Jan 5, 1990 - Department of Pathology and Kaplan Cancer Center, NYU School of. Medicine ...... from the Lucille P.Markey Charitable Trust. T.D. was the ...
The EMBO Journal vol.9 no.4 pp. 1 105 - 1111, 1990
Synergistic interaction between interferon-oa and interferon--y through induced synthesis of one subunit of the transcription factor ISGF3 David E.Levy, Daniel J.Lew1, Thomas Decker', Daniel S.Kessler' and James E.Darnell, Jr1 Department of Pathology and Kaplan Cancer Center, NYU School of Medicine, New York, NY 10016 and 'The Rockefeller University, New York, NY 10021, USA Communicated by G.Stark
Interferon-a (IFNa) and interferon-y (IFN7) each induce in susceptible target cells a state of resistance to viral replication and reduced cellular proliferation, presumably through different mechanisms: these two polypeptides are unrelated by primary sequence and act through distinct cell-surface receptors to induce expression of largely non-overlapping sets of genes. However, acting in concert, they can produce synergistic interactions leading to mutual reinforcement of the physiological response. In HeLa cells, this synergistic response was initiated by cooperative induction of IFNa stimulated genes (ISGs). These normally quiescent genes were rapidly induced to high rates of transcription following exposure of cells to IFNa. Although they were only negligibly responsive to IFN'y, combined treatment of cells with IFN-y followed by IFNcx resulted in an - 10-fold increase in ISG transcription. ISG transcription is dependent upon ISGF3, a positive transcription factor specific for a cis-acting regulatory element in ISG promoters. IFN,y treatment induced increased synthesis of latent ISGF3, which was subsequently activated in response to lFNa to form 10-fold higher levels than detected in cells treated with IFNa alone. ISGF3 is composed of two distinct polypeptide components, synthesis of one of which was induced by IFN-y, increasing its cellular abundance from limiting concentrations to a level which allowed formation of at least 10 times as much active ISGF3. Cell lines vary in their constitutive levels of the inducible component of ISGF3 and in the ability of LFNs to increase its synthesis. The cooperative induction of cytokine-specific transcription factors is one mechanism for producing reinforcing effects of distinct cell-surface ligands while still maintaining the specificities of the individual inducers. Key words: transcriptional control/DNA binding -
factors/ISRE/synergy/priming Introduction Interferons (IFNs) and other cytokines have significant effects on cell proliferation, immune responses and differentiation in diverse cell types (Pestka et al., 1987; Tamm et al., 1987), in large part due to activation of specific gene expression (Revel and Chebath, 1986). Although these extracellular signalling molecules operate through distinct Oxford University Press
cell surface receptors, they can induce similar biological effects and often induce overlapping sets of genes. The mechanisms of gene induction by IFNa and IFNy are probably distinct: they do not induce identical sets of genes (Revel and Chebath, 1986) and their signal transduction pathways are distinct even when inducing the same gene (Lew et al., 1989). In addition, synergistic effects of combined treatment with IFNa and IFN-y have been noted, again suggesting that they operate on similar genes through different mechanisms (Fleischmann et al., 1979; Zerial et al., 1982; Czarniecki et al., 1984; Broxmeyer et al., 1985; Raefsky et al., 1985; Dick and Hubbell, 1987; Hubbell et al., 1987; Carlo-Stella et al., 1988; Morikawa and Fidler, 1989). We have been studying the mechanism of induced expression of genes responsive to IFNa. A set of IFNa stimulated genes (ISGs) was found to be transcriptionally induced by IFNa following receptor binding in a rapid manner not requiring ongoing protein synthesis (Friedman et al., 1984; Larner et al., 1984; Friedman and Stark, 1985). A cis-acting IFNcx stimulated response element (ISRE) has been defined in ISG promoters (Friedman and Stark, 1985; Israel et al., 1986; Levy et al., 1986, 1988; Gribaudo et al., 1987; Reich et al., 1987; Sugita et al., 1987; Wathelet et al., 1987; Cohen et al., 1988; Korber et al., 1988; Porter et al., 1988; Reich and Damell, 1989) which is necessary and sufficient for transcriptional induction. The ISRE serves as the binding site for several nuclear factors, some of which are dependent upon IFNa (Cohen et al., 1988; Kessler et al., 1988a,b; Levy et al., 1988a,b; Porter et al., 1988; Rutherford et al., 1988; Shirayoshi et al., 1988; Dale et al., 1989b). We have identified the positive regulator of ISG promoters, termed ISGF3. This factor binds the ISRE with a defined specificity equivalent to the sequence required for transcriptional induction. It does not require new protein synthesis for activation and appears after treatment with IFNca with the same kinetics as ISG transcriptional induction (Kessler et al., 1988a; Levy et al., 1988a). We and others have recently shown that ISGF3 is activated in the cytoplasm of IFNa treated cells from a pre-existing latent precursor (Dale et al., 1989; Levy et al, 1989a). ISGF3 is composed of two polypeptides and its activation involves subunit association prior to being translocated to the nucleus and binding DNA. In the course of these studies on the activation of ISG promoters, we have analysed a particular example of synergy between IFNa and IFN-y which leads to greatly increased expression of IFNa responsive genes. We have found that maximal synergy is obtained when cells are pretreated with IFNy and allowed to accumulate a necessary protein factor prior to exposure to IFNa. In this report, we show that synergy between IFNa and IFN7y results in increased rates of transcription of IFNa stimulated genes. This increased transcription is brought about by IFN-y induced synthesis of one subunit of ISGF3, leading to greatly increased abundance 1 105
D.E.Levy et al.
