Regulation of Tumor Cell Chemotaxis by Type IV Collagen Is Mediated by a Ca2+-dependent Mechanism Requiring CD47 and the Integrin αVβ3

Studies from our laboratories demonstrated that synthetic peptides from the non-collagenous (NC-1) domain of the α3 (IV) chain of type IV collagen (COL IV) enhanced tumor cell adhesion (Han, J., Ohno, N., Monboisse, J. C., Pasco, S., Borel, J. P., and Kefalides, N. A. (1997) J. Biol. Chem. 272, 20395–20401). We have isolated the receptors for the α3(IV)185–203 peptide from melanoma and prostate tumor cells and identified them as CD47/integrin-associated protein and the integrin αVβ3 (Shahan, T. A., Ziaie, Z., Pasco, S., Fawzi, A., Bellon, G., Monboisse, J. C., and Kefalides, N. A. (1999) Cancer Res. 59, 4584–4590). In the present study we have examined the effect of CD47 and the integrin αVβ3 on in vitro tumor cell chemotaxis and Ca2+ i modulation in response to COL IV, from the anterior lens capsule (ALC-COL IV) and peptides from its NC-1 domain. COL IV as well as the α3(IV) peptide promoted tumor cell chemotaxis with an immediate increase in intracellular [Ca2+]. Treating tumor cells with CD47 and integrin αVβ3-reactive antibodies reduced chemotaxis as well as the rise in [Ca2+] i in response to ALC-COL IV or the α3(IV)185–203 peptide but not to Engelbreth-Holm-Swarm-COL IV or fibronectin. The α3(IV)185–203 synthetic peptide stimulated an increase in calcium from intracellular stores exclusively, whereas ALC-COL IV, Engelbreth-Holm-Swarm-COL IV, and fibronectin stimulated Ca2+ flux from both internal and external stores. Furthermore, treatment of the cells with Ca2+ chelator bis-(O-aminophenoxyl)ethane-N,N,N′,N′-tetraaceticacid-acetomethoxy ester inhibited chemotaxis toward both ALC-COL IV and the α3(IV)185–203 peptide. These data indicate that CD47 and integrin αVβ3 regulate tumor cell chemotaxis in response to COL IV and the α3(IV)185–203 peptide through a Ca2+-dependent mechanism.

Transmigration through the vessel wall by tumor cells requires adhesion and penetration of the basement membrane (BM), 1 followed by movement in or out of the vessel lumen. These processes are mediated through specific adhesion molecules on tumor cells and endothelial cells (1,2). Integrins play an important role in cellular adhesion and motility by allowing cellular adhesion and activation of signal transduction pathways (3,4).
Normal COL IV is a triple-helical molecule formed by the interaction of any two of six different ␣-chains (␣1-␣6) (5)(6)(7). We have shown that synthetic peptides comprising residues 185-203 of the NC-1 domain of the ␣3 chain of COL IV from anterior lens capsule (ALC), enhance melanoma cell adhesion by 50 -60% over controls. On the other hand, peptides from a similar region of the other chains (i.e. ␣1, ␣2, ␣4, or ␣5) were less efficient in promoting adhesion. In these biological activities, the presence of the -SNS-triplet (residues 189 -191) in the above peptide appears to be an absolute requirement (8). Pretreatment of the ␣3(IV)185-203 peptide substrate with a peptide-reactive monoclonal antibody (mAb) inhibited melanoma cell attachment; however, the treatment of a substrate composed of EHS-COL IV did not inhibit attachment.
More recently we have isolated the ␣3(IV)185-203 peptidespecific receptors on melanoma and prostate tumor cells and identified them as CD47/integrin-associated protein and the ␣ V ␤ 3 integrin by affinity chromatography and Western blot analysis (9). COL IV from the Engelbreth-Holm-Swarm (EHS) mouse tumor, which contains only the ␣1 and ␣2 chains of COL IV (10) has been previously demonstrated to induce melanoma cell chemotaxis (11). The role of CD47 in tumor cell chemotaxis has never been explored; however, Cooper et al. (12) and Parkos et al. (13) demonstrated the requirement of CD47 for normal trans-endothelial and trans-epithelial migration of PMN. Work by Parkos et al. (13) showed that PMN, pretreated with CD47reactive antibodies at levels above 2 g/ml, were inhibited from trans-endothelial migration in response to n-formylmethionylleucylphenylalanine. They concluded that the inhibition was due to a yet unknown post-adhesive event following binding to the ␤ 2 integrin subunit.
