (Bio 101, Inc., La Jolla, CA). The c-fos cDNA in pGEM (the generous gift of Dr. Joel Elmquist, Beth Israel-Deaconness Medical Center, Boston, MA) was excised from the vector using BamHI. SOCS3 was amplified from murine brain mRNA using the Titan RT-PCR kit (Roche Molecular Biochemicals) and 27-mer oligonucleotides corresponding exactly to the murine SOCS3-coding region. The SOCS3 cDNA was then reamplified using primers adapted with 5'-BamHI and 3'-EcoRI sites and subcloned into pBS. The SOCS3 cDNA probe was released from pBSSOCS3 by digestion with BamHI and EcoRI. Probes were labeled using the NEBlot Kit (New England Biolabs, Beverly, MA) and [³²P]dCTP (NEN Life Science Products).

Association of Tyrosyl Phosphoproteins with Bacterially Expressed SH2 Domain Fusion Proteins-- GST fusion proteins containing the SH2 domains of GRB-2 have been described previously. GST fusion proteins were prepared as described (42), and integrity and concentration of the fusion proteins was determined by Coomassie Blue staining of fusion proteins resolved by SDS-PAGE.

Lysates of control or Epo-treated 32D or HEK293 cells were prepared as above for immunoprecipitation and incubated at 4 °C for 30 min with 2 µg of each GST fusion proteins immobilized on GSH-Sepharose. Complexes were collected by centrifugation and washed three times with Lysis Buffer before being resolved by SDS-PAGE.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Generation and Expression of ELR and ELR Mutants-- In order to assess signaling by the intracellular region of LRb, we generated a chimeric receptor molecule consisting of the extracellular and transmembrane domains of the murine Epo receptor (EPOR) and the intracellular domain of LRb (ELR chimera) (Fig. 1). Using chimeras such as these enables the examination of the function of the LRb intracellular domain under the control of a different ligand (Epo) thereby facilitating the analysis of intracellular signaling in cells that express endogenous leptin receptors. This chimeric receptor approach has previously proved useful in the study of a variety of receptors, including the insulin receptor and the human LRb, among others (26, 28, 43).

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Schematic diagram of erythropoietin receptor/LRb chimeras. Murine LRb contains three intracellular tyrosine residues (shown with surrounding amino acids) in addition to the conserved Box B1 and Box B2 motifs required for interaction with Jak2. The EPOR similarly contains Box B1, Box B2, and its own unique complement of intracellular tyrosine residues. The ELR chimeras contain the extracellular and transmembrane regions from the EPOR (hatched) and the intracellular sequences (black) from LRb. Four mutant ELR chimeras were constructed in which LRb intracellular tyrosine residues were mutated by substitution individually or in combination.

There are three intracellular tyrosine residues in murine LRb (Tyr⁹⁸⁵, Tyr¹⁰⁷⁷, and Tyr¹¹³⁸) each of which is conserved among all mammalian species examined to date (15). Of these, two (Tyr⁹⁸⁵ and Tyr¹¹³⁸) lie in hydrophilic regions and represent likely sites of tyrosine phosphorylation. We thus generated four mutant ELR constructs in order to examine signaling by each of the LRb intracellular tyrosine residues by replacing them with other amino acids: 1) Tyr⁹⁸⁵  Leu (ELR^L985); 2) Tyr¹¹³⁸  Ser (ELR^S1138); 3) Tyr⁹⁸⁵  Leu, Tyr¹¹³⁸  Ser (ELR^Dbl); 4) Tyr⁹⁸⁵  Leu, Tyr¹⁰⁷⁷  Phe, Tyr¹¹³⁸  Ser (ELR^Triple) (Fig. 1). By using these mutants, we reasoned that we would be able to examine the signaling function of Tyr⁹⁸⁵ and Tyr¹¹³⁸ individually and together and rule out any potential signaling function contributed by Tyr¹⁰⁷⁷. Whereas site-directed mutagenesis of tyrosine residues traditionally substitutes phenylalanine, we chose to substitute other non-tyrosine amino acids for Tyr⁹⁸⁵ and Tyr¹¹³⁸ because these substitutions not only replace the tyrosine residues with amino acids incapable of transmitting phosphotyrosine-dependent signals but also generate novel restriction sites facilitating identification of the mutations. We expressed the ELR and ELR mutants in stably transfected clones of 293 and 32D cells; similar expression of the various mutants was confirmed by comparing cell surface binding of ¹²⁵I-Epo (data not shown).

Activation of Signaling by the EPOR and ELR-- In order to assess the signaling function of ELR, we compared the ability of Epo to stimulate signaling events in control 293 cells and 293 cells expressing EPOR or ELR (Fig. 2). Cells were incubated in the presence or absence of 50 ng/ml Epo for 5 min and lysed. Lysates were immunoprecipitated or resolved directly by SDS-PAGE to determine activation of downstream signals. We determined the extent of tyrosine phosphorylation on Jak2 and ELR by immunoprecipitation with Jak2 or LRb antisera followed by PY immunoblotting. We assayed activation of STAT3, STAT5, and ERK by immunoblotting cell lysates with antisera specific for the phosphorylated/activated forms of these signaling molecules.

View larger version (57K): [in this window] [in a new window]   Fig. 2.   Signaling by EPOR and ELR. Control 293 cells (pcDNA3) or 293 cells stably expressing the ELR or the EPOR as indicated were made quiescent and incubated in the absence () or presence (+) of 50 ng/ml Epo for 5 min (A-E). Cells were lysed, normalized for protein content, and immunoprecipitated (IP) with Jak2 (A) or LRb (B) and resolved by SDS-PAGE or directly applied to SDS-PAGE (C-E) before being transferred to nitrocellulose membranes and immunoblotted (IB) with PY (A and B), STAT3(PY) (C), STAT5(PY) (D), or ERK(PT/PY) and exposed to autoradiography. Locations of phosphoproteins are indicated to the right of the autoradiograms. F, quiescent 32D cell lines expressing the indicated receptor molecules were stimulated with 50 ng/ml Epo for 60 min (the empirically determined maximum, data not shown) before being lysed for isolation of total cellular RNA. RNA was resolved by formamide-agarose gel electrophoresis, transferred to nitrocellulose, and probed with SOCS3-specific probes before exposure to autoradiography. Locations of phosphoproteins/messages are indicated to the right of the autoradiograms. These data are representative of at least three independent experiments.

