Viral-induced encephalitis initiates distinct and
functional CD103+ CD11b+ brain dendritic cell
populations within the olfactory bulb
Paul M. DAgostinoa, Changsoo Kwaka, Haley A. Vecchiarellia, Judit Gal Totha, James M. Millera, Zahrah Masheeba,
Bruce S. McEwena,1, and Karen Bullocha,b,1
a
Laboratory of Neuroendocrinology and bLaboratory of Cellular Physiology and Immunology, The Rockefeller University, New York, NY 10065
Contributed by Bruce S. McEwen, March 6, 2012 (sent for review November 21, 2011)
Dendritic cells (DC) are antigen-presenting cells found in both
lymphoid and nonlymphoid organs, including the brain (bDC) of
Cd11c/eyfp transgenic C57BL/6 mice. Using an intranasal vesicular
stomatitis virus infection, we demonstrated that EYFP+ cells amass
in areas associated with viral antigens, take on an activated mor-
phology, and project their processes into infected neuronal tissue
within the olfactory bulb. These bDC separated into three EYFP+
CD45+ CD11b+ populations, all but one being able to functionally
promote both T lymphocyte proliferation and TH1 cytokine pro-
duction. One population was shown to emanate from the brain
and a second population was peripherally derived. The third pop-
ulation was of indeterminate origin, being both radiosensitive and
not replenished by donor bone marrow. Finally, each EYFP+ pop-
ulation contained CD11b+ CD103+ subpopulations and could be
distinguished in terms of CD115, Gr-1, and Ly-6C expression, high-
lighting mucosal and monocyte-derived DC lineages.
antigen presentation | neuroinammation
D endritic cells (DC) are immune cells specialized in antigen
capture, processing, and presentation to T lymphocytes for
the induction of adaptive immunity (1) or tolerance (2, 3). Since
their initial identication (4) and characterization (5, 6), DC are
now recognized to represent a heterogeneous group consisting of
many distinct subpopulations identiable by a diverse array of
cell-surface markers and functions (7, 8). DC subpopulations are
found in both lymphoid and nonlymphoid tissue, such as the
dermis, intestine, liver, lungs, kidneys, and the heart (912).
Some examples of this phenotypic heterogeneity include: plas-
macytoid DC, the primary function of which is the production of
type I interferons (13); the CD8+ DEC205+ DC in the spleen,
which are involved in cross-presentation of antigen to cytotoxic T
lymphocytes; and CD103+ CD11b+ DC in the intestine, which
migrate to draining lymph nodes in a CCR7-dependent manner
and promote the proliferation of regulatory T cells (Treg) (1416).
DC are also found within the brain (bDC) and meninges of the
Itgax (Cd11c)/eyfp transgenic (Tg) mouse (17, 18). In the steady
state, bDC are found within discrete regions of the CNS (17).
Furthermore, bDC have been shown to respond to physical
trauma and intracranial treatment with the proinammatory
cytokine IFN- (19, 20). Given the ability of bDC to present
antigen ex vivo after stimulus with IFN-, and their association
with blood brain barrier (BBB) compromised or decient ana-
tomical regions of the CNS [e.g., the olfactory bulb (OB) and
circumventricular organs], bDC appear to possess the CNS an-
tigen presenting cell function rst attributed to bone marrow-
derived perivascular microglia by Hickey and Kimura (21). To
further elucidate in vivo bDC function, we capitalized on a nat-
urally occurring infectious route to the brain, namely CNS access
via olfactory nerves within the nasal epithelia.
Using vesicular stomatitis virus (VSV) to elicit an encephalitic
state after intranasal administration (22), we sought to better
understand the bDC response to an acute infection via pheno-
typic and functional characterization. VSV is a member of the
Rhabdoviridae viral family, most notably characterized by its
enveloped bullet-shaped virion, negative sense single-stranded
www.pnas.org/cgi/doi/10.1073/pnas.1203941109
RNA genome, and high sensitivity to IFN-elicited immune re-
sponses (23). Many VSV attributes, including host range, natural
reservoirs, endemic regions, and infectious cycle are well known.
One characteristic in particular, the polarized nature of VSV
virion budding, is the basis for the intranasal VSV infection
model used hereand by many othersto study virally induced
encephalitis in the mouse (24, 25). VSV is capable of accessing
the CNS via olfactory nerve cells that span the cribriform plate;
this, and its neuro-invasive pathology, make it amenable to
studying proinammatory responses within the brain (26, 27).
In this work we found that VSV infection of Cd11c/eyfp Tg mice
results in anatomical changes in CD11c+ cell distribution and
morphology relative to regions associated with viral antigen ex-
pression. Flow cytometric analysis reveals the presence of unique
CD45+, CD11b+, and CD11c+ cell populations within the VSV-
infected OB, each with its own distinct phenotype. Functional
analysis of these populations demonstrates the ability of some to
process and present antigen to T lymphocytes on par with splenic
conventional DC (cDC). Finally, we explore the origin and lineage
of these unique bDC populations within the context of viral in-
fection using radiation chimeras, EdU-labeling, and phenotyping.
