Human Cytomegalovirus Productively Replicates In Vitro in Undifferentiated Oral Epithelial Cells

The ocular epithelial cells currently used to study HCMV infections in vitro have historical significance based upon their role in retinitis, an HCMV disease most often seen in AIDS patients. However, with the successful implementation of highly active antiretroviral therapy (HAART) regimens, the incidence of HCMV retinitis has rapidly declined, and therefore, the relevance of studying ocular epithelial cell HCMV infection has decreased as well. Our introduction here of oral epithelial cells provides two alternative in vitro models for the study of HCMV infection that complement and extend the physiologic relevance of the ocular system currently in use.

ABSTRACT

Human cytomegalovirus (HCMV) productive replication in vitro is most often studied in fibroblasts. In vivo, fibroblasts amplify viral titers, but transmission and pathogenesis require the infection of other cell types, most notably epithelial cells. In vitro, the study of HCMV infection of epithelial cells has been almost exclusively restricted to ocular epithelial cells. Here we present oral epithelial cells with relevance for viral interhost transmission as an in vitro model system to study HCMV infection. We discovered that HCMV productively replicates in normal oral keratinocytes (NOKs) and telomerase-immortalized gingival cells (hGETs). Our work introduces oral epithelial cells for the study of HCMV productive infection, drug screening, and vaccine development.

IMPORTANCE The ocular epithelial cells currently used to study HCMV infections in vitro have historical significance based upon their role in retinitis, an HCMV disease most often seen in AIDS patients. However, with the successful implementation of highly active antiretroviral therapy (HAART) regimens, the incidence of HCMV retinitis has rapidly declined, and therefore, the relevance of studying ocular epithelial cell HCMV infection has decreased as well. Our introduction here of oral epithelial cells provides two alternative in vitro models for the study of HCMV infection that complement and extend the physiologic relevance of the ocular system currently in use.

INTRODUCTION

Human cytomegalovirus (HCMV) is a betaherpesvirus that infects the majority of the population worldwide. In healthy individuals, infection is generally subclinical, but the virus causes severe disease in individuals with compromised, suppressed, or incompletely developed immune systems (1, 2). For example, HCMV causes retinitis and blindness in AIDS patients, organ dysfunction or failure in transplant recipients, and birth defects, including microcephaly and hearing loss, in neonates. Furthermore, research continues into possible roles for HCMV in human cancers such as glioblastoma (3–5). No vaccine exists, but antiviral drugs are available, including the recently approved drug letermovir. The drugs are useful against productive (lytic) infection but select for resistant viruses (6, 7). Furthermore, the approved antivirals do not target the latent reservoirs that permit the virus to infect its host for their lifetime and that reactivate to allow spread within and between hosts.

HCMV infects an array of cell types, including myeloid-lineage cells and endothelial, fibroblast, smooth muscle, glial, and epithelial cells (8). It is speculated, and in many cases known, that infection of a specific cell type has a unique biological or clinical role (9). For example, latency is established in the myeloid and perhaps endothelial cell lineages, and fibroblasts are sites of viral amplification. Furthermore, smooth muscle and glial cell infection may play roles in chronic pathologies associated with HCMV infection (atherosclerosis and cancer, respectively), and epithelial cell infection contributes to both organ disease and viral dissemination through urine, breast milk, and saliva. Therefore, studying HCMV infection in different and relevant cell types can reveal novel insights about replication, persistence, and pathogenesis and point to unique opportunities for novel antiviral interventions.

At the height of the AIDS epidemic, HCMV was a common opportunistic infection in patient cohorts. Retinitis caused by HCMV occurred in 25% to 42% of AIDS patients (10), often leading to blindness. Within the eye, the retinal pigmented epithelium (RPE) was defined as the primary site of infection (11). The severity of the clinical syndrome and the identification of the major relevant cell type sparked interest in studying in vitro HCMV infections in RPE cells (12–14), which were found to be productive rather than latent (see below for further discussion). Importantly, RPE cells were instrumental in determining that the viral pentameric glycoprotein complex (15, 16) was essential for infection of epithelial cells (17). The development and widespread use of highly active antiretroviral therapy (HAART) regimens not only stabilized the AIDS epidemic in developed countries but also dramatically reduced the incidence of retinitis caused by HCMV by ∼80% (10). Therefore, while epithelial cells still remain an import in vivo target of HCMV and thus a priority for in vitro study, both the biological and clinical relevance of studying RPE cells has substantially decreased.

