Role of Protein Phosphorylation in Activation of Interferon-stimulated Gene Factors”

Possible lated gene involvement of protein phosphorylation in (1FN)-mediated activation of IFN-stimu-factor 3 (ISGF3) was investigated. For this purpose, in vivo experiments with specific inhibitors of protein kinases and in vitro experiments with pro- tein phosphatases were carried out. In HeLaM cells, 2-aminopurine, an inhibitor of double-stranded RNA- dependent protein kinase, blocked the induction of ISGF3-y subunit but not the activation of ISGF3a sub- unit. A series of experiments using combinations of protein and RNA synthesis inhibitors and 2-aminopu- rine indicated that the block elicited by 2-aminopurine was at the level of ISGF3r mRNA synthesis. Activation of ISGF3a, although insensitive to 2-aminopurine, was completely blocked by 10 nM staurosporine, an inhib- itor of protein kinase C. On the other hand, even 500 nM staurosporine did not block the induction of ISGF3-y. Incubation of cytoplasmic or nuclear extracts of IFN-treated HeLaM cells in vitro with ,alkaline phosphatase completely eliminated their ability to form the ISGF3 complex but not the ISGF1 complex. Treatment with acid phosphatase, on the other hand, changed the electrophoretic mobility of the ISGF3 complex but did not obliterate it. Complementation experiments revealed that ISGF3a was the alkaline phosphatase-sensitive component of the complex. These results suggest that a protein kinase C-mediated phosphorylation step is involved in ISGF3a activation and a 2-aminopurine-sensitive

At least three trans-acting factors, present in the nuclei of IFN-a-treated cells, bind to ISRE (6,10,11,13,23,24). Electrophoretic mobility shift assays and DNA footprinting analyses have been used to study these IFN-stimulated gene factors (ISGF). The fastest moving complex, termed ISGF1, is present in both IFN-a-treated and untreated cells. ISGF2 is induced by both IFN-a and IFN-y, and its appearance requires ongoing protein synthesis during IFN treatment (10,23). Recently, the cDNA for ISGF2 has been cloned and sequenced, thereby revealing its identity with IRF-1 (25), a trans-acting factor involved in the transcription of the P-IFN gene (26). The third factor, ISGF3, is induced by IFN-a but not by IFN-/, and it appears to be the crucial positive regulatory factor which is responsible for transcriptional activation of the IFN-a-responsive genes (10).
Activation of ISGF3 occurs in the cytoplasm of cells within minutes of contact with 28). In untreated cells, active ISGF3 cannot be detected either in the cytoplasmic or nuclear fractions. Upon IFN-a treatment, active ISGF3 appears first in the cytoplasm and then it translocates to the nucleus where it presumably binds to the ISREs of different IFN-inducible genes and thereby activates their transcription. Thus, ISGF3 itself physically transduces the signal from the cytoplasm to the gene. Understanding the mechanism of activation of ISGF3 in the cytoplasm of IFN-a-treated cells is therefore crucial for defining the signal transduction pathway of IFN-a.
ISGF3 is composed of two functional subunits, ISGF3y and ISGF3a, both of which are needed for efficient DNA binding (28). In most cells, there is a high constitutive level of ISGF37, but HeLaM cells do not contain a detectable level of ISGF3y (21). We have used this cell line to delineate some details of ISGF3 activation by IFN-a. ISGF3a, which is present in an inactive form in the cytoplasm, can be activated within minutes of cellular contact with IFN-a. This activation does not need new protein synthesis (21). As a result, in cells having a high constitutive level of ISGFJy, the overall process of ISGF3 activation and the resultant induction of gene transcription do not need ongoing protein synthesis. This is not the case, however, for HeLaM cells. In these cells, ISGF3 activation requires, in addition to ISGF3a activation, induced synthesis of ISGF3y subunit (21). We and others (21,29) have shown that ISGF3y synthesis can be induced by either IFN-a or IFN-y. IFN-y, however, cannot activate ISGF3a. An in vitro complementation assay has been developed for measuring the levels of active ISGF3a and active ISGF3y in 6389 cell extracts (21). We have operationally defined two distinct signals elicited by IFN-a. Signal 1, which is also produced by IFN-7, induces the synthesis of ISGF37, whereas signal 2 activates the inactive precursor of ISGF3a. The two putative signals have different sensitivities t o several inhibitors such as cytoheximide and 2-aminopurine (2AP). Using the complementation assay, we have shown that ISGF3-y is required for the nuclear translocation of activated ISGF3a (21). We have also shown activated ISGF3a, in the absence of ISGF37, has a relatively short half-life in the cytoplasm. The activation characteristics of ISGF3a, namely its rapidity, its reversibility, and its independence of protein synthesis, strongly suggest that the activation-inactivation equilibrium is regulated by a specific IFN-a-induced post-translation modification of this protein (21). Protein phosphorylation could be such a specific modification.
