Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Varicella-zoster virus (VZV) is a neurotropic human herpesvirus (subfamily Alphaherpesviridae) and the etiological agent of two human diseases, varicella and herpes zoster (reviewed in Cohen et al.(2007) and Zerboni and Arvin(2008)). Varicella, commonly called “chickenpox”, results from primary VZV infection in susceptible individuals. In healthy children, varicella is a self-limiting infection that manifests as successive crops of pruritic lesions accompanied by fever. As a consequence of primary infection, VZV gains access to cells within the sensory cranial nerve and dorsal root ganglia (DRG) in which it appears to persist long term in a non-replicating “latent” state. Herpes zoster is the medical consequence of “reactivation” of latent VZV genomes within sensory nerve ganglion cells in which newly formed progeny virions transit in the anterograde direction to reinfect peripheral skin sites. Unlike the generalized rash of varicella, herpes zoster (commonly called “shingles”) involves a localized region comprising one or more adjacent dermatomes, reflecting the sensory ganglion from which the virus has reactivated and in some cases, neighboring ganglia (Hope-Simpson 1965; Silverstein and Straus 2000). Whereas varicella resolves within 7–10 days, zoster lesions may require several weeks to heal and neuropathic pain may persist for months (Gilden et al. 2006; Silverstein and Straus 2000). In that herpes zoster is contingent upon reactivation of latent varicella within neuronal ganglia, it is a direct consequence of VZV neurotropism.

VZV is highly contagious and is maintained in the human population by shedding from vesicular lesions during varicella and herpes zoster episodes (Hope-Simpson 1965). Prior to widespread vaccination, varicella was a common childhood disease. Resolution of primary infection and herpes zoster corresponds to induction of VZV-specific T cell mediated immune responses (Arvin 2005). Immunoscenescence and immunodeficiency are risk factors for reactivation of latent VZV, and so herpes zoster is most common among the elderly and immunocompromised (Arvin 2005; Oxman et al. 2005). Live attenuated varicella vaccines have been licensed for use in preventing primary VZV infection and restoring protective cellular immune responses in the elderly (Arvin 2001; Oxman et al. 2005; Takahashi et al. 1974). Derived from serial passage of the “Oka” clinical isolate, Oka varicella vaccines retain neurotropism and can reactivate; infection with naturally circulating virus can also occur after vaccination. Therefore, vaccination does not consistently prevent VZV from establishing latency or from reactivating from sensory nerve ganglia (Arvin 2001; Takahashi et al. 1974; Zerboni et al. 2005b).

In this chapter, we review our investigation of VZV neurotropism through experimental infection of human DRG xenografts in severe compromised immunodeficient (SCID) mice (Oliver et al. 2008; Reichelt et al. 2008; Zerboni et al. 2005b, 2007). These experiments provide insight into VZV interactions with neural cells in sensory ganglia and may have potential relevance in the design of neuroattenuated varicella vaccines.

2 The SCID Mouse–Human Xenograft Model for VZV Pathogenesis

The clinical consequences of primary varicella infection and herpes zoster have been investigated for several decades; however, knowledge concerning VZV pathogenesis and virulence within the natural host is limited because infections are rarely fatal and, most importantly, no small animal exists that recapitulates disease in the human (Cohen et al. 2007). Whereas other human alphaherpesviruses readily infect rodents, rabbits, and guinea pigs, progress in the development of a small animal model for the in vivo investigation of VZV pathogenesis has been hampered by a strict restriction for human tissues (Cohen et al. 2007). Experimental infection of mice, rats, rabbits, or guinea pigs with a high potency VZV inoculant fails to produce robust infection (Cohen et al. 2007). Xenotransplanation of immunodeficient SCID mice with human tissues was first reported in 1988 (McCune et al. 1988; Mosier et al. 1988). To address the growing need for an experimental system to carry out “in vivo” infection of tissues with permissivity for VZV, we established a small animal model in which human tissue xenografts with relevance to VZV pathogenesis are infected within the SCID mouse host (Moffat et al. 1995, 1998; Zerboni et al. 2005b). SCID mice are not susceptible to VZV infection due to the host cell restriction, and so the consequences of VZV xenograft infection can be examined over a long interval, which is critical for investigation of latency and reactivation. For more than a decade, studies using human skin and T cell (thymus/liver) xenografts in SCID mice have yielded a wealth of information regarding VZV skin and T cell pathogenesis (reviewed in Arvin(2006)). These SCID xenograft models have also proven to be valuable experimental tools for the assessment of VZV gene function in vivo through experimental infection with recombinant viruses (reviewed in Zerboni and Arvin(2008)).

