Viral Replicative Capacity, Antigen Availability via Hematogenous Spread, and High T_FH:T_FR Ratios Drive Induction of Potent Neutralizing Antibody Responses

Neutralizing antibody response is the best-known correlate of long-term protective immunity for most of the currently licensed clinically effective viral vaccines. However, the host immune and viral factors that are critical for the induction of robust and durable antiviral humoral immune responses are not well understood. Our study provides insight into the dynamics of key cellular mediators of germinal center reaction during live virus infections and the influence of viral replicative capacity on the magnitude of antiviral antibody response and effector function. The significance of our study lies in two key findings. First, the systemic spread of even poorly replicating or nonreplicating viruses to mimic the spread of antigens from replicating viruses due to escalating antigen concentration is fundamental to the induction of durable antibody responses. Second, the T_FH:T_FR ratio may be used as an early predictor of protective antiviral humoral immune responses long before memory responses are generated.

ABSTRACT

Live viral vaccines elicit protective, long-lived humoral immunity, but the underlying mechanisms through which this occurs are not fully elucidated. Generation of affinity matured, long-lived protective antibody responses involve close interactions between T follicular helper (T_FH) cells, germinal center (GC) B cells, and T follicular regulatory (T_FR) cells. We postulated that escalating concentrations of antigens from replicating viruses or live vaccines, spread through the hematogenous route, are essential for the induction and maintenance of long-lived protective antibody responses. Using replicating and poorly replicating or nonreplicating orthopox and influenza A viruses, we show that the magnitude of TFH cell, GC B cell, and neutralizing antibody responses is directly related to virus replicative capacity. Further, we have identified that both lymphoid and circulating T_FH:T_FR cell ratios during the peak GC response can be used as an early predictor of protective, long-lived antibody response induction. Finally, administration of poorly or nonreplicating viruses to allow hematogenous spread generates significantly stronger T_FH:T_FR ratios and robust T_FH, GC B cell and neutralizing antibody responses.

IMPORTANCE Neutralizing antibody response is the best-known correlate of long-term protective immunity for most of the currently licensed clinically effective viral vaccines. However, the host immune and viral factors that are critical for the induction of robust and durable antiviral humoral immune responses are not well understood. Our study provides insight into the dynamics of key cellular mediators of germinal center reaction during live virus infections and the influence of viral replicative capacity on the magnitude of antiviral antibody response and effector function. The significance of our study lies in two key findings. First, the systemic spread of even poorly replicating or nonreplicating viruses to mimic the spread of antigens from replicating viruses due to escalating antigen concentration is fundamental to the induction of durable antibody responses. Second, the T_FH:T_FR ratio may be used as an early predictor of protective antiviral humoral immune responses long before memory responses are generated.

INTRODUCTION

Smallpox eradication through the use of a live-virus vaccine is one of the most successful public health endeavors of modern medicine. Humoral immunity against smallpox in vaccinated individuals, characterized by neutralizing antibody, is stable, lasts for decades (1, 2), and is considered a valuable benchmark of the functional attributes of a good vaccine. Indeed, neutralizing antibody is the best correlate of long-term protective immunity for all the currently licensed clinically effective viral vaccines (3). Despite the success of attenuated live vaccines in preventing disease, very little is known about viral and host factors that drive induction of protective antibody responses that are long-lived. A potential clue resides in findings that humoral immunity following natural infection with variola (4), measles (5), polio (6), or yellow fever (7) viruses persist for decades even in the absence of reexposure to virus. These viruses cause acute generalized infections, with hematogenous spread of the virus and viral antigens to numerous secondary lymphoid organs, including the spleen.

Central to antiviral antibody and immune memory generation is the germinal center (GC) response in secondary lymphoid organs. Here, specialized CD4 T cell subsets, T follicular helper (T_FH), and T follicular regulatory (T_FR) cells provide survival, proliferative, and differentiation cues to B cells, culminating in the production of somatically mutated, high-affinity antigen-specific neutralizing antibody (8, 9). Dysregulation of T_FH cells (10–13), T_FH-B cell interaction (14, 15), or T_FR cells (16–18) has detrimental effects on the GC and subsequent antibody response. The cross talk between the T_FH, T_FR, and B cells dictates the outcome of the GC reaction.

Much of our current understanding of high-affinity neutralizing antibody production is largely based on studies using nonreplicating model antigens (19, 20), inactivated viral vaccines that do not replicate in mice (21), persistent viral infection models (22), and models that do not have a natural host-pathogen relationship (23). For these reasons, we have used the mouse pathogen ectromelia virus (ECTV), which causes a smallpox-like disease termed mousepox, an excellent surrogate for smallpox, and induces long-lived neutralizing antibody responses. The importance of antibodies as a primary correlate of protection in ECTV infection has been established by studies using mice lacking B cells, CD40, or major histocompatibility complex class II (24–27). However, very little is known about the induction or dynamics of the T_FH, T_FR, and GC B cell responses associated with protective neutralizing antiviral antibody generation.

We hypothesized that the escalating concentration of antigens during virus replication or live viral vaccine immunization must reach sufficient concentrations for hematogenous spread to drive T_FH differentiation in secondary lymphoid organs, including the spleen, for induction of potent and durable humoral immunity. Using replication-efficient and -inefficient versions of ECTV and the closely related vaccinia viruses, Western Reserve strain (VACV-WR) and Chorioallantois vaccinia Ankara (CVA), we report here that virus replicative capacity has a significant impact on the magnitude of the T_FH, GC B cell, and protective neutralizing antibody responses. Similar findings were made with live and inactivated influenza A virus (IAV) strains. Increasing the dose of replication-inefficient or replication-deficient poxviruses or IAV strains significantly augmented T_FH, GC, and protective neutralizing antibody titers only when combined with an intravenous (i.v.) route of immunization. Overcoming the dissemination barrier to allow hematogenous spread and optimal antigen availability in secondary lymphoid organs may thus be a key to improving the immunogenicity and protective humoral immunity induced by replication-deficient virus vaccines. Our data also show that T_FH:T_FR ratios at the peak of the GC response in both the spleen and blood may be used as universal biomarkers that are predictive of protective long-lived antibody generation. Manipulation of T_FH:T_FR ratios may thus be a vital step in optimizing the efficiency of viral vaccines and vaccination regimes in generating a potent antiviral antibody response.

