Intracellular Ca2 signaling in endothelial cells by the angiogenesis ...

Am J Physiol Cell Physiol 280: C1140–C1150, 2001.

Intracellular Ca2⫹ signaling in endothelial cells by the angiogenesis inhibitors endostatin and angiostatin LIANWEI JIANG, VIVEKANAND JHA, MOHANRAJ DHANABAL, VIKAS P. SUKHATME, AND SETH L. ALPER Molecular Medicine and Renal Units and The Cancer Center, Beth Israel Deaconess Medical Center and Departments of Medicine and Cell Biology, Harvard Medical School, Boston, Massachusetts 02215 Received 22 September 2000; accepted in final form 8 December 2000

Jiang, Lianwei, Vivekanand Jha, Mohanraj Dhanabal, Vikas P. Sukhatme, and Seth L. Alper. Intracellular Ca2⫹ signaling in endothelial cells by the angiogenesis inhibitors endostatin and angiostatin. Am J Physiol Cell Physiol 280: C1140–C1150, 2001.—Intracellular signaling mechanisms by the angiogenesis inhibitors endostatin and angiostatin remain poorly understood. We have found that endostatin (2 ␮g/ml) and angiostatin (5 ␮g/ml) elicited transient, approximately threefold increases in intracellular Ca2⫹ concentration ([Ca2⫹]i). Acute exposure to angiostatin or endostatin nearly abolished subsequent endothelial [Ca2⫹]i responses to carbachol or to thapsigargin; conversely, thapsigargin attenuated the Ca2⫹ signal elicited by endostatin. The phospholipase C inhibitor U-73122 and the inositol trisphosphate (IP3) receptor inhibitor xestospongin C both inhibited endostatin-induced elevation in [Ca2⫹]i, and endostatin rapidly elevated endothelial cell IP3 levels. Pertussis toxin and SB-220025 modestly inhibited the endostatin-induced Ca2⫹ signal. Removal of extracellular Ca2⫹ inhibited the endostatin-induced rise in [Ca2⫹]i, as did a subset of Ca2⫹-entry inhibitors. Peak Ca2⫹ responses to endostatin and angiostatin in endothelial cells exceeded those in epithelial cells and were minimal in NIH/3T3 cells. Overnight pretreatment of endothelial cells with endostatin reduced the subsequent acute elevation in [Ca2⫹]i in response to vascular endothelial growth factor or to fibroblast growth factor by ⬃70%. Intracellular Ca2⫹ signaling may initiate or mediate some of the cellular actions of endostatin and angiostatin.

of new blood capillaries from preexisting blood vessels, is a complex process required for organogenesis and embryonic development (3). Angiogenesis is also required for tumors to grow beyond several millimeters in diameter (14, 37) and likely contributes to metastasis (49). An extensive network of stimulators and inhibitors regulates angiogensis. Among the stimulators are potent endothelial cell mitogens such as vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (FGF-2).

Inhibitors of angiogenesis have been embraced as antitumor agents due to their high degree of apparent specificity for endothelial cells (35, 36), the absence of desensitization or resistance to their antiangiogenic effects (2), their ability to potentiate radiation therapy (31), and their apparently low systemic toxicity (2, 31, 37). A number of angiogenesis inhibitors have been found to be particularly associated with the presence of tumors, including angiostatin (37) and endostatin (36). Angiostatins comprise the first three, four, or five kringle domains of the plasminogen molecule (37) and are generated by matrix metalloprotease activity upregulated in and secreted by tumor-infiltrating macrophages (13). Endostatin is a 20-kDa COOH-terminal fragment of collagen XVIII (36) of known crystal structure (20), thought to be generated by a two-step process involving matrix metalloproteases and elastases (47). Both angiostatin and endostatin can be found in the circulation of tumor-bearing individuals. Different angiogenesis inhibitors appear to exert their most potent effects against cancers in distinct stages of malignant transformation (1). Most published work on the biological effects of angiostatin and endostatin has focused on antitumor activity in vivo (2, 9, 31, 36, 37) and in vitro (10, 11, 27, 38). Endostatin may inhibit endothelial cell migration and endothelial cell proliferation via G1 arrest (9), and angiostatin inhibits cell migration (6). Both angiostatin and endostatin induce endothelial cell apoptosis (6, 10), but angiostatin, at least in some cells, does not inhibit DNA synthesis (6). Only recently have more acute actions of angiogenesis inhibitors been investigated, but results have differed in different endothelial cell types. Thus prolonged angiostatin treatment has been shown to inhibit activation of the mitogen-activated protein kinases ERK1 and ERK2 by FGF-2 or by VEGF in human dermal microvascular cells (39). In contrast, angiostatin pretreatment did not inhibit FGF-2-induced activation of p42 or p44 mitogen-activated protein kinase in bovine adrenal capillary endothelial cells (6). However, angiostatin itself acutely enhanced ty-

Address for reprint requests and other correspondence: S. L. Alper, Molecular Medicine and Renal Units, Beth Israel Deaconess Medical Center, 330 Brookline Ave., RW763, Boston, MA 02215 (E-mail: [email protected]).

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.

intracellular calcium; fura 2; xestospongin C; inositol trisphosphate; vascular endothelial growth factor; fibroblast growth factor 2

ANGIOGENESIS, THE FORMATION

C1140

0363-6143/01 $5.00 Copyright © 2001 the American Physiological Society

http://www.ajpcell.org

CA2⫹ SIGNALING BY ENDOSTATIN

rosine phosphorylation of p125 focal adhesion kinase in both porcine aortic endothelial cells and mouse pancreatic endothelial cells independent of Arg-Gly-Asp (RGD) recognition by integrins (6). Angiostatin and endostatin have been postulated to exert their biological effects via cell surface receptors yet to be identified. Interactions have been proposed with circulating or matrix-bound signaling proteins such as heparan sulfate proteoglycans (20) (but see Ref. 4). Little is known about second messengers that might convey immediate intracellular signals triggered by binding of angiostatin or endostatin to putative cell surface receptors. Both external ligands and physical forces elicit changes in intracellular Ca2⫹ concentration ([Ca2⫹]i) to convey information across the endothelial cell plasma membrane to intracellular effector proteins. In endothelial cells, [Ca2⫹]i is regulated by laminar and turbulent shear stress (21, 35), hypotonic swelling, hypertonic shrinkage (26), and mechanical deformation (21, 34, 35). Because [Ca2⫹]i regulates cell migration (18, 41), progression through the cell cycle (24, 25), and apoptosis (29, 44), we tested the hypothesis that angiostatin and endostatin trigger Ca2⫹ signals in endothelial cells. In this report, we show that angiostatin and endostatin both elicit Ca2⫹ transients in endothelial cells derived from both large and small vessels. These Ca2⫹ transients are of lower magnitude in undifferentiated epithelial cell lines and are minimal in a fibroblast cell line. The elevation in intracellular Ca2⫹ results both from release of inositol trisphosphate (IP3)-sensitive intracellular Ca2⫹ stores and from entry of extracellular Ca2⫹. Prolonged exposure to endostatin attenuates acute Ca2⫹ signaling in response to subsequent treatment with proangiogenic growth factors. MATERIALS AND METHODS

