Promoter hypermethylation of RAR- β tumor suppressor gene in gastric carcinoma: Association with histological type and clinical outcomes

Abstract

BACKGROUND:

While gastric cancer is a common cancer in the world and Iran, its molecular mechanisms are not fully understood as yet. Epigenetic modifications can lead to alteration of gene expression and development of tumorigenesis mechanisms.

METHODS:

To clarify the difference in DNA methylation pattern of histological types in gastric carcinoma, CpG islands in the promoters of retinoic acid receptor $\beta$ gene (RAR- $\beta$ ) was studied using methylation-specific PCR.

RESULTS:

In gastric cancer tissues, hypermethylation frequency of RAR- $\beta$ gene was respectively 61 and 33% for diffuse and intestinal type. In diffuse type, hypermethylation of RAR- $\beta$ has been significantly associated with invasion ( $P=$ 0.007), differentiation ( $P=$ 0.033) and location ( $P=$ 0.012) of the tumor. However, hypermethylation of RAR- $\beta$ correlated only with tumor size ( $P=$ 0.029) in intestinal type. For adjacent non-tumor samples, hypermethylation of RAR- $\beta$ was not detected and there was no significant association between age of diagnosis and hypermethylation of RAR- $\beta$ in both types of gastric cancer.

CONCLUSIONS:

These results support previous findings denoting a distinct profile of promoter hypermethylation status in the development of the intestinal and diffuse type of gastric carcinoma and the process of the tumorigenesis in these subtypes of gastric cancer is different from each other.

1. Background

Gastric cancer is the second most common cancer and third most common cause of cancer-related death in the world [1].

Successive histopathological alterations in the gastric epithelium in accompany with some genetic and epigenetic aberrations could ultimately result in gastric cancer [2].

Based on Lauren’s criteria, intestinal type and diffuse type adenocarcinoma are the two major histologic subtypes of GC [3] which differ in their epidemiology, etiology, and clinical features. Histologically significant differences between these subtypes of gastric carcinomas [4] may be attributed to the epigenetic and genetic changes in these cells [5].

One of the well-recognized epigenetic mechanisms is DNA methylation at the promoter regions of the genes [6]. Usually, the result of DNA hypermethylation is expression silencing of the genes controlling and suppress cell growth runaway. Recent studies revealed the potential role of DNA hypermethylation in both normal and cancer cell growth and differentiation through its roles in normal gene expression and genomic stability [7]. Therefore, any alteration in the patterns of DNA methylation could result in the abnormal growth of a normal cell toward tumorigenesis [8]. Tumor suppressor genes such as RAR- $\beta$ are the potential targets for DNA hypermethylation during carcinogenesis.

Loss of expression of retinoic acid receptor $\beta$ (RAR- $\beta$ ) during cancer development is associated with tumorigenesis and retinoid resistance. This gene encodes a protein involved in the functions of retinoids which are known to regulate a large number of essentially biological processes and to suppress carcinogenesis [9].

Of the genes responding to RAR- $\beta$ , genes involved in the regulation of apoptosis, cell cycle, transformation, metastasis and cell-matrix interaction can be noted [10, 11]. Accordingly, expression of RAR- $\beta$ mediates retinoic acid-induced growth arrest and apoptosis in cancer cells [12, 13].

The tumor suppressor role of RAR- $\beta$ is established in some cancer, such as breast [14], lung [15], and esophagus [16]. The activity of tumor suppressor genes such as RAR- $\beta$ could be inhibited via epigenetic or genetic mechanisms [17]. Silencing of RAR- $\beta$ through DNA hypermethylation, is reported in several carcinomas such as cervical, lung, breast, colorectal and head & neck cancers [18, 19, 20, 21, 22].

RAR- $\beta$ as a transcription factor recruits some protein complexes such as histone acetyltransferases and histone deacetylases to the promoter of target genes to establish local alterations in chromatin and regulate gene expression [23, 24].

