Membrane-associated RING-CH protein (MARCH8) is a novel glycolysis repressor targeted by miR-32 in colorectal cancer

We collected 85 tumours and paired adjacent tissue from 85 CRC patients who were undergoing treatment at Shanghai Changzheng Hospital, Shanghai, China from January 2007 to December 2008. To avoid interference by other treatment methods, we restricted our study to include patients who only underwent surgical resection and excluded those treated with other methods. Clinicopathological information, including the patient's sex, age, tumour size, and tumour stage, was summarized from the medical records of the patients (Additional file 1: Table S1). All the protocols used in this study followed the Declaration of Helsinki and were approved by the Ethical Review Board of Shanghai Changzheng Hospital. Twenty-five tumors with paired normal tissue of the 85 patients were immediately frozen in liquid nitrogen and stored at − 80 °C until subsequent analysis. We also obtained paraffin-embedded tumor tissues from all 85 patients.

The expression level, survival prognosis, and clinical characteristics information of MARCH1-11 proteins was downloaded through public databases TCGA (https://​www.​cancer.​gov/​about-nci/​organization/​ccg/​research/​structural-genomics/​tcga) [19], and GEO public database (https://​www.​ncbi.​nlm.​nih.​gov/​). The TCGA-COAD included 571 samples and READ-TCGA included 192 samples; GSE39582 contained 443 colon cancer and 19 non-tumoral colorectal mucosas. GSE41258 contained 390 expression arrays from primary colon adenocarcinomas, adenomas, metastasis, and corresponding normal mucosae. GSE12945 included microarray data of 62 CRC patients. GSE17537 and GSE17536 included 55 colorectal cancer patients from Vanderbilt Medical Center (VMC) were used as the training dataset and 177 patients from the Moffitt Cancer Center were used as the independent dataset. GSE14333 contained samples taken from colorectal cancers in surgically resected specimens in 290 colorectal cancer patients. The expression level Boxplot was analysed by ggplot2 R package, the forest R package was used for forest diagram, and survminer R package was carried for survival analysis.

Paraffin-embedded tumour tissue sections from the 85 patients were subjected to IHC analysis. We counted the number of MARCH8-positive cells in five random visual fields under a 40 × microscope, and then calculated an average value of 20% to be the cut-off value. All 85 patients were divided into two groups based on the percentage of MARCH8-positive cells. First, the sections were deparaffinized with xylene and rehydrated in graded alcohol, i.e., 100, 95, 80, and 70% ethanol. After rehydration, we immersed the sections in a 0.01 M citrate buffer (pH 6.0) and heated the samples in a microwave oven for 15 min at 850 W to retrieve the antigen. We then incubated the sections for 30 min in 0.3% hydrogen peroxide and then in a 10% heat-inactivated goat serum. After an incubation with anti-MARCH8 antibody (ab109690, Abcam, USA) at 4 °C overnight, the sections were probed with a horseradish peroxidase-conjugated secondary antibody for one hour at 25 °C. Finally, MARCH8-expressing cells in the tumour sections were visualized with a 3,3-diaminobenzidine solution, and the sections were counterstained for 3 min with haematoxylin.

HEK293T, HIEC-6, CACO2, HT29, HCT116, LOVO, RKO, and SW480 cells were acquired from the Shanghai Cell Bank, Chinese Academy of Sciences, China. The cell lines mentioned above were maintained in Dulbecco's modified Eagle’s medium (DMEM, HyClone, USA) supplemented with 10% foetal bovine serum (Gibco, USA), 100 units/mL penicillin, and 100 μg/mL streptomycin in a 37 °C humidified cell culture incubator with 5% CO₂.

The MARCH8 coding sequence was synthesized by Shanghai Sangon Biotech Co. Ltd. We cloned the MARCH8 coding sequence into the pLVX-puro vector (Clontech, USA) to construct a lentiviral MARCH8-overexpressing plasmid to overexpress MARCH8 in CRC cells. We synthesized three short hairpin RNAs (shRNAs) targeting human MARCH8 (sequences shown in Additional file 1: Table S2) and cloned them into the pLKO.1 vector (Addgene, USA) to silence endogenous MARCH8 in CRC cells. We also synthesized a negative control shNC and cloned it into the pLKO.1 vector. For the lentivirus packaging, each constructed plasmid was transfected into HEK293T cells and packaging plasmids psPAX2 and pMD2.G using the Lipofectamine 2000 cell transfection kit (Invitrogen, Carlsbad, USA).

