Synthesis and Biological Evaluation of Solithromycin Analogs Against Multidrug Resistant Pathogens

Abstract

Novel antibacterial drugs that treat multidrug resistant pathogens are in high demand. We have synthesized analogs of solithromycin using Cu(I)-mediated click chemistry. Evaluation of the analogs using Minimum Inhibitory Concentration (MIC) assays against resistant Staphylococcus aureus, Escherichia coli, and multidrug resistant pathogens Enterococcus faecium and Acinetobacter baumannii showed they possess potencies similar to those of solithromycin, thus demonstrating their potential as future therapeutics to combat the existential threat of multidrug resistant pathogens.

Graphical Abstract

The onset of multidrug resistant (MDR) bacterial infections in both nosocomial and community-acquired settings is a significant public health threat.¹ The Infectious Diseases Society of America has reported that most infections are caused by so-called “ESKAPE” pathogens consisting of Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumonia, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp.² Macrolide antibiotics have proven to be one of the safest and most effective drug classes in medicine but resistance has undermined efficacy. Therefore, the need for novel antibiotics to combat MDR bacteria is a matter of utmost significance.^3,4

Macrolide antibiotics are potent against Gram-positive bacteria, such as Staphylococci and Streptococci, in which the mode of action involves protein synthesis inhibition via binding reversibly to the bacterial ribosome.^5,6 The susceptibility of macrolides toward Gram-negative bacteria, however, is limited and includes Neisseria gonorrhoeae and Haemophilus influenzae.⁷

A timeline of macrolide antibiotic development spanning four generations is shown in Figure 1. Resistance to 1^st generation erythromycin (ERY, 1)⁸ and 2^nd generation macrolide antibiotic clarithromycin (CLA, 2)⁹ and azithromycin (AZY) prompted the discovery of the 3^rd generation ketolides telithromycin (KET, 3)¹⁰ and cethromycin (CET).^7,11 Key innovation elements of the ketolide class included (1) the replacement of the cladinose deoxysugar at C3 on the erythromycin framework with a keto (hence the term ‘ketolide’) and (2) the introduction of a pendant biaryl side-chain from the N11 oxazolidinone moiety, which makes additional π–π contacts with the ribosome. Notably, other approaches relying on Pd-catalyzed glycosylation have been reported for installing modified sugars on the methymycin aglycone in the quest for enhancing antibacterial properties.¹² The discovery of the 4^th generation (i.e., fluoroketolides), represented by solithromycin (SOL, 1),^13–15 was enabled by Cu(I)-catalyzed combinatorial click chemistry.^16–18 However, the development of SOL has been hampered by issues related to liver toxicity, as reported by the FDA in Fall 2016.¹⁹ This finding underscores the urgent need for novel potent and bioactive analogs that display satisfactory safety profiles.

Figure 1.

Structures of Four Generations of Macrolide Antibiotics and Timeline of Discovery.

In 2016, we demonstrated that ribosomes and 50S ribosomal subunits from E. coli were able to template the Huisgen [3+2] dipolar cycloaddition reaction of macrolidetethered azide 5 and 3-ethynylaniline (6) to synthesize solithromycin (4) and several triazole congener (i.e., in situ click chemistry).²⁰ A library of aliphatic, aromatic and heteroaromatic terminal alkynes was employed to probe the extent to which the ketolide side-chain engaged in π-stacking with the ribosomes or the 50S subunit. In 2018, we reported this method could be translated in cellulo, which circumvents the need to isolate bacterial ribosomes.²¹ Herein, we report the Cu(I)-catalyzed synthesis of solithromycin analogs 16–24 (Scheme 1A), several of which were not prepared during the in situ or in cellulo campaigns.

Scheme 1.

(A) Cu-catalyzed click coupling of ketolide azide 5 and terminal alkynes 6–15 to form 1,4-triazoles 4 (SOL) and 16–24. Structures of (B) aromatic and (C) heteroaromatic terminal alkynes in the library with corresponding cycloadducts.

