RETRACTED ARTICLE: miR-195-5p/NOTCH2-mediated EMT modulates IL-4 secretion in colorectal cancer to affect M2-like TAM polarization

For this study, we collected 30 paired CRC samples and ANT (distance to cancer > 5 cm) from patients who had been diagnosed with primary CRC by pathological assessment of tissues and undergone surgeries with complete prognostic information at the Zhongnan Hospital of Wuhan University between January 2016 and July 2018. No any neoadjuvant radiotherapy or/and chemotherapy was managed. The study was endorsed by the Research Ethics Committee of Wuhan University (Wuhan, Hubei, PR China). Informed consents were obtained from all participating patients.

The normal intestinal epithelium cell line NCM460 and CRC cell lines (DLD-1, HCT116, SW480, SW620, HCT116, and HT29) were obtained from the Cell Bank of Wuhan University. The monocyte cell line was purchased from ATCC. Cells were cultured in RPMI 1640 medium (Invitrogen, Shanghai, China) containing 10% heat-inactivated (56 °C, 30 min) fetal calf serum, streptomycin (100 U/mL), and penicillin (100 U/mL), and maintained in a humidified atmosphere of 5% CO₂ at 37 °C.

Hsa-miR-195 mimic and mimic negative control (NC), hsa-miR-195 inhibitor, and inhibitor negative control (NC) were purchased from RiboBio (Guangzhou, China). NOTCH2 siRNAs (Notch2-homo-1815, 2501 and 3130) were purchased from GenePharma (Shanghai, China). Cells were cultured in complete medium at least 24 h before transfection. Cells were washed with phosphate-buffered saline (PBS, pH 7.4) before transient transfection. Transfections were performed by Lipofectamine 2000 (Invitrogen, USA) according to the manufacturer’s protocol with RNA oligonucleotides at a final concentration of 50 nM.

RIP was performed using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, MA, USA) according to the manufacturer’s protocol. In short, 5 × 10⁷ cells were lysed in polysome lysis buffer for each group. The expression of AGO2 protein was detected by western blotting, and then the supernatant was immunoprecipitated with antibody to AGO2 with protein A/G magnetic beads at 4 °C overnight. Magnetic bead-bound complexes with AGO2 were immobilized, and unbound material was washed off six times; after digesting proteins with Proteinase-K, the RNAs were extracted for quantitative real-time PCR and AGE (agarose gel electrophoresis) analyses.

Ten BALB/c athymic nude mice (female, 4–6 weeks old and 16–20 g) were purchased from Hubei Research Center of Laboratory Animals (Wuhan, China). All animal experiments were carried out in accordance with the Guide for the Care and Use of Laboratory Animals of Wuhan University. Each mouse was subcutaneously injected with 5 × 10⁶ HCT116 cells on the right flank region. And the liposomal clodronate was injected intravenously for macrophage depletion [27]. After 8 days, the transplanted nude mice were randomly divided into two groups (n = 5 each), miR-195-5p agomir or miR agomir NC (RiboBio, Guangzhou, China) was directly injected into the implanted tumor at the dose of 2 nmol/50 μL PBS, and THP1-induced macrophages were injected into the caudal veins at the dose of 10⁶ cells/50 μL per mouse every 3 days for eight times. Tumor dimension was measured at indicative time points. After 29 days, the weights and volumes of tumors were analyzed. One milliliter of blood was gathered via cardiac puncture into EDTA-containing tubes per mouse. The xenograft tumor and blood were collected for further experiments.

The mice blood samples were processed within 24 h of collection. Circulating tumor cells (CTCs) were isolated and identified by CTCBIOPSY® (Wuhan YZY Medical Science and Technology Co., Ltd., Wuhan, China) as previously reported by our group [28]. In brief, 1 mL blood was diluted with the 0.9% sodium chloride solution into 5-mL total volumes, then transferred to CTCBIOPSY® tubes with an 8-μm diameter aperture membrane. The samples were filtered by positive pressure from 12 to 20 mmHg through the device tubes. After that, the membranes in the device tubes were stained with Wright’s stain, then the morphology and number were rechecked using an ordinary microscope (BX51-Olympus, Japan).

