Production of Multimeric Prostate-Specific Membrane Antigen Small Molecule Radiotracers using a Solid-Phase ^99mTc Pre-Loading Strategy

Abstract

Small molecule ligands specific for prostate-specific membrane antigen (PSMA) have the potential to improve prostate cancer imaging. However, highly charged ligands are difficult to label with ^99mTc and to purify. In this study, we present an adamantane-trimerized small molecule that has nanomolar binding to PSMA, but that also has 12 negative charges.

Methods

To convert this molecule into a clinically viable SPECT diagnostic, we have developed a simple, cartridge-based, solid-phase pre-labeling strategy that, within 25 minutes, converts readily available and inexpensive ^99mTc-pertechnetate into a chemically pure complex, with a reactive N-hydroxysuccinimide (NHS) ester, in neat organic solvent. This stable intermediate can label any amine-containing small molecule or peptide with ^99mTc in one step, with high specific activity, and without the need for HPLC.

Results

Solid-phase conversion of ^99mTc-pertechnetate to ^99mTc-MAS₃-NHS could be completed in 25 minutes, with >99% radiochemical purity and with no co-ligands present. This intermediate was then conjugated to adamantane-trimerized GPI in one step with >95% yield, and no need for HPLC purification. The final molecule bound specifically to living human tumor cells expressing PSMA on their surface. Quantitative comparison was made among GPI monomer, GPI trimer, and their ^99mTc derivatives.

Conclusion

Our study describes a simple cartridge-based conversion of ^99mTc pertechnetate to a useful, pre-loaded NHS ester intermediate that takes only 25 minutes to prepare and results in >99% radiochemical purity. Using this chemistry, we produce a high specific activity, ^99mTc-labeled, PSMA-targeted small molecule, and demonstrate gamma ray radioscintigraphic imaging of living human prostate cancer cells.

INTRODUCTION

Prostate cancer is diagnosed in approximately 230,000 men each year in the United States, and more than 30,000 of these will die of the disease (1). Every patient with biopsy-confirmed cancer requires clinical and radiographic staging to determine whether cancer has spread outside the capsule or has metastasized to regional lymph nodes or to bone. For patients already diagnosed with metastases, nuclear imaging could be used to pinpoint sites of disease and to quantify response to therapy.

Single photon emission computed tomography (SPECT) is a nuclear medicine technique by which gamma ray emissions from a radioisotope are used to localize cancer and its products. The major radioisotope used clinically is technetium-99m (^99mTc), which emits a monoenergetic, 140 keV photon, has a 6-hour half-life, and has a low absorbed dose. Most importantly, it is readily available at most institutions, and is extremely inexpensive to produce, typically $10.00 for 740 MBq (20 mCi). The major problem with ^99mTc is that it is eluted using saline as sodium pertechnetate (Na^99mTcO₄) in the +VII oxidation state, which requires reduction to the +V oxidation state, as well as exchange chelation, for stability and subsequent chemical manipulation. Any chemistry capable of converting sodium pertechnetate to a more useful form would have immediate impact on diagnostic agent development.

Given its half-life, ^99mTc is an ideal radiotracer for any targeting molecule having a total body clearance ≤6 hours, such as small molecules, peptides, and single-chain antibodies. Of these, small molecules have distinct advantages for tumor targeting due to rapid biodistribution, rapid clearance, and excellent penetration into solid masses. Our group (2) and others (3,4) have reported small molecules with high affinity binding to prostate-specific membrane antigen (PSMA), a type II transmembrane glycoprotein expressed at high levels on normal and malignant prostate epithelial cells (5,6). Our GPI molecule (see Methods) is a 311 Da monomeric small molecule with an affinity of 9 nM and a net charge of −4 after conjugation to contrast agents or radiotracers.

However, two major problems impede the use of GPI and similar molecules in diagnostic imaging. First, in its monomeric form, it is competed for binding to PSMA by the phosphate anion in serum. Second, its chemical structure resembles classic exchange ligands, such as tartaric acid, which are required to stabilize ^99mTc during its reduction. Hence, ^99mTc could be chelated by GPI itself and block binding to PSMA, and HPLC purification of the final product using conventional labeling strategies is extremely difficult (7). In this study, we solve both problems by developing a simple and rapid cartridge-based, solid-phase pre-labeling strategy for ^99mTc, and also describe an adamantane-based trimeric form of GPI that retains full binding to living prostate cancer cells in the presence of serum.

