CD44 Cross-linking induces integrin-mediated adhesion and transendothelial migration in breast cancer cell line by up-regulation of LFA-1 (αLβ2) and VLA-4 (α4β1)

Experimental Cell Research 304 (2005) 116 – 126 www.elsevier.com/locate/yexcr

CD44 Cross-linking induces integrin-mediated adhesion and transendothelial migration in breast cancer cell line by up-regulation of LFA-1 (aLh2) and VLA-4 (a4h1) Hwai-Shi Wanga,*, Ying Hunga, Cheng-Hsi Sub, Shu-Ting Penga, Yi-Jhih Guoa, Mei-Chun Laia, Ching-Yi Liua, Jia-Wei Hsuc a

Institute of Anatomy and Cell Biology, School of Medicine, Yang Ming University, 155, Sec. 2, Lih-Nong Street, Shih-Pai, 112, Peitou, Taipei 112, Taiwan, ROC b Department of Surgery, Veterans General Hospital, Taipei, Taiwan, ROC c Department of Medical Radiation Technology, Yang-Ming University, Taipei, Taiwan, ROC Received 30 March 2004, revised version received 11 October 2004 Available online 26 November 2004

Abstract CD44, a widely expressed cell surface glycoprotein, plays a major role in cell–cell adhesion, cell–substrate interaction, lymphocyte homing, and tumor metastasis. For tumor metastasis to occur through the blood vessel and lymphatic vessel pathway, the tumor cells must first adhere to endothelial cells. Recent studies have shown that high expression of CD44 in certain types of tumors is associated with the hematogenic spread of cancer cells. However, the functional relevance of CD44 to tumor cell metastasis remains unknown. In this study, we investigated the mechanisms of CD44 cross-linking-induced adhesion and transendothelial migration of tumor cells using MDA-MB-435S breast cancer cell line. Breast cancer cells were found to express high levels of CD44. Using flow cytometric analysis and immunofluorescence staining, we demonstrated that cross-linking of CD44 resulted in a marked induction of the expression of lymphocyte function-associated antigen-1 (LFA-1) and very late antigen-4 (VLA-4) by exocytosis. These results were also observed with the Hs578T breast cancer cell line. Furthermore, LFA-1- and VLA-4-mediated adhesion and transendothelial cancer cell migration were also studied. Anti-LFA-1 mAb or anti-VLA-4 mAb alone had no effect on adhesion or transendothelial cancer cell migration, but were able to inhibit both of these functions when added together. This shows that CD44 cross-linking induces LFA-1 and VLA-4 expression in MDA-MB-435S cells and increases integrin-mediated adhesion to endothelial cells, resulting in the transendothelial migration of breast cancer cells. These observations provide direct evidence of a new function for CD44 that is involved in the induction of LFA-1 and VLA-4 expression by exocytosis in MDA-MB-435S cells. Because these induced integrins promote tumor cell migration into the target tissue, it may be possible to suppress this by pharmacological means, and thus potentially cause a reduction in invasive capability and metastasis. D 2004 Elsevier Inc. All rights reserved. Keywords: CD44; Lymphocyte function-associated antigen (LFA)-1; Very late antigen (VLA)-4; Tumor metastasis; Breast cancer cells

Introduction CD44, a surface receptor for hyaluronan (HA), serves as an adhesion molecule in cell–substrate and cell–cell interactions, lymphocyte recruitment to inflammatory sites, and tumor metastasis [1–4]. The size of the CD44 molecule * Corresponding author. Fax: +886 2 28283212. E-mail address: [email protected] (H.-S. Wang). 0014-4827/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.yexcr.2004.10.015

ranges from the standard 85–95 kDa form (CD44s) to larger variant isoforms (CD44v) of 200 kDa or more. These size differences are partially due to post-translational modifications, as all isoforms of CD44 are highly glycosylated [5]. Most variants are generated by the alternative splicing of at least 12 exons [6]. The functional characterization of the different isoforms of this family is still limited. Many cancer cell types express high levels of CD44. Furthermore, it has been shown in animal models that the

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

injection of reagents that interfere with the binding of CD44 to its ligand inhibits local tumor growth and metastatic spread [7–9]. Tumor cell invasion is made up of cell adhesion to the extracellular matrix, degradation of extracellular matrix components, tumor cell motility, and cell detachment [10]. During tumor metastasis, cells detach from the primary tumor, penetrate the basement membrane into the connective tissue, and invade adjacent structures, including lymph and blood vessels. The tumor cells are subsequently transported to metastatic sites via the lymph and/or blood. The loss of adhesive functions and the gain of new ones are thought to play a crucial role in the metastatic cascade [11]. However, the mechanisms by which CD44 promotes tumor metastasis are poorly understood. Leukocyte function-associated molecule 1 (LFA-1) is an aL/h2 heterodimeric integrin belonging to the h2 leukocyte integrin family of adhesion receptors [12,13]. Three ligands of LFA-1 have been identified, namely intercellular adhesion molecule (ICAM)-1, ICAM-2, and ICAM-3 [14]. LFA-1 mediates the adherence of activated leukocytes to endothelial ligands, in particular ICAM-1 [15]. Under most conditions, transendothelial migration of leukocytes is mediated by LFA-1, but very late antigen (VLA)-4 can also be involved [16,17]. VLA-4 (a4/h1), which is expressed on lymphocytes, monocytes, and eosinophils, mediates cell attachment to vascular cell adhesion molecule (VCAM)-1 expressed on activated endothelium at inflammatory sites [18,19]. The VLA-4/VCAM-1 interaction may play a role in melanoma cell metastasis [20].

