The regulation of tyrosine phosphorylation and associated signalling through antigen, growth-factor and cytokine receptors is mediated by the reciprocal activities of protein tyrosine kinases and protein tyrosine phosphatases (PTPases)1. The transmembrane PTPase CD45 is a key regulator of antigen receptor signalling in T and B cells2,3. Src-family kinases have been identified as primary molecular targets for CD45 (ref. 4). However, CD45 is highly expressed in all haematopoietic lineages at all stages of development5, indicating that CD45 could regulate other cell types and might act on additional substrates. Here we show that CD45 suppresses JAK (Janus kinase) kinases and negatively regulates cytokine receptor signalling. Targeted disruption of the cd45 gene leads to enhanced cytokine and interferon-receptor-mediated activation of JAKs and STAT (signal transducer and activators of transcription) proteins. In vitro , CD45 directly dephosphorylates and binds to JAKs. Functionally, CD45 negatively regulates interleukin-3-mediated cellular proliferation, erythropoietin-dependent haematopoieisis and antiviral responses in vitro and in vivo. Our data identify an unexpected and novel function for CD45 as a haematopoietic JAK phosphatase that negatively regulates cytokine receptor signalling.
To investigate a possible involvement of CD45 in cytokine receptor signalling, we generated interleukin-3 (IL-3)-dependent bone-marrow-derived mast cell (BMMC) lines from cd45+/+ mice and mice that carry a mutation in exon 6 of the cd45 gene (cd45-/-)2 (see Fig. 1 in Supplementary Information). BMMCs from both genotypes showed similar expression levels of FcεR1, c-Kit and IL-3 receptor β-chain, indicating that a lack of CD45 does not prevent the emergence and differentiation of BMMCs. During the generation of mast cells with IL-3, we noted that the expansion rate of cd45-/- BMMCs was greater than that of cd45+/+ BMMCs (see Fig. 1 in Supplementary Information). Proliferation of cd45-/- BMMCs was markedly increased in response to stimulation with IL-3 over that of wild-type cells (Fig. 1a). Thus, in contrast to antigen receptor stimulation, in which a loss of CD45 leads to decreased cell growth2,3, our data show that a loss of CD45 expression leads to an increased proliferation of BMMCs in response to the cytokine IL-3.
Stimulation with IL-3 triggers multiple signalling pathways such as the activation of extracellular signed-regulated kinases (ERKs), protein kinase B (PKB)/Akt and JAKs6. To assess the importance of CD45 in regulation of these signalling molecules, we analysed the phosphorylation of these molecules in response to IL-3. Phosphorylation of ERK1/ERK2 (Fig. 1b) and phosphatidylinositol-3-OH kinase (Pl(3)K)-dependent activation of PKB/Akt (see Fig. 1 in Supplementary Information) were comparable between wild-type and cd45-deficient BMMCs. In cd45 +/+ BMMCs, JAK2 tyrosine phosphorylation in response to IL-3 was readily detected. Intriguingly, IL-3-induced JAK2 tyrosine phosphorylation was significantly enhanced in cd45-/- BMMCs (Fig. 1b). Tyrosine phosphorylation of JAK2 was also enhanced in BMMCs derived from a line of independently derived CD45-null mice3 (not shown). Consistent with JAK2 hyperphosphorylation was the observation that JAK2 kinase activity was increased in the absence of CD45 after stimulation with IL-3 (Fig. 1c). Moreover, IL-3-induced tyrosine phosphorylation of the JAK substrates STAT3 (Tyr 705) and STAT5 (Tyr 694) in response to IL-3 was significantly increased in cd45/- BMMCs in comparison with wild-type BMMCs (Fig. 2a). Phosphorylation of STAT3 on Ser 727 in response to IL-3 was comparable in cd45 +/+ and cd45-/- BMMCs (Fig. 2a), indicating that the PTPase CD45 does not regulate the serine phosphorylation of STAT3. Phosphorylation of Tyr 705 in STAT3 and Tyr 694 in STAT5 is required for the dimerization and transcriptional activity of these STATs7. Consistent with the enhanced phosphorylation of STAT3, was the observation that stimulation with IL-3 induced more DNA-binding activity of STAT3 in BMMCs lacking CD45 than in control cd45+/+ BMMCs (Fig. 2b). Furthermore, the induction of cyclin D1, which mediates IL-3-dependent proliferation through STAT5 (ref. 8), was enhanced in cd45-/- BMMCs (Fig. 2c). These results indicate that CD45 negatively regulates the IL-3-triggered JAK–STAT signalling pathway.