Materials. Xestospongin C, SB-220025, PD-98059, SKF96365 HCl, and pertussis toxin (PTX) were obtained from Calbiochem (La Jolla, CA). U-73122 was from Research Biochemicals (Natick, MA). Gadolinium trichloride (Gd3⫹), tetraethylammonium chloride, nitrendipine, neomycin, clotrimazole, EGTA, thapsigargin, and carbachol were from Sigma Chemical (St. Louis, MO). 1,2-Bis(2-aminophenoxy)ethaneN,N,N⬘,N⬘-tetraacetic acid-acetoxymethyl ester (BAPTAAM) and fura 2-AM were from Molecular Probes (Eugene, OR). Human VEGF-165 and human basic FGF-2 were from R&D Systems (Minneapolis, MN). All other reagents were of the highest available grade. Biosynthesis and purification of recombinant endostatin and angiostatin. Recombinant mouse endostatin (9, 11) was expressed in the methanotropic yeast Pichia pastoris (Invitrogen, San Diego, CA), purified by heparin-agarose and gel filtration chromatography, and characterized as previously described (9, 11). Recombinant mouse angiostatin [357 amino acids containing kringle I-IV of plasminogen, relative molecular weight (Mr) ⬃44 kDa with glycosylation] was cloned, expressed in P. pastoris, and purified on lysine-agarose chromatography as described (30). Chemical purity was confirmed by mass spectroscopy. Endostatin purified from P. pastoris in three separate inductions exhibited consistent activity in assays of cell proliferation, induction of apoptosis, and elevation of intracellu-

C1141

lar Ca2⫹. Each preparation was dialyzed against buffers made with Milli-Q reverse osmosis water and tested for endotoxin with the Limulus amebocyte lysate assay (Associates of Cape Cod, Falmouth, MA). Endotoxin concentration in two of the endostatin preparations was ⬍30 units (3–30 ng) of lipopolysaccharide (LPS) per milligram of purified protein. In a third preparation, endotoxin concentration was 10,000 units (ⱕ10 ␮g) of LPS per milligram of protein. Stock preparations of endostatin were diluted at least 500-fold in endotoxin-free diluent before use to yield final endostatin concentrations up to 2 ␮g/ml. Thus working solutions of two endostatin preparations contained ⱕ6 pg/ml of endotoxin, whereas the third preparation contained ⱕ20 ng/ml of endotoxin. However, the three endostatin preparations did not differ in their Ca2⫹-signaling properties (n ⫽ 5). Moreover, addition of 20 ng/ml of purified Escherichia coli LPS (Associates of Cape Cod) to BAEC elevated [Ca2⫹]i only 14 ⫾ 2 nM over baseline (n ⫽ 8 coverslips, each with ⬎40 single cells monitored). This represented ⬍10% of the increase in [Ca2⫹]i elicited by all preparations of endostatin. Endostatin produced from baculovirus-infected insect cells was from Calbiochem. Angiostatin was isolated from a single Pichia preparation and was diluted 200-fold from stock before use. Neutralizing antibody to mouse endostatin. Purified endostatin produced in P. pastoris was used to raise rabbit antiserum that specifically recognized endostatin as previously described (9). IgG was purified from this antiserum by protein A chromatography, stored at 3 mg/ml, and used at a final concentration of 0.3 ␮M. Cell lines. BAEC were harvested from descending thoracic aortas obtained from the local abattoir by collagenase digestion and used at passages 6–12 (gift of A. Malek) and at passages 4–10 (gift of C. Ferran). Calf pulmonary artery endothelial cells (CPAEC), COS-7 SV40 transformed African green monkey kidney cells, and HEK-293 human embryonic kidney cells were obtained from American Type Culture Collection (Rockville, MD). Human microvascular endothelial cells (HMVEC) were purchased from Clonetics (San Diego, CA). All cells were cultured in humidified incubators at 37°C in 5% CO2. HMVEC were maintained in endothelial growth medium-2 containing various growth factors suggested by the manufacturer (Clonetics). All other cell lines were maintained in DMEM (GIBCO BRL) supplemented with 4 mM L-glutamine, 10 U/ml penicillin, and 10 ␮g/ml streptomycin. Measurement of [Ca2⫹]i. Fluorescence ratio digital imaging of [Ca2⫹]i was as described (23, 24, 26, 29, 30), with modifications. Cells were grown to subconfluency on 25-mm coverslips in six-well plates (Corning-Costar, Cambridge, MA). In one experiment as noted, confluent cells were used. Coverslips were coated with 3.5 ␮g/cm2 Cell-Tak (Collaborative BioProducts, Bedford, MA) for HEK-293 and Cos-7 cells to improve adhesion. Cells on coverslips were incubated in medium containing 2 ␮M fura 2-AM for 30–40 min. The coverslips were then washed three times in modified Hanks’ balanced salt solution (HBSS), pH 7.4, containing (in mM) 127 NaCl, 5.6 KCl, 1.27 CaCl2, 1.0 MgCl2, 5.6 glucose, and 11.6 HEPES. Washed coverslips were mounted in a modified chamber positioned on the platform of an Olympus IMT-2 inverted microscope. To conserve recombinant antiangiogenic agents, in most experiments exchange of the 1-ml chamber volume was achieved by suction through an 18G 11⁄2-in. needle affixed to the chamber’s bottom edge, with manual volume replacement within 5 s by pipetting from above. The manual bath exchange procedure (and its associ-

C1142


ated mechanical stimulus to the cells) was identical for medium containing or lacking the antiangiogenic agents. Image acquisition began within 10–15 s after solution change. In some experiments, BAEC were perfused with medium without or with endostatin at a constant rate of ⬃10 ml/min, with chamber volume set at 0.3 ml. [Ca2⫹]i was measured in room air at room temperature by ratio imaging with the Image-1 digital ratio imaging system (Universal Imaging, Westchester, PA) equipped with a Dage-MTI CCD7 camera and a Geniisys image intensifier system. Fura 2 fluorescence images were acquired at 510-nm emission with alternate excitation at 340 and 380 nm at programmed intervals. F340/F380 ratio images of individual cells calculated on a pixel-by-pixel basis were recorded to optical disks on a Pinnacle REO-650 optical disk drive for later data processing. The region of interest for each cell encompassed the entire cell projection area, including the nucleus. The ratio values were calibrated in vitro to free Ca2⫹ concentrations ranging from 0 to 39.8 ␮M (Ca2⫹ calibration buffer kit no. 2; Molecular Probes) using the same imaging parameters. Dissociation constant (Kd) was determined by fitting the experimental R value at various free Ca2⫹ concentrations using the equation [Ca2⫹]free ⫽ Kd(Sf2/Sb2)[(R ⫺ Rmin)/(Rmax ⫺ R)], where the factor Sf2/Sb2 corrects for fura 2 ion sensitivity at 380 nm. A similar procedure was employed for in situ calibration, in which 2 ␮M of 4-Br-A23187 (Molecular Probes), a nonfluorescent Ca2⫹ ionophore, was used to

