miR-34a and miR-34b/c Suppress Intestinal Tumorigenesis

Colorectal cancer is a leading cause of cancer death. In the United States alone, 134,490 new cases and 49,190 deaths from colorectal cancer were expected in 2016 (1). Colorectal cancer is a multistep process driven by mutational activation of several oncogenes and genetic/epigenetic inactivation of tumor suppressors (2). The majority of colorectal cancers originate from benign adenomas, which are caused by inactivating mutations in the adenomatous polyposis coli (APC) gene. APC mutation leads to the constitutive activation of the Wnt/β-catenin pathway, which results in the increased expression of β-catenin, a critical regulator of intestinal epithelial cell homeostasis (3). During colorectal cancer progression, additional mutations in key oncogenes and tumor suppressor genes, such as p53, are acquired. Apc^Min/+ mice inherit a mutant Apc allele that results in a truncation of the APC protein at amino acid 850 (4) and spontaneously lose the wild-type Apc allele, resulting in multiple benign adenomas mainly throughout the small intestine (5). Thus, Apc^Min/+ mice allow to study the influence of putative tumor suppressor genes on the initiation of intestinal tumorigenesis in vivo (6).

The tumor suppressor gene p53 is inactivated in the majority of colorectal cancers and encodes a transcription factor, which is posttranscriptionally induced by DNA damage and a number of additional cellular stresses (7). Besides regulating the expression of mRNAs, p53 also controls the expression of noncoding RNAs, such as lncRNAs and microRNAs (miRNA; refs. 8, 9). Many tumor suppressive functions of p53 are thought to be mediated by p53-induced miRNAs (8, 10). Among the p53-induced microRNA-encoding genes, miR-34a and miR-34b/c display the most consistent and pronounced induction, which may explain why they were identified as one of the first p53-regulated miRNAs (11).

Via downregulation of key-regulators, miR-34a/b/c suppress numerous cancer-associated processes, such as cell proliferation and survival (12), epithelial–mesenchymal transition (EMT; ref. 13), and stemness (14). Moreover, miR-34a/b/c presumably function as cell-fate determinants in colonic cancer stem cells (CSC) by suppressing Notch activity (15). Furthermore, miR-34a/b/c regulate the homeostasis of normal stem cells (16) and suppress the formation of induced pluripotent stem cells (IPSCs) (17).

Consistent with a tumor suppressive role in tumorigenesis, downregulation of miR-34a/b/c expression has been observed in a variety of human cancers (8). miR-34a/b/c expression is also epigenetically silenced by CpG methylation in colorectal cancer cell lines (18) and in primary colorectal cancers (19). These findings suggest an important role of miR-34a/b/c as downstream effectors of p53 and potential tumor suppressors in colorectal cancer. In line with its tumor suppressive potential miR-34a mimetics are currently tested in the clinics for treatment of advanced cancer (20).

Here, we report the generation of Apc^Min/+ mice carrying targeted deletions of the miR-34a and miR-34b/c genes and the characterization of the resulting phenotypes at the organismal and molecular level. Our findings provide genetic proof of a tumor suppressive function of the miR-34a and miR-34b/c genes during intestinal tumorigenesis.

The generation of miR-34a^−/− mice with a C57BL6/SV129 background has been described previously (21). miR-34b/c^fl/fl mice were kindly provided by Dr. Alexander Nikitin (Cornell University, Ithaca, NY; ref. 16). In brief, gene-specific deletions were generated by using homologous recombination with a vector containing miR-34a or miR-34b/c sequences flanked by loxP sites and an intronic Neomycin resistance (Neo) cassette flanked frt sites individually (21). The Neo cassette was removed by crossing with flp-mice and germline miR-34a or miR-34b/c knock-out mice were generated by crossing with CMV-Cre mice. miR-34a^−/−; b/c^−/− compound mice were generated by crossing miR-34a^−/− and miR-34b/c^−/−. The resulting genotypes were obtained in the expected Mendelian ratios and the offspring displayed no overt phenotype. miR-34a^−/−, miR-34b/c^−/−, miR-34a^−/−; miR-34b/c^−/− and miR-34a^+/+; miR-34bc^+/+ mice were crossed with Apc^Min/+ mice to obtain mice with the following genotypes: miR-34a^−/−; Apc^Min/+, miR-34b/c^−/−; Apc^Min/+, miR-34a/b/c^−/−; Apc^Min/+, and miR-34a/b/c^+/+Apc^Min/+. Mice were housed in individually ventilated cages (IVC). Animal studies were approved by the Government of Upper Bavaria, Germany (AZ 55.2-1-54-2532-4-2014).

