Infection order outweighs the role of CD4 + T cells in tertiary avivirus infection

Dengue (DENV) and Zika (ZIKV) are aviviruses that co-circulate throughout the tropical and subtropical regions. The link between CD4 + T and B cells during immune responses to DENV and ZIKV and their roles in cross-protection during heterologous infection is an active area of research. Here we used CD4 + lymphocyte depletions to dissect the impact of cellular immunity on humoral responses during a tertiary avivirus infection. We show that CD4 + depletion in DENV-primed animals followed by ZIKV-DENV resulted in delayed viremia followed by increased viral replication in tertiary infections and dysregulated adaptive immune responses compared to ZIKV-primed animals followed by DENV-DENV infections. We show a delay in DENV-specic IgM/IgG antibody titers and neutralization in the DENV-primed CD4-depleted animals but not in ZIKV-primed CD4-depleted animals. This study conrms the critical role of CD4 + cells in viremia control and priming of early and robust neutralizing antibody responses during sequential avivirus infections. Our results also revealed differential cytokine proles for both infection sequences regardless of CD4 + status. Our work here suggests that the order of avivirus exposure affects the outcome of a tertiary infection. Our ndings have implications for understanding complex immune responses induced by aviviruses that co-circulate and develop effective avivirus vaccines.


Introduction
Flaviviruses, including Dengue virus (DENV) and Zika virus (ZIKV), are principally arthropod-borne viruses that cause mild to severe diseases in humans worldwide. These members of the Flaviviridae family are transmitted primarily through the bite of Aedes spp. mosquitoes, imposing an enormous public health burden in tropical and subtropical areas 1,2 . Whereas ZIKV transmission has decreased in recent years, its initial emergence into the DENV-endemic regions of the western hemisphere 3,4 raised concerns, mainly due to immunological cross-reactivity limiting serological testing and the implications for the development of severe manifestations in populations exposed to sequential infections 5,6 . While ZIKV consists of a single serotype, there are four different serotypes of DENV based on antigenic differences of the envelope protein, all of which are pathogenic in humans [7][8][9][10] . Exposure to one infecting serotype should confer lifelong protection against disease upon secondary homotypic infection. However, heterologous DENV infection can lead individuals to develop dengue or severe dengue, described as hemorrhagic fever or shock syndrome 11,12 . On the other hand, ZIKV cases are generally self-limiting febrile illnesses like dengue fever, but ZIKV has been associated with more severe outcomes such as Guillain-Barré syndrome (GBS) and birth defects 13 .
During the peak of the ZIKV epidemic in 2016, there was little DENV transmission in the Americas. This has been linked to the extensive cross-reactivity between antibodies 6,14,15 and T cells 16 Similarly, CD4 + T cells have been shown to be important in avivirus infections, displaying functional plasticity exerting cytotoxic characteristics as a function of previous infections 24 and contributing to protection [25][26][27] . Recently, Rouers et al. provided a detailed dissection on the balance between protection or harm depending on T cells' phenotype in response to primary or secondary dengue infections 28 . Previously, our group determined that cross-protection is associated with the interval of time between DENV and ZIKV infections and mediated by cellular immune responses, particularly CD4 + T cells 29 . An area of active discussion is their role in the context of primary and secondary avivirus infections where it ranges between being polarized to a T helper 1 cell 30 and aiding B cells in the germinal center (GC) 31 to CD4-restricted HLAs associated with less severe infection outcome, expansion of T follicular cells promoting DENV-speci c antibodies and cytotoxic subpopulations as a result of re-exposures 24,32,33 . Detailed characterization of the maturation of the humoral immune response during secondary infections or vaccinations indicates that during secondary avivirus exposure, the GC reaction, where the CD4 + T cells play a crucial role in naïve B cells activation and immunoglobulin switching, may not be critical for an optimal secondary immune response 34 . Other groups also have con rmed the involvement of memory B cells (MBC) and MBC-derived plasmablasts in the humoral immune response more than the activation of naïve B cells during secondary avivirus infections 14,35−40 and emphasizing the role of cross-reactive CD4 + T cells 31 .
