The microRNA (miRNA) landscape changes during the progression of cancer. We defined a metastasis-associated miRNA landscape using a systematic approach. We profiled and validated miRNA and mRNA expression in a unique series of human colorectal metastasis tissues together with their matched primary tumors and corresponding normal tissues. We identified an exclusive miRNA signature that is differentially expressed in metastases. Three of these miRNAs were identified as key drivers of an EMT-regulating network acting though a number of novel targets. These targets include SIAH1, SETD2, ZEB2, and especially FOXN3, which we demonstrated for the first time as a direct transcriptional suppressor of N-cadherin. The modulation of N-cadherin expression had significant impact on migration, invasion, and metastasis in two different in vivo models. The significant deregulation of the miRNAs defining the network was confirmed in an independent patient set as well as in a database of diverse malignancies derived from more than 6,000 patients. Our data define a novel metastasis-orchestrating network based on systematic hypothesis generation from metastasis tissues. Cancer Res; 75(15); 3010–9. ©2015 AACR.

An estimated 90% of cancer-related deaths are caused by the direct or indirect effects of metastatic dissemination (1). A key question has been to define molecularly the regulatory networks that endow tumor cells with metastatic properties. Several in vivo models have attempted to address this issue (2, 3), but have been limited in their capacity to reflect faithfully human cancer metastasis. Although the direct analysis of human samples would be advantageous for hypothesis generation, few studies have systematically analyzed resected human metastasis tissue in comparison with the corresponding primary tumors and associated normal tissues, largely because such matched tissue sample series are rare.

microRNAs (miRNA) have emerged as an increasingly important class of molecules that can regulate the metastatic process (4) and to prime the metastatic niche (5). To date, however, a systematic analysis of miRNAs whose expression is associated with metastasis, or that act to regulate metastasis, has not been carried out.

To address this issue, we analyzed global miRNA and mRNA expression in a unique series of primary tumors, their matched metastases, and the corresponding normal tissue from the primary and metastatic sites taken from patients with colorectal cancer. To define a metastasis-associated miRNA signature, we identified miRNAs that were differentially expressed in metastases compared with primary tumor tissues and normal tissues, then validated differential expression using qRT-PCR. Putative mRNA targets for these metastasis-associated miRNAs were then identified in silico, and compared with the mRNA transcription profiles obtained from the same original patient samples. This analysis suggested that gene-expression patterns associated with EMT correlate with the metastasis-associated miRNA signature. Using these findings to guide functional studies in vitro and in vivo, we focused on miRNAs that were predicted to regulate the cadherin switch during EMT, and thereby identified a novel miRNA-driven regulatory network that fosters invasion and metastasis. These data functionally validate the metastasis-associated miRNA signature we identified, and suggest that further analysis of the datasets we have generated will generate additional insights into critical molecular regulatory networks that control metastasis.

Patient material

Tumor, metastasis, and corresponding normal and background tissue samples of patients with colorectal cancer were stored and obtained from the tumor bank of the Mannheim Medical faculty, University of Heidelberg, Germany. The Ethical Committee of the Medical Faculty, University Hospital Mannheim, approved the project and informed consent was obtained from patients or their spouses when the former were deceased. Tissue sections and specimen were prepared by pathologists before snap freezing and subsequent storage in liquid nitrogen. Biobanking and handling of the tissues followed the BRISQ guidelines (6).

Microarray and bioinformatics analyses

The human miRNA Sentrix Universal Array Matrix (SAM) V1 arrays were used to determine expression. The full dataset has been deposited at the Gene Expression Omnibus (GEO) with the accession number GSE54088. All statistical analyses were performed within the R statistical software environment (R version 2.15.3 using the R packages limma, version 3.14.4. miRNAs with a fold change of ≥1.5 and P value ≤ 0.1 were considered significant.

The mRNA input into the Ingenuity Pathway core analysis was generated by selecting mRNAs, which were targeted by six or more of the top 10 most significant up- and downregulated miRNAs, and in which the target prediction was made by two or more search engines that comprised TargetScan, MiRanda, RNA22, miRWalk, PICTAR4, PITA, DIANA-mT, and RNAhybrid. This was done using a pearl script that extended miRWalk files created with the “extendAndFilterMiRWalk.pl” and created an output file that showed if an mRNA is a predicted target of the input miRNAs and also how many prediction tools generated the prediction.

Cell culture

The cell lines used for the study; DLD1, 293T, COLO206f, COLO320, RKO, HCT116, SW480, SW620, and A375, were obtained either from the ATCC or the DSMZ (Germany) and cultured routinely in T25 flasks at 37°C in the presence of 5% CO2 and 90% humidity. Culture media were cell line specific and supplemented with 10% FCS and 1% penicillin/streptomycin. The cell lines are authenticated periodically every 2 years using SNP-profiling (Multiplexion GmbH, Germany). The last series of authentication tests were carried out in 2014. All cells taken into culture were Mycoplasma-free and were tested monthly with the Mycoplasma test PCR protocol (Venor GeM-Kit) when required.

Site-directed mutagenesis and RT-PCR

The QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene) was used to create the sequence mutations in the binding motifs of miR-135b, miR-210, and miR-218 in the respective 3′-untranslated regions (UTR) of target genes SETD2, SIAH1, N-cadherin, FOXN3, and ZEB2. RT-PCR was carried out using the Quantitect Primer assays (Qiagen) and SYBR Green detection system on a LightCycler 480 (Roche) and the ΔΔCt algorithm. Details of the all primer sequences used are contained in Supplementary Primer List.

Immunoblotting

Total protein (40 μg) was separated on 10% SDS-PAGE and transferred on to polyvinylidene difluoride membranes. The detailed protocol is as contained in ref. 7 and the respective antibodies used are elaborated in the Supplementary Information.

Reporter gene assays

Promoter assay.

The N-cadherin promoter (a kind gift from Prof Pierre J. Marie, INSERM UMR 606 and University Paris Diderot, Sorbonne Paris Cité Hôpital Lariboisière, France) in pGL3 vector was cotransfected together with a FOXN3 expression plasmid (Source Bioscience) or empty vector. The experimental set up is as described in ref. 8.