Both EPOR and LRb are known to mediate tyrosine phosphorylation and activation of Jak2 (18-20, 44, 45). Although no activation of Jak2 was observed by Epo stimulation of control 293 cells, Jak2 tyrosine phosphorylation was stimulated during Epo treatment of both EPOR- and ELR-expressing cells (Fig. 2A). As expected, no phosphorylation of LRb-reactive proteins was detectable in control or EPOR-expressing cells; a broad 75-85-kDa tyrosyl-phosphorylated band corresponding to tyrosine-phosphorylated ELR was observed in LRb immunoprecipitates from ELR-expressing cells, however (Fig. 2B). In some experiments, this phosphorylated ELR resolved as two distinct bands that likely differ in glycosylation or phosphorylation state (data not shown).

STAT proteins are SH2 domain-containing transcription factors that bind to activated cytokine receptor signaling complexes where they become tyrosine-phosphorylated (46). STAT tyrosine phosphorylation results in nuclear translocation and transcriptional activation of target genes; thus, tyrosine phosphorylation of STAT proteins is required for and correlates with STAT transcriptional activation. The EPOR possesses a number of STAT5-binding sites and is known to mediate robust STAT5 phosphorylation/activation in a variety of systems; EPOR is not known to activate STAT3 (47-50). The LRb contains a STAT3-binding site in its COOH terminus and mediates robust tyrosine phosphorylation and activation of STAT3 (26-30); although LRb-mediated activation of STAT5 has been observed in some systems, it is comparatively weak (28, 29). Consistently, ELR mediated robust tyrosine phosphorylation of STAT3, whereas EPOR-mediated STAT3 tyrosine phosphorylation was virtually undetectable (Fig. 2C). Furthermore, whereas EPOR strongly mediated the tyrosine phosphorylation of STAT5, ELR barely mediated STAT5 tyrosine phosphorylation (Fig. 2D). Consistent with the known abilities of both EPOR and LRb to mediate ERK activation (19, 31, 32, 51), Epo stimulated ERK activation in 293 cells expressing either the EPOR or the ELR (Fig. 2E).

LRb activation also leads to accumulation of message for SOCS3, a direct inhibitor of Jak2 and a potentially important negative regulator of LRb signaling (37, 39, 40, 52). We therefore assessed the ability of ELR to stimulate SOCS3 message accumulation (Fig. 2F). Although we analyzed SOCS3 expression with similar results in 293 (data not shown) and 32D cells, the response was much stronger in the 32D cells (Fig. 2F), which express high levels of Jak/STAT-signaling proteins. Our analysis demonstrated that Epo treatment resulted in increased SOCS3 message in cells expressing ELR but not in control or EPOR-expressing cells.

Together, these data confirm that the ELR mediates LRb-specific signals under the control of Epo. These data further suggest that signaling by the intracellular tail of LRb is not impaired by fusion to the extracellular domain of the EPOR. Indeed, EPOR and ELR stimulate activation of Jak2 and ERK with similar time dependences and sensitivities to Epo (not shown). Additionally, since activation of the ELR, but not the EPOR, promotes SOCS3 mRNA accumulation, neither ERK nor STAT5 are sufficient for the stimulation of SOCS3, although STAT3 activation may be required.

Tyrosine Phosphorylation of Jak/STAT Proteins and ELR Mutants-- We next examined the ability of the ELR mutants to mediate tyrosine phosphorylation of Jak2 and the ELR in response to Epo (Fig. 3). We immunoprecipitated Jak2 and ELR from lysates of control or Epo-stimulated 293 cell lines stably expressing wild type or mutant ELR isoforms, resolved the immunoprecipitates by SDS-PAGE, and analyzed them by immunoblotting with PY. Whereas Jak2 was not phosphorylated during Epo treatment of control cells, it became phosphorylated during Epo treatment of EPOR and all ELR mutant-expressing cells (Fig. 3A). Thus, none of the mutations alter the ability of the ELR to mediate Jak2 activation. In contrast, mutation of certain LRb tyrosine residues in the tail of ELR dramatically altered ELR tyrosine phosphorylation. Whereas ELR, ELR^L985, and ELR^S1138 all became robustly tyrosine-phosphorylated during Epo stimulation, ELR^Dbl and ELR^Triple were not detectably phosphorylated in either 293 cells (Fig. 3B) or 32D cells (data not shown). Thus, mutation of either Tyr⁹⁸⁵ or Tyr¹¹³⁸ alone failed to abrogate ELR tyrosine phosphorylation, and mutation of these two residues together in ELR^Dbl (with Tyr¹⁰⁷⁷ intact) completely abolished ELR tyrosine phosphorylation. This analysis suggests that both Tyr⁹⁸⁵ and Tyr¹¹³⁸ are sites of tyrosine phosphorylation, although Tyr¹⁰⁷⁷ is not tyrosine-phosphorylated during LRb signaling. Interestingly, in both 293 (Fig. 3B) and 32D cells (not shown), tyrosine phosphorylation of ELR^L985 was consistently greatly reduced compared with ELR and ELR^S1138 (which displayed similar levels of tyrosine phosphorylation). Thus, Tyr⁹⁸⁵ may represent the major site of tyrosine phosphorylation on LRb.

View larger version (53K): [in this window] [in a new window]   Fig. 3.   Tyrosine phosphorylation of Jak2, ELR, and STAT3 by ELR mutants. Control 293 cells (pcDNA3) or 293 cells expressing EPOR, ELR, or ELR mutants as indicated were made quiescent and incubated in the absence () or presence (+) of 50 ng/ml Epo (A-C) or various concentrations of Epo (D) for 5 min. Cells were lysed, normalized for protein content, immunoprecipitated (IP) with Jak2 (A) or LRb (B), and resolved by SDS-PAGE or directly applied to SDS-PAGE (C and D) before being transferred to nitrocellulose membranes, immunoblotted (IB) with PY (A and B) or STAT3(PY) (C and D), and exposed to autoradiography. Locations of phosphoproteins are indicated to the right of the autoradiograms. These data are representative of at least three independent experiments.

