Regulation of Erythropoietin-induced STAT Serine Phosphorylation by Distinct Mitogen-activated Protein Kinases*

The STAT proteins are a family of latent transcription factors that are activated by a wide variety of cytokines. Upon receptor engagement, STATs become tyrosine phosphorylated, translocate to the nucleus, and induce expression of target genes. In addition to tyrosine phosphorylation, maximal activation of some STAT proteins requires serine phosphorylation within the transactivation domain. Here we focus on STAT phosphorylation after engagement of the erythropoietin receptor (EPO-R). In Ba/F3-EPO-R cells, EPO induces tyrosine and serine phosphorylation of STAT1, STAT3, STAT5A, and STAT5B. Identical regions of the EPO-R couple to both tyrosine and serine phosphorylation of each cognate STAT protein. A proximal region of the EPO-R lacking cytoplasmic tyrosines couples to STAT1 and STAT3 phosphorylation as well as ERK and p38HOGactivation, but not JNK/SAPK. STAT1 serine phosphorylation was perturbed by inhibition of ERK and p38 pathways, whereas only inhibition of ERK activation blocked STAT3 serine phosphorylation in response to EPO. STAT5A/B phosphorylation is downstream of EPO-R Tyr343, however, STAT5A/B serine phosphorylation is unaffected by either ERK or p38 inhibition. Physiological responses induced by EPO may depend on regulation of serine phosphorylation of the STAT molecules by p38HOG and the ERK family of kinases as well as additional serine/threonine kinases.

dimerization of its receptor, resulting in trans-phosphorylation of receptor-associated Janus Kinase 2 (JAK2) molecules (3,4). Activated JAK2 phosphorylates some or all of the eight tyrosine residues in the intracellular domain of the EPO-R. Phosphorylation of these intracellular tyrosines in turn recruits other intracellular proteins that bind to the EPO-R via their Src homology 2 (SH2) domains (reviewed in Ref. 5). The critical importance of EPO (6), EPO-R (6 -8), and JAK2 (9, 10) for efficient erythopoiesis have been demonstrated by targeted deletion in mice.
STAT proteins are an important class of signaling molecules that are recruited to the erythropoietin receptor. These latent transcription factors are activated at the receptor and transduce signals to the nucleus. While seven STAT molecules have been identified, only STAT5A and STAT5B are consistently activated by EPO in all cell types (13)(14)(15)(16)(17)(18), whereas STAT1 and STAT3 are activated in erythroid cell types or by Friend virus transformation (11,12,25). STAT5A/B knockout mice have defects in prolactin (26), growth hormone (26), and interleukin-2 signaling (27), display anemia during fetal development (28,29) and decreased formation of CD71 lo Ter119 hi cells in STAT5A/B Ϫ/Ϫ adult mice (30). No erythroid abnormalities were been described in the original reports of STAT1 Ϫ/Ϫ mice (31,32), however, a more recent report suggested differences in clonogenic potential of burst forming unit-erythroid cells (33). STAT3 Ϫ/Ϫ mice are embryonic lethal, precluding analysis of their role in erythropoiesis by conventional gene targeting approaches (34).
While tyrosine phosphorylation is required for cytokine-induced STAT dimerization, nuclear translocation, and DNA binding, full transcriptional activity of the homodimer is manifested only when a serine residue in the transcription activation domain is also phosphorylated (35). While the identity of the serine kinases is unclear, these putative proline-directed phosphorylation sites may be targetted by the mitogen-activated protein kinases (MAPKs).
The MAPKs comprise a family of serine/threonine kinases that are activated by extracellular signals. There are at least three distinct types of MAPKs: the classical ERK1/ERK2 kinases, the p38 HOG MAPKs (hereafter referred to as p38), and the stress-activated protein kinase/Jun kinase (SAPK/JNK; hereafter referred to as SAPK) subfamily. All play important roles in erythropoietin-induced differentiation or apoptosis (36 -38). In addition, roles for p38 and ERK1/2 in phosphorylation of STAT1 and STAT3 have been described (39 -42).
