Transduction Lymphopoietin-Mediated Signal Activation in Thymic ...

3 downloads 5427 Views 214KB Size Report
Receive free email-alerts when new articles cite this article. Sign up at: ..... CSF led to activation of the Stat5-responsive promoter and CAT expression.
Uncoupling of Proliferation and Stat5 Activation in Thymic Stromal Lymphopoietin-Mediated Signal Transduction1 Deborah E. Isaksen,2* Heinz Baumann,† Baohua Zhou,* Sebastien Nivollet,* Andrew G. Farr,‡§ Steven D. Levin,‡ and Steven F. Ziegler3*‡ Thymic stromal lymphopoietin (TSLP) is a cytokine that facilitates B lymphocyte differentiation and costimulates T cells. Previous studies have demonstrated that a functional TSLP receptor complex is a heterodimer consisting of the TSLP receptor and the IL-7R ␣-chain. TSLP-mediated signaling is unique among members of the cytokine receptor family in that activation of the transcription factor Stat5 occurs without detectable Janus kinase activation. Using a variety of biological systems we demonstrate here that TSLP-mediated Stat5 activation can be uncoupled from proliferation. We also show that the single tyrosine residue in the cytoplasmic domain of the TSLP receptor is critical for TSLP-mediated proliferation, but is dispensable for Stat5 activation. Our data demonstrate that TSLP-mediated Stat5 activation is insufficient for cell proliferation and identifies residues within the TSLP receptor complex required to mediate these downstream events. The Journal of Immunology, 2002, 168: 3288 –3294.

C

ytokines are secreted proteins that play critical roles in hemopoiesis and immune responses by mediating cell proliferation, differentiation, and survival. Cytokines stimulate target cells via cell surface receptors, leading to the activation of signaling cascades within the target cell. A functional cytokine receptor is often composed of several subunits that form homodimers, heterodimers, or higher order oligomers (1). In addition, specific receptor subunits are often found in the receptor complex of more than one cytokine. For example, the cytokines IL-7 and thymic stromal lymphopoietin (TSLP)4 share the IL-7R␣ subunit. The IL-7R is composed of IL-7R␣ and the common ␥-chain (␥c), whereas a functional TSLP receptor is composed of IL-7R␣ and the recently identified TSLPR subunit (2– 4). This sharing of receptor subunits may explain the similar biological effects of these two cytokines: both support B lymphopoiesis in in vitro culture systems; both sustain the factor-dependent, fetal liverderived NAG8/7 pre-B cell line; and both costimulate thymocytes and mature T cells (5–7, 36). A prevalent feature of type I cytokine-mediated signal transduction includes activation of the Janus kinase (Jak)-Stat signaling cascade (1). For example, engagement of the IL-7R induces activation of Jak1 and Jak3, with subsequent activation of the transcription factors Stat5a and Stat5b (8, 36). TSLP also activates *Virginia Mason Research Center, Seattle, WA 98101; †Department of Molecular and Cellular Biology, Roswell Park Cancer Institute, Buffalo, NY 14263; and Departments of ‡Immunology and §Biological Structure, University of Washington, Seattle, WA 98195 Received for publication November 6, 2001. Accepted for publication January 30, 2002. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by National Institutes of Health Grants AI44259 (to S.F.Z.), CA85580 (to H.B.), and AI44160 (to A.G.F.). D.E.I. was supported by National Research Service Award AI10207. 2

Current address: Deltagen, Redwood City, CA 94063.

3

Address correspondence and reprint requests to Dr. Steven F. Ziegler, Virginia Mason Research Center, 1201 9th Avenue, Seattle, WA 98101. E-mail address: [email protected] Abbreviations used in this paper: TSLP, thymic stromal lymphopoietin; ␥c, common ␥-chain; CAT, chloramphenicol acetyl transferase; hu, human; Jak, Janus kinase; MUP, mouse major urinary protein; wt, wild type. 4

Copyright © 2002 by The American Association of Immunologists

Stat5a and Stat5b, but, unlike IL-7, TSLP fails to induce the tyrosine phosphorylation of any of the four known Jaks (4, 36). Moreover, overexpression of kinase-negative versions of Jak1, Jak2, or Jak3 does not inhibit TSLP-mediated Stat5 activation (Ref. 4 and P. A. Trobridge and S. D. Levin, unpublished observations). Together these data suggest a unique mechanism of Stat5 activation in response to TSLP. In this study we describe three instances in which TSLP-mediated Stat5 activation is insufficient to induce cellular proliferation. We also present the results of a mutational analysis that was undertaken to identify the regions of the TSLPR and IL-7R␣ cytoplasmic domains responsible for TSLP-mediated Stat5 activation and cell proliferation. We show that mutation of the single cytoplasmic tyrosine residue in TSLPR does not affect receptor-mediated Stat5 activation, but ablates receptor-mediated proliferation and cis gene induction. In addition, we find that, unlike IL-7-mediated signaling, TSLP-mediated Stat5 activation and proliferation require the conserved Box1 domains of both TSLPR and IL-7R␣. Collectively, our data demonstrate that TSLP-mediated Stat5 activation can be separated from proliferative responses and identify the residues within the TSLP receptor complex required to mediate these downstream events.

Materials and Methods Cell culture HepG2 cells (9) and Phoenix amphotropic retroviral producer cells (10) were maintained in DMEM supplemented with 10% FCS and antibiotics. BafM7R, NAG8/7, and IxN/2B (11) cells were maintained in RPMI 1640 supplemented with 10% FCS, antibiotics, 1 mM sodium pyruvate, 20 ␮M 2-ME, and, unless otherwise specified, 1 ng/ml murine IL-7 or 20 ng/ml murine TSLP. IL-7 and TSLP were obtained as previously described (4). Recombinant human (hu)GM-CSF was purchased from Immunex (Seattle, WA).

RT-PCR Total RNA from IxN/2B cells was isolated using TRIzol reagent (Life Technologies, Grand Island, NY). First-strand cDNA was made using random primers and a cDNA synthesis kit (Stratagene, La Jolla, CA). Overlapping fragments of TSLPR and IL-7R␣ were PCR-amplified and sequenced using gene-specific primers. 0022-1767/02/$02.00

The Journal of Immunology Flow cytometry Abs used for flow cytometry and purchased from Caltag Laboratories (Burlingame, CA) included anti-huCD2-PE, anti-huCD5-FITC, anti-mouse IgG1-PE, anti-mouse IgG2b-TriColor, and streptavidin-PE. Mouse IgG1 anti-huGM-CSFR␣ and mouse IgG2b anti-huGM-CSFR␤ were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). A biotinylated antiTSLPR Ab (22H9) was obtained from A. G. Farr (manuscript in preparation). Flow cytometry was performed on 5 ⫻ 105 cells using a FACSort flow cytometer and CellQuest software (BD Biosciences, Mountain View, CA).

