The involvement of promoter methylation and DNA methyltransferase-1 in the regulation of EpCAM expression in oral squamous cell carcinoma

Introduction 

The epithelial cell adhesion molecule (EpCAM) is a 40kDa epithelial transmembrane glycoprotein encoded by the GA733-2 gene located on the long arm of chromatin 4.1, 2 This molecule is a homophilic and Ca2+ independent adhesion molecule that appears to be overexpressed by the great majority of human epithelial malignancies.2, 3 However, the biologic function of EpCAM in tumors is variable. It has been proposed as an oncogenic signaling protein and was associated with development of malignancy in certain types of tumors.3, 4 In contrast, other studies suggested that EpCAM could have a tumor suppressor effect in some malignancies.5, 6, 7 Previously, the expression of EpCAM and its clinical significance in OSCC was studied.8, 9, 10 However, the results were not consistent. Given the somewhat controversial results of EpCAM function in tumors, the biologic function and clinical relevance of EpCAM in OSCC needs to be systematically studied. Furthermore, regulation of EpCAM expression in OSCC is still unknown.

Recent evidence has suggested that EpCAM expression could be dynamic during tumor progression. For example, Rao et al. showed that EpCAM expression was relatively lower in circulating tumor cells compared to primary and metastatic tissues11 Jojovic et al. demonstrated EpCAM expression could be transiently lost during development of metastases.12 The expression of some genes can be dynamically regulated by reversible epigenetic events.13 Aberrant DNA methylation is one of the most common epigenetic mechanisms regulating gene expression in human cancers.14 However, information regarding methylation alterations of EpCAM in OSCC is not available.

The methylation of mammalian genomic DNA is catalyzed by DNA methyltransferases (DNMTs), which play a special role in the regulation of chromatin remodeling and gene expression. The mammalian DNMTs are DNMT1, DNMT3A and DNMT3B, which together with accessory proteins such as DNMT3L are responsible for methylation pattern acquisition during gametogenesis, embryogenesis and somatic tissue development.15 DNMT1 is the most abundant DNMT targeted to replication foci and has a 10–40-fold preference for hemi-methylated DNA substrates.16 It seems to be the main enzyme responsible for copying the methylation pattern after each round of DNA replication.17 Recent observations suggest a complicated network of connections between DNMT1 and several cellular proteins that could mediate methylation-dependent and some non-methylation-dependent functions of DNMT1 as cells progress to malignancy. For example, it has been shown that DNMTs are overexpressed in tumorigenic cells18 and upregulation of DNMT1 in vivo occurs in several different tumor types.19, 20, 21, 22 However, its function in OSCC and involvement in EpCAM methylation is not clear.

In the present study, we therefore undertook an investigation to determine the role of EpCAM and DNMT1 in OSCC and to study the possible involvement of DNA methylation in the regulation of EpCAM. We also investigated the correlations among EpCAM methylation, EpCAM and DNMT1 expression, and clinicopathologic parameters.

Materials and methods 

Results 

EpCAM protein expression 

Immunostaining of EpCAM was done in all 112 OSCC cases and in 43 normal tissues from the same patients. Normal and hyperkeratotic mucosal tissues showed negative to faint staining for EpCAM. A distinct margin of EpCAM staining was observed between normal and dysplastic epithelium (Fig. 1A). An obvious increase in staining was observed in most samples of dysplastic epithelium. In our series of OSCC cases, EpCAM staining was heterogeneous, with tumor cells demonstrating a diffuse pattern and an increase in membranous EpCAM staining in comparison with their normal paired control samples (Fig. 1B). In some tumors, EpCAM expression was heterogeneous in that staining was positive in some cells and negative in others (Fig. 1C). In some invasive tumors, we found losses of EpCAM expression in tumors disseminated within vessels or in metastatic lymph nodes (Fig. 1D–E). Lower EpCAM immunostaining was observed in 55/112 tumor samples (49%), in comparison with 57/112 (51%) tumor samples that exhibited higher EpCAM expression. Staining results, which indicates EpCAM expression, in relation clinicopathologic parameters were shown in Table 1. There were no significant correlations between EpCAM expression and clinicopathologic variables for OSCC tissues (Table 1).

