TLR3- and MyD88-Dependent Signaling Differentially Influences the Development of West Nile Virus-Specific B Cell Responses in Mice following Immunization with RepliVAX WN, a Single-Cycle Flavivirus Vaccine Candidate

Abstract

Recognition of conserved pathogen-associated molecular patterns (PAMPs) by host pattern recognition receptors (PRRs) results in the activation of innate signaling pathways that drive the innate immune response and ultimately shape the adaptive immune response. RepliVAX WN, a single-cycle flavivirus (SCFV) vaccine candidate derived from West Nile virus (WNV), is intrinsically adjuvanted with multiple PAMPs and induces a vigorous anti-WNV humoral response. However, the innate mechanisms that link pattern recognition and development of vigorous antigen-specific B cell responses are not completely understood. Moreover, the roles of individual PRR signaling pathways in shaping the B cell response to this live attenuated SCFV vaccine have not been established. We examined and compared the role of TLR3- and MyD88-dependent signaling in the development of anti-WNV-specific antibody-secreting cell responses and memory B cell responses induced by RepliVAX WN. We found that MyD88 deficiency significantly diminished B cell responses by impairing B cell activation, development of germinal centers (GC), and the generation of long-lived plasma cells (LLPCs) and memory B cells (MBCs). In contrast, TLR3 deficiency had more effect on maintenance of GCs and development of LLPCs, whereas differentiation of MBCs was unaffected. Our data suggest that both TLR3- and MyD88-dependent signaling are involved in the intrinsic adjuvanting of RepliVAX WN and differentially contribute to the development of vigorous WNV-specific antibody and B cell memory responses following immunization with this novel SCFV vaccine.

INTRODUCTION

Although originally endemic only in parts of Africa, Asia, and Europe, West Nile virus (WNV) spread to North America and was detected in New York State in 1999. In the following decade, it rapidly spread over the entirety of North America and into Central and South America, causing infection in humans ranging in severity from inapparent infection to encephalitis and death. WNV is considered a significant threat to public health, having caused 34,113 human infection cases and 1,487 deaths between 1999 and 2012 (1). The 2012 WNV outbreak in the United States resulted in 5,387 human disease cases, of which 243 cases resulted in death (1). At present there is no licensed WNV vaccine for humans, although several vaccine candidates have been developed (2, 3). Recently we developed RepliVAX WN, a single-cycle flavivirus vaccine candidate derived from a wild-type WNV strain by introduction of an internal deletion in the virus capsid gene (4, 61). By infecting a packaging cell line that constitutively expresses the WNV capsid protein, the mutated genome of RepliVAX WN can be packaged into WNV capsids and is able to normally infect host cells. However, in the absence of the complete capsid gene, the replicated viral genes from this single-cycle flavivirus (SCFV) fail to be packaged into an infectious particle. RepliVAX WN-infected cells release noninfectious subviral particles (SVPs) and the WNV nonstructural protein NS1, which stimulate vigorous anti-WNV immune responses in mice (5, 6), hamsters (7), and nonhuman primates (8). We have defined the important role of the innate immune response, specifically signaling through the type I interferon (IFN) receptor, in the development of WNV-specific adaptive immune responses (6). However, the manner in which the interplay between host and WNV-expressed pathogen-associated molecular patterns (PAMPs) shapes the developing humoral immune response is still poorly understood. In this study, we investigated the role of signaling through toll-like receptors (TLRs) in the development of B cell responses to RepliVAX WN immunization.

TLRs recognize conserved PAMPs expressed preferentially by viruses, bacteria, and parasites, and the recognition of different PAMPs differentially triggers specific TLR signaling pathways. Subsequently, inflammatory cytokines are released (9), and innate immune cells, including dendritic cells (10), are activated and play an important role in shaping humoral immunity (11). The double-stranded and single-stranded viral RNAs resulting from a WNV infection are recognized by TLR3 and TLR7/8, respectively. TLR3, which is localized in the endosome, recruits the adaptor molecule, TI-domain-containing adaptor-inducing beta interferon (TRIF), whereas activation of TLR7/8 induces TRIF-independent signaling through the myeloid differentiation primary response gene 88 (MyD88) adaptor molecule. Both of these signaling pathways stimulate the transcription of type I IFN and inflammatory cytokines, e.g., tumor necrosis factor (TNF) and interleukin 12 (IL-12) (12), and previous studies have shown that both TLR3/TRIF and TLR7/MyD88 signaling is important in the development of antiviral humoral immunity (13–17). However, the respective roles of these two independent signaling pathways in B cell development, when present together on the same immunogen, are not well understood. Insight into the roles of distinct PRR signaling pathways in development of humoral immunity is important for optimizing and tailoring immune responses induced by single-cycle vaccines and for selection of proper adjuvants to be used in subunit protein vaccines.

The development of an antigen (Ag)-specific B cell response is a complex process and includes B cell activation, expansion, and maturation in the germinal center (GC) and ultimately the development of B cell memory (18). In this study, we utilized TLR3^−/− and MyD88^−/− mice to investigate and compare the roles of TLR3- and MyD88-dependent signaling on these important events in B cell response development and to understand their role in generating an antibody response to RepliVAX WN. Our results demonstrate that loss of signaling through either pathway negatively influences the developing antibody response by impacting different stages of WNV-specific B cell response development and differentially affects the development of B cell memory for this SCFV vaccine.

MATERIALS AND METHODS

Mice.

C57BL/6J (B6) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). TLR3-deficient (TLR3^−/−) and MyD88-deficient (MyD88^−/−) mice on a B6 background were obtained from Michael Diamond (Washington University, St. Louis, MO) and maintained as a breeding colony at the Association for Assessment and Accreditation of Laboratory Animal Care- approved animal research center of the University of Texas Medical Branch. Animals were age and sex matched for all experiments. All animal research was approved by the Institutional Animal Care and Use Committee of the University of Texas Medical Branch with oversight of staff veterinarians.

