An Epidemic Zika Virus Isolate Drives Enhanced T Follicular Helper Cell and B Cell–Mediated Immunity Free

Zika virus (ZIKV) was first isolated in the Zika Forest region of Uganda in 1947 (1, 2). Very little was known about the virus during the 20th century and it was linked to mostly innocuous symptoms. Improving our understanding of the immune response to ZIKV infection has become a major research priority following a series of increasingly severe outbreaks in the Federated States of Micronesia (2007), French Polynesia (2013), and South and Central America (with other outbreaks worldwide; 2015) (3–6). These outbreaks, together with smaller outbreaks that have occurred since 2015, have demonstrated the novel capacity of ZIKV to cause widespread disease and severe symptoms following infection. In particular, ZIKV infection has been associated with Guillain–Barré syndrome (an autoimmune paralysis) in adults, and fetal microcephaly along with a number of other developmental defects when infection occurs during pregnancy (7, 8). Together, these outbreaks and the symptoms that have become associated with infection raise the question of whether recent, epidemic ZIKV isolates have a distinct capacity to modulate host immunity.

CD4 T cells represent an important part of the host immune response against viral infections with multiple roles during primary infection. Following activation, CD4 T cells can differentiate into various Th cell subsets, which serve key functions in supporting the immune response. In the context of viral infection, IL-12 and IFN-γ lead to the generation of Th1 CD4 T cells (9). These cells express the T-box transcription factor T-bet and support the antiviral immune response by secreting effector cytokines such as IFN-γ, TNF-α, and IL-2 and by promoting CD8 T cell activation via dendritic cell (DC) licensing (9, 10). T follicular helper (Tfh) cells are an additional CD4 T cell subset that plays an important role in promoting Ab responses during viral infections. Tfh cells are defined by expression of CXCR5 and PD-1 on the cell surface, as well as the transcription factor B cell lymphoma-6 (Bcl-6) (11). Canonically, pre-Tfh cells must interact with both DCs and B cells for differentiation, and cytokine-dependent and -independent signals contribute to their development (11). Once differentiated, Tfh cells are able to form or enter germinal centers (GCs), distinct structures in secondary lymphoid tissues within which Tfh cells mediate GC B cell affinity maturation and proliferation (11–14).

Several studies have explored CD4 T cell responses during ZIKV infection, in a variety of mouse models (15). In immunocompetent C57BL/6 mice, our group has demonstrated that infection with a pre-epidemic Canadian ZIKV isolate (ZIKV^CDN, PLCal_ZV, GenBank accession no. KF993678, isolated in 2013 from a Canadian traveler returning from Thailand and exhibiting mild symptoms, https://www.ncbi.nlm.nih.gov/nuccore/KF993678) leads to a robust Th1 CD4 T cell response 7 d postinfection (dpi), characterized by upregulation of T-bet, and production of IFN-γ, TNF-α, and IL-2 (16). Experiments using a Cambodian ZIKV isolate (FSS13025) in LysMCre⁺Ifnar^fl/fl mice, which lack IFN-α/β receptor (IFNAR) expression primarily in mature macrophages and granulocytes, revealed a similar Th1 polarization of CD4 T cells, and also identified formation of a Tfh cell response (17). Although CD4 T cells were required for generation of an IgG response, their depletion had no impact on viral burden or CD8 T cell responses (17). In contrast, depleting CD4 T cells from highly susceptible IFNAR knockout (KO) mice caused higher viral loads, more severe paralysis, and reduced survival in 10- to 12-wk-old mice infected with a Puerto Rican isolate (PRVABC59), and lethal infection in 3- to 4-wk-old mice infected with a Brazilian isolate (PE243) (18, 19). The Tfh cell response to ZIKV was more closely examined in a recent paper by Liang et al. (20) who described a robust Tfh cell response to a Chinese isolate (SZ-WIV01) characterized by IFN-γ production, IgG2c production, and long-lasting Ab responses in C57BL/6 mice treated with an IFNAR blocking Ab (20). Although these studies collectively suggest an important role for CD4 T cells during ZIKV infection, most were carried out in immunodeficient mouse models. Furthermore, none of these studies has compared whether pre-epidemic and epidemic ZIKV isolates differ in their capacity to induce Th1 or Tfh cell responses.

