2.1 Clinical specimens and cell lines

The tumour tissues were freshly acquired by surgical resection from 85 patients with CRC at Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China. Normal colorectal mucosae were taken at biopsy from 42 individuals without colorectal malignancies. The specimens were stored at −80°C for further analysis. Each patient provided written consent. The study was approved by the Ethics Committees of Beijing Chao-Yang Hospital, Capital Medical University, which followed the recommendations of the Declaration of Helsinki for biomedical research involving human subjects.

Human colon cancer cell lines HCT116 and SW480 were obtained from the American Type Culture Collection (ATCC, USA) and cultured in Dulbecco's modified Eagle's medium (Sigma, USA) supplemented with 10% foetal bovine serum (Gibco, USA), 100 U/mL penicillin, and 100 μg/mL streptomycin. Cells were cultured in a humidified incubator with 5% CO₂ at 37°C.

2.2 RNA extraction and real-time RT-PCR

Total RNA was isolated from frozen CRC tissues and cell lines using Trizol reagent (Invitrogen, Life Technologies, Carlsbad, CA), according to the manufacturer's instruction. The concentration and purity of RNA were measured using a NanoDrop ND-2000 Spectrophotometer (Thermo Scientific Wilmington, DE, USA). 2 μg of RNA was reverse-transcribed into cDNA using a reverse transcription system (TransGen Biotech, China). The Real-time PCR analysis was conducted on the 7500 Sequence Detection System (Applied Biosystems, China) using SYBR Premix according to the manufacturer's protocol with GPADH as an internal control. The thermal cycling conditions were as follows: 2 seconds at 95°C, followed by 40 cycles of 95°C for 5 seconds and 60°C for 34 seconds. Relative fold changes were normalized with GAPDH and calculated using 2^-ΔΔCT methods. For Cosmc mRNA detection, the forward primer was 5′-ACTGCAGCCCAAAGACTCACATCT-3′, and the reverse primer was 5′-ATGCACCACCATGAGCATCATCAC-3′. As for GAPDH detection, the forward and reverse primers were 5′-AATCCCATCACCATCTTCCA-3′ and 5′-TGGACTCCACGACGTACTCA-3′. The sequences of the primers for ppGalNAc-Ts were shown in the supplementary data.

2.3 Western blot analysis

Frozen tissues or cells were lysed using RIPA lysis buffer (EnoGene, China) supplemented with 1 mM protease inhibitor cocktail (EnoGene, China) and 1 mM phenylmethylsulfonyl fluoride (PMSF, EnoGene, China). After centrifugation at 12 000× g for 20 minutes at 4°C, the supernatant was collected. The concentration of total proteins was measured with the BCA Protein Assay Kit (Thermo Fisher, USA). Samples containing an equal amount of denatured proteins were separated on a 10% SDS polyacrylamide gel and transferred onto a PVDF membrane (Millipore, USA). The membrane was blocked with 5% non-fat milk for 1 hour at room temperature and probed with primary antibodies and HRP-conjugated secondary antibodies. Signals were detected using a chemiluminescent HRP substrate (Millipore, USA) on the Bio-Rad imaging system. The following primary antibodies were used: Cosmc (1:200, Santa Cruz), T-synthase (1:200, Santa Cruz), GRP78 (1:1000, Cell Signaling Technology), CHOP (1:1000, Cell Signaling Technology), ZO-1 (Zonula-occludens-1) (1:1000, Cell Signaling Technology), Vimentin (1:1000, Cell Signaling Technology), Slug (1:1000, Cell Signaling Technology), Snail (1:1000, Cell Signaling Technology) and GAPDH (1:1000, Cell Signaling Technology).

2.4 Lentivirus-mediated Cosmc transfection

The Cosmc overexpression plasmid was constructed by cloning Cosmc CDS (F:5′-GAGGATCCCCGGGTACCGGTCGCCACCATGCTTTCTGAAAGCAGCTCC-3′; R:5′-CACACATTCCACAGGCTAGTCAGTCATTGTCAGAACCATTTG-3′) into the GV367-EGFP lentiviral vector containing restriction sites for AgeI/ NheI, which was purchased from Shanghai Genechem Co. Ltd (China). The empty vector was used as the control. The Cosmc expression vector and the empty vector were transfected into HCT116 and SW480 cells with polybrene respectively, according to the manufacturer's instruction. The transfected cells were cultured for 72 hours and then selected with complete medium containing 2 μg/mL puromycin (Gibco, USA) for 7 days.

