MiR-378a is the most reduced miRNA in CRC tumors of patients and AOM/DSS mice.

CRCs arise from one or a combination of three different mechanisms, namely chromosomal instability (CIN), CpG island methylator phenotype (CIMP), and/or microsatellite instability (MSI) [4]. Most of the miRome studies on colon cancer focus on the colon tumor and adjacent non-tumor tissue [17,18,19]. Such a strategy only identified differentially expressed miRNAs triggered by CIN, CIMP or MSI since cells from both cancerous colon tissues and adjacent normal tissues have the same genome DNA. Importantly, the procedure of tumor resection is a potentially strong contributor to the variability of data among different studies, indicating that it is important to compare miRNAs between CRC patients and normal individuals. Previously, we have identified 37 differentially-expressed miRNAs in 80 colon tumors compared to 28 normal mucosa samples [20]. We further analyzed miRNA profiles from normal colon tissues and CRC tumors from AOM/DSS mice and identified 54 differentially-expressed miRNA (Fig. 1A) [21]. To identify miRNAs that are altered in CRC tumors of both patients and mice, we correlated altered 54 miRNAs in AOM/DSS mice with 37 miRNAs that are differentially-expressed in CRC patients [20]. This correlation analysis identified four miRNAs that were downregulated in both human and mouse CRC tumors (Fig. 1B). miR-378a-3p (miR-378a) was identified as the most reduced among four under-expressed miRNAs in CRC tumor, leading us to focus on miR-378a. qRT-PCR analysis of 28 normal colon tissues and 80 colon tumors further validated that miR-378a expression was robustly reduced in colon tumors versus normal colon tissues (Fig. 1C).

Fig. 1

miR-378a is reduced in CRC tumors of patients and AOM/DSS mice. A 54 differentially expressed miRNAs in CRC tumors of AOM/DSS mice (n = 3). Normal colons from wild-type mice (n = 3) served as the control. Fold change ≥ 2 and p ≤ 0.05. B Four miRNAs that are reduced in CRC tumors of patients and AOM/DSS mice. C qRT-PCR confirming that miR-378a was significantly reduced in 80 CRC tumors versus 28 normal colon tissues. D Reduced expression of Mdm2, Odc1, c-Myc, Tfrc, and Vcam1 in mouse colorectal cancer CT26.WT cells. Oligo Signal Transduction PathwayFinder MicroArray (Qiagen) was used to identify signaling pathways altered by miR-378a. E, F Reduced mRNA levels of c-MYC and ODC1 in 275 CRC tumors versus 41 normal colon tissues in TCGA database. TPM: transcripts per million. Expression levels were shown as Log2 (TPM + 1). P-value was determined by DEseq. Data represent mean ± SEM. *p < 0.05 (A–C: two-tailed student’s t test)

To determine the role of miR-378a in regulating CRC, we overexpressed miR-378a in mouse colorectal adenocarcinoma CT26.WT cells and determined expression of 117 genes related to the onset, development, and progression of tumorigenesis using the Signal Transduction PathwayFinder MicroArray (Qiagen). For this purpose, we constructed an expression vector of miR-378a by cloning miR-378a precursor into mini-circle vectors (MC). MCs are episomal DNA vectors that are produced as circular expression cassettes devoid of any bacterial plasmid DNA backbone [22]. Their small molecular size enables more efficient delivery and offers sustained expression over several weeks compared to standard plasmid vectors that only express for a few days after injection into mice [22]. This vector was referred to as MC-miR378. To rule out a non-specific effect of the plasmid, we generated a miR-378a mis-matched-expression vector by mutating the seed region of miR-378a (MC-miR378-MM). Transfection of MC-miR378 into CT26.WT cells significantly inhibited expression of five genes including Mdm2 (mouse double minute 2 homolog), Odc1, c-Myc, Tfrc (transferrin receptor 1) and Vcam1 (vascular cell adhesion molecule-1) (Fig. 1D). Among five genes reduced by miR-378a, no significant change in levels of MDM2, TFRC, and VCAM1 was observed in human CRC tumors (n = 275) versus normal colon specimen (n = 41) from TCGA database (Additional file 1: Fig. S1A–C), while levels of c-MYC and ODC1 were significantly increased (Fig. 1E, F). In summary, level of miR-378a is inversely correlated with mRNA levels of ODC1 and c-MYC in both human and mouse CRC tumors.

