Introduction

Intrahepatic cholangiocarcinoma (ICC) is the second most common primary liver malignancy, which arises from bile duct epithelium within the liver [1]. The incidence and subsequent mortality of ICC has been growing around the world [2]. Currently, surgical excision remains the most effective treatment for ICC, but most patients are diagnosed at an advanced stage due to atypical symptoms. After surgery, the 5-year overall survival rate is only 30–35%, and the 5-year survival rate for patients who are unable to undergo surgical resection is less than 10% [3].This fact emphasizes the importance of developing new therapeutic strategies for ICC.

Glycosylation is one of the most important post-translational modifications of proteins and plays a role in a variety of biological activities, including cell communication, adhesion, immunological defense, and cell growth [4]. Glycosylation abnormalities also play a vital role in tumor formation and progression [5]. We recently revealed that partial bisecting glycans were commonly increased in ICC and hepatocellular carcinoma (HCC) tumors by comparing differences in aberrant site-specific N-glycosylation between ICC and HCC tumors using a newly developed method (StrucGP) for N-glycoproteomics combined with stable isotope labeling-based quantification [6, 7]. The bisecting GlcNAc structure, a modification mostly found in complex and hybrid N-glycans, is the attachment of an N-acetylglucosamine (GlcNAc) to the core mannose of the N-glycan via a MGAT3-catalyzed β1,4 linkage [8]. Bisecting GlcNAc has been demonstrated to influence cancer development in the past, although whether it acts as an inhibitor or a promoter is still up for debate. In ovarian cancer, hepatocarcinoma, leukaemia and in some cells, including K562 cells, Hela cells and B16-hm cells, it had been demonstrated that MGAT3 had a promoting effect on cancer [9,10,11,12,13,14]. In metastatic ovarian cancer, mammary tumor, and MKN45 cells, MGAT3 had the effect of inhibiting cancer [15,16,17]. All these results illustrate the heterogeneity of MGAT3 on cancer and cancer cells. Bisecting GlcNAc effects and molecular mechanisms in the development of intrahepatic cholangiocarcinoma also remain unclear.

In this study, we comparatively analyzed the bisecting GlcNAc changes between ICC and paracancer tissues using glycoproteomic data and lectin histochemistry. The expression and clinical relevance of MGAT3 were explored based on TCGA and recently published RNA-seq data from 255 ICC tumor and related tissues [18]. Molecular biology experiments were used to examine the effects of MGAT3 on ICC cells. In addition, a comparative glycoproteomic analysis were performed on ICC cells with and without MGAT3 overexpression. Overall, our data preliminarily elucidate the possible mechanisms of bisecting GlcNAc and MGAT3 in the progression of ICC, and provide new ideas for the discovery of possible therapeutic targets for ICC.

Materials and methods

Cell culture

RBE cell line was purchased from Chinese Academy of Sciences Cell Bank. HUCCT1 cell line was purchased from the cell bank of Japanese Collection of Research Bioresource (JCRB, Shinjuku, Japan). RBE and HUCCT1 cells were cultured in RPMI-1640 (VivaCell, Shanghai, China) with 10% fetal bovine serum (VivaCell, Shanghai, China) at 37 ℃ with 5% CO2.

Fluorescence-based lectin histochemistry

Tissue samples were prospectively collected from patients undergoing ICC resection at the First Affiliated Hospital of Xi’an Jiaotong University, China. The study was approved by Human Ethics Committee at the First Affiliated Hospital of Xi’an Jiaotong University, and written informed consents were obtained from all participants. Formalin-fixed paraffin-embedded (FFPE) tissue sections from the paired tumor and paracancer tissue of ICC were dewaxed and hydrated with xylene and gradient alcohol, respectively. For lectin histochemistry, Cy3-labeled PHA-E was applied to detect the Bisecting GlcNAc structures. Briefly, the sections were washed, blocked with 5% (w/v) BSA and 0.02% Triton for 1 h and incubated with 1 µg·µL− 1 Cy3-labeled PHA-E overnight at room temperature in the dark. Then nuclei were stained with DAPI (Solarbio, Beijing, China) for 10 min [19]. The fluorescence images were acquired by a fluorescence microscope, and the relative intensity of fluorescence was quantified by imagej software (http://imagej.net).