3′-UTR assays.

In the 3′-UTR luciferase reporter assay, the 3′-UTR of the mRNA of interest (SETD2, SIAH1, FOXN3, N-cadherin, and ZEB2) was cloned downstream of a reporter luciferase gene (firefly/renilla) into a multiple cloning site of either the pLightswitch 3′-UTR plasmid (Switch Gear Genomics) or the pMIR-REPORT plasmid (Applied Biosystems). Reporter assays were conducted as outlined previously (8). All experimental setups were performed in quadruplicate and repeated in at least three independent trials.

Migration and invasion assays.

Migration and invasion assays were conducted as outlined previously (7, 8).

Chorion-allantoic membrane assay.

Fertilized special pathogen-free eggs were purchased from Charles River Laboratories (Sulzfeld, Germany) and incubated for 10 days in a Marsh incubator (Lyon Electric) at 37°C with automated intermittent rotation. A total of 2 × 106 miR-135b/miR-210/miR-218–overexpressing or control cells in 50 μL of serum-free medium were carefully inoculated in vicinity of a prominent blood vessel and the assay was performed as described previously (9).

Mouse experiments

All animal experiments were approved by the local authorities and performed according to the German legal requirements. Groups of 6 to 8 athymic nu/nu mice were injected i.v. via their tail veins with either miR-218 stably transduced or control SW480 and SW620 cell lines. The mice were monitored regularly and were killed when they became moribund.

Significant differences in the mean survival times for both cell lines was calculated by the Student t test (significance defined as P < 0.05). Kaplan–Meier survival analysis together with Mantel-Cox log rank statistics was applied to calculate significant differences in the course of survival of the mice. Mantel-Cox test results were considered significant with P ≤ 0.05.

A metastasis-associated miRNA signature from matched primary tumors, metastases, and normal patient tissues

To identify miRNAs whose expression is specific for metastatic tumors, we procured a unique series of matched samples from 8 colorectal cancer patients consisting of resected primary colorectal cancers, corresponding normal colorectal mucosa distant from the tumor site, resected metastasis tissue, and corresponding normal surrounding tissue from the metastatic site, all from the same patients. Six patients had liver metastases, and two had lung metastases. All samples had at least 80% tumor cell content. The samples were then used for miRNA expression profiling in parallel to mRNA expression profiling (Fig. 1A). Subsequent bioinformatics analysis identified miRNAs that were up- or downregulated in the primary tumor versus normal colorectal mucosa, and in metastases compared with the relevant normal metastatic site tissue (liver or lung), and in metastases compared with the primary tumor. The most significantly regulated miRNAs were validated by RT-PCR. This allowed us to identify miRNAs that are specifically regulated in metastases versus corresponding primary tumor tissue, and miRNAs that are regulated in both primary tumors and their corresponding metastasis tissue compared with the underlying normal tissues (Fig. 1B). In total, 42 miRNAs were significantly up- or downregulated in the primary tumor compared with normal colonic tissue, 85 miRNAs were significantly deregulated in the metastasis tissue versus the primary tumor, whereas 12 miRNAs were up- or downregulated in the primary tumor and stayed deregulated in the corresponding metastasis tissue (Supplementary Table S1).

Figure 1.

The metastasis-specific miRNA signature. A, Venn diagram showing the overlap of differentially regulated miRNAs in primary tumor and metastatic tissues as compared with normal epithelium and with each other. B, a modified Venn diagram taken from A, with the significant miRNAs listed in the respective compartments. The miRNAs in the nonoverlapping sections of the Venn diagram are further detailed with their fold changes (in color) and significance P values in size, using the associated key. C, Ingenuity Pathway Analysis of the significant pathways resulting from genes targeted by significant miRNAs.

Figure 1.

The metastasis-specific miRNA signature. A, Venn diagram showing the overlap of differentially regulated miRNAs in primary tumor and metastatic tissues as compared with normal epithelium and with each other. B, a modified Venn diagram taken from A, with the significant miRNAs listed in the respective compartments. The miRNAs in the nonoverlapping sections of the Venn diagram are further detailed with their fold changes (in color) and significance P values in size, using the associated key. C, Ingenuity Pathway Analysis of the significant pathways resulting from genes targeted by significant miRNAs.

Close modal

To identify miRNAs that might play a functional role in metastatic progression, we focused on miRNAs that were differentially expressed between metastatic tissue and primary tumors, taking expression in the corresponding normal tissues into account (Fig. 1A and B). In line with our goal of identifying a metastasis-specific signature, the principal component of the analysis was to elucidate miRNAs that were differentially expressed between the metastasis and the primary tumor. Furthermore, because not only the primary colon tumor, but also (indirectly) its metastatic colon cancer lesion principally originated from normal colon tissue, we rationalized that significant differences in the profile between the cancer metastasis tissue and the individual normal colon tissue of the specific patient needed to be considered in the mathematical equation as well.

This analysis generated a signature of 62 miRNAs, which we postulated could potentially have a functional impact on the metastatic process. The signature contains miRNAs whose expression has already been associated with tumor progression, including miR-34 family members (10, 11), miR-10a and b (12, 13), miR-143 (14), miR-122 (15), and miR-200 (16) in addition to novel potential metastasis-relevant miRNAs.

EMT as one predicted target of the metastasis-associated miRNA signature

To identify major pathways that are likely to be regulated by miRNAs in the metastasis-associated miRNA signature, we used in silico methods to identify putative mRNA targets of the miRNAs in the signature. We also reasoned that mRNAs encoding proteins that act as key nodes in regulatory pathways are likely to be targeted by multiple miRNAs. We therefore selected mRNAs that were predicted to be miRNA targets by two or more search engines and that were predicted to be targeted by 6 or more of the top 10 most significantly up- and downregulated miRNAs in the metastasis-associated miRNA signature. Interestingly, when we evaluated the cascades modulated by the mRNAs selected by this process, several pathways known to drive metastasis, including PTEN, Wnt, IGF-1, and HGF signaling were predicted, confirming the validity of our approach (Fig. 1C).

