An intracellular release peptide display technology unveils an antimicrobial peptide with low probability for resistance development

Summary

Discovery of bioactive peptides, including those acting to permeabilize and/or kill bacterial cells (antimicrobial peptides) has drawn extensive interest in recent years. However, current technologies for their identification are limited. To address these limitations, the Intracellular Release Peptide Display (IRPD) technology allowing the recombinant “display” of intracellular linear peptides was developed. IRPD uses the protease domain of the capsid protein from the Semliki Forest virus as a scaffold to express and liberate linear peptides intracellularly in Escherichia coli. IRPD is a universal platform that allows screening of millions of peptides and the discovery of bioactive peptides from direct target interactions and independent of the cell envelope barrier. Here, we identified peptides that cause increased bacterial cell envelope permeability and lysis. The most promising candidate, P38, effectively kills Gram-negative pathogens by disrupting the inner membrane without detectable resistance development. Thus, P38 constitutes an interesting hit peptide for further development.

Graphical abstract

Highlights

• •Intracellular Release Peptide Display to express linear peptides intracellularly
• •IRPD enables high-throughput identification of potential AMPs
• •P38 disrupts the bacterial envelope, killing E. coli without resistance development

Peptides; Microbiology; Applied microbiology

Introduction

The emergence of multidrug-resistant (MDR) bacteria and the decline in discovery of novel classes of antibiotics have made antimicrobial resistance a severe global health challenge and economic burden.¹ The number of bacterial infections caused by MDR bacteria is increasing and in 2019, MDR bacteria caused 1.27 million deaths.² MDR Gram-negative bacteria are particularly challenging due to their limited treatment options³ as their outer membranes provide an effective barrier to many antibiotics.^4,5,6,7

Antimicrobial peptides (AMPs) are common in nature,⁸ and have recaptured attention as potential alternatives to traditional antibiotics and by their ability to permeabilize bacteria at sub-MIC concentrations.^9,10 Naturally occurring AMPs occupy a small fraction of the chemical space available for the de novo design and synthesis of peptide antimicrobials. Currently, the Database of Antimicrobial Activity and Structure of Peptides (DBAASP) contains 3175 ribosomally synthesized and about 717 non-ribosomally synthesized AMPs.¹¹ However, efficient technologies enabling screening of large peptide libraries and de novo discovery of AMPs are limited. Phage display¹² and mRNA display¹³ are in vitro panning-based methodologies that allow the interrogation of vast ensembles of recombinant peptides aimed at selecting peptides that bind a given protein target without directly assessing the functionality. Surface Localized Antimicrobial Display (SLAY) enables the generation and function-based screening for AMPs in particular targeting the bacterial cell envelope.¹⁴ Split-intein circular ligation of peptides and proteins (SICLOPPS) generates cyclic peptides intracellularly,¹⁵ and allows screening of peptide libraries and the identification of cyclic peptides with intracellular targets.

Whereas cyclic peptides are conformationally restricted, linear peptides are more flexible, and therefore cover a larger structural space for a library of a given size. To explore de novo expressed linear peptides as potential AMPs, while benefiting from the recombinant “in cell” advantages of SLAY and SICLOPPS, we used the Semliki Forest virus (SFV) capsid protein (C) protease domain (CP) as a scaffold for the generation of linear peptides intracellularly. C is produced as a polyprotein displaying autocatalytic self-cleavage activity (Figure 1A).¹⁶ CP responsible for cleavage is encoded by amino acids 113–267 of C (Figure 1A). During protein folding, the catalytic triad (H145, D167 and S219) assemble with the C-terminal sequence resulting in autocatalytic cleavage and liberation of a free peptide corresponding to the sequence C-terminal to amino acid W267.^17,18 CP acts exclusively in cis,¹⁹ because the new C-terminus generated following cleavage remains in the active site blocking further proteolysis.^20,21 In live cells, cleavage occurs co-translationally and precedes the complete translation of the ORF encoding the SFV polyprotein.¹⁶

Figure 1.

Intracellular Released Peptide Display (IRPD)

(A) Diagram of the Semliki Forest virus (SFV) polyprotein and capsid protein (C). Following synthesis, C cleaves itself (autocatalytic cleavage). E1, 6K, and p62 are viral envelope proteins. Vertical lines indicate proteolytic cleavage sites. Amino acids 1–112 in C constitute the SFV genomic RNA-binding domain, which is dispensable for auto-protease activity. The protease domain (CP), residues 113–267, is highlighted in blue, while the catalytic triad, residues H145, D167, and S219, are indicated by an asterisk (∗).

(B) Schematic illustration of the IRPD technology. CP amino acids 113–267 encoding the serine protease domain (CP, blue) are cloned on a plasmid under the control of the inducible promoter pBAD. Specific or random peptide sequences (dark blue) are genetically linked either directly to the C-terminus of CP or via the SAP (serine-alanine-proline) linker. Upon induction with L-arabinose, a CP-peptide fusion protein is expressed, followed by folding of CP and autocatalytic cleavage. The vertical red arrows indicate the sites of autocatalytic cleavage (after W267). Linear peptides are liberated from the CP scaffold intracellularly.

(C) Expression of CP-protein and peptide fusion proteins, followed by autocatalytic cleavage and protein/peptide release. SDS-PAGE showing CP, eGFP-CP, CP-eGFP, CP-SAP-eGFP, 3XFlag-CP, CP-3XFlag, and CP-SAP-3XFlag expression and cleavage. CP (16.96 kDa) is indicated with black arrows (→). Samples of MG1655 harboring pCP (only CP without any peptide fused to CP), pGFP-CP, pCP-eGFP, pCP-SAP-eGFP, p3XFlag-CP, pCP-3XFlag and pCP-SAP-3XFlag grown in ABTGly supplemented with casamino acids at 30°C were collected before and after 5 h of induction with 0.2% L-arabinose.

(D) GFP is cleaved off the CP scaffold. MG1655 cultures harboring pCP, pGFP-CP, pCP-eGFP, and pCP-SAP-eGFP grown in ABTGly supplemented with casamino acids at 30°C before and after 5 h of induction with 0.2% L-arabinose. Immunoblots were probed with anti-GFP antibody.

(E) Induction of the synthesis of CP-Oncocin (CP-Onc) or CP-Apidaecin-1B (CP-Api) leads to growth inhibition. Drop dilution assay for evaluating the growth inhibitory effects of CP-Onc and CP-Api upon expression induction with 0.2% L-arabinose. CP, CP-C1, and CP-C2 were used as negative controls. C1 and C2 are random peptides selected from the CP-peptide library. Aliquots of 5 μL from serial 10-fold dilutions prepared in 0.9% NaCl, were spotted onto ABTGly supplemented with casamino acids agar plates with and without 0.2% L-arabinose. The plates were incubated overnight at 30°C.

(F) Expression of CP-Api and CP-Onc fusion proteins, followed by autocatalytic cleavage and peptide release. SDS-PAGE showing CP, CP-Onc, CP-Api, CP-C1, and CP-C2 expression and cleavage. Cleaved CP is indicated with an asterisk (∗).

Recombinant CP-peptide fusion libraries expressed in E. coli were exploited as a high-throughput platform enabling the identification of bioactive peptides. We used this technology (termed “Intracellular Release Peptide Display”, IRPD), to identify recombinant peptides that lead to permeabilization and/or lysis of Gram-negative bacteria upon their intracellular expression. Such putative peptides may act directly on the cellular envelope to destabilize it, but could also act intracellularly to affect biosynthetic pathways involved in envelope integrity. Despite not screening for AMP activity per se, the most promising candidate identified, P38, disrupts the bacterial envelope, including the inner membrane thereby killing E. coli without detectable resistance development. These findings demonstrate the potential of IRPD as an alternative approach for the discovery of peptides for development into novel antimicrobial drug leads and possibly beyond.

