Activation of miR-500a-3p/CDK6 axis suppresses aerobic glycolysis and colorectal cancer progression

According to the data of a cohort of 551 patients from the cBioportal database, we found that the expression of has-mir-500a was significantly associated with the prognosis of CRC patients (Additional file 1: Table S1) Patients who had low miR-500a expression had significantly poor overall survival (P = 0.008, Additional file 2: Fig. S1A) and progression-free survival (P < 0.001, Additional file 2: Fig. S1B). We firstly examined the miR-500a-3p expression by qPCR in CRC cell lines, including HCT116, HT29, RKO, SW620, SW48 and SW480, and normal human colon epithelial cell line NCM460. The expression level of miR-500a-3p was significantly down-regulated in CRC cells than that in NCM460 cells (Fig. 1A). In addition, it was significantly downregulated in CRC tissues compared to adjacent non-tumor tissues (Fig. 1B). Next, we examined miR-500a-3p expression in the cohort that contained 134 continuous CRC specimens. We divided the patients into low or high-expression groups by a cut-off of the median expression value. We found that the expression level of miR-500a-3p was negatively correlated with tumor size (P = 0.022) and differentiation (P = 0.043) (Table 1). Survival analysis showed that low miR-500a-3p expression was significantly correlated with poor overall survival (P = 0.0026, Fig. 1C) and disease-free survival (P = 0.0276, Fig. 1D). To further evaluate the prognostic value of miR-500a-3p, both univariate and multivariate analyses were performed. The results indicated that high miR-500a-3p expression is an independent protective factor for CRC [HR, 0.109; 95% confidence interval (CI), 0.014–0.875; P = 0.037]. (Additional file 1: Table S2). We also analyzed the correlation of miR-500a-3p expression with the biomarkers of cell proliferation (Ki-67) and glucose uptake (Standard Uptake Value, SUV) in CRC tissues. The expression of miR-500a-3p was significantly lower in the high Ki67 group than that in the low Ki67 group (Fig. 1E). Besides, in patients receiving PETCT preoperatively (n = 52), there was a negative correlation between the expression of miR-500a-3p and SUV with the P value of 0.038 and r value of -0.262 (Fig. 1F). Altogether, low expression level of miR-500a-3p predicts poor prognosis in CRC patients and it may suppress CRC by regulating tumor proliferation and glucose metabolism.

HCT116/SW480 cells were transfected with miR-500a-3p mimics to investigate the function of miR-500a-3p in vitro. Transfection efficiencies were confirmed by RT-qPCR (Additional file 2: Fig. S2A). CCK-8 and colony formation assays indicated that CRC cell proliferation and colony formation were significantly inhibited by miR-500a-3p mimics (Fig. 2A–D). Next, fluorescence-activated cell sorting (FACS) analysis revealed an increase in G1-phase cells and a concomitant decrease in S-phase cells by miR-500a-3p (Fig. 2E, F). We also found that miR-500a-3p mimics did not promote apoptosis significantly (Additional file 2: Fig. S2B). Transwell assay showed that the miR-500a-3p mimics markedly inhibited CRC cell invasion (Additional file 2: Fig. S2C). We established xenograft tumor models with SW480 and HCT116 cells stably overexpressing miR-500a-3p (Additional file 2: Fig. S2D, E). The mean volume and weight of tumors were lower in the miR-500a-3p-overexpressing group than the vector group (Fig. 2G–I and Additional file 2: Fig. S2F–H). The protein expression of cell proliferation (Ki-67) in the xenograft tumors was lower in miR-500a-3p overexpression group than vector group (Fig. 2J). Altogether, these results indicated that miR-500a-3p suppresses CRC cell proliferation in vitro and in vivo.

