Induced Factor Binding to the Interferon-stimulated Response Element INTERFERON-a AND PLATELET-DERIVED GROWTH FACTOR UTILIZE DISTINCT SIGNALING PATHWAYS*

Interferon-a (IFNa) and platelet-derived growth factor (PDGF) each rapidly stimulate binding of nuclear factors from Balb/c 3T3 fibroblasts, to a 29-base pair regulatory sequence derived from the 5‘ upstream re- gion of the murine 2-5A synthetase gene. This regulatory sequence contains a functional interferon-stimu- lated response element (ISRE) and also functions as a PDGF-responsive sequence. We show that IFNa in- duces binding of a protein of molecular mass 65 kDa to the ISRE. Constitutively expressed ISRE-binding proteins of 98 and 150 kDa are also demonstrated. Excised complexes were loaded directly onto a 10% sodium dodecyl sulfate-PAGE system (26). The gel was then dried under vacuum and subjected to autoradiography. Plasmid Construction and Transient Transfection Assays-The ISRE oligonucleotide was ligated into the BarnHI site of the 5' poly linker of pBLcat2 (27), located 5' of the bacterial chloramphenicol acetyltransferase gene, driven by the herpes virus thymidine kinase promoter. The forward orientation was confirmed by diagnostic re- striction digests and sequencing. Transient transfections and chloramphenicol acetyltransferase activity assays were carried out as de- scribed (7, 13).

receptor interactions are unknown.
2-5A synthetase is a double-stranded RNA (dsRNA)-dependent enzyme whose expression is transcriptionally induced by IFNa (3). 2-5A synthetase catalyzes polymerization of adenylate residues into a series of 2',5'-linked oligomers. 2-5A oligomers transiently activate a latent cellular endoribonuclease which acts as a translational regulator of gene expression (reviewed in Ref. 2). 2-5A synthetase activity is also induced by epidermal growth factor (4) or dsRNA, and platelet-derived growth factor (PDGF) (5) in fibroblast cells and in rat PC12 cells by nerve growth factor (6). The induction of 2-5A synthetase mRNA by PDGF occurs in the absence of new protein synthesis, suggesting that it is a direct response to this growth factor (5, 7).
2-5A synthetase activity in the liver decreases dramatically shortly after partial hepatectomy in rats (8) and has been shown to accumulate to high levels late in S phase in synchronized mouse embryo fibroblasts (9). These results have led to speculation that 2-5A synthetase activity is involved in regulating fundamental aspects of cellular metabolism such as growth and differentiation. In direct support of this we have shown that plasmid-directed overexpression of 2-5A synthetase activity results in a marked reduction in both growth rate and colony size of a human glioblastoma cell line (10).
Recently, a number of groups have demonstrated the binding of IFNa-modulated nuclear factors to 5' cis-regulatory sequences of Type I IFN-inducible genes. We and others have identified one such regulatory sequence, residing in the 5' upstream region of the 2-5A synthetase gene (11-13), which is highly conserved in a number of IFNa-regulated genes of both mouse and human origin. This sequence functions as an IFN-stimulated response element (ISRE) (14) and specifically binds IFN-modulated nuclear factors in a manner which correlates with transcriptional activation of these genes. Factor binding, contact point analyses, and transcriptional studies have led to the elucidation of a consensus ISRE sequence, (A/G)GAAA(A/G)(N)GAAACT (where N is any nucleotide, Refs. [12][13][14][15][16][17]. Furthermore, an IFNa-inducible DNase I-hypersensitive site lies approximately 50 base pairs upstream of the ISRE in the IFN-inducible ISG-15 gene, implying that ISRE binding factors induce a transcriptionally open chromatin conformation immediately upstream of ISRE-containing genes (18).
