Insights into the pathogenesis of varicella viruses

Abstract

Purpose of review:

Varicella zoster virus (VZV) is a highly contagious, neurotropic alpha herpes virus that causes varicella (chickenpox). VZV establishes lifelong latency in the sensory ganglia from which it can reactivate to induce herpes zoster (HZ), a painful disease that primarily affects older individuals and those who are immune-suppressed. Given that VZV infection is highly specific to humans, developing a reliable in vivo model that recapitulates the hallmarks of VZV infection has been challenging. Simian Varicella Virus (SVV) infection in nonhuman primates reproduces the cardinal features of VZV infections in humans and allows the study of varicella virus pathogenesis in the natural host. In this review, we summarize our current knowledge about genomic and virion structure of varicelloviruses as well as viral pathogenesis and antiviral immune responses during acute infection, latency and reactivation. We also examine the immune evasion mechanisms developed by varicelloviruses to escape the host immune responses and the current vaccines available for protecting individuals against chickenpox and herpes zoster.

Recent findings:

Data from recent studies suggest that infected T cells are important for viral dissemination to the cutaneous sites of infection as well as site of latency and that a viral latency-associated transcript might play a role in the transition from lytic infection to latency and then reactivation.

Summary:

Recent studies have provided exciting insights into mechanisms of varicelloviruses pathogenesis such as the critical role of T cells in VZV/SVV dissemination from the respiratory mucosa to the skin and the sensory ganglia; the ability of VZV/SVV to interfere with host defense; and the identification of VLT transcripts in latently infected ganglia. However, our understanding of these phenomena remains poorly understood. Therefore, it is critical that we continue to investigate host-pathogen interactions during varicelloviruses infection. These studies will lead to a deeper understanding of VZV biology as well as novel aspects of cell biology.

Introduction

The Varicelloviruses genus belongs to the Alphaherpesviridae subfamily (1). Varicella Zoster Virus (VZV) is the prototypic human varicellovirus and Simian Varicella Virus (SVV) is considered its non-human primate homologue (2). Both viruses are neurotropic alphaherpesviruses. VZV is primarily contracted via inhalation of virus laden saliva droplets or by direct contact with varicella or zoster vesicular lesions (3–5). Acute VZV infection results in varicella (commonly known as chickenpox), this is followed by the establishment of latency in sensory ganglia from which it can reactivate resulting in herpes zoster (HZ, also known as shingles), a potentially debilitating disease that affects ~1 million individuals per year in the United States (6). SVV is a natural pathogen of old-world monkeys that also causes the characteristic varicella exanthem. Disease manifestations can vary depending on the macaques species from benign disease (Rhesus macaques) to severe infection (African Green and Cynomolgus) (7, 8). SVV also establishes latency in sensory ganglia and can reactivate in response to immune suppression or stress (9, 10). It is unclear how SVV is transmitted but it is detected in saliva of experimentally inoculated rhesus macaques, suggesting a similar mechanism as VZV (11, 12).

1. Genome organization and virion structure of varicelloviruses in primates

SVV and VZV viral particles are composed of 4 main elements: a linear double stranded DNA molecule contained within an icosahedral nucleocapsid which is surrounded by a tegument and an outer lipid envelope (6, 13). VZV and SVV virions have a diameter of 80–120nm and 170–200 nm and a genome length of 124,138 bp and 124,884 bp, respectively (6, 13). In addition, VZV and SVV genomes are collinear in terms of gene organization sharing approximately 70–75% DNA homology (14). Both genomes encode 71 unique open reading frame (ORFs) (6, 15). Three of these ORFs (ORFs 69, 70 and 71) are duplicated within the repeat regions (6, 15).The major variation between the two genomes is located in the left terminus where SVV does not encode a homolog for ORF2 while VZV lacks a SVV ORFA homolog (16).

