Evaluation of radioiodinated protein conjugates and their potential metabolites containing lysine-urea-glutamate (LuG), PEG and closo-decaborate(2-) as models for targeting astatine-211 to metastatic prostate cancer.

Abstract

Introduction:

The use of lysine-urea-glutamate (LuG) for targeting the PSMA antigen on prostate cancer (PCa) is a promising method for delivering the alpha particle-emitting radionuclide astatine-211 (²¹¹At) to metastatic PCa. High kidney localization has been a problem with radiolabeled LuG derivatives, but has been adequately addressed in radiometal-labeled DOTA-LuG derivatives by linker optimization. Herein, we report an investigation of an alternate approach to diminishing the kidney concentrations of radiolabeled LuG-containing compounds.

Methods:

Our approach involves PEGylated LuG moieties and closo-decaborate (2-) moieties conjugated to streptavidin (SAv) or human serum albumin (HSA). After preparing the LuG conjugates, SAv and HSA conjugates were succinylated to decrease their kidney localization and radioiodinated for evaluation in athymic mice bearing C4–2B osseous PCa tumor xenografts.

Results:

Covalently attaching LuG to succinylated SAv and HSA significantly reduced kidney localization, but unfortunately succinylation resulted in decreased tumor concentrations. In contrast, a potential metabolite [¹³¹I]16b, an unconjugated LuG derivative containing a dPEG₄° linker, provided tumor concentrations of ~15% ID/g at 4 h pi. A second unconjugated LuG derivative with a similar structure, but containing a dPEG₁₂° linker, [¹³¹I]16a had tumor concentrations of ~4 %ID/g at 4 h pi. Those results suggest that long PEG linkers also affect tumor localization in a negative manner.

Conclusions:

Conjugation of PEGylated LuG derivatives to proteins can be an effective approach to diminishing kidney localization of radiolabeled LuG reagents, but the protein, linker and the method of linkage need to be further studied. Additionally, modification of the unconjugated 16b to decrease kidney localization may provide PCa targeting agents for use with radiohalogens, including ²¹¹At.

Advances in knowledge and implications for patient care:

This study is the first to evaluate PEGylated LuG and closo-decaborate (2-) moieties conjugated to proteins as potential methods for diminishing the kidney concentrations of radiolabeled LuG-containing compounds.

1. Introduction

Prostate cancer patients treated in the US recorded an overall increase in their 5-year survival rates from 68% to 99% in the years between 1975 to 2014, but the 5-year survival for prostate cancer patients with metastatic cancer remains unacceptably low (~30%) [1]. For several years, our research group has been interested in developing a radiopharmaceutical that targets the alpha-particle emitting radionuclide astatine-211 (²¹¹At) to metastatic prostate cancer to improve patient survival rates. Early evaluations of a radioiodinated intact anti-PSMA antibody (107–1A4) demonstrated high uptake in athymic mice bearing the human prostate cancer LNCaP tumor xenografts (e.g. ~32+ %ID/g at 24 h post injection (pi)) [2]. However, the pharmacokinetics and tumor penetration of the intact antibody was not matched well with the short half-life (7.21 h) of ²¹¹At. Other studies had shown that ²¹¹At-labeled, antibody fragments F(ab’)₂ provide better pharmacokinetics and higher tumor concentrations than the intact antibody [3]. but it was noted that the conjugated p-[²¹¹At]astatobenzoyl labeling moiety released ²¹¹At in vivo. That assessment was based on concentrations found in tissues that localize that radionuclide (lung, spleen, thyroid and stomach) [4]. Based on the instability of astatinated benzoyl labeling moieties when attached to rapidly metabolized Fab´ carriers, studies were conducted to find alternate ²¹¹At labeling moieties that provided higher in vivo stability.

We hypothesized that use of the stronger aromatic boron-halogen bond might provide the stability sought, so a number of studies were conducted which involved astatination of anionic aromatic boron cage moieties and their Fab´ conjugates. Those studies were followed by biodistribution studies in mice to evaluate in vivo stability. The studies demonstrated that ²¹¹At-labeled aromatic anionic boron cage molecules and their Fab´ conjugates were much more stable towards in vivo stability than observed for astatinated aryl compounds [5–7]. The non-toxic and readily obtained closo-decaborate(2-) moiety was chosen for labeling ²¹¹At, as it provided high ²¹¹At labeling yields and very high stability of the bonded ²¹¹At in vivo, while having minimal effect on the Fab´ distribution. A maleimido-B10 derivative was prepared for conjugation with Fab´ fragments, and while it provided a stable attachment of ²¹¹At the inherent kidney localization of Fab´ was not compatible with the use of ²¹¹At due to potential renal toxicity [8].

In an attempt to circumvent the high kidney concentrations, Fab´ conjugates containing an acid-cleavable hydrazone linker were designed and tested [9]. While the hydrazone linker successfully released the radioactivity from the kidney when it was radioiodinated, the hydrazone linker was not successful in decreasing the ²¹¹At concentrations in kidney. To circumvent the kidney localization of ²¹¹At with Fab´, we have also studied the cancer-pretargeting approach utilizing ²¹¹At-labeled biotin derivatives and antibody streptavidin conjugates [10–14]. In collaborative studies, radiolabeled biotin derivatives targeting therapeutic isotopes, such as yttrium-90, to blood-borne cancers has been quite successful [15, 16], but studies using a number of different ²¹¹At-labeled biotin derivatives have not been successful. Those studies are continuing, but alternate metastatic prostate cancer targeting agents have been studied.

The use of lysine-urea-glutamate (LuG) for targeting the PSMA antigen on prostate cancer has been investigated for a number of years and is currently a very active area of research for targeting both diagnostic and therapeutic radionuclides to metastatic prostate cancer. Some years ago, our research group prepared a radioiodinated biotin compound that also contained a LuG moiety, a dPEG₈° moiety and a closo-decaborate(2-) radiohalogen labeling moiety to target the PSMA antigen on prostate cancer cells. That reagent was labeled with ¹²⁵I and was administered to athymic mice bearing LNCaP prostate cancer xenografts. High tumor xenograft concentrations (e.g. 34% ID/g at 1 h & 39% at 4 h pi) were obtained, however, very high kidney concentrations (160% ID/g at 1 h pi) were also obtained [17]. High kidney concentrations have been obtained in studies by other research groups, leading to a large effort to decrease that localization. Excellent progress has been made in that effort, with kidney concentrations as low as 2.87 %ID/g at 1h post injection being achieved [18]. Herein, we report on an alternate approach to diminishing the kidney concentrations of radiolabeled LuG compounds, that of conjugation of the LuG motif with proteins that are not readily filtered by the kidney. The protein conjugates investigated contain the closo-decaborate(2-) moiety for radiohalogenation, a LuG moiety for binding with the PSMA antigen on prostate cancer cells, and a polyethylene glycol (PEG) linker to allow LuG binding away from the steric protein. Two proteins previously studied in our lab, streptavidin and HSA, were evaluated as succinylated LuG-containing conjugates. The protein-LuG conjugates, as well as potential metabolites, were radioiodinated and administered to athymic mice bearing C4–2B prostate cancer xenografts, and their biodistributions and tumor localization were evaluated.

2. Materials and Methods

2.1. General

All reagents obtained from commercial sources were analytical grade or better and were used without further purification. Recombinant streptavidin was obtained from Prozyme, Inc (Agilent Technologies, Santa Clara, CA) and was a single peak by size exclusion HPLC. Human serum albumin (HSA) (fraction V) was obtained from ICN Biomedicals, Inc (Costa Mesa, CA) as >97% purity. Solvents for HPLC analysis were obtained as HPLC grade and were filtered (0.2 μm) prior to use. Sephadex G-25 desalting columns (PD-10 and NAP-10; GE Healthcare) and Vivaspin 6 centrifugation filters (30 kDa molecular weight cutoff (MWCO); GE Healthcare) were obtained from Fisher Scientific (Houston, TX).