Because Src-family kinases are direct substrates for CD45 (ref. 4), we thought that deregulated Src-family kinase activity leads to hyperactivation of the JAK–STAT pathway in cd45 -/- BMMCs. However, inhibition of Src-family kinases with the Src-family kinase inhibitor PP2 (ref. 9) did not affect JAK2 activation in cd45+/+ or cd45-/- BMMCs. Moreover, in IL-3-stimulated cd45+/+ or cd45 -/- BMMCs, treatment with PP2 had no effect on the increased phosphorylation of the specific residues Tyr 1007 and Tyr 1008 in JAK2, Tyr 705 in STAT3, and Tyr 1022 and Tyr 1023 in JAK1 (see Fig. 2 in Supplementary Information). Thus, Src-family kinases do not account for increased JAK–STAT activation in IL-3-stimulated mast cells. To evaluate whether CD45 could directly dephosphorylate JAKs and/or STATs, we used phosphatase assays in vitro with recombinant CD45 (rCD45). The rCD45 employed contained the intracellular dual PTPase domains with a relative molecular mass of 97,000 (Mr 97 K) (the catalytic D1 and the non-catalytic D2 domains) of CD45 but lacked its extracellular regions. As expected, rCD45 dephosphorylated Lyn in a dose-dependent manner (Fig. 3a). In contrast, tyrosine dephosphorylation of STAT3 (Fig. 3a, b) and STAT5 (not shown) did not occur after incubation with rCD45, even at high concentrations of rCD45. Interestingly, rCD45 could directly dephosphorylate JAK2 in vitro (Fig. 3a). In contrast, the human non-receptor tyrosine PTPase TC-PTP and the bacterial lambda PTPase did not dephosphorylate JAK2 in vitro (see Fig. 2 in Supplementary Information). Incubation of phosphorylated JAK2 with rCD45 in the presence of the PTPase inhibitor vanadate blocked JAK2 dephosphorylation (Fig. 3b). In addition, PTPase-dead rCD45 (D1:C828S mutant)10 did not dephosphorylate JAK2 (not shown), demonstrating that CD45 PTPase activity is responsible for JAK2 dephosphorylation.
Using site-specific and phospho-specific antibodies, we found that the critical tyrosine residues Tyr 1007 and Tyr 1008 of JAK2 are dephosphorylated by rCD45 (Fig. 3b), resulting in decreased JAK2 kinase activity (Fig. 3c). As with JAK2, rCD45 dephosphorylated JAK1 and Tyk2 on tyrosine residues in vitro, including the critical Tyr 1022 and Tyr 1023 residues of JAK1 (Fig. 3d). Although we have not identified all the dephosphorylation sites of JAKs by CD45, these results show that CD45 dephosphorylates functionally important tyrosine residues. It should be noted that, as with our phosphatase assays in vitro, Tyr 1022 and Tyr 1023 of JAK1, Tyr 1007 and Tyr 1008 of JAK2, and Tyr 1054 and Tyr 1055 of Tyk2 are indeed hyperphosphorylated in cd45-deficient cells (Fig. 4a; see Fig. 2 in Supplementary Information). In addition, we observed a physical association between JAK2 and the intracellular portion of a PTPase-inactive CD45 trap mutant (glutathione S-transferase (GST)–CD45 D1:C828S). Of note, the second ‘pseudo-PTPase’ domain of CD45 (GST–D2) could itself bind to JAK2 (Fig. 3e). To extend our findings in vitro to cells and to examine whether the PTPase activity of CD45 is required for the negative regulation of JAKs in vivo, we expressed wild-type and PTPase-dead CD45 (D1:C828S)11 in CD45-negative COS cells (Fig. 3f). Stimulation of COS cells with interferon-α (IFN-α) triggers the tyrosine phosphorylation of JAK1 at Tyr 1022 and Tyr 1023. Ectopic expression of wild-type CD45 resulted in decreased JAK1 tyrosine phosphorylation (Fig. 3f). Importantly, expression of the PTPase-inactive point mutant of CD45 had no effect on JAK1 tyrosine phosphorylation in response to IFN-α. Thus, CD45 can associate with JAK2 and directly regulate the tyrosine phosphorylation of JAK-family kinases.
