Tricepyridinium-inspired QACs yield potent antimicrobials and provide insight into QAC resistance

Abstract

Quaternary ammonium compounds (QACs) comprise a large class of surfactants, consumer products, and disinfectants. The recently-isolated QAC natural product tricepyridinium bromide displays potent inhibitory activity against S. aureus but due to its unique structure, its mechanism of action remains unclear. A concise synthetic route to access tricepyridinium analogs was thus designed and four N-alkyl compounds were generated in addition to the natural product. Biological analysis of these compounds revealed that they display remarkable selectivity towards clinically-relevant Gram-positive bacteria exceeding that of commercially-available QACs such as cetylpyridinium chloride (CPC) and benzalkonium chloride (BAC) while having little to no hemolytic activity. Molecular modeling studies revealed that tricepyridinium and shorter-chain N-alkyl analogs may preferentially bind to the QacR transcription factor leading to potential activation of the QAC resistance pathway found in MRSA; however, our newly synthesized analogs are able to overcome this liability.

A great deal of research has been done to study the properties and occurrence of quaternary ammonium compounds (QACs), cationic species that have found widespread use as disinfectants and other household supplies.^[1,2] The recent emergence of diseases such as COVID-19 has brought these common antimicrobial products into the public eye and made the identification of novel antimicrobials with low eukaryotic toxicity even more essential.^[3] Recent reports of potential toxicity and microbial resistance has motivated further research of new chemical scaffolds in this arena.^[4,5,6] One such compound, tricepyridinium bromide (1), isolated from the marine sponge Discodermia calyx and likely produced by its Escherichia coli symbiote, was recently shown to have promising inhibitory activity against the Gram-positive bacterium Staphylococcus aureus and the fungus Candida albicans.^[7]

Many QACs can attribute their antimicrobial activity to their amphipathic nature, allowing for permeabilization and eventual lysis of plasma membranes.^[8] This often results in selective inhibition of Gram-positive bacteria, as these species lack the additional outer membrane seen in Gram-negative bacteria. Tricepyridinium bromide is intriguing in that while it does show significant and selective inhibitory activity against Grampositive bacteria, it does so despite the absence of a nonpolar region of the molecule, which may suggest a separate mechanism of action. Alternatively, if this compound does indeed function via membrane permeabilization, there may be significant room for improvement via a medicinal chemistry approach. One alternate possibility is that this natural product, which can exist in a planar geometry, may function via nonspecific DNA intercalation, preventing transcription of essential genes and ultimately resulting in cell death, akin to the widely-used compound ethidium bromide (Figure 1).

Figure 1. Hypotheses for tricepyridinium mechanism of action.

Top, intercalation of tricepyridinium’s largely-planar, aromatic structure into the DNA of bacteria may prevent transcription of essential genes, akin to ethidium bromide. Bottom, tricepyridinium-mediated disruption of the plasma membrane of Gram-positive bacteria could lead to cell lysis and death, in a simila manner to cetylpyridinium chloride.

To probe the differences between these mechanistic possibilities, a small library of analogs was envisioned. The isolation report demonstrated that compounds lacking indole moieties at the 3 and 5 positions showed a complete loss of activity against S. aureus, therefore, we decided to focus our efforts on modification of the substituent on the quaternary center itself.^[7] Previous research in our lab has indicated that most membrane-perturbing QACs show the most potent activity with an alkyl chain consisting of 10–14 carbon atoms.^[9] We were therefore interested in synthesizing a decyl (8), dodecyl (9), and tetradecyl (10) analog for further analysis, in addition to the natural product (1) and the previously reported ethyl analog (7).

Attempts at replicating the previously-reported synthesis of tricepyridinium bromide proved unreliable, with low yields and reproducibility; this, coupled with prohibitively high costs of starting materials led us to devise an alternate synthetic route.^[7] Our improved route begins with a regioselective 3-iodination of indole (2), followed by protection of the nitrogen atom with a tert-butoxycarbonyl (Boc) moiety.^[10] Using the lithiation-borylation methodology established by Watson and Aggarwal, the iodine was replaced with a pinacolborane substituent to give 4.^[11] An iterative bis-Suzuki coupling to commercially available 3,5-dibromopyridine (5) furnished the carbon scaffold 6 in high yield on gram-scale. Acid-mediated removal of the Bocprotecting groups and subsequent Menshutkin reaction with various commercially-available alkyl bromides successfully generated the natural product and a subset of analogs, in a total of six synthetic steps and 26–33% overall yield from indole (Scheme 1).