Previously, we have demonstrated that an epidemic ZIKV isolate from Brazil (ZIKV^BR, HS-2015-BA-01, GenBank accession no. KX520666, isolated in 2015 from a symptomatic patient in Salvador, Bahia, Brazil, https://www.ncbi.nlm.nih.gov/nuccore/KX520666) actively suppresses the CD8 T cell response to infection, which correlates with delayed virus clearance, when compared with rapid control of infection with ZIKV^CDN (21). Although virus is detectable by plaque assay in the spleen and kidney at 7 dpi in mice infected with ZIKV^BR, it is cleared from these organs by 14 dpi, indicating that although ZIKV^BR establishes a more sustained infection, it does not become chronic in mice (21). In this study, we sought to determine whether ZIKV^BR modulates CD4 T cell responses, and to determine potential mechanisms for viral clearance following ZIKV^BR infection. Our data demonstrate that ZIKV^CDN induces a more robust Ag-experienced, Th1-polarized CD4 T cell response than does ZIKV^BR at 7 dpi. In contrast, ZIKV^BR-infected mice were more efficient at inducing Tfh and GC B cell responses 10 dpi, which persisted through 14 and 21 dpi. These responses correlated with enhanced formation of GCs in the spleen, as well as production of higher avidity Abs with greater neutralization potential, likely contributing to the control of ZIKV^BR infection. Conversely, ZIKV^CDN infection favored the induction of extrafollicular B cell responses that correlated with the production of lower quality virus-specific Abs with reduced neutralization capacity. Taken together, these data suggest that ZIKV^BR may have evolved to dampen T cell responses that prevent rapid viral clearance, while simultaneously evoking more potent T-dependent humoral immunity that drives a contained, yet more sustained infection.

Vero cells (African green monkey kidney epithelial cells, provided by Steven Varga, The University of Iowa) were cultured in DMEM (Wisent) supplemented with 10% heat-inactivated FBS (Wisent), 1% l-glutamine (Wisent), 1% penicillin-streptomycin (Wisent), and 1% non-essential amino acids (Sigma-Aldrich) at 37°C and 5% CO₂.

Low-passage (p.4) ZIKV^CDN (PLCal_ZV, GenBank accession no. KF993678, https://www.ncbi.nlm.nih.gov/nuccore/KF993678) derived from a ZIKV-infected traveler returning to Canada who acquired the infection in Thailand in 2013 was provided by Gary Kobinger and David Safronetz (National Microbiology Laboratory and Public Health Agency of Canada) and confirmed via sequencing to be of the Asian lineage (22). Low-passage (p.4) ZIKV^BR (HS-2015-BA-01, GenBank accession no. KX520666, https://www.ncbi.nlm.nih.gov/nuccore/KX520666) isolated in August 2015 in Salvador, Bahia, Brazil, was provided by Mauro Teixeira (Universidade Federal de Minas Gerais). The stock was originally passaged three times in C6/36 mosquito cells and once in Vero cells. To propagate ZIKV stocks, 6 × 10⁶ Vero cells were seeded in 150- × 2-mm dishes (Sigma-Aldrich), and 24 h later infected at a multiplicity of infection of 1 for 2 h in unsupplemented Eagle’s MEM (EMEM). After 2 h, infection media were removed and replaced with 15 ml of DMEM supplemented with 2% heat-inactivated FBS, 1% l-glutamine, 1% non-essential amino acids, 1% penicillin-streptomycin, and 15 mM HEPES. Supernatants were harvested after 48–72 h postinfection (hpi), centrifuged for 10 min at 3000 × g, aliquoted, and stored at −80°C. Viral stocks were titered by a plaque assay on Vero cells. Briefly, viral stocks were serially diluted (10-fold) in EMEM (Wisent) and 100 μl of each dilution was used to infect confluent monolayers of Vero cells in 1 ml of unsupplemented EMEM. At 2 hpi, infection media were removed and cells were overlaid with 1.2% carboxymethyl cellulose (Sigma-Aldrich) and 2% heat-inactivated FBS in EMEM for 4 d prior to fixation for 1 h with an equal volume of 10% formaldehyde. Monolayers were rinsed gently with distilled water and stained for 30 min with 0.1% crystal violet prior to counting plaques. ZIKV^CDN was UV-inactivated by transferring 1 ml of ZIKV into one well of a six-well plate and exposing it to 3 J/cm² of UV irradiation in a UVC 500 Crosslinker (Hoefer) as previously described (16). UV inactivation was confirmed by plaque assay.

C57BL/6 mice (wild-type) were purchased from Charles River Laboratories, the In Vivo Therapeutics Core at the Indiana University School of Medicine, or bred at McGill University. Infected mice were housed in biocontainment level 2, and all animal procedures were carried out in accordance with the Canadian Council on Animal Care and were approved by the McGill University Animal Care Committee or the Indiana University Institutional Animal Care and Use Committee. Six- to 12-wk-old mice of both sexes were used for all experiments.

