Acinetobacter baumannii OxPhos inhibitors as selective anti-infective agents
Abstract
The Gram-negative bacterium Acinetobacter baumannii is an opportunistic pathogen in humans and infec- tions are poorly treated by current therapy. Recent emergence of multi-drug resistant strains and the lack of new antibiotics demand an immediate action for development of new anti-Acinetobacter agents. To this end, oxidative phosphorylation (OxPhos) was identified as a novel target for drug discovery research. Consequently, a library of ~10,000 compounds was screened using a membrane-based ATP synthesis assay. One hit identified was the 2-iminobenzimidazole 1 that inhibited the OxPhos of A. baumannii with a modestly high selectivity against mitochondrial OxPhos, and displayed an MIC of 25 lM (17 lg/mL) against the pathogen. The 2-iminobenzimidazole 1 was found to inhibit the type 1 NADH–quinone oxi- doreductase (NDH-1) of A. baumannii OxPhos by a biochemical approach. Among various derivatives that were synthesized to date, des-hydroxy analog 5 is among the most active with a relatively tight SAR requirement for the N0 -aminoalkyl side chain. Analog 5 also showed less cytotoxicity against NIH3T3 and HepG2 mammalian cell lines, demonstrating the potential for this series of compounds as anti-Aci- netobacter agents. Additional SAR development and target validation is underway.
Acinetobacter baumannii (Ab) is a bacterial pathogen that is a major source of Gram-negative bacterial infections within the hospital setting.1–3 Multidrug resistant (MDR) forms of A. bau- mannii, defined as resistance to three or more antibiotic drugs, account for 63% of A. baumannii varieties and are a primary cause of pneumonia or bloodstream infections among critically ill patients (CDC webpage). The risk of mortality is high, especially among ventilator-associated pneumonia (VAP) patients.4 A 2004 study indicated A. baumannii infection in burn wound patients increased the cost of treatment by almost $100,000 per patient.5
A. baumannii is the second most commonly isolated non-ferment- ing bacterium in humans, and infection can result in pneumonia, skin and wound infections, bacteremia and meningitis. In addi- tion, A. baumannii biofilms have been implicated in diseases such as cystic fibrosis, periodontitis and urinary tract infections, partly because of the bacteria’s ability to colonize indwelling medical devices. The continual appearance of strains resistant to b-lac- tams, cephalosporins, aminoglycosides, quinolones and, lately, carbapenems has compromised treatment options. Therefore, the development of new therapeutic agents to treat A. baumannii infection is of great need. We have focused on developing novel antibiotics that selectively target the bacterial oxidative phos- phorylation (OxPhos) system, which plays an essential role in energy production in the form of ATP. Inhibition of this pathway is known to significantly impair viability of pathogenic bacteria. As an obligate aerobe, A. baumannii represents an ideal target organism. High throughput screening for inhibition of
A. baumannii OxPhos with a library of compounds uncovered several promising scaffolds. Derivatives of one of these scaffolds, 1H-benzo[d]imidazol-2(3H)-imine (BDIs), namely 1 (Fig. 1), showed a good spectrum of antibacterial activity against A. bau- mannii. Additional structure–activity relationship studies based upon the synthesis of >100 analogs led to 5 (described below) with improved potency and good activity against a battery of multi-drug resistant clinical isolates of A. baumannii.
