The zinc finger transcription factor Krüppel-like factor 4 (KLF4) is frequently downregulated in colorectal cancer. Previous studies showed that KLF4 is a tumor suppressor in the intestinal tract and plays an important role in DNA damage-repair mechanisms. Here, the in vivo effects of Klf4 deletion were examined from the mouse intestinal epithelium (Klf4ΔIS) in a genetic or pharmacological setting of colonic tumorigenesis: ApcMin/+ mutation or carcinogen treatment with azoxymethane (AOM), respectively. Klf4ΔIS/ApcMin/+ mice developed significantly more colonic adenomas with 100% penetrance as compared with ApcMin/+ mice with intact Klf4 (Klf4fl/fl/ApcMin/+). The colonic epithelium of Klf4ΔIS/ApcMin/+ mice showed increased mTOR pathway activity, together with dysregulated epigenetic mechanism as indicated by altered expression of HDAC1 and p300. Colonic adenomas from both genotypes stained positive for γH2AX, indicating DNA double-strand breaks. In Klf4ΔIS/ApcMin/+ mice, this was associated with reduced nonhomologous end joining (NHEJ) repair and homologous recombination repair (HRR) mechanisms as indicated by reduced Ku70 and Rad51 staining, respectively. In a separate model, following treatment with AOM, Klf4ΔIS mice developed significantly more colonic tumors than Klf4fl/fl mice, with more Klf4ΔIS mice harboring K-Ras mutations than Klf4fl/fl mice. Compared with AOM-treated Klf4fl/fl mice, adenomas of treated Klf4ΔIS mice had suppressed NHEJ and HRR mechanisms, as indicated by reduced Ku70 and Rad51 staining. This study highlights the important role of KLF4 in suppressing the development of colonic neoplasia under different tumor-promoting conditions.
Implications: The study demonstrates that KLF4 plays a significant role in the pathogenesis of colorectal neoplasia. Mol Cancer Res; 14(4); 385–96. ©2016 AACR.
Colorectal cancer is a major cause of cancer mortality in the United States. Several factors play a role in ultimately causing colorectal cancer development. These include mutations, epigenetic changes, and DNA damage. The majority of colorectal cancers contain mutations in the adenomatous polyposis coli (APC) tumor suppressor gene (1). APC is found in the normal intestinal mucosa with an increasing gradient of expression in mature epithelial cells located in the upper crypt region (2). In addition, APC antagonizes the pro-proliferative Wnt pathway by negatively regulating the steady-state level of intracellular β-catenin (3, 4). When APC is inactivated by mutation (which usually leads to a truncated protein), Wnt signaling is unimpeded, resulting in the nuclear accumulation of β-catenin and subsequent activation of downstream target genes, such as cyclin D1 and c-Myc, that promote cell proliferation (5,6). In many cancer types, including colorectal cancer, accumulation of DNA damage has been linked to cancer, and genetic deficiencies in DNA damage repair mechanisms are associated with susceptibility to tumor development (7). Additionally, epigenetic changes that lead to mutations or to silencing of DNA repair genes may promote tumorigenesis (7).
The nuclear transcription factor Krüppel-like factor 4 (KLF4; also known as gut-enriched Krüppel-like factor or GKLF) is highly expressed in the terminally differentiated, post-mitotic intestinal epithelial cells and is an inhibitor of cell proliferation (8,9). We have previously shown that in vitro, Klf4 is required for the prevention of genomic instability (10). In the intestine, the KLF4 promoter has been shown to be regulated by APC in a Cdx2-dependent manner, the latter being an intestine-specific transcription factor that controls intestinal development (11). Conversely, KLF4 has been shown to regulate colonic cell growth by inhibiting β-catenin activity (12, 13). Accordingly, studies have demonstrated a potentially causal relationship between KLF4 and several kinds of human cancers. For example, expression of KLF4 is often reduced in tumors of the gastrointestinal tract (14–16). In addition, loss of heterozygosity (LOH) and promoter hypermethylation have been identified as possible reasons for the reduced expression of KLF4 in a subset of colorectal cancers (16). However, whether KLF4 plays a causal role in the in vivo development of colonic tumors has not been definitively established.
In the current study, we investigated the in vivo role of KLF4 in colonic tumorigenesis in two different systems: the setting of ApcMin mutation and the chemically induced DNA mutations.
Materials and Methods
All animal studies were approved by the Stony Brook University Institutional Animal Care and Use Committee (IACUC). Male C57BL/6J ApcMin/+ founders were originally purchased from The Jackson Laboratory. Mice with floxed Klf4 gene (Klf4fl/fl) and intestine-specific villin-Cre–driven Klf4 deletion (Klf4ΔIS) were previously described (17). ApcMin/+ males were mated with either Klf4fl/fl or Klf4ΔIS females to obtain ApcMin/+ mice with intact Klf4 allele (Klf4fl/fl/ApcMin/+) or with intestine-specific Klf4 deletion (Klf4ΔIS/ApcMin/+).
Azoxymethane (AOM) treatment
Klf4fl/fl and Klf4ΔIS mice were injected with AOM (10 mg/kg), i.p., once a week for 4 consecutive weeks. Mice were euthanized at 4 or 12 weeks after the last injection for tumor development assessment.