Viruses and WNV antigen.

RepliVAX WN was generated in BHK (VEErep/Pac-Ubi-C*) cells as previously described (19). Firefly luciferase-expressing SCFV particles (FLUC-SCFV) containing a WNV replicon genome expressing a humanized FLUC gene were generated in BHK (VEErep/C*-prM-E-Pac) cells as previously described (20). RepliVAX WN and FLUC-SCFV were titrated using Vero cells as previously described (21). WNV subviral particles (SVPs) used as enzyme-linked immunosorbent spot (ELISPOT) and enzyme-linked immunosorbent assay (ELISA) capture antigens were produced by infecting Vero cells with RepliVAX WN. The cell culture supernatant containing the released WNV SVPs was harvested and concentrated using centrifugal filtration, and SVPs were purified on a sucrose gradient. WNV NS1 antigen was harvested and purified from transfected VEErep-bearing BHK cells as previously described (19).

Immunization with RepliVAX WN.

For quantifying WNV-specific serum IgM and IgG antibody titers and WNV-specific antibody-secreting cells (ASCs), mice were immunized by the intraperitoneal (i.p.) route with 10⁶ IU RepliVAX WN. To assess the level of viral gene replication and evaluate early B cell activation in secondary lymphoid tissues, mice were immunized by subcutaneous (s.c.) footpad (FP) injection with 10⁶ IU RepliVAX WN or FLUC-SCFV. Inocula were delivered in L-15 medium containing 10 mM HEPES and 0.5% fetal bovine serum (FBS).

Serological analysis.

An ELISA was used to titrate anti-NS1 and anti-SVP IgM and IgG antibodies as described previously (5). Briefly, the capture antigens, purified WNV SVPs and purified NS1 proteins, were absorbed on ELISA plates (Corning Incorporated, Corning, NY) overnight at 4°C. Plates were blocked, and serial dilutions of serum from B6, TLR3^−/−, and MyD88^−/− immunized and naive mice were added on the coated plates. Plate-bound IgG was developed with horseradish peroxidase (HRP)-IgG (Southern Biotechnology Associates, Inc., Birmingham, AL), HRP-IgM (Southern Biotechnology Associates, Inc.), or biotinylated anti-mouse IgG1 or IgG2c (BD Pharmingen, San Diego, CA), followed by incubation with streptavidin peroxidase (Sigma-Aldrich, St. Louis, MO). Normalized optical density (OD) readings at 490 nm (OD₄₉₀) obtained from serial dilution of serum were analyzed by nonlinear regression. The endpoint titer was defined as the serum dilution resulting in an OD₄₉₀ value equivalent to 3 standard deviations above OD₄₉₀ values from sera of mock-vaccinated animals.

Avidity was measured by a modified ELISA (22) in which serum samples were plated on NS1- or SVP-coated plates for 1 h followed by a 10-min incubation with 4 M, 6 M, or 8 M urea or saline as a control. Plates were developed as described previously (5), and OD₄₉₀ readings for each serum sample were obtained for urea-treated wells (OD_{490 urea}) and saline-treated wells (OD_{490 saline}). The avidity index for NS1- and SVP-specific IgG was calculated as OD_{490 urea}/OD_{490 saline}.

ELISPOT assay.

ELISPOT assays for ASCs were performed as described previously (5) using microtiter filter plates (Millipore Corporation, Billerica, MA) coated with purified WNV NS1 protein or purified WNV SVPs. Ag-specific ASCs were quantified using an ImmunoSpot reader and analyzed with the ImmunoSpot software program (Cellular Technology Ltd., Cleveland, OH).

Flow cytometry analysis and antibodies.

Popliteal lymph nodes (pLNs), inguinal LNs (iLNs), and spleens were collected from immunized and naive B6, TLR3^−/−, and MyD88^−/− mice. Single-cell suspensions were blocked with anti-Fc RII/III monoclonal antibody (MAb) and surface stained with antibodies purchased from BD Biosciences (San Jose, CA): anti-CD19–phycoerythrin (PE) (1D3), anti-CD69–peridinin-chlorophyll proteins-Cy5.5 (Percp-Cy5.5) (H1.2F3), and anti-CD86fluorescein isothiocyanate (FITC) (GL1). Anti-peanut agglutinin (PNA)–fluorescein isothiocyanate was purchased from Sigma-Aldrich (St. Louis, MO). Data were acquired on a BD LSRII Fortessa instrument (BD Biosciences, San Jose, CA) at the UTMB Flow Cytometry Core Facility and analyzed using the FlowJo software program (Tree Star, Ashland, OR).

IVIS.

The posterior half of B6, TLR3^−/− and MyD88^−/− mice was shaven prior to s.c. immunization in the FP with 10⁶ IU FLUC-SCFV. At 14 h postimmunization (hpi) and at 1, 2, 3, 4, 6, 8, and 10 days postimmunization (dpi), real-time in vivo imaging (IVIS) was performed using a Xenogen IVIS 200 imaging system (Caliper LS, Hopkinton, MA) on d-luciferin (Caliper LS)-treated and anesthetized mice (6). Images were analyzed by defining FP regions with FLUC activity and measuring total flux (photons per second [p/s]). Data were acquired using the Living Image 4.0 software program (Caliper LS) and reported as the average total flux from FP of all mice in an experimental group.

Neutralization assay.

Day 28 p.i. serum was pooled from groups of RepliVAX WN-immunized B6 (n = 35), TLR3^−/− (n = 35), and MyD88^−/− (n = 34) mice, and neutralization titers of the serum pools were determined using luciferase-expressing RepliVAX WN (FLUC-SCFV) as described previously (7, 19). The 50% neutralization titer was determined by nonlinear regression using the GraphPad Prism software program, version 5.0 (GraphPad Software, San Diego, CA). For experiments involving normalized serum pools, the endpoint titer of each serum pool was calculated by using ELISA. Serum pools were normalized by dilution of B6 and TLR3^−/− pools to achieve an endpoint titer equivalent to that of the MyD88^−/− serum pool. Equivalent endpoint titers were confirmed by serum titration curves on SVP ELISA plates (see Fig. 7A for an example). Neutralization titers of the normalized serum pools were determined as described previously (7, 19).

