Nucleolin mediates the binding of cancer cells to L-selectin under conditions of lymphodynamic shear stress

Abstract

Head and neck squamous cell carcinoma (HNSCC) cells bind to lymphocytes via L-selectin in a shear-dependent manner. This interaction takes place exclusively under low-shear stress conditions, such as those found within the lymph node parenchyma. This represents a novel functional role for L-selectin-selectin ligand interactions. Our previous work has characterized as-of-yet unidentified L-selectin ligands expressed by HNSCC cells that are specifically active under conditions of low shear stress consistent with lymph flow. Using an affinity purification approach, we now show that nucleolin expressed on the surface of HNSCC cells is an active ligand for L-selectin. Parallel plate chamber flow-based experiments and atomic force microscopy (AFM) experiments show that nucleolin is the main functional ligand under these low-force conditions. Furthermore, AFM shows a clear relationship between work of deadhesion and physiological loading rates. Our results reveal nucleolin as the first major ligand reported for L-selectin that operates under low-shear stress conditions.

INTRODUCTION

Tissue-specific cell lodgment requires adhesion molecules that are capable of overcoming shear stress forces in the microenvironment. Selectins and their ligands are important mediators of cell-cell and cell-matrix interactions under shear stress (7, 18, 32). When a cancer cell arrives in the lymph node via afferent lymphatic vessels, it must establish adhesive interactions under shear stress conditions (53). Within the lymph node, lymphocytes and macrophages express a transmembrane ligand, L-selectin, which is known to support calcium-dependent, shear-resistant interactions (14, 18). Our laboratory has published work demonstrating that head and neck squamous cell carcinoma (HNSCC) cells bind to lymphocytes and macrophages via L-selectin in a shear-dependent manner (46). Moreover, this interaction takes place exclusively under lower shear stress levels such as those found within the lymphatic vascular compartment (12, 15, 29), in which the time-averaged wall shear stress is estimated to range from 0.08 to 1.5 dyn/cm² (29, 45, 46). The binding at a lower shear stress threshold represents a novel functional role for L-selectin-selectin ligand interactions (2). The HNSCC cells do not express any known L-selectin ligand, although those expressed do display the canonical characteristics of other L-selectin ligands, including calcium sensitivity, sulfation, sialylation, and N-glycosylation (4, 8, 23).

Our work now identifies nucleolin, a protein expressed intracellularly as well as on the cell surface, as an L-selectin ligand active under physiologically relevant conditions of lymphodynamic shear stress, not just in unstressed, immunoaffinity assays (22). Furthermore, we show that the dynamic relationship between the work of deadhesion and loading rate between L-selectin-functionalized atomic force microscopy (AFM) tips and the HNSCC monolayer is similar to that reported for selectin-ligand complexes studied in vitro with AFM (17, 21, 57).

EXPERIMENTAL PROCEDURES

Cell lines.

Head and neck cancer cell lines JHU-SCC-012 and JHU-SCC-019 were a gift from Dr. James Rocco (Boston, MA). These cell lines were developed from tumors in patients diagnosed with squamous cell carcinoma of the upper aerodigestive tract. Cells were maintained in RPMI 1640 medium containing glutamine (HyClone, Logan, UT) supplemented with 10% FBS (Invitrogen, Carlsbad, CA) under standard cell culture conditions. SCC-4 and SCC-9 cell lines, derived from patients diagnosed with squamous cell carcinoma of the tongue, were obtained from American Type Culture Collection (ATCC, Manassas, VA). Cells were cultured as recommended by ATCC in DMEM-F-12 medium (50:50) supplemented with 1× penicillin-streptomycin, 10% FBS, 10 mM HEPES, and 400 ng/mL hydrocortisone under standard cell culture conditions. LS174T colon cancer cell lines from ATCC were cultured in MEM-Earle’s balanced salt solution (HyClone) supplemented with 1× penicillin-streptomycin, 10% FBS, and 1% nonessential amino acids under standard cell culture conditions. Cell lines were routinely checked for mycoplasma by PCR.

Reagents.

Recombinant human L-selectin/Fc chimera was purchased from R&D Systems (Minneapolis, MN). Surfact-Amps Nonidet P-40, succinimidyl-6-(biotinamido)hexanoate (sulfo-NHS-LC-biotin), immobilized NeutrAvidin protein, immobilized Recomb Protein A, immobilized Protein G Plus, bicinchoninic acid (BCA) protein assay kit, SuperSignal West Femto chemiluminescent substrate, and StartingBlock T20 blocking buffer were purchased from Pierce (Waltham, MA). Triton X-100 was purchased from Fisher Scientific (Hampton, NH). NuPage Bis-Tris and Tris-acetate polyacrylamide gels were purchased from Invitrogen. Immun-Blot PVDF membrane, 0.2 µm, was purchased from Bio-Rad Laboratories (Hercules, CA). Vectashield Hardset mounting medium with DAPI was purchased from Vector Laboratories (Burlingame, CA). Hanks’ balanced salt solution (HBSS) and HEPES buffer were purchased from Cellgro (Corning, NY).

Antibodies.

