Antimicrobial Peptides Based on the TisB Toxin: Toward Enhanced Activity and Synergy with Antibiotics

Abstract

Antimicrobial resistance is a growing global health crisis, necessitating novel anti-infective agents. The present study explores the potential of toxin–antitoxin (TA) systems as a source of antibiotic candidates. Thus, 44 antimicrobial peptides derived from the TisB type-I TA system were employed to establish initial structure–activity relationships via sequence truncation, cationic modifications, and incorporation of ultrashort PEG-like flexible linkers as well as Ala scanning and partial or full incorporation of D-amino acids. This led to the identification of several hits with promising antibacterial activity (MICs in the range 2–16 μM) against the Gram-negative E. coli, K. pneumoniae, P. aeruginosa, and A. baumannii, while their hemolytic properties were kept at an acceptable level (i.e., approximately 20–30% hemolysis at 400 μM). In contrast, the activity in the Gram-positive bacteria S. aureus and E. faecalis was lower (MICs in the range 8–64 μM). Intriguingly, synergy studies revealed that low micromolar concentrations (0.25–8 μM) of essentially nonhemolytic TisB-derived peptides (i.e., less than 10% hemolysis at 400 μM) could reduce the minimum inhibitory concentrations of azithromycin to therapeutically relevant levels in E. coli, K. pneumoniae, and P. aeruginosa. These findings highlight the potential of the TA system toxins as a source of antibiotic candidates.

Introduction

With antimicrobial resistance (AMR) rendering current treatments ineffective, the discovery of novel antibacterial agents is critically urgent. The clinical impact of AMR in ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp.) is especially severe, as they cause infections with high rates of complications and mortality. , In the search for novel antibiotics, antimicrobial peptides (AMPs) have attracted considerable attention, since they often exhibit broad-spectrum rapid bacterial killing that reduces their propensity for AMR development. The general interest in peptide-based drugs is significant with at least 114 peptides approved as therapeutics. Nevertheless, only a few AMPs (e.g., polymyxins, daptomycin, and bacitracin) have reached the market.

Traditional design strategies for AMP analogues with improved activity profiles comprise site-specific modification (e.g., with D-amino acids), lipidation, glycosylation, PEGylation, and fragment combination to provide hybrids, as well as multimerization and engineering of supramolecular AMP-based therapeutics. − Lipidation often confers enhanced potency, albeit typically with concomitant loss of cell selectivity and reduced compatibility with blood components. , PEGylation and glycosylation typically improves solubility and half-life in vivo, but it may result in reduced binding to bacterial membranes, thereby limiting activity, whereas cytotoxicity and hemolytic properties are diminished. , Recently, helical-wheel rotation of short helical AMPs enabled optimization of pharmacological properties. Most recently, machine-learning and deep-learning models as well as artificial intelligence (AI), including generative methods and large language models, have been explored as tools for predicting and optimizing activity profiles of AMPs. −

Bacterial toxin–antitoxin (TA) systems are genetic loci encoding a stable toxin and its corresponding antitoxin. − The cellular level of antitoxin can be downregulated in response to specific stimuli, including activation of the SOS DNA damage response. When the antitoxin level becomes insufficient to neutralize the toxin, various cellular responses are triggered, ranging from reduced metabolic activity to cell death; moreover, these systems contribute to plasmid maintenance and elimination of bacteriophage infections. − Although some initial research has been conducted (particularly on type-II TA systems), TA systems remain poorly explored with respect to their therapeutic potential. ,− One DNA damage-sensing system found in Escherichia coli is the tisB/istR loci. ,, The corresponding TisB toxin, encoded within this type-I TA system, is predicted to form an α-helix that may have pore-forming properties, thereby leading to dissipation of the proton motive force, which in turn reduces the availability of ATP, resulting in an altered metabolism toward a latent state. ,,− Notably, under native conditions, TisB is expressed at very low levels due to tight post-transcriptional regulation by the antitoxin istR; however, upon overexpression, TisB becomes highly toxic to the host bacterium. These inherent properties of TA-derived toxins may be applied in a strategy that combats multidrug-resistant pathogens.

The present work concerns peptide fragments derived from the TisB toxin. Initially, its active domain was determined by intracellular expression of truncated peptides, which allowed the identification of the minimal active sequence required to exert efficient intracellular killing. This principle has been utilized for other TA systems, such as AapA1, ShoB, and IbsC. , Building on the identified lead sequence, several peptide modifications were explored to achieve enhanced antibacterial activity while attempting to reduce the hemolytic properties.

The first optimization step was to increase the solubility of the initial toxin fragment having a high content of hydrophobic amino acids. Thus, cationic residues were incorporated either as N- or C-terminal blocks or via replacements within the sequence, which were expected also to promote interactions with the negatively charged bacterial membranes. − This approach ensured that most, if not all, of the active core of the peptide toxin remained intact. Other modifications included strategic relocation of cationic residues, insertion of flexible linkers, and positional substitutions. , An Ala scan was performed for the most promising first-generation optimized hit peptide, which guided our design of second-generation analogues.

Biological evaluation comprised an assessment of the antibacterial activity spectrum and basic safety of the peptides. Thus, minimum inhibitory concentrations (MICs) and minimum bactericidal concentrations (MBCs) against reference strains of the ESKAPE pathogens were determined. Also, the degree of hemolysis exerted by the peptides was measured at a high concentration, estimating their safety in vivo by using an optimized assay. Through this characterization in vitro, we aimed to identify hits with a desirable combination of potent antibacterial activity and minimal cytotoxicity, enabling future hit-to-lead development into potential candidates for novel antibacterial agents.

Results and Discussion

First-Generation Analogues

The native TisB toxin is a 29-residue peptide, but plasmid-based expression studies revealed that only the C-terminal part is essential for intracellular antibacterial activity (Figure S1), which is consistent with previous reports on type I toxins. , The hit sequence was truncated further by removing the IV residues at the N-terminus, providing the peptide fragment (bold/underlined part of the hit; Table ) that served as our starting point.

