Described herein is the development of a novel type of covalent inhibitor with selective reaction kinetics for the P. aeruginosa
glutaredoxin (PaGRX). This work was inspired by the need to develop new classes of drug molecules directed against unconventional drug targets using mechanisms that may circumvent common bacterial antibiotic avoidance strategies. Covalent inhibitors are re-emerging as tools for antineoplastic therapies [1
], but there are compounds that have seen widespread use besides the serine β-lactamase inhibitors [2
]; perhaps because of their obvious pharmacokinetic limitations. However, considering the rapid evolution of bacterial resistance to conventional drug compounds, investigation of all therapeutic strategies is required to prevent resistance to or at least extend the lifetime of usefulness of antimicrobial agents. In P. aeruginosa
resistance is often one of two routes: (1) overexpressing efflux pathways, or (2) mutations within target proteins that render a drug ineffective [3
]. The present strategy is to combine a covalently reactive functional group, eliminating the efflux problem, with targeting moieties selective for species specific proteins from biochemical pathways that would have been avoided due to cross-reactivity risks for the host organism. These atypical protein targets are less likely to contain existing mutations that convey resistance and thus may avoid resistance longer. In addition, infectious colonies are less likely to generate a viable mutation if the targeted pathway is central to the bacterial metabolism. Our results show that comparative screening of drug precursors interacting with bacterial proteins and the related human homolog may enable identification of base scaffolds selective for the bacterial protein that can be used to generate covalent inhibitors that avoid cross-reactivity in vivo.
Orthologous proteins are those that share their gene sequence origins in a common evolutionary ancestor. They are usually highly homologous, structurally similar, and perform the same functions, but contain subtle variances in sequence due to natural evolutionary divergence. Of particular interest are differences proximal to conserved enzyme active sites, since the actual active sites are expected to be invariant [4
]. Ligands that have special affinity for these nearby binding sites can be optimized for inhibition by chemical elaboration into the active site. A popular option for ligand growing is the use of an electrophilic functional group (warhead) that can form a covalent bond with a nucleophilic residue on the target protein. When attached at a suitable position on the lead compound (driving group), the warhead is directed toward the nucleophilic side chain and a covalent binding event can occur; usually driven by the Michael addition reaction [5
]. With the lead compound now serving as a driving group, it can increase residence time of the warhead near the active site. This confers a kinetic advantage over the warhead as a free ligand [6
]. The driving group also enables target specificity, eliminating some of the concerns associated with covalent inhibitors [7
]. A comparison of small ligand fragments interacting uniquely with orthologous proteins is the foundation for our approach to discovering species selective lead compounds. Presently, we focus on a set of glutaredoxin proteins (GRXs) as a model system to describe a novel, species selective fragment-based drug discovery (FBDD) strategy.
FBDD is an established method for the discovery of lead molecules that can be developed into higher affinity drug-like compounds for protein targets. This screening method uses libraries of fragment molecules (1 × 104
compounds weighing between 150–300 Daltons) to probe the surfaces of proteins for small, previously unknown binding epitopes. FBDD methods include highly sensitive techniques such as NMR [8
], X-ray crystallography [9
], surface plasmon resonance [10
], weak affinity chromatography [11
], native mass spectroscopy [12
], and calorimetry [13
], among others. FBDD is often considered to be more effective than high throughput screens (HTS) for searching chemical combinatorial space and may produce fewer false negatives than with larger molecules [14
] and when combined with structure-based methods can greatly accelerate the pace of drug discovery [18
While there are several options available for NMR-based FBDD [19
], we chose to use a modified saturation transfer difference (STD) NMR-based screening strategy against three orthologous GRXs [20
]. In this method 1
H-NMR spectra are collected in a pair of experiments where in one a pulse is introduced on the methyl frequency (on-resonance) to irradiate the protein target, completely saturating all protein 1
H signals via spin diffusion. In the second experiment, the saturation pulse is applied off-resonance and does not saturate the protein at all. In the former experiment, saturation is transferred through space to the ligand if it is interacting with the protein in some way. Ligand dissociation after saturation transfer results in a partial loss of signal intensity for the ligand resonances as the saturated ligand population contributes to the overall signal from the free ligand pool [21
]. What is distinctive about our screen is that we are performing it on three different but evolutionarily related protein orthologs using the exact same conditions and panel of fragments to identify sets of common binders, non-binders, and ortholog preferential binders.
