Recruitment of Stat1 to chromatin is required for interferon-induced serine phosphorylation of Stat1 transactivation domain

Abstract

The transcription factor Stat1 plays an essential role in responses to interferons (IFNs). Activation of Stat1 is achieved by phosphorylation on Y701 that is followed by nuclear accumulation. For full transcriptional activity and biological function Stat1 must also be phosphorylated on S727. The molecular mechanisms underlying the IFN-induced S727 phosphorylation are incompletely understood. Here, we show that both Stat1 Y701 phosphorylation and nuclear translocation are required for IFN-induced S727 phosphorylation. We further show that Stat1 mutants lacking the ability to stably associate with chromatin are poorly serine-phosphorylated in response to IFN-γ. The S727 phosphorylation of these mutants is restored on IFN-β treatment that induces the formation of the ISGF3 complex (Stat1/Stat2/Irf9) where Irf9 represents the main DNA binding subunit. These findings indicate that Stat1 needs to be assembled into chromatin-associated transcriptional complexes to become S727-phosphorylated and fully biologically active in response to IFNs. This control mechanism, which may be used by other Stat proteins as well, restricts the final activation step to the chromatin-tethered transcription factor.

The Stat (signal transducers and activators of transcription) proteins are major cytokine-activated transcription factors that play a vital role in the biology of the hematopoietic and immune systems (1). Triggering of the cytokine receptor causes Stat tyrosine phosphorylation by the receptor-associated Jak tyrosine kinases causing the Stat homo- or heterodimers to accumulate in the nucleus and bind DNA. In addition, several Stat proteins are serine-phosphorylated in the C-terminal transactivation domain. Generation of knockin mice bearing alanine instead of serine at position 727, the site of Stat1 and Stat3 serine phosphorylation, and in vivo reconstitution experiments using Stat4 mutated at the phosphorylation site S721 proved the importance of these modifications for the transcriptional activity and biological function (2–4). Stat1 is activated in response to type I and type II interferons (IFNs) by phosphorylation at both Y701 and S727. On type I IFN (IFN-α and IFN-β) stimulation Stat1 is assembled in the ISGF3 complex (Stat1, Stat2, and Irf9 heterotrimer) and, to a lesser extent, in Stat1 homodimers. The type II IFN-γ activates primarily Stat1 homodimers. Several studies revealed that Stat1 complexes are, to a considerable part, preassembled before IFN stimulation (5–7). Phosphorylation of Y701 triggers Stat1 to accumulate in the nucleus in an importin-α5-dependent manner (8, 9). Residues within the Stat1 N terminus as well as in the DNA binding domain were shown to be critical for the IFN-induced nuclear translocation (10–13). Stat1 can also shuttle between the cytoplasm and nucleus independently of IFN stimulation and Y701 phosphorylation (11). Both types of IFNs require S727-phosphorylated Stat1 for biological responses (14–16). The identity of the IFN-induced S727 kinase is not fully resolved. The reported candidate kinases are very diverse with regard to the substrate specificity, and their contribution to S727 phosphorylation in vivo has yet to be evaluated (16–19). S727 of Stat1 can be phosphorylated also independently of IFNs by p38 mitogen-activated protein kinase (MAPK) under conditions of cellular stress (20, 21). Whereas cellular stress causes only S727 phosphorylation, IFNs induce phosphorylation of both Y701 and S727. In fact, Y701 phosphorylation of Stat1 is an essential prerequisite for IFN-γ-induced S727 phosphorylation in mouse fibroblasts (15) indicating that IFN-γ- and p38 MAPK-induced S727 phosphorylations are mechanistically independent events with different biological consequences. The function of the stress-induced S727 phosphorylation is still not completely resolved (22), although several studies have revealed roles for serine- but not tyrosine-phosphorylated Stat1 in apoptosis (23, 24) and in SUMOylation of Stat1 (25).

