|Home | About | Journals | Submit | Contact Us | Français|
Various cytokines utilize Janus kinase (JAK) and the STAT (signal transducers and activators of transcription) family of transcription factors to carry out their biological functions. Among STATs, two highly related proteins, STAT5a and STAT5b, are activated by various cytokines, including prolactin, growth hormone, erythropoietin, interleukin 2 (IL-2), and IL-3. We have cloned a STAT5-dependent immediate-early cytokine-responsive gene, CIS1 (encoding cytokine-inducible SH2-containing protein 1). In this study, we created CIS1 transgenic mice under the control of a β-actin promoter. The transgenic mice developed normally; however, their body weight was lower than that of the wild-type mice, suggesting a defect in growth hormone signaling. Female transgenic mice failed to lactate after parturition because of a failure in terminal differentiation of the mammary glands, suggesting a defect in prolactin signaling. The IL-2-dependent upregulation of the IL-2 receptor α chain and proliferation were partially suppressed in the T cells of transgenic mice. These phenotypes remarkably resembled those found in STAT5a and/or STAT5b knockout mice. Indeed, STAT5 tyrosine phosphorylation was suppressed in mammary glands and the liver. Furthermore, the IL-2-induced activation of STAT5 was markedly inhibited in T cells in transgenic mice, while leukemia inhibitory factor-induced STAT3 phosphorylation was not affected. We also found that the numbers of γδ T cells, as well as those of natural killer (NK) cells and NKT cells, were dramatically decreased and that Th1/Th2 differentiation was altered in transgenic mice. These data suggest that CIS1 functions as a specific negative regulator of STAT5 in vivo and plays an important regulatory role in the liver, mammary glands, and T cells.
The STAT (signal transducers and activators of transcription) proteins are cytosolic latent transcription factors that are rapidly activated by interferons (IFNs), cytokines, and growth factors (4, 9, 26). A total of seven different STAT proteins, many of which play highly specific roles in innate and acquired immunity, have been identified. Mice with STAT1, STAT4, and STAT6 gene knockouts are viable but lack functions that are mediated by IFNs, interleukin 12 (IL-12), or IL-4/IL-13, respectively (5, 13, 20, 29, 33). In contrast, STAT3 knockout mice exhibit fetal lethality (34), a finding consistent with the activation of STAT3 by many cytokines that are important for development, such as leukemia inhibitory factor (LIF), cardiotrophin-1, and IL-6 (4, 9). The highly related proteins STAT5a and STAT5b are activated by various cytokines, including growth hormone (GH), prolactin (PRL), erythropoietin (EPO), IL-2, IL-3, granulocyte-macrophage colony-stimulating factor (GM-CSF), and thrombopoietin. The phenotypes of STAT5a and STAT5b single-gene knockout mice and double-knockout mice revealed important roles of STAT5 in PRL and GH signaling as well as in IL-2-dependent T-cell functions (7, 11, 17, 23, 35, 36). STAT5a is critical in PRL signaling (17), GM-CSF signaling (7), or IL-2 signaling (23), while STAT5b is necessary to maintain the sexual dimorphism of body growth rates and liver gene expression (36), IL-2-mediated T-cell proliferation, and NK cell development (11, 21).
The regulation of the Janus kinases (JAKs) and STATs is a central component in the control of cytokine signaling. Because of the critical role of cytokines in mediating inflammation and immunity, it could be proposed that the constitutive activation of JAKs could contribute to hematopoietic disorders, autoimmunity, and inflammatory diseases. However, the mechanisms that terminate or down-modulate the JAK-STAT pathway have not been fully understood. Considerable evidence suggests that one mechanism involves the recruitment of a tyrosine phosphatase containing the SH2 domain (SHP-1) to receptor complexes, resulting in the dephosphorylation of JAKs (12, 14). The potential importance of this mechanism is strongly suggested by the phenotype of motheaten (me/me) mice lacking SHP-1, which die from a disease with components of autoimmunity and inflammation (30). However, it has also been suggested that a family of small SH2 domain-containing proteins may be involved in the relatively specific regulation of cytokine signaling.
The CIS (cytokine-inducible SH2-containing protein) family of proteins, also referred to as the SOCS (suppressors of cytokine signaling) or SSI (STAT-induced STAT inhibitor) family, has been implicated in regulating signal transduction by a variety of cytokines (1, 31, 43). The gene encoding first member of this family, CIS1, was cloned as an immediate-early gene responding to a number of cytokines, including EPO, IL-2, IL-3, and GM-CSF (41), and is regulated by STAT5 (19). CIS1 binds to the tyrosine-phosphorylated IL-3 and EPO receptors and negatively regulates their signals (41). The second family member was independently cloned by three groups and is termed JAB (JAK-binding protein), SOCS-1, or SSI-1 (6, 22, 31). JAB is induced by IFN-γ and inhibits IFN as well as IL-6 signaling (28, 32). JAB and CIS3 directly bind to the kinase domain of JAKs, thereby inhibiting the kinase activity (32). This family now contains eight members, although most of them have not been well characterized except for CIS1, CIS3, and JAB (18, 43).
