Productive Varicella-Zoster Virus Infection of Cultured Intact Human Ganglia

Abstract

Varicella-zoster virus (VZV) is a species-specific herpesvirus which infects sensory ganglia. We have developed a model of infection of human intact explant dorsal root ganglia (DRG). Following exposure of DRG to VZV, viral antigens were detected in neurons and nonneuronal cells. Enveloped virions were visualized by transmission electron microscopy in neurons and nonneuronal cells and within the extracellular space. Moreover, rather than remaining highly cell associated during infection of cultured cells, such as fibroblasts, cell-free VZV was released from infected DRG. This model enables VZV infection of ganglionic cells to be studied in the context of intact DRG.

Varicella-zoster virus (VZV) is the etiological agent for varicella (chicken pox) and herpes zoster (shingles) (1, 3). During primary VZV infection, VZV accesses nerve axons to reach sensory ganglia, where it establishes latency (4, 5, 7, 14, 15, 16, 19). Reactivation from latency results in new infectious virus and axonal transport of VZV to the skin to cause herpes zoster. During herpes zoster, neural and dermoepidermal inflammations occur, resulting in neuropathic pain and the typical dermatomal rash (8). The complex aberrant repair processes that occur during herpes zoster can result in chronic neuropathic pain (postherpetic neuralgia), which can last for years after resolution of the rash (26).

The high species specificity of VZV has complicated the development of small animal models that mimic productive infection (1, 17), although advances in tissue culture techniques enabled the development of SCID-hu mouse models utilizing grafted human tissue (17, 18, 20, 21). Grafting of neural cells has been used to examine infection of neurons and glial cells (2). In a variation of this model, intact human fetal dorsal root ganglia (DRG) have been grafted into SCID-hu mice to show that after initial productive infection, VZV persisted in a form consistent with the establishment of latency (29).

We have previously shown that single-cell preparations of neurons from dissociated human DRG support virus replication and that unlike productive infection of human fibroblasts (HFs), infected neurons are resistant to apoptosis (13). We also provided evidence that the VZV ORF63 gene product confers resistance to apoptosis during neuronal infection (12). In an extension of these studies, we sought to develop a human model of explant ganglia as a means to study features of VZV interactions with ganglionic cells within the context of intact ganglia.

Human fetal spinal tissue (14 to 20 weeks of gestation) was obtained from the Human Fetal Tissue Distribution Centre (Prince of Wales Hospital, NSW, Australia) following informed consent and with approval by the University of Sydney Human Research Ethics Committee. Individual intact DRG were isolated from fetal spinal tissue and cultured on glass coverslips as previously described (24). Axons typically developed at day 2 postexplant, and only DRG with extensive axonal growth were used for infection (Fig. 1A) . Due to the highly cell-associated nature of VZV in vitro (27), a cell-associated inoculation method was utilized. Single DRG explants, each cultured in 700 μl of neuronal culture media (Dulbecco's modified Eagle's medium with 0.5% fetal calf serum, 100 ng/ml nerve growth factor, 100 U/ml penicillin-streptomycin, 2 mM l-glutamine, and B-27 supplements; GIBCO, CA), were incubated with an inoculum consisting of 100 μl of media containing 1 × 10⁵ VZV strain Schenke-infected HFs at a cytopathic effect of 2+ or an equivalent number of mock-infected HFs. The inoculum was layered on top of the explant, with care taken not to disturb the ganglion, resulting in a total volume of 800 μl per well. DRG were collected at 0, 24, 48, 72, and 96 h postinfection (p.i.), fixed in 10% formalin, and paraffin embedded. DRG incubated with uninfected HFs were collected in parallel at each time point. Analysis of 5-μm sections stained with hematoxylin revealed neurons and nonneuronal cells, such as satellite cells, with sensory neurons being readily distinguished by their large sizes and centrally located nuclei (Fig. 1B and C). In addition, immunohistochemical (IHC) staining was performed for ganglionic cell markers. To detect proteins in formalin-fixed sections, antigens were unmasked using 0.01 M citrate buffer (pH 6) prior to incubation with primary antibodies to the neural cell adhesion molecule (NCAM; monoclonal mouse anti-human antibody; Chemicon Inc., CA) or to S100 proteins, which are expressed on satellite cells and some neurons (2, 9). The antibody used to detect S100 was a rabbit anti-cow S100 polyclonal antibody (Dako, Glostrup, Denmark). As described by the manufacturer, this antibody reacts strongly with human S100B, weakly with S100A1, and very weakly with S100A6 and does not react with other S100 proteins, such as S100A2, A3, and A4. Bound antibodies were detected using horseradish peroxidase and 3,3′-diaminobenzidine. NCAM-positive neurons were readily detected, as were surrounding smaller satellite cells expressing S100 (Fig. 1D and E). In both mock- and VZV-infected ganglia, the number of S100-expressing cells increased over time following explant, consistent with studies of mouse and rat sensory ganglia reporting that the process of explanting ganglia can induce satellite cell proliferation (6, 28).

