Purpose: Endostatin, a peptide derived from proteolysis of collagen XVIII, is an endogenous inhibitor of angiogenesis and tumor growth. We have synthesized five peptide fragments designed to cover the whole length of the endostatin molecule (containing 40–50 amino acids each) with the aim of exploring the possibility that a specific sequence within the molecule might be responsible for its antiangiogenic effects.
Experimental Design: The five peptide sequences, termed fragments I, II, III, IV, and IVox, the latter bearing the original disulfide bond Cys135-Cys165, were investigated for their effects on cultured endothelial cells, on enzyme activities related to angiogenesis, on tube formation in three-dimensional gel matrices, in vivo in the avascular rabbit cornea assay, and in an experimental tumor burden paradigm.
Results: Both the fragment covering the COOH-terminal endostatin region, fragment IV, and particularly fragment IVox, retained the angioinhibitory effects of endostatin. Fragment IVox strongly inhibited endothelial cell migration and proliferation, in vitro tube formation, and in vivo angiogenesis, displaying a potency comparable with that of endostatin. When tested in vivo on tumor growth, fragment IVox demonstrated to be more effective than full-length endostatin. Unexpectedly, fragment III exhibited proangiogenic activity, increasing endothelial cell migration, producing neovascularization to an extent similar to that of vascular endothelial growth factor, and enhancing the angiogenic response to vascular endothelial growth factor in the cornea assay. Peptides encompassing the NH2-terminal region of endostatin (fragments I and II) had negligible effects on angiogenesis.
Conclusions: In view of these results, which show strikingly distinct profiles of endostatin fragments, we propose that the amino acid sequence of endostatin contains both angiosuppressive and angiostimulatory domains.
Angiogenesis, the outgrowth of new capillaries from preexisting vessels, is essential for physiological and pathological conditions, including tumor growth and metastasis (1, 2, 3). Angiogenesis is a complex multistep process that includes the proliferation, migration and differentiation of endothelial cells, degradation of the extracellular matrix, microtubule formation, as well as sprouting of new capillary branches (4). The complexity of the angiogenic processes suggests the existence of multiple controls of the system, which can be transiently switched on and off.
A switch in the angiogenic phenotype of tissues is thought to depend on a local change of the balance between angiogenic stimulators and inhibitors (5). Different families of regulators of angiogenesis have been discovered. Among angiogenic factors, VEGF2 /vascular permeability factor and FGF-2 are the best characterized positive regulators (6, 7). A number of endogenous inhibitors of angiogenesis such as angiostatin, endostatin (8, 9), and thrombospondin (10) are generated by tumors, whereas others such as platelet factor-4 and IFN-inducible protein-10 are not associated with the presence of tumor (11).
Endostatin is a Mr 22,000 COOH-terminal fragment of collagen XVIII that specifically inhibits growth factor-induced endothelial proliferation in vitro and potently inhibits angiogenesis and tumor growth in vivo (9, 12). Both mouse and human endostatin originating from various sources (e.g., yeast Picha pastoris, Escherichia coli expression system, and 293-EBNA human embryonic kidney cells) were shown to possess antiangiogenic activity in endothelial cells stimulated to proliferate and migrate in response to FGF-2 and VEGF. However, large differences in potency were noted in these products. For example, the endostatin produced in P. pastoris was active at μg/ml concentrations, whereas for the human endostatin produced by E. coli and 293-EBNA cells, inhibitory concentrations were 1,000–100,000-fold lower (13, 14). An important finding, emerging from the research on mutant forms of endostatin, is that the antiangiogenic activity seems to be independent of zinc binding (14). Recently, independent groups reported either lack of efficacy of endostatin gene therapy in different tumors or failed to replicate the original findings of tumor growth inhibition with the recombinant protein (15). The transfer of a retroviral vector encoding a secretable form of murine endostatin into emopoietic stem cells gave no inhibition of neoangiogenesis or tumor growth of primary or metastatic tumors (16). Similar results were obtained using a NOD/SCID model of human leukemia (17).
On the basis of these findings, the aim of this work was to study of relationships between the chemical structure and antiangiogenic activity of endostatin with the purpose of determining the presence within the molecule of one or more active domains as described also for other endogenous inhibitors (18). To this end, we have synthesized five peptide fragments containing between 40 and 50 amino acid residues to cover the whole mouse endostatin sequence.
The study of the biological effects of the endostatin fragments was designed as a strategy involving first the evaluation of in vivo angiogenesis to assess the activity of sequences on an integrated system, which provides a measure of their efficacy. The sequences found active in vivo were then investigated on basic functions (migration and proliferation) of cultured postcapillary venular endothelial cells, on enzyme activities implicated in angiogenesis, and in organ culture of explants from small diameter vessels to explore their biochemical and molecular mechanisms of action. Finally, the fragment found to be the most effective in inhibiting angiogenesis was tested for its ability to reduce tumor growth in nude mice.
MATERIALS AND METHODS
Synthesis of Endostatin Peptides
Endostatin fragments were synthesized corresponding to the following sequences, starting at His132 in the NC1 domain as described in Ref. 19: fragment I, sequence 1–39; fragment II, sequence 40–89; fragment III, sequence 90–134; and fragment IV, sequence 135–184. In addition, fragment IVox is the peptide containing the original disulfide bond at Cys135-Cys165.
