|Home | About | Journals | Submit | Contact Us | Français|
Replacing the labile adenosinyl-substituted phosphoanhydride of biotinyl-5′-AMP with a N1-benzyl substituted 1,2,3-triazole gave a new truncated series of inhibitors of Staphylococcus aureus biotin protein ligase (SaBPL). The benzyl group presents to the ribose-binding pocket of SaBPL based on in silico docking. Halogenated benzyl derivatives (12t, 12u, 12w, and 12x) proved to be the most potent inhibitors of SaBPL. These derivatives inhibited the growth of S. aureus ATCC49775 and displayed low cytotoxicity against HepG2 cells.
Biotin protein ligase (BPL) catalyzes the reaction of biotin 1 and ATP 2 to give biotinyl-5′-AMP 3, which then biotinylates and activates essential metabolic enzymes required for fatty acid biosynthesis and gluconeogenesis, specifically acetyl CoA carboxylase and pyruvate carboxylase (Figure Figure11).1−5 A number of analogues of biotinyl-5′-AMP have recently been reported as inhibitors of BPL as shown in Figure Figure22. Some of these compounds have potential as antibacterial agents by inhibiting BPL from clinically important pathogens such as Staphylococcus aureus,6Escherichia coli,7,8 and Mycobacterium tuberculosis.9,10 A range of bioisosteres have been investigated as replacements for the labile phosphoanhydride of biotinyl-5′-AMP 3, including phosphodiester 4,11,12 hydroxyphosphonate 5,13 ketophosphonate 6,13 acylsulfamate 7,11 and sulphonmyl amide 8(10) (Figure Figure22). We have also reported biotin triazoles (e.g., 9–11) as a novel class of BPL inhibitor that selectively targets BPL from the clinically important bacterial pathogen Staphylococcus aureus over the human homologue.3,14,15
Without exception, all isostere-based BPL inhibitors reported to date contain a biotin and an adenine group, or analogue thereof, as discussed above and as shown in Figure Figure22. These two groups occupy well-defined binding pockets in the enzyme as per biotinyl-5′-AMP 3, as supported by X-ray crystallographic and mutagenesis studies.3,16 The ribose group of the triazole series can be removed as in 10, and the adenine can be modified as in 11, which has improved stability and >1000-fold specificity for the BPL from S. aureus over the human homologue.3 We now report the first examples of truncated 1,2,3-triazole-based BPL inhibitors with a 1-benzyl substituent designed to interact with the ribose binding pocket of S. aureus BPL (SaBPL), see 12a–y, Figure Figure33. These derivatives are the first examples of isostere-based BPL inhibitors lacking an appended adenine or analogue thereof and the associated tether as discussed above. The ribose-binding pocket is composed of amino acids that provide potential hydrogen bonding sites, specifically through the side chains of K187, R122, R125, and R227 as well as the backbone peptide atoms from H126 and S128 (Figure Figure44a). This series of inhibitors provides an important starting point for further optimization and antibiotic development.
In silico docking experiments were carried out in order to explore possible binding modes by which this series of benzyl analogues might occupy the active site of SaBPL. Flexible ligand docking was carried out using AutoDockTools (version 1.5.6). The docking protocol was first validated by removing compound 11 from its cocrystal structure with S. aureus BPL (PDB 3V7S) and then redocking. This occurred with a high degree of commonality as revealed by superimposition of the docked and crystallized ligands in the active site. We next docked benzylated triazoles 12a–c (Figure Figure33) into SaBPL. Each of these structures contains a single hydroxyl substituent on the benzyl ring capable of forming a hydrogen bond as per the diol in the ribose of biotinyl-5′-AMP 3. Gratifyingly, the top ranking poses of all three analogues placed the hydroxyl group in the site occupied by the ribose diol of 3 (Figure Figure44a). Rotation around the alkyl linker connecting the triazole and benzyl moieties produced subtly different poses with regards to the ribose-binding site (Figure Figure44b). These binding modes minimized steric clashes with the protein and facilitated hydrogen bonding between the alcohol group and R122 for 12a and R227 for 12b and 12c. This data reflects an apparent openness of the solvent exposed pocket and suggests that this site can accommodate a variety of functional groups as exemplified in the extended series depicted in Figure Figure33. Synthesis and biological testing of this series provides an opportunity to probe potential interaction with the ribose pocket.
