ABSTRACT
Varicella-zoster virus (VZV) is a highly species-specific herpesvirus that targets sensory ganglionic neurons. This species specificity has limited the study of many aspects of VZV pathogenesis, including neuronal infection. We report development of a highly efficient neuroblastoma cell model to study productive VZV infection of neuronal cells. We show that differentiation of SH-SY5Y neuroblastoma cells yields a homogenous population of neuron-like cells that are permissive to the full VZV replicative cycle. These cells supported productive infection by both laboratory and clinical VZV isolates, including the live varicella vaccine. This model may enable rapid identification of genetic determinants facilitating VZV neurotropism.
TEXT
Varicella-zoster virus (VZV) is a human herpesvirus responsible for the two clinically distinct diseases varicella (chickenpox) and herpes zoster (shingles) (2, 8). During primary infection, VZV is transported to the sensory ganglia, where the virus establishes lifelong latency (2, 37, 49). It is thought that neurons are the major reservoir of VZV during latency, when a small subset of viral transcripts and proteins are expressed (14, 19, 26, 28, 30, 31, 34). VZV reactivation is characterized by the transition from latent to productive neuronal infection in which the subsequent anterograde transport of the virus back to the skin results in herpes zoster (18).
Development of suitable animal models for the study of VZV infection has been hampered by the strict human-specific tropism of the virus (2, 38). In the context of neuronal infection, SCIDhu mice grafted with human fetal dorsal root ganglia (DRG) and infected cotton rats have both been reported as models to study VZV persistence or latency (1, 9–13, 29, 40–44, 52–56). In addition, our laboratory has previously described in vitro models to assess VZV neuropathogenesis using dissociated human fetal sensory neurons and intact explant human fetal DRG cultures (21, 24, 25, 47). However, the restricted availability of human tissue required for these models using human ganglia poses limitations on their use and has impeded progress in our understanding of this critical aspect of VZV infection.
Neuroblastoma cell lines have previously been shown to provide highly valuable models in the analysis of the neuropathogenesis and neurotropism of a range of viruses due to the ability of the cells to mimic the morphological and biochemical characteristics of primary neurons (45). Our current study utilized the SH-SY5Y cell line, a derivative of primary neuroblastoma cells obtained by successive cloning (4), as these cells have previously been used successfully for viral tropism studies using human cytomegalovirus (36), assessment of neurovirulent poliovirus strains (33), and as a model of herpes simplex virus latency and reactivation (50). However, few studies have endeavored to enhance these characteristics by using the cells in a fully differentiated form, a step thought to be crucial in the development of the neuronal phenotype (16). Thus, the focus of this study was to develop a highly efficient in vitro neuroblastoma human cell line model available for the study of productive VZV infection within a neuronal context. Such a model would provide a means to more rapidly examine key aspects of the interaction between VZV and human neuronal cells during the productive phase of infection. In particular, a model of this nature may overcome many of the limitations of models requiring fresh primary human neural tissue, including donor heterogeneity and availability, limited and/or variable cell yield, complex and costly experimental procedures, and ethical considerations.
SH-SY5Y cells grown in culture flasks were first induced to differentiate with culture medium (Dulbecco's modified Eagle's medium F-12 [DMEM F-12] containing 5% fetal calf serum and 50 IU/ml penicillin and streptomycin; Gibco, CA) supplemented with 10 μM all-trans retinoic acid (ATRA) (Sigma, Australia) for 10 days. The cells were seeded onto coverslips coated with Matrigel (BD Biosciences, Australia) within 24-well culture plates at a density of 1 × 105 cells/well and allowed to adhere overnight. The cells were then incubated in serum-free DMEM F-12 supplemented with 10 μM ATRA and 0.5 μg/ml brain-derived neurotrophic factor (BDNF) (Invitrogen, Australia) for a further 5 days (16).
