A single nonsynonymous mutation on gene encoding E protein of Zika virus leads to increased neurovirulence in vivo

Zika virus can infect a wide range of tissues including the developmental brain of human fetuses, causing from mild to severe clinical diseases. Whether its genetic characteristics impacts on viral pathogenesis is incompletely understood. We have obtained viral variants through serially passage of a clinical Zika virus isolate (SW01) in neonatal mice in vivo and found some of which exhibited markedly increased virulence and neurotropism. By deep sequencing analysis, the more pathogenic viral variants were found to contain four dominant nonsynonymous nucleotide mutations on genes encoding E and NS2A proteins. Further investigation using molecularly cloned viruses revealed that a single 67D (Aspatic acid) to N (Asparagine) substitution on E protein is sufficient to confer the increased virulence and neurotropism. These findings provide new insight into Zika virus pathogenesis and suggest novel targets for the development of therapeutics. Author Summary Recent large outbreaks of Zika virus infection worldwide have revealed an association between the viral infection and increased cases of specific neurological problems including Congenital Zika Syndrome (including microcephaly) and adult Guillain–Barré Syndrome. However, the determinants of the increased neurovirulence of Zika virus remain uncertain. One hypothesis is that some unique changes across the Zika viral genome have led to the occurrence of these neurological diseases. To test this hypothesis, we continuously propagated a clinical isolate of contemporary Zika virus (SW01) in neonatal mice brain for 11 times to obtain an mouse central nervous system (CNS) adapted Zika virus (MA-SW01) that showed significantly increased neurovirulence in vivo. We then discovered that a single G to A nucleotide substitution at the 1069 site of Zika virus open reading frame leading to a D (aspartic acid) to N (asparagine) in viral Envelope protein is responsible for the increased neurovirulence. These findings improve our understanding of the neurological pathogenesis of Zika virus and provide clues for the development of antiviral strategy.

neurovirulence in vivo than the older African strains (17,18). Other studies have 117 placed more emphasis on the global interations between virus and host. It has 118 been reported that high levels of virus RNA can persist in human fetal and 119 neonatal central nervous system (CNS) in vivo (19,20), and experimentally  (13,(27)(28)(29)(30). Zika virus can also infect wild type 135 neonatal mice and cause diseases that resemble to some extent microcephaly, 136 paralysis, and seizure (31-33). Mechanistically, these neurological 137 manifestations have been linked to Zika virus infection of neuron progenitor 138 cells and other neurocytes in neonatal mice in vivo (11,17,34). Of note,139 comparative analysis between human and rodents has shown that mouse brain 140 at postnatal day 1-2 roughly corresponds to the human fetal brain at mid-141 gestation stage (35,36). Therefore, newborn mice have also been used as a Combining animal study and viral genetic analyses, here we report the isolation

173
To characterize the mouse adapted Zika virus, 100 PFU of parental SW01 virus, 174 mouse adapted MA-SW01 virus, or negative control sterile PBS were i.c.

175
injected into newborn DP2 Balb/c mice and monitored for up to 25 days. All 176 mice in the SW01 group showed a slow and moderate disease progression 177 within 11-25 days ( Fig.2A, 2C, 2D), in comparison, those in the MA-SW01 178 group showed more rapid weight loss, severe morbidity, and even death within 179 6-8 days (Fig.2B, 2C, 2D

234
It is interesting to note that the average viral copy number in the brain of MA-235 SW01 infected mice was dramatically higher (about 488-fold) than that of SW01 236 infected mice at day 6 post infection ( Fig.3A). More viruses in the brain may be 237 explained by two possibilities, one is that MA-SW01 virus replicates more 238 efficiently than SW01 in the central nerve system (CNS); another is that MA-239 SW01 virus has increased neuro-invasion efficiency. To investigate these two  (Fig.3B), but 13.8 fold higher than that of SW01 group at 6 days post 245 infection (Fig.3C), suggesting more efficient replication of MA-SW01 in the 246 brain. Given that viral RNA level in the brain of MA-SW01 group was about 15 247 fold higher than that of SW01 group at day 3, even with s.c. inoculation (Fig.3A), 248 we deduced that the more virulent MA-SW01 virus has great penetration to 249 brain. To more directly visualize viral infection in the brain, immunofluorescence 250 staining of virus E protein in brain tissue sections was performed. Dramatically   i.c. at 10 PFU/mouse, using the parental molecularly cloned CAM-WT virus as 302 a control, and then monitored for 25 days. Results showed that CM1 and CM3, 303 but not CM2, were more virulent than parental virus CAM-WT in neonatal mice 304 (Fig.5C). Since both CM1 and CM3 contain 2 mutations on E protein (D67N, 305 M68I), and CM2 only contains the single NS2A mutation, the above results 306 suggest that the E protein mutations were determinants of increased virulence 307 of MA-SW01, and the NS2A mutation was not. Therefore, CM1 virus containing 308 two E protein mutations was chosen for further investigation. 309 We first examined whether the route of infection alters viral virulence. DP2

354
showed that CM1-A was similar to CM1 in causing 100% mortality of infected 355 mice at 12-13 days post infection (Fig.6E). These data demonstrated that a 356 single D67N mutation is sufficient to account for the increased virulence of CM1. RNA between CAM-WT and CM1 groups; at 11 days post infection, however, 365 viral RNA of CM1 group was significantly higher than that of CAM-WT group in 366 brains, eyes and blood, but not in spleens and kidneys (Fig.7A), confirming the 367 combination of these E protein mutations affect tissue tropism.

368
The fact that more viral RNA was detected in brain and eyes, that are rich in 369 nerve cells susceptible to Zika virus infection, implies that CM1 virus has growth 370 advantage over CAM-WT in these cells. To directly test this notion, DP2 Balb/c 371 mice were infected i.c. with CAM-WT or CM1, and then monitored for virus 372 burden by real-time qPCR. Results showed that brain viral loads of CM1 group 373 was higher than that of CAM-WT group at 11 days post infection, but not 3 days 374 post infection (Fig.7C, 7D), suggesting that CM1 has growth advantage over 375 CAM-WT in brain.

376
To further dissect wthether single D67N mutation in E protein plays the essential 377 role of altering viral virulence and tissue tropism, we next used CM1-A virus 378 (containing only D67N ) to perform viral infection experiments. Results showed 379 that irrespective of throught s.c. infection (Fig.7B), or i.c. infection (Fig.7C, 7D),   (Fig.6A, 6B), it is reasonable to deduce 395 that D67N substitution provides more viral fitness in CNS, and thus enables 396 MA-SW01 to outgrow other viral variants within the SW01 quasispecies.       Survival curves were analyzed by log rank test. Body weight was analyzed by 605 Two-way ANOVA (Turkey correction). All summarized data were compared for 606 statistical differences by student's t test or two-way-ANOVA. All analyses were 607 performed on Graphpad Prism V8.0 platform. Statistical significance levels 608 were reported as the following: "*" for p < 0.05; "**" for p < 0.01; "***" for p < 609 0.001 or less.