Tissue harvesting and tumor assessment
For all groups, following euthanasia with CO2 asphyxiation and cervical dislocation, the entire small intestine and colon were dissected out and flushed with modified Bouin's fixative (50% ethanol/5% acetic acid) then cut open longitudinally. The intestines were examined under a dissecting microscope for the presence of adenomas. The number and size of adenomas in both the small and large intestine were recorded as described previously (18).
Tissue preparation and immunostaining
Following macroscopic examination for tumor formation in harvested tissue, the intestines were Swiss-rolled, fixed in buffered formalin, embedded in paraffin, and 5-mm sections were cut for histological hematoxylin and eosin (H&E) characterization, and for immunostaining.
For immunostaining, sections were deparaffinized in xylene, rehydrated in ethanol, and then treated with 10 mmol/L Na citrate buffer, pH 6.0, at 120°C for 10 minutes in a pressure cooker. The histological sections were incubated with blocking buffer [3% bovine serum albumin, and 0.01% Tween 20 in 1X Tris-buffered PBS (TTBS)] for 1 hour at 37°C. Primary antibodies goat anti-KLF4 (1:300; R&D), rabbit anti-cyclin D1 (1:200; Biocare Medical), rabbit monoclonal anti-Ki67 (1:200; Biocare Medical), rabbit anti-γH2AX (1:400; Abcam), rabbit anti-Rad51 (1:200; Abcam), mouse anti-Ku70 (1:200; Abcam), mouse anti-β-catenin (1:1,000; BD), mouse anti-Cdx2 (1:200; LS Bio), rabbit anti-HDAC1 (1:200; Santa Cruz Biotechnology), rabbit anti-Pp44/42 MAPK (p-ERK; 1:200; Cell Signaling Technology), and mouse anti-p300 (1:200; Santa Cruz Biotechnology) were added at 4°C overnight. For IF, appropriate AlexaFluor-labeled secondary antibodies (Molecular Probes) were added at 1:500 dilution in blocking buffer for 30 minutes at 37°C, counterstained with Hoechst 33258, mounted with Prolong gold (Molecular Probes), and cover slipped. IHC detection of cyclin D1, Ki67, and γH2AX was done using goat anti-mouse or anti-rabbit HRP-labeled (Jackson Immuno Research) secondary antibody at 1:500 dilution in blocking buffer for 30 minutes at 37°C, followed by wash and then DAB color development. For Klf4 and p-ERK detection by IHC, secondary unconjugated rabbit anti-goat antibody and goat anti-rabbit antibody (Jackson Immuno Research) was added, respectively, at 1:500 dilution in blocking buffer for 30 minutes at 37°C. After washing, goat anti-rabbit HRP and donkey anti-goat HRP–labeled tertiary antibodies (Jackson Immuno Research) were then added, respectively, at 1:500 dilution in blocking buffer for 30 minutes at 37°C, followed by DAB color development. Detection of all other primary antibodies for IHC was carried out using either Mach3 rabbit or Mach3 mouse HRP-polymer detection as per manufacturer's recommendations (Biocare Medical). Color development in all IHC staining was followed by hematoxylin counterstaining and mounting.
Anaphase bridging index (ABI) and mitotic index
The ABI was determined as described before (19). A minimum of 30 anaphases per mouse (4 per group) were scored from H&E sections. For Mitotic index, the number of cells undergoing mitosis per crypt (minimum of 60 per mouse) was scored from H&E sections (3 mice per group).
Measurement of LOH of the Apc+ locus
DNA was extracted from paraffin-embedded colon tissues of Klf4fl/fl/ApcMin/+ and Klf4ΔIS/ApcMin/+ mice, and LOH was analyzed as described before (18,20). In brief, five 10-μm-thick sections per mouse were cut and collected in a test tube. Tissue was then deparaffinized in 1 mL xylenes for 30 minutes at room temperature. The xylenes was then discarded, and tissue was washed twice in 100% ethanol followed by 2 times wash in 70% ethanol and 2 times PBS, then spun down and PBS discarded. DNA was extracted from pelleted tissue using the REDExtract-N-Amp Tissue PCR kit (Sigma) and purified using the QIAmp DNA Micro Kit (Qiagen). Equal amount of DNA was used for PCR amplification of the Apc locus followed by HindIII digestion as described before (20). Twenty microliters of each HindIII digestion reaction was electrophoresed through a 2.5% agarose gel. Bands were quantified using ImageJ software (21). Determination of LOH was carried out as described before (18,20).
In vitro overexpression or suppression of KLF4
The colon cancer cell line HCT116 was used. The cell line was originally purchased from ATCC. Cells thawed from frozen stock were used, and all experiments were carried out within 6 months of thawing. We routinely carried out morphology checks on all cell lines and we only passaged the cell lines for three months. In addition, the cell lines were tested by PCR for Mycoplasma contamination. Furthermore, each experiment had appropriate controls to ensure the behavior of tested cell lines. For Klf4 overexpression, plasmid containing pEGFP-Klf4 that expresses EGFP–Klf4 fusion protein was used to transfect the cells as described before (10). For KLF4 suppression, KLF4-specific siRNA (Ambion) was used to transfect cells as per manufacturer's recommendations. For all transfection experiments, the cells were harvested 24 hours after transfection for immunoblot analysis.
