Trade-offs Shaping Transmission of Sylvatic Dengue and Zika Viruses in Monkey Hosts PDF

Article https://doi.org/10.1038/s41467-024-46810-x Trade-offs shaping transmission of sylvatic dengue and Zika viruses in monkey ho...

Article https://doi.org/10.1038/s41467-024-46810-x Trade-offs shaping transmission of sylvatic dengue and Zika viruses in monkey hosts Received: 30 June 2023 Kathryn A. Hanley 1,8 , Hélène Cecilia 1,8, Sasha R. Azar 2,3, Brett A. Moehn 1, Jordan T. Gass 1, Natalia I. Oliveira da Silva 2, Wanqin Yu 1 , Accepted: 8 March 2024 Ruimei Yun 2, Benjamin M. Althouse1,4, Nikos Vasilakis 2,5,6 & Shannan L. Rossi 2,5,6,7 Check for updates Mosquito-borne dengue (DENV) and Zika (ZIKV) viruses originated in Old World sylvatic (forest) cycles involving monkeys and canopy-living Aedes 1234567890():,; 1234567890():,; mosquitoes. Both viruses spilled over into human transmission and were translocated to the Americas, opening a path for spillback into Neotropical sylvatic cycles. Studies of the trade-offs that shape within-host dynamics and transmission of these viruses are lacking, hampering efforts to predict spil- lover and spillback. We infected a native, Asian host species (cynomolgus macaque) and a novel, American host species (squirrel monkey) with sylvatic strains of DENV-2 or ZIKV via mosquito bite. We then monitored aspects of viral replication (viremia), innate and adaptive immune response (natural killer (NK) cells and neutralizing antibodies, respectively), and transmission to mosquitoes. In both hosts, ZIKV reached high titers that translated into high transmission to mosquitoes; in contrast DENV-2 replicated to low levels and, unexpectedly, transmission occurred only when serum viremia was below or near the limit of detection. Our data reveal evidence of an immunologically- mediated trade-off between duration and magnitude of virus replication, as higher peak ZIKV titers are associated with shorter durations of viremia, and higher NK cell levels are associated with lower peak ZIKV titers and lower anti- DENV-2 antibody levels. Furthermore, patterns of transmission of each virus from a Neotropical monkey suggest that ZIKV has greater potential than DENV- 2 to establish a sylvatic transmission cycle in the Americas. Spillover of zoonotic, arthropod-borne viruses (arboviruses) into translocation is not new: yellow fever virus (YFV), which originated in a humans is accelerating, with outcomes ranging from dead-end infec- sylvatic cycle in the forests of Africa involving non-human primates tions to pandemics1. Because of the unparalleled mobility of humans, (NHPs) and arboreal mosquitoes, was carried via sailing ships to the these viruses are being transported across the world, creating the neotropics centuries ago3,4. Soon after, YFV established urban trans- potential for spillback into novel wildlife reservoirs2. Such mission in the Americas and spilled back into a sylvatic cycle 1 Department of Biology, New Mexico State University, Las Cruces, NM 88003, USA. 2Department of Pathology, University of Texas Medical Branch, Galveston, TX 77555, USA. 3Center for Tissue Engineering, Department of Surgery, Houston Methodist Research Institute, Houston Methodist Hospital, Houston, TX 77030, USA. 4Information School, University of Washington, Seattle, WA 98105, USA. 5Center for Vector-Borne and Zoonotic Diseases, University of Texas Medical Branch, Galveston, TX 77555, USA. 6Institute for Human Infection and Immunity, University of Texas Medical Branch, Galveston, TX 77555, USA. 7 Department of Microbiology and Immunology, University of Texas Medical Branch, Galveston, TX 77555, USA. 8These authors contributed equally: Kathryn A. Hanley, Hélène Cecilia. e-mail: [email protected] Nature Communications | (2024)15:2682 1 Article https://doi.org/10.1038/s41467-024-46810-x maintained in neotropical primates and mosquitoes that plagues Results South America to present day3,4. The four serotypes of dengue virus Overview of experimental design (DENV-1-4) underwent a similar journey, albeit their ancestral sylvatic The putative native host in this study was the cynomolgus macaque, cycles occur in Asia, but compelling evidence of a sylvatic cycle of Macaca fascicularis. Rudnick and colleagues isolated sylvatic DENV DENV in the Americas has never been detected despite being actively from sentinel cynomolgus macaques held in the forest canopy