Varicella-zoster virus: molecular controls of cell fusion-dependent pathogenesis

Abstract

Varicella–zoster virus (VZV) is the causative agent of chicken pox (varicella) and shingles (zoster). Although considered benign diseases, both varicella and zoster can cause complications. Zoster is painful and can lead to post herpetic neuralgia. VZV has also been linked to stroke, related to giant cell arteritis in some cases. Vaccines are available but the attenuated vaccine is not recommended in immunocompromised individuals and the efficacy of the glycoprotein E (gE) based subunit vaccine has not been evaluated for the prevention of varicella. A hallmark of VZV pathology is the formation of multinucleated cells termed polykaryocytes in skin lesions. This cell–cell fusion (abbreviated as cell fusion) is mediated by the VZV glycoproteins gB, gH and gL, which constitute the fusion complex of VZV, also needed for virion entry. Expression of gB, gH and gL during VZV infection and trafficking to the cell surface enables cell fusion. Recent evidence supports the concept that cellular processes are required for regulating cell fusion induced by gB/gH–gL. Mutations within the carboxyl domains of either gB or gH have profound effects on fusion regulation and dramatically restrict the ability of VZV to replicate in human skin. This loss of regulation modifies the transcriptome of VZV infected cells. Furthermore, cellular proteins have significant effects on the regulation of gB/gH–gL-mediated cell fusion and the replication of VZV, exemplified by the cellular phosphatase, calcineurin. This review provides the current state-of-the-art knowledge about the molecular controls of cell fusion-dependent pathogenesis caused by VZV.

Varicella-zoster virus

Varicella-zoster virus (VZV) is a medically important, human host-restricted pathogen classified in the subfamily Alphaherpesvirinae of the Herpesvirdae [1]. Herpesviruses have double-stranded DNA genomes that are encapsulated within an icosahedral capsid, which is surrounded by a proteinaceous tegument layer wrapped in a lipid bilayer, termed the envelope. The envelope is studded with virally encoded glycoproteins required for attachment and entry. VZV has a 125 kbp genome encoding 71 open reading frames (ORFs), of which 10 are translated to produce glycoproteins: ORFS/L (ORF0), gK (ORF5), gN (ORF9a), gC (ORF14), gB (ORF31), gH (ORF37), gM (ORF50), gL (ORF60), gI (ORF67) and gE (ORF68) [2–4]. Critically, as for all herpesviruses, VZV relies on a fusion complex comprised of three core glycoproteins, gB, gH and gL, required for entry of virions into host cells. Upon cell entry, the capsid traffics to the cell nucleus where it docks with a nuclear pore to deliver the DNA genome. Simultaneously, regulatory proteins from the tegument translocate to the nucleus where the ORFs of the VZV genome are transcribed in a temporal cascade to produce proteins required for genome replication, capsid assembly and nuclear egress of nascent capsids. Capsids undergo primary envelopment and de-envelopment then traffic to sites of secondary envelopment where the VZV lipid bilayer is acquired from cellular membranes at the trans-Golgi network. Newly synthesized virus particles are then transported by intracellular vesicles to the cell surface and released into the extracellular space. After primary infection, VZV remains in a latent state in dorsal root ganglia (DRG) and can reactivate to cause localized zoster or disseminated infection.

VZV pathogenesis

VZV is transmitted by aerosolized droplets and direct contact with skin lesions, leading to varicella, known as chicken pox, which is usually acquired early in life in the absence of vaccine programs [5]. VZV is highly transmissible with a basic reproduction number (R₀; the expected number of cases directly generated by one case in a population where all individuals are susceptible to infection) reported to be from 3.7 to 5 for varicella [6]. With a typical incubation period of 10–21 days, varicella starts with a mild fever then manifests as a pruritic maculopapular rash distributed across the body that rapidly progresses to vesicular lesions before crusting. These skin lesions are a source of highly infectious virus at the vesicular stage and contain cells that have become fused together forming characteristic polykaryocytes. Complications requiring hospitalization occurred in 2.3–6.3 per 1000 cases before varicella immunization was made universal in the U.S.A. with hospitalization rates declining by 75–88% after mass vaccination programs [7]. Varicella complications are due directly to the virus, including pneumonia, cerebellitis, encephalitis, meningitis, facial palsy, acute retinitis and others, as well as to secondary bacterial infections causing cutaneous complications, arthritis, osteomyelitis, necrotizing fasciitis, pre-septal and orbital cellulitis, and pneumonia [8]. Long-term sequelae from varicella are rare and are primarily due to neurological damage.

VZV can reactivate to cause zoster, which typically occurs later in life. Prior to the availability of the zoster vaccine, the average incidence of herpes zoster was 3.4, and 3.6 per 1000 person per year in the U.K. and U.S. A. respectively [9,10]. The incidence rate increases with age, to more than 10 per 1000 person per year by the age of 75, with 68% of cases diagnosed in people above 50 years of age. Zoster is linked with the reduction in VZV-specific CD4T cell frequencies associated with aging. VZV reactivation can progress to infection of the skin of the dermatome innervated by the sensory nerve ganglion where reactivation has taken place. Zoster lesions appear as vesicles that coalesce to affect much or all of the affected dermatome. Zoster skin lesions also contain polykaryocytes. Importantly, the long-term effects of zoster can be debilitating via the development of a persistent nerve pain known as post herpetic neuralgia (PHN), which occurs in 10–20% of zoster patients aged 60 years or older [10]. For some patients, this pain might last for months and even years after the episode of zoster. Unfortunately, this condition is refractory to treatment with pain medications due to damage to sensory ganglion neuronal cell bodies and satellite cells within the ganglion and to the neuronal axons caused by VZV infection.

VZV is associated with infection of vascular endothelial cells, which could lead to cardiovascular disease, stroke and acute vision loss [11]. Giant cell arteritis (GCA) is a vascular inflammatory disease where multinucleated giant cells derived from cell-cell fusion (abbreviated as cell fusion) of infiltrated activated macrophages are typically identified in the wall of temporal arteries [12]. Importantly, VZV antigen and DNA have been detected in histology sections at areas adjacent to GCA pathology [13,14]. While the etiology of GCA remains obscure, this provides evidence that VZV might be a contributing factor to the pathogenesis of GCA in some cases.

Critically, it has been possible to model the pathogenesis of VZV infection in human tissue xenografts using the severe combined immunodeficiency (SCID) mouse model. Skin or sensory ganglia tissue can be engrafted into SCID mice, and these tissues can be inoculated with VZV. Replication of VZV in the skin produces lesions reminiscent of those produced in natural infection [15]. In DRG xenografts, initial productive infection of VZV is followed by a transition to limited expression of viral transcripts and cessation of viral particle production, [16]. In both of these models of VZV pathogenesis, cell fusion is evident by the abundance of multinucleated cells [17–19].

