Gli Regulates MUC5AC Transcription in Human Gastrointestinal Cells

MUC5AC is a well-known gastric differentiation marker, which has been frequently used for the classification of stomach cancer. Immunohistochemistry revealed that expression of MUC5AC decreases accompanied with increased malignant property of gastric mucosa, which further suggests the importance of MUC5AC gene regulation. Alignment of the 5′-flanking regions of MUC5AC gene of 13 mammal species denoted high homology within 200 bp upstream of the coding region. Luciferase activities of the deletion constructs containing upstream 451 bp or shorter fragments demonstrated that 15 bp region between −111 and −125 bp plays a critical role on MUC5AC promoter activity in gastrointestinal cells. We found a putative Gli-binding site in this 15 bp sequence, and named this region a highly conserved region containing a Gli-binding site (HCR-Gli). Overexpression of Gli homologs (Gli1, Gli2, and Gli3) clearly enhanced MUC5AC promoter activity. Exogenous modulation of Gli1 and Gli2 also affected the endogenous MUC5AC gene expression in gastrointestinal cells. Chromatin immunoprecipitation assays demonstrated that Gli1 directly binds to HCR-Gli: Gli regulates MUC5AC transcription via direct protein-DNA interaction. Conversely, in the 30 human cancer cell lines and various normal tissues, expression patterns of MUC5AC and Gli did not coincide wholly: MUC5AC showed cell line-specific or tissue-specific expression whereas Gli mostly revealed ubiquitous expression. Luciferase promoter assays suggested that the far distal MUC5AC promoter region containing upstream 4010 bp seems to have several enhancer elements for gene transcription. In addition, treatments with DNA demethylation reagent and/or histone deacetylase inhibitor induced MUC5AC expression in several cell lines that were deficient in MUC5AC expression. These results indicated that Gli is necessary but not sufficient for MUC5AC expression: namely, the multiple regulatory mechanisms should work in the distal promoter region of MUC5AC gene.


Introduction
Although the mortality and incidence of gastric cancer has gradually fallen in the last several decades, it is still the fourth most common malignancy and the second leading cause of cancerrelated death worldwide [1]. Gastric cancer mainly occurs from gastric mucosa with intestinal metaplasia, which is mostly caused by chronic infection of Helicobacter pylori [2,3]. Intestinal metaplasia in stomach is typically classified into two types: mixed gastric-and-intestinal type (incomplete type) and solely intestinal type (complete type) [4]. In the process of metaplastic change of gastric mucosa, induction of intestinal differentiation and loss of gastric differentiation occur simultaneously or asynchronously [5,6].
For induction of intestinal differentiation, we and other groups have been reported that Cdx must be an indispensable key molecule by regulating transcription of many intestinal marker genes [7,8,9,10,11,12,13]. On the contrary, the molecular mechanism of loss of gastric genes in the process of metaplastic change in stomach remains poorly understood. It is probably due to inadequate identification of gastric marker genes, and also due to unsolved expression regulation of these gastric genes. Based on these backgrounds, we focused on MUC5AC, which is a wellestablished gastric marker gene [6,14] and is often used for the clinical assessment of gastric cancer [15]. There have been many previous reports investigating the expression of MUC5AC and prognosis of gastric cancer, but the association between expression of MUC5AC and malignant potential of gastric cancer is still controversial [16,17,18,19]. Nevertheless, MUC5AC is one the most evident gastric marker clearly decreased in the process of intestinal metaplasia [14,20]. We believe elucidating the mechanism of MUC5AC gene expression in gastrointestinal cells must be useful to understand the loss of gastric differentiation during the development of pre-malignant atrophic gastritis.
Our aim of this study is to find a critical mechanism of MUC5AC expression regulation in human gastrointestinal cells. To date, HIF-1a [21], Smad4 [21,22], Sp1 [22], GATA-4/-6, and HNF-1/-4 [23] are reported to activate murine MUC5AC. Sp1 and Gli are reported to enhance MUC5AC expression in human lung-epithelial and pancreatic cancer cells, respectively [24,25]. Contrastively, ATBF1 is negatively regulate MUC5AC expression in human gastric cancer cells [26], In spite of these results, molecular mechanisms of MUC5AC expression in human gastrointestinal cells have not been fully elucidated. In the present study, we identified 15 bp sequence in human MUC5AC 59flanking region which plays an important role in MUC5AC expression in gastrointestinal cancer cells, and also found that the region contains a putative Gli-binding site which does not coincide with previous report [25]. Gli is one of transcriptional factors which has a DNA binding zinc finger domain [27]. In this study, we examined molecular roles of Gli on MUC5AC promoter in gastrointestinal cells. Our results demonstrated that MUC5AC expression depends on cooperative regulatory mechanism of Gli, some epigenetic modulation, and other factors in gastrointestinal cells.

