2‑Aminoimidazole-benzimidazole Conjugates Potentiate the Gram-Positive Selective Antibiotic Clarithromycin against Acinetobacter baumannii

Abstract

An increase in antibiotic resistance and the paucity of new treatment options have led to the present-day antibiotic resistance crisis. Acinetobacter baumannii is categorized by the World Health Organization as a critical priority. Previously, we reported halogenated aryl 2-aminoimidazole (2-AI) adjuvants that potentiate macrolide antibiotics against A. baumannii; however, this class of adjuvants exhibits cytotoxicity toward HepG2 cells. In this study we generate a library of 2-AI-benzimidazole adjuvants that potentiate clarithromycin (CLR), and exhibit reduced cytotoxicity. Lead compounds lower the CLR minimum inhibitory concentration (MIC) across a panel of A. baumannii clinical isolates, and have therapeutic indices (TI) up to 6-fold higher than the parent 2-AIs.

Antimicrobial resistance (AMR) is one of the leading global health and development threats to date. It is estimated that in 2019 antibiotic resistance was responsible for 1.27 million deaths worldwide. In addition, AMR poses a considerable economic burden, being responsible for an estimated $4.7 billion in additional health care costs in the United States alone. Specifically, the rise in antibiotic resistance poses a substantial threat to public health, with alarming rates of resistance among common bacterial pathogens. Many of the greatest health care threats are the so-called ESKAPE (E nterococcus faecium, S taphylococcus aureus, K lebsiella pneumoniae, A cinetobacter baumannii, P seudomonas aeruginosa, and E nterobacter species) pathogens. The ESKAPE pathogens have been identified as multidrug resistant (MDR) bacteria that can rapidly “escape” antibiotic action. Four of the six ESKAPE pathogens are Gram-negative pathogens, which are inherently more resistant to antibiotic activity than Gram-positive pathogens due to the presence of an outer membrane (OM) that is impermeable to many antibiotics. Of these, A. baumannii is one of the leading causes of nosocomial infections, making it one of the most dangerous and critical to treat MDR pathogens across the globe according to the World Health Organization. Mortality rates of A. baumannii infections have been reported to range from 29% to 73%. Carbapenem resistant A. baumannii strains that are often responsible for epidemic spread, typically show intermediate resistance to tigecycline, and usually retain susceptibility to colistin, though resistance to colistin is increasing. In the US, the siderophore cephalosporin cefiderocol, and the β-lactam−β-lactamase inhibitor combination sulbactam/durlobactam are also now available to treat MDR A. baumannii. However, concerns over excess mortality and low threshold of resistance to cefiderocol have limited its use in clinical practice, , meanwhile ∼4% of carbapenem-resistant isolates of A. baumannii are resistant to sulbactam/durlobactam, as are all metallo-ß-lactamase (MBL) producing strains.

Coupled with the increase in MDR strains, there has also been a significant decrease in antibiotic investment and development. Since the “golden era” of antibiotic development (1940s–1960s) there has been a lack of novel antibiotic classes developed. The majority of the antibiotic classes currently used in the clinic were developed during this era due to their broad-spectrum activity. Due to the growing prevalence and threat of MDR pathogens, coupled with the lack of novel antibiotic development, alternative approaches to combat this rising issue are urgently needed. One solution is prolonging the lifespan of current antibiotics through the use of antibiotic adjuvants. Antibiotic adjuvants are small molecules that have little to no standalone antibacterial properties; however, they increase the effectiveness of an antibiotic through minimization or blocking of bacterial resistance mechanisms. Such an approach has the potential to restore antibiotic efficacy and lead to a prolonged clinical runway. A notable example of an antibiotic adjuvant in the clinic is the β-lactamase inhibitor clavulanic acid. When combined with amoxicillin (Augmentin) it is used to treat a variety of bacterial infections that are resistant to amoxicillin alone due to the production of β-lactamases. In addition, adjuvants can be utilized to expand the spectrum of activity of an antibiotic, overcoming intrinsic resistance and, in some cases, rendering otherwise Gram-positive selective antibiotics active against Gram-negative bacteria. The activity of macrolides antibiotics against Gram-negative bacteria is limited, and while they have been used clinically to treat infections caused by Haemophilus influenzae and Moraxella catarrhalis, they lack activity against the Gram-negative ESKAPE pathogens. However, it has been shown that macrolides become effective against otherwise nonsusceptible Gram negative bacteria when the OM is compromised. For instance, A. baumannii strains that lack lipooligosaccharide (LOS) are highly permeable and are susceptible to azithromycin. In addition, compounds that physically disrupt the OM, such as polymyxin B nonapeptide and pentamidine, sensitize Gram-negative bacteria to macrolide antibiotics.

