Nitro Sulfonyl Fluorides are a new pharmacophore for the development of antibiotics

Abstract

The development of antibiotics against Gram-negative bacteria is a central problem in drug discovery. In this report, we demonstrate that aromatic sulfonyl fluorides with a nitro group in their ortho position have remarkable antibacterial activity and are active against drug-resistant pathogens, such as methicillin-resistant Staphylococcus aureus (MRSA), multidrug resistant Acinetobacter baumannii, and Pseudomonas aeruginosa.

Infections caused by drug-resistant bacteria are a severe threat to global public health and will cause a public health crisis if new antibiotics are not developed.^1–3 For example, in 2017, the Centers for Disease Control and Prevention (CDC) estimated that more than 2 million people in the world were infected with bacteria that were resistant to antibiotics.⁴ Despite significant efforts, no new classes of antibiotics have been clinically approved during the last 25 years. Developing new antibiotics has been challenging because of the low permeability of Gram-negative bacteria (GNB) to antibiotics,⁵ which dramatically restricts the types of molecules that can be considered for future antibiotic development. Therefore, there is a great need for the development of novel pharmacophores that are both permeable to GNBs and have efficacious antibacterial activity.

Covalent irreversible inhibitors have tremendous potential as antibiotics because of their long residence time and ability to kill dormant bacteria.^6,7 For example, the most effective class of antibiotics, the beta lactams, are covalent inhibitors,⁸ and represent 65% of the world’s antibiotic usage. However, existing covalent inhibitors such as beta lactams, were identified from natural products, and developing new covalent inhibitors, de-novo,⁹ has been challenging due to a lack of suitable electrophiles as starting fragments for antibiotic discovery.

Sulfonyl fluorides (SFs) have great potential as electrophiles¹⁰ for future antibiotic development because they have low reactivity toward nucleophiles and low rates of hydrolysis, and because their electrophilicity is dramatically increased in the presence of local hydrogen bonding.¹¹ This allows SFs to react selectively with protein targets that provide the proper localized hydrogen-bonding environment. SFs, therefore, have the potential to be highly selective for their target and cause low levels of off-target damage and toxicity. In addition, SFs are small in size and should be permeable to cells with the potential to be active against Gram-negative bacteria. Despite their promise as irreversible antibiotics, the ability of sulfonyl fluorides to act as scaffolds for future antibiotic development has never been investigated.

Herein, we screened the antibacterial activity of a library of SF-containing compounds against E. coli to determine if SFs had any intrinsic antibacterial activity. The library contained a variety of chemical structures including aliphatic, aromatic, and heterocyclic SFs, such as thiophene and imidazole SFs, as well as Michael acceptors, such as ethenesulfonyl fluoride (ESF). We also included two known serine protease inhibitors, 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF)¹² and phenylmethylsulfonyl fluoride (PMSF)¹³. All compounds were tested against a Kanamycin-resistant strain of E. coli (plasmid transfected BL21) at an initial concentration of 100 μM. From this screen, 2-nitrobenzenesulfonyl fluoride (1), emerged as the most potent antibacterial compound (Figure 1), and had an MIC₉₉ value of 5.15 μg/mL (see Figure S1 for detailed structures of the initial library).¹⁴ The MIC₉₉ of compound 1 is fairly potent. For example, the FDA-approved nitroaryl-containing drugs metronidazole and ornidazole¹⁵ have MICs against E. coli between 16 – 28 μg/mL and 8 – 128 μg/mL, respectively. We also investigated the mammalian cell toxicity of compound 1 against RAW macrophage cells to determine its selectivity for bacteria. Compound 1 had an IC₅₀ value of 290 μM (Figure S2), suggesting that compound 1 has potential as a future antibiotic.¹⁶ However, there is a possibility that compound 1 will alkylate proteins in vivo and generate nitro aromatic based antigens, which would cause an immune response (see reference for a discussion on the toxicity of nitroaromatic-containing drugs).¹⁷

Figure 1. Ortho-nitro sulfonyl fluorides have antibacterial activity.