Surprisingly, HCMV infection in other types of epithelial cells has received far less study. Sporadic reports of in vitro HCMV infections of epithelial cells from the cervix (13), cochlea (18), kidney (19), mammary gland (20), and thyroid (21) have appeared. These infections were found to be productive. There is a significant body of work examining murine cytomegalovirus (MCMV) replication in the salivary gland (22). However, despite the high concentration of infectious HCMV found in saliva and the identification of HCMV in oral epithelial cells in vivo (23–26), cultured human oral epithelial cell populations have not been used in vitro to study HCMV infections. More work has examined infection by the gammaherpesviruses Epstein-Barr virus (EBV) and Kaposi's sarcoma-associated herpesvirus (KSHV) in oral epithelial cells from stratified squamous epithelia (see below for a description of oral epithelial cell differentiation). For EBV and KSHV, undifferentiated cells supported latent infection, whereas differentiated cells supported productive infection (27–29). Interestingly, this mimics HCMV infection of myeloid cells, where the virus establishes latency within undifferentiated cells but replicates productively in fully differentiated cells.

We examined HCMV infection of two undifferentiated oral epithelial cell cultures, telomerase-immortalized normal oral keratinocytes (NOKs) and telomerase-immortalized gingival cells (hGETs), to define the mode of infection. We determined that these oral epithelial cells support productive HCMV infection. This work establishes NOKs and hGETs as new models for the study of HCMV productive infection, dissemination, and antiviral sensitivity.

RESULTS

HCMV replicates productively in RPE cells.

The accepted and appropriate definition for productive infection is the release of infectious progeny virus. Latency is defined as the maintenance of the viral genome over time without generating infectious progeny with the capacity for future reactivation to a productive infection. These criteria take time to be realized, and therefore, molecular events that happen quickly after infection are often used as surrogates to predict whether an infection will be productive or latent. For example, productive infections are often characterized by the rapid and high-level accumulation of the viral immediate early 1 (IE1) protein (30). IE1 (UL123) transcription is initiated by the action of the tegument-delivered pp71 protein in the nucleus (31). Therefore, productive infection is inferred when tegument-delivered pp71 transits to the nucleus and when the IE1 protein accumulates. Latent infections are often characterized by the absence of IE1 protein accumulation. As IE1 drives the productive cycle and is a target for immune-mediated cell killing, keeping IE1 protein levels low or absent would appear to be a reasonable strategy to establish a latent infection in vitro, and even more so in vivo. IE1 protein accumulation during latency is prevented because the gene that encodes it (UL123) is transcriptionally silenced, in part, because tegument-delivered pp71, the protein that activates its transcription, is sequestered in the cytoplasm (32). While IE1 (UL123) is transcriptionally silenced during latency, other viral loci are not, including the long noncoding RNA (lncRNA) B2.7 (33). B2.7 is also expressed during productive infections. Therefore, latency is inferred when B2.7 (or another latency-associated transcript) accumulates but IE1 (UL123) RNAs fail to accumulate to productive-phase levels and when tegument-delivered pp71 is cytoplasmic.

HCMV requires the viral pentameric protein complex in its envelope to enter RPE cells (17). To test if the pentamer is also required for viral entry into oral epithelial cells, we used a matched pair (13, 17) of pentamer-negative (AD169) and pentamer-positive (ADrUL131) viral strains. To distinguish between productive and latent infections, we monitored tegument-delivered pp71 subcellular localization, B2.7 and IE1 (UL123) mRNA levels, and IE1 protein accumulation.

In tissues in vivo or in organotypic raft cultures in vitro, oral epithelial cells form a stratified squamous epithelium consisting of a basal layer of dividing, undifferentiated cells and multiple suprabasal layers of nondividing, differentiated cells (34). These cells express specific keratin proteins and therefore are classified as keratinocytes. They also differentially express the involucrin protein. Involucrin marks the differentiated cells within the upper (suprabasal) levels of a stratified squamous epithelium but is not found in undifferentiated (basal) cells (35). Prior to rafting, oral epithelial cells can be maintained in vitro as monolayers in their undifferentiated state by subconfluent culture in serum-free medium. As monolayers, they can be differentiated either through treatment with fetal bovine serum (FBS) and calcium or by the addition of methylcellulose to the growth medium (25, 36, 37). Our NOKs and hGETs maintained as monolayers in serum-free media did not express involucrin, but involucrin expression (differentiation) could be induced either by FBS and calcium or by methylcellulose (Fig. 1). Involucrin expression is restricted to keratinocytes and stratified squamous epithelia and therefore is not detected (Fig. 1) in RPEs or normal human dermal fibroblasts (NHDFs), a common model for productive HCMV infection.

