Radiosynthesis and preclinical evaluation of [68Ga]Ga-NOTA-folate for PET imaging of folate receptor β-positive macrophages

Folate receptor β (FR-β), a marker expressed on macrophages, is a promising target for imaging of inflammation. Here, we report the radiosynthesis and preclinical evaluation of [68Ga]Ga-NOTA-folate (68Ga-FOL). After determining the affinity of 68Ga-FOL using cells expressing FR-β, we studied atherosclerotic mice with 68Ga-FOL and 18F-FDG PET/CT. In addition, we studied tracer distribution and co-localization with macrophages in aorta cryosections using autoradiography, histology, and immunostaining. The specificity of 68Ga-FOL was assessed in a blocking study with folate glucosamine. As a final step, human radiation doses were extrapolated from rat PET data. We were able to produce 68Ga-FOL with high radiochemical purity and moderate molar activity. Cell binding studies revealed that 68Ga-FOL had 5.1 nM affinity for FR-β. Myocardial uptake of 68Ga-FOL was 20-fold lower than that of 18F-FDG. Autoradiography and immunohistochemistry of the aorta revealed that 68Ga-FOL radioactivity co-localized with Mac-3–positive macrophage-rich atherosclerotic plaques. The plaque-to-healthy vessel wall ratio of 68Ga-FOL was significantly higher than that of 18F-FDG. Blocking studies verified that 68Ga-FOL was specific for FR. Based on estimations from rat data, the human effective dose was 0.0105 mSv/MBq. Together, these findings show that 68Ga-FOL represents a promising new FR-β–targeted tracer for imaging macrophage-associated inflammation.

, all of which have been investigated in preclinical studies for imaging of FR-overexpressing tumors. Previously, we successfully used 18 F-FOL PET to visualize FR-β-positive macrophages in mouse and rabbit models of atherosclerosis 3 . Atherosclerotic lesions exhibit chronic inflammation associated with accumulation of macrophages in the affected area, providing a rationale for investigating macrophagetargeted tracers.
In radiosynthesis of 68 Ga-radiopharmaceuticals, 68 Ge/ 68 Ga-generators are commonly used to obtain 68 Ga-radionuclides; importantly, these generators can be conveniently implemented in a laboratory setting. In 68 Ge/ 68 Gagenerators, a certain amount of 68 Ge is immobilized on a stationary phase, where the mother radionuclide decays into 68 Ga, which can be eluted out for radiolabeling reactions. In this study, we prepared [ 68 Ga]Ga-NOTA-folate ( 68 Ga-FOL, Fig. 1) with generator-produced 68 Ga and evaluated its potential for imaging of inflammation. 68 Ga-FOL shares the same precursor structure as 18 F-FOL 3 . Al 18 F labeling of NOTA-conjugates requires cyclotron facilities for the production of [ 18 F]fluorine, whereas generator-produced 68 Ga offers a convenient and costeffective option for radiolabeling. First, we determined the binding affinity of 68 Ga-FOL to human FR-β using transfected cells. Next, we investigated the uptake and specificity of intravenously (i.v.) administered 68 Ga-FOL for the detection of inflamed atherosclerotic lesions in mice and compared the tracer with 18 F-FDG. In addition, we determined the whole-body distribution kinetics in healthy rats, with and without the blocking agent folate glucosamine, and estimated the human radiation dose of 68 Ga-FOL.