Cell Culture
Twenty gastric cancer cell lines, ten colorectal cancer cell lines, and two non-gastrointestinal cancer cell lines were maintained in high-glucose DMEM with 10% fetal calf serum (Gibco/Invitrogen, Carlsbad, CA) at 37uC in a humidified 5% CO 2 atmosphere. Names of used cell lines and histological types of gastric cancer cell lines were described in our previous reports [6]. Human T98G and A172 cell lines were purchased from the RIKEN Bio Resource Center (Tsukuba, Japan). For the treatment with DNA demethylation reagent or histone deacetylase (HDAC) inhibitor, 5-Aza-29-deoxycytidine (5-Aza-dC, Sigma-Aldrich) at 2 mg/ml and/ or trichostatin A (TSA, Sigma-Aldrich) at 25-1000 ng/ml were added to the culture medium.

Tumor Samples
For the advanced gastric cancer specimens, we randomly selected 89 gastric adenocarcinoma samples surgically resected at the Fujita Health University Hospital. For the early stage gastric cancer endoscopically resected, we selected 78 specimens banked at the University of Tokyo Hospital. This study was approved by the ethic committees of the University of Tokyo, and also by the institutional ethical review board for human investigation at Fujita Health University. According to the Declaration of Helsinki, written informed consents were obtained from all the study participants for use of resected sample in research.
For Gli1 immunostaining, hydrated heating in 10 mM citrate buffer (pH 6.0) was done in a pressure cooker for 10 min for antigen retrieval. The sections were then incubated for 1 h at room temperature with antihuman Gli1 antibody (H-300, sc-20867, Santa Cruz Biotechnology) at a 1:500 dilution. For the amplification of signals, anti-rabbit antibody (Dako, Hamburg, Germany) was applied to the slides for 30 min at room temperature. This was followed by incubation with FITCconjugated phenol (fluorescyl-tyramide; Dako) for 15 min at room temperature and incubation with anti-FITC antibody conjugated to HRP (Dako) for 15 min at room temperature was then done.

Luciferase reporter assay
Cells were cultured on 96-well plates and transiently transfected with mixtures of Renilla luciferase control vector pGL4.74 (Promega) (15 ng) and the firefly luciferase reporter plasmids (150 ng) by using Lipofectamine and Lipofectamine PLUS (Invitrogen). To examine the effect of Gli on the promoter activities, pcDNA3.1(+)-Gli1, -Gli2, and -Gli3 plasmids were transfected with pGL-based plasmids at the same time. Luciferase assays were performed at 24 or 48 h post-transfected using the Dual luciferase reporter assay system (Promega). Luciferase activities were measured by an AutoLumat Plus LB953 (Berthold, Bad Wildbad, Germany). Luciferase activity was normalized to a Renilla control and the results are shown as mean 6 SD from at least three independent experiments.

