Responses of pseudomonas aeruginosa and other eskape pathogens to antimicrobial peptide dendrimers

(1)

Thesis

Reference

Responses of pseudomonas aeruginosa and other eskape pathogens to antimicrobial peptide dendrimers

BEN JEDDOU, Fatma

Abstract

Antimicrobial peptides (AMPs) are one of the most diverse classes of antibiotics. A prominent example is polymyxin E (colistin), used as last-resort antibiotic against multidrug resistant (MDR) Gram-negative bacteria. However, the emergence of resistance and toxicity of colistin restrict its use. Dendrimer antimicrobial peptides are currently explored as an alternative to colistin, since they show low hemolytic activity and are active against MDR isolates of Pseudomonas aeruginosa and Acinetobacter baumannii. To unravel the mode of action and potential resistance mechanisms of dendrimer AMPs, spontaneous resistant mutants were selected on polymyxin B and the prototypical dendrimer G3KL in ESKAPE pathogens. We propose a model for decreased susceptibility to G3KL in P. aeruginosa that includes two modifications of the lipopolysaccharide resulting in decreased negative charges at the bacterial surface. We hypothesized that mechanisms leading to decreased susceptibility to G3KL may be different from those conferring polymyxin resistance and seem to be species-specific in Gram-negative bacteria.

BEN JEDDOU, Fatma. Responses of pseudomonas aeruginosa and other eskape

pathogens to antimicrobial peptide dendrimers. Thèse de doctorat : Univ. Genève, 2021, no. Sc. Vie - Bioméd. 102

DOI : 10.13097/archive-ouverte/unige:152699 URN : urn:nbn:ch:unige-1526999

Available at:

http://archive-ouverte.unige.ch/unige:152699

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

UNIVERSITÉ DE GENÈVE FACULTÉ DES SCIENCES Section de médecine fondamentale

Département de microbiologie et médecine moléculaire Professeur Christian van Delden Dr. Thilo Köhler

RESPONSES OF PSEUDOMONAS AERUGINOSA AND OTHER ESKAPE PATHOGENS TO ANTIMICROBIAL PEPTIDE DENDRIMERS

THÈSE

présentée aux Facultés de médecine et des sciences de l’Université de Genève pour obtenir le grade de Docteur ès sciences en sciences de la vie,

mention Sciences biomédicales

par

Fatma BEN JEDDOU

de

Sonceboz-Sombeval (Berne)

Thèse N^o 102

GENÈVE 2021

(3)

2

Acknowledgements

This thesis project would have not been possible without the support, advice and help of several persons. I would like to express my sincere gratitude to these people.

First of all, I would like to thank my supervisors Prof. Christian van Delden and Dr. Thilo Köhler for giving me the opportunity to learn and perform research in their laboratory. Thank you for your patience and for providing guidance, motivation and expert knowledge. A million thanks for providing me the greatest support and help during my PhD. I couldn’t imagine having better mentors, you are an inspiration for me. I also thank the group of Prof. Jean-Louis Reymond for their collaboration in this project.

I would like to thank my committee members, Dr. Laurent Poirel and Prof. Karl-Heinz Krause for accepting to be part of the jury and read my PhD thesis. Your suggestions will be of great help to potentially pursue this research.

I thank the department members and my lab members, who made these four years a great journey. I met nice persons who became valuable friends. Thank you for the stimulating discussions, advice and all the fun we had together. Especially Matthieu, Soner, Marie, Gaël, Simon, Fabien, Sunil, Inès, Ronke, Hugo, Caroline, Emilie and Yves. Special thanks to Ezgi, who shared so much with me, and to Jordan for the support and brainstorming sessions. Thank you Alex for your help, sharing your microbiological experience and nice discussions in the P2, Léna for all the moments we shared and always being there for me, Bartosz for being supportive and comforting me.

I would like to thank my second family in Lausanne for the awesome time together and the huge support, especially my flatmates, Samba and Matthias. Loïc, Jérôme, Gaël and Bianca thank you very much for being supportive.

Last but not least, I would like to thank my parents, my brothers and my sisters for their support all along my PhD and my life in general. For this, I will be forever grateful.

(7)

6

Summary

The emergence and spread of antibiotic resistant bacteria have become a priority issue for global health. In particular, the group of ESKAPE pathogens (E. faecium, S.

aureus, K. pneumoniae, A. baumannnii, P. aeruginosa and Enterobacter spp) are considered as a major threat, since they frequently develop multidrug resistant (MDR) phenotypes and

“escape” antimicrobial therapy. An additional concern is the emergence and spread of resistance determinants towards colistin (polymyxin E), a last resort drug used to treat infections by MDR Gram-negative bacteria. These problems emphasize the urgent need of developing novel antimicrobial peptides (AMPs). Among explored synthetic peptides, dendrimer AMPs display promising pharmacokinetic properties. One of the candidate dendrimers, G3KL presents potent activity against multidrug resistant Gram-negative pathogens such as P. aeruginosa, A. baumannii and Enterobacteriaceae harboring extended- spectrum β-lactamases (ESBLs).

This thesis aimed at studying the mode of action and potential mechanisms of decreased susceptibility to the dendrimer G3KL and related AMPs in Gram-negative bacteria. Therefore, we used genetic, biochemical and cellular approaches mainly focusing on P. aeruginosa and later extended the investigations to E. coli and two other important pathogens of the ESKAPE group: A. baumannii and K. pneumoniae.

In the first part of the project, we selected spontaneous mutants on polymyxin B (Pmx-B) and the prototypical dendrimer G3KL in P. aeruginosa. Mutations occurred in sensor kinases of two component regulatory systems (TCSs) leading to lipopolysaccharide (LPS) modifications.

We analyzed the effects of these mutations and showed that two LPS modifications, namely the addition of amino-arabinose to the lipid A and the possible secretion of polyamines, are required to obtain decreased susceptibility to G3KL. These modifications resulted from increased expression of the arnBCADTEF operon as well as the speD2-speE2-PA4775 operon and led to a decrease of negative charges at the bacterial surface thereby preventing interactions with this dendrimer peptide. We further investigated the responses of P.

aeruginosa to two dendrimers and Pmx-B by transcriptome analysis, which suggested that G3KL signals through the CprRS TCS, while Pmx-B triggers the ParRS TCS. This allowed us to establish a model for the response of P. aeruginosa to dendrimer AMPs.

(8)

7

The second part of my PhD thesis project was focused on the regulation of arn and speD2 operons as their expression results in modifications of the LPS involved in the mechanisms of decreased susceptibility to G3KL in P. aeruginosa. We studied the involvement of the CprRS and ParRS TCSs in the regulation of these operons as well as in the susceptibility to G3KL and other AMPs in P. aeruginosa. We analyzed the expression kinetics of these operons using promoter lux-fusions and generated deletions in these TCSs in various strain backgrounds. Our data suggested that, in addition of being upregulated by the PmrAB TCS, which regulates as well the arn operon, the expression of the speD2 operon is also regulated through the CprRS TCS, which recognizes G3KL. We further hypothesized that the contribution of the ParRS TCS occurs through the potential recognition of polyamines that are secreted or present in the medium, leading to the downregulation of the speD2 operon. These observations led us to propose a model for the regulation of arn and speD2 operons and for the role of the CprRS and ParRS TCSs in response to G3KL and Pmx-B.

