A Novel PET Imaging Probe for the Detection and Monitoring of Translocator Protein 18 kDa Expression in Pathological Disorders

A new fluorine-substituted ligand, compound 1 (CB251), with a very high affinity (Ki = 0.27 ± 0.09 nM) and selectivity for the 18-kDa translocator protein (TSPO), is presented as an attractive biomarker for the diagnosis of neuroinflammation, neurodegeneration and tumour progression. To test compound 1 as a TSPO PET imaging agent in vivo, 2-(2-(4-(2-[18F]fluoroethoxy)phenyl)-6,8-dichloroimidazo[1,2-a]pyridin-3-yl)-N,N-dipropylacetamide ([18F]1; [18F]CB251) was synthesized by nucleophilic aliphatic substitution in a single-step radiolabelling procedure with a 11.1 ± 3.5% (n = 14, decay corrected) radiochemical yield and over 99% radiochemical purity. In animal PET imaging studies, [18F]CB251 provided a clearly visible image of the inflammatory lesion with the binding potential of the specifically bound radioligand relative to the non-displaceable radioligand in tissue (BPND 1.83 ± 0.18), in a neuroinflammation rat model based on the unilateral stereotaxic injection of lipopolysaccharide (LPS), comparable to that of [11C]PBR28 (BPND 1.55 ± 0.41). [18F]CB251 showed moderate tumour uptake (1.96 ± 0.11%ID/g at 1 h post injection) in human glioblastoma U87-MG xenografts. These results suggest that [18F]CB251 is a promising TSPO PET imaging agent for neuroinflammation and TSPO-rich cancers.

Microglia constitute up to 10% of the total cell population of the brain and make up the largest resident population of macrophages, which play an important role in the pathogenesis of several neurodegenerative disorders. In fact, in response to a wide variety of central nervous system (CNS) insults, microglia change their morphological state from a resting phenotype to an activated phenotype 1 . To date, various degrees of microglia activation have been observed in neurodegenerative disorders. Relevant cases include the high density of microglia surrounding the amyloid plaques, which are a hallmark pathology in Alzheimer´s disease (AD) and the abundant presence of activated microglia in zones of demyelination in multiple sclerosis 2,3 . Although it is well accepted that microglia activation contributes to the pathogenesis of several neurodegenerative diseases, including AD, multiple sclerosis and HIV-associated dementia, to date its detection is predominantly restricted to histopathological techniques utilizing post-mortem CNS tissues 1 . Computerized tomography and magnetic resonance imaging technologies allow morphometric documentation of late-stage changes in brain volume but are generally not regarded as sensitive or early detectors of brain damage. Recent studies have suggested that activated microglia in the CNS may be detected in vivo using positron emission tomography (PET) with the aid of pharmacological ligands of the Scientific RepoRts | 6:20422 | DOI: 10.1038/srep20422 mitochondrial translocator protein 18 kDa (TSPO) 4 . Furthermore, TSPO levels appear to be enhanced in cancer cells, including brain, breast, colon, prostate and ovarian cancers, as well as in astrocytomas and hepatocellular and endometrial carcinomas 5,6 .
