Potent Inhibitors of Acetyltransferase Eis Overcome Kanamycin Resistance in Mycobacterium tuberculosis

Abstract

A major cause of tuberculosis (TB) resistance to the aminoglycoside kanamycin (KAN) is the Mycobacterium tuberculosis (Mtb) acetyltransferase Eis. Upregulation of this enzyme is responsible for inactivation of KAN through acetylation of its amino groups. A 123 000-compound high-throughput screen (HTS) yielded several small-molecule Eis inhibitors that share an isothiazole S,S-dioxide heterocyclic core. These were investigated for their structure–activity relationships. Crystal structures of Eis in complex with two potent inhibitors show that these molecules are bound in the conformationally adaptable aminoglycoside binding site of the enzyme, thereby obstructing binding of KAN for acetylation. Importantly, we demonstrate that several Eis inhibitors, when used in combination with KAN against resistant Mtb, efficiently overcome KAN resistance. This approach paves the way toward development of novel combination therapies against aminoglycoside-resistant TB.

Graphical Abstract

Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), is a major global health threat. In 2013, approximately 9.0 million people developed TB, and nearly 1.5 million died from the disease.¹ With the emergence of both multidrug-resistant TB [defined as TB with resistance to at least isoniazid and rifampin] (MDR-TB; ~ 10% of new TB cases in 2013¹) and extensively drug-resistant TB [defined as MDR-TB with added resistance to at least a fluoroquinolone and an injectable drug (i.e., kanamycin (KAN), capreomycin, or amikacin)] (XDR-TB), the need for novel strategies to combat drug resistant TB is urgent.²

Among mechanisms of clinically important transmissible drug resistance is the recently identified inactivation of an MDR-TB therapeutic, the aminoglycoside kanamycin, through its acetylation by an upregulated acetyltransferase, the Eis (enhanced intracellular survival) enzyme.³ The unique multiacetylation mechanism by which Eis inactivates KAN and other aminoglycosides has been thoroughly investigated.^4–13

One strategy to combat drug resistance arising as a result of drug-modifying enzymes is to use a combination therapy that includes an antibiotic along with an inhibitor of its associated resistance enzyme.^14–18 In Mtb, including MDR-TB, the combination of the β-lactamase inhibitor clavulanate and the β-lactam meropenem was demonstrated to overcome resistance to β-lactam antibiotics.¹⁴ With this strategy in mind, efforts have also been made toward identifying inhibitors of aminoglycoside acetyltransferases (AACs) present in nonmycobacteria, with limited preclinical progress, but these are not applicable for the mechanistically and structurally distinct Mtb Eis acetyltransferase. For example, aminoglycoside-acetyl coenzyme A bisubstrate inhibitors were found to potently inhibit AAC(6′) enzymes in vitro, but were not effective in cell-based assays.^19–22 Numerous membrane-active cationic peptides that inhibit AAC(6′) enzymes in vitro were previously identified.²³ However, these peptides displayed no synergy with aminoglycosides against resistant bacterial strains due to the inability of these membrane binding peptides to permeate into the cytoplasm. Finally, the natural product aranorosin was found to be an inhibitor of the bifunctional AAC(6′)-Ie/APH(2″)-Ia enzyme, and its combination with the aminoglycoside arbekacin was shown to stop the growth of a methicillin-resistant Staphylococcus aureus strain.²⁴

We set out to identify inhibitors of Eis that could be coadministered with KAN in order to prevent inactivation by Eis in Mtb. Here, we report the identification as well as the biochemical and biological characterization of several potent inhibitors of Eis that are able to restore the antibacterial activity of KAN in a KAN-resistant strain, Mtb K204. To clarify in atomic detail the mode of action of these compounds and their structure–activity relationships, we also determined crystal structures of EisC204A in complex with CoA and two of these inhibitors.

