*2.2. Extrinsic Apoptosis*

"Extrinsic apoptosis is a type of RCD initiated by perturbations of the extracellular microenvironment that are detected by plasma membrane receptors, propagated by CASP8 (with the optional involvement of MOMP), and precipitated by executioner caspases, mainly CASP3" (NCCD [19]).

Extrinsic apoptosis is mainly carried out by the activation of two main receptors types: death receptors and dependence receptors.

Dependence receptors are a functional family of around 20 receptors, characterized by the induction of a positive signal when bound by their ligand (survival, proliferation, differentiation, etc.) while they activate RCD in the absence of the ligand. Among them can be found netrin-1 receptors (deleted in colorectal carcinoma (DCC) [60], uncoordinated 5 homologs (UNC5Hs, UNC5H1,2,3,4 also called UNC5A,B,C,D) [61], the neogenin receptor [62], the low affinity neurotrophin receptor, p75 neurotrophin receptor (p75NTR) [63], and receptors with tyrosine kinase activity (e.g., rearranged during transfection (RET) [64], tropomyosin receptor kinase A and C (TrkA and TrkC) [65], and c-kit (CD117) [66]). Their physiological role is mainly cell guidance and they are mostly involved in tumor progression when dysregulated [67]. Even if Netrin-1 plays a role in inflammation regulation [68–71], dependence

receptors have not been involved, so far, in antiviral response [67]. As such, they will not be described in this review.

Death receptors include Fas (CD95, APO-1), the tumor necrosis factor receptor super family TNFRSF1A (TNFR1), 10a (TNF-related apoptosis-inducing ligand receptor TRAILR1, DR4), and 10b (TRAILR2, DR5) [72,73]. The general mechanism is that ligand binding induces the receptors' oligomerization and the subsequent recruitment through their death domains (DDs), of adapter proteins in the intra-cellular side, to form a "death-inducing signaling complex" (DISC).

Fas ligand or TRAIL binding drives the oligomerization of their receptors, the recruitment of Fas-associated protein with death domain (FADD) through the DD and the subsequent formation of the DISC through interaction with CASP8 via death e ffector domain (DED) and di fferent isoforms of cFlip [74,75].

Bound TNFR1 interacts with TNFR1-dssociated death domain protein (TRADD), through its DD, which enables the formation of "Complex I". The subsequent formations of "Complex II" ("IIa" or "IIb") operate as molecular platforms to regulate the activation and functions of CASP8 (or CASP10, in some cases) [76,77]. CASP8 activation leads to RCD following two di fferent pathways. In Type I cells, CASP8 directly activates CASP3 and -7 thus inducing the execution of the apoptotic pathway. In Type II cells, where CASP3 and -7 are sequestrated by XIAP, extrinsic apoptosis occurs through the cleavage by CASP8 of Bid, a BH3-only protein, and the release of its truncated form, t-Bid. t-Bid acts as an activator on Bax and Bak to provoke MOMP and the subsequent CASP9-dependent RCD, described above in intrinsic apoptosis [78].

CASP8 activation is the key process of extrinsic apoptosis; its regulation is complex and also involved in inflammation and antiviral response [77]. cFlip is one of the key components that promotes or inhibits CASP8 oligomerization and ensuing activation by autoproteolytic cleavage. As cFlip is transcriptionally regulated by NF-κB, it can also participate in a pro-survival pathway induced by TNFR1 in some conditions [79]. It is indeed increasingly clear that the activation of death receptors by their ligands does not necessarily lead to RCD but can also activate pro-survival signals. Specifically, the TNFR1-induced pathway depends on the RIPK1 ubiquitination level, which directly influences the formation of pro-survival versus pro-death complexes.

Briefly, in Complex I, RIPK1 polyubiquitination by cIAP1 and cIAP2 leads to NF-κB activation, pro-survival, and inflammatory genes transcription, where a high level of cFlip is correlated to survival. Subsequently, deubiquitinated RIPK1 is released from Complex I and forms Complex IIa in the cytosol with FADD, TRADD, cFlip, and CASP8. If cFLIP concentration is low, this complex leads to the degradation of RIPK1 and RIPK3, allowing CASP8 dimerization and activation and the subsequent apoptotic cell death through CASP3. In a context of high cFlip concentration, CASP8/cFlip heterodimers are formed and apoptosis is blocked [56,57,78]. Moreover, in the absence of cIAP (after MOMP and IAP inhibitors release for instance), phosphorylated RIPK1 leads to non-canonical NF-κB activation and subsequent association with RIPK3, FADD, and cFlip to activate CASP8 (Complex IIb) [73].

