Cellular density‐dependent increases in HIF‐1α compete with c‐Myc to down‐regulate human EP4 receptor promoter activity through Sp‐1‐binding region

Abstract The up‐regulated expression of E‐type prostanoid (EP) 4 receptors has been implicated in carcinogenesis; however, the expression of EP4 receptors has also been reported to be weaker in tumor tissues than in normal tissues. Indeed, EP4 receptors have been suggested to play a role in the maintenance of colorectal homeostasis. This study aimed to examine the underlying mechanisms/reasons for why inconsistent findings have been reported regarding EP4 receptor expression levels in homeostasis and carcinogenesis by focusing on cellular densities. Thus, the human colon cancer HCA‐7 cells, which retain some functional features of normal epithelia, and luciferase reporter genes containing wild‐type or mutated EP4 receptor promoters were used for elucidating the cellular density‐dependent mechanisms about the regulation of EP4 receptor expression. In silico analysis was also utilized for confirming the relevance of the findings with respect to colon cancer development. We here demonstrated that the expression of EP4 receptors was up‐regulated by c‐Myc by binding to Sp‐1 under low cellular density conditions, but was down‐regulated under high cellular density conditions via the increase in the expression levels of HIF‐1α protein, which may pull out c‐Myc and Sp‐1 from DNA‐binding. The tightly regulated EP4 receptor expression mechanism may be a critical system for maintaining homeostasis in normal colorectal epithelial cells. Therefore, once the system is altered, possibly due to the transient overexpression of EP4 receptors, it may result in aberrant cellular proliferation and transformation to cancerous phenotypes. However, at the point, EP4 receptors themselves and their mediated homeostasis would be no longer required.


| E-type prostanoid 4 receptors; their involvement in colorectal cancer and homeostasis
Increases in the levels of cyclooxygenase-2 (COX-2) and prostaglandin E 2 (PGE 2 ) are well-known biomarkers for the early stage of colorectal cancer. 1,2 The up-regulation of COX-2 expression is associated with the activation of E-type prostanoid (EP4) receptors, 3

and the EP4
receptor-mediated signaling pathway is also associated with increases in cell motility and proliferation 4 in human colon cancer cells. Moreover, the up-regulated expressions of EP4 receptors, 5,6 and PGE 2 synthase 7 have been reported during the progression of colorectal cancer, and increased expression levels of PGE 2 synthase were shown to be mediated by the activation of EP4 receptors. 8 Thus, the activation of EP4 receptors establishes a positive feedback loop that may drive the expression of COX-2 and PGE 2 synthase, followed by the synthesis of PGE 2 ; that is, EP4 receptors are considered to play functional roles in the malignancy of colorectal cancer. 9 While EP4 receptors are generally implicated in colorectal carcinogenesis, they may also be involved in maintaining gastrointestinal homeostasis. 10 Normal colorectal epithelial cells have been shown to express EP4 receptors, 10 with strong expression in lateral crypt epithelia, 11 and have a turnover cycle of 3-5 days. 12 Thus, stem cells at the crypt bottom generate epithelial progenitor cells, these cells proliferate, migrate, and differentiate to the intestinal lumen, and then undergo apoptosis and/or extrusion into the lumen over 3-5 days. 13 New epithelial progenitor cells have been reported to accumulate β-catenin and stimulate β-catenin-mediated transcriptional activity, which induces cells to proliferate and migrate until they reach the midcrypt region. At the midcrypt region, cells inhibit β-catenin-mediated activity, which leads to cell cycle arrest and differentiation, followed by apoptosis and/or extrusion when cells reach the lumen surface. 13 Thus, β-catenin-mediated signaling may function as the dominant switch between the proliferation and differentiation of colorectal epithelial cells.
Although the up-regulated expression of EP4 receptors has been demonstrated during colorectal development, another study showed that the mRNA expression levels of EP4 receptors were higher in normal colon tissues than in cancer tissues. 14 The expression levels of EP4 receptors were previously reported to be regulated by Sp-1, a zinc finger transcription factor, binding to two Sp-1-binding sites in the human EP4 receptor promoter region from −197 to −160 bp. 15 However, another transcription factor, which is induced under hypoxic conditions, hypoxia-inducible factor (HIF)-1α, was shown to down-regulate the expression of EP4 receptors when human colon cancer HCA-7 cells were cultured under high cellular density conditions, the environment of the cells that is equivalent to over proliferated conditions. 16 Thus, the cellular density-dependent induction of HIF-1α protein expression down-regulates the expression of EP4 receptors in HCA-7 cells. 16 Since the expression of HIF-1α is known to be up-regulated and correlates with cancer malignancy, [17][18][19] massively proliferating cancer cells may exhibit the decrease in the expression of EP4 receptors.