Proliferation assay Cells were plated in triplicate into 96-well flat-bottom plates (1.5 ⫻ 104 BafM7R cells/well or 2 ⫻ 104 NAG8/7 cells/well) and cultured for 62 h with the indicated quantity of cytokine and with or without PP1 (obtained from Dr. J. Hanke, Pfizer Central Research, Groten, CT). PP1 was used over a range of concentrations as indicated. Proliferation was measured by [3H]thymidine incorporation following addition of [3H]thymidine to the culture medium for the final 15 h. At the time BafM7R cells were seeded into the 96-well plate, they were also seeded into a flask and grown in the presence of TSLP. These parallel cultures were analyzed for surface expression of GMR␣ and -␤ by flow cytometry on the same day 3H incorporation was measured. BafM7R data are expressed as the mean cpm from triplicate wells minus the mean cpm obtained from wells with medium alone. GM-CSF values were normalized to account for the percentage of cells that did not express both chimeric receptor chains and thus would be able to respond to 20 ng/ml TSLP but not 100 ng/ml GM-CSF.

Immunoprecipitations and Western blots Cells were washed four times with PBS and cytokine-starved for 5–7 h at 37°C. BafM7R cells were also serum-starved. When appropriate, PP1 was added 30 min before stimulation at 20 ␮M. Before proceeding further with BafM7R cells, flow cytometry was used to confirm that ⬎90% of the cells expressed both huGMR␣ and ␤ chimeric receptors. Cells at 1 ⫻ 107/ml were stimulated at 37°C for 20 min with or without GM-CSF (100 ng/ml), TSLP (100 ng/ml), or IL-7 (25 ng/ml). To stop the stimulation, cold PBS plus 1 mM Na3VO4 were added. Cell lysates were generated using TNT buffer (25 mM Tris-HCl (pH 8), 150 mM NaCl, and 1% Triton X-100) plus protease inhibitors as previously described (36). Clarified lysates at 1 ⫻ 108 cells/ml were either loaded onto a 6 or 10% SDS-PAGE gel following the addition of SDS loading buffer or were subjected to immunoprecipitation with an anti-Stat5 antisera as previously described (36). Processing of the Western blots with an anti-phosphotyrosine Ab or anti-Stat5 antisera was also as previously described (36). Blots were developed using chemiluminescent detection and were stripped for reprobing using the Western Blot Recycling kit (Chemicon, Temecula, CA). Chemiluminescent detection using the phospho-Stat5 Ab (SigmaAldrich, St. Louis, MO) on BafM7R whole cell lysates required the SuperSignal West Femto Maximum Sensitivity Substrate (Pierce, Rockford, IL).

Construction of chimeric receptors The huGM-CSFR␤ extracellular domain (with the GM-CSFR␣ signal sequence for improved expression) and the huGM-CSFR␣ extracellular domain (12) were subcloned as NotI-BamHI fragments into pcDNA3.1 Neo⫺ (Invitrogen, Carlsbad, CA). The BamHI site was introduced at the 3⬘ end of the GMR␤ extracellular domain by primer-mediated site-directed mutagenesis using PFU polymerase (Stratagene). This resulted in a single conservative amino acid substitution of a glutamic acid to an aspartic acid at the membrane-proximal end of the GMR␤ extracellular domain. In a similar manner, either a BamHI or an EcoRI site was introduced at the 3⬘ end of the GMR␣ extracellular domain, which resulted in an aspartic acid to glutamic acid substitution one or two residues, respectively, upstream of the predicted start of the transmembrane domain. GM␤/IL-7R␣ chimeric receptors were generated by inserting in-frame the IL-7R␣ transmembrane domain plus cytoplasmic domain (mutated or wild type) as a BamHI-EcoRI cassette into the previously generated pcDNA3.1 Neo⫺ GMR␤ expression vector. The BamHI site upstream of the IL-7R␣ transmembrane domain is naturally occurring. Site-specific mutations of the IL-7R␣ cytoplasmic domain were performed by primermediated site-directed mutagenesis using PFU polymerase. The GMR␣/ TSLPR chimeric receptors were generated in a similar manner. Each TSLPR transmembrane domain plus cytoplasmic domain construct (mutated or wild type) was PCR-amplified and then inserted in-frame as a BamHI-EcoRI or EcoRI-NotI cassette into the previously generated

3289 pcDNA3.1 Neo⫺ GMR␣ vector. The sequence of each construct was confirmed by DNA sequence analysis using the Prism Big-Dye terminator cycle sequencing kit (PE Applied Biosystems, Norwalk, CT).

HepG2 transfection and CAT analysis HepG2 cells were transfected by the calcium phosphate method using 20 ␮g/ml total plasmid DNA and 2–3 ⫻ 105 cells (13). Cells were cotransfected with expression vectors for the internal transfection control mouse major urinary protein (MUP), rat Stat5b (14), GMR␣/TSLPR, and GMR␤/ IL-7R␣ constructs as described above, and p(8xHRRE)-chloramphenicol acetyl transferase (CAT), which contains eight tandem copies of the 27-bp hemopoietin receptor response element in pCAT (15). After an overnight recovery period, cultures were released from the plate with trypsin and divided into six-well culture plates. After an additional 24 h, subcultures were treated for 24 h with serum-free medium containing 100 ng/ml huGM-CSF or were left unstimulated. Medium was then collected and subjected to immunoelectrophoresis to quantitate expression of the cotransfected control MUP plasmid. CAT activity for each culture was determined and normalized to the amount of MUP expression. To compare results from separate experiments, the specific CAT activities within each experimental series were calculated relative to the CAT activity of untreated subcultures (defined as 1.0). Data are presented as the mean of three or more trials.