A library screen of sulfonyl fluorides (SFs) against E.coli identified 2-nitrobenzenesulfonyl fluoride (1) as a lead fragment with antibacterial activity. Further chemical optimization led to the development of compound 2, which had an MIC of 0.66 μg/mL against E.coli, and was active against multidrug resistant Acinetobacter baumannii (MRSA), and Pseudomonas aeruginosa. Ortho-nitro sulfonyl fluorides are a new class of pharmacophores with great potential for drug discovery, given their low molecular weights and high efficacy.

We performed chemical modifications on the hit compound 1 to determine the structural requirements needed for antibacterial activity, and to determine if we could improve its antimicrobial activity (Figure 2). We first investigated if the nitro group in compound 1 had to be ortho to the sulfonyl fluoride, and if the electrophile had to be a sulfonyl fluoride, as opposed to other electrophiles such as an epoxide. Sulfonyl fluoride compounds that contained a nitro group in the meta- and para-position were therefore tested for their antibacterial activity. In addition, compounds containing sulfonyl chlorides and epoxides as the electrophile with a nitro group in the ortho position were also examined. Figure 2 demonstrates that sulfonyl fluorides with a nitro group either in the meta (4) or para (5) position had no antibacterial activity against E. coli. In addition, Changing the electrophile to either a sulfonyl chloride or an epoxide also destroyed antibacterial activity. From the SAR study, we therefore conclude that the nitro group has to be ortho-positioned to the sulfonyl fluoride for the NSFs to be active, and that the electrophile has to be a sulfonyl fluoride.

Figure 2. The nitro group and the sulfonyl fluoride group in compound 1 need to be in the ortho configuration to have antibacterial activity, and the electrophile needs to be a sulfonyl fluoride.

Functional group modifications of compound 1 revealed that the ortho-nitrosulfonyl fluoride geometry is required for antibacterial activity (μg/mL). Compounds 4–5 and 12 are inactive and lack the ortho configuration. In addition, Compounds 10 and 14, have electrophiles composed of epoxides and sulfonyl chlorides and are inactive.

We measured the water hydrolysis rate of 1 to determine its stability in aqueous environments. Compound 1 is remarkably stable in neutral pH aqueous environments, and showed no signs of hydrolysis after three weeks at pH 7.4. In contrast, compound 10, which represents the chloride variant, hydrolyzed orders of magnitude faster, and had a t_1/2=33 h at pH 7.3.

We performed additional studies to determine the tolerance of compound 1 towards electron withdrawing and donating groups. We synthesized 9 new ortho-nitrobenzenesulfonyl fluoride derivatives¹⁷ and screened them against E. coli (Figure 3). Among these molecules, 5 compounds which contained para-substituents to either the sulfonyl fluoride or the nitro group were identified as inhibiting E. coli growth within the range of 5 – 28 μg/ mL concentration. However, no compounds with ortho-substituents to either the sulfonyl fluoride or nitro group showed antibacterial activity. The compound bearing a 4-methyl group (compound 18) was the most potent, exhibiting MIC values of 2.74 μg/mL and 5.45 μg/mL with Kanamycin and Ampicillin-resistant E. coli strains, respectively. Additionally, there was no potency difference between a chloride or a bromide group (compounds 19-22), and para substituents to the sulfonyl fluoride (compounds 18, 19, and 22) had better activity than para substituents to a nitro group (compounds 20 and 21). Sulfonyl fluoride para-substitutions with strong electron withdrawing groups might serve to enhance the electrophilicity of the sulfonyl fluoride. Interestingly, a nitro para-substituent (compound 28, see Figure S3) showed no cytotoxic activity in the MTT assay, although this electrophilic compound and similar electron withdrawing substitutions could harbor undesired off-target reactivity in vivo due to increased broad protein reactivity.

Figure 3. 2-Nitrobenzenesulfonyl fluoride derivatives and their MICs (μg/mL).

