BTP-001 (Innovagen, Sweden and Biosynth, Netherlands) consists of the APIM-motif (in bold) connected to an arginine tail via a linker and has an acetylated N-terminus and an amidated C-terminus; Ac-MD-RWLVK-GILQWRKI-R11-NH₂ (Nepal et al., 2021). Lyophilized peptide was dissolved in water and stored at 4°C. All concentrations are given as net peptide concentrations and 4 μg/mL equals approximately 1 μM. The halogenated pyrrolopyrimidine JK-274 ((R)-4-(4-((1-(4-bromophenyl)ethyl)amino)-7H-pyrrolo[2,3-d]pyrimidin-6-yl)phenol) was synthesized as previously described and 4 μg/mL is approximately equal to 10 μM (Olsen et al., 2022). JK-274 was dissolved in DMSO and stored at 4°C.

Staphylococcus aureus ATCC 29213 cultivated in CAMHB was used for signalomic and metabolomic analysis. To prepare S. aureus pre-cultures, glycerol stock was plated onto blood agar plates and incubated overnight at 37°C. A single colony was picked and incubated in CAMHB overnight. The pre-culture was diluted 1:100 and aliquoted into 6 individual flasks and incubated (37°C, 250 rpm) until an OD₆₀₀ = 0.3. Next, the cultures were treated with either 0.5 μM BTP-001, 1 μM BTP-001, 5 μM JK-274, 7.5 μM JK-274 or 0.5 μM BTP-001 and 5 μM JK-274. One culture was left untreated as a control. After treatment, the cultures were incubated (37°C, 250 rpm) until sampling.

Sampling for signalomics was conducted at 10, 25, 50 and 180 min after treatment. For the first 3 timepoints, 10 mL culture was sampled from 3 independent biological replicas (BRs). At 180 min, 5 mL were sampled from 4 BRs. Samples were centrifuged (4,500 rcf, 10 min, 4°C) and supernatant discarded. The pellet was frozen in N₂ (l) and stored at −80°C until further processing.

Samples for metabolic profiling were harvested from 4 BRs 180 min after treatment. The sampling procedure was adapted for S. aureus from a previously described protocol developed for E.coli (Thorfinnsdottir et al., 2023). Briefly, biomass equivalent to 10 OD units was sampled and filtered through two stacked polyvinylidene fluoride filters with a pore size of 0.65 μm (Durapore^®, DVPP04700; Millipore, USA) exposed to a vacuum pressure of 400 mbar below the ambient pressure. Filtered biomass was rinsed with Milli-Q H₂O (10 mL, 37°C), quenched in 10 mL ice-cold MeOH:ACN:H₂O (20:30:50), frozen in N₂ (l) and kept at −80°C until further processing.

Staphylococcus aureus cell extract was prepared as described for E. coli in MIB optimization, except lysostaphin (1 μg/mL) was used instead of lysozyme. Cell extract (100 μL; 1 mg/mL protein) was added to an equal mix of Sepharose beads coupled to the kinase inhibitors PurvB, L-1 and L-3. The MIB assay and sample preparation for MS was conducted as described for MIB optimization.

For both E. coli (MIB optimization) and S. aureus samples, liquid chromatography (LC)-MS/MS analysis was used for protein identification and MS data were processed using MaxQuant v 2.1.3.0 for label free quantification (LFQ) of proteins (Cox et al., 2014). The following search parameters were used: the digestion enzyme was specified as trypsin with a maximum of 2 missing cleavages, variable modifications were set to oxidation (M), acetylation of protein N-terminal, and deamination (NQ), and fixed modifications were set to carbamidomethyl (C). LFQ min. ratio count was set to 1. Samples were queried against the imported E. coli (strain K12, proteome ID: UP000000625, downloaded February 18, 2021) or S. aureus (strain NCTC 8325/PS 47, proteome ID: UP000008816, downloaded on February 09, 2022) reference proteome from UniProt and Andromeda, MaxQuant’s internal contaminants database. The false discovery rate (FDR) for protein and peptide identification was set to 1%. Only unique peptides were used for definite protein group identification. The area under the peak curve was integrated to obtain peak abundances. The total abundance of all peptides identified for each protein during each run was used to normalize the abundance in every protein group using the LFQ algorithm with minimum peptides ≥1.

