NMR Structural Studies of Antimicrobial Peptides: LPcin Analogs

Abstract

Lactophoricin (LPcin), a component of proteose peptone (113–135) isolated from bovine milk, is a cationic amphipathic antimicrobial peptide consisting of 23 amino acids. We designed a series of N- or C-terminal truncated variants, mutated analogs, and truncated mutated analogs using peptide-engineering techniques. Then, we selected three LPcin analogs of LPcin-C8 (LPcin-YK1), LPcin-T2WT6W (LPcin-YK2), and LPcin-T2WT6W-C8 (LPcin-YK3), which may have better antimicrobial activities than LPcin, and successfully expressed them in E. coli with high yield. We elucidated the 3D structures and topologies of the three LPcin analogs in membrane environments by conducting NMR structural studies. We investigated the purity of the LPcin analogs and the α-helical secondary structures by performing ¹H-¹⁵N 2D HSQC and HMQC-NOESY liquid-state NMR spectroscopy using protein-containing micelle samples. We measured the 3D structures and tilt angles in membranes by conducting ¹⁵N 1D and 2D ¹H-¹⁵N SAMMY type solid-state NMR spectroscopy with an 800 MHz in-house-built ¹H-¹⁵N double-resonance solid-state NMR probe with a strip-shield coil, using protein-containing large bicelle samples aligned and confirmed by molecular-dynamics simulations. The three LPcin analogs were found to be curved α-helical structures, with tilt angles of 55–75° for normal membrane bilayers, and their enhanced activities may be correlated with these topologies.

Introduction

Antimicrobial therapy is used in medical practice for diagnosis and treatment, saving millions of lives (1). Since antimicrobial peptides (AMPs) have such different bacterial-killing mechanisms compared with antibiotics, as well as a versatile ability to kill bacteria, fungi, and even some parasites and viruses, they are considered attractive substitutes for conventional antibiotics (2). They are also recognized as a feasible source of pharmaceuticals for the treatment of antibiotic-resistant bacterial infections such as multidrug-resistant tuberculosis (MDR-TB) and methicillin-resistant Staphylococcus aureus (MRSA). AMPs show a great diversity of primary structures and can be classified into four groups: α-helical peptides, β-sheet peptides, extended peptides, and loop peptides. In particular, the small cationic amphipathic α-helical peptides are recognized as regulators in the innate immune system of the host against microorganisms (2, 3).

Lactophoricin (LPcin) is a cationic amphipathic peptide consisting of 23 amino acids that comprise a portion of proteose peptone 3 (PP3). It has been reported that this small cationic peptide has antibacterial activity against both Gram-positive and Gram-negative bacteria (4, 5). The correlation between the structure and the antimicrobial activity of this peptide was previously studied in detail by means of solution and solid-state NMR spectroscopies (6, 7, 8, 9, 10). As part of our efforts to find novel AMPs with better antimicrobial activity than LPcin, we designed 11 LPcin analog peptides and chose three analog peptides that showed the best antibacterial activity, using the agar hole test against five pathogens. These three analog peptides (LPcin-YK1: NTVKETIKYLKSLFS; LPcin-YK2: NKVKEWIKYLKSLFSHAFEVVKT; and LPcin-YK3: NKVKEWIKYLKSLFS) were effective against two Gram-positive (Listeria and S. aureus) and three Gram-negative (P. aeruginosa, Salmonella, and E. coli) strains (MIC₅₀: YK1 and YK3 < 50 μL, YK2 < 100 μL). The three LPcin analogs were overexpressed in mutant E. coli strains and produced enough peptide for solution and solid-state NMR experiments. Then these peptides (net charge at neutral pH: YK1, +2; YK2 and YK3, +3) were purified and identified by reverse-phase high-pressure liquid chromatography (HPLC), mass spectrometry (MS), and circular dichroism (CD) spectropolarimetry (11, 12).