Results

Development of intracellular release peptide display technology

We cloned the protease domain of the SFV capsid protein (CP) on plasmid pACYC184 under the control of the arabinose inducible pBAD promoter (Figure 1B). We designed the system in such a way that it is possible to genetically link any peptide sequence at the C-terminus of CP (Figure S1). To ascertain the robustness of autocatalytic cleavage activity of CP,²² we tested N- and C-terminal fusions of eGFP and 3XFlag to CP. As control, CP was cloned without a linked peptide or protein. In native SFV, CP is linked at the C-terminus to the peptide p62 via a serine-alanine-proline linker (SAP). Because the SAP sequence was ubiquitous to all previously tested sequences,²² we generated constructs with and without SAP linker (Figure 1B). Following recombinant expression, we investigated protease cleavage by SDS-PAGE (Figure 1C) and immunoblot analysis (Figures 1D and S2A). Induction of CP expression revealed a protein band corresponding to the size of CP (17 kDa) (Figure 1C). Induction of CP-eGFP, CP-SAP-eGFP, CP-3XFlag and CP-SAP-3XFlag resulted in the appearance of a protein band of similar size, which is consistent with CP-protein expression and successful autocatalytic cleavage (Figure 1C). Immunoblot using anti-GFP antibody, revealed a specific band corresponding to the size of eGFP (27 kDa) upon induction of C-terminal CP-eGFP and CP-SAP-eGFP expression, indicative of autocatalytic cleavage and intracellular liberation of eGFP and SAP-eGFP (Figure 1D). Immunoblot analysis using anti-FLAG antibody, failed to detect the 3XFlag peptide (2730 Da) following CP-3XFlag or CP-SAP-3XFlag expression (Figure S2A). Nevertheless, the appearance of CP-sized protein bands detected by SDS-PAGE (Figure 1C) implies synthesis and release of 3XFlag peptide. Immunoblots of N-terminal fusion constructs eGFP-CP (44 kDa) and 3XFlag-CP (20 kDa) expression revealed bands corresponding to the size of the expected fusion proteins (Figures 1D and S2A). Hence, cleavage requires C-terminal localization of the fusion moiety. Finally, CP-eGFP expression was significantly higher in minimal medium compared to LB medium (Figure S2B).

Our results demonstrate that different C-terminal CP-peptide/protein fusions are efficiently expressed and cleaved, and that cleavage is independent of the native SAP linker. These observations suggest limited sequence dependence of the cleavage, properties which are crucial to the use of CP in peptide display. CP detection by SDS-PAGE following cleavage is a robust proxy for peptide expression when the peptide itself cannot be easily monitored, as exemplified by the inability to detect 3XFlag despite CP-3XFlag cleavage. Collectively, these results encouraged the use of CP as a scaffold for the intracellular production and release of peptides in E. coli.

IRPD enables efficient expression of functional AMPs in vivo

To prove that CP is an effective vehicle for the generation of intracellular, functional AMPs, oncocin (Onc), and apidaecin-1B (Api) were chosen as positive controls. Both AMPs bind to the bacterial ribosome inhibiting protein synthesis.^23,24,25 The mechanism of action of Onc involves binding inside the ribosome peptide exit-channel, leading to an arrest of the ribosome at the start codon.^26,27 Api targets translation termination by depleting the cellular pool of free release factors causing the majority of ribosomes to stall at stop codons, and thereby resulting in a global translation termination shut down.^28,29 Onc and Api were cloned C-terminally to CP and expressed in E. coli. Induction of CP-Onc or CP-Api expression, strongly inhibited bacterial growth (Figure 1E). In contrast, neither CP alone (CP) nor CP-eGFP, CP-3XFlag or two random members of the CP-peptide library (CP-C1 and CP-C2) affected bacterial growth (Figure 1E). SDS-PAGE revealed cleavage of CP-Onc, CP-Api, and the controls CP-C1 and CP-C2 (Figure 1F). Notably, CP-Onc expression gave rise to reduced CP amounts as detected via SDS-PAGE, possibly because Onc targets translation initiation and therefore could rapidly inhibit its own synthesis. Collectively, these results corroborate that CP is a suitable vehicle for the display of biologically active AMPs, supporting that IRPD may be exploited more generally for the identification of other, as yet unknown AMPs targeting specific intracellular processes, proteins or other (macro)molecules essential for bacterial survival, thereby providing novel hit structures as well as novel, sensitive molecular targets for traditional small molecule antibiotic discovery.

Identification of peptides causing bacterial cell envelope disruption

In order to obtain proof of principle for identifying AMPs from a random population of peptide sequences, we generated a CP-peptide library encoding random 12-mer peptides (Figure S1). The size of the library was estimated by NGS amplicon sequencing and bioinformatics to consist of a minimum of 2.698.649 different contigs corresponding to peptide sequences with a coverage of 40.752.312 reads (Figure S3 and STAR Methods). Further, 10 random members of the CP-peptide library were selected, sequenced, and investigated for CP-peptide expression and autocatalytic cleavage (Figure S4). All random CP-peptide fusion were expressed and cleaved upon induction with 0.2% L-arabinose regardless of the amino acid sequence.

The bacterial cell envelope is an established and effective target for many antimicrobials. Thus, we modified a previously described genetic screen used to identify factors required for Gram-negative cell envelope biogenesis based on chlorophenol red-β-D-galactopyranoside (CPRG), a substrate for the β-galactosidase (LacZ).³⁰ In cells with increased envelope permeability or frequency of lysis, LacZ cleaves CPRG to CPR, which is detectable by a color change from yellow to dark pink (Figure 2A).

Figure 2.

Screening and identification of CP-peptides conferring increased cell envelope permeability or lysis as determined by the CPRG assay

(A) Coupling of IRPD with the CPRG screen to identify AMPs causing increased cell envelope permeability or higher frequency of lysis. A CP-peptide library is generated, transformed into MG1655ΔaraBAD and plated on CPRG plates supplemented with 0.2% L-arabinose. Cells expressing peptides causing cell lysis or cell envelope permeability are able to cleave CPRG into CPR (dark pink coloring). In contrast, cells with an intact cell envelope will remain yellow because CPRG will not enter these cells. The CP-peptide library plasmids of pink colonies were isolated and sequenced.

(B) Amino acid sequences of hit and control peptides identified in the CPRG screen.

(C) Induction of CP-P5, CP-15, CP-38, and CP-P62 expression leads to increased cell envelope permeability or lysis. Drop dilution assay evaluating cell membrane permeability or lysis upon induction of synthesis of CP-P5, CP-P15, CP-P38, and CP-C3 on CPRG plates containing 0.2% L-arabinose. pCP-C3 is a random CP-peptide library plasmid, which was CPRG negative in the initial screening, and was used as a negative control throughout the CPRG assay. Aliquots of 5 μL, containing serial 10-fold serial dilutions prepared in LB medium and were spotted onto CPRG plates containing 0.2% L-arabinose, incubated overnight at 30°C, and then the plates were photographed.

(D) Induction of the synthesis of CP-P15, CP-P38, and CP-P62 leads to growth inhibition in minimal medium. Drop dilution assay for evaluating the growth inhibitory effect of CP-P5, CP-P15, CP-P38, and CP-P62 upon induction with 0.2% L-arabinose on ABTGly supplemented with casamino acids. CP is used as negative control. 10-fold serial dilutions were prepared in 0.9% NaCl and aliquots of 5 μL were spotted onto agar plates with or without the addition of 0.2% L-arabinose. The plates were incubated overnight at 30°C, and then photographed.

(E) Expression of CP-P15 leads to cell elongation, whereas expression of CP-P38 leads to compromised cell membranes and cell lysis. Cultures of MG1655ΔaraBAD expressing either CP, CP-15, or CP-38 were grown under non-induced or induced (0.2% L-arabinose) conditions for 24h and analyzed by phase contrast and fluorescence microscopy. For fluorescence microscopy cells were stained with LIVE/DEADBacLight. Syto9 stains cells with intact or compromised membranes, whereas propidium iodide only stains cell with compromised membrane (dead cells).