Cells were lysed in complete Laemmli buffer and vortexed for 3 to 4 minutes for homogenization. Insoluble material was removed by centrifugation at 12,000 rpm for 5 minutes, and the supernatant was collected and heated at 95-100°C for 10 minutes, then cooled down to room temperature before loading for gel electrophoresis. For mouse tissues (the entire length of the colon; normal epithelium and tumors), deparaffinization and hydration was done as described above. Reversal of crosslinking was then carried out by adding 1 mL of 10 mmol/L Na citrate buffer, pH6.0, per sample, and heating at 95°C for 20 minutes. Tissues were then spun down, citrate buffer was discarded, washed twice in PBS, then spun down and PBS discarded. Pelleted tissue was then re-suspended in complete Laemmli buffer and heated at 95 to 100°C for 10 minutes. Extracted protein from cells or tissue were resolved using SDS-PAGE gel electrophoresis. For all protein samples, following protein transfer, the membranes were immunoblotted with the following primary antibodies: rabbit anti-KLF4 (Santa Cruz Biotechnology), rabbit anti-HDAC1 (Santa Cruz Biotechnology), rabbit anti-p53 (Santa Cruz Biotechnology), rabbit anti-mTOR (Cell Signaling Technology), rabbit anti-phosphorylated mTOR (Cell Signaling Technology), rabbit anti-phosphorylated p70S6K1 (pS371; Cell Signaling Technology), rabbit anti-p27 (Santa Cruz Biotechnology), mouse anti-p21 (BD), mouse anti-Bax (Santa Cruz Biotechnology), mouse anti-β-actin (Sigma-Aldrich), and mouse anti-GAPDH (Sigma-Aldrich) overnight at 4°C. After being washed in TTBS, the blots were then incubated with appropriate horseradish peroxidase–conjugated secondary antibodies for 1 hour at room temperature. The antibody–antigen complex was visualized by ECL chemiluminescence (Millipore).
PCR and direct DNA sequencing
DNA from AOM-treated Klf4fl/fl and Klf4ΔIS mice was extracted from paraffin-embedded tissue sections as described above and purified using the standard phenol/chloroform/isopropanol method. Thus, the DNA extracted represents the entire length of the colon (normal epithelium and tumors) present in the section collected. PCR primers for detection of β-catenin and K-Ras mutations were designed to amplify the following: exon 3 of the Ctnnb1 gene containing the consensus sequence for GSK-3β phosphorylation and exon 1 of K-Ras (22). PCR amplification conditions were used as described before (22) using RED-Taq Ready PCR Mix (Sigma). Amplified fragments were then electrophoresed in 1% agarose gel, excised, and purified using the QIAEX II gel extraction kit (Qiagen). Sequencing was done at the Genomic Core Facility, Stony Brook University. Thus, the DNA sequenced represents the entire length of the colon (normal epithelium and tumors) present in the section collected.
Counting positively stained cells and statistical analysis
For p300, positively stained colonic epithelial cells were counted in a minimum of 20 crypts per mouse (3 mice/group). For Ku70 and Rad51 in adenomas, positively stained cells were counted per mouse (3 mice/group), per adenoma, per section, per field at 400× magnification. Four fields were counted per adenoma. Where applicable, differences in Rad51 IHC staining intensity in adenomas (2 mice/group) were quantified using ImageJ software (22). Statistical significance between groups was done using paired two-tailed Student t test or one-way ANOVA. Where applicable, box plot for data was done using GraphPad Prism version 5.00 for Windows (GraphPad Software).
Increased adenoma burden and elevated Wnt signaling level in the setting of ApcMin mutation following intestinal epithelium–specific deletion of Klf4
We previously showed that haploinsufficiency of Klf4 in ApcMin/+ mice leads to a significant increase in the number of adenomas formed in the small intestine but with no significant effect on adenoma development in the colon (18). In the current study, we investigated whether complete deletion of Klf4 in the intestinal epithelium (Klf4ΔIS) in the setting of ApcMin mutation has an influence on adenoma development in the colon, in addition to the small intestine. We first compared the number and size of the adenomas that developed in Klf4fl/fl/ApcMin/+ (which contain intact Klf4 loci) and Klf4ΔIS/ApcMin/+ mice between 16 and 20 weeks of age. Klf4fl/fl/ApcMin/+ mice developed on the average 19.88 ± 14.75 adenomas per mouse (N = 8) in the small intestine (Supplementary Fig. S1A). In comparison, the average number of adenomas in the small intestine of Klf4ΔIS/ApcMin/+ mice was 67 ± 27.21 adenomas per mouse (N = 7; Supplementary Fig. S1A). Examining the size of adenomas formed, Klf4ΔIS/ApcMin/+ mice had higher numbers of adenomas in each of the size categories than the Klf4fl/fl/ApcMin/+ mice, with significantly more adenomas of 1 to 2 mm (Supplementary Fig. S1B). There was no significant difference in the distribution of tumor sizes between the two groups of mice.
Because KLF4 has been shown to repress Wnt/β-catenin activity (13), we examined the levels of β-catenin, Ki67 and cyclin D1 by immunohistochemistry in the normal-appearing intestinal tissues and in adenomas of age-matched Klf4fl/fl/ApcMin/+ and Klf4ΔIS/ApcMin/+ mice. Both genotypes showed stronger β-catenin staining in adenomas as compared with their respective normal-appearing intestinal epithelial cells (Supplementary Fig. S2A–S2D). However, Klf4ΔIS/ApcMin/+ showed an overall stronger β-catenin staining as compared with Klf4fl/fl/ApcMin/+mice. A similar pattern of differential staining for Ki67 (Supplementary Fig. S2E–S2H) and cyclin D1 (Supplementary Fig. S2I–S2L) was observed in both mouse genotypes.