Here, we demonstrate that S727 phosphorylation induced by IFNs requires Stat1 to be both Y701-phosphorylated and localized in the nucleus. We further show that Stat1 mutants that display a strongly reduced chromatin association and transcriptional activity (26) are inefficiently S727-phosphorylated in response to IFN-γ. However, IFN-β induces normal S727 phosphorylation of these Stat1 mutants that maintain their transcriptional activity through the assembly into the ISGF3 complex. In agreement with previous results the stress-induced p38 MAPK-dependent S727 phosphorylation is unaltered by mutations affecting nuclear accumulation or chromatin association. These results reveal an unexpectedly complex mechanism that ensures that the Stat1 transactivation domain is modified only after proper assembly of Stat1 into chromatin-associated transcriptional complexes.

Results

IFN-β- and IFN-γ-Induced S727 Phosphorylation of Stat1 Depends on Y701 Phosphorylation.

IFN-γ-induced Y701 phosphorylation of Stat1 reaches the maximal level already after 5 min of treatment, whereas the maximal S727 phosphorylation does not occur earlier than 20–30 min after cytokine stimulation (Fig. 1A). Nuclear accumulation of Stat1 peaks after 10 min of treatment and remains stable during the entire time of observation (Fig. 1B). Thus, both Y701 phosphorylation and nuclear accumulation precede maximal S727 phosphorylation by 15–20 min. In agreement with this, immunofluorescence microscopy revealed that S727-phosphorylated Stat1 is located predominantly in the nucleus of IFN-γ-treated cells (Fig. 1C). In control cells expressing solely the Stat1 S727A mutant, no pS727-S1 signal was detected in the nucleus of IFN-γ-treated cells, thus proving the specificity of the reagent [supporting information (SI) Fig. S2A]. We have previously shown that in mouse fibroblasts phosphorylation of Y701 is required for IFN-γ-induced S727 phosphorylation (15). By using Stat1(−/−) mouse embryo fibroblasts (MEFs) stably reconstituted with Stat1-Y701F mutant (S1-Y701F cells) we now show that the IFN-β-induced S727 phosphorylation also depends on Y701 phosphorylation (Fig. 1D). Similar to IFN-γ, the kinetics of IFN-β-induced S727 phosphorylation is delayed by ≈15 min after Y701 phosphorylation and nuclear accumulation (Fig. S1). These data suggest that the type I and type II IFN-induced signaling pathways causing S727 phosphorylation are mechanistically similar. In contrast, the stress-induced p38 MAPK-dependent S727 phosphorylation occurs independently of Y701 as shown by anisomycin treatment (Fig. 1D). Anisomycin and other cellular stresses (UV light, LPS, and TNFα) were previously shown to cause S727 phosphorylation in a p38 MAPK-dependent manner (20, 22, 27). IFNs do not activate p38 MAPK in mouse fibroblasts (Fig. S2B) and other cell types tested (e.g., mouse primary macrophages, HepG2 cells, 293HEK cells, HUVEC cells; data not shown). The dependence of IFN-induced S727 phosphorylation on previous Y701 phosphorylation is not consistent with one study showing that S727 phosphorylation occurred independently of Y701 in human U3A cells (28). U3A cells display a high constitutive S727 phosphorylation that is not further inducible by IFNs (28). In contrast, other studies demonstrated that IFNs are able to induce S727 phosphorylation in human cells (29, 30). To clarify this issue we stably expressed myc-tagged WT Stat1 or the Y701F mutant in 293HEK and HepG2 cells. These cells were stimulated with IFN-γ or anisomycin, and the exogenous Stat1 was immunoprecipitated by using anti-myc antibody. As shown in Fig. 1 E and F, the S727 phosphorylation of the myc-tagged Y701F mutant was not induced by IFN-γ, whereas that of the myc-tagged WT Stat1 occurred normally in both human cell lines. Both WT and Y701F Stat1 proteins displayed a comparable stress-induced S727 phosphorylation. These data demonstrate that S727 phosphorylation of Stat1 is regulated similarly in mouse and human cells.

Fig. 1.

