Covalent Irreversible Inhibitors of Tetracycline Destructases

Abstract

Next-generation tetracycline antibiotics are threatened by an emerging resistance mechanism — enzymatic inactivation. The relevant enzymes — tetracycline destructases (TDases) — are structural homologs of class A flavin monooxygenase (FMO) that oxidize tetracycline antibiotics leading to various inactive degradation products. Small molecule inhibitors of antibiotic inactivating enzymes are critical clinical therapeutics used to manage bacterial resistance with combination therapy. While reversible TDase inhibitors have been reported, we sought to develop covalent inhibitors that are better aligned with clinically effective covalent β-lactamase inhibitors. Here, we report the design, chemical synthesis, and biochemical characterization of the first covalent irreversible inhibitors of TDases based on C9-derivatives of anhydrotetracycline (aTC). The reactive warheads were installed via a one-step Mannich reaction linking either an N-(1-methyl)cyclopropylamine or N-propargylamine group to the C9-position of the aTC D-ring via an amino methylene linkage. We also synthesized two non-specific FMO inhibitors, N-(1-methyl)cyclopropylbenzylamine (1) and N-methyl-N-benzyl-propargylamine (2) as mechanistic probes to distinguish reactivity with the essential FAD cofactor in TDases via one- or two-electron transfer pathways, respectively. We evaluated the compounds as potential inhibitors of representative TDases from the two major classes – Type 1 (TetX6 and TetX7) and Type 2 (Tet50). The aTC-based compounds 3-5 inhibited both Type 1 and Type 2 TDases with notable differences in potency and inhibition mechanism. The inhibition of Type 1 TDases was more potent but reversible with no time dependence. The inhibition of Type 2 TDases was time-dependent and irreversible even after exhaustive dialysis, consistent with a covalent mechanism of inhibition. Molecular modeling of the inhibitors supports unique inhibitor binding modes for Type 1 and Type 2 TDases that are consistent with the observed differences in the inhibition modes. Blue light irradiation of the Type 2 TDase enhanced this reactivity. Treatment of Tet50 with probe molecules 1 and 2 under blue light exposure enabled the identification of covalent FAD adducts via mass spectrometry that are consistent with the expected one- and two-electron transfer reaction modes of the cyclopropylamine and propargylamine warheads with the FAD cofactor. At concentrations as low as 2 μg/mL, the aTC-based covalent inhibitors 3-5 recovered tetracycline activity against E. coli overexpressing TDases. Our findings suggest that the inhibition of TDases through covalent trapping of the FAD cofactor is a viable strategy for overcoming TDase-mediated antibiotic resistance.

Graphical Abstract

Tetracycline (TC) antibiotics have been an important class of antibacterial agents since the first discovery of aureomycin (chlortetracycline) in the late 1940s.¹ As a family of type-II polyketides, TC antibiotics are constructed with a fused system of four linear rings. TC antibiotics demonstrate impressive antibacterial properties against a wide range of pathogens, including both Gram-positive and Gram-negative bacteria.^2-4 The antibiotic activity of TCs is achieved by interfering with bacterial protein biosynthesis through binding to the 16S rRNA of the 30S bacterial ribosomal subunit and preventing the binding and accommodation of amino-acyl tRNAs.⁵ Extensive research and development efforts have been devoted to optimizing TC antibiotics through semi synthesis and total synthesis to keep pace with resistance caused by efflux and ribosome protection protein mechanisms.^6-9 The development of second-generation TCs, such as doxycycline and minocycline, improved the pharmacokinetic properties and increased the efficacy against resistant bacterial species expressing TC efflux pumps.⁶ Third- and fourth-generations of TC antibiotics, such as tigecycline,^10,11 omadacycline,¹² and eravacycline^13,14 have been developed to overcome resistance via efflux pumps and ribosomal protection proteins. However, all clinically useful TC antibiotics including fourth-generation drugs fail to address enzymatic inactivation as a new clinical resistance mechanism.^15,16

The relevant enzymes, tetracycline destructases (TDases), are members of the class A flavin monooxygenase (FMO) enzyme family.¹⁷ FMOs are single component flavoprotein oxidases that utilize a dynamic non-covalent flavin adenine dinucleotide (FAD) as a redox cofactor. The FAD, known in the family of enzymes as the ‘wavin flavin’, can occupy two distinct conformations referred to as ‘FAD-IN’ and ‘FAD-OUT’. The FAD cofactor rests in the TDase active site in its oxidized form. Substrate binding triggers the two-electron reduction of FAD by NAD(P)H, presumably in the ‘FAD-OUT’ conformation. The resulting reduced form FADH₂ presumably moves to the ‘FAD-IN’ conformation where it is subsequently reoxidized by O₂ to form a reactive C4a-peroxyflavin species.^18-20 Substrate oxidation commonly takes place via oxygen transfer (hydroxylation) or oxygen insertion (Baeyer-Villiger type) resulting in a C4a-hydroxyflavin that dehydrates to the original resting FAD oxidized state (Figure 1).

Figure 1:

Mechanistic basis of MAO/TDase function. Oxidized FAD is the shared resting form of both enzymes. MAO covalent inhibition machinery can be leveraged to develop TDase covalent inhibitor. Cyclopropylamine (1) and propargylamine (2) are proven effective reactive warheads for MAO covalent inhibition. These warheads were incorporated into aTC scaffold for the development of TDase covalent inhibitors (3, 4, and 5).

Two major families of TDases, Type 1 and Type 2, have been reported.¹⁷ Type 1 TDases are more frequently encountered in clinically isolated strains, whereas the Type 2 enzymes are more common in environmental microbes.¹⁶ Type 2 TDases (represented in this study by Tet50) differ from Type 1 TDases (represented in this study by TetX7 and TetX6) in the presence of an extra C-terminal ‘gatekeeper’ helix that allows for closing of the active site during catalysis (Figure 2a, b).²¹ This structural variation between subtypes leads to other distinguishing properties including substrate scope, substrate binding mode, product outcome, susceptibility to inhibitors, and photostability.^22,23 The increased occurrence of TDases in clinical pathogens and ability to inactivate all generations of TC drugs creates an urgent need for effective inhibitors.

Figure 2:

Structural basis of TDase inhibitor binding and ground-state mechanistic basis for the covalent modification of FAD by propargyl and cyclopropyl amines. X-ray crystal structures of aTC bound to Tet50 (a) and TetX6 (b) highlight the different inhibitor binding modes relative to FAD (PDBs: 5TUF and 8ER0, respectively) where the D-ring orients C9 in different directions. The “gatekeeper helix” of Tet50 is highlighted in orange. (c) Proposed mechanism for covalent adduct formation between FAD and a propargylamine via a two-electron pathway starting with a hydride transfer from the α-C–H. (d) Proposed mechanism for the formation of covalent adduct between FAD and a cyclopropylamine via a one-electron pathway starting with a single electron transfer from the lone pair of electrons on nitrogen. R and R' are defined separately for the chemical structures shown throughout the manuscript.

The clinical gold standard of care for treating infections caused by bacterial pathogens expressing antibiotic inactivating enzymes is the use of combination drug therapy.²⁴ β-Lactam antibiotics such as amoxicillin are commonly co-administered with a β-lactamase inhibitor such as clavulanic acid.²⁵ We anticipate that this type of combination therapy can be applied to TC antibiotics against pathogens expressing TDases as a primary resistance mechanism. Our group previously reported that anhydrotetracycline (aTC), a TC biosynthetic precursor and autoinducer,²⁶ is a competitive inhibitor of Type 1 and Type 2 TDases.²⁷ We determined the structural basis for the inhibition of Type 1 and Type 2 TDases by aTC which revealed two distinct inhibitor binding modes for each type (Figure 2a, b).^28,29 The inhibitor binding mode directs the design of various aTC-based inhibitors for different types of TDases. The Type 2 TDase Tet50 (PDB:5TUF) binds aTC with the D-ring pointing toward the active site cavity in the ‘FAD-OUT’ conformation (Figure 2a). The Type 1 TDase TetX6 (PDB: 8ER0) flips the aTC 180° along the relative y-plane to position the A-ring toward the inside of the active site cavity with D-ring exposed to solvent at the entrance of the active site in the ‘FAD-IN’ conformation (Figure 2b). Using structure-based drug design, we developed a series of bivalent TDase inhibitors based on C9-benzamide³⁰ or C10-benzoate³¹ derivatives of aTC that bind competitively and reversibly to the TC and NADPH sites. While these bivalent inhibitors showed promising synergy with TC antibiotics, we hypothesized that the potency and efficacy of the inhibitors might be improved via a covalent inhibition mechanism.