Next, we compared this list of predicted mRNA targets with the mRNA profiles generated from the same patient samples used to identify the metastasis-associated miRNA signature (Supplementary Fig. S1). Of the mRNAs identified in this way, many of them have been implicated in the process of epithelial-to-mesenchymal transition (EMT), including FOXQ1, EPCAM, CEACAM5, CEACAM 6, ELF3, ASCL2, and CDH1. EMT is thought to play a central EMT is a major function of the metastasis-relevant miRNA signature we identified.

Regulation of the EMT-associated cadherin switch by miR-135b, miR-210, and miR-218

Having identified EMT as a putative regulatory target for the metastasis-associated miRNA signature, next we focused on identifying miRNAs that potentially regulate expression of key molecular players in EMT. A pivotal event in EMT is the cadherin switch, in which E-cadherin expression is downregulated and N-cadherin is upregulated (17–21). We therefore performed in silico analysis to compare the expression of E-cadherin and N-cadherin mRNA in the original patient samples with the metastasis-associated miRNA signature. We made several permutations of the individual miRNAs to identify those that had good support vector regression (SVR)/context scores (depending on the in silico tool) for binding to EMT related mRNAs, and based on these scores, miR-135b, miR-218, and miR-210 were seen to be the most significant combination.

Enhanced expression of miR-210 and miR-135b and low expression of miR-218 was found to correlate with downregulation of E-cadherin and upregulation of N-cadherin. This finding allowed us to formulate the hypothesis that these three microRNAs are likely to play a role in miRNA-driven regulation of EMT and metastasis through coordinate control of the cadherin switch. Moreover, N-cadherin was also a significant hit in the list of predicted targets identified in our initial multitool in silico analysis of downregulated miRNAs.

To validate the functional relevance of the metastasis-associated miRNA signature for the EMT cadherin switch, we investigated whether miR-135b, miR-218, and miR-210 regulate E- and N-cadherin expression as predicted. To this end, we transfected SW480, SW620, DLD1 and Colo320 colorectal and the A375 melanoma cancer cell lines with either miR-135b, miR-210, or miR-218, and used RT-PCR and/or Western blotting to assess effects on E- and N-cadherin expression (Fig. 2A and C, Supplementary Fig. S2A) and also assessed the effect on the cell phenotype (Supplementary Fig. S3). As expected, a significant decrease in E-cadherin expression in parallel to a significant increase in N-cadherin was observed upon transfection of the cells with miR-135b or miR-210 (Fig. 2A).

Figure 2.

The cadherin switch as a common target of three miRNAs. A, miR-135b and miR-210 mediate a cadherin switch by suppressing E-cadherin and enhancing N-cadherin expression. SW480 and SW620 cell lines were transfected with the respective miRNA mimics and protein expression was evaluated 72 hours after transfection using Western blotting. B, N-cadherin and ZEB2 are targets for miR-218 in luciferase assays in 293T cells by showing a significant suppression of N-cadherin and ZEB2 3′-UTR reporter activity upon miR-218 expression. C, miR-218 mimics suppress N-cadherin and ZEB2 expression at mRNA and protein levels, whereas miR-218 antagonists have the opposite effect. Messenger RNA and protein expression were evaluated at 48 hours with qRT-PCR and Western blots, respectively. D, potential synergism of miR-218 with the miR-200 family at the ZEB2 3′-UTR shown by luciferase assays in 293T cells.

Figure 2.

The cadherin switch as a common target of three miRNAs. A, miR-135b and miR-210 mediate a cadherin switch by suppressing E-cadherin and enhancing N-cadherin expression. SW480 and SW620 cell lines were transfected with the respective miRNA mimics and protein expression was evaluated 72 hours after transfection using Western blotting. B, N-cadherin and ZEB2 are targets for miR-218 in luciferase assays in 293T cells by showing a significant suppression of N-cadherin and ZEB2 3′-UTR reporter activity upon miR-218 expression. C, miR-218 mimics suppress N-cadherin and ZEB2 expression at mRNA and protein levels, whereas miR-218 antagonists have the opposite effect. Messenger RNA and protein expression were evaluated at 48 hours with qRT-PCR and Western blots, respectively. D, potential synergism of miR-218 with the miR-200 family at the ZEB2 3′-UTR shown by luciferase assays in 293T cells.

Close modal

N-cadherin as a direct target of miR-218

A putative seed sequence for miR-218 was found within the N-cadherin 3′-UTR (Fig. 2B). To determine whether N-cadherin is a direct physical target of miR-218, a luciferase reporter construct containing the wild-type 3′-UTR of N-cadherin mRNA was transfected into 293T cells together with a miR-218 mimic or inhibitor, respectively. An equivalent luciferase reporter construct containing a mutation within the putative miR-218 seed sequence served as a control. A miR-218 mimic specifically inhibited wild-type N-cadherin 3′-UTR activity, but not when the mutated 3′-UTR construct was used (Fig. 2B). N-cadherin mRNA and protein were also significantly downregulated upon transfection with miR-218 mimic (Fig. 2C). These experiments confirm that miR-218 specifically targets the N-cadherin 3′-UTR via a functional seed sequence at position 949–955, thus leading to a downregulation of N-cadherin mRNA and protein.

miR-218 directly targets ZEB2, acting synergistically with the miR-200 family

Suppression of E-cadherin expression is often regulated by ZEB transcription factors (21, 22). We therefore screened the 3′-UTR of ZEB2 for putative miRNA seed sequences, and found a putative seed sequence for miR-218 (Fig. 2B). This seed sequence was not overlapping with published miR-200 seed sequences, an miRNA known to regulate ZEB2 (16, 23). To investigate whether miR-218 suppresses ZEB2 expression, we performed luciferase reporter assays using the ZEB2 3′-UTR wild-type or mutated for the miR-218 seed sequence (Fig. 2B). These studies revealed that miR-218 alone significantly downregulates ZEB2 3′-UTR activity, affects mRNA and protein expression (Fig. 2C), and acts synergistically with the miR-200 family (Fig. 2D). These data suggest that miR-218 acts to suppress EMT by directly targeting ZEB2 mRNA on the one hand, leading to downregulation of ZEB2 and therefore subsequent upregulation of E-cadherin expression, whereas on the other hand it directly targets and suppresses N-cadherin expression. Thus, miR-218 affects a cadherin switch that favors the epithelial phenotype, consistent with its downregulation in the metastasis-associated miRNA signature.

