Bioreductively Activated Reactive Oxygen Species (ROS) Generators as MRSA Inhibitors

Abstract

The number of cases of drug resistant Staphylococcus aureus infections is on the rise globally and new strategies to identify drug candidates with novel mechanisms of action are in urgent need. Here, we report the synthesis and evaluation of a series of benzo[b]phenanthridine-5,7,12(6H)-triones, which were designed based on redox-active natural products. We find that the in vitro inhibitory activity of 6-(prop-2-ynyl)benzo[b]phenanthridine-5,7,12(6H)-trione (1f) against methicillin-resistant Staphylococcus aureus (MRSA), including a panel of patient-derived strains, is comparable or better than vancomycin. We show that the lead compound generates reactive oxygen species (ROS) in the cell, contributing to its antibacterial activity.

Because of the high levels of morbidity and mortality, drug-resistant infections have now become a major public health problem globally.¹ The pathogen Staphylococcus aureus (S. aureus), for example, has acquired resistance to several antibiotics including the methicillin-based drugs. Methicillin-resistant S. aureus (MRSA) strains are fast becoming resistant to other frontline antibiotics as well.² The global pipeline for new antibiotics is weak, and hence, the development of new strategies that specifically address drug resistance is necessary.

Redox-active natural products were long considered as innocuous byproducts of metabolism of microorganisms without any major function.^3,4 It is increasingly clear that these secondary metabolites are produced to mediate a number of cellular processes including gene expression, interspecies communication, and defense.^3,4 The ability of such small molecules to keep competitors in check through the generation of reactive oxygen species (ROS) has attracted attention for new drug development.^3,5 ROS including superoxide radical O₂^–•, hydrogen peroxide H₂O₂, and hydroxyl radical ^•OH are generated as a natural consequence of respiration but can cause cellular damage at elevated levels (Scheme 1).⁶

Scheme 1. Reactive Oxygen Species (ROS) and Their Possible Cellular Effects.

Potentiating these ROS in cancers using small molecules has been considered as a possible drug design strategy.⁷ For example, deoxynyboquinone, a natural product derivative, has potent tumoristatic activity in animal models, and its efficacy is, in part, dependent on generation of ROS (Chart 1).^8,9 The potential for ROS generators to inhibit growth of certain mycobacteria is also reported.^10,11 The oxidative stress-inducing antimalarial drug artemisinin, when conjugated with mycobactin,¹² and the antileprosy drug Clofazimine¹³ have potent inhibitory effects on multidrug-resistant strains of Mycobacterium tuberculosis. Here, we report results of design and synthesis of natural product-inspired ROS generators and their potential to inhibit multidrug-resistant S. aureus.

Chart 1.

We proposed to synthesize and study benzo[b]phenanthridine-5,7,12(6H)-triones of general structure 1 (Scheme 2). This scaffold contained the amide adjoining the quinone of deoxynyboquinone and the aryl ring-like jadomycins,¹⁴ another class of natural products that are known to cleave DNA in the presence of metal ions such as Cu(II)¹⁵ presumably through the generation of hydroxyl radical (Chart 1).¹⁶1 is predicted to undergo bioreduction to its semiquinone,¹⁷ which then is reoxidized by molecular oxygen, forming O₂^–• (Scheme 2). During this oxidation, 1 is regenerated and becomes available for further ROS generation via bioreduction (See Supporting Information, Scheme S3).

Scheme 2. Proposed Mechanism of O₂^–• generation from 1.