Fig 7.

Similar affinities and avidities for IgG antibodies from RepliVAX WN-immunized B6, TLR3^−/−, and MyD88^−/− mice. (A) Day 28 p.i., sera were pooled from groups of RepliVAX WN-immunized B6 (●) (n = 35), TLR3^−/− (■) (n = 35), and MyD88^−/− (▲) (n = 34) mice, and the endpoint titer of each serum pool was calculated by ELISA. B6 and TLR3^−/− serum pools were diluted to achieve a titer equivalent to that of the MyD88^−/− serum pool. Serum titration curves of the normalized serum pools generated by SVP-ELISA are shown. (B and D) Avidity indices of serum IgG from 28 dpi from individual B6 (filled bar) (n = 5), TLR3^−/− (open bar) (n = 5), or MyD88^−/− (striped bar) (n = 5) under the indicated urea wash conditions. (C and E) Avidity indices of serum IgG from 7 and 28 dpi from B6 (filled bar) (n = 8), TLR3^−/− (open bar) (n = 8), or MyD88^−/− (striped bar) (n = 5) under 8 M urea wash conditions. The results shown are from a representative experiment of 2 performed. ∗, P < 0.05 compared to results for B6 mice.

Statistical analysis.

Statistical differences for B lymphocyte assays, serum titer, and frequency and quantity of different cell compartments were determined using Student's t test (unpaired) or analysis of variance (ANOVA) with the Tukey posttest as appropriate. P values of <0.05 were considered significant. All calculations were performed using GraphPad Prism software, version 5.0 (GraphPad Software).

RESULTS

Diminished antibody response to RepliVAX WN in the absence of either TLR3- or MyD88-dependent signaling.

Stimulation of TLRs initiates innate immune responses, including the production of type I interferons (IFNs), which limit viral gene expression and therefore alter availability of viral proteins for induction of adoptive immunity. We therefore tested if SCFV gene expression would be increased in TLR3^−/− or MyD88^−/− mice immunized with RepliVAX WN. B6, TLR3^−/−, and MyD88^−/− mice were immunized with FLUC-SCFV, and FLUC gene expression at the FP site of injection was quantified between 14 hpi and 10 dpi using an in vivo imaging system (IVIS). Expression of SCFV-encoded FLUC was readily detected in FP of all infected mice as early as 14 hpi (Fig. 1A) but was not detected in uninoculated mice (Fig. 1B). In B6 mice, FLUC bioluminescence was maintained at a high level through 4 dpi and diminished thereafter to low levels on 10 dpi (Fig. 1C). Similar patterns of expression of the SCFV-encoded FLUC gene were detected in both TLR3^−/− and MyD88^−/− mice. The FLUC intensity detected in MyD88^−/− mice was initially approximately 2-fold lower at 2 dpi before increasing to levels nearly equivalent to that in B6 mice on 3 dpi. FLUC expression was maintained at levels equivalent to or slightly higher than those in B6 mice on days 4 through 10 p.i. FLUC expression in TLR3^−/− mice was very similar to that in B6 mice on all days tested.

Fig 1.

SCFV Gene expression in B6, TLR3^−/−, and MyD88^−/− mice. Groups of B6 (●), TLR3^−/− (■), or MyD88^−/− (▲) mice were inoculated s.c. in both rear FP with 10⁶ IU FLUC-SCFV and imaged at the indicated time points as described in Materials and Methods. (A) Representative images of FLUC-SCFV-inoculated mice. (B) Representative image of an uninoculated mouse. (C) FLUC gene expression in FLUC SCFV-inoculated B6, MyD88^−/−, and TLR3^−/− mice. Results are expressed as the mean bioluminescence from FP for each group. Error bars represent the standard errors of the means (SEM) for individual mice. ∗, P < 0.05 compared to results at the same time point for B6 mice. The limit of detection for the assay was 10⁴ photons/s.

TLR signaling directs the production of type I IFN and proinflammatory cytokines that have been shown to modulate the development of adaptive immune responses to several viruses (6, 23). Accordingly, we investigated if the TLR signaling pathways induced by the SCFV vaccine RepliVAX WN influenced the development of the WNV-specific antibody response. Groups of B6, TLR3^−/−, and MyD88^−/− mice were immunized i.p. with 10⁶ IU RepliVAX WN, and WNV-specific serum IgG responses were quantified on 7, 14, and 28 dpi. Vigorous anti-NS1 (Fig. 2A) and anti-SVP (Fig. 2B) IgG antibody titers were detected in B6 mice. In comparison, both TLR3^−/− and MyD88^−/− mice produced significantly less NS1-specific IgG (Fig. 2A) and SVP-reactive IgG (Fig. 2B) on 14 and 28 dpi. Although loss of either innate signaling pathway impaired the RepliVAX WN-specific humoral response, WNV antigen-specific serum IgG levels were most compromised in MyD88^−/− mice. Specific antibody responses in MyD88^−/− mice inoculated with a 10-fold-higher dose of RepliVAX WN (10⁷ IU) were still lower than those in B6 mice inoculated with 10⁶ IU RepliVAX WN (Fig. 2C and D), strongly suggesting that the diminished response in MyD88^−/− mice was the result of a cellular defect and not the result of an initially lower antigen load in MyD88^−/− mice (Fig. 1C).

Fig 2.