Finally, in the last part, we determined whether the previously described mechanisms of decreased susceptibility to G3KL were specific to P. aeruginosa or exist also in other important Gram-negative pathogens such as E. coli, A. baumannii and K. pneumoniae. Therefore, we selected spontaneous mutants leading to decreased susceptibility to G3KL in these organisms.

Whole genome sequencing of these mutants identified mutations in asmA, lptC and msbA genes, which are involved in the transport of the LPS at the surface of the outer membrane.

These genes were not previously described in polymyxin resistance in these bacteria and are different from those involved in decreased susceptibility to G3KL in P. aeruginosa. Our data further suggested that the acylation of lipid A could be a potential mechanism for decreased susceptibility to G3KL in E. coli.

In summary, we propose a model for decreased susceptibility to G3KL in P. aeruginosa that includes two modifications of the LPS resulting in a decrease of negative charges at the bacterial surface. We showed that the regulation of these modifications involves the two- component systems PmrAB, CprRS and ParRS. We hypothesized that mechanisms leading to decreased susceptibility to G3KL may be differen from those conferring polymyxin resistance and seem to be species-specific in Gram-negative bacteria.

(9)

8

Résumé

L’émergence et la propagation de souches bactériennes résistantes aux antibiotiques sont devenues une problématique prioritaire à l’échelle mondiale pour la santé publique. En particulier les pathogènes du groupe ESKAPE (E. faecium, S. aureus, K. pneumoniae, A.

baumannnii, P. aeruginosa and Enterobacter spp) sont considérés comme une menace puisque ces bactéries développent fréquemment des phénotypes de multi-résistance et

« échappent » ainsi aux traitements antibiotiques. Une autre inquiétude concerne l’émergence et la dissémination de mécanismes de résistance à la colistine (polymyxin E), un médicament de dernier recours contre les infections provoquées par des souches à Gram- négatif multi-résistantes. Ces problèmes soulignent le besoin urgent de développer de nouveaux peptides antimicrobiens (PAM). Parmi les peptides synthétiques, les PAM dendrimériques présentent des propriétés pharmacocinétiques intéressantes. Un des candidats dendrimers, G3KL, présente une bonne activité contre les pathogènes multi- résistants à Gram-négatif tels que P. aeruginosa, A. baumannii et les Enterobacteriaceae productrices de béta-lactamases à spectre étendu (BLSEs).

Cette thèse avait pour but d’étudier le mode d’action et les possibles mécanismes de diminution de susceptibilité au dendrimer G3KL et autres PAMs associés dans les bactéries à Gram négatif. Nous avons utilisé des approches génétiques, biochimiques et cellulaires en se concentrant principalement sur P. aeruginosa. Par la suite, nous avons étendu l’étude de G3KL à E. coli et deux autres pathogènes importants du groupe ESKAPE : A. baumannii et K.

pneumoniae.

Dans la première partie du projet, nous avons sélectionné des mutants spontanés chez P.

aeruginosa en utilisant la polymyxin B (Pmx-B) et le dendrimer G3KL. Des mutations ont eu lieu dans les senseurs des systèmes à deux composants (TCS) causant des modifications du lipopolysaccharide (LPS). Nous avons analysé les effets de ces mutations et démontré que deux modifications du LPS, à savoir l’addition d’amino-arabinose au lipide A et la possible sécrétion de polyamines, sont requises pour parvenir à une diminution de susceptibilité à G3KL. Ces modifications résultent de l’augmentation de l’expression de l’opéron arnBCADTEF ainsi que de l’opéron speD2-speE2-PA4775. Elles entraînent une diminution des charges négatives présentes à la surface bactérienne et, par conséquent, empêchent les interactions avec le dendrimer. En outre, nous avons aussi étudié la réponse de P. aeruginosa à l’encontre

(10)

9

de deux dendrimers et la Pmx-B par une analyse du transcriptome. Cette analyse a montré que G3KL est reconnu par le le TCS CprRS alors que la Pmx-B est reconnue par le TCS ParRS.

Cela nous a permis d’établir un modèle pour la réponse de P. aeruginosa à l’encontre des PAMs dendrimériques.

La seconde partie de mon projet de thèse était centrée sur la régulation des opérons arn et speD2. L’expression de ces deux opérons est nécessaire pour les modifications du LPS impliquées dans les mécanismes de diminution de susceptibilité à G3KL chez P. aeruginosa.

Nous avons étudié la contribution des TCSs CprRS et ParRS dans la régulation de ces opérons ainsi que dans la susceptibilité à G3KL et autres PAMs. Nous avons analysé les cinétiques d’expression de ces opérons au moyen de fusions lux-promotrices et en générant des mutants de deletion dans ces TCS par recombinaison homologue. Nos données suggèrent que l’expression de l’opéron speD2 est régulée par le TCS CprRS qui reconnaît G3KL, en plus d’être régulée par le TCS PmrAB qui régule également l’opéron arn. De plus, nous avons montré l’implication du TCS ParRS dans cette régulation par le biais d’une potentielle reconnaissance des polyamines sécrétées ou présentes dans le milieu qui engendre une diminution de l’expression de l’opéron speD2. Ces observations nous ont conduites à proposer un modèle pour la régulation des opérons arn et speD2 et pour le rôle des TCSs CprRS et ParRS en réponse à G3KL et à la Pmx-B chez P. aeruginosa.

Finalement, dans la troisième partie, nous avons déterminé si les mécanismes responsables d’une diminution de susceptibilité à G3KL chez P. aeruginosa sont spécifique à cette bactérie ou sont impliqués également chez d’autres pathogènes à Gram négatif tels que E. coli, A.

baumannii et K. pneumoniae. Nous avons donc sélectionné des mutants spontanés présentant une diminution de susceptibilité à G3KL chez ces organismes. Le séquençage du génome de ces mutants a identifié des mutations dans les gènes asmA, lptC et msbA impliqué dans le transport des LPS vers la surface de la membrane externe. Ces gènes n’ont, auparavant, pas été décrit comme contribuant à la résistance à la polymyxine dans ces bactéries et sont différents de ceux impliqués dans la diminution de susceptibilité à G3KL chez P. aeruginosa.

De plus, nos données ont suggéré l’acylation du lipide A comme étant un possible mécanisme de diminution de susceptibilité à G3KL chez E. coli.

En résumé, nous avons proposé un modèle de diminution de susceptibilité à G3KL dans P.

aeruginosa intégrant deux modifications du LPS qui résultent en la diminution des charges négatives à la surface bactérienne. Nous avons démontré que la régulation de ces

(11)

10

modifications inclut les TCSs PmrAB, CprRS et ParRS chez P. aeruginosa. Nous proposons que des mécanismes différents de diminution de susceptibilité existent entre la polymyxin et G3KL et que ces mécanismes sont spécifiques aux espèces bactériennes à Gram négatif étudiées ici.