Y701 is required for S727 phosphorylation induced by type I or type II IFNs. (A) S1-WT cells were treated with IFN-γ for indicated times and phosphorylation of S727 was examined by Western blot analysis of whole-cell extracts with pS727-S1 antibody. The membrane was reprobed with antibody to phosphorylated Y701 (pY701-S1) and Stat1 (S1). (B) Same cells as in A were stimulated with IFN-γ for indicated times and localization of Stat1 was examined by immunofluorescence by using a Stat1 antibody. (C) Localization of S727-phosphorylated Stat1 was examined in untreated or IFN-γ-treated S1-WT cells by double immunofluorescence by using antibodies to pS727-S1 and total Stat1 (S1). (D) S1-WT and S1-Y701F cells were treated for 30 min with anisomycin (an), IFN-γ, IFN-β, or left untreated. Western blot analysis of whole-cell extracts was performed with pS727-S1 antibody, or with Stat1 antibody for loading control. (E) 293HEK cells stably expressing myc-tagged Stat1-WT (293, myc-S1-WT; Upper) or Stat1-Y701F (293, myc-S1-Y701F; Lower) were treated for 30 min with IFN-γ or anisomycin (an). Myc-tagged proteins were immunoprecipitated from extracts by using myc antibody, and analyzed by Western blot analysis with pS727-S1, or S1 (loading control) antibodies. (F) HepG2 cells stably expressing myc-tagged Stat1-WT (HepG2, myc-S1-WT; Upper) or Stat1-Y701F (HepG2, myc-S1-Y701F; Lower) were treated and analyzed as in E.

IFN-γ Fails to Induce S727 Phosphorylation of Nuclear Import-Deficient Stat1, IFN-β Restores Both Nuclear Translocation and S727 Phosphorylation.

We reasoned that, in the IFN-induced S727 phosphorylation pathway, the Y701 phosphorylation serves primarily to enable Stat1 nuclear accumulation. This hypothesis is consistent with the kinetics of S727 phosphorylation that follows the Y701 phosphorylation and nuclear accumulation with a 15- to 20-min delay. To address the role of Stat1 subcellular localization we investigated S727 phosphorylation of Stat1 mutants that are deficient in nuclear import but display normal Y701 phosphorylation in response to IFN-γ. Stat1 and its mutants were fused to GFP and stably introduced into Stat1(−/−) fibroblasts because the regulation of S727 phosphorylation cannot be studied in transient transfections (15). GFP was fused to the N terminus of Stat1 to minimize a potential steric effect of GFP on the critical S727 residue that is located within the C-terminal transactivation domain. Two different nuclear import-deficient Stat1 mutants were used. The first mutant, GFP-S1-L407A, contained a point mutation changing L407 to alanine that is known to prevent binding to importin and, consequently, IFN-γ-induced nuclear accumulation (8). The second construct, GFP-S1-Δ27, contains a deletion of the first 27 N-terminal amino acids. We show here that these 27 aa are required for IFN-γ-induced nuclear translocation (Fig. 2A). All Stat1 constructs display similar IFN-γ-induced Y701 phosphorylation (Fig. S3A). GFP-S1-WT translocated into the nucleus on treatment with either IFN-γ or IFN-β, whereas the mutants GFP-S1-Δ27 and GFP-S1-L407A accumulated in the nucleus only after IFN-β-treatment (Fig. 2A). The IFN-β-induced translocation of the Stat1 mutants occurred presumably because of the complex formation with Stat2, as shown previously for the S1-L407A mutation (8). These data suggest that the mutations are not dominant because they can be complemented by a nuclear localization signal (NLS) of Stat2 in the Stat1:Stat2 heterodimer. Similar to the reported S1-L407A mutant, the GFP-S1-Δ27 mutant remained in the cytoplasm in cells pretreated with the nuclear export inhibitor Leptomycin B (data not shown) indicating that the mutation inhibited nuclear entry rather than increased the rate of nuclear export. Analogous to the L407A mutant, the S1-Δ27 mutant bound DNA, and therefore was not impaired in dimer formation as revealed by EMSA (Fig. S3B). Consistent with a defect in nuclear translocation the mutants did not activate transcription of the IFN-γ target gene Irf1 (Fig. S3C). To investigate the influence of Stat1 subcellular localization on S727 phosphorylation, cells were treated with IFN-γ, IFN-β, or anisomycin. Anisomycin (i.e., p38 MAPK)-mediated S727 phosphorylation served as a control for the integrity of the Stat1 mutants. As expected, GFP-S1-WT is phosphorylated on both S727 and Y701 in cells treated with IFN-γ or IFN-β, and anisomycin caused S727 phosphorylation only (Fig. 2B). However, the nuclear import-deficient Stat1 mutants GFP-S1-L407A and GFP-S1-Δ27 were not S727-phosphorylated in response to IFN-γ, despite phosphorylation on Y701 (Fig. 2C). The mutant Stat1 proteins displayed WT-like anisomycin-induced S727 phosphorylation. Importantly, the S727 phosphorylation of both nuclear import-deficient mutants was induced on treatment with IFN-β that restores nuclear accumulation. The essential role of Stat2 in IFN-β-induced S727 phosphorylation is further supported by experiments with Stat2(−/−) fibroblasts demonstrating lacking S727 phosphorylation in response to IFN-β but a clearly inducible phosphorylation in response to IFN-γ (Fig. S3D). Consistent with previous studies, Stat1 Y701 phosphorylation was not induced by IFN-β (31). These results revealed that (i) Y701 is necessary but not sufficient for IFN-induced S727 phosphorylation, (ii) Y701-phosphorylated Stat1 is not S727-phosphorylated in response to IFN-γ if Stat1 nuclear import is blocked, and (iii) S727 phosphorylation proceeds normally after IFN-β treatment when the nuclear import-deficient Stat1 molecules are carried into the nucleus by a Stat2-mediated piggy-back transport mechanism. We generated several clones for each Stat1 construct and obtained similar results. The expression levels of the GFP-Stat1 constructs were comparable in all cell lines but rather low compared with control MEFs (Fig. S4).