The synthesis of 1 can be achieved in two steps from 2-(2-formyl-phenyl)-1,4-naphthoquinone 2 (Table 1), which in turn can be synthesized from commercially available 2-bromo-1,4-naphthoquinone (Supporting Information).¹⁴ The addition of an amine to 2 would result in the formation of an imine, which sets up the molecule for an intramolecular 6π-electrocyclic ring closure to afford the alcohol 3; subsequent oxidation of 3 should give the desired lactams of structure 1 (Table 1).¹⁴

Table 1. One-Pot Synthesis of 1a–1n by the Reaction of 2 with a Primary Amine.

entry	R	compd	% yielda
1	Me	1a	73
2	Et	1b	50
3	nPr	1c	45
4	cyclohexyl	1d	23
5	allyl	1e	21
6	propargyl	1f	20 (37)b
7	CH2CO2Me	1g	39
8	benzyl	1h	39
9	2-CF3PhCH2	1i	14
10	4-ClPhCH2	1j	20
11	4-OMePhCH2	1k	17
12	4-NO2PhCH2	1l	22
13	4-CF3PhCH2	1m	47
14	(3,4,5-triOMe)PhCH2	1n	13

^aIsolated yield.

^bAddition of H₂O₂ (10 equiv) to the reaction mixture.

When 2 was reacted with methylamine in an open container, we found 1a as the major product in 73% yield implying that the intermediate alcohol (3a, R = Me, Scheme S2, Supporting Information) was oxidized during the reaction (Table 1). Using this one-pot procedure, compounds 1b–1g were prepared from 2 (Table 1, entries 2–7). The use of an aromatic amine such as aniline did not produce the desired product but instead gave an inseparable mixture of products. The one-pot methodology was, however, found to be compatible with benzylamines, and 1h–1n were prepared in moderate yields (Table 1, entries 8–14).

To test if this library of compounds had an effect on the growth of bacterial cells, the minimum inhibitory concentration (MIC) was determined for the series against methicillin sensitive S. aureus (MSSA). Five of the analogues tested were found to inhibit MSSA at 32 μg/mL or less (Table 2). The best inhibitor was 1f with MIC of 0.5 μg/mL against MSSA and MRSA (Table 2, entry 6, and Table 3, entry 1). The activity was comparable to vancomycin, the drug of choice in the clinic for multidrug resistant MRSA infections. This compound was identified as the lead for further studies.

Table 2. Calculated Partition Coefficients (ClogP) and Minimum Inhibitory Concentration (MIC) against Methicillin-Sensitive S. aureus (MSSA) of 1a–1n.

entry	compd	ClogPa	MIC (μg/mL)b
1	1a	2.63	1.0
2	1b	3.16	>32
3	1c	3.69	>32
4	1d	4.66	>32
5	1e	3.40	>32
6	1f	2.73	0.5
7	1g	2.82	>32
8	1h	4.40	>32
9	1i	5.28	>32
10	1j	5.11	>32
11	1k	4.32	32
12	1l	4.14	8.0
13	1m	5.28	>32
14	1n	3.70	8.0
15	tobramycin		1
16	fosfomycin		8
17	vancomycin		0.5

^aCalculated using Chembiodraw Ultra 13.0.

^bMinimum inhibitory concentration (MIC) is defined as the lowest concentration required to inhibit visible bacterial growth; MSSA = MSSA 29213.

Table 3. MICs of 1f against Methicillin-Resistant S. aureus (MRSA) Strains.

entry	strain	MIC (μg/mL)	MIC (μM)	MIC of vancomycin (μg/mL)
1	MRSA 33591	0.5	1.6	0.5–2
2	MRSA 7419	0.12	0.38	1.0
3	B19506	0.06	0.19	0.5–2
4	MRSA K-1	0.25	0.80	1.0
5	MRSA 7425	0.12	0.38	0.12
6	MRSA 7386	1.0	3.19	1.0
7	E9902	0.5	1.60	0.5
8	E151	0.5	1.60	0.5
9	E288	0.5	1.60	0.5
10	B21838	0.5	1.60	0.5
11	B853	1.0	3.19	1.0
12	MRSA B0085	0.5	1.60	1.0

First, we attempted to improve the yield of 1f. A mechanism for the electrocyclic ring closure followed by hydration and oxidation was proposed (Scheme S2, Supporting Information). Aerobic oxidation presumably involved hydrogen peroxide and when the reaction mixture was supplemented with H₂O₂, an increased yield of 1f (37%, Table 1, entry 6) was observed. The overall yield of 1f could thus be improved to 24% in two steps from commercially available 2-bromo-1,4-naphthoquinone. In contrast, jadomycins are synthesized in >20 steps¹⁴ while deoxynyboquinone was prepared in seven linear steps.⁹

