Activation of the Innate Signaling Molecule MAVS
by Bunyavirus Infection Upregulates
Piyali Mukherjee,1Tyson A. Woods,1Roger A. Moore,1and Karin E. Peterson1,*
1Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institutes of Health, 903 S. 4thStreet, Hamilton,
MT 59840, USA
La Crosse virus (LACV), a zoonotic Bunyavirus, is
a major cause of pediatric viral encephalitis in the
United States. A hallmark of neurological diseases
caused by LACV and other encephalitic viruses is
the induction of neuronal cell death. Innate immune
damage, but no mechanism has been elucidated.
By using in vitro studies in primary neurons and
in vivo studies in mice, we have shown that LACV
infection induced the RNA helicase, RIG-I, and mito-
chondrial antiviral signaling protein (MAVS) signaling
pathway, resulting in upregulation of the sterile alpha
and TIR-containing motif 1 (SARM1), an adaptor
molecule that we found to be directly involved in
neuronal damage. SARM1-mediated cell death was
associated with induced oxidative stress response
and mitochondrial damage. These studies provide
an innate-immune signaling mechanism for virus-
induced neuronal death and reveal potential targets
for development of therapeutics to treat encephalitic
The innate immune response protects against virus infections
by production of type I interferons that mediate antiviral host
responses and cytokines that recruit inflammatory cells. This
innate immune response can also induce cellular damage
(Chattopadhyay et al., 2010; Fink et al., 2008; Lei et al., 2009;
McAllister and Samuel, 2009). Studies have demonstrated
direct innate immune signaling-mediated cell death with activa-
tion of pattern recognition receptors such as Toll-like receptors
(TLRs) or RNA helicase receptors (RLRs) leading to cellular
damage and, sometimes, apoptosis (Cameron et al., 2007;
Lathia et al., 2008; Ma et al., 2006, 2007; Tang et al., 2008). In
the brain, innate immune-induced apoptosis following virus
infection may be a contributing factor to neuronal damage
and neuronal dropout. Identification of specific targets that
initiate apoptosis during virus infection of neurons will offer
important insight into the mechanisms of neurodegeneration
and will provide targets for the development of antiviral
One protein that may have a role in innate immune signaling-
mediated cell death is sterile alpha and Toll/interleukin-1 (IL-1)
receptor (TIR) motif-containing 1 protein (SARM1, MyD88-5).
This protein is a member of the TIR-containing adaptor family
and, in immune cells, acts as a negative regulator of TLR-
mediated NF-kB activation (Carty et al., 2006; Peng et al.,
2010) and contributes to T cell apoptosis (Panneerselvam
et al., 2013). In neurons, SARM1 interacts with syndecan 2
and regulates neuronal morphogenesis (Chen et al., 2011).
Studies using GFP-tagged SARM1 show that under condi-
tions of metabolic stress in neurons, SARM1 translocates
to the mitochondria, interacts with c-Jun N-terminal kinase 3
(JNK3), and mediates neuronal apoptosis (Kim et al., 2007).
SARM1 has also been identified as mediating axonal death,
although the mechanism is unknown (Osterloh et al., 2012).
The role of SARM1 in the innate immune response and
neuronal or axon death has prompted questions about the
function for SARM1 in inducing neuronal damage during virus
infections in the central nervous system (CNS) and whether
this damage would be mediated through innate immune
To investigate the role of SARM1 in virus-induced neuronal
death, we utilized La Crosse virus (LACV), an enveloped triseg-
mented negative-sense RNA virus belonging to the family
Bunyaviridae. LACV is a major cause of pediatric viral enceph-
alitis in the USA and is an emerging pathogen due to increased
vector hosts and range (Gerhardt et al., 2001; Haddow and
Odoi, 2009; McJunkin et al., 2001). Neurons are the pre-
dominant cell type infected with the virus in the CNS and
LACV-mediated encephalitis is associated with degenerative
neuronal changes characteristic of apoptotic cells, including
nuclear vacuolization and cell shrinkage (Bennett et al., 2008;
Kalfayan, 1983; Pekosz et al., 1996). In this study, we have
demonstrated a clear role for SARM1 in mediating LACV-
induced neuronal apoptosis. We have also determined the
mechanisms for SARM1 induction and SARM1-induced cell
death during LACV infection. We show that both protective
type I interferon (IFN) and damaging SARM1-induced re-
sponses are generated following virus stimulation of the RNA
helicase, retinoic acid-inducible gene 1 protein (RIG-I), and
subsequent activation of mitochondrial antiviral signaling
protein (MAVS) signaling pathway in neurons.