ZIKV^CDN (p.8 or p.9) or ZIKV^BR (p.5 or p.6) were injected i.v. at 1 × 10⁶ PFU. Mice injected with UV-inactivated ZIKV received a dose of inactivated virus equivalent to 1 × 10⁶ PFU. Mock-infected mice were i.v. injected with an identical volume of mock-infected Vero cell culture supernatant (prepared in parallel to ZIKV stocks).

The following Abs were used in an appropriate combination of fluorochromes: B220 (clone RA3-6B2, BioLegend), Bcl-6 (clone K112-91, BD Biosciences), CD4 (clone GK1.5, BioLegend), CD8α (clone 53-6.7, BioLegend), CD11a (clone M17/4, BioLegend), CD19 (clone 1D3/CD19, BioLegend), CD21 (clone 7E9, BioLegend), CD23 (clone B3B4, BioLegend), CD49d (clone R1-2, BioLegend), CD86 (clone GL-1, BioLegend), CXCR5 (clone L138D7, BioLegend), Foxp3 (clone FJK-16s, eBioscience), GL7 (clone GL7, BioLegend), IFN-γ (clone XMG1.2, BioLegend), IgD (clone 11-26c.2a, BioLegend), IL-2 (clone JES6-5H4, BioLegend), IL-5 (clone TRFK5 BioLegend), IL-17A (clone TC11-18H10.1, BioLegend), PD-1 (clone 29F.1A12, BioLegend), T-bet (clone 4B10, BioLegend), TNF-α (clone MP6-XT22, BioLegend), and appropriate isotype controls.

Spleens were isolated and mechanically disrupted to generate single-cell suspensions. Erythrocytes were lysed with ammonium-chloride-potassium (ACK) buffer (0.15 M NH₄Cl, 1 mM KHCO₃, and 0.1mM Na₂EDTA in dH₂O [pH 7.2]), cells were incubated with TruStain FcX (anti-mouse CD16/CD32, clone 93, BioLegend) and stained with the indicated Abs, followed by fixation using IC (intracellular) fixation buffer (eBioscience). Intracellular staining for transcription factors was performed using the Foxp3/Transcription Factor Staining Buffer Set (eBioscience), as per the manufacturer’s instructions. Samples were analyzed with a BD LSRFortessa flow cytometer (BD Biosciences) and FlowJo software (Tree Star).

For ex vivo restimulation, spleens were harvested and processed as above. Erythrocytes were lysed with ACK buffer and samples were simulated for 5.5 h at 37°C with 5% CO₂ in the presence of 1 μg per well of plate-bound anti-CD3ε Ab (clone 145-2C11, BioLegend) or media alone and brefeldin A (BioLegend). Anti-CD3ε was bound to the plate by incubating overnight in PBS at 4°C, and wells were washed with PBS prior to adding cells. Cells were then stained with surface Abs and fixed with IC fixation buffer (eBioscience), followed by intracellular staining for cytokines in Perm/Wash buffer (eBioscience). Samples were analyzed with a BD LSRFortessa flow cytometer (BD Biosciences) and FlowJo software (Tree Star).

Spleens from uninfected, ZIKV^CDN-infected, and ZIKV^BR-infected mice were cut in half, submerged in optimal cutting temperature compound, flash-frozen with liquid nitrogen vapor, and stored at −20°C. The tissues were sectioned into 5-μm slices using a cryostat. Spleen sections were fixed in an ice-cold acetone solution (75% acetone and 25% ethanol) for 5 min. The sections were then placed in PBS and stained in a dark, humid chamber at room temperature. Sections were blocked for 1 h in 2% BSA with Fc Block (1:100, clone 2.4G2) and then stained in 2% BSA with CD4 (clone GK1.5, eBioscience), GL7 (clone 81E2, BD Pharmingen), and IgD (clone 11-26c, eBioscience) for 1 h. The slides were then washed in PBS for 7 min twice while agitated in the dark and mounted using ProLong diamond antifade (Invitrogen). Images were taken using a Zeiss LSM 780 laser scanning confocal microscope and analyzed using Fiji software. Using the Zeiss tiling software, 4-by-4 images were taken to count GCs. Five to 13 tiled images were taken for each spleen, and the average number of GL7⁺ GCs per image for each spleen was graphed.