Despite these contributions, the changes in the functional quality of avivirus-induced antibodies during immune recall responses remain less well characterized. More importantly, the role of CD4 + T cells in controlling aviviral replication by generating polyfunctional responses and the quality of the antibodies produced by tertiary infections in avivirus experienced humans or non-human primates (NHPs) is largely unknown. NHP models provide advantages such as human-like immune responses and control of external factors, such as injection method, amount of administered viral inoculum, and infection timing 41 . Moreover, their competent immune system resembles those of humans, which is essential for understanding the processes driving disease development and has been broadly used to study DENV and ZIKV responses [41][42][43] . Using this model, our group provided the rst evidence that prior DENV immunity is bene cial against ZIKV infection 44 which was con rmed later on by two other groups 45,46 . These results from NHPs on the limited impact of pre-existing dengue immunity in ZIKV infection outcomes were rst con rmed in humans by Terzian et al. 47 followed by other groups [48][49][50] .
To address the gaps in knowledge regarding the role of CD4 + T cells and the impact of avivirus priming in sequential infections, we performed a longitudinal study focused on the adaptive immune responses.
We assessed the contribution of CD4 + T cells in viral clearance and aid in producing a robust humoral response in rhesus macaques. To test this, DENV or ZIKV-primed CD4 + T cell-depleted, non-depleted, and avivirus-naïve rhesus macaques were exposed to either DENV-2 (n=8) or DENV-4 (n=20). We found that the absence of CD4 + T cells in DENV-primed animals with a secondary ZIKV infection before a tertiary heterologous dengue challenge resulted in a signi cant drawback for the immune responses, including a delay in IgM and IgG responses, as well as a decrease in the overall magnitude of the antibody response, in addition to a reduction in the neutralization capacity of neutralizing antibodies. However, before a third challenge, the lack of CD4 + T cells had no evident impact in ZIKV-primed animals, followed by a sequence of two consecutive heterologous dengue infections.
Furthermore, DENV-primed CD4-depleted cohorts showed plasmablast populations with delayed, impaired isotype switching and reductions in antibody binding during tertiary infections. Importantly, DENV-primed CD4 + T cells are necessary for a robust humoral response against a heterotypic infection with DENV, even in the context of a tertiary infection. However, in ZIKV-primed individuals, where DENV infections were consecutive, the role of CD4 + T cells in modulating the quantity and quality of the immune response was more limited, con rming implications of the order and timing of infections in the hierarchy of the non-neutralizing and neutralizing antibody maturation and function.
Our ndings suggest that CD4 + T cells play a crucial role in shaping the quantity, quality, and magnitude of the humoral immune response during tertiary infections. However, that contribution is modulated by multiple variables, including the relatedness among the infecting viruses and the sequence and time of with the same conditions. After incubation, overlay was added and processed as previously described. Results were reported as the FRNT with a 60% or greater reduction in DENV foci (FRNT60  and side scatter pattern; T cells were selected with a second gate on the CD3 positive population. CD8 + T cells were de ned as CD3 + CD8 + and CD4 + T cells as CD3 + CD4 + . Naive (N=CD28 + CD95 -), effector memory (EM=CD28 -CD95 + ) and central memory (CM=CD28 + CD95 + ) T cell subpopulations were determined within CD4 + and CD8 + T cells. B cells were de ned as CD20 + CD3 -, Memory (CD20 + CD3 -CD27 + ) and Plasma (CD20 -CD3 -CD27 + Ki67 + ) B cell subpopulations were determined within CD20 + B cells. Activation marker CD69 and proliferation marker Ki67 were determined in each different lymphoid cell subpopulation. As we did not label cells with CD38 + or CD19 + markers, we could not differentiate from plasmablast (CD20 -CD27 + CD19 + CD38 + Ki67 + ) or plasma cells (CD20 -CD27 + CD19 -CD38 ++ Ki67 -). From here, throughout the text, the CD20 -CD3 -CD27 + are referred to as Antibodies Secreting Cells (ASC). Data analysis was performed using Flowjo (FlowJo LLC Ashland, OR).

