Design and Synthesis of Covalent Inhibitors of FabA

There is an urgent need for the development of new therapeutics with novel modes of action to target Gram-negative bacterial infections, due to resistance to current drugs. Previously, FabA, an enzyme in the bacterial type II fatty acid biosynthesis pathway, was identified as a potential drug target in Pseudomonas aeruginosa, a Gram-negative bacteria of significant clinical concern. A chemical starting point was also identified. There is a cysteine, Cys15, in the active site of FabA, adjacent to where this compound binds. This paper describes the preparation of analogues containing an electrophilic warhead with the aim of covalent inhibition of the target. A wide variety of analogues were successfully prepared. Unfortunately, these analogues did not increase inhibition, which may be due to a loop within the enzyme partially occluding access to the cysteine.


■ INTRODUCTION
Pseudomonas aeruginosa (P.aeruginosa) is a Gram-negative, rod-shaped bacteria that is capable of infecting humans.P. aeruginosa is an opportunistic infection, which causes an array of life-threatening infections in immunocompromised patients. 1,2P. aeruginosa has developed resistance to multiple classes of antibiotics and is emerging as a worldwide public health threat, so new treatments are urgently needed. 1 To overcome issues with resistance, there is a need for novel classes of antibacterials which work by mechanisms differentiated from current antibiotics. 3atty acids are an essential component of all cells; however, the bacterial type II fatty acid biosynthesis pathway (FASII) is sufficiently different from the type I pathway used in eukaryotic cells that FASII has become an attractive target for antibiotic research. 4,5In the FASII pathway, fatty acids are synthesized in a stepwise manner attached to acyl-carrier-protein (ACP) by a series of enzymes as shown in Figure 1.
First, FabD converts malonyl-coenzyme A (malonyl-CoA) 1 into malonyl-acyl carrier protein (malonyl-ACP) 2. The product from this is then used in one of two steps, either FabH catalyzes the initiation step where malonyl-ACP 2 is condensed with acetyl-CoA, or FabB or FabF condenses malonyl-ACP 2 with the growing fatty acid chain 3 to give the ketone 4. Regardless, the next step involves the reduction of the ketone 4 to the alcohol 5 by FabG using NADPH.Either FabA or FabZ are able to dehydrate this alcohol 5 to the alkene 6.Finally, FabI or FabV reduces the alkene 6 to the alkane 4 using NADH, and the cycle repeats to further lengthen the fatty acid chain. 6Many of the enzymes involved in this pathway have previously been investigated as potential antibacterial targets including FabH, 7 FabG, 8,9 FabZ, 9 FabI, 9−11 and FabF. 12FabA is also able to isomerize E fatty acids to the corresponding Z isomer 6 as shown in Figure 2.
Although both FabZ and FabA are able to catalyze the dehydration shown in Figure 1, only FabA is able to catalyze this isomerization due to subtle changes in the structure of the binding site. 6,13FabA is therefore an essential enzyme in the synthesis of unsaturated fatty acids. 14 covalent inhibitor 7 of FabA has been reported, which is able to react with a histidine residue, His70, in the active site of the enzyme as shown in Figure 3. 14 First the alkyne 7 is converted to the allene 8 by FabA; this intermediate is then able to alkylate His70 via a Michael addition to give 9 and inactivate the enzyme.14 Although this compound covalently inhibits FabA, it was designed as a research tool, not a drug, and likely lacks the properties required to reach the site of action in vivo due to its high molecular weight and lipophilicity.The goal of this work was to investigate the development of a covalent inhibitor of P. aeruginosa FabA that can be used as a tool but also has the physiochemical properties that would make it more suited as a precursor to a novel antibacterial drug.