The CP-peptide library was transformed into MG1655ΔaraBAD and plated colonies were screened for cell envelope permeabilization. Dark pink colonies appeared at a frequency of ∼0.01–0.02%. Sequence analysis of the CP-peptide library plasmids isolated from six dark pink colonies revealed four CP-peptide library plasmids with different sequences (Figure 2B). pCP and pCP-C3, a random, CPRG negative member of the library was used a negative control (Figure 2B). The sequences encoding P5, P15 and P38 were re-cloned and retransformed into MG1655ΔaraBAD, and the positive CPRG phenotype was confirmed ruling out any contribution to the phenotype from other plasmid or genomic mutations (Figure 2C). Therefore, expression of CP-P5, CP-P15, CP-P38 and most likely P62 results in either increased cell envelope permeability and/or frequency of lysis.

Expression of CP-P15, CP-P38 and CP-P62 inhibits bacterial growth

Induction of CP-P15, CP-P38 and CP-P62 expression resulted in bacterial growth inhibition in minimal medium, whereas CP-P5 and CP did not (Figures 2D, S5A, and S6A). CP-P15 induction produced an elongated morphology, whereas CP-P38 generated a “lysis” phenotype as evidenced by the presence of cell debris and ghost cells after induction for 5 h (Figure S5B). The cell lysis phenotype of CP-P38 and the cell elongation phenotypes of CP-P15 and CP-P62 were even more pronounced after 24 h of induction (Figures 2E and S6B). CP induction had no effect on cell morphology neither after 5 h nor upon 24 h of protein expression. Live-dead staining performed after 24 h expression revealed increased propidium iodide staining selectively for cells expressing CP-P38 (Figure 2E), indicative of a compromised cell envelope and cell death in the majority of cells. In contrast, expression of CP-15 resulted in only minor effect on propidium iodide staining, indicative of a functional envelope (Figure 2E).

SDS-PAGE analysis revealed complete cleavage of CP-P15, while CP-P38 was only partially cleaved or not cleaved at all, as evidenced by the CP band size (Figure S7A). To differentiate between non- or partial cleavage, serine 219, essential for autocatalytic cleavage activity of C¹⁸, was replaced by isoleucine. SDS-PAGE analysis showed that the cleavage pattern of CP-P38 and the cleavage deficient CP(S219I)-P38 were indistinguishable (Figure S7A). However, only cells expressing CP-P38, and not CP(S219I)-P38, had a CPRG-positive phenotype with growth inhibition and cell lysis upon induction (Figures S7B–S7D), suggesting that CP-P38 cleavage is required for activity. Thus, it is likely that P38 is very potent given that its release by autocatalytic cleavage was inefficient.

Synthetic P38 peptide shows bactericidal activity against MDR Gram-negative and Gram-positive bacteria

To address whether the AMPs identified were active as externally applied compounds, P15, P38 and P62 were chemically synthesized and tested in trans for antimicrobial activity against Gram-negative and Gram-positive bacteria (Table 1).

Table 1.

Minimum inhibitory concentration [μg/ml] for P15, P38, colistin, oncocin, and rifampicin against Gram-negative and Gram-positive bacteria

	P15	P38	P62	Colistin	Oncocin	Rifampicin
E. colia
MG1655	≥128	8	≥64	0.25–0.5	16	16
ATCC25922	≥128	8	≥64	1–2	16	4
AS19	≥128	4–8	≥64	4	4	0.5–1
ST131	≥128	8	–	0.25	32	8
ESBL20150072	≥128	8	–	32	16	8
K. pneumoniaea
ATCC700603	≥128	32	–	0.5	≥64	32
ST512	≥128	32	–	≥64	≥64	32
P.aeruginosaa
ATCC27853	≥128	32	–	0.5	≥64	4
PAO1	≥128	16	–	2	≥64	4
SNRC	≥128	8	–	≥64	≥64	–
B. subtilisb
168	–	1–2	–	–	–	0.03–0.06

^aCells were grown in AB minimal medium supplemented with thiamin and glucose.

^bCells were grown in Spizizen minimal medium supplemented with 0.5% glucose and 0.05 mg/mL tryptophan.

Because IRPD does not screen for delivery across the bacterial envelope, the lipopolysaccharide (LPS) compromised E. coli strain AS19 with enhanced permeability was included, as this is hypersensitive to a wide range of compounds including multiple antibiotics.^31,32 P15 and P62 did not show antimicrobial activity against any of the strains tested against (Table 1) probably due to poor penetration of the membrane barrier or otherwise insufficient target saturation. In contrast, P38 showed robust antimicrobial activity in minimal medium against E. coli including the MDR clinical isolates ST131, which is sensitive to colistin, and ESBL20150072, which is colistin resistant (MICs 4–8 μg/mL; Table 1). The activity of P38 against E. coli is comparable to that of oncocin (Table 1). P38 showed moderate antimicrobial activity (MIC 32 μg/mL: Table 1) against K. pneumoniae ATCC700603 and K. pneumoniae ST512, both MDR isolates that harbor a capsule promoting increased tolerance to AMPs.³³ Finally, P38 was active against Pseudomonas aeruginosa (MICs 8–32 μg/mL; Table 1). In addition, P38 showed activity against the Gram-positive reference strain B. subtilis 168 (MIC 1–2 μg/mL; Table 1). Peptide P38 shows homology to Temporin-AJ8 isolated from the torrent frog Amolops jingdongensis,³⁴ both with respect to sequence and predicted structure (Figure S8). Interestingly, Temporin-AJ8 has limited activity against Gram-negative bacteria but has a relatively low MIC against Staphylococcus aureus and Bacillus subtilis. The mode of action of Temporin-AJ8 has not been established.

Interestingly, when cultured in Mueller-Hinton II (MHII) medium, P38 had no antimicrobial activity except against E. coli AS19 (Table S4). The reduced activity of P38 in MHII versus ABTG suggests the presence of molecules that interfere with the activity of P38. Indeed, the antimicrobial activity of P38 in ABTG medium is inhibited by increasing concentrations of casamino acids, heat-inactivated human serum, or serum albumin (Table S5), most likely due to non-specific binding.

Because P38 remains active against AS19 in MHII medium, we suspect the outer membrane to comprise a barrier for P38 entry into E. coli. Indeed, mutants deficient in LPS were more susceptible to P38 relative to the isogenic wild-type strain when grown in MHII medium (Figure 3). In particular, loss of WaaC or WaaE showed at least 16-fold increased susceptibility. The LPS of either waaC or waaE mutants, lacking the inner and outer core sugar moieties, consists of Lipid A-KDO₂ only (the minimal LPS structure required for cell viability) revealing that LPS acts as a protective barrier (Figure 3).

Figure 3.

The outer membrane is a permeability barrier of P38 in rich medium

MIC [μg/ml] values of P38 against LPS deficient E. coli strains grown either in MHII medium or ABTG medium. ∗ Represents MIC ≥128 μg/ml. The structure of the LPS is indicated for the individual mutants.

Finally, we evaluated the hemolytic activity of P38 by measuring the lysis of human erythrocytes. At 128 μg/mL, the highest test concentration obtainable due to limited solubility, the hemolytic of activity of P38 was 20 ± 11% (Figure S9). In conclusion, P38 is active against Gram positive and Gram-negative bacteria, particularly E. coli in minimal medium, when administrated externally as a chemically synthesized compound.

P38 rapidly kills exponentially growing cells

To examine the rate of P38 induced cell killing, a time-kill experiment was performed. When P38 (1×MIC) was added to exponentially growing cells, a rapid reduction in the number of viable cells was detected, resulting in efficient killing after 4 h of incubation (Figure 4A). The reduction in CFU is comparable to the membrane-interfering lipopeptide colistin, and it is highly indicative of a mode of action that ultimately leads to the loss of envelope integrity.

Figure 4.

