Orthogonal 19F-Labeling for Solid-State NMR Spectroscopy Reveals the Conformation and Orientation of Short Peptaibols in Membranes

Article abstract of DOI:10.1002/chem.201704307

Peptaibols are promising drug candidates in view of their interference with cellular membranes. Knowledge of their lipid interactions and membrane-bound structure is needed to understand their activity and should be, in principle, accessible by solid-state NMR spectroscopy. However, their unusual amino acid composition and noncanonical conformations make it very challenging to find suitable labels for NMR spectroscopy. Particularly in the case of short sequences, new strategies are required to maximize the structural information that can be obtained from each label. Herein, l-3-(trifluoromethyl)bicyclopent[1.1.1]-1-ylglycine, (R)- and (S)-trifluoromethylalanine, and ¹⁵N-backbone labels, each probing a different direction in the molecule, have been combined to elucidate the conformation and membrane alignment of harzianin HK-VI. For the short sequence of 11 amino acids, 12 orientational constraints have been obtained by using ¹⁹F and ¹⁵N NMR spectroscopy. This strategy revealed a β-bend ribbon structure, which becomes realigned in the membrane from a surface-parallel state towards a membrane-spanning state, with increasing positive spontaneous curvature of the lipids.

Full text of DOI:10.1002/chem.201704307

A Journal of
Accepted Article
Title: Orthogonal ¹⁹F-labeling for solid-state NMR reveals the
conformation and orientation of short peptaibols in membranes
Authors: Stephan L. Grage, Sezgin Kara, Andrea Bordessa, Véronique
Doan, Fabio Rizzolo, Marina Putzu, Tomáš Kubař, Anna
Maria Papini, Grégory Chaume, Thierry Brigaud, Sergii
Afonin, and Anne S. Ulrich
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201704307
Link to VoR: http://dx.doi.org/10.1002/chem.201704307
Supported by

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
1
9
Orthogonal F-labeling for solid-state NMR reveals the
conformation and orientation of short peptaibols in membranes
[
a]
[b]
[c]
[c]
[c]
Stephan L. Grage, Sezgin Kara, Andrea Bordessa, Véronique Doan, Fabio Rizzolo, Marina
[
d]
[d]
[c]
[c]
[c]
[a]
Putzu, Tomáš Kubař, Anna Maria Papini, Grégory Chaume, Thierry Brigaud,* Sergii Afonin,*
Anne S. Ulrich*[a][b]
Abstract: Peptaibols are promising drug candidates in view of their
interference with cellular membranes. Knowledge of their lipid
interactions and membrane-bound structure is needed for
understanding their activity and should be in principle accessible by
solid-state NMR. However, their unusual amino acid composition and
their non-canonical conformations make it very challenging to find
suitable NMR labels. Especially in the case of short sequences, new
strategies are required to maximize the structural information that can
be obtained from each label. Here, we have combined
lipopeptaibols (7 or 11 aa, e.g. trichogins).[1] Many of these
compounds exhibit cytotoxic activity, which is attributed to their
ability to permeabilize membranes and form pores.[
2,3,4,5,6]
To
decipher their mechanism of action, it is necessary to find out how
the molecules are bound to a lipid bilayer, i.e. their conformation
and their orientation relative to the membrane surface. Solid-state
NMR analysis of oriented samples is an established method for
obtaining such structural parameters in a quasi-native lipid
environment, and selective 19F-labels are the most sensitive and
versatile NMR probes for membrane-active peptides.[7] An
appropriate labeling strategy has to fulfill several requirements: (i)
a sufficient number of labels to describe the alignment of any
rigidly folded molecular segment with known conformation; (ii)
these labels have to point into different directions such that
orthogonal axes can be sampled; (iii) the labels have to be rigidly
connected to the peptide backbone in order to translate their local
orientations into the overall peptide alignment; and (iv) the labels
have to be compatible with solid-phase peptide synthesis (SPPS).
Long peptides (>20 aa) usually provide enough labeling sites, but
opportunities for suitable conservative substitutions are very
limited in short peptides/peptaibols.[7,8] It is important to avoid
trifluoromethylbicyclopentyl-L-glycine,
(R)-
and
(S)-
15
trifluoromethylalanine, and N-backbone labels, each probing a
different direction in the molecule, to elucidate the conformation and
membrane alignment of harzianin HK-VI. For the short sequence of
19
1
1 amino acids we obtained 12 orientational constraints using F- and
1
5
N-NMR. This strategy revealed a β-bend-ribbon structure, which
becomes re-aligned in the membrane from a surface-parallel state
towards membrane-spanning state with increasing positive
a
spontaneous curvature of the lipids.
Introduction
labels that point into similar directions, because this makes the
orientational constraints partially redundant.[
9,10]
Another critical
Fungi such as Trichoderma are producers of pharmaceutically
attractive peptaibols that are endowed with improved biostability
due to a high content of unconventional amino acids such as
aminoisobutyric acid (Aib). The family of pept”Aib”ols can be
grouped into long peptaibols (18-20 amino acids (aa), such as
alamethicins), short peptaibols (11-16 aa, e.g. harzianins), and
aspect in the case of peptaibols is their uncertain secondary
structure, which requires extensive labeling to resolve the
underlying backbone scaffold.
To address these challenges for the important class of short
peptaibols, we report and validate a strategy that maximizes the
amount of structural information that can be obtained from a small
number of labelled sites. This approach is based on the use of
complementary NMR reporter groups that provide non-redundant,
[a]
Dr. S. L. Grage, Dr. S. Afonin, Prof. A. S. Ulrich
“
orthogonal” NMR structural constraints for one and the same
Institute of Biological Interfaces (IBG-2), Karlsruhe Institute of
Technology (KIT), POB 3640, 76021 Karlsruhe (Germany)
E-mails: Sergiy.Afonin@kit.edu, Anne.Ulrich@kit.edu
position within a peptide sequence.