2. Viral pathogenesis during acute varicella virus infection

The incubation period of VZV is approximately two weeks and disease is characterized by fever, fatigue, loss of appetite and a characteristic vesicular rash (17). VZV is thought to initially replicate in the upper respiratory tract and tonsillar lymph nodes before the detection of cell-associated viremia (6). Subsequently, viral particles reach the skin leading to the development of the characteristic vesicular rash. Although varicella is a benign disease that resolves within 7–10 days in most children, immunocompromised individuals may develop severe symptoms and complications such as pneumonia, secondary bacterial infections and stroke (18–20). As discussed above, experimental inoculation of different macaque species with SVV can result in a spectrum of disease outcomes. African green monkeys and cynomolgus macaques experience persistent viremia and fatal disease (21–23). In contrast, experimental intrabronchial infection of rhesus macaques with SVV faithfully recapitulates the cardinal hallmarks of VZV infection in humans including the development of varicella rash, generation of cellular and humoral immune responses, establishment of latency in the sensory and enteric ganglia and viral reactivation following immune suppression (24–26, 27*).

Despite extensive studies, the mechanisms by which VZV/SVV traffic to the skin and sensory ganglia following respiratory infection remain poorly understood. Recent data from experiments carried out using either VZV infection in a severe-combined immunodeficient mouse model implanted with human fetal tissues (SCID-hu) or SVV infection in the rhesus macaque model suggest that T cells play a critical role in varicella viruses dissemination from the respiratory mucosa to the skin and sensory ganglia (28–32, 33**, 34**, 35–39). Additional studies have shown that human primary CD4 and CD8 T cells are permissive to VZV infection associated with the formation of infectious virions (28, 40–42). Moreover, human tonsillar CD4 T cells, but not fibroblasts, infected in vitro with the VZV Oka vaccine strain were able to transport the virus to fetal human skin (36) and dorsal root ganglia xenografts (43) following intravenous injection in the SCID-hu model. Furthermore, VZV preferentially infects memory T cells that expressed activation markers and skin homing proteins (35). Collectively, these observations strongly suggest that VZV-infected T cells could facilitate viral dissemination to the cutaneous sites of infection as well as sites of latency.

In accordance with those results, in vitro studies using time of flight mass cytometry (CyTOF) showed increased expression of activation markers (CD69, CD279, CD28, CD11a and CD49d) as well as signaling molecules (phosphorylated ZAP70 and SLP76) in human tonsillar CD4 T cells infected with the attenuated Oka VZV vaccine strain (38). Additionally, a microarray study on human T cells infected in vitro with attenuated VZV vaccine strain reported an increased expression of genes that play a role in antiviral responses, cell cycle progression, and regulation of transcription suggesting that VZV infection induces gene expression changes within T cells (44). Another study reported that VZV could promote viral dissemination through activation of STAT3 to promote survival of infected T cells (45).

Similarly, in vivo studies in the rhesus macaque model showed that T cells in the broncho-alveolar lavage (BAL) are susceptible to and can support SVV viral replication (34). Moreover, memory CD4 and CD8 T cells infiltrate the ganglia as early as 3 days post infection (DPI), before the development of cell-mediated immunity, which correlate with detection of viral loads and viral transcripts, suggesting that SVV uses T cells as a trojan horse to gain access to the sensory ganglia (34**). Similar results were obtained in African green monkeys (46**). More recently, our laboratory reported that SVV infection causes large changes in the expression of genes that play a role in cell cycle, cellular metabolism, and immune processes in T cells isolated from BAL samples collected from rhesus macaques at different days post infection (33**). Taken together, as seen in VZV pathogenesis, these findings suggest that T cells are susceptible and permissive SVV targets that play a critical role in viral dissemination to the ganglia. However, the source of these gene expression changes (infected T cells vs. bystander T cells) and their functional implications for disease pathogenesis have yet to be determined.

Several glycoproteins and tegument proteins were shown to be important viral determinants for VZV T cell tropism. First, glycoprotein I (gI) and the sequence comprising amino-acid residues 51 to 187 of gE are necessary for T cell infection (47, 48). Additionally, viral kinases seem to play a major role in T cell tropism in the SCID-hu model; specifically, loss of ORF47 prevented VZV replication in T cells xenografts while ORF66, although dispensable, was shown to be important for virion assembly in T cells (40). Taken together, these findings suggest that T cells play a critical role in VZV dissemination; however, these studies were conducted using the attenuated Oka vaccine strain, which may not accurately reflect the outcome of infection with wild type strains. Additionally, the strict human tissue tropism of VZV could have altered VZV behavior in the SCID-hu model (44, 49, 50). Additional studies should rigorously test the role of these various ORFs using the SVV model where natural host-virus interactions can be interrogated.

