EGFR Signals Downregulate Tumor Suppressors miR-143 and miR-145 in Western Diet–Promoted Murine Colon Cancer: Role of G₁ Regulators Free

Colon cancer is a leading cause of cancer-related deaths in the Western world (1). Tumors arise from colonic epithelial cells transformed by growth-promoting mutations (2). In prior studies, we showed that epidermal growth factor receptor (EGFR) signals were required for efficient colonic tumorigenesis and tumor progression in the azoxymethane (AOM) and azoxymethane/dextran sulfate sodium (AOM/DSS) models of sporadic and inflammation-associated colon cancer, respectively (3–5). In the AOM/DSS study, we examined the role of EGFR and diet in colonic tumorigenesis, comparing Egfr wild-type (Egfr^wt) mice to mice homozygous for hypomorphic waved-2 Egfr mutations (Egfr^wa2) that abrogate more than 90% of receptor kinase activity in vitro (6). In this model, we showed that tumor promotion by Western diet required EGFR signals (4). In the presence of wild-type EGFR, Western diet upregulated proto-oncogenes MYC and K-Ras (4, 7). These proto-oncogenes control several G₁ cell-cycle regulators and their dysregulations play critical roles in colonic tumor development in humans and experimental animals (8, 9).

Micro RNAs (miRNA) are small, noncoding RNAs that regulate gene expression by base pairing with mRNAs, leading to mRNA destabilization or inhibition of mRNA translation (10). In normal cells, miRNAs control numerous processes including stem cell fate, proliferation, and differentiation (11). Because in general, miRNAs target multiple mRNAs, individual miRNA could potentially alter complex cellular processes such as cell growth and apoptosis. Aberrant miRNA levels occur in many tumors including colon cancer (12). Some miRNAs that are lost during tumorigenesis seem to function as tumor suppressors, whereas other miRNAs are upregulated and might mediate proto-oncogenic signals. EGFR was recently shown to suppress several miRNAs that target proto-oncogenic transcription factors (13). MYC and Ras that are downstream effectors of EGFR in colonic tumorigenesis have been shown to regulate miRNA expression (14, 15). The role of miRNAs in mediating more proximal EGFR oncogenic effects, including tumor-promoting effects of Western diet, however, has not been examined in colonic tumorigenesis. The availability of sporadic and inflammation-associated colonic tumors from Egfr^wt and Egfr^wa2 mice fed standard or Western diet allowed us to ask whether EGFR and diet might regulate miRNAs in colonic tumorigenesis. For these studies, we examined tumors induced by AOM or AOM/DSS and also examined adenomas from Apc mutant Min mice, a genetic model of colon cancer. In addition, we investigated the expression levels of several orthologues of EGFR-regulated miRNAs identified in our mouse models in human sporadic and inflammation-associated colon cancers.

We identified miR-143 and miR-145 as negatively regulated EGFR targets by comparing their expression patterns in AOM and AOM/DSS-induced colonic tumors from Egfr^wt and Egfr^wa2 mice. These miRNAs form a polycistronic cluster located on human chromosome 5 and mouse chromosome 18. Prior investigations reported that these miRNAs were downregulated in human colon cancers (15, 16). Recent studies show that these miRNAs regulate embryonic and smooth muscle stem cell fates (17, 18). Furthermore, in vitro studies showed that transfected miR-143 or miR-145 inhibited the growth of colon cancer cells, consistent with their putative roles as tumor suppressors (19). These miRNAs were later noted to be decreased in other tumors including breast, lung, bladder, and prostate cancers (20–23). The availability of tumors from our experimental models provided an opportunity to investigate EGFR and Western diet regulation of demonstrated or in silico predicted targets of these miRNAs in colonic malignant transformation. We identified several putative targets of these miRNAs as G₁ regulators controlled by EGFR, thus linking EGFR regulation of miRNAs to cell-cycle control in colonic tumorigenesis. Our findings also suggest that tumor suppressor effects of miR-143 and miR-145 are uncovered in the presence of a Western diet that promotes colonic tumorigenesis and upregulates targets of these miRNAs that are no longer restrained in the absence of these miRNAs.

AOM was obtained from Midwest Research, the National Cancer Institute Chemical Carcinogen Reference Standard Repository. Standard chow diet and Western diet were prepared as described (4). FokI restriction enzyme for waved-2 genotyping was purchased from New England Biolabs. EGF was purchased from Calbiochem. C225 anti-EGFR neutralizing antibodies were obtained from ImClone. Gefitinib was provided by AstraZeneca. Polyclonal antibodies to extracellular signal–regulated kinase (ERK5; #3372) were obtained from Cell Signaling. Antibodies to K-Ras, (SC-522), MYC (SC-40, clone 9E10), CCND2 (SC-181), cdk6 (SC-177), and E2F3 (SC-879) were obtained from Santa Cruz Biotechnology. Rabbit polyclonal (#160106) and monoclonal (160112) anti-PTGS2 antibodies were purchased from Cayman Chemicals. MAP/ERK kinase (MEK2) antibodies were purchased from Epitomics. Monoclonal antibodies against Ki67 (clone SP1) were obtained from Neomarkers. Monoclonal ACTB (β-actin) antibodies were purchased from Sigma-Aldrich. Mimics of mature miR-143 and miR-145 were purchased from Ambion. Custom PCR primers were obtained from Integrated DNA Technologies, Inc. Other PCR reagents, including Moloney murine leukemia virus reverse transcriptase and random hexamers, were purchased from Applied Biosystems.

