Content uploaded by Osamu Takeuchi
Author content
All content in this area was uploaded by Osamu Takeuchi
Content may be subject to copyright.
© 2006 Nature Publishing Group
© 2006 Nature Publishing Group
Differential roles of MDA5 and RIG-I helicases in
the recognition of RNA viruses
Hiroki Kato
1,3
*, Osamu Takeuchi
1,3
*, Shintaro Sato
3
, Mitsutoshi Yoneyama
4
, Masahiro Yamamoto
1
,
Kosuke Matsui
1
, Satoshi Uematsu
1
, Andreas Jung
1
, Taro Kawai
3
, Ken J. Ishii
3
, Osamu Yamaguchi
5
, Kinya Otsu
5
,
Tohru Tsujimura
6
, Chang-Sung Koh
7
, Caetano Reis e Sousa
8
, Yoshiharu Matsuura
2
, Takashi Fujita
4
& Shizuo Akira
1,3
The innate immune system senses viral infection by recognizing a
variety of viral components (including double-stranded (ds)RNA)
and triggers antiviral responses
1,2
. The cytoplasmic helicase pro-
teins RIG-I (retinoic-acid-inducible protein I, also known as
Ddx58) and MDA5 (melanoma-differentiation-associated gene 5,
also known as Ifih1 or Helicard) have been implicated in viral
dsRNA recognition
3–7
. In vitro studies suggest that both RIG-I and
MDA5 detect RNA vi ruses and polyinosine-polycytidylic acid
(poly(I:C)), a synthetic dsRNA analogue
3
.Althoughacritical
role for RIG-I in the recognition of several RNA viruses has been
clarified
8
, the functional role of MDA5 and the relationship
between these dsRNA detectors in vivo are yet to be determined.
Here we use mice deficient in MDA5 (MDA5
2/2
) to show that
MDA5 and RIG-I recognize different types of dsRNAs: MDA5
recognizes p oly(I:C), and RIG-I detects in vit ro transcribed
dsRNAs. RNA viruses are also differentially recognized by RIG-I
and MDA5. We find that RIG-I is essential for the production of
interferons in resp onse to RNA vir uses includi ng paramyxo-
viruses, influenza virus and Japanese encephalitis virus, whereas
MDA5 is critical for picornavirus detection. Furthermore, RIG-I
2/2
and MDA5
2/2
mice are highly susceptible to infection with these
respective RNA viruses compared to control mice. Together, our
data show that RIG-I and MDA5 distinguish different RNA viruses
and are critical for host antiviral responses.
Host pattern recognition receptors, such as Toll-like receptors
(TLRs) and helicase family members, have an essential role in the
recognition of molecular patterns specific for different viruses,
including DNA, single-stranded (ss)RNA, dsRNA and g lyco-
proteins
1,9,10
. dsRNA can be generated during viral infection as a
replication intermediate for RNA viruses. TLR3, which localizes in
the endosomal membrane, has been shown to recognize viral dsRNA
as well as the synthetic dsRNA analogue poly(I:C) (refs 11, 12). The
cytoplasmic proteins RIG-I and MDA5 have also been identified as
dsRNA detectors
3–5,7,13
. RIG-I and MDA5 contain two caspase-recruit-
ment domains (CARDs) and a DExD/H-box helicase domain. RIG-I
recruits a CARD-containing adaptor, IPS-1 (also known as MAVS,
VISA or Cardif)
14–17
. IPS-1 relays the signal to the kinases TBK1 and
IKK-i, which phosphorylate interferon-regulatory factor-3 (IRF-3)
and IRF-7, transcription factors essential for the expression of type-I
interferons
18–22
. In contrast, TLR3 activates TBK1 and IKK-i through
the TIR-domain-containing adaptor TRIF (also known as Ticam1)
12
.
In vitro studies have shown that both RIG-I and MDA5 can bind to
poly(I:C) and respond to poly(I:C) and RNA viruses
6
. We have
generated RIG-I
2/2
mice, and show that RIG-I is essential eliciting
the immune responses against several RNA viruses, including
Newcastle disease virus (NDV), Sendai virus (SeV) and vesicular
stomatitis virus (VSV), in various cells except for plasmacytoid
dendritic cells (pDCs)
8
. Hepatitis C virus and Japanese encephalitis
virus are also reported to be recognized by RIG-I in vitro
23,24
.