In addition to their ability to replicate in T cells, varicelloviruses exhibit a strong epithelial tropism. Consistent with the varicella incubation period in humans, VZV-infected human tonsillar CD4 T cells were shown to deliver the virus to skin xenografts in the SCID-hu model leading to the appearance of the characteristic varicella skin lesions within 20 days after intravenous injection (36). In skin xenografts, the first infected cells are keratinocytes near hair follicles; followed by the formation of multinucleated polykaryocytes in adjacent cells which constitute the hallmarks of varicella replication in the skin (36). Several viral proteins were shown to be involved in skin tropism during VZV infection. For instance, gB and the heterodimer formed by gH and gL were shown to be required for cell-cell fusion and polykaryocyte formation in the skin (51, 52). In addition, studies conducted in the SCID-hu model have reported that ORF47, a tegument protein, was necessary for efficient replication in skin xenografts (53, 54). More recently, mutations in the trafficking motifs of gM significantly decreased viral replication in skin xenografts in vivo while exhibiting no effect on VZV neurotropism (55*). Additional viral factors were shown to play a role to some extent in skin tropism, including gE, gI and ORFs 10, 11 and 12 (6).

3. Antiviral immune responses against varicella viruses during primary infection

During varicella and HZ, skin lesions are infiltrated by neutrophils as well as dermal perivascular and perineural macrophages along with some natural killer (NK) cells (56, 57). Importantly, production of type I interferon (IFN) by epithelial cells surrounding infectious foci and infiltrating plasmacytoid dendritic cells (pDCs) in the skin play an important role in restricting VZV replication in the skin (36, 58). Indeed, inhibition of IFNα and β signaling results in significantly larger cutaneous lesions in skin xenografts (36). In addition to type I IFN, promyelocytic leukemia (PML) bodies were also shown to sequester nucleocapsids through interaction with the ORF23 capsid protein to prevent nuclear egress of newly assembled virions thereby restricting viral replication (59). Similarly, the antiviral innate response in the lungs of rhesus macaques following intrabronchial challenge with SVV is associated with increased production of pro-inflammatory cytokines, chemokines and IFNα into the alveolar space that correlates with infiltration of pDCs and a significant reduction in viral loads in the BAL (26, 60**).

While type I interferons are essential in controlling VZV/SVV replication, T cell-mediated immunity is required to resolve primary infection (11, 61). VZV-specific T cells reach detectable levels in the blood around 3 days after varicella onset (61, 62) of which some CD4 T cells were shown to express the skin homing receptor cutaneous lymphocyte associated antigen (CLA) (62). The T cell immune response in children acutely infected with VZV is predominantly Th1 (63) and reaches its peak at 2 weeks after both primary infection and reactivation before a gradual decrease of the specific T cell levels (61, 64, 65). Similarly, results obtained from peripheral blood mononuclear cells (PBMC) of rhesus macaques infected with SVV indicate that all T cell proliferation peaks 14 dpi in the blood (24, 26).

The rhesus macaque model also facilitates the analysis of the immune response in the lungs. These data show that T cell proliferation occurs earlier in the lung peaking at 7 dpi in BAL samples. Although SVV infection resulted in a larger increase in the frequency of CD8 T cells, it was shown that SVV-specific CD4 T cells were more abundant in broncho-alveolar lavage (BAL) samples than CD8 T cells (26). Moreover, several studies have highlighted the critical role played by CD4 T cells in resolving both SVV and VZV primary infection (11, 49, 66, 67) and HZ (68). A recent study showing that production of type II IFN is more potent than type I IFN in restricting VZV replication in primary human fibroblasts corroborates this hypothesis (69*) which could explain why individuals with T cell deficiencies are more likely to develop complications during primary infection (70–74).

During acute infection, the anti-SVV T cell response is broad with most SVV ORFs targeted by both BAL and blood T cells. As described for the anti-VZV response in humans (75–77), T cells specific to SVV ORFs 4, 10, 29, 62, 63 and 67 were detected in infected rhesus macaques. CD8 T cells preferentially recognized immediate-early and early viral proteins, while CD4 T cells were directed against late proteins (78). During latency, the T cell response decreases in both magnitude and breadth, with only 26 ORFs recognized by memory T cells (78). Additionally, a strong systemic pro-inflammatory response associated with an increase in the T cell counts in the peripheral blood is detected following SVV reactivation in rhesus macaques (79).