The expression of miR-500a-3p in CRC patients is negatively correlated with glucose uptake, indicating that miR-500a-3p may have an inhibitory effect on glycolysis. Cellular glycolytic activity was measured. As a result, overexpression of miR-500a-3p significantly decreased glucose uptake (Fig. 3A), lactate production (Fig. 3B) and intracellular ATP levels (Fig. 3C) in both HCT116 and SW480 cells. The extracellular acidification rate (ECAR), measured by the Seahorse analyzer, was downregulated by miR-500a-3p (Fig. 3D). Furthermore, targeted metabolomics focused on the energy metabolism pathway was performed, and lower level of glycolytic metabolites like glucose 6-phosphate, fructose 6-phosphate, fructose 1,6-bisphosphate, dihydroxyacetone phosphate and 3-phospho-glycerate caused by overexpression of miR-500a-3p was found (Fig. 3E and Additional file 2: Fig. S3A–E). We then measured the expression of major genes involved in glucose transport and glycolysis. miR-500a-3p mimics reduced the RNA level of HK2, PFKP, PGK1 and LDHA (Fig. 3F) and the protein level of GLUT1, HK2 and PKM2 (Fig. 3G). Since both RNA and protein levels of HK2 were downregulated by miR-500a-3p, HK2 was evaluated in the CRC cohort. It was negatively correlated with the miR-500a-3p levels (r = − 0.418, P < 0.0001, Fig. 3H, I), confirming the negative regulation of HK2 by miR-500a-3p in CRC tissues. These data suggest that the overexpression of miR-500a-3p causes a decrease in glucose metabolism via the regulation of multiple genes involved in glycolysis.

To explore the mechanisms by which miR-500a-3p inhibits the proliferation and glycolysis of CRC, we firstly used the publicly available algorithms TargetScan and miRDB to find the potential binding sites of glycolytic enzymes, like HK2, PFKP, PGK1 and LDHA, that are downregulated by miR-500a-3p. However, their 3′UTR did not have the conserved binding site of miR-500a-3p, indicating it may regulate glycolysis indirectly (Additional file 2: Fig. S4A). CDK6, a novel glycolysis regulator [17], was predicted as a target of miR-500a-3p by TargetScan and miRDB database (Fig. 4A). qPCR and Western blotting analysis showed that miR-500a-3p reduced both mRNA and the protein level of CDK6, indicating that CDK6 may act as a downstream target of miR-500a-3p (Fig. 4B, C). To further confirm this, we constructed two pairs of luciferase reporter plasmids containing the predicted seed sequence of miR-500a-3p in the 3′-UTR of the CDK6 mRNA and a control reporter containing the mutated sequence of the same fragment (Fig. 4D). miR-500a-3p significantly inhibited the luciferase activity in cells containing the wild-type CDK6 sequence, but not the mutant sequence (Fig. 4E, F). Finally, we evaluated the protein level of CDK6 in clinical CRC samples using IHC (Fig. 4G), and a significant negative correlation was revealed between miR-500a-3p and CDK6 expression (P = 0.0012, r = − 0.285, Fig. 4H). Collectively, these findings suggest that CDK6 is a direct target of miR-500a-3p in CRC.

We examined CDK6 protein levels in 20 paired human CRC tumors and adjacent nontumor tissues using IHC (Fig. 5A). The IHC staining results revealed that it was significantly upregulated in CRC tissues compared to nontumor tissues (Fig. 5B). Survival analysis of a large CRC cohort showed that high CDK6 expression was significantly correlated with poor overall survival (P = 0.0333, Additional file 2: Fig. S5A) and disease-free survival (P = 0.0559, Additional file 2: Fig. S5B). CCK-8 and cell cycle assays indicated that CDK6 enhanced CRC cell proliferation (Additional file 2: Fig. S5C, D). Then, we explored the influence of CDK6 on the aerobic glycolysis of CRC cells. As exhibited in Fig. 5C–F, CDK6 promoted glucose uptake, the production of lactate and ATP and the ECAR in SW480 cells. The expression of major genes involved in glucose transport and glycolysis were also measured. The RNA level of GLUT1, LDHB, HK1 and HK2 and the protein level of GLUT1, HK2 were elevated after overexpressing CDK6 (Fig. 5G, H). In CRC tissues, the IHC scores of HK2 was positively correlated with that of CDK6 (r = 0.411, P < 0.0001, Fig. 5I). These findings indicated that CDK6 plays an important role in promoting aerobic glycolysis in CRC.

