Acetuino—A Handy Open-Source Radiochemistry Module for the Preparation of [1- 11 C]Acetate

Electronics

As a control unit, we used an Arduino/Genuino Uno, which is easily available from commercial suppliers (e.g., RS Components, Conrad, Mouser). To connect the control unit to the peripherals, such as valves, switches, and power supply, we developed an adapter circuit (so-called shield) on a perfboard. The schematic and photographs of the shield are shown in the supplemental data ( Suppl. Figs. S1 and S9 , respectively). The shield’s circuit consisted of an ULN2803APG (Toshiba, Tokyo, Japan) Darlington transistor array, which was used as a driver for the solenoid valves. The status of the module was indicated by three 5 mm LEDs, which were connected to the output pins of the Arduino, each LED connected with a 220 Ω resistor in series. Three miniature rocker switches were attached for power supply (latching), synthesis start (momentary), and selection between cleaning and synthesis mode (latching). A photocoupler (LTV-817, Lite-On, Teipeh, Taiwan) was used to automatically trigger the synthesis on the status (24 V; 1.2 kΩ resistor on pin 1) of FX MeI Valve 9 (as alternative to the momentary switch mentioned before). A 14-pin ribbon cable connected the electronics with the peripherals inside the hot cells. A general-purpose DC power supply (24 V, 48 W) was used to drive the solenoid valves. An OCB350L062Z tube liquid sensor (TT Electronics, Woking, UK) was used to sense the complete transfer of liquids. The logic circuits were supplied by the same supply, converted to 5 V by a DC/DC converter (TSR 1-2450, Traco Electronic, Ismaning, Germany). The firmware was written in C and uploaded using the Arduino IDE 1.8.5. The supplemental information of this article contains the complete code ( Suppl. Fig. S5 ).

Mechanical Parts

The base plate holding the valves and vials was designed in OpenSCAD 2011.12 (http://www.openscad.org) and 3D printed with black acrylonitrile butadiene styrene (ABS) on a PRotos V2 (German RepRap, Feldkirchen, Germany) FDM printer with the following parameters: 0.5 mm nozzle, 0.3 mm layer height, 110 °C bed temperature, 235 °C extruder temperature, and 20% rectilinear fill. The G-code used was generated by Slic3r 1.2.9 (http://slic3r.org). A rendered image of the plate is shown in the supplemental data.

The housing of the electronics was designed with makercase (http://www.makercase.com) and Inkscape 0.91 (https://inkscape.org) and cut from 3 mm medium-density fiberboard (MDF) using a 100 W CO₂ lasercutting machine (1390, AM Laserpoint, Hamburg, Germany).

Fluidic Parts

Seven Type 0127 three-way solenoid valves (120433, Bürkert fluid control systems, Ingelfingen, Germany) were salvaged from the annual maintenance of our commercial radiosynthesizers. Six of them were screwed to the 3D-printed base plate using M3×16 mm screws and the respective nuts, while the remaining valve was installed between V5 and the Ascarite/Sicapent trap of the Tracerlab FX MeI module. To avoid clogging, 1/8-inch tubing was used to connect the loop to the gas waste (no. 1648, P-300, P-308, IDEX Health&Science, Erlangen, Germany), while standard 1/16-inch ETFE tubing with the respective ¼-inch -28 fittings (1528, P-200, P-208, IDEX Health&Science) was used to interconnect the remaining parts. The cartridges, the needles, and the loop (no. 1159.10, Vygon, Aachen, Germany) were attached using Luer connectors (P-655-01, P-658, IDEX Health&Science).

The 3D-printed base plate was designed with two cylindrical wells to accommodate the liquid reservoirs for water and phosphate-buffered saline (PBS), a squared well for the liquid detector printed circuit board (PCB) and a front plate with holes to guide the 1/16-inch tubing. Since the authors replaced clogging-prone parts of the tubing with 1/8-inch tubing during the development, not all these holes are actually used in the latest setup described here. To allow for larger tubing, V3b was mounted on the base plate at the outer left position, followed by V3a, V4a, and V4b.

