Microfluidic radiolabeling of biomolecules with PET radiometals

Abstract

Introduction

A robust, versatile and compact microreactor has been designed, fabricated and tested for the labeling of bifunctional chelate conjugated biomolecules (BFC-BM) with PET radiometals.

Methods

The developed microreactor was used to radiolabel a chelate, either 1,4,7,10-Tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) or 1,4,7-triazacyclononane-1,4,7-triacetic acid (NOTA) that had been conjugated to cyclo(Arg-Gly-Asp-DPhe-Lys) peptide, with both ⁶⁴Cu and ⁶⁸Ga respectively. The microreactor radiolabeling conditions were optimized by varying temperature, concentration and residence time.

Results

Direct comparisons between the microreactor approach and conventional methods showed improved labeling yields and increased reproducibility with the microreactor under identical labeling conditions, due to enhanced mass and heat transfer at the microscale. More importantly, over 90% radiolabeling yields (incorporation of radiometal) were achieved with a 1:1 stoichiometry of bifunctional chelate biomolecule conjugate (BFC-BM) to radiometal in the microreactor, which potentially obviates extensive chromatographic purification that is typically required to remove the large excess of unlabeled biomolecule in radioligands prepared using conventional methods. Moreover, higher yields for radiolabeling of DOTA-functionalized BSA protein (Bovine Serum Albumin) were observed with ⁶⁴Cu/⁶⁸Ga using the microreactor, which demonstrates the ability to label both small and large molecules.

Conclusions

A robust, reliable, compact microreactor capable of chelating radiometals with common chelates has been developed and validated. Based on our radiolabeling results, the reported microfluidic approach overall outperforms conventional radiosynthetic methods, and is a promising technology for the radiometal labeling of commonly utilized BFC-BM in aqueous solutions.

INTRODUCTION

Over the past decade, interest in the use of radiolabeled peptides for use in positron emission tomography (PET) imaging for preclinical development and clinical diagnostics, and radiation-based therapeutics has increased. Such agents are becoming more prevalent since many peptide-targeted receptors are overexpressed in cancer, and the development of routine methods of solid-phase peptide synthesis has allowed the development of highly specific imaging probes and/or therapeutic agents with the appropriate choice of PET radioisotope. These agents are generally composed of two parts, the biomolecule (BM) responsible for receptor targeting and a bifunctional chelate (BFC) for coordinating the radioisotope.

The cyclic pentapeptide cyclo(-Arg-Gly-Asp-(D)-Phe-Lys-) (cyclo(RGDfK)) is a highly potent and selective inhibitor for the α_vβ₃ receptor, an integrin required for tumor-induced angiogenesis, and is over-expressed in various types of tumors. Many radioligands containing the peptide sequence of cyclo(RGDfK) have been prepared to monitor expression of the α_vβ₃ receptor, and their evaluations have been carried out in both murine tumour models and patient studies, with promising results. The preparation of these radioligands requires that cyclo(RGDfK) peptides be labeled with positron-emitting radionuclide such as ¹⁸F, ⁶⁴Cu or ⁶⁸Ga. Among these PET radioisotopes, the distribution of ¹⁸F is limited to relatively small distances from the production cyclotron facility, due to the short half-life of ¹⁸F (t_1/2 =1.83 hr). On the other hand, ⁶⁴Cu and ⁶⁸Ga allow for wider distribution, either due to relatively long half life (t_1/2 =12.7 hr for ⁶⁴Cu) or availability of an on-site, self-contained generator (⁶⁸Ge/⁶⁸Ga generator for ⁶⁸Ga), which has drawn increased attention to ⁶⁴Cu and ⁶⁸Ga as promising radionuclides. ⁶⁴Cu can be effectively produced by both reactor-based and accelerator-based methods, and has suitable decay characteristics (β+: 0.653MeV, 17.4%; β−: 0.578 MeV; 39%) that allow this radiometal to be used as a PET imaging agent, and a nuclear therapy agent. ⁶⁸Ga is a positron emitter with a half-life of 67.6 minutes and decays via positron emission (89%), and its relatively shorter half-life and hydrophilic nature is beneficial for the rapid renal clearance of small peptides labeled with ⁶⁸Ga. In addition, ⁶⁸Ga can be conveniently and economically obtained from the ⁶⁸Ge/⁶⁸Ga generator, and the long half-life of the parent nuclide ⁶⁸Ge (t_1/2 = 280 days) allows PET imaging on-site without the need for a dedicated cyclotron. The ready availability of ⁶⁸Ga for clinical PET has prompted rapid tracer development based on this isotope, and more than 40 centers in Europe have utilized this type of generator, with most of them focusing on either basic or transitional research on radiochemistry of ⁶⁸Ga.

Although widely used conventional methods for labeling with radiometals suffer from several limitations. Due to the radiation emitted by PET radionuclides, the radiolabelings are typically carried out in a heavily lead-shielded fume hood or bunker to minimize radiation exposure to the radiochemists. In addition, radiosyntheses are usually carried out at very low mass amounts (usually less than a few hundred nanograms) of radioactive material. To facilitate convenient handling and efficient mixing, the tiny amounts of radioisotopes need to be dissolved in a relatively large amount of solvent, which result in nanomolar concentrations of the radiometals. To compensate for the slow reaction kinetics resulting from such low concentrations, a very large (~100-fold) excess of the non-radioactive precursor (biomolecule conjugate) is used to promote rapid and efficient incorporation of the radioisotope into the PET imaging agent. However, with such a large stoichiometric excess of precursor, issues such as radioisotope pre-concentration (prior to reaction), side reactions (due to possibly more reactive minor impurities), product purification and product analysis can be problematic. To achieve high specific activities using conventional labeling techniques, careful control over reaction conditions and extensive chromatographic purification is required to separate the radiolabeled compounds from the cold precursors. To address the above issues associated with conventional radiosynthesis, we have developed compact microreactors for rapid, efficient labeling without the need for using large excess reagents.

Microfluidic devices, comprising enclosed micro-channels (normally 10–500 μm wide or tall), mixing units, heaters, pumping systems, are able to control and process chemical or biological reactions in a continuous flow manner or batch mode. When used for radiolabeling biomolecules, microreactors can potentially circumvent many of the existing limitations. Microfluidic radiolabeling offers: (1) the ability to manipulate small volumes, which mitigates issues associated with dilution effects; (2) efficient mixing which greatly improves reaction kinetics; (3) the ability for fine level of control over reaction conditions, such as concentrations, temperature, enabling reliable and reproducible labeling; and (4) a much smaller overall footprint of the system, which drastically reduces the volume that requires shielding.