Covalent irreversible enzyme inhibitors offer several potential therapeutic advantages over reversible inhibitors such as enhanced potency, longer target dwell time, and reduced dosage frequency.³² As a result, the development and application of covalent inhibitors has gained widespread popularity in clinical applications including the inhibition of antibiotic inactivating enzymes. Clavulanic acid,³³ sulbactam,³⁴ avibactam,³⁵ and vaborbactam³⁶ are all FDA-approved covalent inhibitors of β-lactamases that are used in combination with β-lactam antibiotics. β-Lactamases contribute to antibiotic resistance by hydrolyzing the thermodynamically labile β-lactam ring that is conserved among this antibiotic class.³⁷ All clinically useful β-lactamase inhibitors are covalent inhibitors that react with the conserved active serine residue present in all Class A, C, and D β-lactamases. The clinical success of β-lactamase inhibitors suggests that covalent inhibitors could play an essential role in combating antibiotic resistance for a wide range of antibiotic inactivating enzymes including TDases.

As class A FMOs, TDases present two general strategies for covalent inhibition: 1) covalent tagging of amino acid side chains, or 2) covalent reactions with the FAD cofactor. We chose to focus on the second strategy hypothesizing that covalent adduct formation between the inhibitor and the FAD cofactor would result in a tightly bound inhibition complex.³⁸ We tested our hypothesis by functionalizing the known TDase inhibitor aTC with two types of reactive warheads, a propargylamine or cyclopropylamine group, which are known to undergo redox-triggered covalent tagging of the FAD cofactor in monoamine oxidases (MAOs).³⁹ MAOs oxidize amines to imines through direct oxidation by FAD, which is capable of mediating redox reactions via two-electron and/or one-electron transfer pathways (Figure 1).⁴⁰ Propargylamines exploit the two-electron pathway, whereas cyclopropylamines exploit the one-electron pathway. Both pathways lead to the formation of electrophilic intermediates that can react with the nucleophilic N5 of the reduced FAD isoalloxazine ring leading to covalent adduct formation via N5-alkylation.^41,42

During a two-electron process, a reactive FAD anion intermediate, [FADH]⁻, is formed.⁴⁰ Propargylamine inhibitors of MAOs were found to form covalent adducts with the FAD cofactor presumably resulting from a two-electron transfer mechanism involving the [FADH]⁻ anion.⁴² Propargylamine derivatives have been utilized since the 1950s for the treatment of neuropsychiatric and age-related neurodegenerative disorders via the covalent inhibition of MAOs.⁴³ Extensive studies have elucidated the mechanism of covalent adduct formation between propargylamine derivatives and MAOs.^42,44-46 The first step involves a hydride transfer from the propargylamine to N5 of the oxidized FAD. Subsequently, the resulting iminium cation is trapped by the nucleophilic N5 of the reduced [FADH]⁻, forming a substituted 1,3-diaminoallene through a conjugate addition reaction. The resulting intermediate then undergoes a water-mediated 1,3-prototropic rearrangement leading to a highly conjugated covalent adduct (Figure 2c).

During a one-electron process, a reactive FAD semiquinone intermediate, [FADH]^•, is formed.^47,48 Cyclopropylamine inhibitors of pig liver mitochondrial MAOs were found to form covalent adducts with the FAD cofactor presumably resulting from a one-electron transfer mechanism involving the [FADH]^• semiquinone.^41,49 The initial step of this mechanism involves single electron transfer (SET) from the cyclopropylamine nitrogen lone pair to the oxidized FAD, generating the amine radical cation and the [FADH]^• semiquinone after protonation. The amine radical cation activates the adjacent C-C bond of the cyclopropyl ring toward α-cleavage followed by radical recombination with the [FADH]^• semiquinone resulting in FAD N5-alkylation (Figure 2d).

Given that both MAOs and TDases contain an oxidized FAD cofactor in the resting state, we hypothesized that the modes of reactivity observed for MAOs with propargylamines and cyclopropylamines could be extended to TDases to provide covalent inhibitors. We note that both one- and two-electron modes of reactivity can be enhanced via photoactivation of FAD.^50,51 Under visible light irradiation, oxidized FAD can be excited to the singlet state (1Fl*) and subsequently undergo a rapid intersystem crossing (ISC) to the highly reactive triplet state (3Fl*), which is calculated to be a better oxidizing agent than ground-state FAD.⁵² Therefore, we designed a new type of TDase inhibitor merging the scaffolds of the known TDase inhibitor aTC and known MAO inhibitors cyclopropylamine 1 and propargylamine 2 resulting in the synthesis of C9-cyclopropylamine-aTC analog 3 and C9-propargylamine-aTC analogs 4 and 5 (Figure 1). The reactivity of compounds 1-5 towards Type 1 and Type 2 TDases was investigated using time-dependent kinetic assays and exhaustive dialysis under both ‘dark’ and ‘light’ conditions. The ability of compounds 3-5 to synergistically rescue TC activity against E. coli expressing TDases was demonstrated. Findings from this study demonstrate that covalent TDase inhibitors are valuable mechanistic probes for TDase function and have promise as combination therapies with TC antibiotics to overcome TDase-mediated resistance.

Results and Discussion

Cyclopropylamine and propargylamine warheads covalently tag FAD in TDases.

Two mechanistic probe molecules, N-(1-methyl)cyclopropylbenzylamine (1) and N-methyl-N-benzyl-propargylamine (2) (Figure 3), were used to test whether cyclopropylamine and propargylamine warheads are capable of reacting with the FAD cofactors of TDases. Probe 1 was synthesized by S_N2 alkylation of 1-methylcyclopropylamine with BnBr under basic conditions (Figure S1). Probe 2 was synthesized by S_N2 alkylation of N-Me-benzylamine with propargyl chloride under basic conditions. Heterologous expression of TetX6 (Type 1 TDase), TetX7 (Type 1 TDase), Tet50 (Type 2 TDase), and 4-hydroxybenzoate-3-monooxygenase (PHBH; a control class A FMO⁵³) provided the purified enzymes that were used for in vitro enzyme assays (Figure S2). A TC consumption assay was used to measure TDase activity by optical absorbance at 400 nm (λ_max TC). No inhibition of TDase activity was detected at concentrations of the probe molecules 1 and 2 up to 2.5 mM (Table 1 and Figures S3-S5). Blue light exposure was used to enhance the reactivity of the FAD cofactor towards the probe molecules. The Type 1 TDases TetX6 and TetX7 lost ~91% and 82% of their apparent activity, respectively, after 4 minutes of exposure to blue light. However, the Type 2 TDase Tet50 and control FMO PHBH maintained ~76% ± 3% and 88% ± 3% apparent activity, respectively, under the same conditions (Figure S6). Hence, Tet50 and PHBH were chosen for further study with probe molecules 1 and 2 under blue light irradiation.

Figure 3:

Spectroscopic evidence for the photoinduced FAD covalent adducts formation with 1 and 2. (a, d) Structures of two model molecules containing the reactive warheads of interest and the corresponding photoinduced covalent adducts. Optical absorbance spectra measured at 1 h intervals over an 18 h period from Tet50 samples treated with 1 (b) or 2 (e) and protected from external light during the experiment. Optical absorbance spectra measured at 1 min intervals over a 10 min period from Tet50 samples treated with 1 (c) or 2 (f) and exposed to blue light. Arrows show the direction of spectral shifts with increasing time.

Table 1:

Apparent IC₅₀ values of compounds 1–5 and aTC measured through in vitro inhibition of the oxidation of Tet by Tet50, TetX7 and TetX6 under dark conditions.

