Design and Synthesis of a Short-Chain Bitistatin Analogue for Imaging

1068

Bioconjugate Chem. 2004, 15, 1068−1075

Design and Synthesis of a Short-Chain Bitistatin Analogue for Imaging Thrombi and Emboli Kwamena E. Baidoo,*,† Linda C. Knight,‡ Kuo-Shyan Lin,† Jerome L. Gabriel,§ and Jan E. Romano‡ Department of Environmental Health Sciences, 615 North Wolfe Street, Rm. E6632, Baltimore, Maryland 21205, Department of Diagnostic Imaging, Temple University School of Medicine, Broad and Ontario Streets, Philadelphia, Pennsylvania 19140, and Biochemistry Department, Temple University School of Medicine, Broad and Ontario Streets, Philadelphia, Pennsylvania 19140.. Received February 24, 2004; Revised Manuscript Received June 10, 2004

Previously, we showed that labeled bitistatin analogues possessed excellent characteristics for imaging both deep-vein thrombosis and pulmonary embolism. We hypothesized that the N-terminal amino acid sequence of bitistatin, which is different from other disintegrins, likely interacts with the binding site of platelets to confer desirable properties to bitistatin for imaging. In this study, we present the design, synthesis, and initial biological testing of a short-chain analogue of the native 83-amino-acid bitistatin sequence. Our initial molecular modeling of the binding loop of bitistatin showed that the minimal sequence that represented the binding region was a cyclic 10 amino acid sequence cyclo[CysArg-Ile-Ala-Arg-Gly-Asp-Trp-Asn-Cys(S)]. Systematic modeling of a truncated N-terminal sequence of bitistatin fused with the optimized binding region having a thioether sequence through a Gaba spacer ultimately yielded the 24-amino acid peptide, cyclo-[CH2CO-Arg-Ile-Ala-Arg-GlyAsp-Trp-Asn-Cys(S-)]-Gaba-Gly-Asn-Glu-Ile-Leu-Glu-Gln-Gly-Glu-Asp-Ser-Asp-SerLys-OH, 1. The peptide was then coupled to the hydrazino-nicotinic acid bifunctional chelating agent and the purified adduct labeled with 99mTc using tricine as a coligand. Binding of the unlabeled and labeled peptide to stimulated human platelets was assayed in vitro. The 99mTc labeling yield was > 90%. The in vitro binding assays showed that the IC50 for inhibition of platelet aggregation was 3694 nM, while the Kd of the 99mTc labeled peptide was 185 nM, indicating moderate affinity for the receptor. The 99mTc-labeled peptide was able to identify sites of experimental thrombi and emboli in a canine model. The results suggest initial success in attempting to mimic the behavior of bitistatin for imaging thrombi and emboli.

INTRODUCTION 1

Bitistatin is an 83-amino-acid peptide (Figure 1) isolated from the venom of Bitis arietans (1). The peptide binds to the RIIbβ3 (Glycoprotein IIb/IIIa) receptor on platelets that is involved in the process of platelet aggregation (2). Previously, we showed that radiolabeled analogues of bitistatin gave positive images of deep-vein thrombosis and pulmonary embolism in a single test (3, * To whom correspondence should be addressed: Telephone: (410)-955-7706. Fax: (410) 955-6222. E-mail: [email protected]. † The Johns Hopkins Medical Institutions. ‡ Department of Diagnostic Imaging, Temple University School of Medicine. § Biochemistry Department, Temple University School of Medicine. 1 Abbreviations: Boc-Lys(2Cl-Z)-PAM resin ) NR-t-Boc-N(2-chloro-benzyloxycarbonyl)-L-lysine-PAM resin; Boc-Ala-OH ) NR-t-Boc-L-alanine; Boc-Arg(Tos)-OH ) NR-t-Boc-Nγ-tosyl-L-arginine; Boc-Asn(Xan)-OH ) NR-t-Boc-Nβ-xanthylL-asparagine; Boc-Asp(OcHx)-OH ) NR-t-Boc-L-aspartic acid β-cyclohexyl ester; Boc-Cys(Mbzl)-OH ) NR-t-Boc-S-p-methoxybenzyl-L-cysteine; Boc-Gaba-OH ) t-Boc-4-aminobutanoic acid (t-Boc gamma aminobutanoic acid); Boc-Gln(Xan)-OH ) NR-t-Boc-γ -xanthyl-L-glutamine; Boc-Glu(OcHx)-OH ) NR-t-Boc-L-glutamic acid γ-cyclohexyl ester; Boc-Gly-OH ) NR-t-Boc-glycine; Boc-Ile-OH ) NR-t-Boc-L-isoleucine; BocLeu-OH ) NR-t-Boc-L-leucine; Boc-Ser(Bzl)-OH ) NR-t-BocO-benzyl-L-serine; Boc-Trp(CHO)-OH ) NR-t-Boc-NR-formylL-tryptophan; Gaba ) gamma aminobutanoic acid; t-Boc- ) tert-butyloxycarbonyl-.

