Engineered cyclotides with potent broad in vitro and in vivo antimicrobial activity

Abstract

The search for novel antimicrobial agents to combat microbial pathogens is intensifying in response to rapid drug resistance development to current antibiotic therapeutics. Respiratory failure and septicemia are the leading causes of mortality among hospitalized patients. Here, we report the development of a novel engineered cyclotide with effective broad-spectrum antibacterial activity against several ESKAPE bacterial strains and clinical isolates. The most active antibacterial cyclotide was extremely stable in serum, showed little hemolytic activity and provided protection in vivo in a murine model of P. aeruginosa peritonitis. These results highlight the potential of the cyclotide scaffold for the development of novel antimicrobial therapeutic leads for the treatment of bacteremia.

Graphical Abstract

By using a topologically modified sequence of protegrin PG-1, Ganesan et al report the development of novel engineered cyclotides with effective broad-spectrum antibacterial activity against several ESKAPE bacterial strains and clinical isolates. The most active antibacterial cyclotide showed little hemolytic activity and was extremely stable in serum. In addition, this cyclotide was able to provide protection in vivo in a murine P. aeruginosa-induced peritonitis model.

Introduction

The search for novel antimicrobial agents is intensifying, in response to the threat of microbial pathogens and the increasing development of drug resistance to current antibiotic therapeutics. According to the Centers for Disease Control and Prevention, the six ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumanii, Pseudomonas aeruginosa, and Enterobacter species) bacterial species cause two-thirds of health care-associated infections (e.g. pneumonia, septicemia), leading to 99,000 deaths annually in the United States.^[1] A hallmark of these emerging difficult-to-treat clinical superbugs is their ability to “escape” the action of multiple traditional antibiotics, in part due to biofilm formation and mechanisms of drug resistance.^[2]. Antimicrobial peptides are essential host defense molecules found in a wide variety of species and are promising antibacterial therapeutic candidates.^[3] Several hundreds of antimicrobial peptides have been identified in a variety of life forms ranging from bacteria, fungi, plants, amphibians, to mammals, including humans.^[4] In mammals, cathelicidins, protegrins and defensins are the three of major types of host defense peptides.^[5]

Preliminary studies have shown that β-hairpin-containing antimicrobial peptides have potent antimicrobial activity and cell selectivity.^[6] For example, the two-β-strand protegrin 1 (PG-1) (Fig. 1), an 18-amino-acid long peptide, is a prototypic antimicrobial cationic peptide of the protegrin family isolated from porcine leukocytes.^[7] Protegrin PG-1 is smaller in size than α- and β-defensins but shows significant size and structural similarities with another family of antimicrobial peptides, the tachyplesins,^[8] showing also sequence homology with the N-terminal region of α-defensins.^[9] In solution, PG-1 forms a well-ordered antiparallel β-sheet structure stabilized by the presence of two disulfide bonds with disordered N- and C-termini.^[10] The presence of the disulfide bonds is required to maintain potent antimicrobial activity.^[11] PG-1 has been shown to disrupt anionic bacterial membranes and biofilms, showing also a wide range of in vivo immunomodulatory properties like inhibition of LPS and increasing neutrophil clearance.^{[6b, 6c]} This distinct antimicrobial mechanism of action of PG-1 limits potential cross-resistance while providing synergy in combination with other locally produced host defense peptides and/or conventional antibiotics.^[6d] The effectiveness of PG-1 in several different animal infection and inflammation models suggests that this type of peptide may represent a new class of antibiotic and immunomodulatory reagents.^[12] However, their therapeutic use is currently limited by their high cytotoxicity, hemolytic activity and suboptimal biological stability.^[13]

Figure 1.

Scheme depicting the approach used to design the different MCo-PG antimicrobial cyclotides. A circular permuted version of porcine protegrin PG-1, where the original Arg¹ residue in PG-1 was moved to its C-terminus, was grafted into onto of cyclotide loop 6 of cyclotide MCoTI-I between Gly¹ and Ser³³ residues. The backbone cyclized structure of the cyclotide is shown a connecting bond in green. Cys residues are highlighted in yellow and disulfide bonds are indicated in red. The ribbon structures of cyclotide MCoTI-II (PDB: 1IB9)^[33] and porcine protegrin PG-1 (PDB: 1PG1)^[7b] are shown for reference.

