**3. Results**

#### *3.1. ZIKV Does Not Trigger Apoptosis Until the Release of Most of its Progeny*

Our research team had previously demonstrated that a South Pacific epidemic clinical isolate of ZIKV (PF13-25013-18) was able to infect A549 epithelial cells. These cells are particularly permissive to the virus and therefore constitute a suitable model for studying in cellulo host-virus interactions [17]. In order to characterize the cellular death profile that accompanies ZIKV infection more precisely, we conducted a study of the cytopathic effects induced with the viral molecular clone of the epidemic strain from Asian lineage, BeH819015 isolated in Brazil in 2015 (BR15MC) [21]. We infected A549 cells with BR15MC at a multiplicity of infection (MOI) of 1 and followed for 3 days, the characteristics of the viro-induced cell death (Figure 1). We further monitored the induction and execution of apoptosis specifically in infected cells to compare them with the results of viral production (Figure 2).

The measurement of LDH activity in infected cell culture supernatants, which results from a loss of cell integrity mainly reflecting secondary necrosis or estimation of cell viability by measurement of mitochondrial activity by MTT assay, revealed that cell mortality was detected at 48 h post infection (hpi) to reach a high level 72 hpi (Figure 1A,B). At 24 hpi there was no detectable sign of cell death by apoptosis when we looked at the activity or presence of cleaved caspase 3 (Figures 1C and 2C). Relocalization of the pro-apoptotic factor BAX to mitochondria, an early marker of apoptosis (Supplemental Figure S1A), was only observed at 48 hpi (Figure 2A,B) and only occurred in approximately 10% of the cells that were immunolabeled with an antibody directed against the viral envelope protein E (ZIKV-E; Figure 2B). Examination of another signal of engagemen<sup>t</sup> in apoptosis, namely the presence of cytosolic cytochrome-c (Figure 2A, magnified image with (b) arrow on the cyt-c immunodetection panel), led to the same observation (Figure 2C). It should be noted that only some of the cells immunolabeled for ZIKV-E had apoptotic characteristics (Figure 2A). The percentage of uninfected cells with signs of death by apoptosis was always equivalent to that observed in the controlled cell cultures over time (Figure S1B). Analysis of apoptosis execution, such as the measurement of caspase 3/7 activity (Figure 1C), immuno-detection in infected cells of cleaved activated caspase 3 (Figure 2A,D) or by Western blotting (WB; Figure 3C) supported the delay in cell death with respect to the course of viral multiplication. Moreover, significant signs of engagemen<sup>t</sup> in apoptosis occurred when the released viral progeny have already reached their maximum (Figure 2E). A late and low proportion of ZIKV-infected cells engaged in apoptosis were also retrieved during the infection of Vero cells (Figure S2). This apoptosis kinetics differed significantly from the one induced during the Ross River alphavirus (RRV) infection. Thus during RRV infection the maximum of caspase 3 activity was observed at 24 hpi and was at least 10 fold higher than the one measured in ZIKV-infected cells. These observations sugges<sup>t</sup> a ZIKV-infection specific feature of the viro-induced apoptosis, together with a regulatory mechanism implemented by the ZIKV to delay apoptosis.

**Figure 1.** Cell death during a Zika virus (ZIKV) infection of A549 cells. A549 cells were infected with BR15MC at a multiplicity of infection (MOI) of 1. LDH activity was measured in cell supernatant of mock infected cells, BR15 infected cells and in cells treated with triton X-100 as a positive control of total cell lysis value (grey bar) and was normalized to mock infected cells value (**A**), cell viability (MTT assay) (**B**) and caspase 3/7 activity (**C**) were measured at 24, 48 and 72 h post infection (hpi) and normalized to mock infected cells values. Values represent the mean and standard deviation of three independent experiments. Data were analyzed by a one-way ANOVA test with post-hoc Tukey's test (\* *p* < 0.05; \*\* *p* < 0.01; \*\*\*\* *p* < 0.0001; ns = not significant).

