Interferon Regulatory Factor 5-Dependent Immune Responses in the Draining Lymph Node Protect against West Nile Virus Infection

ABSTRACT

Upon activation of Toll-like and RIG-I-like receptor signaling pathways, the transcription factor IRF5 translocates to the nucleus and induces antiviral immune programs. The recent discovery of a homozygous mutation in the immunoregulatory gene guanine exchange factor dedicator of cytokinesis 2 (Dock2^mu/mu) in several Irf5^−/− mouse colonies has complicated interpretation of immune functions previously ascribed to IRF5. To define the antiviral functions of IRF5 in vivo, we infected backcrossed Irf5^−/− × Dock2^wt/wt mice (here called Irf5^−/− mice) and independently generated CMV-Cre Irf5^fl/fl mice with West Nile virus (WNV), a pathogenic neurotropic flavivirus. Compared to congenic wild-type animals, Irf5^−/− and CMV-Cre Irf5^fl/fl mice were more vulnerable to WNV infection, and this phenotype was associated with increased infection in peripheral organs, which resulted in higher virus titers in the central nervous system. The loss of IRF5, however, was associated with only small differences in the type I interferon response systemically and in the draining lymph node during WNV infection. Instead, lower levels of several other proinflammatory cytokines and chemokines, as well as fewer and less activated immune cells, were detected in the draining lymph node 2 days after WNV infection. WNV-specific antibody responses in Irf5^−/− mice also were blunted in the context of live or inactivated virus infection and this was associated with fewer antigen-specific memory B cells and long-lived plasma cells. Our results with Irf5^−/− mice establish a key role for IRF5 in shaping the early innate immune response in the draining lymph node, which impacts the spread of virus infection, optimal B cell immunity, and disease pathogenesis.

IMPORTANCE Although the roles of IRF3 and IRF7 in orchestrating innate and adaptive immunity after viral infection are established, the function of the related transcription factor IRF5 remains less certain. Prior studies in Irf5^−/− mice reported conflicting results as to the contribution of IRF5 in regulating type I interferon and adaptive immune responses. The lack of clarity may stem from a recently discovered homozygous loss-of-function mutation of the immunoregulatory gene Dock2 in several colonies of Irf5^−/− mice. Here, using a mouse model with a deficiency in IRF5 and wild-type Dock2 alleles, we investigated how IRF5 modulates West Nile virus (WNV) pathogenesis and host immune responses. Our in vivo studies indicate that IRF5 has a key role in shaping the early proinflammatory cytokine response in the draining lymph node, which impacts immunity and control of WNV infection.

INTRODUCTION

A canonical model for type I IFN production after virus infection is a two-step positive feedback loop that is regulated by Interferon regulatory factor 3 (IRF3) and IRF7 (1). Detection of viral nucleic acids by Toll-like receptors (TLRs), RIG-I-like (RLRs) receptors, or DNA sensors induces nuclear localization of IRF3, which in concert with NF-κB and ATF-2/c-Jun stimulates transcription, synthesis, and secretion of IFN-β by infected cells. Extracellular interferon beta (IFN-β) binds to the type I IFN receptor (IFNAR) and triggers activation of the JAK-STAT signaling pathway and induction of IFN-stimulated genes (ISGs) (2), which inhibit viral entry, translation, replication, and assembly through a variety of independent mechanisms (reviewed in reference 3). While IRF3 is constitutively expressed in many tissues, IRF7 is an ISG required for the expression of most IFN-α subtypes and is thus a key mediator of the type I IFN amplification loop (1, 4).

Noncanonical signaling pathways also induce type I IFN responses. Even with genetic ablation of IRF3 and IRF7 (Irf3^−/− × Irf7^−/− double-knockout [DKO] cells or mice), type I IFN was produced after infection with West Nile virus (WNV), dengue virus, murine norovirus, or murine cytomegalovirus, albeit at reduced levels compared to wild-type (WT) cells and mice (5–8). Consistent with the sustained production of type I IFN, lethality in Irf3^−/− × Irf7^−/− DKO mice infected with WNV or Chikungunya virus was not as rapid or complete as in Ifnar^−/− mice (5, 9–11). Ex vivo experiments with primary myeloid dendritic cells and macrophages revealed that the IFN-β response after WNV infection was maintained in DKO cells but abrogated in the absence of MAVS (5, 12). The transcription factors ELF4 and IRF5 both have been implicated as participants in the MAVS-dependent induction of IFN-β after WNV infection. A deficiency of ELF4 resulted in reduced type I IFN production after WNV infection in mice (13). ELF4 translocates into the nucleus after MAVS-dependent activation, binds to IFN promoters, and cooperatively increases the binding affinity of IRF3 and IRF7 (13). Mice or cells lacking IRF3, IRF5, and IRF7 (Irf3^−/− × Irf5^−/− × Irf7^−/− triple-knockout [TKO] cells or mice) had abrogated type I IFN and ISG expression in dendritic cells and serum after WNV infection, although an IFN response still persisted in macrophages (14).

Activation of IRF5 in myeloid cells also induces proinflammatory cytokine expression downstream of TLR7 and MyD88 signaling through a TRAF6- and IRAK1-dependent mechanism (15–19). In an original study of immune cells from Irf5^−/− mice, IFN-α levels were normal after TLR stimulation although interleukin-6 (IL-6), IL-12, and tumor necrosis factor alpha (TNF-α) levels were reduced (18). Subsequent experiments suggested that IRF5 regulates type I IFN production in vivo after viral infection (20, 21), and a lack of this function in Irf5^−/− mice was associated with enhanced susceptibility to infection with vesicular stomatitis virus (VSV) or herpes simplex virus (HSV). In response to Newcastle disease virus (NDV) infection, IRF5 induced overlapping and distinct sets of genes compared to IRF7, including the induction of type I IFN and proinflammatory cytokines (20, 22).

Two recent studies revealed the emergence of a spontaneous genomic duplication and frameshift mutation in the guanine exchange factor dedicator of cytokinesis 2 (Dock2) that arose in at least a subset of Irf5^−/− mouse colonies and inadvertently had been bred to homozygosity (23, 24). This mutation complicates interpretation of the relationship between type I IFN and IRF5 expression and activation because Dock2^−/− mice have independent immune system defects, including effects on IFN induction (25–27). After the Dock2 loss-of-function mutation was corrected, plasmacytoid dendritic cells stimulated with TLR agonists or HSV ex vivo showed small defects in IFN-α production, more substantive reductions in IFN-β, and larger decreases in IL-6 and TNF-α levels (23, 24). However, pathogenesis studies or detailed analysis of cytokine production in vivo after viral infection has not been analyzed in these newly generated Irf5^−/− × Dock2^wt/wt mice.

