Augmenting the Activity of Macrolide Adjuvants against Acinetobacter baumannii

Abstract

Approximately 1.7 million Americans develop hospital associated infections each year, resulting in more than 98,000 deaths. One of the main contributors to such infections is the Gram-negative pathogen Acinetobacter baumannii. Recently, it was reported that aryl 2-aminoimidazole (2-AI) compounds potentiate macrolide antibiotics against a highly virulent strain of A. baumannii, AB5075. The two lead compounds in that report increased clarithromycin (CLR) potency against AB5075 by 16-fold, lowering the minimum inhibitory concentration (MIC) from 32 to 2 μg/mL at a concentration of 10 μM. Herein, we report a structure–activity relationship study of a panel of derivatives structurally inspired by the previously reported aryl 2-AI leads. Substitutions around the core phenyl ring yielded a lead that potentiates clarithromycin by 64- and 32-fold against AB5075 at 10 and 7.5 μM, exceeding the dose response of the original lead. Additional probing of the amide linker led to the discovery of two urea containing adjuvants that suppressed clarithromycin resistance in AB5075 by 64- and 128-fold at 7.5 μM. Finally, the originally reported adjuvant was tested for its ability to suppress the evolution of resistance to clarithromycin over the course of nine consecutive days. At 30 μM, the parent compound reduced the CLR MIC from 512 to 2 μg/mL, demonstrating that the original lead remained active against a more CLR resistant strain of AB5075.

The prevalence of infections that stem from multidrug resistant (MDR) bacteria has significantly increased in recent years, which, in addition to a reduction in the involvement of pharmaceutical companies in antibiotic development, has severely limited currently available treatment options.¹ Resistance toward every class of clinically prescribed antibiotics has now been observed.² The Centers for Disease Control and Prevention (CDC) estimates that over 2.8 million people are infected with antibiotic-resistant bacteria each year, and approximately 35,000 deaths result from these bacterial infections.³ A high number of these deaths are associated with infections caused by the ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobactor baumannii, Pseudomonas aeruginosa, and Enterobacter species), where the “ES” represents Gram-positive species and the “KAPE” pathogens represent Gram-negative species.⁴ Gram-negative pathogens are of particular concern because their outer membrane (OM) is impermeable to many antibiotics, further limiting the available antibiotics for treating MDR Gram-negative infections.²

Despite the target being present in both Gram-positive and Gram-negative bacteria, macrolides are generally utilized for the treatment of Gram-positive infections due to low penetration of the Gram-negative OM. It has been shown, however, that if the OM is compromised then macrolides can become efficacious against Gram-negative bacteria. For example, A. baumannii strains that lack lipopolysaccharide (LPS) are highly permeable and are susceptible to treatment with the macrolide antibiotic azithromycin.⁵ Compounds that are able to physically disrupt the outer membrane, such as pentamidine and polymyxin derivatives, are also able to sensitize Gram-negative bacteria to macrolides.⁶

Recently, we reported two small molecules that potentiate the activity of clarithromycin (CLR) (compounds 1 and 2, Figure 1) against a strain of A. baumannii (AB5075).⁷ These compounds, which we and others refer to as antibiotic adjuvants, are relatively nontoxic to AB5075 by themselves (MICs of 100 μM); however at 30 μM, each compound reduced the minimum inhibitory concentration (MIC) of CLR by 128-fold, from 32 μg/mL to 0.25 μg/mL, while at 10 μM they reduced the CLR MIC 16-fold to 2 μg/mL. Investigations into the mechanism of action of these molecules revealed that, unlike other small molecules that compromise the outer membrane through physical disruption, these compounds antagonized colistin and altered Lipid A composition in AB5075. This suggests that alterations in LPS presentation and/or biosynthesis may underpin these adjuvants’ mode of action (MoA).⁷ While research is ongoing to determine the target for these molecules, we herein report a structure–activity relationship (SAR) analysis encompassing structural modifications of the reported lead compounds 1 and 2 (Figure 1) to further optimize activity in combination with macrolide antibiotics, specifically CLR which displays the greatest enhancement in activity of the three macrolide antibiotics studied (erythromycin, azithromycin, CLR) against AB5075.⁷

Figure 1.

