Neonatal Neural Progenitor Cells and Their Neuronal and Glial Cell Derivatives Are Fully Permissive for Human Cytomegalovirus Infection

Abstract

Congenital human cytomegalovirus (HCMV) infection causes central nervous system structural abnormalities and functional disorders, affecting both astroglia and neurons with a pathogenesis that is only marginally understood. To better understand HCMV's interactions with such clinically important cell types, we utilized neural progenitor cells (NPCs) derived from neonatal autopsy tissue, which can be differentiated down either glial or neuronal pathways. Studies were performed using two viral isolates, Towne (laboratory adapted) and TR (a clinical strain), at a multiplicity of infection of 3. NPCs were fully permissive for both strains, expressing the full range of viral antigens (Ags) and producing relatively large numbers of infectious virions. NPCs infected with TR showed delayed development of cytopathic effects (CPE) and replication centers and shed less virus. This pattern of delay for TR infections held true for all cell types tested. Differentiation of NPCs was carried out for 21 days to obtain either astroglia (>95% GFAP⁺) or a 1:5 mixed neuron/astroglia population (β-tubulin III⁺/GFAP⁺). We found that both of these differentiated populations were fully permissive for HCMV infection and produced substantial numbers of infectious virions. Utilizing a difference in plating efficiencies, we were able to enrich the neuron population to ∼80% β-tubulin III⁺ cells. These β-tubulin III⁺-enriched populations remained fully permissive for infection but were very slow to develop CPE. These infected enriched neurons survived longer than either NPCs or astroglia, and a small proportion were alive until at least 14 days postinfection. These surviving cells were all β-tubulin III⁺ and showed viral Ag expression. Surprisingly, some cells still exhibited extended processes, similar to mock-infected neurons. Our findings strongly suggest neurons as reservoirs for HCMV within the developing brain.

Human cytomegalovirus (HCMV) is a ubiquitous betaherpesvirus that is the most common cause of virus-induced birth defects. Primary infection during pregnancy poses a 30 to 40% risk of intrauterine transmission, with adverse outcomes more likely if the infection is within the first half of the gestation period (37). Each year, about 1% of all newborns are congenitally infected, and 5 to 10% of these infants manifest signs of serious neurological defects, including deafness, mental retardation, blindness, microencephaly, hydrocephalus, and cerebral calcification (1, 3, 37). An additional 10% of congenitally infected infants are asymptomatic at birth and subsequently develop brain disorders, the most common of which is sensorineural hearing loss (7, 27). Thus, in the United States, some 8,000 children annually suffer the obvious consequences of congenital HCMV infection. Additionally, recent investigations have suggested that more subtle abnormal changes in human brain development, such as autism (with an incidence of 1:200), may in some cases be related to congenital HCMV infection (38, 43).

Although HCMV has a wide range of permissiveness in vivo (35), the fetal brain is the main site of the drastic manifestations of HCMV infection. It has been suggested that the severity of the neuropathological changes and clinical outcomes may be associated with the stage of central nervous system (CNS) development at which congenital infection occurs (2), yet the mechanism of HCMV pathogenesis in the developing CNS remains poorly understood. Since studies of HCMV in human subjects have obvious limitations, model systems of both in vitro and in vivo HCMV infection have provided insights into HCMV infection of the developing human brain. Most studies of congenital infection have been performed in the mouse model, as HCMV and mouse CMV (MCMV) are similar in genome size and structure, virion morphology, and overall pathogenesis. These studies have shed some light on the permissiveness of different CNS cell types, particularly with respect to the timing of infection.

Studies of mice revealed that early mouse embryos were nonpermissive for MCMV infection, as judged by a lack of viral gene expression after either blastocyst (10) or zygote (40) injection. Mouse embryonic stem cells were also nonpermissive for MCMV infection; however, after cell differentiation, mouse embryonic stem cells became susceptible and permissive for MCMV (21). Later in development, mouse multipotent CNS stem cells isolated from the ventricular/periventricular zones of embryonic and adult mouse brains were also permissive for infection. In fact, MCMV infection inhibited CNS stem cell proliferation and differentiation, with neuronal differentiation being more severely inhibited than glial differentiation (15).

Radial glial cells, which play an important role in guiding neuron migration, were the main targets of the virus during infection in the neonatal mouse (29, 42). In addition, immunostained brain slice cultures indicated that virus-susceptible cells were located in the subventricular zone and cortical marginal regions (areas positive for neural stem cell/progenitor markers) (5, 11). Shinmura et al. (34) found that injection of MCMV into the cerebral ventricles of mouse embryos caused a disturbance of neuronal migration and a marked loss of neurons. They proposed that this might be a cause of microencephaly due to CMV infection. A later paper from the same group (14) suggested that infected neurons in the cortex could persist and produce low levels of virus. These persistently infected neurons appeared to escape detection by natural killer cells and macrophages. Together, these studies of the mouse indicate that progenitor cells, as well as glia and neurons, are permissive for CMV infection and that the neurons may play a role in the long-term ramifications of viral infection within the developing brain.

Due to the inherent difficulty of obtaining and working with human primary neural cultures, much work was initially performed using established/transformed human CNS cell lines. On the basis of this work, it appeared that several neural cell types were permissive (at least to some extent) for HCMV infection (9, 18, 30, 36). Although these studies were important as initial characterizations, they provided only limited information on the permissiveness of untransformed, primary cells within the CNS. Therefore, we pursued a system of studying infection in primary human cells isolated from the CNS to confirm and extend the earlier human and mouse studies.

Although human neural progenitor cells (NPCs) can be extracted from brain tissue and cultured in vitro, scant research has been performed using cells derived from human fetal/neonatal tissue. Published studies have described somewhat conflicting results, but some clear messages can be gleaned: (i) NPCs, from which all of the main CNS cell types are derived, were permissive for infection and likely released virions (6, 22, 25, 26) and (ii) astrocytes were fully permissive and released significant quantities of virus (6, 16, 17, 22). The data on both microglia (16, 17, 31) and neurons (6, 17, 22) were conflicting, with some studies claiming these cell types to be somewhat permissive (shedding small amounts of virus) (16, 22, 31) and others claiming both cell types to be nonpermissive (6, 17).