Inhibitor	Apparent IC50 Values (μM)
	Tet50	TetX7	TetX6
1	none detecteda	none detecteda	none detecteda
2	none detecteda	none detecteda	none detecteda
aTC	410 ± 50	40 ± 6	33 ± 3
3	430 ± 40	11 ± 5	50 ± 4
4	600 ± 40b	74 ± 8	69 ± 7
5	470 ± 60b	70 ± 10	not tested

^aInhibition was not detected for concentrations of compounds 1 and 2 up to 2.5 mM.

^bA poor curve fit was obtained when no pre-incubation was used, so the apparent IC₅₀ value reported here was from a 30 min pre-incubation of inhibitor with enzyme prior to initiating the enzyme assay (see Figure 6 for the time-dependent IC₅₀ plots). All other reported apparent IC₅₀ values are from enzyme assays initiated with no pre-incubation of enzyme and inhibitor (see Figures S3-S5 for these IC₅₀ plots).

Using a higher concentration of Tet50 allowed for the clear visualization of two major absorbance bands centered at ~365 nm and ~450 nm which are characteristic of an oxidized flavoenzyme spectrum (Figure 3).⁵⁴ Under dark incubation using Tet50 in the presence of probes 1 and 2, there was a significant absorbance increase at 365 nm, which also appears in the absorbance spectra of the enzyme alone control (Figures 3b, 3e, and S7). However, treatment of Tet50 with cyclopropylamine probe 1 under blue light irradiation resulted in a continuous decrease in the absorbance of the bands at 450 nm (Figure 3c). Visual inspection of the cuvette revealed near complete quenching of the canonical yellow color of the flavoenzyme solution. These spectral changes are indicative of the reduction of FAD with alkylation at the N5 position of the flavin isoalloxazine ring. The resulting absorbance spectrum was remarkably similar to the reported absorbance spectra of MAOs after treatment with the cyclopropylamine inhibitor tranylcypromine.³⁹ To confirm the identity of the FAD adduct we performed high-resolution mass spectrometry (HRMS) on the treated and control samples with an unbiased comparison of the molecular ion profiles. A denatured sample of Tet50 treated with cyclopropylamine 1 under blue light irradiation produced a unique molecular ion with an observed m/z value of 858.2195 correlating within 2.80 ppm error to the expected m/z value of 858.2219 calculated for the [M+H]⁺ ion of the 4-(flavin-N⁵-yl)-butanone adduct shown in Figure 3a (see also Figure S8a-c).

The incubation of Tet50 with propargylamine probe 2 under blue light irradiation resulted in an increase in absorbance at 320 nm, a decrease in absorbance peak at 365 nm, and a blueshift in the second absorbance peak from ~450 nm to ~414 nm indicating the formation of a new conjugated FAD species (Figure 3f). Visual inspection of the reaction cuvette revealed that the characteristic yellow color of a flavoenzyme was maintained in this sample. These spectral changes are indicative of the reduction of FAD with iminium ion formation at the N5 position of the flavin isoalloxazine ring with strong similarity to the absorbance spectra of MAOs after treatment with the propargylamine inhibitor propargyline.^39,42 A denatured sample of Tet50 treated with propargylamine 2 under blue light irradiation produced a unique molecular ion in the HRMS spectra with an observed m/z value of 945.2667 correlating within 2.64 ppm error to the expected m/z value of 945.2692 calculated for the [M+H]⁺ ion for the (E)-4-(3-(benzyl(methyl)amino)allylidene)-N⁵-flavin-4-ium-4-ide adduct shown in Figure 3d (see also Figure S8d-f). While propargylamine inhibitors of flavoenzymes are proposed to form this adduct via a two-electron mechanism (Figure 2d), under blue light irradiation it is also possible to propose a one-electron mechanism leading to this same FAD adduct (Figure S9). The same pattern of reactivity was observed against PHBH (Figure S10), suggesting that under blue light irradiation cyclopropylamine 1 and propargylamine 2 can serve as general chemical probes for class A FMOs.

Structure-based design of covalent TDase inhibitors.

Structure-based molecular modeling was used to design more specific covalent TDase inhibitors bearing cyclopropylamine and propargylamine warheads. We selected the known TDase inhibitor aTC as the template scaffold for the covalent TDase inhibitors. aTC is a reversible inhibitor of both Type 1 and Type 2 TDases;²⁷ however, the inhibitor binding modes observed in X-ray cocrystal structures with Type 1 and Type 2 TDases are distinct. For Type 1 TDases, the aTC:TetX6 structure (PDB: 8ER0) shows that aTC adopts a substrate-like binding mode with the FAD-IN conformation (Figure 2b). For Type 2 TDases, the aTC:Tet50 structure (PDB: 5TUF) captures a novel binding mode with the FAD-OUT conformation (Figure 4a). A template model based on this published crystal structure recapitulated the crystallographically observed aTC binding mode in Tet50 using molecular docking in AutoDock Vina (Figure S11).⁵⁵ Molecular modeling suggested that the attachment of a cyclopropylamine (compound 3, Figure 4b) or propargylamine (compound 4, Figure 4c) warhead to the C9 position of the aTC D-ring via a methylene linker would orient the warhead favorably for reaction with the FAD cofactor. Based on the proposed mechanisms shown in Figure 2, the reactive atoms in the cyclopropylamine and propargylamine sidechains of compounds 3 and 4 are within 4.7 Å and 4.3 Å of the N5 atom of the FAD isoalloxazine ring system, respectively. Attempts to dock a N-methylpropargylamine derivative (compound 5) failed to recapitulate the aTC binding mode in Tet50 because of a steric clash within the residues that line the solvent exposed channel accessed by the C9-substituent. It is unknown whether aTC-based inhibitors can access this ligand binding mode in Type 1 TDases. However, it is clear that there is ample space for accommodation of the C9-side chains in compounds 3 and 4 should the substrate binding mode observed in the aTC:TetX6 structure (PDB: 8ER0) be adopted. This binding mode is predicted to position the reactive warheads far away from the FAD cofactor (>12 Å; Figure S12). Hence, we hypothesized that the observed inhibition mechanism, irreversible covalent or reversible noncovalent, would be predictive of inhibitor binding modes in Type 1 and Type 2 TDases.

Figure 4:

Molecular modeling and docking predicts that the placement of a cyclopropylamine or propargylamine at C9 of aTC orients the group favorably for reaction with the FAD cofactor in the Type 1 TDase Tet50. (a) Structures of the reversible Tet50 inhibitor aTC and X-ray crystal structure of aTC-bound Tet50, which adopts a binding mode in which the D-ring orients towards FAD-OUT (PDB: 5TUF). Structures of docked aTC cyclopropylamine 3 (b) and aTC propargylamine 4 (c) overlapping with good alignment to co-crystallized aTC in the Tet50 active site. The purple star denotes the predicted location of covalent attachment to N5 of FAD. Docking was performed using AutoDock Vina (Center for Computational Structural Biology, The Scripps Research Institute; v1.1.2), and structures were visualized using PyMOL (Schrodinger, LLC; v2.5.2).

Synthesis of covalent TDase inhibitors.

Prior semi-syntheses of aTC derivatives in our lab demonstrated that selectivity can be achieved through aromatic electrophilic substitution at the C-9 position of the electron-rich D-ring.³⁰ Notably, both reactive warheads earmarked for installation are derived from primary amine precursors. The Mannich reaction between alkylamines, aldehydes, and TC serves as an exceptionally efficient approach for preparing C9-aminomethylcycline derivatives of TC including the FDA-approved third-generation drug omadacycline.⁵⁶ Hence, we reasoned that the Mannich reaction could also be applied to the synthesis of C9-aminomethylcycline derivatives of aTC. Specifically, either 1-methylcyclopropylamine hydrochloride or propargylamine was treated with 1.1 equivalents of formaldehyde in methanol under refluxing conditions to generate the corresponding Schiff base. This mixture was combined with 0.2 equivalents of aTC in 1M HCl_aq and stirred for 16 hours at 40 °C to provide the corresponding Mannich products, 3-5 (Figure 5). The final purification of compounds 3, 4, and 5 was achieved via semi-preparative RP-C18 HPLC resulting in apparent yields of 69%, 30%, and 29%, respectively. The structures of compounds 3-5 were confirmed through multi-dimensional NMR analysis. Signals from the aTC scaffolds were consistent with previous reporting.³⁰ The side chain connection on C9 was proven by HMBC correlation between C10 (~ 155-160 ppm) and the methylene linker which are highlighted in Figures S24, S30, and S36.

