Activation of the beta interferon promoter by paramyxoviruses in the absence of virus protein synthesis.

Abstract

Conflicting reports exist regarding the requirement for virus replication in interferon (IFN) induction by paramyxoviruses. Our previous work has demonstrated that pathogen-associated molecular patterns capable of activating the IFN-induction cascade are not normally generated during virus replication, but are associated instead with the presence of defective interfering (DI) viruses. We demonstrate here that DIs of paramyxoviruses, including parainfluenza virus 5, mumps virus and Sendai virus, can activate the IFN-induction cascade and the IFN-β promoter in the absence of virus protein synthesis. As virus protein synthesis is an absolute requirement for paramyxovirus genome replication, our results indicate that these DI viruses do not require replication to activate the IFN-induction cascade.

Full text

Page 1

Activation of the beta interferon promoter by
paramyxoviruses in the absence of virus protein
synthesis
M. J. Killip,1D. F. Young,1B. L. Precious,1S. Goodbourn2
and R. E. Randall1
Correspondence
M. J. Killip
mjk3@st-and.ac.uk
Received 22 August 2011
Accepted 1 November 2011
1School of Biology, Centre for Biomolecular Sciences, BMS Building, North Haugh,
University of St Andrews, St Andrews, Fife KY16 9ST, UK
2Division of Basic Medical Sciences, St George’s, University of London,
London SW17 0RE, UK
Conflicting reports exist regarding the requirement for virus replication in interferon (IFN) induction
by paramyxoviruses. Our previous work has demonstrated that pathogen-associated molecular
patterns capable of activating the IFN-induction cascade are not normally generated during virus
replication, but are associated instead with the presence of defective interfering (DI) viruses. We
demonstrate here that DIs of paramyxoviruses, including parainfluenza virus 5, mumps virus and
Sendai virus, can activate the IFN-induction cascade and the IFN-b promoter in the absence of
virus protein synthesis. As virus protein synthesis is an absolute requirement for paramyxovirus
genome replication, our results indicate that these DI viruses do not require replication to activate
the IFN-induction cascade.
INTRODUCTION
Viruses activate the interferon (IFN) response in infected
cells through the generation or exposure of molecular
structures, termed pathogen-associated molecular patterns
(PAMPs), which are absent in uninfected cells. During
negative-sense RNA virus infections, these PAMPs are
detected by the cytoplasmic RNA helicases RIG-I and
MDA-5, which activate a downstream signalling cascade
culminating in the activation of ATF2/c-Jun, NF-kB and
IRF3 transcription factors, and subsequent transcription
of the IFN-b gene (reviewed by Kumar et al., 2011;
Onomoto et al., 2010). Secreted IFN establishes an
‘antiviral state’ in both the infected cell and neighbour-
ing uninfected cells by upregulating a large number of
antiviral IFN-stimulated genes (ISGs) (reviewed by
Randall & Goodbourn, 2008). The IFN response is very
effective at limiting virus replication and spread; to
replicate efficiently, therefore, viruses encode antagonists
of the IFN response in order to limit IFN induction or the
ability of IFN to exert its antiviral effects.
Paramyxoviruses are usually poor activators of the IFN
response, but virus preparations that are rich in defective
interfering (DI) viruses are potent inducers of IFN-b
(Chen et al., 2010; Johnston, 1981; Killip et al., 2011;
Poole et al., 2002; Shingai et al., 2007; Strahle et al., 2006).
DI viruses are unable to complete a full replication cycle
due to genome deletions and consequently require co-
infecting non-defective (ND) viruses to supplement the
missing viral factors needed to replicate their genomes.
DIs can also interfere with the replication of ND viruses
through competition for viral or host factors essential for
replication or because of their replicative advantage due to
the smaller size of their genome (reviewed by Marriott &
Dimmock, 2010; Re, 1991).
We have recently developed a GFP reporter cell line that
faithfully reports activation of the IFN-induction cascade
and the IFN-b promoter in individual cells (Chen et al.,
2010). Using this cell line, we have demonstrated that
heterocellular IFN-b promoter activation occurs during
infection with parainfluenza virus 5 (PIV5), even follow-
ing infection with a recombinant PIV5 that lacks a
functional IFN antagonist (Killip et al., 2011). This
suggests strongly that PIV5 does not normally generate
or expose PAMPs capable of activating the IFN-induction
cascade during its replication cycle. Furthermore, we
demonstrated that IFN-b promoter activation in indi-
vidual infected cells infected with PIV5 or mumps virus
(MuV) correlated with the presence of DI viruses in virus
preparations, indicating that DIs are predominantly
responsible for inducing IFN during infections with these
viruses (Chen et al., 2010; Killip et al., 2011). Here we
demonstrate that neither a co-infecting ND virus nor
virus protein synthesis (and therefore genome replication)
Three supplementary figures are available with the online version of this
paper.
Journal of General Virology (2012), 93, 299–307
DOI 10.1099/vir.0.037531-0
037531G2012 SGMPrinted in Great Britain 299

