The MatTek gingival tissue model (EpiGingival) contains normal human oral keratinocytes cultured in serum-free medium to form three-dimensional differentiated tissues. Hematoxylin and eosin staining of tissue cross-sections indicates that the cultured tissue shows an architecture very similar to human gingival mucosa in vivo (Figure 1, see Figure 4A) 22. The cultured tissue is 10–20 cell layers thick and consists of a cornified apical surface and a non-cornified basal region (Figure 1). The thickness and morphology of the apical stratum corneum and the basal cell layers are similar to those in the gingival tissues in vivo. As observed in vivo, cells at the basal region of the cultured tissue continue to divide and differentiate, and apical surface cells continue to cornify to form the stratum corneum. Furthermore, immunohistochemical staining indicates that distributions of different cytokeratins (e.g. K13 and K14) in cultured tissues are like those found in vivo 2233 (data not shown). Thus, the cultured tissue exhibits characteristics in structure (thickness, morphology, and organization), cell type and differentiation, and protein expression and composition as observed in vivo, and can be a model representing the oral tissue 22.

To determine whether the cultured tissues are permissive to HCMV infection and replication, two different HCMV strains (Towne and Toledo) and a mutant (Towne_BAC), were used in our initial experiments. Towne is a laboratory-adopted strain that has been passaged many times in vitro in human fibroblasts; whereas Toledo is an HCMV clinical isolate passaged in limited numbers in vitro 3435. Towne_BAC was derived from Towne by inserting a bacterial artificial chromosome (BAC) sequence into the viral genome and replacing the dispensable, 10 kb US1-US12 region 36. The Towne_BAC DNA, while maintained as a BAC-based plasmid in E. coli, produces infectious progeny in human fibroblasts and retains a wild type-like growth characteristic in vitro (Figure 2A) 2336. Each of these viruses was used to infect the tissues by inoculating at the apical surface with 2 × 10⁴ PFU. The infection through the apical surface serves as a model for HCMV infection via gingival mucosa surface. The infection was carried out for 10 days. We observed that the structure of the tissue remained intact up to 10 days in culture and started to disintegrate after 12 days incubation (data not shown). At different time points post infection, the tissues were harvested and the titers of the viruses were determined. The viral strains were able to grow in the tissues since viral titers increased by at least 300-fold during a 10 day infection period (Figure 2B). Thus, the gingival tissues support active HCMV lytic replication. No differences in growth among these viruses were found, suggesting that the lab-adopted Towne strain and its derivative, Towne_BAC, grow as well as the clinical low-passaged Toledo strain. In subsequent experiments, Towne_BAC was used as an HCMV representative to study viral infection in the gingival tissues. This mutant contains the gene coding for green fluorescence protein (GFP) and therefore, infection can be easily monitored in the tissues by detecting GFP expression 2336.

HCMV oral transmission begins when the virus enters the mucosal (apical) surface of oral tissues (e.g. gingival tissues), replicates in the surface cell layers, and spreads to neighboring cells and tissues in the basal regions 17. To determine whether HCMV infection of the MatTek gingival tissues can be a model for viral infection in vivo, two sets of experiments were carried out. First, Western analysis was used to determine whether viral lytic proteins were expressed, as observed in productive HCMV infection in vivo. Tissues were infected with 2 × 10⁴ PFU of either HCMV Toledo, Towne, or Towne_BAC strains. Protein extracts were isolated from tissues that were either mock-infected or infected with HCMV at 6 days post infection. Viral proteins were separated electrophoretically in SDS-polyacrylamide gels and electrically transferred to identical membranes. One of the membranes was stained with monoclonal antibody against human actin (anti-actin) (Figure 3D) and the other membranes were stained with monoclonal antibodies against viral IE1, UL44, and UL99 proteins (Figure 3A–C). The expression of actin serves as an internal control for the quantitation of HCMV protein expression in the tissues. IE1 is a viral immediate-early (α) protein, while UL44 and UL99 encode viral early (β) and late (γ) proteins, respectively 2. These proteins serve as the representatives for the expression of viral α, β, and γ genes. As shown in Figure 3, IE1, UL44 and UL99 were expressed in infected tissues. Combined with the growth analysis (Figure 2), these results indicate that the cultured tissues are permissive to HCMV infection and can support viral lytic gene expression and replication.

In the second set of experiments, infection of these tissues was studied using both conventional histological and fluorescent microscopy. Two different staining methods were employed. First, tissues were stained with hematoxylin and eosin in order to examine their structures. Second, since Towne_BAC contains a GFP expression cassette 36, fluorescent microscopy was used to detect GFP expression and to visualize infected cells.

