The original isolation paper described the over-expression and not the amplification of EPHA1 in human colorectal cancer[1]. However, the significance of EPHA1 in human cancers is far from being solved. Although EPHA1 was first suspected to be an oncogene (growth factor receptor-like epidermal growth factor receptor), many investigators have recently focused on its down-regulation in human tumors and its possible clinical significance[6,7]. On the other hand, from the standpoint of the pro-angiogenic activity of EPHAs, Chen et al[8] reported that the silencing of EPHA1 induces an anti-angiogenic effect in human hepatocellular carcinoma. Recently, down-regulation by epigenetic silencing was shown to be correlated with a poor survival outcome in patients with colorectal cancer[7,9]. Furthermore, Wang et al[10] extended their observation on colorectal cancer to gastric cancer; that is, they reported the correlation between EPHA1 expression and gastric cancer metastasis and survival. Contrary to the situation reported by Dong et al[6] in colorectal cancer, EPHA1 up-regulation was related to a poor survival outcome and the metastasis of gastric cancer.

EPHA1 expression is possibly regulated by environmental factors. Doleman et al[11] reported that EPHA1 expression and EPHB4 are influenced by n-3 fatty acid eicosapentaenoic acid (EPA). This observation may imply the important involvement of EPH pathways in the mechanism responsible for the presumed health benefits of polyunsaturated fatty acids (PUFA).

Most research published so far about the relationship between EPHA2 expression and human gastrointestinal cancers has indicated that EPHA2 up-regulation in tumor cells results in a more aggressive nature[12-14]. In addition, EPHA2 has been extensively investigated from the standpoint of cell and vascular biology. The ligand for this receptor is EPHRINA1 (EFNA1), isolated as an acute phase reactant induced by TNF in endothelial cells[15]. This observation has tempted many investigators to study the expressions of EFNA1 and its receptor EPHA2 in tumor cells and their relation with tumor angiogenesis. In human cancers, Kataoka[12] demonstrated an increased microvessel density in EPHA2 over-expressing colorectal cancers. The mechanisms by which the overexpression of EPHA2 contributes to the aggressive behavior of cancer cells have been widely debated. Fang et al[16] discussed the importance of receptor phosphorylation and the kinase activity of EPHA2 toward the aggressive and migratory nature of tumor cells. Miao et al[17] on the other hand, reported that the activation of EPHA2 inhibits the Ras/MAPK pathway, that is, the activation of EPHA2 may reduce the aggressive nature of tumor cells. The degradation of EPHA2 is dependent on ligand inducible phosphorylation[18]; thus, the clinico-pathological effects of EPHA2 activation should be assessed, including the complex situation of the genetic profile of the tumor cells themselves and their microenvironment.

EPHA2 and its major ligand EFNA1 are perturbed by various metabolites including deoxycholic acid (DCA) and its derivative. Li et al[19] showed the up-regulation of EPHA2 by DCA in colorectal cancer cells. This may be another example of the involvement of EPH pathways and endogenous metabolites in addition to EPHA1 and PUFA.

There have been few reports on the alteration of EPHA3 in human tumors until a recent high throughput sequencing project identified a high prevalence of a somatic mutation in EPHA3 in human cancers[20,21]. The somatic mutation in EPHA3 resides in D806 where the residue is evolutionally conserved (Table 1). The prevalence does not seem to be high in any population; actually, no mutations of EPHA3 were observed in follow-up studies of 46 Japanese patients with colorectal cancer reported by Shao et al[22]. Cell signaling studies using a culture system disclosed a role of EPHA3 in the formation of a cell’s shape[23]. Thus, changes in EPHA3 are likely to produce particular morphological and biological characteristics in the tumor cells carrying these changes, although no correlation between the EPHA3 status and the clinico-pathological features of gastrointestinal cancers has yet been described. Although the clinical relevance is unknown, there is a report investigating the LINE-1 methylation pattern in the introns of EPHA3 in tumor cells[24].

