Development and Characterization of Annexin V Mutants with

[99mTc]Annexin V can be used to image organs undergoing cell death during cancer chemotherapy and organ transplant rejection. To simplify the preparat...
11 downloads 0 Views 149KB Size
Download PDF
918

Bioconjugate Chem. 2000, 11, 918−925

Development and Characterization of Annexin V Mutants with Endogenous Chelation Sites for 99mTc Jonathan F. Tait,*,†,‡ David S. Brown,† Donald F. Gibson,† Francis G. Blankenberg,§ and H. William Strauss§ Departments of Laboratory Medicine, Medicine (Medical Genetics) and Pathology, University of Washington, Seattle, Washington 98195-7110, and Department of Radiology, Stanford University, Stanford, California 94305. Received June 1, 2000; Revised Manuscript Received August 10, 2000

[99mTc]Annexin V can be used to image organs undergoing cell death during cancer chemotherapy and organ transplant rejection. To simplify the preparation and labeling of annexin V for nuclearmedicine studies, we have investigated the addition of peptide sequences that will directly form endogenous chelation sites for 99mTc. Three mutant molecules of annexin V, called annexin V-116, -117, and -118, were constructed with N-terminal extensions of seven amino acids containing either one or two cysteine residues. These molecules were expressed cytoplasmically in Escherichia coli and purified to homogeneity with a final yield of 10 mg of protein/L of culture. Analysis in a competitive binding assay showed that all three proteins retained full binding affinity for erythrocyte membranes with exposed phosphatidylserine. Using SnCl2 as reducing agent and glucoheptonate as exchange agent, all three proteins could be labeled with 99mTc to specific activities of at least 50-100 µCi/µg. The proteins retained membrane binding activity after the radiolabeling procedure, and quantitative analysis indicated a dissociation constant (Kd) of 7 nmol/L for the annexin V-117 mutant. The labeling reaction was rapid, reaching a maximum after 40 min at room temperature. The radiolabeled proteins were stable when incubated with phosphate-buffered saline or serum in vitro. Proteins labeled to a specific activity of 25-100 µCi/µg were injected intravenously in mice at a dose of 100 µg/kg, and biodistribution of radioactivity was determined at 60 min after injection. Uptake of radioactivity was highest in kidney and liver, consistent with previous results obtained with wild-type annexin V. Cyclophosphamide-induced apoptosis in vivo could be imaged with [99mTc]annexin V-117. In conclusion, annexin V can be modified near its N-terminus to incorporate sequences that form specific chelation sites for 99mTc without altering its high affinity for cell membranes. These annexin V derivatives may be useful for in vivo imaging of cell death.

INTRODUCTION

Membrane phospholipids are distributed asymmetrically in most cells (1, 2). Phosphatidylserine (PS)1 is normally retained on the intracellular face of the plasma membrane, but this distribution is altered in some physiological and pathological states. PS exposure is part of the normal platelet procoagulant response, and it also occurs during apoptosis, where it may serve as a signal to promote phagocytosis of apoptotic cells and their fragments (3). We have recently used annexin V to detect PS exposure in vivo. Annexin V labeled with radioiodine and 99mTc was used to detect thrombus formation in experimental animals (4, 5). [99mTc]Annexin V was used to detect cell death in vivo in animal models (6-10). Because of the clinical applicability of this work in monitoring cancer chemotherapy and rejection of transplanted organs, there is interest in developing new * To whom correspondence should be addressed. Phone (206) 598-6131. Fax: (206) 598-6189. E-mail:[email protected]. † Department of Laboratory Medicine. ‡ Department of Medicine (Medical Genetics) and Pathology. § Department of Radiology. 1 Abbreviations: DTPA, diethylenetriaminepentaacetic acid; HYNIC, 6-hydrazinopyridine-3-carboxylate; IC50, inhibitory concentration 50%; ID, injected dose; ITLC, instant thin-layer chromatography; PBS, phosphate-buffered saline; ROI, region of interest; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

annexin V derivatives with improved imaging properties or greater ease of use. Some naturally occurring proteins have been labeled directly with technetium, which presumably chelates with favorably disposed atoms from the peptide backbone and/or amino acid side chains. However, this approach lacks generality, and the results are unpredictable for any particular protein. More recently, specific peptide sequences, modeled after short peptides such as mercaptoacetyltriglycine (11-14), have been introduced into recombinant proteins in order to create endogenous chelation sites for technetium. For example, George et al. (15) attached the sequence GGGGC to the C-terminus of two recombinant single-chain Fv antibody proteins. They showed that these proteins could be labeled with 99mTc to high specific activity. Bogdanov et al. (16) showed specific Tc chelation to a construct of green fluorescent protein with the added C-terminal sequence LGGGGCGGGCGI. In another approach, Waibel et al. (17) showed Tc chelation to a recombinant protein containing five or six histidine residues at the C-terminus. Pietersz et al. (18) prepared a hybrid protein consisting of metallothionein fused to a single-chain Fv molecule, which was labeled with Tc by transchelation from a precursor protein containing Zn. It would be advantageous to label annexin V directly with 99mTc, without requiring attachment of exogenous chelation groups. The N-terminal region of annexin V is

10.1021/bc000059v CCC: $19.00 © 2000 American Chemical Society Published on Web 10/21/2000

