Fragment screening reveals salicylic hydroxamic acid as an inhibitor of Trypanosoma brucei GPI GlcNAc-PI de-N-acetylase

Graphical abstract


Introduction
Trypanosoma brucei is a protozoan parasite that is transmitted by the bite of an infected tsetse fly and causes the fatal African sleeping sickness in humans and the related wasting disease Nagana in cattle. Current treatments are expensive, toxic, and difficult to administer, leaving an urgent need for improved therapeutic agents. The parasite is able to evade the host immune response by virtue of a dense surface coat composed of 5 Â 10 6 variant surface glycoprotein (VSG) dimers that prevent lysis by the innate immune response. The coat also undergoes antigenic variation to evade the adaptive immune response. 1 The VSG dimer is attached to the plasma membrane by a glycosylphosphatidylinositol (GPI) anchor, and GPI anchor biosynthesis has been genetically and chemically validated as essential for the survival of the clinically relevant bloodstream form of the parasite. [2][3][4][5] The GPI biosynthetic pathway has been extensively studied in both T. brucei and mammalian systems, highlighting differences in both the order of assembly and substrate specificity that can be exploited to produce species-specific inhibitors. The GlcNAc-PI de-N-acetylase is a zinc metalloenzyme that catalyses the second step in biosynthesis of GPI ( Fig. 1), 6 and has been genetically validated as essential in bloodstream form T. brucei. 3 The use of synthetic substrate analogues has revealed that the human enzyme is more fastidious than the trypanosome enzyme, enabling the design of species-specific substrate-based inhibitors. [7][8][9] The substrate analogue approach has shown that the phosphate, 2 0 -NHAc and 3 0 -OH of the substrate GlcNAc-PI are critical for recognition by the T. brucei GlcNAc-PI de-N-acetylase. 10 In contrast, the diacylglycerol is not directly recognised and may be efficiently replaced with an octadecyl chain. 9,11 Recently, we have shown that substrate-based inhibitors containing the zinc binding moieties hydroxamic acid or carboxylic acid can act as inhibitors of the T. brucei de-N-acetylase. 12,13 These substrate-based inhibitors do not possess drug-like physiochemical properties and contain too much stereochemical complexity for tractable synthesis. Here, we report our efforts to identify alternative scaffolds for GlcNAc-PI de-N-acetylase inhibitors that resulted in the identification of salicylic hydroxamic acid as an inhibitor with high ligand efficiency.

Trypanosome cell-free system assay
Trypanosome cell-free system assays, where the formation of GPI precursors is monitored by following the incorporation of [ 3 H]-mannose were analysed using high-performance liquid chromatography and fluorography as described previously. 13 2.4. Compound synthesis 2.4.1. Synthesis of 5-((tert-butoxycarbonyl)amino)-2hydroxybenzoic acid (9) 5-Aminosalicylic acid 8 (6.89 g, 45.0 mmol) was suspended in water (60 mL), and TEA (12.4 mL, 90.0 mmol) was added. A solution of Boc 2 O (10.8 g, 90.0 mmol) in dioxane (120 mL) was added, and the reaction was stirred at room temperature overnight. The solvent was removed, and the solid residue suspended in water (20 mL). HCl (3 M) was added until the greyish pink precipitate stopped forming. The precipitate was filtered off and washed with water. The precipitate was dissolved in boiling acetone and hot filtered before being recrystallized twice from acetone to afford the product (8.61 g, 76%) as a white powder, mp 278°C. 1

Fragment screening
To discover more drug-like scaffolds for zinc-binding de-N-acetylase inhibitors we screened a small focused library of zinc binding fragments. Because we have previously shown that the enzyme may be inhibited by substrate analogues containing hydroxamic acid, N-hydroxyurea or carboxylic acid groups, 12,13 we selected 26 commercially available fragments, which contained these zinc binding moieties, with an average molecular weight of 220 (Supplemental Table S1). We screened the inhibition of recombinant T. brucei de-N-acetylase by the fragments using a mass spectrometry-based activity assay to directly monitor the conversion of the synthetic substrate N-acetyl-D-glucosamine-a-(1-6)-D-myo-inositol-1-octadecyl phosphate (GlcNAc-IPC 18 ) to D-glucosamine-a-(1-6)-D-myo-inositol-1-octadecyl phosphate (GlcNAc-IPC 18 ). The replacement of the diacylglycerol moiety of GlcNAc-PI with a simple C 18 alkyl chain in GlcNAc-IPC 18 has no effect on the substrate recognition, and enables the use of a unique mass-transition to follow the reaction. 6 Initial screening of the fragments at 1 mM in duplicate revealed six compounds with >50% inhibition ( Fig. 2A + D). However, only one compound showed >50% inhibition at 100 lM (Fig. 2B). The most potent compound 1 (Fig. 2C) had an IC 50 of 63 ± 4 lM and a molecular weight of 153, giving an impressive ligand efficiency (-RT.lnIC 50 /N non-H atoms ) 16 of À0.57 kcal mol À1 per non-hydrogen atom, which is over one third of the theoretical maximum affinity for organic compounds of À1.5 kcal mol À1 per non-hydrogen atom. 17  The result of the fragment screen compared favourably to our screening of 16,037 lead-like compounds from the Dundee Drug Discovery Unit's compound collection 18 against the T. brucei de-N-acetylase in a cell-based functional complementation assay, i where no inhibitors were identified (data not shown). Interestingly, compound 1 (also known as SHAM) has previously been identified as an inhibitor of trypanosome alternative oxidase (TAO), and is toxic in vivo to the clinically relevant bloodstream form T. brucei. 19,20

