Potent 1,2,4-Triazino[5,6b]indole-3-thioether Inhibitors of the Kanamycin Resistance Enzyme Eis from Mycobacterium tuberculosis

Abstract

A common cause of resistance to kanamycin (KAN) in tuberculosis is overexpression of the enhanced intracellular survival (Eis) protein. Eis is an acetyltransferase that multiacetylates KAN and other aminoglycosides, rendering them unable to bind the bacterial ribosome. By high-throughput screening, a series of substituted 1,2,4-triazino [5,6b]indole-3-thioether molecules were identified as effective Eis inhibitors. Herein, we purchased 17 and synthesized 22 new compounds, evaluated their potency, and characterized their steady-state kinetics. Four inhibitors were found not only to inhibit Eis in vitro, but also to act as adjuvants of KAN and partially restore KAN sensitivity in a Mycobacterium tuberculosis KAN-resistant strain in which Eis is upregulated. A crystal structure of Eis in complex with a potent inhibitor and CoA shows that the inhibitors bind in the aminoglycoside binding site snugly inserted into a hydrophobic cavity. These inhibitors will undergo preclinical development as novel KAN adjuvant therapies to treat KAN-resistant tuberculosis.

Graphical Abstract

Mycobacterium tuberculosis (Mtb), the etiological agent of tuberculosis (TB), is a dominant worldwide threat to human health.^1,2 The evolution of resistance of this bacterium to conventional antitubercular drugs has exacerbated the problem.^3,4 This is particularly concerning as in the last 50 years only two new antitubercular agents, bedaquiline (Food and Drug Administration (FDA), USA) and delamanid (European Medicines Agency, EU), were approved by these agencies for treatment of multidrug-resistant tuberculosis (MDR-TB) and extensively drug-resistant tuberculosis (XDR-TB), respectively.⁵ The need to combat drug-resistant TB remains a pressing issue, and work toward developing new antitubercular agents continues to be at the forefront of TB research. Several directions are being pursued in TB drug development: (i) investigation of chemical modifications of current TB treatments (e.g., isoniazid and rifampin), (ii) development of agents inhibiting already exploited targets, but capable of avoiding the resistance mechanisms (e.g., KatG and topoisomerase mutations), (iii) development of inhibitors of novel pathways or enzymes that are specific to Mtb or bacteria in general (e.g., mycolic acid pathway, amino acid biosynthesis, PanK, and dihydrofolate reductase), and (iv) development of inhibitors of resistance enzymes (e.g., BlaC and Eis) as adjuvants of currently approved antitubercular drugs.^6,7

Aminoglycosides (AGs) have been used extensively to treat a broad spectrum of bacterial infections since the discovery of streptomycin in the 1950s. The injectable AGs, kanamycin (KAN) and amikacin (AMK), have been effcacious therapeutic agents against Mtb strains for decades. However, KAN- and AMK-resistant Mtb strains have emerged. Mtb has three main mechanisms to overcome the action of KAN and AMK: (i) increased expression of the enhanced intracellular survival (eis) gene as a result of a mutation in its promoter,⁸ (ii) increased expression of Eis and the effux pump, Tap, due to upregulation of transcriptional activator WhiB7,⁹ and (iii) ribosome mutations that interfere with AG binding.¹⁰ Upregulation of the eis gene by either mechanism (i) or (ii) confers resistance to KAN in one-third of KAN-resistant clinical isolates.^8,9 Eis is a unique AG acetyltransferase (AAC) due to its distinct structure and ability to multiacetylate AGs at several different positions including the 3″-amine of 4,6-disubstituted 2-deoxystreptamine AGs and 4‴-amine of AGs containing an (S)-4-amino-2-hydroxybutyryl group that are not modified by other AACs.^11,12 It is also of note that one AG modification may not completely abolish the antibacterial activity of an AG, whereas two or more modifications fully eliminate this activity.¹³ Eis has a larger and more complex active site than that of other AACs, which results in its broader substrate profile and multiacetylating activity.^11,12,14 The kinetics of AG acetylation by Eis have been studied, and this bisubstrate enzyme was-found to operate by a random sequential mechanism.¹⁵

Two potential strategies that may be envisioned to overcome the resistance caused by Eis are (i) identification of new AGs that are not susceptible to inactivation by Eis, and (ii) discovery of structurally unrelated inhibitors of Eis, which would sensitize the KAN-resistant bacteria to KAN when administered as a combination therapy. Because Eis can acetylate multiple amines on clinically relevant AGs of different structures, the first strategy is not likely to be productive. We recently demonstrated that the second strategy is a useful approach by discovering several structural classes of potent Eis inhibitors and demonstrating that they indeed target Eis in Mtb cells.¹⁶⁻²⁰ Diversifying a drug pipeline with different structural classes early on in the drug development process increases the chances of one of the compounds progressing to clinical applications. Development of drugs that need to act in the Mtb cytoplasm is especially complicated by the requirement for these molecules to cross the extremely waxy Mtb envelope as well as the membrane of the macrophage, which harbors persistent Mtb infections. Therefore, we sought a structural class of Eis inhibitors that would be distinct from those that we discovered previously.