FIG. 1.

Culture of intact human explant DRG. (A) Intact DRG cultured on glass coverslips in the presence of nerve growth factor (100 ng/ml) showing extensive axonal growth (arrows) at 48 h postplating. Magnification, ×10. (B, C) Hematoxylin staining of 5-μm paraffin-embedded DRG sections showing large, centrally nucleated neurons (arrow) surrounded by smaller support cells (arrowhead) at ×40 (B) and ×100 (C) magnification. Results for NCAM (D) and S100 (E) staining within VZV-infected DRG are shown. Magnification, ×20. (F) Negative control consisting of VZV-infected DRG section incubated with isotype control mouse immunoglobulin G2a showing no specific staining.

VZV glycoproteins are not expressed at detectable levels during latency, so their detection is a useful indicator of productive infection (10). To assess the extent of infection, DRG sections were examined for VZV antigens by immunofluorescent assay (IFA) staining with either mouse monoclonal antibody to VZV glycoprotein I (Chemicon Inc., CA) or a human hyperimmune serum that recognizes predominantly glycoproteins (kindly provided by A. Arvin, Stanford University), followed by secondary AlexaFluor 594 antibodies. In addition to IFA staining, sections were also subjected to IHC detection with mouse monoclonal antibody to VZV glycoprotein B.

Viral glycoproteins were not detected at 24 h p.i., but by 48 h p.i., distinct individual VZV-positive neurons scattered throughout the DRG were identified (Fig. 2A to C). Viral-antigen-positive cells were also detected around the edge of the DRG body, likely representing infection from direct contact with the infected HF inoculum. The presence of discreet VZV-positive neurons deep within ganglia at 48 h p.i. indicated that infection of these cells may have occurred via transport of virus through neuronal axons that reached beyond the body of the ganglia into the culture media. At 72 and 96 h p.i., infection was much more widespread, with a majority of neurons and nonneuronal cells being VZV antigen positive, indicating that most DRG cells support viral replication (Fig. 2D and data not shown). No staining was observed in mock-infected DRG or VZV-infected DRG stained with isotype control antibodies (Fig. 2E and data not shown). Comparable results were obtained from four replicate experiments using ganglia from different fetal samples.

FIG. 2.

VZV antigen expression in infected explant human DRG. (A) IFA staining of DRG 48 h p.i. with human VZV hyperimmune serum and secondary antibody consisting of fluorescently conjugated anti-human AlexaFluor 594 (red staining). VZV antigen expression on the surfaces of distinct, scattered neurons (white arrows) and around the DRG body (green arrows). The boxed inset shows a magnified image of VZV antigen-positive neurons. (B) IFA staining of DRG at 48 h p.i. with mouse anti-VZV gI monoclonal antibody and secondary antibody consisting of fluorescently conjugated anti-mouse AlexaFluor 594 (red staining). Nuclear blue DAPI (4′,6′-diamidino-2-phenylindole) staining is indicated. (C to E) IHC detection of VZV-infected DRG stained with mouse anti-VZV gB monoclonal antibody at 48 h p.i. (C) and 72 h p.i. (D) (brown staining) or mock-infected DRG (E). Black arrows indicate infected neurons. Sections were counterstained with hematoxylin (blue staining).

The increase in the number of infected cells over time suggested that VZV productively infected and spread within explanted DRG. We next used transmission electron microscopy (TEM) to examine DRG for the presence of VZV particles. DRG were collected at 48, 72, and 96 h p.i. and processed for TEM as described previously (24). Viral capsids or virions were readily identified in cells throughout DRG cells in cell nuclei and cytoplasm, with the highest number detected at 96 h p.i. (Fig. 3). Neurons containing virions were confirmed by their characteristic morphological features, including microtubules in the cytoplasm and adjoining axonal processes. Infected neurons in the intact DRG did not show ultrastructural changes indicative of apoptosis, which is consistent with our previous findings that infection of neurons from dissociated DRG are resistant to apoptosis (12, 13). In addition to being detected intracellularly, enveloped virions were also detected in the extracellular space (Fig. 3C and D).

