The Intestinal Epithelial Cell Differentiation Marker Intestinal Alkaline Phosphatase (ALPi) Is Selectively Induced by Histone Deacetylase Inhibitors (HDACi) in Colon Cancer Cells in a Kruppel-like Factor 5 (KLF5)-dependent Manner*

Background: Differentiation induction represents a potential cancer treatment strategy. Results: Colon cancer cell lines respond differentially to HDACi-mediated induction of the differentiation marker ALPi. HDACi induction of ALPi is KLF5-dependent. Conclusion: HDACi induce ALPi in a subset of colon cancer cell lines in a KLF5-dependent manner. Significance: Colon cancer cell lines are differentially responsive to HDACi-induced differentiation. The histone deacetylase inhibitor (HDACi) sodium butyrate promotes differentiation of colon cancer cells as evidenced by induced expression and enzyme activity of the differentiation marker intestinal alkaline phosphatase (ALPi). Screening of a panel of 33 colon cancer cell lines identified cell lines sensitive (42%) and resistant (58%) to butyrate induction of ALP activity. This differential sensitivity was similarly evident following treatment with the structurally distinct HDACi, MS-275. Resistant cell lines were significantly enriched for those harboring the CpG island methylator phenotype (p = 0.036, Chi square test), and resistant cell lines harbored methylation of the ALPi promoter, particularly of a CpG site within a critical KLF/Sp regulatory element required for butyrate induction of ALPi promoter activity. However, butyrate induction of an exogenous ALPi promoter-reporter paralleled up-regulation of endogenous ALPi expression across the cell lines, suggesting the presence or absence of a key transcriptional regulator is the major determinant of ALPi induction. Through microarray profiling of sensitive and resistant cell lines, we identified KLF5 to be both basally more highly expressed as well as preferentially induced by butyrate in sensitive cell lines. KLF5 overexpression induced ALPi promoter-reporter activity in resistant cell lines, KLF5 knockdown attenuated butyrate induction of ALPi expression in sensitive lines, and butyrate selectively enhanced KLF5 binding to the ALPi promoter in sensitive cells. These findings demonstrate that butyrate induction of the cell differentiation marker ALPi is mediated through KLF5 and identifies subsets of colon cancer cell lines responsive and refractory to this effect.

cancer cell monolayers (4), the formation of dome-like structures that are a feature of vectorial water transport (4), and by the increased expression of intestinal alkaline phosphatase (ALPi) (4).
ALPi is found in high concentrations in the brush border of intestinal epithelial cells, primarily in the small intestine, and catalyzes the hydrolysis of phosphomonoesters with release of inorganic phosphate and alcohol (5). ALPi is also found in surfactant-like particles, which are produced by enterocytes and secreted in two directions, toward the intestinal lumen and toward the bloodstream, suggesting a role in regulating the rate of lipid absorption (6). Coincident with such a role, ALPi levels are elevated in the serum following lipid consumption (6,7), and mice lacking ALPi gain more weight when fed a high fat diet due to accelerated lipid absorption (8). Recently, a role for ALPi in the detoxification of lipopolysaccharide and prevention of bacterial invasion across the gut mucosal barrier has also been demonstrated (9).
The mechanism by which sodium butyrate induces ALPi gene expression has been investigated in detail in colon cancer cells (10 -12) and shown to occur at the level of transcription, to require new protein synthesis, and to be dependent on a KLF/ Sp-related cis regulatory element located 76 bp upstream of the transcription start site (10). Although multiple candidate transcription factors that regulate basal ALPi expression have been identified, including members of the Sp transcription factor family (10), KLF4 (13), ZBP89 (14), and Cdx-1 (15), the transcription factor or factors that drive ALPi induction in response to butyrate treatment remain unknown.