In this study, we present novel 1,2,4-triazino[5,6b]indole-3-thioether-based Eis inhibitors. The structures of these inhibitors are unique as they contain a tricyclic core and a long flexible linker, which allows access to novel binding surfaces of Eis that were unexplored in our previous Eis inhibitor studies. We also introduce a simple and easily scalable synthetic route to derivatize the 1,2,4-triazino[5,6b]indole-3-thioether scaffold. We present kinetic and crystallographic analyses of these inhibitors as well as describe their biological activity as KAN adjuvants. We also perform a thorough comparison of the interactions of this scaffold with those of four previously published divergent Eis inhibitor classes.

RESULTS AND DISCUSSION

Identification of Lead Compounds and Synthetic Optimization.

Inhibitors of Eis were sought by a high-throughput screening (HTS) assay against a collection of ~23 000 compounds from three different libraries: the BioFocus NCC library (~1000 compounds), the ChemDiv library (~20 000 compounds), and the MicroSource MS2000 library (~2000 compounds).¹⁶ The average Z′ score for the HTS was 0.65, indicating assay robustness. Forty-six compounds with a 1,2,4-triazino[5,6b]indole-3-thioether core substituted at three different positions were present in the HTS (Figures 1 and 2). Of these 46 compounds, seven molecules (14a, 23a, 23c, 37b, 37d, 39b, and 40b; Figure 2), some of which displayed Eis inhibition, were selected for additional investigation (Table 1). Four (14a, 37b, 37d, and 39b) of these seven molecules were confirmed to be Eis inhibitors by the dose-response assays. Seventeen additional compounds (Figure 2) were purchased to further explore the utility of this scaffold. On the basis of IC₅₀ data obtained for the commercially available compounds against purified Eis (Table 1), 22 new derivatives (Figure 2) were synthesized. Two of the purchased molecules (37b and 39b) were resynthesized as controls to establish the validity of the data obtained from externally procured materials (Scheme 1). As neomycin B (NEO) was found to be one of the best substrates for Eis in a previous study,^11,16 the HTS was performed with NEO as a substrate for optimal signal-to-noise ratio relative to the background hydrolysis of acetyl coenzyme A (AcCoA). However, as KAN is the clinically relevant substrate for inactivation by Eis in Mtb, we determined IC₅₀ of our synthesized compounds with KAN as a substrate (IC_50,KAN).

Figure 1.

Stepwise cone diagram showing the winnowing of ~23 000 compounds to the four Eis inhibitors that restored KAN activity in Mtb K204.

Figure 2.

Structures of all molecules used in this study along with their origins. Note: At the bottom of the figure, the compound numbers listed in orange indicate that the compounds were tested in the HTS only. Those listed in pink were tested in the HTS and repurchased for validation. Those listed in gray were compounds that were not present in the HTS but were purchased to expand the preliminary SAR study. Those in green were new derivatives synthesized to further expand our SAR study. Compound 37b is in purple as it was present in the HTS, repurchased, and resynthesized to confirm the values obtained for purchased material. Compound 39b is in blue as, not only was it present in the HTS, repurchased, and resynthesized, but was additionally crystallized with Eis. All compounds (except those listed with orange numbers) had an IC₅₀ or/and MIC_KAN value determined as presented in Table 1.

Table 1.

IC₅₀ Values of Synthetic Eis Inhibitors and Compounds Purchased or from HTS with Activity under 28 μM and MIC_KAN Values for Mtb H37Rv and Mtb K204 When Treated in Combination with Inhibitor at Listed Concentration

