Preparation of yeast surface display (YSD) Ang2-BD constructs and RGD loop library
The construct for Ang2-BDWT (amino acids 281 to 496) in the pIDT plasmid was obtained by custom gene synthesis (Integrated DNA Technologies). Amplification of the gene was performed using primers containing NheI and BamHI restriction sites at the 3′ and 5′ ends, respectively. The amplified gene was then introduced into the pCTCON yeast display vector (a generous gift from Dane Wittrup, MIT). The pCTCON vector introduces a cMyc epitope at the C-terminus of the encoded protein, allowing for the detection of expression by antibodies. A loop on the Ang2-BDWT construct between residues 301–308 was chosen for library construction. The library was prepared using the NNS degenerate codons, where N = A, C, T, or G and S = C or G. The loop library was constructed with an RGD sequence flanked by three random residues on each side of the RGD motif (GenScript) and homologous recombination into Saccharomyces cerevisiae EBY100 cells, as previously described . Library size was approximately 1 × 107 transformants, as estimated by the dilution plating on a selective SDCAA medium (2% dextrose, 1.47% sodium citrate, 0.429% citric acid monohydrate, 0.67% yeast nitrogen base, and 0.5% casamino acids, pH 4.5).
Screening of YSD Ang2-BDRGD libraries
Yeast-displayed Ang2-BD RGD loop libraries were grown in a selective medium and induced for expression with 2% (w/v) galactose at 30 °C overnight until OD600 = 10.0, according to the established protocols . The library was subjected to five rounds of screening using high-throughput flow cytometric sorting to isolate clones with high affinity for recombinant αvβ3 integrin [human integrin αv subunit (Phe31-Val992), human integrin β3 subunit (Gly27-Asp718); R&D Systems]. Library screening was performed using decreasing concentrations of αvβ3 integrin (250 nM, 100 nM, 30 nM, and 10 nM) in sorts 2, 3, 4, and 5, respectively; sort 1 was for positive expression, and Tie2 binding was conducted with 100 nM of αvβ3 integrin. A diagonal sorting gate including 1% of the entire yeast pull was used to select Ang2-BD mutants that bind strongly to αvβ3 integrin. The diagonal sorting gate normalized the binding signal to the amount of protein expressed on the yeast surface. For each round of sorting, yeast cells at an amount of approximately ten times the library size were labeled with solubilized αvβ3 integrin (R&D Systems) and a 1:200 dilution of chicken anti-cMyc antibodies (Thermo Fisher Scientific, Cat# A-21281, RRID:AB_2535826) in integrin-binding buffer [IBB, 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM MnCl2, 2 mM CaCl2, and 1% bovine serum albumin (BSA)] for 1 h at room temperature to facilitate fluorescent detection by flow cytometry. Cells were washed and resuspended in ice-cold PBSA (phosphate-buffered saline with 1% BSA) containing a 1:25 dilution of fluorescein isothiocyanate (FITC)-labeled mouse anti-αv integrin (BioLegend, Cat# 327907, RRID:AB_940558) and a 1:100 dilution of phycoerythrin (PE)-conjugated anti-chicken IgY (Santa Cruz Biotechnology, Cat# sc-3748, RRID:AB_634859). After 25 min of incubation on ice, yeast cells were washed in PBSA and sorted using an iCyt Synergy FACS (fluorescence-activated cell sorting) apparatus [Proteomics Unit, National Institute for Biotechnology in the Negev (NIBN), Ben-Gurion University of the Negev (BGU)]. Sixty isolated clones from the two final sorts were sequenced by extraction of plasmid DNA from the yeast clones using a Zymoprep kit (Zymo Research) and transformed into electrocompetent Escherichia coli cells for plasmid miniprep isolation (RBC Bioscience Corp, Taiwan) and DNA sequencing (DNA Microarray and Sequencing Unit, NIBN, BGU). Cells expressing these clones were evaluated for their binding affinity towards αvβ3 integrin by dividing the mean fluorescence intensity (MFI) of the αvβ3 integrin-binding signal by the MFI reflecting expression levels. Binding and expression were detected using anti-αv integrin and anti-cMyc antibodies, respectively. The isolated clones were evaluated for their binding affinity towards Tie2-Fc (R&D Systems) by dividing the MFI of the Tie2 binding signal by the MFI reflecting expression levels. The values obtained were normalized to those obtained with Ang2-BDWT. Of the 60 isolated clones, 5 with the highest affinity for αvβ3 integrin and Tie2 were selected.