3 SCID Mouse–Human DRG Xenografts

Investigations of VZV neurotropism and virus–host interaction involving the various cell types that reside within sensory nerve ganglia have relied primarily upon tissues acquired from deceased individuals (Cohen et al. 2007). Tissues acquired postmortem from individuals who died during primary varicella or reactivation are exceedingly rare, and so information concerning acute VZV replication within sensory nerve ganglia has been limited. Examinations of latently infected cadaver ganglia are fraught with discrepant reports, which may be a consequence of physiological changes that occur during the interval between death and tissue acquisition (Cohen et al. 2007). Alternative in vitro model systems, such as explanted human DRG and primary neuronal cultures, do not survive long term and may also undergo physiological alterations in response to culture conditions. Optimally, a model for VZV neurotropism and neuropathogenesis would contain all cell populations that are potentially relevant to VZV neuropathobiology within their normal tissue microenvironment under physiological (in vivo) conditions. Therefore, we developed a SCID mouse–human xenograft model by establishing intact human DRG xenografts in SCID mice (Zerboni et al. 2005b). This model provides the opportunity to explore VZV infection and tropism for human neurons while preserving the interactions between neurons and non-neuronal cells within the DRG microenvironment.

3.1 Xenotransplantation of Human DRG in SCID Mice

SCID mouse–human DRG xenografts are constructed by implantation of human fetal DRG under the renal capsule of 5–8 week old male scid/scid homozygous CB.17 mice (Zerboni et al. 2005b). DRG are dissected from fetal spinal tissues at 18–24 gestational weeks and are ∼1–2 mm3 at the time of xenotransplantation. Visual inspection at 12 weeks post xenotransplantation reveals a small (2–3 mm3) well-vascularized graft attached to the mouse kidney (Fig. 1a). Histopathic examination reveals a neural tissue with organotypic features of human ganglia including clusters of small and large diameter neural cell bodies surrounded by satellite cells, fibroblasts, and other cell types within a dense supportive matrix comprising collagen and nerve fibers (Fig. 1b). Neural specific cell markers, such as synaptophysin and neural cell adhesion molecule (NCAM), are expressed and retain their expected localization. Human endothelial cell marker, PECAM-1, is also expressed on blood vessels supporting the graft. DRG engraftment is successful in >95% of mice and xenografts survive up to 8 months post xenotransplantation.

Fig. 1
figure 1

VZV infection of DRG xenografts in SCID mice. (a) Human DRG xenograft under the renal capsule of a SCID mouse 12 weeks post xenotransplantation. (b) SCID DRG xenografts retain organotypic features of normal human ganglia, magnification 100×, stained with hematoxylin and eosin. (c) SCID DRG xenograft 14 days after infection, white arrow indicates cytopathology in neurons, magnification 200×, stained with hematoxylin and eosin. (d) SCID DRG xenograft 14 days after infection stained with rabbit polyclonal antibody to VZV IE63 (brown, DAB signal) and counterstained with hematoxylin, magnification 200×. (e) SCID DRG xenograft 56 days after VZV infection showing no evidence of productive infection, magnification 200×, hematoxylin and eosin staining. (f) SCID DRG xenograft 56 days after VZV infection stained with rabbit polyclonal antibody to VZV IE63 (brown, DAB signal) and counterstained with hematoxylin, showing absence of IE63 expression, magnification 200×. (g) TEM of SCID DRG xenograft 14 days after VZV infection, white arrow points to virions in cytoplasm of neuronal cell nucleus, black arrow points to satellite cell nucleus, magnification 3,800×. (h) TEM of SCID DRG xenograft 14 days after VZV infection, virions are present in the nucleus and cytoplasm of a satellite cell

Full size image

4 VZV Neurotropism in Human DRG Xenografts

4.1 Acute VZV Replication in Human DRG

As VZV is a highly cell-associated virus, inoculation of DRG xenografts is accomplished by direct injection of VZV-infected fibroblasts (∼10 ul, 102–104 PFU). Unlike experimental infection of ganglia from non-permissive hosts, VZV infection of DRG is characterized by productive replication of the virus within different cell types that comprise host neuronal tissues. Cytopathic changes, such as denucleated neuronal cell bodies and cytoplasmic inclusion bodies, are apparent 10–14 days post infection (Fig. 1c) (Zerboni et al. 2005b). During this acute stage of infection, VZV regulatory proteins, such as IE63, are expressed within large numbers of neurons and satellite cells (Fig. 1d) (Zerboni et al. 2005b). Infectious virus can be recovered from productively infected DRG xenografts by mechanical disruption and coculture with a permissive cell monolayer up to 14 days post infection and correlates with detection of high viral genome copy numbers (108–109 copies/105 cells) in homogenized ganglia by quantitative real-time DNA PCR (Zerboni et al. 2005b).

Aggregate data from several investigations of VZV latency using cadaver ganglia implicate neurons within sensory ganglia, and, to a much lesser extent, satellite cells, as potential hosts for the latent VZV genome (Gilden et al. 1983; Kennedy et al. 1998, 1999; LaGuardia et al. 1999; Lungu et al. 1995; Mahalingam et al. 1993; Pevenstein et al. 1999; Wang et al. 2005). However, information on targets of VZV replication during acute infection and reactivation is limited. Investigation of acute VZV infection by correlative immunofluorescence-electron microscopy (IF-EM), which permits evaluation of histological and ultrastructural changes within the same cell, demonstrates a role for satellite cells as well as neurons in virion production (Reichelt et al. 2008). Ontogeny of infectious VZV requires assembly and packaging of viral capsids with a DNA core in the cell nucleus, egress of the nucleocapsid through the nuclear membrane to the cytoplasm, association with virion tegument proteins, and acquisition of the virion envelopment followed by release of infectious virus through egress pathways (Cohen et al. 2007). Intranuclear and cytoplasmic viral capsids were observed within neurons as well as satellite cells in acutely infected DRG xenografts by IF-EM, which indicates productive infection in both cell types (Fig. 1g, h; Reichelt et al. 2008). Moreover, extracellular particles were observed external to the plasma membranes of both cell types. VZV IE proteins, the early ORF47 protein kinase, and late viral glycoproteins were expressed on satellite cells as well as neurons (Reichelt et al. 2008; Zerboni et al. 2005b). Evidence of acute replication in satellite cells is interesting given the preponderance of evidence that the neuron is the primary site of latency. It is not clear if VZV is unable to persist long term in satellite cells or if particles detected in satellite cells are non-infectious. By IF-EM, virion particles within satellite cells contained no obvious ultrastructural defects and were similar to those observed in neurons (Fig. 1h) (Reichelt et al. 2008).