MATERIALS AND METHODS

Reagents

Guilford (now MGI Pharma, Baltimore, MD) compound 11245−36 (GPI; 2[((3-amino-3-carboxypropyl)(hydroxy)(phosphinyl)-methyl]pentane-1,5-dioic acid) was synthesized as described previously (8). Ultradry DMSO was purchased from Acros Organics (Geel, Belgium). HPLC grade triethylammnonium acetate, pH 7 (TEAA) was from Glen Research (Sterling, VA). HPLC grade water was from American Bioanalytic (Natick, MA). Triserine was from Bachem (King of Prussia, PA). N-succinimidyl-S-acetylthioacetate (SATA) was from Pierce (Rockford, IL). N,N,N′,N′-Tetramethyl-O-(N-succinimidyl)uronium tetrafluoroborate (TSTU) was from Advanced Chemtech (Louisville, KY). All other chemicals were purchased from Fisher Scientific (Hanover Park, IL) and were ACS or HPLC grade.

HPLC/Mass Spectrometry Platform

The HPLC/mass spectrometry platform used for purification of both non-radioactive and radioactive tumor-targeting small molecules and peptides has been described in detail previously (7). Briefly, the system is composed of a Waters (Milford, MA) model 1525 binary pump, model 2487 UV detector (Waters), Sedex model 75 (Richards Scientific, Novato, CA) evaporative light scatter detector (ELSD) with the nebulizer modified to reduce band broadening at low flow rates, a model FC-3200 high-sensitivity PMT gamma detector (Bioscan, Washington, DC), and a Waters fraction collector, all housed within a Capintec (Ramsey, NJ) hot cell equipped with a model CRC-15R (Capintec) dose calibrator. For non-radioactive reactions, column eluate was split into a Waters LCT electrospray time-of-flight (ES-TOF) mass spectrometer.

Preparative Synthesis of S-Acetylmercaptoacetyltriserine (MAS₃)

10 mg (36 μmol) of triserine were dissolved in 350 μL of water. 1 equivalent (5 μL, 36 μmol) of the base triethylamine (Et₃N) was added, followed by 2 equivalents (16.6 mg, 72 μmol dissolved in 160 μL DMF) of SATA. The reaction mixture was vortexed at room temperature for 3 hours. An additional equivalent (8.3 mg, 36 μmol dissolved in 80 μL DMF) of SATA was then added and vortexing was continued for an additional 2 hours.

To confirm completion of the reaction, a 10 μL sample was analyzed by reverse phase HPLC (RP-HPLC) using a 4.6 × 150 mm Symmetry (Waters) C₁₈ column and a linear gradient from 0% to 15% B over 35 minutes, starting 2 minutes after injection, at 1 mL/min, where A = H₂O + 0.1% formic acid and B = acetonitrile + 0.1% formic acid. MAS₃ eluted at R_t = 11.6 minutes as detected by the ELSD, with its mass confirmed by ES-TOF mass spectrometry. Preparative purification was performed on an HPLC system described in detail previously (7) and equipped with a 5 mL sample loop, after dilution into a final volume of 5 mL of H₂O + 0.1% TFA. The column was a 19 × 150 mm Symmetry C₁₈ column. The gradient consisted of 0% B for 3.5 minutes, then 0% to 15% B over 35 minutes at 7 mL/min, where A = H₂O + 0.1% TFA and B = acetonitrile + 0.1% TFA. MAS₃ eluted at R_t = 21.8 minutes using ELSD detection. Fractions containing product were pooled and lyophilized. MAS₃ was obtained as a white powder in 57% isolated yield (8.0 mg, 20.5 μmol), with expected mass confirmed by ES-TOF mass spectrometry, and purity ≥ 98%.

Solid-Phase Labeling of MAS₃ with ^99mTc-Pertechnetate

150 μL of a 50% slurry of Chelex 100 resin (Bio-Rad, Hercules, CA) in 50 mM 2-(N-morpholino)ethanesulfonic acid (MES) buffer, pH 5.0 were added to an empty micro Bio-Spin (Bio-Rad) chromatography column, washed once with MES buffer, and centrifuged at 3000 g for 10 seconds. 1.2 mg (3 μmol) of MAS₃ were dissolved in 1 mL of water. 1 mg (4 μmol) of stannous (II) chloride dihydrate was dissolved in 1 mL of 10 mM HCl. Then 100 μL of MAS₃ and 35 μL of stannous chloride were mixed well and added to the Chelex resin. 185−370 MBq (5−10 mCi) of ^99mTc-pertechnetate in 100 μL of saline, eluted directly from a ⁹⁹Mo generator with saline, were added to the tube. The tube was capped, wrapped in parafilm, and boiled for 10 minutes in a water bath. ^99mTc loading of MAS₃ was monitored by RP-HPLC using a 4.6 × 75 mm Symmetry C₁₈ column with a linear gradient from 0% to 60% B over 30 minutes, starting 2 minutes after injection, at 1 mL/min, where A = 10 mM TEAA and B = absolute MeOH. ^99mTc-MAS₃ eluted at R_t = 14.1 minutes. In separate experiments, the limits of this reaction were determined to be 370 MBq (10 mCi) ^99mTc-pertechnetate and 1.3 μmol of MAS₃ using 150 μL of 50% Chelex 100 slurry.