Therefore, the successful development of a TSPO-specific PET tracer may help to chart the progression of pathologies in which TSPO is overexpressed, as well as to assess the efficacy of therapies designed to control these diseases. Potent ligands for TSPO have been identified from many different structural classes, such as isoquinoline carboxamides (e.g., PK 11195) 7 , benzodiazepines (e.g., Ro-54864) 8 , phenoxyarylacetamides (e.g., DAA1106) 9 , aryloxyanilides (e.g., PBR28) 10 and 2-phenyl-imidazo[1,2-a]pyridine acetamides (e.g., alpidem) 11 . Many of these ligands have been developed to visualize activated microglia via several imaging techniques. In this context, we recently proposed a practical and easy approach to achieve this objective using fluorescent probes chemically linked to 2-phenyl-imidazo[1,2-a]pyridine acetamide TSPO ligands 12,13 . It should also be mentioned that the first examples of targeted nanocarriers for imaging TSPO in vitro using fluorescent probes have recently been presented 14,15 . Regarding PET, potent TSPO ligands have been explored as potential radioligands. Most of these are carbon-11-radiolabelled TSPO ligands such as the well-known [ 11 C]PK 11195, which has been used to image TSPO in pathological states such as neuroinflammatory diseases and tumours by either autoradiographic methods or PET 16 . Although these studies established the potential to image TSPO in pathologies that overexpress this mitochondrial protein, they also highlighted important limitations of [ 11 C]PK 11195 as a molecular imaging probe, such as a high level of nonspecific binding, a low signal-to-noise ratio and nonspecific uptake 17 . Therefore, a number of alternative TSPO PET radioligands for neuroinflammatory imaging have been developed over the last several years. We recently reported the development of novel imidazopyridine acetamide PET ligands with improved in vivo specificity for TSPO 18,19 . In particular, distribution studies in mice showed that the most promising N-[ 11 C]methylated imidazopyridine acetamide accumulated in TSPO-rich regions and that the radioactivity in brain regions was reduced by co-injection of PK 11195 18 . Nevertheless, the use of such radioligands is necessarily restricted to imaging facilities either at or very close to their site of production because of the short half-life of carbon-11 (20.38 min). The use of fluorine-18 (half-life 109.7 min) as an alternative label offers several advantages over carbon-11. First, high (multi-Ci) activities of fluorine-18, such as [ 18 F]fluoride, can be produced from moderate energy cyclotrons by the 18 O(p,n) 18 F reaction on [ 18 O] water. Second, the longer half-life allows transport over considerable distances to remote PET centres and also allows longer and more accurate imaging sessions. Finally, fluorine-18 decays by emitting a low-energy positron, which permits the high spatial resolution of modern PET cameras to be fully utilized.
To date, only a few 18 F-labelled TSPO ligands have been reported. One of these is a close analogue of [ 11 C] PK 11195, namely, 1-(2-[ 18 F]fluoro-5-nitrophenyl)-N-methyl-N-(1-methylpropyl)-3-isoquinoline carboxamide ([ 18 F]PK 14105), which was shown to be a marker of brain ischaemic lesions in rats but has not yet been evaluated in humans 20 . Another such ligand is the recently reported high-affinity [ 18 F]fluoropropyl analogue of iodo-PK 11195, whose imaging properties have not yet been described 21 . However, even the most promising of these radioligands suffers from extensive metabolism and defluorination. In the last few years, considerable effort has gone into the development of new radioligands that target TSPO. Thus, radioligands from imidazopyridine and other structural classes with improved in vivo specificity for TSPO and with the potential to improve PET imaging of TSPO expression have been synthesized and evaluated in animals 22,23 23,24 . Alpidem has been shown to act on both TSPO and central benzodiazepine receptors (CBRs), with a preference towards TSPO. In an effort to increase this selectivity, some groups have synthesized a series of potent, imidazo[1,2-a]pyridine-based TSPO ligands by introducing various substituents onto the alpidem structure at positions 6 and 8 of the imidazo[1,2-a]pyridine nucleus (Fig. 1). In particular, the main results of our previous structure-activity-relationship studies, combined with a pharmacophore model of TSPO ligands, revealed that substituents at the 8-position (Y, Fig. 1), such as a chlorine atom, lead to compounds with enhanced TSPO selectivity (K i (CBR)/K i (TSPO)> 10 4 ) 11 . To our knowledge, no 18 Fig. 1, as a TSPO-selective PET radiotracer that is useful for imaging activated microglia, as well as TSPO-rich brain tumours.