Figure 5:

Synthesis of C9-cyclopropylamine and C9-propargylamine aTC derivatives 3, 4, and 5, respectively, via a one-step Mannich reaction.

In vitro inhibition of TDases.

The inhibitory properties of compounds 3, 4, and 5 towards the TDase-catalyzed oxidation of TC were evaluated using a steady-state kinetic assay monitored continuously by optical absorbance (A₄₀₀ tracks the disappearance of TC). This in vitro assay was used to determine the apparent IC₅₀ values for TDase inhibitors under a fixed concentration of TC, NADPH, and O₂. We used TetX7 and TetX6 as model Type 1 enzymes and Tet50 as a model Type 2 enzyme. We also evaluated the known TDase inhibitor aTC as a positive control. We found that aTC, compound 3, compound 4, and compound 5 were ~10-fold more potent inhibitors of TetX7 and TetX6 than Tet50 (Table 1). This preference for the inhibition of Type 1 TDases over Type 2 TDases is consistent with prior reports of aTC analogs.^27,30,31 The apparent IC₅₀ values of compounds 3, 4, and 5 are in line with those of the parent inhibitor aTC indicating that substitution by methylene-linked cyclopropylamine or propargylamine at the C9 position of the aTC D-ring is tolerated by both Type 1 and 2 TDases.

Time- and light-dependence of TDase inhibition.

A hallmark of covalent enzyme inhibition is time dependence given that longer exposure of the enzyme to the inhibitor allows for more inactivating turnovers.⁵⁷ Tet50 was incubated with an inhibitor, either aTC, compound 3, compound 4, or compound 5, for 0, 30, 60, 120, or 240 minutes prior to initiating the TDase reaction by the addition of TC and NADPH. The apparent IC₅₀ values were determined for each pre-incubation time point by monitoring the TDase velocity at A₄₀₀ under various inhibitor concentrations. The apparent IC₅₀ values were plotted against the pre-incubation times to determine if the TDase inhibition was time-dependent (Figure 6). When the reactions were shielded from light, the inhibition of Tet50 by aTC was not time-dependent (Figure 6a), whereas compounds 3 (Figure 6b), 4 (Figure 6c), and 5 (Figure 6d) all showed a clear decrease in the apparent IC₅₀ value correlating to longer enzyme–inhibitor pre-incubation times. The apparent IC₅₀ values for aTC remained constant between 410 ± 50 μM and 340 ± 40 μM over 240 min. The apparent IC₅₀ values for compounds 3-5 decreased from 430 ± 40 μM to 13 ± 1 μM, 600 ± 40 μM to 48 ± 7 μM, and 470 ± 60 μM to 70 ± 20 μM respectively, over 240 min. These data were fit to a single-phase exponential decay function ($y=A{e}^{-\frac{x}{\tau }}+y_0$) with apparent decay constants of 0.061 ± 0.003 min⁻¹, 0.013 ± 0.002 min⁻¹, and 0.040 ± 0.014 min⁻¹ for compounds 3-5, respectively. The apparent half-life values for the decay functions were 11.3 min (95% CI: 2.8 – 33.7 min), 54.5 min (95% CI: 32.3 – 175.9 min), and 18.9 min (85% CI: 10.1 – 141.3 min) for compounds 3-5, respectively.

Figure 6:

Time-dependent inhibition of the Type 2 TDase Tet50 and Type 1 TDase TetX7 by aTC, compound 3, compound 4, compound 5. (a-d) IC₅₀ plots for Tet50 under various pre-incubation times with aTC (a), compound 3 (b), compound 4 (c), and compound 5 (d) without exposure to external light. The bottom graph of each panel shows a plot of apparent IC₅₀ values versus time and fit to an exponential decay function ($y=A{e}^{-\frac{x}{\tau }}+y_0$). (e-h) IC₅₀ plots for Tet50 under varying pre-incubation times with aTC (e), compound 3 (f), compound 4 (g), and compound 5 (h) under blue light irradiation. The time domain indicates both the pre-incubation time of the enzyme with inhibitor and the duration of blue light irradiation. The bottom graph of each panel shows a plot of apparent IC₅₀ values versus time and fits to an exponential decay function ($y=A{e}^{-\frac{x}{\tau }}+y_0$). (i-l) IC₅₀ plots for TetX7 under various pre-incubation times with aTC (i), compound 3 (j), compound 4 (k), and compound 5 (l) without exposure to external light. The bottom graph of each panel shows a plot of apparent IC₅₀ values versus time and fit to an exponential decay function ($y=A{e}^{-\frac{x}{\tau }}+y_0$). Error bars represent standard deviations for at least two independent trials. All curve fitting and calculations were performed with GraphPad Prism 10 software.

When these same experiments were conducted under blue light irradiation we observed time-dependent inhibition of Tet50 by aTC and compounds 3-5 (Figure 6e-h). aTC is known to be photosensitive and undergoes rapid photooxidation which might produce a reactive intermediate that inactivates Tet50.^58,59 Hence, the photosensitivity of compounds 3-5 is likely due to the aTC core, and not the side chains. The apparent decay rates under blue light irradiation were significantly faster than the observed rates for reactions that were protected from light. The apparent single-phase exponential decay functions for aTC (Figure 6e), compound 3 (Figure 6f), and compound 5 (Figure 6h) under blue light produced apparent decay constants of 0.6 ± 0.2 min⁻¹, 0.9 ± 0.3 min⁻¹, and 0.5 ± 0.1 min⁻1, respectively, and the apparent half-life values of the decay functions were ~1.2 min, ~0.8 min, and ~1.4 min, respectively. Compound 4 showed slower time-dependent inactivation relative to aTC, compound 3, and compound 5 under these conditions and we were unable to confidently fit the data for compound 4 against Tet50 under blue light to an exponential decay function (Figure 6g). Over 7 min, the apparent IC₅₀ values for aTC, compound 3, and compound 5 decreased from 500 ± 60 μM to 70 ± 10 μM, from 430 ± 40 μM to 41 ± 5 μM, and from ~1200 μM to 107 ± 6 μM, respectively. A similar light-induced enhancement in the dose-dependent inhibition of Tet50 was observed for the nonspecific probe molecules 1 and 2 (Figure S13). Collectively, these results suggest that the substitution of aTC at the C9-position with cyclopropylamine (3) or propargylamine (4 and 5) groups imparts time-dependence for the inhibition of Type 2 TDases without the need for photoirradiation.

Analogous time- and light-dependent experiments were performed on PHBH using an NADPH consumption assay. It is important to note that monitoring for NADPH consumption is not a direct indicator of inhibition since uncoupled formation of H₂O₂ can provide a false positive and there is no direct monitoring for consumption of the natural substrate para-hydroxybenzoic acid with these inhibitors. In the dark, no inhibition of PHBH was detected for compounds 1 and 2 at concentrations up to 2.5 mM. However, with blue light irradiation potent time-dependent and dose-dependent inhibition of PHBH by 1 and 2 was observed (Figure S13). Inhibition was detected for aTC, compound 3, and compound 4 under dark conditions. The apparent IC₅₀ values were similar to those observed for Tet50 (Table 1) and time-dependence was observed for compounds 3 and 4, but not aTC (Figure S14). The apparent PHBH activity decay kinetic were generally slower than those observed for Tet50, but these findings still indicate that propargylamine- and cyclopropylamine-based TDase inhibitors could have off-target effects on other FAD-containing enzymes.

aTC is known to bind Type 1 TDases in a substrate-like binding mode predicted to orient the D-ring toward the opening of the active site and position a C9-substituent away from the FAD cofactor and into the open solvent space (Figures 2b and S12).²⁹ Hence, we were uncertain if the aTC analogs featuring cyclopropylamine (3) or propargylamine (4 and 5) substituents at C9 would adopt a binding conformation that facilitates a reaction with the FAD cofactor. We tested the time-dependence of the inhibition of the Type 1 TDase TetX7 under the same conditions described previously for Tet50. Contrary to Type 2 TDase inhibition, we did not observe any time-dependence for the inhibition of the Type 1 TDase TetX7 by aTC (Figure 6i), compound 3 (Figure 6j), compound 4 (Figure 6k), or compound 5 (Figure 6l). We were unable to explore the time dependence of the effects of the inhibitors under blue light irradiation because of the photoinstability of TetX7 (Figure S6). These findings are consistent with our previous report that C10-benzoate ester derivatives of aTC and C9-glycylcycline derivatives of TC bind to Type 1 TDases in a substrate-like orientation, with the bulky D-ring substituent pointing into the open solvent space (Figure 2b).^31,60,61 Hence, compounds 3-5 could serve as selective chemical probes to distinguish between Type 1 and Type 2 TDases on these basis of covalent tagging of the FAD cofactor (only Type 2 TDases are covalently modified by 3-5).