RESULTS

High-Throughput Screening Leads to the Identification of a Promising Eis Inhibitor Scaffold

In order to identify Mtb Eis inhibitors, we carried out a HTS of molecular libraries comprising 123 000 structurally diverse small molecules using a previously reported Eis acetylation assay in the miniaturized format (Figure 1a).²⁵ The initial HTS assay was performed using the aminoglycoside neomycin B (NEO) as a substrate, which was selected over KAN to maximize the signal-to-noise ratio under the HTS conditions. However, KAN was used in all post-HTS assays, since it is the clinically relevant aminoglycoside. The HTS yielded 617 hits containing an isothiazole S,S-dioxide heterocyclic core that were retested (Figure S1 and Table S1). Of these 617 molecules, 133 showed reproducible inhibition of Eis enzymatic activity in the same single-point HTS assay. Thirty-six of these 133 compounds were chosen for further testing in dose–response assays, based on the potency of their inhibition of Eis in the HTS assay and the structural diversity of their chemical substituents (R₁ and R₂ groups). In a previous study of Eis inhibitors evaluating fewer molecules,²⁵ we demonstrated that active compounds contained (i) a group that may become positively charged, usually an amine, or (ii) an aromatic group. The presence of either of these groups was, therefore, also a criterion for selection of compounds for dose–response testing. Additionally, we included four analogues that were not a part of the original library, but which contained the same structural core (Figure 1b). This process yielded a group of 40 molecules, for which we determined IC₅₀ values against purified Eis. We also measured the effect of these compounds on KAN MIC for KAN-susceptible (H37Rv) and KAN-resistant (K204) Mtb strains (Table 1), as described in the two following sections.

Figure 1.

(a) Schematic representation of the winnowing of ~123 000 small organic molecules to 18 showing inhibition of both Eis enzymatic activity and growth of Mtb K204 in the presence of KAN. (b) Structures of the 41 Eis inhibitors (with an isothiazole S,S-dioxide heterocyclic scaffold) for which IC₅₀ against pure Eis enzyme and MIC values against strains H37Rv and K204 of Mtb have been determined. Note: Compounds highlighted in turquoise are those for which MIC values of KAN against Mtb K204 were found to be <2.5–5 µg/mL. Compounds highlighted in gray are those for which MIC values of KAN against Mtb K204 were found to be >5–10 µg/mL. Compounds highlighted in fuchsia are those for which the X-ray structure in complex with EisC204A and CoA has been determined.

Table 1.

Eis Inhibition IC₅₀ Values of Tested Compounds and the Effect of the Compounds on Kanamycin A MIC Values against H37Rv and K204 Mtb Strains

Cpd	IC50 (µM)a	Concentration tested (µM)b	H37Rv MICKAN (µg/ mL)c	K204 MICKAN (µg/ mL)d
			1.25	10
3i	0.120 ± 0.035	12	0.625	1.25
6b	0.152 ± 0.045	15.2	0.625	2.5
7b	0.102 ± 0.034	10.2	0.625	2.5
8a	2.85 ± 0.26	184.5	0.625–1.25	5
8e	0.200 ± 0.036	20	0.625	2.5–5
11c	0.238 ± 0.089	23.8	0.625	1.25
12e	0.183 ± 0.043	18.3	0.625	2.5
13a	2.82 ± 0.21	282.3	0.625	2.5
13e	0.331 ± 0.071	33.1	0.625	2.5
13g	0.234 ± 0.060	23.4	0.625	2.5–5
13i	0.054 ± 0.002	5.42	0.625	5
14c	0.112 ± 0.008	11.2	0.625	5
15e	0.535 ± 0.045	53.5	0.625–1.25	5
15f	0.148 ± 0.016	14.8	0.625–1.25	10
17i	0.232 ± 0.038	23.2	0.625–1.25	5
33a	0.123 ± 0.023	13.2	0.625	5
35e	> 200	20	0.625	5
35g	2.49 ± 0.49	100	0.625	2.5
35h	3.23 ± 0.99	100	0.625	2.5
35i	0.985 ± 0.190	98.5	0.625	2.5
36d	0.152 ± 0.031	15.3	0.625	2.5–5
37b	0.657 ± 0.105	65.7	0.625	2.5–5
37d	> 200	20	0.625–1.25	5–10
46a	0.432 ± 0.057	43.2	0.625–1.25	10
46b	0.092 ± 0.021	9.21	0.625–1.25	5
46c	0.109 ± 0.022	10.9	0.625	2.5
46h	0.135 ± 0.031	13.5	0.625	5
52i	0.580 ± 0.096	58	0.625	2.5
62i	0.386 ± 0.080	38.6	0.625	5
81b	1.25 ± 0.22	124.8	0.625	2.5–5
81g	1.42 ± 0.42	142.3	0.625	2.5
87b	> 200	20	1.25	10
89b	> 200	20	1.25	10
112b	> 200	20	0.625	10
112i	0.621 ± 0.229	62.1	0.625	5
115i	> 200	20	0.625	10
116i	> 200	20	1.25	10
139b	> 200	20	0.625	10
139e	> 200	20	1.25	10
139i	> 200	20	1.25	10

