Targeting a Highly-Conserved Domain in Bacterial Histidine Kinases to Generate Inhibitors with Broad Spectrum Activity

Abstract

With the rise in antimicrobial resistance and the dearth of effective strategies to combat this threat, the development of novel therapies is of utmost importance. Targeting of bacterial signaling through their the two-component systems (TCSs) may be a viable strategy. TCSs are comprised of a sensory histidine kinase (HK), of which a bacterium can have up to 160 distinct proteins, and a cognate response regulator (RR). The TCSs are generally non-essential for life, but control many virulence and antibiotic-resistance mechanisms. This, along with their absence in animals makes the TCSs an attractive target for antimicrobial therapy, whether as a stand-alone treatments or adjuvants for existing therapies. This review focuses on progress in the development of inhibitors that target the HK ATP-binding domain. Because this domain is highly-conserved, it may be feasible to disrupt multiple TCSs within a single organism to increase effectiveness and reduce pressure for the evolution of resistance.

Graphical abstract

Introduction

The increase in bacterial resistance to all known antibiotics is a global threat to public health. Antimicrobial resistance is projected to cause more than ten million deaths per year by 2050, surpassing cancer as the second leading cause of death.¹ In part, such profound resistance has developed because most currently prescribed classes of antibiotics are derived from compounds that were initially developed four or more decades ago. Thus, new classes of antibiotics are desperately needed.

To survive in the dynamic, complex, and generally hostile environment of a human host, bacteria have developed tools to interact with and circumvent host defense mechanisms. These tools require tight regulation to adapt to particular conditions. One class of proteins used in this regulation is the two-component systems (TCSs). TCSs are the main signal transduction pathways used by bacteria to regulate a variety of processes including development, metabolism, virulence, and antibiotic resistance.^{2, 3} TCSs consist of a homodimeric, typically membrane-bound sensor enzyme, a histidine kinase (HK), and a cognate effector, the response regulator (RR). HKs generally consists of a periplasmic sensor domain, a cytosolic kinase region with a catalytic ATP-binding (CA) domain, and a dimerization and histidine phosphorylation (DHp) domain (Figure 1). Other domains, such as linker domains, can exist, but are not present in all HKs.^{4, 5} The RR is generally comprised of a phosphor-aspartyl receiver domain and one or more output domains. Most RRs are transcription factors, but they can also have enzymatic activity.⁴

Figure 1. Two-component signaling schematic and histidine kinase domain organization.

A) General schematic for signal transduction in two-component systems from histidine kinase protein to a response regulator (RR, purple box), which often controls gene transcription. B) Structural example of domain organization of dimeric orthodox, engineered histidine kinase, YF1, containing a sensor domain (blue box), catalytic ATP-binding (CA) domain (orange box), dimerization and histidine phosphotransfer (DHp) domain (green box), and transmembrane (TM) domain (red line). This protein was created by replacing the PAS domain of Bradyrhizobium japonicum FixL with the sensor domain from Bacillus subtilis YtvA (PDB: 4GCZ).⁶⁸ Image created with BioRender.

The essential role that TCSs play in bacterial signal transduction makes them an attractive target because inhibitors of these proteins could potentially possess broad-spectrum antibacterial activity.⁶ In addition, targeting of virulence mechanisms, as opposed to bactericidal or growth inhibitory compounds, also has the potential to reduce the evolutionary pressure for acquired resistance. ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp.), a group of highly of virulent, infectious, and antibiotic-resistant bacteria, have multiple mechanisms for virulence that are associated with TCSs.^{7, 8} In addition, TCSs are often required for resistance to many commonly used antibiotics such as the ß-lactams and vancomycin, as well as last resort antimicrobials such as colistin.^9–16 Inhibitors that prevent bacterial signal transduction could provide a new mechanism for the treatment of infectious disease. In addition, the absence of HKs in mammals provides the potential to have few off-target effects.

