Bioconjugate Chem. 2007, 18, 254−262
254
3,4,6-Tri-O-acetyl-2-deoxy-2-[18F]fluoroglucopyranosyl Phenylthiosulfonate: A Thiol-Reactive Agent for the Chemoselective 18F-Glycosylation of Peptides Olaf Prante,*,‡ Ju¨rgen Einsiedel,† Roland Haubner,§ Peter Gmeiner,† Hans-Ju¨rgen Wester,# Torsten Kuwert,‡ and Simone Maschauer‡ Laboratory of Molecular Imaging, Clinic of Nuclear Medicine, Friedrich-Alexander University, 91054 Erlangen, Germany, Department of Medicinal Chemistry, Emil Fischer Center, Friedrich-Alexander University, 91052 Erlangen, Germany, Universita¨tsklinik fu¨r Nuklearmedizin, Medizinische Universita¨t Innsbruck, 6020 Innsbruck, Austria, and Department of Nuclear Medicine, Klinikum rechts der Isar, Technische Universita¨t Mu¨nchen, 81675 Munich, Germany. Received October 29, 2006
3,4,5-Tri-O-acetyl-2-[18F]fluoro-2-deoxy-D-glucopyranosyl 1-phenylthiosulfonate (Ac3-[18F]FGlc-PTS) was developed as a thiol-reactive labeling reagent for the site-specific 18F-glycosylation of peptides. Taking advantage of highly accessible 1,3,4,6-tetra-O-acetyl-2-deoxy-2-[18F]fluoroglucopyranose, a three-step radiochemical pathway was investigated and optimized, providing Ac3-[18F]FGlc-PTS in a radiochemical yield of about 33% in 90 min (decay-corrected and based on starting [18F]fluoride). Ac3-[18F]FGlc-PTS was reacted with the model pentapeptide CAKAY, confirming chemoselectivity and excellent conjugation yields of >90% under mild reaction conditions. The optimized method was adopted to the 18F-glycosylation of the Rvβ3-affine peptide c(RGDfC), achieving high conjugation yields (95%, decay-corrected). The Rvβ3 binding affinity of the glycosylated c(RGDfC) remained uninfluenced as determined by competition binding studies versus 125I-echistatin using both isolated Rvβ3 and human umbilical vein endothelial cells (Ki ) 68 ( 10 nM (Rvβ3) versus Ki ) 77 ( 4 nM (HUVEC)). The whole radiosynthetic procedure, including the preparation of the 18F-glycosylating reagent Ac3-[18F]FGlc-PTS, peptide ligation, and final HPLC purification, provided a decay-uncorrected radiochemical yield of 13% after a total synthesis time of 130 min. Ac3-[18F]FGlc-PTS represents a novel 18F-labeling reagent for the mild chemoselective 18F-glycosylation of peptides indicating its potential for the design and development of 18F-labeled bioactive S-glycopeptides suitable to study their pharmacokinetics in vivo by positron emission tomography (PET).
INTRODUCTION Positron emission tomography (PET) using peptide-based radiopharmaceuticals has gained considerable importance in the field of nuclear medicine and radiopharmaceutical sciences (1, 2). In particular, the increasing knowledge about improved automated solid-phase peptide synthesis and advanced methods for structure-activity relationship studies of receptor-specific peptides accelerated the development of suitable candidates as agents for diagnostic and also therapeutic applications. Furthermore, the positron emitter fluorine-18 has been characterized as a suitable nuclide for radiochemical approaches toward peptide labeling. The short-lived positron emitter fluorine-18 has excellent physical characteristics, such as a relatively low positron energy of 635 keV, limiting the patient’s absorbed dose, and a prolonged half-life of 109.7 min allowing extended and multistep radiochemical syntheses and prolonged PET imaging protocols. These features support an exceptional position of 18F as a beneficial PET radionuclide in comparison with alternative positron emitters such as 11C or 13N. Moreover, the development of new 18F-labeled prosthetic groups for the radiofluorination of peptides strongly increased the availability of peptide-based radiopharmaceuticals for PET * Corresponding author. Mailing address: Laboratory of Molecular Imaging, Clinic of Nuclear Medicine, Friedrich-Alexander University, Schwabachanlage 6, D-91054 Erlangen, Germany. Tel: +49-09131-8534440. Fax: +49-09131-8534325. E-mail: olaf.prante@ nuklear.imed.uni-erlangen.de. ‡ Clinic of Nuclear Medicine, Friedrich-Alexander University. † Emil Fischer Center, Friedrich-Alexander University. § Medizinische Universita ¨ t Innsbruck. # Technische Universita ¨ t Mu¨nchen.
over the past years. A commonly used strategy to label peptides relies on the use of N-succinimidyl-4-[18F]fluorobenzoate ([18F]SFB) targeting the N-terminus or lysine side chains of the target peptide. According to Wester et al., [18F]SFB can be efficiently prepared by a simplified method affording an overall radiochemical yield ranging between 50% and 60% based on [18F]fluoride within a preparation time of 35 min (3). Besides [18F]SFB, 4-nitrophenyl 2-[18F]fluoropropionate proved to be an effective and suitable 18F-acylation agent. The prevalent application of these two 18F-labeled prosthetic groups has resulted in a large number of 18F-labeled peptide-based imaging agents, such as [18F](D-Phe1)octreotide (4), [18F]galacto-RGD (5), [18F]FB-PEG-RGD (6), [18F]NT(8-13) (7), [18F](Nle4, D-Phe7)-R-MSH) (8), and 18FB-AHx-insulin (9). Among these, [18F]galacto-RGD already provided promising results for the noninvasive assessment of Rvβ3 expression on tumor and endothelial cells in patients with malignant tumors by PET (10). The development of chemoselective 18F-labeling strategies that are favorable for routine large scale syntheses of shortlived 18F-labeled radiopharmaceuticals have been intensively studied in recent years. For instance, [18F]fluorothiols have been prepared for the chemoselective labeling of N-terminal chloroacetylated peptides (11). Recently, Poethko et al. developed a convenient 18F-labeling method for peptides by oxime conjugation of [18F]fluorobenzaldehyde with unprotected aminooxy peptides yielding chemoselectively labeled RGD peptides and somatostatin analogs (12). Moreover, chemoselective 18Flabeling strategies being based on the radiosynthesis of 18Flabeled maleimide derivatives as thiol-reactive reagents were investigated (13, 14). Toyokuni et al. reported the 18F-radiosynthesis of a new heterobifunctional maleimide derivative, namely, N-[4-(aminooxy)butyl]maleimide ([18F]FBABM, Figure
10.1021/bc060340v CCC: $37.00 © 2007 American Chemical Society Published on Web 12/23/2006
Chemoselective 18F-Glycosylation of Peptides
Bioconjugate Chem., Vol. 18, No. 1, 2007 255
to be well-suited for site-specific glycosylation of cysteinecontaining peptides and proteins. In this study, we present the radiosynthesis of 3,4,6-tri-O-acetyl-2-deoxy-2-[18F]fluoroglucopyranosyl phenylthiosulfonate (Ac3-[18F]FGlc-PTS, Figure 1) as a new 18F-labeled thiol-reactive glycosyl donor for 18Fglycosylation of peptides and its application for bioconjugation using the cyclic peptide c(RGDfC). Furthermore, we studied the in-vitro binding affinities of the glycosylated RGD peptide using immobilized Rvβ3 integrin and human endothelial cells.
EXPERIMENTAL PROCEDURES
Figure 1. Chemical structures of 18F-labeled thiol-reactive prosthetic groups for site-selective 18F-labeling of peptides: N-[4-(aminooxy)butyl]maleimide ([18F]FBABM (13)), 1-[3-(2-fluoropyridin-3-yloxy)propyl]pyrrole-2,5-dione ([18F]FPyME (14)), N-[2-(4-[18F]fluorobenzamido)ethyl]maleimide ([18F]FBEM (15)), and 3,4,6-tetra-O-acetyl-2deoxy-2-[18F]fluoro-glucopyranosyl phenylthiosulfonate (Ac3-[18F]FGlcPTS ([18F]2), this work).