To address whether miR-500a-3p exerted its role through targeting CDK6, we conducted rescue experiments. The CCK-8 and colony formation assays showed that CDK6 promoted CRC cell proliferation and colony formation and offset the inhibitory effect caused by miR-500a-3p (Fig. 6A, B). As shown in Fig. 6C, CDK6 overexpression counteracted the inhibitory effect of miR-500a-3p on the cell cycle of CRC cells. We explored the influence of miR-500a-3p/CDK6 axis on the aerobic glycolysis of CRC cells by measuring glucose uptake, lactate production, intracellular ATP level and ECAR. The inhibition of glycolysis caused by miR-500a-3p could be partly eliminated by CDK6 overexpression (Fig. 6D–G). Together, these results indicated that miR-500a-3p suppressed CRC proliferation and aerobic glycolysis through down-regulating CDK6.

From January 2015 to March 2015, 134 continuous CRC specimens and 20 matching adjacent non-tumor specimens obtained from CRC patients at the Zhongshan Hospital, Fudan University were collected. The inclusion criteria were as follows: with histologically confirmed CRC, receiving radical surgery, having complete follow-up. The exclusion criteria were receiving preoperative chemotherapy or radiation, pregnancy or breastfeeding, previous malignancy within 5 years or no written informed consent. The primary tumor ki67 score is the routine pathological evaluation of our center and was evaluated by two experienced pathologists. Patients in the cohort (134 continuous CRC specimens) who received PETCT preoperatively (n = 52) were included to analyze the association between miR-500a-3p expression and the SUV of the primary tumors. The study was authorized by the Ethics Committee of the Zhongshan Hospital, Fudan University. Written informed consent was obtained from each patient.

We collected gene expression profiles of 551 COADREAD samples from the TCGA cohort (http://​gdac.​broadinstitute.​org/​). Data including mRNA expression level and miRNA expression level. All the data have been standardized. For miRNAs differential analysis, we calculated the mean expression level of these miRNAs and divided them into high and low expression groups.

The R package DESeq2 (version 1.22.2) was used to calculate the differential expressed t statistics for microarray and RNA sequencing data. We used the univariate Cox proportional hazards model to examine the associations between gene expression and overall survival. MiRNAs with P values less than 0.05 were considered to be statistically significant and included in consensus survival analysis.

The whole cohort which contained 134 continuous CRC specimens were stained with the CDK6 antibody (ab124821, Abcam) and HK2 antibody (2867T, Cell Signaling Technology). The 20 matched adjacent non-tumor specimens were stained with CDK6 antibody (ab124821, Abcam). IHC scores were conducted according to the ratio and intensity of positive-staining areas. The staining areas were scored as follow: 0–25%, score 1; 25–50%, score 2; 50–75%, score 3; and beyond 75%, score 4. The signal intensity was scored on a scale of 0–3: 0-negative, 1-weak, 2-moderate and 3-strong. IHC scores were averaged from two experienced pathologists who scored the IHC staining independently.

Human CRC cell lines [HCT116 (RRID: CVCL_0291) and SW480 (RRID: CVCL_0546)] and normal human colon epithelial cell line NCM460 (RRID: CVCL_0460) were obtained from the National Collection of Authenticated Cell Cultures (Shanghai, China). All human cell lines have been authenticated using STR profiling within the last three years and all experiments were performed with mycoplasma-free cells. These cell lines were maintained in Dulbecco’s Modified Eagle Medium (Logan Utah, HyClone, USA) with 10% fetal bovine serum (FBS; Gibco), 1% penicillin (10 U/mL) and 1% streptomycin (10 μg/mL) in an incubator with the environment of 37 °C and 5% CO₂.

