Content uploaded by Stephen M Dombrowski
Author content
All content in this area was uploaded by Stephen M Dombrowski
Content may be subject to copyright.
Control of microglial neurotoxicity by the
fractalkine receptor
Astrid E Cardona
1
, Erik P Pioro
1
, Margaret E Sasse
1
, Volodymyr Kostenko
1
, Sandra M Cardona
1
,
Ineke M Dijkstra
1
, DeRen Huang
1
, Grahame Kidd
1
, Stephen Dombrowski
2
, RanJan Dutta
1
, Jar-Chi Lee
3
,
Donald N Cook
4
, Steffen Jung
5,6
, Sergio A Lira
7
, Dan R Littman
6
& Richard M Ransohoff
1
Microglia, the resident inflammatory cells of the CNS, are the only CNS cells that express the fractalkine receptor (CX3CR1).
Using three different in vivo models, we show that CX3CR1 deficiency dysregulates microglial responses, resulting in
neurotoxicity. Following peripheral lipopolysaccharide injections, Cx3cr1
–/–
mice showed cell-autonomous microglial neurotoxicity.
In a toxic model of Parkinson disease and a transgenic model of amyotrophic lateral sclerosis, Cx3cr1
–/–
mice showed more
extensive neuronal cell loss than Cx3cr1
1
littermate controls. Augmenting CX3CR1 signaling may protect against microglial
neurotoxicity, whereas CNS penetration by pharmaceutical CX3CR1 antagonists could increase neuronal vulnerability.
CX3CR1 is expressed by monocytes, dendritic cells (DCs), and subsets
of T cells and natural killer cells in the circulation and by microglia in
the central nervous system (CNS)
1
. Fractalkine (CX3CL1), the exclu-
sive ligand for CX3CR1, is synthesized as a transmembrane glycopro-
tein, from which a soluble chemokine can be proteolytically released
2
.
CX3CR1/CX3CL1 signaling exerts distinct functions in different tissue
compartments: CX3CR1 deficiency impairs the morphogenesis of
myeloid DCs, which occupy the lamina propria of the small intestine
3
;
further, CX3CR1 contributes to the migration of circulating monocytes
to noninflamed tissues, where they differentiate into macrophages and
DCs. However, peripheral immune and inflammatory reactions are
mostly unaltered in Cx3cr1
–/–
or Cx3cl1
–/–
mice
4,5
. CX3CR1 is respon-
sible for recruiting natural killer cells to cardiac allografts
6
and to the
inflamed CNS of mice with experimental autoimmune encephalomye-
litis
7
. In vitro, CX3CL1 promotes neuronal survival and inhibits
microglial apoptosis
8
, but the roles of CX3CL1/CX3CR1 in the intact
CNS are enigmatic.
Microglia, which have characteristics of immature myeloid cells,
sample the extracellular space of the healthy CNS through continuous
extension, retraction and remodeling of cellular processes
9,10
.In
response to injury, microglial cells undergo rapid morphological and
functional activation and acquire properties of mature myeloid cells,
including antigen presentation, reactive species generation, matrix
metalloproteinase (MMP) expression and phagocytosis, as well as
cytokine and growth factor secretion
11
. Microglial production of
reactive species, MMPs and inflammatory cytokines have been impli-
cated in neurotoxicity in vitro. However, control of microglial neuro-
toxicity in vivo remains poorly understood. Altered microglial function
can cause CNS disease in humans: homozygous deficiency of either
type 2 triggering receptor expressed on myeloid cells (TREM2)orits
intracellular adaptor TYROBP causes adult onset dementing leuko-
encephalopathy, which is recapitulated in Tyrobp
–/–
mice
12
. Knock-
down of TREM2 in mouse microglia impairs their phagocytic activity
and enhances inflammatory gene expression, suggesting a mechanism
for the human disorder
13
.
To investigate the role of CX3CR1 in the intact CNS, we used mice in
which the Cx3cr1 gene was replaced with a cDNA encoding green
fluorescent protein (GFP), such that heterozygous Cx3cr1
+/GFP
(Cx3cr1
+/–
) mice expressed the GFP reporter in cells that retained
receptor function, whereas Cx3cr1
GFP/GFP
(Cx3cr1
–/–
) cells were labeled
and also lacked CX3CR1 (ref. 5). The CX3CR1-GFP reporter identifies
microglia in vivo
5,9,10
and permits their isolation and purification
ex vivo. Our results indicated that in the absence of CX3CR1/
CX3CL1 signaling, microglia had altered responses, both to inflam-
matory and neurotoxic stimuli. In particular, CX3CR1 deficiency was
associated with neuronal cell death after systemic lipopolysaccharide
(LPS) challenge. Cx3cr1
–/–
mice demonstrated more neuronal cell loss
in a toxin-induced model of Parkinson disease and in a model of
genetic motor neuron disease. The results identified CX3CL1 as the
first soluble factor that regulates microglial neurotoxicity.
RESULTS
Neuronal damage in Cx3cr1
–/–
mice after systemic LPS
We first examined the GFP
+
population in the CNS of Cx3cr1
GFP
mice
and found that green fluorescent cells overlapped precisely with brain
cells expressing microglial markers such as ionized calcium-binding
Received 6 March; accepted 5 May; published online 28 May 2006; corrected online 11 June 2006; doi:10.1038/nn1715
1
Neuroinflammation Research Center and Department of Neurosciences, Lerner Research Institute,
2
Department of Neurosurgery, and
3
Department of Quantitative Health
Sciences, Cleveland Clinic, Cleveland, Ohio 44195, USA.
4
National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina 27709, USA.
5
Department of Immunology, Weizman Institute of Science, Rehovot 76100, Israel.
6
Howard Hughes Medical Institute and Skirball Institute for Biomolecular Medicine, New
York University School of Medicine, New York, New York 10016, USA.
7
Immunobiology Center, Mount Sinai School of Medicine, New York, New York 10029, USA.
Correspondence should be addressed to R.M.R. (ransohr@ccf.org).
NATURE NEUROSCIENCE VOLUM E 9
[
NUMBER 7
[
JULY 2006 917
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/natureneuroscience
adapter molecule-1 (IBA-1)
14
(Fig. 1a–c)andGriffonia simplicifolia
isolectin B4 (data not shown). This indicated that all CNS microglia
were GFP
+
in Cx3cr1
+/GFP
and Cx3cr1
GFP/GFP
mice, designated below as
Cx3cr1
+/–
and Cx3cr1
–/–
, respectively. According to a previous report
15
,
various CNS cells express CX3CR1 immunoreactivity. We found no
cells that coexpressed GFP and neuron-specific nuclear protein (NeuN)
(Fig. 1d–f), the proteoglycan NG2 (Fig. 1g–i) or glial fibrillary acidic
protein (GFAP) (Fig. 1j–l), indicating that neither CNS neurons, NG2
+
glia nor astrocytes expressed the CX3CR1 transcription unit in vivo by
this analysis.