We extracted total RNA using TRIzol reagent (Invitrogen, Carlsbad, USA) following the manufacturer’s instructions. The isolated total RNA was then subjected to first-strand cDNA synthesis using the Reverse Transcription Kit (Thermo Fisher, USA). The SYBR Green Mix Kit (Thermo Fisher, USA) was used to run real-time PCR on an ABI 7500 Real-time Platform (Applied Biosystems, USA). Relative mRNA expression was calculated based on the 2^−ΔΔCt method. Human GAPDH was used as the internal reference gene. The primers are shown in Additional file 1: Table S3.

Tissue samples (approximately 0.2 g) were powdered in liquid nitrogen and then resuspended in a RIPA buffer supplemented with protease inhibitor cocktail (Sigma, USA). Cell lysates were prepared by resuspending cell pellets in RIPA buffer supplemented with protease inhibitor cocktail. After blocking with 5% skim milk, the membranes were probed with MARCH8 and metabolism gene primary antibodies at 4 °C overnight. After washing away unbound antibodies with tris-buffered saline and 0.1% Tween® 20 (TBST), the membranes were incubated at 25 °C for one hour with HRP-conjugated rabbit secondary antibody (Beyotime, China). Finally, an enhanced chemiluminescence (ECL) system (Millipore, USA) was utilized to visualize the protein bands.

SW480 and CACO2 cells were lysed with a mixed buffer that contained RIPA, NaF, and PMSF (Beyotime, China) on ice for 30 min. Then the collected protein was denatured in a 95 °C water bath for 10 min and centrifuged at 12,000 rpm at 4 °C for 10 min. The upper clear cell lysates were transferred to new tubes. Equal amounts of protein were loaded on gels and separated by SDS–PAGE. Then, proteins were transferred to PVDF membranes (Immobilon-P; Millipore, Billerica, Massachusetts) and blocked with 5% milk or bovine serum albumin, followed by incubation with primary and secondary antibodies. Primary antibodies were anti-CDK4 (66,950-1-Ig, Proteintech, 1:1000), cycling D1 (60186-1-Ig, Proteintech, 1:20000), cleaved caspase 3 (19677-1-AP, Proteintech, 1:500) and β-actin (66009-1-Ig, Proteintech, 1:10000) antibodies.

For Co-IP analysis, cell lysates were incubated with anti-MARCH8 antibody (Abcam, USA), anti-HK2 antibody (Abcam, USA), or IgG (Abcam, USA) for one hour at 4 °C. After that, protein A/G-agarose beads were added to the lysates, and the mixture was incubated for three hours at 4 °C to pull down the antigen–antibody complexes. Finally, the precipitates were eluted with elution buffer and detected by Western blot analysis.

We utilized the Cell Counting Kit-8 (CCK-8, Dojindo Laboratories, Japan) to evaluate the proliferation of CRC cells. Briefly, 2 × 10³ cells were seeded into each well of 96-well plates. Then, 10 μL CCK-8 reagent was added to each well, and the mixtures were incubated at 37 °C for an additional hour. The number of viable cells was then calculated based on the optical absorbance values measured at 450 nm using the Multiskan MS Plate Reader (Labsystems, Finland).

We measured the OCR and ECAR using a Seahorse XF24 Extracellular Flux Analyser (Seahorse Bioscience, USA). Briefly, 1 × 10⁵ CRC cells were seeded in each well of XF24 Cell Culture Microplates (Seahorse Bioscience, USA). One hour before the assay, we replaced DMEM with XF Assay Medium (Seahorse Bioscience, USA) (pH 7.4). For OCR measurements, 2 mM oligomycin, 3 mM carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone (FCCP), and a mixture of 2 mM rotenone and 5 mM antimycin were sequentially added to the reaction system to determine the basal respiration. For ECAR measurements, 5.5 mM glucose, 2 mM oligomycin, and 100 mM 2-deoxyglucose (2-DG) were added sequentially to the reaction system to determine the glycolytic rate, glycolytic reserve, and nonglycolytic ECAR.