Tyr¹¹³⁸ is predicted to bind the SH2 domain of STAT3, and others (26-29) have shown that this residue is in fact critical for the recruitment and tyrosine phosphorylation of STAT3 as well as the activation of STAT3-mediated transcription by the leptin receptor. We confirmed this result using our panel of ELR mutants expressed in 293 cells by immunoblotting Epo-stimulated cell lysates with antisera specific for the tyrosine-phosphorylated form of STAT3 (Fig. 3C). Indeed, as predicted, mutation of Tyr⁹⁸⁵ did not significantly alter STAT3 tyrosine phosphorylation, whereas any ELR mutant in which Tyr¹¹³⁸ was mutated was unable to mediate tyrosine phosphorylation of STAT3. Since others (26, 27) have suggested that recruitment of SHP-2 by Tyr⁹⁸⁵-mediated dephosphorylation and inhibition of STAT3 during LRb signaling, we analyzed the ability of ELR, ELR^L985, and ELR^S1138 to mediate tyrosine phosphorylation of STAT3 at various Epo concentrations (Fig. 3D). This analysis demonstrated that whereas mutation of Tyr¹¹³⁸ almost completely abrogated the ability of ELR to mediate STAT3 tyrosine phosphorylation, ELR and ELR^L985 mediated STAT3 tyrosine phosphorylation similarly, suggesting that the SHP-2 recruited by Tyr⁹⁸⁵ does not mediate significant dephosphorylation of STAT3 during LRb signaling.

GRB-2-mediated Signaling by ELR-- Control of ERK activation by tyrosine kinase-linked receptors generally involves the activation of the canonical p21^ras  raf  MEK1/2  ERK cascade (33). By and large, tyrosine kinases control this cascade by recruiting the GRB-2·mSOS complex (23, 33, 53, 54). GRB-2 is a small SH2/SH3 domain-containing adapter protein that binds mSOS (a guanine-nucleotide exchange factor for p21^ras) via its first SH3 domain. Binding of the SH2 domain of GRB-2 to tyrosyl phosphoproteins thus recruits the mSOS activator of p21^ras into the tyrosine kinase signaling complex. Although many receptors bind GRB-2 directly, a number phosphorylate intermediary signaling molecules, such as Shc, which similarly recruit GRB-2 to bind via its SH2 domain. We thus determined whether Shc or other proteins became tyrosine-phosphorylated and associated with GRB-2 during activation of the EPOR and ELR by PY immunoblotting Shc and GRB-2 immunoprecipitates from control and Epo-treated cells (Fig. 4A). Shc phosphorylation was readily detected during Epo stimulation of cells expressing the EPOR but not the ELR, suggesting that unlike the EPOR, LRb signaling does not utilize Shc. Activation of the EPOR resulted in the appearance of three tyrosyl phosphoproteins in GRB-2 immunoprecipitates; of these, the smallest comigrated with tyrosine-phosphorylated Shc, and previous studies suggest that the other two phosphoproteins represent the EPOR and GAB-1 (55). Activation of the ELR stimulated the association of GRB-2 with a 75-kDa tyrosyl phosphoprotein (pp75) that demonstrated slightly increased migration on SDS-PAGE compared with the GRB-2-associated EPOR. Thus, pp75 represents a potential mechanism for coupling the activated ELR to the GRB-2 pathway.

View larger version (43K): [in this window] [in a new window]   Fig. 4.   Shc- and GRB-2-associated tyrosyl phosphoproteins during activation of EPOR and ELR. Quiescent 293 cells expressing EPOR or ELR as indicated (A) or ELR (B and C) were incubated in the absence () or presence (+) of 50 ng/ml Epo for 5 min before being lysed. Lysates were normalized for protein content and immunoprecipitated (IP) with Shc or GRB-2 as indicated (A) or incubated with GRB-2, LRb, Jak2, GST, or GST-GRB-2-SH2 as indicated (B and C). Bound proteins were collected on protein A-Sepharose or glutathione-Sepharose, washed, denatured, and resolved by SDS-PAGE before being transferred to nitrocellulose membrane. A and B, the membrane was immunoblotted (IB) with PY. C, the membrane from B was rewet and immunoblotted with SHP-2. Locations of phosphoproteins are indicated to the left of the autoradiograms and the migration of molecular mass standards is indicated on the right. These data are representative of multiple independent experiments.

We furthermore determined whether pp75 was identical to the tyrosine-phosphorylated ELR and whether the SH2 domain of GRB-2 mediated binding of pp75 by examining ELR-stimulated tyrosyl phosphoproteins associated with GRB-2, LRb, and Jak2 immunoprecipitates or with the GRB-2 SH2 domain expressed as a bacterial GST fusion protein (Fig. 4B). This analysis demonstrated that both the tyrosine-phosphorylated ELR and Jak2 migrated well above pp75 on SDS-PAGE and that pp75 associated with the SH2 domain of GRB-2. Thus, pp75 is not identical to the ELR, and the SH2 domain of GRB-2 likely mediates pp75 association by binding phosphotyrosine residues on pp75. Furthermore, although not observed in the high percentage gel in Fig. 4A, phosphorylated Jak2 coprecipitated with GRB-2 and the GRB-2 SH2 domain during signaling by ELR (Fig. 4B). Thus, tyrosine-phosphorylated Jak2 may directly mediate some activation of the GRB-2 pathway during ELR signaling.