Confounding the interpretation of this group of studies was the fact that all parameters of infection were not always clearly reported or thoroughly explored. Often infections were followed for only a very short time, which did not allow for potential differences in the timing of viral antigen (Ag) expression in different cell types. We have performed a thorough analysis of the permissiveness of NPCs and two different cell types we differentiated from them, astrocytes/astroglia and neurons. We have followed morphological changes after infection, studied protein expression of all classes of viral Ags, looked for proper viral Ag subcellular localization and replication center formation, and quantitated the release of functional virions from all three cell types. Importantly, in addition to the laboratory-adapted Towne strain, we have also performed these analyses with a clinical viral isolate (TR, for triple resistant) to determine whether differences occurred in the timing and extent of infection in the three cell types using this clinical strain. We found that in our system, all three cell types were fully permissive for HCMV infection, with varying kinetics of replication center formation, virion release, and ability to survive in the presence of the virus. We discuss our results and their ramifications in the context of prior studies.

MATERIALS AND METHODS

NPC culture.

NPC cultures SC27 and SC30 were obtained postmortem from the brains of premature neonates (estimated gestational age, 23 to 24 weeks) who died of natural causes unrelated to HCMV infection (P. H. Schwartz, National Human Neural Stem Cell Resource, Irvine, CA [33]). The basal medium (DGA) used for all tissue culture experiments described here was Dulbecco's modified Eagle's medium-F12 containing l-glutamine (2 mM Glutamax; Gibco/BRL) and the antibiotics penicillin/streptomycin (100 U/ml and 100 μg/ml), gentamicin (50 μg/ml), and amphotericin B (Fungizone; Gibco/BRL; 1.5 μg/ml). NPCs were cultured in growth medium (GM) (DGA supplemented with 10% BIT9500 [5 mg/ml bovine serum albumin, 5 μg/ml recombinant human insulin, 100 μg/ml human transferrin; Stem Cell Technologies], human basic fibroblast growth factor [Invitrogen; 20 ng/ml], and human epithelial growth factor [Invitrogen; 20 ng/ml]). NPCs were cultured as spheres by seeding cells into uncoated tissue culture dishes (Falcon). NPCs were cultured as monolayers by seeding them onto fibronectin-coated dishes (10 μg/ml fibronectin in Dulbecco's modified Eagle's medium-F12, coated overnight at 37°C and then removed and allowed to dry completely prior to being seeded). For NPC culturing, the medium was refreshed by half every other day. The cells were maintained at 37°C in a humidified atmosphere containing 5% CO₂.

NPC differentiation protocols.

NPCs were differentiated toward a glial phenotype by plating the cells onto uncoated dishes with simultaneous withdrawal of basic fibroblast growth factor and epithelial growth factor from the medium. Instead of GM, the cells were grown in DGA supplemented with 10% non-heat-inactivated fetal bovine serum for 21 days (unless otherwise noted). The cells were fed by a complete medium change twice per week. The media were collected from these cells between 7 and 21 days after the start of differentiation and combined for use as glial conditioned medium (GCM), a component of the neuronal differentiation medium (NDM) described below. GCM was first spun to clear cellular debris and then filtered and frozen at −20°C in aliquots for later use.

NPC differentiation toward a neuronal phenotype was accomplished by plating NPCs onto poly-d-lysine-coated plates (50 μg/ml in double-distilled water, coated at 37°C overnight and then allowed to dry, washed twice with distilled water, and allowed to dry again before being plated). One day after being plated, the cells were switched to NDM (1:1 DGA/GCM ratio supplemented with neurotrophin 3 [Chemicon; 20 ng/ml], recombinant human brain-derived neurotrophic factor [Chemicon; 20 ng/ml], and 1% non-heat-inactivated fetal bovine serum) for 21 days. The medium was refreshed by half every other day. Media were collected from these cells beginning 14 days after the start of differentiation for use as neuronal conditioned medium (NCM).

Enrichment for neurons.

After 21 days of differentiation toward the neuronal phenotype, cells were trypsinized and seeded into uncoated dishes in NDM. The majority (∼80%) of the cells derived from this differentiation protocol were astroglia. These cells quickly and easily settled down onto the uncoated dishes after a short incubation. The remaining 20% (which were neurons) attached only loosely to the uncoated dishes (i.e., differential adherence was displayed by the two populations). With the goal of generating an enriched neuron population, we developed the following protocol. Cells were allowed to settle for 2 h, and then (in a method akin to mitotic shake off) the dishes were firmly struck on their sides in order to dislodge the loosely attached cells. We then collected these cells from the supernatant with a low-speed spin, recounted them, and reseeded them onto poly-d-lysine-coated dishes or coverslips as necessary in NDM. In this way, we obtained 75 to 85% β-tubulin III⁺ (neuron) populations.

Virus strains and infection.

Two different strains of HCMV, the laboratory-adapted Towne (ATCC VR977) and the clinical isolate TR (generously provided by Jay Nelson, Oregon Health Sciences University [24]) were used in these studies. Both virus strains were propagated on human foreskin fibroblasts (HFFs) and maintained at low passage, as previously described (39). To obtain high-titer TR stocks, viral supernatants were pelleted using high-speed ultracentrifugation. All TR isolates were also tested to ensure maintenance of the clinical cassette of gene products as defined previously (24).

For NPC infections, monolayer cultures were lifted with cell dissociation buffer (Gibco), seeded onto poly-d-lysine-coated dishes or coverslips (1 × 10⁶ cells/100-mm dish), and allowed to settle overnight. The cells were then infected with either Towne or TR at a multiplicity of infection (MOI) of 3, allowing 2 h for adsorption. The inoculum was then removed, and the cells were refed. Cells were harvested at multiple time points as outlined below. Both NPC lines SC27 and SC30 were used in the initial rounds of infection to compare their kinetics. As they behaved very similarly, only the SC27 cells were used for further analysis by the differentiation protocols (note that glial differentiation was also performed once in SC30 cells with results almost identical to those presented here).

For astroglial infections, SC27 NPCs were differentiated down a glial pathway for 7 (day-7 astroglia) and 21 (SC27 astroglia) days. The cells were then trypsinized, counted, and reseeded onto uncoated dishes and/or coverslips (1 × 10⁶ cells/100-mm dish). The cells were infected as described above.

For mixed neuron/astroglial infections, SC27 NPCs were differentiated down a neuronal pathway for 21 days (SC27 mixed). SC27 mixed cells were then dissociated (cell dissociation buffer; Gibco), washed with Hanks buffered saline, counted, and reseeded onto poly-d-lysine-coated dishes or glass coverslips (1 × 10⁶ cells/100-mm dish). The cells were infected as described above.

