In virus-infected cells the transcription factors ATF-2, c-Jun, interferon regulatory factor IRF-3, IRF-7 and NF-jB, and the coactivators p300/CBP play critical roles in the activa-tion
Trang 1Mechanism for transcriptional synergy between interferon regulatory
Hongmei Yang1, Gang Ma1, Charles H Lin2,*, Melissa Orr1and Marc G Wathelet1
1
Department of Molecular and Cellular Physiology, University of Cincinnati College of Medicine, Cincinnati, OH, USA;
2
Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA, USA
The interferon-b promoter has been studied extensively as a
model system for combinatorial transcriptional regulation
In virus-infected cells the transcription factors ATF-2, c-Jun,
interferon regulatory factor (IRF)-3, IRF-7 and NF-jB, and
the coactivators p300/CBP play critical roles in the
activa-tion of this and other promoters It remains unclear,
how-ever, why most other combinations of AP-1, IRF and Rel
proteins fail to activate the interferon-b gene Here we have
explored how different IRFs may cooperate with other
fac-tors to activate transcription First we showed in
undiffer-entiated embryonic carcinoma cells that ectopic expression
of either IRF-3 or IRF-7, but not IRF-1, was sufficient
to allow virus-dependent activation of the interferon-b
promoter Moreover, the activity of IRF-3 and IRF-7 was strongly affected by promoter context, with IRF-7 prefer-entially being recruited to the natural interferon-b promoter
We fully reconstituted activation of this promoter in insect cells Maximal synergy required IRF-3 and IRF-7 but not IRF-1, and was strongly dependent on the presence of p300/ CBP, even when these coactivators only modestly affected the activity of each factor by itself These results suggest that specificity in activation of the interferon-b gene depends on a unique promoter context and on the role played by coacti-vators as architectural factors
Keywords: coactivator; interferon; IRF; synergy; virus
Specificity in transcriptional regulation is thought to derive
in part from the combinatorial assembly of unique
complexes of transcription factors at target promoters
Studies of the virus-inducible interferon (IFN)-b gene
promoter support this paradigm but the molecular basis
for its tissue- and stimulus-specific expression remains
incompletely understood (reviewed in [1])
Cells from vertebrate organisms respond to viral infection
by activating antiviral enzymes and by modulating the
expression levels of a set of cellular genes, some of which
encode cytokines such as IFNs (reviewed in [2,3]) These
cytokines signal the occurrence of an infection to other cells,
allowing coordination of the adaptive response at the
organismal level The IFN-b gene plays a crucial role in
initiating and sustaining this response through its early
direct transcriptional activation in infected cells and through
its ability to enhance the induction by virus of the family of IFN-a genes (reviewed in [4])
The molecular basis for the regulation of IFN-b tran-scription has been partially elucidated (Fig 1A) A compact virus-inducible enhancer controls this intronless gene (reviewed in [1]), and flanking scaffold/matrix-attachment regions (S/MARs) insulate the transcription unit from the influence of other regulatory elements (reviewed in [5])
In vivo, IFN-b is essentially silent in uninfected cells, with less than one copy of mRNA detected per 100 000 cells [6] The uninduced state is maintained, at least in part, through the inhibitory effects of an NF-jB regulating factor (NRF; [7,8]), YinYang 1 [9] and nucleosomes Nucleosomes are ordered immediately upstream from the gene [10,11] and their histone tails are hypoacetylated [12] The treatment
of cells with histone deacetylase inhibitors also leads to significant transcription from the IFN-b gene promoter in the absence of virus infection [13,14]
The IFN-b promoter contains binding sites for members
of the AP-1, IRF and Rel families These cis-acting elements are called positive regulatory domains (PRDs) and are located between)99 and )55 relative to the transcription initiation site Virus infection results in the coordinate activation of ATF-2/c-Jun, virus-activated factor (VAF) and NF-jB [15] ATF-2/c-Jun binds to PRD IV ()99 to )91); VAF contains IRF-3/IRF-7 and binds to PRD III-PRD I (known as P31, )90 to )64) while the p50/p65 NF-jB dimer binds to PRD II ()66 to )55) (Fig 1A) VAF also contains the coactivators p300 and CREB binding protein (CBP), which are thought to play a critical role in activation of the IFN-b promoter because interactions between p300/CBP and both ATF-2/c-Jun and NF-jB could compensate for the low intrinsic affinity of IRF-3/7
Correspondence to M G Wathelet, Department of Molecular and
Cellular Physiology, University of Cincinnati College of Medicine,
231 Albert Sabin Way, Cincinnati, OH 45267-0576, USA.
Fax: +1 513 558 5738, Tel.: +1 513 558 4515,
E-mail: marc.wathelet@uc.edu
Abbreviations: CAT, chloramphenicol acetyl transferase; CREB,
cAMP response element binding protein; CBP, CREB-binding
pro-tein; DOC, deoxycholate; GST, glutathione S-transferase; IFN,
interferon; IRF, IFN regulatory factor; ISRE, IFN stimulated
response element; NRF, NF-jB regulatory factor; PRD, positive
regulatory domain; VAF, virus activated factor; WT, wild type.
*Present address: Department of Cellular and Molecular Medicine,
UCSD School of Medicine, La Jolla, CA 92093, USA.
