Roles of phosphatidylinositol 3-kinase in interferon-gamma-dependent phosphorylation of STAT1 on serine 727 and activation of gene expression.

STAT1 must be phosphorylated on serine 727 to be fully active in transcription. We show that phosphatidylinositol 3-kinase (PI3K) and its effector kinase Akt play an important role in the serine phosphorylation of STAT1 and in the activation of gene expression in response to interferon-gamma (IFN gamma). IFN gamma activates PI3K as well as Akt in a variety of cell lines. Specific inhibition of PI3K abrogates IFN gamma-induced, but not interleukin-1- or tumor necrosis factor-alpha-induced, phosphorylation of STAT1 on serine and reduces STAT1-dependent transcription and gene expression by approximately 7-fold. Constitutively active forms of PI3K or Akt activate and their dominant-negative derivatives inhibit STAT1-driven transactivation in response to IFN gamma. In addition to PI3K and Akt, JAK1, JAK2, and the tyrosine 440 STAT1 docking residue of IFNGR1 are required for STAT1 to be phosphorylated on serine. Taken together, these results suggest that the following events lead to the activation of STAT1 upon IFN gamma stimulation: 1) PI3K and Akt are activated by the occupied receptor and Tyr-440 is phosphorylated by the activated JAKs; 2) STAT1 docks to Tyr-440; and 3) Tyr-701 is phosphorylated by the JAKs and Ser-727 is phosphorylated by a kinase downstream of Akt.

Full activation of STAT1 1 by IFN␥ requires two distinct phosphorylation events. Receptor-mediated phosphorylation of STAT1 on tyrosine 701 is required for STAT1 homodimers to form and subsequently to bind to the promoters of IFN␥-responsive genes through ␥-activated sequence (GAS) elements (reviewed in Refs. 1 and 2). The mechanistic basis for STAT1 tyrosine phosphorylation has been established for some time. JAK1 and JAK2, constitutively bound to specific cytoplasmic domains of the IFNGR1 and IFNGR2 subunits, phosphorylate each other when IFN␥ binds and the receptor subunits aggregate. The activated JAKs then phosphorylate tyrosine 440 of IFNGR1, creating a docking site that recruits STAT1 to the receptor. STAT1 binds to phosphorylated Tyr-440 through its SH2 domain, allowing its Tyr-701 residue to be phosphorylated by the JAKs. STAT1 then dissociates from the receptor, dimerizes through reciprocal SH2-phosphotyrosine interactions, and binds to IFN␥-inducible promoters.
The mechanisms through which STAT1 is phosphorylated on serine are not yet well understood. Serine 727 lies within the mitogen-activated protein kinase (MAPK) consensus motif PX n (S/T)P, where S/T is serine or threonine and n is 1 or 2, suggesting that MAPK family members may be STAT1 serine kinases. Through the use of knockout cells and dominantnegative and chemical inhibitors, several kinases have been implicated directly or indirectly (reviewed in Ref. 11). Phosphorylation of STAT1 on serine in response to IFN␥ has been shown to involve JAK2 (4), proline-rich tyrosine kinase 2 (Pyk2), potentially through its activation of ERK2 (12), p38 MAPK (6), and double-stranded RNA-activated protein kinase (PKR) (13).
Phosphatidylinositol 3-kinase (PI3K) is a member of a subfamily of lipid kinases implicated in many physiological processes, including regulation of cell growth, proliferation, survival and differentiation, vesicle trafficking, glucose transport, platelet function, and cytoskeletal remodeling (reviewed in Refs. 14 and 15). PI3K is activated by phosphorylation on tyrosine residues in response to many growth factors and cytokines by receptors with intrinsic tyrosine kinase activity or by receptor-associated tyrosine kinases. A heterodimer comprised of one of three catalytic isoforms and one of seven adaptor/regulatory proteins, PI3K, once active, catalyzes the addition of a phosphate moiety specifically to the 3Ј-OH position of the inositol ring of phosphatidylinositols (reviewed in Refs. 14 and 15). The resulting 3-phosphorylated phosphatidylinositols serve as secondary messengers to activate many downstream signaling targets, initiating the physiological effects of PI3K. One of the best characterized PI3K effectors is the serinethreonine protein kinase Akt(PKB), whose activation following growth factor or cytokine stimulation is directly dependent on PI3K-derived phosphorylated phosphatidylinositols (reviewed in Refs. 16 -18).
