Phosphorylation of JAK2 at Serine 523: a Negative Regulator of JAK2 That Is Stimulated by Growth Hormone and Epidermal Growth Factor

The tyrosine kinase JAK2 is a key signaling protein for at least20 receptors in the cytokine/hematopoietin receptor superfamilyand is a component of signaling for multiple receptor tyrosinekinases and several G-protein-coupled receptors. In this study,phosphopeptide affinity enrichment and mass spectrometry identifiedserine 523 (Ser523) in JAK2 as a site of phosphorylation. Aphosphoserine 523 antibody revealed that Ser523 is rapidly buttransiently phosphorylated in response to growth hormone (GH).MEK1 inhibitor UO126 suppresses GH-dependent phosphorylationof Ser523, suggesting that extracellular signal-regulated kinases(ERKs) 1 and/or 2 or another kinase downstream of MEK1 phosphorylateSer523 in response to GH. Other ERK activators, phorbol 12-myristate13-acetate and epidermal growth factor, also stimulate phosphorylationof Ser523. When Ser523 in JAK2 was mutated, JAK2 kinase activityas well as GH-dependent tyrosyl phosphorylation of JAK2 andStat5 was enhanced, suggesting that phosphorylation of Ser523 inhibitsJAK2 kinase activity. We hypothesize that phosphorylation of Ser523in JAK2 by ERKs 1 and/or 2 or other as-yet-unidentified kinases actsin a negative feedback manner to dampen activation of JAK2 in responseto GH and provides a mechanism by which prior exposure to environmentalfactors that regulate Ser523 phosphorylation might modulatethe cell's response to GH.

INTRODUCTION

Top

Introduction

JAK2 is a member of the Janus family of cytoplasmic tyrosinekinases (JAK1, JAK2, JAK3, and tyk2), critical for transmissionof cytokine-induced proliferative, differentiation, and survivalsignals. Roughly two-thirds of the cytokine/hematopoietin superfamilyof receptors activate JAK2, including the receptors for growthhormone (GH), prolactin, erythropoietin (EPO), gamma interferon,and leptin (4, 61). Dysregulation of JAKs has been linked toimmune diseases (10) and a variety of cancers (6, 33, 38, 40, 41, 49, 70).Upon binding to the GH receptor, GH causes a particularly rapid,transient, and robust activation of JAK2 (3). Activation ofJAK2 is thought to occur when JAK2 molecules are brought intoclose enough proximity to trans phosphorylate on tyrosine residueslocated in the activation loop of each kinase (29). ActivatedJAK2 molecules then phosphorylate themselves, their dimerizedJAK2 partner, associated receptor subunits, and a variety ofsignaling molecules such as signal transducers and activatorsof transcription 3 and 5 (stat3 and stat5) (25).

Auto- and/or transphosphorylation of JAK2 is therefore an importantstep in the transduction of cytokine signaling. Murine JAK2contains 49 tyrosines, of which more than 10 appear to be sitesof autophosphorylation (5, 19, 20, 45; L. S. Argetsinger etal., unpublished data). Phosphorylation of some of these tyrosinesappears to regulate the activity of JAK2, for example, tyrosine1007 (20) and more recently identified tyrosine 221 and tyrosine570 (5, 19). Some phosphorylated tyrosines recruit other signalingmolecules that contain motifs of phosphotyrosine binding (SH2and PTB domains) to JAK2. Phosphorylated tyrosine 1007, whichis critical for activation of JAK2, binds the negative regulatorsSOCS-1 (74), SOCS-3 (57), and phosphatase PTP1B (47); phosphorylatedtyrosine 966 binds p70, an SH3 domain-containing protein ofunknown function (11); and phosphorylated tyrosine 813 bindsSH2-Bß (39), an adapter protein and enhancer of JAK2kinase activity (51, 53).

Little is known about the phosphorylation of JAK2 on serinesand threonines. Protein kinase C- ${delta}$ (PKC- ${delta}$ )-mediated phosphorylationof JAK2 on serine/threonine has been reported to inhibit activationof JAK2 (37). In other proteins, phosphorylation on serinesand threonines has a variety of effects. Phosphorylation atserines and threonines can induce conformational changes witha consequent increase or decrease in the catalytic activityof the protein (65; reviewed in reference 42). Serine/threoninephosphorylation of the receptor tyrosine kinases, ErbB-2 (35)and the insulin receptor (48, 63), as well as insulin receptorsubstrate 1 and 2 (48, 50) and Stat3 (13), decreases tyrosyl phosphorylationof these molecules and downregulates downstream signaling. Serinephosphorylation is reported to modulate the transcriptionalactivity of Stat3 (68). Phosphorylated serines and threoninescan also recruit other signaling proteins: for example, 14-3-3binds phosphoserine/threonine-containing motifs in its numerousbinding partners (reviewed in reference 44). Proteins bind to phosphorylatedserines and threonines via a large variety of domains, includingWW, FHA, WD40, and Polo-box domains (reviewed in reference 71).To begin to understand how serine/threonine phosphorylationof JAK2 affects signaling, we set out to identify sites of serine/threonine phosphorylationin JAK2 and to determine their function in cytokine signaling.

In this study, serine 523 was identified as a site of phosphorylationin JAK2 using mass spectrometry (MS). Using a phosphospecificantibody directed against phosphorylated serine 523, we demonstratedthat phosphorylation of serine 523 can be stimulated by bothGH and epidermal growth factor (EGF) and provide evidence thatGH-dependent phosphorylation of JAK2 at serine 523 is mediatedby extracellular signal-regulated kinases (ERKs) 1 and/or 2or some other kinase downstream of MEK1. Finally, we show thatmutation of serine 523 to alanine increases JAK2 kinase activityand both the basal and GH-stimulated levels of tyrosyl phosphorylationof JAK2 and Stat5b, suggesting that phosphorylation of serine523 inhibits JAK2 kinase activity. We hypothesize that phosphorylationof serine 523 in JAK2 by ERKs 1 and/or 2 or other, as-yet-unidentifiedkinases acts in a negative feedback manner to dampen the activationof JAK2 in response to GH and provides a mechanism by whichprior exposure to other ERK-activating ligands or environmentalfactors that regulate Ser 523 phosphorylation might modulatethe cell's response to GH.