To better define the antiviral and immunoregulatory function of IRF5 in vivo, we infected Irf5^−/− × Dock2^wt/wt mice with WNV in the present study. These mice were highly vulnerable to infection with WNV, with most Irf5^−/− × Dock2^wt/wt animals succumbing to disease. This increase in mortality was associated with enhanced viral replication in peripheral tissues as well as in the brain and spinal cord. The loss of IRF5 was associated with decreased levels of proinflammatory cytokines and immune cell activation with a relatively preserved type I IFN response. Overall, our studies indicate that IRF5 transcriptional activity has key functions in orchestrating the early immune response in the draining lymph node (DLN) after WNV infection.

MATERIALS AND METHODS

Viruses and cells.

The WNV strain (3000.0259) was isolated in New York in 2000 and passaged once in C6/36 Aedes albopictus cells. Mice were inoculated subcutaneously in the footpad with 10² PFU of WNV diluted in Hanks balanced salt solution (HBSS) and 1% heat-inactivated fetal bovine serum (FBS). Virus titers in tissues were analyzed by plaque assay using Vero cells, as described previously (28).

Mice.

C57BL/6J wild-type (WT) mice were commercially obtained from Jackson Laboratories. Irf5^−/− mice (18) originally were a gift from T. Taniguchi (Tokyo, Japan) and obtained from I. Rifkin (Boston, MA) and K. Fitzgerald (Worcester, MA) and had been backcrossed eight generations. After detection of a homozygous Dock2 mutation in this line, we backcrossed the line for an additional five generations and selected animals that were Dock2^wt/wt using PCR-based genotyping (23). Irf5^fl/fl mice (29) on a C57BL/6 background were a gift from P. Pitha (Baltimore, MD) and crossed to congenic CMV-Cre recombinase mice (Jackson Laboratories) to generate an independent gene deletion of IRF5. Deletion of the floxed region results in a frameshift mutation that creates a premature stop codon in exon 3 of the Irf5 gene (29). All mice were housed in a pathogen-free mouse facility at the Washington University School of Medicine and experiments were performed in accordance with federal and university regulations. The protocols were approved by the Institutional Animal Care and Use Committee at the Washington University School of Medicine (assurance number A3381-01). Mice (9 to 10 weeks old) were inoculated subcutaneously via footpad injection with 10² PFU of WNV diluted in 50 μl of HBSS supplemented with 1% heat-inactivated FBS.

For immunization studies, a purified, H₂O₂-inactivated WNV-Kunjin vaccine (30) was used. WT and Irf5^−/− mice were immunized subcutaneously in the footpad with H₂O₂-WNV-KUNV containing ∼10⁷ PFU equivalents/μg protein that was adjuvanted with Matrix M1 (10 μg/mouse; Novavax, Inc.) in phosphate-buffered saline (PBS). Serum samples were obtained on days 14 and 28, and mice were boosted on day 30. At day 60, a final serum sample was obtained, and bone marrow and splenocytes were harvested for long-lived plasma cell (LLPC) and memory B cell (MBC) analysis (see below).

The genotyping primers spanning the Dock2 exon duplication were published recently (ln29.4F, 5′-GAC CTT ATG AGG TGG AAC CAC AAC C-3′; lnR22.3.1R, 5′-GAT CCA AAG ATT CCC TAC AGC TCC AC-3′) (24). Genotyping primers for an internal control gene (CD19) to confirm adequacy of DNA sample preparation also have been published (oIMR1589F, 5′-CCT CTC CCT GTC TCC TTC CT-3′; oIMR1590R, 5′-TGG TCT GAG ACA TTG ACA ATC A-3′) (24).

Measurement of viral burden.

At specified time points (i.e., day 1, 2, 3, 4, 5, 6, 8, or 10) after WNV infection, serum was obtained by intracardiac heart puncture, followed by extensive perfusion (20 ml of PBS) and organ recovery. Organs were weighed, homogenized using a bead-beater apparatus, and titrated by plaque assay on Vero cells (28) or quantitative reverse transcription-PCR (qRT-PCR), as described previously (14).

Quantification of type I IFN activity.

Levels of biologically active type I IFN were determined by using an encephalomyocarditis virus (EMCV)-based cytopathic effect bioassay performed in murine L929 cells as described previously (8). Serum samples were treated with citrate buffer (40 mM citric acid, 10 mM KCl, 135 mM NaCl [pH 3.0]) for 10 min and neutralized with medium containing 50 mM HEPES (pH 8.0). The amount of type I IFN per ml of serum was calculated from a standard curve using IFN-β (PBL InterferonSource).

Cytokine Bio-Plex assay.

WT and Irf5^−/− mice were infected with WNV, and at specified times blood and the draining popliteal lymph nodes (LNs) were collected. Serum was isolated, and an LN homogenate was prepared. Individual LNs were homogenized in 250 μl of PBS using a bead-beater apparatus. The Bio-Plex Pro assay was performed according to the manufacturer's protocol (Bio-Rad).

Flow cytometry analysis of LN and spleen cells.

Popliteal LNs or spleens were dissected from naive or WNV-infected mice. Single-cell suspensions were generated in Dulbecco modified Eagle medium (DMEM) containing 10% FBS. After erythrocyte lysis with ACK buffer (0.15 M NH₄Cl, 10 mM KHCO₃, 0.1 mM Na₂EDTA [pH 7.2]), cells were counted by hemocytometry, washed with PBS supplemented with 1% FBS, and incubated with conjugated antibodies. The following antibodies (clone names are indicated in parentheses) against mouse antigens were purchased from BD Biosciences, eBioscience, or Biolegend: anti-CD3ε-Alexa Fluor 647 (145-2C11), anti-CD4-phycoerythrin (PE; RM4-5), anti-CD8α-Alexa Fluor 700 (53-6.7), anti-CD11b-PE.Cy7 (M1/70), anti-CD11c-BD Horizon V450 (HL3), anti-CD19-PE.Texas Red (1D3), anti-CD69-Brilliant Violet 605 (H1.2F3), anti-NK1.1-PE (PK136), anti-Ly6G-PerCP.Cy5.5 (1A8), anti-MHC-II-APC.Cy7 (I-A/I-E; M5/114.15.2), anti-F4/80-fluorescein isothiocyanate (FITC; BM8), and fixable viability dye Efluor 506. Cells were stained in PBS with 1% FBS supplemented with Fc block (anti-CD16/32) to minimize nonspecific binding, and appropriate isotype control antibodies purchased from Becton Dickinson Biosciences were used. The data were collected on a LSRII flow cytometer and analyzed using FlowJo software (Treestar, Inc.).

qRT-PCR.

WT or Irf5^−/− mice were infected with WNV, and draining popliteal LNs were harvested 2 days later. Total RNA was extracted by using an RNeasy kit (Qiagen). Fluorogenic qRT-PCR was performed using One-Step RT-PCR master mix and a 7500 Fast real-time PCR system (Applied Biosystems) with published TaqMan primers and probes (14). Gene induction was normalized to 18S rRNA levels.