4.2 Acute VZV Infection of Human DRG Xenografts Is Associated with Polykaryon Formation

Our examination of VZV replication in human DRG xenografts in SCID mice showed the presence of VZV DNA, viral proteins, and virion production in both neurons and satellite cells. A hallmark of VZV infection in skin lesions and in cultured cells is cell–cell fusion with resulting polykaryon formation. We examined the role of polykaryon formation during acute infection of DRG between neurons and their surrounding satellite cells, which together form a neuron-satellite cell complex (NSC) (Reichelt et al. 2008).

Sections from VZV-infected DRG (14 days after infection) were examined by staining with antibodies for NCAM and VZV IE62 protein, as a marker for productive infection. NCAM is expressed on both the neuronal membranes encapsulating the NSC and within the NSC, providing separation between the neuronal cell cytoplasm and the satellite cell cytoplasm (Fig. 2a–c). Control (uninfected) DRG were stained with antibody to NCAM and synaptophysin, which is localized to neuronal cytoplasmic vesicles. Examination by confocal IF microscopy revealed a large proportion of NSCs in VZV-infected DRG sections with evidence of polykaryon formation, as judged by the absence of NCAM-expressing membranes within NSC; whereas no fusion was detected within NSC from uninfected DRG, which maintained separation as indicated by intact NCAM-stained membranes. NCAM expression was retained on membranes that surround the NSC in both infected and uninfected sections.

Fig. 2
figure 2

Evidence of polykaryon formation in acutely infected DRG by immunofluorescence and EM analysis. Cryosections of VZV-infected DRG (a–c) were stained with mouse monoclonal anti-NCAM antibody (a–c; green), rabbit polyclonal anti-synaptophysin antibody (a; red), rabbit polyclonal anti-IE62 antibody (b and c; red) and Hoechst-stain (a–c; blue). Texas Red-labeled goat anti-rabbit or FITC-labeled goat anti-mouse antibodies were used for secondary detection (a–c). Nuclei of satellite cells (s) and neurons (N) are marked and nuclei of satellite cells located within a putative polykaryon are indicated with an asterisk (a–c). White arrowheads point to cell boundaries detected by NCAM staining (green). Scale bars are 10 μm (a–c). (d–g) Acutely infected DRG were analyzed by standard EM. Black arrowheads point to the cell membrane surrounding satellite cells (s) or the neuron cell body (N) or a polykaryon (f and g). Black boxes in (d) and (f) indicate the area that is seen at higher magnification in the images (e) and (g), respectively. Note, that despite the normal morphology of the mitochondria, the ER and of the nuclear envelope in the polykaryon (f and g), no cell membranes can be detected between the nuclei of the polykaryon. Panels for composite reprinted from (Reichelt et al. 2008) with permission from American Society for Microbiology

Full size image

To provide additional evidence for polykaryon formation, ultrathin sections of acutely infected DRG were evaluated by TEM (Fig. 2d–g). Ultrastructural analysis revealed examples of both infected NSC in which no fusion between the neuronal cell body and the adjacent satellite cells had occurred, and others in which the neural and the satellite cell nuclei shared the same cytoplasm within a polykaryon. In the absence of fusion, the plasma membranes of both the satellite cells and the neuron were clearly identifiable and a double-membrane like structure resulted when two cells, each surrounded by its own plasma membrane, were adjacent within the same NSC. However, where fusion occurred, plasma membranes separating nuclei within the same cytoplasm were not observed.

Consistent with observations by confocal IF microscopy in VZV-infected NSCs in which polykaryon formation was observed, NCAM expression was detected on membranes that surround the infected NSC. Notably, these NSC polykaryons were engaged in viral replication and assembly, as indicated by the presence of numerous VZV nucleocapsids, intracellular viral particles within vacuoles, and extracellular virions associated with the limiting plasma membrane of the polykaryon. Most of these polykaryons did not show any obvious damage of the nuclear envelope or intracellular organelles, despite the presence of many nucleocapsids and enveloped virions in the cytoplasm, indicating that merging of the satellite and neuronal cell cytoplasm occurred during active VZV replication in NSC and was therefore not a late effect caused by cellular necrosis in VZV-infected sensory ganglia.