Stable knockdown and overexpressed cell lines were generated by using the lentiviral system

To generate MGAT3 knockdown variants, lentiviral shRNA was constructed with the pGreenPuro vector (System Biosciences, CA). The target sequences (5’-GAGTCCAACTTCACGGCTTAT-3’) of the MAGT3 were used as sh-MGAT3, and a scrambled shRNA was used as a control. To generate MGAT3 overexpression cell variants, full length MGAT3 cDNA from humans was amplified using PCR SuperMix High Fidelity (Takara, Japan) and ligated into the eukaryotic expression vector pCDH-CMV-MCS-EF1-Puro plasmid (System Biosciences, CA). An empty vector was used as a control. Lentiviral supernatants were generated by co-transfection of helper plasmids (System Biosciences, CA) into the packaging cell line (HEK293TN). Viral supernatants were harvested at 48–72 h after transfection. RBE and HUCCT1 cells were infected with filtered retroviral supernatant and selected with 2 µg/mL puromycin (Solarbio, China).

Western blot analysis

Proteins were extracted with RIPA lysis buffer (Beijing Solarbio Science & Technology Co., Ltd.). Total protein was separated by 10% SDS-PAGE and then transferred to 0.45 μm PVDF membranes (Millipore, USA). Membranes were blocked with 5% non-fat milk at room temperature for 1 h, followed by incubation with primary and secondary antibodies. Blots were developed with ECL reagent and analyzed on a chemoluminiscence instrument (ChemiScope 6200, CLINX, China).

Clone formation assay

The stable cells (1 × 103) were plated in a six-well plate. Media were changed every 3–4 days. After 10–14 days, surviving colonies were fixed with methanol for 15 min, and then stained by 0.1% crystal violet. The colony formation ability was presented as percentage compared with control group, and three independent experiments were performed.

Transwell assay

Cell invasion was assessed by transwell chambers with 8.0 μm pore membranes (Corning, USA). About 1 × 105 cells in precooled serum-free 1640 were seeded into the upper chamber coated with Matrigel (R&D, USA), and medium supplemented with 10% FBS was added to the lower chamber as a chemoattractant. Cells were then incubated at 37 ℃ with 5% CO2 for 24 h. The invaded cells were fixed with 4% paraformaldehyde for 30 min and stained with 0.1% crystal violet. Finally, different fields of cells were then randomly selected and photographed. The average of cell counts in four microscope fields per condition was used for plotting results.

Gene expression analyses in public datasets

The transcriptome profiling profiles of ICC were downloaded from The Cancer Genome Atlas (TCGA) Data Portal [20] (https://www.cancer.gov), including 33 tumors and 8 normal tissues. Statistical significance of the expression differences between groups was determined using Wilcox’s test.

Glycoproteomic analyses

The analytical methods of glycoproteomics have been described in detail in previous studies [21, 22]. In a nutshell, the details of sample treatment including cell lysis, protein extraction, protein digestion, peptide desalting, and enrichment of intact glycopeptides using MAX columns. LC-MS/MS analysis, database search for intact glycopeptide identification were performed based on StrucGP [22].

Bioinformatic analyses

Protein-protein interaction network was analyzed and performed by the STRING (http://string-db.org/) [23]. Gene Ontology (GO) and KEGG pathway was performed by Database for Annotation, Visualization, and Integrated Discovery (DAVID) V6.8 (https://david.ncifcrf.gov/home.jsp) [24]. Whole genome as background and P-value less than 0.05 was regarded as the significant pathway.

Results

Bisecting GlcNAc is upregulated in ICC tumors

Our existed LC-MS/MS data on ICC tumor (n = 6), HCC (n = 6) and their paired paracancer tissue (n = 12) were re-analyzed in this study [19]. After filtering results by ≥ 5 glycopeptide-spectrum matches (GPSMs), a total of 56 glycopeptides modified by bisecting GlcNAc were identified from ICC tumors and paracancer tissues, which accounted for 5.17% of all intact glycopeptides (1084) (Fig. 1 A, Sup Table S1). The identification of glycopeptide with bisecting GlcNAc was determined by a series of feature Y ions (such as peptide + HexNAc3Hex1 at m/z 1429.69 (z = 2), peptides + HexNAc3Hex1Fuc1 at m/z 1501.71 (z = 2)) (Fig. S1).

