nTZDpa (non-thiazolidinedione PPARγ partial agonist) derivatives retain antimicrobial activity without improving renal toxicity

Abstract

nTZDpa kills both growing and persister Staphylococcus aureus. However, due to toxicity liabilities, our lab conducted two structure–activity relationship (SAR) studies focusing on the core scaffold and obtained a new lead compound that was more potent and less hemolytic. Despite these favorable changes, the new lead displayed toxicity to renal cells. In this SAR study, we sought to improve this renal toxicity by derivatization via changes to sp³ character, the acid moiety, and halogenation of the aryl rings. Presented herein are our efforts that produced potent compounds albeit with no improvement to renal cell toxicity.

Antimicrobial compounds broadly range in their utility, from their use in hospitals and as disinfectants, to other applications, such as in the shipping industry as antifouling reagents or in oral healthcare as mouthwashes.^1–5 Despite the life-saving nature of antibiotics, over- and misuse of antibacterials culminated in resistance.¹ Furthermore, experts predict that broad spectrum use of antimicrobials during the COVID-19 pandemic will lead to a rise in resistance and a second “silent” pandemic.⁶

Previously, our labs discovered that a subset of FDA-approved compounds rescued Caenorhabditis elegans from resistant and persistent Staphylococcus aureus infection (Fig. 1A).^7–9 Persistence differs from resistance as the former have slowed or halted cellular functions, thereby limiting most antibiotics from targeting the bacterium due to the loss of the cellular function of their respective targets.^10–12

Fig. 1.

(A) Previously approved FDA-drugs found to rescue C. elegans in a high-throughput whole animal screen. (B) Our first and second rounds of SAR around the nTZDpa scaffold. (C) Probed hypotheses in our third-generation of SAR.

While promising, one initial hit, nTZDpa (non-thiazolidinedione PPARγ partial agonist), had limited utility due to its inherent hemolytic activity. This led our lab to conduct a series of structure–activity relationship (SAR) studies, which identified specific areas for improvement (Fig. 1B).⁷ A first-generation compound had decreased hemolytic properties; however, the modifications to the structure ablated persister killing ability.⁷ We next performed a second round of SAR studies and identified our second-generation lead, 1, which was more potent at killing both susceptible and persistent S. aureus but lacked hemolytic activity.⁷ However, 1 proved to be toxic to renal cells, which we aimed to address in the third round of SAR studies for this project presented herein.

We endeavored to construct compounds with improved toxicity profiles by answering lingering questions about our scaffold. Hashimoto previously showed that flat, planar compounds tend to stack well and then precipitate out of solution, typically in the kidneys, which could account for the renal toxicity that we observed.¹³ Therefore, we postulated that by increasing the sp³ character of our compounds we could diminish this effect. Furthermore, all the potent analogs that we had developed to date possessed an acid at the 2-position of the indole scaffold and were toxic to renal cells. However, some previously derived analogs lacking the acid had improved renal toxicity numbers albeit with lower potency. Therefore, we hypothesized that we could produce a less toxic, equally potent new lead compound via the increase in sp³ character or by interchanging the acid for other known bioisosteres (Fig. 1C).

First, we examined if the renal toxicity could be improved by modest increases to sp³ character to serve as an intermediate between our lead and our adamantyl analog. Using our previous synthetic route, which included a key Dieckmann condensation to form the indole core, we were able to obtain the methyl and t-butyl analogs in short order (Fig. 2A). Methylation of the resulting indole alcohol, alkylation of the 1-position of the indole, and deprotection of the methyl ether with boron tribromide yields the penultimate intermediate. A Chan-Lam coupling yields the phenolic ether, which following basic hydrolysis, provides analogs 5–6. Unfortunately, although the para-t-butyl analog maintained its potency, there was no improvement to its renal toxicity (Table 1).

Fig. 2.

(A) General route for accessing nTZDpa analogs. (B) Route to homologated adamantyl ether analog. (C) The two pathways under basic conditions in which hydrolysis could produce the observed product.

Table 1.

Minimum Inhibitory Concentrations (MIC) against S. aureus and Renal Toxicity studies for nTZDpa analogs.

