Immunity 24, 633–642, May 2006 ª2006 Elsevier Inc.DOI 10.1016/j.immuni.2006.04.004
The Specific and Essential Role of MAVS
in Antiviral Innate Immune Responses
Qinmiao Sun,2,4Lijun Sun,1,2,4Hong-Hsing Liu,2
Xiang Chen,1,2Rashu B. Seth,2James Forman,3
and Zhijian J. Chen1,2,*
1Howard Hughes Medical Institute
2Department of Molecular Biology
3Center for Immunology
University of Texas Southwestern Medical Center
Dallas, Texas 75390
The mitochondrial antiviral signaling protein (MAVS)
mediates the activation of NFkB and IRFs and the in-
duction of interferons in response to viral infection.
In vitro studies have also suggested that MAVS is re-
quired for interferon induction by cytosolic DNA, but
the in vivo evidence is lacking. By generating MAVS-
deficient mice, here we show that loss of MAVS abol-
ished viral induction of interferons and prevented the
activation of NFkB and IRF3 in multiple cell types, ex-
cept plasmacytoid dendritic cells (pDCs). However,
MAVS was not required for interferon induction by cy-
tosolic DNA or by Listeria monocytogenes. Mice lack-
ing MAVS were viable and fertile, but they failed to in-
duce interferons in response to poly(I:C) stimulation
and were severely compromised in immune defense
against viral infection. These results provide the
in vivo evidence that the cytosolic viral signaling path-
way through MAVS is specifically required for innate
immune responses against viral infection.
Innateimmunity isthe firstlineofdefense against micro-
bial pathogens, including viruses. Viral infection triggers
the induction of type-I interferons (e.g., IFN-a and IFN-b)
and other proinflammatory cytokines through two dis-
tinct signaling pathways (Honda et al., 2005; Kawai
and Akira, 2006; McWhirter et al., 2005; Seth et al.,
2006). One of these pathways utilizes a subfamily of
Toll-like receptors (TLR3, 7, 8, and 9) to detect viral nu-
particles. These TLRs are localized in the endosomal
membranes of specialized cell types, such as pDCs (re-
viewed by Liu ), and they recruit the adaptor pro-
IkB kinase complex (consisting of IKKa, IKKb, and
NEMO/IKKg) and the IKK-related kinases (TBK1 and
IKK3). The IKK complex phosphorylates the NFkB inhib-
itor IkB and targets IkB for degradation by the ubiquitin-
proteasome pathway, thereby allowing NFkB to enter
the nucleus to induce a large array of genes involved
in immune and inflammatory responses (Silverman and
Maniatis, 2001). TBK1 and IKK3 phosphorylate another
transcription factor, IRF3 or IRF7, resulting in its dimer-
ization and nuclear translocation (Fitzgerald et al., 2003;
Sharma et al., 2003). The nuclear IRFs, NFkB, and other
activate the expression of interferons (Maniatis et al.,
1998), which are then secreted to bind to their receptors
on viral-infected as well as neighboring noninfected
cells. The engagement of the interferon receptors acti-
vates the JAK-STAT signaling pathway to induce inter-
lication and assembly (Darnell et al., 1994).
The other viral signaling pathway utilizes the retinoic
acid inducible gene I (RIG-I) to detect viral double-
stranded RNA (dsRNA) in the cytosol (Yoneyama et al.,
2004). RIG-I binds to viral dsRNA through its C-terminal
RNA helicase domain andmediates the activation ofIKK
and TBK1/IKK3 through its N-terminal caspase activa-
tion and recruitment domains (CARD). For unknown rea-
sons, RIG-I knockout mice are embryonic lethal, sug-
gesting that RIG-I has unexpected functions related to
animal development (Kato et al., 2005). Although the in-
ability to obtain viable RIG-I knockout mice precludes
the study of the role of RIG-I in vivo, studies using cells
derived from these mice show that RIG-I is essential for
innate immune responses to several RNA viruses in dif-
ferent cell types. However, in pDCs, loss of RIG-I had no
effect on viral induction of interferons, whereas TLR7
and MyD88 are required for the immune response in
these cells (Diebold et al., 2004; Heil et al., 2004; Kato
et al., 2005; Lund et al., 2004).
The adaptor protein that links RIG-I to IKK and TBK1/
IKK3 activation is the recently identified MAVS (Seth
et al., 2005), also known as IPS-1 (Kawai et al., 2005),
VISA (Xu et al., 2005), or CARDIF (Meylan et al., 2005).
MAVS contains an N-terminal CARD domain that inter-
acts with the tandem CARD domains of RIG-I and a C-
terminal transmembrane domain that localizes it to the
mitochondrial outer membrane (Seth et al., 2005). The
mitochondrial localization of MAVS is essential for its
signaling function, and this property is exploited by hep-
atitis C virus, which deploys the NS3/4A serine protease
to cleave MAVS off the mitochondria to evade the host
innate immune responses (Freundt and Lenardo, 2005;
Li et al., 2005; Meylan et al., 2005). Recent studies
have shown that cytosolic B form DNA and the bacte-
rium Listeria monocytogenes can also induce inter-
ferons (Ishii et al., 2006; Okabe et al., 2005; Perry et al.,
2005; Stetson and Medzhitov, 2006). Cell culture studies
have suggested that MAVS/IPS-1 is required for inter-
feron induction by cytosolic DNA (Ishii et al., 2006); how-
ever, the in vivo evidence is lacking.
In this report, we used MAVS-deficient cells to dem-
onstrate that MAVS is essential for viral induction of in-
terferons and activation of NFkB and IRF3 in multiple
ventional DCs. However, in pDCs, MAVS is not required
for viral induction of interferons and cytokines. Contrary
to previous reports, we found that loss of MAVS did not
affect interferon induction by cytosolic DNA or Listeria
monocytogenes. Furthermore, we found that MAVS-de-
ficient mice failed to induce interferons in response to
4These authors contributed equally to this work.
poly(I:C) stimulation. Interestingly, the MAVS-deficient
mice produced normal amounts of interferons in the
sera when they were infected with vesicular stomatitis
virus (VSV), but they were nevertheless more suscepti-
ble to viral-induced killing. Taken together, these results
demonstrate the specific and essential role of MAVS in
antiviral innate immunity.
