The synthesis of phenylboronic ester precursor was carried out as previously described [10]. Enantiomerically pure (R)-YH168 was synthesized in analogy to (R)-YH132, and the corresponding synthetic route is presented in the supporting information (Supplemental Scheme 1). Chiral separation by supercritical fluid chromatography (SFC) (ChiralPak IH, 5 μm, 250 × 20 mm, 45% MeOH in CO2) was performed on the final racemic product to obtain (R)-YH168 in 98.8% ee. 1H NMR (400 MHz, CDCl3) δ 7.75–7.54 (m, 4 H), 7.49 (s, 1H), 7.42–7.29 (m, 3 H), 7.16 (d, J = 7.3 Hz, 1H), 7.05 (s, 1H), 6.50 (s, 1H), 4.26–3.66 (m, 6 H), 3.38 (s, 1H), 3.01–2.47 (m, 6 H), 2.42 (s, 3 H). HRMS (ESI) calculated for C26H28N3O3+ [M + H]+, 430.2125 m/z; found, 430.2120 m/z.

The half-maximal inhibitory concentration (IC50) of (R)-YH168 was measured using human, cynomolgus monkey or mouse MAGL protein according to the published procedure [11]. Microsomal clearance using commercially available pooled liver microsomes (C57BL/6J mice) and parallel artificial membrane permeability assay (PAMPA) were conducted as previously reported [12].

[11C]CH3I was bubbled into the reaction vial with 0.5 mg precursor, 0.5 mg [Pd2(dba)3], 0.5 mg P(o-tolyl)3, and 0.6 mg K2CO3 in the mixture of DMF/H2O (v/v = 35/4, 390 µL). The reaction was heated at 65 °C for 4 min. After dilution with 0.1% H3PO4 in H2O/MeCN (v/v = 1/1, 1.7 mL), the resulting mixture was loaded to a semi-preparative HPLC for purification (Phenomenex Gemini C18 column, 250 mm × 10 mm, 5 μm, 110 Å, mobile phase A: 0.1% H3PO4 in H2O, mobile phase B: MeCN, gradient with 0–6 min, 10–35% B; 6–8 min, 35–45% B; 8–12 min, 45% B; 12–18 min, 45–50% B; 18–20 min, 50–95% B with a flow of 4 mL/min). The fraction corresponding to the product was collected, diluted with 8 mL of water, and passed through a pre-conditioned C18 light SepPak cartridge (Waters, WAT023501), followed by rinsing with 5 mL of water. The product was eluted off the cartridge with 0.5 mL of EtOH, and formulated with phosphate-buffered saline (9.5 mL, Gibco) to give a neutralized solution (pH = 7.4). The identity of the tracer was confirmed by co-injection with the reference compound using the Agilent 1100 series HPLC system equipped with a UV detector, and a GabiStar radiodetector (Raytest) (ACE XDB-C18 Zobrax column, 75 mm×4.6 mm, 3.5 μm, mobile phase A: 0.1% H3PO4 in H2O, mobile phase B: MeCN, gradient with 0.0–3.0 min, 5 − 30% B; 3.0–4.0 min, 30 − 40% B; 4.0–6.0 min, 40 − 50% B; 6.0–9.0 min, 50 − 95% B with a flow of 1 mL/min, UV detection at 254 nm). The enantiomeric purity of the final product was evaluated by a chiral column (ChiralCel OD, 5 μm, 250 mm×4.6 mm, mobile phase: hepatane/0.5%TEA IsoPrOH = 1/1, v/v with a flow of 2 mL/min). Representative HPLC chromatograms using the chiral column are presented in Supplementary Fig. 2.

For the assessment of plasma stability, (R)-[11C]YH168 (10 µL, 1‒2 MBq) was added to mouse plasma (300 µL). The mixture was incubated at 37 °C with gentle shaking. At 5, 30 and 60 min, 100 µL of the sample were taken out and mixed with ice-cold acetonitrile (200 µL). After centrifugation, the supernatants were filtered and analysed by HPLC equipped with column-switching system. Briefly, the sample was transferred to a pre-column (ReproSil-Pur, 120 ODS-3, 10 μm, 20 × 4.6 mm, Dr. Maisch GmbH, Germany) using 1% MeCN in H2O (1 mL/min, 0–4 min), and subsequently eluted and analyzed using Luna C18 column (5 μm, 250 × 4.6 mm, Phenomenex Inc., Germany) with 55% MeCN in H2O (1 mL/min, 4–15 min) as mobile phase. The UV and radioactive signals were detected by 220 nm diode array detector L-2450 (Hitachi High-Technologies, Japan) and FlowStar, LB 513 (Berthold Technologies GmbH & Co. KG, Germany), respectively. The data were corrected for physical decay, integrated and analyzed using the EZ Chrome Elite Software Package (Version 3.3.1, Agilent Technologies Inc., United States). In vitro autoradiography and free fraction measurement in mouse plasma were conducted as previously reported [10].

