A rhesus macaque model of Asian-lineage Zika virus infection

Infection with Asian-lineage Zika virus (ZIKV) has been associated with Guillain–Barré syndrome and fetal abnormalities, but the underlying mechanisms remain poorly understood. Animal models of infection are thus urgently needed. Here we show that rhesus macaques are susceptible to infection by an Asian-lineage ZIKV closely related to strains currently circulating in the Americas. Following subcutaneous inoculation, ZIKV RNA is detected in plasma 1 day post infection (d.p.i.) in all animals (N=8, including 2 pregnant animals), and is also present in saliva, urine and cerebrospinal fluid. Non-pregnant and pregnant animals remain viremic for 21 days and for up to at least 57 days, respectively. Neutralizing antibodies are detected by 21 d.p.i. Rechallenge 10 weeks after the initial challenge results in no detectable virus replication, indicating protective immunity against homologous strains. Therefore, Asian-lineage ZIKV infection of rhesus macaques provides a relevant animal model for studying pathogenesis and evaluating potential interventions against human infection, including during pregnancy.


44
In contrast, immunocompetent macaque monkeys are widely used in both infectious animals, peaking at over 1 x 10 3 vRNA copies per sample in 3 of 6 animals (Fig. 1b). Notably, 77 as with urine, the kinetics of virus detection in saliva occurred after peak plasma viremia.

78
Cisterna magna punctures were performed at 4 and 14 dpi to quantify viral RNA in CSF; vRNA 79 was detectable at 4 dpi in 3 out of 5 animals from which CSF could be obtained (Fig. 1d).

82
We next characterized the immune response to infection by staining peripheral blood 83 mononuclear cells (PBMC) for multiple lineage and activation markers. Proliferating (Ki-67+) NK 84 cells, CD8+ T cells, and CD4+ T cells expanded significantly above baseline levels by 6 dpi 85 ( Fig. 2a,b). NK and CD8+ T cell expansion increased as plasma vRNA loads decreased starting 86 at 6 dpi. We also enumerated circulating plasmablasts, defined as CD3-/20-/14-/16-/11c-/123-87 and CD80+/HLA-DR+ cells, on 0, 3, 7, 11 and 14 dpi (Fig. 2c) 14 . The peak plasmablast 88 expansion occurred between 7 and 10 dpi in 5/6 animals. Serum neutralizing antibody 89 responses were also measured by plaque reduction neutralization tests (PRNT 90 ). All animals 90 exhibited high neutralizing antibody (nAb) titers as early as 14 dpi (Fig. 2d), the earliest time 91 point tested. Cohort 1 animals were tested at 64 dpi and cohort 2 animals were tested at 14 and 92 28 dpi. Together these data suggest that peak activation of the adaptive immune response and 93 antibody production occurs 5-7 dpi and may both be important to control viral replication as 94 evidenced by reducing vRNA loads in the plasma at these time points.

95
To determine whether activation of T cells correlated with the appearance of Zika virus-96 specific responses, we performed interferon-gamma ELISPOT on PBMCs collected at 4, 10 and 97 14 dpi for cohort 2 animals. Cells were stimulated with pools of 15mer peptides collectively 98 representing the amino acid sequence of the Asian-lineage NS5 protein (GenBank: KU321639).

99
We detected specific IFN-gamma secretion in response to 12 of 16 peptide pools in at least one 100 animal (Extended Data Fig. 4a). Overall, this data supports that there are ZIKV-specific T cell 101 responses in all animals tested.
To determine whether the immune responses that we detected following primary 103 challenge were protective against homotypic rechallenge, we rechallenged the three animals in 104 cohort 1 ten weeks after primary infection with 1 x 10 4 PFU of a homologous virus (Fig. 1b inset   105 and Extended Data Fig. 1). Plasma, urine and saliva vRNA loads remain negative to at least 9 106 dpi (as of 5/4/2016), indicating complete protection against ZIKV re-infection.