Annexin V Mutants with Tc Chelation Sites Table 1. Oligonucleotides Used for Construction of Vectors oligonucleotide ame

vector

oligonucleotide sequence (5′ to 3′)

JT-289 JT-290 JT-295 JT-296 JT-297 JT-298

pJ116 pJ116 pJ117 pJ117 pJ118 pJ118

TATGGCATGTGGCGGTGGCCA TATGGCCACCGCCACATGCCA TATGGCAGGTGGCTGTGGCCA TATGGCCACAGCCACCTGCCA TATGGCATGTGGCTGCGGTCA TATGACCGCAGCCACATGCCA

a suitable site for attachment of additional peptide sequence, since previous work has shown that attachment of small (19) and large (20) sequences does not alter membrane binding affinity, nor does it affect expression or folding of the recombinant protein in Escherichia coli. The three-dimensional structure of annexin V also shows that the N-terminal region is located on the opposite side of the molecule from the membrane binding site (21). In this study, we have prepared three novel derivatives of annexin V that contain N-terminal extensions of seven amino acids that are specifically designed to chelate technetium. One derivative is predicted to form an N2S2 chelation site, and two are predicted to form an N3S site. These derivatives can be produced in high yield by expression in E. coli, and they can be labeled to high specific activity with technetium. MATERIALS AND METHODS

Materials. Recombinant annexin V was produced in E. coli as described (22). Fluorescein-annexin V was prepared as described (23). Sodium glucoheptonate was from Sigma. SnCl2‚2H2O was from Aldrich. Annexin V was derivatized with succinimidyl 6-hydrazinopyridine-3carboxylate (abbreviated HYNIC) as described (6). Erythrocytes with exposed PS (4CPlus Normal Control, part number 7546923) were from Beckman-Coulter (Miami, FL). Construction and Verification of Plasmids. Several plasmids were constructed to express mutant forms of annexin V in E. coli under control of the phage T7 promoter. The parent expression vector was pET12a, available from Novagen Corporation (Madison, WI). Construction of Vector pJ115, Encoding Annexin V-115 with Cys-316fSer Mutation. A 232-bp BstBI-BamHI restriction fragment was first isolated from plasmid pANXVC-S-N6 (19), also known as pJ110, which encodes a modified annexin V with the Cys-316 f Ser mutation. The same restriction fragment was then removed from plasmid pET12a-PAP1 (22), which encodes wild-type annexin V, and the mutated BstBI-BamHI fragment was inserted to form plasmid pJ115. This plasmid was subjected to DNA sequence analysis to verify that the intended mutation had been correctly introduced. Construction of vectors pJ116, pJ117, and pJ118, encoding the mutant molecules annexin V-116, annexin V-117, and annexin V-118. For each vector to be constructed, sense and antisense oligonucleotides (with sequences given in Table 1) were designed to encode the desired peptide sequence and to have NdeI restriction sites at both ends. Equimolar amounts of sense and antisense oligonucleotide were phosphorylated and annealed to form a double-stranded cassette with NdeI-compatible single-stranded ends. The double-stranded oligonucleotide (3.8 ng) was ligated with plasmid pJ115 (0.42 µg) that had been previously digested with NdeI and dephosphorylated. The ligation mixture was transformed into E. coli strain DH5R. Plasmid DNA was prepared from ampicillin-resistant

Bioconjugate Chem., Vol. 11, No. 6, 2000 919

colonies and screened by PCR for the presence of the desired insert. DNA sequence analysis was then performed on positive clones to verify the presence of the desired insertion and the absence of other adventitious mutations. Expression and Purification of Proteins. Plasmids pJ116, pJ117, and pJ118 were each transformed into E. coli strain BL21(DE3) for cytoplasmic expression. Each clone was grown overnight to saturation at 37 °C with shaking in Terrific Broth containing carbenicillin (50 µg/ mL). The cells were separated from culture medium by centrifugation for 10 min at 2560g and washed in icecold buffer (50 mmol/L TrisHCl, 150 mmol/L NaCl pH 8.0). Bacteria were disrupted by sonication in ice-cold 50 mmol/L Tris‚HCl, pH 7.2, 10 mmol/L CaCl2, 1 mmol/L β-mercaptoethanol, and then centrifuged for 20 min at 22530g. The supernatant was discarded and the annexin V bound to bacterial membranes was released by resuspending the pellet in 50 mmol/L Tris‚HCl, pH 7.2, 20 mmol/L EDTA, 1 mmol/L β-mercaptoethanol. Bacterial membranes were removed by centrifugation for 20 min at 22530g and the supernatant containing the annexin V was dialyzed against 20 mmol/L Tris‚HCl, pH 8.0, 1 mmol/L β-mercaptoethanol. The dialysand was applied to a Mono Q HR 16/10 column (Pharmacia, Piscataway, NJ) and eluted with a gradient of 0 to 1 mol/L NaCl in the same buffer. The annexin V mutants all eluted at approximately 0.22 mol/L NaCl. The purified protein was concentrated by ultrafiltration to approximately 5 mg/ mL, dialyzed against 20 mmol/L HEPES, pH 7.4, 100 mmol/L NaCl, and stored in aliquots at -70 °C. The final yield was approximately 10 mg/L of bacterial culture, with a purity of ∼98% as judged by SDS-PAGE. Labeling of Proteins with 99mTc. Unless noted otherwise, mutant proteins were labeled with 99mTc as follows. A 1-mg aliquot of protein was incubated with 1 mmol/L dithiothreitol for 15 min at 37 °C. This material was then purified on a Sephadex G-25 column (0.9 × 2.8 cm) previously equilibrated with deoxygenated 20 mmol/L sodium citrate, pH 5.2, 100 mmol/L NaCl. This material was stored in glass vials at -20 °C in aliquots of 100 µg in 100 µL of the same buffer. For Tc labeling, glucoheptonate was used as the exchange agent. A solution containing 20 mmol/L sodium glucoheptonate and 10 mmol/L sodium HEPES was adjusted to pH 6.6 and deoxygenated with argon. SnCl2‚2H2O was added to 128 µg/mL, and 1-mL aliquots were dispensed into borosilicate glass vials, lyophilized, capped with Teflon-lined screw caps under argon, and stored at -20 °C. For radiolabeling, 100 µL of 99mTcO4 in 0.9% NaCl was added to 200 µL of tin/glucoheptonate (from a lyophilized aliquot freshly reconstituted with 1 mL of deoxygenated H2O). To this was added 100 µg (100 µL) of reduced protein. The reaction was incubated for 60 min at room temperature. For labeling of HYNIC-annexin V, tricine was used as the exchange agent. The tin/tricine reduction/exchange reagent was prepared as previously described (24); the final composition of the reagent was 200 mmol/L tricine, pH 7.1, 128 µg/mL SnCl2.2H2O. This reagent was dispensed in 1-mL aliquots into borosilicate glass vials, lyophilized, capped with Teflon-lined screw caps under an argon atmosphere, and stored at -20 °C. For radiolabeling, a 100-µg lyophilized aliquot of HYNIC-annexin V was reconstituted with 100 µL of H2O, and 100 µL of 99mTcO in 0.9% NaCl was added. Freshly reconstituted 4 tin/tricine reagent (200 µL) was then added and the reaction mixture was incubated for 15 min at room temperature.