Cyclotides are fascinating micro-proteins (≈30 residues long) present in plants from different families including Violaceae, Rubiaceae, Cucurbitaceae, and Fabaceae families, among others.^[14] They have shown a broad array of biological activities such as protease inhibitory, anti-microbial, insecticidal, cytotoxic, anti-HIV, and hormone-like activities.^[15] They share a unique head-to-tail circular knotted topology of three disulfide bridges, with one disulfide penetrating through a macrocycle formed by the two other disulfides and inter-connecting peptide backbones, forming what is called a cystine knot topology^[15a] (Fig. 1). Cyclotides can be considered as natural combinatorial peptide framework structurally constrained by the cystine-knot scaffold and head-to-tail cyclization but in which hypermutation of essentially all residues is permitted with the exception of the strictly conserved cysteines that comprise the cystine knot.^{[15a, 15b]} Cyclotides are characterized by possessing a remarkable stability due to the presence of a backbone cyclized cystine knot topology, a small size making them readily accessible to chemical synthesis ^[16] and heterologous expression,^[17] and exceedingly tolerant to sequence variations and molecular grafting.^[15b] In addition cyclotides have shown to be orally active,^[18] and capable of crossing cell membranes^[19] to efficiently target intracellular targets in vivo.^[20] Altogether, these features make the cyclotide scaffold an excellent molecular framework for the design of novel peptide-based therapeutics,^{[15b, 21]} making them ideal substrates for molecular grafting of biological peptide epitopes.^[15a]

By using a topologically modified sequence of PG-1, we report here for the first time the development of a novel engineered cyclotide with effective broad-spectrum antibacterial activity against several ESKAPE bacterial strains and clinical isolates. The most active antibacterial cyclotide showed little hemolytic activity and was extremely stable in serum. In addition, this cyclotide was able to provide in vivo protection in a murine model of P. aeruginosa peritonitis. These results highlight for the first time the potential of the cyclotide scaffold for the development of novel therapeutic leads for the treatment of bacteremia.

Results and Discussion

In order to produce a cyclotide with PG-1 antimicrobial activity, we employed the naturally-occurring cyclotide MCoTI-I as molecular framework (Fig. 1). MCoTI- cyclotides are potent trypsin inhibitors isolated from the seeds of Momordica cochinchinensis^[22] and show very low toxicity in human cells,^{[19b, 20]} and therefore represent a desirable molecular scaffold for engineering new cyclotides with minimal toxicity and novel biological activities.^{[20, 23]}

According to the solution structure of PG-1,^[7b] its N- and C-termini are very close in space although the N-terminus is slightly more extended (Fig. 1). Therefore, a modified version of PG-1, where the N-terminal Arg residue was moved to the C-terminal position of the PG-1 sequence, was grafted into loop 6 of the cyclotide MCoTI-I (cyclotide MCo-PG2, Fig. 1). We also explored the effect of adding an extra disulfide to the grafted PG-1-derived sequence to further stabilize the grafted β-hairpin structure. This was accomplished by replacing both residues Arg⁴ and Gly¹⁷ in the original PG-1 sequence with Cys residues (MCo-PG3, Fig. 1). Two more cyclotides were also designed with longer and shorter versions of the PG-1-based grafted sequence to explore the effect of the distance of the grafted sequence from the cyclotide and to minimize the size of the PG-1-derived graft (Fig. 1). The elongated version (MCo-PG4) was obtained by adding two extra Gly residues to the N- and C-terminal positions of the modified PG-1 sequence. The shorten version (MCo-PG5) removed the N- and C-terminal Gly and Arg residues from the modified PG-1 sequence, respectively. These analogs were designed to explore the effect of the distance of the grafted sequence from the cyclotide and to minimize the size of the PG-1-derived graft. The different sequences were grafted onto loop 6 of cyclotide MCoTI-I by replacing residue Asp³⁴ (Fig. 1) as this loop has been shown to be less rigid in solution^[24] and quite tolerant to sequence grafting of relatively long peptide sequences.^{[20, 23d, 23e, 25]}