(Received 26 April 2004, revised 20 July 2004, accepted 28 July 2004)
Trang 2for this promoter (reviewed in [1]) ATF-2, c-Jun, IRF-3,
IRF-7, p50 and p65 are found associated with the IFN-b
promoter in vivo in virus-infected cells [15] Their binding
to the IFN-b promoter is accompanied by the localized
acetylation of histone tails in neighboring nucleosomes [12],
remodeling of these nucleosomes, recruitment of the
tran-scriptional machinery and trantran-scriptional activation [11]
Besides virus infection, many stressing or inflammatory
stimuli can coordinately activate members of the AP-1, IRF
and Rel families of transcription factors However, it
remains unclear why only the set of factors activated upon
virus infection (or upon lipopolysaccharide treatment in
some cells [16]) is able to turn the IFN-b gene on
Comparison of the sets of factors activated by different
stimuli suggests that a key determinant in specificity is the
nature of the IRF molecules involved Specifically, most
stimuli that activate AP-1 and NF-jB also induce IRF-1
(e.g interleukin-1, tumor necrosis factor [17]) but fail to
activate the IFN-b gene substantially By contrast, virus infection [15] or lipopolysaccharide treatment additionally activate IRF-3 and/or IRF-7, and consequently the IFN-b gene Moreover, it is not understood why type I IFN genes can be activated by virus infection in most adult cells but not
in pluripotent cells, such as embryonic stem cells or undifferentiated embryonic carcinoma cells [18,19] Here we explore the mechanism by which different IRFs functionally interact with ATF-2/c-Jun and NF-jB to activate IFN-b Ectopic expression of either 3 or
IRF-7, but not IRF-1, was sufficient to allow virus-dependent activation of the IFN-b promoter in undifferentiated embryonic carcinoma cells These cells, as well as insect cells, were used to define the role played by each transcrip-tion factor and coactivator in activatranscrip-tion of the IFN-b promoter We show that activation of the IFN-b promoter was critically dependent on the nature of the IRF involved Moreover, we show that synergy between different
tran-A
B
0
2
4
6
8
10
12
14
16
18
values 0 0 2 0 0 2 0 4 1 0 3 8 0 0 2 1 5 2 0 0 5 0 2 2 0 2 2 1 2 0
0 2 4 6 8 10 12 14
values 0 0 6 0 1 1 0 4 4 0 4 2 0 0 6 0 9 8 0 3 7 3 2 1 0 5 0 9 8 1
Fig 1 The IFN-b gene locus and promoter context dependence (A) Schematic representation of the human IFN-b gene locus, including flanking S/ MARs; the virus-responsive element (VRE) and the factors binding to it in the uninduced and virus-induced states are indicated (B) Activity of P31·2CAT and )110IFNbCAT in P19 cells P19 cells in 6-well plates were cotransfected with 1 lg cytomegalovirus (CMV)-lacZ, 2 lg reporter plasmid P31·2CAT (left panel) or )110IFNbCAT (right panel) and CMV-driven vectors directing the expression of the indicated transcription factors (HuIRF-1, 3 lg; H6HuIRF-3, 1 lg; F3HuIRF-7B, 2 lg) CAT activity and b-galactosidase activity were measured in extracts of transfected cells infected with Sendai virus (SV) or mock-infected as control (Co) CAT activity was normalized to b-galactosidase activity to control for transfection efficiency and is expressed in arbitrary units, rather than fold induction, so that the relative strength of different reporters can be compared Fold induction can be computed from the values listed under the graph.
Trang 3scription factors was strongly dependent on the presence
of p300 or CBP, even when these coactivators had only
a modest effect on the transcriptional activity of each
transcription factor alone
Experimental procedures
Plasmid constructs and sequence analysis
Effector constructs for transient transfections of
mamma-lian and insect cells are in the pcDbA and pPac vectors,
respectively Reporter constructs consist of one or more
copies of a cis-acting element driving expression of the
chloramphenicol acetyl transferase (CAT) gene through the
E1b TATA box, except the IFN-b promoter ()110 to +20)
construct, which is driven by its own TATA box [15,20–23]
Cell culture and transfections
P19 cells were grown at 37C, 5% (v/v) CO2, in Dulbecco’s
modified Eagle’s medium containing 10% (v/v) fetal bovine
serum, 50 UÆmL)1penicillin and 50 lgÆmL)1streptomycin
S2 cells were grown at 26C, in Schneider’s Drosophila
medium containing 12% (v/v) fetal bovine serum,
50 UÆmL)1penicillin and 50 lgÆmL)1streptomycin
Transfections using the calcium phosphate
coprecipitat-ion technique were as described previously [24] P19 cells
were seeded in 6-well plates (300 000 cells in 3 mL),
transfected the next day with 0.3 mL of a precipitate
containing 2 lg reporter, 1 lg pCMV-lacZ and 1–3 lg of
effector plasmid (with pcDbA added to a total of 6 lg) for
18 h Cells were then washed three times with NaCl/Piand
further incubated with medium until harvested 2 days after
transfection Sendai virus was added for the last 18 h of
transfection Sendai virus was obtained from SPAFAS
(North Franklin, CT, USA) and used at 200 hemagglutinin
unitsÆmL)1
S2 cells were seeded in 6-well plates (3 million cells in
3 mL), transfected the next day with 0.3 mL of a precipitate
containing 250 ng hsp82lacZ, 500 ng reporter plasmid and
effector plasmid mixes as indicated in the figure legends
(with pPac added to a total of 5.75 lg), and harvested
2 days after transfection
CAT and b-galactosidase activities were measured in
extracts of transfected cells [24], and CAT activity
was expressed in arbitrary units after normalization to
b-galactosidase activity to control for transfection efficiency
Variation in transfection efficiency between samples was
normal Arbitrary units rather than fold activation was used
in most Figures herein so that the relative strength of
reporters can be compared Basal activity of a reporter
displayed the most variation from experiment to
experi-ment, presumably because the effect of small fluctuations is
most visible on low values of CAT activity As a result, the
net fold activation for the IFN-b promoter is different in
different experiments
Pull-down experiments
The GST–p300/CBP fusions were described [25], as were the
GST–ATF-2 and GST–c-Jun fusions [22] GST–IRF-3/7
were generated by subcloning previously described cDNA
inserts [20,21] and verified by sequencing GST fusions were expressed in Escherichia coli BL21 and purified as recom-mended (Pharmacia), and dialyzed against phosphate-buffered saline/10% (v/v) glycerol
In vitro translation in rabbit reticulocyte lysates was performed as recommended using the TnT kit (Promega), appropriately linearized pcDbA effector plasmids and T7 RNA polymerase
