Antiangiogenic Activity of a Domain Deletion Mutant of Tissue Plasminogen Activator Containing Kringle 2
Objective— The thrombolytic therapy drug, Reteplase, is a domain deletion mutant of tissue plasminogen activator (tPA), comprising the kringle 2 and protease (K2P) domains. Some kringle domains of hemostatic proteins are antiangiogenic and promote apoptosis. The objective of this study was to investigate whether K2P is an angiogenesis inhibitor because of the presence of kringle 2.
Methods and Results— K2P inhibited basic fibroblast growth factor-induced human endothelial cell proliferation and migration. Inhibition was not dependent on the protease activity of K2P because similar results were obtained with catalytically inactivated K2P. Purification of the kringle 2 domain derived from elastase cleavage of K2P at the Arg275–Ile276 bond revealed that inhibition was mediated by this domain. In addition, K2P inhibited angiogenesis in vivo and increased endothelial cell apoptosis.
Conclusions— Wound healing and angiogenesis are severely compromised by K2P. These data provide new mechanistic insights into the bleeding complications observed in some patients while undergoing thrombolytic therapy with this drug. In addition, we identify the kringle 2 domain of tPA as a novel target for antiangiogenic therapy.
Tissue plasminogen activator (tPA) is a serine protease that activates fibrinolysis through the conversion of plasminogen to plasmin. Plasmin, in turn, degrades fibrin in the vasculature during clot dissolution and wound healing.1 tPA is expressed in a variety of cell types including endothelial cells2 and tumor cells.3,4 It is a multidomain protein that consists of a finger domain, an epidermal growth factor-like domain, 2 kringle modules, and a protease domain. Its widespread use for treating thrombosis during acute myocardial infarction has led to the development of tPA derivatives with improved pharmacokinetic properties.5 One of these is the domain deletion mutant comprising just the kringle 2 and protease domains (K2P) of tPA.6
Kringles are triple-loop structures of ≈80 amino acids.7 Recent studies have shown that isolated kringle domains of proteins involved in the hemostatic system are antiangiogenic. These include plasminogen, which has 5 kringle domains,8 hepatocyte growth factor, which has 4,9 prothrombin, which has 2 kringles,10 and urokinase plasminogen activator (uPA), which has just 1.11 Angiostatin, consisting of kringles 1 to 4 of plasminogen, was first shown to inhibit basic fibroblast growth factor (bFGF)-mediated endothelial cell proliferation and tumor angiogenesis.12 Further studies have shown that the individual plasminogen kringles have varying efficacy on endothelial cells. Kringles 1 to 3 and kringle 5 inhibit endothelial cell growth,13,14 whereas the kringle 4 domain is a weaker inhibitor of cell growth but significantly blocks endothelial cell migration.13,15 Angiostatin-specific binding sites have been identified on endothelial cells, including ATP-synthase and angiomotin,16,17 and it has been reported that angiostatin induces apoptosis in endothelial cells,18 suggesting a potential mechanism of action. In support of this is a recent study that links the specific binding of kringles 1 to 5 of plasminogen to ATP-synthase on endothelial cells with induction of apoptosis and caspase activation.19
Recently, the combined kringle 1 to 2 domains of tPA were expressed in a bacterial and yeast expression system and were shown to inhibit bFGF-stimulated endothelial cell proliferation.20 However, the individual contributions of the kringle 1 and 2 domains of tPA on angiogenesis were not investigated in this study or the mechanism of action. In this report, we describe that the fibrinolytic drug, K2P, inhibits endothelial cell growth, migration, and in vivo angiogenesis. We show that inhibition is dependent on the kringle 2 domain and that K2P induces endothelial cell apoptosis in vivo.
Materials and Methods
Reteplase (domain deletion mutant of tPA comprising the kringle 2 and protease modules, otherwise known as Rapilysin or BM 06.022, but hereafter referred to as K2P) was purchased from Roche (Hertfordshire, UK). Alteplase, wild-type recombinant tPA (rtPA) was purchased from Boehringer Ingelheim. For further information and details of cell lines used, see online Methods, available at http://atvb.ahajournals.org.