compound	IC50,KAN (μM)	IC50,NEO (μM)	H37RV MICKAN (μg/mL)	K204 MICKAN (μg/mL)	concentration tested (μM)
no inhibitor			<1.25	>10
9a		5.8 ± 0.8	<1.25	5	100
10k		28 ± 4	<1.25	10	100
11b	0.23 ± 0.02	0.64 ± 0.15	<1.25	5	23
11e	0.27 ± 0.05	1.1 ± 0.2	<1.25	5–10	27
11d		2.4 ± 0.5	<1.25	10	100
11e		4.4 ± 0.4	<1.25	5–10	100
14a	1.5 ± 0.1	5.3 ± 1.0	<1.25	5	100
16a		3.9 ± 0.9	<1.25	10	100
23a		>200
23c		>200
24a		>200
25a		13 ± 2	<1.25, <1.25	10	100
30b		>200
31a		>200
32b	>200
33b	>200
34b	0.056 ± 0.006		<1.25	5	100
				5	5.6
35b	>200
36b	0.43 ± 0.04	0.26 ± 0.07	<1.25	2.5–5	43
37b	0.17 ± 0.04	0.39 ± 0.14	<1.25	2.5	17
37d	0.8 ± 0.1	0.21 ± 0.03	<1.25	5	80
37g	0.43 ± 0.06	0.5 ± 0.1	<1.25	5–10	43
37b	1.3 ± 0.1
37i	0.56 ± 0.03	1.7 ± 0.5	<1.25	10	56
38b	0.07 ± 0.01	0.13 ± 0.04	<1.25–2.5	5–10	7
39b	0.030 + 0.005	0.15 ± 0.02	<1.25	2.5–5	13
39c	0.14 ± 0.01	0.18 ± 0.02	<1.25	5	14
39i	0.05 ± 0.02	0.30 ± 0.14	<1.25	>10	5
40b		>200
41d		>200
42b	>200
43b	0.10 ± 0.02	0.25 ± 0.03			insoluble
43d	1.1 ± 0.2		<1.25	2.5–5	100
				10	10
44b	3.3 ± 0.7		<1.25	>10	100
45b	0.17 ± 0.03		<1.25–2.5	5–10	17
46b	0.14 ± 0.04				insoluble
47b	>200
48b	0.17 ± 0.04	0.51 ± 0.03			insoluble
48c	0.6 ± 0.1
48d	0.15 ± 0.02		2.5	5–10	15
48f	1.2 ± 0.2		1.25	10	100
48b	0.54 ± 0.13
49b	1.0 ± 0.1		<1.25	5	100
50d	0.08 ± 0.02	0.19 ± 0.07	<1.25	10	8
52	0.23 ± 0.03	0.30 ± 0.04	<1.25	5	100
				10	23
53	0.71 ± 0.08	2.27 ± 0.03	<1.25	5–10	71

Scheme 1.

Synthetic Schemes Used for the Synthesis of 24 Compounds Generated in This Study

The synthesis of all compounds started with the N-alkylation with alkyl halides of commercially available compounds 54 and 55 to afford in quantitative yields compounds 56 and 57, respectively. The resulting products were mixed with thiosemicarbazide in the presence of potassium carbonate or cesium carbonate, followed by acidification with concentrated hydrochloric acid to generate the 1,2,4-triazino[5,6-b]indole-3-thione or hydrazinecarbothiamide analogues 58−59 in 22−52% yields. Subsequently, intermediates 58−59 were reacted with various alkyl halides to obtain the desired S-substituted products (32 and 34−49) with a fluoro or a methyl group at the C8-position (indicated as X in Scheme 1) in 14−82% yields. The free hydroxyl group of 32b was methylated in the presence of iodomethane and sodium hydride to generate 33b in 47% yield. Compound 58b was reacted with bromoacetonitrile to afford the desired molecule 47b in 63% yield. Compounds 52−53 were prepared in three steps from 58b.S-Methylation of compound 58b with iodomethane afforded 60b (91% yield), which was oxidized with m-CPBA to give sulfone 61b (45% yield). Compound 61b was reacted with different N,N-dialkylaminoalkylamines to generate the desired compounds 52 and 53.

Structure−Activity Relationship (SAR) Study.

Inhibitory activity (IC₅₀) against purified Eis (Tables 1 and S1 and Figures S79–S85) was determined for a total of 46 compounds. Examination of the activity of this series of derivatives of the 1,2,4-triazino[5,6b]indole-3-thioether scaffold (Tables 1 and S1, Figure 2) revealed several SAR insights. Our chemical library can be divided into four chemical series (I, II, III, and IV; Figure 2). Series I contains the 1,2,4-triazino[5,6b]indole-3-thioether core with a hydrogen atom at the C8-position, while series II and III contain a fluoro and a methyl substituent at this position, respectively. In series IV, a subgroup of series II, we replaced the thioether at the C3-position of the core by secondary amines in an effort to understand the biological importance of a sulfur atom at this position.