Here, we describe the 3D structure and topology of these three LPcin analog peptides as determined by solution and solid-state NMR in membrane environments, and show how they reveal the antibacterial-activity mechanisms of these peptides. Solid-state NMR spectroscopy presents an opportunity to determine the structures of membrane proteins in phospholipid bilayer membranes under experimental conditions that are very close to physiological, without the need for major protein modifications, detergents, or extreme temperatures (13). For studies using solution NMR, mixed protein-detergent micelles are needed to stabilize proteins in water and overcome the limitations of conventional NMR experiments that would obtain spectra with insufficient resolution and low signal/noise ratio (14). Micelles provide reasonable information about the structure the protein may adopt in lipid environments, as well as the motional properties that are advisable for solution NMR. Dodecylphosphocholine (DPC) is well known to be a basic model membrane system for studying proteins that interact with lipids. It forms a stable micelle that rotates freely in solution and mimics anisotropic environments such as the lipid membrane (15). Although DPC is one of the most commonly used detergents in solution NMR assays, many other lipids can be employed (16). Recently, many techniques have been developed to quantify motions that occur over a wide range of correlation times and to measure orientation constraints as indices to determine peptide structures (17). The study of biomolecular dynamics using solution NMR spectroscopy has become widespread thanks to diverse motional timescales, spatial resolution from isotopic labeling of peptide-coupled multidimensional spectroscopy, and continuing advances in both experimental and computational methods (18). To characterize the diverse functions, specificity, and efficacy of the three analog peptides, it is important to determine their high-resolution structure and their dynamics under native conditions. Solid-state NMR spectroscopy is a premier method for investigating immobile and noncrystalline proteins that are difficult to study with x-ray crystallography or solution NMR. Oriented sample (OS) solid-state NMR offers distinctive advantages to examine the local structure and motions, and to delineate the nature of peptide-lipid interactions in physiologically relevant conditions (19, 20, 21, 22, 23). Magnetically oriented bicelles have many advantages. First of all, the solid-state NMR spectra of the peptide in bicelles contain sharp resonance lines because of their high degree of macroscopic alignment. Thus, the assignment of spectra can be readily used to determine the structure of the peptides. Moreover, they provide native membrane environments that fully mimic proteoliposomes with various phospholipids (24). To define the structures of the three analogs of cationic AMPs, we used well-mixed phospholipids of 1,2-di-otetradecyl-sn-glycero-3-phosphocholine (14-O-PC), 1,2-di-O-hexyl-sn-glycero-3-phosphocholine (6-O-PC), and dimyristoyl phosphatidylglycerol (DMPG) to imitate the anionic bacterial membrane (19). We also investigated the solid-state NMR spectra of the three LPcin analog peptides by conducting a polarity index at slanted angle (PISA) wheel pattern analysis (25). The PISA wheel pattern contains structural information regarding membranes, such as the tilt angle and torsion angle. The structures of membranes associated with peptides are not easily determined using the standard method because these structures are greatly affected by a lipid bilayer. Therefore, we verified the 3D structures of the three LPcin analog peptides by using computational modeling based on molecular-dynamics (MD) simulations with implicit membranes (26).

Materials and Methods

Expression and purification

We used mutant E. coli strains with bacteriophage T7 RNA polymerase expression systems for the overproduction of the three LPcin analog peptides. These oligonucleotides were ligated into the pET31b (+) vector (Novagen, Darmstadt, Germany) and the process used for the choice of C41 (DE3) and C43 (DE3) (Lucigen, Middleton, WI) was employed to tailor the expression host cells so as to avoid the toxic effects related to overexpression (27). Uniformly or selectively ¹⁵N-labeled peptides were overexpressed in M9 minimal media for NMR structural studies after the cells were fully grown in M9 media. The three analog peptides were purified by preparative reverse-phase HPLC. Details regarding the process of expression and purification using various biological techniques are described elsewhere (7, 11, 12). The purity of the three refined analog peptides was checked and tandem MS (MS/MS) de novo sequencing was performed on a mass spectrometer (4800+ MALDI TOF/TOF Analyzer; AB Sciex, Foster City, CA). For the secondary structures of LPcin and its three analog peptides in membrane environments, we performed CD experiments using a Jasco J715 CD spectropolarimeter (Jasco, Easton, MD) with a 1-mm-path-length cylindrical cuvette in various environments close to that of a basic membrane (11). The measurements were first obtained in water at ambient temperature and then in DPC micelles at concentrations of 20, 40, 60, 80, and 100 mM.