FIG 1.

Calcium- or methylcellulose-dependent differentiation of monolayer NOKs or hGETs. NOKs or hGETs were left untreated and undifferentiated (U) or differentiated with calcium (Ca) or methylcellulose (MC). Protein lysates were analyzed by Western blotting with the indicated antibodies. Untreated RPE and HF cell lysates served as controls. Results are representative of data from three independent experiments.

As expected based on previous reports of productive HCMV replication, upon infection of RPE cells with ADrUL131, tegument-delivered pp71 entered the nucleus (Fig. 2), both B2.7 and IE1 (UL123) RNAs were expressed (Fig. 3), the IE1 protein accumulated in the nucleus (Fig. 4), and the late proteins UL44 and pp28 also accumulated (Fig. 5A). For all these assays (Fig. 2 to 5), human fibroblasts (HFs) served as a positive control for productive infection. Importantly, infectious progeny were released from RPE cells (Fig. 6A), defining the infection as productive. The parent virus, AD169, does not express the pentamer and, as previously described (13, 17), was unable to enter RPE cells, a conclusion based on an inability to detect intracellular pp71 (Fig. 2) or IE1 (Fig. 4) proteins, B2.7 or IE1 (UL123) RNAs (Fig. 3), or infectious progeny virions (Fig. 6A).

FIG 2.

Pentamer-competent HCMV delivers tegument-incorporated pp71 to the nucleus of differentiated NOKs or hGETs. (A) The indicated cells grown on coverslips were infected at a multiplicity of infection (MOI) of 3 with UV-inactivated pentamer-competent (AdrUL131) or UV-inactivated pentamer-deficient (AD169) HCMV for 24 h. The intracellular localization of tegument-delivered pp71 was determined by indirect immunofluorescence microscopy. (B) Differentiated NOKs or hGETs grown on coverslips were infected with either live virus or the UV-inactivated virus preparations used for panel A. IE1 expression was visualized by indirect immunofluorescence. Results are representative of data from three independent experiments. DAPI, 4′,6-diamidino-2-phenylindole.

FIG 3.

Pentamer-competent HCMV transcribes the viral IE1 (UL123) and B2.7 genes in NOKs or hGETs. (A) The indicated untreated cells were mock infected or infected with the indicated virus at an MOI of 3. After 24 h, RNA was prepared and analyzed by real-time quantitative reverse transcriptase PCR (qRT-PCR) for B2.7. (B) qRT-PCR analysis for IE1 (UL123) with RNA generated as described above for panel A. Results from three biological replicates were analyzed with Student's t test. *, P < 0.05; **, P < 0.01; n.s., not significant. (C) The indicated differentiated cells were mock infected or infected with the indicated virus at an MOI of 3. After 24 h, RNA was prepared and analyzed by qRT-PCR for B2.7. (D) qRT-PCR analysis for IE1 (UL123) with RNA generated as described above for panel C. Results from three biological replicates were analyzed with Student's t test. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

FIG 4.

The IE1 protein accumulates in differentiated NOKs or hGETs infected with pentamer-competent HCMV. The indicated cells grown on coverslips were infected at an MOI of 3 with the indicated virus for 24 h. IE1 expression was visualized by indirect immunofluorescence microscopy. Results are representative of data from three independent experiments.

FIG 5.

Representative members of each productive-phase kinetic class of viral protein analyzed in differentiated NOKs or hGETs infected with pentamer-competent HCMV. (A) The indicated cells were infected with ADrUL131 at an MOI of 3, and protein lysates harvested at the indicated times were analyzed by Western blotting with the indicated antibodies. Lysates from uninfected, differentiated NOKs and hGETs were analyzed as controls. Results are representative of data from three independent experiments. (B) The indicated cells were infected with ADrUL131 at an MOI of 3, and protein lysates harvested at the indicated times were analyzed by Western blotting with the indicated antibodies. Lysates from uninfected, undifferentiated NOKs and hGETs were analyzed as controls. Results are representative of data from three independent experiments.

FIG 6.