35S-Labeled in vitro translated proteins were incubated with GST fusion proteins immobilized on glutathione– sepharose beads in 150 mM KCl, 20 mM Tris, pH 8.0, 0.5 mMdithiothreitol, 50 lgÆmL)1ethidium bromide, 0.2% (v/v) NP-40 and 0.2% (w/v) BSA (binding buffer) for 1 h at
4C, followed by two washes with binding buffer and two washes with binding buffer without BSA Proteins bound to the beads were eluted with SDS loading buffer and analyzed
by SDS/PAGE, visualized by autoradiography and quan-tified with a phosphoimager
Results
Transcriptional activity of IRF-3 and IRF-7 is dependent
on promoter context The activity of a transcription factor depends on the promoter context, which refers both to the specific arrange-ment of the cis-acting elearrange-ments in the promoter and to the nature of the factors they bind To investigate the effect of promoter context on the transcriptional activity of IRFs, we used undifferentiated P19 cells and two reporter plasmids, P31·2CAT and)110IFNbCAT P31 is the binding site for IRFs in the IFN-b gene promoter (Fig 1A) and P31·2CAT contains two copies of P31 driving the expression of the CAT reporter through the E1b TATA box This artificial context isolates the contribution of P31 from that of PRD IV and PRD II In the )110IFNbCAT reporter, in contrast, P31 is in its natural context P19 cells were chosen for these experiments because in the absence of cotrans-fected IRF both reporters had very little intrinsic activity and this activity was not significantly stimulated upon virus infection (Fig 1B) Cotransfection of IRF-1, a constitutive activator, stimulated each reporter to a similar extent and had no effect on their virus-inducibility Cotransfection of either IRF-3 or IRF-7 made both reporters virus-inducible, while cotransfection of both IRF-3 and IRF-7 had a synergistic effect, making both reporters strongly virus-inducible (Fig 1B) Intriguingly, the effects of IRF-3 and IRF-7 were dramatically affected by context IRF-3 stimu-lated P31·2CAT activity in P19 cells infected by Sendai virus 75-fold, as compared to 11-fold for IRF-7 (and
600-fold for IRF-3 + IRF-7) By contrast, IRF-3 stimu-lated virus-induced )110IFNbCAT activity only nine-fold, while IRF-7 stimulated it 29-fold (and 90-fold for IRF-3 + IRF-7) Thus, the ability of IRF-3 to stimulate P31 was about eight times stronger in an isolated context than within its natural context, while the ability of IRF-7 to stimulate P31 was about 2.6 times stronger in its natural context than in isolation We conclude that both proteins are required for maximal activation of the IFN-b gene and that there are interactions unique to the arrangement of regulatory elements in the promoter that favor the involv-ment of IRF-7 in its activation
Trang 4Benefits of using insect cells to reconstitute the
activation of theIFN-b gene
Some of the genes encoding factors involved in IFN-b
expression have been inactivated by gene targeting
(reviewed in [26]) However, functional redundancy in
transcription factor families and the lack of viability
resulting from gene targeting of either p300 or CBP places
restrictions on the use of mammalian cells to dissect the
activation mechanism of the IFN-b gene The IFN system
is restricted to vertebrates and insect cells do not contain
IRFs orthologs Moreover, insect cells contain a p300/CBP
ortholog that is sufficiently distinct from the mammalian
proteins that it cannot substitute for them to enable
IRF-3-dependent transcription [21], making insect cells an
ideal system to dissect the roles of individual factors in
activation of the IFN-b gene Furthermore, mammalian
ATF-2/c-Jun, IRF-1 and NF-jB have been shown to be
transcriptionally active in the Schneider S2 cell line [23] In
contrast to IRF-1, both IRF-3 and IRF-7 require
virus-dependent phosphorylation of specific residues in their
C-termini to display transcriptional activity These
modi-fications cannot take place in S2 cells, as they lack the
relevant kinase(s), but we have shown that mutant forms of
these proteins, IRF-3E7 and IRF-7Di, which are active in
mammalian cells, are also transcriptionally active in S2 cells
[20,21] (Fig 2)
Transcription factor activities are selectively affected
by coactivators Interactions between individual transcription factors or between transcription factors and coactivators that are specific to the IFN-b promoter must account at least in part for the results obtained in P19 cells Therefore, we inves-tigated these interactions both at the physical and functional levels First, we tested the ability of mammalian p300 and CBP, alone or in combination, to affect the activity of mammalian transcription factors in S2 cells (Fig 2) As described previously, IRF-3E7 did not activate transcription from an IFN stimulated response element (ISRE)-driven reporter in the absence of murine (m)CBP in S2 cells [21] Coexpression of either mCBP or human (h)p300 allowed 3-dependent transcription (Fig 2A) By contrast, IRF-7Di displayed intrinsic transcriptional activity [20], which was further stimulated by either hp300 (approximately twofold) or mCBP (approximately fourfold) Coexpression
of p300/CBP had little effect, if any, on IRF-1 transcrip-tional activity in S2 cells Thus, mCBP proved twice as active as hp300 for both IRF-7Di and IRF-3E7, consistent with the observation that both IRFs interact more strongly with mCBP than with hp300 [20,21] (Fig 3D) Interestingly, the combination of p300/CBP was more effective than either coactivator alone in the case of IRF-3E7, while a similar synergy was not observed with IRF-7Di
0 5 10 15 20 25 30 35 40
Vector IRF-3E7 IRF-7 ∆ I F3E7/F7 ∆ I
0 5 10 15 20 25 30 35 40
0.0
0.5
1.0
1.5
2.0
2.5
0
10
20
30
40
50
60
70
ISREx3CAT in S2 cells P31x4CAT in S2 cells
PRDIVx6CAT in S2 cells PRDIIx3CAT in S2 cells
Vector hp300 mCBP hp300/mCBP
Vector hp300 mCBP hp300/mCBP
C
Fig 2 Effects of hp300 and mCBP expression on the activity of transcription factors in S2 cells (A) Transcriptional activity of IRF-1, IRF-7Di and IRF-3E7 (0.5 lg) in the presence or absence of cotransfected hp300, mCBP and hp300/mCBP (1.5 lg) on the ISRE·3CAT reporter The value for vector alone was 0.24 and the value for IRF-3E7 without coactivator was 0.16 (B) Transcriptional activity of IRF-3E7 (1.5 lg) and IRF-7Di (2 lg), alone or in combination and in the presence or absence of cotransfected hp300, mCBP and hp300/mCBP (1.5 lg) on the P31·4CAT reporter The value for vector alone was 0.22 (C) Transcriptional activity of ATF-2 (0.15 lg) and c-Jun (1.5 lg) in the presence or absence of cotransfected hp300, mCBP and hp300/mCBP (1.5 lg) on the PRDIV·6CAT reporter (D) Transcriptional activity of p50 (0.1 lg) and p65 (0.15 lg) in the presence or absence of cotransfected hp300, mCBP and hp300/mCBP (0.5 lg) on the PRDII·3CAT reporter The value for vector alone was 0.06.