To extract serum, blood was collected via cardiac puncture and centrifuged at 7000 rpm for 10 min at 4°C. Serum was diluted 25-, 50-, 100-, 250-, and 500-fold in EMEM, or serially diluted 10-fold in EMEM, and mixed with the corresponding ZIKV isolate. One hundred microliters of each serum/virus mixture was used to infect confluent monolayers of Vero cells in 1 ml of unsupplemented EMEM. Each six-well plate included one virus-only control well. At 2 hpi, infection media were removed, and cells were overlaid with 1.2% carboxymethyl cellulose (Sigma-Aldrich) and 2% heat-inactivated FBS in EMEM for 4 d prior to fixation for 1 h with an equal volume of 10% formaldehyde. Monolayers were rinsed gently with distilled water and stained for 30 min with 0.1% crystal violet prior to counting plaques. Percent neutralization was calculated as: % neutralization = [(no. plaques in no-serum well − no. plaques in indicated dilution well)/no. plaques in no-serum well] × 100%.

To measure virus-specific IgG levels, Thermo Scientific MaxiSorp plates (Thermo Fisher Scientific) were coated with 1 × 10⁵ PFU of heat-inactivated (30 min at 56°C) ZIKV^CDN or ZIKV^BR overnight at 4°C. Following incubation, plates were washed twice with PBS–0.1% Tween 20 and blocked for 2 h in 5% milk diluted in PBS–0.1% Tween 20 at room temperature. Serum was collected from naive or infected mice as described above and diluted in 5% milk PBS–0.1% Tween 20 and incubated for 1 h at room temperature (serum from infected mice was incubated in wells coated with the matching virus isolate). Wells were washed twice with PBS–0.1% Tween 20. For avidity determination, various concentrations of ammonium thiocyanate (0.15–4 M range) were added to each well and incubated for 15 min prior to washing three times with PBS–0.1% Tween 20. Wells were then incubated with HRP-conjugated donkey anti-mouse IgG Ab (Jackson ImmunoResearch Laboratories, 1:20,000 dilution in 5% milk PBS–Tween 20 solution) for 1 h at room temperature followed by washing three times in PBS–0.1% Tween 20. Plates were reacted with tetramethylbenzidine substrate and the reaction was stopped with 4 N sulfuric acid prior to being read at 450 nm using an accuSkan FC plate reader.

Data were analyzed using GraphPad Prism 8 or 9 software. Specific tests for determining statistical significance are indicated in the figure legends. A p value <0.05 was considered statistically significant.

To begin our analysis of the CD4 T cell response to each ZIKV isolate, we compared changes in the total and Ag-experienced CD4 T cell populations in the spleen at 7 dpi. Ag-experienced CD4 T cells were tracked using cell surface expression of CD11a and CD49d, previously established surrogate markers of Ag experience in multiple infection models that we have validated for tracking the global CD4 T cell response to ZIKV infection (16, 21, 23, 24). Consistent with our previous results, the total number of cells in the spleen was significantly higher following infection with ZIKV^CDN than ZIKV^BR (Fig. 1A). Although ZIKV^BR-infected mice had a significantly higher frequency of CD4 T cells, there was no significant difference in the number of CD4 T cells in the spleen between the two infected groups (Fig. 1B–D). This difference in frequency without a corresponding increase in numbers is likely due to the differential CD8 T cell responses induced by these two infections, as we described previously (21). Similar to our previous observations regarding the splenic Ag-experienced CD8 T cell response, we observed a significant reduction in both the frequency and number of CD11a⁺CD49d⁺ CD4 T cells responding to infection with ZIKV^BR in the spleen (Fig. 1B, 1E, 1F), which contrasted to what we previously observed in the blood (21). In addition, the frequency and number of CD11a⁺CD49d⁺ CD4 T cells expressing the Th1 transcription factor T-bet and the expression of T-bet on a per cell basis (as measured by geometric mean fluorescence intensity [gMFI]) were significantly lower following ZIKV^BR infection (Fig. 1G–J). Taken together, these data demonstrate that ZIKV^CDN induces a stronger CD11a⁺CD49d⁺ Ag-experienced CD4 T cell response than does ZIKV^BR, and that those CD4 T cells that are responding to ZIKV^BR infection are less polarized toward the Th1 subset.