Fig. 1
figure 1

Bisecting GlcNAc altered in ICC tumors compared with paracancer tissues. A, Percentage of unique glycopeptides with bisecting GlcNAc in all glycopeptides. B, Changes of bisecting N-glycopeptides in ICC compared with paired paracancer samples. “Up” and “Down” indicate fold change of glycopeptides ≥ 1.7 and ≤ 0.588 in ICC, respectively. “NS” (no significant change) stands for fold change < 1.7 or > 0.588 in ICC tumors compared with paracancerous tissues. C, Heatmap of differentially expressed bisecting N-glycopeptides. TICC, tumors of ICC; PICC, paracancers of ICC; Gene name (glycosite, glycan composition). D, Frequencies of changed site-specific bisecting GlcNAc in ICC (n = 6). E, Lectin histochemistry analysis of ICC tumors and paired paracancer tissues (n = 3, ±SD, P < 0.001 by paired t-test) based on the WFA binding (40×magnification). A-D were derived from reanalysis of the data from previous research [19]

Full size image

The TMT-labeled quantitation results showed that 47 of 56 glycopeptides (95.7%) modified with bisecting GlcNAc were up-regulated, and 2 (3.6%) of them were down-regulated in ICC tumors with 1.7-fold as a cutoff (Fig. 1B), which was determined based on the LC-MS analysis of ICC and HCC paracancers tissues. (95.2% of glycopeptides were within 1.7-fold changes, Fig. S2). These differently expressed glycopeptides were comprised of 21 N-glycan structures and 22 glycosites from 20 glycoproteins (Fig. 1 C-D).

To further confirm the increase of bisecting GlcNAc in ICC, we conducted lectin histochemistry analysis on three ICC tumors and their paired paracancer tissues using Cy3-labeled PHA-E for the detection of bisecting GlcNAc. The histochemical results demonstrated that the binding of PHA-E was significantly stronger (P < 0.001) in ICC tumors than in their adjacent normal tissues (Fig. 1E).

Upregulated MGAT3 is negatively correlated with overall survival of ICC patients

To further confirm the increase of bisecting GlcNAc in ICC, the MGAT3 expression was profiled at both mRNA and protein levels in ICC (Fig. 2 A). Based on the transcriptome data in the TCGA database, the mRNA level of MGAT3 was significantly elevated (P < 0.0001) in ICC tumors (n = 33) compared with normal tissues (n = 8) (Fig. 2B). According to the immunohistochemistry data in the Human Protein Atlas (HPA) database, the staining positivity of GnT-III increased in ICC tissue, which was consistent with the mRNA upregulation in ICC patients identified in TCGA data (Fig. 2 C).

Fig. 2
figure 2

Expression levels ofMGAT3and survival analysis of ICC patients. A, MGAT3 encodes for the glycosyltransferase GnT-III that catalyzes the transfer of N-acetylglucosamine (blue box) to the core mannose of the N-glycan in β1,4 linkage to form the bisecting N-linked glycan structure. B, Transcript levels of MGAT3 in ICC tumors (n = 33) and normal tissues (n = 8) by TCGA database (https://www.cancer.gov), (****P < 0.0001, Wilcoxon’s test). C, Validation of MGAT3 in the translational level. Immunohistochemistry (IHC) staining data were obtained from the Human Protein Atlas (HPA) database. D-F, Comparison of MGAT3 expression in TNM stage (D), tumor size (E), Vascular invasion, regional lymph node metastasis and perineural invasion (F) based on a recently published RNA-seq dataset from 255 ICC tumor tissues [18]. Different groups were separated by the median expression. G, Expression of MGAT3 correlates with survival in 255 ICC tumor tissues (Only 244 patient samples had overall survival information) [18]. The median value of expression was set as the cut-off point. The number of individuals at risk is depicted at the bottom. Significance was determined with log-rank test. P = 0.0195; HR = 1.61. Data are presented as mean ± SEM; P-values were determined by unpaired two-tailed (B-F). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, no significant difference

