The reversible macrocyclization of Tyrocidine A aldehyde: A hemiaminal reminiscent of the tetrahedral intermediate of macrolactamization

Article abstract of DOI:10.1039/b917549k

In spite of the important role of peptide macrocyclizations for the generation of conformationally constrained biological ligands, our knowledge of factors that determine the inclination of a substrate to cyclize is low. Therefore, methods that give access to the thermodynamic characterization of these processes are desirable. In this work, we present the isosteric substitution of the amide ligation site of a cyclopeptide by an imine. Applied to the decapeptide antibiotic Tyrocidine A (TycA), the reversible cyclization of the linear aldehyde TycA-CHO resulted in the unexpectedly stable hemiaminal Ψ[CH(OH)NH]-TycA, which is equivalent to the tetrahedral intermediate of macrolactamization, and which is observed for the first time in a peptidic structure. On the basis of NMR spectroscopy and molecular modeling, we discuss the observed high stereoselectivity of hemiaminal formation, as well as its reluctance to be dehydrated to the imine. As required for thermodynamic analysis, it is possible to establish a pH- and temperature-dependent cyclization equilibrium, which allows determination of the entropy loss of the peptide chain, and quantification of the extent of preorientation of the cyclization precursor.

Full text of DOI:10.1039/b917549k

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
The reversible macrocyclization of Tyrocidine A aldehyde: a hemiaminal
reminiscent of the tetrahedral intermediate of macrolactamization†
Sebastian Enck, Florian Kopp,‡ Mohamed A. Marahiel* and Armin Geyer*
Received 25th August 2009, Accepted 30th October 2009
First published as an Advance Article on the web 27th November 2009
DOI: 10.1039/b917549k
In spite of the important role of peptide macrocyclizations for the generation of conformationally
constrained biological ligands, our knowledge of factors that determine the inclination of a substrate to
cyclize is low. Therefore, methods that give access to the thermodynamic characterization of these
processes are desirable. In this work, we present the isosteric substitution of the amide ligation site of a
cyclopeptide by an imine. Applied to the decapeptide antibiotic Tyrocidine A (TycA), the reversible
cyclization of the linear aldehyde TycA–CHO resulted in the unexpectedly stable hemiaminal
W[CH(OH)NH]–TycA, which is equivalent to the tetrahedral intermediate of macrolactamization, and
which is observed for the first time in a peptidic structure. On the basis of NMR spectroscopy and
molecular modeling, we discuss the observed high stereoselectivity of hemiaminal formation, as well as
its reluctance to be dehydrated to the imine. As required for thermodynamic analysis, it is possible to
establish a pH- and temperature-dependent cyclization equilibrium, which allows determination of the
entropy loss of the peptide chain, and quantification of the extent of preorientation of the cyclization
precursor.
behavior and allows the estimation of the entropy balance of
Introduction
macrocyclic ring closure, which can be quantitatively explained
Macrocyclization induces conformational restriction, which can
by the amount of released water and the number of restricted
increase biological activity and selectivity of protein ligands.1 This
is demonstrated by the multitude of macrocyclic natural products
synthesized by complex enzyme machineries,2 and by the great
effort of medicinal chemists to establish suitable methods for the
cyclization of long molecular chains.3,4 However, concerning the
propensity of a substrate to ring closure, one must rely on a mainly
qualitative understanding of the biological processes and on the
empirical approach to synthetic macrocyclizations. A prerequisite
for a better understanding is thermodynamic analysis, which could
give the free energy difference between linear and cyclic species,
characterizing the inclination of a linear precursor for cyclization
independently of the activation method. Thermodynamic data
can then be used to validate computational predictions of the
cyclization processes and can help to systematically improve
synthetic macrocyclizations, which often suffer from low yields
and thus become the bottleneck of total syntheses. Thermo-
dynamic studies, however, are hampered by the irreversibility
of the macrocyclization processes. The nostocyclopeptides A1
and A2 produced by the terrestrial cyanobacterium Nostoc sp.
ATCC53789 stand out from all other known cyclopeptides due to
their unique backbone imino linkage.5 Recently, we demonstrated
that this structural feature gives the required reversible cyclization
rotational degrees of freedom in the peptide backbone.6 Consid-
ering the great number of macrocyclic compounds of natural and
synthetic origin, it is desirable to expand these possibilities given
by reversible cyclization processes.
