INTRODUCTION
The central nervous system (CNS) is normally protected from pathogens by the blood-brain
barrier (BBB), which is composed of brain microvascular endothelial cells (BMECs)
joined by tight junctions (TJs). TJs are formed by the intercellular association of
transmembrane proteins, including claudins and occludins, whose anchoring to the endothelial
cytoskeletal network by adaptor proteins, including the zonula occludens family, represents
a major regulatory mechanism by which paracellular permeability is controlled (1).
Viruses may cross the BBB via several routes, including direct infection of BMECs,
trans- or paracellular viral trafficking across the endothelium (2), or entry of infected
peripheral leukocytes (3, 4). Trafficking of virus across the BBB is likely made possible
by enhanced permeability of the BBB endothelium, caused either directly by viral factors
or indirectly by host immune factors, including innate cytokines such as tumor necrosis
factor alpha (TNF-α) and interleukin-1β (IL-1β) (5, 6). However, the mechanistic explanation
of this effect and the functions of innate immune responses at the BBB interface during
viral infections have not been explored.
West Nile virus (WNV), a mosquito-borne flavivirus, is the most common cause of epidemic
viral encephalitis in the United States (7). Following infection, the pattern recognition
receptors (PRRs) TLR3, TLR7, RIG-I, and MDA5 are key sensors of WNV pathogen-associated
molecular patterns (PAMPs), inducing the expression of innate Th1 cytokines that limit
viral replication, facilitate antigen presentation, and direct the trafficking of
antiviral leukocytes (8, 9). Studies with mice have shown that TNF-α signaling is
key for CD8+ T cell-mediated clearance of WNV from the CNS (10), as well as protection
from CXCR3-mediated apoptosis in neurons (11). Likewise, IL-1β signaling has been
shown to limit WNV replication in neurons (12) and is essential for dendritic-cell-mediated
reactivation of WNV-specific T cells in the CNS (13). Despite these beneficial effects
in the CNS, Th1 cytokine signaling can also contribute to neuropathology, including
disruption of the BBB (6, 14).
PRR activation also induces the expression of type I interferons (IFN-α and IFN-β)
that share a type I IFN receptor (IFNAR). PRR-mediated expression of IFN ligands results
in the amplification of type I IFN signals via both autocrine and paracrine signaling
through IFNAR, which stimulates further IFN ligand expression and an array of IFN-stimulated
antiviral and immunomodulatory genes (15). Type I IFNs act directly as potent suppressors
of viral replication (16) and indirectly by promoting host adaptive immune responses
(17). Thus, mice lacking type I IFN receptors, ligands, and key transcription factors
exhibit expanded viral tropism and replication, resulting in accelerated and enhanced
death (16, 18, 19). In the context of CNS autoimmunity, type I IFNs can enhance BBB
function and are used clinically in humans to prevent the CNS entry of autoreactive
leukocytes in patients with multiple sclerosis (20, 21). However, the functions of
type I IFN at the BBB interface during viral infections of the CNS have not yet been
explored.
Here, we demonstrate a novel mechanism underlying the dynamic regulation of BBB structure
and function by innate cytokines during viral encephalitis. Using an in vitro BBB
model, we show that, in the presence of WNV, the induction of type I IFN directly
regulates endothelial permeability and TJ formation via regulation of the small GTPases
Rac1 and RhoA and indirectly via suppression of barrier-dysregulating effects of TNF-α
and IL-1β. This regulatory regimen modulates transendothelial WNV trafficking, as
type I IFN responses significantly decreased the movement of virus across an intact
barrier in vitro, while TNF-α and IL-1β increased this crossing. In vivo, genetic
attenuation of type I IFN signaling resulted in enhanced BBB permeability with associated
TJ disruption following peripheral or intracranial (i.c.) inoculation with WNV. Together,
these data demonstrate an additional, previously undescribed, regulatory role for
PRR-mediated cytokine expression at the BBB during WNV infection.
RESULTS
Innate cytokines differentially regulate barrier function following WNV infection.
To assess the kinetics of BBB permeability following murine WNV infection, we utilized
an established model of WNV neuroinvasive disease in which 5-week-old C57BL/6 mice
are subcutaneously infected in the footpad with 102 PFU of a virulent strain of WNV-NY.
Five-week-old mice uniformly develop CNS infection, with detectable CNS viral loads
first emerging ~6 days following infection (22, 23) (see Fig. S1A in the supplemental
material). Fluorometric assessment of BBB permeability following the intraperitoneal
(i.p.) administration of sodium fluorescein revealed that BBB permeability increased
significantly over the course of peripheral infection (days 0 to 4) in all of the
CNS regions analyzed (Fig. 1A). Of interest, BBB permeability exhibited a significant
decrease at day 6 in cortical and subcortical cerebral tissues, coincident with the
timing of detection of virus within the CNS (see Fig. S1A). After this, BBB permeability
rebounded and returned to elevated levels on day 8, coinciding with increased trafficking
of leukocytes into the CNS (22).
FIG 1
WNV infection modulates endothelial barrier integrity via innate cytokine signaling.
(A) Five-week-old C57BL/6 mice were inoculated in the footpad with 100 PFU of WNV
and then assessed for BBB sodium fluorescein permeability in CNS regions on the postinfection
days indicated. The values reported are arbitrary fluorescence values in CNS tissue
normalized to values in serum for individual mice. Group means are normalized to the
mean value for uninfected animals. The values on the x axes are times in days. n.s.,
not significant. (B) In vitro BBBs were treated overnight in the top chamber with
the saline vehicle, TNF-α, IFN-γ, IL-1β, or IFN-β, followed by an additional 6 h of
infection with WNV and subsequent measurement of TEER. (C) BBB cultures were treated
for 2 h with the vehicle or the cytokine indicated, and then the culture medium was
washed away and replaced with medium containing the vehicle or IFN-β for 2 h, followed
by measurement of TEER. (D) In vitro BBBs were constructed with WT or Ifnar
−/− BMECs cocultured with astrocytes of either genotype, as shown on the x axis, with
the BMEC genotype on the left and the astrocyte genotype on the right. TEER measurements
after 6 h of WNV infection are shown. (E) ELISA of cytokine expression in the top
chambers of BBB cultures following 6 h of WNV infection. (F) Quantification of ICC
analysis of colocalization of Zo-1 and Claudin-5 in WT and Ifnar
−/− BMEC monolayers treated for 2 h with the saline vehicle or the cytokine indicated
and subsequently infected for 6 h with WNV at an MOI of 0.01. Values on the y axis
(correlation indexes) represent the probability of an individual pixel staining positive
for both markers if it stains positive for either. (G) Representative ×63 images of
cultures with Zo-1 shown in green, Claudin-5 in red, and Topro3 nuclear staining in
blue. Scale bar = 25 µm. White arrowheads highlight junctions with loss of TJ protein
colocalization. Veh, vehicle.
