Malignant transformation is associated with aberrant N-glycosylation, but the role of protein N-glycosylation in cancer progression remains poorly defined. β4-integrin is a major carrier of N-glycans and is associated with poor prognosis, tumorigenesis, and metastasis. Here, N-glycosylation of β4-integrin contributes to the activation of signaling pathways that promote β4-dependent tumor development and progression. Increased expression of β1,6GlcNAc-branched N-glycans was found to be colocalized with β4-integrin in human cutaneous squamous cell carcinoma tissues, and that the β1,6GlcNAc residue was abundant on β4-integrin in transformed keratinocytes. Interruption of β1,6GlcNAc-branching formation on β4-integrin with the introduction of bisecting GlcNAc by N-acetylglucosaminyltransferase III overexpression was correlated with suppression of cancer cell migration and tumorigenesis. N-Glycan deletion on β4-integrin impaired β4-dependent cancer cell migration, invasion, and growth in vitro and diminished tumorigenesis and proliferation in vivo. The reduced abilities of β4-integrin were accompanied with decreased phosphoinositol-3 kinase (PI3K)/Akt signals and were restored by the overexpression of the constitutively active p110 PI3K subunit. Binding of galectin-3 to β4-integrin via β1,6GlcNAc-branched N-glycans promoted β4-integrin–mediated cancer cell adhesion and migration. In contrast, a neutralizing antibody against galectin-3 attenuated β4-integrin N-glycan–mediated PI3K activation and inhibited the ability of β4-integrin to promote cell motility. Furthermore, galectin-3 knockdown by shRNA suppressed β4-integrin N-glycan–mediated tumorigenesis. These findings provide a novel role for N-glycosylation of β4-integrin in tumor development and progression, and the regulatory mechanism for β4-integrin/PI3K signaling via the galectin-3–N-glycan complex.
Implications: N-Glycosylation of β4-integrin plays a functional role in promoting tumor development and progression through PI3K activation via the galectin-3–N-glycan complex. Mol Cancer Res; 16(6); 1024–34. ©2018 AACR.
This article is featured in Highlights of This Issue, p. 921
Integrins are a large family of heterodimeric transmembrane receptors comprising α and β subunits. The extracellular domains of integrin subunits bind to extracellular matrix proteins, such as collagen, fibronectin, and laminin, and the cytoplasmic domains interact with the actin cytoskeleton and signaling molecules. The interaction of integrin with its ligand can activate intracellular signaling and reorganize the actin cytoskeleton (1). Such integrin signaling regulates various cellular functions, such as cell adhesion, migration, and proliferation, not only in normal tissues but also in tumor tissues (1, 2).
The α6β4-integrin (referred to herein as β4-integrin because the β4 subunit only pairs with the α6 subunit) is a receptor for laminin-332 and is an essential component of the hemidesmosome, an anchoring structure in the basal membrane of stratified epithelial cells (2, 3). In contrast to such a static role in normal epithelial cells, β4-integrin was originally identified as a tumor-associated antigen (4, 5). β4-integrin overexpression was found in several types of metastatic cancers, which correlates with poor prognosis and reduced survival (6, 7). Recent studies have shown that β4-integrin is required for tumorigenesis in several in vivo mouse models (8–10). In addition, β4-integrin promotes cell motility, invasion, and proliferation through activation of the phosphoinositol-3-kinase (PI3K) and ERK pathways (10–14).
Glycosylation is the most common posttranscriptional modification of proteins, and it modulates the folding, stability, and function of glycoproteins (15). Malignant transformation is associated with aberrant glycosylation of cell surface proteins, and the structural change of glycans is mainly due to the altered activity and expression of multiple glycosyltransferases in cancer cells. In addition, some glycans are used as tumor markers for cancer diagnosis (16). Overexpression of β1,6GlcNAc-branched N-glycans, which is due to increased activity of N-acetylglucosaminyltransferase V (GnT-V; Fig. 1A), is often found in tumor tissues, and the increase in β1,6GlcNAc-branched N-glycans is directly associated with malignancy and poor prognosis (17, 18). GnT-V knockout mice exhibit reduced mammary tumor growth and metastasis induced by the polyomavirus middle T oncogene (19). Furthermore, introduction of bisecting GlcNAc residues catalyzed by N-acetylglucosaminyltransferase III (GnT-III) into N-glycans, which can disturb further processing and elongation of N-glycans, such as the formation of β1,6GlcNAc-branched structures (Fig. 1A), suppresses cell migration and cancer metastasis (20).