Retroviral constructs, production, and infection The pMI.2 and pMI.5 retroviral vectors contain the internal ribosome entry site from encephalomyocarditis virus upstream of a cDNA encoding the extracellular and transmembrane domains of huCD2 or huCD5, respectively (16). Subcloning of the chimeric GMR␤/IL-7R␣ or GMR␣/TSLPR receptors as NotI/EcoRI or EcoRI/SalI fragments upstream of the internal ribosome entry site results in transcription of a bicistronic mRNA that directs translation of the chimeric receptor and the truncated huCD2 or huCD5. Wild-type versions of the chimeric receptors were subcloned into pMI.2, while the mutated versions were subcloned into pMI.5. Retroviral vectors were packaged in Phoenix amphotropic cells (10). Twenty-four hours before transfection, 1.75 ⫻ 106 cells were plated in a 60-mm tissue culture dish. The calcium phosphate mammalian cell transfection kit (5 Prime33 Prime, Boulder, CO) was used to transfect 12 ␮g DNA/dish. Ten hours post-transfection the culture medium was replaced with 4 ml fresh medium. Retroviral supernatants were collected 48 h posttransfection and passed through a 0.45-␮m pore size filter syringe. Retroviral supernatants were either used immediately or were stored at ⫺70°C. The generation of BafM7R cells expressing both the GMR␣/TSLPR and GMR␤/IL-7R receptor chains was a two-step process. First, cells that expressed GMR␣/TSLPR wild type (wt) and huCD2 were generated. After enrichment for GMR␣ or huCD2 expression (see below), these cells were used as the recipients for the second round of infection with retrovirus encoding the GMR␤/IL-7R wild-type or mutated receptor. For the infection, 1.5 ⫻ 106 cells (in 100 ␮l) were placed in a single well of a 24-well plate, and 1 ml of the filtered, retroviral supernatant and 8 ␮g/ml polybrene (hexadimethrine bromide; Sigma-Aldrich) were added. The plate was centrifuged (1100 ⫻ g) for 90 min at room temperature. After completion of the spin, 1 ml culture medium, IL-7, and additional polybrene (8 ␮g/ml final) were gently added. Cells were cultured at 32°C overnight. Twentyfour hours postinfection cells were pelleted, washed, and cultured at 37°C in Ba/F3 culture medium plus IL-7 or TSLP.

Magnetic bead selection Two days after the retroviral infection and periodically while the cell lines were being maintained, cells were positively selected for expression of the chimeric receptors or the linked huCD2 or huCD5 markers using mouse IgG1 or pan-mouse Dynabeads as directed by the manufacturer (Dynal Biotech, Oslo, Norway). Abs used for the selection included murine mAbs to huGM-CSFR␣ and -␤ (Santa Cruz Biotechnology) and huCD2 and -5 (Caltag Laboratories).

Analysis of cis expression Total cellular RNA was isolated using TRIzol reagent (Life Technologies) according to the manufacturer’s instructions. Fifteen micrograms of total RNA was separated on a 1% agarose/2.2 M formaldehyde gel, transferred to Transfer-IT membranes (CPG, Lincoln Park, NJ), and cross-linked by UV irradiation. Blots were hybridized as described previously (4), except that washing steps were conducted at 65°C.

3290

Results IxN/2B cells activate Stat5 but do not proliferate in response to TSLP IxN/2B is an IL-7-dependent, bone marrow-derived, pre-B cell line that binds TSLP with high affinity (2). However, unlike IL-7, TSLP failed to support the proliferation of these cells (Fig. 1a). This result was surprising in light of the finding that IxN/2B cells expressed the two chains of the TSLP receptor, TSLPR and IL7R␣, on their cell surface at levels comparable to the NAG8/7 cell line, which does proliferate in response to TSLP (Fig. 1b and data not shown; Ref. 36). Moreover, analysis of cDNAs encoding TSLPR and IL-7R␣ from these cells revealed no sequence abnormalities (data not shown). To determine whether TSLPR signal transduction was intact in IxN/2B cells, we assayed for Stat5 activation following TSLP stimulation. As shown in Fig. 1c, treatment of IxN/2B cells with either TSLP or IL-7 resulted in tyrosine phosphorylation of Stat5, and the levels of phosphorylation were comparable to those seen in the TSLP- and IL-7-responsive NAG8/7 cells. Thus, despite normal Stat5 activation, TSLP failed to drive the proliferation of IxN/2B cells. These data suggest that Stat5 activation is insufficient for TSLP-mediated proliferation, a finding that is novel within the cytokine receptor family. Inhibition of Src family kinases blocks TSLP-mediated proliferation but not Stat5 activation Engagement of the TSLP receptor results in Stat5 activation in the absence of Jak activation (36). To examine the possibility that a Src family kinase may be involved in TSLP-mediated Stat5 phosphorylation, we used the Src family inhibitor PP1 (17). NAG8/7

FIGURE 1. IxN/2B cells activate Stat5 but do not proliferate in response to TSLP. a, IxN/2B cells proliferated in response to IL-7 but not TSLP. b, Cell surface expression of endogenous TSLPR on TSLP- and IL-7-responsive NAG 8/7 cells and on IL-7-responsive IxN/2B cells. Cells were stained with a biotinylated Ab to TSLPR and PE-conjugated streptavidin and were analyzed for fluorescence intensity by flow cytometry. Control cells are IxN/2B cells stained with a biotinylated, isotype-matched, control Ab. c, Cells were starved of cytokine for 4 h, then stimulated with either IL-7 or TSLP. Cell lysates were immunoprecipitated using anti-Stat5 antisera. Shown is an anti-phosphotyrosine blot of the immunoprecipitated material. Reprobing with anti-Stat5 antisera showed equal loading in each lane (data not shown).

