Oral dysplasia and squamous cell carcinoma: correlation between increased expression of CD21, Epstein-Barr virus and CK19

Abstract

OBJECTIVES

Epstein-Barr virus is an orally transmitted human gammaherpesvirus that infects B lymphocytes and epithelial cells. Although most primary infections are asymptomatic, long term carriage of the virus can be associated with either lymphoid or epithelial malignancies. The association of EBV with oral squamous cell carcinomas is sporadic and it is uncertain if the virus is involved in initiation of the tumor or, possibly, in its progression. Complement receptor type 2, CR2 or CD21, is a receptor for the major attachment protein of EBV, which significantly enhances epithelial cell infection, but its expression on normal tissues is restricted to tonsil and adenoid epithelium. As cells become dysplastic they are reported to express higher levels of CK19. We sought to evaluate whether CD21 and CK19 expression change as oral epithelial cells outside Waldeyer’s ring become dysplastic.

MATERIALS AND METHODS

Epithelial cells were isolated by laser capture microdissection and levels of CD21, CK19 and EBV RNA were measured by quantitative reverse transcriptase PCR.

RESULTS

We report that expression of CD21 increases in frequency and intensity as oral epithelial cells become more dysplastic and that expression correlates with an increase in infection by EBV. Tumors or dysplastic lesions that carry EBV also generally express higher levels of CK19 than those that do not.

CONCLUSION

The findings suggest that dysplasia may make cells more susceptible to infection by EBV and that infection by the virus may alter the phenotype of the infected cell in a manner which could affect prognosis.

INTRODUCTION

Epstein-Barr virus (EBV) is an orally transmitted human gammaherpesvirus carried by more than 90% of the world’s population (reviewed in¹). Although primary infections are typically asymptomatic or accompanied by the benign, if temporarily debilitating, syndrome infectious mononucleosis, long term carriage of the virus can be associated with both epithelial and B cell malignancies. These reflect the predominant tropisms of the virus. The latent reservoir of virus is harbored in long lived memory B cells and amplification of virus by replication in epithelial cells is thought to contribute to replenishment of the reservoir and spread of virus to new hosts.^{2, 3, 4}

The principal cancer of the head and neck associated with EBV is nasopharyngeal carcinoma. All non-keratinizing forms of nasopharyngeal carcinoma carry EBV⁵ and typically express the most abundant non-coding EBER RNAs, EBNA1 transcripts and, to varying degrees, transcripts encoding the LMP proteins.⁶ EBV has also been associated with, though not causally linked to, oral squamous cell carcinoma (OSCC). Studies conflict with respect to the percent of cases that show evidence of EBV, only some have convincingly demonstrated that the virus is present in tumor cells rather than in infiltrating lymphocytes and, as EBV is part of the normal oral flora, it is not clear whether its presence simply reflects baseline levels that might be expected in cells in the oral cavity (reviewed in⁷). Several issues are nevertheless raised by the finding of EBV in OSCC at all. First, whether it is possible that there are phenotypes expressed by dysplastic or malignant cells that make them more vulnerable to infection and second, irrespective of whether the virus is causally related to the initiation of a malignancy, whether its presence potentially influences the subsequent behavior of an infected cell.

Infection of any cell by an enveloped virus such as EBV requires attachment of virus and fusion of its envelope with the plasma membrane or with the membrane of an endocytic vesicle. The EBV envelope contains eleven glycoprotein as many as five of which are potentially involved in epithelial cell entry. Attachment may occur as a result of interactions of at least two of them with cellular integrins,^{8, 9, 10, 11} but entry and infection is disproportionally enhanced if a cell expresses the complement receptor type 2, CR2 or CD21^{12, 13}, to which a third and most abundant virus glycoprotein, gp350, can bind.^{14, 15} CD21 is expressed on all B lymphocytes, where it is possibly the exclusive attachment receptor for virus, but it is not typically expressed in epithelial cells in the oral cavity, with the exception of epithelial cells in tonsils,¹⁶ and adenoids (data not shown) which are thought to be among the primary sites of EBV replication. To explore whether dysplastic or malignant cells in the oral cavity might have any increased susceptibility to EBV infection we therefore used laser capture microdissection of archived tissues to determine if mRNA transcripts corresponding to the domain of CD21 to which EBV attaches are expressed in dysplastic or malignant cells. The identity of epithelial cells was confirmed morphologically and by expression of mRNA encoding the cytokeratin CK19. The exclusion of contaminating B cells was determined by lack of expression of mRNA for CD20. We report here not only that expression of CD21 is increased in dysplastic cells, but also that the presence of EBV is associated with increased levels of expression of CK19.