We now find that the phosphorylation of STAT1 on serine in response to IFN␥ requires the activation of PI3K and Akt. Our results reveal that PI3K functions in a novel pathway that plays an important role in signaling and in the activation of gene expression in response to IFN␥ and defines receptorassociated functions that are additionally required for the phosphorylation of STAT1 on Ser-727.
Electrophoretic Mobility Shift Assays-Cells at 80% confluence in 100-mm dishes were pretreated with either solvent (methanol) or LY for 30 min, followed by stimulation with IFN␥ for 20 min. Cells were washed once with phosphate-buffered saline, and cell pellets were lysed for 20 min at 4°C in 100 l of 0.5% Nonidet P-40 buffer containing 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 10% glycerol, 0.1 mM EDTA, 25 mM sodium fluoride, 2 mM sodium orthovanadate, 1 mM dithiothreitol, 0.4 mM phenylmethanesulfonyl fluoride, 3 g/ml aprotinin, 2 g/ml pepstatin, and 1 g/ml leupeptin. Cellular and nuclear debris was pelleted by centrifugation at 16,000 ϫ g at 4°C for 2 min. The high affinity STAT binding site, c-sis inducible element (m67) (24), was end-labeled with polynucleotide kinase (Roche Diagnostics) and [␥-32 P]ATP (Amersham Pharmacia Biotech) and used as a probe. The binding reaction was carried out at room temperature for 20 min with 20 g of whole cell extract, 2 g of poly(dI⅐dC) (Amersham Pharmacia Biotech), and ϳ20,000 cpm of probe in a total volume of 20 l containing 20 mM HEPES, pH 7.0, 10 mM KCl, 0.1% Nonidet P-40, 0.5 mM dithiothreitol, 0.25 mM phenylmethanesulfonyl fluoride, and 10% glycerol. The DNA-STAT1 complexes were separated on 5% polyacrylamide gels by electrophoresis in 0.2ϫ Tris borate/EDTA buffer. The gels were dried, and the labeled complexes were visualized by autoradiography.
Transfection and Reporter Assays-For luciferase reporter assays, cells at 60 -70% confluence, seeded the previous day in 150-mm plates, FIG. 1. Inhibition of PI3K activity with LY294,002 results in specific inhibition of STAT1 Ser-727 phosphorylation in response to IFN␥. A, LY inhibits the IFN␥-induced phosphorylation of STAT1 on Ser-727. T98G cells were pretreated with solvent (methanol) or LY for 30 min and then stimulated with IFN␥ for 10, 20, or 40 min. Cell lysates were separated by 10% SDS-PAGE, and Western analysis was carried out with anti-phospho-Ser-727-STAT1. SP represents singly phosphorylated species of STAT1, while DP represents doubly phosphorylated species. B, LY does not inhibit IFN␥-induced tyrosine phosphorylation or IL-1-or TNF-induced serine phosphorylation of STAT1. Following pretreatment with LY, T98G cells were stimulated as indicated for 20 min. Cell lysates were separated by 8% SDS-PAGE, and Western analysis was carried out sequentially, using anti-phospho-Ser-727-STAT1, anti-phospho-Tyr-701, or anti-Cterminal-STAT1, respectively. C, LY does not affect the IFN␥-induced DNA binding of STAT1. Electrophoretic mobility shift assays were performed with whole cell extracts from T98G cells stimulated with IFN␥ for 20 min following pretreatment with either solvent or LY. Cell lysates were incubated for 20 min with a 32 Plabeled probe corresponding to the c-sis inducible element GAS.
were transfected by the calcium phosphate method (25) with 10 g of 4ϫGBPGAS-luc (Stratagene) and 2 g of pSV2-␤-gal (to verify transfection efficiency). Eight hours after transfection, the cells were divided into four 100-mm plates. Cells were pretreated with methanol or LY for 30 min then stimulated with IFN␥ for 6 h, 40 h after transfection. For experiments involving PI3K expression plasmids, cells at 60 -70% confluence, seeded the previous day in 100-mm plates, were cotransfected with 2 g of 4ϫGBPGAS-luc along with 2, 4, or 8 g of membranetargeted, constitutively active versions of wild-type PI3K, wild-type Akt, kinase-dead PI3K, or dominant-negative Akt. Use of an additional control plasmid allowed equal amounts of DNA to be transfected. Luciferase and ␤-galactosidase activities were determined with the Promega luciferase assay and chemiluminescence reagents, respectively. Results are shown for one of at least three independent experiments.