Scheme 1.

Synthetic approach towards tricepyridinium and N-alkyl analogs.

We next turned our attention to biological investigation of our newly-synthesized analogs. As a preliminary measure of toxicity, a hemolysis assay was performed to determine the propensity for these compounds to lyse ovine erythrocytes.^[12] Compounds were serially-diluted and incubated with sheep’s blood for one hour, then centrifuged. The supernatant was isolated and optical density measurements were taken to determine the concentration of lysed cells present. In stark contrast with many of the QACs previously tested in our lab, tricepyridinium bromide and its related long-chain analogs do not show strong hemolytic activity, indicating a low level of eukaryotic toxicity (Table 1). Remarkably, these compounds are 8 to 16-fold less hemolytic than the structurally-similar compound CPC, which may suggest that these compounds act through a mechanism differing from membrane lysis.

Table 1.

Hemolytic activity of cetylpyridinium chloride (CPC) and N-alkyltricepyridinium compounds.

Compound	CPC	1	7	8	9	10
LC20 [μM]	16	125	250	125	125	125

We next tested our compounds’ ability to inhibit the growth of a panel of bacterial species consisting of Streptococcus mutans (UA159), Pseudomonas aeruginosa (PAO1), and three strains of Staphylococcus aureus (the methicillin-susceptible strain SH1000, a strain with community-acquired methicillin resistance USA300, and a strain with hospital-acquired methicillin resistance ATCC33591 which harbors QAC-resistance genes). The producer of tricepyridinium, Escherichia coli, was also tested (MC4100). These species were specifically chosen for their clinical relevance – S. mutans is the bacterium primarily responsible for the formation of dental cavities and is a known target of CPC;^[13] P. aeruginosa is an opportunistic pathogen commonly found in burn victims, cystic fibrosis patients, and on medical equipment;^[14,15] and S. aureus is the leading cause of soft tissue infections and methicillin resistant strains have been listed as a “serious threat” by the CDC.^[16,17] The latter two are eradicated from hospital settings through the use of high concentration solution of QAC cleaners (benzylalkonium chloride (BAC) and didecylalkonium chloride (DDAC)). The results of this assay are shown in Table 2.

Table 2.

Minimum inhibitory concentration data (μM) for all N-alkylpyr-idinium compounds, cetylpyridinium chloride (CPC), and benzalkonium chloride (BAC). All numbers shown are the average of three replicates

Compound	S. mutans UA159	S. aureus SH1000	S. aureus USA300	S. aureus ATCC33591	P. aeruginosa PAO1	E. coil MC4100
CPC	1	0.5	1	1	250	32
BAC	1	2	4	4	125	64
1	4	2	4	16	>250	>250
7	125	32	64	250	>250	>250
8	2	16	16	16	>250	>250
9	8	32	32	32	>250	>250
10	32	64	64	64	>250	>250

Both tricepyridinium bromide and the decyl analog 8 displayed inhibitory activity against S. mutans and S. aureus at low micromolecular concentrations comparable to those of CPC and BAC. As the alkyl chain length of these analogs increases past ten carbons, the micromolar activity progressively decreases against these bacteria. Further, all of the tested N-alkylpyridinium compounds displayed remarkable selectivity for Gram-positive bacteria surpassing that of CPC and BAC. This result is normally indicative of a mechanism of action involving disruption of the plasma membrane, considering the low hemolytic activity seen in Table 1 it is likely that these compounds are permeabilizing the membrane and not lysing akin to recent discoveries in our lab.^[8] Compound 8 shows an added level of selectivity, showing an 8 fold preference to inhibit the growth of S. mutans relative to S. aureus. This is in contrast to the currently-used oral antiseptic CPC, which shows no selectivity between the two species. As narrow-spectrum antibiotics containing quaternary nitrogens are virtually unknown, the precise in vivo activity of this compound may warrant further exploration in the future.