We have also demonstrated that the 2-5A synthetase ISRE sequence is PDGF-responsive in Balb/c 3T3 cells (7). Since the 2-5A synthetase ISRE represents an early target of both mitogenic (PDGF) and growth-inhibitory (IFNa) signals, we have sought to characterize ISRE-specific binding proteins, as well as the intracellular signaling pathway(s) by which activation of ISRE binding occurs. Although protein kinase 8765 was random primer-laheled to a specific activity of >lo" cpm/pg. Hyhridization and washing was a t 42 "C for this probe.
Oligonucleotides-A double-stranded oligonucleotide, representing nucleotides -8O/-52 (relative to a putative translational start site) of the murine 2-5A synthetase gene (11)? which contains a functional ISRE, was synthesized on an Applied Riosystems 308A DNA synthesizer. An oligonucleotide representing nucleotides -55/-99 of the human IFN/j gene was synthesized for use as a control for ISRE binding reactions.

CTGAAAGGGAGAAGTGAAAGTGGGAAATTCCWTG.
Underlined in the ISRE is the binding element recognizing IFN-induced factors. Underlined in the IFNP IRE sequence are nucleotides involved in recognition of a factor exhibiting characteristics of NF-KB, which is activated by dsRNA (21)(22)(23). Oligonucleotides were synthesized with RarnHI-compatible linkers at the 5' terminus (GATC). Gel-purified oligonucleotides were mixed with an equimolar amount of their respective complements, heated to 65 "C for 15 min, and annealed at room temperature for 18 h. These preparations were used directly in labeling reactions. Extract Preparation and Electrophoretic Mobility Shift Assays-Electrophoretic mobility shift assays (EMSA) were done as previously described, using native 4% polyacrylamide gels run in a Tris/glycine buffer (pH 8.3) (7, 24). Nuclear extracts were prepared according to Dignam et al. (25), or whole cell extracts were prepared as follows. Cell pellets (2-4 X 10' cells) were extracted with buffer A (25) for 5 min on ice. Cells were then pelleted for 10 s a t 10,000 X g, and the supernatant was discarded. Pellets were washed twice more in buffer A and extracted with 100 pl of buffer C (25) for 5 min on ice. Cell debris was pelleted and the supernatant subjected to 15,000 X g for 30 min. This high salt, whole-cell extract (WCE) was dialyzed in 100 volumes of buffer D (25)  extraction steps were carried out a t 4 "C. Oligonucleotides were end-""PIATP.
labeled for use in EMSA by T4 DNA kinase in the presence of [y-UV-induced Cross-linking Assays-For cross-linking experiments, the ISRE synthetic oligonucleotide was random primer-laheled in the presence of [(u-:"'P]~ATP, according to standard procedures. TTP was suhstituted by bromodeoxy-UTP in order to render the labeled binding sites UV-sensitive. After incubation with nuclear or whole cell extracts (40-80 pg) binding reactions were resolved on native 6% polyacrylamide gels. These gels were then exposed to a 302 nm UV light for 60 min (FotoDyne UV 300 transilluminator), under an ice pack. The wet gel was then exposed to Kodak XAR-5 film and visualized complexes excised. Excised complexes were loaded directly onto a 10% sodium dodecyl sulfate-PAGE system (26). The gel was then dried under vacuum and subjected to autoradiography.
Plasmid Construction and Transient Transfection Assays-The ISRE oligonucleotide was ligated into the BarnHI site of the 5' poly linker of pBLcat2 (27), located 5' of the bacterial chloramphenicol acetyltransferase gene, driven by the herpes virus thymidine kinase promoter. The forward orientation was confirmed by diagnostic restriction digests and sequencing. Transient transfections and chloramphenicol acetyltransferase activity assays were carried out as described (7, 13).