To rule out that delayed apoptosis in infected cells was not a feature of the epithelial cell line A549 or Vero cells, we verified the respective courses of infection and cell death in other cell models. In the U251MG line of human brain glioblastoma-astrocytoma cells, infection kinetics was accompanied by a complete absence of apoptosis within the first 48 h of infection before the maximum of viral progeny production (Figure S3). Thus we confirm that death by apoptosis induced by our BR15MC viral molecular clone as for the Asian epidemic clinical isolate PF13 occurs late and is relatively moderate compared to the kinetics of induced viral death that can be observed in the case of infection by other arboviruses like alphaviruses (Figure S2 and [12]).

**Figure 2.** BR15MC does not cause significant activation of apoptosis until late in infection. A549 cells were infected with BR15MC at MOI of 1. (**A**) Cells were immunostained for active mitochondrial BAX, cytochrome c (Cyt c), ZIKV E and cleaved caspase 3 (CASP 3), 48 hpi. The white scale bar represents 10 μm. Right panel series show magnified details of selected cells from the ×200 microscopic field (white square). Arrows indicate (a): an infected cell (stained for ZIKV E) and (b): an infected and apoptotic cell (stained for ZIKV E and with mitochondrial localization of BAX or Cytosolic Cyt c or cleaved CASP3. (**B**) Percentage of A549 infected cells co-immunolabeled for ZIKV E and for active mitochondrial BAX, among the ZIKV E positive cells were determined at 24, 48 and 72 hpi. (**C**) Percentage of A549 infected cells co-immunolabeled for ZIKV E and for cytosolic Cyt c, among the ZIKV E positive cells were determined at 24, 48 and 72 hpi. (**D**) Percentage of A549 infected cells immunostained with anti-cleaved CASP 3 antibody among the ZIKV E positive cells were followed at 24, 48 and 72 hpi. (**E**) The infectious viral particles were collected from infected cell culture supernatants at 24, 48, 72 and 96 hpi and titrated. Values represent the mean and standard deviation of three independent experiments.

#### *3.2. ZIKV Infected Cells Display Resistance to Apoptosis Inducers*

Apoptosis can be initiated by extrinsic or intrinsic pathways, the former being mediated by death receptors located on the cell surface, the latter being driven by various cellular stresses such as DNA damage. Activation of apoptosis-initiating caspases 8 or 9 results in mitochondrial outer membrane

permeabilization (MOMP) via oligomerization and insertion of the proapoptotic factors BAX/BAK in the mitochondria of type II cells such as epithelial cells [26].

The late onset of apoptosis in infected cells led us to postulate that ZIKV may modulate the apoptotic response of the cell, delaying it through transient inhibition. To test this hypothesis, we investigated whether ZIKV could counteract the effect of death inducers added during the infection time course. We induced apoptosis through extrinsic and intrinsic pathways and tested the addition of the inducer at 2 h prior-to and 2 h post infection (Figures 3 and 4).

#### 3.2.1. ZIKV Provides Protection Against Death Receptor Mediated Cell Death

To drive an extrinsic apoptosis, we induced the TNF-receptor using its ligand, TNF-alpha (TNFα), inhibiting the cytoprotective NFkb response by the addition of the translation inhibitor cycloheximide (CHX). Between 6 and 8 h of treatment leads to an estimated 20% cell mortality in A549 cells, when counting BAX positive cells (Figure 3A). No variation in the percentage of cells engaged in early apoptosis (BAX+) 8 h after the onset of treatment was observed when TNFα/CHX was added 2 h prior to ZIKV infection.

**Figure 3.** BR15MC provides a protection against extrinsically induced cell death. A549 cells were infected with BR15MC at MOI of 1 for 8 h and treated with TNFα and cycloheximide (TNFα/CHX) 2 h post infection (2 hpi) or 2 h before infection (2 hbi). (**A**) The percentage of A549 cells immunolabeled with anti-BAX antibody was enumerated from images representative of immunofluorescence observations (panels on the right, white scale bar: 10μm). (**B**) Caspase 3/7 activity was measured after TNFα/CHX treatment 2 hpi and normalized to mock treated and infected cells. (**C**) Immunoblot of active caspase 3 during TNFα treatment 2 hpi and infection time course with BR15MC. Active caspase 3 band intensity was normalized with GAPDH. Western blotting (WB) is representative of three independent experiments. Values represent the mean and standard deviation of three independent experiments. Data were analyzed by a one-way ANOVA test with post-hoc Tukey's test (\*\* *p* < 0.01; \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001, ns = not significant).