1. First-Generation Peptides: Physicochemical Properties,

no.	sequence	no. of AA	net charge
hit	MNLVDIAILI-LKLIV AALQLLDAVLKYLK –OH	29	+ 2
N-terminal Ala-Ala retained
1	RRRRRRAALQLLDAVLKYLK-NH2	19	+ 8
Cationic placement in truncated analogues
2a	RRRRRRLQLLDAVLKYLK-NH2	18	+ 8
2b	LQLLDAVLKYLKRRRRRR-NH2	18	+ 8
2c	KKKRRRLQLLDAVLKYLK-NH2	18	+ 8
2d	KKKKKKLQLLDAVLKYLK-NH2	18	+ 8
2e	LQLLDAVLKYLKKKKKKK-NH2	18	+ 8
2f	RRRLQLLDAVLKYLKRRR-NH2	18	+ 8
2g	KKKLQLLDAVLKYLKKKK-NH2	18	+ 8
3a	RRRLQLLDAVLRRR-NH2	14	+ 6
3b	RRRLQLLRRRKYLK-NH2	14	+ 9
Incorporation of spacers
4a	RRRRRRGSGSGSLQLLDAVLKYLK-NH2	24	+ 8
4b	LQLLDAVLKYLKGSGSGSRRRRRR-NH2	24	+ 8
5a	RRR-sPEG2-RRRLQLLDAVLKYLK-NH2	19	+ 8
5b	RR-sPEG2-RR-sPEG2-RRLQLLDAVLKYLK-NH2	20	+ 8
5c	[R-sPEG2-R]3-LQLLDAVLKYLK-NH2	21	+ 8
5d	RRRRRR-sPEG2-LQLLDAVLKYLK-NH2	19	+ 8
5e	RRRRRRLQLL-sPEG2-DAVLKYLK-NH2	19	+ 8
5f	RRRRRRLQLLDAVL-sPEG2-KYLK-NH2	19	+ 8
5g	RRRRRRLQLL-sPEG2-DAVL-sPEG2-KYLK-NH2	20	+ 8

^aAA = amino acids.

^bsPEG2 = 8-Amino-3,6-dioxaoctanoic acid.

The first set of analogues comprised subseries 1–5 (Table ). Initially, an N-terminal Arg₆ moiety (= R₆) was added to improve solubility and introduce a substantial positive charge (giving 1) in order to promote interactions with negatively charged bacterial membranes. , Omitting the next two N-terminal Ala residues while retaining an R₆ moiety provided 2a and 2b (Table ); the corresponding Arg/Lys- or Lys-based analogues (i.e., 2c–e) were included as well. Analogues with two Arg₃ or Lys₃ motifs (i.e., 2f and 2g) and truncated analogues 3a and 3b were also considered. , Next, a polar GSGSGS motif was incorporated as a spacer between the cationic and hydrophobic segments to give 4a and 4b. Finally, the ultrashort PEG-like ethylene glycol spacer (i.e., sPEG2 = 8-amino-3,6-dioxa-octanoic acid) was inserted in one or two positions (giving 5a–g), inspired by the favorable effect found for the activity profiles of peptidomimetics. Such alterations were expected to act as helix breakers that also confer reduced hydrophobicity and increased flexibility, which might reduce the hemolytic properties. ,

Subseries 1–5 were tested against wild-type strains of ESKAPE pathogens. The R₆-modified peptide 1 exhibited moderate potency across the test panel (MICs: 4–16 μM; Table ).

2. Antimicrobial and Hemolytic Activity of First-Generation Analogues in Wild-Type Strains.

	MIC (MBC) [μM]
no.	E. c.	K. p.	P. a.	A. b.	S. a.	E. f.	hemolysis (400 μM)	% MeCN	GRAVY
N-terminal Ala-Ala retained
1	4 (4)	8 (8)	16 (32)	4 (4)	16 (16)	8 (8)	40%	51.8%	–0.725
Cationic placement in truncated analogues
2a	1 (1)	2 (2)	8 (16)	2 (2)	8 (16)	16 (16)	64%	47.4%	–1.006
2b	2 (4)	8 (16)	8(16)	1 (1)	8 (8)	4 (4)	67%	47.3%	–1.006
2c	2 (2)	4 (4)	8 (8)	1 (1)	16 (16)	16 (16)	38%	46.6%	–0.906
2d	2 (2)	4 (4)	8 (8)	2 (2)	64 (64)	32 (32)	24%	46.1%	–0.806
2e	8 (8)	128 (128)	32 (32)	4 (4)	64 (64)	16 (16)	50%	45.7%	–0.806
2f	16 (32)	4 (4)	32 (32)	1 (1)	16 (16)	16 (16)	46%	45.5%	–1.006
2g	32 (64)	32 (6)	128 (128)	8 (8)	128 (128)	128 (128)	34%	44.3%	–0.806
3a	128 (128)	128 (128)	128 (128)	128 (128)	128 (128)	128 (128)	3%	38.2%	–0.914
3b	128 (128)	128 (128)	128 (128)	128 (128)	128 (128)	128 (128)	2%	30.4%	–1.743
Incorporation of spacers
4a	4 (4)	8 (8)	16 (32)	8 (8)	16 (16)	16 (16)	41%	50.3%	–0.904
4b	16 (16)	128 (128)	64 (64)	2 (2)	4 (8)	16 (16)	46%	47.2%	–0.904
5a	4 (4)	16 (16)	16 (16)	4 (4)	16 (16)	8 (16)	58%	47.7%
5b	8 (8)	32 (32)	32 (64)	16 (16)	32 (32)	16 (32)	42%	47.8%
5c	4 (8)	32 (64)	32 (32)	2 (4)	16 (16)	16 (16)	50%	48.1%
5d	128 (128)	128 (128)	128 (128)	16 (16)	128 (128)	128 (128)	6%	40.3%
5e	64 (128)	128 (128)	128 (128)	32 (32)	128 (128)	128 (128)	9%	40.5%
5f	128 (128)	64 (128)	128 (128)	8 (8)	128 (128)	128 (128)	5%	38.9%
5g	128 (128)	128 (128)	128 (128)	128 (128)	128 (128)	128 (128)	7%	34.3%

^a E. c., E. coli; K. p., K. pneumoniae; P. a., P. aeruginosa; A. b., A. baumannii; S. a., S. aureus; E. f., E. faecalis.

^bHPLC-based measure of hydrophobicity: percentage of MeCN at the peak of elution.

^c http://www.gravy-calculator.de.

^dDue to the content of unnatural residue(s), the GRAVY cannot be applied.

Analogue 2a, devoid of the two original N-terminal Ala residues, had improved or retained activity (cf. 1), and hence these residues were omitted in the subsequent optimization. In Gram-negative pathogens, analogue 2a had improved or retained MICs (i.e., 1–2 μM in E. coli, K. pneumoniae, and A. baumannii and 8 μM in P. aeruginosa), while in the Gram-positive S. aureus and E. faecalis, the MICs remained within a 2-fold change. Positioning of the R₆ moiety at the C-terminus (in 2b) also conferred lower or retained activity. Comparison of 2a and 2b showed the latter to be less active in E. coli and K. pneumoniae, whereas it was more potent in A. baumannii and E. faecalis, while 2a and 2b were equipotent in P. aeruginosa and S. aureus. Peptide 2c with a mixed Arg/Lys moiety had similar activity to 2a. While 2d with an N-terminal K₆ moiety was almost equipotent to both 2a and 2c against the Gram-negative bacteria, the C-terminally K₆-modified 2e had reduced activity (cf. 2d) in E. coli, P. aeruginosa, and A. baumannii and proved inactive in K. pneumoniae. Both 2d and 2e had moderate to weak activity (MICs: 16–64 μM) toward the Gram-positive bacteria. Peptides 2f and 2g, having an R₃ or K₃ moiety at both termini, had lower activity (cf. 2a– d). Peptides 3a,b, incorporating R₃ blocks instead of parts of the original toxin fragment, suffered from a complete loss of activity against all pathogens. Possibly, this is due to too low hydrophobicity estimated as the percentage of MeCN at peak of elution (denoted as % MeCN) in reversed-phase analytical HPLC. Hydrophobicity measured as HPLC retention and via the GRAVY parameters for peptides with identical amino acid composition (but different sequence) deviates somewhat when comparing modification with R₆ (or K₆) with two R₃ (or K₃), which is the main weakness of residue-based calculations. However, the peptides representing the upper and lower limits for hydrophobicity are for both measures the same (i.e., 1 and 3b).

