Cells previously desensitized to type 1 interferons display different mechanisms of activation of stat-dependent gene expression from naïve cells.

Over the past decade, a wealth of knowledge has been obtained concerning the mechanisms by which interferons (IFNs) and other cytokines activate or down-regulate immediate early genes via the Jak/Stat pathway. In contrast, little information is available on interferon-activated gene expression in naïve cells compared with cells that have been desensitized and subsequently resensitized to the actions of these cytokines. In naïve cells, the ISG54 gene is activated via IFN beta-stimulated formation of ISGF3, a heterotrimeric DNA binding complex consisting of p48 (IRF9) and tyrosine-phosphorylated Stat1 and Stat2. In contrast, in previously desensitized cells IFN beta weakly stimulates the assembly of an ISGF3-like complex that lacks Stat1, even though ISG54 mRNA induction is the same as in naïve cells. The lack of Stat1 tyrosine phosphorylation and DNA binding is due to increased activity of a protein-tyrosine phosphatase. In cells that do not express the tyrosine phosphatase Tc-PTP, the rate of Stat1 dephosphorylation is the same in naïve and previously desensitized cells. These results implicate Tc-PTP in a novel role in the regulation of type 1 interferon-stimulated gene expression.

Over the past decade, a wealth of knowledge has been obtained concerning the mechanisms by which interferons (IFNs) and other cytokines activate or down-regulate immediate early genes via the Jak/Stat pathway. In contrast, little information is available on interferonactivated gene expression in naïve cells compared with cells that have been desensitized and subsequently resensitized to the actions of these cytokines. In naïve cells, the ISG54 gene is activated via IFN␤-stimulated formation of ISGF3, a heterotrimeric DNA binding complex consisting of p48 (IRF9) and tyrosine-phosphorylated Stat1 and Stat2. In contrast, in previously desensitized cells IFN␤ weakly stimulates the assembly of an ISGF3-like complex that lacks Stat1, even though ISG54 mRNA induction is the same as in naïve cells. The lack of Stat1 tyrosine phosphorylation and DNA binding is due to increased activity of a protein-tyrosine phosphatase. In cells that do not express the tyrosine phosphatase Tc-PTP, the rate of Stat1 dephosphorylation is the same in naïve and previously desensitized cells. These results implicate Tc-PTP in a novel role in the regulation of type 1 interferon-stimulated gene expression.
The antiviral, antiproliferative, and immunomodulatory activities of type 1 interferons (IFN␣/␤) 1 are controlled, in part, by a set of cellular genes that are rapidly induced upon binding of these cytokines to their specific cell surface receptors. IFN␣/ ␤-stimulated formation of the transcription factor complex ISGF3 mediates gene induction by binding to an interferonstimulated response element (ISRE) located within the promoter of interferon-stimulated genes (ISG) (1,2). ISGF3 is composed of three proteins Stat1, Stat2, and p48 (IRF9). Stat1 and Stat2 are covalently modified by tyrosine phosphorylation after exposure of cells to IFN␣/␤ and subsequently translocate to the nucleus where they interact with IRF9 (p48). The Jak1 and Tyk2 tyrosine kinases also are an integral component of these signaling cascades (3). They are activated by IFN␣/␤ and are responsible for the tyrosine phosphorylation of Stat1 and Stat2. The mechanism governing interferon-induced transcriptional responses has now been extended to include a broad network of cytokine-regulated signaling systems that use tyrosine phosphorylation of Stat proteins to activate transcription of early response genes.
In addition to an increase in our knowledge of the activation of the Jak/Stat pathway, many studies have described mechanisms by which this signaling cascade is down-regulated. Protein-tyrosine phosphatases have been implicated in the inactivation of both the autophosphorylated Jaks as well as the nuclear-localized activated Stats (4 -10). The SH2 domain-containing SOCS family of proteins (suppressors of cytokine signaling) are activated by a variety of cytokines including interferons and act as inhibitors of the Jak/Stat pathway. SOCS proteins bind to both tyrosine-phosphorylated receptors and Jaks, and mediate their ubiquitin-dependent degradation (11). In contrast to the SOCS proteins whose expression is up-regulated as a consequence of cytokine treatment, the PIAS (Protein Inhibitors of Activated STATs) proteins are constitutively expressed. This family of five nuclear proteins binds to tyrosine-phosphorylated Stat dimers and prevents them from binding DNA (12,13). They have an additional function as SUMO E3-ligases whose substrates include c-Jun and p53 (14,15).