MATERIALS AND METHODS

Top

Materials and Methods

Materials. Recombinant 22,000-Da human GH was a gift from Eli Lilly. HumanEGF was from JRH Biosciences (see Fig. 5) or Biosource (seeFig. 8). Recombinant protein A-agarose was from Repligen. Bovineserum albumin (CRG-7) was from Intergen. Bac-to-Bac HT baculovirusexpression system, calf serum, and Dulbecco's modified Eagle'smedium were from Invitrogen. Aprotinin, leupeptin, and TritonX-100 were from Roche. The enhanced chemiluminescence detectionsystem and nitrocellulose paper were from Amersham PharmaciaBiotech. Protein molecular weight standards were from SantaCruz and Invitrogen. X-ray film was from Kodak. p81 paper wasfrom Fisher. The QuikChange site-directed mutagenesis kits were fromStratagene. MEK inhibitor UO126 (stock solution, 20 mM in dimethyl sulfoxide[DMSO]) was from Promega. Phorbol 12-myristate 13-acetate (PMA;stock solution, 10 mM in DMSO) was from Sigma. The protein phosphatasepp2A was from Upstate.

View larger version (35K):
[in this window]
[in a new window]
FIG. 5. EGF promotes the phosphorylation of JAK2 at serine 523. 3T3-F442A cells were stimulated with vehicle or with 125 ng EGF/ml for 5, 10, 15, or 30 min. Cells were lysed, and lysates were immunoblotted with ${alpha}$ pS523 JAK2 (top panel), ${alpha}$ pY1007/1008 JAK2 (middle panel), or ${alpha}$ JAK2 (bottom panel), n = 2. The migration of JAK2 and a nonspecific band (ns) is indicated.

View larger version (21K):
[in this window]
[in a new window]
FIG. 8. Pretreatment with EGF inhibits GH-dependent activation of JAK2. 3T3-F442A cells were treated with 125 ng EGF/ml as indicated or treated with 125 ng EGF/ml or vehicle for 30 min and then treated with vehicle or 50 ng GH/ml as indicated. Cell lysates were blotted with ${alpha}$ pY1007/1008 JAK2, ${alpha}$ PY, or ${alpha}$ JAK2. The intensities of the bands corresponding to phosphorylated JAK2 were quantified, normalized to levels of JAK2, and graphed as the fraction of maximum phosphorylation detected with ${alpha}$ pY1007 (A) and ${alpha}$ PY (B) (error bars denote ranges of values, n = 2).

Antibodies. JAK2 antibody ( ${alpha}$ JAK2) was raised against a peptide correspondingto amino acids 758 to 776 of murine JAK2. The ${alpha}$ JAK2 used forimmunoprecipitation (at a dilution of 1:1,000) was preparedby our laboratory in conjunction with Pel-Freeze Biologicals (3).The ${alpha}$ JAK2 used for Western blotting was from Upstate USA, Inc.(dilution of 1:7,500 for Fig. 2 and 4), or Biosource ( ${alpha}$ JAK2 monoclonalclone no. 829; dilution of 1:2,000 for Fig. 3 and 5 to 8). Antibodythat recognizes a peptide containing phosphorylated serine 523 ( ${alpha}$ pS523JAK2) and antibody to a peptide containing phosphorylated tyrosine570 in JAK2 ( ${alpha}$ pY570 JAK2) (5) were developed in conjunction withUpstate USA, Inc., and used at a 1:2,000 dilution for immunoblotting.Antibody to phosphorylated tyrosines 1007/1008 in JAK2 ( ${alpha}$ pY1007/1008JAK2) was a gift from M. Myers (University of Michigan, AnnArbor) (19). Antibodies to phospho-p44/42 mitogen-activatedprotein kinase ( ${alpha}$ pErk; E10) and total ERK ( ${alpha}$ Erk) were from CellSignaling and used at a dilution of 1:2,000 for immunoblotting.Antibody to phosphotyrosine ( ${alpha}$ PY; 4G10) was from Upstate USA,Inc., and used at a dilution of 1:7,500 for immunoblotting.Antibody to Stat5b against amino acids 711 to 727 of murineStat5b ( ${alpha}$ Stat5b) was from Santa Cruz Biotechnology, Inc., andused at a 1:5,000 dilution for immunoblotting. Antibody to phosphorylatedtyrosine 699 of Stat5b ( ${alpha}$ pStat5b) was from Zymed LaboratoriesInc. and was used at a 1:7,500 dilution for immunoblotting.AlexaFluor 680 anti-rabbit antibody from Molecular Probes andIRdye800 anti-mouse antibodies from Rockland were used at a dilutionof 1:20,000. Horseradish peroxidase-conjugated anti-mouse and anti-rabbitimmunoglobulin G antibodies from Santa Cruz were used at a dilutionof 1:10,000.

View larger version (18K):
[in this window]
[in a new window]
FIG. 2. GH promotes the phosphorylation of JAK2 at serine 523. (A) 293T cells transfected with cDNA (1 µg) encoding either JAK2 (lane 1) or JAK2 S523A (lane 2) were lysed. The lysates were immunoblotted (IB) with ${alpha}$ pSer523 JAK2 (upper panel) or ${alpha}$ JAK2 (lower panel) (n = 4). wt, wild type. (B) 3T3-F442A cells were treated with vehicle or with 500 ng GH/ml for 5, 10, or 15 min. The cells were lysed, and lysates were immunoblotted (IB) with ${alpha}$ pS523 JAK2 (top panel), ${alpha}$ pY1007/1008 JAK2 (second panel), ${alpha}$ pY570 JAK2 (third panel), or ${alpha}$ JAK2 (bottom panel). (C) Replicates for the experiment shown in panel B were normalized to levels of JAK2. Means ± standard errors of the means are shown for n = 6. (D) 3T3-F442A cells were treated with vehicle (lanes 1 and 5) or with 500 ng GH/ml (lanes 2 to 4 and 6 to 8). Cells were lysed, and lysates were incubated without (lanes 1 to 4) or with (lanes 5 to 8) 0.7 U pp2A at 37°C for 60 min. Cell lysates were immunoblotted with either ${alpha}$ pS523 JAK2 (top panel) or ${alpha}$ JAK2 (bottom panel). The migration of JAK2 is indicated. (E) Results from the experiment shown in panel D were quantified and normalized to levels of JAK2.

