All bacterial strains used in this study are listed in Table 1. The reference strains are indicated by their ATCC and CCUG numbers, while the clinical strains, which were obtained from the strain collection at the Department of Medical Microbiology, St. Olav’s (SO) University Hospital, are indicated by their SO codes.

For the clinical strains, this was essentially done as defined by EUCAST Clinical Breakpoints and guidance (EUCAST, 2021).

APIM-peptides (Innovagen, SE) used in this study have the same N-termini but differ in the composition of linkers and/or CPPs as shown in Table 2. Peptides 1 (RWLVK) and 2 (RWLVK*) are previously used in Nedal et al. (2020). A C-terminus FAM-labeled betatide (Innovagen) was used to study intracellular import. All the concentrations of APIM-peptides given in the different figures are net peptide concentrations, and 4 μg/ml equals approximately 1 μM.

HaCaT, a human spontaneous immortalized keratinocyte cell line, was cultured in Dulbecco’s Modified Eagle Medium (DMEM; 4.5 g/L glucose; Sigma-Aldrich), supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich) and 1 mM L-glutamine (Sigma-Aldrich). HEK293, an immortalized embryonic kidney cell line, was grown in DMEM (BioWhittaker, Walkersville, MD, United States) with the same supplements as described above. In addition, Fungizone^® amphotericin B (2.5 μg/ml; Gibco, Thermo Fisher Scientific, Waltham, MA, United States) and 1 mM antibiotic mixture containing 100 μg/ml penicillin and 100 μg/ml Streptomycin (Gibco) were added to the growth media. The cells were incubated at 37°C in a humidified incubator with 5% CO₂.

Minimal inhibitory concentration assay was conducted as recommended by the Clinical and Laboratory Standards Institute (CLSI) (Cockerill et al., 2012), similar to a previous report (Nedal et al., 2020). Briefly, bacterial colonies from blood agar plates were suspended and grown in Cation-Adjusted Mueller-Hinton Broth (CAMHB, 22.5 mg/ml Ca²⁺, 11 mg/ml Mg²⁺). The bacterial suspension was adjusted to 0.5 McFarland standard (∼1 × 10⁸ colony-forming units (CFU)/ml) and serial diluted 1:200 in CAMHB (∼5 × 10⁵ CFU/mL). This was subsequently added to polypropylene microtiter plates (Greiner, 100 μl/well, ∼5 × 10⁴ CFU/well) already prepared with betatide and different antibiotics as single agents or in combinations (11 μl/well, twofold serial dilutions). The suspension was plated out on blood agar plates to confirm the CFU/ml. Both the microtiter plates and the blood agar plates were incubated at 37°C for 24 h. The lowest concentration of antibiotics and/or betatide capable of inhibiting visible bacterial growth was determined as the MIC.

The MICs of ampicillin (Sigma, A9518), cefoxitin (Sigma, C4786), cefotaxime (Sigma, 219504), ceftazidime (Sigma, C3809), ceftriaxone (Sigma, C5793), ciprofloxacin (Sigma, 17850), clindamycin (Sigma, C5269), ertapenem (Sigma, SML1238), gentamicin (Gibco, 1510049), linezolid (Sigma, PZ0014), meropenem (Sigma, M2574), methicillin (Sigma, 51454), and fusidic acid (MedChemExpress, HY-B1350A) were determined in addition to that of betatide.

An overnight culture of Escherichia coli was diluted 1:100 in Luria-Bertani (LB) medium and allowed to grow until an optical density (OD) at 600 nm (OD600) of 0.06–0.1. APIM-peptides were prepared by serial dilution in Milli-Q H₂O and kept at 4°C. Fresh LB medium (60 μl) was added to a flat-bottom microtiter plate. The bacterial suspension was diluted 1:100 in LB, and 75 μl of this suspension was added to the wells. Finally, 15 μl of APIM-peptide solution (to final concentrations 60, 120, and 240 μg/ml) or distilled water (positive control) was added to the wells, reaching a total volume of 150 μl per well. Data are presented only for the dose that separated the effect of the different peptides, i.e., 60 μg/ml. A blank sterile medium was used as negative control. The plates were incubated with shaking at 510 rpm at 37°C inside a plate reader (TECAN, Spark^®), and OD was read every hour over a period of 24 h.