4). Deep-vein thrombosis and pulmonary embolism are often different halves of the same thromboembolic disease since a clot in the extremities can dislodge and relocate to the lungs. A test that can provide complete information on thromboembolic disease relatively quickly and noninvasively, as can be provided by radiolabeled bitistatin analogues through scintigraphy, would greatly impact patient management. For this reason, many efforts have been made to develop radiolabeled tracers for imaging thromboembolic disease. Many of these have been reviewed in a recent article (5). None of the tracers to date consistently provide high-contrast images of both deepvein thrombosis and pulmonary embolism in a single test. Because of the great potential of labeled bitistatin, it would be greatly beneficial if short-chain synthetic analogues that mimic the natural peptide were available. In this study, we have designed and synthesized a 24amino acid analogue, labeled it with technetium-99m, and evaluated its ability to image experimental deep-vein thrombosis and pulmonary emboli. EXPERIMENTAL SECTION

General Methods and Materials. All chemicals were obtained from either the Aldrich Chemical Co. (Milwaukee, WI) or Sigma Chemical Company (St. Louis, MO). Solvents and chemicals were reagent grade and used as received without further purification. The bifunctional chelating agent, succinimidyl-6-hydrazinopyridine-3-carboxylic acid, was synthesized according to literature procedures (6).

10.1021/bc049954v CCC: $27.50 © 2004 American Chemical Society Published on Web 08/28/2004

Short Chain Bitistatin Analogues

Bioconjugate Chem., Vol. 15, No. 5, 2004 1069

Figure 1. Amino acid sequences of selected disintegrins. The sequences are aligned so that RGD domains (shown in bold, KGD for Barbourin) are aligned with bitistatin. Asterisks (***) indicate the position of the RGD or analogue.

Male CD-1 mice were purchased from the Charles River Laboratories (Charles River, MA). Amino acid analyses and MALDI mass-spectral analyses were performed by AnaSpec, Inc. (San Jose, CA). HPLC was performed with a Waters Chromatography Division (Milford, MA) HPLC System equipped with two model 510EF pumps, a model 680 automated gradient controller, a model 490 UV absorbance detector, and a Bioscan NaI scintillation detector connected to a Bioscan Flow-count System. The output from two channels of the UV detector and the output from the Flow-count system were fed into a Gateway 2000 P5-133 computer fitted with an IN/US System, Inc. (Tampa, FL) computer card. HPLC acquisition and analysis were performed with the Winflow software from IN/US. The semipreparative column (C18 Novapak cartridge, 25 mm × 10 cm, 6 µm), the analytical column (C18 Novapak cartridge, 8 mm × 10 cm, 4 µm), and the C18 Light Sep-Pak cartridges were purchased from Waters Chromatography Division. Reversed-phase HPLC solvents consisted of acetontrile containing 0.1% trifluoroacetic acid (solvent A) and water containing 0.1% trifluoroacetic acid (solvent B). The 99mTc activity of tissues from biodistribution studies was counted on an automated gamma counter (Pharmacia-Wallac model 1282). The 99mTc activity for synthetic reactions and doses for biodistribution was measured in a Capintec CRC-7 Dose Calibrator. Molecular Modeling. All molecular modeling was performed on a Silicon Graphics Personal IRIS 4D/25 workstation using Biograf (Accelrys, San Diego, CA), which incorporates the DREIDING II all atom force field. All peptides were constructed using the peptide builder, while all nonpeptide components were similarly constructed using the organic builder; both routines are contained within the main Biograf program. Partial charges for each atom used in calculating the electrostatic energy contribution of the total energy for each structure were derived by the charge equilibration method of Rappe

and Goddard (7). Energy calculations for each peptide or construct consisted of energy minimization and quenched molecular dynamics calculations for 50 ps., followed by energy minimization of the lowest-energy structure obtained from the molecular dynamics simulation. The model of bitistatin used has previously been described (4). Briefly, the model was developed by homology modeling with kistrin, another disintegrin, whose structure has been determined by NMR and chemical analysis. The amino acid sequence of bitistatin is 63% identical and 75% similar to the sequence of kistrin (Figure 1). The amino acid sequences of bitistatin and kistrin were aligned, and then the NMR-derived coordinates of the backbone atoms of kistrin were used as a starting point. Computer site-directed mutagenesis was used to replace each amino acid side chain of kistrin with the homologous side chain of bitistatin. Residues present in bitistatin, but not present in kistrin, were added to the model. The pairing of cysteine residues in kistrin was assumed to be the same in bitistatin. The two additional cysteines present in bitistatin at positions 5 and 24 were assumed to form a crosslink. Subsequent disulfide mapping studies confirmed these disulfide pairings in bitistatin (8). The part of the molecule that is identical to kistrin was kept constant while the domains added as bitistatin were subjected to energy minimization calculations. The entire molecule was then subjected to energy minimization. The RGD loop of bitistatin ranges from C60 to C79, and contains another Cys residue at position 72 branching to an adjacent disulfide loop (4). According to the previous model of bitistatin, a flattened loop structure stretches from A56 to T73 (4). For initial modeling, the minimum number of amino acids needed to simulate the native binding site was determined from systematic deletion of amino acids from either end of the binding loop of native bitistatin. A series of disulfide-constrained