The biphasic kinetics of BBB permeability after WNV infection suggested potentially
differential regulation of the BBB by viral and host factors over the course of infection.
To assess this, we utilized a standard Transwell in vitro BBB system in which primary
murine BMECs are grown on a porous filter membrane in a chamber suspended above primary
murine astrocytes. The integrity of the endothelial barrier is assessed via electrode
recording of transendothelial electrical resistance (TEER). We treated in vitro BBB
cultures in the top (BMEC-only) chamber overnight with cytokines or infected them
for 6 h at a multiplicity of infection (MOI) of 0.01 with WNV that had been purified
via ultracentrifugation through a sucrose gradient. We utilized a low MOI for BMEC
experiments in light of the relatively low viremia present in mammalian hosts during
WNV infection (24) and the fact that high MOIs can trigger fast necrotic cell death
(25), compromising monolayer integrity and confounding the interpretation of TEER
results. Similarly, 6-h infections were chosen because this period of time is insufficient
for completion of the viral life cycle and induction of apoptosis in BMECs, as assessed
via multistep growth curves and terminal deoxynucleotidyltransferase-mediated dUTP-biotin
nick end labeling (TUNEL) staining (see Fig. S1B and C in the supplemental material).
Treatment with Th1 cytokines TNF-α (100 ng/ml), IL-1β (100 ng/ml), and IFN-γ (100 ng/ml)
decreased TEER, while IFN-β (100 pg/ml) and WNV infection both significantly enhanced
TEER (Fig. 1B). TEER effects induced by Th1 cytokines could be rescued by subsequent
infection with WNV, while infection of IFN-β-treated cultures produced no additional
increase in TEER (Fig. 1B). Similar to WNV, addition of IFN-β to cultures pretreated
with Th1 cytokines also rescued TEER (Fig. 1C). These data suggest that WNV infection
may increase TEER either via type I IFN or through convergent mechanisms.
To assess whether type I IFN expression by BMECs and/or astrocytes contributes to
increasing TEER, we performed checkerboard experiments with BMECs and astrocytes isolated
from wild-type (WT) and/or Ifnar
−/−
mice. While TEER increased after 6 h of infection in cultures with WT BMECs, similarly
infected cultures generated with Ifnar
−/−
BMECs instead exhibited significant reductions in TEER (Fig. 1D). While type I IFN
signaling in BMECs robustly controlled TEER responses after infection, type I IFN
signaling in astrocytes produced smaller but significant modulations of TEER responses
as well. Experiments with neutralizing antibodies to IFNAR and IFN ligands recapitulated
the results obtained with Ifnar
−/−
BMECs (data not shown). Given the dramatic change in TEER responses in the absence
of type I IFN signaling, we next assessed the expression of innate cytokines in BMECs
following infection. In WT BBB cultures, the expression of both IFN-β and TNF-α is
induced within 6 h in the top chambers of BBB cultures, with mostly undetectable levels
of IL-1β. However, infection of Ifnar
−/− BBB cultures yielded lower levels of type I IFNs (Fig. 1E; see Fig. S2A and B
in the supplemental material), consistent with a loss of IFNAR-mediated amplification,
as well as modestly enhanced TNF-α and more robustly enhanced IL-1β expression (Fig. 1E).
Baseline expression of cytokines in uninfected cultures did not differ between genotypes
(see Fig. S2), and cytokine expression in bottom culture chambers did not differ substantially
from that in top culture chambers (data not shown). These data indicate that type
I IFN signaling in BMECs can modulate permeability both directly and indirectly through
the suppression of Th1 cytokine expression.
To determine if permeability changes are associated with structural changes involving
TJs, BMEC monolayers treated overnight with cytokines, followed by 6 h of WNV infection,
were evaluated by immunocytochemistry (ICC) for the TJ proteins Zo-1 and Claudin-5.
Infection of WT cells enhanced TJ protein colocalization at intercellular borders,
indicating the recruitment and stabilization of these proteins in functional TJ complexes
(Fig. 1F and G). In contrast and consistent with TEER responses, infection of Ifnar
−/− BMECs resulted in decreased TJ protein colocalization, suggesting that type I
IFN is necessary to enhance TJ formation in the presence of WNV. Also mirroring our
prior TEER results, TNF-α or IL-1β treatment resulted in decreased TJ protein colocalization,
which could be rescued by subsequent WNV infection in WT, but not Ifnar
−/−, BMECs. These data indicate that innate cytokine signaling in BMECs following
exposure to WNV results in the rapid modulation of TJ complexes, which are key regulators
of paracellular permeability, and that barrier dysregulation by Th1 cytokines can
be controlled by the actions of type I IFN.
PRR activation is sufficient to induce cytokine-dependent changes in endothelial barrier
integrity.
As type I IFN signaling enhances both TJ formation and BBB permeability after WNV
infection, we assessed whether this was the result of classical innate immune signaling
through PRRs. First, we inactivated purified WNV via either 2 h of exposure to UV
light (WNV-UV) or incubation with 0.1% (vol/vol) β-propiolactone (WNV-BPL), generating
viral stocks that are unable to replicate within cells, as assessed via standard plaque
assay (data not shown), but do induce the expression of both IFN-β and IFN-α in BMECs
after 6 h (see Fig. S2A and B). Treatment of in vitro BBB cultures consisting of WT
versus Ifnar
−/− BMECs (both over WT astrocytes) with either WNV-UV or WNV-BPL at an MOI of 0.01
(Fig. 2A) for 6 h revealed enhanced TEER in WT cultures but reduced TEER in Ifnar
−/− cultures, largely phenocopying the effect of normal virus. These results suggest
that type I IFN-dependent changes in TEER require viral detection but not replication
in BMECs.
FIG 2
PRR activation is sufficient to induce cytokine-dependent changes in endothelial barrier
integrity. (A, left side) TEER measurements for in vitro BBBs constructed with either
WT or Ifnar
−/− BMECs over WT astrocytes, treated for 6 h at an MOI of 0.01 with WNV inactivated
with UV (WNV-UV), β-propiolactone (WNV-BPL), the TLR3 agonist “naked” poly(I · C),
the TLR7 agonist CL264, the MDA5-biased agonist HMW poly(I · C)-LyoVec, or the RIG-I
agonist 5′ ppp-dsRNA. (A, right side) Quantification of ICC analysis of colocalization
of Zo-1 and Claudin-5 in WT versus Ifnar
−/− BMECs treated as indicated on the left. Values on the x axis (correlation indexes)
represent the probability of an individual pixel staining positive for both markers
if it stains positive for either. (B) Representative ×63 images showing colocalization
of Zo-1 and Claudin-5 in WT versus Ifnar
−/− BMEC monolayers treated for 6 h with inactivated WNV or a PRR agonist (as in panel
A), with Zo-1 shown in green, Claudin-5 in red, and Topro3 nuclear staining in blue.
