Epstein–Barr Virus Infection of Pseudostratified Nasopharyngeal Epithelium Disrupts Epithelial Integrity

Simple Summary Nasopharyngeal carcinoma is associated with Epstein-Barr virus (EBV) infection and originates junction of the oropharynx and nasal cavity, where stratified squamous epithelium and respiratory epithelium are the lining. To elucidate the mechanisms by which EBV transforms the nasopharyngeal epithelium, a pseudostratified multiple-layer model with cilia forming on the apical surface by air-liquid interface (ALI) culture of primary nasopharyngeal epithelial cells was established. We showed: (1) ALI cultures formed stratified epithelia and maintained the diversity of cells found in the airway epithelium, such as ciliated, muco-secretory, and basal cells. (2) Polarized stratified epithelium was more susceptible to EBV infection than monolayer cells. (3) EBV infection in ALI cultures was verified by showing EBV-encoded RNA expressions. (4) EBV infection disrupted the integrity of the epithelium. Thus, our model can be used not only to examine the pathogenesis of pre-neoplastic EBV-infected cells, but also to develop anti-EBV therapy or early stage NPC treatment. Abstract Epstein–Barr virus (EBV) is a human oncogenic virus that causes several types of tumor, such as Burkitt’s lymphoma and nasopharyngeal carcinoma (NPC). NPC tumor cells are clonal expansions of latently EBV-infected epithelial cells. However, the mechanisms by which EBV transforms the nasopharyngeal epithelium is hampered, because of the lack of good in vitro model to pursue oncogenic process. Our primary nasopharyngeal epithelial cell cultures developed pseudostratified epithelium at the air-liquid interface, which was susceptible to EBV infection. Using the highly sensitive RNA in situ hybridization technique, we detected viral infection in diverse cell types, including ciliated cells, goblet cells, and basal cells. EBV-encoded small RNA-positive cells were more frequently detected in the suprabasal layer than in the basal layer. We established the most physiologically relevant EBV infection model of nasopharyngeal epithelial cells. This model will advance our understanding of EBV pathogenesis in the development of NPC.


Introduction
Epstein-Barr virus (EBV) is a ubiquitous gamma herpesvirus that infects both B-lymphocytes and epithelial cells [1]. The EBV infection of these cells can lead to the clonal expansion of a persistently infected cell, and sometimes induces cancer, such as nasopharyngeal carcinoma (NPC) [2,3]. However, the mechanisms by which EBV transforms the nasopharyngeal epithelium remain mostly unknown. To further understand these mechanisms, we need to develop more physiologically relevant in vitro epithelial infection models.
A variety of in vitro epithelial cell models for EBV infection have been developed, ranging from monolayer unpolarized epithelial cells to polarized epithelial cells, using oropharyngeal or tonsillar epithelial cells [4][5][6][7][8]. Although many useful data have been generated using these models, they may not closely represent the physiological condition in the nasopharynx, for the following reasons. First, most epithelial cells were derived from oropharyngeal or tonsillar epithelium, but cells from the nasopharynx were scarcely used, because they were difficult to obtain. Second, the nasopharynx is located at the junction of the oropharynx and nasal cavity, where stratified squamous epithelium and respiratory epithelium are the lining. The stratified squamous epithelium is localized on the inferior anterior, inferior posterior and anterior lateral walls, while the respiratory epithelium is localized around the nasal choanae and roof of the posterior wall. Lastly, abundant lymphoid tissues are found underneath the nasopharynx, which is structurally distinct from other anatomical regions, such as oropharyngeal or tonsillar epithelia.
In this study, we established a conventional monolayer model and a pseudostratified multiple-layer model, using epithelial cells derived from rare and precious nasopharynx specimens, to mimic the squamous and respiratory parts of the nasopharyngeal epithelium. We investigated EBV infection using these models (Figure 1a).

