Obtustatin

Introduction

Angiogenesis, the formation of new vessels, is believed to be central to tumor development and metastasis (1) , and the investigation of suppressors of this process has become a major approach to cancer therapy. At present, many endogenous negative regulators have been identified, including thrombospondin (2) , and a number of proteolytic protein fragments, including angiostatin (3) , endostatin (4) , kininostatin (5) , tumstatin (6) , and arresten (7) . The angiostatic mechanisms of these factors are under intensive investigation, but recent data implicate integrins as potential mediators of these inhibitory processes, e.g., α5β1, αvβ3, and αvβ5 integrins are primary targets for endostatin action (8) , whereas angiostatin and tumstatin interact with αvβ3 integrin (6 , 9) . Arresten interacts with the α1β1 integrin (7) , which is selective for collagen IV, a major component of basement membranes. These integrins are all expressed on vascular cells, and thus, regulation of vascular cell integrin-ligand interactions is becoming an organizing theme within angiogenesis research.

Disintegrins are the largest group of antiadhesive proteins found in viper venom (10) . The characteristic feature common to all disintegrins is the similar pattern of cysteines and presence of the so-called “integrin-binding” loop. Disintegrins may be divided into two groups, monomeric and dimeric. The monomeric disintegrins form the largest class, usually contain the RGD 3 sequence in their integrin-binding loop, and are potent inhibitors of the platelet fibrinogen receptor αIIbβ3 integrin. They have been divided into three subgroups based on the number of cysteines in their molecules (10) . The “short” disintegrins, with only eight cysteines, are represented by the 49 amino acid-containing peptides echistatin and eristostatin, both of which are potent inhibitors of αIIbβ3 integrin. However, only echistatin inhibits other RGD-dependent integrins, such as the vitronectin receptor αvβ3. Here, we describe a novel short monomeric disintegrin called obtustatin. The structure of its integrin-binding loop is novel, and it is a highly selective inhibitor of α1β1 integrin in vitro and of angiogenesis in vivo.

Materials and Methods

Cell Lines and Integrins.

K562 cells transfected with α1 and α2 integrins were from Dr. P. Gotwals (Biogen, Inc., Cambridge, MA) and Dr. M. Hemler (Dana-Farber Cancer Institute, Boston, MA). The Lewis lung carcinoma cells were from Dr. Bruce Ruggeri (Cephalon, Inc., West Chester, PA). K562 cells were from American Type Culture Collection (Manassas, VA). Collagen type I and IV were from Chemicon International, Inc. (Temecula, CA). Synthetic peptides based on the structure of obtustatin were synthesized commercially by Sigma-Genosis (Woodland, TX).

Purification of Disintegrins.

Obtustatin was purified from the venom of Vipera lebetina obtusa using two steps of reverse phase high-performance liquid chromatography as described (11) . Briefly, 10 mg of Vipera lebetina obtusa venom in 300 μl of 0.1% TFA were injected into C₁₈ column and eluted with an acetonitrile gradient (0–80%) in 0.1% TFA over 45 min at a 2 ml/min flow rate. The obtustatin fraction (1 mg in 500 μl of 0.1% TFA) was reapplied to the same column and eluted with a second acetonitrile gradient (20–80%) over 70 min. The yield of purified obtustatin was 12 mg/grams crude venom. Purity was assessed by SDS-PAGE and matrix-assisted laser desoption ionization-time-of-flight MS using an Applied Biosystems DE-Pro spectrometer (Wistar Institute, University of Pennsylvania, MS facility). Eristostatin was purified from the venom of Eristocophis macmahoni (12) .

Structural Characterization of Obtustatin.

Purified obtustatin was reduced and alkylated as described (11) . S-pyridylethylated obtustatin was characterized by NH₂-terminal sequencing (using either an Applied Biosystem 477A or Beckman Porton LF-3000 instrument), amino acid analysis (using a Beckman Gold Amino Acid Analyzer after sample hydrolysis with 6 N HCl, 24 h, 110°C), and MS (as above). The primary structure of obtustatin was deduced from the NH₂-terminal sequence analysis of overlapping peptides obtained by endo-Lys C digestion and purified as described (11) .

Cell Adhesion and CFB Assays.

Adhesion studies of cells labeled with 5-(chloromethyl)fluorescein diacetate were performed as described (11) . For the CFB assay (12) , collagen IV- or collagen I-coated Dynabeads M-280 (1 mg/ml) were blocked with 8% Lewis rat plasma in assay buffer [50 mm HEPES (pH 7.5), 150 mm NaCl, and 0.1% Triton X-100] for 5 min. Beads (10 μg) were combined with α1K562 cell lysate (10⁵ cell equivalents), obtained by lysis of cells with buffer containing 1% NP40 and TS2/16 (0.1 μg/ml) anti-β1 monoclonal antibody labeled with ruthenium (II) tris-bipyridine N-hydroxysuccinimide ester (IGEN, Inc., Gaithersburg, MD) in assay buffer containing 1 mm MnCl₂. In parallel, disintegrins were added. After 1–2-h agitation at room temperature, 200 μl of assay buffer were added, and the samples were read on an ORIGEN electrochemiluminescence detector (IGEN).