Mice were sacrificed at 18 weeks of age. Whole intestines were isolated, separated into four equal parts, flushed with PBS, opened longitudinally, photographed, and fixed as Swiss rolls in 4% buffered formaldehyde. Tumor numbers and size were evaluated using ImageJ software.

Three-micron paraffin sections were used for hematoxylin and eosin staining and periodic acid–Schiff (PAS) staining according to standard protocols. Tumors from Apc^Min/+ mice were classified as adenomas with either low- or high-grade dysplasia, based on nuclear-cytoplasmic ratio, nucleus location, prominence of nucleoli, gland architecture, amount of interglandular stroma, and the presence of mucus secretion. Immunohistochemistry was performed using antibodies and reagents listed in Supplementary Table S1. For each immunohistochemical detection ≥ 40 tumors per genotype were examined. Positive staining was evaluated by Image-Pro plus and ImageJ software.

For detection of intestinal stem cells an Olfm4-specific, DIG-labeled RNA probe was generated using a murine Olfm4 vector (kindly provided by Dr. Hans Clevers, Hubrecht Institute, Utrecht, The Netherlands) in combination with a DIG Northern Starter Kit (Roche Diagnostics). in situ hybridization (ISH) was performed as described (22). To visualize intestinal bacteria by fluorescence in situ hybridization, the universal eubacteria probe (EUB338) and negative control probe (NON338) were employed. The 5′-FITC-labeled EUB338 DNA (5′-GCTGCCTCCCGTAGGAGT-3′) and 5′-Cy3-labeled NON338 DNA (5′-CGACGGAGGGCATCCTCA-3′) probes were synthesized by Metabion (Planegg) and hybridized to 3-μm paraffin sections.

Total RNA from tumors was isolated using the RNeasy Plus Mini Kit (Qiagen) with an on-column DNase digestion (three RNA samples per genotype; each tumor RNA sample represented a pool of 3 tumors isolated from the same mouse). Random primed cDNA libraries were generated and sequenced using the HiSeq2500 (Illumina) platform by GATC (Konstanz). Each sample was covered by at least 35 million single reads of 50 bp length. Data were normalized in R with the RUVSeq (23) module and differential expression analysis was performed using Chipster (24). The overlap of DEseq2 and edgeR results were considered to represent the most significantly, differentially expressed genes. Gene Ontology (GO) and KEGG pathway analysis was performed with DAVID Bioinformatics Resources (25). Gene Set Enrichment Analysis (GSEA) was performed using the GSEA (26) software. miR-34a/b/c targets were predicted with TargetScan 6.2 (27). Expression data were deposited in the Gene Expression Omnibus website (accession no. GSE84138).

H1299 lung cancer cells used for 3′-UTR reporter assays were from own stocks and their negativity for p53 was validated. SW620 colorectal cancer cells used for ectopic pri-miR-34a expression were from own stocks and authenticated by STR analysis in 2014 (Eurofins Medigenomix Forensik GmbH, Ebersberg, Germany). H1299 and SW620 cancer cell lines were kept in Dulbecco's modified Eagle's medium (DMEM) and 10% fetal calf serum (Invitrogen) at 5% CO₂. SW620 cells were transfected with episomal pRTR-pri-miR-34a expression plasmids as described before (13). After 24 hours, cell pools were selected by addition of puromycin (2 μg/mL) for 10 days. GFP expression was evaluated by fluorescence microscopy 48 hours after addition of 100 ng/mL doxycycline (DOX) to the cell pools. Intestinal adenoma cells from Apc^Min/+ mice were isolated and counted using a hemocytometer. Single cells (15,000) were embedded in 50 μL Matrigel per well in 24-well plates. The tumor organoid culture medium was formulated as described before (28).