Increased tumor burden and penetrance of colonic polyps following deletion of Klf4 in the setting of ApcMin mutation
Analysis of adenoma number in the colon revealed that Klf4ΔIS/ApcMin/+ mice developed on average 3 adenomas per colon, while the Klf4fl/fl/ApcMin/+ mice had on average less than 1 adenoma (Fig. 1A). ApcMin/+ mice are known to develop adenomas mostly in the small intestine and very few in the colon. Importantly, 100% (7/7) of the Klf4ΔIS/ApcMin/+ mice developed colonic adenomas as compared with about only 50% (4/8) of the Klf4fl/fl/ApcMin/+mice (Fig. 1A). Also, unlike in the small intestine, deletion of Klf4 from the colonic epithelium (Fig. 1Ba and Bb) did not have an overall effect on β-catenin expression in Klf4ΔIS/ApcMin/+ mice as compared with Klf4fl/fl/ApcMin/+ mice in normal tissue (Fig. 1Bc and 1Bd). Similar finding was observed for β-catenin and cyclin D1 staining in adenomas of both mouse genotypes. However, adenomas of Klf4ΔIS/ApcMin/+ mice tended to have darker staining Ki67 than adenomas of Klf4fl/fl/ApcMin/+mice (Supplementary Fig. S3).
Increased LOH of wild-type Apc locus and increased anaphase bridge index (ABI) in colons of Klf4ΔIS/ApcMin/+ mice
Normal-appearing cells harboring pro-cancer genetic abnormalities are prone to develop into tumor cells if they are not eliminated, such as the case with ApcMin/+ mice. Consequently, we focused on examining normal-appearing colonic epithelial cells to determine the occurrence of preexisting abnormalities that might help explain the increased penetrance of colonic adenoma formation in Klf4ΔIS/ApcMin/+ versus Klf4fl/fl/ApcMin/+ mice. As polyps are initiated by LOH at Apc in ApcMin/+ mice (20), we determined the Apc genotype in the colon of Klf4fl/fl/ApcMin/+and Klf4ΔIS/ApcMin/+ mice, using tissues representing the entire length of the colon. In all mice examined, the amplified band for the wild-type Apc allele was weaker than that of the mutated allele. On the other hand, the gain in the mutated allele was significantly higher in Klf4ΔIS/ApcMin/+ mice than in Klf4fl/fl/ApcMin/+mice (Fig. 2A and B). LOH in ApcMin/+ mice is caused by homologous recombination near the centromere (23). Additionally, a mutation in DNA helicase or telomerase in ApcMin/+ mice increases the number of intestinal polyps because of high frequency of recombination and chromosomal instability (24). In the anaphase of cell cycle, such chromosomal instability can be assessed by the frequency of “anaphase bridges,” extended chromosome bridging between two spindle poles (24,25). We have previously shown that KLF4 suppresses genomic instability (10). To determine the effect of Klf4 deletion on genomic instability in mice colons, we scored ABI in the colonic epithelium (Fig. 2C). We found very few anaphase bridges in wild-type mice, whereas Klf4fl/fl/ApcMin/+mice had significantly more (ABI mean ± SD = 26.3% ± 3.17%; Fig. 2D). In contrast, the Klf4ΔIS/ApcMin/+ mice had ABI almost twice that of Klf4fl/fl/ApcMin/+mice (46.5% ± 2.4%; Fig. 2D). This result is consistent with previous reports that chromosomal instability is enhanced by mutations in Apc (26,27) or by deletion of Klf4 (10).
Deletion of Klf4 in colonic epithelium has no effect on Cdx2 expression in mouse colons
Previous studies have shown that increased tumor burden in the colon in a mouse model with Apc mutation is Cdx2 dependent (19). Additionally, Cdx2 was shown to regulate Klf4 expression in colon cancer lines (11). To determine whether Klf4 deletion has any effect on Cdx2 expression that may have led to the increase in colonic adenoma incidence in Klf4ΔIS/ApcMin/+ mice, we stained for Cdx2. As shown in Fig. 3A and B, we observed no differences in staining between Klf4fl/fl/ApcMin/+ and Klf4ΔIS/ApcMin/+ mice, where there is a decreasing gradient from proximal to distal colon.
Klf4 modulates mTOR pathway activity in colonic epithelium and in colon cancer cell line HCT116
Increased colonic polyposis following reduction of Cdx2 in a mouse model with mutated Apc was linked to an activated mTOR kinase–dependent pathway (19). Because KLF4 was previously shown to be downstream from Cdx2 (11), we tested whether Klf4 deletion in the colonic epithelium of ApcMin/+ mice has any effect on the activity of the mTOR pathway. Western blot analysis of colon tissue lysates showed that there was a relative increase in the level of p70S6K1 (pS371) in Klf4ΔIS/ApcMin/+ mice (Fig. 3C, lanes 4–6) as compared with Klf4fl/fl/ApcMin/+mice (Fig. 3C, lanes 1–3). Increased mTOR pathway activity in mouse colon has been shown to induce cell-cycle acceleration, which was associated with low levels of p27, which is an important cell-cycle regulator (20). Analysis of the level of p27 in colon tissue lysates showed that it was relatively decreased in Klf4ΔIS/ApcMin/+ mice (Fig. 3C, lanes 4–6) as compared with Klf4fl/fl/ApcMin/+mice (Fig. 3C, lanes 1–3).