The library of terminal alkynes featured both aromatic 6–12 (Scheme 1B) and heteroaromatic 13–15 (Scheme 1C) members. The rationale for the above nine side-chain analogs of SOL (4) was to probe the role of H-bonding and π-stacking of the aromatic nucleus within the ribosome binding site and attendant antibacterial activity as measured by MIC. For example, acetamide 7 is expected to enhance H-bonding donation of analog 16 while also decreasing the electron-density of the benzene ring. Analog 17 is a regioisomer of SOL (4) wherein the amino group is in the 4- as opposed to 3-position. The 4-methoxy group in analog 18 possesses diminished π-density vis-à-vis 4 and can only serve as an H-bond acceptor. Analog 19, derived from O-propargyl phenol (10), extends the aromatic nucleus with an ethereal oxygen. Nitroarene analogs 20 and 21 were chosen to test the consequences of markedly decreased electron density in the π-stacking event. As thiophene is a bioisostere for benzene,²² we employed two regioisomers of this heterocycle in analogs 22 and 23 (Scheme 1C). Finally, the use of 2-aminothiazole analog 24 was inspired by the aniline residue in SOL (4).

Solithromycin (4) and analogs 16–24 were synthesized using the “aqueous ascorbate” procedure of Fokin and Sharpless, which uses catalytic CuSO₄ and sodium ascorbate as stoichiometric reductant in tert-BuOH/water (1:1) at room temperature for 24 h (Scheme 1A).¹⁷ Isolated yields of 1,4-triazoles 4 and 16–24 ranged from 70–90% (see supporting information).

To evaluate the potency and scope of 4 (SOL) and analogs 16–24, minimum inhibitory concentration (MIC) assays were performed against Staphylococcus aureus (4 strains) and Escherichia coli (5 strains).²³ The results are shown in Table 1.

Table 1.

MIC (μg/mL) values for SOL (4) and analogs 16–24 against S. aureus (wt), resistant S. aureus (UCN14, UCN17, UCN18), and resistant E. coli bacterial strains (DK, DK3535, DK2058G, SQ171, SQ1712058G).

Results of the MIC evaluation with wild type and resistant S. aureus strains revealed that acetamide analog 16 was four- to eight-fold more potent than SOL against wild type; however, it was not effective against resistant UCN strains (Table 1).²⁴ Regioisomeric SOL analog 17, with its aromatic amino group in the 4- as opposed to 3-position, was equipotent with solithromycin (4) but like 16 did not inhibit the resistant strains. The remaining analogs 18–22 were all equipotent or more potent than SOL against the wild type strain; moreover, analog 20 was equipotent with SOL against the UCN17 resistant strain. On balance, the modifications to the benzene ring were not particularly effective in overcoming resistant S. aureus strains, indicating a need for alternative tactics.

Results of the MIC evaluation with E. coli strains were more promising (Table 1). The 4-amino regioisomer of SOL, analog 17, was two-fold more potent than SOL against DK, DK3535, and SQ171 strains. Nitrophenol analog 20 was four-fold more potent than SOL against wild type DK strain and equipotent with SOL against resistant DK2058G strain. Analog 21 was two-fold more potent than SOL against the SQ171 strain and, significantly, four-fold more potent than SOL against the resistant SQ171(2058G) strain. Heterocyclic analogs 22–24 were largely either more potent than or equipotent with SOL against the tested strains, demonstrating this particular motif is promising and should be investigated further.