According to previous studies, intestinal and diffuse-type of gastric cancer are different from view of genetic [25, 26] and epigenetic [26, 27] patterns.

The status of RAR- $\beta$ methylation in gastric cancer was previously evaluated [28, 29, 30, 31, 17, 32]. However, there is no report on the contribution of RAR- $\beta$ hypermethylation in the development of intestinal and diffuse-type gastric cancer from Iran in databases while gastric cancer is one of the five common cancers in Iranian population [33, 34].

Therefore, the aim of the current study was to explore the methylation status of RAR- $\beta$ in intestinal and diffuse-type adenocarcinoma of gastric cancer and its correlation with clinical and histopathological features in the Iranian population.

2. Methods

2.1 Sampling

This study was approved by the ethics and scientific committee of our Institution. Cancer lesions with matched normal tissue were taken from consenting patients who had undergone endoscopic evaluation of upper gastrointestinal tract or paraffin block archived in the Taleghani hospital (Tehran-Iran). The histological grading was performed using Lauren’s classification [3].

Patients with present or previous neoplastic disease, previous gastric surgery, and gastric or duodenal ulcer were excluded. Also, demographic data such as age and gender were gathered.

2.2 DNA extraction

The DNA from gastric samples was extracted using DNeasy kit (QIAGEN) according to the manufacturer’s instructions. For paraffin blocks, the sections were firstly dewaxed by xylene and rehydrated with graded ethanol.

2.3 Sodium bisulfite treatment

Conversion of unmethylated cytosine to uracil in the genomic DNA was performed using the EpiTect Bisulfite Kit (Qiagen Inc.) according to the manufacturer’s instructions. After the conversion reaction steps, bisulfite-converted DNA was eluted in 20 $\mu$ L elution buffer and used immediately or stored at $-$ 20 ${}^{\circ }$ C for methylation analysis.

Table 1

Primer sequences for methylation-specific polymerase chain reaction

		Sequence	Product length
Methylated	F	5’ ATTGGGATGTCGAGAACGC3’	155 bp.
	R	5’ GACCAATCCAACCGAAACG3’
Unmethylated	F	5’ GAGGATTGGGATGTTGAGAATGT3’	166 bp.
	R	5’ CTTACTCAACCAATCCAACCA3’

Table 2

Clinicopathologic characteristics of gastric cancer patients

Characteristics	No. of patients (%)
	Intestinal-type		Diffuse-type
Gender
Male	23	(69.7)	18	(58)
Female	10	(30.3)	13	(42)
Age, years (Mean $\pm$ SD)	25–83	(59.9 $\pm$ 12.8)	41–58	(56.8 $\pm$ 11.5)
Tumor localization
Body	11	(33.3)	3	(9.6)
Antrum	14	(42.4)	5	(16)
Pylorus	3	(9.6)	4	(13)
Cardia	3	(9)	4	(13)
Fundus	2	(6)	15	(48)
Tumor size (cm)
$⩽$ 5	13/33	(39.4)	12/31	(38.7)
5.1–10.0	7/33	(21.2)	7/31	(22.6)
10.1–15.0	10/33	(30.3)	10/31	(32.3)
$>$ 15.1	3/33	(9.1)	2/31	(6.5)
Differentiation
Poor	2	(6.1)	18	(58.1)
Moderate	17	(51.5)	3	(9.7)
Well	14	(42.4)	10	(32.2)

2.4 Methylation-specific PCR (MSP)

MSP has been designed to distinguish methylated from unmethylated DNA by applying two sets of primers: the methylation-specific and unmethylation-specific, taking advantage of the sequence differences resulting from bisulfite modification [35].