(12)

11

Introduction

1.1 Short history of discovery

The introduction of antibiotics has been the most important step in the fight against bacterial infections. In 1907, Ehrlich and Bertheim synthesized the first antimicrobial agents known as the sulpha drugs or sulfonamides. These molecules are made of a sulfonamide group and affect the folate biosynthesis pathway by inhibiting the dihydropteroate synthetase (DHPS) enzyme (Williams 2009). Nevertheless, it is considered that Domagk discovered the first sulfonamide “Prontosil” that was used in clinical settings and earned him a Nobel Prize in 1939 (Otten 1986). In 1928, sir Alexander Fleming discovered penicillin (Fleming 2001) and its active compound was further purified by Florey and Chain (Ligon 2004). This discovery earned them the Nobel Prize in 1945 and is admitted as the discovery of the first authentic antibiotic when referring to the definition of Waksman: an antibiotic is a molecule synthesized or derived from microorganisms that kills other microorganisms or inhibits their growth (Waksman 1947). The introduction of penicillin enabled the treatment of disease such as streptococcal infections that were until then untreatable. The progressive discovery of different classes of antibiotics occurred during the “golden era of antibiotics” between 1940-1960’s (Fig. 1). Considerable research and industrialization followed these discoveries.

Figure 1. Timeline of antibiotics development and resistance appearance. Top panel illustrates the introduction of main antibiotics classes and bottom panel illustrates when resistance to these classes of antibiotics was observed. Adapted from (Clatworthy, Pierson and Hung 2007).

(13)

12 1.2 Spectrum of activity

Based on their spectrum of activity, antibiotics can be classified in two groups: broad spectrum and narrow spectrum. Broad spectrum antibiotics are useful to treat both Gram-positive and Gram-negative bacteria. Narrow spectrum antibiotics are used when a single pathogen is responsible for an infection and decrease the side effects on the bacterial flora of the patient (Acar 1997, Willing, Russell and Finlay 2011, Blaser 2011). One such example can be found among the cephalosporin antibiotics. Ceftazidime is a 3^rd generation cephalosporin with broad spectrum, whereas Cefuroxime, which is of 2^nd generation, has a narrower spectrum (O'Callaghan 1986, Perry and Brogden 1996). The spectrum of activity relies on structural membrane differences between Gram-negative and Gram-positive bacteria (Fig. 2), as well as the presence of intrinsic resistance mechanisms.

Even if most antibiotics cross the cell membrane through passive transport using porins or other transporters, Gram-negative bacteria remain more challenging to kill as they have an outer membrane (Tommasi et al. 2015). The Gram-negative bacteria are characterized by both an inner and outer membrane composed of a bilayer of phospholipids, and a thin peptidoglycan (cell wall) linked to the membrane through lipoproteins in the periplasmic space. The outer membrane of Gram-negative bacteria also contains lipopolysaccharides (LPS). On the other hand, Gram-positive bacteria only have an inner membrane and a thick peptidoglycan, containing lipoteichoic acids and teichoic acids. Gram-negative and Gram- positive bacteria have common structures that include flagella and for some bacteria a capsule, which is made of glycoproteins or proteins that can provide an additional layer of protection (Cabeen and Jacobs-Wagner 2005). Both Gram-positive and Gram-negative organisms also express different drug efflux systems reducing the intracellular concentration of antibiotics (Poole 2005).

(14)

13

Figure 2. Gram-negative and Gram-positive bacterial membranes. Adapted from (Brown et al. 2015).

Antibiotics can be either bactericidal or bacteriostatic. A bactericidal antibiotic kills bacteria whereas a bacteriostatic antibiotic inhibits the bacterial growth and the infection needs to be cleared by the host’s immune system. It should be noted that these definitions apply at the level of the organism and under in vitro conditions, but in a clinical context they become arbitrary (Pankey and Sabath 2004). Some antibiotics can exert a bactericidal effect against some bacteria and a bacteriostatic effect against others such as chloramphenicol, which has been shown in vitro to be bactericidal against S. pneumoniae and bacteriostatic against S.

aureus (Rahal and Simberkoff 1979).

1.3 Antibiotic classes and targets

Antibiotics can be classified according to their mode of action. Only few bacterial pathways can be targeted due to the fact that the antibiotics should not affect the eukaryotic cells and consequently, only pathways that are specific to the bacteria can be targeted. Based on their targets antibiotic classes include 1) Inhibitors of transcription, 2) Inhibitors of DNA replication, 3) Inhibitors of protein synthesis, 4) Inhibitors of the folic acid synthesis, 5) Inhibitors of the cell wall synthesis (Fig. 3). Most of these molecules were extracted and purified from natural producers (bacteria of fungi), only few are entirely synthetic. Usually new molecules are developed based on the structure of already described antibiotics, but with some modifications or an increased affinity for the same targets (Nathan 2004, Fischbach and Walsh 2009). One such example of this development are the cephalosporins, a class of b-lactams.

(15)

14

They are separated into 1^st, 2^nd, 3^rd, 4^th and 5^th generation according to changes of their structure and the spectrum of activity (Fischbach and Walsh 2009, Bassetti et al. 2013).

The rifamycins, like Rifampicin, target the transcription. They bind to the DNA-dependent RNA polymerase holoenzyme before the DNA unwinds to keep it closed and inhibit RNA synthesis (Hartmann et al. 1967, Naryshkina et al. 2001).

The quinolones and fluoroquinolones, which are derived from quinolones by addition of a fluorine atom like ciprofloxacin, inhibit the DNA gyrase (topoisomerase II) and DNA topoisomerase IV. These enzymes are involved in DNA topology changes, a step required during DNA synthesis. After cleavage and before re-ligation, these enzymes introduce positive (topoisomerase IV) and negative (topoisomerase II) supercoils in the DNA. Fluoroquinolones bind to topoisomerase II and block the DNA re-ligation (Redgrave et al. 2014).

Ribosomal activity is required in translation and involves the 70S complex, which is composed of the two subunits 30S and 50S. Some antibiotics, like aminoglycosides and tetracyclines, inhibit the 30S subunit. Aminoglycosides bind to the 16S rRNA and alter the mRNA-tRNA complex or inhibit the translocation of peptidyl-tRNA. This leads to incomplete translation termination (Davis 1987). Binding to the 16S rRNA of the 30S subunit by tetracyclines prevents the tRNA binding to the A site (Chopra and Roberts 2001). Other antibiotics, like macrolides and chloramphenicol, inhibit the 50S unit by binding nearby the peptidyl transferase center.

Premature protein biosynthesis termination occurs through the dissociation of the ribosome and the peptidyl tRNA (Egorov, Ulyashova and Rubtsova 2018, Menninger and Otto 1982).

The synthesis of folic acid is an essential pathway as organisms require the folate cofactor, which is involved in nucleic acids and amino acids biosynthesis. The dihydropteroic acid is synthesized from para-amino-benzoic acid (PABA) and pteridine by the dihydropteroate synthetase (DHPS) enzyme. Dihydrofolate synthetase then synthesizes dihydrofolic acid from dihydropteroic acid. Dihydrofolic acid is used by dihydrofolate reductase (DHFR) to synthesize tetrahydrofolic acid (Bermingham and Derrick 2002). This pathway is inhibited by the sulfa drugs such as sulfamethoxazole, which inhibits the DHPS enzyme, and trimethoprim, which inhibits the DHFR enzyme (Masters et al. 2003). These molecules also target the eukaryotic enzymes, however the concentrations required to inhibit these enzymes are higher than those used against bacteria leaving a therapeutic window for sulfonamides.

The cell wall synthesis inhibitors include β-lactams, such as carbapenem, penicillin, monobactam and cephalosporin that target the penicillin binding proteins (PBPs) (Holten and

(16)

15

Onusko 2000). In the synthesis of peptidoglycan (PG), the glycosyltransferases (GTases) polymerize the glycan chains, whereas the DD-transpeptidases (DD-TPases), which are also called PBPs are involved in the cross-linking of the peptides. The β-lactams covalently bind to the active site of PBPs and inhibit the synthesis of PG (Typas et al. 2011). In bacteria, various PBPs are involved in the cell cycle depending on which step they act and the localization (Randich and Brun 2015). Due to strong affinity, the β-lactams are most of the time linked with specific PBPs.