RESULTS
Photochemical Cross-linking of IFNa-induced Proteins to the ISRE-We and others have identified both constitutive and IFNa-induced factors, which bind specifically to ISRE sequences found in the upstream of IFNa-inducible genes (12)(13)(14)(15)(16)(17). We sought to characterize the DNA-binding protein components of these factors in Balb/c 3T3 cells (clone A31) by covalent, UV-induced cross-linking of IFNa-induced extracts to a labeled ISRE oligonucleotide. The molecular masses of these ISRE-binding proteins were estimated by resolution of cross-linked complexes on denaturing polyacrylamide gels (Fig. 1). IFNa induced two ISRE-specific complexes ( A and B, of ISRE-binding proteins regulated by IFNa. High-salt extracts (60-70 pg) of quiescent A31 cells treated with rHuIFNnAD (1000 IU/ml) for 15 min were incubated with 10" cpm of hromodeoxyuridine-substituted, random hexamer-labeled ISRE oligonucleotide under standard conditions. a, reaction products from IFN-treated extracts were run on 6% native polyacrylamide gels and UV-irradiated (300 nm) for 1 h. Complexes marked A and H are IFN-induced, and C represents a constitutive binding complex (see Fig. 2). b, cross-linked complexes (A-C) were resolved in small scale binding reactions (10 pg/lane), and slices of native gel corresponding to each complex were then excised from the native gel. Slices corresponding to each complex were collected (5-61 lane) and analyzed on a 10% sodium dodecyl sulfate-PAGE system, for molecular mass estimation. Kd, kilodaltons. As expected, no crosslinked bands were obtained from regions of the UV-exposed native gel which did not correspond to A, B, or C (not shown). Autoradiography was for 3 days with an intensifying screen.
We isolated each of these complexes from native band mobility shift gels, for cross-linking analysis. Fig. 16 illustrates the results of such an experiment. Both IFN-induced complexes, A and B, contained a DNA-binding species of approximate molecular mass of 65 kDa. Constitutive complex C, seen in A31 nuclear and whole-cell extracts regardless of IFN treatment, contained 98-and 150-kDa DNA-binding species (Fig.   l b ) . A and B migrated more slowly than C in native gels but contained a smaller DNA binding component than C, suggesting the presence of additional, non-DNA-binding proteins in the activated A and B complexes.
PK Inhibitors Block PDGF-induced, but Not IFNa-induced, Gene Expression-To determine the nature of signaling pathways mediating ISRE-dependent gene activation by IFNa and PDGF, we have used the protein kinase inhibitors, staurosporine and K252a (28,29). 2-5A synthetase mRNA is induced by PDGF or IFNa/P, with similar kinetics (7). This activation has been shown, in both cases, to be mediated by the ISRE-containing sequence represented by the 29-base pair oligonucleotide used in the above analysis (7). When confluent A31 cells were pretreated with 10 nM staurosporine there was marked inhibition of 2-5A synthetase mRNA induction by PDGF, and no detectable induction with 100 nM staurosporine (Fig. 2). In contrast there was no inhibition of mRNA induction by IFNa at 10 nM staurosporine, with a slight inhibition seen a t 100 nM. Similar results were obtained with a structurally related inhibitor, K252a (28) (Fig. 2b). Direct activation of Ca"/phospholipid-depend- ent protein kinase C by treatment of cells with TPA did not induce detectable 2-5A synthetase mRNA (Fig. 2a). Control experiments indicated that both staurosporine and K252a treatment, at the higher concentration used, resulted in a marked generalized inhibition of protein phosphorylation (as measured by [y-:"P]ATP incorporation in TPA-and Bt2cAMP-stimulated cells) in the 3T3 cultures (data not shown).
In order to confirm that the effect on PDGF-induced 2-5A synthetase mRNA was the result of a transcriptional inhibition, we tested the effect of staurosporine on PDGF-induced, ISRE-dependent transcription in transient transfection assays. We have previously shown that a plasmid carrying the Escherichia coli chloramphenicol acetyltransferase gene under the control of a single ISRE (pMuISREcatF), is inducible by IFNa and PDGF in A31 cells (7). Pretreatment of A31 tranfectants with staurosporine blocked PDGF-induced chloramphenicol acetyltransferase activity (Table I). Thus inhibition of kinase activity by staurosporine inhibited the ISRE-dependent PDGF transcription response in these cells.
