in vivo brain region-specific TSPO imaging by PET/MRI [ 18 F]FEPPA

1 Background: Expression of translocator protein (TSPO) on the outer mitochondrial 2 membrane of activated microglia is strongly associated with neuroinflammation. The 3 second-generation PET ligand [ 18 F]FEPPA specifically binds TSPO to enable in vivo 4 visualization and quantification of neuroinflammation. We optimized an fully 5 automated radiosynthesis method and evaluated the utility of [ 18 F]FEPPA, the second- 6 generation PET ligand specifically binds TSPO, in a mouse model of systemic LPS 7 challenge to detect TSPO-associated signals of central and peripheral inflammation. In 8 vivo dynamic PET/MR imaging was performed in LPS-induced and control mice after 9 [ 18 F]FEPPA administration. The relationship between the [ 18 F]FEPPA signal and the 10 dose of LPS was assessed. The cytokine levels (i.e. TNF-α, Il-1β, Il-6) in LPS-induced 11 mice were measured by RT-PCR. Standard uptake value (SUV), total volume of 12 distribution (VT) and area under the curve (AUC) were determined based on the 13 metabolite-uncorrected plasma input function. Western blotting and immunostaining 14 were used to measure TSPO expression in the brain. ± 0.1, 1.25 ± 0.12, and 1.58 ± 0.09-fold higher in LPS-injected mice than controls. TNF-α, Il-1β and Il-6 mRNA levels were also elevated in the brains of LPS-injected 2 mice. Western blotting revealed TSPO ( p<0.05 ) and Iba-1 ( p<0.01 ) were upregulated 3 in the brain after LPS administration. In LPS-injected mice, TSPO immunoactivity 4 colocalized with Iba-1 in the cerebrum and TSPO was significantly overexpressed in 5 the hippocampus and cerebellum. The peripheral organs (heart, lung) of LPS-injected 6 mice had higher [ 18 F]FEPPA signal-to-noise ratios than control mice. 7 Conclusions: Based on the robust data on ligand specificity and selectivity in both 8 central and peripheral tissues using 7T PET/MR imaging, we demonstrate that the high 9 affinity, stability and high-contrast visualization indicate detection of TSPO using 10 [ 18 F]FEPPA represents a promising, specific biomarker for early diagnosis and 11 neuropathological follow-up of neuroinflammatory processes. [ 18 F]FEPPA rapidly distributed throughout various brain regions and reached a steady-state, enabled detection of higher regional-specific TSPO expression in the brains of mice injected with LPS. These [ 18 F]FEPPA PET/MRI imaging findings confirmed by

The 18-kDa mitochondrial translocator protein (TSPO), originally named peripheral-2 type benzodiazepine receptor (PBR), plays important roles in several physiological 3 processes, including steroidogenesis (1) (2), inflammation (3) and cell proliferation (4). 4 TSPO is endogenously expressed at high levels in some peripheral tissues, including 5 the adrenal gland, heart, lung, kidney and testis (3). However, TSPO is expressed at 6 very low levels in the central nervous system (CNS) under normal conditions, and 7 limited to glial cells (astrocytes and microglia) (3) (5) (6). Neuroinflammation is characterized by the activation of neuroimmune cells and 9 implicated in the pathogenesis of several neurodegenerative diseases. Glial cells are 10 activated in response to brain injury or neuroinflammation, and activated 11 microglial/macrophages are associated with dramatically increased expression of 12 TSPO (7). High expression of TSPO in activated microglia has been reported in several 13 neurodegenerative diseases, including Alzheimer's disease (8), Huntington's disease 14 (9), ischemic stroke (10), multiple sclerosis (11) and epilepsy (12), which indicates 15 TSPO plays essential roles in the progression of these diseases. Therefore, TSPO is 16 12 (Bioscan, Inc., Washington, DC, USA). A delay time of 12 s was observed between the 1 two detectors. In vivo PET/MRI imaging and imaging analysis 19 13 Before PET/MR imaging acquisition, the mice were anesthetized by passive inhalation 1 of a mixture of isoflurane and oxygen (5% isoflurane for induction and 2% for 2 maintenance). A fiber-optic temperature probe was inserted into the rectum to monitor 3 the core temperature of the mice, then each animal was gently positioned in the MR 4 cradle and the MRI coil was secured around the head. A warm water blanket was 5 positioned to maintain the core temperature of the animals, then the mice were 6 positioned inside the scanner. The depth of anesthesia, pulse and respiration were 7 monitored constantly during the imaging procedure; in the unlikely event that an animal 8 regained consciousness, the scanning was immediately stopped and the animal was 9 removed from the scanner, humanely euthanized and eliminated from the study. 10 The animals (n = 6 per dose of LPS; n = 6 controls) were injected with [ 18 F]FEPPA 11 (11.1 MBq; 0.3 mCi) via the tail vein. Dynamic PET images were obtained using a 12 small animal SuperArgus 2r PET (SEDECAL, Madrid, Spain) or 7T PETMR Inline 13 (Bruker, Rheinstetten, Germany) for 30 min with the energy window set to 350-650 14 keV. Images were acquired every 1 s for 10 images, 10 s for five images, 60 s for nine 15 images, 300 s for two images, and 600 s for 1 image; a total of 27 frames were collected. 16 T1 and T2 MRI were performed to determine the anatomical structure of the brain. The 17 14 images (TR = 3455 ms, TE = 36 ms, matrix = 256 × 256, average = 8, slice number = 1