MATERIALS AND METHODS

Tissue specimens

Tissue blocks collected between 1999 and 2011 were obtained from patients treated at Louisiana State University Health Sciences Center-Shreveport after receiving approval from the institutional review board. Seventy-eight mucosal biopsies from the oral cavity and oropharynx were examined. They included 23 samples of normal oral squamous epithelium (NOSE), including 5 from tonsil, 12 from base of tongue and 6 from normal margins of dysplasia. The remaining biopsies were either squamous dysplasia (29 cases) or oral squamous cell carcinoma (OSCC, 26 cases). The cases with squamous dysplasia were further classified as mild dysplasia (MLD, 8 cases), moderate dysplasia (MOD, 8 cases) and severe dysplasia (SED, 13 cases). The cases with OSCC were further classified as: well differentiated (6 cases), moderately differentiated (12 cases) and poorly differentiated (8 cases). All pathology specimens were routinely fixed in 10% neutral formalin and embedded in paraffin. Three biopsies of oral hairy leukoplasia (OHL) embedded in OCT were obtained from the AIDS and Cancer Specimen resource of the NCI and two were a gift of Dr. Sharof Tugizov (University of California, San Francisco). Specimens were sectioned to a thickness of 4 μm for laser capture microdissection and immunohistochemistry.

Tissue processing and Laser capture microdissection (LCM)

Sections were stained with hematoxylin and eosin (H&E), dewaxed in xylene, dehydrated in ethanol, air-dried and stored in desiccant until used for LCM. As previously described,¹⁶ 3,000 to 4,000 epithelial cells, identified by morphology, were captured by a PixCell IIe LCM system (Arcturus Engineering, Inc., Mountain View, CA, USA). CapSure HS LCM caps (Arcturus) were placed over the sections, laser spot size, power, and pulse duration were adjusted 7.5 μm in diameter, 55–60 mW, and 1.5–1.8 ms, respectively. Epithelial cells were captured from whole layers of epithelium in the normal margin of dysplasia, all dysplasia, OHL and OSCC sections. In some tonsil and base of tongue sections epithelial cells were captured above or close to the basal layer to avoid lymphocyte contamination. Each section analyzed was sampled independently two to four times.

Reverse transcription and Quantitative real-time PCR (QPCR)

Total RNA from captured cells was extracted, purified, DNase treated, and reverse transcribed using the Paradise^Plus Whole Transcript Reverse Transcription (WT-RT) Reagent System (MDS, Analytical Technologies) according to the manufacturer’s instructions. A human formalin-fixed universal reference RNA, included in the Reagent System, was reverse transcribed in parallel as a positive control. Quantitative real-time QPCR was performed using an Applied Biosystems 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) with conditions as previously specified.¹⁶ The target gene probes were labeled by FAM (6-carboxy-fluorescein) (Integrated DNA Technologies, Coralville, IA, USA) and a house keeping gene, GAPDH probe was labeled by VIC (glyceraldehyde- 3-phosphate dehydrogenase) (Applied Biosystems). The sequences of primers and probes of target and housekeeping genes were as previously reported.¹⁶ Primers were optimized for QPCR with GAPDH as a control gene. When the optimal primer concentration produced a linear response to input cDNA concentration, RNA samples were analyzed in triplicate for each tested transcript. To normalize the expression levels (ΔCt), the threshold cycle (Ct) for each transcript was subtracted from the Ct of the more abundantly expressed control gene (GAPDH). This value was then used to calculate the 2^−ΔΔCt to compare expression patterns between various tissue samples. A significant change in expression was defined as greater than a fold increase or decrease.

Cell culture and RNA extraction

Akata Burkitt’s lymphoma B cells, which have lost the EBV genome,¹⁷ were grown in RPMI (Sigma), AGS cells (ATCC), CD21-negative gastric carcinoma cells were grown in Ham’s F12 medium (Invitrogen). All media were supplemented with 10% heat inactivated fetal bovine serum (Gibco / BRL). For QPCR analysis, total RNA from cells was isolated and purified using RNA STAT-60 reagent according to the manufacturer instructions (TEL-TEST, INC.). For cDNA synthesis, 5 μg RNA was reversed transcribed with Moloney murine leukemia virus reverse transcriptase (Invitrogen) using random hexamer primers.