White arrowheads highlight junctions with loss of TJ protein colocalization. (C) ELISA
of cytokine production in the top chambers of BBB cultures as in panel A following
6 h of treatment with either “naked” poly(I · C) or CL264. (D) BBB cultures as in
panel A were pretreated with neutralizing antibodies to TNF-α and/or IL-1β or isotype
controls. Cultures were then infected for 6 h with WNV at an MOI of 0.01 (left panel)
or treated for 6 h with “naked” poly(I · C) or CL264 (middle and right panels). Data
are expressed as fold changes in infected/treated cultures versus similarly treated
mock-infected/vehicle-treated cultures under the same conditions.
We next determined if direct activation of PRRs was sufficient to alter TEER. TLR3
agonism via 6 h of treatment with naked (nontransfected) poly(I · C) at 1 µg/ml significantly
decreased TEER in both WT and Ifnar
−/− BBB cultures (Fig. 2A). However, similar 6-h treatments with agonists for TLR7
(CL264, 10 µg/ml), MDA5 [high-molecular-weight poly(I · C), 1 µg/ml, transfected into
cytoplasm], or RIG-I (5′ triphosphate double-stranded RNA [ppp-dsRNA], 1 µg/ml, transfected
into cytoplasm) instead resulted in increased TEER in WT cultures but decreased TEER
in Ifnar
−/− cultures, similar to the results obtained with both infectious and inactivated
WNV. Enzyme-linked immunosorbent assay (ELISA) results confirmed that the expression
of both IFN-β and IFN-α was induced by all of the treatments used, including a TLR3
agonist (see Fig. S2A and B). Consistent with these results, ICC analysis of Zo-1
and Claudin-5 similarly showed enhanced colocalization in WT BMECs treated for 6 h
with WNV-UV, WNV-BPL, and TLR7, MDA5, and RIG-I, but not TLR3, agonists, with decreases
in TJ protein colocalization after these treatments in Ifnar
−/− BMECs (Fig. 2A and B). TLR3 agonism, however, resulted in decreases in TJ protein
colocalization in both WT and Ifnar
−/− cultures.
We next assessed if differences in barrier responses to PRR agonists in WT versus
Ifnar
−/− BBB cultures and between TLR3 and other PRR agonists were due to differential
cytokine expression. Similar to experiments with WNV, Ifnar
−/− BBB cultures had small modulations of type I IFN and TNF-α signaling after stimulation
with inactivated virus and PRR agonists (see Fig. S2A to C) and, strikingly, significantly
enhanced expression of IL-1β compared to that in WT cultures (see Fig. S2D). Moreover,
TLR3 agonism resulted in IL-1β expression even in WT cultures (Fig. 2C), in contrast
to cultures treated with other PRR agonists, including the TLR7 agonist CL264, which
induced detectable levels of IL-1β only in Ifnar
−/− BBB cultures (Fig. 2C; see Fig. S2D). To determine if differential expression
of Th1 cytokines contributed to differences in TEER responses in BBB cultures exposed
to WNV or PRR agonists for 6 h, we measured TEER in the presence of cytokine-neutralizing
antibodies. Blockade of TNF-α augmented increases in TEER in WT cultures following
exposure to WNV or a TLR7 agonist (Fig. 2D) and partially rescued decreases in TEER
in Ifnar
−/− cultures. In contrast, blockade of IL-1β did not impact TEER in WNV and TLR7 agonist-treated
WT cultures, which do not express IL-1β within 6 h, but did partially rescue decreases
in TEER in Ifnar
−/− cultures, which do exhibit IL-1β expression. Similarly, blockade of either TNF-α
or IL-1β partially rescued TEER responses in TLR3 agonist-treated cultures of either
genotype (Fig. 2D). Of note, simultaneous blockade of both TNF-α and IL-1β in Ifnar
−/− cultures following exposure to WNV or PRR agonists held TEER levels at baseline
values. Together, these data suggest that innate cytokine signaling in BMECs orchestrates
barrier responses following exposure to WNV PAMPs and that the relative balance of
type I IFN versus TNF-α and IL-1β governs these changes. Type I IFN signaling in BMECs
is able to prevent BBB dysregulation by Th1 cytokines, which, in the absence of type
I IFN, are able to increase barrier permeability and disassemble TJs. Moreover, PRR
activation in BMECs, even in the absence of direct infection, is sufficient to induce
cytokine-mediated changes in BBB physiology.
WNV modulates BMEC Rho GTPase signaling in an innate cytokine-dependent manner.
To assess the signaling events that regulate BBB physiology downstream of WNV and
cytokine signaling, we examined if WNV infection resulted in changes in the activity
of Rho GTPases, a family of signaling molecules that regulate cytoskeletal dynamics,
including TJs and paracellular permeability (26). WT and Ifnar
−/− BMEC monolayers were infected for 6 h with WNV, and protein lysates were collected
and incubated with Rhotekin or p21-activated protein kinase (PAK) agarose beads to
pull down GTP-bound, activated Rac1, RhoA, and CDC42. Western blot detection of activated
forms of these GTPases revealed a significant increase in activated Rac1 in WNV-infected
WT BMECs (Fig. 3A and B) and a significantly smaller increase in activated Rac1 in
Ifnar
−/− BMECs. In contrast, while activated RhoA was not increased in WNV-infected WT
BMECs, Ifnar
−/− BMECs exhibited significantly increased amounts of activated RhoA (Fig. 3A and
C). Levels of activated CDC42 were not affected by WNV infection in either WT or Ifnar
−/− BMEC cultures (Fig. 3A and D). As these experiments suggested a dynamic range
of Rho GTPase activation responses following WNV infection, we next assessed the manner
in which individual innate cytokines regulate Rho GTPase activation. In agreement
with infection experiments with WT versus Ifnar
−/− BMECs, 2 h of incubation of WT BMECs with recombinant IFN-β resulted in significantly
enhanced Rac1 activation (Fig. 3E and F) and significantly diminished RhoA activation
(Fig. 3E and G). In contrast, 2 h of incubation with TNF-α resulted in decreased Rac1
activation and greater activation of RhoA, while 2 h of IL-1β treatment increased
the activation of both GTPases, with a greater fold change in induction of RhoA activation.
