Novel enantioselective receptors for N-protected glutamate and aspartate

Article abstract of DOI:10.1002/chem.200500444

A series of chiral bisthiourea macrocycles 1-4 have been prepared and their binding properties with various dicarboxylate salts have been examined by using NMR titration and isothermal calorimetry experiments. Macrocycle 1, in particular, favours the 1:1

Full text of DOI:10.1002/chem.200500444

DOI: 10.1002/chem.200500444
Novel Enantioselective Receptors for N-ProtectedGlutamate andAspartate
Andrea Ragusa,[a] Sara Rossi,[a] Joseph M. Hayes,[b] Matthias Stein,[b] and
Jeremy D. Kilburn*[a]
Abstract: A series of chiral bisthiourea
macrocycles 1–4 have been prepared
and their binding properties with vari-
ous dicarboxylate salts have been ex-
amined by using NMR titration and
isothermal calorimetry experiments.
Macrocycle 1, in particular, favours the
1:1 binding of N-protected l-glutamate
and aspartate, but favours 1:2 binding
of the corresponding d-amino acids in
polar solvents (dimethyl sulfoxide and
acetonitrile). The macrocycles, howev-
er, do not bind carboxylates at all in
the less competitive solvent chloro-
form. The binding properties of these
macrocyles are sensitive to small struc-
tural changes as demonstrated by the
altered binding properties of macrocy-
cles 2–4 compared with 1.
Keywords:
enantioselectivity
·
receptors · solvent effects
Introduction
additional amide hydrogens to
act as hydrogen-bonding donors
for the lone pairs of the guest
carboxylate oxygens (Figure 1).
Pyrido units were expected to
help preorganise the macrocy-
cle,[6] and chirality was provided
by the (S,S)-1,2-diphenylethy-
lene diamine spacer.
Molecular recognition lies at the heart of biochemistry. Bio-
chemical processes all inherently require selective molecular
recognition and complexation to be successful. Matching the
degree of exquisite control and efficiency exhibited by natu-
ral systems is a challenging goal for the host-guest chemist
to emulate, and has led to much interest in the design of a
diverse range of synthetic receptors for important biological
substrates.[1] Enantioselective recognition of biologically rel-
evant molecules, however, remains a major challenge.[2] Al-
though numerous receptors have been developed for dicar-
boxylic acids and dicarboxylates,[3] only a fewenantioselec-
tive receptors for chiral dicarboxylates have been de-
scribed.[4] In our efforts to develop novel receptors for dicar-
boxylates, we have targeted N-protected amino acid
dicarboxylates (e.g. glutamate and aspartate). We recently
described the chiral bisthiourea macrocycle 1 which was spe-
cifically designed to bind to glutamate through eight hydro-
gen bonds.[5] It incorporates thiourea moieties intended to
provide the primary interaction with the carboxylates and
Macrocycle
1
formed
a
strong 1:1 complex with N-Boc-
l-glutamate dicarboxylate in
polar solvents (DMSO and
Figure 1. Schematic of pro-
posed mode of complexation
of N-Boc-l-glutamate by mac-
rocycle 1 via eight hydrogen
bonds.
CH3CN). However, only
a
weak 1:1 complex was formed
with N-Boc-d-glutamate dicar-
boxylate, and a 1:2 (host/guest)
complex was favoured instead. Surprisingly the macrocycle
did not bind carboxylates at all in the less competitive sol-
1
vent CDCl3. Indeed H NMR studies indicated that the mac-
rocycle adopted a wrapped conformation with two-fold C2
symmetry in CDCl3, which was unsuitable for binding car-
boxylates.
In order to gain a better understanding of the unusual
host–guest complexation properties of 1 we have now car-
ried out further binding studies with macrocycle 1, and with
[a] A. Ragusa, S. Rossi, Prof. J. D. Kilburn
School of Chemistry, University of Southampton
Southampton, SO17 1BJ (UK)
Fax : (+44)2380-596-805
E-mail: jdk1@soton.ac.uk
1
analogous macrocycles 2–4, using H NMR and isothermal
[b] Dr. J. M. Hayes, Dr. M. Stein
calorimetric titrations, as well as computational studies on
macrocycle 1. Here we describe in detail the results of these
studies.
Anterio consult and research GmbH
Augustaanlage 26, 68165 Mannheim (Germany)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
5674
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5674 – 5688

FULL PAPER
Results andDiscussion
amines 14 and 18, respectively. Removal of the Boc protect-
ing groups and treatment of the resulting diamines with thio-
phosgene produced the bis(isothiocyanates) 16 and 20.
Again, a further equivalent of diamine was added to the cor-
responding bis(isothiocyanate), in the presence of triethyl-
amine, by syringe pump, over three hours to afford the mac-
rocycles 3 and 4 in 17 and 29% yield, respectively.
Synthesis: Macrocycle 1 was synthesised as previously de-
scribed by first coupling (1S,2S)-1,2-diphenylethylene dia-
mine with pyridyl acid 5 to give bisphthalimide 6, which was
deprotected by using hydrazine hydrate. Treatment of the
resulting diamine 7 with CS2 and DCC produced the bis(iso-
thiocyanate) 8. A further equivalent of diamine 7 was added
to bis(isothiocyanate) 8, in the presence of DMAP, by sy-
ringe pump, over three hours to afford macrocyclic bis-
thiourea 1 in 26% yield (Scheme 1).
An analogous, but more conformationally constrained,
macrocycle 2 was readily prepared by an identical sequence,
but starting with (S,S)-1,2-diaminocyclohexane. Pyridyl acid
5 was coupled with the diaminocyclohexane tartrate salt by
using a two-phase (water/dichloromethane) solvent system,
as attempted isolation of the free diamine from the tartrate
salt was frustrated by low yields. The bisphthalimide 9 was
converted to the bis(isothiocyanate) 11 and reaction with a
further equivalent of diamine 10 gave the desired macrocy-
cle 2 in 72% yield. The high yield (relative to the formation
of macrocycle 1) is probably due to the conformational con-
straint of the diamine, which favours the macrocyclisation
over the formation of by-products.
To probe the role of the pyrido unit in preorganising mac-
rocycles 1 and 2, the analogous macrocycles 3 and 4, with
the pyrido units replaced by benzo units, were also pre-
pared. Commercially available 3-cyanobenzoic acid 12 was
coupled with (1S,2S)-(ꢀ)-1,2-diphenylethylenediamine or
(1S,2S)-(ꢀ)-1,2-diaminocyclohexane to afford bis(cyanoben-
zamide) 13 and 17, respectively. The cyano groups were re-
duced using sodium borohydride/NiCl2 and, to prevent di-
merisation by-products, the diamine was trapped in situ
using di-tert-butyl dicarbonate,[7] to provide the protected di-
Conformational studies: As described previously macrocycle
1
1 in [D6]DMSO or in CD3CN gave a simple H NMR spec-
trum consistent with the expected four-fold D2 symmetry of
1
the macrocycle.[5] At room temperature the H NMR spec-
trum of 1 in CDCl3 shows very broad signals, but gives a
well-resolved spectrum at ꢀ408C (Figure 2) indicating two-
fold C2 symmetry, with, for example, two different signals
for the thiourea NH protons (d=9.0 and 8.1 ppm) and for
the amide NH protons (d=10.1 and 7.9 ppm) (Table 1).
1
Further H NMR studies, by using torsion angle dynamics
with NOE and scalar coupling constant constraints,[8] al-
lowed us to determine the conformation of macrocycle 1 in
CDCl3 at ꢀ408C.
The conformation determined by using this approach
(Figure 3) shows that the macrocycle adopts a twisted and
wrapped conformation stabilised by two sets of three intra-
molecular hydrogen bonds between one of the amide car-
bonyls and three NHꢁs from a thiourea and the other amide.
This hydrogen-bonding motif has been observed by us previ-
ously in the dimeric crystal structure of the related acyclic
monothiurea 21 (Figure 4).[6b] Further stabilisation of the
wrapped conformation of 1 is achieved by weak hydrogen
bonds between the pyridine nitrogen and adjacent NH
groups.
The more conformationally constrained macrocycle 2 was
only sparingly soluble in CD3CN, but in [D6]DMSO also
Scheme 1. a) SOCl2; (1S,2S)-(ꢀ)-1,2- diphenylethylenediamine, DMAP, CH2Cl2 (68% for 6, 48% for 13); b) diphenylchlorophosphate; (1S,2S)-(ꢀ)-1,2-di-
aminocyclohexane (tartrate salt), H2O, K2CO3 (52% for 9, 67% for 17); c) N2H4·H2O, EtOH (80–85%); d) CS2, DCC, CH2Cl2; e) simultaneous addition
of 1 equiv of 7 (or 10 or 15 or 19) and 1 equiv of 8 (or 11 or 16 or 20, respectively) by syringe pumps over 3 h to a solution of DMAP or Et3N in CH2Cl2
(26% for 1, 72% for 2, 17% for 3, 29% for 4, over steps d and e or h and e); f) di-tert-butyl dicarbonate, NiCl2, NaBH4 (64% for 14 and 18); g) TFA,
CH2Cl2 (quant.); h) thiophosgene, K2CO3, CH2Cl2, H2O. DMAP=4-dimethylaminopyridine, DCC=N,N’-dicyclohexylcarbodiimide, TFA=trifluoroacetic
acid.
Chem. Eur. J. 2005, 11, 5674 – 5688
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5675

J. D. Kilburn et al.
fluoroethanol, but not in
CH3CN or in CHCl3. Again in
[D6]DMSO
3 and 4 gave a
simple 1H NMR spectrum con-
sistent with the expected D2
symmetry of the macrocycle.
Comparing the 1H NMR spec-
tra of 3 and 4 with the spectra
of 1 and 2 in [D6]DMSO, re-
veals that in the former pair the
signals for the NH protons
(amide and thiourea) are in
each case more upfield than the
corresponding signals for the
latter pair of macrocycles
(Table 1). This is consistent
with the notion that the pyrido
nitrogens are forming weak hy-
drogen bonds with the adjacent
NHꢁs and therefore help to pre-
organise the macrocycle in
DMSO—even though in this
solvent the macrocycles do not
form the wrapped conformation
identified in CDCl3.
Figure 2. 1H NMR spectra of macrocycle 1 in: a) [D6]DMSO at room temperature; b) CDCl3 at room tempera-
ture; c) CDCl3 at ꢀ408C.
Binding studies: Binding studies
were carried out using H NMR
1
Table 1. 1H NMR chemical shifts for NH signals of macrocycles 1–4 in
CDCl3 and [D6]DMSO.
Solvent
CDCl3
Macrocycle
NH1
NH2
NH3
NH4
1[a]
2
9.0
8.7
8.1
8.1
7.9
7.7
10.1
9.2
[D6]DMSO
1
2
3
4
8.3
8.6
7.9
7.9
9.4
9.0
9.0
8.2
[a] Spectrum recorded at ꢀ408C. All other spectra recorded at room
temperature.
1
gave a simple H NMR spectrum consistent with the expect-
ed four-fold D2 symmetry of the macrocycle. In CDCl3
2
1
gave a H NMR spectrum at room temperature similar to
that found for 1 and consistent with the wrapped conforma-
tion identified for 1 at ꢀ408C. Detailed 2D NMR studies
1
(COSY and NOESY H NMR experiments) allowed full as-
1
Figure 3. Conformation of macrocycle 1 in CDCl3 as determined by
signment of all the H NMR signals for 2. In particular, the
1H NMR showing intramolecular H-bond lengths.
thiourea NH protons have two different signals at d=8.7
and 8.1 ppm, while the amide NH protons have signals at
d=9.2 and 7.7 ppm (Table 1). However, in the case of 2 the
spectrum in CDCl3 was well-resolved even at room tempera-
ture, indicating that the greater conformational constraints
imposed by the cyclohexyl diamine units encourages the
wrapped conformation.
titrations and isothermal calo-
rimetry (Table 2). Data from
the 1H NMR titrations were
fitted to 1:1 or 1:2 (host/guest)
binding isotherms by using the
Macrocycles 3 and 4 in which the pyridyl units in 1 and 2
are replaced by benzo units, were soluble in DMSO and tri-
NMRTit HG software.[9] Where
Figure 4. Acyclic thiourea 21.
5676
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5674 – 5688