Reversibility of TDase inhibition.

In addition to their time-dependence, covalent enzyme inhibitors often result in irreversible inhibition.^32,62-64 The reversibility of Type 1 (Tet50) and Type 2 (TetX7) TDase inhibition by aTC, compound 3, compound 4, and compound 5 was tested via exhaustive dialysis of the respective enzyme-inhibitor complexes. Tet50 was pre-incubated with aTC, compound 3, compound 4, or compound 5 at concentrations near the apparent IC₅₀ values (Table 1) for 60 min in the dark or 15 min under blue light irradiation followed by exhaustive dialysis for ~24 h at 4 °C. TetX7 was treated under the same conditions but only in the dark because of the photoinstability of TetX7 under blue light irradiation (Figure S6). TDase activity was measured before and after dialysis relative to that of a no inhibitor control reaction, which was normalized to 100% activity. For Tet50 under dark conditions, we observed a significant recovery of enzymatic activity after dialysis of the aTC-treated samples whereas the samples treated with compounds 3-5 produced a significant decrease in Tet50 activity after dialysis (Figure 7a). Under blue light irradiation, all three inhibitors including aTC produced greater inhibition of Tet50 post-dialysis (Figure 7b). The treatment of TetX7 with all three inhibitors in the dark resulted in a significant recovery of activity after exhaustive dialysis (Figure 7c). Collectively, both exhaustive dialysis and time-dependent inhibition studies support a model of irreversible inhibition by compounds 3-5 against Tet50 under dark conditions and a model of reversible inhibition by aTC and compounds 3-5 against TetX7 under dark conditions. The inhibition of Tet50 by aTC under dark conditions was also reversible, whereas blue light irradiation of Tet50 in the presence of aTC and compounds 3-5 resulted in time-dependent irreversible inactivation. These results are consistent with the proposed inhibitor binding modes for Type 1 TDase TetX7 (Figures 2b and S12) and Type 2 TDase Tet50 (Figure 4) and supports the observed covalent tagging of FAD only in Type 2 TDases by the cyclopropylamine and propargylamine warheads appended to the C9 position of aTC.

Figure 7:

Exploring the reversibility of TDase inhibition. (a) Inhibited Tet50 activity under dark incubation conditions prior- and post- dialysis. (b) Inhibited Tet50 activity under blue light incubation conditions prior- and post- dialysis. (c) Inhibited TetX7 activity under dark incubation conditions prior- and post- dialysis. The error bars represent the standard deviation for three independent trials. *P≤0.05; ns: P>0.05

Covalent TDase inhibitors rescue TC antibacterial activity.

With strong supporting evidence that compounds 3–5 act as selective TDase inhibitors in vitro, these inhibitors were tested in combination with TC against strains of E. coli DH5αZ1 expressing TetX7 (Type 1) or Tet50 (Type 2) from an inducible pZE24 plasmid. Minimum inhibitory concentrations (MICs) were determined using the broth microdilution method according to CLSI guidelines.⁶⁵ Visual inspection of bacterial growth and measurement of the OD₆₀₀ were used to determine the lowest concentration at which bacterial growth is inhibited. The MICs for TC against the empty vector control, TetX7::pZE24, and Tet50::pZE24 were 1, 16, and 16 μg/mL, respectively. In prior work, we reported the MICs of aTC against the empty vector control, TetX7::pZE24, and Tet50::pZE24 as 4, 8, and 8 μg/mL, respectively, and we found that aTC at as low as 2 μg/mL can rescue the growth inhibitory activity of TC at 8 μg/mL against resistant strains.^28,30,31 The MIC values for compounds 3-5 were determined against the empty vector control, TetX7::pZE24, and Tet50::pZE24 strains of E. coli DH5αZ1 in the absence of TC and in the presence of TC at 8 μg/mL. Compounds 3-5 displayed moderate growth inhibitory activity with MICs ranging from 16–64 μg/mL against the three strains (Table 2). The addition of 8 μg/mL TC significantly increased the growth inhibitory activity of the TC/inhibitor combinations against all strains. The apparent MICs of the inhibitors against resistant strains were 4-to-8-fold lower than those of the inhibitor or TC alone. While cyclopropylamine 3 was generally more potent than propargylamines 4 and 5 against TetX7 and Tet50 in vitro (Table 1), propargylamine 4 was the best TDase inhibitor in live cells. Compound 4 had optimal activity against TetX7 with an observed recovery of TC growth inhibition at 2 μg/mL, which is in line with the best TDase inhibitors developed to date.³¹ Whether the whole cell activity of these inhibitors are provided by covalent inhibition is still to be determined. It is relatively difficult to determine the time domain of whole cell activity. Both reversible and irreversible modes of TDase inhibition have promise in rescuing TC activity. It is not clear if there is an advantage to covalent TDase inhibition since it could be possible for TDase-expressing cells to overwhelm covalent inhibitors in vivo by overexpressing the enzyme and destroying the inhibitors in a sacrificial manner, or via cofactor replacement where FAD-inhibitor adducts are replaced with fresh FAD.

Table 2:

Minimum inhibitory concentration (MIC in μg/mL) against E. coli DH5αZ1 expressing TetX7 or Tet50 from the pZE24 plasmid in the absence (− TC) or presence (+ TC) of 8 μg/mL TC.

Strain/ Compound	Empty		TetX7		Tet50
	− TC	+ TC	− TC	+ TC	− TC	+ TC
3	32	<1	64	8	64	16
4	32	<1	16	2	32	8
5	64	<1	64	2	32	16

Conclusions and Mechanistic Considerations

Class A FMOs are well-studied flavoenzymes known to engage primarily in two-electron transfer chemistry initiated by the reduction of the FAD cofactor with NADPH. Subsequent formation of the C4a-peroxy flavin intermediate can lead to oxygen group transfer to a bound substrate such as C11a-hydroxylation of TC by TDases.^66-69 This study shows that in the ground state, class A FMOs including TDases can be forced down either one-electron or two-electron transfer pathways using cyclopropylamine or propargylamine groups as electron transfer agents, respectively. Photoexcitation of TDases generates a photoexcited state FAD* cofactor with increased oxidative potential and enhanced reactivity toward cyclpropylamine- and propargylamine-containing compounds including the nonspecific probe molecules 1 and 2. In the dark, compounds 1 and 2 are not TDase inhibitors. When irradiated with blue light, compounds 1 and 2 inhibit Tet50 in a time-dependent manner. For cyclopropylamine 1, we measured an apparent IC₅₀ of 390 ± 30 μM against Tet50 after 8 min of pre-incubation under blue light irradiation (Figure S13). Since the parent aTC scaffold is a time-dependent and irreversible inhibitor of Tet50 under blue light irradiation, the observed inhibition of Tet50 by compounds 3-5 under photoirradiation is likely a combined effect between covalent FAD tagging by the reactive warheads and the general photoreactivity of the aTC core scaffold.

