Pyrazolopyrimidinones, a novel class of copper-dependent bactericidal antibiotics against multi-drug resistant S. aureus

Abstract

The treatment of methicillin-resistant Staphylococcus aureus (MRSA) infections poses a therapeutic challenge as even last resort drugs become increasingly ineffective. As the demand for antibiotics with novel modes of action is growing, new approaches are needed to probe a greater spectrum of antimicrobial activities for their potential efficacy against drug-resistant pathogens. The use of copper (Cu) by the innate immune system to mount an antimicrobial response against bacterial invaders has created an opportunity to explore a role for Cu in antimicrobial therapy. Here we describe pyrazolopyrimidinones (PZP) as novel copper-dependent inhibitors (CDI) of S. aureus. 5‐benzyl‐3‐(4‐chlorophenyl)‐2‐methyl‐4H,7H‐pyrazolo[1,5‐a]pyrimidin‐7‐one (PZP-915) showed potent bactericidal properties at sub-micromolar concentrations and activity against clinical MRSA isolates and biofilms cultures. This cupricidal activity is founded on the molecule’s ability to coordinate Cu and induce accumulation of Cu ions inside S. aureus cells. We demonstrate that exposure to 915+Cu led to an almost instantaneous collapse of the membrane potential which was accompanied by a complete depletion of cellular ATP, loss of cell-associated K⁺, a substantial gain of cell associated Na⁺, and an inability to control the influx of protons in slightly acidic medium, while the integrity of the cell membrane remained intact. These findings highlight PZP-915 as a novel membrane-directed metalloantibiotic against S. aureus that is likely to target a multiplicity of membrane associated protein functions rather than imposing physical damage to the membrane structure.

INTRODUCTION

Over the last 20 years, the search for broadly acting antibiotics has been largely unsuccessful ¹. Only 4 new antibiotics classes, the oxazolidinones, lipopeptides, fidaxomicin (a novel first in class macrolide with an 18-membered lactone ring ²), and diarylquinolines were approved for use in the United States since 1968 ^{3, 4}. This relatively low output of novel antimicrobial therapeutics in conjunction with a tendency to overuse clinically important antibiotics in both human and veterinary medicine has brought multi-drug resistant pathogens on the verge of being the dominating cause of hospital-acquired infectious diseases ^5–7. This concerning situation necessitates innovative approaches not only in drug discovery, but also in identifying new microbial vulnerabilities.

One strategy of Staphylococcus aureus and other pathogenic bacteria to evade host immunity is to quickly increase in numbers which aids, for example, in the establishment of persistence phenotypes ⁸, activation of cellular defense genes (e.g. quorum sensing ⁹), or in disseminating to and taking up residence at sites of lower immunological exposure (e.g. survival inside host cells or on abiotic prosthetic surfaces ^{10, 11}). While multiple factors contribute to bacterial fitness in vivo, access to essential nutrients is undoubtedly one of the most basic prerequisites for optimal growth. Transition metals (TMs) have been recognized as a critical nutritional resource for S. aureus and other pathogenic bacteria at the host pathogen interphase ^12–15. TMs participate in many biological processes by acting as catalytic or structural components of proteins and other large biomolecules ¹⁶. In proper biochemical environments, their coordination chemistry and catalytic properties can be harnessed ¹⁷. However, these catalytic qualities can be damaging if transition metals react uncontrollably outside their intended environments ¹⁷. As such, deprivation as well as superabundance of select transition metals negatively impacts bacterial fitness in vivo. The innate immune system has evolved to deploy both strategies to contain and weaken the invading pathogen ^17–19. The host attempts to limit the pathogens access to Fe, Zn, and Mn, while Cu, and in some instances Zn, were found to accumulate at high concentrations in close vicinity to the invading pathogen ²⁰. Intriguingly, bacterial Cu resistance mechanism were found to be required by many human pathogens for full virulence. This correlation gave rise to a new paradigm in nutritional immunity which describes the exposure of microbial pathogens to potentially toxic Cu concentrations as a broadly applicable defense strategy of the innate immune system. In consequence, pathogenic bacteria deploy and depend on strategies to defend against Cu overload in vivo ^{19, 21–25}.

To exploit the presence of antibacterial metals in vivo, we searched for molecules that potentiate the antibacterial properties of Cu ions. As Cu has rarely been considered in high-throughput antibiotic discovery screening, and due to its insufficient presence in standard screening media, compounds with Cu related antibacterial properties were largely overlooked in the past, and represent a largely untapped reservoir of unexplored activities ²⁶. The growing interest in the discovery and characterization of Cu-dependent inhibitors has revealed the copper-dependent anti-mycobacterial properties of disulfiram ^{27, 28}, 8-hydroxyquinolines ²⁹, and 1-hydroxy-5-R-pyridine-2(1H)-thiones ³⁰, the anti-staphylococcal activity of the NNSNs ²⁶, the broad spectrum anti-bacterial activity of bis-thiosemicarbazones ^{31, 32}, and the Cu-enhanced anti-fungal activity of Zn-pyrithione and other metal chelators ^33–35. Many of these molecules have potent efficacy on drug resistant microbes suggesting that they act by unique mechanisms.

While searching for novel inhibitors that require Cu for activity, we discovered a cluster of pyrazolopyrimidinone (PZP) derivatives with a wide range of potency against S. aureus. Further analysis revealed that their activity is founded on the scaffolds ability to coordinate Cu, increase the cellular Cu content, and impair essential membrane functions including the maintenance of the membrane potential and ATP production.

MATERIAL AND METHODS

Database searches.

PubChem database analysis was performed by searching for each individual PZP and refining the search to all reported bioactivities (as of August 16^th, 2018). The ChEMBL database was searched using the PZP core structure (Fig. 1) as input and by limiting the search to whole cell based bacterial assays with MICs of 50 μM or below.

Figure 1.

Molecular structures of PZP core and PZP compounds.

Bacterial strains, cell culture, growth conditions, and compounds.