Vaccines

Vaccines are available for the prevention of both varicella and zoster. The attenuated varicella vaccines derived from the Oka strain of VZV (e.g. Varivax, Merck, U.S.A.; Varilrix, GSK, U.K.), have been implemented as a universal routine childhood vaccine in the U.S.A., Canada, Australia, Japan and some countries in Europe and the Middle East [7,20–22]. These attenuated vaccines have reduced the incidence of varicella by >90%, its associated complications by 57–90%, and hospitalizations by 75–88% [7,23]. However, these vaccines are not safe for individuals with immunodeficiencies (e.g. HIV infection, malignancy, immunosuppression after transplantation), and the vaccine virus can reactivate to cause zoster [24–26]. Currently, there are two different zoster vaccines approved by the FDA, an attenuated vaccine derived from the same Oka strain used for varicella vaccines (Zostavax, Merck, U.S.A.) and a vaccine comprised of the recombinant form of VZV glycoprotein gE and the ASO3 adjuvant (Shingrix, GSK, U.K.) [27,28]. While providing 70% protection against zoster and 67% against PHN in adults aged 50–59 years old, the efficacy of the attenuated zoster vaccine decreases significantly as age increases, with only 38% protection in adults aged 70 and older [28,29]. In contrast, the gE subunit zoster vaccine has proven to be extremely efficacious in an age-independent fashion, reducing the incidence of zoster by 97% in individuals aged 50–69 years old, and 90% in individuals aged 70 and older. The gE subunit zoster vaccine has also been shown to reduce the risk of PHN by 89% in individuals 70 years and older [30,31]. The noninfectious nature of the gE subunit vaccine makes it safe for preventing zoster in immunocompromised adults when their immune system is sufficiently capable of responding to immunization. However, it is not clear whether this subunit vaccine will have the same level of efficacy to prevent varicella. Despite these successes, the current vaccines still leave behind vulnerable populations, which need new drugs to reduce pathogenesis or prevent disease. Since cell fusion plays an important role in the pathogenesis of VZV, therapeutic agents that disrupt cell fusion could provide significant advances in alleviating disease.

VZV-induced cell–cell fusion

The virus entry functions of the conserved herpesvirus glycoproteins, gB, gH and gL are a major focus of herpesvirus research [32–49]. VZV gB, the herpesvirus fusogen orthologue, and the heterodimer gH–gL constitute the minimal complex required for VZV-induced cell fusion, the same fusion machinery used by the virus during entry into cells [34,50–52]. This necessary and sufficient requirement is distinct from other herpesviruses that utilize accessory proteins to act together with gB/gH–gL for membrane fusion and virus entry [53–68]. However, until recently there has been little focus on the factors involved in cell fusion, which will be the primary emphasis of this review.

Like many other major human viral pathogens such as measles, respiratory syncytial virus and HIV, VZV overcomes the usual constraint against membrane fusion between fully differentiated host cells [69–74]. VZV is a master in promoting cell fusion during its takeover of host cells, which causes the classic polykaryocytes observed in VZV skin lesions [75–77]. Additionally, as a neurotropic virus, VZV can also induce fusion between differentiated neuronal cells and satellite cells, which allows the virus spread to more neurons within the sensory ganglion. Increasing the number of infected neurons would allow VZV transport along more neuronal axons to skin during zoster [18]. Interestingly, although VZV has been demonstrated to induce cell fusion in a range of cell types, evidence for such a phenomenon in T cells, critical agents of VZV dissemination throughout the body, has not been apparent to date [15,78,79].

Mechanistically, cell fusion is initiated by the formation of a fusion pore between two opposing cell membranes, followed by an expansion of the fusion pore that leads to complete fusion of the plasma membranes, cytoplasmic content mixing and formation of multinucleated syncytia. VZV gB and gH–gL traffic to the cell surface during infection, where interactions with cellular receptors on membranes of adjacent cells likely initiates the fusion pore formation. Syncytia formation in MeWo cells models the polykaryocytes seen within infected skin tissue [17,80]. Live cell imaging of MeWo cells infected with VZV illuminates the speed with which cell fusion occurs [80]. Once cell membrane breakdown occurs during the fusion process, the nucleus from the fusing cell is rapidly sequestered to the center of the syncytium. The molecular basis for this is not understood.

To shed further light on the role of the gB/gH–gL fusion complex in VZV pathogenesis, complementary methods have been developed as outlined in Figure 1. The roles of the fusion complex proteins have been investigated in the context of cell fusion in the absence of other viral proteins, during VZV replication in cultured cells, and infection of human tissue xenografts in the SCID mouse model, in experiments using targeted mutagenesis as a tool to map functional domains as well as evaluation of bioactive compounds and neutralizing antibodies (Figure 1A). Coupled with structural biology, the ability to quantify cell fusion has transformed the understanding of the mechanics underlying herpesvirus entry into cells and the quantitative cell fusion assay provides a powerful virus-free model for assessing the role of individual glycoproteins and their subdomains or glycoprotein complexes in mediating cell fusion independently of other viral proteins (Figure 2; Tables 1 and 2). Advances in recombinant technologies have enabled the development of several bacterial artificial chromosome (BAC) systems for VZV [81–84]. The self-excisable VZV BAC system, pOka-DX, which contains the entire 125kbp DNA genome of the pOka strain of VZV, can be manipulated in bacteria using Red recombination and was designed to allow markerless mutagenesis [83,85]. The purified BAC DNA is then used to transfect MeWo cells to recover infectious VZV (Figure 1B). The ability to perform mutagenesis in the VZV genome, initially implemented with cosmids and now using the BAC system [86,87], provides a necessary complementary experimental approach to study the effects of fusion modifying mutations on viral replication. Finally, the contributions of gB and gH–gL to cell fusion and its regulation has been defined using VZV mutants to assess effects on skin pathogenesis in a SCID-Hu mouse model (Figure 1C). Human skin tissue is engrafted into SCID mice where it becomes vascularized and suitable for inoculation with BAC-derived VZV mutants in order to compare their replication and capacity to form polykaryocytes to wild-type (WT) VZV. These technologies have also paved the way to identify the roles of host cell factors involved in VZV-induced cell fusion.

Figure 1. Quantitative assays used to assess the effects of mutagenesis on the VZV fusion complex.

