Kringle 5 of Human Plasminogen Induces Apoptosis of ... - Hoefer Inc

Research Article

Kringle 5 of Human Plasminogen Induces Apoptosis of Endothelial and Tumor Cells through Surface-Expressed Glucose-Regulated Protein 78 1

1

1

1

Don J. Davidson, Catherine Haskell, Sandy Majest, Abdullah Kherzai, David A. Egan, 2 1 3 2 Karl A. Walter, Andrew Schneider, Earl F. Gubbins, Larry Solomon, 1 1 1 Zhebo Chen, Rick Lesniewski, and Jack Henkin

2

Departments of 1Cancer Research, 2Structural Biology, and 3Molecular Biology, Abbott Laboratories, Abbott Park, Illinois

Abstract Kringle 5 (K5) of human plasminogen has been shown to inhibit angiogenesis by inducing the apoptosis of proliferating endothelial cells. Peptide regions around the lysine-binding pocket of K5 largely mediate these effects, particularly the peptide PRKLYDY, which we show to compete with K5 for the binding to endothelial cells. The cell surface binding site for K5 that mediates these effects has not been defined previously. Here, we report that glucose-regulated protein 78, exposed on cell surfaces of proliferating endothelial cells as well as on stressed tumor cells, plays a key role in the antiangiogenic and antitumor activity of K5. We also report that recombinant K5-induced apoptosis of stressed HT1080 fibrosarcoma cells involves enhanced activity of caspase-7, consistent with the disruption of glucose-regulated protein 78-procaspase-7 complexes. These results establish recombinant K5 as an inhibitor of a stress response pathway, which leads to both endothelial and tumor cell apoptosis. (Cancer Res 2005; 65(11): 4663-72)

Introduction Angiogenesis in normal tissues is tightly regulated by a balance between angiogenesis stimulators, such as vascular endothelial growth factor (VEGF), and angiogenesis inhibitors, such as thrombospondin-1 (1–5). Tumors, such as benign neoplasias, remain dormant when the proangiogenic and antiangiogenic factors are balanced (6–9). However, when a tumor outgrows its blood supply, local hypoxia increases the production of angiogenesis stimulators and decreases the production of angiogenesis inhibitors. This proangiogenic switch releases a tumor from dormancy and drives tumor progression (10–12). Several endogenous angiogenesis inhibitors are protein fragments derived from extracellular matrix (13) or hemostatic system proteins (14). Plasminogen is a blood protein that is proteolysed into potent angiogenesis inhibitors, such as angiostatin (kringles 1-4) and kringle 5 (K5; refs. 15–17). Kringle domains contain f80 amino acids and three similarly linked disulfide bonds (18). Recombinant K5 (rK5) displays the most potent inhibitory activity to endothelial cell proliferation and migration (19–21) of known naturally occurring angiogenesis inhibitors. A recombinant K5 has also been shown to induce apoptosis in proliferating endothelial cells (19). Recently, reports show that rK5 prevents the development and arrests the progression of ischemia-induced retinal neovasculaRequests for reprints: Don Davidson, Department of Cancer Research, Abbott Laboratories, Department 48R, Building AP9, 100 Abbott Park Road, Abbott Park, IL 60064. Phone: 847-937-8547; Fax: 847-937-4150; E-mail: [email protected]. I2005 American Association for Cancer Research.

www.aacrjournals.org

rization in a rat model (21) by the down-regulation of VEGF and upregulation of pigment epithelial-derived factor. The tertiary structure of K5 has been studied in detail and contains a weak lysine-binding pocket. By contrast, the endothelial-binding site through which the antiangiogenic effects of K5 are mediated has not been defined (22). Here, we report that the antiangiogenic and proapoptotic activity of rK5 depends on a high-affinity binding interaction with glucose-regulated protein 78 (GRP78) exposed on the surface of stimulated endothelial cells and on many hypoxic and cytotoxic stressed tumor cells.

Materials and Methods Materials. Recombinant basic fibroblastic growth factor (bFGF) and VEGF were obtained from Invitrogen (San Diego, CA). Polyclonal COOHterminal (C-20) and NH2-terminal (N-20) GRP78 antibodies were obtained from Santa Cruz, Inc. (Santa Cruz, CA). A K5 monoclonal antibody was obained from Green Mountain Antibodies (Burlington, VT). All other antibodies used were obtained from Cell Signaling Technology, Inc. (Beverly, MA). Cell culture. Human microvascular endothelial cells (HMVEC), human umbilical arterial vascular endothelial cells, human umbilical endothelial cells, dermal fibroblasts, and neutrophils were obtained from Clonetics Corp. (San Diego, CA). D54 human glioma tumor cells were obtained from University of Texas-Southwestern Medical Center (Houston, TX). All other cell lines were obtained from American Type Culture Collection (Rockville, MD). Peptide synthesis. All peptides were synthesized using a Symphony (Protein Technology, Inc., Woburn, MA) automated peptide synthesizer. Peptide purification was done using a Gilson high-performance liquid chromatography system equipped with automated liquid handler. The Fmoc-protected amino acids and resins were purchased either from Calbiochem-Novabiochem Corp. (San Diego, CA) or from Bachem, Inc. (Torrance, CA). Mass spectra were recorded using either a Finnigan SSQ7000 (ESI) or JEOL JMS-SX102A-Hybrid (FAB) mass spectrometers. Cell proliferation assay. The effect of rK5 and rK5 peptides on endothelial cells was assessed using a proliferation assay with 1% bovine serum albumin (BSA) and 3 ng/mL bFGF in serum-free medium. Relative cell numbers in each well of a 96-well microplate after incubation for 72 hours in the absence or presence of inhibitors were determined by using the AQueous Cell Proliferation Assay (Promega, Madison, WI). For all other cell lines tested for proliferation, minimal growth medium was used (16). Results are presented as the percent inhibition of control cell (bFGFinduced) proliferation. Expression and purification of recombinant kringle 5. K5 fragment was PCR amplified from a human plasminogen cDNA template (American Type Culture Collection) with the following two primers: 5VCTGCTTCCAGATAGAGA-3V ( forward primer for residues 450-457) and 5V-TTATTAGGCCGCACACTGAGGGA-3V (reverse primer for residues 538543). The PCR fragment was ligated into the pET32a vector (Novagen,