Production and infection of retrovirus vectors
pMXs-Gli1-IRES-Puro, pMXs-Gli2-IRES-Puro and pMXs-Gli3-IRES-Puro and control vector (pMXs-IRES-Puro) were transfected with pCAG-VSVG into the PLAT-GP packaging cell line for retrovirus vector production (Cell Biolabs, Inc. San Diego, CA) using Lipofectamine and Lipofectamine PLUS (Invitrogen). Virus-containing supernatants were harvested at 24 h, 48 h and 72 h post-transfection, pooled, and filtered though a 0.45-mm filter (Millipore, Bedford, MA). To assess the effects of overproduction of Gli, cells were infected by adding retrovirus-containing supernatant at a final concentration of 8 mg/ml Polybrene (Sigma) and were selected over 3 days with puromycin (Sigma).

RT-PCR
Total cellular RNA was prepared using the Isogen RNA isolation reagent (Wako Pure Chemical Industries, Osaka, Japan) as previously reported [29]. RT-PCR was performed via a Superscript One-Step reaction using the Platinum Taq (Invitrogen). The primer sequences used for RT-PCR are shown in Table 2. RNA was reverse-transcribed for 30 min at 50uC, and after an initial denaturation at 94uC for 3 min, cDNA amplification procedures were performed as follows: for MUC5AC, 35 cycles of 94uC for 30 sec, 64uC for 1 min, and 72uC for 1 min; for GAPDH, 30 cycles of 94uC for 30 sec, 60uC for 1 min, and 72uC for 1 min; for Gli1 and Gli2, 35 cycles of 94uC for 30 sec, 60uC for 1 min, and 72uC for 1 min; for Gli3, 40 cycles of 94uC for 30 sec. A commercial RNA panel, Human Total RNA Master Panel II, was purchased from Clontech Laboratories (Palo Alto, CA, USA). siRNA SH-10-TC cells were transfected with siGENOME SMARTpool siRNA against human Gli1 and/or Gli2 (M-003896-00-0005 and M-006468-02-0005, respectively; Dharmacon, Lafayetta, CO, USA) using Lipofectamine RNAiMAX (Invitrogen) according to the manufacture's instructions. RNA was extracted from the cells 48 h after siRNA transfection and used for RT-PCR.

Chromatin immunoprecipitation (ChIP) assay
ChIP analyses were performed using a ChIP assay kit (Upstate Biotechnology Inc., Lake Placid, NY) according to the manufacture's instructions. For crosslinking, cells (2610 6 ) were incubated at 37uC for 15 min in PBS containing 1% formaldehyde. Cells were collected after two washes with ice-cold PBS and subjected to centrifugation for 5 min at 7006g. Cells pellets were resuspended 400 ml sodium dodecyl sulfate (SDS) lysis buffer (ChIP assay kit, Upstate) containing protease inhibitors (0.1 M phenylmethylsulfonylfluoride and 4 ml protease inhibitor cocktail (Sigma)), and incubated for 10 min on ice. DNA was sonicated (setting 5, Handy Sonic, model UR-20P; Tomy Seiko, Co., Ltd., Tokyo, Japan) and then subjected to centrifugation at 13,000 rpm for 10 min. Supernatants were diluted 10-fold with Chip dilution buffer (ChIP assay kit, Upstate). 1% of the supernatant was retained as the input, and the rest was then subjected to immunoprecipitations. Immunoprecipitations were performed overnight at 4uC with 3 mg of anti-Gli1 antibody (H-300, sc-20687; Santa Cruz Biotechnology, Santa Cruz, CA) and nonimmunized rabbit IgG whole molecule (sc-2027; Santa Cruz Biotechnology). After reverse crosslinking, the obtained DNA was purified using PCR product purification kit (Qiagen, Valencia, CA, USA). Immunoprecipitated DNA was analyzed by PCR using primers 59-ctcggaaactgggctctacccgg-39 and 59-gagctttttgtagccccagagctgg-39 to amplify a fragment of the MUC5AC promoter region. Primer pairs for the villin 1 promoter region were used as a negative control [8].

