Original Article

Innate Immunity
2014, Vol. 20(2) 161–172

Neuronally-expressed Sarm1 regulates ! The Author(s) 2013

Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
expression of inflammatory and antiviral DOI: 10.1177/1753425913485877
ini.sagepub.com
cytokines in brains

Chia-Wen Lin1,2, Hsin-Yu Liu2,3, Chiung-Ya Chen1 and

Yi-Ping Hsueh1,2,3

Abstract
Sarm1 is the fifth Toll/IL-1 receptor (TIR) domain-containing adaptor protein identified to regulate TLR downstream
signaling. Unlike the other TIR domain-containing adaptor proteins, Sarm1 is predominantly expressed in the brain. Our
previous study indicated that Sarm1 regulates dendritic growth, axonal extension and neuronal polarity. Here, we
investigated whether Sarm1 is involved in innate immunity in the brain. First, regional and cell-type distribution of
Sarm1 in mouse brains was revealed using double immunostaining. Sarm1 was widely distributed in different regions
of brains, including the cerebral cortex, hippocampus, amygdala, cerebellum and midbrain. Moreover, Sarm1 is present in
both projection and inhibitory neurons, but, interestingly, not in microglial cells—the main immune cells in the brain.
These results suggest that Sarm1 is unlikely to regulate microglial activity in a cell-autonomous manner. However,
compared with wild type littermates, the RNA expression levels of several inflammatory and antiviral cytokines were
altered in the embryonic and adult brains of Sarm1 knockdown transgenic mice. These data imply that Sarm1 influences
cytokine expression in neurons. In conclusion, our findings suggest that Sarm1 regulates the innate immune responses of
the central nervous system through regulating the inflammatory and anti-virus cytokines produced by neurons.

Keywords
Interferon, interleukin-6, Sarm1, TIR domain-containing adaptor protein, central nervous system
Date received: 1 November 2012; revised: 21 February 2013; 13 March 2013; accepted: 18 March 2013