Integrin-binding specificity assay
Flow cytometry analysis of 1 × 106 cells of each of the five isolated clones from the RGD loop library was conducted using a 1:200 dilution of chicken anti-cMyc antibody (Thermo Fisher Scientific, Cat# A-21281, RRID:AB_2535826); 50 nM of solubilized αvβ3, αvβ5, α5β1, α3β1, α4β7, or αIIbβ3 integrins (R&D Systems); and 20 nM of soluble Tie2-Fc (R&D Systems) in parallel for 1 h at room temperature. Cells were washed and resuspended in ice-cold PBSA containing a 1:25 dilution of FITC-labeled mouse anti-αv/α5 integrin (BioLegend, Cat# 327907, RRID:AB_940558, BioLegend Cat# 328308, RRID:AB_2129084), a 1:25 dilution of allophycocyanin (APC)-labeled mouse anti-α3/α4/α2b integrin (BioLegend, Cat# 343807, RRID:AB_10641703, Cat# 304307, RRID:AB_314433, Cat# 303709, RRID:AB_2129464), and a 1:100 dilution of PE-conjugated anti-chicken IgY (Santa Cruz Biotechnology, Cat# sc-3748, RRID:AB_634859). After 25 min on ice, yeast cells were washed in PBSA and analyzed using BD Accuri C6 flow cytometer (BD Biosciences). These clones (Ang2-BDWT, Ang2-BDBC5, Ang2-BDBC6, Ang2-BDBC10, Ang2-BDBC14, and Ang2-BDBC35) were evaluated for their binding affinity towards αvβ3, αvβ5, α5β1, α3β1, α4β7, and αIIbβ3 integrins by dividing the MFI of the αvβ3 integrin-binding signal by the MFI reflecting expression levels.
Purification of soluble Ang2-BD proteins
The Multi-Copy Pichia Expression Kit (Invitrogen K1750-01) was used to produce the soluble Ang2-BDRGD and Ang2-BDWT protein variants, as previously described . Ang2-BDRGD variants were purified from yeast culture supernatants by metal-chelating chromatography using a 5-ml HisTrap FF column (GE Healthcare) equilibrated with 10 mM imidazole and eluted with 500 mM imidazole. Eluted protein fractions were concentrated and buffer exchanged with 20 mM Hepes, 150 mM NaCl, and pH 7.2 buffer using a 5-kDa cutoff Vivaspin concentrator (GE Healthcare). Gel filtration chromatography was performed using a Superdex 200 column (GE Healthcare) equilibrated with 20 mM Hepes, 150 mM NaCl, pH 7.2, and buffer at a flow rate of 0.5 ml/min on an ÄKTA pure instrument (GE Healthcare). Proteins were separated by SDS-PAGE under non-reducing conditions. Concentrations of all the Ang2-BDRGD protein variants were determined by UV-Vis absorbance at 280 nm and an extinction coefficient of 66,500 M−1 cm−1. The molecular weights of the purified proteins were determined using a MALDI-TOF REFLEX-IV (Bruker) mass spectrometer (Ilse Katz Institute for Nanoscale Science & Technology, BGU).
