Time elapsed between Zika and Dengue infections improves the immune response against Dengue without viremia enhancement in rhesus macaques

1 Department of Microbiology and Medical Zoology, University of Puerto Rico-Medical Sciences Campus, San Juan, Puerto Rico, United States of America. 2 Unit of Comparative Medicine, Caribbean Primate Research Center and Animal Resources Center, University of Puerto RicoMedical Sciences Campus, San Juan, Puerto Rico, United States of America. 3 Department of Molecular Microbiology & Immunology, Saint Louis University School of Medicine, Saint Louis, Missouri, United States of America. 4 Department of Biology, University of Puerto Rico-Río Piedras Campus, San Juan, Puerto Rico, United States of America. 5 Texas Biomedical Research Institute, San Antonio, Texas, United States of America. 6 Departments of Microbiology & Immunology, University of North Carolina-Chapel Hill, North Carolina, United States of America. 7 Division of Vaccine Discovery, La Jolla Institute for Immunology, La Jolla, California, United States of America. 8 Department of Internal Medicine, University of Puerto Rico-Medical Sciences Campus, San Juan, Puerto Rico, United States of America.


Introduction
Zika virus (ZIKV) is a re-emerging mosquito-borne Flavivirus that has captivated the attention of the scientific community by its explosive spread in The Americas 1 , and severe neurological sequelae following infection [2][3][4][5] . ZIKV established itself in tropical and sub-tropical regions that are endemic to other closely-related flaviviruses such as Dengue virus (DENV). Both viruses belong to the Flaviviridae family and are transmitted by Aedes spp. mosquitoes. DENV is a global public health threat, having two-thirds of world's population at risk of infection, causing ~390 million infections annually, and possessing one of the highest mortality rate among arboviruses 6,7 . DENV exists as four genetically similar but antigenically different serotypes (DENV1-4) 8 . Exposure to one DENV serotype confers long-lived immunity against a homotypic secondary infection. However, secondary infection with a heterologous serotype of DENV is the major risk factor to induce severe DENV disease [9][10][11] .
Due to the established history regarding the influence of cross-reactive immune interactions in dictating disease outcomes during heterologous infection and the genetic and consequently antigenic similarities between DENV and ZIKV, concerns have been raised regarding the impact of DENV-ZIKV cross-reactive immunity on the development of severe clinical manifestations 12,13 . In the last few years, multiple studies have aimed to understand the role of a prior DENV exposure in the outcome of ZIKV infection. It has been demonstrated that DENVimmune sera from humans can enhance ZIKV infection in vitro 14,15 , and in vivo in immune-deficient mouse models 16 . However, recent results from our group and others have shown that previous flavivirus exposure-including DENV-may have no detrimental impact on ZIKV infection in vivo in non-human primates (NHP) 17,18 and humans 19 . Moreover, these studies and others suggest that previous DENV immunity may play a protective role during ZIKV infection involving humoral and cellular responses [20][21][22][23][24] . On the other hand, little is known about the opposite scenario, the role of a previous ZIKV exposure on a subsequent DENV infection, which is relevant to anticipate the dynamics of forthcoming DENV epidemics.
The recent ZIKV epidemic in the Americas resulted in the development of a herd immunity that may have an impact in subsequent infections with other actively circulating flaviviruses such as DENV. Thus, human sub-populations such as newborns, international travelers from nonflavivirus endemic areas or DENV-naïve subjects could be exposed to a ZIKV infection prior to DENV-since DENV declined in the Americas during ZIKV epidemic 25 . After the epidemic, herd immunity reduced ZIKV transmission and DENV will re-emerge and potentially infect these DENVnaïve ZIKV-immune sub-populations in The Americas or potentially in other geographic areas newly at risk 26,27 . An epidemiological study based on active Dengue surveillance in Salvador, Brazil, suggests that the reduction of DENV cases after the ZIKV epidemic is due to protection from cross-reactive immune responses between these viruses 28 . Prospective experimental studies are needed to confirm this hypothesis. For this purpose, we propose the use of NHPs as a suitable model. NHPs provide advantages such as an immune response comparable to humans, and the normalization of age, sex, injection route, viral inoculum and timing of infection 29 .
ZIKV antibodies (Abs) are capable of enhancing DENV infection in vitro 35 .
Characterization of the specificity of DENV and ZIKV cross-reactive response revealed that ZIKV monoclonal Abs-and more recently-maternally acquired ZIKV Abs can increase DENV severity and viral burden in immune-deficient mouse models 36,37 . However, little evidence is available concerning this phenomenon occurring in vivo in immuno-competent large animal models such as NHPs. George et al., showed that an early convalescence to ZIKV induced a higher peak of DENV viremia and a pro-inflammatory status compared to ZIKV-naïve status in macaques 38 .
Further characterization of ZIKV early convalescent sera from these macaques indicated that there is a delayed induction of the cross-reactive Ab response against DENV, supporting no cross-protection against the outcome of DENV infection 39 . Despite these findings, further studies are needed to dissect the complementary role of the innate, humoral and cellular immune response to mechanistically explain these findings. Particularly, there is no evidence of the modulation and functionality of the T cell immune response in the ZIKV-DENV scenario. Available studies rely upon pathogenesis and antibody studies, but there is no documented evidence as to whether cell-mediated immunity (CMI)-specifically the polyfunctional response of T cells-is modulated in a subsequent DENV infection by the presence of ZIKV immune memory.
The time interval between primary and secondary DENV infections have been shown to be an important predictor for the development of severe clinical outcomes in humans 10 . Shorter time interval between DENV infections result in a subclinical secondary infection, while symptomatic secondary infections and severe DENV cases have been related with longer periods between infections [40][41][42][43] . These findings suggest that high titers of cross-reactive Abs play a timedependent protective role between heterotypic DENV infections. Despite this evidence from DENV sequential infections, it remains poorly understood if the same applies to the time interval between ZIKV-DENV sequential infections. Specifically, how do longer periods of convalescence after ZIKV infection impact the outcome of DENV infection. This scenario will more closely resemble the epidemiological setting and time intervals elapsed between the current circulation of related flaviviruses in the Americas. So currently, the role of multiple convalescent periods to ZIKV in the outcome of DENV and other flavivirus infections is in the forefront of discussions based on the limited studies available in experimental models and a lack of characterized human prospective cohorts of this scenario yet 28,[44][45][46] .