Statistical Methods
Statistical analyses were performed using GraphPad Prism 9.0 software (GraphPad Software, San Diego, CA, USA). For viral burden analysis, the log titers and levels of vRNA were analyzed by multiple unpaired ttests and two-way ANOVA. Also, a Chi-squared test was used to analyze a contingency table created from obtained viremia data. The statistical signi cance between or within groups evaluated at different time points was determined using two-way analysis of variance (ANOVA) (Tukey's, Sidak's, or Dunnett's multiple comparisons test) or unpaired t-test to compare the means. Signi cant multiplicity adjusted pvalues (*< 0.05, **< 0.01, ***< 0.001, ****< 0.0001) show statistically signi cant differences between groups (Tukey test) or time-points within a group (Dunnett test).

Rhesus macaque cohorts, CD4 T cell depletion, and sample collection
The experimental design includes two cohorts of rhesus macaques (Macaca mulatta), subdivided by aviviral immunological background and time of infections and immune depletion status (some were depleted of CD4 + T cells while others were not) and further challenged with DENV-2 NGC-44 (cohort A) or DENV-4 Dominique (cohort B) ( Fig. 1). Experiment 1 served as a preliminary study cohort. Based on the results obtained in experiment 1, and the fact that after the introduction of ZIKV to DENV endemic areas, populations with primary ZIKV infections emerged, we decided to expand the study and the infection sequences. The depletion treatment was e cacious as a 90.1% CD4 + T cell depletion was reported for group A-1 (Supp Fig. 1A and B4), no differences in the enzyme pro le between the CD4 + T cells-depleted and the control group were noted.

DENV-4 RNAemia pro le is modi ed by CD4 T + cells hindrance in DENV-primed individuals
To determine if depletion of CD4 + T cells alters DENV replication kinetics, DENV RNAemia levels were measured in serum using qRT-PCR. For experiment 1, all avivirus-positive animals experienced an early delay in viremia with a late peak compared to the naïve group ( Fig. 2A). Contrary to our expectations, DENV3/4-ZIKV CD4-depleted animals showed the most noticeable delay during the acute phase of the DENV2 tertiary infection with a late peak on day 8 post-infection (p.i.) and detectable viremia by day 10 p.i. compared with the rest of the animals. For the DENV3/4-ZIKV immune-competent group, the RNA detection was consistent with typical heterotypic infection rates with lower replication than the naïve group and a late peak on day 6 p.i. However, when evaluating the area under the curve (AUC), we found that the immune-competent animals with prior avivirus immunity could control the viremia to signi cantly better than the naïve group (p=0.0212). There were no differences in the total viremia between the DENV/ZIKV CD4-depleted and naïve animals. Additionally, the CD4-depleted animals had AUC values with a tendency, although not statistically signi cant, tendency to be higher than the DENV/ZIKV immune-competent animals (Fig. 2B). Next, we evaluated the average RNAemia days de ned as the days with detectable viremia divided by total possible viremia days. We noted that animals depleted of CD4 + T cells had similar viremia days to the naïve group. In contrast, although not statistically signi cant, the DENV/ZIKV immune-competent group showed the lowest viremia days (Fig. 2C). For DENV2-ZIKV-DENV4 groups (Experiment 2, B-1, B-2 & B-5), we observed the same trend to an early delay in viremia with a late peak in all avivirus-experienced animals (Fig. 2D). However, since DENV-4 replication is not as robust and consistent as DENV-2, we did not observe statistical differences in the AUC (Fig. 2E) or mean RNAemia days. Nevertheless, there was a tendency to higher values in the DENV/ZIKV CD4-depleted groups compared to the DENV/ZIKV immune-competent animals (Fig. 2F). Taken together, these results suggest that, although limited statistically signi cant differences among groups were observed, CD4 + T cell depletion negatively impacts the control of DENV viremia in DENVprimed individuals with a secondary ZIKV infection during a tertiary DENV infection regardless of the sequence of infecting DENV serotypes or timing of infections.