For comparisons, one-way analyses of variance, Fisher’s exact tests, chi-squared tests, and two-tailed Student’s t tests were performed, as appropriate. P < 0.05 was considered statistically significant. The results were expressed as the mean ± SD from at least three independent experiments. Kaplan–Meier curves were used to calculate overall survivals, and the differences were analyzed by a log-rank test. All statistical analyses were conducted using the SPSS 20.0 statistical software (SPSS Inc., IL, USA).

We have found miR-195-5p in seven datasets with 387 CRC samples and 386 ANT samples [29]. Slattery et al. performed two newly non-coding RNA profiling analyses with 649 pairs and 28 pairs of colon cancer samples [30] (Additional file 2: Table S1). So we merged together to further confirm the downregulation of miR-195 in CRC (Additional file 3: Figure S1a). We conducted target prediction for validated miR-195-5p with high stringency. Target genes were obtained from both prediction algorithms and experimentally supported databases. Furthermore, gene set enrichment analysis (GSEA) revealed that the predicted targets were enriched in many cancer-related functional pathways (such as EMT, P53, and NOTCH pathways), which were positively correlated with CRC (Additional file 3: Figure S1b-d). To further investigate whether miR-195-5p and NOTCH2 correlate with the survivals of the CRC patients, we performed Kaplan–Meier and Cox’s proportional hazards regression model analyses of TCGA-COAD patients. We found that a low miR-195-5p level and high NOTCH2 level were significantly correlated with poor overall survivals (Additional file 3: Figure S1e-f). We next detected the expression of miR-195-5p and NOTCH2 in 30 pairs of human CRC and ANT samples. Similarly, RT-qPCR showed that miR-195-5p was significantly downregulated in CRC compared with ANT (Fig. 1a, b). Relatively, the expression of NOTCH2 protein was upregulated in CRC (Fig. 1c, d), whereas there was no significant difference in Notch2 mRNA between CRC and ANT (data not shown). Of note, the miR-195-5p level was negatively correlated with the NOTCH2 protein level in the tumors as calculated by Pearson’s correlation (R ² = 0.3837, P = 0.0003) (Fig. 1e).

To assess the role of miR-195-5p in colorectal cancer cells, we first examined the baseline miR-195-5p RNA and NOTCH2 mRNA levels in six cell lines (NCM460, HCT116, SW480, SW620, DLD-1, and HT29) by RT-qPCR (Additional file 3: Figure S1g-h). We found, compared with normal intestinal epithelium cell line (NCM460), a lower expression of miR-195-5p in DLD-1 and other cell lines. HCT116 (lowest level) and DLD-1 (highest level) cells were transfected with miR-195-5p mimic or miR mimic NC (negative control) and miR-195-5p inhibitor or miR inhibitor NC, respectively. The transfection efficiency was assessed by fluorescence microscopy (Additional file 3: Figure S1i). The effects of miR-195-5p on cell proliferation of HCT116 and DLD-1 cells were examined using clone formation assay and EdU immunofluorescence (IF) staining. Clonogenic assay showed that miR-195-5p decreased the clonogenic survivals of HCT116 cells compared with negative control (NC) groups, while miR-195-5p inhibitor-treated HCT116 cells showed a reversed phenotype, so does DLD-1 (Fig. 2a, b). In addition, EdU immunofluorescence staining assay revealed that miR-195-5p inhibited DNA synthesis in two cell lines (Fig. 2c, d). Conversely, the miR-195-5p inhibitor could mitigate this inhibition (Fig. 2c, d).

