An acidic model pro-peptide affects the secondary structure, membrane interactions and antimicrobial activity of a crotalicidin fragment

In order to study how acidic pro-peptides inhibit the antimicrobial activity of antimicrobial peptides, we introduce a simple model system, consisting of a 19 amino-acid long antimicrobial peptide, and an N-terminally attached, 10 amino-acid long acidic model pro-peptide. The antimicrobial peptide is a fragment of the crotalicidin peptide, a member of the cathelidin family, from rattlesnake venom. The model pro-peptide is a deca (glutamic acid). Attachment of the model pro-peptide only leads to a moderately large reduction in the binding to- and induced leakage of model liposomes, while the antimicrobial activity of the crotalicidin fragment is completely inhibited by attaching the model pro-peptide. Attaching the pro-peptide induces a conformational change to a more helical conformation, while there are no signs of intra- or intermolecular peptide complexation. We conclude that inhibition of antimicrobial activity by the model pro-peptide might be related to a conformational change induced by the pro-peptide domain, and that additional effects beyond induced changes in membrane activity must also be involved.

SCientiFiC RepoRts | (2018) 8:11127 | DOI: 10.1038/s41598-018-29444-0 intramolecular interaction of the pro-domain with the antimicrobial peptide domain, but again, no structures have been presented yet from which this is apparent.
Inactivation of antimicrobial and cytotoxic activity are by no means the only possible functions of pro-domains. For example, Ganz and co-workers 11 have shown that the acidic pro-peptide is necessary for the precise sub-cellular trafficking and sorting of proHNP1. The role of pro-peptide domains for cathelicidins is less clear than for the defensins. Cathelicidins 3,12,13 are characterized by their highly conserved cathelin-like domain (CLD), a small folded domain with cystatin-like fold 14 , which is followed by a highly variable domain containing the antimicrobial peptide. One group has reported that the full-length precursor protein for the sole human cathelicidin, LL-37 (known as hCAP18) has no antimicrobial activity 15 , but a more recent report suggests that the antimicrobial activity of hCAP18 is practically identical to that of LL-37 16 . Hence, at this stage it is not yet clear whether for LL-37, its pro-sequence acts as an inhibitor of its antimicrobial activity or not.
We find that attaching the E 10 model pro-peptide leads to a significant change in the secondary structure of the peptide in solution and to a full inactivation of its antimicrobial activity. Surprisingly however, physical properties typically associated with antimicrobial activity, viz. adsorption to lipid bilayers and the induction of liposome leakage are only moderately affected.
Galleria mellonella in vivo toxicity assay. An in vivo toxicity assay was performed using Galleria mellonella 25 . G. mellonella larvae in their final instar stage were purchased (UK Waxworms Ltd, Sheffield, UK), and stored in the dark at 15 °C, and used within 14 days. Larvae were separated by weight, and only larvae between 0.2 and 0.3 g were used for experiments. Larvae were injected with 20 μL of peptide solutions, or controls in the left posterior proleg using Terumo Myjector 29 G insulin syringes (VWR International). The syringes were changed between different treatments. Two negative control groups were included in every experiment; one group was not injected to control for background larval mortality (no manipulation control) and the other group (uninfected control) was injected with PBS to control for the possible effect of physical trauma on mortality. After injection, larvae were stored in Petri dishes in the dark at 37 °C with 5% CO 2 for up to 144 h post-infection (p.i.) and inspected every 24 h for survival; larvae were considered dead if they did not move after shaking of the petri dish. For each sample (non-manipulated control, water control, peptides) fifteen randomly chosen larvae were used. The peptide concentration was 10 mg per kg of body weight.