1070 Bioconjugate Chem., Vol. 15, No. 5, 2004

Baidoo et al.

Figure 2. Minimum binding sequence of bitistatin. The underlined residue in each sequence has been deleted in the subsequent peptide.

peptides was modeled, retaining Cys on each end for cyclization. Starting with the sequence found in native bitistatin, C-R-I-A-R-G-D-W-N-D-D-Y-C, the series of peptides shown in Figure 2 was modeled. To select peptides or constructs for further study, each modeled structure was overlaid onto the structure of the corresponding amino acids of our model of natural Bitistatin shown in Figure 3A (4) while calculating the root-mean-square standard deviation (rms value) between the modeled structures. Structures with the lowest rms value, suggesting the best “goodness of fit”, were selected for further study. In addition to the RGD-containing loop, bitistatin contains an N-terminal domain of 10-12 amino acids that other related disintegrins lack as depicted in Figure 1. We hypothesized that this N-terminal domain may be responsible for the improved binding properties of bitistatin in vivo. Thus, a second domain was modeled to mimic this section of bitistatin. In native bitistatin, the sequence of the N-terminal domain is S-P-P-V-C-GN-E-I-L and it is attached to the rest of the molecule by the intervening 47 amino acids and a disulfide bond between C5 and C24. Modeling was used to design a domain that would contain a minimum of amino acids, which could be attached to the single RGD-containing loop peptide modeled above, and which would simulate the tertiary structure of the native N-terminal domain relative to the native RGD-containing domain. Molecular modeling resulted in the 24 amino acid sequence 1 (Scheme 1), the model of which is depicted in Figure 3B, as a good fit to the structure of bitistatin. Molecular modeling was also performed on the bifunctional chelator-modified peptide, with and without technetium and tricine, to confirm that these modifications did not adversely affect the orientation of the peptide domains (Figure 4). Synthesis of the Short Chain Bitistatin Analogue. cyclo-[CH2CO-Arg-Ile-Ala-Arg-Gly-Asp-Trp-AsnCys(S-)]-Gaba-Gly-Asn-Glu-Ile-Leu-Glu-GlnGly-Glu-Asp-Ser-Asp-Ser-Lys-OH, 1, was synthesized using the BOC technology on an Advanced ChemTech ACT90 automated peptide synthesizer starting with 1.75 g (1.4 mmol, 0.8 mmol/g) Boc-Lys(2Cl-Z)-PAM resin according to Scheme 1. Each Boc-protected amino acid (9 mmol), including Boc-Ser(Bzl)-OH, Boc-Asp(OcHx)OH, Boc-Glu(OcHx)-OH, Boc-Gly-OH, Boc-Gln(Xan)OH, Boc-Leu-OH, Boc-Ile-OH, Boc-Asn(Xan)-OH, Boc-Gaba-OH, Boc-Cys(Mbzl)-OH, Boc-Trp(CHO)OH, Boc-Arg(Tos)-OH, and Boc-Ala-OH was reacted with an equimolar mixture of N-hydroxybenzotriazole (HOBt) and 1,3-diisopropylcarbodiimide (DIC) before coupling to the growing sequence. Deprotection of the Boc group was performed in the presence of 35% trifluoroacetic acid, 3% anisole, 0.2% dithiothreitol in dichlo-

Figure 3. Molecular model of bitistatin (A) compared to the peptide bitistatin analogue (B). The dashed area shows the binding domain.