These data revealed the presence of CD103+ CD11b+ double-
positive cells similar to those found in mucosal DC subsets of the
intestinal lamina propria and the basal lamina of the bronchial
epithelium (10, 14), as well as a population of radio-sensitive, bone
marrow-independent cells resembling monocyte-derived DC.
Results
Cd11c/eyfp-Expressing Cells Respond to Early Stages of Acute VSV
Infection Within the OB. To investigate how CD11c-expressing
cells respond to an infection caused by a pathogens ability to
circumvent the BBB, we intranasally infected Cd11c/eyfp Tg mice
with VSV. Three cohorts of 12 mice received a LD50 of VSV,
UV-inactivated VSV, or vehicle control. Mice were killed fol-
lowing 24, 48, 72, and 96 h postinfection (hpi), and 60-m OB
IMMUNOLOGY
sections were processed for immunouorescence staining with an
anti-VSV antiserum. Over the rst 24 hpi, we observed VSV
antigen within the olfactory nerve layer with no apparent change
in the EYFP+ distribution pattern, relative to naive mice (Fig.
S1A). By 48 hpi, the VSV infection had progressed through the
olfactory nerve and penetrated the glomerular layer, again with
little change in EYFP+ cell distribution. However, by 72 hpi,
EYFP+ cells were found in abundance throughout VSV-infected
areas and continued to increase after 96 hpi. This accumulation
of EYFP+ cells was only observed with viable VSV and was not
reproduced with either UV-inactivated virus or vehicle controls
at any time point postinfection (Fig. S1B). These data indicate
Author contributions: P.M.D. and K.B. designed research; P.M.D., C.K., H.A.V., J.G.T.,
J.M.M., and Z.M. performed research; P.M.D., C.K., H.A.V., B.S.M., and K.B. analyzed data;
and P.M.D., B.S.M., and K.B. wrote the paper.
The authors declare no conict of interest.
1
To whom correspondence may be addressed. E-mail: bulloch@rockefeller.edu or mcewen@
mail.rockefeller.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1203941109/-/DCSupplemental.
PNAS | April 17, 2012 | vol. 109 | no. 16 | 61756180
that a putative bDC response was solely attributable to an active
VSV infection rather than from irritation caused by inhalation of
liquid or innate immune responses within infected neurons
triggered by pathogen-associated molecular patterns. Further-
more, CD11c+ cells that accumulated within these VSV-infected
regions exhibited an activated morphology signied by stout cell
bodies with stunted processes (28), unlike the thin, delicately
branched processes associated with bDC in unchallenged mice
(Fig. S2). Closer examination revealed that CD11c+ cells clus-
tered around VSV-infected neurons within glomeruli (Fig. 1A),
and projected their processes into neuropil-expressing VSV anti-
gens (Fig. 1B). This phenomenon was reminiscent of the hallmark
probing behavior attributed to both brain resident microglia (MG)
and peripheral DC subsets (4, 29, 30).
To correlate the putative bDC response to MG, changes in MG
distribution were analyzed in response to the VSV infection. In
previously published work, MG were discriminated from putative
bDC in the Cd11c/eyfp Tg mouse, primarily based on EYFP ex-
pression in conjunction with either an Iba1+ or F4/80+ phenotype
(17). Using this denition, the vast majority of EYFPneg Iba1+ cells
showed no distributive differences between areas proximal and
distal to the infection relative to vehicle control groups (Fig. S3),
even though EYFPneg Iba1+ displayed an activated morphology
near the site of infection. Given the MGs developmental plasticity
(31), we subsequently sought to phenotypically and functionally
characterize the CD11c+ cell response to verify their role as bDC.
bDC Responding to Acute VSV Infection Separate into Three Pheno-
typically and Functionally Distinct Populations. Given that the BBB
remains uncompromised during an intranasal VSV infection
until 68 d postinfection (22), we chose to analyze bDC accu-
mulation at 96 hpi to maximize cell numbers while minimizing
peripheral immune contributions. As such, bDC were isolated
and phenotypically characterized by ow cytometry at 96 hpi in
the VSV-infected OB. After gating for live, singlet, TCRneg,
and CD19neg cells (Fig. S4), EYFP+ cells resolved into three
distinct CD45+ and CD11b+ populations only prevalent in
infected OB (Fig. 2A). The EYFP+ CD45int CD11bhi (P1)
phenotypic population had a similar CD45 and CD11b pheno-
type as previously described resident MG populations found in
the steady state and inamed brains of wild-type animals (32
35). The two other populations, only found at appreciable levels
in the VSV-infected OB, were dened by EYFP+ CD45hi
CD11bhi (P2) and EYFP+ CD45hi CD11bint (P3) phenotypes. P1
had the greatest number of EYFP+ cells compared with P2
Fig. 1. bDC surround and probe VSV-infected glomeruli. Representative
confocal z-stack analysis of an OB 72 hpi after intranasal VSV infection. (A) A
glomerulus labeled with anti-VSV antiserum (red) showed clustering bDC
(green). (B) Confocal section visualizing a bDC extending cellular projections
(arrowheads) into VSV infected glomerular neuropil. Representative images
from three experiments with an n = 3. (Scale bars, 10 m.)
