Analysis of cell cycle-related proteins in gastric intramucosal differentiated-type cancers based on mucin phenotypes: a novel hypothesis of early gastric carcinogenesis based on mucin phenotype

Progression through the cell cycle and cellular proliferation are under the control of a series of cyclins and cyclin-dependent kinase (cdk) complexes [1‐3]. Accumulating evidence shows that the progression of tumorigenesis frequently involves abnormalities in the expressions of cyclins and other cell-cycle related genes [1‐3]. Abnormalities have been found for cyclins D1, A, E and their co-operating partners, such as cyclin-dependent kinase (cdk), that promote cell cycle progression [1, 3]. Additionally, these progressive factors can be inhibited by blockers, such as p21, p27 and p57, and another group of inhibitor proteins, including p16, p15 and p18 [4‐10]. The uncontrolled proliferation that characterizes tumor cells can be largely explained by the gain and/or loss of protein functions that comprise the cell cycle. Regulation of these cell cycle-related proteins is also governed by other factors, including p53 and β-catenin, and their alterations also impair the cell cycle, resulting in uncontrolled proliferation [11‐15].

Of the above cell cycle-related proteins, key regulators of progression through the G1 phase of the cell cycle are cyclin D1 and cyclin E, p53, p21 and p27 [1, 4, 6, 12]. Their abnormal expressions have been thought to play pivotal roles in the progression of tumorigenesis and have been found to be disturbed in a number of human malignancies. Cyclin A is also a member of the cyclin protein superfamily that can be activated during the transition from the G1 to the S phase of the cell cycle. Abnormal expressions of cyclin A are correlated with poor outcomes in various human cancers [8, 9]. In addition, nuclear expression of β-catenin is implicated in gastrointestinal cancers [14, 15]. β-catenin accumulates in the nucleus due to impairments of the Wnt signal pathway, and its nuclear expression promotes progression of the cell cycle and cellular proliferation [14, 15]. However, to date, its activity has not been shown to affect the pathogenesis of early differentiated-type gastric cancers.

Recent studies have shown that cellular mucin expressions and tumor phenotypes are associated with the clinico-pathological findings and tumorigenesis in differentiated-type gastric cancers [16‐19]. The mucin phenotypes of tumors have been primarily classified into 3 types: gastric, intestinal and mixed phenotypes [16, 17]. The gastric phenotype is characterized by poor outcomes, distinct histological features and a specific subtype of genetic alterations, including microsatellite instability (MSI) [17, 18]. In contrast, the intestinal phenotype is a very well differentiated type, with low proliferative activity and a lack of MSI [17]. The expressions of mucins by tumor cells define tumor characteristics in gastric cancers [16‐19]. Thus, it is important for the understanding of early tumorigeneis of gastric cancers to examine biological alterations according to these mucin phenotypes [16‐19].

Although a number of studies regarding the expressions of cell cycle-related factors have been reported [3‐7], the associations of early differentiated-type gastric cancers and their mucin phenotypes and alterations of cell-cycle-related proteins are not fully understood. In the present study, we examined abnormalities of cell cycle-related proteins of the early phase of differentiated-type gastric cancers based on mucin phenotypes.

In spite of a decreasing trend in incidence, gastric cancer remains an important medical problem throughout the world, and it is the most common neoplasm in Japan. Intramucosal cancer is defined as a tumor at the most early stage of gastric cancer. It is essential to examine intramucosal cancer for evaluating the pathogenesis of gastric cancer. It is widely accepted that a cancerous cell arises due to abnormalities of cell cycle regulation, which is comprised of several cell cycle-related proteins, such as cyclins, cyclin dependent kinase inhibitor, p53 and β-catenin. In addition, recent studies have shown that distinct clinico-pathological findings for gastric cancer are characterized by mucin expressions. We studied abnormal expressions of cell cycle-related proteins that promote gastric carcinogenesis based on mucin phenotypes in gastric intramucosal differentiated-type cancers.