Functional group modifications of lead hit 1 were explored to further understand the electronic and steric requirements which are necessary for antibacterial activity. Compound 1 tolerates halogen modifications, however most modifications resulted in lower activity.

The mechanism by which compound 1 kills bacteria is unknown. We performed random barcode transposon-site sequencing (RB-TnSeq)¹⁹ on bacteria that had been treated with compound 1, using the E. coli strain BW25113, to identify bacterial genes that either protected or sensitized E.coli towards compound 1. The results of our screen with the E. coli BW25113 library are shown in Table 1, with the top 5 genes for both positive and negative fitness listed. Table 1 demonstrates that the inactivation of nitroreductase A confers protection against compound 1, which is similar to other nitro-aromatic drugs such as metronidazole and nitrofurantoin, and suggests that the nitro group in compound 1 is being reduced by nitroreductase A to a toxic intermediate.²⁰

Table 1. Compound 1 is activated by nitroreductase A in bacteria.

Compound 1 was incubated with the E. coli BW25113 library and the deletion of genes that either conferred resistance or sensitization to compound 1 were identified. Positive values indicate that gene deletion confers resistance, and negative values indicate that gene deletion causes sensitization to compound 1. Nitroreductase deletion strongly protects E.coli against compound 1.

Gene	Name	Description	Value
b0851	nfsA	Nitroreductase A, NADPH-dependent, FMN-dependent	6.26
b1531	marA	Multiple antibiotic resistance	4.37
b0464	acrR	DNA-binding transcriptional repressor	3.53
b0912	ihgB	Integration host factor subunit beta	3.05
b1217	chaB	Cation transport regulator	3.03
Gene	Name	Description	Value
b3961	oxyR	DNA-binding transcriptional dual regulator	− 4.66
b0463	acrA	multidrug efflux system	− 4.26
b0462	acrB	multidrug efflux system protein	− 4.20
b3742	mioC	flavodoxin	− 4.09
b3169	nusA	transcription elongation factor NusA	− 3.99

Compound 1 has two potential mechanisms by which it can kill bacteria (Figure 4). It either reacts with an essential cellular protein via the sulfonyl fluoride moiety to generate intermediate 40, and then becomes reduced via nitroreductases, generating reactive intermediates that then destroy the modified protein or neighboring biomolecules (Figure 4, Route A). Or, compound 1 is reduced by nitroreductases first, to generate a hydroxylamine intermediate (15), which attacks the neighboring sulfur atom to generate the oxathiazole intermediate 16 (Figure 4, Route B). The oxathiazole intermediate 16 subsequently reacts with essential bacterial proteins causing cell death. In support of the 2^nd hypothesis is the observation that the nitro group and the sulfonyl fluoride must be ortho to each other for antibacterial activity. In addition, compound 6, which contains an amine ortho to the sulfonyl fluoride was inactive. Compound 6 represents the final reduction product of compound 1, and demonstrates that the active intermediate of 1 is not the fully reduced compound. Attempts were made to synthesize the hydroxyl amine intermediate 16, via reduction of 1 with hydrogen gas, but were unsuccessful, and the reduction reactions always generated the amine (compound 6).²¹

Figure 4. Potential mechanism by which nitroreductase A activates compound 1.

Route A: Compound 1 can react with target proteins to generate 40, and then undergoes reduction to generate reactive intermediates that destroy the protein or neighbouring biomolecules. Route B: Compound 1 is reduced by nitroreductase A to form a hydroxyl amine 15, which can attack the neighbouring sulfur and generate the five-membered oxathiazole intermediate 16. Compound 16 can then react with essential bacterial proteins to kill bacteria.

