Selective Antibacterial Activity and Lipid Membrane Interactions of Arginine-Rich Amphiphilic Peptides

The self-assembly behavior and antimicrobial activity of two designed amphiphilic peptides, R3F3 and R4F4, containing short hydrophobic phenylalanine (F) and cationic arginine (R) sequences, are investigated. The conformation of the peptides was examined using circular dichroism and FTIR spectroscopy, which show that they have a disordered secondary structure. Concentration-dependent fluorescence assays show the presence of a critical aggregation concentration (cac) for each peptide. Above the cac, small-angle X-ray scattering (SAXS) and transmission electron microscopy (TEM) reveal a population of twisted tapes for R3F3 and nanosheets for R4F4. The interaction of the peptides with model bacterial membranes comprising mixtures of the lipids DPPG [1,2-dipalmitoyl-sn-glycero-3-phosphoglycerol] and DPPE [1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine], was studied using SAXS and cryogenic-TEM. Analysis of the SAXS structure factor indicates that R3F3 interacts with lipid bilayers by inducing correlation between bilayers, whereas R4F4 interacts with the bilayers causing an increase in polydispersity of the vesicle wall thickness. Both peptides break vesicles with a 1:3 DPPG:DPPE composition, which is close to the ratio of PG and PE lipids observed in the lipid membrane of Pseudomonas aeruginosa, a pathogen responsible for serious infections and which has developed antimicrobial resistant strains. Both peptides show activity against this bacterium in planktonic form. Peptide R4F4 shows particularly strong bioactivity against this microbe, with a minimum inhibitory concentration (MIC) value in the range of concentrations where the peptide is cytocompatible. It was further shown to have activity against other Pseudomonas species including the common plant pathogen Pseudomonas syringae. Finally, we show that R4F4 inhibits the development of P. aeruginosa biofilms. This was examined in detail and a proposed mechanism involving binding of the signaling molecule c-di-GMP is suggested, based on circular dichroism spectroscopy studies and Congo red assays of extracellular polysaccharides produced by the stressed bacteria. Thus, R4F4 is a promising candidate antimicrobial peptide with activity against Pseudomonas species.


■ INTRODUCTION
The increase in prevalence of multi-antibiotic-resistant pathogens is of great concern and has been listed by the World Health Organization (WHO) as one of the biggest threats to modern day healthcare, food security, and development. There is great interest in the development of new therapeutics to overcome this. Peptides, especially those containing cationic residues or tryptophan, have been shown to have great potential in this regard. 1,2 Many antimicrobial peptides have evolved naturally in species of plants and animals, providing a strong basis to produce biocompatible therapeutics. Peptides with the ability to self-assemble are of interest as they may have increased in vivo stability through increased half-life and resistance to proteolysis.
Antimicrobial peptides are thought to interact with the cell membranes of bacteria through different mechanisms. 2 It has been proposed that some peptides induce pore formation where the pore is lined with peptide (the barrel-stave model), or when the peptide causes curvature of the lipid membrane (the toroidal pore model). 3 Another mechanism involves peptide deposition on the membrane surface (the carpet model), which is followed by a detergent-like action that leads to pore formation. 3 Cationic peptides are able to associate with the membrane and cause an electrical potential difference across the membrane, inducing electroporation (the molecular electroporation model). 4 An important class of antimicrobial peptide are those containing L-arginine. 2 L-Arginine is commonly found as an active site residue in proteins and is common in many cellpenetrating peptides. The antimicrobial activity of L-arginine is due to the cationic charge of this amino acid which interacts with anionic or zwitterionic lipid membranes, which may lead to pore formation. 5,6 Surfactant-like peptides (SLPs) are a group of peptides with promising applications as antimicrobials. Structurally, SLPs have a headgroup of a small sequence of charged residues and a longer hydrophobic tail, which gives them remarkable selfassembling properties. 7−9 Recently, the self-assembly of three SLPs, A 3 K, A 6 K, and A 9 K, was examined and distinct morphologies were observed for the three peptides. 10 The antimicrobial properties were found to be dependent on the length of the hydrophobic chain, with A 9 K being most active against Escherichia coli and Staphylococcus aureus, also strongly disrupting the structure of anionic vesicle DPPG, 10 used as a model system for bacterial cell membranes. Our group has studied the self-assembly properties and antimicrobial activities of several SLPs containing one or two arginine residues. Surfactant-like peptide A 6 R was found to self-assemble into nanofibers, and to have antimicrobial activity. 11,12 More recently, A 6 R in both capped and uncapped forms was examined. CapA 6 R forms nanofibrils, whereas A 6 R assembles into nanotapes, which preferentially interacted with lipid POPG in POPG/POPE vesicles [POPG = 2-oleoyl-1palmitoyl-sn-glycero-3-phosphoglycerol, POPE = 2-oleoyl-1palmitoyl-sn-glycero-3-phosphoethanolamine]. 13 The capped A 6 R variant showed selective activity against Gram-positive Listeria monocytogenes, and the uncapped version showed greater antimicrobial activity against this and also E. coli and S. aureus. 13 Another surfactant-like peptide, RA 3 R, containing arginine on both termini, was found to assemble into a polyproline II helix in water, which interacted strongly with POPG in mixed lipid vesicles. 14 RA 3 R was found to be particularly active against L. monocytogenes through membrane reorganization.
