Zika Virus Encoding Nonglycosylated Envelope Protein Is Attenuated and Defective in Neuroinvasion

Abstract

Zika virus (ZIKV), a mosquito-transmitted flavivirus responsible for sporadic outbreaks of mild and febrile illness in Africa and Asia, reemerged in the last decade causing serious human diseases, including microcephaly, congenital malformations, and Guillain-Barré syndrome. Although genomic and phylogenetic analyses suggest that genetic evolution may have led to the enhanced virulence of ZIKV, experimental evidence supporting the role of specific genetic changes in virulence is currently lacking. One sequence motif, VNDT, containing an N-linked glycosylation site in the envelope (E) protein, is polymorphic; it is absent in many of the African isolates but present in all isolates from the recent outbreaks. In the present study, we investigated the roles of this sequence motif and glycosylation of the E protein in the pathogenicity of ZIKV. We first constructed a stable full-length cDNA clone of ZIKV in a novel linear vector from which infectious virus was recovered. The recombinant ZIKV generated from the infectious clone, which contains the VNDT motif, is highly pathogenic and causes lethality in a mouse model. In contrast, recombinant viruses from which the VNDT motif is deleted or in which the N-linked glycosylation site is mutated by single-amino-acid substitution are highly attenuated and nonlethal. The mutant viruses replicate poorly in the brains of infected mice when inoculated subcutaneously but replicate well following intracranial inoculation. Our findings provide the first evidence that N-linked glycosylation of the E protein is an important determinant of ZIKV virulence and neuroinvasion.IMPORTANCE The recent emergence of Zika virus (ZIKV) in the Americas has caused major worldwide public health concern. The virus appears to have gained significant pathogenicity, causing serious human diseases, including microcephaly and Guillain-Barré syndrome. The factors responsible for the emergence of pathogenic ZIKV are not understood at this time, although genetic changes have been shown to facilitate virus transmission. All isolates from the recent outbreaks contain an N-linked glycosylation site within the viral envelope (E) protein, whereas many isolates of the African lineage virus lack this site. To elucidate the functional significance of glycosylation in ZIKV pathogenicity, recombinant ZIKVs from infectious clones with or without the glycan on the E protein were generated. ZIKVs lacking the glycan were highly attenuated for the ability to cause mortality in a mouse model and were severely compromised for neuroinvasion. Our studies suggest glycosylation of the E protein is an important factor contributing to ZIKV pathogenicity.

Keywords: E protein glycosylation; Zika virus; attenuation; neuroinvasion.

FIG 1

Construction of an infectious clone of ZIKV MR766 and recovery and characterization of the clone-derived virus. (A) The viral genome with the encoded proteins is shown at the top. The solid circle at the 5′ end represents the cap. The genomic RNA from purified virions was amplified by RT-PCR with specific primers to generate four cDNA fragments (A, B, C, and D), which were assembled into a full-length clone using naturally occurring restriction enzyme sites shown above the fragments. The green rectangular box immediately upstream of fragment A represents the bacteriophage T7 RNA polymerase promoter (ϕ10). A unique BssHII site was incorporated immediately following the viral sequences in fragment D to linearize the plasmid prior to in vitro transcription. The blue lines on the left and right sides of the boxed regions represent sequences from pBR322. The entire genome, along with some sequence of pBR322 released by digestion with ScaI and NruI, was cloned in the pJAZZ-OC linear vector (bottom) at the unique SmaI site in the multiple cloning site of the vector. telL and telR, left and right telomeric sequence from coliphage N15; TelN, protelomerase gene of coliphage N15; repA, replication factor gene and origin of replication of coliphage N15; cB, repressor protein of coliphage N15; Cam^r, chloramphenicol resistance gene. The approximate positions of transcriptional terminators, T1 (T7 early transcription terminator), T2 (transcription terminator from the E. coli rrnB gene), and T3 (bidirectional E. coli tonB-P14 transcription terminator) are shown. The arrows indicate the direction of the coding sequence for repA and Cam^r. (B) In vitro transcripts from a full-length infectious clone (pJ-rMR) or from a clone encoding the catalytically inactive polymerase (pJ-rMR/Pol⁻) were transfected into Vero cells, and after 4 days, the cells were fixed and examined by immunofluorescent staining for E protein using 4G2 antibody. Mock, cells treated with the transfection reagent only. The top row shows phase-contrast images; the bottom row shows immunofluorescent images. (C) Infectious virus recovered from the transfected Vero cells at various days posttransfection as determined by plaque assay on Vero cells. The dashed line represents the limit of detection. (D) Silent nucleotide substitutions (shown in red) were introduced into rMR sequences as genetic tags, resulting in elimination of Bsu36I naturally present in the pMR and creation of AatII and BspEI sites in the rMR. The numbers on the left and right of the sequences are the nucleotide positions in the viral genome. Amino acid residues encoded by these sequences (in blue) are shown in the middle. (E) Confirmation of the presence of the genetic tag. A 1,822-bp DNA fragment was amplified by RT-PCR using the genomic RNA from the pMR or rMR virus and primers (Table 2) and digested with the restriction enzymes shown. Upon digestion of the PCR-amplified product with Bsu36I or AatII, two DNA fragments of 1,112 bp and 710 bp were generated. (F) Multistep growth of the pMR and rMR viruses in Vero and C6/36 cell lines. The graphs show mean values with error bars representing SD from three independent experiments. An unpaired Student t test (two-tailed) was used to determine significance. ns, nonsignificant.