(A) The stable reporter fusion assay (SRFA) is a quantitative cell fusion assay based on the constitutive expression of GFP-luciferase dual split proteins (DSP) to identify neutralizing antibodies, compounds that alter fusion and receptors for membrane fusion. Effector cells (square) that constitutively express the DSP1 reporter (black hexagons) and transiently express the fusogen (white circles) are co-cultured with target cells (ellipse) that constitutively express the cellular receptor(s) (orange) or co-receptors needed for membrane fusion and the DSP2 reporter (purple hexagons). The fusion between the effector and target cells (green) leads to the reconstitution of the DSP and GFP and renilla luciferase activity from the reconstituted DSP, allowing for a dual readout of fusion. Flow cytometry can be used to quantify GFP (fusion events) whereas luminescence can be used to quantify the level of fusion in a population of cells. Fusion inhibition — neutralizing antibodies (red Ys) that bind to the VZV fusion complex reduce or inhibit fusion. Bioactive compound screen — in the high-throughput format the SRFA is adapted to 384-well plates to assess the effects of compounds with known biological activity, including those that increase cell fusion. Receptor knockdown — target cells that simultaneously and constitutively express DSP2 and short hairpin RNAs (red hairpin) can be generated using a single lentiviral vector. The knockdown of a required receptor for fusion by shRNA would reduce or inhibit fusion. (B) Mutagenesis of the VZV genome using the BAC system. The 125 kbp VZV genome is carried on a bacterial artificial chromosome (BAC). Transfection of the VZV BAC into permissive cells results in the production of infectious VZV (WT). The VZV BAC can be used for mutagenesis; a VZV ORF carrying a specific mutation, an I-SceI site and a resistance cassette are incorporated into the VZV BAC. A red recombination step is performed that enables markerless removal of the I-SceI site and resistance cassette. The mutant VZV BAC can then be used to transfect cells and produce mutant VZV if the mutation is not inactivating. To evaluate the ‘fitness’ of the mutant, VZV plaque sizes and replication kinetics are quantified. (C) The severe combined immunodeficient (SCID) mouse model of VZV pathogenesis. VZV is a human host-restricted pathogen. To overcome this, fully differentiated human skin or neuronal tissue is engrafted into SCID mice where they become vascularized and can remain in the mice for several weeks. The xenografts are inoculated with WT or mutant VZV. To determine the level of viral replication, the xenografts are processed and the VZV titers are determined.

Figure 2. Mutagenesis of the VZV fusion complex disrupts function.

(A) X-ray crystallography structures of the VZV gB trimer (2.4 Å) and the gH–gL heterodimer (3.2 Å). The structures of VZV gB and gH–gL are represented as ribbons with each domain colored accordingly: gB structure DI (cyan), DII (green), DIII (yellow), DIV (orange), DV (red), linker regions (hot pink); gH–gL structure DI (green), DII (yellow), DIII (orange) and gL (cyan). (B) Linear diagrams for gB, gH and gL mapping the sites of mutagenesis that effect cell fusion. All domains are colored as for the X-ray crystallography structures in (A). Other features include the signal sequence (light gray box) the transmembrane domain (dark gray box) and sites of mutagenesis. Mutations that affect cell fusion are either shown above (single) or below (motifs) the diagrams. Glycosylation sites are indicated by black bars with black circles. For gH, the disrupted α-helices (8, 9, 12 and 14) that abolished fusion are depicted as plum-colored boxes. The disulfide bonds in gH that were mutated are shown with connecting lines.

Table 1.

The effects of gB mutagenesis on cell fusion, VZV replication in cell culture and skin pathogenesis

			Phenotype
Mutation	Domain	Cell fusion	Cell culture	Plaque size	Skin pathogenesis	Reference
ΔgB	N/A	N/A	Inactivating	N/A	N/A	[4]
K109A	N-terminus	Reduced	Inactivating	N/A	N/A	[96]
K109R	N-terminus	Reduced	Inactivating	N/A	N/A	[96]
S110A	N-terminus	Reduced	WT	WT	ND	[96]
Q111A	N-terminus	Reduced	WT	WT	ND	[96]
D112A	N-terminus	Reduced	WT	WT	ND	[96]
109AAAA112	N-terminus	Reduced	WT	Reduced	ND	[96]
W180G	I	Abolished	Inactivating	N/A	N/A	[17,50]
Y185G	I	Abolished	Inactivating	N/A	N/A	[17,50]
491GSGG494	Linker	ND	WT	WT	Reduced	[17,50]
491ΔRSRR494	Linker	ND	WT	WT	Reduced	[17,50]
S589A	IV	WT	WT	WT	ND	[93]
R592A	IV	Reduced	WT	WT	ND	[93]
I594A	IV	Reduced	WT	WT	ND	[93]
589AAA594	IV	Reduced	Inactivating	N/A	N/A	[93]
Q596A	IV	Reduced	WT	Reduced	ND	[93]
N597A	IV	Reduced	Inactivating	N/A	N/A	[93]
596AA597	IV	Reduced	Inactivating	N/A	N/A	[93]
592A/596AA597	IV	Reduced	Inactivating	N/A	N/A	[93]
Y667A	IV	Reduced	WT	WT	ND	[93]
E670A	IV	Reduced	Inactivating	N/A	N/A	[93]
667A/A670	IV	Abolished	WT	WT	ND	[93]
592A/596AA597/667A/A670	IV	Abolished	Inactivating	N/A	ND	[93]
Δ36 (gB-36)	gBcyt	ND	Hyperfusion	ND	N/A	[97]
Y881F	gBcyt	Hyperfusion	Hyperfusion	Reduced	Reduced	[17]
Y881W	gBcyt	Hyperfusion	Hyperfusion	ND	ND	[17]
Y881D	gBcyt	Abolished	Inactivating	N/A	N/A	[17]
Y881E	gBcyt	Abolished	Inactivating	N/A	N/A	[17]
L884G	gBcyt	Abolished	Inactivating	N/A	N/A	[17]
Y920F	gBcyt	WT	WT	WT	WT	[17]
Y881/920F	gBcyt	Hyperfusion	Hyperfusion	Reduced	Reduced	[17]
gB[4R]	gBcyt	WT	Hyperfusion	Increased	ND	[98]
gB[4A]	gBcyt	Hyperfusion	Hyperfusion	Reduced	ND	[98]
gB[Y881F/4R]	gBcyt	Hyperfusion	Hyperfusion	Reduced	ND	[98]
gB[Y881F/4A]	gBcyt	WT	Hyperfusion	Reduced	ND	[98]
N-linked Glyc.1						[95]
N557A	III	Decreased	N/A	N/A	N/A	[95]
O-linked Glyc.1						[95]
T129A	IV	Decreased	N/A	N/A	N/A	[95]
T265A	I	Decreased	N/A	N/A	N/A	[95]
S559A	III	Decreased	N/A	N/A	N/A	[95]

WT, wild-type, not different from parental virus; N/A, not applicable; ND, not determined.

¹Extensive mutagenesis studies that target predicted N- and O-linked glycosylation sites were performed by Suenaga et al. [95]. For brevity, only those with statistical significance (P =< 0.01) for the 7 N-linked and 37 O-linked substitutions performed are provided in Table 1.