The CIS1 promoter contains two pairs of tandem TTCNNNGAA motifs that are capable of binding to STAT5 (19). The essential role of STAT5 in CIS1 expression was confirmed by the observation that CIS1 expression was not observed in the ovaries of STAT5a and -b double-gene knockout mice (35). In addition to STAT5-dependent expression of CIS1, it was interesting that CIS1 could negatively modulate STAT5 activation (19); forced expression of CIS1 partially suppressed EPO-dependent STAT5 activation in 293 and Ba/F3 cells. Thus, we hypothesized that CIS1 acts as a kind of negative feedback regulator of the JAK-STAT5 pathway.
To elucidate the physiological function of CIS1 and to find the relationship of the dysregulated expression of the CIS1 gene to human diseases, we created CIS1 transgenic mice. A striking similarity of phenotypes was found between CIS1 transgenic mice and STAT5 knockout mice. The CIS1 transgenic mice exhibited growth retardation and defects in mammary gland development as well as in T-cell and NK cell development. In addition, we found that helper T cells of CIS1 transgenic mice tend to differentiate into Th2 cells, suggesting an important role of CIS1 in T-cell differentiation. We also showed a decrease in STAT5 activation in response to GH, PRL, and IL-2 in transgenic mice. This study suggests an important regulatory role of CIS1 and STAT5 in T-cell development and differentiation, in addition to PRL and GH functions.
The pCAGGS-CIS1 expression vector was constructed by subcloning the myc-tagged CIS1 cDNA (18) into the EcoRI cloning site of the pCAGGS (24). After digestion with SalI and HindIII, the fragment carrying the promoter, cDNA, and the 3′ noncoding region was used for microinjection into (C57BL/6 × C3H)F1 (B6C3-F1) zygotes. Eggs surviving microinjection were transferred into the oviducts of recipient pseudopregnant females as described elsewhere (10). Transgenic mice were identified by PCR analysis of tail genomic DNA, and protein expression was confirmed by immunoblotting with anti-Myc.
Virgin and lactating mice were sacrificed, and mammary-gland tissue was taken, fixed in 10% formalin, paraffin embedded, and sectioned. Mammary sections were stained with hematoxylin and eosin and used for histological analysis.
The postpartum mammary glands and other tissues were immediately frozen in liquid nitrogen and stored at −80°C. Total-cell extracts were prepared from tissue homogenized in 50 mM Tris-HCl (pH 8.0)–0.5% Nonidet P-40–1 mM EDTA–150 mM NaCl–10% glycerol–1 mM sodium vanadate–50 mM sodium fluoride–10 mM sodium pyrophosphate–1 mM phenylmethylsulfonyl fluoride with protease inhibitor cocktail (Sigma Chemical Co., St. Louis, Mo.). The extracts were cleared by spinning at 15,000 rpm at 4°C for 15 min. The samples were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and proteins were detected by immunoblotting as described elsewhere (19). Anti-STAT5 (C-17) and anti-STAT3 (C-20) were purchased from Santa Cruz Biotechnology, Inc. Tyrosine-phosphorylated STAT3 and STAT5 were detected by using phosphospecific anti-STAT3 (New England Biolabs) and anti-STAT5 (anti-pY-STAT3 and anti-pY-STAT5) (Upstate Biotechnology) as described elsewhere (32). To detect the mobility shift of STAT5, relatively small amounts of cell extracts (1 μg of protein for mammary glands or 5 μg for the liver) were resolved by SDS–6% PAGE.
Total RNA from various tissues was prepared with TRIZOL (Gibco-BRL) according to the manufacturer’s instructions. Total RNA (5 μg) was separated on 1.0% agarose–2.4% formaldehyde gels, then hybridized with digoxigenin-labeled riboprobes prepared by using a digoxigenin-RNA labeling kit (Boehringer Mannheim Japan, Tokyo) as described elsewhere (8, 32, 41).
Cells were suspended in phosphate-buffered saline (PBS) supplemented with 2% fetal calf serum and 0.1% sodium azide. In general, 106 cells were blocked with anti-Fc receptor (2.4 G2) and stained by a standard method described previously (15). Flow cytometry analysis was performed with FACSort (Becton Dickinson, San Jose, Calif.) by using CELL Quest software. The reagents (fluorescein isothiocyanate [FITC]-conjugated anti-CD4 [RM4-5], anti-CD3 [125-2C11], anti-B220 [RA3-6B2], anti-T-cell receptor [anti-TCR] [H57-597], and anti-IL-2 receptor α chain (7D4); phycoerythrin [PE]-conjugated anti-CD8 [53-6.7], anti-immunoglobulin M [anti-IgM], anti-NK1.1 [PK136], and anti-γδ TCR [GL3]) were purchased from Pharmingen (La Jolla, Calif.).