View Large Image Download to PowerPoint

Figure 1. EpCAM expression in the initiation (A,B,C) and progression (C,D,E) of OSCC. (A) No EpCAM expression in normal tissue (right) and increased expression in dysplasic tissue (left). (B) Intense and homogeneous expression in tumor cells of OSCC. (C and D) Heterogeneous staining pattern with positively and negatively staining cancer cells. (E) Disseminated tumor cells (arrow) in the vessel showed negative EpCAM expression as compared to neighboring tumor cells which showed positive staining. (F) Loss of EpCAM expression was noted in tumor cells that had invaded the lymph node. (A, B, C were taken from the same slide. D, E, F were taken from the same patient).

Table 1. Patients’ clinicopathologic features in association with EpCAM methylation status and immunohistochemical expression of EpCAM and DNMT1

	Clinicopathologic features	EpCAM expression		EpCAM methylation		DNMT1 expression
		Low (n=55)	High (n=57)	U (n=35)	M (n=37)	Low (n=59)	High (n=53)
	Gender
	Male	48	45	29	30	50	43
	Female	7	12	6	7	9	10
	Size
	⩽4.0cm	33	33	21	22	44	22⁎
	>4.0cm	22	24	14	15	15	31
	LN involvement
	No	33	31	17	23	38	26
	Yes	22	26	18	14	21	27
	Differentiation
	Well	18	23	14	15	28	13⁎⁎
	Moderate	23	20	12	14	22	21
	Poor	14	14	9	8	9	19
	Staging
	I	7	7	4	4	13	1⁎⁎⁎
	II	17	19	12	15	24	12
	III	10	13	6	4	10	13
	IV	21	18	13	14	12	27

LN, lymph node; U, unmethylation; M, methylation.

⁎ p<0.0001. ⁎⁎ p=0.012. ⁎⁎⁎ p<0.0001.

EpCAM methylation status in OSCC 

To determine whether DNA methylation is involved in EpCAM expression, we examined the methylation status of EpCAM promoter regions in these tumors. Tumor cells were isolated by microdissection, and DNA was extracted and modified using bisulfite. Then methylation-specific polymerase chain reaction (MSP) assays were done to determine the methylation status of EpCAM. Data on DNA extraction and bisulfite modification were available for 98 cases, which were further subjected to MSP analysis. Due to the quality and quantity of extracted DNA, a small portion of cases cannot be further examined. The missingness was random and unrelated to the value of any other clinicopathologic variables. Thus, a total of 72 samples were detected and analyzed. A representative result is shown in Fig. 2. Methylated alleles were present in 37 out of 72 samples (51%). Methylation status of the EpCAM promoter in relation to clinicopathologic features of OSCC patients is presented in Table 1. No significant correlations were found. We next determined whether methylation of the EpCAM promoter is associated with EpCAM gene expression. We analyzed the relationship between expression of EpCAM and promoter methylation in these 72 cases. In the 38 cases with high EpCAM expression, unmethylated and methylated promoters were detected in 24 and 14 cases, respectively. In the 34 cases with low EpCAM expression, unmethylated and methylated promoters were detected in 11 and 23 cases, respectively. There was a significant association between EpCAM methylation and low EpCAM expression in OSCC (Table 2, p=0.008).

View Large Image Download to PowerPoint

Figure 2. Representative cases of EpCAM expression and promoter methylation status in OSCC. (A) Promoter methylation status of EpCAM in OSCC patients. (B) In tumors with high EpCAM expression, the methylation status of the EpCAM promoter showed predominantly an unmethylated status. (C) In cases of heterogeneous expression, MSP assays showed mixed patterns with both methylated and unmethylated alleles. (D) In low EpCAM expression cases, methylation status showed that methylated alleles were predominant. (Original magnification 200× in B, C, D).