^aThe IC₅₀ values in the Eis acetyltransferase assay.

^bAt these concentrations, these compounds did not inhibit the growth of H37Rv or that of K204 Mtb, when tested in the absence of KAN.

^cKAN MIC values for H37Rv Mtb in the absence (first line) and in the presence of each compound at the specified concentrations.

^dSame as c, but for K204 Mtb.

Structure–Activity Relationship Studies

Various R₁ and R₂ side-chains in the identified isothiazole S,S-dioxide heterocyclic scaffold (Figure 1b, Figure S1, and Table S1) were explored to improve anti-Eis potency in vitro. Based on the 617 initial HTS hit molecules, we first drew several unambiguous conclusions about the desirable structure of the R₂ side-chain. Most of the compounds for which R₂ contained an aromatic ring, a branched alkyl group, a cyclohexyl moiety, or a benzyl functionality did not efficiently inhibit Mtb Eis (as indicated by × highlighted in orange in Table S1). Among the 40 compounds that were pursued beyond the dose–response assay, compounds containing R₂ groups 8, 13, 15, 35, 46, and 81 inhibited Eis in vitro when combined with multiple R₁ substituents (as indicated by two checks and an × and three checks, highlighted in dark yellow and dark green, respectively, in Table S1), while other R₂ groups were shown to inhibit Eis in dose–response assays when in combination with only one of the possible R₁ substituents (6b, 7b, 11c, 12e, 14c, 17i, 33a, 36d, 37b, 52i, 62i, and 112i). Overall, two large families of R₂ substituents emerged as potential potent inhibitors of Eis: (i) compounds with R₂ containing two nitrogen atoms separated by three carbon atoms (structures 3, 6–8, 11–15, and 17), particularly compounds with large bulky groups or a cyclohexyl ring attached to the extended nitrogen atom were most often inhibitory, and (ii) compounds containing an R₂ group with two nitrogen atoms separated by two carbon atoms also efficiently inhibited Eis (structures 33, 35–37, and 112). In this series of molecules, only compounds with a nitrogen atom located in a cyclohexyl ring were inhibitory.

We next explored the effect of the R₁ substituents on Eis inhibition. Upon initial inspection of the 617 HTS hits, the identity of the R₁ substituent appeared to have little effect on their Eis inhibitory activity. However, when analyzing these side chains statistically, patterns emerged. The p-fluorophenyl group (b) had the highest percentage (34%) of compounds inhibiting Eis in the initial HTS; this was followed closely by the p-methylphenyl group (e, 33%) and the m,p-dimethylphenyl group (i, 30%), suggesting that these three groups as R₁ substitutions have a better chance of contributing positively to Eis inhibition. The next best R₁ substituent was the p-chlorophenyl group (c) with 27% of the compounds within this group displaying inhibition of Eis. The m-chlorophoenyl group (g) and p-anisole group (d) were next with 19% and 18% of their compounds displaying Eis inhibition, respectively. Finally, those with 13%, 12%, and 9% of compounds showing Eis inhibition were compounds containing the p-isopropylphenyl (f), the m-bromophenyl (h), and the phenyl (a) group as R₁, respectively.