920 Bioconjugate Chem., Vol. 11, No. 6, 2000

Radiolabeled proteins prepared by either method were purified by applying the reaction mixture (400 µL) to a Sephadex G-25 column (0.9 × 2.8 cm) previously equilibrated with PBS and eluting the column with 1000 µL of PBS. Percent protein-bound radioactivity was determined by ITLC on oven-dried silica gel plates with PBS as solvent. For studies of radiolabel stability in serum, percent protein-bound radioactivity was determined by centrifugal ultrafiltration (3 min at 1860g) through a membrane with a 10 000-Da cutoff (Centricon-10 filter, Amicon, Danvers, MA). Radiolabel stability was also assessed by gel filtration on a Superose-12 column (1 × 30 cm) connected to an FPLC system. The column was equilibrated with PBS and eluted at a flow rate of 0.5 mL/min at room temperature. Cell Binding Assays. The affinity of nonradioactive proteins for cell membranes was determined by their ability to compete with fluorescein-annexin V for binding to erythrocytes with exposed phosphatidylserine (20). Mutant proteins were added at various concentrations to 1 mL of a solution containing 5 nmol/L fluoresceinannexin V and 8.3 × 106 erythrocytes in a buffer consisting of 10 mmol/L HEPES-Na, pH 7.4, 136 mmol/L NaCl, 2.7 mmol/L KCl, 5 mmol/L glucose, 1 mg/mL BSA (buffer HNKGB) plus 2.5 mmol/L CaCl2. The sample was incubated for 15 min at room temperature. The cells were centrifuged for 3 min at 8300g, the supernatant removed, and the fluorescein-annexin V bound to the pelleted cells was released by resuspension in 950 µL of HNKGB buffer containing 5 mmol/L EDTA. The sample was centrifuged again, the supernatant removed, and the concentration of fluorescein-annexin V in the supernatant determined by fluorometry. The results of this assay are summarized as the concentration of competitor protein that inhibits 50% of the binding of fluorescein-annexin V, abbreviated IC50. The ability of radiolabeled proteins to bind to cell membranes was determined with a similar procedure. Radiolabeled protein at 10 nmol/L final concentration was added to duplicate tubes containing a final volume of 1 mL of buffer HNKGB plus 2.5 mmol/L CaCl2. One tube then received 4.2 × 108 erythrocytes. After a 15-min incubation at room temperature, both tubes were centrifuged for 3 min at 8320g. Radioactivity was measured in 800 µL of the supernatants. The percentage of radioactivity bound to the cells was calculated from 100 × [1 - (supernatant counts in the presence of cells)/(supernatant counts in absence of cells)]. Quantitative titrations were performed to determine the dissociation constant for protein-membrane interaction as previously described (25). Radiolabeled protein (3-150 nmol/L) was added to 1 × 107 erythrocytes in buffer HNKGB plus 2.5 mmol/L CaCl2. After a 15-min incubation at room temperature, bound and free ligand were separated by centrifugation through a silicone-oil barrier, and the radioactivity bound to the pelleted cells was determined by scintillation counting; this was converted to number of molecules bound per cell by dividing by the known specific activity. The concentration of free ligand was determined by measuring the radioactivity remaining in the supernatant, corrected for the fraction of nonbindable tracer. The dissociation constant was determined by using nonlinear least-squares analysis to fit the data to a model of noncooperative binding to homogeneous sites. Biodistribution and Imaging Studies. Each annexin V mutant protein (100 µg) was labeled with 99mTc (10 mCi) using the glucoheptonate procedure as described above and purified on a Sephadex G-25 column (1.5 × 5 cm)

Tait et al.