A 10 mg Oasis HLB cartridge (catalog #186000383) was activated using methanol, then H₂O, and equilibrated with H₂O, pH 4.0 (pH adjusted with 10 mM HCl). The ^99mTc-MAS₃ solution was diluted with 1 mL of H₂O, pH 4.0 and bound to the cartridge. The cartridge was washed with 10 mL of H₂O, pH 4.0 to remove ^99mTc-pertechnetate, tin, and free MAS₃, then it was purged completely with nitrogen. Finally, ^99mTc-MAS₃ was eluted with 400 μL dry DMSO or DMF.

Synthesis of the [^99mTc-MAS₃]-NHS

After 200 μg (0.7 μmol) of TSTU powder were placed in a 1.5 mL Eppendorf tube, 185−370 MBq (5−10 mCi) of purified ^99mTc-MAS₃ in 400 μL DMSO or DMF and 0.5 μL (2.9 μmol) of neat (5.74 M) N,N-Diisopropylethylamine (DIEA) were added. The top was sealed and the solution was incubated at 60°C for 10 minutes. The reaction was diluted to 1 mL final by the addition of 600 μL of dichloromethane:hexane (6:4). The reactants and byproducts were removed by passage through Waters Oasis MCX (catalog #186000252) and MAX (catalog #186000366) cartridges attached in tandem. The flow-through, containing pure [^99mTc-MAS₃]-NHS ester, was concentrated by loading on an Oasis HLB cartridge, washing with of 1 mL of dichloromethane:hexane (6:4), and eluting using 500 μL of either DMF or DMSO. The purity of the compound was assessed by RP-HPLC using a 4.6 × 75 mm Symmetry C₁₈ column with a linear gradient from 0% to 60% B over 30 minutes, starting 2 minutes after injection, at 1 mL/min, where A = 10 mM TEAA and B = absolute MeOH. [^99mTc-MAS₃]-NHS eluted at R_t = 23.5 minutes. The specific activity of [^99mTc-MAS₃]-NHS was estimated by conjugation to tryptophan and measurement of 280 nm absorbance of the product using the same RP-HPLC conditions as those used to assess purity.

Synthesis and Purification of Radiolabeled PSMA Ligands

Covalent conjugation of GPI derivatives with [^99mTc-MAS₃]-NHS was performed by the addition of 100 μL (10 μmol) of 100 mM triethylamine in dry DMSO to 10 μL (0.1 μmol) of a 10 mM solution of GPI derivative in a total of 400 μL dry DMF/DMSO, followed by the addition of 200 μL of [^99mTc-MAS₃]-NHS (185−259 MBq [5−7 mCi]) in dry DMSO. Constant stirring at room temperature was maintained for 40 to 90 minutes, until the reaction was completed. The radiolabeled ligands were analyzed by RP-HPLC on a 4.6 × 75 mm Symmetry C₁₈ column using a linear gradient from 0% to 60% B over 25 minutes, beginning 2 minutes after injection, at a flow rate of 1 mL/min, where A = 10 mM TEAA, pH 7.0 and B = absolute methanol. Retention times are shown in Figure 3B.

Figure 3. Preparation, Purification, and Analysis of PSMA-Specific Radiotracers.

A) The single nucleophiles (primary amines) of small molecules from (Figure 1) were conjugated in one step to [^99mTc-MAS₃]-NHS using the solid-phase pre-loading strategy described in Methods to create PSMA-specific radiotracers.

B) RP-HPLC analysis of compounds I-IV after solid-phase pre-labeling using either ^99mTc (left) or Re (right). The retention times for compounds Ia-IVa are shown, as are the ES-TOF(-) mass spectra of the peak for Re compounds (insets). For comparison, the retention times for [^99mTc-MAS₃]-NHS and [Re-MAS₃]-NHS on their optimized gradients are 23.5 min and 10.4 min, respectively (see Figures 2B and 2D).

C) Serum protein binding and serum stability of solid-phase pre-labeled ^99mTc PSMA-specific small molecules. Compound ^99mTc-IVa was incubated for 0 or 4 hours at 37°C, in PBS (top) or 100% serum (bottom), then subjected to HPLC analysis. Samples in PBS were resolved on a Symmetry C₁₈ column. Samples in serum were resolved on a 120-Å high-resolution gelfiltration column, and also include absorbance (280 nm) tracings. The retention times for the gel-filtration markers M₁-M₅ (see Methods) are shown as arrows. Marker retention times were M₁ = 6.6 min, M₂ = 8.2 min, M₃ = 9.2 min, M₄ = 11.1 min, and M₅ = 13.4 min.