We examined tissue obtained from different studies to address the influence of EGFR on miRNAs in both sporadic and inflammation-associated colonic tumors. These models included AOM/DSS mouse, AOM (mouse and rat), and Min mouse model.

Flash-frozen normal mouse colonic mucosa and AOM/DSS-induced colonic tumors from mice fed standard rodent chow or Western diet were available from a prior study (5).

To examine the effects of Egfr genotype on miRNAs in a sporadic model of colon cancer, C57BL6/J Egfr^wt/wa2 mice were interbred with A/J Egfr^wt/wa2 mice to generate F1 hybrid C57BL6/J × A/J mice. Mice were genotyped to identify experimental groups Egfr^wt and homozygous Egfr^wa2 mice. We induced tumors by weekly injections with AOM [12.5 mg intraperitoneally (i.p.)/kg body weight] × 6 weeks. We used this AOM dose, as the hybrid mice were relatively resistant to AOM. Mice were fed standard or Western chow (20% fat) diets as described (4). Animals were sacrificed 40 weeks later and tumors were harvested.

Normal rat colonic mucosa and AOM-induced colonic tumors from rats fed standard chow or chow supplemented with gefitinib were available from a previous study (5).

We used frozen Apc mutant Min adenomas and adjacent normal appearing small bowel mucosa to assess miRNA in the Min model.

Human sporadic and ulcerative colitis (UC)-associated colonic adenocarcinomas and adjacent colonic mucosa, as well as normal colonic mucosa from colons resected for diverticular disease, were obtained from the Department of Surgical Pathology at the University of Chicago under an Institutional Review Board-approved protocol. Resected tissues were placed in an ice bath and transported promptly to the Department of Surgical Pathology. Representative tumor sections and adjacent normal appearing colonic mucosal sections were dissected free of underlying muscle and flash frozen in liquid nitrogen. Care was taken to avoid areas with visible necrosis.

RNA was extracted using RNeasy Lipid Tissue Mini Kit (Qiagen). RNA from murine AOM/DSS tumors was labeled with Hy3 and RNA from control mucosa was labeled with Hy5 using miRCURY Hy3/Hy5 power labeling kit (Exiqon). A separate array was used for each Hy3-labeled tumor. Hy5-labeled control RNA from animals matched for Egfr genotype was spotted on each array (n = 4 for each Egfr genotype). Samples were hybridized on the miRCURY LNA Array (v.9.2) using a hybridization station. The quantified signals (background subtraction) were normalized using the global lowess (LOcally WEighted Scatterplot Smoothing) regression algorithm that utilizes within-slide normalization to minimize dye-dependent differences in intensity. Exiqon carried out the miRNA labeling, hybridization, and preliminary array analyses.

RNA was transferred to nylon membranes using a full-immersion system. Oligonucleotides were synthesized with locked nucleic acids (LNA; Exiqon) and miRNA antisense probes generated by oligonucleotide end labeling with [γ³²P] ATP using T4 polynucleotide kinase. Filter hybridization was conducted overnight in hybridization solution containing radiolabeled probes (10⁶ cpm/mL), followed by several washes with 2× SSC/0.1% SDS. Membranes were stripped with several washes in 0.1%SDS at 95°C and re-probed with an LNA probe specific for U6 small nuclear RNA (snRNA) as a loading control.

For mRNA, 1 μg total RNA was reverse transcribed into cDNA using SuperScript III Platinum Two-Step qRT-PCR kit in 20 μL total volume. The resulting first-strand cDNA was used as template for quantitative PCR in triplicate using SYBR Green QPCR Master Mix kit. Primers were designed using Primer3 and sequences are available on request (24). Where possible, one of the primers was designed to span intron–exon junctions. Reverse transcribed cDNA (1 μL of 1:8 dilution) and primers were mixed with SYBR Green dye I master mixture in final volume of 25 μL. Negative controls (reactions lacking either reverse transcriptase or template) yielded no PCR products. Primers and TaqMan probes for the mature forms of miR-143 and miR-145 were obtained from ABI and conditions for reverse transcription and PCR followed the manufacturer's recommendations. Reverse transcriptase reactions were run in duplicate and PCR reactions in triplicate. Data were analyzed using the comparative 2exp(−ΔΔCt) method (25, 26). mRNA levels were normalized to β-actin and miRNA levels normalized to RNU48 (human tissue) or snoRNA202 (murine tissue) and expressed as fold-control (26).

