Splenic macrophages are required for protective innate immunity against West Nile virus

Although the spleen is a major site for West Nile virus (WNV) replication and spread, relatively little is known about which innate cells in the spleen replicate WNV, control viral dissemination, and/or prime innate and adaptive immune responses. Here we tested if splenic macrophages (MΦs) were necessary for control of WNV infection. We selectively depleted splenic MΦs, but not draining lymph node MΦs, by injecting mice intravenously with clodronate liposomes several days prior to infecting them with WNV. Mice missing splenic MΦs succumbed to WNV infection after an increased and accelerated spread of virus to the spleen and the brain. WNV-specific Ab and CTL responses were normal in splenic MΦ-depleted mice; however, numbers of NK cells and CD4 and CD8 T cells were significantly increased in the brains of infected mice. Splenic MΦ deficiency led to increased WNV in other splenic innate immune cells including CD11b- DCs, newly formed MΦs and monocytes. Unlike other splenic myeloid subsets, splenic MΦs express high levels of mRNAs encoding the complement protein C1q, the apoptotic cell clearance protein Mertk, the IL-18 cytokine and the FcγR1 receptor. Splenic MΦ-deficient mice may be highly susceptible to WNV infection in part to a deficiency in C1q, Mertk, IL-18 or Caspase 12 expression.


Introduction
West Nile virus (WNV) is a positive-stranded, enveloped, RNA flavivirus and is a member of the Flavivirus genus that are usually transmitted by mosquitos; some members such as the closely-related Japanese encephalitis virus cause viral encephalitis, and central nervous system (CNS) infection (Zika virus; ZIKV) while other members such as dengue virus, yellow fever virus and ZIKV are also associated with systemic diseases [1]. After its introduction into the New York area in 1999, WNV spread rapidly around the country and into North and South America [2]. It is now endemic in all continents except Antarctica, and its virulence is underscored by the large outbreak in the United States in 2012, where over 5000 people were infected, half of which had neurologic disease [3,4]. Currently, there is neither a preventative vaccine nor an effective antiviral treatment for WNV [5,6]. PLOS  Both innate and adaptive immune responses are required for controlling WNV infections [7,8]. WNV is recognized by innate immune pattern recognition receptors (PRR) including the intracellular RNA sensors retinoic acid-inducible gene 1 (RIG-I) and melanoma differentiation-associated gene 5 (MDA5) and endosomal RNA sensors TLR3 and TLR7, and, like other flaviviruses, employs a number of strategies to counteract its recognition [7][8][9]. WNV is transmitted to humans s.c. through the bites of infected Culex species mosquitoes. An animal model of infection where mice are inoculated with WNV in a footpad mimics the natural inoculation route of WNV and has facilitated the characterization of how innate and adaptive immune responses develop after WNV infection and contribute to protective immunity [7,10]. For example, infection of mice deficient in the adaptor downstream of RIG-I and MDA5, mitochondrial antiviral signaling (MAVS), revealed that RIG-I-like receptors and MAVS are essential for innate protective immunity against WNV [11]. Early control of WNV infection also depends on type 1 and type 3 IFN responses. Type I IFN receptor (IFNAR)-deficient (IFNAR -/-) mice infected with WNV all die within 4 days post-infection (p.i.) with doses of WNV that produce little or no death in wildtype (WT) mice [10], while mice lacking the type 3 IFN receptor chain IL-28Rα, exhibit increased CNS infection concomitant with loss of blood brain barrier integrity [12]. Similar studies using knockout (KO) mice showed that TLR7 and its adaptor protein MyD88 as well as cytokines like IL-12p40 and IL-23 are required for protection against WNV [13,14].
While these and other studies with KO mice have helped to identify key elements involved in global protection against WNV, it is less clear which immune cell types are required for innate immune responses to be most effective. Furthermore, the specific target cells in the spleen and other peripheral lymphoid tissues that are infected and/or respond to WNV are not that well defined. Dendritic cells (DCs) and macrophages (MFs) can be infected by WNV and play a pivotal role in linking innate and adaptive immune responses [7,15]. A recent study using conditional KO mice showed that expression of IFNAR in DCs and in myeloid cells including MFs, monocytes (MOs) and neutrophils (Nphs) is required for protection against WNV [16]. Systemic depletion of monocytes and MFs via injection of clodronate liposomes (CLLs) resulted in greatly accelerated encephalitis and death after WNV infection [17]. Local depletion of subcapsular sinus MFs had little or no effect on either B cell or CTL responses to WNV after it was injected into footpads of mice [18,19]. However, draining lymph node (LN) MFs are required for mice to survive a local WNV infection [18]. While specific IFN stimulated gene products have been shown to control WNV replication and spread within MFs and MOs, relatively little is known about the role of splenic MFs and MOs in the protection against subcutaneous WNV infection.
Although the spleen is a major site for WNV replication and spread, it is unclear which innate cells in the spleen replicate WNV, control viral dissemination, and/or regulate innate and adaptive immune responses. We decided to test if splenic MFs were necessary for control of WNV infection. We selectively depleted MFs from the spleen by injecting mice i.v. with CLL several days prior to infecting them with WNV. This regimen did not affect draining LN MFs. Mice missing splenic MFs succumbed to WNV infection after an increased and accelerated spread of virus to the spleen and the brain. WNV-specific Ab and CTL responses were normal in splenic MF-depleted mice; however, numbers of NK cells and CD4 and CD8 T cells were significantly increased in the brains of infected mice. Splenic MF deficiency led to increased WNV in splenic CD11b-DCs, newly formed MFs and monocytes. Unlike other splenic myeloid subsets, splenic MFs express high levels of mRNAs encoding the complement protein C1q, the apoptotic cell clearance protein Mertk, the IL-18 cytokine and caspase 12. Splenic MF-deficient mice may be highly susceptible to WNV infection in part to a deficiency in the expression of one or more of these proteins.