(HEK) cells and PBMC.5 However, compared with

Introduction
TLRs are the most well-studied type of pattern recog-
nition receptors. They are expressed in a variety of cells
in both vertebrates and invertebrates where they recog-
nize both pathogen- and damage-associated molecular
patterns to initiate innate immunity.1–3 TLRs mainly
interact with Toll/IL-1 receptor (TIR) domain-contain-
ing adaptors, such as Myd88, Myd88-adaptor-like
(Mal), TIR-domain-containing adapter-inducing inter-
feron-b (TRIF) and TRIF-related adaptor molecule
(TRAM) through cytoplasmic TIR domains to activate
the downstream signaling pathways.1–4 Sterile alpha
and HEAT/Armadillo motif containing 1 (Sarm1)
was the fifth TIR domain-containing adaptor to be dis-
covered;4 however, the function of Sarm1 in innate
immune response remains uncertain. Sarm1 was origin-
ally shown to down-regulate TRIF-dependent TLR3
and TLR4 signaling, and attenuate the expression of
TNF-a, IL-8 and CCL5 in human embryonic kidney
wild type mice, lower levels of TNF and reduced micro-
glial activation were found in Sarm1-/- brainstems after
West Nile virus infection,6 suggesting that the function
of Sarm1 in neural innate immune response may not be
simple down-regulation.
Although Sarm1 was originally studied with respect
to innate immunity, it is predominantly expressed in the
1
Molecular Cell Biology, Taiwan International Graduate Program, Institute
of Molecular Biology, Academia Sinica, and Graduate Institute of Life
Sciences, National Defense Medical Center, Taipei, Taiwan
2
Institute of Molecular Biology, Academia Sinica, Taipei, Taiwan
3
Graduate Institute of Life Sciences, National Defense Medical Center,
Taipei, Taiwan
Corresponding author:
Yi-Ping Hsueh, Institute of Molecular Biology, Academia Sinica, 128, Sec 2,
Academia Rd, Nankang, Taipei, 11529, Taiwan.
Email: yph@gate.sinica.edu.tw
162
brain.7,8 In Caenorhabditis elegans, Sarm1/tir-1 has
been shown to regulate left–right asymmetric expres-
sion of odorant receptor genes in olfactory neurons.9
And, in mammals, Sarm1 has been shown to play mul-
tiple roles in controlling neuronal morphology.8 Using
cultured hippocampal neurons, we recently demon-
strated that Sarm1 interacts with transmembrane
heparan sulfate proteoglycan syndecan-2 and mediates
the function of syndecan-2 to control dendritic arbor-
ization.8 In addition, Sarm1 also regulates axonal
extension and the establishment of neuronal polarity.8
The function of Sarm1 in neuronal morphology is likely
through the MKK4-JNK pathway, which influences
microtubule stability and controls neuronal morph-
ology.8 In addition to neuronal morphogenesis,
Sarm1 is also required for the injury-induced axon
death pathway in both flies and rodents.10 All these
studies indicate that Sarm1 is crucial for the regulation
of neuronal morphology.
Our previous study demonstrated that Sarm1 is
expressed in neurons.8 However, the anatomic distribu-
tion of Sarm1 in the brain has not yet been analyzed in
detail. To further explore the function of Sarm1 in the
brain, here we investigated the cell-type expression pat-
terns and regional expression patterns of Sarm1 in
mouse brain using double fluorescence immunostain-
ing. We found that Sarm1 is widely distributed in dif-
ferent brain regions and expressed in a variety of
neurons, but notably not in microglia. However,
knockdown of Sarm1 altered expression of several
cytokines in the brain suggesting that Sarm1 also regu-
lates inflammatory and anti-virus cytokine expression
in the brain. In view of these results, microglial cells are
unlikely to be directly affected by Sarm1. A more likely
scenario is that Sarm1 directly regulates cytokine
expression in neurons to contribute to innate immunity
in the brain.
Materials and methods
Abs
The primary Abs used for immunohistochemistry and
immunoblotting were as follows: rabbit monoclonal
anti-CaMKII (EP1829Y; Abcam, Cambridge, MA,
USA); rat monoclonal anti-CD11b (M1/70; Abcam,
Cambridge, MA, USA); rabbit polyclonal anti-
GAD65/67 (AB1511; Millipore, Taipei, Taiwan);
rabbit monoclonal anti-calbindin (C12D26; Cell
Signaling, Danvers, MA, USA); rabbit polyclonal
anti-tyrosine hydroxylase (H-196; Santa Cruz, Dallas,
TX, USA); rabbit polyclonal anti-Iba1(019-19741;
Wako Chemicals, Osaka, Japan); rabbit polyclonal
anti-IL6 (AAM15G; AbD Serotec, Raleigh, NC,
USA); rabbit polyclonal anti-IFNb (ab140211;
Abcam, Cambridge, MA, USA); mouse anti-valosin-
containing protein (VCP) Ab (612183; BD
Innate Immunity 20(2)
Transduction Laboratories, Taipei, Taiwan); and
HRP-conjugated goat anti-mouse IgG Ab (NA931;
GE Healthcare, Taipei, Taiwan); mouse monoclonal
anti-Sarm1 was generated as described.8
Immunohistochemistry
Adult B6 mice (8 wk old) were anesthetized and trans-
cardially perfused first with PBS and then 4%
paraformaldehyde (PFA) in PBS. After removing the
skull, the brains were post-fixed in 4% PFA at 4 C for
12–15 h (overnight). Vibratome slices (50 mm) were per-
meabilized with 0.3% Triton X-100 in PBS for 10 min
at 23–25 C (room temperature). For blocking, slices
were incubated in a solution containing 3% horse
serum, 2% BSA and 0.3% Triton X-100 in PBS for
2 h at room temperature with gentle shaking. Primary
Abs were diluted in the blocking solution and incubated
with brain slices for 2 days at 4 C with gentle shaking.
After washing out excess Abs, secondary Abs conju-
gated with Alexa 488 and 555 (Invitrogen, Carlsbad,
CA, USA) and DAPI were incubated with the slices
for 2 h at room temperature on a shaker. After washing,
slices were mounted onto glass slides and soaked in
Vectashield Mounting Medium (H-1000; Vector
Laboratories, Burlingame, CA, USA) for imaging.
For fetal brain staining, whole embryos at E14.5 were
washed with PBS and fixed in 4% PFA for 2.5 h at 4 C.
After washing in PBS, the embryos were rotated in 30%
sucrose at 4 C overnight. After OCT embedding and
cryosectioning (20 mm), sections were blocked with 5%
horse serum and 0.2% Triton X-100 in PBS for 30 min
with gentle shaking. Primary Abs were diluted in block-
ing solution and incubated overnight at 4 C. Secondary
Abs were incubated as described for adult brain sec-
tions. Images were acquired using a confocal micro-
scope (LSM700, Carl Zeiss, Oberkochen, Germany),
and Zen2009 (Carl Zeiss, Oberkochen, Germany)
acquisition and analysis software. All samples were
imaged at 20–22 C. For publication, images were pro-
cessed using Photoshop (Adobe, Taipei, Taiwan) with
minimal adjustment of brightness or contrast to the
entire images.
Western blot analysis
Whole brains of adult mice were homogenized with a
glass pestle in PBS-based lysis buffer containing 1 mM
EDTA, 10 mM b-glycerol phosphate, 1 mg/ml leupep-
tin, 1 mg/ml aprotinin, 1 mM PMSF and 20 mM NaF.
After quantification, equal amounts of brain lysates
were boiled with sample buffer containing 100 mM
2-mercaptoethanol at 95 C for 5 min. For microglia
lysate, BV2 cells seeded in 6-well plates were
treated with LPS (InvivoGen, San Diego, CA, USA)
at concentrations of 5 or 10 mg/ml for 24 h and
Lin et al.
harvested for immunoblotting. Ten micrograms of total
lysate was separated by 10% SDS-PAGE and then
transferred onto a polyvinylidene fluoride membrane.
After transfer, the membrane was treated with blocking
solution (2.5% non-fat milk in 0.3% TBST (Tris buf-
fered saline with 0.3% Tween-20)) followed by incuba-
tion with mouse monoclonal anti-Sarm1, mouse anti-
VCP and rabbit anti-Iba1 Ab at 4 C overnight.
Proteins were detected with HRP-conjugated goat
anti-mouse IgG Ab and visualized with Western
Lightning Plus ECL (PerkinElmer, Waltham, MA,
USA), according to the protocol provided.
Quantitative and single-cell RT-PCR
For quantitative RT-PCR, total RNAs were extracted
from E14.5 whole brain and adult cortex by homogeniz-
ing the tissue with Trizol reagent according to the manu-
facturer’s instructions (Invitrogen, Carlsbad, CA, USA).
The RNA of BV2 and Neuro-2 A cells were collected by
directly lysing the cells cultured on 10-cm dishes with
Trizol. For the single-cell RT-PCR assay, the individual
somata of cultured mouse cortical neurons were visua-
lized with an inverted Axiovert 2000 microscope (Carl
Zeiss, Oberkochen, Germany) equipped with a
40  objective lens/NA 0.6 (LD Plan-Neofluar; Carl
Zeiss, Oberkochen, Germany) and collected using a
glass micropipette. The somata were immediately put
into a solubilization solution [0.5% NP-40 and 1 U/ml
RNaseOUT recombinant ribonuclease inhibitor
(Invitrogen, Carlsbad, CA, USA) in diethyl pyrocarbo-
nate (DEPC) water]. After freezing and thawing, total
RNAs were extracted by using Trizol. Extracted total
RNAs were further incubated with DNase I (New
England Biolab, Ipswich, MA, USA) for 30 min at
37 C to remove contaminating DNA. For brain tissues,
BV2 and Neuro-2 A, 5 mg of total RNA was applied for
cDNA synthesis using Transcriptor First Strand cDNA
Synthesis Kit (Roche Diagnostics, Mannheim, Germany)
with an anchored-oligo(dT)18 primer. For single-cell
experiments, all of the RNA underwent reverse transcrip-
tion. The PCR reaction was performed using the
Universal Probe Library (UPL; Roche Diagnostics)
and LightCycler480 system (Roche Diagnostics). The
primers and their paired probes, designed using the
Assay Design Center Web Service (http://qpcr.probefin-
der.com/roche3.html), were as follows: IL-6-F: 50 -GCT
ACC AAA CTG GAT ATA ATC AGG A-30 and IL-6-
R: 50 -CCA GGT AGC TAT GGT ACT CCA GAA-30 ,
with UPL Probe #6; Cyclophilin-F: 50 -TGC CCA GCA
GTT TAG TAC CC-30 and Cyclophilin-R: 50 -TGC TTC
CCT GTC TCC ACA GT-30 , with UPL Probe #64;
IFNb-F: CTGGCTTCCATCATGAACAA and IFNb-
R: AGAGGGCTGTGGTGGAGAA, with UPL
Probe#18; MECP2-F: CAGCTCCAACAGGAT
TCCAT and MECP2-R: TCTTCTGACTTTTCCTCC
CTGA, with UPL Probe#33; Dnmt1-F: CAAATAGA
163
TCCCCAAGATCCAG and Dnmt1-R: CGGAACTA
GGTGAAGTTTCAAAAA with UPL Probe#2;
Sarm1-F: GCTTGCTGGAGCAGATCCT and
Sarm1-R: TCACGCCTAGACCGATGC with UPL
Probe#102. IL1b-F: AGTTGACGGACCCCAAAAG
and IL1b-R: AGCTGGATGCTCTCATCAGG with
UPL Probe#38. IL10-F: CAGAGCCACATGCTCCT
AGA and IL10-R: TGTCCAGCTGGTCCTTTGTT
with UPL Probe#41. IL12b-F: GCTTCTTCATCAGG
GACATCA and IL12b-R: CTTGAGGGAGAAGTAG
GAATGG with UPL Probe#1. TNFa-F: TTGTCTTA
ATAACGCTGATTTGGT and TNFa-R: GGGAGC
AGAGGTTCAGTGAT with UPL Probe#64. CCL5-
F: TGCAGAGGACTCTGAGACAGC and
CCL5-R:GAGTGGTGTCCGAGCCATA with UPL
Probe#110. Iba1-F: GGATTTGACGGGAGGAAAA
G and Iba1-R: TGGGATCATCGAGGAATTG with
UPL Probe#3. Each sample was analyzed in triplicate
to avoid pipetting errors. PCR products after each cycle
were monitored by the fluorescent signal. The PCR
included the following steps: pre-incubation at 95 C
for 10 min, 45 cycles of 10 s denaturation at 95 C, 30 s
annealing at 60 C, 1 s extension at 72 C and a cooling
process at 40 C for 30 s. The relative expression level of
the target gene in each sample was calculated with the
2