There are several reports in the literature suggesting the In this paper, we provide further evidence for the involvement of protein phosphorylation in ISGF3 activation process. Our experiments demonstrated that 2AP specifically inhibited the synthesis of ISGF3-y mRNA, thereby suggesting a role of protein phosphorylation in this process. We also showed that 10 nM staurosporine, an inhibitor of protein kinase C, blocked the activation of ISGF3a. The same inhibitor, even at 500 nM concentration, did not block the induction of ISGF3y by either IFN-a or IFN-7. Treatment of ISGF3 with alkaline phosphatase, in uitro, abolished its DNA-binding ability. The component affected by this treatment was ISGF3a but not ISGF3-y.

EXPERIMENTAL PROCEDURES
Materials-Cell culture materials were obtained from GIBCO. Sources of pure IFN-a, IFN-y, and other reagents were described in our previous publication (39). Cycloheximide (CHX), acid-phosphatase, and alkaline phosphatase were purchased from Boehringer Mannheim. Staurosporin, actinomycin D, 2-aminopurine, and phor-bo1 12-myristate 13-acetate (PMA) were from Sigma. Radioactive chemicals were obtained from Amersham Corp.
CeZZ Culture and Treatments-Cell culture condition for HeLaM cells was described in our earlier publications (38, 39). IFN-a and IFN-y (500 units/ml of medium), 2AP (10 mM), cycloheximide (50 pg/ml), and actinomycin D (2 pg/ml) were used for treatments of cells where indicated.
Oligonucleotide Probe and Gel Mobility Shift Assay-A doublestranded synthetic (Applied Biosystem model 380B) DNA containing the ISRE region (-125 to -93) of the human 56K (561) gene was used as a probe (21). The sequence of the probe was aatt-CTAGCTTTAGTTTCACTTTCCCCTTTCGGTTTg (capital letters represent the sequence from -125 to -93 and underlined se-quences represent the consensus ISRE region). Oligonucleotides were gel-purified, annealed, and labeled either by using [y3'-P]ATP and polynucleotide kinase or by using CY[~*P]~ATP and Klenow enzyme. Cytoplasmic and nuclear extracts were prepared by a modification of Dignam et al. (40). Cells from four 150-mm plates for each treatment were washed three times with cold phosphate-buffered saline, scraped, and harvested by centrifugation. In brief, cells were suspended in 5 ml of buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgC12, and 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) and incubated on ice for 15 min. Cells were centrifuged, and the pellet was resuspended in 0.7 ml of buffer A and homogenized with a loose fitting glass homogenizer with 15 strokes. Nuclei were separated from cytoplasms by centrifugation at 500 x g for 10 min and then cytosols were prepared by centrifugation at 100,000 X g for 1 h. Nuclei were suspended in 300 pl of buffer C (20 mM HEPES, pH 7.9,25% glycerol, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride), shaken for 20 min at 4 "C, and nuclear extracts were obtained by centrifugation at 12,000 x g for 10 min. Complete protocol for the condition of gel mobility shift assay was described in detail in our previous publication (21). Removal of Alkaline Phosphatase after in Vitro Treatments of Extracts-Ten pl of alkaline-phosphatase-agarose beads (Sigma catalog No. P0762) were washed three times with 1 X gel shift buffer (10 mM Tris hydrochloride, pH 8.0, 50 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, and 5% glycerol). ISGF3y or ISGF3a containing cytosolic extracts were incubated with or without packed alkaline phosphatase agarose for 1 h at 4 "C with mild shaking. After incubation, the extracts were centrifuged briefly, and the supernatants were loaded onto Bio-Rad columns (catalog No. 731-1550) at 4 "C to remove any agarose beads left behind. Complementation assays for ISGF3 band formation were performed using flow-through from each column with end-labeled probe. RESULTS 2-Amimpurine Blocks ISGF3y mRNA Synthesis-We have previously shown that ISGF3-y induction by either IFN-a or IFN--y needs ongoing protein synthesis and it is blocked by 2AP (21). In a series of new experiments we have analyzed this induction process further in order to be able to identify the 2AP-sensitive step in this process. A hypothetical working model for the induction of active ISGF3-y is shown in Fig. 1. According t o this model IFN--/ binds to its receptor and elicits a signal in step 1 which persists even after IFN is removed.