Purification of tPA Kringle 2 Domain
K2P was incubated with porcine pancreatic elastase in a molar ratio of K2P:elastase of 3.5:1 overnight at room temperature with gentle shaking. The digest was extensively dialyzed against column binding buffer (50 mmol/L tris-HCl, 50 mmol/L NaCl, pH 7.7) before addition of 1 mmol/L final concentration DTT and incubated for 3 hours at 37°C. Thereafter, the digest was applied to a lysine-Sepharose column previously equilibrated with binding buffer, washed, and the adsorbed kringle 2 domain was eluted in binding buffer containing 0.2 mol/L ε-amino-n-caproic acid. Eluted kringle 2-containing fractions were pooled and dialyzed with binding buffer to remove ε-amino-n-caproic acid. To confirm purity, eluted kringle 2 was analyzed by SDS-PAGE, N-terminal sequence analysis, Western blot with polyclonal anti-tPA IgG #387 (American Diagnostica), and monoclonal anti–kringle 2 antibody 7VPA (Technoclone Inc) at 1:1000 and 1:100 dilutions, respectively. Correct conformation of kringle 2 was confirmed by nuclear magnetic resonance. Possible contaminating endotoxin was removed from K2 preparations with PROSEP-RemTox (Millipore) according to manufacturer’s instructions.
Inactivation of K2P
Twenty-five mU K2P was catalytically inactivated with 2 rounds of either 2 mmol/L PMSF or 100 μmol/L PPACK, each for 1 hour at 37°C in inactivation buffer (50 mmol/L tris-HCl, pH 7.4, 100 mmol/L NaCl, 100 mmol/L l-arginine, 5 mmol/L EDTA, and 0.01% tween 80), followed by extensive dialysis with the same buffer to remove excess inhibitor. To confirm inhibition of protease activity, 10 μL K2P were incubated with 50 μL Spectrozyme tPA chromogenic substrate (American Diagnostica) and 40 μL assay buffer (25 mmol/L tris-HCl, 200 mmol/L NaCl, pH 8.4) for 10 minutes at 37°C. Absorbance was measured at 405 nm.
In Vitro Cell Proliferation Assays
Cells (3×103) were seeded in 96-well plates and allowed to attach overnight. Cells were left in minimal growth medium (M199 with 2% FBS) for 24 hours. Thereafter, reagents were added to wells and incubated for 4 days. Cell number was quantified colorimetrically by the addition of 20 μL CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega) and measured at absorbance 492 nm.
Endothelial Cell Migration Assay
Endothelial cell migration was analyzed in a scratch wound assay as previously described with some modifications.21 Endothelial cells (1×106) were seeded onto gelatin-coated 3.5-cm dishes and left overnight. Media was aspirated the next day and cells were scratched with a P200 pipette tip along the diameter of the dish. Cells were washed twice with PBS before addition of fragments of tPA in M199 and low serum (5% FBS) and incubated at 37°C and 5% CO2. Wound diameters were photographed at t=0, t=24 hours, and t=48 hours. Wound closure was determined by measuring the diameter of wounds at 3 equidistant points along the length of wounds and was expressed as a percentage of wound diameter at the start of the experiment.
Mouse Sponge Angiogenesis Assay
C57 black mice were implanted with a sterile polyether sponge (15×5×5 mm) under the dorsal skin as previously described.22 PBS, bFGF, or test reagents were injected daily through the skin into the sponge in a volume of 100 μL for 20 days. Sponges were dissected on day 21 and fixed in 3.7% formaldehyde, embedded in paraffin, cut in sections of 4 μm, and stained with hematoxylin and eosin. Morphometric analysis of fibrovascular growth (microvessels and fibroblasts) into sponges was performed as previously described with some modifications.23 A grid graticule of 1-mm2 divisions with a total field area of 100 mm2 was used to determine the volume fraction of fibrovascular growth contained in sponges.
Dewaxed and rehydrated paraffin sections were processed for immunohistochemistry as previously described with some modifications.22 Sections were microwaved in antigen retrieval buffer (0.01 mol/L sodium citrate, pH 6) for 15 minutes before incubation with rat–anti-mouse CD34 IgG (Ab8158; Abcam) or anti–Ki-67 (Vector Labs). Sections were subsequently incubated with biotinylated anti-rat IgG (DAKO) and horseradish peroxidase-conjugated avidin–biotin complex (DAKO) before staining with DAB (Sigma) and counterstaining with 50% Mayer hematoxylin. The total number of vessels in CD34-stained sponges were counted at ×20 magnification in 4 representative fields per sponge and data are given as mean±SD.