The synthesis of the 1,2,3-triazoles 12a–y was carried out as summarized in Scheme 1. The key benzyl azides 14a–y were prepared from commercially available benzyl alcohols (13a–i and 13n) and bromides (15j–m and 15o–y). Specifically, commercially available benzyl alcohols 13a–i and 13n were converted directly17 into the corresponding azides 14a–i and 14n on reaction with triphenylphosphine, in the presence of carbon tetrachloride and sodium azide at ambient temperature. The second series of benzyl azides (14j–m and 14o-y) was prepared from commercially available benzyl bromides 15j–m and 15o–y on reaction with sodium azide in DMF as shown. Huisgen cycloaddition of biotin alkyne 16(18) with each of the benzyl azides 14a–y, in the presence of copper sulfate and sodium ascorbate,3 then gave the desired 1,2,3-triazole 12a–y as shown.
The activity profiles of 1,2,3-triazoles 12a–y were determined using established biochemical and microbiological assay protocols.3,19 Compounds displaying inhibitory activity against SaBPL and cytotoxic activity against bacteria, but not mammalian cells, are considered important candidates for further antibiotic development. The in vitro potency and selectivity profiles of the 1,2,3-triazoles 12a–y were measured using recombinant BPLs from S. aureus and Homo sapiens. Here the enzymatic incorporation of radiolabeled biotin onto an acceptor protein was measured in the presence of varying concentrations of each compound with the results shown in Tables 1 and 2. Previous enzymology and X-ray crystallography studies have demonstrated that the biotin triazoles are competitive inhibitors against biotin,3,15,16 and as such, inhibitory constants (Ki) were calculated from IC50 values using the known KM for biotin as previously described.20 The antibacterial activity of the compounds was also determined using S. aureus strain ATCC 49775.18 Growth of the bacteria 20 h post-treatment was measured spectrophotometrically at 600 nm. Finally, selected compounds were assessed for potential toxicity using a cytotoxicity assay with cultured mammalian HepG2 cells (ATCC HB-8065).3
The initial series of alcohol analogues 12a–c docked against SaBPL were first assayed against the enzyme (Table 1) with the compound containing a C4 hydroxyl group (12c) showing modest activity (Ki = 1.59 μM). Interestingly, the C2 and C3 hydroxylated derivatives (12a and 12b, respectively) were devoid of activity. It thus appears that the ribose pocket is sensitive to the position of the hydroxyl group and more so than predicted by the modeling. This observation is supported on analysis of the results for compounds containing other substituents, although there is little consistency regarding which position is most favored. Specifically, derivatives with a methoxyl group at C2 and C3 were both active (12d, Ki = 0.53 μM; 12e, Ki = 1.1 μM), while the C4 analogue (12f) was inactive. For an amino substituent, C2 is active (12g,Ki = 1.49 μM), while both C3 (12h) and C4 (12i) were inactive. Of the other derivatives initially tested, C3 methyl (12j, Ki = 0.71 μM), C4 carboxyl and tertiary butyl (12o, Ki = 0.67 μM; 12n, Ki = 1.1 μM) were active. The 2-bromo and 4-iodo derivatives 12p and 12r showed good activity against S. aureus BPL (Ki = 0.96 and 0.56 μM, respectively), and these compounds provided impetus for the expanded series of halogenated compounds shown in Table 2 and as discussed in detail below. A strongly electron withdrawing trifluoromethyl group at C3 and C4, resulted in compounds (12l and 12m) that were devoid of activity against SaBPL. All of the biotin triazoles tested were inactive against human BPL when 100 μM of compound was included in the assay medium. This is an important finding as it demonstrates that benzyl truncated triazoles retain the selectivity profile (i.e., active against S. aureus but not human BPL) of the earlier and more complex triazoles.3,15
The compounds shown in Table 1 were also assayed for antibacterial activity against S. aureus ATCC 49775. Compounds were designated as antibacterial if they reduced the optical density of the culture by >40% relative to the nontreated controls (Figure Figure55). Of the 18 compounds assessed, only five (those bearing amine, methyl, tertiary butyl, bromo, and iodo substituents, see 12g, 12j, 12n, 12p, and 12r, respectively) were active in the whole cell assays. The fact that these compounds also inhibited SaBPL is consistent with a mechanism of antibacterial action being through the BPL target. It is possible that those BPL inhibitors devoid of whole cell activity, namely, 12c–e, 12g, 12j, and 12p, are unable to penetrate the bacterial membrane, a problem often encountered in antibacterial discovery.21
Of the extended halogenated series (12s–x in Table 2) four compounds displayed good activity against SaBPL, see 12t, 12u, 12w, and 12x that had Ki values of 0.28, 0.6, 0.39, and 1.1 μM, respectively. A 3-halo substituent was most favored for activity as in 12t and 12w. In these cases, the inhibition constants were approximately 2- and 3-fold lower for the 3- vs 4- halogenated analogues, cf. 12t/12u and 12w/12x. A 3-fluoro substituent (12t) provided the most potent compound in this series with a Ki = 0.28 μM. The incorporation of a halogen at C2 removed all activity (see 12s and 12v) as did the introduction of a second fluoro substituent as in 12y. Again all active compounds in this series showed excellent selectivity for SaBPL over the human homologue. In addition, 12t, 12u, 12w, and 12x did not show cytotoxicity toward mammalian HepG2 cells at a single concentration of 40 μg/mL.