Finally, another possible pathway is induced when CASP8 is inhibited by chemical caspase inhibitors or virally encoded proteins. In this case, deubiquitinated RIPK1 and RIPK3 bind in microfilaments, "amyloid-like" complexes called necrosomes (most likely trimers or tetramers) [80,81]. The auto- and transphosphorylation of RIPK1 and RIPK3 and the recruitment of mixed lineage kinase domain-Like (MLKL) to the plasma membrane, triggering membrane permeabilization, initiate what is called necroptosis [82]. It is of interest to note that MLKL oligomers also lead to PS exposure, a feature usually considered as a hallmark of apoptosis [80].

If CASP8 is inhibited, RIPK3 phosphorylation can be triggered by some activated PRR (pathogen recognition receptors), such as TL3 and TL4 [83], nucleic acid sensors, such as RIG-I and MDA5, and some adhesion receptors [84,85]. In addition, IFN α and β receptor subunit 1 (IFNAR1) and IFNγ receptor 1 (IFNGR1) are also able to trigger necroptosis through TRIF and ISGF3 activation [86] (Figure 1).

### *2.3. Inflammasome Activation and Pyroptosis*

"Pyroptosis is a form of inflammatory RCD that critically depends on the formation of plasma membrane pores by members of the gasdermin (GSDM) protein family, often as a consequence of inflammatory caspase (CASP1, 4 or 5) activation" (NCCD [19]).

Pyroptotic cells present PS exposure, chromatin condensation, TUNEL staining but no DNA laddering, and a slight MOMP. Final GSDM-dependent membrane permeabilization allows the release of pro-inflammatory cytokines (IL1β and IL18, both NF-κB-target genes), maturated by interleukin-1β-converting enzyme (ICE/CASP1)-dependent cleavage. Other factors are also released, thus participating in the defense against pathogens through inflammation and the induction of an adaptive response [19,87,88]. In fact, pyroptosis seems to be mainly involved in the innate immunity against intracellular pathogens [88]. Pyroptosis was first thought to be restricted to monocyte/macrophage lineage, but it has been observed in other cells [89].

The inflammasome is activated by different DAMPs or pathogen-associated molecular pattern (PAMPs). It is a multiprotein complex that, like the apoptosome (intrinsic apoptosis) or the DISC (extrinsic apoptosis), acts as a caspase-activating platform. It is formed by a receptor (NOD-like receptor (NLR) family or non-NLR (AIM2)), an adapter protein (ASC, apoptosis-associated speck-like protein containing a CARD) and CASP1 that cleaves pro-IL18, pro-IL1β and GSDM. However, it is now clear that pyroptosis can also be activated by other caspases such as CASP3 [90] (Figure 2).

**Figure 2.** Inflammasome activation and pyroptosis. PM: plasma membrane, TLR: Toll-like receptor, IFNAR1: Interferon associated-receptor 1, DAMP: damage-associated molecular patterns, PAMPs: pathogen-associated molecular patterns, ROS: reactive oxygen species, NLR: NOD-like receptor, IL: interleukin, LPS: lipopolysaccharide.

The recent description of necroptosis and pyroptosis processes has amplified the complexity of RCD understanding. Contrary to general consensus so far, it is now clear that intrinsic as well as extrinsic regulated cell death can be immunogenic, thus participating in the establishment of the adaptive immune response [91,92]. This has been recently underlined by the description of the concomitant activation of apoptosis, necroptosis, and pyroptosis in a context of bacterial or viral infection in macrophages, leading to inflammatory cell death. The phenomenon has been named PANoptosis and would involve molecules of the three RCD pathways (i.e., CASP8, RIPK3, and CASP1) in a single complex called PANoptosome [93].

### *2.4. Autophagy-Dependent Cell Death*

Autophagy-dependent cell death is a type of RCD that depends on components of the macroautophagy machinery [19]. Macroautophagy is a particular form of autophagy where double-membrane vesicles (autophagosomes) sequester a large part of organelles and cytoplasm, leading to their lysis and, in some cases, to cell death. Morphologically, dying cells present an accumulation of autophagosomes and autolysosomes in the cytoplasm, a feature extremely di fferent from apoptotic or necrotic RCD. However, it seems increasingly clear that autophagy usually inhibits rather than induces cell death and has to be considered as a way for the cell to maintain homeostasis after stress signals (hypoxia, ROS, starvation, PRR activation, etc.) [94]. Even if in some cases inhibition of specific autophagy proteins can delay RCD [95], pharmacological or genetic inhibition of macroautophagy components usually accelerates the death of cells rather than protects them. Autophagy may degrade damaged mitochondria or pro-apoptotic complexes thus preventing cell death.