RESULTS

A strong T_FH response drives robust GC B cell and neutralizing antibody responses.

To understand the dynamics of the key cellular players of antibody response generation against ECTV infection, we measured total T_FH (B220^– CD4⁺ CD44^hi CXCR5^hi PD-1^hi Foxp3^–), T_FR (B220^– CD4⁺ CD44^hi CXCR5^hi PD-1^hi Foxp3⁺), and GC B (B220⁺ Fas⁺ GL7⁺) cell responses over 4 weeks following subcutaneous (s.c.) infection of wild-type C57BL/6 mice. T_FH cells (Fig. 1A) expanded rapidly as evident on day 7 postinfection (p.i.), peaked in numbers between days 14 and 16 p.i. (Fig. 1B, left y axis) and accounted for about 4 to 6% of all splenic CD4⁺ T cells. Although numbers contracted after this period, the response was still ongoing at day 28 p.i. In contrast to T_FH cells, there was an initial significant 3-fold drop in total numbers (Fig. 1B, right y axis) of T_FR cells at day 7 p.i. This was followed by a 12- to 16-fold increase in T_FR cell numbers, coincident with the T_FH contraction phase between days 14 and 21 p.i. These changes in numbers of cells, also depicted by T_FH:T_FR and T_FR:T_FH cell ratios (Fig. 1C), revealed an inverse relationship between the two cell subsets from about days 7 to 10 p.i. The T_FH:T_FR ratio was about 1:1 in naive animals but increased to 120:1 at the peak of the T_FH response. The proportion of T_FR cells that expressed CD25, the IL-2 receptor α (IL-2Rα) chain, progressively increased during the course of infection, suggesting a possible IL-2-IL-2-Rα-mediated layer of regulation on T_FH and/or GC B cells (Fig. 1D). GL7⁺ GC T_FH cells (B220^– CD4⁺ CD44^hi CXCR5^hi PD-1^hi GL7⁺; Fig. 1E), reported to have enhanced B-cell help capabilities (28), followed similar kinetics of expansion and contraction as the total T_FH cell response (Fig. 1F), accounting for 50% of all T_FH cells at the peak of the response at day 14 p.i. and beyond (Fig. 1G).

FIG 1.

Kinetics of T_FH and T_FR cells during ECTV-WT infection. C57BL/6 mice (n = 3 to 7 mice per group) infected s.c. with 10³ PFU of ECTV-WT were sacrificed at the time points indicated. (A) Flow cytometry contour plots of splenic T_FH (CD4⁺ CD44^hi CXCR5^hi PD-1^hi Foxp3^–) and T_FR (CD4⁺ CD44^hi CXCR5^hi PD-1^hi Foxp3⁺) cells. (B) Total numbers of T_FH (left y axis) and T_FR (right y axis) cells per spleen. (C) Splenic T_FH:T_FR ratio during the course of infection. The data represent means ± the standard errors of the mean (SEM). (D) Concatenated flow cytometric contour plots of CD25-expressing T_FR cells during the course of infection with a graphical representation of CD25 median fluorescent intensity at the indicated time points. (E) Flow cytometry contour plot of GL7-expressing GC T_FH (CD4⁺ CD44^hi CXCR5^hi PD-1^hi) cells. (F) Total GC T_FH cell numbers per spleen. (G) Comparative analysis of GL7⁺ and GL7^– CXCR5^hi PD-1^hi T_FH cells. The data represent means ± the SEM; data were log transformed, and the statistical significance was determined by one-way ANOVA (****, P < 0.0001).

The GC B cell response (Fig. 2A) was also similar in kinetics to that of T_FH cells, with a peak proliferative response observed at day 14 p.i. (Fig. 2B and C). Histological analysis revealed larger and more GC per spleen section at day 14 p.i. and that GC persisted even at day 28 p.i. (Fig. 2D). Anti-ECTV IgG antibodies were detectable as early as day 7 p.i., with IgG absorbance units increasing progressively over time (Fig. 2E), contemporaneous with increases in the virus-neutralizing activity (Fig. 2F) and the 50% plaque reduction neutralization test (PRNT₅₀) titers (Fig. 2G). The ongoing T_FH and GC B cell responses detected at day 28 p.i. were likely responsible for the continued increase in neutralizing antibodies at days 35 and 50 p.i., which inversely correlated with the viral load in blood (Fig. 2H). Virus-specific antibody levels, neutralization, and PRNT₅₀ titers stabilized between 5 and 6 weeks p.i. and were maintained at 3 months p.i. (Fig, 2I, J, and K, respectively).

FIG 2.

Kinetics of GC B cell and antibody responses during ECTV-WT infection. C57BL/6 mice (n = 3 to 7 mice per group) were infected as described in Fig. 1 and sacrificed at the time points indicated. (A) Flow cytometry contour plots of GC B cells (B220⁺ GL7⁺ CD95⁺) cells at day 14 p.i. (B) Percentage of GC B cells. (C) GC B cell numbers per spleen. (D) Immunofluorescent staining (IgD, red; CD3, green; peanut agglutinin [PNA], blue) of 10-μm frozen spleen sections from uninfected (naive) and ECTV-WT-infected (days 14 and 28 p.i.) mice (magnification, ×10). (E) ECTV-specific IgG absorbance units determined by ELISA (1:200 serum dilution). (F) Kinetics of virus-neutralizing antibody response, curve fitted using four-parameter, nonlinear regression analysis. The dotted horizontal line indicates 50% neutralization. (G) PRNT₅₀ titer calculated as the reciprocal of the plasma dilution at which 50% of the virus is neutralized. (H) Blood viral genome copy numbers determined by qRT-PCR. The dotted horizontal line indicates the minimum level of detection (10 viral genome copies). All graphical data represent means ± the SEM. Data were log transformed, and the statistical significance determined by one-way ANOVA (**, P < 0.005; ***, P < 0.0005; ****, P < 0.0001). For long-term antibody response analyses, C57BL/6 mice (n = 3 mice per time point) were infected s.c. with 10³ PFU of ECTV-WT and bled at time points indicated, and the ECTV-specific IgG levels (I), virus-neutralizing activity (J), and PRNT₅₀ titers (K) were determined. The data represent two independent experiments expressed as means ± the SEM. No statistical difference was detected by one-way ANOVA.

Viral replicative capacity influences T_FH, T_FR, GC B cell, and long-lived protective antibody responses.