Inhibition of the T. brucei cell-free system
The ability of 1 to act as an inhibitor of the T. brucei GPI pathway was confirmed using the trypanosome cell-free system (Fig. 3). Priming the cell-free system with GlcAc-IPC 18 in the presence of GDP-[ 3 H]-mannose stimulates the production of radiolabelled mannosylated GPI intermediates that can be separated by high performance thin-layer chromatography and visualised by fluorography. Priming the cell-free system with GlcNAc-IPC 18 (10 lM) was prevented by incubation with >300 lM of 1 (Fig. 3B), a five-fold increase over the IC 50 value against the recombinant enzyme. To verify that the decrease in potency was not due to a reduction in potency against the endogenous enzyme compared with the recombinant enzyme, we repeated the mass spectrometry based activity assay with the T. brucei cell-free system. We found that it was necessary to reduce the amount of cell-free system by 40-fold in the mass spectrometry-based assay compared with the radiometric assay to achieve measurements in the linear range for the turnover of GlcNAc-IPC 18 . The potency of compound 1 against the cell-free system with an IC 50 = 66 ± 8 lM (Fig. 3B) was comparable to that against the recombinant enzyme. i The assay used the CHO MS2S cell line, which has a defective de-N-acetylase and expresses the GPI anchored human CD55, functionally complemented by transfection with T. brucei de-N-acetylase. Cells were treated with trypsin, plated onto compound, and GPI biosynthesis monitored by aCD55 ELISA at 48 h.

Structure-activity relationship
Substructure searching identified commercially available analogues of 1 that were screened for activity against in the mass spectrometry based assay against T. brucei cell-free system. Replacement of the hydroxamic acid with carboxylic acid decreased the potency of inhibition, potentially due to the carboxylic acid acting as a less efficient zinc chelator. Removal of the 2-hydroxyl group or substitution with bromide or amine completely abrogated the inhibitory activity, whereas substitution at the 4 0 position with bromine was tolerated (Table 1).
To further explore the available chemical space, a small array of analogues of compound 1 were synthesised from the literature compound 11 21 by coupling to the 5-NH 2 position with commercially available acid chlorides (Scheme 1). After deprotection of the hydroxamic acid, the majority of compounds showed decreased inhibitory activity compared with 1 ( Table 2). The two most potent compounds 20 and 21 had IC 50 values of 45 ± 11 lM and 69 ± 6 lM respectively ( Supplementary Fig. 1), comparable to the parent compound albeit with reduced ligand efficiency (À0.30 and À0.29 kcal mol À1 per non-hydrogen atom respectively), suggesting that the 5 0 -amide substitution was tolerated but contributed little to binding.
The structure-activity relationship derived from the data presented here suggest that the 2-OH group of 1 is essential for activity and substitution at the 4-and 5-position is tolerated (Fig. 4). Comparing 1 with the natural substrate GlcNAc-PI could suggest that 1 is mimicking the GlcNAc moiety, with the 2-OH of 1 mimicking the 3 0 -OH that is essential for substrate recognition. In the proposed reaction mechanism for the de-N-acetylase, the catalytic base D43 assists the nucleophilic attack of a zinc-bound water molecule on the activated acetamidocarbonyl, and the subsequent protonation of the tetrahedral intermediate promotes loss of acetic  acid. We propose that 1 could inhibit the enzyme through the hydroxamic acid binding to the catalytic zinc, and replacing both the acetamidocarbonyl and the activated water (Fig. 4).

Conclusions
We have identified salicylic hydroxamic acid 1 as an inhibitor of T. brucei GlcNAc-PI de-N-acetylase with high ligand efficiency, representing the first non-substrate analogue inhibitor of the trypanosome GPI pathway. Although 1 shows modest potency with an IC 50 of 63 ± 4 lM, it has impressive ligand efficiency of 0.57 kcal mol À1 per non-hydrogen atom, and thus may be a useful starting point to develop more potent inhibitors. a Inhibition of T. brucei de-N-acetylase (cell-free system).