Design, construction and properties of peptide N-terminal cap templates devised to initiate α-helices. Part 3. Caps derived from N-[(2S)-2-chloropropionyl]-(2S)-Pro-(2R)-Ala-(2S,4S)-4-thioPro-OMe

Article abstract of DOI:10.1039/a806032k

The construction of a 12-membered macrocyclic template capable of entraining attached peptides in helical conformations from acyclic N-[(2S)-2-chloropropionyl]-(2S)-Pro-(2S)-Pro-(2S,4S)-4-thioPro-OMe precursors has been severely hampered by the problem of simultaneously aligning carboxamide dipoles in the transition state for cyclisation. Previously we provided a detailed conformational analysis of the system and tested two methods for forcing the acyclic precursor into the macrocyclic conformation required for helix initiation. First, the destabilisation of unwanted conformations in the transition state for cyclisation, and second, the stabilisation of the favoured transition state structure through the introduction of a hydrogen-bonding interaction. Both strategies were unsuccessful. A third strategy based upon removing the requirement for all of the carbonyl dipoles to align in the transition state at the same time was also tested and the results are presented here: The relaxation of the highly restrained C^α-N bond torsion for Pro³ in the acyclic precursor, through its substitution for a (2R)-alanine residue, effectively decouples the motion of the second carboxamide group from the C^α-N bond torsion and allows the second carboxamide group to rotate. This rotation allows a helical conformation to develop in the transition state to the macrocycle without the need to align all of the carboxamide dipoles and results in successful cyclisation to give template structures of the all trans (ttt) form. Derivatives of the template were prepared by extending the C-terminus and these were characterised by NMR spectroscopy and restrained simulated annealing. In deuterochloroform solution at low temperature, separate sets of NMR signals were observed for two rapidly interconverting helical conformational isomers of the thioether macrocycle based on (2R)-N-propionyl-(2S)-Pro-(2R)-Ala-(2S)-Pro which possessed an appended trialkylammonium ion. The free energy of activation for the transition (ΔG_c^?) was 48 kJ mol^-1. A similar time-averaged conformation was also observed in aqueous solution. At -80°C in dichloromethane the rate of conformational exchange was slowed sufficiently to obtain resonance assignments and NOE data separately for each isomer. In the minor isomer (40%), the four carbonyl oxygen hydrogen-bond acceptors of the template are aligned in an α-helical conformation and in the major conformer the Pro² carbonyl dipole was anti-aligned with the other three dipoles. Thus, the conformers differ in the orientation of one carbonyl group. Molecular modelling calculations showed that the minor isomer was stabilised by coulombic interactions between the trialkylammonium salt and the carbonyl group dipole moments.

Full text of DOI:10.1039/a806032k

View Article Online / Journal Homepage / Table of Contents for this issue
Design, construction and properties of peptide N-terminal cap
templates devised to initiate á-helices. Part 3.† Caps derived from
N-[(2S)-2-chloropropionyl]-(2S)-Pro-(2R)-Ala-(2S,4S)-4-thioPro-OMe
Arwel Lewis, Trevor J. Rutherford, John Wilkie, Thierry Jenn and David Gani*
School of Chemistry and Centre for Biomolecular Sciences, The Purdie Building,
The University, St. Andrews, Fife, UK KY16 9ST
Received 31st July 1998, Accepted 22nd September 1998
The construction of a 12-membered macrocyclic template capable of entraining attached peptides in helical
conformations from acyclic N-[(2S)-2-chloropropionyl]-(2S)-Pro-(2S)-Pro-(2S,4S)-4-thioPro-OMe precursors
has been severely hampered by the problem of simultaneously aligning carboxamide dipoles in the transition state
for cyclisation. Previously we provided a detailed conformational analysis of the system and tested two methods
for forcing the acyclic precursor into the macrocyclic conformation required for helix initiation. First, the
destabilisation of unwanted conformations in the transition state for cyclisation, and second, the stabilisation of
the favoured transition state structure through the introduction of a hydrogen-bonding interaction. Both strategies
were unsuccessful. A third strategy based upon removing the requirement for all of the carbonyl dipoles to align in
the transition state at the same time was also tested and the results are presented here. The relaxation of the highly
restrained Cα–N bond torsion for Pro3 in the acyclic precursor, through its substitution for a (2R)-alanine residue,
effectively decouples the motion of the second carboxamide group from the Cα–N bond torsion and allows the
second carboxamide group to rotate. This rotation allows a helical conformation to develop in the transition state
to the macrocycle without the need to align all of the carboxamide dipoles and results in successful cyclisation to
give template structures of the all trans (ttt) form. Derivatives of the template were prepared by extending the
C-terminus and these were characterised by NMR spectroscopy and restrained simulated annealing. In deutero-
chloroform solution at low temperature, separate sets of NMR signals were observed for two rapidly interconverting
helical conformational isomers of the thioether macrocycle based on (2R)-N-propionyl-(2S)-Pro-(2R)-Ala-(2S)-Pro
which possessed an appended trialkylammonium ion. The free energy of activation for the transition (∆Gc‡) was 48
kJ molϪ1. A similar time-averaged conformation was also observed in aqueous solution. At Ϫ80 ЊC in dichloro-
methane the rate of conformational exchange was slowed sufficiently to obtain resonance assignments and NOE
data separately for each isomer. In the minor isomer (40%), the four carbonyl oxygen hydrogen-bond acceptors of
the template are aligned in an α-helical conformation and in the major conformer the Pro2 carbonyl dipole was
anti-aligned with the other three dipoles. Thus, the conformers differ in the orientation of one carbonyl group.
Molecular modelling calculations showed that the minor isomer was stabilised by coulombic interactions between
the trialkylammonium salt and the carbonyl group dipole moments.
Introduction
It is well established that local electrostatic fields have a pro-
found effect upon reactivity in proteins. However, it is not cur-
rently possible to quantify the influence of helix polarity upon
the pKa of side chain functionalities. Isolated short peptides
tend towards random coils,1 and in order to increase the prob-
ability of secondary structure initiation the folded peptides
have required stabilisation in environments with an uncertain
dielectric constant. Therefore, access to isolated solvated helices
will be useful in constructing computational forcefields for
modelling chemical reactivity in protein reaction centres.
The design of stable folded polypeptides has been the focus
of many research groups in recent years,2,3 and much of the
work has been reviewed in the previous articles.4,5 Of particular
relevance to our own studies directed towards the construction
of an α-helical cap template was the report of a conformation-
ally constrained triproline macrocycle 1 by Kemp.6,7 This struc-
ture was extremely similar to one of our own targets 2, and was
designed to position four aligned carbonyl oxygen H-bond
acceptors below the plane of the macrocycle pointing towards
the C-terminal end of the helix. For nucleating α-helices the cap
† For Part 2, see ref. 5.
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806
3795

View Article Online
1 must cyclise in a conformation where the carbonyl dipoles are
aligned and all three of the amide bonds exist as trans rotomers,
the so called ‘ttt’ state. Macrocycle 1, however, adopts the ctt
and cct conformers in aqueous solution and does not possess
useful helix-initiating properties. We reasoned that in order to
form stable isolated α-helices the cap design must incorporate
factors to destabilise the potential alternative conformations or
stabilise the required ones. In the previous article we undertook
a detailed analysis of the 12-membered macrocyclic template
system 1 and several of its derivatives (2–5) and tested two
potentially useful methods for coaxing the acyclic precursors
into an α-helical conformation. First, the destabilisation of
unwanted conformations in the transition state for cyclisation
was achieved through the introduction of steric bulk at the C-2
positions of selected residues in the precursors (6–8). This
strategy was partially successful in preventing the formation of
unwanted macrocyclic products containing cis-amide bonds,
but was unsuccessful in that dimerisation rather than cyclis-
ation to the 12-membered macrocycles (e.g. 2) occurred.
Second, the strategy of stabilising the favoured transition state
structure was tested through the introduction of a potential
13-membered hydrogen-bonding interaction of the type
commonly found in α-helices. Preparatively, this necessitated
altering the fourth carbonyl group from a carboxylic ester to a
carboxamide in order to provide a hydrogen bond donor at the
C-terminus of the template 3. This strategy also was unsuccess-
ful because the acyclic precursor 9 was stabilised in a conform-
ation in which the thiol group nucleophile was held away from
the electrophilic α-chloroacylamide group. Thus, the hydrogen
bond stabilised the ground state such that any beneficial hydro-
gen bonding interactions with the first carbonyl group in the
transition state for cyclisation were largely cancelled.
A third strategy, based upon removing the requirement for all
of the carbonyl dipoles to align in the transition state at the
same time, was also tested and the results are presented here.
The relaxation of the highly restrained Cα–N bond torsion for
Pro3 in the acyclic precursor 6, through its substitution for a
(2R)-alanine residue, effectively decouples the motion of the
second carboxamide group from the Cα–N bond torsion and
allows the second carboxamide group to rotate. This leads to
the successful cyclisation of modified acyclic precursor 10 to
template structures (e.g. 11) possessing ttt amide bond stereo-
chemistry. We now report on the synthesis and characterisation
of these modified proline-based macrocycles, and on the prop-
erties of analogues extended at the C-terminus.
Scheme 1 Reagents and conditions: i, (2R)-N-(tert-butoxycarbonyl)-
alanine, BOP-Cl, DIEA, CH2Cl2, 0–5 ЊC, 5 h then rt, 4 days, 55%; ii,
DIAD, PPh3, AcSH, THF, 0 ЊC; 2.5 h then rt, 15 h, 92%; iii, TFA,
CH2Cl2, 0 ЊC, 90 min; iv, N-[(2S)-2-chloropropionyl]-(2S)-proline,
BOP-Cl, DIEA, CH2Cl2, 0–5 ЊC, 3 h then rt, 2 days, 68% (2 steps); v,
KOH, MeOH, H2O, 2.5 h; vi, Cs2CO3, DMF, 4 days, 25%; vii, MeOH,
KOH, H2O, 65 ЊC, 2.5 h, 69%.
Results and discussion
The synthesis of compound 10 followed analogous routes to
those developed for the preparation of the acyclic precursor 6.5
Thus, Boc-(2R)–Ala–OH was coupled with (2S,4R)-4-hydroxy-
proline methyl ester using BOP-Cl to give protected dipeptide
12 in 55% yield, Scheme 1. Mitsunobu inversion gave the thio-
acetate 13, and removal of the Boc group gave amine trifluoro-
acetate salt 14 cleanly in 92% yield over the 2 steps.
All of these compounds displayed the expected analytical
properties and NMR spectroscopy revealed the presence of
small amounts (10–20%) of a conformation with a cis Ala–Pro
amide bond. Such conformations were not observed in anal-
ogous Pro–Pro sequences in the absence of external stabilisation
of the cis-form.5
taken place. The impure material 10 was taken straight on and
subjected to cyclisation using caesium carbonate in DMF under
high dilution conditions at room temperature over 4 days,
Scheme 1. Column chromatography of the crude product
allowed the isolation of a compound displaying NMR spectra
corresponding to those expected of the 12-membered macro-
cycle 11. Gratifyingly, mass spectrometric analysis revealed this
to be the desired monomeric cyclic compound 11 ([M ϩ H]ϩ =
384). Some of the cyclic dimer 12 was also formed, but was
easily separated by chromatography. This methodology allowed
a 25% conversion of thioacetate 15 to the macrocyclic com-
pound 11. Optimisation of the process established that com-
pound 11 could be produced directly from the thioacetate 15
via action of a dilute solution of potassium hydroxide in
methanol–water at 65 ЊC for 2.5 h in an excellent 69% yield.
Peptide bond formation between amine trifluoroacetate 14
and N-[(2S)-2-chloropropionyl]-(2S)-Pro, prepared as des-
cribed previously,4 was mediated by BOP-Cl and gave 68% of
the Pro-(2R)–Ala–Pro peptide 15 which was fully characterised
and displayed the expected properties.
Surprisingly, mild alkaline hydrolysis of the thioester 15 to
the free thiol 10 did not give a pure product. It appeared that
some other process had occurred in addition to cleavage of the
thioester. Apparently, some thiolate alkylation had already
Conformational studies
Macrocycle 11 was crystallised as a hemihydrate from ethyl
acetate–hexane for analysis by X-ray diffraction. In the solid-
3796
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806