B cell and antibody responses.

The levels of WNV-specific IgM and IgG (including IgG1, IgG2b, IgG2c, and IgG3) were determined using an enzyme-linked immunosorbent assay (ELISA) against purified WNV E protein, as described previously (31). Focus reduction neutralization assays on Vero cells were performed after mixing serial dilutions of serum with a fixed amount (50 FFU) of WNV as previously described (32).

Enzyme-linked immunospot (ELISPOT) assays to determine the number of antigen-specific LLPCs were performed as described previously (33). Briefly, mixed cellulose filter plates (Millipore) were coated with 20 μg of recombinant WNV E protein/ml in PBS overnight at 4°C. The plates were washed twice with PBS with 0.1% Tween 20, washed twice with PBS, and blocked for 1 h at room temperature with DMEM supplemented with 10% FBS, penicillin-streptomycin, 10 mM HEPES, 50 μM β-mercaptoethanol, and 10 mM nonessential amino acids. Bone marrow cells from immunized mice were depleted of erythrocytes using ACK lysis buffer. Cells were resuspended at 2 × 10⁶ cells/ml and serially diluted into wells. After 8 h at 37°C, the plates were washed extensively and incubated with 1 μg/ml biotinylated anti-IgG (Sigma-Aldrich) and then incubated overnight at 4°C. After washing, the wells were incubated with 1 μg/ml streptavidin conjugated to HRP diluted in PBS, 0.05% Tween 20, and 1% FBS. The plates were washed, and spots were developed using a chromogen substrate (AEC [3-amino-9 ethyl-carbazole]; Sigma-Aldrich). Spots were enumerated using a Biospot Analyzer (Cellular Technology, Ltd.).

WNV-specific MBC frequencies were determined as described previously (33) by measuring the reactivity of supernatants to recombinant WNV E protein by limiting dilution analysis (LDA) and ELISA. Splenocyte suspensions were generated from immunized mice. Erythrocytes were lysed (ACK lysis buffer; Invitrogen), and CD19⁺ B cells were isolated by positive selection using magnetic beads (Miltenyi Biotec). B cells were resuspended in LDA medium (RPMI 1640, 10% FBS, penicillin-streptomycin, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 10 mM HEPES, and 50 μM β-mercaptoethanol). B cells were cultured in 2-fold serial dilutions starting at 200,000 cells per well in 96-well flat-bottom plates on top of 20,000 3T3 transgenic feeder cells expressing BAFF and CD40L and treated with 5 μg/ml mitomycin C. Plates were incubated at 37°C and 5% humidified CO₂ for 6 days. To calculate the frequency of WNV-specific MBC that produced IgG, 96-well ELISA plates were coated with 5 μg/ml WNV E protein overnight at 4°C. The plates were washed and blocked with PBS–0.05% Tween 20 supplemented with 2% (wt/vol) bovine serum albumin for 1 h at 37°C. Supernatants from LDA plates were added to the ELISA plates (50 μl per well), and detection of WNV-specific antibody was performed as described for the ELISPOT assay. Positive wells were defined as wells that scored 2-fold over the mean optical density of negative-control wells (wells containing feeder cells alone). Supernatant from cultured naive splenocytes did not score positive.

Cellular immune responses.

WT and Irf5^−/− mice were infected subcutaneously in the footpad with 10² PFU of WNV and, at 8 days after infection, the spleens and brains were harvested after extensive cardiac perfusion with PBS. Splenocytes were dispersed into single cell suspensions with a cell strainer. Brains were minced and digested with 0.05% collagenase D, 0.1 μg/ml trypsin inhibitor TLCK, and 10 μg/ml DNase I in HBSS supplemented with 10 mM HEPES (pH 7.4; Life Technologies). Cells were dispersed into single cell suspensions with a cell strainer and pelleted through a 30% Percoll cushion for 30 min (1,200 × g at 4°C). Intracellular IFN-γ staining was performed after ex vivo restimulation with a D^b-restricted NS4B immunodominant peptide using 1 μM peptide and 5 μg/ml brefeldin A (Sigma) as described previously (34). Cells were stained with the following antibodies and processed on an LSR II flow cytometer (Becton Dickinson): CD3 (Becton Dickinson, clone 145-2C11), CD4 (Biolegend, clone RM4-5), CD8β (Biolegend, clone YT5156.7.7), CD45 (Biolegend, clone 30-F11), CD11b (Becton Dickinson, clone M1/70), CD11c (Becton Dickinson, clone HL3) MHC-II (Biolegend, clone M5/114.15.2), CD43 (Biolegend, clone IM7), and IFN-γ (Becton Dickinson, clone XMG1.2). Flow cytometry data were analyzed using FlowJo software (Treestar).

Data analysis.

All data were analyzed using Prism software (GraphPad6, San Diego, CA). Kaplan-Meier survival curves were analyzed by the log-rank test. Differences in viral burden, cytokine levels, and cell numbers were analyzed by the Mann-Whitney test.

RESULTS

Susceptibility of Irf5-deficient mice to WNV infection.

Given that prior viral pathogenesis studies may have used Irf5^−/− mice containing a mutation in the Dock2 gene, we confirmed that our backcrossed Irf5^−/− strain had the WT Dock2 allele (Fig. 1A). To determine whether IRF5 was required for restricting WNV pathogenesis in vivo, we infected 9- to 10-week-old WT and Irf5^−/− × Dock2^wt/wt (here called Irf5^−/−) mice with 10² PFU of WNV (New York 2000 strain) by subcutaneous inoculation and monitored survival over time. Irf5^−/− mice exhibited enhanced mortality (81% versus 22%, P < 0.0001, Fig. 1B) and reduced average survival time (mean time to death: 10.5 and 12.2 days for Irf5^−/− and WT mice, respectively, P < 0.01) compared to WT mice.

FIG 1.

