Edited by: John D. Wade, Florey Institute of Neuroscience and Mental Health, Australia
Reviewed by: Thorsten Wohland, National University of Singapore, Singapore; Stefan W. Vetter, North Dakota State University, United States
This article was submitted to Chemical Biology, a section of the journal Frontiers in Chemistry
†These authors have contributed equally to this work
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
The lipid composition of the cellular membrane plays an important role in a number of biological processes including the binding of membrane-active peptides. Characterization of membrane binding remains challenging, due to the technical limitations associated with the use of standard biophysical techniques and available membrane models. Here, we investigate the lipid binding properties of two membrane-active peptides, VSTx1, a well characterized ion-channel inhibitor, identified from spider venom, that preferentially binds to anionic lipid mixtures, and AA139 an antimicrobial β-hairpin peptide with uncharacterised lipid binding properties, currently in pre-clinical development. The lipid binding properties of these peptides are elucidated using nanodiscs formed by both linear and circularized (sortase-mediated) forms of a membrane scaffold protein (MSP1D1ΔH5). We find that nanodiscs formed by circularized MSPs—in contrast to those formed by linear MSPs—are sufficiently stable under sample conditions typically used for biophysical measurements (including lipid composition, a range of buffers, temperatures and concentrations). Using these circularized nanodiscs, we are able to extract detailed thermodynamic data using isothermal titration calorimetry (ITC) as well as atomic resolution mapping of the lipid binding interfaces of our isotope labeled peptides using solution-state, heteronuclear, nuclear magnetic resonance (NMR) spectroscopy. This represents a novel and general approach for elucidating the thermodynamics and molecular interface of membrane-active peptides toward flat lipid bilayers of variable composition. Our approach is validated by first determining the thermodynamic parameters and binding interface of VSTx1 toward the lipid bilayer, which shows good agreement with previous studies using lipid micelles and liposomes. The method is then applied to AA139, where the membrane binding properties are unknown. This characterization, involved solving the high-resolution structure of AA139 in solution using NMR spectroscopy and the development of a suitable expression system for isotope labeling. AA139 was found to bind exclusively to anionic membranes with moderate affinity (
The composition of the lipid bilayer can have a significant impact on a number of biological processes including the trafficking of soluble proteins, the structure, dynamics and function of integral membrane proteins and the action of membrane-active peptides (Escriba et al.,
Lipid bilayer nanodiscs (ND) have been developed to solubilize and reconstitute membrane proteins in lipid bilayers, and their use is rapidly expanding (Bayburt et al.,
A significant advantage of (c)NDs is in their modularity. Different compositions of synthetic phospholipids can be encapsulated, and these soluble discs can then be studied under a range of solution conditions (pH, temperature, salt etc.). It is often possible to incorporate specific lipid compositions and obtain homogeneous NDs (Lee et al.,
The stability of the NDs were evaluated by size-exclusion chromatography (SEC), electron microscopy (EM) and mass spectrometry (MS), under different storage conditions with variable buffer (pH), temperature and concentration. While non-circularized NDs containing anionic lipids were shown to be poorly stable, excellent stability was found when using the cNDs. The latter was then used for solution-state biophysical studies to investigate the lipid binding interactions of two peptides that are known to exert their function in anionic lipid bilayer environments.
To determine the binding of these membrane-active peptides to model membranes of Gram-negative bacteria in NDs, we first studied the well characterized spider toxin, VSTx1, isolated from the venom of
Here, we first measured the thermodynamics of the binding of the peptide against cNDs using isothermal titrations calorimetry (ITC) experiments. We then mapped the lipid binding interface of the peptide using chemical shift mapping experiments by analysis of, 2D 1H-15N-HSQC, solution-state NMR experiments. Our results show very weak binding of the peptide to POPC bilayers in cNDs–much weaker than that observed when using DHPC micelles. In contrast, the peptide binds very strongly to anionic cNDs, consistent with the previous liposome experiments, suggesting that the binding of the peptide to micelles is different than to bilayers.
