The aminoglycoside multi-acetylating activity of the enhanced intracellular survival (Eis) protein from Mycobacterium smegmatis and its inhibition

Abstract

The enhanced intracellular survival (Eis) protein improves Mycobacterium smegmatis (Msm) survival in macrophages and functions as the acetyltransferase responsible for kanamycin A resistance, a hallmark of extensively drug-resistant (XDR) tuberculosis, in a large number of Mycobacterium tuberculosis (Mtb) clinical isolates. We recently demonstrated that Eis from Mtb (Eis_Mtb) efficiently multi-acetylates a variety of aminoglycoside (AG) antibiotics. Here, to gain insight into the origin of substrate selectivity of AG multi-acetylation by Eis, we analyzed AG acetylation by Eis_Msm, investigated its inhibition, and compared these functions to those of Eis_Mtb. Even though for several AGs the multi-acetylation properties of Eis_Msm and Eis_Mtb are similar, there are three major differences: (i) Eis_Msm di-acetylates apramycin, a conformationally constrained AG, which Eis_Mtb cannot modify, (ii) Eis_Msm tri-acetylates paromomycin, which can be only di-acetylated by Eis_Mtb, and (iii) Eis_Msm only mono-acetylates hygromycin, a structurally unique AG that is di-acetylated by Eis_Mtb. Several non-conserved amino acid residues lining the AG-binding pocket of Eis are likely responsible for these differences between the two Eis homologs. Specifically, we propose that because the AG-binding pocket of Eis_Msm is more open than that of Eis_Mtb, it accommodates apramycin for acetylation in Eis_Msm, but not in Eis_Mtb. We also demonstrate that inhibitors of Eis_Mtb that we recently discovered can inhibit Eis_Msm activity. These observations help define the structural origins of substrate preference among Eis homologs and suggest that Eis_Mtb inhibitors may be applied against all pathogenic mycobacteria to overcome AG resistance caused by Eis up-regulation.

According to the World Health Organization, over 2 billion people, about one-third of the world’s population, are infected with Mycobacterium tuberculosis (Mtb) bacilli that cause the highly contagious and life-threatening tuberculosis (TB). With almost 2 million deaths and 9 million new cases each year, the TB epidemic is one of the most devastating health problems worldwide. The rapidly emerging resistance to all of the available anti-TB drugs presents one of the biggest obstacles in treating the disease. Multidrug-resistant (MDR),(1, 2) extensively drug-resistant (XDR),(3, 4) extremely drug-resistant (XXDR),(5) and more recently totally drug-resistant (TDR)(5) strains of Mtb have been identified.

In order to develop novel strategies to fight this notorious human pathogen, a better understanding of the basic biology of Mtb and its virulence is needed. While many laboratories successfully study Mtb, faster-dividing and non-pathogenic Mycobacterium smegmatis (Msm) has been used as a model mycobacterium to gain insight into Mtb mechanisms.(6–8) The entire genome of both Mtb(9, 10) and Msm have been sequenced and annotated. Msm shares not only more than 2,000 homologs with Mtb and most of the Mtb virulence genes, but also the same unusual cell wall structure.(7) The hyper-transformable Msm str. MC2 155 is now used as a convenient tool for mycobacterial genetics.

It was previously shown that up-regulation of the Mtb enhanced intracellular survival (Eis) protein is responsible for resistance to the aminoglycoside (AG) kanamycin A (KAN), a hallmark of XDR-TB, in a significant fraction of KAN-resistant Mtb clinical isolates.(11) Eis homologs have been found in many pathogens and numerous mycobacterial species (Figures 1 and S1). When transformed into Msm, the eis_Mtb gene was found to increase the intracellular survival of Msm in the human macrophage-like cell line U-937.(12) Based on both biochemical and structural analyses, we recently demonstrated that Eis_Mtb belongs to a novel family of hexameric acetyltransferases, whose tripartite fold is composed of two GCN5 N-acetyltransferase regions, only one of which is active, and a sterol carrier protein fold.(13) All three regions contribute to the structure of the active site and the intricate substrate-binding cavity. We showed that Eis_Mtb has an unprecedented ability to acetylate multiple amines of many AGs. This multi-acetylation is regio-specific; however, the substrate recognition and specificity rules for the N-acetylation remain unclear. Herein, in the hope that functional differences between Eis_Mtb and its closely genetically related, but not identical, homolog from Msm (Eis_Msm) will provide insight into the substrate recognition and N-acetylation specificity rules of the Eis family, we performed biochemical characterization of Eis_Msm. In addition, to probe the active site of Eis_Msm, we used AG-competitive acetylation inhibitors of Eis_Mtb that we recently discovered.(14)

Figure 1.

Sequence alignment of Eis_Msm from M. smegmatis str. MC2 155 and Eis_Mtb from M. tuberculosis H37Rv. The two Eis homologs exhibit 58% sequence identity. Based on structural and mutagenesis studies of Eis_Mtb, the residues proposed to be involved in catalysis, in binding CoA, and in formation of the AG-binding pocket are marked by yellow, orange, and turquoise circles, respectively.(13) Non-conserved residues in the AG-binding pockets of these two proteins are additionally marked by dark blue circles. Conserved residues in the AG-binding pockets of Eis_Msm and Eis_Mtb are marked by green circles.