romethane (2 times; first time 5 min, second time 25 min). The coupling time per cycle was 30 min. Coupling was repeated if necessary after a ninhydrin test was performed to ascertain completeness of the reaction. The coupling of bromoacetic acid to the sequence followed similar procedures for the amino acids, but used only 1,3diisopropylcarbodiimide for activation. At the end of chain lengthening, the resin was dried by high vacuum and the peptide was cleaved from the resin with condensed hydrogen fluoride in the presence of p-cresol at 0 °C for 1.5 h. The HF was removed under a stream of nitrogen and the residue was washed with ether (3 × 30 mL), then dissolved with 10% acetic acid (100 mL), followed by filtration and lyophilization. Cyclization was performed by dissolving 0.5 g of lyophilized powder in 1000 mL water, adjusting the pH to 8.2 with 2 M ammonium hydroxide, and stirring the mixture under nitrogen at room temperature for 24 h. The mixture was then lyophilized again. To remove the formyl (CHO) group on Trp, the mixture was dissolved in water and the pH was adjusted to 11 with 2.5 N sodium hydroxide for 5 min and then readjusted to 7 with 3 N hydrochloric acid. The resulting peptide was purified

Short Chain Bitistatin Analogues

Bioconjugate Chem., Vol. 15, No. 5, 2004 1071

Scheme 1. Synthesis of the Target Peptide

Scheme 2. Synthesis of the

99m

Tc Labeled Peptide

Figure 4. Design of fusion peptide allows for placement of Tc complex on the opposite side of molecule from binding face.

by semipreparative reversed-phase HPLC. The yield was 18%. MALDI MS: calc. MW for expected product 2716.2, found [M + H]+ 2717.1. Amino acid analysis: calc./found Asp (5/5.03), Ser (2/ 1.42*), Glu (4/4.29), Gly (3/2.77), Arg (2/2.24), Ala (1/0.93), Cys (1/0.03*), Lys (1/1.0), Ile (2/1.84), Leu (1/ 0.95), Trp

(1/0**), and Gaba (1/1.0). (*Ser and Cys were partially destroyed during hydrolysis. **Trp was completely destroyed during hydrolysis). Coupling of the Peptide to Succinimidyl-6-hydrazinopyridine-3-carboxylic Acid. The bifunctional chelating agent, succinimidyl-6-hydrazinopyridine-3-carboxylic acid (6), was coupled to the peptide according to Scheme 2. A DMF solution of succinimidyl-hydrazinonicotinate‚HCl (210 mg/mL, 100 µL) was added to a solution of compound 1 (5 mg, 1.8 µmol) in borate buffer (0.4 mL, 0.1 M, pH 9) and the pH was adjusted to 9 with triethylamine. The mixture was incubated at room temperature for 1 h and then extracted with ether (3 × 2 mL). The aqueous fraction was chromatographed by HPLC with a linear gradient from solvent B (100%) to solvent A (80%)/solvent B (20%) over the course of 60 min at a flow rate of 6 mL/min using the semipreparative column monitored on-line for UV absorption at 220 nm. 99m Tc Labeling of the Peptide. The 99mTc derivative of cyclo-[CH2CO-Arg-Ile-Ala-Arg-Gly-Asp-TrpAsn-Cys(S-)]-Gaba-Gly-Asn-Glu-Ile-Leu-GluGln-Gly-Glu-Asp-Ser-Asp-Ser-Lys-OH, was synthesized according to Scheme 2. To a solution of the coupled peptide in ethanol (10 µg in 0.2 mL) a solution of tricine in water (50 mg, 0.4 mL) was added followed by a solution of freshly prepared stannous chloride