To remove B cells and monocytes, spleen cells were cultured in plates precoated with anti-mouse IgG for 1 h. Unattached cells were used as splenic T cells. The splenic T cells were preactivated with a plate-bound anti-TCR monoclonal antibody (MAb) (H57-597) in the presence or absence of an anti-CD28 MAb (PV-1) (1 μg/ml). To isolate CD4+ naive T cells, spleen cells were incubated with an anti-CD8 MAb (53-6.72) at 4°C for 1 h, and the cells were incubated on plates coated with anti-mouse Ig to eliminate B and CD8+ T cells. The CD4+ enriched T cells were incubated with an anti-CD44 MAb (1M7), followed by a cytotoxic killing treatment with Low-Tox-M rabbit complement (Cedarlane Laboratories Limited, Hornby, Ontario, Canada). These CD4+ naive T-cell preparations contained more than 80% CD4+ CD44− T cells. The CD4+ naive T cells were preactivated with a plate-bound anti-TCR MAb. After 36 h, the T cells were cultured with various concentrations of IL-2 either in the presence or in the absence of phorbol myristate acetate (PMA) (50 ng/ml) for 48 h and then pulse-labeled with 1 μCi of [3H]thymidine for an additional 8 h. Recombinant mouse IL-2 was purchased from PeproTech EC Ltd. (London, England).
The CD4+ naive T cells were stimulated with an anti-TCR MAb plus an anti-CD28 MAb in the presence or absence of IL-4 as described previously (2). After 5 days, the primed CD4+ T cells were repurified by panning with an anti-CD8 MAb and anti-mouse Ig and were restimulated with an anti-TCR MAb for 6 h in the presence of 4 μM monensin (Sigma). Then the cells were fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. After the cells were blocked with PBS containing 3% bovine serum albumin, cells were stained with anti-IFN-γ (XMG1.2)–FITC and anti-IL-4 (11B11)–PE as described previously (39). All antibodies were purchased from Pharmingen.
Major urinary proteins (MUPs) were detected by SDS–15% PAGE and staining with Coomassie blue in 0.25 μl of urine as described previously (36).
The STAT5-responsive promoter-luciferase reporter gene, which carries four repeats of the IFN-γ activation site (GAS) sequence with the jun promoter, was described previously (18). GH- and PRL-dependent luciferase activity in 293 cells grown in six-well plates was measured after transfection with the reporter gene (50 ng of plasmid), the β-galactosidase gene (50 ng), murine GH or PRL receptor cDNA (200 ng), the indicated amount of Flag-tagged CIS1 cDNA, and murine STAT5a or STAT5b (5 to 10 ng) as previously described (18, 32, 40). For GH stimulation, serum was omitted from the medium because of the high level of GH in serum. Murine GH receptor cDNA cloned by PCR was subcloned into pcDNA3. Murine PRL receptor cDNA in pECE was donated by Wolfgang Doppler (Innsbruck University, Innsbruck, Austria). Murine STAT5a and -b cDNAs in pMX were provided by Toshio Kitamura (Tokyo University, Tokyo, Japan). Recombinant ovine GH and PRL were kind gifts from Arieh Gertler (The Hebrew University of Jerusalem).
CIS1 transgenic mice were generated by the injection of Myc-epitope-tagged CIS1 cDNA with chicken β-actin promoter into B6C3-F1 eggs. Myc-CIS1 expression was examined by Northern blotting and immunoblotting with either anti-Myc or anti-CIS1 antibodies, and three transgenic lines were established. One of the lines (Tg1) expressed the transgenic CIS1 gene in the mammary glands but not much in the liver and spleen (Fig. (Fig.1A;1A; see also Fig. Fig.4A4A and and6A).6A). The other two lines (Tg2 and Tg3) expressed transgenes in all tissues examined (thymus, spleen, kidney, mammary glands, and liver), although Myc-CIS1 expression levels in most tissues was higher in Tg3 than in Tg2 (Fig. (Fig.1A,1A, see also Fig. Fig.6A).6A). Tg2 and Tg3 lines, which express Myc-CIS1 in the liver, showed a reduction in body weight compared with wild-type mice, while the body weight of the Tg1 line, which did not express detectable Myc-CIS1 in the liver, was normal (Fig. (Fig.1B).1B). Growth retardation was obvious in the Tg3 line from early ages, as shown in Fig. Fig.1C,1C, while Tg2 mice exhibited a normal growth rate until the age of 2 to 3 months but did not gain much weight thereafter (Fig. (Fig.1B1B and data not shown). At the age of 12 weeks, both the male and female Tg3 mice weighed 20 to 30% less than their wild-type littermates. The reduction in body weight was more drastic in male than in female mice. Tg2 and Tg3 mice had significantly less adipose tissue than wild-type mice (data not shown). These phenotypes suggest a defect in GH signaling in CIS1 transgenic mice.