Here we examine phosphorylation of STAT1, STAT3, and STAT5 in erythropoietin signal transduction in Ba/F3-EPO-R and primary murine erythroid progenitors. Tyrosine phosphorylation of STAT1 and STAT3 is JAK2-dependent but does not require EPO-R tyrosines, whereas STAT5A and STAT5B tyrosine phosphorylation is mediated by EPO-R Tyr 343 . STAT1 and STAT3 serine phosphorylation minimally requires a proximal portion of the EPO receptor which is also necessary for ERK and p38, but not SAPK activation. In contrast, STAT5A and STAT5B serine phosphorylation is independent of SAPK, ERK, and p38. We propose that distinct signal transduction pathways couple to EPO-dependent serine phosphorylation of STAT1, STAT3, STAT5A, and STAT5B.
Cell Culture-Ba/F3 cells expressing the wild-type and truncated forms of the EPO receptor have been described. 2 Cells were maintained in RPMI 1640 media supplemented with 10% fetal calf serum and WEHI 3B-conditioned media and grown in a 5% CO 2 environment at 37°C. For cytokine stimulations, cells were washed three times in growth media lacking WEHI, and incubated 4 -6 h in RMPI 1640 supplemented with 0.5% fetal calf serum. After starvation, the cells were centrifuged and resuspended in a minimal volume of the same media. The cells were then stimulated with IL-3 (10 ng/ml) or EPO (dose and time indicated in Fig. legends).
Ba/F3-EPO-R cells were transfected with pTKLucSIE (44) or the (STAT5 RE) 6 tk-luc (gift of Bernd Groner, Frankfurt, Germany) (45), along with pBabe puro in a 10:1 molar ratio using electroporation. (STAT5RE) 6 tk-luc consists of six copies of the GAS luciferase sequence in the rat ␤-casein promoter linked to a thymidine kinase minimal promoter and luciferase (45). Two days after transfection, cells were selected in 3 g/ml puromycin. After 5 days, all non-transfected cells were dead, and the pooled population of puromycin-resistant cells were used in experiments as described.
Inhibitors-Following cytokine depletion, cells were resuspended in a minimal volume and pretreated with PD98059 (100 M), U0126 (30 M), or SB202190 (10 M) for 30 min. Cells were then stimulated with cytokines and lysed as indicated above.
Immunoprecipitations-Cell extracts were subjected to immunoprecipitation with anti-STAT5A or anti-STAT5B antibodies at 4°C for times ranging from 1 to 16 h. Protein A-Sepharose was added to the lysates for 1 h at 4°C. The beads were washed three times with lysis buffer and then boiled in SDS-PAGE loading buffer for 3 min. The samples were resolved on SDS-PAGE and subjected to Western blotting as described above.
Luciferase Assays-For Ba/F3 STAT reporter assays, cells were washed twice with ice-cold phosphate-buffered saline, and aliquoted in 1.5-ml microcentrifuge tubes. The cells were centrifuged at 5000 ϫ g for 2 min, and pellets were lysed in Triton X-100 buffer (100 mM KH 2 PO 4 , pH 7.8, 2 mM dithiothreitol, 1% Triton X-100). The lysates were centrifuged at 15,000 ϫ g for 10 min at 4°C, and supernatants were used in luciferase assays. For each sample, 60 l of lysate was aliquoted into luciferase assay tubes. One-hundred microliters of assay reagent (20 mM Tricine (pH 7. EPO-dependent luciferase activity in 293T cells was performed essentially as described (46). Briefly, 293T cells in 24-well dishes were transfected with 50 ng of pSV-␤-galactosidase, 100 ng of EPO-R construct, 50 ng of STAT construct (pRC STAT1, pRC STAT3, or pcDNA3 STAT5B), and 100 ng of the reporter using the PolyFect transfection reagent (Qiagen). pTKLucSIE was used for STAT1 and STAT3 experiments, whereas (STAT5 RE) 6 Tk-luc was used with STAT5B analysis. Twenty-four hours after transfection, cells were stimulated with 60 units/ml EPO for a further 24 h. ␤-Galactosidase and luciferase assays were performed using the Dual-Light kit (Tropix, Foster City, CA). Luciferase values were normalized for transfection efficiency by ␤-galactosidase readings.