SC27 enriched neurons (derived from the modified shake-off technique described above) were also infected with both strains. To assess long-term infection in these cells, the medium was completely removed at 10 days postinfection (p.i.), and the remaining cells were rinsed, trypsinized (0.125%), and reseeded onto poly-d-lysine-coated coverslips in a 12-well plate in fresh medium (a 1:1 NCM/NDM ratio). The coverslips were harvested at 14 days p.i.

Live-cell imaging.

Live-cell images were taken at the indicated times p.i. using a Nikon TMS-F microscope equipped with a Nikon Coolpix5400 digital camera. The images were representative of time points from at least two infection time courses. The magnification for all live-cell images presented was ×500.

Virus titration (cumulative yields).

Small aliquots (150 μl) of supernatant from infected cultures were collected at the indicated times p.i. and stored at −80°C (after the addition of 1% dimethyl sulfoxide) until all samples were collected. Supernatant aliquots were serially diluted prior to infection of a monolayer of HFFs, following standard plaque-forming assay protocols (4). Averages plus one standard deviations (SD) from at least two independent infections are shown.

Western blotting.

For Western time courses, cells were harvested at the indicated times p.i. by rinsing them with phosphate-buffered saline (PBS), trypsinizing them, and washing them again with ice-cold PBS. The cells were then resuspended in ice-cold PBS, counted, and spun into a pellet. The cell pellets were snap frozen in liquid nitrogen and stored at −80°C until the time course was completed. Cell lysates were prepared as described previously (20). Lysates derived from equivalent numbers of cells (1 ×10⁵ cells) were used for each sample, subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and then transferred to Protran (Schleicher & Schuell BioScience). Detection of cellular and viral proteins was as described previously (20). For each time course, multiple gels were run using appropriate acrylamide percentages to resolve the different-size viral Ags. Blots were stripped a maximum of two times using standard techniques. Due to the difficulty of growing large numbers of these cells, Western time courses were performed only once for each cell type assessed. As these assays were performed solely to assess the presence or absence of each class of viral Ag, they should be taken as approximations of the timing of protein expression.

Immunofluorescence (IF) analysis.

Viral Ags were detected as previously described (18). To detect the cellular proteins used as markers for NPCs, astroglia, and neurons, we slightly modified this fixation/permeabilization protocol. The permeabilization time in 1% Triton X-100 was extended from 5 min to 10 min to access cytoskeletal proteins more efficiently (19). After Ag detection, coverslips were mounted in glycerol containing paraphenylene diamine to inhibit photobleaching. Nuclei were counterstained with Hoechst dye. The images were obtained using a Nikon Eclipse E800 fluorescence microscope equipped with a Nikon DXM camera and Metavue software. Quantitation in the graphs represents the average plus 1 SD from at least two independent experiments. In each experiment, at least 100 cells were scored for each time point.

Abs used in the study.

All primary antibodies (Abs) used in this study (except for Sox2) were mouse monoclonal Abs (MAbs).

(i) IF analysis.

For viral Ag detection, anti-pp65 (immunoglobulin G1 [IgG1]) and anti-UL44 (IgG1) MAbs (Rumbaugh-Goodwin Institute for Cancer Research, Inc.) and anti-IE1 (IgG2a) (a kind gift from Bill Britt, University of Alabama, Birmingham) were used. For NPC identification, anti-nestin MAb (Chemicon) and anti-Sox2 (goat polyclonal; Santa Cruz Biotechnology) were used. For glial cells, anti-GFAP (IgG2b; Fitzgerald) MAb was used, and for neuronal cells, anti-β-tubulin III MAb (IgG2b; Sigma) was used. The secondary Abs used were tetramethyl rhodamine isothiocyanate-conjugated anti-mouse IgG1, IgG2a, and IgG2b (Southern Biotechnology); AlexaFluor 488-conjugated goat anti-mouse IgG1, IgG2a, and IgG2b (Molecular Probes); and fluorescein isothiocyanate-conjugated donkey anti-goat (Jackson Immunoresearch, Inc.).

(ii) Western blotting.

For viral Ag detection, anti-pp65, anti-UL44, and anti-IE1/2 (Ch16.0; IgG1) MAbs (Rumbaugh-Goodwin Institute for Cancer Research, Inc.) and anti-pp28 (IgG1), anti-MCP (IgG2a), anti-MnCP (IgG1), and anti-gB (IgG2b) MAbs (kind gifts from Bill Britt) were used. For cellular proteins, anti-GFAP, anti-β-tubulin III, and anti-actin (clone ACT05; Neomarkers) MAbs were used. The secondary Ab used was horseradish peroxidase-conjugated sheep anti-mouse (Amersham Bioscience).

RESULTS

NPCs showed the hallmarks of multipotent progenitor cells.

In the experiments described in this paper, we set out to fully characterize the parameters of infection for HCMV in NPCs and their neuronal and glial derivatives. The NPCs utilized in this study (SC27 and SC30 cells) were derived from the brains of two neonates born prematurely who subsequently died of natural causes unrelated to HCMV infection. Neural stem cells were harvested soon after death and propagated as previously described in detail by Schwartz and colleagues (33). These cells have been extensively characterized for their expression of neural stem cell markers (33). We chose the progenitor cell markers nestin and Sox2 for our studies to delineate this population of cells. As can be observed in Fig. 1A, all cells in the NPC cultures were nestin⁺ at the start of experiments with both SC27 (shown) and SC30 (not shown). What can also be readily observed in Fig. 1A is the typical bipolar and multipolar morphologies of the NPCs, with the neuritic processes extending from the cell bodies.

FIG. 1.

NPCs displayed stem cell markers and developed classic CPE during infection. (A) SC27 NPCs were seeded onto poly-d-lysine-coated coverslips. The cells were fixed and stained for the stem cell marker nestin as described in Materials and Methods (left). A phase-contrast image of the same field is shown on the right. Scale bar, 5 μm. (B) SC27 NPCs were seeded onto poly-d-lysine-coated dishes the day before being infected with Towne or TR at an MOI of 3, and live-cell images were taken at 24-h intervals beginning at 24 h p.i. Magnification for all live-cell images was ×500.

