Serendipitous Discovery of a Highly Active and Selective Resistance-Modifying Agent for Colistin-Resistant Gram-Negative Bacteria.

Antibiotic resistance is a growing global health concern. Colistin is one of the last-resort antibiotics that treats multidrug-resistant (MDR) Gram-negative bacterial infection. However, bacteria resistant to colistin have become increasingly prevalent. Using a bacterial whole-cell screen of a fragment-based library, one compound was discovered to resensitize MDR Escherichia coli AR-0493 to colistin with low mammalian toxicity. Interestingly, postscreening validation studies identified a highly related yet distinct compound as the actual substance responsible for the activity. Further studies showed that this novel resistance-modifying agent is not only very potent but also highly selective to potentiate the activity of polymyxin family antibiotics in a wide range of MDR Gram-negative bacteria. Thus, it may be further developed as a combination therapy to prolong the life span of colistin in the clinic.

Introduction

Antibiotic resistance is a growing world health concern.[1] In 2019, CDC estimated that over 2.8 million antibiotic-resistant infections occur and cause more than 35,000 people to die in the United States each year.[2] The predicted annual global death toll by 2050 is 10 million.[3] The antibiotic-resistant phenotypes have risen rapidly among clinically relevant pathogens. Gram-negative bacteria are of particular concern, including multidrug-resistant (MDR) Pseudomonas aeruginosa, Acinetobacter baumannii, and Enterobacteriaceae spp.(4) Treatment options for these Gram-negative resistant bacteria are very limited, leading to poor patient prognosis. These pathogens are often associated with severe nosocomial infections with 30–40% of mortality rates. Colistin is currently used as an essential last-resort treatment in clinic for infections with MDR Gram-negative bacteria.[5−8] In 2019, World Health Organization (WHO) designated colistin as “highest priority” among the “critically important antimicrobials for human medicine.”[9]

Colistin (a.k.a., polymyxin E) is a cationic antimicrobial peptide of the polymyxin family. Its antibacterial activity depends on electrostatic interactions between its positively charged moiety and negatively charged phosphate groups of Lipid A, a core component of lipopolysaccharide in the outer membrane of Gram-negative bacteria. Colistin displaces divalent cations from the outer membrane and in turn disrupts its integrity.[10] However, most Gram-negative bacteria, including Escherichia coli, Klebsiella pneumoniae, P. aeruginosa, A. baumannii, and Salmonella spp., have evolved or acquired colistin resistance mechanisms that involve cationic additions of 4-amino-arabinose and/or phosphoethanolamine (pEtN) to the phosphate groups in Lipid A.[11−13] Both modifications impair the binding of colistin and reduce its activity.[14,15] pEtN modification is responsible for resistance in A. baumannii and has rapidly spread across many Gram-negative bacteria, such as E. coli, K. pneumonia, and Salmonella spp., via the mobilized colistin resistance (mcr) plasmids.[16,17]

The discovery of novel antimicrobial agents for Gram-negative bacteria has been proven particularly difficult, mainly due to the prevalence of MDR efflux pumps and the extra barrier of their outer membrane.[18] A variety of novel approaches have been investigated to overcome colistin resistance, including the development of novel polymyxin analogues[19−21] and the use of combinations of colistin with other clinically approved antibiotics (e.g., carbapenems,[22] rifampicin,[23] tigecycline,[24] and fosfomycin).[25] Several novel small molecules have also been discovered to reverse colistin resistance in mcr-positive bacteria and showed promising results in vitro.[26−28] Encouraged by our previous success on the development of a variety of resistance-modifying agents (RMAs) that resensitize methicillin-resistant Staphylococcus aureus (MRSA) to β-lactam antibiotics,[29] we initiated a screening approach to resensitize MDR Gram-negative bacteria to colistin. Herein, we report a serendipitous discovery of a highly active and selective RMA for colistin-resistant Gram-negative bacteria.

Results and Discussion

A MDR E. coli strain AR-0493 (a.k.a., ATCC BAA-3170) was obtained from the CDC and FDA Antibiotic Resistance Bank (https://wwwn.cdc.gov/ARIsolateBank/) and used for the initial screen. We first determined the minimum inhibitory concentration (MIC) of colistin as 16 μg/mL in this strain using the standard Clinical Laboratory Standards Institute (CLSI) broth microdilution method.[30] Considering the difficulties in the discovery of novel antimicrobial reagents against Gram-negative bacteria, we chose to screen the fragment-based library of 3200 compounds from TimTec (Tampa, FL).[31] These compounds have low molecular weights ranging from 110 to 290 and relatively high aqueous solubility, making them an ideal library to screen against Gram-negative bacteria. The screen used a modified broth microdilution assay. Compounds were screened at the final concentration of 75 μM (c.a., 8–20 μg/mL) in 96-well plates in duplicate, and the cation-adjusted MHB (CAMHB) medium was supplemented with 2 μg/mL of colistin, its clinical breakpoint for intermediate resistant Enterobacteriaceae.[32] After incubation at 35 °C over 18 h, 34 compounds, in combination with colistin, inhibited bacterial growth. These compounds were cherrypicked for confirmation using the same assay, where each compound was tested in four concentrations (i.e., 75, 37.5, 18.7, 9.4 μM) in duplicate. The 10 most active compounds were identified as hits and repurchased from TimTec for further validation.

