MicroRNA-26a regulates glucose metabolism by direct targeting PDHX in colorectal cancer cells

The CRC cell lines including SW620, SW480 and HCT116 were ordered from American Type Culture Collection. The cell line NCM460, which is derived from normal human colon mucosal epithelium, was ordered from the INCELL Corporation, LLC, USA (http://​www.​incell.​com). The cells were all cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum at 37°C in humidified atmosphere with 5% (v/v) CO₂.

The alignment of the miR-26a seed region and 3′-untranslated regions (UTR) of PDHX mRNA of different species were analyzed as literatures described previously [6, 18]. The possible targets of miR-26a were screened by bioinformatics tools, Pictar (http://​pictar.​mdc-berlin.​de) and TargetScan (http://​www.​targetscan.​org).

A 461-bp nucleotide fragments in pre-miR-26a sequences (Additional file 1: Table S1) were cloned into the pENTR-CMV-EGFP vector (ordered from Wuhan Cell marker Biotechnology Co., Ltd., Wuhan, P. R. China) between XhoI and EcoRI to construct miR-26a expression plasmid, pENTR-miR-26a. The empty vector pENTR-MIRNA was used as a control in the ectopic overexpression of miR-26a.

The 3′-untranslated region (3′UTR) of PDHX mRNA (Additional file 2: Table S2) was amplified by RT-PCR. The cDNA fragment corresponding to the 3′UTR of PDHX mRNA was cloned in the downstream of the Renilla luciferase gene in the psiCHECK-2 vector (Cat. # C8021, Promega, USA), which contains a reporter gene Renilla luciferase and an intraplasmid transfection normalization gene, a firefly luciferase. The 3′UTR of PDHX mRNA contains eight nucleotides (5′…UACUUGAA…3′), which are corresponding to miR-26a seed sequences (3′…AUGAACUU 5′) (Figure 1A(I)). In the wild type recombinant plasmid pwt-PDHX, the relevant eight nucleotides (…TACTTGAA…) were involved (Figure 1A(II)). Meanwhile, in the mutant recombinant plasmid pmt-PDHX (Figure 1A(III)), the eight nucleotides were mutated into a random nucleotide sequence (…TCACCAAT…).

The miR-26a inhibitor was commercially available chemically modified antisense oligonucleotide inhibitor (Product ID: MH10249, Cat. #4464084, Life Technologies Corporation, USA). The non-targeting oligonucleotides (Cat. #4464076, Life Technologies Corporation, USA), as a negative control, were also ordered from this company.

The miR-26a inhibitor or the negative control was respectively transfected at a working concentration of 30 nM using Lipofectamine 2000 reagent (Cat. #11668-019, Invitrogen). The ratio of Lipofectamine 2000 reagent (μL) versus the oligonucleotides (pmol) was 18:1.

Total RNA was extracted using Trizol reagent (Cat. #15596-026, Invitrogen). The total RNA was dissolved with RNase-free water for use. The first-strand cDNA for miR-26a was generated using a cDNA synthesis kit (Cat. #K1622, Thermo Scientific). Specific primers were designed for the reverse transcription of miR-26a and U6. The primer sequences for miR-26a were TGTCAACGATACGCTACCTAACGGCATGACAGTGTCAGCCTA. And the primer for U6 was GAACGCTTCACGAATTTGC based on literature reports [13]. For PDHX, random primers were used for mRNA reverse transcription (Cat. #170-8890, Bio-Rad).

The real-time PCR was performed to measure the relative expression using the Supermix-Bio-Rad kit (Cat. #172-5261, Bio-Rad) following the protocol. The U6 was used as an internal reference to calculate the relative expression level of miRNA, and the GAPDH mRNA level was taken as a comparison base for the target RNA expression. The relative RNA expression was calculated with the comparative CT method, which was normalized to the internal references. The forward RT-PCR primer for PDHX was 5′-GTCCCTCTAAAGCAGCTCAAAA-3′, and the reverse one was 5′-CTCCCTTCAAAAGATCCAACTG-3′. The forward one for miR-26a was 5′-CTGTCAACGATACGCTAC-3′, and its reverse was 5′-GTAATCCAGGATAGGCTG-3′. In addition, the forward RT-PCR primer for U6 was 5′-CTTCGGCAGCACATATAC-3′, meanwhile its reverse primer was 5′-GAACGCTTCACGAATTTGC-3′.