A similar finding of incomplete boundaries between neurons and satellite cells during acute DRG infection was observed in an ultrastructural examination of a VZV-infected trigeminal ganglion acquired at autopsy from a patient with active infection (Esiri and Tomlinson 1972). Taken together, these observations suggest that VZV replication in a single infected neuron can result in polykaryon formation with associated satellite cells within the NSC. Satellite cell infection and polykaryon formation in NSCs provide a mechanism for amplification of VZV upon entry into neuronal cell bodies or reactivation from latency, and may contribute to VZV spread to adjacent NSCs. This may, in part, explain the neuropathic consequences of VZV ganglionitis that occur during primary infection and episodes of herpes zoster (Gilden et al. 2006).

4.3 Longterm Persistence of VZV in Human DRG Xenografts Following Acute Infection

VZV latency can be characterized as the continued presence of non-replicating VZV genomes within neuronal ganglia. Whereas the extent of VZV gene transcription and translation during latency is controversial, that the VZV genome persists in the absence of infectious virus production, is a defining criterion.

At 8 weeks post infection, several areas within VZV-infected DRG xenografts contain histopathic indicators of a prior infectious process, such as fibrosis and nodules of Nageotte, which indicate neuronal cell loss (Zerboni et al. 2005b). However, clusters of neurons and satellite cells remain that do not exhibit any cytopathic effect (Fig. 1e). At this timepoint, infectious VZV can no longer be recovered from homogenized DRG xenografts; however, VZV genomes persist (Fig. 3a). Thus, as in natural infection of human sensory ganglia, VZV infection of DRG xenografts can be characterized as a biphasic process in which productive infection transitions to a phase in which the virus persists, but in a non-infectious form (Zerboni et al. 2005b; Zerboni et al. 2007). Presumably, these latent genomes retain the capacity to replicate and cause disease upon reactivation. However, axotomized DRG xenografts do not possess any neural connection to the mouse peripheral tissues and so we refer to this stage as “persistence” rather than latency (Zerboni et al. 2005b; Zerboni and Arvin 2008).

Fig. 3
figure 3

Assessment of VZV-infected DRG xenografts in SCID mice. (a) VZV genome copy numbers in VZV-infected (VZV-S strain) DRG xenografts in SCID mice evaluated by quantitative real-time DNA PCR at 14, 28 and 56 days post-infection using a probe to ORF62. (b) VZV mRNA transcripts evaluated by quantitative real-time RNA PCR at 56 days post-infection using probes to ORF31 (VZV gB, a late viral glycoprotein), IE62 and IE63. (c) Two mice with SCID DRG xenografts, the mouse on the left (1) was infected with VZV (pOka virus), the mouse on the right (2) was infected with VZV (pOka-63/70-CBR) expressing click beetle luciferase as an IE63/70 fusion protein. (d) Luminescence values (Photon flux) generated by in vivo expression of IE63. (top) pOka, non-luminescent control virus, (dark line) and pOka-63/70-CBR (grey line). (bottom) pOka-63/70-CBR infection in mice provided with valacyclovir in drinking water (solid line) and untreated mice (broken line). (e) VZV genome copy numbers in rOka and rOka ΔgI-infected DRG xenografts at 14 and 56 days post infection. (f) Immunoblot of DRG extracts using human polyclonal anti-VZV antibody, primarily reactive against VZV gE (top blot), in rOka-infected DRG at day 14 and day 56 post-infection; (bottom blots), immunoblot of DRG extracts using human polyclonal anti-VZV antibody (upper) or rabbit polyclonal antibody to IE63 (lower blot) in rOka and rOka ΔgI-infected DRG at 14 and 70 days post-infection. Panel (a) reprinted with permission from Zerboni et al. (2005b). Panels (c) and (d) reprinted with permission from Oliver et al. (2008). Panels (e) and (f) reprinted with permission from Zerboni et al. (2007)

Full size image

4.4 Viral Genomic DNA Burden in DRG Xenografts Persistently Infected with VZV

Estimates of the VZV DNA burden in latently infected cadaver ganglia have varied according to the methodology employed. Whereas some studies have detected low VZV genome copies (6–31 per 105 cells), other studies have reported 258 ± 38 copies and 557–55, 543 (mean 9,046 copies) per 105 cells (Cohrs et al. 2000; LaGuardia et al. 1999; Levin et al. 2003; Mahalingam et al. 1993; Pevenstein et al. 1999). In a recent study, VZV DNA was detected in 1.0–6.9% of neurons with ∼7 genomes/cell (Wang et al. 2005).

As with latently infected cadaver ganglia, VZV DNA can be readily detected in persistently infected DRG xenografts by PCR and the viral DNA burden can be quantified by real-time PCR (Zerboni et al. 2005b; Zerboni et al. 2007). We routinely detect 105 to 107 VZV DNA copies per 105 human cells in DRG xenografts at 56 days post infection, a 100-fold reduction when compared with VZV genome copies detected during acute infection (Fig. 3a) (Zerboni et al. 2005b, 2007). That this estimate is higher than estimates using cadaver ganglia is not unexpected, considering that cadaver ganglia are often examined decades following primary VZV infection.