β1,6GlcNAc-branched N-glycans are often further modified by additional sugars to form poly-N-acetyllactosamine (polylactosamine) for the elongation of N-glycans consisting of β-galactoside sugars (Fig. 1A). This polylactosamine structure is a preferred ligand for one of the galectin isoforms, galectin-3 (21). Galectin-3 is widely expressed in epithelial and immune cells, and its expression is associated with cancer aggressiveness and metastasis (22–24). Cross-linking between glycoproteins by the binding of galectin-3 to polylactosamine on proteins regulates diverse cellular functions in cancer cells (25–27). We previously reported that N-glycosylation of β4-integrin plays a crucial role in cell adhesion, migration, and galectin-3 binding in human keratinocyte cells (28). However, the contribution of N-glycosylation of β4-integrin to tumor progression remains poorly defined. In the present study, we investigated the contribution of N-glycosylation to β4-integrin-dependent tumor progression.
Materials and Methods
Antibodies and reagents
The following antibodies were used in this study: rat monoclonal antibodies specific for galectin-3 (M3/38; #sc-2393; Santa Cruz Biotechnology), α6-integrin (GoH3; #sc-19622; Santa Cruz Biotechnology), and β4-integrin (439-9B; #555719; BD Transduction Laboratories); rabbit polyclonal antibodies against human β4-integrin (H101; #sc-9090; Santa Cruz Biotechnology), phospho-Akt (pSer473; #9271; Cell Signaling Technology), and Akt (#9272; Cell Signaling Technology); and mouse monoclonal antibodies against Ki-67 (#610968; BD Transduction Laboratories), and β4-integrin (3E1; #MAB1964; Merck Millipore). Alexa Fluor 488-conjugated leukoagglutinating phytohemagglutinin (L4-PHA; #L-11270), Alexa Fluor 546-conjugated goat anti-rabbit IgG secondary antibody (#A11035) and Alexa Fluor 546-conjugated goat anti-rat IgG secondary antibody (#A11081) were purchased from Thermo Fisher Scientific. Biotinylated L4-PHA (#B-1115) was obtained from Vector Laboratories. Hoechst 33342 (#382065) was obtained from Merck Millipore. Biotinylated erythroagglutinating phytohemagglutinin (E4-PHA; #300425) and biotinylated Sambucus sieboldiana agglutinin (SSA; #300442) were purchased from J-OIL MILLS. Purified human laminin-332 and galectin-3 were prepared as described previously (27, 29).
The human cancer MDA-MB435S cell line, which lacks β4-integrin, was purchased from the American Type Culture Collection. Modified human 293 phoenix cells were received as a gift from Dr. M. Peter Marinkovich (Stanford University, Stanford, CA). The human epidermoid carcinoma cell line A431 (RCB0202) was provided by the RIKEN BRC through the National Bio-Resource Project of the MEXT, Japan. These cells were cultured in DMEM supplemented with 10% FBS, penicillin and streptomycin sulfate at 37°C in a humidified 5% CO2 incubator. Primary human keratinocytes isolated from normal skin (NHK) and Ras/IκB-transformed NHK (a gift from Dr. M. Peter Marinkovich; ref. 8) were cultured in a 50:50 mixture of defined keratinocyte serum-free medium (#10744-019; Life Technologies) and medium 154 (#M-154-500; Life Technologies) with human keratinocyte growth supplement (#S-001-5; Life Technologies) containing penicillin and streptomycin sulfate at 37°C in a humidified 5% CO2 incubator. All cells were passaged fewer than 6 months after purchasing or receiving them for all the experiments and tested for mycoplasma contamination.