The aim of this study was to further elucidate the mechanism by which the HDACi, sodium butyrate, induces ALPi expression in colon cancer cells and to determine whether colon cancer cell lines sensitive and resistant to differentiation induction could be identified. Screening of a panel of 33 colon cancer cell lines identified 19 lines (58%) in which butyrate failed to induce ALP enzyme activity. These refractory lines were significantly enriched for those harboring the CIMP phenotype. Furthermore, through gene expression profiling and direct functional studies, we identified KLF5 as a key regulator of butyrate induction of ALPi in colon cancer cells.

EXPERIMENTAL PROCEDURES
Cell Culture-The source of the colon cancer cell lines utilized has been described previously (16). Cells were maintained in minimal essential media supplemented with 10% FBS, 1% HEPES buffer, and penicillin-streptomycin (50 units/ml and 50 g/ml, respectively) at 37°C with 5% CO 2 .
Quantitative Real-time PCR-Total RNA was extracted from untreated or HDACi treated cells using the RNeasy mini Plus kit (Qiagen). Five micrograms of total RNA were reverse transcribed using oligo(dT) primers and Superscript III (Invitrogen) and normalized to actin mRNA levels. Expression of ALPi was determined by quantitative real-time PCR using the Taq-Man assay Hs00897935_gH (Applied Biosystems, Foster City, CA). Expression of a second butyrate target gene, sodium-dependent inorganic phosphate cotransporter, member 7 (SLC17A7), was determined using SYBR green with the following primers: CTACGGGGTCTTTGCTTCTG (forward) and GGCTGAAATGTGCTGTGTGT (reverse) and normalized for actin expression (CACCTTCACCGTTCCAGTTT (forward) and GATGAGATTGGCATGGCTTT (reverse)).
ALP Enzyme Activity Assays-ALP activity was determined on lysed cell pellets by colorimetric assay as described by Young et al. (17) using p-nitrophenyl phosphate as substrate. ALP activity was expressed relative to total cellular protein quantified using the Bradford assay (18). ALP activity was also determined immunocytochemically using the Vector red alkaline phosphatase substrate kit I (Vector Laboratories, Burlingame, CA) according to the manufacturer's instructions.
ALPi Promoter Reporter Assays-ALPi promoter luciferase constructs have been described previously (14). The KLF/Sp site in the pFRL7-IAP-1 construct (renamed pFRL7-ALPi-1 in this work) was mutated by site-directed mutagenesis using the QuikChange II site-directed mutagenesis kit from Stratagene (La Jolla, CA). Cells were transiently transfected with ALPi promoter-reporter constructs using the Lipofectamine 2000 transfection reagent (Invitrogen) and treated with sodium butyrate for 24 h. Reporter activity was measured using the Dual-Luciferase reporter assay kit from Promega (Madison, WI) and normalized for induction of the control pFRL7 vector and total cellular protein.
Bisulfite Conversion and Sequencing-The methylation status of the ALPi promoter was determined by isolation of genomic DNA from each cell line using Qiagen's DNA extraction kit (Qiagen, Valenica, CA), and bisulfite converted using the Qiagen Epitect Bisulfite conversion kit (Qiagen). For each cell line, triplicate PCR amplifications were performed using Platinum TaqDNA High Fidelity Polymerase (Invitrogen). Primer sequences used to amplify ALPi were as follows: GGAA-GATTTAGTTTAGGTTTGGTTGA (forward) and ACCC-AAAACCCCTACATATCTTAAA (reverse). PCR products were pooled and ligated into the pCR2.1 TOPO TA vector (Invitrogen), and 1 l of each ligation used to transform TOP10FЈ competent cells. Clones were assessed for the presence of insert by restriction digestion and positive clones directly sequenced using the M13 primer. Sequence chromatograms were visually inspected then aligned with reference sequences to calculate the percent methylation at each CpG site using BiQ Analyzer (19) and JalView (20). A minimum of 10 clones were analyzed for each cell line.