Potent whole-cell activity of inhibitors 3-5 was observed against strains of E. coli expressing either Type 1 or Type 2 TDases despite differences in the in vitro potency of up to ~40-fold favoring the inhibition of Type 1 TDases (Tables 1 and 2). This trend is consistent with our prior studies on C9- and C10-substituted aTC analogs as reversible TDase inhibitors and adjuvants of TC antibacterial activity.^27,30,31 Biochemical characterization of cyclopropylamine analog 3 and propargylamines 4 and 5 is consistent with a model of non-covalent reversible inhibition of the Type 1 TDase TetX7 and a model of covalent irreversible inhibition of the Type 2 TDase Tet50. We propose that this difference in the inhibition modes is due to the adoption of unique inhibitor binding modes making these inhibitors useful chemical probes for distinguishing Type 1 and Type 2 TDases (see Figures 2a, b, and S12 for inhibitor binding modes). Compound 5 failed to dock via the binding mode shown in Figure 4c due to steric occlusions introduced by the N-methyl group that prevented the C9-substituent from extending into the solvent exposed channel adjacent to the FAD isoalloxazine core positioned in the ‘out’ conformation (Figure S11). Despite this difference, N-methylated derivative 5 generated results similar to those of the parent compound 4 when evaluated in biochemical and whole cell assays. The apparent IC₅₀ value of compound 5 against TetX7 was 71 ± 14 μM and the in vitro inhibition of Tet50 was noticeably weaker (Table 1). However, compound 5 still displayed time- and light- dependent irreversible inactivation of Tet50 (Figure 6d, h), whereas the inhibition of TetX7 was reversed upon exhaustive dialysis (Figure 7). Finally, compound 5 was able to rescue TC activity against E. coli expressing TetX7 and Tet50 at concentrations of 2 μg/mL and 16 μg/mL, respectively (Table 2). We conclude that TDases show plasticity towards ligand binding beyond what is implied by molecular docking studies. We propose that the functionalized aTC-based TDase inhibitors from this study can be used to distinguish Type 1 and Type 2 TDases by testing the reversibility and time-dependence of TDase inhibition. Further optimization of the propargylamine and cyclopropylamine aTC analogs might lead to more selective covalent inhibition of TDases with reduced potential for off-target toxicity. Our results validate that both reversible non-covalent and irreversible covalent inhibition modes are effective in rescuing TC activity against TDase-expressing pathogens.

Experimental Procedures

General materials and methods.

All organic solvents, including deuterated NMR solvents and reagent-grade chemicals used in the preparation or analysis of synthetic compounds, were purchased from Millipore-Sigma (St. Louis, MO) unless otherwise stated and were used without further purification. aTC (HCl salt) was purchased from Chemodex (United Kingdom). 1-Methylcyclopropanamine (HCl salt) was purchased from AmBeed (Arlington Hts, IL). FAD (disodium salt hydrate) was purchased from Chem-Impex (Wood Dale, IL). All in vitro assays were prepared open to air in nondegassed buffer solutions. NMR spectra were obtained on a Varian Unity-Plus 300 MHz, Varian Unity-Inova 500 MHz, or Agilent PremiumCompact+ 600 MHz spectrometer in 5 mm type 1, class A borosilicate glass NMR tubes (Wilman LabGlass part no. 535-PP-8). All free induction decay files (FIDs) were processed using Mestrenova software (version 12.0.4). Chemical shifts (δ) are reported in parts per million (ppm) and referenced to residual nondeuterated solvent. Coupling constants (J) are reported in hertz (Hz). All in vitro TDase and FAD reactions were monitored by measuring optical absorbance (A₄₀₀) on an Agilent Cary 50 UV–visible spectrophotometer using polystyrene cuvettes or a Tecan Infinite M Plex microplate reader using Costar 96-well transparent polystyrene plates. All enzyme concentrations were determined by measuring optical absorbance (A₂₈₀) on a Thermo Fisher NanoDrop 2000c spectrophotometer. Low resolution LC-MS was performed using an Agilent 6130 single quadrupole instrument with G1313 autosampler, G1315 diode array detector, and 1200 series solvent module and separated using a Phenomenex Gemini C18 column, 50 × 2 mm (5 μm) with guard column cassette and a linear gradient of 0% acetonitrile and 0.1% formic acid to 95% acetonitrile and 0.1% formic acid over 25 min at a flow rate of 0.5 mL/min before analysis by electrospray ionization (ESI+). Synthetic compounds were purified by preparative HPLC using an HP1050 system and a reverse phase Luna 10 mm C18(2) 100 Å column (250 mm × 21.2 mm) from Phenomenex fitted with a guard column of the same matrix (15 mm × 21.2 mm). HPLC solvents were 0.1% formic acid in H₂O (A) and 0.1% formic acid in ACN (B), with a gradient formed from 0% B to 95% B over 20 min at a flow rate of 9 mL/min. LC–MS and HPLC data were processed using ChemStation software version B.04.02 SP1. SDS-PAGE analysis was carried out using Bio-Rad Any kD precast polyacrylamide gels with staining by Coomassie brilliant blue and comparison to a Bio-Rad Precision Plus Protein Dual Xtra pre-stained protein standard ladder. The molecular formulas and structures of all synthetic compounds were confirmed through HRMS and multi-dimensional NMR analyses, respectively (Figures S15-S36). The purity of the compounds was confirmed to be >95% by analytical LCMS prior to all biological testing (Figures S37-S41).

Cloning, expression, and purification of TDases.

All genes encoding FMOs used in this study (tetX6 – QHN11884.1; tetX7 – AMP54443.1; tet50 – AKQ05894.1; pobA (PHBH) – P00438) were cloned into the BamHI/NdeI (tetX6, tetX7, and tet50) or NdeI/HindII (pobA) restriction sites of a pET28b(+) vector (Novagen) with a 6-His tag at the N-terminus. The cloned construct was transformed into BL21-Star (DE3) competent cells (Life Technologies). The cloning and expression of TetX6, TetX7, and Tet50 have been described previously by our group.^16,29 The pobA gene encoding for PHBH was ordered from GenScript and was codon optimized for expression in E. coli (Table S1). Cells were cultured at 37 °C in lysogeny broth (LB) containing 0.03 mg/mL kanamycin (final concentration); once the culture reached an OD₆₀₀ of ~0.6, the cells were cooled to 4 °C. Protein expression was induced by the addition of 1 mM IPTG (final concentration), and cells were grown at 15 °C for 12–15 h. To harvest protein, the induced cells were pelleted by centrifugation at 5000 rpm for 15 min (4 °C) and resuspended in lysis buffer (50 mM K₂HPO₄, 500 mM NaCl, 20 mM imidazole, 10% glycerol, 5 mM BME, pH 8.0) containing SIGMAFAST protease inhibitor (Millipore-Sigma) and cooled to 4 °C. Cell suspensions were transferred to falcon tubes, flash frozen in liquid nitrogen and stored at −80 °C. Frozen cell suspensions were thawed and lysed using a Qsonica ultrasonicator, and the resultant lysate was clarified via ultracentrifugation at 45,000 rpm for 35 min at 4 °C. The clarified supernatant was transferred to a fritted column containing washed and equilibrated Ni-NTA resin and incubated for 30–45 min with gentle rocking. The resin was then washed with lysis buffer (2 × 40 mL), and the protein was eluted from the resin with elution buffer (4 × 10 mL elution, 50 mM K₂HPO₄, 500 mM NaCl, 5 mM BME, 300 mM imidazole, 10% glycerol, pH 8.0). Fractions containing the desired proteins (as judged by SDS-PAGE analysis) were combined and transferred to a 10,000 molecular weight cutoff (MWCO) Snakeskin^® dialysis tubing (Thermo Scientific) and equilibrated in dialysis buffer (50 mM K₂HPO₄ pH 8.0, 150 mM NaCl, 1 mM DTT) overnight. Dialyzed protein solutions were concentrated using a 30,000 MWCO Amicon^® centrifugal filter (Millipore-Sigma), and the concentrated protein solution was flash frozen as beads in liquid nitrogen (50 μL portions) and stored at −80 °C. The % FAD for each batch of protein varies and is determined by back calculation from a denaturated aliquot of enzyme using optical absorbance with the published extinction coefficient of 11,300 M⁻¹cm⁻¹ for FAD.⁷⁰ On average, batches of Tet50, TetX7 and TetX6 ranged from 20 to 60% FAD with an average of 30% FAD for batches of Tet50, 50% FAD for batches of TetX7 and 50% FAD for batches of TetX6.

Synthesis and characterization of probes 1 and 2.

N-(1-methyl)cyclopropylbenzylamine (1).