In this study, we present a strategy to access the thermodynamics
of peptide macrocyclizations. The amide bond at the ligation site
of a cyclopeptide is isosterically substituted by a reversibly closing
imino functionality (Fig. 1). By variation of temperature and
pH, we control the ring-chain equilibrium of the 30-membered
macrocycle of an accordingly modified Tyrocidine A (TycA),
which is significantly larger than the 21-membered rings formed by
the nostocyclopeptides. As the entropy balance of the cyclization
process can be estimated, this strategy allows us to quantify the
extent of preorientation and to identify determinants that play a
role in the inclination of a substrate to cyclize.
Results and discussion
Cyclization studies
The well-studied decapeptide antibiotic TycA appeared to be
a promising candidate for the substitution survey because of
its high degree of preorganization, and its ability to cyclize
spontaneously and without the need of an enzyme or an activation
reagent in aqueous solution.7 In order to introduce reversibility
into the irreversible process of native TycA cyclization, we
synthesized an analog of the linear TycA cyclization precursor
with an aldehyde instead of a carboxylic acid function at the
C-terminus (TycA–CHO). While in TycA biosynthesis the macro-
cyclic amide is closed between D-Phe1 and Leu10, cyclization
of the aldehyde precursor should result in reversible formation
Fachbereich Chemie, Philipps-Universita¨t Marburg, Hans-Meerwein-
Strasse, D-35032 Marburg, Germany. E-mail: geyer@staff.uni-marburg.de
† Electronic supplementary information (ESI) available: Peptide synthesis
and purification, characterization of all compounds by HR-ESI-MS,
detailed description of NMR measurements, spectral data, details of
pH/temperature dependence of the macrocyclization equilibrium, as well
as molecular modeling data.
‡ Present address: The Scripps Research Institute, Cravatt Laboratory,
10550 North Torrey Pines Road, La Jolla, CA 92037, USA.
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 559–563 | 559
©

View Article Online
Fig. 1 Amides condense irreversibly in aqueous solution (a) while all steps of macrocyclic imine generation are reversible (b) and thus allow for
thermodynamic analysis. Both pathways have in common the same sequence of intermediates, as the substrate first has to be activated to react with an
amine to form a tetrahedral species, which subsequently eliminates to yield the amide or the imine. While the amide is irreversibly generated from a
tetrahedral intermediate, the imine (red) is in equilibrium with a hemiaminal (green). (c) One main species out of four possible diastereomers is formed
in the reversible cyclization of TycA–CHO. All assigned 1H chemical shifts are written next to their respective positions. The temperature gradients of
the NH proton signals (in ppb K-1) are assigned with bold numbers, and the 1H–1H-TOCSY correlations in the hemiaminal region are marked by blue
arrows. (d) ROESY region (600 MHz, 300 K, KH2PO4/H3PO4 buffer, H2O/D2O 5 : 1 with 0.19 mol L-1 SDS-d25, pH 9.0) of cyclized TycA–CHO showing
correlations in the hemiaminal region (green bar), the trans-Pro2 configuration and the b sheet.
of the TycA imino analog cyclo(Asn5–Gln6–Tyr7–Val8–Orn9–
sulfate (SDS-d25) micelles and aqueous phosphate buffer prevented
precipitation when the pH was increased from 3.0 to 9.0 (see ESI†
for details). As in the case of the nostocyclopeptides, at pH 3.0
only linear species are observed and the C-terminal aldehyde is
mostly hydrated (amount of aldehyde <10%).6 The appearance
of several low-intensity sets of signals may also result from the
epimerization of the aldehyde, as it was observed in the case of the
nostocyclopeptides, as well as from the Pro2 cis/trans isomerism.
8
=
W[C N]Leu10–D–Phe1–Pro2–Phe3–D–Phe4).
An aqueous solution of the precursor TycA–CHO was subjected
to various temperature and pH conditions, enabling us to study the
cyclization equilibrium. Despite the potential presence of several
linear aldehydes and aldehyde hydrates together with diastereo-
meric macrocyclic imines, as well as dimers and oligomers, the
quantities of the dominant contributing species can be clearly
1
1
identified from the H NMR. Unexpectedly, at pH 9.0 no cyclic
A H NMR spectrum recorded 20 min after raising the pH
imine is observed, but the hemiaminal (W[CH(OH)NH]–TycA)
dominates the cyclization equilibrium (Fig. 1c). The cyclization
product of TycA–CHO is the first example of a macrocyclic peptide
with a hemiaminal, which was expected to be unstable in aqueous
solution.