Synthetic P38 kills exponentially growing MG1655 cells and shows low level of resistance development

(A) Time kill assay. Treatment of MG1655 growing exponentially in ABTG with P38, colistin or oncocin at 1×MIC at cell densities comparable to MIC (5 × 10⁵-1×10⁶). Data are means of three biological replicates ±SEM.

(B) Resistance study of MG1655. Cells were evolved in ABTG medium at increasing concentrations of either P38, colistin, or ciprofloxacin for a maximum of 40 days. Non-treated cells were included as a control. MIC values of evolved cultures were determined and the fold-change of MIC calculated. The figure represents the fold-change of MIC for one out of five independently evolved cultures.

(C) Impact of gene overexpression on P38 resistance. MIC values were recorded in ABTG medium for MG1655 carrying either the empty ASKA plasmid pCA24N (control plasmid) or the ASKA plasmid library against P38, ampicillin, and trimethoprim. With increasing IPTG concentration, the overexpression of the ASKA plasmid library leads to increased resistance against ampicillin and trimethoprim, but not against P38.

P38 exhibits low level of resistance development

Interestingly, we were not able to obtain MG1655 mutants resistant to P38 in an adaptive laboratory evolution experiment (Figure 4B). In contrast, when challenged with colistin or ciprofloxacin, highly resistant mutants arose, displaying more than 1000-fold and 4000-fold increased MIC values, respectively (Figures 4B and S10). Next, we applied the ASKA collection,³⁵ which contains each ORF of E. coli cloned into an expression vector, and tested whether gene amplification, in nature often transient,³⁶ confers resistance to P38. Whereas E. coli showed increased resistance to the antibiotics ampicillin and trimethoprim upon induction of the pooled AKSA collection (Figure 4C),³⁷ increased resistance was not detected against P38. Thus, increased expression of any specific native protein does not provide resistance to P38.

P38 disrupts the inner membrane and induces cell lysis

To gain insight into the mode of action of P38, we examined its effect on the four major biosynthetic pathways of E. coli. P38 inhibited the incorporation of radiolabeled precursors into DNA, RNA, protein, and peptidoglycan within minutes of addition (Figures 5A and S11). The bactericidal activity of P38, could be responsible for inhibition of the four tested macromolecular biosynthesis pathways by promoting envelope disruption. As the intracellularly released P38 peptide caused permeabilization of the inner membrane, we tested for P38 induced depolarization and permeabilization of the cell envelope. DiBAC₄(3)³⁸ only enters depolarized cells where it binds the inner membrane and its protein components, upon which it exhibits enhanced fluorescence. The addition of P38 to DiBAC₄(3) treated E. coli cells resulted in significant concentration-dependent increase in fluorescence within 5 min (Figure 5B), suggesting that rapid cellular depolarization occurred. The fluorescent dye Sytox Green only penetrates cells with compromised inner membrane and binds to nucleic acids resulting in increased fluorescence. Upon addition of P38, an increase in Sytox Green fluorescence was similarly detected within 5 min, indicating inner membrane disruption (Figure 5C). This effect is comparable to that of 0.1% detergent Triton X-100 (Figure 5C), but not as strong as that of the model AMP Melittin with pore-forming activity (Figure 5C).³⁹

Figure 5.

P38 compromises the integrity of the inner membrane in E. coli

(A) Effect of P38 on macromolecular biosynthesis in MG1655. Incorporation of ³H-thymidine (DNA), ³H-uridine (RNA), ³H-arginine (protein), and ³H-glucosamine (cell wall) was determined in exponentially growing cells treated for 20 min with P38 (40 μg/mL, 5×MIC). Nalidixic acid (40 μg/mL, 10×MIC), rifampicin (160 μg/mL, 10×MIC), chloramphenicol (80 μg/mL, 10×MIC) and ampicillin (100 μg/mL, 25×MIC) were used as control antibiotics (dark gray bars). Data are means of 3 independent experiments ±SEM.

(B) P38 induces membrane depolarization. MG1655 cells growing exponentially in ABTGly supplemented with casamino acids were treated with DiBAC₄ for 10 min, followed by addition of P38 or melittin (1/4×MIC, 4 μg/mL; 1/2×MIC, 8 μg/mL and 1×MIC, 16 μg/mL). Polymyxin B was added at a concentration of 1×MIC (0.25 μg/mL). Fluorescence was recorded every 10 min. Data are means of three biological replicates ±SEM.

(C) Membrane disruption assay using Sytox Green. MG1655 cells growing exponentially in ABTG supplemented with casamino acids were treated with Sytox Green for 10 min prior to addition of P38 (1/4×MIC, 2 μg/ml), 1/2×MIC, 4 μg/mL; 1×MIC, 8 μg/ml), melittin (1/4×MIC, 4 μg/mL), Polymyxin B (1×MIC, 0.25 μg/mL) or Triton X-100 (0.1%). Fluorescence was recorded every 10 min. Data are means of three biological replicates ±SEM.

(D) Inner membrane permeabilization upon extracellular application of P38. E. coli ML-35p cells growing exponentially in ABTG supplemented with leucine were incubated with either P38, colistin or 5% toluene and of o-nitrophenyl-β-D-galactopyranoside (ONPG). Absorbance at 420 nm was measured every 15 min. Data are means of three biological replicates ±SEM.

(E) Intracellular expression of CP-P38 leads to permeabilization of the inner membrane. ONPG was added to cultures of MG1655ΔaraBAD expressing either CP or CP-38, which were grown under non-induced or induced (0.2% L-arabinose) conditions for 5 h in ABTG, and absorbance at 420 nm (ONPG) and 600 nm was recorded every 15 min. The ratio OD₄₂₀/OD₆₀₀ was calculated (ONPG absorbance normalized to the cell number) was calculated and plotted. Data are means of three biological replicates ±SEM.

Similarly, the inability of ONPG, a β-galactosidase substrate, to cross the inner membrane of E. coli ML-35p was exploited using a lactose permease-deficient strain constitutively expressing cytoplasmic β-galactosidase. Sub-inhibitory concentrations of P38 (0.5×MIC) caused rapid inner membrane permeabilization (Figure 5D). Recombinant expression of CP-P38 similarly increased the inner membrane permeability as measured by ONPG uptake and intracellular cleavage (Figure 5E). Thus, P38 destabilizes the inner membrane regardless of whether it is added extracellularly as a chemically synthesized peptide or derived from intracellular expression of recombinant CP-P38.

Anionic phospholipids added in trans inactivate the antimicrobial activity of P38

Based on the permeabilization activity of P38 along with the inability of generating resistant mutants, we wondered if P38 binds to components of the bacterial inner membrane. We therefore tested the antimicrobial activity of P38 in the presence of components of the bacterial inner membrane including its three major phospholipid components phosphatidylethanolamine (PE), phosphatidylglycerol (PG) and cardiolipin (CL), as well as phosphatidylserine (PS) and phosphatidylcholine (PC) (Table S6). Most notably, PS reduced the antimicrobial activity even at the lowest concentration tested. Similarly, PG and CL display inhibitory activity, whereas the major bacterial membrane component PE did not affect P38 activity.

Discussion

The present study demonstrates that the protease domain of the capsid protein of SFV can be exploited as a scaffold for intracellular display of linear peptides, establishing IRPD as an alternative platform for the discovery and selection of AMPs. This approach facilitates efficient intracellular production of peptides that might be vulnerable to proteolytic degradation and are difficult to express and trace. The discovery of the peptide P38 via IRPD, which retained antimicrobial activity as a chemically synthesized peptide, demonstrates proof of principle and illustrates the advantages of using linear peptide expression and release within cells, and directly avoids complexities associated with peptides that remain part of a larger fusion, and therefore may not reflect the activity of the peptide itself.