As shown in Figure 4, mock-infected tissues maintained the characteristic gingival mucosal structure during the infection period. In these tissues, the cells at the basal surface continue to divide while those at the apical surface differentiate and cornify, forming a characteristic stratum corneum (Figure 4A). In the tissues that were infected through the apical surface, GFP staining was found in the cells near the apical surface, suggesting that the apical cells were infected with HCMV (Figure 4C–F). Compared to mock-infected tissues, the thickness of the stratum corneum in the infected tissues was significantly reduced (Figure 4B), possibly because the active replication of HCMV in apical cells induces cellular lysis and disrupts cellular differentiation and generation of the stratum corneum. Active HCMV replication in the apical surface has been observed in vivo and is associated with reduced thickness and destruction of the oral epithelial surface 1911. Thus, our results suggest that HCMV infection of cultured gingival tissues via the apical surface corresponds to its pathogenesis in vivo.

The ability of HCMV to infect and replicate in cells of the oral cavity is responsible for its pathogenesis in the oral mucosa, including viral-associated gingivitis and oral lesions. However, little is currently known about the mechanism of how HCMV is able to infect and replicate in oral tissues. Equally elusive is the identity of viral determinants responsible for oral infection. Specifically, it is unknown whether HCMV encodes specific genes responsible for its infection in the gingival mucosa. Through the use of a BAC-based mutagenesis approach, we have recently generated a library of HCMV mutants containing deletions in each open reading frame (ORF) 23. If a viral ORF is essential for viral infection in the oral tissue, the corresponding mutant with the deletion of the ORF is expected to be deficient in infecting and replicating in the tissue. Using the gingival tissue as the model, several experiments were performed to determine whether viral mutants that are attenuated in growth in the oral mucosa can be identified.

A collection of eight different mutants was used in our initial screen (Table 1). Each mutant was derived from Towne_BAC and contains a deletion in ORF UL13, UL24, UL25, UL108, US18, US20, US29, or RL9, respectively 23. In these mutants, the deleted ORF sequence was replaced with a kanamycin-resistance gene (KAN) expression cassette, which provides antibiotic resistance for rapid selection and isolation of the bacteria carrying the mutated Towne_BAC sequence. All mutants grew as well as the parental Towne_BAC in primary human foreskin fibroblasts (HFFs), suggesting that these ORFs are not essential for viral replication in vitro in cultured fibroblasts (Table I and Figure 5A). The functions of many of these deleted ORFs are currently unknown. However, they are present in all HCMV strains whose sequences have been determined 1415233738. Hence, these genes may play an important role in HCMV infection in vivo, such as in viral transmission and infection in the oral cavity.

To determine whether any of these HCMV mutants are deficient in growth and infection in cultured gingival tissues, the tissues were infected via the apical mucosal surface with each viral mutant at an inoculum of 2 × 10⁴ PFU. Infected tissues were harvested at 10 days post infection and viral titers in the tissues were determined. The titers of mutant ΔUS18 and ΔUL13 at 10 days post infection were approximately 100- and 10- folds lower than those of the parental Towne_BAC, respectively, while other mutants, such as ΔUL24 and ΔRL9, replicated as well as the parental virus (Table I and Figure 5B). Thus, mutants ΔUL13 and ΔUS18 appeared to be deficient in infecting the tissues via the apical surface. Both ΔUL13 and ΔUS18 were derived from the parental Towne_BAC by replacing the UL13 and US18 ORFs, respectively, with a DNA sequence (KAN) that confers antibiotic resistance to kanamycin in E. coli 23. Because ΔRL9 replicates as well as the parental Towne_BAC (Figure 5), the presence of the KAN cassette in the viral genome per se does not significantly affect the ability of the virus to grow in the tissues. Thus, these results suggest that the growth defect of ΔUS18 may be due to the deletion of the US18 ORF.