The identity of S. aureus strain Newman was validated through whole genome sequencing. Methicillin-resistant S. aureus (MRSA) clinical isolates were characterized and obtained in a de-identified manner from UAB Laboratory Medicine. The antibiotic resistance profiles of the 5 MRSA strains are listed in supplementary table 1. Strains S. aureus JE2 (NR-46543, a USA300 derivative) and its transposon mutants defective in copA (NR-47104) or copX (NR-47133) were obtained from BEI resources and cultured in the presence of 5µg/ml erythromycin. S. aureus MNG3 is a derivative of S. aureus Newman with a G →T mutation in the promoter region of copA causing a slight Cu-resistance phenotype (Fig. S1A). S. aureus strain LAC carrying the pCG44 plasmid encoding the pH sensitive green fluorescent protein (pHluorin) were cultured in the presence of 25 μg/mL chloramphenicol ³⁶. Bacteria were routinely cultured in Mueller Hinton (MH) broth (Oxoid LTD) prior to transferring them into assay specific media. THP-1 monocytes were maintained in Roswell Park Memorial Institute (RPMI) media (Corning) supplemented with 10% fetal bovine serum (FBS, Atlanta Biologicals, Inc.) and 1x penicillin-streptomycin-glutamine and incubated at 37°C in the presence of 5% CO₂. In some instances, phenol red free RPMI medium was used to reduce background in absorbance and fluorescence readings. All PZP compounds were purchased from ChemBridge Corp. Throughout the paper, we refer to individual compounds using a three-digit identifier as defined in Figure 1 and supplementary table 2. PZP compounds were reconstituted and stored in anhydrous DMSO (Sigma-Aldrich) at a 10 mM concentration, aliquoted, and kept at −80°C until use. Compounds were sufficiently stable to withstand repeated freeze-thaw cycles without losing potency. Since the PZP-915 compound was only available in limited quantities from ChemBridge Corp., we re-synthesized this compound as specified in supplemental methods. Biological activity and potency of the re-synthesized PZP-915 batch were identical to the original material (data not shown). Copper sulfate, copper(II)chloride, zinc chloride and iron(II)chloride were purchased from Sigma-Aldrich. Metal salts were resuspended at 100 mM in purified water and kept at 4°C until use. Iron(II)Cl₂ was freshly prepared from powder before each experiment.

Determination of minimum inhibitory concentration (MIC) of PZP, metal specificity, and BCS assay.

For the assays, compounds and the indicated metals salts and/or bathocuproine disulfonic acid (BCS, Acros Organics) were mixed in RPMI medium without phenol red (Corning) and then distributed or further diluted in 96-well plates to reach final concentrations as indicated. Metal concentrations and BCS were kept constant from well to well, at 50 or 500 μM, respectively, while compound concentrations were titrated in 2-fold increments down from 10 μM unless indicated otherwise.

Bacteria harvested from exponentially growing cultures were first washed and diluted in RPMI and then distributed to each well for a final assay OD of 0.005 (~5×10⁶ cells/ml). Sterility controls and no compound controls containing either only media or media supplemented with 50 µM copper were present on each challenge plate. Plates were sealed with parafilm or placed in a sealed plastic bag to reduce evaporation and edge effects and kept at 37°C overnight after which OD₆₀₀ readings were taken on a plate reader (Cytation 3, BioTek). Values were blank subtracted and normalized to the appropriate no drug controls. MIC was defined as the lowest concentration at which growth was reduced by at least 90% in comparison to the untreated controls. Note that CuSO₄ and CuCl₂ boosted the activity of PZP-915 in an identical manner. The presence or formation of the BCS copper complex was monitored by measuring the absorption at 480 nm.

Toxicity testing on THP-1 monocyte cells.

Dilutions of PZP compounds were mixed with CuSO₄ in RPMI medium supplemented with 10% FBS and distributed in 96-well plates in which 20,000 THP-1 monocytes in RPMI with 10% FBS +PSG had been added the previous night. The final CuSO₄ concentration in each well was 25 μM. The copper concentration for toxicity testing was chosen to reflect the higher end of what cells would experience in human serum ³⁷. The highest PZP concentration tested was 40 μM. At this concentration, DMSO was present at 0.4% but had no impact on cell viability in our system (data not shown). The assay plate was incubated for 24 h in a humidified 37°C incubator with 5% CO₂. Following incubation, resazurin (Sigma-Aldrich) was added for a final concentration of 2.5 µg/mL. Fluorescence was measured after 13 hours on a plate reader at 530 nm excitation and 590 nm emission.

Time to death assay and bactericidal activity.

The procedure was similar to the MIC assay, except that small aliquots of 20 μL were taken from the wells at the indicated time intervals after initiation of the assay. These aliquots were diluted in 10-fold increments in phosphate buffered saline (PBS) with 0.05% Tween-20 (PBST) and plated on MH agar plates. Colonies were counted on the next day. In some instances, 5 μL aliquots taken from the wells were directly spotted on MH agar plates and incubated over night at 37°C to assess if cells had survived the treatment.

S. aureus ATP measurement.

The Promega BacTiter-Glo assay was used to measure the ATP content of PZP treated bacteria. Cells at an OD of 0.005 were exposed to treatment conditions in RPMI and then incubated for 1h at 37°C. 25 μL of treated cells and an equal volume of BacTiter-Glo reagent was added to a 384-well plate and shaken for 5 minutes. Luminescence was then read on a Cytation 3 spectrophotometer.

Membrane Potential and Membrane Disruption:

Shifts in membrane potential were measured using the fluorescent dye 3,3’-Diethyloxacarbocyanine iodide [DiOC₂(3)] (Molecular Probes). The proton ionophore CCCP (Fisher, Acros Organics) served as a control. Membrane permeability was assayed using the TO-PRO-3 iodide dye (Molecular Probes). All dyes and ionophores were made up as stock solutions in anhydrous dimethyl sulfoxide (DMSO; Sigma) and stored at −20 °C. The final DMSO concentration during treatment of staining was below 1% (vol/vol). Mid-log phase bacteria were treated as indicated in PBS, and the dyes were added (30 µM DiOC₂(3) and 500 nM TOPRO-3 iodide) 30 minutes before a reading was taken at the indicated timepoints. Stained samples were assessed by flow cytometry (Acuri C6 flow cytometer) following the recommendations of the manufactures. Flow cytometry results were analyzed using FlowJo V10 software package. Membrane potential was measured as a ratio of red to green fluorescence (FL-3/FL-1) from the DiOC₂(3) with CCCP as the positive control. Membrane integrity was measured as an increase in fluorescence (FL-4) from the TO-PRO-3 iodide with heat-killed bacteria as the positive control.

ICP-MS analysis.