miR-378a inhibits expression of ODC1 by binding to its 3’UTR

It is known that dysregulated activities of ODC1 and c-MYC are important promoters of CRC. We, therefore, hypothesized that miR-378a was a potential tumor suppressor by repressing ODC1 and c-MYC expression. We next determined the molecular mechanism by which miR-378a inhibits expression of ODC1 and c-MYC. Combining bioinformatic prediction and mining of Ago HITS-CLIP database (high-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation from argonaute protein complex) [23], we identified a conserved binding site of miR-378a within the 3’UTR of ODC1 (Fig. 2A). Consistent with reduced miR-378a, expression of Odc1 was significantly increased in CRC tumors of AOM/DSS mice (Fig. 2B). To determine if miR-378a inhibits Odc1 directly by binding to the 3’UTR of Odc1, a luciferase reporter was employed, in which the 3’UTR containing wild-type or mutated miR-378a binding site was embedded into the downstream of the luciferase. Indeed, the luciferase activity was reduced by miR-378a in HCT116 cells (Fig. 2C), whereas mutation of the miR-378a binding site within the 3’UTR of Odc1 impaired the ability of miR-378a to reduce luciferase activity (Fig. 2C), suggesting the direct repression of miR-378a on Odc1. Injection of miR-378a into mice also repressed mRNA and protein levels of Odc1 (Fig. 2D). In human DLD-1 cells, overexpression of miR-378a led to reduced mRNA and protein levels of c-MYC and ODC1 (Fig. 2E). In summary, miR-378a is able to inhibit expression of both murine and human ODC1 by binding to their 3’UTRs.

Fig. 2

miR-378a inhibits expression of Odc1 via binding to its 3’UTR. A Graphic representation of the conserved miR-378a binding motif within the 3’UTRs of ODC1 between human and mouse. B mRNA levels of Odc1 in CRC tumors of AOM/DSS mice (n = 6) and normal colon tissues of WT mice (n = 6). C miR-378 significantly reduced luciferase activity of the luciferase reporter construct containing the 3’UTR of Odc1 with either wild-type or mutated miR-378a binding site. Mutation of the miR-378 binding site within the 3’UTR of Odc1 canceled the ability of miR-378a to inhibit luciferase activity. D mRNA and protein levels of Odc1 in AOM/DSS mice injected with MC-miR378 or MC-miR378-MM (MM, control). FC: fold change. E mRNA and protein levels of ODC1 and c-MYC in DLD-1 cells transfected with MC-miR378 or MC-miR378-MM. FC: fold change. Data represent mean ± SEM. **p < 0.01, ns: no significance (B–E: two-tailed student’s t test)

MiR-378a indirectly inhibits expression of c-MYC by inhibiting FOXQ1.

Although miR-378a significantly inhibited expression of c-MYC (Fig. 1D), no binding motif for miR-378a was identified within the 3'UTR of c-MYC. Luciferase assay confirmed that miR-378a had no binding site within the 3’UTR of c-MYC (Additional file 1: Fig. S2A, B). We, therefore, speculated that miR-378a inhibited transcription of c-MYC expression via directly targeting a transcription activator that has a binding site within the promoter of c-MYC. To test this speculation, we used MatInspector software to scan the promoter of c-MYC and identified a highly conserved binding motif for FOXQ1 (Fig. 3A) [24]. FOXQ1 is an oncogenic transcription factor that binds to GTTT core motif [25]. Furthermore, combining bioinformatic prediction and mining of Ago HITS-CLIP database, we identified a potential binding site of miR-378a within the 3’UTR of FOXQ1 (Fig. 3B). Level of FOXQ1 was increased in CRC patients (Fig. 3C). These findings led us to speculate that miR-378a inhibited expression of c-MYC by repressing transcription of FOXQ1 (Fig. 3A). miR-378a significantly reduced luciferase activity of the construct containing the 3’UTR of FOXQ1, whereas disrupting the interaction between miR-378a and FOXQ1 by mutating miR-378a binding site prevented miR-378a from repressing luciferase activity (Fig. 3D). Overexpression of miR-378a in DLD-1 cells reduced mRNA and protein levels of FOXQ1 and c-MYC (Fig. 3E). In murine CT26.WT cells, overexpression of miR-378a significantly inhibited expression of FoxQ1 and c-Myc (Fig. 3F). Together, FOXQ1 is a direct target of miR-378a.