UNCOUPLING OF STAT5 ACTIVATION AND PROLIFERATION cells were treated with PP1 and assayed for Stat5 phosphorylation and cell proliferation. As shown in Fig. 2, both TSLP- and IL-7mediated proliferation were abrogated by PP1, while Stat5 phosphorylation was unaffected. Moreover, this proliferative inhibition is unlikely to reflect nonspecific toxicity, because PP1 treatment had no effect on IL-2-mediated proliferation of HT-2 cells and did not augment the rate of cell death in NAG8/7 cells beyond that observed with simple cytokine deprivation (data not shown). These data provide additional evidence that proliferation and Stat5 activation can be uncoupled in TSLP-treated cells and implicate Src family tyrosine kinases in the transduction of TSLP- and IL-7mediated proliferative signals. Stat5 activation and cell proliferation require the Box1 domain of both TSLPR and IL-7R␣ To uncover sequence domains responsible for TSLP-mediated Stat5 activation and/or proliferation, we began a mutational analysis of TSLPR and IL-7R␣. Mutational analysis of the receptors for erythropoietin, IL-4, IL-5, and IL-6 has identified a membraneproximal domain, rich in hydrophobic amino acids and proline residues, that has been termed the Box1 domain, and this domain is critical for Jak and Stat activation and the induction of proliferation (18 –23). In fact, mutations to the Box1 domain of the erythropoietin or gp130 receptors prevent the physical association of Jak2 with the receptor (19, 24, 25). As described above, TSLP stimulation activates Stat5 without detectable Jak activation (4, 36). In an effort to determine whether the Box 1 domains of TSLPR and IL-7R␣ are involved in TSLPmediated Stat5 activation and proliferation, point mutations were introduced into the Box1 region of both receptor subunits. The

FIGURE 2. Src family kinase inhibitor blocks TSLP-mediated proliferation but not Stat5 activation. a, PP1 inhibits the proliferation of NAG8/7 cells in response to both IL-7 and TSLP. Incorporation of [3H]thymidine was measured in NAG8/7 cells cultured in the presence of either IL-7 or TSLP with the indicated concentration of PP1. Results shown are representative of four different experiments. b, PP1 has no effect on Stat5 activation mediated by IL-7 or TSLP. NAG8/7 cells were stimulated in the presence or the absence of PP1, and immunoprecipitated Stat5 was blotted with an anti-phosphotyrosine Ab. The arrow indicates the position of the tyrosine-phosphorylated Stat5. Reprobing of the same blot with the antiStat5 antisera revealed that all lanes had equivalent amounts of Stat5 (data not shown). The results shown are representative of three separate experiments.

The Journal of Immunology mutated receptor chains were then tested for their ability to activate Stat5. For these experiments we used a receptor reconstitution system that has been previously used to study Stat activation by a number of cytokine receptors, including the TSLP receptor (4). Briefly, this system involves cotransfection of the HepG2 human hepatoma cell line with cDNAs encoding Stat5b, cytokine receptor subunits, and a Stat5-responsive CAT reporter gene (15). The relative ability of the transfected cells to activate Stat5 upon treatment with the appropriate cytokine is measured as CAT activity. Previous studies using this system showed that TSLP-induced CAT activity required the addition of Stat5a or Stat5b and that the endogenous expression of Stat1 and Stat3 by the HepG2 cells was insufficient for TSLP signaling (4). Chimeric receptors consisting of the huGM-CSFR␣ extracellular domain fused in-frame to the murine TSLPR transmembrane and cytoplasmic domain (herein referred to as GMR␣/TSLPR) and the huGM-CSFR␤ extracellular domain fused in-frame to the murine IL-7R␣ transmembrane and cytoplasmic domain (GMR␤/IL7R) were used in these experiments. The use of chimeric receptors allowed us to extend the mutagenesis studies to cell lines that naturally express the TSLP receptor subunits. Treatment of HepG2 cells expressing GMR␣/TSLPR and GMR␤/IL-7R with huGMCSF led to activation of the Stat5-responsive promoter and CAT expression. CAT activity was dependent upon the addition of both GMR␣- and GMR␤-containing receptor chains, GM-CSF, and cDNA encoding Stat5b (Fig. 3 and data not shown) (4). Similar results were obtained using native receptor subunits (data not shown). When the first two conserved proline residues of the Box1 domain of either TSLPR or IL-7R␣ were changed to serine residues, cytokine-induced Stat5 activation was abolished or severely inhibited (Fig. 3). Although disruption of the Box1 domain of either receptor subunit blocked TSLP-mediated Stat5 activation, IL-7mediated Stat5 activation occurred despite disruption of the Box1 domain of IL-7R␣. That is, Stat5 was activated when GMR␤/IL7Rbox1 was paired with a wild-type GMR␣␥c (Fig. 3b). This result highlights yet another difference between the IL-7 and TSLP signaling pathways: TSLP-mediated Stat5 activation requires an intact Box1 domain in both receptor chains, while IL-7-mediated Stat5 activation occurs without an intact IL-7R␣ Box1 domain. Sequence conservation among type 1 cytokine receptor chains, including TSLPR, is also maintained carboxyl-terminal to the Box1 domain (Fig. 3a). Indeed, a highly conserved tryptophan (W) residue within this region in the erythropoietin and gp130 receptors is crucial for Jak-Stat activation (18, 26). Using the CAT assay to test the requirement for the conserved tryptophan of TSLPR (mutation W34R), we found that this residue was required for TSLP-mediated Stat5 activation (Fig. 3b). Next, the Box1 and W34R mutations were tested for their ability to support cytokine-dependent proliferation in the murine, IL-3responsive, pro-B cell line Ba/F3. For these experiments we used a Ba/F3 subline (BafM7R) that had been stably transfected with the IL-7R ␣-chain. This cell line naturally expresses TSLPR and ␥c and, consequently, BafM7R cells respond to TSLP, IL-7, or IL-3 (2). Retrovirus-mediated infection was used to generate BafM7R cells that expressed the wild-type or mutant chimeric receptor chains GMR␣/TSLPR and GMR␤/IL-7R. The use of such a system allowed comparison between huGM-CSF-dependent proliferation (mediated by the mutated chimeric receptors) and TSLPdependent proliferation (mediated by the endogenous TSLPR and IL-7R␣ subunits). In addition to the failure to activate Stat5 (Fig. 3), mutations to the Box1 domain of TSLPR or IL-7R␣ prevented proliferation of the BafM7R cells (Table I). However, when the

3291

FIGURE 3. Conserved residues of the Box1 domain of TSLPR and IL7R␣ are required for Stat5 activation. a, Amino acid alignment of the membrane-proximal region of the cytoplasmic domain of murine TSLP-R to that of closely related cytokine receptor subunits. All sequences begin with the first predicted residue of the cytoplasmic domain and are murine, except for human TSLPR, which is from humans (D. Rawlings and S. F. Ziegler, manuscript in preparation). The Box1 domain is boxed. Residues that are conserved in at least four of the seven sequences are shaded. The two underlined proline residues in TSLPR or IL-7R␣ were changed to serine residues in the Box1 mutations described in the text. The underlined tryptophan residue in TSLPR was changed to an arginine residue in the W34R mutation. b, HepG2 cells were transiently cotransfected with a Statresponsive CAT reporter plasmid, Stat5b cDNA, and cDNA encoding wild-type or mutated chimeric receptor chains as indicated. Subcultures were treated with medium alone or GM-CSF and analyzed for CAT activity. For each trial, the CAT activity detected using wild-type receptor pairs upon GM-CSF stimulation was set at 100% and was used to calculate the relative activity detected from cultures expressing the mutated receptors. Data are presented as the mean ⫾ SD of three to six trials.