Next, the described approach was applied to an antimicrobial peptide, AA139, currently undergoing preclinical trials for the treatment of Gram-negative bacterial infections. AA139 is an analog of arenicin-3, a peptide antibiotic that was originally identified as part of a group of broad-spectrum antimicrobial peptides isolated from the lugworm
The presented approach provides a platform for measurement of both thermodynamic and structural data for membrane-active peptides using a planar bilayer system, using standard NMR experiments. The flexibility and stability of the cNDs as a model system promises to improve our understanding of this important class of molecules.
The construct for the evolved pentamutant of Sortase A in a pET29 vector was a gift from Prof. David R. Liu's laboratory (Harvard University). Expression and purification of the pentamutant Sortase A from
Isotopically single labeled 15N-VSTx1 was expressed and purified as previously described (Lau et al.,
1D 1H NMR experiments were recorded at different pHs and at pH 3.3 all non-exchangeable backbone amide protons could be observed. The spectra at pH 3.3 and 6.5 were superimposable indicating that the structure is unaffected by the change in pH. Subsequently, 2.5 mM synthetic AA139 (provided by Adenium) was analyzed in 20 mM phosphate buffer at pH 3.3 containing 5% D2O. All NMR experiments were performed on a Bruker Avance III spectrometer equipped with a cryogenically cooled triple resonance probe operating at a nominal 1H frequency of 700 or 900 MHz. The excitation sculpting sequence was used to suppress the solvent (H2O) resonance. Two-dimensional TOCSY [tm (MLEV17 spin-lock mixing pulses) = 80 ms], NOESY [tm (mixing time) = 300 ms], 15N-HSQC and 13C-HSQC were recorded at 25°C. Chemical shifts were directly (for 1H) or indirectly (for 13C, 15N) referenced relative to the 2, 2-dimethylsilapentane-5-sulfonic acid (DSS) signal at 0 ppm. The assignment of proton resonances was carried out using TOCSY and NOESY data using the CCPNMR software (Skinner et al.,
All NMR spectra for the hydrogen-deuterium exchange studies were recorded on the spectrometer described above at 25°C. Spectra were referenced to DSS at 0 ppm. Lyophilized peptide was initially solubilized in 20 mM phosphate buffer at pH 3.3. The peptide was then lyophilized and subsequently dissolved in 100% D2O followed by immediate transfer to the spectrometer for measurement. 1D 1H and 2D TOCSY spectra were recorded at specific time points over a 24 h period. Hydrogen-deuterium exchange rates were measured by integrating each exchangeable amide resonance separately.
The temperature dependence of amide proton resonances was derived from 1D 1H to 2D TOCSY spectra recorded on a Bruker ARX 500 MHz spectrometer. Spectra were measured between 15° and 35°C, in 5°C increments, and referenced to DSS at 0 ppm. Assignment of the spectra was performed using the CCPNMR software (Skinner et al.,
NW9 protein expression and purification followed standard procedures and is described in detail in the
To assemble NDs, dH5 or cNW9 and lipids were co-dissolved at [lipid]:[MSP] ratio of 50:1 in reconstitution buffer (20 mM Tris·HCl pH 7.4, 100 mM NaCl, 0.5 mM EDTA and 100 mM cholate) and mixed for 1 h at 4°C. A molar ratio of 1:50 (MSP:lipids) was calculated using the equation: NL×S = (0.423 × M-9.75)2, where NL is the number of lipids per ND, M is the number of amino acids in the scaffold protein and S is the mean surface area per lipid used to form the lipid-nanodisc, measured in Å2 (Ritchie et al.,
0.6 g of Bio-Beads SM-2 (Bio-rad) was added per mL of reaction volume, to absorb the detergent (cholate), and thus initiating ND assembly. The mixture was gently stirred for 4 h at 4°C for complete detergent removal. The solution was filtered through a 0.45 μm PES membrane to remove the Bio-Beads and then concentrated using centrifugal filtration (Amicon Centricon with a 10 kDa MW cut-off). The sample was buffer exchanged using a PD-10 column (GE Healthcare) into three different buffers: (i) 20 mM Tris·HCl pH 7.5, 50 mM NaCl, 1 mM EDTA; (ii) 20 mM NaPO4 pH 6.5, 50 mM NaCl, 1 mM EDTA; (iii) 20 mM Bis·Tris pH 6.5, 50 mM NaCl, 1 mM EDTA. After buffer exchange of NDs into one of three buffers, samples were concentrated by centrifugal filtration to ~5 mg/mL (unless otherwise stated) and stored at 4 °C.