For these subseries displaying different charge distribution, N-terminal modification with R₆ (in 2a) conferred the most favorable activity against Gram-negative bacteria (MICs: 1–8 μM), while C-terminal R₆ modification (in 2b) increased the potency against E. faecalis.

The effect of the [GS]₃ spacer (in 4a,b) depended on the positioning of the R₆ moiety. Thus, peptide 4a had increased MICs as compared to 2a except for E. faecium, in which 2a and 4a were equipotent. In contrast, peptide 4b had at least 8-fold decreased activity (cf. 2b) in three of the Gram-negative pathogens, while its MIC in E. faecalis was 4-fold higher, whereas its MICs remained within a 2-fold change in A. baumannii and S. aureus.

Peptides 5a–c, with sPEG2 residues incorporated into the cationic moiety, had 2- to 8-fold increased MICs (cf. 2a) against E. coli, P. aeruginosa, and A. baumannii, whereas their activity against K. pneumoniae was further reduced. S. aureus was less susceptible to 5a–c, while E. faecalis had unchanged susceptibility (cf. 2a). Conversely, peptides with an sPEG2 unit inserted either as a flexible linker (after R₆ in 5d) or as helix-breaking residue(s) within the hydrophobic segment (in 5e–g) proved to lack activity (MICs: 64–128 μM) in all bacteria, except for A. baumannii in which peptides 5d–f had MICs of 8–32 μM. Loss of activity due to disruption of the hydrophobic segment (also seen for 3a,b) indicates that this segment is critical for antibacterial activity.

For the subseries containing spacer moieties, the general trend was reduced activity as compared to 2a and 2b, although analogues 4a and 5a retained moderate activity (i.e., MICs within the range 4–16 μM) across the panel. Thus, insertion of a [GS]₃ spacer between the N-terminal cationic part and the hydrophobic segment or an sPEG2 residue within the N-terminal cationic part could both be accommodated without severe loss of activity.

Omitting the two Ala residues in peptide 1 (giving 2a) resulted in increased hemolytic properties (i.e., 64% vs 40% for 1 at 400 μM; Table ). Hemolytic properties of 2a and 2b were similar, indicating that positioning of R₆ was not of importance. Peptides 2a, 2c, and 2d, with N-terminal R₆, K₃R₃, and K₆, respectively, exhibited gradually decreased hemolysis (64%, 36%, and 24%), exemplifying that Arg→Lys alterations reduce hemolytic activity. − Dividing the R₆ moiety into two R₃ motifs (in 2f) also resulted in decreased hemolytic properties (46% vs 64–67% for 2a,b). Among Lys-containing peptides 2d,e (with K₆) and 2g (with two K₃), the latter exhibited an intermediate degree of hemolysis. Overall, the hemolytic properties reflected the hydrophobicity as measured by the HPLC retention except for 2d–f that appeared not to follow a clear trend. In contrast, the GRAVY values only clearly inferred the least hydrophobic 3b to possess the weakest hemolytic properties.

In contrast, [GS]₃-modified peptides 4a,b exhibited only moderately decreased hemolysis (41–46% vs 64–67% for 2a,b). Moreover, peptides 5a–c, with sPEG2 residue(s) incorporated into the cationic moiety, remained hemolytic (42–58% at 400 μM). Conversely, when sPEG2 was linking R₆ and the hydrophobic segment (in 5d) or was introduced within the hydrophobic segment (in 5e–g) the degree of hemolysis was reduced considerably (to less than 10%), albeit with concomitant loss of antibacterial activity, which may be ascribed to their substantially lowered hydrophobicity as measured by their HPLC retention. Similar GRAVY values for 4a and 4b were reflected in slightly deviating hemolytic properties, whereas HPLC-based retention lacked correlation. For series 5a– g, GRAVY values could not be obtained, but generally decreased % MeCN was correlated with lowered hemolysis among these analogues.

Within the entire set of first-generation peptides (i.e., series 2–5; Table ), the best balance between antibacterial potency and undesired hemolytic properties was found for the K₆-modified 2d when focusing on activity toward Gram-negative pathogens (MICs of 2–8 μM; 24% hemolysis at 400 μM), whereas K₃R₃-modified 2c retained moderate activity (MICs of 16 μM) with a somewhat higher degree of hemolysis (38% at 400 μM). Both 2c and 2d had a slightly lowered hydrophobicity as compared to the R₆-modified 2a and 2b based on HPLC retention (i.e., 46.1/46.6% MeCN vs 47.3/47.4% MeCN). In this case, the GRAVY values pointed to the nonobserved opposite trend, which indicate that for the present subset of compounds, HPLC-based estimation of hydrophobicity appears most useful.

Second-Generation Analogues

Peptide 2a had the most promising antibacterial profile (Table ), and therefore, its structure–activity relationships (SARs) were explored. Second-generation analogues comprised an initial Ala scan to identify key residues within the hydrophobic segment, while R₆ was kept unchanged (Table ). However, the resulting peptides 6–16 retained an antibacterial activity profile resembling that of 2a, albeit with a few minor deviations (Table ).

3. Second-Generation Peptides: Physicochemical Properties .

no.	sequence	no. of AA	net charge
2a	RRRRRRLQLLDAVLKYLK-NH2	18	+ 8
6	RRRRRRAQLLDAVLKYLK-NH2	18	+ 8
7	RRRRRRLALLDAVLKYLK-NH2	18	+ 8
8	RRRRRRLQALDAVLKYLK-NH2	18	+ 8
9	RRRRRRLQLADAVLKYLK-NH2	18	+ 8
10	RRRRRRLQLLAAVLKYLK-NH2	18	+ 9
11	RRRRRRLQLLDAALKYLK-NH2	18	+ 8
12	RRRRRRLQLLDAVAKYLK-NH2	18	+ 8
13	RRRRRRLQLLDAVLAYLK-NH2	18	+ 7
14	RRRRRRLQLLDAVLKALK-NH2	18	+ 8
15	RRRRRRLQLLDAVLKYAK-NH2	18	+ 8
16	RRRRRRLQLLDAVLKYLA-NH2	18	+ 7

^aAA: amino acid.