Sequenom Assay-For assessment of ALPi gene promoter methylation using the Sequenom assay, 1.0 g of genomic DNA was bisulfite-treated using the Zymo Research EZ DNA Methylation kit with several changes as recommended for Sequenom processing. Two regions of the ALPi promoter were amplified using the following primers created using the Epidesigner program (F1, TTTGGTTTTAGGTTATAGGATTGGG; R1, AAACAAACCTTACTCACTCACCATC; F2, GGAAGATT-TAGTTTAGGTTTGGTTGA; R2, ACCCAAAACCCCTA-CATATCTTAAA). Forward primers were synthesized with a 10-mer tag (5Ј-AGGAAGAGAG-3Ј gene-specific sequence). All reverse primers were created with a T7 promoter tag (5Ј-CAGTAATACGACTCACTATAGGGAGAAGGCT-3Ј gene-specific sequence). PCR products were analyzed using the MASSARRAY platform, and data were analyzed using the EpiTYPER program.
CIMP Status-CIMP status was determined using the MethyLight assay for the five markers RUNX3, CACNA1G, SOCS1, NEUROG1, and IGF2 (21). As a control, we used a non-methylation-dependent reaction designed for a consensus Alu repetitive element sequence (21), which is a sensitive measure of very low amounts of DNA and is less prone to quantitation errors introduced by gene copy anomalies. A CIMP marker was considered methylation-positive by the presence of an amplification product for the methylation marker and the Alu reference control probe. Markers were considered methylation negative if no marker-specific probe amplification was detected in the presence of amplification of the Alu reference probe, with a C t value of Ͻ25. Cell lines were classified as CIMP-positive when three or more markers were positive for methylation (21).
Microarray Analyses-Colon cancer cell lines were treated with 5 mM sodium butyrate for 72 h. Total RNA was extracted, and gene expression changes were determined using Affymetrix U133 Plus microarrays (version 2.0) using standard protocols. For each cell line, experiments were performed on two independent occasions. The mean expression value was computed for each gene and utilized in subsequent analyses.
Statistical Analyses-Unpaired Student's t tests were used to compare groups and were two-sided unless stated otherwise. Chi square tests were used to compare associations between ALP induction and CIMP and microsatellite instability status. Values shown are mean Ϯ S.E., unless otherwise stated.

Identification of Colon Cancer Cell Lines Sensitive and Resistant to Butyrate Induction of ALPi-Sodium butyrate has been
shown to induce expression and activity of the marker of absorptive cell differentiation ALPi in colon cancer cells (23).
To determine the distribution of this effect across multiple colon cancer cell lines, we screened a panel of 33 cell lines for butyrate induction of ALP activity over a 144-h time course.  Table 1). Surprisingly, we also identified a second group of 19 cell lines (58%), which were largely resistant to butyrate induction of ALP activity (defined as Ͻ10-fold induction at all time points) ( Fig. 1B and supplemental Table 1). The differential induction of ALP enzyme activity was confirmed at the protein level ( Fig.  1C) and by immunocytochemical staining (Fig. 1D).
To determine whether the differential induction of ALP enzyme activity corresponded with differential induction of ALPi mRNA expression, four sensitive and four resistant cell lines were treated with butyrate for 72 h, and ALPi mRNA induction determined by quantitative real time PCR. Consistent with the induction of ALP enzyme activity, butyrate induced ALPi mRNA expression to significantly higher levels in sensitive (185.0 ϫ 10 Ϫ4 Ϯ 101.0 ϫ 10 Ϫ4 mean Ϯ S.E.) compared with resistant cell lines (1.0 ϫ 10 Ϫ4 Ϯ 1.0 ϫ 10 Ϫ4 , p ϭ 0.0146 unpaired t test) (Fig. 1E).
We next determined whether sensitivity or resistance to butyrate induction of ALPi was linked to known colon cancer subtypes. Of the sensitive lines, 64% were microsatellite stable and 71% were CIMP low. In contrast, among the butyrate-resistant lines, 58% exhibited microsatellite instability (p ϭ 0.30, Chi square test) and 68% were CIMP high (p ϭ 0.036, Chi square test) (Fig. 1, F and G).