Figures 2–3 and Figure S3 demonstrate that our attempts at improving the MIC of our lead compound 1, using a benzene core, were unsuccessful. We therefore investigated if replacing the benzene core with a hetero-aromatic core would lead to enhanced activity. The rationale for using a hetero-aromatic core was based on the success of hetero-aromatic nitro-containing drugs, such as nitrofurantoin, which appear to have a similar mechanism of activity as compound 1. We therefore synthesized¹⁸ the heteroatom-containing aromatic sulfonyl fluorides, 5-chloro-4-nitrothiophene-2-sulfonyl fluoride 32, 5-methyl-2-nitrothiophene-3-sulfonyl fluoride 33, 5-methyl-4-nitrothiophene-2-sulfonyl fluoride 34, and 2-nitrothiophene-3-sulfonyl fluoride 2, and tested their antibacterial activity (Structures shown in Figure 5). Of these new compounds, we discovered compound 33 had better activity against Kanamycin-resistant E. coli (2.0 μg/mL) compared to compound 1, and we identified 2-nitrothiophene-3-sulfonyl fluoride (compound 2) as a new lead hit with superior performance compared to all previous NSF derivatives. Compound 2 had an MIC value of 0.66 μg/mL and 10.6 μg/mL against Kanamycin and Ampicillin-resistant E. coli strains, respectively (Figure 5). The synthesis of compound 2 was straightforward and readily accessible from the commercially available sulfonyl chloride derivative, via treatment with KFHF in a 1:1 mixture of acetone and water (Scheme S1).

Figure 5. 2-nitrothiophene-3-sulfonyl fluoride (2) is more effective than 2-nitrobenzenesulfonyl fluoride (1) at killing bacteria.

Compound 2 has an MIC (μg/mL) against E. coli that is 10 times lower than compound 1, and is a new sulfonyl fluoride-based pharmacophore for future antibiotic development. Additional modifications to compound 2 lowered its antibacterial activity (compounds 32–34).

We determined the MIC of compound 2 against a panel of drug-resistant bacteria, such as drug-resistant A. baumannii, P. aeruginosa, MRSA, and compared its MIC against nitrofurantoin. These pathogens were selected for investigation because of the significant health risk they pose. The MICs of compound 2 and nitrofurantoin against our panel of bacteria are summarized in Table 2. Compound 2 had an MIC of 21 and 42 μg/mL against two different A. baumannii clinical isolates, which were resistant to almost all antibiotics currently used, including colistin (a last-resort antibiotic). Nitrofurantoin had no activity against these pathogens. Compound 2 also had a better antibacterial activity against methicillin-resistant Staphylococcus aureus (MRSA) than nitrofurantoin. Moreover, compound 2 had better efficacy against both P. aeruginosa strains tested than nitrofurantoin, and had an MIC value of 42 μg/mL, against one of these strains, whereas nitrofurantoin had no activity up to 95 μg/mL. Compound 2 was also active against a standard E. coli strain (ATCC 25922) as well as a drug resistant E. coli clinical isolate obtained from San Francisco General Hospital (SFGH; Extended-Spectrum Beta Lactamase-producer).

Table 2. Compound 2 is active against drug resistant bacteria.

The MIC (μg/mL) values of compound 2 against a panel of clinically isolated bacterial strains was determined and compared against nitrofurantoin and colistin. Compound 2 had better antibacterial efficacy against A. Baumannii 605A than either colistin or nitrofurantoin.

Treatment\Strain	2	Nitrofurantoin Colistin
E. coli KAN-RES	0.66	0.37	0.5
E. coli AMP-RES	10.6	6.0	RES
E. coli SFGH 207	42	6.0	0.5
E. coli ATCC 25922	1.32	3.0	2.0
MRSA	5.28	12	ND
A. baumannii 605 A	21	> 95	4.0
A. baumannii 597 A	42	> 95	256
P. aeruginosa SFGH 266	42	> 95	ND
P. aeruginosa SFGH 427	84	> 95	ND

In conclusion, we present a new electrophile fragment for antibacterial drug discovery based on the nitrosulfonyl fluoride (NSF) fragment. Compound 2, which contains the NSF fragment, was active against MRSA and a variety of drug resistant pathogens, and is also exceptionally atom economical with molecular weights between 200 – 250 Daltons. Compound 2 and the NSF fragment therefore have tremendous potential for further optimization and future applications in drug discovery.