We posited that viral replicative capacity was key to the induction of long-lived humoral immunity and that the escalating concentration of replicating viral antigen provides essential cues for priming T_FH cell differentiation that drives robust GC and long-lived neutralizing antibody response generation. We tested our hypothesis by comparing T_FH, T_FR, GC B cell, and antibody responses in mice infected with ECTV-WT or ECTV-TK^Δ. The latter is a highly attenuated, replication-inefficient virus due to deletion of the thymidine kinase (TK) gene in the parent ECTV-WT (29) and its replication is restricted to the site of inoculation (30). Even when very high doses are used to establish systemic infection, ECTV-TK^Δ does not cause any morbidity in gene knockout mice deficient in type I or type II interferon (IFN), IFN regulatory factor 1, perforin, or granzymes A and B (31, 32). Most of these strains are highly susceptible to ECTV-WT and die from infection with just 1 PFU of virus. ECTV-TK^Δ is attenuated to such an extent that it does not persist in the bone marrow of the ECTV-susceptible BALB/c mice, whereas the other attenuated deletion mutant viruses lacking virus-encoded IFN-γ-binding or IFN-α/β-binding proteins do (33).

At comparable doses (10³ PFU), ECTV-WT-infected mice generated significantly higher proportions and numbers of T_FH (Fig. 3A) and GC T_FH (Fig. 3B) cells compared to ECTV-TK^Δ-infected mice. Viral replicative capacity and viral load were driving the significantly higher T_FH and GC T_FH responses as viral genomes were detected in the blood of mice infected with ECTV-WT at days 7 and 14 p.i. but not in ECTV-TK^Δ-infected mice (Fig. 3C). We also used IL-21 GFP transgenic mice (34) to further establish that replication competent ECTV-WT induced significantly higher numbers of T_FH cells that produced IL-21, the hallmark cytokine associated with T_FH cells (Fig. 3D). Notably, the proportion of T_FH and T_FR cells within the parent CXCR5^hi PD-1^hi population varied distinctly between the two viral infections. Although naive mice had similar proportions of T_FH and T_FR cells (Fig. 3E and F, left panel), there was a clear bias toward a T_FH cell response in ECTV-WT infected mice (Fig. 3E and F, middle panel). In contrast, T_FR cell proportions remained high in ECTV-TK^Δ-infected animals with a small increase in T_FH cell proportions (Fig. 3E and F, right panel). The stronger T_FH cell response to ECTV-WT infection resulted in a >10-fold increase in splenic T_FH:T_FR ratio (Fig. 3G), contemporaneous with a statistically significant 7-fold increase in blood T_FH cell numbers compared to the replication-inefficient virus (Fig. 3H). The circulating T_FH:T_FR ratios in blood paralleled those detected in the spleen (Fig. 3I).

FIG 3.

Influence of viral replicative capacity on splenic and blood T_FH cell populations. C57BL/6 mice (n = 5 to 6 mice per group) infected s.c. with 10³ PFU of ECTV-WT or 10³ PFU of ECTV-TK^Δ or left uninfected (naive) were sacrificed on day 14 p.i., and T_FH cell populations were analyzed by flow cytometry. (A and B) Percentages and total numbers of T_FH (A) and GC T_FH (B) cells per spleen. (C) IL-21 GFP transgenic mice (n = 4 mice per group) were infected as described for panel A, and IL-21-producing (IL-21⁺) T_FH (CD4⁺, CD44^hi, CXCR5^hi, PD-1^hi, Foxp3^–) cells were enumerated by flow cytometry. (D) C57BL6 mice infected as described for panel A were bled on days 0, 7, and 14 p.i. DNA was isolated, and viral genome copy numbers were determined by qRT-PCR. The dotted horizontal line indicates the minimum level of detection (10 viral genome copies). (E) C57BL6 mice infected as described for panel A were sacrificed at day 14 p.i., and proportions of Foxp3⁺ T_FR and Foxp3^– T_FH cells, gated on CD4⁺, CD44^hi, CXCR5^hi, and PD-1^hi cells, are represented by flow cytometric contour plots and as a graphical representation (F). (G) Splenic T_FH:T_FR ratio. (H) Circulating CD4⁺, CD44^hi, CXCR5⁺, PD-1⁺, Foxp3^– T_FH cells in the blood. The data shown are from one of two independent experiments with comparable outcomes. Data are expressed as means ± the SEM; data were log transformed, and the statistical significance was determined by one-way ANOVA (***, P < 0.0005; ****, P < 0.0001). (I) In a separate experiment, C57BL/6 mice (n = 3 to 5 mice per group) infected s.c. with 10³ PFU of ECTV-WT or 10³ PFU of ECTV-TK^Δ or left uninfected were sacrificed, and a comparative analysis of the splenic and blood T_FH:T_FR ratio was performed. Data are expressed as means ± the SEM.

Replication-competent virus infection was associated with significant 10- to 12-fold-higher GC B cell responses (Fig. 4A), substantially larger and higher numbers of GC per spleen section (Fig. 4B) and significantly higher virus-specific IgG absorbance units (Fig. 4C) compared to ECTV-TK^Δ-infected mice. We used B lymphocyte-induced maturation protein 1 (Blimp-1) green fluorescent protein (GFP) transgenic mice (35) to establish that the replication-competent ECTV-WT-infected mice generated a significantly higher frequency of circulating Blimp-1⁺ antibody-secreting cells compared to replication-inefficient ECTV-TK^Δ-infected mice (Fig. 4D). In addition, ECTV-WT induced significantly higher splenic plasma cell (B220^–, CD138⁺) numbers compared to ECTV-TK^Δ at the peak of the GC response (Fig. 4E). These differences were maintained at day 70 p.i. in the bone marrow, a survival niche for long-lived plasma cells (Fig. 4F) and, in serum, the PRNT₅₀ titers were 45-fold higher in ECTV-WT-infected mice compared to ECTV-TK^Δ-infected mice (Fig. 4G). Naive mice passively transferred with immune serum from ECTV-WT-infected mice exhibited far more significant reductions in viral load in liver (Fig. 4H, left panel), spleen (Fig. 4H, middle panel), lungs (Fig. 4H, right panel), and blood (Fig. 4I) compared to mice that received sera from ECTV-TK^Δ-primed mice. Virus replicative capacity is thus very strongly associated with induction of long-lived protective antibody responses. In addition to its effects on humoral immunity, virus replicative capacity is also strongly associated with a greater magnitude of increased numbers of effector memory (CD44^hi, CD62L^lo, and CD127^hi) CD4 and CD8 T cells in the spleen and bone marrow (data not shown).

