Synthesis and Evaluation of Multivalent Integrin Ligand

Materials: Unless otherwise noted, all commercially available reagents and solvents were of analytical grade and were used without further purification. Protected amino acids were purchased from IRIS Biotech (Germany). Cu(OAc)₂⋅H₂O, 4‐pentynoic acid, diisopropylamine (DIPEA) and sodium ascorbate were purchased from Sigma‐Aldrich. 1,4,7‐triazacyclononane‐1,4,7‐triacetic acid (NOTA) was purchased from Chematech (Dijon, France). HATU was obtained from Bachem Holding AG (Bubendorf, Switzerland). HOBt hydrate was obtained from Carbolution (St. Ingbert, Germany). For all radiochemical works, Tracepur® water (Merck) was used. TRAP(azide)₃18 and o‐NBS‐l‐Lys(Fmoc)‐OH47 were synthesized as described previously.
Instrumentation: The synthesis of SDM17‐pentynoic amide was carried out in an ultrasonic bath SONOREX RK 52 H (interior dimensions 150×140×100 mm and operating volume 1.2 L) by BANDELIN electronic (Germany), equipped with timer control for 1–15 min and continuous (∞) operations and built‐in heating control (30–80 °C thermostatically adjustable). Semi‐preparative reversed‐phase HPLC was performed by using a Shimadzu system, consisting of two LC‐20AP quaternary low‐pressure gradient pumps, a SPD‐M30 A photodiode array detector, and a CBM‐20 A system controller. Separations were performed by using a YMC‐Pack ODS‐A, 5 μm, 250×20 mm C₁₈ column. Analytical HESI‐HPLC‐MS (heated electrospray ionization mass spectrometry) was performed on a LCQ Fleet (Thermo Scientific) with a connected UltiMate 3000 UHPLC focused (Dionex) on C₁₈ columns: S1: Hypersil Gold aQ 175 Å, 3 μm, 150×2.1 mm (for 8 or 20 min measurements); S2: Accucore C18, 80 Å, 2.6 μm, 50×2.1 mm (for 5 min measurements; Thermo Scientific). Linear gradients (5 %–95 % acetonitrile content) with water (0.1 % v/v formic acid) and acetonitrile (0.1 % v/v formic acid) were used as eluents. Centrifugation was done with a Heraeus Biofuge 13 benchtop centrifuge. Activities were quantified with a Capintec CRC 15R dose calibrator. Small activities in tissue samples etc. were measured using a PerkinElmer Wizard² 2480 automatic gamma counter. Radio‐TLCs were evaluated using a Bioscan radio‐TLC scanner, consisting of B‐MS‐1000 scanner, B‐EC‐1000 detector with a B‐FC‐3600GM tube.
Synthesis: SDM17‐pentynoic amide: SDM17 functional monomer was synthesized on solid support by conventional Fmoc/tBu approach, employing an ultrasound‐assisted solid‐phase peptide synthesis (US‐SPPS) protocol.17 Rink amide resin (545 mg, 0.3 mmol) was functionalized with o‐NBS‐l‐Lys(Fmoc)‐OH (363 mg, 0.6 mmol, 2 equiv) using HBTU (227 mg, 0.6 mmol, 2 equiv) and HOBt (92 mg, 0.6 mmol, 2 equiv) as coupling partners, and DIPEA (209 μL, 1.2 mmol, 4 equiv) as base, in DMF (3.5 mL). The mixture of reactants was added to the resin in a SPPS reactor and then ultrasonicated for 5 min before washing. Fmoc deprotection was carried out by irradiating the resin with ultrasound in the presence of a 20 % piperidine solution in DMF (2×1 min). The linear aminoacidic sequence was elongated by iterative cycles of the aforementioned amide bond coupling reactions and Fmoc deprotection; the completion of each step was qualitatively determined by Kaiser test or TNBS test. After loading the last amino acid, the resin‐bound peptide underwent a Tsuji‐Trost‐mediated allyl ester removal on the glutamic acid side chain. The resin was treated with a solution of tetrakis(triphenylphosphine)palladium(0) (35 mg, 0.03 mmol, 10 % mol) and DMBA (234 mg, 1.5 mmol, 5 equiv) in anhydrous THF (5 mL) for 1 h at rt under argon, and this procedure was repeated once. After being washed with DMF (3×1 min) and dichloromethane (3×1 min), the resin was suspended in a 0.06 M solution of potassium N,N‐diethyldithiocarbamate in DMF (38 mg in 3 mL of solvent) for 15 min in order to completely remove