RNA was isolated with Trizol reagent. The purity and the concentration of different RNA samples were measured using NanoDrop ND-1000 (Thermo Scientific, USA). Reverse transcription reaction was performed using miRcute miRNA cDNA synthesis Kit (kr211, Tiangen, China) for miRNAs and SuperScript IV VILO cDNA synthesis kit (11756050, Thermo Scientific, USA) for mRNAs. Quantitative RT-PCR (qRT-PCR) was performed using SYBR® Premix Ex Taq™ (Takara) containing mRQ 3′ Primer on an ABI 7500 platform (Applied Biosystems, Carlsbad, CA, USA). The relative quantities of miR-500a-3p in cells were normalized to U6. Beta-actin was used as an internal control for mRNA detection. The primer sequences for miRNA and mRNA detection are listed in Additional file 1: Table S3.

The lentivirus particles overexpressing human miR-500a-3p and CDK6 were purchased from Genomeditech (Shanghai, China). miR-500a-3p mimics were synthesized by Genomeditech (Shanghai, China). The sequences of miR-500a-3p mimic were 5′-AUGCACCUGGGCAAGGAUUCUG-3′. The miRNA oligonucleotides were transfected using Lipofectamine® RNAiMAX™ (50 nmol/L; Invitrogen, CA, USA).

At 72 h after lentivirus infection or at 24 h after miRNA transfection, HCT-116 (2500/well) and SW480 (2000/well) cells were plated into 96-well culture plates. Cell viability was measured at 24, 48, 72 and 96 h post-seeding, using the Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) according to the manufacturer’s instructions.

HCT-116 (1500/well) and SW480 (2000/well) cells were seeded in six-well plates after 72 h lentivirus infection. The culture medium was changed every 3 days. After incubating for 2 weeks, colonies were fixed with 4% paraformaldehyde for 15 min and stained with 5% Giemsa for 20 min. The colonies containing at least 50 cells were scored.

Seventy-two hours after infection, cells were harvested, washed with phosphate-buffered saline (PBS), and fixed in 70% ethanol at 4 °C overnight. After fixation, cells were washed with PBS before suspension in RNase A/propidium iodide solutions (100 mg/mL RNase A and 5 μg/mL propidium iodide). Cells were incubated at room temperature for an hour. Stained cells were analyzed by a FACScan flow cytometer (BD Biosciences, Mountain View, CA, USA).

Cell apoptosis was assessed by annexin V/propidium iodide (BD Biosciences, San Jose, CA, USA). Cells were harvested, washed with PBS, and resuspended in 1× binding buffer at a concentration of 1 × 10⁶ cells/mL. A total of 100 μL solution was transferred to a new tube and added with 5 μL of APC Annexin V and 5 μL of propidium iodide. Cells were incubated at room temperature for 15 min in the dark and then analyzed by a FACScan flow cytometer.

The cell invasion assay was performed in 24-well transwell chambers pre-coated with Matrigel (Corning, NY, USA). HCT-116 cells (10⁵/well). Serum-free medium (200 μL) was placed into the upper chamber, and 600 μL complete medium (with 10% FBS) was filled into the lower chamber. Cells on the inner membrane were removed with a cotton swab after 24 h. The outer membrane was fixed in 4% formaldehyde (in PBS) and stained with 0.5% crystal violet. Cell numbers were counted and averaged in five random fields at a magnification of 100×.

BALB/c nude mice (6 weeks old) were purchased from Shanghai SLAC Laboratory animal co. ltd (shanghai, China) and randomly divided into 2 groups (n = 7/6). SW480 (3 × 10⁶) or HCT116 (4 × 10⁶) cells that stably transfected with miR-NC or miR-500a-3p were injected at the back region of BALB/c mice subcutaneously. The volume of tumors was measured twice a week using a vernier caliper with the method of volume = (length × width²)/2. The mice were killed after inoculation for 30 d, and the tumors were weighed. Tumor tissues were subjected to the expression analysis of ki67. The procedures in this study were permitted by the Animal Research Committee of the Zhongshan Hospital, Fudan University.