Survival and viral burden in WT and Irf5^−/− mice after peripheral inoculation of WNV. (A) Dock2 gene PCR results from genotyping of WT, Irf5^−/− × Dock2^mu/mu, and Irf5^−/− × Dock2^wt/wt mice with primers amplifying the region flanking the duplication. A 305-bp amplicon that spans the genomic duplication of Dock2^mu/mu is observed only in the parent Irf5^−/− mice prior to backcrossing. It is not amplified in the new Irf5^{−/− ×} Dock2^wt/wt mice. Note, the other sized bands are nonspecific amplification products. (B) Irf5^−/− × Dock2^wt/wt (herein referred to as Irf5^−/−) mice were infected with 10² PFU of WNV in the footpad. Survival was monitored for 30 days: n = 31 WT and 45 Irf5^−/− mice from at least three independent experiments. The difference was statistically different (P < 0.0001), as judged by the log-rank test. (C to G) The viral burdens after WNV infection of WT and Irf5^−/− mice were measured by plaque assay in samples from serum (C), spleen (D), kidney (E), brain (F), and spinal cord (G). A scatter plot of data is shown with each individual point representing a single animal. The bars indicate mean values and were obtained from 7 to 11 mice per time point. The dashed lines represent the limit of sensitivity of the assay. Asterisks indicate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001), as judged by the Mann-Whitney test. (H and I) CMV-Cre Irf5^fl/fl and control Irf5^fl/fl (Cre-negative) mice were infected with 10² PFU of WNV in the footpad. (H) Survival was monitored for 30 days: n = 9 CMV-Cre Irf5^fl/fl and 12 Irf5^fl/fl mice from at least three independent experiments. The difference was statistically different (P < 0.002), as judged by a log-rank test. (I) Viremia was assessed by qRT-PCR at day 4 after WNV infection in CMV-Cre Irf5^fl/fl and control Irf5^fl/fl (Cre-negative) mice and expressed as focus-forming unit (FFU) equivalents after comparison to a standard curve. The bars indicate mean values and were obtained from nine mice per group. The difference was statistically significant (*, P < 0.05), as judged by the Mann-Whitney test. (J to N) Viral burdens in the cerebral cortex (I), cortical white matter (J), brain stem (K), olfactory bulb (L) and cerebellum (M) in the brains of WT and Irf5^−/− mice on day 9 after subcutaneous WNV infection. The samples (n = 7) were analyzed as described in panels D to H.

To begin to determine the basis for the increased lethality in Irf5^−/− mice after WNV infection, we measured viral loads in tissues at different time points. An absence of IRF5 resulted in enhanced viremia (9- to 10-fold at days 2 and 4 after infection, P < 0.05; Fig. 1C) and increased WNV infection in the spleen (31-fold at day 4, P < 0.01; Fig. 1D). A small effect was observed in the kidney at day 4, since 5 of 11 Irf5^−/− mice had measurable infectious WNV compared to 0 of 10 WT mice (Fig. 1E).

Higher levels of WNV infection also were observed in central nervous systems (CNS) of Irf5^−/− mice compared to WT controls. At day 4 after infection, 4 of 11 Irf5^−/− mice had measurable viral loads in the brain compared to 0 of 7 WT mice (Fig. 1F). Analogously, 5 of 10 Irf5^−/− mice had detectable WNV in the spinal cord at day 6 compared to 0 of 10 WT mice (Fig. 1G). This pattern resulted in increased titers at later times in the brain (22- to 145-fold increase at days 6, 8, and 10, P < 0.01) and spinal cord (38-fold at day 10, P < 0.01) of Irf5^−/− mice. At day 9 after infection, higher levels of WNV were detected in several different regions of the brain, including the cerebellum, cerebral cortex, brain stem, white matter, and olfactory bulb (Fig. 1J to N, P < 0.05).

To corroborate these results, we crossed Irf5^fl/fl mice (29) to a congenic CMV-Cre recombinase strain to generate an independent line with a targeted deletion of IRF5. Analogous to the data obtained with Irf5^−/− mice, 9-week-old CMV-Cre Irf5^fl/fl mice succumbed to WNV infection at a greater rate than age-matched control Cre-negative Irf5^fl/fl mice (89% versus 25%, P < 0.002, Fig. 1H) and exhibited a higher viremia (8-fold at day 4 after infection, P < 0.04; Fig. 1I).

Because previous studies showed antiviral effects of the TLR and RLR signaling pathways in resident cells of the CNS (12, 35–37), we hypothesized that part of the enhanced WNV infection phenotype in Irf5^−/− mice could be explained by an IRF5-dependent restriction of replication in resident neuronal cells or microglia. To test this, WT and Irf5^−/− mice were infected with 10¹ PFU of WNV directly into the brain via an intracranial route, and viral burden in the cerebral cortex, white matter, brain stem, cerebellum, and spinal cord was measured on days 2, 4, and 6 after infection. In contrast to the increased virus titers observed in the CNS of Irf5^−/− mice after peripheral inoculation, we observed no significant differences in infection between WT and Irf5^−/− mice following intracranial injection (Fig. 2A to E, P > 0.05). These data show that IRF5 is not required to restrict WNV replication directly in the CNS. More likely, the increased WNV titers in the CNS after subcutaneous inoculation reflect the enhanced infection in peripheral organs, which results in dissemination to the CNS prior to evolution of a protective adaptive immune response.

FIG 2.

Viral replication in CNS tissues of WT and Irf5^−/− mice after intracranial infection. (A to F) Mice were infected with 10¹ PFU of WNV by intracranial injection. Viral replication was measured in the cerebral cortex (A), white matter (B), brain stem (C), olfactory bulb (D), cerebellum (E), and spinal cord (F) by plaque assay from 2 to 6 days after infection. A scatter plot of the data is shown, with each individual point representing a single animal. The bars indicate mean values and were obtained from six mice per time point. The dashed lines represent the limit of sensitivity of the assay. None of the differences between WT and Irf5^−/− mice were statistically significant, as judged by the Mann-Whitney test.

Effect of IRF5 on innate immune responses.

Since IRF5 has been reported to be required for an optimal type I IFN response in vivo in the context of VSV, HSV, and NDV infections (20, 21), we assessed its impact on the systemic levels of type I IFN after WNV infection (Fig. 3A). However, we observed no significant difference in type I IFN levels in the serum of Irf5^−/− mice using a sensitive EMCV-based bioassay in L929 cells at days 1, 2, or 4 after infection. A small, yet statistically significant decrease in type I IFN levels was observed in the serum of Irf5^−/− mice at day 3 after infection.

FIG 3.

Systemic cytokine and chemokine responses after WNV infection of WT and Irf5^−/− mice. Mice were infected with 10² PFU of WNV in the footpad, and serum was harvested at days 1, 2, 3, or 4 after infection. (A) The type I IFN activity in serum of WT and Irf5^−/− mice was measured by an EMCV cytopathic effect bioassay. (B to G) Cytokine and chemokine levels in the sera of WNV-infected WT and Irf5^−/− mice were measured by Bio-Plex assay. The data are presented as scatter plots from 6 to 10 mice per group. Bars represent the means, and dotted lines indicate the limit of sensitivity of each assay. Asterisks indicate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001), as judged by the Mann-Whitney test.