Preparation of micelles

Solution NMR studies were carried out in deuterated DPC. We prepared three kinds of micelle samples for solution NMR experiments by dissolving 0.5 mg of the three purified analog peptides uniformly labeled with ¹⁵N and 11 mg of perdeuterated DPC (Cambridge Isotope Laboratories, Andover, MA) in a 300 μL solution (90% H₂O and 10% D₂O) at pH 4.0.

Preparation of bicelles

We used 14-O-PC for long chains (Avanti Polar Lipids, Alabama, AL) and 6-O-PC (Avanti Polar Lipids) for short chains. To determine the prokaryotic membranes related to physical properties such as the presence of negatively charged lipids (28), we used DMPG (Avanti Polar Lipids). All lipids were well dissolved in chloroform and then a stream of N₂ gas was used to remove the solvent. The film of lipid on the walls of the test tube was kept in a vacuum overnight to remove any residual solvent. The lyophilized peptides were codissolved with hydrated 6-O-PC, 14-O-PC, and DMPG in 200 μL sterilized water. The mixture, which was extremely viscous and frosted, was homogenized by vortexing with freeze-and-thaw cycles between 0°C and 45°C until the bicelles became transparent and homogeneous. The final lipid composition of 6-O-PC/14-O-PC/DMPG was in a 1.0:2.6:0.6 molar ratio and the q-value (long chain/short chain) of the lipid was 3.2. The bicelles were transferred into a 5 mm flat-bottom NMR tube (New Era Enterprises, Newfield, NJ), closed with a rubber cap, and then sealed with Parafilm and Teflon tape. The bicelles could be aligned because their bilayer normal was perpendicular with respect to the external magnetic field.

Solution-state NMR experiments

All NMR spectra were recorded with a Bruker Avance spectrometer (Bruker Biospin, Billerica, MA) operating at 800 MHz and equipped with a gradient unit. The experiments were carried out using 0.5 mg uniformly ¹⁵N-labeled peptides dissolved in the presence of 100 mM DPC-d₃₈ micelles at pH 4.0, with 90% ddH₂O and 10% D₂O (v/v). 2D ¹H-¹⁵N heteronuclear single quantum correlation (HSQC) spectra were acquired with time domains of 1024 and 128 complex points for ¹H (F2) and ¹⁵N (F1), respectively. The spectral widths were 8800 Hz for the F2 dimension and 1600 Hz for the F1 indirect dimension. Flip-back pulse techniques included in the HSQC pulse were used for water suppression and sensitivity improvements. For all HSQC NMR spectra, a total of eight scans were used and the total experimental time was ∼20 min. Heteronuclear multiple-quantum correlation spectroscopy-nuclear Overhauser effect spectroscopy (HMQC-NOESY) experiments were performed with the same micelle samples and experimental conditions, differing only in the number of complex points and total scans. We used 2048 complex points for the F2 dimension and 256 points for the F1 indirect dimension, and performed a total of 32 scans. HSQC and HMQC-NOESY spectra were recorded at 313K and processed using Bruker Topspin software.

Solid-state NMR experiments

All NMR experiments were performed with a Bruker Avance spectrometer operating at 800 MHz and equipped with a gradient unit at KBSI (Korea Basic Science Institute). Ammonium sulfate was used as the external reference corresponding to 26.8 ppm. The bicelle samples were placed in an in-house-built, ¹H-¹⁵N double-resonance, solid-state NMR probe with a strip-shielded solenoidal coil for at least 30 min for spontaneous alignment of the bicelle samples in the magnetic field after the temperature had reached 42°C and 60°C (see Supporting Materials and Methods in the Supporting Material). The alignment came about as a result of the interaction between the magnetic field and the magnetic susceptibility anisotropy of the bicelle. 1D ¹⁵N spectra were obtained using a spin-locked ¹H-¹⁵N cross-polarization technique with a 1 ms mixing time. A single level of ¹H radiofrequency power was used at ∼62.5 kHz. The total numbers of scans were 3072 for LPcin-YK1 and -YK3, and 2048 for LPcin-YK2. An improved version of sandwich-based separated local field spectroscopy (SAMMY), SAMPI4, was used for 2D NMR experiments. SAMPI4 generally has the advantage of producing spectra with narrower and more uniform spectral bandwidths and intensities over a relatively wide range of ¹H-¹⁵N dipolar coupling and ¹H and ¹⁵N chemical-shift frequencies (29). The numbers of scans in the F2 dimension were 400, 256, and 300 for LPcin-YK1, -YK2, and -YK3, respectively, and the t1 increments were 34, 34, and 32 for LPcin-YK1, -YK2, and -YK3, respectively. As the temperature was raised, a marked increase in sensitivity and improvement in resolution (∼1.5 ppm line widths) was observed, and a single narrow resonance for each labeled ¹⁵N-¹H bond was observed in the ¹⁵N chemical-shift spectra (19). Thus, the experimental temperature for both the 1D and 2D spectra was 42°C for LPcin-YK1 and -YK3, and 60°C for LPcin-YK2.