All animal experiments in this study were performed under the supervision of the Animal Care Committee of Shanghai Changzheng Hospital. Ten four-week-old male BALB/c nude mice (SLRC Laboratory Animal, China) were randomized into two groups (n = 5 per group) and subcutaneously injected with 5 × 10⁶ oeMAR SW480 cells (the oeMAR group) or vector SW480 cells (the vector group). To evaluate the degree of apoptosis in the tumour tissues, a TUNEL assay kit (Roche, USA) was used. The tumour volume in each mouse was determined every three days for the next 33 days according to the following equation: tumour volume = 1/2 × (the largest diameter) × (the smallest diameter) ². The mice were euthanized after 33 days, and the volumes and wet weights of tumour tissues were determined.

Twenty-four-week-old male nude BALB/c mice were randomized into four groups (n = 5 per group) to investigate whether MARCH8 has a specific inhibitory effect on the functions of HK2 in CRC cells in vivo. The vehicle-shMAR group was subcutaneously injected with 5 × 10⁶ shMAR-1 LOVO cells, and the vehicle-shNC group was subcutaneously injected with 5 × 10⁶ shNC LOVO cells. Fifteen days later, the mice in the above two groups were intraperitoneally injected with DMSO twice a week. The Lonidanmine-siNC group was subcutaneously injected with 5 × 10⁶ shNC LOVO cells and intraperitoneally injected with a lonidamine-DMSO solution at a dose of 100 mg lonidamine/kg body weight (twice a week) fifteen days after LOVO cell injection. The Lonidanmine-siMARCH8 group was injected with 5 × 10⁶ shMAR-1 LOVO cells subcutaneously and intraperitoneally with a lonidamine-DMSO solution at a dose of 100 mg lonidamine/kg body weight (twice a week) fifteen days after LOVO cell injection. The tumour volume in each mouse was then determined every three days using the method described above after the cells were injected.

The amplified MARCH8 3’-UTR-WT and MARCH8 3’-UTR-MUT were cloned into the pGL3 luciferase vector (Promega Corporation). CRC cells were seeded onto 96-well plates over a 24-h period. CRC cells were transfected with MARCH8 3’-UTR-WT or MARCH8 3’-UTR-MUT and miR-32 inhibitor using Lipofectamine 3000. After 48 h, cells were harvested, and the Dual Luciferase Reporter Assay Kit (Promega Corporation) was used to measure luciferase activity.

We used GraphPad Prism software version 6.0 to analyse our data. We used Student’s t test and analysis of variance (ANOVA) to compare data from different groups. Pearson correlation analysis was applied to establish a linear correlation between MARCH8 and HK2. P values less than 0.05 were considered statistically significant.

First, we analysed the expression, survival prognosis, and clinical characteristics of 11 MARCH proteins through public databases TCGA, GEO public databases. As shown in the results, MARCH8 presented lower expression level in CRC tissue compared to normal tissue (Additional file 1: Figs S1, S2). Besides, in general, MARCH8 plays a protective role in CRC (Hazard Ratio < 1). Considering its lower expression level and protective role in CRC, we chosen the MARCH8 for further investigation. Then, immunohistochemistry (IHC) was performed to examine the expression of MARCH8 proteins in tumour tissues. Patients with samples containing more than 20% MARCH8-positive cells (middle panel of Additional file 1: Fig. S3A) were classified into the MARCH8-high group, and those with less than 20% MARCH8-positive cells (right panel in Additional file 1: Fig. S3A) were classified into the MARCH8-low group. We examined MARCH8 mRNA expression in 25 tumours with paired normal tissues stored at − 80 °C. MARCH8 mRNA exhibited significantly decreased expression in the tumour samples, which is consistent with the previous results based on public datasets (Additional file 1: Fig. S3B). Kaplan–Meier analysis indicated that the five-year survival rate of the 34 patients in the MARCH8-high group was significantly higher than that of the 51 patients in the MARCH8-low group (P = 0.007) (Additional file 1: Fig. S3C). Last, chi-square test analysis for MARCH8 protein expression with clinicopathological parameters showed a significant correlation of MARCH8 protein with tumour size and TNM stage (Additional file 1: Table S4). And clinical characteristics of MARCH8 and other MARCH proteins by large samples of public databases TCGA (TCGA- COAD, TCGA-READ), GEO (GSE41258, GSE39582, and GSE12945) were also analysed (Additional file 2: Table S5 and Additional file 3: Table S6).