MATERIALS AND METHODS

Bacterial Strains, Plasmids, Materials, and Instrumentation

The chemically competent E. coli TOP10 and BL21 (DE3) strains were purchased from Invitrogen (Carlsbad, CA). The genomic DNA from M. smegmatis str. MC2 155 used for PCR was a generous gift from Dr. Sabine Ehrt (Weill Cornell Medical College). The pET28a plasmid was purchased from Novagen (Gibbstown, NJ). All restriction enzymes, T4 DNA ligase, and Phusion DNA polymerase were purchased from NEB (Ipswich, MA). PCR primers were purchased from Integrated DNA Technologies (IDT; Coralville, IA). DNA sequencing was performed at the University of Michigan DNA Sequencing Core. Chemical reagents including DTNB, AcCoA, AGs (APR, AMK, HYG, KAN, NEO, SIS, SPT, STR, and RIB) (Figure S2), and chlorhexidine (1) were purchased from Sigma-Aldrich (Milwaukee, WI). The rest of the AGs (NEA, NET, PAR, and TOB) (Figure S2) were purchased from AK Scientific (Mountain View, CA). Compound 2 was purchased from ChemDiv Inc. (San Diego, CA). The pH was adjusted at rt. The spectrophotometric assays were performed on a multimode SpectraMax M5 plate reader using 96-well plates (Fisher Scientific; Pittsburg, PA). Silica gel 60 F₂₅₄ plates (Merck) were used for thin-layer chromatography (TLC) analysis. Liquid chromatography mass spectrometry (LCMS) was performed on a Shimadzu LCMS-2019EV equipped with a SPD-20AV UV-Vis detector and a LC-20AD liquid chromatograph.

Preparation of the pEis-pET28a Overexpression Constructs

The pEis_Mtb-pET28a plasmid encoding the Eis_Mtb protein was constructed as previously reported.(13) In order to prepare the pEis_Msm-pET28a construct, PCR was performed using M. smegmatis str. MC2 155 genomic DNA as a template, a forward primer 5′-TCGAGACATATGATCACGCCGCGCACCCTTC-3′, a reverse primer 5′-CCCGCGGGATCCTCAGAATCCGTATCCCAGC-3′, and Phusion DNA polymerase. The resulting amplified eis_Msm gene PCR fragment was inserted into the linearized pET28a via the corresponding NdeI/BamHI restriction sites (underlined above). The plasmid encoding the Eis_Msm protein was transformed into E. coli TOP10 chemically competent cells. The plasmid bearing the eis_Msm gene insert was sequenced and showed perfect alignment with the reported gene sequence from M. smegmatis str. MC2 155 (locus tag MSMEG_3513).

Overproduction and Purification of Eis Proteins

The Eis_Mtb protein (with a N-terminal His₆-tag) was prepared as previously reported.(13) The overexpression and purification of the Eis_Msm protein was performed exactly as reported for Eis_Mtb.(13) Both proteins were dialyzed in Tris-HCl buffer (50 mM, pH 8.0) and stored at 4 °C. After purification, 1.1 mg of the 46,172-Da Eis_Msm protein was obtained per L of culture.

Determination of the AG Selectivity Profile of Eis_Msm by a Spectrophotometric Assay

The acetyltransferase activity of Eis_Msm was monitored by a UV-Vis assay in which the free thiol group of CoA, generated by the Eis_Msm enzyme catalyzed reaction, was allowed to react with DTNB to produce 2-nitro-5-thiobenzoate (NTB⁻). The production of further ionized NTB²⁻ was monitored by increase in absorbance at 412 nm (ε₄₁₂ = 14,150 M⁻¹cm⁻¹). The reaction mixtures (100 μL) containing AcCoA (0.5 mM, 5 eq), AG (0.1 mM, 1 eq), DTNB (1 mM), and Tris-HCl (50 mM, pH 8.0), were initiated by adding Eis_Msm (0.5 μM) at 25 °C. Reactions were monitored by taking readings every 30 s for 15 min in 96-well plate format.

Determination and Confirmation of Number of Acetylation Sites by the Spectrophotometric Assay and Mass Spectrometry

To determine the number of acetylations performed by Eis_Msm for a given AG, two reactions were carried out with each AG: a calibration reaction with 1 eq of AG and 1 eq of AcCoA as well as a reaction that ensured complete acetylation of the AG with 1 eq of AG and 10 eq of AcCoA. Specifically, reaction mixtures (100 μL) containing AG (0.1 mM), DTNB (1 mM), Tris-HCl (50 mM, pH 8.0), and AcCoA (0.1 mM, 1 eq or 1 mM, 10 eq), were initiated by adding Eis_Msm (1 μM) at 25 °C. Reactions were monitored at 412 nm as described above by taking readings every 30 s until a plateau was reached. Representative plots are shown in Figures 2A and S3. To confirm the number of acetylations on each AG substrate established by UV-Vis assays, a third reaction (20 μL) containing AG (0.67 mM, 1 eq), AcCoA (3.35 mM, 5 eq), Tris-HCl (50 mM, pH 8.0), and Eis_Msm protein (10 μM) was carried out overnight at 25 °C. These reactions were terminated by addition of an equal volume of ice-cold MeOH (20 μL), which was then kept at −20 °C for at least 20 min. The precipitated protein was removed by centrifugation (13,000 rpm, rt, 10 min). The masses of the acetylated AG products present in each sample were determined by LCMS in positive mode using H₂O (0.1% formic acid) after dilution of the supernatant (10 μL) with H₂O (20 μL) and injection of all 30 μL. Mass spectra of all AGs modified by Eis_Msm are provided in Figures 2B and S4. A summary of the level of acetylation is presented in Table 1.

Figure 2.