FIG 3
WNV modulates endothelial Rho GTPase-mediated regulation of the BBB endothelium. (A
to G) Activated GTPase pulldown assay of WT or Ifnar
−/− BMECs either infected with WNV at an MOI of 0.01 for 6 h (A to D) or treated for
2 h with recombinant TNF-α, IL-1β, or IFN-β (E to G). After treatment, BMEC lysates
were incubated with Rhotekin or PAK agarose beads to purify GTP-bound, activated Rac1,
RhoA, and CDC42. Both purified activated GTPases and total unpurified cell lysates
were then probed via Western blot assay to assess activated versus total amounts of
each GTPase. UI, uninfected. (B to D, F, G) Density quantification of activated Rac1
(B, F), RhoA (C, G), or CDC42 (D), normalized to the total amount of each protein
in each sample. Group averages are normalized to the mean values of uninfected/vehicle-treated
WT BMECs. (H) TEER measurements for in vitro BBBs constructed with WT or Ifnar
−/− BMECs over WT astrocytes, pretreated for 2 h with Rac1 inhibitor Z62954982, ROCK/RhoA
inhibitor H1152P, or CDC42 inhibitor ML141 and then infected for 6 h with WNV at an
MOI of 0.01. Data are reported as fold changes in TEER in infected cultures versus
uninfected cultures within each treatment group. (I) Representative ×63 images showing
colocalization of Zo-1 and Claudin-5 in WT versus Ifnar
−/− BMEC monolayers treated for 2 h with GTPase inhibitors (as in panel H), followed
by 6 h of infection with WNV at an MOI of 0.01, with Zo-1 shown in green, Claudin-5
in red, and Topro3 nuclear staining in blue. White arrowheads highlight junctions
with loss of TJ protein colocalization. (J) Quantification of colocalization in images
in panel I. Data are reported as the fold changes in colocalization in infected cells
versus uninfected cells within each treatment group.
Innate cytokine-dependent modulation of Rac1 and RhoA after WNV infection regulates
BBB permeability and TJ formation.
We next examined whether modulation of BMEC Rho GTPase signaling in the context of
WNV infection contributes to type I IFN-dependent regulation of TEER and TJ responses.
Pharmacological blockade of Rac1 signaling via 2 h of pretreatment with its specific
inhibitor, Z62954982, abolished the increase in TEER after 6 h of WNV infection in
WT cells, with TEER values significantly decreased in WNV-infected WT barriers in
the absence of Rac1 signaling, while responses remain unchanged in similarly infected
and treated Ifnar
−/− barriers (Fig. 3H). Conversely, inhibition of RhoA signaling via 2 h of pretreatment
with its specific inhibitor, H1152P, did not impact TEER changes induced by WNV infection
in WT barriers but significantly rescued reductions in TEER in similarly infected
and treated Ifnar
−/− barriers. Inhibition of CDC42 via 2 h of pretreatment with ML141, a specific inhibitor
of CDC42, did not significantly alter TEER in either WT or Ifnar
−/− barriers after WNV infection. As changes in TEER responses could simply be explained
by altered IFN expression by BMECs following infection in the presence of GTPase inhibitors,
we confirmed via ELISA that inhibition of Rac1 or RhoA did not significantly alter
IFN-β levels in luminal chambers following 6 h of infection and that none of the inhibitors
impacted TEER in cultures mock infected for 6 h (see Fig. S3A to C in the supplemental
material).
Consistent with analyses of TEER, ICC analysis further revealed that inhibition of
Rac1 signaling in WT cells completely abolished the increase in the colocalization
of Zo-1 and Claudin-5 observed in WNV-infected WT BMECs (Fig. 3I and J), while inhibition
of RhoA or CDC42 did not impact TJ formation in WT cells. However, inhibition of RhoA
rescues the loss of TJ formation in WNV-infected Ifnar
−/− BMECs, abolishing decreases in TJ protein colocalization observed in untreated
Ifnar
−/− BMECs. Inhibition of Rac1 or CDC42 did not significantly impact TJ formation in
Ifnar
−/− BMECs. Control experiments also confirmed that 2 h of pretreatment with GTPase
inhibitors, followed by 6 h of mock infection, did not change baseline levels of TJ
protein colocalization (see Fig. S3D). Together, these data suggest that cytokines
expressed in BMECs following exposure to WNV differentially influence Rho GTPase activity,
with type I IFN contributing to the preferential activation of Rac1, thereby stabilizing
barrier physiology. This regulation counteracts the activity of the cytokines TNF-α
and IL-1β, which preferentially activate RhoA and contribute to barrier disruption.
Innate cytokine regulation of Rho GTPase signaling controls WNV transmigration across
the in vitro BBB.
We next determined if cytokine and Rho GTPase modulation of TEER and TJ integrity
are relevant regulators of viral trafficking across the BBB, limiting the access of
WNV to the CNS. For these experiments, WNV at an MOI of 0.01 was added to the top
chamber of Transwell BBB cultures; 6 h later, the top chamber was removed, such that
only virus that had crossed during the 6 h of incubation was present in the bottom
chamber. Amounts of migrated virus in the bottom chambers were then assessed via multistep
growth curve analysis. Consistent with changes in TEER and TJ integrity, overnight
pretreatment of the top chamber with TNF-α or IL-1β before infection significantly
enhanced the transendothelial migration of WNV (Fig. 4A and B), while IFN-β pretreatment
of the top chamber significantly reduced viral trafficking (Fig. 4C). Similarly, experiments
with barriers with WT versus Ifnar
−/− BMECs over WT astrocytes revealed that Ifnar
−/− BMECs exhibit significantly enhanced viral migration (Fig. 4D). In order to control
for the possibility that IFN-β treatment or genetic deletion of IFNAR on BMECs alters
the basolateral secretion of type I IFN by BMECs, which may skew the kinetics of replication
in the astrocytes below, we did concurrent experiments with BBB cultures with Ifnar
−/− astrocytes. These experiments yielded similar results, confirming that differential
exposure of astrocytes to IFN did not account for the difference observed in experiments
with WT astrocytes (see Fig. S4A and B in the supplemental material). Pretreatment
of BBB cultures with neutralizing antibodies to TNF-α reduced viral migration in both
WT Ifnar
−/− cultures, while pretreatment with neutralizing antibodies to IL-1β reduced migration
only in Ifnar
−/− cultures, consistent with similar experiments measuring TEER responses (Fig. 4E
and F). Once again, neutralizing antibodies to both cytokines had additive effects
on Ifnar
−/− barriers, suppressing viral migration back to the levels observed in WT cultures.
FIG 4
Innate cytokines differentially regulate WNV transmigration across the in vitro BBB.
(A to H) Detection of replicating virus in the bottom chambers of in vitro BBBs constructed
with the WT BMECs only (A to C) or WT or Ifnar
−/− BMECs (D to H) over WT astrocytes, infected with WNV at an MOI of 0.01 in the
top chambers for 6 h before removal of the top chambers. The cultures in panels A
to C were treated overnight with the saline vehicle or TNF-α (A), IL-1β (B), or IFN-β
(C) prior to infection. The cultures in panels E and F were pretreated for 2 h with
neutralizing antibodies to TNF-α (E) or IL-1β (F) before infection. The cultures in
panels G and H were pretreated for 2 h with Rac1 inhibitor Z62954982 (G) or ROCK inhibitor
H1152P (H) before infection. Values are viral titers present after 0, 12, 24, and
48 h of replication by virus that crossed during the 6 h of infection, prior to the
removal of BMECs from the culture system. Viral titers are reported in plaque-forming
units per milliliter, as determined via standard plaque assay in BHK cells. Each horizontal
dotted line represents the limit of detection of the assay.