The ten amino acid sequence (4n + 2 residues) of TycA has a
high antiparallel amphipathic b sheet content9–11 and is inclined
to self-aggregate in aqueous solution, a fact which hampers its
handling. The combination of perdeuterated sodium dodecyl
to 9.0 indicated the formation of one main product and only
traces (<5%) of the linear species. This fast response of TycA–
CHO to pH increase, as well as the high signal dispersion of
the new set of signals, showed that cyclization had occurred. The
temperature gradients of the NH protons are characteristic for the
Tyc macrocycles, and the strong long-range D-Phe4 a-H/Orn9
a-H NOE indicates the formation of the b sheet. A covalent
linkage of D-Phe1 and Leu10 is proven by NOESY and TOCSY
=
spectra, but no imine signals of N CH, which are expected at
560 | Org. Biomol. Chem., 2010, 8, 559–563
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
chemical shifts of approx. 7 ppm (1H) and 165 ppm (13C), were
detectable in the H and 2D NMR spectra. In fact, the spectral
information identifies a hemiaminal linkage (W[CH(OH)NH]–
TycA), with the hemiaminal-CH signal at 4.43 ppm and the NH
signal at 5.33 ppm, respectively (Fig. 1d).
TycA. However, none of the two hemiaminal epimers could be
unambiguously identified as the more stable one, because similar
geometries are obtained in both cases, and the hemiaminal is
capable of hydrogen bonding with Phe3-NH (Fig. 2).
1
As the a position of the Leu10 aldehyde is subjected to racem-
ization and a new stereogenic center is generated by formation
of W[CH(OH)NH]–TycA, four possible diastereomers can emerge
from the cyclization reaction. The presence of only one cyclic main
species, however, demonstrates that cyclization proceeds along a
single reaction path, and that the hemiaminal is incorporated into
a structurally optimized cyclic structure, which can be dominated
by a distinct main stereoconfiguration. This remarkable behavior
is reminiscent of the nostocyclopeptides, which also deracemize
upon cyclization, yielding stereopure macrocyclic E-configured
imines from a mixture of S- and R-configured aldehydes.6
Stability of the hemiaminal
Fig. 2 NOE-based energy-minimized average structure of W[R-CH(OH)-
NH]–TycA in SDS-d25 environment in H2O at pH 9.0. The structure is
similar to the antiparallel b sheet structure of native TycA. Carbons are
shown in green, oxygens in red, and nitrogens in blue. The three amide
hydrogen bonds are marked in yellow, and the hemiaminal hydrogen bond
is marked in orange.
An important question concerns the origin of the unprecedented
stability of the hemiaminal W[CH(OH)NH]–TycA. The main
difference to the corresponding imine W[C N]–TycA is the sp
3
=
hybridization and the presence of a hydroxyl function, which
is stabilized as part of the hydrogen bonding network of the
antiparallel TycA b sheet, thereby suppressing imine formation.
If the hemiaminal oxygen acts as an electron donor for Phe3-NH
(temperature gradient: -4.3 ppb K-1), a hydrogen bond analogous
to the Phe3-NH/Leu10-CO interaction in native TycA is present,
though the hemiaminal oxygen is expected to be weaker in its
electron-donating ability than the amide oxygen. Of course, the
conditions applied in this case are different, as TycA is investigated
in a membrane mimetic environment in water, while DMSO and
MeOH have been used in earlier works.12 However, the similar
pattern of temperature gradients Dd/DT of the NH protons
(Fig. 1c), which give information on the antiparallel b sheet,
indicates that, in spite of the modifications of TycA and of
its environment, the b sheet with the four hydrogen bonds is
maintained.13
The sp3 hybridized carbon of the hemiaminal resembles the
geometrically similar tetrahedral intermediate of amide formation
(Fig. 1). Hence, W[CH(OH)NH]–TycA, for the first time, allows
the observation of the carbonyl analog of this species formed dur-
ing a macrocyclic amide formation. In previous studies concerning
the cyclization process of native TycA, it was postulated that the
tetrahedral intermediate may be involved in a stabilizing hydrogen
bridge to Phe3-NH, which in turn is a result of the pronounced
precursor preorganization.14 The unexpected hemiaminal stability
fits into this picture as it shows that the preorganization of the
linear chain brings together both termini and thus facilitates ring
closure.