Decrease of MUC5AC expression accompanied with progression of gastric canceration
To evaluate the expression of MUC5AC in gastric cancer, immunostaining was performed using the clinical specimen representing three-grade gastric epithelial cells from the view of tumorigenesis: non-tumorous but precancerous cells of atrophic mucosa around early gastric cancer (Fig. 1A), malignant cells of early gastric cancer endoscopically resected (Fig. 1B), and malignant cells of advanced gastric cancer surgically resected (Fig. 1C). Our results revealed that the stronger the malignant property of gastric epithelial cells is, the weaker the expression of MUC5AC is. In other word, expression of MUC5AC decreases accompanied with increased malignant property of gastric mucosa.
Some previous studies including ours reported that expression of MUC5AC often attenuates during the development of premalignant intestinal metaplasia [6,14]. Our present data further showed that decrease of MUC5AC expression advances in the process of gastric canceration. From these results, we are convinced that it is important to elucidate the regulatory mechanism of MUC5AC gene expression for understanding tumorigenesis of gastric cancer.
A highly conserved region containing a Gli-binding sequence (HCR-Gli) is present in the promoter of MUC5AC gene To examine the transcriptional regulation of MUC5AC gene, the upstream 59-flanking regions of MUC5AC gene of 13 mammal species retrieved from GenBank (dated 8 May, 2013) were aligned using ClustalW (ver 2.1) sequence alignment program [30]. The major transcription initiation site of human MUC5AC gene has been mapped to 48 bp upstream of the ATG translation start site [31]. Alignment of the human sequence with other mammal sequences denoted high homology among orthologs within ,200 bp upstream of the ATG start codon (Fig. 2).
To test the function of the conserved promoter regions identified above, a series of luciferase reporter constructs containing 451 bp or shorter fragments of the human MUC5AC promoter region were generated. Promoter activities of these deletion constructs were measured by luciferase reporter assay in the three MUC5AC-expressing cell lines derived from human gastrointestinal cancer: SW480, SH-10-TC, and KE-39 [6]. As shown in Fig. 3A, the promoter activity of the reporter construct containing upstream 110 bp was much lower than those containing upstream 451, 280, 190, and 158 bp.
To examine the essential region for the luciferase activity between 2158 to 2110 bp more precisely, reporter constructs containing 150, 138, and 125 bp upstream of ATG start codon were generated and examined their activities (Fig. 3B). Although the reporter constructs containing upstream 158, 150, 138, and 125 bp showed high luciferase activities, the construct containing upstream 110 bp denoted very low activities equal to a negative control.
Next we made a deletion mutant construct containing upstream 451 bp but lacking the region between 2125 and 2111 bp, and examined the promoter activity (Fig. 3C). Deletion of the 15 bp led to a marked decrease in the luciferase activity, suggesting that the 15 bp sequence between 2111 to 2125 bp is critical for the MUC5AC promoter activity. Based on the thorough literature search [32,33,34,35,36,37,38], we found a putative Gli-binding site in this short 15 bp sequence (59-GCCCTGCCCACCCAC-39 shown in Fig. 3D). We could not find any other transcription factors which bind to this 15 bp sequence by searching transcription factor database. Consequently, we named this 15 bp sequence as a highly conserved region containing a Gli-binding site (HCR-Gli).