Generation of Mavs2/2Mice
To elucidate the role of MAVS in vivo, we generated
Mavs-deficient (Mavs2/2) mice by homologous recom-
bination in ES cells (Figure 1A). The deletion of Mavs
was verified by Southern and Western blotting (Figures
ratio (Figure 1D), and they developed and bred normally.
These mice displayed no apparent abnormality at the
ages of up to 8 months. We have previously shown
that MAVS is localized in the mitochondrial outer mem-
brane and that it contains a C-terminal transmembrane
proteins such as Bcl-2 and Bcl-xL (Seth et al., 2005). To
examine the potential role of MAVS inapoptosis, we iso-
lated mouse embryonic fibroblasts (MEFs) from the
wild-type and mutant mice and irradiated these cells
with UV (Figure S1 available in the Supplemental Data
with this article online). Immunoblotting experiments
showed that there was no apparent difference in the
UV-induced cleavage of poly (ADP-ribose) polymerase
(PARP) or caspase-3 between the wild-type and
Mavs2/2cells. Thus, MAVS is not essential for mouse
development or survival.
Mavs2/2Embryonic Fibroblasts Are Defective
in Antiviral Innate Immune Responses
To investigate the role of MAVS in antiviral immunity, we
infected MEF cells from the wild-type and mutant mice
with Sendai virus (SeV), an RNA virus of the paramyxo-
viridae family, and then measured interferon production
by ELISA. MEF cells from Mavs2/2mice were com-
pletely defective in the production of IFN-a and IFN-baf-
ter viralinfection (Figure2Aand2B).Theinductionofthe
proinflammatory cytokine IL-6 by Sendai virus was also
abolished in Mavs2/2cells, but this response was unaf-
fected when the cells were stimulated with lipopolysac-
charides (LPS) or double-stranded RNA poly(I:C), which
activates TLR4 or TLR3, respectively (Figure 2C). We
also examined the activation of IRF3 and NFkB by using
gel shift assays, which measure the dimerization of IRF3
and DNA binding of NFkB on native gels, respectively.
Viral infection led to the dimerization and nuclear trans-
location of IRF3 in the wild-type and heterozygous cells,
but not in Mavs2/2cells (Figure 2D and data not shown).
Similarly, the loss of Mavs abolished NFkB activation by
SeV (Figure 2E), but not by LPS (Figure 2F). Because
most viruses produce double-stranded RNA that is de-
tected by the host innate immune system, we examined
the role of MAVS in the cytosolic dsRNA signaling path-
way. As shown in Figure 2G, transfection of poly(I:C) in
MEF cells led to the dimerization of IRF3 in the wild-
type cells, but not in Mavs2/2cells. Furthermore, the in-
duction of IFN-a, IFN-b, and IL-6 by poly(I:C) was abol-
transfection, which introduced the RNA into the cytosol,
addition of poly(I:C) to the media, which is known to
stimulate TLR3, did not induce IFN-a or IFN-b in MEF
Figure 1. Generation of Mavs2/2Mice
the ATG start codon to exon 3 of the Mavs lo-
cus by homologous recombination. Abbrevi-
ation: DTA, diphtheria toxin A.
(B) Southern blotting analysis of BamH1-di-
gested genomic DNA from the mouse tails
using the 50probe as indicated in (A).
(C) Immunoblot analysis of protein extracts
from MEF cells of different genotypes using
an antibody against mouse MAVS. Abbrevia-
tion: N.S., nonspecific.
(D) Offspring from the breeding of Mavs+/2
cells but induced IL-6 through a MAVS-independent
manner (Figure S2 and Figures 2A–2C). Taken together,
by Sendai virus and cytosolic dsRNA in general but is
not required for signaling by TLR3 and TLR4.
To examine the role of MAVS in viral replication and
survival of host cells, we infected MEF cells with vesicu-
lar stomatitis virus (VSV), an RNA virus of the rhabdovir-
idae family. The VSV contains a GFP fused to the cyto-
plasmic domain of the envelope glycoprotein (G) of the
virus, allowing direct visualization of viral replication
(Dalton and Rose, 2001). As shown in Figure 2H,
Mavs2/2cells were much more permissible to viral rep-
lication (GFP fluorescence) and susceptible to viral kill-
ing (Normarski microscopy) as compared to wild-type
cells. To quantify viral infection and killing, we used fluo-
rescent activated cell sorting (FACS) to measure the
numbers of GFP-positive cells as well as apoptotic cells
that can be stained by Annexin V (Figure S3). After infec-
tion with VSV-GFP, the percentages of both GFP- and
Annexin V-positive cells were significantly increased in
Mavs2/2cells as compared to the wild-type cells.
Figure 2. MAVS Is Required for Antiviral Innate Immune Responses in MEF Cells
(A–C) MEF cells were incubated with Sendai virus (SeV), LPS, or poly(I:C) for 16 hr, and the culture supernatants were harvested for ELISA anal-
yses to measure the production of IFN-a, IFN-b, and IL-6 as indicated. Error bars represent standard deviations from the means of duplicated
(D) MEF cells were infected with SeV for the indicated times and then cell lysates were separated on 9% polyacrylamide gels under nondenatur-
ing conditions. The IRF3 dimer and monomer were detected by immunoblotting. Abbreviation: hpi, hours postinfection.
(E and F) Electrophoretic mobility shift assays (EMSA) for NFkB DNA binding using whole-cell extracts from MEF cells infected with SeV or stim-
ulated with LPS for the indicated times. The asterisks (*) indicate nonspecific beads.
(G) MEF cells were transfected with poly(I:C) for the indicated times and then cell extracts were prepared for analyses of IRF3 dimerization by
native gel electrophoresis.
(H) MEF cells were incubated with VSV-GFP at the indicated multiplicity of infection (MOI) for 24 hr, and infection of cells was visualized by fluo-
rescent microscopy. Nomarski microscopy showed that less Mavs2/2cells were detected after VSV-GFP infection, an indication that these cells
were more susceptible to viral killing.
The Role of MAVS in Antiviral Innate Immunity
Thus, MAVS is essential for immune defense against
viral infection and killing.