To determine the effect of altered miR-195-5p expression on the migration and invasion of CRC cells, miR-195-5p mimic, inhibitor, and paired negative control transfected CRC cells were wounded by scratching and maintained for 48 h. In the wound healing assay, cell motility was monitored at designated time points after scratches. The miR-195-5p overexpressed cells migrated toward the wound more slowly than the negative control or the miR-195-5p-inhibited cells (Fig. 2e, f). This result was also confirmed by the transwell migration assay. We found that migration abilities of the miR-195-5p overexpressed cells were significantly inhibited (Fig. 2g, h), whereas the migration ability was higher in the miR-195-5p inhibitor groups. In the transwell invasion assay, we found that invasive abilities of the miR-195-5p overexpressed cells were reduced and could be lifted by the miR-195-5p inhibitor on the contrary (Fig. 2i, j). Taken together, miR-195-5p lowers cell motility, migratory, and invasive abilities.

As mentioned earlier, through bioinformatics analysis, we found that miR-195-5p might target NOTCH2, and the expression of NOTCH2 protein and miR-195-5p was negatively correlated (Fig. 1e), which may affect CRC progression. This reminded that miR-195-5p is likely to modulate NOTCH2 translation. A series of experiments of miR-195-5p and NOTCH2 were performed to further understand the underlying mechanism (Fig. 3 and Additional file 4). After transfecting miR-195-5p mimic, inhibitor, and paired NC, the expression of NOTCH2 mRNA was detected by RT-qPCR. Intriguingly, there was not any difference in the expression levels of NOTCH2 mRNA in the groups of cells (Additional file 5: Figure S3a). Subsequently, the NOTCH2 and activated NOTCH2 (Ad-NOTCH2) protein, EMT-related protein E-cadherin, and Vimentin were verified by western blotting (WB) and immunofluorescence (IF). miR-195-5p mimic-treated cells expressed lower NOTCH2 and Ad-NOTCH2 (Additional file 4: Figure S2a, IF) and Vimentin and higher E-cadherin compared to NC groups, whereas higher NOTCH2, Ad-NOTCH2, and Vimentin and lower E-cadherin were detected in miR-195-5p inhibitor-treated cells (Fig. 3a, b, WB, and Additional file 4: Figure S2b, IF). The results suggest that miR-195-5p negatively regulates NOTCH2, Ad-NOTCH2, and Vimentin levels in CRC cells, while E-cadherin was positively regulated by miR-195-5p, suggesting negative regulations of miR-195-5p on the NOTCH2 pathway and EMT in CRC cells.

To confirm the direct regulation of miR-195-5p on NOTCH2 expression, the dual-luciferase reporter assay was carried out. We cloned the NOTCH2 3′-UTR into a luciferase reporter plasmid. NOTCH2 harbors two conserved miR-195-5p cognate sites (44–50 and 1275–1281 of NOTCH2 3′-UTR, Fig. 3c), which were predicted targets of miR-195-5p. The luciferase reporter vector pmiR-RB-REPORTTM-NOTCH2 3′-UTR or mutant reporter vector carrying point mutations of the putative miR-195-5p binding sites was co-transfected with miR-195-5p mimics or mimic NC, separately. The results indicate that miR-195-5p inhibits the luciferase activity in wild-type NOTCH2 3′-UTR transfected HCT116 and DLD-1 cells (Fig. 3d, P < 0.05), whereas it did not influence the activity of luciferase in carrying mutant NOTCH2 3′-UTR vector cells. These results suggested that miR-195-5p binds directly to NOTCH2 3′-UTR regions of two predicted binding sites.