We first assessed the importance of the substituent at the C8-position (H “series I” vs F “series II” vs methyl “series III”). We compared the Eis inhibitory activity of compounds from series I, II, and III, with matching R₁ and R₂ groups when using KAN as the AG substrate (IC_50,KAN). To understand whether the fluoro substitution was beneficial, we compared compounds from series I and II. We looked at compound 11b (series I, X = H, R₁ = methyl, and R₂ = 2-(piperidin-1-yl)ethyl; IC_50,KAN = 0.23 ± 0.02 μM) and its C8-fluorinated counterpart 39b from series II (IC_50,KAN = 0.030 ± 0.005 μM) and noticed that compound 39b was much more potent. Similarly, we observed that compound 11c (series I, X = H, R₁ = ethyl, and R₂ = 2-(piperidin-1-yl)ethyl; IC_50,KAN = 0.27 ± 0.05 μM) and its C8-fluorinated counterpart 39c from series II (IC_50,KAN = 0.14 ± 0.01 μM) displayed a similar trend, with the fluoro-substituted compound 39c being about twice as potent. These observations showed that adding a fluorine atom at the C8-position increased the potency of Eis inhibition. Next, we explored the effect of replacing the C8-hydrogen of series I or the C8-fluoro of series II by a methyl group (series III). Compounds 48d (series III, X = methyl, R₁ = n-propyl, and R₂ = 2-(N,N-diethylamino)ethyl; IC_50,KAN = 0.15 ± 0.02 μM) and its R₁ = N5-i-pentyl counterpart 48h (IC_50,KAN = 0.54 ± 0.13 μM), both with a methyl at their C8-position, were slightly more potent than their direct respective counterparts with a fluorine at the C8-position, 37d (IC_50,KAN = 0.8 ± 0.1 μM) and 37h (IC_50,KAN = 1.3 ± 0.1 μM). On the other hand, in the case of compounds 37b (series II, X = F, R₁ = methyl, R₂ = 2-(N,N-diethylamino)ethyl; IC_50,KAN = 0.13 ± 0.03 μM) and its C8-methylated counterpart 48b (IC_50,KAN = 0.17 ± 0.04 μM) from series III, the IC_50,KAN values were virtually the same. Overall, we observed that compounds with a methyl substituent at their C8-position displayed the same or more potent activity when compared to those with a fluoro substituent, which in turn were more potent than the C8-unsubstituted compounds.

Having analyzed the effect of three substituents at the C8-position of the 1,2,4-triazino[5,6b]indole-3-thioether scaffold, we next explored the effect of modifying the N5-position with 11 different chemical moieties (indicated by the letters a−k following the compound numbers in Figure 2). We varied the diversity at N5 by not substituting the amine (a) or by introducing various linear alkyl chains (b−e and g), branched alkyl chains (f and h), alkylaryl groups (i and j), and an amido group (k). To establish the favorable N5-substituents, as we did during our analysis of the C8-substituent, we compared pairs of compounds that only differed in their N5-substitutent. We first looked at the N5-unsubstituted compound 11a (Table S1, series I, X = H, R₁ = H, and R₂ = 2-(piperidin-1-yl)ethyl), which was found to not be an inhibitor in our HTS. We next investigated the effect of adding linear alkyl chains at the N5-position. For compounds 11b−d, systematically increasing the size of the R₁ substituent from methyl (b) to ethyl (c) and n-propyl (d) led to two-fold incremental decrease in Eis inhibitory activity (IC5_0,NEO = 0.64 ± 0.15 μM, 1.1 ± 0.2 μM, and 2.4 ± 0.5 μM, respectively). This relationship between the size of R₁ and Eis inhibitory activity did not perfectly translate to compounds 37b−d (X = F, R₂ = 2-(N,N-diethylamino)ethyl), 39b,c (X = F, R₂ = 2-(piperidin-1-yl)ethyl), or 48b−d(X = Me, R₂ = 2-(N,N-diethylamino)-ethyl). However, we noticed by looking at IC50,KAN values that, generally, a methyl group (R₁ = b) at the N5-position of most compounds resulted in the best inhibitors as demonstrated with compounds 11b, 37b, 39b, 43b, and 48b. When examining the effect of branched alkyl chains (f and h) at the N5-position, we observed that these moieties typically decreased or abolished inhibition of Eis. For the alkylaryl (i and j) and amido (k) functionalities, a concrete conclusion could not be drawn from our data due to a small sample size and the limited diversity of the alkylaryl and amido functionalities studied. It is important to note that we did not further explore the alkylaryl and amido functionalities as R₁ substituents because the resulting compounds were not as good as those with the alkyl substituents at that N5-position. Overall, we found that the methyl substituent (R₁ = b) was the most favored out of all N5-modifications tested.