The presence of a single stereoconfiguration results from the
preference for one of the two half-spaces of the aldehyde by
the amine nucleophile. This apparently arises from the steric
shielding by the surrounding side chains, in combination with
a precoordination of the aldehyde to the Phe3-NH (Fig. 3a). As
visible from the structure shown in Fig. 2, this should result in
an orientation of the hemiaminal-NH to the solvent, which is
indicated by the large temperature coefficient of -15.0 ppb K-1.
30 min after cyclization, a low-intensity second set of signals
(<10%) is visible, which grows during the following hours, finally
reaching approx. 25% after 38 h (Fig. S6 and S7 in the ESI†). As the
expected dehydration tothe imine does not occur, twoexplanations
are conceivable. Either this secondary set of signals results from the
epimeric hemiaminal, slowly formed via an imine intermediate by
dehydration and subsequent addition of water, which occurs with
less stereocontrol than the addition of the amine to the aldehyde,
or alternatively, a slow epimerization of Leu10 may also be the
reason for the emerging second set of signals.6 The formation of
dimers in principle is also possible, but was never observed in
cyclization studies of native TycA.7
Thermodynamics of TycA macrocyclization
In order to elucidate the entropy balance of the cyclization process,
we investigated the temperature dependence of the cyclization
equilibrium. In spite of high signal density (two main sets of
decapeptide signals) and the line broadening caused by the SDS-
d25, the Val8 and Leu10 methyl signals are well enough resolved
to be integrated, and thus to give information about the compo-
sition of the equilibrating system (Fig. 3b). The thermodynamic
parameters can be determined from the temperature dependence
of the percentages of linear species (TycA–CH(OH)2) and cyclic
species (W[CH(OH)NH]–TycA). As the composition of the lin-
ear/cyclic peptide mixture at pH 6.6 does not change significantly
between 290 and 330 K (variations of <5% are within measuring
The remarkable selective formation of only one out of four
possible diastereomers upon cyclization raises the question of the
absolute stereochemistry of the cyclization site. The NOE data of
W[CH(OH)NH]–TycA are consistent with the NH temperature
gradients and support the b sheet formation under the NMR con-
ditions. As they are similar compared to native TycA, they suggest
the S-configuration of Leu10, like in the linear precursor. The
NOEs were used as interproton restraints for molecular dynamics
simulations of W[R-CH(OH)NH]–TycA and W[S-CH(OH)NH]–
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 559–563 | 561
©

View Article Online
Fig. 3 (a) Schematic representation of backbone, Pro residue and hydrogen bonds before (Tyc–CHO) and after (W[CH(OH)NH]–TycA) cyclization.
As also assumed for native TycA, residues 3–9 in the linear peptide are preorganized for cyclization by the presence of three hydrogen bonds (black
dotted lines: Phe3-CO/Leu10-NH, Val8-CO/Asn5-NH, Asn5-CO/Val8-NH; from left to right). This enables a precoordination of the aldehyde (orange)
by another hydrogen bond (grey dotted line) and the stereoselective attack by the amine of the predisposed N-terminal segment. The hydrogen bond
between the aminal oxygen and the Phe3-NH is analogous to the hydrogen bond formed by the tetrahedral intermediate in native TycA cyclization and to
the Leu10-CO/Phe3-NH interaction resulting after amide formation. (b) Aliphatic proton range of 1H NMR spectra recorded at different temperatures
and pH values using water suppression by the Gradient-Tailored Excitation (Watergate) method (600 MHz, KH2PO4/H3PO4 buffer, H2O : D2O 5 : 1
with 0.19 mol L-1 SDS-d25). The spectra at pH 6.6, depicted in the middle, show the presence of an approx. 50 : 50 (linear : cyclic) mixture, which is
temperature-independent of composition. For comparison, the spectrum of TycA–CH(OH)2 at pH 3.0 and 300 K is shown at the bottom, while at the
top, the respective section of the freshly cyclized macrocyclic hemiaminal at pH 9.0 and 300 K, showing only one main set of signals, is depicted.
uncertainty), the balance of cyclization enthalpy DH can be set
to zero according to the van’t Hoff equation. The amounts of
linear and cyclic species are approx. 50% each, which indicates
that the free energy difference DG can, in an approximation, also
be set to zero. Consequently, the Gibbs–Helmholtz equation gives
the entropy balance DS of the macrocyclization reaction, which
is equal to zero. This value represents the total entropy change,
which also includes the released water molecule.