All grafted MCo-PG cyclotides were chemically synthesized on a sulfonamide resin using an Fmoc-based solid-phase peptide synthesis protocol.^[19b] The corresponding fully deprotected linear peptide α-thioesters were obtained by alkylation of the sulfonamide linker followed by thiolytic cleavage of the alkylated sulfonamide linker and acidolytic deprotection of the side-chain protecting groups. Cyclization and oxidative folding were accomplished in a one-pot reaction under thermodynamic control using aqueous buffer at pH 7.4 in the presence of 1 mM reduced glutathione (GSH). In all the cases the cyclization/folding reactions were complete in 72–96 h (Figs. 2A and S1). The yields for the cyclization/folding reactions ranged from 16% (MCo-PG3) to 40% (MCo-PG2) (Table S1). All folded cyclotides were purified by reverse-phase HPLC and characterized by ES-MS (Figs. 1B and S1, Table S1). In addition, cyclotide MCo-PG2 was also characterized by homonuclear NMR spectroscopy. The chemical shift Δδ values for most of the backbone protons for the common part shared with the parent cyclotide MCoTI-I were smaller than 0.1 ppm indicating that MCo-PG2 adopts a native cyclotide fold (Fig. 2C and S2, Table S2). Analysis of the through space nuclear Overhauser effect, NOE, connectivities in the 2D ¹H-¹H NOESY spectrum of cyclotide MCo-PG2 revealed long-range NOEs between the backbone H’ protons from residues Leu³⁷ and Val⁴⁸, and residues Tyr³⁹ and Val⁴⁶ in the protegrin-derived graft of cyclotide MCo-PG2; these NOEs are also present in PG-1^[7b] and are characteristic of a native β-hairpin fold (Fig. S3).

Figure 2.

Chemical synthesis and characterization of cyclotide MCo-PG2. A. Analytical HPLC traces of for the linear thioester precursor, GSH-induced cyclization/folding crude after 72 h and purified cyclotide. An arrow indicates the desired peptide. B. ES-MS characterization of pure MCo-PG2. The expected average molecular weight is shown in parenthesis. C. Chemical shifts differences of the backbone, H′ and H^α protons between the common sequence (residues 1 through 34) of MCoTI-I ^[24] and MCo-PG2 (Table S2).

Next, we tested the broad-spectrum antimicrobial activity of the different PG-1-grafted cyclotides against different strains of four ESKAPE pathogens, P. aeruginosa, S. aureus, K. pneumoniae, and E. coli (Table 1). The naturally occurring cyclotide MCoTI-I and the porcine protegrin PG-1 were used as negative and positive controls, respectively. The minimum inhibitory concentration (MIC) values for the different peptides were determined by broth microdilution assay using a cation-adjusted Mueller-Hinton broth (CAMHB).^[26] This growth medium contains 128 mM NaCl supplemented with calcium and magnesium salts providing very similar ionic strength to those found under physiological conditions. As expected, protegrin PG-1 exhibited potent and strong activity against Gram-negative and Gram-positive bacteria, with MIC values ranging from 0.03 μM (E. coli DS377) to 0.4 μM (S. aureus USA300 and HH35, both methicillin resistant strains; and K. pneumoniae BAA1705 and K6) (Table 1). This result is an agreement with published data for this protegrin.^[26] Interestingly, all PG-1-grafted MCoTI-based cyclotides showed antibacterial activity against P. aeruginosa, with MIC values from 25 μM for the less active cyclotide (MCo-PG3) to 1.6 μM for the most active cyclotides (MCo-PG2 and MCo-PG4) (Table 1). Cyclotide MCo-PG3 also showed little activity against S. aureus, K. pneumoniae and E. coli, with MIC values in all the cases above 25 μM, indicating that addition of an extra-disulfide bond to the grafted peptide significantly reduced its antimicrobial activity. Shortening the grafted PG-1-derived sequence also had a detrimental effect on the antimicrobial activity of cyclotide MCo-PG5 although the effect was not as pronounced as the observed for cyclotide MCo-PG3. Elongation of the grafted sequence by adding extra Gly residues had very little impact on the antimicrobial activity, with cyclotides MCo-PG2 and MCo-PG3 showing similar the same antibacterial activity. As shown in Table 1, MCo-PG2 was slightly more active than MCo-PG4 against P. aeruginosa, S. aureus and E. coli, but slightly less active against K. pneumoniae. As expected, the naturally-occurring cyclotide MCoTI-I did not show any antibacterial activity in this assay up to a concentration of 200 μM (Table 1), indicating that the antimicrobial activity of PG-1 grafted cyclotides was specific and comes from the grafted sequence.