We carried out in situ staining for miR-143 and miR-145 in human and mouse tissue from the AOM/DSS study using Exiqon miRCURY LNA oligonucleotide probes labeled 5′ and 3′ with digoxigenin (DIG) as described (27, 28). We used an Exiqon probe with a scrambled sequence as a negative control that gave no specific staining. Briefly, 10 μm cryostat sections were collected on Superfrost Plus slides and dried for 3 minutes at room temperature (RT) and stored at −80°C until staining. Sections were fixed in 4% (w/v) paraformaldehyde for 10 minutes and washed twice in diethyl pyrocarbonate (DEPC)-PBS for 3 minutes. Sections were incubated for 5 minutes in acetylation buffer followed by DEPC-PBS rinses. Sections were incubated for 60 minutes with biotinylated LNA probes (0.025 μmol/L) in hybridization buffer at temperature 25°C below melting temperature (T_m) for LNA probe. Sections were washed 3 times for 10 minutes in 0.1× SSC at 4°C to 8°C above hybridization temperature and then once for 5 minutes in 2× SSC at RT with agitation. Slides were treated for 20 minutes with 3% (v/v) H₂O₂ at RT and then for 30 minutes at RT in blocking buffer. Sections were incubated with primary antibodies (100 μL, 1:4,000; anti-DIG/HRP in blocking buffer) for 30 minutes at RT followed by 3 washes in TNT buffer at RT. For detection, slides were incubated with 1:50 dilution fluorescein isothiocyanate (FITC)-tyramide (100 μL) in amplification buffer for 10 minutes at RT in the dark. Then, 25 μL of ProLong gold antifade reagent was added and sections were secured with glass coverslips. Sections were imaged using an epifluorescence microscope equipped with charge-coupled device camera and image analysis software.

HCT116 colon cancer cells and CCD-18Co human colonic fibroblasts were obtained from American Type Culture Collection (ATCC). HCA-7 colorectal cancer cells were obtained from Susan Kirkland (Imperial Cancer Research Fund, London, UK). Cells were maintained at 37°C in a humidified atmosphere of 5% CO₂–95% air as recommended by ATCC. HCT116 and HCA-7 cells were cultured in McCoy's 5A modified medium containing 10% serum and CCD-18Co cells were cultured in Eagle's minimum essential medium with 15% serum. Young adult mouse colonocytes (YAMC) are a conditionally immortalized murine colonic epithelial cell line isolated from the H-2Kb-tsA58 mouse expressing a heat-labile SV40 large T antigen (29). YAMCs were provided by the Digestive Diseases Research Core Center at the University of Chicago and used between passages 25 and 32. Cells were grown on culture dishes in RPMI 1640 medium under permissive conditions at 33°C in a humidified atmosphere with 5% CO₂ until confluent.

To modulate EGFR signals, cells were stimulated with EGF (10 ng/mL) or EGFR was blocked with C225 neutralizing anti-EGFR antibodies (20 μg/mL) or EGFR kinase inhibitor gefitinib (60 nmol/L) for the indicated times. Control cells were treated with PBS. To examine the effects of miR-143 and miR-145 on cell growth and to assess putative targets of these miRNAs, cells were transfected with mimics of mature miR-143 or miR-145 (Ambion) at the indicated concentrations. Cell proliferation was measured using WST-1 assay (Roche) according to the manufacturer's recommendations. DNA synthesis was measured by incorporation of 5-bromo-2′-deoxyuridine (BrdU) into proliferating cell DNA using the BrdU Proliferation Assay Kit (EMD Biosciences). For BrdU assays, cells were plated in sera in 96-well plates (1 × 10⁵ cells per well) overnight and then serum starved and incubated with indicated reagents for 24 hours. Six hours prior to harvest, BrdU (1:2,000 final dilution) was added to cells. Cells were then fixed and incubated with anti-BrdU antibodies followed by horseradish peroxidase (HRP) substrate. Optical densities were read at 450 and 540 nm using the Synergy HT Microplate Reader (Bio-Tek). Absorbance differences (450 − 540 nm) were calculated as means ± SD and expressed as percentage of EGF treated.

Cdk6 is a putative miR-145 target, as its 3′ untranslated region (UTR) contains a conserved seed match for miR-145 at positions 3,884 to 3,892 (NM_001259.6). A 440-bp fragment flanking this position from base 3,638 to 4,078 was PCR amplified from cDNA generated by reverse-transcriptase PCR using total RNA extracted from HCT116 cells. For PCR, the following primers were used: (forward) 5′-CCCTCGAGGGTCCACAGCATTCAAG-3′ and (reverse) 5′-GCGGCCGCTTCAGAG-AGGCTGAGAT-3′. The PCR product was subcloned into psiCHECK2 dual luciferase reporter vector (Promega). A mutant construct that deleted the 8-bp sequence complementary to the miR-145 seed sequence was generated by 2-step PCR with forward primer 5′-AATGCAGCTGTTCTGTTTTTCAGCATTCTTTAG-3′ and reverse primer 5′-CTAAAGAATGCTGAAAAACAGAACAGCTGCATT-3′. The constructs were confirmed by DNA sequencing. HCT116 cells were cotransfected with psiCHECK2 (EV) or psiCHECK2 containing bases coding for a portion of wild-type or mutant 3′UTR of cdk6, together with a scrambled oligonucleotide (control) or mimic of mature miR-145 (Ambion). Renilla luciferase activity was normalized to firefly luciferase activity.

DNA oligonucleotides coding for pre-miR-143 or pre-miR-145 stem loops were subcloned into pCDH-CMV-MCS-EF1-copGFP (pCDH EV) lentiviral vector for constitutive expression. Lentivirus was prepared in 293T cells by cotransfecting pCDH-miR-143, pCDH-miR-145, or pCDH-EV with vectors expressing gag, rev, and VSV-G. HCT116 cells were transduced with lentivirus expressing pre-miRNA or EV. Transduced HCT116 (5 × 10⁶ cells) were implanted subcutaneously into flanks of nu/nu mice (5 mice per group). Tumor growth was monitored and tumors excised 3 week after implantation. Tissue was fixed in 10% buffered formalin or snap frozen in liquid nitrogen.