Having established general trends for the R₁ and R₂ substituents by analyzing the data from Table S1, we next focused on the measured in vitro potency (IC₅₀) of the 40 selected compounds (Table 1). Several trends consistent with those established above emerged from these quantitative data. With the exception of compound 112i, monosubstituted R₂ amine substituents comprised of a straight alkyl chain (139b, 139e, 139i), an aromatic ring (87b, 89b, 112b), or an amide functionality (115i, 116i) did not inhibit purified Eis. Compounds with R₂ substituents containing a diamine separated by two carbon atoms with the second amine present in a cyclohexyl ring all displayed good to moderate Eis inhibition. However, no conclusion could be formed as to what type of cyclohexyl ring was best, an unsubstituted (33a), a morpholino (35e, g, h, and i), or a piperazinyl (36d or 37b) ring, as most IC₅₀ values were similar for these compounds. Compound 13i with a diamine separated by three carbons at the R₂ position and the m,p-dimethylphenyl group at the R₁ position was found to be the most potent Eis inhibitor (IC₅₀ = 0.054 ± 0.002 µM). Interestingly, compound 46b with a cyclohexyl ring directly attached to the core scaffold followed by a second piperazine ring in the para position was found to be the second best Eis inhibitor (IC₅₀ = 0.092 ± 0.021 µM). These results indicate that either a rotationally free alkyl chain (as in compound 13i) or cyclically restricted alkyl linkers could be useful in designing Mtb Eis inhibitors.

To establish whether the inhibitors identified were specific to Eis over AACs from other families, compounds 46b and 46c were tested against AAC(6′)-Ie/APH(2″)-Ia from Staphylococcus aureus,²⁶,²⁷ AAC(3)-IV from E. coli,²⁸ and AAC(2′)-Ic from Mtb.²⁹,³⁰ Under optimal reaction conditions for each enzyme, no inhibition by either compound was observed with any of these AACs (Table S2). These data demonstrate that the compounds are exquisitely selective for Mtb Eis over other AACs.

Eis Inhibitors Abolish Eis-Mediated Resistance of Mtb to KAN

Having identified inhibitors of the purified Mtb Eis enzyme, we next determined their activity in cellulo. Mycobacteria are notorious for their waxy cell wall that is difficult to penetrate for small molecules. Therefore, we anticipated that not all of the compounds that exhibited inhibition of purified Eis would be effective in cell cultures. We measured the effect of the 40 selected compounds (at a concentration of 100-fold higher than their IC₅₀ values when possible) on KAN MIC values against Mtb K204, a KAN-resistant strain bearing a clinically observed eis promoter mutation known to upregulate Eis production (Table 1). In the KAN-susceptible H37Rv strain, the MIC value of KAN is 1.25 µg/mL, whereas in the KAN-resistant K204 strain, the MIC value of KAN is 10 µg/mL. As expected, compounds that did not inhibit purified Mtb Eis (35e; 37d; 87b; 89b; 112b; 115i; 116i; and 139b, 139e, and 139i) did not overcome resistance to KAN in Mtb K204. Twenty-eight of the remaining 30 Eis inhibitors resulted in at least a 2-fold reduction in KAN MIC against Mtb K204, with two of these compounds, 3i and 11c, lowering KAN MIC 8-fold, down to the level of the KAN-susceptible H37Rv strain. Notably, these compounds did not have any effect on the growth of either mycobacterial strain in the absence of KAN. Therefore, these compounds act synergistically with KAN against KAN-resistant Mtb. The compounds that reduced the MIC value of the KAN-resistant Mtb strain 4-fold had IC₅₀ values that ranged from 0.102 ± 0.034 to 3.23 ± 0.99 µM. Generally, in vitro (Eis inhibition) and in cellulo (KAN-sensitization) activities were correlated among the compounds tested in both assays. This correlation, together with the lack of inhibition of cell growth in the absence of KAN, serve as strong evidence of inhibition of KAN acetylation by Eis as the principal mechanism of action in the mycobacterial cell. Not unexpectedly, among very potent inhibitors of the enzymatic activity of Eis in vitro, a small fraction of the compounds had little or no effect at sensitizing the resistant Mtb strain K204 to KAN. The lack of activity in the cells was likely due to the inability of these compounds to go through the cell envelope or due to their aggregation in the cell culture media at the concentrations that were much higher than those used in the enzymatic assay.

