Sulfonamido-2-arylbenzoxazole GroEL/ES inhibitors are potent antibacterials against methicillin-resistant Staphylococcus aureus (MRSA)

Abstract

Extending from a study we recently published examining the anti-trypanosomal effects of a series of GroEL/ES inhibitors based on a pseudo-symmetrical bis-sulfonamido-2-phenylbenzoxazole scaffold, here, we report the antibiotic effects of asymmetric analogs of this scaffold against a panel of bacteria known as the ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species). While GroEL/ES inhibitors were largely ineffective against K. pneumoniae, A. baumannii, P. aeruginosa, and E. cloacae (Gram-negative bacteria), many analogs were potent inhibitors of E. faecium and S. aureus proliferation (Gram-positive bacteria – EC₅₀ values of the most potent analogs were in the 1–2 µM range). Furthermore, even though some compounds inhibit human HSP60/10 biochemical functions in vitro (IC₅₀ values in the 1–10 µM range), many of these exhibited moderate to low cytotoxicity to human liver and kidney cells (CC₅₀ values >20 µM).

Graphical Abstract

INTRODUCTION:

The persistence of antibiotic resistant pathogens is a significant health and economic burden worldwide. Six of the most problematic drug resistant Gram-positive (Gr+) and Gramnegative (Gr-) bacteria are commonly referred to as the ESKAPE pathogens. This panel of drug resistant bacteria includes Enterococcus faecium (Gr+), Staphylococcus aureus (Gr+), Klebsiella pneumoniae (Gr-), Acinetobacter baumannii (Gr-), Pseudomonas aeruginosa (Gr-), and Enterobacter species (Gr-).^1–5 Of particular significance, the CDC recently estimated that ~80,000 individuals are infected with MRSA in the US annually, with ~11,000 deaths – the highest mortality rate amongst all the antibiotic resistant bacteria. S. aureus-associated infections range from mild skin infections to life-threatening endocarditis, osteomyelitis, or bacteremia, which are usually treated with penicillinase-resistant beta-lactams or vancomycin. For more resistant strains, newer classes of antibiotics can be prescribed, such as linezolid, daptomycin, dalfopristin, telavancin, tigecycline, ceftaroline, oritavancin, dalbavancin, or tedizolid.^6–8 Unfortunately, despite the availability of several antimicrobials, treating MRSA infections remains a significant challenge owing to off-target toxicities of some antibiotics, lack of efficacy in life-threatening clinical infections such as bacteraemia, endocarditis, etc., and the emergence of pan-drug resistant strains.⁷ Thus, there is an urgent need for new antibacterial drugs that function against previously unexploited targets and pathways to circumvent predisposed resistance mechanisms.

Towards developing mechanistically unique antibacterial candidates, we have focused on exploiting bacterial protein homeostasis pathways. A network of molecular chaperones and proteases collectively functions to maintain protein homeostasis by assisting proteins to fold to their native, functional states, or ensuring their proper degradation and recycling.^{9, 10} Since such quality control mechanisms are vital to cell survival, targeting them with small molecule inhibitors should be an effective antibacterial strategy. While recent studies have investigated targeting DnaK (a molecular chaperone belonging to the HSP70 family) and Clp proteases, targeting of the bacterial GroEL/GroES chaperonin system has gone largely unexplored.^11–14 E. coli GroEL, which is the prototypical member of the HSP60 chaperonin family of molecular chaperones, is a homo-tetradecameric protein that forms two, seven-subunit rings that stack back-to-back with one another.^15–17 Through a series of events driven by ATP binding and hydrolysis, unfolded substrate polypeptides are bound within the central cavity of a GroEL ring and encapsulated by the GroES co-chaperonin “lid”, allowing protein folding within the sequestered chamber.^17–21 Since the GroEL/ES chaperonin system is essential for E. coli survival under all growth conditions, as it likely is for other bacteria, it represents an excellent target for antibacterial development.^{22, 23} Furthermore, GroEL is highly conserved in prokaryotes (generally >50% sequence identity between bacterial species, whether Gram-positive or Gramnegative), thus, targeting this molecular machinery has the potential for broad spectrum applicability. However, since human HSP60 is also highly conserved (48% identity with E. coli GroEL), we also need to consider whether GroEL inhibitors have off-target effects against HSP60 in human cells.

Towards our goal of exploiting the GroEL/ES chaperonin system as an antibacterial strategy, we previously reported a high-throughput screen that identified 235 small molecule inhibitors of the E. coli GroEL/ES chaperonin system.²⁴ In a subsequent study, we evaluated 22 of these GroEL/ES inhibitor hits for their antibacterial properties against the ESKAPE pathogens.²⁵ In another study, we evaluated a series of compound 1-based GroEL/ES inhibitors for their antibiotic effects against Trypanosoma brucei parasites – the causative agents of African sleeping sickness (Figure 1A).²⁶ As an extension of those two studies, herein, we have explored additional compound 1 analogs for their ability to selectively inhibit the prototypical E. coli GroEL/ES chaperonin system and growth of the ESKAPE pathogens over human HSP60/10 and liver and kidney cells. Overall, this study enabled us to better understand the bioactivity profiles of this molecular scaffold, which provided invaluable SAR to guide future studies to more rationally optimize the pharmacological properties of these antibacterial GroEL/ES chaperonin system inhibitors.

Figure 1.

Progression of hit-to-lead compound 1 analogs. A. Compound 1 was an initial hit that emerged from our high-throughput screening for GroEL/ES inhibitors.²⁴ We recently reported on a series of pseudo-symmetrical compound 1 analogs (2–14 – R¹ represents a variety of alkyl, aryl, and substituted phenyl substructures) that have antibiotic properties against Trypanosoma brucei parasites.²⁶ B. In the present study, we initially screened the 1–14 analogs for their ability to inhibit a panel of Gram-positive and Gram-negative bacteria, and found that the thiophene (6) and para-chloro-substituted (2c-p) analogs generated the most potent and selective inhibitors of MRSA proliferation. Data plotted in the gray zones represent EC₅₀ and CC₅₀ results beyond the assay detection limits (i.e., >100 µM). C. Based on initial SAR, compound 15 was developed (bearing the 2-chlorothiophene group on both sides) along with two parallel series of asymmetrical 2-chlorothiophene-based analogs (16–34, Right and Left) containing variable alkyl and aryl substructures at the R² positions were developed in an attempt to improve their selectivity indices between efficacy against bacterial proliferation and cytotoxicity in human cell viability assays (structures of the different R² groups are presented in Tables S3A and S4A in the Supporting Information).

RESULTS AND DISCUSSION

Identifying preliminary SAR of previously developed pseudo-symmetrical compound 1 analogs for antibacterial effects against the ESKAPE pathogens.