6176 | www.pnas.org/cgi/doi/10.1073/pnas.1203941109
(P valueP1vsP2 < 0.001) (Fig. 2B) and P3 (P valueP1vsP3 < 0.0001),
and P3 had the least number of EYFP+ cells (P valueANOVA <
0.0001). In summary, most EYFP+ cells within the VSV-infected
OB at 96 hpi fall within the P1 population.
Next, the ability of the three EYFP+ populations to function
as antigen-presenting cells was examined using a recombinant
version of VSV (rVSV-OVA) able to transduce ovalbumin
(OVA) expression in the cells it infects (36). rVSV-OVA was
used to induce the robust increases in EYFP+ cells observed in
wild-type VSV infections, while simultaneously priming those cells
in vivo with OVA. This rVSV-OVA infection model was sub-
stantiated by (i) confocal micrographs demonstrating colocaliza-
tion of OVA and VSV antigens within infected glomeruli (Fig.
3A), and (ii) ow cytometry analysis that revealed an increase
in EYFP+ bDC levels in the OB, which resolved into the three
populations identied during wild-type VSV infections (Fig. 3B).
Using CD45, CD11b, and EYFP as markers, cells from P1, P2,
and P3 were then isolated by FACS at 96 hpi from rVSV-OVA
intranasally infected OB, along with MG (identied as CD45int
CD11bhi EYFPneg). As controls, splenic cDC (B220neg EYFPhi)
were also isolated from both steady state and rVSV-OVA in-
traperitoneally infected Cd11c/eyfp Tg mice. Sorted P2 and P3
bDC, as well as splenic cDC from infected mice, were capable of
directly stimulating CD8 OT-I, but not CD4 OT-II, T-cell pro-
liferation (Fig. 3C). The magnitude of the observed proliferation
was consistent with both the small percentage of OT-I and OT-II
epitopes present within the OVA primary sequence (SIINFEKL
2% and ISQAVHAAHAEINEAGR 4% of total OVA amino
acids, respectively) and the low levels of OVA mRNA produced
by the VSV RNA-dependent RNA polymerase, given the genes
position in the recombinant viral genome (37). P1 and MG from
rVSV-OVA infected OB, as well as splenic cDC from uninfected
mice, were unable to stimulate OT T-cell proliferation without
further antigen supplementation. In summary, these data illus-
trate ongoing in vivo processing of OVA antigens by P2, P3, and
cDC controls, but not P1 and MG, as illustrated by their ability to
facilitate OVA specic T-cell proliferation.
When the EYFP+ populations were supplemented with excess
OVA or the appropriate OT peptide signicant proliferation of
OT T cells was observed, in some cases out-performing splenic
cDC populations. However, in all antigen-presentation assays,
MG were unable to present OVA, OT-I or OT-II peptide anti-
gens to their respective OT T cells (Fig. 3 D and E). Similarly, P1
cells were unable to process and present OVA or OT-II peptide;
however, they were able to facilitate CD8 OT-I T-cell pro-
liferation with the SIINFEKL peptide, albeit at very low levels.
In stark contrast, both P2 and P3 populations were effectively
able to present supplemented peptide antigens and OVA ex vivo
to OT T cells. In most cases, the P3 population induced a more
robust proliferation than that observed with steady state or
rVSV-OVA cDC controls (Fig. 3 D and E). This higher func-
tional ability most likely arose from greater levels of phenotyp-
ically mature DC found within the intranasally infected OB than
spleens from intraperitoneally infected mice, because the spleen
primarily monitors the blood and is not likely to be the draining
lymphoid organ for infections of the peritoneal cavity.
Finally, cytokine proles were analyzed from proliferating
CD4+ OT-II T cells induced during antigen presentation assays
(Fig. S5). Both P2 and P3 populations promoted a proinam-
matory helper T-cell (TH1) biased CD4 T lymphocyte response
characterized by elevated levels of IL-2, IFN-, and TNF, similar
to both splenic cDC populations. MG and the P1 population,
however, had undetectable levels of both pro- and anti-inam-
matory cytokines, which was consistent with the lack of CD4+
T-cell proliferation elicited by those cells. In summary, P2 and
P3, much like their cDC counterparts, were effectively able to
orchestrate both OVA-specic cytotoxic and TH1-mediated
responses after in vitro supplementation with OVA or its pep-
tides. Conversely, P1 and MG did not participate in establishing
such antiviral related responses, although P1 did show a mild
ability to induce cytotoxic effector cells.
DAgostino et al.

Origin of VSV-Induced bDC Populations. Radiation chimeras were
used to ascertain the origin of EYFP+ cell populations that respond
to acute VSV infection. After verifying chimerism by analyzing
CD45.1+ and CD45.2+ leukocytes in the blood, CD11b+ cells within
the OB were assessed for surface CD45.1+ or CD45.2+ expression.
