Structure, toxicity and antibiotic activity of gramicidin S and derivatives

Development of new antibiotics is declining whereas antibiotic resistance is rising, heralding a post-antibiotic era. Antimicrobial peptides such as gramicidin S (GS), exclusively topically used due to its hemolytic side-effect, could still be interesting as therapeutic compounds. By modifying the amino-acid composition of GS, we synthesized GS analogues. We now show that derivative VK7 has a lower MIC (7.8–31.2 μg/ml, median 15.6 μg/ml) against strains of multi-drug resistant (MDR) Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa than GS has (3.9–62.5 μg/ml, median 31.3 μg/ml). Low MICs for both VK7 and GS were observed for Staphylococcus aureus and Enterococcus faecium. VK7 showed reduced haemolysis and less lactate dehydrogenase release. All compounds were fully bactericidal at MIC values. Modification of GS enables production of novel derivatives potentially useful for systemic treatment of human infections.


Introduction
Six bacterial pathogens with a propensity for developing multi-drug resistance (MDR) are specifically warned for by the Infectious Disease Society of America (IDSA) (ESKAPE: Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and extended spectrum beta lactamase (ESBL)producing Enterobacteriaceae). These species are causing the majority of human infections and efficiently acquire additional resistance traits [1], which implies that new antibiotics have to be effective against these actively evolving MDR pathogens. The incidence of vancomycin-resistant Enterococci (VREs) has increased dramatically over recent years [2]. S. aureus, especially methicillin-resistant S. aureus (MRSA), currently causes more deaths in the USA annually than HIV and tuberculosis combined [2]. ESBLs continue to be on the rise and limit treatment options [3]. P. aeruginosa is becoming increasingly resistant to multiple classes of drugs [4], whereas Acinetobacters are naturally resistant to many classes of antibiotics [5]. Increasing antibiotic resistance leads to extended hospitalization, rising treatment costs, and increased morbidity and mortality.
The rise of antibiotic-resistant pathogens has sparked research into currently disregarded antimicrobial peptides including gramicidin S (GS). GS is naturally produced by Aneurinibacillus migulanus [6] and was first discovered in 1941 [7]. GS shows antimicrobial activity against both Gram-positives and Gram-negatives in a MIC range from 4-64 μg/ml [8]. The lowest MICs are seen for Gram-positive bacterial species [9]. Despite its good antimicrobial activity, GS cannot be used systemically due to its haemolytic sideeffect [10] and is therefore only applied topically to treat superficial infections [11]. GS is a cyclic, C2-symmetrical decapeptide with the sequence cyclo(Pro-D Phe-Leu-Orn-Val) 2 . The two Pro-D Phe dipeptides form two type II β-turns, and the two Leu-Orn-Val stretches form an antiparallel βsheet. GS has been reported to kill bacteria by forming pores in the outer membranes [8]. Native GS is a natural scaffold for amino acid alteration in such a way that antimicrobial activity is retained but toxicity is reduced. In previous attempts to modify GS, several strategies have been followed: nonnatural amino-acids were included [12], the size of the ring has been modified [8], and the β-turn region [13] and β-strand region have been changed [14]. Still, few new derivatives of GS have been identified that show retention of antimicrobial activity with reduced toxicity [8,[15][16][17][18]. We here study the βstrand-modified GS analogue VK7 [14] and the β-turn modified derivative 20 [13]. Studying naturally occurring antimicrobial peptides such as GS could help with the design and development of novel derivative drugs to combat multidrug resistance.