Introduction
Epstein-Barr virus (EBV) is a ubiquitous gamma herpesvirus that infects both B-lymphocytes and epithelial cells [1]. The EBV infection of these cells can lead to the clonal expansion of a persistently infected cell, and sometimes induces cancer, such as nasopharyngeal carcinoma (NPC) [2,3]. However, the mechanisms by which EBV transforms the nasopharyngeal epithelium remain mostly unknown. To further understand these mechanisms, we need to develop more physiologically relevant in vitro epithelial infection models.
A variety of in vitro epithelial cell models for EBV infection have been developed, ranging from monolayer unpolarized epithelial cells to polarized epithelial cells, using oropharyngeal or tonsillar epithelial cells [4][5][6][7][8]. Although many useful data have been generated using these models, they may not closely represent the physiological condition in the nasopharynx, for the following reasons. First, most epithelial cells were derived from oropharyngeal or tonsillar epithelium, but cells from the nasopharynx were scarcely used, because they were difficult to obtain. Second, the nasopharynx is located at the junction of the oropharynx and nasal cavity, where stratified squamous epithelium and respiratory epithelium are the lining. The stratified squamous epithelium is localized on the inferior anterior, inferior posterior and anterior lateral walls, while the respiratory epithelium is localized around the nasal choanae and roof of the posterior wall. Lastly, abundant lymphoid tissues are found underneath the nasopharynx, which is structurally distinct from other anatomical regions, such as oropharyngeal or tonsillar epithelia.
In this study, we established a conventional monolayer model and a pseudostratified multiplelayer model, using epithelial cells derived from rare and precious nasopharynx specimens, to mimic the squamous and respiratory parts of the nasopharyngeal epithelium. We investigated EBV infection using these models (Figure 1a).  Green fluorescent protein (GFP)-tagged virus-producing Akata cells were added into monolayer culture or air-liquid interface (ALI) culture of nasopharyngeal epithelial cells for infection. (b) Morphology and marker expression in nasopharyngeal epithelial cell cultures. Representative microscopic images depicting cells in monolayer (i) and cells expressing pan-CK and p63 determined by immunofluorescence (ii,iii). Hematoxylin and eosin staining of the longitudinal section of day 28 ALI culture (iv). It consists of multi-layer of cells with cilia forming on the apical surface. (c) In vitro EBV infection of the monolayer of nasopharyngeal epithelial cells. Merged picture of phase contrast and GFP channel (i). Merged picture of GFP and DAPI (nuclear counterstain in blue) (ii). Infection efficiency was determined by calculating GFP positive cells over total cells in DAPI.

Establishment of Primary Nasopharyngeal Epithelial Cell Cultures
Cells derived from normal nasopharynx biopsies could grow directly on plastic wells without feeder cells. Cells exhibited a typical polygonal epithelial appearance, and strongly expressed epithelial cell markers: pan-cytokeratin (pan-CK) and basal cell marker p63 (Figure 1b). The pseudostratified epithelium was formed in a transwell insert after 3 to 4 weeks of air-liquid interface (ALI) culturing. A movie of ciliated beating cells of the pseudostratified epithelium is presented (Supplementary Video S1). Hematoxylin and eosin (H&E) staining of a longitudinal cryosection showed that the epithelium was composed of multiple layers, with cilia on the apical surface ( Figure 1b).

EBV Infects a Conventional Monolayer of Nasopharyngeal Epithelial Cells
Our previous study indicated that EBV transfer of nasopharyngeal epithelial cells is more efficiently mediated by cell-to-cell contact than by cell-free infection [9]. Therefore, instead of using cell-free infection, we employed cell-mediated infection by adding virus-producing Akata cells infected with recombinant EBV, containing the green fluorescent protein (GFP) gene at the viral BXLF1 locus (Akata GFP-EBV cells) [10], onto monolayer nasopharyngeal epithelial cells. Because Akata GFP-EBV cells produce viruses upon stimulation with anti-human immunoglobulin G (IgG) [11], EBV-infected epithelial cells expressed GFP. As a result, a conventional monolayer of nasopharyngeal epithelial cells was susceptible to EBV infection, in spite of a low efficiency of 0.26%, based on the quantification of GFP positive infected epithelial cells (Figure 1c).