2.2. Radioactivity

All radioactive materials were handled according to approved Radiation Safety protocols at the University of Washington. [¹²⁵I]NaI and [¹³¹I]NaI were purchased from Perkin-Elmer Life and Analytical Sciences, Inc. (Waltham, MA, USA), as high concentration solutions in 0.1 N NaOH Radioiodinations were conducted in a charcoal-filtered Plexiglas enclosure (Biodex Medical Systems, Inc., Shirley, NY, USA) housed in a radiochemical fume hood. The radioiodination reactions were carried out in vials capped with Teflon-coated septa vented through a 10-mL charcoal-filled syringe. Additions of reagents to, or removal of materials from, the radioiodination vessel were conducted by passing a syringe needle through the septa.

Quantification of ¹²⁵I and ¹³¹I was accomplished on a Capintec CRC-15R Radioisotope Calibrator using the manufacturer’s setting for those radionuclides. Tissue samples were counted in a Wallac 140 gamma counter with the following window settings: channels 35–102 (20–90 keV) and 165–185 (300–450 keV) for ¹²⁵I and ¹³¹I. In experiments involving quantification of samples that contain both ¹³¹I and ¹²⁵I, the ¹²⁵I counts were compensated for spillover from the ¹³¹I window (generally ~15%).

2.3. Chromatography

HPLC separations of the nonradioactive compounds were obtained using a system that contained a Hewlett-Packard quaternary 1050 gradient pump, a variable wavelength UV detector (254 nm) and an ELSD 2000 evaporative light scattering detector (Alltech, Deerfield, IL, USA). Analysis of the HPLC data was conducted on Hewlett-Packard HPLC ChemStation software. Reversed-phase HPLC chromatography was carried out on an Alltech Altima C18 column (5 μm, 250 × 4.5 mm) using a gradient solvent system at a flow rate of 1 ml/min. The gradient began with 0.05 M, pH 5.5, aqueous Et₃NHOAc and was increased linearly to 100% methanol (MeOH) over a 15-min period. Following this, the elution continued with 100% MeOH for an additional 5 min. The retention time (t_R) of each iodinated compound is provided with the experimental procedure.

Products were purified from crude reaction mixtures using a Biotage SP Flash Purification System (Charlottesville, VA, USA) on a reversed-phase C18 FLASH 25+M column or 40+M column. The purification used a gradient mixture composed of MeOH and 0.05 M Et₃NHOAc TEAA. The gradient started with 100% 0.05 M TEAA, with MeOH being increased linearly to 100% over the next 20 min. Fractions were collected based on UV detection at 215 or 254 nm. Identification of fractions containing pure products as accomplished by analytical HPLC. The fractions containing pure compound were combined, solvent was evaporated, and the product was isolated to provide the yields listed.

The product mixtures from radioiodination reactions were analyzed by radio-HPLC using a C18 column (Altima C18, 5 μm, 250 × 4.5 mm; Alltech) eluting at a flow rate of 1 ml/min using the same gradient conditions as described for nonradioactive iodination products.

2.4. Spectral Analyses

¹H NMR and mass spectral analyses were obtained on all new compounds. The data obtained is provided with the experimental procedures. Spectral data for compounds previously reported are not listed but were consistent with the reported data. ¹H and ¹¹B NMR spectra were obtained on a Bruker AV-500 (500 MHz for ¹H and 160.4 MHz for ¹¹B). ¹H NMR data are referenced to tetramethylsilane as an internal standard (δ=0.0 ppm), and ¹¹B NMR data are referenced to BF₃·OEt₂ as an external standard. Mass spectral data were obtained on either a Waters Micromass Quattro Micro API tandem quadrupole mass spectrometer, (QHQ) MS/MS, with a liquid chromatograph inlet system and atmospheric pressure chemical ionization or electrospray ionization for low-resolution mass spectra (LRMS), or a Bruker APEX Qe 47e Fourier transform (ion cyclotron resonance) mass spectrometer, [FT (ICR)]MS, using an infusion inlet system and electrospray ionization for high-resolution mass spectra (HRMS).

2.5. Syntheses

Compounds 1 [6], 2 [6], and 10 [19] were synthesized using published procedures. Reagents 6 and 8 were obtained from Sigma Aldrich (St. Louis, MO). Reagents 4a, 4b, and carboxylate precursors to tetrafluoro esters 11a and 11b were provided by Quanta BioDesign, LTD (Plain City, OH, USA).

3-B₁₀H₉-dioxa-5-CO₂TFP-Ph-NH₂, 3.

B₁₀H₉-dioxa-NH₂, 2 (1.0 g, 1.954 mmol) and triethylamine (TEA) (0.817 mL, 5.86 mmol) in anhydrous DMF (20 mL) were added dropwise over 30 min to a solution of 3,5-CO₂TFP-Ph-NH₂, 1 (1.865 g, 3.91 mmol) and anhydrous DMF (20 mL) at room temperature. After volatile materials were removed on a rotary evaporator under vacuum, the crude product was dissolved in 80% MeOH/H₂O (20 mL) and purified using the Biotage SP Flash Purification System to yield 1.15 g (71.5%). HPLC t_R = 12.5 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.09–1.18 (m, 9H), 1.29 (t, J = 7.3 Hz, 18H), 3.20 (q, J = 7.3 Hz, 12H), 3.29–3.32 (m, 4H), 3.57–3.66 (m, 4H), 3.73–3.80 (m, 4H), 6.34–6.46 (m, 1H), 7.52 (s, 1H), 7.59 (s, 1H), 7.89 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.69 (1B), −5.13 (1B), −15.49 (1B), −25.58 (4B), −28.77 (2B), −31.81 (1B). LRMS (ES^–) C₂₁H₃₀B₁₀F₄N₄O₆ (M)^–; calcd: 620.31; found: 620.31.

3-B₁₀H₉-dioxa-5-dPEG₁₂-CO₂H-Ph-NH₂ and 3-B₁₀H₉-dioxa-5-dPEG₄-CO₂H-Ph-NH₂, 5a and 5b.

A quantity of 3 (200 mg, 0.243 mmol) was dissolved in anhydrous DMF (5 mL), then a solution containing NH₂-dPEG₁₂-CO₂H, 4a (150 mg, 0.243 mmol) or NH₂-dPEG₄-CO₂H, 4b (64.5 mg, 0.243 mmol), TEA (102 μL, 0.729 mmol) and anhydrous DMF (5 mL) was added. The reaction solution was stirred at room temperature for 1 h. After volatile materials were evaporated on a rotary evaporator under vacuum, the crude product was dissolved in 50% MeOH/H₂O (6 mL) and purified using the Biotage SP Flash Purification System. Isolation yielded 258 mg (83%) of 5a. HPLC t_R = 9.1 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.15–1.19 (m, 9H), 1.29 (t, J = 7.3 Hz, 18H), 2.44 (t, J = 6.9 Hz, 2H), 3.20 (q, J = 7.3 Hz, 12H), 3.50–3.77 (m, 50H), 7.24 (s, 1H), 7.30 (s, 1H), 7.51 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.60 (1B), −5.22 (1B), −15.48 (1B), −25.66 (4B), −28.79 (2B), −31.75 (1B). LRMS (ES⁻) C₄₂H₈₄B₁₀N₅O₉ (M+H)⁻; calcd: 1072.68; found: 1072.68. Isolation yielded 174 mg (67%) of 5b. HPLC t_R = 7.2 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.10–1.19 (m, 9H), 1.29 (t, J = 7.3 Hz, 18H), 2.49 (t, J = 6.5 Hz, 2H), 3.20 (q, J = 7.3 Hz, 12H), 3.58–3.80 (m, 30H), 7.24 (s, 1H), 7.30 (s, 1H), 7.51 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.63 (1B), −5.18 (1B), −15.54 (1B), −25.62 (4B), −28.85 (2B), −31.87 (1B). LRMS (ES⁻) C₂₆H₅₁B₁₀N₅O₁₁ (M)⁻; calcd: 719.46; found: 719.46.