The in vivo functional relationship between RIG-I and MDA5
remains to be determined. To investigate a functional role for MDA5
in vivo, we generated MDA5
2/2
mice and investigated viral recog-
nition (Supplementary Fig. 1). In contrast to RIG-I
2/2
mice, which
are mostly embryonic lethal, MDA5
2/2
mice are born in a mendelian
ratio, grow healthily and do not show gross developmental abnorm-
alities until 24 weeks of age. Flow cytometric analysis of leukocytes
from the spleen and lymph nodes (staining for CD3, B220 and
CD11c) revealed that the composition of lymphocytes and dendritic
cells is similar in wild-type and MDA5
2/2
mice (data not shown).
TLR3, RIG-I and MDA5 have been implicated in the recognition of
poly(I:C) and the subsequent induction of antiviral responses.
However, their exact contribution to in vivo responses against
dsRNA has yet to be clarified. We therefore examined the in vivo
responses to poly(I:C) in mice lacking RIG-I, MDA5 or TRIF, or both
MDA5 and TRIF. Administration of poly(I:C) led to rapid induction
of the cytokines interferon-
a
(IFN-
a
), IFN-
b
, interleukin-6 (IL-6)
and IL-12 in sera of both wild-type and RIG-I
2/2
mice (Fig. 1a and
Supplementary Fig . 2a). In contrast, MDA5
2/2
mice failed to
produce IFN-
a
and IFN-
b
in response to poly(I:C), and production
of IL-6 and IL-12p40 was also significantly impaired (Fig. 1b).
Although Trif
2/2
mice produced normal amounts of IFN-
a
, they
also showed severely impaired production of IL-12p40 and partial
impairment in IL-6 production. MDA5
2/2
; Trif
2/2
double-knock-
out mice failed to induce IFN-
a
, IL-6 and IL-12p40 in response to
poly(I:C). These results indicate that MDA5 is essential for poly(I:C)-
induced IFN-
a
production and TLR3 signalling is critical for IL-12
production, where as both MDA5 and TLR3 regulate IL-6 pro-
duction.
LETTERS
1
Department of Host Defense,
2
Department of Molecular Virology, Research Institute for Microbial Diseases, Osaka University, and
3
ERATO, Japan Science and Technology
Agency, 3-1 Yamada-oka, Suita, Osaka 565-0871, Japan.
4
Department of Genetics and Molecular Biology, Institute for Virus Research, Kyoto University, 53 Kawahara-cho,
Shogoin, Sakyo-ku, Kyoto 606-8507, Japan.
5
Department of Cardiovascular Medicine, Osaka University Graduate School of Medicine, 2-2 Yamada-oka, Suita, Osaka 565-0871,
Japan.
6
Department of Pathology, Hyogo College of Medicine, 1-1 Mukogawa-cho, Nishinomiya, Hyogo 663-8501, Japan.
7
Department of Medical Technology, Shinshu University
School of Allied Medical Sciences, 3-1-1 Asahi, Matsumoto 390-8621, Japan.
8
Immunobiology Laboratory, Cancer Research UK London Research Institute, Lincoln’s Inn Fields
Laboratories, 44 Lincoln’s Inn Fields, London WC2A 3PX, UK.
*These authors contributed equally to this work.
Vol 441|4 May 2006|doi:10.1038/nature04734
101
© 2006 Nature Publishing Group
© 2006 Nature Publishing Group
When bone-marrow-derived dendritic cells generated by granulo-
cyte–macrophage colony-stimulating factor (GMCSF) were incu-
bated in the presence of poly(I:C), production of IFN-
a
and IFN-
b
was severely impaired in MDA5
2/2
, but not in RIG-I
2/2
or Trif
2/2
,
GMCSF-DCs (Fig. 1c and Supplementary Fig. 2b). Even when
poly(I:C) was transfected into GMCSF-DCs using lipofectamine,
poly(I:C) induced IFN-
b
production in an MDA5-dependent, but
not a RIG-I- or TRIF-dependent, manner (Fig. 1d). IFN-
b
pro-
duction in response to poly(I:C) was also impaired in MDA5
2/2
mouse embryonic fibroblasts (MEFs) (Fig. 1e) , indicating that
poly(I:C) is primarily recognized by MDA5, not RIG-I and TLR3,
in these cells.