Mice
C57BL/6J WT mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and subsequently bred in-house. All mice were housed and maintained in a specific-pathogen-free facility at the University of Washington. B cell-deficient μMT mice were kindly provided by Dr. David Rawlings (Seattle Children's Research Institute).

Ethics statement
All animal experiments were performed in strict accordance with the recommendations in the NIH guidelines listed in the Guide for the Care and Use of Laboratory Animals [20], the Animal Welfare Act, and U.S. federal law. All experiments and procedures were approved by and conducted according to the University of Washington Institutional Animal Care and Use Committee (Assurance Number A3464-01, protocol 2242-08). Footpad injections were performed under anesthesia with ketamine hydrochloride and xylazine, and all efforts were made to minimize suffering. The University of Washington Animal Care and Use Program is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International accreditation number 000523, and registered with the USDA, certificate number 91-R-0001.

Infection with WNV
Female and male 6-to 10-week-old, age-and sex-matched mice were injected under anesthesia in one hind footpad with 10 μL of 100 PFUs of WNV diluted in a vehicle of Hanks buffered salt solution containing 1% FBS. Mice infected with WNV were monitored twice a day for weight, morbidity and mortality. A clinical scoring system was used as follows: 0 = healthy mouse; 1 = ruffled fur, lethargy, hunched posture, no paresis; 2 = very mild to mild paresis; 3 = some paresis in one hind limb or conjunctivitis; 4 = paresis but retains feeling' possibly limbic; 5 = true paresis; 6 = moribund. Mice with a weight loss of !20% of their baseline weight or reaching a score of 5 were euthanized immediately using CO2 followed by cervical dislocation. Mice that were infected with WNV and did not meet the criteria for euthanasia were kept up to 30 days after infection and then sacrificed as described above. A few WNVinfected mice that had not met any criteria for euthanasia at the last check point 12 hours previously ( 5%) were found dead in the cage in spite of twice daily monitoring and added environmental enhancements (gel, moist food and extra nestlet material).

In vivo depletion of splenic MFs
WT mice were injected iv with liposomes in 100 μL containing either 0.5 mg CLLs or 0.5 mg control PBS liposomes (PBSL) as described [17]. To verify that splenic MFs had been depleted, 3 days after CLL or PBSL injection, cell suspensions were prepared from isolated spleens and myeloid subsets quantified as described in the Results.

Virus and quantification of viral burden
All virus stocks were maintained and characterized within the University of Washington's Center for Flavivirus Immunity. The WNV infectious clone (WNV-TX) [21,22] was used at 103 PFU/dose, as determined by a plaque assay using baby hamster kidney (BHK) cells. Viral tissue burden quantification was done as described [23]. Briefly, infected mice at various times p.i. were sacrificed and perfused with PBS, and tissues were harvested into centrifuge tubes containing ceramic beads and PBS, then tissues were homogenized once in a homogenizer except brain tissues which went through two cycles of homogenization. Samples were spun down, and the supernatants were analyzed for viral titer with a plaque assay using BHK cells. Viral RNA levels in sera, tissues and sorted cells were determined by quantitative RT-PCR (qPCR) as previously described using Sybr green master mix (Applied Biosystems) and WNV DNA standards (WNV E protein DNA in a plasmid) [11]. Viral RNA was extracted using a QiaAMP viral RNA extraction kit (Qiagen, Valencia, CA), and total RNA was extracted from tissues using RNeasy kits (Qiagen). Sequences for primers used were as previously published [11,24]: forward primer 5'-TCAGCGATCTCTCCACCA AAG and reverse primer 5'-GGGTCAGC ACGTTTGTCATTG. Threshold cycle values from samples analyzed by Sybr green were made relative to the GAPDH housekeeping gene and normalized to WT uninfected samples.

Quantification of Ab responses
WNV-specific Ab responses were determined as described [23]. Briefly, relative quantities of WNV-specific Abs were measured using a sandwich ELISA and WNV envelope (E) protein purified as described [25]. Serum samples were inactivated for virus by exposure to UV light for 30 min, serially diluted 3-fold starting at 1/20 in 0.1% milk casein in PBS. Goat anti-mouse IgM or goat-anti-mouse IgG sera conjugated to horseradish peroxidase (Southern Biotech, Birmingham, AL) were added at a 1:2,000 or 1:4,000 dilution. A positive reading was determined to be a value above the cutoff value determined from the mean of negative-control wells (naive and/or mock-infected mice) + the standard deviation of the mean X the standard deviation multiplier, f [26]. Negative-control wells contained serial dilutions from at least three individual mice per experiment. To determine neutralizing (nAb) titers, serum samples were analyzed in a plaque reduction neutralization titer (PRNT) assay as previously described] 11, 23].

