Stat1 (signal transducer and activator of transcription 1) regulates transcription in response to the type I interferons IFN‐α and IFN‐β, either in its dimerized form or as a subunit of the interferon‐stimulated gene factor 3 (Isgf3) complex (consisting of Stat1, Stat2 and interferon‐regulating factor 9). Full‐length Stat1‐α and the splice variant Stat1‐β, which lacks the carboxyl terminus and the Ser727 phosphorylation site, are found in all cell types. IFN‐induced phosphorylation of Stat1‐α on Ser727 occurs in the absence of the candidate kinase, protein kinase C‐δ. When expressed in Stat1‐deficient cells, Stat1‐β and a Stat1‐S727A mutant both restored the formation of Stat1 dimers and of the Isgf3 complex on treatment with IFN‐β. By contrast, only Stat1‐α restored the ability of IFN‐β to induce high levels of transcription from target genes of Stat1 dimers and Isgf3 and to induce an antiviral state. Our data suggest an important contribution of the Stat1 C terminus and its phosphorylation at Ser727 to the transcriptional activities of the Stat1 dimer and the Isgf3 complex.
The type I interferons IFN‐α and IFN‐β, which are cytokines with essential roles in innate viral immunity (Sen, 2001), induce expression of their target genes through the JAK (Janus kinase)/Stat (signal transducer and activator of transcription) signalling pathway (Darnell et al., 1994; O'Shea et al., 2002). Receptor‐associated JAK1 and tyrosine kinase 2 (Tyk2) kinases phosphorylate Stat1 and Stat2 on tyrosine residues, leading to the formation of Stat1 homodimers and Stat1/2 heterodimers. The Stat1 homodimer binds to promoters that contain an interferon‐γ‐activated site (GAS). The Stat1/2 hetero‐dimer associates with interferon‐regulating factor 9 (Irf9) to form the interferon‐stimulated gene factor 3 (Isgf3) complex, which binds to interferon‐stimulated response elements (ISREs) in target promoters.
Stat1‐β is a naturally occurring splice variant of Stat1‐α that lacks 38 carboxy‐terminal amino acids. Stat1‐β dimers bind DNA, but do not activate transcription. Therefore, the transactivating domain (TAD) of Stat1‐α is thought to reside in the C terminus of the protein (Shuai et al., 1993), which contains a phosphorylation site at Ser727. Phospho‐Ser727 increases the transcriptional activity of Stat1 dimers in response to type II IFN (IFN‐γ; Wen et al., 1995; Kovarik et al., 2001). Type I IFNs also activate a kinase that phosphorylates Stat1 Ser727, which was recently identified as Pkc‐δ (Uddin et al., 2002). This suggests that a serine‐phosphorylated Stat1 dimer may be required for the induction of gene expression. Several reports suggest that, unlike the Stat1 dimer, the Stat1 TAD and its phosphorylation are not essential for the transcriptional activity of Isgf3, and that the Stat2 TAD provides the Isgf3 transactivation function (Bromberg et al., 1996; Horvath & Darnell, 1996; Qureshi et al., 1996).
We have analysed the effects of expressing Stat1‐β and of mutating Ser727 to alanine on the expression of genes that are induced by type I IFNs and are targets of the following: Isgf3 (Mx), Stat1 dimers (Ifr1) or both of these (Gbp1). We found that the presence of the Stat1 TAD and its phosphorylation on Ser727 are necessary for the complete induction of all the genes analysed, as well as for the establishment of the antiviral state in response to type I IFNs.
Stat1 phosphorylation in Pkc‐δ‐deficient cells
Pkc‐δ is activated by IFN‐α and IFN‐β and catalyses Stat1 Ser727 phosphorylation in vitro (Uddin et al., 2002). We tested whether Pkc‐δ is essential for the phosphorylation on Ser727 that occurs after treatment with IFN‐β and, as controls, with IFN‐γ and lipopolysaccharide (Lps) (Fig. 1). When normalized to Stat1 expression levels, the differences in Stat1 phosphorylation on Ser727 observed in macrophages from wild‐type and Pkc‐δ‐deficient mice were not significant. The same results were obtained using fibroblasts (data not shown).
Transcriptional activity of Stat1‐β and Stat1‐S727A
The importance of the Stat1 C terminus and its phosphorylation at Ser727 was studied in reconstituted fibroblasts derived from Stat1‐deficient mice. Cells that express Stat1‐α or Stat1‐S727A have been described recently (Kovarik et al., 2001). Similar stable cell lines that express Stat1‐β were produced, and two clones (β1 and β8) were selected for further investigation, as the levels of Stat1‐β expression in these cells closely matched those reported for Stat1‐α and Stat1‐S727A in the two cell lines described recently (Fig. 2A; Kovarik et al., 2001).