One aspect of cytokine regulation of the Jak/Stat pathway that has received little attention, is whether there are any changes in the regulation of immediate early genes in cells that have been previously exposed to a cytokine, have become refractory, and subsequently become re-sensitized to the cytokine. Understanding this process is significant in that in many pathophysiological settings cytokines can be repeatedly released in large amounts in response to both inflammatory and infectious insults. It is also relevant in terms of the therapeutic use of cytokines where the frequency of treatment of chronic conditions such as multiple sclerosis with IFN␤, should be optimized to intervals where cellular targets are not desensitized to actions of the drug. In order to understand the cellular effects of repeated cytokine exposure in greater detail, we have examined the IFN␣/␤-stimulated expression of ISG54. This gene is well known to be activated through an ISRE-dependent mechanism by type 1 interferons. It also has the advantage that the half-life of the mRNA is relatively short, so that re-induction of transcription can be easily monitored. The results from these studies clearly demonstrate that compared with naïve cells, cells previously desensitized to the actions of IFN␤ show a dramatically altered response when re-exposed to this cytokine. Although previously desensitized cells display a robust induction of ISGs, this activation shows no detectable IFN␤-stimulated tyrosine phosphorylation of Stat1, and a reduced tyrosine phosphorylation of Stat2. The lack of Stat1 tyrosine phosphorylation is due to the tyrosine phosphatase, T-cell protein-tyrosine phosphatase (Tc-PTP).

MATERIALS AND METHODS
Cells and Culture-Human foreskin fibroblasts (HFFs), Tc-PTP Ϫ/Ϫ and wild-type mouse embryo fibroblasts obtained from matched littermates were all maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Invitrogen, Life Technologies, Inc.), and containing 2 mM L-glutamine, penicillin, and streptomycin (Invitrogen). Fresh human peripheral blood mononuclear cells (PBMC) were isolated as previously described (16) and maintained in RPMI 1640 medium containing with 10% fetal calf serum, and antibiotics. Both human IFN␤ (Avonex) and murine IFN␤ were obtained from Biogen Inc. Human and murine IFN␥ were provided by Genetech Inc.
Western Blot Analysis-Proteins in nuclear extracts, cytoplasmic extracts, or whole cell extracts were separated in SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membranes. The membranes were probed with rabbit polyclonal antibody against either STAT1, phospho-STAT1 (PY 701) (Cell Signaling Technology), STAT2, or phospho-STAT2 (PY 689) (Upstate Biotechnology). Following hybridization, the membrane was washed and incubated for 30 min with peroxidase-conjugated anti-rabbit IgG and subsequently developed by chemiluminescence using the ECL Western blotting system (Amersham Biosciences).
Chromatin Immunoprecipitation PCR (ChIP)-ChIP were preformed as described by Paulson et al. (21). The Stat2 antisera were purchased from Santa Cruz Biotechnology (C-20). Stat1 antibody was generated in rabbits as previously described (22).

RESULTS
To examine the effects of IFN␣/␤ on naïve and previously desensitized cells, we used primary human fibroblasts because the transcriptional activation and inactivation of type 1 interferon-stimulated immediate early genes has been well characterized in these cells (23,24). In these studies we have chosen the mRNA encoded by the ISG54 gene for analysis, because its expression is known to be regulated by IFN␣/␤-stimulated tyrosine phosphorylation of Stat1 and Stat2, and their subsequent binding to an ISRE in its promoter (25). Furthermore, the half-life of the ISG54 mRNA is short (see Fig. 1A). Cells exposed to IFN␤ for 2 h show a robust induction of ISG54 expression which, in the continued presence of IFN␤, remains elevated for 16 h (23,26). Incubation for an additional 6 h after removal of IFN␤ from the media results in the complete disappearance of ISG54 mRNA (Fig. 1A, lane 4). Importantly, reexposure of the same cells to IFN␤ for 2 h yields an induction of ISG54 that is similar to that of naïve cells (Fig. 1A, compare lanes 2 and 5). Incubation of both naïve and previously desensitized cells with actinomycin D and IFN␤ prevents the induction of ISG54 (Fig. 1B), indicating that the increase in ISG54 mRNA levels triggered by the second