of active ISGF3 following IFNa treatment. This paradigm for synergistic interactions suggests that distinct cytokines cooperate by directly modulating the levels and activities of gene-specific transcription factors. Cooperation in the induction of gene expression also has been observed between IFNs and double-stranded RNA (Havell and Vilcek, 1972; Enoch et al., 1986; Tiwari et al., 1987; Horisberger and Hochkeppel, 1987; Goetschy et al., 1989). However no transcription factors which mediate these forms of synergy have yet been identified.
Results IFNa and IFNy synergy induces high level expression of ISG54 ISG54 was originally described as a gene induced to high levels of transcription by IFNa or ,B (Lamer et al., 1984). This induction occurs in several human cell types (Kessler et al., 1988b), is rapid, transient and occurs without a requirement for ongoing protein synthesis (Lamer et al., 1986). Since IFN-y was known to produce very similar biological effects to IFNa, we tested whether IFNy treatment also caused an increase in ISG54 expression. Nuclei were isolated from HeLa cells following various periods of IFN treatment. RNA transcriptional elongation reactions were carried out in vitro in the presence of radiolabeled UTP, and the labeled RNA was hybridized to excess DNA complementary to a portion of the ISG54 second exon and assayed by autoradiography. This 'run-on' assay gives direct assessment of transcriptional activity (Evans et al., 1977; Weber et al., 1977; Derman et al., 1981). Short treatment of HeLa cells with IFNca (2 h) induced high levels of ISG54 transcription from essentially undetectable pretreatment levels (Figure lA). ISG54 showed little or no induction in cells treated for either 2 or 15 h with IFN'y. Although combined treatment with IFNa and IFNy had no discernable effect on peak transcription levels after 2 h of treatment, a 15 h pretreatment with IFN'y produced an 10-fold higher level of transcription following subsequent exposure to IFNa for 2 h. This prior treatment with IFN-y led to levels of ISG54 transcription 10-fold higher than those produced by a 2 h treatment with IFNa alone. The synergistic effect of IFN'y required extended treatment of cells with IFN-y prior to addition of IFNal (not shown). To investigate the nature of the IFN7y synergy with IFNae, we tested whether protein synthesis was required. HeLa cells were treated for 15 h with IFN-y and with IFN-y in the presence of 50,gg/ml cycloheximide, followed by a 1 h exposure to IFNat. Nuclei were isolated following such treatments and were assayed for transcriptional induction of ISG54 and several other IFN stimulated genes (Figure 1B). Three IFNa stimulated genes, ISG54, ISG15 and ISG56, were superinduced by IFNa following IFN-y pretreatment relative to their levels of transcription following treatment with IFNa alone (compare rows 2 and 4). However, when the IFN'y pretreatment was carried out in the presence of cycloheximide, transcriptional induction by IFNa was no different than a treatment with IFNa alone (compare rows 2 and 5) or a treatment with IFNa following cycloheximide alone (row 6). Therefore, the effect of IFN-y on HeLa cells, which creates a condition of greatly increased sensitivity to IFNa, requires ongoing protein synthesis and may involve proteins directly induced in response to IFNy treatment. -
Two other genes also responded to IFN-y pretreatment. The gene for a guanylate binding protein (GBP; Cheng et al., 1983) is itself responsive to both IFNa and IFN-y (Decker et al., 1989; Lew et al., 1989), showing an induction above baseline for both these treatments (rows 2 and 3). However, its transcription was induced to much higher levels by the combined treatment of IFN'y followed by IFNa (row 4) than was observed by IFN-y alone, and again this superinduction was eliminated by inhibition of protein synthesis (row 5). Although the level of GBP transcription increased with continued IFNy treatment, the superinduced level observed after combined treatment with IFN'y followed by IFNoa was significantly higher than that observed following treatment with IFN-y alone (not shown, Lew et al., 1989). The gene for metallothionein IIA was also superinduced by IFNax in HeLa cells primed with IFN-y (data not shown). However, transcription of the oligoadenylate synthetase (OAS) gene was not appreciably affected by IFN>y priming. We have previously observed very poor transcriptional response of this gene to IFN in HeLa cells (Kessler et al.,
1988b).
