|Home | About | Journals | Submit | Contact Us | Français|
The enhanced intracellular survival (Eis) protein from Mycobacterium tuberculosis (Eis_Mtb), a regio-versatile N-acetyltransferase active towards many aminoglycosides (AGs), confers resistance to kanamycin A in some cases of extensively drug-resistant tuberculosis (XDR-TB). We assessed the activity of Eis_Mtb and of its homologue from Mycobacterium smegmatis (Eis_Msm) against a panel of anti-tuberculosis (TB) drugs and lysine-containing compounds.
Both enzymes acetylated capreomycin and some lysine-containing compounds, but not other non-AG non-lysine-containing drugs tested. Modelling studies predicted the site of modification on capreomycin to be one of the two primary amines in its β-lysine side chain. Using Eis_Mtb, we established via nuclear magnetic resonance (NMR) spectroscopy that acetylation of capreomycin occurs on the -amine of the β-lysine side chain. Using Msm, we also demonstrated for the first time to our knowledge that acetylation of capreomycin results in deactivation of the drug.
Eis is a unique acetyltransferase capable of inactivating the anti-TB drug capreomycin, AGs and other lysine-containing compounds.
Causing 1.4 million deaths globally in 2010, tuberculosis (TB) is the second leading pathogen-related cause of death, surpassed only by HIV/AIDS. Although worldwide incidences of TB are beginning to modestly decline, occurrences of drug-resistant TB are increasing rapidly. Drug resistance to TB occurs when infected patients are treated with an improper regimen of antibiotics or when the patients fail to comply with the proper multidrug regimen, which occurs most frequently in areas with sub-standard TB control programmes. Bacteria that are resistant to at least isoniazid and rifampicin, the two most powerful first-line treatments for TB, are designated as multidrug-resistant TB (MDR-TB). Strains of Mycobacterium tuberculosis (Mtb) are classified as extensively drug-resistant (XDR) if they are shown to be resistant to rifampicin and isoniazid, to one of the fluoroquinolones and to at least one of the second-line anti-TB drugs such as kanamycin A, amikacin or capreomycin. The cure rates for MDR- and XDR-TB are ~50% and ~30%, respectively.
In one-third of kanamycin A-resistant Mtb clinical isolates from a large and diverse set, the resistance to kanamycin A was shown to be a result of up-regulation of the enhanced intracellular survival (eis) gene due to mutations in its promoter.1,2 The Eis protein is a unique aminoglycoside (AG) acetyltransferase (AAC)2 that inactivates AGs by unusual regio-versatile acetylation of AGs at multiple amine positions.3 Specifically, kanamycin A and amikacin, which are both multi-acetylated by Eis,3–6 are two important second-line anti-TB drugs. The structural and functional characterization of Eis from Mtb (Eis_Mtb) has been the focus of several recent studies.2,3,5,7–12 Eis homologues can be found in mycobacteria as well as in several non-mycobacterial prokaryotes. We have shown that Eis homologues from Mycobacterium smegmatis (Eis_Msm) and Anabaena variabilis (Eis_Ava) also act as a regio-versatile AG acetyltransferase (AAC) and that they can be inhibited by Eis_Mtb inhibitors.4,6 Additionally, it was recently reported that Eis_Msm and Eis_Mtb are also capable of acetylating lysine residues of dual-protein phosphatase 16/mitogen-activated protein kinase phosphatase-7 (DUSP16/MPK-7).5
The broad specificity of Eis towards a large number of AGs, its acetylation function towards some proteins and previous reports of acetylation by other AACs of drugs other than AGs, including the second-line anti-TB drug ciprofloxacin (Figure S1, available as Supplementary data at JAC Online),13 prompted us to ask whether Eis is capable of conferring resistance to other non-AG anti-TB drugs. As Eis was shown to acetylate protein lysine residues,5 in this study, we sought to explore the potential of Eis to acetylate capreomycin, a cyclic peptide antibiotic with a β-lysine side chain (commercially available as a mixture of capreomycin IA and capreomycin IB; Figure S1, available as Supplementary data at JAC Online), particularly effective against Mtb and used as a second-line anti-TB drug. We also investigated the potential of Eis to acetylate a variety of lysine-containing molecules (Figure S1). We demonstrated that Eis_Mtb and Eis_Msm are both capable of acetylating capreomycin IA and capreomycin IB, but neither shows activity against other non-AG anti-TB drugs that were investigated. The position of acetylation on capreomycin was confirmed as the -amine of the β-lysine side chain using Eis_Mtb. We also showed that, out of the 11 additional lysine-containing molecules tested, Eis_Mtb can efficiently acetylate tetra-lysine and fragment 28–43 of neurogranin, a calmodulin-binding protein.