Cyclic voltammetry analysis revealed that 1e^– reduction potentials (E_red) of 1f was −1.00 V (Table S1, Supporting Information), which appears appropriate for metabolism by bioreductive enzymes.¹⁷ This compound was found to be unreactive with biological thiols such as glutathione (see Supporting Information, Figure S1), possibly due to steric hindrance created by the substituents on the quinone ring. A luminol-based chemiluminescence assay was used to detect O₂^–•, and we found generation of O₂^–• only in the presence of NQO1-naphthoquinone oxidoreductase (DT-Diaphorase, DT-D),¹⁸ a model bioreductive enzyme (Figure 1a).^19,20 A HPLC-based dihydroethidium (DHE) assay was used to independently assess O₂^–• production (Figure 1b).^21,22 Here, the conversion of DHE to 2-hydroxyethidium (2-OH-E⁺) is indicative of O₂^–• generation.²³ During incubation of 1f in the presence of DT-D, the formation of 2-OH-E⁺ was observed, thus, confirming the intermediacy of O₂^–• (Figure 1b). In addition, H₂O₂, the product of 1e^– transfer to O₂^–•, was measured using an Amplex Red-based fluorescence assay. Again, we found that 1f was capable of generating H₂O₂ in the presence of DT-D.²⁴

Figure 1.

(a) Time course of O₂^–• generation was estimated by a luminol-based chemiluminescence assay in the presence of DT-diaphorase. Control (Ctrl) is 1f incubated in pH 7.4 buffer. (b) Superoxide generated during incubation of 1f with DT-D was estimated using a dihydroethidium (DHE) assay. Superoxide specifically react with DHE to produce 2-hydroxyethidium (2-OH-E⁺); ethidium (E⁺) is formed by nonspecific oxidation of DHE and is indicative of a general increase in oxidative species. Ctrl is 1f in pH 7.4 buffer. (c) Hydrogen peroxide produced during incubation of 1f (2 μg/mL) with DT-D for 5 min was quantified by an Amplex Red-based fluorescence assay. (d) O₂^–• generated during incubation of S. aureus with 1f was estimated using a dihydroethidium (DHE) assay. O₂^–• specifically react with DHE to produce 2-hydroxyethidium (2-OH-E⁺); ethidium (E⁺) is formed by nonspecific oxidation of DHE and is indicative of a general increase in oxidative species. (e) Hydrogen peroxide generation during incubation of S. aureus with 1f (2 μg/mL) for 60 min as measured using an Amplex Red-based fluorescence assay. Ctrl is untreated bacteria. (f) ROS accumulation in the cell can damage DNA, leading to the upregulation of the DNA repair enzyme, RecA. Quantitative real time PCR analysis of recA expression in MSSA cells incubated with 6.7 × MIC concentrations, i.e., 3.35 μg/mL for 120 min showed a significant increase in recA levels. Ctrl is untreated bacteria.

Next, the possibility of 1f generating intracellular O₂^–• in bacteria was examined using a DHE assay.^21,22 The bacterial control showed unreacted DHE (Figure 1d), and in the presence of 1f, the formation of 2-OH-E⁺ in a concentration-dependent manner was observed, suggestive of O₂^–• production intracellularly (Figure 1d). Superoxide accumulation intracellularly leads to the production of H₂O₂ through dismutation.⁶ H₂O₂ diffuses out of the cell and can be measured using Amplex Red. In this assay, extracellular H₂O₂ serves as a surrogate marker for intracellular ROS levels.²⁵ When S. aureus cells are exposed to 1f, we find increased H₂O₂ confirming that 1f was capable of enhancing ROS (Figure 1e). The yield of H₂O₂ produced by 1f was nearly quantitative, a testament to efficient ROS generation by this compound (Figure 1e).