Bisulfite-modified DNA was amplified using designed primers (Table 1) according to genbank sequence (accession number: X56849), with cycling program as follows: initial 95 ${}^{\circ }$ C for 5 min (one cycle) and 35 cycles of 95 ${}^{\circ }$ C for 30 s, 53.5 ${}^{\circ }$ C for 60 s, at 72 ${}^{\circ }$ C for 1 min; and, a final extension step, at 72 ${}^{\circ }$ C for 10 min.

All double MSPs were performed with controls for unmethylated and methylated alleles using DNA extracted from normal person’s blood and CpG Methylated HeLa Genomic DNA purchased from new England Biolabs ${}^{\mathrm{\circledR }}$ respectively. PCR products were visualized on a 2% agarose gel with GelRed. The presence of a visible MSP product in those lanes marked M indicated the presence of methylated alleles.

2.5 Statistical analysis

To assessment the results of the study, SPSS 13 was used. Statistical analyses were performed using the Pearson’s $\chi$ 2 test or Fisher’s exact test to assess associ- ations between methylation status and clinicopathological characteristics, such as gender, age, histopathology, tumor extension, and presence of lymph node or distant metastasis. P values less than 0.05 were considered to be statistically significant.

3. Results

3.1 Patients’ findings

In total 31 and 33 histologically confirmed paired samples (normal and cancer) of diffuse and intestinal-type adenocarcinoma were included in the study, age range 41–85 (mean 56.8 $\pm$ 11.5) and 25–83 (mean 59.9 $\pm$ 12.8) years respectively (Table 2). Distribution of age data was normal in both diffuse type and intestinal type adenocarcinoma ( $P=$ 0.2 and 0.2 respectively, Kolmogorov-Smirnov test).

Table 3

Association of histological classification (intestinal and diffuse-type) and promoter methylation status of RAR- $\beta$

Samples	DNA methylation status				P-value ${}^{*}$		P-value ${}^{\mathrm{\square }}$
	Methylated		Unmethylated
Normal ( $N=$ 64)	0	(0%)	64	(100%)	–
Cancer ( $N=$ 64)	30	(46.9%)	34	(53.1%)	$<$ 0.0001
Diffuse ( $N=$ 31)	19	(61.3%)	12	(38.7%)	$<$ 0.0001	$<$ 0.0001
Intestinal ( $N=$ 33)	11	(33.3%)	22	(66.7%)	$<$ 0.0001	0.0004	0.0442

Data are reported as number of individuals with percent within parentheses. ${}^{*}$ Fisher’s exact test. Intestinalor diffuse-type gastric carcinomas vs. Normal tissues. ${}^{\mathrm{\square }}$ Fisher’s exact test. Intestinal-type vs. diffuse-type gastric carcinomas.

Patients with diffuse-type composed of 18 males and 13 females while intestinal-type group include 23 males and 10 females. There was no statistical difference between two groups regarding patients’ gender ( $P=$ 0.332, Chi ${}^2$ test). Clinicopathological characteristics of gastric cancer patients are outlined in Table 2.

Of diffuse-type samples, 58.1% were poorly differentiated while the majority (51.5%) of intestinal-type samples were moderately differentiated and this difference in the differentiation was statistically significant ( $P<$ 0.0001, Chi ${}^2$ test). Regarding the location, diffuse-type and intestinal-type tumors tend to be in the proximal and distal segments respectively ( $P<$ 0.001, Chi ${}^2$ test).

Figure 1.

Gel electrophoresis using RAR- $\beta$ MSP primers on diffuse-type adenocarcinoma. L: size marker 100 bp, M: methylated, U: unmethylated, N: normal tissues, H: CpG Methylated HeLa Genomic DNA and 1, 2, 3, 4, 5: tumor tissues.

Figure 2.

Gel electrophoresis using RAR- $\beta$ MSP primers on intestinal-type adenocarcinoma. L: size marker 100 bp, M: methylated, U: unmethylated, N: normal tissues, H: CpG Methylated HeLa Genomic DNA and 1, 2, 3, 4, 5: tumor tissues.