1), within a two-step synthesis that was followed by a conjugation step with the cysteine-containing peptide in phosphatebuffered saline (13). Using this strategy, radiochemical yields (RCYs) of 70% for the model compound glutathione were possible, whereas for a 5′-thiol-functionalized oligodeoxynucleotide only 5% (decay-corrected) RCY was achieved. Meanwhile, de Bruin et al. developed the alternative pyridinyl-substituted maleimide reagent [18F]FPyME (Figure 1) for the prosthetic 18F-labeling of peptides via conjugation with a thiol function (14). A three-step process is needed for the preparation of [18F]FPyME, advantegeously making use of a nucleophilic heteroaromatic substitution in ortho position of the pyridinyl moiety, followed by cleavage of an N-BOC protective group and ultimately maleimide formation. This strategy provides a RCY of about 30% for the preparation of [18F]FPyME and a non-decay-corrected yield of 65% for the final conjugation of target peptides (14). More recently, a similar approach for the radiosynthesis of a thiol-reactive 18F-labeled maleimide starting from [18F]SFB has been described (15). [18F]SFB was isolated and subsequently converted to the benzamide derivative [18F]FBEM (Figure 1, non-decay-corrected RCY of about 5% within 150 min), which was suitable for 18F-labeling of thiolated RGD peptides. Following this 200 min procedure, the overall decay-corrected RCY was 20% including two HPLC purification steps. 2-Deoxy-2-[18F]fluoroglucose ([18F]FDG) represents by far the most frequently used radiopharmaceutical for PET imaging in nuclear medicine, not least because of its highly efficient radiosynthesis (16). Despite its very high availability, only few approaches toward a chemoselective glycoconjugation of an 18Flabeled glycosyl donor with peptides have been described in previous short communications (17, 18). Such a strategy could be beneficial for the development of 18F-labeled glycopeptides as PET tracers with improved biokinetics, since glycosylation of peptides has often been shown to enhance bioavailability and improve blood-brain barrier (BBB) permeability and in vivo clearance properties (19-21). As an extension to the available 18F-labeling agents, we recently investigated the applicability of tetra-O-acetylated 2-[18F]fluoro-2-deoxy-glucopyranose as a 18F-glycosylation reagent (22, 23). However, this approach did not include a chemoselective ligation of the glycosyl residue and suffered from the necessity of Lewis acids as promotors. The aim of this study was to develop a thiol-reactive 18Flabeled glycosyl donor for a chemoselective 18F-glycosylation approach. Besides maleimide-activated carbohydrates (24), glycosyl 1-phenylthiosulfonates (25) have recently been reported
General. All chemicals and reagents were of analytical grade and obtained from commercial sources if not stated otherwise. N,N′-Dimethylformamide (DMF), dichloromethane (CH2Cl2), acetonitrile, and dimethyl sulfoxide (DMSO) were obtained from Fluka as SureSeal bottles. Benzotriazol-1-yl-oxy-tris-pyrrolidinophosphonium hexafluorophosphate (PyBOP) and N-hydroxybenzotriazole (HOBt) were purchased from Bachem. [18F]Fluoride was obtained from PET Net GmbH (Erlangen, Germany). 125I-Echistatin was purchased from Amersham Biosciences (GE Healthcare, Germany). Purified Rvβ3 integrin was obtained as a Triton X-100 formulation from Chemicon Europe (Southampton, U.K.). Radio-thin layer chromatography (radioTLC) was carried out on plastic sheets using ethyl actetate/nhexane (1:1, v/v) as eluent (Polygram, Sil G/UV254, Macherey Nagel). Analytical HPLC was performed on an Agilent 1100 system with a quarternary pump and variable wavelength detector and radio-HPLC detector D505TR (Canberra Packard). Computer analysis of the HPLC data was performed using FLOOne software (Canberra Packard). Electronic autoradiography (Instant Imager, Canberra Packard) was used to analyze radioTLC data. Electron-spray-ionization (ESI) mass spectrometry analysis was performed using a Bruker esquire 2000 instrument. LC-MS analyses were performed on an Agilent 1100 series analytic HPLC system coupled to a Bruker esquire 2000 mass spectrometer. Matrix-assisted laser desorption/ionization timeof-flight (MALDI) mass spectrometry was performed using a MALDI-3 apparatus (Kratos Instruments) at the Department of Nuclear Medicine of the University of Heidelberg. NMR spectra were recorded on a Varian Gemini-300 system for 19F NMR (Deutero GmbH, Germany) and on a Bruker Avance 600 system for 1H NMR. The syntheses of the 18F-labeling precursor 1,3,4,6tetra-O-acetyl-2-O-trifluoromethanesulfonyl-β-D-mannopyranose (26) and of 3,4,6-tri-O-acetyl-2-deoxy-2-fluoro-R-D-glucopyranose 1-bromide (1) were described elsewhere (22). Peptide syntheses. CAKAY. The peptide synthesis was achieved by manual solid-phase techniques, when microwave irradiation (Discover reactor, CEM corporation) supported short reaction times and enhanced coupling effectivities. Fmoc-Tyr(tBu)-Wang resin (Fluka) with a loading of 0.8 mmol/g was used as the solid support incorporating amino acids as their NR-(9fluorenylmethyloxycarbonyl) (Fmoc)-protected derivatives. For the protection of the amino acid side chains, tert-butoxycarbonyl (Boc) and trityl (Trt) were used for lysine and cysteine, respectively. Elongation of the peptide chain was done by repetitive cycles of Fmoc deprotection applying 20% piperidine in DMF (microwave irradiation 5 × 5 s, 100 W), followed by five washings with DMF and subsequent peptide coupling employing 5 equiv of the corresponding Fmoc-amino acid/ PyBOP/diisopropylamine and 7.5 equiv of HOBt, dissolved in a minimum amount of DMF (irradiation 15 × 10 s, 50 W). Between every irradiation step, cooling of the reaction mixture to a temperature of -10 °C was achieved by sufficient agitation in an ethanol-ice bath. Cysteine was introduced as the pentafluorophenyl ester (Fmoc-Cys(Trt)-OPfp, Fluka) and solely used as a solution in DMF without any additive. After each acylation step, the solid support was washed five times with
256 Bioconjugate Chem., Vol. 18, No. 1, 2007