4663

Cancer Res 2005; 65: (11). June 1, 2005

Cancer Research San Diego, CA) that had been digested with NcoI and XhoI. The NcoI and XhoI cleavage sites of the pET32a had been filled in to form blunt ends with plaque-forming unit DNA polymerase (2.5 units/AL, Stratagene, La Jolla, CA). XL2-Blue ultracompetent cells (Stratagene) were transformed with the ligation mixture as per the manufacturer’s instructions. After sequence confirmation, the pET32a/K5 vector was retransformed into Escherichia coli BL21 cells (DE3, Novagen) for expression as per the manufacturer’s instructions. The recombinant protein was recovered from the cell paste by cell lysis in lysis buffer [50 mmol/L Tris, 300 mmol/L NaCl, 1 mmol/L MgCl2 (pH 7.8)] using a French press. The His-tagged protein was purified over a Probond nickel resin (Invitrogen). The His tag was removed from the rK5 molecule by enterokinase (Invitrogen) and the rK5 was repurified over a second Probond nickel column to remove the His tag. Finally, endotoxin contamination was removed by size filtration (5 kDa) chromatography. Yeast rK5 was expressed as described previously (22). Briefly, the human K5 gene was expressed in the methylotrophic yeast Pichia pastoris (Invitrogen). Genetic transcription of rK5 was under the control of the alcohol oxidase promoter (AOX1). The AOX1 promoter permits high-level expression of heterologous proteins in Pichia. The K5 expression construct also includes a secretion signal sequence to direct transport of the protein to the medium. The plasmid construct was a hybrid of commercially available plasmid sequences from Invitrogen, designated pHIL-S1 and pHILD2. The expressed rK5 was purified by octyl-Sepharose and size exclusion chromatography. Radiolabeled recombinant kringle 5. rK5 was tritiated (3H) by a method published previously (23) Briefly, a carefully controlled particle beam composed of T3+ and T2+ ions and fast T2 molecules were accelerated into rK5 within a vacuum chamber. The 3HK5 was found to be active in the endothelial cell migration assay with an IC50 of 0.2 nmol/L and a specific activity of 8.74 mCi/mg. Human rK5 potency is highly dependent on the extent of iodination because the molecule contains a readily iodinated tyrosine in its binding sequence. Mutations of this tyrosine to phenylalanine resulted in incorrect protein folding. However, dog rK5 has a phenylalanine in this position naturally and is folded and active in our assays. We therefore relied on 125 IrK5 (dog) as a reagent. The radioiodination of rK5 (dog) was done following the procedure published by Markwell (24). The Iodobead reagent (Pierce, Rockford, IL) was used for the radioiodination and the labeling reaction as per protocol. A total of two beads were used with 25 Ag rK5 for the reaction. The separation of labeled rK5 from free iodine was accomplished using an iodine trap and a desalting spin filter (Pierce). 125IK5 (dog) was found to be activity in the migration assay. Endothelial cell migration assays. The effect of rK5 on endothelial cell migration was determined by two different methods. The first assay was done in a 96-well plate with a cellulose membrane between the upper and the lower chambers. HMVECs were starved of growth factors overnight, labeled with fluorescent calcein AM (50-100 nmol/L), plated into a 96-well migration chamber (2.9 104 per well, Neuroprobe), and stimulated to migrate with VEGF (5 ng/mL). After 4 hours, migrated cells were measured by fluorescence (25). In a second assay for cellular migration, a standard Boyden chamber was used (26). HMVEC cells were starved overnight in DMEM containing 0.1% BSA and harvested by scraping and resuspended in DMEM with 0.1% BSA at 1.5 106 cells/mL. Cells were added to the bottom of a 48-well Boyden chamber. The chamber was assembled and inverted, and cells were allowed to attach for 2 hours at 37jC to polycarbonate chemotaxis membranes (5 Am pore size) that had been soaked in 0.1% gelatin overnight and dried. The chamber was reinverted, test substances, including activators, were added to the wells of the upper chamber, and the apparatus was incubated for 4 hours at 37jC. Growth factors were used, where indicated, at concentrations determined in preliminary experiments to give equivalent migration responses of f100 cells migrated per highpower field (400). Growth factors and concentrations used were acidic FGF (aFGF; 50 ng/mL), bFGF (15 ng/mL), interleukin-8 (IL-8; 40 ng/mL), transforming growth factor-h (TGF-h; 1 pg/mL), VEGF (100 pg/mL), hepatocyte growth factor (HGF; 40 ng/mL), and platelet-derived growth factor (PDGF; 250 pg/mL). Membranes were recovered, fixed, and stained

Cancer Res 2005; 65: (11). June 1, 2005

and the number of cells that had migrated to the upper chamber per 10 high-power fields was counted. Background migration to DMEM plus 0.1% BSA was subtracted and the data were reported as the number of cells migrated per 10 high-power fields (400) or, when results from multiple experiments were combined, as the percent inhibition of migration compared with the positive growth factor control (26). Assessment of cellular apoptosis. The effects of rK5 and rK5 peptideinduced apoptosis were determined with a histone ELISA apoptosis assay (Roche, Indianapolis, IN; ref. 27); 5,000 cells per well were grown in 96-well plates. rK5 and/or an antibody to GRP78 were added to plates and incubated overnight. Apoptosis was determined from triplicate samples and the apoptotic index was determined by dividing the absorbance from the treated cells by the absorbance from the untreated cells. Binding of human recombinant kringle 5 to endothelial cells. Tritium-labeled rK5 was added to monolayers of 50,000 HMVEC cells that were either starved or stimulated for 16 hours by 15 ng/mL bFGF and 5 ng/mL VEGF in 96-well plates. The number of counts remaining bound to the cells after extensive washing determined the total amount of rK5 bound (28). Binding of 125I kringle 5 (dog) to endothelial cells and recombinant kringle 5 to recombinant glucose-regulated protein 78. The same methods as described above for the expression and purification of human K5 were used to express dog rK5 in E. coli. 125IK5 (dog) was added to the wells and incubated at room temperature for 1 hour. After 1 hour, the cells were washed and lysed with M-Per (Pierce) and the amount of 125IK5 (dog) bound was counted. Scatchard plot analysis was done using Prism software (GraphPad Software, Inc., San Diego, CA). Competition binding against 5 nmol/L 125IK5 (dog) was tested using a monoclonal antibody against GRP78 (N-20) or a monoclonal antibody against K5. The antibodies were added to HMVECs at various concentrations for 1 hour at room temperature. Cells were washed and the amount of bound 125IK5 (dog) bound was counted. Competition binding was also tested against 2 nmol/L 3 HK5. K5 peptides or a COOH-terminal antibody to GRP78 (A-129, Santa Cruz) were added to HMVECs at various concentrations for 1 hour at room temperature. Cells were washed and lysed and the amount of 3HK5 bound was estimated by scintillation counting. Immunohistochemical analysis of glucose-regulated protein 78 on human microvascular endothelial cells. Cells were starved overnight with medium alone. Complete medium, containing 10% fetal bovine serum plus 15 ng/mL bFGF and 5 ng/mL VEGF, was added at different times to the cells. GRP78 bound antibody was visualized with horseradish peroxidase (HRP)–reactive substrate visualized by a brown color. Binding and pull-down of recombinant kringle 5 binding proteins. Binding of rK5 to endothelial cells was measured as described (29). Briefly, HMVEC cells (50,000 per well) were cultured in 96-well microtiter plates for 2 hours, washed, and then incubated with PBS and increasing concentrations of 3HK5 or 125IK5 (dog) for another 2 hours at 4jC. After washing, cells were lysed and bound labeled rK5 was counted. Cell surface rK5 binding proteins were isolated by two methods. The first method used NH2-terminal biotinylated PRKLYDY active site rK5 peptide with 5 107 endothelial or tumor cell lysates. Cell lysate was passed over an agarose-avidin-biotin-PRKLYDY column. The column was washed with two column volumes of 100 nmol/L of the NH2-terminal rK5 peptide. Bound proteins were eluted with excess unlabeled rK5. Mass spectrometry analysis was used to determine the bound proteins. The second method used to identify cell surface rK5 binding proteins was described previously (29). Surface proteins on 4 106 EaHy cells were labeled with NHS-biotin. The cells were washed and lysed (M-Per, Pierce). Cell lysates were mixed with S-tag K5 for 1 hour at room temperature. S-tag K5 bound proteins were precipitated with S-protein agarose (Pierce). Bound proteins were eluted with excess rK5 or excess PRKLYDY peptide. Eluted proteins were visualized with avidin-HRP and a chemiluminescent substrate. Mass spectroscopic analysis was used to identify the major protein bands. Binding of rK5 to rGRP78 was measured using equilibrium dialysis (30). In the top well of a 96-well equilibrium dialyzer (molecular weight cutoff 50 kDa, Harvard Apparatus, Holliston, MA), 150 AL of 10 nmol/L rGRP78