Enantioselective Receptors
FULL PAPER
Table 2. Binding data for macrocycles (MC) 1–3 with various carboxylate salts in CH3CN and DMSO.
[a]
NMR titration data
ITC titration data[a]
Entry MC Solvent[b] Guest[c]
Ka1:1
Ka1:2
DG1:1
Ka1:1
DH 1:1
TDS1:1
DG1:2
Ka1:2
DH 1:2
TDS1:2
[mꢀ1
]
[mꢀ1
]
[kJmolꢀ1
]
[mꢀ1
]
[kJmolꢀ1
]
[kJmolꢀ1
]
[kJmolꢀ1
]
[mꢀ1
]
[kJmolꢀ1
]
[kJmolꢀ1
]
1
2
3
4
5
6
7
8
1
1
1
1
1
1
1
1
1
1
1
1
CH3CN
CH3CN
DMSO
DMSO
CH3CN
CH3CN
CH3CN
DMSO
DMSO
CH3CN
CH3CN
DMSO
N-Boc-l-Glu
N-Boc-l-Glu
N-Boc-l-Glu
N-Boc-d-Glu
acetate
N-Ac-l-Glu
N-Ac-d-Glu
N-Ac-l-Glu
N-Ac-d-Glu
N-Boc-l-Asp
N-Boc-d-Asp
N-Boc-l-Asp
>104
–
3700
–
ꢀ25.3
–
ꢀ19.2
2.7”104
<100
2300
ꢀ4.5
–
ꢀ10.7
20.8
–
8.5
–
ꢀ26.8
–
<100
4.9”104
<100
–
ꢀ6.7
–
–
20.1
–
>104
–
individual binding constants could not be reliably determined by either method
–
> 104
5100
ꢀ13.2
ꢀ22.4
ꢀ19.6
200
1.0”104
3100
ꢀ2.7
10.5
24.8
13.8
ꢀ20.8
ꢀ17.3
ꢀ23.4
5100
1200
ꢀ5.8
15.0
19.3
35.9
> 104
4500
> 104
800
2.4
2.0
> 104
1000
ꢀ5.9
1.6”104
12.5
binding constants could not be reliably determined by calorimetry
500
binding constants could not be reliably determined by calorimetry
9
> 104
<100
ꢀ15.1
3.5
18.6
ꢀ24.7
2.5”104
ꢀ6.4
18.4
10
11
12
> 104
individual binding constants could not be reliably determined by either method
1200 at
363 K[d]
13
1
DMSO
N-Boc-d-Asp
weak binding
at 363 K[d]
14
15
16
17
18
19
20
21
22
23
24
25
1
1
2
2
2
2
2
2
2
2
3
3
DMSO
DMSO
CH3CN
CH3CN
DMSO
DMSO
CH3CN
CH3CN
DMSO
DMSO
DMSO
DMSO
DiBz-l-tartrate
DiBz-d-tartrate
N-Boc-l-Glu
N-Boc-d-Glu
N-Boc-l-Glu
N-Boc-d-Glu
N-Boc-l-Asp
N-Boc-d-Asp
N-Boc-l-Asp
N-Boc-d-Asp
N-Boc-l-Glu
N-Boc-d-Glu
–
–
-
–
–
211
103
[e]
[e]
–
–
–
ꢀ18.5
–
ꢀ15.0
1800
<100
430
ꢀ21.8
–
ꢀ7.8
ꢀ3.3
–
7.2
–
ꢀ19.0
–
<100
2100
<100
–
ꢀ11.7
–
–
7.2
–
[e]
[e]
[f]
[f]
individual binding constants could not be reliably determined by either method
[e]
[e]
–
–
–
–
ꢀ20.5
3900
<100
ꢀ14.7
5.8
–
19.0
<100
2100
–
–
10.5
[e]
[e]
ꢀ8.4
individual binding constants could not be reliably determined by either method
individual binding constants could not be reliably determined by either method
[f]
[f]
–
–
–
–
ꢀ25.3
2.7”104
260
ꢀ5.9
19.4
9.4
ꢀ14.8
390
ꢀ8.6
6.2
16.5
[f]
[f]
ꢀ13.8
ꢀ4.4
ꢀ24.5
1.9”104
ꢀ8.0
[a] Association constants are reported to two significant figures with estimated errors of ꢁ 20%. The reported value of Ka1:2 refers to the binding constant for
the association of the 1:1 complex with the second equivalent of guest. The overall binding constant is therefore the product of the reported Ka1:1 and Ka1:2. In
general the errors on thermodynamic data are estimated at ꢁ 0.5 kJmolꢀ1, but where 1:1 and 1:2 association constants are both significant (entries 6, 7, 9, 24,
and 25) the errors may be larger. In some cases where both binding stoichiometries are competing the titration data (from one or both of the titration methods)
could not be fitted at all reliably (entries 4, 8, 11, 19, 22 and 23), although strong binding was generally apparent from the data. [b] Deuterated solvents, CD3CN
and [D6]DMSO, were used for NMR titration experiments. [c] All guests were used as the tetrabutylammonium salts. [d] Binding kinetics were slow (on the
NMR timescale) at room temperature and NMR titration experiments were carried out at 908C, see text. Calorimetry experiments were not carried out at this
temperature. [e] NMR titration experiments were not carried out in CD3CN because of solubility problems with macrocycle 2 in this solvent. [f] Binding con-
stants could not be reliably determined by NMR titrations.
binding was dominated by either 1:1 or 1:2 complexation,
reliable association constants (Ka) could be determined and
binding stoichiometry was confirmed using Job plots.[10] In
cases where binding was weak it was often not possible to
distinguish between 1:1 and 1:2 complexation, since the data
could be fitted to both binding isotherms. In some cases it
was apparent that there was strong overall binding, and
both 1:1 and 1:2 complexes were formed, but neither was
dominant. In these cases the individual association constants
could not always be reliably determined. Job plots with
these systems also gave ambiguous results as the plots
showed broad maxima at stoichiometries typically of ~1:1.5.
Some of the data obtained gave binding constants at the
upper limit of detectability by using NMR titrations (104–
When overall binding was weak, or when 1:1 and 1:2 associ-
ation constants were both significant then, as with the NMR
titrations, the individual association constants could not
always be reliably determined.
In general titration experiments were performed more
than once to obtain reproducible data. Where the associa-
tion constant could be determined for the same complex by
using both isothermal calorimetry and 1H NMR titrations,
agreement between the two methods was generally good.
Isothermal calorimetry has the considerable advantage
that it allows direct determination of the association con-
stants and the enthalpy (DH) of binding. Hence the Gibbs
free energy (DG) and entropic contribution (DS) can be cal-
culated. In general we found that using isothermal calorime-
try with these macrocycles, where binding was predominant-
ly either 1:1 or 1:2 complexation, the association constants
(and hence the Gibbs free energy) could be reproduced to a
high degree of accuracy. The enthalpy–entropy balance in
host–guest binding can be very sensitive to solvent composi-
tion and to the presence of small quantities of water, partic-
ularly when the bulk solvent is relatively apolar,[15] so for all
105 mꢀ1) and it was concluded that in these cases Ka
>
104 mꢀ1, but more precise values could not be reliably ob-
tained from these titrations.[11]
Isothermal calorimetric binding data[12–14] was fitted using
a one- or two-site binding model and, again, where binding
was dominated by either 1:1 or 1:2 complexation, reliable
and reproducible association constants could be determined.
Chem. Eur. J. 2005, 11, 5674 – 5688
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5677