In lethally irradiated Cd45.1 genetically engineered host mice
replenished with Cd11c/eyfp (CD45.2) Tg bone marrow, radio-re-
sistant cells were identied at 96 hpi that resembled the P1 pop-
ulation in that they were CD45.1int CD11bhi (Fig. 4A). Fur-
thermore, a second population was identied that expressed the
CD45.2hi CD11bhi donor phenotype and resembled P2. These
observations were veried by EYFP uorescence because the
population resembling P1 was found to be EYFPneg, but the
population similar to P2 was EYFP+ (Fig. 4D). These data were
also anatomically corroborated in the VSV-infected chimeric OB,
because EYFP+ cells were present in low numbers only at the site
of acute viral infection (Fig. S6A). These data were recapitulated
in reciprocally chimeric mice: that is, lethally irradiated Cd11c/eyfp
(CD45.2) Tg host mice replenished with Cd45.1 genetically engi-
neered bone marrow. In short, the P1 population was CD45.2int
CD11bhi, radio-resistant and EYFP+, and a second population
resembling P2 originated from CD45.1hi CD11bhi donor bone
marrow and was EYFPneg (Fig. 4 B and D). These observations
were anatomically represented by the high accumulation of EYFP+
cells typically seen virally infected, nonirradiated Cd11c/eyfp
(CD45.2) Tg mice (Fig. S6B). Taken together, these data suggest
that P1 are radio-resistant and represent a resident population
within the brain, and P2 are derived from bone marrow progenitors.
Unexpectedly, no population resembling P3 was associated
with CD45.1+ or CD45.2+ leukocytes in either chimeric model,
DAgostino et al.
Fig. 2. EYFP+ bDC within the VSV-infected OB exhibit three
phenotypically distinct populations at 96 hpi. (A) Three dis-
tinct EYFP+, CD45+, and CD11b+ populations. (B) Analysis of
the differences in each CD45+ and CD11b+ population shows
that the P1 population represents the majority of EYFP+ cells;
the P2 and P3 populations were less abundant. The EYFP+ cell
numbers found within the P1 population are nearly double
those in the P2 and P3 populations combined. Data repre-
sentative of three separate experiments with n = 3; error bars
represent SEM. ***P < 0.001
even though all three populations were observed in a control in-
fection of Cd11c/eyfp Tg mice (Fig. 4C). Consequently, the source
for P3 is ambiguous at this point and its absence could be in-
dicative of (i) a radiosensitive population that is not replenished
by bone marrow progenitor cells within the time constraints of our
viral infection model (i.e., mice are optimally susceptible to VSV
infection between 5 and 7 wk of age, allowing only 4 wk for chi-
merism), or (ii) a confound of the irradiation model.
Given that the chimera data supports P1 cells as part of the
brain parenchyma, the question still remains as to how these
EYFP+ cells go from very low numbers in the steady-state OB to
such high levels between 72 and 96 hpi of acute VSV infection.
To address this question, EdU-labeled proliferation was assessed
in Cd11c/eyfp Tg mice intranasally infected with VSV (LD50) or
vehicle alone. After 96 hpi, there was an 71.7% increase in
EdU+ cells found within the VSV-infected OB, relative to HBSS
controls (Fig. S7A). However, none of the three identied
EYFP+ CD45+ CD11b+ populations had incorporated EdU (Fig.
S7B); this was interesting considering P1 comprised the largest
proportion of EYFP+ cells responding to the VSV infection.
Lineage of VSV-Induced bDC Populations. In an effort to better
understand the lineage associated with all three EYFP+ pop-
ulations, these cells were separately phenotyped in nonirradiated
mice by ow cytometry using cellular markers commonly asso-
ciated with DC subsets, as well as other immunological cells. As
controls, MG isolated from VSV-infected OB were character-
ized along with cDC (TCRneg, CD19neg, NK1.1neg, and EYFPhi)
from the spleen of VSV-infected mice (Fig. 5A and summarized
in Table S1). As expected, all three EYFP+ populations were
Fig. 3. Recombinant VSV-OVA induced bDC process
and present antigen to OVA-specic T lymphocytes.
(A) Representative confocal images of anti-OVA (red)
and EYFP (green) from OB sections from mice in-
IMMUNOLOGY
tranasally infected with rVSV-OVA at 96 hpi, resulting
in OVA and VSV antigen (blue) colocalization within
glomeruli. (Scale bars, 50 m.) (B) rVSV-OVA infection
elicited EYFP+ populations analogous to wild-type
VSV. (C) P2 and P3 processed in vivo antigens during
a rVSV-OVA intranasal infection and drove ex vivo
proliferation of OVA-restricted CD8+, but not CD4+ T
cells. Bar graphs represent the average number of
proliferating T cells, as determined by a shift in 5-(and
6)-carboxyuorescein diacetate succinimidyl ester
intensity. (D) All three bDC populations facilitated
OVA-restricted CD8+ T-cell proliferation after ex
vivo supplementation with OT-I peptide; however P1
was ineffective at presenting OVA. Representative
histograms from OT-I peptide treated DC populations
plated 1:30 with T cells. (E) P2 and P3 bDC effectively
presented OVA antigens to promote OVA-restricted
CD4+ T-cell proliferation, but P1 was ineffective. His-
tograms from OT-II peptide-treated DC populations
plated 1:30, and OVA treated plated 1:5. Error bars
represent SEM; 612 replicates per treatment group.