Preparation of Primary Erythroid Progenitor Cells-Populations of erythroid progenitor cells were prepared from spleens of mice treated intraperitoneally with phenylhydrazine HCl (Sigma; 60 mg/kg) on day 1 and day 2 (19). On day 5, mice were sacrificed and the disrupted spleen was passed through a 70-m cell strainer (Falcon), exposed for 4 min to 50 mM ammonium chloride (in phosphate-buffered saline) to lyse mature red blood cells, washed twice in phosphate-buffered saline, and starved 4 -6 h in 0.5% fetal bovine serum, ␣-modified Dulbecco's media supplemented with antibiotics. Cells were stimulated with recombinant erythropoietin for the indicated time followed by immunoprecipitation or Western analysis as described above.

RESULTS
Erythropoietin Induces Serine and Tyrosine Phosphorylation of STAT Proteins-We examined phosphorylation of STAT1, STAT3, STAT5A, and STAT5B after erythropoietin stimulation of Ba/F3 cells expressing the murine EPO-R. Cells were depleted of cytokine, and stimulated with EPO, IL-3, or left unstimulated. Lysates were subjected to Western blotting using phosphotyrosine-specific antibodies against STAT1 or STAT3. To evaluate STAT5A or STAT5B phosphorylation, each was immunoprecipitated with isoform-specific antibodies, and subjected to analysis using a phosphotyrosine-STAT5 antibody. EPO stimulated robust and dose-dependent STAT1, STAT3, and STAT5 tyrosine phosphorylation (Fig. 1, A-D). The kinetics of tyrosine phosphorylation were similar for all the STAT proteins, peaking after 15 min of stimulation (Fig. 2

, A-D).
We next examined serine phosphorylation of the STAT proteins upon EPO stimulation. Serine 727 of Stat1 and Stat3 (35), serine 725 of STAT5A, and serine 730 STAT5B (47) have classical MAPK family phosphorylation sites, characterized by flanking carboxyl proline residues (reviewed in Ref. 48). Using antibodies directed against these phosphoproteins, we evaluated EPO-induced STAT serine phosphorylation. EPO stimulation induced serine phosphorylation of STAT1, STAT3, STAT5A, and STAT5B (see Fig. 1, A-D). In contrast to tyrosine phosphorylation which peaked at 15 min, serine phosphorylation persisted until at least 60 min following EPO stimulation (see Fig. 2, A-D). Tyrosine and serine phosphorylation of STAT1, STAT3, STAT5A, and STAT5B was also observed in UT-7/EPO cells (49), a human cell line that expresses endogenous EPO-R (data not shown).
Erythropoietin Leads to Activation of the STATs in Enriched Primary Erythroid Progenitors-To examine STAT activation in response to EPO in primary erythroid progenitors, we prepared splenocytes from phenylhydrazine-treated mice (19). Phenylhydrazine induces hemolytic anemia, leading to enrichment of erythroid progenitors in the spleen. Three days after phenylhydrazine treatment, splenocytes were isolated, depleted of cytokines, and stimulated with EPO. As noted in Ba/F3 cells, EPO stimulation induces serine and tyrosine phosphorylation of STAT1, STAT5A, and STAT5B (Fig. 3). However, no STAT3 phosphorylation can be detected in EPO-stimulated progenitors. STAT1 serine phosphorylation was much less robust compared with Ba/F3-EPO-R cells.
EPO Stimulates the Tyrosine Phosphorylation of JAK2 in a Panel of EPO-R Deletion Mutants-A series of Ba/F3 cell lines expressing the carboxyl-terminal deletions of the EPO-R (Fig.  4) were examined for EPO-dependent JAK2 activation (Fig.  4B). EPO stimulated the tyrosine phosphorylation of JAK2 in EPO-R, EPO-R Ϫ43, EPO-R Ϫ69, EPO-R Ϫ99, and EPO-R Ϫ99, Y343F (EPO-R Ϫ99, F1) cells. However, tyrosine phosphorylation was not observed in EPO-R Ϫ221, consistent with previous findings, as this mutant disrupts the ability of JAK2 to associate with the EPO-R (50). All of the cell lines activated tyrosine phosphorylation of JAK2 in response to IL-3 stimulation.
STAT1 and STAT3 Tyrosine Phosphorylation Is Independent of EPO-R Tyrosine Phosphorylation-The region of the EPO-R that couples to STAT1, STAT3, STAT5A, and STAT5B tyrosine phosphorylation was examined in the deletion mutants described above (Fig. 5A). EPO stimulated STAT1 tyrosine phosphorylation in EPO-R, EPO-R Ϫ43, EPO-R Ϫ69, and EPO-R Ϫ99. Interestingly, tyrosine phosphorylation of STAT1 was also observed in an EPO-R mutant devoid of tyrosine residues, EPO-R Ϫ99, F1. IL-3 stimulated STAT1 tyrosine phosphorylation in all cell lines.