The miR-26a expression plasmid pENTR-miR-26a or the empty vector pENTR-MIRNA was respectively transfected at a working concentration of 100 nM using Lipofectamine 2000 reagent (Cat. #11668-019, Invitrogen). The ratio of DNA (μg) versus the Lipofectamine 2000 reagent (μL) was 1: 2.

For the promoter luciferase reporter assay, the plasmid pwt-PDHX or pmt-PDHX was co-transfected with the miR-26a expression vector pENTR-miR-26a or the empty vector pENTR-MIRNA into HEK293T cells. As mentioned above, Lipofectamine 2000 was used as the transfection reagent. The plasmid was used at a working concentration of 100 nM as well. The ratio of plasmid DNA (μg) versus the Lipofectamine 2000 reagent (μL) was 1: 2.

The protein PDHX was detected against its specific antibody by western blot. The PVDF membranes were respectively incubated with the primary antibody of PDHX (diluted 1:1000, Cat. # S0394, Epitomics, USA) at 4°C overnight, followed by incubating with a secondary HRP-conjugated antibody at 37°C for 1 h. Signal detection was performed with Luminata Crescendo Western HRP Substrate (Cat. # WBLUR0100, Millipore). The detection of GAPDH using its antibody (Cat. # sc-365062, Santa Cruz Biotechnology) was taken as a control.

Either the plasmid pwt-PDHX or pmt-PDHX was respectively co-transfected with the miR-26a expression vector pENTR-miR-26a into HEK293T cells, and cells were harvested for assessment of luciferase activity at 48 hours after transfection. The luciferase activities of cellular extracts were measured with a Dual Luciferase Reporter Assay System (Cat. #E1910, Promega). Both Renilla luciferase and firefly luciferase activities were measured. The luciferase signal was normalized to the firefly luciferase signal as described previously [19].

Either the pENTR-miR-26a or miR-26a inhibitor was transfected into CRC cells. Cell culture media were collected after transfection for 48 h. Glucose uptake and lactate production were measured using Amplex^® Red Glucose/Glucose Oxidase Assay Kit (Cat. #A22189; Invitrogen) and lactate assay kit (Cat. #MAK064; Sigma-Aldrich) respectively. The results were normalized on the basis of total cellular protein amounts.

The concentration of pyruvate in CRC cells, transfected with pENTR-miR-26a or miR-26a inhibitor, was respectively measured using pyruvate assay kit (Cat. #K609-100; BioVision). Briefly, cells were collected after transfection for 48 h and dissolved with 0.5 ml of pyruvate assay buffer. And 50 μl sample was added with 50 μl of reaction mixture to incubate at room temperature for 30 minutes. A standard curve covering a range of 10–0.1 nmol per well was used as control. Absorbance was measured at 570 nm. The pyruvate concentration, which was normalized on the basis of totally cellular protein amounts, was calculated relative to the standard curve.

To analyze the production of acetyl coenzyme A (acetyl-CoA), cell extracts were prepared using the perchloric acid approach as described previously, with minor modification [20]. CRC cells, treated with pENTR-miR-26a or miR-26a inhibitor, were harvested and washed with phosphate-buffered saline. Cells were mixed in 1 mL of washing buffer (10 mM sodium phosphate [pH 7.5], 10 mM MgCl₂, 1 mM EDTA), treated with 200 μL of 3 M ice-cold HClO₄, and incubated on ice for 30 minutes. The mixture was centrifuged for 5 minutes at 10,000 × g at 4°C. The supernatant was neutralized with saturated KHCO₃ and centrifuged as described above. The level of acetyl-CoA in the cell extraction was quantified using the acetyl-CoA assay kit (Cat. # K317-100; BioVision). Fluorescence was measured (Ex/Em = 535/589 nm), and the acetyl-CoA concentration was calculated based on the standard curve. The production of acetyl-CoA was normalized based on the total of cellular proteins.

By analyzing the 3′UTR mRNA of several glycolytic enzymes, we found that the 3′UTR of PDHX is relatively long with around 797 nucleotides, suggesting that PDHX mRNA is a possible target for miRNAs. In order to predict the possible target gene of miR-26a, the miRNA binding sites in the 3′UTR of PDHX were further analyzed through the Pictar and TargetScan software [6, 18]. A conserved miR-26a binding site exists in the 3′UTR of PDHX across Homo sapiens (Hsa), Pan troglodytes (Ptr) and Mus musculus (Mmu) (Figure 2A), further we predicted an exact match to the seed region of mature miR-26a (Figure 2B). The bioinformatics analysis indicated a potential functional link between PDHX and miR-26a.