To identify new classes of adjuvants, we have investigated analogs of sponge-derived marine alkaloids containing a 2-aminoimidazole (2-AI) heterocycle, and demonstrated that properly derivatized 2-AIs potentiate a variety of antibiotics against a multitude of MDR and susceptible pathogens. For example, we have demonstrated that halogenated aryl 2-AIs, dimeric 2-AIs, and halogenated aryl 2-aminobenzimidazoles (2-ABIs) potentiate clarithromycin (CLR) against A. baumannii (representative structures for each class are depicted in Figure ). − Preliminary mechanism of action studies have indicated that macrolide potentiation by 2-AI and 2-ABI adjuvants in A. baumannii is due to altered LOS biosynthesis/presentation, which increases permeability and compromises the OM. While the halogenated aryl 2-AIs display potent CLR potentiation against A. baumannii, they exhibit cytotoxicity against a human hepatoma cell line (HepG2), thus limiting their therapeutic potential. For example, compound 1 (Figure , Figure S1), which is active at 10 μM in combination with CLR against the highly virulent primary clinical A. baumannii isolate AB5075, exhibits an IC₅₀ of 62.8 μM against HepG2 cells, giving a therapeutic index (TI) of 6.3. We recently disclosed that dimeric 2-AIs of the general structure shown in Figure exhibit improve CLR adjuvant activity against A. baumannii, and markedly reduced HepG2 toxicity; the lead compound of this series lowers the CLR MIC to 2 μg/mL at 1.5 μM, and has a HepG2 IC₅₀ of >530 μM, giving a TI of >353.

1.

Representative structure from each class of aryl 2-AIs, dimeric 2-AIs, and 2-ABIs that potentiate CLR, parent compound 1 and the appending of the amide linker to the phenyl core to generate the new 2-AI-benzimidazole class.

Based upon this, we posited that the halogenated benzoyl moiety of compound 1 could play a significant role in toxicity, and sought to explore a bioisostere approach to modify the halogenated aryl 2-AI scaffold. Using compound 1 as an illustrative example, replacing the aryl amide with the benzimidazole bioisostere (compound 2) would eliminate the halogenated benzoyl moiety. This scaffold exchange will also impart restricted bond rotation that may result in reduced entropic penalty upon binding relative to the aryl-2AI class, and allow for more facile diversification relative to the dimeric 2-AI class. Herein, we report a structure activity relationship (SAR) analysis encompassing structural modifications of lead compound 1 (Figure ) to further determine if we can optimize activity in combination with CLR and to reduce cytotoxicity of the aryl 2-AI scaffold. We have further evaluated the lead and parent compounds for cytotoxicity, and colistin antagonism to establish that activity is not dependent upon direct physical disruption of the OM, similar to our previously reported aryl 2-AIs.

To test whether the benzimidazole bioisostere is tolerated in the context of CLR potentiation, we synthesized the 3,5-dichloro cyclized linker analog (2) and evaluated whether compound 2 retains CLR adjuvant activity. Synthesis of 2 (Scheme ) began with a nucleophilic aromatic substitution of the commercially available 4’-fluoro-3′-nitroacetophenone under ammonia/MeOH conditions to yield the corresponding aniline, which was followed by α-bromination of the acetophenone with N-bromosuccinimide, and subsequent cyclization with Boc-guanidine to afford the 2-AI intermediate, 3. Reduction of the nitro group using 10% Pd/C and hydrogen gas yielded the diamino benzene, which underwent cobalt­(II) hydroxide-mediated oxidative cyclization with 3,5-dichlorobenzaldehyde. Finally, Boc-deprotection and counterion exchange afforded the cyclized derivative 2.

1. Synthesis of Cyclized Linker Derivative 2 .

^a Reagents and conditions: a) NH₃, MeOH, 1,4-dioxane, 100 °C, 3 h; b) NBS, p-TsOH, CH₃CN, 80 °C, 16 h; c) Boc-guanidine, THF, 56 °C, 3 h; d) 10% Pd/C, EtOAc, H₂, rt; e) substituted benzaldehyde, 10 mol % Co­(OH)₂, EtOH, rt; f) TFA, DCM; g) 6 M HCl/MeOH.