Table 2. Methylation status of EpCAM promoter in case of low and high EpCAM Expression

	EpCAM expression	EpCAM methylation
		U	M	p value
	Low (n=34)	11	23	0.008
	High (n=38)	24	14
	Total (n=72)	35	37

U, unmethylation; M, methylation.

DNMT1 protein expression and association with EpCAM expression and methylation status 

Immunoreactivity for DNMT1 was detected in nuclei, but not in cytoplasm or cell membranes, of cells in the parabasal layer of epithelia, lymphocytes, and cancer cells (Fig. 3). In cancer tissues, 53 of the 112 cases (47%) showed higher DNMT1 expression. Correlations between the presence of nuclear immunoreactivity for DNMT1 and clinicopathologic features of OSCC are shown in Table 1. DNMT1 protein expression was significantly associated with histologic differentiation (p=0.012), tumor size (p<0.0001) and clinical stage (p<0.0001). When we analyzed DNMT1 expression in relation to EpCAM expression and methylation status, there were no significant correlations between DNMT1 protein expression and EpCAM immunostaining or EpCAM methylation (Table 3).

View Large Image Download to PowerPoint

Figure 3. DNMT1 expression in normal and tumor tissues. (A) DNMT1 positive staining was occasionally observed in the basal layer of normal epithelium. Representative high (B) and low (C) expression of DNMT1 in OSCC. (Original magnification 200× in A, B, C).

Table 3. Relationships between DNMT1 expression and EpCAM expression and methylation status of EpCAM promoter

		EpCAM expression			EpCAM methylation
		Low	High	p value	U	M	p value
	DNMT1
	Low	14	20	0.228	15	10	0.144
	High	23	21		14	19
	Total	37	41		29	29

U, unmethylation; M, methylation.

Discussion 

A large number of studies on EpCAM’s role in cancer have yielded variable results. Present study had examined (1) the relevance of EpCAM expression in the initiation and progression of OSCC and (2) possible epigenetic mechanisms of DNA methylation in the regulation of the EpCAM gene expression. Our study showed that increases in EpCAM expression could be involved in the initiation of OSCC. This was demonstrated by our finding of a significant increase in EpCAM expression in dysplastic epithelium and cancer tissues compared to normal epithelium. In agreement with previous reports that expression of EpCAM increases from baseline levels in normal oral epithelium to that of high levels in dysplasia and OSCC tissues,8 this observation further strengthened the hypothesis that de novo expression of EpCAM is an early event in tumorogenesis. Thus, EpCAM expression is an early step in the malignant transformation of oral epithelium, and could be used as a marker for diagnostic purposes.

Although we did not find a significant association of EpCAM expression with clinicopathologic variables in OSCC, EpCAM expression seems to be flexibly regulated in cancer cells, which was demonstrated by its strong expression in primary tumors and weak or negative expression in disseminated tumor cells or lymph node metastases. In support of this notion, previous reports found that EpCAM expression in circulating and metastatic tumor cells is relatively lower compared to primary tumors from the same patient.11, 23, 24 These observations raise the possibility that in some subpopulations of tumor cells, loss of EpCAM is associated with malignancy and aggressive behavior. Actually, the expectation that EpCAM, like other adhesion molecules, provides invasion-suppressor properties to epithelia through cell-cell aggregation has been demonstrated. Normally non-adhesive cell lines have been induced to aggregate through transfection and overexpression of EpCAM and, in addition, have been shown to have reduced mobility and fewer invasive behaviors.25 Furthermore, negative or limited EpCAM expression in primary laryngeal carcinoma has been linked to the presence of nodal metastases and to a poor prognosis.26 Based on our findings and the functions proposed for EpCAM, we postulate that this adhesion molecule may play distinct and different roles at various stages in the evolution and progression of OSCC. These results may add new information for the application of EpCAM in diagnosis, prevention, and treatment of OSCC.