Analysis of pentamer-competent HCMV productive replication in oral epithelial cells. (A) Monolayer RPE cells were infected at an MOI of 1. Infected cells and media were collected at the indicated times, from which virus preparations were generated and titers were determined. Results from three independent biological replicates are shown. Error bars represent the standard errors of the means. (B) Monolayer differentiated cells were analyzed as described above for panel A. Results from three independent biological replicates are shown. Error bars represent the standard errors of the means. (C) Monolayer undifferentiated cells were analyzed as described above for panel A. Results from three independent biological replicates are shown. Error bars represent the standard errors of the means.

HCMV initiates but fails to complete productive infection in differentiated NOKs and hGETs.

We first compared RPE infections to infections of differentiated NOKs (Diff-NOKs) and differentiated hGETs (Diff-hGETs). Diff-NOKs and Diff-hGETs initiate the hallmarks of a pentamer-dependent productive HCMV infection but fail to complete the process. ADrUL131 infection delivered pp71 to the nucleus (Fig. 2), initiated the transcription of both B2.7 (Fig. 3C) and IE1 (UL123) (Fig. 3D), and resulted in IE1 protein accumulation in the nucleus (Fig. 4). AD169 was unable to achieve any of these hallmarks of productive infection initiation and failed to deliver pp71 to the cytoplasm, indicating that a latent infection was not established. Immediate early (IE1) and early (UL44) proteins accumulated in ADrUL131-infected Diff-NOKs and Diff-hGETs (Fig. 5B), as they did in NHDFs and RPE cells (Fig. 5A). The level of the late protein pp28 remained below our limit of detection. Involucrin expression in Diff-NOKs and Diff-hGETs (Fig. 5B) confirmed that these oral epithelial cells were differentiated. Both Diff-NOKs and Diff-hGETs generated few if any infectious viral progeny after infection with ADrUL131 (Fig. 6B). We conclude that AD169 fails to enter Diff-NOKs and Diff-hGETs because it lacks a functional pentamer complex. Furthermore, we conclude that a pentamer-expressing HCMV initiates a productive infection upon entering Diff-NOKs or Diff-hGETs but fails to successfully complete the process.

HCMV replicates productively in undifferentiated NOKs and hGETs.

Interestingly, undifferentiated NOKs and hGETs also illustrated the hallmarks of a pentamer-dependent productive infection, and in these cells, productive replication was successfully completed. ADrUL131 infection delivered pp71 to the nucleus (Fig. 7), initiated the transcription of both B2.7 (Fig. 3A) and IE1 (UL123) (Fig. 3B), and resulted in IE1 protein accumulation in the nucleus (Fig. 8). For all these assays (Fig. 3, 7, and 8), HFs served as a positive control for productive infection. AD169 was unable to achieve any of these hallmarks of productive infection and failed to deliver pp71 to the cytoplasm, indicating that a latent infection was not established. However, AD169 entered NHDFs and expressed both B2.7 (Fig. 3A) and IE1 (UL123) (Fig. 3B) RNAs, indicating the initiation of a productive infection in fibroblasts, as expected. Immediate early (IE1), early (UL44), and late (pp28) proteins accumulated in ADrUL131-infected NOKs and hGETs (Fig. 9A), as they did in NHDFs and RPE cells (Fig. 9B). Involucrin expression in Diff-NOKs and Diff-hGETs, but not in undifferentiated NOKs or hGETs (Fig. 9A), discriminated differentiated from undifferentiated epithelial cells. Both NOKs and hGETs generated infectious viral progeny after infection with ADrUL131 (Fig. 6C). We conclude that AD169 fails to enter NOKs and hGETs because it lacks a functional pentamer complex. Furthermore, we conclude that a pentamer-expressing HCMV initiates a productive infection upon entering NOKs or hGETs, examples of undifferentiated oral epithelial cells.

FIG 7.

Pentamer-competent HCMV delivers tegument-incorporated pp71 to the nucleus of undifferentiated NOKs or hGETs. (A) The indicated cells grown on coverslips were infected at an MOI of 3 with UV-inactivated pentamer-competent (AdrUL131) or UV-inactivated pentamer-deficient (AD169) HCMV for 24 h. The intracellular localization of tegument-delivered pp71 was determined by indirect immunofluorescence microscopy. (B) Undifferentiated NOKs or hGETs grown on coverslips were infected with either live virus or the UV-inactivated virus preparations used for panel A. IE1 expression was visualized by indirect immunofluorescence. Results are representative of data from three independent experiments.