To validate the differentiation procedure, SH-SY5Y cell morphology was assessed using phase-contrast microscopy (Fig. 1 A to C). While untreated cells possessed large, flat cell bodies, differentiated cells formed vast, branching neuritic networks with small, rounded cell bodies. Quantitation of the proportion of cells exhibiting neurites following each treatment period (Fig. 1D) showed the combination of ATRA and BDNF to be the most effective treatment for inducing neurite formation, with 69% of cells containing at least one projection, but usually multiple projections. In contrast, only 53% of cells treated with ATRA alone or 7% of untreated cells were considered neurite positive (Fig. 1D). A rabbit antisynaptophysin antibody (Invitrogen, Australia) followed by an Alexa Fluor 488-conjugated anti-rabbit IgG antibody (Invitrogen, Australia) were used in immunofluorescence assays (IFAs) to assess the expression of neuron-specific synaptophysin protein (Fig. 1E and F), a neuronal differentiation marker (6, 46). Synaptophysin staining was readily observed localized to the cytoplasm and neuritic processes of the majority of differentiated SH-SY5Y cells (Fig. 1F). In contrast, undifferentiated cells exhibited minimal staining (Fig. 1E).
Evaluation of SH-SY5Y cell differentiation using all-trans retinoic acid (ATRA) and brain-derived neurotrophic factor (BDNF). (A to C) The morphology of SH-SY5Y cells was assessed by phase-contrast microscopy before treatment with any differentiating agent (A) and after 10 days of ATRA treatment (10 μM) (B) or 10 days of ATRA treatment (10 μM) and a subsequent 5 days of treatment with ATRA (10 μM) and BDNF (0.5 ng/ml) (C). (D) The proportion of cells exhibiting neurites was counted at each time point posttreatment. Values are means plus standard errors of the means (error bars) from three independent experiments. Significant difference was determined using a one-tailed, paired Student's t test (*, P < 0.05; **, P < 0.01). RA, retinoic acid. (E and F) Undifferentiated SH-SY5Y cells (E) or SH-SY5Y cells differentiated with ATRA and BDNF (F) were stained with an antisynaptophysin antibody (green) and counterstained with DAPI (4′,6-diamidino-2-phenylindole) (blue).
In order to assess the susceptibility of differentiated SH-SY5Y cells to VZV infection, a commonly used cell-associated inoculation model was utilized to overcome the highly cell-associated nature of VZV in vitro (51). Highly permissive human foreskin fibroblasts (HFFs) infected with VZV strain rOka were first labeled with 1 μM carboxyfluorescein diacetate succinimidyl ester (CFSE) and then inoculated onto differentiated SH-SY5Y cells at a density of 1 × 105 cells/well in parallel with CFSE-labeled mock-infected HFFs. Labeling the inoculating HFFs with CFSE provided an additional means to readily distinguish these cells from SH-SY5Y cells. The cells were harvested at various time points postinfection (p.i.) for analysis by IFA.
To elucidate the expression and localization of VZV antigens within infected differentiated SH-SY5Y cells, cultures were harvested at 24 and 36 h p.i., fixed in 4% paraformaldehyde (PFA), and IFAs were performed using VZV-specific antibodies. As productive VZV infection is characterized by expression of all three kinetic gene classes, while abortive or latent infection results in a premature shutdown of viral gene expression (23), cultures were assessed for expression of representative immediate-early (IE), early (E), and late (L) proteins. Cells were incubated with antibodies against IE62 (Meridian Life Science Inc.), an IE protein, pORF29 (kindly provided by P. Kinchington, University of Pittsburgh), an E protein, and gE (Millipore, Australia), a L protein. Bound antibodies were detected using species-specific Alexa Fluor 594-conjugated antibodies. VZV antigen-positive SH-SY5Y cells (CFSE negative) for each kinetic class were observed at all time points in studies p.i. Furthermore, the localizations of all VZV antigen markers were consistent with productive VZV replication, with IE62 (Fig. 2 A) and pORF29 (Fig. 2B) detected in the nucleus and gE (Fig. 2C) expressed on the cell surface and throughout the cytoplasm (15, 32, 35, 39). As replication centers form within the nucleus in areas of viral genome synthesis during productive infection, our detection of punctate nuclear staining of IE62 and pORF29 in SH-SY5Y cells provides additional evidence of virus replication (39). Mock-infected cultures stained with a VZV gE antibody showed no antigen-specific staining (Fig. 2D). No specific staining was observed in VZV-infected SH-SY5Y cells incubated with isotype control antibodies (data not shown). Comparable results were obtained in three independent experiments. Thus, it was concluded that differentiated SH-SY5Y cells are able to support productive VZV infection.