Next, we studied the effect of modifications at the C3-position of the 1,2,4-triazino[5,6b]indole-3-thioether scaffold. The substituents at the C3-position were more diverse than those at the C8-and N5-positions, as they contained various heteroatoms. Examination of compounds 1−6, 41d, and 42b led us to conclude that any C3-side chain containing only carbon and hydrogen atoms (e.g., aromatic, unsaturated, or aliphatic chains) resulted in compounds with no Eis inhibitory activity. Likewise, substituents at the C3-position containing oxygen atoms (e.g., carbonyl (compounds 13 and 17−30), hydroxy (compound 32b), ether (compounds 8a and 33b), morpholine (compounds 12d, 12e, 12j, 15a, 40b, 40c, 40d, and 51d)), or a cyano group (compounds 31a and 47b) were not tolerated. However, we observed that, with the exception of the morpholino and cyano groups, C3-side chains containing a tertiary nonaromatic nitrogen atom located 3−4 bonds away from the sulfur atom yielded potent Eis inhibitors (Figure 2). We then closely examined the steric effects of these tertiary amines on Eis inhibition. Through comparisons of pairs of compounds that differ by a 2-versus 3-carbon linker, such as compounds 11a and 14a, 36b and 43b, as well as 39b and 45b, we found that, with the exception of 11a, the length of the linker did not greatly influence the IC₅₀ values against purified Eis. To further evaluate the steric effects on Eis inhibition, compounds 36b (R₂ = 2-(N,N-dimethylamino)ethyl; IC_50,KAN = 0.43 ± 0.04 μM), 37b (R₂ = 2-(N,N-diethylamino)ethyl; IC_50,KAN = 0.17 ± 0.04 μM), 38b (R₂ = 2-(pyrrol-1-yl)ethyl; IC_50,KAN = 0.07 ± 0.01 μM), and 39b (R₂ = 2-(piperidin-1-yl)ethyl; IC_50,KAN = 0.030 ± 0.005 μM) were assessed. In these instances, we recognized that increasing the number of carbons attached to the nitrogen atom from the dimethyl- to the diethylamino functionality correlated with increased Eis inhibitory activity as evidenced by comparison of 36b and 37b. Further cyclization of the diethylamino to the pyrrolidinyl moiety led to more potent Eis inhibitors as seen when comparing 37b and 38b. Expectedly, this observation was also consistent when comparing the diethylamino-containing analogue 48d (IC_50,KAN = 0.15 ± 0.02 μM) with the piperidinyl-containing inhibitor 50d (IC_50,KAN = 0.08 ± 0.02 μM). Size expansion of the pyrrolidinyl ring in 38b (IC_50,KAN = 0.07 ± 0.01 μM) to the piperidinyl ring in 39b (IC_50,KAN = 0.030 ± 0.005 μM) additionally suggested that the larger six-membered heterocycle occupied the chemical space more effciently, which resulted in an improved inhibitor (2-fold improvement). Comparison of compounds 43b (IC_50,KAN = 0.10 ± 0.02 μM) and 44b (IC_50,KAN = 3.3 ± 0.7 μM) showed that the chemical space around the tertiary amino group has a limit, and the addition of the phenyl group was not well-tolerated. The rigidity of the tertiary amine was explored by assessing compounds 34b (IC_50,KAN = 0.056 ± 0.006 μM), 43b (IC_50,KAN = 0.10 ± 0.02 μM), and 45b (IC_50,KAN = 0.17 ± 0.03 μM). We observed that additional rigidity in the side chain as in 34b led to improved Eis inhibitory activity compared to 43b and 45b. Further substitution of the piperidinyl group of 45b (IC_50,KAN = 0.17 ± 0.03 μM), as in 46b (IC_50,KAN = 0.14 ± 0.04 μM), was well tolerated. Overall, our study of the R₂ side chains demonstrated that side chains containing tertiary amino groups were favorable for Eis inhibition. We suspected that this phenomenon was the result of these tertiary amino groups possessing positive charges, which mimicked the positively charged nature of the AG substrates. Additionally, we found that tertiary amines with higher degree of bulkiness and rigidity were typically more potent Eis inhibitors.

Finally, we explored the importance of the thioether at C3 (series II) by replacing it with an amine (series IV). When comparing series IV compounds 52 (IC_50,KAN = 0.23 ± 0.03 μM) and 53 (IC_50,KAN = 0.71 ± 0.08 μM) to their counterparts from series II 37b (IC_50,KAN = 0.17 ± 0.04 μM) and 43b (IC_50,KAN = 0.10 ± 0.02 μM), respectively, we found that the thioether afforded compounds with better (1.4−7.1-fold) Eis inhibitory activity.

Inhibition Kinetics.

Once we completed the analysis of the IC₅₀ measurements, we aimed to understand the mechanism by which these Eis inhibitors function. For this analysis, we selected seven inhibitors, all with an IC₅₀ < 1 μM when using NEO as the substrate. Data analysis of kinetics as a function of the concentrations of the inhibitors and the substrate indicated that the inhibitors are competitive with NEO and yielded the inhibition constants (K_i, Table 2) for compounds 36b, 37b, 37d, 37g, 39b, 39c, 39i, and 50d. Representative plots for 39i and 50d as insets in the IC₅₀ curves are presented in Figure S81. All compounds, with the exception of 39i, showed a K_i < 1 μM. From these data, the best inhibitor is 37b, followed by 39b, 39c, 37g, 36b, 37d = 50d, and 39i. Interestingly, on the basis of the IC₅₀ values, 39i is the best followed by 50d, 39b, 37b = 36b, 39c, 37g, and 37d. These observations indicate that the most effective Eis inhibitors (lowest IC₅₀ values) may not bind the tightest to the enzyme (lowest K_i values).