S. aureus was grown in MH media to exponential phase (OD₆₀₀ ~ 0.8), harvested by centrifugation, washed twice, and then resuspended in RPMI. Prior treatment, the bacteria were adjusted to a final OD₆₀₀ of 1 in 40 mL. Treatment conditions were as follows: no-treatment control, 50, 100, 200, and 400 µM copper, 10 µM PZP-915, and 10 µM 915+50 µM copper. The samples were incubated at 37°C for 30 minutes with 180 rpm shaking. Cells were harvested by centrifugation (4000 rpm for 20 minutes at 4°C). The pellet was washed once in 10 mL of cold 10 mM HEPES+500 µM EDTA and then washed twice with chilled sterile water prior to drying the pellet overnight at 65°C. The pellet weight was determined the next day after which 500 µL of concentrated ICP-MS-grade nitric acid (~70%) (Optima; Fisher Scientific) was added for metal dissolution. After incubating again at 65°C overnight, 150 µL of the dissolved pellet was added to 5.1 mL metal free water (Optima, LC/MS grade). Elemental analysis was performed using ICP-MS (Agilent 7700 Series), and the amount of Cu, Zn, Mn, Fe, Na, K, and Mg in each sample was measured. The metal concentrations were determined using metal standards (Millipore) for a calibration curve, then normalized to the dry weight of the bacterial pellets. Samples were analyzed in triplicate, and error bars represent standard deviations. The Agilent instrument was equilibrated using an internal tuning solution containing Li, Y, Ce, Ti, and Co. The detection limit for Cu, Zn, Mn, Fe, Na, K and Mg was 1 ppb. Data were analyzed using Masshunter. In figure S8, Mn levels for the PZP-915 and 915+Cu treatments reached below the limit of detection, and so the bars represent the limit of detection.

Intracellular pH measurements.

A pHIuorin ³⁶ standard curve was generated in a similar manner to previously reported methods ³⁸. Briefly, cells were harvested and concentrated to give an OD₆₀₀ of 20 in CP buffer (0.1 M citric acid, 0.2 M Na₂HPO4 buffer at pH 7.5). 1 ml of this cell suspension was lysed by agitation at 4°C with silica using a cell disruptor (Bead Mill, Fischer Scientific; settings: 10×30s cycles at 5m/s). After centrifugation, the lysate was diluted 1:10 into CP buffers ranging from pH 5.0 and 8.5 in a black-bottomed 96-well plate. Fluorescence emission of the GFP reporter was measured at 510 nm with excitation at 390 nm and 470 nm (Cytation 3, Biotek). The emission ratios (390/470) were plotted and a polynomial equation was fitted to the datapoints to infer the intracellular pH of treated bacteria (Fig. S2).

For experimentation, S. aureus containing the pHIuorin plasmid were treated in RPMI that had its pH stabilized at pH 6.2 or 7.8 using 1 M potassium phosphate buffer. An OD of 0.05 bacteria were then exposed to the reported compounds as described in the MIC determination section for various time points. At the indicated times, fluorescence was measured as described above in a 96-well black bottomed plate. Changes in pH were determined by subtracting the pH value of the untreated sample from the treated samples so that a negative value indicates a drop of the intracellular pH in comparison to the control.

Mass spectrometry studies.

Mass spectrometry samples were prepared in a 30% LC/MS grade methanol, 0.1% formic acid solution. 10 µM PZP-915 and 10 µM copper(II) sulfate pentahydrate, zinc sulfate heptahydrate, or ferric chloride were mixed and diluted to 0.6 µM. A monolithic silicon microchip-based electrospray interface, the TriVersa NanoMate (Advion, Ithaca, NY), served as the source for electrospray ionization (ESI) before analysis on a dual linear quadrupole ion trap Orbitrap Velos Pro mass spectrometer (Thermo Fisher, San Jose, CA). Results from the mass spectrometry analysis were analyzed using the Thermo Xcalibur Qual Browser 2.2 software.

Determination of binding constants

Binding constants were determined by following a UV/Vis based approach (Rose-Drago method) as outlined elsewhere ^{39, 40}. Briefly, the spectra were recorded from various concentrations of PZP-915 added to 1.54×10⁻⁵ M of either CuBr, CuBr₂, FeCl₂ × 4 H₂O or ZnSO₄ dissolved in 25 uM HEPES buffer (pH=7.2). Data were fitted with a non-linear least squares fitting method.

Biofilm eradication assay.

Activity on preformed S. aureus biofilms was determined as previously described ⁴¹. Briefly, S. aureus strain Newman was washed twice in chelexed RPMI (Chelex100 Bio-Rad) (CRPMI), and then inoculated into MBEC (Innovotech) peg plates at OD 0.1. The plates were incubated at 37°C with 5% CO₂ for 48 hours. The plates were washed in PBS and then transferred to fresh 96-well plates containing compound dilutions in CRPMI and incubated at 37°C with 5% CO₂ for 24 hours. Thereafter, the MBEC peg plates were washed in PBS and then transferred to a fresh 96-well plate containing 16 µg/ml resazurin in CRPMI. Fluorescence was measured after 24 hours incubation at 37°C on a plate reader using 530 excitation/ 590 emission.

In vitro ADME Properties.

The distribution coefficient (Log D) was measured after PZP-915 was partitioned between n-octanol and phosphate buffered saline (PBS, pH 7.4). Kinetics solubility was estimated at pH 7.4 after 10 mM PZP-915 (in DMSO) was added to PBS buffer, and the residual DMSO was 1%. LC-MS method was used to quantify PZP-915 in Log D and solubility studies. Metabolic Stability of PZP-915 was evaluated in human liver microsomes. The final concentration of PZP-915 was 1 µM and the microsome protein concentration was 0.5 mg/mL, with 1.0 mM NADPH and 3.3 mM MgCl₂. The disappearance of PZP-915 was measured by LC-MS/MS.

RNA extraction and qRT-PCR.

RNA was extracted from log-phase S. aureus strain Newman and MNG3 after treatment with or without 50 µM CuSO₄ for 30 mins. Extraction was performed by first resuspending bacteria in RNAProtect (Qiagen), and then resuspending in Trizol. The cells were disrupted using BeadBug tubes (Sigma Aldrich) with 0.1 mm glass beads that were placed in a Beadmill (Fisher) and shaken at 5 m/s for 30 secs for a total of 10 cycles. The RNA was then purified using the Zymo Direct-zol RNA purification kit according to the manufacturer’s instructions.

Purified RNA was then used to make a cDNA library with the iScript cDNA synthesis kit (BioRad), according to manufacturer’s instructions. cDNA was diluted and then used to probe for target gene expression. Target expression was analyzed using SYBR-Green and specific primers for the endogenous control gene 16S (forward: GCTCGTGTCGTGAGATGTTG and reverse: CACCTTCCTCCGGTTTGTC) and the target gene copA (forward: TCATCGCAGTGGCAGATAC and reverse: GCAATGGCTTGAGCAGTG). Cycle thresholds were determined with the CFX Connect Real Time System. All qRT-PCR primers were validated for specificity by showing an amplicon with a single melting peak. Results were analyzed using the Δ ΔCT method for normalization. Data are expressed as fold change ± SD relative to control values.

Measuring GAPDH activity.

