1. General Information
All reagents were purchased from TCI Development Co., Ltd (Shanghai). and J & K Scientific (Beijing, China) unless otherwise indicated. Thin layer chromatography (TLC) was performed with silica gel layers, and compounds were visualized under UV light. The 1H NMR (300 MHz) spectra of all compounds were acquired on an Advance (Bruker) spectrometer; chemical shifts (δ) for the proton resonance were reported in parts per million (ppm) downfield from TMS (δ = 0). The analysis of mass spectra under electrospray ionization and purity of precursors were determined by a LC-MS instrument (1200/6120, Agilent Technologies Inc.) with a C18 column (4.6 × 150 mm 5uM; VP-ODS, Shimadzu) at 0.5 ml per minute (ml/min) flow rate. The mobile phase consisted of 60% methanol and 40% H2O containing 1‰ formic acid.
18F− ions were received from a cyclotron (Cyclone 18 Twin, IBA, Belgium), situated at the Molecular Imaging institute, Jiangsu Huayi Technology Co., Ltd., by proton irradiation of 18O-enriched water. For automatic synthesis of MVBF, we used a remote-controlled radiolabeling module (RNplus, Synthra) with slight modifications and created the sequence program, based on manual trials. The scheme of modified RNplus module is presented in Fig. 2. We used six reagent supply vials (A1-A5, B1) at the upper part and two reaction vials (vial I and vial II) at the bottom part.
Analytical HPLC (1260, Agilent Technologies Inc.) with the same type column mentioned above was employed for MVBF characterization and identification. The signal acquisition system consisted of a UV detector (254 nm) and a radio-detector (1IINaI/PMT, Lablogic, USA) in series. The flow rate was 0.8 ml/min and the mobile phase consisted of methanol and H2O containing 0.05% triethylamine and 50 mM ammonium acetate. The percentage of methanol/H2O changed with running time: 0–15 minutes (mins), 15%/85%; 15–25 min, methanol increasing to 100% while H2O decreasing to 0%; 25–30 min, methanol 100%.
A micro-PET/CT equipment (Inveon; Siemens Co., USA) was used for the imaging of mice and marmosets, the body temperature was maintained at 37 °C using a heat pad.
2. Synthesis of Cold Standard Sample of MVBF as Well as Precursors (5) and (6).
We synthesized cold standard sample of MVBF as well as precursors (5) and (6) according to the work of Cline JK, et al., with minor modifications (Supplementary Fig. 1). The purity of the standard sample is > 99.9%, and the purity of two precursors > 99%, identified via LC-MS (Supplementary Fig. 2). The details of the synthetic route were described in the Supplementary text.
3. Automated Radiosynthesis Of Mvbf
We adopted a two-step synthesis route (Fig. 1). The scheme of automated synthesis was shown in Fig. 2. Reagents were added into supply vials as follows: A1: 1.1 ml eluent (3.08 mg KHCO3, 11 mg Kryptofix 2.2.2, 0.88 mL MeCN, 0.22 mL H2O); A2: 1 ml MeCN; A3: 5 mg precursor (6) in 0.5 ml MeCN; A4: 1 ml MeCN; A5: 0.5 ml MeCN; B1: 0.5 ml H2O. 5 mg precursor (5) powder was also added into reaction vial II beforehand. When the first step synthesis finished, the intermediate product [18F]-compound (7) was transferred from reaction vial I to reaction vial II via distillation. The whole automated synthesis duration was 100 min. The details of synthesis and purification were described in the Supplementary text.
4. Characteration And Quality Control Of Mvbf
Radiochemical yield (RCY, decay-correction to the end of bombardment) and radiochemical concentration (RCC) were measured by the radioactivity calibrator (CRC-55tR, CAPINTEC, INC., USA). The identity of MVBF was determined by co-injecting final product with cold standard sample into analytical HPLC. Radiochemical purity (RCP) and specific radioactivity (SA) were calculated by means of the area under curve (AUC) of radio-signals and UV-signals of final product in analytical HPLC, respectively. Bacteria and endotoxin detections were carried out by means of anaerobic/aerobic bacteria media and Limulus reagent gel methods, respectively, according to Chinese Pharmacopoeia.