FIG 2

Pathogenic properties of the pMR and rMR viruses in mice. (A and B) Weight loss and mortality of A129 mice inoculated s.c. with 1,000 PFU of the pMR (n = 6) or rMR (n = 6) virus. PBS (n = 5) was used as a negative control. (C) Genome copies in the plasma of animals described in panels A and B at 3, 4, 5, and 6 dpi. (D) Plasma from blood collected at 3 dpi or 6 dpi from pMR- or rMR-inoculated mice was quantitated for infectious virus. (E) Infectious-virus loads in the brain, spleen, and liver in mice inoculated with the pMR or the rMR. Two-way ANOVA (Tukey's multiple-comparison test) (A), multiple t tests (C and D), and an unpaired Student's t test, two tailed (E) were used to determine significance. ****, P < 0.0001; *, P < 0.05; ns, nonsignificant.

FIG 3

Characterization of the mutant viruses lacking the glycosylation site in the E protein. (A) Three-dimensional structure of dimeric E protein of ZIKV (Protein Data Bank [PDB] code 5IRE). The boxed region in the top diagram is enlarged below, showing the glycan moiety, glycan loop, fusion loop, and dimer interface. (B) Amino acid sequence of the rMR and the mutant (m1MR and m2MR) viruses around the VNDT motif. (C) Examination of the glycosylation status of the E protein in cells infected with the rMR, m1MR, or m2MR virus. Vero cells infected with the viruses (MOI = 1) were radiolabeled with Expre³⁵S³⁵S protein-labeling mix, immunoprecipitated with 4G2 antibody, digested with endo H or PNGase F, analyzed by SDS-PAGE, and detected by fluorography. Relative migration of molecular mass markers (in kilodaltons) is shown on the left. (D) Multistep growth of the rMR, m1MR, and m2MR in Vero and C6/36 cells. The graphs show mean values, with error bars representing standard deviations from the results of three independent experiments. An unpaired Student t test (two-tailed) was used to determine significance. ns, nonsignificant.

FIG 4

Mutant viruses are attenuated in mice. (A) Clinical scores at 6 dpi in A129 mice inoculated s.c. with 1,000 PFU of the rMR (n = 8), m1MR (n = 8), or m2MR (n = 6) virus. See Materials and Methods for clinical symptoms and scores. (B and C) Weight loss and mortality of A129 mice inoculated s.c. with 1,000 PFU of the rMR (n = 13), m1MR (n = 23), or m2MR (n = 12) virus. PBS (n = 6) was used as a negative control. (D and E) Viral-RNA copies and infectious-virus titers in the plasma of infected animals at various times postinfection. The data presented in panels B to E are combined data from two independent experiments. Unpaired Student t tests, two-tailed (A) and multiple t tests (D and E) were used to determine significance. ****, P < 0.0001; ***, P < 0.001; ns, nonsignificant.

FIG 5

Histopathological examination of tissues of animals infected with the wt or mutant virus. H&E staining of sections of brains (meninges and cerebrum) and spleens of animals infected with the rMR, m1MR, or m2MR virus is shown. Representative images from the experiment are shown. Bars, 60 μm in meninges and cerebrum, 30 μm in spleen.

FIG 6

The mutant viruses replicate less efficiently in the brains of infected animals. (A and B) Viral-genome copy numbers (A) and infectious-virus titers (B) in the brains of animals infected with the rMR (n = 8), m1MR (n = 8), and m2MR (n = 8) viruses at 6 dpi. (C) Infectious-virus titers in the spleens and livers of animals infected with the rMR, m1MR, and m2MR viruses at 6 dpi. Data from two separate experiments were combined. An unpaired Student t test (two-tailed) was used to determine significance. ****, P < 0.0001; ns, nonsignificant.

FIG 7

Mutant viruses replicate well in the brain when inoculated i.c. Mice were inoculated i.c. with 1,000 PFU of the rMR (n = 5), m1MR (n = 5), or m2MR (n = 5) virus. (A and B) RNA copies (A) and infectious virus (B) in brains of animals at 6 dpi. (C) Infectious virus in spleens and livers of the same animals at 6 dpi. An unpaired Student t test (two-tailed) was used to determine significance. *, P < 0.05; ns, nonsignificant.