The standalone minimum inhibitory concentration (MIC) of compound 2 was first determined against two A. baumannii strains: AB5075, a hyper virulent clinical isolate, and AB19606, a drug susceptible ATCC strain. Against both strains, compound 2 has an MIC of >200 μM, for comparison, the amide parent 1 has an MIC of 100 μM against AB5075 and AB19606 (Table S1). Next, compound 2 was tested in combination with CLR against AB5075 and AB19606 at 30 μM to allow direct comparison of activity with compound 1 (compounds are not tested above 30% MIC to avoid antibacterial activity from the adjuvant). Since no clinical breakpoint is established for CLR against A. baumannii we opted to use the CLR breakpoint for S. aureus (2 μg/mL) as a guide for activity. We recognize that this is arbitrary, but this breakpoint falls centrally among those species for which CLR is used clinically (8 μg/mL for H. influenzae, and 0.25 for Streptococcus pneumoniae). The CLR MIC against both strains is 32 μg/mL. Compound 2 lowers the CLR MIC to 2 μg/mL in both strains. Fusing the linker to the phenyl core not only leads to retention of CLR adjuvant activity, albeit somewhat reduced (8-fold against AB5075, 4-fold against AB19606) but, also lowered standalone antibacterial activity against A. baumannii. Compound 2 exhibits an IC₅₀ of 69.8 against HepG2 cells, giving a TI of 2.3.

Next a pilot library was synthesized using the same route as for compound 2 (all compounds synthesized in this study are summarized in Scheme ), with various commercially available benzaldehydes, focusing on the chloro substituent and its steric isosteres. We investigated if the chloro group is necessary for CLR potentiation by synthesizing derivative 4, and determined how both positioning and number of the chloro- substituents affects activity (compounds 5-12). In addition, we evaluated both methyl and methoxy steric isosteres (13-17).

The standalone MICs of compounds 4-17 against AB5075 and AB19606 were first determined, with all 14 compounds returning an MIC of ≥200 μM against both strains (Table S1). Next, CLR adjuvant activity was probed at 30 μM (again, to allow direct comparison of activity with compound 1) against both strains (Table ). When there is no substitution on the aryl tail (4), adjuvant activity is essentially lost, returning a CLR MIC of 16 μg/mL against AB5075 and AB19606. When moving from a 3,5-disubstitution pattern to a 3,4-disubstitution pattern (5), the CLR MIC is lowered by 2048-fold and 1024-fold against AB5075 and AB19606, respectively. This is a ≥ 64-fold increase in activity over parent compound 2 in both strains, and a 16-fold increase in activity over parent compound 1 against AB5075. Due to the potent CLR potentiating activity of compounds with a 3,5- and 3,4-disubstituted pattern, we also investigated a 3,4,5-trichloro analog (6). However, activity of this analog was significantly decreased, only lowering the CLR MIC four- and 2-fold against AB5075 and AB19606, respectively. To establish if disubstitution is required for activity, we probed monochlorinated analogs 7-9. It is clear that there is no requirement for a disubstitution pattern because both compound 7 and 8, at 30 μM, display equivalent or superior CLR adjuvant activity in comparison to 2. Compound 7 lowers the CLR MIC to 0.25 μg/mL against AB5075, equivalent activity to parent compound 1, and an 8-fold increase in activity over compound 2. In addition, compound 7 displays potent CLR adjuvant activity against AB19606, lowering the MIC by 256-fold. Compound 8 exhibits comparable activity to compound 2 against AB5075, lowering the CLR MIC to 1 μg/mL. However, against AB19606 compound 8 exhibits an 8-fold increase in CLR adjuvant activity compared to compound 2. Switching from the para- (7) or meta-chlorinated (8) tails to the ortho-chlorinated aryl tail (9) abrogates activity, indicating either a lack of tolerance for the ortho-substitution pattern on the aryl tail, or a requirement for meta- or para-substitution. However, ortho-chlorinated analogs containing an additional chlorine substituent in the meta- or para-position (10–12) exhibit potent CLR activity, thus confirming a requirement for meta- or para-substitution. Lastly, we probed two chlorine isosteres (methyl and methoxy) in both the 4- and 3,5-substitution patterns. Introduction of a strong electron donating group (EDG), such as methoxy, in the 4-position (13) results in reduced activity, only lowering the CLR MIC against AB5075 and AB19606 by four- and 8-fold, respectively. Activity is abolished when the methoxy groups are placed in the 3,5-position (15), effecting no more than a 2-fold drop in CLR MIC against either strain. However, a weak EDG, such as methyl, in either the 4- (14), the 3,5-disubstitution pattern (16), or the 3,4-disubstitution pattern (17) is tolerated with all three compounds suppressing the CLR MIC to ≤ 1 μg/mL against AB5075 and AB19606.