View Article Online
Fig. 2 NOEs observed for macrocycle 11 in [2H6]DMSO solution.
Evaluation of macrocycle 11 as an á-helix initiator
At first glance, the arrangement of carbonyl groups in macro-
cycle 11 may not appear to be favourable for initiating an
α-helix. The Pro2 carboxamide group has the opposite align-
ment to that required, and the Ala3 carbonyl is also appreciably
displaced from the position expected for an α-helix. However,
the first carboxamide is in approximately the correct position to
accept a hydrogen bond from the first amide NH group of an
attached peptide and, thus, induce α-helical geometry in at least
the first appended residue. Note that the formation of such a
H-bond requires the movement of the Ala3 carbonyl O-atom
into the correct position and that, as a result, the entire cap
structure would become α-helical, except the Pro2 carbonyl
group would stay anti-aligned in the absence of its own H-bond
donor, to minimise the molecular dipole. The next (second) NH
group of an attached peptide could be in the correct position to
hydrogen bond to the Pro2 carbonyl group, if this carbonyl
group is able to flip alignment, and there is no steric reason why
it should not do so. If it does align correctly, the Ala3 and Pro4
carbonyl groups would be in the desired position, as defined by
the constrain in the pyrrolidine ring of Pro4. Indeed, sub-
sequent H-bond formation would be to carbonyl groups that
are already positioned correctly through serving as NH H-bond
donors. Thus, after the formation of two H-bonds, the system
should behave as a perfect template for α-helix propagation.
This “conformational co-operativity” is favoured by the form-
ation of hydrogen bonds, but disfavoured on entropic grounds
and by repulsive dipolar interactions. For now then, we should
take heed of the fact that peptide helices display “fraying” at
their termini and, thus, for short added sequences we should
expect to observe significant motion at the C-terminus.
From the analysis above, it is evident that a rotation of the
Pro2 carboxamide (about the dihedral angles ψ of Pro2 and φ of
Ala3) would be essential to bring the carbonyl group into the
correct alignment for helix propagation. Therefore, it was an
important objective to assess the magnitude of activation
energy required for its re-alignment and how the template
might be modified to stabilise the correctly aligned rotamer. In
the study of the feasibility of α-helix induction by template 11,
which would require the synthesis of derivatives with appended
peptide chains, it was decided to address the question of Pro2
carboxamide dipole re-alignment first through the sequential
introduction of H-bond donors and/or positive charges.
One hydrogen bond donor was introduced through conver-
sion of the methyl ester 11 to the methylamide 16 via direct
aminolysis, Scheme 2. The required amide product was
recovered in quantitative yield and displayed the expected
analytical and spectroscopic properties.
Fig. 1 X-Ray structure of compound 11.
state structure 11 (see Fig. 1 and footnotes) all amide bonds
exist in the trans configuration, fulfilling the first objective.
However, the ttt sub-state adopted is apparently incapable of
helix initiation, as the second and third carboxamide carbonyl
group lack the required alignment. In fact, the molecule lacks
any significant dipole and although the first carboxamide group
(in the mercaptopropionyl bridge) possesses an approximately
α-helical geometry, the Pro2 carbonyl group has roughly oppo-
site alignment, and the Ala3 carbonyl is splayed out. Of course,
this is the expected situation in the absence of any H-bond
donors and the structure vindicates our original concerns
regarding the alignment of carboxamides dipoles in synthesis-
ing the macrocycle.4,5 Thus, it now appears certain that the
reason the acyclic precursor can cyclise is because the Pro2
carbonyl group can ‘about turn’ in the transition state and
minimise the repulsive forces associated with bringing together
aligned carbonyl dipoles. Note that in Kemp’s system 1 a sim-
ilar result was observed, but here, because the system was so
rigid, one or more amide bonds needed to rotate to the cis-form
in order to minimise the transition state energy for cyclisation.
Before elaborating the system it was necessary to verify that
the conformer present in the solid-state was not merely a con-
sequence of crystal packing. The solution state conformation
was characterised by NMR spectroscopy and restrained sim-
ulated annealing. In [2H6]DMSO solution, the macrocycle 11
displayed only one conformation, which corresponded almost
exactly to that found in the solid state. The observation of one
set of resonance lines for the macrocycle indicated that there
was no significant conformational exchange that was slow on
the NMR timescale, and any minor conformation was below
the limit of detection (~11% population). The diagnostic NOEs
which indicate the “up” alignment (i.e. anti-alignment) of the
Pro2 carbonyl group all involve the NH proton of the Ala3
residue. This proton shows NOE interactions with the α-CH
hydrogen of Pro2 and both of the δCH2 protons of Pro4 (Fig. 2),
which also characterises a trans configuration for the amide
bond. The absence of NOEs to either Pro2 βCH2 proton argues
against any significant population of the α-helical target con-
formation with the Pro2 carbonyl aligned “down”.
Conformational analysis of compound 16 in H2O–2H2O
(9:1) indicated that there was no change in the conformation of
the macrocycle upon addition of the hydrogen bond donor. The
Pro2 carbonyl group did not flip to form a 310-helical hydrogen
bonding network with the methylamide NH proton. Experi-
ments to establish whether there was an α-helical 13-membered
hydrogen bonding network connecting the methylamide NH to
the mercaptopropionyl carbonyl group were inconclusive. Only
weak NOE cross peaks were observed to support the notion
that the methylamine group may spend some time folded
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806
3797

View Article Online
methylamide. The protected dipeptide 21 was obtained in 67%
yield as colourless needles and displayed the expected proper-
ties. Removal of the Boc protecting group using trifluoroacetic
acid afforded the trifluoroacetate salt 20 which, without further
purification, was coupled to the template cap acid 17 to give the
required conjugate 19 in 33% yield (Scheme 3).
Scheme 2 Reagents and conditions: i, MeNH2, MeOH, 0 ЊC, 2 days,
100%; ii, NaOH, MeOH, H2O, 2 h, 83%; iii, (2S)-phenylalanine methyl-
amide, BOP-Cl, DIEA, CH2Cl2, 0–5 ЊC, 2 h then rt, 20 h, 62%.
underneath the macrocycle. This was the expected result since a
large movement of the Pro4 carboxamide moiety and the proper
alignment of the Ala3 carbonyl dipole would be required to
form a 13-membered H-bonding interaction.
In order to attach more than one hydrogen bond donor site
to the template it was necessary to hydrolyse the methyl ester 11
and perform subsequent peptide coupling steps. The ester was
saponified to afford the acid 17, as colourless crystals in 83%
yield. The acid displayed the expected analytical and spectro-
scopic properties and would be a precursor for several deriv-
atives. The template cap acid 17 was coupled to (2S)-phenyl-
alanine methylamide using BOP-Cl activation to give the
template–peptide conjugate 18 in 62% yield (Scheme 2), which
possessed the expected analytical properties.
In C2HCl3 solution, the macrocyclic portion of compound 18
showed no change in conformational preference relative to that
in the methyl ester 11. However, several NOEs indicate that
there was some interaction between the lower face of the macro-
cycle and the attached peptide. This implied that the Phe5 resi-
due spent some time folded underneath the macrocycle in a
position conducive to the formation of a hydrogen bond
between the Phe NH and the mercaptopropionyl carbonyl
group. The existence of such a hydrogen bonding interaction
requires the Phe5 residue to adopt an α-helical position. This
notion is supported by a 0.2 ppm upfield shift for the methyl
protons of the mercaptopropionyl residue, which is consistent
with a ring current effect caused by the close proximity of the
phenyl ring. There was also a 0.8 ppm downfield shift of the
Phe5 NH proton relative to its position in a model linear peptide
in which it is known intramolecular hydrogen bonding is
absent. While the folded conformation is evidently populated to
some extent, other evidence indicates that the attached peptide
is extremely mobile. For example, some NOE cross-peaks are
present which are not possible for the folded conformation, and
Scheme 3 Reagents and conditions: i, IBCF, NMM, THF, Ϫ15 ЊC→
rt, 2 h, 67%; ii, TFA, CH2Cl2, 0 ЊC, 2 h; iii, 17, BOP-Cl, DIEA, CH2Cl2,
0–5 ЊC, 2 h then rt, 18 h, 33% (2 steps).
NMR studies performed on the template conjugate 19 in
C2HCl3 solution provided convincing evidence for the presence
of a 13-membered intramolecular hydrogen bond network
between the NH of the appended (2S)-Ala5 residue and the first
carbonyl group. There was a relatively strong set of NOEs sim-
ilar to those obtained for compound 18, indicative of the pep-
tide folding underneath the macrocyclic template. The 3JNH–CHα
value for the attached alanine residue was 3.9 0.4 Hz, from
which we estimate a φ dihedral angle of Ϫ57Њ,8 which is consist-
ent with α-helical geometry. These data therefore indicate that
the macrocycle possesses α-helical character and that the first
attached amino acid residue, Ala5, is part of the helical struc-
ture, but not the subsequent residues. The less mobile, tighter
α-helical conformation obtained for this conjugate over deriv-
ative 18 could reflect the higher intrinsic α-helical propensity
of Ala compared with Phe,10,11 or the fact that the fraying
C-terminus had been moved further away by one residue.
3
the JNH-α value for the Phe5 residue is 8.0 Hz, which is not
consistent with α-helical geometry.8 Thus, as expected, the
C-terminus shows the properties of fraying9 and a good deal of
mobility.
In order to provide hydrogen bond donors for each acceptor
carboxamide group in the macrocycle 11, one further derivative
containing an attached dipeptide amide was required, com-
pound 19. The peptide component (2S)-Ala-(2S)-Phe-NHMe
20 was synthesised by activating Boc-(2S)-Ala-OH using the
mixed anhydride method and adding (2S)-phenylalanine
3798
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806

