Determinants of Zika virus host tropism uncovered by deep mutational scanning

Abstract

Arboviruses cycle between, and replicate in, both invertebrate and vertebrate hosts, which for Zika virus (ZIKV) involves Aedes mosquitoes and primates¹. The viral determinants required for replication in such obligate hosts are under strong purifying selection during natural virus evolution, making it challenging to resolve which determinants are optimal for viral fitness in each host. Herein we describe a deep mutational scanning (DMS) strategy^2-5 whereby a viral cDNA library was constructed containing all codon substitutions in the C-terminal 204 amino acids of ZIKV envelope protein (E). The cDNA library was transfected into C6/36 (Aedes) and Vero (primate) cells, with subsequent deep sequencing and computational analyses of recovered viruses showing that substitutions K316Q and S461G, or Q350L and T397S, conferred substantial replicative advantages in mosquito and primate cells, respectively. A 316Q/461G virus was constructed and shown to be replication-defective in mammalian cells due to severely compromised virus particle formation and secretion. The 316Q/461G virus was also highly attenuated in human brain organoids, and illustrated utility as a vaccine in mice. This approach can thus imitate evolutionary selection in a matter of days and identify amino acids key to the regulation of virus replication in specific host environments.

Figure 1.

Deep mutational scanning (DMS) of C-terminal region of ZIKV E protein. a) Schematic of ZIKV E protein and workflow of DMS screening for the C-terminal 204 codons of ZIKV E protein. Amplicons generated using forward (red arrows) and reverse (blue arrows) mutagenesis primers are combined with other ZIKV cDNA fragments to generate infectious cDNA library by Circular Polymerase Extension Reaction (CPER). The CPER cDNA library is then transfected into mammalian (Vero) or mosquito (C6/36) cells, recovered viruses harvested in culture supernatants, deep sequenced and analysed to identify preferentially selected viral mutants. b) Graphical representation of deep sequencing data showing codon variant heatmap, and c) Mutation frequency of cDNA amplicon library illustrating the extent of mutagenesis in the E-DIII stem-anchor region. Mutation frequency is defined as the percentage at which codon mutants are detected in sequenced molecules at each site. For example, at each site, if mutations are present in 100% of sequenced molecules, the frequency is 1. If it is not present in any sequenced molecules, the frequency is 0. d) Mutation frequency of CPER cDNA library, mammalian (Vero) cells-selected virus population and mosquito (C6/36) cells-selected virus population. Identities of preferentially selected amino acid mutations are indicated above each peak.

Figure 2.

In vitro characterization of ZIKV mutants. a) Immuno-plaque assay (iPA) of WT and mutant viruses in Vero cells. b-e) Growth kinetics of WT and mutant viruses infected at a multiplicity of infection (MOI) of 0.1 in b) Vero, c) A549, d) HTR-8, and e) C6/36 cells. f) Growth kinetics of Vero cells infected with WT or 316Q/461G mutant at MOI 0.1, and incubated at 28°C or 37°C. g) WT and mutant viruses, as well as control dengue serotype 2 (strain TSV-01) virus were subjected to heat treatment at the indicated temperatures for 5 hours, and then titered. All culture supernatants above were harvested at their indicated time points after infection and virus titres determined by iPA in Vero cells. n=3 independent experiments for all assays, and statistical analysis was performed by two-way ANOVA with Tukey’s multiple comparisons test against WT virus. * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, **** P ≤ 0.0001. Mean ± standard error of the mean. Limit of detection for iPA is 1.6 Log₁₀FFU/mL. h) Immunoprecipitation with anti-E antibody (6B6C-1) of ³⁵S-labelled cell lysates (C) or supernatants (S) from Vero or C6/36 cells infected with WT or 316Q/461G mutant viruses. Envelope protein indicated by the arrowhead. Antigen-capture ELISA detection of i) extracellular and j) intracellular envelope protein. Dotted lines show the limit of detection, n=3 independent experiments. Detection of k) extracellular and l) intracellular viral RNA by qRT-PCR. N.D. = not detected. Mean ± standard error of the mean, n=3 independent experiments. Statistical analysis for qRT-PCR data was performed using unpaired t-test, two-tailed. *** P = 0.0005.

Figure 3.