30). 2
The PET images were reconstructed through three-dimensional ordered-subset 3 expectation maximization. The regional radioactivity concentration (KBq/cc) of 4 [ 18 F]FEPPA was estimated based on the mean pixel values within the volumes of 5 interest (VOI) corresponding to MR images of various regions of the brain. Image data 6 were decay-corrected to the injection time. The radioactivity concentration (KBq/cc) of 7 the VOI was converted to the standard uptake value (SUV) and the mean and standard 8 error of the mean (SEM) radiotracer accumulation values were calculated for various 9 tissues. PET/MR data were analyzed using PMOD 3.7 software (PMOD Technologies 10 Ltd., Zurich, Switzerland). 11 12

Quantification of Dynamic PET Imaging Using Logan Graphical Analyses 13
The Logan graphical method for reversible uptake (40) was used to assess whether 14 [ 18 F]FEPPA PET/CT (30) or PET/MRI could be used to detect differences in the 15 expression of TSPO. Cardiac blood was used as the reference tissue. The slope of the 16 linear portion of the Logan plot represents the total distribution (VT). If metabolite-17 corrected plasma TACs are not available, a reference region TAC, Cp(t), can be used 18 15 instead of the plasma TAC. Then, the slope of the linear portion of the plot can be 1 calculated using (Eq. 1) 2 3 4 Logan graphical analyses were performed using PMOD 3.7 software. 5 6

Real-time reverse transcription-polymerase chain reaction 7
To assess the transcript levels of the genes encoding the cytokines TNF-α, Il-1β, and Il-8 6 in the brain following LPS-induced systemic inflammation, total RNA was isolated 9 from the brains of the mice after imaging at the indicated time points after LPS injection. 10 RNA was isolated using the RNeasy Micro Kit (Qiagen, Valencia, CA, USA) and cDNA 11 was prepared using the Super Script III First-Strand Synthesis System Kit (Invitrogen, 12 Carlsbad, CA, USA). RT-qPCR was performed for a total of 40 cycles using a Rotor-13 Gene Q cycler (Qiagen, Hilden, Germany), according to the manufacturer's protocol 14 using TaqMan gene expression master mix, primers and MGB probe sets (Applied 15 Biosystems, Foster City, CA, USA). The assay IDs for the PCR primer pairs and 16 TaqMan MGB probes were Mm00443260_gl (Tnf-α), Mm00434228_ml (Il-1β), 17 Mm00446190_ml (Il-6) and Mm00607939Msl (β-actin). Relative expression levels 18 were calculated using the comparative threshold cycle (Ct value) method and 1 normalized to the ΔCt of β-actin. All analyses were conducted in triplicate. 2 3

Western blotting 4
After imaging, brain tissues were lysed in ice-cold radioimmunoprecipitation assay 5 buffer and total protein lysates were fractionated by sodium dodecyl sulfate-6 polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride 7 membranes. The membranes were probed with antibodies against Iba-1, PBR (Cell 8 Signaling Technology, Danvers, MA, USA) and β-actin (Novus Biologicals, Centennial, 9 CO, USA). After washing three times for 10 min in Tris-buffered saline supplemented 10 with 0.1% Tween 20 (TBST), the membranes were incubated with horseradish 11 peroxidase-conjugated secondary antibodies for 1 h, washed with TBST and the signals 12 were visualized and quantified using the GeneGnome Chemiluminescence Imaging 13 System (Syngene, Frederick, MD, USA). 14 15