View Article Online
Whatever the cause, it was evident that the Pro2 carbonyl group
in conjugate 19 was still aligned antiparallel to the aligned
mercaptopropionyl and Ala3 carbonyl groups. From this anal-
ysis we must conclude that the enthalpic benefit of hydrogen
bonding between short appended peptides and the template in
the “helical” conformation appears unable to counterbalance
the repulsive dipolar forces and entropic losses associated with
this conformation.
Hydrogen bonds are not the only stabilising forces acting
upon peptide and protein α-helices in nature. Another common
stabilising feature is the presence of an associated counter
charge near one of the helix termini which can interact favour-
ably with the helix dipole.12–14 The desired alignment of
carbonyl groups in the parent macrocycle is destabilised by
repulsive dipole–dipole interactions but this arrangement could
be stabilised by a positively charged group located on the helix
axis at the C-terminus. The latter effect was demonstrated by
Gellman et al. who appended a tetraalkylammonium group at
the C-terminus of a depsipeptide with the resulting induction
of an α-helical conformation.15 The introduction of an anionic
group at the N-terminus should exert a similar effect, but this is
synthetically challenging and an on-going objective in our
laboratory. However, since the conjugates 18 and 19 were show-
ing helical characteristics, it was clearly of interest to further
investigate the properties of the parent macrocycle 11 by
placing a positive charge at the C-terminus. The question of
whether the Pro3 carboxamide group in conjugates of 11 could
be aligned with the other template derived carboxamide groups
needed to be addressed.
We chose to attach dimethylethylenediamine at the C-
terminus of the macrocycle 17 since this strategy would intro-
duce a hydrogen bond donor, and, after protonation of the
tertiary amine, a suitably disposed cationic group. The amide
bond formation was achieved by reacting the mixed anhydride
of template acid 17 directly with N,N-(dimethyl)ethylene-
diamine to afford the conjugate 22 as colourless crystals in
78% yield, Scheme 4. The compound displayed the expected
analytical and spectroscopic properties and was directly proton-
ated with trifluoroacetic acid in DCM to give the required tri-
alkylammonium salt 23.
Scheme 4 Reagents and conditions: i, N,N-(dimethyl)ethylenediamine,
IBCF, NMM, THF, DMF, Ϫ15 ЊC→rt, 4 h, 78%; ii, TFA, CH2Cl2,
100%.
ation of spin systems by NMR filtration techniques requiring
more than a few milliseconds mixing time. Only one NOE
between two non-exchangeable protons, the NOE between Ala3
Hα and Pro4 Hδ, differs significantly in the two conformers
(Table 2). NOE assignments for conjugate 23 at Ϫ80 ЊC were
accomplished only for the NOEs involving NH protons (Table
2), but both conformers could be characterised sufficiently
from the assignment of amide proton NOEs alone. In the major
conformation the NOE between Ala3 NH and Pro2 Hα was
strong, whilst that between Ala3 NH and Ala3 Hα was weak. In
the minor conformer these relative intensities were reversed,
indicating that the conformational difference was the orienta-
tion of the Pro2–Ala3 amide group. The variation was not,
however, an amide bond isomerisation; a cis-amide bond would
give no NOE between Ala3 NH and any Pro2 protons, whereas
NOEs were observed from Ala3 NH to Pro2 Hα in the major
isomer and to Pro2 Hβ and Pro2 Hδ in the minor isomer.
Relative NOEs were quantified using cross-peak volumes from
a NOESY experiment, with 200 ms mixing time, for use as
modelling restraints during simulated annealing (vide infra).
It was hoped that the helix cap would be capable of tem-
plating helices in neat aqueous solution, without additional
co-solvents. In aqueous solution (95% H2O, 5% D2O, 50 mM
Tris-d11 buffer, pH 7.5) the 13C spectra of compound 23 showed
one set of resonance lines. Acetone-d6 (15% v/v) was added to
facilitate low temperature solution studies, and had an insignifi-
cant effect on 13C chemical shifts, from which we infer that a
low concentration of acetone did not alter the conformational
preference. When cooled to Ϫ15 ЊC, some resonances of Pro2
In deuterochloroform solution the 1H and 13C NMR spectra
of conjugate 23 showed one set of resonances, but with signifi-
cant broadening of the resonance lines for the Pro2 Cα and CO
3
₁
(∆ν = 30 Hz for Cα at 30 ЊC). NOESY assignments for the Ala
₂
amide proton indicated several short H–H distances that could
not be realised simultaneously in a single conformation. On
cooling, the sample exhibited two distinct sets of resonances
(Fig. 3). The free energy of activation at coalescence (∆Gc‡) can
be estimated from the following expression:16
∆Gc‡ = 4.57 Tc [9.97 ϩ log (Tc/∆ν)]
where Tc is the coalescence temperature and ∆ν is the difference
in resonance frequency of the two spins in the slow exchange
limit. For the Ala3 NH, with a coalescence temperature of
243 5 K and ∆ν of 87 Hz, the calculated free energy of acti-
vation for the transition is 48 kJ molϪ1. The corresponding
exchange rate at coalescence was 190 sϪ1.
At Ϫ55 ЊC conformational exchange rates were sufficiently
low to obtain full 1H and 13C resonance assignments for each of
two conformational isomers, with a relative abundance of 3:2
(estimated from peak integrals of resolved amide proton reson-
ances). For a two-site conformational exchange with relative
populations in the ratio 3:2 the difference in free energy of the
two sites is 1.0 kJ molϪ1. However, the conformational exchange
was too rapid to separate the NOEs for each isomer.
₁
In d2-dichloromethane solution at Ϫ80 ЊC the full assignment
of NOE spectra was still complicated by residual exchange
between two conformers with almost degenerate chemical shifts
(Table 1). Conformational exchange also precluded the separ-
broadened appreciably; ∆ν for Cβ and the carbonyl carbon
₂
were ~40 and >100 Hz, respectively, and the Cα resonance was
similarly broadened (but not resolved). The line broadening
was consistent with the observations in deuterochloroform
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806
3799

View Article Online
Table 1 1H Resonance assignments for macrocycle 23 in CD2Cl2 at Ϫ80 ЊC
NH
Hα
HαЈ
Hβ
HβЈ
Hγ
HγЈ
Hδ
HδЈ
Major isomer
Thp1
—
—
8.609
—
8.029
3.631
4.802
4.494
4.504
3.811
—
—
—
—
1.259
2.201
1.098
2.372
3.374
—
1.868
—
2.186
3.241
—
2.392
—
3.828
9.319
—
2.029
—
—
—
—
3.598
—
3.589
2.873
—
3.450
—
3.086
2.804
Pro2
Ala3
Pro4
TAA5,a
3.489
Minor isomer
Thp1
—
—
8.769
—
9.456
3.799
4.801
3.809
4.270
3.911
—
—
—
—
1.274
2.184
1.350
2.402
3.471
—
1.864
—
1.861
3.072
—
2.335
—
3.924
7.737
—
2.009
—
—
—
—
4.027
—
3.594
2.971
—
3.519
—
3.476
2.819
Pro2
Ala3
Pro4
TAA5
3.316
a TAA = trialkylammonium, where NH is the amide proton, and Hγ is the ammonium proton.
Fig. 3 1H NMR spectra of macrocycle 23, cyclic[Pro–Ala–Pro(mercaptopropionyl)] NHCN2CH2NHMe2, in deuterochloroform solution at (a)
30 ЊC and (b) Ϫ55 ЊC.
solution. Additionally, there was broadening of the N-methyl
carbon resonances. Intensities of NOEs (measured from a
NOESY spectrum with 200 ms mixing time) were characteristic
of the two rapidly exchanging conformers observed previously
in dichloromethane.
Models for each isomer were generated by restrained simu-
lated annealing, using the distance restraints derived from the
dichloromethane solution at Ϫ80 ЊC, as shown in Table 2. Two
sets of simulations were performed, applying the constraints
from the two isomers separately. AM1 in vacuo semi-empirical
energy minimizations were performed on the lowest energy
annealed model of each isomer, to give the optimised structures
depicted in Fig. 4. In the AM1 optimised model of the target
conformation [Fig. 4(b)] the ammonium proton is situated
equidistant (2.2 Å) from the carbonyl oxygens of Pro2 and the
mercaptopropionyl moiety, suggestive of possible hydrogen
bonding, but had no interactions with Ala3.
Restrained and unrestrained molecular dynamics simulations
for the minor (40%) conformation using a relative permittivity
of four (approximating to a non-polar environment) showed
no evidence of hydrogen bonding between the macrocycle and
the pendant chain. Instead, the trialkylammonium cation
was positioned beneath the centre of the macrocycle, allowing
maximum charge–dipole interaction with the aligned carbonyl
groups. When the relative permittivity was increased (to 80) to
simulate an aqueous environment, the trialkylammonium
group was found to drift away from the position beneath the
macrocycle.
This striking change in macrocycle conformation shows that
the generation of potential α-helix initiating templates is pos-
sible by judicious use of charge–dipole interactions. Modelling
studies suggest that these initiating forms are stabilised solely by
the charged group effect without contribution from hydrogen
bonding.
3800
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806