PISA wheel pattern analysis

¹H-¹⁵N separated-local-field (SLF) spectra of helical peptides like LPcin and its three analog peptides with bicelles magnetically aligned in the external magnetic fields have characteristic circular patterns called PISA wheels (25, 26). The slanted angle τ of the amphipathic α-helical structure of the three analog peptides was determined by fitting PISA wheel patterns to the ¹H-¹⁵N SAMPI4 data. This fitting is relevant to helical wheel projections and provides indications of the secondary structure and topology. The PISA wheel patterns were calculated using MATLAB & Simulink v. R2010a (The MathWorks, Natick, MA) script. The standard order parameter S = 0.85 and dihedral angles of Φ = −65°, Ψ = −40° were used for the simulation. The principal values for the chemical-shift tensor were applied to σ₁₁ = 64 ppm, σ₂₂ = 77 ppm, and σ₃₃ = 222 ppm for ¹⁵N, and σ₁₁ = 3 ppm, σ₂₂ = 8 ppm, and σ₃₃ = 17 ppm for ¹H (30, 31).

MD simulations

All simulations in this study were performed using the CHARMM-27 (c35b5) all-atom force field (32, 33) and Discovery Studio (DS) 3.1 (Accelrys, San Diego, CA). Also, the SHAKE constraint necessary to keep the position of the bond involving hydrogen atoms at equilibrium was applied to all simulations. The initial structures of the three LPcin analog peptides were generated to the ideal α-helix template using Build and Edit protocols. The initial geometry of the three LPcin analogs in the vacuum was refined by using an energy-minimization protocol with the steepest-descent, conjugate-gradient, and adopted-basis NR algorithm. The positions of the three analog peptides in the membrane were optimized along the bilayer normal axis (z) spaced 0.5 Å apart (z = −11.5 ∼11.5 Å; z = 0 is the center of the bilayer normal axis). In this calculation, the thickness of the low dielectric (hydrophobic) part of the membrane was used to 23 Å, and the generalized Born with a simple switching function (GBSW) implicit membrane model was applied to perform effective calculations of the interactions between proteins and the model membrane (34).

We calculated the MD of the three analogs in the model membrane by using the Standard Dynamics Cascade protocol with the leap-frog Verlet dynamics integrator and canonical NVT ensemble. This ensemble prevents alternation of parameters for this system in thermal equilibrium during an MD simulation. The three LPcin analogs in this protocol were heated to 300 K for 2 ps and then equilibrated for 200 ps. The final MD products of the three LPcin analogs were obtained for 1 ns at 300 K.

Finally, we constructed realistic membrane models for the three analogs using MD products from DS 3.1 and the membrane builder module in CHARMM-GUI (http://www.charmm-gui.org/). The membrane builder module can make realistically packed lipid bilayers around the membrane protein using a certain lipid type. The coordination of the three analogs in the membrane build module was based on the MD simulation. These results were visualized with the use of DS 3.1.

