Automated Synthesis and Initial Evaluation of (4′-Amino-5′,8′-difluoro-1′H-spiro[piperidine-4,2′-quinazolin]-1-yl)(4-[18F]fluorophenyl)methanone for PET/MR Imaging of Inducible Nitric Oxide Synthase

Background Inducible nitric oxide synthase (iNOS) plays a crucial role in neuroinflammation, especially microglial activity, and may potentially represent a useful biomarker of neuroinflammation. In this study, we carefully defined a strategic plan to develop iNOS-targeted molecular PET imaging using (4′-amino-5′,8′-difluoro-1′H-spiro[piperidine-4,2′-quinazolin]-1-yl)(4-fluorophenyl)methanone ([18F]FBAT) as a tracer in a mouse model of lipopolysaccharide- (LPS-) induced brain inflammation. Methods An in vitro model, murine microglial BV2 cell line, was used to assess the uptake of [18F]FBAT in response to iNOS induction at the cellular level. In vivo whole-body dynamic PET/MR imaging was acquired in LPS-treated (5 mg/kg) and control mice. Standard uptake value (SUV), total volume of distribution (Vt), and area under the curve (AUC) based on the [18F]FBAT PET signals were determined. The expression of iNOS was confirmed by immunohistochemistry (IHC) of brain tissues. Results At the end of synthesis, the yield of [18F]FBAT was 2.2–3.1% (EOS), radiochemical purity was >99%, and molar radioactivity was 125–137 GBq/μmol. In vitro, [18F]FBAT rapidly and progressively accumulated in murine microglial BV2 cells exposed to LPS; however, [18F]FBAT accumulation was inhibited by aminoguanidine, a selective iNOS inhibitor. In vivo biodistribution studies of [18F]FBAT showed a significant increase in the liver and kidney on LPS-treated mice. At 3 h postinjection of LPS, in vivo, the [18F]FBAT accumulation ratios at 30 min post intravenous (i.v.) radiotracer injection for the whole brain, cortex, cerebellum, and brainstem were 2.16 ± 0.18, 1.53 ± 0.25, 1.41 ± 0.21, and 1.90 ± 0.12, respectively, compared to those of mice not injected with LPS. The mean area under the curve (AUC0-30min), total volume of distribution (Vt, mL/cm3), and Ki (influx rate) of [18F]FBAT were 1.9 ± 0.21- and 1.4 ± 0.22-fold higher in the 3 h LPS group, respectively, than in the control group. In the pharmacokinetic two-compartment model, the whole brain Ki of [18F]FBAT was significantly higher in mice injected with LPS compared to the control group. Aminoguanidine, selective iNOS inhibitor, pretreatment significantly reduced the AUC0-30min and Vt values in LPS-induced mice. Quantitative analysis of immunohistochemically stained brain sections confirmed iNOS was preferentially upregulated in the cerebellum and cortex of mice injected with LPS. Conclusion An automated robotic method was established for radiosynthesis of [18F]FBAT, and the preliminary in vitro and in vivo results demonstrated the feasibility of detecting iNOS activity/expression in LPS-treated neuroinflammation by noninvasive imaging with [18F]FBAT PET/MRI.


Introduction
Nitric oxide (NO) is a critical, unique mediator of a variety of physiological and pathological processes [1,2]. Two constitutively expressed nitric oxide synthase (NOS) isozymes-neuronal NOS (nNOS) and endothelial NOS (eNOS)-and the inducible isozyme iNOS generate NO by oxidizing Larginine to L-citrulline [3,4]. NOS are detected in a variety of tissues and participate in distinct physiological functions: NO produced by nNOS, which is mainly expressed in the peripheral nerves and brain, acts as a neuromodulator and neurotransmitter, whereas eNOS is primarily expressed in vascular endothelial cells and plays a role in regulation of blood pressure [1,5]. In contrast, iNOS is not constitutively expressed in cells and is only induced when cells are stimulated, typically by proinflammatory cytokines and/or bacterial lipopolysaccharide (LPS) [6,7]. In the central nervous system (CNS), iNOS is expressed in a variety of glial cells, such as astrocytes and microglia. Microglia are considered the "macrophages of the brain." Induction of iNOS expression is thought to be a specific marker of the M1 macrophage and microglial phenotypes [8] and mediates host defense and inflammatory processes induced by various stimuli. The levels of iNOS expression are limited in neurons compared to glial cells.