Similarly, EPO stimulated STAT3 tyrosine phosphorylation in all cell lines examined with the exception of EPO-R Ϫ221 (Fig. 5B). The level of EPO-dependent STAT3 tyrosine phosphorylation was less than IL-3 stimulation, but reproducible in several experiments (data not shown). These data are the first evidence that EPO is capable of activating STAT1 and STAT3, independent of SH2-dependent recruitment to the EPO-R.
STAT5A and STAT5B Require Tyrosine Phosphorylation of EPO-R Tyr 343 -To examine tyrosine phosphorylation of STAT5A and STAT5B, immunoprecipitations were performed with peptide-specific antibodies, followed by Western blotting with a phosphotyrosine-specific STAT5. Analysis of EPO-dependent STAT5A and STAT5B tyrosine phosphorylation revealed that EPO stimulated a robust tyrosine phosphorylation of STAT5A and STAT5B in EPO-R, EPO-R Ϫ43, EPO-R Ϫ69, and EPO-R Ϫ99 cells. STAT5A and STAT5B were also weakly tyrosine phosphorylated in EPO-R Ϫ99,F1 cells, as previously reported (16). Reprobing of the STAT5B immunoblot revealed distinct migration patterns of STAT5B correlating with cytokine stimulation.
Serine Phosphorylation of STAT1 and STAT3 Requires the Proximal Portion of the EPO-R-To determine the region of the EPO receptor required for serine phosphorylation of STAT1 and STAT3, the identical Ba/F3-EPO-R deletion mutants were tested. Serine phosphorylation of STAT1 and STAT3 did not require tyrosine phosphorylation of the EPO-R, as EPO-R Ϫ99,F1 (lacking all tyrosine residues) retained an ability to induce EPO-dependent STAT1 and STAT3 serine phosphorylation (Fig. 6, A and B). However, JAK2 activation is required for both STAT1 and STAT3 serine phosphorylation, as EPO failed to stimulate STAT1 and STAT3 serine phosphorylation in Ba/F3-EPO-R Ϫ221 cells, which lack the JAK2-binding site.
Serine Phosphorylation of STAT5A and STAT5B Requires Tyr 343 of the EPO-R-We evaluated the portion of the EPO receptor that is required for STAT5A and STAT5B serine phosphorylation. After stimulation with EPO, serine phosphorylation of STAT5A and STAT5B was detected in cells expressing EPO-R Ϫ43 (4 tyrosines) and Ϫ99 (1 tyrosine). Mutation of Tyr 343 in the EPO-R Ϫ99 receptor to phenylalanine (EPO-R Ϫ99, F1) extinguished serine phosphorylation at serine 725 and serine 730 of STAT5A and STAT5B, respectively (Fig. 6, C  and D). In contrast, IL-3 stimulated serine phosphorylation in all of the cells tested. These results indicate that EPO-R Tyr 343 is essential for serine phosphorylation of both STAT5A and STAT5B, a situation which mimics the requirements for tyrosine phosphorylation of STAT5 (13)(14)(15)(16)(17)(18).
Dose-and Time-dependent Activation of MAP Kinases after EPO Stimulation-As the MAP kinase family is implicated in serine phosphorylation of some STAT proteins, we examined activation of p38, SAPK, and ERK after EPO stimulation in Ba/F3-EPO-R cells. All of these kinases were phosphorylated by erythropoietin in a dose-dependent manner in Ba/F3-EPO-R cells using phospho-specific antibodies (Fig. 7). The kinetics of ERK activation precedes p38 phosphorylation while SAPK displayed a delayed and less robust pattern of activation (Fig. 8). Activation of these kinases was similar to the kinetics to STAT serine phosphorylation (see Fig. 2).