View Article Online
Table 2 Modelling restraints, and comparison of experimental vs. modelled internuclear distances for amide proton signals of macrocycle 23 in
CD2Cl2 solution at Ϫ80 ЊC
Major isomer
Minor isomer
From
To
Restraintsa
rnoe b/Å
rmodel c/Å
Restraintsa
rnoe b/Å
rmodel c/Å
Ala3 NH
Ala3 NH
Ala3 NH
Ala3 NH
Ala3 NH
Ala3 NH
Ala3 NH
Ala3 Hα
Ala3 Hβ
Pro2 Hα
Pro2 Hβ
Pro2 Hγ
Pro2 Hδ
Pro4 Hδ
weak
weak
strong
—
—
—
2.7
2.7
2.2
—
—
—
3.0
3.1
2.2
4.5
5.9
4.3
3.1
strong
weak
—
weak
weak
weak
strong
2.6
3.7
—
3.4
2.9
3.3
2.7
2.3
3.3
3.6
2.4
4.1
2.7
2.4
weak
2.7
TAA5 NH
TAA5 NH
TAA5 NH
TAA5 NH
TAA5 NH
TAA5 NH
TAA5 NH
TAA5 NH
Pro2 Hα
Pro4 Hα
Pro4 Hδ
weak
medium
weak
strong
strong
weak
3.4
2.7
3.1
2.4
2.2
2.6
2.9
3.8
4.4
3.4
3.6
2.9
3.0
2.7
3.3
4.9
—
—
6.2
2.3
5.1
2.4
3.0
3.7
3.1
4.8
weak
weak
medium
medium
weak
—
4.0
3.7
3.2
2.9
3.7
—
TAA5Hα
TAA5 HαЈ
TAA5 Hβ
TAA5 Hδ
Ala3 Hβ
weak
weak
—
—
TAA5 Hγ
TAA5 Hγ
TAA5 Hγ
TAA5 Hγ
TAA5 Hγ
TAA5 Hγ
TAA5 Hγ
TAA5 Hγ
TAA5 Hα
TAA5 HαЈ
TAA5 Hβ
TAA5 HβЈ
TAA5 Hδ
TAA5 HδЈ
Pro4 Hδ
medium
medium
medium
medium
strong
strong
—
2.9
2.7
2.5
2.6
2.2
2.2
—
2.7
3.5
2.6
2.9
2.4
2.4
4.7
4.0
—
—
—
—
medium
—
weak
—
—
—
—
3.3
2.9
—
3.2
3.8
2.5
3.0
2.4
2.4
5.0
5.9
3.0
Ala1 Hβ
weak
3.2
a Defined as weak (r = 1.8–5.0 Å), medium (r = 1.8–3.3 Å), strong (r = 1.8–2.7 Å); TAA = trialkylammonium. b Effective time-averaged internuclear
distance calculated from relative volume of NOESY cross-peaks, using the isolated spin-pair approximation (i.e. assuming a rigid molecule with
isotropic reorientation). c Internuclear distance in lowest energy model from simulated annealing.
fragments were given as percentages of the base peak intensity
(100%). Infrared spectra were recorded using a Perkin-Elmer
Experimental
1710 FT-IR spectrometer. The samples were prepared as Nujol
mulls or thin films between sodium chloride discs. Absorption
maxima are given in wavenumbers (cmϪ1) relative to a poly-
styrene standard. Melting points were measured using an
electrothermal melting point apparatus and are uncorrected.
Optical rotations were measured on an Optical Activity AA-
1000 polarimeter using 10 cm path length cells at room tem-
perature and are given in units of 10Ϫ1 deg cm2 gϪ1.
All experiments were performed at room temperature (20–
25 ЊC) unless otherwise stated. Flash chromatography was per-
formed according to the method of Still et al.18 using Fluka C60
(40–60 mm mesh) silica gel. Analytical thin layer chromato-
graphy was carried out on 0.25 mm precoated silica gel plates
(Macherey–Nagel SIL g/UV254) and compounds were visual-
ized using UV fluorescence, iodine vapour, ninhydrin, ethanolic
phosphomolybdic acid or aqueous potassium permanganate.
Light petroleum refers to the fraction boiling at 40–60 ЊC.
Solvents and common reagents were purified according to
the method of Perrin et al.19 Thionyl chloride was distilled
over sulfur and the initial fractions were always discarded.
N-Methylmorpholine was distilled over ninhydrin. All other
reagents were used without further purification.
In the molecular modelling studies a wide conformational
space was sampled by generating fifty pseudo-random starting
structures using unrestrained MD simulations at high temper-
ature (750 K). Each model was minimized by restrained simu-
lated annealing (1 ps of MD at each temperature from 500–300
K, in steps of 50 K, then from 290–10 K in steps of 10 K, and
finally at 5 K, followed by conjugate gradients minimization).
The lowest energy model was taken as the starting structure for
a 510 ps MD simulation at 303 K, storing the atomic coordin-
ates after every 250 timesteps of 1 fs. Initial atom velocities were
assigned from a Maxwell–Boltzmann distribution at 303 K, and
the thermal energy was scaled by coupling to a thermal bath
with a time constant of 0.6 ps. The cutoff distance for calculat-
ing non-bonded interactions was 15 Å. NOE restraints were
NMR spectra were recorded on a Bruker AM-300 spectrometer
(1H, 300 MHz; 13C, 75.4 MHz), a Varian Gemini spectrometer
(1H, 200 MHz; 13C, 50.3 MHz) a Varian Gemini spectrometer
(1H, 300 MHz; 13C, 75.4 MHz) and a Varian Unity Plus
1
500 spectrometer (1H, 500.3 MHz; 13C, 125.7 MHz). H NMR
spectra were referenced internally to (C2H3)2SO (δ 2.47), 2HOH
(δ 4.68) or C2HCl3 (δ 7.27). 13C NMR were referenced to
(C2H3)2SO (δ 39.70) or C2HCl3 (δ 77.5). J values are given
in Hz. Carbon and proton resonances of amino acids in NMR
spectra are assigned as α, β, γ and δ according to normal
convention. Where more than one conformational isomer is
present due to the presence of a tertiary amide bond, these are
assigned as c (cis) or t (trans), according to the isomeric state of
the amide bond. If the isomeric states of the amide bonds are
not known, the conformations are assigned as A, B, C etc. For
definitions of Chp and Thp see previous paper.
For conformational analysis of the trialkylammonium salt,
1
23, 500 MHz H NMR spectra were recorded with nominal
1
probe temperatures of 30 and Ϫ80 ЊC. One-dimensional H
spectra were acquired with 0.5 Hz/point digital resolution and a
recycle time of 11 s/transient to allow accurate quantification
from peak integrals. TOCSY experiments were acquired with
80 ms MLEV-17 spin-lock (7.2 kHz field), 4 scans per FID, and
a digital resolution of 2.6 and 10.4 Hz/point in f2 and f1, respect-
ively. The NOESY spectra were acquired with 200 ms mixing
time, 8 scans per FID and a digital resolution of 2.1 Hz/point in
f2 and 8.5 Hz/point in f1. Spectra were processed with cosine-
bell weighting in each dimension. For analysis in aqueous solu-
tion the corresponding pulse sequences were modified with
double pulsed field gradient spin-echo solvent suppression,17
with selective inversion of water magnetization in a 230 Hz
field, and z-gradient pulse pairs of 0.1 and 2.0 ms at 9 G cmϪ1.
Mass spectra and accurate mass measurements were
recorded on a VG 70-250 SE, a Kratos MS-50 or by the SERC
service at Swansea using a VG AZB-E. Fast atom bombard-
ment spectra were recorded using glycerol as a matrix. Major
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806
3801

View Article Online
mmol) and N,N-diisopropylethylamine (10.1 cm3, 58.4 mmol)
in dry dichloromethane (250 cm3) was treated with BOP-Cl
(10.8 g, 42.4 mmol) and the resulting suspension stirred under
nitrogen at 0 ЊC for 25 min. A suspension of methyl (2S,4R)-4-
hydroxyprolinate hydrochloride (6.68 g, 36.8 mmol) in dry
dichloromethane (50 cm3) was then added. The mixture was
stirred at 0 ЊC for 5 h, then allowed to warm to room temper-
ature and stirred for a further 4 days. The solution was washed
with aqueous HCl (0.5 mol dmϪ3, 2 × 150 cm3), aqueous
sodium hydrogen carbonate (5%, 2 × 150 cm3) and brine (150
cm3), then dried (MgSO4). The solvent was removed under
reduced pressure to afford the dipeptide 12 as a white foam
(6.34 g, 55%) (HRMS: Found [M ϩ H]ϩ, 317.1721. C14H25N2O6
requires 317.1713); [α]D Ϫ16.3 (c 1.0 in MeOH); νmax(CH2Cl2)/
cmϪ1 3425 (NH), 2981 (CH), 1747 (ester CO), 1708 (urethane
CO), 1653 (tertiary amide CO) and 1171 (C–O); δH(300 MHz;
C2HCl3) 1.31 [3 H, d, J 6.9, βCH3(Ala)], 1.42 [9 H, s, C(CH3)3],
2.00–2.11 [1 H, m, ¹βCH2(Hyp)], 2.25–2.37 [1 H, m,
¯
²
¹βCH2(Hyp)], 3.56 [1 H, d, J 11.0, ¹δCH2(Hyp)], 3.73 (3 H, s,
¯
²
¯
²
CO2CH3), 3.83 [1 H, dd, J1 11.0, J2 4.4, ¹δCH2(Hyp)], 4.38–4.63
¯
²
[3 H, m, αCH(Hyp), αCH(Ala) and γCH(Hyp)] and 5.47 (1 H,
d, J 8.0, NHCH3); δC(50.31 MHz; C2HCl3) 18.70 [βCH3(Ala)],
28.64 [C(CH3)3], 37.65 [βCH2(Hyp)], 48.22 [αCH(Ala)], 52.65
(CO2CH3), 55.23 [δCH2(Hyp)], 58.30 [αCH(Hyp)], 68.43
[γCH(Hyp)], 80.02 [C(CH3)3], 155.46 (urethane CO), 172.19
(amide CO) and 173.03 (ester CO); m/z (CI) 317 (40%,
[M ϩ H]ϩ), 261 (100, [M Ϫ C4H8 ϩ H]ϩ), 217 (28, [M Ϫ CO-
ϩ
2C4H8 ϩ H]ϩ) and 57 (10, C4H9 ).
Methyl (2S,4S)-N-[(2R)-N-(tert-butoxycarbonyl)alanyl]-4-
(acetylthio)prolinate 13
Diisopropyl azodicarboxylate (7.30 g, 34.3 mmol) was added to
a solution of triphenylphosphine (9.7 g, 37.0 mmol) in THF
(100 cm3) at 0 ЊC. The resulting suspension was stirred for 30
min, after which time a solution of alcohol 12 (6.10 g, 19.0
mmol) and thioacetic acid (2.75 cm3, 37.3 mmol) in THF (50
cm3) was added dropwise. The mixture was stirred at 0 ЊC for
2.5 h and then at room temperature overnight after which the
solvent was removed under reduced pressure. Excess triphenyl-
phosphine and its oxide were removed by crystallisation
(from light petroleum–ethyl acetate) and filtration. The remain-
ing oil was purified by column chromatography using light
petroleum–ethyl acetate as the eluent to afford the thioacetate
13 as a clear colourless oil (6.64 g, 92%); Rf 0.49 (ethyl acetate)
(HRMS: Found [M ϩ H]ϩ, 375.1593. C16H27N2O6S requires
375.1590); [α]D Ϫ24.1 (c 0.2 in MeOH); νmax(CH2Cl2)/cmϪ1 3428
(NH), 2982 (CH), 1749 (ester CO), 1703 (urethane and
thioester CO), 1655 (tertiary amide CO) and 1171 (C–O);
δH(300 MHz; C2HCl3) (2 conformations, tt and tc) 1.26 [3 H, d,
J 6.6, βCH3(Ala)], 1.36 [tc, 9 H, s, C(CH3)3], 1.38 [tt, 9 H, s,
C(CH3)3], 1.84–1.95 [1 H, m, ¹βCH2(Pro)], 2.26 (tc, 3 H,
Fig. 4 AM1-optimised models of (a) the major isomer, and (b) the
minor isomer of the cap macrocycle with attached trialkylammonium
salt, 23.
applied as symmetrical biharmonic potentials, with a force con-
stant of 10.0 kcal molϪ1 ÅϪ2 and an upper limit of 10.0 kcal
molϪ1 per restraint. No energy penalty was applied for res-
traints that were within the distance bounds; strong = 1.8–2.7
Å, medium = 1.8–3.3 Å, weak = 1.8–5.0 Å.
Molecular mechanics computations were performed using
the DISCOVER package (Biosym, San Diego) with the CVFF
force-field parameters. Calculations were performed in vacuo,
with a fixed dielectric constant to partly simulate bulk solvent.
Partial charges for the two conformers were generated from
semi-empirical AM1 energy calculations.
¯
²
SCOCH3), 2.30 (tt, 3 H, SCOCH3), 2.60–2.72 [tt, 1 H, m,
¹βCH2(Pro)], 2.72–2.84 [tc, 1 H, m, ¹βCH2(Pro)], 3.36 [tc, 1 H,
¯
¯
²
²
dd, J1 12.9, J2 4.7, ¹δCH2(Pro)], 3.52 [tt, 1 H, dd, J1 8.8, J2 7.0,
¯
²
¹δCH2(Pro)], 3.69 (tt, 3 H, s, CO2CH3), 3.74 (tc, 3 H, s,
¯
²
CO2CH3), 3.92–4.19 [2 H, m, γCH(Pro) and ¹δCH2(Pro)], 4.35–
¯
²
4.44 [2 H, m, αCH(Ala) and tc, αCH(Pro)], 5.00–5.14 [tt, 1 H,
m, αCH(Pro)] and 5.33 (1 H, d, J 8.2, NHCH3); δC(75.44
MHz; C2HCl3) 18.27 [tc, βCH3(Ala)], 18.86 [tt, βCH3(Ala)],
28.60 [C(CH3)3], 30.79 (SCOCH3), 34.83 [tt, βCH2(Pro)],
37.67 [tc, βCH2(Pro)], 39.04 [tc, γCH(Pro)], 39.83 [tt,
γCH(Pro)] 47.55 [tc, αCH(Ala)], 48.15 [tt, αCH(Ala)], 52.38,
52.70, 52.76 and 53.03 [δCH2(Pro) and CO2CH3], 58.77
[αCH(Pro)], 79.94 [tt, C(CH3)3], 80.18 [tc, C(CH3)3], 155.42
[tt, urethane CO), 155.82 (tc, urethane CO), 171.74 and
172.03 (tt, amide and ester CO), 172.75 and 173.32 (tc, amide
and ester CO) and 195.09 (thioester CO); m/z (CI) 375 (41%,
[M ϩ H]ϩ), 319 (100, [M Ϫ C4H8 ϩ H]ϩ) and 275 (31, [M Ϫ
CO2C4H8 ϩ H]ϩ).
Abbreviations
Boc, tert-butoxycarbonyl; BOP-Cl, N,N-bis(2-oxooxazolidin-3-
yl)phosphinic chloride; DCM, dichloromethane; DIAD, diiso-
propyl azodicarboxylate; DIEA, N,N-diisopropylethylamine;
DMF, N,N-dimethylformamide; DMSO, dimethyl sulfoxide;
IBCF, isobutyl chloroformate; NMM, N-methylmorpholine;
NOE, nuclear Overhauser effect; NOESY, NOE correlation
spectroscopy; TFA, trifluoroacetic acid; THF, tetrahydrofuran.
Methyl (2S,4R)-N-[(2R)-N-(tert-butoxycarbonyl)alanyl]-4-
hydroxyprolinate 12
A solution of (2R)-(tert-butoxycarbonyl)alanine (8.00 g, 42.3
3802
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806