Most microfluidic approaches for the synthesis of radiopharmaceuticals have focused primarily on the synthesis of ¹⁸F- and ¹¹C-labeled agents. For example, Lee et al. developed a poly(dimethylsiloxane) (PDMS)-based microreactor for the multistep synthesis of ¹⁸F-FDG, a radiotracer commonly used to detect cancer via PET. An improved version of this microreactor provided yields of 96% after ~15 minutes of total synthesis time, compared to 75% yield after ~45 minutes using an automated apparatus for conventional synthesis. However, an off-chip purification step was still required to obtain a radiochemical purity of ~99%, and loss of ¹⁸F through reaction with or absorption by the PDMS was noted as a significant issue. In other work, Lu et al. developed a hydrodynamically-driven microreactor and investigated the effect of infusion rate on the yields of several ¹⁸F and/or ¹¹C-labeled carboxylic esters using a T-junction, flow-through glass microfluidic system. They observed that incomplete diffusive mixing at lower residence times adversely affected the yield. Recently, our group has developed PDMS-based microreactors for labeling BFC-BM with radiometals. These microreactors were validated with radiosynthesis of ⁶⁴Cu labeled RGD peptide, and demonstrated higher yields compared to conventional techniques under identical reaction conditions. Following up on the validation with ⁶⁴Cu labeling, we report here the versatility of this microreactor-based method by utilizing two commonly used PET radiometals, ⁶⁴Cu and ⁶⁸Ga, to label two common bifunctional chelates, DOTA and NOTA (more specifically S-2-(4-Isothiocyanatobenzyl)-NOTA (p-SCN-Bn-NOTA)), conjugated to a clinically relevant biomolecule, the pentapeptide cyclo(RGDfK). We present a direct comparison between radiolabeling performance of the microreactor and conventional methods, and results on optimization of the labeling conditions using the microreactor.

EXPERIMENTAL PROCEDURES

General

A 50 mCi ⁶⁸Ge/⁶⁸Ga generator was obtained from Eckert & Ziegler EUROTOPE GmbH. A Hewlett Packard model 1050 high-performance liquid chromatography (HPLC) apparatus was used for purifying synthesized and labeled compounds. The luna HPLC column and Strata X-C column were purchased from Phenomenex. Mass spectra were obtained with an Waters LC-MS API-3000 spectrometer. Three sets of female 1/4-28 to female Luer lock adaptors with ferrules and nuts, PEEK tubing of 1/16″ OD and 0.01″ ID, and a NanoTight kit with 1/16″ × 0.027″ FEP sleeves, obtained from Upchurch Scientific (Oak Harbor, WA) and Microbore PTFE tubing (0.012″ ID × 0.030″ OD) obtained from Cole-Parmer (Vernon Hills, IL) were assembled to connect syringes to the microreactor. The 75 mm × 50 mm × 1 mm glass slide, used to assemble the microreactor, and the Fisher Vortex Genie 2 were obtained from Fisher Scientific (Pittsburgh, PA). Test grade, 3″ silicon wafers, obtained from University Wafer (South Boston, MA) were used to fabricate SU-8 molds for the microreactor. Three microliter flow modular pump components, which included a syringe pump, a pump driver circuit, and a power supply, were obtained from Harvard Apparatus (Holliston, MA). The Kapton-insulated, 2″ × 2″, thin film heater, the Omega CN740 temperature controller, and an Omega SA 1-RTD probe were obtained from Omega Engineering (Stamford, CT). The Harrick plasma cleaner was obtained from Harrick Plasma (Ithaca, NY). Eppendorf tubes were obtained from MIDSCI, Inc. (St. Louis, MO). The BioScan AR-2000 radio-TLC plate reader was purchased from Bioscan, Inc. (Washington, DC). The Thermomixer was obtained from Eppendorf North America (Hauppauge, NY). Millipore Centriplus Centrifugal Concentrator (2 mL, 3000 NMWL) was obtianed from Millipore (Billerica, MA). TLC-silica gel [SG] plates were obtained from Whatman Thin Layer Chromatography (Piscataway, NJ), and iTLC-silica gel [SG] was purchased from Gelman. Gas-tight, microliter syringes were obtained from Hamilton Co. (Reno, NV). The Capintec CRC-712M radioisotope dose calibrator used to measure activities for retention experiments was obtained from Capintec, Inc. (Ramsey, NJ).

All solvents and reagents were purchased from commercial sources and used without further purification. DOTA-NHS-ester and S-2-(4-Isothiocyanatobenzyl)-NOTA (p-SCN-Bn-NOTA) were purchased from Macrocyclics (Dallas, TX). Bovine serum albumin (BAS protein), 3-[4-(2-Hydroxyethyl)-1-piperazinyl]propanesulfonic acid (HEPPS), ammonia acetate, acetonitrile and trifluoroacetic acid (TFA) were purchased from Sigma-Aldrich (St. Louis, MO). Cyclo(RGDfK) was purchased from Peptides International, Inc. (Louisville, KY). De-ionized water (DI-H₂O) was produced using a Millipore Milli-Q water system. RTV 615 poly(dimethyl siloxane) (PDMS) was obtained from General Electric Company (Waterford, NY). SU-8 2050 was obtained from MicroChem Corporation (Newton, MA). (Tridecafluoro-1,1,2,2-tetrahydrooctyl) trichlorosilane was obtained from Gelest, Inc. (Morrisville, PA). ⁶⁴Cu²⁺ was produced at Washington University School of Medicine, and was obtained in 0.1M HCl solution. 6 mL of 0.1M HCl was used to flush ⁶⁸Ga³⁺ (from the ⁶⁸Ge/⁶⁸Ga generator) onto Strata X-C column, and then ~ 20 mCi of ⁶⁸Ga³⁺ was recovered by eluting with 1.0 mL of 0.02M HCl, 98% Acetone.