Nitric oxide (NO), the end product of metabolism catalyzed by the iNOS, is an important mediator of a variety of inflammatory diseases. When induced, iNOS generates significant amounts of NO (in the micromolar range), until the enzyme is degraded, sometimes hours later [9]. Production of NO by iNOS helps to defend against invading pathogens and thus plays critical roles in the inflammatory response and innate immune system. However, production of inappropriately high NO concentrations due to overexpression or dysregulation of iNOS can result in toxic effects and is associated with a variety of human diseases, including septic shock, cardiac dysfunction, pain, diabetes, and cancer [6].
LPS is frequently employed to induce neuroinflammation in animal models. LPS induces the release of immunologically active mediators implicated in inflammatory diseases-including TNF-α, IL-6, IL-8, and IL-1β-from host cells, especially mononuclear phagocytes [10]. Moreover, NO itself can also mediate inflammatory processes, as inhibiting iNOS activity and NO generation reduced the progression and severity of inflammatory disease in experimental models. Therefore, development of a radiolabeled iNOS radiopharmaceutical to noninvasively assess iNOS protein concentrations in living tissues-as a specific biomarker of iNOS activity-would advance our knowledge of NOrelated diseases and may help to identify novel treatments.
Because FFDI demonstrated similar or even superior potency in the cells to that of the standard iNOS inhibitors such as L-NMMA, FEITU, or MITU [14], therefore, in this report, based on FFDI, we developed (4 ′ -amino-5 ′ ,8 ′ -difluoro-1 ′ Hspiro[piperidine-4,2 ′ -quinazolin]-1-yl)(4-[ 18 F]fluorophenyl)methanone ([ 18 F]FBAT) as a tracer to detect iNOS expression in mice injected with lipopolysaccharide (LPS) to induce brain inflammation. We assessed the potential of [ 18 F]FBAT as a prognostic marker for the progression of neuroinflammatory disorders. [ 18 F]FBAT was radiosynthesized from aryl pinacol boronates via the copper-mediated aromatic nucleophilic radiofluorination at an acceptable radiochemical yield and molar activity. Moreover, the [ 18 F]FBAT PET findings were correlated with the results of immunohistochemistry. We anticipate that this study will not only advance our understanding of neuroinflammation diseases from the perspective of molecular imaging but also potentially help to develop novel tools to diagnose and guide the management of these diseases during the early progressive stages in humans.
2.2.6. Synthesis of Compound 5. The solution of 20 mL of methanol and metal sodium (0.364 g, 15.8 mmol) was placed into a 250 mL three-port bottle under N 2 atmosphere to prepare sodium methoxide, and hydroxylamine hydrochloride (1.100 g, 15.8 mmol) was added, the mixture was stirred, 2amino-3,6-difluorobenzonitrile (2.440 g, 15.8 mmol) was 3 Molecular Imaging added gradually, and then, the solution was allowed to equilibrate to room temperature and refluxed for 18 h.
The mixture was concentrated to oil that partitioned between ethyl acetate and 10% sodium hydroxide solution. The basic phase was separated and extracted three times with ethyl acetate, and the organic solution was washed three times with saturated brine and dried over sodium sulfate. The solvent was evaporated, and the product was purified by flash column chromatography on silica gel and eluted in DCM/MeOH = 100/1-50/1 to obtain compound 5 as a white solid product (1.831 g), yield 61.8%. 1  , and 20 mL isopropyl alcohol were added to a 50 mL single-neck bottle under N 2 atmosphere. The reaction mixture was refluxed overnight and cooled to precipitate a large amount of yellow solid. The mixture was filtered, and the filter cake was washed with ether to obtain 0.382 g of yellow solid. The solid was dissolved in 10 mL of methanol and adjusted to pH 8 with ammonia, and the mixture was extracted twice with dichloromethane. The organic phase was washed with saturated sodium chloride solution, dried over magnesium sulfate, filtered, and concentrated under reduced pressure. The crude product was purified by flash column chromatography on silica gel, eluted with DCM/MeOH (20/1 v/v) to obtain the product, which was recrystallized in dichloromethane and n-hexane to obtain the product (0.310 g) as an off-white solid (yield, 85.9%). 1  HCl was added, the solution was heated at 110°C for 10 min and cooled to room temperature, and 1 mL of 6 N NaOH was added to neutralize the solution. The resulting solution was passed through a 0.22 μm PVDF membrane filter (Millipore) into a 5 mL Reacti-vial. The reaction scheme is shown in Figure 1(a).