Here, we examine the self-assembly behavior and antimicrobial activity of two SLPs with sequences RRRFFF (R 3 F 3 ) and RRRRFFFF (R 4 F 4 ) (Scheme 1). These peptides comprise a hydrophobic phenylalanine block and a block of cationic Larginine residues, and were designed as novel arginine-rich SLPs containing aromatic hydrophobic blocks, incorporated to drive self-assembly via π-stacking interactions. Recently, Silva et al. examined the self-assembly of peptides consisting of alternating RF motifs. The short oligopeptide [RF] 4 was found to assemble into highly ordered amyloid-like fibrils above a critical aggregation concentration. 15 Further studies of [RF] n , where n is between 1 and 5, showed an increased structural order and cytotoxicity with increasing n. 16 In another example, Fojan et al. studied the self-assembly of RFFFR, which was shown to self-assemble into fibers stabilized by π-stacking interactions, which further assembled into solid spheres. 17 The two peptides we investigate, R 3 F 3 and R 4 F 4 , are architecturally different from these previously studied RFcontaining peptides and surfactant-like peptides, as they contain blocks of symmetric charged and hydrophobic blocks.
We first present results on the self-assembly and conformation of R 3 F 3 and R 4 F 4 . We, then, examine the interactions of these peptides with DPPG/DPPE vesicles. DPPG (1,2-dipalmitoyl-sn-glycero-3-phosphoglycerol), is anionic, whereas DPPE (1,2-dipalmyitoyl-sn-glycero-3-phos-phoethanolamine) is zwitterionic. Cationic peptides have the ability to distinguish between human cell membranes, which mainly contain lipids based on phosphocholine (PC), as opposed to bacterial cell membranes, which contain significant quantities of PG (phosphoglyercol) and PE (phosphoethanolamine) lipids. 13,14,18,19 Examining the interactions of the two peptides with these bacterial membrane models enables greater understanding of the mechanisms by which these peptides interact with specific membrane constituents of bacteria. Typically, more PG lipids are found in Gram-positive bacteria, whereas PE lipids are generally more abundant in Gramnegative bacteria. 20 We study the interaction of the two designed peptides with model lipid membranes with different ratios of PG/PE, to better understand antimicrobial activity. Specifically, we examine the interactions of the peptides with mixed lipids comprising ratios PG/PE and 3:1, 1:1, 1:3, and 1:0, in line with a previous study. 19 The combination of DPPG and DPPE was chosen due to the accessible melting temperature T m values, which are well separated for the two lipids, being reported to be 41 and 63°C, respectively. 21 Finally, we examine the antibacterial activity of these peptides against four species of bacteria, E. coli, S. aureus, Pseudomonas aeruginosa, and Pseudomonas syringae. Since minimal toxicity to human cells is an essential requirement of an effective peptide therapeutic, we also assay the cytocompatibility of these peptides using a human skin fibroblast model.  ACS Applied Bio Materials www.acsabm.org Article ■ RESULTS Self-Assembly. The critical aggregation concentration (cac) for both R 3 F 3 and R 4 F 4 was determined using 8anilinonaphthalene-1-sulfonic acid (ANS), which is used because its fluorescence changes within a hydrophobic environment. The cac values for R 3 F 3 ( Figure 1a) and R 4 F 4 ( Figure 2a) were found to be (0.32 ± 0.02) and (0.22 ± 0.03) wt % respectively. There is a significant difference in cac values, the cac being lower for R 4 F 4 than for R 3 F 3 , which suggests that the cac depends not just on the balance of hydrophobic and charged residues but is also influenced by the number of charged residues. To further assess the critical aggregation concentration, the amyloid-sensitive fluorophore thioflavin T (ThT) was used. Only R 3 F 3 was found to bind significantly to ThT, indicating that it forms amyloid-like fibers. As shown in Figure 1a, the cac of R 3 F 3 determined by ThT fluorescence assays was found to be (0.27 ± 0.03) wt %, which is consistent within uncertainty with that detected by ANS. This suggests that hydrophobic collapse occurs at the same concentration as the formation of amyloid structures by R 3 F 3 .
To examine the secondary structure of R 3 F 3 and R 4 F 4 , circular dichroism and Fourier transform infrared spectroscopy were conducted above the cac. The CD spectra are dominated by electronic transitions associated with the phenylalanine side chain. The main feature (Figures 1b and 2b) for all spectra is the peak at 220 nm which is due to phenylalanine π-stacking electronic transitions. 22 The molar ellipticity increases when the pH of solutions of R 3 F 3 and R 4 F 4 was adjusted to pH 7.
This indicates an increase in content of chiral structures. FTIR (Figures 1c and 2c) was used to further probe secondary structure. The data shows a peak at 1673 cm −1 , which can be assigned to vibrations of the TFA counterions bound to the peptide. 23 The peak at 1645 cm −1 , characteristic of a disordered secondary structure, is present only at native pH for both peptides. All spectra show a peak at 1608 cm −1 , which is characteristic of arginine side chains. 23,24 The peak at 1588 cm −1 can be assigned to the arginine side chain guanidinium group symmetric stretch. The 1456 cm −1 peak can be ascribed to −CH 2 bond vibrations. 25 Considering the data from both CD and FTIR spectroscopy, we conclude that at native pH, R 3 F 3 , and R 4 F 4 have a disordered secondary structure.