Full size image

To elucidate the relevance of MGAT3 with tumor progression, we further investigated the relationship between MGAT3 expression and clinical features of ICC based on a recently published RNA-seq dataset from 255 ICC tumor tissues [18]. Based on the median MGAT3 mRNA expression, we divided the patients into high and low MGAT3 mRNA-expressing groups. For TNM stage analyses, the transcription levels of MGAT3 were significantly higher in the III-IV than in the I- II (Fig. 2D), while tumor size showed no significant relevance with MGAT3 expression (Fig. 2E). MGAT3 appeared to be highly expressed in patients with other clinicopathological features, including vascular invasion, regional lymph node metastasis, and perineural invasion (Fig. 2 F). Furthermore, survival statistics indicated a significant reduction of overall survival in ICC patients with high MGAT3 expression compared with low MGAT3 expression (Fig. 2G). These results suggested that the MGAT3 expression differed significantly among ICC tumors, and was associated with ICC progression.

Overexpression of bisecting GlcNAc promotes ICC cell invasion

To evaluate the effect of MGAT3 expression on malignant biological behaviors, stable MGAT3 overexpression (OE) and knockdown (KD) of RBE and HUCCT1 cell lines were established via lentiviral transfection. Colony formation and transwell assays were performed to determine the effects of MGAT3 on colony formation and invasion of ICC cells. The colony formation analysis revealed that knockdown of MGAT3 inhibited the clonal formation of RBE and HUCCT1 cells, while overexpression of MGAT3 showed no significant effect on clonal formation (Fig. 3 A). Transwell invasion assays revealed that knockdown of MGAT3 significantly inhibited cell invasion and overexpression of MGAT3 significantly promoted cell invasion in RBE and HUCCT1 cells (Fig. 3B).

Fig. 3
figure 3

Overexpression ofMGAT3promotes invasion of ICC cells A, Representative microphotographs of clone formation experiments in stable ICC cell lines of MGAT3 knockdown and overexpression. B, The Matrigel Trans-well invasion assay (Scale bar = 10 μm) in stable ICC cell lines of MGAT3 knockdown and overexpression. C, Protein expression levels of MGAT3, E-cadherin and vimentin in RBE and HUCCT1 cell lines with MGAT3 knockdown and overexpression (left). Quantifications analysis of the protein expression levels of MGAT3, E-cadherin and Vimentin (right). The grayscale values were measured from the western blotting data with Image J. Analyses were performed in triplicates for each condition. Data are presented as mean ± SEM; P-values were determined by unpaired two-tailed t-test (A and B) or One-way ANOVA (C). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, no significant difference

Full size image

The detection of E-cadherin and vimentin protein expression was conducted by western blot analysis (Fig. 3 C). The results showed that E-cadherin expression was negatively correlated with MGAT3, while expression of MGAT3 did not have a significant effect on vimentin expression. Taken together, these findings demonstrated that MGAT3 promoted ICC cell invasion and was related to metastasis of ICC.

Functional investigation of glycoproteins with bisecting GlcNAc

To explore the biological relevance of bisecting GlcNAc in ICC cells, N-glycoproteome analyses were again performed on the control and overexpressed RBE and HUCCT1 cells. Glycoproteome analysis identified 1,994–2,404 glycopeptides from RBE and HUCCT1 cell lines with/without MGAT3 overexpression (Sup Table S2-S5). The number of glycopeptides modified by bisecting GlcNAc was significantly increased in MGAT3 overexpressed RBE and HUCCT1 cell lines (Fig. 4 A). A total of 38 glycoproteins (204 glycopeptides) and 30 glycoproteins (135 glycopeptides) with bisecting GlcNAc were identified in MGAT3 overexpressed RBE and HUCCT1 cell lines, respectively (PSMs ≥ 5), of which 17 glycoproteins are shared in both cell lines (Fig. 4B). The number of unique glycopeptides from those glycoproteins are increased to varying degrees (Sup Table S6).