To further examine the effect of CIS1 overexpression on GH signaling in the liver, we compared the amount of MUP in the urine of wild-type and transgenic mice. MUP is a family of α2-microglobulin-related liver secretory proteins whose expression is regulated by male-specific pulsatile GH (25). As illustrated in Fig. Fig.2A,2A, the levels of MUP in both male and female Tg3 mice (lanes 4 and 8) were much lower than those in wild-type mice (lanes 1 and 5). The MUP levels of Tg2 CIS1 transgenic mice (lanes 3 and 7) were also lower than those in wild-type mice, but the reduction was less drastic than that in Tg3 mice. We did not see any reduction in MUP levels in Tg1 mice, which marginally expressed transgenic CIS1 in the liver (Fig. (Fig.2A,2A, lanes 2 and 6). These results indicate that exogenous CIS1 expression levels correlate significantly with the inhibition of GH-mediated gene expression in the liver.
GH exerts its function through the activation of STAT5. Since STAT5b single-knockout mice as well as STAT5a,b double-knockout mice also exhibited similar defects in growth retardation and MUP secretion (35, 36), we examined the effect of STAT5 activation in the livers of Tg3 mice, using immunoblotting with a phosphorylated STAT5-specific antibody (Fig. 2Bb) and with the antibody that recognizes both STAT5a and STAT5b (Fig. 2Ba). As shown in Fig. Fig.2B,2B, STAT5 phosphorylation constitutively occurred in the livers of wild-type male mice (lane 1) and was detected by the mobility shift of the STAT5 band (Fig. 2Ba) with high-resolution SDS-PAGE (3, 27), as well as by immunoblotting with anti-phospho-STAT5 specific antibody (Fig. 2Bb). However, activation of STAT5 was not seen in CIS1 transgenic male mice (Fig. (Fig.2B,2B, lane 2). STAT5 was not constitutively activated in the livers of either wild-type or transgenic female mice (Fig. (Fig.2B,2B, lanes 3 and 4). However, injection of GH intraperitoneally into female mice resulted in strong phosphorylation of STAT5 (Fig. (Fig.2B,2B, lane 5). GH-induced phosphorylation of STAT5 in the livers of transgenic female mice was less evident than that in wild-type female mice (Fig. (Fig.2B,2B, lane 6). This suggests that forced expression of CIS1 suppressed GH-dependent STAT5 activation in the liver.
All three transgenic lines expressed Myc-CIS1 in mammary glands (Fig. (Fig.4A),4A), and all of them exhibited defects in mammary-gland development (Fig. (Fig.3),3), although the fertility, length of gravidity, and litter size were normal. All CIS1 transgenic females failed to lactate after parturition, and milk was not detected in the stomachs of the pups despite vigorous suckling by the pups. Upon fostering with wild-type females, these pups thrived, which suggests that CIS1 transgenic females could not produce or eject milk. Whole-mount analysis of mammary tissue from postpartum CIS1 transgenic mice demonstrated incomplete mammopoiesis (data not shown). Fig. Fig.33 shows sections of mammary gland tissues of virgin mice (A and E) and of mice 1 to 3 days postpartum (B through D and F through H). The secretory tissue from transgenic mice was sparse, and alveoli were petite and contained small lumina (representative data for Tg3 mice are shown in Fig. Fig.3).3).
To examine STAT5 activation in mammary glands, tyrosine phosphorylation of STAT5 was detected in tissue extracts postpartum. The expression levels of STAT5 in transgenic mice were slightly lower than those in wild-type mice (Fig. 4Ba). This may be due to a block of mammary-gland maturation during pregnancy, since the STAT5 level has been shown to increase at late pregnancy (16). STAT5 bands were seen as doublets in wild-type mice, probably due to phosphorylation, but not in transgenic mice (Fig. 4Ba). Tyrosine phosphorylation of STAT5 was very low in transgenic mice (Fig. 4Bb). These data suggest that PRL-dependent STAT5 activation is inhibited in CIS1 transgenic mice.
Expression of the endogenous CIS1 gene itself is strictly regulated by STAT5 function (7, 19, 21, 35). Therefore, we measured CIS1 mRNA levels in mammary glands. As shown in Fig. 4Ca, exogenous CIS1 mRNA levels in Tg3 female mice were 3 to 5 times higher than endogenous CIS1 mRNA levels in mammary glands from pregnant female mice. Endogenous CIS1 expression was suppressed in transgenic mice compared with that in wild-type mice, suggesting low STAT5 activity in mammary glands in transgenic mice. We also examined the expression of other CIS family genes because most of them are cytokine inducible (18). CIS2 and CIS3 expression were also detected, and the levels were slightly increased during lactation in wild-type mice (Fig. 4Cb and c). Interestingly, CIS2 and CIS3 expression, like that of endogenous CIS1, was suppressed in transgenic mice, although CIS3 reached normal levels at 3 days postpartum. These data raise the possibility that CIS2 and CIS3 are also regulated by STAT5.