The above results suggest that ISRE-specific trans-acting factors are activated by different biochemical pathways responsive to PDGF-or IFNa-receptor interactions. To test this directly we employed a sensitive EMSA, which we have previously used to identify early PDGF-and IFN-modulated ISRE binding factors (7,13). Confluent A31 cells were treated for 2 and 15 min with PDGF or IFNa, and crude nuclear extracts or WCE were analyzed for ISRE binding, using the 29-base pair ISRE oligonucleotide as probe. Within 2 min of treatment IFNa-induced complexes (i.e. A and B of Fig. 1) were clearly evident (Fig. 3a), and these complexes showed a parallel increase in abundance a t 15 min post-treatment. In extracts from cells treated for 2 min with crude PDGF, no enhanced complex formation was seen, but by 15 min, PDGFinduced ISRE binding was clearly evident (Fig. 3a). T o ensure that equal amounts of active extract were assayed in each reaction, we measured binding to a labeled oligonucleotide representing regulatory sequences in the promoter of the human IFNP gene, and termed IRE anterferon gene _Regulatory Element, Ref. 30). A constant amount of constitutive binding to the IRE was observed regardless of cell treatment (Fig. 3b). PDGF-induced ISRE complexes comigrated in the native PAGE system with the IFNa-induced complexes A and B (Fig. 3a). While this result does not provide proof of identity between PDGF-and IFN-induced factors, the differential kinetic displayed in the induction of binding by IFNa and PDGF does suggest these two agents signal activation of  ISRE binding factor(s) through different receptor-coupled pathways. To examine more closely this apparent separation in signal transduction pathways, we employed the EMSA on extracts made from cells pretreated with the kinase inhibitors, staurosporine and K252a, prior to IFNa or PDGF treatment (Fig. 4a).
We confirmed that authentic murine IFNa//3 induced the same complex pattern as seen with the human hybrid rIFNaAD, used in the experiments shown in Fig. 3 (Fig. 4a,  lanes 1 and 2 ) . The induction of ISRE binding activity by IFNa was not blocked by either staurosporine or K252a (Fig.  4a, lanes 1-4), although staurosporine did inhibit this response to some degree. PDGF-induced ISRE binding, in contrast, was almost completely inhibited by either K252a or staurosporine treatment (Fig. 4a, lanes 5-9). It is possible the partial inhibition of IFN-induced binding by staurosporine is reflected in the inhibition of PDGF-induced binding, a much weaker response. The differential sensitivity of PDGF-and IFN-induced binding to K252a (compare lanes 4 and 9, Fig.  4a) makes this unlikely. rPDGF-BB homodimer and purified human PDGF-AB heterodimer preparations induced ISRE binding identical to crude PDGF, which was predominantly of the AB heterodimer type (Fig. 4a, lanes 5-7). These results indicate that IFNa-and PDGF-receptor interactions trigger different signaling pathways in the activation of ISRE-binding factors. This result is in accordance with the effects of these kinase inhibitors on PDGF-induced, ISRE-dependent transcription (Table I), and induction of 2-5A synthetase gene expression by these two agents (Fig. 2a).