Proteins were extracted in SDS-containing Laemmli buffer, quantified by RC-DC protein assay, and subjected to Western blotting as described (30). Briefly, proteins were separated by SDS-PAGE on 4% to 20% resolving PAGE gradient gels and electroblotted to polyvinylidene difluoride (PVDF) membranes. Prestained molecular markers were included on each gel. Blots were incubated overnight at 4°C with specific primary antibodies followed by 1-hour incubation with appropriate peroxidase-coupled secondary antibodies that were detected by enhanced chemiluminescence using X-OMAT film. Xerograms were digitized using an Epson scanner and band intensity quantified using UN-SCAN-IT gel software (V 5.3, Silk Scientific). Protein levels in tumors were normalized to β-actin levels and expressed as fold of control colonic mucosa (means ± SD), matched for Egfr genotype. Protein lysates from tumors and control colonic mucosa with equal protein abundance as assessed by RC-DC assays showed comparable β-actin levels by Western blotting. Tumors of comparable stage were used for Western blotting comparisons.

Sections (5 μm) were mounted on Vectabond-coated Superfrost Plus slides. Sections were heated to 60°C for 1 hour, deparaffinized by 3 washes × 5 minutes in xylene, hydrated in a graded series of ethanol washes, and rinsed in distilled water. Epitope retrieval for Ki67 was achieved by microwave heating for 15 minutes in Tris-EDTA buffer, pH 9, followed by 3 washes × 2 minutes in TBS with 0.1% Tween-20 (TBST). Endogenous peroxidase activity was quenched with methanol/H₂O₂ solution (0.5%). Sections were washed 3 times in TBST × 2 minutes and blocked in protein block for 20 minutes. Sections were incubated with 1:300 dilution of anti-Ki67 antibodies for 1 hour at RT. After 3 TBST washes, slides were incubated at RT with 1:200 dilution of biotinylated secondary antibodies for 30 minutes. Antigen–antibody complexes were detected using an HRP-labeled DAKO EnVision+ System and 3,3′-diaminobenzidine as substrate. For negative controls, sections were incubated with isotype-matched nonimmune antibodies. After washing in distilled water, slides were stained with Gill's III hematoxylin, rinsed with water, dehydrated in ethanol, and cleared with xylene. It should be noted that tumors of comparable histology were used for immunostaining comparisons between Egfr^wt and Egfr^wa2 mice.

We detected Ki67 by nuclear staining quantified by an automated cellular imaging system (ACIS; Clarient San Juan). Color-specific thresholds were used to determine brown (positive) and blue (negative) nuclei within the outlined regions of interest to calculate the fraction of positively stained nuclei. Proliferation was expressed as the percentage of nuclei positive for Ki67. At least 5 fields per tumor and 3 tumors per group (∼50,000 cells per condition) were scanned for quantitation.

Continuous data were expressed as means ± SD and compared using Student's t test. Real-time PCR samples were run in triplicate and C_t values averaged. Untransformed C_t values were compared using a nonparametric Wilcoxon test (26). Levels of relative abundance, expressed as 2^(−ΔΔC_t), were calculated by exponentiating C_t differences between individual groups. Values of P < 0.05 were considered statistically significant. Unless otherwise indicated, data are representative of at least 3 independent experiments.

To identify potential EGFR-regulated miRNAs, we screened AOM/DSS-induced tumors from Egfr^wt and Egfr^wa2 mice by miRNA arrays. These tumors were available from a prior study (4). We initially focused on mice fed standard chow to assess EGFR effects without Western diet influences (4). Shown in Figure 1A is a heat map of miRNAs that were altered in tumors compared with control mucosa. Relative miRNA expression levels in tumors in the Egfr^wt and Egfr^wa2 mice were normalized to their Egfr genotype–matched respective controls. We observed that miR-143 and miR-145 were downregulated in tumors from Egfr^wt mice, whereas these miRNAs were upregulated in tumors from Egfr^wa2 mice. We quantified differences in expression levels of these miRNAs by real-time PCR. As shown in Figure 1B and as compared with control colonic mucosa matched for Egfr genotype (control expression levels normalized to 1), mature miR-143 and miR-145 were downregulated more than 60% in tumors from Egfr^wt mice, whereas these miRNAs were increased by more than 4-fold in tumors from Egfr^wa2 mice.

We confirmed our array findings by Northern blot analysis that showed reduced mature miR-145 (∼20 bp) in the Egfr^wt tumor, whereas mature miR-145 was increased in the Egfr^wa2 tumor (Fig. 1C). In contrast to the reduction in mature miR-145, in the Egfr^wt tumor, the precursor form (∼100 bp) appeared to be increased (Fig. 1C). This could reflect deregulated posttranscriptional processing of miRNAs that occurs in cancer (31). Interestingly, mature levels of miR-145 appeared comparable in mucosa from control Egfr^wt and Egfr^wa2 mice (Fig. 1C), suggesting that basal levels of these miRNAs are not controlled by EGFR in the absence of neoplastic transformation. These observations, however, will require further study.