FIG 4.

Viral replicative capacity determines the magnitude of GC B cell and protective antibody responses. (A) C57BL/6 mice infected as described in Fig. 3 were sacrificed at day 14 p.i., and the total number of GC B (B220⁺ GL7⁺ CD95⁺) cells was enumerated by flow cytometry. (B) Immunofluorescent PNA⁺ GC staining of 10-μm frozen spleen sections from naive mice or ECTV-WT- or ECTV-TK^Δ-infected mice at day 14 p.i. (magnification, ×10; scale bar, 500 μm). (C) C57BL/6 mice (n = 4 mice per group) infected with 10³ PFU of ECTV-WT or ECTV-TK^Δ were bled at the times indicated, and the ECTV-specific IgG absorbance units were determined by ELISA (1:200 serum dilution). Statistical significance was determined by two-way ANOVA. (D) Blimp-1 GFP transgenic mice (n = 4 mice per group) were infected as described for panel A and sacrificed at day 14 p.i., and the frequency of circulating Blimp-1 GFP⁺ (Blimp-1⁺) antibody-secreting cells was determined by flow cytometry. (E) C57BL/6 mice infected as in panel A (n = 9 mice per group) were sacrificed at day 14 p.i., and the splenic plasma cells (B220^–ve, CD138^+ve) were enumerated by flow cytometry. (F) C57BL/6 mice infected as in panel A were sacrificed at day 70 p.i. (n = 5 to 6 mice per group), and the bone marrow plasma cells were enumerated by flow cytometry. Data are presented as means ± the SEM; the data were log transformed, and the statistical significance was determined by a Student t test. (G) Virus-neutralizing activity and PRNT₅₀ titers in the sera of ECTV-WT- or ECTV-TK^Δ-infected C57BL/6 mice at day 50 p.i. Statistical significance was determined by a two-tailed t test. (H) C57BL/6 (n = 4 mice per group) naive, ECTV-WT, or ECTV-TK^Δ-primed mice were sacrificed at day 50 p.i., and 200 μl of extracted sera from each mouse was passively transferred into a new naive recipient mouse and challenged 24 h later with 10³ PFU of ECTV-WT. Mice were sacrificed at day 5 postchallenge, and the viral load in the liver, spleen, and lungs was determined by a viral plaque assay. (I) DNA isolated from blood was used to quantitate viral genome copy numbers by qRT-PCR. The data shown are from one of two independent experiments with comparable outcomes. Data are expressed as means ± the SEM. Dashed lines indicate the limit of detection. Data were log transformed and, unless indicated otherwise, the statistical significance was determined by one-way ANOVA (*, P < 0.05; ***, P < 0.0005; ****, P < 0.0001).

Increased dose combined with the hematogenous spread of replication-inefficient virus augments the magnitude of T_FH, GC B cell, and antibody responses.

We predicted that increasing the dose of replication-inefficient virus to escalate antigen availability and overcoming the dissemination barrier by using the i.v. route of administration for hematogenous spread would augment T_FH, GC B cell, and antibody responses generated against attenuated viruses. C57BL/6 mice were therefore infected with replication-inefficient ECTV-TK^Δ via the s.c. route at low (10³ PFU) or high (10⁶ PFU) doses or via the i.v. route at a high dose (10⁶ PFU) to investigate the influence of increased inoculum dose and hematogenous viral antigen spread on T_FH and GC B cell responses at day 14 p.i.

Flow cytometry analyses revealed that a 1,000-fold increase in ECTV-TK^Δ given s.c. to mice did not result in any significant changes in the frequencies or numbers of splenic T_FH (Fig. 5A), GC T_FH (Fig. 5B), blood T_FH (Fig. 5C), splenic and blood T_FH:T_FR ratios (Fig. 5D), or GC B (Fig. 5E) cells compared to 10³ PFU of virus. However, when 10⁶ PFU ECTV-TK^Δ was injected i.v., the T_FH, GC T_FH, blood T_FH, GC B cells, and T_FH:T_FR ratios were all significantly increased (Fig. 5A to E). Immunohistochemistry revealed that i.v. inoculation with 10⁶ PFU ECTV-TK^Δ induced more and larger GC responses per spleen section compared to infection with a similar dose through the s.c. route (Fig. 5F). Importantly, there was a positive correlation (r = 0.9916; Pearson correlation coefficient and P < 0.0001) between high dose ECTV-TK^Δ administered i.v. and the magnitude of T_FH and GC B cell responses. Concordantly, at day 50 p.i., significantly higher virus-neutralizing activity (Fig. 5G) and PRNT₅₀ titers (Fig. 5H) were detected in mice infected i.v. with a high dose of ECTV-TK^Δ. These results suggest that i.v. inoculation of the highly attenuated virus, resulting in hematogenous spread to overcome the dissemination barrier, can augment humoral immune responses. In addition, the strong correlation between GC responses and T_FH:T_FR ratios support our idea that splenic and circulating T_FH:T_FR ratios are robust indicators of durable, protective antiviral antibody responses.

FIG 5.

Effect of virus dose and route of infection on T_FH, T_FR, GC B cell, and neutralizing antibody responses. C57BL/6 mice (n = 3 per group) left uninfected or infected with either ECTV-WT (10³ PFU s.c.) or ECTV-TK^Δ (10³ PFU s.c., 10⁶ PFU s.c., or 10⁶ PFU i.v.) were sacrificed at day 14 p.i. (A) Percentage and total number of T_FH cells. (B) Percentage and total number of GC T_FH cells. (C) Circulating T_FH cell numbers. (D) Comparative analysis of the splenic and blood T_FH:T_FR ratio. (E) Percentage and total GC B cell numbers. (F) Immunofluorescent GC staining (IgD; red, CD3; green, PNA; blue) of 10-μm frozen spleen sections from mice infected with ECTV-TK^Δ at 10³ PFU s.c., 10⁶ PFU s.c., or 10⁶ PFU i.v. (magnification, ×10; scale bar, 500 μm). (G and H) Virus-neutralizing activity (G) and PRNT₅₀ titers (H) in sera at day 50 p.i. Data are shown from one of two independent experiments with comparable outcomes. Data are expressed as means ± the SEM; data were log transformed, and the statistical significance was determined by one-way ANOVA (***, P < 0.0005; ****, P < 0.0001).