Here, we have combined four orthogonal NMR probes: (R)-
and (S)-trifluoromethylalanine [(R)-CF
3
-Ala, (S)-CF
3
-Ala][
8,11]
are
[
b]
c]
Dr. S. Kara, Prof. A. S. Ulrich
Institute of Organic Chemistry (IOC), KIT, Fritz-Haber Weg 6, 76131
Karlsruhe (Germany)
used to replace each Aib side chain one-by-one, while 3-
(trifluoromethyl)-bicyclopent-[1.1.1]-1-yl-glycine (CF
-Bpg)[12] in
3
combination with 15N-amide[13] in a suitable backbone position are
used to replace single leucine or isoleucine residues (Figure 1A).
This set of labels fulfils all criteria above: (i) it provides a chemical
variety that allows conservative substitutions of many sites
[
Dr. A. Bordessa, V. Doan, Dr. F. Rizzolo, Prof. A. M. Papini, Dr. G.
Chaume, Prof. T. Brigaud
Laboratoire de Chimie Biologique (LCB), EA4505, Plateform
PeptLab@UCP, Université de Cergy-Pontoise, 5 Mail Gay-Lussac,
Neuville sur Oise, 95000 Cergy-Pontoise cedex (France)
E-mail: Thierry.Brigaud@u-cergy.fr
including Aib; (ii) the orthogonal labels (S)-CF
3
-Ala, (R)-CF
3
-Ala),
1
5
CF
3
-Bpg and amide- N sample four distinctly different director
[d]
M. Putzu, Dr. T. Kubař
axes (Figure 1B); and (iii) they are rigidly connected to the peptide
backbone. Regarding (iv), routine and efficient SPPS has been
Institute of Physical Chemistry (IPC) & Center for Functional
Nanostructures (CFN), KIT, 76131 Karlsruhe (Germany)
demonstrated for CF
coupling reactions were difficult in the case of the -disubstituted
3
-Bpg incorporation.[12] However, the peptide
Supporting information and the ORCID identification number(s) for
the author(s) of this article can be found under http://
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
CF
3
-Ala isomers due to the low nucleophilicity of the nitrogen
enantiomerically pure CF
3
-Ala were prepared and linked to the
atom.[
8,11,14]
This problem has been bypassed by preparing in
preceding amino acid (Leu/Ile) to yield dipeptide building blocks
suitable for automated Fmoc-synthesis, following specific
protocols (see Supplementary Information, SI). In addition, four
selectively 15N-labeled HZ peptides were prepared with either 15N-
Leu or 15N-Ile in one of the positions 3, 4, 7, and 8.
solution enantiomerically pure dipeptide building blocks, in which
Leu or Ile is coupled to CF
3
-Ala.[15] The dipeptides, including the
1
9F-labeled Aib analogues could then be successfully introduced
into peptaibols using microwave-assisted SPPS.
Figure 2. Amino acid sequence of harzianin HK-VI, illustrating the employed
labelling scheme. Dashed boxes indicate the positions of the native amino acids,
which were replaced by NMR labels, color symbols refer to the type of NMR
1
5
19
3
label used in the marked positions:blue - N, magenta - F in (S)-CF -Ala, cyan
-
19
19
3 3
F in (R)-CF -Ala, red - F in L-CF -Bpg).
Before using the 19F-labeled analogues for NMR analysis, we
examined their conformation by circular dichroism (CD) in
zwitterionic detergent micelles (Figure 3); similar results were
obtained in other membrane-mimics, such as 50% 2,2,2-
trifluoroethanol (TFE), micelles of sodium dodecyl sulfate (SDS),
n-dodecylphosphocholine (DPC), lysoMPC (1-myristoyl-2-
hydroxy-sn-glycero-3-phosphocholine), and in unilamellar
vesicles composed of DMPC (1,2-dimyristoyl-sn-glycero-3-
phosphocholine).
Figure 1. NMR labels used in this study. (A) Native amino acids (grey
background) and used isostructural substitutions (with NMR reporter groups
15
19
colored: N – blue, F – magenta in (S)-CF
in CF -Bpg). (B) Sampled directions in space of the employed labels (colors as
above) with respect to an ideal -helical backbone.
3 3
-Ala, cyan in (R)-CF -Ala, and red
3
As a representative short peptaibol, we selected here the
natural 11mer harzianin HK-VI (HZ) with the sequence Ac-Aib-
Asn-Ile-Ile-Aib-Pro-Leu-Leu-Aib-Pro-Leu-ol (Figure 2).[2,16] Even
though the conformation and membrane alignment of the
relatively long alamethicin has been extensively characterized,
short peptaibols have been studied far less. A single experimental
structure has been reported so far (on the 14mer harzianin HC-
IX), revealing an unusual helical β-bend ribbon spiral fold in
methanol.[1] The same non-extended conformation has been
suggested for HZ,[16] but without resolving the actual backbone
structure. To date, there is no structure available of any harzianin-
type peptiaibol in its functionally relevant membrane-bound state.
Results and Discussion
To introduce the selective 19F-labels into HZ (Figure 2), we used
microwave-assisted SPPS to synthesize four HZ analogues with
Figure 3. Representative CD spectra of selectively 19F-labeled HZ analogues in
membrane-mimicking environments. (A) Analogues with CF
3
-Ala in the
CF
3
-Bpg in one of the positions 3, 4, 7, and 8, and six analogues
-Ala in positions 1, 5, and 9. For the latter ones,
presence of lysoMPC micelles, and (B) analogues with CF -Bpg in the presence
3
with (R)- or (S)-CF
3
of DPC micelles. The peptides were mixed with preformed micelles in 10 mM
phosphate buffer, pH 7.4, at a peptide/detergent molar ratio of 1/200.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
All analogues gave the same CD lineshape as the wild type,
hence the secondary structure remained unperturbed. Only the
Ac-Aib substituted analogue showed significant deviations
resolved by liquid-state NMR the structure of HZ in 50% TFE, as
well as in DPC and in SDS micelles.[20] Furthermore, two
computational models were included, which we obtained as
averaged structures from two independent unrestrained all-atom
molecular dynamics (MD) simulations of HZ in a DMPC bilayer on
a µs time scale.[21] (The backbone torsion angles φ/ψ of all models
are listed in the SI Table S4).
Solid-state NMR structure analysis of membrane-bound
peptides performed in macroscopically oriented samples and data
analysis is based on the anisotropic 15N chemical shift and the
1
presumably due to unfolding at the N-terminus. For this reason,
1
position Aib was excluded from further NMR analysis.