We next examined the level of milk protein, as well as c-Fos and c-Myc levels, in mammary glands (Fig. 4Cd through h). CIS1 overexpression in mammary glands resulted in a significant decrease in whey acidic protein (WAP) (e), while α-lactalbumin, β-casein, and Westmead DMBA8 nonmetastatic CDNA1 (WDNM) levels were not much affected (Fig. 4Cd and data not shown). Similar decreases in WAP, but not in other milk proteins, were observed in the mammary glands of STAT5a-deficient mice (17). WAP expression was partially recovered on days 2 and 3 (Fig. 4Ce, lanes 5 and 6). A similar recovery in the WAP level was observed in STAT5a single-knockout mice, suggesting that STAT5 activity remained in CIS1 transgenic mice (17, 35). As shown in Fig. 4Cf and g, neither c-Myc nor c-Fos expression was affected by CIS1 overexpression. These two genes are probably regulated independently of STAT5 (9). Thus, these data suggest that CIS1 overexpression selectively modulates STAT5-regulated gene expression in mammary glands.
To demonstrate a direct inhibition of GH- and PRL-induced STAT5 activity by CIS1, we used a transient reporter gene assay in 293 cells (18, 40). A STAT5 reporter gene construct was transfected into 293 cells with the GH receptor (Fig. (Fig.5A5A and B) or the PRL receptor (Fig. (Fig.5C)5C) and CIS1 cDNAs. Increased CIS1 expression inhibited STAT5 activation in response to GH (Fig. (Fig.5A)5A) and PRL (data not shown). GH induced STAT5 transcriptional activity in a dose-dependent manner, and CIS1 expression suppressed GH-mediated STAT5 activation (Fig. (Fig.5B).5B). At a low dose of GH (10 ng/ml), the suppression of STAT5 activity by CIS1 was almost complete, but it was partial at a higher dose of GH (Fig. (Fig.5B).5B). Therefore, the suppression of STAT5 activity was dependent on both hormone and CIS1 expression levels. CIS1 also strongly suppressed PRL-induced STAT5 activity (Fig. (Fig.5C).5C). When exogenous STAT5a or STAT5b was coexpressed, CIS1 still could suppress reporter gene activation, but the inhibitory effect of CIS1 became partial (Fig. (Fig.5C5C and data not shown). Similarly, inhibition of GH-induced STAT5 activation by CIS1 was overcome by STAT5 expression (data not shown). Therefore, the negative effect of CIS1 was also dependent on STAT5 expression levels. These data are consistent with our hypothesis that CIS1 suppresses STAT5 activation by blocking the binding of STAT5 to the receptor binding sites.
Next, we evaluated the role of CIS1 on the development of T cells, since recent STAT5 gene knockout studies revealed an important role of STAT5 in these cells. Tg2 and Tg3 mice expressed Myc-CIS1 in the spleen, while Tg1 expressed it at lower levels (Fig. (Fig.6A).6A). The expression levels of endogenous and exogenous CIS1 mRNA (Fig. (Fig.6B)6B) were compared in unstimulated splenic T cells (lanes 1 and 3) and splenic T cells stimulated (lanes 2 and 4) with anti-TCR and anti-CD28 MAbs (TCR-plus-CD28 stimulation). Stimulation was confirmed by the induction of IL-2 receptor α chain mRNA (Fig. 6Bb). In unstimulated T cells of Tg3 mice, myc-CIS1 mRNA was expressed at high levels, (Fig. 6Ba, lane 3), while endogenous CIS1 was undetectable in wild-type mice (lane 1). However, the endogenous CIS1 mRNA was strongly induced by TCR-plus-CD28 stimulation in both wild-type and Tg3 mice (Fig. 6Ba, lanes 2 and 4). The level of endogenous CIS1 was comparable to that of exogenous Myc-CIS1 in Tg3 mice (Fig. (Fig.6B,6B, lane 4).
Transgenic expression of CIS1 did not significantly affect the number of thymocytes and splenocytes (data not shown). The CIS1 transgenic mice showed slight reductions in the numbers of CD8+SP cells in the spleen, while the numbers of CD4+SP cells were unaffected (data not shown). The most drastic difference in splenic T cells between wild-type and transgenic mice was the number of γδ T cells (Fig. (Fig.6D).6D). In transgenic mice, the numbers of NK1.1+ CD3− (NK) and NK1.1+ CD3+ (NKT) cells were decreased in the spleen, bone marrow, and liver (Fig. (Fig.6C6C and data not shown). Especially, the number of liver NKT cells in Tg3 mice was about 10% of that in wild-type mice (Fig. (Fig.6C).6C). We did not see much reduction in the numbers of γδ T cells and NK T cells in Tg1 mice, probably reflecting low exogenous CIS1 expression levels (data not shown). These data indicate that overexpression of CIS1 suppresses γδ T-cell, NK cell, and NK T-cell development.