SETD2, SIAH1, and the N-cadherin–suppressing FOXN3 as novel targets of miR-135b/miR-210

To search for putative targets through which miR-135b and miR-210 might regulate N-cadherin expression, we screened for potential miR-135b and miR-210 targets among the mRNAs that were specifically differentially expressed in the metastases from the colorectal cancer patients. From this screen, we found that the 3′-UTRs of SETD2 and SIAH1 mRNAs (24–26), as well as the 3′-UTR of FOXN3, a transcriptional repressor (27) contained putative seed sequences for both miR-135b and miR-210. Luciferase reporter assays using wild-type or mutant seed sequence 3′-UTRs revealed that SETD2, SIAH1, and FOXN3 are all specific targets of miR-135b and miR-210, a finding confirmed by the ability of miR-135b and miR-210 individually to reduce the mRNA and/or protein levels of SETD2, SIAH1, and FOXN3 (Fig. 3A–D).

Figure 3.

SETD2, SIAH1, and FOXN3 transcription factor, a novel transcriptional N-cadherin suppressor, are novel targets of miRs-135b and -210. A, 3′-UTR Luciferase reporter assays showing significant downregulation of reporter activity upon miR-210 transfection on SETD2 and on SIAH1 following miR-135b transfection. Reporter activity was abrogated or significantly decreased when the binding motifs were mutated. B, Western blots showing repression of SETD2 and SIAH by miR-210 and miR-135b, respectively, in SW480 and HCT116 cell lines. C, 3′-UTR luciferase reporter assays showing decreased reporter activity of FOXN3 with miR-210 and miR-135b. This demonstrates that these two miRNAs directly target FOXN3, which translates into loss of FOXN3 protein expression evaluated with Western blots 72 hours after transient transfection (D). E, reporter assay in 293T cells showing that the N-cadherin promoter is significantly suppressed by FOXN3 48 hours after transfection, with concomitant suppression of mRNA and protein expression at 72 hours.

Figure 3.

SETD2, SIAH1, and FOXN3 transcription factor, a novel transcriptional N-cadherin suppressor, are novel targets of miRs-135b and -210. A, 3′-UTR Luciferase reporter assays showing significant downregulation of reporter activity upon miR-210 transfection on SETD2 and on SIAH1 following miR-135b transfection. Reporter activity was abrogated or significantly decreased when the binding motifs were mutated. B, Western blots showing repression of SETD2 and SIAH by miR-210 and miR-135b, respectively, in SW480 and HCT116 cell lines. C, 3′-UTR luciferase reporter assays showing decreased reporter activity of FOXN3 with miR-210 and miR-135b. This demonstrates that these two miRNAs directly target FOXN3, which translates into loss of FOXN3 protein expression evaluated with Western blots 72 hours after transient transfection (D). E, reporter assay in 293T cells showing that the N-cadherin promoter is significantly suppressed by FOXN3 48 hours after transfection, with concomitant suppression of mRNA and protein expression at 72 hours.

Close modal

We hypothesized that FOXN3 might suppress transcription from the N-cadherin promoter, and thus that miR-135b and miR-210 might indirectly upregulate N-cadherin expression by suppressing expression of FOXN3. To test this hypothesis, we performed luciferase reporter experiments using N-cadherin promoter constructs. Indeed, we were able to show that FOXN3 transcriptionally represses N-cadherin promoter activity, and also suppresses N-cadherin mRNA and protein expression (Fig. 3E).

Together, our data suggest a novel regulatory network between miR-218, miR-135b, and miR-210, the miR-200 family and several novel molecular targets for these miRNAs, including ZEB2, N-cadherin, two novel tumor suppressors SIAH1 and SETD2, and the FOXN3 transcription factor, which together coordinately regulate E- and N-cadherin expression, and thus the EMT-associated cadherin switch.

miR-135b and miR-210 induce and miR-218 suppresses invasion and metastasis

To investigate whether these miRNAs functionally contribute to tumor progression, we ectopically expressed them in tumor cells after screening for endogenous expression in SW480 and SW620 colorectal cancer cell lines (Supplementary Fig. S4). We then conducted migration and invasion assays in vitro, and metastasis assays in vivo. In Boyden chamber migration assays, migration of RKO and HCT116 cells was significantly increased when transfected with miR-135b or miR-210 (Fig. 4A), with the converse being the case when SW480 and Colo206f cells were transfected with miR-218 mimics and inhibitors (Fig. 3A). In Matrigel invasion assays miR-135b and miR-210 significantly enhanced the invasive capacity of RKO and HCT116 cells (Fig. 4B), whereas invasion was significantly impaired upon miR-218 transfection in Colo206f and in SW480 cells (Fig. 4C). In vivo, RKO and SW480 cells stably transfected with either miR-135b or miR-210, had a significantly enhanced ability to metastasize into the lungs and livers of chicken embyros, whereas the opposite was observed for miR-218 stably transfected diverse cell lines (Fig. 4E and F, Supplementary Fig. S2).

Figure 4.

Migration, invasion, in vivo intravasation, and metastasis are significantly regulated by miR-218, miR-135b, and miR-210. A and B, enhanced in vitro migration and invasion in RKO and HCT116 cell lines following transfections with miR-135b, miR-210, and in combination. Migration is significantly enhanced with the individual miRNAs and a synergistic effect is seen when both miRNAs are combined. C, suppression of migration and invasion with miR-218 transfection in Colo206f and SW480 cell lines. The induced suppression is reversed when the miR-218 antagonist is used. D, decreased invasion with siRNAs targeting SIAH1 and SETD2 as further evidence that these two tumor suppressors significantly affect invasion. E, distant metastasis to the lungs is significantly enhanced with miR-210 and miR-135b transiently transfected cells in the chicken CAM model. F, decreased distant metastasis with miR-218 stably expressing SW620 and SW480 cell lines using the CAM assay. G and H, enhanced survival in athymic nu/nu mice injected with miR-218 stably transduced SW620 cells as compared with the control group. Significance of differences in the mean survival times for the SW620 cell line was significant as calculated by the Student t test (P ≤ 0.05). Kaplan–Meier survival analysis together with Mantel-Cox log rank statistics was applied to calculate significant differences in the course of survival of the mice.