*P valuevs. T Cell Only < 0.05, **P valuevs. T Cell Only < 0.01,
***P valuevs. T Cell Only < 0.001.
PNAS | April 17, 2012 | vol. 109 | no. 16 | 6177
positive for CD11c and MHC I, although P1 cells expressed the
lowest levels of CD11c relative to all other EYFP+ populations.
All three EYFP+ populations also contained CD103+ cells, with
P3 possessing the least; MG were mostly CD103low cells. Addi-
tionally, all populations were found mostly negative for CD8
and DEC-205, with the exception being a very small P3 sub-
population, and all populations were found negative for CD4
and Siglec H. In summary, the three EYFP+ populations were
distinguishable from MG based on CD11c and CD103 expres-
sion, and from cDC subsets in the spleen associated with cross-
presentation (i.e., CD8+ and DEC-205+).
Further differences between EYFP+ populations were evident
when examining MHC II and the expression patterns of the
costimulatory molecules. First, P1 cells were found to be MHC
IIneg, and a majority of the P2 and P3 populations contained
MHC II+ cells. Second, P1 and MG expressed low levels of the
costimulatory molecules CD40 and CD80, and moderate levels
of CD86, but these markers were found at relatively higher levels
within P2 and P3. These results were consistent with P2 and P3
representing mature DC subsets capable of antigen presentation.
In addition to differences in the expression of DC maturation
markers, there were noted differences in the markers used to
classify DC lineage. Both P1 and P2 populations contained
macrophage colony-stimulating factor receptor (CD115)-positive
and F4/80+ cells, but P3 cells were both CD115neg and F4/80low.
In terms of Ly-6C surface expression, unlike the P1 population,
which was Ly-6Cneg, all of P2 and a majority of P3 cells were
identied as Ly-6C+. Interestingly, the P3 population also con-
tained cells positive for NK1.1, CD49b, CD122, and CD45R,
phenotypes reminiscent of NK cells and a unique IFN-producing
killer DC found within both lymph nodes and spleen (38, 39).
Finally, using a clone capable of discriminating Ly-6G (Gr-1)
from Ly-6C, P2 was the only population found to be Gr-1+.
Subsequent phenotypic analysis using CD103, MHC II, Gr-1,
and Ly-6C, as depicted in Fig. 5 B and C, revealed the three
EYFP+ cell populations could be further divided into CD103+
and CD103neg subpopulations. P1 was composed of 45.3%
7.5% CD103+ cells that were MHC II, Gr-1, and Ly-6Cnega-
tive. However, unlike P1 EYFP+ cells, a majority of P2 was
composed of CD103+ cells (62.8% 2.0%) that were also MHC
II+ and Ly-6C+ (45.0% 9.5%) with Gr-1+ cells associated with
31.7% 4.5% of the CD103+ MHC II+ Ly-6C+ population. Fi-
nally, P3 EYFP+ cells were predominantly CD103neg (85.0%
0.1%) cells that were also MHC II+ and Ly-6C+ (45.5% 4.5%).
Fig. 4. The P1 population is associated
with the brain parenchyma, but P2
EYFP+ cells are part of the inltrating
peripheral response. (A) Lethally irradi-
ated Cd45.1 mice, successfully rescued
with Cd11c/eyfp (CD45.2) Tg bone mar-
row and their (B) reciprocal chimeras
were intranasally infected with VSV at
the LD50. At 96 hpi, blood and OB were
analyzed by ow cytometry using anti-
bodies able to discriminate between
CD45.1 and CD45.2 variants. In both
scenarios a CD45int CD11bhi population
(containing P1) was identied as brain
resident and radio-resistant, and a
CD45hi CD11bhi population (containing
P2) was consistently derived from donor
progenitor cells from the periphery.
The P3 (CD45hi CD11bint) population
was conspicuously absent from each
chimera set, however, as control (C) CD11c/eyfp mice infected with the same batch of VSV possessed all three populations. These ndings were corroborated
by (D) EYFP+ cell populations. Data are representative of three experiments with n = 56 per experiment; error bars represent SEM. *P valuevs. T Cell Only < 0.05,
***P valuevs. T Cell Only < 0.001.
6178 | www.pnas.org/cgi/doi/10.1073/pnas.1203941109
Discussion
Using intranasal VSV infection we show that bDC, normally
positioned near sites associated with olfactory nerve entry into
the steady-state OB, amass in great numbers and surround infec-
ted neuronal tissue. Furthermore, we were able to discriminate
three phenotypically and functionally distinct EYFP+ cell pop-
ulations from the phenotype commonly associated with MG, two
of which with potentially different lineage associations. Because
the pathology associated with intranasal VSV, in terms of viral
progression and leukocyte recruitment, is well established (25
27, 40), we chose to examine a time point (96 hpi) where the
greatest accumulation of bDC coincided with the ability to ana-
lyze their early contributions toward afferent immune responses
before breakdown of the BBB. Our data not only depicts an in-
crease of bDC within VSV-infected areas, but closer examination
revealed these bDC often projected cellular processes directly
into infected tissue. Conversely, EYFPneg Iba1+ MG, although
morphologically activated, did not redistribute in response to the
acute viral infection. These ndings substantiate our hypothesis
that EYFP+ bDC occupy vulnerable CNS access points (in this
case they exist within tissue that have olfactory nerve projections
from the nasal epithelium) to act as rst responders capable of
orchestrating immunological responses. The nature of those im-
mune responses was the focus of our further investigation.