Lipid NDs were diluted to a final concentration of 200 nM in 20 mM Tris–HCl, pH 7.5, 50 mM NaCl and adsorbed to glow-discharged and carbon-coated EM grids. Samples were prepared by conventional negative staining with 1 % (w/v) uranyl acetate. EM images were collected with a Tecnai 12 electron microscope operated at an acceleration voltage of 120 kV.
LC-MS analysis was conducted on lipid nanodiscs using Agilent Technologies 1200 Series Instrument with a G1316A variable wavelength detector set at λ = 210 nm, 1200 Series ELSD, 6110 quadrupole ESI-MS, using an Agilent Zorbax Eclipse XDB-Phenyl column (3 × 100 mm, 3.5 μm particle size, flow rate 1 mL/min, the mobile phases 0.05% formic acid in water and 0.05% formic acid in acetonitrile).
The affinities of AA139 and VSTx1 for cNDs (both POPC and POPC:POPG mixtures) were determined using a Microcal iTC200 instrument (Malvern, UK). Experiments were performed in 20 mM Bis·Tris (pH 6.5), 50 mM NaCl and 1 mM EDTA. The peptides (at 350 μM) were titrated into 25 μM cNDs in 15 × 2.8 μl (AA139) and 19 × 2.2 μl (VSTx1) injections at 25 °C. Considering the symmetry of the cNDs the stoichiometry (
Solution NMR titration experiments between 15N-VSTx1 and 15N-AA139 and unlabelled cNDs [cNW9 (POPC/POPG (4:1))] were performed on a Bruker Avance III spectrometer equipped with a cryogenically cooled triple resonance probe operating at a nominal 1H frequency of 700 MHz. 15N-HSQC spectra were recorded at 25 °C. The concentration of 15N-VSTx1 or 15N-AA139 were kept constant at 20 and 40 μM respectively, while the concentration of cNDs [cNW9 (POPC/POPG (4:1))] was increased from 0 to 20 μM, with a total of up to six concentrations. A second titration of 15N-VSTx1 against cNDs [cNW9 (POPC)] was conducted under identical conditions to those noted above for this peptide. Each 2D experiment was acquired for ~ 45 min (16 scans and 75 complex points for AA139, and 32 scans and 37 complex points for VSTx1). All titrations were performed in 20 mM Bis·Tris buffer (pH 6.5), 50 mM NaCl and 1 mM EDTA.
NMR titration experiments between unlabelled VSTx1 or AA139 and 15N-cNDs [15N-cNW9 (POPC/POPG(4:1))] were performed on a Bruker Neo spectrometer equipped with a cryogenically cooled, triple resonance probe, operating at a nominal 1H frequency of 900 MHz, at 50°C. Three experiments were performed, the concentration of cNDs [15N-cNW9 (POPC/POPG (4:1))] was kept constant at 100 μM, while the concentration of VSTx1 or AA139 was increased from 0 to 50 then 100 μM. Each experiment was acquired over 4 h (128 scans and 50 increments).
All spectra were processed using Topspin (Bruker Biospin) and the Rowland NMR toolkit (University of Connecticut). CCPNMR was used for spectral analysis. The change in peak intensity between titration points is the product of several dynamic processes that occur due to the binding event (for further details see
The change in signal intensity due to this process
where [
MSP circularization following standard Sortase A reaction conditions (Tris·HCl, NaCl and CaCl2 conditions) (Nasr et al.,
Production and circularization of cNW9.