Surface plasmon resonance (SPR) experiments
The binding interactions of Tie2 to Ang2-BDWT, Ang2-BDBC5, Ang2-BDBC6, and Ang2-BDBC10 were analyzed (Proteomics Unit, NIBN, BGU) by SPR using a ProteOn XPR36 instrument (Bio-Rad) as previously described . The binding interactions of αvβ3, αvβ5, αvβ1, αvβ6, αvβ8, α5β1, α4β7, and α3β1 integrins to Ang2-BDBC5, Ang2-BDBC6, and Ang2-BDBC10 with the extracellular domain of recombinant human αvβ3, αvβ5, αvβ1, αvβ6, αvβ8, α5β1, α4β7, and α3β1 integrins were similarly analyzed (R&D Systems). All integrins were immobilized on the surface of a GLC sensor chip (Bio-Rad) using the amine-coupling reagents sulfo-NHS (0.1 M N-hydroxysuccinimide) and EDC (0.4 M 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide; Bio-Rad). αvβ3, αvβ5, αvβ1, αvβ6, αvβ8, α5β1, α4β7, and α3β1 integrins (5.6 μg) in 10 mM sodium acetate, pH 4.0, were allowed to flow over the activated surfaces of the GLC sensor chip channel at a flow rate of 30 μl/min until target immobilization levels (4300, 7800, 5400, 4600, 5900, 3400, 7100, and 4200 RU, respectively) were reached. BSA (3 μg) in 10 mM sodium acetate, pH 4.5, was then allowed to flow over the activated surfaces of a control GLC sensor chip channel six at a flow rate of 30 μl/min until the target immobilization level (3000 RU) was reached. After protein immobilization, the chip surface was treated with 1 M ethanolamine-HCl at pH 8.5 to deactivate any excess reactive esters. All binding experiments were performed at 25 °C in degassed IBB. Since no suitable regeneration conditions were found for the surface with immobilized αvβ3 integrin, a separate channel was used to test the binding of each Ang2-BD protein. To determine αvβ3-integrin binding interactions, 12.5 to 200 nM of Ang2-BD variants were used, while for αvβ5, αvβ1, αvβ6, αvβ8, α5β1, α4β7, and α3β1 integrins, 1 μM of the Ang2-BD variants was used. For αvβ3 integrin binding, the protein analytes were allowed to flow over the surface-immobilized integrins for 600 s at a flow rate of 30 μl/min, and binding interactions were monitored. Following association, dissociation of the various ligand–receptor complexes was monitored for 400 s. For binding other integrins, the protein analytes were allowed to flow over the surface-immobilized integrins for 400 s at a flow rate of 30 μl/min, the and interactions were monitored. Following association, dissociation of the various ligand–receptor complexes was monitored for 600 s. Each analyte sensorgram run was normalized by subtracting the BSA channel (channel six) run and the zero analyte concentration run. Sensorgram data for αvβ3 integrin binding for all of the Ang2-BD bi-specific variants were analyzed using the 1:1 L model for binding kinetics evaluation and kinetic parameters. Ang2-BDBC5 binding to αvβ5, αvβ1, αvβ6, and αvβ8 was analyzed as above. In brief, αvβ5, αvβ1, αvβ6, and αvβ8 integrins (5.6 μg) in 10 mM sodium acetate, pH 4.0, were allowed to flow over the activated surfaces of the GLC sensor chip channel at a flow rate of 30 μl/min until the target immobilization levels (6400, 6300, 4400, and 6100 RU, respectively) were reached. To determine the integrin binding, 62.5 to 1000 nM of Ang2-BDBC5 was used. Ang2-BDBC5 was allowed to flow over the surface-immobilized integrins for 800 s at a flow rate of 30 μl/min, and the interactions were monitored. Following association, dissociation of the various ligand-receptor complexes was monitored for 700 s. SPR sensorgram curves were fitted into a two-state binding model.
Dual receptor binding experiments
A ProteOn GLC sensor chip was prepared as described above with immobilized αvβ3 integrin extracellular domain (R&D Systems). αvβ3 integrin (5.6 μg) in 10 mM sodium acetate, pH 4.0, was allowed to flow over the activated surfaces of the GLC sensor chip channel at a flow rate of 30 μl/min until an immobilization level of 4400 RU was reached. Experiments were performed at 25 °C in degassed IBB. Ang2-BDWT, Ang2-BDBC5, Ang2-BDBC6, or Ang2-BDBC10 (at a concentration of 400 nM) was allowed to flow over the integrin-immobilized surface for 400 s at a flow rate of 30 μl/min. Thereafter, the extracellular domain of recombinant human Tie2 (rhTie2), also at 400 nM, was allowed to flow over the surface for 150 s. Dissociation of the complex was monitored for 650 s. Injection of running buffer followed by rhTie2 in IBB served as negative control.