collapse Ca2⫹ gradients during ⬃15 min of incubation of fura 2-loaded BAEC in a series of Ca2⫹-EGTA buffers of free Ca2⫹ concentrations ranging from 36 to 1,270 nM. Kd in these conditions was 224 nM. The calculated values of resting [Ca2⫹]i determined by in vitro and in situ calibration differed only slightly, and experimental [Ca2⫹]i values were calculated from the in vitro calibration. Data from replayed images were sampled from 9 to 17 individual cells on each coverslip. In Figs. 3–5, data were sampled from a single contoured region of interest encompassing 9 to 17 cells on each coverslip. Mean values of peak [Ca2⫹]i measured in individual cells differed ⬍2.5% from values measured in grouped cells. Resting [Ca2⫹]i values differed ⬍5% by the two methods. Concentration-response data in Fig. 1D were fit to a sigmoidal function (r2 ⫽ 0.99) using Microsoft Excel 98. Fit of the same data to a hyperbolic function (not shown, r2 ⫽ 0.97) did not differ statistically by F test (P ⫽ 0.19). Measurements in Ca2⫹-free EGTA buffer and in BAPTAAM-loaded cells. Measurements in Ca2⫹-free conditions were carried out in modified HBSS supplemented with 2 mM EGTA and lacking the standard 1.27 mM CaCl2. For BAPTA loading, cells were incubated in medium containing 30 ␮M BAPTA-AM for 40 min and were then washed in HBSS. In both conditions, resting [Ca2⫹]i was allowed to stabilize before initiation of data acquisition.

Fig. 1. Endostatin and angiostatin elicit Ca2⫹ transients in bovine aortic endothelial cells (BAEC). A: Ca2⫹ transients in fura 2-AM-loaded BAEC subjected to bath exchanges in the absence (unlabeled arrows) or presence of 2 ␮g/ml endostatin (Endo). B: Ca2⫹ transients in fura 2-AM-loaded BAEC subjected to bath exchanges in the absence (unlabeled arrows) or presence of 5 ␮g/ml angiostatin (ANGI). C: BAEC first exposed to endostatin retain their response to angiostatin while sustaining desensitization to the mechanical stimulus of simple bath exchange. D: magnitude of peak BAEC Ca2⫹ transients in response to increasing concentrations of endostatin, with K ⁄ value of 0.77 ⫾ 0.18 ␮g/ml (n ⫽ 3–4 coverslips; 9 for 2 ␮g/ml), each with 15–30 cells analyzed. Standard deviations ranged from 13 to 23% of the mean at each endostatin concentration. [Ca2⫹]i, intracellular Ca2⫹ concentration. 12


Measurement of IP3. BAEC were grown to ⬃70% confluence in 12-well plates. Quiescent cells were extracted with perchloracetic acid (PCA). Cells were extracted again after five manual washes in HBSS, a protocol that desensitizes the cells to mechanical [Ca2⫹]i transients (Fig. 1). Cells were then exposed to HBSS containing 2 ␮g/ml of endostatin and subsequently PCA-extracted at the times indicated. IP3 concentrations in the BAEC PCA extracts were measured by a radioligand binding displacement assay kit (Biotrak; Amersham Pharmacia Biotech) according to the manufacturer’s instructions. PCA-soluble protein concentration was measured by the bicinchoninic acid method (Pierce). RESULTS

Desensitization of endothelial [Ca2⫹]i transients triggered by solution change. Endothelial cells are sensitive to mechanical forces, including laminar shear stress (5, 28), hypotonic shock (26), and hypertonic shock (26, 29). Among the cellular responses to these stresses is transient elevation of [Ca2⫹]i. Stretch-activated nonspecific cation channels, as well as receptoroperated and/or store-operated Ca2⫹ channels, may be involved in such Ca2⫹ transients (21, 34, 35). Although peak [Ca2⫹]i and d[Ca2⫹]i/dt may differ in response to different stimuli, the triggered increases in [Ca2⫹]i usually return quickly to resting values. Similarly, we noted that simple manual replacement of the extracellular medium bathing BAEC with an identical volume of the same solution triggered substantial but transient elevations in [Ca2⫹]i (unlabeled

C1143

arrows, Figs. 1 and 2). Of 255 BAEC monitored on 17 coverslips, 82 ⫾ 9% exhibited these Ca2⫹ transients, which desensitized after three or four consecutive 1-ml solution changes, as shown in Fig. 1A. Intracellular Ca2⫹ transients triggered in CPAEC and in HMVEC by replacement of extracellular medium exhibited similar desensitization patterns and were abolished in cells preloaded with BAPTA-AM (not shown). In all experiments to follow, antiangiogenic agents were added to cells only after desensitization of the mechanical stimulus-evoked [Ca2⫹]i response by at least five solution changes in the absence of drug. Beyond Fig. 1, only one or two of these ⱖ5 desensitizing solution changes before drug addition are shown. [Ca2⫹]i transients in BAEC induced by endostatin and angiostatin. After desensitization of subconfluent BAEC to the mechanical stimulus of manual bath exchange, the extracellular medium was replaced with buffer containing 2 ␮g/ml endostatin (100 nM) or 5 ␮g/ml angiostatin (120 nM). Within ⬃18–30 s, angiostatin and endostatin both elicited rapidly rising but transient elevations in [Ca2⫹]i (Fig. 1, A–C). Of 585 BAEC on 39 coverslips, 71 ⫾ 19% exhibited endostatinelicited elevations in [Ca2⫹]i, and a similar proportion of BAEC were responsive to angiostatin. Two micrograms per milliliter of endostatin elevated [Ca2⫹]i in subconfluent BAEC by 179 ⫾ 41 nM (mean ⫾ SD for first endostatin exposures on 9 coverslips, each with 12 endostatin-responsive cells imaged) and in confluent

Fig. 2. Endostatin and angiostatin deplete intracellular Ca2⫹ stores but do not depend entirely on these stores for Ca2⫹ signaling. A: after desensitization to bath exchange, BAEC exposed to 2 ␮g/ml endostatin exhibited minimal response to 100 ␮M carbachol (CRB) or to subsequently added 10 ␮M thapsigargin (TG). B: if first exposed to CRB and to TG, BAEC retained responsiveness to endostatin. C: in the continued presence of 5 ␮M TG, cells exhibited significant but diminished response to endostatin. D: after desensitization to bath exchange, BAEC first exposed to 2 ␮g/ml angiostatin exhibited minimal response to subsequently added 100 ␮M CRB or 10 ␮M TG. E: in contrast, prior exposure to CRB and TG did not abolish BAEC responsiveness to angiostatin. F: in the continued presence of 5 ␮M TG, cells exhibited significant but diminished response to angiostatin.