A. Conversion of NEA into its acetylated products, as monitored by the UV-Vis assay when using Eis_Msm with 1 (black circles) and 10 (white circles) equivalents of AcCoA, respectively. B. A mass spectrum confirming the formation of tri-acetyl-NEA by Eis_Msm. C. Multi-acetylation of NEA by Eis_Msm observed by the TLC assay. Lanes 1–7: a time course displaying the mono-, di-, and tri-acetyl- NEA products of the Eis_Msm reaction. Lanes 8–13: a time course of the NEA reaction with Eis_Mtb. Lanes 14–20: Controls for di- and mono-acetylation of NEA performed with AAC(2′)-Ic, AAC(3)-IV, and AAC(6′) sequentially or individually.

Table 1.

Comparison of level of acetylation by Eis from M. smegmatis and M. tuberculosis.

AG	M. smegmatis	M. tuberculosisa
AMK	Tri	Tri
APR	Di	×b
HYG	Mono	Di
KAN	Di	Di
NEA	Tri	Tri
NEO	Tri	Tri
NET	Di	Di
PAR	Tri	Di
RIB	Tri	Tri
SIS	Tri	Tri
SPT	×	×
STR	×	×
TOB	Tetra	Tetra

^aWe previously reported these data (13).

^b× indicates that no acetylation was observed.

Kinetic Characterization of Eis_Msm Activity

The steady-state kinetic assays of acetylation of AGs (AMK, KAN, NEA, NEO, NET, PAR, SIS, and TOB) by Eis_Msm, were performed in two ways: (i) at a fixed concentration of AcCoA (0.1 mM) and varying concentrations of AGs (0, 31.25, 62.5, 125, 250, and 500 μM), and (ii) at a fixed concentration of each AG (0.5 mM) and varying concentrations of AcCoA (0, 16.625, 31.25, 62.5, 125, and 250 μM). All reactions were carried out in Tris-HCl (50 mM, pH 8.0) and contained AcCoA, AG, DTNB (1 mM), and Eis_Msm (0.25 μM). In addition, to distinguish among possible kinetic mechanisms, KAN acetylation reactions with Eis_Msm were carried out at AcCoA concentrations of 100, 200 and 500 μM and for each of these concentrations, at AG concentrations of 0, 20, 50, 100, 250, 500, 1000, and 2000 μM. All reactions were initiated by addition of AcCoA and performed at least in duplicate at 25 °C. As described above, the acetylation reactions were monitored at 412 nm as a function of time and absorbance readings were taken every 15–20 s. The dependence of the initial rates on the concentration of a titrant were used to calculate the apparent Michaelis-Menten kinetic parameters, K_m,AG and k_cat,AG (for reactions of type i above) as well as K_m,AcCoA and k_cat,AcCoA (for reactions of type ii), by non-linear least-squares regression data fitting with SigmaPlot (Systat Software, San Jose, CA) to the Michaelis-Menten equation. These kinetic parameters are given in Tables 2 and 3.

Table 2.

Observed kinetic parameters^a for [AG]-dependent acetylation by Eis from M. smegmatis and M. tuberculosis.

AG	M. smegmatis		M. tuberculosis
	Km,AG (μM)	kcat,AG (s−1)	Km,AG (μM)	kcat,AG (s−1)
AMK	251 ± 55	0.034 ± 0.003	113 ± 35	0.030 ± 0.004
KAN	278 ± 36	0.108 ± 0.005	490 ± 140	0.12 ± 0.02
NEA	359 ± 101	0.176 ± 0.027	137 ± 57	0.064 ± 0.010
NEO	314 ± 77	0.114 ± 0.014	199 ± 87	0.19 ± 0.04
NET	197 ± 30	0.835 ± 0.057	96 ± 15	0.63 ± 0.03
PAR	153 ± 55	0.035 ± 0.006	96 ± 20	0.15 ± 0.01
SIS	82 ± 8	0.206 ± 0.006	159 ± 48	0.38 ± 0.04
TOB	131± 22	0.267 ± 0.016	116 ± 17	0.21 ± 0.01

^aAll parameters are defined in the Materials and Methods section.

Table 3.

Observed kinetic parameters^a for [AcCoA]-dependent acetylation by Eis from M. smegmatis and M. tuberculosis.

AG	M. smegmatis		M. tuberculosis
	Km,AcCoA (μM)	kcat,AcCoA (s−1)	Km,AcCoA (μM)	kcat,AcCoA (s−1)
AMK	58 ± 25	0.142 ± 0.026	8 ± 5	0.015 ± 0.001
KAN	39 ± 17	0.398 ± 0.060	10 ± 3	0.094 ± 0.004
NEA	22 ± 7	0.101 ± 0.008	41 ± 6	0.070 ± 0.004
NEO	76 ± 23	0.546 ± 0.065	32 ± 6	0.116 ± 0.005
NET	108 ± 35	0.819 ± 0.121	124 ± 23	0.36 ± 0.03
SIS	77 ± 19	0.620 ± 0.063	84 ± 11	0.90 ± 0.04

^aAll parameters are defined in the Materials and Methods section.

In order to convert these observed kinetic parameters into the mechanistic rate constants, we considered three mechanisms: 1) the random sequential mechanism, in which either AcCoA or AG can bind to free enzyme followed by binding of the other species, 2) the ordered sequential mechanism, in which AcCoA binds Eis first, followed by AG, and 3) the ordered sequential mechanism, in which AG binds Eis first, followed by AcCoA.

The random sequential mechanism is described by the following kinetic steps:

$$\begin{array}{l}\text{E}+\text{AG}+\text{AcCoA}\rightleftarrows \text{E}•\text{AG}+\text{AcCoA}\rightleftarrows \text{E}•\text{AG}•\text{AcCoA}\\ \text{E}+\text{AG}+\text{AcCoA}\rightleftarrows \text{E}•\text{AcCoA}+\text{AG}\rightleftarrows \text{E}•\text{AG}•\text{AcCoA}\\ \text{E}•\text{AG}•\text{AcCoA}\to \text{AG}-\text{Ac}+\text{CoA}+\text{E},\end{array}$$ (1)

where E is the enzyme and AG-Ac is the acetylated AG product.