As cytokine signaling was observed to influence transendothelial viral migration,
we next performed similar experiments in which cultures consisting of WT or Ifnar
−/− BMECs over WT astrocytes were pretreated for 2 h in the top chamber with antagonists
of Rac1, RhoA, or CDC42 prior to the introduction of WNV. Again, consistent with TEER
and TJ integrity findings, antagonism of Rac1 significantly enhanced the transendothelial
migration of WNV across WT BMECs but did not further augment the enhanced migration
observed across Ifnar
−/− BMECs (Fig. 4G). Likewise, antagonism of RhoA did not significantly alter WNV
migration across WT BMECs but resulted in a partial rescue of the enhanced migration
observed in Ifnar
−/− BMECs (Fig. 4H). As expected antagonism of CDC42 did not impact the transendothelial
migration of WNV (see Fig. S4C). Together, these findings suggest that modulation
of BMEC TEER and TJs caused by type I IFN-dependent regulation of Rho GTPases is consequential
to the ability of WNV to traffic across the BBB endothelium.
Mice with impaired type I IFN responses exhibit enhanced BBB permeability and loss
of TJ integrity after peripheral WNV infection.
Mice lacking a type I IFN receptor, ligands, and regulatory transcription factors
exhibit enhanced susceptibility to WNV infection, including early entry of virus into
the CNS compared to WT controls. Thus, in light of our in vitro observations, we wondered
if mice with impaired type I IFN signaling would also exhibit alterations in BBB permeability
and TJ integrity following peripheral infection. To determine whether type I IFN signaling
is required for the in vivo decrease in BBB permeability observed on day 6 after WNV
infection of the CNS, we examined BBB permeability in the context of i.c. inoculation
(10 PFU) of WNV in 8-week-old WT versus Ifnar
−/− mice. WT animals experienced a significant decrease in BBB permeability to sodium
fluorescein on days 1 and 2 in each CNS region, followed by a large increase in BBB
permeability on day 4 (Fig. 5A), coincident with the detection of large numbers of
inflammatory cells at this time point in this model (27). However, these kinetics
were not observed in Ifnar
−/− animals, which instead exhibited a significant and sustained increase in permeability
to sodium fluorescein each day following infection. For these studies, we examined
permeability in tissues contralateral to the inoculation site, though we ultimately
found that permeability did not differ significantly between hemispheres on any day
following i.c. infection (data not shown). Concurrent experiments with mice that received
mock i.c. inoculations with saline showed only extremely modest increases in BBB permeability,
which did not differ between WT and Ifnar
−/− mice (see Fig. S5 in the supplemental material); thus, the inoculation procedure
alone did not explain changes in BBB permeability following i.c. inoculation with
WNV. In all, these data suggest that local type I IFN responses in the CNS following
WNV entry promote BBB integrity, which may contribute to the decrease in BBB permeability
observed between days 4 and 6 following peripheral infection.
FIG 5
WT mice but not mice with impaired type I IFN signaling exhibit increased BBB integrity
following neuroinvasion. (A) Eight-week-old WT and Ifnar
−/− mice were i.c. inoculated with 10 PFU WNV. On the days indicated after infection,
mice were administered sodium fluorescein and then sodium fluorescein levels in tissue
homogenates taken from CNS regions contralateral to the inoculation site were assessed.
The values reported are arbitrary tissue fluorescence values normalized to the serum
values of individual mice. Group means are normalized to the mean value of mock-infected
WT animals. (B) Sodium fluorescein permeability in CNS regions after footpad inoculation
with 100 PFU WNV. Values taken from CNS regions (both hemispheres) are reported as
in panel A, with normalization to values of uninfected WT mice. (C) IHC detection
of serum IgG in parenchymal CNS tissues after peripheral WNV infection, with quantification
of group mean fluorescence intensity per low-power field. Scale bar = 100 µm. (D)
IHC detection of Zo-1 (green) and Claudin-5 (red) in CNS microvessels on the days
after footpad infection indicated. Scale bar = 10 µm. (E) Eight-week-old WT and Ifnar
−/− mice were administered 50 µg poly(I · C) i.p., followed by sodium fluorescein
24 h later (processed as described for panel B).
As Ifnar
−/− mice rapidly and uniformly die of septic shock (at about 3.5 days after footpad
infection) (16, 28), we next used Irf7
−/− mice, which have impaired expression of IFN-α (29). These mice experience early
neuroinvasion of WNV compared to WT mice following footpad infection (29). Sodium
fluorescein permeability experiments revealed that WNV-infected 8-week-old WT mice
exhibit BBB permeability kinetics similar to those previously observed in 5-week-old
animals, including an increase in permeability during peripheral infection (days 0
to 4), followed by a decrease in permeability at day 6 (Fig. 5B). In contrast, WNV-infected
8-week-old Irf7
−/− mice exhibit a sustained increase in BBB permeability over the entire course of
infection, with significantly higher permeability than that of WT controls on day
6. Immunohistochemical (IHC) detection of extravasated endogenous IgG revealed similar
kinetics of permeability following WNV infection in cortical and cerebellar tissues
(Fig. 5C). Consistent with this, IHC detection of Zo-1 and Claudin-5 in CNS tissues
revealed that both WT and Irf7
−/− mice exhibited loss of TJ integrity on day 4 following WNV infection, characterized
by diminished or diffuse cytoplasmic staining of TJ proteins and substantially decreased
colocalization of TJ proteins at cell-cell borders (Fig. 5D). Alteration in TJ assembly
was reversed at day 6 within the microvasculature of WNV-infected WT mice, while that
of similarly infected Irf7
−/− mice showed sustained dysregulation of TJs. As differences between the viral burdens
of WT and Irf7
−/− mice over time complicate the interpretation of BBB changes due to type I IFN
signaling, we performed control experiments in which WT and Ifnar
−/− mice were treated with the TLR3 agonist poly(I · C), which is known to increase
BBB permeability (6). At 24 h following treatment, Ifnar
−/− mice exhibited greater CNS permeability to sodium fluorescein than WT mice (Fig. 5E),
providing further evidence that type I IFN signaling is required to preserve BBB integrity
in the context of systemic inflammatory activation.
DISCUSSION
While the contributions of innate cytokine responses to virologic control and adaptive
immunity during WNV infection are known, the regulatory potential of these molecules
at the BBB remains poorly understood. Here, we demonstrate that PRR activation in
the presence of WNV in the BBB endothelium results in cytokine-dependent modulation
of Rho GTPase signaling, exerting regulatory control over BBB permeability and TJ
integrity. This, in turn, is consequential for the ability of WNV to traffic across
the BBB endothelium. In particular, our data suggest that the induction of type I
IFN serves to preserve and stabilize BBB structure and function both directly and
indirectly by limiting barrier disruption caused by the cytokines TNF-α and IL-1β.