It is interesting to note that the natural product and ethyl analog 7 both showed approximately 8-fold differences in activity between MSSA and hospital-acquired MRSA. The ATCC33591 strain primarily utilizes the QacC efflux pump as a means of circumventing QAC-mediated cell death. The qacC gene is regulated by the transcription factor QacR, which in the absence of exogenous compounds is bound to QAC resistance efflux genes, preventing transcription. When a suitable QAC substrate is present, it is recognized and bound by QacR, leading to disassociation of the DNA-protein complex and subsequent upregulation of the QAC-efflux pumps including QacC.^[18] We were therefore eager to investigate whether 1 and 7 are possible substrates for QacR.

To examine binding to QacR, the molecular modelling software AutoDock Vina (embedded in PyRx) was employed.^[19] The binding pocket of QacR contains a number of tyrosine and tryptophan residues, facilitating π-stacking interactions with the aromatic moieties commonly seen in QACs.^[20] Previous work by our group has shown that monocationic QACs bearing aryl moieties are favorable substrates for resistance development.^[21]

Computational models reliably localized all of the tested compounds to this recognition region. Compounds 8, 9, and 10 were predicted to bind in virtually identical modes, allowing for stabilizing π-stacking interactions of the indole moieties with Y93 (slip-stacking geometry) and Y103 (T-shaped geometry), as well as hydrogen bonding of the indole protons to T89 and N157 and orientation of the long alkyl chains to the hydrophobic interior of the protein (Figure 2, left). Ethyl analog 7 had a computed binding mode that was rotated nearly 180 degrees in-plane with respect to compounds 8–10; this conformation retained hydrogen bonding to T89 but exchanged its N157 interaction for an additional π-stacking interaction (T-shaped geometry) with W61 (Figure 2, middle). The additional indole moiety of 1 allows it to adopt a third, lower energy conformation in which the molecule forfeits planarity to allow for an intramolecular pi-stacking between two of its indole moieties, as well as to W61, Y93, Y103, and Y123 (Figure 2, right). Hydrogen bonding to T161 is also predicted, though interactions with T89 and N157 are lost.

Figure 2. Computational modelling of pyridinium compounds bound to QacR.

Top, representative image of the three observed binding modes in QacR. Left, binding pocket of QacR with 8 (cyan), 9 (magenta), and 10 (yellow) bound. Center, binding pocket of QacR with 9 and 7 (green) bound. Right, binding pocket of QacR with 1 bound. All docking studies were performed with AutoDock Vina embedded in PyRx, using an exhaustiveness factor of 32. Data output was visualized using PyMOL

The relative predicted affinities place 1 as the strongest binder to QacR ( 12.8 kcal/mol), followed by 7 ( 11.1 kcal/ mol), 8 ( 11.0 kcal/mol), 9 ( 10.8 kcal/mol), and 10 (10.7 kcal/ mol,). This, in addition to the differences observed in binding modes, may account somewhat for the large change in the inhibitory activities of compounds 1 and 7 on SH1000 and ATCC33591, as if 1 and 7 bind to QacR with higher affinity, they are more likely to upregulate the efflux pump machinery found in ATCC33591. Possible explanations as to why compounds 8, 9, and 10 do not suffer from the same resistance issues may result from either stronger association with the plasma membrane or the inability of the efflux pumps to excrete the molecules. Unfortunately, as a transmembrane protein, QacC has not been reliably crystallized to date, and as such a similar computational approach to investigating relative binding affinity is not currently possible.

Through our work we have developed an efficient, reproducible synthetic route to tricepyridinium bromide and its N-alkyl analogs from commercially inexpensive starting materials. Biological assays determined that compounds 1 and 8 are remarkably effective at inhibiting the growth of clinicallyrelevant Gram-positive bacteria, including the oral pathogen S. mutans, while displaying considerably lower hemolytic activity than the currently-used QAC cetylpyridinium chloride. It remains unclear whether these compounds function vis permeabilization of the cell membrane, as suggested by the selectivity towards Gram-positive bacteria, or via another mechanism such as DNA intercalation, similar to known intercalators such as ethidium bromide. Future studies will elucidate this mechanism and allow for targeted design of additional analogs with the potential for use as clinical or household disinfectants, which will be reported in due course. Finally, and arguably most importantly, are the seminal discoveries disclosed herein toward a deeper understanding of QAC-resistance, which to date has been grossly underexplored. Our findings indicate that there may be some design principles that allow one to construct new QAC analogs that are less prone to resistance development or expulsion by existing mechanisms. Based on the tremendous selection pressures currently being put on bacteria during the ongoing pandemic, we expect these issues to be further exacerbated going forward.