The transcription factor p53 was known to be involved in multiple tumor-suppressive functions (28,29) and was shown to require KLF4 for the p53-dependent regulatory effects following DNA damage (30). Also, the ABI observed in Klf4fl/fl/ApcMin/+ and Klf4ΔIS/ApcMin/+ mice (Fig. 2) is a source of DNA damage. Thus, we examined the expression level of p53 in the colonic tissue lysate of both Klf4fl/fl/ApcMin/+ and Klf4ΔIS/ApcMin/+ mice. Compared with Klf4fl/fl/ApcMin/+ mice, the p53 level was elevated in Klf4ΔIS/ApcMin/+ mice (Fig. 3C, lanes 1–3 and 4–6, respectively), and this elevation was accompanied by an elevated Bax level (Fig. 3C, lanes 1–3 and 4–6, respectively).
The effect of Klf4 expression on the activity of the mTOR pathway and on the p53-dependent cell-cycle pathway was determined in the colon cancer cell line HCT116. Overexpression of Klf4 suppressed the activity of the mTOR pathway as indicated by lower levels of phosphorylated mTOR and p70S6K1 (pS371) as compared with control (Fig. 3D, lanes 1–2). In contrast, suppression of KLF4 leads to a relative increase in the phosphorylated mTOR and p70S6K1 (pS371) level (Fig. 3D, lanes 3–4). The effect of the Klf4 expression level on p53 expression was previously shown to be cell-type dependent (31,32). Consistent with previous reports (32), the p53 level was elevated and suppressed following Klf4 overexpression and suppression, respectively (Fig. 3D, lanes 1–4). The differences observed for the p53 expression level in response to suppressed KLF4 expression in the HCT116 cell line versus mouse colon lysates could be due to mutations in some regulatory genes harbored by the colon cancer cell line HCT116 as compared with those in the mouse colon tissue lysates. While the p21 level was not affected by the KLF4 expression level (Fig. 3D), there was an inverse correlation between the KLF4 and Bax levels (Fig. 3D, lanes 1–4), consistent with previous reports (33).
Altered expression of HDAC1 and p300 in colonic epithelium of Klf4ΔIS/ApcMin/+ mice
Acetyl modifications at histone tails constitute a major epigenetic mechanism that regulates chromatin structure and gene expression in cancer development (34). Klf4 was previously shown to suppress genomic instability in vitro (10), and to directly interact with modifiers of histone acetylation such as p300, a member of the histone acetyltransferase family (34). To determine whether deletion of Klf4 in the colonic epithelium of ApcMin/+ mice has any effect on the histone acetylation mechanism, we first examined the level of histone deacetylase 1 (HDAC1) and of p300, a histone acetyltransferase, by Western blot in colonic tissue lysates. While we could not detect p300, the HDAC1 level was elevated in Klf4ΔIS/ApcMin/+ mice (Fig. 3C, lanes 4–6) as compared with Klf4fl/fl/ApcMin/+mice (Fig. 3C, lanes 1–3). In HCT116 cells, the HDAC1 level was not affected by the Klf4 expression level (Fig. 3D), and we could not detect p300. To confirm the change in HDAC1 as observed by Western blot in colonic tissue lysates, we stained for HDAC1 and p300. Consistent with Western blot results, Klf4ΔIS/ApcMin/+ mice showed more colonic epithelium cells staining for HDAC1 as compared with Klf4fl/fl/ApcMin/+mice (Fig. 3E and F). On the other hand, while p300 was detected in Klf4fl/fl/ApcMin/+mice, it was significantly lower in Klf4ΔIS/ApcMin/+ mice (Fig. 3G, H, and I).
Suppressed DNA damage repair mechanisms in colonic adenomas of Klf4ΔIS/ApcMin/+ mice
We have previously identified an important role of Klf4 in DNA damage repair response and in maintaining genomic stability (10). Cells have evolved various strategies to promote genome stability through the precise repair of DNA double-strand breaks (DSB) and other lesions that are encountered during normal cellular metabolism and from exogenous insults (35,36). Given the role of Klf4 in mediating DNA damage repair, and that Klf4ΔIS/ApcMin/+ mice have higher markers of genomic instability compared with Klf4fl/fl/ApcMin/+mice (Fig. 2), we compared the level of γH2AX staining between Klf4fl/fl/ApcMin/+and Klf4ΔIS/ApcMin/+ mice. Normal-appearing colonic crypts of both genotypes had few cells staining positive for γH2AX (Fig. 4A and G, respectively). However, adenomas formed in both mouse genotypes stained strongly for γH2AX (Fig. 4D and J, respectively). To determine whether there is any difference between the two genotypes in the DNA damage repair mechanism, we stained for Rad51 and Ku70, representing homologous recombination repair (HRR) and nonhomologous end joining (NHEJ) repair mechanisms, respectively (37). Both mouse genotypes stained negative for Ku70 in the normal-appearing colonic epithelium (Fig. 4B and H, respectively). On the other hand, adenomas of Klf4fl/fl/ApcMin/+mice stained positive for Ku70, while adenomas of Klf4ΔIS/ApcMin/+ mice had significantly lower staining (Fig. 4E and K, respectively, and 4M). For Rad51, both mouse genotypes stained positive in the normal-appearing colonic epithelium (Fig. 4C and I, respectively), with no observed differences between them. However, adenomas of Klf4fl/fl/ApcMin/+mice had significantly more positive cells staining for Rad51 as compared with adenomas of Klf4ΔIS/ApcMin/+ mice (Fig. 4F and L, respectively, and 4N). Our results suggest an important role of Klf4 in promoting repair of DNA DSB by modulating the HRR and NHEJ mechanisms.