HIF-1α, c-Myc, and Sp-1
The activation of β-catenin-mediated signaling was previously reported following the stimulation of EP4 receptors with PGE 2 . 20 The proliferation of colorectal cancer epithelial cells has also been suggested to be mediated by the effects of c-Myc induced by β-catenin-mediated signaling. 13 Therefore, the c-Myc-mediated proliferation of cells may be regulated via EP4 receptor activation. HIF-1α has been reported to bind to Sp-1 by displacing c-Myc from Sp-1-binding sites to reduce the DNA mismatch repair system in colon cancer cells. 21,22 Thus, regarding the regulation of EP4 receptor expression, c-Myc has been implicated in the up-regulation of receptors, whereas HIF-1α exerts the opposite effects, with the involvement of Sp-1.
We herein demonstrate that the expression of EP4 receptors is tightly regulated by c-Myc and HIF-1α by binding to Sp-1 as cellular density-dependently in HCA-7 cells. This tight regulation of EP4 receptor expression by c-Myc and HIF-1α may be an essential system for maintaining homeostasis in normal colorectal epithelial cells. However, once the system is altered, it may cause aberrant cellular proliferation, the transformation from normal to cancerous phenotypes, which represents the trigger for the early stage of colorectal carcinogenesis.

| Construction of deletion and mutated EP4
promoter luciferase reporter plasmids Deletion mutants, and point mutations introduced into the hypoxia response element (HRE) region, of human EP4 receptor promoter luciferase reporter plasmids were constructed using the human EP4 receptor promoter luciferase reporter plasmid, wild-type (WT) (−1238/ +1), 16 as a template. In order to construct the deletion mutants of human EP4 receptor promoter luciferase plasmids, primers with the following sequences were used for polymerase chain reaction (PCR) The point mutations introduced in HRE containing the human EP4 receptor promoter luciferase plasmid were amplified by PCR using the primers 5′-TCCGCACCCCCGAGGGAATGAAAACCACGGGAGCC-3′ (sense) and 5′-GGCTCCCGTGGTTTTCATTCCCTCGGGGGTGCGG A-3′. The deletion 3 (−197/+1) luciferase plasmids, in which point mutations were introduced at each or both Sp-1-binding sites, were constructed by PCR using the following primers: mut-A-del 3: 5′-GCCCAGCCCTTCCCCAGCCCA-3′ (sense) and 5′-TGGGCTGGGGAA GGGCTGGGC-3′ (antisense), mut-B-del 3: 5′-GCCCAGACACTT CCCCCCGCCA-3′ (sense) and 5′-TGGCGGGGGGAAGTGTCTGGG C-3′ (antisense) and both primer sets for mut-A,B-del 3. 15 The PCR products of each or both Sp-1 site point mutations introduced into del 3 plasmids were digested with Dpn I (Takara Bio) and self-ligated. Each construct was sequenced and verified.