Radiosynthesis. The radiosynthesis of [ 18 F]CB251 was accomplished starting from the tosylate precursor 5 via nucleophilic aliphatic substitution under various reaction conditions. Table 1 summarizes the results of the [ 18 F]fluorine incorporation yields that were achieved using different bases, solvents, reaction times and temperatures. Among the reaction conditions, promising results were obtained when nBu 4 NHCO 3 (4.0 equiv. relative to the precursor) and compound 5 were used in the presence of tert-butanol as the reaction solvent at 120 °C for 10 min (Table 1, entry 7). The reaction mixture was separated by via semi-preparative HPLC and the product was collected after approximately 20 min (Fig. S1). Using the optimized conditions described above, the radiochemical yield of the final formulated [ 18 F]1 product was 11.1 ± 3.5% (n = 14, decay corrected) within a synthesis time of 60 min, including HPLC purification. The specific activity of [ 18 F]1 at the end of the synthesis ranged from 104 to 154 GBq/μ mol. The identity was confirmed by co-injection with authentic compound using analytical HPLC (Fig. S2).
Log D determination and stability of [ 18 F]CB251 in human serum. The ability of [ 18 F]CB251 to cross the blood-brain barrier (BBB) by passive diffusion was assessed by measuring the partition coefficient (log D), whose experimental value was 3.00 ± 0.03. This result suggested that [ 18 F]CB251 can readily cross the BBB and   22 . In addition, the in vitro stability of [ 18 F]CB251 in human serum demonstrated that greater than 99% of the radioactive parent remained after 2 h (Fig. S3).
In vitro binding affinity. CB251 was next assayed for its binding affinity to TSPO and CBR by measuring its ability to displace [ 3 H]PK 11195 and [ 3 H]flunitrazepam, respectively 24 . As shown in Table 2, the affinities for TSPO and CBR, expressed as inhibition constants (K i ), were compared with those of unlabelled PK 11195, PBR28 and flunitrazepam. CB251 is characterized by a subnanomolar binding affinity for TSPO (K i = 0.27 ± 0.09 nM), which was 5 and 22 times higher than that observed for PK 11195 (K i = 1.38 ± 0.42 nM) and PBR28 (K i = 6.1 ± 6.4 nM) 25 , respectively. Furthermore, binding to CBR was below the detection limit, indicating the high selectivity of CB251 for TSPO.

Ex vivo tissue distribution. Purified [ 18 F]CB251 was reconstituted in a 10% ethanol-saline solution.
Animals were killed at various times after tail-vein injection of [ 18 F]CB251 (0.74 MBq/0.2 mL per mouse) and the radioactivity concentrations were measured in various tissues upon dissection. The results of the tissue biodistribution experiment are shown in Table 3. [ 18 F]CB251 accumulated strongly in TSPO-enriched tissues such as the lung, heart and kidney, whereas it exhibited comparatively low uptake in tissues such as the liver and stomach. Additionally, the spleen and adrenal glands, which are particularly rich in TSPO 26 , exhibited high radioactivity (25.11 ± 2.16 and 22.60 ± 5.00%ID/g at 15 min post injection). [ 18 F]CB251 also showed faster blood clearance at early time points (4.39 ± 0.78%ID/g at 5 min and 0.60 ± 0.12%ID/g at 60 min post injection). The brain uptake of [ 18 F]CB251 at 5 min post injection was 2.89 ± 0.23%ID/g, followed by a gradual reduction to significantly lower levels (0.53 ± 0.23%ID/g at 120 min post injection). The radioactivity concentration in the femur remained low for 120 min (1.19 ± 0.50%ID/g at 5 min post injection; 1.49 ± 0.80%ID/g at 120 min post injection).  Figure 3 shows the time-courses of [ 18 F]CB251 and [ 11 C]PBR28 distribution in the ipsilateral (right) and contralateral (left) striatum of the rat brain. Both radioligands had similar pharmacokinetic profiles, but the uptake of [ 11 C]PBR28 in the ipsilateral region was slightly higher than that of [ 18 F]CB251 during the PET scan time. Summed microPET images showed significant [ 18 F]CB251 binding in the ipsilateral striatum in the same inflamed rat brain (Fig. 4).