4.5 Differential Regulation of Viral Gene Expression During VZV Persistence in Human DRG Xenografts

Various techniques have been employed to identify VZV genes transcribed during latent infection. Transcripts corresponding to ORFs 21, 29, 62, and 63 are frequently detected; ORF63 RNAs are most consistently detected and at greater abundance (>2,000-fold) that transcripts mapping to other ORFs (Cohrs et al. 1994, 1996, 2000; Gilden et al. 1987; Hyman et al. 1983; Kennedy et al. 1999; Meier et al. 1993).

We have used quantitative real time RNA PCR to detect viral mRNAs in persistently infected DRG xenografts (Zerboni et al. 2005b, 2007). VZV RNA corresponding to ORFs 62, 63, and 31 (encoding gB) are present in acutely infected DRG xenografts, as expected for productive infection. However, transcripts corresponding to ORF63, but not ORF31, are present at 56 days post infection (Fig. 3b). Detection of IE62 mRNA at late times post infection is inconsistent. This suggests that, as observed in latently infected cadaver ganglia, persistence of VZV in DRG xenografts is characterized by differential regulation of VZV gene expression such that late viral glycoproteins are not expressed. A similar block on glycoprotein expression was observed during VZV infection of human neural stem cells xenotransplanted in the brain of nonobese diabetic SCID mice (Baiker et al. 2004).

4.6 Expression of Viral Proteins During VZV Persistence in Human DRG Xenografts

Evidence of VZV protein expression during latency in human cadaver ganglia has relied exclusively upon detection by immunohistochemical-based methods, and verification by Western blot has not been reported (Cohrs et al. 2003; Grinfeld and Kennedy 2004; Lungu et al. 1998; Mahalingam et al. 1996). In the earliest report, VZV IE63 protein was detected in rare neurons from human ganglia recovered from two of nine individuals with no evidence of VZV reactivation at the time of death (Mahalingam et al. 1996). In more recent reports, IE63 protein – as well as proteins corresponding to ORFs 21, 29, and 62 – have been reported in a high proportion of neurons (5–30%) in all ganglia examined (Grinfeld and Kennedy 2004; Lungu et al. 1998). In that these experiments depend on immunostaining of neurons within cadaver tissues, which may exhibit postmortem physiological or biochemical aberrancies, and which are known to contain an abundance of age-related pigments which can interfere with histochemical tinctorial reactions, they must be carefully controlled (Barden 1981; Double et al. 2008).

Neural pigments are absent from fetal ganglionic tissues, and persistently infected DRG xenografts can be processed and fixed immediately upon recovery of the xenograft. In our examinations of tissue sections from persistently infected DRG xenografts, we have not detected any VZV proteins by immunohistochemical-based methods or immunoblot (Figs. 1f and 3f) (Zerboni et al. 2007). No IE63 protein was detected in DRG xenografts persistently infected with VZV, examined 70 days after infection (Fig. 3f; Zerboni et al. 2007).

To further assess IE63 protein expression in DRG xenografts, we employed live animal imaging and measured IE63 expressed as a luciferase fusion protein during the course of VZV infection (Oliver et al. 2008). Upon inoculation of DRG xenografts with pOka-63/70-luciferase, the progress of VZV infection was measured over a 70-day period by imaging every 3–7 days (Fig. 3c, d). As expected, luminescence increased from day 3 to day 15 during the acute phase of infection, and then declined sharply as the virus transitioned to a longterm persistent phase. At 41 days post inoculation, luminescence had declined to a level below that observed on day 3; by day 56 the level of luminescence was equivalent to that seen in the control virus, pOka, which lacked the luciferase cassette. This experimental system provides further evidence that IE63 protein is not expressed during VZV persistence in human DRG xenografts.

4.7 Evaluation of DRG Xenografts as a System for Assessment of Antiviral Drugs

Investigation of VZV infection in human DRG xenografts by live animal imaging using luciferase tagged viral proteins may also provide an opportunity to assess the efficacy of antiviral drugs in human neuronal ganglia in vivo. We evaluated the potential for this system by delivery of the antiviral drug valacyclovir in the drinking water of SCID mice during the course of DRG xenograft VZV infection (Oliver et al. 2008). Replication of herpes viruses is susceptible to the prodrug valacyclovir, and related nucleoside analogs, through inhibition of the viral thymidine kinase (Cohen et al. 2007). Valacyclovir is a widely used therapeutic for herpes zoster infection in immunocompromised patients.

We provided valacyclovir (1 mg/ml) ad libitum in the drinking water of SCID mice immediately after VZV infection of DRG xenografts and imaged mice over a 70-day interval (Oliver et al. 2008). As previously observed, in both treated and untreated mice, the IE63-luciferase signal increased steadily from days 3 to 15, indicating progressive infection (Fig. 3d). After 15 days, the luciferase signal decreased sharply in untreated mice, as expected for the transition to viral persistence. However, in mice receiving valacyclovir, the luciferase signal plateaued until day 28, when the drug was discontinued from the drinking water. Immediately upon withdrawal of valacyclovir, the IE63-luciferase signal decreased steadily such that signal kinetics were similar to those observed in untreated mice but shifted, with a delay in transition to persistence. Viral mRNA corresponding to ORF31 (gB, a late viral glycoprotein) was detected in DRG of treated but not untreated mice at 70 days post infection, which indicates a delay in cessation of viral replication and the transition to persistence in valacyclovir treated mice. Thus, valacyclovir treatment for 28 days following infection delayed the transition to viral persistence during the treatment interval, but upon withdrawal of the drug, infection kinetics returned to normal. This observation correlates with clinical experience that valacyclovir therapy in severely immunocompromised patients with VZV reactivation often results in relapse upon cessation of drug therapy (Boeckh et al. 2006; Manuel et al. 2008).