Figure 4.

Migration, invasion, in vivo intravasation, and metastasis are significantly regulated by miR-218, miR-135b, and miR-210. A and B, enhanced in vitro migration and invasion in RKO and HCT116 cell lines following transfections with miR-135b, miR-210, and in combination. Migration is significantly enhanced with the individual miRNAs and a synergistic effect is seen when both miRNAs are combined. C, suppression of migration and invasion with miR-218 transfection in Colo206f and SW480 cell lines. The induced suppression is reversed when the miR-218 antagonist is used. D, decreased invasion with siRNAs targeting SIAH1 and SETD2 as further evidence that these two tumor suppressors significantly affect invasion. E, distant metastasis to the lungs is significantly enhanced with miR-210 and miR-135b transiently transfected cells in the chicken CAM model. F, decreased distant metastasis with miR-218 stably expressing SW620 and SW480 cell lines using the CAM assay. G and H, enhanced survival in athymic nu/nu mice injected with miR-218 stably transduced SW620 cells as compared with the control group. Significance of differences in the mean survival times for the SW620 cell line was significant as calculated by the Student t test (P ≤ 0.05). Kaplan–Meier survival analysis together with Mantel-Cox log rank statistics was applied to calculate significant differences in the course of survival of the mice.

Close modal

In further experiments, SW620 cells with stable ectopic expression of miR-218 were injected i.v. into athymic Nu/Nu nude mice. This experiment revealed that mice injected with SW620 cells stably overexpressing miR-218 lived significantly longer than animals injected with control cells (Fig. 4G and H). Collectively, these data show that migration, invasion, intravasation, and metastasis in vivo are significantly induced by miR-135b/210 and suppressed by miR-218. We investigated whether these miRNAs acted synergistically to modulate the observed effects but did not find any significant interaction.

Reciprocal expression of miR-135b/miR-210, miR-218, and their targets in patient tumors

To demonstrate the relevance of our findings for human disease, we used samples from 64 patients with colorectal cancer that were independent from the samples used for the original screen for metastasis-associated miRNAs. The majority of these samples were from patients with advanced tumor stages. Using RT-PCR, we found enhanced expression of miRs-210 and -135b and reduced expression of miR-218 in these samples (Fig. 5A). Furthermore, this correlated with downregulated expression of SETD2, SIAH1, E-cadherin, and FOXN3 (Fig. 5B), consistent with our in vitro findings. Interestingly, when we analyzed miRNA profiles from early-stage tumor samples in the GEO database (GSE35982 and GSE10259), our signature was not visible, suggesting that the metastatically capable clone becomes evident only after a certain advanced tumor stage has been attained.

Figure 5.

Independent in vivo validation of the novel miRNA network in an independent series of resected patient tissues and the Oncomine database. A, composite miRNA expression profile of 64 patients evaluating the expression of miR-135b, miR-210, and miR-218. miR-135b and miR-210 are significantly overexpressed in tumor tissues, whereas miR-218 is significantly downregulated. B, box plots showing the expression of significantly downregulated novel targets; SIAH1, SETD2, FOXN3, and E-cadherin in an independent series of primary tumor/normal tissue samples showing a significant relative downregulation of all four molecules in tumor tissues (P = 0.0012, 0.0007, 0.0012, and 0.047 for SIAH1, FOXN3, SETD2, and E-cadherin, respectively; one-tailed Wilcoxson-matched pairs signed rank test, n = 40). C, top, FOXN3 expression in the TCGA colon dataset (colon adenocarcinoma vs. normal) showing a highly significant downregulation of gene expression (P = 5.8 × 10–24; average fold change, −2.8); columns 1, 2, and 3 represent normal colon, rectum, and colon adenocarcinoma, respectively; bottom, FOXN3 expression in the Su Multicancer dataset showing a significant downregulation across several cancer types (P = 1.33 × 10–6; average fold change, −1.9); columns 1 to 9 represent bladder, breast, colorectal, kidney, liver, lung, ovarian, pancreatic, and prostate cancers in that sequence. D, diagrammatic representation of the novel network implicating miR-210, miR-218, and miR-135b in the mediation of EMT and metastasis resulting from influences on ZEB2, N-cadherin, and FOXN3. FOXN3 and N-cadherin are the central players in this network whose expressions are affected directly or indirectly by the one or more of the three miRNAs in the network.

Figure 5.

Independent in vivo validation of the novel miRNA network in an independent series of resected patient tissues and the Oncomine database. A, composite miRNA expression profile of 64 patients evaluating the expression of miR-135b, miR-210, and miR-218. miR-135b and miR-210 are significantly overexpressed in tumor tissues, whereas miR-218 is significantly downregulated. B, box plots showing the expression of significantly downregulated novel targets; SIAH1, SETD2, FOXN3, and E-cadherin in an independent series of primary tumor/normal tissue samples showing a significant relative downregulation of all four molecules in tumor tissues (P = 0.0012, 0.0007, 0.0012, and 0.047 for SIAH1, FOXN3, SETD2, and E-cadherin, respectively; one-tailed Wilcoxson-matched pairs signed rank test, n = 40). C, top, FOXN3 expression in the TCGA colon dataset (colon adenocarcinoma vs. normal) showing a highly significant downregulation of gene expression (P = 5.8 × 10–24; average fold change, −2.8); columns 1, 2, and 3 represent normal colon, rectum, and colon adenocarcinoma, respectively; bottom, FOXN3 expression in the Su Multicancer dataset showing a significant downregulation across several cancer types (P = 1.33 × 10–6; average fold change, −1.9); columns 1 to 9 represent bladder, breast, colorectal, kidney, liver, lung, ovarian, pancreatic, and prostate cancers in that sequence. D, diagrammatic representation of the novel network implicating miR-210, miR-218, and miR-135b in the mediation of EMT and metastasis resulting from influences on ZEB2, N-cadherin, and FOXN3. FOXN3 and N-cadherin are the central players in this network whose expressions are affected directly or indirectly by the one or more of the three miRNAs in the network.