Diminished WNV-specific serum IgG titers in TLR3^−/− and MyD88^−/− mice. B6 (n = 10) (●), TLR3^−/− (n = 10) (■), and MyD88^−/− (n = 9) (▲) mice were immunized i.p. with 10⁶ IU RepliVAX WN. Anti-NS1 (A) or anti-SVP (B) serum IgG was measured by ELISA on 7, 14, and 28 dpi. Results are pooled from 2 experiments. Data are presented as the mean titer (log 10) ± SEM. ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001 (compared to results at the same time point for B6 mice). (C and D) Anti-NS1 (C) or anti-SVP (D) serum IgG titers from B6 (n = 5) (●) and MyD88^−/− (n = 5) (▲) mice immunized i.p. with 10⁶ IU (solid line) or 10⁷ IU (dashed lines) RepliVAX WN. ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001 (compared to results at the same time point for 10⁶ IU RepliVAX WN-immunized B6 mice).

Diminished B cell activation following RepliVAX WN immunization of MyD88^−/− and TLR3^−/− mice.

We investigated which cellular events during development of a WNV-specific B cell response were affected by loss of these signal pathways. Following events such as cross-linking of B cell receptors (BCRs) with antigen, binding of TLR ligands, or exposure to type I IFN, B cells become activated and express the early activation markers CD69 and CD86 (24–27). To determine if B cell activation was impaired by the lack of either TLR3- or MyD88-dependent signaling, we measured CD69 expression on B cells (CD19⁺ cells) at 0, 3, and 5 dpi (Fig. 3A). In the pLNs of B6 and TLR3^−/− mice, the frequency of CD69⁺ CD19⁺ cells peaked by 3 dpi before diminishing at 5 dpi. In contrast, the frequency of CD69⁺ CD19⁺ cells in MyD88^−/− mice peaked at a significantly lower level on 3 dpi (P < 0.001, ANOVA). The total number of CD69⁺ CD19⁺ B cells in pLNs of B6 mice rose on 3 dpi and was maintained at high levels through 5 dpi. The number of activated B cells for TLR3^−/− mice was similar to that for B6 mice on days 3 and 5 p.i. but was significantly reduced for MyD88^−/− mice on both days (Fig. 3B) (P < 0.05, ANOVA). A similar pattern of expression was also observed for the activation marker CD86. There was a gradual increase in the frequency and quantity of CD86⁺ CD19⁺ cells in pLNs on 3 and 5 dpi in all three mouse strains. TLR3^−/− and MyD88^−/− mice showed a significant reduction in the quantity of CD86⁺ CD19⁺ cells on 5 dpi (Fig. 3C) (P < 0.05, ANOVA). Further, although the quantity of CD86⁺ CD19⁺ cells did not significantly differ among B6, TLR3^−/−, and MyD88^−/− mice on 3 dpi, the level of CD86 expression measured as the mean fluorescence intensity (MFI) was significantly reduced on pLN B cells of TLR3^−/− and MyD88^−/− mice compared to results for B6 mice (Fig. 3D) (P < 0.05, ANOVA).

Fig 3.

Decreased expression of B cell activation markers following RepliVAX WN immunization of TLR3^−/− and MyD88^−/− mice. B6 (n = 4) (filled bar), TLR3^−/− (n = 4) (open bar), and MyD88^−/− (n = 4) (striped bar) mice were inoculated s.c. in the FP with 10⁶ IU RepliVAX WN. (A) Frequency of CD19⁺ B cells expressing CD69 in pLNs from B6, TLR3^−/−, and MyD88^−/− mice. (B) Total number of CD69⁺ CD19⁺ B cells in pLNs from B6, TLR3^−/−, and MyD88^−/− mice. (C) Total number of CD86⁺ CD19⁺ B cells from pLNs of immunized mice. (D) Mean fluorescence intensity (MFI) of CD86 expression on activated B cells (CD69⁺ CD19⁺) in pLNs from B6 (●), TLR3^−/− (■), and MyD88^−/− (▲) mice at 3 dpi. Results are from a representative experiment of 2 performed and are expressed as the means ± SEM (∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001).

To determine if the lack of TLR3- and MyD88-dependent signaling resulted in early functional deficits, mice were immunized s.c. in the FP with 10⁶ IU RepliVAX WN, and anti-SVP IgM antibody-secreting cells (ASCs) were quantified in pLNs, iLNs, and spleens on 0, 3, and 5 dpi. IgM ASCs from immunized mice were readily detected on SVP-coated wells but not on ovalbumin-coated wells, demonstrating the antigen specificity of the ASCs (Fig. 4A). SVP-reactive IgM ASCs were detected at all three lymphoid tissues of RepliVAX WN-immunized B6 mice on 3 dpi, and the number increased through 5 dpi (Fig. 4B, C, and D). SVP-reactive IgM ASCs were detected in the pLNs and spleens of TLR3^−/− mice on day 3, and numbers increased through 5 dpi. In MyD88^−/− mice, SVP-reactive IgM ASCs were not detected at any site on 3 dpi but were detected at low levels at all three sites on 5 dpi. Additionally, the number of anti-SVP IgM ASCs was significantly reduced on 5 dpi in both TLR3^−/− and MyD88^−/− mice (P < 0.001, ANOVA).

Fig 4.

The SVP-specific IgM antibody-secreting B cell response to RepliVAX WN immunization is diminished for TLR3^−/− and MyD88^−/− mice. (A) Specificity of ELISPOT assay for detecting SVP-reactive IgM ASCs. Representative ELISPOT wells coated with SVP (left and middle) or ovalbumin (OVA) (right) for detection of ASCs from naive (left) and RepliVAX WN-immunized (middle and right) mice. (B to D) SVP-specific IgM ASC response in pLNs (B), iLNs (C), or spleens (D) of RepliVAX WN-immunized mice (10⁶ IU RepliVAX WN, s.c. in FP). SVP-specific IgM ASCs from B6 (●) (n = 8), TLR3^−/− (■) (n = 8), and MyD88^−/− (▲) (n = 6) immunized mice were quantified by ELISPOT assay. The results shown are from a representative experiment of 2 performed. Data are presented as the means ± SEM. (∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001 (compared to results at the same time point for B6 mice).