PET study in LPS-induced neuroinflammation and tumour models.
Additionally, there was no significant uptake in the skull. In the blocking experiment with consecutive injection     Immunohistochemical study. Histological examinations of dissected brain and tumour tissues were performed in all of the LPS-induced neuroinflammation rats and U87-MG-bearing mice used for the microPET scans to investigate microglial activation by CD68 staining. The selective PET neuroinflammation images of [ 18 F] CB251 in the ipsilateral area showed a high density of CD68-positive activated microglia compared with the contralateral area (Fig. 7A). In addition, treatment with a CD68 antibody strongly revealed positive immunostaining inside the human glioblastoma U87-MG tumour (Fig. 7B).

Discussion
The aim of this study was to synthesize and evaluate a fluorine-18 labelled analogue of alpidem, namely, [ 18 F] CB251, for in vivo imaging of TSPO in both neuroinflammation and TSPO-rich tumour models. Our initial strategy was to synthesize various potent and selective PBR ligands mainly characterized by various substituents at 6-and 8-positions of the imidazo[1,2-a]pyridine nucleus and several molecules in spired by the novel  11,18,19 . As an extension of our efforts to develop new TSPO-selective PET radiotracers, we have generated the first 18 F-labelled ligand, [ 18 F]CB251, from the 6,8-di-substituted imidazo[1,2-a]pyridin-N,N-dipropylacetamide class and evaluated its performance in vivo. The first step of this study was the synthesis of CB251, which was accomplished with good yields following the well-known synthetic procedures outlined in Fig. 2. As predicted, CB251, which is characterized by the presence of a chlorine atom at the 8-position of the imidazo[1,2-a]pyridine nucleus, displays high affinity and selectivity towards TSPO (K i = 0.27 ± 0.09 nM).
Fluorine-18 labelling of CB251 was performed by nucleophilic aliphatic substitution of the corresponding tosylate 5 (precursor for labelling) using [ 18 F]fluoride (Fig. 2). Various conditions were explored for the preparation of the TSPO ligand [ 18 F]CB251. Above all, when the precursor 5 for labelling was reacted with nBu 4 N[ 18 F] F in aprotic solvents (i.e., CH 3 CN, DMSO or DMF) under general anhydrous radiofluorination conditions, the yield of the desired product was somewhat low (Table 1, entries 1-3). In fact, after radiofluorination, the amount of unreacted precursor 5 was over 90% according to TLC. Given that these low radiochemical yields might be due to the use of aprotic polar solvents, the use of tert-butanol as the reaction solvent increased the yield (Table 1, entry 4) and extending the reaction time did not improve the reaction yield (Table 1, entry 5). In contrast, increasing the concentration of nBu 4 NHCO 3 from 0.8 to 2.0 equivalents relative to the precursor 5 positively affected the radiochemical yield (Table 1, entry 6). By changing the base from nBu 4 NHCO 3 to K 2.2.2 /K 2 CO 3 , a different pattern was observed in our experiment. Indeed, even though a reasonable radiochemical yield was achieved using K 2.2.2 / K 2 CO 3 ( Table 1, entry 9) as the base, suddenly increasing the equivalents of K 2 CO 3 reduced the reaction yield (Table 1, entry 10). Consequently, [ 18 F]CB251 was synthesized in the desired yields by employing a tert-butanol solvent, namely, 4.0 equivalents of nBu 4 NHCO 3 relative to the precursor 5, at 120 °C for 10 min (Table 1, entry 7). After pre-purification on a Sep-Pak cartridge and purification via HPLC, [ 18 F]CB251 was formulated and ready for use, with a 10% EtOH/saline solution and an 11.1 ± 3.5% radiochemical yield (decay corrected for 60 min). In addition, the measured log D value, which predicted good in vitro BBB permeability, together with the high stability in human serum indicated that [ 18 F]CB251 was sufficiently stable for further use in vivo.