2.5 Transwell migration and invasion assays

After being starved for 24 hours, 2 × 10⁵ cells with the serum-free medium were seeded into the upper chamber (8-μm pore size, BD Bioscience, USA) pre-coated with or without Matrigel (BD Bioscience, USA). Complete medium containing 10% FBS was added to the bottom of each well as a chemoattractant. After being cultured for 24 hours, non-migrated or non-invaded cells were removed off the upper chamber using a cotton swab while the migratory or invasive cells were counted after fixation with 4% paraformaldehyde and being stained with 0.1% crystal violet.

2.6 Flow cytometry analysis

To analyse the expression of Tn antigen in Cosmc-overexpressing cells and control cells, flow cytometry analysis was performed. Briefly, 1 × 10⁵ cells were routinely trypsinized, suspended in cold PBS and then incubated with 10 μg/mL mouse anti-Tn IgM mAb or mouse IgM isotype-antibody as a control (Santa Cruz, USA) for 2 hours at 4°C.4 After being washed three times with PBS, cells were incubated with PE-labelled goat antimouse IgM secondary antibody (BD, USA) for 1 hour at 4°C. Cells were washed three times and suspended in 500 μL cold PBS per tube, followed by further analysis on a Flow Cytometer (Canto II, BD Bioscience, USA).

2.7 Statistical analysis

All data were analysed with GraphPad Prism 6.0 (GraphPad Software, La Jolla, CA, USA). All experiments were repeated at least in triplicate, and the data were presented as mean ± standard deviation (SD). Differences were analysed by Student's t test (unpaired, 2-tailed), and P < .05 was considered statistically significant.

3.1 Cosmc expression was up-regulated in human CRC tissues

Cosmc is an essential molecular chaperone for T-synthase and dysfunction of Cosmc results in abnormal exposure of Tn antigen, which is recognized as a hallmark for aberrant O-glycosylation.8, 9 However, the precise role of Cosmc in clinical CRC remains largely unexplored. To determine whether there are aberrant alterations in Cosmc expression in clinical CRC, we examined its mRNA and protein levels in archived frozen human colorectal tissues (42 normal colorectal tissues, 85 CRC tissues). Real-time RT-PCR and Western blot analysis showed that Cosmc mRNA and protein levels were both elevated in human CRC tissues when compared with normal colorectal tissues (both P < .01; Figure 1A,B). This study cannot assess the proper localization of Cosmc, because the anti-Cosmc antibody is not suitable for immunohistochemistry. Besides, we analysed RNA-seq data of TCGA colon adenocarcinoma (COAD) data set including 275 tumour and 41 adjacent normal samples. It exhibited significantly higher Cosmc mRNA levels in CRC patients in comparison with normal controls (P < .05) (Figure 1C), which is consistent with our observations.

3.2 ER stress contributed to elevated Cosmc expression in human CRC

Although loss of functional Cosmc caused by mutation, deletion, or hypermethylation has been widely recognized as a prevailing mechanism underlying aberrant O-glycosylation observed in several types of cancers including CRC,7, 9-14 little is known why there is an elevated expression of Cosmc in CRC, which appears to contradict the data obtained from some tumour-derived cell lines and certain types of malignancies such as pancreatic cancer.18 Here we sought to investigate the molecular mechanisms underlying elevated Cosmc expression in CRC tissues. It has been proposed that expression of Cosmc may be induced by ER stress, which occurs frequently in malignancies.17 Moreover, Tn antigen has been discovered to be abundantly expressed in human CRC, which may promote Cosmc production via a feedback loop.17 To test these possibilities, we examined the expression levels of ER stress markers such as GRP78 and CHOP in another set of frozen CRC tissues (16 normal colorectal tissues, 32 CRC tissues). Western blot results showed that both GRP78 and CHOP were significantly up-regulated in CRC tissues as compared with normal controls, indicative of the occurrence of ER stress, which also correlated with the expression of Cosmc (Figure 2A). Notably, there was a strong correlation between expression of Cosmc and ER stress in human CRC tissues (Figure 2B).

To definitively address this possibility, we treated two types of CRC cell lines (HCT116, SW480) with tunicamycin (TM, Sigma, MO, USA) to induce ER stress. At 24 hours, administration of TM (40 μg/mL) led to the occurrence of ER stress in both cell lines, as evidenced by a remarkable up-regulation of GRP78 and CHOP protein expression (Figure 2C). It is noticeable that Cosmc was also significantly increased at its protein levels following ER stress, thereby supporting the likelihood that ER stress may induce Cosmc expression in cancer cells.