Figure 1. Primary structures of annexin V mutants. Each mutant protein contains the indicated N-terminal extension of seven amino acids, followed by amino acids 2-320 of human annexin V (29). The naturally occurring Cys residue at position 316 has also been mutated to Ser in all three proteins. The initiator methionine residue is not shown because it is removed posttranslationally (30). For convenience, the mutant molecules are given arbitrary numerical designations.

eluted with PBS. The eluate appearing between 2 and 4 mL was pooled. For biodistribution studies, 50 µL ()1.5 µg, 40-150 µCi) of this eluate was injected via the tail vein into 8-10 week old Balb/c mice (Veterinary Services, Department of Comparative Medicine, Stanford University) weighing between 20 and 25 g. Organs and other targets of interest were harvested from groups of 5-7 mice at 60 min after injection; the sample tissues were weighed and then underwent scintillation counting. Samples were counted in a Packard Cobra II gamma counter (Packard, Meriden, CT) with energy windows set between 120 and 170 keV. Scintillation well data were expressed as the percentage of injected dose (%ID) and %ID per gram of tissue (%ID/g) for each sample. Scintillation counting data were adjusted for tail vein infiltration, decay, and background as described previously (6). For imaging studies, young male (150-200 g, 8-10 week old) Sprague-Dawley rats (Simonsens, Gilroy, CA) were housed and treated in a humane manner in accordance to institutional animal subjects guidelines. Animals were treated with 150 mg/kg of cyclophosphamide (Sigma) reconstituted in 1 mL of PBS and injected intraperitoneally. All animals were then imaged at either 0 or 24 h after cyclophosphamide treatment, followed by sacrifice and well counting of selected tissues. Animals were injected via tail vein with 0.5-1 mCi (37 MBq, 2040 µg protein/kg) of radiolabeled annexin V 1 h prior to radionuclide imaging and sacrifice. At the time of imaging, rats were sedated with a mixture of 80 mg/kg acepromazine and 40 mg/kg ketamine injected intramuscularly. A Technicare 420 mobile scintillation camera (Technicare; Solon, OH) equipped with a low energy, high resolution parallel hole collimator was used to record the [99mTc]annexin V distribution. Data were recorded using a 20% window centered on the 140 keV photopeak of 99mTc into a 256 × 256 matrix of a dedicated computer system for digital display and analysis (ICON; Siemens, Hoffman Estates, IL). All images were recorded for a preset time of 10 min. Region of interest (ROI) image analysis was performed to determine relative counts in one femur (corrected for background in the adjacent soft tissue of the thigh) in comparison to that of the whole body. The result was expressed as the percent of the whole body activity. P-Values were calculated using the student t-test with a one-tail test for significance assuming equal variances of test and control groups. P-values of 0.05 or less were considered significant. Average values are presented along with their associated standard deviations. RESULTS

Protein Production and Characterization. Three mutant proteins were designed to have N-terminal sequences potentially suitable for chelation of Tc (Figure 1). Annexin V-116 and annexin V-117 were predicted to form an N3S chelate, while annexin V-118 was predicted

Annexin V Mutants with Tc Chelation Sites

Bioconjugate Chem., Vol. 11, No. 6, 2000 921

Figure 2. SDS-PAGE analysis of mutant proteins. Recombinant wild-type annexin V and the three mutant annexin V molecules were reduced with β-mercaptoethanol, analyzed on a 10 to 15% gradient gel on a Pharmacia PHAST system, and stained with Coomassie Brilliant Blue. Table 2. Membrane Binding Activity of Mutant Proteinsa protein

IC50 (nmol/L)

annexin V annexin V-116 annexin V-117 annexin V-118 HYNIC-annexin V

6.8 ( 0.7 9.3 ( 0.4 10.3 ( 2.5 10.0 ( 2.8 10.1 ( 2.0

Figure 3. Quantitative cell binding curve. Red blood cells with exposed PS were incubated with varying amounts of [99mTc]annexin V-117 (squares) or [99mTc]HYNIC-annexin V (circles); bound and free ligand were then separated as described in the Materials and Methods and the amount of cell-bound and unbound radioactivity was determined by scintillation counting. The data were fit to a model of noncooperative binding to homogeneous sites. The calculated binding parameters for annexin V-117 were Kd ) 7.3 nmol/L, Bmax ) 272 000 sites/cell; corresponding values for HYNIC-annexin V were 6.9 nmol/L and 330 000 sites/cell.

a IC 50 values were determined for unlabeled proteins by competition assay as described in the Materials and Methods. Results are given as mean ( SD with n ) 9 for annexin V and HYNICannexin V and n ) 2 for mutants 116, 117, and 118.