Depletion of CD4 + T cells modi es the serological pro le during heterologous infections in DENV-primed individuals
To further explore the role of CD4 + T cells, we assessed their impact on the quantity of the humoral response against a tertiary DENV infection. All twenty-two DENV-primed animals were tested for seroreactivity against DENV.  . 3H). However, a limited fold-increase in the total IgG from day 0 to 15 and 0 to 30 p.i. and a signi cantly lower total IgG was observed in all DENV/ZIKV CD4-depleted groups compared to the DENV/ZIKV immune-competent animals (Fig. 3E, p=0.0036). Taken together, the IgG pro le suggests that CD4 + T cell depletion impairs the amnestic response during tertiary avivirus infection after sequential DENV-ZIKV infections.
Next, we measured the binding ability of the antibodies in the sera from depleted and non-depleted cohorts at 0, 15, and 30 days after DENV infection against the whole viral particle ( with a signi cant difference with the DENV/ZIKV immune-competent animals on day 30 p.i. (Fig. 5A,  p=0.0046). However, this delay was partially recovered by day 60 p.i., having all avivirus-experienced groups display a similar neutralization pro le. For experiment 2, we observed a similar trend with the DENV2/ZIKV CD4-depleted group, which showed a trend towards lower neutralization magnitude by day serotype, compared to its respective DENV2/ZIKV immune-competent group (Fig. 5D, p=0.0118). Interestingly, the delay in the expansion of the neutralization in the depleted group is con rmed by the statistically signi cant increase in magnitude on day 60 p.i. (p=0.0049), which was apparent only in this group. That late increase resulted in a signi cant difference, with higher values in the depleted group compared to the DENV2/ZIKV immune-competent group (Fig. 5C, p=0.0487). Figures 5E and 5F summarize the neutralization titers for experiments 1 and 2. Per the EC50 values, the same trend was observed on the PRNT values. The DENV/ZIKV CD4-depleted groups showed a delay in neutralizing titers in acute and convalescent periods compared to all avivirus-experienced animals (Fig. 5E). In addition, when analyzing the AUC, there was a slight non-signi cant trend to a lower AUC value in the DENV/ZIKV CD4-depleted groups compared to the immune-competent group (Fig. 5F). Collectively, these results con rm that regardless of the sequence or timing of the DENV infection, depletion of CD4 + T cells negatively impacts the magnitude of neutralization during a tertiary avivirus infection in DENV-primed individuals followed by a secondary ZIKV infection.

Tertiary infection with DENV-4 is not enhanced in ZIKV-primed individuals
Since the introduction of ZIKV in DENV endemic areas, populations with primary ZIKV infections emerge.
Therefore, to explore the role of avivirus infection sequences in the functionality of CD4 + T cells and the outcome of a tertiary infection, we challenged six animals with DENV-4 two years after they had a primary infection with ZIKV and a secondary challenge with DENV-2 three months after (Fig. 1). For experiment 2 (groups B-3, B-4 & B-5), we noted the same tendency for DENV/ZIKV-groups, with an early delay in DENV-4 viremia more evident in the ZIK/DENV2 CD4-depleted animals compared to the ZIKV/DENV2 immune-competent group (Fig. 6A). However, as in the previous experiment, the depleted group showed a higher and late peak viremia on day 5 p.i., suggesting a limited viremia control linked to the lack of CD4 + T cells. Nevertheless, different from the DENV/ZIKV cohorts, we did not observe differences in the AUC (Fig. 6B) or mean viremia days (Fig. 6C) among ZIKV/DENV groups. This result suggests that the role of CD4 + T cells controlling the viremia during a tertiary avivirus infection may have different weights depending on the prior sequence and timing of infections.