Trang 5IRF-3 and IRF-7 each bind with much higher affinity to
the ISRE of IFN- and virus-inducible genes than to the P31
sequence within the IFN-b promoter [15] Accordingly,
IRF-3 and IRF-7 can individually activate an ISRE-driven
reporter, but significant activation of a P31-driven reporter
requires cooperation between IRF-3 and IRF-7 in S2 cells
[20] As shown in Fig 2B, hp300 was relatively ineffective
in promoting synergy between IRF-3 and IRF-7 on the
P31·4CAT reporter as compared to mCBP, and the p300/
CBP combination stimulated activity to an intermediary
level
Expression of ATF-2/c-Jun in S2 cells resulted in
increased activity of a PRDIV-driven reporter, and
coex-pression of p300 and/or CBP further stimulated it up to
twofold (Fig 2C) By contrast, expression of the NF-jB
dimer p50/p65 (known as nfkb1/RelA) led to a strong
activation of a PRD II-driven reporter that, if anything, was
slightly inhibited by coexpression of the p300/CBP
coacti-vators Thus all the transcription factors known to bind the
IFN-b gene promoter in virus-infected cells can be expressed
and activate transcription in insect cells, and the p300 and
CBP coactivators have distinct and specific effects on their
transcriptional activity
Transcription factors interact with multiple domains
of coactivators
We have previously mapped the domains responsible for
interactions between IRF-3 or IRF-7 and p300/CBP [20,21]
(summarized in Fig 3) Similarly, others have mapped
interactions of hATF-2, c-Jun, p65 or IRF-1 with p300/
CBP However, not all domains of p300 and CBP were
tested in these experiments and the results were somewhat
conflicting [25,27–30] Therefore, we conducted a systematic
analysis of the domains within p300 and CBP that interact
with mATF-2 (the shorter murine activating form of
ATF-2), c-Jun, p50, p65 and IRF-1, and the result of these experiments are summarized in Fig 3 Binding of
mATF-2195to GST-p300/CBP was undetectable in our standard assay conditions, but lowering the salt concentration from
150 to 75 mM salt allowed detection of relatively weak (binding£ 4% input) interactions with CBP-N, CBP-C2, p300-N, p300-M and p300-C2 Binding of c-Jun was stronger (up to 40% input) but mapped to the same domains Thus both ATF-2 and c-Jun can bind to p300 and CBP through multiple domains, with a preference of c-Jun for the N- and C-terminal regions and of ATF-2 for the central region of the coactivators
Binding of p50 (amino acids 1–503 of p105) to GST– p300/CBP was very weak overall Binding to CBP–N averaged to 1.5% of input and binding to other GST–p300/ CBP fusions did not exceed 0.5% of input By contrast, p65 bound strongly to the N-, C1- and C2-regions of p300, and
to the N- and C2-regions of CBP Thus, the bulk of the interaction between NF-jB and p300/CBP is mediated by the p65 subunit through the N- and C-terminal regions of the coactivators
The pattern of IRF-1 binding to p300 and CBP domains closely resembled that observed for IRF-7 [20], with relatively strong binding to CBP–N ( 21%), –N2 ( 6%) and –C2 ( 6%) (but weak binding to CBP-C, 1%), and to p300–N ( 11%), –C ( 14%), –C1 ( 7%) and –C2 ( 27%) Synergy between ATF-2/c-Jun and IRF-3/IRF-7 The affinity of ATF-2/c-Jun and of IRF-3/IRF-7 for their target sites within the IFN-b promoter are significantly lower than that for optimal binding sites For example, a reporter driven by a single P31 is not virus-inducible, while a reporter driven by a single ISRE, which binds IRF-3/IRF-7 with higher affinity, is virus-inducible; three copies of P31 are required to make a reporter strongly virus-inducible [15]
nfkb1 (p50) c-Jun ATF-2195 IRF-7∆i
RelA (p65)
hp300
IRF-1 mCBP
IRF-3E7
Fig 3 Mapping of domains of hp300/mCBP interacting with transcription factors Summary of interaction studies between domains of the p300 and CBP coactivators and the transcription factors that can bind to the IFN-b gene promoter (primary data not shown and [20,21]) The following domains were used in pull-down experiments (with the amino acid coordinates indicated in parentheses): CBP–N1(1–267); CBP–N2(267–462); CBP–N3(462–661); CBP–N(1–771); CBP–M(1069–1459, or 1069–1892 for testing ATF-2); CBP–C1(1892–2036); CBP–C2(2036–2231); CBP– C3(2231–2441); CBP–C(1892–2441); p300–N(1–596); p300–M(744–1571); p300–C1(1855–2010); p300–C2(2010–2210); p300–C3(2210–2414); and p300–C(1571–2370) The intensity of binding, expressed as percentage of input bound, is indicated by different shades of gray and only interactions resulting in binding to more than 1.5% of input are shown.
Trang 6However, a single copy of the sequence encompassing the
PRD IV and P31 sites, termed P431, confers significant
virus-inducibility to a reporter gene in mammalian cells [22],
suggesting that ATF-2/c-Jun and IRF-3/IRF-7 cooperate
to synergistically activate this reporter We explored the
mechanism underlying this synergy by coexpressing these
transcription factors and the p300/CBP coactivators in
S2 cells (Table 1) ATF-2/c-Jun, IRF-3 and IRF-7 each
stimulated the P431·3CAT reporter less than threefold in
the absence of mammalian p300/CBP, and less than 10-fold
in their presence However, the combination of
transcrip-tion factors and coactivators resulted in very strong
activation of this reporter (> 1000-fold), indicating that
these proteins bound cooperatively to the P431 element and
synergistically activated transcription Synergy was
compu-ted by dividing the fold induction obtained experimentally
for a given combination of proteins by the value obtained
when the fold induction for each of the proteins present in
the combination were added (Table 1) There was little
synergy in the absence of cotransfected p300/CBP, and this
synergy involved only ATF-2/c-Jun and IRF-7, suggesting
these factors physically interact on the P431 site In the
presence of mammalian p300/CBP, however, very strong
synergy was observed when all the transcription factors
were combined (> 200-fold), and removing a single factor
led to much lower levels of synergy
Proteins binding to theIFN-b promoter interact weakly
with each other
Three forms of IRF-3 were produced by in vitro translation
and tested for their ability to interact with ATF-2, c-Jun,
IRF-3 and IRF-7 immobilized on beads as GST fusion
proteins Wild type IRF-3 (IRF-3wt) interacted poorly, if at
all, with the other proteins (Fig 4) By contrast, IRF-3E7,
which partially mimics virus-activated IRF-3, interacted
weakly with all proteins tested Virus infection leads to a conformational change and dimerization of IRF-3, but IRF-3E7 is mostly a monomer [21] Therefore we also tested
a truncation of IRF-3 that dimerizes more efficiently and we found that indeed IRF-31–328bound much more strongly to the GST fusion proteins than either IRF-3wt or IRF-3E7 Similarly, we tested three forms of IRF-7, namely IRF-7wt, IRF-7Di and IRF-71–388, and found that IRF-71–388bound more efficiently to ATF-2, c-Jun and IRF-3 than either IRF-7wt or IRF-7Di
ATF-2 and c-Jun strongly interacted with each other as expected for these heterodimerization partners, while bind-ing to IRF-3 and IRF-7 was much weaker (Fig 4) Similarly, interactions with p50 or p65 were weak but detectable with all GST fusions tested The strength of the interactions among transcription factors was, with the exception of that between ATF-2 and c-Jun, much weaker than their interactions with the p300/CBP coactivators However, even weak interactions could play a determining role in the context of a given promoter if the arrangement
of cis-acting elements allows them to occur
Synergistic activation of theIFN-b gene promoter
in insect cells The ability of the IFN-b promoter to be activated in S2 cells
in response to various combinations of factors was inves-tigated (Figs 5 and 6) We first tested the effects of ATF-2/ c-Jun, IRF-1, IRF-3/IRF-7 and NF-jB (p50/p65), in the presence or absence of mammalian p300/CBP, on the
Table 1 Synergistic activation by ATF-2/c-Jun, IRF-3 and IRF-7.