EBV Infects Pseudostratified Epithelia Formed by Primary Nasopharyngeal Epithelial Cells
Due to the fact that ALI cultures were generated from primary nasopharyngeal epithelial cells and reflect in vivo epithelial tissue more accurately than monolayer cultures, we expected that EBV would infect ALI cultures more efficiently than monolayer cultures. The differentiation state of day 28 cultures was assessed by transepithelial electrical resistance (TEER) and ciliary beat frequency (CBF) measurements, and only fully stratified epithelium with an average TEER above 4000 Ω × m 2 and CBF > 7 Hz were used for EBV infection. However, direct detection of GFP expression on EBV infected cells was difficult, even after the signals were enhanced by GFP antibody staining.
On the other hand, we could confirm EBV infection by RNAscope in situ hybridization (ISH), because the assay was sensitive enough to detect at single transcript level. The sensitivity and specificity of RNAscope ISH were validated by hybridizing EBV positive cell line or NPC tissue with an EBV-encoded small RNA1 (EBER1) probe, because EBERs (EBER1 and EBER2) are strongly expressed in EBV infected cells, and serve as sensitive targets for detecting EBV infection ( Figure 2) [12]. We used C666-1 cells for EBER1 detection, because C666-1 cell is a representative NPC cell line that expressed EBERs, EBNA1, BamHI-A right transcripts, latent membrane protein 1 (LMP1), and LMP 2A/B [13]. EBER1-staining of C666-1 cells showed both a diffuse pattern and a discrete speckled pattern ( Figure 2b). Similarly, cytokeratin positive NPC tumor cells showed both a diffuse staining pattern and a discrete speckled staining pattern (Figure 2d). Notably, most of the EBER-positive signals were localized in the nucleus, but not elsewhere in the cytoplasm or cilia (Figure 3a,c,e). Substantial numbers of EBER-positive signals could also be observed in EBV-positive NPC tumor cells (Supplemental Figure S1). We then examined whether EBV infects pseudostratified epithelia were formed by primary nasopharyngeal epithelial cells. Five days post-infection (d.p.i.), we performed hybridization with an EBER1 probe that detects EBV infections, and a BRLF1 probe that detects lytic infections (Figures 2 and 3). All cell types, including basal cells (p63-positive) (Figure 3a Notably, most of the EBER-positive signals were localized in the nucleus, but not elsewhere in the cytoplasm or cilia (Figure 3a,c,e). Substantial numbers of EBER-positive signals could also be observed in EBV-positive NPC tumor cells (Supplemental Figure S1).

Predominant EBV Infection of the Suprabasal Layer Cells
Either EBER1 or BRLF1 signals were counted in more than 200 EBV-infected nasopharyngeal epithelial cells (

Predominant EBV Infection of the Suprabasal Layer Cells
Either EBER1 or BRLF1 signals were counted in more than 200 EBV-infected nasopharyngeal epithelial cells (Figure 4). EBER positive cells are 65% with 4.5 puncta/cell and 25.5% with 2.6 puncta/cell in suprabasal layer cells and basal layer cells, respectively (Figure 4a,c). BRLF1 positive cells are 35.5% with 4.0 puncta/cell and 28.3% with 2.0 puncta/cell in suprabasal layer cells and basal layer cells, respectively (Figure 4b,d). EBER positive cells were more frequently observed in the suprabasal layer than in the basal layer (p < 0.05, Figure 4a,c).

EBV Infection Disrupts the Integrity of the Epithelium
Cilia beating plays a key role in airway-defense. CBF was initially reduced by the addition of EBV-infected cells and increased after removing these cells (Figure 5a). The TEER value provides information about the uniformity of the multi-layer cells and the integrity of the tight junctions between the polarized cells. When infected cells were added to the transwell, the TEER dropped sharply. On the other hand, TEER did not decrease with the addition of control cells (Figure 5b). The integrity of the epithelium was disrupted by EBV infection, but not by the addition of cells.

EBV Infection Disrupts the Integrity of the Epithelium
Cilia beating plays a key role in airway-defense. CBF was initially reduced by the addition of EBV-infected cells and increased after removing these cells (Figure 5a). The TEER value provides information about the uniformity of the multi-layer cells and the integrity of the tight junctions between the polarized cells. When infected cells were added to the transwell, the TEER dropped sharply. On the other hand, TEER did not decrease with the addition of control cells (Figure 5b). The integrity of the epithelium was disrupted by EBV infection, but not by the addition of cells.