2.3.3.
Robotic System. All procedures described above were performed using a robotic system (Scanditronix Anatech RB III, Uppsala, Sweden). A diagram of the synthetic system is shown in Figure 1(b). The robotic system included a hand station that gripped the Reacti-vials, a syringe station that dispensed the solutions, a vortex station, a heating block station, cap stations that capped and uncapped the Reacti-vials, a solid phase extraction station for primary column purifications, a movable Reacti-vial rack station that collected the eluted fractions, a filtration station that filtered the raw [ 18 F]FBAT product, a filter check station that checked the integrity of the 0.22 μm membrane filters, a solvent rack station, and a water cooling station.

In Vitro Assessment of LPS-Dependent Nitric Oxide (NO)
Production. Previously, BV2 cells have been shown to respond to LPS stimulation by increasing the expression of iNOS [19] and released nitric oxide (NO) in a concentration-dependent manner [20]. Thus, we investigated the dose-dependent increase in NO production in LPS-challenged BV2 cells.
To assay NO, 100 μL aliquots of culture media were added to 100 μL of Griess reagent (1% sulfanilamide in 5% H 3 PO 4 ) in 96-well plates, incubated at room temperature for 10 min; then, 50 μL NED Griess reagent (0.1% Nnaphthyl-ethylenediamine dihydrochloride) was added and incubated for 10 min, and the absorbance values were determined at 540 nm using a TECAN Sunrise ELISA Reader. Fresh culture medium was used as a blank. The concentrations of nitrite were determined by comparison with a standard curve of sodium nitrite prepared in cell culture media.
2.7. In Vitro [ 18 F]FBAT Uptake Assay. BV2 cells were cultured as described above, washed three times with fresh medium, pretreated with a selective iNOS inhibitor, aminoguanidine (AMG), (0, 1, or 0.1 mM) for 1 h, and then treated with cell culture media containing LPS and the inhibitor. At the indicated time points, the cells were washed thrice with fresh cell culture media, then incubated in fresh cell culture media continuing the radiotracer [ 18 F]FBAT at 0.37 MBq/mL for 15, 30, or 60 min, harvested by gentle scraping, and pelleted by centrifugation at 1500 g for 2 min. The weight of the cell pellet and 0.1 mL of the radioactive supernatant was determined. A Packard 5500 gamma counter (PerkinElmer, Billerica, MA, USA) was used to quantify radioactivity as cpm/g cells or cpm/mL media, respectively. Cell-tomedium radioactivity concentration ratios were calculated and plotted over time to evaluate the kinetics of accumulation of the radiotracer.
iNOS is the enzyme responsible for NO generation in BV2 cells [20,21]; therefore, accumulation of [ 18 F]FBAT in BV2 cells before or after iNOS inhibitor versus NO production was also assessed.
2.8. Animals. All animal handling procedures were approved by Yang-Ming University Institutional Animal Care and Use Committee (IACUC No. 1050910), and animal study was performed according to the Guidelines for Animal Experimentation of National Yang-Ming University. For metabolism study, the ten-week-old male Sprague-Dawley rat (255 ± 12:3 g) or eight-week-old male C57BL/6 mice (22 ± 0:4 g) received food and water ad libitum and were housed under controlled room temperature (22 ± 2°C) and humidity (55-65%) under a 12 : 12 h light-dark cycle. The dark cycle lasted from 19:00 to 7:00.