3-B₁₀H₉-dioxa-5-dPEG₁₂CO₂H-Ph-N=C=S, 7a.

A solution containing 5a (100 mg, 0.078 mmol), 1,1′-thiocarbonyldiimidazole, 6(TCDI) (18.2 mg, 0.102 mmol) and anhydrous DMF (3 mL) was stirred at room temperature for 1 h. The reaction solution was washed with 15 mL of 20% ethyl acetate (EtOAc)/hexanes five times. The light-yellow product was dried under high vacuum for 4 h, then used directly for the next reaction step without further purification. Yield 92 mg (89%). HPLC t_R = 11.9 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.12–1.21 (m, 9H), 1.29 (t, J = 7.3 Hz, 18H), 2.45 (t, J = 6.9 Hz, 2H), 3.20 (q, J = 7.3 Hz, 12H), 3.48–3.79 (m, 50H), 7.88 (s, 1H), 8.25 (s, 1H), 8.58 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.56 (1B), −5.28 (1B), −15.43 (1B), −25.61 (4B), −28.83 (2B), −31.71 (1B). LRMS (ES⁻) C₄₃H₈₁B₁₀N₅O₁₉S (M)⁻; calcd: 1113.62; found: 1113.65.

3-B₁₀H₉-dioxa-5-dPEG₁₂CO₂H-Ph-Lys-Ac, 9a.

A solution of 7a (50 mg, 0.038 mmol), N_α/MeOH (2 mL) was stirred at room temperature for 16 h. The crude solution was purified using the Biotage SP Flash Purification System. Yield 44 mg (77%). HPLC t_R = 9.1 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.09–1.20 (m, 9H), 1.31 (t, J = 7.3 Hz, 18H), 1.41–1.54 (m, 4H), 1.79–1.94 (m, 3H), 1.98 (s, 3H), 2.08–2.17 (m, 3H), 2.38–2.45 (m, 2H), 3.16–3.25 (m, 12H), 3.59–3.83 (m, 60H), 4.21–4.36 (m, 1H), 7.90 (s, 1H), 7.97 (s, 1H), 8.04 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.54 (1B), −5.21 (1B), −15.58 (1B), −25.64 (4B), −28.83 (2B), −31.72 (1B). LRMS (ES⁻) C₅₁H₉₆B₁₀N₇O₂₂S (M-H)⁻; calcd: 1300.72; found: 1300.72.

3-B₁₀H₉-dioxa-5-dPEG₄CO₂H-Ph-Lys-Ac, 9b.

A solution containing 5b (100 mg, 0.108 mmol), thiocarbonyl diimidazole, 6 (25.1 mg, 0.141 mmol) and anhydrous DMF (3 mL) was stirred at room temperature for 1 h. The reaction solution was washed with 18 mL of 20% EtOAc/hexanes five times, dried under high vacuum for 2 h to obtain a light-yellow tacky solid product 7b which was used without further purification in the next reaction step. A solution of compound 7b (50 mg, 0.052 mmol), N-acetyl lysine, 8(14.6 mg, 0.078 mmol), TEA (21.7 μL, 0.156 mmol) and 50% anhydrous DMF/MeOH (2 mL) was stirred at room temperature for 16 h. The crude solution was purified using the Biotage SP Flash Purification System. Yield 52 mg (87%). HPLC t_R = 8.0 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.12–1.23 (m, 9H), 1.30 (t, J = 7.3 Hz, 18H), 1.42–1.54 (m, 4H), 1.78–1.94 (m, 3H), 1.98 (s, 3H), 2.08–2.17 (m, 3H), 2.38–2.46 (m, 2H), 3.15–3.23 (m, 12H), 3.60–3.83 (m, 28H), 4.23–4.28 (m, 1H), 7.89 (s, 1H), 7.98 (s, 1H), 8.02 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.51 (1B), −5.22 (1B), −15.56 (1B), −25.63 (4B), −28.82 (2B), −31.78 (1B). LRMS (ES⁻) C₃₅H₆₅B₁₀N₇NaO₁₄S (M+Na)⁻; calcd: 972.51; found: 972.51.

BOC-dPEG₁₂-CO₂TFP, 11a, and BOC-dPEG₄-CO₂TFP, 11b.

A quantity of CF₃CO₂TFP was added slowly to a solution of TEA and BOC-dPEG₁₂CO₂H or BOC-dPEG₄CO₂H in anhydrous acetonitrile (20 mL) under ice-bath temperature in a molar ratio of approximately 1:0.6:1.2, then the reaction solution was stirred for 30 min. The solution was evaporated to dryness on a rotary evaporator under vacuum, then dried under high vacuum overnight. The TFP esters 11a and 11b were used directly for the next reaction step without further purification. The reactions provided quantitative yields of product as light-yellow oils. 11a. HPLC t_R = 14.0 min. ¹H NMR (CDCl₃, 500 MHz): δ 1.44 (s, 9H), 2.60 (t, J = 6.2 Hz, 2H), 2.95 (t, J = 6.3 Hz, 2H), 3.54 (t, J = 5.1 Hz, 2H), 3.62–3.67 (m, 42H), 3.77 (t, J = 6.2 Hz, 2H), 3.89 (t, J = 6.2 Hz, 2H), 5.12 (s, 1H), 6.46–6.57 (m, 1H). HRMS (ES⁺) C₃₈H₆₃F₄NNaO₁₆ (M+Na)⁺; calcd: 888.3981; found: 888.3991. 11b. HPLC t_R = 14.0 min. ¹H NMR (CDCl₃, 500 MHz): δ 1.44 (s, 9H), 2.96 (t, J = 6.2 Hz, 2H), 3.54 (t, J = 5.1 Hz, 2H), 3.62–3.68 (m, 14H), 3.89 (t, J = 6.2 Hz, 2H), 5.08 (s, 1H), 6.47–6.58 (m, 1H). LRMS (ES⁺) C₂₂H₃₂F₄NO₈ (M+H)⁺; calcd: 514.21; found: 514.21.

LuG-dPEG₁₂-BOC, 12a, and LuG-dPEG₄-BOC, 12b. .

Compound 11a or 11b (0.4g and 0.434 g, respectively) was reacted with lysine-urea-glutamate, 10 (LuG TFA salt) and NaHCO₃ in a molar ratio of approximately 1 : 1 : 7 in a 50 : 50 DMF : H₂O solution. The reaction mixture was stirred at room temperature for 1 h, then acetic acid was added to the reaction mixture to destroy the excess NaHCO₃ and protonate the carboxylate groups. Following acidification, 12a and 12b were purified using the Biotage SP Flash Purification System. 12a. Yield 0.29 g (61.6%). HPLC t_R = 10.9 min. ¹H NMR (CD₃OD, 500 MHz): δ 1.39–1.42 (m, 2H), 1.44 (s, 9H), 1.49–1.56 (m, 2H), 1.62–1.69 (m, 1H), 1.79–1.86 (m, 1H), 1.87–1.95 (m, 1H), 2.08–2.15 (m, 1H), 2.36–2.41 (m, 2H), 2.43 (t, J = 6.9 Hz, 2H), 3.17–3.23 (m, 6H), 3.50 (t, J = 5.6 Hz, 2H), 3.60–3.64 (m, 42H), 3.71 (t, J = 6.2 Hz, 2H), 4.19–4.25 (m, 2H). HRMS (ES⁺) C₄₄H₈₂N₄NaO₂₂ (M+Na)⁺; calcd: 1041.5318; found: 1041.5215. 12b. Yield 0.37 g (65.6%). HPLC t_R = 9.8 min. ¹H NMR (CD₃OD, 500 MHz): δ 1.39–1.43 (m, 2H), 1.44 (s, 9H), 1.50–1.54 (m, 2H), 1.65–1.69 (m, 1H), 1.84–1.88 (m, 2H), 2.10–2.14 (m, 1H), 2.41–2.46 (m, 4H), 3.13–3.18 (m, 4H), 3.33–3.37 (m, 2H), 3.62–3.74 (m, 14H), 4.22–4.35 (m, 2H). HRMS (ES⁺) C₂₈H₅₀N₄NaO₁₄ (M+Na)⁺ Calcd: 689.3221. Found: 689.3219.

