Evolution of Bombesin Conjugates for Targeted PET Imaging of Tumors

Bombesin receptors are under intense investigation as molecular targets since they are overexpressed in several prevalent solid tumors. We rationally designed and synthesized a series of modified bombesin (BN) peptide analogs to study the influence of charge and spacers at the N-terminus, as well as amino acid substitutions, on both receptor binding affinity and pharmacokinetics. This enabled development of a novel 64/67Cu-labeled BN peptide for PET imaging and targeted radiotherapy of BN receptor-positive tumors. Our results show that N-terminally positively charged peptide ligands had significantly higher affinity to human gastrin releasing peptide receptor (GRPr) than negatively charged or uncharged ligands (IC50: 3.2±0.5 vs 26.3±3.5 vs 41.5±2.5 nM). The replacement of Nle14 by Met, and deletion of D-Tyr6, further resulted in 8-fold higher affinity. Contrary to significant changes to human GRPr binding, modifications at the N-terminal and at the 6th, 11th, and 14th position of BN induced only slight influences on affinity to mouse GRPr. [CuII]-CPTA-[βAla11] BN(7–14) ([CuII]-BZH7) showed the highest internalization rate into PC-3 cells with relatively slow efflux because of its subnanomolar affinity to GRPr. Interestingly, [64/67Cu]-BZH7 also displayed similar affinities to the other 2 human BN receptor subtypes. In vivo studies showed that [64/67Cu]-BZH7 had a high accumulation in PC-3 xenografts and allowed for clear-cut visualization of the tumor in PET imaging. In addition, a CPTA-glycine derivative, forming a hippurane-type spacer, enhanced kidney clearance of the radiotracer. These data indicate that the species variation of BN receptor plays an important role in screening radiolabeled BN. As well, the positive charge from the metallated complex at the N-terminal significantly increases affinity to human GRPr. Application of these observations enabled the novel ligand [64/67Cu]-BZH7 to clearly visualize PC-3 tumors in vivo. This study provides a strong starting point for optimizing radiopeptides for targeting carcinomas that express any of the BN receptor subtypes.