We performed a preliminary study to identify the dose of LPS that induces moderate neuroinflammation in mice. Male C57BL/6 mice from BioLASCO (n = 18, weight = 25 ± 0:6 g) were intratracheally administered a bolus containing 5, 10, or 15 mg/kg LPS (Sigma-Aldrich; n = 6 mice per group). Behavior, breathing rate, appetite, and mortality rate were recorded over 48 h. The group receiving 10 and 15 mg/kg LPS exhibited mild to moderate clinical and physical symptoms of adult respiratory distress syndrome (ARDS) within 24 h of administration of LPS. All animals exhibited chills; the mortality rates for the 5, 10, and 15 mg/kg LPS groups were 16.6%, 50%, and 83%, respectively. The mortality rates reduced to 0%, 40%, and 50%, respectively, when the animals were kept on heating pads until they recovered; these animals were less mobile and had decreased appetite. Thus, 5 mg/kg was determined as a suitable dose of LPS to induce moderate neuroinflammation in mice.
For the imaging experiments, twelve C57BL/6 mice were injected i.p. with 0.9% NaCl, eighteen mice were injected i.p. with 0.1 mL of LPS in saline (5 mg/kg), and six mice were injected i.p. with aminoguanidine (Sigma-Aldrich) 30 min prior to 5 mg/kg LPS. In vivo imaging was performed 3 h or 24 h after injection of LPS; then, the mice were immediately humanely euthanized and the brains were removed for immunostaining. 2.10. In Vivo PET Imaging. Mice (n = 30) were anesthetized with 2% isoflurane (in oxygen) and injected with [ 18 F]FBAT (11.1 MBq; 0.3 mCi) via the tail vein. Dynamic PET images were obtained in fully three-dimensional list mode for 30 min using a small animal SuperArgus 2r PET system (SEDE-CAL, Madrid, Spain) or 7T PETMR Inline (Bruker, Rheinstetten, Germany; energy window, 350-650 keV; timing window, 6 ns). Images were acquired every 10 s for 12 images, 60 s for three images, 300 s for three images, or 600 s for four images. The anatomical structure of the brain was imaged using T2 MRI. The MRI sequences included 0. 5  F]FBAT radioactivity in the brain after administration of the iNOS inhibitor was assessed and compared to that of the mice in the LPS 3 h group.
2.12. PET Imaging Analysis. Images were reconstructed by the Fourier rebinning algorithm and two-dimensional filtered back projection using a ramp filter with a cutoff at Nyquist. The regional radioactivity concentration (kBq/cc) of [ 18 F]FBAT was estimated from the mean pixel values within the region of interest (ROI) corresponding to MR images of various organs and regions of the brain. ROI for carotid artery, heart muscle, lung, liver, spleen, and kidney were defined and time activity curves (TACs) generated. TACs for whole brain, cortex, cerebellum, and brainstem were also determined as described above.
The concentration of radioactivity (kBq/cc, μCi/cc) was converted to standardized uptake value (SUV), and the mean and standard deviation (SD) of radiotracer accumulation values were calculated for different organs and regions of the brain. Data were analyzed with PMOD 4.0 software (PMOD Technologies Ltd., Zurich, Switzerland).

Graphic Logan Graphical
Analyses. The dynamic PET imaging data were analyzed using Logan's modelindependent graphical analysis [22] to assess whether [ 18 F]FBAT PET/MR could be used to detect differences in expression of iNOS in brain regions (C bra ðtÞ). Cardiac blood TACs (C ref ðtÞ) were used as the reference tissue. The slope of the linear portion of the Logan plot represents the total distribution (V t ). Then, the slope of the linear portion of the plot can be calculated using Model parameters were estimated for influx constant k1 (mL/cm 3 /min -1 ), efflux (k2) (min -1 ) rate of radioligand diffusion between plasma and brain compartment. Exchange between compartments k3 (min -1 ) and k4 (min -1 ) was also estimated. The net influx constant, K i (min -1 ), parameter that describes the rate of binding to the iNOS was calculated as Compartmental modeling, pharmacokinetic analyses, and generation of pixel-by-pixel parametric images were accomplished using PMOD 4.0 software (PMOD Technologies Ltd., Zurich, Switzerland).

Immunohistochemistry.
After imaging, mice were terminally anesthetized with ketamine/xylazine and perfused with 4% paraformaldehyde, and the brains were dissected, postfixed overnight in 4% paraformaldehyde at 4°C. Fivemicrometer thick paraffin-embedded brain tissues were deparaffinized, rehydrated, microwaved in 10 mM citrate buffer (pH 6.0) at 100°C for 10 min for antigen retrieval, washed, incubated in 3% hydrogen peroxidase for 15 min at RT to inhibit endogenous peroxidases, and blocked in blocking solution for 60 min at RT.