| Luciferase assay
Cells were cultured under low (2 × 10 5 cells/each well) and high (2 × 10 6 cells/each well) cellular density conditions in 6-well plates, and culture medium was replaced with Opti-MEM I (Thermo Fisher Scientific) containing 100 UI/mL penicillin and 100 μg/mL streptomycin. Cells were transiently transfected with 10 μg/each well of firefly WT (−1238/+1), deleted or mutated reporter luciferase plasmids, and with 10 ng/each well of renilla luciferase control plasmids, pRL-CMV (Promega) using Polyethylenimine MAX (MW 40 000) (Polysciences, Warrington, PA) reagent. After approximately 6 hours, the transfection reagent was removed by a medium change using DMEM containing 10% FBS, and cells were incubated for a further 16 hours. To measure the effects of HIF-1α and c-Myc on EP4 receptor promoter activity, cells were cultured under low-density conditions, and culture medium was replaced with Opti-MEM I containing 100 UI/mL penicillin and 100 μg/mL streptomycin. In HIF-1α and c-Myc overexpression experiments, cells were cultured under low cellular density conditions in 12-well plates, and were transiently transfected with 2 μg/each well of firefly WT (−1238/+1) EP4 receptor promoter luciferase plasmids, 3 ng/each well of pRL-CMV renilla luciferase control plasmids, and either the pHA-N1 vector, which was created by replacing EGFP with HA in the pEGFP-N1 vector (Clontech Laboratories, Mountain View, CA), HA-HIF1alpha-pcDNA3 was a gift from William Kaelin (Addgene plasmid #18949) 23 or the pCMFlag_hsc-Myc (RDB06671, RIKEN BRC, Saitama, Japan) expression vector, using the same reagents described above, and cells were then incubated for a further 42 hours. In competition assays, HAtagged HIF-1α expression plasmids alone (0.5 μg/each well), or HAtagged HIF-1α expression plasmids (0.5 μg/each well) plus various amounts of FLAG-tagged c-Myc expression plasmids (0.05, 0.15, and 0.5 μg/each well, respectively); or FLAG-tagged c-Myc expression plasmids alone (0.5 μg/each well), or FLAG-tagged c-Myc expression plasmids plus various amounts of HA-tagged HIF-1α expression plasmids (0.05, 0.15, and 0.5 μg/each well, respectively), and the total amounts of transfected plasmids were adjusted by adding pHA-N1 control plasmids to 1 μg/each well. Reporter plasmid-transfected cells were then lysed and assayed using the Dual luciferase reporter assay system (Promega) according to the manufacturer's instructions with the GL-200 luminometer (Microtech Nichon, Chiba, Japan). Data were normalized by calculating the ratios of firefly luciferase scores to the corresponding renilla luciferase values.

| Western blotting
Regarding the detection of Sp-1, HIF-1α, and c-Myc, HCA-7 cells were cultured under low (2 × 10 5 cells/each well), middle dilution of an anti-HIF-1α antibody (H1alpha67); or a 1:5000 dilution of an anti-β-tubulin-antibody (014-25041; Wako). After being incubated with primary antibodies, membranes were washed three times and then incubated at room temperature for 2 hours with a 1:10 000 dilution of appropriate secondary antibodies conjugated with horseradish peroxidase under the same blocking conditions as those for the primary antibodies. 16 After washing three times, immunoreactivity was detected and visualized with ChemiDoc XRS Plus Image Lab (Bio-Rad Laboratories, Hercules, CA). In order to ensure the equal loading of proteins, membranes were stripped and reprobed with the anti-β-tubulin antibody under the conditions described above. The intensity of chemiluminescence was measured with ImageJ software (National Institutes of Health, Bethesda, MD).