D) Live cell binding assay. PSMA-positive LNCaP cells grown on 96-well filter plates as described in Methods were incubated for 20 min at 4°C with monomeric ^99mTc-Ia (top row) or trimeric ^99mTc-IVa (bottom row) in the presence of increasing concentrations of the homologous non-radioactive test compound. Shown are the results for monomeric Re-Ia (top row) and trimeric Re-IVa (bottom row) in TBS (left), PBS (middle) and 100% serum (right). Also shown are mean affinities ± 95% confidence intervals for n = 3 independent assays. N.D. = None detected.

Synthesis of Re-MAS₃

A procedure reported previously (9) was employed with slight modification. MAS₃ (4.8 mg, 16 μmol) was dissolved in 1.5 mL of water. Stannous chloride dihydrate (5.4 mg, 24 μmol) in 1.5 mL of 0.1 M citrate buffer, pH 5.0 and NaReO₄ (6.6 mg, 24 μmol) in 1.5 mL of H₂O were added to the MAS₃ solution. The reaction mixture was stirred at 90°C for 1 hour. After cooling to room temperature, Re-MAS₃ was purified on an Oasis HLB cartridge, and eluted with DMSO. The purity was assessed by LC-MS on a 4.6 × 150 mm Symmetry C₁₈ column using a linear gradient from 0% to 50% B over 15 minutes, beginning 2 minutes after injection, at a flow rate of 1 mL/min where A = H₂O + 0.1% formic acid and B = acetonitrile + 0.1% formic acid. Retention time was 6.4 minutes.

Synthesis of [Re-MAS₃]-NHS

50 μL (3 μmol) of a 60 mM TSTU solution in DMSO and 200 μL (2 μmol) of a 10 mM Re-MAS₃ solution in DMSO were mixed together, then 25 μL (5 μmol) of 200 mM N,N-diisopropylethylamine and 25 μL of DMSO were added. The solution was vortexed at room temperature for 40 minutes. After dilution with dichloromethane, the NHS ester was purified using Oasis MCX, MAX, and HLB cartridges connected in series as described above. The purity was assessed by LC-MS on a 4.6 × 150 mm Symmetry C₁₈ column using a linear gradient from 0% to 50% B over 15 minutes, beginning 2 minutes after injection, at a flow rate of 1 mL/min where A = H₂O + 0.1% formic acid and B = acetonitrile + 0.1% formic acid. Retention time was 10.4 minutes.

Synthesis and Purification of [Re-MAS₃]-Conjugates

Covalent conjugation of GPI derivatives with [Re-MAS₃]-NHS was performed by the addition of 0.1 mL (10 μmol) of 100 mM triethylamine in dry DMSO to 0.1 mL (1 μmol) of a 10 mM solution of GPI derivatives in dry DMSO or DMF, followed by addition of 0.2 mL (2 μmol) of a 10 mM [Re-MAS₃]-NHS solution in dry DMSO. Constant stirring at room temperature was maintained for 2 hours. The conjugates were purified by preparative HPLC (Symmetry C₁₈ column, 19 × 150 mm, 7 μm) using a linear gradient from 0% to 50% B over 27 minutes, starting 5 minutes after injection, with a flow rate of 10 mL/min. The collected fractions were lyophilized, and purity was assessed by LC-MS on a 4.6 × 150 mm Symmetry C₁₈ column using a linear gradient from 0% to 50% B over 15 minutes, beginning 2 minutes after injection, at a flow rate of 1 mL/min where A = H₂O + 0.1% formic acid and B = acetonitrile + 0.1% formic acid. Retention times are shown in Figure 3B.

Quantification of Serum Stability

The stabilities of [^99mTc-MAS₃]-GPI compounds were tested by incubation in the absence (PBS only) or presence of 100% calf serum for 4 hours at 37°C. Stability and transmetallation were quantified using high-resolution chromatography. For PBS experiments, a 4.6 × 75 Symmetry C₁₈ column was employed as described above for radiolabeled PSMA ligands. For serum experiments, an 8 × 300 mm, 200Å Diol (YMC, Catalog # DL20S053008WT) gel-filtration column was employed using PBS as mobile phase. Gel-filtration molecular weight markers (Bio-Rad) were: M₁= thyroglobulin (670 kDa), M₂ = γ-globulin (158 kDa), M₃= ovalbumin (44 kDa), M₄= myoglobin (17 kDa), and M₅ = vitamin B₁₂ (1.3 kDa).