It is of interest to note that autophagy leads to the degradation by lysosomes of components of endogenous or exogenous origin, which are accessible in the cytoplasm. This has to be distinguished from vesicular tra fficking, which starts in the plasma membrane and also leads to lysosomal degradation (i.e., phagocytosis or receptor-mediated endocytosis). Macroautophagy and vesicular tra fficking pathways interact at numerous regulation points, especially in the late phases of the pathways. Autophagosomes or late endosomes fusion with lysosomes actually require the same machinery. Thus, numerous proteins involved in physiological or lytic autophagy are also essential for viral penetration, from receptor-mediated endocytosis to fusion of the viral envelope with the endosome membrane and subsequent liberation of the viral genome in the cytosol.

### **3. Impact of Alphaviral Infection on Regulated Cell Death**

During a viral infection, RCD is generally described as a defense mechanism, induced to limit virus replication and production, to prevent infection of neighboring cells and, to some extent, to participate in immune response induction. Cell death induced by alphavirus infection has been observed and studied in several cell types infected by di fferent alphaviruses, mainly CHIKV, SFV, SINV, VEEV, and EEV.

### *3.1. Apoptotic Pathways in Alphavirus-Infected Cells*

Infections of baby hamster kidney (BHK), rat prostatic adenocarcinoma (AT-3), and mouse neuroblastoma (N18) cells with SINV result in clear nuclear condensation and membrane blebbing, 24 h post-infection [96]. In the same context, SFV infection of AT-3 induces apoptotic features, correlated to strain virulence. Grandgirard et al. [97] have described, in rat embryonic fibroblasts, a potential caspase-dependent Bcl-2 cleavage in SFV- or SINV-infected cells leading to cell death and viral replication, even in a context of Bcl-2 overexpression. In 293T and BHK cells, the BH3-only protein Bad seems to participate in SINV-induced cell death, through its interaction with some specific anti-apoptotic Bcl-2 proteins, while the other non-binding members also regulate cell death [98]. The dynamics of mitochondria are also highly altered during apoptosis induced by VEEV infection of human astrocytoma cells U87MG [99]. Infection rapidly induces MOMP and ROS increase, followed by perinuclear localization and fission of mitochondria, and then mitophagy. Moreover, VEEV capsid co-localizes with mitochondria and could participate in mitochondria dysregulations.

Lin et al. showed a link between SINV induced cell death, oxidative stress, NF-κB, and Bcl-2 expression. In AT-3 and N18 infected cells, NF-κB activation and cell death were indeed inhibited either by antioxidant agents or Bcl-2 overexpression, with no e ffect on viral entry or replication however [100]. Likewise, MRC5 human fibroblasts present a SINV persistent infection, when manganese-superoxide dismutase (Mn-SOD) is over-expressed, confirming that oxidative pathways are implicated in the e ffects of SINV [101]. Chikungunya virus infection has been studied in the neuroblastoma cell line SH-SY5Y by Dhanwani et al. [102]. Intrinsic apoptosis features (CYC release, CASP3 activation, PARP cleavage) are observed 24 h and 36 h post-infection. Moreover, the infection is followed, 36 h and 48 h after by an elevation of ROS, a decrease of anti-oxidant enzymes expression and glutathione (GSH) depletion.

Another cellular response to stress induced by viral infection is the unfolded protein response (UPR) of the ER due to the accumulation of newly synthetized viral proteins in the ER, leading to translation blocking and intrinsic apoptosis [103]. Endoplasmic reticulum stress response has been described in *flavivirus*-infected cells, but little is known about alphaviruses. However, SFV envelope glycoproteins, but not capsid, seem able to induce and accelerate apoptotic cell death [104], while VEEV glycoproteins induce UPR and apoptosis in primary astrocytes [105,106].