Fig. 3

miR-378a inhibits c-MYC transcription by directly targeting FOXQ1. A A proposed mechanism by which miR-378a inhibits transcription of c-MYC by directly targeting FOXQ1 that serves as a transcription activator of c-MYC. B The predicted miR-378a binding motif within the 3’UTRs of human and mouse FOXQ1. C mRNA levels of FOXQ1 in CRC tumors of 275 patients versus 41 normal colon tissues in TCGA database. TPM: transcripts per million. Expression levels were shown as Log2 (TPM + 1). P-value was determined by DEseq. D Luciferase activities of the reporter constructs containing the 3’UTR of FoxQ1 with either WT or mutated miR-378a binding site after treatment with MC-miR378. HCT116 cells transfected with the luciferase reporter vector and MC-miR378-MM served as a control. E mRNA and protein levels of FOXQ1 and c-MYC in DLD-1 cells transfected with MC-miR378 or MC-miR-378-MM (control). FC: fold change. F Protein and mRNA levels of FoxQ1 and c-Myc in CT26.WT cells transfected with MC-miR378 or MC-miR-378-MM (control). FC: fold change. Data represent mean ± SEM. **p < 0.01, and ns: no significance (D–F two-tailed student’s t test)

We next determined if FOXQ1 was a transcription activator of c-MYC. Luciferase assay was used to evaluate if FOXQ1 was able to inhibit transcription of c-MYC. Overexpression of FOXQ1 significantly increased activity of the c-MYC promoter (Fig. 4A). To determine if the binding site for FOXQ1 within the c-MYC promoter was required for increased expression of c-MYC, we mutated the FOXQ1 binding site within the c-MYC promoter. As expected, this mutation prevented FOXQ1 from driving transcription of c-MYC (Fig. 4A). Overexpression of FOXQ1 strongly enhanced transcription of both c-MYC and ODC1 in DLD-1 cells (Fig. 4B). Overexpression of FoxQ1 in mouse colon cells lead to increased c-Myc and Odc1 (Fig. 4C). As revealed by chromatin immunoprecipitation (ChIP), DNA fragments containing FOXQ1 binding motif within the c-MYC promoter were immune-precipitated from genomic DNA from DLD-1 cells by an FOXQ1 antibody (Fig. 4D), suggesting that FOXQ1 was able to physically bind to the promoter of c-MYC. Delivery of miR-378a into mice reduced mRNA levels of FoxQ1 and c-Myc (Fig. 4E). Our hypothesis is that miR-378a inhibits polyamine synthesis by directly targeting ODC1 and blocking the FOXQ1-MYC-ODC1 axis. We next measured the effect of miR-378a on levels of polyamine in DLD-1 and HCT116 cells. As expected, miR-378a significantly reduced spermine (SPM), putrescine (PUT), and spermidine (SPD) in DLD-1 and HCT116 cells (Fig. 4F). In summary, FOXQ1 is a transcription activator of c-MYC. miR-378a inhibited polyamine synthesis by directly repressing ODC1 and blocking the FOXQ1-c-MYC-ODC1 signaling.

Fig. 4

FOXQ1 activates transcription of c-MYC. A Luciferase activity of the reporter construct containing c-MYC promoter with either WT or mutated FOXQ1 binding site after treatment with FOXQ1 expression vector or empty vector (control). B mRNA levels of c-MYC and ODC1 in DLD-1 cells transfected with FOXQ1 expression vector or empty vector (control). C mRNA levels of c-Myc and Odc1 in CT26.WT cells transfected with FoxQ1 expression vector or empty vector (control). D ChIP assay was performed using genomic DNA isolated from DLD-1 cells; and the binding of FOXQ1 to the endogenous promoter of c-MYC was detected using an FOXQ1 antibody; N.S.: non-specific control, which is located 10 kb downstream of the predicted FOXQ1 binding site. E mRNA levels of c-Myc and Odc1 in AOM/DSS mice injected with MC-miR378 or MC-miR378-MM (MM, control). F Levels of putrescine (PUT), spermidine (SPD) and spermine (SPM) in DLD-1 and HCT116 cells after transfection of MC-miR378 or MC-miR378-MM (control). Data represent mean ± SEM. **p < 0.01, ***p < 0.001, and ns: no significance (A–C, E, F: two-tailed student’s t test; D: two-way ANOVA test)

MiR-378a inhibited growth and proliferation of CRC cells via inhibiting polyamine synthesis.