View Article Online
Methyl (2S,4S)-N-[(2R)-N-{(2S)-N-[(2S)-2-chloropropionyl]-
prolyl}alanyl]-4-(acetylthio)prolinate 15
methane–methanol (24:1) as the eluent to afford macrocycle 11
as colourless crystals (70 mg, 25%), mp 202–204 ЊC (Found: C,
51.9; H, 6.45; N, 10.55. C17H25N3O5Sؒ¹H2O requires C, 52.05;
¯
²
A solution of thioester 13 (3.25 g, 8.7 mmol) in dichloro-
methane (30 cm3) at 0 ЊC was treated with trifluoroacetic acid
(15 cm3) and the solution stirred at 0 ЊC for 90 min. The
solvents were removed under reduced pressure to give the tri-
fluoroacetate salt 14 as a yellow oil.
H, 6.7; N, 10.7%); [α]D Ϫ149.9 (c 0.5 in MeOH); νmax(CH2Cl2)/
cmϪ1 3308 (NH), 2953 (CH), 1763 (ester CO), 1682 (secondary
amide CO), 1663 (tertiary amide CO) and 1450 (CH def.);
δH(500 MHz; [2H6]DMSO) 1.27 [3 H, d, J 6.5, βCH3(Thp1)], 1.28
[3 H, d, J 6.6, βCH3(Ala3)], 1.57–1.62 [1 H, m, ¹βCH2(Pro4)],
¯
²
A solution of (2S)-N-[(2S)-2-chloropropionyl]proline (2.37
g, 11.5 mmol) and N,N-diisopropylethylamine (6.0 cm3, 34.5
mmol) in dichloromethane (120 cm3) was treated with BOP-Cl
(3.10 g, 12.2 mmol) at 0 ЊC. After 45 min, a solution of amine
trifluoroacetate 14 in dichloromethane (40 cm3) was added and
the mixture was stirred at 0–5 ЊC for 3 h, then at room temper-
ature for 2 days. The solution was washed with aqueous HCl
(0.5 mol dmϪ3, 2 × 50 cm3), aqueous sodium hydrogen carbon-
ate (5%, 2 × 50 cm3) and brine (80 cm3). The organic phase was
dried (MgSO4), and the solvent was removed under reduced
pressure. The residual oil was purified by column chrom-
atography using ethyl acetate as the eluent to afford the thioester
15 as a thick yellow oil (2.73 g, 68%), Rf 0.28 (Found: C, 47.85;
H, 6.3; N, 8.65. C19H28ClN3O6SؒH2O requires: C, 47.55; H, 6.3;
N, 8.75%) (HRMS: Found (M ϩ H)ϩ, 462.1457. C19H2935ClN3-
O6S requires 462.1466); [α]D Ϫ312.5 (c 0.9 in MeOH);
νmax(CH2Cl2)/cmϪ1 3404, 3310 (NH), 2955 (CH), 1748 (ester
CO), 1693 (secondary amide and thioester CO) and 1658 (ter-
tiary amide CO); δH(300 MHz; C2HCl3) (2 conformations, ttt
and ttc) 1.28 [3 H, d, J 6.6, βCH3(Ala3)], 1.63 [ttc, 3 H, d, J 6.3,
βCH3(Chp1)], 1.64 [ttt, 3 H, d, J 6.6, βCH3(Chp1)], 1.85–2.26
[5 H, m, βCH2(Pro2), ¹βCH2(Pro4) and γCH(Pro2)], 2.27 (ttc,
1.80–1.88 [1 H, m, ¹βCH2(Pro4)], 1.96–2.00 [1 H, m, ¹γCH2-
¯
²
¯
²
(Pro2)], 2.10–2.14 [1 H, m, ¹βCH2(Pro2)], 2.20–2.28 [1 H, m,
¯
²
¹γCH2(Pro2)], 2.48–2.55 [1 H, m, ¹γCH2(Pro2)], 2.90 [1 H, dd,
¯
²
¯
²
J1 11.8, J2 6.8, ¹δCH2(Pro4)], 3.12–3.17 [1 H, m, γCH(Pro4)],
¯
²
3.61–3.67 [2 H, m, δCH2(Pro2)], 3.65 ( 3 H, s, CO2CH3), 4.09 [1
H, q, J 6.9, αCH(Thp1)], 4.09–4.16 [1 H, m, αCH(Ala3)], 4.31 [1
H, dd, J1 11.7, J2 4.8, ¹δCH2(Pro4)], 4.41 [1 H, dd, J1 9.3, J2 5.4,
¯
²
αCH(Pro4)], 4.88 [1 H, d, J 7.2, αCH(Pro2)] and 8.17 [1 H, d,
J 4.3, NH]; δC(125.73 MHz; [2H6]DMSO) 16.45 [βCH3(Ala3)],
16.73 [βCH3(Thp1)], 24.41 [γCH2(Pro2)], 25.62 [βCH2(Pro2)],
33.56 [βCH2(Pro4)], 40.26 [αCH(Thp1)], 41.56 [γCH(Pro4)],
46.38 [δCH2(Pro2)], 50.85 [δCH2(Pro4)], 51.08 [αCH(Ala3)],
51.79 (CO2CH3), 57.12 [αCH(Pro4)], 57.70 [αCH(Pro2)], 169.26
[CO(Ala3)], 170.17 [CO(Pro2)], 171.05 [CO(Pro4)] and 171.48
(ester CO); m/z (FABϩ) 406 (5%, [M ϩ Na]ϩ), 384 (64, [M ϩ
H]ϩ) and 128 {84, [NC4H6(S)CO]ϩ}.
Compound 11 could be obtained directly from thioester 15
by modification of the initial hydrolysis step. Thioester 15 (200
mg, 0.43 mmol) was dissolved in a mixture of 0.05 mol dmϪ3
aqueous potassium hydroxide (17 cm3) and methanol (400 cm3)
and the mixture heated to reflux for 2.5 h. Aqueous HCl (1.0
mol dmϪ3, 10 cm3) was added, and after the removal of meth-
anol, the aqueous phase was extracted with ethyl acetate
(10 × 15 cm3). The combined organic extracts were dried
(MgSO4) and the solvent was removed under reduced pressure
to give thioester 15 as a clear oil, which was purified as above to
yield the pure product as a colourless oil (115 mg, 69%).
¯
²
3 H, s, SCOCH3), 2.31 (ttt, 3 H, s, SCOCH3), 2.60–2.72 [ttt,
1 H, m, ¹βCH2(Pro4)], 2.74–3.08 [ttc, 1 H, m, ¹βCH2(Pro4)], 3.36
¯
¯
²
²
[ttc, 1 H, dd, J1 12.9, J2 4.7, ¹δCH2(Pro4)], 3.48–3.70 [3 H, m,
¯
²
δCH2(Pro2) and ttt, ¹δCH2(Pro4)], 3.68 (ttt, 3 H, s, CO2CH3),
¯
²
3.74 (ttc, 3 H, s, CO2CH3), 3.92–4.16 [2 H, m, γCH(Pro4) and
¹δCH2(Pro4)], 4.37–4.57 [3 H, m, αCH(Chp1), αCH(Pro2) and
¯
²
αCH(Pro4)], 4.63 [1 H, quintet, J 6.9, αCH(Ala3)] and 7.15 (1 H,
m, NH); δC(75.44 MHz; C2HCl3) 17.82 [ttc, βCH3(Ala3)], 18.04
[ttt, βCH3(Ala3)], 20.98 [βCH3(Chp1)], 25.06 [ttt, γCH2(Pro2)],
25.34 [ttc, γCH2(Pro2)], 28.44 [βCH2(Pro2)], 30.85 (SCOCH3),
34.86 [ttt, βCH2(Pro4)], 37.56 [ttc, βCH2(Pro4)], 39.08 [ttc,
γCH(Pro4)], 39.87 [ttt, γCH(Pro4)], 46.78, 47.05, 47.40 and
47.60 [αCH(Ala3) and δCH2(Pro2)], 51.30 [ttc, αCH(Chp1)],
51.69 [ttt, αCH(Chp1)], 52.33, 52.46, 52.71 and 53.05
[δCH2(Pro4) and CO2CH3], 58.75 [αCH(Pro4)], 60.12 [ttc,
αCH(Pro2)], 60.74 [ttc, αCH(Pro2)], 169.10 [CO(Chp1)],
170.83, 171.03 (2 × amide CO), 171.99 (ttt, ester CO), 172.88
(ttc, ester CO) and 195.13 (thioester CO); m/z (CI) 464 and
462 (21 and 48%, chlorine isotopes, [M ϩ H]ϩ), 371 {26, [M Ϫ
ClCH(CH3)CO ϩ H]ϩ}, 208 and 206 {56 and 100, chlorine
isotopes, [ClCH(CH3)CONC4H7CO2H]ϩ} and 170 (48, [CH2-
CHCONC4H7CO2H ϩ H]ϩ).
Crystallographic data for macrocycle 11‡
C17H26N3O5.5S, M = 392.47, orthorhombic, space group C2221
(#20), a = 9.371(10), b = 16.744(9), c = 25.705(7) Å, V = 4033(4)
Å3, Z = 8, Dcalc = 1.293 g mϪ3, T = 293 K. 1511 Unique re-
flections were collected on a Rigaku AFC7S, diffracting with
graphite monochromated Mo-Kα radiation (λ = 0.71069 Å) of
which 1020 [I > 3σ(I)] were used for refinement. Convergence at
R(F) = 4.9%, Rw(F) = 4.6% for 240 variable parameters. The
structure was solved and expanded using Fourier techniques
and was refined using TEXSAN.21
N-Methyl-[(1S,3R,9S,12R,15S)-3,12-dimethyl-4,10,13-trioxo-
2-thia-5,11,14-triazatricyclo[12.2.1.05,9]heptadecane]-15-carb-
oxamide 16
Methyl ester 11 (42 mg, 0.11 mmol) was dissolved in a saturated
solution of methylamine in methanol (5 cm3) at 0 ЊC. The flask
was stoppered, allowed to warm to room temperature, and
left standing for 2 days. The flask was cooled to 0 ЊC, and the
stopper removed. The excess methylamine was driven off by
bubbling nitrogen through the solution for 30 min. The solvent
was removed under reduced pressure to afford methylamide 16
as white crystals in quantitative yield, mp 280 ЊC (decomp.)
(Found: C, 53.85; H, 6.8; N, 14.5. C17H26N4O4S requires C,
53.4; H, 6.85; N, 14.65%); [α]D Ϫ46.3 (c 0.6 in MeOH);
νmax(CH2Cl2)/cmϪ1 3348 (NH), 2940 (CH), 1671 (secondary
amide CO), 1621 (tertiary amide CO) and 1532 (NH def.);
Methyl (1S,3R,9S,15S)-3,12-dimethyl-4,10,13-trioxo-2-thia-
5,11,14-triazatricyclo[12.2.1.05,9]heptadecane-15-carboxylate 11
Thioester 15 (0.34 g, 0.74 mmol) was dissolved in a mixture of
0.05 mol dmϪ3 aqueous potassium hydroxide (25 cm3) and
methanol (25 cm3). After stirring for 2.5 h, aqueous HCl (0.5
mmol dmϪ3, 2.5 cm3) was added and the methanol was removed
under reduced pressure. The solution was then further acidified
to pH 1 and extracted with ethyl acetate (6 × 40 cm3). The com-
bined organic extracts were dried (MgSO4) and the solvent was
removed under reduced pressure to afford thiol 10 in an impure
state as a colourless oil (0.22 g).
The above oil was redissolved in DMF (150 cm3) and the
solution was treated with caesium carbonate (0.34 g, 1.04
mmol). The suspension was stirred under nitrogen for 4 days
after which it was filtered through Celite, and the solvent
removed under reduced pressure to give a brown oil. The crude
oil was purified by column chromatography using dichloro-
‡ Full crystallographic details, excluding structure factor tables, have
been deposited at the Cambridge Crystallographic Data Centre
(CCDC). For details of the deposition scheme, see ‘Instructions for
Authors’, J. Chem. Soc., Perkin Trans. 1, available via the RSC Web
page (http://www.rsc.org/authors). Any request to the CCDC for this
material should quote the full literature citation and the reference
number 207/268.
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806
3803