Initial lead aryl 2-aminoimidazole adjuvants 1 and 2 and delineation of scaffold regions for SAR examination.⁷

To perform analog synthesis, we divided the lead molecules into three regions: region 1, the aryl tail (boxed green), region 2, the phenyl core (highlighted purple), and region 3, the amide linker (highlighted in blue) (Figure 1). First, we elected to modify the aryl tail of compound 1 in an attempt to delineate the importance of electron-withdrawing and/or inductive effects. Next, we introduced various functional groups on the phenyl core while maintaining the 3,5-dichloro substituent in the aryl tail. After identifying structural components in each region that augmented activity, we combined the most effective phenyl core modification with aryl tails that had shown augmented activity with CLR. Finally, a brief SAR was performed modifying the original amide linker.

Results and Discussion

We initiated our study by screening additional analogs of 1 and 2 for CLR potentiation against AB5075 (Figure 1). Seventeen of the 25 aryl 2-AI analogs had been previously synthesized and screened for potentiation of β-lactam antibiotics against Gram-negative bacteria.⁸ The other analogs were accessed through intermediate 3 by acylation with the appropriate acid chloride, TFA-mediated removal of the Boc-groups, and finally counterion exchange (Scheme 1). Once the pilot library was completed, each compound was first analyzed individually for standalone toxicity against AB5075 by determining the MIC (Table S1). Of the 25 analogs, one registered an MIC of 25 μM (4n), two returned an MIC of 50 μM (4l, 4p), four displayed an MIC of 100 μM (4o, 4r, 4s, and 4v), while the MICs of the rest of the analogs were ≥200 μM. To test for potentiation, the MIC of CLR against AB5075 was determined in the absence or presence of each compound dosed at 30% their MIC or 30 μM, whichever was lower (Table S1). Each analog was compared to the activity of compounds 1 and 2 at 30 μM.

Scheme 1. Aryl Tail Modifications of Previously Reported Adjuvants 1 and 2.

Reagents and conditions: (a) ArCOCl, K₃PO₄, THF, 25 °C, 16 h; (b) TFA, DCM, 25 °C, 3 h; (c) 6 M HCl, MeOH, 25 °C, 5 min.⁸

First, we noted that removal of the halides (4a) abolished activity (Table S1). We also observed that disubstitution patterns on the aryl tail were typically more active and achieved greater CLR potentiation when compared to monosubstituted compounds. The only monosubstituted compounds that displayed CLR potentiation at 30 μM were monochlorinated compounds (4d, 4i) and one monobrominated compound (4h).

Compounds 4k–m, which incorporated a 3,5-disubstiution pattern, allowed a comparison of the effects of replacing Cl/Br with F, Cl/Br with CF₃, and Cl for CH₃ (steric isostere). Compound 4k only exhibited an 8-fold reduction in the MIC of CLR at 30 μM, while compounds 4l and 4m also exhibited relatively modest activity, returning CLR MICs of 8 and 2 μg/mL, respectively, at 15 μM and 30 μM.

Since chlorination appeared to be a promising halogenation choice for adjuvant activity, the positioning of the dichloro substitution pattern was evaluated next. Dichlorination at the 2,3-positions, seen in compound 4q, resulted in only 4-fold potentiation of CLR at 30 μM, while compound 4r in which the chloro substituents were moved to the 3,4-position displayed a 64-fold reduction in the MIC of CLR at 30 μM. Compound 4t, which placed the chloro residues at the 2,4-position, exhibited only an 8-fold reduction in the MIC of CLR at 30 μM. Analysis of the structure–function of tri- and multihalogenated patterns on the aryl tail showed that 4r, 4s, and 4y displayed comparative CLR potentiation to the original adjuvants 1 and 2 when dosed at 30 μM.

Next, a dose response study was conducted with nine of the most active compounds (4d, 4i, 4n, 4o, 4p, 4r, 4s, 4u, 4v, and 4y) to determine CLR potentiation as a function of compound concentration (Table 1). At 10 μM, 4o, 4p, and 4r showed comparable activity to the original leads 1 and 2, returning CLR MICs of 2, 1, and 4 μg/mL, respectively. Compound 4y, however, was significantly more active maintaining a CLR MIC of 0.25 μg/mL (128-fold reduction) at 10 μM.