LFQ values were log-transformed with base 2 before subtracting the control for each timepoint, respectively, to yield the change relative to the control for each treatment group. The values were then tested using a two-sided Wilcoxon signed rank test in R v 4.2.1 to check the directionality of change for each treatment group compared to the control. To reduce the number of hits, LFQ values were also analyzed in R using the DEP package v 1.18.0 from Bioconductor v 3.16 (Zhang et al., 2018). Each timepoint was analyzed separately. Data were normalized using variance stabilizing transformation. Proteins with a missing value in ≥2 BRs for more than 1 sample were filtered out, and remaining missing values were imputed by the k-nearest neighbor (kNN) technique. Differentially pulled down proteins were defined as having either an increased or decreased pulldown compared to the control in all BRs (based on the Wilcoxon signed rank test) and having a p-value ≤0.1. The STRING v 11.5 database was used to create a protein–protein physical interaction network of differentially pulled down proteins with the highest confidence (interaction score > 0.9) (Szklarczyk et al., 2023). The PANTHER v 17.0 classification system was used to run a gene ontology enrichment analysis of biological processes (Mi et al., 2013).

The LC–MS/MS signalomics data have been deposited in the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2021) partner repository with the dataset identifier PXD043757.

Intracellular metabolites were extracted by an extended extraction protocol based on Thorfinnsdottir et al. (2023). Briefly, samples were cycled between −20°C EtOH and N₂ (l) in three consecutive freeze–thaw cycles before filters were removed and remaining cell debris was pelleted (4,500 rcf, 10 min, −9°C). Supernatants were lyophilized, reconstituted in 500 μL cold MQ-H2O and cleared by spin-filteration with a 10 kDa molecular cutoff (20,817 rcf, 10 min, 0°C).

Phosphorylated metabolites, organic acids, and amino acids were absolutely quantified by capillary ion chromatography (capIC-) and LC–MS/MS as previously described (Stafsnes et al., 2018; Røst et al., 2020).

Downstream data processing and absolute quantification was performed as previously described applying the TargetLynx application manager of MassLynx 4.1 (Waters, USA). Absolute concentrations were interpolated from calibration curves prepared from appropriate dilutions of analytical grade standards (Sigma-Aldrich). The response factor of the corresponding U¹³C-isotopologue of each compound was used to correct the standard and sample extract response factors. Extract concentrations were normalized within replicas and to the total metabolite abundance for all treatments. T-tests were conducted in MetaboAnalyst to identify metabolites with a significant change compared to the control (Pang et al., 2020).

Exponentially growing S. aureus were washed 3× with PBS and the pellets were resuspended in PBS containing 2 μg/mL BioTracker Bacterial FtsZ Cell Dye (Sigma-Aldrich; SCT090, excitation: 488 nm, emission: 510 nm) (Ferrer-González et al., 2019) and incubated at RT for 30 min. The samples were washed 2× with PBS and the pellets were resuspended in PBS (volume based on cell density). Then, 8 μL were added to an agarose pad to immobilize the bacterial cells before examination in a LSM 510 Meta laser scanning microscope (Zeiss, Germany) equipped with a plan-apochromat 63×/1.4 oil immersion objective.

ROS were measured using the cell-permeable probe 2′,7′-dichlorofluorescein diacetate (DCFH-DA; Sigma-Aldrich). Briefly, the overnight culture was centrifuged (3,220 rcf, 10 min) and resuspended in PBS to OD₆₀₀ = 0.3. The resuspended culture was incubated with DCFH-DA (5 μM) for 30 min (25°C, 250 rpm). A 96-well plate was prepared with the different concentrations of BTP-001 and positive control wells of H₂O₂ (20 mM), while some wells were left as untreated control wells. The culture with loaded probe was added to the plate to a total volume of 200 μL in each well. The plate was incubated for 30 min (37°C, 400 rpm) before measuring fluorescence (excitation at 485 nm, emission at 520 nm).

Cytotoxicity of JK-274 on human cells was tested on HaCaT (spontaneously transformed aneuploid immortal keratinocyte cell line) using the PrestoBlue™ viability assay (Invitrogen) according to the guidelines provided by the manufacturer. Briefly, 3,000 cells/well were seeded in 96-well plates. After adhesion (minimum 4 h), the cells were treated with JK-274 (2.5–20 μM) and BTP-001 (2 μM) or left untreated and incubated (37°C, 5% CO₂, humidified atmosphere). The media was either left unchanged or exchanged for fresh media 4 h after treatment. Viability was measured 24 h after treatment by adding PrestoBlue™ to a final concentration of 1× and incubating for 2 h at 37°C before reading the fluorescence (excitation at 544 nm, emission at 590 nm).