We also confirmed the differential induction of ALPi mRNA in a sensitive and resistant cell line over an extended time course. As shown in Fig. 2A, butyrate induction of ALPi mRNA in the sensitive T84 cell line was first evident 24 -48 h posttreatment and increased progressively through 120 h. Conversely, in the resistant RKO line, minimal induction of ALPi mRNA was evident at each time point examined ( Fig. 2A). Notably, expression of an independent HDACi target gene, SLC17A7, was robustly induced in both T84 and RKO cells, indicating the failure to induce ALPi expression in RKO cells was specific to this gene and not a reflection of a general resistance of this line to butyrate (Fig. 2B).
Finally, we sought to determine whether the differential response of colon cancer cell lines to butyrate induction of ALPi was also consistent for other structurally distinct HDACi. To test this, the four sensitive and resistant cell lines were treated for 72 h with MS275, a representative of the benzamide class of HDACi. Similar to the effects induced by butyrate, induction of ALPi in response to MS275 was significantly higher in the sensitive compared with the resistant cell lines (51.0 ϫ 10 Ϫ4 Ϯ 16.0 ϫ 10 Ϫ4 versus 6.0 ϫ 10 Ϫ4 Ϯ 4.0 ϫ 10 Ϫ4 means Ϯ S.E., p ϭ 0.036, unpaired t test) (Fig. 3A). As observed for butyrate, MS275 induced expression of SLC17A7 to a similar extent in sensitive and resistant lines, indicating the failure to induce ALPi expression was specific to this gene (Fig. 3B).
Butyrate Induction of ALPi Promoter Activity Is Dependent on a KLF/Sp Regulatory Element in the ALPi Promoter-Butyrate induction of ALPi gene expression has been shown to be transcriptionally mediated and dependent on a KLF/Sp-related  cis regulatory element located 76 bp upstream of the transcription start site (10). To confirm this finding using promoter reporter assays, we transiently transfected butyrate responsive T84 cells with a series of 10 ALPi promoter reporter deletion constructs (14) and determined the effect of butyrate treatment on ALPi reporter activity. As shown in Fig. 4A, butyrate induced activity of all constructs (10.7-to 20.1-fold), however pFRL7-ALPi-10 which lacks the KLF/Sp regulatory element, was only induced 4.3-fold. To further confirm this finding, we performed site-directed mutagenesis of the KLF/Sp element, whereby the central GGC sequence was converted to GGA, GAA, or AAA. In comparison with the WT promoter that was induced 15.3fold by butyrate, activity was induced 12.6-, 3.4-, and 3.7-fold upon introduction of the GGC to GGA, GGC to GAA, and GGC to AAA mutations, respectively (Fig. 4B). These results confirm the previous findings implicating a critical role for the KLF/Sp element in butyrate induction of ALPi promoter activity.
The KLF/Sp-binding Site in the ALPi Promoter Is Selectively Methylated in Resistant Colon Cancer Cell Lines-Although the ALPi promoter does not harbor a bona fide CpG island, the presence of a CpG dinucleotide within the critical KLF/Sp regulatory element (GGGCGGG) and the association between CIMP-high status of colon cancer cell lines and resistance to butyrate induction of ALPi prompted us to examine whether methylation of this site and surrounding region could be the basis for resistance. To address this, genomic DNA was isolated and bisulfite converted from the four sensitive and four resistant cell lines. Two fragments of the ALPi promoter, one spanning the KLF/Sp site, were PCR-amplified and subjected to direct sequencing. Consistent methylation of a number of CpG sites, including the KLF/Sp site, was evident in all four resistant cell lines (Fig. 5, A and B). Conversely, these same sites were largely unmethylated in the four sensitive lines (Fig. 5, A and B). Notably, the differential methylation between sensitive and resistant cells was most prominent at the KLF/Sp site (Ϫ76 bp), where the average percentage methylation was 93.0 Ϯ 6.0 in the four resistant lines compared with 17.8 Ϯ 6.0 in the four sensitive lines (n ϭ 4, p Ͻ 0.00005, unpaired t test). To confirm this finding using an independent assay, DNA from these same cell lines was again bisulfite-converted and methylation status interrogated using the Sequenom assay. Consistent with the direct sequencing results, significantly higher methylation of multiple CpG sites within the ALPi promoter, including the KLF/Sp site, was observed in the refractory cell lines (Fig. 5, C  and D).