Material and methods
ESKAPE panel collection and characterization Except for the S. aureus USA 300 and MRSA 252 strains [19], ESKAPE strains were collected at the Department of Medical Microbiology and Infectious Disease in the Erasmus Medical Centre, Rotterdam, The Netherlands. Thirty clinical isolates of E. faecium (5), S. aureus (5), K. pneumonia (5), A. baumannii (5), P. aeruginosa (5) and E. cloacae (5) were isolated from January 2010 to October 2011 from different wards ( Table 1). All strains were cultured on Columbia agar plates with 5 % sheep blood (Becton Dickinson, Breda, The Netherlands) overnight at 37°C before antibiotic susceptibility testing. Antibiotic resistance was determined using disk diffusion following Clinical and Laboratory Standards Institute (CLSI) guidelines and VITEK2 (bioMérieux, Zaltbommel, The Netherlands) following manufacturer's protocol; susceptibility was determined using EUCAST breakpoints [20].
MIC determination Antimicrobial activity of GS and derivatives ( Fig. 1) was determined following the CLSI guidelines and in triplicate [21]. Briefly, bacterial cells were cultured overnight on Columbia agar plates with 5 % s h e e p b l o o d ( B e c t o n D i c k i n s o n , B r e d a , The Netherlands). Colonies were suspended in 0.9 % NaCl to a density of 0.5 McFarland, then diluted 1:100 in Müller-Hinton Broth (MHB, Oxoid, Badhoevedorp, The Netherlands); 100 μl of this suspension was added to wells containing GS, derivatives 3, 20, or VK7 ranging from a concentration of 0.95 μg/ml to 62.5 μg/ml in MHB. Ninety-six well plates (Greiner Bio One, Alphen aan den Rijn, The Netherlands) were incubated for 18-24 hours at 37°C, and MIC values were determined visually.
To determine whether antimicrobial activity was bactericidal, 200 μl of the suspension was plated onto new Columbia agar plates with 5 % sheep blood, and colonies were counted. Haemolysis assay Haemolysis assays were performed as described before [13]. Freshly drawn heparinised blood from healthy volunteers was centrifuged for 10 min at 1000 g at 10°C. The pellet was washed three times with 0.9 % saline and diluted with saline to a 1/25 packed volume of red blood cells. Triton-X100 (1 %) was used as a positive control. GS or derivative 3, 20, and VK7 were diluted in 100 μl PBS in Ubottom 96-well plates (Greiner Bio One). Serial dilution resulted in a concentrations ranging from 62.5 μg/ml to 0.95 μg/ml. DMSO (Sigma-Aldrich, Zwijndrecht, The Netherlands) was used as a solvent control. Subsequently, 50 μl of the red blood cell suspension was added to the wells, and the plates were incubated at 37°C for 4 h. After incubation, the plates were centrifuged at 1, 000 g at 10°C for 4 min, and 50 μl of the supernatant of each well was dispensed into new flat-bottom 96-well plates (Greiner Bio One), and absorbance was measured at 415 nm in a Bio-Rad 680 spectrophotometer (Bio-Rad, Veenendaal, The Netherlands). OD values were plotted as a percentage of the positive control. Experiments were performed in triplicate. . Plates were centrifuged for 10 min at 1,000 g, and the amount of LDH in the supernatants was determined (LDH release kit, Roche, Woerden, The Netherlands) following the protocol. Cytotoxicity testing was independently repeated three times.

Therapeutic indices
The therapeutic index was defined as a measure of toxicity, either the 50 % haemolysis or the 50 % LDH release, divided by the lowest and highest MIC values seen for each strain tested. Therapeutic indices are given as a range to be compared between GS and its GS derivatives to determine improved performance.
The β-turn variant 20 showed slightly reduced activity against S. aureus and E. faecium as compared to GS. MIC values for compound 20 were in the range of 7.8-62.5 μg/ml (median 7.8 μg/ml). Derivative 20 showed activity against E. cloacae, P. aeruginosa, K. pneumoniae, and A. baumannii in the range of 1.95-62.5 μg/ml (median 62.5 μg/ml), which is slightly less than measured for the parental compound.
Derivative 3 showed activity against S. aureus and E. faecium in the range from 7.8-62.5 μg/ml (median 7.8 μg/ml), which is slightly less than parental GS. Derivative 3 shows activity against the MDR Gram-negative strains E. cloacae, P. aeruginosa, K. pneumoniae, and A. baumannii in the range from 3.9-62.5 μg/ml (median 31.3 μg/ml), which is comparable to the parental compound.
All compounds tested were bactericidal, as sub-culturing of medium from wells without visible growth on agar media did not result in detectable growth.  (Fig. 3a). As the MIC values of GS varied from 3.9 to 62.5 μg/ml, the TI hae for GS was calculated to be between 0.56 and 18.5. Derivative 3 showed 50 % haemolysis at 41.6 μg/ml (Fig. 3a), which is in a similar range as for the parental compound. As derivative 3 had MIC values varying from 3.9 and 62.5 μg/ml, the TI hae of derivative 3 was calculated to be 0.6 to 10.6, also comparable to the parental compound. This shows that derivative 3 is not an improved antibiotic in comparison with GS. VK7 and compound 20 did not show haemolysis at 62.5 μg/ml, which was the highest concentration tested. Hence, the exact TI hae for compound 20 and VK7 could not be determined, but still, these derivatives are clearly less haemolytic than GS (Fig. 2).