Human formalin-fixed paraffin-embedded sections were purchased from BioChain and US Biomax. Specimens were deparaffinized, rehydrated, and immersed in 0.3% hydrogen peroxidase-containing methanol for 20 minutes at room temperature to inactivate the intrinsic peroxidase. After blocking with 5% skim milk for 1 hour at room temperature, the sections were incubated with each biotinylated lectin overnight at 4°C, followed by peroxidase-labeled streptavidin (#426061; Nichirei Corp.) and 3,3′-diaminobenzidine detection using the Histofine DAB substrate kit (#425011; Nichirei Corp.). The slides were then counterstained with Mayer's hematoxylin solution (#131-09665; Wako). Images were obtained using a PROVIS AX-80 microscope (Olympus). For immunofluorescent staining, sections were deparaffinized, rehydrated, and treated with 0.05% protease XXIV in 50 mmol/L Tris-HCl (pH 7.5) for 20 minutes at room temperature. After blocking with 2% BSA in PBS for 1 hour at room temperature, the sections were visualized using a β4-integrin antibody (H101) and Alexa Fluor 488-conjugated L4-PHA. Fluorescence images were obtained using an IX71 fluorescent microscope (Olympus). Lectin reactivity and β4-integrin expression were assessed as follows: positive, ≥10% positive tumor cells; negative, <10% positive tumor cells. For cross-linking inhibitory assay, glass-bottom dishes (#3971-035; IWAKI) were coated with laminin-332 proteins and blocked with 1% BSA in PBS. Cells (5 × 104 cells) were incubated in serum-free medium in the presence of IgG or a functional blocking antibody against galectin-3 for 20 minutes. Then, the cells were plated to the glass-bottom dish and incubated for 1.5 hours at 37°C in a humidified 5% CO2 incubator. The cells were fixed with 4% (w/v) paraformaldehyde in PBS for 10 minutes and blocked with 2% BSA in PBS for 1 hour at room temperature. The fixed cells were stained with a β4-integrin antibody (H101) and Alexa Fluor 488-conjugated anti-rabbit IgG antibody. Fluorescence images were obtained using an A1 confocal microscope (Nikon).
Preparation of cell lysates and immunoprecipitation
Cell lysates were prepared as follows. The cells were washed twice with cold PBS and then lysed with a lysis buffer [1% Triton X-100, 20 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 5 mmol/L EDTA] containing a protease inhibitor cocktail (#25955; Nacalai tesque) and a phosphatase inhibitor cocktail (#07575; Nacalai tesque). After incubation for 10 minutes on ice, the cell lysates were clarified by centrifugation at 15,000 rpm for 10 minutes at 4°C. The resulting supernatant was used in the following experiments. The protein concentration was determined using a protein assay kit (#29449-44; Nacalai tesque). For immunoprecipitation, the primary antibody was added to the supernatant and rotated for 2 hours at 4°C. Then, protein G-Sepharose was added, followed by 3 hours of incubation at 4°C. Immunoprecipitates were washed five times with a washing buffer [50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 2 mmol/L EDTA, 0.2% NP40 (v/v)], suspended in a reducing sample buffer, and heated at 95°C for 5 minutes.
Western blot analyses
Protein samples were resolved by SDS-PAGE under reducing conditions and then transferred to nitrocellulose membranes. The blots were probed with each specific antibody or biotinylated lectin. Immunoreactive bands were detected using an ImmunoStar Zeta (#297-72403; WAKO) or Trident femto-ECL (#GTX14698; GeneTex). Band intensity was calculated using NIH ImageJ software.
Retroviral expression vectors encoding β4-integrin were prepared as previously described (28). Retroviral expression vector encoding active PI3K p110-CAAX was received as a gift from Dr. M. Peter Marinkovich. The cDNA encoding human GnT-III was amplified by PCR using a specific primer set and KOD Plus polymerase (#KOD-201; TOYOBO) for cloning into pENTR-D-TOPO (#K2400-20; Life Technologies) for the Gateway Conversion System according to the manufacturer's instructions. The final construct was recombined from pENTR-D-TOPO to the LZRS blast retroviral vector, including a Gateway cassette, using the LR clonase II Enzyme mix (#11791-020; Life Technologies) by a recombination reaction. The cDNA sequence was verified by sequencing.
Retrovirus vectors were transfected into 293 phoenix cells using FuGENE6 transfection reagent (#11814443001; Roche). After transfection, cells were selected with 5 μg/mL puromycin (#P8833; Sigma-Aldrich). The retrovirus was then produced in 293 phoenix cells. One day before infection, 4 × 105 cells were plated in 6-well plates. After incubation with 5 μg/mL polybrene (#10768-9; Sigma-Aldrich) for 15 minutes, media were exchanged to 3 mL retroviral supernatant and another 5 μg/mL polybrene was added. Plates were centrifuged at 1,200 rpm for 1 hour at 32°C using a Hitachi CR22N centrifuge machine. After centrifuge, the retroviral supernatant was replaced with growth medium and the cells were maintained. To establish a cell line, cells were selected with 5 μg/mL blasticidin S (#203350; Calbiochem).