General Radiochemistry

All solvents and reagents were obtained from commercial suppliers in at least analytical quality. Anhydrous tetrahydrofuran (THF) and methyl magnesium chloride (3 M in THF) were obtained from Sigma-Aldrich (Taufkirchen, Germany) and used without further purification. Chromafix PS-OH^– (200 mg), PS-H⁺ (230 mg), and PS-Ag⁺ (240 mg) cartridges (Macherey-Nagel, Düren, Germany) were conditioned with 5 mL of ethanol and 5 mL of water. NaOH-coated thin-layer chromatography (TLC) plates were prepared by soaking the plates (POLYGRAM SIL G/UV₂₅₄, Macherey-Nagel) in a saturated solution of sodium hydroxide in methanol, and stored in a sealed bag after vacuum drying at ~10 mbar for 30 min. Autoradiography of TLC plates was performed using the Cyclone Plus Storage Phosphor System (PerkinElmer, Waltham, MA).

Radio-high-performance liquid chromatography (radio-HPLC) was performed on a 1260 Infinity HPLC system (Agilent Technologies, Böblingen, Germany) equipped with an autosampler, an NaI(Tl) scintillator setup, and the multiwavelength detector set to 214 nm.

Preparation of the Module for Synthesis

The module was cleaned and dried between syntheses with water, acetone, and THF according to the protocol (see SOP in supplemental data, Suppl. Fig. S4 ). The septa of the reagent vials were renewed after every third synthesis.

Before each synthesis, the CH₄ oven was conditioned at 350 °C with hydrogen and helium for 15 min each, and then cooled to <35 °C before end of bombardment (EOB).

Reagent vials were filled with water and saline, respectively, according to the attached SOP. The product vial was filled with citrate buffer, and the needle was situated at the bottom of the vial. The conditioned cartridges were placed between the respective Luer connectors, and the loop was coated with methyl magnesium chloride by filling and retracting 2 mL of a 2 M solution in THF. After the loop was closed, an unblocked flow path was verified from the flow regulator reading of the FX MeI module at 10 mL/min.

Radiosynthesis of [1-¹¹C]Acetate

[¹¹C]CO₂ was prepared by bombardment of ¹⁴N₂ target gas containing 1% O₂ with 70 µA of 16.5 MeV protons using a PETtrace 860 cyclotron (GE Healthcare). Activity was transferred to the CH₄ oven of an FX MeI module and then released by heating to 380 °C. The released [¹¹C]CO₂ was transferred to the Acetuino module with a helium stream (40 mL/min) and trapped in the synthesis loop coated with Grignard reagent (“CO2 trapping” step, Table 1 ). After a reaction time of 3 min, the [1-¹¹C]acetate was transferred through a PS-H⁺ and a PS-Ag⁺ cartridge to a PS-OH^– cartridge using 40 mL of water (“loop elution” step). When the liquid detector indicated completion of this step, the PS-OH^– cartridge was eluted through a sterile filter into the product vial with 4 mL of 0.9% NaCl (“acetate elution” step). This procedure was executed fully automated from the time point of delivery from the cyclotron, using the trigger connected to V9 (FX MeI) and the liquid detector preceding the PS-OH^– cartridge ( Table 1 ).

OPEN IN VIEWER

Table 1.

Overview of the Automated Synthesis and Cleaning Steps and the Respective LED and Valve States.

Mode	Trigger	Step	V2	V3a/b	V4a/b	V5	V6	LED1	LED2	LED3
Synthesis		CO2 trapping	●	○	○	○	○	●	○	○
	Optocoupler/switch	Loop elution	○	●	○	●	○	●	●	○
	Liquid detector	Acetate elution	○	○	●	○	●	●	●	●
	Liquid detector + 3 min	End	○	○	○	○	○	○	○	○
Cleaning		Filling	○	○	○	○	○	○	○	●
	Switch	Big vial	○	●	○	●	○	○	●	●
	Timer	Small vial	○	○	●	○	●	●	●	●
	Timer	End	○	○	○	○	○	○	○	○

●

= on; ○ = off.

Quality Control

The radiochemical purity (RCP) of the product was assessed by TLC using 0.5 µL of the product solution on NaOH-impregnated silica plates with methanol as the mobile phase as described before. ⁸ [¹¹C]Acetate had Rf values of 0.8–0.9, while the contaminant near the baseline (Rf from 0.0 to 0.1) was considered [¹¹C]bicarbonate.

