Fluorinated Tracers and Chemicals
11-Dehydrodexamethasone was from Steraloids (Newport, Rhode Island, USA) and dexamethasone for pre-clinical studies from Alfa-Aesar (Heysham, UK). Dexamethasone tablets for clinical use were from Essential Generics (Egham, UK). 2-(Phenylsulfonyl)-1-(4-(trifluoromethyl)phenyl)ethanone (Figure 1c, compound c1a) was originally provided by Wyeth (now Pfizer, Sandwich, UK); subsequent batches were synthesised in-house as previously described 32 (Supplementary information). All other non-steroidal polyfluorinated keto compounds (Figure 2) were from Apollo Scientific Ltd (Stockport, UK), with the exception of methyl 6-(trifluoromethyl)-nicotinoyl acetate (c6a) and 4,4’‑bis(trifluoromethyl)benzophenone (c8a) which were from Marshalton Research Laboratories (King, North Carolina, USA). Corresponding hydroxyl metabolites were synthesised in house by reduction with sodium borohydride (Supporting information). The selective 11β-HSD1 inhibitor Merck544 (3-adamantan-1yl-6,7,8,9-tetrahydro-5H-[1,2,4]trazolo[4,3-a]azepine) 33 was from Enamine Ltd (Kiev, Ukraine). All other chemicals were from Sigma-Aldrich (Poole, UK) and used without further purification unless otherwise stated. Solvents (HPLC grade where appropriate) were from Sigma-Aldrich, Fisher Scientific (Loughborough, UK) or Rathburn (Walkerburn, UK).
Library of Polyfluorinated substrates for 11β-HSD1
Putative polyfluorinated non-steroidal keto tracers (Figure 3, compounds c1a to c12a) were selected by consideration of the published literature 34,35 and structural similarity with the library lead, compound c1a 32 and screened for inhibition of cortisone reductase activity. Tracers c1a to c12a (or vehicle) were added to HEK293 cells stably transfected with either human or rat Hsd11b1 and inhibition measured in duplicate using a scintillation proximity assay as described 36. Percentage inhibition was determined relative to vehicle control and IC50 determined using a four parameter logistic equation.
Structural characterisation of steroidal and polyfluorinated tracers
NMR spectra (1H, 13C and 19F) were acquired using an ARX250 Bruker BioSpin NMR spectrometer (Billerca, USA) and referenced to a solvent peak (CDCl3 or d4‑methanol). (Supplementary Figure S1).
Measurement of steroidal and polyfluorinated tracer ex vivo by 19F-MRS at 7 T (pre-clinical scanner) in phantoms
A 7 T pre-clinical scanner (Agilent Technologies, Yarnton, UK) fitted with a 400 mT/m gradient set and with a 30 mm diameter round surface coil for resonance frequency (RF) transmission and signal reception was used. The scanner was tuned to the RF of 19F (280 MHz). Fluorinated keto tracers (including 11-dehydrodexamethasone and c1a) and their hydroxy metabolites were scanned in glass vials (sample volumes 1 to 20 mL) in organic solvent (methanol or chloroform) and whole blood (20:1 in 0.5 M ethylenediaminetetraacetic acid). Care was taken so all the sample volume was contained within the sphere of detection. To determine the linearity of the relationship between signal intensity and concentration and the limit of detection, samples were prepared by serial dilution in chloroform (5 mL) or whole blood, the latter spiked with tracer stock solutions (8-10 mg/mL prepared in chloroform, chloroform/dimethylsulphoxide (DMSO) 9:1 or DMSO). A range of concentrations of 11-dehydrodexamethasone in 5 mL was scanned using 50 to 3200 repetitions per spectra (25, 100, 400 and 1600 s scan time). Scans were carried out at room temperature without stirring or spinning. When using blood as matrix, samples were occasionally shaken between scans. Simple pulse-acquire scans were acquired with TR of 0.5 s and with 50 to 3200 repetitions per spectrum, as indicated in Results.
Dexamethasone measurement in phantoms by 19F-MRS at 3 T (clinical scanner)
A 3 T Siemens Verio scanner (Siemens Healthineers, Erlangen, Germany) with circularly polarized 19F-tuned transmit-receive flexible surface coil (180 mm x 244 mm, Rapid Biomedical, GmbH, Rimpar, Germany) was used for scanning phantom solutions of dexamethasone and 11-dehydrodexamethasone prepared in methanol (20 mL in glass vials). Saline bottles were used to load the coil and the phantom was placed between the loading bottles and the coil. Spectra ranging from 64 to 640 averages were used to measure LODF and signal-to-noise versus tracer content using pulse-acquire scans. TR of 0.5, 1 and 1.5 s were tested and the shortest one that did not affect signal to noise used.