Tissue distribution studies in healthy mice confirmed that [ 18 F]CB251 had a significantly high uptake (> 22%ID/g) in TSPO-enriched tissues, as well as a rapid blood clearance. In the brain, which is the principal target tissue for neuroinflammation, the initial uptake of [ 18 F]CB251 showed a desirable level of 2.89 ± 0.23%ID/g and its radioactivity was rapidly washed out from the brain at 30 and 60 min post injection (1.39 ± 0.13 and 0.63 ± 0.24%ID/g, respectively). However, the low uptake of radioactivity in the femur (< 2%ID/g) indicated negligible defluorination of the radioligand and reflected the high in vitro stability recorded for [ 18 F]CB251 in human serum. Therefore, [ 18 F]CB251 was expected to be sufficiently stable for successive in vivo PET imaging studies. In microPET imaging studies, the accuracies of [ 18 F]CB251 and [ 11 C]PBR28 were determined by comparing the imaging data. Although [ 18 F]CB251 showed a slightly lower uptake in the ipsilateral region compared with [ 11 C]PBR28, it rapidly approached the highest target-to-background ratio at early imaging times and selectively accumulated in the ipsilateral striatum. As a consequence, the uptake ratio of [ 18 F]CB251 between the ipsilateral area and the contralateral area was similar to that of [ 11 Fig. 4.
To confirm the TSPO selectivity and specificity of [ 18 F]CB251, displacement studies were performed in the LPS-induced inflammation rat model by injecting an excess of flumazenil at 15 min post [ 18 F]CB251 injection, which is the time at which the [ 18 F]CB251 uptake displays a steady state in the ipsilateral region. Flumazenil, which selectively binds to CBR, did not prevent the uptake of [ 18 F]CB251 in the ipsilateral region. However, uptake in the ipsilateral area was significantly decreased following PBR28 injection. These displacement studies indicate that the uptake of the major fraction of [ 18 F]CB251 in the inflammatory lesion was selectively and specifically mediated by TSPO, as shown in Fig. 5.
In the TSPO-rich cancer imaging study, the PET image of [ 18 F]CB251 visually reflected the ex vivo biodistribution of 18 F uptake in mice. In fact, [ 18 F]CB251 showed high uptake in TSPO-rich organs after injection, including the lung, heart and kidneys and only minor radioactivity was excreted in the urine (Fig. 6). The high kidney uptake and slow renal clearance might be explained by the strong TSPO binding affinity of [ 18 F]CB251 in these organs, which are characterized by high TSPO expression levels. The pronounced liver uptake is probably correlated with the lipophilic nature of CB251. In particular, [ 18 F]CB251 displayed reasonable tumour uptake in the human glioblastoma U87-MG xenograft model (1.96 ± 0.11%ID/g at 1 h post injection) and was comparable to accumulation of other TSPO ligands 27 . TSPO expression in tumour sections as well as in LPS-induced inflammation brain sections was determined by CD68 immunohistochemical staining (Fig. 7). In particular, the PET neuroinflammation images of [ 18 F]CB251 in the ipsilateral region showed a high density of CD68-positive activated microglia compared with the contralateral area (Fig. 7A). Furthermore, treatment with a CD68 antibody revealed strong positive immunostaining inside the human glioblastoma U87-MG tumours (Fig. 7B). The latter results, together with those obtained on in vitro TSPO binding, in vivo tissue distribution and PET imaging studies with specific and selective binding for TSPO, support the conclusion that [ 18 F]CB251 shows favourable properties as a new TSPO-selective PET radiotracer compared with [ 11 C]PBR28. To our knowledge, ours is the first example of an 18 F-labelled TSPO PET imaging agent, namely, [ 18 F]CB251, which is characterized by a 6,8-di-substituted imidazo[1,2-a]pyridine acetamide structure. [ 18 F]CB251 was prepared with a satisfactory radiochemical yield and in tissue biodistribution experiments in mice, [ 18 F]CB251 showed high accumulation in TSPO-enriched tissues such as the lung, heart, kidney, spleen and adrenal glands, whereas it exhibited low uptake in the liver and stomach. In the PET study, [ 18 F]CB251 produced a clearly visible image of the inflammatory lesion with the highest target-to-background ratio at early imaging times in the brains of LPS-induced neuroinflammation model rats and displayed reasonable tumour uptake in the U87-MG xenograft model. These results suggest that [ 18 F]CB251 is a promising TSPO PET imaging agent for neuroinflammation and TSPO-rich cancers in rodent models. It will be of interest to investigate the translation of these findings to other in vivo models and humans. The imidazopyridine alpidem binds both human and rodent TSPO with similar affinity 28 , although little data exists regarding the binding modes of this alpidem analogue to TSPO harbouring the natural Ala147Thr polymorphism. The recently published TSPO structure may help support further such structural investigations 29-31 .