This signal travels to the nucleus and induces transcription of the ISGF3y gene to produce ISGF3r mRNA in step 2. In step 3, the mRNA is translated into ISGF3-y protein which may be activated by a post-translation modification in step 4.
We wanted to test which of these steps is sensitive to inhibition by 2AP. For this purpose, assays were devised to study these steps individually, using inhibitors of protein and mRNA syntheses. A previously described complementation \ r-Step 3

FIG. 1. Possible sequential steps in the process of ISGF3r induction by IFNr in HeLaM cells.
IFNy binds to a specific receptor on the cell surface and transmits a signal (Step 1 ).to the ISGF3y gene in the nucleus. As a result, in Step 2, ISGF3y mRNA is transcribed from the ISGF3y gene. In Step 3, ISGF3y mRNA is translated to the ISGF3y protein. In Step 4, functional ISGF3y* is formed by a post-translational modification. assay was used for measuring functional ISGF3-y (21). ISGF3a was supplied from a cytoplasmic extract of HeLaM cells treated with IFN-a and CHX. As shown in Fig. 2A, an IFNy-treated cell extract when supplemented with ISGF3a can form the ISGF3 complex (lane 2 ) . ISGF3y was not present in extracts of cells treated with IFN-y + CHX (lane 3). On the other hand, when cells were first treated with IFN-y + CHX, the additives were removed and the incubation was continued in untreated culture medium, active ISGF3y was formed ( l a n e 4 ) . During the follow-up incubation new mRNA synthesis was not needed since the presence of actinomycin D did not inhibit ISGF3y formation (lane 6 ) , although its presence during the IFN-y treatment completely inhibited ISGF3y formation (data not shown). These results are compatible with our working model (Fig. 1). and they show that different steps of the induction process can be studied separately. In the next experiment, effects of 2AP on different steps of the induction process were examined. The overall ISGF3y induction process was inhibited by 2AP (Fig. 2B, lane 1 ). However, if cells were treated with IFN-y and 2AP, the agents were removed by washing, and incubation was continued, ISGF3y induction was not blocked (Fig. 2B, lane 5). This observation indicates that step 1 of our working model for ISGF3-y induction was not blocked by 2AP. When ISGF3y mRNA synthesis was allowed to occur by treating the cells with IFN-y and CHX, 2AP could not prevent the subsequent formation of active ISGF3y after the removal of IFN-y and CHX (Fig. 2B,  lane 3). This result indicates that 2AP does not have any effect on steps 3 and 4 of the induction process. By elimination, it appears that step 2, at which ISGF3 mRNA is synthesized in response to a signal elicited by IFN-y, is the step affected by 2AP. Several observations supported this conclusion. When cells were treated with IFN-y and 2AP followed by an incubation in the presence of CHX only, no active ISGF3-y was formed (Fig. 2R, lane 2 ) , suggesting that the 2APmediated block is prior to step 3. As expected, inclusion of 2AP during the first phase of incubation in the presence of IFN-y and CHX did not block ISGF3y induction (Fig. 2R,  lane 7). These results are consistent with 2AP blocking step 2 of the working model ( Fig. 1). 2AP does not block step 1 in which a IFN-y-elicited signal is produced which lasts after IFN-y is removed. 2AP also does not block the translation of ISGF3y protein or its putative activation by post-translational modification. The step sensitive to 2AP is step 2, at which ISGF3y mRNA is synthesized. In accord with this conclusion we observed that no active ISGF3y was formed if mRNA synthesis was blocked by actinomycin D during the second phase of incubation following a first phase of incuhation with IFN-y and 2AP (data not shown), again suggesting that no ISGF3-y mRNA had been synthesized in the presence of 2AP.