Table 1.

Minimum inhibitory concentrations (MIC) of antimicrobial peptide PG-1 and MCo-PG2 through MCo-PG5 cyclotides. Naturally occurring protegrin PG-1 and cyclotide MCoTI-I were used as a positive and negative controls, respectively. Antimicrobial activities were performed by broth microdilution assays using cation-adjusted Mueller-Hinton broth (CAMHB). This growth medium contains 128 mM NaCl supplemented with Ca²⁺ and Mg²⁺ salts providing a very similar ionic strength to that of physiological conditions.

MIC (µM)
Peptide	P. aeruginosa	P. aeruginosa	S. aureus	S. aureus	S. aureus	S. aureus	K. pneumoniae	K. pneumoniae	E. coli	E. coli
	PAO1	PA27853	USA300[a]	25973	BAA977[b]	HH35[a]	BAA1705	K6	DS377	K12
PG-1	0.2	0.1	0.4	0.1	0.2	0.4	0.4	0.4	0.03	0.1
MCoTI-I	>200	>200	>200	>200	>200	>200	>200	>200	>200	>200
MCo-PG2	1.6	1.6	6.2	3.1	3.1	6.2	12.5	12.5	0.8	0.8
MCo-PG3	25	25	>25	>25	>25	>25	>25	>25	>25	>25
MCo-PG4	1.6	1.6	12.5	12.5	12.5	12.5	12.5	12.5	1.6	1.6
MCo-PG5	3.1	3.1	12.5	12.5	12.5	12.5	6.2	6.2	1.6	1.6

^[a]Methicillin resistant strain

^[b]Clindamycin resistant strain

Based on the superior spectrum of activity of cyclotide MCo-PG2 against three of the four ESKAPE pathogens tested in our study, and in particular P. aeruginosa and S. aureus, which are two ESKAPE pathogens that commonly infect the airways of patients with cystic fibrosis, we decided to further test the antimicrobial activity of cyclotide MCo-PG2 against 20 different clinical isolates of P. aeruginosa and S. aureus. These strains were collected from patients suffering from cystic fibrosis at the Keck Medical Center, University of Southern California (Table 2 and S3). Remarkably, MCo-PG2 retained its antimicrobial activity against P. aeruginosa and S. aureus clinical isolates, with MIC values ranging from 0.4 μM to 12.5 μM (Table 2). The median MIC (MIC₅₀) and MIC 90% (MIC₉₀) values for the P. aeruginosa population (n=20) were 1.5 μM while and 3.1 μM, respectively. For the S. aureus isolates (n=20), the MIC₅₀ and MIC₉₀ were 6.25 μM and 12.5 μM, respectively, indicating that MCo-PG2 shows four times better antimicrobial activity (MIC₉₀ values) against P. aeruginosa than to S. aureus stains. In comparison to protegrin PG-1, cyclotide MCo-PG2 was around four and ten times less active (MIC₉₀ values) against P. aeruginosa and S. aureus than the natural protegrin peptide (Table 2). These results were extremely encouraging indicating cyclotide MCo-PG2 was able to maintain good MIC values against pathogenic clinical isolates. It is important to remark that 30% of the P. aeruginosa clinical isolates were multidrug resistant strains (MDR), while 100% of the S. aureus clinical strains were methicillin-resistant, hence further highlighting the significance of MCo-PG2 MIC values against these pathogens.