An important function of Th1-polarized CD4 T cells is the production of effector cytokines such as IFN-γ, TNF-α, and IL-2 to support the antiviral immune response (9). Thus, to investigate the functionality of responding Th1 cells at 7 dpi with each ZIKV isolate, we analyzed cytokine expression following restimulation with plate-bound anti-CD3ε Ab. Similar to our previous observations (16), and consistent with their Th1 polarization, Ag-experienced CD4 T cells produced Th1 cytokines (IFN-γ, TNF-α, and IL-2) following restimulation (Fig. 2). The frequency and number of IFN-γ⁺ and TNF-α⁺ Ag-experienced CD4 T cells were significantly higher following infection with ZIKV^CDN compared with infection with ZIKV^BR, although among IFN-γ⁺ or TNF-α⁺ cells, their expression on a per cell basis was similar between the two groups (Fig. 2A–H). Although the frequency of IL-2⁺ Ag-experienced CD4 T cells and the gMFI of IL-2 staining were similar between the two groups, we observed a significant reduction of IL-2⁺ Ag-experienced CD4 T cells following ZIKV^BR infection (Fig. 2I–L). These data are consistent with the reduced induction of T-bet expression during ZIKV^BR infection (Fig. 1G–J) and confirm a more robust Th1 response following ZIKV^CDN infection. To determine whether the reduced polarization to the Th1 subset following ZIKV^BR infection was due to polarization toward other Th subsets, we also analyzed expression of IL-5 (Th2) and IL-17A (Th17) following restimulation. Although in some samples we detected small frequencies of IL-5– and IL-17A–producing cells (<1% of CD11a⁺CD49d⁺), we observed no difference between ZIKV^CDN and ZIKV^BR infection (Supplemental Fig. 1). Thus, ZIKV^BR infection induces a CD4 T cell response that is less polarized to the Th1 subset and is less capable of producing Th1 cytokines without inducing polarization toward other Th subsets.

CD4 T cell priming events involve a fate decision-making process directing CD4 T cells toward either effector (e.g., Th1) differentiation or Tfh cell differentiation (11). While the signals that influence this fate decision remain incompletely understood, which subset is favored may depend on the balance of signals that are present. For example, whereas IL-2 or IL-12 signaling and T-bet expression have been shown to favor Th1 differentiation, IL-6 stimulation and strong TCR signaling promote Tfh cell responses (11, 25, 26). To address whether differences exist between the Tfh cell response to ZIKV^CDN and ZIKV^BR, mice were infected with either isolate, or injected with UV-inactivated ZIKV^CDN (UV-ZIKV) as a control, and the Tfh cell response was analyzed in the spleen at 10, 14, and 21 dpi. We observed a small population of PD-1^hiCXCR5⁺ Tfh cells that remained comparable across all time points in mice injected with UV-ZIKV or ZIKV^CDN-infected mice (Fig. 3A–I). However, following ZIKV^BR infection we observed a significantly higher frequency and number of PD-1^hiCXCR5⁺ Tfh cells across all time points tested (Fig. 3A–I). To confirm that these cells represented a bona fide Tfh cell response, we analyzed the PD-1^hiCXCR5⁺ Tfh cells for expression of the transcription factor Bcl-6, a master regulator of Tfh function (11). As expected, Tfh cells from ZIKV^BR-infected mice expressed significantly more Bcl-6 on a per cell basis than did non-Tfh CD4 T cells from the same mouse at all time points tested (Fig. 3J–L), confirming that ZIKV^BR induces a prototypical Tfh cell response. We conducted the same analysis within ZIKV^CDN-infected mice and found that the small frequency of Tfh cells present within these mice expressed significantly more Bcl-6 than did their non-Tfh CD4 T cell counterparts (Supplemental Fig. 2). Nonetheless, at 10 dpi the expression of Bcl-6 (as measured by gMFI) was higher among Tfh cells from ZIKV^BR-infected mice than ZIKV^CDN-infected mice (Fig. 3J, Supplemental Fig. 2). These data indicate that infection with ZIKV^BR promotes a more robust Tfh response than infection with ZIKV^CDN.

Within the Tfh cell population, a subset of Foxp3⁺ T follicular regulatory (Tfr) cells, which develop from T regulatory cells, may also be induced following infection to regulate the Tfh and GC B cell responses (27). To determine whether each ZIKV isolate differentially induced Tfr cell responses, we analyzed Foxp3 expression among the Tfh subset at 10, 14, and 21 dpi with either ZIKV isolate. Although we observed a trend toward a higher number of Tfr cells at 10 and 21 dpi, and significantly more Tfr cells at 14 dpi with ZIKV^BR, this population represented a much smaller proportion of the total Tfh cell population within the ZIKV^BR-infected mice due to the larger increase in overall Tfh cell numbers in those mice (Supplemental Fig. 3). In contrast, the frequency of Tfr cells among CXCR5⁺PD-1⁺ cells in ZIKV^CDN-infected mice showed a trending increase at 10 dpi and was significantly higher at 14 and 21 dpi (Supplemental Fig. 3). Thus, the enhanced Tfh cell response to ZIKV^BR infection is not associated with a significant change in Tfr numbers at most of the time points examined, suggesting that changes in the Tfr population are unlikely to be responsible for the modulation of Tfh populations in this model.