Page 2

is required for the activation of the IFN-induction cascade
by paramyxovirus DIs.
RESULTS
IFN-b promoter activation by PIV5 DI viruses is not
correlated with the level of virus replication
Our previous work has demonstrated that a recombinant
PIV5 virus, PIV5-VDC, that lacks IFN-antagonist activity
due to a C-terminal deletion in its V protein, does not
activate the IFN-b promoter in the majority of infected cells
unless it has been prepared by sequential high-multiplicity
passages, a process that causes the accumulation of DI
viruses (Killip et al., 2011). Thus, a DI-rich preparation of
PIV5-VDC [termed vM2 after the von Magnus effect (von
Magnus, 1951) and the two sequential high-multiplicity
passages required to generate the stock] was very efficient at
inducing IFN compared with our original (vM0) PIV5-
VDC preparation (Killip et al., 2011). The PIV5-VDC vM2
stock was also efficient at inhibiting the replication of ND
PIV5, consistent with the interfering nature of DI viruses.
Subsequent next-generation sequencing of RNA isolated
from CsCl-purified nucleocapsids has revealed that the
number of trailer copyback DIs (Killip et al., 2011) was 20–
40 times greater than that of ND genomes in the PIV5-VDC
vM2 preparation, whilst ND genomes predominated over
DI genomes in the vM0 stock (unpublished data). Thus, the
observation that the vM2 preparation was significantly
better at inducing IFN than the vM0 preparation suggested
a lack of correlation between IFN induction and virus
replication. To examine the relationship between these two
variables more closely, A549/pr(IFN-b).GFP reporter cells
were infected with increasing dilutions of PIV5-VDC vM0
or PIV5-VDC vM2. The levels of GFP and virus NP were
subsequently determined by immunofluorescence (Fig. 1a).
At each dilution of PIV5-VDC vM0 or PIV5-VDC vM2,
cells that were strongly positive for virus NP (indicating
normal virus replication) were usually negative for GFP,
whereas those that were strongly positive for GFP were
generally only very weakly NP-positive. Additionally, as will
be discussed further below, GFP-positive cells could clearly
be observed even at high dilutions (1024) of virus, where
very few cells would have been infected with an ND (i.e.
plaque-forming) virus. Confocal photomicrographs in Fig.
1(b) show very clearly, at a higher magnification, the lack of
correlation between GFP and virus protein expression. In
cells infected with PIV5-VDC vM2, whilst the majority of
cells were positive for GFP expression, the same cells were
generally only weakly NP-positive. In contrast, following
infection with PIV5-VDC vM0, the majority of A549/
pr(IFN-b).GFP cells were strongly positive for NP, but were
negative for GFP expression. Thus, for both DI-rich (PIV5-
VDC vM2) and DI-poor (PIV5-VDC vM0) infections, very
little (if any) virus protein synthesis and thus replication
was occurring in those cells in which the IFN-b promoter
had been activated.
IFN-b promoter activation by PIV5 DIs does not
require co-infection with ND virus
GFP-positive cells could still clearly be observed following
infection with highdilutions (1024) ofPIV5-VDC vM2virus
(Fig. 1a). This dilution of virus corresponded to approxi-
mately 0.01 p.f.u. per cell, an m.o.i. at which very few cells
are infected with an ND virus particle. The vast majority of
cells were therefore unlikely to have been co-infected with
both a DI and an ND virus, suggesting that PIV5-VDC DIs
do not require a co-infecting ND virus particle to activate
the IFN-b promoter. To investigate this further, we
determined the percentage of GFP-positive cells following
infection with increasing dilutions of PIV5-VDC vM2 by
FACS and compared this with the percentage of cells that
would be expected to be infected with ND virus at each
dilutionaspredicted usingthePoissondistribution(Table1;
Supplementary Fig. S1, available in JGV Online). At all
dilutions of PIV5-VDC vM2 examined, the percentage of
GFP-positive cells was higher than the predicted percentage
of ND-infected cells. For example, at a dilution of 1023, only
9.5% of cells were predicted to be productively infected,
whereas 35.7% of cells were positive for GFP by FACS,
indicating clearly that PIV5-VDC DIs can activate the IFN-b
promoter in the absence of a co-infecting ND virus.
PIV5 DIs activate IRF3 and the IFN-b promoter in
the absence of virus protein synthesis
Virus protein synthesis is an absolute necessity for both
DI and ND paramyxovirus genome replication due to
the requirement for concurrent nucleocapsid assembly
(Horikami et al., 1992; Vidal & Kolakofsky, 1989). The
observation that the PIV5-VDC-infected cellsthatwereGFP-
positive were usually negative for virus protein expression
suggested that virus protein synthesis and replication may
not always be required to induce IFN. We therefore sought
to determine whether DI-rich PIV5-VDC virus preparations
could activate the IFN-induction cascade following treat-
ment with protein synthesis inhibitors. To this end, A549
cells were infected with PIV5-VDC vM2 in the presence or
absence of cycloheximide (CHX), and active, phosphory-
lated IRF3 (p-IRF3) was detected by immunoblot analysis
of harvested lysates. Following treatment with 50 mg CHX
ml21, a concentration that inhibits both cellular and viral
protein synthesis (Fig. 2a), p-IRF3 could be detected within
3 h of PIV5-VDC vM2 infection, demonstrating that PIV5-
VDC vM2 infection activates IRF3 and therefore the IFN-
induction cascade in the absence of either cellular or viral
protein synthesis. The ability of PIV5-VDC vM2 to activate
IRF3 in the presence of CHX was not an effect that was
limited to A549 cells, as we observed similar results in other
cell types, including untransformed human lung cells
(MRC-5; Fig.2b)and untransformed humanskin fibroblasts
(HSF; Fig. 2c). An additional striking observation from these
experiments was that p-IRF3 levels were considerably higher
in cells infected in the presence of CHX than in untreated
cells. Further experiments revealed that treatment with the
M. J. Killip and others
300Journal of General Virology 93