cytokine exposure results from a transcriptional re-induction rather than an IFN␤-mediated stabilization of previously transcribed mRNA. Nuclear extracts prepared from naïve fibroblasts incubated with IFN␤ for 30 min contain the transcription factor complex ISGF3 (composed of Stat1 and Stat2) or GRR (which contains only Stat1) (22) (Fig. 1C, lane 1 versus 2). Surprisingly, compared with naïve cells, the extent to which IFN␤ is capable of activating ISRE binding is very much diminished in cells incubated with IFN␤ for 16h followed by an additional 6 h without the cytokine, and subsequent re-stimulation with IFN␤ for 30 min (Fig. 1C, compare lane 2 with 5). Moreover, we have not been able to detect any formation of a GRR binding complex in extracts prepared from previously desensitized cells (Fig. 1C, lower panel). To confirm the results of the EMSAs, we also directly examined tyrosine phosphorylation of Stat1 and Stat2 in the same nuclear extracts using phosphospecific antibodies (Fig. 1D). Naïve cells incubated with IFN␤ for 30 min showed a significant amount of both tyrosine-phosphorylated Stat1 and Stat2 (Fig. 1D, lane 2, upper panels). Cells incubated with IFN␤ for 16 h or those re-exposed to IFN␤ after a 6 h recovery period showed no tyrosine phosphorylated Stat1 (Fig. 1D, lanes 3 and 5, upper panel). In contrast to Stat1, cells that had recovered for 6 h prior to re-exposure to IFN␤ for 30 min showed clear induction of Stat2 tyrosine phosphorylation (Fig.  1D, lane 5). We have confirmed that the weak IFN␤-induced ISRE-binding complex seen in previously desensitized cells contains Stat2, but we have not been able to detect Stat1 in this complex (data not shown). As expected, there is an IFN␤-dependent nuclear translocation of these proteins, although there is some constitutive nuclear Stat1 in untreated cells (data not prior to the addition of IFN␤ for 2 h. C, formation of ISGF3 and GRR binding complexes in nuclear extracts was analyzed by EMSA. Conditions of incubation were the same as in A except the second incubation with IFN␤ was 30 min for lanes 2 and 5. D, IFN␤-stimulated tyrosine phosphorylation of Stat1 (upper panel) and Stat2 (lower panel) was analyzed by immunoblotting of nuclear extracts with phosphospecific Stat1 or Stat2 antisera. The blots were stripped and reprobed with antisera that recognize total Stat1 and Stat2, which are displayed below the phosphospecific blots. E, IFN␤stimulated binding of Stat1 and Stat2 to the ISG54 promoter as analyzed by ChIP assays. Human fibroblasts were incubated with or without IFN␤ as described in C. Lanes 1-5 show a PCR of 1% of the input chromatin used for each immunoprecipitate. shown). It is notable that the amount of total nuclear Stat1 or Stat2 does not appear to diminish with extended incubation with IFN␤ even though these proteins are subject to dephosphorylation.
To directly examine the binding of Stat1 and Stat2 to the endogenous ISG54 promoter we performed chromatin immunoprecipitation-PCR analysis (ChIP). Human fibroblasts were incubated with or without IFN␤ as described in Fig. 1, C and D. Cells were fixed with formaldehyde, chromatin was isolated, and immunoprecipitated with Stat1-or Stat2-specific antisera. Immunoprecipitated chromatin was then subjected to PCR with primers corresponding to the promoter of ISG54 (Fig. 1E). Incubation of cells with IFN␤ for 30 min induced an increase in the binding of both Stat1 and Stat2 to the ISG54 promoter (lanes 6 and 7). After 16 h Stat1 was not associated with the promoter while there was still a small amount of Stat2 binding (lane 8). This is of interest since under these conditions there is likely a very low level of ISG54 transcription (data not shown). After removal of IFN␤ for 6 h both Stat1 and Stat2 binding to the promoter were undetectable (lane 9). Incubation of previously desensitized cells with IFN␤ induced an association of Stat2, but not Stat1 with the promoter. The amount of Stat2 bound to the promoter of previously desensitized cells varied from experiment to experiment, and ranged from 40 to 80% of naïve cells (data not shown). The amount of Stat2 that bound to the ISG54 promoter in previously desensitized cells correlated well with the degree to which Stat2 was tyrosine phosphorylated in previously desensitized cells. These results establish that primarily Stat2 and not Stat1 binds to the endogenous ISG54 promoter in previously desensitized cells.