Liposome leakage assay. Following earlier work 26 , lipids DOPC and DOPG were dissolved in chloroform at a concentration of 25 mg.mL −1 , mixed in a molar ratio of 7:3, and subsequently diluted further to 10 mg.mL −1 using chloroform. The chloroform was evaporated using rotary evaporation (350 mbar, 313 K, 100 rpm). The lipid film was then dried in vacuum for at least 2 h after which the lipids were re-suspended in calcein-containing buffer (70 mM calcein in 10 mM Tris-HCl, pH 7.5) to get a 30 mM lipid suspension by hydration for 1 h in a rotary evaporator (no vacuum, 323 K, 100 rpm). Multilamellar vesicles thus formed were freeze-thawed four times using liquid nitrogen and a 310 K water bath to get unilamellar vesicles. A mini-extruder (Avanti Lipids) equipped with a 200 nm pore size polycarbonate membrane was used to perform 21 extrusions to homogenize the size of the lipid vesicles. Vesicles were separated from free calcein on a gravity driven Sephadex G-50 size exclusion column and eluted with a 10 mM Tris-HCl buffer containing 100 mM NaCl (pH = 7.5). The vesicles were characterized using dynamic light scattering, for which an ALV instrument equipped with an ALV5000/60 × 0 external correlator and a 300 mW Cobolt Samba-300 DPSS laser operating at a wavelength (λ) of 532 nm was applied. A cumulant analysis showed the vesicles to be monodisperse, and to have an average diameter of 190 ± 5 nm. The degree of dilution of the vesicles during size exclusion chromatography was estimated by comparing the intensity of scattered light (count rate) before and after size-exclusion chromatography. For this scattering angles θ ranging from 20° to 140° were used with steps of 5°, five measurements of 30 s were recorded for each scattering angle and the average was taken over all measurements. A fluorescence filter was applied that only transmits the laser light, to prevent the fluorescence light emitted by calcein from disturbing the measurements. Vesicles used for the leakage assay were diluted with a 10 mM Tris-HCl buffer containing 100 mM NaCl (pH = 7.5) to a final lipid concentration of 50 μM. Peptides were dissolved in the same buffer, at a concentration of 6.4 mg.mL −1 . The calcein leakage caused by the AMPs was measured by following the fluorescence intensity of the liposome solutions over time in a Cary Eclipse cuvette fluorimeter using an excitation wavelength of 490 nm and an emission wavelength of 520 nm. The widths of the excitation and emission slits were 2.5 nm. The fluorescence measurements were started with 588 μL of the vesicle solution without AMPs. After two minutes the fluorescence intensity signal was stable and 12 μL of 6.4 mg.mL −1 AMP solution was added, to achieve a final peptide concentration of 128 μg.mL −1 , equal to the highest peptide concentration used in the MIC assays. In total, the measurements lasted 10 minutes. At the end of the experiments, 10 μL of a 10% (v/v) Triton X-100 solution was added, which is a surfactant that disrupts the liposomes. The signal after Triton addition was taken to correspond to 100% liposome leakage. For all peptides, the liposome leakage assays was repeated four times with freshly prepared liposomes and results were found to be reproducible. Reflectometry of peptides adsorbing on supported lipid bilayers. Adsorption measurements were performed using an optical reflectometer with a stagnation-point flow cell and an oxidized silicon surface as a substrate, as described by Dijt et al. 27 . The laser beam reflects from the surface at exactly the point where the jet flow reaches the surface. The reflectometer signal can be directly converted into an adsorbed amount (Γ) if the refractive index increment of the adsorbing species is known 27,28 . The refractive index increment for lipids is dn/dc ≈ 0.145 μg-mL −1 29,30 that of proteins is dn/dc ≈ 0.185 μg-mL −1 31 Clean oxidized silicon strips (with a silica layer of ~62 nm) were prepared as follows. Silica strips were washed with demiwater, sonicated for 1 min in 2-propanol, washed in milliQ water, dried under a nitrogen stream and plasma cleaned for 2 min before being stored in milliQ water. The strips were used in the experiment within 5 h of cleaning. A supported lipid bilayer (SLB) was formed on top of the silica surface by means of vesicle adsorption and subsequent rupture and SLB formation at high surface coverage 32 . Vesicles were prepared as described for the liposome leakage assay, except for the fact that no calcein was used. For preparing the SLB, vesicles were diluted in 10 mM Tris, 100 mM NaCl, pH 4 to a final concentration of 200 μM of lipids. To promote adsorption of the negatively charged vesicles, the silica substrates were first equilibrated against a low pH buffer (10 mM Tris, 100 mM NaCl, pH 4). At this low pH, the silica is less negatively charged as opposed to neutral pH. After a steady baseline was obtained, the flow was switched to the 200 μM vesicle solution for a few minutes, until the signal stabilized again. Next, the flow was switched back to the low pH buffer and rinsed for some minutes until the signal stabilized again and finally, flow was switch to a neutral pH buffer (10 mM Tris, 100 mM NaCl, pH 7.5). This finalized the procedure to attach the SLB. Next, flow was switched to a solution of peptides at a concentration of 16 μg-mL −1 in 10 mM Tris 100 mM NaCl, pH 7.5. Reproducibility of these measurements was typically found to be around 10%.