| Statistical analysis
Data are expressed as the mean ± SD, and statistical analyses were performed using Prism 7 for windows or Prism 5 for Mac OS X software (GraphPad Software, La Jolla, CA). The t test or multiple comparison tests in the analysis of variance (ANOVA) were used to evaluate three or more independent experiments. Additionally, since the original luciferase counts vary greatly among the experiments because of the intrinsic low transfection efficiency of the HCA-7 cells, we normalized each control value as 100%. Therefore, the onesample t test was used to evaluate the experimental means ± SD against the control value (100%). Significance was assumed at P < 0.05. of EP4 promoter-containing reporter gene plasmids concomitantly with either HA-control vector plasmids or HA-tagged HIF-1α expression plasmids. Luciferase activity was assessed, as described in the Materials & Methods. Data are normalized to low cellular density-cultured cells transfected with WT, or HA control vector plasmid-transfected cells under low cellular density conditions as 100%. Data are the mean ± SD of three or more than three independent experiments. *P < 0.05, t test or one-sample t test, significantly different from low cellular densitycultured cells transfected with WT or mutated human EP4 receptor promoter plasmids. † P < 0.05, t test or one-sample t test, significantly different from HA control vector plasmid-transfected cells under low cellular density conditions. n.s.; not significant  each gene between cancer tissues and noncancer tissues were analyzed using the Mann-Whitney U-test. Significance was assumed at P < 0.05.

| Cellular density-dependent EP4 receptor promoter activities are mediated by HIF-1α
The expression of human EP4 receptors was previously reported to decrease in a cellular density-dependent manner in HCA-7 human colon cancer cells, and inversely correlated with HIF-1α expression levels. 9,16 One major HIF-1α-binding sequence is GCGTG, 24 namely HRE, which is located between −230 and −226 bp of the human EP4 receptor promoter region ( Figures 1A and E). In order to confirm HIF-1α-mediated cellular density-dependent decreases in EP4 receptor expression, deletion mutants of the human EP4 receptor promoter region connected to luciferase reporter genes were constructed and translational activities were assessed, as shown in Figure 1A. As we previously reported, 16  reporter gene plasmids, the cellular density-dependent reduction in EP4 promoter activity was canceled; therefore, low and high cellular density-cultured cells exhibited similar promoter activities.
The cellular density-dependent reduction in EP4 promoter activity was shown to be mediated by increases in the protein expression levels of HIF-1α. 16 Therefore, in order to confirm this, HA-tagged HIF-1α expression plasmids were transfected into low cellular density-cultured HCA-7 cells with the WT (−1238/+1), del 3 (−197/ +1), or del 4 (−160/+1) reporter gene plasmids. to those of HA-empty vector plasmid-transfected control cells, as observed for high cellular density-cultured cells shown in Figure 1A.
Similar results were obtained for HA-tagged HIF-1α with the del 3 (−197/+1) reporter gene plasmids in low cellular density-cultured cells ( Figure 1C and Supporting Information 1C). In contrast, when the del 4 (−160/+1) reporter gene plasmids were transfected with the HA-tagged HIF-1α expression plasmids shown in Figure 1D and Supporting Information 1D, no significant decrease or increase was observed. Thus, increases in HIF-1α expression appear to regulate the activation of cellular density-dependent EP4 receptor promoters acting between −197 and −160 bp.

| HRE may not be involved in cellular densitydependent EP4 receptor promoter activity
Cellular density dependency was also detected in the del 3 (−197/+1) reporter gene plasmids, which lack the HIF-1α-binding sequence HRE.
In order to examine whether HRE is involved in cellular densitydependent EP4 receptor promoter activity, point mutations were introduced into the HRE region of WT (−1238/+1) reporter gene plasmids, GCGTG (WT) to GAATG (mut-HRE), 24 as shown in Figure 1E.
Before investigating the cellular density dependency, the binding ability of mut-HRE to HIF-1α was assessed using the ChIP assay. Figure 1F showed that WT, but not mut-HRE, detected the HIF-1αbound DNA sequence, indicating that mut-HRE lost its binding ability to HIF-1α. Cellular density-dependent EP4 receptor promoter activity was then examined using the mut-HRE (−1238/+1) reporter gene plasmids. As shown in Figure 1G Figure 1B. These results indicate that HIF-1α, which directly binds to HRE, may not be involved in cellular densitydependent EP4 receptor promoter activity.