Finally, intrinsic apoptosis may be triggered by alphavirus non-structural protein activity. Indeed, SINV, VEEV, and EEEV nsp2 and nsp3 have been shown to be responsible for viral cytopathic e ffects, enabling persistent infections when specifically mutated [107–109]. Frolov and colleagues have described the nuclear translocation of nsp2 and global transcriptional shuto ff through RNA polymerase II degradation [110–112] and a subsequent nsp3-dependent translational shuto ff for arthritogenic viruses [112], while capsid would play this role for encephalitic viruses [113]. However, CHIKV nsp2 seems to inhibit UPR as well through its transcriptional shuto ff activity [114] and to interfere with the IFNβ signaling pathway [115,116].

However, Sarid et al. characterized a CASP8 and TNF α/TNFR1-dependent PC-12 RCD after infection by the SINV SVNI strain (neurovirulent and cytotoxic). Indeed, they described an upregulation of TNF α expression in infected cells and a cell death inhibition following cFlip overexpression [117]. Nava and colleagues [118] have shown, in SINV-infected BHK cells and in mice, an inhibition of death after treatment with the pan-caspase inhibitor zVAD-fmk and the CASP1 and CASP8 inhibitor CrmA (a serine proteinase inhibitor from Cowpox virus). In addition, by using SFV replicon vectors or a wild-type SFV strain, Kiiver et al. have neither shown any Bcl-2 protective action against virally induced cellular protein synthesis shutdown post-infection nor cell death. Moreover, AT-3 and BHK cells did not present any CYC release after infection [119]. In addition, unlike poxvirus, that blocks CASP1 and CASP8 [120,121], or herpesviruses [122], alphaviruses have never been described as inducing necroptosis.

Thus, on the one hand, alphaviruses appear to induce apoptosis through mitochondria, oxidative, and ER stress or the transcriptional and translational shuto ff induced by nsp (intrinsic pathway). On the other hand, some studies sugges<sup>t</sup> the implication of death receptors and CASP8 activation, without mitochondrial involvement (extrinsic pathway). This apparent discrepancy may be explained by the di fferent apoptotic cells that have been observed (infected or neighbor cells) and the duration of infection (few hours to days post-infection). Indeed, Joubert et al. sugges<sup>t</sup> that, in a first wave, CHIKV-infected cells die through intrinsic apoptosis (CASP9 positive cells) and that, in a second time, during antiviral response, infected, and neighbor cells die through the extrinsic pathway (death receptors and CASP8 activation). These secondary pathways seem to be independent of ER and oxidative stress [123,124].

Other explanations could be found also in two recent studies, which involve two newly described pathways in SFV-induced cell death. Using 3T9 MEFs, Urban et al. [125] characterized SFV-induced RCD. In their study, they first showed that RCD was triggered by SFV replication and not only by viral entry, as previously described [126]. Secondly, SFV-induced cell death occurred in a Bak dependent MOMP, leading to apoptosome activation. Moreover, they excluded the involvement of TRAILR, Fas or TNFR1 in this process. Surprisingly, CASP8 and tBid seemed to be activated downstream of apoptosome, maybe through CASP6 [127], acting as an amplification loop. In fact, although CASP6 has long been considered as an executioner caspase based on its homology with CASP3 and CASP7, recent data

sugges<sup>t</sup> that CASP6 may actually be involved in RCD initiation [127,128]. Additional investigation is required to elucidate the function of CASP6 in mammalian cells. Secondly, in several SFV-infected cell types, El Maadidi et al. recently described a new mitochondrial platform, involving the innate immune factor MAV and the initiator CASP8, comparable to the death-inducing signaling complex (DISC) of the death receptors signaling. However, this complex does not involve FADD but another potential, not ye<sup>t</sup> characterized adaptor. The platform is activated via the dsRNA sensors MDA5 or RIG-I and acts in parallel of the classical Bax/Bak dependent MOMP, also leading to CASP3 activation, independently of type I IFN signaling factors (IRF3, IRF7, IFNβ, PKR, etc.) and mitochondrial depolarization [129]. As MAVs platforms have been described in MAMs, it is tempting to hypothesize that the MAV/CASP8 should be localized in these subcellular structures thus interacting with other metabolic pathways (such as ROS, Ca2+, lipid, autophagy) during antiviral response [130].

Hence, it appears that several alphaviruses induce regulated cell death in numerous cell types, involving mitochondria depolarization, ER stress, and CASP8 activation, maybe in a time-dependent regulation or through alternative processes. However, after nearly 30 years of study, large parts of the molecular pathway leading an infected cell to death remain to be deciphered. It would be of interest in the future studies to focus on the cell types relevant for the viral tropism (e.g., skin cells, muscle cells, neural cells) and to favor innovative cell culture technics that mimic better the natural cell characteristics (e.g., 3D culture, explants, iPS (induced Pluripotent Stem cells)).