FIG. 3.

Transmission electron micrographs of VZV-infected human explant DRG. (A) Detection of numerous VZV capsids (arrow) in a DRG cell nucleus at 96 h p.i. (B) Viral particles present within the nucleus (arrow) and cytoplasm (arrowhead) are indicated. (C) Presence of virions in the cytoplasm (arrowhead) and in the extracellular space between cells (black arrow). Microtubular structures (white arrow) indicative of a neuronal cell. (D) Magnified image of inset from panel C, showing a fully assembled extracellular, enveloped virion (arrow).

VZV remains highly cell associated and is not released during productive infection of cultured cells, such as HFs, and explanted skin tissue infected in vitro (25, 27). However, cell-free VZV is detectable in vesicular fluid from varicella and herpes zoster skin lesions (22) as well as in infected human skin xenografts in SCID-hu mice (21). To determine whether VZV virions remained cell associated or were released from infected DRG explants, pooled culture supernatants from eight individually cultured or VZV- or mock infected DRG were collected at 24, 48, 72, 96, and 120 h p.i. The supernatants were clarified by centrifugation at 10,000 × g for 10 min before being inoculated onto HF monolayers for the detection of cell-free virus by a plaque assay. In parallel, supernatants from infected HFs were similarly assessed for the presence of cell-free virus. Plaques were not detected from infected HF supernatants at any time point, confirming the highly cell-associated nature of VZV infection of these cells. In contrast, plaques were detected from infected DRG supernatants, demonstrating release of infectious virus (Fig. 4). The presence of cell-free virus peaked at 96 h p.i. Comparable results were obtained in an additional three independent replicate experiments. The decline in the number of VZV plaques released at 120 h p.i. may have been due to a general loss of DRG cell viability due to culture conditions, as the histologies of both mock and infected DRG had begun to deteriorate by this time postexplant.

FIG. 4.

Detection by plaque assay of cell-free infectious VZV released from cultured human DRG. Shown are the numbers of plaques formed on HF monolayers after the addition of pooled culture supernatants from eight individually infected DRG or infected HF inoculum-alone samples over a time course of infection. A plaque assay was performed in triplicate, and the mean numbers of plaques (± standard errors of means) from a representative experiment out of four independent experiments are shown. No cell-free virus was detected from the infected HF inoculum control.

In addition to the detection of infectious-cell-free-virus release, the number of infected ganglionic cells was determined by an infectious-center assay. Infected DRG harvested at 48 and 96 h p.i. were washed in Hank's balanced salt solution and dissociated with collagenase and dispase to generate single-cell suspensions, which were inoculated onto HF monolayers. The number of infectious centers increased over time, from approximately 8,000 at 48 h p.i. to approximately 48,000 at 96 h p.i., confirming the spread of productive infection within DRG.

This study provides a model whereby the interaction of VZV with ganglionic cells, which play a critical role in viral pathogenesis, can be studied in the context of intact DRG, using in vitro cell culture techniques. The relative contributions of neuronal and nonneuronal cells to the observed release of cell-free virus remain an important component for future work for better understanding the interaction of VZV with these cells.

These results provide the first evidence of productive VZV infection and release of infectious cell-free virus from cultured human DRG. Extensive neuron-to-neuron spread during reactivation within ganglia in vivo has been suggested to occur in herpes zoster (16), and we observed a similar spread of virus in experimentally infected DRG. These features of DRG infection can now be studied in further detail to better define the molecular mechanisms that underlie VZV infection of ganglionic cells. For example, this model provides a means to rapidly test viral gene mutant viruses and new candidate vaccine strains containing targeted gene disruptions, to define viral genes that may play critical roles in VZV neurotropism, and to examine in detail the outcomes of infection of both neurons and nonneuronal cells with respect to apoptosis and cell function. Infection of cultured intact DRG also enables both anterograde and retrograde axonal VZV transport to be examined for the first time, particularly when combined with two-chamber culture plates, as we have done previously to study herpes simplex virus type 1 (HSV-1) axonal transport (11, 23, 24). In addition, it will now be possible to directly compare VZV and HSV DRG infection to better determine whether the fundamental differences between the natures of the dermatomal rash in herpes zoster and the much more highly localized lesions observed during HSV reactivation are due to differences in the spreads of infection within DRG by these closely related alphaherpesviruses.