Apoptosis was determined on dewaxed and rehydrated sections with an in situ cell death detection kit (Roche) by means of terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) staining according to manufacturer’s instructions. Briefly, sections were microwaved with 0.1 mol/L sodium citrate, pH 6, for 5 minutes before incubation with 0.1 mol/L tris-HCl, pH 7.5, containing 3% bovine serum albumin and 20% normal bovine serum for 30 minutes at room temperature. Slides were rinsed in PBS and incubated with 50 μL TUNEL reaction mixture for 1 hour at 37°C. Slides were rinsed with PBS, incubated with 50 μL converter POD for 30 minutes at 37°C, rinsed again, and processed with metal-enhanced DAB substrate (Sigma). Apoptotic cells displayed a strong blue/black nuclear stain and were counted at ×40 magnification in 3 representative fields per sponge.
Elastase Cleavage of K2P and Purification of Kringle 2
K2P is a bacterially expressed tPA deletion mutant that comprises residues 176 to 527 of native tPA with an additional 3 amino acids Ser-Tyr-Gln at the N terminus.6 To purify the kringle 2 domain, K2P was incubated with increasing concentrations of elastase for 24 hours at room temperature. SDS-PAGE revealed that elastase cleaved K2P into 2 major bands of ≈30 kDa and 10 kDa. Complete cleavage of K2P was obtained at a molar ratio of 10:1 K2P:elastase (Figure IA, available online at http://atvb.ahajournals.org). N-terminal sequence analysis revealed that elastase cleaved K2P at the Arg275–Ile276 bond, the activation cleavage site of single-chain tPA to 2-chain tPA. The N-terminus of the 30-kDa band was identified as IKGGLF corresponding to residues 276 to 281 of the protease domain of tPA and that of the 10-kDa band as SYQGN corresponding to the Ser-Tyr-Gln segment followed by Gly176 and Asn177 of kringle 2 of tPA. We purified this kringle 2 fragment (Ser-Tyr-Gln-Gly176-Arg275) by lysine-Sepharose affinity chromatography. SDS-PAGE revealed that the protease domain initially coeluted with the kringle 2 domain (Figure IB, lane 5), indicating that this domain was still attached to the kringle 2 module most likely by the Cys264-Cys395 disulphide bond. Western blot with an anti-elastase antibody showed that the eluted material was not contaminated with elastase (Figure IC, lane 5). Eluted material from lane 5 was subsequently treated with 1 mmol/L DTT to selectively reduce the Cys264-Cys395 disulphide bond for 3 hours at 37°C. The conditions of DTT treatment were obtained by carefully optimizing the concentration of DTT, kinetics, and temperature conditions to effectively separate the kringle 2 domain from the protease domain without compromising the conformational integrity of the kringle 2 module. Final separation of the kringle 2 domain from the protease domain was achieved after a further purification round on lysine-Sepharose. Purity of kringle 2 preparations was confirmed by Western blot with polyclonal anti-tPA IgG (Figure ID, lane 3) and with monoclonal anti-K2 IgG (data not shown).
Fragments of tPA Inhibit Endothelial Cell Proliferation
K2P was incubated with human umbilical vascular endothelial cells (HUVECs) for 4 days at increasing concentrations in the presence of bFGF. Figure 1A shows that K2P significantly decreased bFGF induced endothelial cell proliferation. Similar results were obtained with HDMEC and BAMEC (data not shown). This effect was specific for endothelial cells because no inhibition of cell growth was observed with vascular smooth muscle cells, normal fibroblasts (MRC-5), or the breast and prostate tumor cell lines MCF-7 and DU145, respectively (Figure 1D). To determine whether the inhibition of bFGF-induced endothelial cell proliferation was dependent on the proteolytic activity of K2P, it was catalytically inactivated so that it was no longer able to activate plasminogen. Catalytically inactivated K2P (K2Pi) also significantly inhibited bFGF-induced HUVEC growth in a similar manner to K2P (Figure 1B), indicating that the proteolytic activity of K2P was not responsible for the inhibition of cell growth observed. The inhibition of bFGF-induced cell growth mediated by K2P and K2Pi was comparable to that of kringles1 to 4 of plasminogen (angiostatin) (Figure 1C). Figure 1E shows the effects of full-length wild-type rtPA on HUVECs, MCF-7, and DU145 cells for comparison. Inhibition of bFGF-mediated HUVEC proliferation was observed with rtPA, but this was not specific for endothelial cells because a similar degree of inhibition was obtained with both MCF-7 and DU145 cells. This indicates there was no specific endothelial cell inhibition by wild-type tPA.