Discussion
The association between EBV and epithelial malignancies, including NPC, has been known for decades. However, the life cycle of EBV during carcinogenesis is not well known, because of the lack of good in vitro models. Here, we successfully established an in vitro model of EBV infection using primary nasopharyngeal epithelial cells. The stratified multi-layer nasopharyngeal epithelial cells with several types of differentiated cells (Figure 1b (iv)) were infected with EBV ( Figure 1c).
Our model is physiologically relevant because of the following aspects. Firstly, we use authentic epithelial cells isolated from the nasopharynx where NPC originates. Secondly, two mechanisms of EBV infection is thought to happen in human body, one is cell-free infection mediated by saliva and the other is cell-to-cell infection mediated by EBV-infected or EBV-loaded B lymphocytes [14,15]. We showed that cell-to-cell infection is much more efficient than cell-free infection in the experimental animal model [9]. B lymphocyte-mediated infection must be important for epithelial EBV infection, because absence of a detectable latent EBV infection in persons with X-linked agammaglobulinemia who lack mature B lymphocytes [16]. Finally, in our previous report, 3 of 10 samples showed less than 10 EBER puncta/infected tumor cells [17]. The range of infection was similar to the current model (Supplemental Figure S1).
A relevant study by Temple et al. demonstrated in vitro EBV infection using primary oral keratinocytes raft cultures, polarized but not completely stratified cells. They showed that EBV infection resulted in viral production and spread throughout the culture, without establishing latent infection [18]. We showed that polarized stratified epithelium is more susceptible to EBV infection than monolayer cells (Figures 1c and 3). Moreover, we detected both latent and lytic infection signals, as evidenced by EBER1-positive and BRLF1-positive cells (Figure 3a,c,e). The high sensitivity and specificity of our probes have been validated in a dozen clinical samples [17]. In addition, because EBER expression is the gold standard to prove EBV infection, we established a model of EBV infection in nasopharyngeal epithelial cells.
Temple et al. used raft cultures of primary oral keratinocytes [18], that are polarized but not completely stratified cells. In contrast, our ALI cultures of nasopharyngeal epithelial cells form stratified epithelia and maintain the diversity of cells found in the airway epithelium, such as ciliated, muco-secretory, and basal cells (Figure 3, Supplementary Video S1).
Another relevant ALI culture by Caves et al. used HK1 cells derived from immortalized NPC cells, but not primary cells [8]. We and Caves et al. showed that a differentiation of latently EBVinfected cells triggered viral lytic replication. However, Caves et al. are different from ours in several ways. First, the HK1 cell was derived from a squamous carcinoma of the nasopharynx and was

Discussion
The association between EBV and epithelial malignancies, including NPC, has been known for decades. However, the life cycle of EBV during carcinogenesis is not well known, because of the lack of good in vitro models. Here, we successfully established an in vitro model of EBV infection using primary nasopharyngeal epithelial cells. The stratified multi-layer nasopharyngeal epithelial cells with several types of differentiated cells (Figure 1b (iv)) were infected with EBV (Figure 1c).
Our model is physiologically relevant because of the following aspects. Firstly, we use authentic epithelial cells isolated from the nasopharynx where NPC originates. Secondly, two mechanisms of EBV infection is thought to happen in human body, one is cell-free infection mediated by saliva and the other is cell-to-cell infection mediated by EBV-infected or EBV-loaded B lymphocytes [14,15]. We showed that cell-to-cell infection is much more efficient than cell-free infection in the experimental animal model [9]. B lymphocyte-mediated infection must be important for epithelial EBV infection, because absence of a detectable latent EBV infection in persons with X-linked agammaglobulinemia who lack mature B lymphocytes [16]. Finally, in our previous report, 3 of 10 samples showed less than 10 EBER puncta/infected tumor cells [17]. The range of infection was similar to the current model (Supplemental Figure S1).
A relevant study by Temple et al. demonstrated in vitro EBV infection using primary oral keratinocytes raft cultures, polarized but not completely stratified cells. They showed that EBV infection resulted in viral production and spread throughout the culture, without establishing latent infection [18]. We showed that polarized stratified epithelium is more susceptible to EBV infection than monolayer cells (Figures 1c and 3). Moreover, we detected both latent and lytic infection signals, as evidenced by EBER1-positive and BRLF1-positive cells (Figure 3a,c,e). The high sensitivity and specificity of our probes have been validated in a dozen clinical samples [17]. In addition, because EBER expression is the gold standard to prove EBV infection, we established a model of EBV infection in nasopharyngeal epithelial cells.
Temple et al. used raft cultures of primary oral keratinocytes [18], that are polarized but not completely stratified cells. In contrast, our ALI cultures of nasopharyngeal epithelial cells form stratified epithelia and maintain the diversity of cells found in the airway epithelium, such as ciliated, muco-secretory, and basal cells (Figure 3, Supplementary Video S1).
Another relevant ALI culture by Caves et al. used HK1 cells derived from immortalized NPC cells, but not primary cells [8]. We and Caves et al. showed that a differentiation of latently EBV-infected cells triggered viral lytic replication. However, Caves et al. are different from ours in several ways. First, the HK1 cell was derived from a squamous carcinoma of the nasopharynx and was negative for EBV [19]. Second, the HK1 cells were infected with EBV by Akata GFP-EBV cells prior to placing on the transwell. Third, Caves' model was a polarized model, but not a real stratified model. A high number of EBER1 and BRLF1 puncta were mostly observed in suprabasal layer (Figure 4). A cellular transcription factor, KLF4 was shown to induce differentiation-dependent lytic EBV infection in the mix culture of normal oral keratinocyte and Akata cells [20]. Though we did not investigate the detailed mechanisms involved, viral lytic reactivation could be induced by such kinds of cellular transcription factors.
Although we did not check the cytopathic effects at the individual cell level, we observed that EBV infection disrupted the integrity of the epithelium (Figure 5b). The cilia beating was temporarily compromised, but was eventually restored (Figure 5a). The disruption of epithelial integrity in EBV-infected cells is reflected in numerous reports, that show increased motility or anchorage independence in EBV-infected gastric, nasopharyngeal, and squamous cells [5,21,22]. Nevertheless, we need to determine whether our observation that EBV infection disrupts the epithelial integrity is unaffected by cytokines/exosomes released from activated Akata cells, by expanding the next experimental size.
Because EBV genomes in NPC tumor cells are monoclonal, tumor cells are clonal expansions of latently EBV infected nasopharyngeal cells [23]. However, the effort to elucidate mechanisms by which EBV transforms nasopharyngeal epithelium are hampered by the lack of in vitro transformation model. The current understanding is that low-grade preinvasive lesions become susceptible to latent EBV infection, because EBV can be detected in precancerous, but rarely in normal, pharyngeal epithelium [24,25]. We have previously established the culturing system of cells derived from normal nasopharynx of NPC patients [26]. Here, we showed that noncancerous stratified pharyngeal epithelia were susceptible to EBV infection, and their differentiation leads to lytic replication. Our new finding that EBV infection disrupted epithelial integrity may predict the carcinogenic process. ALI infection model utilizing either precancerous epithelial cells or epithelial cells that have received transforming stimuli could show a difference not only in EBV susceptibility, but also in epithelial integrity.
Taken together, we established the most relevant EBV infection model of stratified nasopharyngeal epithelia. This model can be used not only to examine the pathogenesis of pre-neoplastic EBV-infected cells, but also to develop anti-EBV therapy or early stage NPC treatment.