3.2 Methylation status of diffuse and intestinal-type adenocarcinoma

The hypermethylation frequency of RAR- $\beta$ promoter was 61% and 33% in diffuse and intestinal-type samples respectively (Figs 1 and 2). Indeed, none of the non-neoplastic samples showed hypermethylation of RAR- $\beta$ gene promoter (Fig. 3). Hypermethylation of RAR- $\beta$ gene was associated with both diffuse-type and intestinal-type compared to matched control samples (Table 3) (respectively $P<$ 0.0001 and $P=$ 0.0004, Fisher’s exact test). The difference in hypermethylation frequency between diffuse and intestinal-type was also significant (the Fisher’s exact test $P=$ 0.0442).

Figure 3.

Gel electrophoresis using RAR- $\beta$ MSP primers on normal tissues. L: size marker 100 bp, M: methylated, U: unmethylated, N: normal tissues, H: CpG Methylated HeLa Genomic DNA and 1, 2, 3, 4, 5: normal tissues.

Figure 4.

Correlation between age of diagnosis and RAR- $\beta$ promoter methylation in diffuse and intestinal-type GC. Data are shown as mean $\pm$ SD.

Figure 5.

Correlation between size of tumor and RAR- $\beta$ promoter methylation in diffuse and intestinal-type GC. Data are shown as mean $\pm$ SD.

3.3 Hypermethylation and clinicopathological findings

We investigated whether RAR- $\beta$ hypermethylation was associated with clinical and pathological characteristics, and detected no significant association of methylation status with age (Fig. 4) in diffuse ( $P=$ 0.18) and intestinal-type ( $P=$ 0.56) gastric cancer. We could find any significant association between gender and RAR- $\beta$ hypermethylation in diffuse ( $P=$ 0.98) and intestinal-type ( $P=$ 0.87) gastric cancer (Table 4).

Table 4

Frequency of RAR- $\beta$ promoter methylation with relation to gender in diffuse and intestinal-GC

Pathology	Methylation status	Gender
		Male		Female
Diffuse	Methylated	11	(61.1)	8	(61.5)	19	(61.3)
	Unmethylated	7	(38.9)	5	(38.5)	12	(38.7)
Total		18	(100)	13	(100)	31	(100)
Intestinal	Methylated	8	(34.8)	3	(30.0)	11	(33.3)
	Unmethylated	15	(65.2)	7	(70.0)	22	(66.7)
Total		23	(100)	10	(100)	33	(100)

Data are shown as number (%) of methylated/unmethylated cases in each gender.

Table 5

Methylation number and frequency of RAR- $\beta$ in gastric samples in comparison with invasion

Pathology	Invasion	Methylation status		P-value ${}^{*}$
		Methylated	Unmethylated
Diffuse-type	Yes	16 (80%)	4 (20%)	0.007
	No	3 (26%)	8 (73%)
Intestinal-type	Yes	9 (36%)	16 (64%)	0.687
	No	2 (25%)	6 (75%)

Data are reported as number of individuals with percent within parentheses. ${}^{*}$ Fisher exact test.

3.4 Hypermethylation and tumor size

We detected a direct association between RAR- $\beta$ gene hypermethylation in intestinal-type gastric cancer and greater tumor size (t-test, $p=$ 0.029) but not in diffuse-type adenocarcinoma (t-test, $p=$ 0.268) (Fig. 5).

3.5 Hypermethylation and invasion

Promoter hypermethylation of RAR- $\beta$ gene was associated with invasion of diffuse type tumors ( $p=$ 0.007 Fisher’s exact test) as 80% of invasive tumors were methylated (Table 5). On the other hand, there was no significant correlation between hypermethylation and invasion in intestinal type ( $p=$ 0.687 Fisher’s exact test).