We assessed the role of IRF5 on systemic production of other cytokines and chemokines at 1, 2, or 4 days after WNV infection (Fig. 3B to G). Notably, serum levels of IL-6, IL-10, IL-12p40, macrophage inflammatory protein 1α (MIP-1α; CCL3), MIP-1β (CCL4), and macrophage chemoattractant protein 1 (MCP-1; CCL2) were lower in serum within the first 2 days of WNV infection in Irf5^−/− compared to WT mice. By day 4, however, differences in these cytokine and chemokine levels in serum were largely absent, indicating that IRF5-independent pathways could compensate for their production. In comparison, levels of other inflammatory cytokines and chemokines (IFN-γ, TNF-α, IL-1β, and RANTES [CCL5]) were similar in WT and Irf5^−/− mice (data not shown). Overall, these results consistently show small defects in the early serum inflammatory cytokine and chemokine response of WNV-infected Irf5^−/− mice and extend results observed previously with VSV, HSV, or NDV infection (20, 21).

Because the differences in serum levels of proinflammatory cytokines between WT and Irf5^−/− mice occurred early after infection, we speculated this might reflect events occurring in the draining popliteal LN (DLN), an initial site of WNV replication and innate immune response after subcutaneous footpad inoculation (38). We first compared levels of viral RNA in the DLN in WT and Irf5^−/− mice. Notably, at day 2 after infection, no difference in WNV infection was observed in the two groups of mice (P > 0.7, Fig. 4A). We next measured the levels of cytokines and chemokines in homogenates prepared from DLN of WNV-infected WT and Irf5^−/− mice. The levels of 11 different cytokines and chemokines were consistently lower in WNV-infected Irf5^−/− mice compared to WT mice (Fig. 4B to L). In comparison, the mRNA levels of IFNβ, IFNα, and two ISGs (Rsad2 and Oas1) were minimally different in DLN samples from WT and Irf5^−/− mice (Fig. 4M).

FIG 4.

Cytokine and chemokine responses in the draining popliteal LNs after WNV infection of WT and Irf5^−/− mice. Mice were infected bilaterally with 10² PFU of WNV in the footpad, and LNs were harvested at day 2 after infection. (A) Levels of viral infection in LN homogenates as measured by qRT-PCR. The differences were not statistically different as measured by the Mann-Whitney test. (B to L) Cytokine and chemokine levels in LN of WNV-infected WT and Irf5^−/− mice were measured by a Bio-Plex assay. The data are presented as scatter plots from 6 to 11 independent measurements, reflecting a total of 12 to 22 LNs over three or four independent experiments for each genotype. Bars represent the mean and dotted lines indicate the limit of sensitivity of each assay. (M) Expression of the indicated IFNs and ISG in the DLN was measured from total RNA 2 days after WNV infection of WT and Irf5^−/− mice by qRT-PCR. Gene expression was compared to levels of 18S rRNA in the sample and displayed as the ΔC_T value. The data represent the averages from eight samples from two independent experiments. In this figure, asterisks indicate statistical significance (*, P < 0.05; **, P < 0.01; ****, P < 0.0001) as judged by the Mann-Whitney test.

While measuring cytokine and chemokine levels in DLN homogenates, we noticed the DLN from WNV-infected Irf5^−/− mice were smaller in size in comparison to their WT counterparts (data not shown). We explored this by assessing whether a deficiency of IRF5 affected the number and activation state of immune cell subsets. In the popliteal LN of Irf5^−/− naive mice, we measured 1.5- to 4-fold fewer cells in several leukocyte subsets, with the differences in NK cells and macrophages attaining statistical significance (Table 1). By day 2 after WNV infection, we observed 2- to 3-fold fewer B cells, CD4⁺ T cells, CD8⁺ T cells, NK cells, F4/80⁺ macrophages, and CD11c⁺ dendritic cells (Fig. 5A to F, P < 0.05) in the DLN of WNV-infected Irf5^−/− mice compared to WT mice. No difference in the numbers of Ly6G⁺ neutrophils was detected in the DLN at this time point (data not shown). Moreover, B cells, CD4⁺ T cells, and CD8⁺ T cells from Irf5^−/− mice expressed the activation marker CD69 on their surface at a lower frequency (P < 0.001, Fig. 5G to I), and Irf5^−/− macrophages expressed lower level of class II major histocompatibility complex (MHC) antigen (P < 0.001, Fig. 5K) than did cells from WT mice. Because of the differences in cellularity of the LNs in Irf5^−/− mice, we performed a similar analysis of splenocyte subsets in naive and WNV-infected animals. In contrast to the differences observed in the LN, in the spleen we did not detect a difference in the number B cells, CD4⁺ T cells, CD8⁺ T cells, NK cells, dendritic cells, or granulocytes in WT and Irf5^−/− mice at baseline or 2 days after WNV infection (Tables 1 and 2). Collectively, these experiments suggest that IRF5 has a key role in shaping the early innate cellular immune response in the DLN after subcutaneous infection with WNV.

TABLE 1.

Number of cells in the naive popliteal LNs and spleens of WT and Irf5^−/− mice^a

Cell type	Mean no. of cells ± SD		P
	WT mice	Irf5−/− mice
Lymph node
CD19+ B cells	2.9 × 105 ± 8.9 × 104	7.5 × 104 ± 3.7 × 104	0.07
CD4+ T cells	1.6 × 105 ± 2.7 × 104	8.1 × 104 ± 4.2 × 104	0.13
CD8+ T cells	1.3 × 105 ± 2.3 × 104	5.7 × 104 ± 2.8 × 104	0.11
NK1.1+ NK cells	4.8 × 103 ± 1.0 × 103	1.2 × 103 ± 3.3 × 102	0.04
CD11b+ F4/80+ macrophages	1.9 × 103 ± 9.9 × 102	3.2 × 102 ± 1.7 × 102	0.02
CD11c+ DCs	1.1 × 103 ± 8.2 × 102	1.9 × 102 ± 1.4 × 102	0.10
Lys6G+ granulocytes	9.0 × 102 ± 3.6 × 102	2.6 × 102 ± 1.1 × 102	0.11
Spleen
CD19+ B cells	2.6 × 107 ± 3.0 × 106	2.7 × 107 ± 3.0 × 106	0.68
CD4+ T cells	9.7 × 106 ± 8.9 × 105	1.1 × 107 ± 1.1 × 106	0.50
CD8+ T cells	5.9 × 106 ± 8.6 × 105	6.7 × 106 ± 7.0 × 105	0.52
NK1.1+ NK cells	2.1 × 106 ± 2.6 × 105	1.8 × 106 ± 2.1 × 105	0.57
CD11c+ DCs	5.3 × 104 ± 6.8 × 103	7.7 × 104 ± 1.0 × 104	0.14
Lys6G+ granulocytes	5.6 × 105 ± 1.0 × 105	1.5 × 106 ± 5.0 × 105	0.14

^aPopliteal lymph node cells and splenocytes were collected from uninfected WT and Irf5^−/− mice. Cells were phenotyped (see Materials and Methods) and processed by flow cytometry. The results are the averages from three to five mice per group. The groups were compared, and the indicated P values of <0.05 (Mann-Whitney test) reached statistical significance. DCs, dendritic cells.