We evaluated the effect of CIS1 overexpression on T-cell responsiveness to IL-2 stimulation, since IL-2 strongly induces CIS1 expression (41). Using STAT5a or -b single-knockout mice, Nakajima et al. (23) and Imada et al. (11) reported that the IL-2 receptor α chain has been shown to be partly regulated by STAT5. Therefore, we measured the increase in the IL-2 receptor α chain level in response to IL-2 by following their methods. First splenocytes from wild-type and CIS1 transgenic mice were preactivated with PMA and concanavalin A (ConA) for 24 h, and then cells were cultured in the presence of IL-2 for an additional 40 h. At each stage, the IL-2 receptor α chain expression was examined with a fluorescence-activated cell sorter (Fig. (Fig.7A).7A). PMA-plus-ConA treatment induced the IL-2 receptor α chain similarly in wild-type and transgenic mice (Fig. 7Ab and e). Further stimulation with IL-2 resulted in the appearance of a population with a much higher level of IL-2 receptor α chain expression in wild-type mice (Fig. 7Ac). However, the appearance of such an IL-2 receptor α chain-high population was retarded in CIS1 transgenic mice (Fig. 7Af).
We examined the effect of CIS1 expression on the IL-2-induced proliferation of T cells. CD4+ T cells were enriched from the spleen and preactivated for 36 h with a plate-bound anti-TCR MAb. After harvest and washing, T cells were incubated with various amounts of IL-2 in the presence or absence of PMA. As shown in Fig. Fig.7B,7B, T cells from CIS1 transgenic mice constantly presented a partial reduction in IL-2-dependent proliferative responses. The responses of T cells from CIS1 transgenic mice were about 50% of those from their wild-type littermates at low doses of IL-2 in the presence of PMA. However, the responses to IL-2 were similar in wild-type and transgenic mice at high concentrations of IL-2.
Next, we examined the effect of transgenic CIS1 on the induction of STAT5-responsive genes (the endogenous CIS1 and oncostatin M [OSM] [41, 42] genes) as well as the phosphorylation of STAT5 in splenic T cells. As shown in Fig. 8Aa and b, IL-2 strongly induced the expression of endogenous CIS1 and OSM in wild-type mice (lane 2). However, induction of these genes was partially suppressed in CIS1 transgenic mice (Fig. (Fig.8A,8A, lane 4), while c-myc was equally induced by IL-2 in wild-type and transgenic mice (Fig. 8Ac), suggesting that CIS1 selectively inhibited the induction of STAT5-responsive genes.
IL-2-dependent STAT5 phosphorylation was markedly reduced in T cells preactivated with anti-TCR from CIS1 transgenic mice compared with those from their wild-type littermates (Fig. 8Bb). On the other hand, LIF-induced tyrosine phosphorylation of STAT3 was not affected by CIS1 overexpression (Fig. 8Cb). These data indicate that the overexpression of CIS1 specifically inhibits STAT5 activation, thereby suppressing IL-2-induced IL-2 receptor α chain expression, proliferation, and STAT5-responsive gene expression.
We observed that stimulation of T cells with the antibodies against TCR and CD28 markedly enhanced the expression of CIS1 mRNA (Fig. (Fig.6B).6B). Therefore, we addressed the question of whether constitutive expression of CIS1 affects the development of helper T cells induced by TCR-plus-CD28 stimulation in the presence or absence of IL-4. After 5 days, the primed CD4+ T cells were restimulated with the anti-TCR MAb for 6 h, and the Th1 and Th2 differentiation profiles were evaluated by intracellular IL-4 and IFN-γ staining (2). In B6C3-F1 wild-type mice, this stimulation generated quite a low IL-4-positive (Th2) population, while IFN-γ-positive (Th1) cells were predominant (Fig. (Fig.9A).9A). The presence of IL-4 in the initial activation stage was required for the development of Th2 cells in wild-type mice (Fig. (Fig.9A).9A). The transgenic expression of CIS1 apparently influenced the proportions of Th1 and Th2 cells. The number of Th1 cells consistently decreased in CIS1 transgenic mice to almost two-thirds of that in their wild-type littermates (Fig. (Fig.9B).9B). In contrast, Th2 development was significantly increased by CIS1 overexpression, and the proportion of Th2 cells in transgenic mice was twice as high as that in wild-type mice (Fig. (Fig.9B).9B). The increase in Th2 development in transgenic mice was confirmed by an enzyme-linked immunosorbent assay. Precultured T cells from CIS1 transgenic mice showed the ability to secrete three-times-higher amounts of IL-4 than wild-type mice (data not shown). Taken together, CIS1 profoundly skews helper T-cell development and shifts the balance from Th1 dominant to Th2 dominant.
We have shown that CIS1 is induced by several cytokines that activate STAT5 and that it binds to the EPO and IL-3 receptors, thereby partially suppressing STAT5 activity. We have also shown that CIS1 overexpression suppressed EPO-, GH-, and PRL-induced STAT5 activation by using a reporter gene assay (18, 19) (Fig. (Fig.5).5). The inhibitory effect of CIS1 depends on the cytokine concentration and the CIS1 level as well as the STAT5 level (Fig. (Fig.5).5). In this study, using transgenic mice, we demonstrated that CIS1 overexpression actually inhibited several cytokine- or hormone-mediated STAT5 functions in vivo. We confirmed phenotypes by using three or two independent transgenic lines, and the extent of phenotypes was dependent on the expression levels of transgenic CIS1.