Conversely, if TNF α/CHX was added 2 h post infection, a drastic and significant drop in the percentage of cells engaged in apoptosis was observed. This drop was corroborated by the measure of caspase 3 activity (Figure 3B). It is worth recalling that in this lapse of time, virally induced apoptosis is undetectable and therefore is unlikely to interfere with the quantification of cell death induced by the action of TNF α/CHX. Moreover, in Western blots, detectable levels of the active form of caspase 3 in BR15MC-infected cells were seen to be reduced by around 75% after treatment with TNF α/CHX (Figure 3C).

#### 3.2.2. ZIKV Provides Protection Against Intrinsically Induced Cell Death

Intrinsically induced apoptosis was stimulated by chemical DNA damage. To do this, we used the genotoxic agen<sup>t</sup> etoposide, a topoisomerase-II inhibitor, which causes chromosome breaks during DNA replication. Overnight treatment with etoposide (16–18 h) resulted in a percentage of cells with a mitochondrial BAX among the remaining adherent cells that was between 7% and 12%, depending on the experiment (Figure 4). Similar to the induction of cell death by TNF α/CHX, although the e ffects are slightly more modest, apoptosis produced in A549 cells after 16 h of etoposide treatment was significantly reduced in the case where ZIKV was added 2 h post-infection (Figure 4A). In addition, both BR15, the molecular clone of ZIKV (Figure 4A) and a clinical isolate of the epidemic strain PF13 (Figure 4B) were capable to interfere with apoptosis induction by etoposide.

**Figure 4.** ZIKV provides a protection against intrinsically induced cell death by etoposide. ( **A**) A549 cells were infected with BR15MC at MOI of 1 for 8 h and treated with etoposide 2 h before infection (2 hbi) or 2 h post infection (2 hpi). The percentage of A549 cells immunostained with anti-BAX antibody was followed. (**B**) A549 cells were infected with PF13 at MOI of 1 for 8 h and treated with etoposide for 16 h, 2 hpi. The percentage of A549 cells immunostained with anti-BAX antibody was followed. Values represent the mean and standard deviation of three independent experiments Data were analyzed by a one-way ANOVA test with post-hoc Tukey's test (\*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001, ns = not significant).

We also tested the e ffect of the cell line used for infection by repeating the procedure in Vero cells, in this instance using blasticidin to induce intrinsic apoptosis in response to the inhibition of translation instead of DNA damage with etoposide. Following cell death by MTT assay and caspase 3/7 activity again suggested that ZIKV infection represses apoptosis, regardless of cell type and death inducers (Figure 5A,B).

**Figure 5.** ZIKV provides a protection against intrinsically induced cell death by blasticidin. Vero cells were infected with ZIKV-PF13 at MOI of 1 for 8 h and treated with blasticidin for 16 h, followed by viability assay (MTT) ( **A**) or caspase 3/7 activity (**B**). Values represent the mean and standard deviation of three independent experiments. Values represent the mean and standard deviation of three independent experiments Data were analyzed by a one-way ANOVA test with post-hoc Tukey's test (\*\* *p* < 0.01; \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001, ns = not significant).

In order to ensure that the ZIKV mediated repression of apoptosis was conserved between ZIKV strains, we looked at the e ffect of apoptosis induction with etoposide after infection with the epidemic clinical isolate PF13 (ZIKV-PF13) but also with a molecular clone of historical strain from African lineage, MR766-NIID isolated in Uganda in 1947 (MR766MC; Supplemental Figure S4). ZIKV infection resulted in reduced apoptosis for all tested strains. Convergent measurements of several parameters that establish the death rate and degree of protection confirm that ZIKV interferes with the achievement of apoptosis in response to a death signal.