Some investigators have screened for inhibitors of HKs without targeting a specific domain.^{17, 18} Others have attempted to block signal transduction by targeting the signaling domain^19–21 or the RR,^{22, 23} which offers the potential advantage of affecting specific TCSs. While there are reviews that have covered development of TCS inhibitors in general, ^{3, 24–26} we focus here on inhibition of the HK CA domain. Because this domain displays a high degree of homology across most HKs, it should be possible to inhibit multiple HKs within a single organism, disrupting several mechanisms of resistance and limiting the likelihood of acquired resistance. Additionally, targeting resistance and virulence, as opposed to systems essential for viability, could reduce the evolutionary pressure for organisms to develop resistance.^{3, 6, 27–29}

CA domain targeting

Several inhibitors have been identified through screening and structure optimization or rational design to target the highly conserved CA domain in HKs.^{6, 30–41} These proteins are sub-categorized into 11 classes based on the sequences of their two major domains (CA and DHp), with most falling into class-1.^{4, 42} Despite these sub-classifications, HKs all contain conserved homology boxes, many of which are contained within the CA domain. The G1-, G2-, G3-, F-, and N-boxes represent a number of conserved residues and mediate crucial interactions with ATP (Figure 2).^{4, 43, 44}

Figure 2. Areas of high conservation within the histidine kinases.

A) Co-crystal structure of ADP with HK853 (PDB: 3DGE). Conserved residues Gly415, Asp411, and Asn380 form key interactions with ADP, as has been shown or postulated for a number of inhibitors. Other highlighted residues have interactions with inhibitors and substrate through π-π stacking and hydrogen bonding. B) Homology boxes conserved throughout the HK family. Conserved regions can be leveraged for targeting of multiple HKs and selectivity over eukaryotic proteins. Image created with BioRender.

2-Aminobenzothiazole-based molecules

Several scaffolds have shown successful CA domain-targeted inhibition both in vitro and in cellulo.^{30, 32–38, 40, 44–47} We identified a 2-aminobenzothiazole scaffold series of putative HK inhibitors including riluzole (Figure 3A), an FDA-approved treatment for amyotrophic lateral sclerosis, using a fluorescence polarization-based screen against HK853 (Thermotoga maritima), a constituently active class-1 HK commonly used as a model HK.³⁵ These compounds inhibit other HKs, Streptococcus pneumoniae VicK (Class-1 HK that is essential in Gram positive bacteria and implicated in peptidoglycan synthesis) and Escherichia coli CheA (Class-9 HK that is responsible for regulation of chemotaxis), without inhibiting bacterial growth or displaying cytotoxicity in eukaryotic cells at relevant concentrations.^48–50 This series of compounds has shown efficacy in reducing several virulence- and pathogenicity-associated phenotypes in P. aeruginosa.³⁷ This Gram-negative opportunistic pathogen harbors intrinsic resistance mechanisms to several common antibiotics and readily develops resistance to the other drugs, including those which are often the “last line” of therapy. Multiple TCSs are required for virulence and regulation of antibiotic resistance.^{2, 51, 52} P. aeruginosa treatment with our compounds decreased several important quorum sensing-related metabolites by as much as 92%. ³⁷ Production of these metabolites is controlled in part by TCSs including GacS/A and PhoR/B, which are critical to the regulation of antibiotic resistance and virulence.^53–55 These molecules promoted additional phenotypic changes such as a 40% decrease in biofilm production and a 70% decrease in rapid surface attachment. These traits are controlled, at least in part, by TCSs and are highly associated with virulence and poorer patient outcomes during infection.³⁷

Figure 3. Structures of HK CA domain inhibitors.

A) Highlighted in yellow and light purple are examples of the heteroatom-amino-heteroatom motifs (top left is general structure, X = heteroatom). These components are potentially important for interaction with a “triad” of conserved residues in the CA domain. Molecules with activity against HKs: Riluzole-based compounds, Adenine-based compounds, repurposed Hsp90 inhibitors, fragment and virtual screen identified compounds, and other molecules. Compounds with in cellulo activity are indicated with a blue star. B) Docking of diaryl-substituted pyrrolo[2,3-b]pyrazine molecule indicates heteroatom-amine-heteroatom motif interacts preferentially with triad residues.³⁰