Experimental section
General materials and equipment. NOTA-folate precursor was synthesized as previously described 14 . 68 GaCl 3 was obtained from 68 Ge/ 68 Ga IGG-100 generators (Eckert & Ziegler, Valencia, CA, USA) by elution with 0.1 M hydrochloric acid (HCl) in water. TraceSELECT water (Honeywell, Morristown, NJ, USA) was used for radiosynthesis. Other chemicals were purchased from commercial suppliers. Chinese hamster ovary (CHO) cells stably transfected with human FR-β (CHO-hFRb; CHO-FR-β + ) were a generous gift from Philip S. Low, Purdue University, USA. FR-β-negative CHO cells (CHO-FR-β − control) were a generous gift from Sirpa Jalkanen, MediCity Research Laboratory, University of Turku, Finland. A dedicated small animal PET/CT (Inveon Multimodality; Siemens Medical Solutions, Knoxville, TN, USA) was used for PET/CT imaging, and a gamma counter (1480 Wizard 3″, PerkinElmer/Wallac, Turku, Finland or Triathler 3″, Hidex, Turku, Finland) was used for radioactivity measurement of ex vivo tissues, blood, and plasma samples. Tracer quality control and plasma metabolite analysis were performed using a LaChrom high-performance liquid chromatography (HPLC) system (Hitachi; Merck, Darmstadt, Germany) equipped with a Radiomatic 150TR flow-through radioisotope detector (Packard, Meriden, CT, USA) (radio-HPLC). Photomicroscopy images were taken with a digital slide scanner (Pannoramic 250 Flash or Pannoramic P1000; 3DHistec Ltd., Budapest, Hungary). 68 Ga-FOL radiosynthesis. A fraction of 68 Ga-eluate (0.5-1.0 mL) was mixed with an aqueous solution of 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid (HEPES, 1.2 g/mL in 50-100 µL). NOTA-folate precursor (10-20 nmol in 20-40 µL water) was added, vortexed, and incubated for 10 min at 80 °C. The mixture was then cooled down and brought to a pH of ~ 6.5 by addition of 55 µL of 1 M sodium hydroxide (NaOH). The product was used without further purification. Radiochemical purity was analyzed primarily by HPLC. The HPLC conditions were as follows: 250 × 4.6 mm Jupiter Proteo 4µ C18 90 Å column (Phenomenex, Torrance, www.nature.com/scientificreports/ CA, USA); flow rate = 1 mL/min; wavelength λ = 220 nm; solvent A = 0.1% trifluoroacetic acid (TFA) in water; solvent B = 0.1% TFA in acetonitrile; gradient: during 0-14 min from 3% B to 35% B; during 14-15 min from 35% B to 3% B. Representative radio-and UV chromatograms are presented in Fig. 1B,C, respectively. During radiosynthesis set-up, the radiochemical purity was also analyzed by instant thin-layer chromatography (iTLC) (Supplementary Figure S1). A 1.0 µL sample of the end product or reaction mixture was applied to a silica gel-based iTLC strip (iTLC-SG; Agilent, Santa Clara, CA, USA) and developed with 50 mM citric acid. Unbound 68 Ga migrated up with the mobile phase with a retention factor (R f ) of 0.8-1.0, while 68 Ga-FOL remained at the application point (R f = 0). To measure the unbound and tracer-bound 68 Ga fractions, the strip was cut into two pieces along a line midway between the baseline and the solvent front, and each piece was measured separately in a gamma counter. 68 Ge breakthrough was monitored by collecting aliquots from the 68 Ge/ 68 Ga generator eluate and measuring their radioactivities at time of collection and again 24 to 48 h later.
The lipophilicity of 68 Ga-FOL (distribution coefficient LogD) was determined as previously described 15 . To evaluate 68 Ga-FOL stability in the injectable formulation, we kept the end product at room temperature (RT) and took samples for radio-HPLC analysis at time intervals of up to 3 h.