View Article Online
δH(300 MHz; C2HCl3) 1.34 [3 H, d, J 6.6, βCH3(Thp1)], 1.37
[1 H, d, J 7.1, βCH3(Ala3)], 1.88–2.01 [2 H, m, ¹βCH2(Pro4) and
needles, mp 258–259 ЊC (Found: C, 58.95; H, 6.65; N, 13.05.
C26H35N5O5S requires C, 58.95; H, 6.65; N, 13.2%); [α]D Ϫ62.7 (c
0.15 in MeOH); νmax(CH2Cl2)/cmϪ1 3364 (NH), 2937 (CH),
1679 (secondary amide CO), 1642 (tertiary amide CO) and
1531 (NH def.); δH(500 MHz; C2HCl3) 1.14 [3 H, d, J 7.2,
βCH3(Thp1)], 1.31 [3 H, d, J 6.6, βCH3(Ala3)], 1.84–1.96 [2 H,
m, ¹βCH2(Pro2) and ¹βCH2(Pro4)], 2.04–2.11 [1 H, m, ¹γCH2-
¯
²
¹γCH2(Pro2)], 2.02–2.16 [2 H, m, ¹CH2(Pro2) and ¹γCH2(Pro2)],
¯
¯
¯
²
²
²
2.37–2.53 [2 H, m, ¹βCH2(Pro2) and ¹βCH2(Pro4)], 2.82 (3 H,
¯
²
¯
²
d, J 4.9, NHCH3), 3.22 [1 H, d, J 11.0, ¹δCH2(Pro4)], 3.49–3.58
¯
²
[1 H, m, γCH(Pro4)], 3.61–3.72 [3 H, m, δCH2(Pro2) and ¹δCH2-
¯
²
(Pro4)], 3.81 [1 H, q, J 7.1, αCH(Thp1)], 4.38–4.51 [2 H, m,
αCH(Ala3) and αCH(Pro4)], 4.83 [1 H, d, J 7.1, αCH(Pro2)],
7.51 [1 H, d, J 6.9, NH(Ala3)] and 7.64 (1 H, m, NHCH3);
δC(75.44 MHz; C2HCl3) 17.10 and 17.15 [βCH3(Thp1) and
βCH3(Ala3)], 25.55 [βCH2(Pro2) and γCH2(Pro2)], 26.71
(NHCH3), 34.70 [βCH2(Pro4)], 43.08 and 43.15 [αCH(Thp1)
and γCH(Pro4)], 47.49 [δCH2(Pro2)], 49.43 [CH(Ala3)], 51.37
[δCH2(Pro4)], 59.73 [αCH(Pro4)], 61.36 [αCH(Pro2)], 169.66
[CO(Pro2)], 172.19 [CO(Pro4)], 172.69 [CO(Ala3)] and 174.63
[CO(Thp1)]; m/z (CI) 383 (100%, [M ϩ H]ϩ) and 351 (23,
[M Ϫ S ϩ H]ϩ).
¯
²
¯
²
¯
²
(Pro2)], 2.37–2.54 [3 H, m, ¹βCH2(Pro2), ¹βCH2(Pro4) and
¯
¯
²
²
¹γCH2(Pro2)], 2.81 (3 H, d, J 4.7, NHCH3), 3.12 [1 H, dd, J1
¯
²
15.9, J2 10.0, ¹βCH2(Phe5)], 3.24 [1 H, d, J 11.9, ¹δCH2(Pro4)],
¯
²
¯
²
3.48 [1 H, q, J 7.9, αCH(Thp1)], 3.58–3.67 [5 H, m, ¹βCH2-
¯
²
(Phe5), γCH(Pro4), δCH2(Pro2) and ¹δCH2(Pro4)], 4.45–4.49
¯
²
[2 H, m, αCH(Ala3) and αCH(Pro4)], 4.84 [1 H, dd, J1 1.5,
αCH(Pro2)], 5.02 [1 H, m, αCH(Phe5)], 6.90 (1 H, q, J 4.3,
NHCH3), 7.15 (1 H, t, J 7.1, Ar-H para), 7.23–7.28 (4 H,
m, Ar-H ortho and meta), 7.60 [1 H, d, J 7.2, NH(Ala3)] and
7.86 [1 H, d, J 8.0, NH(Phe5)]; δC(75.44 MHz; C2HCl3) 17.01
[βCH3(Thp1)], 17.46 [βCH3(Ala3)], 25.52 [βCH2(Pro2)], 25.71
[γCH2(Pro2)], 26.98 (NHCH3), 34.48 [βCH2(Pro4)], 36.59
[βCH2(Phe5)], 43.34 [γCH(Pro4)], 43.61 [αCH(Thp1)], 47.49
[δCH2(Pro2)], 48.85 [αCH(Ala3)], 51.19 [δCH2(Pro4)], 53.60
[αCH(Phe5)], 59.98 [αCH(Pro2)], 62.36 [αCH(Pro4)], 126.59
(Ar-CH para), 128.70 (Ar-CH meta), 129.11 (Ar-CH ortho),
138.78 (Ar-C quaternary), 169.65 [CO(Pro2)], 172.07
[CO(Pro4)], 172.76 [CO(Phe5)], 173.00 [CO(Ala3)] and 174.47
[CO(Thp1)]; m/z (CI) 530 (22%, [M ϩ H]ϩ), 243 {100, [NC4-
H6CONHCH(CH2Ph)CO]ϩ} and 201 {37, [SCH(CH3)CONC4-
H7CONH ϩ H]ϩ}.
(1S,3R,9S,12R,15S)-3,12-Dimethyl-4,10,13-trioxo-2-thia-
5,11,14-triazatricyclo[12.2.1.05,9]heptadecane-15-carboxylic acid
17
A solution of methyl ester 11 (0.30 g, 0.79 mmol) in methanol
(3 cm3) was treated with aqueous sodium hydroxide (1.0 mol
dmϪ3, 3.5 cm3), and the mixture allowed to stir for 2 h. After
addition of aqueous HCl (1.0 mol dmϪ3, 5 cm3), the methanol
was removed under reduced pressure. The solution was acid-
ified with aqueous HCl, saturated with NaCl, and exhaustively
extracted with ethyl acetate (9 × 20 cm3). The combined organic
extracts were dried (MgSO4), and the solvent was removed
under reduced pressure to afford the macrocycle acid 17 as
colourless crystals (0.24 g, 83%), mp 225–226 ЊC (decomp.)
(Found: C, 52.15; H, 6.5; N, 11.0. C16H23N3O5S requires C,
52.0; H, 6.3; N, 11.35%); [α]D Ϫ134.4 (c 0.2 in MeOH);
νmax(CH2Cl2)/cmϪ1 3317 (NH), 1751 (ester CO), 1686 (second-
ary amide CO) and 1672 (tertiary amide CO); δH(300 MHz;
C2HCl3) 1.32 [6 H, m, βCH3(Thp1) and βCH3(Ala3)],
1.90–2.10 [3 H, m, ¹βCH2(Pro2), ¹βCH2(Pro4) and ¹γCH2-
(2S)-N2-[(2S)-N-(tert-Butoxycarbonyl)alanyl]phenylalanine
methylamide 21
A stirred solution of (2S)-N-(tert-butoxycarbonyl)alanine (0.64
g, 3.4 mmol) and N-methylmorpholine (0.38 cm3, 3.4 mmol) in
dry THF (15 cm3) at Ϫ15 ЊC was treated dropwise with isobutyl
chloroformate (0.46 cm3, 3.4 mmol), followed after 2 min by a
solution of (2S)-phenylalanine methylamide (0.60 g, 3.4 mmol)
and N-methylmorpholine (0.38 cm3, 3.4 mmol) in dry THF (5
cm3).
¯
¯
¯
²
²
²
(Pro2)], 2.38–3.43 [3 H, m, ¹βCH2(Pro2), ¹βCH2(Pro4) and
¯
²
¯
²
¹γCH2(Pro2)], 3.11 [1 H, d, J 11.8, ¹δCH2(Pro4)], 3.50–3.66
The suspension was diluted with additional THF to allow
efficient stirring, allowed to reach room temperature, then
stirred for 2 h. The precipitated hydrochloride salts were filtered
off and the solvents were removed under reduced pressure to
yield a solid which was dissolved in ethyl acetate (50 cm3). This
solution was washed with aqueous HCl (0.5 mol dmϪ3, 2 × 30
cm3), aqueous sodium hydrogen carbonate (5%, 2 × 30 cm3),
and brine (30 cm3). The solvent was removed under reduced
pressure to yield the dipeptide methylamide 21 as a white solid
(0.78 g, 67%). A small portion of this was recrystallised from
ethyl acetate to afford colourless needles, mp 157–158 ЊC
(lit.,20 157–158 ЊC); [α]D Ϫ38.7 (c 1.1 in MeOH) [lit.,20 Ϫ37.7 (c
1.0 in MeOH]; νmax(CH2Cl2)/cmϪ1 3411 (NH), 2981 (CH),
1701 (urethane CO), 1670 (amide CO) and 1505 (NH
def.); δH(200 MHz; C2HCl3) 1.27 [3 H, d, J 7.0, βCH3(Ala)],
1.37 [9 H, s, C(CH3)3], 2.70 (3 H, d, J 4.8, NHCH3), 3.01–3.25
[2 H, m, βCH2(Phe)], 4.05 [1 H, m, CH(Ala)], 4.66 [1 H, m,
αCH(Phe)], 5.10 (1 H, m, NH), 6.74 (1 H, m, NH) and
7.14–7.32 (5 H, m, Ar-H); δC(50.31 MHz; C2HCl3) 18.57
[βCH(Ala)], 26.82 (NHCH3), 28.69 [C(CH3)3], 38.36
[βCH2(Phe)], 51.56 [αCH(Ala)], 54.42 [αCH(Phe)], 80.93
[C(CH3)3], 127.39 (Ar-CH para), 129.04 (Ar-CH meta), 129.70
(Ar-CH ortho), 137.11 (Ar-C quaternary), 156.10 (urethane
CO), 171.70 and 173.15 (2 × amide CO); m/z (CI) 351 (100%,
[M ϩ H]ϩ), 294 (93, [M Ϫ C4H8]ϩ) and 250 (12, [M Ϫ
CO2C4H8]ϩ).
¯
²
¯
²
[3 H, m, γCH(Pro4) and δCH2(Pro2)], 3.83–3.91 [2 H, m,
αCH(Thp1) and ¹δCH2(Pro4)], 4.46–4.50 [1 H, m, αCH(Ala3)],
¯
²
4.54–4.58 [1 H, m, αCH(Pro4)], 4.81 [1 H, d, J 7.7, αCH(Pro2)],
7.63 (1 H, d, J 6.6, NH) and 9.77 (1 H, br s, CO2H); δC(75.44
MHz; C2HCl3) 16.94 and 17.00 [βCH3(Thp1) and βCH3(Ala3)],
25.37 and 25.73 [βCH2(Pro2) and γCH2(Pro2)], 34.25 [βCH2-
(Pro4)], 42.62 and 42.75 [αCH(Thp1) and γCH(Pro4)], 47.48
[δCH2(Pro2)], 49.47 [αCH(Ala3)], 51.54 [δCH2(Pro4)], 59.82 and
60.13 [αCH(Pro2) and αCH(Pro4)], 169.98 [CO(Pro2)], 172.56,
173.02 and 174.63 [CO(Thp1), CO(Ala3) and CO(Pro4)]; m/z
(CI) 370 (77%, [M ϩ H]ϩ), 326 (14, [M Ϫ CO2 ϩ H]ϩ), 257 (59,
[M Ϫ NC4H7CO2 ϩ H]ϩ) and 170 (100, SCHCONC4H7COϩ).
(2S)-N2-{(1S,3R,9S,12R,15S)-3,12-Dimethyl-4,10,13-trioxo-2-
thia-5,11,14-triazatricyclo[12.2.1.05,9]heptadecan-15-
ylcarbonyl}phenylalanine methylamide 18
A solution of acid 17 (54 mg, 0.15 mmol) and N,N-diisopropyl-
ethylamine (80 cm3, 0.46 mmol) in dry dichloromethane (2 cm3)
was treated with BOP-Cl (40 mg, 0.16 mmol) and the resulting
suspension stirred under nitrogen at 0 ЊC for 1 h. A solution of
(2S)-phenylalanine methylamide (30 mg, 0.17 mmol) in dry
dichloromethane (1 cm3) was then added. The mixture was
stirred at 0–5 ЊC for 2 h, then allowed to warm to room tem-
perature and stirred for a further 20 h. The solution was diluted
to ~15 cm3 with dichloromethane, washed with aqueous HCl
(0.5 mol dmϪ3, 2 × 5 cm3), aqueous sodium hydrogen carbonate
(5%, 2 × 5 cm3) and brine (5 cm3) and then dried (MgSO4). The
solvent was removed under reduced pressure to afford the
peptide 18 as a white solid (48 mg, 62%). A small portion of
this was recrystallised from ethyl acetate to afford colourless
(2S)-N2-[(2S)-N-{(1S,3R,9S,12R,15S)-3,12-Dimethyl-4,10,13-
trioxo-2-thia-5,11,14-triazatricyclo[12.2.1.05.9]heptadecan-15-
ylcarbonyl}alanyl]phenylalanine methylamide 19
A solution of methylamide 21 (0.50 g, 1.4 mmol) in dichloro-
3804
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806