View larger version (31K):
[in this window]
[in a new window]
FIG. 4. MEK1 inhibitor UO126 inhibits PMA-stimulated phosphorylation of JAK2 on serine 523. (A) 3T3-F442A cells were pretreated with DMSO (lanes 1 to 3) or with 10 µM UO126 in DMSO (lanes 4 and 5) for 30 min prior to stimulation with vehicle (lane 1) or with 1 µM PMA for the indicated times (lanes 2 to 5). Cells were lysed, and the lysates were immunoblotted (IB) with ${alpha}$ pS523 JAK2 (top panel), ${alpha}$ JAK2 (middle panel), or ${alpha}$ pErk (bottom panel). The migration of JAK2, a nonspecific band (ns), and the ERKs is indicated. (B) Replicates for the experiment shown in panel A were quantified and normalized to levels of JAK2. Means and the range are shown for n = 2.

View larger version (32K):
[in this window]
[in a new window]
FIG. 3. MEK1 inhibitor UO126 inhibits GH-stimulated phosphorylation of serine 523 in JAK2. (A) 3T3-F442A cells were incubated with vehicle (lanes 1 to 3) or with 10 µM UO126 (lanes 4 to 6) for 30 min. Cells were then treated with vehicle (lanes 1 and 4) or with 500 ng GH/ml for 5 or 10 min (lanes 2 and 3 and lanes 5 and 6). Cell lysates were immunoblotted (IB) with ${alpha}$ pS523 JAK2 (top panel), ${alpha}$ JAK2 (second panel), ${alpha}$ pErk (third panel), or ${alpha}$ Erk (bottom panel). The migration of JAK2 and the ERKs is indicated. (B) Replicates for the experiment shown in panel A were quantified and normalized to levels of JAK2. Means ± standard errors of the means are shown for n = 4.

Plasmids. cDNA encoding rat GH receptor was kindly provided by G. Norstedt(Karolinska Institute, Sweden) (17). cDNAs encoding murine JAK2or kinase-inactive JAK2 K882E in the prk5 vector were kindlyprovided by J. Ihle (St. Jude Children's Research Hospital,Memphis, TN) (60). cDNA encoding rat SH2-Bß was inthe prk5myc vector (53). cDNA encoding rat Stat5b in the pRc/CMVvector was a gift of L. Yu-Lee (Baylor College of Medicine,Houston, TX) (43). cDNA encoding JAK2 S523A was created fromthe JAK2 expression vector above using the QuikChange site-directedmutagenesis kit from Stratagene. The primer (sense strand, mutationin lowercase) used for JAK2 S523A was 5'-CTGATGTTCAGATCgCACCAACATTACAGAGGC-3'. Mutationswere verified by DNA sequencing. Amino acids in JAK2 are numberedaccording to NCBI accession number NP_032439.

Detection of phosphorylation sites by mass spectrometry. JAK2 was prepared as described previously (5). Briefly, Spodopterafrugiperda (Sf9) cells were infected with baculovirus containingcDNA encoding six-His-tagged JAK2. JAK2 was immunoprecipitatedusing ${alpha}$ JAK2, resolved on 5 to 12% sodium dodecyl sulfate (SDS)-polyacrylamidegels, and stained with Coomassie blue. The JAK2 from 20 immunoprecipitationswas pooled for analysis by mass spectrometry. JAK2 was subjectedto in-gel reduction and S-carboxyamidomethylation followed byin-gel digestion with trypsin (59). To detect phosphorylationat serine 523, detailed analysis of JAK2 tryptic peptides wasperformed in the laboratory of O. N. Jensen at the Universityof Southern Denmark using nanoscale sample preparation methodsin combination with matrix-assisted laser desorption ionization-timeof flight (MALDI-TOF) mass spectrometry (REFLEX IV; Bruker Daltonics,Bremen, Germany) and nanoelectrospray quadrupole time of flighttandem mass spectrometry (QTOF-1; Waters/Micromass, Manchester,United Kingdom) (62). The protein digest was loaded onto twonanocolumns (GELoader tips; Eppendorf, Hamburg, Germany) alignedin series, the first containing QIAGEN Fe(III)-nitrilotriaceticacid-immobilized metal affinity chromatography (IMAC; QIAGEN,Valencia, CA) and the second OligoR3 material (Applied Biosystems,Framingham, MA) (62). The IMAC-enriched phosphopeptide mixturewas eluted from the Fe(III)-IMAC column with 10 µl diluteammonia (pH 10.5) and immediately acidified by addition of 10µl of 5% formic acid. Phosphopeptide mixtures were desaltedusing an OligoR3 nanocolumn and eluted either directly ontothe MALDI target using the MALDI matrix for MALDI mass spectrometryanalysis (REFLEX IV; Bruker Daltonics, Bremen, Germany) or by50% methanol in 1% formic acid directly into a nanoelectrosprayneedle (Proxeon Biosystems A/S, Odense, Denmark) for nanoelectrosprayquadrupole time of flight tandem mass spectrometry (QTOF-1;Waters/Micromass, Manchester, United Kingdom) as described previously (62).

Cell culture and transfection. The stock of 3T3-F442A fibroblasts was kindly provided by H.Green (Harvard University). 293T and COS-7 cells were from theAmerican Type Culture Collection. 3T3-F442A and 293T cells weregrown in Dulbecco's modified Eagle's medium supplemented with8% calf serum, 1 mM L-glutamine, 100 U of penicillin per ml,100 µg of streptomycin per ml, and 0.25 µg of amphotericinB per ml. COS-7 cells were grown in the same medium using fetalbovine serum rather then calf serum. Both 293T and COS-7 cellswere transfected using calcium phosphate precipitation (12)and assayed 24 or 48 h after transfection, respectively. Whencells were to be incubated with a ligand, they were incubatedovernight in serum-free medium containing 1% bovine serum albuminbefore addition of ligand (500 ng GH/ml, 125 ng EGF/ml, 1 µMPMA) or vehicle for the time indicated. When the MEK inhibitorUO126 was used, cells were treated with 10 µM UO126 orvehicle for 30 min prior to the addition of ligand.