In the first experiments, radiochemical identity and chemical purity were analyzed by radio-HPLC on a Dionex CarboPak PA1 column (4 × 250 mm; Thermo Fisher Scientific, Waltham, MA) with a flow of 0.8 mL/min of 0.1 M carbonate-free NaOH (Sigma-Aldrich no. 43617, protected from atmospheric CO₂ by an Ascarite II trap).

Since this column completely retains the main potential contamination, [¹¹C]bicarbonate, we later used an Aminex HPX-87H column (300 × 7.8 mm; Bio-Rad, Munich, Germany) with 1 mM sulfuric acid at a flow of 0.6 mL/min for analysis of radiochemical identity and radiochemical and chemical purity.

pH was analyzed using pH indicator strips (MColorpHast pH 0-6.0, Merck, Darmstadt, Germany) and by visual evaluation.

Imaging

All experiments involving animals were conducted in accordance with the German animal protection laws after approval by the local authorities (Regierungspräsidium Tübingen).

Male C57BL/6 mice were obtained from Charles River (Sulzfeld, Germany) at an age of 6 weeks and kept in individually vented cages under standardized environmental conditions. Food and water were provided ad libitum. Animals anesthetized with 1.5% isoflurane (Abbott, Wiesbaden, Germany) evaporated in oxygen at a flow of 0.5 L/min were placed on an Inveon dedicated small animal PET scanner (Siemens Preclinical Solutions, Knoxville, TN). Fiducial markers were placed in proximity to the animals for better alignment of PET and magnetic resonance (MR) images. Five seconds after initiating the PET scan, ~13 MBq [1-¹¹C]acetate was injected via a tail vein catheter. Body temperature was controlled with a rectal temperature sensor and maintained at 37 °C with a heating pad. The PET scan was stopped after a total scan time of 3600 s, followed by a transmission scan for attenuation correction using the integrated Co-57 source. Subsequently, the animals were transferred into a 7 T BioSpec MR scanner (Bruker, Ettlingen, Germany) and anatomic images were acquired. After the MR scan, animals were killed and several organs were collected, weighed, and measured in a γ-counter (Wizard Wallac 1480, PerkinElmer) to determine the uptake in percent of injected dose per gram of tissue (%ID/g). PET images were histogrammed in 39 frames (12 × 5, 6 × 10, 6 × 30, 5 × 60, 10 × 300 s) and reconstructed with 2 iterations of the ordered subset expectation maximization 3D (OSEM3D) algorithm, followed by 18 iterations of the fast maximum a posteriori algorithm. No scatter correction was performed. Dynamic PET and MR images were aligned and analyzed in Inveon Research Workplace (Siemens Preclinical Solutions) and time–activity curves (TACs) were generated for the percent injected dose per cubic centimeter (%ID/cc) for several organs.

Acetuino Module Setup

The custom assembled Acetuino module described in this work (for photographs, see Suppl. Fig. S9 ) has a very compact footprint of only 12 × 13 × 8 cm³ (width, depth, height) inside the hot cell and 10 × 7 × 7 cm³ outside (excluding tubing and cables); thus, it can be situated together with a standard methylation module (e.g., Tracerlab FX M) in a small hot cell. By using the external valve V2, both the Acetuino and the methylation module can be supplied with ¹¹C by a Tracerlab FX MeI module without additional tubing rearrangement between [¹¹C]CO₂ and [¹¹C]MeI/MeOTf syntheses. The FX MeI time list triggers upon the purification of [¹¹C]acetate by switching MeI_V9, which is otherwise unused in the synthesis of [¹¹C]CO₂ and which is connected to the Acetuino using a photocoupler. Thus, the synthesis of [1-¹¹C]acetate is initiated by the Tracerlab software on the FX MeI and handed over to the Acetuino. Through the use of the tube liquid detector, sensing completion of washing and elution steps, the procedure is completely automated and finished in about 15–20 min after EOB.

Radiosynthesis, Purification, and Quality Control

We used the synthesis and purification procedure described by Soloviev et al. ⁹ as a starting point of our work, automated the procedure, and combined it with [¹¹C]CO₂ trapping and purification in our FX MeI module. In short, the Grignard reagent methyl magnesium chloride dissolved in THF was deposited on the loop walls by brief flushing. Since residual moisture could cause clogging, we found it practical to perform this as the last step of the preparation and to verify the absence of obstructions by analyzing the flow of helium carrier gas from the Tracerlab module after filling.