1072 Bioconjugate Chem., Vol. 15, No. 5, 2004

dissolved in HCl(aq) (20 µg in 20 µL 0.1 M HCl). The mixture was vortexed for 1 min. To this mixture was added 99mTc-pertechnetate in saline solution (10 mCi, 0.4 mL) and the reaction mixture was vortexed for an additional 1 min followed by incubation at 50 °C for 20 min. The product was purified by solid-phase extraction using a C-18 Sep-Pak cartridge. The Sep-Pak cartridge was first washed with ethanol (5 mL) followed by a solution of ammonium acetate (0.05 M, 5 mL). The reaction mixture containing the 99mTc-labeled cyclo[CH2CO-Arg-Ile-Ala-Arg-Gly-Asp-Trp-AsnCys(S-)]-Gaba-Gly-Asn-Glu-Ile-Leu-Glu-GlnGly-Glu-Asp-Ser-Asp-Ser-Lys-OH was diluted five times with ammonium acetate (0.05 M) and then passed through the Sep-Pak cartridge. The Sep-Pak cartridge was washed with an ammonium acetate solution (0.05 M, 5 mL) followed by elution of the 99mTclabeled cyclo-[CH2CO-Arg-Ile-Ala-Arg-Gly-AspTrp-Asn-Cys(S-)]-Gaba-Gly-Asn-Glu-Ile-LeuGlu-Gln-Gly-Glu-Asp-Ser-Asp-Ser-Lys-OH with ethanol (300 µL). The yield of product was >90%. Inhibition of Platelet Aggregation. Blood was obtained from anonymous aspirin-free human donors according to an approved Institutional Review Board protocol and anticoagulant (0.1 vol, 3.8% sodium citrate) was added. The blood was centrifuged (160 x g, 12 min) to obtain platelet-rich plasma. The concentration of platelets in human platelet-rich plasma was determined using a hemacytometer, then diluted with the donor’s own plasma to obtain a platelet concentration of 300 000/ µL. For each measurement, 400 µL of diluted plateletrich plasma was placed in a cuvette and stirred at 37 °C. Vehicle (saline) or various concentrations of peptide were added in a volume of 10 µL and stirred at 37 °C for 1 min while recording visible light transmission. ADP was then added (final concentration of 10 µM) to induce platelet aggregation. Platelet aggregation was determined by change in light transmission through the platelet-rich plasma, which was recorded for 5 min after ADP addition. The IC50 was determined (by least-squares fitting) as the concentration of peptide required to produce 50% inhibition of the response to ADP in the presence of the vehicle. Results were normalized to the results for the tetrapeptide Arg-Gly-Asp-Ser (RGDS). Equilibrium Platelet Binding Assay. Platelet-rich plasma was obtained as described above. Plasma proteins were then removed by filtering the platelets through Sepharose 2B (80 × 25 mm equilibrated with an elution buffer containing 137 mM NaCl, 2.7 mM KCl, 0.1% glucose, 0.2% bovine serum albumin, 0.4 mM NaH2PO4, 12 mM NaHCO3, and 10 mM HEPES, pH 7.45). Fractions containing the highest concentrations were pooled, diluted to a concentration of 108 platelets/mL, and used within 1 h of preparation. Aliquots of the platelets were stimulated with 19 µM ADP, combined with a radioligand (99mTc labeled peptide or 125I labeled rBitistatin) over a range of concentrations, and then incubated for 1 h at 37 °C. Nonspecific binding was determined in the presence of 9 mM EDTA (3). Platelet-bound radioligand was separated from free radioligand by centrifuging (13 000 x g, 3 min) the suspension through 30% (W/V) sucrose. Scatchard analysis was used to analyze the data (9). Biodistribution. The ethanolic solution of 99mTclabeled peptide from the Sep Pak separation was diluted with physiologic saline to make a final injectate (10 µCi/ mL) containing < 0.5% ethanol. Normal male CD-1 mice weighing 24.8 ( 1.3 g were used for biodistribution studies. The labeled peptide (2 µCi, 200 µL) was injected into mice via the tail vein. At

Baidoo et al.

15, 30, 60, and 180 min postinjection, four mice for each time point were euthanized by cervical dislocation. Major organs were dissected, weighed, and counted in an automated gamma counter with aliquots of the injectate (as standards) counted at the same time. The percent injected dose per organ (% ID/organ) and the percent injected dose per gram (% ID/gram) were calculated. The % ID in blood was estimated with the assumption that the total blood volume is 6.5% of the body weight (10). Imaging Experimental Thrombi. Radiolabeled peptide was evaluated in a standardized canine model of deep-vein thrombosis and pulmonary embolism (3, 4). Briefly, under isoflurane anesthesia, an 8-mm embolization coil was placed in the femoral vein of the dog for deep-vein thrombosis while two 5-mm coils were released in the inferior vena cava for pulmonary embolism. The lesions were allowed to develop and mature for 24 h. The 99m Tc-labeled-peptide (8 mCi) was injected through a foreleg vein. Imaging of 99mTc was performed for 4 h using a large-field-of-view camera (General Electric Medical Systems, Milwaukee) interfaced to a Macintosh computer (Apple, Cupertino, CA) and a data acquisition and processing board (NucLear Mac, Scientific Imaging, Denver, CO). Regions of interest were drawn around the thrombus in the leg and compared to a region of about the same size around adjacent muscle, and also around the vessel in the opposite leg corresponding to the location of the thrombus. The average counts per pixel in each region were computed for the calculation of the following ratios: thrombus:muscle and thrombus:contralateral site. Similarly, for the chest images, regions for embolus, adjacent lung, heart, and muscle (behind the lungs, near the vertebral column) were drawn to obtain embolus:lung, embolus:heart, and embolus-tomuscle ratios. The presence of clots in the lungs and leg were confirmed at autopsy. RESULTS AND DISCUSSION