The immunohistochemistry images were converted into 8-bit grayscale images in the ½0 -255 range. Immunostaining intensity was manually measured in the region of interest (ROI). The median (25%, 75% interquartile range) percentage score for each group (control, LPS 3 h, and LPS 24 h) was calculated as the sum of the individual number of slides within the group.
2.16. Statistics. Data are presented as the mean ± SD values for each group, and were compared using one-way ANOVA and the post hoc Bonferroni test or unpaired t -tests with Welch's correction using GraphPad Prism 8 (GraphPad Software, La Jolla, CA, USA). P < 0:05 was considered significant.

Chemistry. The preparation of FBAT precursor and authentic product [ 19 F]FBAT is shown in Schemes 1 and 2.
Briefly, the start material 4-piperidone was protected with ethylene glycol and converted to compound 7 then reacted with 4-iodo-or 4-fluorobenzoyl chloride via acylation to yield compound 1 and compound 8 with excellent yields, respectively. Compound 6 was obtained from 2-amino-3,6difluorobenzonitrile with hydroxylamine condensation and followed by Raney Ni reduction. The quinazolinamines, compound 3 and FBAT, were prepared from the condensation of compound 6 with the hydrolysis products of 1 and 8, respectively. Then, compound 3 was treated with Boc anhydride and modified by boronic ester via catalytic palladium coupling to yield FBAT precursor for radiofluorination. The authentic FBAT can be applied for analytic HPLC quality control.

Automated Radiosynthesis and Characterization of [ 18 F]FBAT.
We successfully completed five runs of n.c.a. [ 18 F]FBAT preparation using a robotic system. We found that the optimization of precursor/Cu(OTf) 2 ratio was 1.46 and temperature was 110°C (Figures 1(a) and 1(b)). On  Fig. 4A). The radioactive product was coinjected with an authentic FBAT standard. The retention time (t R ) of [ 18 F]FBAT (Suppl. Fig.  4B) in HPLC analysis was 14.38 min, which was consistent with that of authentic FBAT (Suppl. Fig. 4C). Molar activity (at the end of synthesis) ranged from 125 to 137 GBq/μmol. The radiochemical purity determined by HPLC was greater than 99%. The total volume of the purified product was 2.0 mL, and the pH was 6.0. Acetonitrile and trimethylamine were not detected in the final product. The Log P of FBAT measured between 1-octanol and phosphate buffer at pH 7.4 was 1:40 ± 0:27, which is a moderate value suitable for a brain imaging agent.

LPS Induces NO Production in a Dose-Dependent
Manner. LPS stimulation of NO release has been well established as an in vitro model of microglia activation. As shown ]FBAT accumulation at 60 min was more than 1.7-fold higher in LPS-induced BV2 cells than control BV2 cells (Figure 2(b)). Pretreatment of the BV2 cells with 0.1 mM or 1 mM AMG for 1 h significantly reduced the LPS-induced [ 18 F]FBAT accumulation (P < 0:05) (Figure 2(b), green or blue lines).
Furthermore, we analyzed NO production in murine microglial BV2 cells after [ 18 F]FBAT cell uptake study. LPS significantly increased NO production compared to control 10 Molecular Imaging cells. LPS-induced NO production was completely abolished by pretreatment with 0.1 or 1 mM of the iNOS inhibitor aminoguanidine (Figure 2(c)).

In Vivo Metabolism Study in Plasma.