■ CONSTRAINED LINKERS TO COVALENT WARHEAD Previously, an in-house, high-throughput screen had identified compound 10 (Figure 4) as a potent inhibitor of FabA, IC 50 = 2.3 μM. 13 The crystal structure of compound 10 bound to FabA revealed that the chlorine atom sat approximately 5 Å from Cys15, which provided a potential target for a covalent warhead (Figure 4).
To take advantage of this, several compounds were designed where the left-hand side of the compound, featuring the furan ring and the triazole ring of 10, was retained and the right-hand side altered with the intention of positioning a covalently    FabA.Cys15, the highlighted cysteine residue, is 5 Å from the chlorine atom.PDB 4CL6. 13eactive group in the correct orientation to react with Cys15.This synthesis is shown in Scheme 1.
Furan carbohydrazide 11 was reacted with methylisourea sulfate 12 in aqueous sodium hydroxide to give 13.The intermediate was suspended in water and heated to 140 °C under microwave conditions to cyclize the intermediate and give the triazole scaffold 14 in quantitative yield.Several compounds were synthesized as shown above including the example benzylamide (Scheme 1) and saturated analogues shown in Table 1 via a reductive amination to give 15 followed by deprotection (16) and addition of an acrylamide (17) as a covalent warhead, or an unreactive isostere of this.
The final compounds were designed to hold the covalent warhead in a slightly different orientation relative to the triazole core and to position it in the right area to form a covalent bond to the target cysteine residue -Cys15.These compounds were screened against FabA and the results are shown in Table 1.
The only compounds which showed significant inhibition of FabA were the known FabA inhibitor 10 and the closely related compound 19 where the chlorine atom had been replaced with a nitrile group.To investigate if compound 19 was able to covalently bind to the target, mass spectrometry was used.Formation of a covalent bond would result in an increase in the mass measured for the appropriate peptide, but no MS-fragment corresponding to this compound bound to a peptide was detected, suggesting there was no covalent binding.
The three most potent compounds, 10, 18, and 19, all contained a benzene ring, and it is known from the crystal structure that this benzene occupies space close to the target cysteine residue -Cys15.Attaching a more reactive warhead to a benzene core may allow covalent bond formation.The vinyl sulfonamide warhead was selected as it is more reactive than the acrylamides 15 previously investigated.
Treating 2-chloroethane sulfonyl chloride with a hindered pyridine derivative gave the required vinyl sulfonyl chloride, which was used to synthesize the vinyl sulfonamides 28 and 29 (Scheme 2).Compound 29 was an analogue of the vinyl sulfonamide which could not form a covalent link to the FabA protein, acting as a control.
Compounds 28 and 29 were screened against FabA but were found to be inactive (IC 50 > 30 μM).This is unlikely to be due to the compound not being reactive enough to form a covalent bond, suggesting that either the target cysteine residue, Cys15, is not reactive or that these types of compounds are not placing the covalent warhead in the correct orientation and/or space to form a covalent bond.
In in silico modeling and docking experiments, it was observed that all predicted interactions between the known inhibitor and the protein, based on the crystal structure, occurred with the triazole ring and the furan ring as shown in Figure 5.