The levels of mRNA for ISG54 were also superinduced in IFN-y treated cells (Figure IC). Cytoplasmic RNA was isolated from HeLa cells following various treatments with IFN and quantitated for ISG expression by RNase protection procedures using radiolabeled complementary RNA probes (Melton et al., 1984; Levy et al., 1986). Barely detectable levels of ISG54 RNA were present following 15 h of IFNy, whereas a strong induction was observed following 2 h of IFNa. IFN'y treatment followed by 2 h of IFNat, however, led to approximately 6-fold higher levels of ISG54 RNA than observed following IFNa alone. ISG56 and ISG15 mRNA were also superinduced by IFNa in IFN-y pretreated cells, although the extent of increase over IFNa alone or IFN'y alone was less dramatic than for ISG54. ISG15 showed a significant level of induction of RNA in response to IFN-y alone, higher than might be expected based on its transcriptional response. In contrast, actin mRNA showed no change in abundance in response to any of the treatments, consistent with unchanged transcription and stability.
-
1106
IFNy pretreatment induces high levels of the latent precursor to ISGF3 We have previously described the dependence of ISG expression in IFNa treated cells on the positive transcription factor termed ISGF3 (Kessler et al., 1988a,b; Levy et al., 1988a). This factor is derived from pre-existing protein components that are activated in response to IFNa treatment to allow binding to the ISRE that is present in ISG promoters (Levy et al., 1988a; Dale et al., 1989a). ISGF3 activation involves at least three distinct steps. Modification of a latent polypeptide in the cytoplasm of IFNa treated cells leads to association of this polypeptide with at least one other factor. These two proteins are subsequently translocated to the nucleus where the complex binds to the ISRE (Levy et al., 1989). The activation of ISGF3 utilizes pre-existing protein components and, like ISG transcription, is resistant to
cycloheximide.
We tested the involvement of ISGF3 in IFN-y synergy with IFNa. Nuclear extracts were prepared from HeLa cells before and after treatment with IFNs and were assayed for the presence of ISGF3 and for the other ISRE binding factors ISGFl and ISGF2 (Levy et al., 1988a). DNA binding factors
A
0
B No
Treatment
IF N,
2h
IFNu..
i.J-
Synergistic induction of ISGF3
C IF\
IEN (a 1
'-,
I
-
I __
gil~~~~~~~~~
I
I IFN-/1S5h.