Ct method using the crossing point (CP) value and
normalizing to the level of cyclophilin or L13. Sample
sizes (numbers of animals) are indicated in the figures.
Results
Sarm1 is mainly expressed in neurons in mouse brain
We used various markers and Sarm1 mAb8 to character-
ize the distribution of Sarm1 in different brain regions.
Sarm1 was widely expressed in different regions of the
cerebral cortex, including the retrosplenial agranular
cortex, motor cortex and sensory cortex (Figure 1A).
Sarm1 was also distributed in every layer of the cerebral
cortex (Figure 1A). Based on the distribution pattern and
cell morphology, we speculated that the Sarm1-positive
cells were neurons. To confirm this, we performed double
staining with Sarm1 and CaMKII, a marker for projec-
tion neurons, and GAD65/67, a marker for interneurons.
Basically, we found that all the CaMKII-positive cells
were Sarm1-positive (Figure 1A, upper panel). In
GAD65/67-positive interneurons, although the Sarm1
immunoreactivity was weaker, Sarm1 was also present
in every GAD65/67-positive interneuron (Figure 1A,
lower panel). We concluded that Sarm1 is widely
expressed in neurons localized in different regions and
different layers of the cerebral cortex.
In the hippocampus, Sarm1 was also expressed in all
regions, including the dentate gyrus, CA3, CA2 and
CA1 (Figure 1B). Similar to the cerebral cortex,
CaMKII- and GAD65/67-positive cells also expressed
Sarm1 (Figure 1B). In the amygdala, Sarm1 was
164 Innate Immunity 20(2)

