Inhibition of Thiamine Diphosphate-Dependent Enzymes by Triazole-Based Thiamine Analogues

Thiamine is metabolized into the coenzyme thiamine diphosphate (ThDP). Interrupting thiamine utilization leads to disease states. Oxythiamine, a thiamine analogue, is metabolized into oxythiamine diphosphate (OxThDP), which inhibits ThDP-dependent enzymes. Oxythiamine has been used to validate thiamine utilization as an anti-malarial drug target. However, high oxythiamine doses are needed in vivo because of its rapid clearance, and its potency decreases dramatically with thiamine levels. We report herein cell-permeable thiamine analogues possessing a triazole ring and a hydroxamate tail replacing the thiazolium ring and diphosphate groups of ThDP. We characterize their broad-spectrum competitive inhibition of ThDP-dependent enzymes and of Plasmodium falciparum proliferation. We demonstrate how the cellular thiamine-utilization pathway can be probed by using our compounds and oxythiamine in parallel.

S. cerevisiae PDC inhibitory activity assay. S. cerevisiae PDC was purchased from Sigma. Its activity was determined by monitoring DCPIP reduction at 600 nm using a microplate reader (CLARIOstar) and conducted as described 1,2 with some modifications. The percentage inhibition of compounds was assayed at a final concentration of 1500 µM. The reaction buffer (50 mM KH2PO4 and 1 mM MgCl2, pH 7) contained 300 µM ThDP (or 750 µM in the competition assay), 0.27 mM DCPIP, and 0.15 mg/ml S. cerevisiae PDC. The reaction mixture was preincubated at 37 o C for 60 min, then reaction was initiated by adding pyruvate to a final concentration of 70 mM. Specific activity was calculated using the molar extinction coefficient of DCPIP, 21 mM -1 cm -1 . 3 E. coli OGDHc E1 inhibitory activity assay. E. coli OGDHc E1 was from our previous work 5 and had been donated by R. Frank. Its activity was determined by monitoring DCPIP reduction at 600 nm using a microplate reader (CLARIOstar) and conducted as described 1,2 with some modifications. The percentage inhibition of compounds against E. coli OGDHc E1 was assayed at a final concentration of 250 µM. The reaction buffer (50 mM KH2PO4 and 2 mM MgCl2, pH 7) contained 50 µM ThDP (or 125 µM in the competition assay), 0.5 mM DCPIP, and 6.7 mg/ml E. coli OGDHc E1. The reaction mixture was preincubated at 37 o C for 60 min, then reaction was initiated by adding α-ketoglutarate to a final concentration of 10 mM. To determine the IC50, ThDP concentration was lowered to 30 µM, and inhibitor concentration was varied (2-1000 µM). Specific activity was calculated using the molar extinction coefficient of DCPIP, 21 mM -1 cm -1 . 3 The enzyme IC50 values were calculated as described for the PDHc E1 assay. KM(ThDP) was found to be 3 µM, consistent with the reported value. 5 A. viridans PO inhibitory activity assay. A. viridans PO and horseradish peroxidase were purchased from Sigma. A. viridans PO activity was determined by monitoring appearance of quinoneimine dye at 550 nm using a microplate reader (CLARIOstar) and conducted as described 1 with some modifications. The percentage inhibition of compounds against A. viridans PO was assayed at a final concentration of 250 µM. The reaction buffer (50 mM KH2PO4 and 10 mM MgCl2, pH 5.9) contained 50 µM ThDP (or 125 µM in the competition assay), 10 µM flavin adenine dinucleotide (FAD), 0.15% 4-Aminoantipyrine, 0.3% N-Ethyl-N-(2-hydroxy-3-sulfopropyl)-m-toluidine (EHSPT), 50 µg/mL horseradish peroxidase and 0.35 U/mL A. viridans PO. The reaction mixture was preincubated at 37 o C for 30 min, then reaction was initiated by adding pyruvate to a final concentration of 50 mM. To determine the IC50, inhibitor concentration was varied (2-1000 µM) with ThDP concentration at 50 µM. 1 unit of PO activity is defined as 1 μmol of hydrogen peroxide produced per minute. The enzyme IC50 values were calculated as described for the PDHc E1 assay. KM(ThDP) was found to be 5 µM.
Z. mobilis PDC inhibitory activity assay. Z. mobilis PDC was expressed and purified following a reported method. 6 Z. mobilis PDC activity was determined by monitoring reduced nicotinamide adenine dinucleotide (NADH) consumption at 340 nm using a microplate reader (CLARIOstar) and conducted as described 6 with some modifications. The reaction buffer (50 mM MES-KOH and 5 mM MgCl2, pH 6.5) contained 10 µM ThDP, 150 µM NADH, 10 U/ml alcohol dehydrogenase (ADH) and 0.5 µM of active sites of Z. mobilis PDC. To determine the IC50, inhibitor concentration was varied (0. 4-200 µM). The reaction mixture was preincubated at 37 o C for 60 min, then reaction was initiated by adding pyruvate to a final concentration of 10 mM. The enzyme IC50 values were calculated as described for the PDHc E1 assay. KM(ThDP) was found to be 0.35 µM, consistent with the reported values. 5,6 Computational calculation of molecular properties (Methods and Results) Many studies have provided guidelines 7 for drug design to predict the oral bioavailability of compounds:  Molecular weight (MW): ≤ 400  Log P: ≤ 4  HB donors (HBDs, i.e. no. of N-H and O-H bonds): ≤ 5  HB acceptors (HBAs, i.e. no. of N and O atoms): ≤ 10  Total polar surface area (TPSA) at pH = 7.4: ≤ 140 Å 2  Rotatable bonds (RBs): ≤ 10-12 Computational docking (Methods and Results) Docking of ThDP and compounds were executed using CCDC GOLD docking program with PDB: 6CFO, 1PVD, 1V5F and 6U3J for human PDHc E1, S. cerevisiae PDC, A. viridans PO and human OGDHc E1, respectively. The binding site of ThDP was selected as the docking site. Our molecules were generated using Mercury. GA runs were set at 20 and was user-defined with population size of 200 and 200000 number of operations. No early termination was permitted. Similarity and scaffold constraint to the original ligand were implemented on our compounds to mimic their binding positions. CHEMPLP and GoldScore were the docking scoring and rescoring respectively. 8 Interactions between docked compounds and protein models are shown using CCDC GOLD.
Compound series 21a-c and 22a-c: Figure S1. Docking of 22a into the ThDP pocket of human PDHc E1 with interactions shown; Mg 2+ shown as yellow-green sphere. Both oxygen atoms were predicted to interact with the Mg 2+ in a sixmembered cyclic form. Only one oxygen atom was predicted to interact with the Mg 2+ with 21a-c and 22b-c (not shown); this may explain their weaker binding. The interactions of the pyrimidine ring are not shown as they are conserved.
Similar binding modes are also present in the docking models of 22a in PDC, OGDHc E1 and PO, thus potentially explaining its inhibitory activities on PDC and OGDHc E1. Interestingly, 22a was found to be inactive on PO. As most of our triazole thiamine derivatives were either inactive or weakly active on PO regardless of the MBG identity, we suggest that the intrinsic affinity of the aminopyrimidine-CH2-triazole motif is low (relative to other three enzymes) and the affinity of the diphosphate is relatively more important. Therefore, the above binding mode is presumably present in the 22a-PO complex, but, at the concentration tested ([inhibitor]:[ThDP] = 5:1), 22a was not potent enough to outcompete ThDP.
Compound series 18a-c: Figure S2. Docking of 18c into the ThDP pockets as surface of human PDHc E1 (left) and human OGDHc E1 (right) with interactions shown; Mg 2+ shown as yellow-green sphere. Instead of the (possibly deprotonated) nitrogen atom, an oxygen of a S=O bond was predicted to interact with the Mg 2+ ; the sulphonamide moiety may pick up some random interactions in the diphosphate pocket. The linkers of 18b and 18c appeared not to be optimal to position a O atom for interaction with the Mg 2+ (not shown), and this may explain their lower affinities. The interactions of the aminopyrimidine are not shown as they are conserved.