4.8 A Role for Innate Immunity During VZV Infection of DRG Xenografts

As discussed, VZV infection of human DRG xenografts is a biphasic process in which a brief replicative phase transitions to a non-replicative phase in which the virus persists longterm. VZV persistence is characterized by cessation of infectious progeny assembly, maintenance of the VZV genome at reduced copy numbers, disappearance of late gene transcripts, and a decline but persistence of low levels of IE63 (and some IE62) gene transcripts, and absence of VZV protein synthesis (Zerboni et al. 2005b, 2007). Remarkably, this transition from productive infection to persistence is achieved in the absence of adaptive immune responses, which SCID mice lack. This suggests that VZV has evolved an intrinsic mechanism to maintain viable neuronal sites in which to establish longterm persistence. Local innate immune responses, such as the production of alpha-interferon, may contribute to this process (Jones and Arvin 2006; Ku et al. 2004). Increased production of IFN-α has been observed in epidermal cells adjacent to VZV-infected cells within skin lesions, which may provide an innate barrier to block spread of VZV within skin xenografts (Ku et al. 2004). Investigation of local innate immune responses in human DRG xenografts in response to VZV infection may provide insight into the delicate balance that must be achieved between VZV and cellular immune responses that favor longterm persistence.

4.9 VZV T Lymphotropism May Facilitate Neurotropism

We investigated the potential for hematogenous spread of VZV in vivo by transfer of infectious virus to DRG xenografts via T lymphocytes. This work was based upon observations by Ku et al. (2004), which refined the prevailing hypothesis that VZV infection resembles the dual viremic model of mousepox (infectious ectromelia) (Grose 1981; Ku et al. 2004). According to this model, primary infection is acquired by inhalation of aerosolized infectious particles, which then replicate within regional lymphoid tissues before hematogenous spread (primary viremia) to reticuloendothelial organs. Following replication in internal organs, a secondary viremic phase delivers infectious particles to deep dermal sites, where they replicate within dermal cells progressively outwards to the epidermis forming cutaneous lesions 10–14 days following initial exposure.

Experimental infection of human T-lymphocytes in thymus/liver xenografts and in vitro demonstrated that VZV exhibits marked tropism for T cells, in particular memory CD4+ tonsil T cells expressing skin homing markers (Ku et al. 2002, 2004; Moffat et al. 1995; Zerboni et al. 2000). Ku et al. (2004) postulated that localized VZV replication in mucosal epithelial cells lining tonsillar crypts may transfer VZV to tonsil T cells and, in turn, VZV-infected memory CD4+ T cells expressing skin homing markers such as cutaneous leukocyte antigen or chemokine receptor-4 may transfer virus directly to skin (without amplification in reticuloendothelial organs) (Ku et al. 2002, 2004). We hypothesized that VZV may spread directly to sensory ganglia through a hematogenous route as well, negating the requirement for skin infection which permits neuronal access via retrograde transport from dermal nerve fibrils. The observation that VZV DNA is present in ganglia recovered from immunocompromised children who died during the incubation period (before the appearance of rash) provides clinical evidence that VZV may be delivered directly to DRG during cell-associated viremia (Cohen et al. 2007).

We investigated this alternative pathway to VZV neurotropism by adoptive transfer of VZV-infected tonsil T cells into SCID DRG mice via the tail vein (Zerboni et al. 2005b). DRG xenografts were homogenized at 14 days after transfer and infectious virus was recovered from one of four xenografts by coculture of the homogenate with a permissive cell line; VZV DNA was detected in two of four DRG by quantitative PCR (8.3 × 105 and 9.5 × 106 VZV genome copies per 105 human cells). These experiments indicate that VZV-infected T cells can traffic to and infect human neuronal ganglion cells in vivo; this is substantial evidence that VZV T cell tropism provides a second mechanism by which the virus reaches DRG sites for longterm persistence.

4.10 VZV “Oka” Varicella Vaccines Retain Neurotropism

Clinical and epidemiological observation indicate that the live attenuated “Oka” varicella vaccine has the capacity to cause herpes zoster in vaccine recipients, but appears to do so less often than wildtype virus (Arvin 2001). Vaccine Oka (vOka) was classically attenuated by serial passage of the Oka parental isolate (pOka) in human and guinea pig embryo fibroblast cell lines (Arvin 2001). The molecular basis for vaccine Oka attenuation is unknown; vaccine formulations appear to comprise heterogeneous genomes (Quinlivan et al. 2007).