Endostatin fragments were synthesized by the solid phase method. The synthesis was carried out on an automated peptide synthesizer (Biolynx plus, mod 4170; Novabiochem, Nottingham, United Kingdom), using continuous flow techniques, Fmoc strategy and polyethylene-oxide-polystyrene resins (20, 21). The functional groups of the lateral chains of amino acids were protected as follows: Arg (2,2,4,6,7-pentamethyl-dihydro-benzofuran-5-sulfonyl, Pbf), Ser, Thr, and Tyr (tert-butyl, tBu), Asp and Glu (OtBu), Lys and Trp (tert-butoxycarbonyl, Boc), His (trityl, Trt), Cys33 and Cys173 by tBu; and Cys135 and Cys165 by Trt.
A total of 0.1 mmol of Fmoc-amino acid-NovaSyn TGA resin was submitted to the following treatments for each amino acid: (a) DMF washing; (b) Fmoc deblocking with 10% piperidine in DMF; (c) DMF washing; (d) coupling in DMF with appropriate Fmoc-amino acid (five equivalents) and HOBt (five equivalents). The greater part of Fmoc-amino acids was coupled using Pfp ester in the presence of anionic dye Novachrome for counter-ion distribution monitoring; only the COOH groups of Fmoc-Arg (Pbf) and Fmoc-His (Trt) were activated by benzotriazol-1-yl-oxy-tris-pirrolidino-phosphonium hexafluorophosphate (PyBop) (reaction time 4 h); and (e) DMF washing.
After the completion of chain assembly, the synthetic peptide-resin was washed sequentially with DMF, t-amyl alcohol, acetic acid, t-amyl alcohol, dichloromethane, diethyl ether, then peptide-resin was dried in vacuum.
Deprotection and cleavage of peptides from resin was performed with a mixture of anhydrous TFA (80%), water (5%), thioanisole (5%), ethanedithiol (2.5%), and phenol 2.5%. The mixture was reacted for 2 h at room temperature with occasional stirring. After filtration under vacuum, the resin was washed twice with TFA. Dry ethyl ether was added slowly at the filtrate, with ice-cooling and under stirring to precipitate the polypeptide. The product was filtered, repeatedly washed with dry ethyl ether, and finally dried under vacuum over KOH.
For disulfide bond formation between Cys135 and Cys165, 22 mg of iodine in 75% methanol (80 ml) were added dropwise with stirring to a solution of 335 mg of fragment IV in 300 ml of 75% methanol. The reaction mixture was left to react for 3 h. The iodine excess was removed with a 10% solution of ascorbic acid, methanol was evaporated in vacuum, and the aqueous solution was lyophilized.
The crude polypeptides were purified by preparative reverse-phase high-performance liquid chromatography using a Aktabasic 100 instrument (Amersham Pharmacia Biotech Europe, Freiburg, Germany) and Jupiter column (250 × 21.2 mm) 15 μ, C 18, 300 Å (Phenomenex, Torrance, CA). The polypeptides were dissolved in solvent A (0.1% TFA +10% acetonitrile) and eluted with a gradient of solvent B (90% acetonitrile +10% of 0.1% TFA). The flow rate was of 20 ml/min and detection at 226, 254, and 280 nm. The main fractions were pooled and lyophilized.
The purified polypeptides were characterized by amino acids analysis, analytical high-performance liquid chromatography (Jupiter column, 250 × 10 mm, 10 μ, C 18, 300 Å), and mass spectrometry using a Termo Finnigan instrument mod. LCQ Deca XP.
All of the peptides were water soluble. Stability of solubilized solutions (200 μg/ml) kept at −20°C was maintained over a period of 2 years, giving overlapping results both in vitro and in vivo. Fragments were chemically stable for at least 15 days when solubilized in sterile water (200 μg/ml) and kept at 37°C.
Cell Lines and Culture Conditions
The CVECs were isolated and characterized as described previously (22). Cells were maintained in culture in DMEM supplemented with 10% bovine CS and antibiotics (100 units/ml penicillin and 100 μg/ml streptomycin) on gelatin-coated dishes. Cells were cloned, and each clone was subcultured up to a maximum of 25 passages. Passages between 15 and 20 were used in these experiments.
BALB/c MAE 22106 cells (23) were grown in DMEM added with 10% FCS. HUVECs were used as primary cultures and grown in M199 supplemented with 10% FCS.
Endothelial cells were characterized by labeling with antibody to factor VIII-related antigen and by uptake of acetylated low-density lipoproteins. At confluence, they form a typical contact-inhibited monolayer with the characteristic cobblestone morphology.
BASMs were freshly isolated form bovine aortas by mincing and collagenase digestion (24). BASMs were grown in DMEM supplemented with 10% CS and used up to passage 6.
The fibroblast NIH-3T3 cells were obtained from American Type Culture Collection (Manassas, VA). NIH-3T3 cells were grown in DMEM in addition with 10% FCS.