3-B₁₀H₉-dioxa-5-LuG-dPEG₁₂-Ph-NH₂,14a.

A solution of LuG-dPEG₁₂-BOC, 12a (0.26 g, 0.255 mmol) and formic acid (3 mL) was stirred at room temperature for 2 h. The reaction solution was evaporated to dryness on a rotary evaporator under vacuum, then dried under high vacuum for overnight. The free amine product 13a was used directly for the next reaction step without further purification. Compound 3 (150 mg, 0.182 mmol) was dissolved in anhydrous DMF (5 mL), then a solution containing LuG-dPEG₁₂-NH₂, 13a (168 mg, 0.182 mmol), TEA (76 μL, 0.547 mmol) and anhydrous DMF (5 mL) was added. The reaction solution was stirred at room temperature for 4 h. After volatile materials were evaporated on a rotary evaporator under vacuum, the crude product was dissolved in 50% MeOH/H₂O (6 mL) and purified using the Biotage SP Flash Purification System. Yield 157 mg (54.7%). HPLC t_R = 8.3 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.11–1.21 (m, 9H), 1.32 (t, J = 7.3 Hz, 18H), 1.39–1.46 (m, 2H), 1.50–1.56 (m, 2H), 1.64–1.71 (m, 1H), 1.79–1.86 (m, 1H), 1.93–2.00 (m, 1H), 2.08–2.15 (m, 1H), 2.34–2.38 (m, 2H), 2.46 (t, J = 6.9 Hz, 2H), 3.03–3.11 (m, 2H), 3.20 (q, J = 7.3 Hz, 12H), 3.62–3.83 (m, 50H), 4.14–4.19 (m, 2H), 7.27 (s, 1H), 7.33 (s, 1H), 7.54 (s, 1H). 11B NMR (CD₃OD, 160.4 MHz): δ −1.57 (1B), −5.22 (1B), −15.54 (1B), −25.65 (4B), −28.82 (2B), −31.76 (1B). LRMS (ES^–) C₅₄H₁₀₂B₁₀N₈O₂₅ (M)⁻; calcd: 1372.80; found: 1373.81.

3-B₁₀H₉-dioxa-5-LuG-dPEG₄-Ph-NH₂, 14b.

A solution of LuG-dPEG₄-BOC, 12b (0.30 g, 0.45 mmol) and formic acid (3 mL) was stirred at room temperature for 2 h. The reaction solution was evaporated to dryness on a rotary evaporator under vacuum, then dried under high vacuum for overnight. The free amine compound 13b was used directly for the next reaction step without further purification. Compound 3 (200 mg, 0.243 mmol) was dissolved in anhydrous DMF (6 mL), then a solution containing LuG-dPEG₄-NH₂ 13b (138 mg, 0.243 mmol), TEA (102 μL, 0.729 mmol) and anhydrous DMF (6 mL) was added. The reaction solution was stirred at room temperature for 4 h. After volatile materials were evaporated on a rotary evaporator under vacuum, the crude product was dissolved in 50% MeOH/H₂O (7 mL) and purified using the Biotage SP Flash Purification System. Yield 176 mg (59.2%). HPLC t_R = 6.3 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.12–1.20 (m, 9H), 1.32 (t, J = 7.3 Hz, 18H), 1.39–1.45 (m, 2H), 1.50–1.55 (m, 2H), 1.65–1.71 (m, 1H), 1.79–1.85 (m, 1H), 1.92–1.99 (m, 1H), 2.09–2.15 (m, 1H), 2.35–2.39 (m, 2H), 2.45 (t, J = 6.9 Hz, 2H), 3.04–3.11 (m, 2H), 3.20 (q, J = 7.3 Hz, 12H), 3.29–3.33 (m, 4H), 3.57–3.83 (m, 26H), 4.15–4.20 (m, 2H), 7.28 (s, 1H), 7.33 (s, 1H), 7.55 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.53 (1B), −5.14 (1B), −15.46 (1B), −25.47 (4B), −28.78 (2B), −31.69 (1B). HRMS (ES⁻) C₃₈H₇₁B₁₀N₈O₁₇ (M+H)⁻ Calcd: 1021.5868. Found: 1021.5889.

3-B₁₀H₉-dioxa-5-LuG-dPEG₁₂-Ph-N=C=S, 15a.

A solution containing 14a (100 mg, 0.063 mmol), thiocarbonyl diimidazole, 6 (16.3 mg, 0.082 mmol) and anhydrous DMF (3 mL) was stirred at room temperature for 1 h. The reaction solution was washed with 15 mL of 20% EtOAc/hexanes five times. The remained light-yellow product was dried under high vacuum for 4 h, then used directly in the next reaction step without further purification. Yield 103 mg (100%). HPLC t_R = 10.7 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.09–1.22 (m, 9H), 1.32 (t, J = 7.3 Hz, 18H), 1.39–1.46 (m, 2H), 1.49–1.56 (m, 2H), 1.65–1.72 (m, 1H), 1.80–1.86 (m, 1H), 1.94–2.00 (m, 1H), 2.09–2.15 (m, 1H), 2.35–2.38 (m, 2H), 2.46 (t, J = 6.9 Hz, 2H), 3.04–3.11 (m, 2H), 3.20 (q, J = 7.3 Hz, 12H), 3.63–3.84 (m, 50H), 4.14–4.20 (m, 2H), 7.89 (s, 1H), 8.26 (s, 1H), 8.59 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.55 (1B), −5.22 (1B), −15.52 (1B), −25.66 (4B), −28.82 (2B), −31.79 (1B). LRMS (ES⁻) C₅₅H₁₀₀B₁₀N₈O₂₅S (M)⁻ ; calcd: 1414.75; found: 1414.76.

3-B₁₀H₉-dioxa-5-LuG-dPEG₁₂-Ph-Lys-Ac, 16a.

A solution of 15a (50 mg, 0.031 mmol), N-acetyl-lysine, 8 (11.6 mg, 0.062 mmol), TEA (21.5 μL, 0.155 mmol) and 50% anhydrous DMF/MeOH (2 mL) was stirred at room temperature for 16 h. The crude solution was purified using the Biotage SP Flash Purification System. Yield 39 mg (70%). HPLC t_R = 8.5 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.10–1.20 (m, 9H), 1.31 (t, J = 7.3 Hz, 18H), 1.42–1.54 (m, 8H), 1.64–1.71 (m, 3H), 1.79–1.93 (m, 4H), 1.98 (s, 3H), 2.07–2.17 (m, 3H), 2.38–2.45 (m, 4H), 3.16–3.25 (m, 14H), 3.58–3.83 (m, 60H), 4.21–4.36 (m, 3H), 7.90 (s, 1H), 7.97 (s, 1H), 8.04 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.53 (1B), −5.22 (1B), −15.55 (1B), −25.61 (4B), −28.79 (2B), −31.72 (1B). HRMS (ES⁻) C₆₃H₁₁₅B₁₀N₁₀NaO₂₈S (M+Na-H)⁻ ; calcd: 1624.8350; found: 1624.8346.

3-B₁₀H₉-dioxa-5-LuG-dPEG₄-Ph-Lys-Ac, 16b.