Figure 1. Expression of Sarm1 in both projection and inhibitory neurons in mouse cerebral cortex and hippocampus. Adult mouse
brains were subjected to double immunostaining using Sarm1 and CaMKII or GAD65/67 Abs as indicated. (A) Cerebral cortex;
(B) hippocampus. CaMKII is a marker for projection neurons; GAD65/67 is a marker for inhibitory neurons. In (A) the right panels are
areas from the motor cortex region shown at higher magnification. All the CaMKII-positive neurons express Sarm1. Similarly, all
GAD65/67-expressing neurons also express Sarm1. In (B) Sarm1 was expressed in all cell layers CA1, CA2, CA3 and the dentate
gyrus (upper panel). The upper right panel is a high magnification image of the CA1 region. As shown in the lower panel, most of the
interneurons are located outside the cell layers. The GAD65/67-positive interneurons also express Sarm1 (lower right panel). Scale
bars: 200 mm in the left panels of (A) and (B); and 20 mm in the right panels of (A) and (B). DG: dentate gyrus; MC: motor cortex; RSA:
retrosplenial agranular cortex; SC: sensory cortex; SL: stratum lacunosum; SM: stratum moleculare; SO: stratum oriens; SR: stratum
radiatum.

present in both CaMKII-positive projection neurons calbindin (Figure 2C), and a weaker Sarm1 signal was
and GAD65/67-positive interneurons (Figure 2A, B). also found in interneurons in the granular cell layer and
In the cerebellum, Sarm1 was predominantly molecular layer (Figure 2C). In addition to the fore-
expressed in Purkinje neurons, which were labeled by brain and cerebellum, Sarm1 was also expressed in
Lin et al. 165

Figure 2. Distribution of Sarm1 in amygdala, cerebellum and midbrain. (A) Sarm1 expression cells were scattered in the amygdala.
Most of these cells are CaMKII-positive projection neurons. Co-localization of Sarm1 and CaMKII in the lateral and basal amygdala is
shown at high magnification in the right panels. (B) GAD65/67-positive inhibitory neurons in the amygdala also express Sarm1. High
magnifications indicating the co-localization are shown in the right panels. Scale bars: 200 mm in the left panels of (A) and (B); and
20 mm in the right panels of (A) and (B). (C) In the cerebellum, Sarm1 is mainly expressed in the Purkinje cells marked by calbindin.
High magnification of lobe VII in the right panel shows that all calbindin-positive Purkinje cells express Sarm1. Calbindin-negative
basket cells located in the Purkinje cell layer also express Sarm1, but at a relatively lower level. Scale bars: 500 mm in the left panels;
100 mm in the right panel. (D) Sarm1 was also expressed in the midbrain. The dopaminergic neurons in the ventral tegmental area
(VTA) were labeled by tyrosine hydroxylase (TH) Ab. High magnification of VTA is shown in the right panel. Scale bars: 200 mm in the
left panels; 20 mm in the right panel. LA: lateral amygdala; BA: basal lateral amygdala. GC: granular cell layer; M: molecular layer;
P: Purkinje cell layer.

the midbrain (Figure 2D). Double -staining with Ab
recognizing tyrosine hydroxylase, an enzyme
catalyzing the conversion of the amino acid L-tyrosine
to L-3,4-dihydroxyphenylalanine, further indicated
that Sarm1 was expressed in the dopaminergic
neurons of the ventral tegmental area of the midbrain
(Figure 2D).
In conclusion, the results of the fluorescent immu-
nostaining covering different areas of mouse brain
indicated that Sarm1 is widely expressed in brain and
is present in a variety of neurons.
Sarm1 is not detectable in microglial cells
The aforementioned results indicate that Sarm1 is mainly
expressed in neurons. However, a previous study showed
that a deficiency of Sarm1 results in the impairment of
microglial activation.6 To examine whether Sarm1 is
166 Innate Immunity 20(2)

(A) (B)

(C) (D)

(E) (F) Sarm 1 3000 lba1

BV2 Brain lysates 1.5
*** *
Relative of expression
Relative of expression

0.5 mg 1 mg *** *
Ctrl LPS LPS Tg WT 2000
1

72 Sarm 1
0.5 1000

97 VCP
0 0
N2A LPS N2A LPS
BV2 BV2

Figure 3. Lack of Sarm1 expression in microglia. (A) CD11b+ microglial cells are dispersed in the cerebral cortex; however, none of
CD11b + cells express Sarm1. (B) High magnification of (A). The arrow indicates an example showing typical microglia morphology,
which is negative for Sarm1 expression in its cell body, nucleus and cell processes. Arrowheads indicate examples of the CD11b-
positive signals located in Sarm1-positive neurons. (C) In the stratum radiatum of the hippocampus, which is mainly comprised of
neuronal fibers, CD11b+ microglias also showed a dispersed pattern and none of them expressed Sarm1 protein. (D) Z-Series analysis
of CD11-positive punta in a neuron. (i) Z-Stack of images in the x–y plane (focal section: 3/9, section thickness: 1mm) (ii) cross-section
of z-axis projection image from y planes. (iii) Cross-section of z-axis projection image from x planes. The series z stacks of (ii) and (iii)
indicates invasion of microglia into neighboring neuron. Scale bars: 50 mm in (A) and (C); 10 mm in (B) and (D). (E) The absence of
Sarm1 proteins in microglial BV2 cells. Sarm1 was not detectable in the cell lysate of BV-2 microglia with immunoblotting, even after
LPS treatment at the doses of 0.5 mg and 1 mg/ml for 24 h. (F) Sarm1 expression in BV2 cells examined by quantitative RT-PCR. The left
panel is a comparison of Sarm1 expression levels in murine neuroblastoma Neuro-2A (N2A) cell line, BV2, and BV2 activated by LPS
treatment. In the right panel, microglia marker, Iba1, was specifically expressed in BV2, but not in N2A. Sample size is n ¼ 3 for each
group. *P < 0.05; ***P < 0.001.