Activation of MAPKs Are Mediated by Distinct Portions of the EPO-R-
To determine the region of the erythropoietin receptor required for EPO-induced MAPK activation, we utilized a series of Ba/F3 subclones each expressing progressive COOHterminal truncations of the EPO-R. p38 and ERK1/2 were activated by all truncation mutants except EPO-R Ϫ221 (Fig. 9,  A and B). In contrast, SAPK activation appeared to require a more distal region of the EPO-R, as it was activated by fulllength EPO-R, but absent in the EPO-R Ϫ43 truncation (Fig.  9C). The proximal region of the EPO-R therefore couples to STAT1 and STAT3 phosphorylation as well as to ERK1/2 and p38 activation, suggesting that ERK1/2 and p38, but not SAPK, may be associated with EPO-dependent STAT1 and STAT3 serine phosphorylation.
Inhibition of MAPKs Blocks STAT Serine Phosphorylation-To examine the requirement of the MAPKs for STAT serine phosphorylation, we used the chemical inhibitors PD98059, U0126 and SB202190. PD98059 specifically binds to

FIG. 3. Erythropoietin induces serine and tyrosine phosphorylation of STAT proteins in a time-dependent manner in phenylhydrazine-treated splenocytes.
Splenocytes were depleted of cytokine, then stimulated with 60 units/ml EPO for the indicated time. Cells were lysed and subjected to Western analysis. Antibodies against the phosphoserine (pSer), phosphotyrosine (pTyr), or total STAT (total) protein are shown.

FIG. 4. EPO-dependent JAK2 activation in Ba/F3-EPO-R cell lines.
A, schematic diagram of EPO-R truncation mutants used in this study. B, Ba/F3 subclones expressing the indicated EPO-R mutant were depleted of cytokine, and stimulated with 60 units/ml EPO or 10 ng/ml IL-3 for 15 min. Cells were lysed and immunoprecipitations were performed with a peptide-specific JAK2 antibody. Western analysis was performed using the monoclonal anti-phosphotyrosine antibody, 4G10. The membrane was stripped and reprobed with a JAK2 antibody.
MEK1, whereas U0126 binds MEK1 and MEK2, thereby disrupting kinase activation and specifically inhibiting activation of the ERK molecules (51). SB202190 is an inhibitor of the p38 ␣/␤ MAPKs, but does not affect the activities of SAPK or ERK molecules (52). Cells were treated with EPO for 15 min, lysed, and subjected to Western blotting using phospho-specific antibodies for STAT1 or STAT3. STAT5A and STAT5B isoforms were immunoprecipitated individually and subjected to Western blotting using a phosphoserine-STAT5A/B antibody.
Treatment with either the MEK or p38 inhibitors does not affect serine phosphorylation of STAT5A or STAT5B (Fig. 10, C  and D). However, MEK inhibition eliminates EPO-induced serine phosphorylation of STAT1 and STAT3 (Fig. 10, A and B). STAT1, but not STAT3, or STAT5A/B serine phosphorylation is also dependent on p38 kinase activity, as it is blocked by preincubation with SB202190. Similar results were obtained with UT-7/Epo cells (data not shown), a human cell line that expresses endogenous EPO-R, suggesting that these results are not unique to Ba/F3-EPO-R cells.
We next examined transcriptional activation by the STAT molecules in the presence or absence of the chemical inhibitors. Ba/F3-EPO-R cells were transfected with genetic reporters downstream of the STAT molecules. To test STAT5 activation, cells were transfected with the promoter of the ␤-casein gene cloned upstream of the luciferase gene. A construct containing the c-fos serum-inducible factor element (SIE), placed upstream of the luciferase gene, was used to gauge activation of STAT1 and STAT3. As the consensus sequence for STAT1 and STAT3 are very similar (53), it is not possible to unequivocally monitor STAT1 and STAT3 transcriptional activation independently.
As predicted by the pattern of serine phosphorylation by EPO-R receptor mutants, inhibition of p38 or ERK after EPO stimulation failed to block STAT5-induced transcriptional activation from the ␤-casein reporter (Fig. 11). In contrast, the SIE reporter was diminished upon treatment with the MEK or p38 inhibitors, consistent with binding of both STAT1 and STAT3. Although we could not demonstrate any effect of p38 inhibition on STAT5 serine (Fig. 9, C and D) or tyrosine (data not shown) phosphorylation, SB202190 slightly affected the STAT5-reponsive GAS promoter. This result may indicate that p38 regulates STAT5-dependent processes independent of Ser 725 /Ser 730 phosphorylation. A promoterless luciferase construct, pTKluc, was not affected by treatment of cells with EPO, or any of the chemical inhibitors.