Table 3. Radiochemical Yield, Purity, and Bioactivity of Labeled Proteins

protein annexin V annexin V-116 annexin V-117 annexin V-118 HYNIC-annexin V

% radioactivity % % bound to radiochemical radiochemical erythrocytesc puritya,b yielda 3.9 ( 1.6 89.5 ( 5.7 88.8 ( 2.0 89.9 ( 1.9 97.0%

not applicable 94.4 ( 2.0 92.2 ( 1.2 94.2 ( 0.6 99.0%

not applicable 77.5 ( 3.5 78.1 ( 3.8 78.7 ( 3.6 83.9 ( 2.5

a Results are given as mean ( SD with n ) 2 for annexin V, n ) 4 for mutants 116, 117, and 118, and n ) 1 for HYNIC-annexin V. 2 % Radiochemical purity was determined by ITLC after purification by gel filtration. 3 Results are given as mean ( SD with n ) 4 for mutants 116, 117, and 118, and n ) 7 for HYNICannexin V.

to form an N2S2 chelate. The single cysteine residue present at position 316 of native annexin V was also mutated to serine to eliminate the possibility that this residue might form a weak binding site for technetium. The structures of all constructs were verified by DNA sequencing. All three proteins could be expressed cytoplasmically in E. coli and purified in high yield (10 mg pure protein per liter of bacterial culture) under conditions similar to those used to produce wild-type annexin V (22, 26). SDS-PAGE analysis (Figure 2) showed that all three mutants could be produced in highly pure form. As expected, all three mutant proteins (326 amino acids in length) also migrated with the same apparent molecular weight as wild-type annexin V (319 amino acids). The bioactivity of the mutant proteins was verified by measuring their ability to compete with wild-type annexin V for binding to erythrocyte membranes with exposed PS (Table 2). All three proteins displaced fluo-

Figure 4. Time course of radiolabeling. Protein (50 µg) was reacted with 99mTcO4 (100 µCi), and ITLC was performed on aliquots taken from the reaction mixtures at the indicated times. (A) Labeling at room temperature (22 °C); (B) Labeling at 37 °C. (4) Annexin V-116, (0) annexin V-117, (]) annexin V-118.

rescein-annexin V with nearly the same relative affinity as wild-type annexin V, indicating that attachment of the N-terminal heptapeptide did not interfere with membrane binding either directly or indirectly. The mutant proteins also had the same IC50 values as HYNICannexin V, which has been previously used to detect PS exposure in vivo. Radiolabeling with 99mTc. The three mutant proteins were tested for their ability to chelate technetium. As

922 Bioconjugate Chem., Vol. 11, No. 6, 2000

Tait et al.

Figure 5. Radiolabeling as a function of technetium/protein ratio. Protein (5-20 µg) was added to technetium (100-500 µCi) to achieve the indicated ratio of technetium to protein. After 60 min at room temperature, ITLC was performed on the reaction mixture to determine percent incorporation of radioactivity into protein. The specific activity was then calculated from the input ratio of technetium to protein multiplied by the percent incorporation of radioactivity into protein. Symbols: as in Figure 4.

Figure 7. Gel filtration analysis of [99mTc]annexin V-117. (A) [99mTc]Annexin V-117 (50 µg in 600 µL of PBS) was incubated for 2 h at room temperature and then analyzed on a Superose12 column as described in the Materials and Methods. Radioactivity was determined in 1-mL column fractions. The arrows indicate the elution positions of globular protein standards with the indicated molecular masses. B. [99mTc]Annexin V-117 (50 µg in 300 µL of PBS) was mixed with 300 µL of human serum, incubated for 2 h at room temperature, and then analyzed on a Superose-12 column.

Figure 6. Stability of radiolabeled proteins in vitro. Protein was radiolabeled to a specific activity of 1-2 µCi/µg and purified by gel filtration chromatography as described in the Materials and Methods. Radiolabeled protein was then incubated at a concentration of 250 µg/mL at room temperature in either PBS (diamonds), PBS + 5 mmol/L DTPA (squares), or a 1:1 mixture of PBS and human serum (circles). Aliquots were taken at the indicated times and percent protein-bound 99mTc was determined.

shown in Table 3, all three proteins incorporated most of the added pertechnetate, as measured by ITLC of the reaction mixture. The percent incorporation was comparable to the value obtained for HYNIC-annexin V under the same conditions. In contrast, there was almost no labeling of wild-type annexin V, indicating that the observed radiolabeling was specifically due to the added N-terminal sequences in the mutant molecules. The radiolabel remained attached to protein after removal of labeling reagents by gel filtration chromatography, as indicated by the high radiochemical purity of gel-filtered proteins. To verify that the radiolabeled proteins retained membrane binding activity, radiolabeled proteins were incubated with erythrocytes in the presence of calcium,

and the percentage of radioactivity bound to cells was measured (Table 3). All three proteins showed comparable levels of binding, and this level was comparable to the binding of [99mTc]HYNIC-annexin V tested under the same conditions. Quantitative titrations were performed to determine the affinity of the radiolabeled protein for cell membranes with exposed PS. As shown in Figure 3, [99mTc]annexin V-117 showed high-affinity saturable binding to erythrocytes with exposed PS; the binding curve was essentially identical to that obtained with [99mTc]HYNICannexin V. The measured dissociation constant was 7 nmol/L, consistent with previous results obtained for binding of [125I]annexin V to platelets and erythrocytes (25, 27). Optimization of Labeling Conditions. The time course of labeling was rapid for all three mutants. As shown in Figure 4A, incorporation of 99mTc occurred rapidly at room temperature, reaching a maximum at 40 min for annexin V-117. The reaction was slightly faster at 37 °C, reaching a maximum at 30 min for annexin V-117 (Figure 4B). For convenience, a 60-min reaction at room temperature was chosen for most subsequent experiments. To determine if proteins could be labeled to high specific activity, the technetium/protein was varied. As shown in Figure 5, the incorporation of 99mTc into protein