S. aureus was grown in MH media to exponential phase (OD₆₀₀ ~ 0.8), harvested by centrifugation, washed twice, and then resuspended in RPMI and adjusted to a final OD₆₀₀ of 1. Treatment conditions were as followed: no-treatment control, 50 µM copper, 10 µM PZP-915, and 10 µM 915+50 µM copper. The samples were incubated at 37°C for 30 minutes while shaking at 180 rpm. Cells were harvested by centrifugation (4,000 rpm for 10 minutes at 4°C). The pellet was washed once in 10 mL of cold 10 mM HEPES+500 µM EDTA and then washed twice with chilled, sterile PBS. Bacteria were then resuspended in GAPDH buffer provided in the BioVision GAPDH Activity Assay Kit. The cells were disrupted using BeadBug tubes (Sigma Aldrich) with 0.1 mm glass beads that were placed in a Beadmill (Fisher) and shaken at 5 m/s for 30 seconds for a total of 10 cycles. The beads were then pelleted by centrifugation at 13,000 rpm for 2 minutes after which the lysate was removed and placed on ice. GAPDH activity was then measured per the manufacturer’s instructions (Biovision).

Data analysis and chemical structure design.

Experiments were repeated at least twice with each experiment containing at least 3 technical replicates. Error bars represent standard deviation of technical replicates unless stated otherwise. Data were analyzed and graphed using Excel (Microsoft), GraphPad Prism 7 (GraphPad), Powerpoint (Microsoft) and Corel Photo Paint (Corel, Inc). Marvin Sketch was used for drawing and displaying chemical structures (MarvinSketch 6.1.4, 2013, ChemAxon (http://www.chemaxon.com)).

RESULTS

Identification of pyrazolopyrimidinones as a promising lead cluster for metalloantibiotics development against S. aureus.

A screen for inhibitors with copper-related modes of action and a subsequent hit cluster expansion effort identified pyrazolo[1,5-α] pyrimidinones (PZP) as a novel scaffold with copper-dependent activity against S. aureus. The cluster was comprised of 7 related compounds that demonstrated Cu-dependent action against S. aureus (Fig. 1), much like the classical Cu-dependent inhibitor GTSM (Table 1). However, all tested PZPs were less toxic to the human monocytic cell line THP-1 than GTSM. In the presence of Cu, the most promising compound, PZP-915, was 32-fold more potent against S. aureus than toxic to eukaryotic cells (in vitro therapeutic index, ^CuSI = 32, Table 1).

Table 1: Anti-staphylococcal activity and eukaryotic toxicity of PZPs.

MIC: Minimal inhibitory concentration against S. aureus, which is defined as 90% or greater inhibition of growth relative to untreated control; ^CuMIC: MIC in 50 μM copper supplemented medium; IC₉₀: Inhibitory concentration against THP-1 monocytic cells defined as 90% inhibition relative to untreated controls; ^CuIC₉₀: IC₉₀ in the presence of 25 μM copper (Cu); SI: selectivity index defined as ratio of IC₉₀/MIC; ^CuSI: defined as ^CuIC₉₀/^CuMIC.

PZP	S. aureus		THP1		SI	cuSI
	MIC	cuMIC	IC90	CuIC90
GTSM	[10]	0.3	[10]	0.15	-	0.5
915	5	0.6	[40]	20	>8	32
716	[40]	1.25	[40]	40	-	32
643	[40]	5	[40]	[40]	-	>8
832	[40]	5	[40]	20	-	4
284	[40]	10	[40]	40	-	4
720	10	2.5	[40]	20	>4	8
894	[40]	2.5	[40]	10	-	4

Compounds were examined for any previously reported bioactivities in the PubChem open chemistry database. All compounds had database records of being tested for potential bioactivities, but never for activity against S. aureus or in the context of Cu. Due to its sub-micromolar activity on S. aureus, low toxicity in mammalian cell culture, and acceptable ADME properties for an early lead (Table S3), PZP-915 was chosen as a representative for subsequent mode of action studies.

PZP-915 requires copper to exert sub-micromolar activity against S. aureus.

From the dose response curve (Fig. S3), we noticed that PZP-915 also inhibited S. aureus without added Cu with a MIC of 5 μM. To distinguish the Cu-dependent from the Cu-independent activity of PZP-915, the non-toxic, cell impermeable Cu chelator bathocuproine disulfonate (BCS) that possesses an exceptionally high Cu-coordination capacity was used to sequester free Cu ions in the media ⁴². In medium supplemented with both Cu and BCS, the activity of PZP-915 remained low and was about the same as in standard medium (Fig. S3) showing that PZP-915 indeed requires Cu for its sub-micromolar activity. The addition of BCS to standard medium in the absence of Cu supplementation had no impact on the activity of PZP-915 suggesting that trace amounts of Cu present in standard medium are not the source of PZP-915’s activity in the absence of Cu supplementation, and that there are at least two distinct types of activity for PZP-915, a copper-dependent and copper-independent (Fig. S3).

PZP-915 forms metal complexes with copper.

The sub-micromolar activity of PZP-915 in Cu supplemented medium would suggest the formation of Cu complexes with an activity that differs from that of its individual components. To produce direct evidence of complex formation, we utilized electrospray ionization mass spectrometry (ESI-MS) as previously described by Dragset et al. ⁴³. Representative m/z spectra of the free ligand and equimolar mixtures with CuSO₄ are shown in Fig. 2A. In those spectra, peaks were identified that correspond to the theoretical mass of either the free PZP-915 molecules (m/z = 350.1039), or PZP-915 associated with Na+ (m/z=372.0857) or K+ (m/z=388.0594) (Fig. 2A). These monovalent ions are considered impurities of the original compound powder and were also observed for the Fe-active PZP-derivative investigated by Dragset et al ⁴³. Importantly, two additional peaks appeared when Cu was added. Based on theoretical mass calculations, the peak at m/z = 412.0223 represents PZP-915 associated with Cu, while the m/z = 761.1210 peak correspond to Cu associated with two PZP-915 molecules (Fig. 2B). To further confirm the identity of the 1:2 (metal:ligand) complex with Cu, the isolated peak was subjected to collision induced fragmentation (CDI) mass spectrometry which confirmed the identity of the Cu(PZP-915)₂ complex (Fig. 2B inset).

Figure 2. Copper complex formation of PZP-915.

Electrospray mass spectra of a solution of PZP-915 (a) or a 1:1 mixture of PZP-915 and copper (b). The 5x represents a 5-fold magnification of the relative abundance in this area to highlight relevant peaks. Collision induced dissociation (CID) fragmentation of peak 761.12 from b was conducted to confirm the identity of the 2:1 complex (b inset). All the reagents were dissolved in 30% methanol and 0.1% formic acid.