The experiments described above were performed using primary human fibroblasts. To determine whether the same pattern of transcriptional responses is seen in other primary cells, a similar experiment was performed using primary human peripheral blood monocytes. Monocytes isolated from healthy volunteers were purified by elutriation and treated as outlined above. IFN␤-stimulated induction of ISG54 mRNA showed a similar pattern in primary monocytes as was seen in primary fibroblasts. Cells incubated for 16 h with IFN␤, followed by a 6-h recovery period, showed a full transcriptional re-induction of the gene when re-exposed to IFN␤ ( Fig. 2A, compare lanes 2 and 5). Whole cell extracts prepared from monocytes under similar conditions were probed for tyrosine-phosphorylated Stat1 and Stat2 (Fig. 2B). As expected, both Stats showed increased tyrosine phosphorylation after 30 min of incubation with IFN␤ (Fig. 2B, lane 2, upper panels), but only Stat2 tyrosine phosphorylation could be detected after 16 h in the presence of IFN␤ followed by a 6 h recovery period, and retreatment with IFN␤ (Fig. 2B, lane 5, upper panels). Interestingly, Stat1 and Stat2 displayed residual tyrosine phosphorylation after 16 h of incubation with IFN␤ (lane 3), which differs from the pattern of tyrosine phosphorylation of these Stats seen in fibroblasts (see Fig. 1D). Probing the blots for total Stat1 and Stat2 (lower panels) revealed that the amount of Stat1 is increased in cells incubated with IFN␤ for 16 h as interferons induce Stat1 gene expression (27).
The observation that IFN␤ does not stimulate tyrosine phosphorylation of Stat1 in previously desensitized cells, would lead to the prediction that genes that are activated by IFN␤ through Stat1 GRR binding would not be induced in such cells. To test FIG. 2. Responses of primary human monocytes to incubation with IFN␤. The treatment conditions prior to preparation of RNA (A and C) and whole cell extracts (B) are identical to those described in Fig.   1, except that IFN␥ (10 ng/ml)was used for re-stimulation in C. A, ISG54, Fc␥R1, and actin mRNAs were analyzed by RPA as in Fig. 1. Immunoblots were probed with anti-phosphoStat1 antisera (upper panel) or anti-phosphoStat2 antisera (lower panel). Both blots were reprobed for total Stat1 or Stat2 as described in Fig. 1. This figure is representative of three experiments. this hypothesis we examined IFN␤-induced expression of Fc␥R1 mRNA in primary human monocytes (Fig. 2C). The promoter region of the Fc␥R1 gene, which is activated primarily by IFN␥ and, to a lesser extent, by IFN␣/␤, harbors a GRR which binds tyrosine-phosphorylated Stat1 (18,28). Monocytes were incubated for 2 h with either IFN␤ or IFN␥, or 16 h with IFN␤ followed by a 6-h recovery period, and re-stimulation with IFN␤ or IFN␥ for an additional 2 h. Fc␥R1 mRNA was then analyzed by RNase protection using actin as an internal control. In naïve cells, both IFN␤ and IFN␥ stimulated the expression of Fc␥R1 (Fig. 2C, lane 1 versus 2 and 4). The induction of Fc␥R1 RNA by IFN␤ is less than that by IFN␥ because this gene is activated by Stat1, and IFN␥ stimulates tyrosine phosphorylation of Stat1 far more effectively that IFN␤ in primary monocytes. 2 Cells that were incubated with IFN␤ for 16 h and subsequently allowed to recover for 6 h showed no Fc␥R1 mRNA (data not shown). Re-exposure of such cells to IFN␤ failed to induce the expression of Fc␥R1 (lane 3). In contrast, IFN␥ did induce the RNA, albeit to a lesser degree than in naïve cells (Fig. 2C, compare lanes 4 and 5). These results strongly support the notion that the IFN␤-stimulated transcriptional re-induction in previously desensitized cell is restricted to genes under the control of the ISRE enhancer.
There are several potential mechanisms that could account for the selective decrease in the tyrosine phosphorylation of Stat1 compared with Stat2 in previously desensitized fibroblasts. One possibility is a selective inhibition of the upstream signaling events leading to Stat1 tyrosine phosphorylation. Conversely, a change in the relative activities of the respective tyrosine phosphatase(s) that dephosphorylate these proteins could also explain the observed results. To examine these pos-sibilities, cytoplasmic and nuclear extracts were prepared from naïve and previously desensitized fibroblasts incubated with IFN␤ for different times. EMSAs were then performed to assay formation of ISRE-binding complexes (Fig. 3A). As previously described (29,30), ISGF3 can be observed in the cytoplasm of IFN␣/␤-treated cells within 10 min (Fig. 3A, left panel, lane 3). Approximately the same amount of ISGF3 can also be seen in the cytoplasm of previously desensitized cells re-exposed to IFN␤ for 10 min (Fig. 3A, left panel, lane 8). However, whereas a robust ISRE-binding activity exists in the nuclei of IFN␤stimulated naïve cells (Fig. 3A, right panel, lanes 3-5), very little or no ISGF3 is detectable in the nucleus of previously desensitized cells (Fig. 3A, right panel, lanes 7-10). Furthermore, Western blotting for nuclear tyrosine-phosphorylated Stat1 and Stat2 in naïve or previously desensitized cells incubated with IFN␤ revealed that only tyrosine-phosphorylated Stat2 was present in the nuclei of previously desensitized cells (Fig. 3B, compare lanes 1 and 2 with 3 and 4, and 5 and 6 with  7 and 8). These results suggest that formation of ISGF3 is not diminished in previously desensitized cells compared with naïve cells. Rather, it appears that the ability of these cells to accumulate the activated complex in the nucleus is impaired, suggesting the elevated activity of a Stat1-specific tyrosine phosphatase in previously desensitized cells.