Eis_Mtb3 and Eis_Msm4 were overexpressed and purified as previously described. Acetyl-coenzyme A (AcCoA), capreomycin (as a mixture of capreomycin IA and IB) (if IA or IB are not indicated, ‘capreomycin’ represents the mixture of both capreomycin IA and IB), isoniazid, pyrazinamide, ciprofloxacin, lysine, di-lysine, tri-lysine, tetra-lysine, lisinopril, tuftsin, thymulin, acetyl-amyloid β-protein fragment 15–20 amide, nisin, neurogranin fragment 28–43, thymopoietin II peptide fragment and 5,5′-dithio-bis-(2-nitrobenzoic acid) (DTNB) were bought from Sigma-Aldrich (Milwaukee, WI, USA) (Figure S1). Thin layer chromatography (TLC) plates (Merck, Silica gel 60 F254) were visualized using iodine. We recorded 1H two-dimensional gradient correlation spectroscopy (gCOSY), gradient heteronuclear single-quantum correlation spectroscopy (gHSQC) and two-dimensional z-filter total correlation spectroscopy (zTOCSY) nuclear magnetic resonance (NMR) spectra on a Varian 500 MHz equipped with a 3 mm OneProbe. All NMR spectra were recorded in D2O or 9:1/H2O:D2O at pH 8.0 (Figures S2–S9, available as Supplementary data at JAC Online). Liquid chromatography–mass spectrometry (LCMS) was performed on a Shimadzu LCMS-2019EV equipped with an SPD-20AV UV-Vis detector and an LC-20AD liquid chromatograph. UV-Vis assays were performed using a multimode SpectraMax M5 plate reader and 96-well plates (Fisher Scientific). Msm str. MC2 155 was a gift from Dr Sabine Ehrt (Weill Cornell Medical College, New York City, NY, USA).
The acetylation of the non-AG anti-TB drugs isoniazid, pyrazinamide, ciprofloxacin and capreomycin and the lysine-containing molecules lysine, di-lysine, tri-lysine, tetra-lysine, lisinopril, tuftsin, thymopoietin II peptide fragment, thymulin, acetyl-amyloid β-protein fragment 15–20 amide, nisin and neurogranin fragment 28–43 was explored using the previously described Ellman method,4,7 monitoring the reaction of coenzyme A (CoA) released by the Eis enzymes with Ellman's reagent (DTNB). Briefly, Eis (from Mtb or Msm, 0.5 μM) was added to a mixture containing Tris (50 mM, pH 8.0, adjusted at room temperature), AcCoA (500 μM), anti-TB drug or lysine-containing compound (100 μM) and DTNB (2 mM) to initiate reactions (200 μL total volume). The reaction progress at 25°C was monitored at 412 nm (412=14150 M−1 cm−1) by taking measurements every 30 s for 1 h (Figure 1c and Figure 2a and b).
The kinetic parameters (Km and kcat) for capreomycin and tetra-lysine were determined as above using varying concentrations of capreomycin or tetra-lysine [0, 20, 50, 100, 250, 500, 1000, 2000, 4000 (for tetra-lysine only) μM] and AcCoA (500 μM). The first 2–5 min of reactions was used to calculate the initial rates and the data were fitted to a Michaelis–Menten curve using SigmaPlot 11.0 to determine the Km and kcat values (Figure 1e for capreomycin and Figure 2c for tetra-lysine). All reactions were performed at least in triplicate.