Flow cytometry analysis
Cells were detached from a 10-cm dish using trypsin with 1 mmol/L EDTA. After quenching trypsinization with a medium that contained 10% FBS, the cells were washed twice with PBS that contained 1 mmol/L EDTA and incubated with a primary antibody or control IgG on ice for 30 minutes. The cells were then washed thrice with PBS that contained 1 mmol/L EDTA, followed by incubation for 15 minutes with the appropriate Alexa Fluor-conjugated secondary antibodies. After washing thrice with PBS containing 1 mmol/L EDTA, the cells were analyzed by flow cytometry using a FACSCalibur and a CellQuest software program (BD Biosciences).
Cell migration assay
Cell migration was assayed using a 24-well chemotaxis chamber (BD Falcon cell culture companion plates; #353504 and 8.0-μm insets; #353097; BD Biosciences). Two hundred microliters of 2.5 × 105 cells/mL (5 × 104 cells in serum-free DMEM) was inoculated into each upper well of a 24-well chemotaxis chamber, and 750 μL of 10% FBS containing DMEM was placed in the bottom chamber to act as a chemoattractant. After incubating for 22 hours, the cells on the upper side of the membrane were removed with a cotton swab, and the cells on the lower side of the membrane were fixed with 4% paraformaldehyde and stained with 0.5% crystal violet in 20% methanol. Random fields were photographed using a phase contrast microscope and the number of migrated cells was counted.
In vivo tumorigenicity assay
All animal studies were performed in accordance with protocols approved by the Fukushima Medical University Animal Care and Use Committee. MDA-MB435S transfectants (1 × 106 cells/mouse) were injected subcutaneously along with Matrigel (#354234; BD Biosciences) into 6-week-old female nude mice. Tumor volume was measured weekly for a total of 6 weeks. Proliferating cells in 5-μm frozen sections were detected with Ki-67 immunofluorescent staining. Proliferation was quantified as the ratio of Ki-67 staining to total nuclear staining.
Cell adhesion assay
Cell adhesion assays were performed as described previously (29). In brief, the wells of a 96-well ELISA plate (#3590; Corning) were coated with 50 μL of indicated concentration of laminin-332 proteins and blocked with 1% BSA in PBS. Cell suspensions in serum-free medium were plated to each well (5 × 104 cells/well) and incubated at 37°C for 20 minutes in a 5% CO2 incubator. After removing the non-adherent cells, the adherent cells were fixed with 25% (w/v) glutaraldehyde for 10 minutes and stained with 0.5% crystal violet in 20% (v/v) methanol for 10 minutes. Random fields were photographed using a phase contrast microscope, and the number of adherent cells was counted.
Cell proliferation assay
Cell suspensions in 1% serum-containing medium were plated to each well of a 96-well ELISA plate (5 × 103 cells/well) and incubated at 37°C for 45 hours in a 5% CO2 incubator. After incubation, the number of growing cells was measured using a cell counting kit-8 (#347-07621; DOJINDO) according to the manufacturer's instructions.
Cell invasion assay
One hundred microliters of Matrigel, diluted to a final concentration of 1.6 mg/mL, was added to the upper chamber of 24-well Transwell plates and dried for 24 hours in a hood. The wells were reconstituted by incubation with 200 μL of serum-free DMEM at 37°C for 1 hour. After removing the medium, 200 μL of 5 × 105 cells/mL (1 × 104 cells in 0.1% BSA containing serum-free DMEM) were inoculated into each upper well of a 24-well chemotaxis chamber, and 750 μL of 10% FBS containing DMEM was placed in the bottom chamber to act as a chemoattractant. After 6 hours of incubation, the cells on the upper side of the membrane were removed with a cotton swab, and the cells on the lower side of the membrane were fixed with 4% paraformaldehyde and stained with 0.5% crystal violet in 20% methanol. Random fields were photographed using a phase contrast microscope and the number of invaded cells was counted.