These 10 compounds were tested for their MICs and minimum resensitizing concentrations (MRCs) in the presence of colistin at 2 μg/mL in E. coli AR-0493. Out of these, three (see structures in Figure ) showed the desired activity and selectivity with MICs >32 μg/mL and MRCs ≤4 μg/mL. To evaluate their toxicity in mammalian cells, we treated human epithelial adenocarcinoma HeLa cells with each compound at a wide range of concentrations. After incubation at 37 °C over 24 h, the remaining cells were analyzed using CellTiter-Glo mammalian viability assay (Promega) and their half growth inhibition concentrations (GI50s) were determined by fitting the data using KaleidaGraph (Synergy Software). The GI50 of ST024336 (1, Figure ) was determined as 10 μg/mL, 5 times of its MRC (i.e., 2 μg/mL), while the GI50s of the other two were 0.9 and 2.0 μg/mL, respectively, lower than their respective MRCs (i.e., 2 and 4 μg/mL). Hence, we decided to focus on ST024336 for further studies.

Figure 1

Structures of the top 3 hits from the colistin RMA screen.

Next, we resynthesized compound 1 by following Santos’ procedure.[33] Compound 1 was obtained by treating meta-toluidine with excess dichloride 2 (Scheme ) in the presence of triethylamine. To our surprise, the purified compound 1 was unable to potentiate colistin in AR-0493 at up to 64 μg/mL, the highest concentration tested. This suggested that some impurity in the commercial sample may be responsible for the desired RMA activity. During the synthesis of 1, we noticed that a small amount of bis-aniline adduct, 3 (Scheme ), was also formed, which was difficult to separate from 1 due to their similar polarity. In addition, compound 1 is also readily hydrolyzed to the hydroxyl analogue 4 under basic aqueous condition.[34] Mass spectrometry analysis of the commercial compound confirmed our hypothesis and detected the mass of both compounds. Therefore, we synthesized and purified these two compounds using their, respectively, optimized procedures.[33,34]

Scheme 1

Synthesis of Compounds 1, 3, and 4

Gratifyingly, compound 3 is highly active with an MRC of 0.5 μg/mL in AR-0493 and showed no antimicrobial activity at up to 64 μg/mL. Compound 4, however, showed much lower RMA activity with an MRC of 8 μg/mL and no antimicrobial activity on its own. To further confirm the synergistic effect of compound 3 and colistin, the checkerboard broth microdilution assay was performed (Figure ). The fractional inhibitory concentration index (FICI) of compound 3 and colistin was calculated as <0.06, which is significantly lower than 0.5 and suggests a strong synergistic effect.[35] The GI50 of compound 3 in HeLa cells was also determined as 15 μg/mL, which is 30 times higher than its MRC.

Figure 2

Checkerboard broth microdilution assay showed dose-dependent potentiation of colistin by compound 3 against MDR E. coli. AR-0493. Data represents the mean OD (600 nm) in two biological replicates.

We next screened a panel of nine commonly used antibiotics for Gram-negative bacteria (Table ) for changes in MICs in the presence of 1 μg/mL compound 3 in E. coli AR-0349. At this concentration, 3 was able to enhance the activity of both colistin and polymyxin B by 16 folds and lower their MICs from 16 to 1 μg/mL. In addition to polymyxins, this strain is also resistant to ceftazidime (a third-generation cephalosporin), azithromycin, ciprofloxacin, rifampicin, and tetracycline. Interestingly, compound 3 did not affect the MICs of any of these antibiotics. In addition, 3 did not affect the MICs of two other antibiotics, gentamycin and meropenem, that this MDR strain is susceptible to. These results suggested that compound 3 is a highly selective RMA for polymyxin family antibiotics but not any other classes of antibiotics.

Table 1

Compound 3 Selectively Potentiates Polymyxins in Multidrug-Resistant E. coli

antibiotics	MICa	MIC (+3)a,b	fold of potentiation	INTa,c
colistin	16	1	16	R
polymyxin B	16	1	16	R
ceftazidime	>64	>64		R
meropenem	0.25	0.25		S
azithromycin	>64	>64		R
ciprofloxacin	32	32		R
rifampicin	16	16		R
tetracycline	>64	>64		R
gentamicin	0.5	0.5		S

MIC values are in μg/mL.

MIC in the presence of 1 μg/mL compound 3.