| Sp-1-binding motifs are required for cellular density-dependent EP4 receptor promoter activity
According to the results shown in Figure 1A, a key sequence for the HIF-1α-mediated cellular density-dependent reduction in EP4 receptor promoter activity may be laid on the sequence between −197 and −160 bp. As shown in Figure 2A and as reported previously, 15  Figure 1A.
These results indicate that at least one Sp-1-binding motif is required for cellular density-dependent EP4 receptor promoter activity. Additionally, loss of both Sp-1 motifs causes a substantial reduction in promoter activity even at low cellular density, as seen also with del 4 (−160/+1).
Based on the involvement of Sp-1 in the regulation of EP4 receptor expression, we then treated HCA-7 cells cultured under low and high cellular density conditions with mithramycin A, an antibiotic that binds to the Sp-1-binding site to displace Sp-1 and inhibit its activity.
As shown in the left two lanes of Figure 2C, EP4 receptor expression levels were significantly lower by approximately 30%-40%, in high density-cultured HCA-7 cells than in low density-cultured HCA-7 cells, and closely correlated with promoter activity, as shown in Figure 1A.
However, EP4 receptor expression levels decreased to similar levels in high and low cellular density-cultured HCA-7 cells treated with mithramycin A, and cellular density dependency was canceled (Figure 2C, right two lanes). These results indicate that the binding of Sp-1 to Sp-1-binding sites in the EP4 receptor promoter region is directly related to cellular density-dependent receptor expression.
We previously reported that HIF-1α protein expression levels increased in a cellular density-dependent manner with a negative correlation with EP4 receptor expression levels in HCA-7 cells. 16 Thus, since Sp-1 appeared to bind directly to the promoter region of EP4 receptors, the cellular density-dependent protein expressions of Sp-1 in HCA-7 cells were examined. As shown in Figure 2D, no significant differences in Sp-1 expression were observed among the different cellular densities tested. On the other hand, the expression of HIF-1α significantly increased in a cellular density-dependent manner ( Figure 2E), similar to that reported previously. 16

| The transcriptional activity of EP4 receptors may be oppositely regulated by HIF-1α and c-Myc on the promoter at Sp-1-binding sites
Previous studies reported that HIF-1α interacts with Sp-1, which binds to the promoter lacking the HRE region. 21,22 Moreover, HIF-1α has been shown to displace c-Myc from the promoter binding Sp-1, resulting in the repression of its promoter activity, that is, the MSH2 promoter of human sporadic colon cancer cells. 21,22 Thus, in order to examine whether cellular density-dependent EP4 receptor  HIF1AN]), copper metabolism domain-containing 1 (COMMD1), heat shock protein (HSP) 70 proteins, such as HSPA1A, and cAMP response element-binding protein-binding protein (CBP)/p300-interacting transactivator 2 (CITED2) (C), between cancer tissues (gray boxes) and paired noncancer tissues (white boxes). ‡ P < 0.05, the Mann-Whitney U-test, significantly lower than noncancer tissues. # P < 0.05, the Mann-Whitney U-test, significantly higher than noncancer tissues. n.s.; not significant (D) Schematic models depict the c-Myc-Sp-1 complex-mediated transcriptional activation of EP4 receptors in low cellular densitycultured HCA-7 cells. In high cellular density-cultured HCA-7 cells, increased HIF-1α competes with and displaces c-Myc for Sp-1 binding and followed by pulling Sp-1 out from its binding site, resulting in the down-regulation of EP4 receptor transcriptional activation increases in HIF-1α or c-Myc reduce the effects of its counterpart.
As shown in Figure 3C and Supporting Information 3C, HIF-1α-overexpressing cells exhibited significantly reduced WT (−1238/+1) reporter gene promoter activity, similar to that shown in Figures 1H and 3A. However, in HCA-7 cells cotransfected with c-Myc plasmid, the suppressed promoter activity recovered to the same levels as the control cells depending on the amount of the c-Myc plasmid.
Conversely, in c-Myc-overexpressing HCA-7 cells cotransfected with HIF-1α, the enhanced promoter activity declined to the same levels as the control cells depending on the amount of the HIF-1α plasmid ( Figure 3D and Supporting Information 3D). Finally, to confirm the cellular density-dependent recruitment of each Sp-1, HIF-1α, and c-Myc to the Sp-1-binding sites was analyzed by ChIP assay. As shown in Figure 3E, the significant cellular density-dependent decrease in binding ability to mut-HRE (−1238/+1) to Sp-1 and c-Myc, but not to HIF-1α, indicating that promoter region of EP4 receptor lost its binding ability to Sp-1 and/or c-Myc when cells were cultured in high cellular density.