Close modal

Independent validation of our network at more than 6,000 patients of diverse cancer entities in the Oncomine database

Using the Oncomine repository for multiple cancer entities, we found that all players of our network were confirmed independently to be deregulated in large databases of patients with different cancer entities. In this regard, FOXN3 was found to be strongly downregulated at highest levels of significance in more than 15 different types of malignancies, including colorectal, breast, prostate, lung cancer, and lymphomas, in a combined sample size of more than 6,000 patients (Fig. 5C; Supplementary Table S2). Similarly, a preliminary survival analysis we conducted with a small patient cohort of The Cancer Genome Atlas (TCGA) colorectal database showed that patients with a high expression of FOXN3 had a significantly better 5-year median survival (Supplementary Fig. S5). Equally interesting was the significant downregulation of SIAH1 in the Ki and colleagues and Hong and colleagues (28, 29) colon cancer datasets. Excitingly, there was also a strong and significant upregulation of miR-210 and a significant downregulation of miR-218 in the dataset that cut across several subtypes of brain tumors (Supplementary Fig. S6). The concept filter analysis for our three miRNAs using the Oncomine tool showed that for the predicted targets, FOXN3 and SIAH1 were significantly downregulated, and correlated inversely with miR-135b. Similarly, SETD2 downregulation was observed and correlated inversely with miR-210 (Supplementary Fig. S7). Together, these data allow us to construct a novel network of regulatory miRNA-driven interactions that regulates expression of key genes associated with EMT and metastasis (Fig. 5D).

The regulation of gene-expression underlies the metastatic properties of tumor cells, and miRNAs have been increasingly implicated in the regulation of metastatic behavior. In this study, we have attempted to systematically identify and characterize miRNAs that orchestrate the metastatic phenotype of colorectal cancer cells. By profiling matched primary tumors, metastases, and corresponding normal tissues coupled to bioinformatic analysis, we identified miR-210, miR-135b, and miR-218 as concerted novel regulators of metastatic dissemination, which all culminate in the regulation of the cadherin switch. Accordingly, we found that upregulation of miR-135b and miR-210 and loss of miR-218, which acts synergistically with miR-200 at the 3′-UTR of ZEB2 mRNA, downregulates E-cadherin and induces N-cadherin expression, central events in EMT, resulting in increased migration and invasion of tumor cells in vitro, and enhanced metastasis in vivo. This is achieved, at least in part, through a novel regulatory molecular network that involves a number of novel miRNA targets, including the tumor-suppressor genes SIAH1 and SETD2, and especially, the transcription factor FOXN3 that we identify here as a novel transcriptional repressor of N-cadherin gene expression, and thus as an interesting novel player in cancer metastasis. We conclude that miR-218 together with miR-200 family is a novel metastasis suppressor and that miR-135b and miR-210 act within a novel regulatory network to promote colorectal cancer metastasis, in part, through fostering EMT.

The relevance to tumor progression of the metastasis-associated miRNA signature, we report here is substantiated on a number of levels. First, a number of miRNAs whose expression has already been associated with tumor progression are contained in the signature, including miR-34 family members (7, 10, 11), miR-10a and b (12, 13), miR-143 (14), miR-122 (15), and the miR-200 family that is an essential regulator of EMT, invasion, chemoresistance, and angiogenesis (16, 30). Second, ingenuity analysis predicted that the miRNA signature regulates major pathways known to drive metastasis (Fig. 1). Additional bioinformatics analysis also identified EMT, an important determinant of metastatic properties, as a central target for the metastasis-associated miRNA signature. Third, functional analysis with selected miRNAs from the signature demonstrated a role for them in the regulation of invasion and metastasis.

Here, we report that miR-218, miR-135b, and miR-210 act together within a novel regulatory network to drive colorectal cancer metastasis. A putative role for miR-135b and miR-210 in cancer biology has previously been reported (31–33). Interestingly, miR-135b targets members of the Hippo pathway that has been associated with determining stemness-like properties (34), which is likely to be relevant to the role of miR-135b in cancer progression. Mechanistic studies on miR-210 have focused on a putative function in hypoxia-associated processes, and a few metastasis-relevant targets of miR-210 have been characterized (35). By demonstrating the targeting of the tumor suppressors SIAH1 and SETD2 and the little studied FOXN3 transcription factor, the findings we report here substantially advance our knowledge about the function of miR-135b and miR-210. Inhibition of SIAH1 and SETD2 is likely to increase tumorigenicity. FOXN3 has previously been studied mainly in the context of embryogenesis (36, 37), it will be highly interesting to determine whether it plays additional roles in tumor progression over and above its novel function in the cadherin switch that we report here. It was striking that our meta-analysis of FOXN3 in the Oncomine dataset revealed a highly significant and impressive downregulation of this molecule not only in a large set of colorectal cancers, but across 15 different types of malignancies, including breast, lung, prostate cancers, and also lymphomas. This clearly indicates that FOXN3 might be an important tumor-suppressor gene that plays a vital role in metastasis and cuts across several cancer entities.

Our finding that miR-218 directly targets N-cadherin and leads to E-cadherin upregulation by targeting ZEB2 transcription factor in synergy with the miR-200 family, together with its ability to suppress invasion and metastasis suggests a novel metastasis-suppressing role for miR-218. This was also corroborated by the Oncomine database in which both ZEB2 and N-cadherin were predicted to be targets of miR-218. We feel encouraged as to the validity of our novel network by the fact that when we screened an independent own series of colorectal cancer tissues, the respective significant up- or downregulation of miRNAs and targets, as well as their inverse correlation were confirmed (Fig. 5B). In addition, the concept filter analysis for our three miRNAs using the Oncomine tool showed that for the predicted targets, FOXN3 and SIAH1 were significantly downregulated and correlated inversely with miR-135b. Also, SETD2 downregulation was associated inversely with miR-210 in this large independent dataset. Taken together, this supports the hypothesis that the regulatory network identified in this article is indeed valid, even across diverse cancer types, and could serve as a guide for future strategies aimed at combating colorectal cancer metastasis.