View Article Online
methane (15 cm3) at 0 ЊC was treated with trifluoroacetic acid
(8 cm3) and the solution stirred for 2 h. The solvents were
removed under reduced pressure to yield the trifluoroacetate
salt 20 as a white solid.
1.23 [3 H, d, J 6.9, βCH3(Thp1)], 1.29 [3 H, d, J 7.1, βCH3-
(Ala3)], 1.81–2.04 [3 H, m, ¹βCH2(Pro2), ¹βCH2(Pro4) and
¯
²
¯
²
¹γCH2(Pro2)], 2.18 [6 H, s, N(CH3)2], 2.31–2.43 [4 H, m,
¯
²
CH2N(CH3)2, ¹βCH2(Pro2) and ¹γCH2(Pro2)], 2.51–2.60 [1 H, m,
¯
²
¯
²
A solution of acid 17 (80 mg, 0.22 mmol) and N,N-diiso-
propylethylamine (0.12 cm3, 0.69 mmol) in dichloromethane (5
cm3) was treated with BOP-Cl (60 mg, 0.24 mmol) at 0 ЊC. After
50 min, amine trifluoroacetate 20 (90 mg, 0.25 mmol) was
added and the mixture was stirred at 0–5 ЊC for 2 h, then at
room temperature for 18 h. The solution was diluted to ~15 cm3
with dichloromethane, then washed with aqueous HCl (0.5 mol
dmϪ3, 2 × 6 cm3), aqueous sodium hydrogen carbonate (5%,
2 × 6 cm3) and brine (6 cm3). The organic phase was dried
(MgSO4), and the solvent removed under reduced pressure. The
crude residue was purified by column chromatography using
dichloromethane–methanol (20:1) as the eluent to afford the
peptide 19 as a white solid (43 mg, 33%), mp 129–131 ЊC
(HRMS: Found (M ϩ H)ϩ, 601.2798. C29H41N6O6S requires
601.2808); [α]D Ϫ25.7 (c 0.3 in MeOH); νmax(CH2Cl2)/cmϪ1 3338
(NH), 2935 (CH), 1680 (secondary amide CO), 1662 (tertiary
amide CO) and 1536 (NH def.); δH(300 MHz; C2HCl3) 1.16
[3 H, d, J 7.4, βCH3(Ala5)], 1.27 [3 H, d, J 6.6, βCH3(Ala3)],
1.30 [3 H, d, J 7.4, βCH3(Thp1)], 1.89–1.98 [2 H, m, ¹βCH2-
¹βCH2(Pro4)], 3.15–3.22 [2 H, m, γCH(Pro4) and ¹δCH2(Pro4)],
¯
²
¯
²
3.35–3.62 [5 H, m, δCH2(Pro2), ¹δCH2(Pro4) and CH2HNCO],
¯
²
3.71 [1 H, q, J 7.1, αCH(Thp1)], 4.37–4.43 [2 H, m, αCH(Ala3)
and αCH(Pro4)], 4.76 [1 H, d, J 8.2, αCH(Pro2)], 7.70 (1 H, t,
J 7.7, NHCH2) and 7.90 [1 H, d, J 8.0, NH(Ala3)]; δC(75.44
MHz; C2HCl3) 16.93 and 17.10 [βCH3(Thp1) and βCH3(Ala3)],
25.47 [βCH2(Pro2) and γCH2(Pro2)], 34.52 [βCH2(Pro4)], 37.63
(CH2NHCO), 43.02 and 43.10 [αCH(Thp1) and γCH(Pro4)],
45.49 [N(CH3)2], 46.97 [δCH2(Pro2)], 48.79 [αCH(Ala3)], 51.16
[δCH2(Pro4)], 57.96 [CH2N(CH3)2], 59.59 [αCH(Pro2)], 61.41
[αCH(Pro4)], 169.53 [CO(Pro2)], 171.92 and 172.30 [CO(Ala3)
and CO(Pro4)] and 173.87 [CO(Thp1)]; m/z (CI) 440 (100%,
[M ϩ H]ϩ), 408 (23, [M Ϫ 2CH3 ϩ H]ϩ), 369 {19, [M Ϫ CH-
CH2N(CH3)2 ϩ H]ϩ} and 241 {82, [CH3CHCONC4H7CONH-
CH2CH2N(CH3)2 ϩ H]ϩ}.
A solution of amine 22 in dichloromethane was treated with
a slight excess of trifluoroacetic acid, followed by removal
of the solvent under reduced pressure to afford the amine
trifluoroacetate 23 as a colourless oil.
¯
²
(Pro2) and ¹βCH2(Pro4)], 2.02–2.09 [1 H, m, ¹γCH2(Pro2)], 2.33–
δH(500 MHz; C2HCl3) 1.36 [6 H, d, J 7.1, βCH3(Thp1) and
βCH3(Ala3)], 2.01–2.08 [2 H, m, ¹βCH2(Pro4) and ¹γCH2(Pro2)],
¯
²
¯
²
2.40 [1 H, m, ¹γCH2(Pro2)], 2.43–2.55 [2 H, m, ¹βCH2(Pro2) and
¯
²
¯
²
¯
¯
²
²
¹βCH2(Pro4)], 2.74 (3 H, d, J 4.7, NHCH3), 2.86 [1 H, dd,
2.11–2.18 (1 H, m, ¹βCH2(Pro2)], 2.28–2.33 [1 H, m, ¹βCH2-
¯
¯
²
¯
²
²
J1 14.4, J2 11.7, ¹βCH2(Phe6)], 3.22 [1 H, d, J 11.8, ¹δCH2(Pro4)],
(Pro2)], 2.39–2.45 [1 H, m, ¹γCH2(Pro2)], 2.45–2.51 [1 H, m,
¯
²
¯
¯
²
¯
²
²
3.44–3.69 [4 H, m, ¹βCH2(Phe6), δCH2(Pro2) and ¹δCH2(Pro4)],
¹βCH2(Pro4)], 2.98 (3 H, s, NCH3), 3.02 (3 H, s, NCH3), 3.25–
¯
²
¯
²
3.74–3.85 [2 H, m, αCH(Thp1) and γCH(Pro4)], 4.15–4.19 [1 H,
m, αCH(Ala5)], 4.37 [1 H, d, J 10.4, αCH(Pro4)], 4.41–4.46
[1 H, m, αCH(Ala3)], 4.60–4.68 [1 H, m, αCH(Phe6)], 4.75 [1 H,
d, J 6.6, αCH(Pro2)], 6.87 (1 H, q, J 4.7, NHCH3), 7.06 [1 H, d,
J 9.1, NH(Phe6)], 7.11–7.22 (5 H, m, Ar-H), 7.58 [1 H, d, J 7.1,
NH(Ala3)] and 8.33 [1 H, d, J 3.9, NH(Ala5)]; δC(125.7 MHz;
C2HCl3) 16.31 [βCH3(Ala5)], 16.79 [βCH3(Thp1)], 17.03
[βCH3(Ala3)], 25.16 [βCH2(Pros)], 25.34 [γCH2(Pro2)], 26.47
(NHCH3), 34.39 [βCH2(Pro4)], 37.31 [βCH2(Phe6)], 43.30
[γCH(Pro4)], 43.57 [αCH(Thp1)], 47.31 [δCH2(Pro2)], 48.24
[αCH(Ala3)], 50.74 [δCH2(Pro4)], 51.01 [αCH(Ala5)], 54.24
[αCH(Phe6)], 59.88 [αCH(Pro2)], 62.11 [αCH(Pro4)], 126.32
(Ar-CH para), 128.22 (Ar-CH meta), 129.14 (Ar-CH ortho),
138.68 (Ar-C quaternary), 169.20 [CO(Pro2)], 171.90, 172.18,
172.48 and 172.60 [CO(Ala3), CO(Pro4), CO(Ala5) and
CO(Phe6)] and 174.24 [CO(Thp1)]; m/z (CI) (100%, [M ϩ H]ϩ),
569 (31, [M Ϫ NH2CH3]ϩ), 439 (10, [M Ϫ CH3NHCOCH-
CH2Ph ϩ H]ϩ) and 372 {100, [CONC4H6CONHCH(CH3)-
CONHCH(CH2Ph)CONHCH3]ϩ}.
3.35 [1 H, m, ¹CH2N(CH3)2], 3.45–3.57 [4 H, m, ¹CH2NHCO,
¯
²
¯
²
¹CH2N(CH3)2, ¹δCH2(Pro2) and ¹δCH2(Pro4)], 3.68 [1 H, dd,
¯
²
¯
¯
²
¯
²
²
J1 12.5, J2 4.5, ¹δCH2(Pro4)], 3.82 [1 H, q, J 7.3, αCH(Thp1)],
3.89–3.98 [3 H, m, ¹CH2NHCO, γCH(Pro4) and ¹δCH2(Pro2)],
¯
²
¯
²
4.23–4.33 [1 H, m, αCH(Ala3)], 4.46 [1 H, d, J 10.5, αCH(Pro4)],
4.55–4.62 [1 H, m, αCH(Pro2)], 8.24 [1 H, d, J 6.8, NH(Ala3)],
8.79 (1 H, br s, NHCH2) and 9.16 [1 H, br s, NH(CH3)2];
δC(125.7 MHz; C2HCl3) 16.10 and 17.07 [βCH3(Thp1) and
βCH3(Ala3)], 25.93 [γCH2(Pro2)], 26.84 [βCH2(Pro2)], 34.39
[βCH2(Pro4)], 35.00 (CH2NHCO), 43.16 [γCH(Pro4)], 43.82
[αCH(Thp1)], 44.02 [NCH3], 45.38 [NCH3], 47.24 [δCH2-
(Pro2)], 49.00 [αCH(Ala3)], 51.81 [δCH2(Pro4)], 57.92 [CH2-
N(CH3)2], 62.37 [αCH(Pro2) and αCH(Pro4)], 116.12 (q,
JCF 289, CF3CO2), 160.88 (q, JCF 38.5, CF3CO2), 173.43, 174.06
and 174.32 (4 × amide CO).
Acknowledgements
We thank the Wellcome Trust for grant 040331, Dr M. Akhtar
for scientific support, Iain Stirling for technical support,
Dr P. Lightfoot for providing the X-ray crystal structure
of compound 11, the Wellcome Trust for a studentship to
A. L. and the BBSRC for grants B03817, B05162 and B06902 to
D. G.
N-{(1S,3R,9S,12R,15S)-3,12-Dimethyl-4,10,13-trioxo-2-thia-
5,11,14-triazatricyclo[12.2.1.05,9]heptadecan-15-ylcarbonyl}-
NЈ,NЈ-(dimethyl)ethylenediamine 22 and amine trifluoroacetate
salt 23
A stirred solution of acid 17 (0.15 g, 0.41 mmol) and N-methyl-
morpholine (0.10 cm3, 0.91 mmol) in dry THF (3 cm3) and
DMF (1 cm3) under nitrogen at Ϫ15 ЊC was treated dropwise
with isobutyl chloroformate (55 µl, 0.42 mmol), followed after
10 min by a solution of N,N-dimethylethylenediamine (60 µl,
0.55 mmol) in dry THF (1 cm3). The solution was allowed to
reach room temperature and stirred for 4 h. The solvents were
removed under reduced pressure, and the residual oil redis-
solved in ethyl acetate (20 cm3). The solution was washed with
aqueous sodium hydrogen carbonate (5%, 2 × 5 cm3), brine (5
cm3) and then dried (MgSO4). The solvent was removed under
reduced pressure to afford the amine 22 as colourless crystals
(0.14 g, 78%), mp 174–176 ЊC (Found: C, 53.95; H, 7.6; N,
15.25. C20H33N5O4Sؒ¹H2O requires C, 53.55; H, 7.65; N, 15.6%);
References
1 D. S. Kemp, T. J. Allen and L. O. Sherri, J. Am. Chem. Soc., 1995,
117, 6641.
2 S. F. Betz, J. W. Bryson and W. F. DeGrado, Curr. Opin. Struct.
Biol., 1995, 5, 457.
3 J. S. Nowick, E. M. Smith and M. Pairish, Chem. Soc. Rev., 1996, 25,
401.
4 A. Lewis, M. D. Ryan and D. Gani, J. Chem. Soc., Perkin Trans. 1,
1998, 3767.
5 A. Lewis, J. Wilkie, T. J. Rutherford and D. Gani, J. Chem. Soc.,
Perkin Trans. 1, 1998, 3777.
6 D. S. Kemp and J. H. Rothman, Tetrahedron Lett., 1995, 36, 3809.
7 D. S. Kemp and J. H. Rothman, Tetrahedron Lett., 1995, 36, 3813.
8 A. Pardi, M. Billeter and K. Wüthrich, J. Mol. Biol., 1984, 180, 741.
9 D. S. Kemp, T. P. Curran, J. G. Boyd and T. J. Allen, J. Org. Chem.,
1991, 56, 6683.
¯
²
[α]D Ϫ48.0 (c 0.85 in MeOH); νmax(CH2Cl2)/cmϪ1 3337 (NH),
2983 (CH), 1668 (secondary amide CO), 1634 (tertiary amide
CO), 1531 (NH def.) and 1269 (C–N); δH(300 MHz; C2HCl3)
10 A. Chakrabartty, T. Kortemme and R. L. Baldwin, Prot. Sci., 1994,
3, 843.
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806
3805