J. D. Kilburn et al.
titration experiments solvents and the samples of host and
guest were carefully dried prior to use.
by the entropic contribution (TDS = 20.8 kJmolꢀ1), indicat-
ing that complexation is promoted by the release of solvent
molecules from the host and guest to bulk solvent. The calo-
rimetric data obtained with the N-Boc-d-glutamate salt indi-
cated a small 1:1 binding constant and strong 1:2 (host/
guest) binding (DG = ꢀ26.7 kJmolꢀ1, Ka1:2 = 4.9”104 mꢀ1),
again confirming the results of the NMR titrations. Confor-
mational changes in the receptor on binding N-Boc-d-gluta-
mate are clearly evidenced by the large shifts in many of the
Binding studies by using macrocycle 1: Initial investigation
of the binding properties of receptor 1 with N-Boc-gluta-
mate dicarboxylate has been described previously.[5] In sum-
mary, addition of the N-Boc-l-glutamate bis(tetrabutylam-
monium) salt to a solution of 1 in CD3CN led to a strong 1:1
association (Ka1:1 > 104 mꢀ1), and significant downfield shifts
1
1
of the H NMR NH signals, consistent with the formation of
CH signals in the H NMR spectrum of the receptor on ad-
strong hydrogen bonds and the proposed mode of binding
(Figure 1). The binding data obtained from the addition of
N-Boc-d-glutamate could not be fitted to a 1:1 binding iso-
therm, but could be fitted to a 1:2 (host/guest) binding iso-
therm and yielded Ka1:2 > 104 mꢀ1. Binding stoichiometry
was confirmed by Job plots[10] and is further evident from
the 1H NMR titrations since saturation is rapidly reached
after addition of ~1 equivalent of the N-Boc-l-glutamate
salt, whereas saturation is reached only after addition of ~2
equivalents of N-Boc-d-glutamate salt (Figure 5). The rela-
dition of the guest.
Binding studies with macrocycle 1 and N-Boc-l-glutamate
in [D6]DMSO gave a similar picture, with a simple 1:1 bind-
ing, obtained by either NMR titration (DG
=
ꢀ20.4 kJmolꢀ1, Ka1:1 = 3.7”103 mꢀ1) or isothermal calorime-
try (DG = ꢀ19.2 kJmolꢀ1, Ka1:1 = 2.3”103 mꢀ1), again with a
large entropic contribution (TDS=8.5 kJmolꢀ1). Binding
data for N-Boc-d-glutamate, however, could not be reliably
fitted to give individual binding constants.
In the less competitive solvent, CDCl3, addition of either
enantiomer of the glutamate
salt did not lead to any discerni-
ble change in the 1H NMR
spectrum of the macrocycle,
which is attributed to the high
energetic cost of breaking the
intramolecular hydrogen bonds
to open the tightly wrapped
conformation adopted by the
receptor in chloroform.
To confirm the above inter-
pretation of 1:2 complexation of
N-Boc-d-glutamate
binding
studies have nowbeen carried
out with tetrabutylammonium
acetate. Significant changes in
the 1H NMR spectrum of the
receptor were noted after the
addition of the acetate guest in
CD3CN. Downfield shifts of the
thiourea NH (Ddsat =1.03 ppm),
~
^
Figure 5. Binding titration curves for macrocycle 1, showing the shift of the amide ( : Boc-l-Glu, : Boc-d- the
amide
NH
(Ddsat =
&
Glu) and the thiourea ( : Boc-l-Glu, ”: Boc-d-Glu) NH protons on addition of the bis(tetrabutylammonium)
salts of N-Boc-l-glutamate and N-Boc-d-glutamate in CD3CN as solvent, indicating the strong 1:1 binding
0.70 ppm) and the benzylic CH
signals (Ddsat =0.47 ppm) were
recorded.
with the former, and 2:1 binding with the latter.
Saturation
was
reached after the addition of
tively smooth titration curve obtained with N-Boc-d-gluta-
mate, and lack of an obvious transition after addition of
only one equivalent of the guest, suggests that binding of
the second guest molecule is stronger than binding of the
first (positive co-operativity), such that the receptor tends to
saturate both binding sites simultaneously, that is, the 1:1
complex is only present in very small amounts relative to
uncomplexed receptor or the 2:1 complex.
Isothermal calorimetric binding data obtained with N-
Boc-l-glutamate salt confirmed the strong 1:1 binding (DG
= ꢀ25.3 kJmolꢀ1, Ka1:1 = 2.7”104 mꢀ1), which is dominated
two equivalents of acetate and the binding data could be
fitted to the expected 1:2 stoichiometry (DG=
ꢀ23.0 kJmolꢀ1, Ka1:2 > 104 mꢀ1). As with N-Boc-d-glutamate,
1
the H NMR titration curve obtained with acetate gave no
indication of an intermediate 1:1 complex which is consis-
tent with a small or negligible 1:1 binding constant. Isother-
mal calorimetric data obtained with the acetate salt also in-
dicated a small 1:1 binding constant (DG ~ꢀ13 kJmolꢀ1,
Ka1:1 ~200mꢀ1) and strong 1:2 (host/guest) binding (DG =
ꢀ20.8 kJmolꢀ1, Ka1:2 = 5100mꢀ1), confirming the results of
the NMR titrations.
5678
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5674 – 5688

Enantioselective Receptors
FULL PAPER
In order to probe the binding properties of macrocycle 1
in more detail binding studies with a series of dicarboxylate
substrates have also nowbeen carried out. Firstly, to assess
the influence of the bulky tert-butyloxycarbonyl protecting
group of the glutamate substrates, binding studies were con-
ducted with both enantiomers of N-Ac-glutamate.
Addition of N-Ac-l-glutamate salt to a solution of 1 in
CD3CN led to significant downfield shifts of the amide NH
and thiourea NH (Ddsat > 1.0 ppm) but in contrast to the ti-
tration with N-Boc-l-glutamate saturation was not reached
after addition of one equivalent of the salt, and the titration
curves showed a significantly sigmasoidal shape, suggesting
that both 1:1 and 1:2 complexes are formed (Figure 6).
Addition of either enantiomer of N-Boc-aspartate tetra-
butylammonium salt to a solution of 1 in CD3CN at room
1
temperature led to dramatic changes in the H NMR spec-
trum of the macrocycle with considerable downfield shifts of
the thiourea NH (Ddsat > 1.5 ppm), the amide NH (Ddsat
>
1.0 ppm) and the benzylic CH signals (Ddsat > 0.5 ppm).
The titration data for N-Boc-l-aspartate showed that satura-
tion was essentially reached after addition of one equivalent
of the substrate and could be readily fitted to a 1:1 isotherm
(Ka1:1 > 104 mꢀ1), but the binding data for N-Boc-d-aspartate
could not be fitted to a simple 1:1 or 1:2 binding stoichiome-
try and indicate that both 1:1 and 1:2 complexes were
formed, but neither was dominant. Similarly isothermal cal-
orimetry data with these guests
could not be fitted reliably.
In [D6]DMSO, addition of N-
Boc-l-aspartate to macrocycle
1 at room temperature caused a
broadening of the signals in the
1H NMR spectrum of the re-
ceptor. During the addition to
the host solution it was possible
to observe one set of peaks for
the unbound macrocycle, which
decreased in intensity, and a
newset of peaks (thiourea NH:
d=9.7 ppm; amide NH: d=
10.3 ppm), which increased in
intensity on addition of the
guest and were attributed to
the bound host–guest complex.
This newset of broad signals
became reasonably well-re-
solved after the addition of four
~
^
Figure 6. Binding titration curves for macrocycle 1, showing the shift of a) the amide ( : Ac-l-Glu, : Ac-d-
equivalents of guest. It was con-
cluded that binding of N-Boc-l-
aspartate by macrocycle 1 in
[D6]DMSO at room tempera-
ture was slow on the NMR
&
Glu) and b) the thiourea ( : Ac-l-Glu, ”: Ac-d-Glu) NH protons on addition of the bis(tetrabutylammonium)
salts of N-Ac-l-glutamate and N-Ac-d-glutamate in CD3CN as solvent (1:1 [host]:[guest] is reached when
[guest]=1.5 mm).
Indeed the data could be fitted to give a large Ka1:1
>
time-scale. In order to overcome the slowbinding kinetics, a
104 mꢀ1, but also a significant Ka1:2 = 5100mꢀ1. Similar down-
field shifts of the amide NH and thiourea NH were ob-
served on addition of N-Ac-d-glutamate salt to a solution of
1 in CD3CN. The NMR titration data could be fitted to give
a large Ka1:1 = 4500mꢀ1 but a larger Ka1:2 > 104 mꢀ1. Similar
results were obtained in [D6]DMSO. The NMR titration
data indicates that for N-Ac-l-glutamate 1:1 complexation is
dominant, whereas for N-Ac-d-glutamate 1:2 binding is the
dominant, but in neither case is the selectivity as pro-
nounced as with the two enantiomers of N-Boc-glutamate.
Isothermal calorimetry data with these guests in both
CH3CN and DMSO confirmed the NMR results and binding
in all cases is dominated by the entropic contribution.
To assess the ability to bind dicarboxylates with shorter
chain lengths, binding studies have been conducted
with both enantiomers of N-Boc-aspartate and dibenzoyl
tartrate.
1H NMR titration was carried out at 908C. At this tempera-
1
ture the H NMR spectrum of macrocycle 1 in [D6]DMSO
was well-resolved and the signals of the amide and thiourea
protons were both ~0.9 ppm upfield compared to the spec-
trum in [D6]DMSO at room temperature. Addition of N-
Boc-l-aspartate to macrocycle 1 at 908C in [D6]DMSO led
to downfield shifts of the thiourea NH (Ddsat =0.64 ppm)
and the amide NH (Ddsat =0.31 ppm) signals. No significant
shift was noted for the benzylic CH protons. The binding
data could be fitted to a 1:1 binding isotherm (DG=
ꢀ17.6 kJmolꢀ1, Ka1:1 = 1.2”103 mꢀ1). Addition of N-Boc-d-
aspartate to macrocycle 1 at 908C in [D6]DMSO produced
less pronounced downfield shifts of the thiourea NH
(Ddsat =0.42 ppm) and amide NH (Ddsat =0.21 ppm) protons
and the slope of the titration curve indicated very weak
binding and only ~25% of saturation was reached after the
addition of five equivalents of guest. The data could not be
Chem. Eur. J. 2005, 11, 5674 – 5688
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5679