In addition, it has been confirmed that Cosmc dysfunction induces Tn antigen expression,8, 9 but it is uncertain whether the increased Tn antigen in turn promotes Cosmc production via a feedback loop. To address this question, we used Tn-expressing CRC cell lines (HCT116, SW480), which were previously generated through forced deletion of the key glycosyltransferase for O-glycosylation T-synthase,19, 20 to assess whether high levels of Tn antigen have an impact on Cosmc expression. T-synthase deficiency is sufficient to induce Tn antigen expression in cells (data not shown). In both cell lines, we observed no appreciable differences in the levels of Cosmc expression between Tn-positive and Tn-negative cells, suggesting that Tn antigen may have little influence on expression of Cosmc (Figure 2D). There may exist other possibilities contributing to Cosmc up-regulation in CRC tissues, which deserves more future work.

3.3 The effects of Cosmc overexpression on CRC cell behaviours

We next studied whether the elevated Cosmc expression influences CRC progression. Two types of human colon cancer cell lines HCT116 and SW480 were chosen to stably transfect with GV367-Cosmc-EGFP (containing Cosmc sequences) and GV367-Control-EGFP lentiviral vectors, respectively. The Cosmc overexpression in both cell lines was confirmed at both mRNA and protein levels. RT-PCR and Western blot analysis showed that compared with the control cells, the mRNA and protein levels of Cosmc in HCT116 and SW480 cells were significantly higher than the controls (Figure 3A,B). We then performed transwell assays to examine cell migratory and invasive abilities in both cell lines. As shown in Figure 3C-F, both Cosmc-overexpressing cell lines exhibited a significant increase in their migratory and invasive abilities in comparison with their respective controls. These results provided evidence that Cosmc overexpression can promote malignant properties of colon cancer cells in vitro.

3.4 Cosmc overexpression-mediated oncogenic alterations were not attributed to aberrant O-glycosylation

Most studies have confirmed that aberrant O-glycosylation mediated by the loss of functional Cosmc contributes to tumour progression and metastasis in a broad range of human cancers including CRC.4, 9, 18 Here we questioned whether the tumour-promoting effects of Cosmc overexpression were also due to aberrant O-glycosylation. To address this issue, we examined the expression of Tn antigen, T-synthase, and ppGalNac-Ts, which are essential components during the process of O-glycosylation, in CRC cells. Alterations in Tn antigen, T-synthase and/or ppGalNac-Ts are indicative of aberrant O-glycosylation.21-24 We first detected Tn antigen expression in both Cosmc-overexpressing cells via flow cytometry. Jurkat cells that harbour mutations in Cosmc, thus expressing abundant Tn antigen, were included as a positive control. The results showed no staining of Tn antigen in both cells transfected with Cosmc and empty vector, respectively. Likewise, Western blot demonstrated that both Cosmc-overexpressing cell lines had comparable levels of T-synthase to the control cells, suggesting that Cosmc overexpression does not necessarily affect T-synthase expression. In addition, ppGalNAc-Ts catalyse the initial O-glycosylation step, so the changes in the ppGalNAc-Ts expression may represent alterations in the process of O-glycosylation.25, 26 Here we used RT-PCR to determine ppGalNac-Ts' mRNA levels in Cosmc-overexpressing cells. We selected 5 types of ppGalNac-Ts (GALNT2, GALNT3, GALNT6, GALNT12, GALNT14), as these ppGalNac-Ts have been reported to be closely associated with CRC.27-31 Our analysis did not detect obvious differences in the mRNA levels of these ppGalNAc-Ts between Cosmc-overexpressing cells and the control cells (Figure 4C). Collectively, these results suggested that aberrant O-glycosylation is unlikely the cause for Cosmc overexpression-mediated oncogenic alterations.

3.5 Cosmc overexpression activated the EMT signalling pathway

We further explored how Cosmc overexpression alters CRC cellular behaviours. The EMT process is an essential candidate signalling pathway associated with cancer progression and metastasis.32 We thereby analysed whether Cosmc overexpression can influence the EMT process. As shown in Figure 5, Cosmc overexpression induced typical characteristics of EMT in both cell lines, as evidenced by a decrease in the expression level of ZO-1, a classical epithelial marker, along with an increase in the expression levels of several mesenchymal markers including Vimentin, Slug and Snail. These data supported that Cosmc overexpression can activate EMT, which may be a major mechanism responsible for enhanced malignancies in colon cancer cells.