The MIC for E. coli in LB medium is higher than that in CAMHB; thus, concentrations higher than the MIC given in Table 2 are used in the growth inhibition and mutagenesis assays.

HEK293 cells (6,000 cells/well) were seeded in 96-well microtiter plates. After 4 h, APIM-peptides (12–48 μg/ml) were added, and the cells were incubated without change of media for up to 4 days. Viability was measured using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay as described (Gilljam et al., 2009). Data show the percentage of viable cells relative to untreated cells for 24 μg/ml at 72 h.

The rifampicin (Rif^R) mutagenesis assay was performed as described (Nedal et al., 2020). Briefly, an overnight culture of E. coli was diluted 1:1,000 and grown until an OD600 of 0.01. APIM-peptides (20 μg/ml) were added to LB media with glass beads (for uniform distribution of APIM-peptides) and incubated for 30 min at 37°C. The pellets were next collected, resuspended in 500 μl PBS, and exposed to UV-C (20 mJ/cm²) in a six-well plate at 4°C. The unexposed bacteria (-UV) were otherwise handled exactly like the UV-exposed bacteria. Next, the cells were resuspended in LB media and incubated in a rotary shaker at 37°C at 250 rpm for 2 h before being harvested, diluted, and plated on LB with soft agar with and without rifampicin (100 μg/ml). Mutation frequency, Rif^R/10⁸, is obtained as follows: (number of colonies on the rifampicin plate (Rif^R)/(number of colonies on LB plates without antibiotics), per milliliter of bacterial suspension.

The epithelialization assay is a modified version of the scratch test (Longaker et al., 1989; Walter et al., 2010). Briefly, HaCaT cells (1 × 10⁶ cells/well) were seeded in six-well plates with Steri-Strips™ (R1540, 3M Healthcare, United States) attached to the bottom. The cells were confluent in monolayer after 24 h (day 1), and the strips were then removed, creating even 3-mm gaps in the middle of the wells. The wells were washed 2× with PBS before fresh medium was added. The cells were next infected with 450 CFU/ml of an antibiotic-resistant S. epidermidis strain (MRSE, see Table 1) and treated with betatide (2–24 μg/ml). All treatments were done at day 1, and the epithelialization of the gaps was examined over a period of 7 days by taking pictures every 24 h in light microscopy (EVOS^® FL, Life Technologies). Bacterial growth was examined by plating of supernatants. The effect of betatide on noninfected HaCaT cells with regard to viability and epithelization was examined in parallel wells without MRSE.

Epithelialization was calculated from the change in the area of the gaps over time using freehand or rectangular selections in the software Fiji ImageJ 1.52p (National Institutes of Health, United States).

The MIC for MRSE in DMEM (0.25 μg/ml) is lower than the MIC given in Table 2; thus, concentrations lower than MIC were used.

The intracellular infection model used was modified from Iqbal et al. (2016) by optimizing the number of multiplicity of infection (MOI: number of bacteria per cell) and time of infection. Briefly, HaCaT cells (6.5 × 10⁴/well) were seeded in 24-well plates and incubated overnight in a medium without antibiotics. The next day, approximately 1.25 × 10⁵ cells/well were infected with S. aureus at an MOI of 100 in antibiotic-free media. The plates were incubated for 3 h. Next, the cells were washed 2× with 500 μl PBS and treated with 100× MIC of gentamicin (100 μg/ml, MIC = 1 μg/ml) for 1 h, before further incubation in media with 10× MIC of gentamicin to kill the extracellular bacteria. Betatide (8–48 μg/ml) was added to the infected cells, and an equal amount of distilled water was added to the untreated control. After 16 h of incubation, the extracellular media (100 μl) from each well were plated to confirm the eradication of extracellular S. aureus. Next, the cells were washed 2× with 500 μl PBS before they were lysed with 0.2% Triton-X (500 μl) for 30 min at room temperature. The lysed samples were placed on ice, and 500 μl cold Tryptic Soy Broth was added before they were plated on blood agar plates. Data are presented for the dose that showed the best intracellular effects and low HaCaT cell toxicity, i.e., 24 μg/ml.