Under the common simplifying assumptions of rapid AcCoA and AG binding equilibria for this mechanism, one obtains the observed kinetic parameters defined above in terms of microscopic constants:

$$k_{\text{cat},\text{AG}}=\frac{k_{\text{cat}}[\text{AcCoA}]}{K_{\text{d},\text{AcCoA}(\text{E}•\text{AG})}+[\text{AcCoA}]}$$ (2)

$$K_{\text{m},\text{AG}}=K_{\text{d},\text{AG}(\text{E}•\text{AcCoA})}\left(\frac{K_{\text{d},\text{AcCoA}(\text{E})}+[\text{AcCoA}]}{K_{\text{d},\text{AcCoA}(\text{E}•\text{AG})}+[\text{AcCoA}]}\right)$$ (3)

$$k_{\text{cat},\text{AcCoA}}=\frac{k_{\text{cat}}[\text{AG}]}{K_{\text{d},\text{AG}(\text{E}•\text{AcCoA})}+[\text{AG}]}$$ (4)

$$K_{\text{m},\text{AcCoA}}=K_{\text{d},\text{AcCoA}(\text{E}•\text{AG})}\left(\frac{K_{\text{d},\text{AG}(\text{E})}+[\text{AG}]}{K_{\text{d},\text{AG}(\text{E}•\text{AcCoA})}+[\text{AG}]}\right),$$ (5)

where each subscript of an equilibrium binding constant K_d designates the step in Mechanism (1); the species outside of the parentheses binds to the species in the parentheses, and k_cat is the microscopic rate constant of the acetylation, i.e. the last step in Mechanism (1).

The ordered sequential mechanism, in which only AcCoA can bind Eis first is obtained by removing the first line in Mechanism (1). For this mechanism, under the assumption of rapid equilibria, one obtains:

$$k_{\text{cat},\text{AG}}=k_{\text{cat}}$$ (6)

$$K_{\text{m},\text{AG}}=K_{\text{d},\text{AG}}\left(1+\frac{K_{\text{d},\text{AcCoA}}}{[\text{AcCoA}]}\right)$$ (7)

$$k_{\text{cat},\text{AcCoA}}=\frac{k_{\text{cat}}[\text{AG}]}{K_{\text{d},\text{AG}}+[\text{AG}]}$$ (8)

$$K_{\text{m},\text{AcCoA}}=\frac{K_{\text{d},\text{AcCoA}}{K}_{\text{d},\text{AG}}}{K_{\text{d},\text{AG}}+[\text{AG}]}$$ (9)

For the other ordered sequential mechanism, in which AG binds Eis first, we obtain by analogy:

$$k_{\text{cat},\text{AG}}=\frac{k_{\text{cat}}[\text{AcCoA}]}{K_{\text{d},\text{AcCoA}}+[\text{AcCoA}]}$$ (10)

$$K_{\text{m},\text{AG}}=\frac{K_{\text{d},\text{AG}}{K}_{\text{d},\text{AcCoA}}}{K_{\text{d},\text{AcCoA}}+[\text{AcCoA}]}$$ (11)

$$k_{\text{cat},\text{AcCoA}}=k_{\text{cat}}$$ (12)

$$K_{\text{m},\text{AcCoA}}=K_{\text{d},\text{AcCoA}}\left(1+\frac{K_{\text{d},\text{AG}}}{[\text{AG}]}\right)$$ (13)

The different functional dependences of the observed Michaelis-Menten parameters on the concentrations of AG and AcCoA for the three mechanisms allow one to distinguish among them.

For the random sequential mechanism, eqs 2–4 yield:

$$\frac{k_{\text{cat},\text{AG}}}{K_{\text{m},\text{AG}}{k}_{\text{cat},\text{AcCoA}}}=\frac{[\text{AcCoA}]}{[\text{AG}]}\left(\frac{1}{K_{\text{d},\text{AcCoA}(\text{E})}+[\text{AcCoA}]}\right)\left(\frac{K_{\text{d},\text{AG}(\text{E}•\text{AcCoA})}+[\text{AG}]}{K_{\text{d},\text{AG}(\text{E}•\text{AcCoA})}}\right)$$ (14)

The left-hand side of the equality is composed of observed constants. The expression in the first parentheses on the right-hand side of the equality is constant for all AGs and, by an assumption, similar for the two Eis proteins, as explained in Results. This allows one to obtain an estimate of K_{d,AG(E•AcCoA)} followed by the determination of the k_cat from eq 4. These values are given in Table 5. Where only bounds on these values could be obtained due to a very low affinity of an AG for AcCoA-bound Eis (K_{d,AG(E•AcCoA)} ≫ [AG]), only catalytic efficiency (k_cat/K_{d,AG(E•AcCoA)}) was calculated from eq 4 and reported.

Table 5.

Rate and equilibrium binding constants^a for the random sequential mechanism of acetylation by Eis from M. smegmatis and M. tuberculosis.