In vivo, BBB permeability increases during peripheral infection but sharply decreases
coincident with viral neuroinvasion in WT animals but not in animals with attenuated
type I IFN responses, suggesting that local CNS type I IFN responses may act on the
BBB to mitigate the access of WNV to the CNS parenchyma.
The Th1 cytokines TNF-α and IL-1β have previously been shown to disrupt barrier endothelium
in vitro and in vivo, increasing permeability (30
–
32) and dysregulating junction complexes (1, 33, 34). During viral infections, TNF-α,
in particular, has been linked to enhanced vascular permeability, poor tissue perfusion,
and endothelial cell death (28, 35). Here, we show that TNF-α signaling in BMECs activates
RhoA, disrupting barrier function and increasing the transendothelial movement of
WNV. This is in line with the study of Wang et al. (6), who reported a decrease in
BBB permeability in Tnfrsf1a
−/− mice following WNV infection. These findings together indicate that Th1 cytokine-mediated
disruption of the BBB provides an access route for neuroinvasion by WNV.
In contrast to Th1 cytokines, type I IFN treatment has been established to promote
BBB function (34, 36). However, for the first time, we demonstrate that in vivo induction
of type I IFN following WNV infection is a key regulator of BBB permeability and TJ
integrity. Type I IFN signaling in BMECs in vitro is shown to enhance barrier function
over baseline conditions and to rescue Th1 cytokine-mediated barrier dysregulation,
suggesting that CNS type I IFN responses may be able to reverse the BBB permeability
caused by Th1 cytokine production during infection in vivo. This interpretation agrees
with in vitro observations suggesting that dengue virus activation of type I IFNs
reduces vascular leakage in peripheral endothelium, even in the presence of the barrier-disrupting
cytokines (35). Type I IFN also preserves barrier integrity by suppression of Th1
cytokine expression, most noticeably, that of IL-1β. While the interactions of type
I IFN and IL-1β during WNV infection vary across tissues and times after infection
(12, 13), type I IFN has been shown to suppress IL-1β expression via mechanisms that
vary by cell type and inflammatory stimulus, including inhibition of inflammasome
activity and pro-IL-1α and pro-IL-1β expression (37, 38). To our knowledge, we are
the first to demonstrate IFN-mediated suppression of endogenous IL-1β in the BBB endothelium.
While this report and others have identified a role for IL-1 signaling in BBB breakdown
(39, 40), we note our previous study showing that IL-1 signaling at the BBB is an
important facilitator of lymphocyte activation and trafficking into the CNS during
WNV infection and was crucial for viral clearance and survival (13). Moreover, Ramos
et al. (12) found that type I IFNs and IL-1β worked synergistically to control WNV
in neurons. Thus, IL-1 signaling during WNV infection may serve both pathological
and protective functions, and the modulation of these effects by type I IFN likely
varies over time and by cell type.
Some studies of murine WNV encephalitis have reported a sustained increase in BBB
permeability over the course of infection, with permeability rising coincident with
neuroinvasion and continuing to increase thereafter (41, 42). However, our findings
more closely resemble those showing markedly enhanced BBB permeability 3 to 4 days
postinfection, prior to neuroinvasion (6, 43, 44). Wang et al. (6) additionally observed
reduced BBB permeability on day 5 postinfection, coincident with neuroinvasion, in
concordance with our findings. A similar biphasic response in BBB permeability was
also observed in a model of Venezuelan equine encephalitis virus infection (45). The
reasons for these discrepancies are unclear and may include variations in the viral
inoculation site and the age of the experimental animals used. Nevertheless, our data
suggest that type I IFN suppresses BBB permeability during viral infection, as loss
of intact type I IFN signaling enhanced BBB permeability following both peripheral
and CNS WNV inoculations. While tissue viral burden and immune activation differences
between Ifnar
−/− and Irf7
−/− mice after peripheral infection may independently contribute to the observed differences
in BBB permeability, we note that, following i.c. inoculation, BBB permeability was
demonstrated to be significantly higher in Ifnar
−/− animals, even 1 day after infection when CNS viral titers have been shown to be
equivalent (16).
Cytokine-mediated regulation of the BBB is likely to arise from many diverse sources,
including peripheral/serum cytokines, CNS cytokines produced by resident and infiltrating
cells, as well as cytokines produced by the BBB endothelium that signal in an autocrine
or paracrine fashion among BMECs. Here, we have focused our investigation specifically
on BMEC-intrinsic cytokine signals, though our data and previous studies provide some
clues as to how cytokine signaling from each of these sources may orchestrate the
observed changes in BBB permeability. In vivo BBB permeability seems to peak initially
at day 4 postinfection, most likely because of the actions of Th1 cytokines that accumulate
in serum during peripheral infection (13, 46). After viral neuroinvasion, type I IFN
is rapidly produced in the CNS, appearing as early as day 5 postinfection (16), while
Th1 cytokines are generally not expressed in significant amounts until day 7 or 8
(12, 13, 47). Thus, the initial interaction of the virus with the BBB endothelium
and early IFN production within the infected CNS on days 5 and 6 likely function together
to produce a decrease in BBB permeability. Increases in inflammatory cytokine levels
during later stages of CNS infection derive from both innate immune responses of CNS
cells and infiltrating leukocytes expressing TNF-α and IFN-γ, which also are first
detectable in significant quantities 7 to 8 days postinfection (47
–
49). Likewise, following i.c. inoculation, type I IFNs are induced within 2 days (29)
as BBB permeability decreases, followed by the infiltration of inflammatory cells
on day 4 (27) and a subsequent increase in BBB permeability. Ultimately, the exact
timing of the expression and interaction of IFNs and inflammatory cytokines expressed
in various compartments following WNV infection is complex and, as in studies of BBB
permeability, sometimes differ among reports. However, on the basis of our findings
and the general trends observed in the literature, it appears that systemic immune
factors contribute to BBB breakdown during peripheral infection, which may contribute
to WNV entry into the CNS. Around the time of CNS neuroinvasion, direct interaction
of the virus with the BBB endothelium and early, local innate immune responses in
the CNS lead to a type I IFN-dependent decrease in BBB permeability, which may be
reversed by infiltration of the CNS by inflammatory cells.
