1. Introduction
The emergence and spread of bacteria that are resistant to many routinely-used antibiotics has driven research into the development of new classes of antimicrobial agents with high antibacterial efficacy and new modes of action. Naturally occurring antimicrobial peptides (AMPs) are essential components of the innate immune system in many different organisms, and they are characterized by diverse targets and mechanisms of action, broad-spectrum activity and only rare instances of resistance [
1,
2,
3,
4]. AMP gene families have expanded and diversified particularly among the arthropods, compensating for the lack of an adaptive immune system [
5].
AMPs have been isolated and characterized from many arthropods, including insects [
6,
7], ticks [
8,
9] and scorpions [
10,
11,
12]. Natural AMPs generally have short, cationic, amphipathic and α-helical structures that interact preferentially with microbial membranes via electrostatic contacts [
13,
14]. Increasing the net positive charge and hydrophobicity enhances the antibacterial activity of AMPs due to the greater affinity for the negatively-charged bacterial membrane [
15]. However, the therapeutic value of AMPs is limited by their hemolytic activity, which is particularly prevalent in the case of scorpion AMPs. The therapeutic index of AMPs can be improved by reducing their hemolytic activity, which can be achieved by engineering new versions with suitable amino acid substitutions [
16].
Here, we describe the functional characterization of natural and engineered versions of the scorpion AMPs UyCT1, UyCT3, UyCT5, Uy17, Uy192 and Uy234 from
Urodacus yaschenkoi, and Um2, Um3, Um4 and Um5 from
U. manicatus [
17,
18,
19,
20]. The AMPs were tested for antifungal, antibacterial, hemolytic and cytotoxic activities in order to identify more potent and selective versions with a higher therapeutic index.
3. Discussion
Scorpion venoms contain hundreds of different compounds that target membranes, ion channels and receptors [
22,
23]. One class of compounds, known as “non-disulfide-bridged peptides”, comprises short peptides with diverse properties including antimicrobial activity, lytic activity and the potentiation of bradykinin and cell signaling [
24]. These peptides are considered to belong to the broader category of antimicrobial peptides (AMPs), an important component of innate immunity in many different organisms, due to their ability to inhibit the growth of pathogens including multidrug-resistant bacteria at MIC values as low as 1 µM [
18,
19,
25]. Scorpion venom AMPs therefore provide a rich source of potential new antimicrobial compounds that could be used to address the increasing prevalence of pathogens resistant to conventional antibiotics. However, many scorpion AMPs also cause undesirable levels of hemolysis and cytotoxicity, so the use of natural peptides would cause unacceptable side effects. It may nevertheless be possible to exploit these peptides if peptide engineering can be used to balance potency with increased selectivity against pathogens [
26].
Natural AMPs in scorpion venom are highly conserved (
Figure 1) as shown by their phylogenetic clustering (
Figure 2), and in some cases the difference between AMPs can be as little as one or two residues. Interestingly, such small differences in sequence often result in much larger differences in bioactivity. For example, UyCT1 from
U. yaschenkoi [
18] and Um3 from
U. manicatus [
20] differ at only one residue (S12N), but this substitution makes UyCT1 more potent that Um3 (
Table 3). The degree of conservation among scorpion venom AMPs is such that some diverse species produce identical AMPs, e.g., StCT2 from
S. tibetanus [
27] is identical to Um3 from
U. manicatus, and UyCT3 from the Australian scorpion
U. yaschenkoi [
18] is identical to OcyC1 from the Brazilian scorpion
Opisthacanthus cayaporum [
28]. UyCT3 differs from Um5 at two positions (L2F and S3K), and the nature of the substitutions increases the net positive charge on Um5, increasing its activity against all the bacteria we tested (
Table 3). These natural substitutions led to our hypothesis that a rational engineering approach could be used to modify scorpion AMPs to increase their potency and specificity for microbial membranes, thereby reducing their hemolytic activity and cytotoxicity towards mammalian cells and thus improving their therapeutic index.
Here, we compared the bioactivity of natural and engineered scorpion AMPs to investigate their suitability as therapeutic leads. In order to preserve the secondary structure of the natural AMPs as far as possible, changes were only made in non-essential residues and their purpose was to increase the positive net charge (
Table 2). The UyCT peptides showed low MIC values (1–15 µM) against several human pathogens, including multidrug resistant bacteria, but they also caused mild hemolysis [
19]. Rational engineering was most successful in the case of UyCT1, where the addition of a C-terminal oligo-lysine tail to generate engineered derivative D11 reduced the hemolytic activity to negligible levels (HC
50 = 1110 µM) compared to the parental UyCT1 (HC
50 = 142.5 µM). The increase in net positive charge and the bulky side chains made the peptide more potent and also more selective towards bacteria (
Table 3).