Spleen and dLN cell isolation
LNs were harvested suspended in digestion buffer (serum-free RPMI 1640 medium, Thermo Scientific, Waltham, MA) in the presence of 170 μg/mL Liberase (Roche, Pleasanton, CA) and 10 μg/mL DNase I (Roche) and incubated for 25 min at 37˚C. LNs were then disrupted mechanically with a needle and syringe and strained through nylon mesh to obtain a single cell suspension. Spleens were injected with 1 mL of digestion buffer then incubated for 30 min at 37˚C. Spleens were transferred to a 70um cell strainer, washed with buffer (1xPBS, 2mM EDTA, 0.5% BSA, 10 μg/mL DNase1) to generate a single cell suspension. Cells were then washed with FCS-containing RPMI 1640 medium and red blood cells (RBCs) were lysed with 1X RBC lysis buffer (BioLegend, San Diego, CA) prior to staining for flow cytometry. For isolation of leukocytes from the brain, tissues were harvested and finely chopped with scissors over a wire screen mesh in cold 5% FCS-containing PBS. Cells were then washed with FCS-PBS and resuspended in digestion buffer and incubated for 45 min at 37˚C. After adding EDTA to 2mM, tissues were further disrupted by pipetting, passed through strainer into 25% Percoll (Sigma-Aldrich, St. Louis, MO) in RPMI with 1% FCS, 25mM HEPES and spun with brake 20 min at 800xg. The myelin/fat layer was aspirated and cell pellet suspended in PBS without serum. Leukocytes were then suspended in 70% Percoll layer that was underlayed with 30% Percoll, and the cells were spun down for 20 min at room temperature, no brake. Lymphocytes were obtained from the 30%:70% interface and washed with serum-containing RPMI 1640 medium prior to staining for flow cytometry.
To detect WNV-specific CD8 + T cells, cells were first stained with mAbs to surface markers, washed the cells and then stained them with a 1:100 to 1:250 dilution of MHC class I tetramer-PE at 4˚C for 25 min. All samples were stained in PBS containing 2% FCS and 0.05% azide (FACS buffer) at 4˚C and fixed with PBS containing 4% paraformaldehyde before acquisition.
The schemes for the analysis of myeloid subsets in the spleen and dLNs are shown in Fig A in S1 File and Fig B in S1 File, respectively. For analyses of splenocytes, debris, doublets and dead cells were excluded based on side scatter (SSC) and granularity. Myeloid cell populations were defined based on their expression of CD11b and other markers: CD11b hi F4/80 hi MFs; conventional F4/80-CD11c hi MHCclassII hi DCs, which were further subdivided into CD11b -DCs and CD11b + DCs; NK1.1 hi CD11b + CD3 -NK cells; NK1.1 hi CD11b + CD3 + NKT cells. NK1.1 -CD11b + CD19/CD3cells were further subdivided based on SSC and relative Ly6C and Ly6G expression into SSC lo Ly6C lo MOs, SSC lo Ly6C hi MOs, Ly6G hi Ly6C + SSC int Nphs and SSC hi Ly6C + Ly6G + eosinophils. CD11b -CD3/CD19 + cells were subdivided into MHC class II hi B cells and MHC class II -T cells; For analyses of dLN cells, after exclusion of debris, doublets and dead cells, cell populations were defined based on their expression MCH class II and/or CD19/CD3 as B cells (CD3/CD19 + MHC class II hi) and T cells (CD3/CD19 + MHC class II -); CD3/CD19cells were further subdivided into NK cells (NK1.1 hi FSC lo ), Nphs (CD11b hi Ly6-G hi ), CD11b + cDCs, CD11b -cDCs, F4/80 hi MFs and Ly6C hi MOs.

Sorting of splenic myeloid subsets
Splenocytes prepared as described above were resuspended in sort-buffer (1 x PBS, 2mM EDTA, 0.5% BSA and >10μg/mL DNase1, 10mM HEPES) and incubated with F4/80-biotin αbiotin magnetic beads on ice for 15-20 min. Then cells were washed, resuspended in sort-buffer and added to an LS column attached to the MACS Separator. The columns were washed with 5 mL of sort-buffer and unbound cells were used to obtain other myeloid subsets. The column was removed from the MACS Separator into a new 15 mL conical tube on ice and the RPMs (F4/80 enriched cells) were eluted with 5 mL sort-buffer. Cells obtained from the flow through fraction were incubated with anti-CD11c and anti-CD11b beads, washed, resuspended in sort-buffer and added to a prepared LS column. The LS column was removed and cells eluted with 5 mL sort buffer. Cells were stained with the surface markers for myeloid cell subsets before sorting, except for Streptavidin-BV605 used to stain F4/80 LS enriched cells. For sorting cells from WNV-infected mice, a BSL3 flow cytometry facility with an Aria1 cell sorter in the Department of Immunology's Cell Analysis Facility Flow Cytometry Core was used. Stained cells were gated to eliminate debris, doublets and dead cells using DAPI. B cells were excluded gating on CD19cells. RPMs were identified as F4/80 hi FITC + (autofluorescent) cells. We used the CD11b vs. CD11c markers to separate out CD11c hi CD11b + DCs from CD11c hi CD11b -DCs. We gated on CD11b hi CD11ccells to isolate Ly6G + neutrophils; then the CD11b hi Ly6Gcells were further gated on SSC lo F4/80cells to sort MOs.

mRNA isolation and qRT-PCR array
Total mRNA was isolated from whole spleen using Qiagen RNeasy minispin columns and using the manufacturer's protocol. In brief, tissues were homogenized and treated with RLT lysis buffer to obtain lysates for RNA isolation. Isolated RNA was reverse transcribed using a cloned avian myeloblastosis virus reverse transcriptase kit (Invitrogen, Grand Island, NY) to make cDNA. qRT-PCR reactions were performed on ViiA7 RT-PCR instrument (Applied Biosystems, USA) using TaqMan PreAmp Master Mix (ThermoFisher Scientific, USA. Custom TaqMan Array cards (pre-loaded 384-well microfluidic cards) were purchased from Thermo-Fisher Scientific (list of primers in Table A in S1 File). All reactions were performed according to the manufacturer's protocols. Gene expression analyses were performed by ΔΔCT method [29] using generated using DataAssist3.1 with 18s and HPRT as endogenous controls as previously described [30]. WNV infected samples were compared to mock-infected samples. PCRamplification of each cDNA sample was done in triplicates.