Fig. 2.

Nuclear import is required for IFN-induced S727 phosphorylation of Stat1. (A) Cells stably expressing GFP-S1-WT, GFP-S1-Δ27, or GFP-S1-L407 constructs were treated for 30 min with IFN-γ, IFN-β, or left untreated. Localization of Stat1 was monitored by GFP fluorescence. GFP-S1-WT (B) or GFP-S1-L407A and GFP-S1-Δ27 (C) cells were treated for 30 min with anisomycin (an), IFN-γ, or IFN-β. Phosphorylation of immunoprecipitated Stat1 was detected by Western blot analysis by using pS727-S1 and pY701-S1 antibodies. Membrane was reprobed with Stat1 antibody (S1).

These data indicate that nuclear accumulation of Stat1 is an essential prerequisite for IFN-induced S727 phosphorylation.

Stat1 Requires Both Y701 Phosphorylation and Nuclear Accumulation to Become S727-Phosphorylated in Response to IFN.

The requirement of Y701 phosphorylation for IFN-induced S727 phosphorylation can be explained in several ways. For example, only Y701-phosphorylated Stat1 may be the substrate of the IFN-induced S727 kinase. Alternatively, the Y701 phosphorylation may serve solely to translocate Stat1 into the nucleus where it becomes S727-phosphorylated by a nuclear serine kinase. This kinase may be IFN-inducible or constitutively active. Stat1 is known to shuttle between the cytoplasm and nucleus without Y701 phosphorylation. The nuclear entry of the non-tyrosine-phosphorylated Stat1 is importin-independent, thus it differs from the cytokine-induced nuclear translocation (11). The shuttling in unstimulated cells might cause Stat1 to become S727-phosphorylated by a constitutively active nuclear kinase hereby explaining the basal level of S727 phosphorylation observed in many cells (e.g., Fig. 1A, lane 1). To elucidate the role of Y701 phosphorylation and nuclear localization in IFN-induced S727 phosphorylation, we inserted the NLS of the SV40 large T protein into GFP-S1-WT and the tyrosine phosphorylation-deficient GFP-S1-Y701F between the GFP part and Stat1 (GFP-NLS-S1 and GFP-NLS-S1-Y701F). SV40 NLS is known to cause constitutive nuclear accumulation of Stat1 (32). Analysis of Stat1(−/−) cells stably transfected with the GFP-NLS-S1 or GFP-NLS-S1-Y701F constructs revealed that GFP-NLS-S1 became S727-phosphorylated in response to IFN-γ, whereas the S727 phosphorylation of the GFP-NLS-S1-Y701F mutant was not induced (Fig. 3). Anisomycin treatment caused S727 phosphorylation of both constructs. Immunofluorescence microscopy confirmed nuclear accumulation of both constructs regardless of treatment with IFN-γ (Fig. 3 C and D). Furthermore, S727-phosphorylated epitopes were present in the nucleus of both cell lines after treatment with anisomycin whereas IFN-γ increased the pS727 signal only in the GFP-NLS-S1 but not in GFP-NLS-S1-Y701F cells (Fig. 3 C and D). The GFP-NLS-S1 was Y701-phosphorylated in response to IFN-γ illustrating that this protein had access to the receptor-associated Jak tyrosine kinases at the cell surface. This finding is consistent with the reported nucleo-cytoplasmic shuttling of the SV40-NLS-modified Stat1 (8).