The OxPhos system is composed of the electron transport chain (ETC) and ATP synthase, and plays a central role in ATP synthesis in various Gram-positive and Gram-negative bacterial pathogens. Recent biochemical6–10 and genetic11,12 studies on the OxPhos pathway in bacteria indicate that these coupled processes of the oxidation of NADH, the transfer of electrons along the ETC and the establishment, and maintenance and utilization of the electro- chemical proton gradient to drive ATP synthesis are essential for bacterial survival under a wide range of conditions. While sequence overlap exists between some bacterial and mammalian mitochondrial OxPhos components, there is excellent evidence that marked differences exist between the components of the bac- terial ETC/ATP synthase and those of mammalian mitochondria. For example, cytochrome bd oxidases and type II NADH:quinone oxidoreductase (NDH-2) have no counterpart in human mitochon- dria, but appear to be nearly ubiquitous in aerobic bacteria. In addition, bioinformatic analyses indicate structural divergence between the enzymes of the OxPhos pathway of bacteria and mito- chondria. For instance, the a and c subunits of ATP synthases have been suggested to be good targets for specific antibacterial agents.13 This approach has been validated by the use of atovaqu- one, which targets complex III of the ETC of malarial-causing Plas- modium, and by the discovery of a potent inhibitor of Mycobacterium tuberculosis ATP synthase, the diarylquinoline bedaquiline,13,14 which was approved in late 2012, although a Phase II study showed this compound to have a higher death rate than comparators.15 In addition, we have reported that inhibitors of ETC component type 2 NADH–quinone oxidoreductase (NDH-
2) are effective anti-M. tuberculosis agents in culture and in animal models.8,9
Isolated A. baumannii membranes were screened against a 10,000 member small-molecule library that was designed and collated based on previous antibiotic screening. Submitochondrial particles (SMP), prepared from bovine heart, were used to estimate selectivity relative to mammalian OxPhos as a secondary assay. The primary screen identified several drug-like scaffolds such as 1 that inhibited A. baumannii ATP synthesis. We have investigated the 1H-benzo[d]imidazol-2(3H)-imines (viz. 1) to the greatest extent, as 1 was the most potent among the original hits with a 1.9 lM IC50 for the inhibition of ATP synthesis, an MIC of 25 lM against A. baumannii, and the best selectivity for A. baumannii OxPhos over mammalian (SMP) OxPhos. A correlation between inhibition of ATP synthesis and MIC was established as more com- pounds were studied. Interestingly, no correlation was found between inhibition of SMP ATP synthesis and cytotoxicity, limiting the utility of this counterscreen in determining the selection of compounds to progress into medicinal chemistry and animal studies.
Replacing the imino NH of 1 with O to afford urea 2 led to a six- fold increase in ATP synthesis IC50 (Ab IC50), and lower antibacterial activity (Ab MIC, Fig. 2 and Table 1). Removing the hydrophobic side chain (3) led to a precipitous drop in potency, while removing the 2-(diethylamino)ethyl side chain (4) retained activity but lost selectivity (SMP IC50). Compound 5 in which the OH moiety was removed from the alkyl dichlorophenoxy side chain showed 3-fold improved activity for ATP synthesis, and greater Ab MIC potency. Time-kill assays at 8 the MIC demonstrated 5 to be bactericidal against Ab. Selectivity was similar to lead 1. Shorter hydrophobic side chains (6 and 7) were 10-fold less potent than 5.
The isolated membranes used for screening contain a fully func- tional OxPhos system and are amenable to biochemical investiga- tion of the mechanism of action of compounds. Consequently, the mechanism of action of 5 has been investigated in these mem- branes. Using this system NDH-1 was identified as a target of 1. Figure 3a shows the result of monitoring O2 consumption by A. baumannii membranes that were exposed to 5. The blue trace shows initial O2 consumption with NADH (1 mM) as the substrate which is abruptly terminated upon addition of 5 to give a concentration of 50 lM. Addition of ascorbate/TMPD (10 mM/0.1 mM in the assay) returned O2 consumption. These results can be rational- ized by 5 inhibiting NDH-1 (preventing transfer of electrons to the quinone (Q) pool) and ascorbate/TMPD directly supplying electrons to the terminal oxidases, cytochrome bo3 oxidase and cytochrome bd oxidase, thereby circumventing the inhibited NDH-1. This is depicted in Figure 3b which shows the sequence of proteins that shuttle electrons through the OxPhos system of A. baumannii to create a membrane potential to facilitate ATP syn- thase function. Similarly, the green trace in Figure 3a shows initial of these findings, the IC50 values obtained by monitoring ATP syn- thesis or by direct monitoring of NADH consumption at 340 nm match closely for both 1 and 5.
In continuing SAR studies of 5, moving the position of the alkyl dichlorophenoxy side chain from the ring benzimidazole nitrogen to 2-CH2NH (8) led to poor activity. Compounds 9–12 were pre- pared that move the point of attachment of the 2-aminoethyl side chain and structurally related side chains from the internal benzimidazole nitrogens of 1 to the external nitrogen. These analogs generally showed good potency but poor selectivity, with little or no MIC activity at 100 lM, possibly due to poor membrane permeability.