S6
Compound series 26a-c: Figure S3. Docking of 26a into the ThDP-pocket of human PDHc E1, shown as surface; Mg 2+ shown as yellow-green sphere. The aromatic ring was predicted to form a cation-π interaction with Mg 2+ while the carbonyl oxygen and the benzamide amino group engage in polar interactions with the surrounding residues. These geometry sensitive interactions were distorted (not shown) with the longer 26b and 26c, and this may explain their lower affinities. The interactions of the aminopyrimidine are not shown as they are conserved.
Oxythiamine diphosphate (OxThDP) 7b:  cerevisiae PDC (D), human OGDHc E1 (E) and A. viridans PO (F); in each case the ThDP-pocket is shown as surface and Mg 2+ is shown as a yellow-green sphere. Collectively, these docking models support the ThDP-competitive nature of 24b which occupied the ThDP pockets with bidentate binding to the Mg 2+ (as summarised in Figure 2). In C, D, E and F, the interactions of the aminopyrimidine are not shown as they are conserved. human OGDHc E1 (E) and A. viridans PO (F); in each case the ThDP-pocket is shown as surface and Mg 2+ is shown as a yellow-green sphere. Collectively, these docking models support the ThDPcompetitive nature of 24b which occupied the ThDP pockets with bidentate binding to the Mg 2+ (as summarised in Figure 2). In C, D, E and F, the interactions of the aminopyrimidine are not shown as they are conserved.

