Degradation of the basement membrane by MMPs (matrix metalloproteinases) is one of the most critical steps in tumour progression. CD147 is a tumour-associated antigen that plays a key regulatory role for MMP activities. In the present study, mass spectrum analysis demonstrated that the purified native CD147 from human lung cancer tissue was N-glycosylated and contained a series of high-mannose and complex-type N-linked glycan structures. Moreover, native glycosylated CD147 existed exclusively as oligomers in solution and directly stimulated MMP production more efficiently than non-glycosylated prokaryotic CD147. The glycosylation site mutation results indicated that, among three N-glycan attachment sites, the N152Q mutants were retained in the endoplasmic reticulum and unfolded protein response signalling was activated. This improper intracellular accumulation impaired its MMP-inducing activity. Increased β1,6-branching of N-glycans as a result of overexpression of GnT-V (N-acetylglucosaminyltransferase V) plays an important role in tumour metastasis. In the present study, we identified CD147 as a target protein of GnT-V and found that overexpression of GnT-V resulted in an elevated level of CD147 at the plasma membrane and in cell-conditioned medium, thereby increasing the induction of MMPs. The present study reveals the important role of N-glycosylation of CD147 in its biological function and implied that targeting aberrant β1,6-branching of N-glycans on CD147 would be valuable for the development of novel therapeutic modalities against carcinoma.

INTRODUCTION

Glycans are involved in each and every aspect of tumour progression, from cellular proliferation to angiogenesis and metastasis [1]. Changes in post-translational modification of proteins are closely associated with malignant transformation of tumour cells.

An aberrant glycosylation induced by GnT-V (N-acetylglucosaminyltransferase V) is a representative example of such modification that is implicated in tumour progression. Up-regulation of GnT-V leading to increased β1,6-GlcNAc branching of N-linked glycans is a major hallmark of cancer progression [2]. Several target proteins for GnT-V were proposed to be involved in cancer progression, including E-cadherin (epithelial cadherin) [3], matriptase [4], integrin [5] and N-cadherin (neural cadherin) [6].

Degradation of the basement membrane by MMPs (matrix metalloproteinases) is one of the most critical steps in tumour progression. A search for MMP-inducing factors in tumour cells led to the identification of CD147, also known as EMMPRIN (extracellular matrix metalloproteinase inducer), a highly glycosylated cell-surface transmembrane protein which stimulates MMP synthesis in neighbouring fibroblasts and tumour cells [7]. CD147 is highly expressed in various human carcinoma tissues and cell lines, correlating with tumour progression under experimental and clinical conditions [8]. The best characterized function of CD147 is its ability to induce the expression of MMPs, including MMP-1, MMP-2, MMP-3, MMP-9 and MMP-11 in stromal cells [911]. Many studies in our laboratory have investigated the involvement of CD147 in tumour progression [1214]. Our preclinical work and Phase I clinical trials have demonstrated that the hepatoma-associated antigen HAb18G/CD147 is a potential target for the development of diagnostic and treatment methods for CD147-associated diseases. Indeed, we have developed the specific mAb (monoclonal antibody) HAb18 against CD147 for use as a radioimmunotherapeutic agent (trade name, Licartin; generic name, metuximab iodine-131, for injection). Licartin has proved safe and effective for the targeted treatment of HCC (hepatocellular carcinoma) in clinical trials and was approved as a new drug for the clinical therapy of primary HCC by the China State Food and Drug Administration (No. S20050039) in April 2005 [15,16].

A significant biochemical property of CD147 is its high level of glycosylation. N-glycosylation contributes to almost half the size of the mature CD147 molecule [17]. The extracellular region of CD147 contains three N-linked glycosylation sites (Asn44, Asn152 and Asn186), which contribute to both the high-glycosylated form HG-CD147 (~40–60 kDa) and the low-glycosylated form LG-CD147 (~32 kDa) [18]. Previous studies were controversial regarding the dependency of CD147 activity on glycosylation. Sun and Hemler [19] suggested that purified deglycosylated CD147 not only failed to induce MMP-1 and MMP-2 expression, but also antagonized the MMP induction by native CD147. However, a previous study has demonstrated that non-glycosylated recombinant CD147 stimulated fibroblasts to express the mRNAs of MMP-1, MMP-2 and MMP-3 [20]. Therefore the present study clarifies further the roles of N-glycosylation on CD147 activity. Moreover, CD147 is a cell-surface carrier of β1,6-branched polylactosamine sugars on HT1080 tumour cells [18]. However, the functional significance of β1,6-glycans on CD147 has not been studied so far.

In view of the high expression of CD147 in malignant tissues and its potential as a target for cancer therapy, many studies have investigated how CD147 modulates MMP in cancer. As a transmembrane glycoprotein, CD147 forms homo-oligomers in both heterotypic and homotypic cell–cell interactions to induce production of MMPs [21]. Moreover, full-length EMMPRIN is released by tumour cells [22], and one possible release mechanism for CD147 is via vesicle shedding [23]. Secreted soluble CD147 in conditioned medium is equally capable of inducing MMP production, either from surrounding fibroblasts or tumour cells themselves. Our previous study, which resolved the crystal structure of CD147 using X-ray crystallography, revealed that recombinant prokaryotic CD147 forms homo-oligomers in the crystal form, as well as in solution [24]. However, the oligomeric status of native eukaryotic CD147 in solution has not yet been investigated.

The present study revealed that CD147 contains a series of high-mannose and complex-type N-linked glycan structures. Native glycosylated CD147 exists as an advanced molecular conformation in solution. The glycosylation site mutation results revealed that N-glycans of CD147 contribute to its MMP-inducing activity by affecting the localization of CD147 on the cell surface. Furthermore, aberrant β1,6-glycans play crucial roles in the biological activity of CD147 and could be a possible marker for tumour metastasis.

EXPERIMENTAL

Materials

Antibodies and reagents were as follows: human hepatoma mAb HAb18 was produced and characterized in our laboratory as described previously [25]. Affi-Gel® Hz Immunoaffinity Kit was from Bio-Rad Laboratories. The protease inhibitor cocktail, N-glycosylation inhibitor tunicamycin, swainsonine, avidin–HRP (horseradish peroxidase) and endoglycosidase F3 were from Sigma. OG (N-octyl-β-D-glucopyranoside) was from Merck. Anti-GRP78 (glucose-regulated protein 78) BiP (immunoglobulin heavy-chain-binding protein) antibody, anti-GADD153 (growth-arrest and DNA-damage-inducible protein 153)/CHOP [C/EBP (CCAAT/enhancer-binding protein)-homologous protein] antibody and anti-α-tubulin antibody were from Santa Cruz Biotechnology. The plasma membrane protein extraction kit was purchased from Biovision. Anti-flotillin-1 mAb was purchased from BD Biosciences. The immunoprecipitation kit was from Pierce. PHA-L (phytohaemagglutinin-L) was purchased from Vector Laboratories. Alexa Fluor® 594-conjugated goat anti-(mouse IgG) and ER (endoplasmic reticulum)-Tracker™ Blue-White were from Invitrogen. PNGase F (peptide N-glycosidase F) was from New England Biolabs. Anti-MMP-2 antibody was purchased from Calbiochem. Human SMMC-7721 and HepG2 cells were obtained from the Institute of Cell Biology (Academic Sinica, Shanghai, China). Human dermal fibroblasts were a gift from the Department of Burn Surgery (Center of Plastic Surgery, Xijing Hospital, Fourth Military Medical University, Xi’an, China). An approximately 20 g tissue specimen of lung squamous cell carcinoma was obtained from a patient undergoing lung resection at Tangdu Hospital affiliated to Fourth Military Medical University (Xi’an, China) with signed informed consent. Study approval was obtained from the Xijing Hospital Institutional Review Board.

Purification of native eukaryotic CD147 from human lung cancer tissue

CD147 was purified from the 20 g tissue specimen of lung cancer from a patient by immunoaffinity chromatography using mAbHAb18. Briefly, lung cancer tissue was stored at −80°C until use. It was minced and homogenized on ice in 20 mM Tris/HCl buffer (pH 8.2), containing 1% OG, 1 mM EDTA and protease inhibitor cocktail. Soluble proteins were separated by centrifugation for 1 h at 12000 g. The clarified supernatant was collected for affinity purification. Then mAb HAb18 was coupled to Affi-Gel Hz gel using the Affi-Gel® Hz Immunoaffinity Kit according to the manufacturer's instructions. Before the soluble extract of human lung cancer tissue was applied to the column, the immunoaffinity column was equilibrated with 10 bed volumes of PBS (pH 7.4). After recirculating the sample through the column for 12 h at 4°C, the column was washed with 1× PBS (pH 7.4) until the A280 was <0.01. CD147 was then eluted from the column with 3 M urea. The eluted protein was immediately monitored with a UV spectrophotometer and ultrafiltrated with an Amicon Ultra Centrifugal Filter Unit (Millipore) to remove the urea. The eluted antigen was then freeze-dried.

