Prior dengue immunity enhances Zika virus infection of the maternal-fetal interface in rhesus macaques

Concerns have arisen that pre-existing immunity to dengue virus (DENV) could enhance Zika virus (ZIKV) disease, due to the homology between ZIKV and DENV and the observation of antibody-dependent enhancement (ADE) among DENV serotypes. To date, no study has examined the impact of pre-existing DENV immunity on ZIKV pathogenesis during pregnancy in a translational non-human primate model. Here we show that prior DENV-2 exposure enhanced ZIKV infection of maternal-fetal interface tissues in macaques. However, pre-existing DENV immunity had no detectable impact on ZIKV replication kinetics in maternal plasma, and all pregnancies progressed to term without adverse outcomes or gross fetal abnormalities detectable at delivery. Understanding the risks of ADE to pregnant women worldwide is critical as vaccines against DENV and ZIKV are developed and licensed and as DENV and ZIKV continue to circulate.


INTRODUCTION
NHP development and placentation resemble those of humans more closely than these 97 processes do in other animal models, making NHPs particularly relevant to understanding 98 viral infections in pregnancy (46). Here we apply our established NHP model (47) to assess 99 the impact of DENV immunity on ZIKV pathogenesis in pregnancy. We do not detect a role 100 for DENV immunity in modulating fetal outcomes in ZIKV-infected pregnant macaques. 101 However, previous exposure to DENV did appear to increase ZIKV infection in tissues of the 102 maternal-fetal interface, a result that warrants further examination. DENV-2 (orange). Approximately 1-3 months following DENV challenge, the eight DENV exposed macaques 106 were bred, became pregnant, and were challenged with 10 4 PFU ZIKV-PRVABC59, an Asian-lineage ZIKV 107 isolate, on gestational day 45. A cohort of four pregnant, DENV-naïve macaques (blue) were challenged with 108 ZIKV-PRVABC59 on gestational day 45. A control cohort of four macaques (green) were mock-challenged with 109 PBS on gestational day 45. All three cohorts underwent the same experimental protocols for blood collection 110 and sedation for ultrasound. At approximately gestational day 160, infants were delivered via cesarean section, 111 and a set of maternal-fetal interface tissues with maternal biopsies were collected. Infants were placed with 112 their mothers for long-term behavioral analysis, data from which is part of a separate study. 113 114

115
Prior DENV immunity does not modulate ZIKV replication kinetics in plasma 116 To characterize the range of pathogenic outcomes of congenital ZIKV infection in DENV-117 immune animals, we subcutaneously (s.c.) inoculated a cohort of eight non-pregnant, 118 Indian-origin rhesus macaques with 10 4 PFU of DENV-2/US/BID-V594/2006, a low-passage 119 human isolate from Puerto Rico (Fig. 1). All eight macaques were productively infected with 120 DENV-2, with peak plasma viral loads ranging from 10 5 -10 7 vRNA copies/mL occurring on 121 days 2-3 post-infection (Fig. 2). Following a biphasic decline in viral loads, all macaques 122 cleared infection by day 11 post-infection.  Following challenge, all three cohorts (DENV-immune, DENV-naïve, and mock) underwent 137 the same blood sampling and fetal monitoring protocols (Fig. 1). All macaques inoculated 138 with ZIKV were productively infected. Peak plasma viremia occurred on days 2-4 post-139 infection, with titers ranging from 10 4 -10 5 vRNA copies/mL in DENV-immune animals and 140 10 3 -10 5 vRNA copies/mL in DENV-naïve animals (Fig. 3A, 3B). An unpaired t-test did not 141 reveal significant differences between cohorts in the peak, area under the curve, or duration 142 of viremia (Fig. 3C). Since prolonged ZIKV viremia >21 days is only observed in pregnancy, 143 we assessed differences in duration both as a continuous variable and as a binary with 144 viremia greater than or less than 21 days. This suggests that prior DENV-2 immunity did not 145 alter ZIKV replication kinetics during gestation. were challenged with 10 4 PFU of ZIKV-PRVABC59 at gestation day 45, which is late in the first trimester. Viral 150 loads were assessed from plasma samples with ZIKV-specific QRT-PCR. All values above the limit of 151 quantification (100 copies vRNA, mL plasma) are shown above. C. Graphs of the values for the peak, duration, 152 and area under the curve of viremia for both DENV-immune and DENV-naïve macaques. An unpaired t-test 153 was used for statistical comparison; ns = not significant (p > 0.05). Only values above the limit of quantification 154 were used in statistical analyses. For DENV-1-4, specific antibody titer ranges have been shown to enhance viral replication.