The human epidermoid carcinoma A-431 cells were obtained from American Type Culture Collection. A-431 cells were maintained in culture in DMEM supplemented with 4500 g/liter glucose and 10% FCS. Cells were split 1:5 twice a week.
The cytotoxic effect of peptides was studied by trypan blue exclusion. Briefly, endothelial and nonendothelial cells were suspended in 0.1% serum DMEM supplemented with 10 or 300 ng/ml peptides and incubated at 37°C for 4 h. Cells were then counted in a hemocytometer, and the percentage of dead cells over the total number of cells was calculated.
Cell migration was assessed in 48-well microchemotaxis chambers (NeuroProbe; Biomap, Milan, Italy) on a polycarbonate filter, 8-μm pore size. The filter was coated either with type I collagen (100 μg/ml) and bovine plasma fibronectin (10 μg/ml). A cell suspension containing 12,500 cells was added to the upper chamber of each well and substances to be tested to the lower one. Endostatin fragments were tested in the presence of angiogenic factors in the lower wells. The chamber was incubated at 37°C for 4 h. At the end of incubation, cells that did not migrate were removed, and the filter was stained with Diff Quik. Migrated cells were counted in 10 random fields/well at a magnification of 400 (25).
CVECs were evaluated for their ability to adhere to a fibronectin coating in the presence of endostatin peptides. A 96-well microtiter plate was coated with 10 μg/ml fibronectin, washed with PBS, and incubated with 1% BSA to remove a specific binding. A cell suspension containing 5 × 104 CVECs in 0.1% CS DMEM supplemented with 10 ng/ml of the peptides was prepared. One hundred μl of the cell suspension were added to each well and incubated for 90 min at 37°C. The plate was then washed three times with PBS, and cells were fixed with methanol and stained with Diff Quik. The total number of adherent cells was counted with the aid of an ocular grid (0.21 mm2; Ref. 25). Each experiment was run in triplicate.
CVECs were cultured in 96-well cell culture plates in 10% CS medium until 90% confluence. Cells were washed with and incubated in serum-free medium. After 24 h incubation with the stimuli, the conditioned medium was collected, clarified by centrifugation, and assayed for zymography (18). Conditioned media were subjected to electrophoresis in 8% SDS-PAGE containing 1 mg/ml gelatin under nondenaturing conditions. After electrophoresis, the gels were washed with 2.5% Triton X-100 to remove SDS and incubated for 48 h at 37°C in 50 mm Tris buffer containing 200 mm NaCl and 20 mm CaCl2 (pH 7.4). The gels were stained with 0.5% Coomassie brilliant blue R-250 in 10% acetic acid and 45% methanol and destained with 10% acetic acid and 45% methanol. Bands of gelatinase activity appeared as transparent areas against a blue background. Gelatinase activity was then evaluated by quantitative densitometry.
Cell proliferation was quantified by total cell number as reported previously (25). Briefly, 1.5 × 103 cells resuspended in 10% CS were seeded in each well of 96-multiwell plates. After adherence (4 h), the medium was replaced with 1% serum DMEM containing increasing concentrations of the peptides in the absence and in the presence of the angiogenic factors VEGF and FGF-2 for endothelial cells and in the presence of 5% FCS for nonendothelial cells and incubated for 48 h. The supernatants were removed from the multiwell plates, and cells were fixed with methanol and stained with Diff-Quik. Cell numbers were obtained by counting in seven random fields at a magnification of 100 with the aid of an ocular grid.
Tube Formation from Vessel Rings
Vessel sprouting form mouse aorta was evaluated as described previously (26).
Vessel Ring Preparation.
Mouse aortas were isolated in sterile conditions, and perivascular fibrous tissue was removed under a dissecting microscope. Rings 1–2 mm long were produced and rinsed in serum-free medium (M199; Sigma Chemical Co.). Vessel fragments were then included in a fibrin gel obtained by adding 400 μl of a bovine fibrinogen solution (3 mg/ml in M199 medium; Sigma Chemical Co.) into each well of 48-multiwell plates. Once vessel rings were positioned in the well with the lumen oriented horizontally in the center of the solution, gelation of the fibrinogen was induced with bovine thrombin (1.5 units/ml; Sigma Chemical Co.). After 20 min, 400 μl of M199 medium was added with antibiotics (100 units/ml penicillin and 100 μg/ml streptomycin), amphotericin B (0.25 μg/ml), and 10% FCS were added to each well. After 24 h, the medium was removed, the gels washed, and substances were added to the medium supplemented with 1% FCS. The organ culture was kept at 37°C, 5% CO2. Each experimental point was run in triplicate.
Quantitative evaluation of newly formed structures was carried out at day 3. Gels were observed with an inverted microscope at a magnification of 200. The area occupied by tubules was measured using an ocular grid (0.21 mm2). The grid was placed over the tubules as many times as necessary to cover the whole area. Each time, the portion of the grid actually corresponding to tubules was recorded. The area covered by tubules in each well was recorded and expressed in microscopic units (0.21 mm2 ).