A solution containing 14b (100 mg, 0.082 mmol), thiocarbonyl diimidazole, 6 (18.9 mg, 0.106 mmol) and anhydrous DMF (3 mL) was stirred at room temperature for 1 h. The reaction solution was washed with 15 mL of 20% EtOAc/hexanes five times to provide 3-B₁₀H₉-dioxa-5-LuG-dPEG₄-Ph-N=C=S, 15b, as a light-yellow compound. The product was dried under high vacuum for 4 h and then was used directly for the next reaction step without further purification. A solution of compound 15b (50 mg, 0.04 mmol), N-acetyl-lysine, 8 (14.9 mg, 0.079 mmol), TEA (27.5 μL, 0.198 mmol) and 50% anhydrous DMF/MeOH (2 mL) was stirred at room temperature for 16 h. The crude product was purified using the Biotage SP Flash Purification System. Yield 42 mg (73%). HPLC t_R = 7.6 min. ¹H NMR (CD₃OD, 500 MHz): δ 0.09–1.19 (m, 9H), 1.30 (t, J = 7.3 Hz, 18H), 1.41–1.54 (m, 8H), 1.65–1.71 (m, 3H), 1.78–1.93 (m, 4H), 1.98 (s, 3H), 2.08–2.17 (m, 3H), 2.38–2.46 (m, 4H), 3.15–3.24 (m, 14H), 3.58–3.82 (m, 28H), 4.22–4.36 (m, 3H), 7.89 (s, 1H), 7.97 (s, 1H), 8.03 (s, 1H). ¹¹B NMR (CD₃OD, 160.4 MHz): δ −1.55 (1B), −5.20 (1B), −15.54 (1B), −25.61 (4B), −28.80 (2B), −31.74 (1B). LRMS (ES⁻) C₄₇H₈₄B₁₀N₁₀NaO₂₀S (M+Na)⁻; calcd: 1273.64; found: 1273.64.

2.6. Protein Conjugation

2.6.1. Preparation of SAv and HSA conjugates, SAv-7a, SAv-15a, HSA-7a and HSA-15a.

An amount (10, 20 or 40 equiv.) of 7a or 15a in H₂O (10 mg/mL) was combined with 3.0–3.5 mg of SAv or HSA (10 mg/mL in HEPES buffer pH 8.6). The reaction mixtures were gently tumbled overnight at room temperature. The excess 7a or 15a was separated from the SAv or HSA conjugates using BioRad EconoPak 10 DG columns eluting with PBS. The collected protein-containing fractions were combined, concentrated and washed 5 times with fresh PBS.

2.6.2. Succinylation of SAv and HSA conjugates, sSAv-7a, sSAv-15a, sHSA-7a and sHSA-15a.

The succinylation reactions were conducted in a manner similar to that previously described [14]. To 1.5– 2.5 mg of LuG-conjugated SAv or HSA in solution (3.7–5 mg/mL) was added 22.4–33.6 μL of 14 mg/mL succinic anhydride in DMSO. After 1 h at room temperature, the succinylated proteins were purified using PD10 columns. The collected protein-containing fractions were combined and concentrated to the desired concentration.

2.7. Radioiodinations

2.7.1. Radioiodination of Compounds 9a, 9b, 16a and 16b

To 90 μL of 1 mg/mL 9a, 9b, 16a and 16b in 1% acetic acid in MeOH, was added 2–5 μL of [¹²⁵I]NaI or [¹³¹I]NaI (18.5–22.2 MBq; 0.5–0.6 mCi), followed by 10 μL of 1 mg/mL N-chlorosuccinimide (NCS) in MeOH. The reaction was allowed to proceed for 1 min at room temperature; then, 10 0L of a 1 mg/mL solution of Na₂S₂O₅ in H₂O was added to quench the reaction. The radioiodination reaction mixture was purified by collection of the radioactive eluate at the appropriate t_R from the radio-HPLC.

2.7.2. Radioiodination of SAv and HSA conjugates

The SAv and HSA conjugates (i.e., SAv-7a, SAv-15a, sSAv-7a, sSAv-15a, HSA-7a, HSA-15a, sHSA-7a and sHSA-15a) were radioiodinated using the following protocol: 20–25 μL of 500 mM sodium phosphate, pH 6.8 was added to 100–200 μg of SAv or HSA conjugate in PBS. To that solution was added 1–5 μL of [¹²⁵I]NaI or [¹³¹I]NaI (7.4–37 MBq; 0.2–1 mCi), followed by 10 μL of chloramine-T (ChT) (1 mg/mL) in H₂O. The reaction was allowed to proceed for 1 min at room temperature; then 10 μL of a 1 mg/mL solution of Na₂S₂O₅ in H₂O was added to quench the reaction. The reaction mixture was placed on a NAP-10 column and was eluted with PBS. The amount of activity associated with the protein was measured in a dose calibrator.

2.8. Animal Studies

Prior to initiation of studies, animal use and procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Washington. Animal care and use was conducted in accordance with the NIH guidelines as described in NIH Publication 86–23 (Guide for the Care and Use of Laboratory Animals). Male athymic mice, obtained from Charles River Laboratories (Hollister, CA, USA) were housed for 1 week in the isolator facility prior to beginning the study. Dual-label experiments were conducted to minimize the number of mice required to obtain the information sought.

2.8.1. Tumor xenografts

The studies employed athymic (BALB/c nu/nu) mice bearing C4–2B prostate cancer xenografts implanted in the hind flank. The C4–2B [20] prostate cancer cells, a LNCaP subline, was maintained in RPMI 1640 medium with 10% fetal bovine serum. Subcutaneous tumor xenografts were formed in 6-week old athymic mice as previously described [21]. Briefly, adequate numbers of mice were injected subcutaneously with 100 μL of 2 × 10⁶ C4–2B prostate cancer cells in a 1:1 mixture with Matrigel (Corning Life Sciences, Tewksbury, MA, USA) into an adequate number of mice to obtain the number of animals required in the study. Groups of 5 mice bearing C4–2B xenografts were used in the studies. Mice were euthanized by cervical dislocation under anesthesia if tumors reached 1000 mm³ or if they were compromised.

2.8.2. Biodistributions

In each experiment, the radioactive peaks corresponding to the desired products were isolated from the HPLC (non-protein) or size-exclusion column (protein) to provide the radioiodinated compounds for study. The isolated compounds (¹²⁵I- and ¹³¹I-labeled) were mixed to prepare an admixture containing predetermined quantities (74–185 kBq; 2–5 μCi) of each radioiodinated compound. The admixture was diluted with PBS to prepare adequate total volume to provide solutions for injection of ~100 μL per mouse. This quantity of radioiodinated compound was injected into five athymic mice per timepoint via the lateral tail vein. The actual amount of injectate each animal received was determined by weighing the administering syringe before and after injection. The mice were sacrificed by cervical dislocation at preset time points (e.g., 1, 4 and 24 h pi). Blood samples were obtained by cardiac puncture immediately before sacrifice. Urine samples were obtained by syringe bladder tap at the time the tissues were excised. The tissues excised included muscle, lung, kidney, spleen, liver, intestines, neck (with thyroid), and stomach. Excised tissues were blotted free of blood, weighed, and counted in a gamma counter. Blood weight was estimated to be 6% of the total body weight. Quantification of percent injected dose (%ID) and percent injected dose per gram (%ID/g) in the tissues was accomplished by counting 4 × 1 μL of the injectate as standards for total ¹²⁵I and ¹³¹I counts administered.

3. Results

3.1. Syntheses

Bifunctional compounds 7a and 7b, and trifunctional compounds 15a and 15b, were synthesized for conjugation to recombinant streptavidin (SAv) and human serum albumin (HSA). Additionally, the potential protein degradation metabolites, 9a, 9b, 16a and 16b, obtained from the bi- and trifunctional compounds, were synthesized. The synthetic paths used to prepare the LuG conjugates studied are shown in Figures 1 and 2. The identities of all new compounds synthesized, except 9b and 16b, were confirmed using combinations of ¹H, ¹¹B NMR, and high-resolution mass spectrometry (HRMS). The methods for the preparation of 9b and 16b are very similar to those for preparation of 9a and 16a, thus for those compounds, and some of the intermediates to them, low-resolution mass spectrometry (LRMS) was used to confirm identity.

Figure 1.