dsRNAs transcribed in v itro (Supplementary Fig. 2c) also stimu-
lated MEFs to produce IFN-
b
. Unlike for poly(I:C), wild-type and
MDA5
2/2
MEFs produced comparable amounts of IFN-
b
(Fig. 1e)
in response to in vitro transcribed dsRNAs. In contrast, RIG-I
2/2
MEFs did not produce detectable amounts of IFN-
b
, indicating that
RIG-I is essential for the detection of in vitro transcribed dsRNAs. As
RIG-I, but not MDA5, is responsible for IFN-
b
production in
response to dsRNAs of various lengths, these helicases probably
distinguish nucleotide structure or sequence, but not length.
Together, these results indicate that MDA5 and RIG-I are involved
in the detection of poly(I:C) and in v itro transcribed dsRNAs,
respectively.
This finding led us to hypothesize that RIG-I and MDA5 are
involved in the detection of different RNA viruses. We have previously
shown that a set of negative-sense RNA viruses are recognized
by RIG-I
8
.WefirstexaminedIFN-
b
and IFN-
a
production in
MDA5
2/2
MEFs in response to a set of negative-sense ssRNA viruses,
including NDV, SeV, VSVand influenza virus. As infection with most
of the wild-ty pe v iruses (except NDV) failed to induce t y pe-I
interferons in MEFs, owing to suppression of interferon responses
by v iral proteins (data not shown), we also used mutant viruses
lacking viral interferon-inhibitory proteins. As shown in Fig. 2a and
Supplementary Fig. 4b, wild-type MEFs produce IFN-
b
and IFN-
a
in
response to these mutant viruses. Production of type-I interferons
was severely impaired in RIG-I
2/2
MEFs compared to wild-type
cells, but MDA5 was dispensable for the production of type-I
interferons. Japanese encephalitis virus ( JEV), a positive-sense
ssRNA virus belonging to the flavivirus family, also required RIG-I,
but not MDA5, for IFN-
b
production (Fig. 2b).
We then examined the interferon responses of MEFs to encephalo-
myocarditis v irus (EMCV), a positive-sense ssRNA virus belonging
to the picornavirus family. EMCV-induced IFN-
b
production was
abrogated in MDA5
2/2
MEFs (Fig. 2c). In contrast, wild-type and
RIG-I
2/2
MEFs produced comparable amounts of IFN-
b
, indicating
that EMCV is specifically recognized by MDA5. The induction of
genes encoding IFN-
b
, IP-10 and IL-6 in response to EMCV was
abrogated in MDA5
2/2
macrophages (Supplementary Fig. 3d). The
synthesis of cellular proteins in MDA5
2/2
MEFs was progressively
inhibited during EMCV infection, to an extent and with kinetics
similar to wild-type MEFs (Supplementary Fig. 5), indicating that
the EMCV infection was established in wild-type and MDA5
2/2
MEFs in a similar manner. Moreover, other viruses belonging to
the picornavirus family (Theiler’s and Mengo viruses) also induced
IFN-
a
through MDA5 (Supplementary Fig. 4d). Furthermore, the
production of IFN-
b
in response to SeV and EMCV was impaired in
RIG-I
2/2
and MDA5
2/2
GMCSF-DCs, respectively (Fig. 2d, e),
indicating that conventional dendritic cells (cDCs) also use these
helicases for the differential recognition of viruses. EMCV-induced
production of IL-6was also abrogated in MDA5
2/2
,butnotRIG-I
2/2
,
cDCs (Supplementary Fig. 4c). Therefore, MDA5 is critical for the
regulation of pro-inflammatory cytokines as well as type-I interferons
in response to EMCV.