Introduction
In recent years, bombesin (BN) receptors have attracted interest as molecular targets for imaging and therapy pertaining to the fact that all three BN receptor subtypes are overexpressed in many human tumor types [1]. For example, gastrin releasing peptide receptor (GRPr) has been shown to be overexpressed in prostate [2,3], breast [4], small cell lung cancer [5] and gastrointestinal stromal tumors [6]. Prostate cancer has been traditionally among the most difficult malignancies to image due to its multifocal nature, demanding imaging solutions that have high sensitivity and good resolution. Radiolabeled BN-based peptides have significant potential as agents for preoperative tumor localization, assessment of lymph node involvement, staging of disease and possibly for therapeutic monitoring of prostate cancer. As such, a number of radiolabeled BN peptide analogs have been developed as targeting vectors for imaging and radionuclide therapy of GRPr positive tumors [7][8][9][10][11][12][13].
Clinical studies with 99m Tc-and 68 Ga-labeled BN-based peptides have been reported for the imaging of metastasized prostate, breast and gastrointestinal stromal tumors [10,[14][15][16]. A potent BN agonist based peptide labeled with 177 Lu has been studied in phase 1 clinical trials [10,17]. More recently, preclinical studies demonstrated that radiolabeled antagonist based BN peptides might even be superior as targeting vectors compared to agonist peptides [18][19][20][21]. Despite these advances, the limitation imposed by peptide pharmacokinetics with respect to binding and clearance demonstrates that significant improvements of these radiolabeled probes are still required.
PET (Positron Emission Tomography) is a powerful diagnostic imaging modality that enables tomographic, whole body, high sensitivity and quantitative imaging of the distribution of positron emitter-labeled molecules, such as peptides. 68 Ga-labeled peptides have been extensively studied and effectively implemented in the clinical setting [22]. On the other hand, copper-64 is an interesting radionuclide as it is both a positron-(17.8%, E b + max = 656 keV) and a bemitter (39.6%, E b 2 max = 573 keV) with a half-life of 12.7 h. Long-lived isotopes such as copper-64 may therefore provide the ability to visualize the anatomy of interest after unbound probe has been cleared from nearby structures, such as the bladder. This has the potential to improve detection of disease [23]. Several 64 Cu-labeled BN analogs have been evaluated as PET tracers targeting GRP receptor positive tumors [24].
In the present study, a series of BN peptides were synthesized and conjugated to DOTA (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) and CPTA (4-(1,4,8,11-tetraazacyclotetradec-1-yl)-methyl benzoic acid) chelator for labeling with radiocopper. We had several goals in mind when designing this study. Initially, we sought to study the influence of charge at the Nterminus of the radiopeptides on their pharmacologic and biologic properties. Earlier work has suggested that the replacement of a tetraamine ligand (antagonistic BN analog) for 99 m Tc-labeling [25] by negatively charged DTPA (diethylenetriaminepentaacetic acid) caused affinity drop by a factor of approximately 10 3 (Reubi, Schmitt, Maecke, unpublished results). A decrease in binding affinity of a BN antagonist was also observed when a tetraamine ligand was replaced by DOTA [26]. We hypothesized that the same effect could also be seen in the class of radiopeptides presented here (agonistic BN analogs). Secondly, the structure of CPTA-ligand allows a modification with glycine to afford a hippurane-like structural spacer. This structure modification was studied here to evaluate its effect on kidney clearance.
In addition, we and others have recently observed that speciesspecific differences may be of significant importance in the evaluation of bombesin receptor ligands [27]. We therefore studied modifications of BN (7)(8)(9)(10)(11)(12)(13)(14) with regard to amino acid substitution at several positions. The final goal of our research is to develop BN-based potent conjugates for labeling with 64 Cu ( 67 Cu: t 1/2 = 61 h, 100%, E b 2 max = 577 keV, Ec = 185 keV), which could be used in imaging and potentially in radionuclide therapy of BN receptor-positive tumors. As CPTA allows for high labeling yields under very mild labeling conditions, it has been chosen as a bifunctional chelator for the labeling of monoclonal antibodies [28], octreotide [29] and CXCR4 (chemokine receptor 4) ligand [30,31] with radiocopper. On this basis, a series of CPTA/DOTA conjugated BN analogs were constructed for these research goals.
This work is the first to investigate the influence of different charges at the N-terminus of BN analogs on binding affinity, of the hippurane-like spacer molecule on kidney clearance, and of a modification at the 6 th , 11 th and 14 th position of these (radio)metallobombesin analogs in different species. Together, we show these investigations play an important role towards optimizing radioligands for diagnosis and targeted radionuclide therapy of bombesin receptor-positive tumors.

Materials and Methods
All chemicals were obtained from commercial sources and used without further purification. 64 Cu/ 67 Cu nuclides were produced for in vitro assays and the biodistribution studies at the 72 MeV accelerator of the Paul Scherrer Institute (Villigen, Switzerland) by irradiating nat Zn with protons [32]. For small-animal PET imaging, 64 Cu was obtained from Washington University in St. Louis. CPTA was synthesized as described previously [33]. Rink amide MBHA resin and all Fmoc-protected amino acids were commercially available from NovaBiochem (Lä ufelfingen, Switzerland). DOTA-tris( t Bu)-ester was purchased from CheMatech (Dijon, France). [ 111 In]Cl 3 was purchased from Covidien plc (Dublin 2, Ireland). Electrospray ionization (ESI) mass spectroscopy was carried out with a Finnigan SSQ 7000 spectrometer, fast atom bombardment (FAB) mass spectroscopy with a VG 70SE spectrometer and MALDI-MS measurement on a Voyager sSTR equipped with an Nd:YAG laser (Applied Biosystems, Framingham, USA). Analytical HPLC was performed on a Hewlett Packard 1050 HPLC system (Waldbronn 2, Germany) with a multiwavelength detector and a flow-through Berthold LB 506 Cl g-detector (Wildbad, Germany) using a Macherey-Nagel Nucleosil 120 C 18 column (Oensingen, Switzerland). Preparative HPLC was performed on a Metrohm HPLC system LC-CaDI 22-14 (Herisau, Switzerland) with a Macherey-Nagel VP 250/21 Nucleosil 100-5 C 18 column. Quantitative gamma counting was performed on a COBRA 5003 gamma system well counter from Packard Instruments (Meriden, CT, USA). Solid phase peptide synthesis was performed on a semiautomatic peptide synthesizer commercially available from Rink Combichem (Bubendorf, Switzerland). The PC-3 cell line was obtained from ATCC (Manassas, VA) and cultured in Dulbecco's minimal essential medium (DMEM) with 10% fetal calf serum (FCS) from BioConcept (Allschwil, Switzerland). Small-animal PET imaging was performed on a R4 microPET scanner (Concorde Microsystems, Knoxville, TN).