DMF. Subsequently, the N-terminal Fmoc residue was deprotected as described above, the resin was rinsed 10 times with CH2Cl2 and dried under vacuum. Cleavage from the resin was performed by using a mixture of trifluoroacetic acid (TFA)/ phenol/water/thioanisole/ethanedithiole/triisopropylsilane 81.5: 5:5:5:2.5:2 for 2 h. After evaporation of the solvent, the peptide was precipitated and washed four times in t-butyl methyl ether, while N2 was bubbled through the suspension before each washing. Purification of the crude material was done using preparative RP-HPLC (Agilent 1100 preparative series, column Zorbax 300 SB-C18, 21.2 mm × 250 mm, 7 µm particles). The eluting system was 0.1% TFA in acetonitrile (A) and 0.1% TFA in water (B) applying a linear gradient from 5.0% to 6.3% A over 15 min at a flow rate of 21.2 mL/min. Chemical purity of CAKAY was 96% as determined by HPLC (Lichrosorb RP18, 250 mm × 4.6 mm, 1.5 mL/min, 10-100% acetonitrile in water (0.1% TFA) over 40 min; tR ) 4.23 min). ESI-MS: m/z 555.3 [M + H]+. Cyclic RGD Peptides. c(RGDfC) and c(RGDfS) were synthesized by Fmoc-assisted peptide solid-phase synthesis using a trityl chloride polystyrene (TCP) resin (PepChem, Tu¨bingen, Germany). Loading of the TCP resin, synthesis of the peptide, subsequent cyclization, and deprotection were carried out following the protocol described elsewhere (5, 19, 27). Sidechain protection was 2,2,4,6,7-pentamethyldihydrobenzofuran5-sulfonyl (Pbf) for arginine, tert-butyl (tBu) for aspartic acid and serine, and trityl (Trt) for cysteine. Chemical purities of cyclic RGD peptides were g96% as determined by HPLC and LC-MS. c(RGDfS): HPLC Lichrosorb RP-18, 250 mm × 4.6 mm, 1 mL/min, 0-50% acetonitrile in water (0.1% TFA) over 30 min, tR(c(RGDfS)) ) 15.5 min; ESI-MS m/z 563.2 [M + H]+. c(RGDfC): HPLC Lichrosorb RP-18, 250 mm × 4.6 mm, 1.5 mL/min, 10-100% acetonitrile in water (0.1% TFA) over 40 min, tR ) 8.8 min; ESI-MS m/z 579.3 [M + H]+, 601.2 [M + Na]+; LC-MS 579.0 [M + H]+. Sodium Phenylthiosulfonate (NaPTS). The synthesis of NaPTS was performed as described by Gamblin et al. (25). Sodium benzenesulfinate (2 g, 12 mmol) and 390 mg (12 mmol) of sulfur were dissolved in 12 mL of anhydrous pyridine to give a yellow suspension. The reaction was stirred in a nitrogen atmosphere at room temperature for 1 h until a white suspension occurred. The mixture was filtered and washed with anhydrous diethyl ether. The crude product was crystallized from ethanol to yield 2.1 g (11 mmol, 92%) of NaPTS as a white solid. 1H NMR (360 MHz, DMSO-d6): δ 7.76-7.71 (m, 2H), 7.387.31 (m, 3H). ESI-MS: m/z 173.2 [M- Na]-, 219.1 [M + Na]+. 3,4,6-Tri-O-acetyl-2-deoxy-2-fluoro-β-D-glucopyranosyl Phenylthiosulfonate, 2 (Ac3-FGlc-PTS). 3,4,6-Tri-O-acetyl-2-deoxy2-fluoro-R-D-glucopyranosyl bromide (207 mg, 558 µmol, 1) was dissolved in 5 mL of acetonitrile and 216 mg (1.1 mmol) of NaPTS and 18 mg (558 µmol) of tetrabutylammonium bromide (TBABr) were added. The mixture was stirred at 70 °C under a nitrogen atmosphere for 1.5 h. Afterward the solvent was removed in vacuo, the residue was partitioned between CH2Cl2 and water, and the aqueous phase was re-extracted with CH2Cl2. The combined organic phases were dried over Na2SO4, and the solvent was removed in vacuo. The resulting residue was purified by flash column chromatography (ethyl acetate/ hexane 1:1) to afford 189 mg (407 µmol, 72%) of Ac3-FGlcPTS (2) as a yellow oil. TLC: ethyl acetate/hexane 1:1, Rf ) 0.6. HPLC (Lichrosorb RP-18, 250 mm × 4.6 mm, 1.5 mL/ min, 10-100% acetonitrile in water (0.1% TFA) over 40 min): tR ) 21.52 min. 1H NMR (360 MHz, DMSO-d6): δ 8.047.94 (m, 2H, Ar-H), 7.81-7.64 (m, 3H, Ar-H), 6.33 (d, 0.2H (R), H-1, J1,2 ) 5.8 Hz), 5.73 (dd, 1H (β), H-1, J1,2 ) 9.4 Hz, J1,F ) 2.0 Hz), 5.62 (dt, 1H, H-3, J3,F ) 14.8 Hz, J2,3 ) J3,4 ) 9.7 Hz), 4.85 (t(dd), 1H, H-4, J4,5 ) J4,3 ) 9.7 Hz), 4.79 (ddd,
Prante et al.
1H, H-2, J2,F ) 50.0 Hz, J2,3 ) 9.7 Hz, J2,1 ) 9.4 Hz), 4.14 (ddd, 1H, H-5, J5,4 ) 9.7 Hz, J5,6a ) 4.6 Hz, J5,6b ) 2.3 Hz), 4.02 (dd, 1H, H-6a, J6a,6b ) 12.6 Hz, J6a,5 ) 4.6 Hz), 3.68 (dd, 1H, H-6b, J6a,6b ) 12.6 Hz, J6b,5 ) 2.3 Hz), 2.03 (s, 3H, OAc), 1.98 (s, 3H, OAc), 1.95 (s, 3H, OAc). 19F NMR (DMSO-d6): δ -192.51 (dd, JF,2 ) 49.8 Hz, JF,3 ) 14.9 Hz, R-anomer), -191.93 (ddd, JF,2 ) 50.0 Hz, JF,3 ) 14.8 Hz, JF,1 ) 2.0 Hz, β-anomer); R/β ) 20:80. ESI-MS: m/z 465.0 [M + H]+, 487.0 [M + Na]+. L-(S-3,4,6-Tri-O-acetyl-2-deoxy-2-fluoro-β-D-glucopyranosyl disulfide)-cysteinyl-L-alanyl-L-lysyl-L-alanyl-L-tyrosine, 3 (C(S,S′-Ac3-FGlc)AKAY). To a solution of 1.3 mg (2.3 µmol) of CAKAY in 1 mL of buffer (50 mM Tris, pH 7.7) was added 19 mg (40 µmol) of Ac3-FGlc-PTS (2) in 100 µL of acetonitrile/ water, 8:2. After the reaction time of 30 min, the absence of free thiol was shown by Ellman’s analysis, and the product was isolated by semipreparative HPLC (Lichrosorb RP-18, 125 mm × 8 mm, 4 mL/min, 10-100% acetonitrile in water (0.1% TFA) over 40 min) and subsequently lyophilized to afford C(S,S′Ac3-FGlc)AKAY (3) (1.8 mg, 2 µmol, 90%) as a white powder. HPLC (Lichrosorb RP-18, 250 mm × 4.6 mm, 1.5 mL/min, 10-100% acetonitrile in water (0.1% TFA) over 40 min, 220 nm): tR ) 9.7 min. 1H NMR (600 MHz, DMSO-d6): δ 9.23 (b, 1H, COOH), 8.69 (d, 1H, JHN/HR ) 7.5 Hz, NH), 8.12 (d, 1H, JHN/HR ) 8.0 Hz, NH), 8.10 (d, 1H, JHN/HR ) 7.8 Hz, NH), 7.83 (d, 1H, JHN/HR ) 7.4 Hz, NH), 7.67 (b, 1H, NH2), 7.01 (m, 2H, Ar-Tyr), 6.65 (m, 2H, Ar-Tyr), 5.53 (dt, 1H, J3,F ) 14.1 Hz, J3,2 ) J3,4 ) 9.5 Hz, H-3), 5.17 (dd, 1H, J1,2 ) 9.5 Hz, J1,F ) 1.8 Hz, H-1, β-anomer), 4.94 (t(dd), 1H, J4,3 ) J4,5 ) 9.5 Hz, H-4), 4.73 (dt(ddd), 1H, J2,F ) 49.7 Hz, J2,1 ) J2,3 ) 9.5 Hz, H-2, β-anomer), 4.54 (q, 1H, JHR/Hβ ) 6.1 Hz, R-CH), 4.39 (p, 1H, J ) 7.5 Hz, R-CH), 4.32-4.07 (m, 5H, 3 × R-CH, H-5, H-6a, H-6b), 3.23 (dd, 1H, JHβ/Hβ ) 14.0 Hz, JHβ/HR ) 6.1 Hz, β-CH), 3.09 (dd, 1H, JHβ/Hβ ) 14.0 Hz, JHβ/HR ) 7.5 Hz, β-CH), 2.90 (dd, 1H, JHβ/Hβ ) 14.0 Hz, JHβ/HR ) 5.4 Hz, β-CH), 2.79 (dd, 1H, JHβ/Hβ ) 14.0 Hz, JHβ/HR ) 8.3 Hz, β-CH), 2.73 (b, 2H, H-Lys), 2.05 (s, 3H, OAc), 2.00 (s, 3H, OAc), 1.99 (s, 3H, OAc), 1.51 (d, 4H, β-CH-Lys, γ-CH-Lys), 1.29 (m, 2H, Hδ-Lys), 1.25 (d, 3H, β-CH3-Ala, JHβ/HR ) 7.0 Hz), 1.19 (d, 3H, β-CH3-Ala, JHβ/HR ) 7.0 Hz). ESI-MS: m/z 877.4 [M + H]+, 439.3 [M + 2H]2+. Cyclo[L-arginyl-L-glycyl-L-asparagyl-D-phenylalanyl-L-(S3,4,6-tri-O-acetyl-2-deoxy-2-fluoro-β-D-glucopyranosyl disulfide) cysteine], 4 (c(RGDfC(S,S′-Ac3-FGlc))). To a solution of 1.3 mg (2.2 µmol) of c(RGDfC) in 1 mL of buffer (50 mM Tris, pH 7.7) was added 18.5 mg (40 µmol) of Ac3-FGlc-PTS (2) in 100 µL of acetonitrile/water, 8:2. After a reaction time of 30 min, the absence of free thiol was shown by Ellman’s analysis, and the product was isolated as described for C(S,S′Ac3-FGlc)AKAY (3) to afford 1.8 mg (2 µmol, 90%) of c(RGDfC(S,S′-Ac3-FGlc)) (4) as a white powder. HPLC (Lichrosorb RP-18, 250 mm × 4.6 mm, 1.5 mL/min, 10-100% acetonitrile in water (0.1% TFA) over 40 min, 220 nm): tR ) 14.9 min. 1H NMR (600 MHz, DMSO-d6: δ 8.71 (1H, b, COOH-Asp), 8.42 (dd, 1H, JHN/HR ) 7.5 Hz, JHN/HR ) 4.2 Hz, NH-Gly), 8.23 (d, 1H, JHN/HR ) 7.7 Hz, NH-D-Phe), 8.09 (d, 1H, JHN/HR ) 7.2 Hz, NH-Cys), 8.06 (d, 1H, JHN/HR ) 7.2 Hz, NH-Asp), 7.73 (d, 1H, JHN/HR ) 8.2 Hz, NH-Arg), 7.44 (t(dd), 1H, J ) 5.6 Hz, HN-Arg), 7.26-7.14 (m, 7H, 5 × H-Ar (DPhe), 2 × NH/NH2-Arg), 5.50 (dt, 1H, J3,F ) 14.3 Hz, J3,2 ) J3,4 ) 9.3 Hz, H-3), 5.02 (dd, 1H, J1,2 ) 9.3 Hz, J1,F ) 1.4 Hz, H-1, β-anomer), 4.92 (t(dd), 1H, J4,3 ) J4,5 ) 9.3 Hz, H-4), 4.70 (dt(ddd), 1H, J2,F ) 49.5 Hz, J2,1 ) J2,3 ) 9.3 Hz, H-2, β-anomer), 4.63 (ddd, 1H, JHR/Hβ ) 6.0 Hz, JHR/Hβ ) 8.5 Hz, JHR/HN ) 7.2 Hz, R-CH-Asp), 4.47-4.40 (m, 2H, R-CH-D-Phe, R-CH-Cys), 4.21-4.01 (m, 5H, H-5, H-6a, H-6b, R-CH-Gly, R-CH-Arg), 3.25 (dd, 1H, JHR/HR ) 14.9 Hz, JHR/HN ) 4.2 Hz,