107
We also challenged two time-mated rhesus macaques at approximately gestation day 31 108 and 38 (mid-first trimester) with 1 x 10 4 PFU of ZIKV (cohort 3; see Extended Data Fig. 1 and 109 Fig. 3a). Both animals were viremic by 1 dpi and exhibited peak plasma viral loads of > 4 x 10 5 110 vRNA copies/ml by 3 or 6 dpi (Fig. 3b). Infectious virus was also quantitated by plaque assay 111 from the serum of 660875 (Extended Data Fig. 5). In contrast to their non-pregnant 112 counterparts, both animals maintained persistent plasma viremia (vRNA copies/ml) to 57+ and 113 29 dpi (Fig. 3b). This is similar to a case described by Driggers et. al., where a pregnant mother 114 had persistent ZIKV vRNA detected from 35 to 70 dpi that did not resolve until termination of 115 pregnancy 7 . The fetus was found to have 2x10 8 copies/ml of virus in brain tissue and it is 116 speculated that the fetus may have been the source of the prolonged maternal plasma viremia.

117
We will continue to monitor these pregnant animals for vRNA in the blood and amniotic fluid and 118 will determine the infection status of the fetus upon termination of the pregnancy either at full 119 term or earlier if necessary for the health and safety of the mother. Amniocentesis using 120 ultrasound guidance was performed at 43 dpi for 827577 and 36 dpi for 660875 and both were 121 negative for ZIKV RNA.

122
Both animals generated similar activation of NK, CD8+ T cell and CD4+ T cell responses 123 above baseline as non-pregnant animals (Fig. 3c). Expansion of plasmablast cells was also 124 observed by 10-21 dpi with one animal expanding more than, and one animal expanding less 125 than, the average non-pregnant animal (Fig. 3d). Neutralizing antibodies were detected by 21 126 dpi for 827577 and 10 dpi for 660875 and were similar to the cohort 2 non-pregnant animals at expected with repeated ketamine sedation and blood collection. The same pregnant animal also 129 developed persistent regenerative anemia characterized by circulating nucleated erythrocytes.