A possible consequence of increased O₂^–• and H₂O₂ is the generation of ^•OH. However, because of its extremely short half-life, detection of ^•OH is challenging. Instead, reported methods measure all elevated oxidative species including ^•OH. Using dichlorofluorescein-diacetate (DCFH₂-DA) fluorescence assay, the levels of oxidative species generated intracellularly in S. aureus was estimated. The results of this assay indicate that 1f at 0.5 μg/mL was capable of generating oxidative species intracellularly (Figure S5, Supporting Information).

Lethality induced by elevated ROS is attributable to DNA damage by ^•OH leading to single and double strand breaks.⁶ In response to DNA damage, DNA rescue pathways are activated.⁶ Symptomatic of one such rescue pathway is the expression of RecA, a repair protein crucial to homologous recombination.²⁶ In the presence of 1f, we found a significant upregulation of RecA expression suggesting that DNA damage repair response is activated in S. aureus upon treatment with 1f (Figure 1f). We next examined the effect of 1f on the viable colony count of MSSA in the presence of thiourea, a ^•OH quencher. The growth inhibition of 1f is reduced in the presence of thiourea, suggesting that quenching the ROS in the cell reduces the antibacterial effect of the compound (Figure S6, Supporting Information). These data provide further support for a ROS based mechanism contributing to antibacterial activity.

To test for synergy between 1f and other known antibiotics, we conducted synergy time kill assays with ciprofloxacin (Figure S7, Supporting Information) and tobramycin (Figure S8, Supporting Information) against MRSA 33591. We found no evidence for synergy suggesting that the mechanism of action of 1f was distinct from these clinically-used antibiotics. The lack of synergy shows that mechanism of action of 1f is independent of the antibiotics tested and could represent a novel way to inhibit bacterial growth. If the mechanism is novel, we hypothesized that it would be able to overcome existing resistance. Hence we tested the activity of the compound against a panel of 11 clinical isolates of MRSA. We found the activity to be well-conserved between the reference strain MRSA 33591 and the clinical isolates (Table 3, entries 2–12). Finally, we tested the growth inhibitory potential of 1f against mammalian cells. When screened against A549 human lung carcinoma cells, the compound showed a 50% growth inhibitory concentration (GI₅₀) of 8.6 μM (Supporting Information). The selectivity index (SI) (GI₅₀/MIC) for MRSA 33951 is >5, which is favorable for further development.

Thus, taken together, our investigations reveal 1f as a potent MRSA inhibitor with a unique mechanism of action that involves enhancement of ROS levels in cells. Several redox-active analogues of 1f (Table S1, Supporting Information) were capable of undergoing bioreduction to generate superoxide in buffer (Figure S3, Supporting Information) and increase hydrogen peroxide levels in MRSA (Figure S4, Supporting Information) but were poor S. aureus inhibitors (Table 2).^{3,4,25,27−29} Hence, generation of ROS by these redox-active small molecules appears necessary but not sufficient for inhibiting MRSA growth (Figure S4, Supporting Information).^3,4 It is noteworthy that even small structural modifications to Jadomycins resulted in significant differences in DNA damaging capability,³⁰ and SCH538415 (Chart 1), the structural analogue with an additional N-methyl group had a 10-fold lower potency in comparison with deoxynyboquinone.⁹

In summary, we report a natural product-inspired redox-active small molecule that is able to overcome drug resistance in MRSA. The in vitro potency of this compound is comparable or better than that of vancomycin, the drug of last resort for such infections.

Glossary

Abbreviations Used

MIC

minimum inhibitory concentration

GI₅₀

50% growth inhibitory concentration

ROS

reactive oxygen species

SI

selectivity index

DHE

dihydroethidium

DT-D

DT-diaphorase

Supporting Information Available

Preparative procedures, assay protocols, NMR spectra, and other experimental data. This material is available free of charge via the Internet at http://pubs.acs.org.

The authors thank IISER Pune and the Department of Biotechnology, India (BT/PR6798/MED/29/636/2012) for financial support. V.S.K. and M.S.C. acknowledge fellowships from Council for Scientific and Industrial Research (CSIR) and Department of Science and Technology (DST), respectively.

The authors declare no competing financial interest.