In the present study, immunohistochemistry was used to investigate the expression levels of cell cycle-related proteins. Immunopositivity for cell cycle-related proteins of tumor tissue was determined by comparing the levels between tumor tissue and its surrounding non-neoplastic tissue. According to these criteria, the level of immunopositivity of tumor tissue may differ for each case. However, separating the parameters within these cancers into "low" and "high" categories based on comparisons of these factors with non-neoplastic tissue may be an unconventional method. The "non-neoplastic" tissue includes gastritis (chronic or chronic active) and/or intestinal metaplasia; the parameters in these various inflammatory or metaplastic tissues could vary considerably. It may be more meaningful to use an absolute rate (i.e., the actual %), rather than this internal relative value. However, we consider that this immunohistochemical criterion is important for finding the potential activities of cell cycle-related proteins at an early stage of gastric differentiated-type cancers, given that the non-neoplastic tissue surrounding tumor tissue is closely associated with tumor development.

It has been believed among pathologists that the gastric phenotype of gastric cancer shows low-grade atypia in gastric intramucosal differentiated-type cancers (19). However, in the present study, in the gastric intramucosal differentiated-type cancers that we examined, the gastric phenotype showed high-grade atypia. Although the reason remains unknown, one possible reason is that a gastric phenotype showing low-grade atypia, which resembles gastric foveolar epithelium histologically, may have been included in previous studies. In the present study, only one such tumor was observed. In addition, these differences could be due to the number of samples or to ethnic or etiologic factors. It is well known that p53 overexpression is correlated with tumor grade. In the present study, we suggest that it is reasonable to show a high frequency of p53 overexpression for the gastric phenotype when compared with other phenotypes.

In this study, p53 overexpression was more frequently found in the gastric phenotype than in the intestinal or mixed phenotypes. This finding suggests that p53 overexpression plays an essential role, at least in the early development of gastric phenotype cancer, although previous studies have shown progressive increases in p53 overexpression in gastric cancers from the early to advanced stages [12].

In this study, reduced p21 expression was observed in approximately 30% of the 3 mucin phenotypes. Previous studies have shown that low expression of p21 is correlated with patient prognosis in gastric cancer [5, 6]. However, little is known regarding p21 expression by gastric cancers based on mucin phenotypes. In addition, any association of p21 expression with tumor stage (tumor depth) remains unknown. Our finding suggests that reduced p21 expression plays a common role in the early development of the 3 mucin phenotypes of gastric differentiated-type cancers. However, a previous study suggested that a progressive decrease of p21 expression occurred in an advanced phase of gastric differentiated-type cancer from the early phase [6]. p21 expression also correlated with the expressions of other cell cycle-related proteins, such as p27 and Ki-67, possibly through a p53-dependent pathway [1, 5, 6]. It has been hypothesized that p21 co-operates with p53 and, as a result, the cell cycle is controlled through a p53-dependent pathway [1, 5]. Furthermore, it is believed that p21 is also inhibited by co-operation with p27 [1, 5]. In the present study, however, we found neither a correlation of p21 reduction with p27 reduction nor an inverse correlation of p27 reduction with p53 overexpression (data not shown). We suggest that a different pathway exists for controlling the cell cycle in human gastric cancers.

It is well known that p27 is a prototypical tumor suppressor gene [1‐3]. However, a recent study has shown that, whereas in some cases, p27 inhibits the cell cycle, in other cases it promotes progression to the next phase of the cell cycle [23]. It has been reported that either a loss or a low expression of p27 is correlated with tumor progression or patient prognosis [24, 25]. In contrast to other studies, we showed that low expression (reduction) of p27 had a very low frequency in our series of gastric cancers. According to the present findings, reduced p27 expression plays a minor role in the development of differentiated type gastric cancers.