Three buffer conditions were chosen for the study of lipid-nanodiscs (ND) following the literature (Shenkarev et al.,
20 mM Tris·HCl pH 7.5, 50 mM NaCl, 1 mM EDTA, a standard buffer-salt condition;
20 mM NaPO4 pH 6.5, 50 mM NaCl, 1 mM EDTA, a common NMR condition at a relatively low pH; and
20 mM Bis-Tris pH 6.5, 50 mM NaCl, 1 mM EDTA, where the low pH is retained while the favorable conductivity of the Tris based buffer is also retained.
Size-exclusion chromatography (SEC) was performed on ND samples at 5 mg/mL concentration, using a Superdex S200 5/150
To start, samples were stored at 4°C and monitored by SEC over intervals from day 1 to 30. The SEC profiles for circular cNW9 NDs were always sharper, indicating greater homogeneity of the sample. With the lipid composition consisting solely of POPC lipids, virtually no change in sample profiles was observed for cNW9 or dH5 NDs up to 30 days in the standard buffer condition (i).
cNDs were stable up to 30 days in all buffers tested. However, dH5 NDs in the lower pH conditions of (ii) and (iii) resulted in SEC profiles having an asymmetric peak, with a shoulder at shorter elution times (indicating either unassembled MSP or aggregates) after only 3 days of storage. It should be noted that unassembled MSP elutes earlier than the assembled NDs, indicating a larger effective molecular radius. Under these conditions the phosphate buffer condition (ii) seemed to be slightly worse for stability than the Bis-Tris buffer condition (iii).
The introduction of 20% POPG into the NDs did not affect the stability of cNW9 ND samples at 4°C. However, significant changes in the profile were observed in all buffer conditions for dH5 NDs at 4°C, with the observed effect getting worse from Tris (i) to Bis-Tris (iii) to phosphate (ii) buffer conditions. In the Bis-Tris buffer condition (iii), significant changes in the profile were observed for dH5 NDs at storage temperatures of 25° and 37°C. However, for cNW9 NDs, no changes were observed at either of these temperatures.
Summary of nanodisc stability under different storage conditions.
i. 20 mM Tris pH 7.5 |
4 | 30/30 | 30/30 | 1/30 | 30/30 |
ii. 20 mM PO4 pH6.5 |
4 | 3/30 | 30/30 | 1/30 | 30/30 |
iii. 20 mM Bis-Tris pH6.5 |
4 | 3/30 | 30/30 | 1/30 | 30/30 |
25 | – | – | 1/2 | 2/2 | |
37 | – | – | 0/2 | 2/2 | |
−20 | – | – | 2/2 | 2/2 |
A detailed study was conducted for both constructs of NDs containing POPC/POPG (4:1) in 20 mM Bis-Tris pH 6.5, 50 mM NaCl, 1 mM EDTA using an analytical Supderdex S200 10/300
Size exclusion chromatography using an S200 Increase 10/300 column. Traces of NDs containing POPC/POPG (4:1) lipids using membrane scaffold proteins dH5
VSTx1 was produced using an MBP-fusion construct as previously described, where the structure of this peptide is also reported (Lau et al.,
AA139 was expressed in
Chemical shifts of the protons of AA139 were measured from 2D 1H–1H TOCSY and NOESY NMR spectra. 13C and 15N shifts were obtained from 2D HSQC spectra (
cNDs containing a zwitterionic (POPC) or an anionic lipid mixture [POPC:POPG (4:1)] in the above described Tris-Bis buffer (20 mM BisTris (pH 6.5), 50 mM NaCl and 1 mM EDTA–used in all subsequent experiments) were used to study the interactions between the cNDs and the membrane-active peptides (VSTx1 or AA139) by isothermal titration calorimetry (ITC) and NMR. Neither peptide showed binding to zwitterionic cNDs by ITC.