To measure viability of S. aureus after 4 h of treatment with JK-274 and BTP-001, 3 × 10⁸ CFU/mL were added to a 96-well plate and treated with either JK-274 (2.5 μM) and BTP-001 (2 μM), isopropanol (100%), or left untreated, and incubated for 4 h (37°C). Viability was measured as described above. CFU/mL was also determined after 4 h.

The proteome contains thousands of proteins with diverse functions, some of which may not be relevant to determining antibiotic MoA. In addition, signal transduction in response to stimuli such as antibiotics is a rapid and transient event that occurs much faster than protein synthesis and degradation. Selecting for the signalome reduces the complexity of the results, making it easier to identify key proteins and signaling pathways, and allows for the detection of dynamic changes caused by post-translation modifications that alter protein activity. The MIB assay has successfully been used to study the signalome in both eukaryotic and prokaryotic cells by pulling down activated signaling proteins, such as kinases, ATP/GTP-binding proteins, and proteins in complex with the latter two (Petrovic et al., 2017; Søgaard et al., 2019; Sumabe et al., 2021; Røst et al., 2023). Increased protein pulldown upon treatment suggests increased activation while reduced pulldown indicates inhibition, destabilization, or reduced activation of the protein. The kinase inhibitors L-5 (in-house), PurB, and Bis-X were used in previous studies, but with a greater ability to pull down proteins in eukaryotes than prokaryotes. To optimize the MIB assay for prokaryotes, we compared the previously used kinase inhibitors to 4 in-house compounds designed to mimic kinase inhibitors (from here on referred to as kinase inhibitors) for their ability to pull down proteins from the model organism E. coli MG1655. The chemical structures for the in-house kinase inhibitors are shown in Figure 1A.

In total, the kinase inhibitors pulled down 1,077 unique proteins, with L-1, L-2, L-3, L-4, and PurvB pulling down the highest number of proteins (Figure 1B). The 5 kinase inhibitors with the highest pulldown had 869 proteins in common (Figure 1C). The two kinase inhibitors not included in the Venn diagram, Bis-X and L-5, did not pull down any unique proteins. The best combination was determined to be a mixture of L-1, L-3 and PurvB, which included 1,075 proteins, accounting for 24% of the E. coli MG1655 proteome. This set of kinase inhibitors yielded even better pulldown in S. aureus as they pulled down 1,568 proteins, covering ~50% of the total proteome (raw data uploaded in PRIDE). The results indicate that the MIB assay can be used for multiple bacterial species but that the proteome coverage may vary.

To check if the kinase inhibitor concentrations used in the MIB assay were sufficient to prevent abundant proteins from saturating the kinase inhibitors and outcompeting scarce proteins, we increased the amount of kinase inhibitor added to the MIB column. The number of detected proteins remained the same; thus, the ratio of pulled down proteins likely reflects the ratio found in the cell.

A putative Tmk inhibitor called JK-274 (Figure 2A) was previously found to have a low MIC (20 μM, 8 μg/mL) toward S. aureus. When combined with a ½ MIC BTP-001 (2 μM, 8 μg/mL), the MIC for JK-274 was reduced 8-fold (Olsen et al., 2022). In this study, we wanted to investigate the molecular basis of the individual and combined effects of JK-274 and BTP-001 by using signalomics supplemented with metabolomics. Our goal was to treat exponentially growing S. aureus with doses that induce stress without substantial lethality to explore the specific effects of JK-274 and BTP-001 while avoiding non-specific changes caused by extensive cell death. Sub-MIC doses also guarantee sufficient biomass for signalomic and metabolomic sampling. To ensure sampling during logarithmic growth, the S. aureus cultures received sub-MIC treatment at OD₆₀₀ = 0.3 (experimental scheme shown in Figure 2B). Growth curves from preliminary studies were used to determine the doses to be used in combination (Supplementary Figure S1). The untreated control reached an OD₆₀₀ = 1.6 in 3 h (Figure 2C). Single treatments with 5 μM JK-274 (1/4 MIC) and 0.5 μM BTP-001 (1/8 MIC) almost reached the same cell density as the untreated control; however, when combined, the two doses yielded a reduction in cell density (Figure 2C). To compare the combination treatment to single treatments with approximately the same cell density, we also treated S. aureus with 1 μM BTP-001 and 7.5 μM JK-274.

