Mucin 16 is a functional selectin ligand on pancreatic cancer cells

Abstract

Selectins promote metastasis by mediating specific interactions between selectin ligands on tumor cells and selectin-expressing host cells in the microvasculature. Using affinity chromatography in conjunction with tandem mass spectrometry and bioinformatics tools, we identified mucin 16 (MUC16) as a novel selectin ligand expressed by metastatic pancreatic cancer cells. While up-regulated in many pancreatic cancers, the biological function of sialofucosylated MUC16 has yet to be fully elucidated. To address this, we employed blot rolling and cell-free flow-based adhesion assays using MUC16 immunopurified from pancreatic cancer cells and found that it efficiently binds E- and L- but not P-selectin. The selectin-binding determinants are sialofucosylated structures displayed on O- and N-linked glycans. Silencing MUC16 expression by RNAi markedly reduces pancreatic cancer cell binding to E- and L-selectin under flow. These findings provide a novel integrated perspective on the enhanced metastatic potential associated with MUC16 overexpression and the role of selectins in metastasis.—Chen, S.-H., Dallas, M. R., Balzer, E. M., Konstantopoulos, K. Mucin 16 is a functional selectin ligand on pancreatic cancer cells.

Metastasis is a multistep process in which cancerous cells separate from a primary tumor and enter the vascular system, where they interact extensively with host cells and tissues. Tumor cells that successfully arrest in distant sites may then exit the vasculature and establish micrometastatic colonies. Mounting evidence suggests that vascular selectins (E-, L-, and P-selectin), a family of C-type lectins, play a prominent role in this process by facilitating the interaction of circulating tumor cells (CTCs), which express selectin ligands, and selectin-expressing host cells. Mice deficient in P- and/or L-selectin display reduced metastasis (1, 2), which may result from diminished aggregation of CTCs with P-selectin-expressing platelets (3–5) and L-selectin-bearing neutrophils (6, 7). E-selectin has also been shown to support metastatic spread in vivo, as E-selectin expressed on the surface of activated endothelial cells facilitates adhesion of free-flowing CTCs and thus their eventual exit from the vasculature and lodging in target organs (8–11).

Selectins recognize sialofucosylated oligosaccharides, such as sialyl Lewis x (sLe^x) and its isomer, sialyl Lewis a (sLe^a) (12). Overexpression of these epitopes is associated with pancreatic tumor growth and metastasis (13–15). In fact, sLe^x is absent from healthy pancreatic tissue, but its expression progressively increases with higher-grade pancreatic intraepithelial neoplasia (PanIN) lesions and pancreatic adenocarcinoma (14, 16). Both sLe^x- and sLe^a-decorated glycoconjugates can mediate selectin-dependent CTC adhesion to host cells (12, 17). Despite the importance of selectin-mediated binding to sialofucosylated target molecules as a potential metastatic determinant for pancreatic cancer, selectin ligands on pancreatic cancer cells have not been well characterized other than by general classifications (i.e., sialofucosylated glycoproteins). The binding affinity of selectins for isolated sLe^x and sLe^a is markedly low. As a consequence, neither sLe^x nor sLe^a itself correlates with the properties of endogenous selectin ligands on cellular targets. As pointed out in the literature (18), the “functional” selectin ligand should fulfill certain requirements: the ligand should bind with some selectivity and relatively high affinity; removal of it should prevent cell adhesive interaction.

In this study, we sought to identify novel selectin glycoprotein ligands on pancreatic carcinoma cells by utilizing a combination of immunoaffinity chromatography and tandem mass spectrometry (MS/MS). Our analysis identified a member of the mucin family of glycoproteins, mucin 16 (MUC16), as an E- and L-selectin ligand overexpressed on metastatic pancreatic cancer cells.

MUC16 is a large (2500–5000 kDa), heavily glycosylated protein that is expressed in mucous membranes of various tissues, including the upper respiratory tract and reproductive organs (19, 20). MUC16 contains an extracellular N terminus adjacent to glycosylated tandem repeats, a transmembrane region, and a short cytoplasmic tail (21, 22). In a clinical context, MUC16 is significantly up-regulated in >80% of patients with ovarian tumors, and the MUC16 epitope CA125 is an accepted tumor marker for ovarian cancer (23, 24). Recent work also suggests that MUC16 is not expressed in the normal pancreatic ducts but is strongly up-regulated in pancreatic cancer and may play a potential role in the progression of this disease (25). Despite this strong clinical correlation, a biological mechanism for how MUC16 enhances the tumor progression has yet to be delineated beyond its role as a mesothelin ligand (26).

To define its functional role as a selectin ligand, we conducted the following set of experiments: blot rolling and cell-free flow-based adhesion assays revealed that MUC16 possesses high E- and L- but low P-selectin-binding activity. Treatment of pancreatic cancer cells with specific glycosidases and metabolic inhibitors revealed that the selectin-binding determinants for MUC16 are sialofucosylated structures displayed on both O- and N-linked glycans. Furthermore, silencing MUC16 with short hairpin RNA (shRNA) significantly suppresses binding to immobilized E- and L-selectin under flow. Collectively, our results suggest that up-regulation of sialofucosylated MUC16 may enhance the ability of CTCs to tether and adhere to host tissues that express E- and L-selectin ligands. Our findings provide a unified perspective for the enhanced metastatic potential associated with overexpression of sialofucosylated structures and MUC16 in pancreatic cancers and the role of selectins in metastasis.