C1144


cells by 173 ⫾ 14 nM (n ⫽ 4 coverslips, each with 30 responsive cells imaged). Similarly, 5 ␮g/ml of angiostatin elevated [Ca2⫹]i by 165 ⫾ 46 nM (n ⫽ 7). After return of [Ca2⫹]i to near resting level, the cellular [Ca2⫹]i response to solution change remained desensitized, whereas repeat exposure to endostatin or angiostatin again elicited Ca2⫹ transients (albeit sometimes slightly blunted; Fig. 1, B and C). Endostatin elicited increases in [Ca2⫹]i in a dosedependent manner, starting at concentrations as low as 0.1 ␮g/ml (5 nM, Fig. 1D) and exhibited a K ⁄ value of 0.77 ⫾ 0.18 ␮g/ml. Endostatin (2 ␮g/ml) produced by insect cells infected with recombinant baculovirus (Calbiochem) produced comparable elevation in [Ca2⫹]i. Acute [Ca2⫹]i increases in response to endostatin and angiostatin were absent in cells preloaded with BAPTA-AM (not shown). To rule out the possibility that the Ca2⫹ signals might be artifactual results of the manual solution change procedure, the response of BAEC to endostatin (2 ␮g/ml) was also tested during continuous superfusion. BAEC exhibited an acute [Ca2⫹]i increase of 130 ⫾ 18 nM over baseline (n ⫽ 10 coverslips). Five minutes after resumption of endostatin-free superfusion, reintroduction of endostatin again elevated [Ca2⫹]i 83 ⫾ 6 nM over baseline (n ⫽ 2). The magnitude of this endostatin-induced [Ca2⫹]i signal was comparable in magnitude to the 146 ⫾ 22 nM [Ca2⫹]i elevation elicited by superfusion under identical conditions with the well-studied agonist, extracellular ATP (50 ␮M), but smaller than that elicited by the agonist thrombin (0.5 U/ml; 319 ⫾ 18 nM over baseline; n ⫽ 4 coverslips in each condition). Endostatin and angiostatin deplete intracellular Ca2⫹ stores responsive to carbachol and thapsigargin. BAEC acutely exposed to 2 ␮g/ml endostatin (Fig. 2A) or to 5 ␮g/ml angiostatin (Fig. 2D) no longer exhibited elevations in [Ca2⫹]i in response to 100 ␮M carbachol or to subsequently applied 10 ␮M thapsigargin. In contrast, when BAEC were first exposed to carbachol and then thapsigargin, they retained the ability to exhibit Ca2⫹ transients in response to subsequent application of endostatin (Fig. 2B) or angiostatin (Fig. 2E). When BAEC were first treated with 5 ␮M thapsigargin, leading to respective peak and steady-state [Ca2⫹]i increases of 70 ⫾ 16 nM and 24 ⫾ 9 nM above the pre-thapsigargin level (Fig. 2, E and F; n ⫽ 6), subsequent addition of 2 ␮g/ml endostatin (Fig. 2E) or 5 ␮g/ml angiostatin (Fig. 2F) in the continued presence of thapsigargin elicited further transient respective increases of 45 ⫾ 10 nM and 25 ⫾ 11 nM (n ⫽ 3), which quickly relaxed to the thapsigargin baseline. These data show that endostatin and angiostatin share the ability to deplete thapsigargin-sensitive intracellular Ca2⫹ stores normally released by muscarinic agonists. Moreover, endostatin and angiostatin retain partial ability to elevate [Ca2⫹]i in cells with Ca2⫹ stores previously depleted by thapsigargin. Role of IP3 in endostatin signaling. The above findings suggest a role for IP3 signaling in the action of endostatin on BAEC. As shown in Fig. 3, preincubation 12

Fig. 3. Xestospongin C (XsC) inhibits peak [Ca2⫹]i elicited in BAEC by 2 ␮g/ml endostatin. A: representative traces showing [Ca2⫹]i responses to endostatin added in the absence (crosses) or presence of 10 ␮M XsC in solution without (䊐) or with EGTA (Œ). Unlabeled arrow indicates bath change without drug addition. B: mean peak increases in [Ca2⫹]i (⌬[Ca2⫹]i) for experiments similar to those in A. Numbers in parentheses indicate the number of coverslips, each with 9–17 cells studied. *P ⬍ 0.01; **P ⬍ 0.001 compared with absence of XsC; 䡠 P ⬍ 0.001 compared with presence of extracellular Ca2⫹.

of BAEC with the IP3 receptor (IP3R) inhibitor xestospongin C (22) at 2 or 10 ␮M inhibited the subsequent endostatin-triggered peak [Ca2⫹]i elevation of 140 ⫾ 13 nM by 61 and 76% to levels of 55 ⫾ 13 and 33 ⫾ 12 nM, respectively. In the absence of extracellular Ca2⫹, 10 ␮M xestospongin C inhibited the peak endostatin response by 95%. When xestospongin C was added to BAEC simultaneously with endostatin, xestospongin C was without effect. However, xestospongin C preincubations of 5–8 min (as in Fig. 3) were maximally effective. The effect of a 40-min preincubation with xestospongin C was reversed within 5 min following xestospongin C removal from the medium (not shown). Thus endostatin Ca2⫹ signaling likely required IP3R-regulated intracellular Ca2⫹ stores. IP3 generation requires activation of phospholipase C (PLC). As shown in Fig. 4A, brief preincubation with the PLC inhibitor U-73122 (10 ␮M) inhibited the peak [Ca2⫹]i response to endostatin by 75%, from 182 ⫾ 9


C1145

Fig. 4. A: U-73112 inhibits peak [Ca2⫹]i elevation elicited in BAEC by 2 ␮g/ml endostatin. Numbers in parentheses indicate number of coverslips. B: endostatin transiently increases inositol trisphosphate (IP3) levels in BAEC. Quiescent IP3 levels (}) were elevated (F at time 0) by the 5⫻ wash protocol that desensitized the solution change-induced elevation in [Ca2⫹]i (Fig. 1). Addition of endostatin (2 ␮g/ml) led to further elevation of IP3 levels (F), peaking at 20–30 s, with subsequent return to basal quiescent levels at 60–120 s. *P ⬍ 0.05; **P ⬍ 0.01 compared with 0 s. The [Ca2⫹]i data (E) are replotted here from Fig. 3A to facilitate comparison.