Cell populations present in tubular structures were identified by immunohistochemistry and quantified at the end of the experiment. The medium was removed, and the vessel rings were extracted from the gel. Tubular structures were retained in the gel, and fibrin was lysed by adding 1 units/ml bovine plasmin. The cell suspension obtained was centrifuged and spread on a slide. Cells were fixed in acetone/chloroform 1:1 and stained with an amplified immunoperoxidase technique. The polyclonal antibody antifactor VIII was used to stain endothelial cells. The slides were observed at a magnification of 1000, and the number of factor VIII positive cells/field was recorded.
Akt Kinase Assay
CVECs were grown to 90% confluence and serum starved for 24 h. Stimulation with angiogenic factors was carried out for 15 min. Where indicated, cells were pretreated with peptides for 30 min. Akt activity was measured in cell lysates by the use of Akt1/protein kinase B immunoprecipitation kinase assay kit (Upstate Biotechnology). Data are expressed as cpm/sample.
Angiogenesis in Vivo: Rabbit Cornea Assay
Angiogenesis was studied in the cornea of albino rabbits because this is an avascular and transparent tissue where inflammatory reactions and growing capillaries can be easily monitored and changes quantitated by stereomicroscopic examination (18, 25). Experiments have been performed in accordance with the guidelines of the European Economic Community for Animal Care and Welfare (EEC Law No. 86/609).
Slow Release Pellets Preparation and Surgical Procedure.
Slow-release pellets were prepared under sterile conditions, incorporating the test substances into a casting solution of an ethynil-vinyl copolymer (Elvax-40; DuPont-De Nemours, Wilmington, DE). In the lower half of New Zealand White rabbit eye (Charles River; Calco, Como, Italy), anesthetized by sodium pentothal (30 mg/kg), a micropocket was surgically produced using a pliable iris spatula. The pellets were implanted in the micropockets.
Endostatin and its partial sequences were tested per se to evaluate their angiogenic potential. In experiments aimed to evaluate their interference on angiogenesis inducers, VEGF (200 ng/pellet) and FGF-2 (200 ng/pellet) and the endostatin peptides were incorporated in the same slow release preparation.
Quantification of Corneal Angiogenesis.
Daily observations of the implants were made with a slit lamp stereomicroscope by two independent operators in a blinded manner. An angiogenic response was considered as positive when budding of vessels from the limbal plexus occurred after 3–4 days and capillaries progressed to reach the implanted pellet according to the scheme reported previously (25). Angiogenic activity was expressed as the number of implants exhibiting neovascularization over the total implants studied. Potency was scored by the number of newly formed vessels and by their growth rate. Data are expressed as angiogenic score calculated as (number of vessels × distance from the limbus).
Corneas were removed at different times (2, 5, and 10 days) and at the end of the experiment, fixed in formalin, and processed by routine histopathological procedures. Sections 5-μm thick were cut next to each pellet and stained with H&E.
In Vivo Tumor Growth
To determine the in vivo antitumor activity of fragment IVox and full-length endostatin, female immunodeficient mice (5–8-week-old BALB/c nude mice; Harlan Nossan, Milan, Italy) were s.c. inoculated in the right flank with 8 × 106 A-431 cells in a volume of 50 μl. After 1 week, when tumors reached a volume of 100 mm3, animals were randomly assigned to three different experimental protocols (five mice/group). Peritumor treatment with equimolar amounts of fragment IVox (62.5 μg/kg/day), full-length endostatin (250 μg/kg/day; Ref. 14), or PBS started. Daily treatment was performed for 7 days. Serial caliper measurements of perpendicular diameters were used to calculate tumor volume using the following formula: shortest diameter × longest diameter × thickness of the tumor in mm. Data are reported as tumor volume in mm3. Experiments have been performed in accordance with the guidelines of the European Economic Community for Animal Care and Welfare and the National Ethical Committee. Animals were observed daily for signs of cytotoxicity and were sacrificed by CO2 asphyxiation.
Results are expressed as mean values ± SE. Multiple comparisons were performed by one-way ANOVA and individual differences tested by Fisher’s test after the demonstration of significant intergroup differences by ANOVA. Differences with P < 0.05 were considered significant.
Fmoc amino acids, Fmoc-amino acids-NovaSyn TGA resin and dye Novachrome were purchased form Calbiochem AG (Laufelfingen, Switzerland). Cell culture reagents and materials were from Sigma Chemical Co. (St. Louis, MO). FCS and CS were from Hyclone (Logan, UT). Human recombinant VEGF, FGF-2, and mouse recombinant (P. pastoris) full-length endostatin were from Calbiochem-Novabiochem Int. (San Diego, CA). Acrylamide, N,N,N′,N′-tetramethylethylenediamine, ammonium persulfate, and Coomassie brilliant blue were from Bio-Rad Laboratories (Hercules, CA). Diff-Quik was from Mertz-Dade AG (Dudingen, Switzerland).
Synthetic Fragments of Endostatin.
The amino acid sequences of fragments derived from mouse endostatin are reported in Fig. 1, and their analytical properties are reported in the legend. The knowledge of tertiary structure (19) was of aid to identify the typology and size of fragments to synthesize. Fragment I possesses an α-helix segment of 14 amino acid residues corresponding to sequence 26–39, where the Phe31 and Phe34 residues are located with side chains fully exposed to solvent. Fragments II and III contain 10 of the 15 arginine residues, almost all clustered on one face of molecule. Fragment IVox contains the disulfide bridge Cys135-Cys165, which circularizes a twisted loop between two β-sheets. The cyclic nature of this compound imposes same conformational restriction to the flexibility of the molecule.