expressed in microglia, we performed double immunos- immunoreactivity in microglial cells in mouse cerebral
taining with Abs recognizing Sarm1 and CD11b or cortex (Figure 3A, B) or hippocampus (Figure 3C).
Iba1, the markers that label both resting and activated Therefore, it seems unlikely that Sarm1 directly regulates
microglia. Unexpectedly, we did not find Sarm1 the innate immune response mediated by microglial cells.
Lin et al.
We noticed that CD11b-positive puncta were occa-
sionally present in Sarm1 positive neurons (Figure 3B,
arrowheads). The signals were not a staining artifact
because Z-series analysis showed that the CD11b-posi-
tive signals in neurons were actually the processes
extending from neighboring microglia (Figure 3D).
Recently, a study in larval zebrafish showed that neu-
rons steer the processes of resting microglia and pro-
mote the contacts between microglia and neurons in a
neuronal activity-dependent manner.11 The tight con-
tact between microglial cells and neurons in zebrafish is
similar to our finding in mouse brains. This suggests
that the reciprocal modulation between resting micro-
glia and neurons is likely an evolutionally conserved
regulation.
To confirm the absence of Sarm1 in microglial cells,
we further examined Sarm1 expression in murine
microglial BV2 cells, which are commonly used as a
cell line representing primary microglial cells.12 By
immunoblotting, Sarm1 proteins were undetectable in
BV2 cell lysates (Figure 3E). In addition to immuno-
blotting, we further performed the more sensitive quan-
titative RT-PCR to examine the Sarm1 expression in
BV2 cells. Compared with the positive control, murine
neuroblastoma Neuro-2A cells, the Sarm1 RNA level
in BV2 cells was extremely low (Figure 3F). The weak
signals of Sarm1 in BV2 cells were not due to the poor
quality of RNA because the microglial marker Iba1
was easily detected in BV2 cells (Figure 3F). As LPS
stimulation induces Sarm1 expression in PBMC,5 we
then wondered whether Sarm1 would be up-regulated
after LPS treatment in BV2 cells. However, in contrast
to PBMC, LPS did not induce Sarm1 expression in BV2
cells at the protein or RNA level, although LPS
induced Iba1 expression (Figure 3E, F). The experi-
ments using BV2 cells supported the earlier double
immunostaining result and together strongly suggest
that Sarm1 is not noticeably expressed in microglial
cells.
In addition to the adult brain examined in the above
experiments, we also investigated the expression pattern
of Sarm1 at embryonic d 14.5 (E14.5). In the cortical
plate, Sarm1 co-existed with MAP2, a neuronal marker
(Figure 4A, B), indicating neuronal expression of
Sarm1 at the embryonic stage. In addition to the cor-
tical plate, Sarm1 was also expressed highly in the mar-
ginal zone and at a moderate level in the intermediate
zone (Figure 4A, B). We then examined whether Sarm1
is expressed in microglia at the embryonic stage. At
E14.5, microglial cells labeled with Iba1 Ab were spar-
sely distributed in mouse brains (Figure 4C). We still
did not find Sarm1 expression in microglial cells at
E14.5 (Figure 4D). Similar to adult brains, at the
embryonic stage some microglial cells tightly interacted
with Sarm1-positive neurons (Figure 4D, lower right
panel). Taken together, the experiments using adult
and embryonic mouse brains and BV2 cells indicate
167
an absence or extremely low level of expression of
Sarm1 in microglial cells.
Forebrain neurons express IL-6 and IFN-
Previous studies have demonstrated the expression of
inflammatory cytokine IL-6 and antiviral cytokine
IFN-b in sympathetic neurons and cerebellar neu-
rons.13–15 Here, we further investigated the expression
of IL-6 in forebrains. Using single-cell RT-PCR to ana-
lyze cultured cortical neurons we found that some
mouse cortical neurons express the IL-6 transcript,
even under normal culture conditions (Figure 5A, B).
In addition, double immunostaining was performed to
investigate the protein expression of IL-6 and IFN-b in
adult mouse brains. Both IL-6 and IFN-b Abs revealed
a small punctate pattern in mouse brain sections
(Figure 5C, D). The puncta were localized in MAP2-
positive cells, indicating neuronal expression of IL-6
and IFNb (Figure 5C, D, left panel). Although micro-
glial cells are recognized as professional immune cells in
brains, we barely detected IL-6 and IFN-b in resting
microglia (Figure 5C, D, right panel). Perhaps, expres-
sion of IL-6 and IFN-b in microglial cells are induced
by immune challenge.
Reduction of Sarm1 dysregulates cytokine expression
in brain
Induction of cytokine expression is the critical step of
TLR activation to induce innate immunity. Previous
studies13–15 and our results (Figure 5) clearly indicate
cytokine expression in neurons. Moreover, Sarm1 nega-
tively regulates TLR signaling.5 Therefore, although
Sarm1 was not expressed in microglia, we still won-
dered whether Sarm1 influences expression of the
inflammatory and antiviral cytokines in the brain.
Several transgenic mouse lines harboring an artificial
miRNA cassette against Sarm1 expression have been
generated in our laboratory.8 Because the miRNA cas-
sette was inserted into the 30 untranslational region of
the GFP transcript, the transgenic mice are easily iden-
tified by the fluorescent signals of GFP (Figure 6A).
In addition to the GFP signal, the knockdown effi-
ciency of Sarm1 was routinely confirmed using
immunoblotting with Sarm1 Ab or quantitative PCR
(Figure 6B–D). We examined the levels of a series of
cytokines in embryonic brain at E14.5 by quantitative
PCR. Among the cytokines examined, we found that
the RNA levels of both IL-6 and IFNb were increased
in Sarm1 knockdown transgenic mice (Figure 6C).
There was no noticeable difference in levels of the
other cytokines, including IL-1b, IL-10, IL-12b,
CCL5 and TNF-a, in Sarm1 knockdown transgenic
mice and their WT littermates (Figure 6C). In addition,
the expression levels of other genes, such as DNA
methyltransferase 1 (Dmnt1) and methyl-CpG-binding
168 Innate Immunity 20(2)