From our previous study exploring the antibacterial effects of our different GroEL/ES inhibitor scaffolds, we found that the parent hit, compound 1 (based on the benzimidazole core, Figure 1A), exhibited no antibacterial effects against any of the ESKAPE pathogens.²⁵ Therefore, before embarking on synthesis of new compound 1 analogs, we tested our previously developed benzoxazole-based GroEL/ES inhibitors (compounds 2-14), which have antitrypanosomal activities, for their antibacterial properties.²⁶ We expected that this initial evaluation would tell us whether or not this scaffold would be worthwhile to pursue for antibacterial development and, if so, allow us to identify which substituents, substitution patterns, and aryl substructures would provide the most potent antibacterial effects. We employed a standard antibacterial efficacy assay in liquid culture as previously reported, with one modification: the media was cation-adjusted by addition of 25 mg/L Ca²⁺ and 12.5 mg/L Mg²⁺ to better mimic the concentrations of these divalent cations in vivo, which can alter the antibacterial effects of some compounds (e.g. daptomycin is a more potent inhibitor when media is supplemented with additional Ca²⁺ and Mg²⁺ ions).²⁷ Briefly, bacteria were grown at 37°C without shaking (stagnant assay) in media stamped with test compounds. Compounds were first tested at single concentrations of 100 µM, then in dose-response for those that exhibited >50% inhibition of bacterial proliferation (refer to Tables S1A and S2A in the Supporting Information for a tabulation of all EC₅₀ results). After 6–8 h (S. aureus, MRSA, K. pneumonia, E. cloacae, and P. aeruginosa) or 24 h (E. faecium and A. baumannii), the absorbance of each well was read at 600 nm to monitor turbidity from bacterial growth, from which EC₅₀ results were obtained by plotting the dose-response measurements and fitting data with non-linear regression. A detailed protocol for this assay is listed in the Supporting Information.

Preliminary results from these proliferation assays indicated that all but one compound (2h-p) were inactive against the four Gram-negative KAPE bacterial species, and only three were able to inhibit E. faecium (the ortho-, meta-, and para-substituted analogs of hydroxylated compound 2h); however, several compounds were able to inhibit S. aureus and MRSA proliferation (Figure 1B, and Tables S1A and S2A in the Supporting Information). Furthermore, these compounds were reasonably selective at killing MRSA in relation to their cytotoxicity CC₅₀ values obtained from liver (THLE 3) and kidney (HEK 293) cell viability assays, which were determined in the previous study targeting T. brucei parasites.²⁶ Preliminary SAR revealed that the sulfonamide end-capping thiophene and para-chlorophenyl groups gave the most potent and selective aryl and substituent/substitution patterns, respectively (Figure 1B). Based on these initial results, we decided to combine these two moieties, reasoning that inhibition properties of each could be additive, and developed two parallel series of 2-chlorothiophene-based asymmetrical analogs (Figure 1C, analogs 15–34) containing variable substructures at the R² positions.

Synthesis of the two parallel series of 2-chlorothiophene-based analogs.

We rationalized that the previously developed pseudo-symmetrical compound 1 analogs (i.e. with the same sulfonamide end-capping substructures on either side of the molecules) may not be truly optimized for binding to and inhibiting the chaperonin system. Thus, to better complement the envisioned asymmetric binding sites, we designed two parallel series of analogs with a variety of alkyl and aryl substructures at the R² positions on the Right (R-series) and Left (L-series) sides of the 2-phenylbenzoxazole scaffold, while keeping the 2chlorothiophenesulfonamide moiety on the opposite sides (Figure 1C). The same R²-groups were used to create the panel of 19 matched-pairs for the different R- and L-series analogs, with an additional pseudo-symmetrical analog (15) that contained the 2-chlorothiophenesulfonamide on both sides (thus, a total of 39 new compounds were synthesized for evaluation). We expected that these matched pairs of analogs would help us to identify whether or not the directionality of the 2-phenylbenzoxazole core would significantly impact inhibitor effects in the various biochemical and cell-based assays they would be tested in. The general syntheses of these analogs are shown in Scheme 1, with detailed procedures and compound characterizations presented in the Experimental section and Supporting Information. Each of the R- and L-series of analogs were synthesized through 5-step linear protocols employing facile reactions. All final test compounds were characterized by ¹H-NMR and LC-MS analyses for structural confirmation, and by two independent sets of HPLC conditions for purity identification (all were >95% pure under both conditions).

Scheme 1^a.

For synthesizing the R-series analogs, 2-chlorothiophenesulfonyl chloride was first coupled to 4-amino-2-nitrophenol in dichloromethane using pyridine as base, giving 35.^{26, 28, 29} Reduction of the nitro group was accomplished by reacting with tin powder in a 1:10 v/v mixture of hydrochloric acid in glacial acetic acid.^28–30 Next, an intermediate Schiff base was formed by refluxing 36 and para-nitrobenzaldehyde with sodium bicarbonate in anhydrous THF, which was then cooled and cyclized to benzoxazole 37 by addition of DDQ.³¹ The nitro group was again reduced using tin powder in HCl/AcOH, affording intermediate 38, from which the final bissulfonamido analogs (15 and 16R-34R) could be rapidly generated by coupling with a variety of differentially-substituted sulfonyl chlorides (R²-SO₂Cl).

For synthesizing the L-series analogs, 2-chlorothiophenesulfonyl chloride was first coupled to methyl-4-aminobenzoate in dichloromethane using pyridine as base, giving 39. The ester was then hydrolyzed with LiOH in a mixture of H₂O/MeOH/THF, affording acid 40.^{28, 30, 32} The acid was then converted to the acyl chloride, concentrated, and coupled to 2-amino-4-nitrophenol in dichloromethane with pyridine as base, affording the intermediate amide, which was then cyclodehydrated to 41 by refluxing with toluenesulfonic acid in xylenes.^{29, 33} The nitro group was subsequently reduced with tin powder in HCl/AcOH, affording intermediate 42, from which the final bis-sulfonamido analogs (16L-34L) could be rapidly generated by coupling with a variety of differentially-substituted sulfonyl chlorides (R²-SO₂Cl).

Evaluating compounds for antibiotic effects against the ESKAPE bacteria and inhibition of GroEL/ES-mediated folding of substrate proteins.

Before proceeding with extensive biochemical and biophysical characterization of the new 2-chlorothiophene-based analogs, we first tested for their antibiotic effects on the ESKAPE bacteria using the proliferation assay described above (refer to Tables S3A and S4A in the Supporting Information for tabulations of all EC₅₀ results). Much like the initial testing with the pseudo-symmetrical analogs, all but one of the 2-chlorothiophene analogs were inactive against the four Gram-negative KAPE bacterial species (27R had an EC₅₀ of 37 µM against A. baumannii). However, 14 of the 39 new analogs (36%) exhibited antibacterial effects against E. faecium, compared to 3 out of 50 (6%) of the pseudo-symmetrical compounds. What was particularly impressive was that 27 out of the 39 new analogs (69%) inhibited S. aureus and MRSA, with greater overall potency, compared to the 20 out of 50 (40%) of the pseudosymmetrical compounds (Figure 2A presents a correlation plot of EC₅₀ values for each compound determined in these two assays). To compare assay results between the R- and Lseries analogs, we analyzed the log(EC₅₀) values of each data set using two-tailed, paired t-tests (95% confidence level) and looked at differences between paired values (results are plotted in Figure S1 in the Supporting Information). For the E. faecium EC₅₀ results, we did not see a statistical difference between results from the R- and L-series analogs (Figure S1A). However, there was a statistically significant difference between results of the two series for inhibiting S. aureus and MRSA proliferation, with the L-series generally providing more potent inhibitors of bacterial proliferation for each strain (Figures S1B and S1C). Although exceptions are evident for individual compounds, there was no statistical difference between inhibiting the drug susceptible S. aureus and MRSA strains when the data sets were analyzed for all 89 compounds as a whole (Figure S5A).

Figure 2.