The data we report here underscore the utility of unbiased screening and analysis of miRNA profiles for hypothesis generation and subsequent functional validation in the cancer context. Importantly, the metastasis-associated miRNA signature we have defined suggests a metastasis-promoting role for a number of miRNAs that have to date not been implicated in tumor progression, including miR-552, miR-218, miR-135, miR-210, and miR-654.

No potential conflicts of interest were disclosed.

Conception and design: G. Mudduluru, M. Abba, J. Batliner, M. Scharp, J.H. Leupold, H. Allgayer

Development of methodology: M. Abba, N. Patil, M. Scharp, T.R. Lunavat, H. Allgayer

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Mudduluru, M. Abba, J. Batliner, N. Patil, M. Scharp, T.R. Lunavat, J.H. Leupold, O. Oleksiuk, W. Thiele, J. Sleeman, H. Allgayer

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Mudduluru, J. Batliner, M. Scharp, T.R. Lunavat, O. Oleksiuk, D. Juraeva, W. Thiele, A. Benner, J. Sleeman, H. Allgayer

Writing, review, and/or revision of the manuscript: G. Mudduluru, J. Batliner, O. Oleksiuk, D. Juraeva, A. Benner, Y. Ben-Neriah, J. Sleeman, H. Allgayer

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.H. Leupold, O. Oleksiuk, M. Rothley

Study supervision: G. Mudduluru, M. Abba, H. Allgayer

The authors thank Jürgen Engel and Dr. Daniel Novak for technical assistance.

H. Allgayer was supported by Alfried Krupp von Bohlen und Halbach Foundation, Essen, Dr. Hella-Buehler-Foundation, Heidelberg, The Hector Foundation, Weinheim, Dr. Ingrid-zu-Solms Foundation, Frankfurt, Walter Schulz Foundation, Munich, the Deutsche Krebshilfe, Bonn (109558), the DKFZ-MOST German Israel Cooperation, Heidelberg (CA149), the HIPO/POP-Initiative for Personalized Oncology, Heidelberg (H032 and H027). H. Allgayer, G. Mudduluru, J.H. Leupold, and J. Sleeman were supported by the Wilhelm Sander Foundation, Munich, Germany (2012.036.1).

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.