Comparison of relative amounts of CK19 mRNA

The relative amounts of target gene CK19 mRNA were calculated by the comparative threshold cycles (Ct) method.^{16, 18} For the ΔΔCt calculation to be valid, amplification efficiencies of the target and reference GAPDH must be approximately equal.¹⁹ ΔCt represents the difference in Ct values between target genes and the GAPDH reference gene from the lesion at the same condition points for each sample. ΔΔCt is described as target gene expression in different tissues and cell lines minus target gene expression in the arbitrary constant. As previously reported,¹⁶ to assess if target and reference amplicons had the same efficiency, serial dilutions of cDNA from AGS cells were amplified by QPCR using CK19 and GAPDH primers. The ΔCt numbers of CK19 and GAPDH were calculated for each cDNA dilution and the absolute value of the slope of the plot of the log cDNA dilution vs. ΔCt was close to zero indicating that the efficiencies of the target and reference genes were similar (Figure 1). The numerical value for ΔΔCt was then used in the 2^−ΔΔCt calculation. The final result represents the differential expression of target CK19 gene in different tissues relative to that in the tissues that were, in each case, used to establish an arbitrary constant of 1.

Figure 1.

Validation of the 2^−ΔΔCt method for determination of relative levels of expression of CK19. The target gene, CK19, and the internal control, GAPDH, were amplified from serial dilutions of cDNA of AGS epithelial cells and the ΔCt (Ct_CK19 – Ct_GAPDH) was calculated for each. The slope of the curve fitted using least squares linear regression analysis was 0.0006.

Immunohistochemistry

Immunohistochemistry was performed on formalin-fixed, paraffin-embedded tissue blocks. Four-micrometer-thick sections were dewaxed in xylene, rehydrated in ethanol and the immunohistochemical staining done in a Ventana Banchmaker XT autostainer (Ventana ULTRA system, Tucson, AZ, USA) using standard peroxidase immunohistochemistry techniques, heat-induced epitope retrieval buffer and primary antibody against CK19 (pre diluted A53-B/A2.26, Ventana). Detection of the staining was achieved by an enzyme-conjugated polymer complex (HRP multimer kit, Ventana). Appropriate positive and negative controls were included.

Statistical Analysis

Statistical analysis was done with Student’s t-test. Results were expressed as means ± SD. A difference of p <0.05 was considered statistically significant.

RESULTS

Compared with normal margin epithelial tissue, which was negative for CD21 transcripts, we found that CD21 mRNA was expressed in 12.5% of mild grade, 37.5% of moderate grade and 69.2% of severe grade dysplastic epithelial specimens (Table 1). Although Ct values in the formalin-fixed paraffin-embedded tissues were high, analysis of the melting curves revealed identical melting temperatures as for those of tissues, such as fresh Akata B cells, which had much lower Ct values. Relative to the levels of CD21 in the Akata B cell line, which were given the arbitrary value of 1, CD21 transcript levels in mild dysplasia averaged 0.35, in moderate dysplasia 1.77, and in severe dysplasia 4.99. The frequency and levels of expression in OSCC epithelial tissues did not differ from those in severe grade dysplastic epithelium (data not shown).

Table 1.

Expression of CD21 and EBER1 mRNA in oral epithelial dysplasia

Tissue source	GAPDH (Ct)	CD21 (Ct)	ΔCt (CR2)	2−ΔΔCt	CD21 expression (%)	EBER positive (%)
Akata B cell line	18.88±0.18	27.47±0.14	8.59	1	100%	-
Normal margin (n=6)	29.26±1.05	-	-	-	0/6 (0%)	0/6 (0%)
Mild dysplasia (n=8)	27.57±0.12	37.76±0.51	10.09	0.35	1/8 (12.5%)	5/8 (62.5%)
Moderate dysplasia (n=8)	29.92±1.70	37.69±0.19	7.77	1.77	3/8 (37.5%)	6/8 (75%)
Severe dysplasia (n=13)	29.53±1.57	35.80±1.51	6.27	4.99	9/13 (69.2%)	5/13 (38.5%)

Since CD21 is a major attachment receptor for EBV, expressed in mature B cells and normal tonsil epithelial cells but not in normal oral tongue, uvula, or buccal mucosa epithelium,¹⁶ we assessed the prevalence of EBV in dysplastic cells by measuring expression of EBER1, one of the EBV small nuclear RNAs. EBER1 transcripts were not detected in normal margin epithelium, but were found in 5/8 mild, 6/8 moderate and 5/13 severe grade dysplasia tissues. There was a strong correlation between the presence of CD21 and EBER1 transcripts (Figure 2). As the expression levels of CD21 increased in those dysplastic cells that expressed any detectable levels of CD21, the likelihood of infection with EBV increased (Figure 3), although the correlation was not as strong.