Chicken CAM Assay.

The chicken CAM assay was performed as described (5) . Briefly, filter disks were soaked in 3 mg/ml cortisone acetate in solution of 95% ethanol and water and air dried. Disks absorbed with FGF2 (1 μg/ml PBS) in the presence (5 μg/disk) or absence of disintegrins were placed on growing CAMs. At 24 h, disintegrins were added to CAMs topically. After 48 h, the CAM tissue directly beneath FGF2-saturated filter disk was resected from the embryo, and the section was placed in a 35-mm Petri dish and examined under an SV6 stereomicroscope at ×50 magnification. Digital images of CAM sections were collected using 3-charge-coupled device camera system (Toshiba) and analyzed with Image-Pro Plus software. The number of vessel branch points contained in a circular region equal to the area of filter disk (angiogenesis index) was counted for each section.

Mouse Model for Tumor Development.

Lewis lung (3) carcinoma cells (1 × 10⁶) were injected under the skin of the C57BL/6 mice (Tacoma, Inc., Germantown, NY). The tumors were allowed to grow for 1 week. The average tumor size at this time, called day 0, was ∼0.08 cm³. At day 0, the first i.p. injection of disintegrins was performed. Obtustatin was injected every other day at a dose of 5 mg/kg. In the control group, the PBS was injected with the same frequency. Tumor volume was measured using the standard formula length × width² × 0.52 (4) . Each group contained four animals.

Results and Discussion

Obtustatin was purified to homogeneity from the venom of the Vipera lebetina obtusa viper using two steps of reverse-phase high-performance liquid chromatography. Mass spectroscopy revealed its molecular mass to be 4395.2 Da. The amino acid sequence of obtustatin was established using automated Edman degradation, as applied previously to other disintegrins (11) . This procedure included NH₂-terminal sequencing of reduced and pyridylethylated obtustatin and of peptides obtained after degradation with endo-Lys C. The calculated molecular mass of obtustatin is 4394.2 Da, which agrees very well with the mass determined experimentally by matrix-assisted laser desoption ionization MS. Obtustatin, containing only 41 amino acids, is the shortest disintegrin reported to date, with a pattern of cysteines nearly identical to two other short monomeric disintegrins, echistatin and eristostatin, placing it in this subgroup (Fig. 1A) ⇓ . However, the sequence of its integrin-binding loop is completely different from these other disintegrins, which contain an RGD sequence (Fig. 1A) ⇓ , suggesting distinct integrin specificity.

Obtustatin was screened against a panel of integrins and found to be a very potent and selective inhibitor of α1β1 integrin. It does not inhibit the structurally closely related collagen receptor α2β1 integrin (Fig. 1, B and C) ⇓ . In other cell adhesion assays, obtustatin did not inhibit αIIbβ3 and αvβ3 integrins, four other β1 integrins (α4, α5, α6, and α9), or α4β7 integrin (data not shown). The anti-α1β1 integrin activity of obtustatin was also confirmed in a cell-free assay (Fig. 1C) ⇓ . Obtustatin inhibited the binding of solubilized α1β1 integrin to collagen type IV (0.8 nm IC₅₀) but was without effect in a cell-free assay using solubilized α2β1 integrin.

Integrin α1β1 contains a so-called “inserted” or I-domain, which is present in the α subunit, contains ∼200 amino acid residues, is localized near the NH₂ terminus (13) , is highly conserved, and plays a necessary and direct role in ligand binding. We found that obtustatin did not inhibit the binding of the recombinant α1 subunit I-domain to collagen IV at concentrations ≤1 μm (data not shown). This result suggests that the α1β1 integrin may contain a second binding site for collagen IV. It has been found (14) that this integrin may bind two distinct fragments of collagen IV, the pepsin-derived triple helical domain and nonhelical NC1 (noncollagenous) domain. One possibility is that binding of the NC1 fragment does not occur through the I-domain but through another site in common with obtustatin.