J. D. Kilburn et al.
fitted unambiguously to a 1:1 or 1:2 (host/guest) binding iso-
therm. Isothermal calorimetry experiments were not per-
formed at these elevated temperatures.
fitted, and indicated relatively weak 1:1 binding (DG =
ꢀ15.0 kJmolꢀ1, Ka1:1 =530mꢀ1).
The addition of the tetrabutylammonium salts of both
enantiomers of dibenzoyl tartrate in [D6]DMSO at room
temperature led to small downfield shifts in the thiourea
NH (Ddsat ~0.3 ppm), suggesting weak binding. The binding
data for both enantiomers could be fitted to a 1:2 (host/
guest) isotherm (Ka1:2 =210mꢀ1 for dibenzoyl-l-tartrate and
100mꢀ1 for dibenzoyl-d-tartrate). Presumably the tartrates
substrates are too bulky to allow1:1 complexation and even
1:2 binding with the receptor is weak compared to the 1:2
binding of glutamate and aspartate carboxylates. With such
weak binding constants isothermal calorimetry was not at-
tempted with these guests.
Binding studies by using macrocycles 3 and 4: Complexation
studies with macrocycles 3 and 4 were restricted, due to
their limited solubility, to titration experiments in DMSO
and used only the enantiomers of N-Boc-glutamate. Addi-
tion of N-Boc-l-glutamate to macrocycle 3 in [D6]DMSO
1
led to significant downfield shifts of the H NMR signals for
the thiourea NH (Ddsat > 0.5 ppm) and the amide NH
(Ddsat > 0.3 ppm), but saturation was not reached after the
addition of the first equivalent of N-Boc-l-glutamate, and
the binding data could not be fitted to a simple 1:1 or 1:2
(host/guest) isotherm. The calorimetric binding data ob-
tained with N-Boc-l-glutamate salt indicated a strong 1:1
binding (Ka1:1 = 2.7”104 mꢀ1) and a moderate 1:2 association
(Ka1:2 =390mꢀ1).
Binding studies by using macrocycle 2: Binding studies with
the chiral bisthiourea macrocycles 2, 3 and 4 were carried
out to probe the effects of the structural modifications on
the receptor on the binding properties. Binding studies with
macrocycle 2 and N-protected glutamate and aspartate, as
their tetrabutylammonium salts, were carried out by using
1H NMR titrations and isothermal calorimetry. 1H NMR
spectra of 1:1 mixtures of both enantiomers of N-Boc-gluta-
mate and N-Boc-aspartate with 2 in CDCl3 were obtained in
chloroform. As expected, addition of either enantiomer of
the guests did not lead to any discernible change in the
1H NMR spectrum of the macrocycle. Thus, as with macro-
cycle 1, macrocycle 2 adopts a conformation unsuitable for
binding in CDCl3 and the energy required to reorganise the
host into a suitable binding conformation, by breaking the
intramolecular hydrogen bonds, is not compensated for by
the host–guest binding interactions that would be established.
1H NMR titration experiments in CD3CN were hampered
by the limited solubility of macrocycle 2 in this solvent. The
problem was addressed by running isothermal calorimetric
titrations, which allowed the use of host solutions at much
lower concentrations. With both N-Boc-glutamate and N-
Boc-aspartate, macrocycle 2 has binding properties very sim-
ilar to those of macrocycle 1. Data for N-Boc-l-glutamate
and N-Boc-l-aspartate indicated formation of strong 1:1
complexes with 2, while both N-Boc-d-glutamate and N-
Boc-d-aspartate form predominantly 1:2 complexes. Nota-
bly, however, the association constants with 2 are at least an
order of magnitude lower than the corresponding associa-
tion constants with 1.
Addition of the N-Boc-d-glutamate tetrabutylammonium
salt to 3 also led to significant downfield shifts of the
1H NMR signals for the thiourea NH (Ddsat ~1 ppm), amide
NH (Ddsat ~0.4 ppm) protons of the macrocycle. Saturation
was reached after the addition of two equivalents of guest
but again the binding data could not be fitted to a simple
1:1 or 1:2 (host/guest) binding model. Isothermal calorime-
try data obtained with the N-Boc-d-glutamate salt and 3,
however, indicated a moderate 1:1 binding constant (Ka1:1
=
260mꢀ1) but a much stronger 1:2 association (Ka1:2 =1.9”
104 mꢀ1).
Addition of either enantiomers of N-Boc-glutamate to a
solution of receptor 4 in DMSO, on the other hand, did not
1
lead to any significant shifts in the H NMR spectrum of the
neat host, suggesting no or very weak binding.
Computational studies: Force field computational studies
were undertaken in order to shed further light on the con-
formational preferences of macrocycle 1 in the solvents
CHCl3, DMSO and CH3CN and the origin of the binding se-
lectivities experimentally determined for macrocycle 1 with
the enantiomers of N-Boc-glutamate. The computational
method used in the calculations is dependent on the solvent
(see Experimental Section). With CHCl3 as the solvent, the
generalized Born/surface area (GB/SA) continuum solvation
model was used[16] and conformational searches performed
using a mixed mode Monte Carlo (MC) search algorithm.
For DMSO and CH3CN, explicit solvation models were used
in molecular dynamics (MD) simulations. With the excep-
tion of the MD simulations for which Impact 2.7[17] was
used, all computations were performed using Macromodel
8.0 or 8.1.[18]
In DMSO, however, binding data for both enantiomers of
N-Boc-glutamate and N-Boc-aspartate could not be reliably
1
fitted to simple 1:1 or 1:2 binding isotherms using H NMR
1
titrations. In H NMR titrations, shifts of the thiourea NH
The global minimum geometry for 1 obtained from the
conformational search with CHCl3 as solvent has a C2 geom-
etry stabilized by strong intramolecular hydrogen bonds
(Figure 7) in agreement with the experimental 1H NMR
data and similar to the conformation previously predicted
by torsion angle dynamics (Figure 3).
(Ddsat > 0.2 ppm), the amide NH (Ddsat > 0.16 ppm) and
the benzylic CH signals (Ddsat > 0.17 ppm) were observed
but saturation was not reached even after addition of several
equivalents of the guest, suggesting that binding was weak,
and again probably weaker than the corresponding complex-
ation with macrocycle 1. Using isothermal calorimetry only
with N-Boc-l-glutamate as the guest could the data be
From the MD simulations in CH3CN and DMSO, a D2-
type conformation is predominant (also in agreement with
5680
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5674 – 5688

Enantioselective Receptors
FULL PAPER
reorganisation energy for binding either enantiomer
(~300 kJmolꢀ1) was found to be much greater than the appar-
ent energy gain from binding (~100 kJmolꢀ1). Of course the
values of energies obtained from force field calculations will
not be quantitatively accurate. However, an energy differ-
ence of ~200 kJmolꢀ1 is substantial and hence, the energeti-
cally unfavourable binding of N-Boc-glutamate enantiomers
to macrocycle 1 in CHCl3 is not experimentally observed.
The computational methods employed were not, however,
able to reproduce the binding selectivity of N-Boc-l-gluta-
mate over N-Boc-d-glutamate for 1:1 binding (computed
DDHL–D ~+7 kJmolꢀ1), reflecting the fact that the force
field parameters used in the MD simulations were insuffi-
cient for accurate calculation of the complexation energies.
Figure 7. Conformation of macrocycle 1 in CDCl3 as determined by mo-
lecular modeling showing intramolecular H-bond lengths.
Discussion
1H NMR data)—a conformation which is more favourable
Macrocycle 1 was previously identified as having rather
unique binding properties.[5] It proved to be highly enantio-
selective forming a 1:1 complex with N-Boc-l-glutamate
with an association constant at
for accepting
(Figure 8).
a substrate into the large open cavity
least two orders of magnitude
larger than the corresponding
association constant for the 1:1
complex with N-Boc-d-gluta-
mate. Furthermore, it is a gen-
erally held viewthat binding in-
teractions between polar func-
tionalities
(e.g.
hydrogen
bonds) will lead to strong com-
plexation in a non-polar solvent
(typically CHCl3), but weak (or
negligible) complexation in
more polar (competitive) sol-
vents or solvent mixtures, and
numerous examples of this phe-
Figure 8. Conformation of macrocycle 1 in DMSO as determined by molecular modeling.
Geometries of the 1:1 binding complex of macrocycle 1
with both enantiomers of N-Boc-glutamate were determined
in both CHCl3 and CH3CN. For each enantiomer the same
binding modes were obtained with CHCl3 or CH3CN as sol-
vent. According to these calculations the dicarboxylate
ligand forms eight hydrogen bonds with the macrocycle,
each carboxylate oxygen forming hydrogen bonds with
either two amide or two thioamide hydrogens (Figures 9
and 10), which differs from the model originally proposed in
which each carboxylate oxygen forms hydrogen bonds to
both an amide and a urea hydrogen (Figure 1).
Although geometries for complexation of 1 with N-Boc-
glutamate enantiomers in CHCl3 were located computation-
ally, no binding could be detected experimentally in this sol-
vent. The energy required to reorganise the tightly wrapped
macrocycle 1 (Figure 7) into a conformation favourable for
accepting the N-Boc-glutamate substrates was calculated
based on the lowest energy conformations of the macrocycle
from the conformational searches in the bound (with
N-Boc-glutamate) and the unbound state (Table 3). The
nomenon exist. For example, N-tolyl-N’-n-butylurea binds
tetrabutylammonium benzoate with Ka = 1300mꢀ1 in CDCl3
and Ka =150mꢀ1 in [D6]DMSO,[19] and thiourea 21 binds the
TBA salt of N-Ac-l-Phe with Ka = 4800mꢀ1 in CDCl3 and
Ka =680mꢀ1 in 10% [D6]DMSO/CDCl3.[6b] The lack of bind-
ing exhibited by macrocycle 1 in CDCl3, in contrast to the
results observed in CH3CN and DMSO, was thus remarka-
ble. However, NMR studies revealed that in CDCl3 the mac-
rocycle adopts a tightly wrapped conformation that is stabi-
lized by
a number of intramolecular hydrogen bonds
(Figure 3). This hydrogen-bonding motif has been previously
observed in the crystal structure of 21 which forms a dimeric
pair in the solid-state.[6b] We postulated that the energy re-
quired to reorganize the macrocycle into a suitable binding
conformation (i.e., to break the intramolecular hydrogen
bonds) was not compensated for by binding interactions that
would thereby be established. The more polar solvents,
CH3CN and DMSO, can solvate the hydrogen bonding func-
tionality of the macrocycle leading to a less rigidly con-
strained molecule, and this is confirmed by the 1H NMR
Chem. Eur. J. 2005, 11, 5674 – 5688
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5681

J. D. Kilburn et al.
Table 3. Reorganisation energy of macrocycle 1 in CHCl3 (Figure 7) into
a conformation suitable for binding of N-Boc-d,l-glutamate (Figure 9)
versus the subsequent energy gain from binding.
DH [kJmolꢀ1
]
Energy type[a]
l complex
d complex
binding energy
reorganisation energy
ꢀ100.0
+296.9
ꢀ100.8
+303.7
[a] As calculated using the MMFFs force field.[25] See text for details.
CHCl3 as previously determined by the NMR studies, and a
more flexible and open cavity in CH3CN and DMSO. The
computational studies also verify that the energy required to
reorganise the wrapped conformation of the macrocycle in
CHCl3 into an open conformation suitable for complexation
of carboxylates far exceeds the binding energy that would
result from such complexation.
Solvation of the macrocycle by the more polar solvents
allows tight binding of the N-Boc-l-glutamate salt—albeit
with some conformational reorganization of the receptor
(and associated energetic cost), as evidenced by the large
1
shifts of various CH signals in the H NMR spectrum on ad-
dition of the guest. Binding of the N-Boc-l-glutamate salt,
in these solvents, is driven to a large extent by entropy. The
same cavity is less willing to accommodate the enantiomeric
guest (N-Boc-D-glutamate) in a simple 1:1 binding mode,
and instead binds two N-Boc-d-glutamate guests with a
small binding constant for the first glutamate and a signifi-
cantly larger binding constant for the second (positive co-
operativity). At first sight this may seem surprising since
binding of the first carboxylate should introduce an electro-
static repulsion which might be expected to disfavour bind-
ing of a second carboxylate. However, the two carboxylate
guests may bind on opposite faces, and at opposite ends of
the macrocycle, in which case the electrostatic repulsion be-
tween the two guests should be small. Furthermore, if bind-
ing of the first N-Boc-d-glutamate requires considerable re-
organization of the receptor, and consequent energetic cost,
then, once the first carboxylate is bound, the receptor may
bind the second carboxylate without significant additional
reorganisation and associated energetic cost. Conformation-
al changes in the receptor, on binding N-Boc-d-glutamate in
acetonitrile, were, again, clearly evidenced by the large
shifts in many of the CH signals of the receptor in the
Figure 9. Geometry of the 1:1 complex of macrocycle 1 with: a) N-Boc-l-
glutamate; b) N-Boc-d-glutamate, in CH3CN as determined by molecular
modeling showing intermolecular H-bond lengths.
1H NMR spectrum, particularly the benzylic CHꢁs (Ddsat
0.5 ppm), on addition of N-Boc-d-glutamate.
>
The binding studies nowreported with acetate showthat
macrocycle 1 forms a relatively weak 1:1 complex with ace-
tate but a strong 1:2 (host/guest) complex. Complexation
with acetate led to significant shifts in the NH signals and
many of the CH signals of the receptor in the 1H NMR spec-
trum, very similar to those observed on complexation with
N-Boc-d-glutamate. These results are entirely consistent
with the hypothesis previously suggested for the complexa-
tion of N-Boc-d-glutamate which cannot be as readily ac-
commodated in the cavity of the macrocyle to form a stable
1:1 complex with both carboxylates bound simultaneously.
Figure 10. Schematic of mode of complexation of Boc-l-glutamate by
macrocycle 1 via eight hydrogen bonds determined by molecular model-
ing.
spectra which reflect the four-fold D2 symmetry of the mole-
cule in these solvents. This hypothesis has nowbeen verified
by the computational studies which identify a similar lowest
energy tightly wrapped conformation for macrocycle 1 in
5682
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5674 – 5688