FIG 8.

The IE1 protein accumulates in undifferentiated NOKs or hGETs infected with pentamer-competent HCMV. The indicated cells grown on coverslips were infected at an MOI of 3 with the indicated virus for 24 h. IE1 expression was visualized by indirect immunofluorescence microscopy. Results are representative of data from three independent experiments.

FIG 9.

Representative members of each productive-phase kinetic class of viral protein accumulate in undifferentiated NOKs or hGETs infected with pentamer-competent HCMV. (A) The indicated cells were infected with ADrUL131 at an MOI of 3, and protein lysates harvested at the indicated times were analyzed by Western blotting with the indicated antibodies. Lysates from uninfected, differentiated NOKs and hGETs were analyzed as controls. Results are representative of data from three independent experiments. (B) The indicated control cells were analyzed as described above. Results are representative of data from three independent experiments.

HCMV replication in NOKs and hGETs is cytopathic.

The numbers of cells in which the IE1 protein accumulated during the first 24 h after ADrUL131 infection were similar between undifferentiated and differentiated hGETs but slightly higher for differentiated than for undifferentiated NOK cells (Fig. 10A). Interestingly, tegument-delivered pp71 migrated to the nucleus much faster in NHDFs, differentiated NOKs, and differentiated hGETs than it did in RPE cells, undifferentiated NOKs, or undifferentiated hGETs (Fig. 10B). The rapid delivery of pp71 to the nucleus does not dramatically affect viral gene expression in epithelial cells, nor does it predict infection outcome, and therefore, its significance, if any, is unknown.

FIG 10.

Differentiated NOKs, but not hGETs, display more IE1-positive cells early after infection than do their undifferentiated counterparts. (A) The indicated cells grown on coverslips were infected at an MOI of 3 with ADrUL131 for 24 h. IE1 expression was visualized by indirect immunofluorescence microscopy, and the fraction of positive cells was determined visually. The averages of data from three biological replicates are plotted. Error bars represent the standard errors of the means. The data were analyzed with Student's t test. **, P < 0.01; ns, not significant. (B) The indicated cells grown on coverslips were infected at an MOI of 3 with pentamer-competent (AdrUL131) HCMV for 6 h. The intracellular localization of tegument-delivered pp71 was determined by indirect immunofluorescence microscopy. Results are representative of data from three independent experiments.

Increased cytopathic effect (CPE) was observed upon infection of NOKs or hGETs compared to RPE cells at 5 days postinfection (Fig. 11). Time course studies (Fig. 12) showed that the CPE in undifferentiated NOKs and hGETs required viral replication, increased with infection time, and was more evident than in similar infections of differentiated NOKs or differentiated hGETs. Viral replication and the subsequent CPE may in part explain why we were unable to raft undifferentiated and infected NOKs or hGETs. While uninfected cultures generated typical stratified squamous epithelia, essentially only the collagen matrix remained with infected cultures after 14 days (Fig. 13). Similar results were obtained after 6 days (Fig. 14), a time frame more similar to a single round of productive HCMV replication. UV-inactivated virus had no effect on the rafting of NOKs. hGETs treated with UV-inactivated HCMV still formed rafts (Fig. 14), although their thickness was reduced compared to that in mock-infected cultures. Thus, HCMV replication-induced cell death of undifferentiated NOKs and hGETs occurs with faster kinetics than does raft generation. In total, we conclude that HCMV productively infects undifferentiated oral epithelial cells in vitro in a pentamer-dependent manner.

FIG 11.

HCMV infection of undifferentiated NOKs or hGETs produces a cytopathic effect. Monolayer cells were mock infected or infected with green fluorescent protein (GFP)-expressing ADrUL131 at an MOI of 1 for 5 days, after which bright-field and fluorescence images were captured. Images are representative of results from at least three independent experiments.

FIG 12.

HCMV-induced cytopathic effect in monolayer NOKs and hGETs requires viral replication. Monolayer cells were mock infected or infected with GFP-expressing ADrUL131 at an MOI of 1 for the indicated times, after which bright-field and fluorescence images were captured. Images are representative of results from three independent experiments. (A) NOKs; (B) differentiated NOKs; (C) hGETs; (D) differentiated hGETs. dpi, days postinfection.

FIG 13.