Notably, the CD spectra do not represent any common
structural motif such as -helix or β-strand, neither do they
correspond to a random coil.[17] Also a dynamic transition between
random coil/α-helix or a mixture thereof seems unlikely, given the
similarity of the CD lineshapes in all the different environments.
The peptide thus seems to adopt a unique and rather stable
secondary structure. Similar CD spectra have been previously
reported for peptaibols folded as 310-helices[18] and β-bend ribbon
spirals.[19]
1
9
19
F- F dipolar coupling within the CF
orientation-dependent 15N chemical shifts δ
from the signal position in the 15N-NMR spectra of 15N-labeled HZ
(Figure 5A). The orientation-dependent intra-CF dipolar
couplings correspond to the triplet splittings ΔFF in the 19F-NMR
3
-reporter group. The
N
are directly available
3
19
spectra of the F-labeled analogues (Figure 5B,5C). The
orientation of the 15N labeled moiety, given by two polar angles α
and β relating the CSA principal axis frame to the membrane
normal, can be deduced from:
2
2
δ
N
= δiso + Smol
δ
CSA (3 cos β – 1 + η sin β cos 2 α)/2,
(1)
with the isotropic chemical shift δiso = 111 ppm, the asymmetry
parameter η = -0.245, and the chemical shift anisotropy δCSA = 98
ppm. Similarly, the orientation of the CF -axis with respect to the
3
membrane normal is related to the splitting by:
2
ΔFF = Smol
Δ
0
(3 cos θ – 1)/2,
(2)
with the maximum splitting Δ
0
= 16 kHz. Partial averaging of the
anisotropies by motion of the peptide was taken into account by
scaling the anisotropic contributions by an order parameter Smol
.
For the analysis of the peptide orientation, the value of the 15
N
N
chemical shift δ or the splitting ΔFF was predicted for each label
within a certain group of labels (using all labels or subsets thereof).
For this, any given backbone model (or segment thereof) is placed
into the membrane coordinate system in an orientation that is
described by two angles  and  (see SI Figure S2 for a graphical
definition). Motional averaging is taken into account by scaling
and the order parameter Smol. To identify the best-fit orientation
(i.e. particular combination of ,  and Smol), we numerically
screened all possible combinations of ,  and Smol values and
Figure 4. Structural models of HZ that were examined in the solid-state NMR
data analysis. Backbone atoms and proline carbons are displayed in the stick
representations (carbon – white, nitrogen – black, oxygen – light grey). The red
stretch in the ribbons marks the fragment Ile -Aib that was experimentally
covered.
3
9
calculated δ
N
or ΔFF for all considered labels. The difference
To analyze the membrane-bound peptaibol by solid-state
NMR, is necessary to start off with a well-defined backbone
structure. The orientation-dependent NMR data will then be
iteratively evaluated in terms of both, the conformation and the
alignment of the molecule within the membrane. Since CD did not
provide a concrete secondary structure that could be used a
model for the backbone, we generated a set of eight plausible
structures (Figure 4) and tested whether they would be able to
self-consistently predict the solid-state NMR data. As 3D models
we considered: a continuous ideal -helix (φ/ψ angles of -57°/-
between the calculated and experimental NMR parameters was
quantified by a root of mean square deviation (rmsd).
47°), a 310-helix (φ/ψ angles of -74°/-4°) and a β-bend ribbon spiral.
The latter was generated as a homology model based on the
liquid-state NMR structure of the 14mer harzianin HC-IX.[1] This
peptaibol contains three similar Xxx-Yxx-Aib-Pro repeats (Xxx,
Yxx = arbitrary amino acid), and the φ/ψ values averaged over all
three repeats were used to model the corresponding “Ile-Ile-Aib-
Pro” and “Leu-Leu-Aib-Pro” segments of HZ. In addition, we have
Figure 5. Solid-state 15N- and 19F-NMR spectra of HZ in oriented 1,2-
diphytanoyl-sn-glycero-3-phosphocholine (DPhPC) bilayers, at 35°C.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
The first, most straightforward attempt of using all eight 19F
labels at the same time in the data analysis resulted in a massive
deviation of the predicted NMR parameters from the experimental
data. This was the case for all of the 8 different structural models,
giving errors well beyond the acceptable margins (see below).
Therefore we decided not to use all NMR data simultaneously in
the analysis, but only a subset of the labels representing one type
of isotope label or a certain region within the sequence of HZ. We
performed such data analyses several times, each time using a
different selection of labels that we refer to as label “patterns”. In
particular, we analyzed the two peptide segments on either side
labels (pattern F3478), on the other hand, allows us to reduce the
set of compatible models to -helix, β-bend ribbon spiral, and the
conformation in DPC micelles. The set of models can be narrowed
down further using the two groups of 19F-labels representing the
N- and C-terminal segments of HZ. Considering the rmsd values
of pattern F345 and F789, both of them clearly favor the β-bend
ribbon spiral as the only model matching the solid-state NMR data
(Table 1). This structure also produces very low rmsd for the other
patterns (N3478, F3478, and F59), hence the β-bend ribbon spiral
can be identified as the most likely backbone structure of HZ in
DPhPC membranes. The same conclusion is reached when
examining the combined set of data for HZ reconstituted in other
lipids (see below, and SI Tables S6-S8). We can clearly rule out
the 310-helix, which had earlier been discussed for Aib-rich
6
of Pro separately, while the other patterns covered the full length
of HZ. Each pattern thus contained four orientational constraints,
which is the mathematically required number of independent
parameters to fully determine 3 structural variables (, ρ, and Smol). structures and would have been compatible with the CD
These patterns include: all four 15N-labels in positions 3, 4, 7, and
analysis.[18,22] Based on these results, we can now address and
interpret the alignment of HZ in different lipid bilayers, by focusing
on the orientation of the β-bend ribbon spiral that is given by the
corresponding , ρ and Smol values.
8
8
5
(pattern N3478); all four CF
(pattern F3478); all four (R)- and (S)-CF
and 9 (pattern F59); the CF -Bpg and CF
3
-Bpg labels in positions 3, 4, 7, and
-Ala labels in positions
-Ala of the N-terminal
3
3
3
half in positions 3, 4, and 5 (pattern F345); as well as the
corresponding CF -Bpg and CF -Ala in the C-terminal half in
positions 7, 8, and 9 (pattern F789).