Annexin V Mutants with Tc Chelation Sites

Bioconjugate Chem., Vol. 11, No. 6, 2000 923

Figure 8. Imaging of apoptosis in bone marrow and spleen. Rats received either no treatment (control) or 150 mg/kg cyclophosphamide (+ cyclophosphamide). After 24 h, they received 0.5 mCi of [99mTc]annexin V-117 injected via the tail vein. One hour later, images were recorded for 10 min from animals in the supine position.

increased with increasing amounts of added 99mTcO4; specific activities of ∼100 µCi/µg were readily obtainable for annexin V-117 and annexin V-118. Stability of Radiolabeled Proteins in Vitro. The stability of the radiolabeled protein in vitro was determined after challenge with phosphate-buffered saline, DTPA (5 mmol/ L), and serum (Figure 6). ITLC analysis showed that all three mutant proteins retained the radiolabel over a period of several hours, indicating that the Tc-protein chelate was of high affinity. These results were confirmed by gel filtration chromatography of [99mTc]annexin V-117 (Figure 7A). After incubation of [99mTc]annexin V-117 with PBS for 2 h, almost all of the radioactivity eluted in the same position as monomeric annexin V-117 (Mr 36 000); there was no evidence for multimeric protein aggregates. Similarly, gel filtration chromatography of [99mTc]annexin V-117 after a 2-h incubation with human serum showed that the radioactivity still eluted in the same position as monomeric annexin V (Figure 7B). Thus, there was no evidence for transchelation of 99mTc to other serum proteins over a time period consistent with the normal blood clearance time of annexin V (see Discussion). Biodistribution of Radiolabeled Proteins in Mice. Labeled proteins were injected intravenously in mice and biodistribution was determined 60 min later (Table 4). Uptake was highest in kidney and liver. The overall pattern of uptake was similar to results obtained with [99mTc]HYNIC-annexin V, indicating that the N-terminal additions to the three mutant molecules had not drastically altered their biodistribution. Annexin V-117 Imaging of Cyclophosphamide-Treated Rats. To verify that annexin V-117 would be suitable to detect apoptosis in vivo, we used a model of cyclophosphamide-induced apoptosis in rats. Imaging of cyclophosphamide-treated rats clearly showed increased uptake of annexin V-117 in the femurs and spleen compared to control animals (Figure 8), consistent with our earlier studies with HYNIC-annexin V (6). Cyclophosphamidetreated rats (N ) 5) demonstrated a 56% increase in femoral uptake determined by ROI analyses 24 h after treatment compared with controls (N ) 4) (0.325 ( 0.084% versus 0.209 ( 0.056% whole body uptake, respectively; p ) 0.0262). Well counting of excised femurs revealed a borderline (p ) 0.065) increase of annexin

Table 4. Biodistribution of Mutant Molecules in Micea % injected dose per organ organ

annexin V-116

annexin V-117

annexin V-118

HYNICannexin V

carcass kidney liver small intestine stomach large intestine tail lung blood spleen bone heart thymus skeletal muscle brain

31.94 17.96 11.24 3.50 3.35 1.79 0.79 0.83 0.24 0.56 0.19 0.11 0.05 0.03 0.02

23.05 6.17 5.97 4.18 1.69 1.15 0.72 0.40 0.36 0.28 0.14 0.07 0.04 0.03 0.02

21.73 5.99 10.36 6.20 4.20 3.22 0.48 0.60 0.49 0.42 0.15 0.07 0.03 0.03 0.02

32.64 39.16 16.59 3.31 0.88 0.88 0.71 0.61 0.39 0.88 0.18 0.14 0.05 0.03 0.01

a

Results are mean values for n ) 5 to 7 animals.

V-117 uptake in treated as compared with control animals (0.239 ( 0.079% ID/g versus 0.167 ( 0.028% ID/g, respectively). Well counting of excised spleens showed a significant 59% increase (p ) 0.0303) in splenic uptake in treated animals compared with control (3.27 ( 1.04% ID/g versus 2.06 ( 0.332% ID/g, respectively). DISCUSSION

We have successfully produced mutant forms of annexin V that have endogenous chelation sites for 99mTc. These mutant molecules retain bioactivity, as measured by their ability to bind with high affinity to cell membranes with exposed PS. Membrane-binding activity is preserved both before and after labeling with 99mTc. Labeling of these molecules with 99mTc is very simple and rapid, and high specific activities can be achieved. Presence of endogenous chelation sites on these proteins greatly simplifies production of material suitable for 99mTc labeling: the recombinant proteins can be used directly, without the need for complex and time-consuming procedures for attachment of organic Tc chelation groups. The annexin V mutants are also chemically homogeneous, in contrast to the heterogeneous labeling that usually occurs with amine-directed bifunctional organic-labeling reagents.