Chemoselective 18F-Glycosylation of Peptides
R-CH-Gly), 3.08 (m, 3H, β-CH-Cys, δ-CH2-Arg), 3.00 (dd, 1H, JHβ/Hβ ) 13.6 Hz, JHβ/HR ) 8.1 Hz, β-CH-D-Phe), 2.86 (dd, 1H, JHβ/Hβ ) 13.6 Hz, JHβ/HR ) 10.0 Hz, β-CH-Cys), 2.79 (dd, 1H, JHβ/Hβ ) 13.6 Hz, JHβ/HR ) 5.7 Hz, β-CH-D-Phe), 2.70 (dd, 1H, JHβ/Hβ ) 16.2 Hz, JHβ/HR ) 8.5 Hz, β-CH-Asp), 2.38 (dd, 1H, JHβ/Hβ ) 16.2 Hz, JHβ/HR ) 6.0 Hz, β-CH-Asp), 2.05 (s, 3H, OAc), 2.00 (s, 3H, OAc), 1.99 (s, 3H, OAc), 1.72 (m, 1H, β-CH-Arg), 1.49-1.35 (m, 3H, γ-CH2-Arg, β-CH-Arg). 19F NMR (282 MHz, DMSO-d6): δ -192.13 (ddd, JF,2 ) 49.5 Hz, JF,3 ) 14.3 Hz, JF,1 ) 1.4 Hz, β-anomer). MALDI: m/z 900.6. ESI-MS: m/z 901.3 [M + H]+, 922.7 [M + Na]+. Ellman’s Analysis. The absence of the free thiol was shown using Ellman’s test (28). An aliquot (10 µL) of the reaction mixture was diluted in 990 µL of phosphate buffer (0.1 M NaH2PO4, pH 8). Subsequently, 90 µL of this solution was added to 60 µL of Ellman’s reagent (10 mM 5,5′-dithio-bis(2nitrobenzoic acid) in phosphate buffer) in a 96-well plate. After incubation for 30 min at room temperature, the optical density was measured at 450 nm using an ELISA reader. A calibration curve was measured using cysteine as the reference in concentrations of 0, 5, 10, 20, 30, 40, 50, and 60 µM in phosphate buffer. Solid-Phase Receptor Binding Assay. The receptor binding assay was performed following the procedure described by Orlando and Cheresh (29) with slight modifications. Commercially available purified Rvβ3 (Chemicon International) was diluted at 250 ng/mL in coating buffer (25 mM Tris-HCl, 150 mM NaCl, 1 mM CaCl2, 0.5 mM MgCl2, 1 mM MnCl2, pH 7.4), and an aliquot of 100 µL/well was added to a 96-well microtiter plate (Maxisorb, Nunc, Wiesbaden, Germany) and incubated overnight at 4 °C. The wells were washed once with 200 µL of blocking buffer (coating buffer + 1% (w/v) bovine serum albumin) and incubated an additional 2 h with 200 µL of blocking buffer at room temperature. The plate was rinsed twice with binding buffer (coating buffer with 0.1% (w/v) bovine serum albumin) and incubated in the presence of varied amounts of competing ligand (0.5 pM to 500 nM echistatin; 0.05 nM to 50 µM RGD peptide) for 3 h with 370 Bq/well (0.05 nM) 125Iechistatin (Amersham, Germany) in a total volume of 100 µL of binding buffer. After the incubation time, the wells were washed three times with binding buffer, and bound 125I-echistatin was solubilized with hot 2 N NaOH. Radioactivity in the resulting suspension was measured by a γ-counter (Wallac Wizard). Nonspecific binding was measured in the presence of 200-fold molar excess of echistatin and was substracted from total binding to yield specific binding. When 125I-echistatin incubations were performed without receptor, negligible nonspecific adsorption (0.04%) to the microtiter well was detected. Each data point is an average of values from triplicate wells. All measurements were repeated at least three times. Ki values were calculated by use of the software program GraphPad Prism in consideration of the equation for “heterologous competitive binding with ligand depletion”. Cell Binding Assay. Human umbilical vein endothelial cells (HUVEC) (Cambrex, Walkersville, MD) were cultured in endothelial growth medium (EGM) (endothelial basal medium (EBM) supplemented with SingleQuotKits (1 mL/L hEGF, 0.4 mL/L hydrocortisone, 1 mL/L VEGF, 4 mL/L hFGF-B (with heparin), 1 mL/L R3-IGF-1, 1 mL/L ascorbic acid, 1 mL/L GA1000, 50 mL/L fetal calf serum, Cambrex) as described previously (30) and used in passage 5-8. For cell binding assay, HUVEC were harvested and suspended in EGM at a concentration of 500 000 cells per Eppendorf tube. Each Eppendorf tube was centrifuged at 210g for 4 min. After the cells were blocked with blocking buffer for 15 min at room temperature, they were washed twice with binding buffer. Subsequently, the HUVEC were incubated on a shaker in the presence of varied amounts
Bioconjugate Chem., Vol. 18, No. 1, 2007 257
of competing ligand (0.5 pM to 500 nM echistatin; 0.05 nM to 50 µM RGD peptide) with 370 Bq/Eppendorf tube (0.025 nM) 125I-echistatin (Amersham, Germany) in a total volume of 200 µL of binding buffer. After the incubation time of 1 h, cells were rinsed three times with binding buffer, the supernatant was sucked off, and the cell pellets with bound 125I-echistatin were transferred to counting tubes for measuring the radioactivity with a γ-counter (Wallac Wizard). Nonspecific binding was measured in the presence of 200-fold molar excess of echistatin and was substracted from total binding to yield specific binding. Each data point is an average of values from triplicate or quadruplicate wells. All measurements were repeated at least three times. Ki values were calculated by use of the software program GraphPad Prism in consideration of the equation for “heterologous competitive binding with ligand depletion”. Radiosynthesis of 3,4,6-Tetra-O-acetyl-2-deoxy-2-[18F]fluoro-glucopyranosyl Phenylthiosulfonate, [18F]2 (Ac3-[18F]FGlc-PTS). Starting from [18F]fluoride (500 MBq), 3,4,6-triO-acetyl-2-deoxy-2-[18F]fluoro-glucopyranosyl bromide ([18F]1) was synthesized in 45 min as described elsewhere (22, 23). The bromide [18F]1 (275 MBq) was fixed on a RP-18 cartridge (Merck Lichrolut, 100 mg), eluted with 1.0 mL of acetonitrile in a new reaction vessel, and dried in a stream of nitrogen. Subsequently, 10 mg (50 µmol) of NaPTS and 5 mg (15 µmol) of dry TBABr in 200 µL of anhydrous acetonitrile and 50 µL of DMF was added. The reaction mixture was stirred at 70 °C for 20 min, diluted with 250 µL of water, and transferred to a semipreparative HPLC system (Kromasil C8, 125 mm × 8 mm, 4 mL/min, 40-100% acetonitrile in water (0.1% TFA) over 40 min). Subsequent solid-phase extraction (SPE, Merck Lichrolut, 100 mg) yielded 95 MBq of Ac3-[18F]FGlc-PTS ([18F]2) after 45 min (related to [18F]fluoride). The chemical identity was confirmed by analytical gradient HPLC by reference to Ac3FGlc-PTS (2) (Lichrosorb RP-18, 250 mm × 4.6 mm, 1.5 mL/ min, 10-100% acetonitrile in water (0.1% TFA) over 40 min, tR ) 21.7 min). Radiosynthesis of L-(S-3,4,6-tri-O-acetyl-2-deoxy-2-[18F]fluoro-glucopyranosyl disulfide)-cysteinyl-L-alanyl-L-lysyl-Lalanyl-L-tyrosine, [18F]3 (C(S,S′-Ac3-[18F]FGlc)AKAY). To a solution of 95 MBq of Ac3-[18F]FGlc-PTS ([18F]2) in 100 µL of acetonitrile/buffer (1:1, 50 mM Tris, pH 7.7) was added 200 nmol of CAKAY dissolved in 100 µL of 50 mM Tris buffer (pH 7.7) at room temperature. The reaction was complete within 15 min (RCY 90-95%) as determined by analytical HPLC (Lichrosorb RP-18, 250 mm × 4.6 mm, 1.5 mL/min, 10-100% acetonitrile in water (0.1% TFA) over 40 min; tR(C(S,S′-Ac3[18F]FGlc)AKAY, [18F]3) ) 9.7 min. The chemical identity of C(S,S′-Ac3-[18F]FGlc)AKAY ([18F]3) was confirmed by analytical gradient HPLC by co-injection with the reference compound C(S,S′-Ac3-FGlc)AKAY (3). Radiosynthesis of Cyclo[L-arginyl-L-glycyl-L-asparagyl-Dphenylalanyl-L-(S-3,4,6-tri-O-acetyl-2-deoxy-2-[18F]fluoroglucopyranosyl disulfide) cysteine], [18F]4 (c(RGDfC(S,S′Ac3-[18F]FGlc))). To a solution of 95 MBq of Ac3-[18F]FGlcPTS ([18F]2) in 100 µL of acetonitrile/buffer (1:1, 50 mM Tris, pH 7.7) was added 200 nmol of c(RGDfC) dissolved in 100 µL of 50 mM Tris buffer (pH 7.7) at room temperature. The reaction was complete within 15 min (RCY 90-95%) as determined by analytical HPLC (Lichrosorb RP-18, 250 mm × 4.6 mm, 1.5 mL/min, 10-100% acetonitrile in water (0.1% TFA) over 40 min; tR(c(RGDfC(S,S′-Ac3-[18F]FGlc)), [18F]4) ) 14.9 min). The reaction mixture was diluted with 300 µL of acetonitrile/water (1:9) and subjected to semipreparative HPLC (Lichrosorb RP-18, 125 mm × 8 mm, 4 mL/min, 10-100% acetonitrile in water (0.1% TFA) over 40 min). The product fraction was diluted with water (1:10), and the product was isolated by solid-phase extraction (C18-cartridge, Merck, 100
