|Home | About | Journals | Submit | Contact Us | Français|
The expression of N-cadherin (NCAD) has been shown to correlate with increased tumor cell motility and metastasis. However, NCAD-mediated adhesion is a robust phenomenon and therefore seems to be inconsistent with the “release” from intercellular adhesion required for invasion. We show that in the most invasive melanoma and brain tumor cells, altered posttranslational processing results in abundant nonadhesive precursor N-cadherin (proNCAD) at the cell surface, although total NCAD levels remain constant. We demonstrate that aberrantly processed proNCAD promotes cell migration and invasion in vitro. Furthermore, in human tumor specimens, we find high levels of proNCAD as well, supporting an overall conclusion that proNCAD and mature NCAD coexist on these tumor cell surfaces and that it is the ratio between these functionally antagonistic moieties that directly correlates with invasion potential. Our work provides insight into what may be a widespread mechanism for invasion and metastasis and challenges the current dogma of the functional roles played by classic cadherins in tumor progression.
During tumor progression, a subset of primary tumor cells undergoes molecular changes leading to an increased ability to survive, proliferate, invade, and often form secondary metastases [1,2]. The mechanisms governing invasion and metastasis are complex and poorly understood. However, it is recognized that, at the cell surface, alterations in classes of adhesion molecules are critical for detachment of tumor cells, mobility through host tissue, and the successful formation of secondary sites [1,3]. These alterations involve not only reduction in surface adhesion molecules but also changes in the profile of adhesion molecule expression.
Classic cadherins are key cell adhesion molecules (CAMs) in epithelia that mediate Ca2+-dependent and generally homophilic intercellular interactions [4,5]. The precursor form of classic cadherins contains a signal sequence that is cleaved to reveal a prodomain of 130 amino acids, which lacks the essential structural features for adhesion . This explains why its presence before cleavage protects from intracellular cadherin interactions [4,7,8]. Processing of the prodomain is necessary to generate functional cadherins at the cell surface .
Classic cadherins play important roles in cancer pathogenesis [3,9,10], and the metastatic potential of tumor cells inversely correlates with cadherin expression [11–14]. An E-cadherin (ECAD)-to-N-cadherin (NCAD) switch has also been shown to take place in several types of carcinomas. For example, in melanoma, malignant vertical growth phase (VGP) cells lose ECAD expression, whereas NCAD levels significantly increase [15,16] and persist throughout transformation. In general, loss of ECAD correlates with high tumor grades and poor prognosis [17,18], and the up-regulation of NCAD correlates with induced cellular motility [19–21].
How does loss of ECAD and the up-regulation of NCAD promote tumor cell invasion and metastasis? Because ECAD functions in “anchoring” normal cells in place , the interpretation has been that loss of ECAD results in the disruption of adhesion junctions between adjacent cells allowing malignant cells to detach from the “ECAD” epithelial cell layer and invade the host tissue. The up-regulation of adhesively competent NCAD is thought to mediate adhesion of malignant cells to NCAD-expressing stromal or endothelial cells, rather than epithelial cells, facilitating invasion of tumor cells to distant sites [23,24].
Other tumors do not undergo an ECAD-to-NCAD shift but exhibit persistence of NCAD in their component cells normally, as well as in the highly malignant state. An interesting model is primary brain tumors, which arise from cells derived from the primitive neuroepithelium . Glioblastoma multiforme is the most aggressive type of malignant glioma . These tumors invade throughout vital brain regions as single cells, with a predilection for existing anatomic structures, such as white matter tracts, the subpial glial space, and the periphery of neurons and blood vessels, and almost never metastasize outside the brain [27,28]. Glioma cell invasion depends on complex interactions, and possibly cooperation with resident brain cells [29–31], and likely correlates with CAM profiles. An up-regulation of NCAD in malignant glioma cells compared with normal brain tissue has been observed .
We studied the cellular localization and functional state of cell surface expressed NCAD in primary glial tumors and during melanoma transformation. Our results demonstrate that in highly invasive glioma cells and during malignant melanoma transformation, in addition to mature NCAD, significant amounts of nonadhesive precursor N-cadherin (proNCAD) are present on the cell surface. It seems that high levels of surface proNCAD promote detachment, tumor cell migration, and invasion. Furthermore, we detect high levels of proNCAD expression in a panel of human primary and metastatic tumors. Mechanistically, we demonstrate that the differential expression of furin, a common proprotein convertase (PC) enzyme  in tumor cells, seems to be implicated in this phenomenon.
Human brain tumor samples were obtained from the Montreal Neurological Institute Brain Tumor Tissue Bank. The institutional ethics committee approved all experimental procedures carried out on human tissue. Melanoma, breast, squamous cell, and prostate carcinoma tissue samples and arrays, in addition to corresponding normal tissue, were obtained from FolioBio (Columbus, OH). Metastatic melanoma arrays were from Imgenex (San Diego, CA).
Human WM115 and WM266 melanoma cell lines, L cells, and HeLa cells were purchased from American Type Culture Collection (ATCC, Manassas, VA). WM115 is a VGP melanoma, and WM266 is a metastatic melanoma isolated from the same patient. The human U343 glioma cell line was generously provided by A. Guha (University of Toronto, Ontario, Canada), and the U251 glioma cell line was generously provided by R. Bjervig (University of Bergen, Norway). U343, U251, and L cells were cultured in Dulbecco modified Eagle medium (DMEM; Invitrogen, Burlington, Ontario, Canada) supplemented with 10% fetal bovine serum (FBS; Invitrogen). Human WM115 and WM266 cells were cultured in modified Eagle medium (Invitrogen) supplemented with 2 mM l-glutamine, Earle's balanced salt solution, and 10% FBS and were adjusted to contain 1.5 g/L sodium bicarbonate, 0.1 mM nonessential amino acids and 1.0 mM sodium pyruvate. All cell lines were cultured in 100 U/ml penicillin and 100 mg/ml streptomycin and maintained at 37°C in a humidified atmosphere of 5% CO2.
NCAD- or ECAD-expressing mouse L cells were previously described . Lipofectamine Plus transfection reagent (Invitrogen) was used to transfect WM115, WM266, U343, and HeLa cells with NCAD constructs and to transfect U251 cells with V5-tagged furin complementary DNA (cDNA) . For the selection of stable cell lines, cells were seeded in complete DMEM containing 600 µg/ml of Geneticin (Invitrogen), the day after transfection. Colonies were isolated and examined for proNCAD-myc or myc-tagged mock vector, or proNCAD-GFP, or GFP-tagged mock vector by immunocytochemistry.
The following primary antibodies were used for immunoblots, immunocytochemistry, and immunohistochemistry: rabbit affinity-purified polyclonal anti-NCAD cytoplasmic domain, and anti-proN  (generated in D.R.C. laboratory), rat monoclonal anti-NCAD extracellular domain (NEC2) (Dr. M. Takeichi, RIKEN, Japan), mouse monoclonal anti-GFP (BD Biosciences, Mississauga, Ontario, Canada), mouse monoclonal anti-myc (clone 9E10; Sigma-Aldrich, Oakville, Ontario, Canada), rabbit polyclonal anti-furin (generated in N.G.S. laboratory), mouse monoclonal anti-ERK, mouse monoclonal anti-tubulin, rabbit polyclonal anti-nestin, mouse monoclonal anti-β-catenin, mouse antiactin, and fluorescent-conjugated secondary antibodies were from Millipore (Billerica, MA). Mouse anti-Na+/K+-ATPase was from Abcam (Cambridge, MA).
Fluorescence mounting medium (DAKO, Burlington, Ontario, Canada) was used to mount coverslips on glass slides. Factor Xa and 5-fluoro-2′-deoxyuridine (FDU) were from Sigma-Aldrich Canada Ltd.
For protein extraction, subconfluent monolayers were washed with PBS, dissociated using 2 mM EDTA in PBS (as previously mentioned), and pelleted at 1000 rpm for 5 minutes. Lysates were obtained using RIPA lysis buffer (150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1% NP-40, 1% Triton X-100) with Complete Mini protease inhibitors (Roche Diagnostics, Laval, Quebec, Canada) on ice for 30 minutes. After cell lysis, samples were centrifuged for 15 minutes at 15,000 rpm, and the supernatants were transferred to clean tubes. Protein concentration was determined using the Lowry assay (Bio-Rad DC protein assay, Mississauga, Ontario, Canada), and samples were run on a 4% to 15% linear gradient SDS-PAGE gel (Bio-Rad), transferred to nitro-cellulose, membrane-blocked with 5% milk protein, and incubated overnight with primary antibodies at 4°C. Blots were then incubated with HRP-conjugated secondary antibodies, and routine washes were carried out. Blots were developed with the chemiluminescence system (Pierce Biotechnology, Rockford, IL). Alternatively, for signal quantification, the chemifluorescence kit (Pierce Biotechnology) and the Storm Imager were used. An independent loading control (actin or Na+/K+-ATPase) was used to normalize signals analyzed by densitometry. Densitometric analysis was carried out using the Image J software (National Institutes of Health, Bethesda, MD). Bands were boxed, and background signal was subtracted from their relative intensities. Intensity values were normalized to reference values (loading control).
Subconfluent monolayers were washed three times with ice-cold PBS containing 2 mM MgCl2 and incubated with 0.2 mg/ml EZ-Link NHS-SS-Biotin (Pierce Biotechnology) solution in PBS for 30 minutes at 4°C to inhibit endocytosis. Excess biotin was quenched by washing three times with ice-cold TBS (25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2 mM MgCl2, and 2 mM CaCl2) followed by three washes with ice-cold PBS. Cells were scraped off the plate with 0.5 ml of RIPA buffer, and lysis was carried out as previously mentioned, followed by protein concentration determination of lysate supernatants. Immuno-Pure Immobilized Streptavidin beads (Pierce Biotechnology) were added to the total protein, and the volume was brought up to 0.5 ml with RIPA buffer. Binding of biotinylated proteins to streptavidin beads occurred during a 2-hour incubation at 4°C, with gentle rocking. Streptavidin beads were pelleted (13,000 rpm at 4°C), the supernatant was discarded, and beads were washed with 1 ml of RIPA buffer three times. The supernatant from the last wash was discarded and 2x SDS sample buffer containing 100 mM DTT was added to dissociate the biotinylated proteins from the streptavidin beads through reduction of the disulfide bond in the biotin molecule. Samples were run on SDS-PAGE gels, and immunoblot analysis was carried out as outlined previously mentioned. Anti-NCAD cytoplasmic antibody was used to detect total NCAD protein (mature and precursor), anti-proN antibody was used to detect proNCAD, anti-Na+/K+-ATPase was used as a loading control, and anti-ERK was used as a cell surface biotinylation control.