AG	M. smegmatis			M. tuberculosis
	Kd,AG(E•AcCoA) (μM)	kcat (s−1)	kcat/Kd,AG(E•CoA) (s−1M−1)	Kd,AG(E•AcCoA) (μM)	kcat (s−1)	kcat/Kd,AG(E•CoA) (s−1M−1)	Kd,AcCoA(E) (μM)b
AMK	>1000	>0.3	284 ± 52	32 ± 14	0.016 ± 0.007	500 ± 300	87 ± 23
KAN	2540 ± 160	1.6 ± 0.5	630 ± 200	350 ± 150	0.16 ± 0.07	460 ± 280
NEA	141 ± 60	0.13 ± 0.05	917 ± 500	95 ± 50	0.08 ± 0.04	870 ± 640
NEO	>1000	>1.0	1090 ± 130	75 ± 41	0.13 ± 0.07	1800 ± 1400
NET	108 ± 35	1.0 ± 0.4	7900 ± 4000	31 ± 10	0.4 ± 0.1	12200 ± 5600
SIS	179 ± 54	0.85 ± 0.25	4700 ± 2050	340 ± 140	1.5 ± 0.6	4500 ± 2600

^aAll parameters are defined in the Materials and Methods section. Determination of the rest of the equilibrium constants of the mechanism by this method was not reliable due to large propagated uncertainties.

^bThis value is AG-independent and assumed to be the same for Eis_Msm and Eis_Mtb (see main text) in the calculations of the Eis_Mtb parameters in this table.

Determination of Positions Acetylated on NEA by Eis_Msm by TLC

To establish which three positions of the NEA scaffold get acetylated by Eis_Msm, reactions (40 μL) were carried out at rt in Tris-HCl buffer (50 mM, pH 8.0) in the presence of AcCoA (4 mM, 5 eq), NEA (0.8 mM, 1 eq), and Eis_Msm (5 μM). Aliquots (4 μL) were loaded and run on a TLC plate after 0, 1, 5, 10, 30, and 120 min as well as after overnight incubation (Figure 2C). The eluent system utilized for the TLC was 3:0.8/MeOH:NH₄OH. Visualization was achieved using a cerium-molybdate stain (5 g cerium ammonium nitrate, 120 g ammonium molybdate, 80 mL H₂SO₄, 720 mL H₂O). The R_f values of the Eis_Msm modified NEA were consistent with those obtained for the Eis_Mtb reaction with NEA (Figure 2C).(13) Control reactions for mono- and di-acetylation of NEA were performed using the AG acetyltransferases AAC(2′)-Ic from M. tuberculosis H37Rv,(13) AAC(3)-IV from E. coli,(15) and AAC(6′) from the AAC(6′)/APH(2″) bifunctional enzyme individually and sequentially.(15)

Determination of Kinetic Parameters and Mode of Inhibition for Inhibitors of Eis_Msm

The IC₅₀ values of two inhibitors, chlorhexidine (1) and compound 2, were determined as previously described.(14) Briefly, the inhibitors were dissolved in Tris-HCl (50 mM, pH 8.0, containing 10% v/v DMSO) (100 μL) and 5-fold dilution series were performed. To these solutions, a mixture (50 μL) containing Eis (1 μM), NEO (400 μM), and Tris-HCl (50 mM, pH 8.0) was added and incubated at rt for 10 min. The reactions were initiated by addition of a mixture (50 μL) containing AcCoA (2 mM; ensuring that all Eis is in the AcCoA-bound form), DTNB (2 mM), and Tris-HCl (50 mM, pH 8.0). The absorbance change at 412 nm was monitored every 30 s for 15 min at 25 °C. Initial rates were calculated and the IC₅₀ values were determined by using a Hill-plot analysis by using the KaleidaGraph 4.1 software (Synergy software; Reading, PA) (Figure 3).

Figure 3.

Inhibition of Eis_Msm. IC₅₀ curves for A. chlorhexidine (1), and B. compound 2. The insets show the competitive and mixed inhibition modes of compounds 1 and 2, respectively.

To determine the mode of inhibition of both compounds tested, varying concentrations of NEO (final concentrations of 50, 75, 100, 150, and 200 μM) and chlorhexidine (1) (final concentrations of 0, 0.5, 1, 2, and 4 μM) or compound 2 (final concentrations of 0, 0.0625, 0.125, 0.25, and 0.5 μM) were used. Chlorhexidine (1) was found to be a competitive inhibitor of NEO whereas mixed inhibition was observed for compound 2. The equilibrium constants K_i for chlorhexidine (1) binding to the Eis•AcCoA complex and constants K_i and K_i′ for compound (2) binding to the Eis•AcCoA and Eis•AcCoA•AG complexes, respectively, were determined from the non-linear least-squares regression data fitting with SigmaPlot, to a Michaelis-Menten equation:

$$V=\frac{(1/{\alpha }^{\prime })V_{max,AG}[\text{AG}]}{(\alpha /{\alpha }^{\prime })K_{\text{m},\text{AG}}+[\text{AG}]}$$ (15)

where V is the steady-state reaction rate, α = 1 + [I]/K_iα′ = 1 + [I]/K_i′ (α′ = 0 for chlorhexidine), and V_max,AG and K_m,AG are the observed kinetic parameters for the reaction in the absence of inhibitor. The reaction rates were shown as Lineweaver-Burk plots (Figure 3, insets of panels A and B).