Transcriptional activity of the indicated combination of the
tran-scription factors (0.5 lg each of ATF-2, c-Jun, IRF-3E7 and IRF-7Di)
in the presence or absence of cotransfected p300/CBP (1.5 lg) on the
P431·3CAT reporter Synergy was computed by dividing the fold
induction obtained experimentally for a given combination of proteins
by the value obtained when the fold induction for each of the proteins
present in the combination were added.
Vector p300/CBP Vector ATF-2/c-Jun Vector ATF-2/c-Jun Fold induction
P431·3CAT
IRF-7DI 1.8 28.8 8.2 116.8
IRF-3E7/
IRF-7DI
2.6 34.6 56.3 1131.3 Synergy vs additive
P431·3CAT
IRF-3E7/IRF-7DI 1.0 7.8 22.7 227.4
p65
GST pull-downs mATF-2
p50
c-Jun IRF-3WT IRF-3E7
IRF-7WT
IRF-1WT
GST fusions Fig 4 Physical interactions among transcription factors.35S-labeled IRF-3WT, IRF-3E7, IRF-3(1–328), IRF-7WT, IRF-7Di, IRF-7(1– 388), mATF-2(195), c-Jun, nfkb1(p50) and RelA (p65) were incubated with the indicated GST fusions of ATF-2, c-Jun, IRF-3 and IRF-7 immobilized on glutathione sepharose Proteins retained on the GST fusions and 20% of the protein input were analyzed by SDS/PAGE and autoradiography; a representative experiment is shown.
Trang 7transcription of the )110IFNbCAT reporter (Fig 5A).
Each transcription factor pair or IRF-1 could activate this
reporter on their own To investigate synergy, their amount
was titrated so that they would each minimally activate the
reporter (< 1.5-fold for all apart from IRF-1, which was
2.7-fold) Pairwise combinations of ATF-2/c-Jun, IRF-1
or NF-jB did not stimulate transcription more than the
sum of their individual effects, whether p300/CBP were
present or not By contrast, IRF-3/IRF-7 with ATF-2/c-Jun
or with NF-jB showed synergy that was entirely dependent
on the presence of mammalian coactivators ( 5.3-fold and
3.3-fold, respectively) The ATF-2/c-Jun, IRF-1 and
NF-jB combination displayed very little synergy (£ 1.4-fold),
which was not augmented in the presence of p300/CBP In
marked contrast, the ATF-2/c-Jun, IRF-3/IRF-7 and
NF-jB combination strongly synergized ( 27.6-fold) but only
in the presence of p300/CBP Thus, maximal activation was
dependent on the simultaneous presence of the mammalian
coactivators and on the set of factors activated upon virus infection
Threshold effect in synergistic activation
We next investigated the mechanism of this synergy Synergy is the functional equivalent of physical
cooperativ-A
B
0
30
60
90
120
150
All - IRFs All + IRF3/7 All + IRF1
IFNβCAT in S2 cells
0
10
20
30
40
50
60
Vector 0 7 0 1 0 5 1 8 7 0 8 1 0 7 7 2 2 6 1 5 4 1 0 9 2 1 2 1 1 7 3 2 1 2 0 5
p300/CBP 0 9 1 1 6 7 1 6 7 1 6 1 1 0 5 2 5 9 7 2 9 1 7 8 1 8 1 7 0 8 3 2 7 3 9 8
IFNβCAT in S2 cells
ATF-2/c-Jun
IRF-1
IRF-3/IRF-7
NF- κB
- + - - - + + + - - + +
+ + + +
+ + + +
- - - - + - - + + + + +
Fig 5 IRF-3/IRF-7 but not IRF-1 synergize with ATF-2/c-Jun, p50/
p65 and p300/CBP in activation of the IFN-b promoter (A) Activity of
the indicated combinations of transcription factors pairs (0.5 lg for
ATF-2, c-Jun, IRF-3E7 and IRF-7Di; 1 lg for IRF-1; 12 ng for p50
and 18 ng for p65) in the presence or absence of cotransfected p300/
CBP (1.5 lg of a 1 : 1 mix) on the )110IFNbCAT reporter (B)
Threshold effect in activation of the )110IFNbCAT reporter by
ATF-2/c-Jun, p50/p65 and p300/CBP (All – IRFs, circles), in the presence of
IRF-1 (All + IRF-1, triangles) or IRF-3E7/IRF-7Di (All + IRF3/7,
squares); 8X corresponds to the amount of transcription factors used
in (A), with or without 1 lg of IRF-1 or of an IRF3/7 mix and 0.75 lg
of CBP; 4X, 2X and 1X correspond to decrease of the amount used in
8X by factors of 2, 4 and 8, respectively.