In contrast, overexpression of p27 was frequently found in the intestinal phenotype compared with the other 2 phenotypes, although the difference between them did not reach a statistically significant level. Whether or not this overexpression of p27 is associated with cell cycle inhibition or cell cycle progression is not known, although several explanations are suggested. First is that p27 that is overexpressed for suppression of cell cycle progression in tumor cells results from a negative feedback mechanism. Conversely, a second explanation is that overexpression of p27 is positively associated with the progression of the cell cycle for tumor cells; that is, p27 acts as an oncogenic factor [23]. In addition, a recent study has shown that expression of p27 is regulated by expression of Skp2, which is a ubiquitin ligase subunit of p27 [26]. According to this report, expression of p27 was inversely correlated with Skp2 expression.

Cyclin D1 overexpression only showed a low frequency in the early stage of gastric differentiated type cancers (27/190, 14.2%). However, although the differences did not reach a significant level, the frequency of cyclin D1 overexpression was higher in the gastric phenotype than in the intestinal and mixed phenotypes. Cyclin D1 was previously studied in several series of gastric cancers, with a reported frequency of 20~56% of overexpression being reported [27‐29], different from our results. As a possible reason for these differences of cyclin D1 overexpression, although early differentiated-type cancers, except for gastric phenotype, do not require an increase of cyclin D1, once the tumor has developed, the tumor cells need a high level of cyclin D1 for tumor progression. This suggests that cyclin D1 overexpression is correlated with patient prognosis in gastric cancers [28].

Overexpression of cyclin A was also more prevalent in the gastric and mixed phenotypes compared with the intestinal phenotype. This suggests that overexpression of cyclin A plays a major role in tumorigenesis of gastric and mixed phenotypes. Thus, in gastric cancers, acquisition of gastric mucin expression may be associated with cyclin A overexpression, which then results in high tumor proliferation. In addition, cyclin A was closely associated with patient prognosis or tumor aggressiveness [8, 9]. This may indicate that cyclin A overexpression is related to both tumor development and progression in approximately half of differentiated-type gastric cancers.

In the present study, cyclin E overexpression showed low frequencies in the 3 phenotypes. In contrast, previous studies have shown that cyclin E overexpression was correlated with poor outcome [4, 10], although contrasting results have been suggested [30]. The present results indicate that cyclin E overexpression plays no essential role in the early tumorigenesis of differentiated-type gastric cancers. In addition, overexpression of cyclin E may primarily promote the progression of gastric cancers. These findings suggest that the high proliferation of the gastric phenotype may be characterized by the overexpression of cyclin D1 and/or cyclin A, and that the mixed phenotype is associated with cyclin A overexpression.

It is well known that β-catenin expression upregulates the cell cycle and plays a central role in the Wnt signal transduction pathway [14, 15]. In the present study, it is interesting to note that β-catenin expression was commonly observed in the early differentiated-type gastric cancers that we examined. It is already accepted that β-catenin expression is critical for colonic adenoma development [31, 32]. The present results and previous studies regarding colorectal adenoma indicate that β-catenin expression may be a common alteration in the early development of gastrointestinal tumors [31]. This could be explained by the findings that, in gastric cancers, gene promoters associated with the Wnt signal pathway are highly methylated [32‐34], which results in the nuclear accumulation of β-catenin expression.

In summary, we propose that the cellular mucin phenotypes of intramucosal differentiated-type adenocarcinomas of the stomach are dependent on distinct cell cycle-related alterations, and that the clinico-pathological findings result from different pathways based on mucin expression: gastric, intestinal and mixed phenotypes. Figure 4 illustrates a novel carcinogenesis model for intramucosal differentiated-type adenocarcinomas that relies on their mucin phenotypes. Based on abnormalities of cell cycle-related proteins, overexpressions of p53 and cyclin A characterize gastric phenotype cancers, whereas overexpression of p27 may be associated with the development of intestinal phenotype cancers. In contrast, mixed-phenotype cancers are characterized by cyclin A overexpression. In addition, mucin phenotypes appear to be characterized by nuclear grade and cell proliferative activity. Three of 190 tumors were of the unclassified type, and this tumor molecular characteristic is not identified. Finally, little is known regarding the distinct molecular alterations that are involved in the pathogenesis of gastric cancer. Further studies will be necessary to address this question.