Table 1. Dose-Response of Active Aryl Tails for CLR Potentiation against AB5075.

Compound	Concentration Tested (μM)	CLR MIC (μg/mL)	Compound	Concentration Tested (μM)	CLR MIC (μg/mL)
		32			32
4d	20	4	4r	20	0.5
	10	16		10	4
				7.5	8
4i	20	4	4s	25	8
	10	16		20	16
				10	32
4n	5	16	4u	20	16
				10	32
4o	10	2	4v	20	8
	7.5	8		10	32
4p	10	1	4y	20	0.25
	7.5	4		10	0.25
	5	16		7.5	8

To probe the SAR of region 2, the original 3,5-dichloro aryl tail was held constant while various functional groups were introduced on the phenyl core. The synthetic approach to these analogs is outlined in Scheme 2. Anilines 7a–e were alloc-protected using allyl chloroformate in a biphasic solvent system under basic conditions at 0 °C to afford compounds 8a–e. Compounds 8a–e and 5a–b were then converted into their corresponding acid chlorides using oxalyl chloride and a catalytic amount of DMF at 0 °C. Each acid chloride was reacted with diazomethane at 0 °C for 2 h followed by addition of hydrobromic acid to generate the corresponding α-bromo-ketone. The α-bromo-ketones were then cyclized with Boc-guanidine to deliver the corresponding mono-Boc-protected 2-AI compounds 9a–e. The alloc protecting groups of 9a–e were then removed by treatment with tetrakis(triphenylphospine)palladium(0) and sodium borohydride. Each aniline was then immediately acylated using commercially available 3,5-dichlorobenzyl chloride under basic conditions, that, following Boc-deprotections using trifluoroacetic acid and subsequent conversion to the hydrochloric acid salt, afforded compounds 10a–e. Intermediates that contained a nitro group as the latent aniline were reduced using palladium on carbon and hydrogen gas. Acylation, Boc-deprotection, and ion exchange as above yielded compounds 6a–b.

Scheme 2. Phenyl Core Modification of Compounds 1 and 2.

Reagents and conditions for 2A: (a) (COCl)₂, DMF_cat, DCM, 0 °C, 2 h; (b) CH₂N₂, Et₂O, DCM, 0 °C, 1.5 h; (c) HBr, 0 °C, 30 min; (d) Boc-guanidine, THF, 56 °C, 3 h; (e) H₂, 10% Pd/C, MeOH, 25 °C, 16 h; (f) 3,5-dichlorobenzoyl chloride, K₃PO₄, THF, 0 °C, 16 h; (g) TFA, DCM, 25 °C, 3 h; (h) 6 M HCl, MeOH, 25 °C, 5 min. Reagents and conditions for 2B: (a) allyl chloroformate, NaHCO₃, DIPEA, 1,4-dioxane/H₂O, 0 °C, 16 h; (b) (COCl)₂, DMF_cat, DCM, 0 °C, 2 h; (c) CH₂N₂, Et₂O, DCM, 0 °C, 1.5 h; (d) HBr, 0 °C, 30 min; (e) Boc-guanidine, THF, 56 °C, 3 h; (f) Pd(PPh₃)₄, NaBH₄, EtOH, 25 °C, 4 h; (g) 3,5-dichlorobenzoyl chloride, K₃PO₄, THF, 0 °C, 16 h; (h) TFA, DCM, 25 °C, 3 h; (i) 6 M HCl, MeOH, 25 °C, 5 min. Difluoro modification on phenyl core of lead adjuvant 10d. Reagents and conditions for 2C: (a) 3,5-dichlorobenzoyl chloride, K₃PO₄, THF, 0 °C, 16 h; (b) 5 M NaOH, MeOH/H₂O, 25 °C, 3 h; (c) (COCl)₂, DMF_cat, DCM, 0 °C, 2 h; (d) CH₂N₂, Et₂O, DCM, 0 °C, 1.5 h; (e) HBr, 0 °C, 30 min; (f) Boc-guanidine, THF, 56 °C, 3 h; (g) TFA, DCM, 25 °C, 3 h, then 6 M HCl, MeOH, 25 °C, 5 min.

