Development of a Novel Lead that Targets M. tuberculosis Polyketide Synthase 13

Abstract

Widespread resistance to first-line TB drugs is a major problem that will likely only be resolved through the development of new drugs with novel mechanisms of action. We have used structure-guided methods to develop a lead molecule that targets the thioesterase activity of polyketide synthase Pks13, an essential enzyme that forms mycolic acids, required for the cell wall of Mycobacterium tuberculosis. Our lead, TAM16, is a benzofuran class inhibitor of Pks13 with highly potent in vitro bactericidal activity against drug-susceptible and drug-resistant clinical isolates of M. tuberculosis. In multiple mouse models of TB infection, TAM16 showed in vivo efficacy equal to the first-line TB drug isoniazid, both as a monotherapy and in combination therapy with rifampicin. TAM16 has excellent pharmacological and safety profiles, and the frequency of resistance for TAM16 is ∼100-fold lower than INH, suggesting that it can be developed as a new antitubercular aimed at the acute infection. PAPERCLIP.

Keywords: Mycobacterium tuberculosis; Pks13 thioesterase domain; benzofuran inhibitors; crystal structure; polyketide synthase; structure-based drug discovery.

Graphical abstract

Figure 1

Novel Benzofurans Inhibit Pks13 Thioesetrase Domain (A) Chemical structure of TAM1 highlighting the convention used for naming the substituent groups (P₁, P₂, P₃, and P₄) and numbering of the benzofuran ring. TAM1 inhibits the esterase activity of Pks13-TE with an IC₅₀ = 0.26 ± 0.03 μM. The graph depicts percent activity relative to DMSO only control (mean ± SD). (B) Overall view of the structure of the Pks13-TE-TAM1 complex showing structural features of the Pks13-TE domain. Catalytic residues His1699 and Ser1533 at the interface of the lid and core domains are shown as ball and sticks. TAM1 is shown as yellow sticks. (C and D) Close-up views of inhibitor interactions show that benzofuran core of TAM1 (yellow sticks) wedges between Phe1670 and Asn1640 with its P₃ group oriented toward the catalytic site. Hydrogen bonds are represented by dashed lines. Surface representation in (C) is colored by electrostatic potential (contoured at ± 5 kT/e, red for negative and blue for positive). See also Figure S1 and Tables S1, S2, and S3.

Figure S1

Structural Features of the Pks13-TE Crystal Structure, Related to Figure 1 (A) Surface representation of the Pks13-TE structure colored by electrostatic potential (contoured at ± 5 kT/e; red for negative and blue for positive) to illustrate the substrate binding groove (∼30 Å, double-headed yellow arrow) on the lid domain. The zoomed view shows the bound PPG fragment (cyan) in the catalytic pocked formed by residues Ser1533, His1699 and Asp1560 along with the residues of the oxyanion-hole (Leu1534 and Ala1477) rendered as sticks; catalytic water shown as red sphere. The 2Fo-Fc electron density map contoured at 1.2σ is shown for the bound PPG fragment. Hydrogen bond interactions are shown as black dashed lines. (B) Predicted tunnels in Pks13-TE structure by CAVER analysis (Chovancova et al., 2012). The three potential tunnels are shown in pink, blue and green surface rendering. The largest of the tunnels (pink) opens onto the substrate binding surface groove and contained the bound PPG fragment. (C) Docking of mycolic acid on Pks13-TE lid domain. A molecule of mycolic acid (shown as yellow sticks) was docked using Molsoft ICM-Pro software to determine a possible binding mode in the substrate binding groove. The zoomed view of the docking indicates that the surface groove can accommodate acyl chains of the mycolic acid precursor attached to the C-terminal ACP domain and position the thioester for cleavage near the catalytic Ser1533 residue.

Figure S2

Structural Changes in TAM1-Bound and D1607N-Mutant Pks13-TE Crystal Structures, Related to Figure 1 (A) Superimposition of Pks13-TE-TAM1 complex structure (purple) with Apo-Pks13-TE structure (yellow) shows that Phe1670 side chain (shown as stick) flips by ∼80° upon TAM1 binding. TAM1 interacting residues are shown in line representation in purple color, and the corresponding residues from Apo structure are shown as lines in yellow color. Catalytic residues (Ser1533 and His1699) are shown as ball and sticks. TAM1 is omitted for clarity of presentation. (B) Superimposition of Pks13-TE-TAM1 complex structure (purple, TAM1 as yellow sticks) with the structure of D1607N mutant (pink) shows the conformational change in Arg1641 of the mutant due to disruption of ion pair interaction with Asp1607. In the wt Pks13-TE structure, Asp1607 carboxylate forms an ion pair interaction with the guanidinium of Arg1641 which helps stabilize the C-terminal end of helix α7. This allows Asp1644 to form hydrogen bond interaction (shown as dashed black line) with TAM1. The mutation D1607N breaks the ion pair interaction mediated anchoring of helix α7 that causes Asp1644 to move away by ∼3 Å (double-headed black arrow), consequently disrupting its interaction with TAM1.