As mentioned above, 1,568 proteins were pulled down from S. aureus using the MIB assay (Supplementary Table S1). Principal component analysis (PCA) revealed that samples from the same timepoint, independent of treatment, cluster together (Figure 3A); therefore, we only compared samples taken at the same timepoint. Many proteins exhibited differential pulldown (DPD) after treatment compared to the control, with BTP-001 yielding the most DPD proteins at 10, 25, and 50 min post-treatment while JK-274 and the combination resulted in the most DPD proteins at 180 min (Figure 3B); thus, indicating that their effects on the signalome are temporally separated. For metabolomics, the intracellular concentrations of selected metabolites were quantified at 180 min post-treatment (Supplementary Table S2), which may have been too late to detect major changes induced by BTP-001 on the metabolome, given the rapid response observed in the signalome. Nonetheless, the log₂ fold-change (LFC) of the metabolite pools in the treated samples compared to the untreated control still show some changes (Figure 3C). However, PCA analysis of these pools revealed that mainly treatment with the high dose of JK-274 (7.5 μM) yielded a significant change in intracellular metabolite pools compared to the control, indicated by the respective 95% confidence regions not overlapping (Figure 3D).

A gene ontology (GO) enrichment analysis of biological processes revealed that JK-274 led to DPD of proteins enriched for tricarboxylic acid (TCA) cycle enzymes, as visualized in a STRING physical interaction network (Figure 5A, burgundy circles). Of the 9 TCA cycle enzymes detected, 8 showed a significantly reduced pulldown at 50 and 180 min (Figure 5B). The first 3 enzymes in the TCA cycle also showed reduced pulldown at 25 min post-treatment. Additionally, pulldown of the proteins PflB (formate acetyltransferase) and Ldh2 (L-lactate dehydrogenase 2), which are associated with anaerobic and fermentative metabolism, increased after treatment (Figure 5B), possibly to compensate for reduced flow though the TCA cycle. Perturbations to the TCA cycle were supported by a decrease in intracellular levels of TCA cycle metabolites like α-ketoglutaric acid (α-KG) and malic acid (Mal), especially after treatment with the high dose of JK-274 (Figure 3C).

JK-274 treatment significantly increased pulldown of nitric oxide dioxygenase (NOD, SAOUHSC_00204), the protein that catalyzes the conversion of nitric oxide (NO) into nitrate and NADPH (Figure 5B). We detected an increase in NOD pulldown from 25 min post-treatment, corresponding to the decrease in TCA cycle enzymes. One of the most sensitive targets of NO is aconitase, which catalyzes the isomerization of citrate to isocitrate in the TCA cycle; thus, reduced pulldown of TCA cycle enzymes may be indirectly due to induction of NO stress or via direct inhibition. Microbes often reduce citrate flux into the TCA cycle under NO stress (Auger et al., 2011; Richardson et al., 2011; Gupta et al., 2012). Accordingly, the metabolic data showed a decrease in α-KG levels after JK-274 treatment (Figure 3C). Increased pulldown of Qox proteins and ScdA provided a second indication that JK-274 triggered NO stress as NO stress is shown to increase cytochrome biosynthesis (QoxA, QoxB, QoxC, and QoxD) and heme repair (ScdA) (Kinkel et al., 2013).

After treatment with JK-274, pulldown of urease subunits (UreA, UreB, UreC) and accessory proteins (UreD, UreE, UreG) increased (Figure 5B). Ureases catalyzes the hydrolysis of urea into ammonia and carbon dioxide, and urease activity has been associated with bacterial virulence (Mobley et al., 1995; Konieczna et al., 2012). Due to possible inactivation of TCA cycle enzymes by JK-274, the urea cycle may provide an alternate carbon source for the TCA cycle via glutamate (DeMars and Bose, 2018) to compensate for reduced activity of citrate synthase, cis-aconitase, and isocitrate dehydrogenase.