0 10 20 30 40 50 60
A
IFNβCAT in S2 cells
p300 CBP p300/CBP
0 10 20 30 40 50 60 70
Vec tor All
- ATF-2
- c-Jun - AJ- IRF-3- IRF-7- F3
/7
- p50 - p65- NFkB - CBP- p300 - p/C
B
IFNβCAT in S2 cells
C
0 10 20 30 40 50
V e c t o r +IRF7 0.1 ug +IRF7 0.25 ug +IRF7 0.5 ug
V e c t o r All - IRF3 All + IRF3 0.25 ug All + IRF3 0.5 ug
IFNβCAT in S2 cells
Fig 6 Mechanism of synergistic activation of the IFN-b promoter in S2 cells (A) Activation of the )110IFNbCAT reporter by the transcrip-tion factors (TFs; 500 ng each of ATF-2, c-Jun, 3E7 and IRF-7Di, 100 ng p50 and 150 ng p65) in the presence or absence of 0.75 or 1.5 lg of the p300, CBP or p300/CBP coactivators The value in the absence of TFs was 0.09 (B) Activation of the )110IFNbCAT reporter by cotransfection with all the transcription factors [All; ATF-2/c-Jun (1 lg), IRF-3E7/IRF-7Di (1 lg), p50 (100 ng), p65 (150 ng) and p300/CBP (1.5 lg)] or All minus the indicated factors The value with vector alone was 0.16 (C) Activation of the )110IFNbCAT reporter by the factors ATF-2/c-Jun (1 lg), p50 (100 ng), p65 (150 ng) and p300/CBP (1.5 lg), in the presence or absence of 0.25 or 0.5 lg of 3 E7 and in the presence or absence of 0.1, 0.25 or 0.5 lg of IRF-7Di, as indicated The value with vector alone was 0.17.
Trang 8ity in the assembly of the components required for the
function Cooperative assembly of transcription factors is
expected to show a strong dependence on small changes
in their concentrations near the threshold at which the
complex can form The amount of transfected plasmids was
serially increased by a factor of two over an eightfold range,
and the experiment was performed in the presence or
absence of IRF proteins (Fig 5B) Transfection of
increas-ing amounts of ATF-2/c-Jun, NF-jB and CBP led to a
linear increase in reporter activity (lower curve, circles)
Remarkably, adding IRF-3/IRF-7 to this mix resulted in an
exponential increase in reporter activity (upper curve,
squares) At the two lowest amounts of transfected
plasmids, the addition of IRF-3/IRF-7 had little effect on
transcription Past that threshold, however, there was a
sharp increase in transcriptional activity over the last two
twofold increases in amounts of expression plasmids
transfected, resulting in an approximately 115-fold increase
in reporter activity over an eightfold increase in the amounts
of transfected plasmids By contrast, when IRF-1 instead of
IRF-3/IRF-7 was used (middle curve, triangles), reporter
activity rose 5.6-fold over an eightfold increase in
amounts of transfected plasmids, and 2.9-fold over the
last twofold increase Thus, these results indicate that
transfection of S2 cells reproduced the essential features of
the specific transcriptional activation of the IFN-b promoter
in response to distinct stimuli
Contribution of each individual factor to synergy
Either p300 or CBP was able to promote the synergistic
activation of the IFNbCAT reporter in S2 cells and
displayed dose-dependent effects (Fig 6A) CBP proved
more efficient than p300, and the combination of p300/CBP
displayed an intermediary efficiency, as was the case for
IRF-3/IRF-7 on the P31·4CAT reporter (Fig 2B) In
Fig 6B, we tested the effect of removing individual factors
Removal of either ATF-2 or p50 led to an increase in
activity from the IFNbCAT reporter, suggesting c-Jun and
p65 homodimers are stronger activators than the ATF-2/
c-Jun and p50/p65 heterodimers in this context By contrast,
removal of either IRF-3 or IRF-7 led to a decrease in
activity from the IFNbCAT reporter, and the decrease was
more significant when IRF-7 was absent
Both IRF-3 and IRF-7 are required for full activation
of theIFN-b promoter Whether the two virus-activable IRFs are both required for transcription from the IFN-b promoter is unclear We examined the dose–response to both IRF-3 and IRF-7 in the context of the IFN-b promoter (Fig 6C) Transfection
of S2 cells with ATF-2/c-Jun, NF-jB and p300/CBP led to a level of reporter activation that was only modestly stimu-lated by the addition of IRF-3 However, in the presence of even small amounts of IRF-7, addition of IRF-3 resulted in
a strong stimulation of the reporter activity IRF-7, together with ATF-2/c-Jun, NF-jB and p300/CBP, could lead to substantial activation of the reporter even in the complete absence of IRF-3 Nevertheless, maximal activation of the IFN-b promoter depended on the presence of both IRF-3 and IRF-7 (Fig 6B,C), as observed in mammalian cells (Fig 1B; [15])
We have previously shown that reporters driven by multiple copies of either PRD III or PRD I fail to respond
to virus infection Two or more copies of P31 (i.e PRD III-PRD I as a single unit), however, confer virus-inducibility, suggesting that interactions between factors bound to PRD III and PRD I are required to activate transcription
in virus-infected cells at physiological levels of IRFs As shown in Fig 7B, overexpression of IRF-3 but not IRF-7 led to virus-dependent activation of the PRDIIIx10CAT reporter By contrast, PRDI·7CAT was more strongly activated by IRF-7 than by IRF-3 in virus-infected cells Importantly, it was the combination of IRF-3 and IRF-7 that proved the most potent for both reporters Taken together, our data strongly suggest that maximal activation
of the IFN-b promoter requires the cooperative assembly of
a nucleoprotein complex containing p300/CBP, ATF-2/ c-Jun, NF-jB and both IRF-3 and IRF-7
Discussion
The current paradigm for specificity in transcriptional activation holds that the physiological concentration of transcription factors typically is such that a single factor does not activate transcription on its own The need for several factors to cooperate allows for a combinatorial principle to operate, which could account for specificity
B
A PRDIII, PRDI and IRF binding sites
0 10 20 30 40
Vector IRF-3 IRF-7 IRF-3/7 Vector IRF-3 IRF-7 IRF-3/7
PRDIIIx10- & PRDIx7-CAT in P19 cells
Fig 7 IRF-3/IRF-7 maximally activates both PRD III and PRD I (A) Sequence of PRD III, PRD I and optimal binding sites for IRF-1, IRF-3 and IRF-7 [47,48] (B) Effect of IRF-3 (1 lg), IRF-7 (2 lg) or IRF-3 and IRF-7 on the activity of PRDIII·10CAT or PRDI·7CAT in P19 cells uninfected (Co) or infected with Sendai virus (SV).