MATERIALS AND METHODS

Cell culture

The human nonmalignant pancreatic epithelial hTERT-HPNE cells and the pancreatic cancer cell lines CFPAC-1 and SW1990 were obtained from the American Type Culture Collection (Manassas, VA, USA), and Pa03C (27) was a generous gift from Dr. Anirban Maitra (Johns Hopkins School of Medicine, Baltimore, MD, USA). hTERT-HPNE was cultured in medium containing 3 vol of glucose-free DMEM, 1 vol of Medium M3 base (InCell, San Antonio, TX, USA), 5% FBS, 5.5 mM glucose, 10 ng/ml human recombinant EGF, and 50 μg/ml gentamicin at 37°C under air with 5% CO₂. SW1990 and Pa03C were cultured in DMEM with 10% FBS. CFPAC-1 was cultured in IMEM with 10% FBS. Before cell lysis, cells were detached from culture flasks using enzyme-free cell dissociation medium (15 min at 37°C; Chemicon, Phillipsburg, NJ, USA). For flow cytometry and flow-based adhesion assays, cells were harvested by mild trypsinization [0.05% trypsin/ethylenediaminetetraacetic acid (EDTA) for 5 min at 37°C] and subsequently incubated (10⁷ cells/ml) at 37°C for 2 h to allow for regeneration of surface glycoproteins (3, 28). E- and P-selectin-expressing Chinese hamster ovary (CHO-E and CHO-P) cells, stably transfected with cDNA encoding full-length E- or P-selectin, respectively, were generously provided by Affymax (Palo Alto, CA, USA) and processed as described previously (29, 30).

Antibodies, reagents, and adhesion molecules

Anti-sLe^a monoclonal antibody (mAb) KM231 was purchased from Millipore (Temecula, CA, USA). Anti-MUC16 mAb X306 was from Santa Cruz Biotechonology (Santa Cruz, CA, USA), X75 was from Abcam (Cambridge, MA, USA), and M11 was from Dako (Burlington, ON, Canada). Anti-MUC1 (VU-4-H5) and MUC4 (1G8) mAbs were from Invitrogen (Carlsbad, CA, USA). Phycoerythrin (PE)-conjugated secondary antibody was from Vector Laboratories (Burlingame, CA, USA). Alkaline phosphatase (AP)- and horseradish peroxidase (HRP)-conjugated anti-mouse IgG and AP-conjugated anti-rat IgM were from Southern Biotech (Birmingham, AL, USA). All other unlabeled antibodies were from BD Biosciences Pharmingen (San Jose, CA, USA), unless otherwise stated. The chimeric form of E-, L-, and P-selectin-IgG Fc (E-, L-, and P-selectin) consisting of the lectin, EGF, and consensus repeat domains for human E-, L-, and P-selectin, respectively, linked to each arm of human IgG1, were purchased from R&D Systems (Minneapolis, MN, USA).

Glycoprotein purification and MS analysis

As described previously (31, 32), the putative selectin ligand corresponding to the glycoproteins with a molecular mass > 460 kDa, which carries HECA-452-reactive sialofucosylated oligosaccharides such as sLe^x and sLe^a, were isolated from SW1990 cell lysate by affinity chromatography using recombinant protein G agarose supports (Invitrogen) cross-linked with bis(sulfosuccinimidyl) suberate (Pierce Biotechnology, Rockford, IL, USA) to the anti-sLe^a mAb KM231. Eluted proteins were then separated by 3–8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gels (Invitrogen) and stained in gel with ProQ Emerald 300 glycoprotein stain (Invitrogen), which binds only to carbohydrate groups at glycosylation sites, thereby leaving the polypeptide core intact (31, 32). The stained band with a molecular mass > 460 kDa was then excised, and trypsin-digested gel fragments were submitted for analysis by nanoflow HPLC interfaced to electrospray ionization MS/MS (HPLC-MS/MS) using a ThermoFinnigan LTQ mass spectrometer (refs. 31, 32; ThermoFinnigan, San Jose, CA, USA). The MS data were searched against all taxonomies in the National Center for Biotechnology Information (NCBI; Bethesda, MD, USA) nonredundant protein database with a 95% significance threshold (P<0.05) using Mascot (Matrix Science, Boston, MA, USA) and with a P < 0.01 confidence using the BioWorks 3.3 software featuring the SEQUEST algorithm (ThermoFinnigan).

Western blot analysis

Whole-cell lysate or immunopurified MUC16 was diluted with reducing sample buffer and separated using 3–8% SDS-PAGE gels. Resolved proteins were transferred to Sequi-blot or Immun-blot PVDF membrane (Bio-Rad Laboratories, Hercules, CA, USA) and blocked with StartingBlock (Pierce Biotechnology) for 30 min. Blots were stained with HECA-452 or anti-MUC16 (X306) mAbs and rinsed with TBS-0.1% Tween 20. Blots were incubated with appropriate AP- or HRP-conjugated secondary antibodies. Western blue AP substrate (Promega) and SuperSignal West Pico chemiluminescent substrate (Pierce Biotechnology) were used to develop the AP- and HRP-conjugated antibody-stained immunoblots, respectively.

Cell lysis and immunoprecipitation (IP) of MUC16

Whole-cell lysate was prepared by membrane disruption using 2% Nonidet P-40 followed by differential centrifugation (30–33). MUC16 was purified via IP from SW1990 and Pa03C cell lysate with an anti-MUC16 mAb (X306, X75, or M11), using recombinant protein G agarose spheres (30–33).