Synthesis
The peptides were synthesized on solid phase using standard Fmoc strategy. The bifunctional chelator CPTA was coupled to the resin-assembled peptide as follows: 6 equivalents CPTA were mixed together with 18 equivalents PyBop, 18 equivalents HoBt and 80 equivalents DIPEA in NMP and immediately incubated with the resin-assembled peptide until the TNBS test was negative (approximately 5 h). DOTA-tris( t Bu)-ester was coupled to the Nterminus of the peptide on resin as follows: 3 equivalents of DOTA-tris( t Bu)-ester, which was pre-activated with 3.3 equivalents of HATU in NMP, was treated with 6 equivalents of DIPEA and immediately incubated with the resin-assembled peptide until the Kaiser test was negative (approximately 4 h).

Preparation of Metallated Conjugates
The peptides used in the following studies are listed in Table 1. Peptide (0.5 mmol) dissolved in 500 mL 0.5 M ammonium-acetatebuffer (pH 5) was incubated with 1.5 mmol CuCl 2 ?2H 2 O predissolved in 0.04 M HCl for 1 h at room temperature, and purified over a SepPak C 18 cartridge (Waters Corp. Milford, MA) preconditioned with 10 mL ethanol and 10 mL water. The cartridge was eluted with 10 mL water followed by 3 mL methanol resulting in Cu II -peptides after evaporation of the methanol. The final product was analyzed with analytical HPLC and MALDI. Using 3 equivalents InCl 3 ?5H 2 O, [In III ]-BZH4 was synthesized at elevated temperature (95uC, 20-25 min) and purified as described above.  Cu), the solution was incubated at room temperature for 1 h. A 1.5 molar excess of CuCl 2 ?2H 2 O was added and incubated again for 0.5 h. Subsequently, radiolabeled peptides were purified utilizing a SepPak C18 cartridge preconditioned with 10 mL methanol and 10 mL water; the cartridge was eluted with 3 mL water, followed by 2 mL ethanol, to afford the pure 64/67 Cu-labeled ligand. For biodistribution studies, the labeling was performed accordingly without adding cold CuCl 2 ?2H 2 O. The solution was prepared for injection by dilution with 0.9% NaCl (0.1% BSA) to afford the radioligand solution. All 64/67 Cu-labeled conjugates were prepared in the same way. The preparation of [ 111 In]-BZH4 was described previously [12]. All radiolabeled peptides were analyzed with HPLC (eluents: A = 0.1% TFA in water and B = acetonitrile; gradient: 0-20 min, 80%-50% A; 20-21 min, 100% B; 21-24 min, 100% B; 25 min, 80% A). Pure 64 Cu was used for small animal PET imaging of PC-3 xenografts.