Bitistatin exhibits superior imaging characteristics among radioiodinated disintegrins evaluated for the imaging of experimental thrombosis (3). The 123I-labeled analogue showed high thrombus uptake, low soft-tissue uptake, and high thrombus and embolus-to-blood and thrombi and emboli-to-soft-tissue differentiation. The 99m Tc analogue showed the highest thrombus-to-blood and thrombus-to-soft-tissue ratios of any 99mTc-labeled thrombus and/or embolus imaging agent studied (4). Radiolabeled bitistatin analogues, therefore, offer the unique potential to generate radiopharmaceuticals capable of rapid positive imaging of both pulmonary emboli and deep-vein thrombosis in a single test. Because native bitistatin is 83 amino acids long and has multiple internal disulfide crosslinks, chemical synthesis is difficult. Thus, we embarked on an effort to generate a short chain analogue that could more conveniently be synthesized. Bitistatin contains seven internal disulfide bridges that help to maintain its tertiary structure. Initial molecular modeling of the native bitistatin sequence shown in Figure 3A (4) indicated that the required RGD binding motif is presented at the turn of a loop represented by the sequence C-R-I-A-R-G-D-W-N-D-D-Y-C (Figure 2). Systematic sequential deletion of single residues from either end resulted in the cyclic decapeptide sequence C-R-I-A-R-G-D-W-N-C as the best match of the native binding sequence. One major difference between bitistatin and other disintegrins is a highly charged domain of additional

Short Chain Bitistatin Analogues

Bioconjugate Chem., Vol. 15, No. 5, 2004 1073 Table 1. Biodistribution of the 99mTc Labeled Peptide, Cyclo-[CH2CO-Arg-Ile-Ala-Arg-Gly-Asp-Trp-AsnCys(S-)]-Gaba-Gly-Asn-Glu-Ile-Leu-Glu-Gln-GlyGlu-Asp-Ser-Asp-Ser-Lys-OH in Normal Mice % injected dose/ga Organ

Figure 5. Inhibition of platelet aggregation by the peptides.

a

amino acids at the N-terminal region. In the model of native bitistatin, this N-terminal sequence appears to be parallel to the binding loop. We hypothesized that this sequence could be interacting with the receptor in some fashion to confer the desirable imaging properties to the molecule. Molecular modeling was performed of components of the N-terminal region fused with the minimal binding site. To confer greater stability to the anticipated peptide and to allow compatibility to a wide variety of bifunctional chelating agents for radiometalation, we replaced the disulfide bridge with the more stable thioether bridge. Furthermore, we attached a Lys residue at the C-terminal to afford the attachment of bifunctional chelating agents to the epsilon amino group. The final molecule having the best fit to the model of bitistatin was the 24-residue peptide cyclo-[CH2CO-Arg-Ile-AlaArg-Gly-Asp-Trp-Asn-Cys(S-)]-Gaba-Gly-AsnGlu-Ile-Leu-Glu-Gln-Gly-Glu-Asp-Ser-Asp-SerLys-OH (Scheme 1) whose model is shown in Figure 3B in comparison to the model of bitistatin (Figure 3A). Figure 4 shows that after Tc derivatization, the Tc chelate would be located far from the RGD binding site and thus is not expected to interfere with receptor binding. Synthesis of the target peptide was performed on an Advanced ChemTech synthesizer ACT90 using the BOC technology as shown in Scheme 1. The final HPLCpurified product was chemically characterized by amino acid analysis and MALDI Mass Spect. Results of in vitro assays of the ability of the new peptide to inhibit platelet aggregation are shown in Figure 5 comparing the activity of the new peptide with a short straight-chain RGD peptide RGDS and recombinant bitistatin (rbitistatin). The IC50 of the recombinant bitistatin, the new peptide,

15 min

30 min

1h

3h

blood 7.67 ( 0.27 5.50 ( 0.45 4.14 ( 0.63 1.94 ( 0.35 brain 0.35 ( 0.03 0.21 ( 0.04 0.15 ( 0.03 0.06 ( 0.01 heart 2.62 ( 0.14 1.96 ( 0.24 1.77 ( 0.13 1.21 ( 0.20 lung 4.17 (0.60 3.06 ( 1.00 2.99 ( 0.56 1.80 ( 1.00 kidneys 73.40 ( 3.95 88.36 ( 7.71 98.26 ( 4.51 112.31 ( 12.10 spleen 2.04 ( 0.37 1.75 ( 0.37 1.92 ( 0.33 1.45 ( 0.18 pancreas 2.38 ( 0.19 1.73 ( 0.18 1.48 ( 0.44 1.06 ( 0.40 stomach 7.81 ( 2.59 10.16 ( 4.10 7.34 ( 1.37 4.12 ( 1.74 intestines 1.63 ( 0.09 1.58 ( 0.17 1.73 ( 0.19 1.77 ( 0.08 liver 4.73 ( 0.52 4.03 ( 0.65 4.47 ( 0.78 3.84 ( 0.45 muscle 1.73 ( 0.21 1.06 ( 0.18 0.71 ( 0.05 0.56 ( 0.24

Average ( 1 SD, n ) 4.