endogenous TSLP receptor complex was engaged, the cells did proliferate, indicating that the TSLP signaling pathway was intact. Single cytoplasmic tyrosine of TSLPR is not required but is sufficient for Stat5 activation For several cytokine receptors, the recruitment of Stat5 is dependent upon phosphorylation of tyrosine residues within the cytoplasmic domain of the receptors (2). The TSLPR subunit contains only one cytoplasmic tyrosine residue (Y103) that resides three amino acids away from the carboxyl terminus. To test the requirement of this residue for TSLP-mediated Stat5 activation, Y103 was substituted with a phenylalanine (mutation Y103F). As shown in Table II, normal levels of Stat5-mediated CAT activity were detected using the GMR␣/Y103F and GMR␤/IL-7Rwt receptor pair. In addition, experiments combining a carboxyl-terminally truncated GMR␤/IL-7R construct (GMR␤/IL-7R125cyt) that lacked all four cytoplasmic tyrosine residues with GMR␣/TSLPRwt activated Stat5 (Table II). However, when GMR␤/IL-7R(125cyt) was paired with GMR␣/Y103F, a receptor combination that has no

3292

UNCOUPLING OF STAT5 ACTIVATION AND PROLIFERATION

Table I. An intact Box1 region in TSLPR or IL-7R␣ is required for cell proliferationa [3H]Thymidine Incorporation (cpm)

Expression Constructs GMR␣/TSLPR

wt Box1 W34R wt

GMR␤/IL-7R

GM-CSF

TSLP

wt wt wt wt Box1

0 93,585 0 0 124

94,717 47,993 64,661 64,941 27,621

a Data from a representative experimental series (repeated at least three times) are presented as the mean cpm from triplicate wells minus those obtained from wells without added cytokine. GM-CSF values are normalized to account for the percentage of cells that did not express both receptor chains, by flow cytometric analysis, and thus would be able to respond to TSLP but not GM-CSF.

cytoplasmic tyrosine residues, Stat5 activation was blocked (Table II). Thus, the four cytoplasmic tyrosines of IL-7R␣ were not required when the cytoplasmic tyrosine of TSLPR was present and vice versa. However, a receptor complex lacking all cytoplasmic tyrosine residues on both chains was unable to activate Stat5. Cytoplasmic tyrosine residue of TSLPR is required for proliferation To determine whether Y103 was required for cell proliferation, we used retroviral transduction to generate BafM7R cells expressing GMR␣/Y103F and GMR␤/IL-7Rwt (yf/wt cells). These cells were unable to proliferate in response to GM-CSF (Fig. 4a), even though we demonstrated that this receptor combination was capable of mediating Stat5 activation in HepG2 cells (Table II). To confirm that Stat5 activation was occurring in yf/wt-expressing BafM7R cells, lysates from these cells were tested for the presence of activated Stat5 following GM-CSF stimulation. As shown in Fig. 4b, tyrosine-phosphorylated Stat5 was detected in yf/wt cells following GM-CSF treatment at levels comparable to those seen following TSLP stimulation. The requirement of Y103 for proliferation, but not Stat5 activation, correlates with results obtained using the IxN/2B cell line (Fig. 1) and NAG8/7 cells treated with PP1 (Fig. 2). To further investigate Stat5 activation in yf/wt cells we examined induction of the cis gene by GM-CSF and TSLP. Previous studies have shown that transcription of the cis gene is rapidly induced following TSLPR engagement (4). For these experiments, yf/wt and wt/wt cells were stimulated with either TSLP or GMCSF for varying times and analyzed for cis gene expression. As shown in Fig. 5, TSLP treatment of both lines resulted in upregulation of the cis gene. However, only in wt/wt cells, and not in

FIGURE 4. The single cytoplasmic tyrosine of TSLPR is required for cell proliferation but not Stat5 activation. a, BafM7R cells stably expressing GMR␤/IL-7Rwt alone, GMR␣/TSLPRwt and GMR␤/IL-7wt (wt/wt), or GMR␣/Y103F and GMR␤/IL-7wt (yf/wt) were tested for their ability to proliferate in the presence of increasing amounts of GM-CSF or, as a control, TSLP. The mean [3H]thymidine uptake from triplicate wells was normalized to account for the percentage of cells that expressed the chimeric receptors, as evidenced by flow cytometry as shown. The percentages listed within the bar graph represent the amount of [3H]thymidine incorporated into the cells following stimulation with 100 ng/ml GM-CSF as a percentage of that seen when stimulated with TSLP (mean of three to six trials ⫾ SD). b, Whole cell lysates from GM-CSF- or TSLP-stimulated wt/wt and yf/wt BafM7R cells were size-separated, and the resulting Western blot was blotted with an anti-phosphotyrosine Stat5 antiserum. Reprobing of the blot with anti-Stat5 antiserum revealed total Stat5 protein levels. Both antisera were capable of recognizing Stat5a and Stat5b.

yf/wt cells, was cis induced in response to GM-CSF treatment. To confirm these results we have used a reporter gene consisting of 305 bp of the cis gene promoter fused to luciferase, as was used previously to analyze TSLP activation of Stat5 (4). Consistent with the Northern blot analysis, we found that TSLP was capable of inducing luciferase activity in both cell lines following transfection of the reporter, while GM-CSF treatment only resulted in induction of luciferase activity in wt/wt cells (data not shown).