Using the SCID mouse–human DRG xenograft model, we directly assessed neurotropism and neurovirulence of vOka and pOka viruses (Zerboni et al. 2005a). Cytopathic effect of vOka and pOka viruses was indistinguishable. VZV genome copies were equivalent at acute and persistent phases of infection with the expected decrease in VZV genomes between days 14 and 28 post infection and cessation of infectious virus production. These DRG experiments suggest that attenuation of pOka has not altered tropism for neurons or neurovirulence. In previous studies in skin and T cell xenografts, we observed that vOka replicates less efficiently than pOka in skin but both viruses retain T cell infectivity (Moffat et al. 1998; Zerboni et al. 2005a). Taken together, these data suggest that attenuation of vOka limits skin infectivity which in turn may limit access of virions to nerve termini. Efforts to completely eliminate vaccine virus latency will likely require development of second generation varicella vaccines with a molecular basis for altered neurovirulence.

5 Investigation of VZV Neurovirulence Using Recombinant VZV Mutants

5.1 Generation of VZV Recombinants

VZV is a highly cell-associated virus. Lack of infectious virus release into the extracellular media of cultured cells makes plaque purification of viral mutants derived by homologous recombination impractical (Cohen et al. 2007). To overcome this hindrance, VZV cosmids that contain overlapping fragments of the VZV genome have been constructed which, when cotransfected into a permissive cell line, yield infectious virus (Mallory et al. 1997). Cosmids used to generate infectious virus can be altered by PCR-directed mutagenesis to contain mutations in genes or sequences of interest. Cosmids spanning the VZV genome have been constructed from the attenuated vaccine Oka strain as well as from the original clinical isolate (parental Oka strain) (Mallory et al. 1997; Niizuma et al. 2003). Infectious VZV derived from bacterial artificial chromosome (BAC) technology has also been utilized (Nagaike et al. 2004; Tischer et al. 2007; Zhang et al. 2007).

Experimental infection of SCID mouse–human tissue xenografts in vivo has facilitated the identification of virulence determinants for skin and T-lymphocytes. Our investigations have demonstrated that in vitro replication phenotypes of recombinant viruses are not predictive of behavior within differentiated human cells within their unique tissue microenvironments. We have often observed that VZV genes and promoter elements that are dispensable for replication in cultured cells are required for, or significantly modulate, VZV infection in vivo. Evaluation of cosmid-derived VZV mutants in SCID DRG xenografts will make it possible to determine if particular gene products or motifs within VZV proteins affect neurotropism and neurovirulence. Our lab has generated over 100 recombinant VZV mutants in various genes including viral glycoproteins, viral kinases, regulatory proteins, and promoter elements (reviewed in Zerboni and Arvin ( 2008 )). So far, we have evaluated only a few of these viruses in SCID DRG xenografts but additional experiments are ongoing. Using these mutants, we hope to identify molecular mechanisms of VZV neuropathogenesis with the aim of defining the contributions of VZV genes and promoter elements to neuropathogenesis. These findings will have direct relevance for the design of neuroattenuated varicella vaccines.

5.2 The Role of Glycoprotein I in VZV Neurotropism

VZV gI (ORF67) and VZV gE (ORF68), encoded in the unique short (US) region of the VZV genome, are the major envelope glycoproteins (Cohen et al. 2007). As in other alphaherpesviruses, gE and gI form a noncovalently linked heterodimer which functions in cell–cell spread and viral envelopment. Whereas VZV gI is similar to orthologous gI proteins in human alphaherpesviruses, VZV gE is unique in that it is indispensable for virus replication (Cohen et al. 2007). VZV gE shares regions that are required for gE/gI heterodimer formation but also possesses a large, non-conserved N-terminal ectodomain. Although gI is dispensable for replication, it is critical for normal syncytia formation and spread in cultured cells (Moffat et al. 2002). Deletion of ORF67 has dramatic effects on intracellular localization of gE, causing aberrant punctate expression on cell surface membranes (Moffat et al. 2002). Virion envelopment in the trans Golgi network is severely affected as assessed by TEM.

Whereas ORF67 is dispensable for replication in vitro, gI functions are critical for pathogenesis of VZV infection in skin and T cell xenografts in vivo (Moffat et al. 2002). As efficient spread of VZV in skin tissues requires polykarocyte formation, the requirement for gI for skin virulence indicates that other fusogenic proteins (gB, gH and gL), cannot compensate for gI functions that facilitate fusion in skin. However, VZV spread in T cells does not involve fusion, and so the requirement for gI suggests that gI functions in normal cytoplasmic envelopment of virions and egress are essential for T cell tropism.

We examined the requirement of gI in VZV neurotropism by inoculation of DRG xenografts with rOkaΔgI, which lacks ORF67 (Zerboni et al. 2007). Whereas infection with the recombinant parental strain (rOka) initiated the expected short replicative phase followed by persistence in DRGs, rOkaΔgI was markedly impaired for replication and showed no transition to persistence up to 70 days after infection (Zerboni et al. 2007). While not strictly required for replication within DRG sensory neurons, replication was significantly altered in that VZV genome copies were 100-fold lower at 14 days after inoculation than with rOka and rose appreciably over an 8-week interval instead of the typical decrease (Fig. 3e). Histopathic effect at day 70 after infection was marginal, with only a few rare cells exhibiting cytopathology (Fig. 4b). Infectious virus production (release of infectious virus from homogenized xenografts) was greater at 70 days post infection (2/4 ganglia) than at 14 days post infection (1/11 ganglia). Membrane localization of gE, which usually surrounds the infected NSC, was greatly reduced with only marginal punctate foci within the neuronal cytoplasm (Fig. 4d). Ultrastructural analysis of rOkaΔgI-infected xenografts revealed unusual Golgi stacks in the cytoplasm of those few neurons that contained virus particles (Fig. 4e). Intracellular trafficking of gE was aberrant, with evidence of retention in the rough ER by immuno-EM (Fig. 4f).