T_FH:T_FR ratios can predict the induction of durable humoral immunity against viruses.

We determined whether the T_FH: T_FR ratio might be used to predict induction of durable humoral immunity more broadly by extending our observations with the ECTV model to other viruses.

We first analyzed humoral immunity to the closely related orthopoxvirus, vaccinia virus (VACV). Unlike ECTV, VACV is not a mouse pathogen, and replication of even the mouse-adapted, neurovirulent VACV-WR strain (referred to as VACV-WT) is restricted to the site of infection after cutaneous inoculation of immunocompetent mice (36). The i.v. route of infection was therefore used to overcome the dissemination barrier and to simulate blood-borne systemic viral spread. Replication-competent VACV-WT induced significantly higher splenic T_FH (Fig. 6A), GC T_FH (Fig. 6B), T_FH:T_FR ratio (Fig. 6C), and GC B (Fig. 6D) cell numbers compared to the replication-inefficient VACV-TK^Δ at a comparable dose. Altering the dose and antigen availability by a 1,000-fold increase in VACV-TK^Δ dose significantly increased the magnitude of T_FH, GC T_FH, T_FH:T_FR ratios, and GC B cells (Fig. 6A to D). We also determined that there was a positive correlation (r = 0.8123; Pearson correlation coefficient and P < 0.05) between increased dose of VACV-TK^Δ and the magnitude of T_FH and GC B cell responses.

FIG 6.

T_FH:T_FR ratios are predictive of protective antibody responses: VACV and MVA infection models. C57BL/6 mice (n = 3 mice per group) left uninfected or infected i.v. with VACV-WT (10³ PFU) or VACV-TK^Δ (10³ or 10⁶ PFU) were sacrificed at day 10 p.i. Blood and spleens were harvested for flow cytometric analysis of T_FH (A), GC T_FH (B), the T_FH:T_FR ratio (C), and GC B cells (D). The data are representative of two independent experiments expressed as means ± the SEM; the data were log transformed, and statistical analysis was performed by one-way ANOVA (*, P < 0.05; **, P < 0.005; ***, P < 0.0005; ****, P < 0.0001). C57BL/6 mice (n = 4 mice per group) left uninfected or infected i.v. with replication-competent CVA (10³ PFU) or replication-inefficient MVA (10³ PFU or 10⁸ PFU) were sacrificed at day 10 p.i. The total T_FH (E), GC T_FH (F), T_FH:T_FR ratio (spleen) (G), and GC B cell (H) numbers per spleen were determined. (I) C57BL/6 mice (n = 4 to 5 mice per group) infected as in panels E to H were bled at day 50 p.i., and the virus-neutralizing activity and PRNT₅₀ titers were determined. (J) C57BL/6 mice (n = 4 to 5 mice per group) left unprimed or primed as described previously (see panels E to H) were challenged with a lethal dose of ECTV-WT (10⁶ PFU i.v.) 50 days later. Mice were sacrificed 5 days p.c., and the viral loads in the spleen, lungs, and liver were determined by viral plaque assay. All data represent means ± the SEM; data were log transformed, and statistical analysis was performed by one-way ANOVA (*, P < 0.05; **, P < 0.005; ***, P < 0.0005; ****, P < 0.0001).

We next used modified vaccinia virus Ankara (MVA), a highly attenuated third-generation smallpox vaccine (37), which is currently being considered as a recombinant vaccine vector for infectious diseases and cancer (38–42). MVA does not cause a productive infection in the mammalian host and is therefore considered safe. However, MVA-based expression vectors have required more than one immunization or combination with prime-boost regimes involving different vector systems or recombinant protein antigens to induce effective levels of protective immunity (43). We posited therefore that a single low dose of MVA would elicit poor GC and protective antibody responses compared to the replication-competent parental CVA but that the MVA response could be augmented by administration of a higher dose and given i.v.

At a dose of 10³ PFU, CVA elicited significantly more robust splenic T_FH (Fig. 6E) and GC T_FH (Fig. 6F) responses, higher T_FH:T_FR ratios (Fig. 6G), stronger GC B cell responses (Fig. 6H) at day 14 p.i., and 25-fold higher PRNT₅₀ titers at day 50 p.i. than a comparable dose of MVA (Fig. 6I). Increasing the dose of replication-incompetent MVA by 100,000-fold (10⁸ PFU; MVA ↑dose) positively correlated (r = 0.7948; Pearson correlation coefficient and P < 0.02) with the significantly increased T_FH and GC responses (Fig. 6E to H), and there was a clear concomitant increase in virus neutralizing PRNT₅₀ titers (Fig. 6I). At 50 days postpriming with CVA (10³ PFU) or MVA (10³ PFU or 10⁸ PFU), mice were challenged with a high dose of ECTV-WT, sacrificed at day 5 postchallenge (p.c.) and viral load in organs determined. CVA-primed or high-dose-MVA-primed mice had a significantly lower viral burden in the spleen (Fig. 6J, left panel), lung (Fig. 6J, middle panel), and liver (Fig. 6J, right panel) compared to naive or low-dose-MVA-primed mice. These results established that a single high dose of replication-deficient MVA given i.v. to allow hematogenous spread induced protective antibody responses that were comparable to responses generated by the replication-competent CVA.

To exclude the possibility that our findings were unique to orthopoxviruses, we extended our studies to influenza A viruses (IAV). IAV is transmitted via the respiratory route causing infection of the upper and lower respiratory tract (44). We reasoned that hematogenous spread of inactivated IAV through the i.v. route of inoculation would also augment the antibody responses, as shown by the previous experiments. At comparable doses, live mouse-adapted influenza A/Puerto Rico/8/1934 (A/PR/8/34; H1N1) elicited significantly higher T_FH (Fig. 7A), T_FH:T_FR ratios (Fig. 7B), and GC B cell (Fig. 7C) numbers at day 10 and significantly higher hemagglutination inhibition (HAI) antibody titers at day 50 p.i. (Fig. 7D) compared to inactivated (replication incompetent) A/PR/8/34 immunization.

FIG 7.