4664


Kringle 5 Induces Apoptosis through GRP78 were added. In the reciprocal (bottom) chamber, 150 AL of increasing concentrations from 0.1 to 50 nmol/L 3HK5 were added. The chambers were shaken at room temperature for 72 hours. The total number of counts from both chambers after dialysis was compared with the number of counts remaining in the 3HK5 chamber. RNA interference. RNA interference of GRP78 expression was induced with small interfering RNA (siRNA) directed against the GRP78 mRNA. Three different nucleotide siRNA primers were made that targeted human GRP78 mRNA sequence. The siRNAs started at position 139 (AAC GGC CGC GUG GAG AUC AUC), position 1,175 (AAG CUG UAG CGU AUG GUG CUG), and position 1,567 (AAG AUC ACA AUC ACC AAU GAC). A scrambled siRNA from position 1,567 was used as negative control (AAA UCA UAG CGU AUG GUG CUG). All oligonucleotides were from Dharmacon Research (Dharmacon RNA Technologies, Lafayette, CO). EaHy or HT1080 cells were seeded at a density of 20,000 cells/cm2 the day before transfection and were f40% confluent when they were transfected with 50 nmol/L positive or scramble oligonucleotides in LipofectAMINE 2000 (Invitrogen) and Opti-MEM (Life Technologies, Gaithersburg, MD) without serum or BSA. Before transfection, the cells were washed once with Opti-MEM. Transfection medium was maintained on cells for 3 hours and was then removed and substituted with complete medium. The reduction in GRP78 protein, 48 hours after transfection, was estimated by Western blot analysis (31). In selected studies, GRP78 siRNA-transfected or scrambled siRNAtransfected EaHy cells were grown in 96-well plates. The medium was changed and 3HK5 in PBS was added to the cells at various concentrations. The cells were incubated at room temperature for 2 hours and then washed thoroughly. Cell counts were measured for bound 3HK5 as described above. Data points were calculated from the average of triplicate samples.

Results Recombinant kringle 5 and kringle 5 active site peptides inhibit endothelial cell activity. We have reported previously that rK5 has antiangiogenic activity in vitro, inhibiting bovine Table 1. Inhibition of VEGF stimulation HMVEC migration Peptides from K5

IC50 values (nmol/L)

K5 (450-543) Ac-VLLPDVETPSEED-NH2 Ac-MFGNGKGYRGKRATTVTGTP-NH2 Ac-QDWAAQEPHRHSIFTPETNPRAGLEKNY-NH2 Ac-RNPDGDVGGPW-NH2 Ac-YTTNPRKLYDY-NH2 Ac-DVPQ-NH2 Ac-RKLYDY-NH2 Ac-KLYDY-NH2 Ac-LYDY-NH2 Biotin-PRKLYDY NH2-S-tag-K5 125 IK5 (dog)

0.20 F >1,000 >1,000 >1,000 0.50 F 0.20 F >1,000 0.12 F 0.10 F >1,000 0.23 F 0.20 F 0.50 F

0.01

0.03 0.02 0.02 0.03 0.02 0.05 0.08

NOTE: HMVECs were added to the bottom of a 48-well Boyden chamber and allowed to attach to the membrane for 2 hours (25). The chambers were inverted and medium containing 3 ng/mL VEGF and peptides was added to each well. The cells were incubated for 4 hours at 37jC in 5% CO2 to migrate. Finally, cells that did not migrate across the membranes were scraped off and the cells that migrated were stained and counted. IC50 values were determined by comparing migrated cells in nontreated wells with treated wells. Each IC50 value listed is an average of triplicates.


Figure 1. rK5 is pleiotropic for migration inhibition. A, HMVEC migration assay was done as described previously (26). Growth factors were used, where indicated, at concentrations determined in preliminary experiments to give a migration response of f100 cells migrated per high-power field (100). Growth factors and concentrations used were aFGF (50 ng/mL), bFGF (15 ng/mL), IL-8 (40 ng/mL), TGF-h (1 pg/mL), VEGF (100 pg/mL), HGF (40 ng/mL), and PDGF (250 pg/mL). ABT-828 was 400 pmol/L, except in the cases of VEGF (rK5 = 100 pmol/L), PDGF (rK5 = 100 pmol/L), and HGF (rK5 = 10 pmol/L). Membranes were recovered, fixed, and stained and the number of cells that had migrated to the upper chamber per 10 high-power fields was counted. Background migration to DMEM + 0.1% BSA was subtracted and the data were reported as the number of cells migrated per 10 high-power fields (100). Results are from triplicate experiments. *, P < 0.01. B, induction of stimulated HMVEC apoptosis by rK5. The rate of apoptosis was measured in HMVEC cells using a histone detection kit (23). HMVECs were grown in 96-well plates. rK5 at various concentrations was added to plates and incubated overnight. Apoptosis was determined from triplicate samples and the apoptotic index was determined by dividing the absorbance from the treated cells by the absorbance from the untreated cells.

endothelial cell proliferation with an IC50 value f50 nmol/L (15). Here, we expand on those results using a yeast expressed rK5 (no detectable endotoxin) as well as synthetic K5 peptides to examine their effects on stimulated human endothelial cell proliferation, migration, and apoptosis assays. In these assays, rK5 inhibits stimulated human endothelial cell migration in a dose-dependent manner with an IC50 value (Table 1) of 0.20 nmol/L. rK5 inhibited endothelial cell migration induced by a wide variety of inducers of angiogenesis, including aFGF, bFGF, IL-8, PDGF, TGF-h, and VEGF (Fig. 1A). It was selective for endothelial cells because it failed to inhibit the migration of neutrophils and fibroblasts even when tested at concentrations up to 1,000-fold higher than that at which it inhibited endothelial cell migration (data not shown). rK5 also showed selectivity for inhibition of proliferation of stimulated endothelial cells and did not cause inhibition of tumor or primary cell proliferation at concentrations as high as 100 Amol/L (Table 2). Apoptosis of stimulated endothelial cells was induced by rK5 (Fig. 1B) in a dose-dependent manner, indicating that the binding of rK5 initiates a cell signaling cascade leading to cell death.