Table II. The sole cytoplasmic tyrosine of TSLPR is sufficient but not required for Stat5 activation GMR␣/TSLPR Construct

wt wt Y103F Y103F a

C-Terminal Residuesa

GMR␤/IL-7R Constructb

Stat5 Activation (CAT activity)c

DSGYMTL* DSGYMTL* DSGFMTL* DSGFMTL*

wt 125cyt wt 125cyt

41.4 ⫾ 4.6 45.0 ⫾ 8.2 46.2 ⫾ 5.4 3.2 ⫾ 1.7d

The asterisk denotes the carboxyl-terminal end of the protein. The 125cyt construct lacks the 70 carboxyl-terminal residues of GMR␤/IL-7R including all four cytoplasmic tyrosine residues. c CAT activity is presented as the fold stimulation following GM-CSF treatment with 1 being the amount of acetylated chloramphenicol detected without added cytokine (mean of three to eight trials ⫾ SE). d Value of p ⬍ 0.005 using an unpaired t test. b

FIGURE 5. Substitution of tyrosine 103 to phenylalanine in TSLPR prevents cis mRNA induction. BafM7R cells expressing GMR␤/IL7R with either GMR␣/TSLPR (wt/wt) or Y103F (yf/wt) were starved of cytokine for 6 h, after which time the cells were cultured in the absence or the presence of GM-CSF or TSLP for the indicated times. RNA was prepared and analyzed by Northern blotting using a probe for cis. The blot was then stripped and reprobed for GAPDH to control for sample loading. The position of cis mRNA is indicated.

The Journal of Immunology

Discussion Unlike all other known members of the cytokine receptor family, engagement of the TSLP receptor complex leads to activation of a Stat protein (Stat5) in the absence of Jak involvement. Although the precise identity of the kinase required for TSLP-mediated Stat5 activation is not known, our previous work has suggested the involvement of Tec family kinases (4). To further define the signal transduction pathway used by the TSLP receptor, we have used targeted mutagenesis of TSLPR and IL-7R␣ to identify sequence elements crucial for TSLP-mediated Stat5 activation and cell proliferation. This approach allowed us to define regions of each receptor subunit required for both processes as well as a single mutation in TSLPR that abolished TSLP-mediated cell proliferation but had no affect on Stat5 phosphorylation. This mutation, along with the IxN/2B cell line and PP1-treated NAG8/7 cells, provides further evidence for the uncoupling of Stat5 activation and cell proliferation resulting from TSLP receptor engagement. Similar to other cytokine receptors, the membrane-proximal Box1 domain of TSLPR and IL-7R␣ was required for Stat5 activation as well as proliferation. This result in itself is not surprising given that Box1 domains in other cytokine receptors act as Jak binding domains and as such are required for Stat activation and proliferation (18, 19, 25, 26). However, unlike other cytokine receptors, TSLP receptor engagement does not activate any of the four known Jaks, and overexpression of kinase-inactive versions of Jak1, Jak2, or Jak3 has no inhibitory effect on TSLP-mediated Stat5 activation (Refs. 4 and 36 and P. A. Trobridge and S. D. Levin, unpublished observations). Thus, the question remains as to why mutations to the Box1 domain of either TSLPR or IL-7R␣ prevent Stat5 activation. These domains, because they are located adjacent to the transmembrane domain, may be required for proper dimerization or folding of the receptor subunits. Conversely, they may be required for docking of signaling proteins, other than Jak family kinases, such as a Tec family kinase. Indeed, overexpression of a kinase-inactive version of Tec was able to partially inhibit TSLP-mediated Stat5 activation (4). Moreover, overexpression of Bmx, another member of the Tec kinase family, in COS cells induced Stat5 activation without activation of endogenous Jaks (27). One interesting byproduct of these studies is the finding that mutations to the Box1 domain of IL-7R␣ do not prevent IL-7mediated Stat5 activation (Fig. 3). A similar observation was made for IL-2R␤ by Higuchi et al. (28). In their study mutations in IL-2R␤ that abrogated Jak1 binding did not prevent IL-2-mediated Jak3 or Stat5 activation. It is therefore likely that the Box1 domain in IL-7R␣, and therefore Jak1 binding, are dispensable for IL-7 signaling. These findings are in contrast to those for mice deficient in Jak1, in which thymocytes have a defect in ␥c-coupled cytokine signaling (including IL-7 and IL-2) (29). However, thymocytes from these mice also have diminished responses to phorbol ester and calcium ionophore as well as to gp130-coupled cytokines and therefore may have a general survival deficit (29). A common theme of the data presented in this study is that TSLP-mediated Stat5 activation can be uncoupled from cell proliferation. We have demonstrated, using three independent systems, that Stat5 is tyrosine-phosphorylated under conditions that inhibit a TSLP-mediated proliferative response. The finding that the Src family kinase inhibitor PP1 can inhibit TSLP proliferation, but not Stat5 activation, suggests that the signaling pathway downstream of the TSLP receptor complex splits, with a Src family kinase involved in proliferative responses and a member of a

3293 different tyrosine kinase family responsible for Stat5 phosphorylation. This finding is consistent with our previous data showing that overexpression of Csk (a negative regulator of Src family kinases) did not affect TSLP-mediated Stat5 activation (4). The lack of a proliferative response to TSLP in IxN/2B cells, despite the presence of normal TSLPR and IL-7R␣ on the cell surface (Fig. 1) and the ability to bind TSLP with high affinity (2), also supports the idea of distinct signaling pathways regulating Stat5 and proliferation. As shown in Fig. 1, while these cells did not proliferate in response to TSLP, TSLP treatment did lead to Stat5 phosphorylation. Thus, it would appear that IxN/2B cells have a defect in the pathway that leads to TSLP-mediated cell proliferation, while the pathway that leads to Stat5 activation is intact. The nature of this defect remains to be determined, but, given the data presented here, it may be due to a defect in or altered expression of a Src family kinase. The final demonstration of the uncoupling of Stat5 activation and cell proliferation comes from the analysis of cells that express a mutated TSLPR cytoplasmic domain. This mutant, Y103F, was expressed as part of a chimeric receptor complex (GMR␣/TSLPR and GMR␤/IL-7R␣) in BafM7R cells (yf/wt BafM7R cells). This cell line, when stimulated with GM-CSF, activated Stat5 but did not proliferate, yet endogenous TSLP responses were intact (Fig. 4). Interestingly, the yf/wt BafM7R cells also did not up-regulate the Stat5-inducible gene cis following GM-CSF treatment. This finding was extended using a reporter plasmid that contained a portion of the cis gene promoter. This is in contrast to the data presented in Table I, showing that the TSLPR Y103F mutant was capable of activating the HRRE-CAT reporter. A possible explanation for this result comes from an examination of the promoters used in these reporter plasmids. The HRRE-CAT reporter has eight copies of a Stat5 binding site (15). The cis reporter contains the sequences ⫺100 to ⫺404 from the human cis promoter (30). In addition to four Stat5 binding sites, this region contains three binding sites for Sp-1, which has been shown to cooperate with Stat5 in the regulation of other genes, including cyclin D2 (31). It is possible that the Y103F mutation in TSLPR results in the inability to activate Sp-1. There are many reports of cytokine receptor mutations that block both Stat activation and proliferation (18, 23, 26, 32–34), including the Box1 mutations described in this work. There are also reports of cytokine receptor mutations that eliminate Stat activation without affecting cell proliferation, including erythropoietin- and IL-2-induced Stat5 activation (18), and thrombopoietininduced Stat1, -3, and -5 activation (35). To our knowledge this is the first report of cytokine receptor-mediated Stat5 activation without corresponding cell proliferation. This unique feature of TSLP receptor signaling may be a consequence of the apparently novel mechanism used by this receptor in the activation of Stat5. In receptor systems that use Janus family kinases for Stat5 activation, signals that result in Stat activation may be sufficient for proliferation. Whatever the mechanism of TSLP-mediated Stat5 activation, it is clearly different at several levels from Jak-activating receptors such as the IL-7R. This is also illustrated by the fact that TSLP activates only Stat5a and Stat5b, whereas IL-7 can activate other Stats as well (4). These differences between the IL-7 and TSLP signaling pathways may explain in part the differences seen in the response of B cell precursors to each cytokine. IL-7 treatment of fetal liver or bone marrow results in the outgrowth of B220⫹IgM⫺ pre-B cells. In contrast, TSLP treatment of the same populations results in the production of B220⫹IgM⫹ early B cells (36).