The riluzole-based scaffold has also demonstrated activity against Salmonella enterica serovar Typhimurium. This species, which can be transferred between humans to animals, can cause gastroenteritis, and in severe infections, full-blown typhoid fever.⁵⁶ S. enterica bears inherent resistance to cationic antimicrobial peptides (CAMPs), which are produced by the host innate immune response. This resistance involves remodeling of the lipopolysaccharide (LPS) to prevent CAMP binding and movement through the outer membrane. This mechanism is controlled by TCSs PhoP/Q and PmrA/B.⁵⁷ We found that under conditions of low [Mg²⁺], in which Salmonella requires PhoQ for growth, several of these compounds significantly reduced organism growth. Conversely, under high Mg²⁺ conditions in which the organism does not require PhoQ, we observed no growth changes likely indicating direct or indirect PhoQ inhibition. Furthering this hypothesis, we found that the expression of pagC and pagK, two genes activated by the RR PhoP, were downregulated by as much as sixteen-fold.⁵⁸ The final exciting experiment in this work demonstrated that S. enterica became more susceptible to CAMPs, polymyxin B and E (colistin), in the presence of the 2-aminobenzothiazole inhibitors with a two-fold decrease in MIC in wild-type S. enterica and a sixteen-fold reduction in a polymyxin-resistant strain.³⁶

Repurposed Hsp90 inhibitors

HK proteins belong to the GHKL family of proteins (DNA Gyrase, Heat shock protein 90 (Hsp90), Histidine Kinase, and Mut L), which is defined by the Bergerat fold, a unique and essential portion found in the ATP-binding domains of these proteins.⁴³ Hsp90 is a eukaryotic chaperone protein responsible for folding and stabilizing of other proteins.^59–63 Vo et al. set out to leverage this structural similarity by repurposing Hsp90 inhibitors for the HKs.⁴⁷ The authors screened six commercially available Hsp90 inhibitors against CckA, an essential HK in Caulobacter crescentus, a non-pathogenic model bacteria. The authors used differential scanning fluorimetry (DSF) to monitor thermal stability of the protein in the presence of inhibitors. Notably, two of these inhibitors, NVP-AUY922 and CCT018159 (Figure 3A), induced thermal stability while also inhibiting kinase activity in vitro. Docking of these compounds in PhoQ and CckA showed similar poses, exhibiting crucial interactions with the conserved ATP-binding site aspartate residue (Asp479 in CckA; Asp415 in PhoQ).⁴⁷

NVP-AUY922 and CCT018159 share a core five-membered heterocycle connecting two substituted aromatic rings. The authors synthesized a series of nine CCT018159 analogs with a core pyrazole ring linking a resorcinol to another substituted aromatic ring. By introducing a chlorine in place of an ethyl group on an aromatic ring, they achieved a decrease in IC₅₀ from 28 μM to 11.2 μM and an increase in thermal stability by as much as 1.3˚C. Despite this minor increase in potency, there was no increase in selectivity as the analogues also had increased activity against Hsp90. Compounds were tested against HKs in S. typhimurium (PhoQ) and C. crescentus (DivJ) and inhibited phosphotransfer activity by as much as 91% in PhoQ with an IC₅₀ as low as 238 μM, but did not inhibit DivJ without causing its aggregation. Finally, the authors found that the inhibitors were bactericidal in three strains (Escherichia coli DC2, C. crescentus NA1000, Bacillus subtilis YB886) at concentrations that were not cytotoxic to mammalian cells. Although these molecules were not selective for the HKs, they provide important structural information that may lead to more selective HK inhibitors in the future.

Fragment and Virtual Screening

Velikova et al. utilized in silico screening of small molecules and an in vitro fragment-based screen by DSF to identify putative CA domain-targeted inhibitors. The authors screened 898 fragment-like ligands using DSF against the HKs of two essential TCSs, WalK from S. pneumonia and NblS of Synechococcus sp. Two molecules thermally stabilized both proteins and were weak inhibitors of WalK autophosphorylation in vitro, but one molecule (F2, Figure 3A) also inhibited E. coli PhoP autophosphorylation, with an IC₅₀ of 400 μM. In addition, these fragments demonstrated antibacterial effects against the Gram-positive organisms S. aureus and S. epidermidis.⁴⁰

In an in silico, structure-based screen of 600,000 drug-like compounds against three HKs yielded additional leads, which were tested for inhibition of autophosphorylation against HK853, PhoR from two organisms, and WalK. The most potent of these compounds inhibited growth of S. epidermidis. Using ligand-based similarity searches, the authors identified an additional set of compounds with structures analogous to the four original hits, leading to compounds with increased inhibition of HK autophosphorylation in vitro. Compound S1.13 (Figure 3A) inhibited growth of several organisms, including multi-drug resistant S. aureus, S. pneumoniae, and S. epidermidis, and yielded the most potent inhibition of PhoR autophosphorylation (low micromolar IC₅₀ value). Moreover, S1.13 did not cause apparent host damage such hemolysis or protein aggregation.⁴⁰