Susceptibility of Klf4ΔIS mice to AOM-induced colonic tumor formation
AOM is a colon carcinogen that is commonly used in rodents to study the pathogenesis of sporadic colorectal cancer. In mice, AOM-induced tumors have been attributed to mutations in β-catenin, while rare mutations in K-Ras are also observed (22,38). To determine whether deletion of Klf4 in the colonic epithelium has any effect on the pathogenesis of AOM-induced colorectal cancer, both Klf4fl/fl and Klf4ΔIS mice were treated with AOM (4 injections, once a week) and then analyzed for tumor development at 4 and 12 weeks after the last injection. At 4 weeks after injection, no difference was found between treated Klf4fl/fl and Klf4ΔIS mice when examined macroscopically. However, H&E staining of the colonic sections revealed the formation of microadenomas in 25% versus 75% of Klf4fl/fl in comparison to Klf4ΔIS mice (Supplementary Fig. S4Aa and S4Ab). The number of microadenomas per section averaged 0.5 ± 0.57 and 2 ± 0.82 in Klf4fl/fl and Klf4ΔIS mice, respectively (data not shown). Staining for β-catenin (Supplementary Fig. S4Ac and S4Ad) showed slightly lighter staining in the normal-appearing colonic epithelium of Klf4fl/fl mice compared with Klf4ΔIS mice, and darker staining in the microadenomas as compared with the surrounding normal-appearing colonic epithelium. While there was no observed difference in cyclin D1 staining at the basal level between Klf4fl/fl and Klf4ΔIS mice, more intense staining was found in the microadenomas as compared with the normal-appearing colonic epithelium (Supplementary Fig. S4Ae and S4Af).
At 12 weeks after AOM treatment, Klf4ΔIS mice developed significantly more colonic adenomas as compared with Klf4fl/fl mice (Fig. 5A). Adenomas of both Klf4fl/fl and Klf4ΔIS mice showed darker β-catenin staining than those of the surrounding normal colon epithelium. Although there was variation in β-catenin staining in adenomas of Klf4ΔIS mice, adenomas of Klf4ΔIS mice had an overall more intense and nuclear staining as compared with adenomas of Klf4fl/fl mice (Figs. 5Ba and 5Bb). Cyclin D1 showed a similar staining pattern between the two groups as in β-catenin (Figs. 5Bc and 5Bd). We also observed an increase in cyclin D1 staining in the normal-appearing colonic epithelium of both genotypes when compared with mice at 4 weeks (data not shown). Because AOM treatment was shown to have a minimum effect on inducing K-Ras mutations in mice, we set to determine whether Klf4 deletion would have any effect on the K-Ras pathway. When stained for p44/42 MAPK (p-ERK), which is a downstream effector of K-Ras, only 20% (1/5) of Klf4fl/fl mice showed any positive staining for p-ERK in the tumors, while 75% (3/4) of Klf4ΔIS mice showed positive staining for p-ERK (Figs. 5Be and 5Bf, respectively).
Analysis of AOM-induced β-catenin and K-Ras mutations showed no mutations in β-catenin in either genotype, contrary to what was reported before (22,38). However, consistent with previous reports (22,38), one AOM-treated Klf4fl/fl mouse (1/4) had K-Ras mutations (Fig. 5C), while 3 of 4 of AOM-treated Klf4ΔIS mice had K-Ras mutations (Fig. 5C; Supplementary Fig. S4B). These mutations were detected at 12 weeks after injection. Codons affected were 14, 16–18, 22 and 23 in the AOM-treated Klf4fl/fl mouse, and codons 14, 17–19, 21, and 22 in AOM-treated Klf4ΔIS mice.
We also examined the DSB repair mechanism in the AOM model of colonic tumorigenesis. Normal-appearing colonic crypts of AOM-treated mice of both genotypes had few cells/crypt staining positive for γH2AX (Fig. 6A and G, respectively). Adenomas formed in both mouse genotypes stained positive for γH2AX (Fig. 6D and J, respectively). Both mouse genotypes stained negative for Ku70 in the normal-appearing colonic epithelium (Fig. 6B and H, respectively). While adenomas of AOM-treated Klf4fl/fl mice stained positive for Ku70, adenomas of AOM-treated Klf4ΔIS mice stained significantly lower (Fig. 6E and K, respectively, and 6M). For Rad51, both mouse genotypes stained positive in the normal-appearing colonic epithelium (Fig. 6C and I, respectively). On the other hand, adenomas of AOM-treated Klf4fl/fl mice had Rad 51-positive staining, while adenomas of AOM-treated Klf4ΔIS mice showed significantly weaker staining for Rad51 (Fig. 6F and L, respectively, and 6N). Our results are consistent with the results shown in Fig. 4, demonstrating an important role of Klf4 in promoting repair of DNA DSB by modulating HRR and NHEJ mechanisms. Additionally, Western blot analysis of the p53 level in colonic tissue lysates of AOM-treated Klf4fl/fl and Klf4ΔIS mice indicated that, compared with Klf4fl/fl mice (Fig. 6P, lanes 1–4), Klf4ΔIS mice had a higher p53 level (Fig. 6P, lanes 5–8). This result is consistent with our observation on p53 in Klf4fl/fl/ApcMin/+and Klf4ΔIS/ApcMin/+ mice (Fig. 4).