Cells were plated onto poly-l-lysine-coated coverslips in supplemented DMEM (see previous discussion). Cells were fixed in 4% paraformaldehyde, permeabilized in 0.3% Triton X, PBS, and blocked in 5% BSA, 5% goat serum, PBS. Cells were then incubated for 1 hour in primary antibody diluted in 1% BSA, 0.02% Triton X, PBS, followed by a 40-minute incubation in fluorescent-conjugated secondary antibodies. Three washes with PBS were performed before fixation, as well as after each step. Coverslips were mounted and examined by confocal laser microscopy using the Zeiss LSM 510 microscope (Carl Zeiss Canada, Ltd., Toronto, Canada) with the Zen image acquisition software and a 60x oil immersion objective. Images were acquired in the same plane of focus between comparisons.
Live cell staining was carried out by incubating cells plated on coverslips with primary antibody diluted in medium without serum at 4°C for 1 hour. The cells were washed with PBS and fixed in 3.7% paraformaldehyde. After washes with PBS, cells were incubated with fluorescent-conjugated secondary antibody diluted in 1% BSA, 0.02% Triton X, PBS, for 40 minutes at room temperature. Coverslips were then mounted and examined as previously mentioned.
Paraffin-embedded tissue was deparaffinized and rehydrated. The antigen retrieval solution consisted of citric acid pH 6. The tissue was subsequently washed with PBS for 5 minutes, blocked in PBS containing 10% FBS and 0.5% Triton X-100 for 90 minutes, and incubated with primary antibody in blocking solution overnight at 4°C in a humidified chamber. Sections were then washed three times in PBS, incubated in secondary antibody in blocking solution for 90 minutes at room temperature in a humidified chamber, and washed two times in PBS. Slides were mounted and examined by confocal laser microscopy using the Zeiss LSM 510 microscope.
Aggregation assays were carried out as previously described . Briefly, monolayer cultures were treated with 2 mM EDTA in PBS for 5 minutes at 37°C. The cells were then washed gently in HCMF (HEPES-buffered Ca2+-Mg2+-free Hank's balanced salt solution) supplemented with 1 mM CaCl2 and 1% BSA for 30 minutes at 37°C to dissociate the monolayer into single cells while leaving cadherins intact on the cell surface. After cell dissociation, 5 x 105 cells per well were transferred to 24-well low-adherent dishes (VWR, Mississauga, Ontario, Canada), and brought up to a final volume of 0.5 ml of HCMF containing 1% BSA with or without 1 mM Ca2+. The plates were rotated at 80 rpm at 37°C for observation of aggregate formation during a 40-minute time course. At t = 0 minute, t = 20 minutes, and t = 40 minutes, 50 µl of the fixed aggregates was removed, placed on a slide, covered with a coverslip, and examined by light microscopy. The number of single cells was counted, and the percentage of single cells during the aggregation assay was determined by the index Nt/N0, where Nt is the total number of single cells after a certain incubation time, and N0 is the total number of single cells at the initiation of incubation. Where applicable, factor Xa (0.4 U/ml; Sigma) was added before and after cell dissociation.
Cells were seeded in a 96-well plate at a density of 1000 cells per well, and at 24-, 72-, and 120-hour time points, MTT (Sigma-Aldrich) was added at a final concentration of 0.5 mg/ml, and the plate was incubated at 37°C for 4 hours. The medium was then removed, 100 µl of DMSO was added per well, the plate was incubated at 37°C for 1 hour, and the absorption was measured at 595 nm using a spectrophotometer. The optical density of the sample was subtracted from that of the blank, and data were plotted versus time. An exponential growth trend line was applied to the data points yielding the following equation: y(t) = y0ekt where y(t) is the optical density at 595 nm at time point t, y0 is the optical density at t = 0 hour, k is the growth constant, and t = time. The doubling time (td) was calculated using the equation td = ln2 / k.
To assess two-dimensional migration of tumor cell lines, 3 x 105 cells were seeded in six-well culture dishes. Before plating in these dishes, two parallel lines were drawn at the underside of the well with a Sharpie marker, serving as fiducial marks for analysis of wound areas. The cell monolayer was 100% confluent at the day of analysis. The growth medium was first aspirated and replaced by calcium-free PBS to prevent death of cells at the edge of the wound by exposure to high calcium concentrations. Monolayers were disrupted with a parallel scratch wound made perpendicular to the marker lines with a fine pipette tip. Migration into the wound was observed using phase-contrast microscopy on an inverted microscope with the 5x objective. Images were taken at regular time intervals of areas flanking the intersections of the wound and the marker lines. The number of cells that migrated into the wound was determined by marking wound edges at t = 0 hour on the underside of the well and counting cells that migrated into the wound at specified time points appropriate for each cell lines using Northern Eclipse software 6.0 (EMPIX Imaging, Inc, Mississauga, Ontario, Canada). Factor Xa was added at a concentration of 0.4 U/ml, where applicable.