Binding Affinity and Receptor Subtype Profile
Using [ 125 I-Tyr 4 ] BN as a GRP receptor preferring ligand, the IC 50 values of the nat Cu/ nat In-labeled peptides were measured by in vitro autoradiography of sections of human prostate cancer tissue overexpressing GRP receptors or mouse pancreas tissue expressing mouse GRP receptors. The prostate cancer tissues originated from samples investigated previously [2] or collected prospectively at the Institute of Pathology of the University of Berne in accordance with international ethical guidelines, including informed written consent and approval by the institutional review board. The binding affinity profile of [Cu II ]-BZH7 for three bombesin receptor subtypes was determined by using [ 125 I-D-Tyr 6 , bAla 11 , Phe 13 , Nle 14 ] BN (6)(7)(8)(9)(10)(11)(12)(13)(14) as an universal radioligand. The procedures were described in detail previously [34].

Serum Stability
The procedures were previously described in details [12]. Briefly, 50 mL, 0.6 nmol 111 In-or 64/67 Cu-labeled conjugates were used and incubated with human serum at different time points (0, 1, 4, and 8 h), in triplicate. The HPLC profiles from sample analysis were used to calculate the half-life of disappearance of intact peptide.

Biodistribution Studies with Mice Bearing PC-3 Tumor
After being brought to the condition of anesthesia with isoflurane in an air/oxygen mixture, female athymic nude mice were implanted subcutaneously with approximately 10 million PC-3 tumor cells, which were freshly expanded in 100 mL sterilized PBS solution. Seven to ten days after inoculation the tumors weighed 60-130 mg. The xenografts were injected via tail vein with 10 pmol radiolabeled peptides (about 0.24 MBq 64 Cu and 0.05 MBq 67 Cu), diluted in 0.9% NaCl (0.1% BSA, pH 7.4, total injected volume = 100 mL). For the determination of nonspecific uptake in the tumor or receptor-positive organs, a group of 4 animals were injected with a mixture of 10 pmol radiolabeled peptide/50 mg [Cu II ]-BZH7 in 0.9% NaCl solution (injected volume 150 mL). Mice were sacrificed at 1, 4 and 24 h, and organs of interest collected, rinsed of excess blood, blotted, weighed and counted in a c-counter. The percentage of injected activity per gram (% IA/g) for each tissue was calculated. The total counts injected per animal were determined by extrapolation from counts of an aliquot taken from the injected solution as a standard. All animal experiments were performed in compliance with the Swiss regulations for animal treatment, as approved by the Federal Veterinary Office (Bundesamt für Veterinä rwesen, approval no. 789). Written consent in the form of an official internal document was given.  view. Mice were maintained at 2% isoflurane/air anesthesia for the duration of the imaging. Data acquisitions were performed for 10 min with an energy window of 250-750 keV and a coincidence-timing window of 6 ns. Analysis of the acquired images was performed using ASIPro software (Siemens Medical Solutions USA, Inc., Malvern, PA). PET imaging studies were conducted at Memorial Sloan-Kettering Cancer Center (MSKCC). All work was evaluated and approved by the Institutional Animal Care and Utilization Committee (IACUC) of MSKCC (approval no. 08-07-011). Written consent of this protocol was provided by IACUC.

Statistical Analysis
Data are expressed as mean 6 SD, calculated using Microsoft Excel. The Student's t-test (Origin 6, Microcal Software, Inc., Northampton, MA) was used to determine statistical significance at the 95% confidence level. Values of P,0.05 were considered significantly different.