1. Potentiation of CLR by Parent Compound 1 and Cyclized Linker Derivatives at 30 μM ^,

Compound	Ar	CLR MIC (μg/mL) against AB5075	CLR MIC (μg/mL) against AB19606
–	–	32	32
1	–	0.25 [128]	0.5 [64]
2	3,5-dichlorophenyl	2 [16]	2 [16]
4	phenyl	16 [2]	16 [2]
5	3,4-dichlorophenyl	0.016 [2048]	0.031 [1024]
6	3,4,5-trichlorophenyl	8 [4]	16 [2]
7	4-chlorophenyl	0.25 [128]	0.125 [256]
8	3-chlorophenyl	1 [32]	0.25 [128]
9	2-chlorophenyl	32 [−]	32 [−]
10	2,3-dichlorophenyl	1 [32]	0.5 [64]
11	2,4-dichlorophenyl	0.125 [256]	0.125 [256]
12	2,5-dichlorophenyl	0.5 [64]	0.5 [64]
13	4-methoxyphenyl	8 [4]	4 [8]
14	4-methylphenyl	1 [32]	0.25 [128]
15	3,5-dimethoxyphenyl	32 [−]	16 [2]
16	3,5-dimethylphenyl	0.25 [128]	0.5 [64]
17	3,4-dimethylphenyl	0.25 [128]	0.25 [128]

^aFull structures shown in Scheme .

^bFold-reduction represents MIC in absence of adjuvant/MIC in the presence of adjuvant. Experiments were performed in three replicates.

After probing chloro-substitution patterns and isosteres, we synthesized a subset of six analogs to probe different halogens (18-23) again using the route depicted in Scheme . Compounds were again tested for their standalone MIC against AB5075 and AB19606 (Table S1) before being tested at 30 μM in combination with CLR (Table ). All six compounds returned standalone MICs of ≥ 200 μM against both strains. Compound 18, containing a 4-fluorophenyl tail, displays equipotent activity to that of the 4-chlorophenyl analog (7). When replacing the 4-substituent with a larger halogen, such as a bromine (19), activity is slightly enhanced, reducing the CLR MIC 32- and 128-fold against AB5075 and AB19606, respectively. This trend continues with compound 20, which contains a 4-iodophenyl tail, and displays the most potent CLR activity of this series, suppressing the CLR MIC to 0.0625 μg/mL against both strains. The 3,5-difluorophenyl analog (21) exhibits an eight- and 4-fold increase in activity against AB5075 and AB19606, respectively, compared to compound 2. The 3,5-dibromophenyl analog (22) displays comparable activity to compound 2, returning CLR MICs of 4 μg/mL in both strains, which again the MIC is increased compared to the 3,5-difluorophenyl analog, 21. When the chloro substituents of compound 5 are replaced with fluoro (23) activity is reduced, with this analog returning a CLR MIC of 0.25 μg/mL (128-fold) against AB5075 and 0.125 μg/mL (256-fold) against AB19606.

2. Potentiation of CLR by Cyclized Linker Derivatives at 30 or 15 μM*.

		CLR MIC (μg/mL) [fold reduction]
Compound	Ar	AB5075	AB19606
–	–	32	32
18	4-fluorophenyl	2 [16]	1 [32]
19	4-bromophenyl	1 [32]	0.25 [128]
20	4-iodophenyl	0.0625 [512]	0.0625 [512]
21	3,5-difluorophenyl	0.25 [128]	0.5 [64]
22	3,5-dibromophenyl	4 [8]	4 [8]
23	3,4-difluorophenyl	0.25 [128]	0.125 [256]
24	3,5-bis(trifluoro­methyl)phenyl	2 [16]*	0.25 [128]*
25	3-trifluoromethyl-4-chlorophenyl	0.5 [64]*	0.125 [256]*

^aFull structures shown in Scheme .