Radiochemical synthesis of [ 18 F]1 ([ 18 F]CB251
). The radiotracer, [ 18 F]CB251, was synthesized in a single-step radiolabelling procedure by nucleophilic aliphatic substitution using tosylate precursor 5 with fluorine-18. Briefly, 18 F was obtained by the 18 O(p,n) 18 F nuclear reaction from [ 18 O]water. Fluorine-18 was isolated from the enriched water by trapping in a Chromafix-HCO 3 cartridge (pre-activated with 2 mL of ethanol and 5 mL of water) and then eluted with methanol:water (1:0.1 mL) dissolved TBAHCO 3 or K 2.2.2. /K 2 CO 3 . The eluted solution with fluorine-18 was dried by azeotropic distillation with acetonitrile (0.3 mL) under a nitrogen stream (× 2) and the tosylate precursor 5 (3.0 mg) in various reaction solvents (0.4 mL), such as CH 3 CN, DMF, DMSO or tert-butanol, was subsequently added. The reaction mixture was heated to the desired temperature for 10 or 30 min. The reaction mixture was cooled to room temperature and was then diluted with 10 mL of water. This solution was loaded into a C18 plus Sep-Pak cartridge, washed with 10 mL of water and eluted with 1.5 mL of CH 3 CN. After dilution with 1.5 mL of water, the combined solution was separated by a semi-preparative HPLC system using a UV detector at 254 nm and a gamma-ray detector. The product fraction was collected after approximately 20 min (Fig. S1). The fraction of [ 18 F]1 collected from the HPLC system was diluted with 20 mL of water. The solution was exchanged to 10% EtOH/saline solution by a C18 plus Sep-Pak cartridge to remove the HPLC solvent.
Log D determination. The log D value was measured four times by mixing a solution of [ 18 F]CB251 (approximately 0.74 MBq) or [ 11 C]PBR28 (approximately 1.5 MBq) in 5% ethanol:saline (10 μ L) with sodium phosphate buffer (0.15 M, pH 7.4, 5.0 mL) and n-octanol (5.0 mL) in a test tube. After vortexing for 1 min, each tube was then stored for 3 min at room temperature and the phases were separated. Samples of each phase (100 μ L) were counted for radioactivity. Log D is expressed as the logarithm of the ratio of the counts from n-octanol versus that of the sodium phosphate buffer.

Stability of [ 18 F]CB251 in human serum.