Expression and purification of the extracellular portion of prokaryotic CD147

A cDNA encoding amino acids 22–205 of CD147 was inserted into pET21a(+) (Novagen) with NdeI and XhoI, and the integrity was confirmed by automated sequencing. This construct was chemically transformed into the Escherichia coli strain and grown in Luria–Bertani broth, yielding soluble CD147 in the bacteria. The bacterial pellet was resuspended and sonicated in 20 mM Tris/HCl, pH 8, and centrifuged at 18000 g for 30 min. The supernatant was applied directly to a HiTrap QHP column, followed by a MonoQ ion-exchange column (GE Healthcare). A Superdex 75 gel-filtration column (GE Healthcare) was used as the final purification step with 20 mM Tris/HCl and 150 mM NaCl, pH 8.0. The eluate was concentrated to 20 mg/ml.

Deglycosylation with PNGase F and Endo F3 (endoglycosidase F3)

Digestion was performed as described by the manufacturer. Briefly, purified native CD147 was denatured in the supplied buffer and was digested with PNGase F for 1 h and 4 h at 37°C in 1× G7 buffer before resolving by SDS/PAGE. In the oligomerization studies, the denaturation procedure was omitted and the protein was digested under native conditions for 24 h. Endo F3 cleaves bi-antennary and tri-antennary complex oligosaccharides under native conditions. The reactions were carried out in the supplied reaction buffer for 12 h and 24 h at 37°C.

Preparation of glycopeptides and release of N-linked glycans

CD147 (1 mg) was digested with trypsin and chymotrypsin for 18 h at 37°C in 0.1 M Tris/HCl, pH 8.2, containing 1 mM CaCl2. The digestion products were enriched and freed of contaminants by Sep-Pak C18 cartridge column. After enrichment, the glycopeptides were digested with 2 μl of PNGase F (7.5 units/ml) in 50 μl of 20 mM sodium phosphate buffer, pH 7.5, for 18 h at 37°C. Released oligosaccharides were separated from the peptide and enzyme by passage through a Sep-Pak C18 cartridge column.

Preparation of the per-O-methylated glycans

The glycan fraction was dissolved in DMSO and then permethylated on the basis of the method of Anumula and Taylor [26]. The reaction was quenched by addition of water and per-O-methylated carbohydrates were extracted with dichloromethane. Per-O-methylated glycans were dried under a stream of nitrogen.

Nanospray ionization-linear ion-trap MS

Mass analysis by NSI-MS (nanospray ionization MS) was performed on a LTQ Orbitrap Discoverer mass spectrometer (Thermo Scientific) equipped with a nanospray ion source. Briefly, permethylated glycans were dissolved in 1 mM NaOH in 50% methanol and infused directly into a linear ion-trap mass spectrometer using a nanospray source at a syringe flow rate of 0.5 μl/min. The capillary temperature was set to 210°C, and MS analysis was performed in positive ion mode. Fragmentation by CID in MS/MS (tandem MS) modes with 50% collision energy was applied. The nomenclature of Domon and Costello [26a] was used to guide the depiction of fragmentation derived from MS/MS spectra.

Plasmid and transfection

The human pcDNA3/GnT-V plasmid was generously provided by Professor Hua-Bei Guo at University of Georgia. This plasmid was characterized using KpnI and XbaI digestion; a 2.3 kb fragment of full-length GnT-V cDNA and a 5.45 kb linearized vector were found after agarose electrophoresis. SMMC-7721 cells were transfected with pcDNA3/GnT-V plasmid using Lipofectamine™ 2000 reagent (Invitrogen) according to the manufacturer's instructions. Transfected 7721 cells were incubated in RPMI-1640 medium containing G418 (0.8 mg/ml). Neomycin-resistant cells were obtained after 2–3 weeks and recloned by serial dilution. The vector pcDNA3 was also transfected into SMMC-7721 cells as a mock control. Site-directed mutagenesis was performed with the QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Stratagene) to introduce asparagine to glutamine mutations in an expression plasmid encoding the CD147-pEGFP-N1 fusion protein. We used the following primers: N44Q (forward, 5′-ctcacctgctccttgcaggacagcgccacagag-3′; reverse, 5′-ctctgtggcgctgtcctgcaaggagcaggtgag-3′); N152Q (forward, 5′-acaaggccctcatgcagggctccgagagcag-3′; reverse, 5′-ctgctctcggagccctgcatgagggccttgt-3′); N186Q (forward, 5′-ccagtaccggtgccagggcaccagctcca-3′; reverse, 5′-tggagctggtgccctggcaccggtactgg-3′). Triple mutants were also produced according to the manufacturer's instructions. Introduction of the mutations into cDNAs was verified by DNA sequencing. The expression vector pcDNA3-CD147 was made in our laboratory. siRNA (small interfering RNA) designed to target CD147 was 5′-GUUCUUCGUGAGUUCCUCdTdT-3′ and 3′-dTdTCAAGAAGCACUCAAGGAG-5′.

Cell culture

SMMC-7721 and HepG2 cells were cultured with RPMI 1640 medium (Gibco) supplemented with 10% FBS (fetal bovine serum; Gibco), 1% penicillin/streptomycin, and 2% L-glutamine at 37°C in a humidified atmosphere of 5% CO2. Conditioned medium was collected and was centrifuged to remove cell debris. Supernatant protein was concentrated in a 10000 Da molecular mass Amicon Ultra Centrifugal Filter Unit (Millipore). Medium concentrates were resuspended in Tris/glycine/SDS solubilization buffer (ICN Biomedicals), and 10% of total samples were typically used for Western blot analysis. Cell lysates were prepared by washing the cells in PBS, before lysis in RIPA buffer (Beyotime) containing a cocktail of proteinase inhibitors.

Western blot analysis

Cell lysates were subjected to SDS/PAGE (10% gels) under reducing conditions. After electrophoresis, proteins were transferred on to PVDF membranes (Millipore) in running buffer with 10% methanol. After non-specific sites were blocked with 5% (w/v) non-fat dried skimmed milk powder in PBST (10 mM phosphate buffer, pH 7.4, 0.15 M NaCl and 0.05% Tween-20) for 30 min at room temperature (25°C), the membrane was then incubated in PBST/5% (w/v) non-fat dried skimmed milk powder with primary antibodies for 2 h at room temperature. After three washes with PBST, the membranes were incubated with the appropriate goat anti-(mouse IgG) (Fc-specific) HRP-conjugated secondary antibody (Pierce) for 1 h in PBST/5% (w/v) non-fat dried skimmed milk powder (1:5000 dilution). Protein bands were visualized using ECL (enhanced chemiluminescence) Plus Western blotting detection kit (Amersham Biosciences) according to the manufacturer's instructions. A PageRuler Prestained protein Ladder Plus (Fermentas Life Sciences) was used for sizing of proteins on Western blots.

Plasma membrane preparation

A plasma membrane protein extraction kit (BioVision) was used according to the manufacturer's protocol to extract total cellular membrane proteins and specifically the plasma membrane from the total cellular membranes. The plasma membrane fraction was dissolved in 0.5% Triton X-100 in PBS and protein concentration was measured using the BCA (bicinchoninic acid) protein assay (Pierce). Flotillin-1 (BD Biosciences) was used as a loading control to show the same amount of plasma membrane protein in each lane.

Lectin blot

Briefly, 10 μg of proteins were subjected to SDS/PAGE (10% gels). After the electrophoresis, the gels were blotted on to PVDF membranes. The membranes were incubated with 3% BSA in TBS (Tris-buffered saline; 20 mM Tris and 0.5 M NaCl, pH 7.5) overnight and then the blots were incubated with biotinylated PHA-L (1:400 dilution) in PBST for 1 h. After washing with PBST, the membranes were incubated with avidin–HRP (1:1000 dilution) for 1 h and then washed with PBST. Staining was detected with ECL Western blot detection reagents. A PageRuler Prestained protein Ladder Plus (Fermentas Life Sciences) was used for sizing of proteins on lectin blots.

SDS/PAGE and native PAGE

The purified CD147 protein was analysed by SDS/PAGE and native PAGE. For SDS/PAGE, 10 μg of purified protein was resuspended in 5× Laemmli buffer (312 mM Tris/HCl, pH 6.8, 10% SDS, 50% glycerol, 25% 2-mercaptoethanol and 0.01% Bromophenol Blue) and boiled for 5 min. For native PAGE, 10 μg of purified protein was mixed with 5× Laemmli buffer without 2-mercaptoethanol. In addition, the boiling step was omitted. Then the samples were loaded on to a 10% native polyacrylamide gel for electrophoresis.

Gelatin zymography

Cells with or without treatment were cultured in serum-free medium for 12 h, the conditioned medium was collected and mixed with non-reducing sample buffer, incubated in a water bath (55°C) for 3–5 min, and then loaded on to a 12% zymographic SDS gel containing gelatin (1 mg/ml). After electrophoresis, the gel was washed in 2.5% Triton X-100 for 15 min twice and incubated in incubation buffer (50 mM Tris/HCl, pH 8.8, containing 10 mM CaCl2 and 0.15 M NaCl) at 37°C for 16–18 h. Then the gel was stained with 0.25% Coomassie Blue in 50% methanol and 10% acetic acid for 4 h, and finally destained. MMPs were detected as a clear band against a blue background.