5. In Vitro Stability
MVBF solution was stored at RT and injected into analytical HPLC for evaluating RCP and peak shape at 0 hour (h), 2 h, 4 h, 6 h, 8 h, and 10 h, respectively after synthesized.
6. Thiamine Deficiency Mouse Model And Marmosets
Eight-week-old male C57BL/6 mice and Institute of Cancer Research (ICR) mice (obtained from the SLAC Laboratory Animal Company, China) were housed in a controlled environment at temperature of 20–26 oC and humidity of 40–70% with free access to food and water. The mice were randomly divided into two groups: TD group (n = 2 for C57BL/6, one died due to anesthesia during micro-PET/CT scanning; n = 3 for ICR), which was fed thiamine-depleted diet (Trophic Animal Feed High-tech Co., Ltd., China), and control group (n = 3 for each strain), which was fed general diet. Twenty-eight days later, all mice received PET/CT scanning.
Four marmosets aged 3.1–10.8 years old (supplied by Jiuting Non-human Primate Facility, Chinese Academy of Sciences, Shanghai) were employed for micro-PET/CT scanning. At the day of experiment, the marmosets were fetched from the facility.
After scanning, the mice were sacrificed and the marmosets were sent back to the facility. The blood of the animals was collected for measuring thiamine levels.
7. Dynamical Micro-PET/CT Images and Evaluation of Cerebral Thiamine Metabolism Status
Micro-PET/CT imaging using MVBF as a tracer for evaluating cerebral thiamine metabolism status was performed in TD mice model and control mice, as well as in marmosets. The animals were anesthetized by inhalation of 1.5%-2% isoflurane in air (1.5 L/min) and received CT scan for acquiring structure image and attenuation correction data. Then, the mice were injected with 7.4–14.8 MBq MVBF in 0.1 ml volume (diluted by normal saline) through the tail vein and immediately scanned dynamically for 90 min with an energy window of 350-650KeV and a time window of 3.438 ns. A total of 35 frames were setup: 20f, 3 s; 4f, 60 s; 5f, 300 s; 6f, 600 s. Dynamic images were reconstructed by OSEM3D/SP-MAP algorithm with two iterations. The marmosets were injected with 46.3–74.0 MBq MVBF in 0.5–0.8 ml volume through the femoral vein and immediately scanned dynamically for 60 min. A total of 18 frames was setup: 6f, 10 s; 4f, 60 s; 5f, 300 s; 6f, 600 s. The other conditions for marmosets were the same as for mice.
Regions of interest (ROIs) were drawn manually over the whole brain (for mice and marmosets) and in the left ventricular cavity (for marmosets) based on the PET/CT co-registered images using IRW 4.2 software (Siemens Medical Solutions USA, Inc.). Radioactivity was expressed as standard uptake value (SUV): (ROI radioactivity/ROI volume)/(injected radioactivity/gram of body weight). The time-activity curve (TAC) and AUC (SUV*mins) were also calculated.
The TACs of the marmosets blood (Radioactivity was expressed as SUV) were taken as input functions (IF) [14–16] for fitting Patlak plots [17, 18], in order to analyze transfer constants (Ki) of brains in marmosets (IRW 4.2 software). The details of Patlak model was described in the supplemental text.