Altered GC cellularity, WNV-specific serum IgG antibody, and ASC responses to RepliVAX WN in TLR3^−/− and MyD88^−/− mice.

The GC is the specialized site in lymphoid follicles for antigen-specific B cell proliferation, maturation, and differentiation into long-lived plasma cells (LLPCs) and memory B cells (MBCs) (28). To determine if lack of TLR3- or MyD88-dependent signaling pathways influenced the development, cellularity, or maintenance of the GC, we quantified CD19⁺ PNA⁺ GC B cells from RepliVAX WN-immunized mice by flow cytometry on 7, 14, and 21 dpi for B6, TLR3^−/−, and MyD88^−/− mice. As shown in Fig. 5A, the number of GC B cells for B6 mice peaked by 14 dpi and diminished by 21 dpi as the GC devolved (29). The development of GC cells in TLR3^−/− mice exhibited similar kinetics and achieved a response magnitude similar to that for B6 mice. In contrast, the number of PNA⁺ GC B cells for MyD88^−/− mice was significantly reduced on 14 dpi compared with that for B6 mice (P < 0.001, ANOVA).

Fig 5.

Altered GC cellularity and WNV-specific ASC responses to RepliVAX WN in TLR3^−/− and MyD88^−/− mice. (A) Decreased cellularity of GC in RepliVAX WN-immunized MyD88^−/− mice. B6 (n = 5) (filled bar), TLR3^−/− (n = 5) (open bar), and MyD88^−/− (n = 5) (striped bar) mice were immunized i.p. with 10⁶ IU RepliVAX WN. Spleen cells were harvested, and the number of GC B cells, measured as PNA⁺ CD19⁺ cells, was quantified by flow cytometry on 7, 14, and 21 dpi. Anti-NS1 (B) or anti-SVP (C) IgG ASCs from spleens of B6 (n = 5) (●), TLR3^−/− (n = 5) (■), and MyD88^−/− (n = 5) (▲) mice immunized i.p. with 10⁶ IU RepliVAX WN were quantified on 7, 14, 21, and 28 dpi. The results shown are from a representative experiment of 2 performed. Data are presented as the mean quantities of GC cells (A) or ASC number (B and C) ± SEM. ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001 (compared to results at the same time point for B6 mice).

As an additional approach to assessing the influence of TLR3- and MyD88-dependent signaling on GC, we quantified NS1- and SVP-specific ASCs on 7, 14, 21, and 28 dpi by ELISPOT assay (Fig. 5B and C). In B6 mice, the quantity of anti-NS1 and anti-SVP ASCs peaked on 14 dpi and gradually decreased thereafter through 28 dpi. Both the NS1- and SVP-specific ASC responses of TLR3^−/− mice rose rapidly and peaked at day seven (NS1-specific ASCs) or 14 dpi (SVP-reactive ASCs) at levels similar to B6 mice. However, the anti-NS1 response dropped rapidly and was significantly lower than that for B6 mice 14, 21, and 28 dpi (P < 0.05 and P < 0.01), and the anti-SVP ASC response was significantly lower beginning on day 21 (P < 0.01). Compared to B6 mice, MyD88^−/− mice showed a significant deficiency in the number of both NS1- and SVP-specific IgG ASCs from 7 to 28 dpi.

Previous studies using TLR3^−/− and MyD88^−/− mice have shown an increased IgG1 component of the serum IgG antibody response to infection or immunization (30, 31). To determine the influence of TLR3- and MyD88-dependent signaling on IgG subclass utilization, we determined the titers of NS1- and SVP-specific IgG_2C and IgG1 antibody in serum of immunized mice on 28 dpi. As shown in Fig. 6A, anti-NS1 IgG_2C titers were significantly lower for TLR3^−/− and MyD88^−/− mice than for B6 mice. In contrast, IgG1 titers were readily detectable for TLR3- and MyD88-deficient mice, but anti-NS1 IgG1 antibody was undetectable for B6 mice (Fig. 6B). Using the limit of IgG1 detection as a conservative titer for B6 mice, the IgG_2c/IgG1 ratio was 3.6 for B6 mice compared to 2.3 and 2.3 for TLR3^−/− and MyD88^−/− mice, respectively (Table 1). For the anti-SVP antibody response, the titers of IgG1 produced by all three mouse strains were similar (Fig. 6C), resulting in a more comparable IgG_2c/IgG1 ratio (Table 1). Neutralization titers were determined to assess potential differences in antibody function. The 50% neutralization titer of immune serum pooled from RepliVAX WN-immunized B6 mice (n = 35) was 1/1,890, compared to 1/1,124 for TLR3^−/− mice (n = 35) and 1/982 for MyD88^−/− mice (n = 34). To compare the neutralizing activity of comparable amounts of antibody from each strain, pooled serum from B6 and TLR3^−/− mice was diluted to achieve a SVP-specific titer equivalent to the MyD88^−/− serum pool, and the endpoint titers were confirmed by ELISA. As shown in Table 1, there was no significant difference in the neutralization activity of serum from the three mouse strains.

Fig 6.

Decreased IgG2c and increased IgG1 expression by NS1-specific IgG antibodies on 28 dpi in B6, TLR3^−/−, and MyD88^−/− RepliVAX WN-immunized mice. Mice were immunized i.p. with 10⁶ IU RepliVAX WN. The endpoint titer of anti-NS1 serum IgG2c antibody (A) or IgG1 antibody (B) or of anti-SVP serum IgG2c antibody (C) or IgG1 antibody (D) was calculated as described in Materials and Methods. Data are presented as the mean endpoint titers ± SEM. ∗, P < 0.05; ∗∗, P < 0.01 (compared to the same time point for B6 mice).

Table 1.