## *3.3. Apoptosis is Repressed Through ZIKV Replication*

Our data sugges<sup>t</sup> that protection against an exogenous induced apoptosis is acquired once the virus has entered the cells and started its multiplication cycle. In order to determine the contribution of the replicative process in the protective mechanism, we exploited the "replicon" systems.

3.3.1. Cells Stably Expressing a ZIKV Replicon are Protected from Intrinsically and Extrinsically Mediated Apoptosis

Since both epidemic Asian (BR15MC) and historical African (MR766MC) strains of ZIKV were found to be able to control apoptosis, we investigated the role of viral replication in the mechanism of apoptosis repression using a previously established MR766 replicon system in HEK-293 cells [25]. The Rep ZIKV-GFP cells have a self-replicating RNA encoding the viral NS proteins from MR766-NIID, a puromycin resistance gene to facilitate selection and GFP as a reporter. HEK-293 cells stably transfected with a plasmid encoding a GFP reporter gene and puromycin resistance without any viral material was used as a control (HEK CT; Figure 6). Apoptosis was induced either with TNF α/CHX, with etoposide or with blasticidin. We monitored apoptosis and in particular measured caspase 3/7 activity 6h post addition of TNF α/CHX or 16 h post addition of etoposide or blasticidin. We compared the values related to caspase 3/7 activity in Rep ZIKV-GFP with those obtained with the control cells (HEK CT; Figure 6A).

**Figure 6.** ZIKV replicon-expressing cells are protected from apoptosis. HEK-293 cells stably expressing a ZIKV replicon (Rep ZIKV-GFP) are protected from intrinsically and extrinsically mediated apoptosis. HEK-293 cells with Rep ZIKV-GFP were treated with TNFα and CHX for 6 h, or treated with etoposide 10 μM or blasticidin 25 μg/mL−<sup>1</sup> for 16 h and analyzed for: caspase 3/7 activity (**A**), cell viability (MTT assay) (**B**) and released LDH activity (**C**). Values represent the mean and standard deviation obtained with three different clones of Rep ZIKV-GFP. Values represent the mean and standard deviation of three independent experiments. Data were analyzed by a one-way ANOVA test with post-hoc Tukey's test (\*\* *p* < 0.01; \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001, ns = not significant).

Monitoring apoptosis by caspase 3/7 activity (Figure 6A), together with the measure of cell viability (Figure 6B) and released LDH activity (Figure 6C) demonstrated that ZIKV replicon resulted in a significant reduction in the indicators of cell death. An inhibition of apoptosis was not observed in the case of the cell control, selected for GFP and puromycin resistance (HEK CT) as well as in untransfected HEK-293 cells (data not shown). It can be concluded that resistance to puromycin or the presence of a GFP encoding gene were not responsible for a protective effect.

It can therefore be proposed that the presence of an autonomous replication of a viral RNA associated with the expression of ZIKV NS proteins makes cells resistant to several extrinsic and intrinsic apoptotic inducers.

3.3.2. A549 Cells Transiently Expressing a ZIKV Replicon are Protected from Different Intrinsically Induced Apoptosis

To address a protective effect of the viral replication in a system consistent with the one in which the effect was revealed through infection, we adapted the ISA method to obtain A549 cells transiently expressing a ZIKV replicon with the non-structural sequences from BeH819015 (REP BR15). We used A549 cells transfected with an incomplete set of amplicons as a negative control (REP-NEG; Figure 7A). Apoptosis was induced 48 h after amplicon transfection either by etoposide treatment (Figure 7B) or by DNA damage through exposure to UV light (Figure 7C).

**Figure 7.** A549 cells transiently expressing a ZIKV replicon are protected against cell death by apoptosis. A549 cells were transfected with ZIKV amplicons (Z genomic overlapping fragments) for production of REP BR15 (Z1, 2, 3 and 4) and REP NEG (Z2, 3 and 4) or with pEGFP-N1 (**A**). At 48 h after transfection A549 cells were treated with etoposide for 16 h (**B**) or UV at 400 mJ (**C**). The percentage of A549 cells immunostained with anti-BAX antibody or anti-cleaved CASP 3 was monitored. Values represent the mean and standard deviation of three independent experiments. Data were analyzed by a one-way ANOVA test with post-hoc Tukey's test (\* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001, ns = not significant).