These NPCs could be propagated under various conditions; generally, when initially cultured, they were grown as neurospheres in uncoated dishes. However, for all of our studies, which involved the analysis of single cells for both morphological changes and expression of viral Ags, it was more desirable to grow the NPCs as monolayer cultures on either fibronectin (for general maintenance) or poly-d-lysine-coated dishes or coverslips (for experiments). The NPCs attached equally well to both substrates.

NPCs were fully permissive for HCMV infection.

The permissiveness of a cell refers to its ability to support the full viral life cycle, from entry at the cell surface to release of newly formed virions. Like those of other herpesviruses, the genome of HCMV is temporally expressed in the permissively infected cell. The immediate-early (IE) genes are transcribed after viral entry and rely mainly on host factors for their expression. Prior to viral DNA synthesis, the early (E) genes (mainly viral replication proteins) are expressed with the aid of one or more IE gene products. Finally, late (L) genes (mainly structural components) are transcribed after the initiation of viral-DNA replication (23).

To test the permissiveness of NPCs for HCMV infection, we plated them onto poly-d-lysine-coated plates and infected them at an MOI of 3 with either a laboratory-adapted strain (Towne) or a clinical isolate (TR). The two strains were used to determine if there were differences in their infection parameters. Since SC27 and SC30 NPCs were used for infections and gave very similar results, only data from SC27 NPCs are shown in Fig. 1 and 2 as representative images (unless otherwise noted). We first monitored morphological changes using live-cell imaging to assess the development of cytopathic effects (CPE) in these cells (Fig. 1B). These initial infections showed that development of CPE with the TR strain progressed more slowly in the NPCs. Development of CPE in the Towne infections began at a pace similar to that in the permissive HFFs we used in our previous studies (4), and all NPCs drew in their processes and rounded up by 24 h p.i. However, it was not until ∼96 h p.i. (compared to 48 h p.i. in HFFs) that the cells flattened and somewhat reextended their processes. Cell death began at 120 h p.i. in a small proportion of cells in these Towne infections. In general, the TR infections lagged approximately 24 h behind in all CPE milestones in the NPCs. Additionally, at 120 h p.i., when the cells had flattened back out, the predominant phenotype observed in the TR-infected NPCs was that of large multinucleated syncitia. Although sporadic small syncitia were observed at late times in the Towne-infected NPCs as well, they were the exception, not the norm. This difference may be a result of the more cell-associated phenotype of the TR strain compared to Towne. It should also be noted that in infections using either strain, the long processes present in mock-infected NPCs were absent once the cells flattened back out.

FIG. 2.

NPCs displayed proper localization of viral Ags and released infectious virions after infection. The NPCs were seeded onto poly-d-lysine-coated coverslips and were infected with Towne or TR at an MOI of 3 (or mock infected), and the coverslips were harvested at the indicated times p.i. Ab sources are listed in Materials and Methods. Scale bars, 5 μm. (A) After 24 h p.i., the majority of NPCs stained strongly within the nucleus for the viral Ag IE1 (left). Small UL44 foci within the nuclei began to develop by 24 h p.i. and progressed to bipolar foci and finally to one large focus as the infection proceeded (middle), as described previously (4, 28). Nuclei stained with Hoechst are shown on the right. Images of Towne-infected SC30 NPCs are shown. SC27 NPCs gave essentially identical results. (B) Quantitation of nuclear IE1 staining over the first 96 h p.i. in both Towne- and TR-infected SC27 NPCs. The bars represent the averages from two experiments plus 1 SD. (C) Graphic representation of the results of scoring for the presence and size of UL44 foci in two experiments for each virus strain in SC27 NPCs. The blue bars represent the percentages of cells with multiple small UL44 foci, the red bars represent the percentages of cells that had bipolar foci, and the yellow bars represent the percentages of cells with one large focus at the given time points. (D) IE1 and pp65 expression at different times p.i. As in panel A, the large majority of cells expressed IE1 by 24 h p.i., and essentially all cells showed staining for the tegument protein pp65 in their nuclei. By 72 h p.i., juxtanuclear localization of pp65 could be seen in many cells, and by 96 h p.i., viral particles containing pp65 could be observed in many cell processes. The boxed images are magnified 2.5 times to better show this localization. Phase-contrast images are shown to emphasize the dramatic morphology changes that occur during infection. Towne infection of SC30 NPCs is shown. Essentially identical results were obtained in SC27 NPCs. (E) Total cellular lysates from equivalent numbers of SC27 NPCs (1 × 10⁵) were separated by SDS-PAGE, transferred to Protran membranes, and probed with the indicated Abs. Representative protein profiles are shown for IE, E, and L proteins. See Materials and Methods for Ab specifics. An actin loading control was included to confirm equal sample loading. (F) The same number of SC27 NPCs (1 × 10⁶) were seeded and infected with either Towne or TR, and small aliquots of supernatant were collected at the indicated times p.i. Viral titers were determined by standard plaque-forming assay as described in Materials and Methods. The bars represent averages from two experiments plus 1 SD.

Since all the cells displayed the classic signs of CPE development, we followed them to determine the extent of expression of different temporal classes of viral Ags and, further, to look for the true test of full permissiveness, the production of infectious virions. We began by assessing the percentage of cells that expressed different classes of viral Ags over time. Figure 2A shows staining for IE1, one of the main viral transactivating proteins, and the E protein UL44, the viral processivity factor, which localizes to viral replication centers. Shown in Fig. 2A are NPCs infected with Towne. Similar results were obtained using TR. The large majority of cells infected with Towne or TR showed strong IE1 positivity in their nuclei by 24 h p.i. Quantitation of the percentage of IE1⁺ cells is shown in Fig. 2B. UL44 was used to assess the expression of E proteins and the development of replication centers in these cells. All stages of viral replication centers that have been previously described (28) were seen in the NPCs infected with either Towne (Fig. 2A) or TR (not shown). It was important to show that all stages of centers, from small multiple foci (which represent prereplication sites) to large single foci (which represent fully developed sites), were seen in these cells to establish the presence of active viral genome replication. However, as can be seen in Fig. 2C, the development of these replication centers lagged significantly behind during infections with the TR strain compared to Towne for the first 48 h. Quite obvious in the figure is the very low percentage of cells expressing viral replication foci at 24 and 48 h p.i. in the TR infections compared to Towne. Additionally, while larger centers (bipolar and single foci) began to be observed after 48 h p.i. in the Towne infections, they were essentially visible only after 72 h p.i. in the TR infections. Unlike the commonly used HFFs, which exhibit a markedly synchronous sequence of replication center development (4, 28), not all NPCs showed large single foci, but instead, a mix of sizes remained present even out to the relatively late time point of 96 h p.i. This could be reflective of the asynchrony of the initial infection, but more likely it is a consequence of slightly different permissiveness of these cells compared to HFFs.