### *3.2. Inflammasome and Pyroptosis in Alphaviral Infection*

Activation of the inflammasome pathway and pyroptosis has been intensively studied for flaviviruses infection, especially for their involvement in pathogenesis [131–133], but little is known for alphaviruses. Even if inflammasome pathways seem to participate in the pathology and the response against alphaviruses, the involvement of pyroptosis has never been described. In dermal fibroblasts, CHIKV and the *flavivirus* West Nile virus (WNV) both induce IL1β production and CASP1 activation through the AIM2 inflammasome sensor, but only CHIKV replication and propagation can be controlled by CASP1 [134]. In PBMC from CHIKV-infected patients, high levels of NLRP3, IL18, and CASP1 are found [135]. Moreover, in mice, NLRP3 activation is correlated to inflammatory symptoms such as bone damage and myositis. NLRP3 inhibition leads to a reduction of the inflammatory pathology induced by CHIKV but not by WNV [135]. The alphavirus Mayaro (MAYV) induces the expression of inflammasome proteins in macrophages, and inflammatory cytokines production through the NLRP3 sensor, activated by ROS and K<sup>+</sup> efflux. In mice, NLRP3 is also involved in MAYV induced pathogenesis [136].

Thus, inflammasome activation has been mainly involved in global inflammatory response to alphavirus in vivo, but the molecular pathways activated in the cell remain to be described. Indeed, it is still unknown if pyroptosis may participate in inflammatory cytokines secretion during alphavirus infection, and, to our knowledge, there is no molecular study of this process.

### **4. Interplay between Cell Death and Alphaviral Replication and Spread in Mammals**

Apoptosis appears to be a strong antiviral process. Indeed, Bcl-2 overexpression converts SINV infection from lytic to persistent in vitro [96] and in vivo [137]. Moreover, Bcl-2 seems to be able to restrict SFV replication by inhibiting early stages of infection and appears to prolong survival of productively infected cells [138].

As described above, autophagy usually blocks apoptosis, and viruses have developed strategies to take advantage of this property. The first connection between alphavirus infection and autophagy has been made by Liang and colleagues [139], when they identified Beclin-1 as a new Bcl-2-interacting protein through a yeas<sup>t</sup> two-hybrids screening. Beclin-1 is a major factor of autophagy, involved in autophagosome initiation and maturation. In this study, Beclin-1 protected SINV-infected mice against fatal encephalitis, with a significantly lower viral replication rate in mice brains. The author correlated these observations with the previously observed protective role of Bcl-2 against in vivo

SINV infection [96,137]. However, years after this study, Bcl-2 was described as a Beclin-1 inhibitor, thus participating in ER stress connected autophagy regulation [140]. Another crucial protein of the autophagy pathway, Atg5, protects mouse neurons from SINV-induced cell death [141], with no apparent impact on viral replication. Moreover, the adaptor protein p62 seems to be linked to viral capsid clearance by direct interaction and target of autophagosomes, thus promoting cell survival [141]. Finally, in SFV-infected cells, autophagosomes accumulate but autophagy modulation has no effect on viral replication, and this autophagosomes accumulation seems to be due to the inhibition of their degradation rather than an induction by SFV infection [142].

In HEK293 cells, CHIKV infection induces autophagy features (LC3 positive vesicles and electron microscopy observation). In this study, autophagy has a clear pro-viral role, increasing the number of infected cells and viral RNA in the cell culture supernatant [143]. Moreover, in vitro and in vivo, CHIKV infection has also been shown to induce an autophagy flux, through ER and oxidative stress [123,124]. In these models, autophagy limits (i) extrinsic and intrinsic RCD induced by CHIKV infection, (ii) mice lethality, and (iii) viral propagation. Autophagy, as a host response to infection, limits indeed the cytopathic effects of CHIKV and regulates the pathogenesis of acute chikungunya disease. However, during late phases of in vitro infection (48 h post-infection), a switch between autophagy and apoptosis is observed and cells die. Finally, in HeLa cells, autophagy promotes CHIKV infection and inhibits cell death. Indeed, in addition to p62-dependent capsid clearing, another autophagy receptor, NDP52, interacts with nsp2, localizes near the CHIKV replication complex and restricts cell shutoff thus promoting viral replication and cell survival [144].