Metabolism study has shown a rapid decrease of the parent fraction in plasma; the parent compound accounted for 77:98% ± 8:36 in plasma 11 Molecular Imaging at the first 5 minutes (Figure 3(a)). Roughly 51:41% ± 14:66 and 94:02% ± 4:75 of the metabolites were observed 15 min and 120 min after radiotracer injection, respectively (Figure 3(b)). [ 18 F]FBAT exhibited biexponential blood (carotid) clearance kinetics after i.v. injection ( Figure 5). The concentration of radioactivity remaining in the blood pool was 0:298 ± 0:019 SUV at 30 min postinjection. This pattern of hepatobiliary clearance was followed by a fast increased radioactivity in the blood and liver and subsequent clearance by the kidneys. The pattern of renal clearance was followed by a rapid increase in radioactivity in the blood and liver and subsequent clearance by the kidneys in the first 5 minutes. Rapid accumulation of [ 18 F]FBAT in the lungs, peaking at 1-2 minutes postinjection, was also observed. The concentration of radioactivity in heart muscle peaked in the first 1 min after injection and then gradually decreased over time ( Figure 5). The time activity curve of carotid as an input function was applied for Logan plot to estimate total volume distribution of [ 18 F]FBAT or pharmacokinetic two-compartment model. One of the aims of the study was to preliminarily measure and analyze [ 18 F]FBAT accumulation in the three major regions: cortex, cerebellum, and brainstem. Whole brain was used as an average uptake per voxel in all regions. PET imaging revealed rapidly cross BBB and accumulation of [ 18 F]FBAT in the whole brain and, to a lesser degree, in the The parametric images were used to determine the standardized uptake value ratios (SURs) after normalization to the control group ( Figure 8). SURs (SUV LPS-3h to SUV con ) were 2:16 ± 0:18 ( * P < 0:05) in the whole brain, 1:53 ± 0:05 ( * P < 0:05) in the cortex, 1:41 ± 0:21 ( * P < 0:05) in the cerebellum, and 1:91 ± 0:12 ( * P < 0:05) in the brainstem.  Figure 9(a) and Table 1.

In Vivo PET Imaging with [ 18 F]FBAT Enables
Mouse brain exposure to [ 18 F]FBAT was estimated in all conditions (control or LPS-induced groups) using the mean area under the curve of the tissue radioactivity from 0 to 30 min (AUC; SUV·min). Integrated activity (0-30 min) showed more [ 18 F]FBAT retention in the LPS 3 h group in comparison to control or LPS 24 h mice (Figure 9(b) and Table 2).

Neuroinflammation Appears Early on PET iNOS
Imaging. In model-independent Logan graphical analysis of dynamic PET imaging data using blood as the reference tissue devoid of iNOS protein expression, the average volume distribution (V t , mL/cm 3 ) for [ 18 F]FBAT in brain regions was 0:40 ± 0:02 (Figure 9(b)). In control animals, V t of [ 18 F]FBAT in the brain regions was 0:29 ± 0:05. The differences in V t between the LPS and control groups were statistically significant in the cortex, cerebellum, and brainstem (Figure 9(c) and Table 3).
Additional to using Logan graphical analysis to calculate volume distribution (V t , mL/cm 3 ), the net influx rate constant (K i , min -1 ) was also calculated using two-tissue compartment model. The whole brain K i of [ 18 F]FBAT (0:007 ± 0:003 min −1 ) in the LPS group was significantly higher than that of controls (4:45E − 06 ± 2:01E − 06 min −1 ). The influx (k1, mL/cm 3 /min) and efflux (k2, 1/min) rate constants of [ 18 F]FBAT in the whole brain were no difference between the control and LPS groups which are consistent with the moderate lipophilicity of the radiotracer. The [ 18 F]FBAT k1, k2, and k3 rate constants and K i for various brain structures are presented in Table 4.  Data are mean ± SEM; n = 6 per group; * P < 0:05, * * P < 0:005, and * * * P < 0:001 compared to controls; # P < 0:05, ## P < 0:005, and ### P < 0:001 compared to LPS 3 h. Data are mean ± SEM; n = 6 per group; * P < 0:05 and * * P < 0:005 compared to the controls; ### P < 0:001 compared to the LPS 3 h group. Quantitative analysis of the immunostained sections confirmed iNOS immunoactivity preferentially accumulated in viable portions of the cerebellum and cortex of mice with LPS-induced inflammatory responses (in the 3 h LPS group; Figure 10(b)). Moreover, higher iNOS immunoactivity was observed in the cortex, granular layer, and Purkinje cells in the cerebellum at 3 h postinjection of [ 18 F]FBAT. In the cortex, iNOS immunoactivity was higher at 3 h after injection of LPS than 24 h after injection of LPS ( * * P < 0:01), whereas iNOS immunoactivity in the cerebellum was not significantly different between the LPS 3 h and 24 h groups (Figure 10(b)). Pretreatment with aminoguanidine significantly reduced LPSinduced iNOS immunoreactivity in the cortex ( # P < 0:05), but not in the cerebellum, compared to the LPS 3 h group.