^bPositively charged moieties and Ala residues are highlighted in bold.

4. Antibacterial Activity and Hemolytic Properties of Second-Generation Analogues.

	MIC (MBC) [μM]
no.	E. c.	K. p.	P. a.	A. b.	S. a.	E. f.	hemolysis (400 μM)	% MeCN	GRAVY
2a	1 (1)	2 (2)	8 (16)	2 (2)	8 (16)	16 (16)	64%	47.4%	–1.006
6	2 (2)	2 (2)	8 (16)	2 (2)	16 (16)	16 (16)	74%	44.7%	–1.117
7	2 (2)	2 (2)	8 (8)	1 (2)	8 (8)	16 (16)	32%	47.7%	–0.711
8	2 (2)	4 (8)	16 (16)	1 (1)	32 (32)	16 (16)	31%	43.2%	–1.117
9	2 (2)	2 (4)	8 (16)	1 (1)	32 (32)	16 (16)	24%	42.7%	–1.117
10	4 (4)	2 (2)	8 (8)	2 (4)	4 (8)	8 (8)	84%	48.3%	–0.711
11	2 (2)	2 (2)	8 (8)	1 (1)	16 (32)	16 (16)	37%	43.4%	–1.139
12	2 (2)	4 (8)	4 (4)	2 (2)	16 (16)	32 (32)	51%	42.8%	–1.117
13	2 (2)	2 (2)	8 (8)	2 (2)	8 (8)	16 (16)	78%	50.2%	–0.689
14	2 (2)	8 (8)	4 (8)	2 (2)	8 (8)	8 (8)	67%	47.1%	–0.833
15	2 (2)	4 (4)	8 (8)	1 (1)	64 (64)	8 (8)	22%	42.4%	–1.117
16	4 (4)	4 (4)	16 (16)	4 (4)	16 (16)	16 (16)	69%	50.0%	–0.689

^a E. c.: E. coli; K. p.: K. pneumoniae; P. a.: P. aeruginosa; A. b.: A. baumannii; S. a.: S. aureus; E. f.: E. faecalis.

^bHPLC-based measure of hydrophobicity: percentage of MeCN at peak of elution.

^c http://www.gravy-calculator.de/.

Within a 2-fold difference, most peptides had potency similar to the parent 2a against E. coli, K. pneumoniae, and A. baumannii (typical MICs: 1–2 μM). Notably, peptide 16, being among the most hydrophobic analogues, had a slightly increased MIC (4 μM) in these bacteria. Against K. pneumoniae, only peptide 14 exhibited a 4-fold increased MIC (cf. 2a). Otherwise, the potency remained essentially unchanged for most peptides: 12 and 14 had 2-fold decreased MICs in P. aeruginosa, whereas 8 and 16 had 2-fold increased MICs.

A few analogues (8, 9, and 15) proved less active (MICs: 32–64 μM) against S. aureus. Notably, these peptides possess a Leu→Ala substitution, reflected in a lower hydrophobicity (in agreement with both HPLC retention and GRAVY values), which infers that these Leu residues only may be replaced with other hydrophobic amino acids if retained activity against Gram-positive bacteria is desired. Peptide 10, with an Asp→Ala substitution that both increased net charge and hydrophobicity (in agreement with both HPLC retention and GRAVY values) was slightly more potent toward S. aureus and E. faecalis (MICs: 4 and 8 μM). Nevertheless, peptides 14 and 15, differing markedly in hydrophobicity (47.1 vs 42.4% MeCN and −0.833 vs −1.117 on the GRAVY scale), both had 2-fold lowered MICs against E. faecalis, emphasizing the complexity of the SAR data.

Parent 2a was relatively hemolytic at 400 μM, which was considered an appropriate test concentration (i.e., 100- to 400-fold higher than the best MICs; Table ) to estimate whether it was possible to reach a safety window approaching that of clinically used antibiotics. Generally, changes in hemolytic properties, induced by Ala replacements, appeared explainable by altered hydrophobicity (for most peptides, an agreement between HPLC retention and GRAVY values was observed). A few peptides had increased hemolytic properties (cf. 2a): peptides 10 (84%), 13 (78%), 6 (74%), and 16 (69%). Thus, in peptides 10, 13, and 16, charged amino acids (Asp or Lys) were replaced with Ala, conferring increased hydrophobicity (48.3–50.2% MeCN and approximately −0.7 on the GRAVY scale vs 2a: 47.4% MeCN and approximately −1.0 on the GRAVY scale) that typically was associated with increased hemolytic properties. Moreover, peptides 13 and 16 had reduced overall charge (Table ), which also may contribute to their increased hemolytic properties. Interestingly, helical-wheel projections (see pages S3–S5 in the SI) did not reveal major differences in amphipathicity. Encouragingly, almost 3-fold reduced hemolysis was seen for peptides 9 and 15 with Leu→Ala replacements, resulting in the lowest hydrophobicity within this subseries (Table ). Also, ∼2-fold lowered hemolysis was found for peptides 7, 8, and 11, of which the last two had low hydrophobicity (43.2% and 43.4% MeCN; −1.12 and −1.14 on the GRAVY scale). Expectedly, within this subseries, low hydrophobicity was associated with reduced hemolytic properties.

Within the set of second-generation (Ala-substituted) analogues, peptides 9 and 15 proved to be the least hemolytic (22–24% at 400 μM) while retaining similar activity in Gram-negative bacteria as the original 2a (i.e., MICs in the range 1–8 μM). Both peptides were among the least hydrophobic within this subseries (as measured by both HPLC and GRAVY). The best antibacterial activity across the entire panel (including S. aureus) was found for peptide 7 (MICs in the range 1–16 μM), which despite a significantly higher hydrophobicity (as measured by HPLC and GRAVY) only was slightly more hemolytic (32% at 400 μM).

Third-Generation Analogues

Among second-generation analogues, peptides 9 and 15 had the most promising activity profile (Table ). The alterations made in these peptides were starting points for a third subseries (17–19; Table ). Peptide 17a was double-substituted with Leu10→Ala and Leu17→Ala, while further increased polarity was achieved by insertion of Ser instead of each of these Ala residues (to give 17b-c), whereas peptide 17d displayed both Ala10→Ser and Ala17→Ser alterations. Finally, peptides with a single Ala→Ser substitution were included (i.e., 18 and 19). Another approach to further reduce the hemolytic properties involved replacement of R₆ with K₆ to give 9a and 15a.