nM (n ⫽ 5) to 46 ⫾ 12 nM (n ⫽ 3, P ⬍ 0.01). U-73122 (1 ␮M) inhibited the peak [Ca2⫹]i response 57% to 78 nM. Addition of 10 ␮M U-73112 simultaneously with endostatin inhibited the [Ca2⫹]i signal by 50% to 92 ⫾ 13 nM (not shown). As suggested by the inhibitory effects of U-73112 and xestospongin C, endostatin itself rapidly increased IP3 levels in BAEC (Fig. 4B). Resting cells had IP3 content of 32 pmol/mg PCA-soluble protein (n ⫽ 2), which increased to 95 ⫾ 16 pmol/mg (n ⫽ 6) after the fivewash densitization protocol. Subsequent addition of endostatin (2 ␮g/ml) further increased IP3 levels to 135 ⫾ 15 pmol/mg after 10 s (P ⬍ 0.05) and to 174 ⫾ 14

pmol/mg after 20 s (P ⬍ 0.01). This elevation in IP3 was sufficiently rapid to contribute to the endostatin-mediated elevation of [Ca2⫹]i. IP3 levels remained elevated at 30 s and relaxed to resting levels at 1 and 2 min. Thus endostatin elevates [Ca2⫹]i in BAEC via PLCmediated IP3 synthesis and subsequent IP3R-mediated release of intracellular Ca2⫹ stores to the cytoplasm. Roles of extracellular Ca2⫹, PTX-sensitive G protein, and protein kinases in endostatin- and angiostatinmediated elevations in [Ca2⫹]i. Endothelial cells respond to depletion of IP3-sensitive intracellular Ca2⫹ stores with activation of a plasmalemmal Ca2⫹ entry via ligand- or store-operated conductive pathways. We

C1146


Fig. 5. Dependence of peak [Ca2⫹]i elicited by 2 ␮g/ml endostatin (A) or 5 ␮g/ml angiostatin (B) on extracellular Ca2⫹ (Ca2⫹ o ) and 1 ␮M pertussis toxin (PTX)-sensitive processes. Numbers in parentheses indicate number of coverslips. *P ⬍ 0.05; **P ⬍ 0.01; ***P ⬍ 0.001 compared with response in presence of extracellular Ca2⫹ and in absence of PTX.

therefore hypothesized that endostatin-induced Ca2⫹entry pathways contribute to the observed elevation of [Ca2⫹]i in endothelial cells. As shown in Fig. 5, chelation of extracellular Ca2⫹ reduced peak [Ca2⫹]i elevation following endostatin and angiostatin exposure by 60% (n ⫽ 4) and 59% (n ⫽ 5), respectively (P ⬍ 0.01 for both conditions). Inhibition of [Ca2⫹]i signaling was further enhanced when extracellular Ca2⫹ removal was combined with inhibitors of intracellular targets. Thus when BAEC were treated with xestospongin C in Ca2⫹-free conditions, the endostatin-induced elevation in [Ca2⫹]i was nearly completely abolished (Fig. 3B). Because G proteins regulate some plasmalemmal Ca2⫹-entry pathways, a role for PTX-sensitive G proteins in endostatin Ca2⫹ signaling was tested. As shown in Fig. 5, a 1-h preincubation of BAEC with 1 ␮M PTX inhibited the endostatin- and angiostatininduced Ca2⫹ signals by 30 and 32%, respectively (n ⫽ 4, P ⬍ 0.05). When extracellular Ca2⫹ chelation was combined with PTX pretreatment of BAEC, peak [Ca2⫹]i elevation following endostatin and angiostatin exposure was inhibited by 79 and 78%, respectively (n ⫽ 4; P ⬍ 0.01 compared with bath Ca2⫹ removal alone, P ⬍ 0.001 compared with control). The mitogen-activated protein kinase pathway may regulate Ca2⫹ signaling. As shown in Fig. 6, inhibition of the endostatin-induced peak Ca2⫹ elevation by the p38 kinase inhibitor SB-220025 (2 ␮M) was modest (P ⫽ 0.05). However, at the less specific concentration of 20 ␮M, SB-220025 inhibited the peak Ca2⫹ response by 71% (n ⫽ 3, P ⬍ 0.001). In contrast, the mitogenactivated protein kinase or ERK1 inhibitor PD-98059 failed to significantly inhibit the endostatin-induced peak Ca2⫹ response at either 2 or 20 ␮M (n ⫽ 3, P ⬎ 0.1).

Effects of Ca2⫹ entry inhibitors on endostatin-induced [Ca2⫹]i elevation. Figure 7 shows that endostatin-induced [Ca2⫹]i elevation in BAEC was not significantly inhibited by the L-type Ca2⫹ channel antagonist nitrendipine (50 ␮M), by the Ca2⫹-sensing receptor agonist neomycin (500 ␮M), or by the K⫹ and Ca2⫹ channel blocker clotrimazole (10 ␮M). In contrast, the Ca2⫹/cation channel antagonists SKF-96365 (50 ␮M) and Gd3⫹ (1 mM) inhibited endostatin-induced [Ca2⫹]i elevation by 70 and 58%, respectively (P ⬍ 0.05). Reduction in the driving force for Ca2⫹ entry via membrane depolarization by the K⫹ channel blocker tetraethylammonium (2 mM) also inhibited intracellular Ca2⫹ signal amplitude by 61% (P ⬍ 0.05). Endostatin and angiostatin elevate [Ca2⫹]i preferentially in endothelial cells. Figure 8 shows that HMVEC exhibited the largest Ca2⫹ transients in response to endostatin (Fig. 8A) and angiostatin (Fig. 8B) among

Fig. 6. Effects on peak [Ca2⫹]i elicited by 2 ␮g/ml endostatin of the p38 inhibitor SB-220025 and the mitogen-activated protein kinase or ERK1 inhibitor PD-98059. Inhibitor concentrations were as indicated. Numbers in parentheses indicate number of coverslips. *P ⬍ 0.001 compared with absence of kinase inhibitor.


C1147

We therefore tested the effect of overnight exposure of BAEC to 2 ␮g/ml endostatin on the ability of both of these growth factors to elevate [Ca2⫹]i. As shown in Fig. 9, endostatin preincubation reduced peak [Ca2⫹]i induced by 30 ng/ml VEGF-165 by 65% (Fig. 9, A and C), from 141 ⫾ 20 to 50 ⫾ 8 nM (n ⫽ 5), and reduced peak [Ca2⫹]i induced by 25 ng/ml FGF-2 by 71% (Fig. 9, B and C), from 126 ⫾ 16 to 37 ⫾ 5 nM (n ⫽ 5, each decrease significant at P ⬍ 0.001). DISCUSSION

Fig. 7. Effects of various inhibitors of Ca2⫹ entry on the endostatinelicited Ca2⫹ signal. Concentrations were 1 mM gadolinium trichloride (Gd3⫹), 2 mM tetraethylammonium chloride (TEA), 50 ␮M SKF-96365, 10 ␮M clotrimazole (CLT), 50 ␮M nitrendipine, and 500 ␮M neomycin. Numbers in parentheses indicate number of coverslips. *P ⬍ 0.05 vs. absence of drug.

the cell lines tested, followed by the large vessel cell lines BAEC and CPAEC. Interestingly, angiostatin was much less active than endostatin in CPAEC. The transformed renal epithelial cell lines COS-7 and HEK-293 exhibited smaller [Ca2⫹]i responses to endostatin. The fibroblast cell line 3T3 was least responsive among tested cell lines. Endostatin attenuates endothelial Ca2⫹ signaling by VEGF-165 and FGF-2. Assessment of the functional importance of Ca2⫹ signaling by endostatin in its antagonism of the angiogenic, proliferative, and cell migratory effects of the endothelial growth factors VEGF and FGF-2 posed methodological difficulties. VEGF165 (7) and FGF-2 (16, 33) themselves both elicit elevation of endothelial cell [Ca2⫹]i. In addition, BAPTA-AM itself blocks endothelial cell migration and proliferation and promotes endothelial cell apoptosis in the presence of growth factors (M. Segal, S. L. Alper, and V. P. Sukhatme, unpublished observations).