catalyst traces, and this procedure was repeated twice. At this stage, the α‐amino group of the Arg residue was released, and the cyclization was carried out by adding a solution of PyAOP (469 mg, 0.9 mmol, 3 equiv) and DIPEA (313 μL, 1.8 mmol, 6 equiv) in DMF (5 mL) and allowing the resin to shake for 12 h. Next, the α‐amino group on C‐terminal lysine residue was released by removing the ortho‐nitrobenzenesulfonyl (o‐NBS) protecting group. This deprotection was performed by adding a clear solution (4 mL) of thiophenol in dry DMF (5 % v/v) in the presence of 1.5 equiv. (relative to thiophenol) of ultrapure K₂CO₃. The obtained suspension was miniaturized by sonication and centrifuged, then the clear supernatant was added to the resin, which was allowed to shake for 10 min. This procedure was repeated a further two times and then the resin was washed exhaustively with DMF (3×1 min), MeOH (3×1 min) and CH₂Cl₂ (3×1 min). Final functionalization with the alkynyl‐bearing building block was carried out in a DMF solution (3.5 mL) of pentynoic acid (59 mg, 0.6 mmol, 2 equiv), HBTU (227 mg, 0.6 mmol, 2 equiv) and HOBt (92 mg, 0.6 mmol, 2 equiv) in the presence of DIPEA (209 μL, 1.2 mmol, 4 equiv) and irradiating with ultrasound for 5 min.
The resin was washed with DMF (2×1 min), CH₂Cl₂ (2×1 min), and diethyl ether (3×1 min), and the peptide was cleaved from the solid support using a solution of TFA/TIS (95:5, 3 mL) for 3 h at room temperature. The suspension was filtered and the crude product precipitated from the TFA solution by diluting to 35 mL with cold diethyl ether, and then centrifuged (4400 g, 15 min). The supernatant was removed, and the precipitate was suspended again in 35 mL ether as described above. The wet solid was dried for 1 h under vacuum, re‐dissolved in water/acetonitrile (9:1) and purified by RP‐HPLC (solvent A: water +0.1 % TFA; solvent B: acetonitrile +0.1 % TFA; from 10 to 60 % of solvent B over 25 min, flow rate: 10 mL min⁻¹). Product‐containing fractions were identified by ESI‐MS, concentrated in vacuo, and lyophilized. The product was characterized by analytical RP‐HPLC (solvent A: water +0.1 % TFA; solvent B: acetonitrile +0.1 % TFA; from 10 to 90 % of solvent B over 20 min, flow rate: 1 mL min⁻¹) and HRMS (ESI‐MS) spectrometry. Overall yield: 179 mg (65 %), purity: >95 %, t_R=12.45 min. MW (calcd for C₃₆H₅₇N₁₁O₁₀): 803.43. HRMS (ESI‐MS): m/z=804.43506 [M+H]⁺ (theoretical value: 804.43626; for MS spectra, see Figure S1)
TRAP(SDM17)₃: The trimer was synthesized employing a previously established method.18 SDM17‐pentynoic amide (16.1 mg, 20.0 μmol, 3.3 equiv) was added to a solution of TRAP(azide)₃ (5.0 mg, 6.1 μmol, 1 equiv) and sodium ascorbate (60 mg, 303 μmol, 50 equiv) in a mixture of water and tert‐butanol (3:1 by volumes, 400 μL). Copper(II) acetate hydrate (1.45 mg, 7.28 μmol, 1.2 equiv) was added, whereupon a brown precipitate formed immediately. Upon vortexing, the solution turned to a transparent green. The solution was allowed to react for 1 h at 60 °C without stirring. Then, all Cu species were sequestered from the TRAP(SDM17)₃ compound and the reaction solution by addition of 1,4,7‐triazacyclononane‐1,4,7‐triacetic acid (NOTA) (55 mg, 180 μmol, 30 equiv) dissolved in water (1.5 mL), adjusting to pH 2.2 by using 1 M aq. HCl, and reacted for 1 h at 60 °C. HPLC‐MS was used in all steps for monitoring of reaction progress. TRAP(SDM17)₃ was obtained as a colorless solid with a yield of 37 % (7.2 mg, 2.2 μmol). RP‐HPLC (gradient: 3–45 % MeCN in water, both containing 0.1 % trifluoroacetic acid, in 20 min, flow rate: 20 mL min⁻¹): t_R=16.6 min. MW (calcd for C₁₃₅H₂₂₅N₄₈O₃₉P₃): 3235.63. MS (ESI, positive mode): m/z=1618.8 [M+2H⁺]²⁺, 1080.2 [M+3H⁺]³⁺, 810.3 [M+4H⁺]⁴⁺, 648.6 [M+5H⁺]⁵⁺, 540.6 [M+6H⁺]⁶⁺ (theoretical values: 1618.8, 1079.5, 809.9, 648.1, 540.3; for MS spectra, see Figure S2).