I
IEN-;
n----
IlFN
2
h
I
ISC54
No Irextliilri
I f-N\
r,
-
1\F -1
fl
-
IFN- 15 h
-
1F-y (CH
1SG -f6
FN u. -I "i
I
4
3ow I
h
-
3
2
-
IlEN .\
a
h
ICsG15
_
15 h -- IFNu {r
I
I
(CI IX
h
IFNcx
-r0
_ r
h
Actini
Fig. 1. Transcriptional superinduction of ISG54 in IFN-y primed cells. (A) and (B) Run-on assays were performed on HeLa S3 cells treated for the indicated times with 5 ng/ml IFN-y, 0.25 ng/ml IFNa or 50 jtg/ml cycloheximide. Co-treatments with cycloheximide involved pretreating cells for 10 min prior to addition of IFN for the times indicated. Co-treatments with IFNs were simultaneous addition of IFNa and IFN-y and addition of IFNa to the medium during the last 2 h (A) or 1 h (B) of IFN-y treatment. (C) Abundance of ISG54, 56 and 15, and -y-actin in untreated cells (lane 1) or in cells treated with 5 ng/ml IFN-y for 15 h (lanes 2 and 3) or with 0.25 ng/ml IFNa for 2 h (lanes 3 and 4) was measured by RNase protection. were analyzed by gel retardation using a DNA fragment containing the ISG15 ISRE sequence, and nuclear protein from equivalent numbers of cells was tested for ISRE binding activity (Figure 2). ISGF1, a constitutive ISRE binding factor, was found in approximately equal amounts in all extracts (lanes 1-6) and was competed by excess, unlabeled ISRE DNA (lane 7). ISGF2 was strongly induced by both IFN-y and IFNa treatment (lanes 2 and 3); we have found that this factor is produced by de novo synthesis from a mRNA transcribed from an IFNcx and IFN-y inducible gene (unpublished results). However, activity of ISGF2 does not correlate with transcriptional activation of ISGs and may rather be involved in some later event in the ISG transcriptional cycle (Levy et al., 1988a). In contrast, ISGF3 binding activity, which is the positive activator of ISGs, was present only in extracts from cells which had been treated with IFNt (lanes 3-6). Moderate levels of ISGF3 activity were observed in extracts from cells treated with IFNa alone (lane 3). IFN^y did not by itself induce ISGF3 activity (lane 2). However, very high levels of ISGF3 were detected in extracts from cells which had been first treated with IFN'y for 15 h prior to treatment with IFNa (lane 4). Thus, production of high levels of ISGF3 in IFN-y pretreated cells following subsequent treatment with IFNa correlated with the high levels of ISG transcription and abundant ISG mRNA produced in these cells. ISGF3 activation in IFNy treated cells was also resistant to inhibition of protein synthesis during IFNa treatment (lane 5), in agreement with previous findings on the cycloheximide insensitivity of IFNa induced transcription. Cells were treated for 14 h with IFN^y followed by a 1 h treatment with IFNa in the presence of cycloheximide. Equal amounts of
es
_
ac
cL
CH (X
A
A
y
y
(
A Clv
I'll --6
ISGF3 >
.s
ISGF2 W IS FIJ
Ac
.:M0
4_0
.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1 -1
3
4
6
7
Fig. 2. Superinduction of ISGF3 in IFN-y primed cells. Nuclear extracts were prepared from HeLa S3 cells following treatment with IFN-y at 5 ng/ml for 15 h (lanes 2, 4-7) and with IFNai at 0.25 ng/ml for 1 h (lanes 3, 4-7) in the presence of 50 ytg/ml cycloheximide during the IFNa treatment (lane 5) or the IFN-y treatment (lane 6). Actinomycin D (1 tg/ml) was included during the IFNcx treatment for the extract shown in lane 6 in order to inhibit further RNA transcription. Co-treatments involved addition of IFNa during the final hour of IFN-y treatments. Extracts were assayed for the presence of ISGFI, ISGF2 and ISGF3 by gel shift analysis using an ISRE containing fragment of the ISG15 promoter. Homologous competitor DNA (80-fold molar excess) was included in the binding reaction analyzed in lane 7 to demonstrate specificity. 1107
D.E.Levy et al.
Actinomycin D (1 jig/ml) was included during the 1 h IFNa treatment in order to prevent any further transcription of IFN-y induced genes. Under this regimen, no increase was observed over the amount of ISGF3 normally activated in non-primed cells. Similar results were obtained by treating cells with IFN-y in the presence of cycloheximide or actinomycin D followed by IFNcx treatment without first removing the inhibitory drugs (not shown). Thus, IFN-y contribution to high levels of ISGF3 activation required ongoing protein synthesis during the IFN^y exposure, in agreement with the requirement for protein synthesis in the induction of high levels of ISG transcription (see Figure 1). The de novo protein synthesis required for IFNy synergy also required continued RNA transcription, suggesting that the gene for the protein produced during pretreatment may itself be induced by IFNy.
ISGF3 activity were produced following this treatment as were produced by IFNa in IFN-y primed cells in which protein synthesis was not blocked. Thus, the conversion of inactive ISGF3 in IFN7y pretreated cells to its active, DNA binding form appears to be qualitatively the same as in untreated cells; only the quantity of active ISGF3 produced is changed. In contrast to the above results, protein synthesis was absolutely required during the IFN-y pretreatment stage. When HeLa cells were treated with IFN7y for 14 h in the presence of cycloheximide, no synergistic effect occurred (lane 6). Extracts were prepared from cells treated simultaneously with IFN-y and cycloheximide, and then washed free of cycloheximide and treated with IFNa. (f.-CYTOSOLNF7 r.1 7f-CYTOSOL C,
A;
)
23
88
7-,/- CYTOSOL -
r; 2
4 r
4
e8
'8.