To a clean, dry 100 mL round-bottom flask, equipped with a magnetic stir bar, was added 1-methylcyclopropylamine hydrochloride (108 mg, 1 mmol), K₂CO₃ (166 mg, 1.2 mmol) and 10 mL of acetone. Benzyl bromide (143 μL, 1.2 mmol) was added to the reaction mixture dropwise with stirring. The reaction mixture was allowed to stir at room temperature overnight. When the reaction was complete as determined by LC-MS analysis, the reaction mixture was filtered through a 0.45 μm PTFE syringe filter. The filtrate was concentrated under reduced pressure to provide the crude product as an oil. Final purification by RP-C18 prep-HPLC provided compound 1 as light-yellow oil (46 mg, 0.29 mmol, 29% yield). 1H NMR (500 MHz, Chloroform-d): δ = 7.56 – 7.41 (m, 2H), 7.33 (d, J = 4.9 Hz, 3H), 4.03 (s, 2H), 1.45 (s, 3H), 1.04 (s, 2H), 0.48 (s, 2H). ¹³C NMR (125 MHz, Chloroform-d): δ = 132.1, 130.2, 128.9, 128.7, 48.8, 35.9, 19.5, 11.6. Prep-HPLC t_R = 6 min. MS (ESI+): [M+H]⁺; expected 162.2, found 161.9.

N-methyl-N-benzyl-propargylamine (2).

To a clean, dry 100 mL round-bottom flask, equipped with a magnetic stir bar, was added N-methylbenzylamine (130 μL, 1 mmol), K₂CO₃ (166 mg, 1.2 mmol) and 10 mL of acetone. Propargyl chloride (87 μL, 1.2 mmol) was added to the reaction mixture dropwise with stirring. The reaction was allowed to stir at room temperature overnight. When the reaction was complete as determined by LC-MS analysis, the reaction mixture was filtered through a 0.45 μm PTFE syringe filter. The filtrate was concentrated under reduced pressure to provide the crude product as an oil. Final purification by RP-C18 prep-HPLC provided compound 2 as colorless oil (35 mg, 0.22 mmol, 22% yield). 1H NMR (500 MHz, Chloroform-d): δ = 7.48 – 7.42 (m, 2H), 7.42 – 7.34 (m, 3H), 3.96 (s, 2H), 3.58 (d, J = 2.5 Hz, 2H), 2.61 (s, 3H), 2.52 (t, J = 2.5 Hz, 1H). ¹³C NMR (125 MHz, Chloroform-d): δ 134.1, 130.1, 128.8, 128.5, 76.1, 75.6, 58.6, 43.5, 40.3. Prep-HPLC t_R = 6.0 min. MS (ESI+): [M+H]⁺; expected 160.2, found 160.2.

General procedure for the synthesis of C9-substituted aTC inhibitors (3-5).

To a clean, dry 25 mL round-bottom flask, equipped with a magnetic stir bar and reflux condenser, was added 1-methylcyclopropylamine hydrochloride or N-methylbenzylamine (0.3 mmol), formaldehyde (25 μL, 0.33 mmol) and 3 mL methanol. The reaction system was heated to 64 °C for 3 h. To another clean, dry 25 mL round-bottom flask, equipped with a magnetic stir bar, was added aTC (46.3 mg, 0.1 mmol) and 2 mL methanol under an argon atmosphere. The reaction system in the first flask was added to second flask dropwise using a syringe while the reaction was heated to 40 °C. ~ 0.2 mL aqueous 1 M HCl was added to dissolve all starting materials. After 16 h, the reaction progress was checked by LC-MS analysis. When the reaction was complete, the reaction mixture was concentrated under reduced pressure to yield a brown residue. The residue was dissolved in MeOH, filtered through a 0.45 μm PTFE syringe filter, and purified by RP-C18 prep-HPLC to provide the desired product, compound 3, 4, or 5, as the corresponding formic acid salts. The compound was then dissolved in a minimal amount of MeOH with ~ 0.1 mL 1M HCl, diluted with 20 mL of Et₂O, and cooled to 4 °C to induce precipitation of the product. After ~3 h, the solvent was removed via pipet, and the yellow solid was triturated with fresh Et₂O. The solid was dried under a stream of N₂ gas to yield compound 3, 4, or 5 as the corresponding HCl salt.

(4S,4aS,12aS)-4-(dimethylamino)-3,10,11,12a-tetrahydroxy-6-methyl-9-(((1-methylcyclopropyl)amino)methyl)-1,12-dioxo-1,4,4a,5,12,12a-hexahydrotetracene-2-carboxamide hydrochloride (3).

Compound 3 was prepared according to the general procedure and was obtained in 69% yield as a yellow solid. ¹H NMR (600 MHz, DMSO-d6) δ 7.88-7.94 (d, 1H)^a, 7.51-7.56 (d, 1H)^a, 4.92-4.94 (d, 1H), 4.28 (s, 2H), 3.54-3.59 (m, 1H), 3.51-3.54 (m, 1H), 2.95-3.01 (m, 11H), 2.90 (s, 6H), 2.37-2.44 (s, 3H)^b, 1.51 (s, 3H), 1.12-1.15 (t, 2H), 0.65-0.68 (t, 2H). ¹³C NMR (150 MHz, DMSO-d6): δ 188.0, 162.9, 156.2, 139.1, 135.3, 131.8, 122.2, 115.1, 113.8, 111.6, 109.0, 76.3, 67.0, 66.0, 42.3, 38.6, 36.5, 26.3, 18.3, 14.0, 10.8. Prep-HPLC t_R = 11.0 min. MS (ESI+): [M+H]⁺; expected 510.2, found 510.2. High-resolution mass spectrometry (HRMS) (time-of-flight (TOF) MS ESI+): calcd for C₂₇H₃₁N₃O₇ [M+H]⁺, 510.2240; found, 510.2227. ^aAppears as two doublets, indicating the formation of the epimer during the synthesis. ^bAppears as two singlets representing the methyl group from each epimer respectively.

(4S,4aS,12aS)-4-(dimethylamino)-3,10,11,12a-tetrahydroxy-6-methyl-1,12-dioxo-9-((prop-2-yn-1-ylamino)methyl)-1,4,4a,5,12,12a-hexahydrotetracene-2-carboxamide hydrochloride (4).

Compound 4 was prepared according to general procedure and was obtained in 30% yield as a yellow solid. ¹H NMR (600 MHz, DMSO-d6) δ 7.81-7.86 (d, 1H)^a, 7.51-7.56 (d, 1H)^a, 4.92-4.94 (d, 1H), 4.30 (s, 2H), 3.90-3.92 (d, 2H), 3.54-3.59 (m, 1H), 3.51-3.54 (m, 1H), 2.95-3.01 (m, 1H), 2.90 (s, 6H), 2.37-2.44 (s, 3H)^b. ¹³C NMR (150 MHz, DMSO-d6): δ 189.9, 188.0, 156.2, 139.1, 134.7, 122.2, 115.2, 113.2, 111.5, 79.6, 77.8, 77.0, 75.0, 66.0, 64.9, 43.3, 38.6, 35.5, 26.4, 14.0. Prep-HPLC t_R = 10.2 min. MS (ESI+): [M+H]⁺; expected 494.2, found 494.2. High-resolution mass spectrometry (HRMS) (time-of-flight (TOF) MS ESI+): calcd for C₂₆H₂₇N₃O₇ [M+H]⁺, 494.1927; found, 494.1909. ^aAppears as two doublets, indicating the formation of the epimer during the synthesis. ^bAppears as two singlets representing the methyl group from each epimer respectively.

(4S,4aS,12aS)-4-(dimethylamino)-3,10,11,12a-tetrahydroxy-6-methyl-9-((methyl(prop-2-yn-1-yl)amino)methyl)-1,12-dioxo-1,4,4a,5,12,12a-hexahydrotetracene-2-carboxamide (5).