Results from the Dragset study in combination with our data suggest that PZPs may interact with a variety of transition metal ions. In fact, this is a common property of many metal coordinating molecules which is reflected by the Irving-William series of metal complex stabilities ⁴⁴. We therefore investigated metal complex formation between PZP-915 and Zn or Fe, as they represent the two most abundant transition metal ions in human physiology and likely competitors of Cu ¹⁷. However, the m/z spectra of 915+Zn or 915+Fe did not show any peaks indicative of a metal complex (data not shown). As metal complexes may undergo rapid fragmentation under ionizing conditions in the utilized buffer, which would prevent their detection by MS, we used UV/Vis spectrometry as an alternative technique to investigate metal complex formation. This method revealed concentration dependent spectral shifts in all cases (Fig. S4A), which demonstrated that PZP-915 forms metal complexes with Cu, Fe, and Zn. Moreover, these shifts conclusively demonstrate the formation of 3^rd order complexes with all tested transition metals (Fig. S4B), with relatively similar binding constants (Table S4). Surprisingly, the binding constants for Zn, Fe and Cu were not in the expected order in reference to the Irving-Williams series of metal complex stabilities, where copper is predicted to bind the strongest to small ligands ⁴⁴. Typically, this is due to changes in binding chemistry, where copper can take on a tetragonal shape due to the Jahn-Teller effect ⁴⁵. However, a tetragonal complex was not observed. Rather, all three complexes were hexagonal in nature. One possible reason why the Cu(II)(PZP)₃ complex was not compliant with the Irving-Williams series could be because there may be a (partial) reduction of Cu(II) to Cu(I). The fact that a 3rd order complex of Cu with PZP-915 was not observed in the MS-spectrum, but in UV-Vis experiments, may be related to instabilities of that complex under ionizing conditions or the solvent used in these MS experiments.

It needs to be pointed out that in our MIC assay, the metal concentration exceeded the concentration of PZP-915 by 5 to 80-fold, depending on the respective concentration of PZP-915. Also, the individual binding constant for the 1^st order reaction of copper binding to PZP-915 (3.8×10⁵ Lmol⁻¹) is exceedingly higher than the binding constants for the 2^nd order (6.18 Lmol⁻¹) or 3^rd order (47.65 Lmol⁻¹) complexes (Table S5). Taken together, applicable thermodynamics will thus likely favor the formation of a 1^st order metal complex in our experiments. We therefore suspect these 1:1 complexes to be the actual biologically active entity in our assays as we have previously proposed for the metal-dependent anti-tuberculosis compound 8-hydroxiquinoline ²⁹. In addition, the availability of Cu to engage in metal complex formation with PZP-915 may be influenced by other medium constituents, such as free amino acids, which can sequester Cu ions or reduce Cu(II) into Cu(I), and therefore also influence the dynamics of the equilibrium or even change the identity of the Cu complex. Because of the uncertainty of which metal complex is present or active, we will refer to the Cu complex as 915+Cu throughout the manuscript.

Transition metal specificity of the antimicrobial properties of PZP-915.

Since the coordination studies indicate metal complex formation with a variety of transition metal ions, we tested PZP-915 for boosted activities in the presence of Fe or Zn. Fe had no effect, while Zn improved the activity of PZP-915 in comparison to standard medium by 4-fold (Fig. 3A). Cu was by far the most potent enhancer of PZP-915’s anti-staphylococcal activity, even when Fe or Zn where added in addition to Cu (Fig. 3B). Overall, the boosting efficacy exerted by Cu and Zn, or lack thereof in the case of Fe, is consistent with the notion that PZP-915 enhances the natural antibacterial properties of the associated metal ion. Iron is the least toxic of the three metals and generally well tolerated by S. aureus with no toxicity observed up to 3.2 mM, while the MICs for Cu and Zn were determined to be 120 μM and 3.2 mM (Fig. S1B & S5). Accordingly, only Cu, and to a lesser extent Zn, boosted the activity of PZP-915 (Fig. 3A).

Figure 3. Impact of transition metals on PZP-915’s activity against S. aureus.

a) Sensitivity of S. aureus to PZP-915 in zinc chloride (Zn, 50 μM), iron chloride (Fe, 50 μM) or copper chloride (Cu, 50 μM) supplemented medium. b) Activity of PZP-915 with 50 μM Cu in combination with Zn or Fe supplementation (50 μM). c) Killing of S. aureus by PZP-915 (0, 0.15, 0.3, 0.6, 1.25, 2.5, 5 μM) in the absence or presence of 50 μM Cu, Fe, or Zn. The white bars represent controls of each condition without PZP-915 added. The PZP-915 concentration is indicated by different shades of grey and increase by two-fold between shades with the lightest representing 0.31 μM and the darkest (black) representing 5 μM. d) Time to kill kinetic of S. aureus exposed to varying concentrations of PZP-915 (0.078 to 1.25 μM) in the absence or presence of 50 μM Cu.

Copper complexation converts the activity of PZP-915 from bacteriostatic to bactericidal.

To distinguish between bactericidal and bacteriostatic modes of inhibition by PZP-915, the number of recoverable viable bacteria were determined after 24h of PZP-915 treatment and then plating on MH-agar. The intrinsic Cu-independent activity of PZP-915 in standard medium or its Zn-enhanced action were accompanied by a lower CFU count in comparison to the untreated control (Fig. 3C), but these numbers were still relatively similar to the input cell count at initiation of the experiment (5×10⁶ cells/ml corresponding to an OD₆₀₀ of 0.005), suggesting a predominantly bacteriostatic mode of action for PZP-915 in the absence of Cu or the presence of Zn. In contrast to these bacteriostatic activities of PZP-915, no viable cells could be recovered when Cu was added, which is consistent with a bactericidal activity (Fig. 3C). In a dose-matrix experiment where Cu was titrated against PZP-915, as expected, an inverse correlation was observed where more PZP-915 required less Cu for killing and vice versa. For example, a 10 µM PZP-915 concentration required only 1.25 µM Cu to exert potent antibacterial properties, whereas 0.6 µM PZP-915 required 50 µM Cu to display a similar potency (Fig. S6). In addition, the time to achieve complete inhibition of growth was concentration dependent and took only 2 hours for 0.6 µM PZP-915 in the presence of 50 µM Cu (Fig. 3D). These findings highlight Cu as an essential facilitator of the bactericidal properties of PZP-915 and establish this compound as a fast-acting cupricide against S. aureus.

PZP-915 activity against MRSA and S. aureus biofilms.

In S. aureus, multi-drug resistance phenotypes can manifest through the acquisition of genetic traits (i.e. MRSA) and/or the formation of biofilms (i.e. periprosthetic orthopedic infections ⁴⁶). A novel antibiotic is therefore expected to show efficacy against drug resistant strains and phenotypes ⁴⁶. As seen in Figure 4A, PZP-915 was as efficacious on biofilms as on planktonic cells and complete eradication of S. aureus in biofilms was accomplished at a PZP-915 concentration of 0.6 µM.