To determine the role of miR-378a in tumorigenesis of CRC, we overexpressed miR-378a in DLD-1 cells, which led to reduced mRNA levels of FOXQ1 and ODC1 (Additional file 1: Fig. S3A). Overexpression of miR-378a significantly reduced ODC activity in DLD-1 cells and polyamine synthesis (Fig. 5A, Additional file 1: Fig. S3B), which subsequently inhibited their growth and viability, as revealed by colony formation and MTT assay (Fig. 5B, C). In addition to inhibiting proliferation and migration, miR-378a also significantly induced apoptosis and death of DLD-1 cells (Fig. 5D). Would healing assay further validated the inhibitory effect of miR-378a on growth of DLD-1 cells (Fig. 5E).

Fig. 5

ODC1 mediates the role of miR-378a in inhibiting proliferation and growth and inducing apoptosis of CRC cells. A Enzyme activity of ODC in DLD-1 cells transfected with MC-miR378 or MC-miR378-MM (control). B Soft agar colony formation assay revealed that miR-378a overexpression inhibited growth and viability of DLD-1 cells. DLD-1 cells were transfected with MC-miR378 or MC-miR378-MM (control). C Reduced viability of DLD-1 cells after MC-miR378 transfection, as revealed by MTT assay. DLD-1 cells treated with MC-miR378-MM served as the control. D miR-378a induced apoptosis of DLD-1 cells, as revealed by flow cytometry. A statistical graph of Annexin V-FITC/PI staining is shown. E Wound healing showing that miR-378a significantly impaired growth and migration of DLD-1 cells. F miR-378a treatment reduced ODC enzyme activity, while additional treatment of TPs of ODC1 and FOXQ1 impaired the ability of miR-378a to inhibit ODC activity. Three groups of DLD-1 cells were transfected with MC-miR378-MM (control), MC-miR378, or a combination of MC-miR378 and TPs of ODC1 and FOXQ1. G miR-378a reduced proliferation and growth of DLD1 cells and induced apoptosis of DLD1 cells, while additional treatment of TPs of FOXQ1 and ODC1 counteracted the ability of miR-378a to repress proliferation and growth and induce apoptosis of DLD-1 cells, as revealed by Wound Healing, MTT assay, and flow cytometry analysis, respectively. Data represent mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 (A–E: two-tailed student’s t test, F–G: two-way ANOVA test)

MiR-378a targets many genes simultaneously. We next determined whether the FOXQ1-MYC-ODC1-mediated polyamine synthesis, at least in part, was responsible for miR-378a to inhibit proliferation and induce apoptosis. For this purpose, we used a novel technique Target Protector morpholinos (TP) to prevent the interaction of miR-378a with FOXQ1 and ODC1. [26] The morpholinos were complimentary to the miR-378a binding sites present within the 3’UTRs of FOXQ1 and ODC1 mRNAs, thereby preventing miR-378a from binding to the 3’UTRs of FOXQ1 and ODC1. Specifically, three groups of DLD-1 cells were transfected with MC-miR378-MM, MC-miR-378 or a combination of MC-miR-378 and TPs of ODC1 and FOXQ1. Overexpression of miR-378a reduced expression of FOXQ1 and ODC1, while additional treatment of TPs of FOXQ1 and ODC1 impaired the ability of miR-378a to prevent their expression (Additional file 1: Fig. S3C). MiR-378a treatment reduced ODC activity and polyamine synthesis, while additional treatment of FOXQ1 and ODC1 TPs offset the effects of miR-378a (Fig. 5F, Additional file 1: Fig. S3D). Treatment of FOXQ1 and ODC1 TPs, which disrupted the interaction between miR-378a and 3’UTRs of FOXQ1 and ODC1, impaired the ability of miR-378a to inhibit proliferation and migration and induce apoptosis of DLD-1 cells (Fig. 5G). In sum, the FOXQ1-MYC-ODC1 axis mediates the roles of miR-378a in inhibiting polyamine synthesis, preventing proliferation and growth, and inducing apoptosis of CRC cells.