Enantioselective Receptors
FULL PAPER
Not surprisingly macrocyle 1 can also bind N-Ac-gluta-
mate dicarboxylate salts, and binds N-Ac-l-glutamate with a
preference for 1:1 binding stoichiometry, whereas 2:1 bind-
ing of N-Ac-d-glutamate is preferred. However, the selectiv-
ities are not as pronounced as found with the sterically more
demanding N-Boc-glutamate substrates. The 1:1 binding of
N-Boc-l-glutamate appears in fact to be stronger than the
binding of N-Ac-l-glutamate and suggests that the former
substrate fits particularly well in the binding pocket. Al-
though the computational studies were unable to calculate
or differentiate between the binding energies of the two
enantiomers of N-Boc-glutamate, the lowenergy structures
found for the 1:1 binding complex of macrocycle 1 and N-
Boc-l-glutamate do indicate that the tert-butyl group is in
close contact with one of the phenyl residues of the macro-
cycle, and may thus benefit from a CH–p interaction provid-
ing additional stabilisation.[20] It is also noteworthy that the
1:1 binding of N-Ac-l-glutamate is slightly endothermic
while the 1:1 binding of N-Boc-l-glutamate is slightly exo-
thermic, but this is compensated to a large extent by the re-
duced gain in entropy on binding the latter. This would
appear to be a typical example of enthalpy–entropy com-
pensation as often found in host–guest systems,[21] and
one can speculate that N-Boc-l-glutamate is more rigidly
perature (which is only obtained at ꢀ408C for macrocycle
1).
Macrocycles 3 and 4 are both insoluble in CHCl3 and
CH3CN, but macrocycle 3 does form strong complexes with
both enantiomers of N-Boc-glutamate in DMSO. The bind-
ing selectivities for the two enantiomers of N-Boc-glutamate
are similar to those with macrocycle 1, but binding is in fact
stronger with macrocycle 3. Macrocycle 4 on the other hand
appeared not to bind N-Boc-glutamate at all. Thus replace-
ment of the pyrido units in macrocycles 1 and 2 with benzo
units in 3 and 4 significantly alters the binding properties.
Analogous changes have been investigated in a number of
related receptor systems and based on this previous work
the pyrido unit was expected to help in preorganising the re-
ceptor for carboxylate recognition (Figure 1) with weak hy-
drogen bonds from adjacent amide and thiourea protons
and the pyrido nitrogen.[6] Conversely, however, the pres-
ence of the pyrido nitrogen can reduce the anion complexa-
tion ability of such systems (relative to the benzo analogue)
due to electrostatic repulsion.[22] The 1H NMR signals for
the NH protons (amide and thiourea) of 1 and 2 are in each
case significantly more downfield than the corresponding
signals for macrocycles 3 and 4, confirming the anticipated
interaction of the pyrido nitrogen and adjacent amide and
thiourea protons in the former pair of macrocycles. The po-
tential electrostatic repulsion between the pyrido nitrogen
and carboxylate oxygens does not, however, appear to
hamper the ability of 1 and 2 to complex carboxylates, and
the computational studies do suggest an alternative mode of
binding to that originally anticipated with each carboxylate
oxygen forming hydrogen bonds with either two amide or
two thiourea hydrogens (Figures 9 and 10) as opposed to
each carboxylate oxygen forming hydrogen bonds to both
an amide and a thiourea hydrogen adjacent to a pyridine
(Figure 1). This alternative mode would appear to minimise
repulsive interactions between carboxylate oxygens and
pyrido lone-pairs.
bound (in comparison with N-Ac-l-glutamate), with
a
stronger enthalpic contribution from hydrogen bonding and
possibly CH–p interactions, but with an entropic penalty as-
sociated with the greater restriction of the motion of the
guest.
Binding studies with N-Boc-l-aspartate (with a shorter
separation of the two carboxylates) reveal that macrocycle 1
can also comfortably accommodate this guest to form a
strong 1:1 complex in both CH3CN and DMSO, although
the binding kinetics are surprisingly slowin DMSO. Howev-
er, again the macrocycle is not able to discriminate so readi-
ly the other enantiomer and titration experiments in both
solvents suggest that with N-Boc-d-aspartate both 1:1 and
1:2 complexes are readily formed. Macrocycle 1 is, however,
sensitive to the overall steric bulk of the guest molecule and
did not form a strong 1:1 complex with either enantiomer of
dibenzoyl tartrate.
Conclusion
We have also nowstudied the binding properties of sever-
al analogous macrocycles 2–4. Macrocycle 2 exhibits some
similarities to the binding properties of macrocycle 1, with a
preference for formation of 1:1 complexes with the l-enan-
tiomers of N-Boc-glutamate and N-Boc-aspartate, and 1:2
complexes with the d-enantiomers. However, binding is gen-
erally weaker with this macrocycle than with 1 and competi-
tion between 1:1 and 1:2 binding stoichiometries is apparent.
Furthermore, whereas binding with macrocycle 1 is general-
ly dominated by the entropic contribution, this is less signifi-
cant for the binding with macrocycle 2. Thus the greater
conformational constraints imparted by the cyclohexyl unit
(in comparison to the diphenylethylene unit in 1) disfavour
binding of carboxylates, and also appear to favour the wrap-
ped conformation observed in CDCl3, as evidenced by the
well-resolved spectrum obtained in this solvent at room tem-
Overall macrocycle 1 is a potent receptor for the N-protect-
ed amino acids glutamate and aspartate, with novel enantio-
selective and solvent-dependent binding properties. Studies
on related macrocycles 2–4 indicate that if the binding prop-
erties of 1 are not entirely unique, they are sensitive to
small structural changes, and it is largely fortuitous that 1
was the first of the series of such macrocycles that we pre-
pared and led to the discovery of the unusual binding prop-
erties. The detailed study described herein gives some in-
sight into the various energetic contributions to the total
binding free energy of even such simple substrates. More
specifically the study has also suggested an alternative mode
of binding of carboxylate anion by the amidopyridyl thiour-
ea motif to that originally conceived, which minimises the
repulsive interaction between the bound anion and the pyri-
dine lone pair (Figures 1 and 10).
Chem. Eur. J. 2005, 11, 5674 – 5688
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5683