4665

Cancer Res 2005; 65: (11). June 1, 2005

Cancer Research

endothelial cells. However, peptides outside of this binding site pocket or tripeptides from the active site pocket were inactive at blocking endothelial cell migration. A NH2-terminally biotinylated PRKLYDY peptide and a S-protein-tagged rK5 were similarly active for the inhibition of stimulated endothelial cell migration.

Table 2. Inhibition of tumor cell proliferation by rK5 Tumor cell lines

HT-29 A549 LNCaP P388 B16F10 PC-3 MDA-445 D54 HT1080 Lewis lung carcinoma

IC50 values for cell proliferation* rK5 (Ag/mL)

Adriamycin (Ag/mL)

>100 >100 >100 >100 >100 >100 >100 >100 >100 >100

0.118 F 0.008 0.051 F 0.006 0.41 F 0.10 0.0026 F 0.0002 0.0032 F 0.0004 c NT NT 0.005 F 0.0004 0.002 F 0.0003 NT

*Tumor cells were plated at 2,000 cells per well in a 96-well plate in full medium and allowed to attached overnight. rK5 or Adriamycin was added to the wells in fresh medium and allowed to grow for 72 hours. The number of cells was measured by AQueous Cell Proliferation kit. IC50 values were determined by comparing treated cells with nontreated cells. Each test was run in quadruplicate. cNT, not tested.

To further explore the binding of rK5 to endothelial cell surfaces, linear peptides made from various regions of K5 were also tested in the endothelial cell migration assay (Table 1). Peptides from the lysine-binding site of K5 displayed the highest activity at blocking stimulated endothelial cell migration. In fact, peptides PRKLYDY, KLYDY, and KLYD were equally potent compared with rK5 in the inhibition of migration of activated

Figure 2. Cell surface K5 binding proteins. A, Coomassie-stained gel of cell surface proteins bound to an immobilized biotin-PRKLYDY peptide. The bound proteins were eluted off with excess rK5. Lane A, eluent from K5 peptide column. Tryptic peptides from the isolated bands were analyzed by mass spectrometric analysis. The major protein was identified as GRP78. Lane B, molecular weight standards. B, avidin-HRP blots of biotinylated cell surface proteins isolated by affinity purification with immobilized S-tag K5. Surface proteins on 4 106 endothelial cells were labeled with NHS-biotin. The cells were washed and lysed. The biotinylated surface proteins were isolated with avidin-agarose beads and eluted from the beads with excess streptavidin. Purified cell surface lysates were mixed with S-tag K5 for 1 hour at room temperature. S-tag K5 bound proteins were precipitated with S-protein agarose. Lane A, bound proteins were eluted with excess rK5; lane B, bound proteins were eluted with excess PRKLYDY peptide. Eluted proteins were visualized with avidin-HRP and a chemiluminescent substrate. Mass spectrometric analysis identified the bands as GRP78 and GRP94.

Cancer Res 2005; 65: (11). June 1, 2005

Figure 3. The active site of K5 binds to endothelial cells through GRP78. For all the assays listed, 50,000 cells were attached to 96-well plates. A, 125IK5 (dog) HMVEC binding assay. Human rK5 potency is highly dependent on the extent of iodination because it contains a readily iodinated tyrosine in its binding sequence. Attempted mutations of this tyrosine to phenylalanine did not allow correct protein folding. However, dog K5 has a phenylalanine in this position naturally and is folded and active in our assays. We therefore relied on 125IrK5 (dog) as a reagent. 125IK5 (dog) was added to the wells and incubated at room temperature for 1 hour. After 1 hour, the cells were washed and the amount of 125 IK5 (dog) bound was counted. Scatchard plot analysis shows 125IK5 (dog) bound with a K d of 0.8 nmol/L with f32,500 binding sites per cell. B, competition binding against 5 nmol/L 125IK5 (dog). Anti-GRP78 or anti-K5 was added to HMVECs at various concentrations for 1 hour at room temperature. Cells were washed and the amount of 125IK5 (dog) bound was counted. C, competition binding against 2 nmol/L 3HK5. K5 peptides or a NH2-terminal antibody to GRP78 were added to HMVECs at various concentrations for 1 hour at room temperature. Cells were washed and the amount of 3HK5 bound was counted.

4666


Kringle 5 Induces Apoptosis through GRP78

Figure 4. rK5 but not rK5(K82A) mutant competes with anti-GRP78 antibody binding to EaHy cells. EaHy cells were grown overnight on glass slides. K5 or mutant rK5(K82A) at concentration of 500 nmol/L was added to the cells for 1 hour. Cells were washed and fixed. The surface GRP78 was visualized with a goat polyclonal GRP78 antibody using a 3,3V-diaminobenzidine system. Intensity of the brown stain indicates amount of GRP78 available for antibody binding.

These probes were important for the isolation of the cell surface receptor for rK5. Glucose-regulated protein 78 is an endothelial cell surface recombinant kringle 5 binding protein. To reduce the cell surface nonspecific protein binding that is often observed with proteins, an immobilized NH2-terminally biotinylated PRKLYDY peptide was used to isolate K5 binding proteins from endothelial cell surfaces. Bound proteins were then eluted with excess rK5. Mass spectrometric sequencing of tryptic peptides from the major protein band (f80 kDa; Fig. 2A) revealed sequences corresponding to GRP78 (78 kDa). That GRP78 is a cell surface binding protein for rK5 was further confirmed by coprecipitation of biotinylated surface proteins with S-tagged K5. The major bound biotinylated proteins eluted with excess rK5 from immobilized S-tagged K5 (Fig. 2B) were identified by mass spectrometric analysis as GRP78 and GRP94.