Since staurosporine and K252a did not block IFN-induced ISRE binding, we tested for the effect of activators of kinasedependent signal transduction on induction of ISRE binding. Neither TPA (Ca"/phospholipid-dependent protein kinase C activator) or Bt2cAMP (CAMP-dependent protein kinase A activator) treatment of intact A31 cells activated ISRE bind- sensitive to protein kinase inhibitors. a, WCE of A31 cells were incubated (5 pg/reaction) with a bodylabeled ISRE oligonucleotide, with binding and gel conditions as in Fig. 3. Lanes represent cells treated (15 min) as follows: 0, untreated; 1 and 2, 1000 IU/ml rHuIFNaAD or murine IFNa/P, respectively; 3 and 4, as in 1, with 15-min pretreatment of 100 nM staurosporine or K252a, respectively. 5, 200 units/ml crude AB-PDGF; 6, 10 ng/ml rBB-PDGF; 7, 20 ng/ml highly purified AB-PDGF; 8 and 9, as in 5, with 15-min pretreatment a t 100 nM staurosporine or K252a, respectively; 10 and 11, as in 5, including 200 neutralizing units/ml 7F-D3 anti-MuIFNn/P or polyclonal anti-HuIFNP, respectively. Induced complexes (AIR, according to Fig. 1) are indicated. The major constitutive band is labeled C, as in Fig. 1. The minor, nonspecific band (starred) is not seen consistently in these experiments. The autoradiogram shown represents a 20-h film exposure. b, confluent cultures of A31 cells were untreated, or treated with 1000 IU/ml IFNa, 100 ng/ml TPA, or 1 mM Bt2cAMP for 15 min, then extracted as in a. WCE (5 pg) from each treatment was analyzed by EMSA using labeled ISRE as probe. TPA treatment of parallel cultures resulted in a marked stimulation of c-jos transcription, as determined by nuclear runoff analysis (not shown). Bt2cAMP activity is routinely assayed in our laboratory by induction of neurite outgrowth in a human neuroblastoma cell line. c, WCE of IFNa-induced A31 cultures were pretreated a t room temperature for 10 min prior to analysis of ISRE binding activity by EMSA with the following: no compound (0), 10 mM NEM, 10 mM dithiothreitol (DDT), or 10 mM each of NEM and dithiothreitol. Subsequent incubation with end-labeled ISRE was under otherwise standard EMSA conditions (4, a-c). Complexes are marked AIR as in Fig. 1. ing factors (Fig. 4b). This result, in conjunction with the kinase inhibitor data, indicates that stimulation of Ca"/ phospholipid-dependent protein kinase C or CAMP-dependent protein kinase A is not sufficient to activate ISRE binding factors, and further, that these two kinases are unlikely to be involved in IFNa signaling of ISRE factors.
IFNcu-induced 65-kDa ISRE Factor Has Characteristics of ISGF3"Recently, an IFNa-induced ISRE binding factor, ISGF3, has been identified in human cells, the activation of which correlates with IFN-induced transcription of the ISG54 and ISG15 genes. ISRE binding activity of HeLa cell ISGF3 is sensitive to treatment in uitro with NEM (31). Since binding of the IFNa-induced A31 complexes A and B (Fig. 1) is competed for by an ISRE derived from the human 2-5A synthetase gene, and the murine ISRE recognizes similar factors in extracts of IFN-treated human fibroblasts (data not shown), there appears to be conservation of IFN-regulated ISRE binding factors between these species. In order to determine whether the IFN-induced 65-kDa protein (Fig. l) of an IFN-inducible Enhancer 8769 might be a component of murine ISGF3, we treated extracts of IFNa-treated A31 cells with NEM in vitro. ISRE binding activity of the complex containing the 65-kDa protein (band Fig. 4c) was abolished by NEM treatment; thus this protein may be the DNA-binding subunit of murine ISGF3 (NEM would not necessarily have to work directly on p65). Identical results were obtained with the PDGF-induced ISRE complex (data not shown). Antibodies to MuIFNP Inhibited PDGF-induced 2-5A Synthetase m R N A Expression but Not ISRE Binding-Previous work on the PDGF-induced 2-5A synthetase response has suggested that it is an indirect result of IFNP induction (5,7). T o address this possibility, we included a neutralizing antibody, 7F-D3, to MuIFNp in the culture medium during PDGF treatment of A31 cells. These cells were treated for 6 h (mRNA, Fig. 2a) or 15 min (ISRE binding, Fig. 4a) with PDGF (with and without 200 neutralizing units of 7F-D3) or an equivalent amount of anti-HuIFNP as control. 7F-D3 efficiently inhibited PDGF-induced accumulation of 2-5A synthetase mRNA (Fig. Za), indicating that IFNp is involved in this response.