To address these knowledge gaps, the objective of our study is to investigate the immune modulatory role of an early-and mid-convalescence after ZIKV infection on the outcome of a subsequent DENV infection in a NHP model. To test this, NHP cohorts who were ZIKV immune for 10 months (mid-convalescence), 2 months (early-convalescence) or naïve for ZIKV were exposed to DENV. In each of these groups we assessed DENV pathogenesis, the elicited Ab response, and characterized the CMI. Based on our knowledge, this is the first characterization of CMI with this scenario-taking into account the synergistic effect between the Ab and cellmediated responses. This study provides evidence that the presence of ZIKV immune memory contributes to improve the immune response-more efficient after longer ZIKV pre-exposureagainst a DENV infection, without promoting enhancement of DENV viremia nor inducing higher levels of pro-inflammatory cytokines.

Results
DENV challenge and clinical status of rhesus macaques. The experimental design includes three cohorts of rhesus macaques (Macaca mulatta) that were challenged with DENV-2 (NGC-44 strain), monitored and bled during three months (Fig. 1). Two were previously exposed to ZIKV: cohort 1 (ZIKVPF-10mo) was comprised of 4 animals that had been exposed to ZIKV H/PF/2013 strain 10 months before DENV-2 challenge (mid-convalescence), and cohort 2 (ZIKVPR-2mo) comprised of 6 animals that had been exposed to ZIKV PRVABC59 strain two months before DENV-2 challenge (early-convalescence). Both ZIKV strains used for previous exposure of these groups are >99.99% comparable in amino acid identity (Supplementary Table 1). An additional cohort 3 (Naïve) included four animals naïve to ZIKV/DENV as a control group. After DENV challenge all macaques were extensively monitored and sample collection was performed at various timepoints up to 90 days post infection (dpi) for serum and PBMCs isolation.
The clinical status was monitored to determine if the presence of ZIKV immunity affected the clinical outcome of DENV infection. Vital signs such as weight (kg) and temperature (°C) were monitored. Also, complete blood cell counts (CBC) and comprehensive metabolic panel (CMP) were performed before (baseline: day 0) and after DENV infection at multiple timepoints (CBC: 0, 6 7, 15 dpi; CMP: 0, 7, 15, 30 dpi). Neither symptomatic manifestations nor significant differences in weight or temperature were observed in any of the animals after DENV infection up to 90 dpi (Supplementary Fig. 1a-b). Likewise, no significant differences between groups were detected in CBC parameters: white blood cells (WBC), lymphocytes (LYM), neutrophils (NEU), monocytes (MON), and platelets (PLT) after DENV infection compared to basal levels of each group (Supplementary Fig. 1c-g). CMP levels showed no differences in alkaline phosphatase and aspartate transaminase (AST) (Supplementary Fig. 1h-i). Although within the normal range, levels of alanine transaminase (ALT) were significantly higher in the ZIKVPR-2mo group compared to its baseline at 7 dpi (p=0.0379, Two-way Anova Dunnett test), at 15 and 30 dpi values returned to baseline levels ( Supplementary Fig. 1j). Overall, except for the isolated increase of ALT at 7 dpi in ZIKVPR-2mo, the clinical profile suggests that the presence of ZIKV-immunity did not significantly influence the clinical outcome of DENV infection.

DENV RNAemia is not enhanced by previous ZIKV immunity. RNAemia levels in NHPs serum
were quantified by qRT-PCR at baseline, 1 to 10, and 15 dpi to determine if the presence of early-(ZIKVPR-2mo) or mid-convalescence (ZIKVPF-10mo) to ZIKV alters DENV kinetics. No significant differences between groups were observed in detected levels of DENV genome copies per ml of serum over time ( Fig. 2a; Supplementary Table 2). We noted that in the ZIKVPF-10mo group 3 out of 4 animals were able to keep the RNAemia level below 10 3 genome copies the next day after DENV infection. This group started an early clearance of the RNAemia at 7 dpi, with only 1 out of 4 animals having detectable levels by days 8 and 9 pi. For ZIKVPR-2mo and naïve animals, the clearance of detectable RNAemia started at 8 dpi, in 4 out of 6 and 1 out of 4 of the animals, respectively. Naïve animals had the most delayed clearance of RNAemia with at least half of the animals with detectable levels of viral RNA until day 9 pi. RNAemia was completely resolved in all animals by 10 dpi. In summary, ZIKVPF-10mo had 7.25, ZIKVPR-2mo 7.5, and Naïve animals 8 mean days of detectable RNAemia after DENV infection ( Fig. 2b; Supplementary Table 2). Together these results show that, although no statistically significant differences among groups were observed, previous immunity to ZIKV is not associated with an increase in DENV RNAemia set point or duration; even more, a mid-convalescence to ZIKV tended to develop a shorter viremic period.
Pro-inflammatory cytokines are not exacerbated by the presence of ZIKV immunity. To determine if the characterized cytokine profile of an acute DENV infection was modulated by ZIKV immunity we assessed the serum concentration (pg/ml) of 8 cytokines/chemokines by Luminex 7 multiplex at baseline, 1, 2, 3, 5, 10, 15 and 30 dpi. The naïve group showed significant higher levels of pro-inflammatory cytokines (Fig. 3a-c). Type I interferon alpha (IFN-α) was highest at 5 dpi ( Fig. 3a: p<0.0001 vs ZIKVPF-10mo and p=0.0003 vs ZIKVPR-2mo, Two-way Anova Tukey test). IFN-α has been demonstrated to be involved in the innate anti-viral immunity and elevated levels are associated with higher viral load and thus antigen availability. Interleukin-6 (IL-6), a multifunctional cytokine involved in immune response regulation and many inflammatory reactions showed the highest levels at 1 dpi in naïve animals ( Fig. 3b:  protein released by activated cytotoxic CD8+ T cells and natural killer (NK) cells. No significant differences between groups were observed in monocyte chemoattractant protein 1 (MCP-1), macrophage inflammatory protein 1-beta (MIP-1β) and IL-1 receptor antagonist (IL-1RA) ( Fig. 3df). Collectively, these results demonstrate that the presence of ZIKV immunity does not exacerbate pro-inflammatory status after DENV infection while mid-convalescence immunity to ZIKV stimulated levels of mediators mainly involved in the activation of cell-mediated immune response.

DENV and ZIKV cross-reactive antibody response is boosted by ZIKV immunity and is influenced by the time span of the previous ZIKV infection. An ELISAs-based serological
profile was performed to determine the contribution of ZIKV immunity in the cross-reactive antibody response before and after DENV infection. We assessed the levels of DENV IgM and IgG, and cross-reactivity with ZIKV (IgM, IgG, NS1-IgG and EDIII-IgG) at multiple timepoints ( Supplementary Fig. 2). Naïve cohort showed a significant higher peak of IgM ( Supplementary   Fig. 2a)  In contrast to the elevated levels of DENV-specific IgM, ZIKV IgM levels were under or near the limit of detection in all groups over time after DENV infection despite several significant differences seen between groups ( Supplementary Fig. 2c). ZIKV IgG levels ( Supplementary Fig.   2d) were similarly high in both ZIKV-immune groups at baseline and 7 dpi compared to naive (p<0.0001 vs naïve, Two-way Anova Tukey test), suggesting that although different pre-infecting ZIKV strains, the previous elicited IgG response against heterologous ZIKV strains is comparable.