Characterization of glucose-regulated protein 78 binding to recombinant kringle 5. The possibility that rK5 binds to GRP78 on endothelial cells was further explored by the examination of the amount of surface-expressed GRP78 compared with rK5 binding. Labeled K5 [125IK5 (dog) or human 3HK5] bound specifically to these cells with a K d of 0.8 nmol/L (Fig. 3A), which is the same order of magnitude for rK5’s inhibition of endothelial cell migration (0.25 nmol/L) and induction of endothelial cell apoptosis (1 nmol/L). A goat polyclonal antibody recognizing the NH2-terminal region of GRP78 inhibited the binding of 125IK5 to proliferating endothelial cells in a concentration-dependent manner (Fig. 3B). In comparison, a goat polyclonal antibody raised against the COOH-terminal region of GRP78 had no effect on cellular 125IK5 binding. Competition binding with the K5 active site peptide, PRKLYDY (residues 80-86), but not the inactive rK5 NH2-terminal peptide, LLPDVETPSEED (residues 1-12), blocked 3 HK5 binding to endothelial cells in a concentration-dependent manner (Fig. 3C). These results show that GRP78 and rK5 associate on the surface of proliferating endothelial cells. To investigate the specificity of GRP78/rK5 interaction, a mutant rK5(K82A) was used to compete with 3HK5 binding to endothelial cell surfaces. Unlike rK5, the mutant rK5(K82A) at concentrations up to 500 nmol/L did not inhibit the binding of 3HK5 or reduced the binding of a NH2terminal GRP78 antibody on stimulated endothelial cell surfaces as determined by immunohistochemical analysis (Fig. 4). The binding of rK5 to endothelial cells is dramatically reduced in starved, quiescent cells compared with VEGF/bFGF stimulated cells (Fig. 5A). Analysis of GRP78 cell surface expression, using immunohistochemical techniques, displayed a readily detectable increase in GRP78 expression on stimulated endothelial cell surfaces compared with cells after starvation (Fig. 5B). These results taken together indicate that the increase in rK5 binding to endothelial cell surfaces correlates with the amount of GRP78 expression on cells. A NH2-terminal antibody to glucose-regulated protein 78 or reduced expression of glucose-regulated protein 78 inhibits the activity of recombinant kringle 5 on endothelial cells. The previous results suggest that GRP78 plays a role in the activity of rK5 on endothelial cells. A NH2-terminal GRP78 polyclonal antibody blocked the inhibition caused by rK5 on stimulated

Figure 5. rK5 binding correlates with GRP78 surface expression on HMVECs. A, specific binding of 125IK5 to HMVEC cells. Tritium-labeled rK5 was added to monolayers of 50,000 HMVEC cells that were either starved or stimulated for 16 hours by 15 ng/mL bFGF and 5 ng/mL VEGF in 96-well plates. The number of counts remaining bound to the cells after extensive washing determined the total amount of rK5 bound. B, immunohistochemical analysis of GRP78 on HMVEC cells at different times after activation with complete medium containing VEGF and bFGF. Cells were starved overnight with medium alone. Complete medium, containing 10% fetal bovine serum + 15 ng/mL bFGF and 5 ng/mL VEGF, was added to the cells for immunohistochemical analysis. GRP78 bound antibody was visualized with HRP-reactive substrate visualized by a brown color.


4667

Cancer Res 2005; 65: (11). June 1, 2005

Cancer Research

endothelial cell proliferation in a concentration-dependent manner (Fig. 6A), whereas antibodies to unrelated proteins, HSP70, fibrin, and avh3, had no effect, except at very high concentrations (Fig. 6B). The same NH2-terminal GRP78 antibody also blocked the rK5induced inhibition of endothelial cell migration (Fig. 6C) and rK5induced apoptosis of endothelial cells (Fig. 6D). The expression of GRP78 on endothelial cells was significantly reduced by the transfection of a siRNA for GRP78 but not by the transfection of a scrambled siRNA (Fig. 7A). The binding of rK5 to GRP78 siRNA-

transfected HMVEC cells was very low compared with the binding of rK5 to scrambled siRNA-transfected cells (Fig. 7B). These results confirm the requirement for GRP78 expression for rK5 inhibitory activity on endothelial cells. Direct binding of recombinant kringle 5 with recombinant glucose-regulated protein 78. To directly observe the binding of GRP78 and rK5, we used yeast-expressed recombinant human GRP78 (rGRP78) protein and 3HK5. Equilibrium dialysis of rGRP78 and 3HK5 revealed direct binding of rGRP78 and rK5 in a concentration-dependent manner with a K d of f0.7 nmol/L (Fig. 8A). This binding could be blocked by the addition of unlabeled rK5 (Fig. 8B). However, the rK5(K82A) mutant in concentrations up to 25-fold excess of GRP78 concentration did not block binding, confirming that GRP78 binds with high affinity and specificity to rK5. Recombinant kringle 5 causes apoptosis of hypoxic tumor cells. Our results show that the inhibitory activity of rK5 on endothelial cells is mediated through cell surface GRP78. We have also shown that K5 does not affect tumor cell proliferation in normal conditions in vitro (Table 2). However, published data indicate that tumor cells up-regulate surface GRP78 expression under stressed conditions (32). When hypoxia and/or the addition of cytotoxic stress are applied to tumor cells, an apoptosisresistant phenotype often emerges. This resistance has been linked to GRP78 surface expression (32). We examined the effect of rK5 on tumor cells under stressed hypoxic conditions. Hypoxia increased GRP78 expression on HT1080 human fibrosarcoma cells at least 4-fold (Fig. 9A). When rK5 was added to these cells under hypoxic conditions (5% CO2, 95% N2), there was a >2-fold increase in apoptosis within 24 hours (Fig. 9B), whereas hypoxia alone or the addition of rK5 without hypoxia had no significant effect on cellular apoptosis. To determine if the proapoptotic effect of rK5 on hypoxic tumor cells was specific to HT1080 cells, we tested apoptosis rates for various tumor cell lines exposed to rK5 under hypoxia. As seen in Fig. 9B, rK5 induced apoptosis of EaHy endothelial cells with and without hypoxia, supporting previous data where the rK5 receptor was up-regulated during endothelial cell stimulation in normoxic conditions. However, rK5 induced apoptosis in seven of nine tumor cell lines tested only under hypoxic conditions. The two tumor cells lines that did not respond to rK5 under hypoxic conditions were Lewis lung and PC-3. For a possible explanation, we examined GRP78 surface expression on