The power of IRPD lies in the ability to discover peptides that inhibit previously uncharacterized or unknown targets by effectively separating the mechanism of cell entry from potential antimicrobial function. Therefore, IRPD identifies peptides acting strictly intracellularly along with those affecting the inner membrane, while it remains unclear whether peptides acting in the periplasm and/or outer membranes may be identified. In contrast, traditional chemically synthesized peptide libraries often miss out on intracellular targets because many peptides cannot penetrate the bacterial cell wall. IRPD, on the other hand, identifies candidate peptides through direct target interaction, irrespective of their ability to enter cells. Recently, Surface Localized Antimicrobial Display,¹⁴ a recombinant bacterial cell surface display technology, has received a great deal of attention. As the name indicates SLAY displays peptide sequences directly on the cell surface. It is therefore conceivable that AMPs identified by SLAY favor those with targets at or close to the bacterial cell surface, whereas it is unclear to which extent peptides affecting truly intracellular processes can be identified. IRPD screening of a very small fraction of the CP-peptide library for cell permeabilization/lysis identified CP-P15, CP-P62 and CP-P38 and demonstrated that the induction of CP-P15, CP-P62 and CP-P38 expression results in growth inhibition. The different morphological phenotypes of CP-P15 and CP-P38 indicated different molecular targets and different modes of action. Based on the frequency of CPRG positive colonies and the library size of 2.4 mio, we theoretically expect 240–480 CPRG positive colonies (and consequently over 3 × 10¹⁷ candidate peptides from the total theoretical library of 12²⁰ = 3 × 10²¹ peptide sequences). This positions IRPD as a distinct tool for tapping into a reservoir of antimicrobial compounds that awaits to be discovered.

However, once hit peptides (sequences) are identified by IRPD, their inability to cross the bacterial envelope might pose a challenge for their application as chemically synthesized AMPs. Thus, peptide P15 did not show measurable antimicrobial activity, when applied as compound, which may indicate that its observed lack of antimicrobial activity is due to limited cellular uptake. This might be addressed by traditional chemical synthesis-based SAR analysis of the hit peptides, or by conjugation to a bacterial penetrating peptide (BPP) or other bacteria delivery moieties. The adaptability and molecular flexibility of IRPD allows the design, generation, and screening of large peptide libraries as well as smaller sub-libraries for optimizing hit peptides.

P38 is an interesting candidate in terms of bacterial delivery and mode of action and it constitutes a concrete example for the successful use of IRPD in tandem with bioactivity assays. The CPRG based screen aims to identify both peptides acting directly on the bacterial envelope to destabilize it, and peptides acting strictly intracellularly on biosynthetic pathways involved in envelope integrity. Our results suggest that P38 belongs to the first group and causes E. coli cells to lyse by compromising the inner membrane’s integrity. Lysis was observed both when P38 was produced intracellularly via IRPD and when it was applied externally as a synthetic peptide. Therefore, the target of P38 is most likely accessible from either side of the envelope. Alternatively, P38 must traverse the envelope to reach its target. This, combined with the observation that the addition of anionic phospholipids to the medium inactivates P38, suggests the inner membrane as the primary target. The fact that we were not able to obtain P38 resistant mutants also suggests that the target of P38 is not a single protein.

It is important to point out that AMPs discovered by IRPD may also identify novel druggable cellular targets for which small molecule inhibitors, including those derived from peptidomimetics of the (optimized) hit peptide, can be designed and developed.

Beyond the discovery of antimicrobials and molecular targets, IRPD has the potential to be used as a platform for the discovery of other types of therapeutic peptides along with their molecular targets, addressing an unmet challenge in early drug discovery. IRPD as the potential to be transferred to other expression systems to be applied more broadly in relation to human diseases. Peptides are increasingly recognized as a valuable class of therapeutics, given their high target selectivity, structural diversity, and large area of target binding (compared to small molecule drugs). When tailored for human cellular use, IRPD could become a cornerstone technology for the discovery, characterization, and advancement of innovative therapeutic peptides.

Limitations of the study

Cell permeabilizing peptides identified by the IRPD technology do not identify AMPs per se. Yet, the most promising peptide P38 efficiently kill bacterial cells. In fact, a common feature of many AMPs is their ability to permeabilize cells at sub-MIC and act bactericidal at MIC and above.^9,10 In order to directly identify those peptides that lead to cell death upon expression, i.e., true AMP, a direct screen for bacterial inviability is required (to be published elsewhere).

There are, however, also a number of shortcomings/caveats to the IRPD technology in its present form. In terms of library randomness, we did not observe a sequence bias arising from a bias in the primer sequences such as reported for other display systems.⁴⁰ Cleavage of peptides from their CP-peptide precursors could vary depending on peptide amino acid sequence. This has currently not been addressed. Because the peptide library is expressed in vivo, an obvious shortage relative to chemically synthesized libraries is the limited ability to incorporate unnatural amino acids in the sequence. Leakage in the expression system is another obvious shortage of the IRPD system, especially when it comes to identifying AMPs since such a leakage results in a loss of the most promising candidates during propagation of the library. In order to minimize this problem we employed the pBAD system for expression, which has minimal leakiness.⁴¹ Finally, as indicated previously, a main obstacle of IRPD is that peptides identified may not enter cells when added from the outside. In the future, addition of a sequence specifying a BPP may alleviate some of these problems.⁴² The inability of certain peptides to enter the bacterial cell is clear from the present work where three peptides (P15, P38 and P62) prevented bacterial growth when produced intracellularly, but only P38 had activity when added as a compound. The SFV primarily infect mosquitos but has a broad host range and can also infect a number of vertebrate hosts.⁴³ Therefore, the CP protein can undergo autocleavage in all these hosts and, as shown here, also in the bacterium E. coli. We therefore believe the IRPD technology can be exported to other bacteria as well, both Gram-negative and Gram-positive, although this has not been tested experimentally at this time.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Anders Løbner-Olesen (lobner@bio.ku.dk).

Materials availability

All plasmids and bacterial strains generated in this study are available from the lead contact with a complemented Materials Transfer Agreement and with a reasonable compensation by requestor for its processing and shipping. There are restrictions to the availability of strains ST131, ESBL20150072, ST512, and SNRC because we want to limit the spread of these multi-drug resistant strains.

Data and code availability

• •Data: all data reported in this article will be shared by the lead contact upon request.
• •Code: this article does not report original code.
• •Any additional information required to reanalyse the data reported in this work will be shared by the lead contact upon request.

Acknowledgments

This work was supported by the Novo Nordisk Foundation Challenge Program [NNF16OC0021700]; and the Lundbeck Foundation [R346-2020-1879 and R436-2023-1224]. All figures were created with BioRender.com.