Page 3

proteasome inhibitor MG132 could restore p-IRF3 levels
to those seen following CHX treatment. Furthermore, a
decrease in total IRF3 levels could be observed following
infection with PIV5-VDC vM2, which could be prevented by
treatment with either CHX or the proteasome inhibitor
MG132 (Fig. 2d). These data are consistent with the
proteasome-mediated degradation of activated IRF3 during
virus infection (in order to prevent continuous activation of
the IFN-b promoter) by a mechanism that requires ongoing
protein synthesis (Lin et al., 1998; Ye & Maniatis, 2011).
We next determined whether the promoters of IFN-b and
other IRF3-responsive genes could be activated in the
absence of virus protein synthesis. A549/pr(IFN-b).GFP
cells were infected with PIV5-VDC vM2 in the presence of
(a)
10
_2
10
_3
10
_4
PIV5-VΔC vM0
PIV5-VΔC vM0
PIV5-VΔC vM2
PIV5-VΔC vM2
GFP Anti-NP/DAPI GFPAnti-NP/DAPI
(b)
GFP Anti-NPMerge
Fig. 1. PIV5-VDC DIs activate the IFN-b promoter in the absence of a co-infecting ND virus. (a) A549/pr(IFN-b).GFP cells were
infected with 10”2, 10”3or 10”4dilutions of PIV5-VDC vM0 or vM2. Eighteen hours later, monolayers were fixed and
immunostained for virus NP expression. GFP and NP (red) were visualized by fluorescence microscopy. Nuclear material was
stained with DAPI (blue). (b) A549/pr(IFN-b).GFP cells were infected with 0.1 p.f.u. PIV5-VDC vM2 or vM0 per cell. Cells were
fixed at 18 h post-infection and immunostained for virus NP expression. GFP and NP (red) were visualized by confocal
microscopy. Selected cells within the panels are highlighted by white arrows; cells that are strongly positive for GFP but very
weakly NP-positive are labelled 1; 2 denotes cells strongly positive for virus NP, but negative for GFP.
IFN induction in the absence of paramyxovirus replication
http://vir.sgmjournals.org 301