Synthetic route to preparation of study compounds that contain the closo-decaborate(2-) and PEG moieties without the lysine-urea-glutamate (LuG) targeting structure. Note that the dianionic aromatic closo-decaborate(2-) cage is enclosed in parentheses represented by a cage structure (B10) without showing the attached protons.

Figure 2.

Synthetic route to preparation of study compounds (PLuG) that contain the closo-decaborate(2-) and PEG moieties and also contain the lysine-urea-glutamate (LuG) targeting structure. Note that the dianionic aromatic closo-decaborate(2-) cage is enclosed in parentheses represented by a cage structure (B10) without showing the attached protons.

3.2. Conjugation and Succinylation

Conjugations of bifunctional compounds 7a, 7b, 15a and 15b were conducted in HEPES buffer at pH 8.6 Varying equivalents, 10, 20 or 40, of the bifunctional compounds were offered in the conjugation reactions. The isolated yields from the conjugations of 7a, 15a with SAv and HSA to prepare HSA-7a, HSA-15a, SAv-7a and SAv-15a were high (~82%). The isolated protein yields for succinylation of the conjugates to form sHSA-7a, sHSA-15a, sSAv-7a and sSAv-15a were lower, being in the range of 54–75%. The SAv and HSA conjugates were analyzed by size-exclusion HPLC. Based on the SE-HPLC chromatograms, HSA-7a, HSA-15a, SAv-7a and SAv-15a were obtained in high purity.

IEF analyses were conducted to confirm the change in pI of the protein conjugates prepared from the starting proteins or protein conjugates. An IEF comparing the conjugates, HSA-7a and HSA-15a and their succinylated derivatives, sHSA-7a and sHSA-15a, is shown in Figure 4 (left panel). As expected, the protein bands corresponding to the modified HSA are more electronegative than those of the unmodified HSA, and the succinylated HSA conjugates have the lowest pI values. The IEF comparing 7a and 15a conjugates of sSAv and unmodified SAv is shown in Figure 4 (right panel). The differences in pI due to succinylation and conjugation of 7a and 15a followed the same trend for both proteins, but unconjugated HSA is more electronegative than SAv, which has a pI near neutral.

Figure 4.

(left panel) IEF of HSA and conjugates, HSA-7a, sHSA-7a, HSA-15a and sHSA-15a that show changes in pI when conjugating 7a or 15a, and after succinylation. In this IEF, Novex IEF 3–7 precast gels were used with Serva standards. (right panel) IEF of SAv and HSA as well as their succinylated conjugates, sSAv-7a, sSAv-15a, sHSA-7a and sHSA-15a. Novex IEF 3–10 precast gels were used with Serva standards.

3.3. Radioiodinations

The potential protein degradation metabolites 9a, 9b, 16a and 16b, as well as the protein conjugates, were radiolabeled with either ¹³¹I or ¹²⁵I for biodistribution studies. The radioiodinations of these unconjugated LuG derivatives were conducted in MeOH/HOAc solution using N-chlorosuccinimide as an oxidant. The reactions provided isolated radioiodination yields of 61–67%. Radioiodination of the SAv and HSA conjugates and their succinylated derivatives were conducted in 500 mM sodium phosphate buffer (pH 6.8) using chloramine-T (ChT) as the oxidant. The isolated radiochemical yields for both HSA and SAv conjugates and their succinylated derivatives were generally above 90%.

3.4. Biodistributions

An initial study was conducted to evaluate the tissue concentrations of [¹²⁵I]9a and [¹³¹I]16a in athymic mice bearing prostate cancer C4–2B tumor xenografts at 1 and 4 h pi. The data obtained are graphed in Figure 5. (See Table S1 in Supporting Information for a tabulation of %ID/g data.) In the study, [¹²⁵I]9a and [¹³¹I]16a were prepared and co-injected into the mice so that a direct comparison could be made. As expected, both reagents demonstrated rapid blood clearance and [¹²⁵I]9a, the compound without the LuG moiety had essentially no tumor uptake. In fact, the concentration of [¹²⁵I]9a in all the tissues except for the intestine (12%ID/g at 1h pi) is lower than 2% ID/g at 1 h pi. The compound containing the LuG moiety, [¹³¹I]16a had relatively low (< 5%ID/g) tumor concentrations at 1 and 4 h pi, but tumor concentrations are 2.3× and 4.7× higher than the blood concentration at 1 and 4 h respectively. However, the kidney concentration of [¹³¹I]16a was very high (e.g. ~94% ID/g) at 4 h pi. The highest spleen concentration of [¹³¹I]16a was 9%ID/g at 1 h pi.

Figure 5.

Graph showing concentrations of the potential protein degradation metabolites [¹²⁵I]9a (without LuG moiety) and [¹³¹I]16a (with LuG moiety) in tissues of athymic mice bearing prostate cancer C4–2B tumor xenografts at 1 and 4 h pi. Data is from 5 mice per time point. See Table S1 for a tabulation of %ID/g data.

A second biodistribution provided a comparison of tissue concentrations of [¹³¹I]sSAv-7a and [¹²⁵I]sSAv-15a, as shown in Figure 6. (See Table S2 for a tabulation of %ID/g data.) The conjugates were prepared for the study using the following steps: (1) SAv was reacted with 20 equivalents of 7a or 15a, (2) the conjugates obtained were purified using DG10 columns, (3) the purified conjugates were reacted with 100 equivalents of succinic anhydride, (4) the succinylated conjugates were purified using DG10 columns, (5) the purified succinylated conjugates were radioiodinated, (6) then repurified using NAP 10 columns. Tissues of athymic mice bearing C4–2B tumor xenografts were collected at 1, 4 and 24 h after the coinjection of [¹³¹I]sSAv-7a and [¹²⁵I]sSAv-15a. It can be noted that most tissue concentrations of the two radioiodinated sSAv conjugates were very similar, suggesting that the tissue localization was primarily directed by the sSAv, not the conjugated moieties. The radioiodinated sSAv conjugates with and without the LuG moiety had very similar tumor localization, indicating that very little specific tumor targeting occurred. The concentrations [¹²⁵I]sSAv-15a in the kidney and spleen were slightly higher than those of [¹³¹I]sSAv-7a, but [¹³¹I]sSAv-7a was slightly higher than [¹²⁵I]sSAv-15a in the liver.

Figure 6.

Graph showing concentrations of succinylated SAv conjugates, [¹³¹I]sSAv-7a (without LuG moiety) and [¹²⁵I]sSAv-15a (with LuG moiety) in tissues of athymic mice bearing prostate cancer C4–2B tumor xenografts at 1, 4 and 24 h pi. Data is from 5 mice per time point. See Table S2 for a tabulation of %ID/g data.

A third biodistribution study compared tissue concentrations of two HSA conjugates obtained by reaction with 10 and 40 equiv. of 15a with HSA. Higher quantities of 15a in the reactions led to higher numbers of conjugates per HSA molecule. The comparison was done in athymic mice bearing C4–2B xenografts. The %ID/g of [¹³¹I]HSA-15a (10 equiv.) and [¹²⁵I]HSA-15a (40 equiv.) in various tissues at 4 h pi are shown in Figure 7. (See Table S3 for a tabulation of %ID/g data.) The concentrations of both [¹³¹I]HSA-15a (10 equiv.) and [¹²⁵I]HSA-15a (40 equiv.) in the kidney are lower (18–25% ID/g) than non-conjugated 15a (~94% ID/g), but are still considered high. Higher quantities of 15a conjugated with HSA did not have much effect on kidney concentration, but it increased liver concentrations substantially. Higher quantities of 15a conjugated to HSA did not affect the %ID/g (~5% ID/g) in tumor. It is interesting though that the % ID/g for the HSA-15a conjugate was the same as that obtained for the non-conjugated potential metabolite 16a.

Figure 7.

Graph showing concentrations of HSA conjugates, [¹³¹I]HSA-15a (10 equiv.) and [¹²⁵I]HSA-15a (40 equiv.) in tissues of athymic mice bearing prostate cancer C4–2B tumor xenografts at 4 h pi. Data is from 5 mice. See Table S3 for a tabulation of %ID/g data.