Properties of NS1- or SVP-specific serum IgG antibodies from B6, TLR3^−/−, and MyD88^−/− mice immunized with RepliVAX WN

Parameter	Value for mouse group
	B6	TLR3−/−	MyD88−/−
IgG2c/IgG1 ratio, NS1a	3.6 ± 0.4	2.3 ± 0.7	2.3 ± 0.9
IgG2c/IgG1 ratio, SVP	2.1 ± 0.8	2.0 ± 0.4	2.0 ± 0.6
50% Neut. titerb	2,934 ± 776	1,654 ± 598	4,126 ± 1,294

^aIgG_2c/IgG1 ratios were calculated for the individual mice shown in Fig. 8. The results shown represent the mean ± SEM for 5 mice.

^bSerum was pooled from groups of 34 or 35 RepliVAX WN-immunized B6, TLR3^−/−, or MyD88^−/− mice, and results were normalized to the equivalent ELISA titer. Results are expressed as the mean titers ± SEM for 5 separate experiments. Neut., neutralization.

To assess antibody affinity, B6 and TLR3^−/− serum pooled from 34 to 35 immunized mice was normalized by dilution to achieve an anti-SVP IgG titer equivalent to that of a MyD88^−/− serum pool (n = 34 mice). The endpoint titration curves of these serum samples are nearly identical (Fig. 7A), strongly suggesting equivalent affinities among the mouse strains. To assess antibody avidity, we determined the avidity index for serum from individual mice of each strain using a urea inhibition assay. The avidity index of anti-NS1 IgG antibodies was not different among the three mouse strains over a range of urea wash concentrations (Fig. 7B). The index was significantly different (P < 0.05) between B6 and MyD88^−/− mice for the SVP response on day 28 at the highest urea dose only, suggesting only minor differences in antibody avidity between the 2 strains (Fig. 7D). As expected, the avidity index for the serum samples from each mouse strain increased similarly between day 7 and day 28 as the NS1-specific (Fig. 7C) and SVP-specific (Fig. 7D) antibody responses matured. Together these data indicate that the affinities and avidities for NS1- and SVP-reactive antibodies were very similar among mouse strains.

Reduced LLPC responses in RepliVAX WN-immunized TLR3^−/− and MyD88^−/− mice.

Persistent humoral protection involves constitutive antibody secretion by LLPCs located predominantly in the bone marrow and the generation of quiescent MBCs located primarily in secondary lymphoid tissues (28, 32). The generation of LLPCs and MBCs has been shown to be closely associated with the GC response (28, 32). Given the impact of TLR3 and MyD88 deficiency on the GC and ASC responses, we investigated whether B cell memory developed normally in TLR3^−/− and MyD88^−/− mice. We quantified antigen-specific LLPCs localized in spleens and bone marrow of TLR3^−/− and MyD88^−/− immunized mice on 56 dpi by ELISPOT. The absence of either TLR3- or MyD88-dependent signaling caused a significant reduction in the quantity of anti-NS1 and anti-SVP ASCs in the bone marrow (Fig. 8A and C) and spleen (Fig. 8B and D), suggesting that both TLR3- and MyD88-dependent signaling played a role in the generation of WNV-specific LLPCs.

Fig 8.

The development of LLPCs and MBCs is differentially impaired in RepliVAX WN-immunized TLR3^−/− and MyD88^−/− mice. NS1-specific (A and B) or SVP-specific (C and D) LLPCs from bone marrow (A and C) or spleens (B and D) of B6 (●), TLR3^−/− (■), and MyD88^−/− (▲) mice (n = 5 mice/group) were quantified by ELISPOT assay on 56 days after i.p. immunization with 10⁶ IU RepliVAX WN. (E and F) The recall response to RepliVAX WN rechallenge by MBCs from B6, TLR3^−/−, and MyD88^−/− mice. RepliVAX WN-immunized B6 (n = 4), TLR3^−/− (n = 4), and MyD88^−/− (n = 4) mice were rechallenged i.v. with 10⁷ IU RepliVAX WN. Anti-NS1 (E) or anti-SVP (F) IgG ASCs from B6, TLR3^−/−, and MyD88^−/− mice were quantified by ELISPOT assay on the indicated days. WNV-specific ASCs in primary responses (dotted lines) following RepliVAX WN immunization of B6 (n = 5), TLR3^−/− (n = 5), and MyD88^−/− (n = 5) mice at 7 dpi are also shown. Data are presented as the mean numbers of ASCs per spleen ± SEM. The results shown are from a representative experiment of 2 performed. ∗, P < 0.05; ∗∗; P < 0.01; ∗∗∗, P < 0.001 (compared to the same time point for B6 mice).

Reduced MBC responses in RepliVAX WN-immunized MyD88^−/− but not TLR3^−/− mice.

Antigen-specific MBCs reside long term primarily in secondary lymph organs (28). To test if TLR3- and MyD88-dependent signaling pathways also played an important role in the generation of MBCs, we evaluated secondary ASC responses after rechallenge of immunized mice with RepliVAX WN. Anti-SVP and anti-NS1 plasma cells were quantified by ELISPOT assay following the RepliVAX WN rechallenge. As expected, recall ASC responses in immunized B6 mice, indicative of MBCs, were more rapid and vigorous than primary B cell responses, which generated only ∼2,000 anti-NS1 ASCs and ∼2,000 anti-SVP ASCs per spleen only on day 7 (Fig. 8E and F). In contrast, immunized B6 mice developed numerous specific ASCs (∼7,000 anti-NS1 ASCs and ∼50,000 anti-SVP ASCs/spleen) as early as day 3 after rechallenge. ASC numbers continued to expand quickly and reached ∼35,000 anti-NS1 ASCs and ∼200,000 anti-SVP ASCs/spleen on day 5 after rechallenge. In contrast, in MyD88^−/− mice, there was a significantly reduced number of anti-NS1 or anti-SVP ASCs on both day 3 and 5 postrechallenge (P < 0.001, ANOVA). Compared with B6 mice, TLR3^−/− mice showed a similar quantity of antigen-specific ASCs following rechallenge, indicating that the lack of TLR3 signaling did not compromise development of WNV-specific MBCs.