demonstrates that the linear precursors of the Tyrocidine peptides
feature such an extraordinary pronounced preorientation that the
cyclization entropy balance is shifted from a negative to an approx.
neutral value. The 10 cal K-1 mol-1 reduction observed for TycA–
CHO cyclization is equivalent to approx. three rotational degrees
of freedom, and the cyclizing substrate acts more like a hepta-
than a decapeptide. These results are in agreement with former
qualitative investigations on native TycA, in which its outstanding
ability to self-cyclize without reagent was demonstrated,7 and
now add a quantitative estimate of the TycA preorientation and
cyclization tendency.
In order to estimate the entropy loss of the peptide chain, several
contributions must be taken into account, as already discussed for
the case of nostocyclopeptide macrocyclization.6 The changes in
the solvation sphere can, in a first approximation, be neglected
in comparison to the water released upon condensation, which is
supported by mostly slight changes of the chemical shifts (Table S2
in the ESI†). While two equivalents of water are released upon
imine formation,6 hemiaminal formation is accompanied by the
loss of only one equivalent of water. Subtracting the entropy of one
released water molecule (16.7 cal K-1 mol-1)15 from the obtained
entropy balance of approx. 0 cal K-1 mol-1 gives an entropy loss
of the TycA peptide chain of approx. 17 cal K-1 mol-1. The rule
of thumb that half of the rotational degrees of freedom are lost
upon cyclization proved to be a well applicable approximation.6
Concerning the cyclization of the decapeptide TycA–CHO, pos-
sessing 18 rotational degrees of freedom in the backbone, we
should expect an entropy loss of 27 (9 ¥ 3) cal K-1 mol-1, which
exceeds the experimentally obtained value by 10 cal K-1 mol-1.
Hence, the overall entropy balance of a decapeptide cyclization
with the release of one equivalent of water should be in the range
of approx. -10 cal K-1 mol-1, and thus entropically disfavored. This
Conclusions
In summary, we use the isosteric substitution of the peptide
bond against a reversibly closing carbonyl functionality for the
structural and thermodynamic analysis of a peptide cyclization
in order to quantify the extent of preorientation of the linear
cyclization precursor TycA–CHO. This readily cyclizes in aqueous
solution, and surprisingly yields a hemiaminal that is not further
dehydrated to the expected imine. The unprecedented stability
of the aminal results from its apparent incorporation into the
hydrogen bonding network. Furthermore, the formation of a
dominating stereoisomer suggests a stereoselective attack, which
is only possible if the cyclization site is structurally well-defined.
The isosteric substitution of an amide by a carbonyl analog
has promising potential for the elucidation of macrocyclization
thermodynamics. A systematic “carbonyl scan” of all amide bonds
in a cyclopeptide could allow the ranking of different precursors
562 | Org. Biomol. Chem., 2010, 8, 559–563
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
with respect to their inclination to cyclize, and to compare the
native with non-native ligation sites. It also has potential for the
determination of the native cyclization site if it is not known. As
an analog of the tetrahedral intermediate in amide formation,
the macrocyclic hemiaminal is of particular interest. Notably,
a comparison of the TycA hemiaminal and the native amide
with respect to biological activity will help the understanding of
the TycA structure–activity relationship. We are confident that
the examination of further lactams should reveal more details,
which may help us to better understand the determinants of
macrocyclization processes.
NOE-based molecular modeling
The structure calculations of W[R-CH(OH)NH]–TycA were car-
ried out as described previously6,18 using the program package
HyperChem19 with MM+ force field and without explicit water
included. The NOE-derived distances that were used as restraints
are listed in Table S4 in the ESI.† Ten snapshots from a 10 ¥
10 ps molecular dynamics simulation (step size 1 fs, 300 K)
were averaged and the resulting structure was allowed to relax
without restraints to verify the plausibility. The resulting structure
is depicted in Fig. 2, and the acquired distances are listed in
Table S4 in the ESI.†
Experimental section
Acknowledgements
Peptide synthesis
We are grateful to Dr U. Linne for his assistance with
HPLC and ESI-MS measurements. Financial support was pro-
vided by the Deutsche Forschungsgemeinschaft (DFG) and by
the Studienstiftung des Deutschen Volkes (Ph.D. scholarship
for S.E.).