Fig. 4
figure 4

Functional enrichment analysis of glycoproteins with bisecting GlcNAc in RBE and HUCCT1 cells. A, Percentages of glycopeptides with bisecting GlcNAc in RBE and HUCCT1 cell. Percentages were calculated based on the unique glycopeptides (left) and PSMs (right) B, Glycoproteins modified by bisecting GlcNAc that were commonly identified in two ICC cell lines (PSM ≥ 5). C and D, Gene Ontology (GO) (C) and KEGG pathway (D) analysis of common glycoproteins modified by increased bisecting GlcNAc between RBE and HUCCT1 cell lines (P < 0.05). Biological process (BP), cellular component (CC), and molecular function (MF). E, Protein-protein interaction network of common glycoproteins modified by bisecting GlcNAc identified in two ICC cell lines. The red lables mark the glycoproteins involved in cell adhesion (GO:0007155). Nodes represent proteins; edges, network connections; line thickness, strength (thicker = stronger). The minimum required interaction score selects medium confidence (0.400)

Full size image

To further explore the biological functions of 17 common glycoproteins with increased bisecting GlcNAc in MGAT3 overexpressed RBE and HUCCT1 cells, GO and KEGG pathway analyses were performed. These glycoproteins were mainly involved in cell adhesion, viral entry into host cell, cell-matrix adhesion, integrin-mediated signaling pathway and cell adhesion mediated by integrin; and were mainly localized in extracellular exosome, cell surface, external side of plasma membrane, integrin complex and azurophil granule membrane. In addition, integrin binding, virus receptor activity, laminin binding, fibronectin binding, and protease binding were the most enriched in terms of molecular function (MF) enrichment (Fig. 4 C). KEGG pathway analysis showed the significant enrichment of ECM-receptor interaction, phagosome, focal adhesion, proteoglycans in cancer, and PI3K-Akt signaling pathway and lysosome of these glycoproteins (Fig. 4D). To obtain a better idea of the potential relationships between those 17 glycoproteins, protein–protein interaction network analysis was also performed by the STRING database. Many glycoproteins have interactions, and 10 proteins are related to cell adhesion (GO:0007155). All these data indicated that bisecting GlcNAC mainly affects cell behavior by biological processes related to cell movement, such as cell adhesion, integrin-related and ECM-receptor interaction in RBE and HUCCT1 cells.

Discussion

In recent years, the incidence and subsequent mortality of intrahepatic cholangiocarcinoma (ICC) have been increasing worldwide despite recent advances in therapeutic approaches [25]. Therefore, an in-depth understanding of its mechanism and a more effective treatment strategy are urgently needed. Aberrant glycosylation has been recognized as a fundamental feature of many cancers and plays an important role in key pathological steps in tumor development and progression [5]. We previously performed precision N-glycoproteomic analysis on both ICC and HCC tumors as well as paracancer tissues to investigate their aberrant site-specific N-glycosylation [26], and found that partial bisecting GlcNAc structures were commonly elevated in ICC and HCC. However, the overall changes of bisecting GlcNAc in ICC and the effects are still unclear.

In this study, we found that the bisecting GlcNAc and its glycosyltransferase MGAT3 were significantly up-regulated in ICC. By silencing or overexpression of MGAT3 in RBE and HUCCT1 cell lines, and combined with phenotypic experiments and glycoproteomic analysis, the effects and possible mechanisms of bisecting glycopeptides on ICC cells were systematically analyzed. In addition, our new method StrucGP can provide overall distributions of bisecting GlcNAc, the modified glycoproteins, glycosites, and possible N-glycan structures [7, 27]. The application of StrucGP in this study greatly facilitates N-glycoproteomic analysis [7].

We identified 47 bisecting N-glycopeptides from 20 glycoproteins that were altered in ICC, with 95.7% of altered glycopeptide were up-regulated. Among the 20 glycoproteins, the highest number of bisecting GlcNAc were identified from IGHG1 and lumican, with nine and eight bisected glycans, respectively. It has been reported that bisecting GlcNAc on IGHG1 were also highly expressed in cirrhosis (CIR) and HCC [28, 29]. Studies showed that bisecting GlcNAc could change the conformation of IGHG1 to influence the binding of IgG to its receptors, and affect the progression of liver diseases [30]. Glycosylation of lumican may protect collagen against MMP-14 proteolysis by decrease MMP-14 activity in tumor cells, and thus influences cell–matrix interaction in tumor progression [31]. Considering the regulatory roles of bisecting GlcNAc in N-glycan biosynthesis, it is possible that bisecting N-glycans on lumican could be involved in the interaction between lumican and MMP-14.