Figure 3. Antibiotic targets. (Madigan, Madigan and Brock 2009).

1.4 Antibiotics resistance

The use of antibiotics is an essential part of modern medicine that has improved health conditions and life expectation (Piddock 2012). It is also of great interest in agriculture and animal production to prevent and treat bacterial diseases (Gustafson and Bowen 1997).

Nevertheless, the extensive use of antibiotics has promoted the selection and proliferation of resistant bacteria (2017). Antibiotic resistance leads to approximately 700’000 deaths per year and is estimated to cause up to ten million deaths per year by 2050 (Blair et al. 2015).

Antibiotic resistance has become a major concern and there is an urgent need to develop novel antibiotics. The emergence of antibiotic resistance is a natural event first observed in

(17)

16

the 1940s and which has spread worldwide. For example, the β-lactamase enzymes, which deactivate the β-lactam antibiotics by hydrolysis, have been described before the introduction of penicillin in 1942 (Spellberg et al. 2011). Historically, the appearance of resistance to a particular antibiotic could be observed shortly after it was introduced (Fig. 1). Antibiotic resistance is a phenomenon that is encoded genetically and can be divided in three categories:

intrinsic, acquired and adaptive resistance. Intrinsic resistance includes all the inherent characteristics of a microorganism that can inhibit the action of an antibiotic. The outer membrane with reduced permeability as in Pseudomonas aeruginosa and Acinetobacter baumannii, or the efflux pumps encoded by bacteria to eject antibiotic from the cell, are examples of intrinsic resistance mechanisms (Lewis 2013). In acquired resistance, the bacteria, which are originally susceptible, become resistant through the incorporation of genetic material (integrons, transposons, plasmids, etc.) or mutations. In adaptive resistance, antibiotic susceptible bacteria respond to the presence of an antibiotic, or a stress condition, by modifying transiently their gene expression leading to target modifications or efflux pump overexpression (Fernández and Hancock 2012).

2. Antimicrobial peptides against Gram-positive bacteria

Since the emergence of antibiotic resistance, the search for alternate antimicrobial drugs has become of high priority. Bacteria that are not susceptible to at least one antimicrobial in more than two antibiotic classes are considered as multi-drug resistant (MDR) strains (Magiorakos et al. 2012). Among Gram-positive bacteria, Staphylococcus aureus is one of the most prevalent MDR strains that causes both community- and hospital-acquired infections (Tong et al. 2015). Methicillin-resistant S. aureus (MRSA) are a great concern as they are difficult to treat (Turner et al. 2019). Enterococcus species such as Enterococcus faecium and Enteroccocus faecalis are also another concern as they cause various infections and are resistant to a broad range of antibiotics (Jabbari Shiadeh et al. 2019). Among enterococcal infections, 85-90% are due to E. faecalis and 5-10% to E. faecium. Vancomycin-resistant Enterococci (VRE) can be found among both species (Ho et al. 2013). Streptococcus pneumoniae is another Gram-positive pathogen causing community-acquired infections such as pneumonia and meningitis. Some strains of S. pneumoniae have been described to be resistant to penicillin (Thummeepak et al. 2015). Antimicrobial peptides (AMPs) were suggested as promising candidates because of their potentiality to fight multi-drug resistant

(18)

17

bacteria (Seyfi et al. 2020). In this section, antimicrobial peptides available against Gram- positive bacteria are reviewed.

2.1 Structure and mode of action 2.1.1 Glycopeptides

Vancomycin is the first glycopeptide antibiotic isolated from Amycolatopsis orientalis in 1955 (McCormick et al. 1955). It consists of a heptapeptide core that is common to all glycopeptides (Fig. 4). Vancomycin binds to the terminal D-Alanine-D-Alanine (D-Ala-D-Ala) of peptidoglycan precursors through five hydrogen bonds, and inhibits transglycosylation and transpeptidation preventing thereby cell wall synthesis (Fig. 5). Vancomycin is used to treat Gram-positive bacterial infections and especially MRSA infections (Courvalin 2006). It displays a good activity against staphylococci, enterococci and streptococci with minimum inhibitory concentration (MIC) of 1 µg/mL, 2 µg/mL and 0.25 µg/mL, respectively (Table 1).

Figure 4. Structures and characteristics of AMPs against Gram-positive bacteria. Structures of glycopeptides, semi-synthetic glycopeptides and antimicrobial peptides are represented with their molecular weight (M.W.), half-life (T1/2) and mode of action (MoA) as well as the number of positive charges in red. Adapted from (Chen and Lu 2020).

(19)

18

Teicoplanin is a glycopeptide similar to vancomycin isolated from Actinoplanes teichomyceticus with a molecular weight varying between 1564 and 1907 g/mol, and carrying one positive charge (Fig. 4). This molecule is made of a mix of several components, five major (teicoplanin A2-1 to A2-5) and four minor (teicoplanin RS-1 to RS-4), and they all possess the same glycopeptide core but the length and conformation of the side chains are different (Barna et al. 1984). Teicoplanin differs from vancomycin by the presence of a hydrophobic acyl side chain (Nicolaou et al. 1999). As for vancomycin, the mode of action of teicoplanin involves inhibition of cell wall synthesis (Khamesipour et al. 2015).

Figure 5. Mode of action of glycopeptides. The binding of the glycopeptide to terminal D-Ala residues of peptidoglycan precursors (A) prevents their interaction with cell wall cross-linking enzymes (B) and results in inhibition of cell wall synthesis.

2.1.2 Semi-synthetic glycopeptides

Telavancin is a semi-synthetic lipoglycopeptide derived from vancomycin. It differs from vancomycin by the presence of a hydrophilic group and a hydrophobic side chain on the vancosamine sugar (Fig. 4). Like vancomycin, telavancin inhibits cell wall synthesis by binding to the terminal D-Ala-D-Ala of peptidoglycan precursors (Attwood and LaPlante 2007). The presence of the hydrophilic group confers to telavancin a second mechanism of action. It can interact with lipid II and disrupt the membrane by causing depolarization (Higgins et al. 2005).

Dalbavancin is a second generation semi-synthetic lipoglycopeptide and a derivative of the A- 40926 antibiotic, which is isolated from Nonmuria spp. and has a similar structure to

(20)

19

teicoplanin (Bailey and Summers 2008, Zhanel et al. 2008) (Fig. 4). The modifications made to the structure by addition of a lipophilic side chain that is not present in teicoplanin increase its half-life and activity against S. aureus. The dalbavancin mode of action also leads to inhibition of cell wall synthesis. The lipophilic side chain increases the affinity of binding to the terminal D-Ala-D-Ala of peptidoglycan precursors (Zhanel et al. 2008, Kim, Kuti and Nicolau 2007).

Oritavancin is derived from chloroeremomycin, which is similar to vancomycin and produced by Amicolatopsis orientalis (Cornaglia and Rossolini 2009). Like vancomycin, it is made of a heptapeptide core but has in addition a lipophilic side chain that provides a longer half-life (Fig. 4). As other glycopeptides, oritavancin binds to the terminal D-Ala-D-Ala of peptidoglycan precursors and inhibits cell wall synthesis. Furthermore, the hydrophobic side chain of oritavancin allows disruption of the membrane (Allen and Nicas 2003).