Two physical properties of membranes, namely the
spontaneous lipid curvature and the bilayer thickness, are known
to have a profound influence on peptide orientation.[
We thus selected a series of phosphocholine systems allowing to
examine both, decreasing thickness and increasing curvature
(Figure 6): DPhPC, DMPC, 1,2-didecanoyl-sn-glycero-3-
phosphocholine (DDPC), and a DMPC/lysoMPC mixture (molar
ratio 2/1).
To evaluate and compare the orientation of HZ in these
different lipid environments, additional NMR data was collected of
all labeled analogues in oriented bilayers of the named lipid
selection (see SI Figures S3-S6 for the spectra). These data were
analyzed in terms of the HZ alignment, i.e. as the best-fit values
3
3
23,24,25,26,27]
In this way, we first validated the backbone structure and
subsequently investigated more thoroughly the orientation of HZ
in the lipid bilayer. When fitting the 8 different structural models,
we obtain a best-fit value for each, though with different rmsd
quality (see Table 1). Based on these numbers it is now possible
to identify certain backbone models that match the experimental
NMR data well, and to rule out others. If we accept all rmsd values
below the experimental error as plausible, only few structures
remain in agreement with the experiments.

, , Smol for the β-bend ribbon spiral. Again, these analyses were
Table 1. Quality of the best-fits to the NMR data, used to assess different
structural models (see Figure 4) for HZ bound to DPhPC bilayers. The
patterns of labels used in each analysis are indicated at the top.
performed several times, each time including a different label
19
15
pattern of F-labels (patterns F3478, F59, F345 and F789) or N-
labels (N3478). The , , Smol values resulting from the individual
label patterns were averaged and are summarized in Table 2.
15
19
Experimental errors are 5 ppm and 1 kHz for the N chemical shifts and F-
19
F dipolar couplings, respectively. The rmsd values within these margins
are shown in bold.
N3478
rmsd/ppm
4.68
F3478
F59
F345
F789
Table 2. Membrane alignment of HZ in its -bend ribbon spiral conformation
rmsd/kHz
(averages over the best-fit results of all patterns, see SI Tables S9-S12 for
the non-averaged values), in oriented bilayers of lipids with different
spontaneous curvature and thickness.
α-helix
0.44
2.46
0.40
0.51
1.75
0.26
2.79
3.21
0.52
1.98
2.27
0.66
310-helix
2.38
DPhPC
DMPC
DDPC
DMPC/lysoMPC
151°±17°
β-bend ribbon
1.15

91°±29°
208°±15°
0.66±0.10
77°±12°
216°±14°
0.68±0.21
103°±30°
222°±17°
0.38±0.15
NMR(DPC)
NMR(SDS)
NMR(TFE)
MD model A
MD model B
2.20
2.75
2.79
2.63
6.20
0.30
1.80
2.79
3.08
2.97
0.44
0.14
0.56
2.00
2.00
1.65
2.46
1.70
1.30
0.15
1.46
1.50
1.70
2.34
2.04

S
285°±26°
mol
0.34±0.12
The compact and rather hydrophobic peptaibol HZ has 
values of around 80°-90° (relative to the membrane normal) in
DPhPC and DMPC, corresponding to an orientation of its long
molecular axis parallel to the plane of the membrane. Such
alignment is typically found for distinctly amphipathic molecules,
such as cationic -helical antimicrobial peptides.[ Hydrophobic
helices, on the other hand, typically adopt a transmembrane
orientation.[27,28]. Interestingly, we find that with decreasing
membrane thickness and increasing positive spontaneous
On their own, the 15N-NMR data are not conclusive in deciding
the backbone structure, as most rmsd values are below or near
the threshold given by the experimental error of ~5 ppm (pattern
23]
N3478). Likewise, the analysis of all CF
is not conclusive, as only the 310-helix and MD models can be
ruled out as inconsistent candidates. Combining all CF -Bpg
3
-Ala labels (pattern F59)
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
curvature, the tilt angle of HZ decreases to 100° in DDPC and to
with an unusual conformation. Here, we observed an orientation
of HZ parallel to the bilayer surface in the majority of lipid systems
investigated, and MD simulations imply that the depth of insertion
is deeper than for the typical amphiphilic -helices. This alignment
nonetheless agrees with the carpet mechanism that is considered
to be one of several modes of action of membranolytic peptides:
the binding and accumulation of peptides - when embedded near
the polar/apolar interface - causes membrane thinning and/or
micellization.[31] However, we also observed a considerable
preference of HZ to tilt upright into the membrane in the presence
of lipids with positive curvature or reduced thickness. Hence, the
propensity of HZ to act via the conventional carpet mechanism
may be questioned, because the surface activity of the peptide
becomes reduced in the inserted state. Whether this ability to
insert into the membrane is sufficient for HZ to assemble into
oligomeric pores similar to those suggested for longer
amphipathic peptides (or for the long peptiaibol alamethicin)
should be debated in the light of its rather small size and its
observed higher mobility in the inserted state. Given the much
higher hydrophobicity of HZ as compared to typical cationic
amphipathic antimicrobial peptides, in conjunction with its small
size, the deeper penetration of HZ may be more effective and by
itself sufficient to destabilize the lipid bilayer. The balance
between a surface-parallel and an inserted alignment could also
indicate a high ability of HZ to cross the lipid bilayer, allowing it to
reach potential other targets inside the cell, different from the
cellular membrane.
150° in DMPC/lysoMPC, getting closer to an upright
transmembrane orientation (Table 2, Figure 6). The peptide
dynamics, too, reveal a clear change with the bilayer properties.