Lentiviral short hairpin RNA (shRNA)
Lentiviral shRNA clones (Dharmacon RNAi consortium Lentiviral shRNA) targeting galectin-3 (#RHS3979-201759611, clone ID: TRCN0000029304) 5′-ATTGTACTGCAACAAGTGAGC-3′ and control shRNA (#RHS4459) 5′-TACAACAGCCACAACGTCTAT-3′ were purchased from GE Healthcare. These vectors were cotransfected with the packaging vectors into HEK293T cells using the Trans-Lentiviral shRNA Packaging kit (#TLP5912; GE Healthcare) according to the manufacturer's instructions. After incubation for 15 hours, media were exchanged to 5% FBS containing DMEM. After further incubation for 48 hours, viral supernatant was harvested and then centrifuged at 1,600 g for 10 minutes at 4°C. The supernatant was filtrated through a 0.45-μm filter. Lentivirus was infected into MDA-MB435S cells using the same method as retrovirus infection. To establish a cell line, cells were selected with 5 μg/mL puromycin.
Results are given as mean ± SEM and are representative of two or three independent experiments. Statistical comparisons were calculated between two groups using unpaired Student t test and among the groups using one-way or two-way ANOVA followed by a Bonferroni posttest, with GraphPad Prism Version 5.0a software. A P value of <0.05 was considered statistically significant.
β1,6GlcNAc residue on β4-integrin was associated with cell migration and tumorigenesis
Expression of β1,6GlcNAc is associated with metastasis and poor prognosis, whereas bisecting GlcNAc suppresses the effect by inhibition of β1,6GlcNAc-branching formation (20). Human cutaneous squamous cell carcinoma (SCC) showed positive expression of β1,6GlcNAc (52 out of 75) and negative expression of bisecting GlcNAc (72 out of 75) by immunohistochemistry using L4-PHA and E4-PHA lectins (Fig. 1B, n = 75), which is consistent with previous observations that β1,6GlcNAc was more likely to be associated with tumor malignancy (17). As overexpression of β4-integrin in SCC has been reported (6, 30), we next examined the expression patterns of β4-integrin and β1,6GlcNAc in human cutaneous SCC. In skin cancer specimens (n = 37), β4-integrin was abundantly present and colocalized with β1,6GlcNAc-branched N-glycans. Thirty-three β4-integrin positive tumor cells, 33 (100%) colocalized β1,6GlcNAc-branched N-glycans (Fig. 1C). In contrast, β4-integrin was clearly localized in the basement membrane, while the expression of β1,6GlcNAc was almost undetectable in all tested normal human skin samples (n = 7). Taking these results into consideration, we hypothesized that the amount of β1,6GlcNAc-branched N-glycans on β4-integrin increases in cancer cells. To test this hypothesis, we examined whether β1,6GlcNAc was attached to β4-integrin upon transformation of NHK cells. Ras/IκB-mediated transformation of NHK cells increased attachment of β1,6GlcNAc to β4-integrin while the extent of the bisecting GlcNAc modification of β4-integrin was decreased after transformation (Fig. 1D). These results suggest that the β1,6GlcNAc modification of β4-integrin is increased in cancer cells.
GnT-III inhibits the formation of β1,6GlcNAc branching on N-glycans by the introduction of bisecting GlcNAc (Fig. 1A). To address the role of β1,6GlcNAc modification of β4-integrin in tumor malignancy, we used MDA-MB435S cells, which do not express β4-integrin, for preparing stable transfectants retrovirally transduced with control lacZ, human β4-integrin, lacZ with GnT-III, or β4 with GnT-III. The FACS analysis exhibited that GnT-III expression did not affect the cell surface expression of β4-integrin (Fig. 2A, lacZ versus lacZ+GnT-III and β4 versus β4+GnT-III). Overexpression of GnT-III in cells expressing β4-integrin significantly reduced β1,6GlcNAc-branched N-glycans on β4-integrin accompanied with increased bisecting GlcNAc (Fig. 2B). Expression of β4-integrin exhibited enhanced cell migration (Fig. 2C) and tumor formation (Fig. 2D) in the MDA-MB435S cells compared with the control cells, which were impaired by GnT-III expression. These results suggest that the β1,6GlcNAc residue in β4-integrin may be associated with primary tumor growth and metastatic cancer cell behavior.