and RGDS were 114, 3694, and 100 000 nM, respectively. Thus, while the biological activity of the new peptide was 27-fold better than the RGDS peptide, it was 32-fold less active than recombinant bitistatin. To label the peptide with 99mTc, coupling with the hydrazinonicotinamidyl bifunctional chelating agent was performed (Scheme 2). After removal of excess uncoupled reagent by HPLC, the modified peptide was labeled in high yield with 99mTc using tricine as the coligand. As shown in the Scatchard analysis of in vitro binding studies depicted in Figure 6, the Kd of the peptide was 185 nM with a Bmax of 117 000 sites/cell compared to a Kd of 8 nM and Bmax of 80 000 sites/cell for 125Irbitistatin. The Kd for 99mTc-labeled bitistatin from a previous study was 32 nM (4). Thus, the 99mTc-labeled peptide has moderate affinity for the receptor. Biodistribution studies in normal CD-1 mice showed that the peptide cleared rapidly from the blood and all normal tissues except the kidneys where a substantial amount of the activity accumulates over an extended period of time (Table 1). At 3 h postinjection, there was still significant soft tissue uptake. The highest soft tissue uptake was in the kidneys, which tends to accumulate 99m Tc-labeled small peptides. Varying levels of kidney activity have been associated with 99mTc-hydrazinonicotinyl-tricine conjugates. The levels of activity exhibited here are higher than usual but similar to those obtained in the kidneys for a 99mTc-labeled polypeptide, EPIHNE-2 (11), using a similar procedure. The high kidney activity levels may be related to the peptide itself in conjunction with the labeling methodology.

Figure 6. Scatchard plots of the binding of the 99mTc-Labeled Peptide (A) compared to the 125I-Labeled Recombinant Bitistatin (B). B/F ) Bound Ligand/Free Ligand, molecs ) molecules.

1074 Bioconjugate Chem., Vol. 15, No. 5, 2004

Baidoo et al.

Figure 7. Images of experimental Deep-Vein Thrombosis in a canine model using the 99mTc-labeled peptide. The images show the hind legs of the dog with the arrows pointing to the clots.

Figure 8. Images of experimental Emboli in the lungs of a dog using the 99mTc-labeled peptide. The images represent lateral positions of the head, neck, and upper torso. The arrows point to the position of the clots in the lungs. The heart is also in the field of view. Table 2. Region-of-Interest Analysis of 99mTc-Short-Chain Bitistatin Analogue Images time postinjection thrombus: muscle thrombus: contralateral site

time postinjection embolus: lung embolus: muscle embolus: heart time postinjection embolus: lung embolus: muscle embolus: heart

legs 45 min 1.56 1.68

1h 1.42 1.47

2h 1.57 1.80

due to the lower affinity of the labeled peptide compared to labeled bitistatin. Currently, 99mTc-Apcitide, a small peptide containing a surrogate for RGD constrained in a small loop, is an approved radiopharmaceutical for imaging deep-vein thrombosis. The thrombus uptake of 125I-bitistatin or 99mTc-bitistatin in the standardized canine model has been consistently much higher than that of 99mTcApcitide. In the same animal model, 99mTc-Apcitide had image-derived thrombus:contralateral vessel and thrombus:muscle ratios of 2.0 and 1.9 at 4 h postinjection (12). The short-chain peptide analogue of bitistatin described here did not exhibit an advantage over 99mTc-Apcitide with respect to imaging deep-vein thrombosis. However, the bitistatin analogue was able to produce images of pulmonary embolism. No images of pulmonary embolism have previously been reported for 99mTc-Apcitide. The RGD-containing loop in bitistatin is very large (more than 13 amino acids). Because of its size, it is inherently flexible and probably depends on the rest of the bitistatin molecule to maintain its orientation. In contrast, the small tetrapeptide loop structure in Apcitide is fairly rigid. Although molecular modeling in this report predicted that a loop structure of 10 amino acids containing RGD would be the best match to the binding loop of bitistatin, it is possible that in the physiologic environment this loop is very flexible and does not have adequately high affinity in all configurations. Previous studies in this animal model with a 99mTc-labeled linear RGD-containing peptide (13) (based upon the binding sequence of an antiplatelet monoclonal antibody) resulted in poorer thrombus binding than the short-chain bitistatin analogue reported here. In this study, we have attempted to simulate the environment created by the N-terminal domain of bitistatin to create the short chain peptide. Other regions of bitistatin may be required to generate a bitistatin analogue that provides the appropriate environment to constrain the orientation of the binding loop. We are currently investigating other structural features of native bitistatin to incorporate in a shorter peptide that would still be amenable to synthesis. ACKNOWLEDGMENT