Cell lysis, immunoprecipitation, and immunoblotting. Cells were washed three times with chilled PBSV (10 mM sodiumphosphate, 137 mM NaCl, 1 mM Na₃VO₄, pH 7.4) and solubilizedin lysis buffer (50 mM Tris, 0.1% Triton X-100, 150 mM NaCl,2 mM EGTA, 1 mM Na₃VO₄, pH 7.5) containing 1 mM phenylmethylsulfonylfluoride, 10 µg/ml aprotinin, 10 µg/ml leupeptin,and 25 mM NaF. The solubilized material was centrifuged at 16,750x g at 4°C for 10 min. For immunoprecipitations, the supernatant(cell lysate) was incubated with the indicated antibody on icefor 2 h. The immune complexes were collected on 30 µlof protein A-agarose for 1 h. The beads were washed three timeswith lysis buffer and boiled for 5 min in a mixture (80:20)of lysis buffer and SDS-polyacrylamide gel electrophoresis (PAGE)sample buffer (250 mM Tris-HCl, 10% SDS, 10% ß-mercaptoethanol,40% glycerol, 0.01% bromophenol blue, pH 6.8). The solubilizedproteins were separated on SDS-polyacrylamide gels, transferredto nitrocellulose, immunoblotted with the indicated antibody,and detected using enhanced chemiluminescence or an OdysseyInfrared Imaging System (Li-Cor Biosciences). With the exceptionof Fig. 2D, results presented were representative of two ormore experiments. For experiments monitoring the effect of pp2A,20 µl of cell lysate was incubated with vehicle or 0.7units of pp2A at 37°C for 1 h. The reaction was stoppedwith the addition of SDS-PAGE sample buffer (80:20 ratio), andthe reaction mixture was boiled for 5 min. Immunoblots were quantifiedusing Bio-Rad MultiAnalyst software (for enhanced chemiluminescencedetection) or Li-Cor Odyssey 2.0 software and normalized forlevels of JAK2 or Stat5 as appropriate. As indicated, for someexperiments, the results of a representative experiment are quantifiedand normalized. In other figures, normalized means ± standarderrors of the means (or range if n equals 2) are shown for resultscarried out on different days.

JAK2 kinase assay. 293T cells were transfected with 0.5 mg cDNA for JAK2 or JAK2S523A and 1.5 µg vector using calcium phosphate precipitation.JAK2 immunoprecipitates were washed and incubated in 50 mM HEPES,0.5 mM dithiothreitol, 100 mM NaCl, 1 mM NaVO₄, 5 mM MnCl₂,pH 7.5, at 30°C, and ATP, 10 µCi [ ${gamma}$ -³²P]ATP (BP Biomedical),and a JAK2 substrate (a Stat5 peptide containing tyrosine 699 [AKAADGY₆₉₉VKPQIKQVV],prepared by the Protein Structure Facility of the Michigan BiomedicalResearch Core Facility, University of Michigan) were added toyield a final concentration of 50 µM ATP and the indicatedStat5 peptide concentration. After 20, 40, and 60 min, the vialswere centrifuged and an aliquot of the supernatant was spottedon p81 paper. The p81 paper was washed in 75 mM H₃PO₄, and bound³²P was counted using a Packard scintillation counter.

RESULTS

Top

Results

Mass spectrometry identifies serine 523 in JAK2 as a site of phosphorylation. Serine/threonine phosphorylation has been shown to regulatethe activity of multiple tyrosine kinases. To begin to determinethe role that phosphorylated serines/threonines in JAK2 playin the actions of cytokines, we set out to identify serines andthreonines that are phosphorylated in JAK2. To obtain sufficient JAK2for analysis, murine JAK2 was overexpressed in Sf9 cells. The overexpressedJAK2 was solubilized, highly purified by immunoprecipitationusing ${alpha}$ JAK2, resolved by SDS-PAGE, and then stained by Coomassiebrilliant blue. The protein was reduced and S-alkylated andthen in-gel digested using trypsin. Phosphopeptides in the trypticdigest of JAK2 were enriched by Fe(III)-IMAC and then desalted/concentratedusing a miniaturized OligoR3 column prior to analysis by massspectrometry. Peptide samples were analyzed by MALDI MS andnanoelectrospray tandem mass spectrometry. The MALDI MS tryptic peptidemass map of JAK2 displayed an ion signal consistent with phosphorylationof the peptide T₅₁₄NGISDVQISPTLQR₅₂₈. Sequencing by nanoelectrosprayquadrupole time of flight tandem mass spectrometry confirmedthe identity of the peptide and located the phosphorylationsite at position 523: T₅₁₄NGISDVQI[pS]PTLQR₅₂₈(Fig. 1).

View larger version (32K):
[in this window]
[in a new window]
FIG. 1. JAK2 is phosphorylated on serine 523. JAK2 was isolated from Sf9 cells infected with baculovirus containing the cDNA for murine JAK2. JAK2 was digested with trypsin and analyzed by nanoelectrospray quadrupole time of flight tandem mass spectrometry. The doubly charged ion corresponding to T₅₁₄NGISDVQI[pS]PTLQR₅₂₈and the Y-ion series corresponding to the C-terminal peptide ion fragments are indicated. The increment corresponding to the phosphorylated serine (pS) is indicated and corresponds to the molecular weight of a phosphorylated serine residue.

Serine 523 in JAK2 is phosphorylated in response to GH. To verify that serine 523 of JAK2 is phosphorylated in mammaliancells, we first tested whether antibody directed against phosphoserine523 ( ${alpha}$ pS523 JAK2) would recognize JAK2 overexpressed in 293Tcells. Cells transiently overexpressing JAK2 or JAK2 S523A weresolubilized, and cell lysates were immunoblotted with either ${alpha}$ JAK2 or ${alpha}$ pS523 JAK2. The ${alpha}$ JAK2 blot revealed that the expressionof JAK2 and that of JAK2 S523A were similar (Fig. 2A, bottompanel). When the lysates were blotted with ${alpha}$ pS523 JAK2, JAK2was recognized, but not JAK2 S523A, suggesting that overexpressed JAK2is phosphorylated on serine 523.

We next examined whether endogenous JAK2 was also phosphorylatedon serine 523 and whether GH, a potent activator of JAK2, regulatesthe phosphorylation of serine 523 in JAK2. 3T3-F442A cells weretreated with vehicle or with 500 ng GH/ml for 0, 5, 10, or 15min. Immunoblotting proteins in cell lysates with ${alpha}$ pS523 JAK2demonstrated a low level of phosphorylation of serine 523 inthe absence of GH. We believe that this signal is due to a lowbasal level of phosphorylation of serine 523, since it was detectedby both our antibody and the antibody of Ishida-Takahashi etal. (32a) and because it was present in ${alpha}$ JAK2 immunoprecipitatesbut not in precipitates using nonimmune serum (data not shown).GH transiently stimulated the phosphorylation of serine 523with maximal phosphorylation of serine 523 routinely detectedby 5 or 10 min (Fig. 2B, top panel, and 2C). Immunoblottingwith ${alpha}$ JAK2 demonstrated that the differences in JAK2 phosphorylationwere not the result of differences in endogenous JAK2 expression(Fig. 2B, bottom panel). Immunoblotting the lysates with ${alpha}$ pY1007/1008and ${alpha}$ pY570 (Fig. 2B, middle panels) demonstrated that GH stimulatestyrosyl phosphorylation of JAK2 within 5 min, as reported previouslyusing ${alpha}$ PY (2, 3). When the lysates from GH-stimulated 3T3-F442Acells were treated with protein phosphatase pp2A, detectionof JAK2 by ${alpha}$ pSer523 was substantially diminished (Fig. 2D and E),indicating that ${alpha}$ pS523 JAK2 detects the phosphorylated form ofserine 523.