Table 2.

K_i Values Determined Using Regression Analysis for Eis Inhibitors Displaying IC₅₀ Value <1 μM Using NEO as Substrate

compound	Ki (μM)
36b	0.6 ± 0.2
37b	0.08 ± 0.01
37d	0.82 ± 0.48
37g	0.45 ± 0.05
39b	0.16 ± 0.03
39c	0.32 ± 0.13
39i	1.4 ± 0.3
50d	0.89 ± 0.23

Selectivity of Inhibitors toward Eis over Other AACs.

Mtb contains another AAC chromosomally encoded in its genome, AAC(2′)-Ic.²¹ To determine the selectivity of the 1,2,4-triazino[5,6b]indole-3-thioether-based Eis inhibitors to-ward Eis, we also tested the potential ability of these compounds to inhibit other AACs. Previously, only one Eis inhibitor, from an unrelated scaffold, displayed inhibitory activity against AAC(2′)-Ic.¹⁶ Compounds 36b and 37b were tested for inhibitory activity with AAC(2′)-Ic, AAC(3)-IV from E. coli,^22,23 and AAC(6′)-Ie from S. aureus.²⁴ No inhibition of these three AACs was observed with these two compounds at concentrations up to 200 μM, indicating that the compounds presented here are highly selective to Eis.

Effect of Inhibitors on KAN MIC for M. tuberculosis.

To determine if the identified Eis inhibitors could restore the activity of KAN in KAN-resistant Mtb, the MIC of KAN (MIC_KAN) was determined in KAN-sensitive Mtb H37Rv and KAN-resistant Mtb strain K204 for all compounds displaying Eis inhibition (<28 μM). Mtb K204 is genetically identical to H37Rv with the exception of one clinically important eis promoter mutation upregulating Eis expression resulting in resistance to KAN;⁸ therefore, Mtb K204 is an excellent tool for validation of Eis inhibition as a mode of action of these compounds. Inhibitors were tested at concentrations of 100-fold of their respective IC_50,KAN or at 100 μM when 100× IC₅₀ was unknown or unachievably high. As expected, the Eis inhibitors did not affect the MIC_KAN in the KAN-sensitive Mtb H37Rv strain. Four compounds (36b, 37b, 39b, and 43d) lowered the MIC_KAN value for K204 from >10 μg/mL in the absence of inhibitors to 2.5−5 μg/mL. Interestingly, compound 43d, while not the most potent Eis inhibitor when tested against purified Eis, reduced the MIC_KAN for Mtb K204 to 2.5− 5 μg/mL. Conversely, two of the most potent inhibitors, 38b and 39i, did not effciently restore the activity of KAN in Mtb K204. Thus, while many Eis compounds aid in the restoration of KAN activity, there is not always a direct correlation between Eis inhibition and MIC_KAN. There may be off-target proteins that bind the compounds and possibly a portion gets caught up in the complex cell wall of Mtb, accounting for the large excess required to restore MIC_KAN.

Several factors validate Eis inhibition as the likely mechanism of action of these molecules. Compounds increased the sensitivity of Mtb K204, but not H37Rv to KAN while not affecting Mtb viability in general. Furthermore, we performed MIC_KAN dose-dependence experiments with compounds 36b and 39b using a double-dilution protocol and a concentration range of Eis inhibitors of 0−32 μM (Table 3). We observed an inversely proportional relationship between the MIC_KAN for Mtb K204 and the concentration of compound implying increasing Eis inhibition as the concentration of compound increased.

Table 3.

MIC_KAN Values for Mtb H37Rv and K204 with Varying Concentrations of Compounds 36b and 39b

	36b		39b
concentration tested (μM)	H37Rv MICKAN (μg/mL)	K204 MICKAN (μg/mL)	H37Rv MICKAN (μg/mL)	K204 MICKAN (μg/mL)
0	1.25, 2.5	20, 20	1.25, 2.5	20, 20
0.5	1.25, 1.25	10, 20	1.25, 1.25	10, 20
1	1.25, 1.25	10, 20	1.25, 1.25	10, 20
2	1.25, 1.25	10, 20	0.625, 1.25	10, 10
4	0.625, 1.25	10, 10	0.625, 1.25	5, 10
8	0.625, 1.25	5, 5	0.625, 1.25	5, 5
16	0.625, 1.25	5, 5	0.625, 1.25	5, 5
32	0.625, 1.25	5, 5	0.625, 1.25	2.5, 2.5

Crystal Structure of Eis-CoA-Inhibitor 39b Complex Confirms Mechanism of Inhibition.