Correlation plots of IC₅₀ and EC₅₀ values for compounds tested in the respective GroEL/ES-dMDH or dRho refolding assays and bacterial proliferation assays. Each data point represents results for individual compounds tested in the respective assays (plotted from results presented in Tables S1-S4 in the Supporting Information). Data plotted in the gray zones represent results beyond the assay detection limits (i.e., >100 µM). A. Compounds inhibit the proliferation of S. aureus (drug susceptible) and MRSA (drug resistant) bacteria nearly equipotently. The L- and R-series asymmetric inhibitors are generally more potent than the previously developed pseudo-symmetrical inhibitors (1-14). B. Compounds are on-target for inhibiting chaperonin-mediated substrate refolding as a correlation is evident between IC₅₀ values obtained from the GroEL/ES-dMDH and -dRho refolding assays, and compounds do not inhibit in both the native MDH and Rho enzymatic activity counter-screens. C. A general trend is observed between inhibiting the GroEL/ES chaperonin system and MRSA proliferation, supporting on-target effects in bacteria.

Having ascertained that the majority of the R- and L-series analogs were potent inhibitors of S. aureus and MRSA proliferation, we next evaluated their abilities to inhibit the biochemical function of the GroEL/ES chaperonin system. For this, we employed our previously reported assays that monitor GroEL/ES-mediated refolding of two denatured substrate enzymes, malate dehydrogenase (dMDH) and rhodanese (dRho).²⁵ Since these were coupled assays that monitor chaperonin-mediated refolding of substrates by virtue of the enzymatic activities of the refolded substrates, we further counter-screened against the native MDH and Rho enzymes to ensure that compounds were not simply false-positives of the reporter reactions. Detailed protocols for these assays and tabulation of all IC₅₀ results are presented in the Supporting Information.

Because the previously developed pseudo-symmetrical analogs (1-14) were not tested in the GroEL/ES-dRho refolding assay, nor the native rhodanese enzymatic reporter reaction counter screen, we also evaluated those compounds in these two assays. When looking at all 89 compounds as a whole, although a correlation is evident, we noticed a statistically significant difference between the GroEL/ES-dMDH and GroEL/ES-dRho refolding IC₅₀ results (Figures 2B and S5B), where compounds were slightly more potent at inhibiting GroEL/ES-mediated refolding of denatured MDH. This difference does not appear to be a result of compounds preferentially inhibiting the native MDH reporter enzyme over rhodanese – while eleven analogs inhibited the native MDH enzymatic reporter reaction, five inhibited native rhodanese, and only to a minor extent. Since only one compound inhibits both native malate dehydrogenase and native rhodanese (28R), and only to a minor extent, this supports that compounds are on-target for inhibiting the GroEL/ES-mediated folding cycle.

For the purposes of categorizing inhibitor potencies in the GroEL/ES-mediated refolding assays, we consider compounds with IC₅₀ values >100 µM to be inactive, 30–100 µM to be weak inhibitors, 10–30 µM moderate inhibitors, 1–10 µM potent inhibitors, and <1 µM very potent and acting near stoichiometrically since the concentration of GroEL tetradecamer is 50 nM during the refolding cycle (i.e. 700 nM of GroEL subunits). Upon further dissection of the GroEL/ES-dMDH refolding assay IC₅₀ results, we observed that the asymmetric R- and L-series analogs are generally more potent than the previously developed pseudo-symmetric analogs: 28 out of 38 (74%) of the asymmetric R- and L-series analogs (16-34) have IC₅₀ values less than 10 µM, whereas only 16 out of 51 (31%) of the pseudo-symmetric analogs (1-15) have IC₅₀ values less than 10 µM. These results are not entirely surprising since the R- and L-series analogs were hypothesized to better complement the envisioned asymmetric binding sites than the pseudosymmetric compounds. However, we caution on over-interpreting these results since the analog groupings did contain different alkyl and aryl substructures on the sulfonamide end-caps. When comparing the R- and L-series analogs with each other, though, we do not see any statistically significant differences in either the GroEL/ES-dMDH or GroEL/ES-dRho refolding assay results (Figure S2A and S2B), suggesting that the orientation of the 2-phenylbenzoxazole core scaffold does not play a significant factor in these compounds binding to and inhibiting the GroEL/ES chaperonin system.

When comparing IC₅₀ results for compounds tested in the biochemical GroEL/ES-dMDH refolding assay with EC₅₀ results for testing in the S. aureus proliferation assay (Figure 2C and Tables S3A-B and S4A-B), we note that no compounds are active against bacteria unless they are able to inhibit GroEL. In general, the more potent compounds are at inhibiting the GroEL/ES chaperonin system, the more potent their antibacterial effects, in particular for the L-series inhibitors. However, several exceptions are evident where potent GroEL/ES inhibitors are poorly effective, or ineffective, against bacteria, which may owe to poor cell wall permeability and/or efflux, as we had previously observed for other GroEL/ES inhibitor scaffolds.²⁵ Also, while correlations between biochemical IC₅₀ and cell-based EC₅₀ are only suggestive of mechanisms of action in cells, they are not confirmatory. Thus, whether compounds are “ontarget” for GroEL within bacteria is still unknown and warrants further investigation. These studies are ongoing and will be reported in the future.

Characterizing GroEL-inhibitor binding interactions using Isothermal Titration Calorimetry.

To further interrogate inhibitor mechanisms of action against GroEL, we used isothermal titration calorimetry (ITC) to identify the thermodynamic parameters, binding affinities, and binding stoichiometries of compounds 20R and 20L. While these two compounds were only moderately active at inhibiting S. aureus and MRSA proliferation, they were strong inhibitors of GroEL/ES-mediated folding functions, and thus likely had high binding affinities. Furthermore, because they bore primary amines that would be charged under physiological conditions, they were much more soluble than other inhibitors and thus more amenable to ITC analysis, which requires high concentrations of both protein and ligands in matched aqueous buffers. We performed ITC analyses by titrating 400 µM GroEL (monomer concentration) into solutions containing either 50 µM 20R or 20L (detailed protocols are presented in the Experimental section). Two representative isotherms for the binding of 20R and 20L to GroEL are presented in Figure 3. After subtraction of background heats of mixing and dilution, plots of the integrated heats from compound binding fit well to a single-site binding model.³⁴ Averaged results for the various binding parameters (K_d, n, ∆H, ∆S, and ∆G) obtained from replicate analyses (six replicates for 20R, and five replicates for 20L) are presented in Table 1.

Figure 3.

Representative analyses of the binding of 20R (Panel A) and 20L (Panel B) to GroEL measured by Isothermal Titration Calorimetry (ITC). Top panels show binding isotherms obtained by titrating GroEL (400 µM monomer concentration) into solutions of either 20R or 20L (50 µM) in the ITC cell. Lower panels show the integrated data (solid squares) fit to a single-site binding model (solid lines). The molar ratio refers to the binding stoichiometry of monomeric GroEL to molecules 20R or 20L. Average results for the various binding parameters (K_d, n, ∆H, ∆S, and ∆G) obtained from replicate analyses are presented in Table 1.

Table 1.

Isothermal Titration Calorimetry (ITC) results for compounds 20R and 20L. Results are averaged from six replicate analyses for 20R, and five replicate analyses for 20L.