Pretreatment of 3T3-F442A cells with a MEK1 inhibitor, UO126, abolishes GH-stimulated phosphorylation of JAK2 on serine 523. Inspection of the amino acid sequence flanking serine 523(DVQISPTLQ) suggeststhat serine 523 is phosphorylated by a proline-directed serine/threoninekinase. Because GH has been shown to rapidly activate the proline-directedserine/threonine kinases ERKs 1 and 2 (1, 9, 46, 69), we examinedwhether ERKs 1 and/or 2 might be responsible for GH stimulationof phosphorylation of serine 523. Because activated MEK1 phosphorylates andthereby activates ERKs 1 and 2 (58) and MEK1 is activated byGH (66), to block the activation of ERKs 1 and 2, we used UO126,a potent inhibitor of MEK1 (18).

3T3-F442A cells were pretreated for 30 min with either DMSOor 10 µM UO126. We then stimulated the cells with eithervehicle or 500 ng GH/ml for 5 or 10 min. Cell lysates were immunoblottedwith an anti-active ERK ( ${alpha}$ pErk) antibody that recognizes thedoubly phosphorylated (pThr²⁰², pTyr²⁰⁴) active form of ERKs1 and 2. GH-stimulated phosphorylation of ERKs 1 and 2 on threonine202 and tyrosine 204 and UO126 inhibited this stimulation (Fig. 3A,third panel from top). Immunoblotting the lysates with ${alpha}$ ERK verifiedthat comparable amounts of ERK were present in each lane (Fig. 3A,bottom panel). Immunoblotting the lysates with ${alpha}$ JAK2 verifiedthat comparable amounts of JAK2 were loaded in each lane (Fig. 3A,second panel). When the lysates were immunoblotted with ${alpha}$ pS523JAK2, GH was found to stimulate the phosphorylation of serine523 to maximal levels by 5 min (Fig. 3A, top panel, lanes 1to 3, and 3B). However, blocking activation of ERKs 1 and 2 withUO126 prior to GH stimulation inhibited GH-dependent phosphorylationof serine 523 (Fig. 3A, top panel, lanes 4 to 6, and 3B). Thisresult suggests that in response to GH, ERKs 1 and/or 2 or anunknown kinase downstream of MEK1 phosphorylates serine 523in JAK2.

PMA induces phosphorylation of JAK2 on serine 523 via a kinase that lies downstream of MEK1. Phorbol esters are potent activators of ERKs 1 and 2. Exactlyhow they activate ERKs 1 and 2 is not known, with some researchershypothesizing the involvement of PKCs and others invoking theinvolvement of the phorbol ester binding proteins RasGRPs (7, 8, 67).If GH-dependent phosphorylation of serine 523 is mediated viaERKs 1 and/or 2, we would predict that activation of ERKs 1and 2 by PMA would also stimulate the phosphorylation of serine523. When 3T3-F442A cells were treated with 1 µM PMA,phosphorylation of ERKs 1 and 2 was detected by 10 min and persistedfor 60 min (Fig. 4A, bottom panel). Blotting with ${alpha}$ pS523 JAK2(Fig. 4A, top panel, and 4B) reveals that PMA also stimulatesthe phosphorylation of serine 523. Neither PMA nor UO126 affectedthe levels of JAK2 (Fig. 4A, middle panel). Consistent withPMA stimulation of phosphorylation of serine 523 in JAK2 involvingERKs 1 and/or 2, pretreatment of the cells with the MEK1 inhibitorUO126 (10 µM for 30 min) decreased both PMA-induced activationof ERKs 1 and 2 and phosphorylation of JAK2 on serine 523 (Fig.4A, lanes 4 and 5, and 4B).

EGF stimulates the phosphorylation of JAK2 on serine 523. EGF, like PMA and GH, is known to activate ERKs 1 and 2 (52;reviewed in reference 34). We predicted, therefore, that EGF,like PMA and GH, would stimulate the phosphorylation of serine523. 3T3-F442A cells, which contain endogenous EGF receptors,were treated with vehicle or 125 ng EGF/ml for 5, 10, 15, and30 min. Cell lysates were prepared, and proteins were blottedwith ${alpha}$ pS523 JAK2. Similar to what was seen with GH, EGF stimulatedthe phosphorylation of serine 523 by 5 min (Fig. 5). Serine523 phosphorylation decreased gradually until it dropped below basallevels by 30 min. As predicted, EGF did not activate JAK2, asassessed by blotting with antibody ( ${alpha}$ pY1007/1008) to the activatingtyrosine in JAK2.

Mutation of serine 523 to alanine in JAK2 increases the activity of JAK2. Having established that serine 523 in JAK2 is phosphorylatedin response to GH as well as PMA and EGF, we began to investigatethe function of phosphorylation of serine 523. Because serine/threoninephosphorylation often has a regulatory function (reviewed inreference 42), we tested whether phosphorylation of JAK2 atserine 523 affects the kinase activity of JAK2. JAK2 or JAK2S523A was expressed in 293T cells and immunoprecipitated using ${alpha}$ JAK2. Kinase activity was then assessed using a Stat5 peptidecontaining the major JAK2 tyrosyl phosphorylation site at aminoacid 699 in Stat5. Immunoblotting with ${alpha}$ JAK2 demonstrated thatJAK2 and JAK2 S523A were expressed at similar levels (Fig. 6A).The time course for phosphorylation of the Stat5 peptide atthe highest (500 µM) Stat5 peptide concentration useddemonstrated that the assay was reasonably linear for at least60 min and that the activity of JAK2 S523A was at least twiceas high as the activity of JAK2 (Fig. 6B). When the concentrationof Stat5 peptide was varied while the ATP concentration waskept at 50 µM, the ratio of Stat5 peptide phosphorylation inthe presence of JAK2 S523A compared to that in the presenceof JAK2 was 2.7 ± 0.35 (n = 3) at 50 µM Stat5 peptide,3.9 ± 0.57 (n = 3) at 150 µM, and 2.7 ±0.34 (n = 2) at 500 µM. Thus, the activity of JAK2 S523Awas consistently higher than the activity of JAK2. Consistentwith JAK2 S523A having higher activity than JAK2, Stat5b wasmore highly phosphorylated when coexpressed with JAK2 S523Athan when coexpressed with JAK2 (Fig. 6D).