Regarding the humoral response to VZV, antibodies directed against a third of the VZV proteome can be detected in sera one week after the development of varicella rash and reach a peak after one month (61, 80). Similarly, intrabronchial infection of rhesus macaques with SVV elicited a humoral response with SVV-specific IgG antibodies detected 7 dpi and peak 18–20 dpi (24). Despite contributing to the control of SVV/VZV infection, the role played by antibody immune responses remains less crucial as compared to T-cell mediated immunity as B-cell immunodeficiencies are not associated with a more severe varicella disease (11, 81).

During primary SVV infection, the development of a strong innate pro-inflammatory response in the sensory ganglia is associated with an infiltration of macrophages (34, 46). In addition to infiltrating immune cells, neurons and neuron-interacting satellite glial cells were also shown to contribute to the local immune response. Indeed, satellite glial cells were reported to upregulate MHC class I and II as well as TFNα and IL6 during primary SVV infection (46**, 82). In the SCID-hu model, VZV infection led to viral replication in neurons and satellite cells from human DRG xenografts followed by viral persistence associated with highly-restricted viral transcription 4–8 weeks post-infection (37, 83). Since adaptive immunity is absent in SCID-hu mice, this observation suggests that innate immunity plays an important role in controlling VZV acute replication in the sensory ganglia (43). T cells also infiltrate sensory ganglia during primary SVV infection (34**, 39). These T cells may be recruited by the upregulation of MHC-II on satellite glial cells or increased expression of the gene encoding CXCL10 chemokine during SVV primary infection (34**, 46**). CD4 T cell responses are important for the establishment of latency as evidenced by impairment in this process in young rhesus macaques depleted of CD4 T cells during acute SVV infection (but not CD8 T cells or B cells) and in aged animals that exhibit a dampened cellular immunity (84). However, the mechanisms by which T cells contribute to the establishment of latency remain poorly understood.

4. Varicella viruses pathogenesis during latency and reactivation and immune responses in the ganglia during reactivation

After primary infection, varicella viruses traffic to the sensory ganglia to establish latency, notably, the dorsal root ganglia (DRG) trigeminal ganglia (TG) and enteric ganglia from where they reactivate to induce HZ (24, 27*, 85**, 86, 87). Two non-mutually exclusive mechanisms are postulated the viral route of entry in the sensory ganglia: 1) retrograde axonal transport from the infected skin and 2) hematogenous dissemination through infected T cells (9, 88). During latency, the viral genome persists in the cellular nucleus in a circular form, termed episome and VZV DNA was shown to be maintained at 5–7 genomes copies in ~ 2–5% of the sensory neurons (89, 90). The gI glycoprotein as well as the formation of the gE-gI heterodimer are essential for neurovirulence as gI deletion led to prolonged acute replication instead of latency and impairing the dimer formation severely disrupted the viral spread between neurons in the sensory ganglia (91).

Latent infection is associated with a restricted viral gene expression program associated with rare protein production. Studies that analyzed the viral expression profile in the sensory ganglia extracted from both humans and rhesus macaques latently infected with VZV and SVV, respectively, have reported sporadic detection of viral transcripts originating from ORF21, 29, 62, 63 and 66 (25, 92–97). In addition, both VZV and SVV ORF63 proteins were identified in the cytoplasm of sensory neurons in latently infected ganglia of humans and rhesus macaques, respectively (24, 93, 98).

However, the most abundant detected transcript in latently infected macaque ganglia is an anti-sense transcript associated with SVV ORF61 (24, 25, 99). Similarly, a recent study, which analyzed human sensory ganglia processed less than 9 hours post mortem, reported a newly identified viral latency-associated transcript called VLT that is antisense to ORF61 (100**). VZV-encoded VLT is a polyadenylated RNA that includes at least 5 exons and is expressed in multiple alternatively spliced isoforms during lytic replication while only one isoform was detected in human trigeminal ganglia during latency (100**). These observations are in line with results obtained for all other alphaherpesviruses that reported the expression of latency-associated transcripts (LAT) antisense to the region comprising homologues of ICP0 suggesting that these transcripts underwent a similar co-evolution (101). However, compared to other alphaherpesviruses LATs, it is important to note that varicella viruses VLTs are expressed in few copies in neurons, and don’t seem to embed micro-RNAs (miRNAs) sequences (100**, 102–104). The current hypothesis to explain the function of the varicella viruses VLTs postulates that these antisense transcripts play a role in shutting off the sense ORF61 expression in order to maintain latency and prevent spontaneous viral reactivation. This hypothesis is further supported by the recent study showing that VZV VLT can selectively repress ORF61 expression when co-expressed in vitro (100**). This mechanism could explain why deletion of the entire SVV ORF61 did not impair the establishment of latency in rhesus macaques (105)