A wide variety of cytokines are known to activate JAKs7,12. The direct dephosphorylation of JAKs by CD45 raised the possibility that, in addition to IL-3 in mast cells, IFN-α-induced activation of JAK1 (Fig. 4a) and Tyk2 (see Fig. 3 in Supplementary Information) was increased in cd45-deficient (CD45-AS) human Jurkat T cells in comparison with the parental wild-type Jurkat cells (Fig. 4a). Re-expression of wild-type CD45 in the CD45-AS cells decreased the levels of JAK1 phosphorylation to that of the parental Jurkat cells. Importantly, re-expression of PTPase-dead CD45 (D1:C828S) in CD45-AS cells did not alter increased JAK1 phosphorylation (Fig. 4a). Thus, as in the COS cell experiments, CD45-PTPase activity is required for the negative regulation of JAK1 phosphorylation in IFN-α-activated Jurkat cells. In freshly isolated thymocytes, stimulation with IFN-α resulted in JAK1 phosphorylation that was higher in cd45-/- cells than in wild-type thymocytes (Fig. 4b). Moreover, JAK1 and JAK3 tyrosine phosphorylation was upregulated in primary cd45-/- B cells in response to the B-cell cytokine IL-4 (Fig. 4c). Finally, in freshly isolated macrophages from cd45-/- mice, Tyk2, STAT1 (Tyr 701) and STAT3 (Tyr 705), but not ERK1/2, were hyperphosphorylated in response to IFN-α (see Fig. 4 in Supplementary Information). It should be noted that our results are in contrast to previous experiments with Jurkat cells and a high dose of IFN-α that suggested that CD45 is a positive regulator of JAK–STAT signalling13. Our data now demonstrate that CD45 is a negative regulator of JAKs in multiple haematopoietic cell types and in response to different cytokines and antiviral IFN-α.
Cytokine-mediated activation of JAKs is important for a wide array of cellular functions such as proliferation, differentiation, survival and host resistance to pathogens7. We first investigated the effect of CD45 deficiency on cytokine-dependent erythropoiesis and myelopoiesis, both of which are regulated by the JAK–STAT pathway14,15. The addition of erythropoietin (EPO) to bone-marrow progenitors induces the growth and differentiation of erythroid colonies (BFU-Es; burst-forming units-erythroid) in a dose-dependent manner. The numbers of EPO-dependent erythroid colonies that differentiated from cd45-/- progenitors were significantly increased in comparison with the numbers derived from cd45+/+ progenitor cells (Fig. 5a). Individual cd45 -/- erythroid colonies were also substantially greater in size and density than EPO-dependent cd45+/+ erythroid colonies. In addition, the formation of IL-3-induced mixed neutrophil/macrophage myeloid colonies was significantly increased when bone-marrow progenitors were isolated from cd45-/- mice as opposed to cd45+/+ littermates (cd45+/+, 139 ± 20; cd45-/-, 206 ± 29 (P < 0.01); 0.1 ng ml-1 IL-3, determined on day 7). Thus, a loss of CD45 in haematopoietic progenitor cells leads to increased cytokine-dependent erythropoiesis and myelopoiesis.
IFN-α and IFN-α receptor (IFN-αR) activation regulate susceptibility to viral infections in vivo and this effect is mediated by the JAK–STAT pathway16. IFN-α and IFN-αR-regulated signalling are critical for replication of the picornavirus Coxsackievirus B3 (CVB3) in vitro and CVB3 infections in vivo17. To test whether CD45 can regulate CVB3 replication directly in vitro, we measured CVB3 titres after infection of parental or cd45-deficient Jurkat cell lines. IFN-α suppressed viral amplification more efficiently in cd45-deficient cells than in parental Jurkat cells; this suppression was dependent on a functional CD45-PTPase domain (Fig. 5b). To test whether reduced CVB3 replication in cd45-deficient Jurkat cell lines in vitro would translate into altered disease pathogenesis of CVB3 infections in vivo, we inoculated cd45+/- and cd45-/- littermate mice with CVB3. Approximately 50% of cd45+/- mice succumbed to CVB3 infections within 7 days after infection owing to the acute cytopathic effects of the virus leading to encephalitis, pancreatitis, myocarditis and hepatitis. In contrast, cd45-/- littermates were completely protected from lethal CVB3 infections (Fig. 5c) and did not show any histological lesions of acute or chronic inflammation in the heart, pancreas, liver or brain (Fig. 5d, and not shown). Thus, the loss of CD45 renders mice resistant to lethal infections with CVB3. The role of CD45 in other viral infections in vivo needs to be determined.
