Structure‐activity relationship studies of shortened analogues of the antimicrobial peptide EeCentrocin 1 from the sea urchin Echinus esculentus

EeCentrocin 1 is a potent antimicrobial peptide isolated from the marine sea urchin Echinus esculentus. The peptide has a hetero‐dimeric structure with the antimicrobial activity confined in its largest monomer, the heavy chain (HC), encompassing 30 amino acid residues. The aim of the present study was to develop a shorter drug lead peptide using the heavy chain of EeCentrocin 1 as a starting scaffold and to perform a structure‐activity relationship study with sequence modifications to optimize antimicrobial activity. The experiments consisted of 1) truncation of the heavy chain, 2) replacement of amino acids unfavourable for in vitro antimicrobial activity, and 3) an alanine scan experiment on the truncated and modified heavy chain sequence to identify essential residues for antimicrobial activity. The heavy chain of EeCentrocin 1 was truncated to less than half its initial size, retaining most of its original antimicrobial activity. The truncated and optimized lead peptide (P6) consisted of the 12 N‐terminal amino acid residues from the original EeCentrocin 1 HC sequence and was modified by two amino acid replacements and a C‐terminal amidation. Results from the alanine scan indicated that the generated lead peptide (P6) contained the optimal sequence for antibacterial activity, in which none of the alanine scan peptides could surpass its antimicrobial activity. The lead peptide (P6) was also superior in antifungal activity compared to the other peptides prepared and showed minimal inhibitory concentrations (MICs) in the low micromolar range. In addition, the lead peptide (P6) displayed minor haemolytic and no cytotoxic activity, making it a promising lead for further antimicrobial drug development.


| INTRODUCTION
Bacterial resistance to commercial antibiotics has increased severely over the last years. Infectious bacteria that were once easily treatable by antibiotics have now become resistant. 1 There is consequently a pressing need to come up with alternatives to the current antimicrobial drugs. Antimicrobial peptides (AMPs) are proteinaceous natural products found in all living phyla examined, and due to their high structural diversity, they are considered attractive hit compounds for development of drug leads for novel antibiotics. AMPs efficiently kill bacterial pathogens with low toxicity to mammalian cells, and often have broad-spectrum antimicrobial activity against pathogenic Grampositive and Gram-negative bacteria. [2][3][4] Most AMPs appear to interact with bacterial membranes, forming pores or aggregates at the membrane surface, causing cooperative permeabilization and loss of membrane integrity. 3 In membrane-like environments, AMPs tend to form amphipathic structures, i.e. structures with separate hydrophobic and hydrophilic domains. Their net positive charge facilitates interactions with the negatively charged bacterial membranes and/or cell walls, whereas their amphipathic character enables membrane permeabilization. 4 Because AMPs in general act on the lipid bilayer structure of bacterial membranes, there seems to be lower propensity of generating resistance against AMPs compared to other antibiotic classes. [5][6][7] There is even recent evidence that antibiotic-resistant Escherichia coli display increased sensitivity towards AMPs. 8 Some AMPs have additional mode of actions, attacking both extracellular and intracellular targets rapidly. 8,9 Several AMPs are currently in the medical pipeline due to these favourable properties. [10][11][12][13] The centrocins are potent marine natural AMPs originally isolated and characterised from the sea urchins Strongylocentrotus droebachiensis 14 and Echinus esculentus. 15 Centrocin-coding genes have also been identified in the genome of S. purpuratus. 16 The centrocins display antimicrobial activities in the sub-micromolar range against both Gram-positive and Gram-negative bacteria, as well as fungi. Centrocins range from 4.4-4.8 kDa in size and have a heterodimeric structure, i.e., they have a heavy chain (HC) of~30 amino acid residues which is linked by a disulphide bridge to a light chain (LC) of~12 amino acid residues. All isolated centrocins contain brominated tryptophan residues in their HC. Previous in vitro experiments have shown that the antimicrobial activity is confined to the HC's and that non-brominated HC peptides are equally potent as the brominated ones. These data imply other purposes than direct involvement in bacterial killing for the LC and the brominated tryptophan residues. 14,15 In the present study, we aimed to design a shorter AMP with improved or similar antimicrobial activity as the non-brominated HC of EeCentrocin 1 derived from E. esculentus. 15 This was done by successive structure-activity relationship (SAR) studiesconcentrated on truncation, sequence modification, C-terminal amidation, and alanine scanning. The EeCentrocin 1 HC was chosen as a framework because of its potent antibacterial activity and low haemolytic activity. 15 The hypothesis was that the bioactive part of EeCentrocin 1 HC could be attributed to the N-terminal region, which contain both hydrophobic tryptophan residues and cationic residuestwo structural properties known to contribute to antimicrobial activity in other AMPs. [17][18][19] 2 | MATERIALS AND METHODS