		20R	20L
Assay IC50(μM)	GroEL/ES-dMDH Refolding	1.3	3.1
	GroEL/ES-dRho Refolding	3.2	11
Stoichiometry n	GroELmonomer : Molecule	0.77 ± 0.16	0.6 ± 0.14
	Molecule : GroELmonomer	1.30 ± 0.27	1.67 ± 0.39
	Molecule : GroELoligomer	18.2 ± 3.8	23.3 ± 5.4
Log(Kd /nM)		1.56 ± 0.20	2.18 ± 0.03
Kd (nM)		36	150
∆H (kcal/mol)		−9.78 ± 0.80	−8.30 ± 0.52
∆S (kcal/molK)		0.001 ± 0.003	0.003 ± 0.002
T∆S (kcal/mol)		0.20 ± 0.82	0.84 ± 0.55
∆G (kcal/mol)		−9.98 ± 0.26	−9.14 ± 0.06

From the averaged results, we found that 20R had a K_d of 36 nM for binding to E. coli GroEL, while the K_d of 20L was about four-fold higher (150 nM), which corresponds reasonably well with the relative IC₅₀ values for inhibiting GroEL/ES-mediated refolding of dMDH and dRho – the IC₅₀ values for 20L were typically two- to four-fold higher than for 20R. For both compounds, binding was predominantly enthalpically driven, with minor entropic contributions to affinity. What was particularly interesting from these analyses was the stoichiometry of binding of each compound to GroEL. Since GroEL consists of 14 identical subunits, we anticipated that compounds could bind with a stoichiometry of 14, or potentially 7 if there is negative cooperativity between the two GroEL rings (also assuming that for inhibitor binding there is no negative cooperativity between subunits within a ring). Since IC₅₀ values were previously found to correlate between GroEL/ES-mediated refolding and ATPase assays for this inhibitor scaffold, the most likely binding sites may be the ATP pockets, of which there are 1 per GroEL subunit (14 per oligomer).²⁶ However, 20R bound with a stoichiometry of roughly 18 molecules per GroEL tetradecamer, and 20L with 23 molecules, which could indicate more than one potential binding site per GroEL subunit, and potentially an unknown site outside of the ATP pockets. We are currently pursuing X-ray crystallographic studies to identify specific inhibitor binding sites, which we will report on in future studies.

Counter-screening compounds for inhibition of HSP60/10-mediated refolding of denatured MDH and for cytotoxicity in human cell viability assays.

Having established that the R- and L-series analogs were potent GroEL/ES chaperonin system inhibitors with anti-staphylococcal properties, we next investigated their selectivity profiles compared to the human HSP60/10 chaperonin system. For this, we evaluated compounds in our previously reported HSP60/10-dMDH refolding assay, which is analogous to that used for determining inhibition of bacterial GroEL/ES, but alternatively employs the human chaperonin system. A detailed protocol and tabulation of all IC₅₀ results for this assay are presented in the Supporting Information. We previously reported that the pseudo-symmetrical analogs (1-14) were highly selective for inhibiting E. coli GroEL/ES over human HSP60/10 (Figure 3A and S3A).²⁶ However, for the asymmetric compounds developed in this study, selectivity for inhibiting GroEL/ES over HSP60/10 was statistically significant for the R-series analogs (Figure S3C), but was lost for the L-series analogs (Figure S3B). Correspondingly, differences in IC₅₀ values between the R- and L-series inhibitors in the HSP60/10-dMDH refolding assay were statistically significant (Figure S4A), which suggests that directionality of the 2-phenylbenzoxazole core does have an effect on compounds binding to and inhibiting the human chaperonin system, unlike the situation for E. coli GroEL/ES.

We next evaluated the cytotoxicity of our compounds to human liver (THLE 3) and kidney (HEK 293) cell lines in a well-established, 48 h, Alamar Blue-based cell viability assay. Detailed protocols and tabulation of all cytotoxicity CC₅₀ results for these assays are presented in the Supporting Information. Intriguingly, even though the L-series analogs are more potent HSP60/10 inhibitors than the R-series, we observed no statistically significant differences between the L- and R-series CC₅₀ values in the HEK 293 kidney cell viability assay (Figure S4B). Furthermore, we observed an opposing trend where the L-series was less cytotoxic to the THLE 3 liver cells compared to the R-series analogs, although the differences were small (Figure S4B & C). These results suggest that the compounds may be unable to penetrate the mitochondrial matrix to engage with the HSP60/10 chaperonin system, and instead may be interacting with other targets in the cytosol or other sub-cellular compartments; however, we have yet to confirm this hypothesis. What is particularly promising with these results is that the cytotoxicity of lead analogs has not concomitantly increased as with antibacterial potency. Compared to the previously developed pseudo-symmetrical analogs, several lead asymmetrical analogs have appreciably higher Selectivity Indices (SI) for inhibiting MRSA proliferation over cytotoxicity in the liver cell viability assay (Figure 4C). The biochemical and cell-based IC₅₀, EC₅₀, and CC₅₀ results for the top four lead inhibitors of the pseudo-symmetrical, R-series, and L-series analogs are presented in Table 2, ranked based on their selectivity indices. Notably, compounds 24L and 25L have selectivity indices of 31 and 46, respectively.

Figure 4.

Correlation plots of IC₅₀, EC₅₀, and CC₅₀ values for compounds tested in the respective GroEL/ES-dMDH and HSP60/10-dMDH refolding assays, liver cell cytotoxicity assay, and MRSA proliferation assay. Each data point represents results for individual compounds tested in the respective assays (plotted from results presented in Tables S1-S4 in the Supporting Information). Data plotted in the gray zones represent results beyond the assay detection limits (i.e., >100 µM). A. The R-series and previously developed analogs selectively inhibit the E. coli GroEL/ES over the human HSP60/10 chaperonin system, but the L-series analogs are not selective. B. No correlation is evident between cytotoxicity of compounds in the liver cell viability assay and the HSP60/10-dMDH refolding assay, suggesting cytotoxicity may be from off-target effects in cells. C. Many compounds are able to selectively inhibit MRSA proliferation with moderate to low cytotoxicity to human liver cells. Results are similar for compounds tested in the kidney cell viability assay.

Table 2.

Biochemical and cell-based IC₅₀, EC₅₀, and CC₅₀ results for the top four lead inhibitors based on Selectivity Index (SI) for inhibiting MRSA proliferation over cytotoxicity in the liver cell viability assay.

	Compound Structure & Number	Cell-Based Assays			Biochemical Assay IC50 (µM)
		MRSA Proliferation EC50 (µM)	Liver Cell Viability Index (SI) CC50 (µM)	Selectivity Index (SI) (CC50 / EC50)	Native Rho Reporter	Native MDH Reporter	GroEL/ES refolding of:		HSP60/10-dMDH Refolding
							dRho	dMDH
L-Series Analogs		0.53	24	46	>100	>63	6.4	5.1	4.7
		1.1	36	31	>100	43	11	6.7	5.0
		2.2	27	12	>100	40	11	5.2	2.4
		0.97	11	12	>100	37	1.0	1.1	1.0
R-Series Analogs		4.5	69	15	72	>63	19	10.0	11
		2.3	27	12	>100	>63	4.5	2.2	4.5
		2.0	22	11	>100	>63	7.3	2.6	6.1
		1.7	17	10	53	34	2.0	1.5	2.1
Pseudo-symmetrical Analogs		2.0	24	12	>100	>63	33	11	58
		4.6	49	11	>100	52	5.6	2.4	5.9
		7.3	47	6.4	>100	>63	22	24	51
		3.5	18	5.2	>100	>63	29	15	96

Investigating the potential for MRSA to gain resistance to lead inhibitors.