The LRb stimulates the tyrosine phosphorylation of the tyrosine phosphatase SHP-2 (26, 27); tyrosine-phosphorylated SHP-2 resolves at approximately 75 kDa on SDS-PAGE and is known to associate with GRB-2 in other systems (56-58). We therefore investigated whether pp75 and SHP-2 are identical (Figs. 4 and 5). In order to determine whether SHP-2 became tyrosine-phosphorylated and associated with GRB-2 during ELR signaling, we immunoprecipitated lysates of control or Epo-stimulated ELR-expressing cells with control, GRB-2, and SHP-2 antisera, and we analyzed the precipitated proteins by immunoblotting with PY or SHP-2 (Fig. 5, A and B). These immunoblots revealed the presence of a pp75 in both GRB-2 and SHP-2 immunoprecipitates from Epo-stimulated cells, although more pp75 was recovered in GRB-2 immunoprecipitates (Fig. 5A). Furthermore, SHP-2 was recovered in SHP-2 immunoprecipitates from both control and Epo-stimulated cells but only from Epo-stimulated cells in GRB-2 immunoprecipitates (Fig. 5B). Interestingly, Epo stimulation resulted in a slight decrease in mobility of a fraction of SHP-2 from SHP-2 immunoprecipitates, and all GRB-2-associated SHP-2 displayed two distinct forms with retarded mobility. The retardation of SHP-2 during ELR activation is consistent with phosphorylation of SHP-2 during ELR activation, since phosphorylated proteins display retarded migration on SDS-PAGE.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Identity of pp75 and SHP-2. Quiescent 293 cells expressing the ELR were incubated in the absence () or presence (+) of 50 ng/ml Epo for 5 min before being lysed. Lysates were normalized for protein content, immunoprecipitated (IP) with nonimmune serum (NI), GRB-2, or SHP-2, as indicated, and resolved by SDS-PAGE. Gels were transferred to nitrocellulose membranes and immunoblotted (IB) with PY (A) or SHP-2 (B). C, the membrane from A was rewet and immunoblotted with SHP-2. The migration of proteins is indicated to the left of the autoradiograms. These data are representative of multiple independent experiments.

These data suggest that ELR activation stimulates the tyrosine phosphorylation of SHP-2, which subsequently associates with GRB-2 (as pp75). Indeed, GRB-2 immunoprecipitates recover a great deal of SHP-2 (all of which is retarded and presumably phosphorylated). The relatively weak recovery of pp75 in SHP-2 immunoprecipitates is consistent with the relatively small amount of highly retarded (and presumably phosphorylated) SHP-2 observed in these same immunoprecipitates and suggests that our SHP-2 antibody may poorly recover tyrosine-phosphorylated SHP-2.

In order to confirm the identity of pp75 and SHP-2, we reprobed the membranes from Figs. 4B and 5A with SHP-2 to determine whether pp75 is detected by SHP-2 (Figs. 4C and 5C). The pp75 recovered in GRB-2 and GRB-2 SH2 domain precipitations was detected by SHP-2 as demonstrated by the direct superimposability of the pp75 and SHP-2 bands in reprobed membranes. Interestingly, both the upper and lower pp75 bands were detected by SHP-2, suggesting that both bands represent forms of SHP-2, but the higher band was more strongly detected with PY. These data suggest that the SHP-2 in the upper band is more highly tyrosine-phosphorylated, consistent with its decreased electrophoretic mobility.

Control of SHP-2/pp75 Phosphorylation and ERK Activation by Tyr⁹⁸⁵-- In order to determine whether the association of SHP-2/pp75 with GRB-2 required a particular tyrosine residue on ELR and to determine whether formation of the SHP-2/pp75 complex with GRB-2 correlated with ERK activation, we assayed the ability of the ELR mutants to mediate the tyrosine phosphorylation of SHP-2, the association of SHP-2/pp75 with GRB-2, and the activation of ERK in 293 cells. Whereas ELR and ELR^S1138 mediated the tyrosine phosphorylation of SHP-2/pp75 (Fig. 6A), and its association with GRB-2 (Fig. 6B), ELR^L985, ELR^Dbl, and ELR^Triple all failed to stimulate the tyrosine phosphorylation of SHP-2/pp75 and its association with GRB-2, suggesting that Tyr⁹⁸⁵ is required for the engagement of this signal but that Tyr¹¹³⁸ is not involved. Others have similarly shown that Tyr⁹⁸⁵ mediates tyrosine phosphorylation of SHP-2 during LRb signaling (26, 27), although the association of SHP-2 and GRB-2 had not been investigated.

View larger version (47K): [in this window] [in a new window]   Fig. 6.   Tyrosine phosphorylation of pp75/SHP-2 and activation of ERKs by ELR mutants. Control 293 cells (pcDNA3) or 293 cells expressing ELR or ELR mutants as indicated were made quiescent and incubated in the absence () or presence (+) of 50 ng/ml Epo (A-C) or various concentrations of Epo (D) for 5 min. Cells were lysed, normalized for protein content, immunoprecipitated (IP) with SHP-2 (A) or GRB-2 (B), and resolved by SDS-PAGE or directly applied to SDS-PAGE (C and D) before being transferred to nitrocellulose membranes and immunoblotted (IB) with PY (A and B) or ERK(PT/PY) (C and D) and exposed to autoradiography. Locations of phosphoproteins are indicated to the right of the autoradiograms. These data are representative of at least three independent experiments.

Analysis of ERK phosphorylation by the ELR mutants demonstrated that ELR and ELR^S1138 mediated similarly robust ERK activation, whereas ELR^L985, ELR^Dbl, and ELR^Triple activated ERK only approximately 20% as well as the wild type ELR (Fig. 6C). We observed similarly decreased maximal levels of ERK activation by ELR^L985 compared with ELR and ELR^S1138 in response to varying levels of Epo stimulation (Fig. 6D); activation of ERK occurred with similar Epo sensitivity during signaling by each of these receptors. This analysis suggests that LRb mediates ERK activation by two pathways as follows: one major pathway dependent upon Tyr⁹⁸⁵ (likely via SHP-2) and another minor pathway entirely independent of LRb tyrosine phosphorylation sites, perhaps mediated by the direct association of GRB-2 with Jak2 (Fig. 4B).