Depletion of splenic MFs by intravenous delivery of clodronate liposomes
To develop a method for depletion of splenic MFs, we injected mice i.v. with CLLs or PBS-filled control liposomes (PBSL) and analyzed spleen cell populations 3 days later (Fig 1). F4/80 hi CD11b int/low splenic MFs were present in naïve and PBSL-treated control mice but were selectively and significantly depleted in the CLL-treated mice (Fig 1A and 1B, gating shown in Fig A  in S1 File). Other myeloid populations including MOs, DCs, eosinophils and Nphs were not significantly depleted (Fig 1B). Splenic MFs remained depleted 6 days after CLL treatment. The depletion was restricted to MFs in the spleen, as MFs were not depleted from inguinal draining LNs (dLNs) (Fig 1C, gating shown in Fig A in S1 File) or popliteal dLNs. In addition, NK, NKT cells, T cells, and B cells were not depleted from the spleen or dLNs after CLL treatment  (Fig A in S1 File). Thus, i.v. delivery of CLL resulted in selective and sustained depletion of splenic MFs.

Mice depleted of splenic MFs are highly susceptible to WNV
To determine whether splenic MFs are required for protection against WNV infection, we compared the susceptibility of mice treated with either CLL or PBSL 3 days prior to footpad inoculation of WNV-TX (Fig 2A). All mice depleted of splenic MFs prior to WNV infection succumbed to infection by day 12 post-infection (median survival time, 11 days), compared to 14% of infected control mice (p = 0.0015) (Fig 2B). Splenic MF-depleted mice began to show weight loss and clinical symptoms within 7-8 days p.i. (Fig 2C and 2D) and also had increased levels of WNV in the blood as early as day 2 p.i. (Fig 2E). These data demonstrate that splenic MFs are required for survival after WNV infection and suggest that they might be essential to control virus replication in the spleen very early after WNV infection.

Control of WNV replication in the spleen and CNS requires splenic MFs
To further examine the role of splenic MFs during WNV infection, we determined the levels of WNV RNA at the WNV dLN inoculation site and in the spleen at 2 days p.i.; although PBSLcontrol-and CLL-treated mice had similar levels of WNV RNA in popliteal and inguinal dLNs, CLL-treated mice had higher levels of WNV RNA in the spleen at day 2 post-WNV infection (Fig 2F). WNV replication in the spleens of WT mice typically peaks about day 4 p.i., and then decreases and is cleared within 6 to 8 p.i.; this course of infection is evident even in innate immune defective MAVS -/mice [7]. However, splenic MF-depleted mice had both a higher frequency of detectable WNV (67% positive versus 17% positive in PBSL controls) and higher infectious virus levels in the spleen even at 2 days p.i. (Fig 2G). Splenic MF-depleted mice all had detectable virus in the spleen by day 4 p.i., compared to 50% of the PBSL-treated mice, and greater than a hundred fold increase in viral load (p = 0.002), and, unlike the control mice, continued to have high levels of virus at day 6 p.i. However, at day 4 p.i. the splenic MF-depleted and control mice did not differ in their ability to control virus in the kidneys and liver.
The lack of splenic MFs also led to an increased viral burden in the brain. Infectious virus was detectable in the brains as early as day 4 p.i. in splenic MF-depleted mice, whereas control mice had no detectable virus at this time ( Fig 2H). As the infection progressed, WNV levels increased and 80% of splenic MF-depleted mice by day 8 p.i. had detectable WNV in the CNS. Together, these data show that splenic MFs are required early during infection for control of virus replication in the spleen and that in their absence, virus rapidly spreads to the brain.

WNV-specific Ab responses are elevated in splenic MF-depleted mice
WNV-specific Abs are important for controlling virus early after infection [31,32], and MFs can contribute to the development of effective humoral responses [15,33,34]. Therefore, we tested whether the lack of splenic MFs might hinder the development of WNV specific Ab responses, which in turn might account for the increased WNV in the spleen and CNS in splenic MF-depleted mice, as occurs in B cell deficient (μMT) mice [31]. We examined sera from infected CLL-and PBSL-treated mice for IgM and IgG responses to recombinant WNV envelope (Env) protein (Fig 3A). Surprisingly, splenic MF-depleted mice, rather than having reduced levels of WNV-specific Ab or nAb, had increased WNV specific IgG and nAb compared to controls. Both Env-specific IgG at days 5 and 8 p.i. and nAb at day 5 p.i. were significantly higher in splenic MF-depleted mice compared to controls. These data suggest that the increased levels of Ab in CLL-treated mice were not sufficient to clear the increased WNV in these mice.
To directly assess the quality of the Abs produced in WNV-infected splenic MF-depleted mice, we compared sera obtained from splenic MF-depleted mice to sera from PBSL-treated mice or immune sera from WNV-infected WT controls for their ability to protect otherwise susceptible B cell deficient (μMT) mice (Fig 3B) [32]. A pool of sera obtained from WNVinfected CLL-treated mice at day 8 p.i. was as effective as a pool of sera from infected control mice in protecting μMT mice from dying. Also, after WNV infection B cell numbers in the spleen were similar between splenic MF-depleted mice and controls at day 8 p.i., although there was a small decrease in numbers of PNA + germinal center (GC) B cells in splenic MFdepleted mice (Fig B in S1 File). Thus, splenic MF-depleted mice that succumb to WNV infection do produce Abs that can protect.