To determine the cell type in the colon expressing miR-143 and miR-145, we assessed their expression by in situ hybridization. As shown in Figure 2, miR-143 and miR-145 were predominantly expressed in colonic epithelial cells especially on the colonic surface and maturation zone of the crypts in both Egfr^wt and Egfr^wa2 mice (see insets in Fig. 2B, C, H, and I). These miRNAs were downregulated in tumors from Egfr^wt mice (Fig. 2E and F) but preserved in tumors from Egfr^wa2 mice (Fig. 2K and L). Consistent with prior reports of miR-143 and miR-145 expression in muscle (17, 32), these miRNAs were also expressed in colonic muscularis mucosa and muscularis propria, albeit at lower levels (e.g., miR-143 in the muscle layers in Fig. 2B, white arrows).

Although expression levels of miR-143 and miR-145 in Egfr^wt and Egfr^wa2 tumors were significantly different, these miRNA differences did not translate into differences in tumor multiplicity or progression in animals on standard chow (4). Rather, the effects of EGFR on tumorigenesis were uncovered in the presence of a Western diet (4). We next examined these miRNAs in tumors from animals on Western diet. As in the case of tumors from mice on standard chow, miR-143 (Fig. 3A) and miR-145 (Fig. 3B) were downregulated in tumors from Egfr^wt mice and upregulated in tumors from Egfr^wa2 mice. Western diet modestly reduced the upregulation of these miRNAs in Egfr^wa2 tumors compared with standard diet (Fig. 3). Taken together, Egfr genotype and to a lesser extent, diet modulated expression of these miRNAs in tumors.

We next assessed changes in miR-143 and miR-145 in the AOM model of sporadic colon cancer to dissect the role of inflammation in the downregulation of these miRNAs. As in the case of the AOM/DSS model, a Western diet promoted tumor development in Egfr^wt but not in Egfr^wa2 mice (Fig. 4A). Similar to our findings in the AOM/DSS model, these miRNAs were downregulated in AOM-induced tumors from Egfr^wt mice and upregulated in AOM tumors from Egfr^wa2 mice on standard chow (Fig. 4B). These results indicate that downregulation of these miRNAs was not dependent on inflammation induced by DSS but mediated by increased EGFR signals in malignant transformation.

In addition to AOM and AOM/DSS mouse models, we examined the AOM rat model and Min mouse model of colon cancer to assess the generality of our findings regarding changes in these miRNAs. Using tumors obtained from a prior study (5), we found that these miRNAs were also decreased in colonic tumors from AOM-treated rats. Furthermore, gefitinib, an EGFR kinase inhibitor partially preserved these miRNAs in rat tumors at a dose that concomitantly inhibited tumorigenesis (5). These miRNAs also appeared to be downregulated in the Apc mutant Min mouse model of inherited colon cancer (Fig. 4D). In this regard, EGFR signals have been shown to be upregulated in Min adenomas (33).

In agreement with others (16, 34–36), we showed by Northern blot analysis that miR-143 and miR-145 were downregulated in sporadic human colon cancers compared with adjacent normal-appearing colonic mucosa (Fig. 5A and Supplementary Fig. S1). We also assessed miR-145 expression levels by in situ hybridization in normal human colon and sporadic colon cancers. Hematoxylin and eosin (H&E)-stained sections are shown in Figure 5B (i) and (iii). As in the mouse, miR-145 was abundant in colonic epithelial cells in the normal human colonic mucosa [Fig. 5B (iii)] but downregulated in colonic tumors [Fig. 5B (iv)]. Because AOM/DSS is a murine model of inflammation-associated colon cancer, we examined these miRNAs in human colon cancers associated with UC. As shown in Figure 5C, miR-143 and miR-145 were significantly downregulated in UC-associated cancers. Thus, in agreement with our findings in the mouse AOM and AOM/DSS models, miR-143 and miR-145 were downregulated in both sporadic and colitis-associated human colon cancers. Interestingly, mucosa adjacent to UC tumors also appeared to have reductions in these miRNAs, suggesting a tumor-related field effect, or an effect of UC on these miRNAs as tumors arose in a field of quiescent UC. We have since verified in a larger study that these miRNAs are downregulated in UC compared with normal colonic mucosa (37). In this regard, UC is a premalignant condition associated with increased colon cancer risk.

As Egfr genotype controlled miR-143 and miR-145 expression in colonic tumors, we asked whether EGFR signals could regulate these miRNAs in vitro in cell culture. As shown in Figure 6, EGF treatment of HCT116 cells did not significantly suppress levels of miR-143 or miR-145. These miRNAs were already significantly downregulated in HCT116 cells compared with normal colonic tissue (38). Because EGFR is constitutively activated by autocrine signals in HCT116 cells, we next blocked these upregulated signals with C225 anti-EGFR neutralizing antibodies. EGFR blockade significantly increased miR-143 and miR-145 compared with vehicle-treated HCT116 cells. In contrast to HCT116 cells, basal EGFR activation is low in human CCD-18Co colonic fibroblasts. We, therefore, activated this receptor with EGF and showed that EGFR signals significantly suppressed these miRNAs in colonic fibroblasts (Fig. 6). We also examined YAMCs, an immortalized murine colonic epithelial cell line with low basal EGFR signals (29). EGF treatment suppressed these miRNAs in this nontumorigenic cell (Fig. 6). Taken together, our studies indicate that EGFR signals downregulate these miRNAs in colonic cells.