Northern Analyses-Cells were stimulated for the indicated times. Total RNA was isolated with the TRIzol reagent (Invitrogen). Twenty micrograms of total RNA was denatured, separated by electrophoresis in a formaldehyde-1.2% agarose gel, and transferred to Hybond-N nylon membrane (Amersham Pharmacia Biotech). GBP-1 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNAs were detected with cDNAs labeled with [␣-32 P]dCTP (Amersham Pharmacia Biotech) by nick-translation using the DNA megaprime labeling system (Amersham Pharmacia Biotech) and visualized by autoradiography.
PI3K Activity-The assay was used as described by Royal and Park (26) and Sizemore et al. (27), with modifications. Cells were washed once with phosphate-buffered saline and lysed for 15 min at 4°C in 200 l of 1% Triton X-100 buffer. Cellular debris was removed by centrifugation at 16,000 ϫ g for 10 min at 4°C. Immunoprecipitations were performed with extracts containing equal amounts of protein, using anti-phosphotyrosine (PY20, Transduction Laboratories) at 4°C for 1.5 h followed by incubation for 1.5 h with protein G-Sepharose (Gamma G Bind, Amersham Pharmacia Biotech). The beads were washed three times with ice-cold lysis buffer and then incubated in kinase buffer containing 0.2 mg/ml phosphatidylinositol (4,5)P 2 (PI(4,5)P 2 ; Sigma) at 25°C for 10 min. The assay was initiated by adding 10 Ci [␥-32 P]ATP and 20 mM MgCl 2 and was terminated after incubation at 37°C for 15 min by adding chloroform, methanol, 12 M HCl (50:100:1). The products were extracted with chloroform, washed twice with methanol, 1 M HCl (1:1), spotted on a silica gel 60 thin layer chromatography plate (EM Science), and separated in chloroform, methanol, 28% ammonium hydroxide, water (86:76:10:14) for 45 min. The phosphorylated products were visualized by autoradiography.

Specific Inhibition of PI3K Abolishes IFN␥-induced Phosphorylation of STAT1 Ser-727 but Not Tyr-701 and Does Not Affect IL-1-or TNF-induced Phosphorylation of STAT1 Ser-727-
Pretreatment of T98G (Fig. 1A) or 2fTGH cells (data not shown) with the PI3K inhibitor LY kept serine phosphorylation of STAT1 at basal levels in response to IFN␥. The bands detected by the phosphoserine-specific STAT1 antibody are specific for residue 727, since the antibody did not detect any phosphorylated STAT1 protein in extracts from IFN␥-treated U3A-S727A cells, STAT1-null cells reconstituted with a STAT1 mutant protein containing a serine-to-alanine substitution (data not shown). Pretreatment with LY of T98G (Fig. 1B) or 2fTGH cells (data not shown) did not affect IFN␥-activated STAT1 tyrosine phosphorylation or the ability of STAT1 to bind to DNA (Fig.  1C). These findings indicate a potential role for PI3K in the serine phosphorylation of STAT1 in response to IFN␥. Because IL-1 (this study) and TNF (7) activate both STAT1 Ser-727 phosphorylation as well as PI3K (27), we tested the effect of LY on STAT1 serine phosphorylation in response to these cytokines. Interestingly, the PI3K inhibitor did not affect IL-1-or TNF-induced STAT1 serine phosphorylation (Fig. 1B).
Specific Inhibition of PI3K Reduces IFN␥-induced, STAT1dependent Activation of Transcription and Endogenous Gene Expression-Because the phosphorylation of STAT1 on serine contributes substantially to its activity as a transcription factor, we investigated the effect of inhibiting PI3K on STAT1driven reporter activity. Pretreatment with LY of T98G cells transfected transiently with the reporter construct 4ϫGBP-GAS-luc resulted in a ϳ7-fold reduction of IFN␥-induced lucif-erase activity compared with control cells (Fig. 2A). This decrease did not reflect toxicity because pretreatment with LY did not increase the fraction of cells stained with trypan blue (data not shown).