Table 1. Antimicrobial activity against Gram-positive bacteria.

MIC (mg/L)

Vancomycin Teicoplanin Dalbavancin Oritavancin Daptomycin Gramicidin

S. aureus 1 0.5 0.06 0.12 0.5 4

E. faecium VSE 2 0.5 0.06 0.008 2 8

VRE 256 >16 8 0.03 4 8

E. faecalis VSE 2 <0.12 0.06 0.015 1 8

VRE 256 >16 4 0.25 1 8

S. pneumoniae 0.25 <0.007 <0.015 0.008 0.25 3

VSE, vancomycin-susceptible enterococci. VRE, vancomycin-resistant enterococci.

2.1.3 Antimicrobial peptides

Daptomycin is a cyclic lipopeptide of 13 amino acids introduced in 2003 and isolated from Streptomyces roseosporus (Debono et al. 1987) (Fig. 4). The mode of action of daptomycin involves the formation of a complex with Ca²⁺ that inserts into the cell membrane of Gram- positive bacteria. Aggregation of daptomycin results in the depolarization of the membrane and subsequently the loss of cell content (Tran, Munita and Arias 2015) (Fig. 6).

(21)

20

Figure 6. Mode of action of daptomycin. Daptomycin get inserted into the cell membrane in a Ca²⁺

dependent manner and results in loss of intracellular components through membrane depolarization.

Adapted from (Bionda, Pitteloud and Cudic 2013).

Gramicidin was first isolated from the soil bacteria Bacillus brevis although its function in these bacteria is unknow and has a structure made of L- and D-amino acids (Sarges and Witkop 1965) (Fig. 4). Two types can be found, the linear pentadecapeptide Gramicidin D and the Gramicidin S, which is a cyclic decapeptide. The activity of Gramicidin D occurs by interaction with the bacterial membrane through monovalent cations resulting in increased permeability.

Gramicidin S acts by disrupting the inner content of the cell (Conti et al. 1997). Even though Gramicidin is among the most effective peptides, its use as an antibiotic peptide is restricted due to a high hemolytic activity (Mogi and Kita 2009).

2.2 Mechanism of resistance

Resistance to vancomycin involves modification of the peptidoglycan synthesis pathway by replacing the terminal D-Ala-D-Ala of peptidoglycan precursors by D-Ala-D-Lac, which prevents binding of vancomycin, and occurs through the vancomycin resistance (van) operon.

In enterococci, this operon can be carried chromosomally or on a plasmid. This operon includes the two component regulatory system VanS/R with vanS being a membrane histidine kinase and vanR the response regulator, the D-lactate dehydrogenase gene vanH, a variable ligase vanA or vanB, the D-Ala-D-Ala dipeptidase gene vanX (Fig. 7). Activation of VanS by vancomycin results in the induction of vanHAX expression through VanR. VanH provides D-Lac from the reduction of pyruvate, whereas free D-Ala are provided through VanX activity, which are incorporated by VanA to produce D-Ala-D-Lac pentapeptides. Furthermore, the D-Ala-D- Ala that is constitutively synthesized is removed by VanX (Faron, Ledeboer and Buchan 2016).

(22)

21

The replacement by D-Lac results in a decrease of affinity to vancomycin of almost 1000-fold (Courvalin 2006).

Figure 7. Vancomycin resistance (van) operon. The VanSR two-component regulatory system responds to vancomycin that leads to expression of vanH, vanA or vanB, and vanX downstream genes.

The D,D-dipeptidase VanX cleaves repeats of D-Ala-D-Ala and thereby provides free D-Ala for VanA/B, whereas the D-hydroxyacid dehydrogenase VanX provides D-Lac through reduction of pyruvate. VanA and VanB ligases enable the production of D-Ala-D-Lac pentapeptides with low affinity for vancomycin.

The D,D-carboxypeptidase VanY cleaves the D-Ala terminal peptide, which results in reduction of pentapeptides with high affinity to vancomycin. VanZ is involved in teicoplanin resistance through unknown mechanism and is present in strains carrying vanA. Adapted from (Faron et al. 2016).

Enterococci bearing the vanA gene show a high level of vancomycin resistance and are the predominant vancomycin-resistant Enterococcus (VRE) among E. faecium and E. faecalis, whereas strains carrying vanB gene display a moderate level of resistance (Faron et al. 2016).

In S. aureus, two phenotypes of decreased susceptibility to vancomycin can be distinguished;

vancomycin intermediate-resistant S. aureus (VISA) and vancomycin-resistant S. aureus (VRSA). VRSA strains are less frequent and acquired vanA operon from a VRE E. faecalis. VISA strains appeared from vancomycin-susceptible S. aureus (VSSA) by accumulation of mutations that affect cell-envelope homeostasis with the most involved genes being walkR, vraRS and graRS. This results in a broader cell wall with increased free D-Ala-D-Ala dipeptides that reduce cross-linking (McGuinness, Malachowa and DeLeo 2017). Similarly, to vancomycin, teicoplanin, dalbavancin and telavancin resistance occurs by modification of the peptidoglycan precursors and involves the expression of the van operon, whereas resistance to oritavancin has not been observed in clinical isolates (Rosenthal et al. 2018).

(23)

22

Table 2. Genes involved in mechanisms of AMP resistance in Gram-positive bacteria.

Function E. faecalis E. faecium S. aureus

Daptomycin

mprF incorporates positive charged amino

acid lysine to PG NI NI Repulsion

dlt operon incorporates D-alanine into cell-wall

teichoic acids NI Repulsion Repulsion

cls cardiolipin synthase Diversion Repulsion Repulsion

lia locus 3-component regulatory syst. as part

of cell envelope stress response Diversion NI NI gdpD glycerophosphoryl diester

phosphodiesterase Diversion NI NI

Vancomycin, Teicoplanin, Telavancin, Dalbavancin van operon production of D-Ala-D-Lac

pentapeptides ✓ ✓ ✓

NI, not involved.

Two mechanisms have been proposed for daptomycin resistance. The first one is diversion of daptomycin from the preferential binding site. The anionic phospholipids, such as cardiolipin, are redistributed away from the septum, resulting in inadequate binding of daptomycin (Fig.

8A). This mechanism is found in E. faecalis and is associated with expression of several genes (Table 2). cls and gdpD encode for enzymes involved in the metabolism of phospholipids, a cardiolipin synthase and a phosphodiesterase, respectively, whereas the liaFSR locus encodes a three-component regulatory system involved in cell-envelope stress response that promotes redistribution of cardiolipin microdomains. The second mechanism occurs through electrostatic repulsion by increasing the net positive charges at the cell surface, preventing the binding of the calcium-daptomycin complex (Fig. 8B). In S. aureus this occurs through the incorporation of positively charged lysine residues into the peptidoglycan through MprF.

Another repulsion mechanism relies on the incorporation of the positively charged amino acid D-alanine into cell wall teichoic acids. This mechanism is encoded by the dlt operon in S. aureus and E. faecium (Tran et al. 2015, Heidary et al. 2018). Lastly, a cardiolipin synthase has also been described to be involved in the repulsion mechanism in S. aureus and E. faecium (Patel et al. 2006) (Table 2).