J. D. Kilburn et al.
ture heated at reflux for 8 h. The solvent was removed in vacuo to afford
a white solid which was redissolved in 2m HCl (5 mL). The solution was
refluxed for 30 minutes, after which the insoluble material was removed
by filtration. The aqueous solution was basified to pH 9 with 1m NaOH
and the precipitated diamine was extracted with CH2Cl2 (5”10 mL). The
organic layer was dried over MgSO4 and the solvent evaporated in vacuo
to afford diamine 7 as a white solid (330 mg, 84%). M.p. 73–758C; [a]D21
= 29.4 (c=1, 10% methanol/CH2Cl2); 1H NMR (400 MHz, CDCl3): d =
9.13 (brs, 2H, NHCO), 7.91 (d, 2H, J=7 Hz, PyrH), 7.64 (t, 2H, J=
7 Hz, PyrH), 7.24 (d, 2H, J=7 Hz, PyrH), 7.19–7.12 (m, 10H, ArH), 5.63
(brs, 2H, CHNH), 3.98 (brs, 4H, CH2NH2), 1.78 (s, 4H, NH2); 13C NMR
(100 MHz, CDCl3): d = 165.0 (0), 149.1 (0), 138.9 (0), 137.9 (0), 128.9
(1), 128.7 (1), 127.9 (1), 127.8 (1), 124.1 (1), 120.4 (1), 59.7 (1), 47.5 (2);
IR (neat): nmax =3362 (w), 1659 (m), 1590 cmꢀ1 (w); MS ES+: m/z: 481
[M+H]+; HRMS (ES+): m/z: calcd for C28H29N6O2: 481.2346, found
481.2359 [M+H]+.
Further studies in our laboratory are aimed at developing
related macrocycles that can bind glutamate and aspartate
derivatives in even more competitive solvents, that is water.
Experimental Section
General methods: Reactions were carried out in solvents of commercial
grade and, where necessary, distilled prior to use. THF was distilled
under nitrogen from benzophenone and sodium. CH2Cl2 was distilled
from calcium hydride, as was petroleum ether where the fraction boiling
between 40 and 608C was used. TLC was conducted on foil backed
sheets coated with silica gel (0.25 mm) which contained the fluorescent
indicator UV254. Column chromatography was performed on Sorbsil C60,
40–60 mesh silica.
1H NMR spectra were obtained at 300 MHz on Bruker AC300 and
Bruker AM 300 spectrometers and at 400 MHz on a Bruker DPX400
spectrometer. 13C NMR spectra were obtained at 75 MHz on Bruker
AC300 and Bruker AM 300 spectrometers and at 100 MHz on a Bruker
DPX400 spectrometer. Spectra were referenced with respect to the resid-
ual solvent peak for the deuterated solvent. Infrared spectra were ob-
tained on BIORAD Golden Gate FTS 135. All melting points were mea-
sured in open capillary tubes using a Gallenkamp Electrothermal Melting
Point Apparatus and are uncorrected. Optical rotations were measured
on a PolAr2001 polarimeter using the solvent stated, the concentration
given is in g per 100 mL. Electrospray mass spectra were obtained on a
Micromass platform with a quadrupole mass analyser. FAB spectra were
obtained on a VG Analytical 70–250-SE normal geometry double focus-
ing mass spectrometer. High-resolution accurate mass measurements
were carried out at 10,000 resolution using mixtures of polyethylene gly-
cols and/or polyethylene glycol methyl ethers as mass calibrants for FAB.
Semi-preparative reverse-phase HPLC was carried out using a Hewlett
Packard HP1100 Chemstation with an automated fractions collector and
Macrocycle 1: Carbon disulfide (414 mL, 6.86 mmol) was added to a mix-
ture of diamine 7 (165 mg, 0.50 mmol) in dry CH2Cl2 (10 mL) at ꢀ108C
and the mixture stirred for
1 hour. N,N’-Dicyclohexylcarbodiimide
(200 mg, 0.49 mmol) was added and the mixture stirred for 45 min at
ꢀ108C, then 30 min at room temperature. The excess carbon disulfide
and solvent were removed in vacuo yielding the crude bis(isothiocyanate)
8 as a pale yellowsolid. Bis(isothiocyanate) 8 in dry CH2Cl2 (10 mL) and
a further one equivalent of diamine 7 (165 mg, 0.20 mmol) in dry CH2Cl2
(10 mL) were simultaneously added over 3 h via syringe pump addition
to a solution of DMAP (20 mg, 10% by weight) in CH2Cl2 (50 mL)
under a slowstream of nitrogen. After stirring at room temperature for
16 h, the excess solvent was removed in vacuo. The resultant yellow resi-
due was purified by column chromatography (30% ethyl acetate/petrol
ether up to neat ethyl acetate) yielding macrocycle 1 as a white solid
(20.4 mg, 20%) Rf =0.48 (ethyl acetate); [a]2D1 = 146.0 (c=1, CH2Cl2);
m.p. 153–1548C; 1H NMR (400 MHz, [D6]DMSO, 908C): d = 9.38 (d,
4H, J=6 Hz, NHCO), 8.27 (t, 4H, J=5 Hz, NHCS), 7.93 (t, 4H, J=
7 Hz, PyrH), 7.84 (d, 4H, J=7 Hz, PyrH), 7.59 (d, 4H, J=7 Hz, PyrH),
7.48–7.26 (m, 20H, Ph), 5.82–5.77 (m, 2H, CHNHCO), 5.03 (dd, 4H, J=
16, 5 Hz, CHAHB(Ar)NHCS), 4.95 (dd, 4H, J=16, 5 Hz, CHAHB(Ar)-
a
Phenomenex C18 prodigy 5 mm (250 mm”10 mm) column. Water
(0.1% TFA) and acetonitrile (0.04% TFA) gradient was employed using
elution conditions: time 0 min: 10% CH3CN/H2O; 25 min: 90% CH3CN/
H2O; 40 min: 10% CH3CN/H2O. Calorimetric experiments were per-
formed on an Isothermal Titration Calorimeter from Microcal Inc.,
Northampton, Massachusetts, USA.
NHCS); 13C NMR (100 MHz, [D6]DMSO): d
= 165.1 (0), 157.3 (0),
149.4 (0), 139.8 (0), 138.7 (0), 128.5 (1), 128.0 (1), 127.6 (1), 127.4 (1),
124.8 (1), 120.8 (1), 58.1 (1), 49.2 (2); IR (neat): nmax =3310 (br), 1670 (s),
1525 (m), 1455 cmꢀ1 (w); MS ES+: m/z: 1045 [M+H]+, 1067 [M+Na]+;
HRMS (ES+): m/z: calcd for C58H52S2N12O4: 1045.3754, found 1045.3748
[M+H]+.
Synthetic details
Diamide 6: Pyridyl acid 5 (1.84 g, 6.5 mmol) was suspended in CH2Cl2
(15 mL) and cooled to 08C. Triethylamine (1.0 mL, 7.15 mmol) was
slowly added and a clear solution obtained. Diphenylchlorophosphate
(1.35 mL, 6.5 mmol) was added and the solution stirred for 1 h at 08C.
Diamide 9: Pyridyl acid 5 (1.23 g, 4.3 mmol) was suspended in CH2Cl2
(10 mL) and cooled to 08C. Triethylamine (0.66 mL, 4.73 mmol) was
slowly added and a clear solution obtained. Diphenylchlorophosphate
(0.89 mL, 4.3 mmol) was added and the solution stirred for 1 hour at
08C. (1S,2S)-1,2-Diaminocyclohexane-l-tartrate salt (564 mg, 2.15 mmol)
was suspended in water (3.3 mL) and K2CO3 (980 mg) added. After
30 min, the solution of diamine was added to the mixed anhydride solu-
tion at 08C and the resulting mixture stirred at 08C for 2 h, then at room
temperature for 14 h. The organic phase was extracted, washed with 2m
HCl (5 mL) and saturated aqueous NaHCO3 (5 mL). The organic layer
was dried over magnesium sulphate and the solvent evaporated in vacuo.
The crude material was purified by column chromatography (80% ethyl
acetate/petrol ether) to afford diamide 9 as a white solid (706 mg, 52%).
Rf =0.49 (ethyl acetate); m.p. 122–1248C; [a]2D1 =150.7 (c=1.1, CHCl3);
1H NMR (400 MHz, CDCl3): d = 8.09 (d, 2H, J=8 Hz, NH), 7.95–7.89
(m, 10H, ArH), 7.76 (dd, 4H, J=7, 3 Hz, PhthH), 7.29 (d, 2H, J=8 Hz,
PyrH), 5.04 (s, 4H, NCH2), 3.81 (m, 2H, NHCH), 2.06 (m, 2H), 1.75 (m,
2H), 1.50 (m, 2H), 1.39 (m, 2H); 13C NMR (100 MHz, CDCl3): d =
168.0 (0), 163.9 (0), 157.2 (0), 149.5 (0), 138.0 (0), 134.1 (1), 132.2 (1),
123.6 (1), 123.3 (1), 121.0 (1), 53.1 (1), 42.6 (2), 31.9 (2), 24.5 (2); IR
(neat): nmax =3260 (m), 1730 (s), 1653 (m), 1535 cmꢀ1 (s); MS ES+: m/z:
643 [M+H]+, 665 [M+Na]+; HRMS (ES+): m/z: calcd for
C36H30N6NaO6: 665.2125, found 665.2122 [M+Na]+.
(1S,2S)-1,2-Diphenyl-1,2-ethylenediamine-l-tartrate
salt
(1.18 mg,
3.25 mmol) was suspended in water (7 mL) and K2CO3 (2.0 g, 10.7 mmol)
added. After 30 min, the solution of diamine was added to the mixed an-
hydride solution at 08C. The resulting mixture was stirred for 2 h at 08C,
and then at room temperature for 14 h. The organic phase was extracted,
washed with 2m HCl (10 mL) and saturated aqueous NaHCO3 (10 mL).
The organic layer was dried over magnesium sulphate and the solvent
evaporated in vacuo. The crude material was purified by column chroma-
tography (30% ethyl acetate/petrol ether up to 50% ethyl acetate/petrol
ether) to give diamide 6 as a white solid (1.1 g, 46%). Rf =0.3 (ethyl ace-
tate); m.p. 106–1078C; [a]2D1 = 15.0 (c=1, CH2Cl2); 1H NMR (300 MHz,
CDCl3): d=8.75 (brs, 2H, CHNHPh), 7.97–7.77 (m, 10H, ArH), 7.73 (t,
2H, J=8 Hz, PyrH), 7.32 (d, 2H, J=8 Hz, PyrH), 7.22–7.02 (m, 10H,
ArH), 5.36 (brs, 2H, CHNHPh), 5.07 (d, 2H, J=16 Hz, PhthNCHAHB),
5.01 (d, 2H, J=16 Hz, PhthNCHAHB); 13C NMR (75 MHz, CDCl3): d=
168.1 (0), 164.1 (0), 154.2 (0), 149.5 (0), 145.8 (0), 138.7 (0), 138.2 (1),
134.2 (1), 132.4 (1), 128.5 (1), 127.7 (1), 123.7 (1), 123.6 (1), 121.2 (1),
59.2 (1), 42.6 (2); IR (neat): nmax =3230 (m), 1770 (w), 1710 (s), 1670 (m),
1510 (m), 1385 (s), 945 cmꢀ1 (s); MS ES+: m/z: 741 [M+H]+, 763
[M+Na]+; HRMS (FAB): m/z: calcd for C44H32N6O6Na: 763.2281, found
763.2281 [M+Na]+.
Diamine 10: Hydrazine monohydrate (59.2 mL, 1.22 mmol) was added to
a solution of diamide 9 (392 mg, 0.61 mmol) in ethanol (3 mL) and the
mixture heated at reflux for 8 h. The solvent was removed in vacuo to
Diamine 7: Hydrazine monohydrate (80 mL, 1.64 mmol) was added to a
solution of diamide 6 (600 mg, 0.82 mmol) in ethanol (5 mL) and the mix-
5684
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5674 – 5688