The involvement of PI3K in STAT1-dependent transactivation indicates that PI3K is likely to play a role in IFN␥-mediated, STAT1-dependent gene expression. Expression of the human guanylate-binding protein-1 (GBP-1) gene is induced by IFN␥. The gene contains overlapping GAS and ISRE sites that bind to STAT1 in association with IRF-9 (28). Consistent with an independent study (29), we find 2 that GBP-1 is not induced in STAT1-deficient U3A cells in response to IFN␥ and that U3A-S727A cells suffer a ϳ5-fold loss in IFN␥-induced GBP-1 expression. Similar to the result with U3A-S727A cells, pretreatment of T98G cells with LY decreased IFN␥-induced activation of GBP-1 by ϳ7-fold compared with controls (Fig. 2B). Similar observations were made in 2fTGH cells (data not shown). These results confirm that PI3K plays an important role in IFN␥-mediated, STAT1-dependent gene expression.
IFN␥ Activates the PI3K Pathway-Assays with phosphatidylinositol diphosphate as a substrate revealed that the lipid kinase activity of PI3K was rapidly and transiently induced (ϳ3-fold) upon IFN␥ treatment of T98G cells, peaking after about 3 min (Fig. 3A), a pattern of activation typical of other PI3K inducers (27,30,31). Similar results were observed in 2fTGH cells (data not shown). The constitutive activity of PI3K observed in untreated cells is most likely due to low levels of 2 H. Nguyen and G. R. Stark, unpublished observations.

FIG. 2.
Inhibition of PI3K activity with LY294,002 dramatically reduces STAT1-driven transcription. A, LY attenuates STAT1 transcriptional activity in a transfection assay. T98G cells were transfected transiently with the GBP4ϫGAS-luciferase construct. Twentyfour hours later, the cells were divided into four plates. Forty hours after transfection, the cells were incubated for 30 min with either solvent or LY and then treated with IFN␥ for 6 h. Cell extracts were then assayed for luciferase activity. B, LY inhibits IFN␥-induced GBP-1 gene expression. Following pretreatment for 30 min with LY, T98G cells were stimulated with IFN␥ for 1, 2, or 4 h. Total RNA was analyzed by the Northern procedure using a GBP-1 cDNA probe. The transfer was reprobed with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA to control for sample loading. growth factors in residual serum from the culture medium. Pretreatment with LY did not significantly reduce the basal activity of PI3K (data not shown) but did reduce the maximum induced activity to about the steady-state level found in untreated cells (Fig. 3A). Akt, a major downstream effector of PI3K, requires the lipid kinase activity of PI3K to be phosphorylated and activated in response to growth factor or cytokine stimulation. Akt was also activated by IFN␥ in a variety of cell lines, as measured by Western analysis of serine-phosphorylated Akt (Fig. 3, B-F).
The PI3K Pathway Regulates STAT1-driven Transcription-To corroborate our observations involving the PI3K inhibitor, we also analyzed the effects of cotransfecting constitutively active, membrane-targeted forms of PI3K or Akt (kind gifts of N. Grammatikakis; described in Refs. 32 and 33) on the STAT1-dependent activation of 4ϫGBPGAS-luc. Increasing concentrations of constitutively active PI3K enhanced IFN␥induced activation by ϳ4-fold (Fig. 4A). Constitutively active Akt also augmented activity by ϳ5-fold in response to IFN␥ (Fig. 4B). Conversely, cotransfection of increasing amounts of a membrane-targeted, kinase-dead form of PI3K reduced the activation of 4ϫGBPGAS-luc in response to IFN␥ by ϳ2-fold (Fig. 4C). Similarly, a dominant-negative derivative of Akt (Akt-CAAX; where A is an aliphatic residue) reduced IFN␥induced reporter activity by ϳ7-fold (Fig. 4D). These findings substantiate the role of the PI3K pathway in STAT1-dependent transcription in response to IFN␥.