Table 2.

Minimum inhibitory concentration (MIC) of antimicrobial peptides MCo-PG2 and PG-1 against clinical isolates of P. aeruginosa (n=20) and methicillin-resistant S. aureus (n=20) collected from patients suffering from cystic fibrosis at the Keck Medical Center, University of Southern California. Antimicrobial activities were performed as described in Table 1. Colistin and vancomycin were used as positive controls for P. aeruginosa and S. aureus, respectively.

	MIC (μM)
	PG-1	MCo-PG2	Colistin
P. aeruginosa MIC50	0.2	1.5	≤ 0.2
P. aeruginosa MIC90	0.8	3.1	0.4
P. aeruginosa MICrange		0.05–1.5	0.4–12.5
≤ 0.2–0.4
	PG-1	MCo-PG2	Vancomycin
S. aureus MIC50	0.4	6.3	0.7
S. aureus MIC90	0.8	12.5	0.4
S. aureus MICrange	0.2–0.8	3.1–12.5	0.5–1.4

Next, we used a time-kill kinetic assay to establish the bactericidal activity of cyclotide MCo-PG2 against P. aeruginosa PAO1 (Fig. 3A). This was accomplished by using different MCo-PG2 concentrations ranging from 0.25 x MIC to 16 x MIC values. The results indicated a rapid and concentration dependent killing kinetics against P. aeruginosa PAO1 by MCo-PG2 with greater than 3 log₁₀ CFU/mL bactericidal activity at concentrations of 4 times the MIC value. It is important to highlight that by using 16 times the MIC value of MCo-PG2 no regrowth of P. aeruginosa after 24 h was observed (Fig. 3A).

Figure 3.

Cytotoxic activities of cyclotide MCo-PG2. A. Bactericidal activity of PG-1 against log-phase P. aeruginosa PAO1. P. aeruginosa was grown to log phase, and aliquots were treated with compounds at incremental concentrations relative to MICs, from to 0.25 x MIC to 16 x MIC. B. Hemolytic activity of protegrin PG-1 and cyclotide MCo-PG2. Hemolytic activity was determined using human erythrocytes in PBS. Peptide concentrations causing 50% hemolysis (HC₅₀) were derived from the dose-response curves. C. Cytotoxic profile of protegrin PG-1 and cyclotide MCo-PG2 to various mammalian cells (A549 and HEK293T). Cells were treated with increasing concentrations of the corresponding peptides. Cell viability was assessed by using the MTT assay. Cyclotide MCoTI-I was used as control. Data are mean ± SEM for experiments performed in triplicate.

We also evaluated the hemolytic activity of cyclotide MCo-PG2. As shown in Fig. 3B, MCo-PG2 exhibited a significantly lower hemolytic activity (HC₅₀ = 88 ± 5 μM) than that of protegrin PG-1 (HC₅₀ = 6.3 ± 1.6 μM). As expected, the control cyclotide MCoTI-I did not have any hemolytic activity up to a concentration of 100 μM. The membranolytic selectivity index (HC₅₀/MIC) is often used as an indicator of the therapeutic potential of a peptide-based antibiotic.^[27] The HC₅₀/MIC₅₀ values for MCo-PG2 and PG-1 against P. aeruginosa clinical isolates were around 60 and 32, respectively. The HC₅₀/MIC₅₀ values for S. aureus clinical strains were found to be similar for PG-1 and MCo-PG2 with value around 15. These results indicate that cyclotide MCo-PG2 has greater therapeutical potential than PG-1 against P. aeruginosa, while showing similar therapeutic potential against S. aureus.

The cytotoxicity profile of cyclotide MCo-PG2 was also studied using two types of human epithelial cells: HEK293T (transformed kidney epithelial cells) and A549 (lung carcinoma). As shown in Fig. 3C, the cyclotide MCo-PG2 was about three times less toxic than PG-1. As previously reported,^[20] the control cyclotide MCoTI-I did not present any cytotoxicity in human cells up to 100 μM.