Natural angiogenesis inhibitors such as angiostatin and endostatin have entered phase 1 clinical trials for cancer treatment (40). Additional trials are planned for other disorders in which unregulated angiogenesis is believed to contribute to pathogenesis (8). However, remarkably little remains known about the mechanisms of action of these proteolytic products of plasminogen and of the nonfibrillar collagen XVIII. We have shown here that angiostatin and endostatin acutely elevate [Ca2⫹]i in primary cultures of bovine and human endothelial cells grown on coverslips. The elevation in cytoplasmic [Ca2⫹]i arose from a combination of Ca2⫹ release from IP3-sensitive intracellular stores and from activation of an extracellular Ca2⫹entry pathway. The [Ca2⫹]i elevation appeared to be regulated by a PTX-sensitive G protein and by an SB-220025-sensitive process that may be a kinase activity. The Ca2⫹ transients were of considerably greater magnitude in endothelial than in nonendothelial cells tested. Endostatin elevated [Ca2⫹]i in BAEC at concentrations of 100 ng/ml (5 nM) and above, higher than required for inhibition of endothelial cell migration (48) (Fig. 1D). However, endostatin concentrations in this range, which have been achieved by adenovirus-mediated gene transfer in intact mice (41), can inhibit endothelial cell proliferation and induce apoptosis in vivo and in cell culture (10, 11, 13). Ca2⫹ signaling by endostatin exhibited specificity by several additional criteria. Contaminating endotoxin

Fig. 8. Cell type specificity of peak [Ca2⫹]i elicited by 2 ␮g/ml endostatin (A) and 5 ␮g/ml angiostatin (B). HMVEC, human microvascular endothelial cells; CPAE, calf pulmonary artery endothelial cells; HEK-293, human embryonic kidney cells. Numbers in parentheses indicate number of coverslips.

C1148


Fig. 9. Endostatin attenuates Ca2⫹ signaling by growth factors. Representative [Ca2⫹]i responses in fura 2-loaded BAEC preincubated for 20 h in the absence (top traces) or presence (bottom traces; arrow) of 2 ␮g/ml endostatin. They were then acutely exposed to 30 ng/ml human vascular endothelial growth factor (VEGF)-165 (A) or to 25 ng/ml human fibroblast growth factor 2 (FGF-2; B). C: mean peak [Ca2⫹]i values (numbers in parentheses indicate number of similar experiments). *P ⬍ 0.001 compared with absence of endostatin.

was not responsible for the elevation in [Ca2⫹]i. In most preparations, endotoxin was present at 6 pg/ml, and purified LPS at this concentration is inactive. One less pure preparation contained 20 ng/ml endotoxin, but nonetheless exhibited indistinguishable Ca2⫹-signaling properties. Consistent with this observation and with earlier studies (46), 20 pg/ml of purified LPS elicited a Ca2⫹ signal ⬍10% the magnitude of that elicited by endostatin. In addition, Ca2⫹ transients were higher in endotoxin-resistant HMVEC than in endotoxin-sensitive CPAEC. Moreover, endostatin produced in insect cells (Calbiochem) and in P. pastoris elicited comparable elevations of [Ca2⫹]i in BAEC. Specificity of the endostatin response was also demonstrated immunologically. Preincubation of endostatin with neutralizing antibody reduced the endostatininduced increase in [Ca2⫹]i from 179 ⫾ 41 nM (n ⫽ 9) to 71 ⫾ 19 nM (n ⫽ 6, P ⬍ 0.01). The antibodyinsensitive component of the endostatin-induced [Ca2⫹]i increase approximated that elicited by antibody alone in BAEC (93 ⫾ 8 nM), but not in HEK-293 T cells (not shown), possibly reflecting activation of endothelial CD32/Fc-␥-RIIa (17). Although the angiogenic pathways activated by VEGF and FGF-2 differ in integrin specificity (15), both VEGF and FGF-2 bind to receptors capable of elevating [Ca2⫹]i. Endostatin pretreatment of BAEC for 18 h reduced by ⬃70% the magnitude of acute Ca2⫹ signals elicited subsequently by VEGF-165 and FGF-2. The potential role of endostatin- and angiostatin-induced acute elevations of [Ca2⫹]i in the antagonism of

longer-term effects of endothelial growth factors such as VEGF and FGF-2 remains unknown. However, endostatin does not inhibit binding of FGF-2 (4) or of VEGF to their respective receptors (R. Ramchandran, B. Knebelmann, and V. P. Sukhatme, unpublished data). All three homomeric receptors for FGF-2 (and putative heteromeric receptors as well) elevate intracellular [Ca2⫹]i in response to FGF-2 (45). Disease-associated mutations in the FGF receptor FGFR3 abolish or alter FGF-2-induced Ca2⫹ signaling (33). Among the three defined VEGF receptors (VEGFR1/flt1, VEGF-R2/KDR/flk1, and VEGF-R3/flt4) and the two coreceptors neuropilins 1 and 2, only VEGF-R2 (KDR/flk1) has been reported to elevate endothelial [Ca2⫹]i in response to VEGF (7). KDR-mediated elevation of [Ca2⫹]i has been attributed to IP3 generation via src-dependent activation of PLC␥1. The requirement of KDR for IP3-mediated release of intracellular Ca2⫹ stores may explain, at least in part, attenuation of VEGF-165-mediated Ca2⫹ signaling by long-term endostatin exposure. This antagonistic effect is consistent with the ability of xestospongin C (Fig. 3) and U-73122 (Fig. 4) to inhibit acute [Ca2⫹]i elevation by endostatin and consistent with the ability of acute endostatin treatment to deplete carbachol- and thapsigargin-sensitive Ca2⫹ stores (Fig. 2). Further studies will assess in greater detail the roles played by depletion of IP3-sensitive or other intracellular Ca2⫹ stores and/or of antagonism of Ca2⫹-entry pathways in the attenuation of acute Ca2⫹ signaling by VEGF-165 and