The magnitudes of T_FH and GC responses to inactivated influenza viruses are governed by the dose and route of administration. C57BL/6 mice (n = 4 mice per group) either uninfected or infected i.v. with 10⁵ HAU of live or formalin-inactivated influenza A/PR/8/34 were sacrificed at day 10 p.i. (A) Total T_FH cells per spleen. (B) Splenic T_FH:T_FR ratio. (C) GC B cell numbers per spleen. (D) HAI titers at day 50 p.i. Data represent two independent experiments expressed as means ± the SEM; statistical analysis was performed using one-way ANOVA (**, P < 0.005; ****, P < 0.0001). C57BL/6 mice (n = 4 mice per group) left uninfected or infected i.v. with 10⁵ HAU of live or formalin-inactivated A/SI/3/06 were sacrificed at day 10 p.i., and the blood and spleens were harvested for flow cytometric analysis. (E) Total T_FH cells per spleen. (F) Comparison of the T_FH:T_FR ratio in spleen and blood. (G) GC B cell numbers per spleen. C57BL/6 mice (n = 4 mice per group) left uninfected or infected with live (10⁵ HAU i.v.) or formalin-inactivated (4 × 10⁶ HAU ↑dose i.v. or i.m.) A/SI/3/06 were sacrificed at day 10 p.i. to evaluate the combined effects of increased dose and route of administration on T_FH and GC B cell populations. (H) Total T_FH cells per spleen. (I) T_FH:T_FR ratio in the spleen. (J) GC B cell numbers per spleen. All data represent means ± the SEM; data were log transformed, and statistical comparison with the live virus-infected group was performed by one-way ANOVA (*, P < 0.05; **, P < 0.005; ***, P < 0.0005; ****, P < 0.0001).

Finally, we measured T_FH and GC B cell responses in mice infected 10 days previously with live or inactivated seasonal vaccine strain A/Solomon Islands/3/2006 (A/SI/3/06; H1N1) i.v. Expectedly, live virus induced significantly higher numbers of T_FH (Fig. 7E) and GC B (Fig. 7F) cells and higher T_FH:T_FR ratios in spleen and blood (Fig. 7G) compared to inactivated virus at a comparable dose. Immunization with a 40-fold-higher dose of inactivated A/SI/3/06 (inactivated ↑dose) augmented the T_FH numbers (Fig. 7H), T_FH:T_FR ratios (Fig. 7I), and GC B cell numbers (Fig. 7J) to levels similar to those induced by live virus. Since current inactivated IAV vaccines are administered via intramuscular (i.m.) or deep s.c. routes, we predicted that i.m. administration of inactivated virus, even at an increased dose (inactivated ↑dose), would limit systemic antigen availability, thereby limiting the GC response. Indeed, compared to the i.v. route of immunization, the i.m. route of immunization with inactivated virus resulted in significant reductions in T_FH numbers (Fig. 7H), T_FH:T_FR ratios (Fig. 7I), and GC B cell numbers (Fig. 7J). We found a positive correlation (r = 0.7692; Pearson correlation coefficient and P < 0.03) between high-dose-inactivated IAV administered i.v. and the magnitude of T_FH and GC B cell responses.

Collectively, our data from experiments using orthopoxviruses and orthomyxoviruses confirm a direct correlation between systemic antigen availability and generation of durable antibody responses. They also strongly support the reliability of T_FH:T_FR ratios for predicting induction of robust GC and long-lived protective antibody responses.

DISCUSSION

The key goal of rational vaccine design is the effective induction and maintenance of immunological memory. This objective often eludes many viral vaccines that progress to clinical trials. While there are no universal prognostic biomarkers which can predict whether a vaccine will induce long-lived immunity, the best correlate of long-term protection is neutralizing antibody (3). This attribute is integrally associated with live viral vaccines. Indeed, the most successful human vaccine, a live viral vaccine against smallpox, elicits stable and life-long humoral immunity (2, 45–47). Much of our understanding of antiviral antibody responses comes largely from studies using protein antigens, inactivated vaccines, viruses that are not natural pathogens, or viruses that cause chronic infection. As a consequence, it has not been entirely clear why replicating viruses or live virus vaccines are superior at eliciting robust and effective long-lived immunity. We have addressed this fundamental question using a model of generalized viral infection in its natural host. ECTV causes a smallpox-like disease termed mousepox and is an excellent surrogate for smallpox and for studying long-lived neutralizing antibody responses.

Using ECTV and other viral models, we have made three key findings in relation to induction of long-lived antiviral antibody responses. First, the replicative capacity of virus or viral vaccine vector shapes the evolution of the GC response and induction of durable humoral immunity. Second, virus replicative capacity has a significant impact on the T_FH:T_FR ratio, a biomarker that can predict induction of protective, long-lived antibody responses. Third, administration of poorly or nonreplicating viruses to allow hematogenous spread generates a significantly stronger T_FH:T_FR ratio and robust T_FH, GC B cell, and neutralizing antibody responses.

In humans, an increase in circulating T_FH cells in blood following IAV vaccination was shown to correlate with increases in preexisting antibody titers but not with the induction of primary antibody responses (48). In another human study, the activation of circulating T_FH cells induced by IAV vaccination was shown to correlate with the development of memory B cells (49). The present study indicates that the magnitude of the splenic or blood T_FH response and T_FH:T_FR ratio can predict the strength of primary, memory, and long-lived antibody responses. In humans, the analysis of blood T_FH subsets has permitted the assessment of ongoing T_FH responses in secondary lymphoid organs, thus making it feasible to monitor circulating T_FH and T_FR responses (48, 50–52). The use of T_FH:T_FR ratios as a predictive indicator of durable antibody response generation would be of substantial value in a clinical vaccine trial, significantly reducing testing time and resources. T_FH:T_FR ratios can be determined within 10 to 14 days after vaccination, long before B cell memory forms and long-lived antibody responses are generated. While it may not be possible to use a single universal T_FH:T_FR ratio value as predictive of durable antibody response generation to different vaccine types, our data from the ECTV-TK^Δ, MVA, and inactivated IAV immunization models indicates that a range between ratios of 5:1 and 10:1 would be sufficient. Although we have only measured total T_FH and T_FR cell responses, which may not necessarily all be antigen specific, our data indicate that in specific pathogen-free mice, the T_FH:T_FR ratios based on total T_FH cell responses to four strains of poxviruses and two strains of IAV are predictive of induction of potent protective antibody responses following immunization/infection. One potential caveat to using total and not antigen-specific T_FH:T_FR ratios in humans is when an individual in a clinical trial has contracted some other infection postvaccination; in that case the measurement of vaccine antigen-specific responses would be more relevant. Under such circumstances, measurement of antigen-specific T_FH and T_FR responses would be more relevant.