258 Bioconjugate Chem., Vol. 18, No. 1, 2007 Scheme 1. Peptidesa
Prante et al.
Synthesis of the Glycosyl Donor Ac3-FGlc-PTS (2) and Subsequent Glycosylation Reaction Using Cysteine-Containing
a Reaction conditions: (a) NaPTS, acetonitrile, 70 °C; (b) 50 mM Tris buffer (pH 7.7, containing 7% CH3CN), 1 mM CAKAY or 1 mM c(RGDfC).
mg). After elution of the product with acetonitrile, the solvent was evaporated in vacuo, and PBS was added to obtain a solution for further experimental use. Starting from 500 MBq of [18F]fluoride, this procedure yielded about 65 MBq of c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4) after a total synthesis time of 130 min. The chemical identity of c(RGDfC(S,S′-Ac3[18F]FGlc)) ([18F]4) was confirmed by analytical gradient HPLC by co-injection with the reference compound c(RGDfC(S,S′Ac3-FGlc)) (4). Metabolic Stability. c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4) dissolved in 200 µL of PBS was added to 2 mL of human serum and incubated at 37 °C. Aliquots (100 µL) were taken at various time intervals (5-90 min) and quenched in 500 µL of methanol/ CH2Cl2 (1:1). The samples were centrifuged, and the supernatants were analyzed by radio-TLC (ethyl acetate/hexane 1:1).
RESULTS AND DISCUSSION Chemistry. Glycosyl phenylthiosulfonates have been introduced as suitable glycosyl donors for the glycoconjugation of thiols, peptides, and proteins (25). These reagents provided enhanced stability under basic conditions compared with the earlier described methanethiosulfonates (31). In this paper, we adopted the synthesis of glycosyl phenylthiosulfonates to provide the 2-deoxy-2-fluoro-glycosyl derivative 2 that served as a reference compound for the corresponding 18F-labeled analog (Ac3-[18F]FGlc-PTS ([18F]2), Figure 1). The synthesis of the glycosylating reagent 3,4,6-tri-O-acetyl-2-fluoro-2-deoxy-β-Dglucopyranosyl 1-phenylthiosulfonate (2) is outlined in Scheme 1. Starting from tetraacetylated 2-deoxy-2-fluoro-β-D-glucopyranose, which was available by acetylation of an anomeric mixture of mannose and subsequent fluorination with diethylamino sulfur trifluoride (DAST), the initial R-bromide (1) was easily obtained in 90% yield by reaction with HBr at room temperature as previously described (23). Following the procedure of Gamblin et al. (25), displacement of the halide from 1 using NaPTS in acetonitrile at 70 °C readily afforded Ac3FGlc-PTS (2) in 70% yield. Ac3-FGlc-PTS (2) consisted of a β/R anomeric ratio of >80:20 as analyzed by 19F NMR spectroscopy. Glycosylation properties of the resulting 2-deoxy-2-fluoroglycosyl donor (2) were evaluated using the model peptide CAKAY and the Rvβ3 targeting c(RGDfC). The glycosylation
of cysteine-containing peptides using glycosyl phenylthiosulfonates proceeds via a highly specific and quantitative coupling reaction of the thiol function, that is, cysteine β-SH, and the thiosulfonate moiety resulting in the formation of a disulfide bond (32). On the one hand, the peptide CAKAY was chosen to demonstrate the chemoselectivity of Ac3-FGlc-PTS (2) and also to optimize 18F-labeling conditions (see paragraph below). On the other hand, the cyclic RGD peptide was selected because of its biological relevance (33) allowing study of the influence of S-glycosylation on Rvβ3 binding affinity in vitro. Glycosylation of CAKAY and c(RGDfC) with Ac3-FGlc-PTS (2) were carried out under similar reaction conditions using an aqueous buffer system (pH 7.7; for details, see Scheme 1). Ligations proceeded rapidly within 30 min and quantitatively, as confirmed by monitoring the reaction progress by titration of residual free thiol groups with Ellman’s reagent (28). Both S-glycosylated peptides, C(S,S′-Ac3-FGlc)AKAY (3) and c(RGDfC(S,S′-Ac3-FGlc)) (4), were isolated by RP-HPLC, and their structures were confirmed by NMR spectroscopy, ESIMS, and, in the case of c(RGDfC(S,S′-Ac3-FGlc)) (4), MALDI mass spectroscopy. Importantly, the prevailing β-anomeric configuration as given by Ac3-FGlc-PTS (2) was retained in the S-glycosidic linked products C(S,S′-Ac3-FGlc)AKAY (3) and c(RGDfC(S,S′-Ac3-FGlc)) (4). Hence, the anomeric stereochemistry was maintained during the glycosylation process, which is consistent with the observation that R- and β-glycosyl thiols also do not mutarotate even under basic reaction conditions (34). Noteworthy, the determination of the anomeric configuration of such 2-deoxy-2-fluoro-glycoconjugates was simplified because of the fluorine substituent at C-2, since the 3J F,1 coupling constant provides an excellent tool to distinguish between R (3JF,1 < 0.5 Hz) and β (3JF,1 ≈ 2.5 Hz) configuration by 19F NMR spectroscopy. In Vitro Binding Assays. In order to compare the affinities of c(RGDfS), c(RGDfC), and the S-glycosylated derivative c(RGDfC(S,S′-Ac3-FGlc)) (4) to Rvβ3 integrin, competitive displacement assays were performed (Figure 2). It is well-known that the disintegrin echistatin displays high affinity for Rvβ3 and also for other integrins, such as R8β1 and R3/5/vβ1 with Ki values ranging between 1 and 4 nM (35, 36). Besides echistatin, we introduced c(RGDfS) as a competitor of 125I-echistatin binding in our experiments, since this cyclic RGD peptide has already been characterized as an inhibitor of vitronectin binding to
Chemoselective 18F-Glycosylation of Peptides
Bioconjugate Chem., Vol. 18, No. 1, 2007 259 Scheme 2. Radiosynthetic Pathway for the Chemoselective 18 F-Glycosylation of CAKAY or c(RGDfC) Using the ThiolReactive Reagent Ac3-[18F]FGlc-PTS ([18F]2)a
a Reaction conditions: (a) [K⊂222]+[18F]F-, 9 mg (20 µmol) of 1,3,4,6-tetra-O-acetyl-2-O-trifluoromethanesulfonyl-β-D-mannopyranose, 500 µL of acetonitrile, T ) 85 °C, t ) 5 min (16); (b) 33% HBr/ AcOH, rt; (c) 200 mM NaPTS, 60 mM TBABr, acetonitrile/DMF, 4:1 (v/v), T ) 70 °C, t ) 20 min; (d) 50 mM Tris buffer (pH 7.7)/acetonitrile, 3:1 (v/v), 1 mM CAKAY or 1 mM c(RGDfC).