RESULTS

Substrate Specificity and Multi-acetylation Profiles of Eis Proteins

Recently, we reported the structural and biochemical characterization of Eis_Mtb, an AG acetyltransferase (AAC) responsible for resistance to KAN in Mtb clinical isolates.(13) We discovered Eis to be a novel AAC capable of unprecedented multi-acetylation of a large number of AG scaffolds. In this study, to gain insight into the unique recognition and specificity rules for the N-acetylation by Eis proteins, we performed a detailed characterization of AG acetylation by Eis_Msm, a homolog of Eis_Mtb, and compared its functional properties to those of Eis_Mtb. To ensure the validity of the comparison, we identically cloned and heterologously expressed the 46,172-Da Eis_Msm and the 46,597-kDa Eis_Mtb in E. coli as N-terminally His₆-tagged proteins. We first investigated the substrate specificity and the multi-acetylation profile of Eis_Msm by spectrophotometric assays (Table 1, Figures 2A and S3) and mass spectrometry (Figures 2B and S4). The data indicated that AMK, APR, HYG, KAN, NEA, NEO, NET, PAR, RIB, SIS, and TOB were substrates of Eis_Msm, whereas SPT and STR were not. We found Eis_Mtb and Eis_Msm to have generally similar AG substrate selectivity and N-acetylation profiles with SPT and STR not modified, KAN and NET di-acetylated, AMK, NEA, NEO, RIB, and SIS tri-acetylated, and TOB tetra-acetylated by both Eis proteins. Interestingly, we observed three major differences in the activity of these two Eis proteins: (i) Eis_Msm di-acetylated APR, an AG with conformationally restricted and extended cylindrical-like structure (Figures 4 and S2),(16, 17) which Eis_Mtb could not chemically modify, (ii) Eis_Msm tri-acetylated PAR, which could only be di-acetylated by Eis_Mtb, and (iii) Eis_Msm mono-acetylated HYG, a structurally unique AG that could be di-acetylated by Eis_Mtb. The identity of several non-conserved amino acid residues lining the AG-binding pocket of the two Eis proteins (highlighted in dark blue in Figures 1 and 4) is likely responsible for these differences in the number of acetylation sites.

Figure 4.

Structural differences between the AG-binding pockets of Eis_Mtb and Eis_Msm. A. The active site and the AG-binding pocket of Eis_Mtb bound to CoA and an acetamide, as seen in the recently reported crystal structure (PDB code: 3R1K (13)). In both panels, CoA is shown in orange, acetamide in red, the conserved AG binding pocket residues in green, and the non-conserved residues in blue. The C-terminal carboxyl group is denoted as C-ter. B. A model of the active site and the AG-binding pocket of Eis_Msm. This model was generated by substituting the non-conserved residues of the AG-binding pocket Eis_Mtb with their Eis_Msm counterparts (Ile268, Trp289, and Gln291 in Eis_Mtb were mutated to Gly266, Ala287, and Ala289 in the Eis_Msm numbering, respectively). As a result of these mutations an APR molecule (in yellow) can be accommodated in the pocket in the orientation appropriate for catalysis. The structure of APR was taken from the high-resolution PDB entry 2OE5.(27)

Regio-specificity of Tri-acetylation of NEA by Eis Proteins

Knowing that the substrate profile for both Eis proteins was generally similar, we sought to compare the regio-specificity of their acetylating activity on an AG scaffold for which the number of sites acetylated was observed to be the same for the two proteins. We recently established by TLC and NMR spectroscopy that Eis_Mtb tri-acetylates NEA first at the 2′-, then at the 6′-, and finally at the 1-position.(13) Here, by using TLC and comparing the retention factor (R_f) values of the mono-, di-, and tri-acetylated NEA derivatives produced over time by Eis_Msm to the R_f values of their counterparts formed by Eis_Mtb as well as to those of mono- and di-acetylated NEA standards obtained by using individually or sequentially mono-acetylating AAC enzymes, we observed that Eis_Msm is also a 2′,6′,1-NEA-tri-acetylating enzyme (Figure 2C). These results suggest that AG scaffolds with the same number of acetylations by Eis_Mtb and Eis_Msm are likely modified by these two enzymes at the same positions.

Steady-state Kinetic Analysis of Eis Proteins

Seeking a deeper understanding of the subtleties differentiating Eis_Mtb and Eis_Msm, we determined the observed Michaelis-Menten catalytic parameters of steady-state acetylation for several AGs. We carried out two sets of reactions, one at a fixed concentration of AcCoA and varying concentrations of AGs (Table 2), and the other at a fixed concentration of each AG and varying concentrations of AcCoA (Table 3). In order to convert the observed kinetic parameters into the mechanistic rate constants, we considered three kinetic mechanisms of Eis acetylation: 1) the random sequential mechanism, in which either AcCoA or AG can bind to free enzyme followed by binding of the other species, most commonly observed for other AAC enzymes (18–22), 2) the ordered sequential mechanism, in which AcCoA binds Eis only before AG does, observed for some AAC enzymes (23–25) and 3) the ordered sequential mechanism, in which AG binds Eis only before AcCoA does, not observed previously for other AAC enzymes.

To distinguish among these three mechanisms, we carried out steady-state kinetic measurements of acetylation of KAN by Eis_Msm at several AcCoA concentrations and for each AcCoA concentration at several KAN concentrations. For each AcCoA concentration, we obtained the [AcCoA]-dependent observed Michaelis-Menten parameters K_m,AG and k_cat,AG (Table 4). Each of the three mechanisms of interest is characterized by a unique dependence of these kinetic parameters on the concentration of AcCoA, as described in Materials and Methods. Notably, the random sequential mechanism is the only one of the three mechanisms for which K_m,AG can increase with increasing [AcCoA] (for K_d,AcCoA(E) < K_{d,AcCoA(E•AG)} in eq 3), which is what we observe (Table 4). Therefore, either ordered mechanism is inconsistent with the data. Furthermore, the dependences of k_cat,AG and K_m,AG on [AcCoA] agree with the random sequential mechanism described by eqs 2 and 3. Fitting the data to these two equations yields the microscopic mechanistic constants k_cat = 1.6 ± 0.5 s⁻¹, K_{d,AcCoA(E•AG)} = 1,460 ± 600 μM, K_{d,AG(E•AcCoA)} = 2,540 ± 160 μM, K_d,AcCoA(E) = 87 ± 23 μM. Indeed, consistent with the above qualitative assessments, K_d,AcCoA(E) is approximately 17-fold smaller than K_{d,AcCoA(E•AG)}. Equilibrium binding constant of AcCoA to free Eis protein (K_d,AcCoA(E)) is independent of the nature of an AG molecule. Moreover, because all the AcCoA binding residues are identical in Eis_Msm and Eis_Mtb (Figure 1), it is reasonable to assume the same value of K_d,AcCoA(E) for both enzymes. This allows us to obtain microscopic constants k_cat and K_{d,AG(E•AcCoA)} or their ratio for Eis_Msm and Eis_Mtb from the data in Tables 2 and 3, as explained in Materials and Methods (Table 5).