Two series of experiments were further carried out to study how ΔUS18 is defective in growth in the cultured tissues. First, viral infection in the tissues was studied by examining hematoxylin and eosin-stained tissues and visualizing GFP expression in infected cells. At 7 days post infection, the structure of the apical region in the ΔUS18-infected tissues was similar to that of uninfected tissues, and the thickness of the stratum corneum was not reduced as observed in the Towne_BAC-infected tissues (Figure 4G–H). Little GFP staining was found in the ΔUS18-infected tissues (Figure 4H) while substantial levels of GFP staining were detected in tissues infected with ΔRL9 and Towne_BAC (Figure 4E–F, data not shown). These observations support the growth analysis results (Figure 5) and show that ΔUS18 is deficient in infection and replication in gingival tissues. Second, Western analyses were used to examine the expression of viral proteins. As shown in Figure 6, at 72 hours post infection, the expression levels of IE1, UL44, and UL99 in ΔUS18-infected tissues were minimal and significantly lower than those in Towne_BAC-infected tissues. Thus, the infection of ΔUS18 appeared to be blocked prior to or at viral immediate-early gene expression, probably during viral entry, decoating, or transporting the capsid to the nuclei. Because similar levels of these proteins were found in tissues that were infected with ΔRL9 and Towne_BAC (Figure 6), the presence of the KAN cassette in the viral genome (e.g. ΔRL9) per se does not significantly affect viral protein expression in the tissues. These observations suggest that the defect in protein expression of ΔUS18 may be due to the deletion of the US18 ORF.

One of our objectives is to establish an in vitro cultured tissue model to screen antiviral compounds and determine their potency in inhibiting HCMV growth and replication in human oral tissue. To determine the feasibility of using the gingival tissue for antiviral compound screening and testing, two sets of experiments were carried out using ganciclovir, which functions as a nucleoside analog and is effective in treating HCMV infection in vivo by blocking viral DNA replication 3132. In the first set of experiment, oral tissues were treated with different concentrations of ganciclovir for 4 hours prior to viral infection. In the second set of experiments, tissues were infected with Towne_BAC for 24 hours and then treated with different concentrations of ganciclovir. The tissues were harvested at different time points post infection and the growth of HCMV was assayed by determining the viral titers. Treatment of ganciclovir reduced the growth of HCMV in HFFs (Figure 7A) 3132. Significant inhibition of HCMV growth was also observed in the gingival tissues when ganciclovir was added 24 hours after viral infection (Figure 7B). Similar levels of inhibition of viral growth in the tissues were found when the tissues were incubated with the drug before viral infection (data not shown). Previous studies have shown that treatment of ganciclovir blocks HCMV infection in cultured fibroblasts regardless whether the drug was added before or 24 hours after viral infection 3132. These results strongly suggest that cultured gingival tissues can be a suitable model for screening and testing antiviral compounds for inhibiting HCMV growth and replication.

Primary human foreskin fibroblasts (HFFs) (CC-2509) from Clonetics (San Diego, CA) were cultured in a humidified incubator at 37°C and in the presence of 5% CO₂. Cells were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% (vol/vol) fetal bovine serum (GIBCO/BRL), 1% (vol/vol) penicillin-streptomycin (GIBCO/BRL), and 0.2% (vol/vol) fungizone amphotericin B (GIBCO/BRL) 23. The HCMV Towne strain was obtained from the American Type Culture Collection (ATCC, Rockville, MD). The Toledo strain was a gift from Dr. Edward Mocarski (Stanford University) 1635. Towne_BAC and all the mutant viruses used in this study have been described previously 2336 and were propagated in HFFs.

Human gingival tissues (EpiGingival), obtained from MatTek Co (Ashland, MA), are living reconstructed oral epithelial tissues of 10–20 layers of cells that are derived from human primary oral keratinocytes and allowed to differentiate to a structure characteristic to that in vivo 22. The tissues arrived in Millipore Millicell CM culture insert wells and were approximately 0.1 mm thick and 9 mm in diameter. After overnight refrigeration (4°C, manufacturer's recommendations), the tissues were equilibrated by transferring them to 6 well plates containing 5 ml of assay media (MatTek Co.) per well and incubated at 37°C and 5% CO₂ for 1 hour. A small volume of 2 × 10⁴ PFU HCMV (0.1~0.2 ml) was then directly added to the apical surface of the tissues. After incubation with the viral inoculum at 37°C and 5% CO₂ for 4 hours, the tissues were washed to remove the inoculum. The tissues were replenished with fresh serum-free media containing growth factors every 48 hours. At different time points post infection, the tissues were collected and processed for determination of viral titers and for histochemical and fluorescent microscopy analysis.