Figure 2. Effect of the concentration of unlabeled echistatin (1), c(RGDfS) (9), c(RGDfC) (b), and c(RGDfC(S,S′-Ac3-FGlc)) (4, 2) on 125I-echistatin binding in a competition binding experiment using (a) immobilized Rvβ3 integrin or (b) human endothelial cells (HUVEC). Data are from four independent experiments each performed in triplicate or quadruplicate (n ) 12-16). Table 1. Binding Affinities of RGD Peptides to Immobilized rvβ3 and to HUVECa compound
Ki (nM) Rvβ3
Ki (nM) HUVEC
echistatin c(RGDfS) c(RGDfC) c(RGDfC(S,S′-Ac3-FGlc)) (4)
0.052 ( 0.001 53 ( 12 41 ( 0.5 68 ( 10
0.43 ( 0.11 541 ( 99 241 ( 40 77 ( 4
a Values are given as mean ( standard deviation from four independent experiments performed in triplicate or quadruplicate.
isolated Rvβ3 (IC50 ) 12 nM (27)). Therefore, c(RGDfS) represents a structurally related RGD peptide that was used as a reasonable reference compound to assess the influence of glycosylation on Rvβ3 binding affinity by comparing c(RGDfC(S,S′-Ac3-FGlc)) (4) with its non-glycosylated analog c(RGDfC). In the presence of echistatin, c(RGDfS), c(RGDfC), and c(RGDfC(S,S′-Ac3-FGlc)) (4) binding of 125I-echistatin to immobilized Rvβ3 integrin was inhibited in a concentrationdependent manner. All of the peptides demonstrated the capability to completely inhibit 125I-echistatin binding to isolated Rvβ3 integrin (Figure 2a). Due to a significant change of the free radioligand concentration during the incubation period, we used an equation for curve analysis that considered this finding. The obtained Ki values are presented in Table 1. Chen et al. also tested RGD peptides using the same in-vitro assay and found Ki values of 30-70 nM (37). Using the Rvβ3 solid-phase binding assay, we obtained Ki values within the same range for all tested cyclic peptides (column 2, Table 1), which are also comparable with Chen’s data. Moreover, these data confirm that Rvβ3 binding affinity is not substantially impaired by Sglycosylation of the cysteine moiety in c(RGDfC).
In addition, we performed a competition study using human endothelial cells, in order to qualitatively assess the binding affinity of the peptides under investigation to interfering receptors in a comparative experiment (Figure 2b). We chose human umbilical vein endothelial cells (HUVEC), since the expression of various integrins, such as R2β1, R3β1, Rvβ1, and Rvβ3 is well-known for this cell type (38). As shown in Figure 2b and Table 1, the peptides c(RGDfS) and c(RGDfC) revealed significant differences in Ki values when comparing Rvβ3 binding results to the values obtained from HUVEC binding experiments. In particular, the curve for inhibition of 125Iechistatin binding to HUVEC by c(RGDfS) showed a decreased steepness with a Hill slope factor lower than 1. This finding clearly indicated the presence of additional low-affinity binding sites for c(RGDfS) and c(RGDfC) on HUVEC. In contrast, the Ki value of the glycosylated derivative c(RGDfC(S,S′-Ac3FGlc)) (4) was determined to be almost unchanged, when comparing the Ki value obtained from competition studies using pure Rvβ3 with that of HUVEC binding experiments (Table 1). However, the assay using HUVEC cells cannot discriminate between different integrins present on the cell surface, but it provided additional information, that the affinity of the Sglycosylated peptide c(RGDfC(S,S′-Ac3-FGlc)) (4) toward 125Iechistatin binding sites was not significantly affected. Radiosynthesis of the 18F-Labeled Glycosyl Donor Ac3[18F]FGlc-PTS ([18F]2). Our radiochemical synthesis of Ac3[18F]FGlc-PTS ([18F]2) was based on the first step of the [18F]FDG synthesis (16), as shown in Scheme 2. Starting from 1,3,4,6-tetra-O-acetyl-2-O-trifluoromethanesulfonyl-β-D-mannopyranose (26), cryptate-assisted nucleophilic substitution was performed at 85 °C for 5 min following the procedure of Hamacher et al. (16). After solid phase (SiO2) purification and semipreparative RP-HPLC (22), no-carrier-added tetraacetylated 2-deoxy-2-[18F]fluoroglucose was converted into the R-bromide [18F]1 using HBr in acetic acid as previously reported (23). [18F]1 was generally obtained in a 55% non-decay-corrected yield within 45 min starting from [18F]fluoride, providing adequate quantities for the following reaction steps. The reaction conditions for the conversion of [18F]1 to the phenylthiosulfonate Ac3-[18F]FGlc-PTS ([18F]3) were studied in detail by varying the reaction solvent, the reaction time, the concentration of the promoter TBABr, and the amount of NaPTS used. As shown in Figure 3, the use of acetonitrile/DMF (4:1) resulted in highest radiochemical yields for the synthesis of Ac3-[18F]FGlc-PTS ([18F]2) (approximately 47% related to [18F]1) within a reaction time of 20 min, whereas acetonitrile alone, acetonitrile/DMSO (4:1), and DMF turned out to be less suitable. Presumably, this is due to an improved solubility of NaPTS in acetonitrile/DMF. Moreover, it turned out that important factors for an optimized
260 Bioconjugate Chem., Vol. 18, No. 1, 2007
Prante et al.
Figure 3. Dependence of the radiochemical yield (RCY, decaycorrected and related to [18F]1) of the 18F-labeled glycosyl donor Ac3[18F]FGlc-PTS ([18F]2) on the reaction solvent.