Author contributions

Conceptualization: A.E., T.B., and A.L.-O.; data curation: A.E. and C.O.; formal analysis: A.E., G.C., and A.L.-O.; investigation: A.E., G.C., and C.O.; methodology: A.E., C.O., and A.L.-O.; project administration: A.E. and A.L.-O.; resources: A.E., H.F., P.E.N., and A.L.-O.; manuscript writing: A.E.; manuscript editing: A.E., C.O., G.C., H.F., T.B., P.E.N., and A.L.-O.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-GFP polyclonal antibody	Invitrogen	A-6455; RRID: AB_221570
Rabbit anti-FLAG polyclonal antibody	Millipore	F7425; RRID: AB_439687
HRP-conjugated anti-rabbit IgG	Millipore	#12-348; RRID: AB_390191
Bacterial and virus strains
Escherichia coli DH5α	Woodcock et al.44
Escherichia coli MG1655	Guyer et al.45
Escherichia coli MC1000	Casadaban and Cohen46
Escherichia coli ATCC25922	ATCC strain collection
Escherichia coli AS19	Sekiguchi and Iida47
Escherichia coli MG1655ΔaraBAD::FRT	Desai and Rao48
Escherichia coli MC1000 harboring pCP-C1	This study
Escherichia coli MC1000 harboring pCP-C2	This study
Escherichia coli MC1000 harboring pCP-C4	This study
Escherichia coli MC1000 harboring pCP-C5	This study
Escherichia coli MC1000 harboring pCP-C6	This study
Escherichia coli MC1000 harboring pCP-C7	This study
Escherichia coli MC1000 harboring pCP-C8	This study
Escherichia coli MC1000 harboring pCP-C9	This study
Escherichia coli MC1000 harboring pCP-C10	This study
Escherichia coli MC1000 harboring pCP-C11	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE92	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pCP-C3	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE96	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE102	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE106	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE108	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE133	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE134	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE59	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE60	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE80	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pCP-C1	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pCP-C2	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE173	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE81	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE213	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE214	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE215	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE216	This study
Escherichia coli MG1655ΔaraBAD::FRT harboring pAE257	This study
Escherichia coli ML-35pΔaraBAD::FRT	This study
Escherichia coli ML-35pΔaraBAD::FRT harboring pAE133	This study
Escherichia coli ML-35pΔaraBAD::FRT harboring pAE80	This study
Escherichia coli BW25113	Datsenko and Wanner49
Escherichia coli BW25113ΔrfaF732::kan	Datsenko and Wanner49
Escherichia coli BW25113ΔrfaE745::kan	Datsenko and Wanner49
Escherichia coli BW25113ΔrfaC733::kan	Datsenko and Wanner49
Escherichia coli MG1655ΔwaaQGPSBOJYZU::cam	This study
Escherichia coli ESBL20150072	Hasman et al.50
Escherichia coli ST131 (CTX-M-15)	Hasman et al.50
Pseudomonas aeruginosa PAO1	Stover et al.51
Pseudomonas aeruginosa SNRC	Jochumsen et al.52
Pseudomonas aeruginosa ATCC27853	ATCC strain collection
Klebsiella pneumoniae ATCC700603	ATCC strain collection
Klebsiella pneumoniae ST512	Grundmann et al.53
Bacillus subtilis 168	ATCC strain collection
Chemicals, peptides, and recombinant proteins
Chloramphenicol	AppliChem	CAS: 56-75-7
Ampicillin sodium salt	AppliChem	CAS: 69-52-3
Kanamycin sulfate	AppliChem	A4789,0010
NuPAGE™ MES SDS Running Buffer (20X)	Invitrogen	NP0002
CPRG / Chlorophenol red-β-D-galactopyranoside	Merck	CAS: 99792-79-7
IPTG	Merck	CAS: 367-703-03
Colistin sulfate	Thermo Scientific Chemicals	J60915.03
Melittin	Merck	M2272-1MG
Polymyxin B sulfate	Thermo Scientific Chemicals	J61763.03
Ciprofloxacin hydrochloride	Merck	CAS: 86393-32-0
Rifampicin	Serva	CAS:13292-46-1
Oncocin	This study
P15	This study
P38	This study
[methyl-3H]-thymidine, [5-3H]-uridine	Perkin Elmer
L-[2,3,4-3H]-arginine monohydrochloride	Perkin Elmer
D-[6-3H(N)]-glucosamine hydrochloride	Perkin Elmer
DiBAC4	Nordic Biosite	M2272-1MG
ONPG	Thermo Fisher Scientific	34055
Sytox Green	Invitrogen	S7020
Phosphatidylethanolamine (PE)	European Pharmacopoeia Reference Standard	Y0001903
Phosphatidylglycerol (PG)	Avanti Polar Lipids	841188P-10mg
Cardiolipin (CL)	Avanti Polar Lipids	841199P-10mg
Phosphatidylserine (PS)	Avanti Polar Lipids	870336P-25MG
Egg phosphatidylcholine (PC); 95%	Avanti Polar Lipids	131601P-1G
Human serum	Sigma-Aldrich	H3667-100ML
Human serum albumin	Sigma-Aldrich	A1653-500mg
Critical commercial assays
Quick-Start Bradford Protein Assay	Bio-Rad	#5000205
SDS Precast Gel: NuPAGE 4-12% Bis-Tris Midi gel	Invitrogen	WG1403A
PageBlue Protein staining	Thermo Scientific	24620
Nylon membrane (GE Healthcare Amersham Hybond-N+)	Amersham	10600021
Pierce ECL Western Blotting Substrate	Thermo Scientific	32109
Phusion Hot Start II Polymerase	Thermo Scientific	F537S
Dream Taq Polymerase	Thermo Scientific	EP0701
Monarch PCR&DNA clean-up kit	NEB	T1130S
EagI-HF	NEB	R3505S
Fast digest BamHI	ThermoFisher Scientific	FD0055
T4 DNA ligase	ThermoFisher Scientific	EL0011
QIAprep Spin Miniprep kit	Qiagen	#27106
QIAquick PCR purification kit	Qiagen	#28506
Qubit	ThermoFisher Scientific	Q32850
LIVE/DEAD BacLight Bacterial viability kit	Invitrogen	L7012
Oligonucleotides
See Table S2
Recombinant DNA
Plasmid:
pBAD-GFPuv	A gift from Jonathan Weissman
pEGFP-C1	Clontech
pACYA184	Chang and Cohen54
pCP	This paper
pCP(S219I)	This paper
pAE59	This paper
pAE60	This paper
pAE80	This paper
pAE81	This paper
pAE92	This paper
pAE96	This paper
pAE102	This paper
pAE106	This paper
pAE108	This paper
pAE133	This paper
pAE134	This paper
pAE173	This paper
pAE213	This paper
pAE214	This paper
pAE215	This paper
pAE216	This paper
pAE257	This paper
pCP-C1	This paper
pCP-C2	This paper
pCP-C3	This paper
pCP-C4	This paper
pCP-C5	This paper
pCP-C6	This paper
pCP-C7	This paper
pCP-C8	This paper
pCP-C9	This paper
pCP-C10	This paper
pCP-C11	This paper
ASKA plasmid collection	Kitagawa et al.35
Other
Semi-Dry electroblotter	JKA Biotech
ImageQuant LAS4000	GE Healthcare
Micro Pulser	BioRad
C1000 Touch Thermal Cycler	BioRad

Experimental model and study participant details

Bacterial strains and plasmids

Bacterial strains used in this study are listed in Table S1. Details on plasmid construction including primer sequences are described in Table S2. DNA and amino acid sequences of CP and the peptides used in this study in are listed in Table S3.

Bacterial growth conditions

Bacteria were grown aerobically at 30°C or 37°C either in Luria-Bertani (LB; 1% tryptone, 0.5% yeast extract and 1% NaCl) unless the salt concentration is specified otherwise, Mueller-Hinton II medium (MHII; 1.75% casein acid hydrolysate, 0.3% beef extract and 0.15% starch, adjusted to final pH 7.3 +/-0.2 at 25°C) or in AB minimal medium⁵⁵ supplemented with 10 μg/ml thiamine and 0.2% glycerol (ABTGly) or 0.2% glucose (ABTG). Bacillus subtilis was grown in Spizizen minimal medium⁵⁶ supplemented with 0.5% glucose and 0.05 mg/ml tryptophan. When required, leucine was added to 10 μg/ml and casamino acids (CAA) to 0.5% unless specified otherwise. For induction of the pBAD promoter, 0.2% L-arabinose was used. When necessary, antibiotics were used at the following concentrations: chloramphenicol, 20 μg/ml; ampicillin, 150 μg/ml; kanamycin, 50 μg/ml. Bacterial cell density was monitored by measuring the optical density at 600 nm for LB medium and at 450 nm for minimal medium.

Method details

Drop assay

Bacterial cultures were incubated overnight in the indicated medium at 30°C or 37°C. Ten-fold serial dilutions were prepared in 0.9% NaCl, and then 5 μl of each bacterial dilution were plated with or without 0.2% L-arabinose. Plates were incubated overnight at 30°C or 37°C.