Mouse anti-GAPDH monoclonal GAPDH-71.1 (cat. no. G8795-200UL) antibody was purchased from Sigma-Aldrich. Mouse IgG1 (cat. no. MG100) antibody was purchased from Caltag (Carlsbad, CA). Mouse anti-human cluster of differentiation 44 (CD44) 2C5 (cat. no. BBA10) antibody was purchased from R&D Systems (Minneapolis, MN). Mouse IgG1κ isotype control (cat. no. 557273), mouse IgG2aκ isotype control (cat. no. 553454), mouse anti-human CD66 B1.1 (cat. no. 551354), mouse anti-human CD66c B6.2 (cat. no. 551355), and mouse anti-human CD66d/e COL rabbit anti-nucleolin polyclonal N2662 (cat. no. N2662-200UL) antibodies were purchased from Sigma-Aldrich (St. Louis, MO). Rabbit anti-nucleolin polyclonal ab70493 (cat. no. ab70493) antibody was purchased from Abcam (Cambridge, MA). Mouse anti-nucleolin 4E2 (KAM-CP100) antibody was purchased from Stressgen (San Diego, CA). Mouse anti-nucleolin MS-3 (cat. no. sc-8031) antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit anti-histone H3 (D1H2 XP; cat. no. 4499S) antibody was purchased from Cell Signaling Technology (Danvers, MA). Mouse anti-human CD66d/e (cat. no. 551477) antibody was purchased from BD Biosciences (San Jose, CA). Mouse anti-human carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1; CD66a, clone GM8G5; cat. no. ab26279), mouse anti-human CEACAM6 (CD66c, clone 9A6; cat. no. ab78029), mouse anti-human pregnancy-specific β-1-glycoprotein 1 (PSG1; CD66f, clone BAP3; cat. no. ab26284), and mouse anti-human CEACAM7 BAC2 (cat. no. ab26281) antibodies were purchased from Abcam (Cambridge, MA). Mouse anti-human CD66b 80H3 (cat. no. MCA216T) antibody was purchased from Bio-Rad (Hercules, CA). Mouse anti-human podocalyxin (PCLP, clone 53D11; cat. no. M084-3) antibody was purchased from MBL International (Woburn, MA). ImmunoPure horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (cat. no. 31430) and HRP-conjugated goat anti-rabbit IgG (cat. no. 32460) antibodies were purchased from Pierce. ChromPure Rabbit IgG (cat. no. 011-000-003), R-phycoerythrin (PE)-conjugated donkey anti-rabbit IgG (cat. no. 711-116-152), and PE-conjugated goat anti-mouse IgG (cat. no. 115-115-206) antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). Alexa Fluor 488-conjugated goat anti-mouse IgG antibody (cat. no. A-11029) and Alexa Fluor 647-conjugated goat anti-mouse IgG antibody (cat. no. A-21236) were purchased from Invitrogen.

Affinity chromatography.

Membrane proteins were prepared from JHU-SCC-019 cells. Cells that were grown in suspension were collected by centrifugation and washed in ice-cold PBS three times. Cells were lysed in buffer B (150 mM NaCl; 0.5 mM Tris, pH 10.4; 2 mM CaCl₂; 20 µg/mL PMSF; and 1× Roche complete EDTA-free protease inhibitor cocktail) by Dounce homogenization on ice. Membrane-enriched fractions [pellet 2 (P2)] were produced by differential centrifugation (34). Pellets of membrane-enriched fractions were resuspended in buffer B with 2% Nonidet P-40 for calcium-dependent binding. For calcium-independent binding, membrane-enriched pellets were resuspended in Nonidet P-40 buffer A (2% Nonidet P-40; 150 mM NaCl; 0.5 mM Tris, pH 10.4; 1 mM EDTA; 20 µg/mL PMSF; and 1× Roche complete EDTA-free protease inhibitor cocktail). L-selectin affinity beads were generated by cross-linking 200 µg L-selectin chimera to 100 µL agarose beads using AminoLink Plus Immobilization Kit (Pierce) according to the manufacturer’s instructions. Control beads were generated by substituting PBS for L-selectin. The L-selectin beads or control beads were washed in Nonidet P-40 buffer A or buffer B before use. P2 samples were precleared by incubation with 50 µL control beads before being incubated with 50 µL L-selectin beads overnight at 4°C with rotation. The beads were then washed three times with 500 µL Nonidet P-40 buffer A or buffer B. The beads were transferred to a new tube and eluted in 50 µL of 2× SDS sample buffer. The samples were electrophoresed and stained with Coomassie blue. Bands of interest were sent to the University of Texas Medical Branch Biomolecular Resource Facility for trypsin digestion and matrix-assisted laser desorption ionization-tandem time-of-flight (MALDI-TOF/TOF) mass spectrometry.

Biotinylation of cell surface proteins.

JHU-SCC-019 cells that were grown in suspension were collected by centrifugation and washed in ice-cold PBS three times. Sulfo-NHS-LC-biotin was dissolved in water and added to the cell suspension according to the manufacturer’s instructions. Biotinylation occurred under rotation at 4°C to prevent internalization. Cells were washed, collected by centrifugation, and resuspended in ice-cold lysis buffer (0.5% Nonidet P-40; 0.5% Triton X-100; 150 mM NaCl; 10 mM Tris, pH 10.4; 1 mM EDTA; 1 mM PMSF; and 1× Roche complete EDTA-free protease inhibitor cocktail). The cell lysis suspension was rotated at 4°C for 4 h to overnight and then clarified by centrifugation (10,000 g, 4°C, 2 min). Biotinylated proteins were isolated with NeutrAvidin beads. The beads were then washed three times, transferred to a new tube, and eluted in 2× SDS sample buffer. The samples were electrophoresed and analyzed by immunoblot.

Immunoblots and immunoprecipitation.

Lysates were prepared by rinsing adherent HNSCC cells twice with ice-cold PBS with 1 mM CaCl₂ and 1 mM MgCl₂ on ice and then scraping the cells into lysis buffer on ice. After a brief incubation on ice, the cells were resuspended in the lysis buffer by gentle pipetting. The samples were rotated at 4°C and then clarified by centrifugation (10,000 g, 4°C, 30 min). For immunoprecipitation, lysates were precleared with protein A or protein G agarose beads. The primary antibody of interest was added to the precleared cell lysate, and the mixture was rotated at 4°C overnight. Control immunoprecipitations were performed with isotype antibodies in place of primary antibody. Antibody-antigen complexes were isolated with protein A or protein G agarose beads. The beads were then washed three times, transferred to a new tube, and eluted in 2× SDS sample buffer. The samples were electrophoresed and analyzed by immunoblot.

Quantitative RT-PCR.