The in vitro stability of [ 18 F]CB251 was assayed by monitoring both the radio-TLC and the radio-HPLC profiles to determine its radiochemical purity. To assess the in vitro serum stability, 100 μ L of [ 18 F]CB251 in 10% EtOH/saline was incubated with 0.5 mL of human serum from a healthy volunteer for 2 h at 37 °C and the solution was analysed at 0, 10, 30, 60 and 120 min by a radio-TLC scanner using MeOH-CH 2 Cl 2 (5:95) as the developing solvent. Separately, 100 μ L of [ 18 F]CB251 in 10% EtOH/saline was also incubated with 0.9 mL of human serum from a healthy volunteer for 2 h at 37 °C. Aliquots (200 μ L) were taken from the sample at time points of 0, 30, 60 and 120 min and mixed with absolute acetonitrile (200 μ L). The mixture was centrifuged (3500 rpm, 5 min) to precipitate out serum proteins. The resulting supernatant including the authentic CB251 (1.0 mg/2.0 mL of CH 3 CN, 20 μ L) was filtered through a 0.45 μ m GH Polypro (GHP) membrane Disc filter before HPLC analysis to determine the fraction of intact radiotracer. Samples were analysed by HPLC (Waters, Xterra RP-18, 5 μ m, 4.6 × 250 mm; eluant: 60% CH 3 CN/H 2 O; flow rate: 1 mL/min; guard column 10 × 10 mm).
In vitro TSPO binding assays. Binding affinity and selectivity to the 18-kDa translocator protein TSPO and to CBR were assessed using in vitro receptor binding assays. These experiments were carried out as described by Denora et al 34 . Ex vivo tissue distribution in normal mice. Tissue distribution studies were performed in normal male ICR mice (7 weeks, 25-30 g) with tail-vein administration of [ 18 F]CB251 (0.74 MBq/mouse) dissolved in 10% ethanol in saline to a final volume of 200 μ L. The animals (n = 5 for each time point) were euthanised with diethyl ether at 5, 15, 30, 60 and 120 min post injection. After the mice were euthanised, blood and other organs such as the heart, liver, lung, stomach, spleen, kidney, intestine, muscle, brain, adrenal and femur were removed, weighed and counted for radioactivity in a NaI (Tl) well counter (Wizard TM 3", PerkinElmer). The injected dose was calculated from standards prepared from the injection solution. The accumulated radioactivity of all tissues is presented as the percentage of the injected dose per gram of tissue (%ID/g). All animal studies were performed with the approval of the Institutional Animal Care and Use Committee of the Seoul National University Bundang Hospital (SNUBH) and the methods were carried out in accordance with the approved guidelines.
LPS-induced neuroinflammation rat model. Male Sprague-Dawley rats weighing 200-250 g were used for the brain inflammation model and neuroinflammation was induced using lipopolysaccharide (LPS; Sigma-Aldrich, St. Louis, MO, U.S.A.) as previously described 22 . In all rats used for sequential PET scans, microglia activation between the ipsilateral and contralateral areas was determined by CD68 immunohistochemical staining.
Human glioblastoma U87-MG xenograft model. U87-MG cells (ATCC) were cultured in Dulbecco's modified Eagle's medium supplemented with 4 mM L-glutamine, 4,500 mg/L glucose, 1 mM sodium pyruvate, 1.5 g/mL sodium bicarbonate, 10% FBS and antibiotic-antimycotic (Gibco, Grand Island, NY, USA) at 37 °C in a humidified atmosphere of 5% CO 2 . U87-MG cells were cultured for 2-3 days without changing the medium and were harvested with trypsin-EDTA in PBS. Male Balb/c nu/nu mice, obtained at 6-7 weeks of age (Orient Bio Inc., Seongnam, Korea), were subcutaneously injected in the right shoulder with 5 × 10 6 U87-MG cells suspended in 100 μ L of 100 mM PBS (pH 7.4). Animal protocols were approved by the Seoul National University Bundang Hospital Animal Care and Use Committee and the methods were carried out in accordance with the approved guidelines. We also performed calliper measurements of the longest perpendicular tumour diameters.