T cell immunity is critical for the maintenance of latency since immunocompromised or older individuals are more prone to develop HZ (86). During SVV and VZV reactivation, T cells infiltrate ganglia in humans and non-human primates, respectively (82, 106, 107). Interestingly, CD4 T cells were the most abundant subset of immune cells infiltrating human sensory ganglia during active VZV reactivation (82). These cells may be recruited via the increased levels of CXCL10 (107) and/or MHC class I and II (46**, 82). NK cells and monocytes were shown to infiltrate sensory ganglia of both cynomolgus macaques experiencing SVV reactivation (107) and humans who suffered from HZ at the time of death (82, 108). Additionally, pro-inflammatory cytokines TFNα and IL6 are also detected during VZV reactivation (46**, 82).

5. Viral Immune evasion mechanisms developed by varicella viruses:

Varicella viruses have developed various strategies to counteract innate and adaptive immune responses. First, both VZV and SVV ORF63 were shown to interfere with type I IFN signaling by inhibiting the JAK-STAT signaling pathway (109–111), degradation of IRF9 (110), or inhibition of eukaryotic initiation factor 2 phosphorylation (114). Second, VZV ORF62 and ORF47 can block IRF3 phosphorylation (112, 113). Additionally, TLR9 signaling is impaired in pDCs infected with VZV (58).

VZV ORF61 was reported to disrupt PML nuclear bodies, an intrinsic antiviral defense, to facilitate viral replication in skin xenografts in the SCID-hu model (115). Additionally, VZV ORF61 interferes with the NFκB pathway in human skin xenografts in the SCID-hu model (116). One proposed mechanisms is by preventing Iκbα degradation (117, 118). Similarly, overexpression of SVV ORF61 in vitro was able to inhibit Ikbα degradation leading to inhibition of NFκB activation (118). However, the deletion of ORF61 from the SVV genome did not restore NFκB signaling in vitro, suggesting that additional viral factors are involved in NFκB inhibition (118). Interestingly, another study showed that overexpression of VZV ORF61 blocks the IRF3-dependent IFNβ pathway while having a minor effect on NFκB signaling (119). Differences between the studies investigating the function of VZV ORF61 could be explained by the different cell lines used (telomerized human fibroblasts in Zhu et al. 2011, compared to human lung fibroblasts in Whitmer et al. 2015); or the types of experimental readouts (NFκB reporter assay following co-transfection with VZV ORF61 expression plasmid and Sendai Virus-mediated activation of IFNβ pathway (119) versus measuring the ubiquitination of Iκbα using cells stably expressing inducible VZV ORF61 and stimulated with TNFα (118)). Deletion of SVV ORF61 led to an increased proinflammatory response in the lungs of SVV-infected macaques, associated with an upregulation in the expression of innate immune genes including type I IFN and increased infiltration of pDCs (105). However, whether this phenotype is due to a direct effect or indirect effect of ORF61, an immediate early transactivator that regulates the expression of other ORFs that could act as immune modulators, remains to be determined.

In addition to their ability to evade the innate immune surveillance, varicella viruses were also reported to be able to escape the host adaptive immune response. First, VZV was shown to downregulate MHC-I in infected cells that may contribute to the viral immune evasion of the CD8 T cell response (120, 121). Interestingly, VZV ORF66 was shown to mediate, at least to some extent, this function through impairing MHC-I transport to the cell surface (121, 122). In addition, both VZV and SVV were shown to inhibit the JAK-STAT pathway leading to the downstream inhibition of IFNγ expression that could contribute to the evasion of the cell-mediated immune response (56, 109, 123*). Importantly, VZV was shown to be able to interfere with the function of NK cells including through the modulation of the cell surface expression of NKG2D ligand (124**, 125).