924 Bioconjugate Chem., Vol. 11, No. 6, 2000

The mutant annexins show good radiolabel stability in vitro. They are stable to challenge by PBS and serum in vitro for at least several hours, and there is no evidence for transchelation of 99mTc to other serum proteins during in vitro incubation (Figures 6 and 7). It should be noted that annexin V is very rapidly cleared from the blood; less than 5% of initial radioactivity is present 2.5 h after intravenous injection (4, 5). Annexin V also does not bind to normal erythrocytes, leukocytes, or platelets in whole blood (28). These results imply that radiolabeled annexins will be confined to the plasma fraction of blood in vivo, and they will be stable in this compartment until they have been cleared into other body compartments. The mutant annexins also showed good in vivo radiostability, as judged by the relatively low excretion of 99m Tc in the stomach. Annexin V-117 showed the greatest radiostability in vivo, as judged by the lowest level of 99mTc present in the stomach (Table 4). This was somewhat surprising, since the N2S2 structure in annexin V-118 might be expected to chelate technetium more tightly than the N3S structure present in annexin V-117. Given its overall combination of properties, annexin V-117 seems the most promising derivative for further development as an in vivo imaging agent. The biodistribution of annexin V-117 may have several advantages for clinical imaging compared to HYNICannexin V. In normal animals, there is only 16% as much uptake in the kidney and 36% as much uptake in the liver, which reduces radiation dose to those organs and may allow the imaging of renal transplant rejection. Bone marrow uptake is also lower in normal control animals, without compromising the relative increase in uptake in the marrow seen in response to cyclophosphamideinduced apoptosis. There is also less uptake in the carcass (probably due to increased excretion in the urine), which will improve target-to-background ratio for most imaging studies. The only comparative disadvantage of annexin V-117 is moderately increased excretion in the stomach and intestines compared to HYNIC-annexin V. The peptide sequences described in this study (Figure 1) also provide new alternatives for introduction of chelation sites into other proteins. It appears that the sequences described here are comparable to the sequences previously used to label other recombinant proteins in the degree of 99mTc incorporation and the stability of the label (15, 16). Although we used an N-terminal location for these peptides because that is optimal for annexin V, they would also be likely to work in either an internal or a C-terminal location. ACKNOWLEDGMENT

We thank Dr. Allan Green for helpful advice and support. This work was partially supported by a grant from Theseus Medical Imaging, Inc., and by NIH Grant HL-61717. LITERATURE CITED (1) Zwaal, R. F., and Schroit, A. J. (1997) Pathophysiologic implications of membrane phospholipid asymmetry in blood cells. Blood 89, 1121-32. (2) Bevers, E. M., Comfurius, P., Dekkers, D. W., and Zwaal, R. F. (1999) Lipid translocation across the plasma membrane of mammalian cells. Biochim. Biophys. Acta 1439, 317-30. (3) Fadok, V. A., Voelker, D. R., Campbell, P. A., Cohen, J. J., Bratton, D. L., and Henson, P. M. (1992) Exposure of phosphatidylserine on the surface of apoptotic lymphocytes triggers specific recognition and removal by macrophages. J. Immunol. 148, 2207-16.

Tait et al. (4) Tait, J. F., Cerqueira, M. D., Dewhurst, T. A., Fujikawa, K., Ritchie, J. L., and Stratton, J. R. (1994) Evaluation of annexin V as a platelet-directed thrombus targeting agent. Thromb. Res. 75, 491-501. (5) Stratton, J. R., Dewhurst, T. A., Kasina, S., Reno, J. M., Cerqueira, M. D., Baskin, D. G., and Tait, J. F. (1995) Selective uptake of radiolabeled annexin V on acute porcine left atrial thrombi. Circulation 92, 3113-21. (6) Blankenberg, F. G., Katsikis, P. D., Tait, J. F., Davis, R. E., Naumovski, L., Ohtsuki, K., Kopiwoda, S., Abrams, M. J., Darkes, M., Robbins, R. C., Maecker, H. T., and Strauss, H. W. (1998) In vivo detection and imaging of phosphatidylserine expression during programmed cell death. Proc. Natl. Acad. Sci. U. S. A. 95, 6349-54. (7) Blankenberg, F. G., Katsikis, P. D., Tait, J. F., Davis, R. E., Naumovski, L., Ohtsuki, K., Kopiwoda, S., Abrams, M. J., and Strauss, H. W. (1999) Imaging of apoptosis (programmed cell death) with 99mTc annexin V. J. Nucl. Med. 40, 184-91. (8) Vriens, P. W., Blankenberg, F. G., Stoot, J. H., Ohtsuki, K., Berry, G. J., Tait, J. F., Strauss, H. W., and Robbins, R. C. (1998) The use of technetium Tc 99m annexin V for in vivo imaging of apoptosis during cardiac allograft rejection. J. Thorac. Cardiovasc. Surg. 116, 844-53. (9) Blankenberg, F. G., Robbins, R. C., Stoot, J. H., Vriens, P. W., Berry, G. J., Tait, J. F., and Strauss, H. W. (2000) Radionuclide imaging of acute lung transplant rejection with annexin V. Chest 117, 834-40. (10) Ogura, Y., Krams, S. M., Martinez, O. M., Kopiwoda, S., Higgins, J. P., Esquivel, C. O., Strauss, H. W., Tait, J. F., and Blankenberg, F. G. (2000) Radiolabeled annexin V imaging: diagnosis of allograft rejection in an experimental rodent model of liver transplantation. Radiology 214, 795800. (11) Fritzberg, A. R., Kasina, S., Eshima, D., and Johnson, D. L. (1986) Synthesis and biological evaluation of technetium99m MAG3 as a hippuran replacement. J. Nucl. Med. 27, 111-6. (12) Fritzberg, A. R., Abrams, P. G., Beaumier, P. L., Kasina, S., Morgan, A. C., Rao, T. N., Reno, J. M., Sanderson, J. A., Srinivasan, A., Wilbur, D. S., and et al. (1988) Specific and stable labeling of antibodies with technetium-99m with a diamide dithiolate chelating agent. Proc. Natl. Acad. Sci. U.S.A. 85, 4025-9. (13) Giblin, M. F., Jurisson, S. S., and Quinn, T. P. (1997) Synthesis and characterization of rhenium-complexed alphamelanotropin analogues. Bioconjug. Chem. 8, 347-53. (14) Stalteri, M. A., Bansal, S., Hider, R., and Mather, S. J. (1999) Comparison of the stability of technetium-labeled peptides to challenge with cysteine. Bioconjug. Chem. 10, 130-6. (15) George, A. J., Jamar, F., Tai, M. S., Heelan, B. T., Adams, G. P., McCartney, J. E., Houston, L. L., Weiner, L. M., Oppermann, H., and Peters, A. M. (1995) Radiometal labeling of recombinant proteins by a genetically engineered minimal chelation site: technetium-99m coordination by single-chain Fv antibody fusion proteins through a C-terminal cysteinyl peptide [published erratum appears in Proc. Natl. Acad. Sci U.S.A. (1996) 93 (18), 9991]. Proc. Natl. Acad. Sci. U.S.A. 92, 8358-62. (16) Bogdanov, A., Jr., Simonova, M., and Weissleder, R. (1998) Design of metal-binding green fluorescent protein variants. Biochim. Biophys. Acta 1397, 56-64. (17) Waibel, R., Alberto, R., Willuda, J., Finnern, R., Schibli, R., Stichelberger, A., Egli, A., Abram, U., Mach, J. P., Pluckthun, A., and Schubiger, P. A. (1999) Stable one-step technetium-99m labeling of his-tagged recombinant proteins with a novel Tc(I)-carbonyl complex. Nat. Biotechnol. 17, 897901. (18) Pietersz, G. A., Patrick, M. R., and Chester, K. A. (1998) Preclinical characterization and in vivo imaging studies of an engineered recombinant technetium-99m-labeled metallothionein-containing anti-carcinoembryonic antigen singlechain antibody. J. Nucl. Med. 39, 47-56. (19) Tanaka, K., Einaga, K., Tsuchiyama, H., Tait, J. F., and Fujikawa, K. (1996) Preparation and characterization of a