TEM was used to image the self-assembled structure of R 3 F 3 and R 4 F 4 ( Figure S1). Self-assembled structures were observed for both peptides, with R 3 F 3 forming a population of twisted nanotapes, that are several microns long, with a width between 50 and 120 nm. The presence of nanotapes agrees with the binding of fluorescent probe ThT. R 4 F 4 was found to assemble into a population of large nanosheets of differing sizes with widths between 342 and 1.7 μm. These non-amyloid structures are not expected to bind ThT, consistent with the fluorescence binding assays.
To further probe the self-assembled structures, SAXS measurements were performed (Figures 1d and 2d) in dilute solutions to provide form factors. These are analyzed to provide information on the shape and dimensions of selfassembled nanostructures. Data from peptide solutions at pH 7 show initial intensity decays with q −2 which indicates planar structures. The data was fitted to a bilayer Gaussian form factor model (previously used by our group to describe the density profile across a bilayer), 23,26 and a generalized Gaussian coil form factor to account for any monomers remaining in the solution (Tables S1 and S2) using the software SASfit. 27 This model fits the data very well. An estimate of the length of the peptide (assuming an antiparallel β-sheet structure) is the number of residues multiplied by 3.4 Å. 28 The approximate length of R 3 F 3 and R 4 F 4 is 20 and 27 Å, respectively. The bilayer thickness obtained from the SAXS fits corresponds approximately to these values implying that the nanosheets and twisted tapes are one molecule thick. This suggests an antiparallel arrangement leading to arginine-coated nanotapes/nanosheets (a nanosheet structure with this configuration was proposed for A 6 R). 12 From these measurements, we conclude that the peptides form a small population of bilayer structures, with a larger population of unaggregated peptides. Adjustment to pH 7 increases the population of these structures.
Interaction with Lipid Vesicles. To understand the antimicrobial activity of R 3 F 3 and R 4 F 4 discussed in the following section, which is due to interaction of the arginine residues with lipid headgroups, we examined whether there were any interactions between the peptides and the model DPPG/DPPE liposomes, a system commonly used as a bacterial membrane model. 10,19 SAXS was used to probe structure changes resulting from interaction of peptides with lipid membranes. We examined the ratios 1:0, 1:3, 1:1, and 3:1 of DPPG/DPPE. In accordance with the phase diagrams recorded previously, 19 the samples with mixtures 1:0, 1:1, and 3:1 DPPG/DPPE are expected to be in the disordered liquid crystal (L α ) phase at 20°C, whereas 1:3 DPPG/DPPE is expected to be in the L α /L β coexistence region at 20°C (where L β is the ordered gel phase). PE alone is not studied because of the shape of the lipid it forms inverted micelles with hexagonal symmetry. 14 Figure 3 shows SAXS intensity profiles obtained with ratios 1:0, 1:3, 1:1, and 3:1 of DPPG/DPPE at 0.5 wt % with and without R 3 F 3 and R 4 F 4 . For all ratios, the SAXS patterns for the liposomes of lipids alone displayed a broad scattering peak centered at q ∼ 0.1 Å −1 , which is associated with unilamellar vesicles. 13 Figure 3 also contains the SAXS data for samples with added peptide. A concentration of 0.08 wt % peptide was used (data being shown for an additional concentration of 0.25 wt % for R 4 F 4 with DPPG), since addition of 0.25 wt % peptide caused precipitation in many samples, as seen with previously studied peptides. 13,14 In the SAXS data for 1:0 DPPG/DPPE, 0.08 wt % R 3 F 3 , there is a small Bragg peak, with a corresponding domain spacing d = 45.9 Å and a distortion of the broad peak shape. This indicates some correlation between bilayers, that is, restructuring of a proportion of vesicles into multilamellar vesicles. Adding 0.08 wt % R 4 F 4 to DPPG leads to a change in the shape of the SAXS profile (broadening of the form factor extrema), and when the concentration of R 4 F 4 is increased to 0.25 wt %, two Bragg peaks with d = 43.4 and 14.6 Å are present. These peaks correspond to the firstand third-order reflections of a multilamellar structure, indicating that restructuring of most liposomes into multilamellar vesicles occurs in the presence of R 4 F 4 at this concentration, indicating peptide-induced lipid restructuring. R 3 F 3 was not examined at this concentration due to precipitation of the sample. For the lipid mixture 3:1 DPPG/DPPE, no Bragg peak was observed in the SAXS profiles in the presence of either peptide, indeed the form factor resembles that for the corresponding mixed liposome, with again only some broadening of form factor features (and a shift to higher q) particularly in the presence of R 4 F 4 . For R 3 F 3 combined with 1:1 DPPG:DPPE, there is again a broadening in the features in the SAXS curve. However, for R 4 F 4 , two Bragg peaks are noted with corresponding spacings d = 96.0 and 68.2 Å, which is approximately a 1: √2 ratio suggesting cubic symmetry, although the presence of two coexisting lamellar structures cannot be excluded in the absence of higher order Bragg reflections. This feature is not observed for R 3 F 3 .