N-Glycosylation of β4-integrin drives proliferation, migration, and invasion through PI3K/Akt activation
GnT-III overexpression modifies not only β4-integrin but also other glycoproteins (Supplementary Fig. S1). β4-integrin contains five potential N-glycosylation sites in its extracellular domain (Fig. 3A). To directly address the biological role of N-glycosylation of β4-integrin in cancer cell behavior, we prepared two β4-integrin constructs; a full-length wild-type (WT) and a mutant lacking all five N-glycosylation sites (ΔN; ref. Fig. 3A). These constructs were retrovirally expressed in MDA-MB435S cells that endogenously expressed no β4-integrin (Fig. 2A, lacZ). To confirm the loss of N-glycans on ΔN, we performed lectin blot analysis using L4-PHA, E4-PHA, and SSA lectins, which recognize β1,6GlcNAc, bisecting GlcNAc, and α2,6 sialic acid in N-glycans, respectively. The WT reacted with all tested lectins, but the ΔN did not react (Fig. 3B), suggesting the loss of N-glycans in the ΔN. Similar to our previous reports using keratinocytes (28), the deletion of N-glycans had no effect on either cell surface expression (Supplementary Fig. S2A) or heterodimer formation of β4-integrin with α6-integrin (Supplementary Fig. S2B) in the MDA-MB435S cells. These results suggest that N-glycans on β4-integrin were not required for either expression or heterodimer formation of α6β4-integrin. We next assessed whether N-glycosylation on β4-integrin can affect cell adhesion and spreading, which are generally associated with cancer progression, using the MDA-MB435S transfectants. Typically, β4-integrin causes cellular signaling through cell adhesion to the extracellular matrix protein laminins in cancer progression (2, 31). Compared with the WT cells, the ΔN cells showed decreased cell adhesion and spreading on a laminin-332 substrate, which was comparable to those in the control lacZ cells (Fig. 3C), suggesting that N-glycans on β4-integrin are related to β4-integrin-dependent cell adhesion and spreading in MDA-MB435S cancer cells.
β4-integrin regulates cell proliferation, migration, and invasion in cancer cell lines (2, 6). Therefore, we examined the cellular functions of MDA-MB435S cancer cells expressing WT and ΔN. The expression of β4-integrin markedly enhanced cell proliferation, migration, and invasion of the MDA-MB435S cells (Fig. 3D–F, WT) compared with the control lacZ cells. The enhanced cellular functions were significantly impaired by the loss of N-glycans on β4-integrin (Fig. 3D–F, ΔN). Previous studies have shown that β4-integrin–dependent cell proliferation, migration, and invasion are closely associated with the PI3K pathway in several carcinoma cell lines (9, 13). To address whether the decreased cellular functions due to the deletion of N-glycosylation were correlated to the downregulation of PI3K signaling, we examined the activation of the PI3K signaling pathway by Western blot analysis with an anti-phosphorylated Akt antibody. The ΔN cells, as well as the lacZ cells, showed decreased phosphorylation of Akt compared with the WT cells (Fig. 3G). We also found that the impairment of cellular functions in the ΔN cells was completely restored through overexpression of the constitutively active PI3K p110α subunit (Fig. 3D–F, ΔN+PI3K). In contrast, overexpression of the active PI3K subunit in WT cells did not significantly affect the WTβ4-dependent cell proliferation, migration, and invasion (Fig. 3D–F, WT+PI3K), suggesting that the WTβ4-mediated PI3K/AKT activation is sufficient for such functions in the cells and higher PI3K/AKT activation does not always promote activities of cancer cells. Together, these findings demonstrate that the N-glycosylation of β4-integrin plays a fundamental role in activating the PI3K pathway, thereby promoting cell proliferation, migration, and invasion.
N-Glycans of β4-integrin promote in vivo tumor growth.