While selectivity indices (and therapeutic windows with regards to in vivo evaluation) are important criteria to consider for antibacterial development, it is also important to determine whether or not bacteria will be able to rapidly develop resistance to lead candidates, especially if they function through new mechanisms of action. To gauge the efficacy of our GroEL/ES inhibitors for eluding acute antibacterial resistance mechanisms, we performed a step-wise, liquid culture resistance assay by consecutively passaging the MRSA strain (ATCC BAA-44) for 12 cycles over 12 consecutive days in the presence of serially-diluted inhibitors 20R, 28R, and vancomycin for comparison (a detailed protocol is presented in the Experimental section).^{35, 36} At the end of each cycle (i.e. each 24 h passage), EC₅₀ values were determined for the test compounds, with the assumption that EC₅₀ values would increase over time if MRSA was able to generate acute resistance. A plot of test compound EC₅₀ values over time is presented in Figure 5A. Despite the fact that 28R was a potent MRSA growth inhibitor (initial EC₅₀ value of 1.7 µM), the bacteria rapidly developed resistance to compound concentrations in excess of 100 µM. However, after culturing in the absence of inhibitors for two days (24 h on agar, and another 24 h in liquid media), bacteria regained sensitivity to 28R (Figure 5B), suggesting resistance is reversible, potentially from up-regulation of efflux pumps. Most importantly, we found that the parent MRSA strain was unable to generate resistance against 20R over the 12 day passage, even though this analog was only a moderate inhibitor of MRSA proliferation (EC₅₀ = 19 µΜ). These results suggest that pan-resistance to these GroEL inhibitor analogs may be difficult to develop, which supports the continued optimization of lead antibacterial candidates based on this molecular scaffold.

Figure 5.

Exploring adaptive tolerance by MRSA bacteria to analogs 20R, 28R, and vancomycin. A. MRSA was serially passaged for 12 days in fresh liquid media supplemented with the selected compounds. Average EC₅₀ values of compounds tested after each 24 h passage are plotted from triplicate analyses. MRSA rapidly evolved resistance to 28R, but retained sensitivity to 20R and vancomycin throughout 12 day experiment. Data plotted in the gray zones represent EC₅₀ results beyond the assay detection limits (i.e., >100 µM). B. After re-culturing the 28R-resistant MRSA strain for two days in the absence of inhibitors (24 h on agar, then another 24 h in liquid medium), the “resistant” MRSA strain regained sensitivity to 28R (average % growth inhibition results (±SD) are plotted from 4 replicate analyses).

CONCLUSIONS.

From exploratory screening of the antibacterial effects of our previously reported pseudosymmetrical compound 1 analogs,²⁶ we were able to develop two parallel R- and L-series of 2chlorothiophene-based asymmetrical analogs with significantly improved antibacterial efficacy profiles against MRSA. While there were no statistical differences between either the R- or Lseries analogs for inhibiting in the GroEL/ES-mediated refolding assays, the L-series analogs were found to be more potent at inhibiting S. aureus and MRSA proliferation – i.e. preferred from an antibiotic development perspective (Table 3). While a general trend is observed between inhibitor potencies in the biochemical GroEL/ES-dMDH refolding assay and the MRSA proliferation assay (in particular for the L-series analogs), further studies are needed to conclusively determine whether or not inhibitors are on-target in bacteria. Using ITC to characterize the binding interactions between GroEL and inhibitors 20R and 20L, we found that inhibitor binding was predominantly enthalpically driven. Furthermore, we observed that >14 molecules were bound per GroEL tetradecamer (i.e. more than one molecule bound per GroEL subunit), suggesting that inhibitors could be binding to unknown sites outside of the ATP pockets. Current studies are underway to identify and characterize these putative binding sites so that more rigorous structure-based optimization of lead candidates can be conducted. Even though the R-series analogs were overall less potent at inhibiting HSP60/10, the L-series was less cytotoxic against the human liver cells tested, and is likely preferential for further antibacterial optimization going forward. However, before pursuing in vivo development of this series, further medicinal chemistry optimization is warranted to increase the selectivity indices for killing MRSA over cytotoxicity to human cells.

Table 3.

Comparison of the L- and R-series analogs for which overall scaffolds exhibited preferred inhibition profiles in the various biochemical and cell-based assays. Lead antibacterial candidates are preferred to potently inhibit GroEL/ES and bacteria, and not inhibit human HSP60/10 and liver/kidney cell viability. The log(IC₅₀), log(EC₅₀), or log(CC₅₀) values for the 20 matched L- and R-series analogs were analyzed by two-tailed, paired t-tests with 95% confidence intervals. Comparative plots are presented in Figures S1-S5 in the Supporting Information.

Assay Inhibition	L-Series	R-Series
S. aureus growth	Preferred
MRSA growth	Preferred
GroEL/ES-dMDH refolding	No statistical difference
GroEL/ES-dRho refolding	No statistical difference
HSP60/10-dMDH refolding		Preferred
Liver cell viability	Preferred
Kidney cell viability	No statistical difference

EXPERIMENTAL.

General Synthetic Methods.

Unless otherwise stated, all chemicals were purchased from commercial suppliers and used without further purification. Reaction progress was monitored by thin-layer chromatography on silica gel 60 F254 coated glass plates (EM Sciences). Flash chromatography was performed using a Biotage Isolera One flash chromatography system and eluting through Biotage KP-Sil Zip or Snap silica gel columns for normal-phase separations (hexanes:EtOAc gradients), or Snap KP-C18-HS columns for reverse-phase separations (H₂O:MeOH gradients). Reverse-phase high-performance liquid chromatography (RP-HPLC) was performed using a Waters 1525 binary pump, 2489 tunable UV/Vis detector (254 and 280 nm detection), and 2707 autosampler. For preparatory HPLC purification, samples were chromatographically separated using a Waters XSelect CSH C18 OBD prep column (part number 186005422, 130 Å pore size, 5 µm particle size, 19×150 mm), eluting with a H₂O:CH₃CN gradient solvent system. Linear gradients were run from either 100:0, 80:20, or 60:40 A:B to 0:100 A:B (A = 95:5 H₂O:CH₃CN, 0.05% TFA; B = 5:95 H₂O:CH₃CN, 0.05% TFA. Products from normal-phase separations were concentrated directly, and reverse-phase separations were concentrated, diluted with H₂O, frozen, and lyophilized. For primary compound purity analyses (HPLC-1), samples were chromatographically separated using a Waters XSelect CSH C18 column (part number 186005282, 130 Å pore size, 5 µm particle size, 3.0×150 mm), eluting with the above H₂O:CH₃CN gradient solvent systems. For secondary purity analyses (HPLC-2) of final test compounds, samples were chromatographically separated using a Waters XBridge C18 column (either part number 186003027, 130 Å pore size, 3.5 µm particle size, 3.0×100 mm, or part number 186003132, 130 Å pore size, 5.0 µm particle size, 3.0×100 mm), eluting with a H₂O:MeOH gradient solvent system. Linear gradients were run from either 100:0, 80:20, 60:40, or 20:80 A:B to 0:100 A:B (A = 95:5 H₂O:MeOH, 0.05% TFA; B = 5:95 H₂O:MeOH, 0.05% TFA). Test compounds were found to be >95% in purity from both RP-HPLC analyses. Mass spectrometry data were collected using either an Agilent analytical LC-MS at the IU Chemical Genomics Core Facility (CGCF), or a Thermo-Finnigan LTQ LC-MS in-lab. ¹H-NMR spectra were recorded on either Bruker 300 MHz or 500 MHz spectrometers. Chemical shifts are reported in parts per million and calibrated to the d₆-DMSO solvent peaks at 2.50 ppm. We previously synthesized compounds 1–14 (including 2a-m-o/m/p)²⁶ and re-synthesized where necessary due to stock depletion. Synthesis and characterization of intermediates 35–42 are presented below. General sulfonamide coupling steps are presented for analogs 15 and 16R-34R and 16L-34L below, with compound characterizations for each analog presented in the Supporting Information.