(24)

23

Figure 8. The two proposed mechanisms of daptomycin resistance. (A) In the first one, cardiolipins are redistributed away from the septum (black arrow), which results in diversion of daptomycin from the binding site and thereby, prevents daptomycin binding. (B) The second mechanism involves electrostatic repulsion of the calcium-daptomycin complex from the cell membrane upon increase of the net positive charges of the cell envelope. Adapted from (Heidary et al. 2018).

3. Antimicrobial peptides with broad spectrum activity 3.1 Host defense peptides

Host defense peptides or innate immunity peptides can be found in nature as part of the immune defense of many living organisms (Wang and Wang 2004, Jenssen, Hamill and Hancock 2006, Zasloff 2002). Host defense peptides able to kill bacteria like the cecropins isolated from Hyalophora cecropia (Saugar et al. 2006, López-Rojas et al. 2011), the human LL- 37 (Wang et al. 2014), and defensins from mammalian and invertebrate hosts have become a main focus for the development of new drugs (Ganz et al. 1985, Verma et al. 2007).

More than 2000 AMPs have been discovered with eukaryotic origin (Wang and Wang 2004, da Costa et al. 2015) and even though they have comparable antimicrobial activity, they are very diversified in their structure and sequence (Jenssen et al. 2006). Main classes based on the structure of these AMPs involve a-helical peptides, b-sheet peptides and looped peptides (Jenssen et al. 2006, Nguyen, Haney and Vogel 2011) (Fig. 9).

(25)

24

Figure 9. Structure of host defense peptides. The human cathelicidin LL-37 has an α-helical conformation whereas the bovine indolicidin is a looped peptide. Adapted from (Haney, Mansour and Hancock 2017).

3.2 Bacteriocins

Bacteriocins are AMPs synthesized by bacteria that can inhibit or kill other bacteria, while the producer bacteria are immune (Cotter, Hill and Ross 2005b). They are generally synthesized under the form of an inactive pre-peptide and transported by ABC-transporters through the membrane. The peptide becomes active when the leader sequence composing the pre- peptide is cleaved. This event can take place either intracellularly, during or after export (Oppegård et al. 2007, Cotter, Hill and Ross 2005a). The co-expression of immunity proteins makes the producer bacteria immune to its own bacteriocins (Cotter et al. 2005b, Yang et al.

2014). Bacteriocins have been categorized into a large number of classes according to their structure, mode of action and producing strain. However, it has been proposed to divide bacteriocins from Gram-positive bacteria into only three classes: Class I (lantibiotics), Class II (non-lanthionine) and Class III (bacteriolysins). Bacteriocins from Gram-negative bacteria have been divided according to their size into colicins and microcins (Cotter et al. 2005a, Yang et al.

2014).

Bacteriocins can target different cellular processes but overall bacteriocins from Gram- positive bacteria act on bacterial membrane such as the lantibiotics Nisin (Wiedemann et al.

2001) and Lactococcocins A (van Belkum et al. 1991, Holo, Nilssen and Nes 1991). Bacteriocins from Gram-negative bacteria act on intracellular pathways like the metabolism of proteins, transcription or DNA replication. Microcin MccC7-51 inhibits protein synthesis (Cotter, Ross and Hill 2013), microcin J25 (MccJ25) inhibits the RNA polymerase (Mukhopadhyay et al. 2004) and microcin B17 (MccB17) inhibits the DNA gyrase (Vizán et al. 1991) (Fig. 10).

(26)

25

Figure 10. Mechanisms of action of representative bacteriocins inhibiting Gram-negative bacteria.

From (Cotter et al. 2013).

4. Antimicrobial peptides against Gram-negative bacteria 4.1 P. aeruginosa, a challenge for antimicrobial therapies

MDR Gram-negative pathogens are a major cause of nosocomial infections and deserve particular attention with respect to therapeutic decisions and epidemiological survey (Livermore 2012). P. aeruginosa, which is the main focus of this thesis, is well known for its ability to develop MDR resistance during prolonged antibiotic treatments. This organism is an ubiquitous bacterium encountered in the environment, but also as an opportunistic pathogen associated with acute and chronic infections (Barbier et al. 2013, Moradali, Ghods and Rehm 2017). The large genome (6.4 to 7.2 Mb) of P. aeruginosa encodes a vast array of regulatory enzymes that provide great adaptability and high versatility (Rocha et al. 2019, Klockgether et al. 2011). P. aeruginosa is the main reason of mortality and morbidity in immunocompromised and cystic fibrosis (CF) patients (Rybtke et al. 2015). Concerns about this pathogen are directed on lung infections as P. aeruginosa chronic infections in CF patients are refractory to antibiotic treatments, resulting in decreased pulmonary functions and consequently to mortality (Lyczak, Cannon and Pier 2002). As it is resistant to many available antibiotics, the treatment

(27)

26

of infections with P. aeruginosa is a great threat and, in addition, antibiotics overuse promoted the emergence of MDR P. aeruginosa (Lister, Wolter and Hanson 2009, Hirsch and Tam 2010).

This pathogen was listed among bacterial species for which development of novel antibiotics are a critical need by World Health Organization (Tacconelli et al. 2018).

Almost all antibiotics used for treatment of P. aeruginosa infection target intracellular components and, therefore, need to cross the outer membrane. Aminoglycosides enter the cells by the self-promoted uptake across the outer membrane through displacement of Mg²⁺

ions that stabilize the outer membrane, whereas β-lactams and quinolones use porin channels or enter the cell by diffusion (Lambert 2002).

However, P. aeruginosa shows intrinsic resistance to these antibiotics through the production of inactivating enzymes, a low permeability of the outer membrane and the expression of broad-spectrum efflux pump systems (Hancock and Speert 2000). P. aeruginosa produces β- lactamases that hydrolyze the β-lactam ring of these antibiotics, the most important being the inducible cephalosporinase AmpC. P. aeruginosa also encodes endogenous aminoglycoside- modifying enzymes, and frequently acquires additional aminoglycoside-modifying enzymes by horizontal gene transfer (Ramirez and Tolmasky 2010, Wright 2005).

The outer membrane of P. aeruginosa includes several types of porins. OprF, which is a non- specific porin and the major porin of P. aeruginosa, enables slow diffusion of small hydrophilic molecules. Specific porins have distinct sites for the binding of small molecules and sugar compounds. They include the carbohydrate-specific OprB, the phosphate-specific OprP, the pyrophosphate-specific OprO and the basic amino acid-specific OprD porin. OprC and OprH are gated porins that are regulated by copper and magnesium ions, respectively, and involved in the uptake of ion complexes.

Finally, P. aeruginosa expresses efflux pumps by which antibiotics are expelled into the external medium. OprM, OprN and OprJ are outer membrane channel proteins that are part of the tripartite efflux pumps (Hancock and Brinkman 2002). Among the five families of efflux pumps, the resistance-nodulation-division (RND) family is mainly responsible for MDR phenotypes in clinical isolates. P. aeruginosa encodes twelve RND type pumps, but only MexAB-OprM, MexCD-OprJ, MexEF-OprN and MexY-OprM were shown to provide clinically relevant resistance levels when overexpressed (Dreier and Ruggerone 2015). RND type pumps are composed of a cytoplasmic membrane transporter, a periplasmic linker protein and an

(28)

27

outer membrane channel (Daury et al. 2016). RND type pumps are chromosomally encoded and have been identified on mobile genetic elements.