The NPCs were then assayed for the localization of structural proteins by staining them for pp65, a major component of the viral tegument (Fig. 2D). Very early localization (4 h p.i.) of input pp65 revealed rapid transit of the protein to the nucleus in the very large majority of Towne-infected cells (data not shown), and by 24 h p.i., all nuclei were strongly positive for pp65 in both Towne (Fig. 2D, top) and TR (not shown) infections. This 24-h p.i. staining likely represents a combination of input and de novo-synthesized proteins and is similar to that observed in permissive fibroblasts at this time point (4). Between 48 and 72 h p.i., pp65 localization began to shift into the cytoplasm, and by 72 h p.i., the viral assembly complex could be clearly visualized as a bolus of pp65 in the cytoplasm close to the nucleus (Fig. 2D, middle). Movement of pp65 viral particles out into the ends of the dendrites could also be visualized by 96 h p.i., indicating potential production and release of infectious particles from these cells (Fig. 2D, bottom). It should be noted that staining in the cytoplasm for this tegument protein was not as pronounced in TR infections. Infection with clinical isolates commonly does not produce large excesses of structural proteins, as is seen with infections by laboratory-adapted strains (12).

Although we had observed several different viral Ags by IF, we also wanted to determine the approximate timing of the appearances of all classes of proteins in the NPCs by Western blotting. Gels were run using lysates made from equivalent numbers of cells loaded into each lane. Viral Ags representative of the three temporal classes, IE (IE1/2), E (UL44 and pp65), and L (MCP, MnCP, pp28, and gB), were examined (Fig. 2E). All temporal classes were expressed during Towne infections of both SC27 (Fig. 2E) and SC30 (not shown) NPCs. Consistent with our live-cell images and IF data, expression of several viral Ags analyzed was slightly delayed in TR infections (data not shown).

Since all classes of proteins were expressed and there were also indications of trafficking of viral particles from our IF analysis, we investigated whether infectious particles were being released from the NPCs by removal of small aliquots from the supernatant of infected cells at 24-h intervals for titer analysis. Figure 2F shows that viral particles were released in relatively large numbers from both Towne- and TR-infected SC27 NPCs beginning at ∼72 h p.i. Preliminary assays performed at earlier times p.i. (24 and 48 h) revealed substantially lower titers (∼10-fold) with both virus strains (data not shown), similar to our previous observations in permissive fibroblast cultures (4). Therefore, we concentrated our further analysis on later time points only. These experiments indicated that the differences between Towne and TR infections encountered in our earlier studies were manifested in substantially lower virus release from the cells infected with TR, with differences sometimes approaching 2 log units at later time points. This could not be explained by cell association of the TR virus, as sonication of the infected cells did not yield dramatically higher titers (data not shown). We also carried out titer assays of both Towne- and TR-infected SC30 NPCs and observed the same trends for both virus strains. The differences between the strains were slightly more pronounced in the SC27 NPCs. Further, these cells attached and spread somewhat better in culture. Therefore, we chose to continue our analyses primarily with these cells, since we felt the latter characteristics could potentially produce clearer images.

NPCs differentiated down a glial pathway showed high permissiveness for HCMV infection.

After establishing that the NPCs were fully permissive for HCMV infection, we sought to determine whether cells differentiated from these progenitors (down either a glial or a neuronal pathway) were also permissive. The literature on neurons' permissiveness has been conflicting on this very important question (6, 17, 22). Among the differences in methodology seen in the literature are the lengths of the differentiation protocols. We defined the length of time for differentiation in our earlier paper as 21 days (33); however, other investigators conducted their differentiation protocols for varying and shorter lengths of time, between 2 and 7 days (6, 22, 25, 26). We chose to first test the ability of our NPCs to fully differentiate in less than 21 days. We compared undifferentiated cells with cells differentiated for 7 and 21 days (down a glial pathway) for the localization of the important stem cell marker Sox2. The literature has clearly shown that in progenitor cells, Sox2 is located exclusively in the nucleus and that after full differentiation, the protein moves into the cytoplasm (8). IF analyses of Sox2 localization at 0, 7, and 21 days postdifferentiation are shown in Fig. 3A. Undifferentiated NPCs showed strong nuclear localization of Sox2 (>95% of cells) (Fig. 3A, top); however, ∼50% of the cells differentiated for only 7 days did, as well (Fig. 3A, middle). Cells differentiated for 21 days displayed movement of Sox2 into the cytoplasm in the large majority of cells, with only ∼12.5% still displaying nuclear localization (Fig. 3A, bottom). Cells at this time point also displayed the large, flattened, pleiomorphic morphology characteristic of astroglia (Fig. 3A, phase-contrast images). We therefore chose to differentiate cells for a full 21 days in all our subsequent experiments.

FIG. 3.

SC27 astroglia differentiated for 21 days showed strong GFAP staining and displayed characteristic CPE after infection. Scale bars, 5 μm. (A) Staining for Sox2 in SC27 NPCs that were undifferentiated (top row), differentiated for 7 days (middle row), or differentiated for 21 days (bottom row; SC27 astroglia). Staining of nuclei with Hoechst is shown in the middle of each row, and phase-contrast images are shown on the right to demonstrate the morphological change that occurred during full differentiation. (B) SC27 astroglia stained for GFAP showing that all cells were strongly positive for this glial marker out to the ends of their processes. (C) Live-cell imaging of the morphological changes associated with infection of SC27 astroglia with either Towne (middle) or TR (right). Mock-infected cells are shown on the left. SC27 astroglia were seeded onto uncoated dishes and infected with Towne or TR at an MOI of 3.

Cells differentiated for 21 days down a glial pathway gave strong cytoplasmic staining for the astrocytic marker GFAP, as can be observed in Fig. 3B. Essentially all cells (>95%) in the cultures differentiated in this fashion stained for GFAP, with particularly strong staining all the way into the ends of the processes, a hallmark of astrocytes (8, 13). We will refer to these cells as SC27 astroglia. When SC27 astroglia were infected with either Towne or TR, typical CPE was observed, although, as can be seen in Fig. 3C, the TR infections once again lagged slightly behind the Towne infections by ∼24 h. As we had noticed in the NPC infections at late times p.i., these cells remained contracted and showed little sign of reextending their processes. Cell death began in the Towne infections at ∼120 h p.i., and the TR-infected cells were ∼24 h behind.