Fig. 3.

Nuclear Stat1 is not phosphorylated on S727 unless it becomes Y701-phosphorylated in response to IFN-γ. (A and B) Stat1 was immunoprecipitated from extracts of GFP-NLS-S1 or GFP-NLS-S1-Y701F cells, respectively, treated for 30 min with IFN-γ or anisomycin (an). Phosphorylation of Stat1 on S727 and Y701 was assayed by Western blot analysis by using antibodies to phosphorylated S727 (pS727-S1) and Y701 (pY701-S1). The membrane was reprobed with Stat1 antibody (S1). (C and D) Localization of GFP-NLS-S1 or GFP-NLS-S1-Y701F (respectively) and their S727-phosphorylated forms was investigated by immunofluorescence microscopy by using double staining with antibodies to pS727-S1 and total Stat1 (S1). Cells were treated with IFN-γ or anisomycin (an).

Our data show that a prolonged presence of Stat1 in the nucleus of untreated cells does not result in S727 phosphorylation suggesting that the basal S727 phosphorylation is not caused by Stat1 shuttling in unstimulated cells. The data also indicate that Y701 phosphorylation is not only required for Stat1 to translocate to the nucleus, but in addition it represents an essential modification needed for the subsequent S727 phosphorylation.

Stable Association of Stat1 with Chromatin Is Required for IFN-Induced S727 Phosphorylation.