The biological stability of cyclotide MCo-PG2 was explored and compared to that of the empty scaffold (MCoTI-I) and protegrin PG-1 (Fig. S4). This was accomplished by incubating the corresponding peptides in human serum at 37° C. The quantitative analysis of undigested polypeptides was performed using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). MCoTI-cyclotides present a very rigid structure,^[24] which makes them extremely stable to proteolytic degradation. Remarkably, cyclotide MCo-PG2 showed slightly greater stability in human serum (τ_1/2 = 60 ± 6 h) than the parent cyclotide MCoTI-I (τ_1/2 = 52 ± 5 h, Fig. S4). More importantly, cyclotide MCo-PG2 displayed minimal degradation within the first 24 h of the serum stability assay, while 40% of cyclotide MCoTI-I was degraded during the first 24 h of the assay (Fig. S4). In contrast, protegrin PG-1 was degraded significantly faster than MCo-PG2 (τ_1/2 = 30 ± 3 h) also showing significant degradation after 24 h of incubation with serum. A linearized, reduced and alkylated version of MCoTI-I was used as positive control and as expected was rapidly degraded (τ_1/2 = 18 ± 6 min). These results highlight the importance of the circular Cys-knot topology for proteolytic stability.

Encouraged by these results, we decided next to explore the biological activity of cyclotide MCo-PG2 in vivo. We first determined the toxicity profile of MCo-PG2 and PG-1 in Balb/c mice (n=3) using intraperitoneal (i.p.) administration (Fig. S5). Colistin was used as a control antibiotic.^[28] The studies revealed that intraperitoneal doses of 5 mg/kg for PG-1, 25 mg/kg for MCo-PG2 and 15 mg/kg for colistin were well tolerated by mice causing only very mild toxicity after 1 h of dosing with all recovering after 24 h (Fig. S5). This maximum tolerated dose found for colistin is consistent with previously published data.^[28] Based on these results, we decided to use the corresponding compound MTDs to test the antimicrobial activity in vivo. For this purpose, we employed a P. aeruginosa bacterial peritonitis model.^[29] This animal model is a well-established acute infection model and is commonly utilized as a common preclinical screening method for new antibiotics.^[30] Peritonitis in Balb/c mice (n=10) was established by intraperitoneal injection of 1.5×10⁷ colony forming units (CFU) per mouse of P. aeruginosa (Schroeter) Migula (ATCC 27853). The animals were then immediately treated by intraperitoneal injection with PBS, PG-1 (5 mg/kg), MCo-PG2 (10 or 25 mg/kg) and colistin (15 mg/kg). As shown in Fig. 4, single-dose administrations of 10 mg/kg and 25 mg/kg of cyclotide MCo-PG2 in the septic mice were associated with high survival rates (hazard ratio [HR]: 0.0875 and 0.048, respectively; p <0.001) comparable to those obtained in animals treated with 5 mg/kg PG-1 and 15 mg/kg colistin ([HR]: 0.040; p <0.001). After day 3 post-treatment, all the animals treated with PBS or the corresponding compound that survived were completely healthy and no further dead or moribund mice were observed over the course of the seven-day experiment (Fig. 4).

Figure 4.

Evaluation of cyclotide MCo-PG2 against P. aeruginosa (Schroeter) Migula (ATCC 27853) in a P. aeruginosa-induced bacterial peritonitis model.^[29] P. aeruginosa was administered to mice by intraperitoneal injection 1.5 × 10⁷ colony forming units (CFU) per mouse. The animals were then immediately treated by intraperitoneal injection with PG-1 (5 mg/kg) and MCo-PG2 (10 or 25 mg/kg). Colistin (15 mg/kg) and PBS were used as positive and negative controls. The numbers of surviving mice were determined daily for 7 days. Single-dose administrations of MCo-PG2 (10 mg/kg, 1.8 μmol/kg; 25 mg/kg, 4.5 μmol/kg) were associated with a high survival rate of septic mice (Hazard ratio (HR): 11.4 and 20.8 respectively, p <0.001) comparable to treatments with PG-1 (5 mg/kg, 2.3 μmol/kg) and 15 mg/kg colistin (15 mg/kg, 12.3 μmol/kg) (HR: 24.8, p <0.001).