Exogenous modulation of Gli affects the endogenous MUC5AC gene expression in gastrointestinal cells
We examined whether Gli1 can upregulate MUC5AC transcription activity in gastrointestinal cell lines. We constructed a retrovirus vectors carrying the Gli1 gene, and transduced it into four gastric (SH-10-TC, MKN1, NCI-N87, AZ-521), one colorectal (WiDr), and three other cancer cell lines (MDA- MB435, T98G, A172). In the resulting stable transductants, the expression levels of MUC5AC, Gli1, and GAPDH were analyzed by RT-PCR (Fig. 4A). Although there were two exceptions of AZ-521 and WiDr, MUC5AC was found to be universally upregulated by Gli1 overexpression. We next compared the three Gli homologs (Gli1, Gli2, and Gli3) identified in human. To examine the effect of these transcription factors on MUC5AC expression, AGS and SW480 cell lines transiently overexpressing Gli1, Gli2 and Gli3 were analyzed by RT-PCR (Fig. 4B). Gli1 strongly upregulated MUC5AC expression, which was consistent with the result of stable transductants in Fig. 4A. Gli2 and Gli3 also increased the MUC5AC expression, but their effects were weaker than Gli1.
To address whether Gli1, Gli2, and Gli3 activate the MUC5AC promoter through the 15 bp of HCR-Gli, we performed the luciferase reporter assay using the construct containing upstream 451 bp of the human MUC5AC gene and the deletion mutant lacking the HCR-Gli (Fig. 4C). In the Gli-transfected SW480, SH-10-TC, and KE-39 cells, all the three Gli homologs enhanced activation of the reporter construct containing upstream 451 bp of MUC5AC, but not of the deletion mutant lacking the HCR-Gli. These results indicate that Gli1, Gli2, and Gli3 can efficiently activate MUC5AC transcription through the HCR-Gli. From these results, we speculate that Gli1, Gli2 and Gli3 may play redundant roles but the effect of Gli1 is stronger than that of others upon the upregulation of MUC5AC expression. We next examined the effect of Gli knockdown by siRNA transfection on MUC5AC expression. After SH-10-TC cells were transfected with Gli1 and/or Gli2 siRNA, the expression levels of MUC5AC, Gli1, Gli2, and GAPDH were analyzed by RT-PCR. As shown in Fig. 4D, double knockdown of Gli1 and Gli2 most efficiently decreased MUC5AC expression, whereas the effects of either Gli1 or Gli2 single knockdown were considerably weaker. These results suggest that Gli1 and Gli2 have redundant functions for MUC5AC expression, which is consistent with our abovementioned speculation.

Gli interacts to HCR-Gli in the MUC5AC promoter region
To determine whether Gli1 directly binds to the MUC5AC promoter, chromatin immunoprecipitation (ChIP) assays were performed using anti-Gli1 antibody. As a negative control, the human villin1 promoter sequence was used because villin1 is a typical intestinal marker gene regulated by Cdx [8], and also because exogenous modulation of Gli1 expression level did not change endogenous villin1 expression (our unpublished observation). As shown in Fig. 4E, the HCR-Gli-containing fragment of MUC5AC promoter was co-immunoprecipitated with Gli1 in SW480 and KE-39 cells, whereas the upstream sequence of villin1 promoter was not. Our results demonstrate that Gli1 activates MUC5AC gene transcription through direct interaction between Gli1 and the MUC5AC promoter region.
Taking all the results from luciferase assays with upstream deletion constructs, overexpression and knockdown of Gli, and    ChIP analyses into consideration, we concluded that Gli enhances the MUC5AC gene transcription via direct protein-DNA interaction on the HCR-Gli region.

Gli is not sufficient for transcription of MUC5AC gene in gastrointestinal cells
To examine the relationship between endogenous expression levels of Gli and MUC5AC in gastrointestinal cells, RT-PCR was performed using the twenty gastric, ten colorectal, and two nongastrointestinal cancer cell lines (Fig. 5A). All the cell lines with endogenous MUC5AC expression were found to be positive for Gli1 and/or Gli2 expression. Conversely, there were several cell lines in which Gli1 and/or Gli2 were expressed whereas MUC5AC was not.
Tissue expression patterns of MUC5AC and Gli were also examined using a commercially available RNA panel of normal human tissues (Fig. 5B). Although several adult tissues such as stomach, lung, trachea, etc. obviously expressed MUC5AC mRNA, about half of the analyzed tissues were deficient in MUC5AC expression. On the other hand, Gli1, Gli2, and Gli3 were ubiquitously expressed in almost all the examined tissues. It should be noted that MUC5AC expression were not detected in adult brain, kidney, liver, placenta, skeletal muscle, spleen, thymus, thyroid gland, etc., despite the expression of the three Gli homolog genes (Fig. 5B).
Using non-malignant gastric tissues derived from the surgical specimens, we further evaluated the protein expression of MUC5AC and Gli1 by immunohistochemistry. As shown in Fig. 5C, Gli1 was weakly but ubiquitously expressed on gastric mucosa, which was different from the spotty immunostaining of MUC5AC. All the gastric tissues derived from the seven surgical specimens showed the same expression pattern.
In total, expression patterns of MUC5AC and Gli clearly support our conclusion that Gli is necessary for MUC5AC expression in human gastrointestinal cells. In addition, it is also obvious that expression of Gli is not enough for MUC5AC expression, suggesting that other mechanism should work cooperatively on the transcription of MUC5AC gene.