The 3′-UTR of Wasf1 was PCR amplified from genomic DNA isolated from mouse tissue, inserted into pGL3-control-MCS vector downstream of a firefly luciferase ORF, and verified by sequencing. H1299 cells were seeded in a 12-well plate with 3 × 10⁴ cells/well and transfected with 100 ng of pGL3-Wasf1 plasmid, 20 ng of Renilla reporter plasmid for normalization, and 25 nmol/L miR-34a or miR-34c pre-miRNAs or a negative control oligonucleotide (Ambion). Forty-eight hours later, luciferase assays were performed with the Dual-Luciferase Reporter 1000 Assay System (Promega). Fluorescence intensities were measured with an Orion II Microplate Luminometer (Titertek-Berthold). Primers used for cloning and sequencing are listed in Supplementary Table S2.

Total RNA was isolated from tumor samples or cultured cell lines using the RNeasy Plus Mini Kit (Qiagen). cDNA was generated by Verso cDNA kit (Thermo Scientific) and quantitative real-time PCR (qPCR) was performed by using Fast SYBR Green Master Mix (Applied Biosystems) and a LightCycler 480 II (Roche Diagnostics). Relative gene expression was determined using the 2^−ΔΔC_t method (29). The individual mRNA levels were normalized to β-actin. Primers used for qPCR are listed in Supplementary Table S3.

For protein lysates tumor samples or cultured cells were lysed in RIPA buffer containing cOmplete Mini protease inhibitor cocktail tablets (Roche Diagnostics). Lysates were sonicated and centrifuged. Whole-lysate proteins (30–60 μg) were loaded per lane. Gel electrophoresis and transfer to PVDF membranes (Millipore) were carried out using standard protocols (Bio-Rad Laboratories). Primary antibodies (Supplementary Table S1) were used in combination with HRP-coupled secondary antibodies. ECL (Millipore) signals were recorded with a 440CF imaging system (Kodak).

The Cancer Genome Atlas (TCGA; ref. 30) gene expression data and follow-up information of colon adenocarcinomas (COAD) were downloaded from NCI's Genomic Data Commons website (GDC; https://gdc.cancer.gov/). Normalized RSEM counts were used to determine the expression of relevant mRNAs. Clinical outcome data were divided into high, intermediate, and low expression groups according to the expression value of individual genes.

A two-tailed Student t test was used to compare continuous variables. Categorical variables were compared using the χ² method. Kaplan-Meier calculations were used to display the overall survival time and the results were compared with a log-rank test. The Sidak method was used to adjust P values when multiple comparisons were performed. P values less than 0.05 were considered significant and indicated by asterisks (*, P < 0.05; **, P < 0.01; or ***, P < 0.0001). Univariate, age/gender/tumor grade-adjusted hazard ratios, and 95% confidence intervals (CI) were estimated with a Cox's proportional hazard model. Prism 6 (GraphPad software) or SPSS (IBM) programs were used for calculations.

We generated mice carrying deletions of the miR-34a or miR-34b/c loci and combinations of these. Deletion of miR-34a, miR-34b/c, or miR-34a/b/c in the germline did not significantly influence the lifespan of mice (Fig. 1A). However, we observed a gender-independent increase in the width and depth of crypts of the small intestine in miR-34a/b/c-deficient mice, which presumably caused the minor increase in the total length of the small intestine (Fig. 1B). Notably, the numbers of Paneth cells per crypt and Goblet cells per villus were significantly increased in miR-34a/b/c-deficient mice (Fig. 1C and D). In addition, the frequency of stem cells at the crypt base was significantly increased in miR-34a/b/c-deficient mice as determined by detection of the stem cell marker Olfm4 using in situ hybridization (Fig. 1E). These phenotypes were independent of the gender (Fig. 1B–E).

In order to determine, whether inactivation of miR-34 genes affects intestinal tumor formation, we generated Apc^Min/+ mice with deletions of miR-34a, miR-34b/c, or miR-34a/b/c. Also in Apc^Min/+ mice deficiency for miR-34a/b/c resulted in an increase of the number of Paneth, Goblet, and Olmf4-positive cells per crypt in untransformed tissue of the small intestine (Supplementary Fig. S1A–D). Similarly, the width and depth of the crypts were increased significantly in miR-34a/b/c-deficient Apc^Min/+ mice, but the minor increase in the length of the small intestine was not statistically significant (Supplementary Fig. S1E). Again, these architectural changes were observed in both, male and female mice. Therefore, deletion of one APC allele did not influence the morphological changes in the intestine caused by deletion of miR-34a/b/c.