FGF-2 mediated by chronic endostatin exposure. These studies will be facilitated in turn by progress in defining additional intracellular signals triggered by both pro- and antiangiogenic agents. [Ca2⫹]i elevation in BAEC by extracellular endostatin and angiostatin suggests that Ca2⫹ may serve as an intracellular messenger of plasmalemmal transmembrane receptors for each antiangiogenic polypeptide. The nature of these receptors (32) has been debated,1 but intracellular signaling pathways are beginning to emerge. Angiostatin has been reported to diminish activation of ERK1 and ERK2 by bFGF and VEGF in human dermal microvascular endothelial cells, but not in bovine adrenal cortex capillary endothelial cells. Angiostatin has also been reported to activate focal adhesion kinase in several endothelial cell types by a pathway not inhibited by soluble RGD peptide (6). Endostatin undergoes endocytosis and promotes tyrosine phosphorylation of the adaptor protein Shb, with promotion of protein complex formation (12). In addition, preliminary experiments show that endostatin can rapidly mobilize from cytosol to nucleus a green fluorescent protein-tagged fusion construct of the Ca2⫹-dependent transcription factor NFAT1 (L. Jiang and S. L. Alper, unpublished observations). The finding that endostatin and angiostatin acutely elevate endothelial cell [Ca2⫹]i expands the intracellular signaling repertoire of these antiangiogenic agents and will contribute to the molecular characterization of their receptors. We thank M. Segal, B. Knebelmann, R. Ramchandran, B. Chan, S. A. Karumanchi, and A. Stuart-Tilley for helpful discussion. This work was supported by National Institutes of Health (NIH) Grants R21-CA-86207, P60-HL-15157 (Boston Sickle Cell Center), and DK-34854 (Harvard Digestive Diseases Center) (to S. L. Alper) and by NIH Grant R01-CA-81151 and seed grants from Beth Israel Deaconess Medical Center (to V. P. Sukhatme). S. L. Alper was an Established Investigator of the American Heart Association during the initiation of these studies. 1 An unusual candidate has been proposed as an angiostatin receptor, a ␤-subunit of the mitochondrial H⫹-ATPase proposed to localize to the plasma membrane (32). The 1:10 dilution of crude antiserum to E. coli mitochondrial H⫹-ATPase ␤-subunit that was shown to partially inhibit [125I]-labeled angiostatin binding to human umbilical vein endothelial cell monolayers (32) could not be tested as an antagonist of angiostatin-induced Ca2⫹ signaling. Even at 1:100 dilution, the crude antiserum itself elicited large elevations in [Ca2⫹]i in BAEC (data not shown).

REFERENCES 1. Bergers G, Javaherian K, Lo KM, Folkman J, and Hanahan D. Effects of angiogenesis inhibitors on multistage carcinogenesis in mice. Science 284: 808–812, 1999. 2. Boehm T, Folkman J, Browder T, and O’Reilly MS. Antiangiogenic therapy of experimental cancer does not induce acquired drug resistance. Nature 390: 404–407, 1997. 3. Bussolino F, Mantovani A, and Persico G. Molecular mechanisms of blood vessel formation. Trends Biochem Sci 22: 251– 256, 1997. 4. Chang Z, Choon A, and Friedl A. Endostatin binds to blood vessels in situ independent of heparan sulfate and does not compete for fibroblast growth factor-2 binding. Am J Pathol 155: 71–76, 1999.

C1149

5. Chien S, Li S, and Shyy YJ. Effects of mechanical forces on signal transduction and gene expression in endothelial cells. Hypertension 31: 162–169, 1998. 6. Claesson-Welsh L, Welsh M, Ito N, Anand-Apte B, Soker S, Zetter B, O’Reilly M, and Folkman J. Angiostatin induces endothelial cell apoptosis and activation of focal adhesion kinase independently of the integrin-binding motif RGD. Proc Natl Acad Sci USA 95: 5579–5583, 1998. 7. Cunningham SA, Tran TM, Arrate MP, Bjerke R, and Brock TA. KDR activation is crucial for VEGF-165-mediated Ca2⫹ mobilization in human umbilical vein endothelial cells. Am J Physiol Cell Physiol 276: C176–C181, 1999. 8. Dawson DW, Volpert OV, Gillis P, Crawford SE, Xu HJ, Benedict W, and Bouck NP. Pigment-epithelium-derived factor: a potent inhibitor of angiogenesis. Science 285: 245–248, 1999. 9. Dhanabal M, Ramchandran R, Volk R, Stillman IE, Lombardo M, Rruela-Arispe ML, Simons M, and Sukhatme VP. Endostatin: yeast production, mutants, and antitumor effect in renal cell carcinoma. Cancer Res 59: 189–197, 1999. 10. Dhanabal M, Ramchandran R, Waterman MJF, Lu H, Knebelmann B, Segal M, and Sukhatme VP. Endostatin induces endothelial cell apoptosis. J Biol Chem 274: 11721–11726, 1999. 11. Dhanabal M, Volk R, Ramchandran R, Simons M, and Sukhatme VP. Cloning, expression, and in vitro activity of human endostatin. Biochem Biophys Res Commun 258: 345– 352, 1999. 12. Dixelius J, Larsson H, Sasaki T, Holmquist K, Lingge L, Engstrom A, Timpl R, Welsh M, and Claesson-Welsh L. Endostatin-induced tyrosine kinase signaling through the Shb adaptor protein regulates endothelial cell apoptosis. Blood 95: 3403–3411, 2000. 13. Dong Z, Kumar R, Yang X, and Fidler IJ. Macrophagederived metalloelastase is responsible for the generation of angiostatin in Lewis lung carcinoma. Cell 88: 801–810, 1997. 14. Folkman J. Tumor angiogenesis: therapeutic implications. N Engl J Med 285: 1182–1186, 1971. 15. Friedlander M, Brooks PC, Shaffer RW, Kincaid CM, Varner JA, and Cheresh DA. Definition of two angiogenic pathways by distinct ␣v integrins. Science 270: 1500–1502, 1995. 16. Gillespie LL, Chen G, and Paterno GD. Cloning of a fibroblast growth factor receptor 1 splice variant from Xenopus embryos that lack a protein kinase C site important for the regulation of receptor activity. J Biol Chem 270: 22753–22763, 1995. 17. Groger M, Sarmay G, Fiebiger D, Wolff K, and Petzelbauer P. Dermal microvascular endothelial cells express CD32 receptors in vivo and in vitro. J Immunol 156: 1549–1556, 1996. 18. Gurubhagavatula I, Amrani Y, Pratico D, Ruberg FL, Albeida SM, and Panettieri RA. Engagement of human PECAM-1 (CD31) on human endothelial cells increases intracellular calcium ion concentration and stimulates prostacyclin release. J Clin Invest 101: 212–222, 1998. 19. He H, Venema VJ, Gu X, Venema RC, Marrero MB, and Caldwell RB. Vascular endothelial growth factor signals endothelial cell production of nitric oxide and prostacyclin through Flk1/KDR activation of c-Src. J Biol Chem 274: 25130–25135, 1999. 20. Hohenester E, Sasaki T, Olsen BR, and Timpl R. Crystal structure of the angiogenesis inhibitor endostatin at 1.5 A resolution. EMBO J 16: 1656–1664, 1998. 21. Hoyer J, Kohler R, and Distler A. Mechanosensitive Ca2⫹ oscillations and STOC activation in endothelial cells. FASEB J 12: 359–366, 1998. 22. Hu Q, Deshpande S, Irani K, and Ziegelstein RC. Cai2⫹ oscillation frequency regulates agonist-stimulated NF-␬B transcriptional activity. J Biol Chem 274: 33995–33998, 1999. 23. Husain M, Bein K, Jiang L, Alper SL, Simons M, and Rosenberg RD. C-Myb-dependent cell cycle progression and Ca2⫹ homeostasis in vascular smooth muscle cells. Circ Res 80: 617–626, 1997. 24. Husain M, Jiang L, See V, Bein K, Simons M, Alper SL, and Rosenberg RD. Regulation of vascular smooth muscle cell proliferation by plasma membrane Ca2⫹-ATPase. Am J Physiol Cell Physiol 272: C1947–C1959, 1997.