High order parameter values of around ~0.7 are observed in
DPhPC and DMPC, comparable to values typically found for α-
helical surface-bound peptides.[29] These values decrease for HZ
to ~0.4 in thin DDPC and DMPC/lysoMPC, where the peptide
becomes even more mobile. The re-alignment of harzianin in
highly curved DMPC/lysoMPC bilayers compares well with similar
findings on amphiphilic antimicrobial peptides. In lipids with a
highly positive spontaneous curvature they can flip into a
transmembrane state and assemble as a stable pore.[25,26] In the
case of HZ, both, surface-parallel and perpendicular alignments
have been recently reported from MD simulations in DMPC,
where a deep penetration of this hydrophobic molecule towards
the core of the bilayer was observed.[21]
Conclusions
We have demonstrated and validated a general solid-state NMR
approach, suitable to determine the structure and orientation of
membrane-bound peptides. It is particularly relevant for
sequences that offer only few positions for labeling, due to either
a short length or an unusual amino acid composition and/or
conformation. We used four orthogonal 19F- and N-NMR labels
to sample different spatial orientations, which allowed us to collect
several complementary NMR constraints even from a single
labeled position in the peptide sequence. This way, the backbone
structure and different alignment states of the peptaibol harzianin
HZ VI could be determined, overcoming the special challenges
imposed by the short sequence and high content of Aib of this
peptide. Despite its hydrophobic character and its short length,
harzianin HK-VI exhibits an alignment behavior that is also
observed for longer and distinctly amphipathic sequences, such
as -helical antimicrobial and pore-forming peptides.[21] In
membranes with negative or low spontaneous curvature harzianin
HK-VI is preferentially aligned parallel to the bilayer surface, and
it becomes tilted in lipids with high positive curvature. The short
non-lipidated peptaibol may thus exhibit a lower membrane-
perturbing activity via a carpet mechanism, as described for many
membrane-perturbing peptides, but it may rather act by
penetrating more deeply into the bilayer, and possess an
enhanced ability to cross the membrane.
15
Figure 6. Orientation of HZ in bilayers of lipids with different spontaneous
curvature and thickness as determined in this study (A). The polarity of HZ lies
between the more polar typical antimicrobial peptides such as PGLa and
hydrophobic transmembrane sequences such as the oncoprotein E5.[28,30] (B).
Color encodes increasing polarity from yellow (hydrophobic)
intermediate) – blue (polar).
– green
(
The observed re-alignment of HZ in bilayers with different
lengths and spontaneous curvatures can provide new
mechanistic insights into the membrane-perturbing activity of HZ.
Compared to many well-known antimicrobial peptides, harzianin
is quite short and more hydrophobic, besides being a peptaibol
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
7
+
1
253.68; Calcd. mass for C60
H
99
F
3
N
12
O
13 (Leu /CF
3
-Bpg) [m+H] : 1253.75,
Experimental Section
8
obs. mass (m/z): 1253.68; Calcd. mass for C60
H
F
99 3
12
N O
13 (Leu /CF
3
-Bpg)
+
[m+H] : 1253.75, obs. mass (m/z): 1253.70.
Peptide synthesis. The synthesis of (R)/(S)-α-trifluoromethylalanine
containing dipeptide building blocks for incorporation in HZ sequences was
performed starting from unprotected enantiopure (R)- and (S)-CF -
3
alanines which were synthesised according to our previously reported
procedure involving a Strecker-type reaction on (R)-phenylglycinol derived
trifluoromethyloxazolidines.[14] The following variations in synthetic
procedures for the assembly of the dipeptides and characterisation of the
compounds are detailed in the SI.
The 15N-Leu-containing HZ analogues were synthesised following the
same procedure as for unlabeled HZ, except that they were synthesised
102 1
H N O
1115N 13 ({15N}Leu7)
on the 0.15 mmol scale. Calcd. mass for C58
[m+H]+: 1176.77, obs. mass (m/z): 1176.68; Calcd. mass for
58 102 1
C H N O
1115N 13 ({15N}Leu ) [m+H] : 1176.77, obs. mass (m/z): 1176.64.
8
+
The 15N-Ile-containing HZ variants were synthesized using a Liberty Blue
CEM automated peptide synthesizer, starting from H-L-Leucinol-2-
Unlabeled HZ was synthesized on a 0.25 mmol scale, using a SyroWave
Biotage automated peptide synthesizer starting from H-L-Leucinol-2-
chlorotrityl resin (loading: 0.5 mmol/g, 500 mg). Fmoc-Asn was coupled
without any protecting group on the side chain. A double coupling was
performed for each amino acid. Coupling reactions were accomplished
with 3.0 equiv. of amino acid, 3.0 equiv. of HATU in DMF, and 4.2 equiv.
of DIPEA in NMP, for 10 min at 70 °C, except for the Aib couplings, which
were performed in two steps. The first step was performed at 55 °C for 15
min, and the second one was performed at 60 °C for 15 min. All Fmoc
deprotections were carried out using 20% piperidine in DMF for 3 min and
15
chlorotrityl resin (loading: 0.5 mmol/g, 500 mg ( N-Ile at position 3) or 280
mg (15N-Ile at position 4)). A double coupling was performed for each
amino acid. Coupling reactions were accomplished with 2.0 equiv. of 15N-
Fmoc-Ile-OH in DMF (0.1 M), 4.0 equiv. of amino acid in DMF (0.2 M), 4.0
equiv. of Oxyma pure in DMF (1.0 M), and 4.0 equiv. of DIC in DMF (0.5
M), for 2 min at 90 °C, except for the Fmoc-Aib-OH (at 50 °C for 10 min).
Fmoc deprotections were carried out using 20% piperidine in DMF for 3
min at 75 °C, except for the Fmoc-Aib-OH (2x 10 min at 25 °C). The
remainder of the synthesis followed the protocols used for unlabeled HZ.
102 1
Calcd. mass for C58H N O
1115N 13 ({15N}Ile ) [m+H] : 1176.77, obs. mass
(
3
+
12 min at room temperature. N-terminal acetylation of the peptide resin
102 1
m/z): 1177.14; Calcd. mass for C58H N O
1115N 13 ({15N}Ile ) [m+H] :
4
+
was performed by a repeated treatment with acetic anhydride (2.5 mmol)
and NMM (2.5 mmol) in DCM for 1 h at room temperature. The 1,2-
aminoalcohol peptide was cleaved from the resin upon four treatments with
1176.77, obs. mass (m/z): 1177.21.
3
0% HFIP in DCM for 30 min. The filtrates were combined and evaporated
The yields, purities, chromatography and mass spectroscopy data of the
synthesized peptides are shown in SI.
to dryness providing the crude peptide. The product was precipitated by
addition of cold diethyl ether and then lyophilized. The crude peptide was
purified by semi-preparative HPLC, using a Phenomenex Jupiter C18 (10
x 250 mm, 10 µm, flow: 4 mL/min) column, to yield the harzianin HK-VI (35
mg, 12 %, purity > 98%). The gradient elution system was composed of
Circular dichroism spectroscopy. The peptides were prepared as 1 mg/mL
stock solutions in MeOH. Aliquots containing 0.1 mg peptide were
transferred to 2 mL microcentrifuge tubes (LoBind Eppendorf), from which
solvent A (0.1 % acetic acid in H
2
O) and solvent B (0.1 % acetic acid in
2
the solvent was removed by a flow of N , followed by incubation under
CH CN). Characterization of the products was performed by UPLC/MS on
3
vacuum overnight. The samples were briefly placed into a bath-sonicator
and subsequently vortexed with 1 mL of the appropriate medium,
consisting of either pure water (milli-Q ultrapure water), PB (10 mM
phosphate buffer, pH 7.4), TFE/PB mixtures (TFE vol%: 40, 50, 75, 100),
a Waters Acquity instrument coupled to a Waters 3100 detector with an
analytical Waters BEH C18 column (2.1 x 50 mm, 1.7 μm, flow: 0.6
mL/min), and by HPLC using an Agilent Technologies 1200 series
instrument with a ZORBAX Eclipse XDB-C18 analytical (4.6 x 150 mm, 5
μm) column operating at 1 mL/min or a Kinetex C18 (3 x 100 mm, 2.6 µm)
column operating at 0.8 mL/min. For UPLC/MS and analytical HPLC the
a
solution of sodium dodecyl sulfate in milli-Q, a solution of n-
dodecylphosphocholine in milli-Q, a solution of 1-myristoyl-2-hydroxy-sn-
glycero-3-phosphocholine (lysoMPC) in milli-Q. The solutions with SDS,
DPC and lysoMPC gave a peptide/detergent ratio of 1/200 mol/mol.
Alternatively, the peptides were co-solubilized with DMPC in excess MeOH
to achieve solutions with peptide/lipid ratios of 1/10, 1/20, 1/100 or 1/200
eluent system was composed of solvent A (0.1 % TFA in H
B (0.1 % TFA in CH CN). Calcd. mass for C58
obs. mass (m/z): 1176.01.
2
O) and solvent
13 [m+H] : 1175.78,
+
3
102 12
H N O
2
mol/mol. The organic solvent was removed by a flow of N , followed by
Aib/CF
3
-alanine substituted HZ analogues were synthesized and purified
incubation under vacuum overnight. The dry mixtures were resuspended
in PB by a brief treatment in an ultrasonic bath and vortexing. Suspensions
underwent 5-7 consecutive freeze/thaw/ultrasonication cycles to achieve
large unilamellar vesicles with homogeneously reconstituted peptide.
the same way, except that upon incorporation of the fluorinated dipeptide
building blocks, a single coupling step (20 min at 60 °C) was performed.
Calcd. mass for (Aib /(R)CF
1
+
3
Ala) [m+H] : 1229.75, obs. mass (m/z):
1
Ala) [m+H]+:
1229.85; Calcd. mass for
C
58
H
99
F
3
N
12
O
13 (Aib /(S)CF
3
1
99 3 12 13
229.75, obs. mass (m/z): 1229.97; Calcd. mass for C58H F N O
The peptide/detergent/lipid mixtures were transferred into Hellma Suprasil
quartz cuvettes with 1 mm path length. CD measurements were performed
on a JASCO J-815 spectropolarimeter. Measurements were performed at
5
+
(
Aib /(R)CF
3
Ala) [m+H] : 1229.75, obs. mass (m/z): 1230.36; Calcd. mass
5
for C58
99
H F
3
N
12
O
13 (Aib /(S)CF
3
Ala) [m+H]+: 1229.75, obs. mass (m/z):
9
Ala) [m+H]+:
1230.30; Calcd. mass for
C
58
H
99
F
3
N
12
O
13 (Aib /(R)CF
3
25 °C (solutions), or at variable temperatures (lipid suspensions). Spectra
1
99 3 12 13
229.75, obs. mass (m/z): 1229.90; Calcd. mass for C58H F N O
were acquired between 185 and 260 nm at 0.1 nm intervals (10 nm/min
rate; 8 s response time; 1 nm bandwidth). Three consecutive scans were
averaged for each sample and for the peptide-free samples (baseline).
Baseline spectra were subtracted, the intensities at 260 nm were assigned
to zero, and no smoothening was applied. All spectral processing was
done using the Jasco Spectra Analysis software.
9
+
(
3
Aib /(S)CF Ala) [m+H] : 1229.75, obs. mass (m/z): 1229.96.
Lxx/CF
same way. Fmoc-CF
syntheses were performed on a 0.15 mmol scale (resin loading: 0.5
mmol/g, 300 mg) and final cleavages succeeded upon a double treatment
with TFA/DCM/water/TIS (47:47:4:2) for 1 h. Crude peptides were
obtained from air flow-dried combined cleavage filtrates, which were
3
-Bpg substituted HZ analogues were synthesized and purified the
3
-(L)-Bpg-OH was treated as Fmoc-Aib-OH. The
Solid-State NMR sample preparation. For solid-state NMR, labeled
harzianin HK-VI was reconstituted into fully hydrated, mechanically
oriented lipid bilayers on glass supports. First, ~0.25 mg of 19F-labeled
harzianin HK-Vi analogues or ~0.5 mg of 15N-labeled harzianin HK-VI
99 3 12 13
lyophilized after addition of 5 mL of water. Calcd. mass for C60H F N O
3
+
(Ile /CF
3
-Bpg) [m+H] : 1253.75, obs. mass (m/z): 1253.78; Calcd. mass
4
+
for C60
99 3 12
H F N O
13 (Ile /CF
3
-Bpg) [m+H] : 1253.65, obs. mass (m/z):
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
analogues were co-dissolved in methanol or methanol/chloroform together
with ~15 mg or 30 mg of lipid, respectively, to achieve a peptide/lipid ratio
of 1/100 mol/mol. The solution was spotted onto about 20 glass slides (7.5
mm x 18 mm x 0.06 mm, Marienfeld, Germany), and the resulting films
were dried under vacuum. The glass slides were then stacked and
hydrated by incubation in 97% relative humidity at 48 °C overnight, which
axis of the 15N CSA tensor was assumed to be rotated 19° away from the
N-H bond.
To identify the best-fit orientation and order parameter, the rmsd was
determined for all combinations of  and within the range of [0°,180°] in
steps of 1°, and for all Smol within the range of [0,1] in steps of 0.1. (Note,
that the values of and  were transformed to equivalent values of 180°-
3
leads to the spontaneous formation of ~10 uniformly aligned and hydrated
bilayers. The sample was wrapped in parafilm and a cling foil to prevent
drying. The orientation of the sample was confirmed using solid-state 31P-
NMR. In all spectra shown here, the membrane normal was aligned
parallel to the static magnetic field direction.

and 180° + ) This analysis was performed for all eight structure models,
to find the best-fit secondary structure with the lowest overall rmds (SI
Tables S6-S8).
Using the -bend ribbon spiral conformation, we calculated the best-fit
values of ,  and Smol for HZ for each of the five different patterns of labels
Solid-state NMR spectroscopy. Solid-state 19F-NMR spectra were
measured on a Bruker Avance III spectrometer with an 11.7 T widebore
magnet, operating at a 19F-resonance frequency of 470 MHz. The sample
was placed into an in-house built F/ H double-resonance probe,
equipped with a flat coil (rectangular cross section, 3 x 9 mm). NMR-
spectra were acquired using an anti-ringing sequence to suppress any
probe background signal, using a 90°-pulse length of 2.5 µs and 20 kHz
1H decoupling.[32] Typically 2000 scans of 4 ms separated by a recycle
delay of 2 s were collected.
(see SI Tables S9-S12). Each group gives a somewhat different solution,
leading to an intrinsic scatter of values that contributes to the error margins.
As the final values reported in Table 2, we took the average over the results
of the individual analyses of the patterns of labels (SI Tables S9-S12).
19
1
Acknowledgements
Solid-state 15N-NMR spectra of the oriented samples were measured on a
Bruker Avance spectrometer with a 14.1 T widebore magnet, operating at
a 15N-resonance frequency of 60 MHz. The sample was placed into an in-
house built 15N/ H double-resonance lowE probe, equipped with a flat coil
The authors acknowledge Markus Schmitt (KIT, Karlsruhe) and
Peter Gor’kov (NHMFL, Tallahassee) for building the probes used
in this study and support of the NMR facility. We thankfully
acknowledge financial support from the DFG-ANR grant
1
(
rectangular cross section, 3 x 9 mm). NMR-spectra were acquired
“
Peptaibols” (DFG: UL 127/6-1; ANR-2011-INTB-1002-01), the
following cross-polarization from 1H (using radio-frequency amplitudes of
1
DFG grant INSTR 121384/58-1 FFUG, the Helmholtz Association
50 kHz) and under 50 kHz H decoupling. Typically, ~20000 scans of 10
ms separated by a recycle delay of 3 s were collected.
program “BIF-TM” and in part the DFG-GRK (Nr2039).
NMR data analysis: The orientation of the backbone structures with
respect to the membrane normal was characterized by two angles  and
Conflict of interest
(SI Figure S2). The angle  is the angle between the membrane normal
n and the long axis of the molecule z, which was defined as the principal
axis of the moment of inertia tensor corresponding to the smallest principal
value. The angle measures the rotation of the molecule around its long
The authors declare no conflict of interest.
6
axis, and we define =0° such that C of Pro lies along (n x z) x z, where
Keywords: amino acids • peptides • membranes • isotopic
labeling • NMR spectroscopy • fluorine
n is the membrane normal and z is the axis of the moment of inertia
corrsesponding to the long axis of the molecule. The splitting of the dipolar
triplet in the 19F-NMR-spectra, ΔFF, or the signal position (chemical shift) in
[
[
[
[
[
[
[
[
[
1]
2]
3]
4]
5]
6]
7]
8]
9]
I. Segalas, Y. Prigent, D. Davoust, B. Bodo, S. Rebuffat, Biopolymers
999, 50, 71-85.
the 15N-NMR spectra, δ
N
, were analyzed as experimental data and
1
compared with predicted (calculated) data to determine the best-fit
backbone structure and membrane-alignment of harzianin HK-VI in lipid
bilayers. Briefly, for a particular peptide orientation given by  and , the
R. Mikkola, M. A. Andersson, L. Kredics, P. A. Grigoriev, N. Sundell, M.
S. Salkinoja-Salonen, FEBS J. 2012, 279, 4172-4190.
L. Beven, D. Duval, S. Rebuffat, F. G. Riddell, B. Bodo, H. Wroblewski,
Biochim. Biophys. Acta Biomembr. 1998, 1372, 78-90.
19
19
15
NMR-parameter ( F- F dipolar coupling, or N chemical shift) of the
considered NMR label is predicted by converting the respective NMR-
interaction tensor (19F-19F dipolar interaction tensor, or 15N chemical shift
anisotropy (CSA) tensor) from its principal axes frame representation to its
laboratory frame representation. Averaging by molecular motions that are
fast on the NMR-timescale was taken into account by scaling the
anisotropic contribution to the resonance frequency with an order
parameter Smol, as indicated in Eq. 1 and Eq. 2. The deviation of the
calculated NMR-parameters from the respective experimental values was
I. Augeven-Bour, S. Rebuffat, C. Auvin, C. Goulard, Y. Prigent, B. Bodo,
J. Chem. Soc., Perkin Trans. 1 1997, 1587-1594.
M. Lucaciu, S. Rebuffat, C. Goulard, H. Duclohier, G. Molle, B. Bodo,
Biochim. Biophys. Acta Biomembr. 1997, 1323, 85-96.
S. Rebuffat, H. Duclohier, C. Auvinguette, G. Molle, G. Spach, B. Bodo,
FEMS Microbiol. Immunol. 1992, 105, 151-160.
K. Koch, S. Afonin, M. Ieronimo, M. Berditsch, A. S. Ulrich, Top. Curr.
Chem. 2012, 306, 89-118.
quantified in terms of the root of mean square deviation (rmsd): rmsd =
D. Maisch, P. Wadhwani, S. Afonin, C. Böttcher, B. Koksch, A. S. Ulrich,
J. Am. Chem. Soc. 2009, 131, 15596-15597.
2
(1/N Σ
k
(dcal,k – dexp,k) )1/2, with dcal,k and dexp,k denoting the calculated and
N
experimental NMR-parameters (ΔFF or δ ), respectively, and N is the
number of NMR-constraints. The principal axis values of the 19F-19F dipolar
tensor and of the 15N chemical shift anisotropy tensor were (-8 kHz, -8 kHz,
P. Wadhwani, E. Strandberg, J. van den Berg, C. Mink, J. Bürck, R. A.
M. Ciriello, A. S. Ulrich, Biochim. Biophys. Acta Biomembr. 2014, 1838,
940-949.
1
6 kHz) and (50 ppm, 74 ppm, 209 ppm), resulting into the maximum 19F
dipolar splitting Δ
= 16 kHz, the 15N isotropic chemical shift of δiso = 111
ppm, the 15N chemical shift anisotropy δCSA = 98 ppm, and the 15
asymmetry parameter of η = -0.245 (See Eq 1 and Eq. 2). The principal z-
[
[
10] O. M. Michurin, S. Afonin, M. Berditsch, C. G. Daniliuc, A. S. Ulrich, I. V.
0
Komarov, D. S. Radchenko, Angew. Chem. Int. Ed. 2016, 55, 14595-
N
14599.
11] F. Huguenot, T. Brigaud, J. Org. Chem. 2006, 71, 7075–7078.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
[
12] P. K. Mikhailiuk, S. Afonin, A. N. Chernega, E. B. Rusanov, M. O.
Platonov, G. G. Dubinina, M. Berditsch, A. S. Ulrich, I. V. Komarov,
Angew. Chem. Int. Ed. 2006, 45, 5659-5661.
[23] S. L. Grage, S. Afonin, A. S. Ulrich, Methods Mol. Biol. 2010, 618, 183-
207.
[24] A. Grau-Campistany, E. Strandberg, P. Wadhwani, J. Reichert, J. Bürck,
F. Rabanal, A. S. Ulrich, Sci. Rep. 2015, 5, 9388.
[
[
[
[
13] S. Afonin, U. H. N.; Dürr, P. Wadhwani, P., J. Salgado, A. S. Ulrich, Top.
Curr. Chem. 2008, 273, 139-154.
[25] E. Strandberg, D. Tiltak, S. Ehni, P. Wadhwani, A. S. Ulrich, Biochim.
Biophys. Acta Biomembr. 2012, 1818, 1764-1776.
14] G. Chaume, N. Lensen, C. Caupene, T. Brigaud, Eur. J. Org. Chem.
2
009, 33, 5717–5724.
15] E. Devillers, J. Pytkowicz, E. Chelain, T. Brigaud, Amino Acids 2016, 48,
457-1468.
[26] E. Strandberg, J. Zerweck, P. Wadhwani, A. S. Ulrich, Biophys. J. 2013,
104, L9-L11.
1
[27] D. Windisch, C. Ziegler, S. L. Grage, J. Bürck, M. Zeitler, P. L. Gor'kov,
A. S. Ulrich, Biophys. J. 2015, 109, 737-749.
16] S. Rebuffat, S. Hlimi, Y. Prigent, C. Goulard, B. Bodo, J. Chem. Soc.,
Perkin Trans. 1996, 1, 2021-2027.
[28] C. Muhle-Goll, S. Hoffmann, S. Afonin, S. L. Grage, A. A. Polyansky, D.
Windisch, M. Zeitler, J. Bürck, A. S. Ulrich, J. Biol. Chem. 2012, 287,
26178-26186.
[
17] L. Whitmore, B. A. Wallace, Biopolymers 2008, 89, 392-400.
18] C. Toniolo, F. Formaggio, R. W. Woody in Comprehensive Chiroptical
Spectroscopy (Eds.: N. Berova, P. L. Polavarapu, K. Nakanishi, R. W.
Woody), John Wiley & Sons, Hoboken, 2012, pp. 499–544.
19] G. Yoder, T. A. Keiderling, F. Formaggio, M. Crisma, C. Toniolo,
Biopolymers 1995, 35, 103–111.
[
[29] R. W. Glaser, C. Sachse, U. H.N. Dürr, P. Wadhwani, A. S. Ulrich J.
Magn. Reson. 2004, 168, 153–163
[
[
[
[
[30] D. S. Radchenko, S. Kattge, S. Kara, A. S. Ulrich, S. Afonin, Biochim.
Biophys. Acta Biomembr. 2016, 1858, 2019-2027.
20] pdb codes: 5MF8 (structure in TFE); 5MF3 (structure in SDS); 5M9Y
[31] Y. Shai, Biopolymers 2002, 66, 236-248.
(structure in DPC); (to be published elsewhere).
[32] S. Zhang, X. Wu, M. Mehring, Chem. Phys. Lett. 1990, 173, 481-484.
21] M. Putzu, S. Kara, S. Afonin, S. L. Grage, A. Bordessa, G. Chaume, T.
Brigaud, A. S. Ulrich, T. Kubar, Biophys. J. 2017, 112, 2602-2614.
22] R. Gessmann, H. Brückner, K. Petratos, J. Pept. Sci. 2016, 22, 76-81.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704307
FULL PAPER
Entry for the Table of Contents (Please choose one layout)
FULL PAPER
A set of four orthogonal 19F- and 15
N
Stephan L. Grage, Sezgin Kara, Andrea
Bordessa, Véronique Doan, Fabio
Rizzolo, Marina Putzu, Tomáš Kubař,
Anna Maria Papin , Grégory Chaume,
Thierry Brigaud*, Sergii Afonin*, Anne S.
Ulrich*
labels was developed for the structure
determination of membrane bound
peptides using solid-state NMR. Using
this novel approach, which maximizes
the information content employing only
few labels, the secondary structure and
]
orientation of
a
membrane-active
Page No. – Page No.
peptaibol bound to lipid bilayers could
be elucidated despite its short length
and high mobility.
Orthogonal 19F-labeling for solid-state
NMR reveals the conformation and
orientation of short peptaibols in
membranes
This article is protected by copyright. All rights reserved.