One of the primary functions of Tfh cells is to promote GC reactions in secondary lymphoid organs, which function to mediate GC B cell isotype class switching and affinity maturation (13, 14). Together, these processes generate B cells with higher affinity for cognate Ag, which in turn leads to production of higher affinity Abs against the pathogen (13, 14). Because we observed a more robust Tfh cell response following ZIKV^BR infection, we next analyzed the GC B cell response to UV-ZIKV, ZIKV^CDN, or ZIKV^BR at 10, 14, and 21 dpi. Similar to the Tfh cell response, we detected only small frequencies of IgD^loGL7⁺ GC B cells at all time points following injection with UV-ZIKV or infection with ZIKV^CDN (Fig. 4A–I). Both the frequency and number of IgD^loGL7⁺ GC B cells were significantly increased following infection with ZIKV^BR across all time points analyzed (Fig. 4A–I). In addition, we analyzed splenic tissue sections for the presence of GL7⁺ GCs by confocal microscopy. We observed similar numbers of GCs in uninfected and ZIKV^CDN-infected mice, whereas the average number of GCs per spleen section was significantly higher in ZIKV^BR-infected mice (Fig. 4J, 4K). Taken together, these data demonstrate that infection with ZIKV^BR leads to enhanced generation of GCs in the spleen and a more robust GC B cell response than with ZIKV^CDN.

One of the primary functions of Abs is to neutralize virus particles, thereby preventing further spread to uninfected cells. Thus, we sought to determine whether the generation of neutralizing Abs differed between infection with each ZIKV isolate. To address this question, we conducted plaque reduction neutralization tests to analyze the capacity of serum from UV-ZIKV–, ZIKV^CDN-, and ZIKV^BR-infected mice at 10 and 21 dpi to neutralize virus and prevent infection of Vero cell monolayers. Serum was serially diluted and incubated with the same isolate with which the mice had been infected (i.e., serum from ZIKV^CDN-infected mice was incubated with ZIKV^CDN virus and serum from ZIKV^BR-infected mice was incubated with ZIKV^BR virus) prior to incubation with Vero cells for plaque assay. Each serum dilution was compared with a no serum control well to calculate the percentage neutralization. Serum from UV-ZIKV–injected mice was unable to neutralize ZIKV^CDN at any serum dilution (data not shown), indicating that there was no non-specific neutralization within mouse serum. Serum from ZIKV^CDN-infected mice neutralized virus only at low dilutions at 10 dpi, which was quickly lost at dilutions of 100-fold or higher (Fig. 5A). Strikingly, serum from ZIKV^BR-infected mice at 10 dpi showed high neutralization across the same range of dilutions, and our data indicate that neutralization capacity was only lost after being diluted 10,000-fold (Fig. 5A, Supplemental Fig. 4). This pattern was maintained at 21 dpi, with serum from ZIKV^CDN-infected mice showing only minor increases in neutralization capacity, and serum from ZIKV^BR-infected mice retaining robust neutralization capacity (Fig. 5B). Thus, ZIKV^BR induces an Ab response that potently neutralizes viral infection in cell culture.

Increased neutralization by Abs from ZIKV^BR-infected mice could be due to a difference in either Ab quantity, quality, or both. To address this question, we harvested serum from naive mice (pooled serum from naive mice was used for these experiments) or on 21 dpi with either ZIKV^CDN or ZIKV^BR and measured ZIKV-specific IgG levels via ELISA. We observed that despite a reduced neutralization capacity, ZIKV^CDN infection induces a stronger virus-specific IgG response at 21 dpi (Fig. 5C). Next, we tested the avidity of the Ab response via ELISA and thiocyanate elution (28). We observed that the concentration of thiocyanate required to remove 50% of IgG binding (IC₅₀, also referred to as affinity index) was significantly greater for ZIKV^BR infection–induced Abs, establishing that these Abs are of higher avidity (Fig. 5D). Taken together, this suggests that the stronger Tfh and GC B cell responses induced following ZIKV^BR infection leads to a higher avidity Ab response that is correlated with a greatly enhanced neutralization capacity despite lower overall production of virus-specific IgG.