5. Third-Generation Peptides: Physicochemical Properties .

no.	sequence	no. of AA	net charge
Double-Ala and Ala→Ser substituted analogues
17a	RRRRRRLQLADAVLKYAK-NH2	18	+ 8
17b	RRRRRRLQLADAVLKYSK-NH2	18	+ 8
17c	RRRRRRLQLSDAVLKYAK-NH2	18	+ 8
17d	RRRRRRLQLSDAVLKYSK-NH2	18	+ 8
18	RRRRRRLQLSDAVLKYLK-NH2	18	+ 8
19	RRRRRRLQLLDAVLKYSK-NH2	18	+ 8
Lys-modified Ala-substituted analogues
9a	KKKKKKLQLADAVLKYLK-NH2	18	+ 8
15a	KKKKKKLQLLDAVLKYAK-NH2	18	+ 8
All-D or partly-D analogues,
9b	RRRRRRlqladavlkylk-NH2	18	+ 8
9c	rrrrrrLQLADAVLKYLK-NH2	18	+ 8
9d	rrrrrrlqladavlkylk-NH2	18	+ 8
15b	RRRRRRlqlldavlkyak-NH2	18	+ 8
15c	rrrrrrLQLLDAVLKYAK-NH2	18	+ 8
15d	rrrrrrlqlldavlkyak-NH2	18	+ 8

^aAA: amino acids.

^bPositively charged moieties and Ala/Ser residues are highlighted in bold.

^cD-amino acids are denoted by noncapitalized 1-letter codes.

To obtain an indication of whether the mode of action of these peptides involved stereospecific interactions, partial or full L→D replacement was performed. These analogues comprised peptides 9b and 15b retaining all-L R₆ moieties combined with an all-D hydrophobic segment and peptides 9c and 15c with reversed stereochemistry of these segments, as well as all-D peptides 9d and 15d.

Subseries 17–19 were explored as to whether antibacterial activity could be retained while hemolytic properties were reduced. Nevertheless, the double-substituted 17a–d lacked potency toward all bacteria (MICs: 128 μM) except for A. baumannii (MICs: 8–32 μM). Conversely, single-substituted 18 and 19 exhibited only 2- to 8-fold higher MIC values (cf. those of 9 and 15) against E. coli, K. pneumoniae, and A. baumannii (Tables and ), whereas they had weak activity toward P. aeruginosa and the Gram-positive bacteria (MICs: 32–128 μM). Likewise, replacement of R₆ with K₆ also conferred reduced activity to peptides 9a and 15a against the Gram-negative pathogens, while they lacked activity in Gram-positive bacteria.

6. Antibacterial Activity and Hemolytic Properties of Third-Generation Analogues.

	MIC (MBC) [μM]
no.	E. c.	K. p.	P. a.	A. b.	S. a.	E. f.	hemolysis (400 μM)	% MeCN	GRAVY
Parent peptides from 1st and 2nd generation analogues
2a	1 (1)	2 (2)	8 (16)	2 (2)	8 (16)	16 (16)	64%	47.4%	–1.006
9	2 (2)	2 (4)	8 (16)	1 (1)	32 (32)	16 (16)	24%	42.7%	–1.117
15	2 (2)	4 (4)	8 (8)	1 (1)	64 (64)	8 (8)	22%	42.4%	–1.117
Double-Ala and Ala→Ser substituted analogues
17a	128 (128)	128 (128)	128 (128)	8 (16)	128 (128)	128 (128)	6%	36.8%	–1.228
17b	128 (128)	128 (128)	128 (128)	16 (16)	128 (128)	128 (128)	4%	34.9%	–1.372
17c	128 (128)	128 (128)	128 (128)	16 (32)	128 (128)	128 (128)	13%	34.5%	–1.372
17d	128 (128)	128 (128)	128 (128)	32 (32)	128 (128)	128 (128)	9%	30.2%	–1.517
18	8 (16)	16 (32)	128 (128)	2 (2)	64 (128)	128 (128)	12%	39.1%	–1.261
19	8 (8)	16 (32)	32 (64)	2 (2)	64 (64)	128 (128)	14%	39.6%	–1.261
Lys-modified Ala-substituted analogues
9a	8 (16)	16 (16)	32 (32)	16 (16)	128 (128)	128 (128)	14%	41.5%	–0.917
15a	8 (16)	64 (64)	8 (16)	8 (16)	128 (128)	128 (128)	12%	40.7%	–0.917
All-D and partial-D analogues
9b	4 (4)	8 (8)	16 (16)	4 (4)	32 (32)	32 (32)	40%	42.4%	–1.117
9c	2 (4)	8 (8)	16 (16)	2 (4)	16 (16)	32 (64)	37%	42.7%	–1.117
9d	4 (4)	4 (4)	8 (16)	2 (2)	16 (32)	16 (16)	57%	42.3%	–1.117
15b	4 (4)	8 (8)	16 (16)	2 (4)	32 (64)	32 (32)	29%	42.9%	–1.117
15c	4 (4)	8 (8)	16 (16)	2 (2)	64 (64)	64 (64)	31%	43.0%	–1.117
15d	2 (4)	4 (4)	16 (16)	1 (1)	32 (32)	8 (8)	52%	42.0%	–1.117

^a E. c.: E. coli; K. p.: K. pneumoniae; P. a.: P. aeruginosa; A. b.: A. baumannii; S. a.: S. aureus; E. f.: E. faecalis.

^bHPLC-based measure of hydrophobicity: percentage of MeCN at peak of elution.

^c http://www.gravy-calculator.de/.

^dAssuming that partial substitution with D-amino acid residues does not affect the hydrophobicity.

Encouragingly, peptides 9b–d and 15b–d with different degrees of L→D replacement retained almost full activity (cf. 9 and 15). Thus, MICs for 9b–d against Gram-negative pathogens were in the range 2–16 μM (cf. 1–8 μM for 9), while in Gram-positive bacteria the MICs were identical for 9 and 9b–d. Also, MICs for 15 and 15b–d were comparable in all species. P. aeruginosa was the Gram-negative bacterium least susceptible to 9b-d and 15b–d (MICs: 8–16 μM), while K. pneumoniae, E. coli, and A. baumannii were increasingly susceptible (MICs: 4–8, 2–4, and 1–4 μM). All-D analogues 9d and 15d were the most potent across the entire test panel. Specifically, peptides 9c and 15d were most active toward E. coli (MICs: 2 μM), while 9d and 15d were most potent against K. pneumoniae (MICs: 4 μM). In P. aeruginosa, 9d was most active (MIC: 8 μM), while 9c and 9d were most potent against S. aureus (MICs: 16 μM), whereas 15d was most active toward A. baumannii and E. faecalis (MICs: 1 and 8 μM). These findings indicate a mode of action that most likely is independent of the stereochemistry of the cationic and hydrophobic parts, albeit optimal potency was seen for all-D 9d and 15d. Moreover, a simple α-helical binding mode may be ruled out on the basis of these findings.