Fabrication and operation of the microreactor

The PDMS-based microreactor for labeling BFC–BMs with radiometals was fabricated as described previously, and is comprised of three major components (Fig. 1a): (1) a staggered herringbone mixing channel to passively mix reagents; (2) a series of incubation reservoirs for the radiometal–ligand mixture and (3) a thin-film heater for heating the mixture. The radiometal and buffer solutions are pumped into the microreactor through inlets 1 and 2, respectively, and then flowed through the mixing channel. As the two solutions flowed along the microchannel, they mix by chaotic advection induced by the geometry of the herringbones. After the buffer and radiometal mixture arrive at the end of the first mixing channel, the biomolecule conjugate (BFC-BM) solution is pumped into the microreactor through inlet 3 and mixed with the radiometal-buffer mixture in a second mixing microchannel with staggered herringbone grooves defined in the ceiling. The mixed reagents then fill a series of five hexagonal reservoirs (holding tanks), and flow is stopped. The mixture is heated to the desired temperature and allowed to react for a specified residence time which is two minutes less than the actual reaction time thereby accounting for the two minutes it takes for reactant introduction and product flushing. After the previously set residence time, buffer solution was pumped into the microreactor to flush the product out, through outlet 1. Figure 1b shows a diagram of the configuration used to operate the microreactor. The flow of each reagent stream; radiometal, buffer, and ligand, into the microreactor was driven by syringe pumps controlled by a LabVIEW virtual instrument (VI). Communication between the VI and the syringe pump drivers was facilitated through a serial connection. The VI was used to set the individual flow rates and regulate the timing of the pumps, with inputs for controlling the incubation time, the total flow rate, the desired volume of product (the maximum being the total volume of the reservoirs, ~50 μL), and the stoichiometric molar ratios of buffer: radiometal and ligand: radiometal. For radiolabeling at elevated temperatures, a thin-film heater, placed in contact with the glass surface of the microreactor, was used to heat the microreactor, and a temperature controller and resistance temperature detector (RTD) probe were used to maintain the temperature within ±1 °C of the desired temperature set-point. The radiometal syringe pump, heater, and microreactor were shielded on four sides and below with 2″ of lead shielding. The radiometal syringe was fitted with a 7.5 cm length of PEEK tubing connected to an Upchurch Luer lock fitting. The PDMS microreactor was fit with a 10 cm length of PTFE tubing connected to an Upchurch NanoTite fitting. The syringe and PEEK “needle” were inserted into the fitting and tightened.

Fig. 1.

(a) Optical micrograph of the microreactor and (b) schematic of the microreactor system for radiolabeling (1b). The mixing channels and incubation reservoirs are filled with blue dye. The length, width, and height of the PDMS portion of the microreactor are ~2.0″ × ~2.0″ × ~0.25″.

Synthesis and purification of NOTA-cyclo(RGDfK)

To a mixture of 6.03 mg cyclo(RGDfK) and 8.26 mg p-SCN-Bz-NOTA (1.2 equiv.) in 0.5 mL dimethylformamide (DMF), 1.0 mL of 0.1 M sodium bicarbonate was added, and then the mixture was allowed to react overnight at room temperature in the dark (Scheme 1a). The crude NOTA-cyclo(RGDfK) was then purified on a Waters HPLC system using an Alltech Econosil C18 semi-preparative column RP18 (10 μm, 4.6 × 250 mm) with a mobile phase of 10 vol% acetonitrile and 90 vol% de-ionized water with 0.1 vol% trifluoroacetic acid (TFA), at a flow rate of 3.0 mL/minute. Under these conditions, the NOTA-cyclo(RGDfK) was eluted at 11.5 minutes. The pooled NOTA-cyclo(RGDfK) fractions were lyophilised overnight, and white powder was obtained as a pure product. The identity and purity of the purified NOTA-cyclo(RGDfK) peptide was confirmed by LC-MS using an Phenomenex Luna C18(2) column (5 mm, 4.6 × 150 mm). The analysis was performed using the following gradient of 0.1 v% TFA in de-ionized water (A) and 0.1 vol% TFA in acetonitrile (B) with a flow rate of 0.5 mL/minute: 0–5 minutes: 90% A, 10% B; 5–20 minutes: 10% A, 90% B; 20–24 minutes: 10% A, 90% B; 25–30 minutes: 90% A, 10% B. A UV chromatogram and a TIC chromatogram are shown in Fig. 2a. The mass spectrum of the peak at ~14.6 minutes is shown in Fig. 2b, and verified that the peak corresponds to pure NOTA-cyclo(RGDfK), with m/z, [M+H]⁺ = 1054.7.

Scheme 1.

Synthesis of NOTA-cyclo(RGDfK) (a) and DOTA-cyclo(RGDfK) (b)

Fig. 2.

Purity of NOTA-cyclo(RGDfK), as indicated by (a) UV (solid line) and Mass (TIC, dashed line) traces of LC-MS chromatograms. (b) The mass spectrum of the material corresponding to the peak with an elution time of 14.6 minutes in (a).

Preparation of NOTA-cyclo(RGDfK) stock solution

Stock solutions of 1.0 mM NOTA-cyclo(RGDfK) were prepared by dissolving 5.28 mg of purified DOTA-cyclo(RGDfK) in 500 μL of a mixture of 10 mM NH₄OAc (pH = 4.50) and acetonitrile (MeCN) (80:20 v%), and then diluting these solutions with 10 mM NH₄OAc (pH = 4.50) to give final concentrations of 200, 100, 40, and 2 μM NOTA-cyclo(RGDfK).

Synthesis and purification of DOTA-cyclo(RGDfK)

A mixture of 6.03 mg cyclo(RGDfK), 8.38 mg DOTA-NHS (1.1 equiv.) and 14.0 μL triethylamine (TEA) (10 equiv.) in 1.0 mL DMF was allowed to react overnight at room temperature in the dark (Scheme 1b), then 3 mL of DI-H₂O were added and stirred for another 30 minutes to hydrolyze the excess DOTA-NHS ester. The crude DOTA-cyclo(RGDfK) was then purified on a Waters HPLC system using an Alltech Econosil C18 semi-preparative column RP18 (10 μm, 4.6 × 250 mm) with a mobile phase of 15 vol% acetonitrile and 85 vol% de-ionized water with 0.1 vol% TFA, at a flow rate of 3.0 mL/minute. Under these conditions, the NOTA-cyclo(RGDfK) eluted at 13.5 minutes. The pooled DOTA-cyclo(RGDfK) fractions were lyophilised overnight, and white powder was obtained as a pure product. The identity and purity of the purified DOTA-cyclo(RGDfK) peptide was confirmed by LC-MS using an Phenomenex Luna C18(2) column (5 mm, 4.6 × 150 mm), which is consistent with the previously reported data.