JAK2 Activation Is Sufficient to Activate p38-and ERK-dependent Gene Expression-To examine transcriptional activation by the STAT proteins by the various EPO-R mutants, we performed a series of transient transfections of 293T cells. A reconstitution system was established by introduction of several EPO-R truncation mutants; STAT1, STAT3, or STAT5B; and the STAT reporters described above (46,54). As predicted by the STAT tyrosine and serine phosphorylation data, maximal activation was observed when cells were reconstituted with the wild type EPO-R in the presence of STAT1, STAT3, or STAT5B. A reduction in reporter activity was observed in EPO-R Ϫ99 and EPO-R Ϫ99,F1. This suggests that the optimal STAT-dependent gene expression requires multiple EPO-R tyrosine residues including Tyr 343 . These data indicate that monitoring STAT tyrosine and serine phosphorylation alone may not reflect the complexity of STAT transcriptional regulation. DISCUSSION STAT proteins have been implicated in a variety of EPOinduced responses, but the mechanisms linking EPO to STAT activation remain less clearly understood. Here we demonstrate that EPO regulates the tyrosine and serine phosphorylation of STAT1, STAT3, STAT5A, and STAT5B in Ba/F3 cells and in primary murine splenocytes. Our data show that tyrosine and serine phosphorylation of STAT proteins require identical regions of the EPO-R. Specifically, tyrosine and serine phosphorylation of STAT1 and STAT3 requires a membraneproximal region of the EPO-R whereas STAT5A and STAT5B tyrosine and serine phosphorylation minimally requires Tyr 343 of the EPO-R. Our results indicate that distinct pathways regulate phosphorylation of each STAT molecule. Using genetic and pharmacologic analyses we show that MEK or p38 MAPK inhibitors perturb STAT1 serine phosphorylation, while only MEK antagonists block STAT3 serine phosphorylation. STAT5A and STAT5B serine phosphorylation is unaffected by either MEK or p38 inhibitors.
Recent data utilizing a genetic knock-in of a mouse EPO-R deletion mutant demonstrated that EPO-R tyrosine residues are dispensable for viability in vivo (29). Surprisingly, we observed that STAT1 and STAT3 tyrosine phosphorylation is observed in a Ba/F3-EPO-R deletion mutant devoid of cytoplasmic tyrosines, but capable of robust EPO-dependent JAK2 activation. We failed to observe tyrosine phosphorylation of STAT3 in primary erythroblasts, suggesting that the role of STAT1 activation may be most relevant in vivo. These results raise the possibility that STAT1 collaborates with STAT5A/B in effective erythropoiesis. It has been recently reported that erythroid burst-forming units from STAT1-deficient mice are impaired compared with normal mice (33) and STAT1 expression is increased in STAT5A/STAT5B knockout mice (26). STAT5A/B nullizygous animals display a modest phenotype with fetal anemia (28) and impaired differentiation of splenic erythroblasts isolated from adult mice (30). This suggests that STAT5 may be required during expansion of the erythropoietic compartment, but may be compensated by STAT1 at other times.
Our data shows that EPO stimulates the tyrosine phosphorylation of STAT1 and STAT3 in a Ba/F3 cell line expressing an EPO-R deletion mutant devoid of tyrosines. In contrast, a recent paper suggests that in UT-7 cells expressing G-CSF-R/ EPO-R chimeras, STAT1 and STAT3 tyrosine phosphorylation is mediated by Tyr 432 of the human EPO-R (corresponding to murine EPO-R Tyr 431 ) (55). It is presently unclear whether these differences from our data represent variations in EPOdependent tyrosine kinase activation in different cell lines or inconsistencies in the phosphorylation of chimeric cytokine receptors examined in the Kirito et al. study (55).