Affinity assays: The integrin affinities were determined by a solid‐phase binding assay, applying a previously described protocol.31 Briefly, flat‐bottom 96‐well enzyme‐linked immunosorbent assay (ELISA) plates (BRAND, Wertheim, Germany) were coated with recombinant human LAP(TGF‐β) in carbonate buffer (15 mM Na₂CO₃, 35 mM NaHCO₃, pH 9.6) at 4 °C overnight. After washing the plates with PBS‐T buffer (phosphate‐buffered saline/Tween20), free binding sites were blocked by incubation with TS‐B buffer (Tris‐saline/BSA). Dilution series of non‐radioactive Ga‐TRAP(SDM17)₃ (20 μM to 6.4 nM) were prepared and incubated in 1 : 1 mixtures with the respective integrin. Surface‐bound integrin was detected by subsequent incubation with a specific primary antibody and a secondary peroxidase‐labeled antibody (anti‐mouse IgG‐POD, Sigma–Aldrich). After addition of the dye SeramunBlau (Seramun Diagnostic, Heidesee, Germany) and quenching of the reaction by addition of 3 M H₂SO₄, the absorbance at λ=450 nm was measured with a microplate reader (Tecan Genius, Männedorf, Switzerland). The IC₅₀ value for each compound was determined in duplicate and the inhibition curves were analyzed by using OriginPro 9.0 software. The measured IC₅₀ values were referenced to the activity of the internal standard RTDLDSLRT:48 α_vβ₆=33 nM, α_vβ₈=100 nM.
Radiochemistry: Fully automated ⁶⁸Ga labeling was done in analogy to a previously described procedure15 by using an accordingly programmed robotic system (GallElut+, Scintomics, Fürstenfeldbruck, Germany) which carried out the following steps. A ⁶⁸Ge/⁶⁸Ga‐generator with TiO₂ matrix (Eckert & Ziegler, Berlin, Germany) was eluted with 0.1 M aq. HCl. A fraction containing the highest activity (1.4 mL, ca. 500 MBq) was collected in a 5 mL conical glass vial, containing 5 or 10 nmol of TRAP(SDM17)₃ or NOTA‐SDM17, respectively, as well as 50 or 100 μL, respectively, of a solution of 4‐(2‐hydroxyethyl)‐1‐piperazineethanesulfonic acid (HEPES) buffer (2.7 M, prepared from 14.4 g HEPES and 12 mL water), resulting in a labeling pH of approximately 2 or 3, respectively. The vial was heated for 5 min to 100 °C. Purification was done by passing the reaction mixture over a solid phase extraction cartridge (SepPak C8 light), which was purged with water (10 mL). The products were eluted with ethanol (0.5 mL), followed by an ethanol/water mixture (1:1 by volumes, 1 mL). The purity of the radiolabeled compounds was determined by radio‐TLC, using silica impregnated glass fiber chromatography paper (ITLC® by Agilent) as stationary phase, and 0.1 M aq. sodium citrate or a mixture of 1 M aq. ammonium acetate and methanol (1 : 1 by volumes) as mobile phases.