After verifying the presence of FR-β on the cells, either CHO-FR-β + or CHO-FR-β − cells were cultured on one side of a 92 mm petri dish in a tilted position in growth medium at 37 °C in a CO 2 incubator. The other side of the petri dish with no cells was used as background control for non-specific binding of 68 Ga-FOL. Once the cells grew to a confluent monolayer, the growth medium from the petri dishes was removed, and phosphate-buffered saline (PBS) containing calcium and magnesium with 10% FBS (binding medium) was added. To starve the cells of folate, the dishes were incubated for 30 min at 37 °C in a CO 2 incubator. After incubation, the cells were rinsed with binding medium (2 × 2 mL). A LigandTracer Yellow instrument (Ridgeview Instruments AB, Uppsala, Sweden) was used to measure the dissociation constant (K D ) for 68 Ga-FOL. The assay protocol with LigandTracer Yellow involves consecutive radioactivity measurements of the target (cell region) and the background (i.e., the cell-free region on the petri dish). Radioactivity was measured in each region for 30 s as raw counts per second (cps) with a delay of 5 s over the time course of the experiment. Target regions (cps) were corrected for background signal and radioactive decay. To detect background radioactivity or noise picked up by the instrument, 5 mL binding medium was added to the cells on the petri dish. After 15 min, 68 Ga-FOL was added stepwise to achieve a concentration range of 1 to 80 nM, followed by replacement with fresh binding medium to measure the dissociation. The ratio of bound 68 Ga-FOL (to the cells) to background (petri dish) and the K D value were calculated using the TraceDrawer software (Ridgeview Instruments AB).
Animal experiments. Low-density lipoprotein receptor-deficient mice expressing only apolipoprotein B100 (LDLR −/− ApoB 100/100 , strain #003000; Jackson Laboratory, Bar Harbor, ME, USA) were used to induce atherosclerosis. The mice were fed a high-fat diet (HFD; 0.2% total cholesterol, TD 88137, Envigo, Madison, WI, USA) starting at the age of 2 months and maintained for 3-5 months. C57BL/6JRj mice (Central Animal Laboratory of the University of Turku) fed with a regular chow diet were used as healthy controls. In total, 17 LDLR -/-ApoB 100/100 (34.7 ± 5.5 g) and six healthy control mice (29.65 ± 1.9 g) were studied. In addition, six Sprague-Dawley rats (135.9 ± 17.1 g) from the Central Animal Laboratory of the University of Turku were studied.
All animals were housed at the Central Animal Laboratory of the University of Turku, and had ad libitum access to water and food throughout the study. All animal experiments were approved by the national Animal Experiment Board in Finland (license number ESAVI/4567/2018) and were carried out in compliance with European Union Directive 2010/63/EU. Mouse studies. PET/CT Imaging. The mice were fasted for 4 h prior to imaging, anesthetized with isoflurane (4-5% induction, 1-2% maintenance), and placed on a heating pad. The mice then received i.v. 18 F-FDG (14.4 ± 0.2 MBq) via a tail vein cannula; the next day under the same conditions, they received 68 Ga-FOL (20.1 ± 1.0 MBq). Immediately after PET, an iodinated contrast agent (100 µL eXIATM160XL; Binitio Biomedical, Ottawa, ON, Canada) was i.v. injected, and high-resolution CT was performed for anatomical reference. The Carimas 2.10 software (Turku PET Centre, Turku, Finland, www.turku petce ntre.fi/carim as/) was used to analyze PET/CT images. We defined regions of interest (ROIs) for the myocardium in coronal PET/CT images using the contrast-enhanced CT as an anatomical reference, as previously described 2 . The results were normalized against the injected radioactivity dose and animal body weight, i.e., the data were expressed as standardized uptake values (SUVs). www.nature.com/scientificreports/ Ex vivo biodistribution. To study the specificity of 68 Ga-FOL uptake, an in vivo blocking study was performed with another group of HFD-fed LDLR -/-ApoB 100/100 mice i.v. injected with 68 Ga-FOL alone or 68 Ga-FOL in conjunction with a 100-fold molar excess of folate glucosamine. Mice were i.v. injected with 68 Ga-FOL (11.3 ± 0.8 MBq) and euthanized after 60 min. Various tissues were excised and weighed, and their radioactivity was measured with a γ-counter (Triathler 3″, Hidex). After compensating for radioactivity remaining in the tail and cannula, the ex vivo biodistribution of 68 Ga-radioactivity results were expressed as SUVs, and blocking and non-blocking results were compared.