The JAK family contains four members, JAK1, JAK2, JAK3 and Tyk2, that are differentially activated in response to different cytokines7. JAKs and JAK-mediated STAT activation have crucial roles in proliferation, differentiation, survival, host resistance to pathogens, and malignancies18. Thus, multiple inhibitory mechanisms and molecular backup systems must exist to regulate cytokine-triggered cellular activation. SOCS (suppressor of cytokine signalling) family proteins bind to phosphotyrosine residues within the activation loop of the JAK proteins to inhibit their kinase activity19. Moreover, the PTPase SHP1 is recruited to certain cytokine receptors and binds to JAKs20. SHP1 can function as a negative regulator of JAK–STAT signalling, although it remains to be shown whether the inhibitory effects of this phosphatase is mediated by dephosphorylating signalling molecules regulating JAKs or by direct dephosphorylation of JAKs1. We provide evidence that CD45 is a novel negative regulator of cytokine-receptor-mediated proliferation and haematopoiesis and IFN-regulated antiviral responses. Genetic inactivation of CD45 leads to hyperactivation of the JAK–STAT pathway in vivo. In humans, loss of CD45 expression has been reported in more than 10% of patients with acute lymphoblastic leukaemia21, and CD45 expression is frequently lost in Hodgkin's lymphoma cells22. Moreover, a genetic mutation of CD45 has been reported in a child that ultimately succumbed to a B-lymphoma23.
JAK expression is not restricted to haematopoietic cells, and JAK–STAT pathways have also been implicated in the homeostasis of cardiac myocytes, mammary gland involution, and responses to insulin, growth hormone or leptin24. Presumably other JAK PTPases exist in addition to the haematopoietic-restricted CD45 that regulate JAKs in non-haematopoietic cells. For example, the transmembrane PTPase LAR, a close structural homologue of CD45, has been shown to downregulate insulin receptor signalling in vivo25. In our preliminary studies in vitro, LAR was indeed capable of dephosphorylating JAK2. It remains to be established by genetic means whether LAR is a JAK PTPase operating in non-haematopoietic cells and whether LAR-regulated JAK-STAT signalling has a role in glucose and/or fat metabolism.
The results of this study provide genetic, functional and biochemical evidence that the transmembrane protein tyrosine phosphatase CD45 can directly dephosphorylate and inactivate JAK-family kinases. CD45 negatively regulates JAK–STAT signalling in response to multiple cytokines, including IL-3, IL-4, EPO and IFN-α. Genetic inactivation of CD45 in cells enhanced cytokine-triggered proliferation, differentiation and antiviral activities. These data identify an unexpected role for the haematopoietic phosphatase CD45 as the elusive JAK phosphatase and a novel negative regulator of cytokine receptor signalling.
Mice, cells and reagents
CD45-exon6 (ref. 2) and CD45-exon9 (ref. 3) mice have been previously generated and were maintained in accordance with institutional guidelines. BMMCs were cultured as described26. Macrophages, thymocytes and splenic B cells were freshly isolated from cd45 wild-type and cd45-mutant littermate mice26. The CD45-negative human Jurkat cell lines CD45-AS (J-AS-1)11 and J45.01 were cultured in RPMI medium plus 5% FCS (fetal calf serum). Phycoerythrin-, fluorescein isothiocyanate- or biotin-conjugated antibodies against pan-CD45, CD19, c-Kit (CD119), Gr1 and Mac1 (CD11b) (PharMingen) were used for flow cytometry. Antibodies against STAT3, STAT1, STAT5, or PKB/Akt and STAT3 (Tyr 705), STAT3 (Ser 727), STAT5 (Tyr 694), STAT1 (Tyr 701), PKB/Akt (Ser 473 or Thr 308), ERK1/ERK2 (Thr 202 and Tyr 204), JAK1 (Tyr 1022 and Tyr 1023), Tyk2 (Tyr 1054 and Tyr 1055) and JAK2 (Tyr 1007 and Tyr 1008) were from QCB. The anti-phosphotyrosine (PY99) antibody and antisera against Lyn, JAK1, JAK2, Tyk2, JAK3 and c-Kit were from Santa Cruz Biotechnology. PP2, recombinant CD45, LAR, TC-PTP and lambda phosphatase were from Calbiochem. Antibodies against JAK3 and IL-3 receptor β-chain and purified JAK2 and Lyn were from Upstate Biotech. One unit of phosphatase activity was defined as the amount of enzyme that hydrolysed 1.0 nmol of p-nitrophenyl phosphate per minute at 30 °C and pH 7.0.