Finally, to demonstrate that compounds inhibited protein biotinylation in vivo, antibacterial susceptibility assays were performed using a S. aureus strain engineered to overexpress the BPL target. Similar approaches to establish the mechanism of action have been employed on M. tuberculosis,10 but not previously for S. aureus. Bacteria were grown in media containing 16 μg/mL of either 12g (series 1) or 12t (series 2) for 16 h, with the optical density of the culture measured every 30 min. Overexpression of the BPL target abolished the antibacterial activity of both compounds, as S. aureus grew at the same rate as nontreated controls (Figure Figure55b,c). Bacteria harboring the parent cloning vector pCN5122 that did not express additional BPL, remained highly sensitive to both inhibitors and failed to grow in their presence. Together these data show that the mechanism of action of 12g (from series 1) and 12t (series 2) is clearly via the inhibition of BPL.
The 1-benzyl substituted 1,2,3-triazoles reported here represent a new class of BPL inhibitors that lack the adenine group, or analogue thereof, found in all other isostere-based BPL inhibitors. These compounds have much reduced molecular weight and are relatively easy to prepare. Importantly, the biochemical and microbiological data provide a clear relationship between in vitro inhibition of BPL and anti-S. aureus activity, with our most potent enzyme inhibitors generally providing our most promising antibacterials (see 12j, 12p, 12r, 12t, 12u, and 12w). In antibiotic drug discovery, this is not always the case as a number of external factors contribute to bioactivity, such as cell permeability and susceptibility to efflux mechanisms and metabolic degradation.
The best lead compound in this new series (12t) with a 3-fluoro substituted benzyl group, has a Ki of 280 nM against SaBPL and demonstrated mechanism of antibacterial activity consistent with the inhibition of protein biotinylation. It is essentially nontoxic to mammalian HepG2 cells and is devoid of activity against human BPL. This compares to the extended 1,2,3-triazole 11 that has Ki of 90 nM against SaBPL. This compound provides clear interactions with both the biotin and adenine pockets of SaBPL and, like the new benzyl series, is essentially inactive against human BPL.3 Our initial SAR data on the new benzyl series provides confidence that further optimization of in vitro inhibition will lead to improved antibacterial activity. In silico docking supports a binding mechanism in which the benzyl group interacts with the ribose pocket of SaBPL. The compounds reported here provide important new scaffolds for further chemical modification and activity optimization, specifically to interact with the adjacent adenyl-binding site in the enzyme. Such studies are currently underway, particularly the inclusion of extended substituents on the benzyl group and also substitution at C5 of the triazole.
We are grateful to the Institute for Photonics and Advanced Sensing (IPAS) for providing access to analytical HPLC and the National Health and Medical Research Council (NHMRC) and Australian Research Council (ARC) for funding. We also thank Dr. Beatriz Blanco Rodriguez for her critical comments on the manuscript.
School of Medical Sciences (Pharmacology) and Bosch Institute, The University of Sydney, Sydney, New South Wales 2006, Australia.
The manuscript was written through contributions of all authors. Medicinal chemistry was performed by J.F., W.T., S.C., and A.D.A., biochemical assays were performed by A.S.P. and S.W.P., antibacterial susceptibility assays were performed by D.H. and A.H. under the guidance of S.W.P. and G.W.B., and cell culture assays were performed by A.H. and S.W.P.
This work was supported by the National Health and Medical Research Council of Australia (application APP1068885), the Centre for Molecular Pathology, University of Adelaide, and Adelaide Research and Innovation’s Commercial Accelerator Scheme. We are grateful to the Wallace and Carthew families for their financial support of this work.
The authors declare no competing financial interest.