130
Updates to the cohort 3 experiments are available in real-time at goo.gl/rmNCqf.

131
Altogether, our study shows the persistence of ZIKV RNA in the plasma of rhesus 132 macaques for approximately 10 days, similar to other vector-borne flaviviruses that cause acute, 133 typically self-limiting infections in humans. This work also shows that natural ZIKV infection 134 elicits a robust immune response including ZIKV-specific T cell response and neutralizing 135 antibody responses that confers protection against reinfection. However, the prolonged 136 detection of vRNA in urine and saliva after apparent clearance from the blood, detection of virus 137 in the CSF, and occasional plasma "blips" after initial clearance, suggest that ZIKV may persist 138 longer, at low levels, in certain tissues. Future work in rhesus macaques will seek to determine if 139 and where these reservoirs may exist and whether they seed virus into fluids that might allow for 140 human-to-human transmission. 141 Our study establishes immunocompetent rhesus macaques infected with physiologically 142 relevant ZIKV as a relevant translational model for infection and pathogenesis. The large 143 immunological toolset available for rhesus macaques will enable investigations of immunity and 144 potential vaccines. Pregnancy, the maternal-fetal interface, and fetal development have been 145 described in detail in rhesus macaques, so this model will also enable assessments of the This was a proof-of-concept study designed to establish the infectivity and viral dynamics of                                       Extended Data Figure 1| Schematic representation of the timeline of infection and sampling for each animal in the presented studies. Cohort 1 received the first ZIKV challenges and were then rested for 6 weeks before a rechallenge. For all studies, samples were collected daily for 10 days and then on 14, 21, and 28 dpi as indicated by hashes in the timelines. Cohort 3 represents the two pregnant animals that were challenged on 2 different days. Both animals are currently in the once weekly sampling phase until the pregnancies come to term (~165 gestational days). Cohort 2 was a repeat experiment of cohort 1 that allowed for additional experiments and sample collection (e.g., serum plaque infectivity) that were not feasible when we initiated cohort 1 studies. These animals are currently in a 6-week rest period and will be rechallenged on June 6, 2016. Ages of all animals are indicated under each macaque identification number.
End of pregnancy  Figure 2| Complete blood counts and serum chemistries for macaques infected with ZIKV. a. Animals were infected with different doses of ZIKV. Cohort 1 animals are represented by solid lines and cohort 2 animals are represented by dotted lines. All non-pregnant animals had serum chemistry analysis performed at -7, 0, 1, 2, 3, 4, 6, and 14 dpi or at -6, 2, 5 and 11 dpi. b. AST blood chemistries c. ALT serum chemistries. d. CK serum chemistries. Complete blood counts were measured prior to infection, daily for 10-11 days after infection and then every 3-7 days until 28 dpi. e. white blood cell counts. f. % lymphocytes. g. red blood cell counts.   Figure 4| Antigen-specific T cell responses by IFNγ-ELISPOT. a. Average spot forming cell counts for PBMC collected from each animal at 4, 10 and 14 dpi. Data were baseline corrected by subtracting the average negative control values from each response. A threshold of 10.0 SFC/100,000 cells was set as the minimum value to be considered a positive T cell response, as indicated by the dashed line. b. Each pool was comprised of 10 overlapping 15mer peptides offset by 4 amino acids. c. Peptide pools eliciting T cell responses at 4, 10 and 14 dpi for each animal. The region of the NS5 protein that is represented by each pool of overlapping 15mers is provided. MHC class I haplotypes of each cohort 2 animal are also presented. All three animals shared the A004 and B012b major histocompatibility complex haplotypes and two animals shared the A023 haplotype. Therefore, it was not surprising that 3 pools were recognized by 2 different animals likely sharing the MHC class I allele that is presenting one of the peptides in those pools. Grayed pools were positive in more than one animal and bolded pools were positive at more than one time point in the same animal. To detect signs of morbidity, animals were evaluated daily for evidence of 1 disease, injury, or psychological abnormalities (e.g., inappetence, dehydration, diarrhea, 2 depression, inactivity, trauma, self-injurious or stereotypical repetitive behaviors (e.g.,

3
pacing) often seen in captive animals). Five of six animals exhibited mild to moderate 4 inappetence, which resulted in mild weight loss in four animals. Two animals (912116 5 and 393422) also developed a very mild rash around the inoculation site at 1 dpi that 6 persisted for 4-5 days. No other abnormal clinical signs were noted (e.g., increased body 7 temperature, joint pain, lymphadenopathy, lethargy). 8 Daily complete blood counts (CBCs) were evaluated for all six non-pregnant 9 animals for 10 dpi and then every 3 to 7 days thereafter and serum chemistry analyses 10 were performed intermittently post-infection as per protocol (Extended Data Fig. 2).

11
Reference intervals (RI) developed for the WNPRC colony for species, gender, and age 12 were used to evaluate results. All six animals developed elevated serum creatine kinase 13 (CK), which peaked by 5 dpi (Extended Data Fig. 2d). Increases in serum CK are 14 strongly associated with muscle damage and myositis (skeletal, smooth, and cardiac),

26
. CC-BY-ND 4.0 International license peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/046334 doi: bioRxiv preprint first posted online Mar. 30, 2016; All the animals displayed decreased total WBC numbers following infection, but 27 only one animal fell below the RI with a value of 2.88 ths/ul (3.70-15.70). WBC numbers 28 rebounded almost completely to pre-infection levels around 10 dpi in all six animals 29 (Extended Data Fig. 1e). Both animals that received the highest dose of inoculum 30 developed persistent mature neutrophilia around 7-14 dpi that lasted through 28 dpi.