Table 4.

Observed kinetic parameters for KAN acetylation by Eis from M. smegmatis.

[AcCoA] (μM)	Km,AG (μM)	kcat,AG (s−1)
100	278 ± 36	0.108 ± 0.005
200	450 ± 31	0.18 ± 0.01
500	756 ± 66	0.40 ± 0.02

These values indicate that generally AG binds the complex of Eis with AcCoA more weakly for Msm than for Mtb. Interestingly, this weaker binding is compensated by faster catalytic rate constants k_cat, which results in comparable catalytic efficiencies k_cat/K_{d,AG(E•AcCoA)} for the two proteins, irrespective of the nature of an AG substrate. For the two AGs used against Mtb in clinic, KAN and AMK, this effect appears to be most dramatic; they bind Eis_Mtb•AcCoA with much higher affinity than Eis_Msm, but display much higher k_cat values with Eis_Msm. Catalytic efficiencies varied in the 10–20 fold range among different AG substrates. For both Eis proteins, NET and SIS displayed the highest catalytic efficiencies among the AGs. For these two AGs, the AG affinity and k_cat values were similar for the two Eis homologs.

Inhibition of Eis Proteins

To further probe the differences in the active sites of Eis_Mtb and Eis_Msm, we next investigated with Eis_Msm the activity of the two inhibitors, chlorhexidine (1) and compound 2, that we recently identified from a high-throughput screening study with Eis_Mtb (Figure 3).(14) We previously showed that these two Eis inhibitors display high selectivity towards the Eis_Mtb AG-binding site over that of any other types of AAC enzymes. Here, we found both inhibitors of Eis_Mtb to be effective against Eis_Msm in an analogous NEO acetylation assay. Interestingly, with an IC₅₀ value of 1,848 ± 389 nM, chlorhexidine (1) appeared to be a 10-fold worse inhibitor of Eis_Msm than of Eis_Mtb (IC₅₀ = 188 ± 30 nM). In contrast, with an IC₅₀ value of 108 ± 19 nM, compound 2 appeared to be a 3-fold better inhibitor of Eis_Msm than of Eis_Mtb (IC₅₀ = 364 ± 32 nM). Consistently with the previously reported mode of inhibition for Eis_Mtb,(14) we observed that compounds 1 and 2 displayed the AG-competitive mode and the mixed mode of inhibition against NEO, respectively, when tested against Eis_Msm. The Michaelis-Menten analysis of the inhibition kinetics for compound 1 yields K_i values of 470 ± 150 nM for Eis_Msm and 105 ± 24 nM for Eis_Mtb, where K_i is the affinity of the inhibitor to the Eis•AcCoA complex. For compound 2, which displays the mixed inhibition mode, K_i and K_i′ values are 990 ± 520 nM and 320 ± 170 nM for Eis_Msm and 203 ± 47 nM and 547 ± 19 nM for Eis_Mtb, where K_i′ is the affinity of the inhibitor to the ternary complex Eis•AcCoA•AG. Thus, this compound binds with comparable affinities to the ternary complex Eis•AcCoA•AG and to the binary complex Eis•AcCoA, for both Eis_Msm and Eis_Mtb (for either Eis protein), indicating that it does not bind directly in the AG-binding site. These results demonstrate the potential of inhibitors of Eis_Mtb for use against other mycobacterial species.

DISCUSSION

Eis is an Mtb protein whose up-regulation leads to KAN resistance, a hallmark of XDR-TB, in a large set of clinical isolates from different regions of the world.(11) With its hexameric structure and its AG multi-acetylating capabilities, Eis_Mtb is an AAC that is structurally and functionally highly divergent from all other characterized AACs.(13) Eis homologs are found in a number of pathogenic and non-pathogenic mycobacteria including Mtb and Msm (Figures 1 and S1). The low 10% sequence identity among Eis homologs from ten mycobacteria presented in Figure S1 may prompt one to assume that substrate recognition rules used by Eis_Mtb may not apply to other mycobacteria. Upon closer inspection of these sequence alignments, we find that the amino acid residues involved in CoA and AG binding as well as in catalysis are highly or even strictly conserved. Therefore, we can hypothesize that the discoveries made and information gained for Eis_Mtb may find application to other diseases of mycobacterial origin. In fact, the generally minor variation of the putative AG-interacting residues sets up different mycobacteria as good systems for studying the poorly understood AG-recognition rules of Eis. For several decades, Msm has been used as a model system to study mechanisms of Mtb. Eis_Msm and Eis_Mtb share 58% sequence identify (Figure 1). All amino acid residues involved in CoA binding and catalysis are strictly conserved between Eis_Msm and Eis_Mtb. However, there are four amino acid residues (highlighted in dark blue in Figures 1 and 4) in the AG-binding pockets of these two proteins that differ.