Figure 4. Expression pattern of Sarm1 in embryo brains at E14.5. Sarm1 was double stained with MAP2, which delineates the
regions of the cortical plate. In this caudal section, Sarm1 was expressed in the cortical plate, where it was co-localized with MAP2. In
addition, Sarm1 was also expressed in the marginal zone and the intermediate zone. (B) Higher magnification of (A). (C) Sarm1 was
double stained with microglia marker Iba1. The cell population of microglia at E14.5 was smaller than that in the adult brain shown in
Figure 3. (D) Z-Series analysis of Sarm1 expression in Iba1-positive microglia. (i) Z-Stack of images in the x–y plane (ii) cross-section of
z-axis projection image from the y planes. (iii) Cross-section of z-axis projection image from x planes. The series z stacks of (ii) and (iii)
suggest that microglia per se do not express Sarm1, but are often surrounded by cell extensions containing Sarm1 signals from
neighboring cell. Scale bars: 200 mm in (A) and (C); 10 mm in (B) and 5mm in (D). mz: marginal zone; cxp: cortical plate; sp: subplate;
iz: intermediate zone; svz: subventricular zone; vz: ventricular zone; cp: caudoputamen; LV: lateral ventricle.

protein 2 (MeCP2), were not changed by Sarm1 knock-
down (Figure 6C), supporting the specific effect of
Sarm1 on IL-6 and IFNb expression in embryonic
brains.
We also investigated whether Sarm1 influences cyto-
kine expression in adult mice. Unexpectedly, the cyto-
kine expression profile regulated by Sarm1 in adult
brains was different from that in embryonic brains.
The same as in embryonic brains, IL-10 expression
was unchanged in the WT and Sarm1 knockdown
adult brains; however, there was also no obvious differ-
ence in the level of IL-6 in the WT and Sarm1 knock-
down adult brains (Figure 6D). IFN-b and TNF-a were
down-regulated in Sarm1 knockdown brains (Figure
6D). The result showing TNF-a reduction in Sarm1
knockdown brains is similar to the data collected from
Sarm1-/- mice by Szretter et al.6 In contrast, IL-1b,
CCL5 and IL-12B were all significantly up-regulated in
Sarm1 knockdown adult brains (Figure 6D). The CCL5
up-regulation in adult brains echoes the results collected
by Carty et al. using HEK cells,5 which showed that
Sarm1 attenuated CCL5 expression. Thus, compared
with embryonic brains, Sarm1 has a broader effectt on
cytokine expression in adult brains.
In conclusion, although it is unclear why Sarm1
knockdown has a different effect on cytokine expression
in adult and embryonic brains, our data suggest that
Sarm1 expressed in neurons influences expression of
inflammatory and antiviral cytokines in the absence
of immune challenge.
Discussion
This report examined the distribution of Sarm1 in
mouse brain and the role of Sarm1 in the regulation
of cytokine expression. Sarm1 was widely expressed
in all the brain regions examined, including the cere-
bral cortex, hippocampus, amygdala, cerebellum and
midbrain. Further, the results of double immunos-
taining clearly showed that Sarm1 is expressed in
Lin et al. 169

Figure 5. Neurons express IL-6 and IFN-b. (A), (B) Single-cell RT-PCR was performed to investigate the expression of IL-6 mRNA in
cultured mouse cortical neurons at 4 day in vitro. (A) The representative images before and after collecting the soma of a neuron for
single-cell PCR. Scale bar: 20 mm. (B) Single-cell RT-PCR. The results of eight single neurons are shown. Two negative controls, medium
(M) and non-template control (NTC), are shown to indicate the specificity of PCR. (C) IL-6 was double stained with neuronal marker,
MAP2, or with microglia marker, CD11b. (D) IFN-b was double stained with MAP2 or with CD11b. Scale bars: 10 mm in (C) and (D).
SR: stratum radiatum.