In summary, we report here for the first time the design and synthesis of a novel cyclotide with broad-spectrum antimicrobial activity in vitro against different ESKAPE pathogens (P. aeruginosa, S. aureus, K. pneumoniae, and E. coli), including 20 clinical isolates for the human pathogens P. aeruginosa and S. aureus, and more importantly in vivo using a murine model of acute P. aeruginosa peritonitis. This was successfully accomplished by grafting a series of topologically modified peptides based on the porcine protegrin PG-1 sequence onto loop 6 of the cyclotide MCoTI-I. Structural studies in solution by ¹H-NMR also revealed that the new antimicrobial cyclotide adopts a native cyclotide scaffold, allowing the grafted PG-1-based sequence to assume a bioactive native conformation. This emphasizes the tolerance of this loop in the MCoTI-based cyclotide family for the molecular engraftment of long peptide sequences.^{[15b, 31]} For example, the sequence engrafted in the bioactive cyclotide MCo-PG2 was 18 residues long containing two extra-disulfide bonds. The most active cyclotide, MCo-PG2, displayed good antimicrobial activity against different ESKAPE pathogen strains, including P. aeruginosa, S. aureus, K. pneumoniae, and E. coli (Table 1), in addition to 20 clinical strains of P. aeruginosa and S. aureus isolated from patients with cystic fibrosis (Tables 2 and S3). All the S. aureus clinical isolates were methicillin-resistant (MRSA), while around 30% of the P. aeruginosa were classified as multi-drug (MDR) strains, i.e. showing antimicrobial resistance to at least three or more antimicrobial agents from different groups of antibiotics. Cyclotide MCo-PG2 showed strong activity against these clinical strains with MIC₅₀ values of 1.5 μM against P. aeruginosa (n=20) and 6.25 μM against S. aureus (n=20) indicating its potential therapeutic value (Table 2). More importantly, MCo-PG2 (25 mg/kg, 4.5 μmol/kg; 10 mg/kg, 1.8 μmol/kg) provides a similar level of protection to that of PG-1 (5 mg/kg, 2.3 μmol/kg) and colistin (15 mg/mol, 12.3 μmol/kg) when used as single dose treatment in a murine P. aeruginosa-induced bacterial peritonitis model (Fig. 4). These results reveal that although cyclotide MCo-PG2 was in general less active than protegrin PG-1 in vitro displayed a similar level of activity to that of PG-1 in vivo. Cyclotide MCo-PG2 also exhibited 14 times less hemolytic activity than PG-1, while was only about three times less cytotoxic than PG-1 to human epithelial cells. In vivo toxicity studies also revealed that cyclotide MCo-PG2 was approximately 4 times less toxic than PG-1 in mice. These results are extremely encouraging, and open the possibility to improve even more the antimicrobial activity of cyclotide MCo-PG2 in future studies. Cyclotides contain multiple loops that are amenable to variation using different molecular evolution techniques.^[32] Hence, more active cyclotides could be produced by modifying adjacent loops to loop 6 in MCo-PG2, mainly loops 1, 3 and 5 (Fig. 1). It is also worth noting that cyclotide MCo-PG2 showed a remarkable resistance to biological degradation in serum, with a τ_1/2 value of ≈60 h and not showing any significant degradation for the first 24 h (Fig. S4). In contrast, protegrin PG-1 was significantly degraded (≈55% degradation) after the first 24 h under the same conditions, hence revealing the superior proteolytic stability of the circular cystine-knot topology of MCo-PG2 versus the disulfide-stabilized β-hairpin structure of PG-1.

Altogether, our results show that engineered cyclotides hold great promise for the development of a novel type of peptide-based broad spectrum antimicrobial agents able to efficiently target specific bacterial targets. Our results demonstrate for the first time the design of an engineered cyclotide able to show potent antimicrobial activity in vitro using physiological-like conditions and more importantly in vivo using a murine P. aeruginosa-induced peritonitis animal model, thereby providing a promising lead compound for the design of novel antibiotics.