MAGL knockout (n = 3) and wild type (n = 3) mice were administrated with 8.02‒14.6 MBq (R)-[11C]YH168 (5.78‒10.71 nmol/kg) via tail vein injection for PET imaging. Data acquisition and reconstruction were performed as previously described [12]. The time-activity curves were generated with predefined volumes of interest using an MRI T2 template by PMOD software (version 4.201; PMOD Technologies, Fällanden, Switzerland).

For in vivo metabolite analysis, (R)-[11C]YH168 was injected intravenously to wild-type mice, and the animals were sacrificed at 5 and 30 min post-injection. After decapitation, blood was rapidly collected in a tube (BD Vacutainer, LH Lithium heparin) and mixed. The blood samples were centrifuged at 5000 RCF for 3 min at 4 °C to separate the plasma. The supernatant was collected, mixed with equal volume of cold acetonitrile and centrifuged at 5000 RCF for 3 min at 4 °C for deproteinization. The brain was dissected and homogenized using a polytron (PT 2100) in PBS and cold acetonitrile was added (v/v = 1/1 in the final mixture). The homogenate was vortexed and centrifuged at 4800 rpm for 5 min at 4 °C. All supernatants were aspirated with a syringe and filtered through a 0.45 μm filter unit (Whatman, SPARTAN 13/0.45 RC). The resulting samples were analyzed by the column-switching HPLC system as mentioned above. The in vitro stability of (R)-[11C]YH168 in brain homogenates was further examined by co-incubating the tracer with freshly prepared brain tissue at 37 °C. Samples were collected at 5 and 30 min after incubation and analyzed as previously described.

Two scans were performed in one monkey, a baseline scan to evaluate the kinetic properties, and a blocking scan with MAGL inhibitor PF-06795071 at 0.1 mg/kg dose to evaluate the binding specificity. The radiotracer or blocking drug was given as a 3 min bolus injection by an infusion pump, and a dynamic PET scan was conducted on the Focus 220 scanner (Siemens Medical Solutions, Knoxville TN USA) as previously reported [13]. The data acquisition lasted for 120 min. Arterial blood samples were collected at preselected time points, separated and measured in a gamma counter (Wizard 1480/2480, PerkinElmer, Waltham, MA, USA) to obtain the radioactive uptake in whole blood and plasma overtime. Heart rate, blood pressure, respirations, SpO2, electrocardiogram, end-tidal CO2, and body temperature of the animal were continuously monitored during the scan.

For metabolite analysis, arterial blood samples at 0, 3-, 8-, 15-, 30-, 60- and 90-min post-injection were centrifuged at 3900 g at 4 °C for 5 min to separate the plasma. The supernatant was collected, mixed with 8 M urea to denature plasma proteins and passed through a 1.0 μm filter (Whatman 13 mm CD/X). The filtrate was then analyzed by HPLC with a column-switching system. Self-packed column (4.6 × 19 mm with C18 sorbent) and Phenomenex Luna C18 (2) (4.6 × 250 mm, 5 μm) were used for capture and analysis, respectively. The mobile phase consisted of 55% MeCN in 45% 0.1 M ammonium formate (pH = 6.4, v/v) at a flow rate of 1.4 mL/min. The eluent fractions were collected with an automated fraction collector. The radioactivity in the whole blood, plasma, filtered plasma-urea mix, filter and collected fractions were measured in a gamma counter. The arterial plasma input function (AIF) was calculated and corrected by the percentages of parent tracer, filtration efficiency and the ratio of plasma/whole blood.

Data reconstruction was carried out using a Fourier rebinning and filtered back projection algorithm. High-resolution magnetic resonance image (Siemens 3T Trio) was registered to the PET images and used as a template to draw regions of interest (ROIs). The time activity curves (TACs) of ROIs overtime were generated. Regional TACs were fitted into one-tissue and two-tissue compartment (1TC, 2TC) models [14]. Akaike information criterion (AIC) was used to evaluate the goodness-of-fits of three models [15].

Regional distribution volume (VT, mL/cm3) was calculated from kinetic analysis. The occupancy plot was used to obtain MAGL occupancy by PF-06795071 (0.1 mg/kg) and non-displaceable volume of distribution (VND) of (R)-[11C]YH168 [16]. Non-displaceable binding potential (BPND) was calculated from regional VT under control conditions and VND, i.e. BPND = (VT, ROI - VND)/ VND [17].