Immunofluorescence and confocal microscopy

Cells were allowed to attach to pre-coated glass coverslips overnight. They were fixed the following day in 4% paraformaldehyde and then blocked with 2% BSA in PBS for 0.5 h. Coverslips were incubated with the mAb HAb18 in PBS for 1 h. Primary antibody-treated cells were washed in PBS and then incubated with Alexa Fluor® 594 goat anti-mouse secondary antibodies at a 1:500 dilution in PBS for 1 h. Cell nuclei were dyed with DAPI (4′,6-diamidino-2-phenylindole) (Invitrogen) for 3 min. Finally, the cells were mounted using glycerol and observed using a FV1000 laser-scanning confocal microscope (Olympus).

Statistical analysis

All experiments were performed in triplicate, and the results were expressed as means±S.D. Statistical significance was determined using one-way ANOVA analysis or Student's t test. GraphPad Prism software (Cricket Software) was used for the above analyses and differences were deemed significant if P<0.05. ** indicates P<0.01, and * indicates P<0.05.

RESULTS

Purified native eukaryotic CD147 was N-glycosylated and contains a series of high-mannose and complex-type N-linked glycan structures

To determine the glycosylation feature of native CD147, we purified native CD147 from human lung cancer tissue by immuno-affinity chromatography. The recombinant extracellular portion of prokaryotic CD147 was also purified. Coomassie Blue staining showed that the molecular mass of purified eukaryotic CD147 was approximately 50–60 kDa, whereas the purified prokaryotic CD147 was 21 kDa (Figure 1A). Furthermore, deglycosylation treatment using PNGase F revealed that the purified eukaryotic CD147 was N-glycosylated (Figure 1B). A prolonged incubation time of 4 h resulted in the appearance of a positive 27 kDa band, which is consistent with the size of the CD147 core protein (Figure 1B). Endo F3 also reduced the size of CD147, although not dramatically, indicating that eukaryotic CD147 contained N-linked bi-antennary and tri-antennary complex oligosaccharides (Figure 1C). We then performed a lectin blot to determine the nature of glycans in purified native CD147. PHA-L binds preferentially to GlcNAc residues on β1,6 branches of tri- or tetra-antennary sugar chains and is indicative of β1,6-GlcNAc branching and polylactosamines [27]. As shown in Figure 1(D), PHA-L binding to the HG form of purified native CD147 (~55 kDa) was detected. This result is consistent with a previous study [18], and suggests that the mature form of CD147 contains β1,6-branched polylactosamines, which are specifically catalysed by GnT-V [28,29]. The identification and characterization of proposed structures of N-linked glycans of CD147 by NSI-MS are presented in Figure 1(E), in which the major structure of the N-glycans was the presence of a series of high-mannose and complex-type N-linked glycan structures. In addition, structural assignments in the table of Figure 1(E) indicate that native CD147 from human lung cancer tissue contains a high percentage of core fucosylated structures (28.8%).

Purified native eukaryotic CD147 is N-glycosylated and contains a series of high-mannose and complex-type N-linked glycan structures

Figure 1
Purified native eukaryotic CD147 is N-glycosylated and contains a series of high-mannose and complex-type N-linked glycan structures

(A) Native eukaryotic CD147 was purified from 20 g of human lung cancer tissue by immunoaffinity chromatography. The recombinant extracellular portion of prokaryotic CD147 was also purified. Both eukaryotic and prokaryotic CD147 were resolved by SDS/PAGE. A total of 10 μg of protein were resolved per lane. (B) Purified eukaryotic CD147 was treated with or without PNGase F and then analysed by SDS/PAGE and Western blotting. Left-hand panel: Coomassie Blue staining after treatment with PNGase F for 4 h. Right-hand panel: Western blotting (WB) after treatment with PNGase F for 1 h or 4 h. (C) Purified eukaryotic CD147 was treated with or without Endo F3 and then analysed by SDS/PAGE and Western blotting. Left-hand panel: Coomassie Blue staining after treatment with Endo F3 for 24 h. Right-hand panel: Western blotting after treatment with Endo F3 for 12 h or 24 h. (D) Purified native eukaryotic CD147 was blotted using lectin PHA-L, suggesting that the mature form of CD147 contains β1,6-branched polylactosamines, which are specifically catalysed by GnT-V. Immunoblotting of native eukaryotic CD147 was used as a molecular mass reference. Molecular masses are indicated to the left-hand side of the Western blots in kDa. (E) N-linked glycan profiling of native eukaryotic CD147. The proposed structures of N-linked glycans of CD147 by NSI-MS are presented over the NSI-MS spectra of permethylated N-glycans from CD147. Glycans are detected as singly charged species [M+Na]+. Structural assignments are based on MS/MS fragmentation and known biosynthetic limitations (bottom panel). #, non-glycan signal generated by mass spectrometer. Only high-mannose and the complex-type N-glycans were detected.

Figure 1
Purified native eukaryotic CD147 is N-glycosylated and contains a series of high-mannose and complex-type N-linked glycan structures

(A) Native eukaryotic CD147 was purified from 20 g of human lung cancer tissue by immunoaffinity chromatography. The recombinant extracellular portion of prokaryotic CD147 was also purified. Both eukaryotic and prokaryotic CD147 were resolved by SDS/PAGE. A total of 10 μg of protein were resolved per lane. (B) Purified eukaryotic CD147 was treated with or without PNGase F and then analysed by SDS/PAGE and Western blotting. Left-hand panel: Coomassie Blue staining after treatment with PNGase F for 4 h. Right-hand panel: Western blotting (WB) after treatment with PNGase F for 1 h or 4 h. (C) Purified eukaryotic CD147 was treated with or without Endo F3 and then analysed by SDS/PAGE and Western blotting. Left-hand panel: Coomassie Blue staining after treatment with Endo F3 for 24 h. Right-hand panel: Western blotting after treatment with Endo F3 for 12 h or 24 h. (D) Purified native eukaryotic CD147 was blotted using lectin PHA-L, suggesting that the mature form of CD147 contains β1,6-branched polylactosamines, which are specifically catalysed by GnT-V. Immunoblotting of native eukaryotic CD147 was used as a molecular mass reference. Molecular masses are indicated to the left-hand side of the Western blots in kDa. (E) N-linked glycan profiling of native eukaryotic CD147. The proposed structures of N-linked glycans of CD147 by NSI-MS are presented over the NSI-MS spectra of permethylated N-glycans from CD147. Glycans are detected as singly charged species [M+Na]+. Structural assignments are based on MS/MS fragmentation and known biosynthetic limitations (bottom panel). #, non-glycan signal generated by mass spectrometer. Only high-mannose and the complex-type N-glycans were detected.

Native eukaryotic CD147 exists as oligomers in solution and stimulated production of MMPs more efficiently than non-glycosylated prokaryotic CD147

To investigate whether N-glycans affect the oligomerization of soluble CD147, we first conducted native PAGE to detect the oligomeric status of purified native eukaryotic CD147 and prokaryotic CD147. To our surprise, all of the native eukaryotic CD147 existed as oligomers (~120–175 kDa) in native PAGE, whereas the oligomer formation of prokaryotic CD147 was barely observable in native PAGE (Figure 2A). In fact, we observed that a small fraction of prokaryotic CD147 formed oligomers (~40 kDa) in native PAGE using Western blotting, a more sensitive analysis than Coomassie Blue staining (Figure 2A). Furthermore, even when native eukaryotic CD147 was denatured by boiling, oligomerization was still observed in non-reducing PAGE (Figure 2B). To determine whether N-glycans participate in the oligomerization of soluble eukaryotic CD147, we deglycosylated the native eukaryotic CD147 using PNGase F under native conditions. Both native PAGE and the subsequent Western blot results confirmed that PNGase F treatment dramatically attenuated the oligomerization of eukaryotic CD147 (Figure 2C). To evaluate the involvement of glycosylation in CD147's MMP-inducing activity, native eukaryotic CD147 and recombinant prokaryotic CD147 were added to fibroblasts, and gelatin zymography was used to analyse MMP activity. The results showed that both eukaryotic and prokaryotic CD147 directly induced MMP-2 and MMP-9 secretion by fibroblasts (Figure 2D). Moreover, the stimulatory activity of native eukaryotic CD147 on both MMP-2 and MMP-9 production (P<0.01) was more significant than that of prokaryotic CD147 (P<0.05) at the same concentration (5 μg/ml). These results suggest that maintenance of a native advanced conformation is essential for the activity of CD147.

Native eukaryotic CD147 exists as oligomers in solution and stimulates production of MMPs more efficiently than non-glycosylated prokaryotic CD147

Figure 2
Native eukaryotic CD147 exists as oligomers in solution and stimulates production of MMPs more efficiently than non-glycosylated prokaryotic CD147

(A) Purified native eukaryotic CD147 and recombinant prokaryotic CD147 were subjected to native PAGE followed by Western blotting. A total of 10 μg of protein was loaded in each lane for native PAGE, and 5 μg of protein was loaded in each lane for Western blotting. (B) Purified native eukaryotic CD147 (with or without boiling) was resolved on SDS/PAGE (10% gel) under non-reducing conditions. Proteins were stained with Coomassie Blue. (C) Purified native eukaryotic CD147 was treated in the presence of PNGase F under native conditions for 24 h. The oligomeric status of native eukaryotic CD147 with and without PNGase F treatment was then analysed by native PAGE and Western blotting. Molecular masses are indicated to the side of the Western blots in kDa. (D) Fibroblasts were cultured in the presence of native eukaryotic CD147 or recombinant prokaryotic CD147 at the indicated doses for 20 h, and medium samples were then collected and analysed for MMP activity by gelatin zymography. Upper panel: a representative image. Lower panels: densitometric analysis of MMP-2 and MMP-9 activity. Results are means+S.D. (n=3) *P<0.05, compared with control. **P<0.01, compared with control. Statistical significance was determined using the Prism 5 software for Windows. conc, concentration.