| High cellular density-cultured HCA-7 cells showed significantly larger hypoxia-positive area than in low cellular density-cultured cells
As shown in Figures 2D, E, and 3B, the protein expression levels of Sp-1 and c-Myc were not altered in a cellular density-dependent manner, in contrast to HIF-1α. Hypoxia is one of the critical factors inducing HIF-1α, which is frequently overexpressed in many cancers including colon cancer. [17][18][19] Thus, in order to confirm whether the cellular density-dependent induction of HIF-1α was due to hypoxic cellular conditions, hypoxic areas were measured using the hypoxia probe. As shown in Figure 3F, the hypoxia-positive area in low cellular density-cultured HCA-7 cells was approximately 0.087% whereas that in high cellular density-cultured cells was approximately 1.076%, an area that was approximately 10-fold more significant than that in low cellular density-cultured cells. Thus, elevations in hypoxia under high cellular density conditions are a plausible reason for the upregulated expression of HIF-1α. Although the hypoxia-positive area was approximately 1% of the total area, the high density-cultured HCA-7 cells were in sphere-like/multilayers phase, so that there was a possibility that the hypoxia probe might not penetrate all the way down to the underlying and/or bottom cells. Therefore, even the majority of the cells did not appear to be hypoxia-positive, it was difficult to conclude that the bulk of the HCA-7 cells were not in hypoxia.

| HIF-1α protein levels may increase with decreases in the activity of its degradation pathways, which induce the HIF-1α-mediated down-regulation of EP4 receptor mRNA expression
In order to confirm the relevance of these results, an in silico analysis was performed using the Cancer Genome Atlas database; http:// xena.ucsc.edu. 25 As shown in Figure 4A, the expression levels of EP4 receptor mRNAs were significantly lower in colorectal cancer tissues (gray boxes) than in the corresponding normal tissues (white boxes). In contrast, the expression levels of c-Myc mRNAs were significantly higher in colorectal cancer tissues than in normal tissues.
The expression levels of Sp-1 as well as HIF-1α mRNAs were similar, with no significant differences being observed between cancer and normal tissues.
We previously reported that reductions in EP4 receptors in high cellular density-cultured HCA-7 cells by the induction of HIF-1α switched the responsible primary EP receptor subtypes from EP4 receptors to stationary EP3 receptors. 9, 16 We also demonstrated that the stimulation of EP3 receptors induced vascular endothelial growth factor (VEGF)-A 165 (VEGF-A) and VEGF receptor-1 (VEGFR-1, also known as FLT-1) in HCA-7 cells and HEK-293 cells stably expressing human EP3 receptors. 26,27 Thus, as shown in Figure 4B, the expression levels of VEGF-A and FLT-1 mRNAs were significantly higher in cancer tissues (gray boxes) than in normal tissues (white boxes). Although we have not yet proven whether increases in the induction of VEGF-A and FLT-1 mRNAs in cancer tissues, as shown in Figure 4B, are EP3 receptor-mediated events, the expression of these angiogenic-related factors is known to be regulated by HIF-1α. 28 Another well-known example of HIF-1α-regulated mRNA expression is glucose transporter (GLUT)-1 (also known as SLC2A1) mRNA, because the transport of glucose plays essential roles in the development of embryos in the relatively hypoxic environment of the placenta. 29 As shown in Figure 4B, the mRNA expression levels of SLC2A1 were also significantly higher in cancer tissues (gray box) than in normal tissues (white box). The stabilized protein expression levels of HIF-1α are also regulated by degradation rates through the ubiquitin-proteasome system under normoxic conditions. 28 Thus, the protein expression levels of HIF-1α shown in Figure 2E