As detailed above, all compounds were first screened against AB5075 alone to determine their MICs followed by an initial screen of each compound at ≤30% of their MIC in combination with CLR (Table 2). Except for 6a, which returned a standalone MIC of 100 μM, all derivatives were essentially nontoxic showing MIC values >200 μM. When the carbon ortho to the amide linker (R₁, Scheme 2) contained either a methyl or methoxy substituent, similar levels of CLR potentiation were noted in comparison to compounds 1 and 2. Both methyl and methoxy compounds, 10a and 6a, displayed a 64-fold reduction in CLR MIC at 30 μM compared to parent 128-fold activity at 30 μM. When either Cl or F is placed at this position, activity decreased significantly, with only a 2- and 4-fold reduction in CLR MIC observed for compounds 10b and 10c, respectively, at 30 μM. We noted a substantially different trend when substituents were placed meta to the amide linker (R₂, Scheme 2). Derivatives containing a chloro (10e), fluoro (10d), or methyl (6b) potentiated CLR activity by a similar degree to compound 1, lowering the MIC of CLR by 64-fold, from 32 to 0.5 μg/mL, when dosed at 30 μM.

Table 2. CLR Potentiation Data against AB5075 Using Phenyl Core Derivatives 6a–b and 10a–e.

Compound	R1	R2	Compound MIC (μM)	Concentration Tested (μM)	CLR MIC (μg/mL)	Fold Reduction
					32
6a	OCH3	H	100	30	0.5	64
				20	4	8
				10	16	2
6b	H	CH3	>200	30	0.5	64
				20	0.5	64
				10	8	4
10a	CH3	H	>200	30	0.5	64
				20	1	32
				10	16	2
10b	Cl	H	>200	30	16	2
10c	F	H	>200	30	8	4
10d	H	F	>200	30	0.25	128
				20	0.25	128
				10	0.5	64
				7.5	4	8
				5	8	4
10e	H	Cl	>200	30	0.5	64
				20	0.5	64
				10	8	4

The five analogs (6a, 6b, 10a, 10d, and 10e) showing comparable activity to 1 were then subjected to a dose response study (Table 2). Compounds 6a, 6b, 10a, and 10e were not as active as 1 at 10 μM, effecting an MIC reduction of 2-, 4-, 2-, and 4-fold, respectively. The fluoro analog 10d, however, surpassed the activity of 1 and displayed a 64-fold reduction of the CLR MIC at 10 μM.⁷

Given the activity of the monofluoro compound (10d), the difluoro analogue 13 was synthesized to probe if either symmetry or additional electronic effects would further modulate activity (Scheme 2C). Interestingly, compound 13 was essentially inactive at 10 μM, displaying only a 2-fold reduction in the MIC of CLR. At 20 and 30 μM, compound 13 lowered the MIC of CLR to 8 and 4 μg/mL, respectively.

As the monofluoro analog 10d was significantly less toxic to AB5075 in comparison to 1 and 2, we probed whether this type of modification could attenuate the standalone toxicity of some of the initial tail derivatives we studied.⁷ To explore this, we chose some of the aryl tails that exhibited the ability to potentiate CLR yet were also more toxic, registering standalone MIC values (i.e., no antibiotic added) of ≤100 μM. We postulated that the addition of a fluoro substituent would reduce the overall toxicity of these adjuvants while either retaining or improving activity with CLR against AB5075.

The aryl tail from compound 2 in Figure 1 was chosen first for further analog synthesis and was accessed by coupling 3,5-dibromobenzoyl chloride to mono-Boc-protected 2-AI aniline 14. After TFA-mediated Boc-deprotection followed by counterion exchange, the fluorinated analog of 2 (15a) was screened for standalone toxicity and CLR potentiation (Scheme 3). Fluorination did indeed attenuate toxicity as 15a returned an MIC of >200 μM (compound 2 has a standalone MIC of 100 μM); however, it also compromised adjuvant activity. At 30 μM, compound 15a only reduced the CLR MIC 16-fold against AB5075 (Table 3), while compound 2 effected a 128-fold reduction at the same concentration.

Scheme 3. Evaluation of Active Aryl Tails in Combination with Fluorinated Phenyl Core.