The production of higher levels of low-avidity virus-specific Abs following ZIKV^CDN infection in the absence of a robust Tfh response suggests that this viral isolate may favor the induction of an extrafollicular response. This response involves the early production of lower avidity Abs and the activation of non-GC B cell subsets such as marginal zone B (MZB) cells, which can rapidly produce IgM and IgG Abs in response to pathogens, including viruses (29–33). At 7 dpi, we observed that infection with ZIKV^CDN led to a significant increase in the activation of MZB cells, as measured by their surface expression of the costimulatory molecule CD86, compared with mock-infected or ZIKV^BR-infected mice (Fig. 6A, 6B). In addition, we observed that ZIKV^CDN infection, but not ZIKV^BR infection, led to the induction of a virus-specific IgG response at this early time point (Fig. 6C). Taken together, this suggests that infection with ZIKV^CDN primarily drives an extrafollicular B cell response that is associated with greater production of virus-specific IgG Abs with lower avidity and less capacity to neutralize infection.

Since the South and Central American outbreak, many studies have sought to improve our understanding of the immune response to ZIKV infection. However, few of these studies have directly compared the immune response induced by distinct ZIKV isolates, an approach that may help further our understanding of how ZIKV has changed in its interactions with the host immune system as it acquired epidemic capacity. In this study, we have conducted an analysis of CD4 T cell differentiation following infection with a pre-epidemic isolate (ZIKV^CDN) or an epidemic isolate (ZIKV^BR). Our data demonstrate that although ZIKV^CDN induces more robust Th1 and global Ag-experienced CD4 T cell responses than does ZIKV^BR at 7 dpi, there is a stronger Tfh cell response to ZIKV^BR at 10 dpi, which persists through to 21 dpi. This Tfh cell response correlates with increased GC formation and increased generation of GC B cells, which in turn leads to a more potent neutralizing Ab response in the serum and correlates with the induction of a higher avidity virus-specific Ab response. Conversely, ZIKV^CDN induces the production of more virus-specific IgG at various time points postinfection, even as early as 7 dpi. However, these Abs are of lower avidity, have poor neutralization capacity, and correlate with the stronger activation of extrafollicular B cells such as MZB cells. Taken together, these data suggest that although ZIKV^BR may have evolved to subvert antiviral Th1 CD4 T cell responses and CD8 T cell responses that develop earlier following infection (21), this ultimately results in stronger GC B cell–mediated immunity that develops later in the infection and likely prevents the establishment of chronic infection. One caveat that remains is that our study centered on investigating the total Ag-experienced CD4 T cells using surrogate markers rather than defined Ag-specific CD4 T cells using ZIKV-specific peptides or tetramers. Our findings regarding a change of polarization within the total pool of virus-activated CD4 T cells provide an overview of the virus-specific CD4 T cell response to these two ZIKV isolates. However, it will be of interest to determine whether these two isolates induce CD4 T cells that target different viral epitopes, as we have recently observed regarding the CD8 T cell response following infection with these ZIKV isolates, and this will be the focus of future investigation (21).

Impaired CD4 T cell immunity following ZIKV^BR infection could have a broad impact on the host’s capacity to respond to infection. The cytokines produced by Th1 CD4 T cells play an important role in activating other immune cells and enhancing the antiviral immune response (9). If the reduced presence of Th1 CD4 T cells impacted DC licensing as well, it could in turn impact the generation of protective CD8 T cell immunity, which would be consistent with our previous observations of an impaired CD8 T cell response to ZIKV^BR infection, which correlated with delayed viral clearance in some tissues (21). CD4 T cells clearly play an important role in protection against ZIKV infection in susceptible mouse strains, as their depletion leads to more severe paralysis, higher viral loads, and increased mortality (18, 19). In addition, transferring memory CD4 T cells from ZIKV-immune to naive IFNAR KO mice protected these mice from lethal ZIKV challenge (18, 19). The fact that this protection was lost in the absence of IFN-γ receptor signaling or B cells implies an important role for both Th1 and Tfh-dependent B cell responses in controlling ZIKV infection (19).