After DENV infection, an increase of ZIKV IgG was shown and remain constantly high at 15, 30, 60 and 90 dpi in both ZIKV-immune groups (p<0.0001 vs naïve for all timepoints, Two-way Anova Tukey test), suggesting that DENV has the potential to stimulate ZIKV-binding Ab-producing plasmablasts. In addition, to elucidate the composition of similar ZIKV IgG levels in ZIKV-immune groups, we measured ZIKV-specific NS1 IgG ( Supplementary Fig. 2e) and ZIKV-specific EDIII IgG ( Supplementary Fig. 2f) levels. Although ZIKVPR-2mo showed significant differences compared to naïve at 30, 60 and 90 dpi (p<0.0001, p=0.0001, p=0.0159; Two-way Anova Tukey test), we observed a significantly higher expansion and long-lasting response of ZIKV NS1specific Abs in the ZIKVPF-10mo group compared to the ZIKVPR-2mo group at baseline, 60 and 90 dpi (p=0.0036, p=0.0071, p=0.0294; Two-way Anova Tukey test) and also compared to naïve animals at all timepoints (p<0.0001, Two-way Anova Tukey test). Moreover, higher magnitude of ZIKV-specific EDIII IgG levels in the ZIKVPF-10mo group than in the ZIKVPR-2mo group was observed compared to naïve at baseline (ZIKVPF-10mo only), 15, 30 and 60 (ZIKVPF-10mo vs Naïve: p=0.0092, p<0.0001, p<0.0001, p=0.0034; ZIKVPR-2mo vs Naïve: p=0.0003, p=0.0014, p=0.0055; Two-way Anova Tukey test), suggesting that ZIKV mid-convalescence promotes an expansion of higher magnitude of ZIKV EDIII IgG Abs from ZIKV memory B cells. However, those higher cross-reacting levels decrease over time as expected. In summary, a boost of DENV and ZIKV Abs is triggered by the presence of ZIKV immunity and the expansion of specific-and cross-. CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint reactive Abs is of higher magnitude and long lasting when a mid-convalescence immunity to ZIKV is present.
Neutralizing antibody response against DENV-2 and heterologous serotypes is higher in magnitude and durability in presence of mid-convalescence to ZIKV. Neutralizing antibodies (NAbs) are essential to combat DENV and ZIKV infection. The maturation and potency of this response is known to define to a great extent the infection outcome 12,47 . Accordingly, we tested the neutralization capacity of NAbs in serum from ZIKV-immune and naïve animals before and after DENV infection, to determine whether an early-or mid-convalescence to ZIKV affected the neutralizing Ab response. Plaque Reduction Neutralization Test (PRNT) was performed to elucidate the NAb titers of all groups against all DENV serotypes and both ZIKV pre-infecting strains. Before infection with DENV the naïve groups had no detectable NAb levels (<1:20 PRNT60 titers) against all DENV serotypes, while ZIKV-immune groups showed low cross-NAb titers against DENV-2 and DENV-4 (Fig. 4a). These cross-reactive levels were higher in the ZIKVPF-10mo group than in the ZIKVPR-2mo group for both viruses. The peak of high NAb titers occurred at 30 days after DENV infection for all groups (ZIKVPF-10mo>ZIKVPR-2mo>Naïve) against all DENV serotypes (DENV-2>DENV-4>DENV-3>DENV-1) (Fig. 4b). The ZIKVPF-10mo group neutralized all DENV serotypes with significant higher potency than naïve animals (p<0.0001, p=0.0337, p<0.0001, p<0.0001 for DENV1-4; Two-way Anova Tukey test) and the ZIKVPR-2mo group, except for DENV-2, that both ZIKV-immune groups have comparable neutralization magnitude at 30 dpi (p=0.0002, p=0.7636, p=0.0016, p=0.0004; Two-way Anova Tukey test). However, the neutralization kinetics by sigmoidal response curves suggest higher percent of neutralization against DENV-2 overtime in the group with mid-convalescence to ZIKV ( Supplementary Fig. 3). On the other hand, the ZIKVPR-2mo group showed significantly higher potency of the NAb response only against DENV-1 compared to naive animals (p=0.0146; Twoway Anova Tukey test) (Fig. 4b).
In addition, we tested whether the NAb titers that peak at 30 dpi for all groups remain constant over time (up to 90 dpi) against all DENV serotypes ( Fig. 4c-f). In general, the neutralizing response of the ZIKVPF-10mo group was more long-lasting, maintaining higher NAb titers up to 90 dpi compared to ZIKVPR-2mo and naïve groups. Significant differences between ZIKVPF-10mo and ZIKVPR-2mo groups were observed against DENV-1,-3 and -4 at day 30 pi (p=0.0002, p=0.0016, p=0.0004; Two-way Anova Tukey test) and at day 60 pi against DENV-2 and DENV-3 (p=0.0179, p=0.0047; Two-way Anova Tukey test). The neutralizing Ab response of the ZIKVPF-10mo group was even more significantly higher compared to the naïve group at day . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint 15 (only performed for the infecting serotype to monitor early neutralizing activity), day 30, 60 and 90 pi against DENV-2 (p=0.0022, p=0.0337, p=0.0146, p=0.0337; Two-way Anova Tukey test); at day 30 pi against DENV-1 (p<0.0001, Two-way Anova Tukey test); at day 30 and 60 pi against DENV-3 (p<0.0001, Two-way Anova Tukey test); and at day 30 pi against DENV-4 (p<0.0001, Two-way Anova Tukey test). In contrast, the ZIKVPR-2mo group showed a neutralizing Ab response with a magnitude and long-lasting levels comparable to the naïve group, except at day 15 and 30 pi against DENV-2 and DENV-1, respectively (p=0.0067, p=0.0146; Two-way Anova Tukey test). The neutralizing response was long-lasting in the ZIKVPF-10mo group compared to other groups as supported by the data from days 30 and 60 p.i. At day 90 pi, although no significant differences were observed between all groups, the ZIKVPF-10mo group showed a consistent trend to maintain higher NAb titers against all DENV serotypes indicating a higher and long-lasting breadth of cross-neutralization within DENV serocomplex.