Analysis of VZV antigen expression by immunofluorescence staining (IFA) of VZV-infected SH-SY5Y cells. Differentiated SH-SY5Y cells were inoculated with carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled VZV rOka-infected human foreskin fibroblasts (HFFs) (green) and fixed 36 h p.i. IFA staining was performed using antibodies against VZV IE62 (A), pORF29 (B), or gE (red) (C), followed by a DAPI counterstain (blue). Insets show enlarged images of VZV antigen-positive cells. (D) No VZV-specific staining was observed in parallel mock-infected cultures.
To extend the IFA-based analysis of VZV infection of SH-SY5Y cells, flow cytometry was utilized. In addition, the proportion of SH-SY5Y cells infected with VZV strain rOka was compared with infection of these cells using a clinical isolate, VZV S. Furthermore, in parallel with this analysis of SH-SY5Y infection, uninfected HFFs were seeded 24 h prior to infection at a density of 1 × 105 cells/well and inoculated in the same manner as differentiated SH-SY5Y cells to allow a direct comparison of infection efficiency between cell types. In both infection settings, the VZV-infected HFFs used as the inoculum were labeled with CFSE to enabled newly infected cells to be readily distinguished from the inoculating HFFs.
At 24, 48, and 72 h p.i., the cells were fixed in 1% paraformaldehyde and incubated with either a mouse anti-VZV IE62- or gEgI-specific antibody (Meridian Life Science Inc.) followed by an anti-mouse IgG Alexa Fluor 647-conjugated antibody, and the proportion of VZV-infected SH-SY5Y cells (i.e., VZV antigen positive, CFSE negative) was determined. The use of an alternate glycoprotein-specific antibody served as an additional means of confirming productive infection in these cells. VZV antigen-positive SH-SY5Y cells were readily detected following infection with both VZV rOka and VZV S, whereas no specific staining was observed in isotype control or mock-infected cultures (Fig. 3 A). The combined results of three independent experiments showed that the proportions of IE62- and gEgI-positive SH-SY5Y cells were comparable in cells infected with VZV rOka and VZV S at each time point, with no significant difference in the efficiency of infection of these two viruses (Fig. 3B). Furthermore, VZV antigen detection in SH-SY5Y cells increased between 24 and 72 h p.i., peaking at an average value of 39% for VZV rOka and 36% for VZV S, although this was not as high as the rates of infection achieved in HFFs assayed in parallel (Fig. 3B). Cytopathic effect (CPE) was detected at later times postinfection, as indicated by rounding of cells and detachment from the culture flask surface (data not shown). Taken together, these experiments demonstrated that differentiated SH-SY5Y cells were permissive to productive infection by both VZV rOka and VZV S.
Kinetic analysis of VZV antigen expression by flow cytometry of VZV-infected SH-SY5Y cells. HFFs that were mock infected or infected with VZV strain rOka or clinical isolate VZV S were labeled with CFSE and inoculated onto cultures of differentiated SH-SY5Y cells and harvested for immunostaining and flow cytometry. In parallel, cultures of HFFs were infected using the same method. (A) Flow cytometry scatter plots of cells 72 h after infection with rOka or VZV S following staining with antibodies against VZV IE62 or gEgI. Mock-infected cultures and VZV-infected cultures stained with isotype control antibody were included as controls. The percentages of SH-SY5Y cells (i.e., CFSE-negative cells) that were VZV antigen positive are shown in the top left corners of the graphs. (B) Graph showing the percentage of VZV antigen-positive (VZV+) SH-SY5Y cells and HFFs over a time course of infection with VZV rOka or VZV S. Values are means plus standard errors of the means (error bars) from three independent experiments.