ACS Omega
To take advantage of this, it was proposed that developing a set of compounds with flexible linkers of a variety of lengths should allow the covalent warhead to adopt a wide variety of conformations in the binding site.This should better facilitate the formation of a covalent bond.However, these compounds would have a lot of rotational freedom on binding and, therefore, a larger entropic penalty to binding than compounds synthesized thus far.Any of these compounds that seemed to be binding covalently would need to be studied using X-ray crystallography to understand how they bind.Using this information, the linker would be constrained to reduce the entropy while maintaining the ability to form a covalent bond.

■ FLEXIBLE LINKERS TO COVALENT WARHEAD
Previously, directly forming amines on this core via a reductive amination was successful so this method was employed to access these intermediates.The required aldehydes (34−37) were synthesized (Scheme 3) and used to add appropriate alkyl chains to the triazole core to give intermediates (38−41) which were subsequently deprotected (42−45).
The resulting compound library was screened against FabA; however, in all cases, no significant inhibition of the enzyme was found (IC 50 > 30 μM).This suggested that it was extremely challenging to covalently inhibit FabA with this scaffold.

■ PROBING THE BINDING SITE
To investigate why FabA could not be covalently inhibited by compounds based on the triazole scaffold, a molecular dynamics simulation of the protein was performed using the crystal structure of the known inhibitor bound to FabA.This revealed that there is a loop in the active site which sits between the cysteine residue being targeted, Cys15, and the binding site.This loop was observed to not significantly move at any point during the simulation, as shown in Figure 6.
To investigate if this loop was blocking the cysteine residue, a small set of compounds was designed in order to probe the binding site.These were designed to put groups of various sizes and polarity in the active site in approximately the position of the loop and observe any flexibility of the protein.
Two simple alkane compounds 62 and 63 were synthesized from the corresponding aldehydes (Scheme 5).Two alcohols were also synthesized (Scheme 6).
The diols were reacted with tert-butyl(chloro)diphenylsilane to give the protected alcohols 66 and 67 in high yield.A Swern oxidation was used to convert these to the aldehydes 68 and 69 which were used in reductive aminations with the amino triazole core to give the amines 70 and 71.The silyloxy protecting group was removed in the final step to obtain the desired alcohols 72 and 73.Another route was required to synthesize the desired carboxylic acids and esters (Scheme 7).
The lactone 74 was treated with methanol to give the alcohol 75, and a Swern oxidation was used to convert this to the aldehyde 76.Performing a reductive amination with this aldehyde and the amino triazole core gave the required ester 77.However, attempting to hydrolyze this ester to the carboxylic acid 79 using lithium hydroxide instead gave the lactam 78.Subsequently this was converted to the desired straight chain carboxylic acid 79 by treating with aqueous acid.
For the shorter chain ester, the acetal was converted to the aldehyde using Amberlite acidic resin, and the reductive amination of the amino triazole core with the aldehyde gave the desired ester (Scheme 8).

ACS Omega
The resulting library of potential inhibitors was screened against FabA (Table 2).
A longer alkyl chain in the molecule gave better potency, e.g., 63 compared to 62. Adding a polar group to this resulted in a decrease in the potency, e.g., 73 vs 63.The log D was calculated in Stardrop and found not to be correlated well with the potency meaning, this is not just related to the lipophilicity.Ultimately, these data suggest that these compounds are binding in a hydrophobic pocket.A covalent warhead has heteroatoms which induce polarity, and even if the loop sitting in the binding site in front of the target cysteine residue, Cys15, can move it seems unlikely that a polar covalent warhead will be tolerated in this pocket.These data, along with the previous sets of data, suggest that identifying a covalent inhibitor of FabA is extremely challenging.Therefore, attempts to develop a covalent inhibitor of FabA by attaching a covalent warhead to the known inhibitor were abandoned.