Previously, JK-274 showed inhibitory activity toward E. coli Tmk in an enzymatic assay, and based on structural similarity to known Tmk inhibitors, we postulated that JK-274 also inhibits Tmk in S. aureus (Olsen et al., 2022). Here, Tmk was pulled down, but with no significant difference between JK-274 and the untreated control. Conversely, pulldown of the essential protein Ndk (nucleoside diphosphate kinase) was significantly reduced 180 min after treatment (Figure 5B), indicating a late-stage or secondary response as JK-274 affected growth before 180 min (Supplementary Figure S1). Furthermore, pulldown of several proteins in the pyrimidine biosynthesis pathway, including the highly regulated PyrB protein, was reduced (Figure 5B). This is also reflected in decreased intracellular CMP and UMP pools after JK-274 treatment (Figure 3C).

A GO enrichment analysis of DPD proteins, as visualized in a STRING physical interaction network, showed cell division as a significantly enriched biological process after treatment with BTP-001 (Figure 6A, light green circles). The data revealed a strong reduction in pulldown of FtsA, an essential cell division protein that anchors the Z-ring, as the most significant effect of BTP-001 on the S. aureus signalome (Figure 6B; Supplementary Table S4). Even though S. aureus rapidly responded to BTP-001, a 10-fold reduction in FtsA pulldown was still detected 180 min after treatment. We also detected reduced pulled down of FtsZ, the protein that forms the Z-ring, 25 min after treatment with the high dose of BTP-001 (1 μM, Figure 6B). Interestingly, FtsA and FtsZ physically interacted with the APIM-motif and β-clamp, respectively, in parallel studies conducted in E. coli. Immunoprecipitation of the APIM-motif fused to EYFP (RWLVK-EYFP) from overexpressing E. coli pulled down FtsA from three independent BRs and from none of the EYFP-only controls (Figure 6C). This suggests that the APIM sequence (as in BTP-001) either directly or indirectly via the β-clamp binds to FtsA. FtsZ was also strongly pulled down from E. coli using a His-tagged β-clamp in 2 of 3 BRs, while much lower levels were pulled down in the corresponding control, indicating that the β-clamp and FtsZ likely bind or are in the same complex (Figure 6C). Together, these results indicate that the β-clamp might act as a scaffold in Z-ring formation and cell division, and that BTP-001 might disrupt this interaction.

FtsA is considered a potential antibiotic target as FtsA inhibition is shown to prevent Z-ring formation and promotes delocalization of FtsZ from the midcell (Mura et al., 2017; Ragunathan et al., 2019). In an attempt to correlate reduced FtsA pulldown with impaired Z-ring formation, we quantified Z-rings in BTP-001 (1 μM) treated and untreated S. aureus. However, no visual difference in Z-ring formation was detected between the treated and untreated cells at this dose (Supplementary Figure S2). Because BTP-001 did not completely prevent growth at this concentration, the data indicates Z-ring formation despite a substantial decrease in FtsA pulldown (Supplementary Figure S1). This suggests that FtsA levels were not critically low at this dose or that S. aureus has an unknown redundant protein or mechanism (Mura et al., 2017).

High dose BTP-001 (1 μM) appeared to induce oxidative stress as we detected a strong reduction in the pulldown of Rex, a redox sensing transcriptional regulator (Figure 6B). Rex operates by binding NADH during oxidative stress and subsequently becoming deactivated and disassociating from the repressor site (Sickmier et al., 2005). Since the BTP-001 doses were below MIC, S. aureus was able to overcome the oxidative stress and at 180 min, Rex pulldown was the same as the control. Consistent with the potential role of ROS in the MoA, BTP-001 generated a dose-dependent increase in levels of intracellular ROS (Figure 6D). A common theory postulates that ROS formation plays a role in the MoA of most, if not all, bactericidal antibiotics (Dwyer et al., 2009, 2014). Other antimicrobial peptides have exhibited a higher anaerobic than aerobic MIC, indicating a central role of ROS formation in cell death (Choi et al., 2016, 2017). To test if ROS formation is critical for BTP-001 MoA, we performed an anaerobic MIC assay on S. aureus. However, MIC remained the same (data not shown). ROS might still contribute to BTP-001’s MoA as the 2-fold dilutions and 24-h incubation time used in the MIC assay cannot detect small changes and rapid effects.