| Purification and characterization
The peptides (P4-P18) were purified by preparative RP-HPLC using a

| Antifungal assay
The synthetic peptides were also screened for antifungal activity against Candida albicans (ATCC 10231), Aureobasidium pullulans, and Rhodotorula sp. (the last two were obtained from Professor Arne Tronsmo, The Norwegian University of Life Sciences, Ås, Norway).
The antifungal assay was performed as previously described. 21 Briefly, fungal spores were dissolved in potato dextrose broth (Difco) to a con-

| Haemolytic assay
The synthesised peptide analogues were screened for eukaryotic celltoxicity with a haemolytic activity assay using human red blood cells (RBC) as described previously. 21 The assay was performed in 96-well U-shaped microtiter plates (Nagle Nunc) with 50 μL peptide sample,

| Cytotoxicity assay
The

| RESULTS AND DISCUSSION
In the present study, a series of shortened peptides, based on the marine heterodimeric and brominated AMP EeCentrocin 1 (originally isolated from the red sea urchin E. esculentus), were chemically synthesized and screened for antimicrobial activity. Initial experiments involved the stepwise truncation of the non-brominated HC of the peptide. The LC and bromination of the Trp-residues within the HC were excluded from this study because of their minor importance for antimicrobial activity. 15

| Truncation of non-brominated HC
In silico modelling of the EeCentrocin 1 HC (P1, Figure 1) revealed that the N-terminal part of the sequence most likely forms an α-helix, a well-known conformation of AMPs. The N-terminal has an abundance of hydrophobic and cationic residues, which is a characteristic known to be of importance for the activity of AMPs. 19 Furthermore, the Nterminal region of many α-helical AMPs is shown to be important for antimicrobial activity. 17 Based on this qualified sequence evaluation of the HC of EeCentrocin 1, the last 14 C-terminal amino acid residues were removed, resulting in the peptide HC(1-16) (P2, Table 1). The peptide contains two Trp and six cationic residues (Arg/Lys). Truncation of the heavy chain led to reduced antibacterial activity, especially against S. aureus. However, the Gram-positive C. glutamicum was still sensitive to P2 albeit at a slightly higher concentration. An eight-fold decrease in potency was also observed against the Gram-negative test bacteria, maybe due to reduced charge of P2 compared to the full HC peptide (Table 1).
Since an electrostatic interaction between cationic AMPs and the negatively charged surface of bacteria is important for antibacterial activity, 26 (Table 1). Thus, all subsequent synthesised AMPs were prepared with the Asp8 residue substituted with Ala (or Arg) as well as having a C-terminal amide. Carboxyamidation, which increases the overall positive charge of the peptides, is previously shown to increase the antimicrobial activity of AMPs. 27 Early in the process, it was discovered that eliminating the N-termi-  (Table 1). However, the activity against the other test strains remained the same. The importance of Gly1 has been discussed beforethe presence of an N-terminal Gly-residue was recorded in 70% of 150 α-helical AMPs collected from the AMSDb database, perhaps serving as a capping residue for α-helices or providing resistance to aminopeptidases. 17 As the N-terminal truncation proved unsuccessful against P.
aeruginosa, the focus was directed towards further truncation of the  (Table 1). To improve the antibacterial activity of this 12-residue peptide, the C-terminal Arg-Lys-motif, which was recognised in the 16-residue peptides, was reinstated. This also made it possible to replace the original Asn12-residue, which can compromise peptide integrity by forming aspartimide side-products in SPPS involving chain-elongation through its side-chain and not the peptide-backbone. 28 The resulting peptide HC(1-12)A8K12 (P6) was the most potent AMP produced so far with antibacterial activities towards C. glutamicum and P. aeruginosa close to the original HC (P1) peptide ( Table 1).
The next analogue HC(1-9)R8 (P7) was synthesised using the same argument as with HC(1-12)A8 (P3) and HC(1-12)A8K12 (P6) to further shorten the peptide sequence and reinstate the C-terminal Arg-Lysmotif. However, the potency of this 9-residue peptide (P7) was much lower than the previous peptides, maybe due to reduction in net positive charge (Table 1)