Reagents and conditions: Scheme A protocol: (a) ArCOCl, K₃PO₄, THF, 25 °C, 16 h; (b) TFA, DCM, 25 °C, 3 h; (c) 6 M HCl, MeOH, 25 °C, 5 min. Scheme B protocol: (a) ArCOCl, K₃PO₄, THF, 25 °C, 16 h; (b) 5 M NaOH, MeOH/H₂O, 25 °C, 3 h; (c) (COCl)₂, DMF_cat, DCM, 0 °C, 2 h; (d) CH₂N₂, Et₂O, DCM, 0 °C, 1.5 h, followed by HBr, 0 °C, 30 min; (e) Boc-guanidine, THF, 56 °C, 3 h; (f) TFA, DCM, 25 °C, 3 h, then 6 M HCl, MeOH, 25 °C, 5 min.

Table 3. CLR Potentiation Data for Compounds 15a–b, 17a–b, and 20a–c against AB5075.

Compound	Compound MIC (μM)	Concentration Tested (μM)	CLR MIC (μg/mL)	Fold Reduction
			32
15a	>200	30	2	16
		20	8	4
		10	16	2
15b	>200	30	0.5	64
		20	0.5	64
		10	0.5	64
		7.5	1	32
		5	8	4
17a	50	15	0.25	128
		10	8	4
		7.5	16	2
17b	50	15	0.25	128
		10	2	16
		7.5	8	4
20a	50	15	0.125	256
		7.5	0.25	128
		5	4	8
20b	25	7.5	0.5	64
		5	8	4
20c	100	30	≤0.0625	≥512
		20	≤0.0625	≥512
		10	0.25	128
		7.5	2	16
		5	8	4

The 3,4-dichlorobenzoyl tail 15b showed both a reduction in the standalone toxicity and improvement in CLR potentiation over the original motif 4r. Compound 15b afforded a standalone MIC of >200 μM (vs 100 μM for 4r) and delivered a 32-fold reduction in CLR at a concentration of 7.5 μM, exceeding the activity of the parent compound 4r as well as 10d. When the activities of 17a and 17b were directly compared to their nonfluorinated variants 4n and 4o, we noted that standalone toxicity was essentially unchanged (2-fold changes are not considered to be significant in a standard MIC assay). The ability of 17a to potentiate CLR activity at 7.5 μM was severely compromised in comparison to compound 4n, showing an insignificant 2-fold reduction in the MIC of CLR compared to a 16-fold reduction at 7.5 μM for 4n. The adjuvant activity of compound 17b was identical to compound 4o (Table S1, 1, and 3).

In a previous report from our lab,⁹ we described related aryl 2-AI analogs that inhibit E. coli biofilm formation. Within that report, compound 4k was determined to be the lead biofilm inhibitor, indicating that biofilm inhibition activity was greater when the amide linker of the 3,5-difluoro aryl tail was para in relation to the 2-AI headgroup opposed to meta (18) or ortho (19) (Figure 2).⁹ We screened these same compounds for CLR potentiation against AB5075 and observed a similar trend, with the para amide linker 4k (highest adjuvant activity) > meta amide linker 18 > ortho amide linker 19 (lowest adjuvant activity), when dosed at 30 μM (Figure 2). Based on this we maintained the para position and explored diversifying the linker itself instead. We focused on incorporating a urea due to past reports showing that urea linkers are active in the context of antibiotic potentiation against Gram-negative species, including A. baumannii.^10,11

Figure 2.

Previously reported 3,5-difluoro aryl 2-AI analogs with varying amide linker connectivity to the phenyl core.⁹

Three of the most active aryl tails from the initial screenings, the aryl tails seen in compounds 1, 2, and 4r, were utilized for the synthesis of para urea derivatives (Scheme 4). Compounds 20a–c were synthesized by converting the 2-AI aniline intermediate 3 into its corresponding isocyanate using sodium carbonate and triphosgene in a biphasic solvent system at room temperature. Each isocyanate was then reacted with the appropriate halogenated aniline derivative to deliver the tri-Boc protected 2-AI intermediates. Finally, each urea derivative was exposed to trifluoroacetic acid to remove the Boc protecting groups and converted to the hydrochloric acid salt for biological testing (Scheme 4).