Levels of Stat1 tyrosine phosphorylation were similar in cells that express Stat1‐α, Stat1‐β and Stat1‐S727A after treatment with IFN‐β for various durations. One of the Stat1‐β‐expressing clones (β8) showed a slightly more transient response (Fig. 2B). All of the variants of Stat1 that were expressed localized to the nucleus after IFN treatment, as visualized by immunofluorescence (data not shown). Similar amounts of the Isgf3 complex were formed in all lines after treatment with IFN‐β (Fig. 2C). Therefore, signalling from the type I IFN receptor was intact in all of the clonal lines that were used in subsequent experiments.
To study gene expression, Stat1−/− cells and cells that express Stat1‐α, Stat1‐β or Stat1‐S727A were treated with IFN‐β for 4 h or 6 h. Complementary DNAs were synthesized from total RNA, which was normalized by agarose gel electrophoresis and used for real‐time PCR analysis using TaqMan technology. Fig. 3 summarizes the effects of the absence of the Stat1 C terminus and of the mutation of its Ser727 phosphorylation site on the inducibility of the Mx1, Gbp1 and Irf1 genes after 4 h or 6 h of treatment with IFN‐β. Consistent with previous studies, none of the genes were induced in Stat1‐deficient cells. Cells that express Stat1‐β or Stat1‐S727A showed residual inducibility, but, for both, the reduction in induction of gene expression compared to cells that express Stat1‐α was ∼75–90%. Importantly, a requirement for the phosphorylation of the Stat1 C terminus was shown for the Isgf3 target gene, Mx1, as well as for the Stat1‐dimer target gene, Irf1, and for the Gbp1 gene, the promoter of which contains both a GAS and an ISRE sequence (Hug et al., 1988; Pine et al., 1994; Briken et al., 1995).
To rule out any effects of clonal variation, a second pair of clones, which expressed similar levels of Stat1 variants to each other, but higher levels of Stat1‐α and Stat1‐S727A than in the previous clones, was analysed and gave similar results (data not shown). Furthermore, a transient transfection assay was carried out to analyse the effects of Stat1‐α and Stat1‐S727A expression in a non‐clonal population of Stat1‐deficient cells. Comparable gene‐transfer efficiency was controlled for by using a co‐transfected β‐galactosidase expression vector. Half of each transfected population was treated with IFN‐β for 4 h and used for real‐time quantitative PCR analysis of the endogenous Mx1, Gbp1 and Irf1 genes. Compared to cells that express Stat1‐α, an ∼50% reduction in IFN‐β‐mediated induction of the Mx1 and Gbp1 genes was seen in cells that express Stat1‐S727A (Fig. 4). The reduced effect of the S727A mutant protein compared with that seen in the stable lines is probably due to high levels of expression of the Stat1 proteins during transient transfection, and has also been seen for the response to IFN‐γ (K.R. and T.D., unpublished data). Any induction of Irf1 in the transient transfection experiments was below detectable levels.
Antiviral response in cells that express mutant Stat1
To study the biological consequences of mutations in the Stat1 TAD, Stat1‐deficient fibroblasts and lines that express Stat1‐α, Stat1‐β or Stat1‐S727A were treated with serial dilutions of IFN‐β and were infected with vesicular stomatitis virus (VSV). The cytopathic effect (CPE) of viral infection was evaluated by staining surviving cells with crystal violet, followed by taking an absorbance measurement. As expected, cells that express Stat1‐α were protected from CPE even at low levels of IFN‐β and Stat1‐deficient cells were not protected from viral lysis at all. Stat1‐β‐ and Stat1‐S727A expressing cells showed a reduction in the ability to establish the antiviral state (Fig. 5). At high IFN‐β levels, a residual antiviral response was seen in cells that express Stat1‐S727A, whereas it was almost abolished in cells that express Stat1‐β.