To determine whether N-glycosylation of β4-integrin is related to tumorigenesis in vivo, we subcutaneously injected MDA-MB435S transfectants into nude mice and analyzed tumor growth. Compared with the WT cells, the ΔN cells showed significantly impaired tumorigenesis, which was comparable with the β4 null (lacZ) cells (Fig. 4A and B). Impairment of the tumorigenesis in the ΔN cells was fully restored through overexpression of the constitutively active PI3K p110α subunit (Fig. 4A, ΔN+PI3K). In comparison with the WT tumors, staining with anti–Ki-67 antibody demonstrated that the ΔN tumors included notably fewer proliferating cells, which was fully increased to WT tumor levels by overexpression of the constitutively active PI3K p110α subunit (Fig. 4C and D). These results demonstrate that the ability of β4-integrin to drive tumorigenesis, and tumor proliferation is associated with activation of PI3K pathway via N-glycans.
β4-integrin–dependent tumor progression is associated with interaction between galectin-3 and β1,6GlcNAc residue on β4-integrin
Galectin-3 is known to be associated with cancer aggressiveness and metastasis (22–24). To examine the role of galectin-3 in β4-integrin–mediated tumor progression, we performed inhibition assays using a functional blocking antibody against galectin-3 (M3/38). The functional blocking antibody against galectin-3 suppressed β4-integrin clustering probably through inhibition of galectin-3 multimerization, suggesting that galectin-3 actually cross-links β4-integrin (Fig. 5A). The antibody suppressed WT cell motility, whereas no effect was observed in the ΔN cells (Fig. 5B). Corresponding with the decreased cell motility of the WT cells, the galectin-3 antibody reduced Akt phosphorylation compared with control IgG in the WT cells (Fig. 5C). Furthermore, we tested the tumorigenicity of the WT cells expressing control shRNA or galectin-3 shRNA to confirm the importance of galectin-3 expression on tumor formation (Supplementary Fig. S3). The result showed that reduced expression of galectin-3 in WT cells suppressed tumor formation, suggesting that galectin-3 plays an important role in β4-integrin–promoting tumor formation (Fig. 5D).
Galectin-3 has a high affinity for β1,6GlcNAc-branched N-glycans modified by polylactosamine consisting of β-galactoside sugars. The oligomerization of integrins by galectin-3–mediated cross-linking between the β1,6GlcNAc-branched N-glycans on integrins promotes integrin function (25, 27, 32). To understand why deletion of N-glycosylation suppressed the ability of β4-integrin to promote cancer cell progression, we examined the relationship between galectin-3 binding to β4-integrin and cellular functions of β4-integrin. β4-integrin immunoprecipitates from the WT cells contained high amounts of galectin-3 compared with those from the lacZ and ΔN cells, indicating that N-glycans on β4-integrin promote complex formation by the two molecules (Fig. 5E). The association of β4-integrin with galectin-3 was inhibited by a competitive inhibitor of galectin binding, β-lactose but not by control sucrose, suggesting that galectin-3 is bound to β1,6GlcNAc-branched N-glycans modified by polylactosamine on β4-integrin (Fig. 5F). Similar results were obtained from the squamous cell carcinoma cell line, A431 cells, which endogenously express β4-integrin. The addition of galectin-3 significantly enhanced cell adhesion of the WT cells but not the lacZ or ΔN cells to a laminin-332 substrate (Fig. 5G). Furthermore, the addition of galectin-3 to the cell culture significantly promoted cell motility of the WT cells but not that of either the WT with GnT-III cells or ΔN cells (Fig. 5H). These findings suggest that galectin-3 binding to β4-integrin through β1,6GlcNAc-branched N-glycans promotes the ability of β4-integrin to drive cell adhesion and motility.
In the current study, β4-integrin was colocalized with β1,6GlcNAc-branched N-glycans in tumor tissues. The suppression of the modification of β4-integrin by GnT-III was associated with reduced cancer cell migration and tumorigenesis in MDA-MB435S cells expressing β4-integrin. Deletion of N-glycosylation sites in β4-integrin, which was accompanied with downregulation of the PI3K signaling pathway, inhibited β4-integrin–dependent cancer cell migration, invasion, proliferation, and tumor formation. Furthermore, loss of association between galectin-3 and β4-integrin via β1,6GlcNAc-branched N-glycans abolished galectin-3–promoting cancer cell adhesion and migration. These results provide evidence that N-glycosylation of β4-integrin plays a functional role in promoting tumor development and progression through the PI3K activation.