Cytokine-mediated BBB regulation is shown to occur via PRR pathways, independently
of viral replication, as detection of viral PAMPs in BMECs was sufficient for the
regulation of barrier physiology. In particular, agonism of WNV-sensing PRRs RIG-I,
MDA-5, and TLR7 induced cytokine responses dominated by type I IFN, thereby enhancing
barrier function. Prior studies have revealed that deficiencies in each of these pathways
result in increased disease susceptibility, including early neuroinvasion and greater
CNS viral burdens during WNV infection (27, 50, 51). Of note, unlike other PRRs, TLR3
agonism increased barrier dysregulation in both WT and Ifnar
−/− BMECs. Compared to other PRRs, TLR3 induced cytokine responses more heavily skewed
toward Th1 cytokines and, uniquely, stimulated rapid expression of IL-1β, even in
WT cells. These data align with a previous report (6) suggesting that TLR3 activation
contributed to BBB breakdown and subsequent WNV neuroinvasion. In our study, in vivo
TLR3 agonism also induced BBB permeability and did so more efficiently in Ifnar
−/− than WT mice, again suggesting that type I IFN expression may counterregulate
the effects of TNF-α and IL-1β at the BBB. The specific role of TLR3 during WNV encephalitis
remains uncertain, as another report (3) was unable to link TLR3 activation to BBB
breakdown. Ultimately, TLR3 signals through an adaptor different from that used by
other PRRs (52) and thus may exhibit different regulatory properties at the BBB, as
has been shown in other tissues (53). As demonstrated by Verma et al. (2), we found
that WNV is capable of trafficking across the BBB endothelium in vitro, though this
study used high MOIs over several days, leading to increased virus in bottom chambers
because of replication within endothelium, independently of paracellular permeability.
Alternatively, we show that the virus can cross the BBB endothelium quickly after
exposure, independently of replication, and that trafficking is correlated with paracellular
permeability. Movement of virus in this manner is likely relevant in vivo, as studies
have linked host factors that contribute to BBB disruption to WNV neuroinvasion (44,
54, 55). We also demonstrate a rapid (<6 h), cytokine-dependent regulation of TJ protein
localization after infection. Type I IFN regulation of BMEC TJs was due to the activation
of Rac1 and suppression of RhoA activity, a regulatory regimen well known to control
TJ integrity (26). This finding is in line with a report demonstrating that RhoA contributed
to the dysregulation of BMEC TJs, enhancing the trafficking of HIV-1-infected monocytes
across the BBB (56). Thus, regulation of this pathway may serve to limit CNS trafficking
of both free and cell-associated WNV.
Understanding how viral and host factors contribute to the development of neuroinvasive
disease is a critical step in developing treatment and prevention strategies for WNV
encephalitis. IFN therapy has been used with some success in immunocompetent patients
with WNV encephalitis (57, 58), while IFN therapy in immunosuppressed patients, such
as recipients of infected transplant organs, has been less successful, according to
case reports (59). A limitation of the human treatment literature, however, is that
patients have typically developed neuroinvasive disease before treatment; this caveat,
combined with the broad systemic effects of IFN therapy, limits our understanding
of how IFN may regulate BBB function in human patients. A more promising line of investigation
includes the identification of genetic differences in innate immune pathways that
may influence disease susceptibility to encephalitis-causing viruses. Studies thus
far have associated polymorphisms in several type I IFN pathway elements, including
IRF3, MX1, and OAS1, with susceptibility to symptomatic WNV infection (60, 61). In
addition, enhanced TLR3 activity and inflammatory cytokine production after WNV infection
have been shown in macrophages obtained from elderly patients compared to those of
young controls, which may contribute to the enhanced susceptibility of older individuals
to neuroinvasive disease (62). Thus, understanding how variations in innate immune
factors contribute to susceptibility to viral encephalitis may inform new strategies
for prevention and targeted therapeutics.
MATERIALS AND METHODS
Murine model of WNV encephalitis.
Five- and 8-week-old C57BL/6 mice were commercially obtained (Jackson Laboratories).
Congenic Ifnar
−/− and Irf7
−/− mice were the generous gift of Michael Diamond (Washington University in St. Louis).
All animals were housed under pathogen-free conditions in the animal facilities of
the Washington University School of Medicine. All experiments were performed in compliance
with Washington University animal studies guidelines. Mice were inoculated subcutaneously
via footpad injection (50 µl) or i.c. (2 µl) with either 100 or 10 PFU of WNV, respectively,
as previously described (23). WNV strain 3000.0259 (isolated in New York in 2000 [63])
was used in all experiments. Viral titers in all experiments were determined via standard
plaque assay in BHK21-15 cells as previously described (64).
In vivo assessment of BBB permeability.
At various day intervals after infection, mice were injected i.p. with 100 µl of 100 mg/ml
fluorescein sodium salt (Sigma-Aldrich) in sterile phosphate-buffered saline (PBS).
After 45 min, mice underwent extensive cardiac perfusion with PBS, followed by collection
of blood and harvesting of CNS tissues. Tissue homogenates and serum were incubated
overnight at 4°C at a 1:1 dilution in 2% trichloroacetic acid (Sigma-Aldrich) to precipitate
protein, which was pelleted by 10 min of centrifugation at 4,000 rpm at 4ºC. Supernatants
were then diluted in equal volumes of borate buffer, pH 11 (Sigma-Aldrich). Fluorescence
emission at 538 nm was determined via a Synergy H1 microplate reader and Gen5 software
(BioTek Instruments, Inc.). Tissue fluorescence values were standardized against plasma
values for individual mice.
In vitro BBB cultures and TEER recordings.
Primary murine BMECs were prepared from 8- to 10-week-old WT and Ifnar
−/− mice as previously described (65). Primary neonatal murine astrocytes were prepared
from WT and Ifnar
−/− pups (postnatal days 1 to 3) from mixed glial cultures as previously described
(66). In vitro BBB cultures were then prepared as previously described (67). Briefly,
105 BMECs were cultured on the apical side of a 0.9-cm2 fibronectin-coated polyethylene
terephthalate filter insert with a 3.0-µm pore size (BD Falcon). A total of 105 astrocytes
were cultured concurrently in 12-well plates until they reached confluence (~2 days),
at which point BMEC inserts were added to astrocyte cultures. Resistance recordings
were made via chopstick electrode with an EVOM voltmeter (World Precision Instruments).
Resistance values are reported in Ω/cm2 (recorded values minus values for cell-free
inserts).
ICC and IHC analyses.
ICC analysis on primary BMECs was performed after 10 min of fixation in ice-cold methanol,
followed by blocking for 30 min in 10% goat serum in PBS at room temperature (RT).
Cells were then incubated with primary antibodies to Zo-1 (rat anti-mouse monoclonal
antibody, clone R40.76; Millipore) and Claudin-5 (rabbit anti-mouse polyclonal antibody;
Invitrogen) in blocking buffer for 1 h at RT, washed three times in PBS, and then
incubated for 15 min in goat anti-rat Alexa Fluor 488-conjugated and goat anti-rabbit
Alexa Fluor 555-conjugated secondary antibodies in blocking buffer at RT. CNS tissue
sections were obtained from WT and Irf7
−/− mice after cardiac perfusion with 4% paraformaldehyde. Tissues were frozen, blocked,
and stained with the primary and secondary antibodies listed above as previously described
(47). Nuclei in all preparations were stained with Topro3. All images were acquired
via confocal microscopy (Carl Zeiss USA). Colocalization statistics were obtained
via analysis with ImageJ software.