Intracellular import of betatide in HaCaT cells was examined using a fluorescent-tagged betatide (betatide-FAM, ∼20 μg/ml). Vybrant^® DyeCycle™ Ruby stain (VDR, 5 μM, V10273, Life Technologies™), which can penetrate plasma membranes, was used to stain DNA of live cells. Both betatide-FAM and VDR were added to live HaCaT cells immediately before (<2 min) examination in a Zeiss LSM 510 Meta laser scanning microscope equipped with a plan-apochromat × 63/1.4 oil immersion objective. FAM was excited at λ = 514 nm and detected above 515 nm, and VDR was excited at λ = 633 nm and detected above 650 nm. We used consecutive scans, and the optical slices were 1 μm.

Bacteria (E. coli K-12 MG1655 and S. aureus ATCC 29213 and SO-SAU19-1) were serial passaged in CAMHB as described by Silverman et al. (2001) with some small modifications. Briefly, bacteria were passaged for up to 32 days in a round-bottom polypropylene microtiter plate (Greiner) in media containing 0.25, 0.5, 1, and 2× MIC of betatide or other antibiotics. In E. coli, the MIC was 0.06 μg/ml for ciprofloxacin and 16 μg/ml for ampicillin. In S. aureus, the MIC was 1 μg/ml for gentamicin, 0.25 μg/ml for ciprofloxacin in ATCC 29213, and 0.5 μg/ml for SO-SAU19-1. For every passage (each day), the new MIC was determined, and bacterial cells from the 0.5× MIC culture were continued for passage by adjusting this culture to ∼0.5 × 10⁶ CFU/ml in CAMHB. Fresh preparations of betatide/other antibiotics (0.25–2× MIC) were finally added to the diluted cultures, adjusted to the new MIC.

The antibacterial effect of APIM-peptides was previously shown to be partly caused by the CPP part, although MIC was 2×–3× higher for the CPP only than for the full-length APIM-peptide variants (Nedal et al., 2020). We have also previously found that an APIM sequence linked to a CPP containing 11 arginines (R11) had higher antibacterial activity than the same sequence linked to HIV-TAT, transportan, and penetratin CPPs (data not shown). Peptide 1, which is based on the sequence of the original anticancer peptide (Muller et al., 2013), is previously shown to have lower antibacterial efficacy (higher MIC) than the same peptide with a different linker, peptide 2 (Nedal et al., 2020), and this was verified here (Table 2). The number of arginines (Rs) required to facilitate uptake into the nucleus of mammalian cells has been found to be eight, while an increased proportion of the peptide was detected in the cell membrane when Rs were increased to 16 (Futaki et al., 2001). Here, we therefore explored if the number of Rs in the CPP domain of peptide 2 affected the growth of bacteria and human cells differently. Table 2 includes comparison of performance of the peptides in more assays than previously reported (Nedal et al., 2020); therefore, we also included peptide 1 in these tests.

Reduction in the number of Rs from 11 to 8 (peptides 2–5) did not affect MIC in the MRSE strain, while peptides 2 and 5 had the lowest MIC in E. coli (K-12 MG1655) (Table 2). The MIC assay determines visual growth inhibition after 24 h; therefore, to explore potential differences in antibacterial efficacy in more detail, we examined these peptides’ inhibitory effect on the growth of E. coli over 24 h (for growth curves, see Supplementary Figure 1). We found that peptide 2 inhibited bacterial growth more than peptide 5 did in this assay (Table 2, reduction of viability, E. coli); thus, the superior antibacterial efficacy based on these two assays was determined to be that of peptide 2.