WNV-infected splenic MF-depleted mice generate WNV-specific T cells that enter the CNS but fail to control WNV replication
Since both CD4 and CD8 T cells are required for protective immunity against WNV [7,10], we assessed if T cell responses were compromised after WNV infection of splenic MF-depleted mice. We measured WNV-specific CD4 and CD8 T cell responses (Fig 4). Splenocytes were  isolated from either untreated naïve mice or from CLL-or PBSL-treated mice day 8 p.i. and cultured in vitro with immunodominant peptides from either WNV NS3 protein or NS4B protein, which stimulate WNV-specific CD4 T cells or CD8 T cells, respectively (Fig 4A and 4B). CLL-treated mice had somewhat lower numbers of total CD4 and CD8 T cells than control mice in the spleen (see Fig 5B); however, levels of WNV-specific IFN-γ-producing CD4 and CD8 T cells were not significantly different between groups (Fig 4A). Likewise, NS4B tetramer-binding CD8 T cells in the spleen were not different between splenic MF-depleted mice and PBSL-treated controls (Fig 4B). In contrast, NS4B tetramer-binding CD8 T cell levels within the brain were significantly increased in splenic MF-depleted mice 8 days p.i. (p = 0.01) (Fig 4C). Together these data show that WNV-specific T cell responses were generated in the splenic MF-depleted mice and that a greater number of WNV specific CD8 T cells trafficked to the brain in splenic MF-depleted mice compared to controls. The brains of the splenic MF-depleted mice had increased levels of virus (Fig 2H) in spite of the significant recruitment of WNV-specific CD8 T cells.

Increased cellular infiltration in the brains of WNV-infected splenic MFdepleted mice
We next compared the number of leukocyte subsets in the brains of WNV-infected mice pretreated with either CLL or PBSL. Total numbers of leukocytes, B220 + B cells and microglia cells in the brains of splenic MF-depleted mice were not significantly different from control mice. However, numbers of Ly6C hi MOs, NK cells, CD4 T cells and CD8 T cells were significantly increased at day 8 pi (Fig 5). Thus, the high level of WNV in the brains of splenic MFdepleted mice (Fig 2H) is not simply due to lack of immune cells; rather the numbers of potential effector cells were increased. Interestingly, at the same time point a different pattern was evident in the spleens of MF-depleted mice: the cell populations elevated in the brain-Ly6C + monocytes, NK cells, CD4 T cells and CD8 T cells-were instead decreased in the spleen (Fig B  in S1 File). F4/80 hi MFs were present in the spleens of MF-depleted mice at day 8 p.i. (11 days after CLL treatment) but remained significantly lower than MF levels in control mice (Fig B  in S1 File). Together these data suggest that the decreased numbers Ly6C + MOs and other subsets in the spleen may be due to increased trafficking of these same populations to the brain.

Splenic MF deficiency leads to increased WNV expression in splenic CD11c + CD11b -DCs, newly formed MFs and monocytes
To compare expression of WNV in splenic myeloid populations obtained from WNV-infected splenic MF-depleted mice to infected control mice, we enriched splenic myeloid populations from WNV-infected mice and examined in detail WNV expression in five myeloid subsets isolated by cell sorting (see Materials and Methods). Little or no WNV RNA was found in splenic B cells and other non-myeloid splenic cells. We extracted RNA from each purified myeloid subset and quantified WNV RNA expression on either a per cell basis (Fig 6A) or on a per population basis (Fig 6C) by qPCR. In PBSL-treated control mice infected 3-4 days previously, WNV RNA was not detectable in splenic Nphs, was low in splenic MOs and was expressed a little higher in CD11c + CD11b + DCs and F4/80 hi MFs. However, it was expressed at high levels in CD11c + CD11b -DCs (Fig 6A). These data demonstrate that splenic CD11c + CD11b -DCs, which are mainly located in T cell zones and active in cross-presentation of antigens [35], are a major target for WNV infection.
During infection of splenic MF-depleted mice, the same pattern of WNV expression was evident (Fig 6, white bars), except that overall levels of viral RNA were significantly increased (2-way ANOVA p = 0.002). CD11c + CD11b -DCs continued to express more WNV than CD11c + CD11b + DCs. However, after splenic MF depletion, significant WNV RNA was now evident in both MOs and Nphs. Thus, when mature F4/80 hi MFs are depleted, WNV control is altered, making some myeloid cells more susceptible to WNV. Furthermore, F4/80 hi MFs, which had begun to repopulate the spleen 7 days after initial MF depletion, had similar levels of WNV RNA on a per cell basis as the CD11b -DCs, whereas in control mice, the mature splenic MFs had much lower level of WNV RNA than CD11b -DCs (Fig 6A). These same patterns of WNV expression were evident when we calculated the total amount of WNV RNA
https://doi.org/10.1371/journal.pone.0191690.g005 contributed by each splenic cell population by multiplying the frequency of each splenic myeloid subset (Fig 6B) by the amount of WNV RNA/cell each subset produced (Fig 6A). The CD11c + CD11b -DC and F4/80 hi MF populations contributed the most WNV RNA in both control and CLL-treated mice and at similar levels (Fig 6C and 6D). The MOs and Nphs also contributed significant amounts of WNV but only in the CLL-treated mice. Overall the infected splenic MOs in splenic MF-depleted mice had 17.5 times more WNV RNA than in controls (Student's t test p = 0.02). These infected MOs may be a source of MOs infiltrating and bringing WNV to the brains of splenic MF-depleted mice (Fig 5). The fact that mature F4/80 hi MFs in control mice and repopulating F4/80 hi MFs in CLL-treated mice differed in their level of WNV expression suggested that the MFs repopulating the spleen might be phenotypically distinct from resident MFs. Indeed, the newly repopulating F4/80 hi MFs from CLL-treated mice compared to resident MFs from PBSL-treated mice had significantly increased size and granularity and higher levels of CD11b.