We further analyzed the specific role of miR-195-5p for NOTCH2 expression. HCT116 cells were added with actinomycin D after 24 h after being transfected with miR-195-5p mimics or mimic NC. Then the NOTCH2 mRNA levels were detected at designated time points. The NOTCH2 mRNA levels were normalized by GAPDH, and the decay model was fit in a one-phase exponential decay model [31]. The half-lives of miR-195-5p and mimic NC transfected cell mRNA were 4.098 h and 5.102 h (Fig. 3e, no significance), which revealed that miR-195-5p did not affect NOTCH2 mRNA stability. Then AGO2-RNA immunoprecipitation assay (RIP) was performed. There was no difference in the AGO2 protein expression levels of miR-195-5p mimic and mimic NC groups (Additional file 5: Figure S3b, no significance). NOTCH2 mRNA was strongly enriched in the AGO2 immunoprecipitates compared with negative control IgG immunoprecipitates (Fig. 3f). And the miR-195-5p-treated group enriched more NOTCH2 mRNA compared with the NC group (Fig. 3f). Taken together, our results revealed that miR-195-5p might inhibit CRC cell EMT though binding to the NOTCH2 3′-UTR region and suppressing the NOTCH2 protein in a post-transcriptional manner.

We also assessed the carcinogenicity of NOTCH2 in CRC cells. Three siRNAs (Notch2-homo-1815, 2501 and 3130) were transfected to HCT116 cell, and the expressions of NOTCH2 mRNA and protein were detected with RT-qPCR and WB (Additional file 5: Figure S3c-d). The best siRNA (Notch2-homo-2501) was applied in subsequent experiments. Loss of NOTCH2 expression also contributed to inhibition of CRC cell clone formation (Fig. 4a, b), DNA synthesis (Fig. 4c, d), migration (Fig. 4e, f), and invasion (Fig. 4g, h). As previously mentioned, miR-195-5p inhibitor can facilitate CRC cell clone formation, DNA synthesis, proliferation, migration, and invasion. However, after silencing NOTCH2, the miR-195-5p inhibitor could not play the above role. Moreover, after NOTCH2 was overexpressed by the pGV219 vector, the abilities of CRC cell clone formation, DNA synthesis, migration, and invasion were enhanced (Additional file 6: Figure S4c-f). These effects could be partly abrogated by miR-195-5p mimic. Taken together, these results further confirmed the powerful carcinogenicity of NOTCH2 in CRC and indicated that the anticancer efficacy of miR-195-5p may be partly attributed to its suppressive role on NOTCH2.

Focusing on tumor cells and M2-like TAMs in TME, we detected NOTCH2 and CD163 expression and distribution in the CRC and paired ANT tissues of two CRC patients by immunofluorescence. NOTCH2 (red) was highly expressed in CRC tissues (Fig. 5a), especially the invasive tumor front (ITF; white dashed, Additional file 7: Figure S5a), whereas it was lowly expressed in adjacent normal tissues. Interestingly enough, the CD163 (green) expression was concentrated in the invasive tumor front, rather than the tumor nest (Fig. 5a, Additional file 7: Figure S5a). Moreover, miR-195-5p was negatively related with the expression of NOTCH2 and CD163 in CRC and ANT (Fig. 5a). These results might hint that high expression of NOTCH2 can recruit macrophages to ITF and induce M2 polarization, as several studies have shown that NOTCH2 was capable of enhancing IL-4 production though promoting GATA3 expression in mast cells or T cells [22, 32]. However, the mechanism of NOTCH2 inducing IL-4 production in CRC and whether affecting TAM polarization was rarely understood. To verify the hypothesis that NOTCH2 signaling enhances GATA3-mediated IL-4 production in CRC cells, we tested the expression of NOTCH2, GATA3, and IL-4 in patients’ tumor tissues and transfected CRC cells. For two patients, NOTCH2, GATA3, and endogenous IL-4 were strongly expressed and positively correlated in cancer tissues, whereas these were weakly expressed in ANT (Fig. 5b). The results were also confirmed in immunohistochemical staining serial sections (Additional file 7: Figure S5b). In CRC cells, inhibition of miR-195-5p increased the protein levels of NOTCH2, GATA3, and IL-4, but miR-195-5p mimic did abrogate these effects (Fig. 5c). Moreover, after silencing NOTCH2, relative expressions of GATA3 and endogenous IL-4 were decreased (Fig. 5). Comparatively, the levels of GATA3 and endogenous IL-4 were increased after NOTCH2 overexpression (Additional file 6: Figure S4a). Exocrine IL-4 levels in all cell supernatants were further detected by ELISA. The IL-4 level in the miR-195-5p inhibitor group was higher than that in the paired control group, while the IL-4 levels in the miR195-5p mimic-treated, si-NOTCH2, and IL-4 inhibitor (Suplatast Tosilate, IPD 1151 T, a cytokine inhibitor which can inhibit IL-4 expression by cells) groups were lower than those in the negative control groups (Fig. 5e). Moreover, mesenchymal cell lines (SW480 and SW620) expressed higher IL-4 compared to epithelial cell lines (HT29, HCT116, and DLD-1) (Additional file 8: Figure S6a). In conclusion, these results suggested that miR195-5p-mediated IL-4 production relied on the NOTCH2/GATA3/IL-4 signaling pathway in CRC.