The physicochemical properties of the synthesized polypeptides are as follows: fragment I (sequence 1–39), [α]20D −43.0°(c = 0.5, water); mass spectrum, molecular peak (M +1) = 4376 Da; fragment II (sequence 40–89), [α]20D −12.5°(c = 0.04, water); mass spectrum, molecular peak (M +1) = 5514 Da; fragment III (sequence 90–134), [α]20D −60.5°(c = 0.06, water); mass spectrum, molecular peak (M +1) = 5125 Da; fragment IV (sequence 135–184), [α]20D −14.5°(c = 0.2, acetic acid 80%); mass spectrum, molecular peak (M +1) = 5504 Da; and fragment IVox (sequence 135–184), [α]20D −14.5°(c = 0.2, acetic acid 80%); mass spectrum, molecular peak (M +1) = 5502 Da.
Endostatin Fragments Affect in Vivo Angiogenesis.
When pellets containing 200 ng of each endostatin fragment were implanted in the avascular rabbit cornea to study their effects on vessel growth, two distinct patterns of activity were seen. Fragments I, II, IV, and IVox produced no angiogenic effects (Fig. 2,A). In contrast, fragment III induced a strong and sustained angiogenic activity (three positive implants of four performed, angiogenic score of ∼9 at 14 days), not accompanied by inflammatory cell infiltrate. Inspection of Fig. 2 and comparison of Fig 2, A and B, indicates that the response to fragment III has a similar temporal pattern and is of a similar magnitude (angiogenic score) to that elicited by VEGF (200 ng/pellet).
In experiments in which fragments were coreleased in the same cornea with VEGF (200 ng/pellet), fragment IVox produced a strong inhibition of neovascularization, which judged from the angiogenic score at 14 days was ∼75% (Fig. 2,B, see also E and G). Fragment IV reduced VEGF-induced angiogenesis (four positive implants of five performed), whereas fragment II was ineffective (Fig. 2,B). In sharp contrast, corelease of fragment III and VEGF in the same cornea elicited a very fast and pronounced angiogenic response, resulting from the sum of the effects observed (four positive implants of four performed, angiogenic score of ∼13 at 10 days; Fig. 2, B, E, and F). Fragment I when challenged with either VEGF or FGF-2 produced an inflammatory cell infiltrate detectable both at macroscopic and histological level in 50% of the implants, and therefore, data were not further taken in consideration.
Fragment IVox and fragment IV also completely abolished FGF-2 (200 ng/pellet) induced neovascularization (zero positive implants of three performed in both experimental conditions; Fig. 2,D). Fragment III (200 ng/pellet) did not modify FGF-2 induced angiogenesis when the peptides were coreleased (Fig. 2 D).
Mouse endostatin was tested at two doses (200 or 800 ng/pellet). The native molecule, which exhibited a weak angiogenic activity per se, inhibited in a dose-dependent manner VEGF-induced neovascularization (Fig. 2 C, only 800 ng shown). Fragment III retained its angiogenic activity even in the presence of excess full-length endostatin, suggesting an independent mechanism in the control of angiogenesis (data not shown).
In view of these results, showing that influence on angiogenesis is confined to fragment III, IV, and IVox, investigations with all five fragments on cell functions and on molecular mechanisms were performed but only for the above fragments are reported (see below).
Endostatin Fragments Do Not Have Cytotoxic Effects.
The cytotoxic effect of each peptide on different endothelial and nonendothelial cell lines was evaluated in cell suspensions after 4 h incubation in the presence of 10 or 300 ng/ml of the peptides. None of the peptides produced increase of cell death either at high (300 ng/ml; Table 1) or low concentration (10 ng/ml; data not shown) in all of the cell lines examined after 4 h (Table 1).
Effects of Endostatin and Fragments on Endothelial Cell Migration and Proliferation.
As expected, mouse full-length endostatin reduced migration and proliferation promoted by angiogenic factors in a very consistent manner, without any effect on quiescent cells (Fig. 3 A). The effect of endostatin fragments was then compared with the native molecule.
The effect of a gradient of peptides on cell migration in quiescence and toward a gradient of the angiogenic factors FGF-2 and VEGF (both at 20 ng/ml) was investigated in CVECs. Fragments (only fragments III, IV, and IVox reported) did not modify migration of quiescent cells, except fragment IV, which induced cell migration at 10 ng/ml. In cells stimulated to migrate by FGF-2 and VEGF, the effect exerted by fragments was mainly inhibitory, being particularly marked for fragment IVox, which inhibited migration at all concentrations examined toward FGF-2 (Fig. 3, C and D). In contrast, fragment III had no inhibitory effect on stimulated migration (Fig. 3 B).