The tissues were suspended in a small volume of 10% skim milk, followed by sonication. The tissue homogenates were titered for viral growth on HFFs in 6-well tissue culture plates (Corning Inc., Corning, NY) 23. Cells were inoculated with 1 ml of the sonicated tissues in 10-fold serial dilutions. After two hours of incubation at 37°C and 5% CO₂, cells were washed with complete media, overlaid with fresh complete medium containing 1% agarose, and cultured for 7–10 days. Plaques were counted under an inverted microscope. Each sample was titered in triplicate and viral titers were recorded as PFU/ml of tissue homogenates. The limit of virus detection in the tissue homogenates was 10 PFU/ml of the sonicated mixture. Those samples that were negative at a 10^-1 dilution were designated a titer value of 10 (10¹) PFU/ml.

Human oral tissues were fixed in Streck Tissue Fixative (Streck Laboratories, La Vista, NE) and then placed in 30% sucrose overnight. To prepare for cryostat sectioning, tissues were embedded in Histo Prep (Fisher Scientific, Fair Lawn, NJ) and frozen in 2-methylbutane submerged in liquid nitrogen. Tissues were cross-sectioned at 9 μm using a LEICA cryostat LC1900 sectioner, placed on Superfrost Plus microscopic slides (Fisher Scientific, Pittsburgh, PA), air-dried at room temperature, and frozen at -80°C until further use.

In the experiments using hematoxylin and eosin staining, the tissue slides were rehydrated in ethanol baths, immersed in Gill's Hematoxylin 3 and 1% eosin Y (Fisher Scientific, Fair Lawn, NJ), and then dehydrated in ethanol. Slides were mounted in permanent media and examined using a Nikon TE300 microscope with a SPOT camera attached (Diagnostic Instruments, Inc., Detroit, MI). For experiments using fluorescence staining, the tissue slides were permeabilized with 1:1 acetone:methanol and blocked with 0.1% BSA. For direct visualization of GFP staining, the slides were counterstained with DAPI (Molecular Probes, Portland, OR) and mounted with Vectashield (Vector Laboratories, Inc., Burlingame, CA). For staining with anti-HCMV antibody, the permeabilized slides were stained with anti-IE1 monoclonal antibody (Goodwin Institute of Cancer Research, Plantation, FL), and then with secondary anti-mouse IgG conjugated to FITC and/or Texas-Red (Vector Laboratories, Inc., Burlingame, CA), prior to counterstain with DAPI. Images were visualized on a Nikon PCM2000 confocal microscope system 46. The monoclonal antibodies against cytokeratins K13 and K14 were purchased from United States Biological (Swampscott, MA).

The tissues were either mock-infected or infected with 2 × 10⁴ PFU of different HCMV strains and mutants, then incubated for 0–10 days. Viral proteins were isolated as described previously 47. The polypeptides from cell lysates were separated on either SDS/7.5% polyacrylamide gels or SDS/9% polyacrylamide gels cross-linked with N,N"methylenebisacylamide, and transferred electrically to nitrocellulose membranes. We stained the membranes using the antibodies against HCMV proteins and human actin in the presence of a chemiluminescent substrate (Amersham Inc, Arlington Heights, IL), and analyzed the stained membranes with a STORM840 phosphorimager. Quantitation was performed in the linear range of protein detection 47. The monoclonal antibodies c1202, c1203s, and c1207, which react with HCMV proteins UL44, IE1, and UL99; were purchased from Goodwin Institute for Cancer Research (Plantation, FL). The monoclonal antibody against human actin was purchased from Sigma Inc (St Louis, MO).

Two different sets of experiments were carried out to study the effect of ganciclovir (GCV) 3132 on HCMV replication in the oral tissues. First, the tissues were first pre-incubated with different concentrations (i.e. 10 μM and 100 μM) of GCV for 2 hours, and then incubated with the viral inoculum in the presence of GCV for 4 hours to initiate HCMV infection. In the second set of experiments, the tissues were incubated with viral inoculum for 4 hours in the absence of GCV, and then incubated in fresh media in the absence of GCV for additional 24 hours before adding different concentrations of GCV to the culture. The infected tissues were incubated in the GCV-containing media for different periods of time and harvested, and viral titers in these tissues were determined by plaque assays on HFFs.

Growth analyses of different HCMV strains and mutants in vitro in primary human foreskin fibroblasts (HFFs) were carried out as described previously 23. Briefly, 1 × 10⁶ human foreskin fibroblasts were infected at an MOI of 0.05 PFU per cell. The cells and media were harvested at 0, 2, 4, 7, 10 and 14 days post infection, and viral stocks were prepared by adding an equal volume of 10% skim milk, followed by sonication. The titers of the viral stocks were determined by plaque assays on HFFs in triplicates.