Many factors such as mutations, epigenetic changes, and DNA damage contribute to the development of colorectal cancer. In humans, familial adenomatous polyposis is a disorder in which germline mutations of the APC gene lead to florid colonic polyposis early in life (1). The ApcMin/+ mouse carries a nonsense germline mutation in the murine Apc gene (39). A key difference between the human disorder and the mouse model is that adenomas mostly occur in the small intestine compared with the colon in ApcMin/+ mice (39), whereas it is the reverse in humans. We have previously shown that heterozygous whole-body deletion of Klf4 in the setting of ApcMin/+ mutation leads to significant increase in intestinal adenomas as compared with ApcMin/+ mice (18). In support of our previous finding, our current results show that complete deletion of Klf4 in the intestinal epithelium of ApcMin/+ mice (Klf4ΔIS/ApcMin/+) leads to a significant increase in the number of intestinal adenomas but not in size as compared with ApcMin/+ mice (Klf4fl/fl/ApcMin/+; Supplementary Fig. S1). These results indicate that in the small intestine Klf4 plays a role in adenoma multiplicity (and possibly initiation) rather than their progression.
The Wnt signaling pathway is hyperactivated in the setting of Apc mutation, both in humans and in mice (40,41). KLF4 was previously shown to negatively regulate the Wnt signaling pathway (13,42). Additionally, we have shown that ApcMin/+ mice with Klf4 haploinsufficiency have higher levels of Wnt signaling components, such as β-catenin and cyclin D1 (18). Consistent with these findings, Klf4ΔIS/ApcMin/+ had higher levels of β-catenin, proliferation, and cyclin D1 in both the normal-appearing intestinal epithelium and adenomas, as compared with Klf4fl/fl/ApcMin/+ (Supplementary Fig. S2). These results indicate an important role of Klf4 in suppressing the Wnt signaling pathway in the intestinal epithelium, and thus in adenoma formation, in the setting of ApcMin/+ mutation.
Mice with ApcMin/+ mutation develop adenomas mainly in the small intestine, with much lower incidence and multiplicity in the colon (39,41). We have previously shown that there was no significant difference in colonic adenoma incidence and multiplicity between ApcMin/+ mice and ApcMin/+ mice with Klf4 haploinsufficiency (18). However, an analysis of adenoma development in Klf4fl/fl/ApcMin/+ and Klf4ΔIS/ApcMin/+ mice showed a significant increase in colonic adenoma number and penetrance in Klf4ΔIS/ApcMin/+ as compared with Klf4fl/fl/ApcMin/+ mice (Fig. 1A), suggesting that Klf4 might be a key factor in suppressing colonic tumorigenesis. In mice, factors that determine tumor development in the small intestine versus the colon are not very well understood. Analysis of chromosomal instability in the colonic tissues Klf4ΔIS/ApcMin/+ mice showed significant increase both in LOH of the Apc locus and in aberrant mitosis as indicated by elevated ABI (Fig. 2). These in vivo results lend support to our previous finding about the role of KLF4 in the regulation of chromosomal instability in vitro (10). It was previously reported that Cdx2, a mouse homolog of Drosophila melanogaster caudal1 (43) and a key transcription factor for intestinal development and differentiation (44), is important in regulating colonic tumorigenesis through mTOR-mediated chromosomal instability (19). KLF4 has been shown to be downstream from Cdx2 (11), and this was supported by our finding that deletion of Klf4 had no effect on Cdx2 expression in the colonic tissue (Figs. 3A and B). However, analysis of the mTOR pathway revealed a potential role for Klf4 in negatively regulating mTOR signaling because there was an inverse relationship between the levels of Klf4 and an active mTOR pathway in mouse colonic epithelial cells and in colon cancer cell lines (Figs. 3C and D). This result is in line with a recent finding of the inverse relationship between Klf4 and mTOR signaling in mouse embryonic fibroblasts (45). Several studies have implicated a positive cross-interaction between mTOR signaling and epigenetic events, such as histone acetylation, regulated by HDAC1 (46,47). Additionally, an inverse interplay between Klf4 and HDAC1 has been previously reported (48). Our assessment of histone acetylation regulation in normal colonic epithelial cells of both Klf4fl/fl/ApcMin/+ and Klf4ΔIS/ApcMin/+ mice indicated elevated HDAC1 and reduced p300 staining levels in Klf4ΔIS/ApcMin/+ as compared with Klf4fl/fl/ApcMin/+ mice (Fig. 3). Taken together, our findings are strongly suggestive of a role for KLF4 in suppressing colon tumorigenesis, in an ApcMin/+ setting, through the modulation of the mTOR signaling pathway and epigenetic changes via the regulation of HDAC1 and p300 and consequently chromosomal instability.