Acquired resistance mechanisms involve mutational alterations in chromosomally encoded antibiotic target or regulator genes as well as acquisition of resistance determinants by horizontal gene transfer (Breidenstein, de la Fuente-Núñez and Hancock 2011). Mutations in the transcriptional regulators such as nalB, nalC, nalD or mexR lead to overexpression of the MexAB-OprM efflux pump and increase fluoroquinolones and β-lactams resistance, whereas mexZ mutation causes the overexpression of MeXY-OprM and results in fluoroquinolone and aminoglycoside resistance (Braz et al. 2016, Srikumar, Paul and Poole 2000, Muller, Plésiat and Jeannot 2011). Antibiotic target modifications such as the quinolone resistance determining regions (QRDR) of the gyrase genes occur frequently upon quinolone exposure. Mutations in gyrA and gyrB (DNA gyrase), and parC and parE (topoisomerase IV) lead to decreased binding of the quinolones to the gyrase or topoisomerase IV complexes (Bruchmann et al. 2013).

Mobile genetic elements carrying enzymes that modify aminoglycosides, resulting in decreased affinity for the 30S subunit are an example of horizontal gene transfer (Vakulenko and Mobashery 2003).

Adaptive resistance occurs when antibiotics or environmental stimuli modify expression of specific resistance determinants. Prominent examples are the induction of the MexXY efflux pump by macrolides and aminoglycosides, the expression of the arn operon upon exposure to polymyxin, or the induction of AmpC beta-lactamase by imipenem (Taylor, Yeung and Hancock 2014, Fernández et al. 2010).

4.2 Polymyxins, a potent class of natural antimicrobial peptides

As the development of MDR P. aeruginosa as well as carbapenemase-producing Enterobacteriaceae and A. baumannnii strains has been increasingly reported, polymyxins have been reintroduced in clinical settings (Dijkmans et al. 2015). Polymyxins are positively charged cyclic AMPs isolated from Paenibacillus polymyxa (Fig. 11). Due to their toxicity, colistin (polymyxin E) and polymyxin B were limited to topical applications and are currently used as last-resort drug to treat MDR Gram-negative bacteria (Koch-Weser et al. 1970, Bialvaei and Samadi Kafil 2015). However, the overuse of polymyxins and the recent report of a mobile colistin resistance gene (mcr-1) have promoted the emergence of polymyxin resistance (Liu et al. 2016), which emphasizes the urgent need of developing novel antimicrobial peptides.

(29)

28

Figure 11. Structure of polymyxin B (A) and polymyxin E (colistin) (B). From (Zhang et al. 2019).

4.3 Synthetic peptides

The wide diversity of AMPs underscores the capacity of peptides to be further developed as drug therapies. Even though a lot of AMPs have been described, not all of them passed clinical trials and only some of them have been approved. Most of the AMPs under clinical trials are limited to topical utilization as they are susceptible to degradation by proteases, lead to systemic toxicity, and are rapidly cleared by kidneys (Andersson, Hughes and Kubicek- Sutherland 2016, Vaara 2009, Jenssen et al. 2006). To prevent these issues and to increase the potency of AMPs, new approaches to generate varied structures and applications have been investigated (Fjell et al. 2009, Fjell et al. 2011, da Costa et al. 2015). For example, in the case of polymyxins, attempts to enhance their efficacy and decrease their toxicity have been made (Vaara 2013). Manipulations on lantibiotics have as well been undertaken (Cotter et al.

2005a). AMPs can be made as linear or cyclic molecules by chemical synthesis (Badosa et al.

2007, Scott et al. 1999). Peptides can be exploited to synthesize peptidomimetics like peptoids (Mojsoska, Zuckermann and Jenssen 2015, Chongsiriwatana et al. 2008). Peptoids arising from the circularization of linear peptides present advantages such as being less subject to degradation (Mojsoska et al. 2015, Tan et al. 2008). In the same way, glycosylation, phosphorylation or replacement of L-amino acids by D-amino acids have been applied to decrease the susceptibility to degradation by proteases (Jenssen and Aspmo 2008, Wade et al. 1990).

One strategy, which is of great interest to this thesis, is the solid phase peptide synthesis that allows the synthesis of many peptides (Amblard et al. 2006). Solid phase peptide synthesis is mainly used for peptide sequences with natural origin, but can also be applied to non-natural

(30)

29

peptides. The introduction of branching points in the peptide chain topology generates cyclic and dendrimeric peptides displaying promising pharmacokinetic properties (Mintzer et al.

2012).

Using this type of synthesis, our collaborators from University of Bern, the group of Prof.

Reymond, have developed dendrimer AMPs. These molecules are made of amino acids, mostly leucine and lysine, which are alternating in the branches of the sequence (Fig. 12).

Dendrimer AMPs can be of 1^st, 2^nd or 3^rd generation depending on the sequence that grows by addition of an amino acid layer (generation) after each branch point. They are characterized by the presence of 13 to 17 positive charges and a molecular weight (MW) varying between 4500 and 4900 g/mol (Stach et al. 2012, Stach et al. 2014, Pires et al. 2015).

Figure 12. Structure of a dendrimeric peptide. 1^st, 2^nd or 3^rd generation dendrimers depend on the sequence that grows one generation after each branch point. G, generation. L, leucine. K, lysine.

Adapted from (Siriwardena et al. 2018).

Among these synthesized dendrimer AMPs, G3KL, which is a 3^rd generation dendrimer with a MW of 4531.38 g/mol and carrying 15 positive charges (Fig. 13), shows potent activity against A. baumannii and MDR P. aeruginosa strains and a faster killing compared to polymyxin of P.

aeruginosa cells (Siriwardena et al. 2018, Pires et al. 2015). This thesis involves studies with the AMP dendrimer G3KL.

(31)

30

Figure 13. Structure and sequence of G3KL. G3KL is a 3^rd generation AMP dendrimer with a MW of 4531.38 g/mol and 15 positive charges. 1^st, 2^nd and 3^rd generation are represented in red, green and black color, respectively. The branching lysine in the sequence are in italic. L, leucine. K, lysine. Adapted from (Pires et al. 2015).

4.4 Mode of action of antimicrobial peptides

Since the discovery of AMPs, their mode of action (MOA) has been largely studied as this is crucial for further development as therapeutic drugs. The common features of AMPs such as amphipathicity, cationicity and short length are important for their MOA (Hancock and Sahl 2006). The binding and insertion of AMPs are more specific to the plasma membrane of fungi and bacteria than to the membranes of eukaryotes. This is mainly due to the fact that the outer leaflets of fungal and bacterial plasma membranes contain more exposed anionic phospholipids in comparison to the plasma membranes of animals and plants, which have the phospholipids negatively charged headgroups localized on the inner leaflet (Matsuzaki 1999).

Other components that are negatively charged like lipopolysaccharide (LPS), wall teichoic acids (WTA) and lipoteichoic acids (LTA) can also be found but only on bacterial cell surfaces (Münch and Sahl 2015). AMPs can bind and accumulate to the surfaces of membranes through electrostatic interactions between these negatively charged components of the cell envelope and the positively charged amino acids (Münch and Sahl 2015, Pasupuleti, Schmidtchen and

(32)

31

Malmsten 2012, Epand et al. 2010). In addition, other components can be linked to the activity and selectivity of AMPs like cholesterol present in the target membrane modifying its fluidity and stability, and the transmembrane potential (Matsuzaki 1999). In the case of polymyxins, the integrity of the membrane is disrupted by electrostatic interactions with the negatively charged LPS of the outer membrane. This results in the displacement of calcium and magnesium ions from the LPS phosphate groups, and leads to membrane disruption, cell lysis and death (Bialvaei and Samadi Kafil 2015).