Phenotypic characterization demonstrated that VSV-induced
EYFP+ cells resolved into three distinct populations based on
CD45 and CD11b expression not observed in the steady state. One
of these populations, described as a radio-resistant resident
CD45int CD11bhi (P1) population, was similar in phenotype to MG
(34, 35), except that it contains a CD103+ subpopulation, a phe-
notype resembling migratory DC (10, 11). P1, however, differed
from migratory DC cell types based on surface MHC II and cos-
timulatory molecule expression. Furthermore, this population
failed to incorporate EdU during an acute VSV infection, all of
which strongly suggests P1 may differentiate from the CD103neg
CD11clow MG population found within the VSV-infected OB.
P1 and MG populations were also functionally distinguishable
from one another. Unlike MG, P1 cells were able to weakly fa-
cilitate CD8+ OT-I T-cell receptor-restricted T lymphocyte
proliferation. Given these observations, it would be interesting
to assess the phenotypic and functional characteristics of P1 during
later stages of the infection (e.g., between 6 and 8 d postinfection).
Such an analysis might reveal either a reversion to an inactivated
microglial phenotype, or provide evidence for this populations
further differentiation. Given that we have not identied an im-
munogenic role for P1 and that it represents the largest population
DAgostino et al.

responding early on during the acute VSV infection, we are cur-
rently exploring other functional possibilities for these cells.
The second EYFP+ population of radio-sensitive CD45hi
CD11bhi (P2) cells was reminiscent of peripherally inltrating
leukocytes identied during Theilers murine encephalitis virus
infection of rats, and experimental autoimmune encephalomyelitis
mouse models (34, 35). P2 cells differed from the P1 and MG
populations in terms of MHC II and costimulatory molecule ex-
pression. P2 cells were also found to contain CD103+ CD11c+ F4/
80+ Gr-1+ and Ly-6Chi subpopulations, a phenotype similar to
mucosal DC (10), and were found to be functionally active in most
T-cell proliferation assays. The P2 bDC were also able to promote
a TH1 bias in proliferating CD4+ T lymphocytes. Taken as a whole,
these observations have led us to formulate the hypothesis that P2
derive from a population of mucosal DC (10), which inltrates
DAgostino et al.
Fig. 5. bDC populations contain a diverse
array of CD103+ CD11b+ and CD103neg
CD11b+ cells. (A) Phenotypic analysis of
TCRneg, CD19neg, CD45+, and CD11b+ cells
represented by the three EYFP+ populations
using a variety of markers characteristic of
DC and other immunological subsets. MG
(CD45int, CD11b+ and EYFPneg) and common
DC (Linneg EYFPhi) from the spleen of infec-
ted mice were used as a control. bDC pop-
ulations were further discriminated by (B)
CD103+ and (C) CD103neg subpopulations
possessing MHC II, Gr-1, and Ly-6c+ cells. P1 is
composed of a more-or-less equal number of
CD103+ and CD103neg cells, but all are MHC
II, Gr-1, and Ly-6cnegative. P2 contains sev-
eral populations of cells with phenotypes
analogous to mucosal DCs. Finally, P3 is
predominantly composed of CD103neg cells
IMMUNOLOGY
with phenotypes similar to monocytes-de-
rived DC. These histograms and dot plots are
representative of three separate experi-
ments, with two mice OB pooled per anti-
body per experiment.
the OB of VSV intranasally infected mice, perhaps from the nasal
epithelium.
The third EYFP+ population observed was classied by
a CD45hi CD11bint (P3) phenotype. We believe P3 cells represent
a unique population of DC yet to be described in CNS viral in-
fection models. This population was the least abundant of the
EYFP+ cells, but had costimulatory molecule expression on par
with splenic cDC and P2 bDC, was able to process antigen and
promote T-cell proliferation (often more robustly than splenic
cDC controls), and induced a TH1 bias in proliferating CD4+ T
lymphocytes. The P3 population, however, was different from the
other two EYFP+ populations in that, although it possessed both
MHC II+ and Ly-6Chi populations, it was also mostly negative for
CD103, CD115, F4/80, and Gr-1. These phenotypes draw simi-
larities to inammatory monocytes (41, 42), which have the ability
PNAS | April 17, 2012 | vol. 109 | no. 16 | 6179
to form monocyte derived DC populations. Taken as a whole, the
exact DC-subset and source of P3 remains uncertain, as our efforts
to identify the origin and source of the P3 population were con-
founded by those cells being completely absent from VSV-infec-
ted irradiation chimeras. We suspect this might be because of P3
representing a radiosensitive population, the source of which is
some other organ (e.g., spleen or liver) that is not immediately
replenished by bone marrow. However, an alternate hypothesis is
that the absence of P3 is because of a lingering proinammatory
condition caused by the lethal irradiation (43).