Figure 4. Growth in biofilms or multi-drug resistance phenotypes of clinically isolated MRSA does not protect against 915+Cu toxicity.

a) Susceptibility of preformed S. aureus biofilms to PZP-915. Survival was quantified by the metabolic indicator resazurin and normalized to the untreated control. b) Sensitivity of 5 clinically isolated, methicillin resistant S. aureus strains to PZP-915 in the presence of 50 μM Cu.

Similarly, PZP-915 was highly potent against MRSA. When tested against a panel of 5 clinically isolated MRSA strains with differing drug resistance profiles (Table S1), no difference in the MIC of PZP-915 activity against the MRSA strains in comparison to the non-resistant laboratory strain S. aureus Newman was observed (Fig. 4B), demonstrating that PZP-915’s activity is not impaired by disease-relevant, multi-drug resistance phenotypes of S. aureus.

PZP-915 enhances copper accumulation in S. aureus.

After demonstrating that PZP-915 forms complexes with Cu ions and Cu is required for the bactericidal activity of PZP-915, we investigated the impact of 915+Cu on the cell associated metal content of S. aureus. Cu levels increased by 65.7-fold in 915+Cu treated cells relative to the untreated control or PZP-915 only treatment (Fig. 5A). Equimolar Cu only treatment (50 μM) increased the cell associated Cu content by a comparatively small 4.9-fold relative to the untreated control (Fig. 5A). For comparison, in the absence of PZP-915, an 8-fold higher extracellular Cu concentration (400 μM) was needed to obtain an equally pronounced rise in the cell associated Cu content compared to 915+Cu treatment. These data are consistent with the idea that PZP-915 acts as a Cu-ionophore.

Figure 5. PZP-915 alters cytoplasmic copper levels.

a) Copper (Cu) content of treated S. aureus as determined by ICP-MS. Treatments are as followed: untreated, 50, 100, 200, and 400 μM Cu, 10 μM PZP-915, and 10 μM 915+50 μM Cu. b and c) Transposon mutants with a defective copA (b) or copX gene (c) were tested for sensitivity to 915 with and without 50 μM Cu. Data for wildtype (wt) in b and c are identical. d) A copA overexpressing strain, MNG3, was tested for sensitivity to 915+50 µM Cu.

Due to the lack of reliable chemical probes that can distinguish between intracellular and cell wall-associated Cu ions, we used a genetic approach to test if mobile Cu ions accumulate in the cytoplasm. S. aureus JE2 mainly uses two Cu-specific efflux pumps, CopA and CopX, to remove excess Cu ions from its cytoplasm, ^47–49. To determine if these Cu-efflux pumps play a role in protecting S. aureus against PZP-915-mediated Cu-intoxication, transposon mutants of each efflux pump were tested for increased sensitivity ⁴⁹. The previously published general Cu-sensitivity phenotype of the ΔcopX mutant ⁴⁷ reproduced well in our assay system, but different from this previous study, we also detected a copper sensitivity phenotype for the ΔcopA mutant (Fig. S1B). That we noted these phenotypes is probably a benefit of our adapted dose-response curve assay were the 10 μM increments offer a greater resolution than the traditional 2-fold concentration increments. Importantly, both mutants displayed increased sensitivity to 915+Cu (Fig. 5B & 5C). If the lack of Cu-specific efflux pumps enhanced sensitivity, we reasoned that overexpression of CopA should increase resistance. For this purpose, a mutant of S. aureus Newman, where the copA gene is constitutively overexpressed due to a 1-bp mutation in the binding site of the Cu-sensing transcriptional repressor CsoR, which represses copA expression in low Cu environments, was tested for increased resistance (Fig. S1C). As predicted, this mutant showed a Cu resistance phenotype and was also mildly (2- to 4-fold) resistant to 915+Cu (Fig. S1D & 5D) relative to wild-type. Together, these data suggest that the intracellular accumulation of Cu ions is an important component of PZP-915’s Cu-dependent mode of action and that Cu efflux pumps help shape baseline tolerance levels to 915+Cu (and Cu only) treatment in S. aureus.

PZP-915 impairs energy metabolism and metal ion homeostasis without loss of membrane integrity.

To identify potential targets that are impacted by PZP-915, we decided to focus our investigations on the cell membrane as the first point of contact between the inhibitor and the bacterial cell. One important function of the membrane is to establish and maintain an electrochemical gradient which powers many transport processes across the cytoplasmic membrane and drives ATP-synthesis via proton-driven ATP-synthase complexes ⁵⁰. The electrochemical gradient consists of an electrical component, also known as membrane potential Δψ, and various ionic gradients with the most important being the gradients of protons (H+), potassium ions (K⁺), and sodium ions (Na⁺). To test whether 915+Cu has an impact on Δψ, we used the chemical probe DiOC₂(3) ⁵¹. In response to 915+Cu treatment, a decrease in DiOC₂(3)’s red/green fluorescence ratio within 30 min of treatment was observed indicating membrane depolarization (Fig. 6A). Interestingly, treatment with an inhibitory concentration of PZP-915 (10 μM) in the absence of Cu also decreased the membrane potential, although to a much lesser extent, while the equimolar Cu mono-treatment had no impact (Fig. 6A). To exclude the possibility that membrane depolarization is a consequence of membrane rupture, treated and untreated cells were stained with TO-PRO-3 (Fig. 6B), a dye commonly used to assess membrane damage of bacterial cells ⁵². While heat-killed cells showed high TO-PRO-3 fluorescence intensity as expected, treated cells, including the one’s exposed to 915+Cu, remained TO-PRO-3 negative after a 30 min exposure (Fig. 6B). This remained true after 8 hours of treatment, demonstrating that even after extended treatment periods, 915+Cu did not impair the general barrier function of the plasma membrane (Fig. S7). The deteriorating membrane potential was therefore unlikely the consequence of a ruptured membrane which points towards dysregulation of cellular ionic gradients as the likely mode of action for 915+Cu.

Figure 6. Effect of PZP-915 treatment on membrane potential, membrane integrity, and ion flux.

a) Changes in membrane potential of S. aureus Newman in response to 50 μM Cu, 10 μM PZP-915, and 10 μM 915+50 μM Cu treatment were assessed using the red/green fluorescence ratio of the fluorescent dye 3,3’-Diethyloxacarbocyanine iodide (DiOC₂(3)). CCCP served as a positive control. Error bars represent the coefficient of variation for the ratios. b) Membrane integrity staining of treated S. aureus Newman using the fluorescent indicator probe TO-PRO-3. Heat killed S. aureus cells served as a positive staining control and were spiked with a small number of live cells prior to staining for control purposes. The data are representative of a 30 min treatment period. Treatments were 10 μM 915+50 μM Cu (blue), 10 μM PZP-915 (green), 50 μM Cu (yellow), untreated (orange), and Heat-killed (HK, grey). Figure S5 shows the corresponding data after an 8h treatment time. Na⁺ (c) and K⁺ (d) content of treated S. aureus cells as determined by ICP-MS. Treatments are as followed: untreated (ctrl, white), 50 μM Cu (light grey), 10 μM PZP-915 (dark grey), and 10 μM 915+50 μM Cu (black). e) Changes over time of the intracellular pH in S. aureus cells treated with 50 μM Cu, 10 μM PZP-915, or 10 μM 915+50 μM Cu in medium adjusted to e) pH 7.8 or f) pH 6.2. CCCP was added as a positive control. Values shown represent the calculated pH subtracted from the pH of the untreated control.