Peptides 9 and 15 as well as 9b– d and 15b– d, with a hydrophobicity of 42–43% MeCN (and identical GRAVY values of −1.117), exhibited potent to moderate antibacterial activity against Gram-negative pathogens (MICs: 1–16 μM; Tables and ), whereas the less hydrophobic peptides 9a and 15a (40.7–41.5% MeCN and GRAVY values of −0.917) were less active (MICs: 8–32 μM). Thus, the most potent activity in E. coli was generally seen above a critical hydrophobicity threshold at ∼41.5–42% MeCN (Figure ), explaining the weak activity of peptides 17a–d (eluting below 37% MeCN), while peptides 18 and 19 with intermediate hydrophobicity (39.1 and 39.6% MeCN) had moderate potency.

1.

Plots illustrating the influence of hydrophobicity on activity and hemolysis. All compounds tested in the present work are included. Upper panel: plot of MICs (E. coli) vs hydrophobicity (estimated as % MeCN at peak of elution in HPLC); lower panel: plot of hemolysis (at 400 μM) vs % MeCN.

Peptides 17a– d were weakly hemolytic (4–13% hemolysis at 400 μM) but with concomitant loss of antibacterial activity. This activity profile appears to arise from their lowered hydrophobicity (30–37% MeCN and GRAVY values below −1.2) as compared to the parent 9 and 15 (42–43% MeCN and GRAVY values of approximately −1.1) that were more hemolytic (24% and 22% hemolysis at 400 μM). Hence, two polarity-enhancing substitutions (i.e., Leu→Ala/Ser) abolished their membrane-disruptive properties toward both bacterial and mammalian cells. The same trend was seen for Lys-based 9a and 15a vs Ser-substituted 18 and 19.

Peptides 9b,c and 15b,c, containing an all-D part, were more hemolytic than all-L 9 and 15 (29–40% vs 22–24%), whereas all-D 9d and 15d were even more hemolytic (>50% at 400 μM). Thus, all-D peptides may resist degradation by enzymes secreted from (or present on the surface of) erythrocytes during the assay, whereas partial degradation of all-L peptides into less hemolytic fragments may occur.

While the all-L Ala/Ser-substituted third-generation analogues (i.e., 9a, 15a, and 17a– d) essentially are inactive against the Gram-positive species, peptides 9a and 19 possess the best balance between activity in Gram-negative pathogens and hemolytic properties (MICs in the range 2–32 μM and 14% hemolysis at 400 μM). Interestingly, even though the K₆-modified 9a was more hydrophobic than 19 (i.e., 41.5 vs 39.6% MeCN and GRAVY values of −0.9 vs −1.3), they were equally hemolytic, possibly reflecting that Arg generally is more strongly membrane-interacting than Lys.

Among the analogues (9b–d and 15b–d) with partial or full L→D replacement in one or both segments (i.e., the cationic moiety and original hydrophobic toxin fragment), 15b and 15c with mixed stereochemistry possess the most favorable activity profiles: MICs of 2–16 μM against Gram-negative pathogens combined with the weakest hemolytic properties within this subset (i.e., 29–31% hemolysis at 400 μM). These peptides possess identical moderate hydrophobicity (ca. 43% MeCN and a GRAVY value of −1.1).

Antibiotic Synergy Studies

Rapid spreading of AMR necessitates both development of novel treatment options and optimal use of existing antibiotics. The latter may be achieved by identifying synergistic combinations of AMPs and antibiotics with balanced efficacy vs toxicity, enabling potential clinical application. Antibiotic combination therapy is already applied in clinical settings for difficult-to-treat infections. , Typically, antibiotic synergy is assessed by using the fractional inhibitory concentration index (FICI): FICI = [MIC_A,comb/MIC_A,alone] + [MIC_B,comb/MIC_B,alone], where FICI ≤ 0.5 represents synergy.

To ensure that only clinically relevant AMP-antibiotic combinations would be identified, a strict selection criteria was applied for inclusion in the checkerboard assays: only AMPs with less than 10% hemolysis at 400 μM (except for 18 giving rise to 11% hemolysis) were considered.

Antibiotics comprised rifampicin (RIF) and azithromycin (AZT), which primarily are employed to treat infections with mycobacteria or Gram-positive bacteria (e.g., S. aureus), respectively. These antibiotics constitute well-known examples of hydrophobic anti-Gram-positive antibiotics that can be repurposed to exhibit potent activity against Gram-negative pathogens, in which pre-existing resistance is less likely. Thus, previous studies employing membrane-permeabilizing compounds demonstrated that both RIF and AZT can be potentiated by AMPs and peptidomimetics in Gram-negative bacteria. − To be of clinical relevance, potentiated MIC values of the antibiotic should be close to or below its clinical breakpoint: for AZT, this is 4 μg/mL against E. coli,, while for RIF, it is 0.06 μg/mL against Staphylococcus spp.

Peptides exhibiting synergy with rifampicin in E. coli (FICI < 0.5) are shown in Table . Combinations including peptides 17a and 18 (at 2 and 1 μM) enabled a lowering of the MIC for RIF to 0.50 μg/mL (FICI of 0.19). Although 18 exhibited synergy at the lowest concentration, it was more hemolytic (Table ). Weakly hemolytic peptides 3a,b and 17b (2–4% at 400 μM) also lowered the MIC for RIF to 0.50 μg/mL, when coapplied at 4, 2, and 4 μM, respectively (∼ FICIs of 0.09 or 0.13). The best potentiation of RIF (to a MIC of 0.25 μg/mL) was achieved with peptides 5d and 5f (at 2 and 4 μM), reaching FICIs of 0.09 and 0.16. Notably, peptides 5d and 5f were weakly hemolytic (5–6%) even at 200- and 100-fold higher concentrations than applied in the combinations with RIF. However, RIF did not reach its clinical breakpoint (0.06 μg/mL) in any of the combinations.

7. Synergy between Rifampicin (RIF) and Selected Peptides in E. coli .

	MIC alone		MIC in combination
no.	RIF (μg/mL)	peptide (μM)	RIF (μg/mL)	peptide (μM)	FICI	hemolysis (at 400 μM)
3a	8	128	0.50	4	0.09	3%
3b	8	32	0.50	2	0.13	2%
5d	8	32	0.25	2	0.09	6%
5f	8	32	0.25	4	0.16	5%
17a	8	16	0.50	2	0.19	6%
17b	8	64	0.50	4	0.13	4%
18	8	8	0.50	1	0.19	11%

^aRatios between MICs (either in μg/mL or μM) infer FICs to be independent of the units.

AZT was also tested against E. coli in combination with these peptides as well as analogue 17d (Table ), whereas peptide 18 proved not to exert synergy with AZT (Table S2 in the Supporting Information). Peptides 3a,b (at 8 μM) as well as 5f and 17a (at 4 μM) lowered the MIC for AZT to 0.50 μg/mL (∼ FICIs of 0.19–0.25). These peptides were all weakly hemolytic (2–6% at 400 μM). Moreover, peptides 5d (at 4 μM), 17b (at 8 μM), and 17d (at 16 μM) lowered the MIC of AZT 16-fold to 0.25 μg/mL (∼ FICIs of 0.13–0.19); also these peptides were low-hemolytic (4–9% at 400 μM).