HCMV infection of NOKs or hGETs prevents stratification upon raft culturing. Monolayer cells were mock infected or infected with pentamer-competent (AdrUL131) or pentamer-deficient (AD169) HCMV at an MOI of 3 for 1 h and then rafted. Rafts were cultured for 14 days prior to sectioning, staining, and imaging. Images are representative of results from two independent experiments.

FIG 14.

HCMV-induced cytopathic effect in rafted NOKs or hGETs requires viral replication. (A) Monolayer cells were mock infected or infected with live or UV-inactivated ADrUL131 at an MOI of 3 for 1 h and then rafted. Rafts were cultured for 6 days prior to sectioning, staining, and imaging. Images are representative of results from three independent experiments. (B) NHDFs were infected with the indicated viral stocks used for panel A at an MOI of 3, and protein lysates harvested at 24 h postinfection were analyzed by Western blotting with the indicated antibodies. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

DISCUSSION

Epithelial cells from different anatomical sites are not identical. For example, in vivo, RPEs are postmitotic, monolayer cells (38), whereas oral keratinocytes form multilayer structures composed of differentially located dividing or postmitotic (differentiated) cells (30). Therefore, it might not be surprising if specific characteristics of in vitro infection with HCMV were to differ depending upon the type of epithelial cell studied. Furthermore, the relevance of the results to pathogenesis or dissemination would also seem to rely on the origin of the epithelial cell used for in vitro study. The classic epithelial cell type used to study HCMV infection in vitro has been RPE cells, mainly because of the incidence of HCMV retinitis in pre-HAART-era AIDS patients. As the prevalence of retinitis continues to diminish, epithelial cells from other anatomical sites that model other aspects of HCMV disease or transmission should supplant RPE cells as models of choice for studying HCMV infection in vitro.

We chose to examine in vitro HCMV infection of oral epithelial cells because we were intrigued at the possibility that, in vivo, such infections might provide a latent reservoir important for viral transmission from person to person through saliva. HCMV is known to reside latently in bone marrow hematopoietic progenitor cells and circulating monocytes and may form an additional latent reservoir in microvascular endothelial cells. Reactivation from these sites of carriage seems to reasonably explain intrahost dissemination. However, without a subsequent amplification step (for example, in differentiated epithelial cells), it is unclear if reactivation from the known sites of latency could be directly responsible for most interhost dissemination events. Conversely, reactivations in the oral cavity could easily explain high-level host-to-host transmission through saliva.

There were two major reasons why we hypothesized that oral epithelial cells could represent an undiscovered site of HCMV latency. First, the oral cavity sustains protracted infections, undergoes a high incidence of reactivation events, and appears to be an immune-privileged site for viral persistence, all contributing to viral dissemination via saliva (22). Second, both EBV and KSHV establish latent infections in undifferentiated, basal-layer oral keratinocytes but reactivate upon cellular differentiation (23–25). The known dependence upon differentiation of HCMV reactivation from the myeloid lineage (39) as well as the known characteristics of oral cavity infection mentioned above prompted us to ask if undifferentiated oral epithelial cells supported a productive or a latent infection.

Our results unequivocally demonstrate that undifferentiated oral epithelial cells support productive HCMV infection in vitro. While previous reports of HCMV in vitro infection of epithelial cells sourced from other anatomical sites have all reported productive infection, the cells examined do not generate stratified epithelial layers and thus did not afford the possibility to examine HCMV replication within an undifferentiated subset of cells. The CPE elicited by in vitro infection may contribute to the low titers of infectious progeny generated by these oral epithelial cells. If such CPE were also characteristic of in vivo infections of the oral epithelium, loosely attached infected cells may spread the virus from person to person through saliva. CPE might also provide a recruiting signal for monocytes, which could then carry latent virus to distant anatomical sites within the original host, thereby promoting intrahost viral dissemination. Therefore, these cells represent an opportunity to study and understand the mechanisms and consequences of CPE during an HCMV infection. Immortalization with telomerase does not increase CPE or decrease viral titers during productive infection of fibroblasts (40) and therefore is unlikely to be the cause of these phenotypes in NOKs and hGETs. In addition, the apparent abortive infection in fully differentiated oral epithelial cells provides a chance to identify host factor-dependent late steps critical for productive viral replication that become immediate targets for antiviral drug development.