Fig. 4
figure 4

Comparison of rOka-infected and rOka ΔgI-infected DRG xenografts. (a) Histopathic effect in rOka-infected DRG xenografts (white arrow) at 14 days post infection, magnification 400×, hematoxylin and eosin staining. (b) Rare neurons exhibiting cytopathic effect in rOka ΔgI-infected DRG xenografts (white arrows) at 70 days post infection, magnification 400×, hematoxylin and eosin staining. (c) rOka-infected DRG neuron at 14 days post-infection and rOka ΔgI-infected DRG neuron (d) stained by confocal IF with rabbit polyclonal antibody to IE63 (green) and mouse monoclonal antibody to gE (red) demonstrates punctate and cytoplasmic retention of gE in the absence of gI. (e) Immuno-EM of rOka ΔgI-infected DRG neuron with gold-labeled gE (black dots), white arrow denotes strange convolutions in Golgi structures. (f) Immuno-EM of rOka ΔgI-infected DRG neuron with gold-labeled gE (black dots), white arrow denotes gE retention in rough ER. Composite from panels reprinted with permission from Zerboni et al. (2007)

Full size image

In contrast to the lethality of deletion of gI in skin and T cell xenografts, absence of gI did not alter VZV neurotropism in that rOkaΔgI was infectious for human sensory neurons. However, without gI, the virus lifecycle was significantly impaired, in particular at late stages. Although gE was present in the virion envelope, gE intracellular trafficking was aberrant, Golgi stacks were disorganized, and gE was retained in the rough ER. These experiments illustrate that genetic requirements for VZV replication in skin and T lymphocytes may be more stringent than in sensory neurons and, most importantly, that targeted mutations that limit viral replication in vitro may have unintended consequences on replication within sensory neurons in vivo.

5.3 Effect of Targeted Mutation of gI Promoter Elements on VZV Neurotropism

The activities of herpes viral gene promoters, like cellular gene promoters, can be modulated by cellular transactivating proteins through ubiquitous promoter elements. Host cell regulatory proteins recognize consensus binding sites in VZV gene promoters. Many consensus binding sites of VZV gene promoters have been mapped and examined for their role in enhancing viral transactivation via IE62 or other VZV regulatory proteins. We have examined consensus sites in the promoter of gI by generation of reporter constructs as well as targeted mutation (Ito et al. 2003). The VZV gI promoter contains consensus binding sequences for interaction with specificity factor 1 (Sp1) and upstream stimulatory factor (USF) cellular proteins. Base pair substitution of the Sp1 and USF cellular transactivating binding sites alone or in combination had a dramatic effect on VZV infection of skin xenografts and T cell xenografts (Ito et al. 2003). Disruption of both Sp1 and USF promoter sites significantly reduced viral titers as well as plaque size in cultured cells. This dual Sp1/USF promoter mutant was unable to replicate in skin xenografts and produced only low viral titers upon infection of T cell xenografts. The overall effect of the Sp1/USF mutations were reduced gI expression, which indicate that less gI is required for VZV infection of T cells than in skin, where gI is required for fusion (Ito et al. 2003).

We examined the contributions of the Sp1 and USF binding site consensus sequences in the gI promoter on VZV neurotropism (Zerboni et al. 2007). Whereas disruption of both Sp1 and USF sequences significantly reduced skin virulence, we observed no effect on acute replication or persistence in DRG xenografts. Recombinant Oka with substitutions in the Sp1 sequence (rOka-gI-Sp1) and both the Sp1 and USF sequences (rOka-gI-Sp1/USF) replicated equally well and was equivalent to the parental rOka strain. No differences were observed in release of infectious virus or in histopathic effect. These observations demonstrate tissue-specific differences in the requirements for cellular transactivators known to enhance viral gene expression. Whereas virulence in skin and T cells depends upon functions of cellular transactivators, in this example VZ neurovirulence is independent of cellular transactivators. Understanding the requirements of tissue-specific VZV gene expression will facilitate a better understanding of the requirements for VZ neurovirulence.

6 Conclusion

The SCID mouse–human xenograft model for VZV pathogenesis has provided an opportunity to examine complex virus–host biological interactions within human tissues that are relevant to VZV pathogenesis, and within a small animal model that can be experimentally manipulated. Experimental infection of human skin, T cells, and sensory ganglia with recombinant VZV viruses that have targeted mutations of specific genes and regulatory elements enables the assessment of factors that influence virulence and tropism. In this chapter, we have focused on initial investigations of VZV infection of human DRG xenografts. We hope that our investigation of VZV neurotropism and neuropathogenesis using this model can improve our understanding of VZV neuropathogenesis with an aim to improving the clinical management of herpes zoster and the development of neuroattentuated varicella vaccines.