8. Synergy between Azithromycin (AZT) and Selected Peptides in E. coli .

	MIC alone		MIC in combination
no.	AZT (μg/mL)	peptide (μM)	AZT (μg/mL)	peptide (μM)	FICI	hemolysis (at 400 μM)
E. coli
3a	4	64	0.50	8	0.25	3%
3b	4	64	0.50	8	0.25	2%
5d	4	32	0.25	4	0.19	6%
5f	4	64	0.50	4	0.19	5%
17a	4	32	0.50	4	0.25	6%
17b	4	128	0.25	8	0.13	4%
17d	4	128	0.25	16	0.19	9%
K. pneumoniae
5f	4	16	1	0.5	0.28	5%
17a	4	8	1	0.25	0.28	6%
P. aeruginosa
5f	16	32	4	2	0.31	5%

^aRatios between MICs (either in μg/mL or μM) infer FICs to be independent of the units.

In addition, a subset of these peptides (i.e., 3a, 3b, 5f, and 17a) were analyzed for potential synergy with rifampicin and azithromycin in K. pneumoniae and P. aeruginosa; however, only 5f exhibited synergy with AZT in both species, while 17a and AZT had a synergistic effect in K. pneumoniae.

Overall, AZT combinations with the low-hemolytic peptides 3a,b (only 2–3% hemolysis at a 50-fold higher concentration than applied for potentiation) as well as 5d and 17b are considered most promising in E. coli, since MIC of AZT in these combinations were 16-fold below its clinical breakpoint (4 μg/mL in E. coli). In neither K. pneumoniae nor P. aeruginosa, there is an official clinical breakpoint for AZT. Nevertheless, the considerable lowering (8- to 16-fold) of the MICs of AZT toward these bacteria to or below its breakpoint in E. coli may be of clinical value. In particular, the very low potentiating concentrations of 5f and 17a (i.e., 0.5 and 0.25 μM, respectively) may indicate a potential for repurposing AZT, e.g., as a therapy against K. pneumoniae lung infections.

All peptides identified as potentiators of RIF and/or AZT have a high content of Arg residues, which appears important for facilitating improved uptake in Gram-negative bacteria, where the outer polar lipopolysaccharide layer serves as a barrier for penetration of such hydrophobic antibiotics. Thus, a contributing factor in the mechanism for potentiation may well be membrane permeabilization arising from simultaneous interactions with multiple Arg residues, although this occurs at much lower concentrations than their MICs. The low inherent antibacterial activity of peptides 3a and 3b may be related to the short lengths of their hydrophobic segments that may limit their capacity for pore formation. For 5f, it may be hypothesized that the incorporated helix-breaking sPEG2 unit induces flexibility that is not compatible with membrane disruption via pore formation.

Conclusions

In total, 44 peptides, derived from a TisB toxin fragment, were studied. Following hit identification and hit-to-lead optimization, analogues with promising antibacterial activity against Gram-negative pathogens were identified. Notably, Arg-based peptide 2a and its Lys-based counterpart 2d demonstrated MICs in the range 1–8 μM against E. coli, K. pneumoniae, P. aeruginosa, and A. baumannii. A few second-generation analogues from the Ala scan of 2a (i.e., 9 and 15) exhibited reduced hemolysis while retaining activity against Gram-negative pathogens. However, further improvement of selectivity is required prior to testing in animal models, and thus future studies will focus on further reduction of the hemolytic properties.

Intriguingly, TA-derived peptides proved effective as potentiators of antibiotics (rifampicin and azithromycin). Specifically, four essentially nonhemolytic peptides (3a,b, 5d, and 17b) proved capable of reducing the concentration of azithromycin to below its clinical breakpoint when coapplied at low micromolar concentrations. Collectively, our findings highlight the potential of a design approach based on hit peptides derived from TA systems. Future studies of TA-derived AMPs will focus on their mechanism, which so far is hypothesized to involve membrane-disruptive interactions that are independent of stereochemistry, since partial or full L→D replacement in one or both segments (i.e., the cationic moiety and original hydrophobic toxin fragment) only had a minor effect on the antibacterial potency.

Materials and Methods

General

Starting materials and solvents were purchased from commercial suppliers and used without purification. Water for HPLC was filtered through a 0.22 μm filter on an Evoqua LaboStar Pro TWF UV system. Preparative HPLC was performed on Phenomenex Luna Omega Polar C₁₈ columns (250 × 30 mm or 250 × 21.2 mm; particle size: 5 μm; pore size 100 Å) using a Shimadzu Prominence system. Gradient elution was performed with eluents A (5:95 MeCN–H₂O + 0.1% TFA) and B (95:5 MeCN–H₂O + 0.1% TFA) with UV detection at 220 nm. Depending on the compound and column size, gradients of either 0–30%B, 0–50%B, 0–60%B, 20–60%B, or 20–50%B over 20 min (flow rate: 20 mL/min; 21.2 mm column) or 30 min (flow rate: 40 mL/min; 30 mm column) were applied. Purity was tested via analytical HPLC on a Phenomenex Luna C₁₈ HST column (same eluents as for preparative HPLC) using a 0–60% gradient during 10 min (flow rate: 0.5 mL/min; UV detection at 220 nm).

Automated Synthesis of Peptides

Peptides were prepared by Fmoc-based SPPS (0.1 mmol scale) either on a microwave-assisted CEM Liberty Blue synthesizer or on a Gyros Protein Technologies synthesizer. An H-Rink-amide resin (Matrix Innovation; loading 0.50 mmol/g; 100–200 mesh) was used as solid support. Coupling of N^α-Fmoc-protected building blocks with acid-labile side-chain protecting groups (5.0 equiv for the CEM Liberty Blue and 3.0 equiv for the Gyros Protein Technologies) was performed with N,N-diisopropylcarbodiimide (0.5 M in DMF; 5.0 equiv) and ethyl (hydroxyimino)­cyanoacetate (0.5 M in DMF, 5.0 equiv). Fmoc-Arg­(Pbf)–OH was triple-coupled at rt for 30 min followed by 75 °C for 2 min on the CEM Liberty Blue, while 50 °C for 10 min was applied on the Gyros synthesizer. Other building blocks were double-coupled for 10 min (at 75 or 50 °C on Liberty Blue and Gyros, respectively), while Fmoc-PEG2-OH was single-coupled. Fmoc deprotection: 20% piperidine-DMF (2 × 3 min at 75 °C). Side-chain deprotection and cleavage were performed with TFA-H₂O-triisopropylsilane (95:2.5:2.5; 2 × 40 min, each with 3 mL) followed by elution of the resin with CH₂Cl₂ (3 mL); the combined filtrates were concentrated in vacuo, and the crude was purified by preparative HPLC. Identity of fractions was verified by MALDI-TOF MS, and their purity was determined by analytical HPLC (pages S9–S24 in the SI); fractions with appropriate purity were lyophilized.