in sought5–7. Thus, when ZIKV was detected in the Americas in 2015, there Malaysia and also showed that ≥ 50% of wild, forest-living cynomolgus was great concern, but equally great uncertainty, about its potential to macaques in Malaysia were seropositive for at least one serotype of establish a neotropical sylvatic cycle8–10. While multiple instances of DENV37–39. Other studies in Asia have also detected seropositive cyno- ZIKV infection in neotropical primates have now been documented11,12, molgus macaques (reviewed in ref. 40). The role of NHPs as hosts for the potential for such spillback events to launch a sylvatic cycle ZIKV in Asia, and indeed the existence of an Asian sylvatic ZIKV cycle, is remains unclear. Furthermore, the sylvatic cycles of DENV and ZIKV murkier because it is less studied. Wikan and Smith41 have proposed that persist in Asia and Africa, where both continue to spill over into the sparsity of human ZIKV infections is Asia is consistent with spillover humans and are sometimes translocated to new continents, e.g.13–19. from a sylvatic cycle. Of 234 wild cynomolgus macaques tested in Mathematical models using within-host viral dynamics to predict Malaysia42, 1.3% were seropositive for ZIKV, and, in a singular case, an between-host transmission are key for predicting arbovirus spillover, Australian traveler became infected with ZIKV after being bitten by a spread, and spillback20. These models assume within-host trade-offs cynomolgus macaque in Indonesia43. Moreover, machine learning that influence virus transmission21, particularly the trade-off between models have pinpointed cynomolgus macaques as having the highest instantaneous pathogen transmission and infection duration22,23. Host risk of all primate species worldwide to serve as a host for ZIKV44. mortality is often invoked in theoretical studies as the mechanism that The novel host was the neotropical squirrel monkey, Saimiri regulates infection duration, however most pathogens are cleared by boliviensis boliviensis, which machine learning models have identified the immune response. Indeed, for sylvatic DENV and ZIKV there is no as a high risk for spillback of ZIKV44. The two sylvatic viruses were evidence of disease in infected Old World monkeys24–26, although syl- DENV-2 strain P8-1407 and ZIKV strain DakAR 41525; the former was vatic ZIKV was first isolated in Uganda from a febrile sentinel rhesus isolated from a sentinel cynomolgus macaque in the forest canopy in macaque, an Asian species imported to Africa for viral surveillance16, Malaysia37–39 and the latter was isolated from an Aedes africanus, a and poses similar risks to the developing fetus in Old World monkeys primatophilic mosquito species that is generally confined to forest as do human-endemic lineages of the virus27–29. Thus, transmission of habitats45, in Senegal. Virus was delivered by infected Ae. albopictus these viruses likely depends on the trade-off between the magnitude of mosquitoes, which are native to Asia but distributed across much of viral replication and immune clearance30. the world, a likely bridge vector for spillover and spillback of both Studies of replication-clearance trade-offs in arboviruses are DENV and ZIKV46, and a vector in outbreaks of sylvatic ZIKV in Gabon47. scarce. The majority of publications on the intra-host dynamics of Figure 1 shows the experimental design for this study. To infect arboviruses during experimental infection of vertebrate reservoir macaques, cartons of either 1 (low dose, n = 5 monkeys) or 10 (high hosts found an inverse relationship between the magnitude (peak dose, n = 5 monkeys) Ae. albopictus that had been intrathoracically (IT) titer) of infection and duration of infection31,32. However, most of these inoculated with a sylvatic Malaysian strain of DENV-2, or 15 (high dose, studies delivered virus to the host via needle injection, which differs n = 3 monkeys) Ae. albopictus IT inoculated with a sylvatic African strain from natural transmission via mosquito bite33–35. We formulated a of ZIKV, or 10 uninoculated, control Ae. albopictus (n = 3 monkeys) were dynamical model comparing the effect of a “tortoise” strategy of low placed upon a monkey’s ear for approximately 10 minutes. Engorged magnitude, long-duration viremia and a “hare” strategy of short- mosquitoes were separated, incubated for 2 days, and force-salivated to