Interestingly, comparing DENV-2 NAb titers and DENV genome copies at early and late phase of RNAemia timeframe (baseline NAb titer vs 1 dpi genome copies: early phase; and 7 dpi NAb titers vs 8 dpi genome copies: late phase), an inverse proportion was observed among elevated NAb levels and a subsequent lower (early phase) or no detection (late phase) of DENV genome copies ( Supplementary Fig. 4a). This proportion (>NAb titers: <genome copies) was stronger for the ZIKVPF-10mo group. Therefore, this observation is consistent with the trend of ZIKVPF-10mo group to develop a shorter viremic period (Fig. 2). Although no statistically significant differences between groups in NAbs at baseline and 7 dpi, we observed that only the ZIKVPF-10mo group have PRNT60 cross-reactive NAb titers capable to neutralized 60% or more (~75%) of DENV-2 infection at baseline, but only at the more concentrated serum dilution (1:20) ( Supplementary Fig. 4b). All groups had the property to neutralize 60% or more of DENV-2 one week after infection. However, the ZIKVPF-10mo and ZIKVPR-2mo groups retained this property even at higher serum dilutions, 1:20-1:160 and 1:20-1:80, respectively, compared to naïve animals ( Supplementary Fig. 4c). Collectively, these results demonstrate that a midconvalescence to ZIKV provokes a boost of the magnitude and durability of the neutralizing response against all DENV serotypes more effectively than in animals with an earlyconvalescence to ZIKV, but also higher compared to a de novo DENV-specific NAb response of the naïve animals. Also, our results suggest that the low-to-intermediary levels of crossneutralizing Abs to DENV-2 detected prior DENV infection (induced by previous ZIKV infection) may play an early role controlling DENV set point RNAemia. This qualitative effect seems to be associated to a longer period of convalescence (10 months) but not to a recent ZIKV infection (2 months earlier). The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint Potency of ZIKV cross-neutralizing antibody response is strain-independent and higher in magnitude and durability in the presence of mid-convalescence to ZIKV. Previous exposure to heterologous ZIKV strains in ZIKV-immune groups developed long-lasting high levels of crossreactive, non-neutralizing, and neutralizing Abs before DENV infection (baseline). To determine if this memory Ab response is strain-specific and if the difference in convalescence period to ZIKV alters the efficacy and modulation after DENV infection, we assessed the NAb levels in ZIKVimmune (ZIKVPF-10mo and ZIKVPR-2mo) and ZIKV-naïve (Naïve) serum with both pre-infecting contemporary Asian-lineage H/PF/2013 and PRVABC59 ZIKV strains at multiple timepoints after DENV infection. At baseline, both ZIKV-immune groups showed high NAb titers against H/PF/2013 strain, which suggest that irrespective of pre-exposure to heterologous ZIKV strains and different convalescent periods the Ab response remains similarly effective (Fig. 5a). As early as day 15 after DENV infection, a potent boost of NAb titers in both ZIKV-immune groups was developed. However, elevated NAb titers were significantly higher in the ZIKVPF-10mo group compared to the ZIKVPR-2mo and naïve groups at day 15 pi (p=0.0005, p<0.0001; Two-way Anova Tukey test) and day 30 pi (p=0.0067, p=0.0012; Two-way Anova Tukey test). As expected, this elevated ZIKV cross-reactive NAb levels decreased gradually over time after 15 dpi in both ZIKV-immune groups. Nevertheless, the ZIKVPF-10mo group retained higher NAb titers until 90 dpi while the titers of the ZIKVPR-2mo group returned to baseline levels. In addition to this trend in NAb titers, the difference in durability is clearly demonstrated by the calculation of half maximal (50%) effective concentration (EC50). The serum of the ZIKVPF-10mo group maintained the neutralization capacity at significantly more diluted concentrations compared to the ZIKVPR-2mo group at later timepoints (p<0.0001 for 60 dpi and p=0.0028 for 90 dpi; Two-way Anova Tukey test) (Fig. 5b). Of note, the NAb titers of the naïve group were considered as negative in all timepoints and failed to neutralize ZIKV throughout DENV infection even at concentrated levels of the antibodies ( Fig. 5a-b). These results are confirmed by the behavior of neutralization kinetics by sigmoidal response curves where the ZIKVPF-10mo group retained elevated magnitude of ZIKV neutralization overtime ( Supplementary Fig. 5).
To determine if the strain of ZIKV used in the previous exposures played a role in the modulation of the cross-NAb response triggered by a subsequent DENV infection, NAb titers were measured against both ZIKV strains before and 30 days after DENV infection. The ZIKVPF-10mo group showed significant higher NAb titers to the pre-infecting and heterologous ZIKV strains compared to the ZIKVPR-2mo group before DENV infection (p=0.0093, p=0.0141; Two-way Anova Tukey test) (Fig. 5c). Subsequently, DENV infection promote an equally 8-fold increase of . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint NAb titers against both strains in the ZIKVPF-10mo group, significantly higher than the 4-fold increase in the ZIKVPR-2mo group (p=0.0025, p=0.0011; Two-way Anova Tukey test) (Fig. 5d).
Altogether, these results demonstrate that DENV infection results in a significant increase in the magnitude and durability of the cross-neutralizing Ab response against ZIKV in animals with a mid-convalescent period from ZIKV infection. The elicited changes in neutralization capacity were likely driven more by the longevity of the immune memory maturation and the associated memory recall of the ZIKV immunity than by a strict dependency of the specific pre-exposed ZIKV strain.  Fig. 8). In general, no differences were detected between baseline and after DENV infection in all groups for all NK subpopulations and receptors with the exception of the ZIKVPR-2mo group that showed a significant increases in the following subpopulations: NKG2A + NKp30 and NKp30 + NKp46 + at 7 dpi (p=0.0495, p=0.0006; Two-way Anova Dunnett test) and NKp46 + NKp30 + at 7 and 10 dpi (p=0.0005, p=0.0001; Two-way Anova Dunnett test) ( Supplementary Fig. 8j, o, s).
We next investigated cell subsets that are part of the bi-phasic (humoral/cellular) adaptive immune response such as B (CD20 + CD3 -) and T (CD3 + CD20 -) cells, to determine if convalescent immunity to ZIKV alters responses to subsequent DENV infection (Supplementary Fig. 6 for gating strategy). No differences were detected in total B cells between groups following DENV infection compared to baseline levels ( Supplementary Fig. 9a), but ZIKV-immune groups had elevated levels of activated B cells (CD20 + CD3 -CD69 + ) since baseline and a trend to increase these levels more than the naïve group over time ( Supplementary Fig. 9b). We detected a significant decrease of proliferating B cells (CD20 + CD3 -Ki67 + ) in naïve animals at 7 and 10 dpi (p=0.0031, p=0.0345; . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint Two-way Anova Dunnett test), while ZIKV-immune groups retained their proliferating levels ( Supplementary Fig. 9c). Interestingly, the ZIKVPF-10mo group showed a significant increase of B cells that were proliferating and activated simultaneously (CD20 + CD3 -CD69 + Ki67 + ) as early as in day 1 pi (p=0.0240; Two-way Anova Dunnett test) and maintained higher levels up to 10 dpi Previous studies have demonstrated that DENV and ZIKV specific CD4 + and CD8 + T cells are enriched in certain memory subsets 24,48 . Thus, we measured whether the early activation of T cell subpopulations, such as effector memory (CD3 + CD4 + CD28 -CD95 + ) and central memory (CD3 + CD4 + CD28 + CD95 + ) T cells (T-EM and T-CM), within each T cell compartment was modulated following DENV infection in presence or absence of convalescence to ZIKV (Fig. 6).