Blot rolling assay

Western blots of SW1990 and Pa03C cell lysate or immunopurified MUC16 from the lysate were stained with anti-MUC16 or HECA-452 mAbs and rendered translucent by immersion in 90% Dulbecco's modified medium-PBS (D-PBS)-10% glycerol as described previously (34). The blots were placed under a parallel-plate flow chamber, and human peripheral blood lymphocytes or CHO transfectants, resuspended at 5 × 10⁶ cells/ml in 90% D-PBS-10% glycerol, were perfused at a shear stress of 0.5 dyn/cm² (30–33). Molecular mass markers were used as guides to aid placement of the flow chamber over stained bands of interest. The number of interacting cells per lane was averaged over 5 ×10 fields of view (0.55 mm² each) within each stained region. Nonspecific adhesion was assessed by perfusing 5 mM EDTA in the flow medium.

Preparation of MUC16-coated microspheres

MUC16 purified from SW1990 and Pa03C cell lysate was diluted to desired concentrations with binding buffer (0.2 M carbonate/bicarbonate buffer, pH 9.2) and incubated with 10 μm polystyrene microspheres (2.5×10⁷ microspheres/ml; Polysciences, Warrington, PA, USA) overnight at 4°C with constant rotation, as described previously (30–33). Microspheres were washed 2 times with D-PBS and subsequently blocked with D-PBS/1% BSA for 30 min at room temperature. Microspheres were resuspended (10⁶ microspheres/ml) in D-PBS/0.1% BSA for use in flow cytometric and flow chamber assays. Site densities of MUC16-coated microspheres were determined by flow cytometry (30–33).

Enzymatic and inhibitor treatments

MUC16-coated microspheres (1×10⁷ microspheres/ml) were treated with 0.1 U/ml Vibrio cholerae sialidase (Roche Molecular Biochemicals, Indianapolis, IN, USA) for 90 min at 37°C to remove terminal sialic acid residues (30, 32, 33). In select experiments, MUC16-coated microspheres suspensions (5×10⁶ microspheres/ml) were incubated for 2 h at 37°C with 120 μg/ml of O-sialoglycoprotein endopeptidase (OSGE; Accurate Chemical & Scientific, Westbury, NY, USA) to specifically cleave glycoproteins with O-glycosylation on serine and threonine residues (35). To cleave N-linked glycans from MUC16, MUC16-coated microspheres were treated with PNGase F (New England Biolabs, Beverly, MA, USA) according to the manufacturer's protocol. For metabolic inhibitor studies, cell suspensions (10⁷ cells/ml) were pretreated with 0.1 U/ml sialidase for 60 min at 37°C to remove terminal sialic acid residues and ensure de novo synthesis of newly generated HECA-452-reactive carbohydrate structures (30, 32, 33). Complete removal of sialylated structures was confirmed via flow cytometry using the mAb HECA-452 that recognizes sialic acid-bearing epitopes. Subsequently, cells were cultured for 48 h at 37°C in medium containing either 2 mM benzyl-2-acetamido-2-deoxy-α-d-galactopyranoside (benzyl-GalNAc) to inhibit O-linked glycosylation or 1 mM deoxymannojirimycin (DMJ) to disrupt N-linked processing (30, 32, 33); DMSO dilution was used for control untreated cells. Site densities of MUC16 adsorbed on microspheres following enzymatic or inhibitor treatments were determined by flow cytometry before used in flow-based adhesion assays.

Flow cytometry

MUC16 and HECA-452 site densities on microspheres were assessed by single-color immunofluorescence and flow cytometry (FACSCalibur; BD Biosciences) using primary anti-MUC16 mAb (X306) with appropriate PE-conjugated secondary antibodies or FITC-conjugated HECA-452 mAb. Similarly, surface expression of MUC1, MUC4, and sLe^a was monitored using anti-MUC1 (VU-4-H5), MUC4 (1G8), or sLe^a (KM231) mAbs with appropriate PE-conjugated secondary antibodies. CD29 expression on cells was monitored by using the PE-conjugated anti-CD29 antibody (MAR4). Background levels were determined by incubating cell or microsphere suspensions with properly matched FITC- or PE-conjugated isotype control antibodies (30, 32, 33).

Flow-based adhesion assays

Wild-type, mammalian scramble control, and MUC16-knockdown (MUC16-KD) cells or MUC16-coated microspheres suspended in D-PBS-0.1% BSA (1×10⁶/ml) were perfused over immobilized E-, L-, or P-selectin-coated dishes at prescribed wall shear stresses using a parallel-plate flow chamber (250 μm channel depth, 5.0 mm channel width) as described previously (3, 5). The number of interacting cells or microspheres was then quantified by averaging over multiple ×10 fields of view during a 2-min period. In select experiments, E-, L-, or P-selectin-coated dishes were preincubated with the 20 μg/ml anti-E-, -L-, or -P-selectin function-blocking mAb for 1 h before being used in flow assays. Nonspecific adhesion was assessed by adding 5 mM EDTA to the perfusion medium.

Preparation of MUC16 short interfering RNA (siRNA) oligonucleotides

siRNA oligonucleotides targeting MUC16 were generated using the siRNA design program from Whitehead Institute (Massachusetts Institute of Technology, Cambridge, MA, USA) as described previously (30, 31). The siRNA sequences were used to construct 60-mer shRNA oligonucleotides, which were then synthesized (Invitrogen) and ligated into the pSUPER.puro expression vector (Oligoengine, Seattle, WA, USA) under the control of the H1 promoter. The following oligonucleotide was used: 5′-GATCCCCCAGCAGCATCAAGAGTTATTTCAAGAGAATAACTCTTGATGCTGCTGTTTTTA-3′ (underscored, sense and antisense sequences; boldface, restriction enzyme sites; italicized, polymerase III termination signal; boldface italicized, loop with linker). The ligated product was transformed into competent DH5α Escherichia coli cells and amplified in the presence of ampicillin, and the plasmid was purified using the EndoFree Maxi kit (Qiagen, Valencia, CA, USA). Sequence insertion was verified by restriction digestion and confirmed by direct sequencing. A mammalian scramble sequence (Oligoengine) was used as a negative control in all siRNA experiments.