To assess the growth consequences that accompany loss of miR-143 and miR-145 in colonic tumorigenesis, we transfected mimics of mature miR-143 or miR-145 into HCT116 colon cancer cells. These transfected miRNAs did not alter basal proliferation but inhibited EGF-induced cell proliferation (Fig. 7A). Presumably, targets of these miRNAs do not regulate basal proliferation. Alternatively, compensating mechanisms might be activated in the presence of transfected miRNAs. These transfected miRNAs also inhibited EGF-induced DNA synthesis (Fig. 7B), suggesting that EGFR-induced proliferation is mediated in part by downregulating these miRNAs in colonic tumorigenesis.

We searched miRNA target databases and prior reports for potential targets of miR-143 and miR-145 that might mediate the antiproliferative effects of these miRNAs. We focused on G₁ cell-cycle regulators implicated as in silico targets, as miR-143 and miR-145 strongly inhibited DNA synthesis (Fig. 7B). We identified K-Ras (miR-143) and MYC, CCND2, cdk6, and E2F3 (miR-145) as putative targets, as they possessed 3′UTR sequences that complemented seed sequences for these miRNAs. K-Ras and MYC were previously identified as targets of miR-143 and miR-145 (39, 40), respectively but not studied in an in vivo model of colonic tumorigenesis. As shown in Figure 7C (left), transfected miR-145 inhibited protein expression of MYC, cdk6, CCND2, and E2F3 cell-cycle regulators in HCT116 cells. To assess the potential direct regulation of cdk6 by miR-145, we cloned a portion of cdk6 3′UTR that contained the target sequence complementing the miR-145 seed sequence, downstream of a luciferase reporter (Fig. 7D). We also mutated the 3′UTR to abrogate the predicted miR-145-cdk6 3′UTR interaction. As shown in Figure 7E, miR-145 significantly inhibited the expression of Renilla luciferase regulated by wild-type cdk6-3′UTR but not mutant cdk6-3′UTR. We examined the effect of EGF on transcripts of these targets and found that EGF significantly induced cdk6 and E2F3 mRNA in HCT116 cells, whereas gefitinib blocked these inductions (Supplementary Data and Fig. S2). As noted earlier, however, EGF did not further suppress miR-143 and miR-145 in these cells, which were already significantly downregulated compared with normal colonocytes. Based on these results, we speculate that miR-143 and miR-145 downregulations contribute to tumorigenesis but are not sufficient for induction of these G₁ regulators in transformed colon cancer cells.

We also investigated the effects of miR-143 transfection on K-Ras, MEK2, ERK5, and PTGS2, as these proteins are demonstrated or predicted targets of this miRNA (39, 41–43). For this purpose, we selected HCA-7 cells, which express PTGS2 protein. (HCT116 cells express only PTGS2 transcripts). In agreement with these predictions, transfected miR-143 downregulated K-Ras, MEK2, ERK5, and PTGS2 in HCA-7 cells compared with a control oligonucleotide (with irrelevant sequence; Fig. 7C, right).

Our cell culture experiments suggested that miR-143 and miR-145 drive growth-inhibiting signals mediated at least in part by suppressing G₁ cell-cycle regulators. Our animal studies suggested that miR-143 and miR-145 may be tumor suppressors under Western diet conditions. To directly assess the tumor suppressor phenotype of these miRNAs in vivo, we transduced HCT116 cells with lentivirus alone (EV) or lentivirus-expressing pre-miR-143 or pre-miR-145. We confirmed that mature miR-143 and miR-145 were significantly upregulated in pre-miRNA–transduced cells, compared with EV-transduced HCT116 cells (Fig. 8A). Upregulation of these miRNAs decreased tumor xenograft weight supporting their postulated tumor suppressor phenotype (Fig. 8B). Consistent with cell-cycle effects of these miRNAs in cell culture, cells transfected with pre-miRNAs formed tumors with significantly reduced Ki67 immunostaining (Fig. 8C). Representative tumors stained for Ki67 are shown in Figure 8D–F.

To identify possible proteins mediating the growth-suppressive effects of these miRNAs in vivo, we examined expression levels of putative targets in tumor xenografts. As shown in Figure 8G, protein levels of CCND2, cdk6, E2F3, and MYC were downregulated in tumor xenografts from miR-145 expressing cells compared with EV-transduced HCT116 cells. These G₁ regulators were not significantly altered in tumor xenografts from miR-143–transduced cells (∼80%–90% of EV tumors, P > 0.3; Supplementary Data and Fig. S3), supporting the specificity of miR-145 to downregulate these proteins. In addition, we observed downregulation of PTGS2, K-Ras, and ERK5 in tumors with upregulated miR-143 (Fig. 8H). The latter are putative or established targets of miR-143 (39, 41, 42).