Synthesis and Labeling
All conjugates (

Receptor Binding Affinity
Competitive binding assays were performed with human GRP receptor-positive cancerous tissue and mouse pancreas tissue (expressing mouse GRP receptor) using [ 125 I-Tyr 4 ]bombesin as radioligand. Table 2 summarizes the binding affinities of the metallopeptides to GRP receptors of mouse and human origin. The Cu II -and In III -complexed peptides exhibit a wide dynamic range of binding (high to moderate affinity) to human GRP receptor (0.4260.13 nM to 41.562.5 nM). In contrast, all of these probes display similar high affinity to mouse GRP receptor (0.2260.07 nM to 1.160.33 nM).
The most promising peptide, Cu-BZH7, was also studied with respect to BN receptor subtype profiles using human cancerous tissues shown to express predominantly a single bombesin receptor subtype. The peptide showed very high binding affinity to all 3 human BN receptor subtypes (0.2760.16 nM to NMB-R; 0.3060.07 nM to GRP-R; 1.460.6 nM to BNRS-3).
The efflux kinetics were studied in PC-3 cells that were exposed to radioligand for 2 h as described for internalization, followed by an acid wash, and then incubated with medium (1% FCS). The results are summarized in Figure 4. To identify the composition of the externalized peptides, [ 111 In]-BZH4 ( 111 In-DOTA-GABA-D-Tyr-Gln-Trp-Ala-Val-bAla-His-Leu-Nle-NH 2 ) was used as a surrogate peptide with high specific activity, required for metabolic studies. Upon 2 h of internalization and acid wash, the externalized radioactivity after 2 h incubation already consisted of approximately 84% metabolites ( 111 In-DOTA: 14%; 111 In-DOTA-GABA-D-Tyr-Gln: main peak, 64%; 111 In-DOTA-GABA-D-Tyr-Gln-Trp-Ala-Val-bAla: 6%) and 16% intact peptide. These results indicate that the internalized ligands can be decomposed quickly in the targeted cells; and their retention in cells is determined mainly by their metabolic stability.

Stability in Human Serum
Serum stability was studied to determine the half-life of disappearance of intact peptides in serum (Table 3). There was less than 3% of radiometal transfer to serum proteins during serum incubation studies. Using the equation of A = A 0 *exp(-k1*t), the half-lives (t 1/2 ) of disappearance of intact peptides in serum were calculated [12]; they varied between 0.55 and 5.    (Table 4). There was also a rapid clearance from GRP-R-negative organs except kidneys and liver. High uptakes were observed both in human prostate tumor xenografts and mouse GRP-R-positive organs; e.g. at 4 h p.i., tumor uptake was 3 Cu]-BZH7 resulted in a .89% reduction of tumor uptake and also a reduction of uptake in GRP-R-positive organs, e.g. .97% in pancreas, 96% in adrenals, 91% in bowel, 84% in stomach, 76% in spleen and bone. These uptake values through co-injected peptide blocking were all significantly decreased (P,0.05). The co-injection of [Cu II ]-BZH7 led to a somewhat increased liver uptake, whereas the uptake in kidneys was partially blocked. The injection of a blocking dose had no significant influence on the uptake in other non-target organs (p.0.05). Due to rapid clearance of the peptides, high tumor-tobackground ratios were obtained (Tables 4 and 5). [ 64/67 Cu]-BZH7 showed the greatest tumor-to-background and tumor-toliver ratios. The kinetics of [ 64/67 Cu]-BZH7 (Table 5) showed a high initial accumulation in the tumor (11.261.5%IA/g at 1 h p.i.), followed by a decreased uptake to 6.6360.80%IA/g at 4 h and to 4.1460.55%IA/g at 24 h p.i., indicating a rapid initial wash out. The ratios between tumor and background (blood and muscle) were .20 at 1 h p.i. and increased to .40 at 24 h. The ratios between tumor and liver or kidney also increased slightly from 1 to 24 h.

Small Animal PET Imaging
Whole-body PET scanning of PC-3 tumor bearing mice was performed with [ 64 Cu]-BZH7, as shown in Figure 6. The PC-3 tumor on the right shoulder was clearly visualized at 1, 4, and 24 h p.i.; and it could be distinguished well from other organs. [ 64 Cu]-BZH7 displayed a high uptake in gut. Pancreas (mouse GRP receptor positive organ), liver, kidneys and the urinary bladder also displayed some activity. The low uptake of [ 64 Cu]-BZH7 was found in the blood pool, which resulted in a high tumor-tobackground ratio. There was a negligible hepatobiliary elimination of the radiopeptide, as implied by a low accumulation in intestine.