Immunity 38, 705–716, April 18, 2013 ª2013 Elsevier Inc. 705
SARM1 Is Induced in LACV-Infected Neurons
SARM1 is highly conserved from chordates to humans (Mink
et al., 2001), suggesting a conserved function. Given that
SARM1 influences neuronal death following oxygen and glucose
deprivation (OGD) or axonal injury (Osterloh et al., 2012; Yuan
et al., 2010; Kim et al., 2007), we investigated whether SARM1
also influenced virus-mediated neuronal damage. Primary
cortical neurons were infected with LACV at different multiplici-
ties of infection (MOI) and followed for gene expression and
cell death. LACV RNA was detectable in infected neuronal
cultures as early as 6 hr postinfection (hpi) and increased loga-
rithmically until reaching a plateau between 18 to 36 hpi,
depending on MOI (Figure 1A). LACV-induced neuronal death
started at 24 hpi in cultures infected with the highest MOI of
virus with widespread cell death by 72 hpi for all MOIs tested
(Figure 1B). Dying neurons were TUNEL positive (Figure 1C)
and associated with increased caspase-3 activity, typical of
apoptotic neurons (Figure 1D). Analysis of SARM1 expres-
sion during LACV infection demonstrated increased Sarm1
messenger RNA (mRNA) as early as 18 hpi at the highest MOI
and by 24 to 30 hpi for the lower MOIs (Figure 1E). A correspond-
ing increase was observed with SARM1 protein in whole-cell
lysates of LACV-infected neurons compared to mock-infected
controls (Figure 1F). Thus, LACV infection resulted in the upregu-
lation of Sarm1 mRNA and protein expression.
SARM1 Contributes to LACV-Mediated Neuronal Death
Because SARM1 was induced during LACV infection and upre-
gulated prior to apoptosis, we examined whether SARM1 had
a role in LACV-mediated cell death. Transfection of neurons
with small interfering RNA (siRNA) targeted to Sarm1 prior to
infection significantly reduced LACV-mediated neuronal death
(Figure 2A). Similar results were observed with neurons from
Sarm1?/?mice as determined by both MTT assay and TUNEL
staining (Figures 2B–2D), with a significant inhibition or delay in
neuronal death compared to C57BL/6 wild-type (WT) neurons.
SARM1 deficiency did not inhibit virus replication with cells
from Sarm1?/?cultures expressing high amounts of virus (red
fluorescence) (Figure 2E), but not undergoing the same amount
of cell death (green fluorescence) as observed in infected cells
Figure 1. LACV Infection of Primary Cortical Neurons Induces Apoptotic Death and Increased Production of SARM1
(A) Cortical neurons were infected with LACV at MOIs of 1, 0.1, and 0.01. Virus RNA was measured at 4, 8, 12, 18, 24, 36, and 48 hpi by real-time PCR and
calculated as percent of Gapdh expression.
(B) Cell viability of neurons following LACV infection was measured by MTT assay.
(C) Neurons infected with LACV (MOI 0.01) were stained with TUNEL reagent. Cells were counted from five to six images for each group and a percentage of
TUNEL-positive cells per total number of nuclei per image determined.
(D) At 36 hpi, neurons were analyzed for caspase-3 activity.
(E) RNA from cells in (A) was analyzed for Sarm1 mRNA by real-time PCR and calculated as percentage of Gapdh mRNA expression. (A–E) Data are the mean ±
SEM for three or more samples per group per time point and are representative of duplicate experiments.
(F) Immunoblot analysis of SARM1 and b-actin in whole cell lysates of mock and LACV infected neurons from WT and Sarm1?/?mice at 36 hpi.
Neuronal Death Mediated by MAVS Induction of SARM1
706 Immunity 38, 705–716, April 18, 2013 ª2013 Elsevier Inc.
from WT cultures (Figures 2C–2E). Analysis of viral RNA expres-
sion in neuronal cultures, controlled for the number of cells by
Gapdh mRNA expression, showed similar amounts of virus
RNA (Figure 2F). Thus, SARM1 mediates neuronal death during
LACV infection, by a mechanism unrelated to suppression of
virus replication in neurons.