The same set of parameters examined in the NPCs was analyzed in the SC27 astroglia (Fig. 4). As can be seen in Fig. 4A, the SC27 astroglia were permissive and very quickly began expressing viral Ags. By 24 h p.i., essentially all cells were IE1⁺. Quantitation of the protein's expression in both Towne and TR infections of SC27 astroglia showed essentially no differences in initiation of gene expression in the cells.

FIG. 4.

SC27 astroglia were somewhat more permissive than SC27 NPCs. SC27 astroglia were seeded onto uncoated dishes and infected with Towne and TR at an MOI of 3. Scale bars, 5 μm. (A) IE1 protein expression was strong in the large majority of cells by 24 h p.i. in both Towne (left) and TR infections. Quantitation of IE1 staining at different times p.i. is shown on the right. (B) UL44 focus formation and development in SC27 astroglia. On the left are shown representative images of SC27 astroglia infected with Towne and displaying multiple small, bipolar or single large foci with increasing time p.i. Note also that the GFAP staining, although remaining cytoplasmic, became less distinct than that seen in mock-infected cells (Fig. 3B). Scoring of SC27 astroglia for development of different-size foci in both Towne and TR infections is shown on the right. (C) Western blots of viral IE (IE1/IE2), E (UL44), and L (pp28, gB, and MnCP) protein expression in Towne- and TR-infected SC27 astroglia. Equal amounts of cell lysates were applied to all lanes in all blots. GFAP levels were not affected by viral infection. An actin loading control was included to confirm equal sample loading. (D) Viral titers from Towne- and TR-infected SC27 astroglia at the indicated times p.i. The bars represent the averages of two experiments plus 1 SD.

Figure 4B demonstrates that UL44 foci could be seen beginning at about 24 h p.i. in the SC27 astroglia. The progression of these foci to the typical bipolar and single-large-focus phenotypes was also observed. Note that all cells were still GFAP⁺ in the cytoplasm, although the morphology of the cells had markedly changed (no processes extended from the cells after infection). TR-infected cells again lagged behind in development of these foci in SC27 astroglia, as had been observed in the NPCs, although more cells showed foci in TR-infected SC27 astroglia at 48 h p.i. than were seen in TR infections of their parental SC27 NPCs.

When the expression patterns for viral proteins were examined, we saw that all three classes of proteins were expressed in the SC27 astroglia (Fig. 4C). As expected from the delay in replication focus development observed in Fig. 4B, L protein expression was delayed by ∼24 h in the TR infections of SC27 astroglia compared to Towne. Interestingly, GFAP levels were not affected by infection of the SC27 astroglia with either strain of virus.

Titer data in the SC27 astroglia (Fig. 4D) showed that TR infections again produced markedly fewer virions than the Towne-infected cells at all time points tested. Over the first 96 h, titers were very similar in NPCs and astroglia infected with either strain. However, at later time points (120 to 168 h p.i.), the astroglial titers were much higher in both strains, indicating higher production/output in these astroglial derivatives (compare Fig. 2F and 4D). The literature has suggested that astroglia are very permissive for infection (6, 16, 17, 22), and this was certainly borne out in our results.

Differentiation down a neuronal pathway generated a mixed population of cells, which were permissive for HCMV infection.

When the SC27 NPCs were differentiated for 21 days using trophic factors that promote the survival of neurons (neurotrophin 3 and recombinant human brain-derived neurotrophic factor), a mixed population developed (Fig. 5A shows a live-cell image). Using lineage-specific Abs to define this population, we found that approximately 15 to 25% of the cells had neuronal morphology and stained strongly with the neural-specific markers β-tubulin III (Fig. 5B) and MAP2A/2B (not shown). The other 75 to 85% were GFAP⁺, indicating their astroglial descent. The more “tenuous” attachment of the neurons than of the astroglia (described in Materials and Methods) can be observed by examining Fig. 5A and B. The neurons appear more phase bright in the live image and appear to be on top of the astroglia.

FIG. 5.

Differentiation of SC27 NPCs down a neuronal pathway produced a mixed population of cells, some of which showed somewhat delayed development of CPE. (A) SC27 NPCs were seeded into poly-d-lysine-coated dishes and differentiated down a neuronal pathway as described in Materials and Methods. A live-cell image 21 days after the start of differentiation shows the mixed morphology of this population. (B) IF staining for β-tubulin III shows the presence of neurons (∼15 to 25%) in this SC27 mixed population. The characteristic morphology of these cells is demonstrated in the phase-contrast image of the field. Scale bar, 20 μm. (C) Live-cell images of SC27 mixed populations infected with either Towne or TR at an MOI of 3 the day after they were seeded onto poly-d-lysine plates. Note how the neurons (cells with morphologies demonstrated in panel B) within the population appeared more refractory to infection in both the Towne and TR infections, with the latter population lagging further behind in development of CPE.

When live-cell imaging was used to examine the morphological changes of this mixed neuron population (which we have termed SC27 mixed) (Fig. 5C), the TR infections lagged behind in CPE development by ∼24 h but caught up by 48 h p.i. (not shown). However, in both infections the neurons appeared to be a bit more refractory to the effects of the virus than the astroglia and did not round up as quickly (not until 72 h p.i. were all cells rounded), even though they appeared more loosely attached to the substrate (see below for more discussion).

The SC27 mixed cells demonstrated CPE more slowly, which led us to assess whether viral Ag expression was delayed as well. IE1 expression was still relatively early in the SC27 mixed populations (Fig. 6A). By 48 h p.i., >90% IE⁺ cells were observed in both Towne and TR infections, implying that the neurons were also susceptible to infection. We tested this assumption by assaying for UL44 focus formation. In Fig. 6B, we show that β-tubulin III⁺ neurons were capable of supporting all three levels of UL44 focus development. However, development of foci in the infected (both strains) SC27 mixed populations was delayed until ∼48 h p.i. Not all cells developed foci during the 96-h time course, indicating that there might have been a subset of cells in this mixed population that did not express E viral Ags or was very slow to do so. Viral Ag expression was checked by Western blot analysis in the SC27 mixed populations. All classes of viral proteins were expressed, with similar profiles observed in both Towne and TR infections (Fig. 6C). Interestingly, the level of β-tubulin III was not affected by viral infection. Viral titers were assessed in the SC27 mixed populations (Fig. 6D); we found virtually no differences compared to infected SC27 astroglia. The profiles looked very similar, and the peak titers were almost identical (compare Fig. 4D to 6D). Following the trend we had already observed, the TR infections produced lower levels of virus in these SC27 mixed populations at all time points assessed.