Recent studies revealed that many kinases function as chromatin-associated enzymes that modulate different steps of gene transcription (33). Because our data on the mechanism of IFN-induced S727 phosphorylation are consistent with a chromatin-associated phosphorylation event we decided to analyze the S727 phosphorylation of Stat1 mutants that are lacking DNA binding activity or are deficient in chromatin recruitment. Mutations of amino acids that are in contact with DNA or are part of the Stat linker domain were shown to affect the ability of Stat1 to bind target DNA (26). Stat1 mutated at lysine 336 (K336A) completely lacks DNA binding activity whereas the N460A mutant and the linker domain double-mutant K544A/E545A display a highly increased off-rate that prevents stable association of these mutants with target promoters in vivo. Consequently, these mutants are devoid of transcriptional activity despite robust Y701 phosphorylation and nuclear accumulation in response to IFN-γ. The K544A/E545A was shown to support type I IFN-inducible transcription from IFN-stimulated response element (ISRE)-driven promoters indicating that within the ISGF3 complex the Irf9 subunit provides a sufficient DNA binding activity. We have stably introduced the K336A, K544A/E545A, and N460A mutants into Stat1(−/−) cells by using both GFP-tagged (with the same results as untagged proteins; data not shown) as well as untagged versions to take into account the recently published effects of GFP on the dynamics of nucleo-cytoplasmic trafficking of Stat1 (34). IFN-γ failed to induce S727 phosphorylation of the K336A mutant whereas IFN-β strongly stimulated this modification (Fig. 4A). This result was obtained for two different time points (30 and 60 min) indicating that the reduced IFN-γ-induced S727 phosphorylation is not caused merely by a delayed kinetics. Both Y701 phosphorylation and nuclear accumulation occurred normally (Fig. 4 A and B). The analysis of the transcriptional activity (Fig. 4C) revealed that the K336A mutant was transcriptionally inactive on the predominantly GAS-driven Irf1 and Tap1 genes in response to both IFN-γ and IFN-β. Gbp2 transcription (known to depend on both GAS and ISRE) was abolished on IFN-γ treatment and strongly reduced on IFN-β stimulation. K336A retained transcriptional activity on the ISRE-driven Mx2 gene in response to IFN-β (Fig. 4C). Consistently, only IFN-β was able to induce robust S727 phosphorylation. Chromatin immunoprecipitation assays (ChIPs) and quantitative ChIP (qChIP) confirmed that the K336A mutant was not recruited to the GAS in Irf1 promoter (Fig. 4D). The analysis of the off-rate mutant K544A/E545A revealed that, similarly to the K336A mutant, IFN-γ-induced S727 phosphorylation was strongly reduced compared with the IFN-β-stimulated S727 phosphorylation (Fig. 5A). Analogous to other Stat1 mutants used in this study, anisomycin strongly induced S727 phosphorylation. Immunofluorescence experiments confirmed nuclear accumulation of the K544A/E545A mutant after IFN-γ treatment (Fig. 5B). The recruitment of the K544A/E545A mutant to the Irf1 promoter was still detectable yet considerably less efficient than that of Stat1-WT (Fig. 5C). Similar data were obtained for the other off-rate mutant N460A that was efficiently S727-phosphorylated on IFN-β but not IFN-γ treatment, consistent with strongly reduced recruitment to the Irf1 promoter (Fig. S5C). The requirement of stable chromatin association for IFN-induced S727 phosphorylation is further demonstrated by a differential extraction of S727-phosphorylated molecules from isolated nuclei. A considerable fraction of S727-phosphorylated Stat1 is retained in the high-salt-extracted nuclei, thus resembling the behavior of the DNA-bound fractions of RNA polymerase II and Cdk9 (Fig. S5D). In conclusion, these data demonstrate that the final activation of the transactivation domain does not occur before the assembly of Stat1 into higher-order transcriptional complexes in the chromatin of IFN-inducible genes.

Fig. 4.

Stat1 deficient in DNA binding displays strongly reduced S727-phosphorylation in response to IFN-γ. (A) Stat1 was immunoprecipitated from extracts of S1-K336A cells treated with anisomycin (an), IFN-γ, or IFN-β for times indicated. Phosphorylation of Stat1 was assayed by Western blot analysis by using antibodies to phosphorylated S727 (pS727-S1) and Y701 (pY701-S1). The membrane was reprobed with Stat1 antibody (S1). (B) Localization of the S1-K336A mutant in cells treated for 30 min with IFN-γ as revealed by immunofluorescence using S1 antibody. (C) S1-WT and S1-K336A cells were treated with IFN-γ or IFN-β for 1 h (for Irf1 and Mx2) or 4 h (for Tap1 and Gbp2), and total RNA was isolated for qRT-PCR. The graphs show induction of mRNA of Irf1, Tap1, Ggb2, and Mx2. (D) S1-K336A mutant is not recruited to the Irf1 GAS on treatment with IFN-γ or IFN-β (for 15 or 30 min) as revealed by ChIP (Upper) or qChIP (Lower).

Fig. 5.

Stable association of Stat1 with chromatin is required for efficient IFN-induced S727 phosphorylation. (A) Stat1 was immunoprecipitated from extracts of S1-K544A/E545A cells treated for 30 min with anisomycin (an), IFN-γ, or IFN-β. Phosphorylation of Stat1 was assayed by Western blot analysis by using antibodies to phosphorylated S727 (pS727-S1) and Y701 (pY701-S1). The membrane was reprobed with Stat1 antibody (S1). (B) Localization of the S1-K544A/E545A linker domain mutant in cells treated for 30 min with IFN-γ as revealed by immunofluorescence by using S1 antibody. (C) Recruitment of the S1- K544A/E545A mutant to the Irf1 GAS is strongly reduced compared with S1-WT in cells treated with IFN-γ or IFN-β (for 15 or 30 min) as revealed by ChIP (Upper) or qChIP (Lower).