Immunohistochemistry and immunofluorescence 16
After imaging, the mice were terminally anesthetized, perfused with 4% 17 paraformaldehyde, the brains were dissected, post-fixed overnight in 4% 18 paraformaldehyde at 4 C, embedded in optimal cutting temperature compound, and 19 frozen coronal sections (30-μm thick) were prepared. The sections were permeabilized 1 with 1% Triton X-100 in phosphate-buffered saline (PBS) for 10 min, blocked with 1% 2 bovine serum albumin in PBS-T for 1 h, incubated with antibodies against Iba-1 3 (GeneTex; to detect microglia) and PBR (GeneTex). The sections were incubated with 4 3'-diaminobenzidine for 20 sec or fluorescein isothiocyanate-conjugated donkey anti-5 rabbit antibody (1:200) at room temperature for 1 h, and counterstained with 4',6-6 diamidino-2-phenylindole for 5 min at room temperature. 7 Paraffin-embedded sections were cut at 5-µm thickness, deparaffinized and rehydrated. 8 Antigen retrieval was carried out by microwaving in 10 mM citrate buffer (pH 6.0) at 9 100 ºC for 10 min. Sections were washed, incubated in 3% hydrogen peroxide for 15 10 min at room temperature, incubated in blocking solution for 60 min at room temperature,  regions of interest (ROI) were manually selected and the intensity of immunostaining 19 was measured using the software. The median (25%, 75% interquartile range) 1 percentage score for each group was calculated as the average of all mice in each group. 2 3

Statistical analysis 4
Data are expressed as mean ± SEM. One-way ANOVA with the post-hoc Bonferroni 5 or unpaired Student's t-test were used for statistical evaluation. P < 0.05 was considered 6 statistically significant. Statistical analyses were performed using GraphPad Prism 8 7 (GraphPad Software, La Jolla, CA, USA).

Quality control and stability tests of [ 18 F]FEPPA 17
The retention time (Rt) of [ 18 F]FEPPA in semi-preparative HPLC was 13.07 min ( Fig.  18 2A, radio-peak). The radioactive product was also co-injected with an authentic 19 FEPPA standard. The retention time (Rt) of [ 18 F]FEPPA (Fig. 2B, radio-peak) in 1 HPLC analysis was 6.8 min, which was consistent with that of authentic FEPPA (Fig.  2

2C, UV peak). 3
One typical production batch of [ 18 F]FEPPA was assessed using all of the quality 4 control tests required for human use. In this production, the levels of residual solvents 5 and K2.2.2 were below the levels set by the Taiwan FDA and the values for all other 6 tests all met the specifications for human use. 1-2 min post-injection, was also observed. The concentration of radioactivity in heart 1 muscle peaked in the first 3 min after injection, and then gradually decreased over time 2 (Fig. 3B) The 3D PET/MRI brain images shown in Figure 4 illustrate the [ 18 F]FEPPA signals in 7 the control group and LPS groups at 24 h. As shown in Figure 5, significantly higher 8 [ 18 F]FEPPA uptake was observed in all brain regions of the LPS group at 24 h 9 compared to the control group. The highest uptake was detected in the hypothalamus, 10 followed by the midbrain, hippocampus, thalamus, brain stem, whole brain, cortex, 11 striatum, cerebellum and amygdala. In control animals, the regional [ 18 F]FEPPA time-12 activity curves indicated radioactivity peaked during the first 3 min, followed by rapid 13 washout. In contrast, [ 18 F]FEPPA accumulated more rapidly in the brain regions of the 14 LPS group in the first 2-3 min, and then continued to progressively increase to reach a 15 steady state. The brain regions of the LPS group exhibited significantly higher SUV, VT and AUC 1 values compared to the control mice. Generally, the [ 18 F]FEPPA SUV were 2 significantly different between control and LPS animals from the first 5 min onwards, 3 and the average SUV ratio for all brain regions between the two groups was 1.61 ± 0.1 4 (Fig. 6A). 5 PET images were quantified using Logan graphical analysis. The VT values for 6 [ 18 F]FEPPA PET in the majority of brain regions were significantly higher in the LPS 7 group than the controls, except for the hypothalamus, brain stem and cerebellum. On 8 average, the [ 18 F]FEPPA VT values were 1.25 ± 0.12-fold higher in the LPS group than 9 the control group (Fig. 6B).

Quantitative analysis of LPS dose response and brain distribution 1
To assess whether the [ 18 F]FEPPA uptake dose-response correlates with LPS exposure, 2 we quantified [ 18 F]FEPPA uptake in various brain regions after administration of 3 different doses of LPS. The time-activity curves for various brain regions after injection 4 of LPS are presented in Figure 7. There was no significant correlation between 5 [ 18 F]FEPPA uptake and the dose of LPS at concentrations up to 2.5 mg/kg; though the 6 SUV values significantly increased at 5 mg/kg LPS (Fig. 8). 7 8

LPS stimulates proinflammatory cytokine secretion 9
To investigate the proinflammatory reaction induced by LPS, we measured the mRNA 10 expression levels of the proinflammatory cytokines TNF-α, IL-1β and IL-6 in brain 11 homogenates at 24 h after injection of LPS. As shown in Figure 9A, the mRNAs 12 encoding TNF-α and IL-1β were expressed at higher levels in the brain homogenates 13 of the LPS-group than the control group (LPS 2.5 mg/kg vs. control, P < 0.01; LPS 5 14 mg/kg vs. control, P < 0.001), demonstrating that injection of LPS induced 15 inflammation and the release of proinflammatory cytokines. Thus, we next examined 16 the dose-dependence of cytokine expression and release. 17