The observation that IFN␤-stimulated formation of ISGF3 could be detected in about the same amounts in the cytoplasm, but not in the nuclei of re-sensitized cells suggests an enhanced nuclear tyrosine phosphatase activity that dephosphorylates Stat1, and to a lesser extent, Stat2. To further examine this possibility, we assayed the rate of Stat1 dephosphorylation in naïve and previously desensitized cells under pulse-chase type conditions. IFN␤ does not induce tyrosine phosphorylation of Stat1 in previously desensitized cells, thus we used IFN␥ to induce Stat1 tyrosine phosphorylation. Naïve or previously  5 and 11-15). Alternatively, cells were incubated with IFN␤ for 16 h and an additional 6 h in the absence of IFN␤, and re-exposed to the cytokine for 5, 10, 15, or 30 min (lanes 6 -10 and 16 -20). B, immunoblots of nuclear extracts from cells incubated with or without IFN␤ for 30 min were probed with anti-phospho Stat1 (upper panels) or anti-phospho Stat2 (lower panels) antibodies. This figure is representative of two experiments. desensitized human fibroblasts (incubated with IFN␤ for 16 h followed by a 6 h recovery period), were incubated with IFN␥ for 30 min after which the kinase inhibitor staurosporin was added to halt any further tyrosine phosphorylation of Stat1. Staurosporin has been used in a number of reports to analyze the rate of decay of tyrosine phosphorylated Stats (10,31). Whole cell extracts were prepared at various times after the addition of staurosporin and Stat1 DNA binding was assayed by EMSA using a probe corresponding to the GRR of the Fc␥R1 gene ( Fig. 4) (18). The GRR binding activity elicited following 30 min of IFN␥ stimulation was slightly less in previously desensitized cells compared with naïve cells (Fig. 4A, compare  lanes 2 and 7). After addition of staurosporin, GRR binding significantly decayed but was still detectable after 50 min in naïve cells (Fig. 4A, compare lanes 2 and 5). However, in previously desensitized cells the GRR binding activity disappeared within 5 min after the addition of staurosporin (Fig. 4A,  compare lanes 7 and 8). These results strongly support our model that an elevated tyrosine phosphatase activity directed against Stat1 is present in previously desensitized cells. To examine the subcellular location of this enhanced PTPase activity, nuclear extracts were prepared from naïve and previously desensitized cells incubated with IFN␥ followed by the addition of staurosporin (Fig. 4B). In a manner similar to whole cell extracts there is a differential loss of the GRR binding complex in nuclear fractions from cells incubated 16 h with IFN␤.
The enhanced PTPase activity toward Stat1 in cells exposed to IFN␤ for extended periods suggests that these cells would also show altered sensitivity to IFN␥. The results presented in Fig. 2C support this model since IFN␥ induction of Fc␥R1 RNA is diminished in monocytes incubated with IFN␤ for 16 h com-pared with naïve cells. To examine this in greater detail, human fibroblasts were incubated with or without IFN␤ for 16 h, allowed to recover for 6 h and then stimulated with different doses of IFN␥ for 30 min. Nuclear extracts were prepared and analyzed for activation of Stat1 by EMSA using the GRR probe (Fig. 5). In accordance with our hypothesis, a clear decrease in the amount of activated Stat1 is observed in cells exposed to IFN␥ that had been previously desensitized toward IFN␤. The difference in the amount of activated Stat1 between naïve and previously desensitized cells is most prominent when cells are incubated with submaximal doses of IFN␥. These results not only provide independent confirmation of increased tyrosine phosphatase activity in cells exposed for extended periods to IFN␤, but also suggest that such exposure can decrease the ability of other cytokines to activate Stat1.