ALI Culture
We resuspended 1 × 10 5 primary cultured EBV negative cells in 100 µL of B-ALI TM growth medium (Lonza, Walkersville, MD, USA) and pipetted it onto 24-well transwell inserts (0.4 µm pore size, Thermo Fisher Scientific), with 400 µL of B-ALI TM growth medium, that was added to the basal chamber. On day 3 after seeding, the B-ALI TM growth medium from the apical and basal chambers was removed, and 400 µL of B-ALI TM differentiation medium was added to the basal chamber only. The differentiation medium of the basal chamber was changed every other day during the 4-week culture period [19][20][21]. The ALI cultures were observed with an inverted microscope (Olympus, Tokyo, Japan) at 400X magnification. TEER and CBF were measured [18].

TEER and CBF Measurements
CBF was recorded and analyzed automatically by using the Sisson-Ammons Video Analysis system (SAVA, Ammons Engineering, Clio, MI, USA) [29,30].
TEER was measured with EVOM2 instrument (World Precision Instruments, Sarasota, FL, USA). Before measurement, electrodes were equilibrated and sterilized according to the manufacturer's recommendations. Two hundred microliters of culture medium were added in the upper compartment of the cell culture system. The Ohmic resistance of a blank (culture insert without cells) was measured in parallel. To obtain the sample resistance, the blank value was subtracted from the total resistance of the sample.

Virus Production and Infection
Two Burkitt's lymphoma cell lines, one is EBV-negative Ramos cells and the other is GFP-tagged EBV infected Akata cells [10], were maintained in RPMI 1640 (Sigma-Aldrich), supplemented with 10% FBS. Akata cells were stimulated by 100 µg/mL of rabbit IgG-Fab fraction against human IgG (Jackson Immunoresearch, West Grove, PA, USA) for 24 h, to induce lytic EBV production [11]. Cell suspensions were centrifuged for 5 min at 300× g and washed once with phosphate buffered saline (PBS). The resultant Akata cells were resuspended in PBS to a concentration of 0.25 × 10 6 cells/100 µL, and then 200 µL was added to each well (24 well plate) of monolayer cells or 200 µL to the top chamber of the transwell (Figure 1a). EBV-negative Ramos cells were similarly treated with rabbit anti-human IgG-Fab and added into wells as a sham control.

Statistical Analysis
A Mann-Whitney test was used to analyze and to test whether there was a difference between two independent groups, which was expressed as mean ± standard deviation (SD). All experiments were repeated three or five times. A probability (p) value < 0.05 was considered statistically significant.

Conclusions
The present study showed that noncancerous stratified pharyngeal epithelia generated by ALI culture were susceptible to EBV infection, and cell differentiation leads to lytic viral replication. Moreover, EBV infection disrupted the integrity of the epithelium that may increase cell motility. The ALI infection model is useful for investigating the tumorigenic process of NPC and for developing anti-EBV chemotherapeutic agents to treat NPC patients.