3.6 Hypermethylation and differentiation

There was an inverse relationship between RAR- $\beta$ hypermethylation and degree of differentiation in both diffuse and intestinal-type (Table 6) so that the frequency of hypermethylation increased in accompany with a decrease in differentiation, but the associations was only statistically significant for ( $p=$ 0.033 and 0.345 Fisher’s exact test).

Table 6

Correlation of promoter methylation status and differentiation in intestinal and diffuse-type gastric carcinoma

Pathology	Differentiation	Methylation status
		Methylated		Unmethylated
Diffuse-type	Well	3	(30)	7	(70)
	Moderate	2	(66.7)	1	(33.3)
	Poor	14	(77.8)	4	(22.2)
Intestinal-type	Well	3	(21.4)	11	(78.6)
	Moderate	7	(41.2)	10	(58.8)
	Poor	1	(50)	1	(50)

Data are reported as number of individuals with percent within parentheses.

3.7 Hypermethylation and tumor location

RAR- $\beta$ hypermethylation was observed more often in the fundus segment of diffuse-type gastric cancer tissues than in the other segments ( $p=$ 0.012 Fisher’s exact test). There was no difference in the location of methylated and unmethylated intestinal-type gastric cancer ( $p=$ 0.169 Fisher’s exact test), although the body and antrum samples constitute 54.5% and 50% of methylated and unmethylated samples. The interesting observation of our study was that samples of pylorus segment were not methylated both in diffuse and intestinal-type cancers (Table 7).

Table 7

Correlations between RAR- $\beta$ status and tumor location in diffuse and intestinal-type gastric cancer

Pathology	Tumor location	Methylation status
		Methylated		Unmethylated
Diffuse	Body	2	(10.5)	1	(8.3)
	Antrum	4	(21.1)	1	(8.3)
	Pylorus	0		4	(33.3)
	Cardia	1	(5.3)	3	(25)
	Fundus	12	(63.2)	3	(25)
Intestinal	Body	6	(54.5)	5	(22.7)
	Antrum	3	(27.3)	11	(50)
	Pylorus	0		3	(13.6)
	Cardia	2	(18.2)	1	(4.5)
	Fundus	0		2	(9.1)

Data are reported as number of individuals with percent within parentheses.

4. Discussion

Gastric cancer is one of the most prevalent and cause of death in the world and Iran [1, 33, 34]. However, the molecular mechanism of gastric cancer is not fully understood as yet. Indeed, there is little work to search the molecular characterization of two histologically and clinically different types of gastric cancer i.e. intestinal and diffuse-type adenocarcinoma. So, this study focused on the status of the RAR- $\beta$ gene methylation and its correlation with clinicopathological aspect of these subtypes of gastric cancer.

We showed that gastric cancer samples were approximately 46.87% methylated for RAR- $\beta$ . According to previous reports, the methylation frequencies of RAR- $\beta$ vary from 17% [28] to 45–66.8% [32, 17, 36, 37, 38]. Therefore, it could be concluded that suppression of RAR- $\beta$ gene expression through DNA hypermethylation has some advantages for the development of gastric cancer.

According to our finding, the frequency of hypermethylation in diffuse and intestinal type adenocarcinoma was significantly different (61 vs. 33%). This shown that DNA hypermethylation of RAR- $\beta$ occurs preferentially in diffuse type gastric carcinomas rather than intestinal type gastric carcinomas. The distinct pattern of promoter methylation were reported previously for RAR- $\beta$ [36, 37, 38] and some other genes such as CDH1 [26]. Therefore, it could be inferred that these types of gastric cancer have a different pattern of gene activation and/or silencing during carcinogenesis, so that in each type, activation of (an) oncogene(s), induces hypermethylation of specific gene(s) and ultimately different clinicopathological properties.