Experiments in vitro in colon cancer cells and in vivo in tumor xenografts using cells transfected with miR-145 suggested that this miRNA negatively controls G₁ regulators cdk6, CCND2, and E2F3. We next examined these G₁ regulators in colonic tumors from Egfr^wt and Egfr^wa2 mice to assess their regulation by EGFR, and their potential control by miR-145 in colonic tumorigenesis. As shown in representative tumors in Supplementary Figure S4A and quantified in Supplementary Figure S4B, CCND2 and cdk6 were significantly increased in tumors from Egfr^wt mice compared with Egfr^wa2 mice, consistent with loss of miR-145 in Egfr^wt tumors. Increases in E2F3 in Egfr^wt tumors compared with Egfr^wa2 tumors did not reach statistical significance (P = 0.3). Because tumor promotion occurred with Western diet only in Egfr^wt mice, we examined the expression of K-Ras and MYC, targets of these miRNAs in tumors from mice on Western diet. As shown in Figure 9A and B, K-Ras and MYC were significantly increased in tumors from Egfr^wt but not Egfr^wa2 mice. In prior studies, we also showed that the MYC increase was significantly higher in tumors from Egfr^wt mice on Western diet compared with standard diet. Tumors from Egfr^wt mice also had significantly higher proliferation as assessed by Ki67 staining and shown in Figure 9C (compare Fig. 9E with G). In contrast, colonic epithelial cell proliferation was comparable in control Egfr^wt and Egfr^wa2 mice (Fig. 9D and F).

EGFR is increased in many tumors including colon cancer (44). Furthermore, studies in experimental models of colon cancer have shown that this receptor plays a key role in colonic tumorigenesis (3–5, 45). Beginning with a miRNA array discovery approach, in the current report, we showed for the first time that miR-143 and miR-145 were negatively regulated by EGFR signals in murine colonic tumors. Specifically, miR-143 and miR-145 were decreased in AOM/DSS-induced tumors in Egfr^wt mice, whereas they were significantly increased in colonic tumors from Egfr^wa2 mice on either standard or Western diet. We confirmed these miRNA array results by real-time PCR and Northern blot analysis (miR-145). We also showed that miR-143 and miR-145 were more highly expressed in colonic epithelial cells than in stromal cells in normal mouse colon. We speculate that EGFR mediators required for tumor promotion by Western diet include targets of miR-143 or miR-145. This provides an example of epigenetic gene regulation by environmental cues (Western diet). In support of this hypothesis in the current study, we showed that Ras and MYC were upregulated by Western diet in colonic tumors from Egfr^wt but not Egfr^wa2 mice and that these proto-oncogenes are targets of miR-143 and miR-145, respectively in tumor xenografts. Prior studies have shown gene–dose thresholds for these proto-oncogenes in controlling tumorigenesis (46, 47). A gene–dose effect of cdk6 controlling proliferation has also been reported (48). Thus, upregulation of miR-143 and miR-145 in Egfr^wa2 mice would be predicted to restrain Western diet–induced increases in these proto-oncogenes and inhibit tumor promotion. This hypothesis is also supported by our findings that miR-143 and miR-145 decreased MYC and K-Ras and potently suppressed colon cancer cell growth in tumor xenografts.

As in the case of AOM/DSS-induced tumors, we observed that miR-143 and miR-145 were downregulated in tumors induced by AOM alone in Egfr^wt mice, whereas they were upregulated in AOM-induced tumors in Egfr^wa2 mice. Thus, expression levels of these miRNAs were negatively controlled by EGFR in murine models of both sporadic and inflammatory colon cancer. EGFR was also required for tumor promotion by Western diet in the AOM model. Targets of these miRNAs likely also block diet-related AOM tumor promotion in Egfr^wa2 mice. Consistent with genetic experiments showing that EGFR suppresses these miRNAs in colonic tumors, we observed that EGFR blockade with gefitinib partially inhibited miR143 and miR-145 downregulation in AOM-induced rat colonic tumors. The EGFR inhibitor effects, moreover, are in agreement with studies showing that gefitinib reduced miR-143 downregulation in lung cancer cells (49). These miRNAs also appeared to be downregulated in Min adenomas, another model with upregulated EGFR (45). Interestingly, recent in vitro studies in breast cancer cells identified several other miRNAs that target oncogenic transcription factors and that were suppressed by EGFR (13). Our studies have directly established that EGFR negatively regulates miR-143 and miR-145 in vivo in models of colonic tumorigenesis.

Downregulation of miR-143 and miR-145 were previously reported in colon cancer (16, 50–53). Subsequently, other tumor types were noted to have reduced levels of these miRNAs (20–23). In this study, we confirmed this finding and showed by in situ hybridization that miR-145 miRNA was predominantly expressed in epithelial cells in normal human colon and downregulated in sporadic human colon cancers. We also showed for the first time that miR-143 and miR-145 were downregulated in UC-associated colon cancers. These results extend our recent study showing that these miRNAs were downregulated in chronic UC (37). Further investigation will be required to determine whether mechanisms mediating suppression of these miRNAs in sporadic versus UC-associated colon cancer are different.

To assess whether EGFR suppression of these miRNAs could occur in a cell autonomous manner, we examined cells in culture. Consistent with findings in tumors, EGF stimulation of CCD-18Co colonic fibroblasts and murine colonocytes decreased these miRNAs. Conversely, EGFR blockade in HCT116 colon cancer cells increased miR-143 and miR-145. These findings are in agreement with a recent study showing that miR-145 was inhibited by cellular proliferation (54). The failure of EGF to further reduce these miRNAs in HCT116 cells likely reflects their downregulated state under basal conditions. It should be noted that these cells possess autocrine-activated EGFR signals. In the current study, we also showed that these miRNAs suppressed EGFR-induced colon cancer cell growth. Thus, while EGF-induced growth in HCT116 cells did not require further suppression of miR-143 and miR-145, increases in these miRNAs suppressed EGF-induced growth in vitro and in tumor xenograft growth in vivo. A growth-promoting circuit regulated by EGFR and involving miR-145 suppression was recently described in lung cancer cells (22). Taken together, we postulate that EGFR signals driving colonic tumorigenesis are mediated in part by EGFR suppression of these miRNAs. The recent demonstration that oncogenic Ras suppresses these miRNAs in pancreatic cancer cells suggests that Ras might mediate their downregulation by EGFR in colon cancers (15). The role of Ras in EGFR suppression of these miRNAs, however, will require further study.