18
Upregulation of TSPO expression after LPS induction in mice 19 23 The TSPO band was detected at 18 kDa in Western blotting (Fig. 9B). A marked 1 increase in TSPO expression was observed in the brain 24 h after administration of LPS 2 (P < 0.05); Iba-1 expression also increased in the LPS group at this time point (P < 0.01; 3 TSPO activity was also analyzed in the brain cortex, hippocampus and cerebellum at 9 24 h after LPS injection (Fig. 10A). TSPO immunoreactivity was significantly higher 10 in the CA1, CA3 and cerebellum at 24 h after administration of LPS, whereas TSPO 11 expression was similar in the CA2 and dentate gyrus between the control and LPS 12 groups. A tendency towards higher expression of TSPO was observed in the cortex of 13 LPS mice; however, this trend was not significant (Fig. 10B). 14 Cerebral expression of TSPO colocalized with the microglial marker Iba-1 in mice 15 injected with LPS (Fig. 10C). Both TSPO and Iba-1 expression were upregulated at 24 16 h after LPS injection. final products of both manual and automated synthesis were stable, maintaining a 1 radiochemical purity of more than 99% at 6 h after the end of synthesis. The process is 2 also economical, as no disposable components are required. Quality control analysis 3 confirmed the identity, strength, quality and purity of the radiocompound, and the 4 stability of the product is sufficient for both animal and clinical studies. 5 Mice were systemically injected with LPS (5 mg/kg, i.p.) to evaluate the optimized 6  However, PET imaging enables assessment of biological disease processes in the heart 19 or lungs, whereas magnetic resonance (MR) scanning provides detailed anatomic 1 imaging and tissue characterization. In this study, advanced 7T PET/MR imaging was 2 used to identify the heart muscle and detect significant [ 18 F]FEPPA uptake in the heart 3 muscle 24 h after LPS challenge compared to controls (Table 1, p = 0.05). 4 In vivo imaging of TSPO in the peripheral tissues can also help to detect inflammation, whereas repeat measurements using other invasive methods (e.g., lung biopsy and 1 bronchoalveolar lavage) are unacceptable. The in vivo profiles for [ 18 F]FEPPA reached 2 a steady state in the heart muscle and lungs 10 min after injection of [ 18 F]FEPPA, and 3 then gradually decreased (Fig. 3). On the one hand, this results demonstrated that 4 [ 18 F]FEPPA significantly increased in peripheral organs or tissues (~3 folds in heart 5 and lung) in the LPS-inducted group as compared to control mice; but on other hand, 6 the fast metabolic rate in plasma (60% metabolite in 30 min imaging time-window) (30) 7 could increase the difficulty of the quantification accuracy of PET imaging and limit 8 its application for the peripheral inflammation. 9 The brain exhibited relatively lower [ 18 F]FEPPA uptake compared to peripheral organs 10 such as the heart, lung or kidney. Western blotting, real-time PCR and IHC analyses of 11 brain tissue sections for TSPO confirmed the PET/MR imaging detection of 12 significantly higher accumulation of [ 18 F]FEPPA in the brains of mice injected with 13 LPS. To investigate whether the specific binding of [ 18 F]FEPPA in the brain was due 14 to LPS-induced microglial activation, we used western blotting to detect increased co-15 expression of TSPO and Iba-1 at 24 h after injection of LPS (Fig. 9, p = 0.01 and p =  16 0.05, respectively). 17 Regional-specific accumulation of [ 18 F]FEPPA in the brain reflected the changes in 18 TSPO expression. TSPO ligands have been reported to exert beneficial effects as anti-19 29 inflammatory drugs in experimental models of various neurodegenerative diseases and 1 anxiety disorders (55). More than a dozen anti-inflammatory drugs have been 2 developed (55), (56). However, the in vivo pharmacokinetic and pharmacodynamics of 3 those anti-inflammatory drugs in specific regions of the brain remain unclear, due to 4 the lack of quantitative or repeatable imaging agents, as described above. In the current 5 study, we used PET/MR-based image segmentation to calculate regional-specific 6 [ 18 F]FEPPA accumulation. At 24 h after injection of LPS, the PET SUV images 7 revealed significant increases in [ 18 F]FEPPA uptake in the majority of brain regions 8 assessed (Fig. 6A). A pharmacokinetic parameter, total distribution volume VT, 9 revealed a significant increase in [ 18 F]FEPPA distribution in the various brain regions 10 of LPS-injected mice compared to control mice (Fig. 6B). Moreover, the AUC0-30 min 11 value for [ 18 F]FEPPA increased by an average of 1.3-fold compared to the controls (Fig.  12   6C). These results are in good agreement with previous publications (30).