MAVS Is Required for Antiviral Innate Immune
Responses in Macrophages
Next, we examined whether MAVS is required for inter-
feron induction in macrophages. We isolated both
bone marrow-derived macrophages (BMDMs) and peri-
toneal macrophages and infected these cells with SeV.
The viral induction of IFN-a, IFN-b, and IL-6 was com-
pletely abolished in Mavs2/2macrophages (Figures
3A–3F). Furthermore, Mavs2/2macrophages failed to
activate IRF3 or NFkB in response to Sendai virus (Fig-
ures 3G and 3H). In contrast, the induction of IFN-
b and IL-6 by LPS or poly(I:C) was normal in Mavs2/2
macrophages (Figures 3C–3F). Therefore, MAVS is
specifically required for antiviral responses in macro-
MAVS Is Not Required for Interferon Induction
by Cytosolic DNA or Listeria monocytogenes
Recently, it was reported that cytosolic B form DNA
could elicit the induction of interferons (Ishii et al.,
2006; Okabe et al., 2005; Stetson and Medzhitov, 2006)
through a mechanism dependent on MAVS/IPS-1 (Ishii
et al., 2006). To investigate whether MAVS is required
for IFN induction by cytosolic DNA, we transfected
(dI:dC). Both wild-type and Mavs2/2cells had a robust
induction of IFN-b after DNA transfection (Figure 4A).
Similarly, the induction of IFN-a and IL-6 by cytosolic
DNA was intact in Mavs2/2cells (Figure S2). No cytokine
induction was detected when the dsDNA was added to
culture media directly without transfection, indicating
that there were no contaminating TLR ligands in the
DNA preparations. Native gel analyses showed that cy-
tosolic DNA-induced activation of NFkB and dimeriza-
tion of IRF3 was not affected by Mavs deficiency (Fig-
ures 4B and 4C). Thus, MAVS is not required for the
induction of interferons or the activation of NFkB and
IRF3 by cytosolic DNA.
Listeria monocytogenes is an intracellular bacterium
that depends on TBK1 and IRF3 (O’Connell et al., 2005;
Stockinger et al., 2004). Recent studies have shown that
Listeria induces interferons by releasing bacterial DNA
into the cytosol (Stetson and Medzhitov, 2006), but the
signaling pathway that links Listeria infection to IRF3
activation is not understood. To determine if MAVS is
involved in the induction of interferons by Listeria, we in-
fected BMDMs with Listeria and found that comparable
amounts of IFN-b and IL-6 were produced in wild-type
Figure 3. MAVS Is Essential for Antiviral In-
nate Immune Responses in Macrophages
(A–F) Bone marrow-derived macrophages
(BMDMs) or peritoneal macrophages (Mfs)
were incubated with SeV, LPS, or poly(I:C)
for 16 hr and then culture supernatants were
collected for measurement of IFN-a, IFN-b,
or IL-6 by ELISA as indicated. The asterisks
(*) indicate levels that were not detectable.
Error bars represent standard deviations
from the means of duplicated experiments.
(G) Cell extracts from BMDMs infected with
SeV for the indicated times were resolved
by native gel electrophoresis and then ana-
lyzed by immunoblotting with an IRF3-spe-
(H) Cell extracts as described in (G) were in-
cubated with g-32P-ATP-labeled NFkB oligos
and then resolved by native gel electrophore-
sis (EMSA). Abbreviation: N.S., nonspecific.
and Mavs2/2macrophages (Figures 4D and 4E). The
loss of MAVS also did not affect IL-6 induction by Liste-
4G) nor did it affect the dimerization of IRF3 (Figure 4H).
Taken together, these results indicate that MAVS is dis-
pensable for interferon induction and IRF3 activation by
Listeria, further reinforcing the conclusion that MAVS is
not required for interferon induction by cytosolic DNA.
MAVS Is Required for Interferon Induction in cDCs,
but Not pDCs
DCs play a pivotal role in bridging innate and adaptive
responses, and these cells can be classified into con-
ventional (cDCs) and plasmacytoid dendritic cells
(pDCs), the latter being high producers of IFN-a/b (Liu,
2001). We isolated cDCs and pDCs from the bone mar-
row cultured with GM-CSF and Flt-3 ligand, respec-
tively, and purified them by FACS sorting (the purities
of cDCs and pDCs were 90%–95%; Figure S4). These
cells were stimulated with Sendai virus to measure cyto-
kine production by ELISA. Although cDCs derived from
the wild-type and heterozygous mice were fully capable
of producing IFN-a, IFN-b, and IL-6, cDCs from Mavs2/2
mice were severely defective in producing these cyto-
kines (Figures 5A–5C). When cDCs from Mavs2/2mice
were stimulated with LPS, poly(I:C), or CpG DNA (a
TLR9 ligand), normal production of IL-6 was detected
(Figure 5C). In sharp contrast to cDCs and other cell
types, pDCs from Mavs2/2mice produced comparable
levels of IFN-a and IFN-b to those in wild-type mice in
Figure 4. MAVS Is Not Required for Interferon Induction by Cytosolic DNA or Listeria Monocytogenes
(A) MEF cells were transfected with poly(dA:dT) or poly(dI:dC) for 16 hr and then culture supernatants were harvested for measurement of IFN-b
(B) MEF cells were transfected with 10 mg/ml of poly(dA:dT) DNA, and the cell lysates were resolved by native gel electrophoresis followed by
immunoblotting with an IRF3 antibody (top). The same cell lysates were also analyzed for NFkB DNA binding by EMSA (bottom). Abbreviation:
(C) Similar to (B), except that poly(dI:dC) was used to stimulate cells.
(D and E) BMDMs were infected with Listeria monocytogenes (L.M.) for 16 hr, and the culture supernatants were harvested for measurement of
IFN-b and IL-6 by ELISA.
(F and G) MEF cells or peritoneal macrophages were infected with L.M. for 16 hr, and the induction of IL-6 was measured by ELISA.