35: 5-Chloro-N-(4-hydroxy-3-nitrophenyl) thiophene-2-sulfonamide.

To a stirring mixture of 4-amino-2-nitrophenol (5.34 g, 24.6 mmol) in anhydrous CH₂Cl₂ (50 mL) was added 5-chlorothiophene-2-sulfonyl chloride (4.21 g, 27.3 mmol) and pyridine (2.40 mL, 29.4 mmol). The reaction was allowed to stir at room temperature for 18 h and was then diluted with hexanes and the precipitate was filtered, rinsed with 1 M HCl and water, collected, and dried to afford 35 as a reddish-brown solid (7.70 g, 94% yield). ¹H-NMR (300 MHz, d₆-DMSO) δ 10.98 (br s, 1H), 10.62 (br s, 1H), 7.60 (d, J = 2.7 Hz, 1H), 7.39 (d, J = 4.1 Hz, 1H), 7.30 (dd, J = 8.9, 2.7 Hz, 1H), 7.22 (d, J = 4.1 Hz, 1H), 7.10 (d, J = 8.9 Hz, 1H); MS (ESI) C₁₀H₆ClN₂O₅S₂ [M-H]^- m/z expected = 332.9, observed = 332.8; HPLC-1 = >99% (RT = 8.1 min).

36: N-(3-Amino-4-hydroxyphenyl)-5-chlorothiophene-2-sulfonamide.

Tin powder (8.73 g, 73.6 mmol) was added slowly to a stirring mixture of 35 (8.11 g, 24.2 mmol) in a 1:10 mixture of HCl:AcOH (24 mL). The reaction was allowed to stir at R.T. for 18 h, then diluted with EtOAc and H₂O, neutralized with NaHCO₃, and filtered. The filtrate was extracted with EtOAc and the organics dried over Na₂SO₄, filtered, and concentrated. The residue was diluted in a 50% mixture of DCM in hexanes and the precipitate was filtered, rinsed with hexanes, collected, and dried to afford 36 as a brown powder (6.08 g, 82% yield). ¹H-NMR (300 MHz, d₆-DMSO) δ 9.81 (br s, 1H), 8.99 (br s, 1H) 7.28 (d, J = 4.1 Hz, 1H), 7.18 (d, J = 4.1 Hz, 1H), 6.49 (d, J = 8.3 Hz, 1H), 6.41 (d, J = 2.5 Hz, 1H), 6.13 (dd, J = 8.3, 2.6 Hz, 1H), 4.64 (br s, 2H); MS (ESI) C₁₀H₁₀ClN₂O₃S₂ [M+H]⁺ m/z expected = 305.0, observed = 304.9; HPLC-1 = 97% (RT = 6.0 min).

37: 5-Chloro-N-(2-(4-nitrophenyl)benzo[d]oxazol-5-yl)thiophene-2-sulfonamide.

Compound 36 (2.94 g, 9.65 mmol), 4-nitrobenzaldehyde (2.03 g, 13.4 mmol), NaHCO₃ (2.08 g, 24.8 mmol), and Na₂SO₄ (3.35 g) were stirred in THF (40 mL) for 4 h at reflux (under Ar), then cooled to R.T. DDQ (2.85 g, 12.6 mmol) was then add portion-wise and the reaction was left to stir for 2 h, then filtered. The filtrate was extracted into EtOAc, rinsed with saturated NaHCO₃, 1 M HCl, and brine. The organics layer was dried over Na₂SO₄, filtered, and concentrated. The residue was diluted in a 25% mixture of DCM in hexanes and the precipitate was filtered, rinsed with hexanes, collected, and dried to afford 37 as a brown powder (4.01 g, 95% yield). ¹H-NMR (300 MHz, d₆-DMSO) δ 10.74 (s, 1H), 8.37–8.48 (m, 4H), 7.82 (d, J = 8.8 Hz, 1H), 7.60 (d, J = 1.9 Hz, 1H), 7.43 (d, J = 4.0 Hz, 1H), 7.25 (dd, J = 8.8, 2.1 Hz, 1H), 7.19 (d, J = 4.1 Hz, 1H); MS (ESI) C₁₇H₉ClN₃O₅S₂ [M-H]^- m/z expected = 434.0, observed = 433.8; HPLC-1 = 91% (RT = 7.7 min).

38: N-(2-(4-Aminophenyl)benzo[d]oxazol-5-yl)-5-chlorothiophene-2-sulfonamide.

Tin powder (2.74 g, mmol) was added slowly to a stirring mixture of 37 (2.43 g, 5.58 mmol) in a 1:10 mixture of HCl:AcOH (15 mL). The reaction was allowed to stir at R.T. for 18 h, then diluted with EtOAc and H₂O, neutralized with NaHCO₃, and filtered. The filtrate was extracted with EtOAc and the organics dried over Na₂SO₄, filtered, and concentrated. The residue was chromatographed over silica (hexanes:EtOAc gradient) and concentrated to afford 38 as an orange solid (936 mg, 41% yield). ¹H-NMR (300 MHz, d₆-DMSO) δ 10.54 (s, 1H), 7.78–7.86 (m, 2H), 7.60 (d, J = 8.7 Hz, 1H), 7.35–7.40 (m, 2H), 7.18 (d, J = 4.1 Hz, 1H), 7.04 (dd, J = 8.6, 2.2 Hz, 1H), 6.64–6.71 (m, 2H), 6.04 (br s, 2H); MS (ESI) C₁₇H₁₃ClN₃O₃S₂ [M+H]⁺ m/z expected = 406.0, observed = 405.9; HPLC-1 = 97% (RT = 7.6 min).

General sulfonamidation procedure for the synthesis of 16R-34R.

To a stirring mixture of compound 38 (1 eq.) in anhydrous CH₂Cl₂ (5 mL) was added the respective R² sulfonyl chloride (1.2 eq.) followed by anhydrous pyridine (1.2 eq.). The reaction was allowed to stir at room temperature for 18 h and was then chromatographed over silica (hexanes:EtOAc gradient) and concentrated. If necessary, the product was further purified by preparatory RP-HPLC (H₂O:CH₃CN gradient), concentrated, and lyophilized. Refer to the Supporting Information for individual compound characterization data.

39: Methyl 4-((5-chlorothiophene)-2-sulfonamido)benzoate.

To a stirring mixture of methyl-4-aminobenzoate (2.76 g, 18.3 mmol) in anhydrous CH₂Cl₂ (50 mL) was added 5-chlorothiophene-2-sulfonyl chloride (4.78 g, 22.0 mmol) and pyridine (1.80 mL, 22.1 mmol). The reaction was allowed to stir at room temperature for 18 h and was then diluted with hexanes and acidified with 1 M HCl. The precipitate was then filtered, rinsed with 1 M HCl and water, collected, and dried to afford 39 as a pinkish-orange solid (5.94 g, 98% yield). ¹H-NMR (300 MHz, d₆-DMSO) δ 11.16 (br s, 1H), 7.86–7.93 (m, 2H), 7.56 (d, J = 4.1 Hz, 1H), 7.25–7.31 (m, 2H), 7.22 (d, J = 4.1 Hz, 1H), 3.80 (s, 3H); HPLC-1 = 99% (RT = 6.2 min).