The effect exerted by fragments on proliferation of quiescent cells was rather variable, only fragment III being capable to significantly increase proliferation at all of the concentrations tested (P < 0.05). In cells stimulated to proliferate by FGF-2 and VEGF, fragment IVox produced inhibition at all doses examined against both factors, showing a dose-dependent effect toward FGF-2 (IC50 = 8 ng/ml, ∼1.5 nm; Fig. 3,D). Fragment IV showed inhibition only against FGF-2 at the highest concentration (Fig. 3,C). Again, fragment III exerted no influence on proliferation activated by growth factors (Fig. 3 B).
When mouse endostatin (Mr 22,000) and fragment IVox (Mr 5,500) are compared on a molar basis for their effect on proliferation, it appears that the parent molecule is approximately four times as potent as the fragment (Fig. 3, A and D). It is worth to note that in all migration and proliferation experiments the molar excess of fragments relative to FGF-2 and VEGF ranged from a value of 2 to >50 (10 and 300 ng/ml, respectively).
The results obtained in migration and proliferation experiments indicated that only fragment IVox retained the inhibitory effects of the parent molecule, and therefore, its activity was additionally investigated.
The selectivity of fragment IVox was evaluated by studying proliferation in endothelial cells different from CVECs, and in nonendothelial cell lines. Proliferation of MAEs (Fig. 4,A) and primary HUVECs (data not shown) stimulated by FGF-2 or VEGF was strongly inhibited by fragment IVox (300 ng/ml). However, this peptide had no effect on proliferation of nonendothelial cell lines such as BASM, NIH-3T3, and A-431 cells stimulated to grow by incubation in 1 or 5% FCS (Fig. 4 B, data shown for A-431 only). Thus, the action of fragment IVox seems to be selectively addressed to endothelial cells.
Adhesion of CVEC to polystyrene plates coated with fibronectin, either in presence or absence of FGF-2 and VEGF, was not modified either by the fragments or by mouse endostatin (data not shown).
Inhibition of Endothelial Cell Sprouting.
The formation of capillary-like structures from mouse aortic fragments cultured in three-dimensional fibrin gels was evaluated after 3 days of incubation. Cell composition of the newly formed structures, characterized by immunohistochemistry, showed that >90% of the cells stained positive for the endothelial marker factor VIII (data not shown). The presence of fragment IVox (100 ng/ml) reduced tubule growth induced by FGF-2 and VEGF, without affecting spontaneous budding (Fig. 5, A–C). The most significant inhibition was observed in response to VEGF (67% inhibition, compared with 50% inhibition in the presence of FGF-2). Fragment IVox was able to reduce the number of endothelial cells stimulated by either one of the growth factors (95% inhibition, n = 3).
Effect of Fragments on Serine-Threonine Kinase (Akt) and MMP.
Among the molecular mechanisms implicated in angiogenesis, Akt and matrix MMP (MMP-2) activities were selected as signals of endothelial cell survival and of the microenvironment status under the influence of endostatin peptides.
The serine-threonine kinase Akt, also known as protein kinase B, is a downstream prosurvival signal of phosphatidylinositol 3′-kinase. As shown in Fig. 6, Akt activity measured after exposure of CVEC to VEGF was markedly inhibited by fragment IVox (300 ng/ml). Other fragments had no influence on this enzyme (data not shown).
MMP activity (MMP-2) was predictably induced by FGF-2 and VEGF challenge (20 ng/ml; Fig. 7). Fragment III was the only fragment found capable of increasing the enzyme activity above that of quiescent cells and that obtained by exposure to FGF-2 and VEGF (Fig. 7,A). As shown in Fig. 7, B and C, both fragments IV and IVox had no influence on basal MMP-2 activity, whereas they inhibited FGF-2 and VEGF-stimulated activity.
Effect of Fragment IVox on in Vivo Tumor Growth.
Having shown that fragment IVox acted as a strong antiangiogenic agent, we were interested in determining whether this fragment would exhibit an antitumor effect in vivo on human tumor cell line grown in nude mice. Treatment of A-431 tumors with fragment IVox (62.5 μg/kg/day) reduced the growth of tumors as compared with control. Tumors in fragment IVox-treated mice were significantly smaller (∼50%) than in control mice beginning day 5, whereas the inhibition of tumor growth by native endostatin was characterized by high variability (Fig. 8). Finally, the inhibitory effect of the fragment persisted after discontinuation of animal treatment (data at day 3 after treatment ending is reported in Fig. 8).
The investigation of endostatin partial sequences reveals that these synthetic peptides exhibit distinct biological properties on endothelial cells and on in vivo angiogenesis and tumor growth in comparison with native endostatin.
The in vivo evaluation of endostatin fragments in the avascular rabbit cornea model provided the means to uncover the unexpected activity of one endostatin fragment. In fact, fragment III proved to be a potent angiogenic molecule in the in vivo assay, inducing florid neovascularization of the rabbit cornea, totally devoid of inflammation. The activity of fragment III in promoting angiogenesis was comparable in intensity and in its temporal profile to that of VEGF. Indeed, this fragment enhanced (in an additive fashion) the angiogenic response to VEGF when the two molecules were coreleased in the same cornea, providing firm evidence of its proangiogenic activity. The FGF-2-induced angiogenic response, which was far more intense (almost double) than that obtained with VEGF, was not enhanced by coadministration of fragment III: this might reflect the intrinsic limitation of the rabbit cornea having reached maximal vascularization or might suggest the ability of fragment III to synergize preferentially with selected signaling pathways. Fragment IVox inhibited in a very significant manner angiogenesis promoted by either VEGF or FGF-2. Other fragments were either ineffective (fragments I and II, see details in “Results”) or produced a moderate inhibition (fragment IV). Angiogenesis induced by VEGF was also very significantly reduced by mouse endostatin (800 ng), which appears to exhibit inhibitory potency comparable with that of fragment IVox when the two molecules are compared on a molar basis.