CD4 + T cells depletion did not change the anti-dengue humoral pro le in ZIKV-primed individuals Next, we wanted to evaluate how the viremia of CD4 + depleted animals correlated to their serological pro le, assessing the quantity and quality of the humoral response. Animals were tested for seroreactivity and binding ability of the sera-derived antibodies against the whole DENV-4 viral particle by an endpoint dilution approach. As expected, all avivirus-experienced animals had a limited level of anti-DENV IgM in comparison to the induction of a primary IgM response in the naïve groups. Total anti-DENV IgM levels were slightly lower in the ZIKV/DENV CD4-depleted group than the ZIKV/DENV immune-competent group ( Fig. 7A) but not as prominent as in the DENV/ZIKV-immune animals (Fig. 3B). Interestingly, there was no signi cant difference in the total IgG levels between ZIKV/DENV2 CD4-depleted and ZIKV/DENV2 immune-competent animals (Fig. 7B), as reported in the DENV/ZIKV-exposed groups (Fig. 3D). When we characterized the functional properties of the antibodies against DENV in ZIKV-primed individuals, at baseline, all avivirus-experience animals showed cross-reactivity with titers up to a dilution of 1x10 4 with no differences in their AUC values, respectively (data not shown). In addition, on day 15 p.i., no differences were observed between ZIKV/DENV2 groups (Fig. 7C & 7D), indicating that CD4 + T cell depletion did not affect the binding of the antibodies to DENV-4 when ZIKV was the initial avivirus priming the immune system followed by a secondary DENV preceding a tertiary heterologous DENV infection. Lastly, we evaluated the antibody neutralizing capabilities during a tertiary infection against DENV4. Interestingly, ZIKV-DENV2-DENV4 CD4-depleted animals had a similar neutralizing pro le to their control immune-competent animals against DENV4, with no differences observed in any of the timepoints tested (Fig. 7E) as it was in the case for the DENV2-ZIKV-DENV4 sequence (Fig. 5C). These results suggest that the role of CD4 + T during a tertiary heterologous DENV infection can be modi ed by the sequence and timing of avivirus infection.

Lack of CD4 + T cells does not affect the recall humoral immune response to previous infecting aviviruses
Next, we reviewed if depleting CD4 + T cells changed the functional properties of the antibodies during the recall memory response to previous infecting serotypes. We measured the binding ability of the antibodies in the sera from depleted and non-depleted cohorts before and 15 days after DENV infection against the whole viral particle ( To further explore how the role of CD4 + T cells can be modulated by order of infections, neutralizing capabilities against previous infecting aviviruses (DENV-4, DENV-3, DENV-2, and ZIKV) were tested for experiments 1 and 2 (Supp Fig S6). The EC 50 of nAbs for experiments 1 and 2 are shown (Fig. 8). For experiment 1, DENV3/4-ZIKV-DENV2 immune animals had similar neutralizing responses against their previous infecting serotypes (DENV-3 or DENV-4) regardless of the status of CD4 + T cells (Fig. 8A & B).  Fig. 8C-D). We detected an expansion in antibody-secreting cells (ASC) (CD20 − CD3 − CD27 + ) in DENV/ZIKV groups at 7-and 10 p.i., while ZIKV/DENV groups maintained their levels (Supp Fig. 8E-F). No differences were detected in the memory B cells (CD20 + CD3 − CD27 + ) between both infecting sequences. However, the ZIKV/DENV groups showed a higher frequency of activated cells than the DENV/ZIKV groups. Moreover, a higher frequency of (ASC) was detected in ZIKV/DENV-primed, but not in DENV/ZIKV-primed individuals, with a statistically signi cant difference at day 7 p.i. with their respective baselines (Fig. 9E-F p= 0.0003, p=0.0001). Remarkably, there were no differences between the depleted and the non-depleted groups in that sequence of infection (Fig. 9F). To further explore the role of CD4 + T cells and the hierarchy of infections in shaping the frequency, activation, and proliferation of cellular immune cell subsets for experiment 2 we measured whether T-cell subpopulations, such as naïve (CD3 + CD8 + CD28 + CD95 − ) effector memory (CD3 + CD8 + CD28 − CD95 + ) and central memory (CD3 + CD8 + CD28 + CD95 + ) T cells, within CD8 + T cell compartment were modulated by CD4 + T cell depletion and infection sequences following a tertiary DENV infection (Supp Fig. 10; Supp Fig. 11 for gating strategy). We observed different CD8 + T cell pro les depending on the sequence of infections (Supp Fig. 9. A-L). No major differences were observed when evaluating naïve (Supp Fig. 10. experienced animals. On the other hand, signi cantly higher CXCL10 (IP-10) levels (Supp Fig. 13E), a Tcell activating chemokine, and chemoattractant for many other immune cells were detected in ZIKV/DENV2 animals at day 7 p.i. (p=0.0111) than DENV2/ZIKV animals. No signi cant differences in the naïve group were observed. Collectively, these results demonstrate that the order of infecting aviviruses prior to a tertiary infection can outweigh the role of the CD4 + T cells modulating the proin ammatory cytokine/chemokine pro le.