To examine microglial activation in the absence of CX3CR1, we
induced a systemic inflammatory reaction by means of intraperitoneal
(i.p.) injections of LPS (ref. 16). Cx3cr1
+/–
mice showed moderate
morphological transformation of microglia (Fig. 2a), whereas
Cx3cr1
–/–
mice showed intense and widespread microglial activation
(Fig. 2b). Cx3cr1
–/–
but not Cx3cr1
+/–
mice showed numerous annexin
V–positive cells with neuronal morphology, throughout the cortex
and hippocampus (Fig. 2c–f). We quantified the annexin V–positive
cells in the dentate gyrus and found that they were significantly
more numerous in Cx3cr1
–/–
mice (Fig. 2g; P ¼ 0.0012) that received
LPS than in controls (Fig. 2g). We obtained compatible results from
TdT-mediated dUTP nick end labeling (TUNEL) analyses (data not
shown). Based on these findings, we formed the hypothesis that
systemic inflammation, induced by LPS, activated the neurotoxic
potential of microglia in Cx3cr1
–/–
mice. We asked whether direct
local effects of LPS in Cx3cr1
–/–
mice might differ from those in
Cx3cr1
+/–
mice. RNase protection assay (RPA) of peritoneal macro-
phages from LPS-challenged Cx3cr1
+/–
and Cx3cr1
–/–
mice demon-
strated equivalent production of representative inflammatory cytokines
(Fig. 2h). This finding suggested that direct local effects of LPS were
equivalent in heterozygous and knockout mice and that the neurotoxic
consequences of peripheral LPS injections in Cx3cr1
–/–
mice arose
within the CNS.
Cell-autonomous neurotoxicity of Cx3cr1
–/–
microglia
To address whether microglial neurotoxicity induced by systemic
inflammation was cell autonomous, we developed an adoptive transfer
protocol using purified populations of GFP-labeled activated micro-
glial cells (Fig. 3a). These were isolated from LPS-injected Cx3cr1
–/–
or
Cx3cr1
+/–
mice (n ¼ 6–8 donor mice per inoculum), washed and
microinjected into the frontal cortex of wild-type recipient mice
(Fig. 3b). At 36 h after transfer, recipient brains were analyzed for
the presence of microglia in serial free-floating
horizontal sections through the needle track,
and 1–2 mm ventral to the end of the needle
mark. The injection sites were defined as the
last section containing a visible needle track
and the next serial section without the needle
artifact. Examination of injection sites
(Fig. 3c–f) showed divergent phenotypes for
Cx3cr1
+/–
(Fig. 3c) as compared with
Cx3cr1
–/–
microglial cells (Fig. 3d). Microglia
from Cx3cr1
+/–
mice, as previously described
for myeloid cells
17
, were not observed at
injection sites (Fig. 3c) and migrated widely
throughout the CNS parenchyma, preferen-
tially in white matter tracts (Fig. 3g). In
contrast, microglial cells from Cx3cr1
–/–
mice
remained localized at the injection site
(Fig. 3d) and migrating cells (Fig. 3g–j)
were not detected in wild-type recipients of
activated Cx3cr1
–/–
microglia (Fig. 3h). Furthermore, we analyzed
neuronal cell death (Fig. 3k–n) and readily identified TUNEL/NeuN
double-positive neurons near the injection site (60–200 mm deep) in
wild-type recipients of Cx3cr1
–/–
microglia (Fig. 3l) but not in
recipients of Cx3cr1
+/–
microglia (Fig. 3k). The CNS tissues of wild-
type recipients of Cx3cr1
–/–
microglia contained significantly fewer
migrating cells (Fig. 3o; P ¼ 0.02 compared with Cx3cr1
+/–
microglia).
Additionally, significantly more TUNEL-positive neurons were
detected in the CNS of wild-type recipients of Cx3cr1
–/–
microglial
cells (Fig. 3p; P ¼ 0.02). The microglial phenotype in these adoptive
transfer experiments was dependent both on genotype and activation
status, as unactivated microglia from Cx3cr1
–/–
control mice (that
received i.p. saline injections) distributed throughout the white matter
tracts of recipient wild-type mice (data not shown). These findings
demonstrated that activated Cx3cr1
–/–
microglia are neurotoxic in the
CNS of wild-type mice and provided an assay for evaluating putative
mediators of this toxic effect.
Our initial results suggested that CX3CR1 modulated the response of
CNS microglia to systemic inflammation. This hypothesis was further
tested by ex vivo RPA, using highly enriched preparations of microglia
(Fig. 3a)fromCx3cr1
+/–
or Cx3cr1
–/–
mice, purified from CNS tissues
after induction of systemic inflammation by i.p. LPS injections. These
experiments demonstrated increased expression of interleukin (IL)-1b,
but not tumor necrosis factor (TNF)-a, IL-6 or lymphotoxin, by
microglia from LPS-injected Cx3cr1
–/–
mice (Fig. 3q, P ¼ 0.02).
Microglia from LPS-injected Cx3cr1
–/–
mice also demonstrated
elevated production of IL-1b–associated signaling intermediates such
as Myd88 (data not shown). These findings were noteworthy, because
IL-1 action through IL-1 receptor type 1 (IL-1R1) has been consistently
associated with neurodegeneration in vivo
18
. We tested whether IL-1
was a mediator of neurotoxicity caused by Cx3cr1
–/–
microglia, by using
IL-1 receptor antagonist (IL-1RA), which blocks the assembly of the IL-
1R1/IL-1 receptor accessory protein (IL-1RAcP) signaling complex.
When IL-1RA was included in the adoptive transfer inoculum along
with activated microglia from LPS-injected Cx3cr1
–/–
mice, we
observed a reversal of the knockout phenotype (compare Fig. 3d,h,l
with Fig. 3e,i,m). In particular, blockade of IL-1 signaling restored
migration of Cx3cr1
–/–
microglia throughout the CNS of wild-type
recipients (Fig. 3o; P ¼ 0.03 comparing Cx3cr1
–/–
microglia with and
without IL-1RA), and the number of TUNEL-positive neurons was
significantly reduced (Fig. 3p; P ¼ 0.02 comparing wild-type recipients
of Cx3cr1
–/–
cells with and without IL-1RA). Coinjection of Cx3cr1
–/–
ab de gh j
cf i l
k
Figure 1 Microglial cells comprise the CX3CR1/GFP
+
population. (a–l) Brain sections from Cx3cr1
+/–
mice immunostained with antibodies to IBA-1 (a–c), NeuN (d–f), NG2 (g–i) and GFAP (j–l), showing
that the GFP
+
population overlaps precisely with IBA-1
+
microglial cells (c, merged image). Cells
expressing the CX3CR1-GFP reporter (d,g,j) did not colocalize to markers of neurons (e:NeuN
+
cells,
red), NG2
+
glial cells (h:NG2
+
, red) or astrocytes (k:GFAP
+
, red) as shown in merged images (f,i,l).
Scale bar, 25 mm.
918 VOLUME 9
[
NUMBER 7
[
JULY 2006 NATURE NEUROSCIENCE
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/natureneuroscience
microglia with inert carrier protein did not enhance microglial migra-
tion or reduce microglial neurotoxicity (Fig. 3o,p).