The expression level of miR-26a in NCM460, HCT116, SW480 and SW620 cells was examined using q-PCR analysis (Figure 2C). Compared with the normal epithelial cells NCM460, the expression of miR-26a was significantly increased to 1.3, 1.7 and 4.6-fold for the colon cancer cell lines HCT116, SW480 and SW620 respectively (p < 0.05). Generally, the expression level of miR-26a was significantly higher in malignant CRC cell lines HCT116, SW480 and SW620 than in a normal colorectal mucosal epithelial cell line NCM460 [21], especially it was almost increased over 4-fold in SW620 cells, which has high lymph node metastatic potentials [22]. Among the three CRC cell lines, the expression levels of miR-26a were gradually increased from HCT116 to SW480 and SW620 cells. Compared with HCT116 cells, which is derived from a human colorectal adenocarcinoma [22], SW480 and SW620 cells are typical model systems for CRC metastasis. The SW480 is derived from the primary site of CRC and SW620 comes from the recurrent lymph node metastasis [23]. Therefore, it has a positive correlation between the expression levels of miR-26a and the malignant degree of CRC cells.As miRNA expression is often inversely correlated with those of their specific target mRNAs, we further analyzed the protein expression of the miR-26a target —PDHX, in CRC cell lines. We found that the expression of PDHX was stepwise decreased in NCM460, HCT116, SW480 and SW620 cells (Figure 2D). The expression level of miR-26a in NCM460 was the lowest among the four cell lines (Figure 2C), but the protein level of PDHX exhibited the highest (Figure 2D). While in SW620 cells, the highest level of miR-26a expression and the lowest level of PDHX expression were observed. By comparing the expression levels of miR-26a and PDHX, an inverse correlation between them was observed in CRC cell lines, which supported the bioinformatics prediction on miR-26a target as above.

Among these CRC cell lines, HCT116 cells were chosen to perform the following studies since it showed the media expression of miR-26a. Firstly, the miR-26a expression plasmid pENTR-miR-26a was transfected into HCT116 cells to detect its influence on the expression level of PDHX. As a result, a marked reduction of PDHX was observed at both mRNA and protein level when miR-26a was overexpressed with almost 2.5-fold (Figure 3A). Compared to the empty vector pENTR-MIRNA, the expression of PDHX mRNA reduced almost 33% (P < 0.001) in HCT116 cells transfected with pENTR-miR-26a (Figure 3B), and PDHX protein was decreased by almost 40% (Figure 3C).Furthermore, the suppression of miR-26a expression was performed in HCT116 cells treated with 30 nM of miR-26a inhibitor, and the random oligonucleotides were taken as negative controls. Compared with the negative control, the mRNA expression of PDHX had a 1.3-fold upregulation (Figure 3E) in HCT116 cells, along with about 70% inhibition of miR-26a (Figure 3D), and the protein expression of PDHX was increased 1.5-fold in the treated HCT116 cells (Figure 3F).We also performed the loss-of function studies in SW620 cells, which have a highly endogenous expression level of miR-26a and a low level of PDHX (Figure 2C-D). Under the condition of a great down-regulation of miR-26a in SW620 cells (Figure 4A, p < 0.01), the mRNA expression of PDHX was significantly increased with 2-fold (Figure 4B, p < 0.01), while the protein level of PDHX was improved to 1.56-fold (Figure 4C). The down-regulation of endogenous miR-26a in SW620 cells could restore PHDX expression, which is similar to the effects with a knock-down of miR-26a in HCT116 cells (Figure 3).

Except to the gain and loss-of function studies, the direct regulatory effect of miR-26a on PDHX expression was also verified by Renilla luciferase reporter gene assays. Either the plasmid pwt-PDHX or pmt-PDHX was co-transfected with the miR-26a-expressing pENTR-miR-26a into HEK293T cells respectively. Cells were harvested for assessment of luciferase activity at 48 hours after transfection. Compared with the mutated reporter plasmid pmt-PDHX, when the wild type plasmid pwt-PDHX was co-transfected with the miR-26a-expressing plasmid pENTR-miR-26a, the ectopically expressing miR-26a bound to the 3′ UTR of the wild type plasmid pwt PDHX, then the fused luciferase and the 3′UTR of PDHX was cleaved and subsequently degraded, which resulted in an effective decrease of luciferase activity (Figure 1B). By contrast, in the case of the co-transfection of the mutant plasmid pmt-PDHX with the pENTR-miR-26a, the ectopically expressing miR-26a couldn’t bind to the 3′ UTR of the mutant type plasmid pmt-PDHX, furthermore the luciferase activity did not be decreased. These data showed that miR-26a modulates the PDHX expression by direct targeting the 3′UTR mRNA of PDHX.