Initial NMR experiments were conducted to find concentration ranges where clear intensity changes could be observed (data not shown). Based on these concentrations, a titration was conducted where the concentrations of uniformly isotopically labeled 15N-VSTx1 or 15N-AA139 was fixed at 20 and 40 μM, respectively, and NMR spectra were acquired in the presence of increasing concentrations of cNDs.
The ITC binding isotherms using anionic cNDs produced results consistent with weak binding in the μM range (
Thermodynamic parameters for VSTx1 and AA139 binding to cNDs.
VSTx1 | 2.02 ± 0.81 | −1.92 ± 0.73 | −5.88 ± 0.94 | −7.80 ± 0.22 | 2.00 ± 0.20 |
AA139 | 1.16 ± 0.28 | −6.29 ± 0.61 | −1.83 ± 0.75 | −8.12 ± 0.16 | 4.09 ± 0.09 |
15N-VSTx1 at a concentration of 20 μM was titrated in a series against anionic cNDs [POPC:POPG (4:1)] with increasing concentration: 0, 5, 10, and 20 μM (20 mM BisTris (pH 6.5), 50 mM NaCl and 1 mM EDTA). Each of the four samples was then used to acquire a 2D 1H-15N-HSQC experiment. A concentration dependent broadening of the signals was observed (see
1H-15N HSQC titrations of membrane-active peptides against nanodiscs containing bacterial model membranes. (Left) The titration of ion-channel inhibitor VSTx1 (20 μM) against increasing concentration of nanodiscs [POPC:POPG (4:1)]. The three superimposed spectra correspond to a [peptide]:[nanodisc] ratio of 1:0 (black), 1:0.25 (green) and 1:1 (red). (Right) The titration of antimicrobial peptide AA139 (40 μM) against increasing concentration of nanodiscs [POPC:POPG (4:1)]. The three superimposed spectra correspond to a [peptide]:[nanodisc] ratio of 1:0 (black), 1:0.25 (green) and 1:0.375 (red). In both cases 1D traces are shown at 15N frequencies corresponding to peaks that show a small change in intensity that does not fit to the binding isotherm (N14 for VSTx1 and Ala13 for AA139), and a signal (Arg25 for VSTx1 and Trp4 for AA139) having a strong intensity change that fits the binding isotherm. The latter also display an observable chemical shift and linewidth change consistent with intermediate exchange due to binding. Assignments are provided in 1D traces.
The initial fit of individual residues to Equations (2) and (3) showed that residues Cys3, Asn14, Asp15, Cys16, and Phe35 as well as the sidechain resonance of Asn14 did not produce reliable fits to the model. The remaining residues fit the binding isotherm having,
Further to the above titration, a second series of experiments were conducted with cNDs containing zwitterionic lipids. It had previously been observed that VSTx1: 1) binds to zwitterionic micelles, and 2) binds to the MSP of NDs (Shenkarev et al.,
15N-AA139 (40 μM) was titrated in the presence of cNDs [POPC:POPG (4:1)] at 0, 5, 7.5, 10, 12.5, 15 and 20 μM. For each sample a 2D 1H-15N HSQC spectrum was acquired. The increasing concentration of the ND added did not result in any significant chemical shift changes, however, significant broadening of the signals was observed in a concentration dependent manner (see
Further, a qualitative analysis of the likely binding interface of the peptide was performed by comparing the intensity of the signals at the first two titration points (in the absence of cNDs and in the presence of 5 μM of cND). Residues that are most strongly affected at this low concentration of ND, are most likely to be at the binding interface. These residues include the hydrophobic and acidic residues at the termini of the β-hairpin loop (see also
Lipid interaction of membrane-active peptides. Cartoon representation of the peptides [
As noted above it had previously been suggested that there is an interaction between the anionic MSP and cationic peptides such as VSTx1 and AA139 (Shenkarev et al.,
The procedure described here for production of cNDs, departs from the original protocol described (Nasr et al.,
In this work, we investigated the stability of NDs containing anionic lipid mixtures commonly used to model the bacterial membrane [POPC:POPG (4:1)], and compare this to NDs containing zwitterionic lipids (POPC). The stability of the NDs is evaluated based on EM images and SEC profiles, where in the latter aggregation or disassembly results in asymmetric elution profiles with increased absorbance at early elution times. For the linear NDs there is a general trend that the lower pH buffers result in higher heterogeneity of the NDs (