Table 2.

The effects of gH mutagenesis on cell fusion, VZV replication in cell culture and skin pathogenesis

			Phenotype
Mutation	Domain	Cell fusion	Cell Culture	Plaque Size	Skin Pathogenesis	Reference
ΔgH	N/A	N/A	Inactivating	N/A	N/A	[4,51]
αX	I	Reduced	WT	WT	Reduced	[51]
S42A	I	ND	WT	WT	WT	[51]
N45A	I	ND	WT	WT	WT	[51]
S47A	I	Reduced	WT	WT	Reduced	[51]
S47T	I	ND	WT	WT	WT	[51]
T127A	I	ND	WT	WT	WT	[51]
W291A	I	WT	ND	ND	ND	[104]
F292A	I	WT	ND	ND	ND	[104]
291AA292	I	WT	ND	ND	ND	[104]
Loop A	I	WT	ND	ND	ND	[104]
C327A	II	ND	WT	WT	WT	[51]
T351A	II	ND	WT	WT	WT	[51]
α8	II	Abolished	Inactivating	N/A	N/A	[51]
α9	II	Abolished	Inactivating	N/A	N/A	[51]
α12	II	Abolished	Inactivating	N/A	N/A	[51]
α14	II	Abolished	Inactivating	N/A	N/A	[51]
C540A	II	Abolished	Inactivating	N/A	N/A	[51]
C575A	II	Abolished	Inactivating	N/A	N/A	[51]
C647A	III	Abolished	Small Plaque	Reduced	Reduced	[51]
S694F	III	Increased	ND	ND	ND	[51]
C703A	III	Abolished	Small Plaque	Reduced	Reduced	[51]
C724A	III	Abolished	Inactivating	N/A	N/A	[51]
S694F/C724A	III	Reduced	ND	ND	ND	[51]
C727A	III	Abolished	Small Plaque	Reduced	Reduced	[51]
S694F/C727A	III	Reduced	WT	WT	Reduced	[51]
T751A	III	Reduced	Small Plaque	Reduced	WT	[51]
781FPNG784	III	Abolished	Inactivating	N/A	N/A	[51]
gH-TL (834 stop)	gHcyt	WT	Hyperfusion	WT	Reduced	[105]
Δ834–841	gHcyt	WT	Hyperfusion	WT	Reduced	[105]
834StopV5	gHcyt	WT	Hyperfusion	WT	ND	[105]
gH-V5	gHcyt	Inactive	WT	WT	WT	[105]
gH[WT]	gHcyt	Inactive1	WT	WT	WT	[105]
Y835A	gHcyt	Inactive1	WT	WT	WT	[105]
Y835F	gHcyt	Inactive1	WT	WT	WT	[105]
gB[Y881F]/gH[Δ834–841]	gBcyt/gHcyt	Hyperfusion	Inactivating	N/A	N/A	[105]

WT, wild-type, not different from parental virus; N/A, not applicable; ND, not determined.

¹The VZV fusion assay is based on the gH-TL (834 stop) construct. This substitution is required for fusion activity in the cell fusion assay. Both Y835A and Y835F substitutions were based on the gH[WT] construct that does not have the 834 stop mutation.

VZV gB functions in cell–cell fusion and its regulation

The gB molecule

VZV ORF31 encodes gB, producing a 931aa N- and O-linked glycosylated protein of ~130 kDa that is proteolytically cleaved by the protease furin at the ⁴⁹¹RSRR⁴⁹⁴ recognition site [50,88]. The molecular structure of VZV gB in its post fusion conformation is similar to other herpesvirus orthologues (Figure 2), forming a trimer as determined by both cryogenic electron microscopy (cryo-EM) and X-ray crystallography [89–94]. In addition to the five structurally solved domains (DI-DV), there are the N-terminal domain and the carboxyl-terminal domain (CTD), for which the structure remains elusive to date due to flexibility in these regions. DI contains the two fusion loops that embed into the opposing membrane, predicted to trigger the outer lipid leaflets of the cell and virion membranes to undergo hemifusion followed by fusion pore formation during a conformational change of the gB trimer. VZV gB is a target for human neutralizing antibodies (Figure 3A).

Figure 3. VZV neutralizing antibodies target the gB/gH–gL fusion complex.

(A) Cryo-EM structure (2.8 Å) of native VZV gB in complex with human neutralizing mAb 93k Fab fragments [93]. Representative 2D class averages (left hand upper and lower panels) used to generate the 2.8 Å cryo-EM structure of the gB-93k complex (middle upper and lower panels). The gB trimer (gray) and the 93k Fab fragments (blue) are segmented. The right hand panel shows a segmentation of the cryo-EM map for one VH and VL chain of a 93k Fab fragment bound to a protomer of VZV gB. The structures of VZV gB and 93k VH and VL are represented as ribbons with each domain colored accordingly: DI (cyan), DII (green), DIII (yellow), DIV (orange), DV (red), linker regions (hot pink), 93k VH (blue) and 93k VL (light blue). Expanded view of the gB-93k interface represented as a ribbon diagram with amino acids in stick format showing amino acid interactions between gB and mAb 93k. The β23, β25–26 and β29–30 of gB are highlighted with orange boxes, and the VHCDR1, VHCDR3, VLCDR1 and VLCDR2 are highlighted by blue boxes; VH — dark blue, VL — light blue. The gB DIV residues R592, I594, Q596, N597, Y667 and E670 central to 93k binding and gB function are highlighted by orange boxes. (B,C) X-ray crystallography structures of VZV gH–gL in complex with human neutralizing mAbs RC (3.1 Å) and 94 (3.6 Å) Fab fragments [104]. The gH–gL heterodimer and either mAb RC (B) or mAb 94 (C) are represented as ribbons and each domain colored accordingly: DI (green), DII (yellow), DIII (orange), gL (cyan), VH (blue) and VL (light blue). Expanded views on the right show the interface for gH–gL–RC (B) and gH–gL–94 (C) with amino acids in stick format showing amino acid interactions between gH–gL and the mAbs. The gH DI loop A residues W291 and F292 critical for RC and 94 binding are highlighted by the green boxes.

gB functions in cell fusion

Mutagenesis of the VZV gB ectodomain has targeted glycosylation sites, the fusion loops, the furin cleavage site, DIV, the flexible N-terminus and the CTD (Figure 2; Table 1). For the most part, the substitution of VZV glycosylation sites had little effect on gB fusion function with the exception of four substitutions at N-linked glycosylation site N557A, and O-linked glycosylation sites T129A, T265A and S559A [95]. Although these substitutions reduced cell fusion their effects on virus replication was not studied. In contrast, substitutions in the gB fusion loop, W180G and Y185G, both rendered VZV noninfectious and abolished cell fusion without affecting the level of gB on cell surfaces, demonstrating the importance of the fusion loops in both virus-induced cell fusion and the virus entry process [17,50]. A more subtle phenotype was uncovered by the mutagenesis of the gB furin cleavage site. Two VZV mutants, ⁴⁹¹GSGG⁴⁹⁴ and ⁴⁹¹ΔRSRR⁴⁹⁴, which prevent furin cleavage of gB, did not affect cell fusion and replicated in MeWo cell monolayers to similar titers compared with WT pOka [17,50]. Conversely, these mutants were impeded in their pathogenic capacity as both replicated to significantly reduced titers in human skin xenografts compared with WT pOka [50]. However, the role of furin in VZV replication within differentiated skin tissue remains unclear.