It remained uncertain whether neurotoxic properties of IL-1 in these
experiments proceeded through autocrine effects of IL-1 on Cx3cr1
–/–
microglia or paracrine effects on recipient microglia, astrocytes and
neurons, as IL-1RA blocked signaling to all cells within the injection
site. Adoptive transfer of activated Cx3cr1
–/–
cells into the CNS of
Il1r1
–/–
recipients demonstrated that defective migration by Cx3cr1
–/–
microglial cells could be partially rescued by abrogation of the host
response to IL-1. In Il1r1
–/–
recipients, transferred Cx3cr1
–/–
microglial
cells remained partially localized at the injection site (Fig. 3f)but
migrating donor-derived GFP
+
cells were readily detected (Fig. 3j)and
were significantly more numerous than in the wild-type recipients of
Cx3cr1
–/–
microglia (Fig. 3o; P ¼ 0.04 comparing Il1r1
–/–
and wild-type
recipients of Cx3cr1
–/–
cells). Notably, TUNEL-positive cells were
virtually absent from the CNS of Il1r1
–/–
recipients of Cx3cr1
–/–
microglia (Fig. 3n,p; P ¼ 0.02 comparing Cx3cr1
–/–
microglia trans-
ferred to wild-type and Il1r1
–/–
mice).
To exclude the possibility that these results emerged from variability
in the placement of the microinjected cells, we performed stereotactic
adoptive transfer experiments. These experiments required that
8-week-old recipients be used, in order to accommodate the stereo-
tactic apparatus (Fig. 4a). We observed that Cx3cr1
–/–
microglia
(Fig. 4b) but not Cx3cr1
+/–
microglia (Fig. 4c)remainedatthe
injection sites. Cx3cr1
+/–
microglia but not
Cx3cr1
–/–
microglia were found migrating
throughout the white matter of recipients
(Fig. 4d,e). Following transfer of Cx3cr1
–/–
microglia but not Cx3cr1
+/–
microglia, we
detected annexin V–positive neurons (com-
pare Figs. 4f and g).
Together, these results established that
CX3CR1 governs critical components of the
microglial response to systemic inflammation
in vivo, as suggested by previous in vitro
studies
8,19
, and that IL-1 is a mediator
of microglial neurotoxicity induced by
systemic inflammation.
SNpc neurons in Cx3cr1
–/–
or Cx3cl1
–/–
mice after MPTP
Microglial neurotoxicity has also been pro-
posed to augment the severity of neurodegen-
erative processes including Parkinson disease.
We addressed a potential role for CX3CR1 in
the microglial reaction to neurodegeneration
by evaluating the responses of gender-matched
littermate Cx3cr1
–/–
, Cx3cr1
+/–
and Cx3cr1
+/+
mice to the administration of the dopaminergic
neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetra-
hydropyridine (MPTP), which recapitulates
selected features of Parkinson disease.
CX3CR1 deficiency aggravated the pathological
outcomes of MPTP injection in mice. We
performed a stereological analysis of the sub-
stantia nigra pars compacta (SNpc) of saline-
(Fig. 5a–c) and MPTP-injected (Fig. 5d–i)
Cx3cr1
+/+
and Cx3cr1
–/–
littermate mice 7 d
after challenge (Tab l e 1). We found a signifi-
cant MPTP-induced loss of tyrosine hydroxy-
lase–immunoreactive (TH-IR) cells and a
similar loss of Nissl-stained neurons in the SNpc of Cx3cr1
+/+
mice,
with no change in the percent of Nissl-stained cells that were TH-IR.
The loss of both TH-IR cells and Nissl-stained cells was significantly
(P o 0.001) worse in Cx3cr1
–/–
mice (Tabl e 1 ). We also evaluated the
effect of CX3CL1 ligand deficiency in this model: the results were nearly
identical to those observed in receptor-deficient Cx3cr1
–/–
mice
(Table 1 ). Saline-injected mice showed no genotype-related differences
(P ¼ 0.86 for TH-IR cells; P ¼ 0.47 for Nissl-stained cells). Transport
and metabolism of MPTP were not altered in Cx3cr1
–/–
mice, as
concentrations of MPTP in the SN were equivalent in heterozygous
and knockout mice, and striatal concentrations of 1-methyl-4-phenyl-
pyridinium (MPP+) were not significantly different, either 90 min or
3 h after injection (data not shown). Inspection of coronal sections of
the SNpc of Cx3cr1
+/–
and Cx3cr1
–/–
mice (Fig. 5j,k) suggested a greater
degree of morphological transformation of knockout microglia, as
compared to those in heterozygous mice, in response to MPTP
injections. Taken together, these results demonstrated that the response
to MPTP in mice lacking microglial CX3CR1 signaling caused greater
loss of TH-IR neurons.
Worsened disease in Cx3cr1
–/–
SOD1
G93A
transgenic mice
Microglial cells have also been implicated in the loss of motor neurons
during the fatal neurodegenerative disorder amyotrophic lateral sclero-
sis (ALS). Transgenic mice that overexpress a mutant form of the
50 µm
Cx3cr1
+/–
Cx3cr1
–/–
50 µm
25 µm
50 µm
50 µm
25 µm
a
c
e
b
d
f
Cx3cr1
–/–
Cx3cr1
+/–
Annexin V–positive cells (mm
2
)
Ratio cytokine/GAPDH
gh
Saline
550
450
250
200
150
100
50
0
300
LPS Saline
0.0012
LPS
IL-6
0
0.15
0.35
LT
IL-1α IL-1β TNF-α TNF-β
Figure 2 Cx3cr1
–/–
mice show increased microglial activation and enhanced neuronal damage after
systemic inflammation. (a,b) IBA-1 immunohistochemistry revealed highly ramified microglia in the
hippocampus of LPS-treated Cx3cr1
+/–
mice (a), compared to increased microglial activation in
Cx3cr1
–/–
mice (b) whose cells have shorter and thicker processes and bigger cell bodies. (c–f)AnnexinV
immunostaining showed numerous annexin V–immunoreactive cells in Cx3cr1
–/–
mice (d,f:brown
staining, arrows) with the nuclear morphology of neurons, but not in Cx3cr1
+/–
mice. (g) Quantitation of
annexin V–positive cells showed a significant (P ¼ 0.0012) increase of annexin V–positive cells in the
dentate gyrus of Cx3cr1
–/–
mice after LPS administration. (h) RNase protection analysis of RNA from
peritoneal macrophages showed low cytokine levels in saline-injected Cx3cr1
+/–
mice (gray), and equal
responses in LPS-injected Cx3cr1
+/–
(white) and Cx3cr1
–/–
mice (black). Error bars represent s.d.
NATURE NEUROSCIENCE VOLUM E 9
[
NUMBER 7
[
JULY 2006 919
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/natureneuroscience
human SOD1 gene (SOD1
G93A
) encoding copper/zinc superoxide
dismutase show age-dependent degeneration of motor neurons accom-
panied by limb weakness and provide a useful model of ALS (ref. 20).
We established a colony of mice that were SOD1
G93A
/Cx3cr1
+/+
,
SOD1
G93A
/Cx3cr1
+/–
, SOD1
G93A
/Cx3cr1
–/–
or nontrangenic/Cx3cr1
–/–
,
and performed serial neurobehavioral and survival assessment of
SOD1
G93A
/Cx3cr1
+/–
and SOD1
G93A
/Cx3cr1
–/–
mice, with histological
analysis of a cohort of surviving mice from all four genotypes at 133 d.