^bFold-reduction represents MIC in absence of adjuvant/MIC in the presence of adjuvant. Experiments were performed in three replicates.

Finally, two additional analogs, 24 and 25, incorporating a trifluoromethyl substituent, a strong electron-withdrawing group (EWG), were synthesized using the same route depicted in Scheme . These analogs were tested for their standalone MIC, as well as in combination with CLR against AB5075 and AB19606. Both compounds display higher standalone antibacterial activity than the previous analogs, returning MICs of 50 μM against both strains (Table S1). Due to the increased standalone antibacterial activity, these compounds were tested at 30% their MIC (15 μM) in order to avoid toxicity from the adjuvant.

Both trifluoromethyl containing compounds lower the CLR MIC in both strains. Compound 24, containing a 3,5-bis­(trifluoromethylphenyl) tail, lowers the CLR MIC to 2 μg/mL (16-fold) against AB5075 and to 0.25 μg/mL (128-fold) against AB19606 (Table ). Compound 25, containing a 3-trifluoromethyl-4-chlorophenyl tail, has slightly increased activity compared to compound 24, reducing the CLR MIC 64- and 256-fold against AB5075 and AB19606, respectively.

Following this initial evaluation, a dose response study was conducted to determine the lowest active concentration at which the adjuvant reduces the CLR MIC to 2 μg/mL. All compounds, with the exception of compounds 5, 20, and 25, lost activity against both AB5075 and AB19606 at 15 or 10 μM (Table S2). Compounds 5, 20, and 25 exhibit potent CLR adjuvant activity at 7.5 μM, returning CLR MICs of ≤2 μg/mL against both AB5075 and AB19606 (Table ), however none of the three compounds lowered the CLR MIC to 2 μg/mL at 5 μM. This is comparable to our previously reported lead CLR aryl-2-AI adjuvant, which lowers the AB5075 CLR MIC to 1 μg/mL at 7.5 μM, and the lead 2-ABI adjuvant which lowers the MIC to 2 μg/mL at 5 μM, while the lead dimeric 2-AI is more active, lowering the CLR MIC to 2 μg/mL at 1.5 μM.

3. Dose Response Activity of Lead Compounds against AB5075 and AB19606.

		CLR MIC (μg/mL) [fold reduction]
Compound	Concentration Tested (μM)	AB5075	AB19606
–	–	32	32
5	15	0.125 [256]	0.5 [64]
	7.5	2 [16]	1 [32]
	5	8 [4]	8 [4]
20	15	1 [32]	0.125 [256]
	10	1 [32]	0.5 [64]
	7.5	2 [16]	1 [32]
	5	8 [4]	8 [4]
25	10	0.25 [128]	1 [32]
	7.5	1 [32]	1 [32]
	5	8 [4]	16 [2]

^aFold-reduction represents MIC in absence of adjuvant/MIC in the presence of adjuvant. Experiments were performed in three replicates.

Following determination of the dose response activity, the three most active compounds (5, 20, 25) were tested in combination with CLR, at 10 μM, against a panel of A. baumannii clinical isolates that comprises all major and most minor clinically relevant clades , (Table ). The standalone CLR MIC for each strain was first determined, with three strains returning an MIC of 16 μg/mL, five strains returning an MIC of 64 μg/mL, and the remaining strains returning a CLR MIC of 32 μg/mL. Compound 5 lowers the CLR MIC to or below the breakpoint of 2 μg/mL against 15 of the 26 strains tested, while compound 20 lowers the CLR MIC to the breakpoint against 17 strains. Compound 25 displayed the most potent CLR activity across the AB panel, dropping the CLR MIC to ≤ 2 μg/mL in 21 of 26 strains.

4. Potentiation of CLR against Select A. baumannii Clinical Isolates by Lead Cyclized Linker Derivatives 5, 20 and 25 at 10 μM.