FIG. 6.

SC27 mixed populations were fully permissive for HCMV infection. SC27 mixed cells were seeded onto poly-d-lysine-coated plates or coverslips the night before infection at an MOI of 3. Scale bars, 5 μm. (A) IE1 protein expression was strong in the large majority of cells by 24 h p.i. in both Towne (left) and TR (not shown) infections. Quantitation of IE1 expression is shown on the right. (B) UL44 focus formation and development in SC27 mixed populations. On the left are shown representative images of SC27 mixed cells infected with Towne, displaying multiple small, bipolar or single large foci with increasing time p.i. These images show that β-tubulin III⁺ cells also displayed UL44 foci. Scoring of SC27 mixed populations for development of different-size foci in both Towne and TR infections is shown on the right. (C) Western blots of viral IE (IE1/IE2), E (UL44), and L (pp28, gB, and MnCP) protein expression in Towne- and TR-infected SC27 mixed populations. Lysates from equal numbers of cells were applied to all lanes in all blots. β-Tubulin III levels did not vary with infection. An actin loading control was included to confirm equal sample loading. M, mock-infected cells. (D) Viral titers from Towne- and TR-infected SC27 mixed populations at the indicated times p.i. The bars represent the averages of two experiments plus 1 SD.

NPCs differentiated down a neuronal pathway could be enriched to ∼80 to 85% β-tubulin III⁺ neurons.

Although we had evidence that β-tubulin III⁺ cells were permissive for viral infection, the parameters of this infection could not be studied efficiently in such a mixed population. We needed to be able to differentiate between the neurons and their support astroglia and, if possible, to separate them. We took advantage of the differential adherences of these two cell types (astroglia attach easily without extra coating, while neurons take longer and need coating of the substrate to adhere). We performed a modified “mitotic-shake-off” procedure (see Materials and Methods for a description) and were able to enrich the population to ∼80 to 85% β-tubulin III⁺ neurons after this process (Fig. 7A). We then carried out viral Ag expression analyses and assayed for release of functional virus from this more homogeneous neuron population (termed SC27 enriched).

FIG. 7.

Employing a modified mitotic-shake-off technique greatly enriched the population of β-tubulin III⁺ cells. (A) SC27 NPCs differentiated for 21 days down a neuronal pathway were subjected to a modified mitotic shake off as described in Materials and Methods. IF staining for β-tubulin III showed these cells to be greatly enriched for neurons (∼80%). Scale bar, 5 μm. (B) SC27 enriched neurons were seeded to poly-d-lysine-coated plates and infected with either Towne or TR at an MOI of 3. Live-cell images are shown at different times p.i. for TR-infected SC27 enriched neurons.

Figure 7B shows live images of morphology changes after infection with TR virus. The Towne and TR infections showed virtually identical characteristics, in contrast to what we observed for NPCs or astroglia. Most notably, the enriched neurons were much more refractory to the effects of infection and showed little CPE for the first several days. Between 4 and 7 days p.i., the cells became swollen and their processes were reduced in length (see the 7-day p.i. image in Fig. 7B). We find it interesting that the infections of SC27 enriched neurons generally lasted longer as well, with 25 to 30% of the cells alive even after 10 days p.i. (when infected NPCs and astroglia were all dead). Small numbers of syncitia (many fewer than in the NPC infections) could be observed in both Towne- and TR-infected cells. As noted previously, syncitia were smaller and less frequent in the Towne-infected cells (than in TR-infected cells).

Single-cell and total-population viral Ag expression was checked in these SC27 enriched neurons to assess any differences from the SC27 mixed population. As depicted in Fig. 8, SC27-enriched neurons showed patterns of IE, E, and L gene expression very similar to those of their SC27 mixed counterparts. The most notable difference was in UL44 focus development, with the TR infections lagging behind the Towne infections more noticeably at 48 h p.i. in the SC27 enriched neurons (Fig. 8B). Additionally, focus formation did increase between 48 and 72 h p.i. in the TR-infected cells, but no further increases in foci were observed out to 96 h p.i. SC27 mixed and SC27 enriched lysates, loaded onto the same gel to emphasize the enrichment of β-tubulin III, showed virtually identical viral Ag expression profiles (Fig. 8C).

FIG. 8.

SC27 enriched neurons were fully permissive for HCMV infection but produced fewer infectious virions than the SC27 mixed population when infected with Towne virus. (A) Quantitation of IE1⁺ cells in SC27 enriched neurons infected with Towne or TR. (B) Quantitation of UL44 foci in SC27 enriched neurons showed a lag in the development of UL44 foci, with TR infections displaying very low levels of foci until 72 h p.i. (C) Western blot analysis of viral Ag expression in Towne-infected SC27 mixed and SC27 enriched populations. β-Tubulin III levels were higher in the enriched population and were not affected by HCMV infection. An actin loading control was included to confirm equal sample loading. (D) Release of infectious virions from SC27 enriched cells compared to SC27 mixed cells (as shown previously in Fig. 6). Note that while the virus output from Towne-infected populations differed by as much as 10-fold in the mixed versus enriched cells, the TR-infected cells released almost equivalent amounts of virus. (E) Fourteen days p.i., there were still cells surviving in both Towne-infected (shown) and TR-infected SC27 enriched populations. These surviving cells were still positive for both IE1 and UL44 and were all β-tubulin III⁺. Scale bar, 5 μm. The error bars represent one standard deviation.

The titer data (Fig. 8D) are shown with the SC27 mixed and SC27 enriched populations side by side for comparison. This comparison illustrates several important points. First, the TR infections did not produce as much virus as the Towne at any given time point. In the Towne infections, particularly at later time points, the SC27 mixed populations consistently yielded at least 10-fold more virus than the SC27 enriched populations. However, in the TR infections, SC27 mixed and SC27 enriched populations yielded essentially equivalent amounts of virus at all time points tested (out to 168 h p.i./7 days p.i.). In addition, the SC27 enriched cultures continued to yield high levels of virus for at least an additional 3 days without a drop in titers (data not shown). This indicated the long duration of infection in the enriched neuron populations.