In addition to screening our library of SOL analogs for susceptibility against S. aureus and E. coli, Gram-positive Enterococcus faecium and Gram-negative Acinetobacter baumannii were also targeted as these are ESKAPE pathogens (vide supra). E. faecium accounts for enterococcal infections such as urinary tract infections, endocarditis, intra-abdominal, and bacteremia.^25,26 There is a strong demand for novel antibacterial agents against E. faecium,²⁶ particularly as MDR strains have emerged that thwart treatment by vancomycin, penicillin, and aminoglycoside antibiotics.^27,28

Despite the fact that macrolides in general are less active against Gram negatives, screening against Acinetobacter baumannii revealed some promising results (Table 2). Curiously, both thiophene analogs 22 and 23 were four-fold and two-fold more potent, respectively, than SOL (4). The remaining triazoles were either equipotent or two-fold less potent than 4. A. baumannii is a fast-evolving, opportunistic pathogen that thrives in a hospital setting leading to urinary tract infections, meningitis, and wound infections.²⁵ Developing antibacterials that treat A. baumannii is challenging due to the diverse repertoire of resistance mechanisms it has acquired, which include drug-modifying enzymes against β-lactams and aminoglycosides, efflux mechanisms and modifications to outer membrane proteins.²⁹

Table 2.

MIC (μg/mL) values for SOL (4) and analogs 16–24 against multidrug resistant E. faecium (NCTC 7171) and A. baumannii (ATCC 19606) bacterial strains.

Compound	A. baumannii (ATCC 19606)	E. faecium (NCTC 7171)
SOL (4)	25	0.024
16	50	0.024
17	25	0.024
18	25	0.0015
19	25	0.012
20	25	0.0000059
21	50	0.0015
22	6.5	0.049
23	12	0.000094
24	>50	0.024

The results of MIC evaluation of SOL and analogs 16–24 against E. faecium are shown in Table 2. While analogs 16, 17, and 24 were equipotent with SOL (4), we found that analog 19 was two-fold more potent, analogs 18 and 21 were sixteen-fold more potent, analog 23 was 256-fold more potent, and significantly, analog 20 was 4096-fold more potent SOL (4).

The extremely low MICs for analogs 20 and 23 against E. faecium encouraged us to investigate the biological activity of other known macrolides such as ERY (1) and AZY. Results are shown in Table 3. MIC values for both ERY (1) and AZY were 12 and 25 μg/mL, respectively, which are in fact high compared to analogs 20, 23, and even SOL (4). This emphasized the need for investigating analogs 20 and 23 in downstream assays to better understand their value as drug discovery entities. We, thus, began by assessing the mammalian cell cytotoxicity with Vero cells (African green monkey kidney epithelial cell line). The CC₅₀ was quantified as the minimum concentration of compound resulting in 50% growth inhibition of this model cell line.³⁰ Accordingly, the CC₅₀ values of 20 and 23 were 12 and 25 μg/mL, respectively. The selectivity index (SI = CC₅₀/MIC) for 20 and 23 was then calculated for both to be in significant excess of 10, which is a minimum value for demonstrating a reasonable toxicity window in early drug discovery compounds.³¹

Table 3.

MIC (μg/mL), CC₅₀ (μg/mL), and SI values for analogs 20 and 23 against multidrug resistant E. faecium (NCTC 7171) in comparison to ERY (1) and AZY.

Compound	E. faecium (NCTC 7171)	Vero cell CC50	SI = CC50/MIC
ERY (1)	12	ND	ND
AZY	25	ND	ND
20	0.0000059	12	2.0 × 106
23	0.000094	25	2.7 × 105

ND = Not Determined

In summary, we have synthesized solithromycin (4) and nine analogs thereof 16–24 using Cu(I)-catalyzed click chemistry and evaluated all ten compounds against wild type and resistant isolates of Gram-positive S. aureus, E. coli, and E. faecium, in addition to Gram-negative A. baumannii. While many compounds were equipotent with 4, we found several analogs that displayed enhanced activity against E. coli, E. faecium, and A. baumannii strains that justify in vivo studies to develop novel antibacterials that can address the existential threat of multidrug resistant pathogens.

Abbreviations

MDR

multidrug resistance

MIC

Minimum Inhibitory Concentration

ERY

erythromycin

CLA

clarithromycin

AZY

azithromycin

TEL

telithromycin

CET

cethromycin

SOL

solithromycin

CC₅₀

Cytotoxicity Concentration 50%

SI

Selectivity Index