(H) Cell lysates from MEF cells infected with L.M. or SeV were resolved by native gel electrophoresis and then immunoblotted with an antibody
The Role of MAVS in Antiviral Innate Immunity
response to SeV infection (Figures 5D and 5E). Thus, as
shown for RIG-I (Kato et al., 2005), the role of MAVS in
interferon induction is cell type dependent (see Discus-
MAVS Is Essential for Antiviral Immune Defense
To investigate the role of MAVS in antiviral responses
in vivo, we injected wild-type and mutant Mavs mice
with VSV-GFP through the tail vein and then collected
sera to measure viral titers and interferon production.
At 12–48 hr after viral infection, the viral titers in the
Mavs2/2mice were significantly higher than those in
the wild-type and heterozygous mice(Figure 6A).Never-
theless, the virus was largely cleared in both wild-type
and mutant mice at 72 hr postinfection, suggesting
that the immune system was still effective in clearing
the virus in the absence of MAVS. Consistent with this
notion, the sera of the Mavs2/2mice contained similar
amounts of IFN-a and IFN-b to those of wild-type mice
(Figure 6B), indicating that some cells in mice, likely
pDCs, could still produce sufficient amounts of inter-
ferons when the RIG-I-MAVS pathway was crippled.
mice after viral infection, we infected mice with the wild-
shown in Figure 6C, whereas the majority of wild-type
mice (4/6) survived VSV infection, all of the Mavs+/2
tion. The high mortality rate of Mavs+/2mice was sur-
were capable of inducing interferons. To determine if
there is a quantitative difference in the viability of mice
carrying different copies of Mavs, we infected these
mice with VSV-GFP, which is less virulent (Figure S5).
When 2 3 108pfu of the virus was used to infect each
Figure 5. MAVSIsEssential forInterferonInductionincDCs,but Not
(A–C) cDCs were isolated from bone marrow cells after stimulation
with GM-CSF. These cells were incubated with SeV (A–C), LPS,
poly(I:C), or CpG DNA (C), and the production of IFN-a, IFN-b, and
IL-6 was measured by ELISA.
(D and E) pDCs were isolated from Flt-3L-stimulated bone marrow
cells and purified by FACS. These cells were stimulated with SeV
for 16 hr and then culture supernatants were harvested for measure-
ment of IFN-a and IFN-b by ELISA.
Error bars represent standard deviations from the means of dupli-
Figure 6. MAVS Is Required for Antiviral Immune Defense In Vivo
(A) Wild-type (n = 7), Mavs+/2(n = 5), and Mavs2/2(n = 5) mice were
infected with VSV-GFP (2 3 108pfu) via tail vein injection. The sera
were collected from the mice at different time points as indicated
and used to measure viral titers by plaque assays. The error bars in-
dicate the standard error of the mean (SEM).
(B) Sera collected as in (A) were used for measurement of IFN-a and
IFN-b by ELISA. The error bars indicate SEM.
(C) Mavs+/+, Mavs+/2, and Mavs2/2mice (n = 6 for each genotype)
were infected with wild-type VSV (Indiana strain; 5 3 107pfu) via
tail vein injection, and the survival of the mice was monitored for 5
weeks. The mice that survived the viral infections at 2 weeks re-
mained alive after 5 weeks.
mouse, all Mavs2/2mice died within 8 days (n = 5),
whereas two out of five Mavs+/2and five out of seven
wild-type mice survived at 12 days. At 1 3 108pfu per
mouse, 50% of Mavs2/2mice (n = 6) succumbed to viral
infection at day 12, whereas three out of four Mavs+/2
and all of wild-type (n = 4) mice remained alive. When
the viral titer was further reduced to 5 3 107pfu per
mouse, none of the mice died of viral infection within
20 days, although the Mavs2/2mice appeared sick
and later recovered (data not shown). Thus, MAVS pro-
tects the mice from VSV-induced mortality in a manner
thatdepends on its gene dosage as wellas theviral titer.
These results suggest that interferons produced by
pDCs and other cells through the TLR pathway are not
sufficient to protect mice from viral killing and that the
RIG-I-MAVS pathway may provide innate immunity
through local production of interferons and/or other an-
tiviral molecules. In the absence of MAVS, the initial high
viralload mighthave causedirreversible damages tothe
mice so that they failed to recover even after the virus
was cleared. The observation that Mavs+/2mice were
vulnerable to VSV killing is particularly interesting, be-
cause it suggests that the expression level of MAVS is
critical to antiviral immune defense (see below).
MAVS Is Required for Interferon Induction
by Poly(I:C) In Vivo
tribution of interferon production by other cell types, we
sought to examine the role of MAVS in interferon induc-
tion by poly(I:C), which activates the cytosolic RNA
sensing pathway in most cell types. Intravenous injec-
tion of poly(I:C) led to a rapid and robust induction of
IFN-a, IFN-b, and IL-6 in the sera of wild-type mice (Fig-
ures 7A–7C). In contrast, the Mavs2/2mice failed to pro-
duce IFN-a and IFN-b, and the induction of IL-6 within
the first 4 hr of poly(I:C) injection was also impaired.
However, there was a late induction of IL-6 in Mavs2/2
mice, albeit at a much reduced level as compared to
the wild-type (Figure 7C; 8 hr time point). This may be
due to the induction of IL-6 through TLR3, but not
MAVS (see also Figure 3C). Interestingly, Mavs+/2mice
produced intermediate levels of IFN-a, IFN-b, and IL-6,
indicating that the gene dosage of Mavs is important
for the cytosolic RNA signaling pathway. As the amount
of MAVS protein in Mavs+/2cells is about half of that in
the wild-type cells (Figure 1C), these results indicate
that the steady-state level of MAVS protein is a limiting
factor in mounting an effective antiviral response in vivo.