Autoradiography, histology, and immunostainings. Following PET/CT imaging, the dissected aortic arch was processed into 20 and 8 µm cryosections. The 20 µm cryosections were used for digital autoradiography analysis as previously described 3 . Briefly, the sections were apposed on an Imaging Plate BAS-TR2025 (Fuji, Tokyo, Japan), and the plates were subsequently scanned on Fuji Analyzer BAS-5000 after an exposure time of 3 h for 68 Ga-FOL and at least 4 h for 18 F-FDG. After scanning, sections were stored at -70 °C until staining with hematoxylin-eosin (H&E) and then scanned with a Pannoramic digital slide scanner. Autoradiographs were analyzed using the Tina 2.1 software (Raytest Isotopemessgeräte, GmbH, Straubenhardt, Germany), and the uptake of 68 Ga-FOL and 18 F-FDG was corrected for injected radioactivity dose per unit body mass and radioactive decay during exposure; data were expressed as photostimulated luminescence per square millimeter (PSL/ mm 2 ). For immunohistochemistry, adjacent 8 µm sections were used to investigate co-localization of 68 Ga-FOL with Mac-3-positive macrophages. The sections were incubated with anti-mouse Mac-3 antibody (1:1,000; BD Biosciences, Franklin Lakes, NJ, USA), and a color reaction was subsequently developed using 3.3′-diaminobenzidine (Bright-DAB, BS04-110).

In vivo stability.
To determine the in vivo stability of 68 Ga-FOL, plasma samples collected from atherosclerotic mice (n = 3) at 60 min post-injection were analyzed using radio-HPLC. Blood samples were collected in heparinized tubes and centrifuged at 4 °C for 5 min at 2,118 × g. Plasma proteins were precipitated with 10% sulfosalicylic acid (1:1 v/v), followed by centrifugation at RT for 2 min at 14,000×g. The supernatant was analyzed by radio-HPLC. Standard samples were prepared by addition of 68 Ga-FOL tracer to 500 µL plasma supernatant collected from mice that had not received tracer. Both standard and metabolite samples applied to radio-HPLC analysis were normalized to a final volume of 1 mL by dilution with radio-HPLC solvent A if necessary. The radio-HPLC conditions were as follows: 250 × 10 mm Jupiter Proteo 5µ C18 90-Å column (Phenomenex); flow rate = 5 mL/min; solvent A = 0.1% TFA in water; solvent B = 0.1% TFA in acetonitrile; gradient, 0-11 min from 3% B to 25% B, 11-12 min from 25% B to 100% B, 12-14 min 100% B.

Rat studies.
To determine distribution kinetics and estimate human radiation dose, dynamic whole-body 68 Ga-FOL PET/CT was performed in six healthy rats. In addition, three of the rats were also subjected to a blocking experiment with co-injection of a 100-fold excess of folate glucosamine. The rats were injected with 10.3 ± 0.4 MBq of 68 Ga-FOL and PET imaged for 60 min. After imaging, the rats were euthanized, and various tissues were excised, weighed, and measured for radioactivity. Plasma samples were analyzed by radio-HPLC as described above. Using CT as an anatomical reference, quantitative PET image analysis was performed by defining ROIs on the main organs, and time-activity curves were extracted with the Carimas software. Human radiation dosimetry was estimated from the rat data using the OLINDA/EXM 2.2 software 17 .

Statistical analysis.
Results are presented as means ± SD. Differences between groups were analyzed by the unpaired Student t test using Microsoft Excel. P values < 0.05 were considered statistically significant.