Role of ERK Signaling in the Regulation of c-fos Message Accumulation-- Leptin is known to induce expression of the immediate early gene/transcription factor c-fos in vivo (34-38); ERK signaling is known to activate transcription of c-fos during signaling by a number of growth factors (33). In order to determine the role of ERK signaling in the control of c-fos message levels, we examined the ability of ELR to mediate accumulation of c-fos mRNA in cells treated with PD98059, an inhibitor of the kinases (MEK1/2) that phosphorylate and activate the ERKs (Fig. 7). As expected, Epo stimulation of control 293 cells failed to mediate ERK activation, STAT3 phosphorylation, or accumulation of c-fos message. Similarly, ELR activation mediated all of these events. Preincubation with PD98059 abrogated ELR-induced ERK activation but did not alter tyrosine phosphorylation of STAT3, confirming that ERK activation lies downstream of MEK1/2 during ELR signaling and that STAT3 tyrosine phosphorylation occurs independently of the MEK1/2  ERK pathways. Furthermore, PD98059 inhibited c-fos mRNA accumulation, suggesting that c-fos mRNA accumulation lies downstream of ERK activation during ELR signaling in 293 cells.

View larger version (36K): [in this window] [in a new window]   Fig. 7.   ERK dependence of c-fos mRNA accumulation during ELR activation. Quiescent 293 cells or 293 cells expressing the ELR were preincubated for 30 min in the absence () or presence (+) of PD98059 and then incubated in the absence () or presence (+) of 50 ng/ml Epo for 5 (A and B) or 30 (C) min. A and B, cells were lysed, normalized for protein content, and resolved by SDS-PAGE before being transferred to nitrocellulose membrane for immunoblotting (IB) with STAT3(PY) (A) or ERK(PT/PY) (B). C, total cellular RNA was extracted, and equivalent amounts of RNA were loaded onto agarose gels, transferred to nitrocellulose, and probed with cDNA probes for c-fos. Locations of phosphoproteins/messages are indicated to the left of the autoradiograms. These data are representative of multiple independent experiments.

Role of ELR Tyrosine-dependent Signaling in the Regulation of c-fos and SOCS3 Message-- As well as stimulating c-fos message accumulation, LRb mediates accumulation of mRNA encoding the Jak2 inhibitor, SOCS3, in cultured cells and in vivo (19, 34-38). In order to determine the mechanisms by which the leptin receptor mediates c-fos and SOCS3 message accumulation, we investigated the ability of ELR and the ELR mutants to mediate c-fos and SOCS3 mRNA accumulation by Northern blotting total RNA from Epo-treated cells (Fig. 8). In 293 cells, increased c-fos expression during Epo stimulation was detected in cells expressing ELR and ELR^S1138 but not in cells expressing any mutant in which Tyr⁹⁸⁵ was mutated (ELR^L985, ELR^Dbl, and ELR^Triple) (Fig. 8A). Thus, Tyr⁹⁸⁵ of LRb likely mediates c-fos message expression by controlling activation of the ERK pathway.

View larger version (32K): [in this window] [in a new window]   Fig. 8.   Stimulation of c-fos and SOCS3 message accumulation by ELR mutants. A, control 293 cells (pcDNA3) or 293 cells expressing ELR or ELR mutants as indicated were made quiescent and incubated in the absence () or presence (+) of 50 ng/ml Epo for 30 min. Total cellular RNA was extracted and equivalent amounts of RNA were loaded onto agarose gels, transferred to nitrocellulose, and probed with cDNA probes for c-fos. B, control 32D cells (pcDNA3) or 32D cells expressing ELR or ELR mutants as indicated were made quiescent and incubated in the absence () or presence (+) of 50 ng/ml Epo for 60 min. Total cellular RNA was extracted, and equivalent amounts of RNA were loaded onto agarose gels, transferred to nitrocellulose, and probed with cDNA probes for SOCS3. Locations of messages are indicated to the right of the autoradiograms. These data are representative of multiple independent experiments.

We examined induction of SOCS3 mRNA by ELR and ELR mutants in 32D cells (Fig. 8B), in which strong increases in SOCS3 message are observed during ELR activation due to high levels of Jak/STAT signaling proteins in these cells. In contrast to what was observed with c-fos message, while ELR and ELR^L985 mediated increased SOCS3 message levels during Epo stimulation, any ELR mutant in which Tyr¹¹³⁸ was mutated (ELR^S1138, ELR^Dbl, and ELR^Triple) failed to mediate increased SOCS3 message levels. Similar results were observed in 293 cells (data not shown). Thus, the ability of ELR to mediate increases in SOCS3 message correlated with the presence of Tyr¹¹³⁸ and the activation of STAT3, suggesting that regulation of SOCS3 message levels by LRb is transcriptionally mediated by STAT3.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The discovery of leptin and the leptin receptor has illuminated not only the phenotypes of the obese (ob) and diabetes (db) mutations in mice but has added to our understanding of the mechanisms by which body energy status is sensed (2). The model that has rapidly emerged over the past several years is that leptin is secreted from adipocytes in amounts proportional to body energy reserves and that leptin reports energy status to the central nervous system and other tissues by binding and activating LRb. Leptin signals permit energy-expensive events such as growth and reproduction in addition to suppressing appetite and increasing metabolic rate to regulate body energy storage. The observation that high leptin levels in most cases of obesity fail to inhibit weight gain suggests the possibility of leptin resistance in obesity. In order to understand potential mechanisms of leptin resistance, it is critical to identify positive and negative mediators of leptin signaling.

The early recognition of LRb as a class I cytokine receptor similar to members of the IL6R family facilitated the rapid identification of several of the signaling mechanisms employed by LRb (10, 18, 22). Cytokine receptors contain no intrinsic enzymatic activity but transmit signals via non-covalently associated Jak family tyrosine kinases (22). Numerous studies have shown that the Jak2 tyrosine kinase is activated during LRb signaling (18-20); Jak1 may also play a role in LRb signaling (19). A rapid second step in signaling by IL6R family members is the activation of the STAT3 transcription factor by Jak kinase-dependent tyrosine phosphorylation (22). LRb similarly mediates STAT3 tyrosine phosphorylation and STAT3-dependent transcription during leptin signaling (26-30). Interestingly, whereas most IL6R family members undergo ligand-induced heterodimerization with other IL6R family members, such as gp130, LIFR, or OSMR, LRb appears to exist on the plasma membrane as a preformed homodimer that is activated by conformational changes during ligand binding as opposed to oligomerization with other IL6R family members (16, 17, 22).