Figure 2
Native eukaryotic CD147 exists as oligomers in solution and stimulates production of MMPs more efficiently than non-glycosylated prokaryotic CD147

(A) Purified native eukaryotic CD147 and recombinant prokaryotic CD147 were subjected to native PAGE followed by Western blotting. A total of 10 μg of protein was loaded in each lane for native PAGE, and 5 μg of protein was loaded in each lane for Western blotting. (B) Purified native eukaryotic CD147 (with or without boiling) was resolved on SDS/PAGE (10% gel) under non-reducing conditions. Proteins were stained with Coomassie Blue. (C) Purified native eukaryotic CD147 was treated in the presence of PNGase F under native conditions for 24 h. The oligomeric status of native eukaryotic CD147 with and without PNGase F treatment was then analysed by native PAGE and Western blotting. Molecular masses are indicated to the side of the Western blots in kDa. (D) Fibroblasts were cultured in the presence of native eukaryotic CD147 or recombinant prokaryotic CD147 at the indicated doses for 20 h, and medium samples were then collected and analysed for MMP activity by gelatin zymography. Upper panel: a representative image. Lower panels: densitometric analysis of MMP-2 and MMP-9 activity. Results are means+S.D. (n=3) *P<0.05, compared with control. **P<0.01, compared with control. Statistical significance was determined using the Prism 5 software for Windows. conc, concentration.

Expression of CD147 on the cell surface and in conditioned medium is in the highly glycosylated mature form in both SMMC-7721 and HepG2 hepatoma tumour cells

To investigate whether the glycosylated forms of CD147 was related to its plasma membrane expression, we separated plasma membrane proteins from SMMC-7721 and HepG2 hepatoma tumour cells using a plasma membrane protein extraction kit and detected the expression of CD147 by Western blotting assay. As shown in Figure 3(A), both the HG-CD147 (~55 kDa) and the LG-CD147 (~40 kDa) were detected in the total cell lysate, whereas only the mature form of CD147 (~55 kDa) was detected in the plasma membrane fraction. The expression of soluble CD147 in the conditioned medium of SMMC-7721 and HepG2 hepatoma tumour cells was also analysed, and we found that, although the expression level of secreted CD147 in the conditioned medium was rather low compared with that in the cell lysate, the glycosylated form was also the mature form, HG-CD147 (~55 kDa; Figure 3B). From these results, we presumed that the glycosylation may be essential for CD147's translocation to the cell surface and secretion into the conditioned medium.

Expression of CD147 on the cell surface and in conditioned medium is as the HG mature form in both SMMC-7721 and HepG2 hepatoma tumour cells

Figure 3
Expression of CD147 on the cell surface and in conditioned medium is as the HG mature form in both SMMC-7721 and HepG2 hepatoma tumour cells

(A) Western blot showing CD147 in total cell lysate and plasma membrane fractions of SMMC-7721 and HepG2 hepatoma tumour cells. The plasma membrane (PM) protein fraction was extracted using a plasma membrane protein extraction kit (BioVision). The results are representative of three experiments with similar results. (B) Western blot showing CD147 in 10× concentrated conditioned medium from SMMC-7721 and HepG2 hepatoma tumour cells. Total cell lysate from SMMC-7721 cells was used as a molecular mass reference. Molecular masses are indicated to the left-hand side of the Western blots in kDa.

Figure 3
Expression of CD147 on the cell surface and in conditioned medium is as the HG mature form in both SMMC-7721 and HepG2 hepatoma tumour cells

(A) Western blot showing CD147 in total cell lysate and plasma membrane fractions of SMMC-7721 and HepG2 hepatoma tumour cells. The plasma membrane (PM) protein fraction was extracted using a plasma membrane protein extraction kit (BioVision). The results are representative of three experiments with similar results. (B) Western blot showing CD147 in 10× concentrated conditioned medium from SMMC-7721 and HepG2 hepatoma tumour cells. Total cell lysate from SMMC-7721 cells was used as a molecular mass reference. Molecular masses are indicated to the left-hand side of the Western blots in kDa.

Glycosylation mutants were retained in the ER and improper intracellular accumulation of CD147 impaired its MMP-inducing activity

To evaluate how the removal of individual N-glycans affects the functional expression of CD147, we produced N-glycosylation- defective mutants by substituting glutamine for asparagine through site-directed mutagenesis of the three predicted N-glycosylation sites. EGFP (enhanced green fluorescent protein)-tagged WT (wild-type) CD147 and its glycosylation mutants were transfected into SMMC-7721 cells. As shown in Figure 4(A), expression of EGFP–CD147 in both N152Q and N44Q/N152Q/N186Q were present at a lower level in total cell lysates after transfection than in the WT. Then we measured the expression levels of WT and mutated CD147 at the cell surface. As shown in Figure 4(B), among the single site mutated forms, either the N44Q or the N186Q N-glycan mutant displayed a similar cell-surface expression level compared with WT CD147, whereas the N152Q mutant exhibited a reduction in cell-surface expression compared with WT CD147. To confirm further the altered cell-surface expression of mutated CD147, the subcellular distributions of WT and glycosylation mutants were examined using confocal laser-scanning microscopy. Figure 4(C) reveals that both N44Q and N186Q were expressed at the cell surface at levels similar to those of WT CD147. However, the N152Q and N44Q/N152Q/N186Q mutants accumulated in the ER, as demonstrated by co-localization of EGFP-tagged CD147 with ER-Tracker Blue-White DPX (p-xylenebispyridinium bromide), an ER marker. Previous studies have shown that a highly conserved UPR (unfolded protein response) signalling pathway is activated by ER stress in response to misfolded protein accumulation [30]. To determine whether abnormal accumulation of the N152Q and N44Q/N152Q/N186Q mutants was able to induce UPR in SMMC-7721 cells, we performed Western blotting to detect the expression of ER stress markers. We found that the UPR was activated in N152Q and N44Q/N152Q/N186Q mutants, as indicated by elevated levels of GRP78/BiP (Figure 4D). Tunicamycin, a known inducer of ER stress, was used as a control. In addition, we used the C/EBP homologous protein (CHOP, also known as GADD153), a transcription factor that is activated at multiple levels during ER stress, as another marker of UPR activation [31,32]. The results showed that the CHOP expression level was also up-regulated in cells expressing the N152Q and N44Q/N152Q/N186Q mutants, similar to what we observed in cells treated with tunicamycin (Figure 4D). To determine whether the reduced functional expression of CD147 with defective N-glycosylation resulted in a reduction in MMP secretion, we analysed MMP activity and expression in glycosylation mutants. Figure 4(E) shows that the removal of the glycans on Asn152 and all three glycosylation sites on CD147 in SMMC-7721 cells significantly suppressed both MMP-9 and MMP-2 activity in a co-culture system (P<0.01). In addition, the protein expression of MMP-2 was also suppressed in SMMC-7721 cells transfected with N152Q and N44Q/N152Q/N186Q mutants compared with WT CD147 (Figure 4E).

Glycosylation mutants were retained in the ER and improper intracellular accumulation of CD147 impaired its MMP-inducing activity

Figure 4
Glycosylation mutants were retained in the ER and improper intracellular accumulation of CD147 impaired its MMP-inducing activity

(A) Western blot (WB) of WT CD147–EGFP and its three glycosylation mutants. Cell lysates from SMMC-7721 cells expressing WT CD147–EGFP or its glycosylation mutants, N44Q, N152Q and N186Q, were analysed by Western blotting with an anti-GFP (green fluorescent protein) antibody. (B) Cell-surface expression of CD147 on SMMC-7721 cells transfected with WT CD147–EGFP or the glycosylation mutants. Plasma membrane (PM) proteins were extracted using the plasma membrane protein extraction kit. Flotillin-1 was used as loading control to show equal amount of plasma membrane proteins in each lane. Molecular masses are indicated to the left-hand side of the Western blots in kDa. (C) Co-localization analysis of WT CD147–EGFP or its glycosylation mutants with the ER. SMMC-7721 cells were transiently transfected with WT CD147–EGFP or the glycosylation mutants and analysed by confocal microscopy. CD147–EGFP was green, and the ER was stained with ER-Tracker DPX (red). (D) Cells expressing WT CD147–EGFP, CD147 (N152Q)–EGFP and CD147 (N44Q/N152Q/N186Q)–EGFP were analysed by Western blotting to determine the levels of the ER-localized chaperone protein GRP78 and the pro-apoptotic ER stress indicator protein CHOP. Cells in the presence of 5 μg/ml tunicamycin for 24 h were used as an ER stress-induced control. The total protein amount was 15 μg per lane. Tubulin was employed as a loading control. (E) WT CD147–EGFP, CD147 (N152Q)–EGFP and CD147 (N44Q/N152Q/N186Q)–EGFP were transiently transfected into SMMC-7721 cells. The transfected cells were then co-cultured with fibroblasts, and conditioned medium samples were analysed by gelatin zymography to detect MMP-9 and MMP-2 activity. Upper panel: a representative image. Middle panel: Densitometric analysis of MMP-2 and MMP-9 activity. Results are means+S.D. (n=3) **P<0.01, compared with WT. Bottom panel: the protein expression of MMP-2 in glycosylation mutants was also determined by Western blotting. Tubulin was employed as a loading control.