One of the most important factors to consider when selecting and developing a drug is low toxicity for human cells. The ideal situation would be to develop an APIM-peptide variant with a lower effect on mammalian cells and a larger effect on bacteria, i.e., to increase the therapeutics window. However, when the viability of HEK293 cells was tested after treatment with the peptides using the MTT assay, the peptides reduced the viability of HEK293 cells similarly, with peptide 5 (R8) possibly inhibiting the viability slightly more than the other peptide variants did (Table 2 and Supplementary Figure 2).

Peptides 1 and 2 are previously shown to inhibit TLS at sub-MICs via inhibition of polymerase V (Pol V) binding to the β-clamp (Nedal et al., 2020). Because inhibition of TLS is an important trait of these peptides, we compared these two peptides with the peptides with shorter CPPs for their ability to reduce the mutation frequency in E. coli using the Rif^R assay. We found that peptide 2 reduced the mutation frequency more efficiently than the other peptides did (Table 2 and Supplementary Figure 3).

In summary, these results show that the peptide with the GILQ-WRK-I linker and the R11 CPP is superior to the peptide with the W-KKKRK-I linker and to peptides with shorter arginine chains, with regard to antibacterial and antimutagenic properties, while the toxic effects on human cells are similar in all the peptide variants tested. Based on the results summarized in Table 2, peptide 2 was selected as the lead antibacterial peptide candidate and hereafter named betatide (beta-clamp targeting peptide).

Next, the activity of betatide against a wider selection of bacterial species from the ESKAPE list, i.e., MDR clinical isolates and corresponding reference strains, was examined (Table 3A). MICs for conventional antibiotics in the different MDR strains were determined in parallel with betatide, and this showed that the MDR strains had a 4× to >16,000× increase in MIC relative to their reference strains. However, betatide had an overall low MIC for all species (8–16 μg/ml) and was equally efficient against the reference strains as the clinical MDR isolates (except in one case: Enterococcus faecalis, 2× MIC). These results show that betatide has broad antibacterial activity and has no cross-resistance with the other antibiotics tested. This was expected as betatide has a ubiquitous bacterial target that is not shared by the other antibiotics. In some strains, a 2×–4× additive effect of the commercial antibiotic was detected when combined with 0.5× MIC of betatide, and an additive effect was observed more often for the MDR clinical isolates than for the reference strains (Table 3A).

Because fusidic acid is commonly used in the treatment of wound infections, we examined if betatide enhanced the effect of fusidic acid against S. aureus. An 8× reduction in MIC of fusidic acid was observed when combined with 0.5× MIC betatide in a fusidic acid-sensitive MRSA strain (Table 3B). Further, a 2×–4× additive effect was detected in three fusidic acid-resistant MRSA strains (FR-MRSA).

Altogether, these results support the potential of betatide, both as a single antibacterial agent and in combination with commonly used antibiotics.

In a mouse MRSA wound infection model, a gel containing a variant of the APIM-peptide significantly reduced the bacterial load with no visible toxicity on the skin (Nedal et al., 2020). As wound infections could be a suitable indication for these peptides and mouse skin may differ from human skin, we next more closely examined the efficacy of betatide and its effect on the epithelialization in an in vitro wound infection model. For this, we developed an epithelialization assay that is similar to the scratch test (Longaker et al., 1989; Walter et al., 2010), but where the gaps were made identical for accurate quantification by using strips. HaCaT cells infected with MRSE without treatment were all dead by day 2 after an exponential growth of the bacteria (Figure 1A, second panel). However, betatide (2 μg/ml) already eradicated MRSE from the culture wells at day 2 (Figure 1A, fourth panel). The epithelialization was completed at day 4, similar to the uninfected cells (Figure 1A, first and third panels). Thus, the epithelialization capacity of the cells was not affected at doses that completely abolished infection (epithelialization quantified in Figure 1B, CFU/ml depicted on the image in Figure 1A). The cells were cultured for up to 7 days without the infection re-emerging (Figure 1A, day 7, fourth and first panels).