Human telomerase-immortalized microvascular endothelium (TIME) cells (ATCC, Cat# CRL-4025, RRID:CVCL_0047) were cultured in growth-factor-depleted Vascular Cell Basal Medium (ATCC) supplemented with 2% fetal bovine serum (FBS) and growth factor supplements (ATCC). For binding assays, 105 cells were suspended and incubated with different concentrations of Ang2-BD variants in a total volume of 200 μl PBSA, followed by incubation at 4 °C for 2 h with gentle agitation. Cell suspensions were centrifuged at 150g at 4 °C for 5 min and washed with 100 μl PBSA, followed by centrifugation at 150g at 4 °C for 5 min twice more. Cells were then resuspended in 100 μl PBSA containing a 1:200 dilution of APC-conjugated anti-FLAG antibodies (BioLegend, Cat# 637308, RRID:AB_2561497). After 30 min on ice, the cells were washed twice in PBSA and analyzed by flow cytometry with a BD Accuri C6 flow cytometer (BD Biosciences). Mean fluorescence values were generated using FlowJo software (Treestar). For the competitive binding assay, cells were treated as described above with added full-length human Ang1 (FL-Ang1), cRGD peptide, or a combination of the two. Since FL-Ang1 exists in different oligomeric states, the FL-Ang1 concentration is reported in this work as mass concentration units instead of molar concentration units. MFI was detected using PE-conjugated anti-FLAG antibody (BioLegend, Cat# 637309, RRID:AB_2563147) and analyzed by flow cytometry with a BD Accuri C6 flow cytometer. For receptor level detection, 105 cells were harvested, resuspended in 100 μl PBSA with 1:100 Alexa Fluor 647-labeled anti-human Tie2 antibodies (BioLegend, Cat# 334210, RRID:AB_2203206) or (FITC)-labeled anti-human αvβ3 integrin antibodies (Millipore, Cat# MAB1976F, RRID:AB_94482), incubated at 4 °C for 30 min, and then analyzed by flow cytometry.
Docking modeling and simulation of αvβ3 and Ang2-BDBC5 complex
Molecular coordinates of the αvβ3 binding domains were taken from the 1L5G PDB structure  (residues 1–438 of the αv subunit and 55–432 of the β3 subunit). The coordinates of the binding domain of Ang2 were obtained from the 1Z3S PDB structure  (residues 280–495). The Ang2-BDBC5 mutant was created by replacing residues 301–308 of the native protein with residues NTCRGDCLP using the PyMOL Molecular Graphics System, Version 1.8 (De Lano). Each structure was energy-minimized using the Gromacs 4.6.7 package of programs , and the receptor–ligand docking procedure was performed by a PatchDock server . To avoid irrelevant structures, potential binding sites both for the receptor and the ligand were defined according to PatchDock recommendations. Slight variations in the interaction restraints yielded a total of 421 structures. Docking solutions were clustered with a 0.6-nm cutoff using Gromacs. The most prominent cluster (41% of total) was subjected to molecular dynamics (MD) simulation with Gromacs 4.6.7. Two identical simulations were carried out with the GROMOS 53a6 force field , yielding similar results. The protein was immersed in a dodecahedral box, filled with simple point charge (SPC)  water molecules and ions that extended to at least 1.2 nm from the edge of the protein. The whole system was subjected to energy minimization using the steepest descent algorithm until the force component of the system was smaller than 1000 kJ mol−1 nm−1. Equilibration with the solvent was initiated by a 40 ps position-restrained simulation under a constant force of 1000 kJ mol−1 nm−1 at 300 K. Next, the system was simulated without restraints for 3 ns, allowing for equilibration. The final structure was used for a 10-ns MD simulation as detailed below.