Figure 2

Structure-Guided Development of TAM16 Substitution of P₁ and P₂ groups in TAM16 with phenol and methyl amide, respectively, increased potency and metabolic stability. Calculated log(P), calculated log(partition coefficient); Mouse Cli., intrinsic clearance in mouse liver microsomes.

Figure S3

Metabolic Stability of TAM1 Analogs, Related to Table 3 (A) Metabolic stability studies of TAM12 in mouse liver microsomes showed that TAM12 was hydroxylated at P1 phenyl. Graph represents mean values ± SD of two independent assays. (B) Cartoon of the Pks13-TE-TAM16 complex structure showing hydrogen bond interaction between P₁ 4-OH of TAM16 (cyan) and the side chain carbonyl oxygen atom of Gln1633 (green). The gray mesh represents the 2mF_o - DF_c maximum-likelihood omit map, contoured at 1.2σ. Hydrogen bonds are shown as black dashed lines. (C) Glucuronidation of TAM16 was assessed in mouse liver microsomes with UDPGA (5 μM for 60 min). At 0, 30 and 60 min, 100 μL aliquots of the reaction mixture were removed and placed in 100 μL of acetonitrile to terminate the reaction. Analysis by mass spectrometry for metabolite identification showed little metabolism of the parent compound (TAM16), the conjugate was barely detectable after 60 min incubation (peak A). (D and E) Metabolic stability analysis of TAM16 incubated with glutathione and methoxylamine (50 μM for 180 min) in the presence of mouse plasma and HLMs for possible retro-Mannich metabolites or adducts. The parent compound was seen to decrease slightly over time and two metabolites were detected. For metabolite B the main ion see was at m/z 379 but a barely detectable ion was seen at m/z 397 indicating the metabolite may be due to oxidation which dehydrates readily in the MS. No trace of the quinone-methide or potential GSH adducts was seen in any of the samples. ^a% of total peak area for parent and metabolite peaks of the proposed [M+H]⁺ ions at 60 min time point. ^b% of total peak area for parent and metabolite peaks of the proposed [M+H]⁺ ions at 180 min time point.

Figure S4

In Vivo Pharmacokinetics of TAM16, Related to Table 3 Mean blood concentration profiles of TAM16 following oral (p.o) and iv dose of 10 mg/kg and 3 mg/kg, respectively, in female BALB/c mouse. PK parameters (inset) were determined after the administration of a single dose (both p.o and iv) to mice. C_max, maximum concentration; T_max, time to reach C_max, ${\text{t}}_{\left(1/2\right)}$, half-life; AUC, area under the concentration curve; V_ss, volume of distribution at steady state.

Figure 3

Efficacy of TAM16 in Mouse Models of TB (A) In vivo activity of TAM16 against acute TB infection in immunocompetent BALB/c mice. Data represent mean M. tuberculosis burden (log₁₀ CFU) in the lungs of mice (n = 5 per time point) expressed as mean ± SD. Week 0 indicates CFU counts in the lungs at treatment initiation (2 weeks after infection). Drugs were administered via oral gavage 5 days/week for 2 weeks. The mice in the untreated group were moribund after 3 weeks of infection and were euthanized in accordance with institutional animal care regulations. ^∗p < 0.05 by Dunnett’s multiple comparison tests, as compared to the untreated (vehicle-only) control group; ns, no statistical significance. (B) Efficacy of TAM16 in reducing M. tuberculosis burden in chronically infected immunocompetent mice after 4 weeks of treatment. Treatment was initiated 27 days after infection, and drugs were administered once daily via oral gavage for 5 days/week for 4 weeks. Data show the bacterial loads (mean log₁₀ CFU ± SD) in the lungs and spleen of the infected mice (n = 5 per group). ^∗p < 0.05 and ^∗∗∗p < 0.001 by pairwise multiple comparison procedures (Tukey test); ns, no statistical significance. (C) Efficacy of TAM16 administered in combination with anti-TB drug rifampicin (RIF) in chronically infected BALB/c mice. Treatment was initiated 28 days after infection, and drugs were administered once daily via oral gavage (5 d/wk) for 4 and 8 weeks. Data represent mean M. tuberculosis burden (log₁₀ CFU) in the lungs of the infected mice (n = 6 for vehicle-only control group and n = 7 for each treatment group) after 4 and 8 weeks of therapy (mean ± SEM). In combination studies, TAM16 and isoniazid (INH) were administered 1 hr following prior administration of RIF. Dotted horizontal line indicates the limit of detection. See also Figure S5 and Table S7.

Figure S5

Effect of TAM16 Treatment on Mice in Acute BALB/c Model, Related to Figure 3A (A) Lung gross pathology images from untreated control (vehicle only) and treated mice. Untreated mice were moribund 1 week after treatment initiation (3 weeks post-infection) and were euthanized in accordance with institutional animal care regulations. INH, isoniazid. (B and C) Effect of drug treatment on mean lung weights and, (C) mean body weights in M. tuberculosis infected mice (n = 5 per time point per group) after 2 weeks of treatment. Mice were infected on Day −13 and treatment was initiated on D0 (2 weeks after infection). Graphs represent mean values ± SD.