IFN␥-dependent Induction of STAT1 Ser-727 Phosphorylation, but Not Activation of PI3K or ERK, Is Abrogated in JAK1deficient Cells-To examine the receptor-associated components involved in IFN␥-induced STAT1 serine phosphorylation, we analyzed the ability of JAK1-deficient U4A cells to activate STAT1 and PI3K, as well as ERK1(p42)/ERK2(p44) and p38MAPK, other kinases previously implicated in STAT1 Ser-727 phosphorylation in response to IFN␥ (Fig. 5). As expected, U4A cells did not phosphorylate STAT1 on Tyr-701 in response to IFN␥ treatment (Fig. 5A). Serine phosphorylation of STAT1 was not induced in U4A cells stimulated with IFN␥, but was in IFN␥-treated control U4A cells reconstituted with JAK1 (Fig. 5B), as well as in TNF-treated U4A cells (data not shown). Interestingly, the lack of JAK1 did not affect the acti-vation of PI3K (data not shown), Akt (Fig. 5C), or ERK (Fig.  5D). The latter observation is consistent with the report of Hu et al. (34). p38 was not phosphorylated in response to IFN␥ but was activated in response to TNF in the 2fTGH parents of U4A (data not shown). These results show that, in addition to and independently of PI3K activation, JAK1 is required for STAT1 serine phosphorylation in response to IFN␥.
IFN␥-dependent Induction of STAT1 Ser-727 Phosphorylation, but Not Activation of PI3K or ERK, Is Abrogated in JAK2deficient Cells-As expected, STAT1 was not phosphorylated on Tyr-701 in JAK2-deficient ␥2A cells in response to IFN␥ (Fig. 6A). Consistent with previous findings (4), STAT1 serine phosphorylation in response to IFN␥ was also not observed in ␥2A cells, and this defect was reversed upon reconstitution with JAK2 (Fig. 6B). Similar to observations in JAK1-null U4A cells, the activation of Akt in response to IFN␥ was not affected by the absence of JAK2 (Fig. 6C). The phosphorylation of ERK appeared to be deficient in ␥2A cells compared with parental 2C4 cells; however, ERK activation was not restored upon reconstitution of the cells with JAK2 (Fig. 6D). ERK was activated by IFN␥ in an independent JAK2-deficient clone (B9; Fig.  6E) as well as in JAK2-null MEFs (Fig. 6F). Therefore, the defect in ERK activation is peculiar to the ␥2A clone, and ERK activation in response to IFN␥ does occur in the absence of JAK2 in other cells. These findings show that JAK2 is necessary, in addition to and independently of PI3K, for the induction of STAT1 serine phosphorylation in response to IFN␥.
The Y440F Mutation of IFNGR1 Prevents STAT1 Ser-727 Phosphorylation but Not PI3K or MAPK Activation in Response to IFN␥-JAK1 and JAK2 phosphorylate IFNGR1 Tyr-440 to provide a docking site for STAT1, facilitating its subsequent tyrosine phosphorylation in response to IFN␥. Because JAK1 and JAK2 are both required for STAT1 serine phosphorylation, we tested the role of IFNGR1 Tyr-440. Activation of STAT1 and other responses to human (h)IFN␥ were analyzed in SCC16-5 murine fibroblast cells stably expressing either wild-type human IFNGR1 (hGR) or the Y440F mutant form (Fig. 7). STAT1 was not phosphorylated on Tyr-701 in Y440F cells in response to hIFN␥ (Fig. 7A). Interestingly, the induction of serine phosphorylation of STAT1 in Y440F cells was attenuated significantly (ϳ3-fold) in response to hIFN␥ (Fig. 7B) but not in FIG. 3. IFN␥ stimulates PI3K activity. A, T98G cells were serum-starved for 18 -24 h before treatment to reduce constitutive levels of PI3K. Cell lysates were immunoprecipitated with anti-phosphotyrosine, and the immunoprecipitates were assayed for PI3K activity. PI3K phosphorylates phosphatidylinositol (4,5)P 2 to phosphatidylinositol (3,4,5)P 3 (PI(3,4,5)P 3 ). The sample labeled ϪAb was processed in the absence of anti-phosphotyrosine. The sample labeled ϩLY was pretreated with LY for 30 min before stimulation with IFN␥. B-F, the PI3K effector Akt is activated by IFN␥. T98G (B), 2fTGH (C), HeLa (D), primary MEFs (E), or immortalized MEFs (F) were treated with IFN␥ for 5, 10, or 20 min. Cell lysates were separated by 10% SDS-PAGE, and Western analysis was performed sequentially using antiphospho-Ser-473 or anti-Akt, respectively. response to murine IFN␥, which activates the wild-type murine receptor in these cells (data not shown). The IFN␥-dependent activation of Akt (Fig. 7C), ERK (Fig. 7D), or p38 (Fig. 7E) by IFN␥ was not affected in Y440F cells. Taken together, these results reveal that in addition to and independently of PI3K STAT1 serine phosphorylation requires tyrosine phosphorylation of IFNGR1 Tyr-440, mediated by JAK1 and JAK2. DISCUSSION PI3K and Akt Are Important for STAT1 Serine Phosphorylation in Response to IFN␥-The PI3K effector Akt has been implicated, either directly or indirectly, in the phosphorylation and subsequent regulation of several transcription factors, including the forkhead factors (35)(36)(37)(38)(39), NF-B (27,40), and E2F (41,42). The role of PI3K in the serine phosphorylation of STAT1, rather than in its tyrosine phosphorylation and subsequent binding to DNA in response to IFN␥, very much resembles the role of PI3K in the IL-1-mediated stimulation of NF-B-dependent transactivation, which is also independent of DNA binding, through serine phosphorylation of the transactivating p65 subunit of NF-B (27).