Notably, miR-34a/b/c^−/−; Apc^Min/+ mice showed signs of morbidity sooner and exhibited a shorter life-span than wild-type Apc^Min/+ mice (Fig. 2A). In contrast, Apc^Min/+ mice deficient for either the miR-34a or the miR-34b/c allele did not show a statistically significant change in life-span (Fig. 2A). When mice were sacrificed at the age of 18 weeks and the entire small intestinal tract was examined, miR-34a/b/c^−/−; Apc^Min/+ and, to a lesser extent, miR-34a^−/−; Apc^Min/+ knockout mice showed a significantly increased number of tumors when compared with Apc^Min/+ mice (Fig. 2B and C). Furthermore, the total tumor area in miR-34a/b/c^−/−; Apc^Min/+ mice was ∼4 times larger than in wild-type Apc^Min/+ mice (Fig. 2D). The frequency of large tumors (>6 mm²) was significantly higher when miR-34a/b/c, but not when only miR-34a or miR-34b/c genes had been deleted in Apc^Min/+ mice (Fig. 2E). We also noted a significantly increased number of adenomas with high-grade dysplasia in miR-34a/b/c-deficient Apc^Min/+ mice, whereas the inactivation of either miR-34a or miR-34b/c alone did not have this effect (Fig. 2F). Taken together, our findings show that the inactivation of miR-34a/b/c promotes intestinal tumor formation by increasing tumor initiation and enhancing the growth of tumors in Apc^Min/+ mice. As the combined inactivation of miR-34a and miR-34b/c genes was necessary for these effects, these miRNAs may have overlapping functions and compensate each other. In addition, these results are in accordance with the combined epigenetic inactivation of miR-34a and miR-34b/c by DNA methylation, which has been detected in more than 75% of analyzed colorectal cancer samples (31). In the subsequent analyses, we therefore focused on studying mice with simultaneous deletion of miR-34a and miR-34b/c.

Notably, tumors in miR-34a/b/c knockout Apc^Min/+ mice showed a significant increase in proliferation and a significantly reduced rate of apoptosis (Fig. 3A). The frequency of stromal cells within the tumors was not affected by miR-34a/b/c deletion as determined by detection of vimentin. Because it was recently shown that bacterial infiltration promotes tumorigenesis in Apc^Min/+ mice (32), we asked whether miR-34a/b/c deletion affects the presence of bacteria in intestinal tumors. By detection of bacterial 16S rRNA, we found that adenomas from miR-34a/b/c^−/−; Apc^Min/+ mice display extensive bacterial infiltration regardless of tumor size (Fig. 3A) or tumor grade (Supplementary Fig. S2A). However, the adjacent normal intestinal epithelium of all mice analyzed here was devoid of bacteria. Therefore, miR-34a/b/c-deficient mice do not display a general intestinal barrier defect. Unexpectedly, the frequency of immune cells, such as macrophages, T- and B-cell, was decreased in miR-34a/b/c-deficient adenomas (Fig. 3B; Supplementary Fig. S2B). Therefore, the increased bacterial infiltration of adenomas may be due to a defect in the tumor-associated immune defense in miR-34a/b/c-deficient mice.