To determine whether the kringle domain of K2P was responsible for the inhibition observed, K2P was incubated with HUVECs in the presence of an anti–kringle 2 antibody. Figure 2A shows that the inhibitory activity of K2P was reversed in the presence of anti–kringle 2, indicating that the kringle 2 domain mediated the inhibition of HUVEC proliferation. This was confirmed with the purified kringle 2 domain obtained from elastase cleavage of K2P. Purified kringle 2 significantly inhibited both vascular endothelial growth factor (VEGF)-induced and bFGF-induced HUVEC growth (Figure 2B). Specificity of the kringle 2-mediated effects for endothelial cells was confirmed as no inhibition of K2 was observed with the nonendothelial cell types, human vascular smooth muscle cells, Bowes melanoma cell line, and the human fibrosarcoma cell line HT-1080 (Figure 2C).
Fragments of tPA Inhibit Endothelial Cell Migration
Endothelial cell migration was analyzed in a scratch wound assay. Figure 3A shows the effects of K2P on microvascular PMBEC migration. All wound diameters were similar after initial scraping of cells (t=0 hour). K2P inhibited cell migration in a time-dependent and concentration-dependent manner. Graphical representation is given in Figure 3B. At 24 hours and 48 hours after cell scraping, statistically significant inhibition of cell migration was observed with 2.5 μmol/L K2P. Figure 3C shows the effects of K2P, K2Pi, and K2 on VEGF-stimulated and bFGF-stimulated HUVEC migration. All fragments of tPA significantly inhibited HUVEC migration to a similar extent.
Fragments of tPA Inhibit Angiogenesis
After confirming that both human and mouse endothelial cell migration was inhibited in vitro by kringle 2-containing fragments of tPA, we analyzed their effects in a mouse sponge angiogenesis model in vivo. Purified kringle 2 at 0.1 μmol/L, 1 μmol/L, or 5 μmol/L was administered daily into sponges for a total of 20 days, either in the absence or presence of daily injections of bFGF. Control treated sponges received daily injections of PBS alone or daily injections of bFGF as a positive control. Specific CD34 vessel staining allowed quantification of the number of microvessels contained within sponge sections. The mean number of vessels±SD determined in 0.1 μmol/L, 1 μmol/L, or 5 μmol/L K2-treated sponges in the absence of bFGF was 7.5±4.7, 4.5±1.3, and 1.5±0.6, respectively, which was statistically significant (P<0.05, ANOVA, n=4). A similar trend was observed in the presence of bFGF, where the mean number of vessels±SD determined in 0.1 μmol/L, 1 μmol/L, or 5 μmol/L K2-treated sponges was 13.8±1.9, 5.3±1.5, and 1.5±1.0, respectively (P<0.01, ANOVA, n=4). Figure 4 shows representative hematoxylin and eosin–stained (left panel) and CD34-stained (right panel) sections of sponges treated with PBS, bFGF, or bFGF and 5 μmol/L K2. As well as a significant decrease in bFGF-induced angiogenesis in the sponge assay, it was also observed that the extent of invading mouse stromal tissue that had migrated into the sponge was also inhibited by K2P and purified kringle 2. The volume fraction of fibrovascular growth of K2P-treated sponges was consistently <50% of the total available growth area. Control sponges from mice treated with PBS or bFGF typically displayed volume fractions of fibrovascular growth in the region of 70% and 95%, respectively.
We wanted to identify potential mechanisms of action of K2P on angiogenesis inhibition. Kringle domains of plasminogen have been shown to induce apoptosis of endothelial cells.18 To identify whether K2P could also induce apoptosis in vivo, TUNEL staining was performed on K2P-treated sponges. Representative sections of sponge are given in Figure 5 for TUNEL (left panel) and Ki-67, which was used as a marker of proliferation (right panel). No difference in the number of Ki-67–stained cells was found between K2P-treated sponges and control treated; however, the mean number±SD of apoptotic cells was significantly greater in K2P-treated sponges (34±8) as compared with PBS-treated controls (17±4, P<0.05) (Figure II and Table I, available online at http://atvb.ahajournals.org). We also compared the effects of wild-type rtPA, both in the presence and absence of bFGF, and found no endothelial cell apoptosis in rtPA-treated sponges.