The signal transduction pathways by which CD47 and inte- 1 The abbreviations used are: BM, basement membrane; ALC-COL IV, type IV collagen from anterior lens capsule; BAPTA-AM, bis-(Oaminophenoxyl)ethane-N,N,NЈ,NЈ-tetraacetic acid-acetomethoxy ester; IAP, integrin-associated protein; COL IV, type IV collagen; EHS, Engelbreth-Holm-Swarm tumor; FBS, fetal bovine serum; HBSS, Hanks' balanced salt solution; i, intracellular; mAb, monoclonal antibody; NC-1, noncollagenous domain; HPF, high power field; PMN, polymorphonuclear leukocyte. grin ␣ V ␤ 3 regulate chemotaxis are mostly unknown. Schwartz et al. (14) first described the role of CD47 on Ca 2ϩ flux in endothelial cells. Based on the primary sequence homology of CD47 with known Ca 2ϩ channel proteins, they hypothesized that it may play a role in Ca 2ϩ modulation. Their data showed that CD47 is specifically required for integrin-mediated Ca 2ϩ flux in endothelial cells (14). The role of Ca 2ϩ modulation in the alteration of the cell cytoskeleton to accommodate cell movement for spreading and motility has been demonstrated (15,16). However, the effect of CD47 on the regulation of Ca 2ϩ flux and motility in tumor cells has not been previously reported.
The primary aim of this study was to examine the influence of CD47 and integrin ␣ V ␤ 3 on tumor cell chemotaxis in response to COL IV and the NC-1 domain of its ␣3 chain. The data do indicate that both CD47 and integrin ␣ V ␤ 3 influence tumor cell chemotaxis in response to COL IV and to synthetic peptides from the ␣3(IV) chain. Furthermore, we show that the ␣3(IV)185-203 peptide causes an immediate rise in [Ca 2ϩ ] i that was inhibited by pretreatment of tumor cells with CD47 or integrin ␣ V ␤ 3 -reactive mAbs. The Ca 2ϩ chelator BAPTA-AM inhibited chemotaxis toward COL IV, indicating an association between the receptors, cell motility, and Ca 2ϩ flux in response to COL IV and its peptides.
Type IV Collagen Isolation from Bovine Anterior Lens Capsule and Preparation of Its Synthetic Peptides-COL IV was extracted from bovine ALC according to the protocol of Brinker et al. (24). EHS-COL IV, which contains only the ␣1 and ␣2 chains, was a gift from Dr. Hynda Kleinman, National Institutes of Health (25). Synthetic peptides corresponding to the primary sequence regions of the NC-1 domain of human COL IV ␣1 and ␣3 chains (26) were synthesized and purified at the Protein Core Facility at the University of Pennsylvania (Philadelphia, PA) essentially according to the method of Barany and Merrifield (27). One of the peptides, ␣3(IV)L5 1-15, was derived from an alternatively spliced human ␣3(IV) collagen clone missing exon IV as described previously (28). The peptides were solubilized in medium without FBS (overnight at 4°C) or with 10 l of dimethyl sulfoxide and diluted to the appropriate concentration in medium with FBS (2%). Vehicle control consisted of only medium into which peptides were diluted. Primary sequences of synthetic peptides used in this study are listed in Table I.
Chemotaxis Analysis-Tumor cell chemotaxis toward COL IV or the ␣3(IV)185-203 peptide was measured using multiple 12-well chemotaxis chambers from Neuroprobe Inc. (Cabin John, MD) (29). Cells were detached from culture by incubation with Sigma cell dissociation medium and washed three times with Hanks' balanced salt solution (HBSS) with 0.3% bovine serum albumin and incubated for 1 h (37°C at 5% CO 2 ) in the same. In some cases, cells were pretreated for 30 min with integrin-reactive mAbs and washed (twice) with HBSS prior to their addition to the upper well. In other studies, cells were pretreated with BAPTA-AM (145 M) for 25 min at 37°C or dotarazine (10 M) or nifedipine (50 M), Ca 2ϩ channel modulators, for 60 min before use. Chemoattractants (158 Ϯ 1 l of various concentrations diluted in medium with 1% FBS) were loaded into the lower compartment, and 110 Ϯ 1 l of cell suspension (1.5 ϫ 10 5 cells/ml) was added to the upper compartment. The two compartments were separated by a polyvinylfluoride-free polycarbonate filter with 8-m pores (Poretics, Livermore, CA) precoated with a 0.05% gelatin solution (15 min, room temperature). The apparatus was incubated at 37°C in 5% CO 2 for 60 min. Filters were then scraped, fixed, stained using a Leukostat kit (Fisher Scientific Inc.), and mounted on glass microscope slides according to manufacturer's instructions. Motility was analyzed by manually counting 6 high power oil-immersion fields using an Olympus microscope (model BH-2) equipped with a Toshiba CCD color video camera (model IXTU40a, Laser and Motion Development, Union City, CA) connected to a PC running Image tool software for Windows (University of Texas, version 1.28).