Microglial reaction, as judged by IBA-1 immunohistochemistry,
showed morphological transformation in the lumbar spinal cords of
all transgenic mice (Fig. 6a–d), without significant genotype-related
differences in the tissue area showing IBA-1 immunoreactivity (46% in
SOD1
G93A
/Cx3cr1
–/–
compared to 39% in SOD1
G93A
/Cx3cr1
+/+
mice;
P ¼ 0.12). SOD1
G93A
/Cx3cr1
–/–
mice showed decreased neuronal cell
density at this time point (Figs. 6e–i) when compared with either
littermate SOD1
G93A
/Cx3cr1
+/+
(P ¼ 0.02) or SOD1
G93A
/Cx3cr1
+/–
(P ¼ 0.03) mice (Fig. 6i). Comparisons of neuronal cell density
between transgenic heterozygous and transgenic wild-type mice did
not show statistically significant differences (Fig. 6i; P ¼ 0.09 compar-
ing SOD1
G93A
/Cx3cr1
+/–
and SOD1
G93A
/Cx3cr1
+/+
mice).
Behavior and survival analyses also revealed differences between
SOD1
G93A
/Cx3cr1
+/–
and SOD1
G93A
/Cx3cr1
–/–
mice. Hindlimb grip
strength showed evident decline after 7–9 weeks in all SOD1
G93A
mice,ascomparedtonontransgenic/Cx3cr1
–/–
littermate control mice
*
*
*
*
*
*
180
0
10
0
10
1
10
2
10
3
GFP
Cx3cr1
+/–
Cx3cr1
+
/
–
Cx3cr1
–
/
–
Cx3cr1
–
/
–
Wild-type Cx3cr1
–/–
Cx3cr1
–/–
+ IL-1RA
Cx3cr1
–/–
IL-1RA
Cx3cr1
–/–
carr
ier
Il1r1
–/–
Cx3cr1
–/–
Wild-type
Wild-type
92.10%
g–j
100 µm
100 µm
*
*
*
Counts
10
4
Anterior
Posterior
(IS)
Injection site
Il1r1
–/–
Donor
Recipient
Wild-type
Cx3cr1
+
/
–
Cx3cr1
–/–
Cx3cr1
–/–
Cx3cr1
–/
–
IL-1RA
Cx3cr1
–/–
carr
ier
Il1r1
–/–
Donor
Recipient
Wild-type
IL-6
0.02
0.02
0.02
0.03
0.04
Ratio cytokine/GAPDH
Apoptotic neurons
Migrating microglia
2.0
60
20
15
10
5
0
40
20
0
1.0
0
IL-1α IL-1β TNF-α TNF-βLT
a
cdef
ghi j
k
o
p
q
lmn
b
Figure 3 Adoptive transfer studies of microglia by intracranial microinjection. (a) Flow cytometry shows microglial preparations containing 490% GFP
+
cells
(purple overlay, GFP
–
Cx3cr1
+/+
microglia), used for RNA isolation or adoptive transfer. (b) Recipient brains (shown in axial section) were stained using NeuN
antibodies (blue) and TUNEL (red). (c–j) Merged GFP-NeuN images show injection sites within dotted ovals (c–f) and points of migration (g–j). (k–n) Apoptotic
neurons in merged GFP-NeuN-TUNEL images. Cx3cr1
+/–
microglia (c) were not detected at the injection site, but were distributed along white matter tracts
(g, arrows), whereas Cx3cr1
–/–
microglia (d) remained localized at injection site, did not distribute throughout the brain (h) and were associated with apoptotic
neurons 60–200 mm from the injection site (i, arrows). When LPS-activated Cx3cr1
–/–
microglia were transferred with IL-1RA into wild-type recipients (e,i,m),
few GFP
+
microglia were found at the injection site (e) and migrating cells were seen (i, arrowhead) without neuronal apoptosis near the injection site (m). In
Il1r1
–/–
recipients (f,j,n), many Cx3cr1
–/–
microglia persisted at the injection site (f), but were also detected at points of migration (j, arrowhead) without
associated neuronal cell death (n) near the injection site. (o) Number of migrating microglia, counted at sites shown in b (asterisks). (p) Apoptotic neurons
were counted in serial sections at a depth of 60–200 mm from the injection site, using transfer preparations and recipients as indicated. Scale bars, 25 mm.
(q) Cytokine profiling by RPA revealed an increased expression of IL-1b by activated Cx3cr1
–/–
microglia. Error bars represent s.d.
920 VOLUME 9
[
NUMBER 7
[
JULY 2006 NATURE NEUROSCIENCE
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/natureneuroscience
which became stronger as they attained adulthood (data not shown).
Between weeks 15 through 20, when weakness became progressive,
random coefficient modeling disclosed a strong age group interaction
(P o 0.01) indicating that SOD1
G93A
/Cx3cr1
–/–
mice had a faster
decline in hindlimb grip strength than SOD1
G93A
/Cx3cr1
+/–
mice.
With further analysis of the data, we found that the male Cx3cr1
–/–
group showed a much steeper decline (slope ¼ –19.92 ± 2.1 (s.e.m.),
n ¼ 5) than the other groups (male Cx3cr1
+/–
: –6.01 ± 2.5, n ¼ 5; female
Cx3cr1
+/–
: –6.72 ± 3.5, n ¼ 6; and female Cx3cr1
–/–
: –8.41 ± 2.4, n ¼ 7),
accounting for the group differences (Fig. 6j). Male SOD1
G93A
/
Cx3cr1
–/–
mice also showed faster loss of body weight (P o 0.01) and
forelimb grip strength (P ¼ 0.02) than male SOD1
G93A
/Cx3cr1
+/–
mice
(data not shown). Using a defined indicator of terminal state and
Kaplan-Meier analysis (Fig. 6k), survival was significantly (log rank P ¼
0.003) reduced in male SOD1
G93A
/Cx3cr1
–/–
mice as compared with
male SOD1
G93A
/Cx3cr1
+/–
miceandwithfemaleSOD1
G93A
/Cx3cr1
+/–
or SOD1
G93A
/Cx3cr1
–/–
mice. Notably, there is precedent for a gender
effect in this model, as absence of the p75
NTR
receptor was selectively
beneficial for female, but not male, SOD1
G93A
mice
21
. In the aggregate,
the results of these experiments further supported the role of CX3CR1
as a key regulator of microglial neurotoxicity in the contexts of either
inflammation or neurodegeneration.