Generation of stable MUC16-KD pancreatic carcinoma cell lines

As described previously (30, 31), SW1990 and Pa03C cells were plated in 100-mm dishes and grown to reach 90% confluence. Cells were then transfected with pSUPER.puro. MUC16 using Lipofectamine 2000 (Invitrogen) according to manufacturer's instructions. On reaching confluence, transfected cells were passed and 10⁶ cells/Petri dish were seeded in growth medium in triplicate. After 24 h, the medium was replaced by a fresh aliquot containing 0.5 μg/ml puromycin (Invitrogen). Cells were then grown continually without passaging for 2 wk, replenishing the puromycin-containing medium every 3 d. Single-cell colonies were isolated and cultured using standard techniques.

Statistical analysis

Data are expressed as means ± se for ≥3 independent experiments. Statistical significance of differences between means was determined by ANOVA. The statistical significance was set at probability values of P < 0.05.

RESULTS

Sialofucosylated MUC16 is expressed on metastatic pancreatic cancer cells

Sialofucosylated glycoproteins presented on the surface of metastatic pancreatic cancer cells are important mediators of selectin-dependent adhesion (36, 37). However, the selectin ligands on pancreatic cancer cells have yet to be identified. Using a flow-based adhesion assay, we observed that metastatic pancreatic cancer cells, such as SW1990, Pa03C, and CFPAC-1, expressing sialofucosylated glycoproteins (Supplemental Fig. S1A), adhered to E-selectin and transiently tethered to L-selectin in shear flow (Table 1). In contrast, the nonmalignant hTERT-HPNE pancreatic epithelial cells are devoid of sialofucosylated structures (Supplemental Fig. S1A) and failed to interact with selectins (Table 1). In an attempt to characterize the sialofucosylated glycoproteins that function as selectin ligands on metastatic pancreatic cancer cells, SW1990 cells were initially chosen for further studies because of their potent selectin binding activity. Western blot analysis of whole SW1990 cell lysate was carried out to examine the expression of all sialofucosylated (i.e., HECA-452-reactive) glycoproteins. Two HECA-452-reactive protein bands were identified in a 3–8% gradient gel: an intense band broadly distributed at >460 kDa and a weak signal at ∼200 kDa (Fig. 1A). With the use of a blot rolling assay, the selectin binding activity of these HECA-452-reactive bands was evaluated. This technique identifies selectin-ligand interactions under physiological shear conditions (34). To this end, CHO-E and CHO-P cells and L-selectin-expressing lymphocytes were perfused at 0.5 dyn/cm² over HECA-452 immunostained blots, which were rendered translucent by immersion in 90% D-PBS/10% glycerol (34). CHO-E (Fig. 1B) and lymphocytes (not shown) bound predominantly to the >460- and ∼200-kDa regions, whereas CHO-P cells failed to interact with either band.

Table 1.

Extent of adhesion of pancreatic nonmalignant or carcinoma cells to E-, L-, and P-selectin under physiological fluid flow

Cell line	Pancreatic cell type	Interaction with selectins
		E	L	P
hTERT-HPNE	Nonmalignant	−	−	−
CFPAC-1	Metastatic carcinoma	+	*	−
SW1990	Metastatic carcinoma	++++	***	−
Pa03C	Metastatic carcinoma	+++	**	−

Pancreatic nonmalignant cells or pancreatic cancer cells were perfused at a wall shear stress level of 0.5 dyn/cm² over the surface coated with 20 μg/ml E-, L-, and P-selectin. +, Firm cell adhesion; *, transiently cell tethering/rolling; −, no interaction.

Figure 1.

Sialofucosylated glycoproteins expressed by SW1990 pancreatic carcinoma cells support selectin-mediated adhesion. A) Western blot of SW1990 cell lysate was stained with a HECA-452 mAb. Two major bands were observed at ∼200 and >460 kDa. B) Selectin-dependent adhesion to SDS-PAGE resolved SW1990 lysate under physiological flow conditions. CHO-E cells were perfused at the wall shear stress level of 0.5 dyn/cm² over the HECA-452 stained Western blots. Number of interacting cells per square millimeter was tabulated as a function of molecular weight to compile an adhesion histogram. Data represent means ± se of n = 5 experiments.

We recently reported that sialofucosylated CD44 variant isoforms, carcinoembryonic antigen, and podocalyxin are functional selectin ligands on colon carcinoma cells (29–33). Because of their lower molecular masses (∼150–180 kDa), it is unlikely either of these molecules constitutes the HECA-452-reactive band observed in the >460-kDa region. To identify this novel selectin ligand, we purified the putative sialofucosylated glycoprotein from SW1990 cell lysates by affinity chromatography using protein G agarose beads coated with an anti-sLe^a mAb (Fig. 2). Eluted proteins were separated by SDS-PAGE and stained in gel with the ProQ Emerald 300 stain, which fluorescently labels periodate-oxidized glycans while leaving the polypeptide backbone intact (31, 32). The gel band corresponding to the >460-kDa HECA-452-reactive protein was excised and digested in gel with trypsin. Extracted peptides were analyzed by HPLC-MS/MS (Fig. 2). Bioinformatics analysis identified peptide fragment matches for MUC16 (NP_078966.2).

Figure 2.