6. Current vaccines against chickenpox and herpes zoster.

A live attenuated vaccine against chickenpox is licensed and given routinely in several countries including, Japan (1988), the United States (Varivax®, 1995), Canada (1999), Australia (2005) and Germany (2004) (126). The VZV vaccine was derived from the Oka strain which was initially isolated in Japan from vesicles of a 3-year-old boy with chickenpox (127). The live attenuated vaccine was subsequently obtained through passaging the virus multiple times in different cell types, including human embryonic lung fibroblasts and guinea pig fibroblasts (127). Although the precise mechanisms of attenuation remain unknown, a comparison of the attenuated Oka VZV (vOka) and the WT VZV strain revealed several nucleotide substitutions, deletions and insertions located mainly in the major viral transactivator ORF62 as well as in the internal repeat regions (50). The varicella vaccine is safe and used routinely worldwide to prevent chickenpox disease. However, the attenuation does not impair vOka from establishing latency in the sensory ganglia and reactivating (128–131). The VZV vOka vaccine was initially administered as a single dose in the United States, however, due to numerous chickenpox outbreaks in the late 1990s and early 2000s, a routine second dose of vaccine was added to the immunization schedule of children in 2006 to prevent varicella disease. As reported in a case-control study, this second dose of vaccine increased protection from 85% to 98% (132).

Since the VZV vOka vaccine strain does not provide sterile immunity but rather protects against clinical disease, vaccinated individuals may acquire WT virus that establishes life-long latency. Therefore, there was a need to develop a more potent vaccine to prevent HZ in the elderly. A first vaccine (Zostavax®) was licensed in 2006 to prevent HZ in the elderly (133). This vaccine contains 19,400 pfu/dose of the same live attenuated virus as Varivax® vaccine (>14x more potent) (134). Routine vaccination in the elderly with Zostavax® vaccine led to a decrease in the overall incidence of HZ by 51% while the associated complications of HZ, including postherpetic neuralgia, were reduced by 66.5% (135). However, vaccine efficacy drastically decreases with age reaching only 18% of efficacy in individuals over the age of 80 years old (136) potentially due to a failure to induce a robust VZV-specific T cell-mediated immunity. Indeed, the magnitude of the T cell response correlated negatively with the age of the patient (137). Moreover, the cell-mediated immune response decreases significantly reaching pre-vaccination levels within 3 years (135). These observations led to the FDA, reducing the age requirement for the Zostavax® vaccine to 50 years old.

More recently, a novel and more effective vaccine, called Shingrix® has been developed to prevent HZ. This recombinant vaccine contains an adjuvanted form of the VZV gE protein encoded by the ORF68 gene which was shown to be the most abundant and highly immunogenic viral glycoprotein (138**). Shingrix® vaccine is highly effective and demonstrated 97.2% of efficacy after the inoculation of 2-doses vaccination administered between a 2-month interval in patients over 50 years old. Moreover, it is also safe for patients where live attenuated vaccines are contraindicated. Immunization with Shingrix® leads to the development of both strong humoral and cell-mediated immune responses against gE. Importantly, the efficacy of this recombinant vaccine lasts at least 4 years post vaccination, including in patients over 70 years old without showing a significant loss in efficacy correlated with age. Although the magnitude of the VZV-specific adaptive immune response induced by the vaccine decreases over time, it was shown to remain significantly higher than the pre-vaccination levels for up to 9 years in healthy patients that received 2 doses of vaccine.

Conclusion

The mechanisms by which VZV disseminates from the respiratory system to the skin and the sensory ganglia remain poorly understood. Data from several studies suggest a critical role for T cells in this process, but the mechanisms by which varicella viruses usurp T cells to disseminate to the ganglia and skin remain to be elucidated. Similarly, the transition from lytic infection to latency and then reactivation is poorly understood, including the role played by the VLT and infiltrating T cells. In addition, the viral and host factors that promote the establishment of latency in sensory neurons are not fully understood. Although several in vitro studies have reported various immune evasion strategies by VZV/SVV, the biological significance of these pathways has yet to be determined in vivo. Although the newly developed Shingrix® vaccine confers better protection against HZ, there is so far no vaccine to prevent initial VZV infection without establishing latency. Thus, it is critical that we improve our current understanding of VZV pathogenesis.