We showed that transfected miR-143 and miR-145 inhibited the expression of K-Ras, MYC, cdk6, CCND2, and E2F3 in colon cancer cells growing in vitro or in tumor xenografts in vivo. Furthermore, transfection of miR-143 or miR-145 inhibited proliferation and DNA synthesis in cell culture and reduced Ki67 staining in tumor xenografts. These results suggest an important role for these miRNAs in cell-cycle regulation. cdk6, a major cyclin D–dependent kinase, is increased in human colon cancers (55). We confirmed that cdk6 is a direct target of miR-145 by showing that this miRNA significantly downregulated a luciferase reporter linked to wild-type but not mutant cdk6 3′UTR containing a target sequence for miR-145. CCND2 is another putative target of miR-145 and a member of the cyclin D family that can activate cdk6. CCND2 was linked to early progression in human colonic carcinogenesis (56, 57). CCND2–cdk6 complex can phosphorylate the retinoblastoma protein, Rb causing the release of E2F3 (58). E2F3 in turn drives cell-cycle progression by upregulating genes controlling G₁ to S-phase transition (59). Pathways involving E2F3 are active in many tumors including human colon cancer (60–62). Our in vitro and in vivo studies suggest that K-Ras, MYC, cdk6, CCND2, and E2F3 are targets of miR-143 or miR-145 that link these miRNAs to EGFR and cell-cycle control. Experiments using luciferase reporters linked to 3′UTR for CCND2 and E2F3 are in progress to further characterize their miR-145 target status in colon cancer cells.

To further assess several of these G₁ regulators as in vivo targets of miR-145 in colonic tumorigenesis, we compared their expression levels in colonic tumors arising in Egfr^wt and Egfr^wa2 mice. CCND2 and cdk6 were upregulated in tumors from Egfr^wt compared with Egfr^wa2 mice. These results are consistent with the differential effects of mutant and wild-type Egfr on miR-145 expression in colonic tumors and with in silico predictions of these G₁ regulators as targets of miR-145. Clinical differences in tumor multiplicity and tumor stage in Egfr^wt and Egfr^wa2 mice, however, required a Western diet that enhanced additional EGFR mediators, including K-Ras and MYC that are also targets of miR-143 and miR-145, respectively.

While miR-143 and miR-145 appear to drive tumor suppressor phenotypes based on our tumor xenograft studies, development of tumors in Egfr^wa2 mice suggests that alternative mechanisms allow tumor escape in these mutant mice. Such tumors likely exploit EGFR-independent oncogenic pathways. Increases in ErbB2, ErbB3, and c-Met were implicated as escape mechanisms in other tumors (63–65).

Taken together with earlier reports, our results indicate that miR-143 and miR-145 are tumor suppressors in colon cancer and play important restraining roles on cell-cycle progression (40, 66). Loss of these miRNAs may be especially critical for tumor development under stress conditions of a Western diet. Our results have uncovered a potentially important role of miRNAs in EGFR regulation of the cell cycle. The findings that EGFR signals coordinately control multiple G₁ regulators via miR-143 (K-Ras) and miR-145 (MYC, cdk6, CCND2, and E2F3) are consistent with studies showing that EGFR regulates other orchestrated events via coordinate miRNA changes (13). We would predict that by targeting multiple G₁ regulators, these cotranscribed tumor suppressor miRNAs would play an important restraining role on cell-cycle progression. Future studies restoring levels of individual targets could assess whether loss of a specific target plays a dominant role in cell-cycle arrest induced by miR-143 or miR-145 as suggested for other targets (67).

In summary, we have shown that EGFR signals negatively regulate expression of miR-143 and miR-145 in colonic tumorigenesis. Because these miRNAs drive tumor suppressor phenotypes in tumor xenograft models, their downregulation by EGFR uncovers an important miRNA-dependent mechanism mediating oncogenic effects of this receptor. The tumor suppressor role of these miRNAs appears to be conditional in the AOM and AOM/DSS models and revealed by Western diet that activates EGFR signals and promotes tumorigenesis. A schema summarizing this pathway is shown in Supplementary Figure S5. This mechanism involves increased G₁ cell-cycle regulators cdk6, CCND2, and E2F3, as well as enhanced K-Ras and MYC that promote proliferation in colonic tumorigenesis. Approaches that prevent downregulation of these miRNAs could provide new strategies for colon cancer prevention. The recent demonstration that these miRNAs can be upregulated by small molecules, moreover, offers an exciting potential strategy for cancer prevention in high-risk individuals (68).

No potential conflicts of interest were disclosed.

The authors thank Dr. John Kwon (University of Chicago) for critical reading and helpful suggestions.

These studies were funded in part by the following grants: P30DK42086 (Digestive Diseases Research Core Center), and CA036745 (M. Bissonnette), as well as the Samuel Freedman Research Laboratories for Gastrointestinal Cancer Research. The University of Chicago Department of Pathology Research Core provided additional funding.