Two tyrosine phosphatases have been reported to be responsible for dephosphorylating Stat1 (10,32). One is SHP2, which was identified using an antibody array (32), and the other, Tc-PTP, was purified from nuclear extracts of HeLa cells using tyrosine-phosphorylated Stat1 as a substrate (10). At the moment it is not evident whether these PTPases function to simultaneously dephosphorylate Stat1 or whether they each function in different physiological conditions. To examine this issue, we compared the rates of Stat1 dephosphorylation in naïve and previously desensitized Tc-PTP-deficient immortalized MEFs (10,32). MEFs were subject to pulse-chase type experiments as previously outlined. Extracts were prepared at various times after the addition of staurosporin and immunoblots probed with anti phospho-Stat1 antiserum. In wild-type MEFs there was a rapid disappearance of tyrosine-phosphorylated Stat1 after the addition of staurosporin (Fig. 6A, lanes  2-5). In cells that were incubated overnight with IFN␤, allowed to recover for 6 h and then exposed to IFN␥, the rate of Stat1 tyrosine dephosphorylation was enhanced (Fig. 6A, lanes 7-10). Similar results were seen in Tc-PTP Ϫ/Ϫ cells that were reconstituted with the appropriate PTPase (10, 32) and data not shown. A similar pattern of Stat1 dephosphorylation was seen in naïve and previously desensitized SHP2-null cells (data not shown). In contrast to wild-type MEFs, little Stat1 dephosphorylation was observed after exposure of Tc-PTP-null cells to staurosporin for 30 min (Fig. 6B, lanes 3-5 and 8 -10). Prolonged tyrosine phosphorylation of Stat1 was seen in both naïve Tc-PTP-deficient cells and those exposed to IFN␤ for 16 h. In fact, Stat1 remained tyrosine-phosphorylated after continuous incubation of cells with IFN␤ for 16 h followed by an additional 6 h in the absence of IFN␤ (Fig. 6B, lane 6).
Because our preliminary findings indicated that IFN␤ stimulated tyrosine phosphorylation of Stat2 was decreased, but FIG. 4. Rate of Stat1 dephosphorylation is enhanced in previously desensitized human fibroblasts. A, naïve cells or those treated with IFN␤ for 16 h and without IFN␤ for an additional 6 h, were incubated with IFN␥ for 30 min. Staurosporin (1 M) was then added to cells to prevent further activation of Stat1 and whole cell extracts were prepared 5, 10, or 20 min later. Tyrosine phosphorylation of Stat1 was analyzed by EMSA using the GRR element as a probe. B, the same experiment was performed as above except nuclear extracts were used for the EMSAs instead of whole cell extracts. This figure is representative of multiple experiments.
FIG. 5. IFN␥-stimulated Stat1 activation is attenuated in cells pretreated with IFN␤. Human fibroblasts were incubated with or without IFN␤ for 16 h followed by an additional 6 h in the absence of IFN␤. Cells were then exposed to increasing concentrations of IFN␥ for 30 min. Nuclear extracts were prepared and analyzed for Stat1 activation by EMSA as described in the legend to Fig. 1. This figure is representative of two experiments. still was detectable in previously desensitized cells, we wanted to examine whether the expression of Tc-PTP altered tyrosine phosphorylation of Stat2. Tc-PTP-null cells or their reconstituted counterparts were incubated with or without IFN␤ for 16 h prior to preparing cell extracts to analyze tyrosine phosphorylation of Stat1 and Stat2 (Fig. 7). While we were unable to detect tyrosine phosphorylation of Stat1 in previously desensitized reconstituted cells re-exposed to IFN␤ (Fig. 7A, lane 2  versus 5), a re-induction of Stat1 tyrosine phosphorylation equal to the initial response was observed in the absence of TcPTP (Fig. 7A, lane 7 versus 10). IFN␤-stimulated tyrosine phosphorylation of Stat2 was equally strong in naïve compared with previously desensitized cells (Fig. 7B, lane 2 versus 5), regardless of TcPTP expression (Fig. 7B, lanes 2, 5, 7, and 10). These results not only confirm that activation of Stat1 occurs in previously desensitized cells that do not express Tc-PTP, they also emphasize that expression of Tc-PTP selectively controls the tyrosine phosphorylation of Stat1 without altering Stat2 activation.