Crystal Structures of Eis-Inhibitor Complexes Reveal the Inhibition Mechanism

To elucidate the mechanism of Eis inhibition by the isothiazole S,S-dioxide heterocyclic core-containing compounds, rationalize the observed SAR, and guide future rational inhibitor design, we determined 2.2-Å-resolution crystal structures of Eis in complex with CoA and inhibitors 13g and 11c (Figure 2a,c), which displayed both potent inhibition of Eis acetylation in vitro and the KAN resistance abolishing effect in the Mtb cultures. Both inhibitors and their chemical features were clearly distinguishable and modeled into strong F_o−F_c electron density (Figure 2b,e). The only exception was the 1,2,3,4-tetrahydroisoquinoline group of 11c, for which only partial electron density was observed (Figure 2e), likely due to a somewhat dynamic behavior of this highly hydrophobic group in a partially polar environment, as allowed by the flexible diamine linkage. No other significant difference electron density that could be attributed to a bound inhibitor was observed elsewhere. Both inhibitors exhibited high steric complementarity with the Eis surface. The binding site for both inhibitors overlapped with the aminoglycoside substrate-binding site, as was recently established by the structure of Mtb Eis in complex with CoA and tobramycin (Figure 2c,f; Figure S2).¹⁰ In addition, the R₂ groups of both inhibitors reached the C-terminal carboxyl group of Eis, which was proposed to serve as the general base in the acetyl transfer reaction. Therefore, these crystal structures indicate that the inhibitors block access of aminoglycosides to the active site of the enzyme. Explaining the critical role of the isothiazole S,S-dioxide heterocyclic core, this group is bound in the same location and orientation in both Eis-inhibitor structures (Figure 2). Remarkably, despite the nearly identical positions of this group, there are differences in how this group interacts with the protein for the two inhibitors. These differences are due to significant conformational changes in the region spanning residues 26–31 of Eis (Figure 2b,e), which uniquely adapts to bind the two structurally different R₂ groups. For inhibitor 13g (Figure 2a,b,c), one of the sulfonyl oxygens forms a direct hydrogen bond with the main chain amide nitrogen of Phe84, and the ring nitrogen forms a water-mediated hydrogen bond with the main chain amide nitrogen of Ile28. In this case, the Eis conformation resembles that observed in a previously reported crystal structure with the aminoglycoside-binding pocket unoccupied.⁴ Inhibitor 11c does not make any water-mediated contacts due to a small positional shift of the inhibitor and a conformational change of Eis. Instead, the two sulfonyl oxygens are at a distance of ~4 Å, consistent with the stabilizing van der Waals and electrostatic interactions. The isothiazole ring is sterically sandwiched between the side chains of Ile28 and Ser83 in one direction and between the side chains of Phe24 and Trp36 in the orthogonal direction. The R₁ groups of the two compounds are almost coplanar and fit neatly into a nearly fully nonpolar environment of side chains Leu63, Trp13, and Ala33 and the aliphatic stems of Arg37, Met65, and Phe84, explaining the nonpolar phenyl ring with small substitutions as effective R₁ groups (Figure 1). Such an environment, devoid of hydrogen bond donors or acceptors, is appropriate for accommodating halogen substituents, as is the case with 11c and 13g. The halogens can be accommodated in the para and/or one, but not both, meta positions. Substitutions at both meta positions would not be accommodated, as reflected in the list of potent inhibitors (Figure 1), since one of them would clash sterically with the side chain of Met65. In contrast with R₁, the conformations of the R₂ groups of the two inhibitors are drastically different with one exception—the positively charged ring nitrogen of R₂ occupies nearly the same location where it forms a salt bridge with the carboxyl group of Glu401 in both structures. Besides this interaction, the R₂ group of compound 13g makes steric and weak electrostatic interactions with His119, Ser121, Glu401, and the C-terminal carboxyl group (weakly electrostatically interacting with the stem N) on one side and makes water-mediated steric interactions with region 26–31 on the other side. In contrast, for the R₂ of the compound 11c, while the interactions with one face of this group are made with the same amino acid residues, the other face interacts directly through hydrophobic and steric interactions with the region 26–31, which is in a different conformation. Most prominently, the side chain of Asp26 is flipped to interact with the stem, and Phe24 orthogonally stacks against the double ring. Because the R₁ group abuts the hydrophobic wall, whereas the R₂ group points toward the more extended end of the substrate binding channel, there is more variability in both the identity and the size of R₂ substituents, as exhibited in the list of effective R₂ groups (Figure S1).