An aliquot (5 μL) of the capreomycin acetylation reaction used for NMR analysis [10 mM in Na2HPO4 buffer (25 mM, pH 8.0 adjusted at room temperature)] and an aliquot of the capreomycin standard used for NMR analysis (10 mM in D2O) were loaded onto a silica gel TLC plate and eluted in a mixture of phenol:H2O:NH4OH/30:10:1. After drying for 2 h, the TLC plate was stained for visualization in a sealed chamber containing I2/SiO2 for 8 h (Figure 1b).
To confirm the single N-acetylation of both capreomycin IA and IB, the masses of the novel acetylated capreomycin were determined using the NMR sample (20 μL) (see below for NMR sample preparation) by LCMS in positive mode using H2O (0.1% formic acid). The mass spectrum obtained is provided in Figure 1(d).
The capreomycin IB structure was built with Sybyl-X software and minimized to 0.01 kcal/mol by the Powell method, using Gasteiger–Hückel charges and the Tripos force fields. The protein coordinates (PDB code 3R1K)3 were downloaded from the Protein Data Bank web site. With the exception of CoA, the H2O molecules, acetamide and all other substructures were removed from the structure. An acetyl group was added to CoA by modifying the natural ligand using Sybyl-X. The surrounding protein residues were then kept frozen while the AcCoA was subjected to energy minimization using the Steepest Descent method to get rid of steric clashes. Hydrogen atoms were added to the Eis_Mtb-AcCoA complex and its energy was minimized using the Amber force fields with Amber charges. The energy-optimized capreomycin IB was docked into the AG binding site in the Eis_Mtb protein using GOLD.14 The parameters were set as the default values for GOLD. The maximum distance between hydrogen bond donors and acceptors for hydrogen bonding was set to 3.5 Å. After docking, the top three capreomycin IB conformations were individually merged into the minimized Eis_Mtb–AcCoA complex. To relax the protein side chains, the three new Eis_Mtb–AcCoA–capreomycin IB complexes were subsequently subjected to energy minimization using the Amber force fields with Amber charges. During the energy minimization, the structures of capreomycin IB and residues within an 8 Å radius were allowed to move. The remaining residues were kept frozen in order to reduce calculation time. The energy minimization, in all three cases, was performed using the Powell method with a 0.05 kcal/mol energy gradient convergence criterion and a distance-dependent dielectric function. The model of the Eis_Mtb–AcCoA–capreomycin IB complex is depicted in Figure S10 (available as Supplementary data at JAC Online).
To determine which position of capreomycin becomes acetylated by Eis_Mtb, the reaction was performed by monitoring the solution via NMR in a manner that did not require the need for purification of the acetylated products. A solution (200 μL) containing capreomycin (10 mM, 1 eq), AcCoA (25 mM, 2.5 eq) and Eis_Mtb (5 μM) in Na2HPO4 buffer (25 mM, pH 8.0 adjusted at room temperature) containing 10% D2O was prepared and allowed to proceed to completion (overnight) at room temperature. 1H gCOSY, zTOCSY and gHSQC NMR experiments were performed on the reaction solution, applying PURGE (presaturation utilizing relaxation gradients and echoes) solvent suppression.15 Proton connectivity was assigned using zTOCSY and gCOSY spectra. Representative spectra for 6′-N-acetyl-capreomycin products are provided in Figures S6–S9, available as Supplementary data at JAC Online. Note that according to standard nomenclature the -amine of the β-lysine is notated as the 6′-position, which is indicated on the spectra of the appropriate Figures S6–S9.
To unambiguously establish the position acetylated on capreomycin, the NMR spectra of the products were compared with those of a standard of pure capreomycin (as a mixture of capreomycin IA and IB) (10 mM in 10% D2O) (Tables S1 and S2, available as Supplementary data at JAC Online). Representative spectra for the capreomycin standard are provided in Figures S2–S5, available as Supplementary data at JAC Online.