Figure 2.

Percent of normal margins or epithelial tissues with different grades of dysplasia that express both CD21 and EBER1 transcripts.

Figure 3.

Levels of expression of CD21 mRNA in EBER1 positive epithelial tissues with different grades of dysplasia.

Expression of CK19 was used as a marker of epithelial cells. CK19 transcripts were detected in all samples of normal tonsil epithelium or normal base of tongue epithelium from patients with lymph node metastases of unknown origin, although expression levels were low. Relative to the CK19 expression in these tissues, set at an arbitrary value of 1, CK19 transcript levels (Table 2) were increased in all grades of dysplasia, in all but well differentiated OSCC and in oral hairy leukoplakia, an epithelial lesion caused by lytic replication of EBV which carries a very high virus load.²⁰ The high expression of CK19 in oral precancerous lesions and OSCC was consistent with other studies, which have reported upregulation of CK19 in head and neck dysplasias and cancers ^{21, 22}, but the highest levels were clearly seen in the EBV infected OHL tissue.

Table 2.

Comparison of CK19 mRNA in normal oral epithelium with expression in OHL, all grades of dysplasia and OSCC

Tissue source	Case (n)	GAPDH (Ct)	CK19 (Ct)	ΔCt (CK19-GAPDH)	P-value	2−ΔΔCt
Normal epithelium	23	28.85±1.93	36.38±1.40	7.91±1.64		1
OHL	5	28.31±3.12	33.62±2.34	5.31±1.26	0.0001	6.06
Mild dysplasia	7	28.99±2.68	36.30±2.46	7.28±1.20	0.355	1.55
Moderate dysplasia	6	30.09±1.81	36.97±1.50	6.93±1.29	0.186	1.97
Severe dysplasia	7	29.45±1.08	36.95±2.04	7.50±1.80	0.579	1.33
Well differentiated OSCC	6	27.70±3.38	36.70±1.96	8.83±3.92	0.598	0.53
Moderately differentiated OSCC	12	26.15±2.57	33.04±2.18	6.10±2.03	0.007	3.51
Poorly differentiated OSCC	8	29.14±3.71	35.27±3.24	6.13±1.64	0.013	3.43

To evaluate further whether EBV infection overall was associated with increased expression of CK19, EBV positive and EBV negative cases were separated and CK19 transcript levels were re-calculated with the EBV negative normal epithelium (normal tonsil, normal margins of dysplasia and normal base of tongue) set at the arbitrary value of 1 (Table 3). Compared with the EBV negative normal epithelium, CK19 expression was almost twice as high in EBV positive normal epithelium, but the difference did not reach statistical significance. In contrast, CK19 transcript levels were significantly higher in EBV positive, but not in EBV negative dysplastic epithelium. Both EBV positive and negative OSCC expressed significantly higher levels of CK19 transcripts than that of normal epithelium, but again, expression levels in EBV positive tissues were higher than in EBV negative tissues.

Table 3.

Comparision of CK19 mRNA expression in EBV−ve and EBV+ve tissues.

Tissue source	GAPDH (Ct)	CK19 (Ct)	ΔCt (CK19-GAPDH)	P-value	2−ΔΔCt
Normal epithelium
EBV−ve (n=16)	29.33±1.75	37.53±1.49	8.20±1.15		1
EBV+ve (n=7)	28.36±2.10	35.23±1.31	7.23±2.41	0.343	1.96
Dysplasia
EBV−ve (n=9)	29.54±1.20	37.43±1.22	7.96±1.00	0.593	1.18
EBV+ve (n=11)	29.47±2.24	36.11±2.46	6.67±1.46	0.005	2.89
OSCC
EBV−ve (n=10)	27.55±3.07	34.57±2.89	6.83±1.70	0.021	2.58
EBV+ve (n=11)	27.81±3.06	33.23±2.14	5.42±1.72	0.0003	6.87

We also compared CK19 transcript levels in EBV positive and EBV negative cases within the same groups of dysplasia and OSCC. In each group, CK19 mRNA expression in EBV negative cases was now set as an arbitrary value of 1. CK19 levels were 1.96 times higher in EBV positive normal epithelium than in EBV negative normal epithelium, 2.45 times higher in EBV positive dysplasia and 2.66 times higher in EBV positive OSCC (Figure 4).

Figure 4.