We generated chimeric receptors containing the extracellular and transmembrane domains of the EPOR fused to wild type or mutant intracellular domains of LRb (ELR chimeras). We thereby facilitated the functional assessment of LRb-mediated signals in cells expressing isoforms of the LR and circumvented difficulties with the expression of the intact LRb. We chose the EPOR extracellular domain since the EPOR, like the LR, exists on the membrane as a pre-homodimerized receptor activated by a ligand-induced conformational change and does not heterodimerize with other receptors that might confound the analysis of the resulting signal (59, 60). Others (26, 28, 43) have used such chimeric receptor approaches to study signaling by a number of receptors, including cytokine receptors such as LRb.

Indeed, our analysis demonstrates that the ELR mediates LRb-specific signals, whereas the EPOR mediates a distinct set of intracellular signals. EPOR and LRb are known to mediate activation of Jak2 and ERKs as we demonstrated for both EPOR and ELR. EPOR mediates activation of STAT5, but not STAT3, and LRb mediates activation of STAT3, weakly (if at all) activating STAT5 (26-30, 47-50). Similarly, EPOR activates STAT5 but not STAT3 in our system, and ELR mediates the LRb-specific STAT3 activation and weakly activates STAT5. We chose to analyze signaling in two cell types, HEK 293 and 32D myeloid progenitor cells. Since the results of all assays were similar in both cell lines (data not shown), we chose to present the data from the 293 cells, with the exception of the SOCS3 data, which is from 32D cells. Whereas the SOCS3 data were similar in both cell lines, our mouse SOCS3 probe gave much stronger and more easily interpretable data in the mouse-derived 32D cells than in the human-derived 293 cells.

In general, tyrosine kinase-linked receptors transmit signals by recruiting specific downstream signaling proteins to individual sites of tyrosine phosphorylation (23-25). SH2 domains in the downstream signaling proteins mediate the specificity of this interaction by binding at high affinity to phosphotyrosine in the context of surrounding amino acids. Different signaling proteins contain alternate isoforms of the SH2 domain; each isoform requires phosphorylated tyrosine for binding but differs in its preference for the amino acids surrounding the phosphotyrosine residue. The specificity of a tyrosine kinase signal is thus determined by the phosphorylated tyrosine-containing motifs that it contains.

Ligand binding to cytokine receptors results in the activation of the associated Jak tyrosine kinases and phosphorylation of motifs on the receptors and the Jak kinase (21, 22). Murine LRb contains three tyrosine residues (Tyr⁹⁸⁵, Tyr¹⁰⁷⁷, and Tyr¹¹³⁸). Of these, Tyr⁹⁸⁵ and Tyr¹¹³⁸ reside in hydrophilic motifs likely to be accessible to the Jak2 tyrosine kinase, whereas Tyr¹⁰⁷⁷ is surrounded by hydrophobic sequences and is likely to be hidden within the folded protein and inaccessible. Indeed, whereas mutation of neither Tyr⁹⁸⁵ nor Tyr¹¹³⁸ alone abolished the ability of the receptor to become phosphorylated, simultaneous mutation of these two residues in the presence of an intact Tyr¹⁰⁷⁷ completely abrogated tyrosine phosphorylation of the receptor. These results suggest that Tyr⁹⁸⁵ and Tyr¹¹³⁸ represent the unique sites of tyrosine phosphorylation on murine LRb, in agreement with the observation that the homologous tyrosine residues are the sites of phosphorylation on human LRb (27). The observation that the tyrosine phosphorylation of ELR^L985 is greatly reduced compared with ELR and ELR^S1138 suggests that Tyr⁹⁸⁵ may be the major site of LRb phosphorylation in intact cells. Furthermore, Tyr⁹⁸⁵ and Tyr¹¹³⁸ also lie in motifs predicted to bind the SH2 domain-containing tyrosine phosphatase SHP-2 and the SH2 domain-containing transcription factor STAT3; consistently, mutation of these tyrosine residues in LRb and ELR abrogates the binding and phosphorylation of these proteins (26-28). Thus, numerous data point to these two residues as sites of LRb tyrosine phosphorylation.

As well as being unable to demonstrate phosphorylation of Tyr¹⁰⁷⁷ in receptors containing no other intracellular tyrosine residues, we have not observed any signaling function attributable to Tyr¹⁰⁷⁷. Thus, as suggested by the hydrophobic amino acid context of this residue, Tyr¹⁰⁷⁷ is unlikely to become phosphorylated or function directly in signal transmission.

In addition to the direct recruitment of SHP-2 and STAT3, activation of LRb mediates a number of other intracellular events both in cultured cells and in physiologically important regions of the central nervous system. These signals include activation of the extracellular signal-regulated kinases (ERKs), c-fos message accumulation, and SOCS3 message accumulation (Fig. 9) (19, 28, 29, 31, 32, 34-37). Whereas LRb activates these signals and they may mediate physiologically important events, the mechanisms whereby LRb mediates them were previously unclear.

View larger version (21K): [in this window] [in a new window]   Fig. 9.   Model of LRb signaling. Murine LRb contains three intracellular tyrosine residues (shown with surrounding amino acids) in addition to the conserved Box B1 and Box B2 motifs required for interaction with Jak2. Upon ligand stimulation, the associated Jak2 tyrosine kinase becomes activated, autophosphorylating and phosphorylating Tyr985 and Tyr1138 of the LRb. Phosphorylated Tyr1138 recruits STAT3, which is then tyrosine-phosphorylated by Jak2, whereupon it translocates to the nucleus to mediate the transcription of socs3 and other genes. SOCS3 ultimately feeds back upon and inhibits Jak2/LRb signaling (dotted line). Phosphorylated Tyr985 recruits SHP-2, which is then tyrosine-phosphorylated by Jak2. Phosphorylated SHP-2 (identical to pp75) binds GRB-2 and mediates the majority of ERK activation during LRb signaling. An additional minor amount of GRB-2 binding and ERK activation is mediated directly by Jak2 (thin line). The activation of ERK results in the transcription and accumulation of c-fos message.