Another potential way to improve selective antimicrobial activity is to modulate the charge distribution of amphipathic helices, i.e., increase the hydrophobicity of the hydrophobic face and/or the hydrophilicity of the hydrophilic face [
29]. We increased the net positive charge of UyCT1 by adding lysine residues either at the C-terminus or distributed along the parent sequence (
Table 2). As stated above, the engineered peptide D11 (C-terminal oligo-lysine tail) had a much lower MIC than the parental peptide UyCT1 (~1 µM compared to ~4 µM) against all seven bacteria tested (
Table 3), probably reflecting the formation of a more effective membrane-transducing domain [
30]. In contrast, D12 (based on UyCT1 but with distributed lysine residues) was totally inactive (data not shown). Similar results were observed with UyCT1 when we replaced the negatively charged glutamic acid residue with lysine (E8K) to generate the D4 derivative. The D4 peptide was less potent than UyCT1 (
Table 3) but also showed no hemolytic activity (
Table 3). Similarly, D5 was derived from UyCT1 by replacing tryptophan with leucine (W6L) and it showed only ~25% of the activity of the parental AMP (
Table 3), probably because the presence of the large tryptophan side chain enhances the membrane purterbation [
31]. These findings suggest that an increase in the net positive charge can increase the potency of AMPs but only if the charge is concentrated at the end of the peptide and not distributed throughout the sequence.
The engineered peptides D5, D10 and D11 showed a significant loss of hemolytic activity compared to their parent sequences (
Table 3), substantially increasing the corresponding therapeutic index, especially in the case of D11 (
Table 4). D5 is a derivative of UyCT1 with W7L and N12K substitutions, and D10 is a derivative of Uy234 with the addition of four scattered lysine residues (
Table 2). The hemolytic activity of these engineered peptides was 100-fold lower than the parent peptides (
Table 3). However, more sophisticated engineering approaches are required to reduce HC
50 values. For example, the naturally occurring peptides UyCT3 and Um5 differ at two residues (positions 2 and 3) but their HC
50 values are almost identical (~60 µM), whereas UyCT1 and Um3 differ only at position 12 and the HC
50 values are 142.5 and 126.2 µM, respectively. The hemolytic effect of AMPs may depend more on the stereochemistry of the residues than the nature of the substitution [
16], so L-to-D replacements may offer a better strategy to reduce the hemolytic activity of scorpion AMPs [
32].
UyCT3 and D1 differed at two positions (L2F and S3G) but behaved similarly in terms of antibacterial and hemolytic activities (
Table 3), possibly because both leucine and phenylalanine are lipophilic with neutral nonpolar side-chains. Similarly, replacing serine with glycine has only a limited effect, probably because these residues have similar hydrophobicity. In contrast, Uy192 and D2 also differed at two positions (G11S and L13F) but the engineered version was more potent and the hemolytic activity was lower, even though the same pairs of residues were exchanged (
Table 3), suggesting that the size of the side chains may have affected the conformation of the peptide in a favorable manner [
19].
Although these peptides showed promising antibacterial activity, they were inactive towards phytopathogenic fungi. Similarly, opisin from
Opistophthalmus glabrifrons was able to inhibit a range of bacteria such as
S. aureus,
M. luteus,
Bacillus megaterium, B. thuringiensis and
Nocardia coralline, but showed only a weak activity against
Candida tropicalis [
33]. One possible reason for such specificity towards bacteria is the effective electrostatic interaction of scorpion AMPs with the bacterial membrane—as a determinant factor for the bacterial viability—whereas acting against fungi via targeting cell wall components and/or intracellular molecules needs more complex interactions, which might be achieved through the synergism between different peptides [
34,
35].
In summary, our rational engineering approach yielded a number of derivative peptides with therapeutic indices better than the natural parental peptides. In particular, peptides D4, D5, D10 and D11 showed more potent antimicrobial activity and lower hemolytic and cytotoxic activity. D10 and D11 showed the highest activity against diverse bacteria and would be suitable as leads for the development of a new generation of antimicrobial products to address the prevalent threat of multidrug-resistant bacterial pathogens.