That EpCAM expression is heterogeneous in primary tumors and transitional loss occurs in disseminated and lymph node metastases, suggests that regulation of EpCAM expression involves a reversible mechanism rather than irreversible genetic alterations. Cytosine methylation within the context of CpG dinucleotides in the genome is a molecular mechanism that causes epigenetic changes in chromatin structure and leads to transcriptional silencing of genes in many human cancers. This epigenetic alteration is heritable but does not alter the nucleotide sequence, making the modification potentially reversible. Previously, our work in lung cancer had shown that DNA methylation coupled with alterations in chromatin structure and transcription factor activity could be attributed to the reversible regulation of EpCAM.6 Spizzo et al. also found the promoter and exon 1 was highly methylated in an EpCAM negative breast cancer cell line compared to few methylated CpG sites in EpCAM positive cell lines, which implicated that methylation of EpCAM is, in part, responsible for EpCAM expression in some cancer cells.27 In the present study, our results show that promoter methylation is significantly correlated with the expression of EpCAM in OSCC, in which a high proportion of methylated alleles are detected in low EpCAM expressing tumors. These findings therefore suggest that an epigenetic mechanism via methylation of the EpCAM promoter might be responsible for dynamic regulation of this gene in OSCC.

Although the importance of CpG island methylation in cancer is now becoming apparent, the mechanisms that lead to this phenomenon in tumors are still unknown. The most widely investigated potential cause is the increased expression in tumors of one or more of the DNMT enzymes. DNMT1 was the first enzyme to be isolated as a mammalian DNA methyltransferase.28 Hyperactivation of DNMTs in parallel with aberrant patterns of DNA methylation was reported in many tumor types.22, 29 However, in other studies investigating CpG island methylation status failed to confirm this relationship.30, 31, 32, 33 Here, we investigated the association of DNMT1 expression and promoter methylation status of EpCAM. Nonetheless, the expression level of DNMT1 did not correlate with the methylation status of the EpCAM gene in OSCC. This could be explained in part by the fact that aberrant promoter hypermethylation is caused by gene-specific regulation involving DNMTs and its interaction proteins. Actually, DNMT1 can interact with a number of different proteins that could be involved in the intracellular delivery and regulation of the catalytic activity of DNMT134, 35, 36, 37 as well as access to its DNA target sites in chromatin.17, 38 Thus, DNMT1 overexpression alone may not explain gene-specific DNA hypermethylation in cancers. Certain unknown components of the DNA methylation machinery may potently target DNMT1 to gene-specific methylation in OSCC.

We showed that the expression of nuclear DNMT1 immunoreactivity correlates significantly with histologic differentiation and tumor malignancy in OSCC. Protein interactions with DNMT1 not only regulated its methylation-dependent but were also involved in methylation-independent function.32 For example, inhibition of DNMT activity in the human lung cancer cell line A549 by a DNMT antagonist or by antisense oligonucleotides induces a rapid upregulation of cyclin-dependent kinase inhibitor p21, which does not involve the demethylation of the p21 promoter. These results suggest that the DNMT1 protein might have a direct and immediate effect on the state of cell growth and transformation. Taken together, our results indicated DNMT1 expression may involve in the histogenesis and aggressiveness of OSCC and protein expression analysis of DNMT1 may be a biologic marker of tumor progression for OSCC patients.

In conclusion, we demonstrated that promoter methylation is a major event involved in the loss of EpCAM expression in OSCC. Our study also indicated that upregulation of DNMT1 is involved in the progression of OSCC. Although expression of DNMT1 is elevated, additional mechanisms may be required for specifying specific cystosine residues for methylation and that other factors are therefore involved in gene silencing of EpCAM. Further studies on the cooperation between DNMT1 and other components of the DNA methylation machinery in tissue specimens may extend our understanding of the basis of regional DNA methylation during OSCC carcinogenesis.

Conflict of interest statement 

None declared.