RNA isolation and purification were achieved using RNeasy Plus Mini Kit (Qiagen, Valencia, CA), per the manufacturer’s protocol. Quantification of the RNA concentration was performed using the NanoVue Plus spectrophotometer (GE Healthcare Bio-Sciences, Marlborough, MA). Reverse transcription of the RNA was achieved using High-Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems, Foster City, CA), per the manufacturer’s protocol. PCR was then performed on the cDNA with SYBR Green PCR Master Mix (Applied Biosystems) and recorded using the StepOnePlus Real-Time PCR system (Applied Biosystems). The forward and reverse primers for nucleolin and housekeeping gene ribosomal protein L13A (RPL13A) were ordered from Integrated DNA Technologies (Coralville, IA). Nucleolin gene expression levels were normalized to the housekeeping RPL13A gene (11) using the Pfaffl method (44).

Flow cytometry.

Cells were harvested by incubation with EDTA (5 mM EDTA in PBS, 37°C, 15 min) and washed in ice-cold PBS with 1 mM CaCl₂ and 1 mM MgCl₂. Cell suspensions (10⁷ cells/mL in PBS with 5% horse serum) were incubated with anti-nucleolin antibodies or with equivalent amounts of isotype control antibody for 30 min on ice. After washing in PBS, the cells were treated with PE-conjugated secondary antibody. After washing with PBS, fluorescence was detected on a BD FACSArray bioanalyzer (BD Biosciences) or a BD Special Order Research Product cytometer/sorter (BD Biosciences).

Immunofluorescence.

HNSCC cells were seeded at 10,000 cells per well in a 96-well, black-walled plate a day in advance of staining. Cells were washed with Dulbecco’s PBS with Ca²⁺ and Mg²⁺ (DPBS+), fixed in methanol-free 4% paraformaldehyde-DPBS+ for 20 min at room temperature, and washed again in DPBS+. Cells were incubated for 30 min on ice with either 10 μg/mL N2662 anti-nucleolin antibody or rabbit IgG isotype antibody diluted in 0.1% BSA-DPBS+, washed, and then incubated for 30 min on ice with 20 μg/mL Alexa Fluor 488-conjugated goat anti-rabbit antibody diluted in 0.1% BSA-DPBS+. Stained cells were washed three times with PBS and mounted with ProLong Gold with DAPI (Life Technologies) for qualitative imaging under wide-field fluorescence using a Leica DMI 6000 inverted microscope (Leica Microsystems, Wetzlar, Germany).

Microsphere preparation.

Protein A-coated polymeric superAvidin microspheres (9.77-μm diameter) were purchased from Bangs Laboratories (Fishers, IN). Human L-selectin was adsorbed to microspheres (i.e., L-selectin microspheres). These were prepared as previously described (5). In short, microspheres were spun out of stock solution, washed in pH 4.0 TBS, and blocked for 30 min in 1% BSA in DPBS+ at 5 million microspheres per milliliter. Blocked microspheres were then washed in pH 8.0 TBS and allowed to incubate for 1 h in 20 μg/mL human L-selectin in pH 8.0 TBS. L-selectin microspheres were then washed in 1% BSA in TBS before being checked for L-selectin “expression” via flow cytometry. The same procedure was used to generate human IgG isotype control beads.

Shear-dependent binding assays.

L-selectin-coated microspheres (0.5 × 10⁶ microspheres/mL) were perfused in a parallel-plate flow chamber (254-μm channel height, 5-mm channel width) at a wall shear stress of 0.07 dyn/cm² (corresponding to a wall shear rate of 7.7 s⁻¹, or ~8 s⁻¹, at 70°F with an estimated buffer viscosity of 0.095 mPa·s) over cellular monolayers of SCC-4, SCC-9, JHU-SCC-012, and JHU-SCC-019 cell lines. Cells were plated in 5-mm-inner diameter gaskets on cell culture dishes 24 h before experiments. In antibody blockade experiments, each cell monolayer was washed and treated with 10 µg/mL of anti-nucleolin N2662, rabbit IgG isotype control, or culture media 30 min before use. L-selectin microspheres were perfused over each of the three conditions for 3 min while adhesion events were recorded.

In other flow chamber experiments, JHU-SCC-019 cancer cells in suspension were introduced into the flow chamber over L-selectin chimera adsorbed to a plastic surface. Cells were grown under standard culture conditions, harvested by incubation with EDTA (5 mM EDTA in PBS, 37°C, 15 min), and washed with binding buffer (HBSS, 10 mM HEPES, and 2 mM CaCl₂). Cells were resuspended to a final concentration of 10⁶ cells/mL in binding buffer with 5% horse serum. L-selectin plates were prepared by overnight incubation with L-selectin/Fc chimera (0.5 or 1 µg/mL in PBS) at 4°C and blocked with heat-inactivated FBS for 24 h at 4°C. Attachment was assayed by introducing tumor cells into the parallel flow chamber at 0.25 mL/min followed by decreasing flow to 0.025 mL/min (0.07 dyn/cm², ~8 s⁻¹). Nucleolin specificity of interactions was established by pretreating cancer cells with anti-nucleolin or isotype antibodies (30 µg/mL, 20 min on ice). Cells were examined, and digital pictures were taken using a Nikon TE2000-U microscope with Photometrics Cool SNAP EZ digital camera (Roper Scientific, Tucson, AZ) and the NIS-Elements BR software (Nikon, Melville, NY).

Atomic force microscopy.

The mechanical properties between L-selectin-functionalized AFM tips and HNSCC monolayers were studied using a home-built AFM that consists of a detector head (Digital Instruments, Tonawanda, NY) mounted on top of a single-axis piezoelectric positioner with a strain gauge sensor (P841.10; Physik Instrumente, Auburn, MA). This system has a z-axis resolution of a few nanometers and can measure forces in the range of 5–10,000 pN (3, 20, 38, 42). The monitoring of the force reported by the cantilever and the control of the movement of the piezoelectric positioners are achieved by means of two data acquisition boards (PCI 6052E and PCI 6703; National Instruments, Austin, TX) and controlled by custom-written software (LabView, National Instruments; and Igor, WaveMetrics).