We next examined whether viral RNAs derived from VSV and
EMCV recapitulate the production of interferons through MDA5
and RIG-I. When transfected into GMCSF-DCs by lipofection,
RNAs prepared from VSV or EMCV induced production of IFN-
a
in a RIG-I- or MDA5-dependent manner, respectively (Fig . 2f). We
also performed reconstitution experiments by transfecting RIG-I or
MDA5 expression vectors into RIG-I
2/2
; MDA5
2/2
MEFs, in which
IFN-
b
induction was completely abrogated in response to infection
with EMCVor SeVCm (SeV with a mutated C protein) (Fig. 2g). The
ectopic expression of human RIG-I, but not MDA5, activated the Ifnb
promoter in response to SeV Cm . Reciprocally, cells expressing
human MDA5, but not RIG-I, activated the Ifnb promoter in
response to EMCV in a dose-dependent manner (Fig. 2h). These
results indicate that human RIG-I and MDA5 recognize different
RNA viruses by recognizing viral RNAs.
Previous studies have shown that pDCs use mainly the TLR system
instead of RIG-I in the recognition of several RNA viruses
8
. MyD88
is an adaptor protein essential for TLR signalling (except through
TLR3). We purified B220
þ
pDCs from Flt3L-generated bone-
marrow-derived dendritic cells (Flt3L-DCs) and infected them with
EMCV. pDCs from Myd88
2/2
, but not MDA5
2/2
, mice showed a
profound defect in IFN-
a
production (Supplementary Fig. 6).
Reciprocally, MDA5, but not MyD88, is required for the production
of IFN-
a
in B220
2
cDCs purified from Flt3L-DCs (Supplementary
Fig. 6). These results indicate that both MDA5 and RIG-I are
Figure 1 | Roles of MDA5, RIG-I and TRIF in the recognition of synthesized
dsRNAs and dsRNA analogues. a, b, RIG-I
2/2
and littermate RIG-I
þ/2
mice (a) or wi ld-type (WT), MDA5
2/2
, Trif
2/2
or MDA5
2/2
; Trif
2/2
double-knockout mice (b) were injected intravenously with 200
m
g poly(I:C)
for the indicated periods, and IFN-
a
, IL-6 and IL-12p40 production was
measured in serum by ELISA. Data show mean ^ s.d. c, GMCSF-DCs from
RIG-I
2/2
, MDA5
2/2
, TRIF
2/2
and littermate control mice were incubated
in the presence of 50 or 250
m
gml
21
poly(I:C) for 24 h. IFN-
b
production in
the cell culture supernatants was measured by ELISA. Med, medium only.
d, GMCSF-DCs were treated with 1
m
gml
21
poly(I:C) complexed with or
without lipofectamine 2000 for 24 h, and IFN-
b
production was measured.
e, Wild-type, RIG-I
2/2
and MDA5
2/2
MEFs were treated with poly(I:C) or
in vitro transcribed dsRNAs of indicated lengths complexed with
lipofectamine 2000 for 12 h, and IFN-
b
production was measured. Error
bars indicate s.d. of triplicate wells in a single experiment; data are
representative of three independent experiments. ND, not detected.
LETTERS NATURE|Vol 441|4 May 2006
102
© 2006 Nature Publishing Group
© 2006 Nature Publishing Group
dispensable for the viral induction of IFN-
a
in pDCs.
We next examined the in vivo roles of MDA5 and RIG-I in host
defence against viral infection. Although most RIG-I
2/2
mice are
embryonic lethal
8
, we could efficiently obtain live adult mice by
intercrossing the RIG-I
þ/2
mice obtained after RIG-I
þ/2
£ ICR
crosses (Supplementary Table 1). When the mice were infected with
JEV, serum IFN-
a
levels were markedly decreased in RIG-I
2/2
mice
compared to littermate RIG-I
þ/2
mice. In contrast, MDA5
2/2
mice
did not show a defect in JEV-induced systemic IFN-
a
production
(Fig. 3a). IFN-
a
production was partially impaired in Myd88
2/2
mice compared to wild-type mice, but the extent of this impairment
was far less than in RIG-I
2/2
mice (Fig. 3a). These data suggest that
the TLR system is not critical for the induction of serum IFN-
a
in vivo
in response to JEV. Consistent with this finding, RIG-I
2/2
mice, but
not MDA5
2/2
or Myd88
2/2
mice, were more susceptible to JEV
infection than control mice (Fig. 3b). Furthermore, RIG-I
2/2
mice,
but not MDA5
2/2
mice, succumbed to VSV infection, consistent
with abrogated interferon responses (Supplementary Fig. 7). Thus,
RIG-I-mediated recognition of a specific set of viruses has a critical
role in antiviral host defence in vivo.