We first examined the expression of MARCH8 in multiple cell lines (Fig. 1A) to investigate the function of MARCH8 in CRC cells. The expression of MARCH8 in LOVO cells was high, and LOVO cells were used for MARCH8 knockdown (shMAR). MARCH8 overexpression significantly enhanced MARCH8 abundance in CACO2 and SW480 cells, as shown in Fig. 1B. Three short hairpin RNAs (shMARs) targeting MARCH8 were designed to rule out off-target effects in LOVO cells. Of the three, shMAR-1 and shMAR-2 showed better inhibitory effects (Fig. 1B) and were used for the following phenotype assays. As shown in Fig. 1C, compared to that of other cells, CACO2 and SW480 oeMAR cells showed significantly decreased proliferation, and LOVO shMAR cells exhibited increased cell numbers, suggesting that MARCH8 may inhibit cell proliferation in CRC. Consistently, CACO2 and SW480 oeMAR cells showed increased apoptosis rates, and LOVO shMAR cells demonstrated slightly reduced apoptosis rates (Fig. 1D), implying that MARCH8 may also induce cell apoptosis. Furthermore, as seen in Fig. 1E, the WB results showed that the proliferation related markers CDK4 and cyclin D1 were decreased while the MARCH8 was overexpression. The apoptosis-related proteins caspase3 was significantly increase in the MARCH8 overexpression group, which was consistent with the previous results. Together, these results indicated that the MRCHA8 was a tumor suppressor in CRC in vitro.

We generated xenograft mouse models by subcutaneously injecting SW480 cells into nude BALB/c mice to verify the in vitro effect of MARCH8. As shown in Fig. 1F, G, tumour tissues in the oeMAR group grew significantly slower than those in the vector group. Tumour tissues were then subjected to a TUNEL assay to study the impact of MARCH8 on cell apoptosis. As shown in Fig. 1H–I, the ratio of apoptotic cells in the oeMAR group was significantly higher than that in the control group. Collectively, the in vitro and in vivo results suggested that MARCH8 is a tumour suppressor in CRC.

Studies have shown that E3 ubiquitin ligases regulate glycolysis in CRC [21, 22], but no studies could be found regarding MARCH8 in cancer cell glycolysis. Therefore, the basal cellular respiration and maximal respiration rate were examined with ECAR and OCR assays to explore the role of MARCH8 in glycolysis. Figure 2A–D shows that MARCH8 overexpression inhibited glycolysis and respiration in CACO2 and SW480 cells. In LOVO cells, MARCH8 knockdown with shMAR-1 and shMAR-2 significantly increased the glycolytic rate and basal cellular respiration level (Fig. 2E, F). The phenotype of MARCH8-mediated inhibition of glycolysis prompted us to explore its mechanism in CRC. The expression of canonical glycolysis enzymes [23] was examined in oeMAR cells, including lactate dehydrogenase A (LDHA), glucose transporter 1 (GLUT1), glucose transporter 4 (GLUT4), hexokinase 2 (HK2), uncoupling protein 2 (UCP2), and pyruvate kinase M2 (PKM2). As shown in Fig. 3A, the protein levels of LDHA, GLUT4, HK2, and PKM2 were decreased upon MARCH8 overexpression. Given that MARCH8 mediates substrate protein ubiquitination and degradation [14, 24], we next investigated the binding of MARCH8 with LDHA, GLUT4, HK2, and PKM2. Immunoprecipitation experiments showed that MARCH8 only interacted with HK2 (Fig. 3B). Therefore, we sought to investigate how MARCH8 regulates the HK2 protein. MARCH8 overexpression resulted in decreased HK2 protein levels without affecting the transcription of HK2 mRNA in CACO2 and SW480 cell lines (Fig. 3C, D). Consistent with these results, MARCH8 knockdown led to elevated HK2 protein levels without affecting the transcription of HK2 in LOVO cells (Fig. 3C, D). Interestingly, treating oeMAR SW480 cells with MG132, a potent proteasome inhibitor [25], offset the effect of MARCH8 overexpression on the HK2 protein (Fig. 3E), indicating that MARCH8 regulates the expression of HK2 in a proteasome-dependent manner. Therefore, we investigated the effect of MARCH8 on HK2 ubiquitination levels in SW480 cells. As shown in Fig. 3F, we indeed observed a higher HK2 ubiquitination level in oeMAR SW480 cells than in other cells. Finally, Co-IP was conducted to validate the MARCH8-HK2 binding complex (Fig. 3G). Together, these results demonstrate that MARCH8 promoted ubiquitination-mediated proteasome degradation to reduce HK2 protein expression.