Fortunately, the introduction of head-to-tail cyclisation of the MSP resulted in excellent (c)ND stability regardless of the lipid composition or sample conditions–i.e., no changes were observed to the SEC elution profiles or EM images. Our conclusion based on these results is that small linear NDs are unsuitable for studies of anionic lipid bilayers. We further note that, although it is known that the presence of anionic lipids does not alter the membrane thickness or its fluidity, it has a very significant effect on the stability of lipid nanodiscs in solution. This conclusion may shed some light into previous work on the use of anionic NDs for studies of membrane-active peptides, where it was found that VSTx1 binds very effectively to anionic NDs, and based on analysis of NMR titration data a stoichiometry of ~35 peptides to each ND was found (Shenkarev et al.,
Although protein interactions with NDs have been studied previously by ITC (Agamasu et al.,
Isotope labeling of both membrane-active peptides was achieved through recombinant protein expression in bacteria. In the case of AA139, a SUMO-fusion tag produced high soluble yields (~1 mg per liter of culture) when expressed in SHuffle cells, in contrast to the MBP fusion expressed in the BL21 strain, where no soluble protein was found. It is worth noting the inherent challenge in finding a suitable bacterial expression system for an antimicrobial peptide. In this case it appears that the SUMO-fusion may have reduced the toxicity of the peptide. Although the bacterial expression of VSTx1 had previously been described, we also found that VSTx1 could be produced in relatively high yields using this strategy (SUMO-fusion in SHuffle cells–yielding 0.5 mg per liter of culture).
Transferring the results of the analysis of the NMR data (see discussion in
Those that fit the binding isotherm but show smaller changes in their intensity [category (a) in
The lipid binding data for VSTx1 are consistent with previous findings, and show, perhaps unsurprisingly, that the positively charged peptide binds more strongly to lipids containing negative head-groups. What is surprising, however, is that the same amino acids show the strongest perturbations in both titrations (against anionic and zwitterionic bilayers). This would suggest that the binding is driven by the hydrophobic patch of the peptide (loops 1 and 4) and that the affinity is enhanced by the basic residues in these loops in the presence of acidic moieties at the bilayer surface. This is particularly interesting as this peptide has been found to interact with acidic residues on its ion-channel receptor, in what appears to be a conserved mode-of-action in the inhibitory function of gating-modifier toxins (Lau et al.,
The structure of AA139 reveals a twisted β-hairpin structure similar to other members of this family of antimicrobial peptides (Edwards et al.,
The ITC experiments revealed that both peptides had binding affinities in the low μM range. Interestingly, although the binding free energy (ΔG) is similar in both cases (~-8 kcal/mol), the relative contribution of entropic and enthalpic terms is different. The binding of VSTx1 appears to be largely driven by an entropic component (~-6 kcal/mol) rather than an enthalpic component (~-2 kcal/mol). The larger entropic term is consistent with increased disorder either in the peptide or the lipid bilayer upon binding. Although it is difficult to deduce the relative contributions of each of these components, we do see a very significant broadening of residue L31 of VSTx1 even in the presence of small concentrations of cNDs. This is consistent with a conformational change upon binding, and is likely to account for some of the observed gain in entropy. The change in conformation of a leucine residue is likely to occur in a highly apolar environment and the data would suggest that this face of the molecule partitions into the lipid bilayer.
In contrast to VSTx1, the ITC results for AA139 reveal a reversal of the relative thermodynamic terms, with a greater enthalpic term (~-6 kcal/mol). AA139 has a net charge of +5 which is due to the presence of five arginine residues accounting for almost a quarter of the total amino acid composition. This would suggest that the binding of the peptide is driven by charge-charge interactions at the lipid interface, involving some of these residues. As noted above the masking of the C-terminal charge through amidation would increase the net charge to +6 which may further strengthen the enthalpic contribution to the binding energy.