We next determined whether 7F-D3 had an inhibitory effect on PDGF induction of ISRE binding factors. This antibody completely inhibits PDGF-induced chloramphenicol acetyltransferase expression from pMuISREcatF (7). Surprisingly, PDGF induced ISRE binding activity regardless of the presence or absence of 7F-D3 antibody (Fig. 4a, compare lanes 5  and IO). Thus, induction of ISRE binding by PDGF did not appear to be mediated by IFNP. This apparent discrepancy with the 2-5A synthetase gene expression data (Fig. 2a) is discussed below.

DISCUSSION
A number of recent studies indicate that ISRE sequences are highly conserved in the 5' upstream region of a number of IFNa/P-inducible genes. This sequence confers inducibility by IFNy (32), IFNa (12)(13)(14)(15)(16)(17), PDGF (7), and dsRNA (data not shown) on promoters normally unresponsive to these agents. Here we have used photoaffinity labeling experiments to reveal a minimum of three nuclear proteins which specifically contact the ISRE, exhibiting apparent molecular masses of 150,98, and 65 kDa. Only the 65-kDa protein bound to the ISRE sequence in an IFN-dependent manner. PDGF induced a n identical pattern of ISRE-specific complex formation as IFNa (Figs. 3 and 4) on nondenaturing gels, but the abundance of the PDGF-induced complex was much less than was seen with IFN ( Fig. 3a). Thus we have not been able to definitively cross-link a 65-kDa (or any other) induced protein in extracts from PDGF-treated cells. Proof of the identity between PDGF-and IFNa-induced ISRE binding factors will require more rigorous biochemical analyses of these induced complexes. We propose the identity of the IFNa-induced 65-kDa ISRE-binding protein as p65ISp, for 65-kDa IFN-stimulated protein.
~65'" is almost certainly a component of the E (early) factor described by Stark, Kerr, and colleagues (15), which binds to the ISRE sequence of the IFNa-inducible 6-16 gene. A similar factor, ISGF3, has been described (31), which mediates activation of the ISRE sequence of ISG-54 and ISG-15 genes (16). Active ISGF3 is a heteromeric complex of two identified protein subunits, ISGF3a and ISGF3y. IFNa-stimulates the stoichiometric association of ISGF3a and ISGF37 in the cytoplasm of HeLa cells, followed by nuclear translocation (31). It has not been established which ISGF3 subunit binds to the ISRE. Similarly, active E factor is found in purified cytoplasts of human lymphoblastoid cells after IFNa treatment, in a form that is rapidly translocated to the nucleus in intact cells (15). The 6-16 gene ISRE sequence competes for induced factor binding to the 2-5A synthetase ISRE in vitro,' These reports are consistent with identity between ISGF3 and E factor. Formation of HeLa cell ISGF3 i n vitro was sensitive to NEM treatment, and NEM treatment of A31 extracts prevented formation of the IFNa-induced ~65'" complex in uitro (Fig. 4c). Thus ~65'"' may represent a DNAbinding subunit of murine ISGF3.
Another broadly inducible factor, NF-KB, undergoes nuclear translocation after TPA treatment of intact cells (21). Activation of NF-KB and nuclear translocation can be catalyzed in vitro by addition of the purified subunit of either Ca2+/ phospholipid-dependent protein kinase C or CAMP-dependent protein kinase A, directly implicating these two kinases in signal transduction and physiological activation of NF-KB (33). NF-KB has recently been shown to be dsRNA-inducible (22,23) and binds to the positive regulatory domain I1 in the dsRNA-inducible IFNp gene IRE (30, 34). The 2-5A synthetase ISRE is also highly inducible by dsRNA in transient transcription assays, suggesting that NF-KB might be interacting with the ISRE (data not shown), consistent with the proposed overlap in factors mediating inducibility of IFN genes and IFN-responsive genes (35). A non-DNA-binding cytoplasmic protein of 65 kDa has been identified as an integral part of inducible NF-KB (36). However, ~65'" is distinct from the 65-kDa NF-KB subunit on the basis of its DNA binding activity. This distinction is further evident from our experiments with a dsRNA-inducible IRE sequence, shown in Fig. 3b. This sequence spans nucleotides -99/-55 of the IFNP gene, and contains four GAAANN motifs (35), including one centered in the NF-KB-binding site (23). Two of these GAAANN motifs are also present, as direct repeats, in the core ISRE. Neither IFNa or PDGF induced factor binding to the IFNp IRE site. In addition, the IFNP IRE sequence did not compete with the ISRE for binding of ~65'" in vitro (data not shown). Finally, in contrast to NF-KB activity, which is unmasked by deoxycholate (21) IFNa-induced ISRE binding (E factor) is abolished by treatment of cytoplasmic extracts with deoxycholate (15).