It is unclear why infection of undifferentiated oral epithelial cells with HCMV, a betaherpesvirus, is substantially different from infection with the gammaherpesviruses EBV and KSHV, or even human papillomavirus, where productive replication is limited to suprabasal cells. In vitro infection of oral epithelial cells with alphaherpesviruses is also understudied; however, two studies with herpes simplex virus 1 (HSV-1) generated data consistent with productive replication in undifferentiated cells (41, 42). Furthermore, HSV-1 productive replication in epithelial cells from other anatomical locations has been reported (43, 44).

Here we introduce NOKs and hGETs as novel models to study in vitro epithelial cell infection. Our initial results with these two new models confirm and extend the absolute dependence for epithelial infection of the viral glycoprotein pentameric complex. In the future, NOKs and hGETs will be useful to us and others for the continued search for the elusive receptor(s) for the HCMV pentamer and in screens for neutralizing antibodies, vaccines, or antiviral drugs that may inhibit epithelial cell infection by alpha-, beta-, and gammaherpesviruses and thereby curtail or eliminate both oral disease and viral dissemination through saliva.

MATERIALS AND METHODS

Cells and viruses.

NOKs are human telomerase reverse transcriptase (hTERT)-expressing cells from human gingival tissue (45), generously provided by Karl Münger. hGETs are hTERT-expressing normal human gingival epithelial progenitor cells (46). NOKs and hGETs were cultured as monolayers in keratinocyte serum-free medium (KSFM; Life Technologies) supplemented with 0.14 μg/ml epidermal growth factor and 0.33 mg/ml bovine pituitary extract at 5% CO₂ and 37°C. To maintain their undifferentiated state, the culture medium is changed every other day, and cells are never allowed to approach confluence. NOKs and hGETs were differentiated by treatment with either calcium chloride (1.2 mM CaCl₂ and 10% [vol/vol] fetal bovine serum [FBS] in KSFM) or methylcellulose (1.6 g/ml in KSFM without supplements) (catalog number M-0512; Sigma) for 48 h at 37°C. Rafting of NOKs and hGETs was performed as previously described for NIKs (near-diploid immortalized keratinocytes) (47). RPE cells are human retinal pigmented epithelial cells (ARPE-19; ATCC CRL-2302) and were provided by Tom Shenk. RPE cells were cultured in Dulbecco's modified Eagle medium 12 (DMEM-12; Life Technologies) supplemented with 10% (vol/vol) FBS, 15 mM HEPES (pH 7.3), 100 U/ml penicillin, and 100 μg/ml streptomycin at 5% CO₂ and 37°C. NHDFs are normal human dermal fibroblasts (Clonetics) and were cultured in DMEM supplemented with 10% (vol/vol) FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin at 5% CO₂ and 37°C. The HCMV strains used were AD169 and an AD169 derivative (ADrUL131, generously provided by Tom Shenk) in which the frameshift mutation in the UL131 gene was repaired (13). Viruses were grown and titers were determined on NHDFs. Infections were performed in a minimal volume of culture medium at 37°C for 1 h in a tissue culture incubator, after which additional medium was added.

Assays and antibodies.

Western blot analyses (48), indirect immunofluorescence microscopy (49), and quantitative real-time PCR (50) were performed with primer pairs (51) as previously described. For plaque assays, infected cells and media were harvested, subjected to 1 freeze (−80°C)-thaw (37°C) cycle, and sonicated three times (20 s each, 50% duty cycle, and output control 5 on a Branson 450 sonifier). Debris was removed by low-speed centrifugation. Serial dilutions were used to infect fibroblasts for 1 h, after which they were overlaid with 1% low-melting-point agarose in minimum essential medium Eagle (EMEM) without phenol red and l-glutamine (catalog number 12-668E; BioWhittaker) supplemented with 10% FBS and 0.34% NaHCO₃. After 17 days, plates were stained with 0.03% methylene blue in phosphate-buffered saline (PBS), and plaques were counted. Antibodies for tubulin (catalog number DM 1A; Sigma), UL44 (catalog number CA006-100; Virusys), and involucrin (catalog number 19018; Sigma) were obtained from commercial sources. Antibodies against pp71 (IE-233), IE1 (1B12), pp150 (CMV127), and pp28 (CMV157) were described previously (52). Secondary antibodies for immunofluorescence conjugated with Alexa Fluor 594 were obtained from Molecular Probes. Secondary antibodies for Western blots conjugated with IRDye-680 were obtained from LiCor.