To establish if acetylation of capreomycin results in deactivation of the drug, reactions (250 μL) were done with four concentrations of capreomycin (100 μM, 500 μM, 1.0 mM and 2.5 mM) with an ~5-fold excess of AcCoA (500 μM, 2.4 mM, 5 mM and 12 mM, respectively) in Tris–HCl buffer (50 mM, pH 8.0 adjusted at room temperature) in the presence of Eis_Mtb (3.5 μM) and incubated at room temperature overnight (~16 h). A second series of identical reactions were setup in the absence of the Eis_Mtb enzyme as controls. Prior to testing the activity of capreomycin and acetyl-capreomycin in disc diffusion assays, completion of the acetylation reactions was confirmed by taking an aliquot of each of the above reaction mixtures (10 μL) to which a solution (100 μL) containing AcCoA (500 μM), Eis_Mtb (0.5 μM), Tris–HCl buffer (50 mM, pH 8.0 adjusted at room temperature) and DTNB (2 mM) was added prior to monitoring potential additional acetylation by UV-Vis assays. The reaction progress was monitored at 412 nm taking measurements every 30 s for 1 h. As expected, acetylation was observed for the reactions lacking the Eis_Mtb, while no further acetylation was observed for the reactions containing already acetylated capreomycin, indicating that acetylation of capreomycin was complete. To test the antibacterial activity of capreomycin and acetyl-capreomycin, 10 μL of the reaction mixture was spotted on sterile filter discs and allowed to air dry for 30 min. The discs were then placed on 7H9 plates and covered with soft 7H9 agar (0.75% agar) containing Msm str. MC2 155 (1 mL of a turbid culture of Msm str. MC2 155 per 25 mL of soft 7H9 agar). The bacteria were allowed to grow overnight (~16 h) at 37°C, and were then stained with 1.5 mL of an MTT solution (0.5 mg/mL of H2O). Visualization was performed after 5 min of staining (Figure S11, available as Supplementary data at JAC Online).
In addition to capreomycin, we chose other first- and second-line non-AG anti-TB drugs as candidate Eis acetylation substrates, as follows. Previous studies have shown that isoniazid (Figure S1) becomes acetylated by arylamine N-acetyltransferase 2 on the hydrazide functional group, demonstrating its potential for deactivation via acetylation.16 Pyrazinamide (Figure S1) was chosen for its structural similarity to isoniazid, allowing comparison of isoniazid with a similar drug that is not likely to be acetylated. Ciprofloxacin was selected because its acetylation by another resistance-conferring AAC was previously reported.13
We first tested acetylation of capreomycin, ciprofloxacin, isoniazid and pyrazinamide by Eis proteins from Mtb and Msm by using a UV-Vis spectrophotometric assay. We found that Eis_Mtb and Eis_Msm did not acetylate ciprofloxacin, isoniazid and pyrazinamide to any observable levels. For pyrazinamide, this result is not surprising, given the lack of nucleophilic amines likely to undergo acetylation. In contrast, we observed efficient acetylation of capreomycin by both Eis_Mtb and Eis_Msm (Figure 1a and c). Even though Eis can acetylate AGs at multiple positions, we confirmed, by TLC (Figure 1b) and LCMS (Figure 1d), that both capreomycin IA and capreomycin IB were converted to mono-acetylated products. Steady-state kinetic experiments (Figure 1e) yielded the following values of Michaelis–Menten parameters for capreomycin acetylation by Eis_Mtb: Km=654±45 μM and kcat=1.25±0.03 s−1. Both Km and kcat values are slightly greater than the respective values for AG acetylation by Eis_Mtb or Eis_Msm.3–5 The catalytic efficiency (kcat/Km) of 1911±139 M−1 s−1 is equivalent to or better than that for many of the AGs previously tested,3,4 suggesting that a somewhat lower affinity of Eis for capreomycin than for AGs is over-compensated by the more efficient acetyl transfer onto capreomycin or a quicker product release.