DISCUSSION
We demonstrated a role for the chemokine receptor CX3CR1 in
microglial neurotoxicity in three clinically relevant models: CNS
response to systemic inflammation, the MPTP model of Parkinson
disease, and the SOD1
G93A
model of ALS. Based on complementary
expression of CX3CL1 on neurons and CX3CR1 on microglia, it has
been proposed that neuron signaling to microglia might be mediated
through this receptor
1
. Previous in vitro results support this concept:
excitotoxic injury is a potent stimulus for release of CX3CL1 from
cultured neurons
22
; CX3CR1 supports neuronal survival
23
; microglia
cultured with CX3CL1 are protected from Fas-mediated apoptosis
8
;
and CX3CL1 suppresses neuronal cell death in LPS- and IFN-g–
stimulated microglial and neuronal cocultures
24
. Our current
findings confirm this hypothesis in vivo and establish models for
investigating mechanisms of microglial neurotoxicity. For example,
we used an adoptive transfer protocol to examine neurotoxic
mechanisms in LPS-injected Cx3cr1
–/–
mice: the GFP reporter was
used to verify purification of relatively large numbers of microglia
that had been activated in vivo, and our results implicated IL-1 in
a
b
c
jk
d
e
f
g
h
i
Saline Cx3cr1
–/–
MPTP Cx3cr1
+/+
Cx3cr1
+/–
MPTP Cx3cr1
–/–
MPTP
MPTP Cx3cr1
–/–
IS
50 µm 50 µm
b
de
fg
c
Cx3cr1
–/–
Wild-type
IS
d,e
Bregma –0.10 mm
Cx3cr1
+/–
Wild-type
Cx3cr1
–/–
Wild-type
Migrating
microglia
Apoptotic
neurons
Cx3cr1
+/–
Wild-type
a
Figure 4 Adoptive transfer studies using stereotaxic placement of microglial
cells. (a–g) Thirty-six hours after stereotaxic placement of microglial cells
(a, stereotactic coordinate), recipient brains were analyzed for the presence of
GFP
+
microglia (green microglial cells in b,c; merged GFP–4,6-diamidino-2-
phenylindole (DAPI) images in d,e). Sections were stained for NeuN (blue)
and annexin V (red), shown in merged GFP-NeuN–annexin V images (f,g).
Cx3cr1
–/–
microglia (b) remained clustered at the injection site (dotted oval),
without evidence of migration along white matter or subventricular zones (d),
and apoptotic neurons near the injection site were easily detected
(f, arrowheads). In contrast, Cx3cr1
+/–
microglial clusters were not detected
at the injection site (c), no evident association with neuronal cell death was
observed (g), and these microglia distributed along white matter tracts and
subventricular zones (e). Scale bars in d–g,25mm.
Figure 5 Enhanced neurotoxic effects of MPTP in Cx3cr1
–/–
mice.
(a–i) Matched sections at the level of the SNpc were TH-immunostained
and counterstained with cresyl violet in saline- (a–c) and MPTP-treated
mice (d–i). Seven days after MPTP administration, we observed a reduction
in the number of TH-IR cells in the SNpc (outlined) of Cx3cr1
+/+
mice
(d–f), with more pronounced effects in mice lacking CX3CR1 (g–i).
(j,k) Immunofluorescent labeling of TH neurons (red) shows decreased
numbers of neurons and robust microglial activation (green) in the SNpc
(arrows) of Cx3cr1
–/–
mice (inset in k, showing high-power confocal imaging
of TH-IR neurons and GFP
+
microglia) compared to Cx3cr1
+/–
mice (inset
in j). Panels a–k: same magnification, scale bars 200 mm.
NATURE NEUROSCIENCE VOLUM E 9
[
NUMBER 7
[
JULY 2006 921
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/natureneuroscience
microglial neurotoxicity following systemic inflammation. We propose
that roles of CX3CR1 in microglial activation will differ depending on
the nature and chronicity of the activating stimulus. For example,
Cx3cr1
–/–
mice show a normal level of microglial activation after
facial nerve axotomy
5
and after laser-induced injury
9
. Furthermore,
Cx3cl1
–/–
mice show relative protection from cerebral ischemia
25
,and
intrathecal injection of CX3CL1 enhances nociception
26,27
, possibly by
activating microglia.
The molecular foundations of microglial neurotoxicity have been
obscure. Here, we identify CX3CR1 as the first selective regulator of
microglial neurotoxicity in vivo. Mice deficient for CD200, a neuronal
glycoprotein whose receptor, CD200R, is expressed by all myeloid
cells, show aberrant microglial physiology including morphological
activation of microglia in the resting CNS and accelerated response
to facial nerve transection
28,29
. None of these attributes of altered
microglial function are observed in Cx3cr1
–/–
mice
5,7,30
, indicating
different functions for CD200/CD200R and CX3CL1/CX3CR1 in
regulating microglia.
We propose that tonic neuronal release of CX3CL1 provides specific
restraint of microglia in the healthy CNS, where the blood-brain barrier
is intact. In support of this concept, we found 4300 pg of soluble
CX3CL1 per milligram of aqueous extracts of the adult mouse brain
Table 1 Stereological counts of neurons in the SNpc
Saline MPTP 2d MPTP 7d
Wild-type Cx3cr1
–/–
Cx3cl1
–/–
Wild-type Cx3cr1
–/–
Wild-type Cx3cr1
+/–
Cx3cl1
+/–
Cx3cr1
–/–
Cx3cl1
–/–
n 44 4 7773 3 77
TH 11,549 ± 612 12,088 ± 752 11,958 ± 851 6,550 ± 640 2,907 ± 551 7,403 ± 632 7,086 ± 900 7,616 ± 1,761 3,564 ± 615 3,728 ± 604
Nissl 15,081 ± 730 13,851 ± 968 16,678 ± 2,506 8,448 ± 1,031 4,295 ± 741 11,309 ± 880 10,258 ± 943 12,829 ± 2,514 6,073 ± 906 6,791 ± 735
SNpc neurons (mean ± s.e.m.) were counted by stereology. P o 0.001 when comparing wild-type mice (saline-treated versus MPTP-treated), and when comparing MPTP-treated groups
(wild-type versus Cx3cr1
–/–
or Cx3cl1
–/–
) at day 2 or day 7 after MPTP administration (ANOVA with Newman-Keuls test).
Cx3cr1
+/+
Cx3cr1
+/–
SOD1
G93A
Cx3cr1
–/–
Cx3cr1
–/–
Cx3cr1
+/–
Cx3cr1
–/–
0
10
20
30
40
50
60
70
80
90
0.09
Number of neurons per section
0
14 15 16 17
Age (weeks)
18 19 20
10
20
30
40
50
60
70
80
90
100
Male SOD1
G93A
Hindlimb strength (grams)
0
20
40
60
80
100
Percentage at terminal stage
0.03
0.02
i
jk
abcd
ef gh
0 50 100
Age (d)
150 200
Cx3cr1
–/–
Male SOD1
G93A
Cx3cr1
+/–
Male SOD1
G93A
Cx3cr1
–/–
Female SOD1
G93A
Cx3cr1
+/–
Female SOD1
G93A
Figure 6 Microglial activation, neuron loss, hindlimb grip strength and survival in SOD1
G93A
/Cx3cr1 mice. (a–d) IBA-1 immunohistochemistry shows a
progression of resting microglial cells in nontransgenic Cx3cr1
–/–
spinal cord (a) to microglial activation in SOD1
G93A
/Cx3cr1
+/+
(b), SOD1
G93A
/Cx3cr1
+/–
(c)
and SOD1
G93A
/Cx3cr1
–/–
(d) mice. (e–h) Nissl staining shows healthy-appearing neurons in nontransgenic spinal cord (e, arrows). In contrast, healthy-
appearing neurons in SOD1
G93A
transgenic mice were fewer in CX3CR1-deficient mice. Many surviving cells showed chromatin condensation (f,g, arrowheads).