| Alanine-scan of the lead peptide HC(1-12) A8K12
In order to investigate the importance to antibacterial activity of individual residues of the lead peptide HC(1-12)A8K12 (P6), each amino acid was substituted by Ala and antibacterial activity was recorded for each peptide in an alanine scan. Ala, along with Leu and Lys are known stabilisers of peptide α-helicity. 29 The peptides were named (apart from the lead peptide, P6) according to the original amino acid, position and substitution, i.e. the peptide where Gly was substituted with Ala in position 1, was named G1A. All Ala-scan peptides displayed antibacterial activity, but with different potencies against different strains (Table 3) Data obtained from Solstad et al. 15 2 Data obtained from Li et al. 14 3 Data and reference number obtained from NCBI. and only two AMPs were antibacterial against S. aureus at 50 μg/ml concentrations; R4A (P11) and K12A (P18) ( Table 3). The antibacterial profiles against the two Gram-negative strains were quite similar: the lead peptide HC(1-12)A8K12 (P6), T6A (P13), K9A (P15) and K12A (P18) were antibacterial at concentrations ≤6.3 μM, whereas W2A (P9) and W3A (P10) were the least antibacterial with MICs ≥50 μM against the Gram-negative bacterial strains (Table 3).
In general, all Ala-substitutions of the lead peptide HC(1-12)A8K12 (P6), except T6A (P13), resulted in reduced antibacterial activity (Table 3). This indicated that all residues except Thr6 (and perhaps Ala8) were of importance in P6 to maintain optimal antibacterial activity. The peptide T6A (P13) was generally the second most potent AMP after P6. However, while T6A (P13) was only marginally less potent than the lead peptide against S. aureus, P. aeruginosa, and E. coli, a four-fold dilution separated T6A (P13) and P6 against C. glutamicum (Table 3) Table 3), compared to the other peptides that were either more or less hydrophobic. Others also report a hydrophobicity-window for optimal antibacterial activity of AMPs. 30 Replacement of the positively charged residues with Ala in R4A  (Table 3). This reduced activity can be explained by reduction of net positive charge of these peptides. A high net positive charge is shown to be important for cationic AMPs for the initial electrostatic interaction with bacterial cell membranes. 19,29 In addition, all the positively charged residues were in the hydrophilic and charged face of the proposed α-helix. Insertion of an additional hydrophobic residue in this region may therefore alter the amphipathic character of the peptide.
Residue Lys12, which was positioned at the C-terminal end of the peptide, seemed to be the least important cationic residue according to the alanine scan experiment.
A noteworthy pair when considering individual amino acid substitutions were the Trp substitutions represented by the peptides W2A (P9) and W3A (P10) ( Table 3). Trp is a bulky hydrophobic residue, commonly accepted as a contributor to antimicrobial activity in AMPs. 19,31 This was also apparent in the current alanine scan experiment, where replacement of the Trp-residues with Ala resulted in a dramatic loss of antibacterial activity. However, a few points can be made regarding the loss of antibacterial activity when these amino acids were replaced. The peptides W3A (P10) and G1A (P8) showed similar high potencies against the Gram-positive bacterial strains, but W3A (P10) was noticeably less potent than G1A (P8) against the Gram-negative bacterial strains ( Table 3). The one AMP that was consistently least potent against all strains was W2A (P9), which indicate that Trp2 was a more important residue for antibacterial activity than Trp3 (Table 3). The exception being against E. coli where W2A (P9) was somewhat less potent than W3A (P10). As shown in Table 3, the retention times (hydrophobicity) on a C18 RP-HPLC column were reduced for W2A (P9) and W3A (P10) compared to the lead peptide HC(1-12)A8K12 (P6). This illustrates the importance of the hydrophobic character contributed by the two Trp-residues, which are located at the hydrophobic face of the predicted α-helix  (Table 3). Val contains an isopropyl side chain, in contrast to the methyl side chain in Ala. Replacement of Val with Ala would therefore slightly alter the hydrophobicity in the hydrophobic sector and thereby the overall amphipathicity of the peptide, an important characteristic of AMPs. 4 As shown in Table 3, the retention times were reduced for V7A (P14) and V10A (P16), indicating reduced hydrophobicity for these two peptides compared to the lead peptide HC(1-12)A8K12 (P6).

| Bacterial killing experiments
On the basis of our SAR studies (Tables 1 and 3

| Antifungal activity
The synthesised peptides were subjected to antifungal screening against the moulds A. pullulans and Rhodotorula sp., and the yeast C.
albicans. The lead peptide HC(1-12)A8K12 (P6) was superior in activity compared to the other peptides, including the full HC (P1) peptide (

| Haemolytic and cytotoxic activity
The lead peptide HC(1-12)A8K12 (P6) and a selection of the other synthetic peptides were screened for haemolytic activity against    Table 4).

| CONCLUSIONS
Pathogenic bacteria are becoming resistant to most antibiotics to an ever-increasing extent. This has spurred the discovery of novel antibacterial compounds such as AMPs isolated from natural sources.