Results and Discussion

Solution-state NMR studies of three analog peptides

Each crosspeak in the 2D ¹H-¹⁵N HSQC spectrum presents an amino acid in a special backbone conformation of peptides. Thus, the ¹H-¹⁵N HSQC spectrum of peptides can be expected to be unique (35). The ¹H-¹⁵N HSQC spectra of the three analog peptides show that the N-H crosspeaks of the backbone were well resolved (Fig. 1). The total resonances correspond to the number of amino acids comprising the side-chain N-H bond of arginine (N1) (Table S1) The extensive chemical-shift changes observed for the peptides in water or micelles reflect changes in the conformation of the peptide. In Fig. 1, the crosspeaks from each labeled amide site within the narrow dispersion (red) of ¹H chemical shifts indicate that the three analog peptides did not show distinct structures in aqueous environments. However, the crosspeaks spread out through a wide range (black, 7–9 ppm) of ¹H chemical shifts, demonstrating that the three analog peptides adopted stable conformations in membrane environments. The dispersion of the peaks demonstrates the unique chemical environments of residues in the folded protein, with the greatest dispersion evidenced in β-sheets, followed by α-helices. The ¹⁵N-¹H HSQC spectrum of the peptides showed limited HN dispersion at ∼7.1– 8.8 ppm, which is characteristic of all α-helical proteins. Unfolded proteins or regions of a folded protein without regular secondary structures have amide proton resonances between 8 and 8.5 ppm (36, 37). This result agrees well with the CD data mentioned above (11).

Figure 1.

(a–c) Comparison of ¹H-¹⁵N HSQC spectra for three analog peptides (LPcin-YK1 (a), LPcin-YK2 (b), and LPcin-YK3 (c)) in water (red) and micelle (black) environments. The concentration of all peptides was 0.5 mM in the presence of 100 mM DPC micelles and in water. Each crosspeak in the spectrum represents an amide proton in the N-H bond from the backbone of the peptide. The constructs included crosspeaks from an arginine (Arg) side chain. Residues exhibiting a wide range of chemical shifts provide evidence that the peptide adopted a stable conformation in micelle environments, but not in water. To see this figure in color, go online.

Selectively labeled peptides in DPC micelles were also prepared to confirm the assignments of the three analog peptides. Various 2D ¹H-¹⁵N HSQC spectra were recorded from uniformly ¹⁵N-labeled LPcin-YK1, LPcin-YK2, and LPcin-YK3 peptides, as well as selectively ¹⁵N-labeled peptides with micelle environments (Fig. S1). Since the approximate peak assignments were conducted using Fig. S1, it was easy to confirm the full sequence with the ¹H-¹⁵N HMQC-NOESY spectra of each peptide (Fig. S2).

Solid-state NMR experiments

Even though some membrane protein structures have been determined by x-ray crystallography, the technique has limitations related to purification and crystallization (28, 38, 39, 40). Since the function of membrane proteins depends highly on their environment, investigations of protein structure should be done with a membrane milieu (41). This can be achieved by solid-state NMR experiments with the appropriate isotope-labeled proteins in bicelle samples (42). Also, the effect of environment on the structure of membrane proteins is particularly significant. Therefore, the surfactants used should be considered to mimic the physical features of the biomembrane (43). To investigate the topology of the membranes of the three analog peptides, we made large bicelle samples that mimicked the membrane environments. To measure the chemical-shift anisotropy (CSA) of the peptides and dipolar couplings, investigators have used bicelles over the past few decades as an alignment medium for various membrane proteins in solution (44). Results obtained with ether-linked phospholipids have demonstrated the stability of these model membranes with a wide range of pH values (45).

We obtained 1D ¹⁵N solid-state NMR spectra of the three analog peptides using uniformly ¹⁵N-labeled peptides in aligned bicelles. Fig. 2 shows the ¹H-¹⁵N cross-polarization spectra of LPcin-YK1 (Fig. 2 a), LPcin-YK2 (Fig. 2 b), and LPcin-YK3 (Fig. 2 c). ¹⁵N solid-state NMR spectroscopy is widely used to determine the membrane topology of peptides and proteins attached to lipid membranes, in addition to their rotational diffusion rates. Since the size and orientation of the principal value (σ₁₁, σ₂₂, σ₃₃) and the chemical-shift tensor elements define the CSA pattern, which is an index of the MD in the membrane, ¹⁵N solid-state NMR experiments can be used to determine the orientation of helical peptides in bicelles aligned with their normal parallel or perpendicular to the external magnetic field direction (46, 47). The single-line resonances observed in the OS solid-state NMR spectra obtained with uniaxial OSs coincide with the frequencies of motionally averaged powder patterns as the peptides undergo uniaxial rotational diffusion with respect to the bilayer normal, and are much faster (>10⁵ Hz) than their static spin interaction tensor (17).

Figure 2.