In vitro quantification of 68 Ga-FOL binding affinity to FR-β. FACS analyses confirmed that FR-β was
clearly expressed on CHO-FR-β + cells but not on CHO-FR-β − cells ( Fig. 2A-C). In binding assays, with a stepwise increase in the concentration of 68 Ga-FOL from 1 to 80 nM, binding of 68 Ga-FOL to CHO-FR-β + cells was gradually increased and exhibited a K D of 5.1 ± 1.1 nM (n = 3). By contrast, we observed no clear accumulation of 68 Ga-FOL on CHO-FR-β − cells, even at concentrations up to 40 nM (Fig. 2D). 68 Ga-FOL detects macrophage-rich lesions in atherosclerotic mice. We evaluated the biodistribution of i.v.-administered 68 Ga-FOL in mice using in vivo PET/CT, ex vivo gamma counting of excised tissues, and ex vivo autoradiography of aorta cryosections. To study the specificity of 68 Ga-FOL to FR-β, we blocked by co-injecting a molar excess of folate glucosamine. In addition, we compared 68 Ga-FOL with 18 F-FDG in a headto-head PET/CT imaging setting, as well as by ex vivo autoradiography.
Our ex vivo results revealed that the aortic uptake of 68 Ga-FOL was higher in atherosclerotic mice (SUV 0.75 ± 0.12) than in healthy controls (SUV 0.41 ± 0.10, P = 0.004) or atherosclerotic mice from the blocking study (SUV 0.09 ± 0.03, P = 0.001). Furthermore, the concentration of radioactivity was threefold greater in www.nature.com/scientificreports/ atherosclerotic aorta than in blood (SUV 0.23 ± 0.09). The highest radioactivity uptake was in FR-positive kidneys 18 in both atherosclerotic and control mice (SUV 22.30 ± 3.28 and 20.27 ± 5.48, respectively, P = 0.49), and uptake was significantly reduced in a blocking study performed in atherosclerotic mice (SUV 2.65 ± 1.80, P = 0.0002). The radioactivity of other tissues was much lower than that of the kidneys. Besides the kidneys, blocking with folate glucosamine in atherosclerotic mice decreased the radioactivity concentration in many other tissues as well, including an 88% reduction in the aorta (Supplementary Table S1). A comparison of the two tracers by in vivo PET/CT revealed that myocardial uptake of 68 Ga-FOL (SUV 0.43 ± 0.06) was significantly lower than that of 18 F-FDG (SUV 10.6 ± 1.88, P = 0.001, Fig. 3A,B).
To further elucidate 68 Ga-FOL and 18 F-FDG uptake in the aortas of atherosclerotic mice in greater detail, we analyzed radioactivity using autoradiography and H&E staining of aortic cryosections followed by macrophagedetecting immunohistochemistry on adjacent tissue cryosections. The results revealed that 68 Ga-FOL and 18 F-FDG radioactivity co-localized with Mac-3-positive macrophage-rich plaques (Fig. 3C). The plaque-to-healthy vessel wall ratio of 68 Ga-FOL (2.44 ± 0.15) was significantly higher than that of 18 F-FDG (1.93 ± 0.22, P = 0.005, Fig. 3D).
Distribution kinetics in rats and estimation of the human radiation dose of 68 Ga-FOL. In rats as in mice, 68 Ga-FOL underwent fast renal excretion and highest uptake in kidneys, urine, salivary glands, liver, and spleen (Fig. 4). Co-injection of 68 Ga-FOL along with a molar excess of folate glucosamine clearly decreased tracer uptake in several organs but increased urinary excretion. Radio-HPLC analysis of plasma samples revealed that 68 Ga-FOL was relatively stable in vivo (Supplementary Figure S2); at 60 min post-injection, 71.8% ± 1.5% of the total radioactivity was from the intact tracer in healthy rats (n = 3) without blocking, and 88.0% ± 0.7% (n = 3, P = 0.0002) when blocked with folate glucosamine. Extrapolating from the rat PET data, the estimated human effective dose for a 73 kg man was 0.0105 mSv/MBq. The most critical organ was the kidney (0.1420 mSv/MBq) (Supplementary Table S2).

Discussion
In this study, 68 Ga-FOL was conveniently prepared with 68 Ga-eluate from a 68 Ge/ 68 Ga-generator based on a fractionation elution method. We observed that 68 Ga-FOL binds to FR-β with high affinity and accumulates in atherosclerotic lesions in mice following i.v. administration. Importantly, 68 Ga-FOL exhibited lower myocardial uptake and higher plaque-to-healthy vessel wall ratio than 18 F-FDG. Based on estimation from the rat data, the human radiation dose of 68 Ga-FOL was low.
