**3. Results**

## *3.1. Characterization of Mutant ZIKV Molecular Clones*

To determine the contribution of E-152/156/158 residues in ZIKV E protein functions, we generated two mutant molecular clones: MR766E-152I/156T/158H, hereafter called MR766E-MUT, in which E-152/156/158 residues of BR15 epidemic strain were introduced, and BR15E-152T/156I/158Y, hereafter called BR15E-MUT, in which E-152/156/158 residues were replaced with their counterparts from the MR766 historical African strain (Figure 1a). Genomes were assembled using the infectious subgenomic amplicon method [29]. Briefly, Vero cells were electroporated with overlapping fragments, in which appropriate mutations have been previously introduced. The two recovered clones were viable and twice amplified on Vero cells. Titres of P2 working viral stocks were determined in Vero cells and were ranging from 5 × 10<sup>5</sup> to 1 × 10<sup>8</sup> PFU·mL−<sup>1</sup> (Figure 1b). MR766E-MUT and BR15E-MUT gave plaque morphologies that resembled those of respective MR766 and BR15 parental clones (Figure 1c), which is in agreemen<sup>t</sup> with previously published data [27]. In addition, we confirmed that the introduced mutations affected the electrophoretic mobility of ZIKV E proteins, suggesting that E-152/156/158 residues from BR15 E protein might enable its glycosylation (Figure 1d + Figure S1).

**Figure 1.** ZIKV mutant molecular clones. In (**a**), schematic representation of mutant viral clones BR15E-MUT and MR766E-MUT and their respective parental clones. In (**b**), histograms showing viral titres. Values represent means and standard errors of three independent experiments. In (**c**), examples of infectious plaques developed for BR15E-MUT andMR766E-MUT, and parental clones, after plaque-forming assay on Vero cells. In (**d**), Vero cells were infected with parental and mutant molecular clones (MR766 and BR15) at a MOI of 1. 24 h post-infection (hpi), cells were then lysed and subjected to an immunoblot, in non-reducing conditions. Anti-ZIKV EDIII immunoblot shows differences of electrophoretic mobility associated with residue mutations.

#### *3.2. Residues E-152*/*156*/*158 from BR15 Potentiate Viral Infectivity*

We first analysed infectivity of P2 virus stocks as described above. Particle-to-PFU ratios obtained from parental clones were around 900–1000 (Table 1), which is consistent with our previous observations [23]. We then analysed particle-to-PFU ratios of the two E mutant clones. Addition of residues E-152/156/158 from BR15 to MR766 resulted in a 2.3-fold decrease in the particle-to-PFU ratio. In contrast, when residues E-152/156/158 from MR766 were introduced to BR15, the particle-to-PFU ratio was markedly increased with more than ten-folds. These results sugges<sup>t</sup> that residues E-152/156/158 from BR15 potentiate virion infectivity.

**Table 1.** Table showing particle-to-PFU ratios. Viral RNA extracted from viral stock P2 were subjected to quantification by RT-qPCR using E primers. Obtained Ct values were plotted in a standard curve (serial dilutions of plasmid copies) in order to ge<sup>t</sup> the number of viral RNA molecules per mL. These results were compared to viral stock quantifications by standard plaque-forming assay, which then gave the particle-to-PFU ratios, also named vRNA-to-PFU ratios. Values represent means and standard errors of two to four independent experiments.


#### *3.3. Alteration of Residues E-152 to E-158 of ZIKV E Protein Does Not A*ff*ect Virus Binding to Host Cells but May A*ff*ect Virus Progeny Production*

We previously showed that historical and epidemic ZIKV strains display differences in their abilities to bind host cells, leading to differences in cell susceptibility to infection (18, 19). Here, we further investigated the ability of the described mutant clones to bind onto A549-Dual™ cells. Virus binding assays were performed and analysed by RT-qPCR to determine virus particle binding onto cell surface after an incubation period of 1 h. Panels A and B show no difference between mutant clones and their respective parental clones (Figure 2). These results contrast with other studies in mosquito cells [27], suggesting that viral receptors may vary between vertebrate and invertebrate cells. These data sugges<sup>t</sup> that alteration of residues E-152 to E-158 of ZIKV E protein does not affect virus bindings to A549-Dual™ cells.