To establish the structural basis for Eis inhibition by compounds with a 1,2,4-triazino[5,6b]indole-3-thioether core, we determined the crystal structure of the enzyme in complex with CoA and one of the potent inhibitors, 39b (Figure 3). Compound 39b (Figure 3A,B) binds at a site that partially overlaps the site occupied by the AG tobramycin (TOB) as well as other Eis inhibitors in the literature (Figure 3C).^12,17,18,20 This binding site supports the mode of action of inhibitor 39b as an AG competitive inhibitor. The crystal structure shows that the aromatic 1,2,4-triazino-[5,6b]indole core of compound 39b is stabilized by interactions with two neighboring aromatic residues, Trp36 and Phe84 (highlighted in orange) via sandwich π−π stacking interactions (Figure 3B). We have previously observed similar π−π interactions with Trp36 and Phe84 during studies with Eis inhibitors of other structural classes (inhibitors a, b, and c; Figure 3D–F). Additionally, the fluorine atom at the C8-position of the 1,2,4-triazino[5,6b]indole-3-thioether core projects toward a hydrophobic pocket and interacts with the Cγ of Arg37 (~3.3 Å away from the fluorine atom) and the Cδ atoms of Leu63 (~3.6 Å away from the fluorine atom) by hydrophobic interactions (Figure 3B). Hydrophobic interactions with Arg37 and Leu63 were also seen with other Eis inhibitors (Figure 3B,D–F).

Figure 3.

(A) Crystal structure of one of the six monomers of the EisC204A-CoA-inhibitor 39b (shown as yellow sticks in a box) complex. Note: For the sake of simplicity, the CoA is omitted in this figure, but is shown in panel C. (B) Zoom-in view of the binding pocket of compound 39b. The strong omit F₀-F_c electron density map contoured at 3σ generated without compound 39b is shown by a magenta mesh. The amino acid residues interacting with compound 39b are depicted in dark turquoise. The conserved residues that interact with 39b and previously published Eis inhibitors are depicted as orange sticks. The Asp26-Ser32 loop is shown as dark turquoise sticks. The C-terminus is labeled as C-ter. A water molecule is shown as a green sphere. The distances between atoms of the inhibitor 39b and those of amino acid residues of the Eis protein are shown by a dark blue dashed line and are in Å. For compound 39b, the carbon, oxygen, nitrogen, fluorine, and sulfur atoms are colored pale green, red, blue, dark green, and orange, respectively. (C) Compound 39b is overlapped with tobramycin (TOB) and previously published Eis inhibitors. The previously published structure of bound TOB (PDB ID 4JD6¹²) is in blue. Bound CoA is depicted as red sticks. (D) Inhibitor “a”, a pyrrolo[1,5-a]pyrazine-based Eis inhibitor (labeled 2k* in ref 20, PDB ID 5TVJ), is depicted as green sticks. (E) Inhibitor “b”, a sulfonamide-based inhibitor (labeled 39 in ref 18 PDB ID 5IV0), is depicted as gray sticks. (F) Inhibitor “c”, an isothiazole S,S-dioxide heterocyclic core (labeled 11c in ref 17 PDB ID 5EBV) is depicted as purple sticks. Note: In panels D−F, amino acid residues that interact specifically with the inhibitors presented are depicted in lilac. (G) Chemical structures of TOB as well as inhibitors 39b and “a−c”.

While sharing some interactions with other Eis inhibitor classes, this structural scaffold enabled us to explore a novel inhibitor-Eis interface. The unique tricyclic core of 39b allows the flexible C3-linker to extend toward a large opening in the AG-binding pocket such that the piperidinyl group on the end of the linker comes into a hydrophobic contact with the aliphatic stem of Glu401. In this conformation, the linker and the piperidinyl group at the end are juxtaposed against the Asp26-Ser32 loop (Figure 3B,C). Interactions with Asp26-Ser32 loop appear to be essential for Eis binding of this inhibitor class because analogues 1b, 32b, and 33b, which all have shorter C3 substituents, do not inhibit Eis. A water molecule located 2.8 Å away from the N2-nitrogen of the tricyclic core could form a hydrogen bond with either the hydroxyl of Ser32 or the backbone carbonyl oxygen of Gly29. Furthermore, the sulfur atom of 39b is in a hydrophobic contact with the Cε of Ile28, and the linker makes extensive van der Waals contacts with the backbone of residues 26−28. Two carbons away from the sulfur, the nitrogen atom of the piperidinyl moiety bearing a positive charge at the physiological pH forms a salt bridge with the negatively charged carboxylate of Asp26 (located 3.4 Å away). Even though none of the previous Eis inhibitor classes exploits interactions with the Asp26-Ser32 as extensively as this class, among the previously discovered inhibitor classes, inhibitor b interacts with this region by forming hydrogen bonds with the carboxyl group of Asp26 and the backbone N of Ile28 (Figure 3E).