The synthesis of Tyrocidine A aldehyde was carried out ac-
cording to a protocol described previously,16 starting from H–
R
Thr–Gly–NovaSynꢀ TG resin (Novabiochem) preloaded with
leucine amino aldehyde. The chain extension was carried out
by automated Fmoc solid phase synthesis. After assembly of
the target sequence and removal of protecting groups, the crude
peptide aldehyde was cleaved from resin, precipitated in hexane,
and purified by preparative HPLC. The identity of the product
obtained after lyophilization was confirmed by analytical HPLC-
ESI-MS. A sample of cyclized TycA in basic aqueous solution
was also controlled via ESI-MS. Further experimental details, the
HPLC trace of the purified peptide, as well as high-resolution ESI-
MS spectra of all linear and cyclic compounds, are provided in the
ESI.†
References
1 E. M. Nolan and C. T. Walsh, ChemBioChem, 2009, 10, 34–53.
2 F. Kopp and M. A. Marahiel, Nat. Prod. Rep., 2007, 24, 735–
749.
3 E. M. Driggers, S. P. Hale, J. Lee and N. K. Terrett, Nat. Rev. Drug
Discovery, 2008, 7, 608–624.
4 J. Blankenstein and J. Zhu, Eur. J. Org. Chem., 2005, 1949–1964.
5 T. Golakoti, W. Y. Yoshida, S. Chaganty and R. E. Moore, J. Nat.
Prod., 2001, 64, 54–59.
6 S. Enck, F. Kopp, M. A. Marahiel and A. Geyer, ChemBioChem, 2008,
9, 2597–2601.
NMR analytics
7 X. Bu, X. Wu, G. Xie and Z. Guo, Org. Lett., 2002, 4, 2893–2895.
=
8 This modified TycA is abbreviated to W[C N]–TycA. The fact that
Homo- and heteronuclear NMR measurements were carried
out using a Bruker Avance DRX600 spectrometer operating at
600.13 MHz (1H) and 150.90 MHz (13C), respectively, with a 5 mm
BBI probe head. All chemical shifts are given in parts per million
(ppm). For water suppression, the double pulsed field-gradient
spin echo (WATERGATE) sequence17 was applied, which had
also been appropriate in preceding studies.6 All measurements
described were carried out with the same NMR sample to prove
the reversibility of the system. TycA–CHO was dissolved in a
partially deuterated KH2PO4/H3PO4 buffer solution (H2O/D2O
5 : 1) at pH 3.0. The pH was set to the desired value by addition
of solid Na2CO3 and aqueous H3PO4, respectively. For the
temperature-dependent series of measurements, 10 K steps (290–
330 K) were chosen and the sample was allowed to equilibrate in
the spectrometer for 1 h. In the ESI,† further details, including
assigned NMR spectra, as well as chemical shifts and temperature
gradients, are given.
no position for the modified tether is given indicates that the native
cyclization position is retained.
9 M.-C. Kuo and W. A. Gibbons, J. Biol. Chem., 1979, 254, 6278–
6287.
10 M.-C. Kuo and W. A. Gibbons, Biophys. J., 1980, 32, 807–836.
11 R. M. Kohli, J. W. Trauger, D. Schwarzer, M. A. Marahiel and C. T.
Walsh, Biochemistry, 2001, 40, 7099–7108.
12 A. P. Tonge, P. Murray-Rust, W. A. Gibbons and L. K. McLachlan,
J. Comput. Chem., 1988, 9, 522–538.
13 N. Zhou, P. Mascagni, W. A. Gibbons, N. Niccolai, C. Rossi and H.
Wyssbrod, J. Chem. Soc., Perkin Trans. 2, 1985, 581–587.
14 J. W. Trauger, R. M. Kohli and C. T. Walsh, Biochemistry, 2001, 40,
7092–7098.
15 L. M. Amzel, Proteins: Struct., Funct., Genet., 1997, 28, 144–149.
16 F. Kopp, C. Mahlert, J. Gru¨newald and M. A. Marahiel, J. Am. Chem.
Soc., 2006, 128, 16478–16479.
17 T.-L. Hwang and A. J. Shaka, J. Magn. Reson., Ser. A, 1995, 112,
275–279.
18 M. Haack, S. Enck, H. Seger, A. Geyer and A. G. Beck-Sickinger,
J. Am. Chem. Soc., 2008, 130, 8326–8336.
19 HyperChem, Hypercube, Inc., Gainesville, FL, 2000.
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 559–563 | 563
©