We next challenged the mice with EMCV as a model virus that is
recognized by MDA5. Induction of IFN-
b
,IFN-
a
, RANTES and IL-6
was severely impaired in the sera of MDA5
2/2
mice (Fig. 4a and
Supplementary Fig. 8). MDA5
2/2
mice and mice null for the IFN-
a
/
b
receptor (Ifnar1
2/2
) were highly susceptible to EMCV infection
(viral titre of 1 £ 10
2
plaque-forming units (p.f.u.)) compared to
littermate controls (P , 0.01) (Fig. 4b). In contrast, deficiency of
neither RIG-I nor TLR3 affected the survival of mice infected with
EMCV. Consistent with a previous report
22
, Myd88
2/2
mice were
modestly susceptible to EMCV infection compared to wild-type
mice, implying that pDC-mediated responses are not critical for
eliminating EMCV (Fig. 4b).
It is known that EMCV preferentially infects cardiomyocytes and
causes myocarditis. Consistent with increased susceptibility to
EMCV, viral titre in the heart was much higher in MDA5
2/2
mice
compared to control mice (Fig. 4c). Histological analysis of hearts
two days after EMCV infection revealed that focal necrosis of
Figure 2 | Differential viral recognition by RIG-I and MDA5. a, Wild-type,
RIG-I
2/2
and MDA5
2/2
MEFs were exposed to negative-sense ssRNA
viruses, including NDV, VSV lacking a variant of M protein (NCP), SeV with
a mutated C protein (Cm), SeV lacking V protein (V
2
), and influenza virus
lacking the NS1 protein (DNS1) for 24 h. IFN-
b
production in the culture
supernatants was measured by ELISA. b, c, Wild-type, RIG-I
2/2
and
MDA5
2/2
MEFs were exposed to the positive-sense ssRNA viruses JEV (b)
and EMCV (c), and IFN-
b
production was measured. d, e, GMCSF-DCs
from RIG-I
2/2
and MDA5
2/2
mice and their littermate wild-type mice
were infected with an increasing m.o.i. of SeV V
2
(d) or EMCV (e) for 24 h,
and IFN-
b
production was measured. f, Wild-type, RIG-I
2/2
and MDA5
2/2
GMCSF-DCs were treated with RNAs directly prepared from VSV and
EMCV (complexed with lipofectamine 2000) for 24 h, and IFN-
a
production was measured. g, Wild-type and RIG-I
2/2
; MDA5
2/2
MEFs
were transiently transfected with a reporter construct containing the Ifnb
promoter and exposed to SeV Cm or EMCV for 24 h. Cell lysates were then
prepared and subjected to a luciferase assay. h, RIG-I
2/2
;MDA5
2/2
MEFs
were transiently transfected with the Ifnb promoter construct together with
expression plasmids encoding human RIG-I or MDA5. The cells were then
infected with EMCV or SeV Cm for 24 h and were subjected to a luciferase
assay. Error bars in a–g indicate s.d. of triplicate wells in a single experiment;
data are representative of three independent experiments. ND, not detected.
Figure 3 | Susceptibility of RIG-I
2/2
and MDA5
2/2
mice to JEV infection.
a, RIG-I
þ/2
, RIG-I
2/2
, MDA5
þ/2
and MDA5
2/2
mice (n ¼ 8), and
Myd88
þ/þ
or Myd88
2/2
mice (n ¼ 6), were injected intravenously with
2 £ 10
7
p.f.u. JEV. Sera were collected 24 h after injection, and IFN-
a
production levels measured by ELISA. Circles represent individual mice,
bars indicate mean values. Asterisk, P , 0.05 versus controls (t-test). b, The
survi val of 6-week-old mice (genotypes as indicated) infected intravenously
with 2 £ 10
7
p.f.u. JEV. Mice were monitored for 15 days (P , 0.01 between
RIG-I
2/2
mice and their littermate controls, generalized Wilcoxon test).
NATURE|Vol 441|4 May 2006 LETTERS
103
© 2006 Nature Publishing Group
© 2006 Nature Publishing Group
cardiomyocytes had developed in MDA5
2/2
mice, but wild-type
hearts showed no histological abnormalities at this time point
(Fig. 4d). Notably, no infiltration of immune cells was observed in
either wild-type or MDA5
2/2
heart sections at this time point.
However, when cardiac performance was analysed by echocardio-
graphy two days after infection (Fig. 4e), cardiac contractility was
severely depressed in MDA5
2/2
mice (fractional shortening
48.2 ^ 4.9% in MDA5
þ/2
mice, 21.2 ^ 5.8% in MDA5
2/2
mice),
indicating that MDA5
2/2
mice developed severe heart failure due to
virus-induced cardiomyopathy. Thus, MDA5-mediated recognition
of EMCV is a prerequisite for triggering antiviral responses as well as
for prevention of myocardial dysfunction.
Together, our results demonstrate that RIG-I and MDA5 have
essential roles in the recognition of different groups of RNA viruses,
as well as in the subsequent production of type-I interferons and pro-
inflammatory cytokines. We have found that poly(I:C) and in v itro
transcribed dsRNA are recognized by MDA5 and RIG-I, respectively;
this is in contrast to results from previous in vitro studies. RIG-I
probably recognizes dsRNA generated over the course of RNA virus
replication, as all in vitro transcribed dsRNAs tested except for
poly(I:C) induced type-I interferons through RIG-I. In contrast,
the endogenous ligand of MDA5 remains enigmatic. Moreover,
how RIG-I and MDA5 differentially recognize natural dsRNAs is
undetermined. Given that the helicase domains of RIG-I and
MDA5 bind to dsRNA, analyses of the cr ystal structures of these
domains should to help achieve a better understanding of the
molecular mechanisms underly ing this differential recognition.
Furthermore, it is still possible that unk nown dsRNA-binding
proteins also function as direct receptors for viral RNAs.
Finally, the picornavirus family contains several viruses that are
pathogenic for humans, including poliovirus, rhinovirus and the
virus causin g foot-and-mouth-disease. Our studies suggest that
human MDA5 and RIG-I also recognize RNA viruses. Thus, identi-
fication of therapeutic agents that modify RIG-I or MDA5 may lead
to antiviral strategies against selected viruses.
METHODS
Mice, cells and reagents. The generation of MDA5
2/2
mice is described in the
Supplementary Information. Myd88
2/2
, Tlr3
2/2
and Trif
2/2
mice have been
described previously
12
. Ifnar1
2/2
mice have also been described previously
25
.
RIG-I
þ/2
mice in a 129Sv £ C57BL/6 background were crossed with ICR mice,
and the resulting RIG-I
þ/2
mice were further intercrossed. Interbreeding of
these RIG-I
þ/2
mice produced healthy and fertile RIG-I
2/2
offspring, although
their number was less than half that of RIG-I
þ/þ
progeny (Supplementary
Table 1). RIG-I
2/2
and RIG-I
þ/2
littermate mice were used for in vivo
experiments. RIG-I
2/2
; MDA5
2/2
mice in a 129Sv £ C57BL/6 background
were lethal at embryonic day 12.5. Additional details regarding cells, reagents and
the preparation of in vitro transcribed dsRNA are provided in the Supplementary
Information.
Viruses. NDV (ref. 3), VSV, VSV lacking a variant of M protein (NCP) (ref. 8),
influenza virus lacking the NS1 protein ( DNS1) (ref. 26), JEV (ref. 27) and
EMCV (ref. 3) have been described previously. SeVand SeV lacking the V protein
(V
2
) or with mutated C proteins (Cm) were provided by A. Kato
28
.
Luciferase assay. Wild-type or RIG-I
2/2
; MDA5
2/2
MEFs were transiently
transfected with a reporter construct containing the Ifnb promoter together with
an empty vector (mock), or RIG-I or MDA5 expression vectors. As an internal
control, a Renilla luciferase construct was transfected. Transfected cells were
untreated or infected with EMCV or SeV Cm (m.o.i. 20) for 24 h. The cells were
lysed and subjected to a luciferase assay using a dual-luciferase reporter assay
system (Promega) according to the manufacturer’s instructions.
Analysis of mice after EMCV infection. Methods for plaque assays, histological
analysis and echocardiography are described in the Supplementary Information.