It has previously been reported that pure neuronal cell cultures inoculated with cell-free VZV undergo persistent infection, in contrast to cell-associated inoculation, which results in productive VZV infection (5). We therefore sought to investigate infection of differentiated SH-SY5Y cells with cell-free VZV (at a multiplicity of infection [MOI] of 0.014) using the commercially available varicella vaccine (Varivax; Merck) which comprises a cell-free preparation of VZV strain vOka. Seven days p.i., the cultures were fixed and stained with VZV-specific antibodies and appropriate fluorescently conjugated secondary antibody as described above. The presence and cellular localization of cells positive for IE62 (Fig. 4 A), pOR29 (Fig. 4B), and gEgI (Fig. 4C) in these cultures indicated that the cells were undergoing productive VZV replication. In contrast, no specific staining was observed in isotype control cultures (Fig. 4D and E). In order to confirm that these infected SH-SY5Y cells were capable of transmitting infectious VZV to another permissive cell type and thus complete the cycle of productive infection, infectious center assays were performed. Single wells containing vOka-infected SH-SY5Y cultures were trypsinized and inoculated onto uninfected HFF monolayers. Immunofluorescent staining performed 7 days p.i. for VZV gEgI revealed the presence of VZV antigen-positive plaques within the HFF cultures (mean of 40 and range 37 to 41 from 3 independent experiments), thus verifying the spread of VZV from infected SH-SY5Y cells to HFFs (Fig. 4F). Our demonstration that inoculation of SH-SY5Y neuronal cells with cell-free VZV resulted in productive infection rather than a latent infection contrasts with a study using guinea pig enteric neurons (5). This may be due to differences in the nature of the inoculum used, the different properties of cell line cultures compared with primary tissues, and the species from which the cells or tissues were derived. Nonetheless, it may be possible in the future to modify the SH-SY5Y cell productive infection model (e.g., by inclusion of viral DNA synthesis inhibitors) to derive a model which would abort viral replication infection and enable viral genome persistence.
Infection of differentiated SH-SY5Y cells with cell-free Varivax vaccine. (A to C) At 7 days p.i., the cells were fixed and stained with antibodies against VZV IE62 (A), pORF29 (B), or gEgI (C) (red), in conjunction with a DAPI counterstain (blue). (D and E) No antigen staining was observed in cultures stained with corresponding isotype control antibodies. (F) To confirm the spread of VZV from Varivax-inoculated differentiated SH-SY5Y cultures, the cells were trypsinized and inoculated onto HFF monolayers. After 7 days, HFF monolayers were stained with a VZV gEgI (red) antibody and DAPI counterstain (blue) to visualize plaques.
This study describes an efficient neuronal cell culture model of VZV replication which may overcome many of the limitations accompanying studies in human ganglionic tissue or animal models (48). As little is known so far regarding the impact of VZV on neuronal gene regulation, this model will enable a more detailed transcriptomic and/or proteomic analysis of VZV-neuronal cell interactions. In addition, the high-throughput nature of this model renders it ideal for testing compounds with potential antiviral functions. The current VZV vaccine strain still retains the capacity to establish latency within the DRG and subsequently reactivate (17, 20). Although infrequent, a number of cases have recently been reported involving severe complications following reactivation of the vaccine strain (7, 22, 27). Therefore, identification of VZV genes responsible for neuronal infection is considered a priority in order to aid the design of a safer “second-generation” vaccine, which would contain targeted mutations of neurotropic genes (3). Thus, this neuroblastoma cell culture model is likely to provide a means to rapidly screen recombinant viruses with targeted gene deletions or modifications to identify roles in neurotropism, which would then enable selective examination in models of infection in primary ganglionic tissue.
ACKNOWLEDGMENTS
We thank Louise Cole of the Bosch Institute, University of Sydney, for assistance with microscopy and Paul Kinchington, University of Pittsburgh, for kindly providing the pORF29 antibody.
FOOTNOTES
- Received 14 March 2011.
- Accepted 24 May 2011.
- Accepted manuscript posted online 1 June 2011.
- ↵* Corresponding author. Mailing address: Department of Infectious Diseases and Immunology, University of Sydney, Blackburn Building, Room 601, Sydney 2006, New South Wales, Australia. Phone: 61 2 93516867. Fax: 61 2 93513968. E-mail: allison.abendroth{at}sydney.edu.au.
REFERENCES
- Copyright © 2011, American Society for Microbiology. All Rights Reserved.