Figure 4
Glycosylation mutants were retained in the ER and improper intracellular accumulation of CD147 impaired its MMP-inducing activity

(A) Western blot (WB) of WT CD147–EGFP and its three glycosylation mutants. Cell lysates from SMMC-7721 cells expressing WT CD147–EGFP or its glycosylation mutants, N44Q, N152Q and N186Q, were analysed by Western blotting with an anti-GFP (green fluorescent protein) antibody. (B) Cell-surface expression of CD147 on SMMC-7721 cells transfected with WT CD147–EGFP or the glycosylation mutants. Plasma membrane (PM) proteins were extracted using the plasma membrane protein extraction kit. Flotillin-1 was used as loading control to show equal amount of plasma membrane proteins in each lane. Molecular masses are indicated to the left-hand side of the Western blots in kDa. (C) Co-localization analysis of WT CD147–EGFP or its glycosylation mutants with the ER. SMMC-7721 cells were transiently transfected with WT CD147–EGFP or the glycosylation mutants and analysed by confocal microscopy. CD147–EGFP was green, and the ER was stained with ER-Tracker DPX (red). (D) Cells expressing WT CD147–EGFP, CD147 (N152Q)–EGFP and CD147 (N44Q/N152Q/N186Q)–EGFP were analysed by Western blotting to determine the levels of the ER-localized chaperone protein GRP78 and the pro-apoptotic ER stress indicator protein CHOP. Cells in the presence of 5 μg/ml tunicamycin for 24 h were used as an ER stress-induced control. The total protein amount was 15 μg per lane. Tubulin was employed as a loading control. (E) WT CD147–EGFP, CD147 (N152Q)–EGFP and CD147 (N44Q/N152Q/N186Q)–EGFP were transiently transfected into SMMC-7721 cells. The transfected cells were then co-cultured with fibroblasts, and conditioned medium samples were analysed by gelatin zymography to detect MMP-9 and MMP-2 activity. Upper panel: a representative image. Middle panel: Densitometric analysis of MMP-2 and MMP-9 activity. Results are means+S.D. (n=3) **P<0.01, compared with WT. Bottom panel: the protein expression of MMP-2 in glycosylation mutants was also determined by Western blotting. Tubulin was employed as a loading control.

Overexpression of GnT-V resulted in an elevated level of CD147 on plasma membrane and in cell-conditioned medium, thereby increasing the induction of MMPs

To determine the functional significance of aberrant β1,6-branches on CD147, we established SMMC-7721 tumour cells stably transfected with GnT-V. A lectin blot using PHA-L confirmed an increase in the production of GnT-V in cell lysates and on CD147 (Figure 5A). The protein expression of CD147 was then analysed. In addition to GnT-V transfection, cells were treated with tunicamycin, an inhibitor of N-acetylglucosamine transferases [33]. As shown in Figure 5(B), the expression of the HG form of CD147 was significantly elevated in GnT-V transfectants. Tunicamycin treatment affected both LG and HG forms. The LG form completely disappeared, and the level of the HG form was greatly diminished. The new 27 kDa band is consistent with the size of the core protein. Then we detected the CD147 expression at the mRNA level, at the protein level on the plasma membrane and in the cell-conditioned medium after GnT-V transfection and swainsonine treatment. Swainsonine inhibits Golgi α-mannosidase II activity and thus eliminates the substrate for GnT-V and prevents β1,6-branching [2]. As shown in Figure 5(C), CD147 mRNA transcription levels were not altered after GnT-V transfection or swainsonine treatment. The Western blot in Figures 5(D) and 5(E) reveals that overexpression of GnT-V resulted in increased CD147 on the plasma membrane and in cell-conditioned medium compared with the mock cells, whereas swainsonine treatment suppressed this process. Immunofluorescence staining (shown in Figure 5F) revealed that, in GnT-V transfected SMMC-7721 cells, CD147 cell-surface staining was increased compared with the mock cells. These results verified that β1,6-branches are crucial for CD147 to translocate to the plasma membrane. Then gelatin zymography was used to determine whether the introduction of GnT-V resulted in altered CD147-induced MMPs activity. As shown in Figure 5(G), overexpression of GnT-V dramatically increased MMP-2 activity in SMMC-7721 tumour cells compared with mock cells (panel a). Since GnT-V transfection not only increases the β1,6-GlcNAc branching on CD147, all proteins that contain β1,6 branching will be affected. We conducted further experiments to determine whether the β1,6-GlcNAc branching on CD147 was involved in increased MMP-2 activity caused by GnT-V transfection. The second panel of Figure 5(G) shows that, after transient overexpression of CD147 in mock and GnT-V-overexpressing 7721 cells, the MMP-2 activity of both GnT-V-overexpressing cells and mock cells was enhanced compared with normal 7721 transfectants. Knockdown of CD147 (third panel of Figure 5G) inhibited MMP-2 activity of GnT-V-overexpressing cells and mock cells compared with normal 7721 transfectants, demonstrating that aberrant β1,6-branches on CD147 is crucial for the induction of MMPs in SMMC-7721 HCC cells.

Overexpression of GnT-V resulted in an elevated level of CD147 at the plasma membrane and in cell-conditioned medium, thereby increasing the induction of MMPs

Figure 5
Overexpression of GnT-V resulted in an elevated level of CD147 at the plasma membrane and in cell-conditioned medium, thereby increasing the induction of MMPs

(A) Stable transfection of GnT-V in SMMC-7721 cells was confirmed by lectin blot (LB) analysis using PHA-L to detect the β1,6-branched polylactosamines after GnT-V transfection. Left-hand panel: the production of GnT-V in SMMC-7721 cell lysates. Right-hand panel: the production of GnT-V on CD147. (B) Western blotting (WB) analysis to evaluate protein expression of CD147 after GnT-V transfection and tunicamycin treatment (5 μg/ml for 20 h). (CE) RT–PCR and Western blotting analysis evaluating CD147 expression at the mRNA transcription (C), cell-surface (PM, plasma membrane) protein expression of CD147 (D), and secreted CD147 in 10× concentrated conditioned medium (E) after GnT-V transfection and swainsonine treatment (1 μg/ml for 20 h) of SMMC-7721 cells. Plasma membrane proteins were extracted using the plasma membrane protein extraction kit. Flotillin-1 was used as loading control to show equal amount of plasma membrane proteins in each lane. The PVDF membrane was stained with Ponceau S to show the same amount of protein in 10× concentrated conditioned medium in each lane (lower panel). Molecular masses are indicated to the left-hand side of the Western blots in kDa. (F) Immunofluorescence labelling of CD147 in mock cells (left) and GnT-V-transfected cells (right). Mock cells and GnT-V-transfected cells were fixed, subjected to immunofluorescence staining, and analysed by confocal microscopy for CD147 (red). Nuclei were stained with DAPI (blue). (G) Upper panel: gelatin zymography analysis to detect the MMP-2 activity in the conditioned medium of SMMC-7721 tumour cells after GnT-V transfection (7721). 7721-CD147: effects of overexpression of CD147 on MMP-2 activity in mock and GnT-V-transfected 7721 cells. The pcDNA3-CD147 plasmid was transiently transfected to mock 7721 and GnT-V transfected 7721 cells, and cell-conditioned medium samples were analysed by gelatin zymography to detect MMP-2 activity. 7721-siCD147: effects of knockdown of CD147 on MMP-2 activity in mock and GnT-V-transfected 7721 cells. The results in 7721-CD147 and 7721-siCD147 demostrate that the β1,6-GlcNAc branching on CD147 was involved in increased MMP-2 activity caused by GnT-V transfection. The zymographies show a representative image. Lower panel: densitometric analysis of MMP-2 activity. Results are means+S.D. (n=3).