| HRE-bound HIF-1α may be responsible for positively regulating basal EP4 receptor promoter activity
The up-regulated expression of EP4 receptors has been implicated in carcinogenesis. 5,6 However, the expression levels of EP4 receptors were also found to be significantly lower in tumor tissues than in normal tissues. 14 We previously demonstrated that EP4 receptor expression levels might be altered in a cellular density-dependent manner. 16 Thus, the cellular density-dependent down-regulation of EP4 receptors was previously shown to be regulated via the up-regulation of HIF-1α in HCA-7 cells. 16 In the present study, we elucidated the underlying mechanisms by which HIF-1α down-regulates EP4 receptor expression.
Although the existence of the HRE site, to which HIF-1α binds directly, was confirmed in the EP4 promoter region ( Figures 1E and   F), HRE was not responsible for the cellular density-dependent down-regulation of EP4 receptor expression. Since the knockdown of HIF-1α by siRNA altered the cellular density-dependent down-regulation of EP4 receptors as shown previously, 16 HIF-1α is a crucial factor for this regulation. Whereas, as shown in Figures 1G and H, mutation-induced HRE, which did not bind to HIF-1α ( Figure 1F), exerted negligible effects on cellular density dependency. However, based on the results shown in Figure 1A Figures 1G and H). Thus, in contrast to Sp-1-bound HIF-1α, HRE-bound HIF-1α may be responsible for positively regulating basal EP4 receptor promoter activity.

| HIF-1α and c-Myc may be critical factors for maintaining the homeostasis of colorectal epithelial cells by regulating EP4 receptor expression
As described, while EP4 receptors are generally known to be involved in colorectal carcinogenesis, they have also played a role in maintaining gastrointestinal homeostasis. 10 Thus, during the 3-5 days turnover of epithelial cells, β-catenin-mediated signaling is activated for proliferation and migration in the first half, followed by its inactivation for differentiation and apoptosis in the last half. Since EP4 receptors are known to activate β-catenin-mediated signaling, 20 the proliferation and migration of colorectal epithelial cells appear to be mediated by the activation of EP4 receptor-expressing cells in the first of the cycle; whereas, in the last half of the cycle, β-cateninmediated signaling is inhibited for differentiation and apoptosis, 13 and the present results may explain the underlying mechanisms.
Thus, due to the up-regulation of proliferation by EP4 receptor-acti- proliferation and, ultimately, a cancerous phenotype. 13 As shown in Figure 4A, the induction of c-Myc mRNA was significantly stronger in cancer tissues than in normal tissues, and the overexpression of c-Myc may be one of the first steps in carcinogenesis, which may be because of the overexpression of EP4 receptors. 5,6 Furthermore, an increase in the induction of c-Myc was not detected in HCA-7 cells, as shown in Figure 3B. This may have been because HCA-7 cells retain some features of normal colon epithelial cells. 33 However, it is more likely that HCA-7 cells are cancer cells in which c-Myc may already be overexpressed to nearly maximal levels. ously. 26,27 Thus, at the stage of HIF-1α abundance in colorectal epithelial cells, cells appear to progress to cancerous phenotypes and EP4 receptor-mediated homeostasis is no longer required.

| CONCLUSIONS AND PERSPECTIVES
Although we could not show the "direct" competition between HIF-1α and c-Myc on Sp-1 transcriptional factor, when the cellular switch for maintaining homeostasis mediated by c-Myc and HIF-1α for the Sp-1-binding balance is altered, such as the overexpression of EP4 receptors, the cells continue to grow aberrantly and become cancerous phenotypes. EP4 receptors appear to be required for the first step of carcinogenesis, as we have discussed previously, 34

DISCLOSURE
The authors declare no conflict of interest.