No healthy-appearing neurons were observed in the ventral horns of the lumbar spinal cords of SOD1
G93A
/Cx3cr1
–/–
mice, and the remaining neuronal nuclei
showed abnormally condensed chromatin (h, arrowheads). (i) Quantitation showed significantly (P ¼ 0.03) more remaining neuronal cells in SOD1
G93A
/
Cx3cr1
+/–
mice than in SOD1
G93A
/Cx3cr1
–/–
mice, and much higher neuronal counts in healthy nontransgenic Cx3cr1
–/–
controls. (j) Graphs of hindlimb grip
strength measurements showing that male SOD1
G93A
/Cx3cr1
–/–
mice lost grip strength more rapidly between weeks 15–18 than did SOD1
G93A
/Cx3cr1
+/–
males. (k) Kaplan-Meier survival curves revealed that male SOD1
G93A
/Cx3cr1
–/–
mice died sooner than any of the other groups. Scale bars in a–h,50mm.
922 VOLUME 9
[
NUMBER 7
[
JULY 2006 NATURE NEUROSCIENCE
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/natureneuroscience
(data not shown), and others have reported the presence in the CNS of
ADAM10, the catalyst of constitutive CX3CL1 release
31
.Whenthe
blood-brain barrier is disrupted, the roles of CX3CL1/CX3CR1 may be
quite different: as noted above, CX3CL1-deficient mice are moderately
protected from focal ischemic stroke
25
and mice lacking CX3CR1
develop increased disease severity after the induction of experimental
autoimmune encephalomyelitis (EAE)
6
, due at least in part to the
deficient recruitment of regulatory natural killer cells
7
.
The activities of CX3CL1/CX3CR1 in various tissues are remarkably
varied. Recent evidence from animal models implicates CX3CL1 as a
mediator of atherosclerosis
32
, and an allele of CX3CR1 that exerts
impaired adhesive function and blunted signaling is associated with
dominantly inherited, reduced risk for atherosclerotic end points
33
.
Based on these and other findings, efforts to develop pharmacological
inhibitors of CX3CR1 are in progress
34
. Our current findings suggest
that impaired CX3CR1 function in the CNS may worsen neurodegen-
eration. Therefore, agents designed to block CX3CR1 might show en-
hanced safety profiles if they were excluded from the CNS. Further, the
dominantly acting atheroprotective CX3CR1
I249/M280
variant may merit
evaluation as a risk factor for susceptibility or severity in neurodegen-
erative disorders. Finally, the signaling by which CX3CR1 regulates
microglial neurotoxicity in vivo can now be further addressed, as such an
effort may lead to new therapeutic strategies for neuroprotection.
METHODS
Mice. Cx3cl1
+/+
, Cx3cl1
–/–
, Cx3cr1
–/–
, Cx3cr1
+/–
and Cx3cr1
+/+
mice were
generated from heterozygous breeding pairs, backcrossed for more than
10 generations to C57BL/6 (refs. 2,5). Experimental protocols were performed
in accordance with US National Institutes of Health guidelines on animal care
and were approved by the Cleveland Clinic Animal Care and Use Committee.
Lineage marker analyses of the CX3CR1-GFP
+
population in the CNS.
Cx3cr1
+/–
mice were perfused and 30-mm sections were stained with rabbit
polyclonal antibody to IBA-1 (anti–IBA-1), antibody to NG2 (anti-NG2) or
antibody to GFAP (anti-GFAP), followed by Cy3-conjugated secondary anti-
bodies (mounted in FluorSave, Calbiochem). For IBA-1 and GFAP stainings,
sections were incubated overnight at 4 1C. For NG2 staining, sections were
incubated with primary antibody (gift from W. Stallcup) at room temperature
(22 1C) for 5 h, followed by incubation at 4 1C for 48 h. Sections were imaged
by confocal microscopy, and projections of 20-mm z-stacks at 40 magnifica-
tion are shown.
Induction and analysis of systemic inflammation by lipopolyssacharide
administration. Mice were injected daily (i.p.) for 4 d with LPS (Sigma;
20 mgin100ml phosphate-buffered saline, PBS) or mock-injected with PBS.
Four hours after final injection, mice were killed and peritoneal cells were
aspirated with Hank’s Balanced Salt Solution(HBSS), washed and resuspended
in TRIZOL (Invitrogen). RNA was extracted and subjected to RPA with
32
P-labeled antisense riboprobes (BD Pharmingen) with results for each
cytokine normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
signals in two independent experiments. Sections from LPS- or saline-injected
mice were stained with rabbit polyclonal anti–IBA-1 or antibody to annexin V
(Novus Biologicals), counterstained with 0.5% cresyl violet (Sigma), and
positive cells in the dentate gyrus were counted at 40 magnification in
6–8 matched 30-mm sagittal sections. Results are presented as annexin
V–positive cells per mm
2
, with areas measured from 10 images using Image
J.1.34vi software (National Institutes of Health).
Isolation of microglial cells, RNase protection and flow cytometry. After
perfusion in HBSS without Ca
2+
and Mg
2+
, brains were collected in 10 ml
HBSS per brain, containing 0.05% collagenase D (Roche), 0.1 mgml
–1
N-tosyl-
L-leucine chloromethyl ketone (TLCK, Sigma), 10 mgml
–1
dispase (Roche) and
10 mM HEPES buffer (Invitrogen). Brain tissues were dispersed with a glass
dounce homogenizer, and cells were separated over discontinuous per-
coll gradients (Supplementary Methods online). Cells were washed and
resuspended in TRIZOL for RNA extraction. RNA was assessed using a
Bioanalyzer (Affymetrix). Representative cytokines were determined by RPA
as described for peritoneal cells. For flow cytometry, cells were diluted in PBS
(4 10
5
cells ml
–1
), and GFP
+
cells were acquired on an LSR (BD Immuno-
cytometry), gated by forward and side light-scattering properties and analyzed
with WinList software (Verity Software).
Adoptive transfer of microglia. For intracranial microinjections, microglia
were resuspended at 8 10
6
cells ml
–1
(Supplementary Methods) and trans-
ferred within 30 min of isolation. Recipient mice (4–5 weeks old) were
anesthetized with ketamine (30 mg per kg body weight) and xylazine (4 mg
per kg body weight) and injected intracranially with a 60-mlcellsuspension
(B5 10
5
cells). Using 1-ml syringes, 25-gauge 5/8’’ needles were inserted to a
2-mm depth in the frontal cortex with handmade needle-cap adaptors. C57BL/
6 recipients received Cx3cr1
+/–
,orCx3cr1
–/–
microglia alone or with 3 ng of
IL-1RA or human albumin carrier protein. Il1r1
–/–
mice received Cx3cr1
–/–
microglia. Sections were analyzed by confocal microscopy (GFP/NeuN-Cy5/
TUNEL-rhodamine). Microglia and apoptotic neurons were counted in four
high-power fields (HPF) per section (4 sections per mouse, n ¼ 4miceper
group), and results are shown as number of cells per HPF. For stereotactic
placement of microglia, recipient mice 8–10 weeks old were anesthetized with
ketamine (200 mg per kg body weight) and xylazine (10 mg per kg body
weight), and heads were secured in a stereotaxic head frame (Kopf Instru-
ments). A 10-ml Hamilton syringe with a 29-gauge needle was inserted into the
left motor cortex through a small hole drilled through the skull. Cells (2–3
10
5
cells in 10 ml) were injected at a flow rate of 1 mlmin
–1
at the following
coordinates anteroposterior, –0.12 mm; lateral, 1.7 mm; dorsoventral, 2 mm.