Only a few studies have directly analyzed Tfh and GC B cell responses to ZIKV infection. In LysMCre⁺Ifnar^fl/fl mice, ZIKV infection induces Tfh and GC B cell responses that are important for IgG isotype class switching (17). Liang et al. (20) similarly detected a robust “Th1-like” (IFN-γ⁺) Tfh cell response following ZIKV infection in mice that had been treated with an IFNAR-blocking Ab prior to infection. Similar to our observations in the present study, this Tfh response was readily detectable at 10 and 14 dpi, and correlated with generation of GC B cells, IgG2a production, and neutralizing Ab responses (20). Our previous work has identified that type I IFN production is reduced following ZIKV^BR infection, consistent with the role that Liang et al. (20) identified for IFNAR signaling in modulating the Tfh cell response to ZIKV (21). Whether IFNAR signaling directly or indirectly influences Tfh cell differentiation following ZIKV infection remains an intriguing question for future study. We also demonstrated previously that ZIKV^BR remains detectable by plaque assay in the spleen and kidney at 7 dpi but is cleared by 14 dpi (21). Thus, the enhanced Tfh and GC B cell responses we observe, as well as the resulting increase in Ab avidity and neutralization capacity following ZIKV^BR infection, could provide a potential mechanism for the eventual clearance of this virus.

Our observation of a balance between the induction of Th1- or Tfh-mediated immunity following infection with ZIKV^CDN or ZIKV^BR is similar to observations that have been made following Salmonella enterica and Toxoplasma gondii infection (25, 26). Following S. enterica infection, IL-12 enhances Th1 differentiation at the expense of the Tfh cell response, which is dependent on IL-12 receptor signaling and T-bet (26). Furthermore, this contributes to the inhibition of GC formation observed following S. enterica infection. During T. gondii infection, although IL-12 initially induces genes that may contribute to both Th1 and Tfh cell phenotypes, T-bet and IFN-γ antagonize Tfh cell responses and promote Th1 differentiation, which in turn limits GC formation and Ab responses (25). Thus, the robust Th1 response to ZIKV^CDN could be limiting the development of Tfh cell and GC B cell responses, whereas the lack of a Th1 response and Th1 cytokines may have the opposite impact following ZIKV^BR infection. However, Tfh cell differentiation does not represent a default differentiation pathway and requires more than an absence of conditions or cytokines that promote Th1 differentiation for their generation (11). For example, Tfh cell responses are promoted by strong TCR signals, such as those that occur following high-affinity TCR–peptide/MHC class II interactions, or by repeated stimulation (11). Fahey et al. (34) demonstrated that although both LCMV Armstrong and clone 13 initially drive comparable Th1 responses, over time as LCMV clone 13 persists, the CD4 T cell response becomes increasingly Tfh cell dominant. This suggests that delayed viral clearance (longer Ag exposure) following ZIKV^BR infection could contribute to the more robust Tfh cell response we observe. Interestingly, infection with ZIKV^CDN yielded the production of higher levels of virus-specific Abs, but these Abs had significantly lower avidity for ZIKV Ags and lower neutralization capacity. This was associated with stronger activation of MZB cells, suggesting that this virus isolate primarily activates extrafollicular B cell responses. The signals that determine the activation of extrafollicular and follicular B cell responses following ZIKV infection remain unknown and will be the focus of future investigation. In addition, future studies will seek to determine the cytokine-dependent and -independent signals that impact the Th1/Tfh cell balance following ZIKV infection. Nonetheless, the ability to compare the Th1 and Tfh cell responses between two isolates of the same virus suggests that ZIKV could provide a useful model for dissecting other signals that promote or restrain Tfh cell differentiation.

In conclusion, our data demonstrate that while ZIKV^BR infection induced a weaker Th1 response than did infection with ZIKV^CDN, this contrasts with more robust Tfh and GC B cell responses. As a result, ZIKV^BR infection generates a potent neutralizing Ab response, which persists as late as 21 dpi. In contrast, infection with ZIKV^CDN induces only a minor neutralizing Ab response despite producing high levels of virus-specific IgG Abs. Taken together, these data suggest that ZIKV^BR may have evolved to subvert Th1 CD4 T cell responses, but at the cost of promoting B cell–mediated immunity, which develops later in the infection. However, the difference in timing of these responses could provide a crucial window of viremia within which ZIKV^BR continues to circulate, which may have important implications during ZIKV infection, including potentially increasing the transmission window of the virus and increased risk of pathogenesis in the context of pregnancy.

We thank David Safronetz and Gary Kobinger at the National Microbiology Laboratory and Public Health Agency of Canada for providing ZIKV^CDN, and Mauro Teixeira (Universidade Federal de Minas Gerais) for providing ZIKV^BR. We also thank Steven Varga (University of Iowa) for providing Vero cells. Thank you to Noah Butler, Alexander Dent, Stefanie Valbon and Miranda Yu for discussion and feedback. We thank the members of the Indiana University Melvin and Bren Simon Cancer Center Flow Cytometry Resource Facility for outstanding technical support.

The authors have no financial conflicts of interest.