Enantioselective Receptors
FULL PAPER
afford a white solid which was redissolved in 2m HCl (3 mL). The solu-
tion was refluxed for 30 min, after which the insoluble material was re-
moved by filtration. The filtrate was basified to pH 9 with 1m NaOH and
washed with CH2Cl2 (5”10 mL). The combined organic extracts were
dried over MgSO4 and the solvent evaporated in vacuo to afford diamine
10 as a white foam (193 mg, 83%). 1H NMR (400 MHz, CDCl3): d =
8.54 (brs, 2H, NHCO), 7.96 (d, 2H, J=8 Hz, PyrH), 7.70 (t, 2H, J=
8 Hz, PyrH), 7.29 (d, 2H, J=8 Hz, PyrH), 4.10–3.90 (m, 6H, CH2NH2
and CHNH), 2.27 (m, 2H), 2.13 (m, 6H, CH2 and NH2), 1.85 (m, 2H),
1.48 (m, 2H); 13C NMR (100 MHz, CDCl3): d = 163.8 (0), 148.2 (0),
136.8 (0), 126.5 (1), 122.8 (1), 119.3 (1), 52.9 (1), 45.9 (2), 31.3 (2), 23.8
(2); IR (neat): nmax =3430 (m), 1703 (s), 1547 cmꢀ1 (m); MS ES+: m/z:
383 [M+H]+; HRMS (ES+): m/z: calcd for C20H27N6O2: 383.2195, found
383.2188 [M+H]+.
with CHCl3 to afford dicarbamate 14 as a white solid (618 mg, 64%).
Rf =0.83 (5% MeOH in CH2Cl2); m.p. 187–1888C; [a]2D3 =ꢀ40.5 (c=1.0,
DMSO); 1H NMR (400 MHz, [D6]DMSO): d = 8.97 (d, 2H, J=8 Hz,
NHCH), 7.59–7.54 (m, 4H, ArH), 7.46–7.32 (m, 10H, ArH), 7.31–7.25
(m, 4H, ArH), 7.15 (t, 2H, J=7 Hz, NHCO), 5.67 (d, 2H, J=8 Hz,
CHNH), 4.14 (d, 4H, J=8 Hz, CH2NH), 1.37 (s, 18H, CH3); 13C NMR
(100 MHz, [D6]DMSO): d = 166.5 (0), 155.7 (0), 140.6 (0), 140.4 (0),
134.8 (0), 129.6 (1), 128.1 (1), 127.8 (1), 127.2 (1), 126.7 (1), 125.9 (1),
125.3 (1), 77.8 (0), 57.2 (1), 43.1 (2), 28.2 (3); IR (neat): nmax =3295 (m),
2973 (w), 1689 (s), 1630 (s), 1528 (s), 1361 (m), 1243 cmꢀ1 (m); MS ES+:
m/z (%): 679 [M+H]+, 701 [M+Na]+ (100), 1379 (15) [2M+Na]+;
HRMS (ES+): m/z: calcd for C40H46N4NaO6: 701.3315, found 701.3320
[M+Na]+.
Diamine 15: Dicarbamate 11 (312 mg, 0.46 mmol) was dissolved in TFA/
CH2Cl2 1:1 (10 mL) and the solution stirred at room temperature for 2 h.
Toluene (10 mL) was added and the solvent evaporated in vacuo to give
the bis-TFA salt as a clear oil (294 mg, quantitative). The salt was dis-
solved in CH2Cl2 (10 mL), washed with 10% aqueous K2CO3 (10 mL),
dried with MgSO4 and evaporated in vacuo to give the diamine 15 as a
white foam (206 mg, 70%). M.p. 2838C; [a]2D3 = ꢀ89.1 (c=0.7, DMSO);
1H NMR (400 MHz, [D6]DMSO): d = 9.29 (brs, 2H, NHCH), 7.77–7.53
(m, 4H, ArH), 7.43 (d, 2H, J=8 Hz, ArH), 7.35–7.10 (m, 12H, ArH),
5.69 (d, 2H, J=8 Hz, CHNH), 4.02 (brs, 4H, CH2NH2), 1.77 (s, 4H,
NH2); 13C NMR (100 MHz, [D6]DMSO): d = 166.7 (0), 141.4 (0), 140.6
(0), 134.7 (0), 127.7 (1), 127.5 (1), 127.4 (1), 126.7 (1), 126.5 (1), 126.0 (1),
125.2 (1), 57.4 (1), 45.6 (2); IR (neat): nmax =3298 (w), 1633 cmꢀ1 (s); MS
ES+: m/z: 479 [M+H]+, 501 [M+Na]+, 979 [2M+Na]+; HRMS (ES+):
m/z: calcd for C30H31N4O2: 479.2448, found 479.2442 [M+H]+.
Macrocycle 2: Carbon disulfide (207 mL, 3.7 mmol) was added to a mix-
ture of diamine 10 (100 mg, 0.26 mmol) in dry CH2Cl2 (5 mL) at ꢀ108C
and the mixture stirred for 1 h. N,N’-Dicyclohexylcarbodiimide (107 mg,
0.50 mmol) was added and the mixture stirred for a further 45 min at
ꢀ108C then 30 min at room temperature. The excess carbon disulfide
and solvent were removed in vacuo yielding the crude bis(isothiocyanate)
11 as a white solid. Bis(isothiocyanate) 11 in dry CH2Cl2 (5 mL) and a
further one equivalent of diamine 10 (100 mg, 0.26 mmol) in dry CH2Cl2
(5 mL) were simultaneously added over 3 h via syringe pump addition to
a solution of DMAP (10 mg, 10% by weight) in CH2Cl2 (25 mL) under a
slowstream of nitrogen. After stirring at room temperature for 16 h, the
excess solvent was removed in vacuo and the resultant yellow residue pu-
rified by column chromatography (ethyl acetate) to yield macrocycle 2 as
a
white solid (156 mg, 72%). Rf =0.54 (ethyl acetate); m.p. 2408C
1
(decomp); [a]2D1 =165.5 (c=1.1, CHCl3); H NMR (400 MHz, CDCl3): d
= 9.24 (brs, 2H, NHCO), 8.73 (brs, 2H, NHCS), 8.15 (d, 2H, J=8 Hz,
NHCO), 8.00 (d, 2H, J=7 Hz, PyrH), 7.79 (t, 2H, J=7 Hz, PyrH), 7.72
(brd, 2H, J=7 Hz, NHCS), 7.36 (d, 2H, J=7 Hz, PyrH), 7.23 (brs, 4H,
PyrH), 6.85 (brs, 2H, PyrH), 5.42 (m, 2H, CHNH), 5.14 (brd, 2H, J=
17 Hz, CHAHBNH), 4.88 (brd, 2H, J=17 Hz, CHAHBNH), 4.34 (d, 2H,
J=12 Hz, CHCHDNH), 4.12 (m, 2H, CHCHDNH), 3.69 (m, 2H, CHNH),
2.57 (m, 2H), 2.12 (m, 2H), 1.82 (m, 4H), 1.56 (m, 2H), 1.53 (m, 2H),
1.31 (m, 4H); 13C NMR (100 MHz, [D6]DMSO): d = 182.0 (0), 164.1 (0),
154.2 (0), 148.6 (0), 138.5 (1), 120.0 (1), 117.9 (1), 54.5 (2), 49.0 (1), 30.8
(2), 24.0 (2); MS ES+: m/z: 850 [M+H]+; IR (neat): nmax =3310 (br),
2360 (w), 2330 (w), 1670 (m), 1525 (m), 1450 (w), 1360 (w), 1255 cmꢀ1
(w); HRMS (ES+): m/z: calcd for C42H49N12O4S2: 849.3444, found
849.3448 [M+H]+.
Macrocycle 3: Diamine 15 (200 mg, 0.42 mmol) was dissolved in CH2Cl2
(5 mL) and 0.5m aqueous K2CO3 added (5 mL). After stirring for 20 min,
thiophosgene (127 mL, 1.67 mmol) was added directly to the organic layer
and the solution stirred for 16 h. The organic layer was washed with 2m
aqueous HCl (10 mL), dried over MgSO4 and the solvent evaporated in
vacuo to give crude bis(isothiocyanate) 16 as a pale yellowfoam. The bi-
sisocyanate in dry CH2Cl2 (5 mL) and a further one equivalent of dia-
mine 15 (200 mg, 0.42 mL) in dry CH2Cl2 (5 mL) were simultaneously
added over 3 h via syringe pump addition to a solution of triethylamine
(116 mL, 0.84 mmol) in dry CH2Cl2 (50 mL) under a slowstream of nitro-
gen. The mixture was stirred at room temperature for 16 h. Macrocycle 3
precipitated out of solution as a yellowsolid. The compound was re-
moved by filtration and recrystallised (DMF/diethyl ether) to give macro-
cycle 3 as a pale yellowsolid (68 mg, 17%). M.p. 240 8C (decomp);
1H NMR (400 MHz, [D6]DMSO): d = 9.00 (brs, 4H, NHCO), 7.95 (brs,
4H, NHCS), 7.61–7.57 (m, 8H, ArH), 7.36 (m, 16H, ArH), 7.22–7.20 (m,
8H, ArH), 7.14–7.12 (m, 4H, ArH), 5.66 (4d, H, J=5 Hz, CHNH), 4.68
(brs, 8H, CH2NH); 13C NMR (100 MHz, [D6]DMSO): d = 168.5 (0),
164.4 (0), 142.7 (0), 136.8 (0), 131.9 (1), 130.2 (1), 129.6 (1), 129.4 (1),
128.9 (1), 128.4 (1), 127.9 (1), 59.4 (1), 50.6 (2); IR (neat): nmax =3271 (b),
1637 (s), 1529 cmꢀ1 (s); MALDI MS: m/z: 1041 [M+H]+, 1063 [M+Na]+;
elemental analysis calcd (%) for C62H56N8O4S2: C 71.54, H 5.42; N 10.76,
S 6.16; found C 71.11, H 5.28, N 10.36, S 5.66.
Bisnitrile 13: 3-Cyanobenzoic acid (867 mg, 5.9 mmol) was refluxed in
neat thionyl chloride (12 mL) for 4 h. The solvent was removed in vacuo
and the residue dissolved in dry CH2Cl2 (5 mL). (1S,2S)-(ꢀ)-1,2-Dipheny-
lethylenediamine (400 mg, 1.8 mmol) and DMAP (1.15 g, 9.40 mmol)
were added and the mixture stirred at room temperature for 2 d. The sol-
vent was removed in vacuo and the residue purified by column chroma-
tography (50% ethyl acetate/petrol ether) to give bisnitrile 13 as a white
solid (420 mg, 48%). Rf =0.33 (50% ethyl acetate/petrol ether); m.p.
243–2448C; [a]2D3
=
ꢀ102.7 (c=0.8, DMSO); 1H NMR (400 MHz,
[D6]DMSO): d = 8.06 (s, 2H, ArH), 7.98 (d, 2H, J=8 Hz, NH), 7.81 (d,
2H, J=8 Hz, ArH), 7.56 (t, 2H, J=8 Hz, ArH), 7.32 (d, 2H, J=8 Hz,
ArH), 7.20–7.10 (m, 10H, Ph), 5.58 (d, 2H, J=8 Hz, NHCH); 13C NMR
(100 MHz, [D6]DMSO): d = 171.6 (0), 166.5 (0), 140.3 (0), 136.4 (0),
135.9 (1), 132.6 (1), 132.2 (1), 130.6 (1), 129.3 (1), 128.9 (1), 128.6 (1),
119.0 (0), 58.4 (1); IR (neat): nmax =3287 (b), 2981 (w), 2927 (w), 1696 (s),
1631 cmꢀ1 (s); MS ES+: m/z: 471 [M+H]+, 963 [2M+Na]+; HRMS (ES+):
m/z: calcd for C60H44N8NaO4: 963.3377, found 963.3369 [2M+Na]+.
Bisnitrile 17: Diphenylchlorophosphate (4.15 mL, 0.02 mol) was added to
a solution of 3-cyanobenzoic acid (3 g, 0.02 mol) and triethylamine
(3.1 mL, 0.022 mol) in CH2Cl2 (15 mL) and the solution stirred for 1 h at
08C. (1S,2S)-1,2-Diaminocyclohexane-l-tartrate salt (2.62 g, 0.01 mol)
was suspended in water (15 mL) and K2CO3 (4.5 g, 0.03 mol) added.
After 30 min, the diamine solution was added to the mixed anhydride so-
lution at 08C. The resulting mixture was stirred at 08C for 2 h and then
at room temperature for 14 h. The organic phase was washed with 2m
HCl (50 mL), saturated aqueous NaHCO3 (50 mL), and then dried over
MgSO4. The solvent was evaporated in vacuo and the residue purified by
flash column chromatography (ethyl acetate) to give bisnitrile 17 as a
white solid (3.61 g, 97%). Rf =0.60 (ethyl acetate); m.p. 236–2378C; [a]D22
= 191 (c=1.0, DMSO); 1H NMR (400 MHz, [D6]DMSO): d = 8.50 (d,
2H, J=8 Hz, NH), 8.05 (s, 2H, ArH), 7.95 (d, 2H, J=8 Hz, ArH), 7.88
(d, 2H, J=8 Hz, ArH), 7.57 (t, 2H, J=8 Hz, ArH), 3.91 (m, 2H,
CHNH), 1.72 (m, 2H), 1.50 (m, 4H), 1.26 (m, 2H); 13C NMR (100 MHz,
[D6]DMSO): d = 164.5 (0), 135.9 (0), 134.4 (0), 131.9 (1), 130.7 (1), 129.6
Dicarbamate 14: Di-tert-butyl dicarbonate (1.25 g, 5.72 mmol) and NiCl2
(371 mg, 2.86 mmol) were added to a solution of 13 (673 mg, 1.43 mmol)
in THF (15 mL) and MeOH (10 mL). Sodium borohydride (757 mg,
20.0 mmol) was carefully added to the mixture at 08C and the mixture
stirred at room temperature for 16 h. The solvent was removed in vacuo
and the residue was dissolved in ethyl acetate (20 mL) and saturated
aqueous NaHCO3 (10 mL). The insoluble precipitate was removed by fil-
tration and washed with ethyl acetate (2”15 mL). The combined organic
layers were dried over MgSO4 and evaporated in vacuo and triturated
Chem. Eur. J. 2005, 11, 5674 – 5688
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5685