We have presented genetic evidence that MAVS is es-
sential for antiviral innate immune responses in fibro-
blasts, macrophages, and cDCs, but not in pDCs. The
cell type-specific requirement of MAVS in antiviral im-
munity is similar to that of RIG-I (Kato et al., 2005). How-
ever, RIG-I knockout mice are not viable, suggesting
not understood. In contrast, Mavs2/2mice displayed no
apparent developmental abnormality, but they were se-
verely compromised in immune responses against viral
infection. Thus, our results provide the genetic evidence
that the cytosolic viral sensing and signaling pathway
through MAVS is essential for antiviral immune defense
in vivo. Interestingly, normal levels of interferons are still
detected in the sera of Mavs2/2mice, suggesting that
other cell types such as pDCs can still produce inter-
ferons in the absence of MAVS. Nevertheless, Mavs2/2
mice have higher initial viral loads and are much more
vulnerable to viral killing. Thus, although the host im-
mune system is endowed with different cell types that
provide two partially redundant antiviral pathways,
TLR and RIG-I/MDA-5, each ofthese pathways is impor-
tant for effective antiviral responses in vivo.
Previous studies have presented conflicting data con-
cerning the role of MAVS in TLR signaling (Seth et al.,
2006). Although two reports showed that RNAi of
MAVS/IPS-1 did not affect interferon induction by TRIF
or poly(I:C) (Kawai et al., 2005; Seth et al., 2005), another
report found that MAVS/VISA interacts with TRIF and is
required for TLR3 signaling (Xu et al., 2005). Using
MAVS-deficient cells, we have now provided the genetic
evidence that MAVS is not required for interferon
Figure 7. MAVS Is Required for Interferon Induction By Poly(I:C)
Wild-type (n = 5), Mavs+/2(n = 5), and Mavs2/2(n = 6) mice were in-
jected intravenously with 200 mg of poly(I:C) and then sera were col-
lected at indicated times for ELISA to measure the concentrations of
IFN-a (A), IFN-b (B), and IL-6 (C). Littermates of 10- to 12-week-old
mice were used for this experiment. The error bars indicate SEM.
The Role of MAVS in Antiviral Innate Immunity
induction byTLRsinmultiple celltypes.Therefore, TLRs
and RIG-I are parallel antiviral signaling pathways that
detect viral RNA in topologically different locations.
The TLRs involved in viral sensing, including TLR3, 7,
8, and 9, are localized on the endosomal membrane,
with the ligand sensing domain facing the lumen of the
endosome (O’Neill, 2004). These TLRs detect viral RNA
after the endocytosis and disassembly of the viral parti-
cles and then transduce signals through the cytosolic
Toll-IL1 receptor (TIR) domain, which recruits MyD88
or TRIF to activate NFkB and IRFs. In contrast to
TLRs, RIG-I is a cytosolic receptor that detects dou-
ble-stranded RNA generated during viral replication in
the cytosol. It has been proposed that some RNA vi-
ruses such as respiratory syncytial virus (RSV) and
SeV, which enter host cells through fusion with the
plasma membrane, induce type-I interferons through
cytosolic RNA sensing pathway instead of TLR path-
ways in most cells, including pDCs (Hornung et al.,
2004). Indeed, SeV has been shown to induce inter-
ferons in pDCs derived from Myd882/2and Pkr2/2
mice (Hornung et al., 2004). Because signaling by
TLR7 and TLR9 in pDCs is strictly dependent on
MyD88, our finding that Mavs2/2pDCs are still fully ca-
pable of inducing interferons (Figures 5D and 5E) raises
the interesting possibilities that either MAVS and MyD88
function redundantly or there is a third pathway (MAVS
and MyD88 independent) that mediates the induction
of type-I interferons in pDCs in response to infection
by some viruses such as SeV.
The TLR and RIG-I pathways utilize distinct adaptors
to activate IKK and TBK1/IKK3. The use of MAVS as an
essential adaptor in the RIG-I pathway is particularly in-
teresting, as the function of MAVS depends on its local-
ization on the mitochondrial membrane. The importance
of the mitochondrial localization of MAVS is under-
scored by the recent discovery that hepatitis C virus
(HCV) employs a serine protease, NS3/4A, to cleave
MAVS off the mitochondrial membrane, thereby block-
ing interferon induction by the viral RNA (Li et al., 2005;
Meylan et al., 2005). However, the in vivo situations ap-
pear to be more complicated, as microarray analyses
have revealed abundant intrahepatic expression of
ISGs in HCV-infected chimpanzees, suggesting that
IFNs are produced even though MAVS might have
been cleaved in these animals (Bigger et al., 2004; Wie-
land and Chisari, 2005). This conundrum is reminiscent
of our observation that Mavs2/2mice remain capable
of producing high levels of interferons after VSV infec-
tion. Thus, it is possible that both VSV and HCV can in-
duce interferons through the TLR pathway in some cell
types such as pDCs. Although HCV is a hepatotropic vi-
rus, it can also infect and replicate in extrahepatic tis-
sues such as B cells (Machida et al., 2006). In addition,
a recent study has shown that the hepatoma cell line
Huh7 expresses TLR7 (Lee et al., 2006), which could in-
duce interferons in response to viral infection. Although
the source of interferons in HCV-infected host remains
to be investigated, the cleavage of MAVS by the HCV
protease likely contributes to the pathogenesis of this
widespread virus, as suggested by our finding that
Mavs2/2mice were highly susceptible to viral killing.
The importance of MAVS in antiviral defense may par-
tially explain why the HCV protease inhibitors are highly
effective in rapidly reducing HCV viral loads in early-
are expected to not only inhibit viral replication and as-
sembly but also restore the host antiviral innate immu-
nity by blocking the cleavage of MAVS.
Recent studies have uncovered a signaling pathway
that induces interferons in response to cytosolic DNA,
which canbeintroducedtothehostcells through bacte-
rial or viral infection or under certain pathological condi-
et al., 2006; Okabe et al., 2005; Stetson and Medzhitov,
2006). It has also been shown that Listeria monocyto-
genes induces interferon by releasing the bacterial
DNA into the host cytoplasm (Stetson and Medzhitov,
2006). In addition, it was reported that RNAi of MAVS/
IPS-1 partially inhibited the induction of an interferon re-
porter in cultured cells, leading to the proposal that
MAVS/IPS-1 is required for signaling by cytosolic DNA
(Ishii et al., 2006). However, our experiments using
MAVS-deficient cells have shown clearly that MAVS is
not required for interferon induction by cytosolic DNA
or Listeria monocytogenes. Previous studies have also
shown that RIG-I is not required for interferon induction
by cytosolic DNA (Ishii et al., 2006) and that TLR, NOD1,
NOD2, andRIP2 aredispensable forinterferon induction
by Listeria monocytogenes (O’Connell et al., 2005;
Stockinger et al., 2004). Thus, further studies are re-
quired to identify the sensor(s) of cytosolic DNA and
the adaptor(s) that transduce the DNA signals to the ac-
tivation of IKK and TBK1/IKK3.