Adenine-based molecules

Wilke et al. screened a large library of synthetic compounds using a fluorescence polarization-based assay to identify ATP-competitive molecules.³⁵ Nine hits showed good affinity across three HKs, HK853 (T. maritima), VicK (S. pneumoniae) and CheA (E. coli), four of which contain a common adenine core. These molecules differ from native adenine by substitution on the N6 exocyclic amine, which we postulated may block important interactions with the highly conserved Asp in the G1 box (HK853 D411; equivalent to Asp479 in CckA and Asp415 in PhoQ discussed above).³³ Docking studies indicated that these molecules may be flipped in the active site in comparison to the adenine in ATP, which binds to the Asp through the 6-NH2 group, and instead are hydrogen bonded to Asp through the 9-NH. The importance of Asp411 to inhibitor binding was confirmed by DSF as the wild-type protein was stabilized by the adenine-based inhibitors, but the D411N mutant was not, in addition to it being catalytically inactive.

Modifications around the pyrimidine and imidazole portions of the adenine core structure were investigated for changes in inhibitory activity and selectivity over eukaryotic kinases and Hsp90. The molecules that were the most potent and selective shared a N-NH-N motif, which by molecular docking appears to enable binding to three residues that are conserved across the HKs: Gly415, Asp411, and Asn380 (HK853 numbering).^{4, 30, 33, 42} Although these compounds have not yet been tested for activity in live bacteria, they provide crucial information about the residues that are likely to be important for inhibitor binding.

Inhibitor-HK Binding Interactions

Although the number of studies that have identified ATP-competitive inhibitors for the HKs is relatively small, these investigations provide essential information about the amino acids within the active site that are likely essential for the continued development of potent HK inhibitors and molecular features that can be used to access these amino acids within the protein active site.

Based on several co-crystal structures and docking of adenine-based inhibitors, Goswami et al. identified key residues that are necessary for binding of ATP and inhibitors, along with important motifs present in potent inhibitors. One of the major residues, a highly conserved aspartate in the G-1 box (D411 in HK853), forms crucial hydrogen bonds with the exocyclic nitrogen on the adenine ring in crystal structures of ATP-like molecules, and by docking, is also projected to have strong interactions with inhibitors.^{30, 33, 34, 47} Other key residues in HK853, such as the conserved glycine (G415) and asparagine (N380), interact with ATP analogues and preferentially form key hydrogen bonds with several types of inhibitors. This creates an important triad for inhibitor binding along with the aspartate in the conserved G-1 box (Figure 2).^{30, 33} These three residues appear to interact with the N-NH₂-N motif in the adenine-based molecules that exhibit inhibitory properties.³³ These observations have been supported by NMR-based analysis of HK853, with luteolin, a flavonoid inhibitor of HKs, or ADP by Zhou et al.³⁴ These experiments, in conjunction with additional docking studies, have shown that Asp, Gly, and Asn, along with others located in the conserved homology boxes of HKs, are crucial points for interactions with substrate as well as inhibitors.

Although the scope of the riluzole-based scaffold SAR is limited, it is clear that substitutions at the 2-amino position of the core structure are not well tolerated.^{36, 37} This exocyclic nitrogen may very well be mediating an important hydrogen bond with the conserved G-1 aspartate, while the proximal nitrogen and sulfur may also interact with the conserved glycine and asparagine as postulated for adenine-based molecules (vide supra).^{30, 33} Indeed, docking studies performed with a compound that also contained a heteroatom-amino-heteroatom triad, diaryl-substituted pyrrolo[2,3-b]pyrazine (Figure 3B),^{30, 33} demonstrated preferential interactions with the “triad” residues. A series of guanidine-containing molecules (Figure 3A) with in vitro activity also exhibited preference for salt bridge interactions with the conserved aspartate in docking experiments, along with a π-π interaction between the central ring of these inhibitors and a tyrosine in the conserved N-box (Tyr384 in HK853).³² Similar bonding with the conserved Asp and π-π interactions were observed in the putative HK inhibitors identified by Marina and co-workers, as well as association with the N-box asparagine (Asn380 in HK853) and backbone residues in the G-2 box (Figure 2).⁴⁰ One of the originally discovered inhibitors, 3,6-diamino-5-cyano-4-phenyl-thieno[2,3-b]pyridine-2-carboxylic acid (4-bromo-phenyl)-amide (TEP, Figure 3A), has not been explored with structural or docking experiments,⁴⁵ but could also fit the reoccurring heteroatom-amine-heteroatom motif (Figure 3B).^{30, 33, 37} Another limiting factor in the cellular activity of these molecules could be penetration into the cytosol through the cell envelope. Improving their ability to enter and/or stay inside the cell would likely yield more successful antibacterial agents.^{64, 65} Several of the inhibitors in Figure 3A have in cellulo activity (blue stars), but more consideration should be made into properties that enable greater cytoplasmic accumulation.