To further illuminate the mechanisms by which miR-34a/b/c loss promotes intestinal tumorigenesis in Apc^Min/+ mice, we obtained mRNA expression profiles of adenomas from 18-week-old miR-34a/b/c-deficient Apc^Min/+ mice and wild-type Apc^Min/+ mice. For each mouse (n = 3 per genotype), RNAs derived from three different tumors were pooled, libraries were generated and subjected to RNA-Seq to obtain more than 35 million reads per library. Thereby, we detected the upregulation of 1773 mRNAs, when mRNAs with a log₂-fold change ≥ 0.6 and RPKM ≥ 0.25 in miR-34a/b/c^−/−; Apc^Min/+ mice versus wild-type Apc^Min/+ mice were included (Fig. 4A). As expected, the percentage of mRNAs upregulated in the miR-34a/b/c-deficient tumors was higher for mRNAs harboring miR-34a/b/c seed-matching sequences than for mRNAs lacking these (Fig. 4B). Three hundred and twenty-four mRNAs were identified as differentially regulated due to miR-34a/b/c loss when the raw-count methods DESeq2 (33) and edgeR (34) were applied (Fig. 4C). After unsupervised clustering of these differentially expressed genes, one of the miR-34a/b/c-deficient tumor samples showed a divergent expression profile, which was presumably due to biological variation (Fig. 4D). Next, we used Gene Ontology (GO) and KEGG pathway analyses to group the differentially expressed transcripts into functional classes. The GO analysis indicated that the upregulated mRNAs are involved in cell adhesion, extracellular matrix, and growth factor binding functions (Supplementary Fig. S3A). The KEGG analysis showed that several pathways were significantly enriched, such as focal adhesion, ECM-receptor interaction, and pathways related to cancer (Supplementary Fig. S3B). On the contrary, both GO and KEGG analyses indicated that the downregulated genes are mainly involved in immune response functions (Supplementary Fig. S4A and S4B). A subsequent Gene Set Enrichment Analysis (GSEA) revealed that miR-34a/b/c deletion results in gene expression signatures related to EMT, hypoxia, angiogenesis, inflammatory response, Kras signaling, and apical junction processes (Fig. 4E). Interestingly, the expression signatures of miR-34a/b/c-deficient adenomas also showed a significant overlap with signatures of intestinal stem cells (ISC) (35), Lgr5⁺ positive cells (36) and the GO Wnt signaling signature. In addition, several regulators of barrier function in IEC were also detected as downregulated by NGS analysis in miR-34a/b/c-deficient adenomas (e.g., Muc1, Tff3, and Retnlb; fold change: −4.74, −1.44, and −3.34, respectively). The downregulated mRNA expression of these barrier components was validated by qPCR (Supplementary Fig. S5A). Furthermore, we confirmed the decreased protein expression of MUC1 in miR-34a/b/c-deficient adenomas by immunohistochemistry (Supplementary Fig. S5B). The repression of these barrier components may also contribute to the observed infiltration of bacteria into miR-34a/b/c-deficient adenomas.

To obtain functional evidence for an enhanced stemness of adenoma cells with deletion of miR-34a/b/c, we performed a tumor organoid formation assay. Indeed, an increased tumor organoid formation rate was observed for miR-34a/b/c-deficient adenoma derived cells when compared with those derived from miR-34a/b/c-proficient adenomas (Fig. 5A).

Wnt/β-catenin signaling is critically involved in regulating the homeostasis and neoplastic transformation of the intestine. Because 34a/b/c-deficient adenomas showed a Wnt signaling signature (Fig. 4E) and the Wnt/β-catenin pathway is directly regulated by miR-34a/b/c (37) and by APC, we determined whether the expression of β-catenin is also affected in the intestinal epithelial cells of miR-34a/b/c knockout mice. Interestingly, we detected an increased nuclear accumulation of β-catenin/CTNNB1 protein in normal crypts after miR-34a/b/c deletion in Apc^Min/+ mice, but not in miR-34a/b/c-deficient wild-type mice (Fig. 5B), indicating that miR-34a/b/c deficiency in the presence of hemizygous Apc allows the nuclear accumulation of β-catenin in untransformed cells, including the stem cells located at the crypt base, which represent the preferred cells of origin for intestinal adenomas (38). Therefore, this effect of miR-34a/b/c loss may explain or at least contribute to the increased rate of tumor initiation in miR-34a/b/c-deficient Apc^Min/+ mice.

By bioinformatics analyses of the mRNA profiles obtained here, we identified a set of 11 upregulated mRNAs with miR-34 seed-matching sites in their 3′-UTR, which may, at least in part, mediate the tumor suppressive function of miR-34a/b/c (Supplementary Table S4; Fig. 6A). These mRNAs and a selected set of mRNAs with known pro-tumorigenic functions were analyzed by qPCR to exemplarily validate the NGS results (Fig. 6B). Several of these factors have already been characterized as direct miR-34 targets by others and us (Pdgfra (39), Pdgfrb (39), and Axl (40)). In addition, components of the Wnt pathway and EMT regulators were analyzed. Notably, all of the tested mRNAs were expressed at significantly higher levels in the tumors of miR-34a/b/c knockout Apc^Min/+ mice. Therefore, our screen identified numerous additional miR-34a/b/c targets, which may be upregulated in colorectal cancer and relevant for tumor progression.