SARM1 Deficiency Suppresses Viral Pathogenesis
of LACV Infection In Vivo
To examine whether SARM1 deficiency would alter LACV path-
ogenesis in vivo, we infected C57BL/6 WT and Sarm1?/?mice
with 103plaque forming units (PFU) of LACV intraperitoneally,
a dose that induces 100% incidence of neurological disease in
3-week-old WT mice (Figure 3A). Sarm1?/?mice developed
neurological disease at a significantly lower incidence than WT
mice,indicating adetrimental roleforSARM1duringLACVinfec-
tion. In comparison, deficiency in MyD88, a family member of
SARM1, did not significantly affect neurological disease devel-
opment. Analysis of viral RNA from brain tissue of wild-type
and Sarm1?/?mice at 5 dpi, just prior to the onset of disease
in wild-type mice, revealed similar amounts of viral RNA in
Sarm1?/?mice compared to wild-type controls (Figure 3B). In
in neuronal cell bodies (see Figure S1A available online).
However, in focal areas in the cortex of LACV infected mice,
SARM1 was observed localized to the axons of neurons (Fig-
ure S1B). Analysis of 5 dpi preclinical WT mice demonstrated
numerous TUNEL-positive cells (green) in areas of virus infection
(red) (Figure 3C; Figure S1G). In contrast, Sarm1?/?mice at 5 dpi
had significantly fewer TUNEL-positive cells, despite large areas
of virus infection (Figures 3D and 3E; Figure S1H). Thus, SARM1
deficiency inhibits LACV-induced damage and death in vivo,
through a mechanism independent of virus replication, similar
to the results from neuronal cultures in vitro.
SARM1 Localizes to the Mitochondria and Is Associated
with Mitochondrial Damage
The N-terminal domain of SARM1 contains a mitochondrial
localization signal sequence (Panneerselvam et al., 2012).
Analysis of LACV-infected neurons demonstrated increased
controls, which was not observed in the cytosol fraction (Fig-
ure 4A). Similar results were observed in the mitochondrial
fraction of brain tissue from LACV-infected mice (Figure S1C).
The c-Jun N-terminal kinase 3 (JNK3), which associates with
SARM1 following metabolic stress (Kim et al., 2007), was also
increased in mitochondrial fractions following LACV infection
as was phosphorylated JNK (Figure 4A).
Despite mitochondrial localization, the mechanism by which
SARM1 induces neuronal apoptosis is unknown. Analysis of
neurons from WT mice showed numerous neurons containing
swollen and/or degenerative mitochondria in cell bodies and
axons in LACV-infected cultures (Figure 4C), but not in mock-
infected controls (Figure 4B). In contrast, LACV-infection of
neurons did not induce mitochondrial damage
(Figures 4D and 4E). Mitochondrial damage can be induced via
reactive oxygen species (ROS) formation, which has been impli-
cated in some cases of neuronal damage (Pan et al., 2009).
Staining with Mitosox red, an indicator of mitochondrial super-
oxide production, indicated that LACV-infected neurons had
increased superoxide production following LACV infection
(Figures 4F and 4G). However, superoxide production was not
detected in LACV-infected neurons from Sarm1?/?mice indi-
cating that SARM1 was necessary for LACV-induced generation
of ROS (Figure 4H). Furthermore, analysis of genes that are often
induced in response to oxidative stress demonstrated increased
mRNA expression following LACV infection in WT neurons,
but not in Sarm1?/?neurons (Figure S2A). Differences in
mRNA expression of oxidative stress response genes were
Figure 2. SARM1 Deficiency Limits LACV-
Mediated Neuronal Death but Not Virus
Infection or Replication in Neurons
(A) Neurons were treated with Sarm1 siRNA or
nontargeting (NT) siRNA starting at 3 days
postculture. Twenty-four hours later, cells were
infected with LACV and cell viability wasmeasured
in neurons treated with NT or Sarm1 siRNA is
shown in inset on right.
(B) Neurons from WT and Sarm1?/?mice were
infected with LACV, and cell viability was mea-
sured at 24, 30, and 36 hr by MTT assay.
(C–E) Cells were stained for DNA fragmentation by
TUNEL analysis at 36 hpi and counted as percent
positive as described in Figure 1. (D and E)
Immunofluorescence analysis of neurons from (D)
WT and (E) Sarm1?/?mice infected with LACV
at 36 hpi with TUNEL-stained nuclei (green) and
LACV (red). The relative lack of cells in (D)
compared to (E) is due to cell death. Scale bar
represents 25 mm.
(F) Viral RNA expression in neurons at 36 hpi as
determined by real-time PCR. All data are shown
as mean ± SEM of 3–6 samples per group per time
point. Data are representative of 2 or 3 replicate
experiments per panel. See also Figure S3.