Three-dimensional collagen gel invasion assays. Confluent monolayers of tumor cell lines were dissociated, and spheroids were prepared using the hanging drop method as previously described [36,37]. Spheroids were implanted into four-well culture dishes containing 0.5-ml aliquots of a collagen type I solution (Vitrogen 100; Cohesion, Palo Alto, CA) using a Pasteur pipette. After hardening of the gel at 37°C for 60 minutes, the gel was overlaid with 0.5 ml of supplemented DMEM. Cell invasion was assessed daily using an inverted phase-contrast light microscope. The number of cells invading at increasing distances away from the spheroid on day 5 after implantation was assessed using a concentric grid system (Northern Eclipse 6.0; EMPIX Imaging, Inc). Where applicable, factor Xa was added at 0.4 U/ml during spheroid preparation and postimplantation into the medium overlaying the collagen gel. In addition, where applicable, FDU (Sigma-Aldrich) was added after implantation into the medium overlaying the collagen gel at a concentration of 0.2 µM.
Boyden chamber invasion assays. To assess cellular invasion, 3 x 105 cells were seeded on the upper chamber of Matrigel-coated membranes (8 µm pore size; Millipore). Conditioned medium was made by incubating NIH3T3 cells in DMEM with 0.1% bovine calf serum and 50 µg/ml ascorbic acid for 24 hours and was applied to the bottom chamber, serving as a chemoattractant. The cells were allowed to invade the Matrigel substrate for 12 hours in the tissue culture incubator. The cells were fixed by replacing the culture medium with 4% formaldehyde in PBS for 15 minutes at room temperature. The chambers were then rinsed with PBS and stained with 0.2% crystal violet for 10 minutes followed by another series of rinses. The remaining cells that did not migrate through the membrane pores were removed with several cotton swabs from the upper chamber, and the number of invaded cells was counted using an inverted microscope and a 4x objective. Factor Xa was added at a concentration of 0.4 U/ml, where applicable.
Predesigned, small interfering RNA (siRNA) oligos targeting two different regions of the human furin sequence (nos. 105594 and 112945), glyceraldehyde 3-phosphate dehydrogenase, and Cy3-labeled negative control no. 1 were purchased from Ambion (Austin, TX). U343 cells were transfected with furin siRNA (80 nM), respectively, using Lipofectamine Plus reagent (Invitrogen). Cells were used in experiments 3 days after transfection.
RNA was extracted from human cell lines and frozen tissue samples using RNeasy mini kit (Qiagen, Missisauga, Ontario, Canada). The RNeasy FFPE kit was used to extract RNA from paraffin-embedded tissue samples. cDNA were prepared using the RevertAid H Minus First Strand cDNA Synthesis Kit (Fermentas, Burlington, Ontario, Canada).
Semiquantitative polymerase chain reaction (PCR) was carried out to determine furin expression, and GAPDH was used as a normalizing control. Real-time PCR was carried out to quantify furin expression relative to hS14 expression, as previously described . Primers are listed in Table W1.
Descriptive statistics were analyzed using GraphPad Prism 4. Mean, SEM, and Student's t-test were used to determine significant differences between pairs. Comparisons of migrating and invading single cells and comparisons of furin levels in human tissue samples were performed using a parametric or nonparametric analysis of variance and Tukey or Dunnett multiple comparison tests, respectively. P < .05 was considered significant.
We previously observed that during early development, nonadhesive precursor NCAD (proNCAD) is on the surface of immature neurons (unpublished results), presumably facilitating the “slipping” of membranes past one another. NCAD is also a major CAM during tumorigenesis, as it is expressed on the surface of various types of tumors. The migration and invasion of tumor cells may be considered to be analogous to the motility of membranes during neurite outgrowth; therefore, we hypothesized that changes in NCAD-mediated adhesion due to surface proNCAD expression may be important during tumor progression . It is conceivable that the correlation observed between up-regulation of NCAD during tumor progression and increased tumor cellular motility [19–21] is largely due to the aberrant expression of cell surface nonadhesive proNCAD, decreasing cell-cell adhesion and promoting migration and invasion.
To test the role of proNCAD in tumorigenesis, we used two different tumor models in which NCAD expression plays an important role during tumor progression. We investigated the cellular distribution and functional state of the NCAD molecule at the cell surface in melanoma and malignant glioma. Melanoma is a tumor model that undergoes an ECAD-to-NCAD transition whereby ECAD expression is downregulated in VGP cells and NCAD expression is upregulated and maintained during cancer progression [15,16]. Malignant glioma is a tumor model that exhibits persistence of NCAD during malignant progression, and compared with normal brain tissue, an increase in NCAD expression has been shown in malignant glioma .
We made use of melanoma cell lines representing different stages of transformation. WM115 was derived from VGP melanoma at the primary tumor site, and WM266 was derived from metastatic melanoma at a secondary site from the same patient. We also used the U343 and U251 glioma cell lines, isolated from malignant gliomas. U251 cells are more invasive in a collagen gel implantation assay than U343 cells (Figure W1A). Because these cell lines exhibit significantly different proliferation rates (please refer to Figure W6 and Table W2), we also carried out the invasion assay in the presence of FDU, an antiproliferative agent to exclude the effect of proliferation in the assay, and our results show that the difference in invasion between U251 and U343 cells is maintained (Figure W1B).
We first wanted to evaluate the levels of total NCAD in the glioma and melanoma cell lines. To this end, we made use of an antibody against the cytoplasmic domain of NCAD to detect total NCAD levels in U343, U251, WM115, and WM266 cell lines by Western blot analysis. We found that total NCAD levels were comparable in both glioma cell lines (Figure 1A) and in both melanoma cell lines (Figure 1A). Using a previously characterized rabbit polyclonal antibody specifically against the NCAD prodomain (anti-proN) , we next investigated whether a proportion of the NCAD expressed at the cell surface was unprocessed, in the nonadhesive precursor form. We carried out cell surface biotinylation experiments where proteins at the surface of cells were labeled with biotin and affinity purified using a streptavidin resin and then analyzed by immunoblot analysis. Surprisingly, we found a high proportion of proNCAD on the surface of highly invasive U251 cells compared with the indolent U343 cells in cell surface biotinylation experiments (Figure 1B). Similarly, expression of proNCAD was estimated to be three times greater in metastatic WM266 cells compared with VGPWM115 cells (Figure 1B). We also examined the immunolocalization of proNCAD and found that it could be detected intracellularly in permeabilized glioma and melanoma cells (Figure 1C, top panels). Under nonpermeabilizing conditions, proNCAD was detected on the surface of U251 and WM266 cells and, to a lesser extent, on the surface of WM115 cells (Figure 1C, bottom panels) and coexisted with mature NCAD (Figure W2). Surface proNCAD staining of U343 cells was negligible (Figure 1C, bottom panels), in agreement with our immunoblot analysis.
Our results show that NCAD is an abundant component of melanoma and glioma cell lines and that proNCAD is highly expressed on the surface of invasive U251 cells and metastatic WM266 cells. Because proNCAD is nonadhesive, we would expect U251 and WM266 cells to exhibit lower intercellular adhesive activity compared with U343 and WM115 cells. To test differences in cellular adhesion between these cell lines, we carried out cell aggregation assays. In these assays, cells were dissociated while leaving cadherins intact at the cell surface, and aggregation was monitored over time. We observed greater aggregation in less invasive U343 glioma cells and in WM115 VGP melanoma cells compared with highly invasive U251 cells and WM266 metastatic melanoma cells, respectively (Figures 1D and W3). There was no aggregation in the absence of calcium for all cell lines (data not shown). These results suggest that higher levels of cell surface nonadhesive proNCAD correlate with less cell aggregation in the U251 and WM266 cell lines.
Our data demonstrate an inverse correlation between surface-expressed proNCAD and adhesive behavior. Because NCAD expression correlates with increased motility and proNCAD lacks adhesive function, we hypothesized that loss of adhesion due to aberrant surface expression of proNCAD may serve as a mechanism for enhanced motility in tumor cells, even in the presence of mature NCAD. To address how proNCAD might influence invasion and metastasis in these tumor cells, we engineered an NCAD construct (proNCAD) where the endogenous consensus PC cleavage site [6,39] was replaced with a serum coagulation factor Xa recognition site in the linker sequence (Figure 2A). Glioma and melanoma cell lines were transfected with mutant myc-tagged proNCAD and mutant GFP-tagged proNCAD, respectively, and clones were selected and expanded. The expression of mutant proteins at the surface of transfected cells was verified by immunofluorescence. Myc and proNCAD colocalized extensively at the plasma membrane of transfected glioma cells, and GFP and proNCAD showed a similar localization in transfected melanoma cells (Figure 2B). Cleavage of the prodomain by factor Xa was detected by immunoblot analysis of total cell lysates or conditioned medium of transfected cells. ProNCAD-myc levels were decreased on treatment with factor Xa (Figure W4A), and cleavage did not compromise the integrity of the mature protein (Figure W4B). Accumulation of the 17-kDa fragment in the conditioned medium indicates specific cleavage of the mutant prodomain by factor Xa (Figure W4C).
We investigated the effect of surface proNCAD expression on intercellular adhesion by carrying out cell aggregation assays with the transfected cells described previously. We found that cells transfected with mutant proNCAD formed considerably smaller aggregates compared with mock-transfected cells (Figure 2C). Treatment with factor Xa restored cell-to-cell adhesion, resulting in aggregates comparable to those observed with mock-transfected cells (Figure 2C). Results were quantified as percent of single cells over time, demonstrating low aggregation for transfected cells and high aggregation for mock-transfected cells in the presence or absence of factor Xa, as well as high aggregation of transfected cells in the presence of Xa. Therefore, these data support the notion that cell surface expression of proNCAD leads to a decrease in cell-cell adhesion.
It is conceivable that decreased intercellular adhesion due to surface expression of proNCAD would affect cellular motility. To examine the role of proNCAD in cell motility, we performed a wound healing assay in which confluent monolayers were disrupted by scraping with a fine pipette tip, and migration of cells into the wound was monitored and quantified over time. WM115 and WM266 cells transfected with mutant proNCAD exhibited increased migration during a 6-hour period compared with mock-transfected controls (Figure 3A). This effect was significantly decreased on treatment of proNCAD-transfected cells with factor Xa (Figure 3A). The most dense cell monolayer was observed with WM266 cells transfected with mutant proNCAD because these cells express both transfected and endogenous surface pro-NCAD (Figure 3A). Similarly, U343 cells overexpressing mutant proNCAD exhibited increased migration into the wound during a 12-hour period compared with mock-transfected cells, and this effect was abolished by treatment with factor Xa (Figure 3A). Furthermore, to show the direct involvement of proNCAD in cell migration, we carried out migration assays using HeLa cells, a cell line lacking detectable expression of classic adhesion molecules, including NCAD (Figure W5A) [40,41]. HeLa cells were transfected with mutant proNCAD or mock construct, and experiments were carried out in the presence or absence of factor Xa. Compared with mock-transfected HeLa cells, proNCAD-transfected cells migrated 1.7 times more 10 hours after monolayers were disrupted (Figure W5, B and C). On addition of factor Xa, which cleaves the prodomain at the engineered consensus site rendering the NCAD molecule adhesive, we observed a significant decrease in migration (Figure W5, B and C).
Because the effect of cellular proliferation is not taken into account in migration assays, we tested whether proNCAD expression affects proliferation of transfected cells. We performed MTT proliferation assays where the reduction of tetrazolium salt (MTT) in metabolically active cells to a purple formazan product was quantified spectrophotometrically, and doubling times were calculated. These assays yielded a doubling time of ~23 hours for both mock- and mutant proNCAD-transfected U343 cells, ~28 hours for both mock- and mutant proNCAD-transfected WM115 cells, ~26 hours for both mock- and mutant proNCAD-transfected WM266 cells, and ~24 hours for both mock- and mutant proNCAD-transfected HeLa cells (Figure W6 and Table W2). Therefore, although migration was quantified at time points far from the doubling time of these cell lines, we can exclude the effect of growth in this assay because proNCAD does not significantly alter the proliferation rate of transfected cells.
The implication of proNCAD in cellular migration warranted the investigation of its role in more complex cellular behaviors such as invasion. To this end, cellular invasion was assessed using three-dimensional collagen gel invasion assays and Boyden chamber invasion assays. Spheroids of mutant proNCAD- or mock-transfected glioma cells were prepared using the hanging drop method and implanted into a collagen gelmatrix in the presence or absence of factor Xa, and invasion was quantified on day 5 using a phase-contrast light microscope and Northern Eclipse software (Figure 3B). Compared with the control, transfection with mutant proNCAD resulted in more than double the number of U343 cells invading up to distances greater than 300 µm from the edge of the spheroid. This effect was most pronounced at distances greater than 300 µm, where there were nine-fold more proNCAD transfected cells invading compared with control cells (Figure 3B). This effect was reversed on treatment with factor Xa. Because the melanoma cell lines did not form spheroids, Boyden chamber assays were carried out to assess the effect of proNCAD on invasion of melanoma cells. Cells were seeded in the upper chamber of a Matrigel coated filter, and NIH3T3 conditioned medium was used as a chemoattractant in the lower chamber. Cells that invaded the Matrigel substrate were quantified 12 hours after monolayer disruption using a light microscope. ProNCAD surface expression substantially increased invasiveness relative to the parental WM115 and WM266 lines (Figure 3B). Treatment with factor Xa reduced invasion to levels observed with parental lines. As mentioned above for migration assays, the effect of cellular proliferation is not taken into account in invasion assays. However, we can exclude the effect of growth in these assays because proNCAD does not significantly alter the proliferation rate of transfected cells (Figure W6 and Table W2).
Collectively, our data demonstrate that inhibition of cell-to-cell adhesion by surface-expressed proNCAD promotes malignant tumor cell behaviors such as migration and invasion in glioma and melanoma cells.
Classic cadherins are synthesized as inactive propeptide precursors, which become functional mature proteins on posttranslational processing. The subtisilin-like basic amino acid-specific PCs are a family of Ca2+-dependent endoproteases, responsible for the activation of precursor proteins by cleavage at a consensus recognition site (Arg/Lys-(X)n-Lys/Arg-Arg, n = 0, 2, 4, or 6) [42,43]. ProNCAD, like other classic cadherins , has a consensus cleavage site for furin [6,44], at the C-terminal end of the prodomain. Furin is one of the most common and widely expressed PCs and processes precursor proteins in the constitutive secretory pathway .
Because furin is a ubiquitously expressed PC and a putative convertase that cleaves the precursor form of NCAD, we analyzed its expression in the tumor cells predicting that differences in levels of this enzyme might underlie the mechanism leading to surface expression of proNCAD. Interestingly, furin expression was lower in U251 cells relative to U343 cells (Figure 4A). This difference was quantified by quantitative real-time PCR and found to be more than 30-fold less in U251 compared with U343 cells (Figure 4A). Low furin levels in U251 cells might explain proNCAD surface expression in these cells because the immature N-cadherin protein would not be properly cleaved at the consensus site. High furin levels would be expected to render tumor cells less invasive and more adhesive to one another because N-cadherin would be properly cleaved.
Because we observed a marked difference in furin expression between U251 and U343 cell lines, we predicted that altering the expression of furin would affect the migration and aggregation behaviors of these cell lines. To this end, we carried out knockdown experiments using siRNAs specifically targeting furin and gain-of-function experiments where cells were stably transfected with furin. U343 cells were successfully transfected with siRNA targeting the furin sequence (Figure W7A), and an 80% reduction of furin messenger RNA (mRNA) levels was demonstrated by reverse transcription-polymerase chain reaction (RT-PCR; Figure W7, B and C). Furin siRNA did not affect PC7 or NCAD mRNA levels (Figure W7B). In addition, GAPDH levels were not affected but were reduced by a GAPDH-specific siRNA (Figure W7B). Immunocytochemistry also demonstrated a reduction in furin levels in U343 cells (Figure W7D), but there was no reduction in tubulin, nestin, or β-catenin expression (Figure W7D). Knockdown of furin in U343 cells resulted in a substantial increase in cell migration (Figure 4B) and a decrease in cell aggregation (Figure 4C) compared with control cells. Knockdown of furin also resulted in an increase in surface proNCAD levels under nonpermeabilizing conditions (Figure 4B). In our gain-of-function studies, using a bicistronic pIRES-EGFP vector expressing a C-terminally V5-tagged furin, transfected cells exhibited colocalization of furin immunoreactivity with EGFP fluorescence (data not shown). Compared with mock transfections, overexpression of furin in U251 cells resulted in a substantial decrease in the number of migrated cells at 12 hours in a migration assay (Figure 4B) and a decrease in surface proNCAD levels under nonpermeabilizing conditions (Figure 4B). U251 cells transfected with furin aggregated to a much greater extent compared with control cells (Figure 4C). Because the effect of cellular proliferation is not taken into account in these migration assays, we tested whether siRNA transfections and furin expression affect the proliferation of U343 and U251 cells, respectively. We performed proliferation assays, which yielded a doubling time of ~24 hours for U343 cells transfected with control siRNA and furin-targeted siRNAs. Proliferation assays testing U251 cells transfected with mock or furin constructs yielded doubling times of 22.6 and 22.2 hours, respectively (Figure W6 and Table W2), which were not significantly different. Thus, although migration was quantified at time points far from the doubling time of these cell lines, we can exclude the effect of growth in this assay because neither furin knockdown or overexpression significantly alters the proliferation rate of transfected cells.
These results demonstrate that furin expression seems to inhibit glioma cell migration by affecting cell-to-cell adhesion; conversely, lack of furin expression leads to proNCAD accumulation at the cell surface, a decrease in cell-to-cell adhesion, and an increase in cell migration.
Together, our in vitro results demonstrate that, in highly invasive and metastatic cancer cells, the NCAD protein is aberrantly processed so that the precursor protein is expressed at the cell surface, rendering these cells less adhesive, more migratory, and more invasive. Furthermore, our data suggest an important role for furin in posttranslational processing of the precursor NCAD molecule.
Because N-cadherin is commonly expressed in several human tumors, we investigated the clinical relevance of this phenomenon by screening the expression of proNCAD and furin in a panel of human epithelial derived tumors as well as in corresponding normal control tissues. ProNCAD immunoreactivity was determined in paraffin-embedded human tissue samples and demonstrated an extremely strong direct correlation between tumor grade and proNCAD expression. ProNCAD immunoreactivity was strikingly elevated in high-grade glioma, melanoma, squamous cell carcinoma (SCC), breast carcinoma, and prostate carcinoma (Figure 5A). ProNCAD expression was also elevated in metastases of melanoma, SCC, breast carcinoma, and prostate carcinoma (Figure 5A). Conversely, proNCAD expression was relatively weak in low-grade tumors (glioma, grade 2; SCC, grade 2; breast carcinoma, grade 1; prostate carcinoma, grade 2) and was negligible in corresponding normal control tissues (Figure 5A). NCAD expression was also verified by immunohistochemistry (data not shown).
We also examined furin levels in normal and cancerous human tissues. Quantitative real-time PCR results revealed an overall decrease in furin expression during tumor progression (Figure 5B). More specifically, there was a significant decrease in furin expression when comparing normal dermal tissue with primary melanoma or with metastatic melanoma (Figure 5B). Similarly, there was a significant decrease in furin expression when comparing normal squamous tissue with primary SCC or with metastatic SCC (Figure 5B). There was also a significant decrease in furin expression between normal breast tissue and primary breast carcinoma or metastatic breast carcinoma (Figure 5B). We observed a decrease in furin expression between normal prostate tissue and primary prostate carcinoma, or metastatic prostate carcinoma (Figure 5B). Although proNCAD expression was high in gliomas, furin levels remained relatively constant (Figure 5B), possibly because most invasive cells quickly migrate out of the tumor core .
In summary, we show that cell surface proNCAD potentiates invasiveness in indolent tumor cells. Furthermore, we provide very strong evidence supporting the relevance of our findings in human primary tumors and metastases. Together, our results support the conclusion that during malignant transformation, the extent of tumor cell invasion is determined by the ratio of surface expressed nonadhesive proNCAD to functional NCAD (Figure 6).
Destabilized cell contacts, cellular reorganization, and metastatic dissemination are all associated with changes in cell adhesion. NCAD is a major CAM in normal physiology and during tumorigenesis, and it possesses a range of adhesive strengths. However, this hierarchy of adhesion is thought to be regulated solely by monomer-dimer ratios , “overlapping” domains , clustering , and by mass amounts of cadherin on cell surfaces. Persistence of the prodomain on the surface of tumor cells has not been observed previously, and it is believed that the N-terminal prodomain in classic cadherins is completely removed by members of the PC family, resulting in a mature, adhesively competent molecule at the cell surface [6–8,49]. However, we have observed proNCAD on the surfaces of neurites in the developing brain (unpublished results).
We report here that during malignant transformation, the NCAD molecule undergoes altered proteolytic processing. This results in a mixture of NCAD molecular forms at the cell surface and functionally enhances cellular migration and invasion. Our results show unequivocal proNCAD labeling in a panel of high-grade human tumor tissues. This correlates well with down-regulation of furin RNA levels for all tumors examined except gliomas. In the total pool of cells in normal brain and glioma tissues, we observed relatively constant levels of furin during malignant progression. This may be explained by the fact that these tumors are extremely heterogeneous and that samples were taken from the tumor core, well after detachment and exit of tumor cells with the highest migratory potential . To date, the role of furin in tumorigenicity has been controversial, as it is responsible for cleaving not just proNCAD but numerous other substrates as well [48,50–52].
NCAD is fundamentally involved in other dynamic cellular systems. For example, it is important for physiological invasion in various nonneoplastic cells such as during neural growth cone extension [53–56], as well as during gastrulation when cells invaginate to form the mesoderm [53,57], and the migration of neural crest cells [58,59] over large distances during embryonic development. In the last two cases, an ECAD-to-NCAD switch similar to that which takes place in tumor cells is important for the invasive phenotype [53,57–59]. It is conceivable that differentially processed cell surface NCAD plays an important function in these physiological processes.
Other groups have shown that NCAD plays a role in tumor progression, typically in tumor cells that undergo a switch from ECAD to NCAD. The current hypothesis in the field is that NCAD expression is sufficient to promote malignant cell transformation [19,21,60–62]. Yet, this body of work does not shed light on the stages of malignancy during tumor progression. We propose that whereas NCAD activity promotes malignant transformation, nonadhesive surface proNCAD determines the degree of cell invasiveness in later stages of tumor progression. Thus, the switch from ECAD to NCAD is in reality a switch to a mixture of NCAD molecules where only a certain proportion is functionally adhesive.
Classic cadherins can exist in a weakly adhesive monomeric form, or as dimers, which are strongly adhesive [35,63]. We speculate that proNCAD may disrupt NCAD arrays emanating from juxtaposed cells by intercalating between lateral dimers (Figure 6). It has been shown that the prodomain of ECAD prevents dimer formation, and it has also been shown that dominant-negative forms of cadherins with extracellular domain deletions prevent cadherin self-association. It is conceivable that by altering cadherin molecular forms and/or composition at the cell surface, the adhesive strength of nascent cell-cell contacts may be regulated, allowing for fine-tuning of malignant intercellular connections.
In summary, down-regulation of CAMs is one manner by which cells decrease functionally adhesive intercellular interactions. In many carcinomas, epithelial junctions are no longer detectable due to the loss of close cell-cell contacts. Another way that intercellular adhesion can be modulated, as we demonstrate here, is by surface expression of a nonadhesive CAM. We provide insight into the mechanism by which NCAD can promote both stable and dynamic cell-cell interactions, and our work revises the current dogma of cadherin posttranslational processing and surface expression. Perhaps differences in relative cell adhesion, rather than absolute cadherin specificity, dictate cell sorting and migration in vivo.
We speculate that differential expression of PC enzymes may also be a common mechanism in many types of tumors to regulate cellular motility as well as other malignant traits. In addition, precursor forms of other types of cadherins at the cell surface may play a similar role in tumor cell motility. Determining the major form of NCAD at the cell surface may serve as a prognostic tool for the staging and progression of malignant tumors. Furthermore, our work sheds light on putative novel treatment strategies with the potential to attenuate the extensive infiltration and metastatic spread by invasive tumor cells.
The authors thank Carmen Sabau, Rosica Bolovan (Montreal Neurological Institute), and Ann Chamberland (Institut de Recherches Cliniques de Montréal) for technical assistance.
1D.M. was supported by a Doctoral Canada Graduate Scholarship from the Canadian Institutes of Health Research (CIHR) and a Richard H. Tomlinson Doctoral Fellowship. R.F.D. holds the William Feindel Chair of Neuro-Oncology. This work was supported by The Raymonde and Tony Boeckh, the Goals for Lily, the Alex Pavanel Family, and the Franco Di Giovanni Brain Tumor Research Funds, the Maggie DeFontes Foundation, and the Brain Tumor Foundation of Canada. N.G.S. is supported by CIHR grant MOP-44363 and a Canadian Chair No. 216684. D.R.C. is supported by CIHR grant RMF-79028 and National Institutes of Health grant NS-20147-20.