FIG 5.

Leukocyte infiltration and activation in popliteal LNs of WNV-infected WT and Irf5^−/− mice. WT and Irf5^−/− mice were infected with 10² PFU of WNV in the footpad. Two days later, the draining popliteal LNs were harvested, and the cells were counted and stained with monoclonal antibodies to detect specific cell populations, including CD19⁺ B cells (A), CD4⁺ T cells (B), CD8⁺ T cells (C), NK1.1⁺ NK cells (D), CD11b⁺ F4/80⁺ macrophages (E), and CD11c⁺ dendritic cells (F). B cells (G) and T cells (H and I) were costained with MAbs to CD69 and macrophages were stained with MAbs to class II MHC (MHC II) (J) to establish activation state. The results are the pooled from two independent experiments with a total of 9 to 10 mice per group, and bars indicate the standard errors of the mean. Asterisks indicate statistical significance as judged by the Mann-Whitney test (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

TABLE 2.

Number of immune cells in the spleens of WT and Irf5^−/− mice at day 2 after WNV infection^a

Cell type	Mean no. of cells ± SD		P
	WT mice	Irf5−/− mice
CD19+ B cells	2.2 × 107 ± 2.0 × 106	2.3 × 107 ± 2.1 × 1046	0.67
CD4+ T cells	9.2 × 106 ± 1.0 × 106	1.1 × 107 ± 7.8 × 105	0.25
CD8+ T cells	5.8 × 106 + 5.7 × 105	6.4 × 106 ± 2.9 × 105	0.20
NK1.1+ NK cells	1.7 × 106 ± 2.1 × 105	1.8 × 106 ± 2.1 × 105	0.91
CD11c+ DCs	4.2 × 104 ± 7.5 × 103	6.5 × 104 ± 1.0 × 104	0.07
Lys6G+ granulocytes	6.2 × 105 ± 1.7 × 105	1.0 × 106 ± 4.8 × 105	0.43

^aSplenocytes were collected from WT and Irf5^−/− mice 2 days after WNV infection. Cells were phenotyped (see Materials and Methods) and processed by flow cytometry. The results are the averages from five to seven mice per group. None of the differences between groups attained statistical significance (P > 0.05). DCs, dendritic cells.

Effect of IRF5 on adaptive cellular immune responses.

Prior studies revealed that an optimal T cell response to WNV requires MAVS signaling (12) and that IRF5-dependent transcription in dendritic cells is regulated in part by MAVS-dependent signaling events (14). Given these findings and our current results showing that Irf5^−/− mice had lower levels of proinflammatory cytokines in the DLN and higher levels of WNV infection in organs that are cleared by effector T cells (39–42), we evaluated whether defects in cellular immune responses in peripheral tissues or the brain in Irf5^−/− mice contributed to the susceptibility phenotype after WNV infection.

We initially harvested splenocytes from WT and Irf5^−/− mice at 6 days after WNV infection, a relatively early time point in the evolution of the cellular response, and performed immunophenotyping analysis. Cells were stained with D^b-NS4B peptide tetramers to establish antigen specificity and antibodies to granzyme B to monitor effector function (Fig. 6A to C). No significant difference in the percentage of number of NS4B-tetramer positive cells expressing Granzyme B in WNV-infected WT and Irf5^−/− mice was detected at day 6 (P > 0.1). We also analyzed the cellular response in the spleen at a later time, day 8 after infection. Cells were stained with antibodies to detect T cells (CD3, CD4, and CD8) and the production of effector molecules (IFN-γ) after WNV peptide restimulation and to detect D^b-NS4B peptide tetramers. Again, we observed no significant differences in the percentage or total number of CD4⁺ or CD8⁺ T cells (Fig. 6D and E, P > 0.4) or antigen-specific tetramer-positive CD8⁺ T cells (Fig. 6F and G, P > 0.7) in the spleens of WNV-infected WT and Irf5^−/− mice. Consistent with this, after ex vivo peptide restimulation, we detected no difference in the percentages or numbers of CD8⁺ T cells producing IFN-γ (Fig. 6H and I, P > 0.5) or TNF-α (data not shown).

FIG 6.

T cell responses in the spleen after WNV infection. WT and Irf5^−/− mice were infected with 10² PFU of WNV in the footpad. Splenocytes were harvested and analyzed by flow cytometry at day 6 (A to C) or 8 (D to I) after infection. The percentages and numbers of NS4B tetramer⁺ CD8⁺ T cells (A and F), granzyme B⁺ CD8⁺ T cells (B), total CD8⁺ T cells (D), total CD4⁺ T cells (E), or IFN-γ-expressing CD8⁺ T cells after restimulation with an immunodominant NS4B peptide (H) are shown. Symbols represent individual mice. (C, G, and I) Representative flow cytometry plots of NS4B tetramer⁺ (C and G), granzyme B⁺ (C), and IFN-γ⁺ CD8⁺ (I) T cells from WT and Irf5^−/− mice at days 6 or 8 after infection. Numbers indicate the percentage of cells in each quadrant. The results are pooled from two independent experiments with a total of 5 to 11 mice per group depending on the experiment. None of the comparisons were statistically different, as judged by the Mann-Whitney test.

Although T cell responses were intact in the spleens from Irf5^−/− mice, it remained possible that the viral burden phenotype in the CNS was due, in part, to impaired immune cell trafficking across the blood-brain barrier or altered function of cells within the CNS. To address this, we analyzed immune cell accumulation in the brain at day 8 after WNV infection. We observed no difference in the percentages or numbers of NS4B-specific CD8⁺ T cells (Fig. 7A and B, P > 0.2) or IFN-γ-producing CD8⁺ T cells after peptide restimulation (Fig. 7C and D, P > 0.2) in the brain. Analogously, the percentages or numbers of CD11b⁺CD45^lo microglia or infiltrating CD11b⁺CD45^hi macrophages (Fig. 7E and F, P > 0.2) in the brains of WNV-infected WT and Irf5^−/− mice were similar. Thus, the increased WNV titers observed in the CNS of Irf5^−/− mice were not due to major defects in the generation or localization of antigen-specific T cells or other inflammatory cells.

FIG 7.

Immune cell responses in the brain after WNV infection. WT and Irf5^−/− mice were infected with 10² PFU of WNV in the footpad. Brain leukocytes were harvested and analyzed by flow cytometry at 8 days after infection. Percentages and numbers of NS4B tetramer⁺ CD8⁺ T cells (A), IFN-γ expressing CD8⁺ T cells after restimulation with an immunodominant NS4B peptide (C), and microglia (CD11b⁺CD45^lo) and macrophages (CD11b⁺CD45^hi) (E) are shown. Symbols represent individual mice. (B, D, and F) Representative flow cytometry plots of NS4B tetramer⁺ (B), IFN-γ⁺ CD8⁺ T cells (D), and microglia/macrophages (F) from the brains of infected WT and Irf5^−/− mice. Numbers indicate the percentage of cells in each quadrant. The results are the pooled from two independent experiments, with a total of 11 mice per group. None of the differences were statistically significant, as judged by the Mann-Whitney test.