neurons—both projection neurons and interneurons—
in mouse brain. In contrast, although microglial cells
are critical players in innate immune responses in the
brain, expression of Sarm1 in resting microglial cells
was undetectable. This neuron-specific distribution of
Sarm1 in mouse brain echoes our previous finding,
suggesting that Sarm1 plays a critical role in neuronal
morphogenesis. Although only hippocampal neurons
were investigated in our previous study, it is likely
that Sarm1 plays a general role in controlling the
morphogenesis of a variety of neurons.
Using RT-PCR to analyze the primary cultures, a
previous study showed Sarm1 expression in all of neu-
rons, astrocytes and microglial cells.6 However, the
detailed methods used to prepare the primary cultures
and the purities of the primary cultures were not
described in the previous work. It is possible that neur-
onal contamination contributed to the PCR signals of
Sarm1 in the previous study because Sarm1 is highly
expressed in neurons. In this study, we did not detect
noticeable RNA levels of Sarm1 in microglial BV2 cells
compared with Neuro-2A cells. Immunostaining did
not reveal significant signals in resting microglial cells
in mouse brains either. Although it might still be pos-
sible that resting microglial cells in mouse brains
express a very low level of Sarm1 proteins that is
under the detection threshold of our Sarm1 Ab, our
results favor the conclusion that Sarm1 proteins are
predominantly expressed in neurons and not in micro-
glial cells.
Although Sarm1 is undetectable in microglia, our
data showed that the expression levels of antiviral
and inflammatory cytokines are altered in Sarm1-
knockdown transgenic mice. Here, using single-cell
PCR and double immunostaining, we demonstrated
that murine neurons also express IL-6 and IFN-b. In
170 Innate Immunity 20(2)

(A) WT Tg WT Tg (B) WT Tg
#1 #2 #1 #2 #3 #4

72 Sarm 1

97 VCP
Visual light GFP
(C) E14.5
Sarm 1 IL-6 IFNb Dmnt1 MeCP2
1.6 **
* ns. ns.
Relative expression level

1.4 ***
1.2
1.0
0.8
0.6
0.4
n =26

n =26

n =33

n =29

n =35

n =31

n =14

n =14

n =14

n =14
0.2
0
WT Tg WT Tg WT Tg WT Tg WT Tg
IL-1b IL-10 IL-12b CCL5 TNFa
1.6 ns. ns.
ns. ns. ns.
Relative expression level

1.4
1.2
1.0
0.8
0.6
0.4
n =19

n =21

n =19

n =21

n =19

n =21

n =19

n =21

n =19

n =21
0.2
0
WT Tg WT Tg WT Tg WT Tg WT Tg
(D) Adult
Sarm 1 IL-6 IFNb IL-1b
1.4 2
ns. **
Relative expression level

1.2 *** *
1.0 1.5

0.8
1
0.6
0.4
0.5
n =10

n =11

n =14

n =11

n =14

n =11

n =14

0.2
n =7

0 0
WT Tg WT Tg WT Tg WT Tg
IL-10 IL-12b 2 CCL5 1.4 TNFa
1.4 5
* **
Relative expression level