C1150


26. Jiang L, Vandorpe DH, Malek AM, and Alper SL. Hypotonicity and hypertonicity trigger distinct Ca2⫹ signalling pathways in bovine aortic endothelial cells (Abstract). J Am Soc Nephrol 8: 52A, 1997. 27. Lu H, Dhanabal M, Volk R, Waterman MJF, Ramchandran R, Knebelmann B, Segal M, and Sukhatme VP. Kringle 5 causes cell cycle arrest and apoptosis of endothelial cells. Biochem Biophys Res Commun 258: 668–673, 1999. 28. Malek AM, Alper SL, and Izumo S. Hemodynamic shear stress: an overview of its role in atherosclerosis. JAMA 282: 2035–2042, 1999. 29. Malek AM, Goss GG, Jiang L, Izumo S, and Alper SL. Mannitol at clinical concentrations activates multiple signalling pathways and induces apoptosis in endothelial cells. Stroke 29: 2631–2640, 1998. 30. Malek AM, Jiang L, Lee I, Sessa WC, Izumo S, and Alper SL. Shear stress-mediated upregulation of endothelial nitric oxide synthase mRNA requires intracellular Ca and G protein signals, and is modulated by PI-3 kinase. Biochem Biophys Res Commun 254: 231–242, 1999. 31. Mauceri HJ, Hanna NN, Beckett MA, Gorski DH, Staba MJ, Stellato KA, Bigelow K, Heimann R, Gately S, Dhanabal M, Soff GA, Sukhatme VP, Kufe DW, and Weichselbaum RR. Combined effects of angiostatin and ionizing radiation in antitumor therapy. Nature 394: 287–291, 1998. 32. Moser TL, Stack MS, Asplin I, Enghild JJ, Hojrup P, Everitt L, Humchak S, Schnaper HW, and Pizzo SV. Angiostatin binds ATP synthase on the surface of human endothelial cells. Proc Natl Acad Sci USA 96: 2811–2816, 1999. 33. Nguyen HB, Estacion M, and Gargus JJ. Mutations causing achondroplasia and thanatophoric dysplasia alter bFGF-induced calcium signals in human diploid fibroblasts. Hum Mol Genet 6: 681–688, 1997. 34. Niggel J, Sigurdson W, and Sachs F. Mechanically induced calcium movements in astrocytes, bovine aortic endothelial cells, and C6 glioma cells. J Membr Biol 174: 121–134, 2000. 35. Nilius B, Viana F, and Droogmans G. Ion channels in vascular endothelium. Annu Rev Physiol 59: 145–170, 1997. 36. O’Reilly MS, Boehm T, Shing Y, Fukai N, Vasios G, Lane WS, Flynn E, Birkhead JR, Olsen BR, and Folkman J. Endostatin: an endogenous inhibitor of angiogenesis and tumor growth. Cell 88: 277–285, 1997. 37. O’Reilly MS, Holmgren L, Shing Y, Chen C, Rosenthal RA, Moses M, Lane WS, Cao Y, Sage EH, and Folkman J. Angiostatin: a novel angiogenesis inhibitor that mediates the

38.

39.

40.

41.

42.

44.

45.

46.

47.

48.

49.

suppression of metastases by a Lewis lung carcinoma. Cell 79: 315–328, 1994. Ramchandran R, Dhanabal M, Volk R, Waterman MJF, Segal M, Lu H, Knebelmann B, and Sukhatme VP. Antiangiogenic activity of restin, NC10 domain of human collagen XV: comparison to endostatin. Biochem Biophys Res Commun 255: 735–799, 1999. Redlitz A, Daum G, and Sage EH. Angiostatin diminishes activation of the mitogen-activated protein kinases ERK-1 and ERK-2 in human dermal microvascular endothelial cells. J Vasc Res 36: 28–34, 1999. Ryan DP, Penson RT, Ahmed S, Chabner BA, and Lynch TJ. Reality testing in cancer treatment: the phase I trial of endostatin. Oncologist 4: 501–508, 1999. Sauter BV, Martinet O, Zhang WJ, Mandeli J, and Woo SLC. Adenovirus-mediated gene transfer of endostatin in vivo results in high level of transgene expression and inhibition of tumor growth and metastases. Proc Natl Acad Sci USA 97: 4802–4807, 2000. Schwab A, Wojnowski L, Gabriel K, and Oberleithner H. Oscillating activity of a Ca2⫹-sensitive K⫹ channel. A prerequisite for migration of transformed Madin-Darby canine kidney focus cells. J Clin Invest 93: 1631–1636, 1994. Suc I, Escargueil-Blanc I, Troly M, Salvayre R, and NegreSalvayre A. HDL and ApoA prevent cell death of endothelial cells induced by oxidized LDL. Arterioscler Thromb Vasc Biol 17: 2158–2166, 1997. Ueno H, Gunn M, Dell K, Tseng A, and Williams L. A truncated form of fibroblast growth factor receptor 1 inhibits signal transduction by multiple types of fibroblast growth factor receptor. J Biol Chem 267: 1470–1476, 1992. Umans J, Salvi D, Murray PT, and Wylam ME. Selectivity of endotoxin-induced defect in endothelial calcium mobilization. Kidney Int 54: 1063–1069, 1998. Wen W, Moses MA, Wiederschain D, Arbiser J, and Folkman J. The generation of endostatin is mediated by elastase. Cancer Res 59: 6052–6056, 1999. Yamaguchi N, Anand-Apte B, Lee M, Sasaki T, Fukai N, Shapiro R, Que I, Lowik C, Timpl R, and Olsen BR. Endostatin inhibits VEGF-induced endothelial cell migration and tumor growth independently of zinc binding. EMBO J 18: 4414– 4423, 1999. Zetter B. Angiogenesis and tumor metastasis. Annu Rev Med 49: 407–424, 1998.