This crystal structure and the SAR will guide future rational fine-tuning of the inhibitors with this new scaffold. First, other C3 substitutions bearing a positive charge could favorably interact with another negatively charged residue (e.g., Glu401 or the terminal carboxyl group). These electrostatic interactions support our previous SAR observations that positively charged tertiary amine functionalities such as those in compounds 37b or 36b are more tolerated compared to the neutral or partially negatively charged C3 substituents. In the R₁ substitution of the tricyclic core, the carbon of the N5-methyl is in close proximity to the terminal carboxyl group of Eis (located 3.4 Å away) and the hydroxyl group of Ser83 (located 3.1 Å away), which explains why N5 alkylation with bulkier groups is not well tolerated. For instance, the IC50,NEO values for compounds 11b, 11c, and 11d increase almost two-fold for every carbon atom added to the N5-alkyl chain. Additionally, there are a variety of ways to install different hydrogen bond donors at the N5 to take advantage of the polar protein C-terminus and side chain of Ser83. In summary, the crystal structure of the Eis-CoA-inhibitor 39b complex allowed us to compare and contrast the current class of inhibitors with previous inhibitors in the literature, explain some of the observed SAR trends, and gain insight into further optimization of this class of inhibitors. Even though these 1,2,4-triazino[5,6b]indole-3-thioether-based inhibitors are competitive KAN inhibitors as are the other previously published Eis inhibitors, the new inhibitors interact extensively with an underexplored surface of the substrate binding cavity of Eis. This property may help combat resistance to other Eis inhibitor classes that could arise due to point mutations in other inhibitor interacting surfaces.

Mammalian Cytotoxicity of Eis Inhibitors.

The ability of some molecules bearing the 1,2,4-triazino[5,6b]indole-3-thioether scaffold to restore the activity of KAN in Mtb K204 is highly encouraging. However, to understand the utility of these compounds as potential therapeutic agents, the mammalian cell cytotoxicity needed to be examined. Four of the best compounds (34b, 37b, 39b, and 43d) were tested against three mammalian cell lines (A549 (lung), HEK-293 (kidney), and J774A.1 (macrophage)) alone and in the presence of KAN (50 μg/mL = 86 μM). Compounds were evaluated for mammalian cytotoxicity at concentrations ranging from 0 to 100 μM. Cells displayed 100% survival rate from 0 to 12.5 μM; therefore, we showed the cytotoxicity data only in the range from 12.5 to 100 μM (Figure 4). The negative and positive controls used for these experiments were 0.5% DMSO (100% cell survival) and 1% v/v Triton-X 100 in 0.5% DMSO (<5% cell survival), respectively. Overall, we found that the compounds did not display significant toxicity until concentrations of 50 μM with and without KAN (Figure 4). Because compounds 34b, 37b, and 39b sensitize KAN-resistant Mtb cells to KAN when used at concentrations between 5.6 to 17 μM, where they are not cytotoxic, these molecules are good candidates for further development into therapeutically useful KAN adjuvants for treatment of KAN-resistant Mtb.

Figure 4.

Mammalian cytotoxicity of selected compounds (34b, 37b, 39b, and 43d) alone (depicted as dark color columns) or in the presence of 50 μg/mL (86 μM) KAN (depicted as light color column directly to the right of the dark color column for the compound alone) against (A) A549, (B) HEK-293, and (C) J774A.1 cells. Note: No cytotoxicity was observed from 0 to 12.5 μM.

CONCLUSION

In summary, by chemical, biochemical, biological, and structural studies, we discovered, optimized, and extensively characterized Eis inhibitors with the 1,2,4-triazino[5,6b]indole-3-thioether scaffold. We discovered analogues with potent IC₅₀ values and verified that these inhibitors were able to restore the MIC_KAN in cellular assays. We also determined that these analogues inhibit Eis by competing with KAN via Lineweaver−Burk plots and determining the crystal structure of Eis in complex with inhibitor 39b. The crystal structure further revealed that the scaffold enabled inhibitor interactions with a previously underexploited region of the Eis active site. The inhibitors were shown to not be cytotoxic at concentrations needed to restore MIC_KAN. These compounds act as KAN adjuvants against KAN-resistant Mtb. This study lays the groundwork for the further optimization of these compounds for clinical applications.

ABBREVIATIONS

AAC

aminoglycoside acetyltransferase

AcCoA

acetyl coenzyme A

AMK

amikacin

Eis

enhanced intracellular survival

HTS

high-throughput screening

KAN

kanamycin

MDR

multidrug-resistant

MIC

minimum inhibitory concentration

Mtb

Mycobacterium tuberculosis

NEO

neomycin B

SAR

structure–activity relationship

TOB

tobramycin

TB

tuberculosis

XDR

extensively drug-resistant