Effect of IRF5 on humoral immune responses.

Prior studies have suggested that IRF5 regulates Ikaros expression in B cells to generate an optimal class-switched antibody response to T-dependent and -independent antigens (29) and that an absence of IRF5 alters B cell maturation (43), which prevents the development of antinuclear antibodies in a murine model of systemic lupus erythematosis (SLE) induced by pristine injection (44). Given that recent studies suggested that at least some of the cell-intrinsic B cell defects originally described could be due to the adventitious homozygous mutation of Dock2 that was present in many Irf5^−/− mouse colonies (23, 24), we analyzed humoral responses to WNV infection in our Irf5^−/− mice, which have a WT Dock2 allele. This analysis also was important because the B cell response to WNV is critical for controlling viremia and dissemination into the CNS (28, 32, 45).

To assess the effect of IRF5 on early WNV-specific antibody responses, we analyzed serum from Irf5^−/− and WT mice 4 and 8 days after infection for binding to WNV E protein. At day 4, we observed a small decrease in the magnitude of the IgM response in Irf5^−/− mice (2-fold, P < 0.01, Fig. 8A). By day 8, no difference was observed in the IgM response (Fig. 8B). At day 8, we also observed slightly lower levels (2-fold, P < 0.01, Fig. 8C) of WNV-specific IgG in Irf5^−/− mice. However, this difference was subtle and did not impact the neutralizing capacity of serum (Fig. 8D).

FIG 8.

B cell responses to live and inactivated WNV in WT and Irf5^−/− mice. (A to C) Serum was obtained from WNV-infected WT and Irf5^−/− mice, and the IgM levels at days 4 (A) and 8 (B) or the IgG levels at day 8 (C) after infection were measured by ELISA for reactivity with the WNV E protein. (D) Neutralizing antibody titers were determined by a focus-reduction assay from serum at day 8. The results are shown as a scatter plot and represent samples from seven to eight mice per group. The data are plotted as the reciprocal log₁₀ titer, and dotted lines indicate the limit of sensitivity of the ELISA. (E to H) B cell immune responses to inactivated WNV-Kunjin vaccine. WT and Irf5^−/− mice (n = 10 to 15 from two to three independent experiments) were immunized with adjuvanted WNV-Kunjin vaccine at day 0 and boosted at day 30. After vaccination, serum was obtained (days 14, 28, and 60), and splenic MBCs and bone marrow LLPCs (day 60) were harvested. (E) WNV-specific MBCs were determined by limiting dilution assay 6 days after polyclonal stimulation, as described in Materials and Methods. (F) WNV-specific LLPCs were determined by ELISPOT analysis. (G) Total Anti-WNV E IgG titers were determined from serum at the indicated days by ELISA. (H) Isotype analysis of WNV-specific MAbs was determined by ELISA on serum 28 days after immunization. Asterisks indicate statistical significance (*, P < 0.05; **, P < 0.01), as judged by the Mann-Whitney test.

One caveat to this analysis of the humoral response was that we observed 2-fold lower antibody titers in the setting of substantially increased viral replication, especially in the spleen (see Fig. 1D). In prior studies with mice deficient in IFN-γ (46), increased replication in peripheral organs was associated with higher antibody responses possibly secondary to the increased amount of viral antigen present. Thus, the higher level of antigen associated with enhanced infection of Irf5^−/− mice could mask defects in the evolution of the humoral response. Given this possibility, and the prior studies suggesting a B cell intrinsic effect of IRF5 (29), we assessed anti-WNV B cell responses in WT and Irf5^−/− mice after immunization with a nonreplicating H₂O₂-inactivated vaccine candidate (30). Mice were immunized with adjuvanted H₂O₂-inactivated WNV-Kunjin strain and boosted at day 30. One month later, the spleens and bone marrow were harvested for analysis of WNV-specific MBCs and long-lived plasma cells (LLPCs) by limiting dilution and ELISPOT analyses (33), respectively. Consistent with a role for IRF5 in optimal B cell responses (29), Irf5^−/− mice had fewer WNV-specific MBCs and LLPCs at day 60 after initial vaccination (Fig. 8E and F). Lower titers of WNV-specific antibodies in serum also were observed in immunized Irf5^−/− mice at day 60, 1 month after boosting (Fig. 8G). As reported previously (29), part of this defect reflected isotype skewing away from virus-specific antibodies of the IgG2c subtype (Fig. 8H). Similar to that seen with infectious WNV, the differences in serum and cellular antibody responses with nonreplicating inactivated virus between WT and Irf5^−/− mice were reproducible yet relatively small. Although IRF5 contributes to an optimal humoral response, it is not required for its generation.

DISCUSSION

Although the roles of IRF3 and IRF7 in regulating IFN-dependent and IFN-independent antiviral responses are well established, the significance of the transcription factor IRF5 in this response has remained less certain, in part due to variation in results obtained from different groups with Irf5^−/− mice. One explanation for these conflicting results was the recent discovery of a spontaneous genomic duplication and frameshift mutation in the guanine exchange factor Dock2 that had arisen spontaneously in several colonies of Irf5^−/− mice and inadvertently been bred to homozygosity (23, 24). Because Dock2-deficient mice have independent immune system defects, including effects on type I IFN induction (25–27), the relationship between IRF5 expression, type I IFN production, and viral pathogenesis required reassessment. In more recent studies, we used Irf3^−/− × Irf7^−/− DKO mice and Irf3^−/− × Irf5^−/− × Irf7^−/− TKO mice to assess whether IRF5 was a possible source of the residual induction of IFN-β and ISGs in dendritic cells (14). These experiments suggested that, in the absence of the transcription factors IRF3 and IRF7, IRF5 was responsible for mediating the type I IFN and ISG response after WNV infection. However, these studies did not explore the nonredundant functions of IRF5 and their effects on restriction of virus infection.