Page 4

CHX and, at various times post-infection (p.i.), the CHX
block was reversed in the presence of actinomycin D (to
prevent any further transcription of cellular genes). The cells
were cultured until 12 h p.i. and subsequently examined for
the presence of GFP-positive cells; under these conditions,
any GFP synthesized must have been made from mRNAs
that accumulated in the presence of CHX, and therefore in
the absence of virus protein synthesis. Fig. 2(e) shows that
GFP-positive cells could readily be detected under these
conditions, confirming that the IFN-b promoter must have
been activated prior to the onset of virus protein synthesis.
Furthermore, our results were not restricted to our GFP
reportergene,aswecouldalsodetectexpressionoftheIRF3-
upregulated ISG56 protein under the CHX/actinomycin D
reversal conditions (Fig. 2f). These results demonstrate that
PIV5-VDC DIs can activate the IFN-induction cascade and
the IFN-b promoter in the absence of virus protein
synthesis, and therefore virus genome replication.
Virus protein synthesis is not required for IRF3
activation by DI-rich stocks of other
paramyxoviruses
DI-rich stocks of other paramyxoviruses, including MuV
and Sendai virus (SeV), have previously been shown to be
good inducers of IFN (Chen et al., 2010; Johnston, 1981;
Strahle et al., 2006). To determine whether virus protein
synthesis was also dispensable for IFN induction by these
viruses, we repeated our CHX experiments with DI-rich
stocks of MuV (termed MuV bulk) and SeV (vM5). As we
observed above for PIV5-VDC, both MuV bulk and SeV
vM5activated IRF3strongly inthepresenceof CHX(Fig.3a,
b), indicating that DIs of other paramyxoviruses can activate
the IFN-induction cascade without virus protein synthesis
and genome replication.
Virus protein synthesis is required for optimal
activation of IRF3 by virus preparations that have
not been enriched for DIs
We have shown above that DI-rich paramyxovirus
preparations strongly activate the IFN-induction cascade
and the IFN-b promoter in the presence of protein synthesis
inhibitors. We investigated next whether preparations of
paramyxoviruses that were generated by low-multiplicity
passage, and therefore were not deliberately enriched for DI
viruses (i.e. our working stocks of virus), could activate the
IFN-induction cascade in the absence of virus protein
synthesis. IRF3 activation could be detected in untreated
cells infected with our vM0 preparations of PIV5-VDC and
SeV, but, in contrast to infections with DI-rich preparations
of these viruses, treatment with CHX had an inhibitory
effect on IRF3 activation in infected cells (Fig. 4a, b). For
these virus preparations, therefore, inhibition of virus
protein synthesis limited activation of the IFN-induction
cascade. This inhibition was not complete, however, and
small amounts of p-IRF3 were detectable even in the
presence of CHX. This low-level activation of the IFN-
induction cascade by these virus stocks in the absence of
protein synthesis suggests that DIs are present even in virus
preparations that have been generated so as to minimize DI
generation. Our working stock of MuV (termed MuV cl.3)
was generated by plaque purification of our DI-rich MuV
preparation (MuV bulk) [for further characterization, see
Chen et al. (2010)] and could therefore be considered ‘DI-
poor’; consistent with this, IRF3 phosphorylation was not
detectable in MuV cl. 3-infected cells in the presence of CHX
(Fig. 4c). Replication was required for IRF3 activation by
this virus preparation, as p-IRF3 was detectable in untreated
but not CHX-treated cells.
We next examined transcription of the IFN-b gene at
various times p.i. with SeV in the presence or absence of
CHX. At 2 h p.i., levels of IFN-b transcript were similar
in untreated or CHX-treated cells (Supplementary Fig. S2,
available in JGV Online). At later times, however, tran-
scription of IFN-b increased steadily in untreated cells
between 4 and 8 h p.i., whereas levels of IFN-b transcript
remained stable in CHX-treated cells. As a result, IFN-b
transcription was considerably higher in untreated than in
CHX-treated cells by 8 h p.i. These results are consistent
with DIs already being present in this SeV preparation,
inducing IFN in the absence of virus protein synthesis and
replication. However, if virus replication is allowed to
Table 1. Activation of the IFN-b promoter by different multiplicities of PIV5-VDC vM2
Duplicate dilutions of PIV5-VDC vM2 were titrated on Vero cells or used to infect A549/pr(IFN-b).GFP cells in order to determine GFP expression
by FACS 16 h later. The percentage of cells expected to be infected with an ND virus at each dilution was calculated using the Poisson distribution.
Dilution of PIV5-VDC
vM2
m.o.i. (p.f.u. per cell)
P(0)* (%)
P(¢1)D (%) GFP-positive cells (%)
1022
1023
1024
1
0.1
0.01
36.8
90.5
99.0
63.2
9.5
1.0
75.1
35.7
7.9
*The probability of a cell remaining uninfected by a plaque-forming particle, P(0), was defined as P(0)5e2m(where m5m.o.i.) and expressed as a
percentage. Further explanation of the derivation of this formula is given in Methods.
DThe probability of a cell being infected by ¢1 p.f.u. was defined as P(¢1)512P(0) and expressed as a percentage.
M. J. Killip and others
302Journal of General Virology 93