After completion, the needle was left in place for 5 min, then withdrawn at
0.2 mm min
–1
. Brains were removed 36 h after transfer, sectioned serially from
the level of the injection site to a depth of 1,000 mm and double-stained for
NeuN and TUNEL or annexin V.
Administration of MPTP and analysis of effects. Male mice, 8–10 weeks old
and weighing 22–28 g, were injected i.p. four times at 1-h intervals with saline
or MPTP (Sigma; 10 mg per kg body weight) and killed 2 d or 7 d after
injection. Brains were serially sectioned through the substantia nigra and
stained with antibody to TH (Chemicon) followed by Cy3-conjugated or
biotinylated secondary antibodies; this was followed by diaminobenzidine
(DAB) substrate development and cresyl violet counterstaining. Confocal
images of TH
+
neurons and GFP
+
microglia were obtained in 25-mm z-stacks.
We counted the number of TH-IR and Nissl-stained SNpc neurons in seven
mice per group, using the optical fractionator method
35,36
(Supplementary
Methods). The total number of SNpc neurons was calculated as the product of
neuron (TH-IR or Nissl-stained) densities and the volume of the SNpc (ref. 36).
Generation and analysis of SOD1
G93A
/Cx3cr1 mice. Male G93A-SOD1
mutation (SOD1
G93A
)transgenicmice
20
from Jackson Laboratories were bred
with Cx3cr1
–/–
females. F1 nontransgenic/Cx3cr1
–/–
or nontransgenic/Cx3cr1
+/–
females were bred with SOD1
G93A
/Cx3cr1
+/–
males to produce F2 SOD1
G93A
/
Cx3cr1
–/–
, SOD1
G93A
/Cx3cr1
+/–
, SOD1
G93A
/Cx3cr1
+/+
and nontransgenic/
Cx3cr1
–/–
mice.
Limb strength was assessed between the ages of 7 and 23 weeks (n ¼ 6–20
per group) as described
37
. Survival was recorded for each SOD1
G93A
/Cx3cr1
–/–
and SOD1
G93A
/Cx3cr1
+/–
mouse, with the terminal stage defined by righting
response Z 20 s.
Histopathology was analyzed at 133 d in terminal SOD1
G93A
mice and in
four age-matched nontransgenic SOD1
G93A
/Cx3cr1
–/–
littermates. Motor neu-
rons in lumbar spinal cords were counted in Nissl-stained sections, and
microglial cells were visualized by IBA-1 immunohistochemistry. The area
occupied by IBA-1 immunoreactivity was obtained using Image J. 1.34vi
software, in six images acquired at 20 from three spinal cords per mouse
(n ¼ 5micepergroup).
Statistics. We used analysis of variance (ANOVA) to analyze numbers of
annexin V–positive cells after LPS administration, neuronal apoptosis in the
adoptive transfer experiments and motor neuron counts and microglial area in
spinal cords of SOD transgenic mice. Cytokine analyses by RPA were compared
using Student’s t-test. The numbers of TH-IR and Nissl-stained neurons in the
NATURE NEUROSCIENCE VOLUM E 9
[
NUMBER 7
[
JULY 2006 923
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/natureneuroscience
SNpc were determined by stereology and compared by ANOVA with Newman-
Keuls post-test. Behavior data (body weight, forelimb grip strength and
hindlimb grip strength) were collected for males and females in three groups
of mice, SOD1
G93A
/Cx3cr1
+/–
, SOD1
G93A
/Cx3cr1
–/–
and nontransgenic
Cx3cr1
–/–
, with separate analyses of data from weeks 9 to 14 and weeks 15 to
20. We used a random coefficient mixed model to estimate the effects of
genotype, gender and age (Supplementary Methods).
Note: Supplementary information is available on the Nature Neuroscience website.
ACKNOWLEDGMENTS
We acknowledge B. Trapp (Cleveland Clinic, Cleveland) for IBA-1 antibodies,
W. Stallcup (Burnham Institute, La Jolla, California) for NG-2 antibodies,
C. Canasto (Mount Sinai School of Medicine, New York) for technical assistance
with CX3CL1 mice, R. Zhang (Mass Spectrometry Core II, Cleveland Clinic) for
assistance with MPP+ measurements, C. Shemo (Flow Cytometry Core, Cleveland
Clinic) for assistance with flow cytometry, and J. Drazba (Lerner Research Institute
Imaging Core, Cleveland Clinic) for assistance with confocal microscopy. R.H.
Miller (Case Medical School, Cleveland) provided helpful comments about the
manuscript. This work was supported by the US National Institute of Health
(NS32151), the Charles A. Dana Foundation, the National Multiple Sclerosis
Society (fellowship FG1528-A-1 to A.C.), the Robert Packard Foundation for
ALS Research at Johns Hopkins University and the Boye Foundation.
AUTHOR CONTRIBUTIONS
A.E.C. performed the experimental design of the LPS and MPTP models, and
carried out the microglia isolation, tissue staining and microglial transfer
experiments. E.P.P. and V.K. carried out the experiments with SOD
G93A
transgenic mice and assisted with manuscript preparation. M.E.S. and S.M.C.
assisted in the maintenance of the mouse colony, genotyping, histopathological
staining and neuronal counting. I.M.D. assisted in the development of the
stereotaxic protocol. D.H. collaborated in the colocalization of lineage markers
with the GFP reporter. G.K. assisted with the confocal analyses and imaging.
S.D. assisted with stereology methods. R.D. collaborated in the analysis of the
gene expression data from nuclease protection assays. J.-C.L. performed the
statistical analyses for all experiments. D.N.C., S.J., S.A.L. and D.R.L. generated
the highly inbred receptor- and ligand-deficient mouse strains, and assisted with
the experimental design and manuscript preparation. R.M.R. provided the basis
for the development of the experimental designs. A.E.C. and R.M.R. analyzed
the data, interpreted the results and prepared the manuscript.
COMPETING INTERESTS STATEMENT
The authors declare that they have no competing financial interests.
Published online at http://www.nature.com/natureneuroscience
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/
1. Harrison, J.K. et al. Role for neuronally derived fractalkine in mediating interactions
between neurons and CX3CR1-expressing microglia. Proc. Natl. Acad. Sci. USA 95,
10896–10901 (1998).
2. Cook, D.N. et al. Generation and analysis of mice lacking the chemokine fractalkine.
Mol. Cell. Biol. 21, 3159–3165 (2001).
3. Niess, J.H. et al. CX3CR1-mediated dendritic cell access to the intestinal lumen and
bacterial clearance. Science 307, 254–258 (2005).