Our in vivo studies of Mavs mutant mice showed that
both Mavs2/2and Mavs+/2mice were highly vulnerable
to killing by VSV, suggesting that the dosage of MAVS is
critical for antiviral immune defense. Because the Mavs
mutant mice had normal levels of interferons in the cir-
culation, the sensitivity of these mice to viral killing is
likely due tothe defect in the local production of antiviral
molecules (Levy, 2002). Interferons and cytokines may
be such molecules, as the production of IFN-a, INF-b,
and IL-6 was abolished in Mavs2/2mice that were stim-
ulated with poly(I:C). A recent study has shown that
poly(I:C) induces interferons through Mda5, but not
RIG-I (Kato et al., 2006). Thus, our results provide the
in vivo evidence that MAVS is essential for signaling
downstream of Mda5, and it serves as a convergent
point for both Mda5 and RIG-I pathways. Interestingly,
the induction of interferons and IL-6 by poly(I:C) was
that one copy of Mavs is not sufficient to mediate full an-
tiviral responses. That Mavs+/2mice are severely im-
mune compromised may have important implications
in human immunogenetics. It will be of great interest to
analyze the Mavs gene locus in human populations to
determine whether individuals homozygous and hetero-
zygous for Mavs mutations have increased susceptibil-
ity to viral diseases.
Generation of Mavs-Deficient Mice
Genomic DNA containing the Mavs gene was isolated from 129/Sv
mouse ES cell genomic DNA by PCR. The targeting vector was con-
structed by replacing a 1.7 kb fragment spanning from the ATG start
cassette, which also contained two loxP sites flanking PGK-Neo.
After electroporation of the Mavs targeting vector, three indepen-
dently targeted ES cell clones were injected into C57BL/6 blasto-
cysts to produce chimeric mice. Chimeric mice obtained from two
targeted ES clones were bred to C57BL/6 mice to obtain germline
transmission. The heterozygous F1 progenies were intercrossed to
obtain Mavs2/2mice. Mice from these independent clones dis-
played indistinguishable phenotypes. The mice used in this study
were 129/Sv/C57BL/6 hybrids. However, only littermates from the
crossing of heterozygous mice were used in the same experiments.
All mice described in this report were engineered and housed in an-
imal facilities at the University of Texas Southwestern Medical Cen-
To generate polyclonal antibodies against mouse MAVS, a recombi-
nant protein containing residues 127–276 of mouse MAVS was ex-
pressed in E. coli as a His6-tagged protein and affinity purified.
This protein fragment was used to immunize rabbits, and the sera
were further purified by using the antigen column to obtain the
MAVS antibodies. The antibody for mouse IRF3 was purchased
from Zymed Inc, and antibodies for PARP and caspase 3 were
from Cell Signaling Inc. FITC-conjugated antibodies against
CD11c and CD11b, PE-conjugated antibody against B220, and
APC-Annexin V were purchased from BD Pharmingen.
Embryonic fibroblasts from wild-type and mutant mice were pre-
pared from day 13.5 embryos and cultured in DMEM supplemented
with 10% FBS. Bone marrow cells were prepared from the femurs
and tibiae of mice. These cells were cultured in RPMI 1640 contain-
100 ng/ml human Flt3 ligand (peproTech) or 10 ng/ml murine GM-
CSF (peproTech). After 6–8 days, the cells were collected and
used as Flt3L-induced BMDCs or GM-CSF-induced BMDCs, re-
spectively. Flt3L-induced BMDCs were stained with antibodies
against CD11c and B220 and sorted by FACS. FACS sorting was
carried out with FACSVantage SE (with DIVA upgrade) after CD11c
and B220 staining. CD11c+B220+cells and CD11c+B2202cells
were used as bone marrow pDCs and cDCs, respectively. The purity
of pDCs and cDCs was greater than 90% based on FACS analysis.
To isolate BMDMs, 1 3 107bone marrow cells were cultured in
DMEM containing 10% FBS and 10 ng/ml CSF-1 (Sigma). Twenty-
four hours later, nonadherent cells were transferred to a new flask
and cultured for 3 days before 10 ml fresh media containing CSF-1
were added and cells were cultured for another 4 days. Mature mac-
rophages were harvested by collagenase (Roche) digestion and cul-
tured on 96-well or 12-well plates for experiments. For peritoneal
macrophages, cells were obtained by lavage of the peritoneal cavity
rophages were collected and resuspended in DMEM containing
10% FBS for further experiments.
Stimulation of Cells and Functional Assays
To stimulate MEF cells or macrophages with cytosolic DNA, poly
(dA:dT) or poly(dI:dC) (10 mg/ml; GE Biosciences) was incubated
with lipofectamine 2000 (Invitrogen; 1 ml LF2000/mg DNA) at room
temperature for 20 min and then added to cultured cells at the final
cells were incubated with poly(I:C) (10 mg/ml; GE Biosciences), LPS
(10 mg/ml; Sigma), or CpG-2084 DNA (50-TCCTGACGTTGAAGT-30; 5
mg/ml) (Lund et al., 2004), respectively. After incubation for indicated
time periods, cell extracts were prepared to measure NFkB activa-
tion by electrophoretic mobility shift assays (EMSA) or IRF3 activa-
tion by native gel dimerization assays (Seth et al., 2005). For
EMSA, whole-cell extracts were incubated with
NFkB oligos (50-AGTTGAGGGGACTTTCCCAGG-30), and the pro-
tein-DNA complex was resolved by native gel electrophoresis. To
lected for ELISA. The ELISA kits for mouse IFN-a and IFN-b were
purchased from PBL Biomedical Laboratories (Piscataway, NJ),
and the IL-6 ELISA kit was from BD Biosciences (San Diego, CA).