DISCUSSION

WNV-specific antibody is an important component in immune protection against WNV disease, as shown by WNV infection of B cell-deficient mice and by passive transfer of WNV-specific IgM and IgG antibodies (33–36). Therefore, in addition to generating polyfunctional T cell responses, effective WNV vaccines will need to elicit vigorous and durable antibody responses to protect against WNV disease. We have previously shown that immunization with the candidate vaccine RepliVAX WN results in the development of strong antibody responses to the E glycoprotein and the nonstructural protein NS1 (5), which represent important targets for a protective antibody response (36). These previous studies also showed that signaling through the type I IFN receptor influences the nature of the antibody response (6). RepliVAX WN is intrinsically adjuvanted with ligands for several PRRs, including RIG-I, TLR3, TLR7/8, and MDA-5. In these studies, we focused on the roles of the TLRs and specifically on TLR3- and MyD88-dependent signaling on the development of the WNV-specific B cell response to the single-cycle vaccine RepliVAX WN.

There are conflicting conclusions about the role of TLR3 signaling resulting from WNV infection. In a study by Daffis et al. (37), TLR3 played a protective role against WNV infection, and TLR3^−/− mice displayed higher virus titers in the brain and experienced increased mortality than wild-type controls. In contrast, Wang et al. (38) concluded that TLR3 signaling led to high lethality and facilitated neuroinvasion of virus following WNV infection. Studies with MyD88-deficient mice have clearly demonstrated a protective role for MyD88-dependent signaling in preventing WNV invasion of the central nervous system (39). WNV titers in the brain were significantly increased in MyD88^−/− mice, although titers in peripheral tissues remained generally very similar to those for wild-type mice in the early stages of infection. Although these previous studies provide evidence for TLR3- and MyD88-dependent signaling pathways in protection, the increased susceptibility to and high mortality of WNV infection for MyD88^−/− and TLR3^−/− mice at early times after infection (37, 39) complicate a comprehensive examination of the roles of these pathways in development of long-term adaptive immune responses to WNV. Our results using an SCFV vaccine demonstrated that the IgM ASC response was significantly lower at early times after immunization and that the serum IgG response to both SVP and NS1 antigens was significantly lower in RepliVAX WN-immunized TLR3^−/− and MyD88^−/− mice. Interestingly, studies by others showed that the early WNV-specific IgM and IgG responses to wild-type WNV infection (days 6 to 10) were less affected by lack of MyD88 (39) or TLR3 (37) than was the response to the SCFV RepliVAX WN in the present study. While the single-cycle particle RepliVAX WN initiates a WNV infection identical to that of wild-type virus and induces a vigorous anti-WNV humoral response (5), it is possible that the presence of high expression of PAMPs, increased antigen loads, and heightened inflammatory responses during infection with fully infectious WNV may provide redundant stimulatory signals to the developing B cell response, rendering it less dependent on specific TLR signaling than is the response to the SCFV RepliVAX WN. While the roles of TLR3- and MyD88-dependent signaling in B cell responses to infectious WNV remain uncertain, the results of the present study clearly demonstrate the important role of intrinsic PAMPs in development of optimal antibody responses under the more modest inflammation and antigen availability conditions following immunization with a single-cycle vaccine. Further, these results demonstrate the dependence of B cell memory development on the intrinsic adjuvanting by RepliVAX WN-associated PAMPs.

The development of antibody responses can be divided into several steps, including B cell activation, B cell expansion, and affinity maturation in the GC and the development of the B cell memory response. TLR3- and MyD88-dependent signaling pathways have been shown to be pivotal in the development of antiviral humoral responses (40–42). Injection of various TLR agonists has been shown to impact B cell activation and/or proliferation. For example, injection of TLR3 agonists, in synergy with BCR signaling and CD40L, aids in B cell activation and proliferation, whereas TLR7 agonists promote development of antibody-secreting cells (43). Previous studies using virus infection models have shown that early B cell activation in draining lymph nodes as manifested by surface expression of CD69 and CD86 occurs within the first few days after infection, is polyclonal, and is transient (26, 44). Similarly, we observed a transient increase in the frequency and total number of CD19⁺ B cells from the pLNs of B6 mice that expressed increased CD69 and CD86 levels at 3 days after RepliVAX WN inoculation. In agreement with the findings of Purtha et al. (26) following WNV infection, the number of CD69-expressing B cells detected in RepliVAX WN-inoculated MyD88^−/− mice was moderately but significantly lower than that of wild-type mice. Additionally, the mean fluorescence intensity of CD86 expression by MyD88^−/− and TLR3^−/− B cells was significantly decreased on 3 dpi, and the total number of CD86⁺ B cells was significantly lower at 5 dpi in these strains. Consistent with the diminished activation phenotype of B cells in MyD88^−/− and TLR3^−/− mice, there was a significant reduction of anti-SVP IgM-secreting cells in the draining LNs and spleens of these mice over the initial 5 dpi. The detection of IgM ASCs in these mouse strains on 3 dpi in the current study compared to the detection at 7 dpi by Purtha et al. (26) using a WNV infection model may reflect a higher assay detection sensitivity due to the higher availability of epitopes or different epitopes expressed by the WNV SVP capture antigen in our studies compared to recombinant E protein used in the previous studies (26). Although the diminished serum IgG responses in MyD88^−/− mice in response to high-dose RepliVAX WN inoculation strongly suggest an overall defect in the B cell response, it is also possible that the early B cell activation events in MyD88^−/− mice may have reflected a lower initial antigen load. This caveat does not hold for TLR3^−/− mice, since the IgM ASC response in TLR3^−/− mice was significantly diminished although the antigen load appeared equivalent to that of B6 mice.