Figure 2.

Crystal structures of EisC204A from Mtb in complex with 13g (a and b) and 11c (d and e). The overall views of the Eis monomer with the inhibitors bound (a and c) and zoom-in views of the active site (b and e) show that the inhibitors (green sticks with atoms labeled as follow: C = green, N = blue, O = red, S = orange, and Cl = dark green) occupy a site overlapping with the aminoglycoside substrate-binding site. The F_o−F_c omit map contoured at 3σ is shown by the purple mesh. Inhibitor interacting residues are shown as orange sticks (with N and O in blue and red, respectively). Water molecules mediating inhibitor binding are shown as navy blue spheres. The side chain of Asp26 and the backbone of this residue and its neighbors adopt different conformations in the two complexes. The CoA molecule is not shown in panels a, b, d, and e to simplify the view. Superposition of these respective inhibitor structures (in green) with the previously reported crystal structure of EisC204A (pale orange) in complex with CoA and tobramycin is shown in panels c and f. The tobramycin is shown in red, and the ordered part of CoA (from the structures of Eis-inhibitor-CoA complexes) is in blue. The CoA molecule from the complex of Eis with tobramycin is bound similarly and is not shown.

DISCUSSION

While it has been met with initial skepticism, target-based rational drug design is now gaining momentum and is widely used in industrial and academic drug discovery pipelines, due to advances in HTS technology, robust assay development practices, and availability of large and diverse chemical libraries. Prominent examples of successful rational drug design are HIV protease inhibitors ritonavir and the anticancer drug Gleevec. Here, we report the discovery and initial preclinical development of novel first-in-class isothiazole S,S-dioxide heterocyclic inhibitors of a unique acetyltransferase Eis from Mtb as agents that overcome one mechanism of resistance to KAN in Mtb. Among the diverse pool of compounds tested, we found agents that we validated as potent inhibitors of Eis both in the test tube and in the relevant Mtb cultures. Using a ~5-fold larger and more structurally diverse compound library than that used in our initial effort²⁵ yielded compounds that were 10-fold and more potent than those reported in that previous study. Potencies in the mid-nanomolar range appear to be achieved through binding of the compounds via steric complementarity, hydrophobic, and hydrogen bonding interactions. While one substituent group (R₁) is relatively unchangeable in terms of its size and physicochemical properties, the other group (R₂) can vary greatly. The crystal structures of Eis-inhibitor complexes demonstrate that binding to variable structures is accompanied by conformational plasticity of the aminoglycoside-binding site of Eis, which adapts to different R₂ groups, while the protein–R₁ interactions are rigid body-like. Binding of the two inhibitors of Eis investigated structurally differs not only in protein conformation but also in the presence or absence of water-mediated contacts. Interestingly, the exquisite selectivity and potency of Gleevec toward its target Abl kinase have been recently elegantly shown to be a consequence of an induced-fit mechanism, where Abl undergoes unique conformational changes that stabilize Gleevec in the binding site of this kinase and result in a nanomolar affinity.³¹ Our Eis inhibitors also appear to be selective in that they do not inhibit other AAC enzymes. This selectivity must be due to the structurally different substrate binding pocket of Eis from that of regiospecific AAC enzymes. Indeed, we demonstrated previously that even though Mtb Eis shares main catalytic residues with Eis homologues from different bacteria and even AAC(2′)-Ic, their substrate binding pocket topographies and the surface charges differ dramatically.¹² For example, the C-terminal domain of Eis contributing to the shape of the substrate binding pocket (and containing inhibitor interacting residue Glu401 and the C-terminal carboxyl group) is absent in the single-domain AAC(2′)-Ic and other AAC enzymes.⁴