Instead, these results sugges<sup>t</sup> that E-152/156/158 residues might influence ZIKV progeny production. Indeed, the progeny production of MR766E-MUT was modestly but reproducibly increased in comparison to that of MR766 (3 × 10<sup>7</sup> PFU·mL−<sup>1</sup> vs. 1 × 10<sup>7</sup> PFU·mL−1) at 72 hpi [23]. Conversely, kinetics of the BR15E-MUT progeny production were strongly altered compared to BR15 (2 × 10<sup>6</sup> PFU·mL−<sup>1</sup> vs. 4 × 10<sup>7</sup> PFU·mL−1) at 72 hpi, respectively [23]. Differences in progeny production were observed at as early as 24 hpi. Similar differences were also seen on the percentages of the infected cell at 48 hpi. Taken together, these results indicate that E-152/156/158 residues from BR15 potentiate viral infectivity, independently of the virus binding to host A549-Dual™ cells.

**Figure 2.** Analysis of virus binding and viral growth in A549-Dual™. In (**a**) and (**b**), for virus binding assays, cells were incubated with viral clones at the MOI of 1 for 1 h at 4 ◦C. The number of virus particles bound to cell surface was measured by RT-qPCR. Values represent means and standard errors of three independent experiments. In (**c**) and (**d**), A549-Dual™ were infected with BR15E-MUT and MR766E-MUT at MOI of 1. Infectious virus released into the supernatants of infected A549-Dual™ cells were quantified at 24, 48 and 72 hpi. Error bars represent standard errors of at least two independent experiments. In (**e**) and (**f**), A549-Dual™ were infected with BR15E-MUT and MR766E-MUT and parental clones at MOI of 1. Percentages of ZIKV-infected cells were determined at 48 h by flow cytometry using anti-E mAb 4G2 as primary antibody. Error bars represent standard errors of two independent experiments in duplicates. *ns*: not significant, \*\*\*: *p* value < 0.001

#### *3.4. Mutations at E-152*/*156*/*158 Residues Have No E*ff*ect on ZIKV-Induced Cell Death or Interferon Pathways*

To determine whether differences described with the mutant viruses were associated with specific host-cell responses, we first analysed virus-induced cell death at 48 h and 72 h post-infection. No difference in cytotoxicity measured by LDH release was observed between wild-type and mutant viruses (MR766 and BR15) (Figure 3, panels a and b). We then took advantage of the properties of A549-Dual™ cells to test whether mutant viruses can trigger different host cell innate immunity. A549-Dual™ cells were derived from A549 cells by stable integration of two reporter genes: *SEAP* gene (Secreted Embryonic Alkaline Phosphatase) and *Lucia* luciferase gene under the respective transcriptional control of an IFN-β minimal promoter, which is fused to NF-κB binding sites or an ISG54 minimal promoter in conjunction with interferon-sensitive response elements. We examined possible activation of the IRF pathway by monitoring production of *Lucia* luciferase at 48 dpi and 72 hpi. Similar responses were observed in both wild-type and mutant clones (Figure 3, panels c and d). The NF-κB pathway was not investigated, as we showed previously that this pathway is not activated upon ZIKV infection [23]. These results indicate that differences in the mutant virus properties could not be explained by specific host-cell responses, which are consistent with our previous observations suggesting a link between host-cell responses and ZIKV nonstructural proteins [23].

**Figure 3.** Analysis of infection-induced cell death and immune responses. A549-Dual™ were infected with BR15E-MUT and MR766E-MUT and parental clones at MOI of 1. In (**a**) and (**b**), LDH activity was measured at 48 and 72 hpi respectively. Values represents mean and standard errors of two independent experiments in triplicates. In (**c**) and (**d**), analysis of IRF pathway activation in response to viral infection. Activity of secreted *Lucia* luciferase was measured using QUANTI-Luc substrate at 48 and 72 hpi. Results are expressed as raw data of luminescence arbitrary units. Error bars represent standard errors of two independent experiments in triplicates. *ns*: not significant.