The phenotypes observed in CIS1 transgenic mice are strikingly similar to those of STAT5a and/or STAT5b knockout mice. CIS1 transgenic mice exhibited defects in growth, mammary-gland development, and T-cell response and NK cell development. This is a mixture of the phenotypes found in STAT5a and STAT5b single-knockout mice or is close to that of STAT5a,b double-knockout mice. For example, STAT5a knockout results in retardation of mammary-gland development but not of growth, while in STAT5b knockout mice, the body growth of males was retarded but mammary-gland development was not much affected (35). NK cell development was more severely affected in STAT5b null mutant mice than in STAT5a null mutant mice (11). It has been reported that the major STAT5 form activated by GH in the liver is STAT5b (25, 35), while the major form activated in mammary glands is STAT5a (3); such a different localization may account for the separate effects of STAT5a and -b single-gene knockouts in mice, as suggested by Teglund et al. (35). Since CIS1 can suppress both STAT5a and STAT5b (data not shown), the phenotype of CIS1 transgenic mice could be similar to that of STAT5a,b double-knockout mice.
However, there are some differences between CIS1 transgenic mice and STAT5a,b double-knockout mice. For example, CIS1 transgenic female mice are fertile, while STAT5a,b double-mutant female mice are infertile and have altered ovarian development (35). T cells from STAT5a,b double-knockout mice fail to proliferate in response to anti-TCR and IL-2 (21), while suppression in CIS1 transgenic mice is only partial. There is a detectable reduction in the number of hematopoietic colonies induced in response to IL-3, IL-5, and GM-CSF in STAT5a,b double-knockout mice (35), while we did not see any difference in hematopoietic colony formation in vitro in CIS1 transgenic mice (data not shown). These differences are probably due to the partial inhibitory effect of CIS1 on STAT5a and -b, which depends on cytokine concentrations as well as CIS1 and STAT5 expression levels. Nevertheless, the similarity of the phenotypes of CIS1 transgenic mice to those of STAT5 knockout mice strongly supports our notion that CIS1 is a specific inhibitor of STAT5.
We have reported on other CIS1-related proteins, including JAB and CIS2 to CIS6 (6, 18, 28, 32). We have tried to create JAB, CIS2, CIS3, and CIS4 transgenic mice under the control of the β-actin promoter, which is also used for CIS1 transgenic mice. We have established CIS2 and CIS4 transgenic mice with levels of protein expression comparable to that of CIS1. However, we have not seen any defects so far in growth, lactation, and T-cell development in these transgenic mice. Therefore, the phenotype of CIS1 transgenic mice is not due to overexpression of an SH2 domain; rather, CIS1 seems to specifically inhibit STAT5 activity. Consistent with this notion, CIS2 and CIS4 did not inhibit EPO-dependent STAT5 activation (18). The target molecules of CIS2 and CIS4 have not been identified as yet; however, detailed analysis of the phenotype of transgenic mice will give us a clue to the functions of CIS2 and CIS4. On the other hand, we could not obtain JAB and CIS3 transgenic mice using the β-actin promoter. These two molecules can bind to JAKs directly and strongly inhibit kinase activity. Therefore, JAB and CIS3 overexpression is probably embryo lethal, like JAK2 gene disruption.
The inhibitory effect of CIS1 on cytokine-dependent STAT5 activation is partial after stimulation with very high doses of cytokines (see Fig. Fig.5).5). On the other hand, CIS3 can strongly inhibit PRL-dependent STAT5 activation, and the inhibitory effect of CIS3 on STAT5 activation was much more profound than that of CIS1 (8). In this sense, it is notable that CIS3 is expressed at high levels in mammary glands. Furthermore, CIS3 expression was partially suppressed in CIS1 transgenic mice, suggesting that these two molecules are also regulated by STAT5 in this tissue. We recently found that CIS3 binds to the same cytokine receptors as CIS1, including the EPO receptor and the IL-2 receptor β chain. Moreover, the binding region of the EPO receptor to CIS3 is similar to that to CIS1. Therefore, CIS3 may have an overlapping function with CIS1.