chest 2 h 29 min right lateral 1.15 4.08 1.53

3 h 35 min right lateral 1.30 3.91 1.44

2 h 40 min left lateral 1.13 3.05 1.31

3 h 22 min left lateral 1.33 3.04 1.47

To investigate the utility of the 99mTc-labeled peptide for imaging thrombi and emboli, we performed imaging of experimental deep-vein thrombosis and pulmonary emboli in a dog. Figures 7 and 8 show representative images obtained for the deep-vein thrombus and the emboli in lungs of the same dog, respectively. There appeared to be two clots in the lung and one main clot in the leg. The clots in the lung and the leg were discernible within the 4-h imaging period. The clot in the leg could be seen earlier within 45 min postinjection. The targetto-nontarget ratios obtained from the image are presented in Table 2. Comparison of results obtained with bitistatin labeled with 99mTc in the same fashion (4) indicated that the peptide under study was less effective than the native peptide for imaging thrombi and emboli. The lower contrasts obtained with this peptide may be

The authors would like to thank Dr. Alan Maurer for creating the model of experimental thrombi and emboli. This work was supported in part by NIH grant HL-54578 (L.C.K.). LITERATURE CITED (1) Shebuski, R. J., Ramjit, D. R., Bencen, G. H., and Polokiff, M. A. (1989) Characterization and platelet inhibitory activity of bitistatin, a potent arginine-glycine-aspartic acid containing peptide from the venom of the viper Bitis arietans. J. Biol. Chem. 264, 21550-21556. (2) McLane, M. A., Kowalska, M. A., Silver, L., Shattil, S. J., and Niewiarowski, S. (1994) Interaction of disintegrins with the alpha IIb beta 3 receptor on resting and activated human platelets. Biochem. J. 301, 429-436. (3) Knight, L. C., Maurer, A. H., and Romano, J. E. (1996) Comparison of Iodine-123-disintegrins for imaging thrombi and emboli in a canine model. J. Nucl. Med. 37, 476-482. (4) Knight, L. C., Baidoo, K. E., Romano, J. E. Gabriel, J. L., and Maurer, A. H. (2000) Imaging pulmonary eemboli and deep venous thrombi with 99mTc-Bitistatin, a platelet-binding polypeptide from viper venom, J. Nucl. Med. 41, 1056-1064. (5) Knight, L. C. (2001) Radiolabeled peptide ligands for imaging thrombi and emboli. Nucl. Med. Biol. 28, 515-526. (6) Abrams, M. J., Juweid, M., tenKate, C. I., Schwartz, A. D., Hauser, M. M., Gaul, F. E., Fucello, A. J., Rubin, R. H., Strauss, H. W., and Fischman, A. J. (1990) Technetium-99m-

Short Chain Bitistatin Analogues human polyclonal IgG radiolabeled via the hydrazino nicotinamide derivative for imaging focal sites of infection in rats. J. Nucl. Med. 31, 2022-2028. (7) Rappe, A. K. and Goddard, W. A. (1994) “Biograf Reference Manual”, Chapter 3, Molecular Simulations, Inc; pp 25. (8) Calvete, J. J., Schrader, M., Raida, M., McLane, M. A., Romero, A., and Niewiarowski, S. (1997) The disulphide bond pattern of bitistatin, a disintegrin isolated from the venom of the viper Bitis arietans. FEBS Lett. 416, 197-202. (9) Burt, D. R. (1986) Receptor binding methodology and analysis. In Receptor Binding in Drug Research. (R. A. O’Brien, Ed.) pp 4-29, Marcel Dekker, New York. (10) Diehl, K-H., Hull, R., Morton, D., Pfister, R., Rabemampianina, Y., Smith, D., Vidal, J-M., and Van De Vorstenbosch, C. (2001) A good practice guide to the administration of substances and removal of blood, including routes and volumes J. Appl. Toxicol. 21, 15-23.

Bioconjugate Chem., Vol. 15, No. 5, 2004 1075 (11) Rusckowski, M., Qu, T., Gupta, S., Ley, A., and Hnatowich, D. J. (2001) A comparison of 99mTc Labeled to a peptide by 4 methods. J. Nucl. Med. 42, 1870-1877. (12) Lister-James, J., Knight, L. C., Maurer, A. H., Bush, L. R., Moyer, B. T., and Dean, R. T. (1996) Thrombus imaging with a technetium-99m-labeled, activated platelet receptorbinding peptide. J. Nucl. Med. 37, 775-781. (13) Knight, L. C., Radcliffe, R., Maurer, A. H., Rodwell, J. D., and Alvarez, V. L. (1994) Thrombus imaging with technetium99m synthetic peptides based upon the binding domain of a monoclonal antibody to activated platelets. J. Nucl. Med. 35, 282-288. (14) Lehninger, A. L., Nelson, D. L., and Cox, M. M. (1992) Amino acids and peptides. Principles of Biochemistry, 2nd edition. pp 111-133. Chapter 5. Worth Publishers, New York.

BC049954V