duration, high-magnitude viremia and showed that arboviruses estimate dose of virus delivered. Squirrel monkeys were infected with adopting a tortoise strategy had higher rates of persistence in both either a high dose (15 mosquitoes) of a sylvatic Malaysian strain host and vector populations32. Ben-Shachar and Koelle36 investigated of DENV-2 (n = 10 monkeys) or a sylvatic African ZIKV strain the replication-clearance trade-off in DENV using within-host simula- (n = 10 monkeys) or control mosquitoes (n = 4 monkeys). Because the tion models based on data from a human cohort study in which small size of squirrel monkeys constrained blood draw frequency, mosquitoes fed on infected individuals, and found that a replication- monkeys were assigned to one of two cohorts and bled prior to infec- clearance trade-off selects for intermediate DENV virulence. Impor- tion and on alternating days in the first week pi. NK cells, neutralizing tantly for the current study, they invoked natural killer (NK) cells to antibody titers via plaque reduction neutralization titers (PRNTs), represent the innate immune response driving this effect. weight and temperature were measured on designated days prior to In the current study, we investigated trade-offs between replica- and post-infection (dpi). On designated dpi, uninfected Ae. albopictus tion and clearance of sylvatic DENV and ZIKV, as well as specific were allowed to feed on each animal to monitor transmission. immune responses associated with these trade-offs and potential dif- All raw data from macaques and squirrel monkeys infected with ferences in trade-offs between native and novel NHP hosts. This study DENV-2 or ZIKV, save for temperature and PRNT values against non- tested three a priori hypotheses, of which the first is fundamental and infecting viruses, are provided in Supplementary Datasets 1, 2, 3, and 4, the latter two consider mechanism and evolutionary context, respec- respectively. Following infection with either DENV-2 or ZIKV, body tively: Hypothesis 1 - sylvatic arboviruses experience a replication- temperatures of macaques48,49 and squirrel monkeys50 stayed within clearance trade-off in both native and novel hosts, as evidenced by a the normal range of each species (Fig. S1) and there were no differ- positive relationship between the dose of virus delivered and peak ences among treatment groups, nor were there any differences among virus titer, a positive relationship between virus titer and transmission treatment groups in changes in body weight change (Fig. S2). to mosquitoes, and a negative relationship between peak virus titer and duration of viremia; Hypothesis 2 - NK cells can serve as a proxy for Sylvatic DENV-2 delivered by mosquito bite to both macaques the innate immune responses that shape this trade-off, as evidenced by and squirrel monkeys resulted in nearly 100% seroconversion a negative relationship between NK cell mobilization immediately with high neutralizing antibody titers post-infection (pi) and peak virus titer as well as levels of neutralizing In IT-inoculated mosquitoes used to deliver DENV-2 to macaques, virus antibody, and Hypothesis 3 - sylvatic arboviruses have achieved an was not initially detectable via serial dilution and immunostaining of optimal replication-clearance trade-offs in native hosts but have not saliva samples in monolayers of mosquito cells. After saliva were sub- had evolutionary opportunity to reach this optimum in novel hosts, jected to one blind passage in mosquito cells, the virus was detected in a resulting in less transmission from novel hosts. subset of salivations (Table 1). Thus, we used the number of mosquitoes Nature Communications | (2024)15:2682 2 Article https://doi.org/10.1038/s41467-024-46810-x salivating detectable virus as a proxy for the total dose of DENV deliv- detectable virus, but this macaque seroconverted, demonstrating that it ered to macaques. For macaque, UG253A, the single mosquito that fed had become infected. Of the remaining eight macaques in the low and produced no detectable virus in the saliva, and UG253A never produced high-dose treatments, seven seroconverted, but one, BC407, did not, detectable viremia or seroconverted (defined here as a PRNT50 titer, even though it received a bite from a mosquito with detectable virus in the most permissive cutoff, ≥ 20; Supplementary Dataset 5). Hence the saliva (Table 