3294

Acknowledgments We are grateful to Dr. Brad H. Nelson (Virginia Mason Research Center) for the GM-CSFR␣ and -␤ plasmids, to Erin Kinzie for assistance with the CAT assay system, to Dr. Michael Bevan (University of Washington) for the pMI.2 and pMI.5 expression vectors, and to Drs. Deborah Kasprowicz and Brad H. Nelson for critical reading of the manuscript.

References 1. Leonard, W. J., and J. X. Lin. 2000. Cytokine receptor signaling pathways. J. Allergy Clin. Immunol. 105:877. 2. Park, L. S., U. Martin, K. Garka, B. C. Gliniak, J. P. DiSanto, W. Muller, D. A. Largaespada, N. G. Copeland, N. A. Jenkins, A. G. Farr, et al. 2000. Cloning of the murine thymic stromal lymphopoetin (TSLP) receptor: formation of a function heteromeric complex requires interleukin 7 receptor. J. Exp. Med. 192:659. 3. Pandey, A., K. Ozaki, H. Baumann, S. D. Levin, A. Puel, A. G. Farr, S. F. Ziegler, W. J. Leonard, and H. F. Lodish. 2000. Cloning of a novel rceptor subunit required for signaling by thymic stromal lymphopoietin. Nat. Immunol. 1:59. 4. Isaksen, D. E., H. Baumann, P. A. Trobridge, A. Farr, S. D. Levin, and S. F. Ziegler. 1999. Requirement for Stat5 in thymic stromal lymphopoietinmediated signal transduction. J. Immunol. 163:5971. 5. Sims, J. E., D. E. Williams, P. J. Morrissey, K. Garka, D. Foxworthe, V. Price, S. L. Friend, A. Farr, M. A. Bedell, N. A. Jenkins, et al. 2000. Molecular cloning and biological characterization of a novel murine lymphoid growth factor. J. Exp. Med. 192:671. 6. Friend, S. L., S. Hosier, A. Nelson, D. Foxworthe, D. E. Williams, and A. Farr. 1994. A thymic stromal cell line supports in vitro development of surface IgM⫹ B cells and produces a novel growth factor affecting B and T lineage cells. Exp. Hematol. 22:321. 7. Ray, R. J., C. Furlonger, D. E. Williams, and C. J. Paige. 1996. Characterization of thymic stromal-derived lymphopoietin (TSLP) in murine B cell development in vitro. Eur. J. Immunol. 26:10. 8. Foxwell, B. M. J., C. Beadling, D. Guschin, I. Kerr, and D. Cantrell. 1995. Interleukin-7 can induce the activation of Jak 1, Jak 3 and STAT 5 proteins in murine T cells. Eur. J. Immunol. 25:3041. 9. Knowles, B. B., C. C. Howe, and D. P. Aden. 1980. Human hepatocellular carcinoma cell lines secrete the major plasma proteins and hepatitis B surface antigen. Science 209:497. 10. Pear, W. S., G. P. Nolan, M. L. Scott, and D. Baltimore. 1993. Production of high-titer helper-free retroviruses by transient transfection. Proc. Natl. Acad. Sci. USA 90:8392. 11. Park, L. S., D. J. Friend, A. E. Schmierer, S. K. Dower, and A. E. Namen. 1990. Murine interleukin 7 (IL-7) receptor: characterization on an IL-7-dependent cell line. J. Exp. Med. 171:1073. 12. Lord, J. D., B. C. McIntosh, P. D. Greenberg, and B. H. Nelson. 1998. The IL-2 receptor promotes proliferation, bcl-2 and bcl-x induction, but not cell viability, through the adapter molecule Shc. J. Immunol. 161:4627. 13. O’Mahoney, J. V., and T. E. Adams. 1994. Optimization of experimetnal variables influencing reporter gene expression in hepatoma cells following calcium phosphate transfection. DNA Cell Biol. 13:1227. 14. Ripperger, J. A., S. Fritz, K. Richter, G. M. Hocke, F. Lottspeich, and G. H. Fey. 1995. Transcription factors Stat3 and Stat5b are present in rat liver nuclei late in an acute phase response and bind interleukin-6 response elements. J. Biol. Chem. 270:29998. 15. Morella, K. K., C.-F. Lai, S. Kumaki, N. Kumaki, Y. Wang, E. M. Bluman, B. A. Witthuhn, J. N. Ihle, J. Giri, D. P. Gearing, et al. 1995. The action of interleukin-2 receptor subunits defines a new type of signaling mechanism for hematopoietin receptors in hepatic cells and fibroblasts. J. Biol. Chem. 270:8298. 16. He, Y. W., M. L. Deftos, E. W. Ojala, and M. J. Bevan. 1998. ROR␥ T, a novel isoform of an orphan receptor, negatively regulates Fas ligand expression and IL-2 production in T cells. Immunity 9:797. 17. Hanke, J. H., J. P. Gardner, R. L. Dow, P. S. Changelian, W. H. Brissette, E. J. Weringer, B. A. Pollok, and P. A. Connelly. 1996. Discovery of a novel, potent, and Src family-selective tyrosine kinase inhibitor: study of Lck- and FynT-dependent T cell activation. J. Biol. Chem. 271:695.