Figure 6. GRP78 antibody inhibits rK5 cellular activity. A, anti-GRP78 blocks K5’s inhibition of HMVEC proliferation. HMVEC cellular proliferation assay was done with 10,000 cells added to each well of a 96-well plate. After the cells had attached, complete medium containing VEGF (5 ng/mL) and bFGF (15 ng/mL) along with rK5 with or without the GRP78 antibody was also added. The cells were grown for 72 hours. 3-[4,5,Dimethylthiazol-2-yl]-5-[3carboxymethoxy-phenyl]-2-[4-sulfophenyl]-2H-tetrazolium, inner salt assay was used to determine the number of live cells. Results were determined from triplicate samples. B, nonspecific antibodies do not block rK5’s inhibition of HMVEC proliferation. Various antibodies (100 Ag/mL) to unrelated proteins were added to HMVEC cellular proliferation assays containing 100 ng/mL K5 as mentioned above. Inhibition of cellular proliferation was determined from quadruplicate tests. C, anti-GRP78 blocks rK5’s inhibition of stimulated HMVEC migration. Stimulated HMVEC cell chemotaxis toward VEGF (5 ng/mL) was done in 96-well plates for 4 hours at room temperature. Casein-AM prelabeled HMVEC cells that had migrated to the bottom of the membrane were measured by fluorescence. D, anti-GRP78 inhibits the increase in HMVEC cellular apoptosis caused by rK5. The rate of apoptosis was measured in HMVEC cells using a histone detection kit. HMVECs were grown in 96-well plates. rK5 and an antibody to GRP78 were added to plates and incubated overnight. Apoptosis was determined from triplicate samples and the apoptotic index was determined by dividing the absorbance from the treated cells by the absorbance from the untreated cells.

Cancer Res 2005; 65: (11). June 1, 2005

4668



should show an increase in caspase-7 activity compared with untreated cells. Our data (Fig. 10) show a significant increase in caspase-7 activities in HT1080 cells when treated with rK5 and hypoxia, consistent with this hypothesis.

Discussion These results show that rK5 binds with high affinity to endothelial cells through an interaction on the lysine-binding site of rK5 and GRP78. The inhibition of this interaction with an antibody against GRP78 blocks the induction of endothelial cell apoptosis. Exposure of quiescent endothelial cells to VEGF or of tumor cells to a stress-like hypoxia increases surface-expressed GRP78 and as such increases the number of binding sites for rK5. Peptides from the lysine-binding site of rK5 potently compete with rK5 cellular binding and induce endothelial cell apoptosis. As further evidence that the rK5 lysine-binding site is important for the binding of rK5 to GRP78, the Lys82 of K5 was mutated to alanine [rK5(K82A)], which abolished f90% of its ability to inhibit in vitro migration, although the kringle was still correctly folded as

Figure 7. Reduced EaHy cell surface expression of GRP78 by siRNA results in reduced binding of 3HK5. A, LipofectAMINE 2000 has been used to transfect EaHy cells with the siRNA oligonucleotides. Transfection has been carried out with LipofectAMINE 2000 at 12 AL per six-well plate in Opti-MEM followed by normal medium after 3-hour transfection and then looked for the reduction in GRP78 protein 48 hours after transfection. siRNA-7 significantly reduced the expression of GRP78. B, GRP78 siRNA or scrambled siRNA-transfected EaHy cells were grown in 96-well plates. The medium was changed and 3HK5 in PBS was added to the cells. The cells were incubated at room temperature for 2 hours and then washed thoroughly. Cell counts were measured for bound 3 HK5. Points, average of triplicate studies.

PC-3 cells under hypoxic conditions. Immunohistochemical analysis of PC-3 and Lewis lung cells showed no increase in GRP78 expression under hypoxia (data not shown), which is in sharp contrast with the stress-induced GRP78 expression on HT1080 cells. To make the expression of GRP78 on HT1080 cells appear more like PC-3 cells, we used a siRNA to GRP78 for the knockdown of protein expression. There was an f90% decrease in GRP78 expression in transfected HT1080 (GRP78 ) cells (data not shown), which was very similar to the siRNA knockdown of GRP78 expression we observed in HMVECs (Fig. 7A). This decrease in GRP78 protein expression significantly eliminated the proapoptotic activity of rK5 on hypoxic HT1080 (GRP78 ) cells (Fig. 9C). However, rK5 significantly induced apoptosis of the hypoxic HT1080 cells transfected with scrambled siRNA and control nontransfected cells. These data show the necessary GRP78 expression for rK5 induction of apoptosis on tumor cells. Recent reports have shown that the ATPase domain of GRP78 binds to procaspase-7, blocking its activation and decreasing stressinduced cellular apoptosis. Our nuclear magnetic resonance studies with a recombinant GRP78 ATPase domain show direct binding of rK5 with the ATPase domain of GRP78.4 We have also shown that the GRP78 ATPase domain on stressed tumor cells is extracellular (Fig. 9A). If rK5 is then internalized and competes with procaspase-7 for GRP78 binding, then stressed tumor cells treated with rK5

4

In preparation.


Figure 8. Equilibrium dialysis with rGRP78 and rK5. 3HK5 was added to one side of an equilibrium chamber at concentrations from 0.1 to 3 nmol/L. GRP78 (2 nmol/L) was added to the other side of the chamber with a 50,000-kDa MWCO filter between. 3HK5 was allowed to equilibrate to both sides of the chamber for 72 hours at room temperature. A, GRP78 and rK5 bind with high affinity. The amount of 3HK5 on the GRP78 side of the chamber was compared with chamber with 3HK5 alone to obtain the bound versus free rK5 distribution. B, mutant rK5(K82A) does not compete with binding of rK5 and rGRP78. Competition binding dialysis was done with 2 nmol/L GRP78 and 2 nmol/L 3HK5 using rK5(K82A) mutant and unlabeled rK5 at various concentrations from 0.5 to 50 nmol/L. The amount of 3HK5 on both sides of the chamber was measured after 72 hours and compared with the control.

4669

Cancer Res 2005; 65: (11). June 1, 2005

Cancer Research

Figure 9. rK5 inhibited proliferation and increased apoptosis of tumor cells in hypoxic conditions by the up-regulation of GRP78. A, HT1080 fibrosarcoma tumor cells were plated on slides overnight with complete medium. Control slides were put in the incubator at 5% CO2 at 37jC. The other slide of cells was put in a hypoxic chamber with 95% N2, 5% CO2 at 37jC for 24 hours. GRP78 on the cell surface was detected by immunohistochemical analysis with a polyclonal antibody to GRP78 conjugated to HRP. The GRP78 was visualized with the 3, 3V-diaminobenzidine reagent displaying a brown color. B, cells were plated in 96-well plates overnight. Half the plates were put in hypoxic chambers overnight at 37jC with and without 500 nmol/L rK5. Increased apoptosis was shown by an ELISA technique that measures the number of nucleosome fragments. The apoptotic index was calculated from the apoptosis rate of control cells (without K5 and with or without hypoxia). The samples were run in quadruplicate. C, HT1080 tumor cells were transfected with a siRNA to GRP78. The transfected cells were plated in 96-well plates and placed at 5% CO2 at 37jC. rK5 (500 nmol/L) was added to the cells and half were incubated in an hypoxic chamber with 95% N2, 5% CO2 at 37jC and half were incubated at 5% CO2 at 37jC for 24 hours. Apoptosis was shown by an ELISA technique mentioned above. Samples were run in quadruplicate. *, P < 0.01.