Figure 5
Overexpression of GnT-V resulted in an elevated level of CD147 at the plasma membrane and in cell-conditioned medium, thereby increasing the induction of MMPs

(A) Stable transfection of GnT-V in SMMC-7721 cells was confirmed by lectin blot (LB) analysis using PHA-L to detect the β1,6-branched polylactosamines after GnT-V transfection. Left-hand panel: the production of GnT-V in SMMC-7721 cell lysates. Right-hand panel: the production of GnT-V on CD147. (B) Western blotting (WB) analysis to evaluate protein expression of CD147 after GnT-V transfection and tunicamycin treatment (5 μg/ml for 20 h). (CE) RT–PCR and Western blotting analysis evaluating CD147 expression at the mRNA transcription (C), cell-surface (PM, plasma membrane) protein expression of CD147 (D), and secreted CD147 in 10× concentrated conditioned medium (E) after GnT-V transfection and swainsonine treatment (1 μg/ml for 20 h) of SMMC-7721 cells. Plasma membrane proteins were extracted using the plasma membrane protein extraction kit. Flotillin-1 was used as loading control to show equal amount of plasma membrane proteins in each lane. The PVDF membrane was stained with Ponceau S to show the same amount of protein in 10× concentrated conditioned medium in each lane (lower panel). Molecular masses are indicated to the left-hand side of the Western blots in kDa. (F) Immunofluorescence labelling of CD147 in mock cells (left) and GnT-V-transfected cells (right). Mock cells and GnT-V-transfected cells were fixed, subjected to immunofluorescence staining, and analysed by confocal microscopy for CD147 (red). Nuclei were stained with DAPI (blue). (G) Upper panel: gelatin zymography analysis to detect the MMP-2 activity in the conditioned medium of SMMC-7721 tumour cells after GnT-V transfection (7721). 7721-CD147: effects of overexpression of CD147 on MMP-2 activity in mock and GnT-V-transfected 7721 cells. The pcDNA3-CD147 plasmid was transiently transfected to mock 7721 and GnT-V transfected 7721 cells, and cell-conditioned medium samples were analysed by gelatin zymography to detect MMP-2 activity. 7721-siCD147: effects of knockdown of CD147 on MMP-2 activity in mock and GnT-V-transfected 7721 cells. The results in 7721-CD147 and 7721-siCD147 demostrate that the β1,6-GlcNAc branching on CD147 was involved in increased MMP-2 activity caused by GnT-V transfection. The zymographies show a representative image. Lower panel: densitometric analysis of MMP-2 activity. Results are means+S.D. (n=3).

DISCUSSION

Glycosylation is an important post-translational protein modification. As more than 50% of all eukaryotic proteins are glycosylated, it is not possible to understand the functions of many proteins without considering the post-translational modifications. CD147, a tumour-associated antigen that is highly expressed on the cell surface of various tumours, is a potential target for cancer diagnosis and therapy. A significant biochemical property of CD147 is its high level of glycosylation. In the present study, we resolved the N-linked glycan profiling of CD147 and confirmed the importance of N-glycosylation for CD147's activity and maturation process.

We report the first mass spectrometric structural determinations of N-glycans of CD147 from human lung cancer tissue. The results show that purified native CD147 displayed high-mannosetype and bi-antennary complex-type oligosaccharides. This was consistent with the results of Yu et al. [17], who also found CD147 from human hepatoma cells contains complex-type carbohydrate (sensitive to PNGase F) and high-mannose-type carbohydrate (sensitive to Endo H). In addition, we observed a high percentage of core fucose (28.8%) in the purified CD147. Core fucosylation (α1,6-fucosylation) of glycoproteins is widely distributed in mammalian tissues and is altered under pathological conditions, such as tumorigenesis of liver, lung and stomach [34]. Serum α-fetoprotein was reported to be core fucosylated in patients with hepatoma and has thus been employed as an early implication of diagnosis [35]. Also, core fucosylation is highly associated with functions of cell-adhesion molecules such as E-cadherin and integrins [36,37]. We found that CD147 is also a cell-surface carrier of core fucose in lung cancer tissue, but its role in tumour metastasis needs to be determined further.

Many biological studies have emphasized the critical role of oligomerization in the multiple functions of CD147. Studies have shown that homo-oligomers of CD147 are present in chicken and mouse tissues [21,38]. Further research strongly suggests that the oligomerization-dependent biological functions of CD147 requires both cis- and trans-homophilic interactions [19]. However, most of the previous studies did not consider the role of glycosylation, and all of these studies suggest that non-glycosylated CD147 forms dimers in solution or in crystals [20,24]. The present study indicates that, although CD147 can form oligomers in a glycan-independent fashion at a low level, carbohydrate chains enhance the oligomerization of native eukaryotic CD147. One possible explanation is that the glycans on CD147 preserve the stability of the tertiary and quaternary protein structure and help maintain an active molecular conformation. The present functional study demostrated that native glycosylated CD147 stimulated production of MMPs more efficiently than non-glycosylated prokaryotic CD147. Collectively, we supposed that maintenance of a native advanced conformation is essential for CD147's activity.

As a transmembrane protein, mature CD147 on the cell surface is considered capable of exerting its biological function. Analysis of plasma membrane proteins extracted from SMMC-7721 cells and HepG2 cells indicated that only the mature glycosylated form of CD147 could be found on the cell surface. This result was consistent with a previous study in COS-7 cells by cell-surface biotinylation [21]. Although only a small fraction (2–3%) of active CD147 was released from the surface of tumour cells into the conditioned medium, the secreted soluble CD147 acted as an inducer to stimulate MMPs from surrounding stromal cells and tumour cells themselves. Therefore we further analysed CD147 secreted into the culture medium by tumour cells. Not surprisingly, only the mature glycosylated form of CD147 was detected in the conditioned medium. These results raised a question: is the process of glycosylation important for CD147 translocation to the cell surface?

A previous study in our laboratory resolving the crystal structure of CD147 provided the proof of three potential N-linked glycosylation sites of CD147: Asn44, lying at the end of strand B, and the other two glycosylation sites, Asn152 and Asn186, are located at the middle of the C’D loop and strand F [24]. Mutation of the N-linked glycosylation site in the present study caused a decrease in molecular mass, confirming that all three potential glycosylated sites could be glycosylated. Interestingly, we also found that, among the three N-glycosylation sites, Asn152 is crucial for the cell-surface expression of CD147. Analysis of subcellular localization showed that the N152Q mutant lacking N-glycosylation sites is retained in the ER compartment. Most proteins synthesized in the rough ER are glycosylated by the addition of a common N-linked oligosaccharide (composed of N-acetylglucosamine, mannose and glucose and containing a total of 14 sugars) [39]. Chaperones, such as calnexin, calreticulin and BiP, interact with this precursor oligosaccharide, which is critical for proper protein folding and ER quality control [40,41]. Our results of the present study suggested that initial N-glycans on Asn152 play vital roles in the quality control of CD147 in the ER. The possible explanation might be that Asn152 is located in a specific domain of the protein, which interferes with the protein folding; or the N152Q mutation disrupts the interaction between CD147 and partner proteins that assist protein folding. Furthermore, we presume that misfolded CD147 retainined in the ER induces an UPR [42,43]. The UPR is characterized by ER chaperones such as GRP78, which are up-regulated in an attempt to restore ER homoeostasis [44,45]. Prolonged activation of this pathway ultimately leads to apoptosis. An elevated level of the chaperone protein GRP78 and the pro-apoptotic indicator protein CHOP after N152Q mutation confirm the induction of UPR in response to accumulation of misfolded CD147. Intriguingly, the sequence alignment among CD147 from various species revealed that all three glycosylated sites of CD147 are highly conserved across species [24], suggesting unknown functions for the remaining two sites.

In vivo and in vitro studies indicate that β1,6-GlcNAc branching is a key structure associated with cancer metastasis [46,47]. Specifically, increased β1,6-branching glycans are highly associated with biological functions of cell adhesion molecules such as E-cadherin and integrins, thereby affecting cancer metastasis (reviewed in [48]). In the present study, we found evidence that β1,6-GlcNAc branching is crucial for the function of an additional cell-surface molecule, CD147. Our results showed that overexpression of GnT-V elevated the levels of the fully glycosylated mature form of CD147, resulting in the up-regulation of the expression of CD147 on the cell surface. However, it should be noted not only that GnT-V transfection increased the β1,6-GlcNAc branching on CD147, but also that all the proteins that contain β1,6-branching will be affected. Kim et al. [49] have demonstrated that, in GnT-V-overexpressing WiDr cells, the aberrantly glycosylated TIMP-1 (tissue inhibitor of matrix metalloproteinase-1) showed weaker inhibition on both MMP-2 and MMP-9, and was closely associated with cancer cell invasion and metastasis. Our results are therefore consistent with the phenotypic changes in GnT-V-overexpressing cells. Although numerous studies have demostrated that up-regulation of CD147 on tumour tissue was closely related with cancer progression, the effects of CD147 are taken into account only in terms of ‘quantity’ without consideration of ‘quality’. The results from the present study implied that β1,6-branching is a crucial alteration on N-glycans of CD147 in HCC cells, which could be a possible marker for tumour metastasis.

In summary, the present study provides evidence of the critical role that N-linked glycans on Asn152 play in determining CD147 expression and function. On the basis of the glycan structure, we revealed aberrant core fucose and β1,6-GlcNAc branching as cancer-associated glycans present on CD147. We confirmed further that β1,6-branching of CD147 reinforces the degradation of the extracellular matrix, creating opportunities for the development of targeted therapies for carcinoma.

Abbreviations

     
  • BiP

    immunoglobulin heavy-chain-binding protein

  •  
  • CHOP

    C/EBP (CCAAT/enhancer-binding protein)-homologous protein

  •  
  • DAPI

    4′,6-diamidino-2-phenylindole

  •  
  • DPX

    p-xylenebispyridinium bromide

  •  
  • E-cadherin

    epithelial cadherin

  •  
  • ECL

    enhanced chemiluminescence

  •  
  • EGFP

    enhanced green fluorescent protein

  •  
  • EMMPRIN

    extracellular matrix metalloproteinase inducer

  •  
  • Endo F3

    endoglycosidase F3

  •  
  • ER

    endoplasmic reticulum

  •  
  • GADD153

    growth-arrest and DNA-damage-inducible protein 153

  •  
  • GnT-V

    N-acetylglucosaminyltransferase V

  •  
  • GRP78

    glucose-regulated protein 78

  •  
  • HCC

    hepatocellular carcinoma

  •  
  • HG

    high-glycosylated

  •  
  • HRP

    horseradish peroxidase

  •  
  • LG

    low-glycosylated

  •  
  • mAb

    monoclonal antibody

  •  
  • MMP

    matrix metalloproteinase

  •  
  • MS/MS

    tandem MS

  •  
  • NSI-MS

    nanospray ionization MS

  •  
  • OG

    N-octyl-β-D-glucopyranoside

  •  
  • PHA-L

    phytohaemagglutinin-L

  •  
  • PNGase F

    peptide N-glycosidase F

  •  
  • UPR

    unfolded protein response

  •  
  • WT

    wild-type

AUTHOR CONTRIBUTION

Wan Huang, Wen-Juan Luo, Ping Zhu and Juan Tang contributed equally to this work. Jian-Li Jiang and Zhi-Nan Chen are corresponding authors. Wan Huang and Juan Tang conducted most of the experiments and finished the paper. Wen-Juan Luo and Ping Zhu designed the experiments and helped with data analysis. Xiao-Ling Yu instructed the purification of eukaryotic CD147 by immunoaffinity chromatography and designed the site mutation of glycosylated sites. Hong-Yong Cui purified the prokaryotic CD147. Bin Wang helped with freeze-drying of eukaryotic CD147. Yang Zhang helped to transport the protein samples to the University of Georgia. Jian-Li Jiang and Zhi-Nan Chen designed the study and provided helpful discussion.

We thank the Complex Carbohydrate Research Center at the University of Georgia for providing the analytical services to characterize N-linked glycan profiling of CD147. We thank Professor Hua-Bei Guo at University of Georgia for providing the human pcDNA3/GnT-V plasmid. We thank Dr James Griffin for a critical reading of the paper prior to submission.

FUNDING

This study was supported by the National Natural Science Foundation of China [grant numbers 81030058 and 30671099], by the National Basic Research Program of China [grant number 2009CB521705] and the National Science and Technology Major Project [grant numbers 2012ZX10002015 and 2013ZX09301301].