Immunoblotting, immunoprecipitation and transient expression of CD45
Murine BMMCs, splenic B cells, thymocytes, peritoneal macrophages and Jurkat T-cell lines were stimulated with various cytokines. For immunoprecipitations, lysates were incubated with antibodies conjugated with agarose or Sepharose beads for 2 h at 4 °C. Total cell lysates or immunoprecipitates were separated using SDS–polyacrylamide gel electrophoresis, transferred onto membranes and immunoblotted. COS cells were transfected using diethylaminoethyl-dextron transfection technique with wild-type or PTPase-dead CD45 (D1:C828s) as described11.
For tyrosine phosphatase assays, immunoprecipitated Lyn, JAKs or STATs from lysates of 107 BMMCs (for Lyn, JAK1, JAK2, STAT3 and STAT5) or macrophages (for Tyk2) were washed in lysis buffer and phosphatase reaction buffer10. The amounts of Lyn and JAK2 in the immunoprecipitates used in Fig. 3a were 72 ng and 48 ng, respectively, as determined by silver staining. Recombinant CD45 was added to substrates in the presence or absence of 0.1 mM Na 3VO4 followed by incubation at 30 °C for 20 min. Tyrosine phosphorylation was monitored by immunoblotting with phosphotyrosine-specific antibodies. The extent of tyrosine phosphorylation and amount of substrates were determined by quantifying the intensities of the immunoblots with NIH Image 1.61 software. Kinase activity in anti-Lyn or anti-JAK immunoprecipitates was assessed by an in vitro kinase assay with Raytide (Oncogen Science) as a substrate or using authophosphorylation26. In vitro binding assays were performed as described10, using 1 µg of soluble GST, GST–CD45-D2, or GST–PTPase-dead CD45 (D1:C828S). EMSAs were as described27. The sequences of the EMSA probes were: STAT3, 5′-GATCCTTCTGGGAATTCCTAGATC-3′; STAT3 mutant, 5′-GATCCTTCTGGGCCGTCCTAGATC-3′.
Proliferation, cell death, colony formation and CVB3 infections
For proliferation of BMMCs, cells were deprived of IL-3 for 12 h and then stimulated with IL-3 for 24 h and pulse-labelled with [3H]thymidine for 12 h. The percentage of cell death was determined by 7-amino-actinomycin D (7AAD) and staining with Annexin V. Haematopoietic precursors were isolated from the bone marrow of 6-week-old cd45-/- and cd45+/+ mice and the formation of EPO-dependent erythroid colonies (BFU-E) and IL-3-dependent mixed neutrophil/macrophage myeloid colonies was determined. Total bone-marrow cells were plated in methylcellulose containing 5% fetal bovine serum. EPO (0.1, 0.3 and 1 unit ml-1) or IL-3 (0.1 ng ml -1) was added to the culture to induce erythroid or myeloid colony formation, respectively. The sizes and numbers of colonies were analysed 7 and 12 days after the start of culture.
For Coxsackievirus infections in vitro, Jurkat cells were pretreated with IFN-α (1,000 units ml -1) for 4 h and inoculated in triplicate with 3 × 105 CVB3 for 1 h at 37 °C. Cells were then washed to remove excess IFN-α and excess CBV3 virus, and resuspended in RPMI medium containing 10% FBS. After 24 h, culture supernatants were analysed for the presence of infectious CVB3 by using plaque-forming assays in HeLa cell monolayers. For CVB3 infections in vivo, the cd45 mutation was backcrossed onto an A/J (H2k/k) background for five generations. A/J (H2k/k) mice are highly susceptible to acute and chronic CVB3 infections. Four-week-old mice were inoculated intraperitoneally with 105 plaque-forming units of CVB3. For lethality scores, cd45-/- (n = 25) and cd45 +/- (n = 22) littermate mice were monitored daily after infection. CVB3-infected cd45+/- mice showed signs of severe systemic illness, including lethargy, ruffled coats and anorexia, before death, but cd45-/- mice failed to show these signs. At defined time points, mice were killed and organs were processed for histology. Sections from eight mice per group were stained with haematoxylin/eosin and Masson blue. In all experiments, the CVB3 strain Gauntt-Chow was used.
We thank M. Saunders for scientific editing; T. Mak for providing CD45 mutant mice; and T. Mak, M. Reth, W. Yeh, D. Barber, V. Stambolic, C. Mirtsos, K. Bachmaier, A. Oliveira-dos-Santos, M. Crackower, Y. Kong, N. Joza, I. Kozieradzki and Q. Liu for comments and advice. This work was supported by grants from the Canadian Institutes for Health Research (CIHR) and the National Cancer Institute (NCI) of Canada and Amgen to J.M.P., and from the NIH to D.M.R. J.M.P. holds a Canadian Research Chair in Cell Biology.