In general, there is good qualitative agreement between the ITC and the NMR data. The binding constants derived from the NMR data were as expected exaggerated when compared to the ITC data, thus, although not quantitative provide a reasonable estimate of the binding. The stoichiometry was consistent when comparing data from the two methods. We note that in both cases the relatively weak binding observed, required careful analysis of the data, where a priori knowledge regarding the symmetry of the disc was used to improve the fitting of the data. Thus, although there is good agreement in the values derived, we note that these data are near the limits of binding interactions that can be detected by the two methods and some uncertainties are likely associated with the inherent sensitivity of these techniques in this binding regime.
We present a method to obtain high yields of cNDs in 3 days and assess their suitability for biophysical studies in solution. We performed size-exclusion chromatography (SEC) of linear and cyclised NDs of different compositions under varying conditions. The cNDs consistently displayed significantly improved stability over their noncircular counterparts. The study revealed that linear NDs can be unstable under common experimental conditions, particularly in phosphate buffers at a low pH and when containing anionic lipids. In contrast, cNDs were stable at all solution conditions and lipid compositions tested. In addition, the sharper peaks observed in the SEC profiles indicate greater homogeneity. Finally, we describe a method for high-yield (~mg per liter of culture) recombinant production of two membrane-active peptides, notably including an antimicrobial peptide–AA139.
These materials are then used to evaluate the use of biophysical methods to study the membrane binding properties of membrane-active peptides against cNDs containing anionic lipid mixtures, that approximate the charge distribution of bacterial membranes. We use ITC to measure the binding thermodynamics, and heteronuclear 2D NMR to characterized the binding of 15N-labeled peptides against cNDs. We first study the well characterized VSTx1 peptide to validate the proposed approach, and find good agreement with previous reports while revealing new information regarding the thermodynamics of the binding event. We then apply our method to gain insights into the activity of AA139, an antimicrobial peptide with an unknown mode-of-action. ITC and NMR data show that AA139 binds to the cNDs with low μM affinity, which is driven by a significant enthalpic contribution (−6 kcal/mol) and a stoichiometry of 4 peptides per disc. We further solved the structure of AA139 by NMR, which revealed a twisted β-hairpin fold, and allowed us to determine the likely lipid-binding interface of the peptide, which included the tails of the antiparallel β-sheets. These results establish the use of cNDs in combination with ITC and solution-state NMR as a novel and general method for investigating the membrane binding properties of membrane-active peptides.
AZ, IE, and MM conceived the study. AZ, IE, BM, GS, MH, and XJ performed the experiments. AZ, IE, BM, MH, BC, and MM analyzed the data. IE, AZ, and MM wrote the paper with input from all authors. All authors contributed to different components of the study design.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This project was supported by the Australian Research Council (ARC grants: DP140101098, FTl10100925) and the National Health and Medical Research Council (NHMRC grants: APP1102267, APP1080405, APP1106590 and APP1136021). IE and AZ are supported by an International Postgraduate Award and an Australian Postgraduate Award Ph.D. scholarship, respectively. We thank Adenium Biotech for allowing this work using AA139 and Dr. Frank Sainsbury for his technical assistance. We are also grateful to Dr. Mahmoud L. Nasr and Prof. Gerhard Wagner for MSP1D1ΔH5 expression plasmid and Prof. David R. Liu for sortase A pentamutant expression plasmid. Finally, we thank Dr. Zhenling Cui and Prof. Kirill Alexandrov for the advice and reagents regarding SUMO-tag cleavage by SUMO protease.
The Supplementary Material for this article can be found online at:
The supporting information include further details on the NMR analysis, ITC data, size exclusion chromatography profiles, thin layer chromatography images, EM negative staining images and LC-MS traces.
ND
cNW9
dH5
AMPs
MSP
PDB
DPC
DMPC
DMPG
POPC
POPG
SEC
EM
MS
CSP
PCR.