We, and others, have recently reported serine phosphorylation of the Stat1 TAD during responses to type I IFNs (Kovarik et al., 1998; Goh et al., 1999). Here, we show that phosphorylation occurs in the absence of Pkc‐δ, a kinase that is reported to be stimulated by type I IFNs and that is able to phosphorylate Stat1 at Ser727 in vitro (Uddin et al., 2002). Our results do not rule out a function for Pkc‐δ in the phosphorylation of Stat1, but they do show that Pkc‐δ is not the only kinase able to do this in cells stimulated with type I IFN. Potential candidates for other enzymes that can carry out this function are other Pkc isoforms or calcium/calmodulin‐dependent protein kinase II, which is thought to phosphorylate Stat1 at Ser727 in response to IFN‐γ (Nair et al., 2002).
Previous reports have suggested that the Stat1 TAD and its phosphorylation are important for the function of Stat1 dimers (Shuai et al., 1993; Kovarik et al., 2001); however, their impact on Isgf3 activity has not been recognized previously. Consistent with our findings, serine‐phosphorylated Stat1 participates in Isgf3 formation (Goh et al., 1999). Phospho‐Ser727 greatly enhanced the ability of Stat1 to interact with the transcriptional machinery. One explanation for this might be an increased association of Ser727‐phosphorylated Stat1 with the minichromosome maintenance 5 (Mcm5) protein, the variation in expression of which during the cell cycle correlates with the ability of IFN‐γ to induce the expression of a Stat1 reporter gene (Zhang et al., 1998). An interaction of Mcm5 with the Isgf3 complex has not been tested; however, the data suggest that Mcm5 functions as a transcriptional coactivator in the context of the Stat1 dimer. Alternatively, or in addition, phosphorylation might influence the association of Stat1 complexes with nuclear chromatin, a possibility that is not ruled out by the apparent lack of requirement for Stat1 phosphorylation on Ser727 for its association with DNA in vitro (Wen et al., 1995).
The prevailing view in the literature is that Stat2 provides the essential TAD for the Isgf3 complex. This is based on genetic experiments in human fibrosarcoma cells made deficient for either Stat1 (U3A cells) or Stat2 (U6A cells) by chemical mutagenesis (Darnell et al., 1994). The importance of the Stat2 TAD for Isgf3 activity was established in experiments using U6A cells that express a Stat2 protein with a truncated C terminus (Qureshi et al., 1996), and was proven further using GAL4‐based reporter systems (Park et al., 1999; Paulson et al., 1999) and by the demonstration of coactivator recruitment (Paulson et al., 1999, 2002). Expression of Stat1‐β in U3A cells in the presence of wild‐type Stat2 had variable effects on IFN‐induced gene expression. Most genes analysed, notably Gbp1, were significantly affected by the absence of the Stat1 TAD (Muller et al., 1993). The latter study is therefore consistent with our results in showing a function of the Stat1 TAD in IFN‐type‐I‐induced transcription, but emphasizes that Isgf3 complexes that contain Stat1‐β retain some transcriptional activity.
Our study provides evidence for an important function of a phosphorylated Stat1 TAD in the establishment of the antiviral state in response to IFN‐β. U3A cells that express Stat1‐β or Stat1‐S727A have been reported to establish an antiviral state in response to treatment with saturating concentrations of type I IFNs (Horvath & Darnell, 1996). In our study, the already reduced ability of Stat1‐S727A to induce the antiviral state in the presence of saturating concentrations of IFN‐β decreased rapidly to even lower levels at lower concentrations of IFN‐β. As seen for the IFN‐γ response (Kovarik et al., 2001), discrepancies between results from reconstituted U3A cells and those from reconstituted fibroblasts from Stat1‐deficient mice might be due to the origin of the cells (human versus mouse; transformed versus immortalized), or to previously unnoticed effects of the chemical mutagenesis that preceded the selection of U3A cells. In the 11.1 cell line, which is like the U3A mutagenized progeny of 2fTGH human fibrosarcoma cells, Tyk2 deficiency abolishes responsiveness to IFN‐α (John et al., 1991; Velazquez et al., 1992). By contrast, cells derived from Tyk2‐deficient mice show reduced IFN‐α signalling, but establish an antiviral state comparable to that of wild‐type cells (Karaghiosoff et al., 2000).
In most cells, the amounts of Stat1‐α present exceed those of Stat1‐β, but the ratio of these proteins varies between cell types. Whether, and how, the splicing of alternative Stat1‐α or Stat1‐β mRNAs is regulated is still unknown. Our results suggest that the variable Stat1‐α : Stat1‐β ratio might pre‐set the level at which IFN‐α‐ and IFN‐β‐inducible genes are transcribed. A further implication of our study is that the extent to which Stat1 molecules are phosphorylated at Ser727 determines the transcriptional output in cells responding to type I IFNs. Because stress and inflammatory signals cause Stat1 phosphorylation on Ser727 (Decker & Kovarik, 2000), an incoming virus might increase the proportion of serine‐phosphorylated Stat1 that is available for a response to IFN. The potential function of the Ser727 phosphorylation site as an environmental sensor for the biological context of an IFN response will be addressed in future research.