Scheme 4. Linker Modification of Active CLR Adjuvants 1, 2, and 4r.

Reagents and conditions: (a) Na₂CO₃, H₂O, DCM, then triphosgene, 25 °C, 2 h; (b) ArNH₂, DCM, 25 °C, 1 h; (c) TFA, DCM, 25 °C, 3 h; (d) 6 M HCl, MeOH, 25 °C, 5 min.

Modification of the amide linker to a urea led to an increase in adjuvant activity, yet at the cost of increasing their standalone toxicity toward AB5075 (Table 3). All three urea analogs 20a–c potentiate CLR ≥16-fold at concentrations as low as 7.5 μM, with 20a showing the greatest CLR potentiation of 128-fold and becoming our lead urea derivative (Table 3). However, compounds 20a and 20b displayed higher toxicity, returning MIC values of 50 and 25 μM, respectively. The least toxic urea was compound 20c, which displayed a standalone MIC of 100 μM. Compound 20c had superior adjuvant activity at 30 μM of all of the screened compounds, including the amide linkers, lowering the MIC of CLR to ≤0.0625 μg/mL (≥512-fold, lowest concentration tested). Interestingly, when compound 20c was dosed at 7.5 μM, its adjuvant activity was comparable to the most active amide linker 15b activity, lowering the MIC of CLR by 16-fold from 32 to 2 μg/mL (Table 3).

In our initial report describing the macrolide potentiation activity of adjuvants 1 and 2, we examined their activity against a panel of A. baumannii isolates that encompasses all major and most minor clinically relevant clades.^7,12,13 Both adjuvants suppressed macrolide resistance in 23 different AB strains at 30 μM, lowering CLR MIC values to ≤0.125 to 1 μg/mL.⁷ Since the amide linker compound 15b and the urea analog 20a displayed greater CLR potentiation than compounds 1 and 2 in AB5075, we chose to screen the same 23 isolates at lower concentrations with compounds 15b and 20a (Table S3). To begin, compounds 15b and 20a were screened individually to determine their standalone MIC values to ensure that toxicity is avoided during antibiotic potentiation screening (Table S2). Compound 15b returned MIC values of 50 μM in three of the AB isolates, though it was relatively nontoxic (≥100 μM) in the remaining 20 strains; therefore, compound 15b was tested at 10 μM when combined with CLR against all 23 strains. Compound 20a displayed MIC values ranging from 25 to 100 μM and thus was tested at 7.5 μM for all of the 23 AB isolates to avoid potential toxic effects. Compound 15b at 10 μM lowered the MIC of CLR in all 23 AB isolates, with reductions ranging from 8- to 32-fold (Table S3). Compound 20a at 7.5 μM exhibited even greater CLR potentiation, lowering the MIC of CLR between 16- and 128-fold in all 23 AB isolates (Table S3).

We previously reported that compounds 1 and 2 antagonize the activity of the polymyxin colistin against AB5075, effecting an increase in colistin MIC from 1 to 4 μg/mL at 30 μM.⁷ The mechanism of action of colistin is well-documented, and it is known that binding to LPS plays a significant role in polymyxin activity.⁵ In addition, a genome sequencing study of some colistin resistant strains of A. baumannii revealed that loss of LPS imparts increased colistin resistance.⁵ It was posited that adjuvants 1 and 2 could be interfering with LPS production or assembly and thus affecting the ability of colistin to bind to its target. Therefore, we wanted to evaluate the effect of these second-generation adjuvants at their active concentrations on colistin activity against AB5075 (Table S4). Some compounds that display moderate CLR potentiation at 30 μM, such as compounds 4h, 4m, or 4u, did not potentiate nor antagonize colistin activity when dosed at 30 μM (Table S4). However, the most active compounds from the CLR potentiation screenings did increase the MIC of colistin by 2- to 8-fold when dosed at their active concentrations against AB5075 (Table S4), as seen with adjuvants 1 and 2.