CK19 mRNA expression levels in normal epithelium, dysplastic epithelium and OSCC samples grouped according to whether tissues do or do not express EBV EBER1 mRNA.

To confirm that increased CK19 mRNA expression levels were reflected in protein expression levels, CK19 was examined by immunohistochemistry in samples of normal mucosal epithelium, dysplasia and OSCC. CK19 staining was weak in the basal layer of normal epithelium (Figure 5, a and b). There was an increase in staining intensity in dysplasia and OSCC (Figure 5, c–f), which was greatest in tissues that were EBV positive (Figure 5, d and f).

Figure 5.

Immunohistochemical staining for CK19 expression in normal oral tissues, dysplastic epithelium and OSCC that are positive (EBV+) or negative (EBV−) for expression of EBV EBER1 mRNA.

DISCUSSION

The finding that CD21 mRNA was expressed in mild, moderate and severe grade dysplasia with a frequency that increased with grade, was unexpected. Confirmation of simultaneous expression of protein was not possible since at least one monoclonal antibody to CD21 cross-reacts with an unrelated epithelial cell protein,²³ only a subset of antibodies that recognize CD21 on B cells is reactive with all epithelial cells and reports from different groups using the antibodies are not consistent.^{24, 25, 26, 27} Even subclones of a single epithelial cell line were found to be recognized differently.¹⁶ However, the good correlation between the expression of CD21 mRNA and infection with EBV suggests that CD21 protein itself is expressed. Such findings are consistent with those in vitro which demonstrated that infection can occur in the absence of CD21, but is more likely in its presence.^{9, 13} They raise the interesting possibility that, relative to normal tissue, tissues that have taken the first potential steps towards malignancy become more susceptible to infection with a virus which is known to express proteins that can impact cell growth and behavior.

The observation that CK19 expression at both the mRNA and protein level increases in all grades of dysplasia was not unexpected. Several studies have noted that expression of CK19, which has been proposed to be a stem cell marker,^{28, 29, 30} in suprabasal layers of epithelium is associated both with premalignant changes and OSCC. Moderately and poorly differentiated OSCC express more CK19 than well-differentiated OSCC which in turn correlates with poorer prognosis.^{21, 22, 31} However, the finding that those lesions that carried EBV expressed even higher levels of CK19 relative to the EBV negative tissues with the same grade of dysplasia or malignant differentiation was of new interest.

The variable presence of EBV in head and neck tumors complicates a single cohesive interpretation of its role in tumor growth. However, one potential threat of infection is that it provides an additional growth advantage to a premalignant cell. This could perhaps be a transient event, instigating epigenetic changes, or one mediated by the facilitation of persistence of latent virus in the cell. There is a rich and abundant literature documenting the potential roles that EBV latent proteins and transcripts may play in altering cell behavior^{1, 32} and a recent report has implicated the EBV latency protein LMP1 in increased expression of CK19 in nasopharyngeal epithelial cells and a concomitant increase in proliferation rates.³³ The fact that EBV-positive tumors carry episomes with identical numbers of terminal repeats has provided strong support for the contention that, in cancers consistently associated with EBV, such as nasopharyngeal carcinoma, virus infection provides the critical, initiating, oncogenic insult.³⁴ Infection of a cell is followed by circularization of the virus genome as a result of random recombination of variably reiterated terminal repeat sequences. The number of terminal repeats remaining in the retained episome is thus particular to each infecting event. Infection of a preexisting tumor would produce cells carrying episomes with variable and not identical numbers of repeats. However, there is also evidence for evolution towards a predominant episomal form as one cell gains a growth advantage over its peers.³⁵ Pre-invasive NPC has been shown to contain monoclonal EBV termini, supporting an early role for EBV in tumorigenesis,³⁶ but other studies find the genome in only a small portion of tumor cells in early NPC and progressively increased numbers of cells at more advanced stages.^{37, 38} EBV infection has been detected in all high grade dysplastic lesions and NPC, but not in low grade lesions or nasopharyngeal tissue of normal adults.³⁹

The finding of a virus in a malignancy may have some prognostic value. In recent years there has been a dramatic increase in the number of carcinomas of the tonsil and base of tongue associated with infection with human papillomavirus.^{40, 41} The tumors do, however, have a better prognosis than those more typically seen in heavy smokers and drinkers which are virus negative. Whether this reflects an increased immunogenicity or less profound abnormality is not clear. In view of the sporadic, but significant association of EBV with tumors of the oral cavity it may be equally appropriate to evaluate EBV status and its impact on long term outcome.