determined by nuclear magnetic resonance analysis (22). This mutant K5(K82A) did not compete with 3HK5 binding to endothelial cells or with direct binding of 3HK5 to recombinant GRP78 (K d = 0.25 nmol/L). To further elucidate this binding,

Cancer Res 2005; 65: (11). June 1, 2005

immunohistochemical analysis of cells with anti-GRP78 were examined with and without K5 or K5(K82A) mutant. Wild-type rK5 reduced the binding of the GRP78 antibody to cells, but the K5(K82A) mutant had no effect on GRP78 antibody binding. This specificity of binding on cell surfaces was also examined using the siRNA for GRP78 on endothelial cells. When siRNA transfection reduced the GRP78 protein concentration by f90% on EaHy and HT1080 cells, there was greatly reduced binding of 3HK5 to cell surfaces. These observations show that the effects of rK5 on proliferating endothelial cells require direct, specific, high-affinity binding to GRP78. This conclusion is not excluded by the results of a previous study reporting that the K5 endothelial cell surface binding protein was voltage-dependent anion channel (VDAC; ref. 33). That study used confluent, static monolayers of endothelial cells to examine rK5 binding. However, our findings show that under confluent conditions GRP78 is minimally exposed on the endothelial cell surface; hence, other binding proteins may play a more prominent role for rK5 binding under these static growth conditions. Our studies were carried out under subconfluent conditions. There are also reports that show VDAC is chaperoned to cell surfaces by heat shock or glucose-regulated proteins. It is possible that rK5 binds to GRP78, which is associated with VDAC on cell surfaces (34). VDAC siRNA knockdown experiments are under way to determine if rK5 binding to GRP78 may depend on VDAC expression. However, our results challenge the direct binding of rK5 to VDAC because of the tight binding of rK5 to GRP78 (K d = 0.25 nmol/L) compared with the reported rK5 binding to VDAC (K d = 28 nmol/L). A recent study showing that rK5 is more effective on hypoxic endothelial cells supports our data of increased rK5 surface binding to endothelial cells with stress and supports our finding of increased apoptosis by rK5 on hypoxic tumor cells. The GRP78 antibody and GRP78 siRNA provide nearly quantitative inhibition of rK5 binding to endothelial cell surfaces providing additional evidence for GRP78 as the dominant rK5 binding protein. In addition to identifying and establishing the importance of an uncharacterized endothelial cell binding protein for rK5, our results also show that rK5 directly induces tumor cell apoptosis under stressed conditions. Reports have shown that the induction of GRP78 is critical for tumor cells to maintain viability during stress-like hypoxia and glucose deprivation. In addition, the induction of tumor cell surface–expressed GRP78 by hypoxic conditions has been linked to resistance to cytotoxic or radiation therapy in vitro and in vivo (35). Surface-expressed GRP78 was induced by hypoxia on the surface of 8 of the 10 tumor cell lines tested here. Under these conditions, rK5 had a very potent proapoptotic effect on all eight of the hypoxia-stressed tumor cell lines that induced surface-expressed GRP78. The two tumor lines, PC-3 and Lewis lung, on which rK5 had no proapoptotic effect, did not up-regulate GRP78 under hypoxic conditions. These results show the selectivity of rK5 for most stressed tumor cells and activated endothelial cells but not for normal or quiescent endothelial cells. The identification of GRP78 as an endothelial and tumor cell– binding site required for the antitumor and antiangiogenic activity of rK5 raises several questions concerning the mechanism by which rK5 induces cellular apoptosis. Although rK5 binds to GRP78, there is not a predicted transmembrane domain in GRP78 to transfer signal to the cytosol. However, published reports indicate that GRP78 may protect cells from apoptosis during stress by blocking procaspase-7 activation (36). We hypothesize that rK5 binds to

4670



Figure 10. A, caspase-7 activity is increased by rK5 in HT1080 cells. HT1080 cells were plated in 48-well plates. rK5 at 100 nmol/L was added to the plates. HT1080 cells were incubated overnight in hypoxic conditions (95% N2, 5% CO2) compared with controls, which were incubated in 5% CO2. Caspase-7 activity in cell lysates was detected by the activation of an ELISA fluorescent substrate (R&D Systems, Minneapolis, MN). B, rK5 reduces the surface-expressed GRP78 on HMVEC cells. HMVECs were plated on coverslips overnight. rK5 (100 nmol/L) was added to the cells the next day. Immunohistochemical analysis of cell surface-expressed GRP78 was detected by a noncompeting GRP78 antibody at 0, 6, and 24 hours after treatment. The concentration of GRP78 on the surface of endothelial cells greatly decreased after rK5 addition.

GRP78 and is endocytosed into the cell where rK5 blocks GRP78 binding to procaspase-7. Evidence supporting this hypothesis is shown by our finding that when rK5 is added to stressed endothelial cells or stressed HT1080 cells caspase-7 activity is increased at least 2-fold within 2 hours. We also show that the cell surface concentration of GRP78 is greatly reduced within a few hours after rK5 binding (Fig. 10B). Expression of recombinant procaspase-7 is under way, which will help to determine if rK5 can inhibit the binding rGRP78 to procaspase-7. Although rK5 could block the protection of procaspase-7 leading to active caspase-7 generation, the appearance of other caspases within the cell necessary to activate caspase-7 argues that additional mechanisms for rK5’s antiangiogenic activity exist. In summary, we have shown that an interaction between rK5 and GRP78 on stimulated endothelial and stressed tumor cells underlies the proapoptotic activity of rK5. To our knowledge, this is the first evidence that an antiangiogenic compound is effective directly against stressed tumor cells. Finally, we predict that a

References 1. Folkman J. Angiogenesis in cancer, vascular, rheumatoid and other disease. Nat Med 1995;1:27–31. 2. Folkman J. Tumor angiogenesis. Adv Cancer Res 1985; 43:175–203. 3. Shweiki D, Itin A, Soffer D, Keshet E. Vascular endothelial growth factor induced by hypoxia may mediate hypoxia-initiated angiogenesis. Nature (London) 1992; 359:843–5. 4. Thomas KA. Vascular endothelial growth factor, a potent and selective angiogenic agent. J Biol Chem 1996;271:603–6. 5. Wang Y, Becker D. Antisense targeting of basic fibroblast growth factor and fibroblast growth factor receptor-1 in human melanomas blocks intratumoral angiogenesis and tumor growth. Nat Med 1997;3:887–93. 6. Bussolino F, Mantovani A, Persico G. Molecular mechanisms of blood vessel formation. Trends Biochem Sci 1997;22:251–6. 7. Dawson DW, Volpert OV, Gillis P, et al. Pigment epithelium-derived factor: a potent inhibitor of angiogenesis. Science 1999;285:245–8. 8. Jimenez B, Volpert OV. Mechanistic insights on the


two-stage inhibition of in vivo tumor growth by rK5 could exist. First, rK5 would induce apoptosis in activated tumor endothelial cells. Subsequently, the consequent hypoxia would cause tumor cells to present surface-expressed GRP78, allowing rK5 to enhance tumor cell apoptosis. The second state of rK5 action could go beyond tumor stasis and possibly lead to tumor regression. Further efforts to characterize the biology of GRP78 as well as the binding interaction between GRP78 and rK5 may yield insight into both the regulation of angiogenesis and the survival of hypoxic tumor cells. These findings could lead to the development of novel agents targeted toward pathologic angiogenesis and tumor cell survival.