View larger version (17K):
[in this window]
[in a new window]
FIG. 6. Mutation of serine 523 to alanine in JAK2 elevates JAK2 kinase activity. (A) 293T cells were transfected with cDNA (0.5 µg) encoding JAK2 (lane 1) or JAK2 S523A (lane 2). Cell lysates (top panel) or ${alpha}$ JAK2 immunoprecipitates (bottom panel) were immunoblotted with ${alpha}$ JAK2. The migration of JAK2 is indicated. wt, wild type. (B) Immunoprecipitated JAK2 or JAK2 S523A was subjected to an in vitro kinase assay for 20, 40, and 60 min using 50 µM ATP and 500 µM of a peptide containing tyrosine 699 of Stat5 as substrate. (C) Immunoprecipitated JAK2 or JAK2 S523A was subjected to an in vitro kinase assay at Stat5 peptide substrate concentrations between 0 and 500 µM. The reaction velocity was determined by averaging the rate of ³²P incorporation into the Stat5 peptide at 40 and 60 min, n = 2. (D) COS-7 cells were transfected with cDNAs encoding GH receptor (100 ng), Stat5b (200 ng), and either JAK2 (20 ng) or JAK2 S523A (20 ng). Cells were incubated in serum-free medium overnight and resolved on a 5 to 12% gradient gel. Cell lysates were immunoblotted with ${alpha}$ pStat5b, ${alpha}$ Stat5b, or ${alpha}$ JAK2 as indicated, n = 2.

Mutation of serine 523 to alanine in JAK2 increases GH-dependent tyrosyl phosphorylation of JAK2 S523A and Stat5b. We then examined the effect of mutating serine 523 to alanineon the ability of GH to stimulate the tyrosyl phosphorylationof both JAK2 and Stat5b. COS-7 cells were transfected with cDNAencoding GH receptor, Stat5b, and either JAK2 or JAK2 S523A andtreated with vehicle or 500 ng GH/ml. JAK2 in cell lysates was immunoprecipitatedwith ${alpha}$ JAK2. Blotting with ${alpha}$ JAK2 revealed that expression levelsof JAK2 and JAK2 S523A were comparable (Fig. 7A, second panelfrom top). When cell lysates were blotted with ${alpha}$ PY, tyrosyl phosphorylationof wild-type JAK2 was observed by 5 min. In contrast, with JAK2S523A, both basal and GH-dependent phosphorylation were increased(Fig. 7A, top panel, and 7B). When cell lysates were blottedwith antibody to phosphorylated tyrosine 699 in Stat5b, phosphorylationof tyrosine 699 in Stat5b was detected by 5 min in the presenceof wild-type JAK2. In the presence of JAK2 S523A, low-levelphosphorylation of Stat5 at tyrosine 699 was detectable at thezero time point (visible in darker exposures) and GH-dependentphosphorylation of Stat5 at tyrosine 699 was enhanced comparedto the Stat5 phosphorylation seen in the presence of JAK2 (Fig.7A, third panel, and Fig. 7C). Unfortunately, the levels ofGH receptor required for us to reconstitute GH-dependent activationof exogenous JAK2 were too low for us to assess whether mutationof serine 523 in JAK2 alters phosphorylation of the GH receptor.Taken together, these data suggest that phosphorylation of JAK2on serine 523 inhibits JAK2 activity.

View larger version (36K):
[in this window]
[in a new window]
FIG. 7. Mutation of serine 523 to alanine in JAK2 enhances GH-dependent phosphorylation of JAK2 and Stat5b. (A) COS-7 cells were transfected with cDNAs encoding GH receptor (100 ng), Stat5b (200 ng), and either JAK2 (10 ng) (lanes 1 to 6) or JAK2 S523A (10 ng) (lanes 7 to 12). Cells were treated with vehicle or 500 ng GH/ml for the indicated times. Cell lysates were immunoprecipitated with ${alpha}$ JAK2 and immunoblotted with ${alpha}$ PY (top panel) or ${alpha}$ JAK2 (second panel). Cell lysates were blotted with ${alpha}$ pStat5b (third panel) or ${alpha}$ Stat5b (bottom panel), n = 2. The migration of JAK2 and Stat5b is indicated. wt, wild type. (B and C) The intensities of the bands corresponding to phosphorylated JAK2 and phosphorylated Stat5b were quantified, normalized for levels of JAK2 and Stat5, and plotted as the fraction of maximum phosphorylation detected with ${alpha}$ PY (B) and with ${alpha}$ pStat5 (C).

Pretreatment with EGF inhibits GH-dependent activation of JAK2 as measured by phosphorylation at tyrosine 1007 of JAK2. Because EGF stimulates the phosphorylation of JAK2 at serine523 and phosphorylation of serine 523 appears to inhibit JAK2activity, we examined if prior exposure of cells to EGF mightdampen the cellular response to GH. 3T3-F442A cells were treatedwith 125 ng EGF/ml for 30 min and then treated with vehicleor 50 ng GH/ml for 0, 5, 10, or 20 min. For comparison cells weretreated with 50 ng GH/ml or 125 ng EGF/ml for 0, 5, 10, 20,or 30 min. Lysates were then immunoblotted with ${alpha}$ pY1007/1008JAK2 and ${alpha}$ PY. Previous studies indicated that the tyrosyl-phosphorylated bandsthat comigrate with JAK2 are primarily if not exclusively JAK2 (3).Tyrosyl phosphorylation that comigrates with JAK2 was quantified,normalized to amounts of JAK2, and graphed. Figures 8A and B,illustrate that EGF did not stimulate overall tyrosyl phosphorylation ofJAK2 or tyrosyl phosphorylation of Y1007/1008. When cells stimulated withGH alone were compared to cells pretreated with EGF and then stimulatedwith GH, the ability of GH to stimulate phosphorylation at tyrosines1007 and 1008 as well as total tyrosyl phosphorylation of JAK2(and any proteins that comigrate with JAK2) was similar priorto 10 min. However, by 20 min, pretreatment with EGF suppressed phosphorylationat tyrosines 1007 and 1008 (Fig. 8A) as well as total tyrosylphosphorylation of JAK2 (and any comigrating proteins) (Fig.8B) by more than 50%. Thus, pretreatment with EGF appears todampen the ability of GH to maintain JAK2 activity.

DISCUSSION

Top

Discussion

Little is known about the sites of tyrosine phosphorylationwithin JAK2. Even less is known about the sites of serine/threoninephosphorylation within JAK2. We report here the first identificationof a site of serine/threonine phosphorylation in JAK2: serine523. For the initial identification, JAK2 was expressed in abaculoviral expression system and phosphorylation of serine523 in JAK2 was identified using tandem mass spectrometry. Westernblotting with an antibody specific for phosphorylated serine523 demonstrated that serine 523 is also phosphorylated whenJAK2 is overexpressed in mammalian (293T and COS-7) cells. Thissame antibody was used to demonstrate that GH stimulates thephosphorylation of serine 523 in endogenous JAK2 in 3T3-F442A cells.Phosphorylation of serine 523 in response to GH was rapid but transientwith maximal phosphorylation detected at 5 to 10 min and returningtowards basal values by 15 min. As reported in reference 32a,Ishida-Takahashi et al. have also used tandem mass spectroscopicanalysis to independently identify serine 523 in JAK2 as a siteof phosphorylation in activated JAK2 expressed in 293T cells.These findings, using two different phosphospecific antibodies,that overexpressed JAK2 is phosphorylated on serine 523 in COS-7cells, 293T cells, HEK 293 cells, and Sf9 cells and that endogenousJAK2 is phosphorylated on serine 523 in 3T3-F442A fibroblasts,32D cells, and mouse spleen (this paper and reference 32a) providestrong evidence that serine 523 in JAK2 is a bona fide siteof phosphorylation.

Analysis of the amino acids surrounding serine 523 in JAK2 (DVQISPTLQ)indicates that the kinase that phosphorylates JAK2 at serine523 is a proline-directed serine/threonine kinase. GH has beenshown to activate at least four proline-directed serine/threoninekinases, mitogen-activated protein kinases, ERKs 1 and 2 (1, 9, 46, 69),p38 (28, 78), and Jun N-terminal kinase (77), making these kinaseslogical candidates. The time course of activation in response toGH of all four of these kinases is compatible with any of thembeing the kinase that phosphorylates serine 523. When 3T3-F442Acells are treated with the MEK1 inhibitor UO126, GH-dependentphosphorylation of JAK2 at serine 523 is suppressed. This suggeststhat ERKs 1 and/or 2 (or another kinase downstream of MEK1)phosphorylates serine 523 in response to GH. Consistent withERKs 1 and/or 2 being able to phosphorylate serine 523, serine523 was also phosphorylated when 3T3-F442A cells were incubatedwith the phorbol ester PMA, a potent activator of ERKs 1 and2 (52, 67). Like GH-induced phosphorylation of serine 523, PMA-inducedphosphorylation of serine 523 was blocked by preincubation ofthe cells with the MEK1 inhibitor. Whether ERKs 1 and/or 2 oranother serine/threonine kinase is responsible for the low levelof serine 523 phosphorylation seen in serum-deprived 3T3-F442Acells (before GH addition) is not known. Because basal levelsof phosphorylation were found to vary from day to day and withtime of serum deprivation (data not shown and reference 32a),we suspect that there are other kinases capable of phosphorylatingserine 523. Our finding that PMA stimulates the phosphorylationof serine 523 raises the possibility that PKC could be sucha kinase. Although our results do not specifically exclude PKCfrom consideration, Ser523 does not lie in a PKC substrate consensussequence and PMA-stimulated phosphorylation of serine 523 isinhibited by MEK1 inhibitor to a similar extent as PMA-inducedERK 1 and 2 activity. We also think it unlikely that PKC isthe kinase in 3T3-F442A cells that phosphorylates serine 523in response to GH because we are unaware of any evidence thatGH activates PKC in 3T3-F442A cells (unpublished observation).

Because serine phosphorylation is often regulatory in nature,we examined whether phosphorylation of serine 523 might regulatethe kinase activity of JAK2. Mutating serine 523 to alanine markedlyincreased the kinase activity of JAK2, as assessed in an in vitrokinase assay using Stat5 peptide as substrate, suggesting that phosphorylationof serine 523 in vivo inhibits the overall kinase activity ofJAK2. It may therefore play a critical role in suppressing basallevels of JAK2 tyrosyl phosphorylation. Consistent with this, basalJAK2 autophosphorylation was increased when serine 523 was mutated.Phosphorylation of serine 523 may also dampen the ability of JAK2to respond to ligand activation, as suggested by our findingthat mutation of serine 523 to alanine increased GH-stimulatedtyrosyl phosphorylation of both JAK2 and Stat5b. Based uponthese findings, we hypothesize that phosphorylation of serine523 in JAK2 by ERKs 1 and/or 2 acts in a negative feedback mannerto dampen the initial activation of JAK2 in response to GH.Phosphorylation of serine 523 would also provide a mechanismby which prior exposure to other ERK-activating ligands mightdampen the cellular response to GH. Ishida-Takahashi et al.(32a) have also found that mutation of serine 523 to alanineresults in constitutive JAK2-dependent signaling in the absenceof cytokine stimulation. In support of the hypothesis that phosphorylationof serine 523 acts in a negative feedback manner to dampen activationof JAK2 and downstream signaling pathways in response to cytokinebinding, they observed that mutating serine 523 to alanine enhancedand prolonged JAK2 activation in response to EPO binding toa hybrid EPO-leptin receptor and increased EPO-dependent stimulationof expression of a Stat3 reporter.

A negative regulatory function for phosphorylation of serine523 would be consistent with what is known thus far about the roleof serine phosphorylation in the regulation of JAK2 and the receptortyrosine kinases. PKC- ${delta}$ -dependent phosphorylation of JAK2 onserine and threonine has been reported to abrogate interleukin-3-dependenttyrosyl phosphorylation and activation of JAK2, with inhibitionof JAK2 either by PKC- ${delta}$ -dependent serine/threonine phosphorylationor by interleukin-3 deprivation, leading to differentiationof 32D myeloid progenitor cells (37). Serine/threonine phosphorylationof insulin receptor has been reported to reduce insulin-stimulatedtyrosine phosphorylation of the receptor and its targets andinhibit propagation of the insulin receptor signal (48, 63).Protein kinase C-dependent phosphorylation of EGF receptor isreported to decrease autophosphorylation of EGF receptor andreduce tyrosine kinase activity (14, 16). Phosphorylation of theplatelet-derived growth factor receptor ß on serine1104 by G protein-coupled receptor kinase 2 (GRK2) is thoughtto reduce signaling pathways downstream of platelet-derivedgrowth factor receptor at least in part by reducing receptorautophosphorylation and reducing receptor interaction with the Na⁺/H⁺exchanger regulatory factor (NHERF, also known as EBP50) (27).

How phosphorylation of serine 523 inhibits the kinase activityof JAK2 is not known. JAK2 is composed of seven JAK homologydomains (JH1 to JH7) (24, 72). The C-terminal JH1 domain isthe kinase domain (20, 76). JH2, the pseudokinase domain, ishypothesized to be autoinhibitory (21, 54-56). The entire JH3domain combined with the second half of the JH4 domain of JAK2has structural homology to an SH2 domain but does not appearto function as an SH2 domain (26). The first half of the JH4domain of JAK2 and the JH5 to JH7 domains comprise the FERM (band4.1, ezrin, radixin, and moesin) domain (23), responsible for bindingto the GH receptor (22, 64), erythropoietin receptor (15, 30),gamma interferon receptor (15, 36), and granulocyte-macrophagecolony-stimulating factor ßc subunit (75). Serine523 lies in the linker connecting the pseudokinase domain (JH2)to the JH3 domain. Thus, one can envision phosphorylation ofserine 523 inducing a conformational change in JAK2 that wouldstrengthen or promote the autoinhibitory action of the JH2 domainon the kinase domain of JAK2. Mutation of serine 523 to alaninewould be expected to release phosphorylation-dependent inhibition,resulting in increased kinase activity and tyrosine phosphorylationof JAK2 and its substrates as demonstrated in this report. Insupport of this general region of JAK2 playing a critical rolein regulating the enzymatic activity of JAK2, autophosphorylationof tyrosine 570 in JAK2 similarly appears to inhibit JAK2 activity(5, 19). It is also possible that phosphorylation of serine523 and/or tyrosine 570 recruits a negative regulatory protein,such as a phosphatase, to JAK2 or prevents the binding of apositive regulatory protein. These possibilities are currentlyunder investigation.

The finding that serine 523 appears to be phosphorylated byERKs 1 and/or 2 suggests that phosphorylation of serine 523may be a site of cross talk between ligands such as GH thatactivate JAK2 and ligands that activate MEK1 and ERKs 1 and/or2 by a JAK2-independent pathway. Our finding here that EGF stimulatesthe phosphorylation of JAK2 on serine 523 and decreases theability of GH to stimulate phosphorylation of the activatingtyrosine, tyrosine 1007, in JAK2 as well as the overall tyrosylphosphorylation of JAK2 provides support for this hypothesis. Interestingly,the cross talk between GH receptors and EGF receptor familymembers extends to both receptors, even in regards to the role ofERK activation. Thus, GH is thought to stimulate the phosphorylation ofErbB-2 by ERKs 1 and/or 2 and cause a decrease in EGF-induced tyrosylphosphorylation of ErbB-2 and DNA synthesis (35). The crosstalk between EGF family receptors and GH receptors appears multifaceted.GH has also been reported to protect EGF receptor from degradationand thereby potentiate EGF signaling and to decrease EGF bindingto its receptor by decreasing the affinity of EGF for its receptorin a MEK1/ERK pathway-dependent manner (31, 32). GH has alsobeen reported to stimulate phosphorylation of the EGF receptoron tyrosines (35, 73). It seems likely that EGF regulation ofGH signaling may also be multifaceted, with phosphorylationof serine 523 being just one aspect of that regulation. Similarly,we would anticipate that other ligands that activate receptortyrosine kinases and ERKs 1 and/or 2, such as platelet-derived growthfactor, as well as ligands that activate ERKs 1 and/or 2 by otherprocesses, such as via G_i-coupled receptors, would also stimulatephosphorylation of serine 523 in JAK2 and thereby downregulateGH signaling. The degree to which this would affect GH signalingwould depend upon the timing of the stimulation via the variousligands and whether these ligands have other effects on GH signaling.

In summary, we have identified serine 523 in JAK2 as a siteof phosphorylation. Using an antibody that specifically recognizes JAK2phosphorylated at serine 523, we demonstrate that serine 523is rapidly but transiently phosphorylated in response to GHand EGF and provide evidence that serine 523 is phosphorylatedby ERKs 1 and/or 2. We also report that mutating serine 523to an alanine increases the enzymatic activity of JAK2 as wellas basal and GH-stimulated tyrosyl phosphorylation of JAK2 andthe JAK2 substrate Stat5b, suggesting that phosphorylation ofserine 523 inhibits JAK2 activity. We hypothesize that phosphorylationof serine 523 in JAK2 by ERKs 1 and/or 2 or another kinase downstreamof MEK1 acts in a negative feedback manner to dampen the initialactivation of JAK2 in response to GH. Phosphorylation of serine523 also provides a mechanism by which exposure to other ERK-activatingligands or environmental factors that regulate serine 523 phosphorylationmight modulate the cell's response to GH.

ACKNOWLEDGMENTS

We thank Baljeet Deo and Xiaqing Wang for technical assistanceand Barbara Hawkins for assistance with the manuscript.

This work was supported by NIH grant DK34171 (to C.C.-S.) anda grant from the Danish Natural Sciences Research Council (toO.N.J.). cDNA sequencing was supported by the Cellular and MolecularBiology Core and peptide synthesis by the Protein StructureFacility of the Michigan Biomedical Research Core of the MichiganDiabetes Research and Training Center (P60 DK20572), Universityof Michigan Rheumatologic Diseases Comprehensive Center (P60-AR20557),and the University of Michigan Comprehensive Cancer Center (NIHP30 CA46592). A.M.M.-M. was supported by a Predoctoral Fellowshipfrom the Cellular and Molecular Biology Training Grant T32-GM 07315.

FOOTNOTES

* Corresponding author. Mailing address: Department of Molecular and Integrative Physiology, The University of Michigan Medical School, Ann Arbor, MI 48109-0622. Phone: (734) 763-2561. Fax: (734) 647-9523. E-mail: cartersu{at}umich.edu.

These two authors contributed equally to this work.

REFERENCES

Top