For the purpose of the additional SAR analysis presented here, the dichlorophenoxy moiety was kept constant; however, o-chlo- rophenyl and biphenyl substitution here has also looked promising so far (data not shown). The SAR aspects of 5 were explored fur- ther. Diverse side chain replacements were made to the 2-(diethyl- amino)ethyl moiety with the reasoning that this side chain was likely to contribute to binding with the OxPhos protein NDH-1 involved in activity. A dialkylamino group is often subject to phase I oxidative metabolism via cytochrome P450 mediated N-dealkyla- tion, which was observed for 5 (see below). An assortment of ana- logs varying the 2-aminoethyl side chain is compared in Table 2 keeping the rest of the molecule constant, including ethyl 13, butyl 14 and phenylpropyl 15, di(alkyl)aminoethyl derivatives 16 and 17, heterocycles 18–27, amide 28, and diether 29. Of this group, most of the compounds showed low micromolar activity in the A. baumannii ATP synthesis screen, and a few had antibacterial activity at or below 50 lM. With this in mind we focused on more conservative changes to 2-(diethylamino)ethyl moiety of 5.
Compound 5 (highlighted in grey in Table 2) was evaluated in a panel of 23 clinical isolates of A. baumannii including multi-drug resistant (MDR) strains from various regions around the US, Europe and Latin America. The compound showed impressive activity for this early stage (Table 3). The MIC90 for 5 was 8 lg/mL (15.5 lM) and the range was tight (4–16 lg/mL). The compound performed better against the drug-resistant strains than commercial antibiot- ics ceftazidime and imipenem, and was generally only a few fold less potent than standard of care colistin (MIC90 = 2 lg/mL), which displays dose-limiting toxicities. Similar results were obtained with 4 in a second screen of the same panel of clinical isolates; MIC90 = 8 lg/mL and MIC range = 8–16 lg/mL (data not shown). Importantly, susceptibility to either compound was comparable between drug-sensitive and drug-resistant strains, indicating that the resistance phenotypes were not resistant to the OxPhos inhibitors.
Compounds 4 and 5 were also screened against drug-suscepti- ble (ATCC 17978) and multi-drug resistant (BAA-1606) strains of
A. baumannii17,18 and there were no significant differences in sus- ceptibility (MICs = 6.25 and 12.5 lg/ml, respectively). These results coupled with those from the clinical isolate screens argue that cel- lular responses, such as overexpression of efflux pumps, that con- fer resistance against multiple classes of antibiotics, including cephalosporins, carbapenems, fluoroquinolones and tetracyclines are not effective against the OxPhos inhibitors.
The spectrum of activity of 5 was investigated by determining its MIC against bacterial species other than A. baumannii. The MIC values obtained with Staphylococcus aureus, methicillin-resis- tant Staphylococcus aureus (MRSA), Escherichia coli, Klebsiella pneu- moniae and Pseudomonas aeruginosa are listed in Table 4. Comparable MIC values to that obtained for A. baumannii were obtained with MRSA and one strain of K. pneumonia. In addition,two species (E. coli and K. pneumonia) were tested in the presence and absence of 40% serum. In both cases the MIC rose in the pres- ence of serum suggesting that free drug was reduced upon serum addition.
Compound 5 showed low metabolic stability in mouse and human liver microsomes (13 and 9.2 min t1/2, respectively), which translates to the potential for high hepatic clearance rates in vivo. It was determined that the major metabolite generated in mouse liver microsomes was N-dealkylation of the diethyl amine, and minor amounts of hydroxylated products were also observed. Because of this finding, other compounds not possessing the dial- kylamino side chain were evaluated in mouse liver microsome sta- bility assays including 15, 25 and 26 (indicated in blue in Table 2). Each of these three compounds showed considerably greater sta- bility than 5 (48.2, 25.2 and >90 min, respectively). Although these compounds were not in themselves potent, this result suggests that we are able to improve metabolic stability through structural modifications, and that decreasing the dealkylation metabolic pathway did not result in some other pathway becoming overly prevalent.
Herein, a new mechanistic approach to target A. baumannii by inhibition of bacterial OxPhos, specifically NDH-1, has been dem- onstrated. Selectively targeting OxPhos in bacteria, without effect upon host OxPhos, represents a new approach to attack the grow- ing problem of resistant Gram-negative bacteria, such as A. bau- mannii. It is possible that this novel mechanism of action may afford stand alone agents or agents that are useful in combination with other agents to minimize potential drug resistance over time. We have developed a validated and robust assay for identifying inhibitors of the OxPhos pathway. Further biochemical and genetic characterization of the interactions of test compounds with target membranes are being done to substantiate the role of NDH-1 in the antibacterial activity and identify other potential OxPhos targets.
The application of the ‘drug-like’ 2-iminobenzimidazole scaffold for Gram-negative bacteria is a relatively novel approach. The demonstrated activity for 4 and 5 in a battery of >20 multi-drug resistant A. baumannii clinical isolates is a promising indicator. With this in mind further SAR MS-L6 and target validation studies are underway.