RCY are (a) the use of recrystallized and intensively dried NaPTS for the reaction with [18F]1 (in general, moisture has to be excluded, since limited RCY (100 °C should be avoided, due to an accelerated degradation of Ac3-[18F]FGlc-PTS ([18F]2) with higher temperatures, and (c) the presence of TBABr as promoter (data not shown). Under optimized reaction conditions (200 mM NaPTS, 60 mM TBABr, acetonitrile/DMF (4:1), 70 °C, 20 min), the 18F-labeled glycosylating reagent Ac3-[18F]FGlc-PTS ([18F]2) was isolated in 33% decay-corrected RCY (related to [18F]fluoride) after semipreparative HPLC. 18F-Glycosylation of CAKAY and c(RGDfC). In a set of preceding experiments in order to assess the 18F-labeling ability of Ac3-[18F]FGlc-PTS ([18F]2), varied concentrations of CAKAY were used in the conjugation step (Scheme 2, step d). As shown in Figure 4, a peptide precursor concentration of 1 mM in aqueous buffer (pH 7.7) was adequate to achieve a RCY of C(S,S′-Ac3-[18F]FGlc)AKAY) ([18F]3) of about 95% (related to Ac3-[18F]FGlc-PTS ([18F]2)) as determined by radio-HPLC within a reaction time of 15 min. With adoption of these optimized coupling conditions for the chemoselective S-glycosylation of c(RGDfC) using Ac3-[18F]FGlc-PTS ([18F]2), quantitative ligation yielding c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4) was successfully demonstrated. It should be mentioned that a possible complication in the synthesis of mixed disulfides, such as c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4), is the formation of a peptide-peptide disulfide (32). This could occur via a disulfide exchange reaction between c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4) and the peptide thiol, which is present in excess under no-carrieradded reaction conditions. However, such a reaction pathway would result in the degradation of the 18F-glycosylated product, which we did not observe by radio-HPLC analysis in our experiments, even after a prolonged reaction time of 30 min (Figure 4a). Thus, the conjugation proceeded efficiently in 95% coupling yield (decay corrected), illustrating the excellent ability of Ac3-[18F]FGlc-PTS ([18F]2) as a thiol-reactive 18F-labeling agent for cysteine-containing peptides under mild reaction conditions. Both 18F-labeled peptides, C(S,S′-Ac3-[18F]FGlc)AKAY ([18F]3) and c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4), were characterized by reference to the respective 19F-substituted standards C(S,S′-Ac3-FGlc)AKAY (3) and c(RGDfC(S,S′-Ac3FGlc)) (4) and co-injection using HPLC (Figure 5). The overallradiochemical yield of C(S,S′-Ac3-[18F]FGlc)AKAY ([18F]3) and c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4) was about 13% (decay-uncorrected and related to [18F]fluoride) after a total synthesis time of 130 min, following a three-step procedure (Scheme 2). Finally, to verify the stability of the formed glycosidic bond toward degradation, we performed incubation
Figure 4. (a) Dependence of the radiochemical yield (RCY) of C(S,S′Ac3-[18F]FGlc)AKAY ([18F]3) for 18F-glycosylation of CAKAY using Ac3-[18F]FGlc-PTS ([18F]2) on the concentration of CAKAY (10 µM (2), 0.1 mM (b), and 1 mM (9)) and (b) representative HPLC analysis of a sample withdrawn from the reaction mixture containing 1 mM CAKAY after 15 min.
experiments with c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4) using human serum at 37 °C. Radio-TLC of the incubation mixture revealed no cleavage of the carbohydrate moiety of the glycopeptide c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4), thereby indicating high in-vitro metabolic stability of c(RGDfC(S,S′Ac3-[18F]FGlc)) ([18F]4) for at least 90 min (>98%). This finding is consistent with the known resistance of thioglycosides to chemical and enzymatic hydrolysis (39). In comparison with chemoselective 18F-radiolabeling methods using maleimide derivatives (13-15), our approach provided radiochemical yields within the same range, additionally gaining the opportunity to perform glycosylation and 18F-labeling of peptide target compounds simultaneously within the same reaction step. In our study, we successfully developed a suitable radiosynthesis for the novel 18F-labeled glycosyl donor Ac3[18F]FGlc-PTS ([18F]2) and focused on the optimization of the ligation of [18F]2 with cysteine-containing peptides. In view of the rapid and chemoselective conjugation under mild reaction conditions, together with the improved stability of S-linked glycopeptides toward acid/base or enzyme-mediated hydrolysis compared with their O-linked counterparts (39), the 18Fglycosylation reaction described herein could find valuable use in the radiosynthesis of 18F-labeled S-glycopeptides to study their pharmacokinetics by PET. The development of a facilitated radiochemical procedure for an efficient automated synthesis of fully unprotected 18F-labeled glycopeptides taking advantage of this mild ligation procedure is a topic of further studies.
Chemoselective 18F-Glycosylation of Peptides
Figure 5. Analytical HPLC profiles of (a) the 18F-labeled CAKAY [18F]3 coinjected with the standard C(S,S′-Ac3-FGlc)AKAY (3) and (b) the 18F-labeled peptide c(RGDfC(S,S′-Ac3-[18F]FGlc)) ([18F]4) coinjected with the standard c(RGDfC(S,S′-Ac3-FGlc)) (4). HPLC column and elution protocol: Lichrosorb RP-18, 250 mm × 4.6 mm, 1.5 mL/ min, 10-100% CH3CN in water (0.1% TFA) over 40 min.
CONCLUSION Ac3-[18F]FGlc-PTS, a novel phenylthiosulfonate derivative of [18F]FDG, has been introduced as a new thiol-reactive 18Fglycosylation reagent for labeling of peptides. This strategy combines both site-specific 18F-labeling and glycosylation of a cysteine-containing target peptide. As illustrated by glycosylation of c(RGDfC), the in-vitro Rvβ3 binding affinity of this representative peptide remained unaffected, providing evidence for the suitability of this 18F-glycosylation method. Due to the excellent yield for the conjugation step between cysteinecontaining peptides and the 18F-labeled glycosyl phenylthiosulfonate, this method is promising for the design and development of 18F-labeled bioactive glycopeptides.
ACKNOWLEDGMENT The authors thank Dr. Walter Mier (Department of Nuclear Medicine, University of Heidelberg) for performing MALDI mass spectrometry and Stephan Schwarz (Medizinische Universita¨t Innsbruck) for excellent technical assistance. This work was supported by the ELAN-Fonds of the Friedrich-AlexanderUniversita¨t Erlangen-Nu¨rnberg (Grant 06.01.04.1).
LITERATURE CITED (1) Massoud, T. F., and Gambhir, S. S. (2003) Molecular imaging in living subjects: seeing fundamental biological processes in a new light. Genes DeV. 17, 545-580. (2) Okarvi, S. M. (2004) Peptide-based radiopharmaceuticals: future tools for diagnostic imaging of cancers and other diseases. Med. Res. ReV. 24, 357-397. (3) Wester, H. J., Hamacher, K., and Sto¨cklin, G. (1996) A comparative study of N.C.A. fluorine-18 labeling of proteins via acylation and photochemical conjugation. Nucl. Med. Biol. 23, 365-372. (4) Wester, H. J., Brockmann, J., Ro¨sch, F., Wutz, W., Herzog, H., Smith-Jones, P., Stolz, B., Bruns, C., and Sto¨cklin, G. (1997) PET-
