**4. Discussion**

ZIKV has attracted tremendous attention in the field of medically important flaviviruses, which are responsible of world epidemics that are di fficult to control [29]. To date, there is no e ffective treatment against this emerging pathogen despite significant e fforts made to succeed in providing an effective vaccine. The characterization of the interaction modalities between the pathogen and host cells are important in order to better understand the strategies adopted by the virus to be e ffective in its replication and dispersal of its progeny. Among the responses of infected cells, a high priority must be given to virally induced apoptosis since its completion can significantly interfere with the virus's multiplication cycle and hinder its spread [30]. Moreover, crosstalk between innate immune signaling and cell death pathways and how the viruses are able to manipulate each other are essential for viral clearance or persistence and for the global outcomes of the infection. Many studies support that ZIKV infection induces apoptosis in vivo as well as in vitro [31–33] However, several of them also assert that ZIKV induced cell death could be delayed [17,18]. We therefore wanted to define whether this capacity was specific to the strain of ZIKV responsible for the current epidemic and whether ZIKV was able to interfere with the induction of apoptosis.

Using a molecular clone of the BeH819015 from the Brazilian 2015 outbreak (BR15MC) and its comparison to the clinical isolate from the French Polynesian 2013 outbreak (PF13), our work has highlighted that ZIKV strains from the actual epidemic Asian lineage are particularly inclined to delay the onset of apoptosis in infected human epithelial cells (A549) as well as in human brain glioblastoma–astrocytoma cells (U251 MG; Figure 1, Figure 2 and Figure S3). These viruses are also characterized by a rather slow viral growth compared to the molecular clone of the historical strain of ZIKV MR766 from the African lineage [21]. Infection with MR766MC was marked by a higher cytotoxicity over the duration of infection, with 10% of infected cells showing signs of entry into apoptosis as early as 24 h, whereas epidemic strains had no signs of mortality before 48 h. However, despite this behavior of MR766MC, which seems more aggressive, it must be admitted that the mortality rate among infected cells remained limited until 48 h post infection (Figure 2 and Figure S4). We proposed from these results that both viruses had the ability to interfere with the onset of apoptosis even though faster and more e ffective viral growth in the early hours of infection for MR766MC than for BR15MC had resulted in faster and more cytopathic e ffects. In both cases, the maximum mortality was only recorded when the virion production had reached its highest titer. This observation suggests a manipulation of apoptosis orchestrated by ZIKV in order to give it enough time to complete its entire production cycle. This phenomenon has already been described for other flaviviruses such as DENV, JEV and WNV as they can delay apoptosis through activation of the phosphatidylinositol 3-kinase (PI3K) and Akt pathway [16,34].

In a rather unexpected way, our work mainly shows that in the early stages of infection, ZIKV infection provides a solid protection against an exogenous induced cell death. We provided supporting evidence that protection is acquired both with the Asian epidemic strains and with the African strain (Figure 3, Figure 4, Figures S4 and S5). This protection is e ffective against apoptosis mediated by an extrinsic death inducer (TNF α) as well as by an intrinsic signal (provided by the action of etoposide or blasticidin). Resistance to these induction modes has also been found to characterize cells expressing replicons, either HEK 293 cells stably expressing a MR766 replicon or A549 cells transiently expressing BR15 replicons or MR766 replicons (Figure 6, Figure 7 and Figure S5). The data obtained with the use of these replicons are in support of a greater protection granted by BR15. This is consistent with the data obtained with the whole virus that is responsible for the longest delay in apoptosis entry. Cell death inhibition ability acquired with the ZIKV replicons would imply that the single presence of a viral RNA leading to the production of the NS proteins and allowing its self-replication is the driving force behind the protection acquired against apoptosis.