The ZIKVPF-10mo group showed significant higher frequency of activated CD4 + and CD8 + T-EM (CD3 + CD4 + CD28 -CD95 + CD69 + and CD3 + CD8 + CD28 -CD95 + CD69 + ) following DENV infection compared to basal levels (CD4 + T-EM at 7 and 10 dpi: p=0.0001, p=0.0072; CD8 + T-EM at 2 and 7 dpi: p=0.0291, p=0.0001; Two-way Anova Dunnett test) (Fig. 6a, d). Interestingly, the ZIKVPR-2mo group showed a very limited frequency and activation of the CD4 + and CD8 + T-EM compared to the ZIKVPF-10mo and naïve groups. However, this group with an early convalescent period to ZIKV, contrary to the other two groups, showed a very limited but significant activation of CD8 + T-CM (CD3 + CD8 + CD28 + CD95 + CD69 + ) at day 7 and 10 pi (p=0.0007, p=0.0147; Two-way Anova Dunnett test). (Fig. 6e).In contrast, naïve animals did not show any significant activation of these cell subsets after DENV infection. Collectively, the B and T cell results suggest that following DENV infection: (i) animals with a mid-convalescence ZIKV immunity have a more dynamic B cell response and are able to rapidly produce more activated effector memory T cells from both T cell compartments; (ii) animals with an early-convalescence to ZIKV induced activation of central memory T cells in the CD8 + compartment with a very limited T-EM frequency and activation profile compatible with a contraction phase of the T cells compartments; (iii) and animals without previous exposure to ZIKV exhibited a limited B cell response and minimal modulation of T cell subpopulations at early timepoints as the ZIKV-immune groups. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint T cell polyfunctional effector response against DENV and ZIKV is shaped by the longevity of ZIKV-immunity. To further characterize the cross-reactive T cell response we investigated if different convalescent periods of ZIKV immunity impacted the outcome of the effector role of CD4 + and CD8 + T cells following DENV infection. PBMCs were isolated and stimulated with peptide pools from DENV and ZIKV envelope (E) proteins and from ZIKV non-structural proteins (ZIKV-NS) (Supplementary Table 5 for peptide sequences). Then, intracellular cytokine staining using flow cytometry analysis ( Supplementary Fig. 11 for gating strategy; Supplementary Table 4 for Ab panel) was performed to quantify the production of effector immune markers such as the cytotoxic marker CD107a, IFN-γ, and TNF-α by CD4 + and CD8 + T cells at baseline, 30, 60, and 90 days after DENV infection (Fig. 7).
After DENV infection, we were able to determine the modulation of the ZIKV-primed effector CD4 + and CD8 + T cell responses of ZIKV-immune groups and the de novo response of ZIKV-naïve animals. The ZIKVPF-10mo and naïve groups significantly boosted their CD107a expression in both T cell compartments stimulated mainly by DENV E protein at 30  Anova Tukey test) (Fig. 7b, c, n, o, p). Also, these groups boosted the CD107a cytotoxic signature reacting against ZIKV E and NS proteins by cross-reactive CD4 + T cells 30 days after DENV infection (ZIKVPF-10mo vs ZIKVPR-2mo: p=0.0025 for ZIKV E, p<0.0001 for ZIKV NS; Naïve vs ZIKVPR-2mo: p=0.0025 for ZIKV E, p=0.0002 for ZIKV NS; Two-way Anova Tukey test) (Fig. 7b).
Although all groups showed a boosted TNF-α effector response in the CD8 + T cell compartment up to 90 days after DENV infection, no significant differences between groups were observed.
Collectively, these results after DENV infection suggest that a mid-convalescence to ZIKV translate in a more complete polyfunctional T cell response characterized by: (i) a cytotoxic CD107a + phenotype directed to DENV E protein for both T cell compartments comparable to the DENV-specific de novo response of the naïve group, (ii) developed CD107a, IFN-γ and TNF-α producing CD8 + T cell effector response that cross-react efficiently with DENV E protein since baseline and is boosted after DENV infection, (iii) and promoted the higher T cell effector response against ZIKV NS proteins. An early-convalescence to ZIKV results in (iv) a very limited cytotoxic activity (limited expression of CD107a marker) which is in line with a very limited activation of the T-EM observed and with failed capability to react efficiently against E or NS proteins. The ZIKVnaïve group response was characterized by: (v) production of a DENV-specific de novo functional T cell response with similar magnitude between both T cell compartments, (vi) capable to crossreact against ZIKV E and NS proteins, (vii) and able to mount a DENV-specific cytotoxic CD107a + phenotype.
. CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint

Discussion
We found that previous infection to ZIKV improved the immune response against subsequent DENV infection without an enhancement of DENV viremia nor pro-inflammatory status, and that this improvement relies in the longevity of ZIKV convalescence-more efficient after longer ZIKV pre-exposure. This scenario seems to be independently of the ZIKV preexposure strain when those strains belong to the same phylogenetic group. The results presented herein provide insights on the anticipated immune modulation by previous ZIKV immunity in the expected re-emergence of DENV in the same geographical regions. The aftermath of the recent ZIKV epidemic has been related to a remarkable decrease in DENV cases in Brazil 28 , and also  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint within genotypes. Another possible explanation is the genetic heterogeneity of rhesus macaques used in these two studies as they are derived from different breeders. The importance of selecting genetic well-characterized macaques have been discussed previously 49   The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint DENV cross-neutralization prior DENV infection. The highly conserved identity of E protein between ZIKV and DENV suggests the development of cross-binding/cross-neutralizing functional Abs against both viruses 56 . However, there is no data yet that delineates shared crossneutralizing epitopes between ZIKV and DENV-2/-4, but it is known that DENV-4 genotypic diversity impact the capacity of its neutralization 57 .