Measurement of Intracellular [Ca 2ϩ ] by Fura-2 in Melanoma
Cells -To determine if COL IV and synthetic peptides stimulate a rise in [Ca 2ϩ ] i , fura-2 fluorescence was measured in melanoma cells. Cells were removed from adherence by incubation with cell dissociation media, washed twice in HBSS without phenol red, and placed in the same appended with 0.1% bovine serum albumin. Fura-2 acetoxymethyl ester was solubilized in Me 2 SO (99%) and added to the cell suspension (1.67 ϫ 10 6 /ml) at a final concentration of 25 M and incubated for 15 min at 37°C, washed and resuspended in Ca 2ϩ -free HBSS containing 1 mM EGTA, and used immediately. For extracellular Ca ϩ2 depletion studies, cells were incubated in Ca 2ϩ -free HBSS containing 1 mM EGTA for 25 min. For intracellular depletion studies, cells were incubated in Ca 2ϩ free HBSS containing BAPTA-AM (145 M) for 20 min. In these experiments, COL IV or its synthetic peptides were added to cells alone or to cells that had been previously treated with integrin-reactive mAbs (30 min) and washed (twice) prior to their addition.
Fura-2 fluorescence was measured using a Photon Technologies Inc. (South Brunswick, NJ) fluorescent spectrophotometer (model MP-2) equipped with a mercury arc lamp and a Windows-based PC running PTI Alpha Scan software (version 2.050). Incubation temperature was automatically maintained at 37°C. Software was set to allow alternating excitement wavelengths at 340 and 380 nm every 2 s while emission spectra passed through a 510-nm filter before measurement with a photon detector. Uptake of Fura-2 AM among cells was uniform, as determined by fluorescence microscopy immediately after loading and between 1 and 2 h after loading (30).
Calibration of the Fura-2 Signal in Fura-2-loaded Melanoma Cells-Melanoma cells were washed twice and resuspended in Ca 2ϩ -and Mg 2ϩ -free HBSS with 25 M ionomycin and 7 mM EGTA. To the cell suspension, an equal volume of HBSS containing 5 mM EGTA, 1.0 mM Ca 2ϩ , and 20 M ionomycin was added. Free [Ca 2ϩ ] i concentrations were calculated using the equation of Grynkiewicz (30). Data were then expressed as nanomolar Ca 2ϩ .
Statistical Analysis-Data were analyzed using Sigmaplot (Jandel Scientific Software, San Rafael, CA) and were expressed as mean Ϯ S.E. To evaluate the effect of the receptors on agent-mediated chemotaxis, comparisons were performed using Student's t test. Data were considered statistically significant if p values were Ͻ0.05 (31).

The Effect of COL IV and Its Synthetic Peptides on Tumor
Cell Chemotaxis-Tumor cell chemotaxis in response to COL IV and synthetic peptides was studied using a modified Boyden chamber. The melanoma cell line W-164 migrated toward ALC-COL IV and the bioactive ␣3(IV)185-203 synthetic peptide (which contains the triplet -SNS-) in a concentration-dependent manner (Fig. 1). On the other hand, the ␣3(IV)190 -203 peptide (which contains -NS-, rather than the -SNS-sequence) failed to induce chemotaxis; similarly, cells responded poorly to the ␣1(IV)185-203 peptide, which has the -ANS-sequence, even at 1000 g/ml. Chemotaxis of melanoma cells toward ALC-COL IV peaked at approximately 10 g/ml at 533 Ϯ 41 cells/HPF, toward EHS-COL IV peaked at 1000 g/ml with 332 Ϯ 31 cells/HPF and toward the ␣3(IV)185-203 synthetic peptide peaked at 10 g/ml with 373 Ϯ 30 cells/HPF (Fig. 1). Motility toward fibronectin peaked at 100 g/ml with 213 Ϯ 24 cells/HPF for fibronectin (data not shown). It is evident that chemotaxis of melanoma cells toward ALC-COL IV and its ␣3(IV) peptides was more efficient as compared with EHS-COL IV or synthetic peptides lacking the -SNS-triplet.