Noteworthy in the SC27 enriched neuron infections was the survival of ∼5 to 10% of these cells to at least 14 days p.i. (21 days p.i. for TR-infected cells). These cells expressed viral IE and E Ags (Fig. 8E) and continued to shed very low levels of virus (not shown). These cells were all β-tubulin III⁺, and many showed the extended morphology of the original, uninfected neurons of these cultures (Fig. 8E). Future experiments will further study these populations to determine their continuing survival and characteristics in the presence of low-level infection.

DISCUSSION

We have shown that NPCs isolated from neonatal tissue are fully permissive for both laboratory-adapted and clinical isolates of HCMV. This result is in general agreement with several other studies that have isolated NPCs from fetal/abortus tissue (6, 22, 25, 26). Our results are important, as they show that NPCs remain susceptible to infection even as they age developmentally and suggest that this population of cells may be susceptible even in the adult brain. In general, infections of NPCs using the clinical TR isolate progressed more slowly. However, all classes of viral Ags were expressed, and infectious virions were released from the TR-infected NPCs, albeit at a significantly lower level. Interestingly, the predominant phenotype observed during infection of NPCs with TR was large multinucleate syncitia. Although also observed during infection of NPCs with the laboratory-adapted Towne strain, syncitia were much smaller and less frequent in these infections. Scattered syncitia were also observed during enriched neuron infections with both strains. The difference in numbers and size of these syncitia could be due to the more cell-associated nature of the clinical isolate, but it is intriguing that the progenitors showed a more pronounced phenotype. Two other studies mentioned the formation of large multinucleate cells. One used directly excised brain aggregates, not progenitor cells (31), and reported that only microglia showed viral Ag positivity. The other (22) infected NPCs as they were differentiating and reported that neurons, not their progenitors, formed syncitia. Differences between our study and these may be accounted for by their use of different viral isolates (these studies both used the laboratory-adapted strain AD169) and/or lower-MOI infections.

We have also been able to differentiate our NPCs down two different pathways. Using established techniques (33), we were able to isolate populations of virtually pure astroglia/astrocytes (GFAP⁺) and a mixed population of astroglia (75 to 85%) and neurons (β-tubulin III⁺; 15 to 25%). Taking advantage of differential adherence, we could enrich the latter population so that we obtained 75 to 85% neurons. Using the same criterion we used for the NPCs, we have shown clearly that both astroglia and neurons are also fully permissive for HCMV infection. Our results for astroglia agree with previously published results (6, 16, 17, 22). Our results with neurons agree with some authors (22), but not with others (6, 17). We provide reasons for this discrepancy below.

In the course of studying these differentiated populations, we discovered several important issues. First, we found that differentiation of NPCs in culture for less than 21 days did not lead to sufficient changes in an important stem cell marker's localization to be able to delineate a clear difference from progenitor to differentiated cells. Namely, analysis of the localization of the neural stem cell-associated transcription factor Sox2 in NPCs differentiated down a glial pathway showed only partial (∼50%) movement out of the cytoplasm after 7 days but almost complete (∼87%) change in localization after 21 days. This may have played a role in the results from several previous studies (6, 25, 26). Second, we found that infections using the clinical isolate TR progressed more slowly than with the laboratory-adapted strain, even when infections were performed at a relatively high MOI of 3. These differences were most pronounced in the enriched neurons. Other studies have utilized different clinical isolates (16, 25, 26, 31), and although some differences were noted, they were not characterized as extensively as we have done here. Understanding the differences in HCMV interaction with these neural cell types is crucial when trying to extrapolate information derived from tissue culture to circumstances during congenital infection. Third, although all three cell types were fully permissive for the virus in our hands, in some cases the time scale of infection was quite protracted. This was most noticeable in neuronal populations, where our results differed most from previous studies.

In general, events occurred more slowly in the neurons. Neurons appeared to be quite refractory to morphological changes that normally occur during permissive infection. They took much longer to display the typical CPE of cell rounding and persisted with extended processes for a longer time. The neurons also took longer to develop replication centers than did astroglial derivatives. Although we cannot definitively say that the β-tubulin III⁺ neurons in our enriched cultures shed large numbers of infectious virions, the fact that (at least in the TR-infected populations) the amount of virus shed did not vary after enrichment indicated that the neurons were responsible for the release of a proportion of the virus shed. We cannot discount the differences in developmental age between our cells (neonatal, but well before term) and those used in previous studies (isolated from fetal tissue of a few weeks earlier estimated gestational age then our NPCs), but the most likely reasons that some authors have not observed neurons to be permissive for infection (6, 17) are that they infected at a lower MOI and, more importantly, they did not follow their infections for as long as we found necessary to observe the full characteristics of permissiveness of these cells.

Also of interest was a subset of neurons that had the capability for long-term survival in the presence of the virus. These cells continued to express viral Ags and to release small quantities of virus into the medium. Many of these surviving neurons had rather long processes compared to the large majority of the population, which had contracted processes. There are similarities in these results to those obtained by Tsutsui's group working in the MCMV model. Low-dose injections of MCMV during either day 8.5 or 15.5 of gestation produced Ag⁺ neurons in the cortex and hippocampus out to 30 days p.i. (41). Also, in later studies, these authors saw the presence of low-level-infected neurons within the cortex, which did not attract NK cell or macrophage clearance and thus escaped immune system detection (14). Either of these scenarios could be extended to the neurons that we saw surviving for relatively long periods while expressing viral Ags.

In vivo, CMV latent/persistent infection is principally found in white blood cells, the genital tract, salivary glands, etc. (reviewed in reference 32). Latently/persistently CMV-infected cells may serve as a reservoir for infecting neighboring cells, tissues, or organs. In the isolated compartment of the brain, where most cell types are susceptible to CMV (at least to some degree), such a virus reservoir could be very harmful, especially during fetal development or during compromises of the immune system, such as by stress, organ transplantation, or human immunodeficiency virus infection. Future studies will determine the characteristics of the long-term-surviving neurons and the persistence of viral Ag expression within them. Planned studies will also seek to determine whether long-term viral infection influences the spectrum of cellular proteins expressed by these cells. The fact that these long-term-surviving neurons can exist in the absence of their support astroglia implies potential changes in cellular gene expression in these cells. Although differences may exist between infections performed in vitro and those performed in vivo, the potential of these clinically significant cells to act as a reservoir for viral infection inside the brain is certainly intriguing.