S9
Compound series 15a-c, 16a-c and 17a-c: Docking studies of 17c with the PDC and PO enzymes (not shown) gave similar outcomes to the above with human OGDHc E1, suggesting that the lack of activity of 17c on all three enzymes is due to a steric clash between the diphosphate-binding pocket and the bulky p-methoxyphenyl terminus of 17c.

Cell-based assays (Results)
Anti-plasmodial activity assay. This study used human malaria parasite P. falciparum strain 3D7 (chloroquine-sensitive) and the same strain expressing an extra copy of TPK with a GFP-tag (PfTPK-GFP) generated as previously described. 9 The intraerythrocytic stage of the parasites were maintained essentially as previously described. 10 Compounds were tested at concentrations up to a highest final concentration depending on their solubility (between 25 µM and 200 µM). Compound stock solutions were prepared in dimethyl sulfoxide (DMSO) followed by dilution in RPMI 1640 medium in the absence of thiamine or in the presence of 2.97 µM (or 297 µM) thiamine. The final concentration of DMSO that the parasites were exposed to never exceeded 0.05%. Two-fold serial dilutions were then performed, with each concentration tested in triplicate. The assay was performed as described 11 with some modifications. Experiments were initiated with parasites in the ring-stage, a parasitemia level of 0.5% and a haematocrit of 2%. Chloroquine (0.5 µM) was used as the positive control (i.e. complete inhibition of parasite proliferation), and parasites maintained in the absence of any inhibitor represented 100% parasite proliferation. The final volume in each well was 200 µL. Plates were incubated at 37 °C, under an atmosphere of 96% nitrogen, 3% carbon dioxide and 1% oxygen. Parasite proliferation was measured by performing SYBR-Safe assay. 12 The concentration at which the compound suppresses parasite proliferation by 50% (i.e. parasite IC50) was determined from data calculated using Microsoft Excel and analysed by non-linear regression plot using GraphPad Prism. The data were averaged from three independent experiments.
Cytotoxicity assay. This study used HFF cells (human foreskin fibroblasts) as described 13 with some modifications. The HFF cells were seeded in 96-well plates at a density of about 13 x 10 4 cells/mL. Cycloheximide (10 µM) was used as a control to indicate complete inhibition of HFF cell proliferation.
Plates were incubated at 37 °C in a humidified 5% carbon dioxide incubator for 96 h. A sample of the supernatant (150 µL) was then carefully aspirated from each well and discarded. The plates were then stored at -80 °C. SYBR-Safe assay was used. The plates were thawed, SYBR-Safe lysis solution (150 µL) was added to each well and mixed via pipetting to ensure the HFF cells were detached from the plate and lysed. The plates were then processed as described for the anti-plasmodial assay. Figure S8. In vitro anti-plasmodial activity of the compounds against 3D7 parasites in thiamine-free medium (white circles), medium with 2.97 μM thiamine (black circles), and 297 μM thiamine (white triangles) under assay conditions as described above. Data are averaged from three independent experiments, each carried out in triplicate. Error bars represent SEM and, where not visible, are smaller than the symbols. Data presented for compounds 17c and 24b are shown in Figure 4 in the main text but are included here for completeness. Figure S9. In vitro cytotoxicity result of the compounds against HFF cells. Data are averaged from three independent experiments, each carried out in triplicate. Error bars represent SEM and, where not visible, are smaller than the symbols. Data presented for compounds 17c and 24b are shown in Figure  4 in the main text but are included here for completeness. Figure S10. In vitro anti-plasmodial activity of compounds 17b,c and 24b,c against P. falciparum 3D7 parasites transfected with an empty plasmid (black circles), and 3D7 parasites expressing PfTPK-GFP (white circles) in the thiamine-free medium. Data are averaged from three independent experiments, each carried out in triplicate. Error bars represent SEM and, where not visible, are smaller than the symbols. Data presented for compounds 17c and 24b are shown in Figure 4 in the main text but are included here for completeness.