The putative miR-500a-3p binding sequence and the matching mutant binding sequence in the 3′untranslated region (3′UTR) of CDK6 were amplified and inserted into pGL3 luciferase reporter vector (Genomeditech, shanghai, China). SW480 cells were transfected with miR-NC or miR-500a-3p and the above-constructed reporter plasmids. After transfection for 48 h, the luciferase activities in different groups were determined by the dual-luciferase reporter assay system (Promega).

SW480 and HCT116 cells were cultured with an FBS-free medium and culture medium was collected after 24 h. Glucose Uptake Colorimetric Assay kit (Biovision, Milpitas, California, USA), Lactate Assay Kit (Biovision) and ATP Colorimetric Assay kit (Biovision) were utilized to detect the glucose consumption, the lactate production and ATP in CRC cells according to the manufacturer’s instructions. The glucose, lactate and ATP levels were normalized to total cell protein.

Extracellular acidification rate (ECAR) was detected using a Seahorse XF96 analyzer (Seahorse Biosciences, USA). SW480 cells (10⁵/well) were seeded in a 96-well XF96 microplate (Seahorse Biosciences, USA). Before experiments, cell culture medium was replaced and cells were then incubated with assay medium for 1 h at 37 °C in a CO₂-free incubator. ECAR was detected using a sequential injection of 10 mM glucose, 2 mM oligomycin (Sigma-Aldrich) and 50 mM 2-deoxyglucose (2-DG, Sigma-Aldrich). Each cycle of measurement involved mixing (3 min), waiting (2 min), and measuring (3 min) cycles.

The targeted metabolomics analyses were performed using an HPLC system (Agilent 1290, Agilent Technologies) and a mass spectrometer (Agilent 5500, Agilent Technologies). A 10-cm dish of cultured tumor cells was collected, adding 1 mL acetonitrile/methanol water (v, 2:2:1) and storing at − 80℃ after quick freezing in liquid nitrogen. Sample preparation processes were performed in accordance with the above method of parallel preparation of QC samples. MRM transitions representing the metabolites were simultaneously monitored, and the positive/negative polarity switching was used. Data analyses were performed as instructed by Shanghai Applied Protein Technology [18].

Whole protein extracts were lysed by radioimmunoprecipitation assay (RIPA) buffer (Thermo Fisher Scientific) according to the manufacturer’s protocol. At that time, 30 μg proteins were run on a 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel at 100 V for 2 h and transferred to a polyvinylidene fluoride membrane at 80 V for 2 h. Membranes were blocked with 5% bovine serum albumin (BSA) in Tris-buffered saline with Tween-20 (TBST) at room temperature for 1 h. They were then incubated overnight at 4 °C with one of the following primary antibodies: anti-CDK6 (ab124821, Abcam) (1:1000), anti-GLUT1 (ab115730, Abcam) (1:1000) and anti-HK2 (2867T, Cell Signaling Technology) (1:1000) and anti-beta actin (1:1000) from Santa Cruz Biotechnology (Dallas, Texas, USA). After washing with TBST, membranes were incubated with secondary antibodies for 1 h at room temperature and signals were developed using an enhanced chemiluminescence kit (Pierce, Waltham, MA, USA).

SPSS (version 22.0, SPSS Inc.) or GraphPad Prism software (version 7.0, USA) were used to analyze the data and generate the graphs. The differences between the miR-500a-3p expression and the clinical characteristics of CRC patients were analyzed using χ² test. Survival curves were generated using the Kaplan–Meier method and log-rank tests. Univariate and multivariate Cox regression analyses were conducted to identify the independent factors. The correlation among the expression of miR-500a-3p, CDK6 and HK2 was analyzed using Spearman’s correlation test. Student’s t-test or the Mann–Whitney U test was used to calculate the P values between two groups. A value of P < 0.05 was identified as statistically significant. R software 3.5.1 was applied to this study for the statistical analyses. All miRNA expression data were normalized.