Immunofluorescence and transmission electron microscopy of infected C6/36 or Vero cells. a) Immunofluorescence microscopy of Vero or C6/36 cells infected with WT or 316Q/461G mutant, and immuno-stained with 4G2 anti-E antibody (green channel), and DAPI (blue channel). Scale bars are 5 μm. Representative images of n=3 biologically independent samples. b) Transmission electron microscopy of C6/36 or Vero cells infected with WT, 316Q/461G mutant, or uninfected (mock). Virus particles are indicated by arrowheads. CM – convoluted membranes, VP – vesicle packets, V –vesicles, N - nucleus, mt – mitochondria, ER – endoplasmic reticulum. Scale bars in C6/36 slides are 200 nm, and scale bars in Vero slides are 500 nm. n=3 biologically independent samples.

Figure 4.

Molecular analysis and modelling of the S461G and K316 mutations. a) Cryo-EM structure of H/PF/3013 strain ZIKV E dimer (PDB accession number: 5IZ7) showing the position of the S461G mutation (red spheres) in the mature structure. E and M proteins represented by green and yellow ribbon structures, respectively. b) Residue S461 sits in a pocket -in contact with six M protein residues; Y74, S75 (from chain D in M protein) and S8, K11, E24, H28 (from chain F in M protein). Contact atoms are surrounded by pink highlights. Protein contact analysis by Molecular Operating Environment (MOE) software package (using 5IZ7). c) The S461G substitution results in only 3 residues in contact with G461. The S461G mutation was introduced on the 5IZ7 structure followed by energy minimization and protein contact analysis by MOE. d) Homology model of ZIKV Natal pr protein (based on DENV structure PDB:3C5X and generated using Swiss-Model) fitted to the cryo-EM structure for low pH immature DENV (EMD-5006), showing a potential salt bridge forming between D57 in pr and K316 in E-DIII (right inset). A S66L mutation can potentially result in hydrophobic interaction between L66 in E-DII and A47 in pr (left inset).

Figure 5.

Infection of iPSC-derived human brain organoids. a) Growth kinetics of WT or 316Q/461G viruses in iPSC-derived human brain organoids. Three different organoids per virus were infected. Mean ± standard deviation. b) Morphology of iPSC-derived human brain organoids infected with WT virus, 316Q/461G mutant virus, or uninfected visualized by light microscopy. Scale bars are 0.355 mm. Representative images of n=3 biologically independent samples per timepoint. c) Quantification of organoid sizes after infection. Circumference of four organoids per group were measured. Mean ± standard deviation. Immunohistochemistry (IHC) analysis of iPSC-derived human brain organoids co-stained for ZIKV E protein and d) SOX2, e) BRN2, f) MAP2, g) TBR1, h) Ki67, i) CASP3 at the 4 and 18 dpi timepoints. Scale bars are 100 μm. Paired images represent the original image on the left, with the zoom-in of the inset (dotted box) on the right. Representative images of n=3 biologically independent samples per timepoint.

Figure 6.

316G/461Q mutant as an attenuated ZIKV vaccine candidate. a) Timeline of mouse experiment. b) Viraemia in IFNAR1^−/− C57BL/6 mice vaccinated with 10⁴ FFU/mouse of WT (n=5) or 316Q/461G virus (n=5). Daily tail bleeds were conducted up to 7 days after vaccination and virus titres in the collected sera were determined by iPA in Vero cells. Mean ± standard error of the mean. c) Neutralising antibody titers against ZIKV-MR766 virus in the mouse sera collected at 21 days post-vaccinations with WT virus (n=5), 316Q/461G mutant virus (n=5) or naïve (n=4). The neutralizing antibody titres were determined by plaque reduction neutralization test (PRNT) on Vero cells. Mean ± standard error of the mean. d) Viraemia in IFNAR1^−/− C57BL/6 mice vaccinated with WT virus (n=4) or 316Q/461G virus (n=5) or naïve (n=4) and challenged with 10³ CCID₅₀/mouse of ZIKV-MR766. Daily tail bleeds were conducted up to 7 days post-challenge, and virus titres in collected mouse sera were determined by iPA in Vero cells. Mean ± standard error of the mean. e) Percentage body weight loss of IFNAR1^−/− C57BL/6 mice vaccinated with WT (n=4) or 316Q/461G (n=5) or naïve (n=4) was determined up to 21 days after challenge with ZIKV-MR766. Mean ± standard error of the mean. f) Survival of IFNAR1^−/− C57BL/6 mice vaccinated with WT (n=4) or 316Q/461G (n=5) or naïve (n=4) and challenged with ZIKV-MR766. Statistical analysis performed by Log-rank (Mantel-Cox) test, two-sided, compared to WT. ** P = 0.0025.