Gene expression in splenic F4/80 hi MFs from naïve uninfected or WNVinfected mice
To assess antiviral genes that may be deficient in splenic MF-depleted mice, we used microfluidic qPCR arrays to analyze gene expression in splenic myeloid populations isolated by cell sorting from naïve mice or PBSL-treated mice 4 days after WNV infection. We compared expression of 108 immune-associated genes (Table A in S1 File) in F4/80 hi MFs, CD11c + CD11b -DCs, CD11c + CD11b + DCs or CD11b + SSC lo MOs (gating used for cell sorting in Fig A in S1 File) and calculated the relative contribution of a myeloid subset to the gene's expression by multiplying the gene expression levels times the frequency of the splenic population. A number of genes were highly expressed in F4/80 hi MFs isolated from uninfected mice and were significantly much higher compared to the expression in other subsets (Fig 7A,  Table B in S1 File). Thus, the myeloid cells remaining in splenic MF-depleted mice are deficient in these genes. Some of these genes, as expected, encoded known MF-associated markers like CD163 and Clec4n (Dectin-2). Several genes known to be important for control of WNV infection such as C1q genes were selectively expressed in splenic MFs. Both C4 -/and C1q -/mice succumb to WNV infection [36]. Other genes expressed at high levels in splenic MFs have not yet been directly implicated in the control of WNV but function either in innate immune (e.g., TLR8, IL18, Clec4n and Casp12) or adaptive immune responses (e.g., Tgfb2 and CD86).
CD11b -DCs, unlike the other myeloid subsets we examined, contained high levels of WNV 4 days after infection of control mice (Fig 6A). This suggested that the CD11b -DCs might have reduced constitutive levels of genes known to play a role in limiting WNV infection. Indeed, compared to MFs, MOs and even CD11b + DCs, the CD11b -DCs from naïve mice express much lower levels of RIG-I, MDA5 and Mavs (Fig 7A, Table B in S1 File), which are known to play important roles in resisting WNV infection [7,8].
We also evaluated gene expression in these same splenic myeloid subsets, isolated 4 days after WNV infection of PBSL-treated control mice. The expression of a set of genes that increased about 2.5 fold more compared to expression in myeloid cells from uninfected mice is shown in Fig 7B and Table C in S1 File. These genes fell roughly into three groups. The first group was expressed only or mainly in splenic MFs and included, as expected, genes highly expressed in splenic MFs from uninfected mice such as Mertk, C1q proteins and IL18. Notably, C4b was dramatically and almost exclusively upregulated in MFs from infected mice. The genes encoding the regulatory cytokines IL-10 and IL-33 were also expressed principally in MFs before infection and upregulated after infection. A second group of genes that were upregulated after WNV infection were expressed in splenic MFs but not exclusively. Some of the genes expressed in MFs were also expressed in MOs but with little or no expression in the DC subsets, e.g., those encoding FcRγl, Cfb, IRF7 and TGFßR2. These data suggest that in the presence of splenic MF deficiency, splenic MOs in particular may provide proteins that otherwise would be deficient in mice missing splenic MFs. A third group of genes were increased after WNV infection but expressed mainly in non-MFs. Notably, Nos2 was strongly induced after WNV infection but mainly in splenic MOs.

Expression of genes in splenic F4/80 hi MFs that function to sense WNV RNA
To assess further how splenic myeloid populations may sense WNV RNA, we compared in more detail the expression in F4/80 hi MFs, CD11b -DCs, CD11b + DCs, and MOs obtained from WNV-infected mice of genes involved in sensing viral RNA and programming type I IFN responses. (Fig 8). F4/80 hi MFs and monocytes, unlike DCs, were the major myeloid sources of basal expression of the genes encoding the viral sensors RIG-I (Ddx58) and MDA5 (Ifih1), and F4/80 hi MFs expressed more than 2/3 of the genes encoding the adaptor downstream of RIG-I and MDA5, MAVS (Fig 8A). Expression of MAVS on a per cell basis was also higher in F4/80 hi MFs than in DC subsets (Fig 8B). And as noted above, CD11b -DCs expressed very low mRNA levels of these sensors. A comparison of expression after WNV infection to expression in subsets from uninfected mice (Fig 8C) revealed that as, expected, both Ddx58 and Ifih1 expression increased to varying degrees in several myeloid subsets after infection, but MAVS gene expression did not. And even after infection, the CD11b -DCs expressed relatively low amounts of mRNAs encoding the RIG-I and MDA5 sensors.
A different pattern was evident for the genes encoding the RNA sensors TLR3 and TLR7 and the downstream adaptor, Myd88: monocytes contributed most of the Myd88 and expressed significantly higher levels per cell of Myd88 (Fig 8A and 8B). Like MAVS, Myd88 expression did not change significantly in splenic myeloid subsets after WNV infection ( Fig  8C). As with the components upstream of MAVS, genes encoding TLR3 and TLR7 upstream of MyD88 increased after WNV infection: TLR3, while constitutively expressed at high levels in CD11b -DCs, was increased in F4/80 hi MFs after infection; however, on a per cell basis, most of the TLR3 was found in CD11b -DCs (Fig 8B and 8C). In contrast, TLR7 was strongly induced in CD11b + DCs after WNV infection, and on a per cell basis was strongly expressed in all the myeloid subsets except CD11b -DCs.
We conclude that splenic F4/80 hi MFs are especially well programmed to rapidly respond to cytosolic RNA sensing through Mavs, whereas MOs, for example, may utilize more signaling through TLRs. Collectively our data suggest that depleting splenic MFs leads not only to a deficiency in complement components but also to a lesser extent in the RIG-I/MDA5/MAVS pathway, whose components are expressed to a relative lower levels in the other splenic myeloid cell subsets. This could contribute to the susceptibility of splenic MF-deficient mice to WNV infection. Furthermore, the major RNA sensor pathways apparently are differentially Expression of immune-associated genes in splenic myeloid subsets before and after WNV-infection. F4/80 hi MFs, CD11b -DCs, CD11b + DCs or CD11b + SSC lo MOs splenic subsets were isolated by cell sorting from either naïve B6 mice or from PBSL-treated mice infected 4 days previously with 1000 pfu WNV/Tx. RNA was isolated, after which expression of 108 immuneassociated genes were determined (see Materials and Methods and Table A in S1 File). Heat maps show the gene expression/ population for the 4 subsets relative to the total gene expression in myeloid cells obtained from (A) naïve mice (mean of 2 independent experiments) or (B) WNV-infected mice (mean of 3 experiments). Ratios were loaded into R statistical programming language (version 3.2.0). Pearson correlation was performed on ratios and displayed in R/Bioconductor using the gplots package. Gene expression/population shown is calculated as expression of the given population X the frequency of the same population in the spleen. Details of expression levels are available in Tables B and C in S1 File.