Previous work has been shown that RAR- $\beta$ hypermethylation maybe has an important role during the initial steps of some cancer, such as esophagus cancer [39] and it is suggested that this epigenetic modification occurs early in the diffuse type of GC [32]. In our previous studies some early (epi)genetic alteration in gastritis lesion was detected [40, 41, 42, 43]. However, there is no previous report on the RAR- $\beta$ hypermethylation during precancerous lesions such as gastritis or intestinal metaplasia. Consequently, it is impossible to determine in which steps of carcinogenesis, RAR- $\beta$ undergoes DNA hypermethylation processes. Therefore, revealing the status of RAR- $\beta$ hypermethylation in precancerous lesions seems necessary.

In diffuse-type, RAR- $\beta$ hypermethylation has been associated with the invasion of the cancer cells to the adjacent tissues. So far there is no report on the correlation of RAR- $\beta$ hypermethylation and invasion in gastric cancer. Degradation of extracellular matrix is one important prerequisite of invasion, a process by matrix metalloproteinases, which their synthesis and gene expression is under control of retinoids through RAR- $\beta$ [43]. On the other hand, increased invasion of diffuse type can be attributed to down regulation of nm23-H gene, which its up-regulation by some types of retinoids (through RAR- $\beta$ ) reduces cell motility [44]. It will be interesting to determine whether nm23-H1 expression is correlated with RAR- $\beta$ promoter hypermethylation status.

Therefore, it is possible that RAR- $\beta$ hypermethylation and inhibition of its expression facilitate invasion and expansion of tumor cells to the adjacent tissues, through a decrease in the expression of gene involved in the establishment and rearrangement of cell adherence and motility.

In accordance with some previous work suggestive of location-specific pattern of DNA methylation [27], our results showed that in diffuse-type GC, hypermethylation of RAR- $\beta$ correlates with fundus location. These findings support the notion that fundus tumors have different behavior from non-fundus tumors. Maybe the presence of a specific factor or carcinogenic process in a defined anatomical location could be able to induce specific genetic and epigenetic pattern. Accordingly eradication of H. Pylori would be changed aberrant DNA hypermethylation in a gene-specific manner [31]. It will be interesting to search for a factor in the fundus region which directly or indirectly influences on the pattern of DNA methylation and subsequent carcinogenesis.

In intestinal-type GC, hypermethylation of RAR- $\beta$ promoter had a direct association with the size of tumor. As tumor size is an important factor in determining prognosis and survival of patients [45, 46, 47], it could be expected that in intestinal-type adenocarcinoma, RAR- $\beta$ hypermethylation results in increased tumor size, decreased survival, and worse prognosis.

In both tumor types, there was a reverse correlation between hypermethylation and degree of differentiation. Although this association was not statistically significant for intestinal-type cancer, nevertheless these results reveal that hypermethylation of RAR- $\beta$ and subsequent inhibition of gene expression is to some extend essential for differentiation of gastric tumor cells and is associated with a more aggressive behavior. Our results should be interpreted critically because of the few patients in each subgroup. A comprehensive study organized of all differentiation degree will illuminate this matter.

DNA hypermethylation in some genomic regions is age-related, for example hypermethylation of the hMLH1 promoter in gastric cancers [48]. However, in our study, RAR- $\beta$ hypermethylation was not associated with age of diagnosis and normal adjacent tissues did not exhibit any RAR- $\beta$ hypermethylation in both types of gastric cancer. These indicate that RAR- $\beta$ hypermethylation plays no role in gastric carcinogenesis in old people and its hypermethylation results from some cancer-specific process, not from age-related changes in DNA methylation.

5. Conclusion

In the present study, we evaluated the methylation status of RAR- $\beta$ promoter in gastric adenocarcinoma samples and their possible associations with clinical and pathological characteristics. The results of our study show that hypermethylation of RAR- $\beta$ gene promoter occurred frequently in gastric cancer however, two histological types of GC differ in their tumorigenesis processes. Some of the different morphological and clinicopathological aspects of intestinal and diffuse-type adenocarcinoma could be attributed to the genetic and epigenetic profile of their progenitor cell.