To verify positive hits obtained from the STD screen, we employed transfer nuclear Overhauser effect (trNOE) and 15
N heteronuclear single quantum coherence (HSQC) chemical shift perturbation (CSP) as complementary forms of cross-validation for each fragment [22
]. In trNOE, when a fragment binds transiently to a large molecule it inverts the sign of the NOE cross-peaks. This change in sign is due to change in molecular tumbling rate of the ligand as it moves in congruence with the target and confirms that it is a binder of that protein [19
N HSQC CSP screening identifies residue specific changes in the chemical environment of the protein target when bound to the ligand. This approach observes spectral perturbations (i.e., CSPs) to characterize a binding event with a ligand from the protein’s perspective [25
]. As each peak in the HSQC corresponds with a unique residue on the protein, perturbations exceeding the standard deviation can be said to correlate with residues that interact with the ligand [26
]. Figure 1
depicts these three methods used in congress to identify an exemplary fragment, RK246.
After screening and validation of hits, the next important step in any drug design campaign is lead optimization [27
]. A good hit in FBDD is not only determined by its binding affinity with the target macromolecule, but also by its ligand efficiency (LE), which is essentially affinity normalized for size. There are several effective drug candidates that are known to show dissociation constants in the nanomolar range but have LE of only 0.3
= heavy atoms) or larger [28
]. Hits with higher LE values are believed to be more useful as leads and are often good candidates for chemical elaboration or optimization. This can involve linking or merging weak fragments that probe adjacent binding epitopes on a target molecule for large payoffs in affinity [29
]. In the present study, growing the lead into the thiolate-containing active site afforded consideration of a chemically reactive cysteine trap moiety that should allow covalent inhibition of the protein target.
GRXs are characterized by a dithiol containing active site with a Cys-X-X-Cys motif at the N-terminal end of their enzymatically active helix [30
]. Using one or both active site Cys thiolates, GRXs can catalyze the reversible reduction of a range of disulfide containing molecules. By taking advantage of the nucleophilic properties of these active site thiolates, it is possible to develop inhibitors for GRXs. Thiols and thiolates are prime targets for chemical modification, where the right compound could produce covalent inhibition of a target protein’s enzymatically important cysteine residues. Indeed, covalent alkylation of enzymatically important cysteines have already been explored to produce irreversible inhibitors of epithelial growth factor receptors (EGFR) [5
], sortase [33
], and some receptor tyrosine kinases [34
]. Cys alkylation is achieved by bringing a vinyl containing functional group into close proximity with the active site thiol(ate). Once in place, a covalent bond is formed between protein and ligand via a Michael addition reaction, where the Michael donor is the thiol or thiolate sulfur anion of a target cysteine and the Michael acceptor is the vinyl group [35
]. In the case of EGFR inhibitors, acrylamide-based functional groups were chemically synthesized into existing drug compounds creating vinyl containing warhead groups and increasing potency. Such functional groups are termed acrylamide warheads [5
Major concerns associated with warhead-based inhibitors are toxicity, cellular efficacy, and off-target reactivity. In early drug design campaigns researchers tried to avoid smaller fragments containing reactive electrophilic groups. These were considered to be promiscuous hits that can react with any target protein and are very difficult to optimize into selective leads [34
]. More recently covalent inhibitors have been regaining researchers’ interest as a final optimization step in designing therapeutic compounds [36
]. Warheads based on highly selective compounds can maintain the parent molecule’s specificity and potency for a particular target protein, setting the stage for further optimization into drug-like compounds with the potential for regulatory approval [37
Compounds containing acrylamide warheads share two basic characteristics: (1) an electrophile reactive moiety, i.e., the warhead, and (2) a driving group. The driving group, typically a heterocyclic molecule, forms the bulk of the compound and provides target specificity and affinity. This driving group is also responsible for positioning the reactive warhead into the active site. The attached acryl electrophilic warhead should have reasonable fit in the active site, low off-target reactivity, and have its reactive centers properly oriented to interact with a target nucleophile [38
]. Together these produce highly selective, irreversible inhibitors. While these types of covalent inhibitors have been met with resistance from the medicinal chemistry community in the past, it is becoming a necessity to explore every option to create new antibiotics in the war against bacterial resistance. Covalent inhibitors side-step many of the conventional mechanisms that bacteria use to deal with damaging molecules. Adding species selective elements to covalent inhibitors also affords the advantage of targeting proteins involved in critical metabolic pathways, rendering resistance by mutation a less successful option for bacteria. In the present study, we describe the interaction of a novel acrylamide warhead ligand derived from a fragment driving group with PaGRX. The fragment lead identified as PaGRX selective via the three NMR methods was chemically linked with an acrylic acid moiety to form a more potent lead molecule as an acrylamide warhead. PaGRX selectivity was maintained even after fusion to a chemically reactive warhead and a significant increase in warhead reaction rate was monitored via kinetic NMR experiments.