The findings that the lysine-containing anti-TB drug capreomycin can be acetylated by Eis proteins and the previously reported ability of Eis to acetylate lysine residues on the DUSP16/MPK-7 proteins5 raised the question of whether Eis proteins could accommodate a large number of substrates that feature a lysine side chain. To address this question, we tested 11 lysine-containing molecules. We first explored if lysine, di-lysine, tri-lysine and tetra-lysine could act as Eis substrates (Figure 2a). It appeared that a minimum of four amino acid residues are required for lysine N-acetylation by Eis as only tetra-lysine was found to be a good substrate of Eis_Mtb. Steady-state kinetic experiments (Figure 2c) yielded the following values of Michaelis–Menten parameters for tetra-lysine acetylation by Eis_Mtb: Km=1130±168 μM and kcat=0.208±0.010 s−1. The catalytic efficiency (kcat/Km) of 184±29 M−1 s−1 is 10-fold worse than that for capreomycin, indicating a higher specificity for capreomycin. We next investigated the N-acetylation of a variety of molecules containing at least one lysine residue: (i) the angiotensin-converting enzyme inhibitor lisinopril; (ii) the tetra-peptide tuftsin, a product of the enzymatic cleavage of the Fc-domain of the heavy chain for immunoglobulin G; (iii) the thymopoietin II peptide fragment; (iv) the nona-peptide thymulin, believed to be involved in T cell differentiation; and (v) the Aβ aggregation inhibitor acetyl-amyloid β-protein fragment 15–20 amide. We found that Eis_Mtb did not acetylate any of these molecules. We next explored the N-acetylation of the food preservative nisin and of the fragment 28–43 of the calmodulin-binding protein neurogranin, each containing three lysines. We found neurogranin fragment 28–43 to be a good substrate of Eis_Mtb (Figure 2b). Collectively, these results demonstrate the specificity of Eis proteins towards a selected group of lysine-containing compounds, including capreomycin, tetra-lysine and neurogranin fragment 28–43.
In order to analyse which amine of capreomycin could be acetylated by Eis_Mtb, we performed computational docking of capreomycin IB onto Eis_Mtb in complex with AcCoA (PDB code 3R1K)3 (Figure S10, available as Supplementary data at JAC Online). The docking simulation indicated that the large binding pocket of Eis_Mtb could accommodate the scaffold of capreomycin. A number of hydrogen bonding interactions between capreomycin and the side chains of amino acid residues Arg192, Glu203, Asp211, Ser239, Arg242 and Glu401 of Eis_Mtb are predicted to play an important role in binding of capreomycin in the active site of the enzyme (Figure S10C and D). More specifically, capreomycin is predicted to bind in a manner that places both amines of the β-lysine side chain in the proximity of the acetyl thioester of AcCoA (Figure S10B). This suggests the potential for two acetylations, yet capreomycin is only mono-acetylated (Figure 1a and b). It has been reported that the lysine 55 residue in DUSP16/MPK-7 is acetylated at the α- and -amine by Eis_Msm and Eis_Mtb, respectively.5 Given these observations, acetylation of the -amine of the β-lysine side chain of capreomycin by Eis_Mtb appeared more likely.
In order to unambiguously establish the site of acetylation by Eis_Mtb on capreomycin, 1D- and 2D-NMR experiments were performed with the Eis_Mtb reaction products (Figures S6–S9) and with non-acetylated capreomycin as a standard (Figures S2–S5). Tables were constructed for a side-by-side comparison of the chemical shifts of the protons of interest (Tables S1 and S2), to facilitate the interpretation of the data. The analysis of the Eis_Mtb reaction with capreomycin indicated that the -amine of the β-lysine side chain was acetylated (Figure 1a), as manifested by a change in the chemical shift of the protons directly adjacent to the -amine of the β-lysine side chain.