8. Measurement of Thiamine, TMP, and TDP in Whole Blood Samples of Mice and Marmosets
Thiamine, TMP, and TDP levels in whole blood samples were measured using HPLC, based on the established mothed  with slight modification. Briefly, blood samples were collected using heparin-anticoagulated tubes, 150 ul sample was vibrated for 30 s with equal volume 5.7% (for mice) or 5.2% (for marmosets) perchloric acid (PCA) added dropwise for deproteinization. Then, the mixture was stored at -80oC until assay within one month. The mixture was centrifuged at 12000 rpm for 8 min at 4 oC, the supernatant was pipetted. Thiamine, TMP, and TDP in supernatant were derivatized into thiochromes using potassium ferricyanide and analyzed by gradient elusion with C18 reversed-phase analytical column (250 × 4.6 mm). The derivatives were identified by HPLC fluoroscopy (1100, Agilent Technologies Inc., ex: 367 nm, em: 435 nm). The thiamine, TMP, and TDP levels were quantified using standard samples (Sigma-Aldrich, St. Louis, MO). The analyzers were blinded to samples information.
9. Pharmacokinetic and Metabolic Kinetics Studies in Liver and Kidney of Mice
Nine-week-old male ICR mice (n = 5, SLAC Laboratory Animal Company) were dynamically scanned using micro-PET/CT for 60 min. A total of 18 frames was reconstructed: 6f, 10 s; 4f, 60 s; 5f, 300 s; 6f, 600 s. The housing environment and the other scanning conditions were the same as mentioned above. ROIs of the liver and renal parenchyma as well as the left ventricular cavity were manually drawn.
For Pharmacokinetic Study, the TACs of the blood were fitted. The radioactivity was evaluated as %ID/g (the percentage of injected dose per gram of blood). Pharmacokinetics parameters were counted through the software PKSolver (version 2.0, China Pharmaceutical University) .
For metabolic kinetics study, SUV, TAC, AUC, maximum radioactivity (Cmax), and time to Cmax (Tmax) were calculated. The TACs of the blood (Radioactivity expressed as SUV) were taken as IFs for fitting Logan plots [20, 21], in order to analyze the distribution volumes (VD) of MVBF in liver and kidney, respectively (IRW 4.2 software). The details of Logan model were described in the supplemental text.
10. Biodistribution Study
The biodistribution of MVBF was studied on ICR mice (n = 36 in total; 18 males, nine-weeks-old, 33.5 ± 4.0 g; 18 females, seven-week-old, 27.7 ± 5.0 g). MVBF solution was diluted to 37 MBq/ml, and for each mouse, 0.1 ml MVBF was injected into the tail vein under isoflurane anesthesia. The mice were sacrificed at 2 min, 10 min, 30 min, 1 h, 2 h, and 4 h after administration (3 males and 3 females for each time point). The tissues of heart, liver, spleen, lung, kidney, stomach, duodenum, pancreas, femur, muscle (from thigh), blood, brain, fat, and gonad (ovary or testicle) were harvested, weighted, and measured for radioactivity by γ-counter. %ID/g was calculated referring to the counts of standard samples.
11. In Vivo Stability and the Renal Excretion Rate
After metabolic kinetics study and waking from anesthesia, the urine samples from three ICR mice were collected and measured in the radioactivity calibrator. The duration from MVBF injection to urine collection was about 85 min. Then, 0.1 ml urine for each mouse was added in an Eppendorf tube, vibrated for 30 s with equal volume PCA added dropwise for deproteinization. After centrifuging at 12000 rpm for 8 min at 4 oC, the supernatant was filtered and analyzed using HPLC.
12. Statistical Analysis
For the continuous data, mean ± SEM was applied for statistical description. Student t test was employed to compare the AUC values between TD mice and controls in ICR strain. The Pearson correlation was utilized to analyse the correlation between the cerebral accumulation of MVBF and the blood levels of thiamine, TMP, and TDP in marmosets. Repeated measurement of ANOVA with Trukey’s post-hoc was used to analyse the AUC values of MVBF between brain, liver, and kidney in ICR mice. All statistical analyses were performed using SPSS (Statistical Package for the Social Sciences) software (version 22.0; SPSS Inc., Chicago IL).