Compound 5 was prepared according to general procedure and was obtained in 29% yield as a yellow solid. ¹H NMR (300 MHz, Methanol-d4) δ 7.73 (d, J = 8.8 Hz, 1H), 7.61 (d, J = 8.6 Hz, 1H), 4.65 (dd, J = 12.9, 3.8 Hz, 1H), 4.50 (dd, J = 12.9, 4.0 Hz, 1H), 4.28 (d, J = 3.1 Hz, 1H), 4.16 (d, J = 6.7 Hz, 2H), 3.68 (dd, J = 17.7, 4.8 Hz, 1H), 3.52 – 3.45 (m, 1H), 3.44 (s, 1H) 3.13-3.09 (d, 1H), 3.08 (s, 6H), 2.96 (s, 3H), 2.48 (s, 3H). ¹³C NMR (150 MHz, Methanol-d4): δ 160.0, 137.3, 134.8, 127.7, 125.6, 118.5, 113.4, 111.8, 82.9, 80.5, 79.2, 74.3, 71.0, 56.3, 47.4, 45.5, 44.3, 41.8, 38.5, 31.7, 15.7. Prep-HPLC t_R = 10.7 min. MS (ESI+): [M+H]⁺; expected 508.5, found 508.7.

Optical absorbance spectroscopy and LC-MS detection of covalent FAD adducts.

A solution of 20 μM Tet50 (final protein concentration based on A₂₈₀; working concentration of active enzyme is predicted to be 4–12 μM based on FAD content) in the presence of 6 mM probe, either 1 or 2, in 100 mM TAPS buffer (pH = 8.5) in polystyrene cuvettes was analyzed by scanning optical absorbance (280-550 nm) using the Agilent Cary 50 UV–visible spectrophotometer. The unexposed ‘dark’ samples were protected from external light by keeping the cuvettes inside the spectrophotometer throughout the entire data collection period. Absorbance spectra (280-550 nm) were acquired at 1 min interval for a total of 10 min. For the ‘exposed’ samples, the cuvette was removed from the spectrophotometer between spectral acquisitions (1 min) and illuminated by a 700 lumen blue LED blub from a distance of 20 cm for 45 s (cumulative light exposure time of 7.5 min). All data generated from the UV-Vis spectrophotometer were replotted with GraphPad Prism 10 software. Consecutive acquisitions were stacked in one graph with gradient blue lines (from 0 min to 10 min, darkest to lightest).

The samples after spectral acquisition under blue light exposure were then quenched with enough methanol to denature the enzymes. The solid precipitates were removed by centrifugation at 13,000 rpm for 5 min. The supernatant was removed, filtered through a 0.2 micron syringe filter, and analyzed by LC-MS and HRMS to detect the covalent FAD adducts.

Time-dependence of TDase inhibition.

TDase activity was determined from the apparent velocities of TC degradation as judged by a decrease in A₄₀₀ over time with variable enzyme-inhibitor preincubation times and in the presence of varying concentrations of inhibitors at room temperature. PHBH activity was determined similarly using para-hydroxybenzoic acids as substrate and monitoring for NADPH consumption via A₃₄₀. Reaction samples were prepared in 100 mM TAPS buffer (pH 8.5) with 500 μM NADPH, 5 mM MgCl₂, 100 μM TC, varying concentrations of inhibitor, and 1 μM TDase (final protein concentration judged by A₂₈₀; working concentration of active enzyme is predicted to be 0.2–0.6 μM based on FAD content). Tet50 (1 μM final concentration) was incubated with inhibitors at final working concentrations of 0, 4, 8, 16, 31, 63, 125, 250, and 500 μM. TetX7 (1 μM final concentration) was incubated with inhibitors at final working concentrations of 0, 0.5, 1, 2, 4, 8, 16, 32, and 64 μM. Reaction samples were prepared in a Costar 96-well flat bottom clear, polystyrene plate and incubated under dark or a 700 lumen blue LED bulb held at a distance of 20 cm for various durations. Incubation times were 0, 30, 60, 120, and 240 min under dark or 0, 1, 2, 4, and 8 min under blue light. Following incubation, the enzyme-inhibitor samples were diluted two-fold with substrate solution containing TC and NADPH, and the initial rate was measured by monitoring optical absorbance spectroscopy at 400 nm for 3 min. Initial enzyme velocities were determined by linear regression over the linear range of the reaction (typically between 0 and 1 min). The rates were expressed as percent activity remaining by normalizing to the rate of TDases that had not been treated with inhibitors. The velocities were plotted against the logarithm of inhibitor concentration, and apparent IC₅₀ plots were analyzed using GraphPad prism 10 software. The apparent IC₅₀ values from 0 min incubation were summarized in Table 1 as an indicator of the inhibitors’ reversible binding affinity compared with substrates. All apparent IC₅₀ values were plotted against the duration of incubation time and fitted to an exponential decay function. All data points are shown as means and error bars represent standard deviation at least two independent experiments.

Testing the reversibility of TDase inhibition by exhaustive dialysis.

The reversibility of TDase inhibition was directly assessed by exhaustive dialysis of TDase-inhibitor complexes followed by analysis in the in vitro activity assay. TDases (40 μL of 110 μM protein stock concentration as judged by A₂₈₀; working concentration of active enzyme is predicted to be 22–66 μM based on FAD content) were mixed with 760 μL of inhibitor stock solution, 1.64 mM stock for Tet50 or 0.49 mM stock for TetX7, targeting a final concentration of inhibitor near the pre-determined IC₅₀ values of 284 μM for Tet50 or 45 μM for TetX7 (Table 1). The mixtures were incubated in the dark for 60 min or under blue light for 15 min to induce covalent inhibition at room temperature. The samples were dialyzed in a 50 mL Thermo Scientific^™ Slide-A-Lyzer^™ MINI Dialysis Devices, 10K MWCO overnight (~18 h) at 4 °C against 100 mM TAPS buffer (pH 8.5). TDase activity was determined before and after dialysis by addition of 200 μL enzyme-inhibitor pre-incubation mixture to 900 μL reaction mixture prepared in 100 mM TAPS buffer (pH 8.5) with 500 μM NADPH, 5 mM MgCl₂, and 100 μM TC (all final concentration) in a polystyrene cuvette (1 mL working volume). Reaction progress was monitored at A₄₀₀ (decrease in TC absorbance band) for 3 min using the Agilent Cary 50 UV–visible spectrophotometer. Initial enzyme velocities were determined by linear regression using Agilent Cary WinUV Software over the linear range of the reaction (typically between 0 and 1 min). The remaining enzymatic activity was normalized as a percent of the non-inhibited controls. All data points are shown as means and error bars represent standard deviations from three independent experiments.

Minimum inhibitory concentration (MIC).

MIC panels were prepared in 96-well flat-bottom microplates (Corning) by two-fold serial dilution of inhibitors in cation-adjusted MH-II broth (BD) supplemented with 50 μg/mL kanamycin (KAN50), 1 mM IPTG, and with or without 16 μg/mL of TC. The panels were stored at −80°C before use. Single colonies of E. coli DH5αZ1 containing TetX7, Tet50, or empty vector in pZE24 expression system were grown in MH-II broth with KAN50 and growth for overnight at 37°C. On the day of the experiment, MIC panels were thawed at room temperature. Overnight cultures were subcultured in fresh MH-II broth with KAN50 and 1 mM IPTG, grown to exponential phase (OD600 of 0.3–0.8), then diluted in MH-II broth with KAN50 and 1 mM IPTG. The diluted cells were inoculated into the MIC panel at a 1:1 ratio, resulting in a final concentration of ~ 5 × 10⁵ CFU/mL cells in each well. The panels were incubated at 37°C for 20 h, then scored by visual inspection. Potential synergistic effects between inhibitors and TC were evaluated by comparing the MIC of inhibitors alone to the MIC of inhibitors in combination with TC. Each test was performed in triplicate, with no-antibiotic and no-cell control wells.

Supplementary Material

Experimental methods, supplementary figures, supplementary tables, plasmid sequences, NMR spectra, and LC–MS chromatograms

Abbreviations

aTC

anhydrotetracycline

FAD

flavin adenine dinucleotide oxidized form

FMO

flavin monooxygenase

KMO

kynurenine-3-monooxygenase

MAO

monoamine oxidase

NADPH

nicotinamide adenine dinucleotide phosphate reduced form

NADP⁺

nicotinamide adenine dinucleotide phosphate oxidized form

PHBH

4-hydroxybenzoate-3-monooxygenase

TC

tetracycline

TDase

tetracycline destructase