Discussion

In this work we have used a series of Stat1 mutants to investigate the role of intracellular localization, Y701 phosphorylation, and chromatin recruitment of Stat1 in the IFN-induced S727 phosphorylation. All constructs, except the S1-Δ27 mutant, are based on published work. In all features (i.e., Y701 phosphorylation and DNA binding) examined, the nuclear import-deficient S1-Δ27 mutant, which lacks 27 N-terminal amino acids, was indistinguishable from the reported S1-L407A mutant thus confirming studies on the involvement of the N terminus and the central region of Stat proteins in nuclear import (10, 35, 36). The step-by-step analysis of (i) the Y701 mutant, (ii) nuclear import-deficient mutants, (iii) constitutively nuclear Stat1 constructs and (iv) Stat1 with diminished DNA binding or recruitment to chromatin led us to the conclusion that Stat1 needs to be assembled into chromatin-bound transcriptional complexes to become S727-phosphorylated. The kinetics of S727 phosphorylation and nuclear translocation is, in all cells tested, consistent with this model. In addition, in two human cell lines that are routinely used in many laboratories the IFN-induced S727 phosphorylation requires Y701 phosphorylation. Thus, we believe that the proposed mechanism of S727 phosphorylation is generally applicable despite the above-mentioned report showing Y701-independent S727 phosphorylation in U3A cells (28). For the following reasons we assume that the lack of IFN-γ-induced S727 phosphorylation of the Stat1 mutants used in this study is not caused by a grossly aberrant conformation: (i) anisomycin (i.e., p38 MAPK) induces S727 phosphorylation of all mutants; (ii) the nuclear import-deficient mutants are phosphorylated on Y701 and bind DNA on stimulation with IFN-γ; (iii) IFN-β causes the nuclear import mutants to become localized in the nucleus and phosphorylated on S727; and (iv) the mutations affecting nuclear translocation are localized in two domains separated by 400 aa and are distant from the critical S727 residue; (v) despite nuclear accumulation, the Stat1 K336A, K544A/E545A, and N460A DNA binding mutants are not efficiently S727-phosphorylated on IFN-γ treatment.

Throughout this study we used IFN-β either to complement the nuclear import deficiency or to circumvent defects in chromatin recruitment. This approach is justified by our finding that Y701 phosphorylation is required for both type I and type II IFN-induced S727 phosphorylation. This result indicates a mechanistically similar nature of both pathways because the Y701F mutation prevents nuclear accumulation of Stat1 after type I and type II IFN stimulation, and precludes the structural rearrangement from antiparallel to parallel that is required for the generation of transcriptionally active Stat1:Stat1 homodimers and possibly the ISGF3 complex as well (37–40). Our study did not address the regulation of dephosphorylation of the S727 residue although the dephosphorylation may also be a compartmentalized process. However, as the K336A, K544A/E545A, or N460A mutants accumulate in the same compartment (i.e., the cell nucleus) after treatment with both types of IFNs, yet only type I IFN induces S727 phosphorylation, it seems unlikely that a phosphatase accounts for the observed differences.

This work revealed an unusually complex nature of IFN-induced S727 phosphorylation. The proposed mechanism ensures that fully transcriptionally competent Stat1 will form only after successful assembly into chromatin-bound complexes that also contain the S727 kinase. We speculate that S727 phosphorylation, on one hand, triggers the transcriptional machinery, and on the other hand, it may accelerate the disassembly of the transcriptional complexes once the mRNA synthesis has been initiated. Such a mechanism would enable a more rapid shuttling of Stat1 to the IFN receptor to initiate another activation cycle. Consistent with this hypothesis, a role of S727 phosphorylation in nuclear export has been proposed recently (41). Future experiments will clarify whether similar mechanisms control serine phosphorylation of other Stat proteins and how the reported IFN-induced Stat1 S727 kinases match with the mechanism proposed in this work.