Several models have been proposed for the MOA of AMPs. They are all based on the same first event, which is stoichiometric AMP accumulation by binding to the phospholipids on the outer leaflet of the cell membrane, and once a threshold is reached, cell membrane permeabilization (Melo, Ferre and Castanho 2009). The models can be divided in two categories: the transmembrane pore and the non-pore models.

The barrel-stave pore and toroidal pore models are the two models of the transmembrane pore category. In the barrel-stave model, the AMPs are located in parallel to the membrane and get inserted in a perpendicular way in the lipid bilayer (Ehrenstein and Lecar 1977) (Fig.

14). This enables lateral peptide-peptide interactions in a similar way to the protein ion channels of the membrane. In this model of pore formation, the amphipathic structure (a and/or b-sheet) of the peptide is important because the hydrophobic regions can interact with the lipids of the membrane and the hydrophilic residues can form the channels lumen (Brogden 2005, Breukink and de Kruijff 1999). A minimum length of about 22 residues for α- helical or about 8 residues for β-sheet are essential properties for the AMPs of this category in order to cross the lipid bilayer. Only few AMPs like protegrins (Brogden 2005), pardaxin (Rapaport and Shai 1991), alamethicin (Wimley 2010) have been described to act through the barrel-stave model.

(33)

32

Figure 14. Proposed models for the mode of action of AMPs. When a threshold concentration of AMPs is reached, the permeabilization of the membrane occurs through the barrel-stave, toroidal or carpet model.

In the toroidal pore model, the peptides still get inserted in a perpendicular way to the lipid bilayer, but without specific peptide-peptide interactions (Wimley 2010). The peptides rather cause a local curvature of the lipid bilayer and the pores are formed partially by the peptides and partially by the headgroup of the phospholipids (Fig. 14). This temporary lipid-peptide molecule is called the “toroidal pore”. The arrangement of the bilayer is a characteristic that differentiates between the toroidal and the barrel-stave pore models. In the toroidal pore model, the hydrophobic and hydrophilic distribution of the lipids is disturbed, while in the barrel-stave model, the lipid distribution is preserved. This contributes to the interactions of the lipid headgroup and tail with the surfaces. Some peptides can displace to the inner cytoplasmic leaflet and enter the cytoplasm to possibly target intracellular components as the formed pores are temporary upon disintegration (Uematsu and Matsuzaki 2000). The selectivity of ions and the discrete size are other characteristics of the toroidal pore model (Yeaman and Yount 2003). AMPs like lacticin Q (Lee, Hall and Aguilar 2016), magainin 2 (Lee et al. 2016), and melittin (Lee et al. 2016, Wimley 2010) act through the toroidal pore model.

Both barrel-stave and toroidal pore models lead to the depolarization of the membrane and cell death.

(34)

33

AMPs can also act without pore formation. Such AMPs are classified in the category of non- pore models. In the so called carpet model (Lee et al. 2016, Shai 2002, Yeaman and Yount 2003), the AMPs are absorbed in a parallel way to the lipid bilayer and cover the surface of the membrane and therefore create a “carpet” until reaching a threshold concentration (Fig.

14). This results in disadvantageous interactions on the membrane surface and leads to the loss of the membrane integrity. This loss generates a detergent-like effect that will form micelles and disintegrate the membrane. The structural disruption of the membrane bilayer in micelles is known as well as the detergent-like model. The specific peptide-peptide interactions and their insertion in the hydrophobic core to form a transmembrane channel are not required in the carpet model (Yeaman and Yount 2003). Several peptides act as AMPs regardless of their length of the sequence or their specific amino acid composition. They generally act through the carpet model (Shai 2002) and because of their amphiphilic character, they are used at high concentrations (Jenssen et al. 2006). Cecropin (Sitaram and Nagaraj 1999), indolicidin (Sikorska et al. 2009) and LL-37 (Shai 2002) are examples of AMPs using the carpet model.

There are many models to describe the mode of action of AMPs, and besides those described above, there are other models including the electroporation, the interfacial activity and the Shai-Huang-Matsazuki models (Lee et al. 2016). The specific distinctions illustrated in Figure 14 are not taken into account in some models. For example, the carpet model has been suggested to be a prerequisite step for the toroidal model (Lee et al. 2016). The described membrane models have been used in almost all studies on the mode of action of AMPs, and only few AMPs have been studied with the whole bacterial cells using imaging techniques (Gee et al. 2013, Choi, Rangarajan and Weisshaar 2016).

4.5 Mechanisms of resistance

For a long time, it has been thought that resistance against AMPs would not arise as the bacteria would need to modify the organization and/or composition of structures that are well conserved, such as the membrane and cell-wall (Zasloff 2002). This idea was supported by the fact that, until recently, there was no report on resistance against AMPs (Zasloff 2002, Pasupuleti et al. 2012). Nevertheless, resistance to colistin (polymyxin E) in A. baumannii and K. pneumoniae has been reported and is a major concern (García-Quintanilla et al. 2014, Li et al. 2006, Moffatt et al. 2010, Jayol et al. 2015). Diverse mechanisms of resistance against AMPs

(35)

34

have emerged as a result of exposure to host and environmental AMPs. Some of these mechanisms involve proteolytic digestion, efflux activity, modification of charges at the cell surface and alteration of the membrane fluidity (Cole and Nizet 2016).

In the case of polymyxins, four main resistance mechanisms that are conserved among Gram- negative bacteria can be mentioned: 1) modification of the lipid A of the lipopolysaccharide (LPS) on the outer membrane, 2) loss of the LPS such as in A. baumannii (Moffatt et al. 2010), 3) extrusion by efflux pumps such as in K. pneumoniae and 4) expression of capsule polysaccharides that can trap polymyxins (Llobet, Tomás and Bengoechea 2008). The most common resistance mechanism to polymyxins is the modification of the LPS by the addition of either 4-amino-4-deoxy-L- arabinose (L-Ara4N), phosphoethanolamine (pEtN) or galactosamine to the lipid A or the LPS core by enzymatic reactions (Trent 2004, Moskowitz, Ernst and Miller 2004, Pelletier et al. 2013). For example, in the case of L-Ara4N addition, a positive charge on the amino-arabinose will replace a negatively charged phosphate. These LPS modifications lead to a reduction of the total negative charge provided by the phosphate residues thereby decreasing the binding of the positively charged polymyxin molecule to the bacterial surface (Fig 15).

Figure 15. LPS modifications. The addition of L-Ara4N and pEtN through ArnT and EptA, respectively, to the lipid A of LPS leads to decrease of negative charges at the membrane surface, whereas the addition of a palmitate chain (PagP) or the removal of an acyl chain (PagL) lead to decrease of permeability. Adapted from.

Responses of pseudomonas aeruginosa and other eskape pathogens to antimicrobial peptide dendrimers

Thesis

Reference

Responses of pseudomonas aeruginosa and other eskape pathogens to antimicrobial peptide dendrimers

RESPONSES OF PSEUDOMONAS AERUGINOSA AND OTHER ESKAPE PATHOGENS TO ANTIMICROBIAL PEPTIDE DENDRIMERS

Table of Contents

Acknowledgements

Summary

Résumé

Introduction