In a fourth biodistribution study, a comparison of the tissue distributions of succinylated HSA conjugates [¹³¹I]sHSA-7a and [¹²⁵I]sHSA-15a was obtained. Similar to succinylation of SAv conjugates, the HSA was reacted with 20 equivalents of 7a or 15a, then reacted with 100 equivalents of succinic anhydride, followed by radioiodination. Two different quantities of both sHSA-7a and sHSA-15a (10 and 30 μg injected protein per mouse) were evaluated in athymic mice bearing C4–2B xenografts. Selected tissues were collected at 4 h pi and counted. The concentrations of radioiodinated succinylated HSA conjugates in tissues are shown in Figure 8. (See Table S4 for a tabulation of %ID/g data.) The results indicated that the differing quantities of the sHSA conjugates injected did not have a significant influence on the tissue concentrations. The conjugates that had LuG moieties attached, [¹²⁵I]sHSA-15a, had higher concentrations in kidneys and liver than [¹³¹I]sHSA-7a conjugates without that moiety. The results indicate that essentially no tumor targeting was obtained with the [¹²⁵I]sHSA-15a. This was unexpected and appeared to indicate that there was a problem with succinylation of the HSA conjugates.

Figure 8.

Graph showing concentrations of two different quantities; 5 μg (red and blue bars) or 15 μg (green and orange bars) of succinylated HSA conjugates, [¹³¹I]sHSA-7a and [¹²⁵I]sHSA-15a in tissues of athymic mice bearing prostate cancer C4–2B tumor xenografts at 4 h pi. Data is from 5 mice. See Table S4 for a tabulation of %ID/g data.

A fifth biodistribution study was conducted to determine if succinylation affected the tumor localization of HSA conjugates. A comparison of tissue concentration for radioiodinated non-succinylated HSA conjugate, [¹²⁵I]HSA-15a, and HSA conjugate that has been succinylated, [¹³¹I]sHSA-15a, in athymic mice bearing C4–2B xenografts at 4 h pi is shown in Figure 9. (See Table S5 for a tabulation of %ID/g data.) It is clear from the data that succinylation of HSA to provide [¹³¹I]sHSA-15a resulted in reducing kidney concentrations from 17%ID/g to 7%ID/g, but it also decreased tumor localization from 4%ID/g to <1% ID/g. In fact, the tumor uptake of [¹²⁵I]sHSA-15a was the same as that of [¹³¹I]sHSA-7a (without LuG moiety), indicating that sHSA was detrimental to PSMA binding. Importantly, the succinylation of HSA also increased the liver uptake by ~21%ID/g.

Figure 9.

Graph showing concentrations of succinylated and non-succinylated HSA conjugates, [¹²⁵I]HSA-15a and [¹³¹I]sHSA15a in tissues of athymic mice bearing prostate cancer C4–2B tumor xenografts at 4 h pi. Data is from 5 mice. See Table S5 for a tabulation of %ID/g data.

In a sixth biodistribution study the potential protein degradation metabolites [¹²⁵I]9b and [¹³¹I]16b containing a shorter dPEG₄° moiety (vs. dPEG₁₂° in 9a and 16a) were evaluated in athymic mice bearing C4–2B xenografts at 1, 4, and 24 h pi. A graph of the data obtained is shown in Figure 10. (See Table S6 for a tabulation of %ID/g data.) Interestingly, the blood concentration of the radioiodinated compound containing the LuG moiety was significantly higher at all time points, which was not the case for the potential metabolites, 9a and 16a, with a longer PEG moiety. The kidney concentrations were quite high but appeared to be similar to those obtained with the longer dPEG linker. The most striking difference between the compound containing a LuG moiety, [¹³¹I]16b, and the similar compound with a longer dPEG linker, [¹³¹I]16a comes from the fact that [¹³¹I]16b has ~4× higher tumor concentrations (14–15%ID/g) than obtained with of [¹³¹I]16a (~4 %ID/g), suggesting the length of the (PEG)_n linker may play an important role in PSMA binding.

Figure 10.

Graph showing concentrations of the potential protein degradation metabolites [¹²⁵I]9b (without LuG moiety) and [¹³¹I]16b (with LuG moiety) in tissues of athymic mice bearing prostate cancer C4–2B tumor xenografts at 1, 4 and 24 h pi. Data is from 5 mice per time point. See Table S6 for a tabulation of %ID/g data.

4. Discussion

It has been known for over two decades that the extracellular regions of PSMA antigen contain an aminopeptidase region that removes the terminal glutamate from folate and the neuropeptide N-acetylaspartylglutamate. Since PSMA expression is highly upregulated on prostate cancer cells [22] it was suggested that specific small molecule inhibitors might be prepared to inhibit folate hydrolysis [23]. Early inhibition studies were conducted with organophosphorous derivatives of glutamic acid [24, 25] and alkylphosphonamidates [26], but a simpler and perhaps more robust inhibitor was designed using a glutamate dipeptide coupled by a urea bond [27–29]. The dipeptide lysine-urea-glutamate (LuG) has been used in the development of radiopharmaceuticals for prostate cancer targeting of ^99mTc [30] and radiohalogens [19, 31] for imaging and therapy. An issue with using the urea-coupled glutamate dipeptides to target radionuclides to prostate cancer is that high kidney concentrations are obtained. More recently, there have been a number of research groups investigating urea-coupled glutamate dipeptide derivatives in an effort to obtain lower kidney concentrations when targeting the PSMA antigen on prostate cancer cells [18, 19, 32–37]. One of the derivatives referred to as “PSMA-617” [33], which is radiolabeled with ¹⁷⁷Lu or ²²⁵Ac, is currently undergoing clinical trials for treating patients with metastatic castration-resistant prostate cancer [38].

Our overall goal has been to develop a radiopharmaceutical that can effectively treat metastatic prostate cancer. We believe that targeting the alpha-emitting radionuclide, astatine (²¹¹At), to prostate cancer cells can provide an effective therapy for that disease if the appropriate carrier molecule can be found. We started our studies with monoclonal antibodies to the PSMA antigen, however, intact antibody pharmacokinetics did not match with the short half-life (7.21 h) of ²¹¹At. In other studies, we attempted to use smaller antibody fragments that have more favorable pharmacokinetics, but those studies were not successful due to inherently high kidney concentrations. We also conducted a preliminary investigation of a small molecule containing the lysineurea-glutamate (LuG) moiety. Importantly, it was found that the LuG derivative targeted prostate cancer xenografts as effectively as the intact anti-PSMA antibody being studied. Those study results were encouraging, but using that approach also resulted in very high kidney localization. As an alternative approach to circumvent the kidney localization, we have studied antibody-based pretargeting. Many attempts to develop ²¹¹At-labeled reagents for pretargeting approaches have been unsuccessful. Our pretargeting studies demonstrated that the protein tetramer streptavidin, when highly succinylated, is not readily filtered by the kidney resulting in dramatically decreasing its concentration in that organ [14]. This made us consider the prospect of conjugating LuG moieties to SAv, then succinylating that conjugate to keep it from being concentrated in the kidney. In the design, it was thought that a long water-soluble (PEG) linking arm would make the LuG moiety available to bind without steric encumbrance from the protein. Another important part of the conjugate design was inclusion of a moiety that provided stable attachment of ²¹¹At and other radiohalogens (a theranostic approach). Thus, we chose closo-decaborate(2-) as a radiohalogenation moiety since it has been shown to stably attach ²¹¹At to carrier molecules for in vivo use. At the initiation of this investigation, our hypothesis was that an appropriately designed protein-LuG conjugate could target ²¹¹At to the PSMA antigen without being filtered by the kidney.