Antimicrobial activity and toxicity. All three peptides were tested against different bacterial strains, both Gram negative and Gram positive, in order to determine their in vitro antibacterial activity: E. coli ATCC 25922, K. pneumoniae ATCC 13883 and S. aureus ATCC 25923. Additionally, we considered Gram-negative clinical isolates from KPC positive E. coli (KPC + 001812446) and K. pneumoniae (KPC + 001825971) due the medical relevance related to KPC strains by their resistance to conventional antibiotics. MIC and MBC results are given in Table 2.
Interaction of the peptides with model membranes. The peptide physical-chemical properties that have been suggested to correlate most strongly with antimicrobial activity are their affinity towards model membranes, and their ability to induce leakage in model liposomes. Therefore, we have tested whether the E 10   model pro-peptide prevents adsorption of the peptides on model membranes, and whether it prevents leakage from model liposomes. As a model for the predominantly negatively charged bacterial membranes, we use a 7:3 mixture (molar ratio) of the lipids DOPC and DOPG. Liposomes were prepared with an average diameter of 190 ± 5 nm, as determined using DLS. In a first experiment, supported lipid bilayers (SLB) were prepared on a silica wafer by adsorbing the liposomes to the silica at low pH. Subsequently, the amount of peptide (mg/m 2 ) adsorbing to the SLB was monitored using reflectometry, as a function of time (Fig. 6). Note that in these experiments, we have also included the full-length crotalicidin peptide (Ctn). The reflectometry experiments demonstrated that, under the conditions of the experiment (10 mM Tris 100 mM NaCl, pH 7.5, and a peptide concentration of 16 μg.mL −1 ) both full length Ctn and the fragment Ctn [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] adsorb to the model membranes at the same surface density of approximately 0.5 mg.m −2 . Attaching the (GS) 4 sequence reduces the adsorbed amount to approximately 0.3 mg.m −2 . Attaching the E 10 model pro-peptide sequence leads to a further reduction of the adsorbed but the E 10 model pro-peptide sequence by no means completely prevents adsorption: the adsorbed amount remains significant at approximately 0.15 mg.m −2 .
Both experiments on model membranes therefore present results that do not correlate precisely with the observed antimicrobial activities: the addition of the (GS) 4 leads to a significantly lower adsorption, but leakage and antimicrobial activity are not affected. Most strikingly, the addition of the E 10 model pro-peptide leads to a complete inactivation of the antimicrobial activities, but only to a somewhat lower adsorption on model membranes, and to a only a somewhat lower induced leakage of model liposomes, in contrast to other studies 8 where the presence of a pro-peptide was shown to lead not only to inhibition of antimicrobial activity, but also to a much lower adsorption on-and leakage of model liposomes.
In summary, our peptide model for the inhibition of antimicrobial activity by acidic pro-peptides partly calls into question mechanisms for the inhibition suggested earlier (charge neutralization, inter-and intramolecular complexation, leading to loss of membrane activity and hence loss of antimicrobial activity). We suggest that conformational changes induced by attaching the pro-peptides may be an additional factor determining the loss of antimicrobial activity.