Aminoglycosides are among the antibiotics of last resort for MDR and XDR-TB patients, and resistance to them strongly decreases the chances of a favorable treatment outcome. The strategy of using KAN in combination with an Eis inhibitor is meant not only to overcome KAN resistance due to Eis upregulation, but also to curb emergence of new resistant strains by mutagenesis of the eis promoter, as such mutation would not have a survival benefit in the face of combination therapy. This study not only validates AACs as drug targets, but it sets an important precedent for HTS-driven discovery for AACs that cause resistance in clinically useful non-TB pathogens, such as AAC(3) and AAC(6′) in Klebsiella pneumoniae.^32–34 As with TB, such inhibitory agents could play an important role in both overcoming existing resistance and curbing the acquisition of resistance.

Conclusions

In conclusion, we identified potent inhibitors of Eis enzymatic activity with resulting sensitization of KAN-resistant Mtb cells, in which the resistance to KAN is caused by Eis upregulation. The inhibitors bind in the aminoglycoside binding pocket, blocking the access of aminoglycosides to the active site of the enzyme. The inhibitor binding is accompanied by conformational changes of the protein. These compounds have great potential for further development as KAN adjuvants in Mtb.

EXPERIMENTAL SECTION

Detailed versions of the abbreviated protocols below are provided in the Supporting Information.

Protein, Reagents, and Small-Molecule Libraries

The wild-type Eis from Mtb,⁴ EisC204A,¹⁰ AAC(6′)/APH(2″),³⁵ AAC(3)-IV,³⁵ and AAC(2′)-Ic⁴ were overexpressed and purified as previously reported, with minor modifications. All reagents were purchased from Sigma with the exception of albumin-dextrose-catalase, which was from BD Biosciences. Eis was screened at the Center for Chemical Genomics (CCG, University of Michigan) against 123 000 compounds from three libraries. Compounds were retested as fresh powder samples purchased from ChemDiv.

Eis Chemical Library Screening and Hit Validation

Screening of small-molecule libraries, hit validation, and dose response analysis (represented by Figure S3) were performed as previously described.²⁵

Inhibitor Selectivity

Compounds 46b and 46c were tested against three additional AAC enzymes: AAC(6′)/APH(2″), AAC(3)-IV, and AAC(2′)-Ic at optimal reaction conditions for each enzyme (Table S2).

Mycobacterial MIC Determination by Alamar Blue Assay

The assay was done with Mtb strains H37Rv and K204 in 7H9 broth. Compounds were tested at either 100× IC₅₀ (if known) or 20 µM for their inhibitory effect on Mtb growth and on KAN MIC in these two strains.

Crystallization, Diffraction Data Collection, and Structure Determination and Refinement of EisC204A-CoA-inhibitor Complexes

Crystals were grown by the hanging drop method and then underwent a complex harvesting process. The cryoprotectant solution contained 0.5 mM inhibitor. The diffraction data were collected at synchrotron beamline 22-ID of the Advanced Photon Source at the Argonne National Laboratory (Argonne, IL) at 100 K. The data were processed and the structures were built and refined according to previously described methods.¹⁰ The data collection and structure refinement statistics are given in Table S3.

ABBREVIATIONS

AAC

aminoglycoside acetyltransferase

CoA

coenzyme A

Eis

enhanced intracellular survival

HTS

high-throughput screening

KAN

kanamycin

MDR

multidrug-resistant

MIC

minimum inhibitory concentration

Mtb

Mycobacterium tuberculosis

NEO

neomycin B

TB

tuberculosis

XDR

extensively drug-resistant