A 2.8 Å resolution cryo-EM structure of native, full-length VZV gB in complex with Fab fragments from a neutralizing mAb, 93k, revealed that gB DIV is critical for fusion initiation (Figure 3A). Mutagenesis of residues in gB at the mAb 93k binding site disrupted membrane fusion measured by the virus-free stable reporter fusion assay (SRFA) [93]. In addition, BAC mutagenesis of the VZV genome demonstrated the significance of these residues for gB fusion functions necessary to produce infectious extracellular VZV virions. These findings have implications for modeling the transition of gB from prefusion to postfusion conformations. Along similar lines, the VZV gB N-terminus was also shown to have a role in fusion function via mutagenesis of residues ¹⁰⁹KSQD¹¹² that were predicted to form an α-helix [96].

gB functions in the regulation of cell fusion

Evidence for the regulation of VZV-induced cell fusion first emerged via mutagenesis of VZV gB cytoplasmic domain (gBcyt). A truncation mutant, gB-36, exhibited a hyperfusion phenotype for VZV replication in cell culture, with giant fused cells containing hundreds of nuclei that resulted in larger plaques [97]. In subsequent studies, point mutations in the gBcyt also exhibited a hyperfusion phenotype pinpointing specific motifs of the gBcyt in the regulation of cell fusion [17,80,98]. VZV gBcyt contains two endocytosis motifs YXXΦ (X, any amino acid; Φ, a larger hydrophobic residue), ⁸⁸¹YMTL⁸⁸⁴ and ⁹²⁰YSRV⁹²³, which are necessary for gB intracellular trafficking, cell surface retrieval and incorporation into the virion envelope [99]. Oliver et al. [17] reported that two substitutions of tyrosine at position 881, Y881D and Y881E, both abolished cell fusion due to defective intracellular trafficking and limited quantities of cell surface gB, suggesting that the polarity and charge of the side chain at position 881 is essential for gB trafficking and localization. Similarly, L884G prevented cell fusion attributed to the decreased surface expression of gB and corroborating the critical role of the hydrophobic residue at position 884 in the trafficking of gB to plasma membranes. Importantly, Y881F and Y881W induced exaggerated cell fusion without significantly affecting the trafficking of gB to the plasma membrane, suggesting that the hydroxyl moiety in the aromatic sidechain on tyrosine 881 is critical for the regulation of cell fusion. In contrast, Y920F, did not significantly affect cell fusion nor gB surface expression. The ⁸⁸¹YMTL⁸⁸⁴, but not ⁹²⁰YSRV⁹²³, also overlaps with a canonical immunoreceptor tyrosine-based inhibition motifs (ITIMs) [ILV] XYXX[LV], ⁸⁷⁹IKYMTL⁸⁸⁴; ITIMs in host cell proteins are involved in cell signaling cascades that affect cell motility. ITIMs must be tyrosine-phosphorylated to be functional, as was confirmed by the detection of phosphorylation of the gB Y881 during infection by mass spectrometry [17].

Unexpectedly, despite the enhanced syncytium formation in cell culture, gB[Y881F] and gB[Y881/920F] mutant viruses were severely impaired for propagation demonstrated by reduced virus plaque size and decreased virus titers recovered from infected skin xenografts compared with WT pOka [17]. This challenged the long-accepted concept that syncytia formation was beneficial to the spread of VZV, and instead, suggested that cell fusion must be tightly regulated to sustain optimal VZV propagation. Taken together, these studies revealed that gBcyt modulates cell fusion via an ITIM-mediated Y881 phosphorylation-dependent mechanism, which is essential for VZV skin pathogenesis. Additionally, Yang et al. [98] reported that the conserved lysine cluster K894, K897, K898 and K900, downstream of the ITIM motif in gBcyt also played a role in cell fusion regulation. An arginine substitution which has a larger positively charged sidechain in gB[4R], did not affect cell fusion despite reduced quantities of gB on the cell surface, and showed enhanced syncytium formation and a slight increase in plaque size. However, substitutions to alanine, gB[4A], increased cell fusion significantly despite less surface gB. This mutant also resulted in markedly exaggerated syncytia, but reduced plaque size. Combined mutants, gB[Y881F/4R] and gB[Y881F/4A], both had a phenotype similar to gB[4A], resembling the hyperfusion mutant gB[Y881F] [98]. These studies further elucidate how the gBcyt is implicated in the self-regulation of cell fusion to assure VZV propagation.

VZV gH–gL functions in cell–cell fusion and its regulation

The gH–gL heterodimer

VZV ORFs 37 and 60 encode gH (841aa; 118 kDa) and gL (160aa; 20 kDa), respectively, which are N- and O-linked glycosylated and form a heterodimer that is necessary for the activation of gB [3,4,51,100,101]. The molecular structure of VZV gH–gL is similar to herpesvirus orthologues, where gH has three domains (DI-III) and gL co-folds with the gH N-terminal domain (DI) as determined by X-ray crystallography[32,102–104]. In contrast with gB, VZV gH has a short CTD of only 18aa. Like gB, the ectodomain of the gH–gL heterodimer is also a target for neutralizing antibodies (Figure 3B).