Acknowledgments Received 9/21/2004; revised 12/17/2004; accepted 3/22/2005. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

inhibition of tumor angiogenesis. J Mol Med 2001;78: 663–72. 9. Holmgren L, O’Reilly MS, Folkman J. Dormancy of micrometastases: balanced proliferation and apoptosis in the presence of angiogenesis suppression. Nat Med 1995;1:149–53. 10. Hanahan D, Folkman J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996;86:353–64. 11. Dameron KM, Volpert OV, Tainsky MA, Bouck N. Control of angiogenesis in fibroblasts by p53 regulation of thrombospondin-1. Science 1994;265:1582–4. 12. Fang J, Yan L, Shing Y, Moses MA. HIF-1a-mediated up-regulation of vascular endothelial growth factor, independent of basic fibroblast growth factor, is important in the switch to the angiogenic phenotype during early tumorigenesis. Cancer Res 2001;61:5731–5. 13. Sage EH. Pieces of eight: bioactive fragments of extracellular proteins as regulators of angiogenesis. Trends Cell Biol 1997;7:182–6. 14. Browder T, Folkman J, Pirie-Shepard S. The hemostatic system as a regulator of angiogenesis. J Biol Chem 2000;275:1521–4. 15. O’Reilly MS, Holmgren L, Shing Y, et al. Angiostatin: a

4671

novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma. Cell 1994;79:315–28. 16. Cao Y, Ji R, Davidson D, et al. Kringle domains of human angiostatin. Characterization of the antiproliferative activity on endothelial cells. J Biol Chem 1996;271:29461–7. 17. Castellino FJ, McCance SG. The kringle domains of human plasminogen. Ciba Found Symp 1997;212:46–60. 18. Cao Y, Chen A, An SSA, et al. Kringle 5 of plasminogen is a novel inhibitor of endothelial cell growth. J Biol Chem 1997;272:22924–8. 19. Lu H, Dhanabal M, Volk R, et al. Kringle 5 causes cell cycle arrest and apoptosis of endothelial cells. Biochem Biophys Res Commun 1999;258:668–73. 20. Ji WR, Barrientos LG, Llinas M, et al. Selective inhibition by kringle 5 of human plasminogen on endothelial cell migration, an important process in angiogenesis. Biochem Biophys Res Commun 1998;247: 414–9. 21. Zhang D, Kaufman PL, Gao G, Saunders RA, Ma JX. Intravitreal injection of plasminogen kringle 5, an endogenous angiogenic inhibitor, arrests retinal neovascularization in rats. Diabetologia 2001;44:757–65.

Cancer Res 2005; 65: (11). June 1, 2005

Cancer Research 22. Chang Y, Mochalkin I, McCance SG, Cheng B, Tulinsky A, Castellino FJ. Structure and ligand binding determinants of the recombinant kringle 5 domain of human plasminogen. Biochemistry 1998; 37:3258–71. 23. Bush GA, Yoshida N, Lively MO, et al. Ion beam tritium labeling of proteins and peptides. J Biol Chem 1981;256:12213–21. 24. Markwell MAK. A new solid-state reagent to iodinate proteins. I. Conditions for the efficient labeling of antiserum. Anal Biochem 1982;125:427–32. 25. Frevert CW, Wong VA, Goodman RB, Goodwin R, Martin TR. Rapid fluorescence-based measurement of neutrophil migration in vitro . J Immunol Methods 1998;213:41–52. 26. Polverini PJ, Bouck NP, Rastinejad F. Assay and purification of naturally occurring inhibitor of angiogenesis. Methods Enzymol 1991;198:440–50. 27. Mayr M, Li C, Zou Y, Huemer U, Hu Y, Xu Q. Biomechanical stress-induced apoptosis in vein grafts

Cancer Res 2005; 65: (11). June 1, 2005

involves p38 mitogen-activated protein kinases. FASEB J 2000;14:261–70. 28. Dudani AK, Ganz PR. Endothelial cell surface actin serves as a binding site for plasminogen, tissue plasminogen activator and lipoprotein(a). Br J Haematol 1996;95:168–78. 29. Zhang JC, Donate F, Qi X, et al. The antiangiogenic activity of cleaved high molecular weight kininogen is mediated through binding to endothelial cell tropomyosin. Proc Natl Acad Sci U S A 2002;99:12224–9. 30. Kariv I, Cao H, Oldenburg KR. Development of a high throughput equilibrium dialysis method. J Pharm Sci 2001;90:580–7. 31. Spangnolo R, Corada FO, Zanetta L, et al. Gas 1 is induced by VE-cadherin and vascular endothelial growth factor and inhibits endothelial cell apoptosis. Blood 2004;8:3005–12. 32. Koomagi R, Mattern J, Volm M. Glucose-related protein (GRP78) and its relationship to the drugresistance proteins P170, GST-pi, LRP56 and angioge-

4672

nesis in non-small cell lung carcinomas. Anticancer Res 1999;19:4333–6. 33. Gonzalez-Gronow M, Kalfa T, Johnson CE, Gawdi G, Pizzo SV. The voltage-dependent anion channel is a receptor for plasminogen kringle 5 on human endothelial cells. J Biol Chem 2003;278:27312–8. 34. Schwarzer C, Barnikol-Watanabe S, Thinnes FP, Hilschmann N. Voltage-dependent anion-selective channel (VDAC) interacts with the dynein light chain Tctex1 and the heat-shock protein PBP74. Int J Biochem Cell Biol 2002;34:1059–70. 35. Namiki A, Brogi E, Kearney M, et al. Hypoxia induces vascular endothelial growth factor in cultured human endothelial cells. J Biol Chem 1995;270:31189–95. 36. Reddy RK, Mao C, Baumeister P, Austin RC, Kaufman RJ, Lee AS. Endoplasmic reticulum chaperone protein GRP78 protects cells from apoptosis induced by topoisomerase inhibitors: role of ATP binding site in suppression of caspase-7 activation. J Biol Chem 2003; 278:20915–24.