Viral and Bacterial Infection of Cells
SeV (Cantell strain) and VSV (Indiana strain) have been described
previously (Seth et al., 2005). VSV-GFP virus (kindly provided by
Dr. Genhong Cheng; UCLA) was propagated in BHK21 cells (Oga-
nesyan et al., 2006). Listeria monocytogenes (10403 serotype) was
cultured in 3.7% Brain-Heart Infusion overnight (Berg et al., 2003).
The bacteria were washed three times in PBS before being used
to infect cells.
For viral infection, cells grown in media containing 1% FBS were
incubated with viruses at the indicated MOIs for 1 hr before replace-
ment with the complete media containing 10% FBS. For Listeria in-
fection, cells were incubated with the bacteria at MOI of 10 or 100 in
antibiotic-free DMEM containing 10% FBS. After infection for 1 hr,
excess bacteria were washed away and cells were incubated in
complete media containing 50 mg/ml gentamycin.
Viral Infection and Poly(I:C) Injection in Mice and Measurement
of Viral Titer
Mice of different genotypes were infected with VSV (5 3 107pfu per
mouse) or VSV-GFP (1 3 108or 2 3 108pfu per mouse) via tail vein
injection. The viability of the infected mice was monitored for 2–5
weeks. Sera were collected at different time points to measure inter-
feron induction by ELISA, as described above, and viral titers by
plaque assays. For plaque assays, BHK21 cells were incubated
with viral samples at serial dilutions for 1 hr and then overlaid with
1.5% methylcellulose in MEM containing 1% FBS. Forty-eight hours
later, cells were fixed in methanol and stained with 0.1% crystal
violet. Plaques were counted to calculate viral titer. For poly(I:C)
injection, 200 mg of poly(I:C) was injected into each mouse intrave-
nously, and the sera were collected for ELISA as described above.
Supplemental Data include five figure and can be found with this ar-
ticle online at http://www.immunity.com/cgi/content/full/24/5/633/
We thank Dr. Robert Hammer and the Transgenic Core Facility at
University of Texas Southwestern Medical Center for ES cell target-
ing, blastocyst injection, and generation of chimeric mice. We are
grateful to Dr. Michelle Tallquist (UT Southwestern) for providing
ingstrategy. We also thank Dr.Genhong Cheng(UCLA)forproviding
the VSV-GFP virus. This work was supported by grants from the Na-
tional Institutes of Health and the Welch Foundation. Z.J.C. is an In-
vestigator of the Howard Hughes Medical Institute and a Burroughs
Wellcome Fund Investigator in Pathogenesis of Infectious Diseases.
Received: March 9, 2006
Revised: April 18, 2006
Accepted: April 25, 2006
Published: May 23, 2006
Berg, R.E., Crossley, E., Murray, S., and Forman, J. (2003). Memory
CD8+ T cells provide innate immune protection against Listeria
monocytogenes in the absence of cognate antigen. J. Exp. Med.
Bigger, C.B., Guerra, B., Brasky, K.M., Hubbard, G., Beard, M.R.,
Luxon, B.A., Lemon, S.M., and Lanford, R.E. (2004). Intrahepatic
gene expression during chronic hepatitis C virus infection in chim-
panzees. J. Virol. 78, 13779–13792.
Dalton, K.P., and Rose, J.K. (2001). Vesicular stomatitis virus gly-
coprotein containing the entire green fluorescent protein on its
cytoplasmic domain is incorporated efficiently into virus particles.
Virology 279, 414–421.
Darnell, J.E., Jr., Kerr, I.M., and Stark, G.R. (1994). Jak-STAT path-
tracellular signaling proteins. Science 264, 1415–1421.
The Role of MAVS in Antiviral Innate Immunity
Diebold, S.S., Kaisho, T., Hemmi, H., Akira, S., and Reis e Sousa, C.
(2004). Innate antiviral responses by means of TLR7-mediated rec-
ognition of single-stranded RNA. Science 303, 1529–1531.
Fitzgerald, K.A., McWhirter, S.M., Faia, K.L., Rowe, D.C., Latz, E.,
Golenbock, D.T., Coyle, A.J., Liao, S.M., and Maniatis, T. (2003).
IKKepsilon and TBK1 are essential components of the IRF3 signal-
ing pathway. Nat. Immunol. 4, 491–496.
Freundt, E.C., and Lenardo, M.J. (2005). Interfering with interferons:
Hepatitis C virus counters innate immunity. Proc. Natl. Acad. Sci.
USA 102, 17539–17540.
Heil, F., Hemmi, H., Hochrein, H., Ampenberger, F., Kirschning, C.,
Akira, S., Lipford, G., Wagner, H., and Bauer, S. (2004). Species-spe-
cific recognition of single-stranded RNA via toll-like receptor 7 and
8. Science 303, 1526–1529.
Honda, K., Yanai, H., Takaoka, A., and Taniguchi, T. (2005). Regula-
tion of the type I IFN induction: a current view. Int. Immunol. 17,
Hornung, V., Schlender, J., Guenthner-Biller, M., Rothenfusser, S.,
Endres, S., Conzelmann, K.K., and Hartmann, G. (2004). Replica-
tion-dependent potent IFN-alpha induction in human plasmacytoid
dendritic cells by a single-stranded RNA virus. J. Immunol. 173,
Ishii, K.J., Coban, C., Kato, H., Takahashi, K., Torii, Y., Takeshita, F.,
Ludwig, H., Sutter, G., Suzuki, K., Hemmi, H., et al. (2006). A Toll-like
receptor-independent antiviral response induced by double-
stranded B-form DNA. Nat. Immunol. 7, 40–48.
Kato, H., Sato, S., Yoneyama, M., Yamamoto, M., Uematsu, S., Mat-
sui, K., Tsujimura, T., Takeda, K., Fujita, T., Takeuchi, O., and Akira,
S. (2005). Cell type-specific involvement of RIG-I in antiviral re-
sponse. Immunity 23, 19–28.
Kato, H., Takeuchi, O., Sato, S., Yoneyama, M., Yamamoto, M., Mat-
sui, K., Uematsu, S., Jung, A., Kawai, T., Ishii, K.J., et al. (2006). Dif-
ferential rolesof MDA5 andRIG-I helicases in the recognition of RNA
viruses. Nature 441, 101–105.