Cmpd #	R1	R2	R3	n	MICa	Renal Toxicityb
1 (prev. lead)	p-I	p-CI	COOH	1	1	15%
5	p-I	p-Me	COOH	1	8	15%
6	p-I	p-t-Bu	COOH	1	1	10%
18	p-I	p-CI	COOH	0	16	20%
19	p-I	m, p-CI	COOH	1	4	15%
20	m-I	p-CI	COOH	1	1	10%
21	p-I	p-Br	COOH	1	2	15%
22	m-I	p-Br	COOH	1	8	15%
15	p-H	p-CI		1	8	20%
23	p-I	p-CI		1	>64	25%
17	p-H	p-CI		1	4	20%
24	p-I	p-CI		1	32	15%

^aconcentrations in μg/mL

^b% viability of cells at 64 μg/mL

Vancomycin and DMSO used as positive and negative controls, respectively. Vanco MIC: 1 μg/mL and 65% cells viable at 64 μg/mL. DMSO MIC: >64 μg/mL and 95% cells viable at 64 μg/mL.

We next sought to develop synthetic routes toward adamantyl derivatives to increase sp³ character (Fig. 2B). This was based on recent successes in our lab toward optimizing the antimicrobial activity of CD437.⁸ Toward this end, we only observed moderate success generating the homologated adamantyl derivative. We endeavored to access this adamantyl analog via Vilsmeier-Haack formylation of the commercially available indole ethyl ester.¹⁴ Next, alkylation of the nitrogen, reduction of the aldehyde with sodium borohydride, and alkylation would yield the adamantyl ether. Much to our dismay, we were never able to realize our desired analog due to loss of adamantanol during ethyl ester hydrolysis. Under basic conditions, we observed both hydrolysis of the adamantyl ether and the ethyl ester. With the homologated adamantyl ether analog, one could see two pathways that this hydrolysis goes awry. In the first pathway, hydroxide could potentially add in an S_N2 fashion at the methylene bridge between the indole core and the adamantyl ether to expel adamantanoxide (Fig. 2C, first pathway). In a second pathway, direct hydrolysis of the methyl ester could provide the acid, which under basic conditions would then be deprotonated. This carboxylate could add in to the σ* orbital of the adamantyl ether to form a 5-membered lactone and expel adamantanoxide. This lactone could then undergo typical basic hydrolysis to reform the acid and create a homologated alcohol at the 3-position of the indole core (Fig. 2C, second pathway). Because of the issues we encountered with basic hydrolysis conditions, we next turned to acidic conditions, however, these conditions also proved unsuccessful.

We next focused on introducing bioisosteres to replace the acid, with the goal of improving the renal toxicity profile (Fig. 3). We aimed to access a tetrazole analog, which would act as an acid bioisostere.¹⁵ In addition, we sought to make the amide-acid analog, which would homologate the acid farther away from the indole core. From our mechanism of action studies and previous SAR work, we knew that the acid was necessary for activity,⁷ but we wanted to see if the placement of the acid further away from the indole core would retain activity while introducing more sp³ character. Beginning with previous lead 1 as our starting material, we first transformed 1 to the amide, cyclized with TMS-N₃, and then eliminated the alkyl chain to generate the free tetrazoles, 15 and 23. With 17 and 24 (our amide-acid analogs), we first formed the amide and then deprotected the methyl ester to obtain the desired analog. Interestingly, the p-iodophenyl ether analogs were inactive in both series, while the non-functionalized aryl ether analogs were active hinting at more nuance in the mechanism of action than was initially expected.

Fig. 3.

Route to access tetrazole analogs (15, 23) and amide-acid analogs (17, 24).

In addition, to these compounds, we also synthesized five additional compounds by our general method presented in Fig. 2A. These compounds combined preferable aspects of previous analogs (i.e. location of heteroatoms) to test if these changes improved their renal toxicity (analogs 18–22) (Table 1). These compounds largely retained their potency; however, none of these compounds significantly improved the renal toxicity as all were toxic to >75% of renal cells at 64 μg/mL.

In conclusion, we synthesized and tested an eleven compound library of our third-generation of SAR studies to improve the toxicity profile of our lead membrane-targeting antimicrobial. These analogs explored modest changes to sp³ character, acid functionalization, and combinations of previous SAR leads. Our studies have further broadened our understanding of bioactivity with new two compounds being as potent as 1; unfortunately, however, we were unable to improve their renal toxicity profile thereby limiting future antibacterial applications to external use.