However, the DENV/ZIKV/DENV CD4-depleted animals showed a noticeable rebound on day 8 p.i., no resolution of the viremia by day 10, and a strong trend to higher viremia AUC values even in the presence of prior immunity, con rming that CD4 + T cells are needed for sustained viremia control when a secondary ZIKV infection precedes a tertiary DENV infection. This agrees with our previous work, Prior results indicate that sequential immunizations for aviviruses sharing CD4 + epitopes should promote protection during a subsequent heterologous infection. However, as supported by our work, sequential infections with different Yellow Fever serocomplex viruses (DENV and JEV) result in different immune pro les 31 . Additionally, similar amounts of circulating ASC in primary and secondary viral infections helped postulate that pre-existing CD4 + T helper cells may not be required for optimal responses in some viral infections [61][62][63][64] , and may depend on the nature of the antigen 65 . Correspondingly, using a mouse model, Yauch et al. showed that during a primary DENV infection without prior avivirus immunity, CD4 + T cell depletion did not impact the DENV-speci c IgM/IgG Ab titers and their neutralizing activity 66 . Our data add to the role of the CD4 + T cells in a more complex immune background, showing that CD4 + T cells depleted groups from the DENV-ZIKV-DENV infection sequence have a faster increase in total DENV-speci c IgM and IgG antibodies, with lower a nity than CD4 + competent groups. Conversely, we were unable to identify the same effect in the ZIKV-DENV-DENV sequence. These differing results con rm a multifactorial setting controlling the contribution of CD4 + T cells in avivirus infections.
It is known that the cross-reactive humoral response is broader in secondary DENV infections derived from MBC clonal expansion compared to predominant ZIKV-speci c antibodies in primary ZIKV or DENV-ZIKV scenarios 67,39 . We hypothesized that sequential secondary and tertiary DENV infections induce a CD4 + T cell-independent clonal expansion of DENV-speci c MBC producing antibodies that, at a functional level, keep their high a nity contributing most strongly to DENV neutralization. On the other hand, during a secondary ZIKV infection, a virus outside the dengue serocomplex and group 68,69 and with different immunodominant epitopes 49,70,71 , more DENV-speci c MBC generated during the primary DENV infection, may undergo hypersomatic mutation and lose cross-reactivity lowering the a nity, determined by the magnitude and speci city of the neutralization to the tertiary infecting DENV. This effect is con rmed by the lower a nity of the anti-DENV and ZIKV antibodies in the absence of CD4 + T cells. As shown before, the immune responses induced by ZIKV and DENV as secondary infecting viruses are different. ZIKV infection in DENV-naïve subjects or with prior DENV immunity induces both DENV-MBC and naïve B cells with the production of ZIKV type-speci c antibodies in both cases. In contrast, cross-reactive MBC predominates after DENV infection 14,67 . Essentially our results indicate that the dynamic of the CD4 + T and B cells interaction during a tertiary aviviral infection is modulated by the prior, secondary, infecting virus. Results for the recall memory were also interesting. We found that the neutralization hierarchy in both sequences of experiment 2 during the tertiary DENV-4 challenge were similar (DENV2>ZIKV>DENV4) resembling the Original Antigen Sin (OAS) postulate. In both cases, it was broadly cross-reactive but of higher magnitude against DENV-2, which was the primary infecting DENV serotype. However, in the sequence DENV3/4-ZIKV-DENV2 (experiment 1), despite the limited number of animals, the neutralization was clearly dominated by DENV-2, the infecting serotype. The neutralization hierarchy was DENV2>ZIKV≥DENV3/4. Others have reported that during secondary DENV infections, the response is primarily directed to the current infecting serotype, most likely due to a nity maturation in the GCs, resulting in the selection of MBCs with antibodies directed to the current infecting serotype 72 .