When examining how epithelialization was affected by higher doses of betatide, a gradual decrease in percentage of epithelialization with increasing doses of betatide was observed (Figure 1C, quantified in Figure 1D). An approximately 70% decrease in epithelialization at day 4 was detected when using a betatide dose that was 12× higher than what is needed for a total eradication of the bacteria (24 μg/ml); however, the cells were not dying, and the epithelialization was re-established on day 7. Epithelialization was also re-established on day 7 after treatment with up to 40 μg/ml betatide (data not shown). Overall, these data indicate that doses that are more than 12× higher than the antibacterial dose could be used without severely affecting epithelialization.

Betatide is a variant of the APIM-peptide ATX-101, which is previously shown to be rapidly imported into multiple cells and to be distributed to all tissues upon intravenous infusion (Muller et al., 2013; Søgaard et al., 2018). Here, we show that fluorescent-labeled betatide (betatide-FAM, green) is rapidly taken up by live HaCaT cells (Figure 2A) and has similar subcellular localizations as ATX-101 (Muller et al., 2013); i.e., it is found in the cytosol, in the nuclei, and in the nucleoli. In addition, betatide-FAM in S. aureus-infected cells is found in small circular dots in the cytosol (highlighted by white arrows in Figure 2A, upper panel) that also are stained with live-cell DNA staining (magenta). These dots are not detected in uninfected HaCaT cells stained with live-cell DNA staining (lower panel), suggesting that these circular dots represent S. aureus.

To examine if betatide, which co-localizes with intracellular S. aureus (Figure 2A, merged image, white arrows), has antibacterial activity against intracellular S. aureus, we next measured intracellular bacterial counts in infected HaCaT cells treated with the peptide. Because S. aureus produces toxin that kills mammalian cells (Fraunholz and Sinha, 2012), optimization of the infection period and the number of infecting bacteria per cell was vital. We found that 100 MOI and infection for 3 h gave an intracellular infection without severe HaCaT cell cytotoxicity. The remaining extracellular bacteria were killed by gentamicin treatment prior to treatment of the infected cells with betatide (confirmed by plating at the time of harvest of the cells). A 4× reduction in intracellular bacterial load was found in betatide-treated cells (24 μg/ml), compared to untreated control (Figure 2B). Treatment with lower concentrations of betatide did not cause a significant reduction in CFU, while higher concentrations killed the infected HaCaT cells. Toxins from S. aureus likely sensitized the infected cells as uninfected HaCaT cells tolerated up to 40 μg/ml betatide. The maximum tolerated dose of betatide may therefore be different in different types of bacterial infections. In conclusion, these results show that betatide is rapidly taken up in mammalian cells where it retains its antibacterial activity.

Long-time exposure to sub-MIC levels of antibiotics are known to increase TLS and induce resistance development (Kreuzer, 2013; Raeder et al., 2021). To directly examine the resistance development against betatide, we exposed E. coli and S. aureus (both MDR and reference strain) to sub-MIC and 1×–2× MIC levels of betatide through serial passage and measured MIC over 20–30 days. Compared to bacteria exposed to gentamicin, ampicillin, and ciprofloxacin, those exposed to betatide had a much lower capacity to develop resistance (Figure 3). In E. coli, betatide showed only a temporary 2× increase in MIC during these 32 days, compared to up to a 64× stable increase in MIC for ciprofloxacin and ampicillin (Figure 3A). S. aureus developed a higher increase in MIC toward all the treatments compared to E. coli: up to a 256× increase in MIC for ciprofloxacin and gentamicin, while only an 8× increase in MIC for betatide was detected in the first experiment (Figure 3B, parallel 1). As mutations are stochastic events, we repeated this experiment and found no detectable resistance development with betatide after 30 days (Figure 3B, parallel 2), while ciprofloxacin was more similar to parallel 1). Furthermore, an MDR strain of S. aureus (MRSA) did not show any signs of resistance development against betatide during 20 passages even though 8× and 32× increases in MIC against gentamicin and ciprofloxacin, respectively, were detected (Figure 3C). These experiments show reduced ability of the bacteria to develop resistance against betatide compared to commonly used antibiotics.