Kawai, T., and Akira, S. (2006). Innate immune recognition of viral in-
fection. Nat. Immunol. 7, 131–137.
Kawai, T., Takahashi, K., Sato, S., Coban, C., Kumar, H., Kato, H.,
Ishii, K.J., Takeuchi, O., and Akira, S. (2005). IPS-1, an adaptor trig-
gering RIG-I- and Mda5-mediated type I interferon induction. Nat.
Immunol. 6, 981–988.
Lamarre, D., Anderson, P.C., Bailey, M., Beaulieu, P., Bolger, G.,
Bonneau, P., Bos, M., Cameron, D.R., Cartier, M., Cordingley,
M.G., et al. (2003). An NS3 protease inhibitor with antiviral effects
in humans infected with hepatitis C virus. Nature 426, 186–189.
Lee, J., Wu, C.C., Lee, K.J., Chuang, T.H., Katakura, K., Liu, Y.T.,
Chan, M., Tawatao, R., Chung, M., Shen, C., et al. (2006). Activation
Acad. Sci. USA 103, 1828–1833.
Levy, D.E. (2002). Whence interferon? Variety in the production of in-
terferon in response to viral infection. J. Exp. Med. 195, F15–F18.
Li, X.D., Sun, L., Seth, R.B., Pineda, G., and Chen, Z.J. (2005). Hep-
atitisC virus protease NS3/4A cleaves mitochondrial antiviralsignal-
ing protein off the mitochondria to evade innate immunity. Proc.
Natl. Acad. Sci. USA 102, 17717–17722.
Liu, Y.J. (2001). Dendritic cell subsets and lineages, and their func-
tions in innate and adaptive immunity. Cell 106, 259–262.
Lund, J.M., Alexopoulou, L., Sato, A., Karow, M., Adams, N.C., Gale,
N.W., Iwasaki, A., and Flavell, R.A. (2004). Recognition of single-
stranded RNA viruses by Toll-like receptor 7. Proc. Natl. Acad. Sci.
USA 101, 5598–5603.
Machida, K., Cheng, K.T., Sung, V.M., Levine, A.M., Foung, S., and
Lai, M.M. (2006). Hepatitis C virus induces toll-like receptor 4 ex-
pression, leading to enhanced production of beta interferon and in-
terleukin-6. J. Virol. 80, 866–874.
Maniatis, T., Falvo, J.V., Kim, T.H., Kim, T.K., Lin, C.H., Parekh, B.S.,
and Wathelet, M.G. (1998). Structure and function of the interferon-
beta enhanceosome. Cold Spring Harb. Symp. Quant. Biol. 63, 609–
McWhirter, S.M., Tenoever, B.R., and Maniatis, T. (2005). Connect-
ing mitochondria and innate immunity. Cell 122, 645–647.
Meylan, E., Curran, J., Hofmann, K., Moradpour, D., Binder, M., Bar-
tenschlager, R., and Tschopp, J. (2005). Cardif is an adaptor protein
ture 437, 1167–1172.
O’Connell, R.M., Vaidya, S.A., Perry, A.K., Saha, S.K., Dempsey,
P.W., and Cheng, G. (2005). Immune activation of type I IFNs by Lis-
teria monocytogenes occurs independently of TLR4, TLR2, and re-
ceptor interacting protein 2 but involves TNFR-associated NF kappa
B kinase-binding kinase 1. J. Immunol. 174, 1602–1607.
O’Neill, L.A. (2004). Immunology. After the toll rush. Science 303,
Oganesyan, G., Saha, S.K., Guo, B., He, J.Q., Shahangian, A., Zarne-
gar, B., Perry, A., and Cheng, G. (2006). Critical role of TRAF3 in the
Toll-like receptor-dependent and -independent antiviral response.
Nature 439, 208–211.
Toll-like receptor-independent gene induction program activated by
mammalian DNA escaped from apoptotic DNA degradation. J. Exp.
Med. 202, 1333–1339.
Perry, A.K., Chen, G., Zheng, D., Tang, H., and Cheng, G. (2005). The
host type I interferon response to viral and bacterial infections. Cell
Res. 15, 407–422.
Seth, R.B., Sun, L., Ea, C.K., and Chen, Z.J. (2005). Identification and
characterization of MAVS, a mitochondrial antiviral signaling protein
that activates NF-kappaB and IRF 3. Cell 122, 669–682.
Seth, R.B., Sun, L., and Chen, Z.J. (2006). Antiviral innate immunity
pathways. Cell Res. 16, 141–147.
Sharma, S., tenOever, B.R., Grandvaux, N., Zhou, G.P., Lin, R., and
Hiscott, J. (2003). Triggering the interferon antiviral response
through an IKK-related pathway. Science 300, 1148–1151.
Silverman, N., and Maniatis, T. (2001). NF-kappaB signaling path-
ways in mammalian and insect innate immunity. Genes Dev. 15,
Stetson, D.B., and Medzhitov, R. (2006). Recognition of cytosolic
DNA activates an IRF3-dependent innate immune response. Immu-
nity 24, 93–103.
Stockinger, S., Reutterer, B., Schaljo, B., Schellack, C., Brunner, S.,
Materna, T., Yamamoto, M., Akira, S., Taniguchi, T., Murray, P.J.,
et al. (2004). IFN regulatory factor 3-dependent induction of type I
IFNs by intracellular bacteria is mediated by a TLR- and Nod2-inde-
pendent mechanism. J. Immunol. 173, 7416–7425.
Wieland, S.F., and Chisari,F.V. (2005).Stealth and cunning: hepatitis
B and hepatitis C viruses. J. Virol. 79, 9369–9380.
VISA is an adapter protein required for virus-triggered IFN-beta sig-
naling. Mol. Cell 19, 727–740.
Yoneyama, M., Kikuchi, M., Natsukawa, T., Shinobu, N., Imaizumi,
T., Miyagishi, M., Taira, K., Akira, S., and Fujita, T. (2004). The RNA
helicase RIG-I has an essential function in double-stranded RNA-
induced innate antiviral responses. Nat. Immunol. 5, 730–737.