Nevertheless, the hierarchy orders from our results, raise inquisitiveness on the potential immunodominance of DENV-2 over the other serotypes. Our results from experiment 2 agree with other works reporting cross-reactive plasmablasts more reactive towards the previous infecting DENV serotype 40,73,74 . It has been documented that the plasmablast response during secondary DENV infection is mainly derived from MBC 34,65,67 . However, there is no information on MBC or plasmablasts dynamic during tertiary avivirus infection. We found a higher frequency of activated ASC in the ZIKV-DENV2-DENV4 cohorts, regardless of the CD4 + T cells status, compared to the DENV2-ZIKV-DENV4 groups (experiment 2) increasing by day 7 p.i. resembling the plasmablast dynamic after DENV infections 39,40,73,74 . Also, the fact that in the ZIKV-DENV2-DENV4 sequence, with two consecutive dengue infections the neutralizing magnitude was higher against the priming DENV serotypes and not against the current infecting dengue serotype and that the neutralization against the infecting serotype was also independent of CD4 + T cells, strongly supports (i) that the antibodies originated from reactivation of MBC plasmablasts after the tertiary infection 40 and (ii) that the strong plasmablast response may neutralize the infecting agent without a secondary a nity maturation in GC 14,34 , suggesting that the infection's order plays a crucial role in regulating the a nity maturation process in the GC.
The fact that the hierarchy of neutralization was unaffected regardless of the order of ZIKV infection related to the two DENV infections is also a noteworthy nding. However, it is in line with prior observations showing that DENV/ZIKV cross-reactive MBC response decreased over time post-ZIKV infection 67 . In our case, the ZIKV infections occur at the same time, 13 months before the tertiary DENV infection in both sequential infections.
We evaluated the CD8 + T cell response on day 7 p.i. in both sequences of infections in experiment 2 by assessing the frequency of virus-speci c CD8 + T producing IFN-γ + or TNF-α + or expressing CD107a + , a marker associated with cytotoxicity. We were unable to identify any statistically signi cant differences when the cells were stimulated with the whole DENV2, DENV4 or ZIKV. However, in animals with the sequence DENV2-ZIKV-DENV4, there was a trend towards lower IFN-γ + producing cells in the CD4 + T cells, a trend that was not apparent in the ZIKV-DENV2-DENV4 sequence. Previous studies showed that expanded activated CD4 + T cells located near CD8 + T cells in the spleen after a primary DENV-2 infection did not affect the induction of DENV-2-speci c CD8 + T cells 66  Experimental design for CD4 + T cell depletion and heterologous DENV challenge in avivirus experienced and naïve macaques. Two cohorts of rhesus macaques (Macaca mulatta) were exposed to DENV and ZIKV virus at different time points. Cohort A: A-1 (n=2) and A-2 (n=2), were exposed to DENV-3 or DENV-4 (5x10 5       The quantity and quality of the humoral response are not affected by CD4 + T cell depletion in ZIKVprimed individuals. The quantity of the humoral response was assessed using different commercial