Annexin V Mutants with Tc Chelation Sites disulfide-linked bioconjugate of annexin V with the B-chain of urokinase: an improved fibrinolytic agent targeted to phospholipid-containing thrombi. Biochemistry 35, 922-9. (20) Tait, J. F., Engelhardt, S., Smith, C., and Fujikawa, K. (1995) Prourokinase-annexin V chimeras. Construction, expression, and characterization of recombinant proteins. J. Biol. Chem. 270, 21594-9. (21) Huber, R., Berendes, R., Burger, A., Schneider, M., Karshikov, A., Luecke, H., Romisch, J., and Paques, E. (1992) Crystal and molecular structure of human annexin V after refinement. Implications for structure, membrane binding and ion channel formation of the annexin family of proteins. J. Mol. Biol. 223, 683-704. (22) Wood, B. L., Gibson, D. F., and Tait, J. F. (1996) Increased erythrocyte phosphatidylserine exposure in sickle cell disease: flow-cytometric measurement and clinical associations. Blood 88, 1873-80. (23) Tait, J. F., Gibson, D., and Fujikawa, K. (1989) Phospholipid binding properties of human placental anticoagulant protein-I, a member of the lipocortin family. J. Biol. Chem. 264, 7944-9. (24) Larsen, S. K., Solomon, H. F., Caldwell, G., and Abrams, M. J. (1995) [99mTc]tricine: a useful precursor complex for the radiolabeling of hydrazinonicotinate protein conjugates. Bioconjug. Chem. 6, 635-8.

Bioconjugate Chem., Vol. 11, No. 6, 2000 925 (25) Tait, J. F., and Gibson, D. (1994) Measurement of membrane phospholipid asymmetry in normal and sickle- cell erythrocytes by means of annexin V binding. J. Lab. Clin. Med. 123, 741-8. (26) Tait, J. F., and Smith, C. (1991) Site-specific mutagenesis of annexin V: role of residues from Arg-200 to Lys-207 in phospholipid binding. Arch. Biochem. Biophys. 288, 141-4. (27) Thiagarajan, P., and Tait, J. F. (1990) Binding of annexin V/placental anticoagulant protein I to platelets. Evidence for phosphatidylserine exposure in the procoagulant response of activated platelets. J. Biol. Chem. 265, 17420-3. (28) Tait, J. F., Smith, C., and Wood, B. L. (1999) Measurement of phosphatidylserine exposure in leukocytes and platelets by whole-blood flow cytometry with annexin V. Blood Cells Mol. Dis. 25, 271-8. (29) Funakoshi, T., Hendrickson, L. E., McMullen, B. A., and Fujikawa, K. (1987) Primary structure of human placental anticoagulant protein. Biochemistry 26, 8087-92. (30) Maurer-Fogy, I., Reutelingsperger, C. P., Pieters, J., Bodo, G., Stratowa, C., and Hauptmann, R. (1988) Cloning and expression of cDNA for human vascular anticoagulant, a Ca2+-dependent phospholipid-binding protein. Eur. J. Biochem. 174, 585-92.

BC000059V