The presence of cryptic antiangiogenic domains located within larger molecules that do not possess antiangiogenic activity is an emerging theme in angiogenesis biology. In the hemostatic system of particular interest are the kringle domains that are present within a number of molecules that regulate coagulation and fibrinolysis.7 We were interested in whether the thrombolytic therapy drug, Reteplase, could also inhibit angiogenesis caused by the presence of the kringle 2 domain. Previous studies have shown that kringle domains do not possess antiangiogenic activity in the intact parent molecule, but only when isolated after enzyme cleavage or expressed as recombinant domains. However, the K2 domain is more exposed in this tPA mutant because of the deletion of the preceding finger and kringle 1 domains. We observed that this drug had very significant effects on endothelial cell proliferation, migration, and angiogenesis. We found that catalytically active K2P and inactivated K2P inhibited bFGF-induced endothelial cell growth to a similar extent. This indicated that inhibition of endothelial cell proliferation was unlikely to be caused by plasminogen activation by K2P, but rather caused by a region located within the K2 domain. This was confirmed after elastase cleavage of K2P and purification of the K2 domain. K2 dose-dependently inhibited both VEGF and bFGF endothelial cell growth. To our knowledge, elastase cleavage of tPA at the Arg275–Ile276 bond has not been reported previously. However, this enzyme has been used extensively to produce fragments of plasminogen14 and to generate angiostatin.12 Our data show that elastase can also be used for limited proteolysis of K2P. Reduction of the Cys264–Cys395 disulphide bond was necessary to completely separate the protease domain from the kringle 2 domain. Other studies have shown that selective reduction of disulphide bonds is possible with limited use of DTT.24 We found that 1 mmol/L DTT was sufficient to selectively reduce the Cys264–Cys395 disulphide bond without compromising the conformational state of the kringle 2 module, as evidenced by the fact that K2 was still able to bind lysine-Sepharose after DTT treatment.
Human and mouse endothelial cell migration was significantly reduced in a scratch wound assay after treatment with K2 containing tPA fragments, as was in vivo angiogenesis in a mouse sponge angiogenesis assay. Of particular interest was the very significant reduction of mouse stromal fibroblasts that had invaded into sponges after tPA fragment treatment. We found no effects of K2P on normal fibroblast proliferation in vitro or Ki-67 proliferating cells in vivo indicating that the reduction of total fibrovascular growth in sponges was not caused by toxicity, but most likely caused by the significant loss of vascularity in the sponges. In support of this are previous studies using disc25 and sponge23 angiogenesis models, which have shown that fibrovascular growth in these models is directly proportional to vessel growth, suggesting a dependency of fibroblast density on the number of blood vessels. In addition, we have also demonstrated increased overall apoptosis in the K2P-treated sponges, both in the absence and presence of bFGF, suggesting a potential mechanism of action of K2P. Although we observed inhibition by wild-type tPA in our assays, this was not specific for endothelial cells and may have been caused by other mechanisms, such as proteolytic degradation. That different mechanisms were in operation is supported by the fact that we found no endothelial cell apoptosis with wild-type tPA in the sponge assay. This indicates that the antiangiogenic activity is differentially available in K2P. Figure III (available online at http://atvb.ahajournals.org) compares the amino acid sequence of the K2 domain of tPA with kringle domains of plasminogen, uPA, and with the kringle 1 domain of tPA. K2 of tPA has the highest identity with K1 of tPA (53%) and with the kringle domain of uPA (47%). Considerable sequence identity is also observed with the kringle domains of plasminogen (37% to 43%). Despite these similarities, a common structural motif that might confer the antiangiogenic properties of kringle domains has not been determined. However, it cannot be excluded that the antiangiogenic activity is caused by the secondary structure of these domains in their interaction with the endothelial cell surface.
tPA fibrinolytic drugs are used to treat acute myocardial infarction, pulmonary embolism, and stroke.5 However, complications of fibrinolytic therapy include bleeding, of which intracranial hemorrhage is the most serious. K2P has a reduced plasma clearance and prolonged half-life compared with full-length tPA as the finger, epidermal growth factor-like domain, and the kringle 1 domain, which confer recognition by clearance receptors, have been deleted. Loss of the finger domain also results in decreased fibrin specificity of K2P that can lead to more extensive systemic plasminogen activation and bleeding. Our data suggest that inhibition of endothelial cell growth and migration of cells into areas of damaged endothelium may contribute to bleeding complications observed during thrombolytic therapy with Reteplase. In addition, our findings suggest that the kringle 2 domain of tPA may have use as an angiogenic inhibitor, which we are currently evaluating as a therapeutic target.
Funds from Cancer Research UK and charitable donations supported this work. Our thanks also go to Dr Nick Totty for N-terminal sequence analysis, to Chris Selkirk for endotoxin assays, and to George Elia for hematoxylin and eosin staining.