To further substantiate these findings, we investigated if other intracellular mono and divalent metal ion concentrations would be impacted by PZP-915. ICP-MS analysis revealed substantial shifts in the Na⁺ and K⁺ content of 915+Cu treated cells (Fig. 6C & 6D), while other low (Mn, Fe, and Zn), or high abundance metal ions (Ca and Mg) showed only minor fluctuations (Fig. S8A), which were seen under both, PZP-915 alone and 915+Cu conditions. As Na⁺ and K⁺ ions have important functions in establishing and maintaining a membrane potential, their altered concentrations in the presence of 915+Cu are consistent with a collapsing membrane potential. In such a scenario, we would anticipate these ions to equilibrate along their concentration gradients across the cellular membrane. This was indeed what was observed (Fig. 6C & 6D).

When comparing the changes in the sum of all cations measured of the 915+Cu treated cells to the control cells, the cell associated cation concentration was increased by approximately 10.5% (Fig. S8B). Of these changes in cation concentrations, Cu caused the largest change in terms of fold-change and third largest in terms of absolute amounts (Figs. S8A & S8C). In the untreated samples, Cu only made up 0.04% of the total cations and was the least abundant of the cations measured, whereas after 915+Cu treatment, Cu drastically increased to make up 2.3% of the total cations and became the fourth most abundant cation (Fig. S8C). These shifts in metal concentrations coincide with a collapsing membrane potential, but whether they are the cause or consequence of this collapse requires further investigation.

Because membrane potential and proton gradient are tightly intertwined, as they both define the proton motive force (PMF) that drives ATP-synthesis ⁵⁰, we next investigated a potential impact of 915+Cu on cellular proton homeostasis. For this purpose we used an S. aureus strain expressing a pH sensitive green-fluorescent protein ³⁶. When the assay was performed in medium at pH ~7.8, the intracellular pH dropped by ~0.2 units in response to treatment (Fig. 6E). As previously noted ⁵³, this relatively small shift is indicative of a dissipating proton gradient under conditions where the extracellular and intracellular pH are relatively similar, as is the case here. Under these conditions, the PMF is mainly driven by the electrical potential where positively charged protons are attracted by the negatively charged interior of the cell. Once the electrical membrane potential collapses in the presence of 915+Cu (Fig. 6A), very few protons, if any, are expected to enter the intracellular space. Given that under the experimental conditions the outside pH (medium pH 7.8) and the intracellular pH in S. aureus is pH 7.5–8.0 ⁵⁴, any drop in intracellular pH would be expected to be small (Fig. 6D). Therefore, to demonstrate that in 915+Cu treated cells protons begin to diffuse into the cell, we repeated the experiment in mildly acidic medium (pH 6.2). Here, the intracellular pH dropped gradually within the first 3 hours in 915+Cu exposed cells, while CCCP (a proton specific ionophore) treated cells, as expected, experienced an almost instantaneous pH drop (Fig. 6E). PZP-915 or Cu single treatments had no or only a minor impact on the intracellular pH relative to the untreated controls, showing that in the case of 915+Cu treatment, cells lose their ability to regulate the proton flux.

As 915+Cu treatment resulted in the collapse of both the membrane potential and proton gradient, and therefore the proton motive force, we expected that 915+Cu would lower the cellular ATP levels. Indeed, ATP concentration declined rapidly within 60 min of 915+Cu treatment, and to a much greater degree than for cells treated with the positive control CCCP, which specifically uncouples the proton gradient (Fig. 7A). However, despite the observed lower ATP levels, most 915+Cu treated cells remained viable for about 2 hours post exposure (Fig. 3D), suggesting that cells initially had a limited ability to recover from these collapsing ATP levels, if treatment was interrupted. In comparison, high concentrations of Cu also reduced ATP levels after 1 hour of treatment, but not even 800 µM Cu resulted in the same level of ATP depletion observed in the 915+Cu treated cells (Fig. 7B). However, the killing kinetic for high levels of Cu is similar to 915+Cu, with complete bacterial eradication observed after 2 hours (Fig. S9).

Figure 7. ATP levels and GAPDH activity of treated S. aureus cells.

a) Intracellular ATP levels of S. aureus Newman after 1h exposure to PZP-915 in standard or 50 μM Cu supplemented media. Values are normalized to the untreated controls of each series. Carbonyl cyanide m-chlorophenyl hydrazine (CCCP) served as a positive control for this assay. b) ATP was measured after 1 hour of treatment. Values were normalized to the untreated control. Treatments are as followed: untreated, 50, 100, 200, 400, and 800 μM Cu, 10 μM PZP-915, and 10 μM 915+50 μM Cu. c) S. aureus was treated for 30 minutes, washed to remove extracellular copper, and then lysed. GAPDH activity was measured from the bacterial lysates. Treatments are as followed: control, 50 µM Cu, 10 µM PZP-915, and 10 µM 915+50 µM Cu.

Given the observed drop in ATP concentration in the presence of 915+Cu, and based on a recent publication describing an inhibitory effect of sub-inhibitory concentrations of Cu on GAPDH activity in S. aureus ⁵⁵, we explored whether 915+Cu would potentially target GAPDH. To investigate a PZP-915 effect on GAPDH activity, bacteria were treated for 30 minutes with Cu, PZP-915, 915+Cu, or left untreated. GAPDH activity was measured from the bacterial lysates. GAPDH activity in cell lysates from Cu and 915+Cu treated cells were below detection, whereas PZP-915 only treatment did not alter GAPDH activity relative to the untreated control (Fig. 7C). These results were consistent with previously published data on the effect of sublethal Cu concentrations on GAPDH activity ⁵⁵. However, the results do not suggest that the bactericidal activity of 915+Cu is due to a specific inhibitory effect on GAPDH activity, as 50 µM Cu treatment also reduced GAPDH activity, but did not significantly affect ATP levels, did not disrupt the membrane potential, and did not kill the bacteria (Fig 6A & 7B). Taken together, our data suggest that in S. aureus exposure to 915+Cu interferes broadly with bioenergetic processes, which leads to a catastrophic depletion of cellular ATP levels.