Hence, alphaviruses may exploit autophagy to delay cell death through (i) direct inhibition of intrinsic and extrinsic apoptosis and (ii) a limitation of viral proteins production, allowing cell survival and a longer viral replication.

However, several Old World Alphaviruses, such as CHIKV, SFV, and RRV, seem able to activate the phosphatidylinositol-3-kinase (PI3K)–AKT–mTOR pathway, involved in cell survival and autophagy inhibition. Furthermore, inhibition of this pathway has a negative effect on viral replication [145]. This apparent discrepancy with the previous observations may indicate that, more than autophagy per se, cell survival is the key process which favors viral replication.

Finally, it is of interest to note that, in some cases, apoptosis has been shown to enhance viral spread. Indeed, in their study, Krejbich-Trotot and colleagues [146] first confirmed the dual nature of the alphavirus-induced apoptosis (intrinsic and extrinsic) in HeLa and primary fibroblasts infected with CHIKV accompanied by CASP8 activation in neighbor cells. More interestingly still, inhibitors of blebbing or engulfment drastically reduced infection rates. Finally, they detected infective CHIKV in apoptotic corpses and in the macrophages which phagocyted them, leading to macrophages infection and viral production. As macrophages are refractory to CHIKV infection in vitro, this study highlights a possible role of apoptotic blebs in viral propagation. This phenomenon, called "apoptotic mimicry", is used by a large number of viruses to exploit the PS receptors present on numerous cells membranes, enhancing viral spread and limiting immune response [147].

### **5. Impact of Apoptosis on Virus Pathogenesis in Mammals**

### *5.1. Alphavirus Encephalitis*

The first cell death analysis was documented in vivo, using SINV, VEEV, and SFV, three alphaviruses causing encephalitis. In SINV-infected mice, the apoptotic cells were detected principally in the brain and contained viral antigens, suggesting that apoptosis was correlated to neurovirulence [148]. The in vivo mouse infection of VEEV was also associated to cell death in brain, demonstrated by TUNEL assay (DNA fragmentation) and morphological changes [149].

Comparing SINV infections with SVNI (neurovirulent and cytotoxic) or SVA (avirulent and leading to persistent infection) strains in PC-12 cells and astrocytes, revealed that SVNI induces Bax overexpression while SVA induces Bcl-2 expression [150].

Intranasally SFV-infected rats develop encephalitis, where infiltrating leucocytes and neural precursor cells undergo apoptosis while productively infected neurons present necrotic features, apparently due to the local inflammation [151].

Hence, it appears that alphaviruses pathogenicity is linked to its cytopathic e ffects in infected cells, at least in the case of the encephalitic group.

### *5.2. Alphaviral Chronic Infection: What about Cell Death?*

One characteristic of the *Alphavirus* genus is the ability of some of them (CHIKV, SINV, MAYV, RRV, etc.) to induce chronic pain, such as arthritis and myalgia, which may last for years, with detectable viral genome in the organism. This persistent infection implies that some cells may be chronically infected, and in some way able to delay or block cell death. However, despite an intense immune response observed in chronic patients, damaged synovial tissues present strong apoptosis features. Chikungunya virus has been found in synovial macrophages several months after infection but joints do not seem to be the viral reservoir [152]. In addition, RRV-infected human monocyte acute leukemia MM6 cell line presents very low replication rates, without innate immune control, and apoptosis features at late stages of infection. This indicates that monocytes could be persistently infected and participate in the chronic form of RRV or CHIKV [153]. Young et al. [154] propose that dermal and muscular fibroblasts, as well as myofibers, may survive the acute CHIKV infection and harbor persistent CHIKV RNA during chronic phase of the disease. Moreover, they observe that synovial cells are not infected in large numbers in vivo and sugges<sup>t</sup> that synovial cells may be infected but do not survive.

How these cells survive remains to be understood. The mechanisms involved in alphaviral persistence are mostly unknown. They may depend on infected cell type and the highly complex interplay between virus, immune response, and di fferent RCD pathways.

### **6. Interplay between Cell Death and Susceptibility of Mosquito Species to Arboviruses**

The interplay between arbovirus and arthropods is still poorly understood and the primary point of study concerns flaviviruses (mainly Dengue virus (DENV)) in mosquito and/or in *Drosophila*, used as a genetic model for insect immunity. Hence, very little is known concerning the impact of alphavirus infection in mosquitoes and which factors may explain the tolerance versus resistance of di fferent mosquito strains.