UNCOUPLING OF STAT5 ACTIVATION AND PROLIFERATION 18. Quelle, F. W., D. Wang, T. Nosaka, W. E. Thierfelder, D. Stravopodis, Y. Weinstein, and J. N. Ihle. 1996. Erythropoietin induces activation of Stat5 through association with specific tyrosines on the receptor that are not required for a mitogenic response. Mol. Cell. Biol. 16:1622. 19. Witthuhn, B. A., F. W. Quelle, O. Silvennoinen, T. Yi, B. Tang, O. Miura, and J. N. Ihle. 1993. JAK2 associates with the erythropoietin receptor and is tyrosin phosphorylated and activated following stimulation with erythropoietin. Cell 74: 227. 20. Murakami, M., M. Narazaki, M. Hibi, H. Yawata, K. Yasukawa, M. Hamaguchi, T. Taga, and T. Kishimoto. 1991. Critical cytoplasmic region of the interleukin 6 signal transducer gp130 is conserved in the cytokine receptor family. Proc. Natl. Acad. Sci. USA 88:11349. 21. Narazaki, M., B. A. Witthuhn, K. Yoshida, O. Silvennoinen, K. Yasukawa, J. N. Ihle, T. Kishimoto, and T. Taga. 1994. Activation of JAK2 kinase mediated by the interleukin 6 signal transducer gp130. Proc. Natl. Acad. Sci. USA 91:2285. 22. Takaki, S., H. Kanazawa, M. Shiiba, and K. Takatsu. 1994. A critical cytoplasmic domain of the interleukin-5 (IL-5) receptor ␣ chain and its function in IL-5mediated growth signal transduction. Mol. Cell. Biol. 14:7404. 23. Friedrich, K., K. Kammer, I. Erhardt, S. Brandlein, W. Sebald, and R. Moriggl. 1999. Activation of STAT5 by IL-4 relies on Janus kinase function but not on receptor tyrosine phosphorylation, and can contribute to both cell proliferation and gene regulation. Int. Immunol. 11:1283. 24. Miura, O., N. Nakamura, F. W. Quelle, B. A. Witthuhn, J. N. Ihle, and N. Aoki. 1994. Erythropoietin induces association of the JAK2 protein tyrosine kinase with the erythropoietin receptor in vivo. Blood 84:1501. 25. Tanner, J. W., W. Chen, R. L. Young, G. D. Longmore, and A. S. Shaw. 1995. The conserved box 1 motif of cytokine receptors is required for association with JAK kinases. J. Biol. Chem. 270:6523. 26. Haan, C., H. M. Hermanns, P. C. Heinrich, and I. Behrmann. 2000. A single amino acid substitution (Trp6663 Ala) in the interbox 1/2 region of the interleukin-6 signal transducer gp130 abrogates binding of JAK1, and dominantly impairs signal transduction. Biochem. J. 349:261. 27. Saharinen, P., N. Ekman, K. Sarvas, P. Parker, K. Alitalo, and O. Silvennoinen. 1997. The Bmx tyrosine kinase induces activation of the Stat signaling pathway, which is specifically inhibited by protein kinase C␦. Blood 90:4341. 28. Higuchi, M., H. Asao, N. Tanaka, K. Oda, T. Takeshita, M. Nakamura, J. Van Snick, and K. Sugamura. 1996. Dispensability of Jak1 tyrosine kinase for interleukin-2-induced cell growth signaling in a human T cell line. Eur. J. Immunol. 26:1322. 29. Rodig, S. J., M. A. Meraz, J. M. White, P. A. Lampe, J. K. Riley, C. D. Arthur, K. L. King, K. C. Sheehan, L. Yin, D. Pennica, et al. 1998. Disruption of the Jak1 gene demonstrates obligatory and nonredundant roles of the Jaks in cytokineinduced biologic responses. Cell 93:373. 30. Verdier, F., R. Rabionet, F. Gouilleux, C. Beisenherz-Huss, P. Varlet, O. Muller, P. Mayeux, C. Lacombe, S. Gisselbrecht, and S. Chretien. 1998. A sequence of the cis gene promoter interacts preferentially with two associated STAT5A dimers: a distinct biochemical difference between STAT5A and STAT5B. Mol. Cell. Biol. 18:5852. 31. Martino, A., J. H. Holmes, J. D. Lord, J. J. Moon, and B. H. Nelson. 2001. Stat5 and Sp1 regulate transcription of the cyclin D2 gene in response to IL-2. J. Immunol. 166:1723. 32. Schmitz, J., H. Dahmen, C. Grimm, C. Gendo, G. Muller-Newen, P. C. Heinrich, and F. Schaper. 2000. The cytoplasmic tyrosine motifs in full-length glycoprotein 130 have different roles in IL-6 signal transduction. J. Immunol. 164:848. 33. Doyle, S. E., and J. C. Gasson. 1998. Characterization of the role of the human granulocyte-macrophage colony-stimulating factor receptor ␣ subunit in the activation of JAK2 and STAT5. Blood 92:867. 34. Nelson, B. H., B. C. McIntosh, L. L. Rosencrans, and P. D. Greenberg. 1997. Requirement for an initial signal from the membrane-proximal region of the interleukin 2 receptor ␥c chain for Janus kinase activation leading to T cell proliferation. Proc. Natl. Acad. Sci. USA 94:1878. 35. Dorsch, M., P. D. Fan, N. N. Danial, P. B. Rothman, and S. P. Goff. 1997. The thrombopoietin receptor can mediate proliferation without activation of the JakStat pathway. J. Exp. Med. 186:1947. 36. Levin, S. D., R. M. Koelling, S. L. Friend, D. E. Isaksen, S. F. Ziegler, R. M. Perlmutter, and A. G. Farr. 1999. Thymic stromal lymphopoietin (TSLP): acytokine that promotes the development of IgM⫹ cells in vitro and signals via a novel mechanism. J. Immunol. 162:677.