Herpesvirus Regulation of Selective Autophagy

Selective autophagy has emerged as a key mechanism of quality and quantity control responsible for the autophagic degradation of specific subcellular organelles and materials. In addition, a specific type of selective autophagy (xenophagy) is also activated as a line of defense against invading intracellular pathogens, such as viruses. However, viruses have evolved strategies to counteract the host’s antiviral defense and even to activate some proviral types of selective autophagy, such as mitophagy, for their successful infection and replication. This review discusses the current knowledge on the regulation of selective autophagy by human herpesviruses.


Introduction
Macroautophagy (simply called autophagy) is a conserved cellular degradation and recycling system that is essential for cell survival during normal physiological conditions and in response to different types of stress, such as starvation [1]. In its common form, autophagy proceeds through the formation of double-membrane vesicles, termed autophagosomes, that engulf portions of the cytoplasm and deliver them to lysosomes. In addition to the non-selective autophagy, forms of selective autophagy have evolved to remove specifically damaged, superfluous, or unwanted materials including protein aggregates and subcellular organelles, thereby providing cellular quality control and homeostasis [2][3][4]. Selective autophagy is mediated by autophagy cargo receptors, which contain an LC3-interacting region (LIR) [5] and can therefore bind directly to the autophagy-related gene 8 (ATG8) family proteins, including MAP1LC3A (LC3A), LC3B, LC3C, GABARAP, GABARAPL1, and GABARAPL2, on the phagophore [6].

Mitophagy
Mitochondria have been known to act as a platform for antiviral signaling that leads to apoptosis and type-I interferon (IFN) expression in response to virus infection [16,17]. Therefore, a variety of RNA and DNA viruses have evolved strategies to attenuate mitochondria-mediated antiviral responses for their own benefit [13,[16][17][18]. On the other hand, mitochondria normally generate reactive oxygen species (ROS) from the mitochondrial respiratory chain reaction [19]. Accumulation of damaged or dysfunctional mitochondria leads to the generation of excessive ROS, which can potentiate apoptosis through mitochondrial membrane depolarization and augment innate immune signaling [20][21][22]. Thus, viruses have to eliminate altered mitochondria to promote their successful infection and replication. Indeed, recent studies have demonstrated that RNA and DNA viruses can eliminate infection-altered mitochondria via mitochondria quality control, which is achieved by balanced actions among mitochondrial biogenesis, dynamics (fission and fusion events), and mitophagy, potentially related to attenuation of apoptosis and the innate immune response (Table 2).  [41] Mitophagy is a selective form of autophagy that eliminates dysfunctional or superfluous mitochondria in a ubiquitin-dependent or independent manner [42]. The mitochondrialocalized kinase PINK1 and the ubiquitin (Ub) E3 ligase Parkin (PARK2) are the best known mitophagy-associated proteins. Upon the loss of mitochondrial membrane potential, the PINK1 and Parkin complex induces ubiquitination of mitochondria [43], which is recognized by ubiquitin-binding SLR autophagy receptors [44] and then targeted to autophagosomes via interaction with ATG8 family proteins. In addition, the ubiquitinationindependent mitophagy pathways via mitophagy receptors, such as NIX/BNIP3L, BNIP3, FUNDC1, and PHB2, have been identified under certain physiological and pathological conditions, including erythrocyte differentiation, paternal mitochondrial elimination, hypoxia, and virus lytic replication [41,[45][46][47][48]. It is believed that mitochondrial fission/fragmentation is prerequisite for mitophagic removal of damaged parts of mitochondria. Here, we describe recent findings on the regulation of mitochondrial fragmentation and mitophagy by human herpesviruses. Most herpesviruses, except for human herpesvirus 8 (see below), have not been proven to activate mitophagy receptors directly from their gene products, but we discuss some human herpesviruses showing activity which induces mitochondrial fragmentation, demonstrating their potential for mitophagy activation.

Herpes Simplex Virus 1 (HSV-1) and Herpes Simplex Virus 2 (HSV-2)
HSV-1, also known as human herpesvirus 1 (HHV-1), and HSV-2, also known as HHV-2, are so ubiquitous that two third of the human population in the world are infected by at least one of these viruses [49]. Although infections by HSV-1 and HSV-2 are usually asymptomatic or associated with mild symptoms, the viruses have been considered as causative agents in severe neuronal diseases, such as herpes simplex encephalitis (HSE) and meningitis [50]. The severe manifestations are postulated to be associated with mitochondrial dysfunction in cultured neuron cells after HSV-1 and HSV-2 infection [51]. HSV-1 and HSV-2 infection-induced mitochondrial dysfunction occurs at multiple levels, including upregulation of mitochondrial fission, decrease in the mitochondrial membrane potential, and increase in ROS levels. The changes in the organization of the mitochondrial network in infected neurons provide appropriate conditions for HSV-1 and HSV-2 replication and are required for effective viral dissemination [51]. Furthermore, in brain tissues of HSE cases, the levels of mitochondrial DNA (mtDNA) and mtDNA-encoded proteins are significantly reduced, suggesting that mitophagy may be activated following HSV-1 infection in neurons. Together, these results suggest that HSV-1 and HSV-2 may activate mitochondrial quality control mechanisms for their productive replication, and that this may contribute to pathogenesis. However, there is no direct evidence to date that HSV-1 and HSV-2 activate mitophagy.

Varicella-Zoster Virus (VZV)
Primary infection by VZV, also known as HHV-3, causes varicella (chickenpox) during which VZV becomes latent in ganglionic neurons. When reactivated, it causes zoster (shingles), which can be complicated by chronic pain and other serious neurological disorders, such as meningoencephalitis, myelitis, and vasculopathy [52]. Although one report showed that VZV infection induces a time-dependent significant change in mitochondrial morphology (fragmentation), potentially via its immediate-early gene product IE63, in human fetal lung cells [53], whether mitophagy activation is associated with VZV infection and associated pathogenesis has not been demonstrated.

Epstein-Barr Virus (EBV)
Infection by EBV, also known as HHV-4, causes various human malignancies, including posttransplant lymphoproliferative diseases, nasopharyngeal carcinoma, Hodgkin's lymphoma, Burkitt lymphoma, and gastric carcinoma [54]. The EBV-encoded product BHRF1, a BCL2 homolog, is known to activate mitophagy via modulating mitochondrial dynamics or mitochondrial membrane permeabilization transition (MMPT) in BHRF1transfected cervical (HeLa) or nasopharyngeal (CNE1 and 5-8F) cancer cell lines and EBV-reactivated AKATA cells [55,56]. BHRF1 expression stimulates DRP1-mediated mitochondrial fission and, concomitantly, autophagic flux by interacting with Beclin 1, thereby attenuating type-I IFN induction [55]. The BHRF1-induced mitochondrial fission and autophagy stimulation lead to mitochondrial network reorganization to form juxtanuclear mitochondrial aggregates, which then cause the induction of mitophagy and the accumulation of PINK1 at the mitochondria. In addition, BHRF1 expression is known to induce cyclophilin D-dependent MMPT, which in turn increases ROS production and activates mitophagy in nasopharyngeal carcinoma cell lines [56]. In addition, the viral latent membrane protein 2A (LMP2A) has been shown to promote DRP1-mediated mitochondrial fission in stably transfected gastric and breast cancer cell lines, AGS and MCF7 respectively, suggesting that LMP2A-induced mitochondrial fission may contribute to tumorigenesis of EBV-associated epithelial cancers [57]. Overall, BHRF-1-or LMP2Aregulated mitochondrial dynamics and mitophagy are implicated in the promotion of EBV tumorigenesis [56].

Human Cytomegalovirus (HCMV)
HCMV, also known as HHV-5, infects most individuals in the world, mostly without showing overt symptoms. Modern advanced diagnostic methods reveal that HCMV is a common opportunistic infection in fetuses and immunocompromised individuals, including organ transplant patients and AIDS patients [58]. It is not known if mitophagy is activated in response to HCMV infection. Instead, HCMV infection promotes mitochondrial biogenesis and respiration to support bioenergetic and biosynthetic requirements early, during the viral replication cycle in human foreskin fibroblasts and U373 astrocytoma cells [59,60]. The immediate-early protein pUL37x1, also known as the viral mitochondrial inhibitor of apoptosis (vMIA), is known to induce mitochondrial fragmentation in human fibroblasts [61][62][63]. However, it is unclear to date whether pUL37x1-induced mitochondrial dynamics are associated directly with the inhibition of BAX-dependent or -independent apoptosis [64,65], or with HCMV infection-induced mitochondrial biogenesis. Future studies are warranted to understand how HCMV infection or pUL37x1 regulates mitochondrial quality control in infected cells.

Human Herpesvirus 8 (HHV-8)
HHV-8, also known as Kaposi's sarcoma-associated herpesvirus (KSHV), is the etiological agent that is causally associated with at least three human malignancies, Kaposi's sarcoma (KS), primary effusion lymphoma (PEL), and multicentric Castleman's disease (MCD) in immunodeficient individuals with HIV infection or AIDS [66,67]. Like other herpesviruses, HHV-8 has two distinct infection stages in the host: latency (persistent infection) and lytic replication, both of which are implicated in HHV-8 pathogenesis. In particular, PEL and MCD are likely associated with lytic replication [68][69][70]. Recently, it has been demonstrated that HHV-8-encoded viral IFN regulatory factor 1 (vIRF-1) localizes in part to mitochondria, where it interacts directly with the mitophagy receptor NIX and promotes NIX-mediated mitophagy to remove dysfunctional mitochondria during lytic replication in HHV-8-infected PEL cells [41]. vIRF-1 also has the ability to induce DRP1-dependent mitochondrial fragmentation in transfected HeLa cells [41], which may contribute to the promotion of mitophagy. Although, the exact mechanisms by which vIRF-1 orchestrates mitochondrial dynamics to induce mitophagy remains to be determined. This pro-mitophagic activity of vIRF-1 is linked to the inhibition of apoptosis and the promotion of virus productive replication.

Aggrephagy
It is believed that soluble misfolded proteins are repaired or removed by protein quality control machineries, such as folding complex-associated chaperons and the ubiquitin-proteasome-mediated degradation system (UPS), respectively. Unfolded or misfolded proteins in the lumen of the endoplasmic reticulum (ER) are resolved by the unfolded protein response (UPR) that enhances the expression of ER chaperones and components of ER-associated degradation [71]. When misfolded proteins fail to be rescued, however, they form protein aggregates, which are then removed by a selective form of autophagy, called aggrephagy [72]. Aggrephagy is mediated by autophagy receptors, TOLLIP, a human homologue of Saccharomyces Cue5 that contains a CUE domain, and specific SLRs, including p62/SQSTM1, NBR1, and OPTN [73][74][75][76][77]. Intriguingly, OPTN-mediated aggrephagy is independent of ubiquitin [77]. In addition, WDFY3, also known as ALFY [78], is involved in the clearance of aggregated polyQ proteins by interacting with a number of different partners, including p62/SQSTM1 [79], GABARAP subfamily members [80], and phosphatidylinositol-3-phosphate [81], a prominent lipid in the regulation of autophagosome membrane formation. These results suggest that WDFY3 acts as a scaffold protein in p62/SQSTM1-dependent degradation of ubiquitinated aggregates. Interestingly, in yeast, the mechanistic determination of protein removal by aggrephagy versus degradation appears to be independent of the ubiquitin-binding properties of the receptors and largely determined by their oligomerization potentials [82]. In fact, oligomerization of p62/SQSTM1 and Cue5 is required for selective autophagy of ubiquitinated cargo [83,84]. Therefore, it is likely that WDFY3 promotes the oligomerization of p62/SQSTM1 for aggrephagy activation.
Recently, aggrephagy was proposed to play a proviral role in herpesvirus infection. Murine cytomegalovirus (MCMV)-encoded M45 protein, a potent cell death inhibitor [85], interacts with two cellular signaling proteins, NEMO and RIPK1, and induces their aggregation via the induced protein aggregation motif (IPAM) of M45 [86]. The aggregation of NEMO blocks NF-kB signaling, and the aggregation of RIPK1 blocks necroptosis, thereby contributing to immune evasion and cell viability, and to increased virus replication in mouse embryonic fibroblasts (MEF) [86]. Mechanistically, the complex of M45 with NEMO and RIPK1 recruits the retromer component vacuolar protein sorting 26B (VPS26B) and TBC1 domain family member 5 (TBC1d5), the latter containing two LC3-interacting region (LIR) motifs, to facilitate the selective autophagy.

Nucleophagy
Nucleophagy is a selected form of autophagy that removes nuclear components [88]. While autophagic degradation of entire nuclei was reported in the filamentous fungus Aspergillus oryzae that has multi-nucleated hyphae [89], it is unlikely to occur in any mononuclear cell, as this would be lethal. Some studies have suggested that selective autophagic degradation of chromatin and nuclear lamina could play a role in preventing tumorigenesis [90,91].
Autophagic degradation of nuclear materials may be either proviral or antiviral in herpesvirus infection. The nucleus is the site of herpesvirus latency and replication. Therefore, it is conceivable that the host cell may activate nucleophagy to directly remove herpesvirus genomic DNA or proteins, located within the nucleus, that are essential for infection. However, such antiviral autophagy has not been reported to date. Instead, autophagic epitope processing of EBV EBNA1, a nuclear protein that maintains the viral genome of EBV as an episome in the host cell nucleus, was reported to be involved in MHC class II presentation of EBNA1 for acquired immune response to EBV-infected cells [92]. In fact, EBNA1 is the first viral protein identified as being degraded by autophagy. However, whether EBNA1 is recognized by the selective autophagy machinery, such as p62 or LC3B, in the nucleus or the cytoplasm after nuclear export, is unknown. Thus, the precise mechanism by which EBNA1 is processed by autophagy remains to be elucidated.
On the other hand, EBV BFRF1, a major component of nuclear egress complexes, is expressed at the early stage of lytic replication of EBV, localizes at the nuclear envelope, and facilitates the translocation of vesicles from the nucleus to cytoplasm [93][94][95][96]. Thus, some aggregated nuclear proteins, which may be harmful to virus-infected cells, can be engulfed by the BFRF1-derived vesicles and moved to the cytoplasm, where they are degraded in an autophagy-dependent manner [97], suggesting that BFRF1-mediated nucleophagy plays a proviral role in EBV productive replication. Mechanistically, WDFY3 and p62/SQSTM1, which are involved in aggrephagy activation as mentioned above, recognize aggregated nuclear proteins and facilitate BFRF1-induced protein degradation. It is noteworthy that BFRF1 is found to have an important role for virion egress from the nucleus via BFRF1-derived vesicles [98]. Therefore, it is necessary to determine how BFRF1-mediated trafficking of EBV virions evade autophagic degradation. On the other hand, it is known that autophagic degradation of nuclear laminar proteins, including lamins A/C, B1, and B2, facilitates the nuclear egress of HSV-1 in permissive immature dendric cells (DCs), but not mature DCs [99], suggesting that nucleophagy is required for HSV-1 replication in a cell-type dependent manner.

Ferritinophagy
Ferritinophagy is a selected form of autophagy that degrades the iron-sequestering protein ferritin [100]. Iron is a crucial component of various enzymes and proteins, making it essential for several cellular processes. However, excessive free iron induces oxidative stress by promoting ROS production and is harmful to the cell [101]. Ferritin acts as a buffer for iron when cellular iron levels are high. Conversely, when iron levels are low, ferritin is degraded by ferritinophagy to release iron [100]. Upon iron depletion, nuclear receptor coactivator 4 (NCOA4) is stabilized, specifically binds to the heavy chain of ferritin, and marks ferritin as autophagic cargo, allowing it to be selectively degraded [102]. However, NCOA4 has no conventional LIR motif, which is found in other autophagy receptors and required for binding to the ATG8-containing phagophore membrane. Interestingly, a recent study discovered that the NCOA4-ferritin complex is targeted to lysosomes via the non-canonical autophagy pathway mediated by ATG9A and VPS34 proteins [103].
Ferritinophagy is proposed as a potential host antiviral response to HCMV infection. A recent study showed that premature cell death induced by HCMV infection in MRC5 human lung fibroblasts is inhibited by its gene product pUL38, which interacts with the host protein ubiquitin-specific protease 24 (USP24) and then leads to a decrease in NCOA4 stability, thereby resulting in the inhibition of ferritinophagy [104]. Consistent with this, blockage of ferritinophagy by genetic ablation of NCOA4 prevented pUL38-deficient HCMV infection-induced cell death [104]. Therefore, pUL38 plays an important role in the inhibition of NCOA4-mediated ferritinophagy, leading to increased cell viability and successful virus infection.

Viral Xenophagy (Virophagy)
Xenophagy is a common term used for selective autophagic degradation of intracellular pathogens, including bacteria, viruses, and fungi, which comprises an important part of the host immune response [105]. Therefore, viral xenophagy (virophagy) refers to the host antiviral response that removes virions or, conceivably, viral components (proteins or genetic materials) in response to virus infection or replication. Following bacterial infection, proteins on the surface of bacteria are rapidly ubiquitinated and recognized by SLRs, including p62/SQSTM1, NDP52, and OPTN [106][107][108]. However, the molecular mechanisms by which whole virion particles are recognized and targeted to autophagosomes by the autophagy machinery have been relatively understudied.
A recent genome-wide screening study showed that the DNA repair genes, Fanconi anemia complementation group A (FANCA) and C (FANCC), are involved in the autophagic degradation of Sindbis virus (enveloped RNA virus) as well as HSV-1 [109]. Importantly, Sumpter and colleagues demonstrated that FANCC binds biochemically to the Sindbis virus capsid protein, indicating that FANCC recognizes the de-enveloped nucleocapsid of Sindbis virus in the cytosol [109]. However, it is not known whether FANCC binds to the nucleocapsid or tegument proteins of HSV-1, whereas an in vivo study verified that FANCC-deficient mice are more susceptible to lethal central nervous system infection by HSV-1, compared to wild-type mice [109]. It is noteworthy that the study used the HSV-1 mutant virus (ICP34.5 ∆68-87 ), in which amino acids 68-87 required for Beclin 1 binding of and autophagy inhibition by the HSV-1 neurovirulence protein ICP34.5 is deleted. On the other hand, FANCC is recruited to mitochondria and interacts with Parkin for mitophagy activation, indicating that FANCC-activated mitophagy is dependent on ubiquitination [109]. However, it is not known if FANCC-mediated virophagy is dependent on ubiquitination. An image-based genome-wide siRNA screen found that the HECT-type E3 ubiquitin ligase SMURF1 is required for the autophagic targeting of both Sindbis and HSV-1 viruses [110]. Therefore, it is likely that FANCC and SMURF1 physically and/or functionally interact each other to facilitate virophagy. Intriguingly, a recent genome-wide short interfering RNA screen study showed that virion-containing endosomes can be targeted for virophagy. The endosomal protein sorting nexin 5 (SNX5) interacts with ATG14-containing class III phosphatidylinositol-3-kinase complex 1 (PI3KC3-C1) upon HSV-1 (ICP34.5 ∆68-87 ) infection and increases the lipid kinase activity of PI3KC3-C1, which in turn promotes endosomal generation of phosphatidylinositol-3-phosphate (PtdIns(3)P), and recruitment of (PtdIns(3)P)-binding protein WIPI2 to virion-containing endosomes [111].
In addition to the autophagic recognition of virions, viral components, such as proteins and the genomes of RNA and DNA viruses, may be recognized by autophagy receptors or associated proteins for degradation. DN-autophagy and RN-autophagy are selected forms of autophagy that remove DNA and RNA, respectively, in the cytosol [112]. The lysosomal proteins LAMP2C and SIDT2 are known to bind to both cytosolic DNA and RNA and mediate the direct transport of nucleic acids into lysosomes [113,114]. However, to date, it is unknown whether viral RNA and DNA in the cytosol are cleared by selective autophagy. Interestingly, HSV-1 infection-induced autophagy in macrophages is known to be dependent on STING (stimulator of IFN genes) [115], which is an adaptor protein that is activated by cyclic guanosine monophosphate-adenosine monophosphate (cGAMP), produced by the DNA sensor cGAMP synthase [116]. STING-mediated autophagy is known to restrict HSV-1 infection independently of a STING-mediated innate immune response that leads to the expression of type-I IFNs [117]. In fact, RNA interferencemediated knockdown of TBK1 and IRF3, key components of innate immunity, does not affect STING-mediated autophagy in HeLa cells [118]. Instead, STING dimerization and binding to LC3 via its LIR motifs are required for autophagy activation [118]. Overall, these results suggest that viral DNA in the cytosol can elicit the host antiviral autophagy mediated by STING.
On the other hand, the role of TBK1 in STING-mediated autophagy seems controversial in HSV-1 infection [118,119]. A recent study demonstrated that TBK1 is required for early autophagy induction, in a STING-dependent manner, upon HSV-1 infection using TBK1 +/− dermal fibroblasts from herpes simplex encephalitis with autosomal dominant TBK1 containing a dominant-negative G159A mutation [119]. Interestingly, Ahmad and colleagues found two LC3B phenotypes: perinuclear LC3B puncta (also known as nuclear envelope-derived autophagy, NEDA) and cytoplasmic LC3B puncta. The latter puncta are formed early upon HSV-1 infection or cyclic diguanylate monophosphate (c-di-GMP) transfection in wild-type, but not TBK1 +/− dermal fibroblasts and involved in cytoprotection from HSV-1 infection. This study highlights a possibly cytoprotective role for TBK1 in HSV-1-induced cytoplasmic autophagy. Other evidence supports the role of TBK1 in autophagy-mediated restriction of HSV-1; tripartite motif protein 23 (TRIM23), a RINGtype E3 ubiquitin ligase with ADP-ribosylation factor (ARF) GTPase activity, is known to activate TBK1 on the autophagosome membranes, which in turn phosphorylates p62 to induce autophagy [120,121]. Overall, TBK1 requirement of STING-mediated autophagy may be dependent on cell type.