1). All statistical analyses of neutralizing antibody levels UG253A was reassigned, post hoc, to the control treatment. Macaque were conducted using PRNT80 values, a more stringent measure con- SB393 received a single bite from a mosquito that did not salivate sistent with our previous study of sylvatic DENV-2 in African green Nature Communications | (2024)15:2682 3 Article https://doi.org/10.1038/s41467-024-46810-x Fig. 1 | Overview of experimental infections of cynomolgus macaques and and all squirrel monkeys, (ii) stool was collected daily from macaques to screen for squirrel monkeys with dengue virus serotype 2 (DENV-2, in green) and Zika occult blood; stool was not collected from squirrel monkeys due to pair housing of virus (ZIKV, in blue) and subsequent sampling. Blood samples were used to animals, (iii) mosquitoes that fed on infected squirrel monkeys on dpi 3 and 4 were measure viremia, natural killer (NK) cells, and neutralizing antibody as plaque force-salivated after a 14 day incubation, but this was not done in macaques, and reduction neutralization titer (PRNT). Monkey body temperature was monitored (iv) squirrel monkeys were euthanized at the conclusion of the experiment and continuously via implanted transponder and weight was monitored on designated necropsies were conducted; euthanasia was not performed on macaques. In one days. Infection and monitoring were similar in macaques and squirrel monkeys, instance, blood sampling (dpi 7) and mosquito feeding (dpi 8) occurred on dif- save that: (i) the sample size of infected and uninfected mosquitoes was increased ferent days for macaques. Images used under license from Shutterstock.com. from 10 in DENV-2 infected and control macaques to 15 in ZIKV-infected macaques monkeys24. The harmonic mean PRNT80 value (values LOD), with a harmonic mean PRNT80 value of cells were correlated with % NK cells prior to infection (P = 0.04), but 91.4 (Table 2, Fig. 2B). There were no notable findings in necropsy later % NK cells were not (P = 0.21). Thus, we analyzed only early % NK reports for squirrel monkeys infected with DENV-2. cells (1 dpi for macaques or 1 and 2 dpi for squirrel monkeys, batched due to the cohort structure of the experiment). Sylvatic DENV-2 replicated to low levels in macaques and Early % NK cells were not significantly different among control, squirrel monkeys and showed unexpected patterns of low dose macaques (1 mosquito) and high dose macaques ( > 1 mos- transmission quito) macaques (ANOVA, DF = 2, F = 2.74, P = 0.11) (Fig. 2C), and Day We initially predicted, based on hypothesis 1, that increasing dose of 28 PRNT80 values were not significantly different between low and virus would lead to increasing peak viremia, which in turn would be high dose macaques (t-test on log10 transformed data, DF = 7, t = 2.29, associated with greater instantaneous transmission to mosquitoes but P = 0.056) (Table 1). Thus, we combined low and high dose macaques a shorter duration of viremia. It was not possible to test most of these for analysis. Early % NK cells did not differ between control and high predictions in macaques infected with DENV-2, as viremia was detec- dose squirrel monkeys (ANOVA, DF = 1, F = 0.87, P = 0.37; note that all ted from serum via cell culture-based assay in only three macaques, four control squirrel monkeys were used in comparisons with both one from the low dose and two from the high dose treatment (Table 1, DENV and ZIKV infected squirrel monkeys; Fig. 2C). Fig. 3A). Furthermore, only three macaques transmitted DENV-2 to Due to the low level of replication of DENV-2, it was not possible to mosquitoes, all from the high dose treatment (Table 1, Fig. 3A, Sup- analyze peak titer for either macaques or squirrel monkeys. The effects plementary Dataset 6). Intriguingly, virus was not detected from raw or of early % NK cells and host species, as well as the interaction between passaged serum in any macaque on the day of transmission to mos- the two factors, on PRNT80 values in infected animals was tested using quitoes or, in the case of macaques FR840 and BC167, at any point a linear model: the only significant effect was early % NK cells (ANOVA, during the study. It was possible to analyze the number of infectious DF = 1, F = 10.8, P = 0.005), and the relationship between early % NK bites received and the likelihood of becoming viremic, as evidenced by cells