Staurosporine Blocks ISGF3a Activation-Staurosporine is a specific inhibitor of protein kinase C. A t a relatively high concentration, it has been shown to inhibit IFN-inducible gene transcription (32). This inhibition has been traced to a lack of activation of ISGF3 in cells treated with IFN-a in the presence of staurosporine. In the experiments shown in Fig.  3, we examined if ISGF3a activation or ISGF3y induction or both were sensitive to staurosporine in HeLaM cells. ISGF3a activation was blocked by staurosporine a t 10 and 50 nM concentrations (Fig. 3A, lanes 6 and 7, respectively). ISGF3y induction by IFN-y, on the other hand, was not blocked by staurosporine a t 10 and 50 nM concentrations (Fig. 3A, lanes  4 and 5 ) or even at 500 nM concentration (data not shown). When cells were treated with IFN-a and CHX for 30 min followed by a treatment with staurosporine (500 nM) for 30 min, active ISGF3a was not present in the extract (Fig. 3A,  lane 8), suggesting that the functional half-life of active ISGF3a is short and its reactivation is blocked by staurosporine. Induction of ISGF3y followed by a treatment with staurosporine, on the other hand, did not obliterate i t s activity (Fig. 3A, lane 9 ) .
ISGF3y can be induced not only by IFN-y but also by IFNa. We investigated if its induction by IFN-a was also insensitive to staurosporine. As expected, IFN-a induced ISGFB and either CHX or staurosporine blocked this induction process (Fig. 3B, lanes 3-5). However, when assayed for ISGF3y induction or ISGF3a activation individually, it was apparent that ISGF3y was induced by IFN-n in the presence of 500 nM staurosporine (Fig. 3B, lane 2 ) , whereas ISGF3a was activated in the presence of CHX. These results demonstrated that although a very low dose of staurosporine (10 nM) can block the activation of ISGF3a, even a relatively high dose (500 nM) of it cannot block the induction of ISGF3y by either IFN-a or IFN-y.
PMA Does Not Affect ISGF3a Activation-Since ISCF3n activation was blocked by staurosporine, a known inhibitor of protein kinase C, we wondered whether PMA, which activates protein kinase C and eventually down-regulates it from the cell surface, will affect ISGF3a activation. HeLaM cells were treated with PMA for 1 h to activate protein kinase C and then treated with IFN-a and CHX for ISGF3a activation. The short treatment with PMA did not block ISGF3a activation (Fig. 4, lane 3). It also did not block ISCF3-y induction by IFN-a, since treatment with PMA followed by IFN-a resulted in active ISGF3 formation (Fig. 4, lane 4 ) . PMA treatment by itself, however, did not activate ISGF3a (Fig. 4,  lane 5). When cells were treated for 24 h with PMA to completely down-regulate protein kinase C from the cell surface, activation of ISGF3a (Fig. 4, lane 7) and induction of ISGF3r (Fig. 4, lane 8 ) were not affected.