The regulation of STAT1 by a PI3K effector has been found recently in Dictyostelium. DdSTATa, the Dictyostelium homolog of mammalian STATs, is phosphorylated on serine by GskA, the Dictyostelium homolog of glycogen synthase kinase-3 (43), a protein kinase that is phosphorylated and inactivated by Akt following treatment of mammalian cells with insulin (44). Extracellular cAMP induces the tyrosine phosphorylation and subsequent import of DdSTATa into the nucleus as a prerequisite for the transition from a unicellular amoeba to a multicellular mound. GskA, required for establishing the correct cell-type pattern in the mound, is concomitantly activated by cAMP and then phosphorylates DdSTATa on serine, enhancing its nuclear export (43). There are several Dictyostelium homologues of Akt, some of which are activated by cAMP through a G-protein or PI3K-dependent pathway and one of which plays a substantial role in multicellular development (45), although no association between Dictyostelium Akt and DdSTATa has yet been reported. These findings suggest that the involvement of PI3K signaling in the regulation of STAT1 activity may be evolutionarily conserved.
Is Akt the STAT1 serine kinase? Peptide substrates that are effectively phosphorylated by Akt contain the minimum amino acid sequence RXRXX(S/T)*, where X is any amino acid and * is a bulky hydrophobic residue (Phe or Leu) (46). This motif is found in most of the sequences surrounding the phosphorylation sites in proposed Akt substrates so far (reviewed in Ref. 47). The residues near STAT1 Ser-727 do not conform to this motif, suggesting that Akt probably does not phosphorylate STAT1 directly on serine, although the possibility cannot be ruled out without a direct experiment. Alternatively, the STAT1 serine kinase may be an Akt target, such as the mammalian target of rapamycin, mTOR/FRAP, a serine kinase involved in regulating protein synthesis and the cell cycle (48,49).
Another scenario is that PI3K and Akt lie upstream of either ERK, p38, or both, because both were proposed previously to be involved in STAT1 serine phosphorylation in response to IFN␥ (6,12). Two unpublished observations 2 make this possibility unlikely. First, although p38 was activated in response to IL-1 or TNF, phosphorylation of p38 was undetectable in response to IFN␥ in T98G and 2fTGH cells, diminishing the possibility that this kinase could phosphorylate STAT1 on Ser-727 in response to IFN␥ in these cells. This finding does not necessarily eliminate the possibility that p38 is involved in IFN␥-me- FIG. 4. The PI3K pathway helps to regulate STAT1-driven transcription. T98G cells were cotransfected transiently with the GBP4ϫGAS-luciferase construct and increasing amounts of either constitutively active p110 (A), constitutively active Akt (B), a kinase-dead p110 derivative (C), or a dominant-negative Akt derivative (D). Twenty-four hours after transfection, the cells were divided into two plates each. Forty hours after transfection, the cells were treated with IFN␥ for 6 h. Cell extracts were assayed for luciferase activity. Fold change is the ratio of the luciferase activity of each IFN␥-treated sample to the activity of each untreated sample.