To measure the interaction forces, we used cantilevers with a 10-µm latex bead glued to the tip (Novascan Technologies, Ames, IA). We incubated the cantilevers with 50 μL of L-selectin/Fc at 100 μg/mL in 0.1 M NaHCO₃ (pH 8.6) overnight at 4°C. Unbound proteins were removed by rinsing with PBS; BSA (Sigma) at 500 μg/mL in PBS was used to block the exposed surface of the latex bead.

The spring constant of each individual cantilever was calculated using the equipartition theorem (16). The cantilever spring constant varied between 20 and 50 pN/nm. Interaction forces were measured by pressing down the cantilever onto the cell monolayer for ~500 ms and then stretching several hundred nanometers. We obtained at least 30 force-extension curves at random locations on the cell monolayer. We used serum-free HBSS supplemented with 10 mM HEPES, 2 mM CaCl₂, and 1 mM glucose. In the experiments using blocking antibodies, cells were pretreated with different antibodies (25 μg/mL) for 15 min before the AFM measurements. Unless noted, the pulling speed of the different force-extension curves was ~1.0 μm/s.

Statistical analysis.

Quantitative data shown are reported as means ± standard error, unless otherwise indicated in the figure legends. The number of independent experiments, statistical tests used, and P values are reported in the figure legends. Qualitative data (e.g., micrographs and Western blot images) are representative of n = 3 independent images, unless otherwise indicated in the figure legends.

RESULTS

Identification of nucleolin as an L-selectin ligand on HNSCC cells.

Our previous work has shown that HNSCC cells express novel unidentified ligands known to support shear stress-dependent binding to L-selectin (46). The unidentified ligands expressed on the HNSCC cells display canonical characteristics of selectin ligands, including calcium sensitivity and glycosylation (46, 47). To identify this glycoprotein(s), L-selectin affinity beads were used to isolate putative ligands from JHU-SCC-019 membrane preparations. Binding was performed in the presence of calcium or EDTA to identify calcium-dependent interactions. The putative ligands were separated by SDS-PAGE (Fig. 1A). Bands that were calcium sensitive were analyzed by MALDI-TOF/TOF mass spectrometry. Two bands near 100 kDa (Fig. 1A) were identified as nucleolin (Tables 1 and 2 and Fig. 1C). This result is substantiated by an earlier report describing nucleolin as an L-selectin ligand on the KG1a hematopoietic progenitor/stem and the U937 lymphoma cell lines through immunoaffinity experimentation under static conditions, but not dynamic flow conditions, in vitro (22). To confirm the identity of those bands, an immunoblot of the membrane proteins pulled down by L-selectin beads in the presence of calcium or EDTA was performed. This confirmed that L-selectin specifically bound nucleolin from HNSCC cells in the presence of calcium (Fig. 1B). Nucleolin was detected at the protein level in whole cell lysates (Fig. 1, D and E) and at the mRNA level (Fig. 1F) in four HNSCC cell lines (SCC-9, SCC-4, JHU-SCC-012, and JHU-SCC-019) by immunoblots using Western blot analysis and quantitative RT-PCR, respectively. Altogether, these data confirm that HNSCC cells express nucleolin that is captured by L-selectin in the presence of calcium.

Fig. 1.

Nucleolin is isolated from head and neck squamous cell carcinoma (HNSCC) membrane preparation by L-selectin in a calcium-dependent manner. A: L-selectin affinity pulldown of JHU-SCC-019 membrane preparation with 2 mM CaCl₂ (first lane) or 5 mM EDTA (second lane), Coomassie-stained gel. Arrows indicate bands that were identified as nucleolin by mass spectrometry (see Table 1). B: L-selectin affinity pulldown of JHU-SCC-019 membrane preparation with 2 mM CaCl₂ (first lane) or 5 mM EDTA (second lane) [molecular weight marker (third lane); pellet 2 (P2, fourth lane) is the starting material], immunoblot with 4E2 mouse monoclonal anti-nucleolin antibody. C: conserved domains and predicted modifications of nucleolin based on the primary sequence from the Universal Protein Resource (UniProt) database and the ExPASy Bioinformatics Resource Portal. The locations of the tryptic peptides identified by mass spectrometry (mass spec) are marked with black boxes (see Table 2). D: immunoblots of SCC-9, SCC-4, JHU-SCC-019, and JHU-SCC-012 cell lysates with N2662 anti-nucleolin antibody (top) and loading control GAPDH-71.1 anti-GAPDH antibody (bottom). Data are representative of n = 3 independent experiments. E: quantification of the intensity of the nucleolin band at the expected ~100 kDa relative to the intensity of the GAPDH band for each cell line in every independent experiment, using ImageJ. Values are means ± SE; n = 3 independent experiments. P value >0.05 by one-way ANOVA with post hoc Tukey’s HSD test. F: quantitative RT-PCR was performed on HNSCC mRNA using primers designed for detection of nucleolin activity. Data were analyzed using the Pfaffl method and are presented as relative expression to the ribosomal protein L13A (RPL13A) housekeeping gene (n = 3 independent experiments, 3 technical replicates per experiment). CT, threshold cycle. Values are means ± SD. P value >0.05 by one-way ANOVA with post hoc Tukey’s HSD test.

Table 1.

Nucleolin is identified by MALDI-TOF/TOF mass spectrometry from L-selectin-interacting proteins

No.	Protein Identification (Accession No.)	Peptides Matched, n	Sequence Coverage, %	Score*
1	Nucleolin/protein C23 (P19338)	7	13.7	80
2	Nucleolin/protein C23 (P19338)	9	14.5	422

MALDI-TOF/TOF, matrix-assisted laser desorption ionization-tandem time-of-flight.

^*The threshold score for P < 0.05 is 54.

Table 2.

Tryptic peptides identified by mass spectrometry

Start	End	Sequence
No. 1
348	361	FGYVDFESAEDLEK
370	381	VFGNEIKLEKPK
403	419	VTQDELKEVFEDAAEIR
457	466	SISLYYTGEK
577	596	GLSEDTTEETLKESFDGSVR
610	623	GFGFVDFNSEEDAK
681	690	GGFGGRGGFR
No. 2
333	341	NDLAVVDVR
347	361	KFGYVDFESAEDLEK
403	419	VTQDELKEVFEDAAEIR
410	419	EVFEDAAEIR
457	466	SISLYYTGEK
467	473	GQNQDYR
577	596	GLSEDTTEETLKESFDGSVR
599	609	IVTDRETGSSK
610	623	GFGFVDFNSEEDAK

Nucleolin is expressed on the surface of HNSCC cells.

Although it was first reported to be an intracellular protein (19, 43), recent studies have also placed nucleolin on the cell surface (22, 24, 30, 33, 50). Multiple assays were used to confirm that HNSCC cell lines natively express nucleolin on the cell surface. Cell surface proteins were labeled with water-soluble and membrane-impermeable sulfo-NHS-LC-biotin. The biotinylated cells were lysed, applied to NeutrAvidin-agarose, separated by SDS-PAGE, and subjected to immunoblot analysis. Two nucleolin bands near 100 kDa were detected in the fraction of biotinylated cell surface proteins (Fig. 2A). To control for contamination with intracellular proteins, the membranes were also probed with anti-histone H3 antibody (Fig. 2A). No biotinylated histone H3 was identified.

Fig. 2.

Nucleolin is displayed on head and neck squamous cell carcinoma (HNSCC) cell surface. A: JHU-SCC-019 cell surface proteins were biotinylated, and then cells were lysed for detection of nucleolin expressed on the cell surface vs. intracellularly. NeutrAvidin beads captured biotinylated proteins, i.e., those displayed on the cell surface. The flow-through represents the fraction of lysate that was not biotinylated. Left: immunoblot with N2662 rabbit anti-nucleolin antibody, biotinylated nucleolin indicated by arrows. Right: immunoblot with rabbit anti-histone H3 antibody, intracellular protein control. B–D: flow cytometry with anti-nucleolin antibodies confirmed that nucleolin is expressed on the HNSCC cell surface. The expression of nucleolin was assessed on unfixed JHU-SCC-019 cells using flow cytometry with anti-nucleolin antibodies (solid black lines): N2662 rabbit anti-nucleolin (B), 4E2 mouse anti-nucleolin (C), and MS-3 mouse anti-nucleolin (D). Staining with isotype control antibody (dashed gray lines) was used as negative control. E: flow cytometry with N2662 anti-nucleolin antibody (solid black lines) confirms nucleolin expression on different cancer cell lines compared with the isotype control (solid gray lines). Left: SCC-9. Middle left: SCC-4. Middle right: JHU-SCC-019. Right: JHU-SCC-012. F: same as in E, repeated with the ab70493 rabbit anti-nucleolin antibody. Data are representative of n ≥ 3 independent experiments.

Flow cytometry with multiple anti-nucleolin antibodies indicated that the nonpermeabilized JHU-SCC-019 cell line expresses nucleolin on the cell surface, although the different antibody clones had different levels of reactivity (Fig. 2, B–D). The cell surface expression of nucleolin was confirmed in all HNSCC cell lines by flow cytometry using the antibody with the highest apparent reactivity with JHU-SCC-019 cells, the N2662 rabbit polyclonal antibody (Fig. 2E).

Additionally, nucleolin was qualitatively visualized on the cell surface by wide-field fluorescence microscopy. Immunofluorescence imaging of fixed, nonpermeabilized HNSCC cells revealed distinct nucleolin staining on the cell surface, as opposed to the control condition with isotype antibody, in which no signals were apparent (Fig. 3).

Fig. 3.

Immunofluorescence imaging of nucleolin expression on the head and neck squamous cell carcinoma cell surface. A–D: imaging of intact cells using wide-field immunofluorescence microscopy confirmed that nucleolin is expressed on the surface of SCC-9 (A), SCC-4 (B), JHU-SCC-019 (C), and JHU-SCC-012 (D) cell lines. Cells were labeled with N2662 anti-nucleolin antibody followed by Alexa Fluor 488-conjugated secondary antibody (green) and DAPI (blue). E: cells were labeled with negative control isotype control antibody followed by Alexa Fluor 488-conjugated secondary antibody and DAPI. Left: SCC-9. Middle left: SCC-4. Middle right: JHU-SCC-019. Right: JHU-SCC-012. Images are representative of n = 3 independent experiments. All scale bars = 20 µm.

Further flow cytometry experimentation was performed to determine expression levels of HNSCC cell surface proteins that could possess L-selectin ligand activity when appropriately glycosylated, as previously reported for colon cancer cell lines at hematogenous shear stress levels: CD44, CD66/CEACAM (also known as CEA), splice variants of CD66/CEACAM, and podocalyxin (PCLP, also known as PODXL; 4, 9, 51, 52). As shown in Table 3, all four HNSCC lines express CD44, CD66, and PCLP but do not express CD66b or CEACAM7. The cell lines variously express other CD66 splice variants (Table 3). All relative protein expression levels on the LS174T colon cancer control cell line (Table 3) are consistent with prior publications (4, 9, 51, 52).

Table 3.

Flow cytometry screening for cell surface expression of other potential L-selectin ligands

	Geometric Mean Fluorescence Intensity
	SCC-9	SCC-4	JHU-SCC-019	JHU-SCC-012	LS174T
CD44 (2C5)	55,460 ± 7,791 (100 ± 0)	13,811 ± 872 (100 ± 0)	73,473 ± 18,514 (100 ± 0)	43,529 ± 27,460 (100 ± 0)	23,617 ± 3,950 (100 ± 0)
CD66 (B1.1)	3,242 ± 288 (92 ± 3)	3,815 ± 518 (87 ± 5)	1,131 ± 424 (61 ± 14)	408 ± 112 (47 ± 4)	10,778 ± 839 (100 ± 0)
CD66a (GM8G5)	303 ± 49 (33 ± 7)	597 ± 33 (45 ± 5)	167 ± 24 (14 ± 2)	130 ± 25 (3 ± 2)	104 ± 12 (5 ± 4)
CD66b (80H3)	134 ± 6 (2 ± 0)	214 ± 15 (3 ± 1)	110 ± 14 (2 ± 1)	117 ± 27 (2 ± 2)	59 ± 7 (1 ± 1)
CD66c (9A6)	218 ± 36 (17 ± 5)	415 ± 64 (27 ± 0)	138 ± 24 (8 ± 1)	196 ± 41 (13 ± 4)	1,472 ± 106 (79 ± 2)
CD66c (B6.2)	266 ± 50 (23 ± 7)	590 ± 81 (41 ± 2)	173 ± 34 (14 ± 3)	214 ± 26 (16 ± 6)	2,335 ± 200 (89 ± 1)
CD66d/e (Col-1)	140 ± 14 (5 ± 2)	302 ± 53 (12 ± 4)	139 ± 29 (3 ± 2)	180 ± 62 (28 ± 24)	4,460 ± 652 (97 ± 1)
CD66f (BAP3)	185 ± 12 (6 ± 1)	300 ± 51 (18 ± 13)	377 ± 48 (60 ± 6)	208 ± 67 (15 ± 9)	97 ± 5 (2 ± 1)
CEACAM7 (BAC2)	145 ± 13 (3 ± 1)	186 ± 18 (3 ± 1)	128 ± 17 (4 ± 1)	135 ± 32 (4 ± 3)	67 ± 18 (3 ± 2)
PCLP (53D11)	4,041 ± 1,262 (94 ± 5)	30,649 ± 2,966 (100 ± 0)	38,835 ± 8,625 (100 ± 0)	40,135 ± 8,694 (100 ± 0)	925 ± 101 (79 ± 1)

Values are means ± SE; positive population percentages are in parentheses; n = 3 independent experiments. Head and neck squamous cell carcinoma (HNSCC) cells or LS174T colon cancer cells (control line) were incubated with various primary antibodies [cluster of differentiation 44 (CD44), 2C5; CD66, B1.1; CD66a, GM8G5; CD66b, 80H3; CD66c, 9A6; CD66c, B6.2; CD66d/e, Col-1; CD66f, BAP3; carcinoembryonic antigen-related cell adhesion molecule 7 (CEACAM7), BAC2; and podocalyxin (PCLP), 53D11] followed by isotype-matched Alexa Fluor 647-conjugated secondary antibodies to screen for surface expression of proteins that may possess L-selectin ligand activity (4, 9, 51, 52). CD44 (2C5), CD66 (B1.1), and PCLP (53D11) are highly expressed on SCC-9, SCC-4, JHU-SCC-019, JHU-SCC-012, and LS174T cell lines, but CD66c (B6.2), CD66c (9A6), and CD66d/e (Col-1) are only highly expressed on the LS174T cell line. None of the cell lines express detectable levels of CD66b (80H3) or CEACAM7 (BAC2), whereas expression levels varied among the lines for the remaining proteins.

Nucleolin mediates L-selectin binding under low shear stress.

Using a parallel plate flow chamber, we investigated the ability of anti-nucleolin antibodies to block L-selectin microsphere binding to HNSCC cell monolayers under low shear stress. L-selectin microspheres showed reduced adhesion to N2662-treated monolayers of SCC-9, SCC-4, JHU-SCC-019, and JHU-SCC-012 cells compared with isotype control-treated cells (Fig. 4B).

Fig. 4.

Antibodies to nucleolin block the binding of head and neck squamous cell carcinoma (HNSCC) cells to L-selectin under low shear stress. A: flow cytometry detection of L-selectin on protein A microspheres coated with L-selectin (solid line with fill), negative control protein A microspheres coated with human IgG (dashed line), or blank uncoated microspheres (solid line without fill). B: N2662 anti-nucleolin antibody reduced the binding of L-selectin-coated microspheres to HNSCC monolayers under low shear stress. SCC-9, SCC-4, JHU-SCC-019, and JHU-SCC-012 cell lines were cultured to confluence and assessed for shear-dependent L-selectin binding through the perfusion of coated microspheres over the monolayers at wall shear stress of 0.07 dyn/cm² (wall shear rate ~8 s⁻¹, see experimental procedures) in a parallel plate flow chamber. Prior to being perfused, cell monolayers were incubated with fresh cell culture media containing 10 µg/mL anti-nucleolin N2662 antibody, 10 µg/mL rabbit IgG, or plain media for 30 min. Microsphere adhesion events were recorded in real time and then quantified off-line. Adhesion of L-selectin microspheres to antibody-treated and negative isotype control-treated cells was normalized to respective untreated control cell data (SCC-9 = 26.2 ± 3.8 cells/mm², SCC-4 = 23.8 ± 2.3 cells/mm², JHU-SCC-019 = 25.4 ± 1.5 cells/mm², and JHU-SCC-012 = 27.7 ± 1.5 cells/mm²). Data shown are means ± SE; n = 3 independent experiments. Statistical analysis was performed via two-sided, one-way ANOVA with post hoc Tukey’s test comparison. *P < 0.05 vs. isotype control (P = 0.046 for SCC-9, P = 0.011 for SCC-4). **P < 0.01 vs. isotype control (P = 0.004 for JHU-SCC-019). The result for JHU-SCC-012 was above the significance threshold, P = 0.17. C: multiple antibodies to nucleolin inhibited JHU-SCC-019 cell binding to L-selectin at 0.07 dyn/cm² (~8 s⁻¹). Data shown are means ± SE; n = 3 independent experiments. Statistical analysis was performed by t test. The interaction between JHU-SCC-019 cells and the L-selectin/Fc chimera is blocked by pretreating the cancer cells with N2662 rabbit anti-nucleolin antibody (***P < 0.01 vs. isotype control, P = 0.00013), with 4E2 mouse anti-nucleolin antibody (P = 0.061), or with MS-3 mouse anti-nucleolin antibody (**P < 0.01 vs. isotype control, P = 0.006).

The ability of multiple anti-nucleolin antibodies (N2662, 4E2, and MS-3) to block JHU-SCC-019 cell binding to immobilized L-selectin under shear stress was also tested. All three antibodies significantly inhibited the ability of JHU-SCC-019 cells in bulk flow to bind L-selectin (Fig. 4C), confirming the role of cell surface nucleolin as a low-shear stress ligand for L-selectin.

Atomic force microscopy.

Although a trend was observed for decreased L-selectin microsphere binding to anti-nucleolin (N2662)-treated JHU-SCC-012 monolayers compared with isotype control-treated monolayers, the difference was not statistically significant. These cells may express other L-selectin ligands that are functionally overlapping or redundant in multiple-bond adhesion phenomena typically encountered in flow chamber assays. Thus, the role of nucleolin as an L-selectin ligand under applied force was investigated using single-molecule force spectroscopy in AFM. An L-selectin-coated bead was brought into contact with a monolayer of unfixed, intact JHU-SCC-012 cells (Fig. 5A), creating a compression force that increased gradually. The maximum compression force was set to 100 pN. After ~0.5 s, the AFM cantilever was retracted from the cell layer at a speed of 1.0 µm/s.

Fig. 5.

Atomic force microscopy (AFM) measurements of the adhesion forces between L-selectin and head and neck squamous cell carcinoma (HNSCC) cells. A: schematic of a typical AFM experiment. An L-selectin-coated bead is brought into contact with a JHU-SCC-012 cell monolayer. The maximum compression force was set to ~100 pN. After ~0.5 s, the AFM cantilever is retracted from the cell layer at a speed of 1 μm/s. The arrows show the direction of cantilever movement. B: typical force-distance curves obtained between HNSCC cells and L-selectin-coated beads. The dashed lines indicate zero force. The top force-distance curve was obtained in standard recording solution; the approach and retraction traces are marked by the arrows. The shaded area in the top trace indicates the work of deadhesion. Adhesion is significantly reduced by antibodies against nucleolin and L-selectin (2nd and 3rd traces, respectively). C: work of deadhesion between L-selectin beads and the HNSCC monolayer using different antibodies. Statistical analysis was performed by t test, with a P value of <0.05 considered statistically significant. *P < 0.05 compared with control (medium); ns, not significant (P > 0.05; P = 0.023). D: effect of the separation speed on the work of deadhesion. Data are shown as means ± SE; n = 30 observations in each case.

Figure 5B shows typical force-separation curves for HNSCC monolayer and L-selectin-coated beads obtained under different experimental conditions (Fig. 5B). The work of deadhesion, as indicated by the shaded area in the top recording of Fig. 5B, represents the work done by the cantilever to break adhesive bonds, which is obtained by integrating the force over the separation distance. It includes the work needed to break molecular linkages and the viscoelastic deformation of the cell. The detachment of the cantilever from the cell typically occurs in steps, which may correspond to the rupture of one or more L-selectin-ligand bonds. However, most of these events are obscured by the viscoelastic properties of the cell. The unbinding forces of individual selectin-carbohydrate bonds are in the range of 20–100 pN (13, 57). We estimate that the top trace in Fig. 5B may represent the unbinding of 10–20 L-selectin-ligand bonds. Our data show that it takes, on average, 1 × 10⁻¹⁵ J of work for the L-selectin-functionalized cantilever to detach from a monolayer of JHU-SCC-012 cells. Similar values for adhesion energies have been reported for E-selectin- and VCAM-1-mediated endothelial cell interactions (58, 60). We found that the work of deadhesion reduces to ~4 × 10⁻¹⁶ J in the presence of anti-nucleolin or anti-L-selectin antibodies (Fig. 5C). This difference is statistically significant with a P value <0.05.

Previous studies have shown that the work of deadhesion between selectins and their ligands depends on their separation rates (17, 21, 57). We found a nonlinear dependence of the deadhesion energy and the separation speed. Figure 5D shows the deadhesion work as a function of the logarithm of the separation speed and the loading rate (Fig. 5D, inset). Above ~1 μm/s a 10-fold increase in pulling speed increases the deadhesion energy by ~5-fold. Below 1 μm/s a 10-fold decrease in speed decreases the deadhesion energy by only ~2-fold. This nonlinear dependence of the unbinding energy and pulling speed is indicative of a change of unbinding mechanism at lower speeds, as, for example, reported for selectins and sialyl Lewis X (13, 57).

DISCUSSION

For many cancers, including colon cancer, breast cancer, prostate cancer, and HNSCC, the presence of lymph node metastasis is among the most important prognostic factors for staging and for determining appropriate treatment (49, 54, 62). Lymph node metastasis characterizes tumors that are more aggressive, predicts resistance to therapy, and indicates the likelihood of distant metastasis (36). For patients with HNSCC, lymph node metastasis is associated with a 50% decrease in survival (39). More than half of patients with HNSCC present with regional lymph node disease (25, 55).

Thus far, research efforts have provided insight into how the tumor cells enter afferent lymphatic vessels and ultimately enter the lymph node (10, 41, 48). Little, however, is known about the molecular mechanistic details supporting initial tumor cell lymph node interactions. Our work puts forth a distinct ligand-receptor interaction specifically active under conditions of lymphodynamic shear stress as is found within the lymph node’s lymphatic compartment. As such, it provides an initial avenue by which to study this process at a molecular level.

Prior to this study, the specific receptor-ligand interactions that are responsible for binding under the low range of lymphodynamic shear stress have not been examined. We have identified nucleolin as a ligand on head and neck squamous cell carcinoma (HNSCC) cells responsible for the major portion of low shear stress-dependent binding to L-selectin. Our biochemical and imaging data show that a portion of nucleolin in these cells is localized on the outer leaflet of the plasma membrane. Furthermore, we show that pretreating HNSCC cells with antibodies to nucleolin blocks their ability to bind L-selectin under shear stress in the flow chamber and under applied force in AFM. With the latter, our laboratory has previously shown that HNSCC cells tether to L-selectin at wall speeds between 3 and 100 μm/s (46), which is in the range of the pulling speeds used in our AFM experiments. We found that the work required to disrupt binding between the L-selectin-functionalized cantilevers and the monolayer of HNSCC cells increases with the separation speed (Fig. 5D), a characteristic of all known selectin ligands (1, 13, 57). However, this relationship is nonlinear. There are two regimes: a slow-loading regime below ~2 × 10⁴ pN/s where the force is weakly dependent on the pulling speed and a fast-loading regime above ~2 × 10⁴ pN/s where there is a stronger dependence on the pulling speed. This complex dynamical response is similar to those found for selectins-carbohydrate ligands (13, 57), leukocyte function-associated antigen-1 (LFA-1)-ICAM-1 (59), and α₅β₁-integrin-fibronectin (35) interactions. One possible explanation for the nonlinear relationship is that L-selectin undergoes a conformational change triggered by mechanical force, similar to the “catch”-to-“slip” transition observed for P-selectin binding to its ligand (40, 61).

Selectins and their ligands have been implicated in cancer metastasis (31). For example, sialyl Lewis a, as well as its positional isomer sialyl Lewis X, serves as a ligand for selectins and facilitates hematogenous metastasis (26, 27). More recently, it has been shown that human HT 29 colon cancer cells formed fewer spontaneous lung metastases in E- and P-selectin-deficient severe combined immunodeficient (scid) mice than in wild-type scid mice (28). These data describe interactions that take place at shear stress levels consistent with hematogenous flow rates. Our description of nucleolin as a ligand for L-selectin on HNSCC cells is unique in that the interaction takes place at lower shear stress levels that are experimentally consistent (i.e., within experimental uncertainty) with lymphodynamic flow, wherein time-averaged wall shear rates are estimated to range from 0.08 to 1.5 dyn/cm² (29, 45, 46). Interestingly, HNSCC predominantly metastasizes to lymph nodes, and to the best of our knowledge, our report is the first of a specific low shear stress-dependent ligand for L-selectin. Nucleolin was previously identified through immunoaffinity techniques as an L-selectin on KG1a and U937 cells (22), but experimentation was not performed under flow conditions. In our previous investigation of low-shear binding of HNSCC cells to L-selectin, we evaluated CD34 and CD44 as possible L-selectin ligands, since both have been reported to operate as ligands at higher shear levels consistent with hematogenous flow (46). CD34 was not expressed on the cell surface, and although the HNSCC cells tested expressed CD44, the hematopoietic cell E-/L-selectin ligand (HCELL) glycoform responsible for CD44’s selectin ligand activity was not detected (46). As revealed in Table 3, all four HNSCC lines in the present study express CD66/CEACAM (and, to varying degrees, its splice variants) plus PCLP, proteins first reported as L-selectin ligands on colon cancer cell lines at hematogenous shear stress levels (9, 51, 52). Characterization of these molecules and evaluation of new candidates as low-shear L-selectin ligands are warranted in future, since faint L-selectin-reactive, Ca²⁺-dependent bands were found in affinity chromatography of the JHU-SCC-019 cell line (Fig. 1) and function-blocking nucleolin antibodies were not completely inhibitory to L-selectin adhesion in multiple cell lines (Figs. 4 and 5).

Whereas the significance of cell surface nucleolin continues to be documented, the mechanism by which nucleolin is trafficked to the cell surface remains a mystery. One study has shown that trafficking of nucleolin to the cell surface in HeLa cells was reduced by low temperatures and low serum, but not by transport inhibitors (brefeldin A, monensin, methylamine, or A23187 Ca²⁺ ionophore; 24). Nucleolin expressed on the cell surface is reported to be glycosylated. Specifically, cell surface nucleolin discovered on Jurkat cells is N-glycosylated (6), and N-glycosylation of nucleolin is necessary for surface expression in Jurkat, HuT 78, and PC3 cells (37). This is in keeping with our previously published report demonstrating that N-glycosylation is necessary for HNSCC cells to bind L-selectin under shear stress (46).

Future work aimed at characterizing the signal pathway(s) initiated by nucleolin-L-selectin binding may increase our understanding of how cancer cells thrive in the lymph node. Additionally, animal studies assessing a direct role for nucleolin-L-selectin interactions in the development of lymph node metastasis will be critical. Our attempts at knocking down nucleolin expression for purposes of functional assessment in vitro and in vivo have confirmed an essential cellular function for nucleolin, thus precluding this approach toward in vivo testing in animal models of lymphatic metastasis. We are currently working on alternative approaches to circumvent this problem. Studying lymphatic metastasis may provide new diagnostic tools or therapeutic targets to improve the poor prognosis associated with regionally metastatic disease.

GRANTS

These studies were funded by NIH National Cancer Institute Grant T32-CA-117834 to T. M. Goldson, NIH Grant R01-DK-073394 and the John Sealy Memorial Endowment Fund for Biomedical Research to A. F. Oberhauser, an Honors Tutorial College Apprenticeship award to K. L. Turner and M. M. Burdick, a John J. Kopchick Translational Biomedical Sciences award and a Provost’s Undergraduate Research Fund award to E. G. Caggiano, an Ohio University Baker Fund award and 1804 Fund award to M. M. Burdick, and NIH National Cancer Institute Grant K08-CA-13298801A2 and the Howard Hughes Medical Institute Early Physician Scientist award to V. A. Resto.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

T.M.G., A.F.O., M.M.B., and V.A.R. conceived and designed research; T.M.G., K.L.T., Y.H., G.E.C., E.G.C., A.F.O., S.M.F., and M.M.B. performed experiments; T.M.G., K.L.T., Y.H., G.E.C., E.G.C., A.F.O., S.M.F., M.M.B., and V.A.R. analyzed data; T.M.G., K.L.T., M.M.B., and V.A.R. interpreted results of experiments; T.M.G. and K.L.T. prepared figures; T.M.G., K.L.T., M.M.B., and V.A.R. drafted manuscript; T.M.G., K.L.T., M.M.B., and V.A.R. edited and revised manuscript; T.M.G., K.L.T., Y.H., G.E.C., E.G.C., A.F.O., S.M.F., M.M.B., and V.A.R. approved final version of manuscript.