Effects of in Vitro Dephosphorylation of ISGF3"Our observations that inhibitors of different protein kinases block the activation of ISGF3 in uiuo suggested that this process involves phosphorylation of ISGF3. consequently, we argued that dephosphorylation of activated ISGF3 should affect its ISRE binding activity. We tested this possibility by incubating in vitro ISGF3-containing nuclear extracts with acid phosphatase or alkaline phosphatase before examining their ISRE binding activities (Fig. 5). Acid phosphatase treatment enhanced the mobility of the ISGFB complex as well as that of part of the ISGFl complex (Fig. 5 , lane 2 ) . Alkaline phosphatase treatment, on the other hand, did not affect the ISGFl complex, but formation of the ISGF3 complex was totally blocked in such an extract (Fig. 5 , lane 3).
Next, we examined if the cytoplasmic form of ISGF3 was also sensitive to the action of phosphatases and if the sensi-

ISGF3-
Nuclear E x t r a c t s FIG. 5. Effects of phosphatase treatments of nuclear extracts on ISGF.7 complex formation. Nuclear extracts were prepared from cells treated with IFNy for 16 h followed hv IFNn for 30 min. The extract was either untreated ( l a w 1 ) or treated with 10.0 pg/ml of acid phosphatase (lane 2 ) or 12.5 pg/ml of alkaline phosphatase (lane 3) for 20 min a t 22 "C before the addition of the prohe. Acid phosphatase was dissolved in 50 mM sodium acetate. pH 5.2. and alkaline phosphatase was either added directly or diluted with 20 mM Tris hydrochloride, pH 8.0, immediately hefore use. Since phosphatases would have removed the laheled terminal phosphate from a kinase-labeled prohe, for this experiment the prohe WAS Inheled with ['"PIdATP by filling in using Klenow polymerase. tivity remained after the formation of the protein-DNA complex. In one experiment, cytoplasmic extracts of IFN-treated cells were treated with increasing concentrations of acid phos-phatase and then challenged with the labeled DNA probe for complex formation (Fig. 6, lanes 2-4). As was observed with the nuclear extract, acid phosphatase treatment caused a clear shift in the mobility of the ISGF3 complex. The new complex moved slightly faster than ISGF3. Increasing the phosphatase concentration increased the proportion of the faster form, which appeared as a distinct tight band. The acid phosphatase treatment, even at the highest concentration, did not perturb the formation or the mobility of the constitutive complex which had a much faster mobility. Exactly similar results were obtained when the extracts were first incubated with the labeled probe to allow complex formation and then treated with acid phosphatase a t increasing concentrations (Fig. 6,  lanes 5-7). These results suggest that acid phosphatase treatment removes specific and probably the same phosphate residues from the ISGF3 complex, irrespective of whether it is bound to DNA or not, and the removal of these residues does not affect the DNA binding property of the complex, although it changes its electrophoretic mobility. In another experiment, we tested the sensitivity of cytoplasmic extracts t o alkaline phosphatase treatment before and after incubating with the labeled DNA probe (Fig. 6, lanes 8-1 I ). This treatment obliterated the ISGF3 complex in a dose-dependent manner, both before and after DNA binding, suggesting that it removed specific phosphate residues from the proteins in ISGF3 which are essential for binding of this complex to DNA. It is curious to note that the formation of the faster constitutive complex was enhanced by alkaline phosphatase treatment.