In this work we have focused on gaining a better understanding of the evolution of the novel multi-acetylation mechanism of Eis and on probing the potential application of novel Eis_Mtb inhibitors against other mycobacterial organisms. Our steady-state kinetic measurements indicated that Eis_Msm, like Eis_Mtb, is able to efficiently multi-acetylate many AGs and that it employs the random sequential mechanism of acetylation, similarly to the majority of other mechanistically characterized AACs.(18–22) Interestingly, the binding affinities and the catalytic rate constants differed between the two enzymes for most AGs, whereas the catalytic efficiencies were very similar. Structurally very similar NET and SIS were acetylated with the highest efficiencies among AGs for both proteins and displays comparable kinetic properties. This suggests that these two AGs must interact mostly with the features of the binding pockets that are conserved between the two enzymes. Despite the overall similarities between Eis_Msm and Eis_Mtb, we observed a major variation in the acetylating activity of APR by these two enzymes. Eis_Msm di-acetylated APR whereas Eis_Mtb did not accept APR as a substrate. APR is a structurally unique AG that contains four rings, with the two central rings fused (Figure S1). This chemical structure limits the conformational freedom of this molecule and, as a result, APR has an extended cylinder-like conformation, as observed both in a crystalline form(16) and in solution.(17) Furthermore, this extended conformation of APR is essentially unchanged in several crystal structures of APR bound to RNA ligands.(26–28) Therefore, the AG-binding cavity of an enzyme needs to be sufficiently wide and long to accommodate a bound APR molecule for its acetylation. Because Eis_Mtb is unable to acetylate APR and Eis_Msm acetylates APR at two positions, the stiff APR scaffold serves as an excellent AG specificity discriminator between the two homologs. In order to explain this major specificity difference, we looked for structural features that could allow Eis_Msm to accommodate an APR molecule in the Eis_Msm, but not in the Eis_Mtb binding site. Because three out of the four primary amines of APR lie on the “terminal” rings, at least one of them is modified, which means that the AG-binding pocket should be wide and long enough to accommodate an APR cylinder with one end of it placed at the active site.

As seen in the crystal structure of Eis_Mtb in complex with CoA and an acetamide determined recently by our group,(13) the AG-binding cavity of Eis_Mtb is bifurcated, with Glu401 separating the cavity into two channels (with borders highlighted in green and in blue in Figure 4A). One channel (green) is not straight enough to fit an APR molecule without steric clashes in both Eis_Mtb and Eis_Msm. Most residues lining this channel are the same in Eis_Mtb and Eis_Msm (in Eis_Msm numbering in Figure 1: Phe26, Asp28, Trp38, and the carboxyl terminus denoted C-ter in Figure 4), In particular, it is not possible to avoid clashes between the APR and the catalytic carboxyl terminus, whose position is likely highly conserved,(13) and with Trp38 (Trp36 in Eis_Mtb). The other channel (blue) is fairly shallow in Eis_Mtb due to Glu401. In addition, this channel is capped at the end by Trp289 and Ile268 in Eis_Mtb imposing additional steric restrictions on the length of the ligand. Interestingly, all differences in the AG-binding pockets of Eis_Mtb and Eis_Msm lie in the residues that line this channel. Moreover, in Eis_Msm, the residues in this channel are smaller than those in Eis_Mtb thus likely resulting in a longer and wider channel in Eis_Msm than in Eis_Mtb and creating a cavity large enough to bind an APR molecule in a proper orientation for acetylation (Figure 4B). Specifically, Trp289 in Eis_Mtb is changed to an alanine (Ala287 in Eis_Msm) and Ile268 and Glu401 in Eis_Mtb to glycines (Gly266 and Gly401 in Eis_Msm), respectively (Figures 1 and 4). In addition, another residue lining this channel, Gln291 in Eis_Mtb is replaced by a smaller alanine (Ala289 in Eis_Msm) contributing to the channel widening. An APR molecule readily fits into this channel in the model of an Eis_Msm structure (Figure 4B). Even though the structural details of binding of APR to Eis_Msm may not all be accurately represented by this model, when taken together, our biochemical results and these structural considerations suggest that APR binding occurs in this channel.

Finally, by demonstrating that inhibitors of Eis_Mtb can also prevent acetylation of NEO by Eis_Msm, we showed that the Eis_Mtb inhibitory properties of these compounds could potentially be extended to other mycobacteria. As we demonstrated, the inhibitory power of each compound was somewhat different between the two mycobacteria, suggesting that these inhibitors would display varying potency among mycobacteria. Further investigations are currently underway in our laboratory to determine appropriate inhibitor-mycobacterium pairs. In the future, testing the activity of these inhibitors against other Eis-containing pathogenic bacteria to evaluate the extent of the application of these molecules will be of interest.

ABBREVIATIONS

AcCoA

acetyl-coenzyme A

AMK

amikacin

AG

aminoglycoside

AAC

aminoglycoside acetyltransferase

APH

aminoglycoside phosphotransferase

APR

apramycin

DTNB

5,5′-dithiobis-(2-nitrobenzoic acid)

Eis

enhanced intracellular survival

eq

equivalent or equation (depending on the context)

HYG

hygromycin

KAN

kanamycin A

LCMS

liquid chromatography mass spectrometry

Msm

Mycobacterium smegmatis

Mtb

Mycobacterium tuberculosis

NEA

neamine

NEO

neomycin B

NET

netilmicin

PAR

paromomycin

RIB

ribostamycin

rt

room temperature

SIS

sisomicin

SPT

spectinomycin

STR

streptomycin

TLC

thin-layer chromatography

TOB

tobramycin

TB

tuberculosis