View Article Online
11 M. Blaber, X. Zhang and B. W. Matthews, Science, 1993, 260, 1637.
12 D. Sali, M. Bycroft and A. R. Fersht, Nature, 1988, 335, 740.
13 S. Dasgupta and J. A. Bell, Int. J. Pept. Protein Res., 1993, 41, 499.
14 K. R. Shoemaker, P. S. Kim, E. J. York, J. M. Stewart and
R. L. Baldwin, Nature, 1987, 326, 563.
15 E., A. Gallo and S. H. Gellman, J. Am. Chem. Soc., 1994, 116,
11560.
16 J. M. Lehn, F. G. Riddell, B. J. Price and I. O. Sutherland, J. Chem.
Soc. B, 1967, 387.
17 T. L. Hwang and A. J. Shaka, J. Magn. Reson., A, 1995, 112, 275.
18 W. C. Still, M. Kahn and A. Mitra, J. Org. Chem., 1978, 43, 2923.
19 D. D. Perrin, W. L. F. Armarego and D. R. Perrin, Purification of
Laboratory Chemicals, Pergamon, Oxford, 1980.
20 P. Kaur, K. Uma, P. Balaram and V. S. Chauhan, Int. J. Pept. Protein
Res., 1989, 33, 103.
21 TEXSAN Single Crystal Structure Analysis package, 1992,
Molecular Structure Corporation, MSC, 3200 Research Forest
Drive, The Woodlands, TX 77381, USA.
Paper 8/06032K
3806
J. Chem. Soc., Perkin Trans. 1, 1998, 3795–3806