Discussion
Here we have shown that splenic MFs are required for successful resistance to WNV infection and that splenic MFs most likely contribute to resistance by sensing WNV RNA and restricting viral spread. Previous studies have shown that MFs contribute to resistance to WNV infection, but just how MFs affect WNV infection and whether MFs in the spleen are essential for protective immunity have been less clear. Systemic depletion of all MFs and monocytes with CLL increases susceptibility to WNV and viral loads after infection [17]. Local depletion of subcapsular sinus (SCS) MFs in LNs prior to footpad inoculation of WNV also increased susceptibility to WNV and serum and brain virus levels but did not affect B cell responses significantly [18]. Furthermore, depletion of LN MFs led to increased and more rapid dissemination of WNV single-cycle viral particles while having no effect on CD8 T cell responses [19]. This suggests that, as is the case for vesicular stomatitis virus [37], SCS MFs play a major role in preventing the dissemination of WNV. Thus far several studies have shown that local depletion of LN MFs does not affect antibody responses to viral infections [30,38]. Similarly, our results demonstrate that selective depletion of splenic MFs does not significantly affect either humoral or T cell immunity to WNV after infection.
Whether or not depletion of splenic MFs would affect humoral immunity to WNV was unclear since MF subsets are functionally distinct, and some MF subsets do play key roles in B cell responses [39]. For instance, some splenic MFs, most likely MZ MFs, are essential for T cell-dependent Ab responses to particulate bacterial Ags [40]. Within the spleen phenotypically distinct MFs have been identified in the red pulp, outer MZ and white pulp [39,41]. Furthermore, known receptors for flaviviruses such as MDL-1 and mannose receptors and CLEC5A are differentially expressed on splenic MFs [42][43][44]. Splenic MFs can replicate flaviviruses after taking up the virus directly or with the help of Ab or complement [45,46]. Nevertheless, global depletion of splenic MFs did not affect adaptive immune responses to WNV. These findings are similar to Seiler et al. [47], who found that while splenic MFs are essential for preventing spread of lymphocyte choriomeningitis virus, they are not required for generation of CTL responses.
The selective depletion of splenic MFs, while leaving LN MFs intact, led mice to become susceptible to WNV and succumb within 12 days of infection. This course was slower than that seen in mice with profound innate immune defects such as IFNAR -/-, MAVS -/or RIG-I -/-MDA5 -/mice that die in <8 days [11,48,49]. Rather the rate of death was similar to that seen in mice deficient in complement components or adaptive immune cells such as B cells or CD8 T cells [31,36,50,51]. The humoral immune response in splenic MF-depleted mice was robust and was generated early during WNV infection, and sera from WNV-infected splenic MF-depleted mice was capable of protecting otherwise susceptible B cell-deficient mice (Fig 3). Yet the high level of virus specific Ab was unable to control virus replication occurring in the spleen or prevent spread to the CNS. As the infection progressed, the level of WNV specific IgG and nAb became more similar to that of controls, even though WNV levels were increased in the spleen and brain. Similarly, WNV-specific CD4 and CD8 T cell responses in the splenic MF-depleted mice were robust (Fig 4), and the brains of WNVinfected splenic MF-deficient mice had significant numbers of immune cells (Fig 5). Apparently, the adaptive immune responses, while functional in the splenic MF-deficient mice, simply could not keep up with the accelerated infection.
The fact that WNV was significantly elevated in both the sera and spleen of splenic MFdepleted mice as early as 2 days p.i. suggests that splenic MFs play a key role early on in preventing WNV spread to other cells in the spleen and elsewhere. Indeed, in the absence of splenic MFs, after WNV infection very high levels of WNV were detected particularly in splenic CD11c + CD11b -DCs and MOs and in MFs that had repopulated the spleen after MF depletion (Fig 6) and was detected even in Nphs. In infected control mice little or no WNV was detectable in splenic MOs or Nphs. Apparently, splenic MFs normally prevent the spread of WNV to MOs and Nphs that otherwise can be infected. The increased numbers of infected MOs in splenic MFs-depleted mice is noteworthy as they may efficiently deliver WNV to the brain as well as potentially contribute to the accelerated WNV encephalitis in these mice [52,53]. Also, WNV-infected Nphs may contribute to pathogenesis by transporting virus to the CNS [54,55].