After irradiation of the target gas, [¹¹C]CO₂ was trapped in the CH₄ oven of the Tracerlab FX MeI, followed by elution at an elevated temperature of 380 °C in a flow of helium. In contrast to the recommendations by the manufacturer of the MeI module, in our experience conditioning the oven with hydrogen and helium (at 350 °C for 15 min each) between syntheses led to a more stable [1-¹¹C]acetate yield. However, we did not thoroughly follow up on this hypothesis due to our scheduled regular productions of ¹¹C-methylated tracers requiring daily conditioning of the oven. We therefore included this in our protocol for acetate production.

As expected and reported by Soloviev et al., trapping of the activity on the loop coated with Grignard reagent was almost quantitative, as no appreciable activity (<50 MBq) was found on Ascarite traps mounted to the gas waste tube. We calibrated the activity detector of the Tracerlab CH₄ oven by replacing the Acetuino loop with two subsequent Ascarite traps and measuring the radioactivity on both traps after release from the CH₄ oven using a dose calibrator. No activity reached the second trap; the activity on the first trap was thus used to calculate the radioactivity released from the CH₄ oven.

After transfer of [¹¹C]CO₂ to the loop and reaction with the Grignard reagent, the purification program flushed the loop with water; by this the reaction solution was guided over a stack of a PS-H⁺ and a PS-Ag⁺ cartridge for the hydrolysis and reduction of halide (here, chloride from the Grignard reagent) content, respectively. Afterwards, the flow was guided to the PS-OH^– cartridge where the product was trapped, and in the next step eluted into the product vial with isotonic sodium chloride solution.

After the first syntheses, we discovered that the “unknown anionic impurity” that Soloviev et al. observed also occurred in our setup in unpredictable amounts (Rf of 0.0–0.1 on TLC, not eluted on radio-HPLC with the CarboPak system). When we acidified the tracer solution and bubbled air, helium, or argon through the solution, the impurity was removed ( Fig. 2 ), leading us to the assumption that the contaminant might be [¹¹C]bicarbonate (bicarbonate is tightly retained on CarboPak PA1 columns at 0.1 M NaOH, requiring exclusion of CO₂ from the eluent). After we modified our synthesis and added 1 mL of 139 mM sodium citrate buffer (pH 4.7) to the product vial and bubbled air through the product solution for 3 min, contamination by [¹¹C]bicarbonate was generally avoided and the RCP of [¹¹C]acetate was at 94.4 ± 6.7% (n = 9). A similar procedure has been described before for acetate elution.^{12–14,18,22} Two out of nine syntheses showed an RCP below 95% (78% and 91%). In these cases, the septum of the NaCl vial was worn out, leading to inefficient bubbling. Based on this experience, we included a regular exchange of the septa (every third synthesis) into our routine.

OPEN IN VIEWER

Figure 2.

Quality control of [1-¹¹C]acetate by HPLC on an Aminex HPX-87H column before (A) and after (B) spiking with sodium acetate for peak identification. Radioactivity (upper graphs) and UV signal at 214 nm (lower graphs) are shown. (C) TLC of [1-¹¹C]acetate before (1) and after (2) helium bubbling, demonstrating the efficient removal of [¹¹C]bicarbonate (low Rf) from [1-¹¹C]acetate (high Rf).

For estimation of the RCP as well as the detection and identification of contaminants, we relied on both TLC and CarboPak-HPLC since the different literature-known contaminants, [¹¹C]bicarbonate, [¹¹C]acetone, and [¹¹C]tert-butanol, are not covered by only one of these methods. The latter two contaminants are separated from [¹¹C]acetate by HPLC, ⁹ but these contaminants were efficiently removed by the cartridge purification and not detected in the final tracer. The more problematic contaminant, [¹¹C]bicarbonate (as it was not always completely removed), was not detectable using this column since it is completely retained by the stationary phase. TLC in turn separated carbonate well from acetate, making both methods complementary for estimation of the RCP. The Aminex HPX-87H column used by Kruijer et al. and Le Bars et al. is reportedly able to resolve all the abovementioned compounds,^13,16 making this system an attractive option and justifying the extended analysis time of 30 min instead of 10 min, and the additional time needed to ramp up the flow rate according to the manufacturer’s instructions. We thus switched to this column for routine quality control of [1-¹¹C]acetate (Fig. 2A,B).

In summary, after optimization of the procedure, we were able to reliably produce [1-¹¹C]acetate in suitable amounts (3.2 ± 1.3 GBq at EOS) and quality (typically >95% RCP) for small animal imaging from 5 min target irradiations at 70 µA in only 15 min of synthesis and purification time. Radiochemical yields (relative to [¹¹C]CO₂ released from the CH₄ oven) were in the range of previously reported procedures (51.4 ± 28.2%, decay corrected).

Imaging and Ex Vivo Analysis

In our in vivo experiments, [1-¹¹C]acetate showed a strong perfusion peak in heart and kidneys. The radioactivity was quickly washed out, possibly after oxidative metabolism to bicarbonate. Already, ~400 s after tracer injection the liver became the organ with the highest uptake. A stable equilibrium of the tracer distribution was reached 1800 s after tracer injection. In the last time frame analyzed after the start of the PET scan (3300–3600 s), values of 0.5 ± 0.1, 1.5 ± 0.4, 0.9 ± 0.2, and 0.5 ± 0.2 %ID/cc for muscle, liver, kidney, and heart, respectively, were determined in the PET images (Fig. 3A,B). The biodistribution data confirmed the PET data and indicated the highest tracer uptake in the liver, followed by kidney, heart, and muscle. Relatively high uptake in the biodistribution was also observed in the lung and spleen, which we did not analyze with PET ( Fig. 3C ). Overall, the results for heart and liver are in good accordance with other studies of [1-¹¹C]acetate in this animal strain, although our uptake values are generally lower. ³⁰ These differences might be explained by the different PET system, different reconstruction algorithm, and animal handling.

OPEN IN VIEWER

Figure 3.

Results of in vivo experiments in healthy animals. (A) One single slice of MRT, PET, and the overlay of both modalities of one representative animal at 3300–3600 s after the start of the PET scan. (B) TACs for muscle, liver, kidney, and heart derived from PET. (C) Biodistribution data for several organs. Shown are mean values and SD (SD not shown in B for graphical reasons) from a total of six animals.

Here we describe a low-cost and low-footprint synthesizer for the production of [1-¹¹C]acetate for preclinical imaging. It is small enough to fit into an already-occupied hot cell, and we are using this setup routinely to start the synthesis of a ¹¹C-methylated tracer (e.g., for neuroimaging), directly followed (and eventually outpaced) by [1-¹¹C]acetate for oncological studies.

Numerous synthesizers for [1-¹¹C]acetate have been described, but most of them rely on closed-source hardware and software for controlling the valves and lack crucial information about their automation. Automation solutions often require the use of separate hot cells, hampering their use in research labs with limited space and resources. Our setup aims to be easily replicated at other sites using inexpensive equipment by anyone with basic skills in electronics. The growing network of FabLabs worldwide^27,28 aims to enable access to the production methods we used in this work, such as 3D printing, lasercutting, and soldering, without significant additional investments.

Due to the simplicity of the synthesis and purification task, we chose not to include a display and instead only used LEDs to indicate the progress of the synthesis. We did not include a radioactivity detector into the setup because the actual benefit would be low in such a rapid synthesis. We also chose to rely on existing infrastructure for [¹¹C]CO₂ trapping and release (Tracerlab FX MeI) instead of using nitrogen-cooled steel loops for direct connection to the cyclotron, since automated CO₂ purification capabilities are typically already available at cyclotron facilities.

Our device is able to produce no-carrier-added [1-¹¹C]acetate with high radiochemical yield and purity. This is not only shown by chemical analysis with TLC and HPLC, but also supported by the in vivo and ex vivo distribution of the tracer, as shown in dynamic PET scans and γ-counter studies, which show a similar biodistribution of our [1-¹¹C]acetate to what has been reported by others. ³⁰

We thus claim that the use of microcontrollers like the Arduino platform enables versatile construction and control of radiochemical synthesizers. Their low cost and small footprint allow dedication of the module for individual tasks. Setups based on this platform have become valuable tools in our lab, automating processes where full radiochemistry synthesizers are too bulky and uneconomical.