The GC is the site of B cell proliferation, isotype class switch, affinity maturation, and differentiation of antigen-specific B cells into MBCs or LLPCs. In the present study, using SCFV particles as a vaccine immunogen, the cellularity of the GC reaction was significantly impaired in MyD88^−/− mice. The involvement of MyD88-dependent signaling in development or maintenance of the GC reaction is apparently pathogen dependent. No defects in GC response development were observed following infection of MyD88^−/− mice with influenza virus virus-like particles (VLPs) (45), murine polyomavirus (40), or Salmonella enterica serovar Typhimurium (46). However, our observations of reduced GC cellularity in RepliVAX WN-immunized MyD88^−/− mice are consistent with those of others following immunization of MyD88^−/− mice with inactivated respiratory syncytial virus (RSV) (42) or 2009 pandemic influenza split vaccine (15), murine gamma herpesvirus 68 infection (47), or infection with the murine gammaretrovirus Friend virus (48). As an additional assessment of GC development in TLR3^−/− and MyD88^−/− mice, we quantified WNV antigen-specific ASC responses during the initial 28 dpi. Consistent with the evidence of diminished GC cellularity in RepliVAX WN-immunized MyD88^−/− mice, the number of antigen-specific IgG ASCs was consistently, significantly lower than that for B6 mice. These results are similar to those reported by Jeisy-Scott (15) following infection of MyD88^−/− or TLR7^−/− mice with influenza virus and suggest that MyD88-dependent signals are important early for development of the GC response and antigen-specific ASC response. Interestingly, in immunized TLR3^−/− mice, although the total number of PNA⁺ CD19⁺ B cells in the GC was unaltered, the WNV-specific ASC response was initially equivalent to that of B6 mice but was not maintained and diminished thereafter to levels significantly lower than those in wild-type mice. These results are consistent with the notion that TLR3-dependent signals were necessary for optimal development or maintenance of the ongoing ASC response in the GC. In support of this idea, the presence of TLR7 ligand in bacteriophage QB VLP immunogens was shown to rescue the development of GC responses in IL-21-deficient mice, indicating that TLR ligands can cooperate in association with other immune products to facilitate strong GC responses (49).

The impact of the lack of TLR3- or MyD88-dependent signaling was also manifested in the development of B cell memory. Given the relatively weak GC response and ASC response of RepliVAX WN-immunized MyD88^−/− mice, the detection of significantly diminished numbers of NS1- and SVP-specific LLPCs in the bone marrow and spleen on 56 dpi for these mice was not surprising. Similarly, the recall IgG ASC response to intravenous high-dose RepliVAX WN challenge, reflecting stimulation of memory B cells, was significantly smaller in MyD88^−/− mice. Consistent with these results, diminished development of LLPCs and MBCs has been observed previously in TLR7^−/− and MyD88^−/− mice following immunization with inactivated influenza vaccines (15, 45).

The role of TLR3 in development of antibody responses has not been studied as extensively. In the present studies, while diminished LLPC responses to NS1 and SVP were observed in spleen and bone marrow of immunized TLR3-deficient mice, the recall IgG ASC response by MBCs was nearly equivalent in magnitude and kinetics to that of wild-type mice. This differential effect of TLR3 deficiency on development of LLPCs and BMCs may reflect the predisposition of antigen-specific B cells produced early in the GC response to differentiate to MBCs, whereas those B cells produced later in the GC differentiate primarily into LLPCs (50). Such a model is consistent with the developed and then rapidly diminishing IgG ASC response we detected in the spleens of RepliVAX WN-immunized TLR3^−/− mice in the present studies.

The skewing of the IgG subclass profile has been found previously in various infection and immunization models using MyD88^−/− or TRIF^−/− mice (30, 31) and may reflect diminished class switch DNA recombination resulting from decreased type I IFN responses (51–53) or with a decreased availability of synergistic TLR/BCR signals (54, 55). In the present study, we detected decreased levels of serum anti-NS1 IgG_2c and increased levels of anti-NS1 IgG1 antibodies in RepliVAX WN-immunized MyD88^−/− and TLR3^−/− mice. However, it is interesting that the IgG_2c/IgG1 ratio was less affected for the SVP-specific response among mouse strains. The reason for this difference is unclear, but it may reflect differences in the epitope density or bioavailability between SVP and NS1 antigens in vivo. In this light, it is possible that enhanced BCR signaling due to the multivalent nature of SVP particles (56) may require less TLR-dependent signaling to drive switch of antibodies from the IgG1 subclass, whereas class switching of B cells in response to the less-complex antigen NS1 (57, 58) may be more dependent on TLR signaling. Serum antibody from all mouse strains demonstrated virus neutralization activity, and the affinities and avidities of the serum IgG antibodies appeared to be very similar among strains. While perhaps less important for the ability of antibody to neutralize virus, the expression of appropriate IgG subclasses is important for protection manifested through complement-dependent or Fcγ receptor-dependent mechanisms known to play a role in protection against WNV (59, 60).

Taken together, the results of the current studies indicate that both TLR3- and MyD88-dependent signaling play important roles in shaping the development of humoral responses to the single-cycle vaccine RepliVAX WN. MyD88-dependent signaling affects the development of humoral responses by impacting B cell activation, development of the GC reaction, and the generation of LLPCs and MBCs, whereas GC responses develop, but are not maintained, in the absence of TLR3 signaling, which ultimately reduces differentiation of WNV-specific B cells into LLPCs. The results of this study enhance our understanding of the link between specific TLR pathways and development of hte B cell response. Specifically, they have important implications for the role of individual PAMPs on intrinsic adjuvanting of single-cycle vaccines as it relates to development of vaccine-specific B cell memory. This understanding will be important for the rational development of new vaccines and for improving the efficacy of existing vaccines.