FIG. 5. JAK1 is required for IFN␥-induced serine and tyrosine phosphorylation of STAT1 but is dispensable for the activation of PI3K or ERK. 2fTGH, U4A, or U4A-JAK1 cells were serum-starved for 18 -24 h before treatment to reduce constitutive levels of Ser-727 phosphorylation. Cell extracts were separated by 10% SDS-PAGE, and Western analyses were carried out sequentially with anti-phospho-p42/ p44ERK, anti-phospho-Ser-473-Akt, anti-phospho-Ser-727-STAT1, and anti-phospho-Tyr-701-STAT1. The membranes were then stripped and analyzed sequentially with anti-p42/p44ERK, anti-Akt, and anti-Nterminal-STAT1 to control for loading. diated phosphorylation of STAT1 on serine in HeLa S3 cells, in which the role of p38 in the serine phosphorylation of STAT1 was originally demonstrated (6). Secondly, peak activation of ERK preceded peak Akt phosphorylation and was not inhibited by LY in T98G cells. Moreover, pretreatment of T98G cells with the MEK1 inhibitor PD98059 prevented the activation of ERK but did not inhibit the serine phosphorylation of STAT1 in response to IFN␥, indicating that ERK cannot be the PI3Kactivated STAT1 serine kinase. These observations are consistent with other studies that present evidence against the role of either p38 or ERK in STAT1 serine phosphorylation in response to IFN␥ in certain cell lines (4,7,29,50).
The Activation of PI3K Is a Novel Signal in Response to IFN␥; Similarities between the IFN␥ Receptor and Growth Factor Receptors-This report has shown that the PI3K pathway is activated by IFN␥, and PI3K is known to be activated by IFN␣/␤ as well (51,52). Activation of PI3K is required for induction of STAT1 serine phosphorylation in response to IFN␥ and for full activation of gene expression. We found that STAT1 was phosphorylated very weakly on Ser-727 in response to IFN␣ and that induced levels were insignificantly affected by LY in T98G cells. 2 These results complement previous studies that demonstrate that inhibition of PI3K by wortmannin does not affect the IFN␣-induced formation of ISGF3, a complex composed of STAT1, STAT2, and p48 transcription factors, or IFN␣-mediated gene transcription driven by ISRE, the DNA element to which ISGF3 binds (53). These findings are not surprising, since the transactivation domain of STAT1 is dispensable for ISGF3 transcriptional activity, which is provided primarily by STAT2 (54 -56) and is very different from the situation with IFN␥-induced, STAT1-dependent gene expression, which depends entirely on the ability of STAT1 dimers to activate transcription (54).
The activation of PI3K by IFNs presents a new perspective in understanding the full function of the IFN receptors. Evidence is accumulating that reveals the activation in response to IFNs of many proteins that are involved in growth factor-dependent signaling. MAPK and PI3K are activated by growth factors such as platelet-derived growth factor (reviewed in Ref. 57). In addition, both IFN␣ and IFN␥ activate vav-and Crk-dependent signaling pathways in hematopoietic cells (reviewed in Ref. 58), and these proteins are also activated in response to epidermal growth factor, colony-stimulating factor-1, and B-cell or T-cell activation (59 -68). Participants in pathways that are unique FIG. 6. JAK2 is required for IFN␥induced serine and tyrosine phosphorylation of STAT1 but is dispensable for the activation of PI3K or ERK. 2C4, ␥2A, ␥2A-JAK2, and B9 cells and JAK2Ϫ/ϪMEF were serum-starved for 18 -24 h before treatment to reduce constitutive levels of Ser-727 phosphorylation. Cell extracts were separated by 10% SDS-PAGE, and Western analyses were carried out sequentially with antiphospho-p42/p44ERK, anti-phospho-Ser-473-Akt, anti-phospho-Ser-727-STAT1, and anti-phospho-Tyr-701-STAT1. The membranes were then stripped and analyzed sequentially with anti-p42/p44ERK, anti-Akt, and anti-N-terminal-STAT1 to control for loading. FIG. 7. Phosphorylation of the STAT1 docking site of IFNGR1 is important for the induction of both serine and tyrosine phosphorylation of STAT1 in response to IFN␥ but is dispensable for the activation of PI3K or MAPK. SCC16-5 murine fibroblasts stably expressing wild-type (hGR) or mutant (Y440F) human IFNGR1 were serum-starved for 18 -24 h before treatment to reduce constitutive levels of STAT1 Ser-727 phosphorylation. Cells were separated by 10% SDS-PAGE, and Western analyses were carried out sequentially with anti-phospho-p38MAPK, anti-phospho-p42/p44ERK, anti-phospho-Ser-473-Akt, anti-phospho-Ser-727-STAT1, and