The requirement of identical EPO-R domains for tyrosine and serine phosphorylation suggests that the phosphorylation events may be functionally linked. However, only in certain contexts has tyrosine phosphorylation been associated with serine phosphorylation. For example, simultaneous tyrosine and serine phosphorylation bas been observed on expression of oncogenic RhoA mutants (56), and STAT serine phosphorylation does not occur in JAK2-deficient cells (57). In contrast, serine STAT1 phosphorylation is not associated with detectable tyrosine phosphorylation in lipopolysaccharide-treated macrophages (58), and Chung et al. (39) reported that serine phosphorylation negatively modulates tyrosine phosphorylation. In all cases rigorously examined, tyrosine phosphorylation is not a prerequisite for STAT serine phosphorylation (59) even when activation of JAK2 is required for both phosphorylation events (57,60). Together, these results indicate that both JAK2dependent and -independent pathways can contribute to STAT serine phosphorylation in different contexts.
Despite our evidence that STAT1/3 tyrosine and serine phosphorylation requires JAK2 activation, full STAT-induced gene expression appears to require additional regions of the EPO-R (Fig. 12). Compared with the EPO-R Ϫ99 mutant, EPO-R Ϫ99,F1 mutants have impaired EPO-induced STAT-dependent gene expression, suggesting an important role of Tyr 343 in STAT regulation. These results indicate that STAT tyrosine and serine phosphorylation are not sufficient for maximal activity, and point to other mechanisms of regulation. Recruitment of co-activators including p300/CBP (61-63), glucocorticoid receptor (45,64), NMI (65), as well as inhibitors of STAT activity such as SOCS (66) or PIAS (67, 68) may add to the complexity of STAT regulation.
Since phosphorylated serine residues in STAT1, STAT3, and STAT5A/B lie within potential MAPK phosphorylation motifs, we examined the role of the p38, ERK, and JNK in this process. p38 and ERK targeting of STAT1 and STAT3 is suggested by their activation via the same region of the receptor. Specific inhibition of p38 abolishes STAT1 serine phosphorylation, but not phosphorylation of other STAT molecules. In contrast, ERK pathway inhibitors block STAT1 and STAT3 serine phosphorylation. STAT5A and STAT5B serine phosphorylation does not correlate with activation of any of the MAPKs tested, and none of the inhibitors affected STAT5A/B serine phosphorylation.
As STAT1 and STAT3 share similar sequences surrounding the serine phosphorylation site, the differences in STAT kinase pathways suggests there may be additional determinants of STAT kinase specificity. One possibility is sequences distant from the phosphorylation site may interact with signaling molecules and confer specificity. It would be interesting to examine STAT1/3 chimeras to localize determinant regions. STAT5A and STAT5B serine phosphorylation occurs via the same portion of the EPO-R and are both insensitive to the MAPK pathway inhibitors. Thus, we cannot exclude the possibility that serine phosphorylation of STAT5A and STAT5B molecules are regulated identically.
Our results linking JAK2 and ERK/p38 activity are consistent with previous reports in interferon-␥ (69), interleukin 2 (70), erythropoietin (70), and growth hormone signaling (71,72). We have shown that treatment of Ba/F3-EPO-R with the JAK inhibitor, AG490, blocks ERK and p38-dependent gene expression (data not shown). The requirement of both ERK and p38 activation for STAT1 serine phosphorylation suggests that these MAPKs may not directly phosphorylate the STAT molecules, but rather share a common effector. This possibility is consistent with the relatively inefficient phosphorylation of STAT1 by p38 (57,58). Alternatively, ERK and p38 may lie on the same pathway, as has been described previously (40,73).
The physiological role of STAT serine phosphorylation is unclear. STAT serine phosphorylation may prime the cell for subsequent transcriptional activity, as several reports indicate a requirement for serine phosphorylation in maximal cytokinestimulated transcriptional activity (35,60,74). Serine phosphorylation of STAT3 may also decrease its transcriptional response (75,76). Mutation of Ser 727 of STAT1 differentially affects basal and induced expression of target genes (59), although it is unknown if other STAT molecules behave similarly.
Erythropoietin-induced STAT activation can have both positive and negative effects on erythroid differentiation (77)(78)(79)(80), proliferation (13,14), and apoptosis (81). MAPKs may also have roles in these physiological processes. In SKT6 murine erythroleukemia cells, SAPK and p38 are required for stress-and EPO-induced differentiation, whereas inhibition of MEK stimulates stress-induced apoptotic cell death (36,37,43). Moreover, induced activation of p38 is sufficient for erythroid differentiation. Thus, STAT serine phosphorylation may be one mechanism by which MAPKs impinge on erythropoietin-induced physiological responses.