ELISA.
Sandwich ELISA kits were used for detection of cytokine levels in cell culture supernatants.
ELISA kits for IFN-β and IFN-α (PBL InterferonSource), TNF-α (EBioscience), and IL-1Β
(R&D Systems) were used according to the manufacturers’ instructions. Colorimetric
reading of ELISA plates was performed with a Synergy H1 microplate reader and Gen5
software (BioTek Instruments, Inc.).
Virus inactivation.
Stocks of WNV (NY-2000, 2×104 PFU/ml) were incubated in 0.1% (vol/vol) β-propiolactone
(Sigma) for 30 min at 4ºC. BPL in viral stocks was then inactivated by incubating
the stocks for 2 h at 37ºC, followed by hydrolysis of BPL via subsequent overnight
incubation at 4ºC. Alternatively, similar stocks were placed in 35-mm2 culture plates
and incubated in a dark chamber above a UV transilluminator box (Stratagene) for 2 h
at RT. Inactivated viral stocks were stored at −80ºC.
Neutralizing-antibody studies.
Neutralizing-antibody studies were performed after 2 h of pretreatment with purified
anti-mouse TNF-α (1 µg/ml, clone TN3-19; EBioscience) and anti-mouse IL-1β (1 µg/ml,
clone B122; Leinco) antibodies. IgG isotype antibodies were used as negative controls.
GTPase studies.
GTPase pulldown experiments were performed with an activated Rac1/RhoA/CDC42 Rhotekin
and PAK agarose bead kit (Cell BioLabs, Inc.) according to the manufacturer’s instructions.
Purified, GTP-bound protein and unpurified BMEC protein lysates were separated via
gel electrophoresis on 10% bis-Tris gels (Invitrogen) and transferred onto iBlot nitrocellulose
transfer membranes (Invitrogen) according to standard protocols. Blots were probed
with primary antibodies against Rac1, RhoA, and CDC42 (Cell Biolabs), followed by
incubation with IRDye-conjugated secondary antibodies (LI-COR). Blots were imaged
with the Odyssey fluorescent scanning system (LI-COR). TEER and viral trafficking
experiments involving inhibition of GTPase signaling were performed as previously
described (68), with in vitro BBBs after 2 h of pretreatment in the top chamber at
37ºC with the following agents: the Rac1 inhibitor Z62954982 (1 mM; Cayman Chemical
Co.), the Rho/ROCK inhibitor H-1152P (10 nM; Cayman Chemical Co.), and the CDC42 inhibitor
ML141 (100 nM; Tocris-R&D Systems).
Statistical analysis.
All of the data reported here are mean values ± the standard errors of the means.
Data reported for in vitro TEER and cytokine expression experiments include 6 to 12
replicates from two or three independent experiments. In vitro ICC analyses include
five or six replicates from two independent experiments. In vivo fluorescein experiments
included six mice/group/day from two independent experiments. In vivo IHC analyses
included five mice/group/day from two independent experiments. All group effects were
compared via one- or two-way analysis of variance; Bonferroni’s post hoc test was
subsequently used to compare individual means. Correction for repeated measures was
also used where appropriate. All statistical analysis was performed with GraphPad
Prism software (v 5.0). P < 0.05 was considered significant for all comparisons. Statistical
significance is indicated as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001.
SUPPLEMENTAL MATERIAL
Figure S1
(A) Five-week-old WT C57BL/6 mice were infected via footpad injection with 100 PFU
WNV. Data are viral titers from whole brain homogenates harvested on the days indicated.
(B) TUNEL staining for apoptotic cells in BMEC monolayers following 6 h of infection
with WNV at an MOI of 0.01, compared to a DNase (100 U/ml)-treated positive control.
(C) Multistep growth curves in BMEC monolayers following infection with WNV at the
MOIs indicated. Supernatant viral titers were assessed via plaque assay, and results
are reported in plaque-forming units per milliliter. Download
Figure S1, TIFF file, 0.9 MB
Figure S2
(A to D) ELISA was used to determine cytokine levels in top/apical chambers of in
vitro BBBs constructed with either WT or Ifnar
−/− BMECs over WT astrocytes treated for 6 h with WNV, WNV-UV, or WNV-BPL at an MOI
of 0.01; the TLR3 agonist “naked” poly(I · C); the TLR7 agonist CL264; the MDA5-biased
agonist HMW poly(I · C)-LyoVec; or the RIG-I agonist 5′ ppp-dsRNA. Download
Figure S2, TIFF file, 0.5 MB
Figure S3
(A, B) ELISA was used to determine IFN-β levels in the top chambers of in vitro BBBs
constructed with WT or Ifnar
−/− BMECs over WT astrocytes pretreated for 2 h with the Rac1 inhibitor Z62954982
(A) or the ROCK/RhoA inhibitor H1152P (B) and then infected for 6 h with WNV at an
MOI of 0.01. (C) TEER measurements in BBB cultures generated with WT versus Ifnar
−/− BMECs (both over WT astrocytes). Cultures received 2 h of pretreatment with the
inhibitors indicated, followed by 6 h of mock infection with saline. (D) ICC analysis
of colocalization of Zo-1 and Claudin-5 in WT versus IFNAR
−/− BMEC monolayers treated for 2 h with inhibitors of Rac1 (Z62954982), ROCK/RhoA
(H1152P), and CDC42 (ML141). Following 2 h of pretreatment, cells were mock infected
with saline for 6 h. Colocalization is quantified as the index of correlation between
green and red staining (see Fig. 1). Download
Figure S3, TIFF file, 4.3 MB
Figure S4
Detection of replicating virus in the bottom/abluminal chambers of in vitro BBBs constructed
with WT BMECs over Ifnar
−/− astrocytes with pretreatment of the top/luminal chamber overnight with 100 pg/ml
IFN-β (A) or WT or Ifnar
−/− BMECs over Ifnar
−/− astrocytes (not treated with a cytokine) (B, C). The cultures in panel C were
pretreated for 2 h with the CDC42 antagonist ML141 before infection. The cultures
in all of the panels were infected with WNV at an MOI of 0.01 in the top/luminal chamber
and allowed to incubate for 6 h before removal of the top/luminal chamber. The values
shown represent viral titers present after 0, 12, 24, and 48 h of replication by virus
that crossed during the 6 h of infection, prior to the removal of BMECs from the culture
system. Viral titers, determined via standard plaque assay in BHK cells, are reported
in plaque-forming units per milliliter. The dotted line represents the limit of detection
of the assay. Download
Figure S4, TIFF file, 0.5 MB
Figure S5
Eight-week-old WT and Ifnar
−/− mice received mock i.c. inoculations with saline. On the days indicated following
infection, mice were administered sodium fluorescein (see Fig. 5) and CNS regions
contralateral to the inoculation site were assessed for BBB permeability. Download
Figure S5, TIFF file, 0.2 MB