One factor that plays a critical role in the induction of enhancement and disease severity is the time elapsed between sequential heterologous DENV infections 10  However, they found similar increased DENV severity in pups from ZIKV early-and lateconvalescent mothers. This is in contrast with our results and human findings in DENV sequential infections where the window of cross-protection and immune memory maturation was influenced by the longevity of convalescence 40,41,54,58 . Of relevance, in contrast to DENV sequential infections-suggesting cross-neutralizing protection by short time intervals-our results by ZIKV-DENV sequential infections reveal that longer convalescence to ZIKV induce higher boost of cross-neutralizing response with potentially less risk of DENV pathogenesis than short period of convalescence. However studies using NHP cohorts with longer period between ZIKV and DENV infections and prospective human cohorts, when available, are necessary to provide more conclusive answers.
Early studies of T cells associate their contribution towards immunopathogenesis in DENV secondary infections explained by the original antigenic sin 59 , but increasing evidence suggest their protective role during primary and secondary DENV infections 60 . Recently, with the introduction of ZIKV into The Americas, T cells from DENV immunity are being implicated in mediating cross-protection against ZIKV [22][23][24] . However, the role and kinetics of T cells from ZIKV immunity in a subsequent DENV infection remains unknown. We found that animals with a midconvalescence to ZIKV developed an early activation of CD4 + and CD8 + effector memory T cells after DENV infection. This early activation has been observed for the opposite scenario in DENVimmune ZIKV-infected patients 24 . Interestingly, the ZIKV early-convalescent group display a . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint modest activation (T-CM>T-EM) early after DENV infection. Since this group was infected with ZIKV only two months before DENV it is possible that after viral clearance and development of ZIKV-specific T cell response, the T cell compartments were still under the contraction phase at the time of the DENV challenge. Yellow fever virus (YFV) and vaccinia virus vaccinations in humans demonstrate that T cell contraction start as early as approximately one-month postvaccination and at least for almost three months is still ongoing 61 . Also, a study shows that restimulation using alphavirus replicons during T cell response contraction does not have significant impact modulating the pre-existing T cell response 62 .
The profile of ZIKV-specific CD8 + T cells in humans with convalescence to ZIKV is characterized by the production of IFN-γ, and expression of activation and cytotoxic markers 63 .
Presence of sustained levels of IFN-γ prior and early after DENV challenge in vaccinees has been associated with protection against viremia and/or severe disease 64,65 . We observed a similar phenotype of the polyfunctional response of CD8 + T cells prior DENV infection in animals with longer convalescence to ZIKV. Strikingly, this response recognize more efficiently peptides from DENV E protein than from ZIKV E protein. However, ZIKV-specific CD8 + T cells direct 57% of their response against structural proteins, which may suggest these cells can recognize conserved epitopes between ZIKV and DENV structural proteins. Cross-reactivity of T cells between heterologous flavivirus infections is explained by selective immune recall of memory T cells that recognize conserved epitopes between DENV and ZIKV 24 , which also has previously been demonstrated during secondary heterotypic DENV infections 66,67 . In addition an increased cytotoxic profile as demonstrated by the higher frequency of CD107a-expressing CD4 + and CD8 + T cells in the ZIKV mid-convalescent group correlates with the synchronously early activation of CD4 + and CD8 + effector memory T cells and elevated levels of perforin release. Since CD8 + T cells are known for their role in early clearance of DENV in heterologous secondary infections 68 , it is possible that the T cell cross-reactivity with DENV E detected in the ZIKV mid-convalescent group at baseline together with the basal levels of cross-neutralizing Abs against DENV-2 before DENV infection, provide a synergistic and partial cross-protective effect.
Higher proportion of IFN-γ and TNF-α producing T cells before a secondary heterologous DENV infection has been associated to a subsequent subclinical outcome 69 . Herein, we observed that the ZIKV mid-convalescent group had elevated levels of IFN-γ and TNF-α producing T cells since baseline. In this group, DENV infection stimulated a higher frequency of these cells, but remarkably, also increased highly cross-reactive IFN-γ-producing CD4 + T cells directed to DENV E, and ZIKV E/NS proteins. A study showed that cross-reactive ZIKV-primed CD4 + T cells recognized conserved homologous sequences of other related flaviviruses such as West Nile The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint virus (WNV), YFV, and of relevance for our study, cross-react with E protein epitopes of all DENV serotypes 70

. Moreover, IFN-γ-producing CD4 + T cells have a role in providing help to B cells while B cells present DENV antigens to CD4 + T cells to produce IFN-γ and other immune mediators
that induced B cell activation and subsequent efficient Ab production 71 . Memory CD4 + T cells are also required to generate an effective humoral response against ZIKV 72 73 . On the other hand, we showed that naïve animals with DENV de novo response did not cross-neutralized ZIKV at all, which state that although similar, antigenic differences are sufficient to mount predominantly type-specific rather than crossreactive responses during a primary infection 50,55 .
A lack of ZIKV immunity promoted a more pro-inflammatory profile characterized by significant elevated levels of IFN-α, IL-6 and MIG/CXCL9. These are part of the cytokine storm produced by DENV infection and correlate with DENV severe disease. IFN-α is known to be actively produced by human pDCs-increasing frequency of pDCs in naïve animals-during acute DENV infection in vitro and in vivo 74 . Elevated levels have been correlated with severity in DHF patients, and to act as a marker for elevated DENV replication 75,76 . IL-6 has been detected in high levels during secondary DENV infections in children 77 , and the day patients suffer from shock (DSS) 78 or died from DHF 79 . MIG/CXCL9 is known to be a risk factor for DENV severity involved in vascular permeability 80  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint producing CD8 + T cells, required to control DENV infection in vivo 82,83 . Levels of CXCL10 are amplified in presence of DENV-specific T cells 84 . This correlated with higher proportion and activation of both T cell compartments and subsequent polyfunctional T cell response against DENV-E-specific peptides in the group with longer convalescence to ZIKV. Perforin is involved in the cytotoxic degranulation process against virus-infected cells. In DENV infection, perforin is part of the anti-DENV cytotoxic phenotype of CD8 + and CD4 + T cells 48,85 . Perforin levels were significantly elevated only in the ZIKV mid-convalescent group after DENV infection. Accordingly, this coincide with a significant activation of CD8 + and CD4 + effector memory T cells, and degranulation functional response of both T cell compartments, suggesting an enhanced perforinproducing cytotoxic role of T cells in presence of longer convalescence to ZIKV.
In summary, dissecting our main findings per previous ZIKV-immune status we found that a ZIKV middle-convalescence: (i) results in shorter DENV viremic period, (ii) lowest proinflammatory status with upregulation of cellular immune response mediators, (iii) robust neutralizing antibody response higher in magnitude and durability against ZIKV strains and DENV serotypes, (iv) elevated activated and proliferating B cells, (v) early activation of cross-reactive CD4 + and CD8 + effector memory T cells, (v) and a major breadth of polyfunctional T cell response. This proof-of-concept and other prospective studies of ZIKV/DENV pathogenesis and crossimmune relationships are urgently needed even as the peak of the ZIKV epidemic has passed as there is a high probability for ZIKV to establish a sylvatic transmission cycle using neotropical primates and mosquitoes in the Americas that will sustain ZIKV circulation and potential reemergence 86,87 . Our data show a positive scenario that supports the implementation of ZIKV vaccine programs, since it suggests that a vaccine-acquired ZIKV-immunity will not worsen DENV . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint pathogenesis and may ameliorate immune response against a subsequent infection with DENV.