A comparison of the integrin-binding loop of obtustatin to related RGD-containing disintegrins (Fig. 1A) ⇓ suggested that the motif homologous to RGD in obtustatin lies within the sequence KTSLT. The active site of obtustatin was localized by analysis of short peptides. Fig. 2A ⇓ shows the effects of native obtustatin, EP-obtustatin, and two obtutastin-derived linear synthetic peptides on adhesion of α1K562 cells to immobilized collagen IV. EP-obtustatin is a refolded form of this protein with reduced S–S bonds and cysteines blocked with vinylpyridine. EP-obtustatin still retained inhibitory activity, but the IC₅₀ increased from 2 nm to 30 μm. The synthetic peptide CWKTSLTSHYS, containing the entire integrin-binding loop, gave an IC₅₀ of 600 μm. In contrast, the peptide CKLKPAGTTC, synthesized based on another part of obtustatin, was not active even at 20 mm. These data indicate that the active site of obtustatin is localized, as expected, within the integrin-binding loop. The essential amino acids for activity within this loop were then localized through alanine scanning (Fig. 2B) ⇓ . The data revealed that first threonine is critical for activity. The peptides with mutated lysine and serine adjacent to this amino acid lost only partially their inhibitory activity at a 1 mm concentration. These results suggest that the KTS sequence may be a new biologically active motif relevant for the α1β1 integrin.

Recent studies suggest that α1β1 integrin is important in new vessel development. Vascular basement membrane collagens are proving a rich source of antiangiogenic fragments (6 , 7) . A fragment of the NC1 collagen IV domain, designated arresten, inhibits endothelial cell proliferation in vitro and angiogenesis in vivo, and its mechanism of action is linked to interactions with α1β1 integrin (7) . In addition, proteolytic exposure of cryptic sites within collagen type IV, required for angiogenesis and tumor growth in vivo, is associated with the loss of α1β1 integrin binding and the gain of αvβ3 integrin binding (15) . Furthermore, α1 knockout mice show a significant reduction in vascularization of skin tumors, an imbalance in their collagen/collagenase ratios, and circulating angiostatin (16) . In a direct investigation of the mechanisms through which α1β1 integrin supports angiogenesis driven by VEGF (17) , it was shown that dermal microvascular EC attachment through α1β1 integrin supported robust activation of the Erk1/Erk2 (p44/42) mitogen-activated protein kinase signal transduction pathway driving EC proliferation. Haptotactic EC migration toward collagen I was partially dependent on α1β1 integrin, as was VEGF-stimulated chemotaxis of ECs in a uniform collagen matrix. In addition, monoclonal antibody antagonism resulted in ∼45% inhibition of VEGF-driven angiogenesis in mouse skin. These studies show a critical role for α1β1 integrin in VEGF-driven angiogenesis in the dermis but do not address other agonists or organ compartments. Therefore, to evaluate obtustatin as an angiogenesis inhibitor, we used FGF2, an agonist distinct from VEGF, and an established angiogenesis system not used previously, the chicken CAM. As shown in Fig. 3A ⇓ , as little as 5 μg of obtustatin potently inhibits FGF2-stimulated new vessel development in the CAM by ∼80%. Eristostatin, a disintegrin structurally related to obtustatin (Fig. 1A) ⇓ but with no integrin receptors expressed on endothelial cells (data not shown), was used as a control and showed no inhibitory activity. Our experiments directly confirm and extend the importance of α1β1 integrin in angiogenesis. Moreover, obtustatin significantly inhibited the development of Lewis lung carcinomas in a syngeneic mouse model (Fig. 3B) ⇓ . On therapeutic treatment of established tumors, obtustatin reduced tumor sizes ≤50% after 1 week of treatment. These results agree with genetic studies with α1-null mice in which tumor growth was also reduced (16) .

Additional studies on obtustatin and obtustatin-derived compounds may have a significant impact on development of novel drugs with angiostatic activity. New peptide and peptidomimetic inhibitors have been developed for the therapy of platelet thromboembolism, based on early critical data with RGD/KGD(lysine-glycine-aspartic acid)-containing disintegrin structures (18) . The identification of the novel tri-peptide sequence KTS for α1β1 integrin may similarly provide lead compounds for drug development. In addition, although many inhibitors of angiogenesis are known, a detailed understanding of their molecular mechanisms has been lacking. Recent work shows that tumstatin is an endothelial cell-specific inhibitor of protein synthesis, via a novel interaction with αvβ3 integrin (6) . The availability of a well-defined angiogenesis inhibitor of α1β1 integrin should also provide insights into the molecular role of this integrin in vascular cell function. Finally, after the discovery of disintegrins in snake venom, a large family of related molecules has been found in mammalian systems, which contain metalloprotease and disintegrin domains, some of which are known to bind integrins (19) . These molecules play critical roles in fertilization, neurogenesis, protein ectodomain shedding, and amyloid precursor protein processing. We suggest that novel mammalian disintegrins may be found, which, like obtustatin, are inhibitors of α1β1 integrin and may form a new group of endogenous angiogenesis inhibitors. Interestingly, we note that one member of this family, MDC-9/ADAM-9, does in fact contain the sequence KTS in the disintegrin domain, in a position analogous to that of the obtustatin sequence (20) . We are currently investigating whether MDC-9 binds α1β1 integrin.