Dexamethasone measurement in human liver by 19F-MRS at 3 T (clinical scanner)
With local ethical approval (South East Scotland Research Ethics Committee 01) and written informed consent, three male healthy volunteers (ages 24, 26 and 32 years old) were recruited by advertisement. The study was performed in accordance with relevant guidelines/regulations, and in accordance with the Declaration of Helsinki. Inclusion criteria were age 16-60 years. Exclusion criteria were known hypersensitivity to dexamethasone or contraindications to MRI scanning; glucocorticoid therapy in the previous 3 months; diabetes mellitus; body mass index >40 kg/m2; alcohol intake >28 units/week; renal, thyroid or liver dysfunction on biochemical screening; history of dyspepsia or peptic ulcer disease; history of, or current treatment for, mental illness; pregnancy or lactation. Volunteers attended in the morning. A 20G IV cannula was inserted into the right antecubital fossa for serial venous sampling for ex-vivo measurement of dexamethasone and 11-dehydrodexamethasone (by liquid chromatography tandem mass spectrometry (LC-MS/MS)). Dexamethasone was administered orally (volunteer 1, 10 mg; volunteer 2, 12 mg; volunteer 3, 14 mg) with the aim of detecting a signal for dexamethasone in liver and determining the lowest dexamethasone dose which could be detected.
Volunteers were then scanned in the Verio 3 T scanner as for in vitro phantoms. The flexible surface coil was placed directly over the liver. A 90 degree excitation 100µs block pulse was used with a vector size of 128, and spectra obtained at isocentre with a second order semi-automated shim applied before acquisition. Pulse calibration was not possible and higher flip angles we tried empirically without any obvious increase in signal. No phase cycling was used, with an acquisition bandwidth of 10000 Hz. TR of 1500 ms and echo time (TE) of 0.15 ms were used with 400 repetitions (scan time 10 minutes) per spectrum. Anticipating a transient rise in hepatic dexamethasone concentrations within two hours post administration, six 10 minute scans were conducted at intervals from each participant between 30 and 100 minutes post dexamethasone administration. In two of the three subjects, a vial with dexamethasone solution in methanol was placed between the coil and patient liver to confirm coil operation and 19F-MRS signal detection prior to oral dexamethasone administration. The phantom was then removed, dexamethasone administered, and in vivo scanning of the liver commenced.
LC-MS/MS analysis of dexamethasone
Dexamethasone and 11-dehydrodexamethasone were measured by liquid chromatography tandem mass spectrometry (LC-MS/MS). Stock solutions of analytes and internal standard were prepared (10 µg/mL in methanol) and stored at -20°C. Standards were prepared on the day of analysis by serial dilution of stock solutions. A standard curve was prepared representing a concentration range of 0-300 ng/mL for all analytes. 4,6α,21,21-Dexamethasone (d4-dexamethasone, internal standard, 15ng; C/D/N/ Isotopes. Quebec, Canada) was added to plasma (200 µL) and standard curve samples. Calibration curves and samples were processed by solid-phase extraction (Sep-Pak® C18, 200 mg cartridges (Waters, Wilmslow, UK), conditioning with 5 mL methanol, equilibration 5 mL water followed by sample loading, washing with 5 mL water and analyte elution with 2mL methanol). Eluates were dried under nitrogen (60 °C), resuspended in water (200 µL) and then extracted with ethyl acetate (2 mL). The supernatant was dried and dissolved in mobile phase. Injection volume was 10 µL. Analysis was performed on a Waters Acquity™ UPLC with autosampler (10 °C), coupled to an AB Sciex QTRAP® 5500 mass spectrometer (Warrington, UK), and operated with Analyst Software version 1.5.1. Separation was achieved on a SunFire™ C18 column (150x4.6 mm, 5 µm; Waters) at 20 °C, with a linear gradient from 60:40 to 45:55 (acetonitrile with 0.1% formic acid (FA): water with 0.1%FA) at a flow rate of 1.5 mL/min with a total run-time of 6 minutes. Ionisation was performed in positive electrospray mode with curtain gas 20 psi, collision gas medium, source temperature 500 °C and source gas 40 psi. Mass transitions of protonated ions monitored were (Declustering Potential, DP; Collision Energy, CE; Cell Exit Potential, CXP): dexamethasone m/z 393→373 (DP 71; CE 11; CXP 16); 11-dehydro dexamethasone m/z 391→253 (DP 71; CE 27; CXP 12); d4-dexamethasone m/z 397→377 (DP 51; CE 11; CXP 16). Intra-assay precision and accuracy respectively of the assay were 10.1% and 13.1% for dexamethasone and 7.7% and 0.3% 11-dehydrodexamathasone in the relevant concentration range (n=6 replicates), with a lower limit of quantitation of 0.25 ng/mL, measured against a 1/x weighted calibration line.