Trang 9Model for synergistic activation of theIFN-b promoter
The IFN-b promoter is approximately six times more potent
when PRD IV is converted to a higher affinity site [22]
Likewise, IRF-3 and IRF-7 bind to P31 with much less
affinity than to the ISRE present in some IFN-inducible
genes [15] and the IFN-b promoter is 28 times more
potent when P31 is converted to such an ISRE By contrast,
PRD II binds NF-jB with high affinity [31,32] Thus, in
order to activate the IFN-b promoter, a physiological
stimulus must not only activate transcription factors of the
AP-1, IRF and Rel families, but also a specific combination
that can bind the promoter cooperatively to overcome the
low intrinsic affinities of PRD IV and P31 for their cognate
factors
IRF-1-dependent activation of theIFN-b promoter
ATF-2/c-Jun, IRF-1 and NF-jB could each activate the
IFN-b promoter in insect cells However, the combination
of these factors activated the IFN-b promoter to a level that
did not exceed the sum of individual contributions, under
conditions where each factor is limiting (Fig 5) This result
is consistent with the observations that (a) stimuli that
activate this combination of transcription factors in
mam-malian cells do not activate the IFN-b gene, and that (b)
these factors bind the IFN-b promoter anticooperatively
in vitro, due to steric hindrance between IRF-1 and NF-jB
In the latter experiments, the high mobility group (HMG)-I/
Y protein is able to neutralize this anticooperativity but
binding remains noncooperative [23] We tested the effect of
expressing HMG-I in S2 cells on activation of the IFN-b
promoter by ATF-2/c-Jun, IRFs, NF-jB and coactivators
We found no statistically significant effect, one way or the
other, over a wide range of HMG-I concentrations, whether
IRF-1 or IRF-3/7 were used (H.Yang and M G Wathelet,
unpublished data) However, we note that D1, a Drosophila
ortholog of HMG-I [33], is present in S2 cells in large
amounts [E Kas (CNRS, UMR5099, Toulouse, France),
personal communication] and thus could mask any effect
of transfected HMG-I
IRF-3/IRF-7-dependent activation of theIFN-b promoter
Unlike IRF-1, IRF-3E7/IRF-7Di strongly synergized with
ATF-2/c-Jun, NF-jB and p300/CBP to activate the IFN-b
promoter (Fig 5A) Presumably, the difference between the
level of activation achieved with the IRF-3/7-containing set
of factors vs that achieved with the IRF-1-containing set
would be much greater if the virus-activated IRF-3/7
proteins could be used instead of the mutant forms, not only
because of the difference in affinity for DNA, but also for
the coactivators Nevertheless, the observation of a strong
threshold effect, even with the IRF-3E7/IRF-7Di-containing
set (Fig 5B), further suggests that the binding of the set of
virus-activated transcription factors to the IFN-b promoter
is highly cooperative
Some synergy was evident in the absence of either ATF-2/
c-Jun or NF-jB This is consistent with the observation that
it is possible to bypass the requirement for both factors
provided that the concentration of IRF-3 is above
physio-logical levels [34] Interestingly, no synergy was observed in
the absence of either p300/CBP or IRF-3/IRF-7, suggesting that VAF serves as a keystone in the assembly of a functional activator/coactivator complex at the IFN-b promoter VAF formation depends on multiple protein–protein interactions: (a) virus-dependent homodimerization of IRF-3 [35] and of IRF-7 [36]; (b) constitutive interactions between IRF-3 and IRF-7 [15,37,38]; and (c) virus-dependent modifications of these factors that result in their association with several domains of the coactivators p300 and CBP [20,21] The role of promoter context in ensuring specificity
We show that IRFs could interact with ATF-2, c-Jun, p50 and p65 (Fig 4) These interactions were rather weak but could be important in the context of the IFN-b promoter if the arrangement of the PRDs allows them to occur The importance of these interactions was tested functionally and our results indicate that the balance of positive and negative interactions between IRF-1 and ATF-2/c-Jun or NF-jB prevented cooperative binding in the context of the IFN-b promoter By contrast, such a balance favored cooperative binding when IRF-3/IRF-7 was used instead of IRF-1 (Fig 5) IRF-7 but not IRF-3 drives synergy with ATF-2/ c-Jun when binding to P431 (Table 1), which suggests that IRF)7 has unique interactions with ATF-2/c–Jun Such interactions might account, at least in part, for the observation that IRF-7 is a stronger activator when binding
to P31 in its natural context than in isolation (while the reverse was true of IRF-3, Figs 1B and 6)
The role of coactivators in promoting synergy and specificity
Synergistic activation of the IFN-b promoter was entirely dependent on the presence of mammalian coactivator (Figs 5 and 6), consistent with the inhibitory effect of E1a on induction of the IFN-b gene in response to dsRNA [39] Importantly, ATF-2/c-Jun, IRF-7Di and NF-jB had intrin-sic transcriptional activities that were only moderately stimulated by coexpression of the mammalian coactivators (Fig 2) Nevertheless, this combination of factors had little activity in the absence of coactivators but strongly synergized
in their presence (Figs 6B.C) Thus, the ability of c-Jun, IRF-7Di and RelA to interact with coactivators was more important to their ability to synergize with other transcrip-tion factors than to activate transcriptranscrip-tion by themselves Taken together, these data suggest that in the activation of the IFN-b promoter, coactivators not only serve as an adaptor between the general transcription machinery and the activators, but also act as a scaffold by stabilizing the formation of a nucleoprotein complex through simultaneous interactions with transcription factors The flexible nature of p300 and CBP may be crucial for accommodating the specific arrangement of activator proteins on the IFN-b promoter as well as on other complex gene regulatory elements [40] Such
a scaffolding role for these coactivators has been hypothes-ized [1,41], but not demonstrated Our data lend strong support to this important paradigm
If p300 or CBP bind simultaneously to two or more transcription factors, it must do so through different domains It is therefore somewhat puzzling that all the factors tested interacted most strongly with the N2 and C2
Trang 10fragments Moreover, we have shown that in the case of
IRF-3, all of its interactions with the coactivators were