This discovery is rather unusual when one considers the literature mainly in favor of pro apoptotic functions for the NS proteins [35]. However, studies have also shown that some NS may help ZIKV to evade antiviral immunity and cell death. NS2B in particular was proposed to be responsible for blocking (RIG-I)-like receptors triggered apoptotic cell death [36]. A thorough identification of the viral protein responsible for an inhibitory effect needs to be confirmed and further investigated. While our work suggests a possible role for NS proteins in the control of apoptosis, we cannot rule out the possibility that structural proteins may also act. Anti-apoptotic activities have been previously reported in relation to the capsid [16,18]. It should be noted that in the construct we used to generate the replicons, the polyprotein produced retains the first 33 amino acids of the capsid. It would be interesting to see if this part of the structural protein, released after the cleavage of the GFP has a role in protection. We could not also exclude a role of the viral RNA by itself. Recent work has also investigated the effects of viral RNAs, as flaviviruses are known to produce multiple small RNAs that may have interference activities in the cell physiology. It was recently described that recent epidemic Asian lineage display more negative-strand replicative intermediates than the historical African strain [37]. This important production could be a key element in the search for which viral determinants are crucial in the control of apoptosis. We also know how important are the viral genomic 3 UTR regions and the Subgenomic Flavivirus RNA (sfRNA) produced, in the implementation of cellular responses to infection [38]. It remains to be discovered which factor associated with the replication process of ZIKV viral RNA is involved in the protection mechanism. What we already know from our study is that this mechanism required the Bcl-2 family protein as ABT-737 abrogates the protection acquired against apoptosis by ZIKV infection (Figure 8). The anti-apoptotic capacity of the pro-survival Bcl-2 proteins is known to depend mainly on their ability to sequester pro-apoptotic proteins by binding their BH3 domains. A decrease in Bcl-2 leads to the disruption of associated pro-survival and pro-apoptotic Bcl-2 proteins and will promote apoptosis whereas overexpression of Bcl-2 will inhibit mitochondrial OMP [39,40]. A control of the stability and degradation of Bcl-2 is therefore a key in the subtle balance that takes place between pro and anti-apoptotic suits to determine the cell's fate [40]. A previous study showed the importance of Bcl-xL for cells survival, in deficient Bcl-2 cells during ZIKV infection [41], but here we could not exclude a role for Bcl-2 and/or Bcl-xL protein as our model expressed these two anti-apoptotic proteins. Based on the fact that Bcl-2 family protein was involved in ZIKV infection with a Bcl-2 protein level quantitatively maintained (Figure 8), we formulated the hypothesis that ZIKV might allow a stabilization of the Bcl-2 protein over time and inhibit MOMP formation (Figure 9). The mechanism by which the virus allows the Bcl-2 stabilization and blocks apoptosis needs further investigation.

**Figure 9.** Model depicting the protective action of ZIKV infection against apoptosis. Intrinsic or extrinsic activation of cell death occurs through the formation of mitochondrial outer membrane pore (MOMP) via the BAX/BAK complex. The anti-apoptotic family Bcl-2 members Bcl-2/Bcl-XL interfere with the complex formation by sequestering BAX via their BH3 domain. ZIKV replication interferes with BAX relocalization at the mitochondria. ABT-737, a BH3 mimetic, which inhibits Bcl-2/Bcl-XL, abrogates the inhibition induced by ZIKV. Indeed, the virus delays apoptosis during infection by modulating the homeostasis of Bcl-2/Bcl-XL.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/8/11/1338/s1, Figure S1. IF imaging to validate markers of apoptosis in A549 cells. Figure S2. Alphavirus RRV induces early and massive apoptosis compared to Zika virus. Figure S3. ZIKV-PF13 does not cause significant activation of apoptosis until late in infection in U251MG cells. Figure S4. ZIKV-MR766 does not cause significant activation of apoptosis until late in infection and ZIKV-MR766 is able to control cell death. Figure S5. A549 cells transiently expressing a ZIKV-MR766 replicon are protected against cell death by apoptosis.

**Author Contributions:** J.T., E.F., W.V., P.K.-T. designed the research J.T., E.F., W.V., P.K.-T. performed the research; J.T., E.F., P.D., W.V., P.K.-T. contributed new reagents/analytic tools; all authors analyzed the data; J.T., E.F., W.V., P.K.-T. wrote, revised and edited the manuscript.

**Funding:** This work was supported by the Federation BioST from Reunion Island University (Zikapone project), the ZIKAlert project (European Union-Région Réunion program under gran<sup>t</sup> agreemen<sup>t</sup> n◦ SYNERGY: RE0001902). E.F. holds a fellowship from the Regional Council of Reunion Island (European Union-Région Réunion program under gran<sup>t</sup> agreemen<sup>t</sup> n◦ SYNERGY: RE0012406). J.T. has a PhD degree scholarship from La Réunion Island University (Ecole Doctorale STS), funded by the French ministry MEESR.

**Acknowledgments:** We thank the members of PIMIT and DéTROI laboratories for helpful discussions. We are grateful to David A. Wilkinson for the critical review of the manuscript and Valérie Ferment-Van de Moortele for the English spelling. We also thank Valérie Choumet for providing us the rabbit anti-ZIKV EDIII antibodies. Servier Medical art provided the icons used in Figures 7 and 9.

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