gH functions in VZV cell fusion

Mutagenesis of the gH ectodomain has revealed several motifs important for cell fusion [51,104]. Mutation of the first β-strand at ³⁸LREY⁴¹ (αX) by glycine substitutions, ³⁸GRGG⁴¹, resulted in reduced cell fusion without affecting gH surface expression (Figure 2; Table 2). Although plaque formation was unaltered, replication of the mutant virus was impaired in skin xenografts [51]. These data were consistent with the observation that the ³⁸LREY⁴¹ to ³⁸GRGG⁴¹ mutation also significantly reduced the binding affinity of gH neutralizing antibody mAb 206 that prevented gB/gH–gL-mediated cell fusion efficiently [19,51,104]. Thus, ³⁸LREY⁴¹ (αX) is likely part of the epitope recognized by the neutralizing antibody to elicit its inhibition on cell fusion. Disruption of the helices α8 and α14 by introducing proline, and α9 and α12 by alanine substitutions also reduced cell fusion with significantly decreased surface expression of gH, accompanied by the inability to recover mutant viruses from transfected cells [51]. The function of β-turn residues ⁷⁸¹FPNG⁷⁸⁴, T751 within α22, as well as S47 within the predicted N-linked glycosylation motif were also assessed via alanine substitutions. The ⁷⁸¹AAAA⁷⁸⁴ substitutions resulted in abolished cell fusion and rendered VZV noninfectious likely due to diminished quantities of cell surface gH. The T751A had cell surface quantities of gH equivalent to WT but exhibited reduced fusion and smaller plaque formation in cell culture. The S47A substitution reduced cell fusion and significantly impaired virus replication in the skin without decreasing cell surface gH. Cysteine to alanine substitutions that disabled the predicted intra-molecular disulfide bonds between C540/C575, C647/C703 and C724/C727, all abolished cell fusion likely due to largely reduced cell surface quantities of gH resulting from improper folding of the molecule. These mutant viruses also replicated poorly in cell culture or skin xenografts. Interestingly, replication of the C727A virus in the skin was unexpectedly robust at 10 and 21 dpi compared with other cysteine mutants, and the virus recovered from skin produced plaque morphologies similar to WT VZV. DNA sequencing of the VZV recovered from the C727A inoculated skin implants identified a S694F substitution in gH. To confirm whether the S694F substitution was the cause of the reversion to a WT phenotype, S694F was incorporated into C724A or C727A constructs. Cell fusion was restored in both mutants, although to a lesser extent for C724A. Curiously, the introduction of phenylalanine at position S694 into WT gH increased cell fusion. Importantly, the inactivating properties of C724A and the small plaque phenotype of C727A were rescued by incorporating the S649F into the two gH mutants. These data demonstrated the coordination of S694 and the C724/C727 disulfide bond in influencing cell fusion via stabilizing gH structure [51].

gH functions in the regulation of VZV cell fusion

The cytoplasmic domain of gH (gHcyt) also has a mechanism to regulate cell fusion [105]. Yang et al. reported that the expression of VZV WT gB, gH (gH[WT]) and gL induced fusion that was only marginally greater than vector alone when quantified in the cell fusion assay. However, the introduction of a stop codon in the gHcyt at E834 (gH[TL]/834stop) produced significantly higher levels of fusion and subsequently used as a WT positive control. Similarly, the deletion of the amino acids 834–841 in gHcyt (gH[Δ834–841]) caused cell fusion comparable to WT levels. Interestingly, a mutant with the 8 amino acids 834–841 replaced by a 14 amino acid V5 epitope (gH[V5]) resembled gH[WT] in having very limited capacity to induce cell fusion, in spite of the absence of amino acids 834–841. However, a mutant that contained both the E834stop and V5 epitope (gH[834stopV5]) increased fusion compared with gH[V5], similar to the increase in fusion observed by gH[TL]/834stop relative to gH[WT]. Of note, all of these mutants had protein surface quantities similar to gH[WT], suggesting the change in fusion was not attributed to the expression and trafficking of gH [105]. These data indicated that the full-length gHcyt limits fusion via a mechanism dependent on the physical length of the amino acids 834–841, not on its sequence. The presence of amino acids 834–841 is akin to a latch on a gate that negatively self-regulates the function of gH in gB-triggered fusion. With the latch in place, a co-ordinated set of events are required for fusion to occur; when the latch is released the gate can then be pushed open. The truncation of amino acids 834–841 removes the latch, releasing the constraint on fusion; the gate can be pushed open freely. In the context of VZV infection, gH[TL], gH[Δ834–841] and gH[834stopV5] mutant viruses all exhibited a hyperfusion phenotype with larger syncytia formation compared with WT virus. It is possible that during WT virus infection, the negative self-regulation on gH via gHcyt 834–841 can be modulated by unidentified viral or host factors that allow gB fusion to occur in a regulated fashion, whereas in infection with the gH[TL], gH[Δ834–841], and gH[834stopV5] mutant viruses, the latch from gHcyt 834–841 was removed, so gH is in a constitutively ‘permissive’ state for gB fusion, resulting in a hyperfusion phenotype.

Although the plaque sizes were not significantly affected, the gH[TL] and gH[Δ834–841] mutant viruses both produced significantly reduced virus titers in infected skin xenografts [105]. These findings corroborated the effects of the gHcyt on cell fusion regulation associated with amino acid 834–841 in the presence of other VZV proteins in the skin microenvironment. Interestingly, expression of gB[Y881F], gH[TL] and gL increased cell fusion in comparison with gB[WT], gH[TL] and gL, while gB[Y881F] did not rescue fusion with gH[WT] and gL. This again confirms that amino acids 834–841 of gHcyt limits fusion independent of the ITIM domain in the gBcyt. Critically, the introduction of the two hyperfusogenic mutations, gB[Y881F]/gH[Δ834–841], into the genome rendered VZV noninfectious. Transfection of MeWo cells with the pOka gB[Y881F]/gH[Δ834–841] BAC construct induced cell fusion but VZV capsid assembly in the nucleus was not apparent by EM and infectious VZV was not produced [105]. This particular mutant is a perfect example of the effects of unregulated cell fusion where exaggerated fusion not only impairs VZV spread but restricts the assembly of progeny virions.

Functions of other VZV glycoproteins in cell fusion

Besides the gB/gH–gL core fusion complex, VZV encodes seven other glycoproteins, ORFS/L, gC, gE, gI, gK, gM and gN. VZV gE and gI form a heterodimer important for VZV replication. VZV gE differs from its orthologues in the other alphaherpesviruses in having a unique long N-terminus and being required for replication [106,107]. A VZV ΔgI mutant yields a small plaque phenotype and gI is required for replication in human skin and its deletion results in aberrant replication in sensory ganglia xenografts [106–109]. Although gE binding to insulin-degrading enzyme (IDE) plays a significant role in VZV cell–cell spread [110], neither gE or gI nor the gE/gI heterodimer are required for cell fusion (Figure 4). Although the remaining five glycoproteins, ORFS/L, gC, gK, gM and gN, have not been directly implicated in VZV-induced membrane fusion, orthologues from other alphaherpesviruses have been reported to affect membrane fusion. For instance, the HSV-1 gK/UL20 (homolog of VZV ORF39) heterodimer is known to physically interact with gB [111,112]. When coexpressed with HSV-1 gD, gB and gH–gL, gK blocked cell fusion [113]. In a subsequent study, coexpression of UL20 and gK was shown to inhibit cell fusion by down-regulating the cell surface expression of the other glycoproteins, further demonstrating the potential regulatory role of the gK/UL20 complex on cell fusion [114]. Pseudorabies virus (PRV) gM and HSV-1 gM/UL49A (homolog of VZV gN) have been shown to inhibit HSV-1 gD and gB/gH–gL-mediated cell fusion in a cell-based fusion assay by limiting trafficking of gD and gH–gL to the plasma membrane [115,116]. VZV gM is encoded by ORF50 and a partial deletion of this ORF (nucleotides 408–1058) in VZV pOka to generate rpOkaΔgM yielded infectious virus but resulted in a small plaque phenotype and decreased levels of spread in fibroblast monolayers [117]. Further supporting these findings, the cytoplasmic tail of VZV gM was found to be a determinant of virulence in skin [118]. This suggests that gM, highly conserved among herpesviruses, could act as a modulator in cell fusion by altering trafficking of the core fusion glycoproteins. Whether VZV gM/gN and gK/ORF39 also have similar functions in regulating cell fusion warrants future investigation.