High-Throughput, Radioactive Live Cell Binding and Affinity Assay

Human prostate cancer cell lines, PSMA-positive LNCaP and PSMA-negative PC-3, were obtained from the ATCC (Manassas, VA). The PSMA-negative human bladder cancer cell line TsuPR1 was a generous gift of Dr. John T. Isaacs (Johns Hopkins University, Baltimore, MD). Cell lines were cultured at 37°C under humidified 5% CO₂ in RPMI 1640 medium (Mediatech Cellgro, Herndon, VA) supplemented with 10% fetal bovine serum (Gemini Bio-Products, Woodland, CA) and 5% penicillin/streptomycin (Cambrex Bioscience, Walkersville, MD). Cells were split onto 96-well filter plates (model MSHAS4510, Millipore, Bedford, MA) and grown to 50% confluence (approximately 35,000 cells per well) over 48 hours.

To assign absolute affinity to each compound, a homologous competition assay was employed using the ^99mTc-labeled version as tracer and the Re-labeled version as test compound. To avoid internalization of the radioligand due to constitutive endocytosis (2), live cell binding was performed at 4°C. Cells were washed 2 times with ice-cold tris-buffered saline (TBS), pH 7.4 and incubated for 20 minutes at 4°C with 0.02 MBq (0.5 μCi) of radiotracer in the presence or absence of the test compound. Cells were then washed 3 times with TBS using a Millipore vacuum manifold (Catalog # MSVMHTS00) and the well contents was transferred directly to 12 × 75 mm plastic tubes placed in gamma counter racks. Transfer was accomplished using a modified (Microvideo Instruments, Avon, MA) 96-well puncher (Millipore MAMP09608) and disposable punch tips (Millipore MADP19650). Well contents were counted on a model 1470 Wallac Wizard (Perkin Elmer, Wellesley, MA) ten-detector gamma counter, and curves fit using Prism version 4.0a (GraphPad, San Diego, CA) software.

Near-Infrared Fluorescence and Gamma Radioscintigraphic Imaging

To assess viability and to verify confluence, living cells were loaded with the NIR fluorophore IR-786 by adding it to the cell culture medium at 1 μM for 30 minutes at 37°C prior to the start of the experiment (10). Cells were incubated as described above with 3.7 MBq (100 μCi) of ^99mTc-labeled compound per well, in the buffer being tested, for 20 minutes at 4°C, prior to extensive washing of the filter plate. Simultaneous white light and NIR fluorescence imaging of plates was performed as described in detail previously (11,12). Gamma radioscintigraphy was performed with an Isocam Technologies (Castana, IA) Research Digital Camera equipped with a 1/2″ NaI crystal, 86 photomultiplier tubes, and high-resolution low energy lead collimator.

RESULTS

High-Affinity, Multimeric Small Molecules for Targeting PSMA

Our group has previously developed a high affinity (9 nM), monomeric small molecule (311 Da) termed GPI, which is engineered to bind the active site of PSMA (2). Unfortunately, monomeric GPI is competed by inorganic anions in serum, such as phosphate, rendering it marginal for use as an in vivo diagnostic (2). The work of Whitesides and colleagues (13) suggests that by decreasing off-rate, multivalency can permit a small molecule to out-compete endogenous competitors. To test this hypothesis, while maintaining overall small size, we have developed a tri-NHS ester derivative of adamantane (14), which permits conjugation of up to three targeting ligands (such as GPI) in addition to a contrast agent or radiotracer. Recently, we have added a 6-carbon linker to improve radiotracer conjugation, and have purified a series of GPI derivatives with increasing valency (Figure 1). As shown in Table 1, conversion of monomeric GPI (I) to an adamantane trimer (IV) results in an affinity improvement of over one and one half logs, and retention of affinity in 100% serum.

Figure 1. Highly Anionic PSMA-Targeting Small Molecule Ligands.

Chemical structures and molecular weights for the PSMA-targeting ligands employed in this study.

TABLE 1.

Live Cell Binding Affinity of Small Molecules Targeted to Prostate-Specific Membrane Antigen

						Live Cell Binding Affinity (nM)a
Molecule	Description	Metal Conjugate	M.W.	Net Charge (pH 7.0)	Charge to Mass Ratio (X 10−3)	LNCaP Cells (TBS)	LNCaP Cells (PBS)	LNCaP Cells (Serum)	PC-3 Cells (TBS)	TsuPR1 Cells (TBS)
I	GPI	None	311.2	−3	−9.6	10.4 ± 0.3	N.D.	N.D.	N.D.	N.D.
Ia	GPI	Re-MAS3	844.7	−4	−4.7	35.5 ± 0.1	N.D.	N.D.	N.D.	N.D.
IIa	AdamGPI Monomer	Re-MAS3	1307.3	−6	−4.6	31.3 ± 0.4	N.D.	N.D.	N.D.	N.D.
IIIa	AdamGPI Dimer	Re-MAS3	1600.5	−9	−5.6	14.4 ± 1.4	20.3 ± 1.5	17.4 ± 1.0	N.D.	N.D.
IV	AdamGPI Trimer	None	1359.5	−11	−8.1	0.4 ± 0.1	0.5 ± 0.2	0.7 ± 0.2	N.D.	N.D.
IVa	AdamGPI Trimer	Re-MAS3	1893.5	−12	−6.3	3.3 ± 0.1	2.9 ± 0.1	2.1 ± 0.1	N.D.	N.D.

^aMean affinity ± 95% confidence interval from n = 3 independent experiments is shown.

N.D. = None detected.

Solid-Phase Pre-Labeling for the Preparation for ^99mTc-MAS₃ Conjugates

Unfortunately, highly anionic small molecules such as IV are difficult to use as ^99mTc diagnostic radiotracers because the 12 net negative charges after conjugation to a chelator can potentially act as co-ligands for ^99mTc. Furthermore, traditional pre- and post-labeling approaches for ^99mTc (Figure 2A) involve the use of excess quantities of an exchange ligand such as Na⁺/K⁺ tartrate. This creates two major problems. First, separation of ^99mTc-tartrate away from a desired radiolabeled chelator is extremely difficult and requires high-resolution HPLC (Figure 2B). For this reason, the radiolabeled tartrate peak is typically just accepted as “contaminant” and the entire labeling reaction used (even in clinical settings). Second, non-radioactive exchange ligand is used in excess, hence, all subsequent chemical steps, such as solution-phase pre-labeling (Figure 2A) are competed by the exchange ligand's carboxylic acids, unless great care, and time-consuming separation, is employed. Finally, if not purified away to completion, non-radioactive tartrate can form reactive NHS esters and conjugate covalently to the targeting ligand, again requiring careful purification, and risking competition with radiotracer.

Figure 2. ^99mTc Labeling Strategies, Solid-Phase Pre-Labeling, and Purification of ^99mTc/Re-Labeled MAS₃ and MAS₃–NHS.

A) General approaches for creating ^99mTc radiolabeled derivatives of small molecules and peptides include post-labeling (Route 1) in the presence of an excess exchange ligand such as tartrate, solution-phase pre-labeling using an excess of exchange ligand (Route 2), or solid-phase pre-labeling (Route 3) using Chelex 100 resin as described in this study.

B) C₁₈ HPLC radiochromatograms of conventional solution-phase pre-labeling of ^99mTc-MAS₃ with excess tartrate (top left) and solid-phase pre-labeling with Chelex 100 (bottom left). Solid-phase formation and purification of [^99mTc-MAS₃]-NHS (top right) and hydrolysis of the NHS ester at pH 10 (bottom right). ^99mTc-MAS₃ and [^99mTc-MAS₃]-NHS elute at 14.1 min and 23.5 min, respectively. ^99mTc-tartrate elutes at 13.5 min.

C) A simple, cartridge-based labeling and purification protocol to produce [^99mTc-MAS₃]-NHS is shown. Total elapsed time (brackets) and the yield for each step are also shown.

D) C₁₈ HPLC ELSD tracings (top) and mass spectra of the major peak (bottom) for Re-MAS₃ (left) and its NHS ester (right). The expected isotopic patterns for these Re derivatives are shown as insets.

To solve the problems associated with traditional radiolabeling, we have developed a simple and rapid, cartridge-based method for converting readily available and inexpensive ^99mTc-pertechnetate (Na^99mTcO₄) into a chemically pure, reactive NHS ester in organic solvent (Figure 2C). The first key to this strategy is the elimination of soluble exchange ligand in favor of a solid-phase source of carboxylic acids (Chelex 100 resin). As described in detail in Methods, only 8 minutes is required to fully-load (99% radiochemical yield) a ^99mTc chelator, such as MAS₃ (9), with ^99mTc (Figure 2C). MAS₃ was chosen over MAG₃ and other ^99mTc chelates given its low background binding in primates after conjugation to targeting molecules (15).

The second key to this strategy is the transfer of ^99mTc-MAS₃ from aqueous buffer to non-aqueous buffer, since the final NHS ester is highly susceptible to attack by the hydroxyl ion of H₂O. To accomplish this, and to remove unwanted molecules, the supernatant from the Chelex 100 resin is bound to a C₁₈ mixed-bed cartridge, washed, then eluted with desired organic solvent (DMSO or DMF). The NHS ester of ^99mTc-MAS₃ is then formed rapidly using TSTU and DIEA as described in Methods. Excess reactants are removed using tandem mixed-mode anion exchange and cation exchange cartridges, and the final product concentrated on a C₁₈ mixed bed cartridge (Figure 2C). Using this chemical strategy [^99mTc-MAS₃]-NHS, is prepared in ultrapure form (>99%; Figure 2B) in DMSO or DMF, in only 25 minutes, with an overall radiochemical yield of 70−75% relative to ^99mTc-pertechnetate starting material. Reaction of the final product with basic water (pH 10) results in the rapid re-formation of ^99mTc-MAS₃, thus proving the presence of an NHS ester (Figure 2B). The typical specific activity of [^99mTc-MAS₃]-NHS produced using this strategy was 4.1 × 10⁸ MBq/mmol [1.1 × 10⁴ Ci/mmol].

To confirm, absolutely, the chemical structures produced, as well as to prepare non-radioactive derivatives of diagnostic agents for affinity measurements, the chemistry was repeated using Re. Although the exact procedure shown in Figure 2C worked well for Re-MAS₃ chelates, the yield was low (≈20%), likely reflecting the difficulty others have had in preparing Re chelates (8). Hence, for preparative purification of Re-MAS₃, solution-phase labeling was used followed by formation and purification of [Re-MAS₃]-NHS using our solid-phase approach (see Methods). A high-resolution LC/MS HPLC system (7) was also used to verify that all cartridges performed as designed, and to verify the purity of final compounds (Figure 2D).

Solid-Phase Pre-Labeling of Highly Anionic PSMA Small Molecules

Using this solid-phase pre-loading strategy, GPI-containing molecules of increasing ligand valency were labeled with either ^99mTc or Re in one chemical step as shown in Figure 3A. For ^99mTc-labeled molecules, no further purification was necessary, since the NHS ester reaction with an amine in dry DMSO or DMF goes to completion without side reactions (Figure 3B). For Re labeled molecules, which were to be used for affinity measurements, LC/MS analysis with evaporative light scatter detection (7) was used to ensure the highest possible purity and to confirm the expected isotopic pattern (Figure 3B).

The stability of adamantane trimerized GPI molecule (IVa) was tested by incubation at 37°C for 4 hours in either PBS or 100% serum. HPLC analysis (Figure 3C) revealed 100% stability in PBS and 99% stability in serum, with 1% appearing as an early-eluting peak.

Bioactivity and Radioscintigraphic Imaging of PSMA-Specific Radiotracers

Using the non-radioactive Re derivatives as test compounds, ^99mTc derivatives as radiotracers, and a high-throughput 96-well filter plate assay ((7); see also Methods), the absolute affinity of each molecule for the surface of living prostate cancer cells was measured using homologous competition (Table 1). Actual raw data from the assay using monomeric [^99mTc-MAS₃]-GPI (Ia) as radiotracer (Figure 3D, top) and trimeric [^99mTc-MAS₃]-AdamGPI (IVa) as radiotracer (Figure 3D, bottom) confirms that both phosphate and serum compete effectively with monomeric radiotracers for the active site of PSMA, but are unable to compete with trimeric radiotracers, resulting in nearly identical affinities of IVa under all physiological conditions. The measured B_Max for compounds Ia/IIa, IIIa, and IVa was 1.6 × 10⁵, 2.1 × 10⁵, and 2.6 × 10⁵ PSMA receptors per cell, respectively, which is consistent with previously published reports of 1.8 × 10⁵ to 8.0 × 10⁵ for PSMA-specific monoclonal antibodies (16,17).

To confirm that the synthesized radiotracers permit sensitive and specific radioscintigraphic imaging of prostate cancer cells, approximately 35,000 PSMA-positive (LNCaP) or PSMA-negative (PC-3 and TsuPR1) cancer cells were grown in each well of a 96-well filter plate and incubated at 4°C for 20 minutes with the indicated tracer. As shown in Figure 4, and confirming the results in Table 1 and Figure 3D, ^99mTc-labeled trimeric AdamGPI (IVa) bound with high affinity to living prostate cancer cells expressing PSMA in all buffers including 100% serum, and no binding was detectable on PSMA-negative cells.

Figure 4. Radioscintigraphic Imaging of ^99mTc-Labeled PSMA-Specific Radiotracers.

Live cell binding of ^99mTc-Ia (GPI monomer; M) and ^99mTc-IVa (AdamGPI Trimer, T) was performed in TBS (top row), PBS (middle row), and 100% serum (bottom row) for 20 min at 4°C using PSMA-positive LNCaP cells (left), PSMA-negative PC-3 cells (middle), and PSMA-negative TsuPR1 cells (right) grown on 96-well filter plates, followed by extensive washing. Cells were independently loaded with NIR fluorophore IR-786 to assess viability and confluence. Shown are white light (top), NIR fluorescence (middle), and gamma detector (bottom) images of cells grown on 96-well filter plates.

DISCUSSION

Patients with prostate and other cancers are in desperate need of targeted diagnostic and therapeutic agents. One common strategy for the development of such agents is to utilize a small molecule or peptide targeting ligand, conjugated to a desired functional molecule (e.g., contrast agent, radiotracer, or therapeutic) via an isolating linker (7). This robust strategy requires conjugation chemistry that is simple and rapid. Although other pre- and post-labeling strategies for preparing ^99mTc SPECT radiotracers have been described (8,18), none has the combined features of taking less than one hour with inexpensive disposable cartridges, eliminating exchange ligand contaminants, producing ultrapure reactive NHS esters in organic solvent, producing high specific activity radiotracers, producing a stable amide bond, and eliminating the need for HPLC purification of final labeled agents. Eliminating the need for any HPLC purification steps makes the strategy amenable to high-throughput small molecule and peptide screening, and will also hopefully enable SPECT imaging, and especially pre-clinical SPECT imaging, by investigators who would otherwise not have access to the expensive equipment and chemistry expertise needed with other approaches. Another key feature is that it begins with ^99mTc-pertechnetate, an inexpensive reagent available at virtually every medical center. Finally, our solid-phase strategy is inherently automatable.

An important general limitation of NHS ester chemistry is that the hydroxyl ion of water is a potent competitor. NHS esters of fluorophores are typically used in mM concentrations and in excess of the targeting molecule. This is not true of radioactive NHS esters, such as [^99mTc-MAS₃]-NHS, which are used in nM concentrations. Hence, the use of [^99mTc-MAS₃]-NHS in aqueous buffer will result in low labeling efficiencies. When used in dry organic solvent, however, [^99mTc-MAS₃]-NHS reacts rapidly and completely with primary amines. Another potential limitation is the availability of the nucleophile within the targeting ligand. Each of our GPI derivatives has been engineered to have a flexible 6-carbon spacer, however, more sterically hindered amines could be difficult to conjugate.

Even with careful chemical engineering, small molecule and peptide targeting ligands often have poor affinity or are competed by endogenous ligands. A general approach to solving these problems was proposed by Whitesides' group (13), and involves multimerizing binding epitopes into an optimally-spaced rigid structure. To accomplish this, we have created a tri-NHS adamantane derivative (14), which permits the conjugation of up to three (same or different) targeting ligands. The molecule also has an isolating linker and a deprotectable primary amine for subsequent conjugation to contrast agents, radiotracers, or therapeutics. Adamantane, being the core structure in diamond, has maximal rigidity. In this study, we conjugated GPI through a propionyl acid spacer and made no attempt to optimize intermolecular spacing or overall rigidity. Future studies will focus on these issues.

Nonetheless, the results to date are encouraging. First, conjugation of a single GPI monomer to a single adamantane core had only negligible effect on affinity (Table 1; compare Ia to IIa). Second, increasing multivalency from 1 to 3 led to dramatic improvements in both affinity and competition with endogenous anions. Indeed, affinity not only increased almost two logs when measured in TBS, but as a trimer, was unaffected by serum (Table 1). The results also reveal, however, that the GPI molecules are not as isolated from the effect of metal chelate as one would like. For example, when Re-MAS₃ is conjugated to either GPI monomer or AdamGPI Trimer, there is a 3- to 8-fold decrease in affinity. Nevertheless, the final affinity of IVa (3 nM) is equivalent to previously described PSMA monoclonal antibodies (16).

Since it is known that PSMA undergoes both constitutive and inducible endocytosis (19), we were careful to perform all affinity measurements on living cells at 4°C. However, both NIR fluorescent GPI derivatives (2), and ^99mTc-labeled derivatives (data not shown) are concentrated inside cells via endocytosis, which serves to “amplify” radiotracer signal over time. This important feature of PSMA will likely improve in vivo performance, and also needs to be incorporated into pharmacokinetic modeling.

CONCLUSION

Trimeric GPI conjugated to ^99mTc-MAS₃ (IVa) may someday be a valuable diagnostic agent for clinical use. It has a 3 nM affinity for the surface of living prostate cancer cells in serum; 99% of it is unchanged after incubation in warm serum for 4 hours; and covalent conjugation to ^99mTc-MAS₃ is performed in one step without the need for subsequent purification. Ongoing experiments will characterize its biodistribution and clearance in vivo.