Multiple regulatory mechanisms work in the promoter region of MUC5AC gene
To analyze the far distal promoter region of MUC5AC gene, we generated reporter constructs containing 4010, 3000, 2000, and 1433 bp upstream of ATG start codon and examined their promoter activities (Fig. 6A). Deletion from 4010 bp to 1433 bp region revealed gradual decrease in transcriptional activity, suggesting the presence of enhancer elements in this distal promoter region. However, even in the reporter construct including long 4010 bp, deletion of the 15 bp HCR-Gli region resulted in the drastic decrease of luciferase activity (Fig. 6A). Together, our results demonstrated that there are at least two regulatory elements in the MUC5AC promoter region: the 15 bp sequence between 2111 bp to 2125 bp which critical for the promoter activity, and the wider regulatory region in 24010 bp to 21433 bp which can enhance the promoter activity.
To further examine the possible epigenetic regulation on MUC5AC expression, four gastrointestinal cell lines and two glioblastoma cell lines were treated with demethylating agent (5Aza-dC) and/or HDAC inhibitor (TSA). Transcription of MUC5AC was slightly upregulated in all the four MUC5ACdeficient cell lines (Fig. 6B), suggesting that epigenetic regulation such as methylation and/or histone acetylation additionally works as the regulatory mechanism of MUC5AC expression. From the results of four MUC5AC-deficient cell lines, methylation seems to play some universal role on suppression of MUC5AC expression. Synergistic effects of 5-Aza-dC and TSA in A172 and T98G cells also suggested that modification of histone acetylation may have some influence on MUC5AC transcription.