DISCUSSION

PZP-915 emerged from our drug-discovery and development pipeline as a promising early chemical lead with sub-micromolar activity against MRSA, activity against S. aureus biofilms, low toxicity against mammalian cells, and encouraging in vitro ADME properties. As such, its drug-like qualities will likely set the benchmarks for future lead optimization and development efforts of the PZP scaffold.

PZPs are another promising example of copper-dependent inhibitors (CDIs), a group so far comprised of novel thiourea-based inhibitors dubbed the NNSNs ²⁶, 1-hydroxy-5-R-pyridine-2(1H)-thiones ³⁰, and Cu-dependent antimicrobial peptides such as DAB-10, piscidin 1, and ovispirin-3 ^56–58. These inhibitors improve upon the first generation of CDIs, which include GTSM, disulfiram, and 8-hydroxiquinolines, and show a higher Cu-specific in vitro selectivity towards bacteria over eukaryotic cells.

One potential future of these compounds lies in the developing field of nutritional immunity which focuses on the antimicrobial and nutritional properties of metal ions during an infection. Several in vitro and in vivo infection models have demonstrated that Cu ions accumulate at the site of infection leaving microbes to depend on Cu-resistance mechanisms for full virulence ^{21, 22, 59–61}. For example, macrophages are thought to exploit Cu’s Fenton-like chemistry to kill phagocytosed bacteria by presumably exacerbating oxidative stress levels in acidified phagolysosomes ^{62, 63}. In addition, Cu-ions are also known to displace iron from iron sulfur cluster proteins in E. coli ⁶⁴, to inhibit heme biosynthesis in Neisseria gonorrhoeae ⁶⁵ and to impair aerobic nucleotide synthesis in Streptococcus pneumoniae ^{65, 66}. In this ongoing battle between the innate immune system and invading pathogens, CDIs may have a special niche as they could potentiate Cu’s antimicrobial properties at the site of infection. As such, CDI’s have created opportunities to probe for specific synergies between antibiotics and Cu-dependent innate immune functions and therefore deliver an intriguing opportunity for antimicrobial drug-discovery and development.

CDIs likely depart from the one-drug-one-target paradigm, which has guided antimicrobial drug discovery for decades ⁶⁷. This is especially true for PZP-915 which combines at least 3 distinguishable components, including Cu-dependent bactericidal activity, metal-independent bacteriostatic activity, and bacteriostatic Zn-dependent activity (Fig. 3C). It is possible that these activities could exploit or synergistically enhance Cu- and Zn-dependent innate immune functions, as previously shown for 8-hydroxyquinoline derivatives in macrophage in vitro infection models of Mycobacterium tuberculosis or Cryptococcus neoformans ^{29, 35}.

This spectrum of inhibitory properties is further enhanced by the prospect of Cu itself acting on multiple targets which would create an even richer polypharmacology against S. aureus. This notion is supported by our finding that loss of membrane potential and ATP depletion were not all exclusive consequences of 915+Cu, but also, at least in part, of exposure to excessive Cu-concentrations. These alterations in cell homeostasis may be due to targeting of membrane associated properties or may involve yet unrecognized protein targets or functions.

Simultaneously addressing multiple targets or exhibiting several activities could help overcome a well-recognized shortcoming of many single target antibiotics, which is that they tend to lose clinical efficacy relatively quickly due to the microbe’s inherent ability to develop or acquire resistance ⁶⁸. Examples are the sulfonamide Prontosil ⁶⁸ or the diarylquinoline Bedaquiline. The latter is a novel second line antibiotic for the treatment of drug-resistant M. tuberculosis ⁶⁹, where resistance emerged only 3 years after FDA-approval ⁷⁰. While combination therapy, as often advocated by antimicrobial stewardship efforts, could slow down resistance development, it would require the parallel development of at least two new antibiotics, which in itself poses a challenge, given the marginal success in developing even one novel broad-spectrum antibiotic. Polyfunctional metalloantibiotics that are based on PZP-915 or other CDIs may therefore be an attractive alternative to traditional single target drugs as the prospect of a presumably longer clinical half-life could partially alleviate some of these current economic challenges in antibiotic development.

While the low occurrence of resistant mutants is highly desirable, it poses a challenge for mode of action studies. Consistent with this, we were unable to isolate mutants of S. aureus that show enhanced resistance to 915+Cu. The only gene(s) we could link to resistance/sensitivity were the Cu-efflux pumps, copA and copX using preexisting mutants that are either deficient in or overexpress these genes (Fig. 5B, 5C, and 5D). However, it is encouraging to see that even these resistance genes are of limited efficacy, as overexpression of copA only increased the tolerable dose by 4-fold. This is most likely due to the requirement of ATP for the pumps to function, which is quickly depleted in Cu overloaded cells (Fig. 7B). In consequence, this depletion of ATP could also negatively impact the homeostasis of other cations including Na⁺ and K⁺, as their gradients are also, in part, maintained by ATP-dependent transporters. This could be a downstream source of toxicity from 915+Cu treatment. Of note, it was of interest that 5 different MRSA strains that have antibiotic resistances to not only beta-lactam antibiotics, but also to clindamycin, tetracycline, or erythromycin (Table S1) were still equally sensitive to PZP-915 as the Newman laboratory strain. This implies that multi-drug resistance phenotypes of circulating MRSA strains do not protect against PZP-915-mediated Cu-related toxicity.

Overall, our data capture an intriguing scenario for PZP-915’s mode of action. S. aureus tolerates Cu concentrations of up to 100 μM in our assay system, but it takes only 0.6 μM of PZP-915 to increase the cell associated Cu content by ~13-fold relative to the equimolar Cu-only treatment (Fig. 5A). This comparison highlights the importance of the cell envelope for Cu tolerance in S. aureus, which has evolved to limit the influx of toxic Cu ions. In consequence, cells comfortably tolerate physiologically achievable Cu concentrations in the host organisms (incl. humans). In contrast, PZP-915 likely acts by disguising Cu ions, allowing significant amounts of this metal to enter the cell. We think that this process subsequently overwhelms the intracellular buffering and efflux capacity for Cu. If PZP-915 functions as a classical Cu shuttle that continues to transport more and more Cu into the cell over time remains unknown as it will be challenging in our assay system to distinguish between Cu ion that entered the cell in a PZP-915 mediated fashion or by other means. However, the presented findings highlight PZP-915 as a novel membrane-directed metalloantibiotic against S. aureus that is likely to target a multiplicity of metabolic functions rather than simply imposing physical damage to the membrane structure. PZP-915 adds to the increasing pool of second generation cupricidal antibiotics with high selectivity for bacteria that efficiently overcome preexisting bacterial antibiotic resistance.