S-R interpretation (INT) derived from CLSI 2021, M100, Ed 31.[30]

To investigate the scope of the colistin-potentiation effect of compound 3, we screened a panel of MDR Gram-negative bacteria obtained from the CDC and FDA Antibiotic Resistance Bank. These include a variety of Enterobacteriaceae spp., such as E. coli AR-0346 and AR-0538, K. pneumoniae AR-0497, Salmonella Enteritidis AR-0496, and Salmonella Typhimurium AR-0635. These strains all express the mcr genes (e.g., mcr-1, mcr-2, and mcr-4) and are resistant to colistin with its MICs of 8–16 μg/mL (Table ). When used at 1 μg/mL, compound 3 showed 16 folds of potentiation for colistin and lowered its MICs to 0.5–1 μg/mL for almost all of these strains. The only exception is S. Typhimurium AR-0635, which expresses mcr-4 gene. In this strain, 1 μg/mL compound 3 was only able to lower the MIC of colistin to 4 and 2 μg/mL of 3 was required to lower the MIC of colistin to 1 μg/mL. We next tested another two A. baumannii strains from the same source. Compound 3 showed even more potent effect in this species with 32 folds of potentiation for colistin. Therefore, at 1–2 μg/mL, compound 3 resensitize all of these MDR Gram-negative bacteria to colistin and lower its MICs below 2 μg/mL, the clinical breakpoint. Furthermore, we also tested the activity of colistin in the presence or absence of 1 μg/mL compound 3 in a wild-type E. coli K-12. The MICs of colistin were found to be 0.25 μg/mL, regardless of the presence of compound 3. The MICs of compound 3 in all of these strains were >64 μg/mL, the highest concentration tested. Taken together, these results suggested that compound 3 is a highly active and selective RMA for polymyxin family antibiotics in resistant Gram-negative bacteria.

Table 2

Colistin MICs in the Presence and Absence of Compound 3 in a Panel of Gram-Negative Bacteria

bacterial strain	resistance mechanism	colistin MICa	colistin MIC (+3)b	folds of potentiation
E. coli AR-0493	mcr-1	16	1	16
E. coli AR-0346	mcr-1	8	0.5	16
E. coli AR-0538	mcr-2	8	0.5	16
S. Enteritidis AR-0496	mcr-1	16	1	16
S. Typhimurium AR-0635	mcr-4	16	1c	16
K. pneumoniae AR-0497	mcr-1	8	0.5	16
A. baumannii AR-0300		8	0.25	32
A. baumannii AR-0310		16	0.5	32
E. coli K-12		0.25	0.25

Colistin MIC in μg/mL.

Colistin MIC in μg/mL, in the presence of 1 μg/mL compound 3, unless otherwise noted.

Colistin MIC in μg/mL, in the presence of 2 μg/mL compound 3.

Conclusions

In summary, we have carried out a colistin RMA screen of a fragment-based library in MDR E. coli using a bacterial whole-cell assay. This screen identified three novel potentiators of colistin. Follow-up validation studies discovered that an impurity in the best hit, compound 3, was the actual active substance responsible for the RMA activity. Further evaluation of compound 3 in combination with a variety of antibiotics showed that it only potentiates the activity of polymyxin family antibiotics but not other classes of antibiotics. Evaluation of compound 3 in a wide range of MDR Gram-negative bacteria suggested that it potentiates colistin in both mcr-positive Enterobacteriaceae and chromosomally encoded colistin-resistant A. baumannii strains. In addition, compound 3 has low antibacterial activity and mammalian toxicity on its own. Therefore, compound 3 may be used in combination with colistin or other polymyxins, clinically approved or under development, to treat MDR Gram-negative bacterial infections and prolong the life span of these last-resort antibiotics. Further the structure–activity relationship and mechanistic studies are ongoing and will be reported in due course.

Methods

General Methods

MDR strains E. coli AR-0493, AR-0346, and AR-0538, S. Enteritidis AR-0496, S. Typhimurium AR-0635, K. pneumoniae AR-0497, and A. baumannii AR-0300 and AR-0310 were all obtained from the CDC & FDA Antibiotic Resistance Isolate Bank. E. coli K-12 was a gift from the Copley lab (Department of MCD Biology, University of Colorado Boulder). HeLa cells were purchased from the ATCC. All bacterial strains were grown on trypticase soy agar supplemented with 5% sheep blood (BAP). Bacteria were then cultured in Luria-Bertani (LB) medium at 35 °C until they achieved the log phase before screening or antimicrobial testing using CAMHB media. MIC, MRC, and checkerboard analyses and mammalian growth inhibition analyses in HeLa cells were performed as described previously.[29] More details are given in the Supporting information.

Compound Screening

The TimTec fragment-based library was screened using a modified broth microdilution assay. The MDR E. coli strain AR-0493 was used to screen the fragment-based library for RMA activity. Ninety-six-well assay plates were prepared containing 50 μL of CAMHB supplemented with 4 μg/mL colistin. Five hundred nanoliters of each compound (15 mM in dimethyl sulfoxide (DMSO)) was pinned to the assay plate using the CyBi-Well 96-channel simultaneous pipettor (Cybio). These plates were inoculated with 50 μL E. coli AR-0493 diluted in CAHMB to OD600 0.002. The final concentration of colistin in the screen was 2 μg/mL, the final concentration of each screening compound was 75 μM, and the final inoculum concentration was OD600 0.001. All plates were incubated at 35 °C with shaking for 18 h before the results were interpreted.