We have previously shown that, like staurosporine and K252a, antibodies to murine IFNp block PDGF induction of the ISRE/chloramphenicol acetyltransferase hybrid gene in A31 cells (7). Therefore, since IFNp is active at picomolar concentrations, significant concentrations of the polypeptide may be synthesized in response to PDGF, even in the presence of cycloheximide, which inhibits protein synthesis incompletely (up to 95%; see "Discussion" in Ref. 37). This suggested the possibility that PDGF induces 2-5A synthetase gene expression indirectly, through the kinase-dependent induction of IFNP. In order to test whether such an indirect mechanism is responsible for PDGF modulation of ISREbinding factors, we determined the effect of a monoclonal antibody to murine IFNp on the induction of ISRE factors by PDGF (Fig. 4). This is one of the antibodies shown to block PDGF-induced ISRE/chloramphenicol acetyltransferase activity in transient transcription assays (7). Surprisingly, we observed that PDGF stimulated ISRE binding in the presence of this antibody, even though PDGF-induced 2-5A synthetase mRNA accumulation was blocked under these same conditions (Fig. 2a). Similarly, it has recently been shown that the protein kinase inhibitor, H-7, blocks IFNinduced 2-5A synthetase mRNA accumulation in human lymphoblastoid cells without affecting IFN-induced transcription (38). H7 does not inhibit IFNa induction of ISRE binding factors (data not shown). Thus, early signals leading to acti-It has recently been shown that binding of heat shock factor the case of the PDGF-induced 2-5A synthetase gene, we conclude that activation of ISRE binding factors is a direct, early response to PDGF, but that this activation alone is inadequate for initiation of transcription from the ISRE (our data do not exclude the involvement of non-ISRE sequences in the overall response of 2-5A synthetase to PDGF). This observation may hold true for other ISRE-containing genes, and suggests that PDGF and IFNp cooperate in the full induction of ISRE-directed gene expression, early in PDGFinduced mitogenesis. We have previously shown that small amounts of IFN (detectable only by radioreceptor assay and antibody neutralization) are produced by confluent, arrested cultures of a human glial cell line (3). IFN is likely also present in the culture medium of the confluent 3T3 cells used here.
IFNa has been proposed to signal cells through rapid generation of diacylglycerol and inositol 1,4,5-triphosphate second messengers, suggesting activation of a Ca*+/phospholipiddependent protein kinase C in IFNa signal transduction (40). In support of this notion, IFNa has been reported to induce c-fos transcription (41) and diacylglycerol production (39,42) similar to PDGF. These results imply that mitogenic and growth inhibitory factors share early transmembrane signal transduction pathways.
We have shown here that PDGF and IFNa utilize signaling pathways exhibiting distinct kinase dependencies, in the transcriptional activation of a common pattern of ISRE-dependent gene expression in 3T3 cells. Our data leave open the question of whether signaling events upstream of kinase activation might be shared by PDGF and IFNa. Clearly, there is an advantage to the cell in distinguishing transmembrane signal transduction pathways triggered by activation of specific growth stimulatory (PDGF) and growth inhibitory (IFNa) receptors.