A previous study in mice using nonreplicating antigens showed that the T_FR cell responses follow the kinetics of the T_FH cells (16). However, using the natural mouse pathogen ECTV, we have found that there is, in fact, an inverse relationship with respect to cell numbers and kinetics between the two subsets from about days 7 to 10 p.i., just prior to the peak GC response at day 14. T_FH numbers increase beginning around day 7, as shown in our study, as well as by others (48, 49, 53), whereas T_FR numbers are reduced during this time. The fact that there is an inverse relationship between T_FH and T_FR cell would suggest that the bigger the ratio, the more durable the antibody response generated. The precise molecular mechanisms that control this inverse relationship are not known but must be important in order for expansion of T_FH cells just before the peak GC response. In a recent study, high concentrations of IL-2 produced at the peak of IAV infection were reported to prevent T_FR cell development by a Blimp-1-dependent mechanism (54). When IL-2 levels were reduced following resolution of IAV infection, T_FR cells were found to differentiate and move to B cell follicles to maintain B cell tolerance. In mice infected with ECTV-WT, IL-2 production rapidly increases from day 4 p.i., peaks at day 8, after which the levels are somewhat reduced but remain high even at day 12 (55). In addition to the early drop in T_FR numbers, we observed an increase in the proportion of IL-2Rα-expressing T_FR cells over the course of infection, particularly during the contraction phase of the T_FH and GC responses. It is possible that the IL-2:IL-2Rα axis contributes to the suppressive activity of T_FR cells following the peak T_FH and GC B cell responses. Although IL-2 can mediate inhibition of early T_FH differentiation (56, 57), it is conceivable that the cytokine could play a context-dependent paradoxical role in T_FH regulation by T_FR cells.

The biosafety aspects of attenuated viruses, either replication inefficient or replication defective, have prompted their use as viral vaccine vectors. However, our current understanding of the mechanisms of immune protection provided by attenuated viruses is confounded by the use of various doses, routes, animal models, and challenge strategies that have been reported. In this study, the antiviral humoral immune responses elicited by replication-efficient and -inefficient ECTV were compared, with a deliberate attempt to ensure that the comparisons were carried out under similar conditions. At the same dose and route of administration (s.c.), significant increases in the magnitude of T_FH and GC responses and neutralizing antibody titers were observed following infection with replication-competent virus compared to replication-inefficient virus. This finding can be primarily attributed to increased viral antigen availability, hematogenous spread, activation of potentially numerous secondary lymphoid organs, including the spleen, and most likely longer antigen retention by the GC follicular dendritic cells in its native form promoting sustained presentation and subsequent antibody affinity maturation (58). Our findings are consistent with previous studies examining the role of antigen levels in early T_FH proliferation and terminal differentiation to GC T_FH, albeit using protein antigens (19), bacteria (23), or persistent viral infection models (22). However, it should be noted that the chronic persistence of antigen leads to an abnormal T_FH expansion, resulting in dysregulated antibody responses (59).

In our experiments with poxvirus and IAV infection models, we found a direct correlation between attenuated/inactivated virus dose administered i.v. and the magnitude of T_FH and GC B cell response. A previous study established that mice inoculated i.v. with 10⁸ PFU of recombinant MVA expressing IAV HA and boosted 35 days later generated significantly higher titers of HAI antibody compared to the s.c., i.m., or intranasal routes of inoculation with an equivalent dose (60). The findings from our study and that of Sutter et al. (60) are consistent with results from a clinical trial that used nonreplicating sporozoite vaccine against malaria which demonstrated superior protection when administered i.v. compared to s.c. or intradermal routes (61). Generalized viral infections such as variola, measles, and polio result in viremia and hematogenous dissemination, which could explain the generation of robust and durable protective antibody responses.

In order to understand the molecular changes in antigen-specific B cells associated with virus replicative capacity, dose, and route of virus inoculation, we have utilized the SWHEL B cell receptor transgenic (Tg) mice (62) in combination with recombinant ECTV (WT and TK^Δ) engineered to express the model protein antigen, hen egg lysozyme (HEL). SWHEL Tg mice have HEL-specific Tg B cells that can class switch and can be tracked as they undergo affinity maturation and class switch recombination in response to virus-expressed HEL. We have found that systemic spread of a replication-inefficient virus through i.v. inoculation significantly increases the induction of somatic hypermutation events in antigen-specific B cells and affinity maturation of antibody (unpublished data).

Our data suggest that the T_FH/T_FR balance may shape the evolution of the GC response. Neutralizing antibody responses were strongly correlated with lymphoid tissue resident T_FH:T_FR ratios across four virus infection models. Furthermore, the circulating T_FH:T_FR ratios paralleled those detected in the lymphoid tissue and may be used reliably to predict induction of potent virus neutralizing antibody responses, provided that the terminal T_FH differentiation and GC formation steps are not affected. Modulating the T_FH/T_FR balance could thus, represent one of the axes along which the quality and duration of antibody response may be manipulated during or postvaccination. However, any strategy to regulate the T_FR cells for optimal GC response must consider the level and timing of the modulation so as to prevent generation of autoreactive antibodies. Altered T_FR:T_FH ratios have been observed in several human and animal models of autoimmunity (63–65). Although there is evidence that T_FR cells can govern the GC response by suppressing T_FH and B cells (16, 18), the exact mechanism still remains unclear. A better understanding of the molecular mechanisms of IL-2Rα-mediated T_FR suppression of GC responses will be critical for augmenting protective antibody responses following vaccinations.

Recently, transcriptomics, combined with systems biology using peripheral blood leukocytes from vaccinees have revealed molecular signatures of protective antibody responses associated with vaccine-induced immunity, but no universal signatures have been found (66–69). Another recent study has demonstrated a direct relationship between GC activity and levels of GC T_FH expressed CXCL13 in plasma (70). CXCL13 is the ligand for CXCR5, and the authors have suggested it may be potentially used as a plasma biomarker of GC activity. We postulate that the judicious use of the circulating T_FH:T_FR ratio, with or without plasma CXCL13 (as a marker for terminal T_FH differentiation), as a correlate that can predict antiviral antibody response generation within days or weeks of vaccination will facilitate efficient screening of the efficacy of new and novel viral vaccine formulations.