Considering the SAXS data for the samples containing most DPPE (1:3 DPPG/DPPE), the liposome form factor is lost upon addition of either peptide, this being replaced with form factor profiles with a shape dominated at low q by scattering from the peptides alone (Figures 1d and 2d), especially for R 4 F 4 , although there is a Bragg peak for R 3 F 3 , showing some multilamellar vesicles are still present. Cryo-TEM images of samples prepared with lipids at this ratio show that R 3 F 3 causes a change in morphology of liposomes, seen by the appearance of rougher looking surfaces (Figure 4b). Cryo-TEM images of samples with this lipid composition in the presence of R 4 F 4 show that the peptide induces complete disruption of liposome formation, in agreement with conclusions from the SAXS data ( Figure 4c). The Cryo-TEM image for vesicles of DPPG ( Figure 4d) shows that vesicles are generally larger when exposed to R 3 F 3 with some fibers (presumably peptide) also present (Figure 4e), whereas with R 4 F 4 , the vesicles are more swollen and, moreover, are much more angular in shape and have a multilamellar structure (Figure 4f). This is consistent with the SAXS data shown in Figure 3a.
Circular dichroism spectroscopy is often also used to examine the interactions of lipid membranes with peptides. 13,14,29 In some instances, peptide secondary structure is shown to be influenced in the presence of lipid membranes which can induce the formation of more ordered secondary structure. This moreover gives information about whether the peptide interacts with the lipid as a monomer or as a selfassembled structure. Figure S2 shows the circular dichroism spectra for the mixtures of the two peptides with different composition lipid mixtures. Because of presence of the broad maximum around 220 nm in the spectra, it is concluded that the peptides interact with the lipids in a monomeric form.
Before investigating the potential antimicrobial activity of R 3 F 3 and R 4 F 4 , it is desirable to assay cytocompatibility, to determine the concentration range of peptides over which viability of human cells is retained. We therefore performed cytotoxicity assessments using the MTT assay on 161br human skin fibroblasts ( Figure S3). The IC 50 of R 4 F 4 was calculated to be 2.15 mM (or 2.64 mg/mL or 0.264 wt %). Interestingly, this is in the error range of the cac value determined for R 4 F 4 ( Figure 1a) implying that the self-assembly into nanosheet structures correlates to the onset of increased toxicity. The IC 50 of R 4 F 4 is higher than the previously studied [RF] 4 peptide, which has an IC 50 value of less than 1 mM, meaning that R 4 F 4 is less cytotoxic. Clearly, sequence has a significant effect on the cytotoxicity of these compositionally identical peptides. The IC 50 of R 3 F 3 was found to be 4.31 mM (4 mg/ mL or 0.4 wt %), the maximum concentration used in the concentration series and above the cac of this peptide. The finding that an increased number of residues leads to increased toxicity agrees with studies conducted with [RF] n repeating peptides, 16 although R 4 F 4 is significantly less cytotoxic than its compositional homologue RFRFRFRF. Having determined conditions where viability of human fibroblasts in the presence of these peptides is acceptable, we then studied their antimicrobial activity. Compounds that are active against bacteria have either inhibitory or bactericidal properties. 30 Inhibitory compounds are less likely to cause bacterial resistance but are more susceptible to the innate immune system and antibiotics. Bactericidal compounds are useful for fighting bacterial infections but can increase the resistance of bacteria. Both activities are valuable for different purposes. To examine whether R 3 F 3 and R 4 F 4 are bactericidal, the two peptides were inoculated with several strains of bacteria for 24 h. Figure 5 shows the antimicrobial activity of these peptides against Gram-positive S. aureus and two Gramnegative bacteria, one strain of P. aeruginosa (PA01) and one of E. coli (K12). A concentration of 0.05 wt % was selected, based upon the viability of human skin fibroblasts at this concentration, which was shown to be 77% and 81% for R 3 F 3 and R 4 F 4 , respectively ( Figure S3). The antimicrobial activity data indicate that R 3 F 3 shows no effect on any of these bacteria; however, R 4 F 4 causes a small, not statistically significant, reduction of 1 order of magnitude in colony forming units/milliliter (CFU/mL) of E. coli (K12) after 24 h. This effect, however, is not seen for the clinically relevant E. coli (0157) strain ( Figure S4). Both peptides showed no activity against S. aureus, in contrast to previously studied arginine-rich peptides, 11,14 although this could be due to the lower concentrations used. Interestingly, both peptides showed a strong antimicrobial effect against P. aeruginosa (Figure 5c), with a 3.2 log reduction in CFU/mL upon addition of R 3 F 3 , and complete death of the strain after 2 h on exposure to 0.05 wt % R 4 F 4 (Figure 5c). To investigate this further, a series of concentrations of R 3 F 3 ( Figure S5) and R 4 F 4 (Figure 5d) were examined to locate any MIC. The data indicate that R 3 F 3 does not have an MIC below 0.1 wt %, and R 4 F 4 has a MIC value between 0.045 and 0.05 wt % for P. aeruginosa.