ShMAR-1 LOVO and shNC LOVO cells were treated with the HK2 inhibitor lonidamine to determine whether HK2 mediated the functions of MARCH8 in CRC. As shown in Fig. 4A–C, the effects of MARCH8 knockdown on LOVO cell proliferation, glycolysis, and respiration were blocked by lonidamine treatment. Next, LOVO xenograft mouse models were constructed and treated with vehicle or lonidamine. Tumour tissues in the shMAR-1 + vehicle mice grew significantly faster than those in the shNC + vehicle mice. The growth rates of shMAR-1 + LND tumour tissues were significantly lower than those of shMAR-1 + vehicle tumour tissues (Fig. 4D–F).

Next, the correlation of MARCH8 with HK2 protein in CRC patients was analysed. We selected 8 patients and determined the protein levels of MARCH8 and HK2. CRC tumour samples had significantly lower MARCH8 levels (P < 0.0001) and significantly higher HK2 levels (P < 0.05) than those in normal colon tissues (Fig. Additional file 1: S4A, B). Pearson correlation analysis revealed that there is a significant negative correlation between MARCH8 and HK2 in CRC tumour tissues (R = -− 5672, P < 0.05) (Additional file 1: Fig. S4C). These results demonstrate that the protein level of MARCH8 is negatively correlated with that of HK2 in human CRC tumour tissues.

Finally, the molecular mechanism underlying the decreased expression of MARCH8 in CRC was investigated. First, we investigated the histone activation marks of MARCH8 in CRC cells. As shown in Fig. 5A, weak or no H3K27ac peaks were found in the promoter regions of HCT116, LOVO, SW480, and RKO cells, suggesting that chromatin may be in a poised state and that the transcription process is controlled. We then investigated the role of miRNAs, which are posttranscriptional regulators, in the regulation of MARCH8. Three online tools (MiRwalk, TargetScan, and miRDB) were used to predict the binding of miRNA to sequences in MARCH8 3’UTR regions, and all 22 common miRNAs were acquired as a result (Fig. 5B). The expression of these miRNAs and MARCH8 in CRC samples was then analysed, and two miRNAs, miR-199b and miR-32, showed a significant negative correlation with MARCH8 (Fig. 5C). Because miR-32 acts as an oncogene and unfavourable prognostic marker in CRC [26‐30] while miR-199b plays the opposite role [31, 32], we chose miR-32 for further validation studies. Then, we transfected miR-32 inhibitors into CRC cells, and as shown in Fig. 5D, MARCH8 expression was significantly increased. Finally, a dual luciferase assay was conducted to demonstrate the binding of miR-32 to the MARCH8 3’UTR sequence. The MARCH8 3’UTR wild-type (WT) group showed increased luciferase activity, while no significant changes were observed in the mutant (MUT) group (Fig. 5E). In summary, in CRC, poised chromatin and miR-32 decreases the expression of MARCH8, which induces HK2 degradation, represses glycolysis and acts as a tumour suppressor in CRC (Fig. 5F).