The over-expression of EPHA4 has been reported in gastric and colorectal cancers[25,26]. In both cancers, the over-expression of EPHA4 is an ominous sign with a shorter survival period and frequent liver metastasis respectively. EPHA4 is the only type A receptor that binds a B family ligand, EPHRIN(EFN)B2, in addition to a type A ligand, EPHRIN(EFN)A2 (Table 3). A structural study has been conducted to reveal the stereoscopic interactions between several members of EPH receptors and EPHRIN(EFN)s[27]. The potential significance of EPHA4 over-expression in clinical oncology and the possibility of its use as a therapeutic target remain unknown.

There is no information regarding alterations in EPHA5 in human gastrointestinal cancers. EPHA5 is not expressed in the intestine at any age, as reported by Islam et al[28].

Research on EPHA6 in the gastrointestinal tract is sparse. EPHA6 is commonly expressed in the testis and brain[29].

Since the first description of the down-regulation of EPHA7 in colorectal cancer[30], several papers have assessed the expression of EPHA7 in human gastrointestinal cancers[31], human lung cancer[32] and prostate cancer[33]. The biological basis for these clinicopathological observations and their significance in oncology remain to be investigated. The promoter methylation of EPHA7 was the first example of down-regulation by methylation in EPH receptors but a subsequent survey of other EPH receptors, including EPHB receptors in colon cancer, produced negative results[34]. Another topic concerning EPHA7 is its secretory form. The secretory form of EPHA7 contains only the extracellular part of the molecule and does not anchor at the cell membrane. Its biological and clinical significance remain unknown. A secretory form of EPHA7 is known to exist in malignant lymphoma[35] and lung cancer[32] but no study has been conducted on the presence of the secretory form of EPHA7 in clinical gastrointestinal cancer.

Although the clinical significance is still unclear, Kim et al[36] reported a single nucleotide polymorphism (SNP) at the EPHA7 locus, rs2278107; this SNP was related to the chemoresponsiveness to fluoropyrimidine-based adjuvant chemotherapy for colorectal cancer[36].

EPHA8 was screened for mutation in Japanese colorectal cancer but no mutations were found[22], similar to other EPHA receptors such as EPHA3 and EPHA7. The EPHA8 receptor induces the sustained up-regulation of MAP kinase; thus, it is supposed to play a role in tumor cell growth and proliferation[37]. EPHA8 is expressed during the fetal period of intestinal morphogenesis[28] and missense mutations in stomach cancer and colon cancer are known (Table 1).

EPHB1 has been investigated in terms of signal transduction involved in the biological behavior of tumor cells[38,39], but little information is available on its status in human clinical cancer. An EPHB1 mutation was recently identified in ovarian cancer and missense mutations have also been found in gastric cancer[40] (Table 1).

EPHB2 is the most extensively studied member of EPH receptors in the field of oncology. Kiyokawa et al[2] reported the overexpression of EPHB2 in human gastric cancer and assigned it to the chromosomal locus at 1p36 which many investigators have assumed to be a tumor suppressor locus of human colon cancer because of the frequent loss of heterozygosity that has been documented[41]. Subsequently, Oba et al[42] demonstrated the loss of heterozygosity of the EPHB2 locus in human colorectal cancer. Furthermore, Batlle et al[43] argued that EPHB receptor activity could suppress the progression of colorectal cancer and EPHB2 is now viewed, at least in some contexts, as a tumor suppressor or a suppressor against tumor progression[26,44-48], although different aspects have also been discussed[49]. A group led by Hans Clevers put forward the comprehensive idea of EPHB2-EPHRINB1 interplay at the bottom of human colon crypts[50,51]. They showed the clear territory of EPHB2 and EPHRINB1 in a human colorectal crypt, its important role in cell positioning and the ordered developmental migration of intestinal cells using EphB2/EphB3 knockout mice[51]. This view is now prevalent[52] and they have further refined the concept of a stem cell unit in human gastrointestinal crypts[53,54]. Based on the mutually exclusive localization of EPHB2 and EPHRINB1, Cortina suggested that tumor compartmentalization arising from the repulsive action of cells expressing EPHB2 and EPHRINB1 is a possible mechanistic basis for tumor suppression by the EPHB2-EPHRINB1 system[55].