The picture emerging from these in vivo experiments is that the endostatin molecule contains peptide sequences exerting different activities on angiogenesis. The sequence covering the COOH-terminal of the molecule exhibits marked effects on neovascularization, particularly when the disulfide bond Cys135-Cys165 of the parent molecule is maintained as in fragment IVox.
The cellular and biochemical mechanisms of inhibition targeted by endostatin fragments were then evaluated. In quiescent endothelial cells, exposure to these peptide sequences had negligible effects on migration and proliferation, the exception being fragment III, which significantly stimulated cell proliferation at all concentrations. Clearer differences in the activity of the fragments emerged when cells were stimulated to migrate and proliferate under FGF-2 or VEGF challenge. Fragment IVox exerted a pronounced inhibition of cell migration promoted by FGF-2 and VEGF, whereas the other peptides were either ineffective or showed only a modest inhibitory effect. FGF-2-induced migration was nearly suppressed by fragment IVox, the level of migrated cells below that of quiescent cells at concentrations of 100 and 300 ng/ml. Similarly, fragment IVox markedly inhibited cell proliferation stimulated by FGF-2 and VEGF at all concentrations examined, with a clear dose-response effect seen toward FGF-2. Other fragments showed inhibitory effects mainly toward FGF-2-induced proliferation at the highest concentration (300 ng/ml), again with fragment III having no influence on proliferation induced by either growth factors.
Inhibition of cell proliferation by fragment IVox extended to other endothelial cells (e.g., MAEs and HUVECs) but not to nonendothelial cell lines (e.g., smooth muscle and tumor cells), indicating that the effect of this peptide is selective because it targeted only endothelial cells and specific because no changes of cell morphology or viability were observed in trypan blue exclusion experiments. Full-length mouse endostatin exhibited a pattern of activity on migration and proliferation of quiescent and stimulated cells comparable with that observed for fragment IVox. However, when concentrations are expressed on a molar basis, mouse endostatin in in vitro experiments appeared to be more potent (four to five times) than fragment IVox, particularly toward FGF-2-induced proliferation.
The inhibitory effects of fragment IVox were also manifested in the organ culture of vessels and on molecular events stimulated by growth factors. The increase of tubule sprouting from aortic rings under FGF-2 and VEGF challenge was significantly prevented by fragment IVox, which reduced specifically the endothelial compartment. This mirrors the effect of full-length endostatin, which also is capable of blocking tubule formation by affecting the endothelial compartment (27).
The inhibition of stimulated Akt activity seen with this fragment can be interpreted as an indication of its specific action on mechanisms favoring apoptosis because they have been proposed for endostatin (28, 29).
Fragment III exhibited effects on microvascular endothelial cells that were not very pronounced but distinct from all of the other peptides: it increased proliferation in quiescent endothelial cells at low concentration (10 ng/ml, ∼2 nm) and had no inhibitory activity on stimulated cells. Also, the enhancement of stimulated MMP-2 activity was a peculiarity of this peptide not shared by other sequences. These findings point to a direct proangiogenic profile of this fragment and confirm the results obtained in the in vivo cornea assay.
Collectively, the in vitro and in vivo data reveal striking differences in the activity profile of sequences contained in the endostatin molecule. Fragment IVox, covering the COOH-terminal of the molecule, clearly contains the inhibitory domain of endostatin, being capable of inhibiting stimulated endothelial cell functions and in vivo angiogenesis. On the other hand, fragment III displays noninhibitory properties showing clear-cut proangiogenic activity in vivo.
The demonstration that peritumoral treatment with fragment IVox significantly reduced the tumor burden in nude mice definitely proves the efficacy, at least for this route of administration, of the COOH-terminal domain of endostatin as antitumor agent, apparently more efficacious than the full-length molecule.
Thus, it appears that the endostatin molecule may contain different domains endowed with distinct biological properties on endothelial cells. In our view, these findings provide the opportunity to put forward new hypotheses on the action of endostatin and its putative domains. It is very plausible that the inhibition exerted by endostatin may be attributed entirely to its COOH-terminal sequence as other parts of the molecule are either poorly effective for the inhibition or they exert opposite effects. The available evidence indicates that endostatin interacts with its proposed molecular targets as an intact molecule, although the biological activity has been attributed to specific parts of its molecule (14, 30). However, in light of the results of this study, the endostatin inhibitory activities may be also explained by postulating the generation of smaller sequence(s) through a proteolytic process. In fact, recent evidence shows that the proteolytic enzymes cathepsin L and B, which cleave endostatin from collagen, are also capable to quickly degrade endostatin, producing smaller products that have not been characterized (31).