The Effect of Calcium Modulation on Melanoma Cell Chemotaxis in Response to COL IV and Its Synthetic Peptides-To
Calcium Mobilization in Response to ALC-, EHS-COL IV, and Synthetic Peptides in Melanoma Cells-We have shown that Ca 2ϩ channel modulators as well as the Ca 2ϩ chelator BAPTA-AM influence chemotaxis toward COL IV and its synthetic peptides (Fig. 4) it allowed [Ca 2ϩ ] i to peak at 797 Ϯ 57 nM in response to ALC-COL IV and 1127 Ϯ 107 nM in response to EHS-COL IV (Fig. 5B). Treatment of the cells with the CD47 mAb at levels above 10 g/ml inhibited Ca 2ϩ i mobilization in response to both EHS-COL IV or fibronectin (data not shown), indicating that high levels of the CD47-reactive mAb may inhibit Ca 2ϩ i modulation in a nonspecific manner. Comparison of these data to studies without antibodies was made using the data presented in Fig. 5A. As an additional control, the cells were treated with either integrin ␤ 1 or ␤ 2 subunit-reactive mAbs. These mAbs had little effect on [Ca 2ϩ ] i at the concentration used. Treatment of the cells with the antibodies alone had no effect on resting [Ca 2ϩ ] i (data not shown). These data indicate that the ␣3(IV)185-203 peptide initiates Ca 2ϩ mobilization by virtue of the -SNS-at residues 189 -191; they also demonstrate that both CD47 and the ␣ V ␤ 3 integrin regulate Ca 2ϩ i mobilization in response to the ␣3(IV)185-203 peptide, ALC-COL IV, and EHS-COL IV.
To determine whether the ␣3(IV)185-203 peptide stimulated Ca 2ϩ flux was from internal or external sources, specific channel modulators were used. The calcium channel modulator dotarazine (Fig. 5B) had no effect on Ca 2ϩ mobilization, with (Fig. 5B) or without (data not shown) external Ca 2ϩ on melanoma cells stimulated with the ␣3(IV)185-203 synthetic peptide, further suggesting that the primary source of Ca 2ϩ is from intracellular stores. In contrast, ALC-COL IV or EHS-COL IV failed to increase [Ca 2ϩ ] i in the presence of dotarazine (Fig.  5B). Treatment with nifedipine had no effect (data not shown) on [Ca 2ϩ ] i mobilization. As expected, treatment of cells with BAPTA-AM inhibited Ca 2ϩ i mobilization in response to all agents (data not shown).

DISCUSSION
The ability of CD47 and integrin ␣ V ␤ 3 to function as receptors mediating chemotaxis in response to the ␣3 chain of COL IV is a novel observation. Type IV collagen is found in BM and contributes to its structural and functional properties. Our studies have discovered additional biological properties of BMs that can be attributed to COL IV. Because of their critical location, underlying all endothelial and epithelial surfaces, BMs can be considered as our first line of defense. As with the enhancement of PMN chemotaxis toward BM components (32), the present studies demonstrate that COL IV and the ␣3(IV)185-203 peptide enhance tumor cell chemotaxis. The chemotactic potential of other BM components such as fibronectin, laminin, and entactin has also been demonstrated (34,35).
In previous studies, we demonstrated that the ␣3(IV)185-203 peptide bound specifically to CD47 and the ␣ V ␤ 3 integrin (9). CD47 is known to influence PMN motility; however, the mechanism of this process is unknown (12,13). In this report, we present evidence which demonstrates that these receptors are necessary for unabated tumor cell chemotaxis toward COL IV and the ␣3(IV)185-203 peptide. Treatment of these cells with either CD47 or integrin ␣ V ␤ 3 mAbs alone partially inhibited chemotaxis; however, treatment of the tumor cells with both mAbs together markedly inhibited chemotaxis toward the ␣3(IV)185-203 peptide and ALC-COL IV. On the other hand, the mAbs had little or no effect in the presence of other chemoattractants such as EHS-COL IV or fibronectin. These data indicate that CD47 and integrin ␣ V ␤ 3 are involved in chemotaxis in response to ALC-COL IV and the ␣3(IV)185-203 peptide.
CD47 has been previously shown to associate with and modulate integrin activation. Data support its association with the ␤ 1 (33), ␤ 2 (34, 35), and ␤ 3 (36 -38) integrin subunits. Previous data from our laboratory suggest that the interaction of CD47 with ␣ V ␤ 3 is necessary for the complete inhibition of tumor cell proliferation and normal tumor cell adhesion to a substrate composed of the ␣3(IV)185-203 peptide (9). In the present study, we have presented evidence that both these receptors are also necessary for normal chemotaxis toward the ␣3(IV)185-203 peptide. The same argument holds true for the rise in [Ca 2ϩ ] i in response to the above peptide.