indispensable for transcriptional activity [21] Similarly, we
have shown that the ability of IRF-7 to synergize with
either c-Jun or IRF-3 was dependent of its contacts with
the coactivators [20] These observations could be reconciled
considering that these domains in p300 and CBP are
relatively large and it is thus possible for more than one
factor to bind at once Furthermore, more than one
molecule of coactivator is likely recruited to the IFN-b
promoter in virus-infected cells, because p300 and CBP can
interact with each other (H Yang, C H Lin & M G
Wathelet, unpublished data) and VAF contains at least two
molecules of p300/CBP [15] Thus, we favor a model
(Fig 1A) where at least two molecules of coactivator
contribute to the cooperative assembly of a nucleoprotein
complex at the IFN-b promoter
Tissue-specificity in activation of theIFN-b promoter
It has been documented that the IFN-b promoter is not
induced by virus in embryonic stem cells or undifferentiated
embryonic carcinoma cells [18,19] The data presented in
Fig 1B suggest that in P19 cells (a pluripotential
teratocar-cinoma line) the failure to activate the IFN-b promoter was
not due to the absence of the pathway leading to activation of
3 and 7 The endogenous levels of 3 and
IRF-7 in P19 cells were apparently too low to support induction of
the transiently transfected)110IFNbCAT reporter by virus,
but ectopic expression of either factor was sufficient to confer
virus-inducibility to this reporter Additional experiments
will be required to determine if this observation holds true
for the endogenous IFN-b gene In early passage primary
embryonic fibroblasts, by contrast, IRF-3 is expressed
at normal levels while IRF-7 is expressed at low levels
Elimination of IRF-3 by gene targeting does not block IFN-b
mRNA induction but results in lower levels, indicating that
these low IRF-7 levels are biologically significant
Inactiva-tion of the IRF-9 gene in these IRF-3 null cells results in
undetectable levels of IRF-7 mRNA and a complete block in
IFN-b induction [42] However, in later passage embryonic
fibroblasts or in spleen cells of IRF-3 null mice, which express
higher levels of IRF-7, induction of the IFN-b mRNA is
similar to wild type [43] These results are congruent with our
observations in P19 and S2 cells (Figs 1B and 6) that indicate
that (a) either IRF-3 or IRF-7 is sufficient to activate the
IFN-b promoter; (b) maximal activation is achieved in the
presence of both IRF-3 and IRF-7; and (c) IRF-7 preference
for the context of the IFN-b promoter favors its recruitment
to the promoter even when expressed at low levels Because
viruses can interfere with antiviral defenses [44], including
production of IFN and targeting of IRF-3 [45] or IRF-7 [46],
the existence of some redundancy in the function of IRF-3
and IRF-7 might help minimize the influence of this later
class of virulence factors
Acknowledgements
We would like to thank T Collins, R Goodman and D Livingston for
kindly providing reagents and N Horseman for critical reading of the
manuscript This work was supported by a Dean Research Award to
M.G.W.
References
1 Maniatis, T., Falvo, J.V., Kim, T.H., Kim, T.K., Lin, C.H., Parekh, B.S & Wathelet, M.G (1998) Structure and Function of the Interferon-b Enhanceosome Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
2 Sen, G.C (2001) Viruses and interferons Annu Rev Microbiol.
55, 255–281.
3 Stark, G.R., Kerr, I.M., Williams, B.R., Silverman, R.H & Schreiber, R.D (1998) How cells respond to interferons Annu Rev Biochem 67, 227–264.
4 Levy, D.E., Marie, I., Smith, E & Prakash, A (2002) Enhance-ment and diversification of IFN induction by IRF-7-mediated positive feedback J Interferon Cytokine Res 22, 87–93.
5 Bode, J., Benham, C., Knopp, A & Mielke, C (2000) Tran-scriptional augmentation: modulation of gene expression by scaffold/matrix-attached regions (S/MAR Elements) Crit Rev Eukaryot Gene Expr 10, 73–90.
6 Tovey, M.G., Streuli, M., Gresser, I., Gugenheim, J., Blanchard, B., Guymarho, J., Vignaux, F & Gigou, M (1987) Interferon messenger RNA is produced constitutively in the organs of normal individuals Proc Natl Acad Sci USA 84, 5038–5042.
7 Nourbakhsh, M & Hauser, H (1997) The transcriptional silencer protein NRF: a repressor of NF-jB enhancers Immunobiology
198, 65–72.
8 Nourbakhsh, M & Hauser, H (1999) Constitutive silencing of IFN-b promoter is mediated by NRF (NF-jB-repressing factor),
a nuclear inhibitor of NF-jB EMBO J 18, 6415–6425.
9 Weill, L., Shestakova, E & Bonnefoy, E (2003) Transcription factor YY1 binds to the murine beta interferon promoter and regulates its transcriptional capacity with a dual activator/repres-sor role J Virol 77, 2903–2914.
10 Bode, J., Pucher, H.J & Maass, K (1986) Chromatin structure and induction-dependent conformational changes of human interferon-beta genes in a mouse host cell Eur J Biochem 158, 393–401.
11 Lomvardas, S & Thanos, D (2001) Nucleosome sliding via TBP DNA binding in vivo Cell 106, 685–696.
12 Parekh, B.S & Maniatis, T (1999) Virus infection leads to loca-lized hyperacetylation of histones H3 and H4 at the IFN-b pro-moter Mol Cell 3, 125–129.
13 Bode, J., Hochkeppel, H.K & Maass, K (1982) Links between effects of butyrate on histone hyperacetylation and regulation of interferon synthesis in Namalva and FS-4 cell lines J Interferon Res 2, 159–166.
14 Shestakova, E., Bandu, M.T., Doly, J & Bonnefoy, E (2001) Inhibition of histone deacetylation induces constitutive derepres-sion of the beta interferon promoter and confers antiviral activity.
J Virol 75, 3444–3452.
15 Wathelet, M.G., Lin, C.H., Parekh, B.S., Ronco, L.V., Howley, P.M & Maniatis, T (1998) Virus infection induces the assembly
of coordinately activated transcription factors on the IFN-b enhancer in vivo Mol Cell 1, 507–518.
16 Navarro, L & David, M (1999) p38-dependent activation of interferon regulatory factor 3 by lipopolysaccharide J Biol Chem 274, 35535–35538.
17 Fujita, T., Reis, L.F., Watanabe, N., Kimura, Y., Taniguchi,
T & Vilcek, J (1989b) Induction of the transcription factor IRF-1 and interferon-beta mRNAs by cytokines and activators
of second-messenger pathways Proc Natl Acad Sci USA 86, 9936–9940.
18 Burke, D.C., Graham, C.F & Lehman, J.M (1978) Appearance
of interferon inducibility and sensitivity during differentiation of murine teratocarcinoma cells in vitro Cell 13, 243–248.
19 Harada, H., Willison, K., Sakakibara, J., Miyamoto, M., Fujita,
T & Taniguchi, T (1990) Absence of the type I IFN system in EC