Haemolysis assay and therapeutic indices Haemolysis is clearly concentration-dependant. Canonical GS showed
LDH release assay and therapeutic indices The values of 50 % LDH release were 18.7 μg/ml for GS (Fig. 3b). As the Fig. 3 Erythrocyte lysis and LDH release under the influence of GS and the derivatives MIC values were between 3.9 and 62.5 μg/ml, TI LDH was calculated to be between 0.3 and 9.6. Derivative 3 showed 50 % LDH release at 49.8 μg/ml. As the MIC values were between 3.9 and 62.5 μg/ml, TI LDH was calculated to be between 0.8 and 12.8, in the same range as documented for GS. Derivate 20 showed 50 % LDH release at 62.5 μg/ml. As MIC values were between 1.95 and 62.5 μg/ml, TI LDH was calculated to be between 1 and 32.1 μg/ml, which shows slight improvement compared to GS. VK7 did not reach 50 % LDH release at concentrations tested here. Therefore, no TI LDH could be calculated, indicating again that VK7 is less cytotoxic than GS (Table 1).

Discussion
Beta-strand modification of GS seemed to be promising for the development of new systemically applicable antibiotics. Derivative VK7 showed activity against E. cloacae which was equal to that of GS. Against all P. aeruginosa and K. pneumoniae and most A. baumannii there was a 2-8-fold increase in activity. The antimicrobial activity of VK7 against Gram-positive MDR pathogens such as S. aureus and E. faecium was similar to that of GS. In addition, we observed reduced toxicity for VK7 towards human erythrocytes and the human colorectal adenocarcinoma cell-line HT-29. The βstrand-modified VK7 has the same overall secondary structure as GS, but probably displays an elevated cationic character counterbalanced by two robust hydrophobic adamantane groups. The data indicate that β-strand modification of GS can generate interesting new antibiotics combining antimicrobial activity and lowered toxicity.
The β-turn-modified derivative 20 showed reduced toxicity compared to the parental compound, with a slight decrease in antibiotic activity compared to GS, especially when used against Gram-negatives. This β-turn-modified derivative encompasses a substituted sugar amino acid (SAA) dipeptide isoster as turn mimetic. The six-ring SAA in our lead was found to have better conformational and hydrophobic characteristics than a 4-ring (oxetane) and 5-ring (furanoid) SAA [9]. Beta-turn modification could still be promising, as at least some reduction of toxic potential is observed. Not all modifications of GS have a positive effect on antimicrobial activity and toxicity: derivative 3 shows comparable antimicrobial activity to the parental compound but a similar toxicity profile.
The synthesis of modified GS derivatives has been reported by other groups [8,12,15,17]. One of the key factors important in the balance between cytotoxicity and antimicrobial activity is overall hydrophobicity-hydrophilicity. Derivatives that are slightly less hydrophobic than the parental GS generally show good antimicrobial activity, while showing reduced haemolysis. Using solid-and liquid-phase organic synthesis, derivatives of GS can be obtained containing non-naturally occurring amino acids, which show reduced cytotoxicity and reduced haemolysis, while retaining antimicrobial activity.
Antimicrobial peptides may have a bright future in combating infection, as they generally do not have a single conserved target, but affect multiple bacterial processes. Modifications of the β-strand of GS in which the hydrophobic side chains have been varied are promising leads for the development of novel compounds. New derivatives of GS can possibly address the growing problem of multi-drug resistant bacteria and lead to new therapeutic compounds for systemic use, as is suggested on the basis of our current data.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.