Identification of MUC16 as the sialofucosylated glycoprotein > 460 kDa expressed by SW1990 cells. 1) HECA-452-reactive molecules were purified from SW1990 cell lysate by immunoaffinity chromatography using anti-sLe^a mAb immobilized on recombinant protein G agarose supports. HECA-452-reactive molecules were eluted with low-pH elution buffer. 2) Samples were separated by SDS-PAGE and stained in gel with ProQ Emerald 300 glycoprotein stain. Fluorescently labeled band >460 kDa was subsequently excised from the gel. 3) Proteins were extracted, trypsin-digested, and subjected to tandem MS. 4, 5) bioinformatics analysis of the MS data revealed the presence of MUC16 in the HECA-452-reactive band >460 kDa.

A series of experiments was then carried out to confirm the MS data. Western blot analysis using an anti-MUC16 mAb, X306, revealed the presence of MUC16 in SW1990 cell lysates, and MUC16 was retained in the affinity chromatography eluate (Fig. 3A). Flow cytometry confirmed the expression of MUC16 on the surface of SW1990 cells (Supplemental Fig. S1B). In addition, sLe^a and sLe^x epitopes on MUC16 were disclosed by staining immunopurified MUC16 with HECA-452 mAb (Fig. 3A).

Figure 3.

MUC16 carries sialofucosylated glycans and supports selectin-mediated adhesion. A) Western blots of whole-cell lysate or IP MUC16 or sLe^a from SW1990 pancreatic carcinoma cells. HECA-452 (lanes 1, 3, 5) or anti-MUC16 (lanes 2, 4, 6) mAbs were used to stain Western blots of cell lysate (lanes 1, 2), IP MUC16 (lanes 3, 4), and IP sLe^a (lanes 5, 6) from the cells. B) Selectin-dependent adhesion to SDS-PAGE resolved immunopurified MUC16 from SW1990 lysate. CHO-P cells, lymphocytes, or CHO-E cells were perfused at the wall shear stress level of 0.5 dyn/cm² over Western blots of MUC16 immunopurified from SW1990 lysate. For control experiments, CHO-P cells, lymphocytes, or CHO-E cells were pretreated with anti-P-, L-, or E-selectin function-blocking mAbs (20 μg/ml), respectively, before blot rolling assays. This concentration was maintained throughout the experiment. Data represent means ± se of n = 3 experiments. *P < 0.05 vs. untreated cells.

We next sought to confirm the ability of MUC16 to mediate selectin binding under physiological levels of shear stress. To this end, CHO-E and CHO-P cells and L-selectin-expressing human peripheral blood lymphocytes were perfused over the immunopurified MUC16 band at a wall shear stress of 0.5 dyn/cm². CHO-E cells firmly adhered to the MUC16 protein band (Fig. 3B); lymphocytes tethered transiently to immunopurified MUC16, which is in line with the nature of selectin-mediated binding, whereas negligible binding of CHO-P cells was detected (Fig. 3B). The specificity of these adhesive interactions was confirmed by incubating CHO-E, CHO-P, or lymphocyte suspensions with an anti-E-, -P-, or -L-selectin function-blocking mAb, respectively (Fig. 3B). Moreover, CHO-E cells or lymphocytes suspended in flow medium containing EDTA, which sequesters Ca²⁺ (indispensable in the selectin-ligand interaction), failed to adhere to any region of the blot (data not shown). Taken together, the data suggest that MUC16 possesses E- and L- but not P-selectin ligand activity. These observations are consistent with flow-based adhesion assays using intact SW1990 cells, which revealed appreciable binding to E- and L- but not P-selectin (Table 1).

To eliminate the presence of potential non-MUC16 selectin ligands in the >460-kDa region (Fig. 1A), whole SW1990 cell lysate was depleted of MUC16 by repeated IP and the immunodepleted fraction was measured for selectin ligand activity in blot rolling assays (29). After 4 rounds of IP with anti-MUC16 mAb, no MUC16 glycoprotein and HECA-452 reactivity were detectable by Western blot analysis (Fig. 4A). In addition, blot rolling assays revealed the number of interacting CHO-E cells over the >460-kDa region of the MUC16-depleted blot was essentially abolished (Fig. 4B). As a control, no significant change was detected in the extent of selectin-mediated adhesion to the ∼200-kDa region of the MUC16-depleted blot (Fig. 4B). Collectively, these observations suggest that MUC16 is the main selectin ligand in the >460-kDa HECA-452-reactive species from SW1990 cell lysates. Of note, other membrane-bound mucins often implicated in cancer progression are either modestly expressed (e.g., MUC1) by or absent (e.g., MUC4) from SW1990 cells (Supplemental Fig. S1C).

Figure 4.

Immunodepletion of MUC16 eliminates selectin-mediated binding >460 kDa. A) Western blots of SW1990 lysate depleted of MUC16 by sequential IP. Anti-MUC16 mAbs (X306 or X75) were used to immunodeplete MUC16 from the cell lysate by 4 rounds of IP. Anti-MUC16 mAb and HECA-452 were used to stain the untreated lysate (ctrl) and sequentially depleted lysates (depletions 1–4). B) CHO-E cells were perfused at the wall shear stress level of 0.5 dyn/cm² over 2 major HECA-452-reactive bands on SDS-PAGE Western blots of SW1990 cell lysate depleted of MUC16 by sequential rounds of IP. Number of interacting cells per square millimeter was tabulated as described in Fig. 1. Data represent means ± se of n = 3 experiments. *P < 0.05; ANOVA.