Viral Evasion of Antiviral Autophagy
Viruses have evolved the strategies to evade antiviral autophagy or virophagy by targeting the autophagy core machinery [8][9][10][11]. For example, the HSV-1 neurovirulence protein ICP34.5 prevents formation of the phagophore by binding to Beclin 1, a key factor involved in the elongation of the phagophore membrane, and by redirecting the protein phosphatase 1α (PP1α) to dephosphorylate translation initiation factor 2α (eIF2α) in neuron cells and fibroblasts [122,123]. The Us11 protein is also known to inhibit autophagy by direct interacting with and inhibiting the double-stranded RNA-dependent protein kinase PKR in HeLa cells and fibroblasts, thereby reducing the level of phosphorylated eIF2α [124]. In addition, the ICP0 protein downregulates SLR autophagy receptors p62/SQSTM1 and OPTN during the early stages of HSV-1 infection in various cell lines, including human embryonic lung fibroblasts (HEL), human epithelial cells (HEp-2), human microglia cell line 3 (HMC3), human osteosarcoma cell line (U2OS), but not human monocyte cell line (THP1) [125]. Furthermore, the Us11 protein excludes TBK1 from the TRIM23/HSP90 complex by binding to the ARF domain of TRIM23, thereby inhibiting autophagy-mediated restriction of HSV-1 infection in MEF and HEL [126]. These results indicate that HSV-1 has developed multiple strategies to incapacitate autophagy to promote virus replication. For HCMV, autophagy is rapidly induced upon virus infection in MRC5 cells but inhibited by the HCMV-encoded proteins TRS1 and IRS1, which blocks autophagosome biogenesis by interacting with Beclin 1 [127]. For HHV-8, the viral proteins vFLIP and vBcl-2 can suppress autophagy by preventing ATG3 from binding and processing LC3 [128], and by inhibiting Beclin 1 [129], respectively.

Conclusions
A variety of selective autophagy mechanisms have been identified, and more will be revealed upon the discovery of specific cargo and autophagy receptor partners. Selective autophagy contributes to cellular homeostasis via quality and quantity control of the specific cargo. Viruses have evolved strategies to activate or inhibit selective autophagy for their own benefit, that is, to achieve successful infection and replication (Figure 1). Complex and varied mechanisms of herpesvirus regulation of autophagy are likely a consequence of cell-type specificities, infection type (de novo, lytic, latent), kinetic stage, and particular subfamily associated herpesvirus biologies. In this paper, we have focused on the molecular mechanisms and functional implications of herpesvirus regulation of mitophagy, aggrephagy, nucleophagy, ferritinophagy, and virophagy. In addition to elucidating further details about these processes, it is important to also investigate the role and regulation of other selective autophagy pathways, including pexophagy, lysophagy, lipophagy, reticulophagy, and glycophagy, in herpesvirus infection and replication.