In cells treated with etoposide, REP BR15 expressing cells showed approximately half the number of apoptotic cells of the replicon control (REP NEG; Figure 7B). The percentage of dying cells was even lower in cells expressing REP BR15 after UV treatment (Figure 7C). Thus, REP BR15 was able to confer resistance to apoptosis. This resistance was also observed in an experiment conducted with A549 cells transiently expressing a ZIKV replicon from the MR766 strain of ZIKV (Figure S5). All together these results sugges<sup>t</sup> that replication of the ZIKV was capable of inhibiting apoptosis.

#### *3.4. ZIKV Promotes an Anti-Apoptotic Prevailing Status in Infected Cells Through the Bcl-2 Family Protein*

The long delay in the onset of viral apoptosis induced by ZIKV infection, particularly in the case of Asian viral strains responsible for current epidemics, and the demonstration that, when the viral RNA is present and replicating there is an inhibition of apoptotic induction, sugges<sup>t</sup> that cells have acquired with the virus a status in which anti-apoptotic activity prevails. The protective effect acquired with ZIKV could be at the level of convergence of the intrinsic and extrinsic pathways. As we followed BAX relocalization we could argue in favor of protection around the mitochondria events and the control of the outer membrane permeabilization (OMP). A prominent anti-apoptotic factor involved in the regulation of early apoptosis, by operating mainly at the mitochondrial level for the control of its permeabilization is Bcl-2 and the related Bcl-XL protein [27]. To identify to which extent these anti-apoptotic factors play a role in the protection provided by ZIKV, we examined the effect of ABT-737 on cell death outcomes, with or without BR15MC (Figure 8A,B). ABT-737 is a BH3 mimetic molecule that can bind to the hydrophobic groove of the members of the anti-apoptotic Bcl-2 protein family, Bcl-2 and Bcl-xL, and therefore inhibits their activity by shifting oligomerization mechanisms in favor of BAX/BAK dimerization [28].

**Figure 8.** Inhibition of anti-apoptotic Bcl-2 family proteins abrogates the protection mediated by ZIKV. A549 cells were infected with ZIKV at MOI of 1. TNFα and CHX were added 2 h post infection for 6 h with or without ABT-737. A549 cells were immunostained with an anti-BAX antibody (representative images in panels on the right, white scale bar: 10μm) (**A**) and caspase 3/7 activity was followed after treatment (**B**). A549 cells were infected with ZIKV at MOI of 1 and level of Bcl-2 was followed by western blot 24 h post infection (**C**). Values represent the mean and standard deviation of three independent experiments. Data were analyzed by a one-way ANOVA test with post-hoc Tukey's test (\* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001, ns = not significant).

When inducing apoptosis with TNFα/CHX, addition of ABT-737 restored the percentage of ZIKV-infected cells with mitochondrial BAX (Figure 8A) and caspase 3/7 activity (Figure 8B) to levels that were similar to cells that were not infected with ZIKV.

These observations sugges<sup>t</sup> that ABT-737 has counteracted the protective e ffect acquired with the virus. It can legitimately be deduced that the viro-induced protective e ffect probably depends on the anti-apoptotic activity of Bcl-2 family proteins. When Bcl-2/Bcl-XL is inhibited, ZIKV no longer allows a quantitative reduction of apoptosis.

As we used cycloheximide or blasticidin in our assays, we could therefore assume that the mechanism implemented by ZIKV did not involve a "de novo" synthesis of proteins. Based on these remarks, we formulated the hypothesis that ZIKV might allow stabilization of the Bcl-2 protein over time. We performed a western blot assay to measure Bcl-2 protein level upon ZIKV infection. Upon ZIKV infection, we observed an increase of Bcl-2 at the protein level (Figure 8C). This stabilization of Bcl-2 could support the anti-apoptotic capacity of ZIKV.