MD simulations were run under NPT (constant number of particles, pressure, and temperature) conditions, relying on Berendsen’s coupling algorithm for maintaining constant temperature and pressure (P = 1 bar, τp = 0.5 ps, T = 300 K, τR = 0.1 ps) . A LINCS (linear constraint solver) algorithm  was used to constrain the lengths of all bonds; the water molecules were restrained by the SETTLE algorithm. Long-range electrostatic interactions were treated by the particle mesh Ewald method . Distances and electrostatic and Lennard–Jones potentials were analyzed with the tools provided by the GROMACS package, while snapshots were prepared by the VMD program .
The Tie2 structure was obtained from the 2GY7 PDB structure (residues 23–445), and its coordinates were aligned to a simulated Ang2-BDBC5–αvβ3 complex to show the possibility of simultaneous binding of Ang2-BDBC5 to αvβ3 integrin and Tie2.
Cell adhesion assays
Inhibition of adhesion of TIME cells to vitronectin was determined in 96-well microplates coated with human vitronectin (R&D Systems). Ang2-BDWT, Ang2-BDBC5, Ang2-BDBC6, Ang2-BDBC10, or cRGD peptide (Merck Millipore) (1 μM) was mixed with 5 × 104 TIME cells and plated on vitronectin-coated wells either with or without 500 ng/ml of FL-Ang1, incubated at 37 °C/5% CO2 for 2 h, and washed twice with PBS. A solution of 0.2% crystal violet in 10% ethanol was added to the wells for 10 min, which were then washed three times with PBS. Solubilization buffer (a 1:1 mixture of 0.1 M NaH2PO4 and ethanol) was added, and the plate was shaken gently for 15 min. Absorbance was measured at 600 nm using a microtiter plate reader (BioTek Instruments). Background signals generated with a negative control containing no cells were subtracted from the data.
Tie2, Akt, and FAK phosphorylation assays
Confluent TIME cells were cultured in growth-factor-depleted Vascular Cell Basal Medium supplemented with 0.5% FBS for 12 h at 37 °C/5% CO2 on human vitronectin-coated 12-well plates prior to experimentation. The cells were then washed with PBS, and the medium was exchanged with fresh Vascular Cell Basal Medium-depleted of growth factors and serum. After pre-treatment with 1 mM sodium orthovanadate (Na3VO4; Sigma) for 15 min, the cells were co-incubated for 15 min for Tie2 and FAK and for 30 min for Akt at 37 °C with either commercial full-length rhAng1 as positive control (R&D Systems) or a combination of full-length rhAng1 and the Ang2-BD bi-specific variants. Non-stimulated cells served as negative control. The cells were then washed twice with PBS plus 1 mM Na3VO4 and lysed in ice-cold lysis buffer [20 mM HEPES, pH 7.4, 150 mM NaCl, 1% TritonX-100, 1 mM Na3VO4, and 1× complete protease inhibitor cocktail tablet (Roche)]. The cells were scraped from the culture plate wells, and the lysates were clarified by centrifugation (13,000 rpm for 30 min at 4 °C). Protein concentration was measured by the BCA assay (Thermo Fisher Scientific), and equivalent amounts of each lysate sample were analyzed by duplicate 10% SDS-PAGE and transferred to duplicate PVDF membranes (BioRad). Blots were blocked (5% BSA, 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20) for 1 h at room temperature and probed with 1:500 dilution anti-phospho-Tie2-specific rabbit polyclonal (R&D Systems, Cat# AF2720, RRID:AB_442172) and 1:1000 dilution anti-Tie2-specific rabbit monoclonal antibodies (Cell Signaling Technology, Cat# 7403S, RRID:AB_10949315), 1:1000 dilution anti-phospho-Akt-specific (Cell Signaling Technology, Cat# 4060, RRID:AB_2315049), and 1:1000 dilution anti-Akt-specific antibodies (Cell Signaling Technology, Cat# 4691, RRID:AB_915783) or 1:1000 dilution anti-phospho-FAK-specific (Cell Signaling Technology, Cat# 8556S, RRID:AB_10891442) and 1:1000 dilution anti-FAK-specific antibodies (Cell Signaling Technology, Cat# 3285S, RRID:AB_10694068) overnight at 4 °C. Membranes were washed three times with TBST (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20) and probed with 1:1000 dilutions HRP-linked anti-rabbit antibodies (Cell Signaling Technology, Cat# 7074, RRID:AB_2099233) for 1 h at room temperature. Membranes were washed three times with TBST and then visualized and quantified using chemiluminescence (ECL, Biological Industries) and ImageJ software, respectively. The intensities of the phospho-Tie2, phospho-Akt, and phospho-FAK bands were adjusted for the expression of total Tie2, Akt, and FAK for each experiment. Blots were stripped and re-probed with 1:1000 dilution anti-β-actin antibodies (Cell Signaling Technology, Cat# 4970, RRID:AB_2223172) for further normalization. Each condition was repeated in triplicate. Phosphorylated protein (Tie2, Akt, and FAK) band intensities (as measured by ImageJ software) were normalized to the respective total protein levels, and this value was subsequently normalized to the total amount of β-actin for each sample. For each condition, a representative band is shown.