The glucose consumption and lactate production were gradually increased among the NCM460, HCT116, SW480 and SW620 cells under physiological conditions (Figure 5). Compared with the normal cell line NCM460, each CRC cell line showed a high level of glucose consumption. The glucose uptake was 30.94, 37.38, 44.42 and 110.99 μmol respectively for NCM460, HCT116, SW480 and SW620 cells. Moreover, there is a positive correlation between the lactate production and the malignant potential of CRC cells. The lactate production was 38.78 μmol, 70.24 μmol, 75.63 μmol and 195.10 μmol from NCM460, HCT116, SW480 to SW620 cells. The lactate content in NCM460 cells was much lower than each of CRC cells. Especially for SW620 cells, it showed more than 5-fold of lactate production than NCM460 cells. In combination with the miR-26a endogenous level in these cells (Figure 2C), it is concluded that the overexpression of miR-26a in CRC cells may cause a preference to glycolysis.In order to investigate miR-26a effects in glucose usage for colon cancer cells, the glucose uptake in HCT116 cells was compared either under the condition of miR-26a overexpression or inhibition. The glucose consumption was increased from 28.86 μmol to 53.58 μmol in miR-26a-overexpressing HCT116 cells by transfecting with pENTR-miR-26a (Figure 6A), showing an almost 2-fold elevation (p < 0.05). Furthermore, we performed a loss-of-function analysis of miR-26a in glucose consumption by using its inhibitor to silence endogenous miR-26a expression. With loss of miR-26a expression, the glucose usage was decreased from 18.82 μmol to 11.25 μmol after being treated with miR-26a inhibitor compared to the control (Figure 6B), indicating a 0.59-fold change (p < 0.005). These evidences strongly demonstrated that the expression level of miR-26a is positively associated with the glucose utilization in colon cancer cells. The overexpression of miR-26a enhanced glucose usage. On the contrary, the inhibition of miR-26a expression would reduce glucose consumption.

Considering the direct target molecule of miR-26a, we further detected whether the effects of miR-26a on glucose metabolism were mediated its target gene PDHX. PDHX, also known as E3 binding protein (E3BP), is a non-catalytic subunit of PDH complex [18, 24] to catalyze the conversion of pyruvate to acetyl-CoA, thereby linking glycolysis to the citric acid cycle [25]. Since PDHX plays a key role in the conversion of pyruvate to acetyl-CoA, we further detected the concentration changes of pyruvate and acetyl-CoA in miR-26a-overexpressing HCT116 cells. As expected, a 1.2-fold increase of pyruvate accumulation (from 8.90 μmol to 10.55 μmol) was detected with the ectopic overexpression of miR-26a (Figure 6C), while the amount of acetyl-CoA (from 474.87 μmol to 410.26 μmol) in HCT116 cells showed a 14% decrease (Figure 6E) (p < 0.05). Besides, the expression inhibition of endogenous miR-26a could induce the decrease of pyruvate content. In detail, the pyruvate content in HCT116 cells was decreased from 1.83 μmol to 0.66 μmol, reducing about 63% (p < 0.05) after being treated with miR-26a inhibitor (Figure 6D). While in the same conditions, the production of acetyl-CoA was effectively increased from 464.87 μmol to 554.90 μmol, raising to almost 1.2 fold (p < 0.05) (Figure 6F).Similarly, we had also performed the loss-of-function studies in SW620 cells to measure the glucose utilization, pyruvate content and acetyl-CoA production. As shown in Figure 7, when the endogenous miR-26a was inhibited, the glucose uptake was decreased from 47.66 μmol to 39.50 μmol and the pyruvate content was decreased from 14.36 μmol to 2.56 μmol, with about 20% (p < 0.05) and 75% (p < 0.01) reduction respectively. While the acetyl-CoA conversion was increased from 200.28 μmol to 252.46 μmol, with almost a 1.3-fold upregulation (p < 0.05). It indicated that the inhibition of miR-26a in SW620 cells could cause similar glucose metabolic effects that were obtained in HCT116 cells.All these results have demonstrated that miR-26a inhibits the conversion of pyruvate to acetyl-CoA though directly targeting the PDHX (Figure 8), which accelerates the glucose consumption to undergo aerobic glycolysis for meeting their increased energy and biosynthesis in colon cancer cells.