Expression of CP-protein/peptide fusions & Western-blot analysis

Exponentially growing cells (OD₄₅₀=0.2) were diluted 1:10 in medium with or without 0.2% L-arabinose and incubated for 5 h. Cells from 3 ml samples were harvested by centrifugation, resuspended in SDS sample buffer,⁵⁷ and boiled for 15 min. Samples for total protein assay were centrifuged at 13000×g for 15 min at 4°C to remove cell debris, followed by quantification using Quick-Start Bradford Protein Assay (Bio-Rad) according to the manufacturer's instructions. Total protein (6-8 μg per lane) was separated by using SDS-PAGE (Precast Gel: NuPAGE 4-12% Bis-Tris Midi gel; Invitrogen) and stained with PageBlue Protein Staining (ThermoScientific) for protein visualization. For western-blot analysis, SDS-PAGE gels were blotted on nylon membrane (GE Healthcare Amersham Hybond-N⁺) using a Semi-Dry electroblotter (JKA Biotech). After transfer, the membrane was blocked with 5% non-fat milk in TBS with 0.1% Tween-20 (TBST) overnight at 4°C. The blot was then probed with either rabbit anti-GFP polyclonal antibody (Invitrogen) or rabbit anti-FLAG polyclonal antibody (Millipore), followed by immunodetection using HRP-conjugated anti-rabbit IgG (Merck) and Pierce ECL Western Blotting Substrate (Thermo Scientific). Chemiluminescence was detected by using ImageQuant LAS4000 (GE Healthcare).

CP-peptide library construction

The CP-peptide library was generated by modifying a previously described PCR-based method in which the random oligonucleotides encoding the peptide library are incorporated in the reverse primer.⁵⁸ The library construction is described in Figure S1. Briefly, a 1^st PCR was performed with the reverse primer #134 (library primer introducing a random nucleotide sequence encoding 12 amino acid peptide library) and the forward primer #133 using pCP as a template (Phusion Hot Start II polymerase; 1×(98°C 3min), 30×(98°C 10s, 60°C 30s, 72°C 30s), 1×(72°C 10min)). The PCR product from the 1^st PCR reaction was purified using Monarch PCR&DNA clean-up kit (NEB), and then used as template in a 2^nd PCR using forward primer #130 and reverse primer #133. The purified PCR product (Monarch PCR&DNA clean-up kit (NEB) from the 2^nd PCR was sequentially digested with EagI (NEB) and BamHI (ThermoFisher), and ligated into pCP, which was also cut with EagI and BamHI, resulting in the CP-peptide library. The ligation mix was purified using Monarch PCR&DNA clean-up kit (NEB) before transformation into electrocompetent E. coli MC1000. All colonies were pooled, followed by either plasmid DNA extraction or frozen as glycerol stocks at -80°C for storage. Next-generation amplicon sequencing was used the estimate the size of the CP-peptide library (Figure S3). Briefly, CP-peptide library plasmid DNA was isolated from non-induced E. coli MC1000 cells using QIAprep Spin Miniprep kit (Qiagen) and used for amplication generation by PCR with forward primer #234 and reverse primer #256 (Phusion Hot Start II polymerase; 1×(98°C 3min), 30×(98°C 10s, 65°C 10s, 72°C 10s), 1×(72°C 10min)). The PCR product was purified using QIAquick PCR purification kit (Qiagen) and the DNA concentration was measured by Qubit (ThermoFisher Scientific), and 30 ng were sent for amplicon sequencing with primer #256 (Eurofins Genomics; Illumina, paired end). Mothur's make.contig was used to create contigs from pair ends, and duplicate contigs were removed.⁵⁹

CPRG screening and sequencing

MG1655ΔaraBAD transformed with CP-peptide library were plated on LB (2% NaCl) agar supplemented with 20 μg/ml CPRG, 30 μM IPTG, 0.2% L-arabinose and 20 μg/ml chloramphenicol, and was then incubated at 30°C. After 18-20 h incubation at 30°C, colonies displaying a pink CPRG-positive phenotype were restreaked on LB supplemented with chloramphenicol (20 μg/ml). Plasmids from CPRG-positive colonies were isolated and retransformed into MG1655ΔaraBAD to exclude genomic mutations causing a CPRG-positive phenotype. Finally, the part of the plasmid encoding CP and the random peptide sequence was amplified using primer #120 and primer #121, and the PCR product was sequenced using primer #121.

Microscopy

For intracellular peptide expression, overnight cultures were diluted 1:100 in ABTGly supplemented with casamino acids and grown at 30°C until OD₄₅₀=0.2. The cultures were then diluted 1:10 in 10 ml medium with or without 0.2% L-arabinose and induced for 5h at 30°C. Cells were stained with LIVE/DEAD™BacLight™ Bacterial Viability Kit (Invitrogen). Imaging was performed on a Nikon Eclipse Ti inverted microscope attached to a Zyla 5.5 sCMOS camera. Images were analyzed with NIS elements and ImageJ.

Automated SPPS Protocol

Peptides (P15, P38 and oncocin) were assembled via MW-assisted Fmoc-based solid-phase peptide synthesis by using an automated CEM Liberty™ MW peptide synthesizer as reported recently.⁶⁰ In brief, couplings were performed with 0.2M solutions Fmoc-protected amino acid building blocks (5.0 equiv), N,N′-diisopropylcarbodiimide (DIC) (0.5 M in DMF; 5.0 equiv), and ethyl (hydroxyimino)cyanoacetate (OxymaPure®; 2M in DMF; 5.0 equiv). Fmoc deprotection was performed with 20% piperidine in DMF at 75°C (37 W for 30 sec followed by 40 W for 180 sec), while amino acid couplings were conducted at 75°C (25 W for 300 sec). All peptides were released by treatment with TFA–TIS–H₂O (95:2.5:2.5; 4 mL) for 2 × 1 h, eluting the resin with additional CH₂Cl₂. The combined eluates were then co-evaporated with toluene under vacuum. The obtained residues were carefully triturated with diethyl ether to remove impurities derived from protecting groups.

Purification by preparative HPLC and characterization

Water used for analytical and preparative high-performance liquid chromatography (HPLC) was filtered (0.2 μm) in-house by using a Labo Star® Pro TWF system. Analytical HPLC was performed on a Phenomenex Luna® Omega Polar C18 column (150 mm × 4.6 mm; particle size: 3 μm; pore size: 100 Å) using a Shimadzu Prominence and Shimadzu Nexera system, eluting with H₂O-acetonitrile (MeCN) gradients with 0.1% trifluoroacetic acid (TFA) added to the eluents (A: 5:95 MeCN–H₂O + 0.1% TFA; B: 95:5 MeCN–H₂O + 0.1% TFA) with UV detection at λ = 220 nm. Preparative HPLC was performed by using a Phenomenex Luna® Omega Polar C18 column (250 × 21.2 mm; particle size: 5 μm; pore size 100 Å) on a Shimadzu Prominence system using the same eluents as for analytical HPLC. MALDI-TOF MS analysis was performed by using a Bruker Microflex LT detector.

Crude peptides were purified by preparative HPLC using a linear gradient of 0-30% B during 20 min for P15, and 0-40%B during 20 min for P38. The purity (>95%) and identity of the peptides were verified by analytical HPLC (and MALDI-TOF MS analysis). P15: RT= 11.36 min (0-40% B during 15 min); P38: RT = 11.20 min (0-60% B during 15 min). The purified peptides were freeze-dried and stored at -20°C until biological testing.

Minimal inhibitory concentration (MIC) assay

The broth microdilution assay for antimicrobial susceptibility testing was performed in 96-well polypropylene microtiter plates. Each well contained 10⁵-10⁶ CFU/ml bacteria in the exponential growth phase, and the peptides prepared by two-fold serial dilutions, in a total volume of 200 μl. The plates were incubated at 37°C for 18-20 h. The MIC is defined as the lowest concentration that inhibits bacterial growth upon visual inspection.

