**4. Discussion**

The role of structural proteins in determining infectivity of human cells to ZIKV infection has been previously reported [23]. Zika viruses of historical or epidemic strains display differences in their abilities to bind host cells leading to differences in cell susceptibility to infection. To further characterise biological properties of contemporary ZIKV strains, which have been associated with recent epidemics and severe forms of human disease, we investigated the role of E-152/156/158 residues of the envelope protein, residues located around a unique *<sup>N</sup>*-glycosylation site (E-154) in viral entry into host cells. Our mutagenesis data showed that E-152/156/158 residues are responsible for the differences we observed in the infectivity of the virus and progeny production kinetics, without affecting viral attachment and host-cell responses. Further characterisations identified the E protein conformational changes in the fusion loop, supported by E-152/156/158 residues, as a major event in virus fusion and release of viral RNA into cell cytoplasm.

The first step in viral entry pathway involves nonspecific viral binding to cellular attachment factors. Negatively charged glycosaminoglycans, which are abundantly expressed on numerous cell types, are considered as low-affinity attachment factors by flaviviruses. These interactions serve to concentrate viruses on the cell surface and are mediated by the EDIII domain of E proteins [38]. Our data demonstrate that E-152/156/158 residues of ZIKV E protein do not influence virus binding, which suggests that domain EDIII conformation is not strongly affected by E-152/156/158 residues. We conclude that this initial step of ZIKV entry into cells does not depend on E-152/156/158 residues of the E protein.

Viral particle internalisation could occur through distinct routes, including clathrin-mediated endocytosis or non-classical clathrin-independent endocytic pathways. These distinct entry modes depend not only on host cells but also on viral serotype or strain [39]. Despite these differences in the internalisation process, genome release into the cytoplasm always occurs through E protein-mediated membrane fusion [40]. The low-pH environment within endosomes triggers a series of molecular changes within flavivirus E protein resulting in fusion of viral membrane with endosomal membrane and subsequent release of the nucleocapsid into cell cytoplasm [13,14]. The initial step in membrane fusion involves protonation-dependent disruption of E protein rafts at viral surface, resulting in conformational changes and formation of a fusion pore from which the nucleocapsid is released into the cytosol. Evidences suggested that the glycan loop modulates the overall Flavivirus E protein conformation and, specifically, fusion loop exposure [41,42]. Sevvana and colleagues proposed that, given the close proximity of ZIKV glycan and fusion loops, any interaction of the glycan loop with a receptor on a host-cell surface might promote exposure of the fusion loop and facilitate formation of fusogenic trimers leading to membrane fusion [43]. Another study based on antibody neutralizing activities demonstrated that residues surrounding ZIKV E protein glycan regulate virus antigenicity [44]. In a previous study, we also evaluated immunogenicity of a chimeric viral clone ZIKBeHMR-2, in which the region encoding envelope proteins of MR766 African strain backbone was replaced with its counterpart from BeH819015 epidemic strain [45,46]. Amino-acid substitutions I152T, T156I, and H158Y were introduced in the glycan loop of the E protein, making chimeric ZIKBeHMR-2 a nonglycosylated virus. Those results sugges<sup>t</sup> that, rather than just determining the glycosylation, amino-acid residues at position 152, 156 and 158 play a pivotal role on accessibility of neutralizing antibody epitopes on mature virus particles. In this study, we demonstrated that changes in the glycan loop can modulate accessibility to the fusion loop. Although these observations were made independently of the presence of a glycan, it is conceivable that E protein *<sup>N</sup>*-glycosylation could provide another level of regulation on the access of the fusion loop. The role of *<sup>N</sup>*-glycosylation on ZIKV E protein has also been investigated using pseudoviral particles, showing that reduced infectivity was observed with mutant viral particles lacking the N-glycan [47]. Altogether, these studies sugges<sup>t</sup> that conformational changes induced at the glycan loop most probably modulate fusion loop exposure and subsequent fusion of viral and cellular membranes, which strongly supports our observations.

One interesting finding from this study is that chloroquine treatment results in less BR15E-MUT entries than it does for BR15. This finding suggests that E-152/156/158 residues of BR15 increase pH sensitivity of E protein. To generate BR15E-MUT, its sequence was modified so that the coding region of the E protein IVNDTGH (amino acids 152 to 158) in BR15 was replaced with TVNDIGY motif from MR766, meaning that not only *<sup>N</sup>*-glycosylation motif was abrogated but also that histidine E-158 was changed into a tyrosine. Histidine residues have been described as pH sensors in flavivirus membrane fusion [48]. Fusion initiation is crucially dependent on protonation of conserved histidine residues at the interface between domains EDI and EDIII of E protein, leading to dissolution of domain interactions and to fusion peptide exposure. Given the fusion di fferences we observed between wild-type and mutant BR15 molecular clones, further investigations on histidine E-158 protonation are required to determine its exact contribution to membrane fusion.

Finally, our analysis of virus inocula generated in Vero cells showed di fferences in particle-to-PFU ratios indicating that E-152/156/158 residues of BR15 E protein facilitate release of more infectious particles. In addition, we demonstrated with recombinant proteins that E-152/156/158 residues of epidemic ZIKV E protein also facilitate production of more soluble proteins. These results are supported by works of Mossenta and colleagues [47]. Whether these observations are due to conformational changes occurring during virion production remains undetermined. However, in a study on flavivirus cross-reactive epitopes, Crill and Chang suggested that close packing of fusion peptide against its subunit partner and glycan on the upper surface protects the fusion loop from irreversible pH-induced conformational changes during maturation and secretion [49]. All these observations sugges<sup>t</sup> that E-152/156/158 residues of epidemic ZIKV E protein could also be an advantage during virion maturation process.

Altogether, our data indicate that the envelope residues E152/156/158 of Zika virus influence early stages of virus infection in human cells. This study highlights the importance of E-152/156/158 residues in ZIKV biology and specifically in their roles in supporting viral fusion. These new findings could potentially help to design innovative strategies for future ZIKV infection control.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/8/11/1444/s1, Figure S1: Tunicamycin treatment of Vero cells infected with ZIKV mutant molecular clones, Figure S2: structures of ZIKV E wild-type and mutant proteins.

**Author Contributions:** Conceptualization, P.D. and G.G.; Funding acquisition, P.D.; Investigation, S.B., W.V., E.F. and G.G.; Visualization, S.B., W.V., E.F. and G.L.; Writing—original draft, G.G.; Writing—review & editing, S.B., W.V., E.F., P.D., R.Y.Z. and G.G.

**Funding:** This work was supported by ZIKAlert project (POE FEDER 2014–2020 Ile de la Réunion action 1.05 under gran<sup>t</sup> agreemen<sup>t</sup> n SYNERGY:RE0001902). G.L. and R.Y.Z. were supported in part by funding from the National Institute of Health (NIH R21 AI129369) and an intramural fund from the University of Maryland Medical Center (to R.Y.Z.). E.F. received funding from FEDER <sup>R</sup>égion (ZIKALIVax 2.0 program, N SYNERGIE RE0012406) and La Reunion University. S.B. has a PhD degree scholarship from La Réunion Island University (École Doctorale STS), funded by the French ministry, MESRI.

**Acknowledgments:** We gratefully acknowledge Pascale Krejbich-Trotot and Chaker El Kalamouni for helpful discussions. We also thank Hervé Pascalis for his kind help on structure comparison.

**Conflicts of Interest:** The authors declare no conflict of interest.