ISGF3a Is the Alkaline Phosphatase-sensitive Component of ISGF3-Once we established that the DNA-binding activity of ISGF3 could be eliminated by alkaline phosphatase treatment, we tested if the ISGF3a component or the ISGF3y component or both were affected by this treatment. For this purpose, ISGF3y-containing extract of IFN-y-treated HeLaM cells and ISGF3a-containing extract of IFN-a and CHX-treated HeLaM cells were separately treated with alkaline phosphatase and tested for activities using the complementation assay (Fig. 7). T o carry out this experiment successfully, it was necessary to completely remove alkaline phosphatase from the treated extract before the complementation assay was performed. This was achieved by using alkaline phosphatase bound to an insoluble matrix as described in detail under "Experimental Procedures." When neither component was phosphatase-treated, ISGF3 complex  lanes 4 and 7), 5.0 pg/ml (lunes R and IO), and 12.5 pg/ml (Ianes 9 and 1 1 ) . Lune I was not treated with nny phosphatase. Other conditions were the same as in Fig. 5. was formed, as expected, upon complementation (Fig. 7 , lane   I). Treatment of the ISGFBy component with alkaline phosphatase did not affect its ability to form the ISGFB complex (Fig. 7, lane 2). Supplementation with additional untreated extract containing ISGFBy did not have anv major effect (Fig.  7 , lane 3). Treatment of the ISGF3n-containing extract with alkaline phosphatase, on the other hand, completely abolished the formation of the ISGFB complex (Fig. 7 , l a m 4 ) , suggesting that ISGF3n was the phosphatase-sensitive component. T o establish that complete removal of phosphatase had been achieved by our procedures, the same extract was supplemented with a source of active untreated ISGFBa, in addition to the usual addition of ISGF.77, and tested for ISGF.3 complex formation (Fig. 7 , lane 5 ) . The usual amount of ISGFB Protein Phosphorylation in IFN Signal Transduction complex was formed thereby confirming the validity of our assay system. It is also worth noting that unbound labeled DNA probe was present in all lanes. If alkaline phosphatase had not been removed completely, it would have removed the terminal labeled phosphate of the DNA probe, and the released labeled phosphate would have migrated out of the gel under the conditions of electrophoresis used here. As mentioned earlier, formation of the faster migrating constitutive complex was not affected by the phosphatase treatment.

DISCUSSION
In our early studies with HeLaM cells, we observed that gene induction by IFN-a required ongoing protein synthesis which could be obviated by pretreating the cells with 41). According to our working model (21,38,39), both IFNs produce a hypothetical signal, signal 1, which induces the transcription of a gene, X. The product, mRNAX, needs to be translated to the corresponding protein, protein X for functioning of signal 1. As would be expected from this model, inhibition of either mRNA synthesis or protein synthesis blocked this pathway. Our recent results strongly indicate that protein X is identical to . We have shown that induction of protein X or ISGF3y by either IFN-a or IFN-y is blocked by 2AP. In the experiments reported here we further dissected this process and tried to identify the exact step in the induction pathway of ISGF3-y which is sensitive to 2AP. We operationally divided this induction process into four discrete steps which can be assayed individually. Our results clearly indicated that the 2AP-mediated block is at the level of ISGF3-y mRNA synthesis. There are obviously multiple steps within what has been called step 2 here. Such steps may involve activation of specific transacting factors, their binding to cognate cis-acting elements of ISGF3y gene, and their interactions with other proteins involved in the transcription process. Testing of each of these steps for 2AP sensitivity awaits the development of reagents such as ISGF3y cDNA and genomic clones. Nonetheless, the experiments reported here established that the action of a specific protein kinase is required at one or more steps of this process. Whether the kinase in question is the doublestranded RNA-dependent protein kinase remains to be seen.
Several reports in the literature implicated the involvement of protein kinase C in IFN-a's signal transduction (30-35).