Bioconjugate Chem., Vol. 18, No. 1, 2007 261 pharmacokinetics of 18F-octreotide: A comparison with 67Ga-DFOand 86Y-DTPA-octreotide. Nucl. Med. Biol. 24, 275-286. (5) Haubner, R., Kuhnast, B., Mang, C., Weber, W. A., Kessler, H., Wester, H. J., and Schwaiger, M. (2004) [18F]Galacto-RGD: synthesis, radiolabeling, metabolic stability, and radiation dose estimates. Bioconjugate Chem. 15, 61-69. (6) Chen, X., PArk, R., Hou, Y., Khankaldyyan, V., Gonzales-Gomez, I., Tohme, M., Bading, J. R., Laug, W. E., and Conti, P. S. (2004) MicroPET imaging of brain tumor angiogenesis with 18F-labeled PEGylated RGD peptide. Eur. J. Nucl. Med. Mol. Imaging 31, 10811089. (7) Bergmann, R., Scheunemann, M., Heichert, C., Mading, P., Wittrisch, H., Kretzschmar, M., Rodig, H., Tourwe´, D., Iterbeke, K., Chavatte, K., Zips, D., Reubi, J. C., and Johannsen, B. (2002) Biodistribution and catabolism of 18F-labeled neurotensin(8-13.) analogs. Nucl. Med. Biol. 29, 61-72. (8) Vaidyanathan, G., and Zalutsky, M. R. (1994) Improved synthesis of N-succinimidyl 4-[18F]fluorobenzoate and its application to the labeling of a monoclonal antibody fragment. Bioconjugate Chem. 5, 352-356. (9) Guenther, K. J., Yoganathan, S., Garofalo, R., Kawabata, T., Strack, T., Labiris, R., Dolovich, M., Chirakal, R., and Valliant, J. F. (2006) Synthesis and in vitro evaluation of 18F- and 19F-labeled insulin: A new radiotracer for PET-based molecular imaging studies. J. Med. Chem. 49, 1466-1474. (10) Haubner, R., Weber, W. A., Beer, A. J., Vabuliene, E., Reim, D., Sarbia, M., Becker, K. F., Goebel, M., Hein, R., Wester, H. J., Kessler, H., and Schwaiger, M. (2005) Noninvasive visualization of the activated Rvβ3 integrin in cancer patients by positron emission tomography and [18F]Galacto-RGD. PLoS Med. 2, e70. (11) Glaser, M., Karlsen, H., Solbakken, M., Arukwe, J., Brady, F., Luthra, S. K., and Cuthbertson, A. (2004) 18F-fluorothiols: A new approach to label peptides chemoselectively as potential tracers for positron emission tomography. Bioconjugate Chem. 15, 14471453. (12) Poethko, T., Schottelius, M., Thumshirn, G., Hersel, U., Herz, M., Henriksen, G., Kessler, H., Schwaiger, M., and Wester, H. J. (2004) Two-step methodology for high-yield routine radiohalogenation of peptides: 18F-labeled RGD and octreotide analogs. J. Nucl. Med. 45, 892-902. (13) Toyokuni, T., Walsh, J. C., Dominguez, A., Phelps, M. E., Barrio, J. R., Gambhir, S. S., and Satyamurthy, N. (2003) Synthesis of a new heterobifunctional linker, N-[4-(aminooxy)butyl]maleimide, for facile access to a thiol-reactive 18F-labeling agent. Bioconjugate Chem. 14, 1253-1259. (14) de Bruin, B., Kuhnast, B., Hinnen, F., Yaouancq, L., Amessou, M., Johannes, L., Samson, A., Boisgard, R., Tavitian, B., and Dolle´, F. (2005) 1-[3-(2-[18F]Fluoropyridin-3-yloxy)propyl]pyrrole-2,5dione: design, synthesis, and radiosynthesis of a new [18F]fluoropyridine-based maleimide reagent for the labeling of peptides and proteins. Bioconjugate Chem. 16, 406-420. (15) Cai, W., Zhang, X., Wu, Y., and Chen, X. (2006) A thiol-reactive 18 F-labeling agent, N-[2-(4-18F-fluorobenzamido)ethyl]maleimide, and synthesis of RGD peptide-based tracer for PET imaging of Rvβ3 integrin expression. J. Nucl. Med. 47, 1172-1180. (16) Hamacher, K., Coenen, H. H., and Sto¨cklin, G. (1986) Efficient stereospecific synthesis of no-carrier-added 2-[18F]-fluoro-2-deoxyD-glucose using aminopolyether supported nucleophilic substitution. J. Nucl. Med. 27, 235-238. (17) Prante, O., Hamacher, K., and Coenen, H. H. (1999) Chemoenzymatic n.c.a. synthesis of the coenzyme uridine diphospho-2deoxy-2-[18F]fluoro-R-D-glucose. J. Labelled Compd. Radiopharm. 42, S111. (18) Berndt, M., Pietzsch, J., Bergmann, R., and Wuest, F. (2006) A novel role for [18F]FDG: Synthesis and application of a [18F]FDGbased prosthetic group for peptide and protein labeling. J. Nucl. Med. 47, 29P. (19) Haubner, R., Wester, H. J., Burkhart, F., Senekowitsch-Schmidtke, R., Weber, W., Goodman, S. L., Kessler, H., and Schwaiger, M.
262 Bioconjugate Chem., Vol. 18, No. 1, 2007 (2001) Glycosylated RGD-containing peptides: Tracer for tumor targeting and angiogenesis imaging with improved biokinetics. J. Nucl. Med. 42, 326-336. (20) Schottelius, M., Rau, F., Reubi, J. C., Schwaiger, M., and Wester, H.-J. (2005) Modulation of pharmacokinetics of radioiodinated sugar-conjugated somatostatin analogues by variation of peptide net charge and carbohydration chemistry. Bioconjugate Chem. 16, 429437. (21) Egleton, R. D., Davis, T. P. (2005) Development of neuropeptide drugs that cross the blood-brain barrier. NeuroRx 2, 44-53. (22) Maschauer, S., Pischetsrieder, M., Kuwert, T., and Prante, O. (2005) Utility of 1,3,4,6-tetra-O-acetyl-2-deoxy-2-[18F]fluoroglucopyranoside for no-carrier-added 18F-glycosylation of amino acids. J. Labelled Compd. Radiopharm. 48, 701-719. (23) Maschauer, S., Kuwert, T., and Prante, O. (2006) 18F-glycosylation using Koenigs-Knorr conditions: A comparative study. J. Labelled Compd. Radiopharm. 49, 101-108. (24) Ni, J., Singh, S., and Wang, L. X. (2003) Synthesis of maleimideactivated carbohydrates as chemoselective tags for site-specific glycosylation of peptides and proteins. Bioconjugate Chem. 14, 232238. (25) Gamblin, D. P., Garnier, P., Ward, S. J., Oldham, N. J., Fairbanks, A. J., and Davis, B. G. (2003) Glycosyl phenylthiosulfonates (GlycoPTS): Novel reagents for glycoprotein synthesis. Org. Biomol. Chem. 1, 3642-3644. (26) Hamacher, K. (1984) Phase-transfer catalysed synthesis of 4-Sβ-D-glucopyranosyl-4-thio-D-glucopyranose (thiocellobiose) and 2-Sβ-D-glucopyranosyl-2-thio-D-glucopyranose (thiosophorose). Carbohydr. Res. 128, 291-295. (27) Haubner, R., Gratias, R., Diefenbach, B., Goodman, S. L., Jonczyk, A., and Kessler, H. (1996) Structural and functional aspects of RGDcontaining cyclic pentapeptides as highly potent and selective integrin Rvβ3 antagonists. J. Am. Chem. Soc. 118, 7461-7472. (28) Ellman, G. L., Courtney, K. D., Andres, V., Jr., and FeatherStone, R. M. (1961) A new and rapid colorimetric determination of acetylcholinesterase activity. Biochem. Pharmacol. 7, 88-95.
Prante et al. (29) Orlando, R. A., and Cheresh, D. A. (1991) Arginine-glycineaspartic acid binding leading to molecular stabilization between integrin Rvβ3 and its ligand. J. Biol. Chem. 266, 19543-19550. (30) Maschauer, S., Prante, O., Hoffmann, M., Deichen, J. T., and Kuwert, T. (2004) Characterization of 18F-FDG uptake in human endothelial cells in vitro. J. Nucl. Med. 45, 455-460. (31) Davis, B. G., Maughan, M. A. T., Green, M. P., Ullman, A., and Jones, J. B. (2000) Glycomethanethiosulfonates: Powerful reagents for protein glycosylation. Tetrahedron: Asymmetry 11, 245-262. (32) Wynn, R., and Richards, F. M. (1995) Chemical modification of protein thiols: Formation of mixed disulfides. Methods Enzymol. 251, 351-356. (33) Ruoslahti, E., and Pierschbacher, M. D. (1987) New perspectives in cell adhesion: RGD and integrins. Science 238, 491-497. (34) MacDougall, J. M., Zhang, X. D., Polgar, W. E., Khroyan, T. V., Toll, L., and Cashman, J. R. (2004) Design, chemical synthesis, and biological evaluation of thiosaccharide analogues of morphine- and codeine-6-glucuronide. J. Med. Chem. 47, 5809-5815. (35) Thibault, G. (2000) Sodium dodecyl sulfate-stable complexes of echistatin and RGD-dependent integrins: A novel approach to study integrins. Mol. Pharmacol. 58, 1137-1145. (36) Kumar, C. C., Nie, H., Rogers, C. P., Malkowski, M., Maxwell, E., Catino, J. J., and Armstrong, L. (1997) Biochemical characterization of the binding of echistatin to integrin avb3 receptor. J. Pharmacol. Exp. Ther. 283, 843-853. (37) Chen, X., Hou, Y., Tohme, M., Park, R., Khankaldyyan, V., Gonzales-Gomez, I., Bading, J. R., Laug, W. E., and Conti, P. S. (2004) Pegylated Arg-Gly-Asp peptide: 64Cu labeling and PET imaging of brain tumor Rvβ3-integrin expression. J. Nucl. Med. 45, 1776-1783. (38) Luscinskas, F. W., and Lawler, J. (1994) Integrins as dynamic regulators of vascular function. FASEB J. 8, 929-938. (39) Pachamuthu, K., and Schmidt, R. R. (2006) Synthetic routes to thiooligosaccharides and thioglycopeptides. Chem. ReV. 106, 160-187. BC060340V