anti-phospho-Tyr-701-STAT1. The membranes were then stripped and analyzed sequentially with anti-p38MAPK, anti-p42/p44ERK, anti-Akt, and anti-N-terminal-STAT1 to control for loading. in comparison to JAK-STAT signaling have now assumed increased importance with the discovery of substantial IFN␥induced, STAT1-independent gene expression, the mechanism of which remains to be elucidated. In the absence of STAT1, IFN␥ retains the ability to confer resistance to viral infection and to regulate the growth of primary bone marrow-derived macrophages. Furthermore, IFN␥ activates a surprisingly large number of genes in STAT1-deficient MEFs or bone marrow-derived macrophages, some of which exert important roles in immunomodulation and cell growth regulation. Interestingly, in the presence of STAT1, expression of some of these genes is actually suppressed (13,69,70). Expression of the c-myc gene by the novel pathway requires Raf1, since it is inhibited by the HSP90-specific inhibitor geldanamycin and by expression of a mutant form of p50 cdc37 that is unable to recruit HSP90 to the Raf1 complex (13). Our finding that PI3K is activated by IFN␥ makes it an attractive candidate for participation in the novel pathway. Interestingly, Akt interacts with the cdc37/HSP90 complex (18), an association that positively regulates Akt kinase activity in response to various stimuli (71). It is plausible that PI3K can activate signaling components independently of STAT1 to activate the novel pathway and subsequent gene expression and concurrently stimulate STAT1 serine phosphorylation to suppress induction of some of the same genes.
STAT1 Tyrosine and Serine Phosphorylation in Response to IFN␥ Requires the Same Receptor-associated Functions-It is interesting that the activation of PI3K and ERK in response to IFN␥ was not reduced in cells lacking either JAK1 or JAK2. Although we can conclude that STAT1 serine phosphorylation requires both JAK1 and JAK2, we cannot state that both JAKs are dispensable for the activation of PI3K or ERK. It is possible that either one of the JAKs is capable of catalyzing activation and that they can substitute for each other when one of them is missing. Analysis of a cell line in which the function of both JAKs are lacking would truly determine whether they play a role in the activation of PI3K or ERK.
Our observation that STAT1 serine phosphorylation requires the JAK-mediated phosphorylation of IFNGR1 Tyr-440 suggests that STAT1 must dock on the receptor in order to be phosphorylated not only on tyrosine but on serine as well. A relevant consideration is that Akt is activated at the plasma membrane (72)(73)(74), where it can act on its target substrates. To test this hypothesis, we analyzed IFN␥-mediated induction of STAT1 serine phosphorylation in STAT1-deficient cells reconstituted with STAT1 containing an arginine3lysine substitution at residue 602, which is critical for STAT1 interaction with the receptor via its SH2 domain (U3A-SH2), as well as in STAT1-deficient cells reconstituted with a STAT1 variant containing a tyrosine3phenylalanine substitution at residue 701, which is also critical for STAT1 interaction with the receptor, as well as for STAT1 dimerization (U3A-Y701F). Although STAT1 serine phosphorylation was not induced by IFN␥ in both of these cell lines, it was difficult to make a definite conclusion since the basal levels of serine-phosphorylated STAT1 were high. 2 However, a recent study also reported that in STAT1-deficient murine cells expressing the same STAT1 mutants, STAT1 was not phosphorylated on serine in response to IFN␥ (29). Several conclusions can be drawn from these observations. First, STAT1 may have to dock on the IFN␥ receptor to be phosphorylated on serine. Second, phosphorylation of IFNGR1 Tyr-440 may signal to a yet unidentified component that is required in addition to PI3K for STAT1 serine phosphorylation. Third, STAT1 may have to be dimeric in order to be recognized by its serine kinase, or dimerization and subsequent nuclear translocation of STAT1 may be required to move it to its serine kinase. In any case, our work reveals a two-step process leading to STAT1 serine phosphorylation in response to IFN␥: 1) PI3K and Akt are activated, leading to the activation of the STAT1 serine kinase; and 2) in order to be phosphorylated on serine, STAT1 has to dock to the receptor and perhaps also dimerize and translocate to the nucleus; a process initiated by JAK-mediated phosphorylation of IFNGR1 Tyr-440.