Neuronal Death Mediated by MAVS Induction of SARM1
Immunity 38, 705–716, April 18, 2013 ª2013 Elsevier Inc. 707
also observed in vivo to a lesser extent (data not shown). These
data indicate that SARM1 is necessary for oxidative stress
response to virus infection and contributes to LACV-induced
SARM1-Mediated Damage Is Independent of Interleukin
1 or Type I IFN Pathways
Mechanisms of innate immune signaling-mediated cell death
have been described including caspase 1-dependent IL-1 medi-
ated pyroptosis, IFN-mediated apoptosis, and IFN regulatory
topadhyay et al., 2010; Fink et al., 2008). We therefore investi-
gated whether SARM1-mediated neuronal death was influenced
not related to pyroptosis because caspase-1 inhibitors did
not suppress neuronal death in vitro, SARM1 deficiency did not
affect Il1a or Il1b mRNA expression, and LACV infection did
not induce pro-IL-1b cleavage (Figures S3A–S3C, data not
Figure 3. SARM1 Deficiency Protects Mice from LACV-Induced Neuronal Damage
(A) Mice at 20 to 21 days of age were infected with 103PFU of LACV by intraperitoneal infection and followed for the development of clinical disease. Survival
curve for each strain was determined using Kaplan-Meier analysis of 32 WT, 16 Irf3?/?and Irf7?/?, 38 Myd88?/?, 19 Mavs?/?, and 29 Sarm1?/?mice. P value
for Sarm1?/?versus WT was < 0.01.
(B) Viral RNA and Ifna4 mRNA was quantified from brains of WT and Sarm1?/?mice at 5 dpi. No difference was observed in viral RNA between WT and Sarm1?/?
mice. A decrease in Ifna4 mRNA was observed in Mavs?/?mice. Data are shown as mean ± SEM of 5–8 samples per group.
(C–E) Immunohistochemical analysis of TUNEL (green) and LACV (red) in the cortex of (C) WT and (D) Sarm1?/?mice at 5 dpi. Scale bar represents 100 mm. (E)
Number ofTUNEL-positivecells per fieldofview (2003)fromfocalareaofvirusinfection inbrain tissuefromWT andSarm1?/?mice.Dataareshownasindividual
points with mean +/? SEM plotted. See also Figure S1.
Neuronal Death Mediated by MAVS Induction of SARM1
708 Immunity 38, 705–716, April 18, 2013 ª2013 Elsevier Inc.
shown). Stimulation of neurons with type I IFNs did not signifi-
cantly alter Sarm1 mRNA expression either in the presence or
absence of LACV infection (Figure S3D). Additionally, deficiency
in Irf3 and Irf7 or IFN-a receptor (Ifnar1) did not influence LACV-
induced cell death or SARM1 expression, suggesting that
SARM1-induced neuronal death was not dependent on IRF3 or
type I IFN responses (Figure S3E; data not shown). SARM1 defi-
ciency did not suppress type I IFN responses of neurons either
in vitro or in vivo (Figure S2B; data not shown), indicating that
SARM1 does not mediate antiviral type I IFN responses.
Figure 4. LACV Infection Induces Mitochondrial Localization of SARM1 and Mitochondrial Damage
at 36 hpi. CoxIV and b-actin were used as loading controls. SARM1 whole-cell lysate is shown in Figure 1F. Data are representative of duplicate or triplicate
experiments. Graphs below are the mean amounts of SARM1 protein in the mitochondrial and cytosol fractions from mock and LACV-infected neurons. Data are
the mean ± SEM of densitometry readings from three experiments.
(B–E)Morphologicalchanges inthemitochondriaassociatedwithLACVinfection.TEMimageswereacquiredat36hpifrom(Band D)mock- and(CandE)LACV-
infected neurons. Data are representative of 5–6 fields for each group. Scale bars represent 100 mm (F–J). Images of (F) mock- and (G–J) LACV-infected neurons
from (F–G) WT or (H) Sarm1?/?, (I) Mavs?/?, or (J) Mavs?/?cells treated with imiquimod at 36 hpi stained with 5 uM of Mitosox red. First panel for each image is
single channel for Mitosox red and the second panel is Mitosox red plus DAPI. Images are representative of cells in culture. Scale bar represents 10 mm. See also
Neuronal Death Mediated by MAVS Induction of SARM1
Immunity 38, 705–716, April 18, 2013 ª2013 Elsevier Inc. 709