40: 4-((5-Chlorothiophene)-2-sulfonamido)benzoic acid.

LiOH•H₂O (7.02 g, 167 mmol) was added to a stirring mixture of 39 (5.67 g, 17.1 mmol) in THF (20 mL), MeOH (20 mL), and H₂O (20mL). The reaction was allowed to stir at room temperature for 2 days and was then diluted with 1M HCl. The precipitate was filtered, washed with H₂O, collected, and dried to afford 40 as an off-white solid (5.20 g, 96% yield). ¹H-NMR (300 MHz, d₆-DMSO) δ 12.85 (br s, 1H), 11.09 (br s, 1H), 7.83–7.90 (m, 2H), 7.54 (d, J = 4.1 Hz, 1H), 7.19–7.28 (m, 3H); HPLC-1 = >99% (RT = 6.7 min).

41: 5-Chloro-N-(4-(5-nitrobenzo[d]oxazol-2-yl)phenyl)thiophene-2-sulfonamide.

Compound 40 (2.96 g, 9.31 mmol) was stirred in SOCl₂ (10 mL) for 6 h at 60°C, then was concentrated to a solid. This was refluxed with 2-amino-4-nitrophenol (2.96 g, 19.2 mmol) and pyridine (1.50 mL, 18.4 mmol) in anhydrous CH₂Cl₂ (50 mL) for 6 h, then stirred at R.T. for 3 days. The reaction was then extracted into EtOAc and rinsed with 1 M HCl and brine, dried over Na₂SO₄, filtered, and concentrated. This amide intermediate was then refluxed with TsOH•H₂O (3.61 g, 19.0 mmol) in xylenes (50 mL) using a Dean-Stark apparatus to remove the residual H₂O. After 18 h, the reaction was cooled, the xylenes decanted off, the sludge extracted with EtOAc, and the combined organics dried over Na₂SO₄, filtered, and concentrated. Flash chromatographic purification over silica (hexanes:EtOAc gradient) afforded 41 as a peach solid (1.81 g, 45% yield). ¹H-NMR (300 MHz, d₆-DMSO) δ 11.25 (br s, 1H), 8.64 (br s, 1H), 8.32 (d, J = 8.8 Hz, 1H), 8.18 (d, J = 8.1 Hz, 2H), 8.01 (d, J = 8.7 Hz, 1H), 7.60 (br s, 1H), 7.41 (d, J = 8.2 Hz, 2H), 7.24 (br s, 1H); MS (ESI) C₁₇H₁₁ClN₃O₅S₂ [M+H]⁺ m/z expected = 436.0, observed = 436.1; HPLC-1 = 88% (RT = 5.5 min).

42: N-(4-(5-Aminobenzo[d]oxazol-2-yl)phenyl)-5-chlorothiophene-2-sulfonamide.

Tin powder (1.53 g, 12.9 mmol) was added slowly to a stirring mixture of 41 (1.78 g, 4.08 mmol) in a 1:7 mixture of HCl:AcOH (8 mL). The reaction was stirred and heated at 60°C for 3 h, then cooled to R.T., diluted with EtOAc and H₂O, neutralized with NaHCO₃, and filtered. The filtrate was extracted with EtOAc and the organics dried over Na₂SO₄, filtered, and concentrated. The residue was diluted in a 10% mixture of DCM in hexanes and the precipitate was filtered, rinsed with hexanes, collected, and dried to afford 42 as a pale-yellow powder (1.59 g, 96% yield). ¹H-NMR (300 MHz, d₆-DMSO) δ 8.01–8.09 (m, 2H), 7.55 (d, J = 4.1 Hz, 1H), 7.30–7.41 (m, 3H), 7.22 (d, J = 4.1 Hz, 1H), 6.83 (d, J = 2.0 Hz, 1H), 6.64 (dd, J = 8.7, 2.2 Hz, 1H); MS (ESI) C₁₇H₁₃ClN₃O₃S₂ [M+H]⁺ m/z expected = 406.0, observed = 406.1; HPLC-1 = 86% (RT = 6.8 min).

General sulfonamidation procedure for the synthesis of 16L-34L.

To a stirring mixture of compound 42 (1 eq.) in anhydrous CH₂Cl₂ (5 mL) was added the respective R² sulfonyl chloride (1.2 eq.) followed by anhydrous pyridine (1.2 eq.). The reaction was allowed to stir at room temperature for 18 h and was then chromatographed over silica (hexanes:EtOAc gradient) and concentrated. If necessary, the product was further purified by preparatory RP-HPLC (H₂O:CH₃CN gradient), concentrated, and lyophilized. Refer to the Supporting Information for individual compound synthesis and characterization data.

Cell information for compound evaluation.

The ESKAPE bacteria were purchased from the American Type Culture Collection (ATCC): E. faecium (Orla-Jensen) Schleifer and Kilpper-Balz strain NCTC 7171 (ATCC 19434); S. aureus subsp. aureus Rosenbach strain Seattle 1945 (ATCC 25923); Multi-drug resistant S. aureus (MRSA) subsp. aureus Rosenbach strain HPV107 (ATCC BAA-44); K. pneumonia, subsp. pneumoniae (Schroeter) Trevisan strain NCTC 9633 (ATCC 13883); A. baumannii Bouvet and Grimont strain 2208 (ATCC 19606); P. aeruginosa (Schroeter) Migula strain NCTC 10332 (ATCC 10145); E. cloacae, subsp. cloacae (Jordan) Hormaeche and Edwards strain CDC 442–68 (ATCC 13047). HEK 293 kidney cells (ATCC CRL-1573) and THLE-3 liver cells (ATCC CRL-11233) were used for compound toxicity assays.

Evaluation of compounds for inhibition of bacterial cell proliferation.

All compounds were evaluated for inhibiting the proliferation of each of the ESKAPE bacteria as per previously reported procedures with one minor modification: Liquid growth media was cation-adjusted by addition of 25 mg/L Ca²⁺ and 12.5 mg/L Mg²⁺ to reflect the free concentrations of these divalent cations in vivo.²⁷ Detailed protocols for bacterial growth assays are presented in the Supporting Information.

Protein Expression and purification.

E. coli GroEL and GroES, and human HSP60 and HSP10, were expressed and purified as previously reported.²⁵ Detailed protocols for these protein purifications are presented in the Supporting Information.

Evaluation of compounds in GroEL/ES and HSP60/10-mediated dMDH and dRho refolding assays.

All compounds were evaluated for inhibiting E. coli GroEL/ES and human HSP60/10mediated refolding of the denatured MDH and denatured Rho reporter enzymes as per previously reported procedures.²⁶ Detailed protocols for these assays are presented in the Supporting Information.

Counter-screening compounds for inhibition of native MDH and Rho enzymatic activity.

All compounds were counter-screened for inhibiting the enzymatic activity of the native MDH and native Rho reporter enzymes as per previously reported procedures.^{25, 26} Detailed protocols for the assays are presented in the Supporting Information.

Evaluation of inhibitors 20R and 20L binding to GroEL using Isothermal Titration Calorimetry (ITC).