It is classically admitted that alphaviruses do not induce any major pathology in their vectors. However, several lesions are observed in tissues which are critical for viral propagation and transmission. Indeed, after feeding on infected blood, cellular response in the midgut plays a decisive role in vector competence. EEEV infection of *Culiseta melanura* mosquito induces severe lesions in midgut epithelial cells and basal lamina, associated to viral spread [155]. Likewise, infection of more or less susceptible *Culex tarsalis* strains with WEEV revealed lesions and apparent necrotic cell death only in the sensitive mosquito's gu<sup>t</sup> [156]. Transcriptomic analysis of *Aedes Aegypti* fed with CHIKV in blood or di fferent bu ffers reveals the over-expression of matrix metallo proteinases (MMP) and other peptidases in the midgut, as well as the decrease of Collagen IV, a component of the basal lamina [157]. Intrathoracically SINV-injected *Aedes Albopictus* present colocalization of virus antigen with structural lesions and TUNEL positive cells in salivary glands [158] and midgut-associated visceral muscles [159]. Furthermore, organ-associated muscles respond di fferently to SINV [160]: 10 days post-infection, the virus has cleared from the midgut, is persistent in the hindgut, and unable to infect ovary associated muscle cells. High viral titers induce pathology limited to gu<sup>t</sup> associated muscles and gu<sup>t</sup> epithelium. Finally, in *Aedes Aegypti* mosquitoes, AeIAP1 (IAP ortholog) downregulation leads to a higher replication of SINV in the midgut, while AeDronc (CASP9 ortholog) inhibition is associated to a lower viral replication and dissemination towards salivary glands [161] (see Box 1 for RCD pathway description in insect).

**Box 1.** Comparative cell death pathways in mammal, *Drosophila* and *Aedes.* PM: plasma membrane; CYC: cytochrome C; IAP: Inhibitors of apoptosis; *Ae*: *Aedes*; RHG: Reaper, Hid and Grim; IMD: immune deficiency.

Recent knowledge concerning apoptosis in mosquitoes has been acquired through gene homology with *Drosophila melanogaster*. Apoptosis is under the control of initiator and e ffector caspases [162], expressed ubiquitously and synthesized as inactive procaspases. The Apaf-1-related killer (Ark) molecule [163] assembles itself into an apoptosome-like complex [164] to activate Dronc, but the role of CYC in the insect apoptosome is very controversial [165]. Two Bcl-2 orthologues have been identified in *Drosophila*: Bu ffy and Debcl [166], whose pro- or anti-apoptotic roles are not clear [166,167]. Finally, mitochondrial fission, through Drp1 activation, seems to be required for e fficient cell death [168]. However, the pro-apoptotic activity of caspases is mainly regulated by members of insect IAP family [169,170]. IAP antagonists (dOmi/HtrA2 and RHG proteins in drosophila; Michelob\_X in mosquito) are localized in the mitochondrial intermembrane space in living cells and released into the cytosol, but remain near the mitochondria, after an apoptotic stimulus [170,171], where they compete for caspase binding through their IAP-binding domain. Additionally, RHG proteins can induce DIAP1 ubiquitination and degradation [168,172]. Interestingly, RHG and Mx promoters present di fferent response elements regulated by transcription factors, such as dmp53, activated by developmental or environmental signals, leading to cell death [173]. In insects, no clear distinction can be made between intrinsic or extrinsic apoptotic pathways. Nevertheless orthologues of TNF (Eiger) and TNFR (Wengen) have been described in *Drosophila* [174] and induce cell death through a JNK-mediated pathway, requiring apoptosome components [175]. Immune response against pathogens is triggered by the NF-κB (Relish) pathway induction through the Immune Deficiency IMD/dFADD/Dredd (CASP8) pathway [176] in *Drosophila* and IMD/AeFADD/AeDredd in *Aedes* [177].

Finally, autophagy-dependent cell death is finely controlled in insects, and its role in development has been largely studied in *Drosophila* [178,179]. To our knowledge, pyroptosis and necroptosis have not ye<sup>t</sup> been described in insects.