Discussion
18 F-labeled aromatics are the most commonly prepared using S N Ar reactions [23,24]. These reactions typically require high temperatures (often >150°C) and are restricted to electron-deficient substrates. In this study, we initially used a nitro compound as a precursor [15], though the radiochemical yield (<0.1%, uncorrected for decay) and molar activity (2 GBq/μmol) were very poor. It would be very difficult to carry out radiofluorination by substituting nitro group on the benzene ring if there were no electron-withdrawing groups in the nitrobenzene ring. Recent advances have expanded the scope of nucleophilic aromatic radiofluorination by using triarylsulfonium diarylselenone [25] [26] or iodonium ylide precursors [27][28][29]. The Sanford group previously reported Cu-mediated fluorination of aryl trifluoroborates, arylboronate esters, and arylboronic acids with KF [30] which provides an alternative way to perform radiofluorination of benzene ring with the lack of electron-withdrawing groups and obtain an acceptable yield. Copper-mediated aromatic nucleophilic radiofluorination was used to radiosynthesize [ 18 F]FBAT by [31] a boron ester 2-N-protected precursor. Pyridine is essential for copper-mediated radiolabeling. The radiochemical yield increased when we increased the pyridine/Cu(II)(OTf) 2 molar ratio from 15 to 25, in agreement with a previous study [32]. However, future studies will explore the potential strategies for reducing the time (200 min) and cost of radiosynthesis for [ 18 F]FBAT.
Radiolabeled [ 18 F]F-NOS has been previously investigated and demonstrated that iNOS activity correlated significantly with lung tissue measurements in healthy volunteers and in myocardial tissue of orthotopic heart transplant patients undergoing surveillance for rejection [33,34]. Expression of iNOS in activated macrophages was assessed in vivo using [ 18 F]F-6-(2-fluoropropyl)-4-methyl-pyridin-2amine, both in a LPS-induced lung inflammation of mouse model and endotoxin-induced lung inflammation in healthy volunteers. Significantly higher [ 18 F]F-6-(2-fluoropropyl)-4methyl-pyridin-2-amine uptake was reported in the lungs of mice treated with LPS than control mice [15]. However, no studies of neuroinflammation in animal or humans using this approach have yet been reported.
To our knowledge, this is the first report of [ 18 F]FBAT synthesis and the first histologically validated study of [ 18 F]FBAT for in vitro and in vivo PET imaging. We Data are mean ± SEM; n = 6 per group; * P < 0:05 and * * P < 0:005 compared to the controls; # P < 0:05 compared to the LPS 3 h group.   Our metabolism study showed a rapid decrease of the parent fraction in plasma where roughly 50% of the metabolites was detected at 15 min postinjection of [ 18 F]FBAT. However, most of the iNOS radioligands available are also known to show similar metabolism. For example, [ 11 C]L-NAME revealed that at 10 min postinjection 65% of the radioactivity was [ 11 C]methanol (metabolite) in dog [36]; within 5 min postinjection, only 40.3% of the activity in the blood was parent [ 18 F]9 [16]. This metabolism in the plasma may be a disadvantage in terms of [ 18 F]FBAT use since it makes its quantification more difficult. Nevertheless, the contribution of the major metabolites of [ 18 F]FBAT circulating in blood which led to accumulation of [ 18 F]FBAT-derived radioactivity in the brain needs to be accounted for in future investigations.
The results of in vivo biodistribution studies showed that the highest uptake in both groups was observed in the liver and kidneys, which are likely to be the major metabolic and/or excretory sites for [ 18 F]FBAT. The rapid whole-body redistribution and fast renal clearance of [ 18 F]FBAT from circulation make this radiotracer especially suitable for imaging of the chest/abdominal area, specifically of the lung. No significant accumulation of [ 18 F]FBAT-derived radioactivity was detected in the skeletal structures up to 30 min postinjection of [ 18 F]FBAT, which suggests that the radiotracer was not catabolized or defluorinated in vivo. Our results of wholebody biodistribution of [ 18 F]FBAT-derived radioactivity were in agreement with iNOS distribution in organs in male BLAB/c mice that reported iNOS mRNA and protein expression 6 h after LPS stimulation was observed in many organs. Among them, the highest iNOS expression was in the lungs followed by that in the kidneys, heart, gut, and liver [37].