To elucidate whether miR195-5p-mediated IL-4 in CRC cells affects macrophage alternative polarization, co-culture and migration assay of macrophages and CRC cells were performed. After being cultured with transfected CRC cell-conditioned medium, macrophages were inclined to the M2 phenotype (CD163^high CD80^low IL-10^high TGF-β^high) in the miR-195-5p inhibitor-treated group compared to the inhibitor NC (Fig. 5f, g). After adding IL-4 neutralizing antibody (Anti IL-4) in medium of the miR-195-5p inhibitor-treated group (miR-195-5p + Anti IL-4 group), this effect has disappeared. In the miR-195-5p mimic-treated groups, macrophages seemed to be M1-like (CD163 ^low IL-10 ^low) (Fig. 5f, g). With silenced NOTCH2 or inhibited IL-4 expression by IL-4 inhibitor, macrophages tended to be M1 activated (CD80^high CD163^low IL-10^lowTGF-β^low) and downregulation of CD163 (RT-qPCR and FCM, Fig. 5h, i, Additional file 6: Figure S4c-d). These results indicated that IL-4 from CRC cells may be the reason of M2-like TAM polarization. In addition, through wound healing and co-culture cell migration assay, we found that miR-195-5p inhibitor-treated cancer cells would like to favor macrophages while miR-195-5p mimic, si-NOTCH2, and IL-4 inhibitor reversed it (Additional file 8: Figure S6b-g). All in all, our finding indicated that miR195-5p-mediated IL-4-related M2-like TAM polarization and recruitment were a function of the NOTCH2-GATA3-IL-4 signaling pathway in CRC.

To further explore the role of miR-195-5p in vivo, we performed a xenograft nude mouse experiment via subcutaneously injecting HCT116 cells. We found that miR-195-5p upregulation significantly delayed tumor growth (Additional file 8: Figure S6h). And the tumor weights were also decreased on the 29th day post injection after miR-195-5p accumulation (Additional file 8: Figure S6i-j). Furthermore, EMT abilities of formed tumors derived from miR-195-5p-upregulated tumor cells were also impaired (Fig. 6a, b). These results confirmed the suppressing role of miR-195-5p in tumor progression. More importantly, the proposed regulations of miR-195-5p on the expression of NOTCH2, GATA3, and IL-4 in miR-195-5p-agomir-treated tumors were similar to the in vitro results, which substantiated the suppressor function of miR-195-5p in the NOTCH2/GATA3/IL-4 pathway inducing TAM recruitment and M2 polarization (Fig. 6a, b). Since the tumor EMT program may promote CTC generation, we had detected CTCs from blood of every mouse using blood film and CTC capture technology. CTCs were further identified by immunofluorescence staining (Fig. 6d) that miR-195-5p suppressed CTC generation (Fig. 6c, d). Summarily, our research showed that miR-195-5p inhibited the NOTCH2 signaling pathway leading to suppression of EMT and CTCs. miR-195-5p/NOTCH2-mediated IL-4 secretion during CRC EMT is responsible for M2-like TAM polarization.