Measurement of polyfluorinated tracer in perfused via rat liver by 19F-MRS at 7 T (pre-clinical scanner)
In an initial ex vivo experiment, 32.8 mg c1a was dissolved in 4 mL DMSO:PEG400 1:1 and diluted in Krebs buffer (1 L) to a final concentration of 100 µM before perfusion into an excised rat liver (further details below) via the portal vein at 25-30 mL/min for 30 minutes. Subsequent scanning was carried out with 800 repetition (400 s) per spectrum (11 scans acquired) with the vial containing excised liver placed inside the coil.
Measurement of polyfluorinated tracers in rat liver in vivo by 19F-MRS at 7 T (pre-clinical scanner)
Male Wistar rats (Harlan Olac, Bicester, UK), weighing between 415 and 515 g were studied under license from the UK Home Office and approved by the Animal Welfare and Ethical Review Board committee, University of Edinburgh. All procedures were performed under the UK Animals (Scientific Procedures) Act, 1986 and carried out in compliance with the ARRIVE guidelines. Rats were anesthetized with 4% isoflurane (Merial Animal Health Ltd., Harlow, UK), followed by 1.2-1.4% in oxygen enriched air for maintenance, and placed in an MRI compatible rat holder (Rapid Biomedical GmbH, Rimpar, Germany). To minimize the scanning time and maximize the signal‑to‑noise ratio of the 19F spectra, no localization was applied. Positioning of the animal and coil were checked by the acquisition of fast gradient echo scout images (TR 30 ms, TE 2 ms, FOV 50 x 50 mm, matrix 128 x 128, and slice thickness 3 mm) in 3 orthogonal orientations. The homogeneity of the magnetic field was optimized using the 1H signal before the coil was tuned to the resonance frequency of 19F (280 MHz). Untriggered 19F-MRS spectra using pulse-acquire sequence were acquired with following parameters: TR 0.5 s; 800 signal averages (400 s) per spectra unless otherwise indicated; spectral width 20000 Hz centred near the tracer signal. After scanning was completed, rats were killed by cervical dislocation while still under anesthesia.
Preliminary validation used a range of concentrations (10-20 mg/kg of c1a and 20 mg/kg of c1b with up to 800 s (1600 repetitions) per spectrum. 19F-MRS experiments to assess pharmacodynamics of tracer c1a and to test the effect of 11β‑HSD1 inhibition used c1a at 15 mg/kg (administered by gavage, dissolved in olive oil:PEG400:DMSO 95:5:5 as vehicle at 5 mg/mL). After overnight fast, rats were gavaged with inhibitor (Merck544, 5 mg/mL in vehicle) at 30 mg/kg (n=3) or vehicle (n=3, weight matched as with inhibitor) then anesthetized and positioned prone on top of the surface coil and scout images obtained as above. The animals were removed from the scanner and dosed with tracer c1a. The tracer was administered 35-40 minutes after Merck544. The animals were repositioned into the holder (taking care to avoid including stomach or gut within the coil inner detection volume), scout images re‑taken and the instrument switched to 19F-MRS mode. Spectra (400 s per spectrum) were acquired sequentially from 25 (±2) minutes until 83 (±2) minutes post c1a gavage.
Data analysis
7 T-MRS signal processing
19F-MRS spectra were processed using jMRUI 37 AMARES algorithm. For phantom work, line broadening of 20 or 50 Hz was applied as it gave more reliable signal intensity results.. For animal work, 50 Hz line broadening was applied unless stated otherwise. The offset spike was removed; when present, the highest isofluorane peak was used as a reference for signal position and isofluorane peaks were filtered out using the HLSVD algorithm before peak fitting and signal integration. If no isofluorane was present, a 0 ppm value was given to the centre of the spectrum. Overlapping tracer/metabolite 19F-MRS peaks were resolved by using soft constraints on signal position and line width. Prior knowledge parameters were based on the values obtained from 19F‑MRS experiments when only one of the signals was present.
Statistical Analysis
19F-MRS data are given as mean ± SEM and compared by repeated-measures ANOVA, with Fisher post-hoc tests as appropriate, with significance accepted at p<0.05.