ns. * 1.2
1.2
4 1.5
1.0 1.0
0.8 3 0.8
1
0.6 0.6
2
0.4 0.5 0.4
1
n=11

n=14

n=11

n=14

n=11

n=14

n=11

n=14

0.2 0.2
0 0 0 0
WT Tg WT Tg WT Tg WT Tg

Figure 6. Dysregulated cytokine expression in Sarm1 knockdown mice. (A) Gross morphology of Sarm1 knockdown brain. Left
panel, brain morphology of a Sarm1 knockdown (Tg) mouse and a wild type (WT) littermate. Right panel, GFP signal labeling the
Sarm1 knockdown brain. (B) The Sarm1 protein levels were lower in Sarm1 knockdown brains. Immunoblotting was performed using
Sarm1 and VCP Abs. (C), (D) Quantitative RT-PCR was carried out to quantify the mRNA levels of several inflammatory and antiviral
cytokines in Sarm1 knockdown (Tg) and WT littermates in (C) E14.5 mouse brains and in (D) adult mouse cerebral cortex. Before
collecting the brains, GFP signals were used to distinguish Sarm1 knockdown and WT mice. Some RNA samples were also measured
to confirm the efficacy of Sarm1 knockdown at the RNA level. Two unrelated genes, Dmnt1 and MeCP2, were used as negative
controls. Data represent means  SEM. The sample sizes (n) indicating the number of animals are shown in the columns. *P < 0.05;
**P < 0.01; ***P < 0.001.
Lin et al.
Sarm1 knockdown embryonic brains, IL-6 and IFN-b
were up-regulated, indicating that Sarm1 may nega-
tively regulate the innate immune responses of embry-
onic neurons. Sarm1 has been shown to attenuate the
activity of TRIF in the TLR3 and TLR4 pathways in
innate immune response. Expression of Sarm1 reduces
TRIF-induced NF-kB activation and cytokine produc-
tion in HEK cells.5 As TLR3 and NF-kB are also
expressed in neurons Sarm1 likely uses the same mech-
anism to down-regulate cytokine production in embry-
onic neurons, although direct evidence supporting the
involvement of NF-kB downstream of Sarm1 in neu-
rons is still lacking.
It is puzzling that the Sarm1-regulated cytokine pro-
files are different in adult and embryonic brains.
Perhaps, expression of cytokines in neurons is also con-
trolled by developmental stage or neuronal activity.
For instance, IL-1b, IL-12b and CCL5 may be up-regu-
lated by signal of neuronal activation. In mature neu-
rons, Sarm1 may antagonize the signaling pathway
from neuronal activation to cytokine production.
Therefore, removal of negative signal derived from
Sarm1 increases the expression of these three cytokines
in adult brain. Another explanation is that the cytokine
profile at E14.5 is mainly determined by neurons them-
selves. In the adult, a significant population of astro-
cytes and microglial cells in the brain and the peripheral
immune responses may interact with Sarm1-expressing
neurons and thus alter the cytokine expression profiles
in brains. This possibility may be applied to both up-
regulated cytokines (IL-1b, IL-12 and CCL5) and
down-regulated cytokines (IFN-b and TNF-a) in
Sarm1 knockdown brain. However, these two possibi-
lities are not mutually exclusive of each other. More
investigation needs to be conducted to address the
detail regulation.
Interestingly, Sarm1 deficiency was previously
reported to result in impairment in microglial activa-
tion.6 However, because our current data indicate
that microglial cells do not obviously express Sarm1,
the impairment of microglial activation in Sarm1-defi-
cient mice is unlikely to be a cell-autonomous effect.
Rather, it is more likely that the neuronal innate
immune response influences the activation of the micro-
glia. Actually, a recent study and our data demonstrate
the tight contact between microglial cells and neu-
rons.11 It would not be surprising to find that neurons
can also influence microglial activation in mouse
brains.
In conclusion, our study reveals the role of TIR-
domain-containing adaptor Sarm1 in innate immunity
in the brain. We demonstrated that Sarm1 is widely
expressed in different types of neurons in mouse
brains, but notably not in microglial cells; and also
that knockdown of Sarm1 alters the expression levels
of antiviral and inflammatory cytokines in brains. As
Sarm1 is present in neurons, the regulation of cytokines
171
is expected to occur in neurons, suggesting that Sarm1
regulates the innate immune responses of neurons.
Further, as IL-6 is also known to negatively regulate
dendrite growth, in addition to the JNK-microtubule
pathway, as reported in our previous study,8 it is likely
that IL-6 may also act downstream of Sarm1 in embry-
onic brains to regulate dendrite growth, although fur-
ther investigation is need to address this speculation.
Because prenatal infection is frequently associated
with schizophrenia and autism, and because Sarm1
protein levels are lower in the cortex of patients with
autism,16,17 our data also suggest a possible role
for Sarm1 in inflammation-related neuropsychiatric
disorders.
Funding
This work was supported by the National Science Council
(grant numbers NSC 98-2311-B-001-012-MY3, NSC-100-
2321-B-001-022, NSC-101-2321-B-001-010).
Conflict of interest
The authors do not have any potential conflicts of interest to
declare.
Acknowledgements
We thank Ms Miranda Loney for English language editing of
the manuscript.
References
1. Takeda K and Akira S. TLR signaling pathways. Semin Immunol
2004; 16: 3–9.
2. Kawai T and Akira S. TLR signaling. Cell Death Differ 2006; 13:
816–825.
3. Kumar H, Kawai T and Akira S. Toll-like receptors and
innate immunity. Biochem Biophys Res Commun 2009; 388:
621–625.
4. O’Neill LA and Bowie AG. The family of five: TIR-domain-
containing adaptors in Toll-like receptor signalling. Nat Rev
Immunol 2007; 7: 353–564.
5. Carty M, Goodbody R, Schroder M, et al. The human adaptor
SARM negatively regulates adaptor protein TRIF-dependent
Toll-like receptor signaling. Nat Immunol 2006; 7: 1074–1081.
6. Szretter KJ, Samuel MA, Gilfillan S, et al. The immune
adaptor molecule SARM modulates tumor necrosis factor
alpha production and microglia activation in the brainstem and
restricts West Nile Virus pathogenesis. J Virol 2009; 83:
9329–9338.
7. Kim Y, Zhou P, Qian L, et al. MyD88-5 links mitochondria,
microtubules, and JNK3 in neurons and regulates neuronal sur-
vival. J Exp Med 2007; 204: 2063–2074.
8. Chen CY, Lin CW, Chang CY, et al. Sarm1, a negative regulator
of innate immunity, interacts with syndecan-2 and regulates neur-
onal morphology. J Cell Biol 2011; 193: 769–784.
9. Chuang CF and Bargmann CI. A Toll-interleukin 1 repeat pro-
tein at the synapse specifies asymmetric odorant receptor expres-
sion via ASK1 MAPKKK signaling. Genes Dev 2005; 19:
270–281.
10. Osterloh JM, Yang J, Rooney TM, et al. dSarm/Sarm1 is
required for activation of an injury-induced axon death pathway.
Science 2012; 337: 481–484.
172
11. Li Y, Du XF, Liu CS, et al. Reciprocal regulation between rest-
ing microglial dynamics and neuronal activity in vivo. Dev Cell
2012; 23: 1189–1202.
12. Henn A, Lund S, Hedtjarn M, et al. The suitability of BV2 cells
as alternative model system for primary microglia cultures or for
animal experiments examining brain inflammation. ALTEX
2009; 26: 83–94.
13. Ward LA and Massa PT. Neuron-specific regulation of major
histocompatibility complex class I, interferon-beta, and anti-viral
state genes. J Neuroimmunol 1995; 58: 145–155.
14. Paul S, Ricour C, Sommereyns C, et al. Type I interferon
response in the central nervous system. Biochimie 2007; 89:
770–778.
Innate Immunity 20(2)
15. Marz P, Cheng JG, Gadient RA, et al. Sympathetic neurons can
produce and respond to interleukin 6. Proc Natl Acad Sci U S A
1998; 95: 3251–3256.
16. Pardo C, Azhagiri A, Lawler C, et al. Expression profiling of
TLR signaling pathway genes in brain tissue from patients with
autism. In: International Meeting for Autism Research, Chicago,
IL, USA, 7–9 May 2009, abstract no. 131.05. West Hartford, CT,
USA: The International Society for Autism Research.
17. Azhagiri A, Lawler C, Zea-Vera A, et al. TLR signaling path-
ways in brain tissue from patients with autism. In: Annual meet-
ing of the Society for Neuroscience, Chicago, IL, USA, 12–16
November 2009, poster no. 437.13/M31. Washington, DC,
USA: The Society for Neuroscience.