Measurement of cytokine production. Cell culture supernatants were collected
and analysed for IFN-
b
, IFN-
a
, IL-6 or IL-12p40 production using enzyme-
linked immunosorbent assays (ELISAs). ELISA kits for mouse IFN-
a
and IFN-
b
were purchased from PBL Biomedical Laboratories, and those for IL-6, IL-12p40
and RANTES were obtained from R&D Systems.
Statistical analysis. Kaplan–Meier plots were constructed and a generalized
Wilcoxon test was used to test for differences in survival between control and
mutant mice after viral infection. Statistical significance of any differences in
cytokine concentration and ECMV titres was determined using Student’s t-tests.
Received 30 January; accepted 20 March 2006.
Published online 9 April 2006.
1. Akira, S., Uematsu, S. & Takeuchi, O. Pathogen recognition and innate
immunity. Cell 124, 783–-801 (2006).
2. Katze, M. G., He, Y. & Gale, M. Jr. Viruses and interferon: A fight for
supremacy. Nature Rev. Immunol. 2, 675–-687 (2002).
3. Yoneyama, M. et al. The RNA helicase RIG-I has an essential function in
double-stranded RNA-induced innate antiviral responses. Nature Immunol. 5,
730–-737 (2004).
4. Kang, D. C. et al. mda-5: An interferon-inducible putative RNA helicase with
double-stranded RNA-dependent ATPase activity and melanoma growth-
suppressive properties. Proc. Natl Acad. Sci. USA 99, 637–-642 (2002).
5. Andrejeva, J. et al. The V proteins of paramyxoviruses bind the IFN-inducible
RNA helicase, mda-5, and inhibit its activation of the IFN-
b
promoter. Proc. Natl
Acad. Sci. USA 101, 17264–-17269 (2004).
6. Yoneyama, M. et al. Shared and unique functions of the DExD/H-box helicases
RIG-I, MDA5, and LGP2 in antiviral innate immunity. J. Immunol. 175,
2851–-2858 (2005).
7. Rothenfusser, S. et al. The RNA helicase Lgp2 inhibits TLR-independent sensing
of viral replication by retinoic acid-inducible gene-I. J. Immunol. 175,
5260–-5268 (2005).
8. Kato, H. et al. Cell type-specific involvement of RIG-I in antiviral response.
Immunity 23, 19–-28 (2005).
9. Iwasaki, A. & Medzhitov, R. Toll-like receptor control of the adaptive immune
responses. Nature Immunol. 5, 987–-995 (2004).
10. Beutler, B. Inferences, questions and possibilities in Toll-like receptor signalling.
Nature 430, 257–-263 (2004).
11. Alexopoulou, L., Holt, A. C., Medzhitov, R. & Flavell, R. A. Recognition of
double-stranded RNA and activation of NF-
k
B by Toll-like receptor 3. Nature
413, 732–-738 (2001).
Figure 4 | Role of MDA5 in host defence against EMCV infection.
a, MDA5
þ/2
and MDA5
2/2
mice (n ¼ 5) were inoculated intravenously
with 1 £ 10
7
p.f.u. EMCV. Sera were prepared 4 h after injection and IFN-
a
production levels determined by ELISA. b, The survival of 6-week-old mice
(genotypes as indicated) infected with 1 £ 10
2
p.f.u. EMCV intraperitoneally
was monitored every 12 h for six days (P , 0.01 between MDA5
2/2
or
Ifnar1
2/2
mice and their littermate controls, generalized Wilcoxon test).
c, MDA5
þ/2
and MDA5
2/2
mice were infected intraperitoneally with
1 £ 10
2
p.f.u. EMCV. After 48 h, mice were killed and virus titres in hearts
were determined by plaque assay. d, Heart sections of MDA5
þ/2
and
MDA5
2/2
mice, two days after infection, were assessed for histological
changes using haematoxylin and eosin staining. Arrow indicates the focal
necrosis of cardiomyocytes. e, Cardiac function of mice 48 h after EMCV
infection was assessed by echocardiography (see Supplementary Fig. 8b).
The fractional shortening (FS) after infection determined by transthoracic
M-mode echocardiographic tracings is shown. Asterisk, P , 0.05 versus
MDA5
þ/2
mice (t-test).
LETTERS NATURE|Vol 441|4 May 2006
104
© 2006 Nature Publishing Group
© 2006 Nature Publishing Group
12. Yamamoto, M. et al. Role of adaptor TRIF in the MyD88-independent toll-like
receptor signaling pathway. Science 301, 640–-643 (2003).