Next, we performed an exemplary characterization of Wasf1 as a direct miR-34a/b/c target, because Wasf1 is known to regulate cancer-relevant processes (41). A luciferase-based reporter containing the full-length 3′-UTR of the murine Wasf1 mRNA was significantly repressed by cotransfection of pre-miR-34a or pre-miR-34c (Fig. 6C). In addition, we detected a significant reduction of Wasf1 mRNA expression after ectopic expression of pri-miR-34a in SW620 colorectal cancer cells (Fig. 6D). The WASF1 protein showed a delayed repression 72 hours after activation of the pri-miR-34a allele, which may be due to an unusually stable WASF1 protein (Fig. 6E). Consistent with being a miR-34a/b/c target, the WASF1 protein showed increased expression in three out of four tumors isolated from miR-34a/b/c knockout Apc^Min/+ mice, when compared with miR-34a/b/c-proficient tumors (Fig. 6F). Taken together, these results show that Wasf1 is a direct target of miR-34a/b/c.

In order to determine whether the miR-34 targets described above are also clinically relevant, we analyzed their expression within the expression profiles of 460 human colorectal adenocarcinomas deposited in TCGA database (30). Interestingly, INHBB, PDGFRB, STC1, and COL4A2 showed a significantly elevated expression in primary colorectal cancers (Fig. 7A). Furthermore, a significantly increased expression of INHBB, AXL, FGFR1, and PDGFRB was detected in primary colorectal cancers from patients with nodal status N2 when compared with samples from patients with lower nodal status (Fig. 7B). Except for AXL, a significantly increased expression of these mRNAs was associated with an increased tumor stage (Supplementary Fig. S6A). In addition, increased expression of INHBB and AXL in primary colorectal cancers was significantly associated with decreased patient survival (Fig. 7C; Supplementary Table S5). Elevated expression of COL4A2, WASF1, STC1, and PDGFRB was also associated with poor survival, although not significantly (Supplementary Fig. S6B). Because poor survival is mostly due to metastatic spread, these results suggest that the elevated expression of these mRNAs, which may be due to loss of miR-34a/b/c, promotes metastasis.

Finally, we compiled the regulations identified here and combined them with previously published results on miR-34a/b/c-regulated pathways in a summarizing model shown in Fig. 7D. Notably, several of the miR-34 targets displaying elevated expression after miR-34a/b/c deletion are involved in transmembrane signal transductions, which also activate the Wnt signaling pathway by inhibition of GSK-3β. Taken together, miR-34a/b/c presumably suppresses intestinal tumorigenesis caused by loss of Apc by downregulating the expression of a large number of pro-tumorigenic factors. In case miR-34a/b/c expression is lost or silenced during tumor progression, these pathways may be further activated and thereby contribute to colorectal cancer progression.

The increase in the pool of intestinal stem cells (ISC) in the miR-34a/b/c-deficient mice observed here may underlie the increased rate of tumor formation in Apc^Min/+ mice, because ISC were shown to serve as efficient tumor initiating cells during intestinal tumorigenesis (38). Paneth cells provide a niche for ISCs and communicate with these via multiple signaling pathways, among them the Delta/NOTCH and Wnt/APC/β-catenin pathways, which are under negative control by miR-34a/b/c (15, 37): that is, several key components of the Wnt signaling pathway are miR-34a/b/c targets (WNT1, WNT3, LRP6, LEF1, and β-catenin (37, 40). Moreover, GSK-3β, which phosphorylates β-catenin and thereby leads to its poly-ubiquitination and proteasomal degradation, is inhibited by the PKB/AKT and PI3K pathways, which are also under control of miR-34a/b/c. Therefore, the concerted deregulation of the NOTCH and WNT pathways caused by the loss of miR-34a/b/c is a likely cause for the increased number of Paneth and stem cells. In support of this scenario, we also noticed an increased tumor organoid formation rate in miR-34a/b/c^−/−; Apc^Min/+ adenomas as well as an increased accumulation of nuclear β-catenin in the untransformed, epithelial cells of intestinal crypts in miR-34a/b/c-deficient Apc^Min/+ mice. In addition, a recent study suggests that activation of the Wnt/β-catenin pathway correlates with T-cell exclusion (42). Therefore, miR-34 may regulate immune responses through Wnt signaling. In addition, the expression of barrier proteins was decreased in miR-34a/b/c^−/−; Apc^Min/+ mice (e.g., Muc1, Tff3, and Retnlb). In combination with the decreased presence of immune cells, these effects of miR-34a/b/c loss may contribute, at least in part, to the increased bacterial infiltration observed in miR-34a/b/c^−/−; Apc^Min/+ adenomas.