		CLR MIC (μg/mL) [fold reduction]
Strain	–	5	20	25
AB2828	16	2 [8]	0.5 [32]	1 [16]
AB3927	32	2 [16]	1 [32]	0.25 [128]
AB4025	32	2 [16]	2 [16]	2 [16]
AB4027	32	2 [16]	1 [32]	1 [32]
AB4448	32	2 [16]	2 [16]	1 [32]
AB4878	32	2 [16]	2 [16]	1 [32]
AB5001	32	2 [16]	1 [32]	1 [32]
AB5256	32	0.0625 [512]	1 [32]	0.25 [128]
AB5674	64	1 [64]	1 [64]	0.5 [128]
AB5711	32	2 [16]	2 [16]	0.5 [64]
AB8967	32	1 [32]	2 [16]	0.5 [64]
AB3560	32	8 [4]	8 [4]	2 [16]
AB3785	64	1 [64]	16 [4]	2 [32]
AB3806	16	4 [4]	4 [4]	0.5 [32]
AB4026	32	1 [32]	8 [4]	1 [32]
AB4052	32	2 [16]	4 [8]	8 [4]
AB4269	32	1 [32]	4 [8]	16 [2]
AB4490	32	8 [4]	2 [16]	0.5 [64]
AB4498	64	8 [8]	8 [8]	8 [8]
AB4556	16	4 [4]	4 [4]	2 [8]
AB4795	32	4 [8]	2 [16]	2 [16]
AB4857	32	8 [4]	2 [16]	2 [16]
AB4957	32	8 [4]	1 [32]	16 [2]
AB4991	32	4 [8]	2 [16]	1 [32]
AB5197	64	8 [8]	2 [32]	2 [32]

^aFold-reduction represents MIC in absence of adjuvant/MIC in the presence of adjuvant. Experiments were performed in three replicates.

To probe the spectrum of activity of these compounds outside of A. baumannii, we tested compounds 5, 20, and 25 in combination with CLR against three other Gram-negative species: K. pneumoniae (strain ATCC BAA-2146 (KP2146)), P. aeruginosa (strain PAO1), and Escherichia coli (strain ATCC 25922 (EC25922)). KP2146 and PAO1 are highly resistant to CLR, returning standalone MICs of 512 μg/mL and 256 μg/mL respectively, while EC25922 has a CLR MIC of 32 μg/mL. No adjuvant activity (greater than 2-fold reduction in MIC) was seen against these additional Gram-negative species, indicating a selectivity for A. baumannii for this series of CLR adjuvants.

We have previously demonstrated through a series of observations including: altered LOS composition, a lack of activity in an LOS BODIPY-cadaverine displacement assay, limited reduction in activity upon addition of MgCl₂ or exogenous LOS, and antagonism of colistin (COL), that 2-AI and 2-ABI macrolide adjuvants do not directly bind LOS, and instead impact LOS biosynthesis/presentation. Specifically, compound 1 antagonizes COL against AB5075, raising the MIC from 1 μg/mL to 4 μg/mL at 30 μM, as does a representative benzimidazole adjuvant (MIC raised from 1 to 16 μg/mL at 60 μM). The antimicrobial activity of COL is driven by membrane disruption through binding LOS (or lipopolysaccharide (LPS) in other Gram-negative species). Some strains of A. baumannii can stop production of LOS, which eliminates the interaction of COL with the OM that is required for antibacterial activity, thus imparting colistin resistance. Adjuvants that bring about this LOS- phenotype will effect a reduction in COL susceptibility, whereas adjuvants that disrupt the OM through direct interaction with LOS should synergize with COL, as we have previously demonstrated for pentamidine. This mechanism of action would also explain the lack of activity against other the Gram-negative species examined above, as for most Gram-negative species, LPS/LOS is essential. Similarly, differences in essentiality of LOS across divergent A. baumannii strains could also play a role in the discrepancies in activity.

To investigate whether these cyclized linker analogs are acting in a similar manner to the aryl 2-AIs and 2-ABIs we have previously reported, we probed the relationship between colistin (COL) and compound 5 (Figure S2). In the presence of 50 μM 5, the COL MIC is raised from 1 μg/mL to 4 μg/mL against AB5075, potentially indicating a similar mechanism of action to compound 1.

To establish the TI of these compounds, the cytotoxicity of the lead cyclized linker derivatives, as well as the parent amide compounds, toward HepG2 cells was determined. The half-maximal inhibition activity (IC₅₀) was determined for the three most active analogs (5, 20, 25) and the parent compounds (26–28 Figures S3–4, Table ).