STAT5a and STAT5b have been shown to play an important role in T-cell and NK cell development (11, 21, 23). T cells from STAT5b single-knockout mice do not respond to anti-TCR stimulation, and expression of the IL-2 receptor β chain is diminished (11). NK cell development is also impaired in STAT5b-deficient mice. The response of T cells to IL-2 was also partially suppressed in CIS1 transgenic mice. However, these defects in CIS1 transgenic mice and STAT5a knockout mice are modest compared to those in STAT5b single-knockout or STAT5a,b double-knockout mice. Thus, CIS1 transgenic mice rather resemble STAT5a knockout mice in the response of T cells to IL-2. Interestingly, the number of γδ T cells was decreased in CIS1 transgenic mice as in STAT5a-deficient mice, while these cells were normal in STAT5b-deficient mice. In contrast, the decreases in the numbers of NK cells and NKT cells in CIS1 transgenic mice were more drastic than those found in STAT5b knockout mice, while the number of NK cells was not much affected in STAT5a knockout mice (11, 23). Therefore, again, defects of lymphocytes in CIS1 transgenic mice are a mixture of those found in mice lacking either STAT5a or STAT5b. Such different effects of STAT5a and STAT5b may be explained by their distinctive expression patterns in specific lymphocyte populations and by different target genes of the two proteins. Importantly, STAT5a,b double-knockout mice develop splenomegaly and have T cells with an activated phenotype, which were not seen in mice lacking STAT5a or STAT5b alone or in our CIS1 transgenic mice. This is probably because T cells in CIS1 transgenic mice as well as STAT5a or -b signal knockout mice retain some STAT5 activity.
It is notable that endogenous CIS1 was normally induced by TCR-plus-CD28 stimulation in CIS1 transgenic mice (Fig. (Fig.6B,6B, lane 4), while it was marginally induced through TCR-plus-IL-2 stimulation (Fig. (Fig.8A,8A, lane 4). The IL-2 receptor α chain was also equally induced by TCR-plus-CD28 stimulation in wild-type and Tg3 mice (Fig. 6Bb). This may be because signals from TCR and CD28 induced the endogenous CIS1 gene and the IL-2 receptor α chain without using STAT5 or because CIS1 did not inhibit TCR- and CD28-dependent signals, including TCR-mediated STAT5 activation (38).
In this study, we found that Th1/Th2 development was altered by CIS1 constitutive expression. Helper T-cell development has not been examined in STAT5a or -b single-knockout mice. Th1/Th2 development has been shown to be profoundly regulated by IL-4–STAT6 and IL-12–STAT4 systems. However, it has been well known that CD28 costimulation induces the secretion of large amounts of IL-2 in naive T cells, and IL-2 signals have been shown to be required for the proliferation and maintenance of Th1 cells. Therefore, Th2 cells may develop in CIS1 transgenic mice, since IL-2 signals would be blocked by CIS1 overexpression. On the other hand, Moriggl et al. (21) showed that STAT5a,b double-deficient lymphocytes consistently produced less IL-2 and more IFN-γ. As suggested by the authors, high IFN-γ production in the T cells of double-knockout mice may be due to predominance of a CD62L− CD44+ population. However, as an alternative explanation, an increase in the Th1-type cytokine level in STAT5a,b double-knockout mice may be attributed to the lack of CIS1 expression, since CIS1 was not expressed in T cells from STAT5a,b double-knockout mice. This is consistent with our data of Th2 increase in CIS1 transgenic mice. Our data raise the interesting possibility that CIS1 rather than STAT5 plays an important role in regulating Th1/Th2 balance.
CIS1 protein has been shown to associate with the tyrosine-phosphorylated EPO and IL-3 receptors (41). We recently found that CIS1 protein can also bind to the PRL receptor and the IL-2-receptor β chain but not to gp130, a signal-transducing subunit of the IL-6 receptor. Selective binding of CIS1 to the receptors which activate STAT5 may explain the selective negative effect of CIS1 on STAT5 activation.
However, the molecular mechanism of STAT5 inhibition by CIS1 has not been clarified yet. We reported previously that CIS1 binds to the region of the EPO receptor containing the second tyrosine residue (Y401) (37). One of the mechanisms is simply masking the STAT5 binding sites on the receptor. This model is supported by our observation that overexpression of STAT5 can overcome the negative effect of CIS1 (Fig. (Fig.5C).5C). The other possibility is that CIS1 accelerates the degradation of the receptor-CIS1 complex by the ubiquitin-proteasome pathway (37). Since proteasome inhibitors stabilize the tyrosine-phosphorylated form of the EPO receptor as well as STAT5, and since CIS1 is rapidly ubiquitinated, this model is an attractive hypothesis. However, this model cannot explain the selective negative effects of CIS1 overexpression on STAT5, since CIS1 overexpression did not affect EPO or IL-2-dependent c-myc induction (19) (Fig. (Fig.8).8). Further study is necessary to elucidate the mechanism of STAT5 inhibition by CIS1.
We thank H. Ohgusu and M. Sasaki for excellent technical assistance, M. Chikushi for preparing the manuscript, A. Gertler for donating purified GH and PRL, W. Doppler for providing PRL receptor cDNA, and T. Kitamura for making available to us STAT5a and -b cDNAs.
Part of this work was supported by grants from the Ministry of Science, Education, and Culture of Japan, the Uehara Memorial Foundation, the TORAY Research Foundation, the Naito Memorial Foundation, the Sumitomo Foundation, and the Kanae Research Foundation.