In addition, we found that the deletion of miR-34a/b/c in adenomas leads to the upregulation of previously described miR-34a/b/c target mRNAs, such as Pdgfra, Pdgfrb, and Axl, which are known to enhance colorectal cancer formation (43). Furthermore, a set of novel, putative miR-34a/b/c target mRNAs (Wasf1, Fgfr1, Igf1, Stc1, Cacna2d2, Col6a2, Col4a2, and Inhbb), which were reported to be involved in tumorigenesis (41, 44–49), was upregulated in miR-34a/b/c-deficient adenomas. The factors encoded by these genes form complex signaling and functional networks (see also Fig. 7D): PDGFRA, PDGFRB, FGFR1, AXL, and IGF1, which are either ligands or receptors of the tyrosine kinase family, activate the PI3K and MAPK signaling pathway, and thereby promote cell growth, survival, EMT, and metastasis. Notably, CACNA2D2, a voltage-dependent calcium channel, deregulates calcium homeostasis when ectopically expressed (47). The resulting calcium release may induce the activation of PKB/AKT and RAS to promote cell proliferation and angiogenesis (47). INHBB assembles into activins, which are critical modulators of growth and survival (49). COL4A2 is a major structural protein of the basement membrane and COL6A2 has an anchoring function. Both collagens stimulate integrin signaling and are involved in cell growth, angiogenesis, and tumor metastasis (48). WASF1, a new miR-34a/b/c target characterized here, mediates actin polymerization, lamellipodia formation, and plays a critical role in cancer cell migration and invasion (41). STC1 is an endocrine regulator of calcium and phosphate homeostasis, which promotes cell proliferation and inhibits apoptosis through CCND1 and CDK2/4 (50). Taken together, miR-34a/b/c presumably act as global regulators to fine tune multiple cellular functions, which are necessary for the homeostasis of intestinal epithelia. Because miR-34a has been shown to form bimodal switches with at least some of its targets (15), the loss of miR-34a/b/c during colorectal cancer progression may have significant effects on the regulation of these signaling and expression networks, which ultimately promote intestinal tumorigenesis. miR-34a/b/c have a multitude of targets in intestinal epithelial and colon cancer cells as shown here and in previous studies (10). Therefore, it seems plausible that the effects of miR-34a/b/c loss are mediated by the combined activity of multiple upregulated targets and not by one or a few targets.

A subset of the direct targets of miR-34a/b/c may significantly contribute to human colorectal tumorigenesis because their elevated expression was associated with poor clinical outcomes of colorectal cancer patients. However, further analysis will be necessary to corroborate the role of miR-34a/b/c silencing for their increased expression in colorectal tumors. Our results suggest that miR-34 mimetics may be used to target multiple key pathways simultaneously and could thereby potentially prevent the emergence of resistance caused by mutations of single pathways. Therefore, miR-34a/b/c replacement therapy may represent a potential option for the treatment of colorectal cancer.

No potential conflicts of interest were disclosed.

Conception and design: H. Hermeking

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Jiang, H. Hermeking

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Jiang, H. Hermeking

Writing, review, and/or revision of the manuscript: L. Jiang, H. Hermeking

Study supervision: H. Hermeking

Other (experimental design): L. Jiang

We are grateful to Alexander Nikitin for providing miR-34b/c^fl/fl mice, Marlon Schneider for Apc^Min/+ mice, Hans Clevers for the Olfm4 plasmid, Matjaz Rokavec for initial help with setting up the Apc^Min/+ mouse cohorts, and Markus Kaller for advice on bioinformatics analyses. We would also like to thank Peter Jung for discussions and Ursula Götz for technical assistance.

H. Hermeking received support from a Deutsche Krebshilfe Grant (No. 109531) and the Rudolf Bartling Stiftung (V/124). L. Jiang received a fellowship from the China Scholarship Council.

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.