1.
Sporn
MB
. 
The war on cancer
.
Lancet
1996
;
347
:
1377
81
.
2.
Cimpean
AM
,
Ribatti
D
,
Raica
M
. 
The chick embryo chorioallantoic membrane as a model to study tumor metastasis
.
Angiogenesis
2008
;
11
:
311
9
.
3.
Viney
JL
. 
Transgenic and gene knockout mice in cancer research
.
Cancer Metastasis Rev
1995
;
14
:
77
90
.
4.
Lee
YS
,
Dutta
A
. 
MicroRNAs in cancer
.
Annu Rev Pathol
2009
;
4
:
199
227
.
5.
Peinado
H
,
Aleckovic
M
,
Lavotshkin
S
,
Matei
I
,
Costa-Silva
B
,
Moreno-Bueno
G
, et al
Melanoma exosomes educate bone marrow progenitor cells toward a pro-metastatic phenotype through MET
.
Nat Med
2012
;
18
:
883
91
.
6.
Moore
HM
,
Kelly
AB
,
Jewell
SD
,
McShane
LM
,
Clark
DP
,
Greenspan
R
, et al
Biospecimen reporting for improved study quality (BRISQ)
.
Cancer Cytopathol
2011
;
119
:
92
101
.
7.
Mudduluru
G
,
Ceppi
P
,
Kumarswamy
R
,
Scagliotti
GV
,
Papotti
M
,
Allgayer
H
. 
Regulation of Axl receptor tyrosine kinase expression by miR-34a and miR-199a/b in solid cancer
.
Oncogene
2011
;
30
:
2888
99
.
8.
Leupold
JH
,
Yang
HS
,
Colburn
NH
,
Asangani
I
,
Post
S
,
Allgayer
H
. 
Tumor suppressor Pdcd4 inhibits invasion/intravasation and regulates urokinase receptor (u-PAR) gene expression via Sp-transcription factors
.
Oncogene
2007
;
26
:
4550
62
.
9.
van der Horst
EH
,
Leupold
JH
,
Schubbert
R
,
Ullrich
A
,
Allgayer
H
. 
TaqMan-based quantification of invasive cells in the chick embryo metastasis assay
.
Biotechniques
2004
;
37
:
940
2
.
10.
He
L
,
He
X
,
Lim
LP
,
de
SE
,
Xuan
Z
,
Liang
Y
, et al
A microRNA component of the p53 tumour suppressor network
.
Nature
2007
;
447
:
1130
4
.
11.
Hwang
CI
,
Matoso
A
,
Corney
DC
,
Flesken-Nikitin
A
,
Korner
S
,
Wang
W
, et al
Wild-type p53 controls cell motility and invasion by dual regulation of MET expression
.
Proc Natl Acad Sci U S A
2011
;
108
:
14240
5
.
12.
Liu
Z
,
Zhu
J
,
Cao
H
,
Ren
H
,
Fang
X
. 
miR-10b promotes cell invasion through RhoC-AKT signaling pathway by targeting HOXD10 in gastric cancer
.
Int J Oncol
2012
;
40
:
1553
60
.
13.
Meseguer
S
,
Mudduluru
G
,
Escamilla
JM
,
Allgayer
H
,
Barettino
D
. 
MicroRNAs-10a and -10b contribute to retinoic acid-induced differentiation of neuroblastoma cells and target the alternative splicing regulatory factor SFRS1 (SF2/ASF)
.
J Biol Chem
2011
;
286
:
4150
64
.
14.
Pagliuca
A
,
Valvo
C
,
Fabrizi
E
,
di
MS
,
Biffoni
M
,
Runci
D
, et al
Analysis of the combined action of miR-143 and miR-145 on oncogenic pathways in colorectal cancer cells reveals a coordinate program of gene repression
.
Oncogene
2013
;
32
:
4806
13
.
15.
Wang
B
,
Wang
H
,
Yang
Z
. 
MiR-122 inhibits cell proliferation and tumorigenesis of breast cancer by targeting IGF1R
.
PLoS ONE
2012
;
7
:
e47053
.
16.
Brabletz
S
,
Bajdak
K
,
Meidhof
S
,
Burk
U
,
Niedermann
G
,
Firat
E
, et al
The ZEB1/miR-200 feedback loop controls Notch signalling in cancer cells
.
EMBO J
2011
;
30
:
770
82
.
17.
Behrens
J
. 
Cadherins and catenins: role in signal transduction and tumor progression
.
Cancer Metastasis Rev
1999
;
18
:
15
30
.
18.
Meyer
T
,
Hart
IR
. 
Mechanisms of tumour metastasis
.
Eur J Cancer
1998
;
34
:
214
21
.
19.
Sleeman
JP
,
Thiery
JP
. 
SnapShot: the epithelial–mesenchymal transition
.
Cell
2011
;
145
:
162
.
20.
Thiery
JP
,
Chopin
D
. 
Epithelial cell plasticity in development and tumor progression
.
Cancer Metastasis Rev
1999
;
18
:
31
42
.
21.
Wheelock
MJ
,
Shintani
Y
,
Maeda
M
,
Fukumoto
Y
,
Johnson
KR
. 
Cadherin switching
.
J Cell Sci
2008
;
121
:
727
35
.
22.
Comijn
J
,
Berx
G
,
Vermassen
P
,
Verschueren
K
,
van
GL
,
Bruyneel
E
, et al
The two-handed E box binding zinc finger protein SIP1 downregulates E-cadherin and induces invasion
.
Mol Cell
2001
;
7
:
1267
78
.
23.
Gregory
PA
,
Bert
AG
,
Paterson
EL
,
Barry
SC
,
Tsykin
A
,
Farshid
G
, et al
The miR-200 family and miR-205 regulate epithelial to mesenchymal transition by targeting ZEB1 and SIP1
.
Nat Cell Biol
2008
;
10
:
593
601
.
24.
Cho
HS
,
Kelly
JD
,
Hayami
S
,
Toyokawa
G
,
Takawa
M
,
Yoshimatsu
M
, et al
Enhanced expression of EHMT2 is involved in the proliferation of cancer cells through negative regulation of SIAH1
.
Neoplasia
2011
;
13
:
676
84
.
25.
Duns
G
,
van den Berg
E
,
van
DI
,
Osinga
J
,
Hollema
H
,
Hofstra
RM
, et al
Histone methyltransferase gene SETD2 is a novel tumor suppressor gene in clear cell renal cell carcinoma
.
Cancer Res
2010
;
70
:
4287
91
.
26.
Newbold
RF
,
Mokbel
K
. 
Evidence for a tumour suppressor function of SETD2 in human breast cancer: a new hypothesis
.
Anticancer Res
2010
;
30
:
3309
11
.
27.
Scott
KL
,
Plon
SE
. 
CHES1/FOXN3 interacts with Ski-interacting protein and acts as a transcriptional repressor
.
Gene
2005
;
359
:
119
26
.
28.
Hong
Y
,
Downey
T
,
Eu
KW
,
Koh
PK
,
Cheah
PY
. 
A ‘metastasis-prone’ signature for early-stage mismatch-repair proficient sporadic colorectal cancer patients and its implications for possible therapeutics
.
Clin Exp Metastasis
2010
;
27
:
83
90
.
29.
Ki
DH
,
Jeung
HC
,
Park
CH
,
Kang
SH
,
Lee
GY
,
Lee
WS
, et al
Whole genome analysis for liver metastasis gene signatures in colorectal cancer
.
Int J Cancer
2007
;
121
:
2005
12
.
30.
Pecot
CV
,
Rupaimoole
R
,
Yang
D
,
Akbani
R
,
Ivan
C
,
Lu
C
, et al
Tumour angiogenesis regulation by the miR-200 family
.
Nat Commun
2013
;
4
:
2427
.
31.
Arigoni
M
,
Barutello
G
,
Riccardo
F
,
Ercole
E
,
Cantarella
D
,
Orso
F
, et al
miR-135b coordinates progression of ErbB2-driven mammary carcinomas through suppression of MID1 and MTCH2
.
Am J Pathol
2013
;
182
:
2058
70
.
32.
Camps
C
,
Buffa
FM
,
Colella
S
,
Moore
J
,
Sotiriou
C
,
Sheldon
H
, et al
hsa-miR-210 is induced by hypoxia and is an independent prognostic factor in breast cancer
.
Clin Cancer Res
2008
;
14
:
1340
8
.
33.
Huang
X
,
Ding
L
,
Bennewith
KL
,
Tong
RT
,
Welford
SM
,
Ang
KK
, et al
Hypoxia-inducible mir-210 regulates normoxic gene expression involved in tumor initiation
.
Mol Cell
2009
;
35
:
856
67
.
34.
Lin
CW
,
Chang
YL
,
Chang
YC
,
Lin
JC
,
Chen
CC
,
Pan
SH
, et al
MicroRNA-135b promotes lung cancer metastasis by regulating multiple targets in the Hippo pathway and LZTS1
.
Nat Commun
2013
;
4
:
1877
.
35.
Rothe
F
,
Ignatiadis
M
,
Chaboteaux
C
,
Haibe-Kains
B
,
Kheddoumi
N
,
Majjaj
S
, et al
Global microRNA expression profiling identifies MiR-210 associated with tumor proliferation, invasion and poor clinical outcome in breast cancer
.
PLoS ONE
2011
;
6
:
e20980
.
36.
Samaan
G
,
Yugo
D
,
Rajagopalan
S
,
Wall
J
,
Donnell
R
,
Goldowitz
D
, et al
Foxn3 is essential for craniofacial development in mice and a putative candidate involved in human congenital craniofacial defects
.
Biochem Biophys Res Commun
2010
;
400
:
60
5
.
37.
Schuff
M
,
Rossner
A
,
Wacker
SA
,
Donow
C
,
Gessert
S
,
Knochel
W
. 
FoxN3 is required for craniofacial and eye development of Xenopus laevis
.
Dev Dyn
2007
;
236
:
226
39
.