Bacterial Strains

Bacterial reference strains, originating from the American Type Culture Collection (ATCC), were obtained from the Department of Microbiology, Oslo University Hospital, and comprised E. coli ATCC 25922, K. pneumoniae ATCC 13883, A. baumannii ATCC 17978, S. aureus ATCC 29213, and E. faecalis ATCC 29212.

Preparation of Inoculum

Bacteria were plated on Mueller-Hinton (MH) II agar plates and incubated at 37 °C overnight. The next day, single colonies were suspended in sterile-filtered saline (0.9%). The optical density (OD) of each bacterial suspension was adjusted to McFarland standard 0.5 by using a spectrophotometer (UV-1800, Shimadzu) followed by 1:100 dilution in MH II broth to reach 10⁶ colony-forming units (CFU) per mL. The inoculum suspensions were used in MIC, MBC, and combination assays.

Minimum Inhibitory Concentrations (MICs)

Peptides were serially diluted (2-fold) directly into 96-well plates (Greiner, Microplate, 96-well, PP, U-Bottom, Natural, 650261) containing MH II broth, giving concentrations in the range 256–0.5 μM. An equal volume of bacterial suspension was added to each well, resulting in a final inoculum of 5 × 10⁵ CFU/mL and final peptide concentrations of 128–0.25 μM. The plates were incubated for 18–24 h at 37 °C. Then, the OD₆₀₀ of each well was measured in a plate reader (Victor Nivo, PerkinElmer). MICs were identified as the lowest concentration of peptide producing OD₆₀₀ measurements similar to control wells without bacteria. Experiments were performed in biological triplicate (each in technical replicate). The MIC values reported in this study correspond to the highest concentration observed across all replicates, when minor 2-fold variations were noted. This provides a conservative estimate of antimicrobial activity under the conditions applied.

Minimum Bactericidal Concentrations (MBCs)

To assess MBCs, 5 μL from each well of the 96-well plate used for MIC analysis were plated onto MH II agar and incubated at 37 °C for 16 h. The day after, the agar plates were visually inspected to identify the lowest concentration of peptide that prevented regrowth of bacteria. Experiments were performed in biological triplicate (each in technical triplicate).

Hemolysis Assay

Hemolytic activity was evaluated using an established protocol. Briefly, human erythrocytes from freshly collected peripheral blood were washed in PBS and diluted to a 1% suspension. Equal volumes of the erythrocyte suspension and peptide solutions were incubated for 1 h at 37 °C in 96-well plates. Following centrifugation, supernatants were transferred to clear plates, and hemoglobin release was quantified at 405 nm. Hemolysis (%) was determined relative to PBS (0%) and 10% Triton X-100 (100%) controls. Assays were performed in biological triplicate (each in technical triplicate).

This study was conducted in accordance with the principles of the Declaration of Helsinki and the applicable Norwegian regulations. Human blood samples were obtained from healthy volunteers, following written informed consent. The use of human blood for in vitro experimental purposes was reviewed by the Regional Committees for Medical and Health Research Ethics (REK), Norway (application no. 538503), and was determined not to fall within the scope of the Norwegian Health Research Act. No interventions were performed on human participants as part of this study.

Combination Assay

Synergy of antibiotic-peptide combinations was examined via checkerboard assays. In a 96-well plate, 50 μL of MH II was added to column 1 as negative control. Peptides were serially diluted (2-fold) from column 3 starting at 128 μM (with 25 μL in each well), resulting in an initial concentration of 32 μM after 4× dilutions (subsequently with antibiotic solution and bacterial suspension). Then, 25 μL of MH II was added in rows A1–A12. A serial dilution (2-fold) of the antibiotic was made in Eppendorf tubes starting at a concentration of 32 μM, and 25 μL was added to the remaining wells, excluding column 1 and rows A1–A12. Also, 50 μL of bacterial suspension was added to each well, resulting in a final inoculum of 5 × 10⁵ CFU/mL. The plates were incubated for 18–24 h at 37 °C. Then, the OD₆₀₀ of each well was measured in a plate reader (Victor Nivo, PerkinElmer). For wells producing OD₆₀₀ measurements similar to control wells containing antibiotics only (and without visible growth), FICI was calculated. Experiments were performed in biological duplicate.

Plasmid-Based Truncation Assay

Constructs encoding the full-length TisB peptide and its truncated derivatives were synthesized by Genscript Inc. (Piscataway, NJ, USA) using the pET28b­(+) plasmid as the expression scaffold. To minimize the appearance of suppressor variants caused by basal toxin leakage, the plasmids were introduced into electrocompetent E. coli ER2566 cells immediately before running the assay. Following electroporation, cultures were transferred directly into Luria–Bertani (LB) medium supplemented with kanamycin (50 μg/mL) and were then allowed to grow for approximately 6 h. Cell suspensions were then adjusted to an OD₆₀₀ of 0.5 and serially diluted 1:10 to 1:100,000. Each dilution was spread in triplicate onto LB agar plates containing kanamycin (50 μg/mL) and 0.2 mM IPTG to induce TisB production. Plates were incubated overnight at 37 °C, and colony-forming units were quantified the following day. Survival was evaluated relative to cells harboring the empty pET28b­(+) vector. The entire assay was conducted independently three times.

Glossary

ABBREVIATIONS

AAs

amino acids

A. b.

A. baumannii

AMP

antimicrobial peptide

AMR

antimicrobial resistance

AZT

azithromycin

E. c.

E. coli

E. f.

E. faecalis

ESKAPE

E. faecium, S. aureus, K. pneumoniae, A. baumannii, P. aeruginosa, and Enterobacter spp.

FICI

fractional inhibitory concentration index

K. p.

K. pneumoniae

MBC

minimum bactericidal concentration

MeCN

acetonitrile

MIC

minimum inhibitory concentration

MH

Mueller-Hinton

P. a.

P. aeruginosa

Pbf

2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl

RIF

rifampicin

sPEG2

8-amino-3,6-dioxaoctanoic acid

S. a.

S. aureus

SPPS

solid-phase peptide synthesis

TA

toxin-antitoxin.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c02137.

• Data from truncation and synergy studies; HPLC data (chromatograms, retention and % MeCN at peak of elution) and MALDI-TOF mass spectra (PDF)

Conceptualization: E.D., J.A.B., E.H., and H.F. Investigation: E.D., I.P.S., and I.M.M.R. Methodology: E.D. and I.P.S. Writing – original draft: E.D. Writing – review and editing: J.A.B., E.H., M.B., and H.F. Funding acquisition: J.A.B., E.H., H.F., and M.B. Supervision: J.A.B., E.H., and H.F.

The authors declare no competing financial interest.