Current affiliation for V.A.C.: Cell Growth Regulation and Angiogenesis Laboratory, Cancer Research UK Centre for Cancer Therapeutics, The Institute of Cancer Research, Surrey, UK.
- Received April 9, 2004.
- Accepted January 19, 2005.
Carmeliet P, Collen D. Molecular analysis of blood vessel formation and disease. Am J Physiol. 1997; 273: H2091–H2104.
Larsson A, Astedt B. Immunohistochemical localisation of tissue plasminogen activator and urokinase in the vessel wall. J Clin Pathol. 1985; 38: 140–145.
Paciucci R, Tora M, Diaz VM, Real FX. The plasminogen activator system in pancreas cancer: role of t-PA in the invasive potential in vitro. Oncogene. 1988; 16: 625–633.
Cao R, Wu H-L, Veitonmaki N, Linden P, Farnebo J, Shi G-Y, Cao Y. Suppression of angiogenesis and tumor growth by the inhibitor K1–5 generated by plasmin-mediated proteolysis. Proc Natl Acad Sci U S A. 1999; 96: 5728–5733.
Kuba K, Matsumoto K, Date K, Shimura H, Tanaka M, Nakamura T. HGF/NK4, a four-kringle antagonist of hepatocyte growth factor, is an angiogenesis inhibitor that suppresses tumor growth and metastasis in mice. Cancer Res. 2000; 60: 6737–6743.
Lee T-H, Rhim T, Kim SS. Prothrombin kringle-2 domain has a growth inhibitory activity against basic fibroblast growth factor-stimulated capillary endothelial cells. J Biol Chem. 1998; 273: 28805–28812.
Kim KS, Hong Y-K, Joe YA, Lee Y, Shin J-Y, Park H-E, Lee I-H, Lee SY, Kang D-K, Chang S-I, Chung SI. Anti-angiogenic activity of the recombinant kringle domain of urokinase and its specific entry into endothelial cells. J Biol Chem. 2003; 278: 11449–11456.
Cao Y, Ji RW, Davidson D, Schaller J, Marti D, Sohndel S, McCance SG, O’Reilly MS, Llinas M, Folkman J. Kringle domains of human angiostatin. J Biol Chem. 1996; 271: 29461–29467.
Cao Y, Chen A, An SSA, Ji RW, Davidson D, Cao Y, Llinas M. Kringle 5 of plasminogen is a novel inhibitor of endothelial cell growth. J Biol Chem. 1997; 272: 22924–22928.
Ji W-R, Castellino FJ, Chang Y, DeFord ME, Gray H, Villarreal X, Eghtedarzadeh Kondri M, Marti DN, Llinas M, Schaller J, Kramer RA, Trail PA. Characterization of kringle domains of angiostatin as antagonists of endothelial cell migration, an important process in angiogenesis. FASEB J. 1998; 12: 1731–1738.
Moser TL, Stack MS, Asplin I, Enghild JJ, Hojrup P, Everitt L, Hubchak S, Schnaper HW, Pizzo SV. Angiostatin binds ATP synthase on the surface of human endothelial cells. Proc Natl Acad Sci U S A. 1999; 96: 2811–2816.
Troyanovsky B, Levchenko T, Mansson G, Matvijenko O, Holmgren L. Angiomotin: an angiostatin binding protein that regulates endothelial cell migration and tube formation. J Cell Biol. 2001; 152: 1247–1254.
Lucas R, Holmgren L, Garcia I, Jimenez B, Mandriota SJ, Borlat F, Sim BKL, Wu Z, Grau GE, Shing Y, Soff GA, Bouck N, Pepper MS. Multiple forms of angiostatin induce apoptosis in endothelial cells. Blood. 1998; 92: 4730–4741.
Veitonmaki N, Cao R, Wu LH, Moser TL, Li B, Pizzo SV, Zhivotovsky B, Cao Y. Endothelial cell surface ATP synthase-triggered caspase-apoptotic pathway is essential for k1–5-induced antiangiogenesis. Cancer Res. 2004; 64: 3679–3686.
Cattaruzza S, Schiappacassi M, Ljungberg-Rose A, Spessotto P, Perissinotto D, Morgelin M, Mucignat MT, Colombatti A, Perris R. Distribution of PG-M/versican variants in human tissues and de novo expression of isoform V3 upon endothelial cell activation, migration, and neoangiogenesis in vitro. J Biol Chem. 2002; 277: 47626–47635.
Hague S, MacKenzie IZ, Bicknell R, Rees MCP. In-vivo angiogenesis and progestogens. Human Reprod. 2002; 17: 786–793.