Figure 4. VZV gE and gI are not required for cell fusion. Cell fusion was quantified using the cell fusion assay developed in Vleck et al.[51].

Effector cells were co-transfected with combinations of vectors that express either gB (B), gH (H), gL (L), gE (E) or gI (I) or vectors only (V), and mixed with reporter cells. Flow cytometry was used to quantify individual fusion events identified by the production of a GFP signal and normalized to the BHL positive control. These data have not been published previously.

Interactions of VZV with host cell factors in cell fusion and its regulation

VZV-induced cell fusion is emerging as a complex interplay between host cell and viral factors modulated during infection that function via positive and negative modalities in the regulation of gB/gH–gL-mediated fusion. Myelin-associated glycoprotein (MAG), a cell surface molecule that is preferentially expressed in neural tissues, was first reported to be a receptor that bound VZV gB, promoted gB/gH–gL-mediated cell fusion, and enabled VZV infection when MAG was overexpressed in fibroblasts [34]. However, MAG was also shown to bind gE making the role of MAG in cell fusion in the context of infection uncertain. Integrins also play a role in gB/gH–gL-mediated cell fusion as demonstrated by the reduced levels of fusion in MeWo cells when the αV subunit was knocked down [52]. Integrins have been associated with viral entry for several herpesviruses [119–124]. However, their role in cell fusion is less well understood.

Both viral and host cell transcriptomes were concurrently investigated during WT VZV and hyperfusion mutant gB[Y881F] infection [80]. Altered cellular gene expression profiles were associated with the hyperfusogenic mutant compared with WT virus, including up-regulation of a subset of Ras GTPase genes linked to membrane remodeling (Figure 5). This study suggested a role of host proteins in VZV-induced cell fusion regulation. More recently, a high-throughput SRFA was developed to evaluate the effect of bioactive compounds on VZV gB/gH–gL-mediated cell fusion [125]. This recent study identified the cellular phosphatase, calcineurin, as a critical host factor that modulates VZV-induced cell fusion (Figure 6A). Tacrolimus, one of the fusion enhancing compounds, and the structurally and functionally related compound pimecrolimus, are known to form a complex with FKBP1A that then binds to calcineurin and inhibits its phosphatase activity [126,127]. This specific pimecrolimus-based inhibition of calcineurin phosphatase activity was functional in VZV infected cells (Figure 6B). The knockdown of FKBP1A significantly reduced the enhancement of gB/gH–gL dependent fusion by calcineurin inhibitor pimecrolimus (Figure 6C). Calcineurin inhibition caused more extensive syncytium formation during VZV infection. Consistent with gBcyt and gHcyt hyperfusogenic mutants, exaggerated cell fusion induced by calcineurin phosphatase inhibition also led to suppressed virus propagation (Figure 6D). Importantly, knockdown of FKBP1A released the drug-dependent suppression of VZV spread. Calcineurin acts as a center piece in host cell calcium signaling pathways, so it is not surprising that calcineurin has been reported to be involved in host cell membrane fusion [128–131]. However, calcineurin has not previously been reported to play a central role in regulating virus-induced cell fusion. Critically, pimecrolimus alters the MeWo cell phosphoproteome, which led to the discovery of seven potential calcineurin substrates that might be involved in the regulation of VZV-induced cell fusion, including five putative targets not previously reported (Figure 6E).

Figure 5. Dysregulated gBcyt-mediated cell fusion affects the host transcriptional response to VZV infection.

(A) Hierarchical clustering of the RPKM gene expression values for the 537 genes associated with wild-type-like (upper heatmap), or the 267 genes associated with hyperfusion (lower heatmap) that were differentially expressed in melanoma cells at 36 hpi. Gene expression data were normalized by reallocating gene RPKM values to a mean of zero, scaled to a standard deviation of one, and expressed as a heat map with a range from −4.5 (blue) to +4.5 (red). The dendrograms above the heat maps show the grouping of RNA-seq samples. The colored boxes highlight the groupings of uninfected cells (orange) at 12–36 hpi and the wild-type phenotype (blue) and the hyperfusion phenotype (red) at 24 and 36 hpi. (B) Venn diagrams of frequencies for up-regulated genes (red arrow), and down-regulated genes (blue arrow) associated with VZV phenotype in melanoma cells at 36 hpi. (B) Validation of differentially transcribed Ras GTPase genes associated with the gB[Y881F] hyperfusion phenotype. Transcript abundance of Ras GTPase genes ARL4D, GEM, RASD1, RASL11B, RHOV, RND1 and RRAD associated with the hyperfusion phenotype calculated by qRT-PCR (ΔCq) for the cellular Ras GTPase genes associated with the hyperfusion phenotype in cells infected with pOka and gB[Y881F] at 48 hpi. Adapted from [80]; Copyright © American Society for Microbiology.

Figure 6. Calcineurin phosphatase activity regulates varicella-zoster virus-induced cell fusion.