It has been unclear which splenic cells after WNV infection are most responsible for efficiently producing WNV and furthering viral spread [56]. Our data demonstrate that after WNV infection of WT mice, splenic CD11b -DCs express significantly more WNV than splenic MFs, monocytes, CD11b + DCs or Nphs (Fig 6A). One reason this DC subset may produce more WNV is that they express relatively low levels of the RIG-I and MDA5 sensors and thus may be relatively ineffective at activating innate immune protection against WNV. Previous studies have revealed that CD11b -CD8α + DCs play a key role in the priming of CD8 + cytotoxic T cells to viruses [57]. Furthermore, BATF3 -/mice that do not develop resident CD11b -CD8α + DCs or migratory CD11b -CD103 + DCs have defective virus-specific T cell responses after WNV infection or influenza infections [58][59][60]. Thus, WNV may target a DC subset that is essential for its control.
The presence of resident MFs may facilitate the ability of splenic MOs and Nphs to control virus replication. Viral replication in MO populations has been shown to change depending on which host cells first replicate the virus or on viral strains [61,62]. Ultimately the ability of MOs to control viral replication is dictated by their ability to mount an effective Type I IFN response vs. lower induction of pro-inflammatory cytokines. It is possible that the absence of splenic MFs skews this balance toward increased viral replication in MOs populations.
Previous studies have implicated innate immune genes in the control of WNV tissue tropism [63]. To determine how splenic MFs might reduce the spread of WNV, we compared the expression of a set of innate immune-associated genes in purified splenic MFs, CD11b -DCs, CD11b + DC and MOs in both naïve mice and in PBSL-control mice 4 days after WNV infection. C1q genes were selectively expressed in splenic MFs and C4b was dramatically upregulated after WNV infections (Fig 7, Tables B and C in S1 File). Both C4 -/and C1q -/mice succumb to WNV infection [36]; our results suggest that splenic MFs may be a key source of C4 and C1q that protects against lethal infection. The receptors for complement expressed on MFs are known to normally facilitate uptake and replication of WNV [45]. Also noteworthy is our finding that Caspase 12 expression both before and after WNV infection was restricted to splenic MFs and not detected in other splenic myeloid subsets. Caspase 12-deficient mice are highly susceptible to WNV infection, and like splenic MF-deficient mice die about 10-12 days after infection [64]. Caspase 12 may function to promote anti-WNV responses by promoting TRIM24 ubiquitination of RIG-I that is essential for RIG-I to mediate type I interferon responses. A deficiency in Caspase 12 and combined with lower levels of splenic RIG-1, thus, may contribute to the susceptibility of splenic MF-deficient mice to WNV.
Also, most of IL18 expression in uninfected mice was restricted to splenic MFs while after infection splenic MOs expressed high levels of the IL18 gene. IL18 plays a key non-redundant role the generation of virus-specific NK cells [65], and NK cells may contribute to resistance to WNV [66]. Finally, other immunoregulatory cytokines like IL-10 and IL-33 that play important roles in the control of viral infections are highly expressed by MFs but not by other myeloid subsets, particularly after WNV infection. In the early phase of antiviral innate immunity IL-10 counterbalances pro-inflammatory signals and protects from tissue damage [67]. IL-33 is essential to attenuate viral-encephalitis by downregulating iNOS expression in the CNS [68]. The absence of these protective anti-inflammatory signals in splenic MF-deficient mice may contribute to the increased WNV encephalitis in these mice. While additional studies are required to define further what factors produced by splenic MFs are essential for anti-flaviviral immunity, our studies underscore how important this myeloid subset is for resistance to WNV.
Supporting information S1 File. Fig A. Gating strategies for quantifying splenic and dLN cell subsets. Splenic and dLN cell suspensions were analyzed by flow cytometry. Debris and doublets were excluded based on scatter profile. Cells were stained for CD11b expression and other markers. (a) Spleen cells were identified as follows: CD11b hi F480 hi MFs; F480 -CD11c hi MHCclassII hi DCs further subdivided into CD11b -DCs and CD11b + DCs; NK1.1 + CD11b + NK cells further subdivided into CD3 -NK cells and CD3 + NKT cells; CD11b+CD19/CD3cells subdivided based on side scatter (SSC) and relative Ly6C and Ly6G expression into SSC lo Ly6C lo MOs, SSC lo Ly6C hi MOs, SSC hi Ly6C + Ly6G + eosinophils and SSC hi Ly6C + Ly6G ++ Nphs; CD11b -CD3/CD19 ++ cells further subdivided into MHC class II ++ B cells and MHC class II -T cells; or PBSL (black bar), 3 days prior to s.c. viral (WNV, 1000 PFU) inoculation (footpad), spleens were harvested at day 8 post-WNV. Splenocytes from naïve mice served as a negative control (grey bars). The frequency of myeloid and lymphocyte populations in the spleen were determined by flow cytometry and applied to total splenocytes counts to determine cell numbers for each population. The results shown are the combined result of five experiments. Statistics shown are for Two-tailed Student's t test, Ã p < 0.05, ÃÃ p <0.01, ÃÃÃ p<0.001. Table A. List of primers for the immune-associated genes tested in the microfluidic qPCR Array Table B. Relative expression of immune-associated genes in splenic myeloid subsets isolated from naïve mice Table C