Then, what happened to the previous interpretation for the over-expression of EPHB2 in human cancer[2,56,57]? Mao[58] reported EPHB2 as a therapeutic antibody drug target for EPHB2 over-expressing tumors. Mutation analyses in kinase genes have been very popular and somatic mutations of EPHB2 have also been reported in many cancers[59,60], including GI tract cancers[61,62].

However, these mutations occur mostly in the microsatellite repeats of tumors with microsatellite instability or nonsense mutations causing RNA decay. No naturally occurring missense mutation that may positively or negatively influence the kinase activity of EPHB2 has ever been reported. At this moment, we can only say that individual tumors may have an individual EPHB2 status in an individual environment. The prevalence of methylation in the EPHB2 promoter, on the other hand, is low compared with RASSF2 and O-6-methylguanine-DNA methyl transferase (MGMT) in early colorectal tumors[63].

There are reports investigating the possible contribution of germline EPHB2 variants to rare polyposis syndrome[64,65]. The detailed mechanistic basis controlling the EPHB2-EPHRIN (EFN) B1 system has also been investigated. Tanaka et al[66] reported that C-terminal EFNB1 regulates matrix metalloproteinase secretion and that the phosphorylation of EFNB1 regulates the dissemination of gastric cancer cells in an animal model[66]. He also showed the successful suppression of peritoneal dissemination in an animal model using an EFNB1-derived peptide[67]. The translational approaches using this method (use of EFNB1 peptide to suppress human cancer dissemination) have not yet been shown.

The localization and function of EPHB3 partially overlaps with EPHB2 in a Paneth cell compartment. EPHB3 also has EFNB1 as a ligand. Both are controlled by the beta-catenin/Tcf4 pathway[51].

Chiu reported that the over-expression of EPHB3 enhanced cell-cell contact and suppressed tumor growth in HT-29 human colon cancer cells[68]. The defect in the positioning of Paneth cells is thought to arise from the disruption of the EPHB2-EPHB3 system[69]. Clinicopathological information on EPHB3 alone (not accompanied with EPHB2) in human gastrointestinal tract cancers remains limited. A clinical interpretation of the over-expression and/or amplification of EPHB3 (Figure 1) in gastrointestinal cancer[70,71] awaits further investigations.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Fluorescence in situ hybridization of EPHB3 (bacterial artificial chromosome RP11-328G15, red) in gastric cancer cells. Numerous red signals (more than 5) in a cell with two centromeres (green), indicating EPHB3 amplification, are shown. The methodological details have been previously reported[69,70].

Kumar reported that EPHB4 over-expression is more prevalent than EPHB2 over-expression and the cyclic AMP-responsive element binding protein-binding protein (CBP) complex reciprocally regulates EPHB2 and EPHB4 (CBP complex suppresses EPHB2 and induces EPHB4 expression)[48]. EPHB4 is thought to act in an EPHB4-EPHB6 system[72] to regulate cancer cell invasiveness. The structure and dynamism on EPHB4-EFNB2 was investigated[73,74] and the translational application of this basic knowledge awaits further investigation.

EPHB6 is the oldest EPH family member to attract enthusiastic interest from cancer researchers, especially neuroblastoma researchers. EPHB6 is unique in that there is no kinase activity. It is one of the major genes involved in the clinico-biological behaviors of neuroblastomas[75-77]. Unlike other EPHBs, a suppressor role of EPHB6 has been pointed out from an early stage of research[78-80], although its over-expression has been identified in leukemic cells[81]. A functional enigma of EPHB6, a kinase defective receptor affecting tumor invasiveness, has been gradually clarified in the fields of lung cancer research[82] but the role of EPHB6 in carcinogenesis in the human digestive tract is not clear, although its alteration such as promoter methylation in lung adenocarcinoma, has been recently reported[83]. Recently, some missense variants have been reported in familial colorectal cancer[84]. Somatic changes in colorectal cancers according to ethnic stratification have revealed EPHB6 to be one of the most frequently deleted genes in African Americans[85].