Based on previous work showing that APIM-peptides block replication and inhibit TLS at sub-MIC doses, we expected proteins associated with the replisome to be affected by treatment with BTP-001 (Nedal et al., 2020). Analysis of DPD proteins supported this as they showed increased pulldown of the DNA damage recognition proteins UvrA and UvrC, decreased pulldown of the DNA repair associated proteins PcrA (DNA helicase) and PolA (DNA polymerase), and lastly, reduced pulldown of RecA, the protein responsible for SOS induction by binding ssDNA (Figure 6B). Multiple proteins in DNA replication and DNA repair must interact with the DNA sliding clamp for optimal function (Moldovan et al., 2007). Reduced pulldown of PcrA and PolA could be due to BTP-001 blocking their direct interaction with the β-clamp, leading to protein destabilization and degradation. As expected, we mainly saw these changes with the high dose of BTP-001 which led to reduced growth compared to the control (Figure 2C; Supplementary Figure S1).

During growth, protein translation accounts for over 50% of the energy use in the bacterial cell (Russell and Cook, 1995). Therefore, many factors interact with the ribosome to tightly regulate protein synthesis in response to both internal and external stressors. Ten minutes after BTP-001 treatment, we detected 17 DPD proteins that belong to the cellular process translation (Figure 6A, blue circles), which consisted mostly of proteins from the large and small subunit of the 70S ribosome (Figure 6B). BTP-001 binds to the β-clamp at the same site as many replicative proteins and thereby stalls replications (Nedal et al., 2020). Therefore, a possible explanation for decreased ribosomal protein pulldown might be an increased degree of transcription-replication collision and thereby a collapse of translational complexes and degradation of these proteins (Nedal et al., 2020).

After combination treatment, the S. aureus signalome predominantly exhibited characteristics that were also observed after BTP-001 and JK-274 single treatment (Supplementary Figure S3). Although the signalome and metabolome data do not easily explain the strong additive effect, we observed that several DPD proteins involved in stress responses were only significant following combination treatment (Figures 4A, 7). These proteins include Holliday junction resolvases (RuvA and RecU), peptidases and proteases, virulence proteins (Hly and KdpD), and multiple other stress response proteins, e.g., the ribosomal hibernation protein (Hpf, inhibits translation), the PrsA foldase protein, and the metal-binding stress regulator McsA. This suggests that the combination treatment induces more stress than the high dose single treatments even though they had a similar growth rate. Additionally, at 50 min post-treatment, pulldown of riboflavin biosynthesis proteins was strongly reduced following combination treatment (Figure 7). Riboflavin is the precursor for flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN), both of which participate as cofactors in several redox reactions such as electron transfer, lipid metabolism, and oxidative stress response (Pinto and Zempleni, 2016). Reduced riboflavin biosynthesis might impact the ability of S. aureus to prevent and repair damage caused by NO (induced by JK-274) and ROS (induced by BTP-001) as many of the proteins involved in these processes require either FAD or FMN for activity (Chen et al., 2013; Ashoori and Saedisomeolia, 2014).

Overall, the data suggests that BTP-001 and JK-274 have very distinct MoAs that occur at different times, but when combined, lead to excessive stress. Perhaps, the bacterial adaptation toward one drug candidate is comprised of elements perturbed by the other. This could in many cases be an advantage to hamper resistance and to increase the efficacy.

An antibiotic candidate should exhibit low or no toxicity toward mammalian cells at concentrations required to kill the host pathogen. To evaluate the toxicity of JK-274, we measured the viability of HaCaT cells, an immortalized human keratinocyte cell line, by measuring the fluorescence in a resazurin based viability assay. When treated for 24 h, 20 μM JK-274 decreased viability of the HaCaT cells by 35%; however, 2.5 μM JK-274, the dose needed for activity in combination with BTP-001, only reduced viability by 10% (Figure 8A). In S. aureus, combination treatment with JK-274 (2.5 μM) and BTP-001 (2 μM) reduced viability to the same level as the negative control (isopropanol, Figure 8B) after only 4 h and 100% lethality was confirmed by CFU assays (data not shown). Because the in vivo drug concentration is not constant over time, we also measured HaCaT cell viability after changing the media containing the treatment after 4 h. In this case, the combination dose, which was 100% bactericidal, did not affect cell viability after 24 h (Figure 8C). This shows that the cells tolerated the treatment; however, further modifications of JK-274 to increase specificity toward bacterial targets and reduce cytotoxicity is likely needed for use as a single agent.