Discussion
This study describes synthesis, characterization and evaluation of radiocopper-chelated BN analogs for PET imaging. [ 64 Cu]-BZH7 has shown to be a potential candidate for further development as PET tracer. This is due to its high affinity to NMB, GRP and BB3 receptors and a high rate of internalization into GRP receptor expressing cells.

Binding Affinity
Our earlier unpublished work has suggested that charge differences at the N-terminus of BN targeting peptides could effect drastic changes on receptor binding. Here, charges were introduced at the N-terminus of BN peptides in a dual-purpose strategy to change the charge and functionalize the peptides for radiolabeling by attaching different metal-chelate complexes. Compared to negative-or neutral charges, positive charge was found to significantly improve IC 50 values of BN peptides by a factor of 8.2 and 13.0, respectively, for human GRP receptor. This result indicates that bifunctional chelators which serve to introduce an N-terminal positive charge may be a good choice for the development of copper-64 labeled BN analogs. These findings explain why the positively charged [ 99 m Tc]-labeled bombesin analog [25] displays high affinity to GRP receptor and high accumulation in PC-3 tumors whereas non-positively charged analogs, such as 64 Cu-labeled DOTA-Aoc-BN(7-14) [35] or 111 Inlabeled DOTA-[Lys 3 ] BN [36] show low affinities.
In contrast to the effect of these charge differences on binding to human GRP receptor, the three differently charged peptides ( Table 2 and Figure 1) show only slight differences in the affinity to mouse GRP receptor. This surprising result indicates that mouse GRP receptor is not sensitive to the modifications at the Nterminus. Further, we have shown that the other BN peptide modifications performed in this study are also of little effect. This is a strong indication that the mouse pancreas, which is commonly used to screen the performance of new BN based ligands, may in fact not be a good predictor of probe utility.
A hippurane-type molecule might facilitate clearance through kidney (as is generally preferred for radioligands), and the introduction of a glycine after CPTA leads to a hippurane-type spacer molecule. Therefore, this structural modification was expected to show similar behaviour. Our results ( Figure 5) confirmed this hypothesis. Even when GRP receptor expressed in mice was blocked by an excess of [Cu II ]-BZH7, [ 64/67 Cu]-BZH8 was excreted faster from the kidney than [ 64/67 Cu]-BZH7. This clearly indicates that the hippurane-type spacer molecule integrated in BN analogues hastens renal excretion of injected radioactivity. However, the tumor-to-kidney ratio of [ 64/67 Cu]-BZH8 was not improved because this modification also resulted in a lower GRP receptor affinity and concomitantly a lower tumor uptake.

Conclusion
In this work, we pursued a series of modifications of bombesin receptor targeting peptides in order to generate improved PET and therapeutic radioligands for pre-and clinical investigation. These studies show that charge at the N-terminus of radiometal labeled BN peptides has a significant influence on the rate of internalization and the binding affinity to human GRP receptor. Interestingly, this effect is substantially less significant for binding to mouse GRP receptor. This observation supports earlier findings that the careful selection of animal species and tumor origin is absolutely mandatory in order to evaluate new radioligands in general [46], and for the radiolabeled bombesin analogs in particular [27].
Pharmacokinetic considerations were also evaluated through the introduction of a hippurane-like spacer into a 64/67 Cu-labeled BN analogue. This led to an improved kidney clearance. The replacement of Met 14 by Nle 14 lowered the binding affinity of BN analog to human GRP receptor, which might be a potential cause for the lower internalization rate, as well as decreased tumor and pancreas uptake. The 64/67 Cu labeled BZH7 (CPTA-[bAla 11 ] BN (7)(8)(9)(10)(11)(12)(13)(14)) showed favorable qualities as a targeting vector, which suggest its potential for localization and treatment of GRPreceptor positive tumors. The relatively low in vivo stability of the Cu II -CPTA complexes may be improved by cross-bridging CPTA, a strategy which we are pursuing currently.