Thus, caspase activity may be required for dissemination of SINV from the midgut to the secondary organs by participating in the remodeling of the basal lamina, as suggested in baculovirus-infected lepidopteran, where caspase and MMP activity is necessary to cross the midgut barrier [180]. However, cell death modulation in vitro, in mosquito cells, does not seem to alter alphaviral replication. In fact, recombinant SINVs, expressing Reaper (*Drosophila* IAP inhibitor) or Michelob\_X (Mx, *Aedes* IAP inhibitor), induce apoptosis in infected *Aedes Albopictus* C636 cells, with no inhibitory e ffect on viral production in the initial phase of infection. Moreover, in these conditions, inhibition of caspase activity has no e ffect on viral replication neither [181]. However, recombinant SINV expressing Reaper induces cell death in vivo in *Aedes Aegypti*'s midgut, a delayed infection and propagation in the saliva [182]. More importantly, this last study also describes a rapid genetic selection of SINV variants in vivo against Reaper expression.

Hence, cell death, with tissue degradation features, seems important for alphavirus propagation in mosquito organism, with no clear e ffect on replication in vitro. However, as described above for mammal cells, RCD is also associated to e fficient immune response against viruses and may be one of the key processes involved in mosquito resistance to virus.

How cell death is modulated in infected tolerant mosquito cells remains to be understood. Oxidative stress response may play an important role in the mosquito's response. Indeed, CHIKV infection induces upregulation of antioxidant pathways in mosquito midgut, which delays cell death [183]. During arbovirus infection, oxidative stress is actually detected in both mammal [101,123,184] and insect cells. Oxidative stress is defined by loss of homeostasis between accumulation of ROS and production of antioxidant enzymes such as superoxide dismutase (SOD), catalase (CAT) or glutathione transferase and reductase [185]. After blood feeding, the midgut is in contact with sugar, iron, heme and other components of vertebrate blood. Mosquitoes have developed protective adaptation against the damage caused by heme and iron uptake. Indeed, the heme can induce lipid peroxidation, protein degradation, and ultimately cell death. Once in the epithelial cells, these components are detoxified, and a strong antioxidant and protective response is engaged [186]. Concomitantly, pathogens present in the blood could take advantage of this antioxidant response, blocking cell death, to infect and replicate into midgut epithelial cells.

The pro-survival pathway PI3k–Akt–mTOR may also be involved in insect tolerance. Indeed, in drosophila, activation of the PI3k–Akt–mTOR pathway is associated to an increase of SINV infection, potentially through apoptosis and autophagy inhibition and a more e fficient cap-dependent translation of viral genome [187]. Finally, the role of autophagy seems to be limited in CHIKV infection of mosquito cells. Even infection induces autophagy in Aag2 cells, every pharmacological modulation of autophagy (inducer or blocker) leads to a replication increase in mosquito cells [188].

Hence, the sensitivity may depend on the better resistance of midgut cells to oxidative stress induced by viral infection, leading to a delayed cell death but the involvement of autophagy in these regulations remains to be understood.

Few studies have been conducted to investigate the interplay between alphaviruses and their vector. However, recent findings in mosquito and drosophila underlie the role of p53 isoforms in cell response to oxidative stress and to DENV infection. The balance between a rapid apoptosis and a delayed, secondary necrosis may explain in part the di fferences between tolerant and resistant mosquitoes' strains. For more details see [177,189–196] and Figure 3 for suggested mechanisms which may be involved in alphavirus infection in sensitive versus resistant mosquitoes, extrapolated from other arboviruses.

**Figure 3.** Possible interplay between cell death and alphavirus infection in *Aedes* mosquitoes: extrapolation from other arboviruses studies. BL: basal lamina, *Ae*IAP: *Aedes* inhibitor of apoptosis protein. See Box 1 for RCD pathway description in insect.

In addition to the virus' ability to penetrate the cells of a particular mosquito (specific receptor, lipid membrane composition), the difference between resistant and tolerant strains may also lie in the rapidity of cell response to viral infection in the first targeted tissue, i.e., the mosquito gut. This response has to be apoptotic and not necrotic to ensure mosquito resistance. A p53/Mx- [189] and caspase- [177] dependent cell death has been linked to *Aedes Aegypti* resistance to DENV infection. Phagocytosis of dead cells and apoptotic bodies also seem to be important for a virus specific immune response, at least in drosophila [190].

Sensitive mosquitoes may tolerate viral infection through IAP-dependent apoptosis inhibition, as shown for the arbovirus bluetongue virus (BTV) [191,192]. Another tolerance mechanism to DENV seems to rely on resistance to oxidative stress through CAT protection in mosquito gu<sup>t</sup> [193] or p53 isoforms regulation leading to cell survival in drosophila [194] and mosquito [195,196]. Delayed apoptosis may then lead to secondary necrosis, impairing a proper innate immune response, favoring basal lamina damage and viral spread [189].