(A) Bioactive compounds that affect VZV gB/gH–gL-mediated cell fusion identified using HT-SRFA. Scatter plot of cell fusion and cell viability derived from the HT-SRFA. The Y-axis and X-axis indicate fusion efficiency and cell viability values normalized to the mean of positive controls (no drug). The mean of the percentage (% positive control) from two biological replicates are shown: blue circles are all 4846 compounds screened, red and black circles are positive and negative controls and the black crosses are medium only. Gray and orange boxes show ± 3 CV % of positive controls for fusion efficiency and cell viability respectively. (B) Calcineurin phosphatase activity is functional in VZV infected cells. Fluorescence microscopy of MeWo cells transfected with the calcineurin responsive GFP-NFATC1 construct were inoculated with pOka-TK-RFP at 24 h post transfection. At 16 hpi, cells were untreated (medium) or treated with DMSO or pimecrolimus (Pim, 10 μM; 30 min) before ionomycin (Iono, 1 μM; 30 min); TK-RFP virus-infected cells (red), GFP-NFATC1 (green) and nuclei stained with Hoechst 33342 (blue), with a composite image with three channels merged (scale bar = 15 μm). (C) Pimecrolimus disruption of calcineurin-dependent regulation of gB/gH–gL fusion requires FKBP1A. Box and whisker plots of cell fusion (SRFA) in MeWo-DSP1 control (ctrl) or shRNA FKBP1A cells untreated (medium; Med) or treated with DMSO or pimecrolimus (Pim, 10 μM). Cell fusion was normalized to the untreated (% medium) for each MeWo-DSP1 cell line. Boxes represent 25–75 percentile, whiskers extend to 10–90 percentile, the median is the horizontal band and the mean is indicated by the ‘+’ and the statistical differences were analyzed by two-way ANOVA (ns, not significant; ****, P < 0.0001). (D) Inhibition of calcineurin phosphatase activity suppresses VZV spread. Immunohistochemistry staining of VZV plaques (upper panel) and plaque sizes (lower panel) of VZV pOKa infected MeWo cells treated with medium, DMSO or pimecrolimus (10 μM) at 4 dpi. IHC images show VZV plaques with the mean plaque size per condition (scale bar = 0.3 mm). Box and whisker plot of plaque sizes for MeWo cells, MeWo-DSP1 control cells or shRNA FKBP1A MeWo-DSP1 cells at 4 dpi with VZV pOKa, untreated (medium; Med) or treated with DMSO or pimecrolimus (Pim, 10 μM). Plaques were normalized to the untreated (% medium). Boxes represent 25–75 percentile, whiskers extend to 10–90 percentile, the median is the horizontal band and the mean is indicated by the ‘+’ from two independent experiments (n = 60). Statistical differences were evaluated by two-way ANOVA (ns, not significant; ****, P < 0.0001). (E) Inhibition of calcineurin phosphatase activity alters the phosphoproteome of MeWo cells. Heat map of uniquely phosphorylated proteins detected in MeWo cells treated with pimecrolimus (10 μM) for 2 h. Total cellular proteins were extracted in the presence of phosphatase inhibitor and subject to Zr-IMAC phosphopeptide enrichment, followed by orbitrap mass spectrometry. Relative frequency of spectra for uniquely phosphorylated proteins detected in pimecrolimus treated samples but not in DMSO. Nuclear factor of activated T cells cytoplasmic 1 (NFATC1); ETS transcription factor (ELK1); phosphatase and actin regulator 2 (PHACTR2); interleukin enhancer-binding factor 3 (ILF3); nuclear receptor co-activator 1 (NCOA1); transmembrane channel-like protein 8 (TMC8); methyl-CpG binding domain protein 1 (MBD1). Previously known calcineurin substrates are highlighted with asterisks (*). Adapted from [125]; reproduced under the Creative Commons Attribution (CC BY) license.

The goldilocks principle of VZV cell fusion

Syncytia formation is a typical characteristic of VZV replication in vitro which recapitulates polykaryocyte formation during pathogenesis in vivo. VZV-induced cell fusion follows the Goldilocks principle of not too much, not too little but ‘just right’; it needs to be tightly regulated via coordination through specific motifs on viral glycoproteins, gB or gH, that execute membrane fusion and also through host cell factors that are implicated in the breakdown of the intrinsic barriers against cell fusion. Limited fusion leads to suppressed virus replication in cell culture and in skin tissue, likely due to less efficient virus spread. However, enhanced cell fusion does not necessarily lead to enhanced virus replication and spread, instead, frequently exaggerated cell fusion limited or even abolished virus replication in vitro and reduced pathogenesis in the skin in vivo. The suppressed VZV spread caused by enhanced cell fusion could be due to at least three factors. Firstly, VZV cell-to-cell spread does not solely rely on cell fusion. This is supported by the observations that VZV spread preceded the earliest cell fusion events detected during replication [132]. Secondly, exaggerated cell fusion likely hampers viral assembly caused by the severe disruption of the intracellular environment as evidenced by the dual gB/gH cytoplasmic domain mutant (gB[Y881F]/gH[Δ834–841]) that dramatically enhanced cell fusion but failed to produce any detectable viral capsid assembly within infected cells [105]. Separately, virus-induced cell fusion has been shown to function as a ‘danger’ signal sensed by neighboring cells to elicit innate antiviral immune responses (i.e. type-I interferon or inflammatory response) [133,134]. VZV infection in the epidermis was accompanied by down-regulation of type I interferon (IFN-α) in infected cells but IFN-α expression was up-regulated in adjacent cells [79]. Thus, it is possible that cell fusion elicits a potent innate immunity in the neighboring cells that protects the uninfected cells from virus infection, overall limiting virus spread. However, the adaptive VZV-specific T cell responses, including the production of IFN-γ, are required to prevent life-threatening VZV infections [135–137]. Therefore, the outcome of VZV infection is governed by the complex environment of the host. Well-modulated cell fusion sustains the optimal levels of replication for viral propagation, a result from the approximately 70 million years of co-evolution between VZV and its human host [138,139]. A better understanding of the molecular mechanisms underlying the regulation of VZV-induced cell fusion and its relationship with host factors that inhibit or support fusion and its regulation has the potential to provide novel antiviral strategies by targeting either viral and host factors that disrupt the homeostasis of VZV-induced cell fusion.

Perspective.

• Importance of the field: VZV infection produces skin lesions that contain multinucleated cells, polykaryocytes, and cause neurons and satellite cells within sensory ganglia to fuse, likely contributing to the debilitating condition of PHN. A better understanding of the molecular mechanisms that drive cell fusion during varicella and zoster have the potential to lead to treatments that target the pathology which results from VZV-induced cell fusion.

• Current thinking: VZV replication and spread is governed by the core fusion machinery of gB/gH–gL. Inactivation of this complex prevents replication whereas hyperfusion is detrimental to VZV spread; fusion is regulated by both viral and cellular factors.

• Future directions: Targeting the gB/gH–gL fusion complex directly or cellular processes that regulate fusion might have significant advantages over current therapies that rely solely on targeting VZV genome replication and pain management. In addition to the viral fusion complex, further insights into the cellular factors that regulate cell fusion will support this goal.

Abbreviations

CTD

carboxyl terminal domain

DRG

dorsal root ganglia

GCA

giant cell arteritis

IDE

insulin-degrading enzyme

ITIMs

immunoreceptor tyrosine-based inhibition motifs

MAG

myelin-associated glycoprotein

ORFs

open reading frames

PHN

post herpetic neuralgia

PRV

pseudorabies virus

SCID

severe combined immunodeficiency

SRFA

stable reporter fusion assay

VZV

varicella-zoster virus