We have recently identified HOPX as genes specifically methylated in human cancers [7
] after developing algorithm utilizing pharmacological unmasking microarray (PUM) [5
]. Among the identified candidates of TSGs, HOPX is of particular interest in terms of methylation and functional involvement in tumor aggressiveness. Other groups also recapitulated the similar finding that HOPX promoter DNA is hypermethylated specifically in endometrial cancer [15
]. In this present study, we for the first time added pancreatic cancer to the list of organs in which HOPX is involved in carcinogenesis.
HOPX harbors 2 discrete promoter regions, promoter A and promoter B. Promoter B has CpG islands, while promoter A does not have them, and cancer-specific hypermethylation is recognized in the promoter B in primary PC tissues as well as other GI cancers [7
]. Such independent regulation of the discrete promoter regions was reported in other critical methylation genes such as RASSF1 [19
], and possession of the complex promoter regions may indicate their functional importance in biological relevance. On the other hand, epigenetic reactivation of HOPX gene expression was much less than expected in PC cell lines as compared to other GI cancer cell lines. Allowing for actual expression in primary cancer tissues, constitutive HOPX expression signal was derived from carcinoma-stroma interaction in primary PC cells.
Pancreatic cancer is a ductal carcinoma, however it is controversial which normal components (ductal cells, acinar cells, or islet cells) of the pancreatic tissues are precursor cells for PC [20
]. Pour et al. proved that transplantation of islets into the submandibular gland of Syrian golden hamsters followed by treating with nitrosamine N-nitrosobis-(2-oxopropyl)amin (BOP), a carcinogen for PC led to the development of ductal pancreatic adenocarcinoma in this site, while PC did not occurred after transplanting ductal and acinar cells into this gland [22
]. Schmied et al. has also insisted that islet cells contribute to pancreatic carcinogenesis in an animal model and disease exploration [23
]. In mice with hamster islets implanted in the splenic lobe of the mouse pancreases, pancreatic ductal adenocarcinomas developed in the implanted animals, but not in control mice, after BOP treatment [25
]. These findings strongly supported the hypothesis that PC is generated from islet cell origin. In this current study, we for the first time revealed that islet cells expressed abundant HOPX protein in primary PC tissues as well as the normal pancreas. It is intriguing hypothesis that cancer cell with low expression of HOPX is derived from islet cells which constitutively express abundant HOPX, and that promoter DNA hypermethylation is causative for gene silencing.
Clinical findings also supported hypothesis that the islet cell is alternatively involved in PC carcinogenesis [23
], in which remarkable alteration of quality of islet cells was observed in primary PC tissues. Ten out of the 14 cancer specimens showed a significant loss of beta cells (P
0.005) and eight of them also showed a significant increase of alpha cells (P
0.005), all of them from hyperglycemic patients. Most affected islets were found within zone 1 (intratumoral) and zone 2 (peritumoral), to a lesser extent in zone 3 (acini close to tumor) and none in zone 4 (acini remote from tumor). The incidence of 72% with alteration of islets in their material correlates with the frequency of abnormal glucose levels in human pancreatic cancer patients. In our study, HOPX is remarkably increased in primary PC tissues, and it was predominantly expressed in the islet cells. These findings suggested that alteration of HOPX expression in the islet cells may explain the link of PC to diabetes mellitus, and this mechanistic possibility should be paid attention in the next future, as oncogenic role of islet cells remains elusive during PC carcinogenesis.
HOPX actually suppressed tumor aggressiveness of PC cells (PANC-1 and MIA Paca2). WST assay showed that HOPX suppressed cell viability putatively representing cell proliferation ability. In cell cycle analysis, HOPX increased subG1 and G0/G1 phases, representing apoptotic induction and inhibition of DNA synthesis, suggesting that cell cycle abnormalities may be linked to cell viability. More importantly, HOPX could inhibit tumor-forming ability in soft agar, which is supposed to represent metastatic trait of tumor cells [26
]. Interestingly, HOPX has been demonstrated to suppress tumorigenesis in soft agar in ESCC and gastric cancer as well as pancreatic cancer, hence anchorage independent growth suppression is the common feature of HOPX expression in human cancers. Finally, HOPX also affects Matrigel invasion less than other phenotypes in PC. These findings may directly show gene silencing of HOPX involved in PC aggressiveness.
Such tumor suppressive effects as shown in Figure might include artifact effect, because expression level of HOPX protein in transfectants may not correspond to the physiological level of the originated normal cell, if precursor cells of the PC were ductal or acinar cells. On the other hand, HOPX expression level of the islet cells reached similar level of the transfectants in our current study. More importantly, the level of expression in the transfectants of the current study was comparable with those of normal mucosa of other tissues such as gastric [8
] and colorectal mucosa [27
]. As compared to such common solid tumors, PC exhibited uniquely dismal prognosis, which is consistent with low expression of HOPX in PC. As constitutive expression of HOPX in human cancer cell lines including PC cell lines was infrequently found, RNA knockdown experiments was impossible to verify the endogenous role of HOPX in human pancreatic cancer cells, however we previously investigated RNA knockdown effects of HOPX by using esophageal cancer cells, TE15 that is a rare control cell which constitutively expressed HOPX [7
], and tumor suppressive role was confirmed. DNA hypermethylation of HOPX with gene silencing is therefore likely to affect PC phenotypes as in other cancers. On the other hand, there were some limitations of the conclusions that can be made based on our functional assay. In MIA Paca2 cells, HOPX was unlikely to be inactivated by methylation, and transfected HOPX protein of PANC-1 cells was expressed relatively weakly. Hence, our conclusion on tumor suppressive role of HOPX on PC was based largely on epigenetic characteristics in primary PC, and results of PC cell lines remained supplementary. We would like to know more specific and definitive conclusions as to these concerns in the near future.
HOPX affects gene transcription through recruitment of HAT and/or HDAC activity for specific transcriptional factors [28
]. Yeast two hybrid identified enhance of polycomb-1 (Epc-1), a critical component of NuA4 HAT complex, as a binding partner of HOPX, and augments transcription of heart differentiation genes. Interestingly, Epc1 was demonstrated to be associated with EZH2 which is required for cellular proliferation, E2F6-PcG complex (E2F6-EPC1) that interacts with EZH2 and may regulate genes required for cell cycle progression [31
]. Thus, HOPX may therefore affect critical process of chromatin conformation change to affect expression of onco-molecules.
Collectively, we found that HOPX methylation is a very frequent and cancer specific event in PC development. We further elucidated that HOPX is a putative tumor suppressor gene critical for tumor aggressiveness in PC. We are also interested in alternate aspects of HOPX in terms of a role in islet cells. We must confirm more detailed mechanism involved in remarkable phenotype alteration by HOPX abnormalities in PC in future study.