Butyrate Induction of Exogenous ALPi Promoter Activity Parallels Induction of Endogenous ALP Expression in Colon Cancer
Cell Lines-Although these findings demonstrate an association between ALPi promoter methylation status and the lack of inducibility of ALPi mRNA by butyrate, they do not directly establish methylation as the primary determinant for the failure to induce ALPi. To investigate this further, we transiently transfected an ALPi promoter reporter construct (pFRL7-ALPi-9) into the four butyrate sensitive and resistant cell lines and determined the change in promoter reporter activity in response to butyrate treatment. Butyrate induction of ALPi promoter activity was significantly higher in the four sensitive cell lines at all three concentrations tested, paralleling induction of the endogenous ALPi gene (Fig. 6). These findings suggest that although endogenous ALPi promoter methylation status was associated with inducibility, it may not be the primary determinant of this effect. Instead, these data suggest that butyrate induction of endogenous ALPi gene expression may be dependent on the presence or absence of specific transcriptional regulators.

Identification of Transcription Factors Differentially Expressed between ALP-sensitive and -resistant Cell
Lines-To identify this putative transcription factor or factors, we treated the four ALPsensitive and -resistant cell lines with 0 or 5 mM butyrate for 72 h and identified genes altered in expression using Affymetrix microarrays. Each cell line was profiled in duplicate. In addition, we used an existing in-house Affymetrix microarray database to identify transcription factors with basal differential expression between butyrate-sensitive and -resistant colon cancer cell lines. We applied the criteria that the candidate transcription factor(s) should be (1) selectively induced by butyrate in sensitive cell lines and/or (2) basally differentially expressed between sensitive and resistant cell lines. Furthermore, we focused specifically on those transcription factors known to bind KLF/Sp response elements, as this was an essential regulatory element required for butyrate induction of ALPi. The transcription factor that satisfied the first of these criteria with the strongest statistical significance was KLF5 (Fig. 7A). KLF5 mRNA expression was significantly induced following butyrate treatment in the four sensitive cell lines (p ϭ 0.01) but not the resistant lines (p ϭ 0.49) (Fig. 7A). The preferential induction of KLF5 by butyrate was independently confirmed by qRT-PCR and Western blotting in a time course analysis in two sensitive (T84, SKCO1) and two resistant (LIM2405, RKO) cell lines (Fig. 7, B and C). Furthermore, as shown in Fig. 7D, basal KLF5 expression was significantly higher in sensitive lines (p ϭ 0.004, unpaired t test).
KLF5 Overexpression Induces ALPi Promoter Activity-To determine whether KLF5 can induce ALPi promoter activity, we co-transfected the butyrate-resistant RKO cell line with a KLF5 expression vector in combination with an ALPi promoter reporter construct. As shown in Fig. 8A, KLF5 overexpression induced a 2-fold increase of ALPi promoter activity in RKO cells.
To determine whether KLF5 physically interacted with the KLF/Sp binding site in the ALPi promoter following butyrate treatment, we performed ChIP analysis in the butyrate sensitive T84 cell line. Consistent with induction of ALPi mRNA expression in this cell line, we observed a 27-Ϯ 6-fold increase in KLF5 occupancy of the ALPi promoter following butyrate treatment. We also observed a parallel increase in the active transcription marks AcH3 and H3K4Me3 following butyrate treatment. In contrast, no increase in KLF5 occupancy or of the active transcription marks H3Ac and H3K4Me3 were observed in the resistant RKO cell line following butyrate treatment (Fig. 8B).