To further understand the specificity of these peptides against Pseudomonas bacteria, the antibacterial activity of the peptides against four other species was examined ( Figure S6). Since R 4 F 4 is particularly active against Pseudomonas species, data were obtained for an additional three species. Data for Pseudomonas putida, Pseudomonas agarici, and Pseudomonas f luorescens are shown in Figure S6a−c, and this shows that all bacteria displayed an initial die off when exposed to R 4 F 4 , but interestingly, P. f luorescens shows some recovery in bacterial colony counts after 24 h, perhaps because the bacteria are able to metabolize the peptide, however further studies would be needed to understand this effect. There is also a significant (log 2.6) reduction in CFU/mL of plant pathogen P. syringae when exposed to R 3 F 3 and a complete bacterial kill when exposed to R 4 F 4 , similar to the effects seen with P. aeruginosa after 2 h of exposure. The MIC concentration is ∼0.001 wt % for P. syringae, which is approximately 50× lower than that for P. aeruginosa. Our data suggests that R 4 F 4 is selectively active against specific Pseudomonas species.
Finally, we examined whether our peptides can prevent biofilm formation of P. aeruginosa, which is a major problem in infections, in particular those of the respiratory tract. An important regulator of biofilm formation in P. aeruginosa is the second messenger signaling molecule bis(3′5′)-cyclic dimeric guanosine monophosphate (c-di-GMP). 31−33 High levels of cdi-GMP in P. aeruginosa are associated with the formation of biofilms, which leads to virulence and persistence of bacteria in hosts. Compounds that can reduce and prevent biofilm formation by targeting c-di-GMP are, therefore, attracting ACS Applied Bio Materials www.acsabm.org Article attention. We first examined whether R 3 F 3 and R 4 F 4 inhibit biofilm growth formation and then examined interactions of these peptides with c-di-GMP. Crystal violet was used to assay biofilm formation, this cationic dye binding to anionic polysaccharides in the biofilm. 34 The assay results ( Figure  6a) show that R 4 F 4 causes a reduction of biofilm formation in two strains of P. aeruginosa, PA01 and ATCC-12600. At a concentration 0.05 wt % of R 3 F 3 , we observe 84% survivability of biofilm ( Figure S7), whereas R 4 F 4 significantly reduces biofilm viability to 37% of P. aeruginosa strain PA01 and 33% for strain ATCC-12600 (Figure 6a). Figure 6b shows images of vials showing the Crystal violet taken up by biofilms formed at the air−water interface, after removal of the aqueous phase. The residual purple coloration of the vial surfaces is due to dye taken up by bacteria in planktonic form.
We used a series of biophysical techniques to observe the interaction of both peptides with ci-d-GMP, including SAXS, CD, and TEM. At concentrations of 100 μM or more, c-di-GMP exists in equilibrium with its homodimer. 33 Interestingly, the CD spectra of R 4 F 4 and R 3 F 3 show that c-di-GMP causes a strong increase in molar ellipticity ( Figure S8a) at a 255 nm maximum for R 3 F 3 and a slight red-shifted maximum of R 4 F 4 of 259 nm and an increase in negative ellipticity at the minimum at 281 nm for R 3 F 3 and 284 nm for R 4 F 4 , which are due to the stacking of the four guanine rings present in dimers of c-di-GMP, showing that addition of the peptides stabilizes the homodimer. 33 The red shifting of peak positions for R 4 F 4 is evidence of altered π-stacking in the complex between c-di-GMP and R 4 F 4 , which is not observed for the c-di-GMP complex with R 3 F 3 . TEM images for c-di-GMP with R 3 F 3 show the formation of nanosheets ( Figure S8c). In contrast, the TEM image for c-di-GMP with R 4 F 4 shows the formation of a fibrous network structure ( Figure S8e). Differences in structure also change the slope of the SAXS profiles in the low q region ( Figure S8b,d), I ∼ q −n . The slope of the low q intensity for cdi-GMP alone is n = 0.8 consistent with locally extended structures. Upon addition of R 3 F 3 , the slope changes to n = 1.8 which suggests the formation of layered structures, such as bilayers as seen in the TEM images. The peptide alone has a SAXS profile with n = 3.5. For R 4 F 4 , the slope changes to n = 4.2, suggesting the presence of compact globular structures.
Congo red staining can be used to quantify the amount of polysaccharides produced by bacteria. 35 This has been used as an indirect measurement of c-di-GMP levels since these are correlated to polysaccharide expression. 35 Congo red binds to polysaccharides resulting in a decreased optical density (OD) when polysaccharide matrices are formed by the bacteria. 35 When adding R 4 F 4 to P. aeruginosa, we observe a decrease in optical density of the supernatant (Figure 6c). This indicates enhanced biofilm growth, however, it was observed that, from 0.00613 to 0.025 wt %, there was a precipitate at the bottom of the tube, and some film growth occurred at the interface of the tube between the air and the solution (Figure 6d). P. aeruginosa is a motile organism, therefore is expected to form a biofilm at the interface. 36 At 0.05 wt % (0.5 mg/mL), a concentration that is the above the MIC in water and for which 37% biofilm viability is observed (Figure 6a), there is no biomass on the side of the tube, but there is a deposit on the bottom of the tube (Figure 6d). P. aeruginosa has a mucoidal phenotype, which means it can form a polysaccharide capsule around itself in response to stressed conditions. 37 Therefore, we conclude that the reduction in the OD is due to stressed bacteria producing polysaccharides as part of the stress response.