Here, using Irf5^−/− mice with WT Dock2 alleles, we established a protective role for IRF5 in the context of WNV infection. These results were corroborated with CMV-Cre Irf5^fl/fl mice, an independently generated mouse line with a targeted deletion of IRF5. An absence of IRF5 in vivo resulted in increased lethality and higher viral loads in the CNS compared to WT mice. Detailed virological and immunological analyses revealed that IRF5 contributed to the control of WNV at the earliest stages, since infection was increased in the blood and the spleen within a few days of inoculation, which ultimately resulted in enhanced replication in the brain and spinal cord. IRF5 had a subordinate role in the induction of type I IFN since relatively minor defects in the serum or DLN levels were detected in WNV-infected Irf5^−/− mice. In comparison, mice lacking IRF5 showed blunted production of proinflammatory cytokines and chemokines at early time points, which was associated with fewer immune cells being recruited to the DLN. Vaccination and infection studies in Irf5^−/− mice also revealed defects in the generation of an optimal acute and memory B cell response, although the antigen-specific CD8⁺ T cell response remained preserved. Collectively, our results suggest that the nonredundant role of IRF5 is primarily immunomodulatory, since its transcriptional activity shapes the magnitude, composition, and activation state of cellular infiltrates in lymphoid tissues. This activity modulates the early antiviral response and the transition to adaptive immunity, especially in the B cell compartment.

Our experiments suggest that in vivo, IRF5 does not have a major nonredundant role in regulating systemic type I IFN production, at least after WNV infection. These data are consistent with the original characterization of Irf5^−/− mice (18) in which defects in proinflammatory cytokine production but not IFN-α production were observed after the treatment of dendritic cells or animals with TLR agonists. These results vary with those reported after infection with other viruses, which showed reduced levels of IFN-α and IFN-β in serum of infected Irf5^−/− mice (20, 21), although some of the differences were small and did not attain statistical significance (20). While the basis for the discrepancy in nonredundant effects of IRF5 on the systemic type I IFN response warrants further analysis, there are at least two possible explanations. First, cell-type-specific effects of IRF5 in the context of infection by different viruses could impact the induction of type I IFN. In the context of NDV infection, an absence of IRF5 in cultured dendritic cells but not peritoneal macrophages was associated with reduced levels of type I IFN in the supernatant (20). Irf5^−/− plasmacytoid dendritic cells obtained directly from mice produced less IFN-α and IFN-β in response to TLR4 and TLR9 agonists (24, 47, 48) or myxoma virus (49). Analogous cytokine induction studies with WNV have shown greater type I IFN production from conventional myeloid compared to plasmacytoid dendritic cells (50). Alternatively, some of the earlier studies may have used mice carrying the Dock2 mutation, which independently affects migration or activity of IFN-producing immune cells (26, 27).

We did observe a role for IRF5 in regulating expression of proinflammatory cytokine and chemokine responses, especially at early times after WNV infection in serum and the DLN. This result agrees with studies in Irf5^−/− mice that used other inflammatory agonists or viruses as stimuli (18, 20, 21, 23). Nonetheless, the signaling pathway that induces these proinflammatory molecules remains uncertain. Although IRF5-dependent cytokine induction after TLR agonist stimulation occurs via MyD88- and/or TRIF-dependent pathways (18, 51), after viral infection it also depends on RLR- and MAVS-dependent signals (14, 52). Indeed, IRF5 has been shown to interact in a complex with RIG-I and MAVS via coimmunoprecipitation and immunoblotting experiments (14, 21, 53). Under certain circumstances, RLR signaling can antagonize IRF5-dependent cytokine production as activated IRF3 can bind dominantly, relative to IRF5, to the Il12b gene promoter and interfere with the TLR-induced assembly of a transcription-factor complex (54). As an additional level of regulation, histone deacetylases and acetyltransferases associate directly with IRF5 and modulate its transcriptional activity (55). Further studies in primary cells are needed to define the mechanistic intermediates that link IRF5 activation and signaling to cytokine induction after WNV infection.

Among the most significant phenotypes observed was the impact of IRF5 on leukocyte numbers in the DLN before and after WNV infection. This unexpected phenotype was not due to differences in virus trafficking or local replication because WNV titers were similar between WT and Irf5^−/− mice within 2 days of infection. Rather, our data suggest that both homeostatically and downstream of virus infection and pathogen detection pathways, IRF5 regulates expression of cytokines and chemokines that shape cellular trafficking into the DLN and activation of key cell types once they enter. In the absence of IRF5, fewer T cells, B cells, NK cells, macrophages, and dendritic cells were present in the DLN at day 2 after infection, and several lymphocyte subsets expressed smaller amounts of the activation marker CD69. Given that we observed no difference in WNV replication in dendritic cells from WT and Irf5^−/− mice (14) or in neurons after intracranial infection of WT and Irf5^−/− mice, our data are most consistent with a cell-extrinsic indirect antiviral function of IRF5 that reflects its immunomodulatory activity. Proinflammatory and leukocyte recruitment functions of IRF5 in the DLN have implications for its linkage to SLE, which can feature upregulated expression of IRF5 (56, 57), and for the metastatic potential of tumors where IRF5 expression is downregulated (58, 59).

Part of the enhanced vulnerability of Irf5^−/− mice to WNV infection also may be associated with defects in B cell responses. Despite the increase in viral infection and antigen in peripheral tissues, we observed blunted IgM and IgG responses at early time points. This could impact WNV pathogenesis, since the early humoral response is critical for controlling viremia. Even small differences in WNV-specific IgM levels at early times after infection affect viral dissemination to the brain and clinical outcome (28, 32). Because of the possible confounding influence of enhanced WNV infection and antigen burden on B cell responses in Irf5^−/− mice, we also performed immunization studies with a H₂O₂-inactivated nonreplicating WNV-Kunjin virus strain (30). In these experiments, we observed defects in Irf5^−/− mice after boosting in the magnitude and isotype skewing of the serum IgG responses, as well as the total numbers of antigen-specific LLPCs and MBCs. These results are consistent with the concept that IRF5 has a B-cell-intrinsic function through its ability to modulate the plasma cell maturation factor, Blimp-1 (43). While our data agree with the trends identified in a prior study showing altered isotype skewing in response to T-dependent or polyomavirus infection (29), the magnitude of the phenotype after WNV infection was subtle. Indeed, in a study by Fang et al., the antibody phenotype was more pronounced in the Irf5^−/− mice compared to Irf5^cre/cre mice (29), which could reflect the effect of the adventitious Dock2 mutation on T_H2-type effects (24).

In summary, our study shows that Irf5^−/− mice were more vulnerable to lethal WNV infection. This phenotype was associated with defects in accumulation and activation of immune cells in the DLN that likely reflected the loss of IRF5-dependent expression of proinflammatory cytokines and chemokines, with possible subordinate effects on the type I IFN response. The altered DLN response and deficits in the early antibody response correlated with increased WNV infection in the spleen and the bloodstream, which resulted in higher peak titers in the brain and spinal cord. Future studies with conditionally targeted mice are planned to define the specific cell types that use IRF5 to shape LN responses after viral infection or immunogen exposure. Given the wide range of effects of loss-of-function and gain-of-function IRF5 alleles, an improved understanding of its specific functions in innate and adaptive immunity could facilitate strategies that ameliorate clinical outcomes in the context of viral infection, autoimmunity, and neoplastic disease.