117

In this report, we have identified a new function for CD44, namely, its involvement in the induction of LFA-1 and VLA-4 expression by exocytosis in MDA-MB-435S cells and the subsequent mediation of cellular adhesion and migration in invasion processes. Based on these new findings, we suggest that the activation signals resulting from CD44 stimulation may be involved in the tumor cell/ endothelial cell interaction and in metastasis. We propose that one function of CD44 in tumor cells may be to induce integrin expression, which, at least for some tumor cell types, may be a critical step in the formation of metastatic colonies.

Materials and methods Cell culture The human breast carcinoma cell lines, MDA-MB-435S and Hs578T, and human umbilical vein endothelial cells (HUVECs) were obtained from the Food Industry Research and Development Institute (Hsinchu, Taiwan). MDA-MB435S cells were grown in Leibovitz’s L-15 medium supplemented with 15% fetal bovine serum (FBS) (Hyclone, Logan, UT), 10 Ag/ml of insulin, 100 U/ml of penicillin, and 100 Ag/ml of streptomycin. Hs578T cells were cultured in Dulbecco’s modified Eagle’s medium (Gibco, Gaithersburg, MD) supplemented with 5% FBS, 2 mM glutamine, 100 U/ml of penicillin, and 100 Ag/ml of streptomycin.

Fig. 1. (A) Phenotypic analysis of MDA-MB-435S cells. Cells were stained with CD44 mAb (A), LFA-1 mAb (B), or VLA-4 mAb (C). Flow cytometric analyses were performed using FACScan. The shaded area shows the profile of the negative control. The histograms shown are representative of five experiments. (D) LFA-1 expression on MDA-MB-435S cells induced by CD44 cross-linking. Cells were cross-linked with anti-CD44 mAb for 1 h and secondary Ab for 30 min, then LFA-1 expression was analyzed by FACScan. The figure shows one representative histogram out of five.

118

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

HUVECs were cultured in Medium 200 (Cascade Biologics, Portland, OR) supplemented with 2% FBS, 1 Ag/ml of hydrocortisone, 10 ng/ml of human epidermal growth factor, 3 ng/ml of basic fibroblast growth factor, 10 Ag/ml of heparin, 100 U/ml of penicillin, and 100 Ag/ml of streptomycin. Cross-linking of CD44, aVb3, and CD29 Cells were cultured until confluence, then incubated for 1 h at 378C with mouse monoclonal anti-human CD44 antibody (10 Ag/ml) (clone 5F12, which blocks the binding of hyaluronic acid to the CD44 receptor, Cat. #MS-178-p1, Newmarker, Fremont, CA), mouse anti-human integrinaVh3 (vitronectin receptor) monoclonal antibody (10 Ag/ml) (clone LM609, Cat. #MAB1976, Chemicon Int. Inc., Temecula, CA), or mouse anti-human CD29 (fibronectin receptor) monoclonal antibody (10 Ag/ml) (clone S6, Cat. #AHE2014, Biosource, Camarillo, CA). After three washes, 1 Ag/ml of goat anti-mouse IgG-Fc (Bethyl, Montgomery, TX) was added for different lengths of time as the second antibody (Ab) allowing CD44 cross-linking to occur.

the primary antibodies were omitted, negligible immunofluorescence was seen. Monolayer cell adhesion assay using HUVECS The adhesion assay was performed in 96-well plates as previously described [21–23]. Briefly, 1.5  104 HUVEC cells were grown to confluence in 96-well plates, and then treated for 12 h with IL-1h (100 U/ml). MDA-MB-435S cells were grown to subconfluence on 35-mm dishes, then sequentially incubated with mouse anti-human CD44 mAb (10 Ag/ml) for 1 h at 378C and goat anti-human IgG Fc antibody (10 Ag/ml) for another 30 min before loading with BCECF/AM (2 AM) (Molecular Probe, Eugene, OR) for 30 min. They were then harvested using trypsin/EDTA, resuspended for 1 h in culture medium with/without mouse anti-human LFA-1 mAb (40 Ag/ml) (Newmarker) and mouse anti-human VLA-4 mAb (40 Ag/ml) (R&D, Minneapolis, MN), washed with PBS, centrifuged at 120 g for

Flow cytometric analysis Suspended tumor cells (5  106 cells; 1 ml) with or without CD44 cross-linking, were incubated for 45 min at 48C with 150 Al of appropriate R-Phycoerythrin- or FITCconjugated antibody solution (10 Ag/ml), washed with PBS, spin down the cells. Control samples were incubated with phosphate-buffered saline (PBS) in place of primary antibody. A Becton Dickinson FACScan was used to analyze antibody binding. Immunocytochemistry of MDA-MB-435S cells Staining was performed on fixed, non-permeabilized monolayers of MDA-MB-435S cells grown to confluence on glass tissue culture slides. For exocytosis experiment, CD44 cross-linking cells were rapidly cooled on ice before fixation, or after 1 h on ice transient to 378C for 30 min. The cells were then washed with PBS, fixed for 30 min at 378C in 0.05% glutaraldehyde, then incubated for 25 min at room temperature with a 1:9 dilution of normal goat serum in PBS to block non-specific binding of the primary antibody. The slides were incubated for 75 min at 48C with mouse monoclonal anti-human LFA-1 conjugated FITC antibody (clone number: CRIS-3, which recognize an antigen of 170/90 kDa, Cat. #AHS1118, Biosource) or mouse monoclonal anti-human VLA-4 conjugated FITC antibody (10 Ag/ml) (clone number: 44H6, which recognizes VLA-4a, Cat. #AHS4948, Biosource). They were then mounted with mounting medium (Vector, North Hollywood, CA), and viewed with a confocal laser scanning microscope (Leica) using a 100/1.30 oil immersion objective and an appropriate filter. In controls in which

Fig. 2. (A) Time-course of CD44-triggered induction of LFA-1 expression on MDA-MB-435S cells. (A) Cells were treated with 10 Ag/ml of antiCD44 mAb for 1 h, then cross-linked with goat anti-mouse IgG Fc for 30 s, 1, 10 min, or 1 or 6 h, then LFA-1 expression was analyzed by FACScan. (B) VLA-4 up-regulation by CD44 cross-linking on MDA-MB-435S cells. Cells were cross-linked with 10 Ag/ml of anti-CD44 mAb for 1 h, followed by secondary Ab for 30 min. VLA-4 expression was analyzed by FACScan.

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

119

3 min, and resuspended at a concentration of 8  104 cells/ ml. Five hundred microlitres of the suspension was added to each well of the washed HUVEC monolayers. The plates were incubated for 30 min at 378C to allow adhesion, then non-adherent cells were removed by three washes with PBS, and the plate read using a Fluorescence Measurement System (Millipore cytofluor 2300) with a 485 nm excitation filter and a 538 nm emission filter. The data are expressed as the mean F SEM of at least three experiments performed in quadruplicate. Statistical analysis of the significance of differences between groups was carried out using a Student’s t test. A P b 0.05 was considered significant.

grow to confluence. Afterward, 2  104 breast carcinoma cells (MDA-MB-435S), cultured in a 35-mm dish with or without CD44 cross-linking as described above, were prepared in suspension using fresh Medium 200. Next, the cell suspension was added to the well containing the monolayer of HUVEC. The impedance of the challenged HUVEC was monitored via ECIS for the next 20 h to determine the invasive ability of the MDA-MB-435S cells, as a highly metastatic cell reduces the total resistance across the HUVEC monolayer.

ECISk (electric cell–substrate impedance sensing)

The migration assay used was a transwell, the upper chamber of which consisted of cell culture inserts coated with 5 Ag/ml of collagen I at 48C overnight. The wells were washed three times with PBS, and then HUVEC cells (1  105 cells/well) were added. After the cells reached confluence, 100 U/ml of IL-1h was added for 12 h to induce expression of ICAM-1 and VCAM-1. Subconfluent breast cancer cell lines were trypsinized and resuspended in cell culture medium containing 5% FBS, then 2  105 cells were added to the upper chamber and incubated for 24 h at 378C in

To detect the invasive activities of the metastatic cells in vitro, an electric cell–substrate impedance sensing (ECIS) assay is used as reported previously by Keese et al. [24]. Electrode arrays with gold film electrodes supplied by Applied Biophysics (Troy, NY) are prepared by rinsing with PBS. A HUVEC suspension was prepared at 5  105 cells/ml, and 200 Al were added to each well, resulting in a final surface concentration of 1.25  105 cells/cm2 and cells allowed to

Migration assay

Fig. 3. Immunofluorescent staining of LFA-1 and VLA-4 on MDA-MB-435S cells. Cells with and without CD44 cross-linking. (A and B): stained with antiLFA-1 mAb. (C and D): stained with anti-VLA-4 mAb. (A and C): no cross-linking. (B and D): after CD44 cross-linking with anti-CD44 mAb for 1 h, followed by secondary Ab for 30 min. Scale bar = 10 Am.

120

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

a humidified 5% carbon dioxide environment. When used, LFA-1 and VLA-4 antibodies were added to the cell suspension before plating into the upper chamber. Cells that had invaded through the matrix and become adherent to the undersurface of the filter were quantified using Giemsa stain and a microscope.

on the cell membranes of CD44 cross-linked cells (Fig. 3D). The results show that CD44 cross-linking caused a marked increase in LFA-1 and VLA-4 expression on MDA-MB435S cells. These results suggest that CD44 plays a role as a signaling molecule that induces the expression of integrins. Lack of effect of a protein synthesis inhibitor on CD44 cross-linking-induced LFA-1 and VLA-4 expression

Results Phenotypic analysis and cross-linking of CD44 on MDA-MB-435S cells up-regulates LFA-1 and VLA-4 expression We first characterized the phenotype of the human breast carcinoma cell lines, MDA-MB-435S and Hs578T. Flow cytometric analysis showed that MDA-MB-435S highly expressed CD44 (Fig. 1A), but not LFA-1 (Fig. 1B). The fluorescence histograms also showed that MDA-MB-435S cells expressed VLA-4 (Fig. 1C). Similar results were also observed with Hs578T cells (data not shown). We hypothesized that, for some tumor cell types, one function of CD44 may be to facilitate integrin expression, which might be a critical step in establishing metastatic colonies. To characterize the function of CD44, we examined the effects of crosslinking CD44 on cell surface molecule expression, using a specific mAb and a secondary cross-linking Ab. As shown in Figs. 1C and D, LFA-1 was not expressed on nonstimulated MDA-MB-435S cells, but CD44 cross-linked cells showed significant expression. The controls without primary or without secondary antibodies showed no effect on the LFA-1 expression (data not shown). We then examined the time-course of CD44-triggered LFA-1 expression on MDA-MB-435S cells. Cells were treated for 1 h with 10 Ag/ml of anti-CD44 mAb, and then cross-linked with goat anti-mouse IgG Fc for 30 s, 1 or 10 min, or 1 or 6 h, finally, LFA-1 expression was analyzed by FACScan. As shown in Fig. 2A, the addition of the secondary antibody for as little as 30 s resulted in an increase in LFA-1 expression, which continued to increase with time of exposure. When CD44 on MDA-MB-435S cells was cross-linked by applying 10 Ag/ml of anti-CD44 mAb for 1 h and the secondary cross-linking Ab for 30 min, VLA-4 expression was up-regulated (Fig. 2B). Similar results were obtained for Hs578T cells (data not shown). We further checked if the elevated cell surface expression of LFA-1 and VLA-4 is caused by cross-linking CD44 specifically, or by clustering of cell surface receptors in general. The results showed that cross-linking aVh3 or CD29 (fibronectin receptors) did not induce LFA-1 or VLA-4 expression (data not shown). Immunofluorescent studies also showed that untreated MDA-MB-435S cells were LFA-1-negative (Fig. 3A), but that CD44 cross-linking resulted in LFA-1 expression (Fig. 3B). Similar results were obtained for VLA-4, which was not expressed on untreated cells (Fig. 3C), but highly expressed

The above experiments showed that up-regulation of LFA-1 and VLA-4 by CD44 cross-linking occurred in a very short period of time. To determine whether these effects required new protein synthesis, MDA-MB-435S cells were pretreated for 1 h with the protein synthesis inhibitor, cycloheximide (20 Ag/ml), before cross-linking with anti-CD44 mAb and secondary Ab, and then LFA-1 and VLA-4 expression was analyzed. As shown in Fig. 4, CD44 cross-linking-induced LFA-1 and VLA-4 expression was not affected by the presence of cycloheximide. Similar results were observed with Hs578T cells (data not shown). These results suggested that new protein synthesis was not required.

Fig. 4. Effect of cycloheximide on CD44 cross-linking-induced LFA-1 and VLA-4 expression on MDA-MB-435S cells. Cells were pretreated with 20 Ag/ml of cycloheximide for 1 h before cross-linking with antiCD44 mAb (10 Ag/ml) for 1 h, followed by secondary Ab for 30 min, then expression of LFA-1 (A) or VLA-4 (B) was analyzed by FACScan.

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

121

Fig. 5. (A) Effect of temperature transition, MDC, BAPTA-AM, and 3-methyladenine on CD44 cross-linking-induced LFA-1 and VLA-4 expression. MDAMB-435S cells were culture on cover slips. Immunostaining of LFA-1 expression of: (a) Cells were incubated at 378C. (b) Cells were cross-linked with anti-CD44 mAb for 1 h, followed by secondary Ab for 30 min. (c) Cells were shifted to ice for 20 min before CD44 cross-linking and left on ice for 1.5 h of cross-linking then fixed for immunostaining. (d) Cells were shifted to ice 20 min before CD44 cross-linking and stayed on ice for 1.5 h cross-linking, then shifted to 378C for 30 min before immunostaining for LFA-1. Scale bar = 10 Am. (B and C) Effect of MDC and 3-methyladenine on CD44 crosslinking induced LFA-1 and VLA-4 expression. The figure shows one representative histogram out of five showing VLA-4 expression with/without MDC treatment. Cells were pretreated with 50 AM of MDC (B), or 3-methyladenine (C) for 1 h before cross-linking with anti-CD44 mAb (10 Ag/ml) for 1 h, followed by secondary Ab for 30 min, then expression of LFA-1 (a) or VLA-4 (b) on MDA-MB-435S cells was analyzed by FACScan.

122

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

Cross-linking of CD44 on MDA-MB-435S cells up-regulates LFA-1 and VLA-4 expression through exocytosis We further investigate whether vesicle trafficking is involved in CD44-induced LFA-1 and VLA-4 expression. As exocytosis is an energy-dependent process, we supposed that exocytosis would be terminated by cooling on ice. The effect of a temperature transition on CD44 induced LFA-1 expression is seen in Fig. 5A. When the temperature of the incubation medium was shifted from 378C to 08C before CD44 cross-linking, upregulation of LFA-1 was markedly inhibited (Fig. 5A(c)). When the cross-linking cells were shift from 48C to 378C, LFA-1 reappeared on the membrane (Fig. 5A(d)). The reduced temperature also inhibited CD44 cross-linking-induced VLA-4 expression (data not shown). Monodansylcadaverine (MDC), an inhibitor of calcium-dependent transglutaminase, has been used previously to study the endocytosis and exocytosis by blocking release by exocytic transfer in reticulocytes [25,26]. In this study, MDC was also able to suppress LFA-1 and VLA-4 expression induced by CD44 cross-linking on the cell membrane (Fig. 5B). Actin filaments have been implicated in vesicle trafficking. In our study, LFA-1 induced by CD44 cross-linking was decreased by pretreatment with cytocalasin B, or by calcium chelator, BAPTA-AM (data not shown). Moreover, pretreatment of the cells with the lipid kinase PI3K inhibitor, 3-methyladenine, which has been shown to interfere with the normal trafficking of exocytosis organelles [27], resulted in CD44 cross-linking-induced VLA-4 but not LFA-1 expression being decreased (Fig. 5C). These results suggested exocytosis is involved in CD44 cross-linking-induced LFA-1 and VLA-4 expression on the cell membrane. CD44 cross-linking augments integrin-mediated adhesion of MDA-MB-435S cells to IL-1b-activated HUVECs We investigated whether CD44-induced LFA-1 and VLA-4 up-regulation on MDA-MB-435S cells was involved in their adhesion to HUVECs. ICAM-1 and VCAM-1 expression was induced on HUVEC cells by incubation with IL-1h (100 U/ml) for 12 h before using the cells in an adhesion assay. Cells were either incubated with control medium or cross-linked for 6 h with CD44 mAb, then loaded with BCECF/AM in the presence or absence of the indicated blocking mAb. As shown in Fig. 6, cells on which CD44 had not been cross-linked were able to adhere to ICAM-1-/VCAM-1-expressing HUVECs, but cross-linking resulted in a 1.7-fold increase in adhesion. Adhesion of CD44 cross-linked MDA-MB435S cells to HUVECs was partially inhibited by antiVLA-4 or anti-LFA-1 mAbs alone, but the effect was reduced to the same level seen using non-CD44-crosslinked cells by co-addition of the two mAbs. Thus, the adhesion of CD44 cross-linked MDA-MB-435S cells to

Fig. 6. Effect of CD44 cross-linking on MDA-MB-435S cells on their adhesion to IL-1h-activated HUVEC cells. Cells were incubated with control medium or cross-linked with CD44 mAb for 30 min and secondary antibody for 5.5 h, loaded with BCECF/AM, then incubated in the presence or absence of the indicated blocking mAb. The data are representative of four different experiments. *P b 0.05 compared to control.

HUVEC cells is mediated by LFA-1 and VLA-4 together. Invasion of the breast cancer cell determined by ECIS after CD44 cross-linking The penetration of the cells through the endothelial monolayer has been suggested to represent similar invasive activities that take place during the metastatic process in vivo. An electric cell–substrate impedance sensing (ECISk) cell-electrode system was adopted to study the invasive behavior of the breast cancer cell in terms of impedance change, when HUVEC cells were seeded to attach at the bottom of electrode well. MDA-MB-435S cells with or without CD44 cross-linking were then inoculated into the electrode well to challenge the HUVEC cells. Individual electrodes were followed to measure the timecourse of any impedance changes from the time of inoculation to 20 h after inoculation. In Fig. 7, real time impedance changes are shown for several attachment curves, where each curve represents the time-course impedance change measured for one electrode. In this experiment, HUVECs were inoculated and grown on electrode arrays until confluent. MDA-MB-435S cells with or without CD44 cross-linking were able to drop the intact HUVEC layer resistance. Moreover, there is a substantial further drop elicited by CD44 cross-linked cells, as well as a quieting of the typical endothelial cell impedance fluctuations. The changing of the impedance started at 15 to 20 h is

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

123

Fig. 7. CD44 cross-linking-induced transendothelial migration. Resistance changes in the impedance as confluent layers of HUVEC cells are challenged with MDA-MB-435S cells. Real-time impedance changes were recorded, wherein each curve represents the time-course impedance change measured from each electrode. The cancer cells with or without CD44 cross-linking were added to the HUVECs monolayer after 1 h recording. Electrode 1 (black) and 2 (pink) measured impedance change from the MDA-MB-435S cells without CD44 cross-linking, electrode 3 (blue) and 4 (red) measured impedance change from the cells with CD44 cross-linking. After the cancer cells were added, the impedance started to drop for both electrodes. After the CD44 cross-linked cancer cells had been present for 15 h, a drastic drop of impedance was observed in electrodes 3 and 4. This indicates that CD44 cross-linked cancer cells have a higher metastatic ability.

because of the tumor cells had to attach to the HUVECs first, and then invade through the HUVEC monolayer to breakdown the cell–cell contact site. Once the cell–cell contact between HUVECs was broken, the resistance drops fast; the real time invasion happens in this study between about 15 to 20 h. Integrin regulation of breast cancer cell invasion is triggered by cross-linking of CD44 To characterize the function of CD44 cross-linkinginduced LFA-1 and VLA-4 in cell invasion, we examined the effects of CD44 cross-linking on breast cancer cell migration. As shown in Fig. 8, the transendothelial migration of MDA-MB-435S cells was markedly increased by CD44 cross-linking, and this effect was blocked by the co-addition of anti-LFA-1 and anti-VLA-4 mAbs. These results suggest that CD44-induced cross-linking reduction in expression of LFA-1 and VLA-4 is involved in cell migration through the endothelial cells.

Discussion In 1992, Tanaka et al. [28] found that exposure of CD31+ T cells to anti-CD31 mAbs triggers the adhesive function of h1 the integrins, particularly VLA-4, and proposed that CD31 functions in an badhesion cascadeQ by amplifying the integrin-mediated adhesion of CD31+ T cells to other cells, particularly endothelial cells. Cytokines, such as IL-1 or

Fig. 8. Transmigration of MDA-MB-435S cells through IL-1h-activated HUVECS. Transmigration of MDA-MB-435S cells was assessed in 3 Am pore, 24-well transwells precoated overnight with HUVECS, which were then activated with IL-1h (100 U/ml) for 12 h at 378C. The insert wells contained: (A) control MDA-MB-435S cells, (B) cells cross-linked with CD44 mAb for 30 min, followed by secondary antibody for 5.5 h, (C) cross-linked cells with mouse anti-human LFA-1 mAb and mouse antihuman VLA-4 mAb added. After incubation for 30 min at 378C, cells that had migrated into the lower wells were stained (Geimsa stain). The data are representative of four different experiments. Scale bar = 40 Am. (D) mAb inhibition of CD44 cross-linking-induced adhesion. Data are expressed as mean percentage and SEM for the binding of MDA-MB-435S cells from four replicate wells. *P b 0.05 compared to control.

124

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

TNF-a, can stimulate endothelial cells to express VCAM-1 and bind lymphocytes [18]. In a rheumatoid synovial cell line, cross-linking of ICAM induces activation of the transcription factor, AP-1, and transcription of the IL-1h gene [29], showing that ICAM functions not only as an adhesion molecule for integrin binding, but also as a signaling molecule. These studies suggest that adhesion molecules function not only as cellular adhesion molecules, but also as signal transducers. An immunohistochemical study showed that high levels of CD44 are found in metastatic breast carcinoma [30]. Our present results showed that the two breast carcinoma cell lines, MDA-MB-435S and Hs578T, expressed high levels of CD44 (Fig. 1). A number of lines of evidence implicate CD44 as a cell adhesion molecule with a possible role in tumor progression [31–33]. However, how CD44 contributes to the malignant phenotype is not yet known. In 1999, Fujisaki et al. [23] found that stimulation of CD44 by crosslinking as well as 6.9 kDa hyaluronan can induce LFA-1 expression on colon cancer cells. In order to study the role of CD44 in tumor metastasis, we cross-linked CD44 using mouse anti-human CD44 mAb and goat anti-mouse IgG Fc antibody, and found that LFA-1 and VLA-4 expression was increased (Figs. 1 and 2). Anti-human CD44 mAb alone had no effect. This suggests that cross-linking made the CD44 molecules cluster and this phenomenon transduces a signal into the cell; the results are consistent with the physiological conditions, in which a large molecule, HA, can bind several CD44 molecules, leading to clustering and thus signal transduction. These results suggest that CD44 is a signal molecule for the induction of the expression of other integrins. Recently, the function of CD44 as a signaling molecule has been demonstrated. In rheumatoid synovial cells, stimulation of CD44 with mAbs or hyaluronan transmits the signal into the cells, leading to up-regulation of VCAM-1 [34]. The time-course of CD44 cross-linkinginduced expression of LFA-1 showed that 1 h of crosslinking induced significant cell surface expression of LFA-1, suggesting that LFA-1 was translocated from the cytosol to the membrane. This up-regulation was shown not to involve new protein synthesis (Fig 4). These findings indicate that CD44 cross-linking is probably responsible for transducing the signal into the cell and for the translocation of LFA-1 to the membrane. The fact that CD44 stimulation up-regulated LFA-1 and VLA-4 expression prompted us to investigate how this phenomenon might contribute to tumor metastasis. The best-known ligands for LFA-1 and VLA-4 are, respectively, ICAM-1 and VCAM-1. ICAM-1 expression on endothelial cells is strongly regulated by various inflammatory cytokines [35]. IL-1h or TNF-a can up-regulate Ig-superfamily adhesion molecules, such as ICAM-1 and VCAM-1 [36]. Since, during tumorigenesis, activated NK cells and macrophage release IL-1h and TNF-a and induce endothelial cells to express more adhesion molecules, we used IL1h to induce expression of ICAM-1 and VCAM-1 on

HUVECs and studied the effect of CD44 cross-linking on tumor adhesion to, and migration through, endothelial cells. This model therefore closely resembles the situation in tumors. In addition to CD44, LFA-1, and VLA-4, breast tumor cell lines express other adhesion molecules, including a2h1 and a3h1, their ligands being collagen and laminin on the basement membrane [37,38]. Breast cancer cells also express av integrins. By binding to fibronectin and vitronectin, tumor cells can invade the extracellular matrix [39,40]. Herrera-Gayol and Jothy [41] suggested that CD44 (especially CDv6) on breast cancer cells regulates binding to extracellular HA and is involved in metastasis. Since most integrins bind to the extracellular matrix, most studies on metastasis have focused on this interaction. However, in our study, we were particularly interested in examining the signaling caused by CD44 cross-linking and found that it resulted in induction of LFA-1 and VLA-4 expression. LFA-1 and VLA-4 expression on cell membrane appeared as early as 30 s after CD44 cross-linking. By cold treatment and MDC treatment, both LFA-1 and VLA-4 expression can be inhibited (Fig. 5). Furthermore, pretreatment with cytochalasin B, and BAPTA-AM decreased the LFA-1 expression induced by CD44 cross-linking on the membrane. The actin cytoskeleton has been implicated in exocytosis and we propose that exocytosis of adhesion molecules is involved in the CD44 cross-linking-induced up-regulation of LFA-1 and VLA-4. However, pretreatment with PI3K inhibitor 3-methyladenine has less effect on VLA-4 expression induced by CD44 cross-linking. Since the phenotypic differences between LFA-1 and VLA-4 expression on non-stimulated cells, CD44 cross-linkinginduced exocytosis of LFA-1 and VLA-4 might be through different signaling pathways. Under most conditions, transendothelial migration is mediated by LFA-1, but VLA-4 can also be involved. In contrast to LFA-1, this requires exogenous chemokines and the activation of endothelial cells [16]. The fact that CD44 stimulation up-regulated other integrins prompted us to investigate how this phenomenon might contribute to the metastasis process. We found that CD44 cross-linking caused MDA-MB-435S cells to adhere to HUVEC and that this adhesion was blocked by co-addition of mouse antihuman LFA-1 and anti-human VLA-4 mAbs together, but not by addition of either mAb alone (Fig. 6). These results suggest that both integrins play a role in tumor cell adhesion to endothelial cells, and that, if one is blocked, the other still has adhesive function. The invasiveness of the MDA-MB-435S cells was further characterized using the ECIS assay proposed by Keese et al. [24] to follow the real-time invasive activity of metastatic cells in culture. The MDA-MB-435S cells with or without CD44 cross-linking, within 1 h after being challenged would result in the impedance of the confluent HUVEC layer being substantially reduce. Interestingly, a drastic drop of impedance 15 h after addition to the

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126

monolayer (Fig. 7), suggests that the breast cancer cell after CD44 cross-linking has gained higher invasive ability. Furthermore, in the migration assay by using transwells, when HUVEC were grown to confluence in the upper part of the wells and then MDA-MB-435S cells with/without CD44 cross-linking were co-cultured with HUVEC cells, CD44 cross-linking enhanced the migration capability of the tumor cells. This effect was significantly attenuated by the co-addition of mouse anti-human LFA-1 and mouse antihuman VLA-4 mAb together (Fig. 8). These findings indicate that the adhesion of tumor cells to the endothelium via LFA-1 and VLA-4 may contribute to tumor cell migration and are consistent with the adhesion assay data, suggesting that the migration process also requires tumor cells to attach to the endothelial cells. Based on the findings presented in this work, we propose that stimulation of adhesion molecules may induce expression of other adhesion molecules on the same cell, because we have observed that CD44 stimulation by cross-linking led to an increase in LFA-1 and VLA-4 expression. Furthermore, the concomitant expression of CD44, LFA-1, and VLA-4 on tumor cells and the increased induction of LFA-1 and VLA-4 by CD44 stimulation suggest that CD44 may play a pivotal role in the amplification of LFA-1 and VLA-4 by membrane trafficking and the subsequent adhesion of tumor cells to endothelial cells. Thus, CD44mediated up-regulation of LFA-1 and VLA-4 appears to be very relevant to metastasis. Our study demonstrated a specific new role for CD44 in tumor cell adhesion and invasion. Furthermore, we suggest that firstly, CD44 plays a role in adhesion of tumor cells to endothelial cells and migration through endothelial cells and that secondly, CD44 may participate in adhesion not only by functioning as an adhesion receptor, but also by increasing the expression of other integrins on cell membranes. Our findings warrant further studies into CD44 stimulation and its involvement in the critical steps of the metastatic process, which might allow the identification of new pharmacological approaches to metastasis control. Expression of this receptor may be an important factor in the ability of a cell to metastasize.

Acknowledgment This work was supported by a research grant from the National Science Council, Taiwan (NSC 91-2320-B-010066) to H.-S. Wang.

References [1] J. Lesley, R. Hyman, P.W. Kincade, CD44 and its interaction with extracellular matrix, Adv. Immunol. 54 (1993) 271 – 335. [2] D. Naot, R.V. Sionov, S.D. Ish, CD44: structure, function, and association with the malignant process, Adv. Cancer Res. 71 (1997) 241 – 319.

125

[3] Y. Shimizu, G.A. Van Seventer, R. Siraganian, L. Wahl, S. Shaw, Dual role of the CD44 molecule in T cell adhesion and activation, J. Immunol. 143 (1989) 2457 – 2463. [4] U. Gunthert, M. Hofmann, W. Rudy, S. Reber, M. Zoller, I. Haussmann, S. Matzku, A. Wenzel, H. Ponta, P. Herrlich, A new variant of glycoprotein CD44 confers metastatic potential to rat carcinoma cells, Cell 65 (1991) 13 – 24. [5] C.B. Underhill, CD44: the hyaluronan receptor, J. Cell Sci. 103 (1992) 293 – 298. [6] G.R. Screaton, M.V. Bell, D.G. Jackson, F.B. Cornelis, U. Gerth, J.I. Bell, Genomic structure of DNA encoding the lymphocyte homing receptor CD44 reveals at least 12 alternatively spliced exons, Proc. Natl. Acad. Sci. U. S. A. 89 (1992) 12160 – 12164. [7] U.-J. Guo, J. Ma, J. Wang, X. Che, J. Narula, M. Bigby, M. Wu, M.-S. Sy, Inhibition of human melanoma growth and metastasis in vivo by anti-CD44 monoclonal antibody, Cancer Res. 54 (1994) 1561 – 1565. [8] M.A. Zahalka, D. Naor, Beta 2-integrin dependent aggregate formation between LB T cell lymphoma and spleen cells: assessment of correlation with spleen invasiveness, Int. Immunol. 6 (1994) 917 – 924. [9] M.A. Zahalka, E. Okon, U. Gosslar, B. Holzmann, D. Naor, Lymph node (but not spleen) invasion by murine lymphoma is both CD44and hyaluronate-dependent, J. Immunol. 154 (1995) 5345 – 5355. [10] L.A. Liotta, Tumor invasion and metastases—role of the extracellular matrix: Rhoads Memorial Award Lecture, Cancer Res. 46 (1986) 1 – 7. [11] I.R. Hart, N.T. Goode, R.E. Wilson, Molecular aspects of the metastatic cascade, Biochim. Biophys. Acta 989 (1989) 65 – 84. [12] T.A. Springer, Adhesion receptors of the immune system, Nature 346 (1990) 425 – 434. [13] R. Pardi, L. Inverardi, J.R. Bender, Regulatory mechanisms in leukocyte adhesion: flexible receptors for sophisticated travelers, Immunol. Today 13 (1992) 224 – 230. [14] M.E. Binnerts, Y. van Kooyk, D.L. Simmons, C.G. Figdor, Distinct binding of T lymphocytes to ICAM-1, -2 or -3 upon activation of LFA-1, Eur. J. Immunol. 24 (1994) 2155 – 2160. [15] M.P. Stewart, C. Cabanas, N. Hogg, T cell adhesion to intercellular adhesion molecule-1 (ICAM-1) is controlled by cell spreading and the activation of integrin LFA-1, J. Immunol. 156 (1996) 1810 – 1817. [16] Z. Ding, K. Xiong, T.B. Issekutz, Chemokines stimulate human T lymphocyte transendothelial migration to utilize VLA-4 in addition to LFA-1, J. Leukocyte Biol. 69 (2001) 458 – 466. [17] K. Selmaj, A. Walczak, M. Mycko, T. Berkowicz, T. Kohno, C.S. Raine, Suppression of experimental autoimmune encephalomyelitis with a TNF binding protein (TNFbp) correlates with down-regulation of VCAM-1/VLA-4, Eur. J. Immunol. 28 (1998) 2035 – 2044. [18] L. Osborn, C. Hession, R. Tizard, C. Vassallo, S. Luhowskyj, G. ChiRosso, R. Lobb, Direct expression cloning of vascular cell adhesion molecule 1, a cytokine-induced endothelial protein that binds to lymphocytes, Cell 59 (1989) 1203 – 1211. [19] G.E. Rice, J.M. Munro, M.P. Bevilacqua, Inducible cell adhesion molecule 110 (INCAM-110) is an endothelial receptor for lymphocytes. A CD11/CD18-independent adhesion mechanism, J. Exp. Med. 171 (1990) 1369 – 1374. [20] G.E. Rice, M.P. Bevilacqua, An inducible endothelial cell surface glycoprotein mediates melanoma adhesion, Science 246 (1989) 1303 – 1306. [21] A. Takada, K. Ohmori, N. Takahashi, K. Tsuyuoka, A. Yago, K. Zenita, A. Hasegawa, R. Kannagi, Adhesion of human cancer cells to vascular endothelium mediated by a carbohydrate antigen, sialyl Lewis A, Biochem. Biophys. Res. Commun. 179 (1991) 713 – 719. [22] E.A. Price, D.R. Coombe, J.C. Murray, Beta-1 integrins mediate tumour cell adhesion to quiescent endothelial cells in vitro, Br. J. Cancer 74 (1996) 1762 – 1766. [23] T. Fujisaki, Y. Tanaka, K. Fujii, S. Mine, K. Saito, S. Yamada, U. Yamashita, T. Irimura, S. Eto, CD44 stimulation induces integrinmediated adhesion of colon cancer cell lines to endothelial cells by up-

126

[24]

[25] [26]

[27]

[28]

[29]

[30] [31]

[32]

[33]

H.-S. Wang et al. / Experimental Cell Research 304 (2005) 116–126 regulation of integrins and c-Met and activation of integrins, Cancer Res. 59 (1999) 4427 – 4434. C.R. Keese, K. Bhawe, J. Wegener, I. Giaever, Real-time impedance assay to follow the invasive activities of metastatic cells in culture, BioTechniques 33 (2002) 842 – 850. H. Kohno, R. Tokunaga, Transferrin and iron uptake by rat reticulocytes, J. Biochem. 97 (1985) 1181 – 1188. J.A. Grasso, M. Bruno, A.A. Yates, L.T. Wei, P.M. Epstein, Calmodulin dependence of transferring receptor recycling in rat reticulocytes, Biochem. J. 266 (1990) 261 – 272. R. Carini, R. Castino, M.G. De Cesaris, R. Splendore, M. Demoz, E. Albano, C. Isidoro, Preconditioning-induced cytoprotection in hepatocytes requires Ca2+-dependent exocytosis of lysosomes, J. Cell Sci. 117 (2003) 1065 – 1077. Y. Tanaka, S.M. Albelda, K.J. Horgan, G.A. van Seventer, Y. Shimizu, W. Newman, J. Hallam, P.J. Newman, C.A. Buck, S. Shaw, CD31 expressed on distinctive T cell subsets is a preferential amplifier of beta 1 integrin-mediated adhesion, J. Exp. Med. 176 (1992) 245 – 253. Y. Koyama, Y. Tanaka, K. Saito, M. Abe, K. Nakatsuka, I. Morimoto, P.E. Auron, S. Eto, Cross-linking of intercellular adhesion molecule 1 (CD54) induces AP-1 activation and IL-1beta transcription, J. Immunol. 157 (1996) 5097 – 5103. N. Iida, L.Y. Bourguignon, New CD44 splice variants associated with human breast cancers, J. Cell. Physiol. 162 (1995) 127 – 133. S. Seiter, R. Arch, S. Reber, D. Komitowski, M. Hofmann, H. Ponta, P. Herrlich, S. Matzku, M. Zoller, Prevention of tumor metastasis formation by anti-variant CD44, J. Exp. Med. 177 (1993) 443 – 455. M. Culty, M. Shizari, E.W. Thompson, C.B. Underhill, Binding and degradation of hyaluronan by human breast cancer cell lines expressing different forms of CD44: correlation with invasive potential, J. Cell. Physiol. 160 (1994) 275 – 286. S.B. Wallach-Dayan, V. Grabovsky, J. Moll, J. Sleeman, P. Herrlich, R. Alon, D. Naor, CD44-dependent lymphoma cell dissemination: a

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

cell surface CD44 variant, rather than standard CD44, supports in vitro lymphoma cell rolling on hyaluronic acid substrate and its in vivo accumulation in the peripheral lymph nodes, J. Cell Sci. 114 (2001) 3463 – 3477. K. Fujii, Y. Tanaka, S. Hubscher, K. Saito, T. Ota, S. Eto, Crosslinking of CD44 on rheumatoid synovial cells up-regulates VCAM-1, J. Immunol. 162 (1999) 2391 – 2398. M.L. Dustin, R. Rothlein, A.K. Bhan, C.A. Dinarello, T.A. Springer, Induction by IL 1 and interferon-gamma: tissue distribution, biochemistry, and function of a natural adherence molecule (ICAM-1), J. Immunol. 137 (1986) 245 – 254. C.W. Marlor, D.L. Webb, M.P. Bombara, J.M. Breve, M.L. Blue, Expression of vascular cell adhesion molecule-1 in fibroblast-like synoviocytes after stimulation with tumor necrosis factor, Am. J. Pathol. 140 (1992) 1055 – 1060. A. Lundstrom, J. Holmbom, C. Lindqvist, T. Nordstrom, The role of alpha2 beta1 and alpha3 beta1 integrin receptors in the initial anchoring of MDA-MB-231 human breast cancer cells to cortical bone matrix, Biochem. Biophys. Res. Commun. 250 (1998) 735 – 740. R.B. Lichtner, A.R. Howlett, M. Lerch, J.A. Xuan, J. Brink, B. Langton-Webster, M.R. Schneider, Negative cooperativity between alpha 3 beta 1 and alpha 2 beta 1 integrins in human mammary carcinoma MDA MB 231 cells, Exp. Cell Res. 240 (1998) 368 – 376. G.P. Gui, J.R. Puddefoot, G.P. Vinson, C.A. Wells, R. Carpenter, In vitro regulation of human breast cancer cell adhesion and invasion via integrin receptors to the extracellular matrix, Br. J. Surg. 82 (1995) 1192 – 1196. T. Meyer, J.F. Marshall, I.R. Hart, Expression of alphav integrins and vitronectin receptor identity in breast cancer cells, Br. J. Cancer 77 (1998) 530 – 536. A. Herrera-Gayol, S. Jothy, CD44 modulates Hs578T human breast cancer cell adhesion, migration, and invasiveness, Exp. Mol. Pathol. 66 (1999) 99 – 108.