■ EXPERIMENTAL SECTION
FabA RapidFire High-Throughput Mass Spectrometry Assay.Details of FabA RapidFire high-throughput mass spectrometry (HTMS) assay development will be published separately.An overview is shown in Figure 7. Briefly, compounds in DMSO stock were dispensed into 384-well assay plate (Greiner 781101) through an ECHO 550 acoustic liquid-handling system (LabCyte).Assays were performed by adding 5 μL 40 nM FabA protein solution in reaction buffer (50 mM Tris, pH 7.5, 1 mM DTT, 0.1% BSA, 0.005% NP40), and the reaction was initiated with the addition of a 5 μL 720 μM substrate 3-OH decanoyl-N-acetylcysteamine (3OH-NAC) in assay buffer.The plates were incubated in the plate shaker at 300 rpm at room temperature for 30 min, followed by addition of 90 μL of 1% formic acid to quench enzyme reaction.The reaction mixture was subjected to RapidFire HTMS analysis.
RapidFire HTMS was performed using a RapidFire 365 system (Agilent) coupled to a triple quadrupole mass spectrometer 6470 (Agilent).The samples were loaded onto a C4 cartridge (Agilent) using deionized water containing 5 mM ammonium formate at flow rate of 1.5 mL/min and eluted to the mass spectrometer using acetonitrile/deionized water (90/10, v/v) containing 5 mM ammonium formate in at a flow rate of 1.25 mL/min.The sipper was washed to minimize carryover with deionized water followed by acetonitrile.Aspiration time, load/wash time, elution time and re-equilibration time were set to 600, 3000, 5000, and 500 ms, respectively, with a cycle time of approximately 10 s.The triple-quadrupole mass spectrometer with electro-spray ion source was operated in positive multiple reaction monitoring (MRM) mode.The detailed setting for the mass spectrometer parameters was as follows: capillary voltage: 3000 V; gas temperature: 350 °C; gas flow: 7 l/min; nebulizer: 40 psi; sheath gas temperature 300 °C; sheath gas flow: 11 l/min; and nozzle voltage 1500 V.The MRM transitions (Q1 and Q3) for 2-decenoyl-N-acetylcysteamine (2DE-NAC), as a reaction product, were set as 272.1/153.1.The mass resolution window for both parental and daughter ions was set at as unit (0.7 Da).The dwell time, fragmentor, and collision energy for each transition were 50 ms, 100 V and 8 eV, respectively.
The inhibitory activity was calculated using the reaction product peak area.The peak area of the reaction product without enzyme was defined as 100% inhibitory activity, whereas that of the complete reaction mixture as 0% inhibitory activity.Curve fittings and calculations of IC 50 values were performed using ActivityBase XE version 9.2.0.106 from IDBS with a four-parameter logistics model.
General Methods.Chemicals and solvents were purchased from commercial sources and were used without any further purification unless noted otherwise.Air and water sensitive reactions were carried out under an inert nitrogen atmosphere in oven-dried glassware.Analytical thin-layer chromatography (TLC) was performed on precoated TLC plates (layer 0.20 mm silica gel 60 with fluorescent indicator UV254, from Merck).Developed plates were air-dried and analyzed under a UV lamp (UV254/365 nm) and by staining with permanganate or ninhydrin.Flash column chromatography was performed on prepacked silica gel cartridges (230−400 mesh, 40−63 μm, from SiliCycle) using a Teledyne ISCO Combiflash Rf or Combiflash Rf 200i. 1 H (500 MHz), 13 C (125 MHz), 1 H (400 MHz), 13 C (100 MHz), and 2D NMR spectra were recorded in DMSO-d 6 , MeOD-d 4 , or CDCl 3 using a Bruker Avance spectrometer.Proton chemical shifts are reported in ppm relative to the residual DMSO peak (δ = 2.50 ppm), methanol peak (δ = 3.31 ppm), or chloroform peak (δ = 7.26 ppm).Multiplicities are given as s (singlet), d (doublet), t (triplet), q (quartet), sept (septet), m (multiplet), brs (broad singlet), dd (doublet of doublets), or as a Table 2. Final Library of Compounds Was Screened against FabA a a Where a compound was not potent enough to measure a pIC 50 , the percent inhibition at 100 μM is reported.

Figure 2 .
Figure 2. FabA is also able to isomerize E fatty acids to Z fatty acids.This is utilized to synthesize specific unsaturated fatty acids.

Figure 3 .
Figure3.A covalent inhibitor 7 of FabA was developed and used as a tool to investigate its activity.NAC, N-acetylcysteine.

Figure 5 .
Figure 5. Interactions between known inhibitor and FabA based on the obtained crystal structure (PDB 4CL6).Scheme 3. Preparation of a Range of Cores (42−45)

Scheme 4 .Figure 6 .
Scheme 4. Covalent Warheads and Their Unreactive Isosteres Were Attached to the Free Amines of 42−45