Tyrosine phosphorylation of Stat1 remains elevated in Tc-PTP-null cells even after incubation with IFN␤ for 16 h followed by a 6 h wash (see Figs. 6 and 7). We wanted to examine whether the continued presence of tyrosine phosphorylated Stat1 in Tc-PTP-null cells correlated with continued expression of ISG54. RNAs were prepared from naïve or previously desensitized cells Tc-PTP-null cells and their reconstituted counterparts. ISG54 expression was assayed by RT-PCR (Fig. 7C). In a manner similar to human fibroblasts and monocytes, treatment of these cells with IFN␤ for 2 h stimulated ISG54 RNA accumulation (Fig. 7C, lane 2 and 7) compared with GAPDH, which was used as an internal control. ISG54 RNA remained present in cells continuously exposed to IFN␤ for 16 h, and disappeared after incubation of cells in the absence of IFN␤ for 6 h. Interestingly, disappearance of ISG54 RNA occurred in both wild-type and Tc-PTP-null cells (Fig. 7C, lane 4 and 9), even though Tc-PTP-null cells continued to contain significant amounts of tyrosine-phosphorylated Stat1. The re-induction of A and B, cells were exposed to IFN␤ as described in the legend to Fig. 1, C and D. Extracts were separated by SDS-PAGE, and the immunoblots were probed with antisera specific for tyrosine-phosphorylated Stat1 (A) or tyrosine-phosphorylated Stat2 (B). Blots were also probed for total Stat1 and actin to ensure equal loading. C, RNAs were prepared under the same conditions except the treatment times with IFN␤ were 4 h. ISG54 and GAPDH RNAs were analyzed by RT-PCR as described under "Materials and Methods." This figure is representative of two experiments.
ISG54 was also the same in both the reconstituted and TcPTPnull cells (Fig. 7C, lanes 5 and 10). These results suggest that although tyrosine-phosphorylated Stat1 is needed to initially induce the transcription of ISG54 in naïve cells, once transcription has been initiated it is not needed for the continued expression of the gene, and it plays no significant role in initiating the transcription of the gene in previously desensitized cells. DISCUSSION Under physiological and pathological conditions, levels of cytokines and growth factors can fluctuate. The mechanisms by which cells respond to these stimuli and down-regulate an initial response is well documented. Much less is known about how cells that have seen a particular stimulus respond to the same stimulus upon re-exposure. Previous studies examining IFN␤-stimulated genes indicate that cells become desensitized after incubation with IFN␤ for about 8 h (24,33). The experiments in this report demonstrate changes in the mechanisms by which IFN␤ induces ISRE-dependent genes in cells that have been previously exposed to IFN␤. To our surprise, the nature of the response in previously desensitized cells compared with naïve cells is significantly different in two aspects. 1) The amount of the ISRE-binding complex formed in previously desensitized cells is far less than that seen in naïve cells under conditions where the induction of a well-described IFN␤activated ISRE-dependent gene (ISG54) is the same. 2) The relative ability of IFN␤ to stimulate the tyrosine-phosphorylation of Stat1 is far less than the IFN␤-stimulated tyrosinephosphorylation of Stat2 in previously desensitized cells (see Figs. 1 and 2), and this results correlates well with the binding of these transcription factors to the endogenous ISG54 promoter as analyzed by ChIP assays (Fig. 1E). In fact, in most experiments we have not detected any IFN␤-mediated activation of Stat1 in re-stimulated cells. This observation is also confirmed when we examined induction of the Stat1-dependent gene Fc␥R1 in human monocytes (see Fig. 2C). While IFN␤ activation of ISG54 is the same in naïve and previously desensitized monocytes, we were not able to detect any IFN␤-induced expression of Fc␥R1 RNA in previously desensitized cells. These experiments suggest that there are different rates of recovery of previously desensitized cells in their ability to activate sets of early response genes induced by Stat1 compared with those induced by Stat1 and Stat2. It also should be noted that we have not seen any changes in the expression of these proteins in naïve compared with previously desensitized cells (data not shown).
Published studies have documented a dose-dependent relationship between the levels of induction of interferon stimulated genes and activation of Stat1 in naïve cells (18). We have confirmed these observation in human fibroblasts (data not shown). The "ISGF3" transcription complex that we observed in previously desensitized cells does contain Stat2 as determined by antibody supershift analysis, but we have not been able to detect Stat1 in the complex (data not shown). Several reports have indicated that Stat2 is the primary component of the ISGF3 transcription complex responsible for inducing the expression of ISRE-driven genes. Deletions of the C-terminal region of Stat2 abrogate the transcriptional activity of ISGF3 (34). Recently it has been demonstrated that an IRF9-STAT2 hybrid can both induce ISRE gene expression and allow for an antiviral response in the absence of type 1 interferons (35). Our results indicate that Stat2 can also activate ISRE-dependent genes in the absence of tyrosine phosphorylation of Stat1 in a variety of cells under conditions that are physiological and likely occur within the setting of innate immune responses. The fact that there is no difference in the pattern of IFN␤ stimu-lated ISG54 RNA expression in wild type compared with Tc-PTP-null cells highlights the probable importance of Stat2 and lack of importance of Stat1 in cells repeatedly exposed to these cytokines (Fig. 7C).