Numerous tyrosine kinase-based signaling systems activate the ERK Ser/Thr kinases (33). ERKs translocate to the nucleus following phosphorylation/activation, whereupon they phosphorylate a set of nuclear transcription factors to mediate gene expression and control cell physiology. In many tyrosine kinase signaling systems, ERK activation begins with the recruitment of the small SH2/SH3 domain containing adapter protein GRB-2 in complex with the p21^ras activator mSOS (33, 53, 54). Binding of GRB-2 to the complex of tyrosine-phosphorylated signaling proteins results in the activation of p21^ras and the subsequent activation of the raf  MEK1/2  ERK pathway, wherein each kinase activates the downstream kinase by phosphorylation. We thus examined potential roles for GRB-2 and the well known GRB-2-binding protein Shc in signaling by the intracellular tail of LRb. Whereas EPOR appears to recruit GRB-2 via at least three pathways (direct recruitment via the tyrosine-phosphorylated EPOR and indirect recruitment via Shc and GAB-1) (55), ELR stimulated the association of GRB-2 with a 75-kDa tyrosyl phosphoprotein (pp75) and with Jak2. As expected, the SH2 domain of GRB-2 mediated the binding of GRB-2 to pp75 and Jak2.

The size and GRB-2 binding ability of pp75, as well as the previous observations that LRb recruits SHP-2 (26, 27), led us to examine closely the potential identity of SHP-2 and pp75. Both GRB-2 and SHP-2 recover PY-detectable pp75, and SHP-2 immunoblotting reveals that pp75 and a retarded (likely phosphorylated) portion of SHP-2 comigrate on SDS-PAGE during activation of ELR. Furthermore, mutation of Tyr⁹⁸⁵ abrogates not only the tyrosine phosphorylation of SHP-2 (our data and see Refs. 26 and 27) but also the association of pp75 with GRB-2, whereas mutation of Tyr¹¹³⁸ has no affect upon either. Thus, pp75 appears to represent tyrosine-phosphorylated SHP-2. Indeed, SHP-2 contains a tyrosine residue in a GRB-2-binding motif, and others (41, 56) have demonstrated the association of GRB-2 with tyrosine-phosphorylated SHP-2 via this tyrosine residue during signaling by other receptors.

In order to determine whether the SHP-2 pathway indeed mediates the activation of ERK signaling by the LRb intracellular domain, we also determined the ability of the ELR mutants to activate ERKs. This analysis demonstrates that approximately 80% of ERK activation by ELR is attributable to Tyr⁹⁸⁵-dependent signals. These results suggest that the recruitment of the SHP-2·GRB-2 complex by Tyr⁹⁸⁵ is required for the majority of ERK activation during LRb-mediated signaling. Since even the non-phosphorylated ELR^Dbl and ELR^Triple mediate a small (approximately 20%) amount of ERK activation observed in cells expressing wild type ELR, some signal independent of LRb tyrosine phosphorylation must mediate this portion of ERK activation. A pathway mediated directly by Jak2 or other Jak kinases may account for this activation; indeed we show some GRB-2 appears to interact directly with tyrosine-phosphorylated Jak2 during ELR signaling. Furthermore, the relative contribution of Tyr⁹⁸⁵ to the activation of ERK appears to be affected by the level of Jak2 expression in the cell.²

Numerous stimuli result in the accumulation of immediate early gene products such as c-fos (33). In various systems, c-fos message accumulation lies downstream of Ca²⁺ release, membrane depolarization, or ERK activation. During ELR signaling, inhibition of ERK activation by treatment with PD98059 (an inhibitor of the upstream ERK activators MEK1/2) or by mutation of Tyr⁹⁸⁵ results in greatly diminished c-fos message accumulation. This analysis suggests that c-fos message accumulation during LRb signaling is mediated by ERK activation.

The SOCS/CIS/JAB family of small SH2 domain-containing proteins is a recently described set of proteins that inhibit cytokine signaling (37, 39, 40). Although the individual members of this family may employ slightly differing mechanisms of action, they generally appear to operate by binding to the phosphorylated tyrosine residues in cytokine signaling proteins and either mediate their direct inhibition or ultimate degradation (40, 61). SOCS3 has been shown to bind specifically to Jak2, inactivating the kinase and blocking signaling transmission by a variety of cytokine receptors, including LRb (62). Importantly, SOCS3 expression is increased during activation of LRb in the hypothalamus, potentially mediating feedback inhibition of the leptin signal (37, 38). Our initial observation that ELR activation mediates accumulation of SOCS3 message, while EPOR activation does not, suggests that ERK and STAT5 (signals activated by the EPOR) activation are not sufficient to mediate SOCS3 message accumulation, although STAT3 activation, being ELR-specific, could conceivably mediate the stimulation of SOCS3 mRNA. We thus investigated the role of LRb-mediated signals in mediating SOCS3 message accumulation by analyzing the ability of the ELR mutants to stimulate SOCS3 message accumulation. Although mutation of Tyr⁹⁸⁵ and inhibition of the SHP-2  ERK pathway does not alter SOCS3 accumulation, mutation of Tyr¹¹³⁸ blocks the ability of ELR to activate STAT3 and its ability to control SOCS3 message accumulation. Thus, SOCS3 transcription is likely mediated by activated STAT3 independently of any input from the SHP-2/ERK pathway. We do not expect to observe inhibition of signaling by SOCS3 in our system, since we are observing signaling events within a few minutes of receptor activation, prior to the accumulation of SOCS3 in the cell.

Although our present data suggest that the recruitment of SHP-2 to the tyrosine-phosphorylated LRb tail may mediate ERK activation via the association of GRB-2 with SHP-2, data in other systems also suggest that SHP-2 mediates ERK activation via pathways that are independent of the SHP-2/GRB-2 interaction. Indeed, during signaling by insulin and a variety of cytokines, overexpression of a catalytically inactive SHP-2 mutant with intact tyrosine phosphorylation sites results in the inhibition of mitogen-activated protein kinase signaling, while overexpression of an active SHP-2 mutated for the GRB-2 binding site does not alter mitogen-activated protein kinase signaling (57, 63). Recent data from Drosophila torso receptor and the IL6R family member gp130 suggest that both SHP-2 catalytic activity and GRB-2 binding may be important for different reasons (58, 64). The catalytic activity of SHP-2 dephosphorylates the tyrosine residue critical for recruiting the p21^ras inhibitor rasGAP in the torso receptor pathway, thereby playing a permissive role in activation of the p21^ras  ERK pathway (58). In the gp130 and torso systems, the GRB-2 docking tyrosine residue on SHP-2 also appears to play a critical role in p21^ras  ERK activation (58, 64). Redundancy of pathways that engage GRB-2 in some signaling systems (insulin, EPOR, IL-3) but not others (torso, gp130, LRb) may explain the observation that direct SHP-2/GRB-2 association is sometimes required and other times not required. Our data demonstrating that ELR activation recruits GRB-2 predominantly via SHP-2 and that blocking this recruitment by mutation of Tyr⁹⁸⁵ also blocks the majority of ERK activation suggest that the binding of GRB-2 by SHP-2 may be critical in the activation of the p21^ras  ERK pathway by LRb. More data are needed to resolve this issue decisively, however.