In conclusion, we identied immune cells acting as sentinels at
vulnerable points of the CNS. These cells have the capacity to
orchestrate adaptive immune responses in the face of a viral
pathogen and demonstrate characteristics of DC. Furthermore,
our data provide insight by building upon our previous obser-
vations of CD11c+ cells with a dendritic morphology to show
a heterogeneous group of immune cells with complex immuno-
logical functions. Our hope is that further dissection of the in-
teraction between bDC and disease, resulting from viral stimuli
or in other present day models of neurodegenerative diseases,
may ultimately lead to the identication of therapeutic targets
for ameliorating neuro-inammatory diseases.
Materials and Methods
Viral Infection Model. VSV, Indiana Serotype (ATCC VR-158) and rVSV-OVA
(36) were isolated from infected BHK (ATCC CCL-10) supernatants and pu-
ried on a 540% sucrose gradient. Endotoxin levels were determined using
1. Steinman RM, Idoyaga J (2010) Features of the dendritic cell lineage. Immunol Rev
234:517.
2. Steinman RM, et al. (2003) Dendritic cell function in vivo during the steady state: A
role in peripheral tolerance. Ann N Y Acad Sci 987:1525.
3. Steinman RM, Hawiger D, Nussenzweig MC (2003) Tolerogenic dendritic cells. Annu
Rev Immunol 21:685711.
4. Steinman RM, Cohn ZA (1973) Identication of a novel cell type in peripheral lymphoid
organs of mice. I. Morphology, quantitation, tissue distribution. J Exp Med 137:11421162.
5. Steinman RM, Witmer MD (1978) Lymphoid dendritic cells are potent stimulators of
the primary mixed leukocyte reaction in mice. Proc Natl Acad Sci USA 75:
51325136.
6. Nussenzweig MC, Steinman RM (1980) Contribution of dendritic cells to stimulation of
the murine syngeneic mixed leukocyte reaction. J Exp Med 151:11961212.
7. Shortman K, Liu YJ (2002) Mouse and human dendritic cell subtypes. Nat Rev Immunol
2:151161.
8. Dudziak D, et al. (2007) Differential antigen processing by dendritic cell subsets
in vivo. Science 315:107111.
9. Merad M, Ginhoux F, Collin M (2008) Origin, homeostasis and function of Langerhans
cells and other langerin-expressing dendritic cells. Nat Rev Immunol 8:935947.
10. Iwasaki A (2007) Mucosal dendritic cells. Annu Rev Immunol 25:381418.
11. Ginhoux F, et al. (2009) The origin and development of nonlymphoid tissue CD103+
DCs. J Exp Med 206:31153130.
12. Choi JH, et al. (2009) Identication of antigen-presenting dendritic cells in mouse
aorta and cardiac valves. J Exp Med 206:497505.
13. Liu YJ (2005) IPC: Professional type 1 interferon-producing cells and plasmacytoid
dendritic cell precursors. Annu Rev Immunol 23:275306.
14. Sung SS, et al. (2006) A major lung CD103 (alphaE)-beta7 integrin-positive epithelial
dendritic cell population expressing Langerin and tight junction proteins. J Immunol
176:21612172.
15. Johansson-Lindbom B, et al. (2005) Functional specialization of gut CD103+ dendritic
cells in the regulation of tissue-selective T cell homing. J Exp Med 202:10631073.
16. Jaensson E, et al. (2008) Small intestinal CD103+ dendritic cells display unique func-
tional properties that are conserved between mice and humans. J Exp Med 205:
21392149.
17. Bulloch K, et al. (2008) CD11c/EYFP transgene illuminates a discrete network of
dendritic cells within the embryonic, neonatal, adult, and injured mouse brain.
J Comp Neurol 508:687710.
18. Anandasabapathy N, et al. (2011) Flt3L controls the development of radiosensitive
dendritic cells in the meninges and choroid plexus of the steady-state mouse brain.
J Exp Med 208:16951705.
19. Gottfried-Blackmore A, et al. (2009) Acute in vivo exposure to interferon-gamma
enables resident brain dendritic cells to become effective antigen presenting cells.
Proc Natl Acad Sci USA 106:2091820923.
20. Felger JC, et al. (2010) Brain dendritic cells in ischemic stroke: Time course, activation
state, and origin. Brain Behav Immun 24:724737.
21. Hickey WF, Kimura H (1988) Perivascular microglial cells of the CNS are bone marrow-
derived and present antigen in vivo. Science 239:290292.
22. Bi Z, Barna M, Komatsu T, Reiss CS (1995) Vesicular stomatitis virus infection of the central
nervous system activates both innate and acquired immunity. J Virol 69:64666472.
23. Vogel SN, Fertsch D (1987) Macrophages from endotoxin-hyporesponsive (Lpsd) C3H/
HeJ mice are permissive for vesicular stomatitis virus because of reduced levels of
6180 | www.pnas.org/cgi/doi/10.1073/pnas.1203941109
a ToxinSensor Chromogenic LAL Endotoxin Assay Kit (GenScript). See SI
Materials and Methods for intranasal administration procedure.