MiR-378a prevented overproduction of polyamine and growth of CRC in AOM/DSS mice

Carcinogen-induced colon cancer in mice simulates the phases of initiation and progression of tumor that occur in humans [27]. To evaluate the potential of miR-378a to inhibit CRC in vivo, we exposed mice to AOM and DSS to induce CRC (Fig. 6A). [27] The AOM/DSS mice recapitulate a chemical-induced carcinogenesis, inflammatory effects, colitis-driven initiation stage and sequential tumor progression [27]. Therefore, we next determined gain-of miR-378a function against CRC in AOM/DSS mice. The small molecular size of MCs enables efficient delivery and offers a sustained expression for a few weeks, as described above. We therefore used MC-miR378, as diagramed in Fig. 6B, to overexpress miR-378a in mice. Two groups of AOM/DSS mice were injected with either MC-miR378-MM (control) or MC-miR378 at a dose of 2.5 μg/g by tail vein weekly for ten weeks. Injection of MC-miR378 into AOM/DSS mice led to increased miR-378a in the colon (Additional file 1: Fig. S4). MiR-378a treatment resulted in a significant reduction in tumor size and number (Fig. 6C). We next performed the histopathological evaluation of intestinal inflammation, hyperplasia and tumorigenicity using H&E staining. Consistent with previous reports [28], diffuse mucosal hyperplasia was observed in the distal colon of AOM/DSS mice (control) (Fig. 6D). In contrast, no or much less hyperplasia was observed in the colonic mucosae of AOM/DSS mice treated with miR-378a (Fig. 6D). Inflammation is a major driver of CRC development in AOM/DSS mice. A significant reduction in inflammation and hyperplasia score was observed in miR-378a-treated AOM/DSS mice (Fig. 6E). MiR-378a treatment also reduced adenocarcinoma (Fig. 6F). These observations indicated that miR-378a was able to alleviate colonic hyperplasia and adenocarcinoma. H&E staining revealed that CRC tumors were well differentiated with an acinar pattern in MC-miR378-treated AOM/DSS mice, whereas CRC tumors in control AOM/DSS mice exhibited poor differentiation, indicating the strong inhibitory effect of miR-378a on colon cancer development (Fig. 6D). Mechanistically, miR-378a led to reduced mRNA levels of FoxQ1, c-Myc and Odc1, enzyme activity of ODC and polyamine synthesis (Fig. 7A-C). K67 staining (general marker of cell proliferation) confirmed the inhibitory effect of miR-378a on growth of colon cancer, which was reflected by a significant reduction in proliferating colon cells (Fig. 7D-E). In summary, miR-378a maintains an appropriate level of polyamine by precisely fine tuning polyamine synthesis, which subsequently prevents CRC development.

Fig. 6

miR-378a significantly prevented CRC development in AOM/DSS mice. A Scheme for AOM/DSS-induced colon carcinogenesis in mice. Mice were kept on normal chow. Five-weeks old mice were given AOM, followed by 1% DSS in water for 4 days. At 13 weeks of age, mice were injected with MC-miR378-MM (n = 6, control) or MC-miR378 (n = 6) weekly for 10 weeks. B Diagram of miR-378a mini-circle expression vector (MC-miR378). C Representative photos of CRC tumors in mice treated with MC-miR378 and MC-miR378-MM. Black arrows indicate colon tumors. Both total tumor number and tumors with the size of ≥ 2 mm were reduced in AOM/DSS mice treated with MC-miR378. D H&E staining of colon from two groups of mice. Red arrowhead: acinar regions; green arrowhead: inflammation regions; and blue arrowhead: hyperplasia regions. E, F Scores of hyperplasia, inflammation and adenocarcinoma in AOM/DSS mice treated with MC-miR378 or MC-miR378-MM. Data represent mean ± SEM. **p < 0.01 and ***p < 0.001 (Fig. 6C, E–F: two-tailed student’s t test)

Fig. 7

miR-378a reduced enzyme activity of ODC and proliferation of colon cells in AOM/DSS mice. A Levels of miR-378a and mRNA levels of FoxQ1, c-Myc and Odc1 in AOM/DSS mice injected with MC-miR378 and MC-miR378-MM. B Reduced enzyme activity of ODC in AOM/DSS mice treated with MC-miR378. C Levels of putrescine (PUT), spermidine (SPD) and spermine (SPM) in AOM/DSS mice injected with MC-miR378 or MC-miR378-MM (control). D, E A significant reduction in proliferation of colon cells in AOM/DSS mice treated with MC-miR378, as revealed by Ki67 staining. Data represent mean ± SEM. **p < 0.01 and ***p < 0.001 (Fig. 7A-E: two-tailed student’s t test)