13. Kovacsovics, M. et al. Overexpression of Helicard, a CARD-containing helicase
cleaved during apoptosis, accelerates DNA degradation. Curr. Biol. 12, 838–-843
(2002).
14. Kawai, T. et al. IPS-1, an adaptor triggering RIG-I- and Mda5-mediated type I
interferon induction. Nature Immunol. 6, 981–-988 (2005).
15. Seth, R. B., Sun, L., Ea, C. K. & Chen, Z. J. Identification and characterization of
MAVS, a mitochondrial antiviral signaling protein that activates NF-
k
B and IRF
3. Cell 122, 669–-682 (2005).
16. Xu, L. G. et al. VISA is an adapter protein required for virus-triggered IFN-
b
signaling. Mol. Cell 19, 727–-740 (2005).
17. Meylan, E. et al. Cardif is an adaptor protein in the RIG-I antiviral pathway and
is targeted by hepatitis C virus. Nature 437, 1167–-1172 (2005).
18. Fitzgerald, K. A. et al. IKK
1
and TBK1 are essential components of the IRF3
signaling pathway. Nature Immunol. 4, 491–-496 (2003).
19. Sharma, S. et al. Triggering the interferon antiviral response through an
IKK-related pathway. Science 300, 1148–-1151 (2003).
20. Hemmi, H. et al. The roles of two I
k
B kinase-related kinases in
lipopolysaccharide and double stranded RNA signaling and viral infection. J. Exp.
Med. 199, 1641–-1650 (2004).
21. Sato, M. et al. Distinct and essential roles of transcription factors IRF-3 and
IRF-7 in response to viruses for IFN-
a
/
b
gene induction. Immunity 13, 539–-548
(2000).
22. Honda, K. et al. IRF-7 is the master regulator of type-I interferon-dependent
immune responses. Nature 434, 772–-777 (2005).
23. Chang, T. H., Liao, C. L. & Lin, Y. L. Flavivirus induces interferon-beta gene
expression through a pathway involving RIG-I-dependent IRF-3 and
PI3K-dependent NF-
k
B activation. Microbes Infect. 8, 157–-171 (2006).
24. Melchjorsen, J. et al. Activation of innate defense against a paramyxovirus is
mediated by RIG-I and TLR7 and TLR8 in a cell-type-specific manner. J. Virol.
79, 12944–-12951 (2005).
25. Hoshino, K., Kaisho, T., Iwabe, T., Takeuchi, O. & Akira, S. Differential
involvement of IFN-
b
in Toll-like receptor-stimulated dendritic cell activation.
Int. Immunol. 14, 1225–-1231 (2002).
26. Diebold, S. S., Kaisho, T., Hemmi, H., Akira, S. & Reis e Sousa, C. Innate
antiviral responses by means of TLR7-mediated recognition of single-stranded
RNA. Science 303, 1529–-1531 (2004).
27. Mori, Y. et al. Nuclear localization of Japanese encephalitis virus core protein
enhances viral replication. J. Virol. 79, 3448–-3458 (2005).
28. Kato, A. et al. Characterization of the amino acid residues of sendai virus C
protein that are critically involved in its interferon antagonism and RNA
synthesis down-regulation. J. Virol. 78, 7443–-7454 (2004).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements We thank all colleagues in our laboratory, K. Takeda,
T. Shioda, E. Nakayama and K. Kiyotani for helpful discussions, A. Kato, T. Abe,
Y. Mori, B. S. Kim and A. Palmenberg for viruses and plasmids, M. Hashimoto
for secretarial assistance, and Y. Fujiwara, M. Shiokawa, N. Kitagaki and
A. Shibano for technical assistance. This work was supported by grants from the
Ministry of Education, Culture, Sports, Science and Technology in Japan, and
from the 21st Century Center of Excellence Program of Japan.
Author Information Reprints and permissions information is available at
npg.nature.com/reprintsandpermissions. The authors declare no competing
financial interests. Correspondence and requests for materials should be
addressed to S.A. (sakira@biken.osaka-u.ac.jp).
NATURE|Vol 441|4 May 2006 LETTERS
105