4. Geissmann, F., Jung, S. & Littman, D.R. Blood monocytes consist of two principal
subsets with distinct migratory properties. Immunity 19, 71–82 (2003).
5. Jung, S. et al. Analysis of fractalkine receptor CX(3)CR1 function by targeted deletion and
green fluorescent protein reporter gene insertion. Mol. Cell. Biol. 20, 4106–4114 (2000).
6. Haskell, C.A. et al. Targeted deletion of CX(3)CR1 reveals a role for fractalkine in cardiac
allograft rejection. J. Clin. Invest. 108, 679–688 (2001).
7. Huang, D. et al. The neuronal chemokine CX3CL1/fractalkine selectively recruits NK
cells that modify experimental autoimmune encephalomyelitis within the central
nervous system. FASE B J. 20, 896–905 (2006).
8. Boehme, S.A., Lio, F.M., Maciejewski-Lenoir, D., Bacon, K.B. & Conlon, P.J. The
chemokine fractalkine inhibits Fas-mediated cell death of brain microglia. J. Immunol.
165, 397–403 (2000).
9. Davalos, D. et al. ATP mediates rapid microglial response to local brain injury in vivo.
Nat. Neurosci. 8, 752–758 (2005).
10. Nimmerjahn, A., Kirchhoff, F. & Helmchen, F. Resting microglial cells are highly
dynamic surveillants of brain parenchyma in vivo. Science 308, 1314–1318 (2005).
11. Kreutzberg, G.W. Microglia: a sensor for pathological evets in the CNS. Trends Neuros ci.
19, 312–318 (1996).
12. Nataf, S. et al. Brain and bone damage in KARAP/DAP12 loss-of-function mice correlate
with alterations in microglia and osteoclast lineages. Am. J. Pathol. 166, 275–286
(2005).
13. Takahashi, K., Rochford, C.D. & Neumann, H. Clearance of apoptotic neurons without
inflammation by microglial triggering receptor expressed on myeloid cells-2. J. Exp.
Med. 201, 647–657 (2005).
14. Ito, D. et al. Microglia-specific localisation of a novel calcium binding protein, Iba1.
Brain Res. Mol. Brain Res. 57, 1–9 (1998).
15. Hulshof, S. et al. CX3CL1 and CX3CR1 expression in human brain tissue: noninflamma-
tory control versus multiple sclerosis. J. Neuropathol. Exp. Neurol. 62, 899–907 (2003).
16. Rivest, S. Molecular insights on the cerebral innate immune system. Brain Behav.
Immun. 17, 13–19 (2003).
17. Carson, M.J., Reilly, C.R., Sutcliffe, J.G. & Lo, D. Disproportionate recruitment of
CD8
+
T cells into the central nervous system by professional antigen-presenting cells.
Am.J.Pathol.154, 481–494 (1999).
18. Rothwell, N. Interleukin-1 and neuronal injury: mechanisms, modification, and ther-
apeutic potential. Brain Behav. Immun. 17, 152–157 (2003).
19. Zujovic, V., Benavides, J., Vige, X., Carter, C. & Taupin, V. Fractalkine modulates TNF-
alpha secretion and neurotoxicity induced by microglial activation. Glia 29, 305–315
(2000).
20. Facchinetti, F. et al. Lack of involvement of neuronal nitric oxide synthase in the
pathogenesis of a transgenic mouse model of familial amyotrophic lateral sclerosis.
Neuroscience 90, 1483–1492 (1999).
21. Kust, B.M., Brouwer, N., Mantingh, I.J., Boddeke, H.W. & Copray, J.C. Reduced p75NTR
expression delays disease onset only in female mice of a transgenic model of familial
amyotrophic lateral sclerosis. Amyotroph. Lateral Scler. Other Motor Neuron Disord. 4,
100–105 (2003).
22. Chapman, G.A. et al. Fractalkine cleavage from neuronal membranes represents an
acute event in the inflammatory response to excitotoxic brain damage. J. Neurosci. 20,
RC87 (2000).
23. Meucci, O., Fatatis, A., Simen, A.A. & Miller, R.J. Expression of CX3CR1 chemokine
receptors on neurons and their role in neuronal survival. Proc. Natl. Acad. Sci. USA
[erratum appears in Proc. Natl. Acad. Sci. USA 98, 15393 (2001)] 97, 8075–8080
(2000).
24. Mizuno, T., Kawanokuchi, J., Numata, K. & Suzumura, A. Production and neuroprotec-
tive functions of fractalkine in the central nervous system. Brain Res. 979,65–70
(2003).
25. Soriano, S.G. et al. Mice deficient in fractalkine are less susceptible to cerebral
ischemia-reperfusion injury. J. Neuroimmunol. 125, 59–65 (2002).
26. Milligan, E. et al. An initial investigation of spinal mechanisms underlying pain
enhancement induced by fractalkine, a neuronally released chemokine. Eur. J. Neurosci.
22, 2775–2782 (2005).
27. Milligan, E.D. et al. Evidence that exogenous and endogenous fractalkine can induce
spinal nociceptive facilitation in rats. Eur. J. Neurosci. 20, 2294–2302 (2004).
28. Hoek, R.M. et al. Down-regulation of the macrophage lineage through interaction with
OX2 (CD200). Science 290, 1768–1771 (2000).
29. Wright, G.J. et al. Lymphoid/neuronal cell surface OX2 glycoprotein recognizes a
novel receptor on macrophages implicated in the control of their function. Immunity
13, 233–242 (2000).
30. Lesnik, P., Haskell, C.A. & Charo, I.F. Decreased atherosclerosis in CX3CR1
–/–
mice
reveals a role for fractalkine in atherogenesis. J. Clin. Invest. 111, 333–340 (2003).
31. Hundhausen, C. et al. The disintegrin-like metalloproteinase ADAM10 is involved in
constitutive cleavage of CX3CL1 (fractalkine) and regulates CX3CL1-mediated cell-cell
adhesion. Blood 102, 1186–1195 (2003).
32. Cybulsky, M.I. & Hegele, R.A. The fractalkine receptor CX3CR1 is a key mediator of
atherogenesis. J. Clin. Invest. 111, 1118–1120 (2003).
33. McDermott, D.H. et al. Association between polymorphism in the chemokine receptor
CX3CR1 and coronary vascular endothelial dysfunction and atherosclerosis. Circ. Res.
89, 401–407 (2001).
34. Charo, I.F. & Ransohoff, R.M. The many roles of chemokines and chemokine receptors in
inflammation. N. Engl. J. Med. 354, 610–621 (2006).
35. Sugama, S. et al. Age-related microglial activation in 1-methyl-4-phenyl-1,2,3,6-
tetrahydropyridine (MPTP)-induced dopaminergic neurodegeneration in C57BL/6
mice. Brain Res. 964, 288–294 (2003).
36. West, M.J. New stereological methods for counting neurons. Neurobiol. Aging 14,
275–285 (1993).
37. Mitsumoto, H. et al. Effects of cardiotrophin-1 (CT-1) in a mouse motor neuron disease.
Muscle Nerve 24, 769–777 (2001).
924 VOLUME 9
[
NUMBER 7
[
JULY 2006 NATURE NEUROSCIENCE
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/natureneuroscience