In summary, the results presented in this study demonstrate that the replicative capacity of virus or viral vaccine vector governs the induction of neutralizing antibody responses by influencing antigen availability. Our data clearly shows that allowing dissemination of highly attenuated (MVA) or inactivated (IAV) vaccines through the hematogenous route in mice can significantly increase the magnitude of T_FH and GC B cells and subsequent neutralizing antibody responses. Although the i.v. route may not be feasible for mass vaccination, it may be an important alternative for inducing better and long-lived antibody responses against pathogens, especially in instances where the other routes of vaccination are not effective.

MATERIALS AND METHODS

Ethics statement.

All animal experiments were conducted in strict accordance with the good animal practice for care and use defined by the Australian code for National Health and Medical Research Council guidelines and protocols approved by the Animal Ethics and Experimentation Committee of the Australian National University.

Mice.

Inbred, specific-pathogen-free C75BL/6J, Blimp-1 GFP, and IL-21 GFP transgenic mice on a C57BL/6 background were bred and maintained at the Australian Phenomics Facility, Australian National University, and used at 6 to 8 weeks of age.

Animal experiments.

All animal infections were performed under intraperitoneal avertin anesthesia, which was freshly prepared before use. Mice were infected either subcutaneously (s.c.) in the flank of the left hind limb (hock), intramuscularly (i.m.) in the left thigh, or intravenously (i.v.) in the lateral tail veins. Mice were monitored daily during the entire experimental period and scored against a clinical matrix taking into account hair coat, posture, breathing, activity, foot swelling, and body weight.

Cell lines and culture.

African green monkey kidney epithelial (B-SC-1), Canis familiaris kidney cell line (MDCK), and Syrian hamster kidney fibroblast cell line (BHK21) were maintained in Eagle minimum essential medium (EMEM) and Roswell Park Memorial Institute (RPMI) 1640 medium, respectively, supplemented with 10% fetal calf serum, 2 mM l-glutamine, 120 μg/ml penicillin, and 200 μg/ml streptomycin and neomycin sulfate.

Virus stocks.

The Moscow strain of ECTV (referred to as ECTV-WT; ATCC VR-1734), the TK deletion mutant of ECTV-WT (ECTV-TK^Δ), the Western Reserve strain of VACV (referred to as VACV-WT; ATCC VR-1354), and the TK deletion mutant of VACV-WT, VACV-TK^Δ (71), were propagated in BS-C-1 monolayers, and titers were determined by viral plaque assay as described previously (72, 73). CVA virus was kindly provided by Jürgen Hausmann at the Bavarian Nordic GmbH, Germany. Titration of MVA stocks was carried out on BHK-21 cell monolayers by immunostaining as described previously (74). IAV A/PR/8/34 and A/SI/3/06 stocks were propagated in specific-pathogen-free embryonated chicken eggs, and viral titers were determined by plaque assay using MDCK cells (75). Inactivation of influenza virus stocks was performed by incubation with formalin (0.02% final concentration) for 18 h at 37°C, followed by dialysis against four changes of PBS (76). Inactivation of the virus was confirmed by plaque assay.

Viral load in organs and blood.

Organs harvested were snap-frozen and stored at –80°C until processed. Briefly, serial dilutions of organ homogenates (Polytron homogenizer; Pro-Scientific, Inc.) were plated onto confluent BS-C-1 monolayers, and virus titers were determined using the conventional viral plaque assay as described previously (77). Viral titers in blood were measured by real-time qualitative PCR and expressed as viral genome copy numbers correlated with the copy numbers of ECTV156, a late gene encoding viral hemagglutinin (73).

Immunofluorescent staining protocol.

Harvested spleen was frozen in O.C.T and stored at –80°C until required. Air-dried cryosections (10 μm) were fixed in ice-cold acetone and blocked with PBS containing 3% bovine serum albumin (BSA) before staining with peanut agglutinin (PNA; Sigma-Aldrich), rabbit anti-mouse (clone 11.26 C; Southern Biotech), and rat anti-mouse CD4 (clone GK1.5; BioLegend) in PBS containing 0.5% BSA and 0.1% Tween 20. Sections were washed, mounted with Vectashield mounting medium (Vector Laboratories), and imaged using a Leica-SP5 confocal laser scanning microscope.

ELISA.

Virus-specific IgG levels were measured by enzyme-linked immunosorbent assay (ELISA) as described elsewhere (31). Briefly, high-binding ELISA plates (Thermo Fisher Scientific) were coated with sucrose-cushion purified ECTV-WT virus, washed, and incubated with serial dilutions of the serum samples. Antibodies were detected using horseradish peroxidase conjugated to anti-mouse IgG or IgG subclass antibodies (Southern Biotech), and the colorimetric reaction (tetramethyl benzidine; Invitrogen Life Technologies) absorbance was measured at 650 nm using Tecan plate reader (Tecan).

Flow cytometry.

Cells harvested from spleen or blood were incubated with Fc block (clone 2.4G2; BD Biosciences) and surface stained as required with anti-mouse CD4 (clone RM4-5; BD Biosciences), B220 (clone RA3-6B2; BD Biosciences), CD44 (clone IM7, BioLegend), CXCR5 (clone DG8; BD Biosciences), PD-1 (clone J43; eBioscience), GL7 (clone GL7; BD Biosciences), CD95 (clone Jo2; BD Biosciences), CD138 (clone 281.2; BD Biosciences), CD62L (clone MEL-14; BD Biosciences), or CD62L (clone H1.2F3; eBioscience). For Foxp3 staining, cells were fixed and permeabilized according to the manufacturer’s instructions (Foxp3 transcription factor buffer kit; eBioscience) and incubated with anti-mouse Foxp3 (clone FJK-16s; eBioscience). All samples were acquired on BD LSRII or BD LSR Fortessa and subsequently analyzed using FlowJo software (Tree Star, Inc.).

Statistical analysis.

Statistical analyses of experimental data, as indicated in figure legends, were performed using Prism V6 (GraphPad Software, San Diego). Where indicated, single comparisons were analyzed using the unpaired student two-tailed t test with Welch correction, and multiple comparisons were analyzed using one-way or two-way analysis of variance (ANOVA), followed by the Holm-Sidak multiple-comparison test.