Flow Cytometry and FACS. After 96 hpi, mice were rapidly decapitated, brains
extracted, and placed in ice-cold HBSS. OB were subsequently removed and
single cell suspensions obtained by digesting with collagenase D (Roche),
dispase II (Roche), and DNase I (Invitrogen) in conjunction with repeated
trituration using a GentleMACS Dissociator (Miltenyi Biotec) and serological
pipetting. See SI Materials and Methods for antibodies, ow cytometry
staining procedure, and spleen DC isolation.
Antigen Presentation Assays. Because of the limited number of viable bDC
collected per OB, DC to T-cell ratios were titered, alongside antigen con-
centrations and T-cell plating number, to maximize both proliferation sen-
sitivity and sample replicates within each experiment. See SI Materials and
Methods for details about the antigen presentation assay.
See SI Materials and Methods for additional details on the mouse models,
antibodies, immunouorescence, radiation chimeras, EdU incorporation
assays, and the statistics used in this article.
ACKNOWLEDGMENTS. We thank Drs. Carol Shoshkes Reiss, Margaret
MacDonald, Charles Rice, and Leo Lefranois for providing useful reagents;
Drs. Juliana Idoyaga, Andres Gottfried-Blackmore, Elizabeth Waters, Ulrike
W. Kaunzner, Cheolho Cheong, and Jae-Hoon Choi for their valuable rec-
ommendations; and The Rockefeller Universitys Comparative Bioscience
Center, Flow Cytometry Resource Center, and the Bio-Imaging Resource
Center for their expert technical assistance. This research is supported by a
gift from the Peter Deane Trust (to K.B.).
endogenous interferon: possible mechanism for natural resistance to virus infection.
J Virol 61:812818.
24. Rodriguez Boulan E, Sabatini DD (1978) Asymmetric budding of viruses in epithelial
monlayers: A model system for study of epithelial polarity. Proc Natl Acad Sci USA 75:
50715075.
25. Grodums EI (1976) Ultrastructure of mouse periventricular and choroid plexus tissues
in experimental vesicular stomatitis virus infection. Arch Virol 51:7585.
26. Lundh B, Kristensson K, Norrby E (1987) Selective infections of olfactory and re-
spiratory epithelium by vesicular stomatitis and Sendai viruses. Neuropathol Appl
Neurobiol 13:111122.
27. Huneycutt BS, et al. (1994) Distribution of vesicular stomatitis virus proteins in the
brains of BALB/c mice following intranasal inoculation: An immunohistochemical
analysis. Brain Res 635:8195.
28. Raivich G, et al. (1999) Neuroglial activation repertoire in the injured brain: Graded
response, molecular mechanisms and cues to physiological function. Brain Res Brain
Res Rev 30:77105.
29. Lindquist RL, et al. (2004) Visualizing dendritic cell networks in vivo. Nat Immunol 5:
12431250.
30. Nimmerjahn A, Kirchhoff F, Helmchen F (2005) Resting microglial cells are highly
dynamic surveillants of brain parenchyma in vivo. Science 308:13141318.
31. Santambrogio L, et al. (2001) Developmental plasticity of CNS microglia. Proc Natl
Acad Sci USA 98:62956300.
32. Mildner A, et al. (2007) Microglia in the adult brain arise from Ly-6ChiCCR2+ mono-
cytes only under dened host conditions. Nat Neurosci 10:15441553.
33. Getts DR, et al. (2008) Ly6c+ inammatory monocytes are microglial precursors
recruited in a pathogenic manner in West Nile virus encephalitis. J Exp Med 205:
23192337.
34. Greter M, et al. (2005) Dendritic cells permit immune invasion of the CNS in an animal
model of multiple sclerosis. Nat Med 11:328334.
35. Sedgwick JD, et al. (1991) Isolation and direct characterization of resident microglial
cells from the normal and inamed central nervous system. Proc Natl Acad Sci USA 88:
74387442.
36. Kim SK, et al. (1998) Generation of mucosal cytotoxic T cells against soluble protein by
tissue-specic environmental and costimulatory signals. Proc Natl Acad Sci USA 95:
1081410819.
37. Barr JN, Whelan SP, Wertz GW (2002) Transcriptional control of the RNA-dependent
RNA polymerase of vesicular stomatitis virus. Biochim Biophys Acta 1577:337353.
38. Chan CW, et al. (2006) Interferon-producing killer dendritic cells provide a link be-
tween innate and adaptive immunity. Nat Med 12:207213.
39. Taieb J, et al. (2006) A novel dendritic cell subset involved in tumor im-
munosurveillance. Nat Med 12:214219.
40. McGavern DB, Kang SS (2011) Illuminating viral infections in the nervous system. Nat
Rev Immunol 11:318329.
41. Geissmann F, et al. (2008) Blood monocytes: Distinct subsets, how they relate to
dendritic cells, and their possible roles in the regulation of T-cell responses. Immunol
Cell Biol 86:398408.
42. Geissmann F, et al. (2010) Development of monocytes, macrophages, and dendritic
cells. Science 327:656661.
43. Linard C, et al. (2004) Acute induction of inammatory cytokine expression after
gamma-irradiation in the rat: effect of an NF-kappaB inhibitor. Int J Radiat Oncol
Biol Phys 58:427434.
DAgostino et al.