159
As measured by a DENV inhibition ELISA (iELISA) assay, an intermediate antibody titer 160 range of 1:21-1:80 was associated with a greater risk of developing severe dengue disease 161 upon secondary exposure in a human cohort study (2). In a separate human cohort study, a 162 plaque reduction neutralization test (PRNT50) titer of <1:100 was associated with an 163 increased risk of severe DENV disease upon secondary exposure (48). In order to assess 164 how cross-reactive DENV antibodies impact ZIKV outcomes during pregnancy, we     in DENV-immune macaques than DENV-naïve animals 28-35 days after ZIKV challenge, but 210 no significant differences were noted in PRNT90 titers between groups. Together, these 211 data provide evidence that antibodies capable of cross-reacting with ZIKV were present at in blue. Samples labeled "ND" were not detected. Using an unpaired t-test, PRNT50, but not PRNT90, titers 220 from the DENV-immune group were significantly higher than the PRNT50 titer of DENV-naïve animals at 28 221 days post-ZIKV-challenge (**p<0.01). Neutralization curves can be found in Supplementary Fig. 1. diameter of the mock-infected cohort was significantly lower than the normative data 236 (p=0.01713). There were no significant differences noted in pairwise comparisons of growth 237 trajectories between groups. Taken together, these extensive fetal growth measurements 238 suggest that there was no significant reduction in fetal growth in ZIKV-exposed macaques, 239 regardless of their DENV immune history.  Table 1). Fetal tissues are not available for virological analysis because 248 infants were placed with their mothers for long-term behavioral analysis, data from which 249 will be part of a forthcoming study. We collected fetal plasma, umbilical cord plasma, and  Table 1). There was no robust evidence to support direct infection of  fetal health and perform four measurements of fetal growth: biparietal diameter and head circumference to 257 evaluate head size; abdominal circumference and femur length to evaluate overall fetal growth. Using 258 normative data from the California National Primate Research Center, a z-score was calculated for each 259 measurement and the change in z-score from baseline is plotted for each measurement with an open circle. 260 The overall growth trajectory for each group was quantified by calculating the regression slope parameters 261 from baseline (solid line). When compared to the normative data, mock-infected animals had significantly 262 reduced biparietal diameter growth (p=0.01713). No other significant differences were detected in 263 comparisons to the normative data or in comparisons between the experimental groups. 264 265 Enhanced infection of the maternal-fetal interface in DENV-immune macaques 266 We performed an extensive dissection of both discs of the placenta in order to understand 267 the distribution of ZIKV in placental tissues. Positive tissue samples were detected above Abdominal Circumference (z-score) To determine whether the presence of vRNA in the MFI was associated with duration, peak, immune cohort (Fig. 6C). There was a significant correlation between area under the curve 287 and presence of vRNA in the MFI in both cohorts (Fig. 6D, 6E). There was a significant   also present in mock-infected animals. There were no statistically significant differences 324 between any of the groups for any of these pathologic features or placental weight. This  We did observe an association between prolonged viremia, defined as lasting >21 days, and tempting to speculate that prior DENV immunity may lead to longer viral replication and 357 therefore greater ZIKV burden in the placenta. However, since we did not observe any 358 statistically significant differences in the duration of viremia between the two groups, perhaps due to a small sample size, we cannot make any definitive conclusions about the 360 impact of prior DENV immunity on the duration of ZIKV viremia. serotypes has an effect on subsequent enhancement or protection (for review see (20)).

381
There is also considerable evidence that the pre-existing antibody titer at the time of    Tissues were dissected as previously described (47) using sterile instruments that were 590 changed between each organ and tissue type to minimize possible cross contamination.

591
Each organ/tissue was evaluated grossly, dissected with sterile instruments in a sterile 592 culture dish, and sampled for histology, viral burden assay, and/or banked for future assays.

593
A comprehensive listing of all specific tissues collected and analyzed is presented in Fig. 6A 594 (maternal-fetal interface tissues) and Supplementary