Adaptive Laboratory Evolution (ALE)

MG1655 was evolved 40 days to P38, colistin, and ciprofloxacin. Twenty colonies of MG1655 were inoculated separately in ABTG medium and allowed to grow overnight at 37°C. On day one of the experiment, the overnight cultures were diluted 1:100 in 2 ml ABTG and incubated at 37°C for 24 h. The next day, the cultures were again diluted 1:100 in 2 ml ABTG, but this time containing either no treatment or 0.25×MIC of P38, colistin, or ciprofloxacin. Five replicates were included for each treatment. Every 24 h, the cultures were transferred (1:100 dilution) to new tubes containing 50% higher drug concentration, such that the MIC would be surpassed at day 6. Starting from day 6, the cultures were transferred to two tubes, one with the same concentration as the previous day and one with a 50% higher concentration of antibiotic/AMP. If the cultures failed to grow in both tubes after 24 h incubation, they were left to incubate for another 24 h. Cultures were saved at -80°C with 25% glycerol during every transfer. The evolution experiment was stopped either after 40 days or when the cultures were able to grow at more than 1000×MIC.

Overexpression pooled ASKA collection

The individual clones of the ASKA library³⁵ were pooled, plasmid DNA extracted and transformed into MG1655 via electroporation. Transformants were selected on LB plates containing 20 μg/mL chloramphenicol, and the cells harboring the pooled ASKA plasmids were scraped and pooled in ABTG, and aliquots were stored at -80°C with 50% glycerol. MICs were determined for MG1655 harboring pooled ASKA plasmids or the empty pCA24N plasmid in the presence of 1, 10, or 100 μM IPTG to induce the expression of the genes from the ASKA plasmids.

Time-kill assay

Overnight cultures of MG1655 prepared in ABTG were diluted 1:100 and grown at 37°C until log phase (OD₄₅₀∼0.2). Exponentially growing cultures at 5×10⁵ CFU/ml in a total volume of 5 ml were then treated with P38, colistin or oncocin. To test the peptide activity on stationary phase cultures, P38 or colistin were added directly to 5 ml overnight culture. Each treatment was performed in triplicate. The bacterial cultures were incubated at 37°C with shaking at 200 rpm. At the time points 0, 1, 2, 4, 6, 8, and 24 h after peptide addition, 100 μl samples were taken from each culture, and 10-fold serial dilutions prepared in 0.9% NaCl were drop-plated (5 μl) on LB agar plates. Plates were incubated at 37°C overnight before counting colonies.

Macromolecular synthesis assay

Overnight cultures of MG1655 grown either in ABTG or ABTGly were diluted 1:100 in 15 ml fresh medium. The cultures were grown at 37°C to an OD₄₅₀=0.1 before P38 or control antibiotic were added. At time points 0, 10, 20, 30, 40, 50, and 60 min, a 500 μl sample was taken and added to a tube containing 0.375 μCi [methyl-³H]-thymidine, [5-³H]-uridine, L-[2,3,4-³H]-arginine monohydrochloride, or D-[6-³H(N)]-glucosamine hydrochloride (PerkinElmer) in order to quantify DNA, RNA, protein, and cell wall synthesis, respectively. OD was also measured at these time points to monitor cell growth. The radiolabeled precursors were allowed to incorporate at 37°C for 4 min (thymidine), 2 min (uridine), 1 min (arginine), or 15 min (glucosamine), before adding 5 ml ice-cold trichloroacetic acid (TCA) containing 0.1M NaCl to stop the reaction. Samples were kept on ice for at least 30 min, before the TCA-precipitated material was collected by filtering through 0.5 μm glass fiber filters. The filters were washed twice with 5 ml 10% TCA, placed in scintillation vials and allowed to dry overnight. Finally, 5 ml scintillation liquid (Ultima Gold) were added and the incorporation of radiolabeled precursors was quantified using a Tri-Carb 2910TR liquid scintillation analyzer (PerkinElmer).

Membrane depolarization studies using DiBAC₄(3)

Overnight cultures of MG1655 were diluted 1:100 in ABTGly supplemented with casamino acids and grown at 37°C until OD₄₅₀=0.3-0.6 was reached. Bacterial cells were washed and resuspended with 5mM HEPES supplemented with 20 mM glucose to OD₄₅₀=0.1. The resulting suspension was mixed with DiBAC₄ (3) (to reach a final concentration of 1 μg/ml) and distributed into a 96-well microtiter plate (90 μl/well), followed by the addition of 10 μl of 10× compound solutions (1/4× to 1×MIC P38, Melittin, or Polymyxin B) after 10 min. DiBAC₄ (3) and compound only (without cells) were included to check for any interaction of the compounds with the fluorescence probe. Excitation and emission wavelengths on the microtiter plate reader were set to 490 nm and 515 nm, respectively.

Inner membrane permeabilization assay using o-nitrophenyl-β-D-galactopyranoside

Peptide-induced inner membrane permeability was measured as entry of o-nitrophenyl-β-D-galactopyranoside (ONPG) into cells. Overnight cultures of E. coli ML-35p were diluted in ABTG supplemented with leucine and grown at 37°C until OD₄₅₀=0.2. The cells (100 μl) were added to 96-well microtiter plate containing 50 μl compound (1/2× to 16×MIC P38 or colistin) and 50 μl ONPG were added to a final concentration of 1.5 mM. Toluene (5%) was used as a positive control. The absorbance at 420 nm was recorded every 15 min for 90 min at 37°C using SYNERGY H1 microplate reader (BioTek).

For intracellular peptide expression, exponentially growing cells (OD₄₅₀=0.2) were diluted 1:10 in ABTgly supplemented with casamino acids and with or without 0.2% L-arabinose and incubated for 5 h at 30°C. Cells (100 μl) were added to a 96-well microtiter plate containing 50 μl medium with or without 0.2% L-arabinose and 50 μl ONPG added to a final concentration of 1.5 mM. The absorbance at 420 nm was recorded every 15 min for 1h at 30°C using SYNERGY H1 microplate reader (BioTek). The absorbance at 600 nm was also recorded in order to normalize the ONPG absorbance to the cell number.

Inner membrane disruption studies using Sytox Green

Overnight cultures of MG1655 were diluted 1:100 in ABTGly supplemented with casamino acids and grown until mid-log phase at 37°C. Cells were harvested, washed twice, and resuspended in 5mM HEPES supplemented with 20 mM glucose (pH 7.4) to reach OD₄₅₀ of 0.1. Sytox Green (Invitrogen) was added to the cells to a final concentration of 1μM. The cells (90 μl) were added to the wells of a black 96-well microtiter plate, and the fluorescence (λ_ex = 480 nm, λ_em = 520 nm) was measured every 5 min. After 10 min, 10 μl compound (1/4× to 1×MIC P38, Melittin, or Polymyxin B) was added, and the fluorescence was measured continuously every 5 min for 30 min. Triton X-100 (0.1%) was used as a positive control, and Sytox Green and compound only (without cells) were included to check for any interaction of the compound with the fluorescence probe.

Hemolysis assay

The hemolytic activity of the peptides was determined by measuring hemoglobin release from human erythrocytes. Briefly, human blood was centrifuged for 15 min at 1000×g. Plasma was discarded and the erythrocytes were washed three times with PBS. The washed erythrocytes were then suspended in 10% v/v PBS. The erythrocytes (100 μl) were added to a 96-well microtiter plate containing equal volume of peptide (final concentration of 0.25-128 μg/ml P38 or rifampicin) in PBS. Triton X-100 (0.2%) and PBS only were used as positive and negative control, respectively. The plate was incubated for 1 h at 37°C before centrifuging for 10 min at 1300×g. The supernatants were transferred to a new microtiter plate and the hemoglobin release was determined by measuring absorbance at 540 nm. The percentage hemolysis was calculated as (A_sample – A_PBS)/(A_Triton – A_PBS) × 100%, where A is the absorbance at 540 nm.

Quantification and statistical analysis

Statistical parameters are reported in the figure legends and figures. The number of independent experimental replicates is specified in each figure legend.

Published: May 9, 2025