■ CONCLUSIONS
The two "blocky" surfactant-like peptides R 3 F 3 and R 4 F 4 are both able to self-assemble in aqueous solution at concentrations in excess of critical values, obtained from fluorescence probe assays. The cac = 0.3 wt % (≡ 3.2 mM) for R 3 F 3 is higher than that previously obtained for the compositional homologue peptide RFRFRF, for which the cac = 2.3 mM. 16 On the other hand, the cac = 0.22 wt % (≡ 1.8 mM) for R 4 F 4 is significantly higher than that previously reported for the compositional homologue RFRFRFRF, for which the cac = 0.34 mM. 16 This shows a significant influence of sequence on aggregation propensity. In addition, the nanostructure that selfassembles is distinct, since R 3 F 3 forms a population of twisted nanotape structures whereas RFRFRF adopts a globular morphology. In the case of the longer peptides, nanosheets are formed by R 4 F 4 , whereas RFRFRFRF adopts a nanofiber morphology. The self-assembly of these peptides is likely driven by π-stacking interactions. This is supported by the observation in the CD spectra that the molar ellipticity of the 220 nm peak due to phenylalanine π-stacking increases from native pH to pH 7, as shown in Figures 1b and 2b.
The CD and FTIR spectra measured for R 3 F 3 and R 4 F 4 show that, even above the cac, these peptides do not adopt well-defined secondary structures, although there are πstacking interactions. Furthermore, the SAXS intensity profiles suggest a significant component of unaggregated peptides in solution, also above the cac. Combined with the relatively high cac values (compared to typical surfactant-like peptides, such as A 9 R 38 or lipidated peptides), this indicates that R 3 F 3 and R 4 F 4 have a weak aggregation propensity. This may be beneficial in terms of the observed antimicrobial activity, which is observed at concentrations an order of magnitude below the cac.
We found that both peptides have antimicrobial activity against Pseudomonas species, with R 4 F 4 particularly active, killing all colony forming units of P. aeruginosa in planktonic form at 0.05 wt % (0.5 mg/mL). At this concentration, the cytotoxicity to human cells (fibroblast model) is relatively low. This peptide has a MIC of 0.045 wt % (0.45 mg/mL, 0.37 mM). This peptide also showed activity against other Pseudomonas species, in particular P. syringae, which is a widespread plant pathogen. We propose that the antimicrobial activity against Pseudomonas in planktonic form may be due to disruption of the bacterial lipid membrane, which contains ∼21% PG lipids and 60% PE lipids (the remainder being cardiolipin). 39,40 This 1:3 ratio of PG: PE corresponds to the composition where model vesicle membrane disruption is observed by SAXS and cryo-TEM (Figures 3 and 4) and this composition of vesicle has been used in prior studies of cationic antimicrobial peptide activity. 40 We were then led to further examine the activity of lead candidate peptide R 4 F 4 against Pseudomonas biofilms and a significant reduction in biofilm viability was observed through Crystal violet assays although biofilm formation was not completely inhibited. Since c-di-GMP is known to be an important signaling molecule involved in Pseudomonas biofilm formation, we then examined the interaction of the two peptides with this nucleotide. Circular dichroism confirmed an interaction with the peptides which stabilizes the homodimer form of c-di-GMP and in the case of R 4 F 4 , modification of π-ACS Applied Bio Materials www.acsabm.org Article stacking interactions was observed. These observations point to a possible role of these arginine-rich peptides as binders of c-di-GMP, and this is a possible mechanism of biofilm inhibition. The arginine guanidinium group is known to bind c-di-GMP through electrostatic interactions and hydrogen bonding, and this was further probed using Congo red staining of polysaccharides, which were overexpressed by bacteria exposed to the antimicrobial activity of R 4 F 4 .
In summary, R 4 F 4 is a promising candidate peptide with pronounced antimicrobial activity against Pseudomonas species. Although R 3 F 3 has some activity against Pseudomonas aeruginosa, the increased molecular weight of R 4 F 4 seems to confer higher activity. It would be expected that longer diblock peptides would potentially have higher antimicrobial activity, due to the increased cationic charge. However, this increased charge might result in increased toxicity to human cells, thus future work could focus on finding an optimum length of peptide with a balance of cytocompatibility, and antimicrobial activity. Peptide R 4 F 4 is more active against Pseudomonas species than RA 6 R and RA 9 R studied in our previous work. 28 Also promising is the fact that the specific activity against Pseudomonas species is likely to slow antimicrobial resistance to the peptide because wide spectrum actives often develop resistance faster.
Further work could explore chemical modification to enhance the activity of R 4 F 4 against biofilm formation and cdi-GMP binding capacity, along with other modifications, such as insertion of protecting groups/lipidation/use of D-amino acids or other non-natural residues, to enhance stability in vivo for potential future application.

■ MATERIALS AND METHODS
Materials. R 3 F 3 and R 4 F 4 were synthesized by Peptide Synthetics (Fareham, UK). Purity was determined by HPLC, and for R 3 F 3 was found to be 97.83%, the molar mass from ESI-MS was M w = 928.089 g mol −1 . The purity of R 4 F 4 was 97.98%, and the molar mass from ESI-MS was M w = 1231.452 g mol −1 .