#### *3.5. E-152*/*156*/*158 Residues from BR15 Facilitate Viral Fusion*

We showed earlier that E-152/156/158 residues from BR15 have a growth advantage without apparent association with cellular attachment (Figure 2) or specific host-cell responses (Figure 3). We then studied viral fusion to test whether it could explain the observed growth advantage. Viral fusion of flaviviruses is commonly triggered from endosomes upon low-pH by a series of molecular changes within the E protein, resulting in the release of the nucleocapsid into cell cytoplasm. Chloroquine, a 4-aminoquinoline, is a weak base that inhibits endosome acidification and consequently restricts viral replication of many viruses through inhibition of pH-dependent steps. Recently, chloroquine was shown to inhibit Zika virus infection in different cellular models [36,37]. As BR15 and BR15E-MUT

showed significant differences in the percentage of infected cells (Figure 2f), we decided to focus on these two molecular clones for the following virus fusion experiments. We treated A549-Dual™ cells infected with BR15 or BR15E-MUT with 100 μM of chloroquine 1 hpi for 2 h and then cells were moved back to regular medium. Intracellular viral RNA was quantified 30 hpi by RT-qPCR. BR15E-MUT fusion was significantly restricted by chloroquine treatment compared to that of BR15 (Figure 4). These data sugges<sup>t</sup> that E-152/156/158 residues from BR15 favour viral fusion with host-cell membranes.

**Figure 4.** Viral fusion in A549-Dual™ cells. Pre-chilled cells were incubated at 4 ◦C with ZIKV at MOI of 1. After 1-h incubation, cells were shifted to 37 ◦C. Chloroquine was then added to the culture medium. Viral RNA was measured by RT-qPCR 30 h at 37 ◦C. Error bars represent standard errors of two independent experiments. \*: *p* value < 0.1

#### *3.6. E-152*/*156*/*158 Residues from BR15 Favour Conformational Changes within the Fusion Loop*

As virus fusion with host cells is highly dependent on conformational changes of the E protein triggered at low-pH, we hypothesize that the reduced fusion we observed with the mutant molecular clone BR15E-MUT bearing E-152/156/158 residues from MR766 could be the consequence of conformational differences between E proteins of the two molecular clones. In order to test this hypothesis, BR15, BR15E-MUT and MR766 sequences coding for TMD2-prM/E were codon-optimised for expression in mammalian cells and cloned into a pcDNA3.1 vector. HEK-293T cells were transfected with different plasmids and positive cells were selected with antibiotics. The resulting stable cell lines were fractionated to evaluate the capacity of recombinant E proteins to fold properly, insolubility been a hallmark of misfolded proteins [32,35]. Resulting fractions were subjected to an immunoblot analysis. We first used a rat antibody developed in-house, specifically raised against E protein domain EDIII [28]. Figure 5a revealed that BR15E-MUT and MR766 TMD2-prM/E overexpression resulted in a greater E protein propensity to accumulate in insoluble fractions than that of BR15 overexpression, as shown by inversion of soluble/insoluble ratios. Interestingly, differences observed between BR15 and mutant BR15 TMD2-prM/E suggested that ZIKV E proteins bear different conformations that only depend on E-152/156/158 residues. To verify these observations, the same samples were immunoblotted using a 4G2 monoclonal antibody, which recognises a highly conserved fusion loop sequence of most flaviviruses. As shown in Figure 5b, 4G2 monoclonal antibody strongly reacts against E protein from BR15, whereas we could barely detect any signal with the two E proteins bearing E-152/156/158 residues from MR766. Preliminary in silico modelling of ZIKV E proteins sugges<sup>t</sup> that changes in the glycosylation motif could affect structure of the glycosylation and fusion loops as well as interactions with surrounding residues (Figure S2) and surface hydrophobicity (not shown). These data confirm that E-152/156/158 residues in the EDI domain support conformational changes on the ZIKV E protein, which could be detected in the fusion loop of EDII domain. Finally, these results sugges<sup>t</sup> that conformational changes occur in BR15 E protein upon mutation of E-152/156/158 residues.

**Figure 5.** Conformational changes induced by residues E-152/156/158 of ZIKV E protein. In (**a**) and (**b**), HEK-293T cells were transfected with TMD2-prM/E constructs and antibiotics that were selected to raise stable cell lines. Cells were harvested and proteins extracts subjected to a fractionation. Protein fractions were immunoblotted with anti-ZIKV EDIII (**a**) or anti-E 4G2 (**b**) antibodies. S, soluble proteins; I, insoluble proteins. Band intensities were determined with ImageJ software and S/I ratios were calculated. Apparent discrepancies with cytometry experiments (Figure 2) regarding antibody reactivity are explained by experimental and recombinant protein overexpression versus viral infection conditions.