β4-integrin promotes cell proliferation, migration, and invasion (31, 33), and plays pivotal roles in tumorigenesis (8). In the present study, deletion of N-glycosylation sites in β4-integrin suppressed those cellular functions in vitro as well as cell proliferation and tumorigenesis in vivo. GnT-V knockout mice have been reported to suppress polyomavirus middle T oncogene-induced mammary tumor growth and metastasis (19). Previous studies have shown that an increase of β1,6GlcNAc in the α3-, α5-, or β1-integrin subunit resulted in increased cancer cell migration (34, 35). Our findings indicate that modification of β4-integrin with β1,6GlcNAc may be upregulated in tumor tissue. In addition, suppression of the β1,6GlcNAc modification in β4-integrin by overexpression of GnT-III reduced β4-integrin–dependent cancer cell migration and tumor formation. Therefore, this loss of β4-integrin function by the N-glycosylation defect may be mainly due to a lack of β1,6GlcNAc modification in β4-integrin.
Galectin-3 has a high affinity for β1,6GlcNAc-branched N-glycans through multivalent binding, and thereby cross-links glycoproteins on the cell surface and in the extracellular matrix to form molecular complexes (24). The formation of the macromolecular complexes on the cell surface by galectin-3 affects the distribution of glycoproteins and cellular signaling. Furthermore, integrin clustering mediated by galectin-3 promotes integrin activation (25, 27). In our study, we found that the deletion of N-glycans of β4-integrin, resulting in impaired galectin-3 binding to β4, suppressed β4-integrin functions mainly through downregulation of the PI3K pathway. These results support the hypothesis that the cross-linking between β1,6GlcNAc-branched N-glycans of β4-integrins by galectin-3 directly or indirectly affects PI3K activation and cellular functions such as cell motility and tumorigenesis (Fig. 6); however, further investigation is required.
The association of β4-integrin with laminin-332 is known to induce PI3K activation, thereby promoting cell adhesion and migration (2). Our previous study showed that a decreased level of β1,6GlcNAc-branched N-glycans on laminin-332 by introduction of bisecting GlcNAc suppressed its cell adhesion and migration activity as well as galectin-3–mediated β4-integrin clustering (27). The present study showed that a defect of N-glycosylation in β4-integrin suppressed the association with laminin-332 and PI3K activation, suggesting that N-glycosylation of β4-integrin played a critical role in PI3K activation through the interaction with laminin-332. These results suggest that the association between laminin-332 and β4-integrin through β1,6GlcNAc-branched N-glycans may also be important for β4-integrin–dependent tumor progression.
In the present study, we cannot exclude the potential effect of the lack of N-glycosylation on β4-integrin protein folding. N-Glycosylation on integrins is important for such protein folding, which is required for heterodimer formation. In fact, the lack of N-glycans on α5-integrin causes its misfolding and loss of heterodimer formation with β1-integrin (36). In contrast, ΔNβ4-integrin could form a heterodimer with α6-integrin (Supplementary Fig. S2B). Furthermore, FACS analysis showed that ΔNβ4 expressed on the cell surface, which is comparable with WTβ4 (Supplementary Fig. S2A). In essence, noncomplexed integrin is degraded immediately or remains in the endoplasmic reticulum (37). Therefore, the lack of N-glycans on β4-integrin seems to affect its function and the association with other molecules, rather than β4-integrin folding.
In conclusion, our study suggests that N-glycosylation of β4-integrin is associated with tumor development and progression through β4-integrin/PI3K signaling via the galectin-3–N-glycan complex. N-Glycosylation of β4-integrin may therefore represent a potential therapeutic target for cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: J. Gu, Y. Kariya
Development of methodology: Y. Kariya, Y. Kariya
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Hashimoto, Y. Kariya
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Kariya, J. Gu, Y. Kariya
Writing, review, and/or revision of the manuscript: Y. Kariya, J. Gu, Y. Kariya
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Kariya, M. Oyama, Y. Kariya
Study supervision: J. Gu, Y. Kariya
This work was supported by the Japan Society for Promotion of Science KAKENHI grant number 25860243, 17K08665 (Y. Kariya), 15H04354 (J. Gu), and the Fukushima Medical University Research Project from Fukushima Medical University No. KKI28001 (Y. Kariya).
The authors are grateful to Dr. M. Peter Marinkovich (Stanford University, Standford, CA) for providing the cells. We thank the Fukushima Medical University English Editing Service for their English language review.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.