Matrigel endothelial tube formation assay
Serum-reduced Matrigel (10 mg/ml; BD Biosciences) was thawed overnight at 4 °C, and 150 μl was added to each well of a 48-well microtiter plate and allowed to solidify for 1 h at 37 °C. The wells were incubated with 3.25 × 104 TIME cells plus 500 ng/ml rhAng1 either alone or with 1 μM of Ang2-BDWT, Ang2-BDBC5, Ang2-BDBC6, Ang2-BDBC10, or cRGD peptide (Merck Millipore). The cells were incubated for 16–18 h at 37 °C/5% CO2 and then washed twice in HBSS (Hanks’ balanced salt solution; Sigma). Capillary tube formation was observed using EVOS Cell Imaging Systems microscope (ThermoFisher Scientific). Images were collected with an EVOS × 2 Objective. The total number of meshes and the number of junctions of the tubes were quantified by the analysis of digitized images of the capillary-like structures using ImageJ software and the Angiogenesis Analyzer plugin.
Cell viability assay
The effects of Ang2-BD bi-specific variants on the growth and survival of TIME cells were assessed by an XTT assay (2,3-bis [2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanilide inner salt assay; Biological Industries). TIME cells were seeded (7500 cells per well) on a human vitronectin-coated 96-well microplate (R&D Systems) and incubated at 37 °C/5% CO2 for 24 h. The medium was then replaced with fresh Vascular Cell Basal Medium supplemented with 2% FBS and growth factor supplements, and the cells were incubated with 500 ng/ml of rhAng1 either alone or with 2 μM Ang2-BDWT, Ang2-BDBC5, Ang2-BDBC6, Ang2-BDBC10, or cRGD peptide (Merck Millipore). The cells were incubated for 16–18 h at 37 °C/5% CO2. Viable cells from each condition were measured by XTT at UV 450 nm, as described in the manufacturer’s protocol. The UV readings of the cell-only control were normalized to 100%, and readings from cells treated with the Ang2-BD variants were expressed as a percentage of the control.
An in vitro Boyden chamber assay was performed using ThinCert 24-well inserts (Greiner Bio-One). ThinCert cell culture insert membranes were coated with Matrigel (Corning) diluted in Vascular Cell Basal Medium (ATCC) at a 1:30 ratio. The lower compartment was filled with 600 μl of Vascular Cell Basal Medium supplemented with 2% FBS. TIME cells (2 × 104), with or without Ang2-BD variants and Ang1, were incubated in 200 μl supplement-free Vascular Cell Basal Medium, added to the pre-coated ThinCert cell culture inserts, and incubated for 20 h at 37 °C/5% CO2. Invasive cells were stained with a DippKwik stain kit (American MasterTech Scientific) and detected by an EVOS FL Cell Imaging System at × 20 magnification. Quantification was performed by counting 16 fields for each membrane. Analysis of digitized images was performed using ImageJ software and a Cell Colony Edge Analyser.
Data were analyzed with GraphPad Prism version 5.00 for Windows (La Jolla, CA). Data shown in all the figures are the means of triplicate from independent experiments, and error bars represent the standard error of the mean. Statistical significance was determined by column statistics and one-way ANOVA analysis. A P value < 0.05 was considered statistically significant.