References

References
1
Dube
D. H.
Bertozzi
C. R.
Glycans in cancer and inflammation: potential for therapeutics and diagnostics
Nat. Rev. Drug Discovery
2005
, vol. 
4
 (pg. 
477
-
488
)
2
Dennis
J. W.
Granovsky
M.
Warren
C. E.
Glycoprotein glycosylation and cancer progression
Biochim. Biophys. Acta
1999
, vol. 
1473
 (pg. 
21
-
34
)
3
Pinho
S. S.
Reis
C. A.
Paredes
J.
Magalhaes
A. M.
Ferreira
A. C.
Figueiredo
J.
Xiaogang
W.
Carneiro
F.
Gartner
F.
Seruca
R.
The role of N-acetylglucosaminyltransferase III and V in the post-transcriptional modifications of E-cadherin
Hum. Mol. Genet.
2009
, vol. 
18
 (pg. 
2599
-
2608
)
4
Ihara
S.
Miyoshi
E.
Ko
J. H.
Murata
K.
Nakahara
S.
Honke
K.
Dickson
R. B.
Lin
C. Y.
Taniguchi
N.
Prometastatic effect of N-acetylglucosaminyltransferase V is due to modification and stabilization of active matriptase by adding β1-6 GlcNAc branching
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
16960
-
16967
)
5
Gu
J.
Isaji
T.
Sato
Y.
Kariya
Y.
Fukuda
T.
Importance of N-glycosylation on α5β1 integrin for its biological functions
Biol. Pharm. Bull.
2009
, vol. 
32
 (pg. 
780
-
785
)
6
Guo
H. B.
Lee
I.
Kamar
M.
Pierce
M.
N-acetylglucosaminyltransferase V expression levels regulate cadherin-associated homotypic cell–cell adhesion and intracellular signaling pathways
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
52412
-
52424
)
7
Kataoka
H.
DeCastro
R.
Zucker
S.
Biswas
C.
Tumor cell-derived collagenase-stimulatory factor increases expression of interstitial collagenase, stromelysin, and 72-kDa gelatinase
Cancer Res.
1993
, vol. 
53
 (pg. 
3154
-
3158
)
8
Riethdorf
S.
Reimers
N.
Assmann
V.
Kornfeld
J. W.
Terracciano
L.
Sauter
G.
Pantel
K.
High incidence of EMMPRIN expression in human tumors
Int. J. Cancer
2006
, vol. 
119
 (pg. 
1800
-
1810
)
9
Toole
B. P.
Emmprin (CD147), a cell surface regulator of matrix metalloproteinase production and function
Curr. Top. Dev. Biol.
2003
, vol. 
54
 (pg. 
371
-
389
)
10
Nabeshima
K.
Iwasaki
H.
Koga
K.
Hojo
H.
Suzumiya
J.
Kikuchi
M.
Emmprin (basigin/CD147): matrix metalloproteinase modulator and multifunctional cell recognition molecule that plays a critical role in cancer progression
Pathol. Int.
2006
, vol. 
56
 (pg. 
359
-
367
)
11
Xu
J.
Xu
H. Y.
Zhang
Q.
Song
F.
Jiang
J. L.
Yang
X. M.
Mi
L.
Wen
N.
Tian
R.
Wang
L.
, et al. 
HAb18G/CD147 functions in invasion and metastasis of hepatocellular carcinoma
Mol. Cancer Res.
2007
, vol. 
5
 (pg. 
605
-
614
)
12
Zhao
P.
Zhang
W.
Wang
S. J.
Yu
X. L.
Tang
J.
Huang
W.
Li
Y.
Cui
H. Y.
Guo
Y. S.
Tavernier
J.
, et al. 
HAb18G/CD147 promotes cell motility by regulating annexin II-activated RhoA and Rac1 signaling pathways in hepatocellular carcinoma cells
Hepatology
2011
, vol. 
54
 (pg. 
2012
-
2024
)
13
Jiang
J. L.
Zhou
Q.
Yu
M. K.
Ho
L. S.
Chen
Z. N.
Chan
H. C.
The involvement of HAb18G/CD147 in regulation of store-operated calcium entry and metastasis of human hepatoma cells
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
46870
-
46877
)
14
Tang
J.
Wu
Y. M.
Zhao
P.
Yang
X. M.
Jiang
J. L.
Chen
Z. N.
Overexpression of HAb18G/CD147 promotes invasion and metastasis via α3β1 integrin mediated FAK-paxillin and FAK-PI3K-Ca2+ pathways
Cell. Mol. Life Sci.
2008
, vol. 
65
 (pg. 
2933
-
2942
)
15
Xu
J.
Shen
Z. Y.
Chen
X. G.
Zhang
Q.
Bian
H. J.
Zhu
P.
Xu
H. Y.
Song
F.
Yang
X. M.
Mi
L.
, et al. 
A randomized controlled trial of Licartin for preventing hepatoma recurrence after liver transplantation
Hepatology
2007
, vol. 
45
 (pg. 
269
-
276
)
16
Chen
Z. N.
Mi
L.
Xu
J.
Song
F.
Zhang
Q.
Zhang
Z.
Xing
J. L.
Bian
H. J.
Jiang
J. L.
Wang
X. H.
, et al. 
Targeting radioimmunotherapy of hepatocellular carcinoma with iodine (131I) metuximab injection: clinical phase I/II trials
Int. J. Radiat. Oncol. Biol. Phys.
2006
, vol. 
65
 (pg. 
435
-
444
)
17
Yu
X. L.
Jiang
J. L.
Li
L.
Feng
Q.
Xu
J.
Chen
Z. N.
The glycosylation characteristic of hepatoma-associated antigen HAb18G/CD147 in human hepatoma cells
Int. J. Biochem. Cell Biol.
2006
, vol. 
38
 (pg. 
1939
-
1945
)
18
Tang
W.
Chang
S. B.
Hemler
M. E.
Links between CD147 function, glycosylation, and caveolin-1
Mol. Biol. Cell
2004
, vol. 
15
 (pg. 
4043
-
4050
)
19
Sun
J.
Hemler
M. E.
Regulation of MMP-1 and MMP-2 production through CD147/extracellular matrix metalloproteinase inducer interactions
Cancer Res.
2001
, vol. 
61
 (pg. 
2276
-
2281
)
20
Belton
R. J.
Chen
L.
Mesquita
F. S.
Nowak
R. A.
Basigin-2 is a cell surface receptor for soluble basigin ligand
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
17805
-
17814
)
21
Yoshida
S.
Shibata
M.
Yamamoto
S.
Hagihara
M.
Asai
N.
Takahashi
M.
Mizutani
S.
Muramatsu
T.
Kadomatsu
K.
Homo-oligomer formation by basigin, an immunoglobulin superfamily member, via its N-terminal immunoglobulin domain
Eur. J. Biochem.
2000
, vol. 
267
 (pg. 
4372
-
4380
)
22
Taylor
P. M.
Woodfield
R. J.
Hodgkin
M. N.
Pettitt
T. R.
Martin
A.
Kerr
D. J.
Wakelam
M. J.
Breast cancer cell-derived EMMPRIN stimulates fibroblast MMP2 release through a phospholipase A2 and 5-lipoxygenase catalyzed pathway
Oncogene
2002
, vol. 
21
 (pg. 
5765
-
5772
)
23
Sidhu
S. S.
Mengistab
A. T.
Tauscher
A. N.
LaVail
J.
Basbaum
C.
The microvesicle as a vehicle for EMMPRIN in tumor-stromal interactions
Oncogene
2004
, vol. 
23
 (pg. 
956
-
963
)
24
Yu
X. L.
Hu
T.
Du
JM
Ding
J. P.
Yang
X. M.
Zhang
J.
Yang
B.
Shen
X.
Zhang
Z.
Zhong
W. D.
, et al. 
Crystal structure of HAb18G/CD147: implications for immunoglobulin superfamily homophilic adhesion
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
18056
-
18065
)
25
Liu
Y.
Localization of hepatocellular carcinoma with monoclonal antibodies
Zhonghua Yixue Zazhi
1991
, vol. 
71
 (pg. 
362
-
365
)
26
Anumula
K. R.
Taylor
P. B.
A comprehensive procedure for preparation of partially methylated alditol acetates from glycoprotein carbohydrates
Anal. Biochem.
1992
, vol. 
203
 (pg. 
101
-
108
)
26a
Domon
B.
Costello
C. E.
A systematic nomenclature for carbohydrate fragmentations in FAB-MS/MS spectra of glycoconjugates
Glycoconjugate J.
1988
, vol. 
5
 (pg. 
397
-
409
)
27
Cummings
R. D.
Trowbridge
I. S.
Kornfeld
S.
A mouse lymphoma cell line resistant to the leukoagglutinating lectin from Phaseolus vulgaris is deficient in UDP-GlcNAc: α-D-mannoside β 1,6 N-acetylglucosaminyltransferase
J. Biol. Chem.
1982
, vol. 
257
 (pg. 
13421
-
13427
)
28
Dennis
J. W.
Laferte
S.
Oncodevelopmental expression of –GlcNAcβ1–6Manα1–6Manβ1– branched asparagine-linked oligosaccharides in murine tissues and human breast carcinomas
Cancer Res.
1989
, vol. 
49
 (pg. 
945
-
950
)
29
Taniguchi
N.
Miyoshi
E.
Ko
J. H.
Ikeda
Y.
Ihara
Y.
Implication of N-acetylglucosaminyltransferases III and V in cancer: gene regulation and signaling mechanism
Biochim. Biophys. Acta
1999
, vol. 
1455
 (pg. 
287
-
300
)
30
Bertolotti
A.
Zhang
Y.
Hendershot
L. M.
Harding
H. P.
Ron
D.
Dynamic interaction of BiP and ER stress transducers in the unfolded-protein response
Nat. Cell Biol.
2000
, vol. 
2
 (pg. 
326
-
332
)
31
Wang
X. Z.
Ron
D.
Stress-induced phosphorylation and activation of the transcription factor CHOP (GADD153) by p38 MAP kinase
Science
1996
, vol. 
272
 (pg. 
1347
-
1349
)
32
Marciniak
S. J.
Yun
C. Y.
Oyadomari
S.
Novoa
I.
Zhang
Y.
Jungreis
R.
Nagata
K.
Harding
H. P.
Ron
D.
CHOP induces death by promoting protein synthesis and oxidation in the stressed endoplasmic reticulum
Genes Dev.
2004
, vol. 
18
 (pg. 
3066
-
3077
)
33
King
I. A.
Tabiowo
A.
Effect of tunicamycin on epidermal glycoprotein and glycosaminoglycan synthesis in vitro
Biochem. J.
1981
, vol. 
198
 (pg. 
331
-
338
)
34
Miyoshi
E.
Uozumi
N.
Noda
K.
Hayashi
N.
Hori
M.
Taniguchi
N.
Expression of α1–6 fucosyltransferase in rat tissues and human cancer cell lines
Int. J. Cancer.
1997
, vol. 
72
 (pg. 
1117
-
1121
)
35
Miyoshi
E.
Noda
K.
Yamaguchi
Y.
Inoue
S.
Ikeda
Y.
Wang
W.
Ko
J. H.
Uozumi
N.
Li
W.
Taniguchi
N.
The α1–6-fucosyltransferase gene and its biological significance
Biochim. Biophys. Acta
1999
, vol. 
1473
 (pg. 
9
-
20
)
36
Geng
F.
Shi
B. Z.
Yuan
Y. F.
Wu
X. Z.
The expression of core fucosylated E-cadherin in cancer cells and lung cancer patients: prognostic implications
Cell Res.
2004
, vol. 
14
 (pg. 
423
-
433
)
37
Zhao
Y.
Itoh
S.
Wang
X.
Isaji
T.
Miyoshi
E.
Kariya
Y.
Miyazaki
K.
Kawasaki
N.
Taniguchi
N.
Gu
J.
Deletion of core fucosylation on α3β1 integrin down-regulates its functions
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
38343
-
38350
)
38
Fadool
J. M.
Linser
P. J.
Evidence for the formation of multimeric forms of the 5A11/HT7 antigen
Biochem. Biophys. Res. Commun.
1996
, vol. 
229
 (pg. 
280
-
286
)
39
Kukuruzinska
M. A.
Lennon
K.
Protein N-glycosylation: molecular genetics and functional significance
Crit. Rev. Oral Biol. Med.
1998
, vol. 
9
 (pg. 
415
-
448
)
40
Lederkremer
G. Z.
Glycoprotein folding, quality control and ER-associated degradation
Curr. Opin. Struct. Biol.
2009
, vol. 
19
 (pg. 
515
-
523
)
41
Roth
J.
Zuber
C.
Park
S.
Jang
I.
Lee
Y.
Kysela
K. G.
Le Fourn
V.
Santimaria
R.
Guhl
B.
Cho
J. W.
Protein N-glycosylation, protein folding, and protein quality control
Mol. Cells
2010
, vol. 
30
 (pg. 
497
-
506
)
42
Kozutsumi
Y.
Segal
M.
Normington
K.
Gething
M. J.
Sambrook
J.
The presence of malfolded proteins in the endoplasmic reticulum signals the induction of glucose-regulated proteins
Nature
1988
, vol. 
332
 (pg. 
462
-
464
)
43
Cao
S. S.
Kaufman
R. J.
Unfolded protein response
Curr. Biol.
2012
, vol. 
22
 (pg. 
R622
-
R626
)
44
Lee
A. S.
Mammalian stress response: induction of the glucose-regulated protein family
Curr. Opin. Cell Biol.
1992
, vol. 
4
 (pg. 
267
-
273
)
45
Wilhovsky
S.
Gardner
R.
Hampton
R.
HRD gene dependence of endoplasmic reticulum-associated degradation
Mol. Biol. Cell
2000
, vol. 
11
 (pg. 
1697
-
1708
)
46
Hakomori
S.
Tumor malignancy defined by aberrant glycosylation and sphingo(glyco)lipid metabolism
Cancer Res.
1996
, vol. 
56
 (pg. 
5309
-
5318
)
47
Murata
K.
Miyoshi
E.
Kameyama
M.
Ishikawa
O.
Kabuto
T.
Sasaki
Y.
Hiratsuka
M.
Ohigashi
H.
Ishiguro
S.
Ito
S.
, et al. 
Expression of N-acetylglucosaminyltransferase V in colorectal cancer correlates with metastasis and poor prognosis
Clin. Cancer Res.
2000
, vol. 
6
 (pg. 
1772
-
1777
)
48
Zhao
Y.
Sato
Y.
Isaji
T.
Fukuda
T.
Matsumoto
A.
Miyoshi
E.
Gu
J.
Taniguchi
N.
Branched N-glycans regulate the biological functions of integrins and cadherins
FEBS J.
2008
, vol. 
275
 (pg. 
1939
-
1948
)
49
Kim
Y. S.
Hwang
S. Y.
Kang
H. Y.
Sohn
H.
Oh
S.
Kim
J. Y.
Yoo
J. S.
Kim
Y. H.
Kim
C. H.
Jeon
J. H.
, et al. 
Functional proteomics study reveals that N-acetylglucosaminyltransferase V reinforces the invasive/metastatic potential of colon cancer through aberrant glycosylation on tissue inhibitor of metalloproteinase-1
Mol. Cell. Proteomics
2008
, vol. 
7
 (pg. 
1
-
14
)

Author notes

1

These authors contributed equally to this work.