Others (26, 27) have examined the potential role of SHP-2 in the negative regulation of LRb-mediated signaling. As in our present study, alterations in LRb-mediated tyrosine phosphorylation of Jak2 or STAT3 following mutation of the SHP-2-binding Tyr⁹⁸⁵ were not observed at endogenous levels of SHP-2. Overexpression of SHP-2 decreased tyrosyl phosphorylation of Jak2 and STAT3 in a Tyr⁹⁸⁵-dependent manner, however (27). Additionally, an increase in transcription from a STAT3-responsive vasoactive intestinal peptide promoter-luciferase reporter plasmid was observed following mutation of Tyr⁹⁸⁵ (26). Thus, others (26, 27) have suggested that, especially at high levels of SHP-2 expression, SHP-2 may decrease phosphotyrosine-dependent signaling by LRb and that SHP-2 may directly or indirectly decrease STAT3-mediated transcription from some promoters. In contrast, our data clearly demonstrate that at the endogenous levels of SHP-2 and other signaling proteins found in 293 and 32D cells SHP-2 mediates primarily positive signals to the ERK  c-fos pathway during LRb signaling.

SHP-2 binding to gp130 (which is highly homologous to LRb) both negatively regulates transcription from the haptoglobin promoter and positively mediates ERK activation (64); ERK activation requires both GRB-2 association by SHP-2 and the catalytic activity of SHP-2. Thus, both in the LRb and gp130 signaling systems, data suggest relevance of SHP-2 for activating ERKs and inhibiting STAT3-mediated transcription. In one model that accounts for much of the data, receptor-bound SHP-2 becomes tyrosine-phosphorylated, allowing the binding of GRB-2 and the activation of the pathway leading to ERK activation. At endogenous levels of SHP-2, the receptor-bound SHP-2 may selectively dephosphorylate one or a few specific sites on Jak2, selectively blocking signals that inhibit ERK activation, as in the D. torso system. Indeed, the SHP-2 phosphatase selects substrate phosphotyrosine residues based on specific steric and contextual requirements (57). The dephosphorylation of one or few tyrosine residues on Jak2 is unlikely to be observed in PY immunoblots due to the numerous (>15) individual phosphorylation sites on activated Jak2 (65). At high levels of SHP-2 expression (as achieved with transient overexpression), more sites on Jak2 are likely to be dephosphorylated. The effects of SHP-2 on STAT3-mediated transcription are unlikely to be mediated by tyrosyl dephosphorylation of STAT3, since mutation of the SHP-2-binding site on LRb does not detectably alter STAT3 tyrosine phosphorylation (our data and Refs. 26 and 27). Thus, the transcriptional effects of SHP-2 may be mediated indirectly through ERK-mediated serine phosphorylation of STAT3 or by ERK-regulated transcriptional events. Indeed, the vasoactive intestinal peptide-luciferase construct for which STAT3-dependent transcription was shown to be increased in LRb variants mutant for SHP-2 signaling (26) contains an ERK-responsive element (66, 67).

Knowledge of the mechanisms by which LRb controls both positive and inhibitory intracellular signals is critical to our understanding of how these signals regulate physiologic function and possibly leptin resistance. Previous models of LRb signaling suggested that the engagement of SHP-2 by LRb mediates inhibition of the leptin signal, whereas STAT3 activation effects positive signals (4, 15, 26, 27). We find no inhibition of receptor-proximal signaling events (i.e. Jak2 and STAT3 tyrosine phosphorylation) attributable to dephosphorylation by the Tyr⁹⁸⁵  SHP-2 pathway. In contrast, our present data suggest positive signaling to ERKs and c-fos via SHP-2 and the accumulation of the inhibitory SOCS3 via STAT3. It is now critical to examine the potential interplay of the SHP-2 and STAT3 pathways on other aspects of leptin physiology to determine whether both signals can play positive and negative signaling roles.

	ACKNOWLEDGEMENTS

We thank Dr. Mark Goldsmith for the murine erythropoietin receptor cDNA and Dr. Joel Elmquist for the c-fos cDNA probe. We also thank Drs. Morris White, Christian Bjorbaek, and Diane Fingar for helpful discussions.

	FOOTNOTES

^* This work was supported by National Institutes of Health Grant DK 56731 and a new investigator grant from The Medical Foundation/Harcourt General Charitable Trust (to M. G. M.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ Both authors contributed equally to this work.

^¶ Present address: Dept. Microbiology, College of Physicians and Surgeons, Columbia University, New York, NY 10032.

To whom correspondence should be addressed: Research Division, Joslin Diabetes Center, 1 Joslin Place, Boston, MA 02130. Tel.: 617-735-1967; Fax: 617-735-1970; E-mail: martin.myers@joslin.harvard.edu.

² C. Bjorbaek, personal communication.

	ABBREVIATIONS

The abbreviations used are: LR, leptin receptor; Epo, erythropoietin; IL, interleukin; IL6R, interleukin-6 receptor; ERK, extracellular signal-regulated kinase; EPOR, Epo receptor; PCR, polymerase chain reaction; BSA, bovine serum albumin; DMEM, Dulbecco's modified Eagle's medium; FBS, fetal bovine serum; PAGE, polyacrylamide gel electrophoresis; GST, glutathione S-transferase; MOPS, 4-morpholinepropanesulfonic acid.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES