HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 24, 21930-21938, June 14, 2002

This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 277/24/21930    most recent M201630200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Pilch, J. Articles by Felding-Habermann, B. Search for Related Content PubMed PubMed Citation Articles by Pilch, J. Articles by Felding-Habermann, B. Social Bookmarking                      What's this?

Unique Ability of Integrin _v3 to Support Tumor Cell Arrest under Dynamic Flow Conditions^*,

Jan Pilch, Rolf Habermann, and Brunhilde Felding-Habermann

From the Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla, California 92037

Received for publication, February 18, 2002, and in revised form, April 2, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Shear-resistant arrest of circulating tumor cells is required for metastasis from the blood stream. Arrest during blood flow can be supported by tumor cell interaction with attached, activated platelets. This is mediated by tumor cell integrin _v3 and cross-linking plasma protein ligands. To analyze the mechanism of tumor cell ligand interactions under dynamic flow conditions, we used real-time video microscopy and tested human melanoma cell binding to fibrinogen, von Willebrand Factor, or fibronectin matrices in a buffer perfusion system. When perfused at venous flow, melanoma cells arrested abruptly and began to spread immediately. This was uniquely mediated by integrin _v3 on all tested ligands, and required _v3 activation and actin polymerization. Under static conditions, _v3 cooperated with _v1 and ₅₁ in supporting melanoma cell adhesion to fibronectin. But even when activated, ₁ integrins did not contribute to melanoma cell arrest during flow. Soluble ligand served as a cross-linker between attached and circulating tumor cells and enhanced melanoma cell arrest. Cohesion of activated melanoma cells was restricted to the matrix surface and did not occur in suspension. We conclude that the presence of _v3 in a functionally activated state provides a unique advantage for circulating tumor cells by promoting tumor cell arrest in the presence of flow-dependent shear forces.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Metastasis to distant organs is a key characteristic of malignancy. This often involves the blood stream, where circulating tumor cells are exposed to flow-dependent shear forces that physically oppose tumor cell anchorage. Tumor cell arrest within the vasculature is required for intravascular growth and extravasation within target organs of metastasis (1-3). Arrest at the vessel wall depends on specific adhesive mechanisms rather than passive entrapment, because tumor cell variants that differ in their adhesion receptor outfits have distinct metastatic activities, and blocking of specific adhesion receptors can inhibit metastasis (4). Tumor cell arrest during blood flow can be supported by tumor cell interaction with activated platelets (5-10). We showed earlier that human melanoma and breast cancer cells use integrin _v3 to bind to platelet integrin _IIb3 via cross-linking plasma protein ligands (5, 6). Support of tumor cell-platelet interaction requires activation of integrin _v3. Importantly, tumor cells that stably express activated _v3 metastasize very aggressively, in contrast to those expressing the non-activated receptor (6). Thus, the regulation of _v3 expression in tumor cells and determinants that control the activation state of the receptor may directly affect the metastatic activity of circulating tumor cells.

In leukocytes and platelets, the ability to arrest within the vasculature is tightly regulated and depends on integrin activation (11-13). Generally, integrin activation results in increased affinity for ligand and is accompanied by conformational changes within the / heterodimer (12). In addition, the cellular avidity of the receptors can be enhanced by controlled lateral diffusion within the plasma membrane and by interaction with the cytoskeleton (11). Integrins with increased affinity and avidity promote arrest of blood-borne cells at matrices by supporting rapid and stable receptor-matrix interaction. It is possible that the functional state of tumor cell integrins is controlled in a similar manner and determines the ability of circulating metastatic cells to arrest within the vasculature.

Analysis of adhesive tumor cell functions in a blood perfusion system allowed us to identify distinct functional states of tumor cell integrin _v3 (6). When blood containing tumor cells streams past reactive matrices, a multitude of complex cell-cell and cell-ligand interactions can occur, because blood cells and plasma proteins may affect the adhesive behavior of the tumor cells. Consequences of leukocyte and platelet activation during blood flow are difficult to control individually and add to the complexity of the experimental system. We therefore simplified the analytical conditions to investigate mechanisms of tumor cell ligand interaction in the presence of flow dependent shear forces. We chose a buffer perfusion system combined with real-time video microscopy and examined adhesive interactions of human melanoma cells with individual matrix proteins under a variety of defined flow conditions. We found that integrin _v3 has the unique ability to support melanoma cell arrest during flow on different matrix proteins. Other _v and ₁ integrins participated in melanoma cell adhesion to the same ligand(s) under stationary conditions, but were unable to contribute to melanoma cell arrest, when the cells were in motion. To support melanoma cell arrest, integrin _v3 had to be activated. Rapid _v3-ligand interaction broke the flow of circulating melanoma cells, and the receptor immediately colocalized with polymerizing actin to promote cell spreading. Soluble ligand enhanced melanoma cell arrest during flow by supporting melanoma cell cohesion at the matrix surface. This did not occur in suspension and did not interfere with cell arrest at the matrix. These findings contribute to an understanding of specific adhesive functions that characterize tumor cells, which successfully metastasize from the circulation.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Reagents-- Function blocking monoclonal antibodies (80-100 µg/ml) were used to analyze the contribution of individual integrins to melanoma cell adhesion: VNR1 27.1 (anti-_v3) (14), 7E3 (anti-₃) (15), P1D6 (anti-₅), and P5D2 (anti-₁) (16). Monoclonal antibodies AV-8 (anti-_v), AV-10 (anti-₃) (5), LJ-CP8 (anti-_IIb3) (17) were used to verify melanoma cell integrin expression by flow cytometry. The peptide antagonist GRGDSPK and control peptide GRGESPK were from The Torrey Pines Institute of Molecular Studies (La Jolla, CA). Human fibrinogen was from Enzyme Research Laboratories Inc. (South Bend, IN). The integrity of the fibrinogen A, B, and  chains was verified by SDS-PAGE analysis. Human plasma fibronectin was from Calbiochem (San Diego, CA) and human von Willebrand Factor was a gift from Z. M. Ruggeri (The Scripps Research Institute). Cytochalasin D and Hoechst 33342 were from Sigma Chemical Co. (St. Louis, MO); Alexa 546-phalloidin and Prolong antifade mounting medium were from Molecular Probes (Eugene, OR).

Cells and Culture-- M21 human melanoma cells, their _v-integrin lacking variant M21-L (18), and the _v-reconstituted, _v3-expressing transfectant M21-L4 (19) were grown in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum, 1 mM pyruvate, and 2 mM L-glutamine in 5% CO₂. The cells tested free from mycoplasma during this study (Mycoplasma Plus, Stratagene, La Jolla, CA). In preparation for functional tests, the cells were starved in serum-free medium overnight, harvested with EDTA (0.02% in PBS¹), and resuspended in Hepes/Tyrode's buffer, pH 7.4 (10 mM Hepes, 140 mM NaCl, 2.7 mM KCl, 0.4 mM NaH₂PO₄ × H₂O, 10 mM NaHCO₃, 5 mM dextrose). As specified for each experiment, MnCl₂ (0.2 mM) and/or CaCl₂ (1 mM) were added directly before the adhesion or perfusion assays.

Static Cell Adhesion Assay-- Melanoma cell adhesion under static conditions was measured in 48-well plates (Costar, polystyrene, non-tissue culture treated). The plates were coated overnight at 4 °C with 5-10 µg/ml fibrinogen, von Willebrand Factor, or fibronectin, and then blocked with 2% BSA (1 h at room temperature). The plates were washed, and melanoma cell suspension was added (2 × 10⁵ in 200 µl/well) with our without function blocking antibodies or peptides. When inhibitors were used, the cells were incubated with the inhibitors for 15 min at 22 °C before adding the cell suspension to the adhesion plates. The cells were allowed to attach for various time periods at 37 °C and 5% CO₂. The incubation periods were stopped by aspirating the cell suspension and removing unattached cells by three gentle washings with 200 µl of PBS. Attached cells were quantified by measuring cellular phosphatase with para-nitrophenol phosphate (5 mg/ml in 50 mM sodium acetate, 1% Triton X-100, pH 5.2) as substrate. The reaction was stopped with NaOH, and the reaction product was measured at 405 nm in an enzyme-linked immunosorbent assay plate reader. Nonspecific cell adhesion was measured on BSA-coated wells (less than 5% of adhesion on specific substrates). Given values were corrected for nonspecific binding.

Measurement of Cell Adhesion under Flow-- Tumor cell arrest during flow was measured by real-time video epifluorescence microscopy as described previously (5) but with specific modifications. The melanoma cells were stained with hydroethidine (Polysciences Inc., Warrington, PA) (20 µg/ml in Hepes/Tyrode's buffer at 37 °C for 30 min) and washed twice. Glass coverslips (24 × 50 mm) were used as the bottom of a parallel plate flow chamber and coated with a 200-µl solution of fibrinogen, fibronectin (100 µg/ml), or von Willebrand Factor (20 µg/ml) in PBS, pH 7.4, at 22 °C for 2 h in a humid atmosphere. Before assembling the flow chamber, the coverslips were rinsed with Hepes/Tyrode's, the chamber filled with this buffer and connected to a pump system (Harvard Apparatus Inc., Holliston, MA). The cell suspension was prewarmed to 37 °C and kept at this temperature throughout the perfusion experiments. During the initial period of the experiments (referred to as the "ON" phase in the videos; see Supplemental Material), cell suspensions (5 × 10⁵/ml) were perfused at a constant wall shear rate of 50 s¹. After 10 min, the cell suspension was replaced by buffer (containing the same cation concentrations as the initial cell suspension), without interrupting the flow. After 1 min of continued perfusion at 50 s¹, the wall shear rate was stepwise increased every 2 min (250, 500, 1000, and 1500 s¹) by proportionally rising the flow rate (referred to as the "OFF" phase in the videos; see Supplemental Material). During the perfusion experiments, cell-cell and cell-matrix interactions were visualized and recorded at 543/590 nm (excitation/emission, red fluorescence of the hydroethidine-stained tumor cells). Cell adhesion was quantified by directing the automated stage of the microscope to predefined positions and by automatically capturing images at the same positions, each minute during the initial ON phase of the experiment, and every other minute during the OFF phase of the experiment. Thus, the area was monitored at the same positions after it had been exposed to an increased wall shear rate for 2 min. To quantify cell adhesion and to measure the size of attached particles (single cells, or multi-cell aggregates), the captured images were analyzed by image processing (MetaMorph, Universal Imaging Corp., Downingtown, PA). The given data represent the average number of arrested cells in 5 optical fields ± standard deviation. Each experiment was carried out at least twice with very similar results. The attached videos were generated with video editing software (Premiere 6.0, Adobe Systems Inc., San Jose, CA) on a Targa 3000 editing system (Pinnacle Systems, Mountain View, CA).

Deconvolution Microscopy-- Deconvolution microscopy was used to analyze the localization of integrin _v3 in relation to polymerizing actin filaments in melanoma cells just after arrest on a matrix during flow (20). Unlabeled M21 melanoma cells were perfused at a venous wall shear rate of 50 s¹ over a fibrinogen or fibronectin matrix. Briefly after starting the perfusion, the flow chamber was disassembled while submersed in perfusion buffer to avoid unwanted increase in shear exposure. The coverslips were gently rinsed, cells fixed with 3.5% formaldehyde (20 min), permeabilized with 0.5% Triton X-100 (10 min), blocked with 2% BSA (20 min), and stained with a combination of anti-₃ mAb AV-10 (10 µg/ml) and Alexa Fluor 546-phalloidin, followed by fluorescein isothiocyanate-anti mouse and nuclear stain Hoechst 33342 (14 µg/ml) (all steps at 22 °C). The coverslips were mounted onto glass slides in Prolong AntiFade mounting medium and stored at 20 °C until analyzed on a Delta Vision Optical Sectioning Microscope Model 283. Images of optical sections were acquired along the z-axis of arrested cells in 0.5-µm increments. Three sets of images were acquired at each x, y, z position with filters set to detect green, red, or blue fluorescence. The digitized images were processed with deconvolution software softWoRx version 2.5 to reduce out-of-focus fluorescence in the three-dimensional (3D) reconstruction of the combined images. To analyze the subcellular localization of integrin _v3 and F-actin, the specific signals were assigned pseudo colors (green for _v3 and red for F-actin). The distribution of green and red signal was analyzed by image processing (MetaMorph). Overlap of green and red signals, indicating colocalization of _v3 and F-actin, was assigned a yellow color, and the relative distribution of green, red, and yellow signals was determined and quantified.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Melanoma Cells Arrest on Immobilized Plasma Proteins under Dynamic Flow Conditions-- We previously demonstrated that human tumor cells can attach to adherent, activated platelets during blood flow, and that plasma proteins are required for this interaction (5, 6). To analyze the mechanism of tumor cell ligand binding during flow, we chose a buffer perfusion system and individual matrix proteins to reduce the complexity of possible cell-cell and cell-ligand interactions that can occur, when tumor cells are suspended in blood and allowed to stream past reactive surfaces. When suspended in buffer and perfused over fibrinogen, von Willebrand Factor or fibronectin matrices at a venous wall shear rate of 50 s¹ (4 dynes/cm²), M21 human melanoma cells arrested at each of these matrices. An increasing number of cells was recruited to the surface as perfusion continued (Fig. 1). Melanoma cells that came in contact with the matrix arrested abruptly without previous rolling and began to spread immediately (see video of Fig. 1 in Supplemental Material). This initial period of the experiment is referred to as the ON phase in the videos in Figs. 2 and 8 (see below). To test the stability of melanoma cell attachment during flow, the cell suspension was replaced by buffer without interrupting the flow, and the wall shear rate was stepwise increased every 2 min to 250, 500, 1000, and 1500 s¹, respectively. This period of the experiment is referred to as the OFF phase in the videos in Figs. 2 and 8 (see Supplemental Material). Once established, melanoma cell attachment to each of the tested matrix proteins resisted increasing wall shear rates, including those at arterial levels (1500 s¹) (Fig. 1). This indicates that melanoma adhesion receptor(s) can rapidly engage in matrix interaction and break the flow of cells, which stream past the matrix under flow conditions as found in the venous circulation.

View larger version (33K): [in this window] [in a new window]   Fig. 1.   Melanoma cells arrest on immobilized plasma proteins under dynamic flow conditions. M21 melanoma cells were perfused over immobilized fibrinogen (Fg), von Willebrand Factor (VWF), or fibronectin (FN) in Hepes/Tyrode's buffer containing 0.2 mM Mn2+ as detailed under "Experimental Procedures." During the first 10 min, the wall shear rate was kept constant at a wall shear rate of 50 s1 (4 dynes/cm2) (venous flow). Then, the cell suspension was replaced by buffer (same cation milieu) without interrupting the flow, and the wall shear rate was stepwise increased every 2 min. Cell-cell and cell-matrix interactions were recorded by real-time video microscopy. Images were acquired every minute at identical x,y positions to determine the number of arrested cells. This figure includes a video (see the Supplemental Material for the video demonstration). It shows in real time (original velocity) how M21 melanoma cells that come in contact with a fibrinogen matrix arrest abruptly without previous rolling. The cells were perfused at a wall shear rate of 50 s1.

View larger version (48K): [in this window] [in a new window]   Fig. 2.   Integrin v3 mediates M21 melanoma cell arrest during flow. M21 (v3 +), M21-L (v3 ), or M21-L4 (v3 +) cells were perfused over fibrinogen (Fg), von Willebrand Factor (VWF), or fibronectin (FN) in Hepes/Tyrode's buffer containing 0.2 mM Mn2+ at a venous wall shear rate of 50 s1. The top panel shows images acquired during flow after 1-, 5-, 10-, and 18-min perfusion of M21 or M21-L cells over fibronectin. After 10 min, the wall shear rate was stepwise increased and reached 1500 s1 after 18 min. In the bottom panel, M21, M21-L, or M21-L4 cells were perfused at 50 s1 in the presence or absence of function blocking anti-v3 mAb VNR1 (80 µg/ml) or GRGDSPK peptide (100 µM). M21-L4 cells are variants of M21-L, in which v3 expression was restored by transfection. The experimental set-up, image acquisition during flow, and evaluation are detailed under "Experimental Procedures." This figure includes a video (see the Supplemental Material for the video demonstration). It shows in twice the original velocity that v3 supports M21 cell arrest during flow on a fibronectin matrix during venous flow and that cell arrest is stable in the presence of increasing wall shear rates up to arterial levels. A split screen shows M21 cells (v3+) in the top portion and M21-L cells (v3) in the bottom portion of the screen. The first half of the video shows the ON phase of the experiment at various time points, where the cell suspensions are perfused at a wall shear rate of 501. The second half of the video shows the OFF phase, where the cell suspensions were replaced by buffer without interrupting the flow, and the wall shear rates were stepwise increased. All frames of the video were recorded at the same x,y position.

Integrin _v3 Mediates Melanoma Cell Arrest during Flow-- We next examined which adhesion receptor(s) supported melanoma cell arrest on fibrinogen, von Willebrand Factor, or fibronectin matrices under dynamic flow conditions. We previously showed that integrin _v3 can mediate tumor cell interaction with platelets during blood flow and thereby support tumor cell arrest. To test whether _v3 can directly support melanoma cell arrest at immobilized matrix proteins, we compared _v3-expressing M21 cells against their _v-integrin lacking variant M21-L (Fig. 2 and video from Fig. 2 in Supplemental Material). The _v3-positive M21 cells arrested efficiently on each of the tested matrix proteins, and the attachment was resistant to increasing wall shear rates. In contrast, _v3-lacking M21-L cells failed to attach to any of the tested matrix proteins. The ability to arrest during flow was fully restored in M21-L4 cells, which were transfected to restore _v3 expression. This indicates that integrin _v3 function is required for melanoma cell arrest under flow conditions. This was confirmed by blocking M21 cell arrest on each of the tested matrix proteins with a function blocking anti-_v3 antibody. Melanoma cell arrest was also abolished in the presence of GRGDSPK peptide, indicating that the RGD recognition motif within each of the tested ligands is necessary and sufficient to support _v3 mediated ligand binding and cell arrest during flow (Fig. 2).

Integrin _v3 Cooperates with Other _v Integrins and ₁ Integrins in Static Melanoma Cell Adhesion-- Melanoma cells, specifically M21 cells, contain more than one adhesion receptor that can support cell attachment to some of the tested matrix proteins (18, 19). Fibrinogen is recognized by _v3 and _v1, von Willebrand Factor is recognized by _v3, and fibronectin is recognized by _v1 and ₅₁ in addition to _v3 (21, 22). Under stationary conditions, M21 cells attached to fibrinogen and von Willebrand Factor in a strictly _v-integrin-dependent manner, because _v-lacking M21-L cells failed to attach to these matrix proteins within the measured time period (90 min) (Fig. 3). On fibronectin, _v3 cooperated with other integrins to support melanoma cell adhesion under static conditions. The _v3-positive M21 cells and _v3-negative M21-L cells attached to fibronectin almost equally well. In M21 cells, which express _v3, _v5, _v1, and ₅₁, adhesion to fibronectin could not be inhibited by individual function blocking antibodies to any of these receptors. But a combination of anti-₃ and anti-₅ reduced M21 cell adhesion to fibronectin by 30%, and a combination of anti-₃ and anti-₁ antibodies reduced adhesion by more than 80%. In contrast, in M21-L cells, which express ₅₁ as their only fibronectin receptor, adhesion to this protein was abolished either by function blocking anti-₅ or anti-₁ antibody (Fig. 3). This indicates that integrin _v3 cooperates with other _v integrins and ₁ integrins in supporting M21 melanoma cell adhesion to fibronectin under static conditions. Each of these adhesion receptors was capable of supporting static adhesion, when the other receptors were inhibited. In the absence of _v integrin expression, ₅₁ alone efficiently supported melanoma cell adhesion to fibronectin during stasis. Thus, while cooperating with other integrins in supporting static melanoma cell adhesion, _v3 is the only receptor on M21 melanoma cells that supports cell arrest under dynamic flow conditions. During stasis, melanoma cell adhesion to fibronectin was only partially inhibited by RGD peptide (Fig. 3).

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Integrin v3 cooperates with other v integrins and 1 integrins in supporting static melanoma cell adhesion. M21 (v3+) or M21-L (v3) melanoma cells were allowed to attach to fibrinogen (Fg), von Willebrand Factor (VWF), or fibronectin (FN) under static conditions in the presence or absence of function blocking anti-integrin antibodies (80 µg/ml) or GRGDSPK peptide (100 µM). The cells were incubated with inhibitory antibody or peptide for 15 min before plating. Adhesion time was 30 min. M21-L cells did not attach to fibrinogen and von Willebrand Factor even after 90 min.

Integrin _v3 Activation Is Required for Support of Melanoma Cell Arrest during Flow-- Integrin activation is required for leukocyte and platelet attachment during blood flow (12, 13). We found earlier that arrest of blood-borne tumor cells also depends on integrin activation (6). To analyze directly, whether the activation state of tumor cell integrin _v3 determines the ability of the receptor to support cell arrest on defined matrices, we allowed _v3-positive M21 melanoma cells to stream past immobilized fibrinogen, von Willebrand Factor or fibronectin in the presence of Mn²⁺ to activate the tumor cell integrins. Control perfusions were done in Ca²⁺. To test, whether a general activation of integrins with Mn²⁺ would promote melanoma cell arrest in the absence of integrin _v3, we analyzed _v3-lacking M21-L cells under the same conditions. When perfused at a venous wall shear rate of 50 s¹, melanoma cell arrest was measurable only with _v3-positive M21 cells in the presence of Mn²⁺. Mn²⁺-treated M21 cells arrested efficiently on each of the tested matrix proteins. But M21 cells were unable to interact with any tested matrix when perfused in Ca²⁺. M21-L cells lacking _v3 could not attach to any of the tested matrices under dynamic flow conditions, regardless of the cation environment (Fig. 4, top panel). This indicates that integrin _v3 needs to be activated to support tumor cell arrest during flow and that no other integrin shared by M21 and M21-L cells can do the same, not even when activated. Integrin ₅₁ is shared by M21 and M21-L cells and readily supported M21-L cell adhesion to fibronectin under static conditions. To analyze whether the functionality of ₅₁ changed in response to Mn²⁺ stimulation, we tested time-dependent melanoma cell adhesion during stasis in Mn²⁺ versus Ca²⁺. In the presence of Ca²⁺, M21-L cell adhesion to fibronectin was measurable after 30 min, and the number of attached cells continued to increase by 90 min of incubation. In Mn²⁺, M21-L cells began to attach to fibronectin during the first minutes of incubation, and the number of attached cells reached a plateau after 60 min (Fig. 4, bottom panel). This indicates that ₅₁, the only fibronectin receptor of M21-L cells, changed its functionality in response to Mn²⁺ treatment, because M21-L cells attached to fibronectin faster in Mn²⁺ than in Ca²⁺. However, this change in ₅₁ activation was not sufficient to support M21-L cell arrest under dynamic flow conditions (Fig. 4, top panel). A general tendency to support melanoma cell adhesion faster in Mn²⁺ than in Ca²⁺ was also observed with _v3-positive M21 cells on all tested matrix proteins. M21-L cells lacking _v did not recognize fibrinogen, vitronectin, or von Willebrand Factor as adhesive substrates, regardless of the cation environment. Together, stimulation with Mn²⁺ improved the ability of ₅₁ to mediate accelerated melanoma cell adhesion to fibronectin (Fig. 4, bottom panel), but this was not sufficient to support melanoma cell arrest during flow (Fig. 4, top panel).

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Integrin v3 activation is required for support of melanoma cell arrest during flow. In the top panel, M21 (v3+) or M21-L (v3) cells were perfused over immobilized fibrinogen (Fg), von Willebrand Factor (VWF), or fibronectin (FN) at a venous wall shear rate of 50 s1 in Hepes/Tyrode's buffer containing 0.2 mM Mn2+ or 1 mM Ca2+. During flow, images were acquired every minute at identical x,y positions as detailed under "Experimental Procedures." In the bottom panel, M21 or M21-L cells were allowed to attach to fibrinogen (Fg), von Willebrand Factor (VWF), fibronectin (FN), or vitronectin (VN) for increasing time periods in the presence of 1 mM Ca2+ (left) or 0.2 mM Mn2+ (right).

Flow-resistant Melanoma Cell Arrest through Activated _v3 Occurs at Physiologic Calcium Concentrations-- The ability of tumor cells to arrest under dynamic flow conditions is relevant for circulating metastatic cells, which depend on specific mechanisms that support their attachment within the vasculature of target organs. Our results show that integrin _v3 has the unique ability to support melanoma cell arrest during flow, when the receptor is activated. Physiologic Ca²⁺ concentrations in the millimolar range are potentially antagonistic for _v3 activation, because Ca²⁺ may suppress cell adhesion by reducing the affinity of _v3 for certain ligands (23-25). Therefore, we tested whether experimentally activated _v3 can support melanoma cell arrest during flow at physiologic Ca²⁺ levels. To do this, we first determined under static conditions which Mn²⁺ concentration best enhanced short term M21 cell adhesion to fibrinogen, as a measure of integrin activation. We then allowed the cells to adhere to fibrinogen in the presence of the optimal Mn²⁺ concentration plus increasing levels of Ca²⁺ (Fig. 5, top panel). Ca²⁺ inhibited Mn²⁺ dependent cell adhesion by 50% at 500 µM Ca²⁺, but inhibition was not increased, when more Ca²⁺ was added. Under venous flow, M21 cell arrest on fibrinogen was reduced in the presence of 1 mM Ca²⁺ (less with increasing perfusion time), but cell arrest was still measurable at significant levels (Fig. 5, bottom panel). As a control, Ca²⁺ alone did not support melanoma cell arrest. Similar results were obtained on fibronectin and von Willebrand Factor. This indicates that activated integrin _v3 can support tumor cell arrest at immobilized matrices in the presence of physiologic Ca²⁺ concentrations as found in blood.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Flow-resistant melanoma cell arrest through activated integrin v3 occurs at a physiologic calcium concentration. Top panel, M21 (v3+) melanoma cells were allowed to attach to a fibrinogen matrix under static conditions in the presence of increasing Mn2+ concentrations (left) or increasing Ca2+ concentrations and a constant Mn2+ concentration of 0.2 mM. Adhesion time was 45 min. Bottom panel, M21 (v3+) cells were perfused over a fibrinogen matrix at a venous wall shear rate of 50 s1 in Hepes/Tyrode's buffer containing either 0.2 mM Mn2+, 0.2 mM Mn2+ plus 1 mM Ca2+, or 1 mM Ca2+. Images were captured during flow at the indicated time periods at identical x,y positions as detailed under "Experimental Procedures."

Actin Polymerization Is Required for _v3-mediated Melanoma Cell Arrest during Flow-- Activated integrin _v3 supports arrest of melanoma cells under venous flow conditions. When touching the matrix, the tumor cells arrested abruptly, without previous rolling. This indicates that activated _v3 mediates instant, firm adhesion and does not require other adhesion receptors, which slow the tumor cells down. Once attached, adherent melanoma cells were resistant to increasing wall shear rates, including rates at arterial levels. This is possible only if the adhesion receptor binds the immobilized ligand very quickly and if the interaction is stabilized instantly. During the perfusion experiments, we observed that melanoma cells, which came in contact with the matrix surface, began to spread immediately. Cell spreading involves rearrangement of the cytoskeleton and actin polymerization (26). In M21 melanoma cells that had just arrested at a matrix surface during venous flow, integrin _v3 and filamentous actin distributed toward the cell-matrix contact site and colocalized at the perimeter of the area, where cells touched the matrix (Fig. 6). Inhibition of actin polymerization by cell treatment with cytochalasin D (0.5 or 5 µM) reduced or abolished melanoma cell arrest at venous flow and shear resistant cell attachment depending on the drug concentration (Fig. 7). This indicates that 1) melanoma cell arrest through activated _v3 and the establishment of shear-resistant adhesion require the actin cytoskeleton in post-receptor events and 2) activated _v3, as the mediating adhesion receptor, colocalizes with the actin cytoskeleton. This likely stabilizes tumor cell arrest.

View larger version (25K): [in this window] [in a new window]   Fig. 6.   Integrin v3 colocalizes with filamentous actin at the perimeter of matrix contact when melanoma cells arrest during flow. M21 (v3 +) cells were perfused over a fibrinogen matrix at a venous wall shear rate of 50 s1 for 5 min, fixed, permeabilized, and stained for v3 (fluorescein isothiocyanate-mAb AV-10) and for filamentous actin (Alexa Fluor 546-phalloidin) and analyzed by deconvolution microscopy as detailed under "Experimental Procedures." A, three-dimensional reconstruction of optical sections through a melanoma cell briefly after matrix contact (side view). The nucleus is shown in blue (Hoechst 33342). v3 and F-actin localize toward the matrix contact site. B, optical cross-section through the same cell as shown in A at the z-plane of the cell-matrix contact. Shown is the distribution of F-actin (red). C, same optical section as in B, but showing the distribution of v3 (green). D, overlay of B and C (signal intensified), overlap of red and green signal was assigned a yellow color, indicating colocalization of F-actin and v3.

View larger version (22K): [in this window] [in a new window]   Fig. 7.   Actin polymerization is required for v3-mediated melanoma cell arrest during flow. M21 (v3 +) cells were perfused over a fibrinogen matrix (in 0.2 mM Mn2+) at a venous wall shear rate of 50 s1 for 10 min in the presence or absence of 0.5 or 5 µM cytochalasin D. After 10 min, the cell suspension was replaced by buffer without interrupting the flow, and the wall shear rate was stepwise increased every 2 min. Images were acquired during flow at identical x,y positions in 1- or 2-min intervals. The experimental set-up and image processing are detailed under "Experimental Procedures."

Soluble _v3 Ligand Enhances Melanoma Cell Arrest under Flow-- Activated tumor cell integrin _v3 binds soluble ligand (6, 27). Circulating metastatic cells are surrounded by plasma proteins, several of which are ligands for _v3. We therefore asked whether soluble ligands would compete with immobilized ligands for _v3 binding and possibly interfere with _v3-mediated tumor cell arrest at a matrix, when the adhesion receptor is activated. To test this, M21 melanoma cells were perfused over a fibrinogen matrix in the presence of soluble fibrinogen (0.5 mg/ml). Under venous flow conditions, melanoma cell arrest was not inhibited but enhanced by the addition of soluble fibrinogen, which served as cross-linking ligand between attached and circulating tumor cells (Fig. 8 and the video for Fig. 8 in Supplemental Material). Cohesion of activated melanoma cells was restricted to the matrix surface and did not occur in suspension. In the presence of soluble ligand, melanoma cell aggregates, which formed at the matrix surface during venous flow, broke apart when the wall shear rate was increased to arterial levels. But tumor cells that were attached to the matrix remained attached. Thus, cell-matrix interaction is resistant to arterial shear rates, but cell-cell cohesion that depends on cross-linking _v3 ligand is not. Together, this indicates that binding of soluble ligand by activated integrin _v3 does not interfere with tumor cell arrest, which also depends on the function of the activated receptor. Binding of soluble ligand rather enhances tumor cell arrest by promoting cell-cell cohesion at the matrix surface. This is potentially relevant for the arrest of metastatic tumor cells within blood vessels at low wall shear rates.

View larger version (39K): [in this window] [in a new window]   Fig. 8.   Soluble ligand for v3 enhances melanoma cell arrest during flow. M21 (v3+) cells were perfused over a fibrinogen matrix (in 0.2 mM Mn2+) at a venous wall shear rate of 50 s1 in the presence or absence of 0.5 mg/ml soluble fibrinogen. After 10 min, the cell suspension was replaced by buffer without interrupting the flow, and the wall shear rate was stepwise increased every 2 min. Images were acquired during flow at identical x,y positions in 1- or 2-min intervals. The experimental set-up and image processing are detailed under "Experimental Procedures." This figure includes a video (see the Supplemental Material for the video demonstration). It shows in twice the original velocity that soluble fibrinogen enhances M21 cell arrest at a fibrinogen matrix during venous flow by supporting cell-cell cohesion at the matrix surface. It also shows that cell-matrix, but not cell-cell interaction, is stable in the presence of increasing wall shear rates. A split screen shows M21 cells (v3+) perfused at 50 s1 in the presence (top portion) or the absence (bottom portion) of soluble fibrinogen. The first half of video shows the ON phase of the experiment at various time points, where the cell suspensions were perfused at a wall shear rate of 501. The second half of the video shows the OFF phase, where the cell suspensions were replaced by buffer without interrupting the flow, and the wall shear rates were stepwise increased. All frames of the video were recorded at the same x,y position.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Tumor cell arrest within the vasculature is a prerequisite for metastasis from the blood stream. Tumor cells must attach to components of the vessel wall, regardless of whether they begin to proliferate within the vasculature, or whether the cells extravasate and start to grow within the parenchyma of their target organ (1-3). Circulating tumor cells are exposed to shear stress, which is generated by blood flow and physically opposes cell attachment. Tumor cell arrest depends on specific adhesive tumor cell functions (4), and these are likely different from those that mediate tumor cell attachment under static conditions. We demonstrated previously that tumor cells can arrest during blood flow by binding to the surface of attached and activated platelets. This interaction requires plasma proteins as cross-linking ligands (5). Here, we analyzed the mechanism of tumor cell matrix interaction during flow in a defined buffer perfusion system, to contribute to a basic understanding of tumor cell adhesive functions in the presence of flow-dependent shear forces. From our results, we draw the following conclusions: 1) Melanoma cells can attach to immobilized plasma proteins under venous flow conditions. In the tested system, melanoma cell arrest during flow is uniquely mediated by integrin _v3. On some of the tested matrix proteins, _v3 cooperates with other _v integrins and with ₁ integrins to support melanoma cell adhesion under static conditions, but _v3 is the only receptor on M21 melanoma cells that mediates cell arrest during flow. 2) Integrin _v3 activation is required for support of shear-resistant melanoma cell arrest. 3) Actin polymerization is needed to sustain _v3-initiated melanoma cell arrest during flow. 4) The ability of activated _v3 to bind soluble ligand enhances melanoma cell arrest under dynamic flow conditions by promoting adhesion of circulating tumor cells to matrix attached cells.

For our studies, we chose human melanoma cells, because these are known to utilize the blood stream for metastatic dissemination. To analyze adhesive tumor cell properties under flow conditions, we perfused melanoma cells over matrices of immobilized fibrinogen, von Willebrand Factor, or fibronectin. These plasma proteins are relevant, because they were shown to contribute to hematogenous metastasis (7, 28). The ligands may serve as adhesive matrices, when immobilized at the surface of activated platelets, leukocytes, or endothelial cells, or as constituents of the subendothelial matrix. Under static conditions M21 human melanoma cells attach to fibrinogen via _v3 and likely _v1, to von Willebrand Factor via _v3, and to fibronectin via _v3, _v1, and ₅₁ (18, 19, 29). Under flow conditions, M21 cell arrest was uniquely mediated by integrin _v3, whereas the other receptors could not contribute to this event. M21 cells do not express the fibronectin receptor ₄₁ (not shown), which can contribute to leukocyte arrest on this matrix (30). Melanoma cell arrest in the presence of venous wall shear rates occurred abruptly and without previous rolling. This indicates that _v3-matrix interaction can break the flow of a tumor cell and immediately stabilize firm arrest. It is possible that other integrins help to stabilize cell attachment, once matrix contact was established through _v3. However, our results on von Willebrand Factor show that _v3 is sufficient to do both, because it it is the only receptor that recognizes this matrix. Although the RGD recognition motif and additional synergy sequences within certain ligands, such as fibrinogen and fibronectin, may jointly support static melanoma cell adhesion when several integrins are engaged (31, 32), M21 melanoma cell arrest during flow depended entirely on the RGD motif recognized by activated _v3.

Circulating leukocytes slow down through tethering and rolling before they attach firmly to the vessel wall (33). Similarly, platelet attachment at very high wall shear rates requires initial transient matrix interaction (34, 35). Evidence suggests that blood-borne tumor cells can also tether and roll before they arrest firmly, and this may involve tumor cell interaction with platelets (9, 10, 36). Our results show that tumor cells can engage immediately in firm arrest through _v3-ligand binding under venous flow conditions, without previous rolling. This is true for platelet-supported tumor cell arrest during blood flow (5, 6) and for direct tumor cell arrest on individual matrix proteins. This unique ability of _v3 may help tumor cells to arrest under higher than venous wall shear rates, if supportive adhesive mechanisms slow the cells down.

Attachment of circulating blood cells is regulated by the activation state and avidity of their adhesion receptors (11, 12). We showed previously that tumor cell integrin _v3 must be activated to support platelet-dependent arrest of human breast cancer cells (6). Importantly, breast cancer cells can stably express _v3 in an activated state, and the cells are highly metastatic only if they bear the activated receptor (6). M21 melanoma cells can arrest during blood flow through _v3-mediated interaction with platelets (5). This indicates that at least a subpopulation of M21 cells expresses _v3 in an activated state, or in a state in which _v3 becomes activated during blood flow without other exogenous stimuli. In the buffer perfusion system, melanoma cell _v3 had to be activated exogenously, here done with Mn²⁺, to support melanoma cell arrest at immobilized matrices. This allowed us to demonstrate that _v3 activation was required for support of melanoma cell arrest during flow, whereas activation of other integrins through Mn²⁺, as seen by accelerated adhesion during stasis, was not sufficient to contribute to matrix interaction during flow. The fact that the buffer perfusion system provides a reliable readout for the activation state of _v3 makes this experimental approach useful to analyze factors that regulate or permit tumor cell integrin _v3 activation. The consequences of _v3 activation determine the adhesive properties of tumor cells in the complex situation of blood flow and promote metastasis (6). The arrest-competent state of integrin adhesion receptors is regulated by integrin affinity, receptor diffusion within the plasma membrane, and linkage to the cytoskeleton. The latter promote integrin clustering and control receptor avidity (11, 12). Integrin-mediated cell-matrix contact must be followed immediately by post-ligand strengthening to sustain cell adhesion in the presence of flow-dependent shear forces. Our results show that _v3, but none of the other melanoma cell integrins that recognize the tested ligands, has these specific qualities. Melanoma cells that came in contact with any of the tested matrices during flow arrested and began to spread promptly. This was accompanied by redistribution of _v3 to the perimeter of the cell contact surface and colocalization of the adhesion receptor with filamentous actin, whereas inhibition of actin polymerization prevented melanoma cell arrest. Instant rearrangement of _v3 molecules within the membrane likely involves disassembly of actin filaments, as shown for _v3-mediated leukocyte adhesion (38), as well as assembly of actin filaments and engagement of adaptor proteins that link the integrin to the cytoskeleton (39, 40). Interestingly, it was recently shown that activated _v3 is selectively recruited to the leading edge of migratory cells (37). Together with our finding that activated but not non-activated _v3 supports tumor cell arrest under flow, this indicates that the activated receptor has specific qualities that support dynamic matrix interactions and mechanisms that guide the receptor to critical sites of cell-matrix contact.

A hallmark of integrin activation is the ability to bind soluble ligand. We showed that activated melanoma cell integrin _v3 binds soluble fibrinogen, and this resulted in fibrinogen-dependent cell-cell cohesion, reminiscent to that of platelet cohesion in thrombus formation. In accordance with our previous finding that fibrinogen or other multivalent plasma protein ligands of _v3 support tumor cell cohesion with matrix-attached platelets during blood flow (5), we observed in the buffer perfusion system that soluble fibrinogen supported recruitment of circulating melanoma cells to cells already attached to the matrix but did not promote aggregation of suspended melanoma cells. Thus, adhesion and cohesion of circulating tumor cells seems to follow a mechanism similar to that known for platelet thrombus growth (13). It therefore seems possible that circulating tumor cells undergo sequential steps of activation. 1) Initial activation may enable a critical adhesion receptor, like integrin _v3, to support tumor cell arrest at a reactive matrix or on ligand proteins immobilized at the surface of vascular cells. 2) The arrest event could trigger signals that enable the receptor to bind soluble ligand. 3) The fully activated form of the adhesion receptor may further support new or enhanced functions, such as migration and invasion, which are critical properties downstream of tumor cell arrest. In conclusion, the expression of activated integrin _v3 and/or its potential to undergo activation provides a unique advantage for circulating tumor cells by promoting tumor cell arrest in the presence of flow-dependent shear forces. This could be crucial during metastasis from the bloodstream and select tumor cells that can proceed further in the metastatic cascade.

	ACKNOWLEDGEMENT

We thank Dr. Z. M. Ruggeri of The Scripps Research Institute for providing access to the video microscopy system and for stimulating discussions.

	FOOTNOTES

^* This work was supported by Grants DAMD 17-99-1-9368 from the United States Army Department of Defense, 5JB-0143 from the California Breast Cancer Research Program, and RO1 CA95458 from the National Institutes of Health (to B. F. H) and by fellowship PI 402 from the Deutsche Forschungsgemeinschaft (to J. P.). This is manuscript 14779-MEM of The Scripps Research Institute.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The on-line version of this article (available at www.jbc.org) contains the video demonstrations for Figs. 1, 2, and 8.

To whom correspondence should be addressed: Dept. of Molecular and Experimental Medicine, The Scripps Research Institute, 10550 North Torrey Pines Rd., Mail Drop MEM 175, La Jolla, CA 92037. Tel.: 858-784-2021; Fax: 858-784-2030; E-mail: brunie@scripps.edu.

Published, JBC Papers in Press, April 4, 2002, DOI 10.1074/jbc.M201630200

	ABBREVIATIONS

The abbreviations used are: PBS, phosphate-buffered saline; BSA, bovine serum albumin; mAb, monoclonal antibody.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES