Membrane Topology and Oligomerization of Peptaibols by Solid‐State NMR Spectroscopy

ABSTRACT

Here, we review solid‐state NMR investigations of a variety of peptaibols in phospholipid bilayer environments. The peptides were prepared using different approaches such as microbial expression, solid‐phase peptide synthesis or the chemical synthesis in solution and a modular fragment condensation approach. Importantly, the techniques allowed the preparation of peptaibols labelled with ¹⁹F as well as with ¹⁵N either uniformly or specifically. In a first approach, ¹⁵N‐labelled peptides were reconstituted into uniaxially oriented bilayers to determine membrane alignment, structure and tilt angles. Whereas alamethicin (20 residues) tends to adopt transmembrane orientations, shorter peptaibols such as zervamicin IIB, ampullosporin A and trichogin show increasing preference for orientations parallel to the membrane surface. In a second set of experiments, magic angle spinning solid‐state NMR spectroscopy experiments were performed. The CODEX experiments show the formation of oligomers in the membrane that increase in a concentration‐dependent manner. Whereas for alamethicin transmembrane dimers (P/L 1/30) and pentamers (1/13) were observed, trichogin occurs as in‐plane oriented tetramers (1/8). Interestingly, even the long alamethicin is oriented parallel to the membrane surface in a lipid composition mimicking bacterial membranes. These findings suggest that alternative mechanisms exist for the antimicrobial action of peptaibols.

Peptaibols of 11 to 20 residues were reconstituted in lipid bilayers of varying lipid composition and investigated by solid‐state NMR spectroscopy. Their membrane alignment is revealed by ¹⁵N spectra of static‐oriented samples and their oligomerization using ¹⁹F CODEX magic angle spinning spectra.

Abbreviations

Aib

α‐aminoisobutyric acid

ALM

alamethicin

ALM F50/5

alamethicin isomorph (sequence given in Table 1)

ALM F30/3

alamethicin isomorph (sequence given in Table 1)

ALM9

alamethicin analogue (sequence given in Table 1)

ATR FTIR

attenuated total reflection Fourier transform infrared

CD

circular dichroism

CODEX

centerband only detection of exchange

Dab(Tfa)

4‐trifluoroacetyl‐2,4‐diaminobutyric acid

DOPC

1,2‐dioleyl‐sn‐glycero‐3‐phosphocholine

DMPC

1,2‐dimyristoyl‐sn‐glycero‐3‐phosphocholine

DPhPC

1,2‐diphytanoyl‐sn‐glycero‐3‐phosphocholine

DPPC

1,2‐dipalmitoyl‐sn‐glycero‐3‐phosphocholine

EPR

electron paramagnetic resonance

IR

infrared

HPLC

high‐performance liquid chromatography

MAS

magic angle spinning

MD

molecular dynamics

MHC

major histocompatibility complex

NMR

nuclear magnetic resonance

OMe

methoxy

PC

1, 2‐diacyl‐sn‐glycero‐3‐phosphocholine

PE

1, 2‐diacyl‐sn‐glycero‐3‐phosphoethanolamine

PELDOR

pulsed electron–electron double resonance

PISEMA

polarization inversion spin‐exchange at the magic angle

POPC

1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine

POPE

1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphoethanolamine

POPG

1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phospho‐rac‐glycerol

P/L

peptide‐to‐lipid ratio

TFA

trifluoroacetic acid

TOAC

2,2,6,6‐tetramethyl‐piperidine‐1‐oxyl‐4‐amino‐4‐carboxylic acid

Peptides are highly dynamic and thereby hard to grasp by structural techniques. Membrane‐active peptides are particularly difficult to investigate because their conformation, membrane interactions, alignment relative to the membrane surface and their oligomerization state are strongly dependent on lipid composition, peptide‐to‐lipid ratio and other environmental factors. When interacting with each other, peptides and lipids form soft supramolecular ensembles where the components can adopt to each other and to the experimental conditions. Therefore, predictions or investigations of their structure, oligomerization and macromolecular arrangement remain challenging.

Whereas solution NMR approaches have been used to investigate peptide structures in membrane‐mimetic environments such as detergent micelles or isotropic bicelles, an understanding of the peptide–lipid interactions requires studies in lipid bilayers, at best in their liquid‐crystalline state. Whereas polypeptides are investigated in lipid vesicles by a variety of biophysical approaches including CD‐, fluorescence‐ and IR spectroscopies [1], solid‐state NMR spectroscopy is probably the only technique that can provide structural information in liquid‐crystalline bilayers at atomic resolution [2, 3, 4, 5]. Here, we review solid‐state NMR studies on peptaibols to gain insight about their structure, topology, oligomerization and dynamics in phospholipid bilayers and compare the findings to other hydrophobic polypeptides.

1. Peptaibols Are Antimicrobial Peptides and Models for Pore‐Forming Proteins

Peptaibols compose a family of peptides that are produced by fungi and that have their name from bearing several, non‐coded α‐amino isobutyric acid (Aib) residues. They are of different lengths encompassing 4 to 21 amino acid residues and end on a C‐terminal alcohol [6]. The peptides exhibit antimicrobial and, in some cases, antitumor activities [7, 8, 9] and, because of their resistance to proteolytic enzymes, have the potential to be developed into agents that are orally accessible [10]. Furthermore, they have been proposed to be used as crop protective agents [11]. As a consequence, the search for new peptaibols is ongoing [12, 13, 14, 15].

The best known representatives of the family are alamethicins (ALM), identified as early as 1967, whose membrane interactions have been studied extensively (reviewed in [8, 16, 17, 18, 19, 20, 21, 22]). The isoform ALM F50/5 is characterized by a linear sequence of 19 amino acids, of which eight are Aib residues, acetylated at the N‐terminus and a C‐terminal extension by an 1,2‐aminoalcohol (Phl) (making it a 20‐mer). This membrane‐active peptide antibiotic is of interest as it serves as a model for more complex antimicrobial sequences but also in analogy to larger voltage‐gated channels (reviewed in [16, 17]). Alamethicin has been studied extensively and serves as a paradigm to explore new biophysical methods [23, 24], including MD approaches such as enhanced sampling methods [25] or exploring the free energy profiles of ion conduction across oligomeric bundles [26]. In electrophysiological recordings, alamethicin causes successive increases into well‐defined conductance levels, which persist a few milliseconds [27, 28]. As many as 20 distinguishable conductance states have been described [29]. They have been interpreted to reflect transmembrane helical bundle channels of different sizes [16, 17], ranging from at least three [30], four [16] or five subunits [31]. In this model, the few polar side chains of this amphipathic peptide [Gln⁷, Gly¹¹, Gln¹⁸ (or Glu¹⁸) and Gln¹⁹ (or Glu¹⁹)] face the center of a pore and allow the passage of polar solutes, with positions 18 and 19 being at the very end of the sequence and probably not involved in the core of the channel. However, channel properties are affected by modifications of Gln⁷ [32], Gly¹¹ or Pro¹⁴ [33]. At high concentrations of alamethicin, also bulky compounds such as N ^α ‐benzoyl‐L‐arginine‐para‐nitroanilide or carboxyfluorescein are able to pass the membrane independently of an electric potential [34, 35, 36].

Zervamicin II and ampullosporin A are polypeptides also rich in Aib but of shorter length being composed of 15 and 14 amino acids, respectively, to which a C‐terminal amino alcohol is added. (i.e., a 16‐ and 15‐mer, respectively, Table 1). For comparison, they will be included in this review [37, 38].

TABLE 1.

Peptaibol sequences investigated and discussed in this paper. ^a

Trichogin GA IV	nOct‐Aib‐Gly‐Leu‐Aib‐Gly‐Gly‐Leu‐Aib‐Gly‐Ile‐Lol
[19F‐Phe10]‐trichogin GA IV	nOct‐Aib‐Gly‐Leu‐Aib‐Gly‐Gly‐Leu‐Aib‐Gly‐(19F)Phe‐Lol
[15N‐Gly2]‐trichogin GA IV	nOct‐Aib‐(15N)Gly‐Leu‐Aib‐Gly‐Gly‐Leu‐Aib‐Gly‐Ile‐Lol
Ampullosporin A	Ac‐Trp‐Ala‐Aib‐Aib‐Leu‐Aib‐Gln‐Aib‐Aib‐Aib‐Gln‐Leu‐Aib‐Gln‐Lol
Zervamicin IIB	Ac‐Trp‐Ile‐Gln‐Iva‐Ile‐Thr‐Aib‐Leu‐Aib‐Hyp‐Gln‐Aib‐Hyp‐Aib‐Pro‐Phl
ALM F30/3	Ac‐Aib‐Pro‐Aib‐Ala‐Aib‐Ala‐Gln‐Aib‐Val‐Aib‐Gly‐Leu‐Aib‐Pro‐Val‐Aib‐Aib‐Glu‐Gln‐Phl
ALM F50/5	Ac‐Aib‐Pro‐Aib‐Ala‐Aib‐Ala‐Gln‐Aib‐Val‐Aib‐Gly‐Leu‐Aib‐Pro‐Val‐Aib‐Aib‐Gln‐Gln‐Phl
ALM9	Ac‐Aib‐Pro‐Aib‐Ala‐Aib‐Ala‐Gln*‐Aib‐Dab(Tfa)‐Aib‐Gly‐Leu‐Aib‐Pro‐Val‐Aib‐Aib‐Gln*‐Gln*‐Phl

^a ALM, alamethicin; nOct, n‐octanoyl; Lol, leucinol; Aib, α‐aminoisobutyric acid; Iva, isovaline; Hyp, trans‐4‐hydroxyproline; Phl, phenylalaninol; Ac‐, acetyl, Gln*, Glu(OMe).

An even shorter peptaibol is the decameric trichogin GA IV from the fungus Trichoderma longibrachiatum [39, 40]. The hydrophobic character of trichogin is enhanced by the sequence ending in an amino alcohol and derivatization of the N‐terminus with an n‐octanoyl group. Because of their versatility, shortness and relative stability peptaibols have experienced some interest as a platform for drug development [41], including analogues with selective cytotoxic and antimicrobial activity [6, 36]. Trichogin has been shown active against several methicillin‐resistant strains of Staphylococcus aureus , to exert low hemolytic and potent antitumor activity against several cell lines [42].

To understand how trichogin can form pores despite being too short to span the membrane, a number of studies have been performed using, for example, NMR [39, 43], EPR [44, 45, 46, 47], fluorescence [48, 49, 50, 51, 52], X‐ray crystallography [53], scanning probe microscopy [54], neutron reflectivity [55], electrochemistry [56, 57, 58] and molecular dynamics simulations [55, 59]. The presence of three Aib residues has been shown to result in a mixed 3₁₀‐/α‐helical conformation under a variety of experimental conditions [36, 42, 49, 53]. Replacing Aib by Leu destabilizes the helical conformation concomitant with a reduction in bioactivity [60].

2. Solid‐State NMR Investigations of Peptaibols

Membrane association and insertion, folding into a secondary structure and self‐assembly into a membrane‐associated quaternary structure, are thought to be at the basis of channel formation and membrane leakage activities [16, 17]. Here, we will review solid‐state NMR data that allowed one to investigate the alignment of alamethicin and other peptaibols as a function of lipid composition and peptide‐to‐lipid ratio. Whereas the closely related amino acids composition, on first view, is suggestive that other peptaibols behave in the same manner, a stable transmembrane configuration is much more probable for alamethicin, which fits the hydrophobic thickness of many natural lipid bilayers, when compared to some of the shorter variants.

Therefore, we will also review data that were obtained with shorter variants such as ampullosporin, zervamicin or trichogin [37, 38, 61] and include an early study where the helix tilt of model peptides was measured as a function of membrane thickness. When being studied by membrane biophysical and structural approaches, it is important to choose an environment that presents the biological situation reasonably well. Solid‐state NMR spectroscopy is probably the only technique capable of investigating polypeptides in fluid liquid‐crystalline bilayers at concentrations that come close to the natural P/L ratio [62, 63].

3. Preparation of Peptides for Solid‐State NMR Studies

Notably, the quantitative preparation of such hydrophobic sequences carrying bulky Aib residues for structural studies is far from trivial. An additional difficulty arises because for solid‐state NMR experiments peptides need to be prepared carrying NMR isotopic labels such as ¹³C, ¹⁵N, ²H and/or ¹⁹F. The choice of isotope, the preparation method (chemical or microbiological) and the labelling strategy (specific, selective and uniform) are all intrinsically interlinked and determinants for the feasibility and resolution of the planned solid‐state NMR experiments. For the peptaibols presented here, three fundamentally different approaches were chosen.

On the one hand, the overexpression in microorganisms and their biochemical preparation provides means for polypeptide production. Whereas bacterial overexpression systems with E. coli are well established (e.g., [64] and references cited therein), the use of other microorganisms often needs much optimization. Therefore, peptaibols that are uniformly labelled with ¹⁵N were made available through collaborations, by isolation from microbiological cultures grown on culture media enriched in the isotopic labels. These approaches allow for the enrichment with ¹³C and ¹⁵N at reasonable cost, but it has to our knowledge not been used to introduce specific changes in the amino acid composition. While the methods are outlined shortly below, detailed information needed to reproduce the experiments can be found in the corresponding references.

Using a microbiological/biochemical approach alamethicin was isolated from Trichoderma viride secretions and purified by HPLC [65]. The peptide sequence and structure were analysed by multidimensional solution NMR spectroscopy and were found to contain a glutamine at position 18.

Ampullosporin A was prepared from a culture of Sepedonium ampullosporum HKI‐053 grown on ¹⁵N‐enriched medium (¹⁵N‐Celtone‐Powder 2) [38].

Zervamycin IIB was prepared from the Emericellopsis salmosynnemata, strain 336, IMI 58330 using a previously prepared ¹⁵N‐enriched biomass obtained from yet another microorganism [37].

On the other hand, alamethicin and trichogin peptides were prepared by the Padova laboratory headed by Claudio Toniolo using two very different chemical approaches. The chemical synthesis provides full flexibility in labelling specific residues and for modifying the primary sequence.

ALM9 was prepared by chemical synthesis in solution [66]. Utilizing a segment condensation approach, it is modular in nature, which allows the preparation of many variants by recombination of shorter fragments [67, 68]. ALM9 is an analogue of the natural alamethicin isoform F50/5 where valine 9 is replaced by the hydrophobic 4‐trifluoroacetyl‐2,4‐diaminobutyric acid (Dab(Tfa)) residue, which carries a CF₃ group for ¹⁹F solid‐state NMR spectroscopy [69, 70, 71, 72, 73] (Table 1). Furthermore, it carries methoxy glutamates (Glu(OMe)) instead of Gln at positions 7, 18 and 19, substitutions that do not affect the conformation and functionality of native alamethicin F50/5 [67, 74, 75, 76].

The trichogin GA IV peptides were prepared by solid‐phase peptide synthesis with good yield and purity including selectively labelled sequences for solid‐state NMR spectroscopy [77], a method also used for other peptaibols [78, 79]. The ¹⁵N isotopic label was incorporated to study the peptide alignment relative to the membrane surface, while ¹⁹F is used to determine the oligomerization number and intermolecular distances by CODEX MAS solid‐state NMR spectroscopy.

4. Solid‐State NMR to Investigate Membrane‐Associated Peptides

Solid‐state NMR spectroscopy works in liquid‐crystalline bilayers, which is particularly important when investigating pore‐forming peptides that can adopt many different conformations, topologies and oligomerization states. Whereas in solution NMR, interactions are averaged to their isotropic values by fast molecular motions, interaction anisotropies relative to the magnetic field direction of the NMR spectrometers persist in the solid or in semi‐solid states. Therefore, chemical shifts, dipolar and quadrupolar interactions result in broad and/or orientation‐dependent spectral line shapes.

Two fundamentally different solid‐state NMR approaches have been used to investigate the peptaibols presented in this review. First, from samples that have been oriented in an uniaxial manner relative to the magnetic field direction, it is possible to exploit this anisotropy and extract angular, that is, structural information [3, 80].

A second and more commonly used approach consists in fast rotation of the sample around the ‘magic angle’, which results in the mechanical averaging and elimination of the NMR interaction anisotropies. The resulting spectra resemble those obtained in isotropic solution albeit the intrinsic conformational and environmental heterogeneities result in considerably broader line widths.

In particular, combining NMR studies of uniaxially oriented samples and by magic angle sample spinning (MAS) is quite unique, which is due to the complexity of both methods and the need for distinct and dedicated equipment (e.g., [81]). Whereas the first approach has been shown to provide detailed information about the structure, topology and dynamics of membrane‐associated peptides, here it is presented together with MAS solid‐state NMR studies to reveal the oligomerization behaviour of the same peptaibols in fluid lipid bilayers.

5. Solid‐State NMR of Peptides in Uniaxially Oriented Membranes: Peptide Alignment

The topology of the peptaibols was investigated by static‐oriented ¹⁵N solid‐state NMR spectroscopy of peptides where one or several ¹⁵N nuclei were introduced in the sequence to replace the naturally abundant ¹⁴N (cf. above), a minor modification that has no effect on the physico‐chemical character of the sequences. Homogeneous solutions of the peptides and the lipids were prepared, much of the organic solvent removed by a stream of nitrogen gas, the viscous solution applied to ultra‐thin glass plates and the organic solvent completely removed. After hydration in a chamber of high relative humidity (typically 93%), the glass plates were stacked on top of each other, sealed, stabilized and inserted into dedicated flat‐coil solid‐state NMR probe heads [81, 82] with the sample normal parallel to the magnetic field of the NMR spectrometer (except when indicated differently). In contrast to many other biophysical techniques, this sample preparation offers a wide flexibility in changing the lipid composition and the physical state of the membrane and works with a considerable range of peptide concentrations.

From complementary investigations, it is known that the peptaibols adopt helical conformations in membrane environments [38, 83, 84]. In such cases, the ¹⁵N chemical shift measured from peptides reconstituted into uniaxially aligned bilayers is a direct indicator of the approximate angle of the ¹⁵N‐¹H bond amide bond vector relative to the magnetic field [80]. Furthermore, the ¹⁵N‐¹H bond is almost parallel to the helix long axis; thereby, the ¹⁵N chemical shift provides a direct measure of the helical tilt angle relative to the membrane surface with transmembrane helices resonating around 200 ppm and in‐plane oriented ones at < 100 ppm (schematically illustrated in Figures 1 and 2) [80]. Additional details can be extracted by measuring orientation‐dependent chemical shifts, dipolar and quadrupolar couplings [86, 87, 88, 89]. Furthermore, the spectral line shapes, intensities and exact chemical shift positions allow one to distinguish labelled sites that are part of structured helices or unstructured mobile regions [90, 91].

FIGURE 1.

Proton‐decoupled ¹⁵N solid‐state NMR spectra of U‐¹⁵N‐alamethicin reconstituted into uniaxially oriented lipid bilayers made of (A–D) POPC, (E) DMPC, (F, G) DPPC or (H) POPE/POPG 3/1 mole/mole. The P/L ratios and the temperatures are indicated next to the spectra. On top of each column, the correlation between ¹⁵N chemical shift and helix orientation is schematically illustrated for sample normal alignments parallel to B_o (i.e. not valid for panel D). Panels A and D are from [37], B, C, F and G from [76], E from [38] and H from [85].

FIGURE 2.

Proton‐decoupled ¹⁵N solid‐state NMR spectra of U‐¹⁵N ampullosporin A (i.e., labelled uniformly with ¹⁵N) reconstituted at P/L 1/100 into glass‐plate supported phosphatidylcholine bilayers. The sample normal is oriented parallel to the magnetic field direction of the NMR spectrometer. (A) POPC (C16:0, C18:1‐PC), (B) di C‐14:1‐PC, (C) di‐C12:0‐PC and (D) di‐C10:0 PC. On top, the correlation between ¹⁵N chemical shift and helix orientation is illustrated. From [38].

Alamethicin was prepared in microorganisms grown on medium enriched with ¹⁵N. Therefore, in this preparation, all amide bonds (and amines) occur in their ¹⁵N‐labelled form. The oriented ¹⁵N solid‐state NMR spectra of uniformly labelled alamethicin are shown in Figure 1. The spectra are from 19 amide bond vectors, all of them with close‐by chemical shifts; therefore, they are not resolved in the one‐dimensional ¹⁵N chemical shift spectrum, and a broad spectral line shape is observed. Spectra 1A–D are from alamethicin in POPC at P/L molar ratios ranging from 1/15 to 1/237, with A, B and D in the liquid‐crystalline state. POPC carries a fully saturated palmitoyl (C16:0) and an unsaturated oleoyl chain (C18:1) with a transition temperature T_m around 270 K. Therefore, spectrum C that was recorded at 248 K represents the same sample as B but in the gel state. The chemical shifts of A–C are all around 210 ppm indicative of transmembrane alignments of the alamethicin helix. Panel D exhibits a chemical shift of about 80 ppm, but in this case the sample was tilted by 90° and aligned with the normal perpendicular to the magnetic field direction. In this configuration, a transmembrane helix exhibits chemical shifts < 100 ppm. In a next step, other PC lipid compositions were investigated. Transmembrane alignments were also observed for liquid‐crystalline DMPC (Figure 1E, T_m = 297 K), in agreement with previous investigations using single‐site labels [84, 92], and in liquid‐crystalline DPPC (Figure 1F, T_m = 314 K). However, the alignment of alamethicin changed for gel‐state DPPC (Figure 1G). An alignment parallel to the membrane surface is also observed for a bacterial membrane mimetic, POPE/POPG 3/1 (mol/mol), hydrated at 93% r.h. and investigated at 310 K, that is, in the fluid state (T_m around 288 K [93]). The physical state of the membrane is validated by the chemical shift anisotropy visible in the ³¹P spectra of the phospholipids [85]. Interestingly, when CD‐ and solid‐state NMR spectroscopic data of novocidin or alamethicin reconstituted into bicellar systems are compared to those from vesicles, the structure and dynamics of the peptides were significantly different [94].

The well‐defined steps in electrophysiological recordings, the hydrophobic character of the peptaibols, together with the ensemble of biophysical and structural data have established alamethicin as paradigm for pores formed by transmembrane helical bundles of variable size [16, 17]. However, it is notable that other membrane lipid compositions or changes in experimental conditions favour helix orientations parallel to the surface (Figure 1G,H). This observation is in agreement with the two‐state model for alamethicin where in‐plane and transmembrane alignments have been observed in DiPhPC or DOPC membranes as a function of P/L and hydration [34, 83]. Interestingly, a transition from in‐plane to transmembrane has also been observed for the DQA1 hydrophobic (transmembrane) domain of the MHC II receptor [95]. In order to explore such in‐plane–TM equilibria further, we also investigated the membrane interactions of shorter peptaibols such as zervamicin IIB (16 building blocks), ampullosporin (15, Figure 2) or trichogin GA VI (11, Figure 3) using ¹⁵N solid‐state NMR spectroscopy. Whereas 20 building blocks of alamethicin make up a helix of an estimated length of ca. 3 nm, these smaller peptaibols are too short to span physiological membranes in a stable fashion. The palmitoyl or oleoyl fatty acyl chains of POPC, POPE and POPG phospholipids are considered to represent well the physico‐chemical properties of many lipid bilayers found in nature, with a hydrophobic thickness around 28 Å [96], and a mixture of saturated and unsaturated fatty acyl chains.

FIGURE 3.

Proton‐decoupled ¹⁵N‐ (A, C, E, G) and ³¹P‐ (B, D, F) solid‐state NMR spectra of [¹⁵N‐Gly2]‐trichogin in POPC membranes oriented with the sample normal parallel to the magnetic field direction at P/L = 1/120 (A, B), 1/20 (C, D) and P/L = 1/8 (E–G). The spectra were acquired at ambient temperature, except G where the temperature was set to 250 K. From [61].

¹⁵N solid‐state NMR spectra of ampullosporin reconstituted into PC phospholipid bilayers of variable thickness are shown in Figure 2. Whereas this peptaibol adopts an in‐plane alignment in POPC (C16:0, C18:1) and di‐14:1‐PC (Figure 2A,B), it is transmembrane when the hydrophobic thickness of the membrane is reduced (Figure 2C,D). The hydrophobic thickness of C12:0‐PC is estimated to be 20 Å, which fits reasonably well to the length of a helix made up of ≤ 15 units (≤ 23 Å). Similar observations are made with zervamicin IIB where a transition between in‐plane and transmembrane alignments is observed when shortening the fatty acyl chains from 14 to 10 carbons [37]. In such thin membranes, the transmembrane alignment of zervamicin becomes more homogeneous when increasing the lipid‐per‐peptide ratio.

Trichogin GA IV is made up of 10 amino acid residues and an amino alcohol at the C‐terminus (11‐mer). Thereby, its peptidic part is even shorter than zervamicin or ampullosporin (Table 1). Figure 3 shows ¹⁵N and ³¹P solid‐state NMR spectra of [¹⁵N‐Gly2]‐trichogin GA IV in POPC membranes at P/L ratios of 1/120 (A and B), 1/20 (C and D) and 1/8 (E–G). For the more dilute sample compositions, a sharp line at 79 ± 2 ppm is indicative of a well‐defined alignment approximately parallel to the membrane surface (Figure 3A,C). An additional broad intensity appears at the highest concentration reaching from 50 to 115 ppm with a maximum at 62 ppm (Figure 3E), which probably arises from oligomers. The narrow and broad signal contributions are well separated, showing that the two populations are in slow exchange at the time scale of the ¹⁵N chemical shift anisotropy (10⁻⁴ s). This intensity is broadened further when the sample temperature is lowered to 250 K (Figure 3G). The broad signal represents a distribution of alignments of the NH vector relative to the magnetic field direction, where the helix tilt covers angles from about 50° up to 90°, that is, predominantly parallel to the membrane surface. An alignment parallel to the membrane surface was also observed for harzianin VI in DMPC, another peptaibol made up of 11 residues [79].

For all samples, the most intense ³¹P NMR intensity occurs at 28 ppm (Figure 3, right column), confirming that the phospholipid bilayers are well oriented while the chemical shift anisotropy reaching out to −15 ppm agrees with phosphatidylcholine membranes in their liquid‐crystalline state [4, 97]. A more detailed analysis of how ³¹P solid‐state NMR line shapes represent membrane deformations and lipid dynamics is found in, for example, [98, 99]. Polarized ATR‐FTIR and CD spectroscopies confirm the helical conformation in the bilayer [100]. For all analogues, an average angle of about 60°–80° was determined for the helical domain regardless of the P/L ratio [61].

The response of peptides under hydrophobic mismatch conditions was also tested in a systematic study where a series of Ala‐Leu‐peptides covering between 10 and 30 hydrophobic residues were investigated in membranes of variable hydrophobic thickness (C10 to C22). The peptides were prepared with a ¹⁵N label at a central residue to allow ¹⁵N solid‐state investigations and two lysines at each terminus, which improve the solubility during the sample preparation and serve as interfacial membrane anchors [96].

The ³¹P solid‐state NMR spectra of the phospholipids show that the bilayer order is maintained as long as the hydrophobic helical domain of the peptides does not deviate by more than −3 Å (too short) or 14 Å (too long) from the calculated hydrophobic thickness. Beyond these limits, considerable disruption of the bilayer packing is observed where the peptides lose their transmembrane alignment. For peptides that ‘match’ the membranes within the above range (−3 to +14 Å), the ¹⁵N chemical shifts obtained from the peptide helices change in a systematic manner, indicating that their tilt angle adjusts to the membrane hydrophobic thickness in a continuous manner (Figure 4).

FIGURE 4.

The ¹⁵N chemical shifts measured from helical model peptides reconstituted in uniaxially oriented PC lipid bilayers are plotted as a function of the calculated hydrophobic mismatch. The peptides encompass a hydrophobic core of alanine‐leucine residues between 10 and 25 in length and the lipid fatty acyl chains between 10 and 22 carbons. The sample normal is parallel to the magnetic field direction. The calculated hydrophobic length of the peptide does not take into consideration a possible hydrophobic extension by the lysine side chains. The data obtained from peptides with a hydrophobic Ala‐Leu region extending over 10 (solid squares), 16 (solid triangles), 20 (solid circles) or 25 residues (open inverted triangles) as well as a peptide carrying 20 leucines (crosses) are shown. From [96].

A more detailed investigation of the energy contributions that determine the alignment of amphipathic helical sequences has been possible from measurements of histidine‐containing model peptides that change from in‐plane to transmembrane in a pH‐dependent manner [101]. This peptide family allowed to explore the effect of amino acid replacements [102, 103] and more recently also revealed strong effects from lipid saturation [104, 105].

Whereas the one‐dimensional ¹⁵N solid‐state NMR spectra from uniaxially oriented samples provide information about the approximate membrane alignment of the peptides, a more detailed analysis about the structure and tilt angle can be obtained from two‐dimensional local field experiments. In the best case, the NMR spectra are well resolved where each residue is characterized by orientation‐dependent ¹⁵N chemical shift/¹⁵N–¹H dipolar coupling correlations [106]. Polypeptide helices labelled at multiple sites reveal a wheel‐like arrangement of peaks with information on the helix tilt angle and its motional averaging [86]. To deduce structural details from such spectra, knowledge of the very details of the chemical shift tensors is essential, and much work has been invested in their characterization including for Aib residues [107, 108, 109].

The two‐dimensional PISEMA solid‐state NMR spectra from peptaibols in oriented membranes show ¹⁵N chemical shifts around 200 ppm and large dipolar couplings, indicative of transmembrane helical tilt angles (Figure 5). They are also characterized by relatively broad lines, indicating that the peptide structures are heterogeneous. The experimental spectra in Figure 5 were compared to simulations where the chemical shift/dipolar interaction positions were calculated from published structures at different helical tilt angles. Notably, the alamethicin x‐ray structural units A, B or C, a continuous α‐helix, a continuous 3.10 helix or a mixed α‐/3.10 helix were all tested to fit the spectra (Figure 5A) [38]. Whereas pure α‐helical structures do not explain important features of the spectra, a continuous 3.10 helix gives a fit of similar quality as a fit using Unit B of the crystal structure exhibiting a mixed α‐/3.10‐helical conformation (Figure 5A).

FIGURE 5.

Experimental and simulated PISEMA spectra of ¹⁵N uniformly labelled peptaibols. (A) Alamethicin reconstituted into oriented POPC at a peptide/lipid ratio of 1/100. Superimposed on the spectrum are the simulated peak positions of an ideal 3.10‐helix (Φ = − 50°, Ψ = − 31^o) at a 14° tilt angle. (B) Ampullosporin A reconstituted into oriented di‐C10:0‐PC at a peptide/lipid ratio of 1/100. The simulation shown is from an ideal 3.10‐helix at a tilt angle of 31.3^o. (C) Ampullosporin A in di‐C12:0‐PC at a peptide/lipid ratio of 1/75. The simulation shown is of a mixed α‐/3.10‐helical peptide at a tilt angle of 31.8° (the first eight residues are α‐helical and the last seven in a 3.10‐helical conformation). Spectra are taken from [38].

In the crystal structure of alamethicin, Units A and B show a 25°–35° kink at the proline‐14 site; thereby, in the simulated PISEMA spectrum, the N‐terminal helix aligns at about 8° while the C‐terminus exhibits a 20° tilt [38], in excellent agreement with another solid‐state NMR analysis of the N‐terminal domain using fluorinated Aibs [73]. A combined solid‐state NMR/MD analysis of the ¹³C chemical shift anisotropy of alamethicin, singly labelled with ¹³C at 11 carbonyls and reconstituted into DMPC at a 1/10 M ratio, is in agreement with an N‐terminal α‐helix, a kink followed by a C‐terminal 3.10 helix at somewhat larger tilt angles [110]. Notably, using four complementary orientational constraints of the Aib‐8 site, the alignment and dynamic nature of the peptide domain were characterized in considerable detail [111]. Related analysis tools to describe their dynamic nature were also developed and used for other membrane‐associated peptides including rocking and wobbling motions of the peptides [85, 87, 112, 113] or their rotational diffusion (and oligomerization) in the membrane normal [80, 114, 115, 116].

The two‐dimensional PISEMA spectra for ampullosporin A reconstituted into uniaxially oriented C10:0 or C12:0 phosphatidylcholine bilayers, respectively, are also shown (Figure 5B,C). The best fits are indicative of helical tilt angles around 30^o (i.e., transmembrane).

6. Solid‐State NMR Investigations of the Oligomeric State of Peptaibols

In order to get information about the oligomerization state of alamethicin, a ¹⁹F‐labelled analogue of alamethicin, ALM9 (Table 1), was reconstituted into POPC bilayers at a peptide‐to‐lipid ratio of 1/13. Under related conditions, a transmembrane helical alignment is observed for alamethicin by oriented ¹⁵N solid‐state NMR experiments (Figure 1A) [38, 76].

First, control experiments were performed to check the functionality of the ¹⁹F‐labelled alamethicin analogue. ALM9 exhibits somewhat reduced ability in calcein release experiments from DOPC/cholesterol 7/3 vesicles [117] when compared to ALM 50/5, [Glu(OMe)^7,18,19]ALM F50/5 or ALM analogues with Gln to Glu(OMe) substitutions [67, 74, 75, 76], suggesting differences in pore size, duration and frequencies [16, 17, 118].

Furthermore, in single‐channel electrophysiological recordings, similar pore conductance levels were observed for the ALM9 peptide (Figure S1 in [117]) when compared to alamethicin F50/5 or [Glu(OMe)^7,18,19]‐ALM F50/5 [119] albeit somewhat higher voltages were needed. These data suggest a closely related structure of the channels where the equilibrium between the open and closed states is shifted, while the high peptide‐to‐lipid ratio used in the centerband‐only detection of exchange (CODEX) experiment shown in Figure 6A,B favours the oligomeric state.

FIGURE 6.

(A) The squares show ¹⁹F CODEX experimental data for ALM9 in POPC vesicles at 240 K and 15 kHz MAS at a molar P/L of 1/13. The inset shows the S ₀ and S centerband intensities obtained with a mixing time of 600 ms. Fitting of the S/S ₀ decay using a biexponential where the equilibrium is indicative of oligomer size and the decay values of the ¹⁹F–¹⁹F distances (cf text for details). (B) MD simulation of the [Dab(TFA)9] analogue of Alm F30‐3 in its pentameric assembly [31] with numerical simulations of the resulting magnetization exchange. (C) Experimental data at P/L 1/30, the inset represents the data set at 400 ms mixing time. (D) Simulation of the data shown in C for a dimer. From [117].

7. Calibration of the CODEX Solid‐State NMR Experiment

The CODEX solid‐state NMR technique is based on the proton‐driven diffusion of ¹⁹F‐spin polarization under magic angle sample spinning (MAS) conditions. Although at room temperature the experiment has originally been set up to probe in a quantitative manner motional regimes on time scales of 1 ms or longer [120], when such motions are suppressed at low temperatures, the same experiment allows one to analyse the aggregate number and intramolecular distances up to 15 Å between ¹⁹F nuclei within oligomeric complexes [117, 121, 122]. During the experiment, polarization exchanges through the direct ¹⁹F–¹⁹F dipolar couplings, which allows one to determine the number of spins (N) that interact within the oligomer [121, 123]. The underlying theory and the equations used during data analysis (e.g., Figures 6 and 7) are presented in reference [121]. To take into account the dynamics and interaction properties of fluoromethyls, in a first step, calibration measurements were performed [122] using bis (trifluoromethyl)benzenesulfonyl chloride as a reference compound, which is characterized by a fixed 5.5 Å intramolecular distance [117]. The exchange of magnetization due to spin exchange causes the signal intensity S/S ₀ , to diminish as a function of the mixing time τ _m. The reference experiment (S₀) is conducted on the same sample under conditions under which spin diffusion is absent and the signal decay is governed solely by T₁ relaxation.

FIGURE 7.

(A) CODEX curve of [¹⁹F‐Phe10]‐trichogin GA IV in POPC at P/L = 1/8 (MAS 15 kHz, 240 K) and fit of the experimental data with an exponential function. The inset shows ¹⁹F spectra of S and S ₀ for the mixing time of 200 ms. (B) The best fit with a tetramer geometry and a nearest distance between spins of 9 Å (red curve) or 9.5 Å (green curve). From [61].

8. Size of Membrane‐Inserted Oligomers by CODEX Solid‐State NMR Experiments

The time‐dependent CODEX decay of ALM9 in POPC is shown in Figure 6A together with a representative set of spectra at a sample temperature of 240 K. The ¹⁹F spectra show spinning side bands in agreement with the full chemical shift anisotropy of about 48 ± 2 ppm (asymmetry parameter η = 0) similar to the static powder limit for this label [124].

It is important to suppress other exchange mechanisms, such as slow peptide motions, in order for the CODEX experiment to provide the oligomer size in a correct manner. For the set‐up under investigation, it has been tested that motional averaging sets in at temperatures around 262 K [117]. This is also the temperature limit when membrane‐associated water freezes [125], concomitant with motions of the amino acid side chains and the peptides as a whole. Therefore, at < 250 K, dipolar spin diffusion remains the predominant mechanism of magnetization exchange, while fast rotation of the CF₃ group assures that all three fluorine atoms are magnetically equivalent [126].