General synthesis (Methods)
Oxygen-and moisture-sensitive reactions were carried out in flame-dried glassware under a nitrogen atmosphere. Unless otherwise stated, all chemicals and reagents were purchased from commercial suppliers and used without further purification. Reaction progress was monitored by analytical thinlayer chromatography (TLC). TLC was conducted using Merck glass plates with silica Kieselgel 60 F254 of thickness 0.25 mm and visualised under 254 nm UV lamp or potassium permanganate staining solution (with light heating). Flash column chromatography was carried out in the indicated solvent system using prepacked silica gel cartridges for use on the Biotage Purification System. All solvents were removed under reduced pressure using a Büchi rotary evaporator with dry ice traps.
All yields refer to chromatographically and spectroscopically pure compounds unless otherwise stated. Known compounds were characterised by, at minimum, 1 H NMR spectroscopy. New synthetic intermediates were characterised by, at minimum, 1 H NMR spectroscopy, 13 C NMR spectroscopy (except 10a-c) and ESI-MS unless otherwise stated. Compounds subjected to biological assays were characterised by, at minimum, 1 H NMR spectroscopy, 13 C NMR spectroscopy and HRMS. 1 H NMR spectra were recorded at 400 MHz or 700 MHz in CDCl3, CD3OD or CD3SOCD3 solution on a Bruker 400 MHz or 700MHz spectrometer and chemical shifts were recorded in parts per million (ppm). 13 C NMR spectra were recorded on either a Bruker 400 MHz or 700 MHz spectrometer. 19 F NMR spectra were recorded on either a Bruker 400 MHz or 700 MHz spectrometer. Resonances are described using the following abbreviations: s (singlet), d (doublet), t (triplet), q (quartet), quin.
(quintet), m (multiplet), br (broad), dd (doublet of doublets), etc. Coupling constants (J) are given in Hz and are rounded to the nearest 0.1 Hz. All NMR data were collected at 25 °C. Mass spectra used electrospray ionisation (ESI). Melting points of compounds were measured using a Reichert apparatus and are uncorrected.

Experimental procedures (Synthesis)
Preparation of the common precursor -5-(azidomethyl)-2-methylpyrimidin-4-amine 8 To a stirred solution of thiamine hydrochloride (10 g, 30 mmol) and NaN3 (5 g, 77 mmol) in water (100 mL, 0.3 M) was added Na2SO3 (0.4 g, 3.2 mmol). The resultant mixture was stirred at 70 °C overnight, then acidified with citric acid to pH 4-5, washed with DCM (200 mL), and basified with potassium carbonate to pH 8-10. Upon product precipitation, the suspension was cooled in an ice bath and filtered under reduced pressure. The residue was rinsed with cold water and dried under reduced pressure to yield azide 8 as a white solid. To increase the yield, the filtrate was extracted with EtOAc (100 mL x 2). The combined organic phases were dried over MgSO4, filtered, and evaporated under reduced pressure. The solid residues were pooled with the precipitate (collected upon filtration) and recrystallised from EtOAc-hexane and dried under reduced pressure to yield the product as a white solid (3.6 g, 75%).

S16
General procedure for preparation of 11a-c: To a stirred suspension of tosylate 10a-c (3 mmol, 1 equiv.) in dry DMF (1 M) under nitrogen at 0 o C was added NaN3 (2 equiv.). The resultant mixture was stirred at r.t. for 2 days. The reaction mixture was quenched with aqueous phosphate buffer (pH 7) and extracted with n-BuOH. The organic phase was dried over MgSO4, filtered, and evaporated under reduced pressure. The residue was purified by silica flash chromatography (10% MeOH in DCM) to yield azide 11a-c as a solid.
White solid (513 mg, 66%). 1  General procedure for preparation of 12a-c: A stirred solution of azide 11a-c (1 mmol, 1 equiv.) in MeOH (0.1 M) at r.t. was treated with 10% Pd/C (25 mg) under nitrogen. The flask was evacuated and flushed with hydrogen gas (three times). The resultant mixture was stirred vigorously at r.t. under an atmosphere of hydrogen (1 atm, H2 balloon) for 4 h. The reaction mixture was filtered through Celite and concentrated under reduced pressure to yield amine 12a-c as a solid, which was used in the next step without further purification.
General procedure for preparation of 21a-c: To a stirred solution of 3-methoxybenzoic acid (0.65 mmol, 1.3 equiv.) and DCC (3 equiv.) in dry DMF (0.2 M) under nitrogen at 0 °C was added DMAP (1.3 equiv.) and corresponding amine 12a-c (0.5 mmol, 1 equiv.). The resultant mixture was stirred at r.t. for 2 days, then diluted with DCM, filtered through cotton wool (to remove DCC/DCU), washed with aqueous phosphate buffer (pH 7), and the washings extracted with DCM. The organic phases were dried over MgSO4, filtered, and evaporated under reduced pressure. The residue was purified by silica flash chromatography (12% MeOH in DCM) to yield amide 21a-c as a solid.