A trifunctional compound, 15a, that contained a closo-decaborate(2-) moiety, a LuG moiety with a lengthy dPEG₁₂° linker, and an isothiocyanate moiety for conjugation with amines on proteins, was synthesized. Another bifunctional compound 7a, which had the same structure as 15a, except did not contain the LuG moiety, was also prepared. Compound 7a was used to evaluate the binding of the LuG moiety in the mice bearing tumor xenografts. An initial study was conducted to evaluate the tumor localization and biodistribution of potential metabolites, 9a and 16a, of proteins conjugated with 15a and 7a. Compounds 9a and 16a, which contain conjugated N-acetyl lysine moieties, were synthesized, radioiodinated and evaluated in vivo. It should be noted that some metabolites of protein conjugates have previously been shown to contain a N-acetyl lysine moiety [39]. The results from that animal study showed a large difference in tumor localization between 16a, which had a prostate cancer-targeting LuG moiety, and 9a, which did not contain a LuG moiety (Fig. 5). Importantly, the LuG containing compound 16a had very high kidney concentrations (e.g. 94%ID/g at 4 h pi) as expected, whereas 9a did not localize in the kidney. This result establishes that the kidney localization is due to the LuG moiety. However, the tumor concentration was disappointing, with only ~4% ID/g at 1 h pi. It was not apparent as to whether the low tumor localization was due to the nature of 16a, or due to the C4–2B tumor model itself.

Despite the low tumor concentration of 16a, the corresponding succinylated SAv conjugate, sSAv-15a was prepared, as was the corresponding sSAv conjugate without a LuG moiety, sSAv-7a. We had previously shown that succinylation of SAv restricts filtration by the kidney [14]. This decreased filtration by kidney is believed to occur because the size of the protein, which is only a little smaller than the size of the openings in the glomerulus, and the highly anionic nature of the protein when succinylated, ultimately restricting its movement through the anionic basement membrane [40, 41]. The sSAv-15a and sSAv-7a conjugates were radioiodinated and evaluated in the mouse C4–2B xenograft model. Unfortunately, while the kidney concentrations were quite low, elevated liver and spleen concentrations were noted. More importantly, low concentrations of the conjugates were present in the tumor xenografts and very small differences in tumor concentrations were obtained between the conjugates with the LuG and without the LuG moiety. This result indicated that there might be an issue with the LuG binding to the PSMA antigen for this conjugate.

SAv was used as a model protein because of our prior pretargeting studies, however, its highly immunogenic nature precludes its general use. We chose a second protein, human serum albumin for further studies. HSA holds the potential for use in humans since it is non-immunogenic and has a long circulation half-life (~19 d) [42]. HSA has been used as a drug delivery system in a number of other approaches, so we believed that its use as a carrier molecule might diminish kidney localization, and due to half-life might increase tumor uptake. Although in this investigation the PEG-linked LuG moiety was covalently bonded to HSA, an alternative method is incorporating an HSA binder that reversibly binds to HSA. Other investigators have tried to conjugate a para-substituted phenyl-butanoic acid binding motif linked to an N_α-acetylated D-lysine to various radiolabeled peptides and obtained encouraging results [43]. In this investigation, HSA conjugates containing 15a and 7a were prepared, radioiodinated and evaluated in the mouse C4–2B xenograft model. One of the questions that came up was whether the number of molecules conjugated with the protein made a difference in the biodistribution, so HSA-15a conjugates were prepared by reaction with 10 or 40 equivalents of 15a per HSA molecule. The tissue concentrations at 4 h post injection where essentially the same in most tissues, but quite different in liver (Fig. 7). It appears that higher loading of the LuG conjugate resulted in substantially higher liver concentrations (~22%ID/g vs. 9%ID/g) at the 4 h timepoint. The subsequent studies were conducted by reacting 20 equivalents of 15a in the conjugations. The kidney concentrations in the biodistribution study were high for both conjugates, so succinylation of HSA-15a and HSA-7a was carried out in an attempt to decrease kidney concentrations. Another question that arose was whether the quantity of protein conjugate affected its’ in vivo biodistribution. Thus, a biodistribution study was conducted in mice bearing C4–2B xenografts, where animals were injected with 10 or 30 μg succinylated HSA-15a and HSA-7a (Fig. 8). Succinylation of the HSA-15a and HSA-7a conjugates decreased the kidney concentrations at 4 h pi, but also dramatically decreased blood concentrations and increased the liver and spleen concentrations. The most striking outcome was the low tumor concentrations obtained. A question of whether the low tumor concentrations were caused by succinylation was addressed in a subsequent animal study. In that study, succinylated HSA-15a was compared with HSA-15a that was not succinylated. It was very clear that succinylation decreased the blood and kidney concentrations, but also led to a large decrease in tumor concentration (Fig. 9). The low concentration of HSA-15a in tumors, even when not succinylated, made us question whether the size of the PEG linker was too large. It is possible that its 3-dimensional structure caused steric encumbrance and/or might have led to other interactions occurring with the 12 ethylene oxide units.

A new trifunctional molecule, 15b, containing a closo-decaborate(2-) moiety for radiohalogenation, a phenyl isothiocyanate moiety for conjugation and a shorter dPEG₄ linking the LuG moiety to the rest of the molecule, was synthesized. As previously done, a similar molecule, 7b, which had the same structure but without the LuG moiety was also synthesized to allow assessment of the effect of the LuG moiety in tissue distribution and tumor localization. Those compounds were reacted with N-acetyl-lysine to prepare potential metabolites 16b and 9b. The small molecules 16b and 9b were radioiodinated and co-injected into athymic mice bearing C4–2B xenografts. The biodistribution results were similar to those for 16a and 9a, however, kidney and spleen were higher (Fig. 10). The most interesting result was that the tumor concentration was considerably higher, being almost 15% ID/g. These results suggest that the long PEG linker did interfere with the tumor localization. Because the results from the succinylated SAv conjugates and succinylated HSA conjugates had significant liver concentrations, no further conjugations with those proteins have been done.

5. Conclusions

It is clear from the results obtained that conjugation of LuG moieties with proteins that are not readily filtered by the kidneys can be an approach to decreasing kidney concentrations of LuG derivatives. However, conjugation of LuG moieties with the two proteins used in this study appear to trade high kidney concentrations for high liver concentrations. It is also apparent from the study that making the protein highly negative by succinylation dramatically decreases and even alleviates tumor localization. The reason for this is not readily apparent, however it is clear that this is not a reasonable approach for future studies. Similar studies may be warranted with other proteins that are not filtered by the kidney, but larger proteins might still have the problem of not being able to access tumor cells in vivo before the ²¹¹At decays. Alternate approaches where either a non-covalent interaction of the radiolabeled LuG compound with HSA is obtained [43] or modification of the small molecule 16b in a manner that decreases the kidney localization are warranted. Based on the fact that other investigators have been successful in decreasing kidney localization by making derivatives of small molecule LuG compounds, it appears that making derivatives of 16b is the most promising approach at this time. However, along with decreasing kidney localization, the issue of targeting agent size and nature may be important in decreasing or alleviating the serious side effect of targeting salivary glands, which results in xerostomia or “dry mouth” syndrome seen in some patients administered radiolabeled LuG agents. Thus, there may be a rationale to continue to design and evaluate protein or nanoparticle-bound LuG reagents [44] that include the closo-decaborate(2-) labeling moiety. Such reagents could be labeled with ¹²³I for SPECT imaging or ¹²⁴I for PET imaging, and can also be labeled with ²¹¹At for therapy of metastatic prostate cancer.

Supplementary Material

1

Tables S1 - S6 of %ID/g from biodistributions studies

Figure 3.

SE-HPLC chromatograms showing changes in SAv and HSA before and after reaction with 20 equivalents of 7a and 15a. Note that the major peak broadens and elutes earlier (SAv-7a; 9.8 min & SAv-15a; 10.0 min) from that of the starting SAv (10.5 min). This observation does not hold with HSA (8.9 min), where HSA-7a (8.9 min) has the same retention time and HSA-15a elutes later (9.4 min).