Knockdown of KLF5 Attenuates Butyrate Induction of ALPi Gene Expression-To further determine whether KLF5 is directly required for butyrate induction of ALPi gene expression, we knocked down KLF in T84 cells using a pool of KLF5targeting siRNAs. Transfection with this siRNA pool resulted in a 45% down-regulation of KLF5 mRNA expression. To reduce the possibility of off-target effects, we also transfected T84 cells with the four individual siRNAs comprising the siRNA pool and identified siRNAs KLF5.2, KLF5.3, and KLF5.4 as able to inhibit KLF5 expression, whereas KLF5.1 was found to be non-functional (Fig. 9A). Transfection of T84 cells with the KLF5 siRNA pool, or siRNAs KLF5.2, 5.3, and 5.4 but not KLF5.1, significantly attenuated basal and butyrate induction of ALPi mRNA and protein expression (Fig. 9, B and C) and butyrate induction of ALP enzyme activity (Fig. 9, D and E). No effect on basal ALP enzyme activity was observed (Fig. 9D), possibly due to the contribution of other ALP isoforms (tissue-nonspecific (liver/bone/kidney) alkaline phosphatase, placental alkaline phosphatase, and placental-like alkaline phosphatase) to ALP activity. Collectively, these findings demonstrate that butyrate induction of the differentiation marker ALPi is dependent on induction of KLF5 and occurs in a subset of colon cancer cell lines.

DISCUSSION
ALPi is a widely used marker of absorptive intestinal epithelial cell differentiation (4). Here, we report that induction of ALPi mRNA and enzyme activity in response to butyrate treatment is heterogeneous among colon cancer cell lines, with sensitive and resistant cell lines readily identifiable.
The differential sensitivity of colon cancer cell lines to butyrate induction of ALP was also evident for the structurally distinct HDACi, MS-275, indicating butyrate induction of ALPi is most likely a consequence of HDAC inhibition. This finding is also important because although the clinical utility of butyrate is limited by its rapid metabolism, MS-275 (entinostat) is undergoing clinical testing for anti-tumor activity and was recently shown to improve overall survival in post-menopausal women with ER ϩ advanced breast cancer, when combined with the aromatase inhibitor exemestane (24).
Notably, cell lines resistant to butyrate induction of ALPi were significantly enriched for those harboring the CpG island methylator phenotype (CIMP), which manifests as the co-ordinate methylation of multiple loci (25). CIMP high colon cancers have a number of distinct clinicopathologic features, including preponderance in females, higher frequencies of BRAF mutations, microsatellite instability, and location of tumors within the proximal colon (26). The current findings demonstrate that an additional feature of this subset of colon cancers is increased resistance to differentiation induction by HDACi. Notably, CIMP high colon cancers are also associated with mucinous and poorly differentiated histology (26), raising the possibility that failure to undergo HDACi-induced differentiation may be linked to the poorly differentiated basal state of these tumors.
Consistent with the association with CIMP status, we also observed that resistant cell lines had significantly higher levels of ALPi promoter methylation, particularly of a CpG dinucleotide located within the key KLF/Sp regulatory element. However, butyrate induction of an exogenous ALPi promoter reporter paralleled induction of the endogenous gene across the cell line panel, suggesting the presence or absence of a key transcriptional regulator is likely to be the key determinant of butyrate induction of ALPi.
Butyrate-induction of ALPi mRNA is dependent on new protein synthesis and a key KLF/Sp regulatory element within the ALPi promoter (27). We therefore sought to identify transcription factors capable of binding this site, which were either selectively induced by butyrate in sensitive cells or which were basally differentially expressed between sensitive and resistant cell lines. This screen identified the C 2 H 2 zinc finger transcription factor KLF5 as both preferentially induced by butyrate and basally more highly expressed in sensitive cell lines. Consistent with ALPi being a direct target of KLF5, KLF5 overexpression induced ALPi promoter activity, KLF5 knockdown inhibited endogenous ALPi expression, and KLF5 was localized to the ALPi promoter following butyrate treatment.