These studies mostly used various inhibitors of this enzyme and showed that they block the actions of IFN-a. Pfeffer et al. (31) also showed that protein kinase C is activated by IFNa, and the @-isozyme is translocated from the cytoplasm to the particulate fraction in response to IFN-a treatment of HeLa cells. Reich and Pfeffer (32) showed that staurosporine, a specific inhibitor of protein kinase C, prevented the formation of ISGF3 complex. In their studies, relatively high doses of this inhibitor (200-1000 nM) were needed for displaying this effect. In the experiments reported here we showed that even 10 nM staurosporine blocked ISGF3 activation in HeLaM cells. At this low concentration, the inhibitor is highly specific for protein kinase C. Our results also demonstrated clearly that ISGF3a activation was the staurosporine-sensitive step. Involvement of protein kinase C in this step, either directly or indirectly is consistent with the nature of activation of ISGF3a. Although inhibiting the activity of protein kinase C blocked ISGF3a activation, activating the enzyme by treating the cells with PMA was not sufficient for its activation. Neither did down-regulation of cell surface protein kinase C by prolonged treatment with PMA impair the activation of ISGF3a by IFN-a. It is conceivable that the staurosporine-sensitive isozyme of protein kinase C is different from the one activated and down-regulated by PMA. Alternatively, protein kinase C activity may be a necessary but not sufficient requirement for ISGF3a activation in response to IFN-a treatment of cells.
Our in vitro experiments with alkaline phosphatase clearly indicated that active ISGF3a contain phosphate residues which are essential for its activity. Alkaline phosphatase removed these residues both from DNA bound and unbound ISGF3 complex as well as from the uncomplexed ISGF3a subunit. Acid phosphatase did not remove the same phosphate residues, since the DNA-binding activity of ISGF3 was not affected, although its mobility was changed presumably as a result of the removal of other phosphate residues. The alkaline phosphatase-sensitive and the acid phosphatase-sensitive phosphate residues could also be on different components of the ISGF3 complex. The in vitro inactivation of ISGF3a by alkaline phosphatase and the in vivo blockade of its activation by staurosporine strongly suggest that ISGF3a activation occurs via specific phosphorylation of the protein. As a corolary, physiological inactivation of ISGF3a may occur by specific dephosphorylation as mimicked by the in vitro alkaline phosphatase treatment. This process may contribute to the mechanism of desensitization of IFN-inducible gene transcription.
In HeLaM cells and probably in other cells, ISGF3-y is not only induced by IFN-y but by IFN-a as well. The ISGF3-y gene, therefore, is an IFN-a-inducible gene. Hence the dilemma, how is the transcription of this gene induced in HeLaM cells in the absence of the ISGF3-y protein which is needed for the formation of the crucial ISGF3 complex? In this respect, therefore, the ISGF3y gene is different from other known IFN-a-inducible genes. It is also the only known gene whose transcriptional induction in HeLaM cells by IFNa is a one-step process not requiring ongoing protein synthesis. Experiments similar to those shown in Fig. 2, but using IFN-a instead of IFN-y, led to the above conclusion (Ref. 21 and data not shown). Another major distinction of the ISGF3y gene as an IFN-a-inducible gene was unraveled by experiments reported in this paper. Staurosporine inhibited ISGF3a activation by IFN-a. As a result, transcriptional induction of IFN-a-inducible genes was blocked by this inhibitor. Induction of ISGF3-y by IFN-a was, on the other hand, totally insensitive to staurosporine. It appears therefore that, at least in HeLaM cells, transcriptional induction of the ISGF3y gene by IFN-a is mediated by an alternative route which requires neither the ISGF3-y protein nor active ISGF3a. In principle, there could be other as yet untested IFN-ainducible genes which share the induction pathway of ISGF3-y gene. One can speculate that these genes are activated not by ISGF3 but by some other trans-acting factor which binds to their ISREs. It is also conceivable that their induction is mediated not by ISRE but by a different cis-acting element. All genes carrying this putative cis-element may be directly inducible by both IFN-a and IFN--y without the involvement of protein kinase C.
Drawing upon the experimental observations noted in this paper and elsewhere (3), one can draw the following model for gene induction by IFN-a in HeLaM cells (Fig. 8). IFN-a binds to its cell surface receptor and initiates a chain reaction which involves a specific isozyme of protein kinase C. This leads to the phosphorylation of ISGF3a, in the cytoplasm, which activates this protein and results in its binding to ISGF3y and translocation to the nucleus. Activated ISGF3a can be readily inactivated by dephosphorylation, and hence it has a short functional half-life. It is not known whether the phosphorylation state of ISGF3a affects its binding to