Selectin-binding determinants of MUC16 on pancreatic carcinoma cells are displayed on sialofucosylated O- and N-linked glycans

We next investigated the mechanism of interaction between MUC16 and selectins using a cell-free flow-based adhesion assay (30–33). Microspheres were coated with MUC16 immunopurified from SW1990 and perfused over E-, L-, and P-selectin-coated substrates at a wall shear stress of 1 dyn/cm². In agreement with blot rolling assays (Fig. 3B), a substantial number of MUC16-coated microspheres bound to E- and L-selectin (Fig. 5A). The specificity of MUC16-selectin binding in these assays was confirmed by the use of nonspecific IgG-bearing microspheres and by preincubating the selectin-coated dishes with the respective anti-selectin function-blocking mAb prior to perfusion of MUC16-coated microspheres (Fig. 5A). As an additional control, MUC16-coated microspheres failed to adhere to any of 3 different selectin substrates in the presence of EDTA (Fig. 5A).

Figure 5.

MUC16-coated microspheres interact with immobilized selectins, and the binding determinants of MUC16 are sialofucosylated structures displayed on both O- and N-linked glycans. A) Extent of interaction of microspheres (10⁶/ml) coated with nonspecific IgG or MUC16 purified from SW1990 cells with E-, L-, or P-selectin (20 μg/ml) at a wall shear stress level of 1.0 dyn/cm² for 2 min. In control experiments, MUC16-coated microspheres were incubated with anti-E-, L-, or P-selectin function-blocking mAbs (20 μg/ml) before use in flow assays. Nonspecific adhesion was assessed by adding EDTA (5 mM) to the perfusion medium. B) Extent of interaction of MUC16-coated microspheres pretreated with V. cholera sialidase, PNGase F, or OSGE with E- or L-selectin at a wall shear stress level of 1.0 dyn/cm². C) Extent of interaction of microspheres coated with MUC16 immunopurified from SW1990 cells cultured in DMJ or benzyl-GalNAc-containing medium with E- or L-selectin at a wall shear stress level of 1.0 dyn/cm². Data represent means ± se of n = 3–4 experiments. *P < 0.05 vs. control MUC16-coated microspheres.

To determine the location of the sialofucosylated glycans responsible for selectin binding of MUC16, MUC16-coated spheres were treated with enzymes that selectively cleave carbohydrate moieties and examined for their ability to interact with E- and L-selectin in shear flow. After treatment of MUC16-coated spheres with sialidase, which cleaves the sialic acid residue on sLe^x and sLe^a, adhesion of MUC16-bearing microspheres to E- and L-selectin was reduced by ∼90% (Fig. 5B). The removal of sialic acid residues in sialidase-treated MUC16-coated microspheres was confirmed by flow cytometry whereas control MUC16-bearing spheres retained their HECA-452 reactivity (Supplemental Fig. S2A). To assess the relative contributions of N- vs. O-linked glycans on MUC16-selectin binding, MUC16-bearing spheres were treated with either PNGase F, which cleaves N-linked glycans from the protein backbone, or OSGE, which cleaves the polypeptide backbone of heavily O-glycosylated glycoproteins, followed by flow-based adhesion assays, as described previously (32, 35). PNGase F treatment reduced the HECA-452 reactivity of MUC16-coated spheres (Supplemental Fig. S2A) and diminished the interaction between MUC16-coated spheres and E- and L-selectin (Fig. 5B). OSGE treatment of MUC16 completely eliminated sLe^x and sLe^a detection on the bead surface in flow cytometric analysis (Supplemental Fig. S2A), resulting in dramatic reduction in adhesion of MUC16-coated spheres to E- and L-selectin (Fig. 5B). To validate these data, microspheres were coated with MUC16 immunopurified from SW1990 cells pretreated with metabolic inhibitors DMJ, an N-linked processing inhibitor, or benzyl-GalNAc, which interferes with O-glycosylation. The site density of MUC16 on microspheres from DMJ- or benzyl-GalNAc-treated SW1990 cells was assessed to be similar to those of untreated controls (data not shown). However, the expression of sLe^x and sLe^a epitopes on MUC16 was partially reduced after DMJ incubation or treatment with benzyl-GalNAc (Supplemental Fig. S2B). Microspheres coated with MUC16 from DMJ- or benzyl-GalNAc-derived lysates displayed reduced interaction with E- or L-selectin compared with spheres treated with vehicle control lysates (Fig. 5C). To ensure the generality of our biochemical observations, MUC16 was also immunopurified from metastatic Pa03C pancreatic cancer cells and used to coat microspheres. Pa03C MUC16-coated microspheres displayed significant E- and L-selectin-binding activity in shear flow (Supplemental Fig. S3). These adhesive interactions were markedly reduced by sialidase treatment (Supplemental Fig. S3). Furthermore, N-linked glycosidase PNGase F and O-linked processing inhibitor benzyl-GalNAc attenuated the binding of Pa03C MUC16-bearing microspheres to E- and L-selectin (Supplemental Fig. S3). Taken together, these data suggest that the selectin-binding determinants on MUC16 immunopurified from metastatic SW1990 and Pa03C pancreatic cancer cells are sialofucosylated epitopes displayed on both O- and N-linked glycans.

MUC16 is a functional E- and L-selectin ligand on pancreatic carcinoma cells

To assess the functional role of MUC16 in the adhesion of pancreatic carcinoma cells to selectins under flow conditions, stable MUC16-KD cell lines were generated by transfecting wild-type SW1990 cells with MUC16 shRNA plasmids. As shown in Fig. 6A, this procedure resulted in the generation of MUC16-KD cells with markedly reduced MUC16 surface expression (>92% decrease in mean fluorescence intensity) relative to SW1990 cells transfected with scramble control or mock-transfected cells. The specificity of this interaction was confirmed by flow cytometric analysis of CD29, for which expression levels remained unaltered (Fig. 6A). MUC16 knockdown efficiency was confirmed by Western blot (Fig. 6B). Probing MUC16-KD with a HECA-452 mAb showed a dramatic reduction in staining intensity at the >460-kDa region (Fig. 6C, lane 2), whereas the HECA-452-reactive band at the ∼200-kDa region remained unchanged between wild-type and MUC16-KD cells. When CHO-E cells were perfused over the >460-kDa region of Western blots of MUC16-KD cell lysate, a marked reduction in CHO-E binding was observed in comparison with binding to blots of wild-type cell lysate. In contrast, the interacting number of cells over the ∼200-kDa region remained unaltered (Fig. 6D). Similar observations were made using L-selectin-expressing human peripheral lymphocytes (data not shown).