Cells and cytokines.
3T3 fibroblasts from Stat1‐deficient mice (Durbin et al., 1996), either untransfected or stably transfected with Stat1 variants, were cultured in DMEM containing 10% FCS. IFN‐β (Calbiochem) was used at a concentration of 500 units ml−1. Bone marrow was isolated from femurs of wild‐type or isogenic Pkc‐δ‐deficient mice (Leitges et al., 2001). Macrophages were obtained by culturing bone marrow in L‐cell‐derived colony‐stimulating factor 1.
A Stat1‐β cDNA was inserted into pEF–Zeo, as described recently for Stat1‐α and Stat1‐S727A (Kovarik et al., 2001).
For transient transfection, Effectene reagent (Qiagen) was used in accordance with the manufacturer's instructions. Stable transfectants were isolated as described in Kovarik et al. (2001).
Antisera against the Stat1 C terminus and against phospho‐S727‐Stat1 have been described recently (Kovarik et al., 1998). Rabbit antiserum against Tyr701‐phosphorylated Stat1 and a monoclonal antibody against the Stat1 amino terminus were purchased from New England Biolabs and Transduction Laboratories, respectively. The monoclonal antibody against extracellular‐signal‐regulated protein kinases (ERKs) (pan‐ERK) was purchased from Transduction Laboratories.
Western blot analysis.
A protocol for this procedure has been described recently in Kovarik et al. (1998).
Electrophoretic mobility‐shift assay.
The conditions used for this assay have been described previously (Eilers et al., 1994); this was carried out using a double‐stranded oligonucleotide corresponding to the ISRE sequence from the interferon‐stimulated gene 5 promoter as a probe.
RNA preparation, complementary DNA synthesis and real‐time PCR.
Total RNA extraction and cDNA synthesis have been described in Kovarik et al. (2001). TaqMan real‐time PCR was carried out using a light cycler (Roche). For normalization, the housekeeping gene that encodes hypoxanthine phosphoribosyltransferase (Hprt) was amplified (Heid et al., 1996). TaqMan probes labelled with the reporter dye 6‐carboxy‐fluorescein (FAM) and the quencher dye 6‐carboxy‐tetramethyl‐rhodamine (TAMRA) were obtained from Metabion. The sequences of the probes and the forward (f) and reverse (r) primers were as follows: Hprt/FAM, 5′‐TGGGAGGCCATCACATTGTGGC‐3′; Hprt‐f, 5′‐ttgctcgagatgtcatgaagga‐3′; Hprt‐r,5′‐tgagagatcatctccaccaataactt‐3′; Irf1/Fam, 5′‐CAGTCTGAGTGGCAGCCGACACACA‐3′; Irf1‐f, 5′‐CCGAAGACCTTATGAAGCTCTTTG‐3′; Irf1‐r, 5′‐GCAAGTATCCCTTGCCATCG‐3′; Gbp1/Fam,5′‐AAAATCAAGAACATGCCTCCACCTCG‐3′; Gbp1‐f, 5′‐ATCATATCCCTTAAACTTCAGGAACAG‐3′; Gbp1‐r, 5′‐GTGGAAACAGGGTAGAGAGCTTTAGT‐3′; Mx1/Fam, 5′‐AGCCTACTACCAGGAGTGCAGACGGAA‐3′; Mx1‐f,5′‐GACTACCACTGAGATGACCCAGC‐3′; Mx1‐r, 5′‐ATTTCCTCCCCAAATGTTTTCA‐3′.
2 × 104 fibroblasts were seeded into microtitre wells containing serial dilutions of IFN‐β. After overnight incubation with the cytokine, the cells were infected with the Indiana strain of vesicular stomatitis virus (VSV) at an MOI (multiplicity of infection) of 0.1. Twenty‐four hours after infection, remaining viable cells were stained with a 0.2% solution of crystal violet in 20% methanol. After repeated washing, cell‐associated dye was dissolved in methanol and the absorbance was measured at 620 nm.
The authors thank M. Baccarini for critical reading of the manuscript. Support was obtained from the Austrian Science Foundation (FWF) through grant P‐15680 to T.D.
- Copyright © 2003 Nature Publishing Group