There are several possibilities that might account for the full activation of ISG54 under conditions where there is reduced activation of Stat2 and the lack of activation of Stat1. The chromatin in the regions of interferon-stimulated genes might remain relaxed even after the transcriptional response has dissipated. Under these conditions, even a reduced activation of Stat2 (Stat2 provides the primary transcriptional activation domain of ISGF3) and binding to the ISG54 promoter could be sufficient to stimulate a full response. The other possibility is that another IFN␤-regulated transcription factor(s) complements or substitutes for ISGF3 to drive the response in previously desensitized cells. Although we have no evidence for this scenario, it is known that the interferon regulatory factors (IRFs) bind ISREs, and IRF1 and IRF3 can activate or contribute to the activation of ISGs including ISG54 (36,37). We are presently exploring whether either or both of these mechanisms account for full activation of ISG54 in previously desensitized cells.
The other clear difference in IFN␤-stimulated signaling in re-sensitized cells compared with naïve cells is that tyrosine phosphorylation of Stat1 is absent or very diminished. Decreased IFN␤-stimulated tyrosine phosphorylation of Stat1 could be due to either decreased production of tyrosine-phosphorylated protein or enhanced dephosphorylation of Stat1.
Our results indicate that in previously desensitized cells there is an elevated PTPase activity that dephosphorylates Stat1 (Figs. 4 and 5). It has been known for several years that a nuclear PTPase plays an important role in down-regulating IFN-stimulated gene expression (4). Recently, two PTPases have been described to be involved in dephosphorylating Stat1 (10,32). Using cells that do not express Tc-PTP, we demonstrated that decreased levels of phosphorylated Stat1 in previously desensitized cells are likely due to the expression of this PTPase. Tc-PTP is ubiquitously expressed and has two isoforms. TC48, (the 48-kDa isoform), is localized to the endoplasmic reticulum, and TC45 is primarily targeted to the nucleus, although there is a small fraction in the cytoplasm (38). Although we see enhanced dephosphorylation of Stat1 in nuclear fractions of previously desensitized cells, cytoplasmic fractions also show enhanced activity (data not shown). This is not surprising since both nuclear and cytoplasmic Tc-PTP have been reported to dephosphorylate Stat1 (10). We are investigating whether the activity or subcellular localization of Tc-PTP is affected by extended exposure to IFN␤. We have detected no changes in the total amount of Tc-PTP in cells incubated with IFN␤ for extended periods (data not shown). It has been reported that methylation of arginine 31 of Stat1 enhances its interaction with Tc-PTP and decreases its interaction with PIAS1, a negative regulator of Stat1-dependent transcription (39). It is therefore possible that previously desensitized cells will contain altered levels of arginine methlyated Stat1 and/or changes in its affinity for PIAS1. Although we have not completely eliminated the role of SOCS protein expression in blunting IFN␤-stimulated tyrosine phosphorylation of Stat1 in previously desensitized cells, the fact that tyrosine phosphorylation of Stat2 is not nearly as inhibited as tyrosine phosphorylation of Stat1 argues for SOC proteins playing a less prominent if any role in the phenomenon that we are studying (see Fig. 7). The PTPase responsible for dephosphorylating Stat2 has not been defined although it does not appear to be Tc-PTP (Fig. 7).
It is notable that cells exposed to IFN␤ for extended times show a blunted response to IFN␥ with regard to activation of Stat1, especially at suboptimal doses of IFN␥ (Fig. 5). IFN␥induced expression of Fc␥R1 is also decreased in primary monocytes previously treated with IFN␤ (see Fig. 2C). Although the physiological significance of extended exposure of cells to type I interferons is not fully appreciated, there have been a variety of reports which indicate that type 1 interferons can inhibit the effects of IFN␥ with respect to induction of MHC Class II and inducible nitric-oxide synthase expression on macrophages and endothelial cells (40 -43). Furthermore, the ability of IFN␥ to inhibit histoplasmosis infection of macrophages is blunted by IFN␣/␤ (44). IFN␤ activation of Tc-PTP, resulting in attenuated responses to IFN␥, is a plausible mechanism to explain these observations. We speculate that the ability of IFN␤ to regulate the actions of IFN␥ will also occur with other cytokines that activate common sets of Stats. The ability of prolonged exposure to one cytokine to attenuate responses to another cytokine by a post-receptor down-regulation of Stat activation is a novel mechanism of cytokine "cross-talk." Such a mechanism may contribute to altered patterns of gene regulation in mammals that are repeatedly exposed to bursts of cytokine release under conditions of infection and inflammation.