Materials and Methods

Cell Culture.

Immortalized Stat1(−/−) MEFs and all of their derivatives, Stat2(−/−) and Stat1(+/+) MEFs, and human HepG2 and 293HEK cells were maintained in DMEM containing 10% FCS. Mouse bone marrow-derived macrophages were isolated as described in ref. 29. Stat1(−/−) cells reconstituted with Stat1-wild-type (S1-WT), Stat1-Y701F (S1-Y701F), or Stat1-S727A (S1-S727A) have been described (15). Cell lines stably expressing Stat1 constructs GFP-S1-WT, GFP-S1-L407A, GFP-S1-Δ27, GFP-NLS-S1, GFP-NLS-S1-Y701F, GFP-S1-K544A/E545A, S1-K544A/E545A, S1-K336A, and S1-N460A were generated by transfection of Stat1(−/−) MEFs by using Exgen500 (Fermentas) according to the manufacturer′s protocol. After transfection, cells were selected for two weeks with 100 μg/ml zeocin (Stratagene). GFP positive cells were isolated from the bulk cultures by using FACS Aria (BD Biosciences) followed by limiting dilution cloning. 293HEK and HepG2 cells stably expressing N-terminally myc-tagged constructs Stat1-WT or Stat1-Y701F inserted in the expression vector pCI-Neo (Promega) were obtained by transfection and selection with 200 μg/ml G418 (Invitrogen) for two weeks.

Plasmids.

To generate Stat1 constructs wild-type Stat1 or Stat1 mutants were inserted either into the pEF-Zeo or pEF-GFP expression vectors as described in SI Material and Methods.

Immunoprecipitation (IP), Western Blot Analysis, Electrophoretic Mobility Shift Assay.

A protocol for these procedures has been described (29). In brief, to prepare extracts cells were washed with cold PBS and lysed in buffer containing 10 mM Tris·HCl (pH 7.5), 50 mM NaCl, 30 mM NaPP_i, 50 mM NaF, 2 mM EDTA, 1% Triton X-100, and protease inhibitor mixture (Roche). Extracts were cleared by centrifugation at 15,000 rpm. Stat1 IPs were carried out with S1-C-terminal antibody (29), and protein A Sepharose (Amersham Biosciences). Phosphorylation of Stat1 was detected by using antibodies to pS727-S1 (29) and pY701-S1 (Cell Signaling Technology–New England BioLabs). For IP, Western blot analysis or EMSA experiments, cells were treated with 5 ng/ml mouse IFN-γ (kindly provided by G. Adolf, Boehringer Ingelheim), 100 U/ml IFN-β (Calbiochem–Merck) or 100 ng/ml anisomycin (Sigma). Infrared imaging system Odyssey (LI-COR) was used for detection in Western blot experiments. Fluorescence-labeled secondary antibodies IRDye800 and IRDye700 (Rockland) were used for infrared detection. For EMSA, radioactively labeled double-stranded oligonucleotide corresponding to the GAS element of the Irf1 gene (TCGATGATTTCCCCGAAATGA) was used (29).

Microscopy.

Cells were fixed with methanol, incubated with monoclonal Stat1 antibody (BD Transduction Laboratories) or rabbit pS727-S1 antibody, or both antibodies in the case of double labeling. After washing the samples were stained with anti-mouse Alexa488 or anti-rabbit Alexa584 secondary antibodies (Molecular Probes). GFP-Stat1 was visualized in cells fixed with 1.8% formaldehyde for 5 min. Samples were mounted by using 15% Mowiol (Sigma) and inspected by using a Zeiss Axioplan 2 microscope.

Chromatin Immunoprecipitation (ChIP).

ChIP was performed as described in SI Material and Methods.

Quantitation of Gene Expression by Quantitative RT-PCR (qRT-PCR).

qRT-PCR experiments were performed by using SYBR green for detection, and Irf1, Mx2, Tap1, Gbp2, and HPRT primers as described in SI Material and Methods.