(a–c) 1D ¹⁵N solid-state NMR spectra of LPcin-YK1 (a), LPcin-YK2 (b), and LPcin-YK3 (c). The resonances near and to the left of 120 ppm for LPcin-YK1, LPcin-YK2, and LPcin-YK3 show that they contain in-plane helices. This suggests that the surface of the membrane and small portions of the peptide are outside of the in-plane. The experiments were carried out at 42°C and 60°C for LPcin-YK1, -YK3, and -YK2 to obtain high-resolution solid-state NMR spectra. The peptide/lipid molar ratio was ∼1:70.

For an embedded peptide in transmembrane bilayers or flipped bicelles, the 1D ¹⁵N solid-state NMR spectrum appears through the region from 185 to 225 ppm. However, for the unflipped bicelles oriented with their normal perpendicular to the magnet field, this region is inverted with respect to 120 ppm, which is the isotropic value and is scaled by a factor of 0.5 (48, 49, 50). Although it is possible to make their bilayer normal to be parallel with the addition of lanthanide ions, unflipped bicelles are favored because they are more stable over 6 months and yield sharper resonance lines. The sharper resonance lines reflect the fact that the proton CSA and motionally averaged dipolar couplings have a narrower range than the normal perpendicular orientation. In addition, the effect of mosaic spread (i.e., the misalignment of proteins and phospholipids) is less experimentally observed at the normal perpendicular orientation. Therefore, it is possible to obtain spectra with narrower line widths because the ¹H decoupling is more efficient (24, 51, 52, 53).

For in-plane helixes such as the three analog peptides in unflipped bicelles, ¹⁵N resonances are sited near or to the left of the isotropic frequency of 120 ppm. The resonances near the area of 105–140 ppm for LPcin-YK1, LPcin-YK2, and LPcin-YK3 indicate that they have some residues in an in-plane helix and some outside the in-plane helix, as shown in Fig. 2, a–c.

The 2D ¹H-¹⁵N SLF-based SAMPI4 spectra of LPcin-YK1, LPcin-YK2, and LPcin-YK3 are shown in Fig. 3, a–c. These spectra provided most of the orientational restriction used for structure determination (47). The information regarding their topologies and structures was obtained from the ¹⁵N chemical-shift and ¹H-¹⁵N dipolar coupling frequencies of SAMPI4 spectra (20, 21, 22). The 1D and 2D solid-state NMR spectra were processed using TopSpin software (Bruker) and NMRPipe (54), respectively.

Figure 3.

(a–c) 2D ¹H-¹⁵N SAMPI4 spectra of LPcin-YK1 (a), LPcin-YK2 (b), and LPcin-YK3 (c) overlap with the PISA wheel pattern analysis results (indicated by dotted lines). The experimental SAMPI4 spectra of LPcin-YK1 (a), LPcin-YK2 (b), and LPcin-YK3 (c) in magnetically aligned bicelles provided the orientational restriction used for structure determination. The PISA wheel pattern analysis indicated that LPcin-YK1 was tilted at 50° and 48.5°, LPcin-YK2 was tilted at ∼67° and 76°, and LPcin-YK3 was tilted at 47° and 54° with respect to the bilayer normal. To see this figure in color, go online.

PISA wheel pattern analysis

Uniaxially aligned peptides in lipid bicelles with their bilayer normal lying perpendicular to the external static magnetic field produce a single-line resonance for each of their ¹H-¹⁵N amide bonds with well-resolved PISA wheel patterns. Peptides that are immobilized interact with the anionic preformed surface of the bicelles, but lack the fast rotation. To determine the rotational diffusion coefficients around the bilayer normal, one can compare the experimental results with numerical simulations. If there is fast rotational diffusion around the axis of an α-helix, the PISA wheels will be projected into a single frequency around the center of the ideal wheel when there is no fast motion around the α-helical axis but there is fast rotation around the axis of the bilayer normal (55). The 2D ¹H-¹⁵N SAMPI4 spectra of the three analog peptides exhibited characteristic circular patterns of resonances. The calculated PISA wheel pattern showed an ideal α-helix with tilt angles of 50° and 48.5° for LPcin-YK1, 76° and 67° for LPcin-YK2, and 54° and 47° for LPcin-YK3. The 2D ¹H-¹⁵N SAMPI4 spectra for LPcin-YK1, LPcin-YK2, and LPcin-YK3 overlapped with the calculated PISA wheel, as shown in Fig. 3. All three LPcin analog peptides showed a kinked α-helical conformation characterized by two helical axes in lipid membrane environments.

MD simulations

We confirmed the results obtained from the NMR spectrum for the three LPcin analogs by comparing them with the computational MD simulation studies. The initial modeled structures of the peptides were prepared by energy minimization in a vacuum. Before the MD simulations, the initial positions of the energy-minimized structures in the model membrane were found by using the Add Membrane and Oriented Molecules protocol. All three analog peptides were located on the surface region of the model membrane.

During the MD simulations, the three analog peptides in the membrane were driven by heating from 50 K to 300 K and equilibrated at 300 K. Finally, the dynamics of the peptides were obtained with NVT ensembles during the final production step. The three analog peptides formed a curved α-helix structure with amphipathy and were oriented perpendicular to the bilayer normal. These structures are represented in Fig. 4. The orientation for LPcin-YK2 in the membrane corresponded with the PISA wheel pattern base on the SAMPI-4 spectra. However, the orientations of LPcin-YK1 and YK3 showed little difference between the MD simulations and PISA wheel patterns. These differences may be due to several factors, such as the deformation effect of the membrane lipids, hydrophobic mismatch, protein aggregation, and ionic interaction (56, 57, 58). However, these factors could not be explored in the MD simulations due to limitations of the implicit-solvation model.

Figure 4.

Orientations of the three LPcin analog peptides on the membrane after an MD simulation. All LPcin analog peptides were curved amphipathic α-helixes. (a–c) Angle θ represents the tilt angle for the membrane bilayer normal axis (z): (a) LPcin-YK1 (θ = 62°, 52°), (b) LPcin-YK2 (θ = 67°, 76°), and (c) LPcin-YK3 (θ = 54°, 64.5°). To see this figure in color, go online.

Conclusion

The three mutated and sequence-shuffled analog peptides developed in our lab based on LPcin, a cationic amphipathic AMP, are good candidates for antibacterial peptides. These peptides were successfully overexpressed in E. coli, purified, and structurally characterized in micelles and lipid bilayers. Resonance assignments of the 2D ¹H-¹⁵N HSQC spectra for the uniformly ¹⁵N-labeled LPcin analogs in micelles were performed based on 2D ¹H-¹⁵N HMQC-NOESY spectra and 2D ¹H-¹⁵N HSQC spectra for the selectively labeled LPcin analogs in DPC micelles. The three LPcin analog peptides magnetically oriented in anionic phospholipid bilayers were also studied in 1D ¹⁵N cross-polarization mismatch-optimized IS transfer and 2D ¹H-¹⁵N SAMPI4 solid-state NMR experiments using an in-house-built, ¹H-¹⁵N double-resonance probe with a strip-shielded coil for the 800 MHz narrow-bore magnet to characterize the absolute orientation in membrane environments. The characteristic circular patterns of resonances of 2D ¹H-¹⁵N SAMPI4 spectra for the three analog peptides were used to calculate an ideal α-helix with a tilt angle by PISA wheel pattern analysis. 1D and 2D ¹⁵N solid-state NMR spectra of the three analog peptides were used as indices of the location of the peptides in the membrane. The NMR spectroscopy results were compatible with the results obtained from MD simulations using computational modeling methods (59). All of the LPcin analog peptides showed bending α-helical conformations with specific tilt angles for LPcin-YK1 (θ = 62°, 52°), LPcin-YK2 (θ = 67°, 76°), and LPcin-YK3 (θ = 54°, 64.5°) (angle θ represents the tilt angle for the membrane bilayer normal axis (z)) in lipid membrane environments. The curved α-helical conformation would play an important role in penetrating and disrupting the membrane, and eventually killing the bacteria. Structural and topological studies of AMPs are required due to the remarkable antibacterial efficacy of these peptides against a wide range of microbes.

Author Contributions

J.H.J. contributed to expression, purification, and solid-state NMR spectroscopy. J.S.K. contributed to expression, purification, and solution NMR spectroscopy. S.S.C. contributed to the MD simulations. Y.K. designed research, analyzed data, and wrote the manuscript.

Editor: Francesca Marassi.

Supporting Citations

References (60, 61, 62) appear in the Supporting Material.