Carcinogens such as AOM are known to induce colonic tumors in mice by promoting mutations in β-catenin and to a lesser extent in K-Ras (22,38). In our model, AOM-treated C57BL/6 Klf4fl/fl and Klf4ΔIS mice, though both developed colonic tumors with Klf4ΔIS mice having significantly more than Klf4fl/fl mice, yet neither group had β-catenin mutations, nor did they have K-Ras mutations in the commonly affected codons 12 and 13 (38). This could either be due to differences in the mouse strain used in our study versus those used in other studies (22,38), or due to differences in treatment regimen and tissue collection time point (22,38). Our results indicate that the increase in β-catenin expression in tumors of AOM-treated C57BL/6 mice is independent of β-catenin mutations, and in Klf4ΔIS mice corresponds to an absence of Klf4. Additionally, our results show that following exposure to certain carcinogens such as AOM, Klf4 is important in suppressing mutations in pro-oncogenic genes such as K-Ras.
DNA damage normally occurs in cells as a consequence of both environmental and endogenous insults. During the normal course of DNA replication or following exposure to DNA damaging agents, DSBs may arise and are considered one of the most cytotoxic forms of DNA damage (49). Deficiencies in DSB repair can lead to mutations and chromosomal aberrations that ultimately may result in genomic instability and tumorigenesis (49). Consequently, cells have effective mechanisms for the accurate and timely repair of DSBs in DNA (49). In mice, the importance of Klf4 is demonstrated by its ability to modulate DSB events and regulate DSB repair mechanisms (Figs. 4 and 6). Our results indicate that in the mouse intestinal epithelium, under basal conditions, the NHEJ repair mechanism is inactive, while the HRR mechanism is active and is independent of Klf4 expression (Figs. 4 and 6). However, in colonic adenomas both NHEJ and HRR mechanisms are active, and are dependent on Klf4 expression as both mechanisms were suppressed in the absence of Klf4. This differential activation of HRR versus NHEJ in normal epithelium versus adenomas points to a crucial role for Klf4 in regulating DNA damage repair pathways under pathologic conditions. Several factors play important roles in regulating DNA damage repair mechanisms, for example, the tumor suppressor p53. p53 acts as an important link between upstream signaling and activation of downstream signaling cascades depending on the extent of DNA damage and can activate cell-cycle arrest and allow the damage to be repaired, or it could transactivate genes involved in the apoptotic machinery (50). Surprisingly, the primary effect of p53 on DNA repair mechanisms was shown be the inhibition of the HRR and of NHEJ repair mechanisms (50). Our previous work has shown Klf4 to be a crucial mediator of p53-dependent cell-cycle arrest and DNA damage repair machinery and to suppress p53-dependent apoptosis (30,33). Our data from Klf4ΔIS/ApcMin/+ mice, where there was an elevated p53 level and suppressed HRR and NHEJ mechanisms in the absence of Klf4, are in line with these reports. Elevated p53 level and suppressed HRR and NHEJ repair mechanisms in the absence of Klf4 were also observed in AOM-treated mice, adding validation to our observation on the p53 level in Klf4fl/fl/ApcMin/+ and Klf4ΔIS/ApcMin/+ mice. Together, our data hint to a potential mechanism by which Klf4 promotes damaged DNA repair. It is possible that one way Klf4 promotes DNA damage repair, at least in the colonic epithelium, is by counteracting the suppressive effects of p53 on HRR and NHEJ repair mechanisms, thus allowing for the DNA repair to proceed. The exact mechanism by which Klf4 differentially regulates HRR and/or NHEJ DNA DSB repair pathways remains to be investigated.
Our findings are summarized in Fig. 7, where it is shown that Klf4 suppresses colonic tumor formation in association with ApcMin/+ mutation by both suppressing the mTOR pathway and reducing rates of precancerous epigenetic alterations. In response to the colon carcinogen AOM, Klf4 suppresses carcinogen-induced K-Ras mutations. In both mouse models, Klf4 is involved in DNA DSB repair by differential regulation of HRR versus NHEJ in normal versus tumor tissue. The exact mechanism by which Klf4 suppresses tumorigenesis under these seemingly different conditions requires further investigation. In conclusion, in the mouse intestinal mucosa, KLF4 plays an integral role in suppressing tumor formation under conditions of genetic mutations, epigenetic alterations, aberrant DNA damage repair, and carcinogen-induced mutations. Thus, modulation of KLF4 expression in the intestinal epithelium represents a potential therapeutic approach to prevent colon cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: A.M. Ghaleb, V.W. Yang
Development of methodology: A.M. Ghaleb, V.W. Yang
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.M. Ghaleb, E.A. Elkarim, V.W. Yang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.M. Ghaleb, V.W. Yang
Writing, review, and/or revision of the manuscript: A.M. Ghaleb, A.B. Bialkowska, V.W. Yang
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.M. Ghaleb, V.W. Yang
Study supervision: A.M. Ghaleb, V.W. Yang
The authors thank pathologist Dr. Kenneth R. Shroyer, Department of Pathology, Stony Brook University, for examining stage of adenomas formed.
This work was supported by grants from the NCI (CA084197 and DK052230) awarded to V.W. Yang.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.