Small animal PET imaging study. Five LPS-induced neuroinflammation rats (219.2 ± 14.1 g) were used for brain PET imaging at 4 days post-LPS injection. Before the scan, the rats were anesthetized with 2% isoflurane in a 7:3 mixture of N 2 /O 2 . PET imaging of the rat brain was performed in a dedicated small animal PET/CT (NanoPET/CT, Bioscan, Inc., U.S.A.) with a 10-cm axial field-of-view (FOV) and a 12-cm transaxial FOV. PET acquisition in the list-mode was concomitantly started with the intravenous injection of either [ 11 C]PBR28 or [ 18 F]CB251 and was performed for 90 min, followed by a CT scan of the head or body for attenuation correction. The first exam was performed with [ 11 C]PBR28 (approximately 18.5 MBq) and the second exam followed 3.5 h later with [ 18 F]CB251 (approximately 18.5 MBq) in same rat models (Fig. 8). To determine the TSPO selectivity and TSPO specificity of the PET images, flumazenil (5 mg/kg intravenously) was administered at 15 min post [ 18 F]CB251 injection via jugular catheter. Subsequently, PBR28 (5 mg/kg intravenously) was injected 30 min later via jugular catheter in the same LPS-induced neuroinflammation rat model. Three human glioblastoma U87-MG-bearing mice were used for TSPO-rich cancer PET imaging. PET scans of the mice, anesthetized with 2% isoflurane, were performed with a 9.3-cm axial, 8.4-cm trans-axial FOV scanner. At the end of each study, the list-mode data were sorted into a dynamic scan consisting of 54 frames for 90 min. The acquired images were reconstructed by the 3-D Adjoints Monte Carlo method, which included scatter and random corrections. A region of interest (ROI) with a 2-mm radius was delineated in the ipsilateral lesioned striatum by the intensely visualized region in the summed image of all the frames. Regional uptake of radioactivity was decay-corrected to the injection time and was expressed as the standardized uptake value (SUV), which was normalized to the injected radioactivity and animal body weight. Analysis of the PET images was performed using PMOD software v.3.1 (PMOD Technologies Ltd., Zurich, Switzerland). The time activity curve (TAC) representing the variation in radioligand concentration according to the time-course was estimated for the ipsilateral striatum. The binding potential of the specifically bound radioligand relative to the non-displaceable radioligand in tissue (BP ND ) for both [ 11 C]PBR28 and [ 18 F]CB251 on the ipsilateral striatum (right) was calculated with Logan plot with reference tissue input model using the contralateral striatum (left) as the reference region 35,36 . Immunohistochemical staining. Immunohistochemical staining was performed using serial sections of brain and tumours that had undergone PET/CT imaging. Briefly, the isolated brains and tumours were embedded in a tissue-freezing medium (Triangle Biomedical Sciences, Durham, NC, USA), frozen and consecutively cryosectioned in 10-μ m segments using a Cryocut Microtome (CM3050S, Leica, Solms, Germany). The tissue sections were thaw-mounted onto silane-coated microscope slides (Muto Pure Chemicals Co., Tokyo, Japan), dried in an aeration room and stored at − 80 °C until use. The tissue sections were rinsed with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde for 20 min at room temperature. After an additional series of washes with PBS, the sections were cleared with 3% sodium deoxycholate solution for 2 h at room temperature, blocked with 20% normal goat serum in 1% BSA-PBS for 2 h at 37 °C and incubated with the primary antibodies at 4 °C overnight (> 17 h). A total of 100 μ L DAPI solution (1 μ g/mL) was added to completely cover the section and incubated for 5 min at room temperature. Specific antibodies against macrophage marker CD68 (1:150, Abd Serotec, MCA1957) were used on the brain and tumour sections. Subsequently, the slides were rinsed three times with PBS and incubated with secondary antibodies (1:400, Alexa Fluor 488 labelled, Life Technologies, Carlsbad, CA, USA) for 2 h at 4 °C. Then, the slides were washed several times with PBS, counterstained with Hoechst 33342 dye (1:750, Life Technologies) and mounted with Prolong Gold antifade reagent (Life Technologies). Images were captured with the use of a confocal microscope (A1, Leica, Solms, Germany). Statistical analysis. The data are expressed as the mean ± standard error of the mean (SEM) and were analysed using the Mann-Whitney U-test. Statistical significance was defined as p < 0.05. Statistical analyses were performed using SPSS 12 software (SPSS, Inc., Chicago, IL, USA).