Sample Preparation. Samples of R 3 F 3 and R 4 F 4 were dissolved in water. The native pH was found to be 2.22 and 2.28, respectively. Adjustment of pH to pH 7.0 was done using 0.5 M NaOH.
Fluorescence Assays. The critical aggregation concentrations (cac) of R 3 F 3 and R 4 F 4 were determined using fluorescence spectroscopy. Fluorescence spectra were recorded with a Varian Cary Eclipse fluorescence spectrometer with samples in 4 mm inner width quartz cuvettes. ANS (8-anilinonaphthalene-1-sulfonic acid) was used to probe the aggregation as a probe that is sensitive to hydrophobic environments making it suitable to locate the cac. 41 ANS assays were performed using a 66.8 μM ANS solution (1 mg of ANS dissolved in 50 mL water and sonicated for 30 min) to solubilize R 3 F 3 and R 4 F 4 . Fluorescence spectra were recorded between 400 and 650 nm (λ ex = 356 nm). Additionally, ThT was used to determine the cac of R 3 F 3 , as it is sensitive to the formation of amyloid fibrils at the cac. 42 ThT assays were performed using 5.0 × 10 −3 wt % solution to solubilize the peptide. Spectra were recorded between 460 and 600 nm (λ ex = 440 nm).
Circular Dichroism (CD). CD spectra were recorded using a Chirascan spectropolarimeter (Applied Photophysics, UK) in the wavelength range 180−260 nm (or 340 nm for samples with c-di-GMP). Samples 1 wt % of R 3 F 3 and R 4 F 4 in pure H 2 O at native pH and pH 7 were pipetted into a quartz plate cell with 0.1 mm path length. Data with absorbance less than 2 only are reported. Measurements were recorded with a 0.5 nm bandwidth, 1 mm step and 1 s collection time per point. The CD signal for the background solution was subtracted from the CD signal of the sample, and molar ellipticity was calculated.
Fourier Transform Infrared Spectroscopy (FTIR). Spectra were recorded using a Thermo Scientific Nicolet iS5 instrument equipped with a DTGS detector, with a Specac Pearl liquid cell (sample contained between fixed CaF 2 plates). Aliquots 80 μL of 1 wt % R 3 F 3 and R 4 F 4 at native and pH 12 dissolved in D 2 O were prepared and added into the liquid cell. Spectra were scanned 128 times over the range 900−4000 cm −1 Transmission Electron Microscopy (TEM). Imaging was performed using a JEOL 2100Plus TEM microscope operated at 200 kV. Droplets of sample were placed on Cu grids coated with a carbon film (Agar Scientific, UK), stained with uranyl acetate (1 wt %) (Sigma-Aldrich, UK) and dried. Specimens were then loaded into the microscope.
Cryo-TEM. Vitrified specimens were prepared using an automated FEI Vitrobot device using Quantifoil 3.5/1 holey carbon copper grids with a hole size of 3.5 μm. Prior to use, grids were plasma cleaned using a Gatan Solarus 9500 plasma cleaner and then transferred into the environmental chamber of a FEI Vitrobot at room temperature and 100% humidity. Thereafter, sample solution was applied onto the grid, and it was blotted twice for 5 s and then vitrified in a 1/1 mixture of liquid ethane and propane at a temperature −180°C. The grids with vitrified sample solution were maintained at liquid nitrogen temperature and then cryo-transferred to the microscope. Imaging was carried out using a field emission cryo-electron microscope (JEOL JEM-3200FSC) operating at 200 kV. Images were taken in bright field mode and using zero loss energy filtering (Ω type) with a slit width of 20 eV. Micrographs were recorded using a Gatan Ultrascan 4000 CCD camera. Specimen temperature was maintained at −187°C during the imaging.
Liposome Preparation. Liposomes were prepared using the thinlayer hydration method. Weighed quantities of DPPG and DPPE were prepared by dissolution in chloroform, and thin lipid films were prepared by drying the lipids under a stream of nitrogen gas. Films were then placed under vacuum for 2 h to remove any remaining organic solvent. After this, lipids were resuspended in water to a final concentration of 0.5 wt % lipid, heated above their T m and vortexed for 5 min. Liposome mixtures were then left to equilibrate before experiments. DPPG/DPPE vesicles were prepared using this method at different molar ratios of 1:3, 1:1, 3:1, and 1:0 DPPG/DPPE. To obtain peptide/lipid mixtures, resuspended lipids were added to peptide powders to make up a 0.08 wt % concentration of peptide, heated above the T m and vortexed for 5 min, then left to equilibrate.
Small-Angle X-ray Scattering. Solution small-angle X-ray scattering (SAXS) data was collected on the bioSAXS beamline B21, at Diamond Light Source, Harwell, United Kingdom and beamline ID02 of the European Synchrotron Radiation Facility (ESRF), Grenoble France. On beamline B21, samples of 1 wt % R 3 F 3 and 1 wt % R 4 F 4 were loaded into PCR tubes in an automated sample changer. The samples were then delivered into a temperaturecontrolled quartz capillary and exposed for 1s, acquiring 30 frames at 20°C. Data was collected using a Pilatus Dectris 2 M detector with a 3.9 m sample−detector distance and X-ray wavelength λ = 1.0 Å. Background was manually subtracted using ScÅtter. Form factor modeling was done using SASfit.