5. Cytotoxicity against HepG2 Cells for Lead Compounds and Their Parent Amides .

Compound	HepG2 IC50 (μM)	Active concentration in combination with CLR (μM)	TI
5	73.5 ± 1.1	7.5	9.7–9.9
20	97.1 ± 1.1	7.5	12.8–13.0
25	141.6 ± 1.1	7.5	18.7–19.0
26	53.8 ± 1.6	20	2.6–2.8
27	182.0 ± 1.1	–	–
28	32.5 ± 1.2	10	3.1–3.4

^aExperiments were performed in three replicates.

The TI in the case of conventional antibiotic development is typically defined as (mammalian cell IC₅₀/antibiotic MIC), with a TI of ≥ 50 desirable for further development. Since adjuvants are typically nontoxic to bacteria by themselves, that definition is not directly applicable. For adjuvant development, we define TI as (mammalian cell IC₅₀)/(adjuvant concentration that reduces the antibiotic MIC to breakpoint level). Similar to compound 1, parent amides 26 and 28 displayed moderate toxicity, exhibiting IC₅₀ values of 53.8 ± 1.6 μM and 32.5 ± 1.2 μM and TIs of 2.7 and 3.3, respectively. In comparison to parent amide 26, the cyclized linker analog 5 displays a slightly reduced cytotoxicity, returning an IC₅₀ of 73.5 ± 1.1 μM and an almost 4-fold increase in TI. Lead compound 20 has an IC₅₀ of 97.1 ± 1.1 μM and a TI of 12.9; the parent amide 27 returned an IC₅₀ of 182.0 ± 1.1 μM; however, due to its lack of CLR adjuvant activity, a TI could not be calculated. Compound 25 is the least cytotoxic, returning an IC₅₀ of 141.6 ± 1.1 μM and a TI of 18.9, approximately 6-fold higher than parent amide 28. While these compounds do not achieve a TI of ≥ 50, and have lower TIs than our lead dimeric 2-AIs (TIs > 250), they represent an improvement in cytotoxicity in comparison to the halogenated aryl-2-AI scaffold.

In conclusion, it is well understood that MDR A. baumannii infections create a serious burden worldwide, and approaches to eradicate such infections are required. Here we have identified and synthesized a series of novel 2-AI-benzimidazole analogs for potentiation of CLR against A. baumannii. An SAR study revealed analogs 5, 20, and 25 to be the most active 2-AI-benzimidazole adjuvants, lowering the CLR MIC to or below 2 μg/mL at a concentration as low as 7.5 μM. These adjuvants, similar to the parent amide compound, antagonize COL against AB5075, potentially indicating they do not directly bind LOS. Further mechanistic studies are needed to address whether these compounds are affecting assembly or biosynthesis of LOS. Finally, lead 2-AI-benzimidazole analogs exhibit overall reduced cytotoxicity compared to the parent amides, with lead compound 25 returning an IC₅₀ of 141.6 ± 1.1 μM against HepG2 cells, corresponding to a TI of ∼19. These results indicate that appending the linker to the phenyl core of the aryl 2-AI compounds reduces cytotoxicity, highlighting the potential for further exploration of this scaffold to increase TI. We anticipate that further refinement of these 2-AI-benzimidazole conjugates to increase CLR adjuvant activity will move these compounds closer to the targeted TI of 50, and enable evaluation in vivo.

Glossary

Abbreviations

2-AI

2-aminoimidazole

2-ABI

2-aminobenzimidazole

AB

Acinetobacter baumannii

AMR

antimicrobial resistance

CLR

clarithromycin

COL

colistin

EC

Escherichia coli

EDG

electron-donating group

EWG

electron-withdrawing group

HepG2

human hepatoma cell line

IC₅₀

half-maximal inhibition activity

KP

Klebsiella pneumoniae

LOS

lipooligosaccharide

MDR

multidrug resistant

MIC

minimum inhibitory concentration

NBS

N-bromosuccinimide

OM

outer membrane

TI

therapeutic index

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.5c00201.

• Experimental procedures, compound characterization, spectra, LC traces, bacterial strains, IC₅₀ curves, additional MIC data, checkerboard heat maps, and additional compound structures (PDF)

We would like to thank the National Institutes of Health (AI167284) for support.

The authors declare no competing financial interest.

Safety Statement: No unexpected or unusually high safety hazards were encountered