MGAT3 expression was also consistently up-regulated in ICC tissues at both the mRNA and protein levels. Using RNA-seq data from 255 ICC tumor tissues, we found that increased expression of MGAT3 is associated with decreased overall survival. Some clinical events, including vascular invasion, regional lymph node metastasis, and perineural invasion, were shown to be related to MAGT3 expression. Vascular invasion and regional lymph node metastasis are associated with overall survival, and perineural invasion is associated with recurrence of cholangiocarcinoma [3, 32]. However, there was no significant correlation between MGAT3 expression and tumor size, which is consistent with the conjecture that MGAT3 is involved mostly in cell differentiation rather than cell proliferation during cell culture [33]. Similar with our results in ICC, the over-expression of MGAT3 was also found to be associated with enhanced tumor progression in spleen, which was partially mediated by CD44-hyaluronan interactions enhanced by the bisecting GlcNAc on CD44 [13].

Our finding shows that MGAT3 expression is positively correlated with cell invasion. Meanwhile, MGAT3 expression levels were slightly negatively correlated with the expression levels of E-cadherin. According to previous studies, different glycan structures were described to regulate E-cadherin expression through interfering with its basic mechanisms including E-cadherin-mediated cell-cell adhesion, E-cadherin cellular trafficking, and E-cadherin-dependent signaling pathways [34]. As for bisecting GlcNAc, its over-expression induced by MGAT3 might affect the expression of E-cadherin by regulating its turnover rate and its release from cell surface [35] [36]. In present study, glycosylated E-cadherin was not identified in RBE and HUCCT1 cell lines, which may be due to the low abundance of glycosylated E-cadherin or the expression of E-cadherin is only slightly affected by other pathways without glycosylated. Altering the expression of glycosyltransferases has a significant impact on the glycosylation remodeling of various glycoproteins, thereby affecting various cellular behaviors such as cell invasion. The detailed mechanism behind this process is complex and worthy of further investigation. In previous studies, MGAT3 also has been shown to promote the progression of malignant behavior in other cells and cancers. MGAT3 overexpression in K562 cells promotes cell spleen colonization [9]. In HeLa cells, overexpression of MGAT3 promotes cell metastasis by increasing EGFR signaling and decreasing cell adhesion [11]. In addition, it has also been observed that MGAT3 is proportional to invasive ability in ovarian cancer [10, 37]. However, several other studies have demonstrated that MGAT3 can also act as a tumor suppressor [17] [38]. All these results illustrate the heterogeneity of MGAT3 on cancer cells, and the reason remains to be further investigated.

By using high-throughput glycoproteomics, we systematically charactered bisecting GlcNAc in RBE and HUCCT1 cells, including the corresponding glycoproteins information. We detected a serial of glycoproteins with increased bisecting GlcNAc in MGAT3 overexpressed RBE and HUCCT1 cells. Evidences have shown that bisecting GlcNAc might modulate the biological activity of its carrier proteins. For example, overexpression of bisecting GlcNAc on CD44 enhances its adhesion to hyaluronate, and affects local tumor growth and metastatic growth in the spleen [13]. Bisecting GlcNAc on EGFR could regulate endocytosis of EGFR and further modulate downstream signaling [39]. So far, details of molecular function of bisecting GlcNAc in regard to the regulation of its carrier proteins associated with cellular invasiveness (especially promotion of invasive) are complex and still remain unclear due to the varied consequences and possibly mechanisms of bisecting GlcNAc expression among different cell types. Further studies are still required to fully explore molecular functions of bisecting GlcNAc associated with cellular invasiveness. ECM-receptor interaction is the most prominent pathway enriched in the KEGG pathway. ECM-receptor interaction mainly controls cell adhesion, migration, proliferation, and coagulation cascade activation [40]. Our research provides rich data for exploring mechanisms between bisecting GlcNAc and their carrier proteins.

Based on all the above analyses, MGAT3 plays an important role in ICC carcinogenesis. Although the exact molecular mechanism of MGAT3 on cell behavior is still unclear, the glycoproteomics data presented here provides useful insights for future researches. These achievements will be greatly beneficial for further in-depth functional investigation and specific regulatory mechanisms of bisecting GlcNAc and MGAT3 in ICC.