Discussion
MUC5AC, a secreted mucin highly detected in the superficial gastric epithelium, is a well-known gastric marker often used for classification of stomach cancer [20,39]. We have previously reported that expression of MUC5AC in stomach decreases in association with development of intestinal metaplasia [6]. In the present study, we further showed that MUC5AC expression is apparently related to the tumor stage: advanced gastric cancers present reduced levels of MUC5AC compared with early gastric cancer. Despite the anticipated importance of MUC5AC regulation, the molecular mechanisms underlying MUC5AC expression in gastrointestinal cells remain poorly understood. In this study, we searched 59-upstream of MUC5AC gene and identified the HCR-Gli at 2125/2111 bp in its promoter region. Our overexpression/knockdown analyses and luciferase reporter assays revealed that Gli induced MUC5AC expression through the HCR-Gli in gastrointestinal cells. In addition, ChIP analysis showed that Gli1 directly binds to HCR-Gli. The results of RT-PCR using 30 gastrointestinal tumor cell lines and an RNA panel from systemic normal tissues suggested that Gli is necessary but not sufficient for MUC5AC expression, which was consistent with the results from immunohistochemistry of clinical specimens. Furthermore, we found that the far distal upstream region from 21433 to 2 4010 bp enhances the Gli-dependent MUC5AC expression, at least partly based on the epigenetic modulation.
Recently, Inaguma et al. reported that Gli1 activates the MUC5AC promoter through two putative Gli-binding sites (GBS1 and GBS2) in pancreatic cancer cells [25]. However, the significance of MUC5AC expression is different in gastrointestinal cells and pancreatic cells. Namely, abundant expression of MUC5AC is observed in normal epithelial cells of stomach but its expression is decreased accompanied with development of intestinal metaplasia [14,20]. On the other hand, MUC5AC is generally undetectable in normal pancreas tissue and is sometimes ectopically induced in pancreatic tumor cells. In fact, there are two differences between Inaguma's report and our results. First, Inaguma et al. showed that both GBS1 and GBS2 respond to upregulation of MUC5AC expression by Gli. The GBS1 overlaps to our identified HCR-Gli, but our results indicated that the GBS2 (corresponding to 2159/2168 bp in Fig. 2) has no effect on MUC5AC expression (Fig. 3A). Second, Inaguma et al. reported that Gli1 expression correlates with MUC5AC induction by immunohistochemistry. However, our results revealed that expression of Gli does not always accompany with MUC5AC expression in gastrointestinal cells ( Figure 5). These inconsistencies might be due to a difference of originated tissues (pancreatic or gastrointestinal).
siRNA. (E) ChIP analyses of the human MUC5AC gene in SW480 and KE-39 cells. Anti-Gli1 antibody (Gli1-Ab) and non-immunized rabbit IgG whole molecule were used for immunoprecipitation. PCR was performed with primers recognizing the HCR-Gli-containing fragment of MUC5AC promoter or recognizing the promoter region of villin1 gene. doi:10.1371/journal.pone. 0106106.g004 No induction of MUC5AC expression in AZ521 cells (Fig. 4A), which completely lack the SWI/SNF complex-based chromatin remodeling activity [28], may suggest the interaction between Gli and the SWI/SNF complex. Although there have been no previous reports concerning the interaction with Gli, SWI/SNF chromatin remodeling complex has been reported to bind with various transcription factors such as AP-1 [40], CREB [41], MyoD [42], Cdx [8], GATA1 [43], and so on. The 59-upstream distal region from 21433 to 24010 bp in MUC5AC promoter showed obvious enhancing effect on MUC5AC expression (Fig. 6A). Interestingly, we noticed that the region from 21455 to 23187 bp contains 86 copies of tandem sequences 59-TCA(C/T)TCA(C/T)-39 (Fig. 6C) [44]. The tandem repeats are conserved among mammals especially among primate and cetartiodactyla. Human, olive baboon, rhesusmonkey, killerwhale, and dolphin have more than 60 repeats of the tandem repeats in their 4010 bp upstream of MUC5AC promoter region. In contrast, horse, dog, cat, panda, and hamster have only less than 10 repeats of the tandem repeats. We have no information for this tandem repeat sequences at present, but it might has some roles on regulation of MUC5AC expression.
Recently, Yamada et al. showed that the CpG methylation status of around 23.7 kb in the MUC5AC promoter is associated with MUC5AC expression in some cancer cell lines, suggesting that the expression of MUC5AC is epigenetically regulated in the distal promoter region [45]. In fact, we found increase in MUC5AC expression by 5Aza-dC and/or TSA treatment, indicating some epigenetic mechanism involved in MUC5AC gene regulation (Fig. 6C).
To understand gastric tumorigenesis mainly occurred in the atrophic mucosa with chronic gastritis and intestinal metaplasia, we believe the disrupted balance between intestinal and gastric differentiation must be important [8,28]. For intestinal differentiation, many previous reports indicated that Cdx must be an indispensable key molecule based on the transcriptional regulation of many intestinal differentiation marker genes [8,9,10,11,12]. On the contrary, the critical key molecule for gastric differentiation has not been identified yet. Despite the through screening, we have not found the essential regulator for our identified gastric marker Cathepsin E (data not shown) [6], and we have elucidated that typical gastric marker MUC5AC gene is partly regulated by Gli, a universal transcription factor which alone cannot realize the tissuespecific expression of MUC5AC gene. Unlike intestinal differentiation, we now speculate that gastric differentiation may depend on cooperative regulatory mechanism of some universal transcription factors and epigenetic modulations. In summary, the present study has shown that Gli regulates MUC5AC gene expression via direct protein-DNA interaction through the highly conserved 15 bp sequence between 2125 and 2111 bp in the promoter region of MUC5AC. Furthermore, immunohistochemical analysis and RT-PCR using systemic normal tissue revealed that Gli is necessary but not sufficient for MUC5AC expression. We conclude that MUC5AC expression is regulated by combination of multiple regulatory mechanisms such as universal transcription factors and epigenetic modulations.