All ITC experiments were performed on a MicroCal VP-ITC system. To minimize background heats of dilution, matched solutions of GroEL and compounds 20R or 20L were prepared in buffer containing 20 mM HEPES, pH 7.4, 50 mM KCl, 10 mM MgCl₂, and 1% DMSO. All buffers were freshly prepared from stock solutions on the day of use and degassed under vacuum. For each analytical run, a solution of 400 µM GroEL (monomer concentration) in the syringe was titrated into 50 µM solutions of either compound 20R or 20L in the sample cell (equilibrated at 20°C for 300 seconds): the first injection was 0.4 µL (discarded during analysis) and subsequent injections (2–20) were each 2.0 µL. Each injection was made over 4 s durations with 3-min intervals between subsequent injections. The reference power of the experiment was set to 6 µcal/s and a 5 s filter period. Sample stirring was set at 600 rpm for all measurements. Six independent replicates were conducted to analyze compound 20R binding, and five replicates for 20L. Raw thermograms from each replicate were analyzed independently using the ORIGIN instrument software (MicroCal Inc., version 7.0). In each experiment, the upper baseline was collected after the binding reaction was saturated, and the terminal 7–8 points in the linear region were fit to a straight line and subtracted from the entire data set to remove contributions from background heats of mixing and dilution. The ∆H was obtained by non-linear least-squares fitting of the plot of ∆H (mol of injectant)⁻¹ versus molar ratio using a single site binding model.³⁴ The ∆H, ∆S, K_d (K_A= 1/ K_d), and binding stoichiometries (n) in each titration were obtained, and ∆G was calculated using the Gibbs equation at a temperature of 20°C (293.15 K). Standard deviations were determined based on results from each of the replicate analyses.

Evaluation of compound effects on HEK 293 and THLE-3 cell viability.

Evaluation of compound cytotoxicities to HEK 293 kidney and THLE-3 liver cells were performed using Alamar Blue-based viability assays. HEK 293 cells were maintained in MEM medium (Corning Cellgro, 10–009 CV) supplemented with 10% FBS (Sigma, F2242). THLE-3 cells were maintained in Clonetics BEBM medium (Lonza, CC-3171) supplemented with the BEGM bullet kit (Lonza, CC-3170) and 10% FBS. All assays were carried out in 384-well plates (BRAND cell culture grade plates, 781980). Cells at 80% confluence were harvested and diluted in growth medium, then 45 µL of the HEK 293 cells (1,500 cells/well) or THLE-3 cells (1,500 cells/well) were dispensed per well, and plates were sealed with “Breathe Easy” oxygen permeable membranes (Diversified Biotech) and incubated at 37°C, 5% CO₂, for 24 h. The following day, 1 µL of the compound stocks (10 mM to 4.6 µM, 3-fold dilutions in DMSO) were pre-diluted by pin-transfer into 25 µL of the relevant growth mediums. Then, 15 µL aliquots of the diluted compounds were added to the cell assay plates to give inhibitor concentration ranges of 100 µM to 46 nM during the assay (final DMSO concentration of 0.1% was maintained during the assay). Plates were sealed with “Breathe Easy” oxygen permeable membranes and incubated for an additional 48 h at 37°C and 5% CO₂. The Alamar Blue reporter reagents were then added to a final concentration of 10%, the plates incubated at 37°C and 5% CO₂, and sample fluorescence (535 nm excitation, 590 nm emission) was read using a Molecular Devices FlexStation II 384-well plate reader (readings taken between 4–24 h of incubation so as to achieve signals in the 30–60% range for conversion of resazurin to resorufin). Cell viability was calculated as per vendor instructions (Thermo Fisher - Alamar Blue cell viability assay manual). Cytotoxicity CC₅₀ values for the test compounds were obtained by plotting the % resazurin reduction results in GraphPad Prism 6 and analyzing by non-linear regression using the log(inhibitor) vs. response (variable slope) equation. Results presented represent the averages of CC₅₀ values obtained from at least triplicate experiments.

Evaluation of MRSA resistance generation against lead inhibitors.

To identify potential resistance toward inhibitors, a liquid culture, 12-day serial passage assay was employed as per procedures reported by Kim. S et al., using the MRSA strain (ATCC BAA-44).^{35, 36} Briefly, MRSA bacteria were streaked onto a Tryptic Soy agar plate and grown overnight at 37°C. A fresh aliquot of Tryptic Soy Broth (TSB) was inoculated with a single bacterial colony and the cultures were grown overnight at 37°C with shaking (250 rpm). The overnight culture was then sub-cultured (1:5 dilution) into a fresh aliquot of media and grown at 37°C for 1 h with shaking, then diluted into fresh media to achieve a final OD₆₀₀ reading of 0.01. Aliquots of the diluted culture (200 µL) were dispensed to 96 well plates along with addition of 2 µL of test compounds in DMSO (20R, 28R, and vancomycin as a control). The inhibitor concentration range during the resistance assay was 100 µM to 48.8 nM (2-fold dilution series). Plates were sealed with “Breathe Easy” oxygen permeable membranes (Diversified Biotech) and left to incubate at 37°C without shaking (stagnant assay). OD₆₀₀ readings were taken at the 24 h time point to monitor for bacterial growth. A second set of baseline control plates were prepared analogously, without any bacteria added, to correct for possible compound absorbance and/or precipitation, as well as plate and media baseline effects. For inoculations on subsequent days, bacteria from the wells with the highest drug concentration where the OD₆₀₀ was >0.2 were diluted with fresh media to OD₆₀₀ of 0.01 and dispensed into a new 96-well plate. Test compounds were added, and the bacteria propagated again as described above. This procedure was repeated each day for a total of 12 days to observe changes in EC₅₀ values over each passage. EC₅₀ values for the test compounds were obtained by plotting the OD₆₀₀ results from each passage in GraphPad Prism 6 and analyzing by non-linear regression using the log(inhibitor) vs. response (variable slope) equation. Results presented represent the averages of EC₅₀ values obtained from at triplicate experiments.

Calculation of IC₅₀ / EC₅₀ / CC₅₀ values and statistical considerations.

All IC₅₀ / EC₅₀ / CC₅₀ results reported are averages of values determined from individual dose-response curves in replicate assays as follows: 1) Individual I/E/CC₅₀ values from replicate assays were first log-transformed and the average log(I/E/CC₅₀) values and standard deviations (SD) calculated; 2) Replicate log(I/E/CC₅₀) values were evaluated for outliers using the ROUT method in GraphPad Prism 6 (Q of 10%); and 3) Average I/E/CC₅₀ values were then backcalculated from the average log(IC₅₀) values. To compare statistical differences between log(I/E/CC₅₀) values between the matched R- and L-series of analogs, two-tailed, paired t-tests were performed using GraphPad Prism 6 (95% confidence level) and looking at differences between paired values (results are plotted in Figures S1-S5 in the Supporting Information). For compounds where log(I/E/CC₅₀) values were greater than the maximum compound concentrations tested (i.e. >1.8 and >2.0, or >63 and >100 µM, respectively), results were represented as 0.1 log units higher than the maximum concentrations tested (i.e. 1.9 and 2.1, or 79 and 126 µM, respectively), so as not to overly bias comparisons because of the unavailability of definitive values for these inactive compounds.

ABBREVIATIONS.

MDH

malate dehydrogenase

Rho

rhodanese

HPLC

high-performance liquid chromatography

¹H-NMR

Proton nuclear magnetic resonance

MS

Mass spectrometry

EtOH

ethanol

AcOH

acetic acid

EtOAc

ethyl acetate

MeOH

methanol

DMSO

dimethyl sulfoxide

IC₅₀

Inhibitory concentration for half-maximal signal in biochemical assay

EC₅₀

effective concentration for half-maximal signal in bacterial proliferation assays

CC₅₀

cytotoxicity concentration for half-maximal signal in human cell viability assays

ITC

isothermal titration calorimetry