During the mixing time, the initial magnetization distributes between the different ¹⁹F within the oligomer, thereby reducing the CODEX echo intensity to 1/N where N is the number of magnetically non‐equivalent ¹⁹F‐nuclear spins [121]. Therefore, the equilibrium magnetization (S/S ₀ ) after extended mixing times is indicative of the subunits within the oligomer. The solid line in Figure 6A is an empirical fit using a bi‐exponential function S/S ₀ = 0.186 + 0.334 e^‐t/1ms  + 0.510 e^‐t/200ms where the plateau corresponds to N = 5.3 ± 0.4 subunits, that is, indicative of a pentamer.

In a second step, the S/S ₀ decay was simulated to obtain the distances r between the spatially different coordinates of the ¹⁹F nuclei within a symmetric pentamer (35). The fit represents distances to the next neighbour subunits at 9 Å and to two subunits that are further away [117]. Structural heterogeneity or asymmetry of the spatial distribution of ¹⁹F nuclei can explain the bi‐exponential nature of the decay. A good fit is also obtained from a Gaussian distribution of the inter‐spin distances. Furthermore, starting from the MD structure of the transmembrane pentamer in POPC [31], the simulation was continued for another 100 ps during which the ¹⁹F–¹⁹F distances were extracted and used to simulate the CODEX data (Figure 6B). The MD simulations and the experimental CODEX data are indicative that the peptide exhibits a heterogenous arrangement of helices and side chain conformations [31, 38, 127], which also reflect the electrophysiological properties of the resulting channels [128]. It should be noted that tetra‐ or pentameric oligomers are in average separated by several layers of lipids making random encounters within the 15–20 Å detection limit of the ¹⁹F–¹⁹F CODEX experiment negligible.

The CODEX experiments at reduced peptide‐to‐lipid ratios of 1/30 are indicative of a dimer with a 9 Å distance between the ¹⁹F labels (Figure 6C,D). An additional data point recorded at 235 K coincides within experimental error the curve obtained at 247 K, indicating that slow motions are sufficiently suppressed even at the higher temperature.

Finally, the oligomer size of [¹⁹F‐Phe10]‐trichogin GA IV in POPC bilayers was investigated at a peptide‐to‐lipid ratio of 1/8 (Figure 7A). Because rotations and flipping motions around bonds persist even at the low temperatures used here, it was necessary to first calibrate the overlap integral under the conditions of the experiment for the ¹⁹F‐Phe side chain incorporated into this peptide [61]. The S/S ₀ intensity reaches a plateau at around ¼ in agreement with a tetrameric structure. A mono‐exponential decay with a characteristic time constant of 160 ms fits to a nearest distance of 9–9.5 Å between subunits (Figure 7B).

9. Structural Models From the Solid‐State NMR Data

For peptaibols and other membrane‐active peptides, models have been proposed where equilibria interconnect a number of states. The peptides can be found in aqueous solution, aligned along the membrane surface or in a transmembrane fashion and occur as monomers or as oligomers of different size, where the latter have been proposed to form ion conducting helical bundles (Figure 8D). The distribution between states is governed by hydration, membrane phase, lipid composition, membrane thickness and the presence of other peptides [34, 37, 85, 129, 130, 131]. For the peptaibols under scrutiny, the membrane‐associated states have been shown to be characterized by a high degree of helicity (e.g., Figures 5 and 6).

FIGURE 8.

Models for peptaibols membrane interactions with increasing P/L. (A–D) Alamethicin configurations and oligomers observed in solid‐state NMR experiments. Panel B also defines the helical tilt angle. (E–G) Trichogin GA IV in POPC membranes at peptide‐to‐lipid ratios of (E) ≤ 1/100, (F) 1/20 and (G) 1/8. Panel F shows the distances of 15–40 Å measured by PELDOR EPR [44] (blue arrows), and Panel G demonstrates the four ≈9–10 Å distances that agree with the CODEX experiment (green arrows).

Based on oriented CD spectroscopy, it has been proposed that alamethicin switches from an in‐plane arrangement into the transmembrane alignment once a lipid‐dependent threshold concentration is reached [16, 132]. Indeed, previous publications including the experiments presented in Figure 1A–F show that in liquid‐crystalline phosphatidylcholine membranes, a transmembrane orientation is observed at P/L ratios ranging from 1/8 to 1/237 [37, 38, 76, 84, 85, 92]. With increasing P/L, oligomeric structures become more predominant and have been observed in the CODEX experiments (Figures 6 and 8C,D).

However, in gel phase DPPC or in POPE/POPG 3/1 mol/mol, the ¹⁵N solid‐state NMR spectra are indicative of alamethicin in‐plane alignments (Figures 1G,H and 8A). Whereas an alignment parallel to the surface has been considered an intermediate configuration during the insertion into PC membranes (Figure 8A–D), it seems a stable topology for POPE/POPG 3/1 (Figure 8A). Because the latter lipid composition is a good mimetic of bacterial membranes, it is well possible that alternative mechanisms for the alamethicin antibiotic activity are associated with the destabilization of the membrane integrity by such in‐planar helix orientations, similar to observations made with cationic amphipathic antimicrobial peptides [133]. The alignment parallel to the surface is possibly favoured by the smaller head group of PE when compared to PC, which leaves more space for and stabilizes the interfacial positioning of amphipathic peptides as suggested by the SMART model [133]. The latter was introduced to describe the multiple supramolecular arrangements of peptides and lipids as phase diagrams, including pore formation and membrane disruption.

Notably, not only the lipid head group composition but also the fatty acyl chain length makes an important energetic contribution with regard to the membrane topology of peptaibols. This is particularly evident when shorter peptaibols such as zervamicin (16 building blocks), ampullosporin (15) or thrichogin (11) are investigated in PC membrane of different hydrophobic thickness.

Alamethicin is long enough to span a POPC or DMPC membrane in a transmembrane configuration where it can associate in a step‐wise manner into pores made from oligomeric helical bundles of different sizes [16, 17]. The modest helical tilt angles of the MD structures shown in Figure 6B [31] agree well with oriented solid‐state NMR of ¹⁵N‐labelled alamethicin in POPC membranes [38] (Figure 5A).

Our CODEX experiments show that a pentameric structure forms at P/L 1/13 [38]. Notably, when alamethicin modified with TOAC spin labels at Positions 1, 8 or 16 [TOACⁿ, Glu(OMe)^7,18,19] has been investigated at 77 K in egg PC vesicles at molar ratios ranging from 1/70 to 1/200, an oligomer size of 4.3 ± 0.6 (tetramer) has been found [134, 135]. The MD simulations show considerable conformational fluctuations within the pentamer model represented in Figure 6B [31] in agreement with the data that suggest a rather dynamic behaviour of the membrane‐associated peptides at room temperature in agreement with a heterogeneous structure when frozen (Figures 5 and 6). This agrees well with functional assays such as electrophysiological single‐channel measurements, where different levels in conductivity are observed, suggesting that different oligomeric states exist within a membrane [16, 17]. It is therefore of interest that the CODEX data of alamethicin (Figure 6A,B) could also be explained by linear combinations of tetra‐, penta‐ and hexamers [117]. The energetic barrier between such different states is small; thus, changes in membrane composition, temperature, hydration or buffer composition may also modulate the detailed properties of the pores. At P/L of 1/30, dimeric structures predominate (Figure 6C,D), which are probably non‐conductive.

In contrast, the shorter peptaibols adopt in‐planar orientations also in PC membranes (Figures 2 and 3). For zervamicin and ampullosporin, it has been shown that by reducing the fatty acyl chain length to C12 or C10 transmembrane, alignments are observed also for these shorter peptaibols (Figure 2) [37, 38]. However, it should be noted that these thin membranes are made up of lipids carrying only fully saturated fatty acyl chains whereas the phospholipids making up the thicker bilayers carry an unsaturation. Therefore, the transition between in‐planar to transmembrane comes along with a change in fatty acyl chain saturation, which has more recently been found to have a pronounced effect on the topology of another helical peptide [104].

The shortest peptaibol investigated here adopts in‐planar alignments in POPC with little membrane disordering effects at high or low P/L (up to 1/20) in agreement with EPR [44, 46, 47, 136] and fluorescence spectroscopies [49, 50, 51]. The latter techniques also show penetration of the TOAC spin labels and of the fluorophores, respectively, into the hydrophobic part of the membranes. Above a threshold concentration of 1/30 to 1/20, the EPR data are indicative that the N‐terminus moves deeper into the membrane [44, 46] suggesting a rotation around the helix long axis when the oligomers form [61]. PELDOR experiments show an intradimer distance of about 25 Å for the most populated conformers, however, within a wide range of distances [44] (Figure 8F).

At the highest P/L of 1/8 investigated, the CODEX experiments agree with a trichogin tetrameric structure where the ¹⁹F‐Phe10 sites are about 9–9.5 Å apart (Figure 7) when at the same time a predominantly in‐planar alignment with considerable variation is observed (Figure 3G), which has been associated with a dimer of dimer arrangement (Figure 8G) [137, 138]. A tendency for dimer and higher oligomer formation has also been observed in MD calculations [139]. However, when these latter simulations were performed under the influence of high transmembrane voltages (200–400 mV) and at temperatures of 120°C, small conducting oligomeric barrels stave structures were also observed involving about a third of the trichogin peptides. Notably, not only has the voltage a pronounced effect by interacting with the polypeptide helix dipoles but the membrane thickness estimated from the water density was only 34.6 Å, which is considerably thinner than 42 Å for POPC found at 25°C [140].

Because the peptide concentration dependence of trichogin‐induced leakage of ions from liposomes suggests that three to four molecules are involved in pore formation [141], it is tempting to speculate that the tetrameric structure is indeed responsible for the membrane openings (Figure 8G). Notably, trichogin is about 200 times less active than the longer peptaibol zervamicin. Trichogin‐induced membrane leakage has also been demonstrated by fluorophore release experiments [39, 52, 55, 142] where leakage of large organic molecules and dextran polymers of about 10 kDa was observed [49]. Notably, when the membrane hydrophobic thickness is comparable to that of biological membranes, trichogin is about 30 times less effective than alamethicin. When compared to alamethicin, trichogin was also less active in natural cell membranes [28] whereas the difference diminished in membranes that match the length of helical trichogin [55] or the diameter of a tetramer (Figure 8G).

Notably, the ³¹P solid‐state NMR spectra indicate that the lipids are well oriented even by the highest trichogin concentrations (Figure 3B,D,F), which agrees with observations from other peptaibols [59, 143]. However, this observation is in contrast to the lipid disordering observed for cationic amphipathic peptides [98, 144, 145] that have been shown to reside within the interface [1, 146, 147]. Because the peptaibols lack the charged lysine or tryptophan amino acids, which often anchor at the membrane interface [148], trichogin has been positioned deeper inside the membrane interior (Figure 8E–G).

10. Conclusions

Two fundamentally different solid‐state NMR approaches applied to the investigations of peptaibols of variable length are presented in this review. These methods are unique in being performed in fluid‐ or gel‐state liquid‐crystalline bilayers providing structural information at atomistic resolution. On the one hand, the investigations of peptides reconstituted into lipid bilayers that are oriented relative to the magnetic field provide important information on their structure, membrane alignment and dynamics. Depending on the length of the peptide and the thickness of the membrane, predominant alignments either parallel to the surface or transmembrane orientations are observed, indicating that hydrophobic matching conditions make an important contribution. On the other hand, MAS solid‐state NMR experiments reveal the oligomeric association of peptaibols in membranes and provide complementary information on the secondary structure of the polypeptide chain.

Funding

This work was supported by Agence Nationale de la Recherche (Biosupramol 17‐CE18‐0033‐3, Naturalarsenal 19‐AMRB‐0004‐02, AmphiPep 20‐CE18‐0021, SAFEST 21‐CE18‐0043, AMLAA 21‐CE11‐0038‐02 and ANR‐10‐IDEX‐0002), COFECUB (88881.879071/2023‐01), Indo‐French Centre for the Promotion of Advanced Research (62T10‐1), Centre National de la Recherche Scientifique, Conseil régional du Grand Est and Université de Strasbourg.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available upon request from the corresponding author. The data are not publicly available because of privacy or ethical restrictions.