Preparation of DOTA-cyclo(RGDfK) stock solution

Stock solutions of 1.0 mM DOTA-cyclo(RGDfK) were prepared by dissolving 4.96 mg of purified DOTA-cyclo(RGDfK) in 500 μL of a mixture of 10 mM NH₄OAc (pH = 6.8) and acetonitrile (MeCN) (80:20 vol%), and then diluting these solutions with 10 mM NH₄OAc (pH = 6.8) to give final concentrations of 200, 100, 40, and 2 μM DOTA-cyclo(RGDfK).

Preparation of 200 μM DOTA conjugated BSA protein (DOTA-BSA)

To a solution of 2.68 mg BSA protein in 1.80 mL PBS buffer (10 mM, pH = ~ 8.50), 306 μg DOTA-NHS in 200 μL DMSO was added. After the pH was adjust to ~9.50 by addition of 1.0 M Na₃PO₄, the mixture was incubated for 2 hour at room temperature. Then, the mixture was concentrated to 200 μL with 5-kDa molecular weight cutoff ultrafiltration membrane centrifugal filter devices (Millipore). The residue was diluted with 1.80 mL NH₄OAc buffer (10 mM, pH = ~ 6.80), and then concentrated to 100 μL; this sequence was repeated an additional 5 times. The pure DOTA-BSA was obtained in 200 uL of NH₄OAc buffer (10 mM, pH = ~ 6.80), and then was diluted with 200 uL of NH₄OAc buffer (10 mM, pH = ~ 6.80) to give the final concentrations of 200 μM DOTA-BSA, which was stored in −20 °C freezer. The number of DOTA moieties on each BSA protein was determined to be 4 using a published method.

Preparation of no-carrier added ⁶⁸Ga³⁺ and carrier-added ⁶⁸Ga³⁺ stock solution

The 1.0 mL ⁶⁸Ga³⁺ (~20 mCi, in 0.02M HCl, 98% Acetone) was dried, and the residue was re-dissolved in 2.0 mL of 1.0 M HEPES bufffer (pH = 4.50) to give the no-carrier-added ⁶⁸Ga³⁺ stock solution with specific activity of 10 μCi/μL (decay corrected).

Various carrier-added Gallium solutions were prepared from this no-carrier-added stock solution. 200 μM carrier-added ⁶⁸Ga³⁺ solution was prepared by mixing 100 μL ⁶⁸Ga³⁺ solution (10 μCi/μL) with 20 μL of 10 mM non-radioactive Gallium solution and 880 μL of 1.0 M HEPES bufffer (pH = 4.50). The carrier-added ⁶⁸Ga³⁺ solutions with different concentrations were obtained by the dilution of the above 200μM carrier-added ⁶⁸Ga³⁺ solution with 1.0 M HEPES bufffer (pH = 4.50).

Preparation of no-carrier added ⁶⁴Cu²⁺ and carrier-added ⁶⁴Cu²⁺stock solution

The cyclotron-produced ⁶⁴Cu²⁺ (5 mCi, in 0.1M HCl) was diluted with 10 mM NH₄OAc (pH = 6.50) to obtain the no-carrier added ⁶⁴Cu²⁺ stock solution with specific activity of 10 μCi μL⁻¹. Various carrier-added copper solutions with different concentrations were then prepared from this no-carrier-added stock solution. 200 μM carrier-added ⁶⁴Cu²⁺ solution was prepared by mixing 100 μL ⁶⁴Cu²⁺ solution (10 μCi μL⁻¹) with 20 μL of 10 mM non-radioactive copper solution and 880 μL of 10 mM NH₄OAc (pH = 6.50). The carrier-added ⁶⁴Cu²⁺ solutions with different concentrations were obtained by the dilution of the above 200μM carrier-added ⁶⁴Cu²⁺ solution with 10 mM NH₄OAc (pH = 6.50).

Measurement of retention of ⁶⁸Ga³⁺

Before use in retention measurements, the fabricated microreactors were cleaned with 1.0 mL of 1N nitric acid to remove trace amount of metals, and then flushed with 5 mL of 10 mM NH₄OAc buffer solution (pH = 4.50).

Pre-treatment with Ga³⁺/Na⁺

100 μL of 10 mM NaCl/GaCl₃ solution was injected into the microchannel and allowed to sit for 30 minutes. The microchannel was then flushed with 5 mL of 10 mM NH₄OAc buffer solution (pH = 4.50).

Retention measurement

30 μL of no-carrier added ⁶⁸Ga³⁺ in 10 mM NH₄OAc buffer solution (pH= 4.50) was injected into the microchannel. After 10 minutes, the ⁶⁸Ga³⁺ solution was displaced from the microchannel by 1.0 mL of air, via syringe. The microchannel was then flushed with 1.0 mL of 10 mM NH₄OAc buffer (pH= 4.50). The percent of injected activity retained in the microchannel was calculated from the known activity of the injected solution, and the difference between the activity left behind in the microchannel after flushing and the initial activity of the microchannel measured before the ⁶⁸Ga³⁺ solution was injected, all measured with the Capintec radioisotope dose calibrator.

Measurement of retention of ⁶⁴Cu²⁺

Before using in retention mesurements, the fabricated microreactors were cleaned with 1.0 mL of 1N nitric acid to remove trace metals, and then flushed with 5 mL of 10 mM NH₄OAc buffer solution (pH = 4.50).

Pre-treatment with Cu²⁺

100 μL of a 10 mM CuCl₂ solution was injected into the microchannel and allowed to sit for 30 minutes. The microchannel was then flushed with 5 mL of 10 mM NH₄OAc buffer solution (pH = 4.50).

Retention measurement

30 μL of no-carrier added (only radioactive) ⁶⁴Cu²⁺ in 1.0 M HEPES bufffer (pH = 4.50) was injected into the microchannel. After 10 minutes, the ⁶⁴Cu²⁺ solution was displaced from the microchannel by 1.0 mL of air, via syringe. The microchannel was then flushed with 1.0 mL of 10 mM NH₄OAc buffer (pH= 4.50). The percent of injected activity retained in the microchannel was calculated from the known activity of the injected solution, and the difference between the activity left behind in the microchannel after flushing and the initial activity of the microchannel measured before the ⁶⁴Cu²⁺ solution was injected, were all measured with the Capintec radioisotope dose calibrator.

Conventional radiolabeling of the DOTA-cyclo(RGDfK)/NOTA-cyclo(RGDfK) with ⁶⁴Cu²⁺

30 μL of 100 μM DOTA-cyclo(RGDfK)/NOTA-cyclo(RGDfK) solution was firstly added to 15 μL of 10 mM NH₄OAc (pH = 6.50) in a 1.6 mL Eppendorf tube. 60 μL of 50 μM ⁶⁴Cu²⁺ solution (in 10 mM NH₄OAc, pH = 6.50) was then added to the mixture. The mixture was vortexed for 10 seconds, and incubated in Eppendorf thermomixer for 12 minutes at a temperature of 23, 37 or 47°C (Scheme 2a and 2b). The final concentration of ⁶⁴Cu²⁺ and NOTA-cyclo(RGDfK)/DOTA-cyclo(RGDfK) was 28.6 μM in both methods.

Scheme 2.

Radiolabeling of DOTA-cyclo(RGDfK) with ^64Cu (a); NOTA-cyclo(RGDfK) with ^64Cu (b); DOTA-cyclo(RGDfK) with 68Ga (c); NOTA-cyclo(RGDfK) with 68Ga (d)

Conventional radiolabeling of DOTA-cyclo(RGDfK)/NOTA-cyclo(RGDfK) with ⁶⁸Ga³⁺

30 μL of 100 μM DOTA-cyclo(RGDfK)/NOTA-cyclo(RGDfK) solution was added to 15 μL of 10 mM NH₄OAc (pH = 4.50) in a 1.6 mL Eppendorf tube. 60 μL of 50 μM ⁶⁸Ga³⁺ solution (in 1.0 M HEPES buffer, pH = 4.50) was then added to the mixture. The mixture was vortexed for 10 seconds, and incubated in Eppendorf thermomixer for 12 minutes at a temperature of 23, 37 or 47°C (Scheme 2c and 2d). The final concentration of ⁶⁸Ga³⁺ and NOTA-cyclo(RGDfK)/DOTA-cyclo(RGDfK) was 28.6 μM in both methods. The labeling results were analyzed by radio-HPLC and/or iTLC(SG).

Conventional labeling of DOTA-BSA with ⁶⁴Cu²⁺/⁶⁸Ga³⁺

10 μL of 100 μM DOTA-BSA solution was mixed to 5 μL of 10 mM NH₄OAc (pH = 6.50 for ⁶⁴Cu, pH = 4.50 for ⁶⁸Ga) in a 1.6 mL Eppendorf tube. 50 picomole (10 μL of 5.0 μM ⁶⁸Ga, 50 μL of 1.0 μM ⁶⁴Cu) of the above no-carrier added radiometal stock solution (BSA to M = 20:1) was then added to the mixture. The mixture was vortexed for 10 seconds, and then incubated in Eppendorf thermomixer for 22 minutes at a temperature of 37 °C. The labeling results were analyzed by radio-FPLC and/or iTLC(SG). The concentration of the no-carrier added radiometal was caculated based on its E.S.A. as reported.

Microreactor radiolabeling procedure

The microreactor is operated as previously described in Fabrication and operation of the microreactor section. For the radiolabeling experiments summarized in Fig. 5a (effect of residence time on extent of reaction) and Fig. 5b (effect of mixing method and temperature on extent of reaction), the following solutions were used: (1) 50 μM carrier-added ⁶⁸Ga³⁺/⁶⁴Cu²⁺ solution, (2) 100 μM NOTA-cyclo(RGDfK)/DOTA-cyclo(RGDfK) stock solution, and (3) 10 mM NH₄OAc (pH = 6.8 for ⁶⁴Cu, and pH = 4.50 for ⁶⁸Ga) buffer. Using the LabVIEW interface, the pumps were programmed to combine the radiometal (1), ligand (2), and buffer (3) solutions with the same volumetric ratios as those used for the conventional radiolabeling procedure, giving a ligand-to-metal molar ratio of 1:1 and a final concentration of carried-added radiometal and BFC-BM of 28.6 μM. The total flow rate was set to 25 μL/minute, the total volume was set to 25 μL. An additional 25 μL of buffer was pumped into the microreactor to fill all the reservoirs. For the experiments in Fig. 6a, the temperature was set to 37 °C, and the residence time was set to 5, 10, and 20 minutes (resulting in total reaction times, t_REA = 7, 12, and 22 minutes: e.g., 1 minute for filling + 10 minutes for residence + 1 minute for flushing). For the experiments in Fig. 6b, the residence time was set to 10 minutes (t_REA = 12 minutes), and the temperature was set to 23, 37 and 47 °C. After the collection of product, the yield of radiolabeling raction was determined as described in the Measurement of radiolabeling yield section.

Fig. 5.

Effect of (a) residence time, and (b) temperature on the labeling of DOTA/NOTA-cyclo(RGDfK) with carrier-added ⁶⁴Cu²⁺/⁶⁸Ga³⁺.

Fig. 6.

Comparison of radiolabeling results between two approaches, microreactor and conventional method: (a) labeling DOTA-RGD with carrier-added ⁶⁸Ga³⁺ at different temperature (23, 37 and 47 °C) with a residence time of 10 minutes; (b) labeling NOTA/DOTA-RGD with carrier-added ⁶⁴Cu²⁺/⁶⁸Ga³⁺ with a residence time of 10 minutes at 37 °C.

For the radiolabeling experiments summarized in Fig. 5 (effect of concentration on extent of reaction), combinations of stock solutions and buffer-to-metal molar ratio inputs were used to give the following final concentrations of radiometal and ligand (1:1 molar ratio): 1 μM (2 μM stock solutions, 1:1 buffer-to-radiometal ratio), 10 μM (40 μM stock solutions, 500:1 buffer-to-radiometal ratio), 20 μM (40 μM stock solutions, 1:1 buffer-to-radiometal ratio), 30 μM (100 μM stock solutions, 130:1 buffer-to-radiometal ratio), 50 μM (100 μM stock solutions, 1:1 buffer-to-radiometal ratio), and 90 μM (200 μM stock solutions, 11:1 buffer-to-radiometal ratio). Experiments were performed with the same settings as listed above, a residence time of 10 minutes (t_REA = 12 minutes) and at a temperature of 37°C. After the collection of product, the yield of radiolabeling raction was determined as described below in the Measurement of radiolabeling yield section.

For the radiolabeling experiments summarized in Fig. 7 & 8 (label the DOTA-BSA protein with no-carrier added radiometals: ⁶⁸Ga³⁺/⁶⁴Cu²⁺): 100 μM ligand (DOTA-BSA), 10 mM NH4OAc buffer (pH = 6.50 for ⁶⁴Cu, and pH = 4.50 for ⁶⁸Ga), 1:1 buffer-to-radiometal ratio and 20: 1 ligand-to-radiometal ratio. Experiments were performed with the same settings as listed above, and with t_RES = 22 and 32 minutes and at a temperature of 37 °C. After the collection of product, 1.0 μL of 50 mM EDTA solution (pH = 5.00) was added and the mixture was incubated for a minute to chelate with unreacted radiometals. Then, the yield of radiolabeling raction was determined as described below in Measurement of radiolabeling yield section.

Fig. 7.

Preparation of no-carrier added ⁶⁴Cu²⁺/⁶⁸Ga³⁺ labeled BFC-BM with different L: M ratio at 37 °C, and a residence time of 10 minutes for DOTA/NOTA-RGD peptides and a residence time of 20 minutes for DOTA-BSA proteins.

Fig. 8.

Direct comparison of the labeling yields obtained from two approaches, microreactor and conventional methods under identical conditions: 37 °C; L:M = 20: 1; and a residence time of 10 minutes for DOTA/NOTA-RGD peptides and a residence time of 20 minutes for DOTA-BSA proteins.

Throughout the course of radiolabeling experiments, three trial runs were performed before data was recorded to ensure proper functioning of the microreactor after syringe pump or heater settings were changed and after depleted syringes were replenished.

Measurement of radiolabeling yield

In the case of the ⁶⁴Cu labeling of NOTA/DOTA conjugated cyclo(RGDfK), at the end of each reaction run, 1.0 μL aliquot of product was spotted onto a TLC(SG) plate and developed using a mobile phase of methanol/10% ammonium formate buffer (2:1 volume ratio). After developing, the radio-TLC plates were then counted using the Bioscan radio-TLC scanner to quantify free and bound ⁶⁴Cu²⁺. The radiolabeling yields were calculated by dividing the area of the radioactivity peak obtained for the bound ⁶⁴Cu²⁺ by the total area of the radioactivity peaks obtained for both bound and free ⁶⁴Cu²⁺.

For the ⁶⁸Ga labeling of NOTA/DOTA conjugated cyclo(RGDfK), at the end of each reaction run, a 2.0 μL aliquot of product was spotted onto a iTLC(SG) plate and developed using 0.1M citric acid as the mobile phase. After developing, the iTLC plates were then counted using the Bioscan radio-TLC scanner to quantify free and bound ⁶⁸Ga³⁺. The radiolabeling yields were calculated by dividing the area of the radioactivity peak obtained for the bound ⁶⁸Ga³⁺ by the total area of the radioactivity peaks obtained for both bound and free ⁶⁸Ga³⁺.

For the ⁶⁴Cu/⁶⁸Ga labeling of DOTA-BSA, after challenging with EDTA, a 2.0 μL aliquot was spotted onto a iTLC(SG) plate and developed using a mobile phase of methanol/10% ammonium formate buffer (2:1 volume ratio). After developing, the iTLC plates were then counted using the Bioscan radio-TLC scanner to quantify free and bound radiometal. The radiolabeling yields were calculated by dividing the area of the radioactivity peak obtained for the bound radiometal by the total area of the radioactivity peaks obtained for both bound and free radiometal.

RESULTS AND DISCUSSION

As mentioned previously, the microfludic approach employed in this work provides several advantages as discussed above, leading to overall better performance in the radiolabeling of biomolecules compared to conventional methods. In the following sections, the compatilibilty of PDMS/glass with radiometals, the optimization of labeling conditions for NOTA/DOTA–(RGDfK) with radiometals using the microreactor, and comparisons between microfluidic and conventional radiosynthesis will be discussed in detail.

Compatibility of PDMS/glass with ⁶⁸Ga³⁺/⁶⁴Cu²⁺

One of the major problems with PDMS-based microreactor designed for radiosynthesis of ¹⁸F-FDG is the loss of ¹⁸F through reaction with (or absorption by) the PDMS. Therefore, the compatiblity of the designed microreactor with radioisotopes is critical for its utilization in radiolabeling. In order to demonstrate the compatibility of our microreactor with ⁶⁴Cu, we have previously examined the retention of ⁶⁴Cu²⁺ and the ability of PDMS/glass to withstand the emissions of this radionuclide. Here, we present additional data on compatibility of ⁶⁸Ga³⁺ with PDMS/glass-based reactors.

The retention studies of ⁶⁸Ga³⁺/⁶⁴Cu²⁺ in a PDMS/glass microchannel (Fig. 3) were performed in single microchannels (length = 3 cm, volume = 0.6 μL) with staggered herringbone grooves. Before the retention studies, the microchannels were washed with nitric acid to remove trace amounts of metal ions. The data shown in Fig. 3, clearly show that the non pre-treated microchannels retained a substantial portion of the activity of the first injection (~70%), but retained only ~5% of the activity of subsequent injections. The microchannel that was pre-treated with non-radioactive Cu²⁺ solution retained only ~15% of the activity of the first injection and also retained only ~5% of the activity of subsequent injections. The activity retention of subsequent injections decreased to around 5% for both pretreated and non-pretreated microchannels. The retention of ⁶⁴Cu²⁺ on microchannel was hypothesized to be due to the non-specific electrostatic interactions between ⁶⁴Cu²⁺ and the microchannel surface, and this hypothesis is consistent with the observation that pretreatment with Na⁺ decreases the ⁶⁴Cu²⁺ retention similar to pretreatment with non-radioactive Cu²⁺(data not shown).

Fig. 3.

Retention of ⁶⁸Ga³⁺/⁶⁴Cu²⁺ in a PDMS/glass microchannel.

Interestingly, very little difference in the ⁶⁸Ga³⁺ retention (~15%) between the gallium-pretreated and untreated microchannels was observed, and no changes were evident after the microchannel was pre-treated with Na⁺ solution (data not shown). To determine whether the buffer influences the retention of ⁶⁴Cu²⁺/⁶⁸Ga³⁺, ⁶⁴Cu²⁺ stock solution was prepared in HEPES buffer (pH = 4.50) following the precedure for preparation of ⁶⁸Ga³⁺ stock solution (refer to experimental section). We observed that ⁶⁴Cu²⁺ retention performance in HEPES buffer (pH = 4.50) is similar to that in NH₄OAc buffer (pH = 6.50). Compared with ⁶⁸Ga³⁺, ⁶⁴Cu²⁺ has a much higher retention in the PDMS/glass microchannel in both two buffers, which indicates that the non-specific electrostatic interactions between gallium and PDMS/glass is very weak.

The PDMS-based microfluidic chip has withstood 260 mCi of radiometal and more than 720 hours of radiation exposure, and has shown no sign of degradation or loss of function. The only observed sign of damage with the microreactor, which has been used for six months, is leakage caused by mechanical failure of the seal between the PDMS and glass. Based on this delayed observation of failure and the preceding results on retention, we conclude that the microreactor is compatible with ⁶⁸Ga³⁺/⁶⁴Cu²⁺, and the interaction between the microreactor and ⁶⁴Cu²⁺ can be minimized by saturating the walls of the microchannels with cations. Thus, the microreactor is sufficiently robust and reliable for its envisioned role as a single or multiple use for radiometal labeling under aqueous condition.

Microfluidic radiolabeling of NOTA/DOTA–(RGDfK)

A variety of radiometal labeled cyclo(RGDfK) peptides have been used to monitor expression of the α_vβ₃ receptor, which is an integrin required for tumor-induced angiogenesis, and very promising results were obtained. The radiometal labeling of cylco(RGDfK) was conducted following the procedure described in the Experimental section. The labeling results were analyzed by radio-HPLC and/or radio-TLC(iTLC), and various TLC/iTLC developing solvents were used in order to obtain the best separation. The radiolabeling conditions were optimized by varying the reaction temperatures, concentrations and residence times, and the labeling results obtained using the microreactor were directly compared with those achieved via conventional radiosynthesis.

When radio-TLC was used to monitor the ⁶⁴Cu incorporation yield obtained after a short residence time (0.5 minute), two complex isomers were observed for commonly used chelates, such as Cyclen, Cyclam, DOTA and TETA (data not shown). However, once the residence time was prolonged to 5 minutes or longer, the appearance of one of the isomers was totally eliminated. We speculate that these isomers are probably the ⁶⁴Cu bound intermediate, which undergoes further rearrangement to form the final complex with stable conformation. In the case of ⁶⁴Cu-NOTA-RGD, the ratio of the two complex isomers (R_f on developed TLC were 0.46, 0.53 respectively) did not change much even after the residence time was prolonged from 2 to 20 minutes, although the total incorporation yield was significantly increased. Regardless of the changes in ligand to metal ratio, reactant concentration and reaction temperature, only very small variations were observed in the ratios of the two isomers formed from ⁶⁴Cu-NOTA-RGD when comparing the conventional method and the microfluidic approach, which suggests that both the isomers are stable and may have adopted different geometrical configurations due to the bulky p-NH₂ benzyl substituent in NOTA chelator. These observations are consistent with recent reports using the same or similar radioligands. Further investigation on how to produce a radiolabeled complex with a desired geometrical configuration is currently under progress, because different complex isomers may have different biological properties and applications in vitro and in vivo.

To optimize the labeling conditions using the microreactor, we first performed experiments with a fixed ligand to metal ratio (1:1), residence time (10 minutes) and incubation temperature (37 °C). The final concentration of radiometal, identical to the concentration of the ligand, was varied from 1 μM to 200 μM by adjusting both, the stock solutions and buffer-to-metal molar ratio, as previously described in the Experimental section. As shown in Figure 4, with increasing final concentrations of carrier-added ⁶⁴Cu²⁺/⁶⁸Ga³⁺, the reaction yields also increased dramatically from less than 10% with a final concentration of 1 μM to more than 90% with a final concentration of 90 μM. This increase is consistent with classical collision theory in that reaction rates (as well as reaction yields for a given period of time) generally increases with increasing concentrations in the reaction mixtures. In conventional radiosynthesis with 1: 1 ligand to metal ratio, the final concentration of radiometal is typically less than 1 μM (20 mCi ≈ 85 picomoles of ⁶⁴Cu or 100 mCi ≈ 100 picomoles of ⁶⁸Ga in 100 μL reaction mixture), which will result in low labeling yields at a 1:1 stoichiometric ratio. In addition, trace metal impurities found in the radiometal adversely affects the labeling yield. Only very low labeling yields (<2%) could be obtained in such a short reaction time. Therefore, large excess of ligands are usually employed in conventional radiosynthesis to compensate for dilution effects and metal impurities. Microfluidic approaches can alleviate the issues associated with dilution of cyclotron/generator produced radiometals by (1) using small volumes of concentrated radiometal, ligand and buffer solutions, and (2) implementing a microfluidic means to concentrate the reactants on-chip, such as evaporation of reaction solvent.

Fig. 4.

Radiolabeling yields as a function of the final concentration of carrier-added ⁶⁴Cu²⁺/⁶⁸Ga³⁺ on the microreactor using a 1: 1 stoichiometric ratio of radiometal to DOTA/NOTA-cyclo(RGDfK). Error bars represent standard deviation in the extent of reaction for three experiments (residence time = 10 minutes, T = 37 °C), using radiometal solutions made from the same batch of ⁶⁴Cu²⁺/⁶⁸Ga³⁺.

To study the effect of residence time, the labeling reactions were conducted by mixing 100 μM carrier-added ⁶⁴Cu²⁺/⁶⁸Ga³⁺ solution and 100 μM DOTA/NOTA-RGD solution at a 1: 1 stoichiometric ratio (100 μM concentration provided maximum yield) at 37 °C for different residence times (5 minutes, 10 minutes and 20 minutes); the data is summarized in Fig. 5a. Under these labeling conditions, at least a 10% increase in radiolabeling yield was observed for all four complexes when the residence time was prolonged from 5 minutes to 10 minutes. However, when the residence time was further extended, the increases in radiolabeling yields were relatively smaller compared to the increases obtained from 5 to 10 min. These observations were expected, since the use of longer residence times leads to better radiolabeling yields, because both reactants have longer reaction times in the reservoirs before elution from the system. As time elapses, the amounts of available reactants are reduced along with the formation of the complexes, and the reaction rates increase at a slower rate and finally reach a plateau.

To study the effect of temperature, we used a radiometal concentration of 100 μM (based on Fig. 4), and residence time of 10 min (based on Fig. 5a), but varied the temperature. Figure 5b shows radiolabeling yields obtained from the chelation reactions between carrier-added radiometals and the corresponding BFC-BMs (NOTA/DOTA-RGD) at different temperatures (23, 37 and 47 °C). At room temperature (23 °C), moderate labeling yields (55–75%) were obtained, except in the case of ⁶⁸Ga radiolabeling of DOTA-RGD (≤ 2.0%). When the temperature was increased, higher labeling yields (80–90%) were observed at 37 °C and 47 °C. The increase in labeling yield with temperature is due to the fact that the energy activation barrier is lowered for the reaction at higher temperatures, which leads to improved reaction kinetics. Especially in the case of ⁶⁸Ga labeling of DOTA-RGD, a significant improvement in labeling yield (from ≤ 2.0 % to 80 %) is obtained since it is a very temperature sensitive reaction.

To demonstrate the advantages of enhanced mass and heat transfer at the microscale, we compared the labeling performance obtained using the microreactor and conventional methods with carrier-added radiometals (containing over 99% non-radioactive copper/gallium, referred to as radiometal solutions). The efficient mixing by staggered herringbone grooves enables effective handling of small reagent volumes. The enhanced heat transfer due to shorter path lengths will lead to the improved incorporation yields of the radiolabeling reactions. To compare the performance obtained using microreactor and conventional methods, we used a radiometal concentration of 100 μM (based on Fig. 4), a residence time of 10 min (based on Fig. 5a), and a temperature of 37 °C (Fig. 5b). These parameters were chosen, because under these conditions labeling yields were very high but not maximum, so there was still room to observe changes in performance.

The comparison of radiolabeling yields at different temperatures between microfluidics and conventional radiosynthesis for ⁶⁸Ga-DOTA-RGD is shown in Fig. 6a, while the comparison for all four labeling reactions under above-mentioned conditions is shown in Fig. 6b. For radiolabeling of DOTA-RGD with carrier-added ⁶⁸Ga³⁺, the most temperature sensitive reaction of all the four reactions, the yields were greatly improved when the temperatures were increased using either method, but microreactor-based labeling led to a much larger enhancement in the radiolabeling yields than conventional labeling at elevated temperature (37 and 47 °C), which validates the hypothesis of enhanced heat transfer at the microscale. For all four reactions, the labeling yields obtained using the microreactor were higher than those obtained using conventional methods, due to improved heat and mass transfer (Fig. 6b).

Preparation of no carrier–added radiometal-labeled BFC-BM

All the above radiometal labeling experiments have been performed with carrier-added radiometals. The labeling with carrier-added radiometal is different from the labeling with no-carrier added radiometal, because the total amount of metal impurities in the latter case is typically high, in most cases even higher than the amount of radiometal of interest. For clinical and pre-clinical applications, typically no-carrier added radiometal solutions are used. To validate the performance of the microreactor for these applications, no-carrier added radiometal labeled BFC-BMs were prepared.

The four previously used radioligands (using no-carrier added radiometals), DOTA(⁶⁴Cu)-RGD, DOTA(⁶⁸Ga)-RGD, NOTA(⁶⁴Cu)-RGD and DOTA (⁶⁸Ga)-RGD were prepared by using a microreactor under the optimized conditions described in the previous section (at 37 °C, and a residence time of 10 minutes, while the concentration of radiometals varied from batch to batch: [⁶⁴Cu]= 3 – 5 μM; [⁶⁸Ga] = 5 – 8 μM). In addition to RGD, we also studied labeling of DOTA-BSA, as a representative protein for larger macro-biomolecules (proteins and antibodies). The purpose for labeling BSA was to evaluate the microreactor’s ability to label large molecules compared to peptides (smaller molecules), since larger molecules may mix slower due to lower diffusivity.

The results for these no-carrier added radiolabeling experiments are summarized in Figure 7. For all reactions: (1) BFC-BM was successfully radiolabeled using microreactors at 37 °C in a relatively short time (with a residence time of 10 minutes for RGD peptide), except for the labeling of BSA, where the residence time was 20 minutes; (2) excess ligand was generally required for efficient radioisotope incorporation in such a short time and the radiolabeling yield increased with increasing L:M ratio. This increase in yield is due to higher concentration of BFC-BM and increased probability of chelation of the BFC-BM with the radiometal of interest. Note that the values of L: M ratio in Fig. 7 may not be accurate, as the concentrations of no-carrier added radiometal were estimated from the average effective specific activity. In addition, the labeling performance of DOTA-RGD peptide and DOTA-BSA protein with no-carrier added radiometal using microreactor and conventional methods (Fig. 8) were also compared. Again, the microfluidic approach greatly improved the radiolabeling yields compared to the yields obtained with conventional methods, similar to the carrier-added radiolabeling results reported above. Figure 9 compares the calculated specific activites of the no-carrier added preparations of DOTA(Cu)-RGD, DOTA(Cu)-BSA and DOTA(Ga)-BSA based of the specific activities of the radiometals and the labeling yields. These no-carrier added labeling results suggest the tremendous potential of the microreactor for the preparation of radiopharmaceuticals in pre-clinical and/or clinical environments.

Fig. 9.

Direct comparison of the calculated specific activites obtained from two approaches, microreactor and conventional for no-carrier added radiolableling of DOTA(Cu)-RGD, DOTA(Cu)-BSA and DOTA(Ga)-BSA.

CONCLUSION

In summary, a robust, reliable, compact microreactor capable of chelating radiometals with common chelates has been developed and validated. The microreactor was used for the preparation of a variety of ⁶⁴Cu²⁺/⁶⁸Ga³⁺ labeled BFC-BMs, in which the BM included both peptide and protein, and the BFC included two common ligands DOTA and NOTA. The radiolabeling conditions were optimized with carrier-added radiometals by varying the residence time, concentration and temperature, and comprehensive comparisons between microfluidic and conventional methods were conducted with both carrier-added and no-carrier added radiometals. Based on our radiolabeling results, the reported microfluidic approach overall outperforms conventional radiosynthetic methods, and is thus a promising technology for the radiometal labeling of commonly utilized BFC-BMs in aqueous solutions.