At ID02, samples were loaded into a glass capillary (internal radius = 2 mm) using a syringe. The beamline operated with an X-ray wavelength λ = 1.0 Å. 2D data was collected using a Rayonix MX170 detector, with a 1.474 m sample−detector distance.
Cell Viability Assays. The cytotoxicity of both R 3 F 3 and R 4 F 4 was examined. In vitro cell culture was conducted using 161Br (European Collection of Authenticated Cell Cultures, ECACC) cells, a human skin fibroblast cell line. Cells were cultured in EMEM (Eagle's minimum essential medium), with 2 mM glutamine, enriched with 15% fetal bovine serum (FBS) and 1% nonessential amino acids (NEAA). Cells were maintained in a humidified atmosphere of 37°C, 5% CO 2 .
Potential cytotoxicity effects were examined using the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay. The peptides were dissolved in complete media. Cells were seeded into a 96-well plate at 4 × 10 4 cells/mL and allowed to adhere for 24 ACS Applied Bio Materials www.acsabm.org Article h in 100 μL complete medium. After this, 100 μL of either complete media and/or peptide solution was added, to give either control solution, or solutions of peptide between the concentrations of 0.0625−4 mg/mL. Cells were incubated for 67 h. After this, 20 μL MTT (5 mg/mL, in PBS) was added to each well plate and allowed to incubate for 5 h. After a further 5 h (72 h total), the solution was removed from the wells and replaced with 100 μL of DMSO per well, which dissolves the formazan crystals. Plates were incubated for 30 min, and then analyzed using a UV microplate reader (λ = 570 nm). Results are reported as a percentage cell viability compared to control (untreated) values, and fitted to a sigmoidal function using Origin Lab as a guide to the eye.
Antimicrobial Studies. The antimicrobial assays were performed with three types of bacteria which cause disease in humans, two being Gram-negative, Escherichia coli (K12) and Pseudomonas aeruginosa (PA01), and one being Gram-positive, Staphylococcus aureus (8325-4). Plant bacteria species P. syringae, P. putida, P. agarici, and P. fluorcences were also used to examine specificity against Pseudomonas species. Stock cultures were stored in −80°C in 7% (v/v) DMSO. Prior to experiments, samples of all human bacteria were streaked out onto LB (Lysogeny Broth) agar and grown overnight at 37°C, whereas the plant bacteria were grown overnight at 27°C.
Bacterial Survival Assays. To examine whether the peptides were bactericidal, one colony from streaked plates was transferred into 7 mL of sterile lysogeny broth (LB), and grown overnight at 37°C under agitation at 150 rev/min, on an orbital shaker, and these cultures were used for ongoing experiments. Cultures were then transferred into a 15 mL Falcon tube and cells were harvested by centrifugation at 9000 rpm at 4°C for 10 min. The supernatant was discarded, and the pellet was resuspended in 1.5 mL of ice-chilled PBS (phosphate buffered saline).
After this, 30 μL of this solution was transferred into 300 μL of 0.5 mg/mL (or 0.05 wt %) of R 3 F 3 or R 4 F 4, in sterile water or control solutions of 300 μL of sterile water. Solutions were then vortexed for 5 s, and three 20 μL samples were taken at times 0, 120, and 1440 min. These samples were then serially diluted in PBS, and 10 μL of each dilution was platted onto LB agar and incubated at 37°C overnight before colony counting.
Biofilm Assays. To form a biofilm, Pseudomonas aeruginosa PA01 and ATCC2363 strains were grown overnight in LB media. Following this, the culture was diluted 1:100 in M63 minimal medium supplemented with arginine and magnesium sulfate, with and without R 3 F 3 or R 4 F 4 , with four technical replicates. After this the plate was incubated for 4−24 h.
Following incubation, the supernatant was discarded and then gently washed in water several times to decrease background staining. After this, 125 μL of 0.1 wt % crystal violet solution was added to each well of the microtiter plate and incubated for 10 min. Following this, the plate was washed several times and left to dry overnight.
Next, 125 μL of 30% acetic acid in water was added, to solubilize the biofilm, and the plate was incubated for further 10 min to allow solubilization. Absorbance was measured using a plate reader at 550 nm. Anova and Tukey were used to assess statistical significance.
Indirect Measurements of c-di-GMP Using Congo Red. Pseudomonas aeruginosa PA01 was grown overnight in LB media. After this, the culture was diluted 1:100 in 1 mL of M63 minimal medium supplemented with 0.4% L-arginine and 1 mM magnesium sulfate, with 40 μg/mL Congo red with and without R 4 F 4. Control solutions were also prepared using the described medium above with no bacterial cells present. Congo red provides an indirect measure of c-di-GMP levels as it binds to the matrix of the biofilm, production of which is correlated to c-di-GMP levels. 35 Samples were shaken at 200 rpm overnight in a shaking incubator. After this, 200 μL of supernatant was added in triplicate to a 96 well place, and measured using a plate reader at 490 nm. Results are reported as (control solution OD − bacteria containing solution OD). Student's t tests were used to assess significance.
SAXS fit parameters, TEM images, CD spectra of peptides with DPPG, cytotoxicity profiles, additional antimicrobial and biofilm activity assay results, and peptide binding with c-di-GMP data (PDF)