Page 5

proceed, activation of the IFN-induction cascade increases
as more PAMPs are generated, presumably due to the
amplification of existing DIs and/or the de novo generation
of DIs during genome replication.
Virus binding is insufficient to activate the IFN-b
promoter
The data presented above demonstrate clearly that the IFN-b
promoter can be activated by paramyxovirus DIs in the
absence of virus protein synthesis. It is also clear that virus
binding is a prerequisite for IFN induction, as IFN induction
by PIV5-VDC vM2 is prevented completely by treatment with
neutralizing antibody to virushaemagglutinin–neuraminidase
(Supplementary
Furthermore, this result also rules out that the IFN induction
seen with this virus is caused by free RNA present in the virus
inoculum. It has previously been reported that binding and
entry of some enveloped viruses, e.g. herpesviruses, can
activate an IRF-dependent antiviral response that is triggered
by increasing amounts of input virus (Mossman et al., 2001;
Preston et al., 2001). This was unlikely to be the case for PIV5,
as we observed activation of the IFN-induction cascade even
at very low multiplicity (Fig. 1a; Table 1). Nevertheless, we
tested whether virus binding alone was sufficient to trigger the
IFN-inductioncascadeusing a UV-inactivation approach;UV
exposure impairs virus replication by inducing uracil dimers
in regions of the RNA genome that prevent the progress of
Fig.S3, availableinJGVOnline).
UI
12
_
1212
_
123
_
6
_
9
_
369 1212 h p.i.
+
_
+
_
_
__
_
_
+
_
0_1
1_12
0_2
2_12
0_4
4_12
0_6
6_12
0_8
8_12
0_12
_
0_12
CHX
ActD
CHX
MG132
CHX 0_12 h
ActD 0_12 h
CHX 6 h/
ActD 6 h
p-IRF3
IRF3
CHX
p-IRF3
IRF3
P
CHX
p-IRF3
IRF3
P
CHX
p-IRF3
Actin
Actin
Actin
Actin
Actin
NP
ISG56
_
_
++++
+++
_
+
_
+
_
+
+
_
+++
+
UI UI
PIV5 wt
(a)
(b)(c)
(d)
(e)
(f)
PIV5-V C
vM2
UI
UT
PIV5-V C
vM2
PIV5-V C
vM2
PIV5-V C vM2
Fig. 2. PIV5-VDC DIs can activate IRF3 and
the IFN-b promoter in the absence of virus
protein synthesis. (a) A549 cells were mock-
infected or infected with 10 p.f.u. PIV5 wt vM0
or PIV5-VDC vM2 per cell in the presence or
absence of cycloheximide (CHX). At various
times p.i., the cells were harvested and the
presence of phosphorylated IRF3 (p-IRF3),
viral NP and actin was detected by immunoblot
analysis. Uninfected (UI) cells were included
as a negative control. Note: for PIV5-VDC vM2
infections, the amount of NP corresponds to
input virions only, as PIV5-VDC DIs inhibit ND
virus replication significantly in infected cells.
As a result, although CHX markedly reduces
PIV5 wt NP synthesis, it does not significantly
affect NP expression in PIV5-VDC vM2-
infected cells. (b, c) MRC-5 (b) or HSF (c)
cells were mock-infected or infected with
10 p.f.u. PIV5-VDC vM2 per cell in the
presence or absence of CHX. Sixteen hours
p.i., cells were harvested and lysates were
immunoblotted for p-IRF3, total IRF3, viral P
protein and actin. (d) A549 cells were infected
with 10 p.f.u. PIV5-VDC vM2 per cell in the
presence of CHX and/or the proteasome
inhibitor MG132. Sixteen hours p.i., cells were
harvested and lysates were immunoblotted for
p-IRF3, total IRF3 and actin. (e) A549/pr(IFN-
b).GFP cells were infected with 10 p.f.u. PIV5-
VDC vM2 per cell and cultured in the presence
of CHX for the times indicated, after which
the CHX block was removed and any further
transcription was prevented by culturing the
cells in medium containing actinomycin D
(ActD). At 12 h p.i., GFP-positive cells were
visualized by fluorescence microscopy. (f)
A549 cells were infected with 10 p.f.u. PIV5-
VDC vM2 per cell in the presence of CHX or
ActD for 12 h, or 6 h CHX followed by 6 h
ActD treatment. Monolayers were harvested
and ISG56 and actin were detected by
immunoblot analysis.
IFN induction in the absence of paramyxovirus replication
http://vir.sgmjournals.org 303