Similarly, the implementation of DENV vaccines is also supported in the context of previous ZIKV immunity, since ZIKV convalescence may boost the vaccine-acquired anamnestic immune response to DENV without predisposing to an enhanced pathogenesis. However, the selection of the vaccine schedule may be critical to induce the optimal immune response when more than one dosis is planned. Figure 1 | Experimental design for DENV-2 challenge of ZIKV-immune and naïve rhesus macaques. 14 young adult male rhesus macaques (Macaca mulatta), matched in age and weight, were divided in three cohorts. Cohort 1 (ZIKVPF-10mo, n=4, blue): composed of four animals (5K6, CB52, 2K2, and 6N1) that were inoculated with 1x10 6 pfu/500 ul of the ZIKV H/PF/2013 strain subcutaneously 10 months before (mid-convalescence) DENV-2 challenge. Cohort 2 (ZIKVPR-2mo, n=6, orange): composed of six animals (MA067, MA068, BZ34, MA141, MA143, and MA085) that were inoculated with 1x10 6 pfu/500 ul of the contemporary ZIKV PRVABC59 strain two months before (early-convalescence) DENV-2 challenge. Both ZIKV strains used for previous exposure of these groups are >99.99% comparable in amino acid identity (Supplementary Table 1). Cohort 3 (Naïve, n=4, black): composed of four ZIKV/DENV naïve animals (MA123, MA023, MA029, and MA062) as a control group. Prior to DENV-2 challenge all animals were subjected to quarantine period. All cohorts challenged subcutaneously (deltoid area) with 5x10 5 pfu/500 ul of DENV-2 New Guinea 44 strain (NGC44). After DENV-2 challenge all animals were extensively monitored for evidence of disease and clinical status by vital signs such as external temperature (°C), weight (Kg), CBC and CMP panels at the Caribbean Primate Research Center (CPRC). Blood samples were collected longitudinally at baseline, 1 to 10, 15, 30, 60 and 90 days after DENV infection (gray arrows). In all timepoints the blood samples were used for serum separation (yellow), and for PBMCs isolation (red) at baseline, 1, 2, 3, 7, 10, 15, 30, 60, and 90 days after DENV infection. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint

Figure 2 | Previous ZIKV immunity does not contribute to an increase of DENV RNaemia.
(a) DENV-2 RNA kinetics in ZIKV-immune and naïve animals at baseline, sequentially from day 1 to day 10, and day 15 after DENV infection. Genome copies (Log10) per ml of serum were measured by qRT-PCR. Symbols represent individual animals per cohort: blue squares (ZIKVPF-10mo), orange squares (ZIKVPR-2mo) and black circles (Naïve). Lines indicate the mean number of genome copies detected for each cohort over time. Statistically significant differences between groups were determined using Two-Way Anova (Tukey's multiple comparisons test). (b) Mean of days that DENV-2 RNAemia was detected per cohort. Bars represent mean values for ZIKVPF-10mo (blue), ZIKVPR-2mo (orange) and Naïve (black). The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint between groups were calculated using Two-Way Anova Tukey's multiple comparisons test. Significant multiplicity adjusted p values (* ˂0.05, ** ˂0.01, *** ˂0.001, **** ˂0.0001) are shown colored representing the cohort against that particular point where is a statistically significant difference between groups.
. CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint were assigned with one-half of the limit of detection for graphs visualization (1:10). Statistically significant differences between groups were calculated using Two-Way Anova Tukey's multiple comparisons test with 95% CI. Significant multiplicity adjusted p values (* ˂0.05, ** ˂0.01, *** ˂0.001, **** ˂0.0001) are shown. Blue and orange asterisks represent significant difference between the corresponded ZIKV immune groups and naive group, and gray asterisks indicate a significant difference between ZIKV immune groups.
. CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint and 90 days after DENV infection. Comparison of NAb titers between pre-infecting and heterologous ZIKV strains was performed (c) before and (d) after DENV infection. Symbols connected with full lines indicate mean levels of NAb titers detected per cohort over time: blue squares (ZIKVPF-10mo), orange squares (ZIKVPR-2mo) and black circles (Naïve). Error bars represent the standard error of the mean (SEM). PRNT60: NAb titer capable of reduce 60% or more of ZIKV strains plaque-forming units (pfu) compared with the mock (control of virus without serum). A PRNT60 1:20 titer was considered positive, and <1:20 as a negative Neut titer. Dotted line mark <1:20 for negative results. Non-neutralizing titers (<1:20) were assigned with one-half of the limit of detection for graphs visualization (1:10). Statistically significant differences between groups were calculated using Two-Way Anova Tukey's multiple comparisons test with 95% CI. Significant multiplicity adjusted p values (* ˂0.05, ** ˂0.01, *** ˂0.001, **** ˂0.0001) are shown. Blue and orange asterisks represent significant difference between the corresponded ZIKVimmune groups and naive group, and gray asterisks indicate a significant difference between ZIKV-immune groups. Half maximal (50%) effective concentrations (EC50) were calculated based on the % of neutralization of ZIKV for all timepoints. The EC50 was defined as the Ab dilution factor that is successfully capable of block 50% of the viral infection in the PRNT assay. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint  Fig. 6 for gating strategy). Blue, orange and black squares represent T-EM for ZIKVPF-10mo, ZIKVPR-2mo and Naïve, respectively. Gray squares represent T-CM for each group. Short black lines mark mean value for each group per timepoint. Cutted line divide % of T-EM and T-CM cells quantified before and after DENV infection. Statistically significant differences between groups were determined using Two-Way Anova Dunnett's multiple comparisons test (comparison of each cohort response at each timepoint versus baseline) and reported as multiplicity adjusted p values (* ˂0.05, ** ˂0.01, *** ˂0.001, **** ˂0.0001) with 95% CI. Asterisks represent significant difference between the corresponded timepoint and baseline within the same group.
. CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint Figure 7 | Longevity of ZIKV immunity shapes the polyfunctional response of CD4 + and CD8 + T cells. T cell polyfunctional effector response was determined by the quantification (%) of (a-d; m-p) CD107a-expressing and (e-h; q-t) IFN-γ or (i-l; u-x) TNF-α producing CD4 + and CD8 + T cells before (0) and 30, 60 and 90 days after DENV infection. Responses to several peptide pools that encode for DENV and ZIKV envelope (E) proteins or ZIKV non-structural (NS) proteins were quantified. After antigenic stimulation intracellular cytokine staining was performed using flow cytometry analysis ( Supplementary Fig. 11 for gating strategy). Individual symbols represent each animal per antigenic stimulation over time: blue squares (ZIKVPF-10mo), orange squares . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint (ZIKVPR-2mo) and black circles (Naïve). Short gray lines mark mean value for each group. Statistically significant differences between groups were calculated using Two-Way Anova Tukey's multiple comparisons test with 95% CI. Significant multiplicity adjusted p values (* ˂0.05, ** ˂0.01, *** ˂0.001, **** ˂0.0001) are shown. Asterisks represent significant difference between indicated groups. and also naïve cohorts were available as well. After our laboratories prioritized ZIKV research since 2016, DENV pre-exposed and naïve cohorts were infected with ZIKV and pre-exposed animals became available for this study. All animals were housed within the Animal Resources  For the PRNT, serum samples were inactivated, diluted (2-fold), mixed with a constant inoculum of virus (volume necessary to produce ~35 pfu/well) and then incubated for 1 hr at 37°C and 5% CO2. After incubation, virus-serum mix dilutions were added to Vero-81 cells monolayer in flat bottom 24-well plates seeded the day before for 1 hr at 37°C and 5% CO2, finally overlay medium was added and incubated by several days (serotype dependent). Results were reported as PRNT60 titers, NAb titer capable of reduce 60% or more of DENV serotypes or ZIKV strains pfu compared with the mock (control of virus without serum). A PRNT60 1:20 titer was considered a positive Neut titer, and <1:20 as a negative Neut titer. Non-neutralizing titers (<1:20) were assigned with one-half of the limit of detection for graphs visualization. In addition, the half maximal (50%) effective concentrations (EC50) were calculated based on the % of neutralization of DENV-2 and ZIKV infection for all timepoints. The EC50 was defined as the Ab dilution factor that is successfully capable of block 50% of the viral infection in the PRNT assay. For this, the data was transformed into Log10, non-linear regression (curve fit) analysis was performed and calculated by the sigmoidal dose response (variable slope) equation in Prism 7. Reported values were validated by a R squared < 0.75 and Hill Slope absolute value < 0.5, and the geometric mean with 95% confidence interval was also calculated. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint Multiplex cytokine profile analysis. A total of 8 cytokines/chemokines were measured (pg /ml -1 ) by Luminex at baseline, 1, 2, 3, 5, 10, 15 and 30 dpi, including: interferon alpha (IFN-α), interleukin-6 (IL-6), monokine induced by IFN-gamma (MIG/CXCL9), monocyte chemoattractant protein 1 (MCP-1/CCL2), macrophage inflammatory protein 1-beta (MIP-1β/CCL4), IL-1 receptor antagonist (IL-1RA), C-X-C motif chemokine 10 (CXCL10/IP-10) and perforin. The multiplex assay was conducted as previously described 17,88 .  Table 4 for Ab panel) 17,89 . Polyfunctional effector response of CD4 + and CD8 + T cells was measured before and after DENV infection. Antigen-. CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint specific CD4 + and CD8 + T cell effector responses were measured at baseline to determine basal levels in presence (ZIKVPF-10mo, ZIKVPR-2mo) or absence (Naïve) of previous immunity to ZIKV. Also, 30, 60 and 90 dpi were assessed to determine how this pre-existing functional response is modulated after DENV infection and if is maintained over time. PBMCs were thawed one day before stimulation. For peptide pools stimulation, PBMCs were stimulated for 6 hr at 37°C and 5% CO2. The peptides used for DENV-E, ZIKV-E and ZIKV-NS were 15-mers overlapped by 10 amino acids at 1.25 ug/ml -1 , 2.5 ug/ml -1 , 475 ng/ml -1 per peptide, respectively. PBMCs were ex vivo stimulated with DENV and ZIKV supernatants of infection, envelope (E) proteins, and ZIKV-NS peptide pools (Supplementary Table 5 for peptide sequences). The stimulation with peptides was performed in presence of brefeldin A at 10 ug/ml -1 . After stimulation, the cells were stained for the following markers: CD3, CD4, CD8, CD20 (excluded), CD107a (functional cytotoxicity).

Methods
Levels of IFN-γ and TNF-α also were measured in gated lymphocytes cell populations. Samples were measured and data was collected on a LSRII (BD). The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint orange squares (ZIKVPR-2mo) and black circles (Naïve). Lines connect mean values detected over time. Error bars indicate the standard error of the mean (SEM) for each cohort per timepoint. Statistically significant differences between groups were determined using Two-Way Anova Tukey's multiple comparisons test. For differences in ALT levels Two-Way Anova Dunnett's multiple comparisons test (comparison of each cohort response at each timepoint versus baseline) was performed due to levels divergence between cohorts since baseline. Statistically differences are reported as multiplicity adjusted p values (* ˂0.05) and 95% CI. Results were read at OD 450, 405 or using ISR (Immune Status Ratio) following manufacturer's instructions. Statistically significant differences between groups were calculated using Two-Way Anova Tukey's multiple comparisons test with 95% CI. Significant multiplicity adjusted p values (* ˂0.05, ** ˂0.01, *** ˂0.001, **** ˂0.0001) are shown. Blue and orange asterisks represent significant difference between the corresponded ZIKV immune groups and naive group, and gray asterisks indicate a significant difference between ZIKV immune groups. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint (* ˂0.05, ** ˂0.01, *** ˂0.001, **** ˂0.0001) with 95% CI. Asterisks represent significant difference between the corresponded timepoint and baseline within the same group. CI. Asterisks represent significant difference between the corresponded timepoint and baseline within the same group. ND (Not Done) in panels 8o and 8s refers that for ZIKVPF-10mo and Naïve the NKp30 + NKp46 + and NKp46 + NKp30 + subpopulations were not measured. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/621094 doi: bioRxiv preprint compartments (% of total T cells) frequencies and their activation (d-f) were quantified at baseline and following DENV infection up to 30 dpi by immunophenotyping using flow cytometry. Symbols represent individual animals per group for each timepoint: blue squares (ZIKVPF-10mo), orange squares (ZIKVPR-2mo) and black circles (Naïve). Short gray lines mark mean value of T cells percent in each cohort per timepoint. Cutted line divide % of T cells quantified before and after DENV infection. Statistically significant differences between groups were determined using Two-Way Anova Dunnett's multiple comparisons test (comparison of each cohort response at each timepoint versus baseline).

Legends for Supplementary Tables
Supplementary Table 1

Competing interests:
The authors declare no competing financial interests.