and PRNT80 values was negative, as predicted (Fig. 2D). detectable virus in raw or passaged serum or mosquitoes (n = 5 viremic macaques total). Consistent with Hypothesis 1, detectably viremic Sylvatic DENV-2 dynamics, transmission, and neutralizing anti- animals received a significantly greater number of infectious bites than body were similar between native and novel hosts animals that were not detectably viremic (logistic regression, DF = 1, Based on Hypothesis 3, we predicted that transmission of DENV-2 n = 9, χ2 = 4.63, P = 0.03). would be less efficient from squirrel monkeys than cynomolgus Only three squirrel monkeys produced viremia that could be macaques. Squirrel monkeys and macaques received similar doses of detected from serum (Table 2, Fig. 3B). Two monkeys that produced DENV-2 (Supplementary Text S1.1, S1.2, S1.3). When considering only viremia detectable in serum transmitted DENV-2 to mosquitoes, but infected animals, neither early % NK cells (t test, DF = 17, t = −0.39, only when their viremia was at its lowest point. Another four monkeys P = 0.70) nor PRNT80 values (t test, DF = 17, t = 0.35, P = 0.73) differed that did not produce detectable viremia nonetheless transmitted between macaques and squirrel monkeys. Viremia was too low DENV-2 to mosquitoes (Table 2, Fig. 3B). Notably, for mosquitoes fed (Fig. 2A, B) and transmission was too infrequent (Fig. 3A, B) from both on both macaques and squirrel monkeys, DENV-2 could be detected macaques and squirrel monkeys for statistical comparison but did not more often in legs than in bodies (Supplementary Dataset 6). Contra support the claim that transmission is lower from a novel host. Intri- Hypothesis 1, dose of DENV-2 delivered had no significant impact on guingly, DENV-2 transmission from both macaques and squirrel mon- the likelihood of squirrel monkeys becoming detectably viremic keys was associated with low or undetectable viremia (Fig. 4A). (logistic regression, N = 7, Table 2, DF = 1, χ2 = 1.0556, P = 0.30). Given the small number of monkeys for which peak titer could be quantified, Sylvatic ZIKV delivered by mosquito bite to both macaques and we did not analyze the association of dose with peak titer. DENV-2 was squirrel monkeys resulted in 100% seroconversion, but antibody not detectable in any of the salivations from the mosquitoes that fed responses differed between the species on squirrel monkeys at 3 and 4 dpi. ZIKV was detectable in the saliva of 22/25 intrathoracially inoculated mosquitoes that fed on macaques. All three macaques ser- Following infection with sylvatic DENV-2, higher NK cells on day oconverted (Table 2, Supplementary Dataset 5) with a minimum 1 were associated with lower neutralizing antibody titers in both mean harmonic PRNT80 of 320 (treating the > 640 value as 640) macaques and squirrel monkeys (Fig. 2A). ZIKV was detectable in the saliva of 58/70 IT inoculated We predicted, based on Hypothesis 2, that NK cells mobilized early in mosquitoes that fed on squirrel monkeys. There was a tight rela- infection would show a negative relationship with peak virus titer and tionship between number of mosquitoes salivating ZIKV and total neutralizing antibody titer. To analyze variation in NK cells, we used the dose delivered (DF = 1, F = 12.6, P = 0.008, adjusted R2 = 0.56, slope percent NK cells of the total number of peripheral blood mononuclear 0.14 (0.05; 0.22) in log10 units). All squirrel monkeys that received cells (PBMCs). After Bonferroni correction for multiple testing, in infectious mosquito bites seroconverted based on PRNT50 values Nature Communications | (2024)15:2682 4  Article Table 1 | Assignment of each cynomolgus macaque to low dose (1 infected mosquito) or high dose (10 infected mosquitoes) of DENV, or high dose (15 infected mosquitoes) of ZIKV, or control (10 uninfected mosquitoes) treatments and subsequent viremia, transmission to mosquitoes, and neutralizing antibody response NHP ID Sex A priori Virus Post hoc Virus No. infected mosqui- No. infected mosquitoes with detectable Days on which virus was detectable in Days on which virus was trans- PRNT80 treatment Treatmenta toes engorged virus in saliva post passageb; (total dose virus serum, day of peak titer (peak titer mitted to mosquitoes (% on Day c delivered (log10 pfu)) log10pfu/ ml) infectedd) −10 28 Nature Communications | (2024)15:2682 NV289 M Control Control 8 0 – –