A role for KLF5 in regulating expression of ALPi is consistent with its high level of expression in the intestinal tract (28) and animal studies demonstrating its requirement for normal intestinal cell differentiation (29,30). Specifically, variegated deletion of KLF5 in the intestinal tract results in loss of expression of the differentiation regulators Cdx-1 and Cdx-2, aberrant  localization of differentiated cells along the crypt-villus axis, and impaired barrier function (29). Similarly, a more recent study demonstrated that deletion of KLF5 in the fetal intestinal epithelium reduced expression of several regulators of intestinal cell differentiation (Elf3, Atoh1, Ascl2, Neurog3, HNF4␣, and Cdx1), impaired villus morphogenesis, and resulted in loss of the apical brush border characteristic of differentiated enterocytes (30). These findings highlight a key role for KLF5 in the regulation of intestinal cell differentiation, a role that has also been ascribed to KLF5 in adipocytes (31) and leukemic cells (32).
It is important to note, however, that KLF5 expression is maximal in the crypt base (33), more consistent with a role in cell proliferation. In agreement with such a role, intestine-specific inactivation of KLF5 resulted in neonatal lethality due to an absence of epithelial cell proliferation (29), and a recent study in which KLF5 was specifically deleted in LGR5 ϩ stem cells also demonstrated reduced intestinal cell proliferation (33). Collectively, these findings demonstrate that the role of KLF5 in the intestinal epithelium is complex, with roles in both proliferation and differentiation.
The mechanistic basis for the differential induction of KLF5 among colon cancer cell lines remains unknown. Notably, the KLF5 promoter is GC-rich, and its basal expression has been shown to be Sp1-dependent (34). Furthermore, methylation of the KLF5 promoter has been reported in acute myelogenous leukemia (32,35) and medulloblastoma (36). However, our examination of the methylation status of the KLF5 promoter in resistant colon cancer cell lines did not identify any evidence of methylation (data not shown), suggesting other mechanisms such as loss of key transcriptional regulators may be responsible.
Our finding that methylation of the ALPi promoter is linked to CIMP high status of colon cancer cells may provide some insight into the manifestation of the CIMP phenotype. The molecular basis for CIMP is currently unknown, although potential mechanisms, including overexpression and aberrant targeting of DNA methyltransferases, have been suggested (25). In the current study, we observed that the basal levels of KLF5 expression were significantly lower in resistant cell lines. This raises the interesting possibility that the reduced levels of KLF5 in resistant cell lines may predispose its target genes (ALPi) for methylation due to reduced occupancy of the promoter. Extending this further is the possibility that the CIMP phenotype could at least in part, manifest due to loss of transcription factors such as KLF5, which results in the coordinate predisposition of their binding sites to promoter methylation. This model is supported by several previous studies, which demonstrate that transcription factors can play an important role in protecting against promoter methylation (37). First, in colon cancers, a role for Sp1/Sp3 binding sites in methylation protection was recently suggested in a genome-wide ChIP on chip screen (38), as well as a locus specific study interrogating the RIL gene, where the presence of a 12-bp polymorphic sequence containing a Sp1/Sp3-binding site was linked to reduced methylation (39). Second, it has been shown that transient transfection of exogenously methylated DNA fragments containing Sp1 sites resulted in rapid demethylation in ES cells, whereas methylated fragments in which the Sp1 site was mutated remained methylated (40,41). Finally, several epigenetically silenced regions identified in prostate cancer comprise coordinately regulated gene families, including the Hox, metallothionein, keratin, and SERPINB families (42). Whether the coordinate methylation of these gene families is primarily mediated by their genomic proximity or linked to their coordinate transcriptional regulation is worthy of investigation.
In summary, we demonstrate that ALPi is selectively induced by HDACi in a subset of colon cancer cell lines. In particular, resistant cell lines were enriched for those with the CIMP high phenotype. Although we observed consistent methylation of the ALPi promoter in resistant cell lines, we also established a direct role for the transcription factor KLF5 in butyrate induction of ALPi and observed preferential induction of KLF5 by butyrate in sensitive cell lines. A possibility raised by these observations is that the methylation of ALPi may be a consequence of the absence of KLF5, which may have implications for understanding the manifestation of the CIMP phenotype in colon cancer.