Finally, parent adjuvant 1 was evaluated for its ability to suppress the evolution of CLR resistance in AB5075 in comparison to that occurring upon exposure to CLR alone over the course of nine consecutive days. Evaluation of resistance evolution was conducted by serially passaging AB5075 in the presence of CLR with and without adjuvant 1 and determining CLR MICs of the resultant bacterial populations. When exposed to CLR alone, after day 2 of the evolution assay, the MIC of CLR increased from 32 μg/mL to ≥128 μg/mL in AB5075 (Table S5). By day 6 of serial exposure, the MIC of CLR had risen to 512 μg/mL, and the MIC of CLR remained approximately 512 μg/mL throughout the remaining 3 days of the assay (Table S5). We then tested whether adjuvant 1 could reduce the MIC of this mutant. At 30 μM, compound 1 reduced the CLR MIC from 512 to 2 μg/mL, demonstrating that the adjuvant was still active against a more CLR resistant strain.

The evolution assay was next conducted via serial passaging of AB5075 in the presence of CLR and 30 μM of adjuvant 1. We observed that adjuvant 1 suppressed the increase in macrolide resistance in AB5075 (Table S6) throughout the entire nine-day course. On day 2 and day 9 of passaging, the CLR MIC remained unchanged at 32 μg/mL, and the combination of adjuvant 1 at 30 μM lowered the CLR MIC 128-fold to 0.25 μg/mL, identical to what we observe for the activity of 1 against the parent AB5075 (Table S6).

In summary, we report optimized adjuvants within the aryl 2-AI class for macrolide potentiation against a virulent strain of Gram-negative bacteria, AB5075 (Figure 3). Compounds 4o, 4p, 4r, and 17b retained significant adjuvant activity at 10 μM, lowering the MIC of CLR 16-, 32-, 8-, and 16-fold, respectively, which was comparable to the 16-fold reduction at 10 μM for original adjuvants 1 and 2.⁷ Adjuvants that surpassed the activity of 1 and 2 were the amide-linked derivatives 4y, 10d, and 15b as well as urea analogs 20a–c. At 10 μM, compound 4y reduced the MIC of CLR by 128-fold, where compounds 10d and 15b reduced the MIC of CLR by 64-fold at 10 μM. However, compounds 10d and 15b both surpassed compound 4y’s activity at 7.5 μM, lowering the MIC of CLR by 8- and 32-fold, respectively.

Figure 3.

Structure–activity relationship trends observed and representative clarithromycin potentiation results against AB5075.

The urea analog 20c exhibited comparable adjuvant activity as amide compounds 10d and 15b at 7.5 μM, lowering the MIC of CLR by 16-fold. Interestingly, compound 20c was able to lower the MIC of CLR to ≤0.0625 μg/mL when dosed at higher concentrations (≥20 μM), which surpassed the fold reductions seen thus far with CLR. Urea derivatives 20a and 20b were more toxic, displaying standalone MIC values of 50 and 25 μM; however, they maintained the highest adjuvant activity at 7.5 μM out of all of the screened compounds. Compound 20a lowered the MIC of CLR 128-fold from 32 to 0.25 μg/mL at 7.5 μM, and compound 20b lowered the MIC of CLR 64-fold to 0.5 μg/mL at 7.5 μM.

Lead adjuvants 15b and 20a potentiated CLR activity against 23 additional clinically relevant A. baumannii strains at 10 and 7.5 μM, respectively. An evolution assay indicated that original adjuvant 1 suppresses the acquisition of CLR resistance in AB5075. Overall, these compounds represent valuable tools in ongoing mechanism of action studies, as well as future murine model studies to determine whether these compounds can potentiate CLR in vivo.

Glossary

Abbreviations

2-AI

2-aminoimidazole

CLR

clarithromycin

CLSI

Clinical and Laboratory Standards Institute

HCAI

hospital care-associated infection

MDR

multidrug resistance

MIC

minimum inhibitory concentration

OM

outer membrane

SAR

structure–activity relationship.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.0c00276.

• Biology experimental, chemistry experimental, ¹H and ¹³C nuclear magnetic resonance spectra, bibliography (PDF)

The authors declare the following competing financial interest(s): Dr. Melander is a co-founder and a board of directors member of Agile Sciences, a biotechnology company that is seeking to commercialize the antibiofilm and antibiotic sensitization activity of 2-aminoimidazoles.