Figure 6.

Knockdown of MUC16 decreases E- and L-selectin-mediated interaction of SW1990 cells. A) Representative flow cytometric histograms of MUC16 and CD29 surface expression by wild-type, mammalian scramble control, and 2 clones of MUC16-KD SW1990 cells. Cells were stained by indirect single-color immunofluorescence using the anti-MUC16 or CD29 mAbs and their isotype control antibodies. B) Western blot of whole-cell lysate from wild-type and MUC16-KD SW1990 cells using anti-MUC16 or anti-β-actin antibody. C) Western blot of whole-cell lysate from wild-type and MUC16-KD SW1990 cells using HECA-452 mAb. D) Selectin-dependent adhesion under physiological flow conditions to SDS-PAGE resolved wild-type and MUC16-KD SW1990 cell lysates. CHO-E cells were perfused at the wall shear stress level of 0.5 dyn/cm² over the HECA-452 stained Western blots. Number of interacting cells per square millimeter was tabulated as described in Fig. 1. E) Extent of adhesion of wild-type, mammalian scramble control, and MUC16-KD SW1990 cell lines (10⁶/ml) to E-selectin (20 μg/ml) under physiological flow conditions at the wall shear stress levels of 0.5 and 1.0 dyn/cm². F) Extent of tethering of wild-type, mammalian scramble control, and MUC16-KD SW1990 cell lines (10⁶/ml) to L-selectin (20 μg/ml) under flow conditions at shear stress levels of 0.5 and 1.0 dyn/cm². Data represent means ± se of n = 3–5 experiments.*P < 0.05 vs. wild type.

To investigate whether MUC16 is a functional selectin ligand on pancreatic cancer cells, MUC16-KD SW1990 cells were perfused in flow-based adhesion assays. The results showed that SW1990 cells transfected with mammalian scramble control plasmids interacted with E- and L-selectin substrates under flow at levels comparable with those of wild-type controls (Fig. 6E, F). However, MUC16-KD cells displayed a markedly reduced capacity to bind to E-selectin under a shear stress of 0.5 and 1.0 dyn/cm² (∼50% of control; Fig. 6E). Similarly, two distinct MUC16-KD clones exhibited ∼40% reduction in cell tethering to L-selectin relative to appropriate controls under shear (Fig. 6F). To generalize these observations, MUC16 was also knocked down in Pa03C cells by shRNA. The knockdown efficiency was confirmed by flow cytometric and Western blotting analysis (Fig. 7A, B). Immunoblots stained with HECA-452 mAb revealed the absence of an immunoreactive band at the >460-kDa region in MUC16-KD Pa03C cell lysates (Fig. 7C). Blot rolling assays revealed a ∼90% decrease in selectin-dependent adhesion to the >460-kDa region of Western blots of MUC16-KD cell lysate relative to that observed in the corresponding band of the wild-type cell lysate (Fig. 7D). Furthermore, a marked reduction in the number of interactions between MUC16-KD Pa03C cells and E- or L-selectin was observed under flow (Fig. 7E, F). Altogether, these data provide clear evidence that MUC16 is a functional E- and L-selectin ligand in pancreatic carcinoma cells.

Figure 7.

MUC16 is a functional E- and L-selectin ligand on metastatic Pa03C pancreatic cancer cells. A) Representative flow cytometric histograms of MUC16 and CD29 surface expression by wild-type, mammalian scramble control, and 2 clones of MUC16-KD Pa03C cells. Cells were stained by indirect single-color immunofluorescence using the anti-MUC16 or CD29 mAbs and their isotype control antibodies. B) Western blot of whole-cell lysate from wild-type and MUC16-KD Pa03C cells using anti-MUC16 or anti-β-actin antibody. C) Western blot of whole-cell lysate from wild-type and MUC16-KD Pa03C cells using HECA-452 or anti-β-actin antibody. D) Selectin-dependent adhesion under flow conditions to SDS-PAGE resolved wild-type or MUC16-KD Pa03C cell lysates. CHO-E cells were perfused at the wall shear stress level of 0.5 dyn/cm² over the anti-MUC16 mAb-stained Western blots. E) Extent of adhesion of wild-type and MUC16-KD Pa03C cell lines (10⁶/ml) to E-selectin (20 μg/ml) at the wall shear stress levels of 0.5 dyn/cm². F) Extent of tethering of wild-type and MUC16-KD Pa03C cell lines (10⁶/ml) to L-selectin (20 μg/ml) at the wall shear stress level of 0.5 dyn/cm². Data represent means ± se of n = 3 experiments. *P < 0.05 vs. wild type.

DISCUSSION

CA125, an epitope expressed on the extracellular domain of MUC16, is the focus of active research as a clinical marker for the diagnosis of epithelial cancers, due to its selective up-regulation in tumor tissue (23, 24). It is presently the only serum biomarker used routinely for the diagnosis and prognosis of ovarian cancer (38). Recent work reveals that MUC16 is nearly absent from normal pancreatic ducts but is strongly up-regulated in pancreatic cancer (25), suggesting a role for MUC16 in the progression of this disease. As for a biological mechanism underlying this clinical presence, MUC16 has been implicated in different adhesive processes. It has been shown that MUC16 binds mesothelin, which is expressed on the surface of peritoneal mesothelial cells, providing an adhesive capacity for metastasizing carcinoma cells (26). In addition, recent studies have reported that MUC16 may act in an immune-suppressing capacity by binding to galectin-1 and Siglec-9, both of which are expressed by human immune cells. This binding may mask MUC16-expressing tumor cells from immunosurveillance (39, 40).

While the adhesive properties of MUC16 have been broadly appreciated as regulators of tumor progression, the potential interaction between MUC16 and selectins in the vascular system has been largely overlooked. Several lines of evidence reveal the role of selectins and their cognate ligands during metastasis (2, 8, 41, 42), both of which are important for a thorough understanding of the metastatic process. Although different types of tumor cells may express selectin-binding sialofucosylated glycans on distinct protein backbones, identifying these specific ligands, especially those overexpressed by tumor cells, is important for the design of chemotherapeutics that target circulating and disseminating tumor cells.

sLe^a antigen, also known as CA19–9, is the only diagnostic marker approved by the U.S. Food and Drug Administration for pancreatic cancer (43), and its expression can mediate selectin-dependent carcinoma cell adhesion to vascular endothelium during hematogenous metastasis (12, 17). In addition, sialylated oligosaccharides have been reported to be overexpressed on mucins from pancreatic tumors (44). The importance of the mucin family in selectin adhesion has been indirectly demonstrated, as treatment of cultured pancreatic cancer cells with the mucin-rich sera of pancreatic cancer patients reduced tumor cell binding to E-selectin (37). sLe^a was shown to be vital in this interaction through treatment of SW1990 cells with anti-sLe^a antibodies. This intervention resulted in the dramatic reduction of SW1990 binding to E-selectin-expressing endothelial cells and inhibition of liver metastasis (36). Nevertheless, the protein carriers of these sLe^a epitopes remained unidentified before our study. Without this vital information, therapeutics aimed at disrupting tumor cell-selectin binding may not be possible, as targeting all sLe^x and sLe^a epitopes would also disrupt leukocyte-selectin interactions (45), thereby compromising an otherwise healthy immune system. In support of this targeting strategy, several therapeutic antibodies or vaccines are designed to target carcinoma mucins, such as MUC1, and tumor mucin-associated oligosaccharides, such as O-linked glycan sialyl Tn antigen, as a strategy for cancer treatment (46). With the use of immunoaffinity chromatography and MS analysis, MUC16 was identified as the sialofucosylated glycoprotein with a molecular mass > 460 kDa and confirmed as a functional ligand of selectins in 2 distinct metastatic pancreatic cancer cell lines.

MUC16 is heavily glycosylated with both O- and N-linked oligosaccharides (47); however, the biological significance of the glycans expressed on MUC16 is not fully understood. It has been shown that the binding of MUC16 to galectin-1 depends on its β-galactose-terminated, O-linked glycans (39), whereas the MUC16-mesothelin interaction is dependent on N-linked oligosaccharides (48). Treatment of pancreatic cancer cells with glycosidases or carbohydrate metabolic inhibitors demonstrated that the sialofucosylated epitopes from MUC16 are displayed on both O- and N-linked glycans. This finding is supported by prior MS data, which suggested the presence of sialyl Lewis epitopes on O- as well as N-linked glycans on MUC16 (47). Taken together, these data suggest that the ability of MUC16 to support selectin-mediated adhesion may help to confer metastatic potential to MUC16-expressing tumor cells.

It has been argued that distinctions must be made between ligands that can bind to selectins under static condition in vitro and functional ligands that interact with selectins under dynamic condition in vivo (18). Consequently, the functional role of MUC16 in selectin-dependent adhesion warrants further exploration. We achieved this by stably silencing MUC16 expression in pancreatic cancer cells, which bound to E-selectin with markedly reduced capacity. These cells also displayed a significantly decreased ability of transiently tethering to L-selectin. To our knowledge, this is the first characterization of the capacity for MUC16 to interact with E- and L-selectin in a dynamic flow environment. It is noteworthy that depletion of MUC16 from SW1990 or Pa03C pancreatic cancer cells resulted in a 40–60% inhibition of cell binding to E- and L-selectin in shear flow. Thus, other molecules may also contribute to the selectin-dependent adhesion of metastatic pancreatic cancer cells. Immunostaining of SW1990 (Fig. 1) or Pa03C (data not shown) cell lysates with HECA-452 mAb reveals two distinct bands: one corresponding to MUC16 at >460 kDa and another at ∼200 kDa. The latter band corresponds to podocalyxin (32, 42, 49) and is capable of supporting selectin-dependent adhesion (Fig. 4B). In addition, SW1990 cells express MUC1 but at lower levels relative to MUC16 (Supplemental Fig. S1C). Thus, MUC1 may also contribute, albeit modestly, to selectin-dependent adhesion under dynamic flow conditions. Finally, we cannot exclude the possibility that glycolipids may play a role in E-selectin-dependent binding (50).

In summary, by employing biochemical/bioengineering approaches involving SDS-PAGE analysis, blot rolling, and flow-based adhesion assays, we demonstrate that MUC16 is a functional E- and L-selectin ligand on pancreatic carcinoma cells. Our findings offer a unifying perspective on the apparent enhanced metastatic potential associated with the overexpression of MUC16 and sialofucosylated structures on many types of tumor cells, including pancreatic carcinoma, and the critical role of selectins in metastatic spread (Table 1). Our data enhance our understanding of this ubiquitous molecule, and warrant further investigation into the development of novel therapies which target MUC16 to combat metastasis.