One of the main complications of targeting the CA domain is the shared substructure with several important eukaryotic proteins. As noted, HK proteins are members of GHKL family of proteins.⁴³ Vo et al. leveraged this similarity by repurposing Hsp90 inhibitors for HKs.⁴⁷ Their inhibitor series contained a core pyrazole ring linking a resorcinol to another substituted aromatic ring. Docking of these compounds to the CA domains of C. crescentus CckA and E. coli PhoQ demonstrated similar poses, both interacting with the conserved aspartate residue through a hydroxyl group on the resorcinol ring.⁴⁷ Additionally, these compounds appear to have π-π interactions with a tyrosine (Tyr393 in PhoQ) in the conserved N-box, as seen with other inhibitors (Figure 2).^{33, 40, 47} While the generated inhibitors were also active against Hsp90, the authors suggest that exploitation of a conserved phenylalanine in the F-box that is not present in Hsp90 may yield selective HK binders in future inhibitors.⁴⁷ Hsp90 and MutL are crucial for eukaryotic cell growth and viability, regulating the conformation of signaling molecules and protein homeostasis (Hsp90) and DNA mismatch repair (MutL). Failure to achieve selectivity over these proteins could cause significant off-target toxicity.^{6, 43} DNA gyrase is itself an antibacterial target as it is an essential protein that catalyzes DNA supercoiling. As such, confirmation that observed antibacterial/antivirulence activity is due to HK inhibition alone would be important in mechanism of action studies, as well as for minimizing potential effects on beneficial bacteria such as the gut microbiota.^{43, 66}

Selectivity for the HKs will most likely be achieved by interactions within the ATP lid. The conserved F-box motif is characteristic only to the HKs, while the ATP lids of the other GHKL family members vary in length, amino acid composition, and the presence of various secondary structures.⁴³ Our work has demonstrated that selective inhibition of the HKs over Hsp90 is possible as the 2-aminobenzothiazole scaffold nor the adenine-based molecules were active against Hsp90, even at millimolar concentrations.^{33, 37} We postulate that this may be due to interactions with hydrophobic residues in the lid that are not present in Hsp90.³³ Additionally, Boibessot et al. showed that CA domain inhibitors can be selective over other GHKL family members, such as DNA gyrase.⁶⁷ Taking HK inhibitors towards clinical development will require further testing against all members of the GHKL family to minimize potential off-target toxicity.

Conclusions and Future Directions

Taken together, findings to date suggest that inhibition of the HKs is a promising strategy for interfering with virulence and resistance to host defenses and existing antibiotics.²⁴ HK inhibitors, including those described herein, appear to have substantial potential as antibiotic adjuvants or even stand-alone alternatives to antibiotics and provide hope for solutions to the current global crisis of antimicrobial resistance. One of the emerging strategies to target the HKs is through the broad inhibition of multiple proteins through their highly conserved ATP-binding domain. While still a relatively new approach, a growing body of evidence indicates that specific structural motifs are likely to be crucial in the development of potent and selective inhibitors, which may ultimately yield new antimicrobials with a novel mechanism of action and provide a defense against the rise of antibiotic resistant organisms. As we gain a greater understanding of how we can effectively inhibit HK activity through the CA domain and progress towards animal infection models, we must start to consider other factors that may translate to greater success in the clinic, such as features to improve solubility and cytoplasmic accumulation, as well as to minimize potential off-target effects and host metabolism and clearance.

Highlights:

• CA domain of bacterial HKs presents a novel anti-virulence or antibiotic target.

• CA domain targeted molecules can inhibit multiple HKs, increasing effectiveness.

• Structural data suggests several key conserved residues involved in inhibition.

• Selectivity over eukaryotic kinases and other GHKL proteins is possible.