In this initial evaluation, we used AUC and V t of [ 18 F]FBAT to quantify [ 18 F]FBAT accumulation in the brain from the PET images. The AUC 0-30min and V t values of [ 18 F]FBAT in the whole brain were averagely 1:9 ± 0:21and 1:4 ± 0:22-fold lower in control mice than in the LPS 3 h group. After treatment with selective iNOS inhibitor, AUC 0-30min and V t values showed a statistically significant reduction in AUC and V t of [ 18 F]FBAT in the LPS group (P < 0:001 and P < 0:05, respectively).
Further studies such as coupling imaging of PET [ 18 F]FBAT and dynamic contrast-enhanced MRI (DCE-MRI) assess dose dependence of LPS on BBB permeability/disruption in rodent model or neuroinflammation produced by intracerebral injection of LPS.
The higher resolution 7T PET/MR system was used to enable more detailed assessment of the heterogeneous accumulation of [ 18 F]FBAT radioactivity. The PET/MR results indicated that the radioactivity of [ 18 F]FBAT correlated with the levels of iNOS expression induced by cellular inflammatory responses in the brain observed using quantitative immunohistochemistry (QIHC). Pretreatment of animals with the selective iNOS inhibitor aminoguanidine significantly decreased [ 18 F]FBAT accumulation in all structures of the brain, consistent with the QIHC studies.
In comparison with previous reports of iNOS imaging agents [11][12][13][14][15][16], the results of our in vivo PET imaging studies with [ 18 F]FBAT in mouse brain expressing different levels of iNOS after LPS inducement are more selective. Predominant accumulation of [ 18 F]FBAT in iNOS-expressing brain regions could be implied to be responsive to therapy with selective iNOS inhibitor (e.g., aminoguanidine) [38].
The LPS animal model of neuroinflammation used in this study has several essential advantages, including technical ease and high reproducibility, particularly with respect to the magnitude of the inflammatory response elicited. Shortly after administration of LPS, high levels of proinflammatory cytokines are released and lead to rapid development of systemic inflammatory response syndrome (SIRS) and subsequent dose-dependent mortality [39,40]. Thus, a well-characterized method is needed to monitor the inflammatory reactions in vivo after LPS treatment. Hou et al. [41] used [ 18 F]-ROStrace PET imaging targeted to superoxide and reported the level of neuroinflammation induced by LPS (5 mg/kg) was strongly associated with the severity of the pain and distress scores [42].
Another critical issue that needs to be addressed is whether [ 18 F]FBAT in the systemic circulation crosses the BBB and directly activates cells within the CNS. Based on the current results, we consider that [ 18 F]FBAT crosses the BBB and cell membranes bidirectionally by nonfacilitated diffusion, due to its physicochemical characteristics (Log P 1.40). Therefore, the rate constants of [ 18 F]FBAT influx (k1 FBAT ) and efflux (k2 FBAT ) across the BBB and cell membranes are assumed to be high and not rate limiting. The critical issue of whether LPS actually disrupts the BBB and results in overestimation of [ 18 F]FBAT accumulation in the brain can be answered using the pharmacokinetic twocompartment model. Based on our preliminary evaluation, [ 18 F]FBAT influx (k1 FBAT ) and efflux (k2 FBAT ) were similar with or without LPS induction; however, k3 was significantly different between LPS-injected and control mice. This suggests that the higher [ 18 F]FBAT V t observed in the brain of LPS-injected mice is most likely due to increased expression of iNOS in the brain parenchyma. Also, due to the low brain uptake of the tracer (1 to 2 SUV), cerebral passage of [ 18 F]FBAT does not depend on cerebral blood flow. Thus, V t represents the most robust parameter for quantifying [ 18 F]FBAT uptake and is not affected by changes in cerebral blood flow. Nevertheless, further metabolite analysis of [ 18 F]FBAT in the brain and knowledge of the dose-response relationship between LPS and [ 18 F]FBAT accumulation in the brain are required to validate the implications of BBB disruption by LPS.

Conclusion
An automated robotic method was established for radiosynthesis of [ 18 F]FBAT from a boron ester precursor in a