Can this acetylation mechanism of capreomycin by Eis play a role in vivo? A recent systematic review of Mtb mutations that are associated with resistance to capreomycin, as well as kanamycin A and amikacin, found that Eis promoter mutations have been reported in as many as 9% of capreomycin-resistant strains.17 In this review,17 a weak correlation between a C12T mutation in the eis promoter strongly associated with amikacin resistance and resistance to capreomycin has been observed,17 suggesting that Eis may play a role in acetylating capreomycin in vivo. The authors of this review17 suggested that capreomycin-resistance of Mtb may be predicted by concurrent mutations in the rrs gene and eis promoter region, suggesting that Eis may inactivate capreomycin in vivo, although evidence suggests that mutations in the eis promoter region alone are insufficient to cause significant levels of resistance to capreomycin in Mtb. Furthermore, capreomycin acetylation by an N-acetyltransferase (albeit at a yet unknown position) was shown to be a resistance mechanism ensuring protection from capreomycin of the capreomycin-producing organism, Streptomyces capreolus.18 The X-ray crystal structure of capreomycin IA in complex with the 70S ribosome (PDB code 3KNN) shows that the -amine of the β-lysine residue of capreomycin makes electrostatic interactions with the backbone phosphates of the tRNA at the A-site while the capreomycin core is bound to the ribosome.19 As capreomycin may act by blocking translocation of the ribosome,20–23 the acetylation of the -amine observed here is predicted to abolish the electrostatic interaction of capreomycin with tRNA, possibly inactivating capreomycin. Replacing the β-lysine with ornithine was shown to increase the MIC of capreomycin against Mtb and other pathogens,24 and -amine acetylation of the β-lysine of viomycin, a related cyclic peptide antibiotic produced by Streptomyces vinaceus, was shown to abolish viomycin activity. These observations, although all indirect, suggest that -amine acetylation may significantly decrease or abolish the activity of capreomycin variants containing the β-lysine residue.
In order to investigate directly the effect of acetylation of capreomycin on activity of the drug, we performed disc diffusion assays of capreomycin and acetyl-capreomycin at 0.8×, 4×, 8× and 18× MIC of capreomycin against Msm str. MC2 155 (Figure S11). Msm str. MC2 155 was selected because it is a faster-dividing and non-pathogenic mycobacterial strain that is most commonly used as a model to gain insight into Mtb mechanisms.25–27 Doing so, we demonstrated that acetylation of capreomycin abolishes its activity against the bacteria.
We have assessed the potential of Eis proteins from Mtb and Msm to acetylate various first- and second-line non-AG anti-TB drugs, finding that capreomycin is the only non-AG anti-drug tested to date that can be modified by either protein. We have also established that tetra-lysine and the lysine-containing neurogranin fragment 28–43 can be efficiently acetylated by Eis_Mtb. These observations expand upon our understanding of Eis proteins as acetyltransferases, specifically in terms of their ability to modify a broad range of small molecules as well as proteins; all known acetylations occur at a primary amine. We conclusively confirmed via NMR spectroscopy that capreomycin is modified at the -amine of its β-lysine residue. In silico assessment of the binding mode of a novel peptide scaffold such as capreomycin with Eis will aid the discovery of novel inhibitors of Eis, an emerging strategy for overcoming the resistance conferred by Eis.4,7 Our finding that acetylation of capreomycin inactivates the drug against Msm supports the potential role of acetylation in resistance to capreomycin proposed in the MIC studies of Mtb that have shown that A1401G rrs mutations combined with mutations in the eis promoter may confer some low-level resistance to capreomycin.
This work was supported by the National Institutes of Health (NIH) grant AI090048 (S. G.-T.). J. H. L. was supported by the Cellular Biotechnology Training Program (CBTP) and an American Foundation of Pharmaceutical Education (AFPE) Fellowship. R. E. P. and J. H. L were supported by Rackham Merit Fellowships at the University of Michigan. R. E. P. was supported by the Chemistry Biology Interface (CBI) Training Program at the University of Michigan.
No financial conflicts of interest to declare.
We thank Dr Sabine Ehrt (Weill Cornell Medical College, New York City, NY, USA) for providing Msm str. MC2 155. We thank Dr Oleg V. Tsodikov for critical reading of this manuscript and helpful suggestions (prior to submission).