To determine the n‐octanol/PBS distribution coefficients (log D_7.4), 650 μL octan‐1‐ol and 650 μL phosphate‐buffered saline (PBS, pH 7.4) were combined in a 1.5 mL Eppendorf tube. Approximately 0.5 MBq of the radiolabeled compound was added and vortexed for 2 min at 2850 rpm using a Vortex Genie2 (Scientific Industries). The samples were centrifuged (11 500 g, 10 min), after which 100 μL of the organic phase and 10 μL of the aqueous phase were taken out and the activities of the aliquots were quantified in a γ‐counter. The log D values were calculated from the quotients of the measured activities and are given as averages±standard deviation (n=10).
Cell culture: H2009 human lung adenocarcinoma cells (CRL‐5911; American Type Culture Collection (ATCC), Manassas, VA, USA) were cultivated as recommended by the distributor. Cells were subcultivated after trypsination in a ratio of 1:2–1:5, two to three times weekly in culture medium (DMEM : F12, Biochrom FG4815; 5 % fetal bovine serum, FBS Superior Biochrom S0615; 1 % ITS‐G, ThermoFisher 41400045; 4.5 mM l‐glutamine (final conc.), Biochrom K 0282; 10 nM Hydrocortisone, Sigma H0888, 10 nM β‐estradiol, Sigma E2758; penicillin/streptomycin, Biochrom A 2213).
In‐vivo studies: All animal experiments were performed in accordance with general animal welfare regulations in Germany and the institutional guidelines for the care and use of animals. Keeping of the animals, generation of respective tumor xenografts, and ex‐vivo biodistribution studies49 as well as μPET imaging50 were done following previously described protocols, which are briefly summarized below.
To generate tumor xenografts, 6‐ to 8‐week‐old female CB17 severe combined immunodeficiency (SCID) mice (Charles River, Sulzfeld, Germany) were inoculated with a maximum of 10⁷ H2009 cells (the best results were obtained with 5–7×10⁶) in Matrigel® (CultrexBME, Type 3 PathClear, Trevigen, Gentaur, Aachen, Germany; discontinued in 2019, hence switched to Geltrex™ LDEV‐Free Reduced Growth Factor Basement Membrane Matrix, A1413202, Life Technologies). Mice were used for biodistribution or PET when tumors had grown to a diameter of 8–10 mm (4–5 weeks after inoculation). β₆ Integrin immunohistochemistry (IHC) was performed as described before.37PET was recorded on a Siemens Inveon small‐animal PET system under isoflurane anesthesia. The animals were injected with between 9 and 18 MBq (200–400 pmol) of the ⁶⁸Ga‐labeled compounds into the tail vein, whereupon PET was either continuously recorded in list mode for 90 min while anesthesia was maintained (dynamic scan, reconstructed as multiple frames), or the animals were allowed to wake up with access to food and water and scanned 75 min p.i. for 20 min with refreshed anesthesia (static scan, reconstructed as single frame). Time between scans shown in Figure 5 was 1 day. Data were reconstructed using Siemens Inveon Research Workspace software, employing a three‐dimensional ordered subset expectation maximum (OSEM3D) algorithm without scatter and attenuation correction. Images of static scans were exported as maximum intensity projections (Figure 5). Time‐activity curves (Figure 3) were obtained by generating isocontour regions of interest (ROI) for the tumor and the heart content (i.e., blood), as well as defining two spherical ROIs (each 23.4 mm³) in the thigh area (muscle), followed by plotting of average activity per volume in these ROIs over time.
For biodistribution studies, the mice were administered approximately 120 pmol (3–8 MBq, depending on radiolabeling yield and decay) of the radiopharmaceuticals into the tail vein and allowed to wake up with access to food and water. For blockade, 50 nmol of TRAP(SDM17)₃ was administered 10 min before tracer injection. Animals were sacrificed 90 min after injection, blood was immediately taken from the heart with a syringe, and the organs of interest (heart, lung, liver, spleen, pancreas, stomach (empty), small intestine (empty), large intestine (empty), kidneys, adrenals, muscle, tongue, tumor, tail) were dissected. The activity in weighed tissue samples was quantified by using a γ‐counter. Injected dose per gram tissue (%ID/g) was calculated from the organ weights and counted activities.