The requirement of Ca 2ϩ flux for cell movement has already been described (15,16). Our data demonstrate that COL IV and the ␣3(IV)185-203 peptide cause an immediate, unsustained rise in [Ca 2ϩ ] i . Furthermore, treatment of melanoma cells with either CD47 or integrin ␣ V ␤ 3 -reactive mAbs alone inhibited the rise in [Ca 2ϩ ] i (Fig. 5, A and B). In comparison, treatment of melanoma cells with either CD47 or integrin ␣ V ␤ 3 -reactive mAbs alone inhibited chemotaxis partially in response to ALC-COL IV and markedly in response to the ␣3(IV)185-203 peptide but had no effect on chemotaxis toward EHS-COL IV. These data suggest that the rise in [Ca 2ϩ ] i in response to the ␣3(IV)185-203 peptide is mediated through the same receptors for the peptide as those required for chemotaxis.
Based on the predicted structure of CD47 (5 transmembrane domains) and its primary sequence homology to known channel proteins, it has been hypothesized that the receptor may function in the capacity of an ion channel or be closely associated with one (11). To further investigate the relationship among these factors, namely the peptide, the two receptors, and Ca 2ϩ modulation, we examined the source of the Ca 2ϩ flux. We found that the rise in [Ca 2ϩ ] i in response to the ␣3(IV)185-203 peptide was primarily from intracellular sources. In comparison, the rise in [Ca 2ϩ ] i in response to ALC-COL IV and EHS-COL IV was mostly from extracellular sources. Since we have demonstrated that the receptors for the ␣3(IV)185-203 peptide are CD47 and integrin ␣ V ␤ 3 , these data suggest that CD47 may not act as an ion channel in melanoma cells in response to the ␣3(IV)185-203 peptide; however, the receptors are likely linked to a signal transduction pathway that leads to the release of intracellular Ca 2ϩ stores. On the other hand, treatment of melanoma cells with the CD47 mAb (B6H12) at concentrations above 10 g/ml inhibited the rise in [Ca 2ϩ ] i in response to either EHS-COL IV and fibronectin, each of which were shown to cause a rise in [Ca 2ϩ ] i mostly from extracellular stores. This indicates that CD47 may function as a Ca 2ϩ channel in response to certain agents, but not in response to the ␣3(IV)185-203 peptide.
In this report we have demonstrated a regulatory effect of normal type IV collagen from ALC and peptides derived from the NC-1 domain of the ␣3(IV) chain on tumor cell chemotaxis in vitro. Specifically, we have shown that CD47 and the ␣ V ␤ 3 integrin are necessary for melanoma cell chemotaxis in response to the ␣3(IV)185-203 peptide. Our current findings also FIG. 5. Effect of melanoma cell treatment alone and with CD47 and/or integrin ␣ V ␤ 3 integrin-reactive mAbs on i[Ca 2؉ ] in response to ALC-COL IV and its peptides. Melanoma cells (W-164) were isolated, loaded with Fura-2, and analyzed as described under "Experimental Procedures." Cells were incubated in HBSS before the addition of agent. The arrow in each plot designates the time when agent was added in each experiment. A description for each plot is in the upper left corner. In A, cells were incubated with ALC-COL IV, EHS-COL IV, the ␣3(IV)185-203, ␣2(IV)185-203, ␣1(IV)185-203, or the ␣3(IV)190-203 peptides alone. In B, prior to the addition of the agents, the cells were first incubated (30 min, room temperature) with a mAb to CD47 or integrin ␣ V ␤ 3 or both mAbs together and washed (twice) before [Ca 2ϩ ] measurements. Agents used in this experiment were ALC-COL IV, EHS-COL IV, and the ␣3(IV)185-203 peptide (100 g/ml). Antibodies by themselves failed to increase [Ca 2ϩ ] i . Spectrophometric values were converted to nanomolar [Ca 2ϩ ] i as described by Grynkiewitz et al. (27). Representative plots are provided for each experiment. The data are representative of each experiment performed in triplicate on three separate occasions.
indicate that calcium flux is an important regulator of chemotaxis toward these ligands and suggest the presence of a Ca 2ϩsensitive signal transduction pathway through CD47 and the integrin ␣ V ␤ 3 .