To produce 68 Ga-FOL, we used a fractionation method to obtain 68 GaCl 3 from a 68 Ge/ 68 Ga-generator for a chelation reaction with the precursor compound NOTA-folate. This is a well-established method that we used previously, and the total radiosynthesis takes less than 20 min.
Previously, two folate-based imaging agents for single-photon emission computed tomography (SPECT), 99m Tc-EC20 and 111 In-EC0800, were used to detect atherosclerotic lesions in mice 19,20 . Additionally, the PET tracer 3′-aza-2′-18 F-fluorofolic acid can detect FR-β-positive macrophages in human atherosclerotic plaques in vitro 8 . In our previous studies, we demonstrated that 18 F-FOL is specific for FR-β-positive macrophages and can detect inflamed atherosclerotic plaques in mice and rabbits, as well as in human tissue Sects. 3 . However, the affinity of 68 Ga-FOL binding to human FR-β has not been previously evaluated, and its ability to detect atherosclerotic lesions has not been compared with that of 18 F-FDG. Our in vitro binding assay of 68 Ga-FOL with CHO-FR-β + and CHO-FR-β − cells revealed specificity and high affinity for FR-β (5.1 ± 1.1 nM), close to the binding affinity of 18 F-FOL (1.0 nM) for FR-positive tumor xenografts reported earlier 14 . Blocking studies in mice and rats further supported the tracer's specificity for FR, and mouse studies confirmed the ability to detect macrophage-rich inflammatory lesions. The observed low myocardial uptake is beneficial for detection of atherosclerotic lesions in coronary arteries in prospective PET/CT studies of patients with coronary heart disease. When compared with our earlier 18 F-FOL study, 68 Ga-FOL exhibited similar plaque-to-healthy vessel wall ratios (2.44 ± 0.15 for 68 Ga-FOL and 2.60 ± 0.58 for 18 F-FOL). However, the in vivo stability of 68 Ga-FOL in mice (63% ± 1% intact tracer at 60 min post-injection) was slightly lower than in our previous studies with 18 F-FOL (85% ± 6% at 60 min postinjection) 3 . The human effective dose of 68 Ga-FOL extrapolated from the rat data (0.0105 mSv/MBq) is low and within the same range as other 68 Ga-tracers [21][22][23] .
We have obtained 68 Ga-FOL with moderately high molar activity but the impact of molar activity on the imaging performance was not evaluated. On the other hand, we are also aware that high > 1,000 GBq/µmol molar activity does not always improve in vivo imaging, as reported by Wurzer and co-workers 24 . Our tracer, 68 Ga-FOL, has directly conjugated NOTA, i.e. one carboxylic arm is used to form an amide bond, which makes the structure less optimal for 68 Ga chelation than the N 3 O 3 hexadentate coordination 25 . We did not explore if a higher molar activity could have been achieved by replacing the NOTA with for example a 1,4,7-triazacyclononane,1-glutaric acid-4,7-acetic acid (NODAGA) 26 or triazacyclononane-phosphinate (TRAP) 27 chelator, which conserve the N 3 O 3 coordination for 68 Ga binding. This is our first-generation conjugate and further optimizations with better chelators such as NODAGA, or even more recent ones such as tris(hydroxypyridinone) (THP) and desferrioxamine (DFO) 28 are indeed warranted. In this particular lead compound, NOTA was chosen since it is a commonly used chelator for radiolabeling with both [ 18 F]AlF and 68 Ga. Although NOTA is not the most optimized chelator for 68 Ga, it is indeed applicable in the preparation of 68 Ga-radiopharmaceuticals for clinical use 29 .

conclusions
In summary, we have prepared 68 Ga-FOL and evaluated its FR-β targeting and imaging performance in vitro and in vivo. The preclinical results of 68 Ga-FOL are in line with those of our previous studies using 18  www.nature.com/scientificreports/