A recent study on synthetic fragments of human endostatin describes the antiangiogenic properties of the COOH-terminal peptide sequence, corresponding to fragment IV of mouse endostatin of this study (32). At variance with the results of this study, the above report also indicates the activity of the NH2-terminal sequence (amino acids 6–49 of human endostatin, corresponding to fragment I of this study). In our hands, this peptide produced a very pronounced inflammatory response when tested in vivo and therefore was not additionally investigated.
Endostatin has been shown to be internalized by an endocytotic process and promotes tyrosine phosphorylation of apoptotic proteins (29). Interestingly, the internalization process of endostatin is followed by a rapid clearance of the molecule leading to its disappearance within hours from its cellular uptake (29). Thus, cleavage of endostatin during the endocytotic process appears likely because the extracellular matrix proteins, including proteases, are in a very dynamic state during the activation of vessel formation (33). In this context, a recent study from this and other laboratories has shown that large protein components of extracellular matrix can encode small peptides able to affect endothelial cell functions in opposite ways (18, 34, 35). The ratio of intact to cleaved molecules might constitute a physiological and transient angiogenic switch regulated by the expression level of the protein and of the tissue-specific proteases responsible for their processing (18, 36).
A number of hypotheses have been advanced to explain the mechanism of action of endostatin on angiogenesis (28, 29, 37, 38). Whether these mechanisms are operant for the fragments examined in this study is not known. However, the hypotheses implicating binding of endostatin to heparin/heparan sulfate, which requires tight structural requirements of the molecule (14, 19, 30) seems the least plausible to explain the inhibition exerted by fragment IVox.
Although controversial results on the efficacy of various preparations of endostatin have appeared in the literature (15, 16, 17), it is arguable whether the findings of this study may offer an explanation for the discrepancies observed.
This study offers interesting clues on the potential of the proteolytic cascade existing in tissues, which might be capable of generating both inhibitors and stimulators of angiogenesis. In addition, the peptide sequences described in this study may be exploited for the therapeutic applications proposed for other inhibitors of angiogenesis, having the advantage of smaller molecular weight.
This work was supported by Italian Association for Cancer Research Grant 593/2001 (to M. Z.), Italian Ministry of University and Research Grants MM06037341 and 2002062519, University of Siena Grant PAR 2002, and the National Research Council (Consiglio Nazionale delle Ricerche, Agenzia 2000) (to S. D.). Fragments described in this article are covered by an European patent application: PCT EP000/03236–April 4, 2000. (Polipeptides derived from endostatin exhibiting antiangiogenic activity.)
The abbreviations used are: VEGF, vascular endothelial growth factor; FGF-2, fibroblast growth factor-2; Fmoc, 9-fluorenylmethoxycarbonyl; DMF, dimethylformamide; TFA, trifluoroacetic acid; BASM, bovine aortic smooth muscle cell; CS, bovine calf serum; CVEC, coronary venular endothelial cell; MAE, mouse aortic endothelial cell; HUVEC, human umbilical vein endothelial cell; MMP, matrix metalloproteinase; mEnd, mouse recombinant full-length endostatin.
|Compound .||Cytotoxicity (% of dead cells/total)a .||.||.||.||.|
|.||CVEC .||MAE .||BASM .||NIH-3T3 .||A431 .|
|None||25 ± 8||15 ± 2||6 ± 1||8 ± 3||10 ± 3|
|mEnd||25 ± 7||13 ± 6||3 ± 2||8 ± 2||8 ± 2|
|Fragment III||19 ± 5||16 ± 5||4 ± 1||9 ± 3||5 ± 4|
|Fragment IV||27 ± 9||14 ± 2||7 ± 2||5 ± 1||6 ± 2|
|Fragment IVox||24 ± 7||18 ± 4||6 ± 2||7 ± 2||9 ± 3|
|Compound .||Cytotoxicity (% of dead cells/total)a .||.||.||.||.|
|.||CVEC .||MAE .||BASM .||NIH-3T3 .||A431 .|
|None||25 ± 8||15 ± 2||6 ± 1||8 ± 3||10 ± 3|
|mEnd||25 ± 7||13 ± 6||3 ± 2||8 ± 2||8 ± 2|
|Fragment III||19 ± 5||16 ± 5||4 ± 1||9 ± 3||5 ± 4|
|Fragment IV||27 ± 9||14 ± 2||7 ± 2||5 ± 1||6 ± 2|
|Fragment IVox||24 ± 7||18 ± 4||6 ± 2||7 ± 2||9 ± 3|
Incubation of cell lines (CVECs, MAEs, BASMs, murine fibroblasts NIH-3T3, and HS epidermoid A431 cells) was carried out for 4 h at 37°C with 300 ng/ml of the compounds. Data are reported as percentage of dead cells over the total number of cells (n = 5 for CVECs and n = 2 experiments for the other cells, run in quadruplicate).
We thank Professor Marco Presta, University of Brescia (Italy) for providing MAEs. We thank Lifetech Srl (Florence, Italy) for technical support and Professor Fabrizio Ledda, University of Florence (Florence, Italy), for the permission to perform some of the experiments in his laboratories, and Professor Richard Schulz, University of Alberta (Alberta, Edmonton, Canada), for his comments and criticisms.