J. D. Kilburn et al.
(1), 118.3 (0), 111.3 (1), 53.1 (1), 31.4 (2), 24.7 (2); IR (neat): nmax =3273
(m), 1630 (s), 1541 cmꢀ1 (s); MS ES+: m/z: 373 [M+H]+, 410 [M+Na]+;
HRMS (ES+): m/z: calcd for C44H40N8NaO4: 767.3065, found 767.3065;
elemental analysis calcd (%) for C22H20N4O2: C 70.95, H 5.41, N 15.04;
found C 70.77, H 5.41, N 14.93.
steps (LMCS) is used.[24] For MCMM steps, 2 to 23 (Nꢀ1) of the defined
macrocyclic torsions (where N is the number of defined torsions) were
randomly adjusted; for LMCS steps, the lowfrequency eigenvectors
(“soft” vibrational modes) in the system were used to direct the search.
The definition of ring opening/closures allowed for the torsional varia-
tions within the macrocycle. A 1:1 ratio of Monte Carlo torsional moves
to lowmode moves was used. The Monte Carlo Structure Selection
(MCSS) option used was a “random walk” search where the most recent
structure within a 75 kJmolꢀ1 energy window of the current “global”
minimum was used as the starting structure for the subsequent MC step.
Finally, the MMFFs force field[25] was used in combination with the GB/
SA solvation model for CHCl3.[16] Computations were performed using
MacroModel 8.0.[18]
Dicarbamate 18: Di-tert-butyl dicarbonate (2.34 g, 10.7 mmol) and NiCl2
(695 mg, 5.36 mmol) were added to a solution of bisnitrile 17 (1.0 g,
2.68 mmol) in THF (10 mL) and MeOH (15 mL). Sodium borohydride
(1.42 g, 38.0 mmol) was carefully added at 08C and the mixture was stir-
red at room temperature for 16 h. The solvent was removed in vacuo and
the precipitate dissolved in ethyl acetate (20 mL) and saturated aqueous
NaHCO3 (10 mL). The insoluble precipitate was filtered off and washed
with ethyl acetate (2”10 mL). The combined organic layers were dried
over MgSO4 and evaporated in vacuo to afford a white solid. The crude
material was purified by column chromatography (2% MeOH/CHCl3) to
yield dicarbamate 18 as a white solid (925 mg, 91%). Rf =0.58 (ethyl ace-
tate); m.p. 101–1028C; [a]D22=58.1 (c=0.5, CHCl3); 1H NMR (400 MHz,
CDCl3): d = 7.64 (s, 2H, ArH), 7.58 (d, 2H, J=8 Hz, ArH), 7.41–7.30
(m, 4H, ArH), 6.75 (brs, 2H, NH), 5.00 (brs, 2H, NH), 4.28 (d, 4H, J=
6 Hz, CH2N), 3.99 (m, 2H, CHNH), 2.21 (m, 2H), 1.83 (m, 2H), 1.43 (m,
22H); 13C NMR (100 MHz, CDCl3): d = 168.1 (0), 167.9 (0), 139.7 (0),
134.6 (0), 130.6 (1), 128.9 (1), 126.1 (1), 125.8 (1), 79.8 (0), 54.7 (1), 44.5
(2), 32.5 (2), 28.5 (2), 24.9 (3); IR (neat): nmax =3303 (b), 2981 (w), 2927
(w), 2852 (w), 1696 (s), 1631 cmꢀ1 (s); MS ES+: m/z: 581 [M+H]+, 603
[M+Na]+; HRMS (ES+): m/z: calcd for C32H4N4NaO6: 603.3153, found
603.3148 [M+Na]+.
Since there is no GB/SA parametrization available within MacroModel
for the solvents DMSO and CH3CN, for studies of macrocycle 1 in these
solvents explicit solvation models were used. The macrocycle was there-
fore imbedded in previously equilibrated cubic boxes containing either
512 DMSO or CH3CN solvent molecules with experimental densities. For
CH3CN, fully flexible six-site all-atom OPLS-AA(2001)[26] CH3CN
3
models were equilibrated in a 35.42”35.42”35.42 Š dimensional box
over a series of simulations and temperatures first in the isothermal–iso-
choric (NVT) ensemble (T=50, 150, 225 and 298.15 K) and then the iso-
thermal-isobaric (NPT) ensemble (T=298.15 K) so that the experimental
density[27] of acetonitrile at 298.15 K and 1 atm pressure (0.7857 gcmꢀ3
)
3
was close to reproduced. In the case of DMSO, a 39.28”39.28”39.28 Š
[27]
box was used corresponding to an experimental density of 1.096 gcmꢀ3
.
Final boxes for each solvent were within 2% of the experimental densi-
ties. Two starting conformations for macrocycle 1 were used for the simu-
lations: the C2 conformation found as the most stable comformation in
CHCl3; and a more planar open type conformation where the intramolec-
ular hydrogen bonds for the C2 conformation did not have to be broken.
Overlapping solvent molecules and those giving rise to repulsive interac-
tions with macrocycle were removed and short minimizations (1000 steps,
steepest descent algorithm) performed to remove bad contacts. MD sim-
ulations by using periodic boundary conditions (PBC) and the Verlet al-
gorithm in the isothermal-isobaric ensemble (NPT) were subsequently
performed (298.15 K, 1 atm). A 30 ps equilibration period was followed
by a 60 ps data collection phase using a timestep of 1 fs. The OPLS-AA-
(2001) force field[26] was used due to the unavailability of MMFFs for use
with Impact 2.7.[17] Conformational preferences of the macrocycle were
analyzed from the resulting MD trajectories for the data collection
phase. For both solvents a D2 type conformation located starting from
the open “unwrapped” conformation was found to be energetically more
stable. The C2 conformation remained close to its wrapped conformation
over the time period of the simulations—which in DMSO and CH3CN is
unfavorable compared to having D2 symmetry. Finally, as stated MD sim-
ulations were performed using Impact 2.7.[17]
Diamine 19: Dicarbamate 18 (500 mg, 0.86 mmol) was dissolved in 1:1
solution of TFA and CH2Cl2 (10 mL). The solution was stirred at room
temperature for 2 h. Toluene (10 mL) was added and the solvent evapo-
rated in vacuo to give a clear oil. Trituration of the oil with diethyl ether
provided the trifluoroacetate salt of diamine 19 as a white solid (450 mg,
86%). M.p. 1708C (decomp); [a]D22
=
61 (c=1.0, DMSO); 1H NMR
(400 MHz, [D6]DMSO): d = 8.26 (brs, 2H, NHCO), 7.86 (s, 2H, ArH),
7.71 (d, 2H, J=8 Hz, ArH), 7.54 (d, 2H, J=8 Hz, ArH), 7.45 (m, 2H,
ArH), 4.47 (brs, 6H, NH3+), 4.03 (brs, 4H, CH2NH3+), 3.95 (m, 2H,
CHNH), 1.90 (m, 2H), 1.70 (m, 2H), 1.52 (m, 2H), 1.30 (m, 2H);
13C NMR (100 MHz, [D6]DMSO): d = 165.6 (0), 134.8 (0), 133.7 (0),
130.9 (1), 128.1 (1), 127.7 (1), 126.4 (1), 52.6 (1), 41.7 (2), 31.2 (2), 24.2
(2); IR (neat): nmax =3290 (m), 2935 (b), 1667 (m), 1635 (s), 1533 (s),
1179 (s), 1120 cmꢀ1 (s).
Macrocycle 4: Diamine 19 (200 mg, 0.53 mmol) was dissolved in CH2Cl2
(10 mL) and 0.5m aqueous K2CO3 added (5 mL). Thiophosgene (160 mL,
2.1 mmol) was added directly to the organic layer and the solution stirred
for 16 h. The organic phase was extracted and washed with 2m HCl
(10 mL), dried over MgSO4 and the solvent evaporated in vacuo to give
bis(isothiocyanate) 20 as a pale yellowfoam. The bisisocyanate in dry
CH2Cl2 (5 mL) and a further one equivalent of diamine 15 (200 mg,
0.53 mmol) in dry CH2Cl2 (5 mL) were simultaneously added over 3 h via
syringe pump addition to a solution of triethylamine (183 mL, 1.3 mmol)
in dry CH2Cl2 (50 mL) under a slowstream of nitrogen. The mixture was
stirred at room temperature for 16 h. Macrocycle 4 precipitated out of so-
lution as a pale yellowsolid and purified by semi-preparative RP-HPLC
(127 mg, 29%). M.p. 228–2298C; [a]2D2 = 104 (c=0.5, DMSO); 1H NMR
(400 MHz, [D6]DMSO): d = 8.20 (d, 4H, J=7 Hz NHCO), 7.92 (brs,
4H, NHCS), 7.66 (m, 4H, ArH), 7.33 (brs, 8H, ArH), 4.63 (s, 8H,
CH2NH), 3.94 (m, 4H), 1.91 (m, 4H), 1.75 (m, 4H), 1.53 (m, 4H), 1.31
(m, 4H); 13C NMR (100 MHz, [D6]DMSO): d = 167.0 (0), 162.8 (0),
137.5 (0), 134.8 (0), 128.0 (1), 126.5 (1), 125.4 (1), 120.8 (1), 52.9 (1), 35.7
(2), 30.7 (2), 24.6 (2); IR (neat): nmax =3276 (b), 3056 (w), 2983 (w), 2862
(w), 1636 (s), 1534 cmꢀ1 (s); MS MALDI in DMSO: m/z: 867 [M+Na]+;
elemental analysis calcd (%) for C46H52N8O4S2: C 65.38, H 6.20, N 13.25,
S 7.59; found C 65.07, H 5.87, N 13.02, S 7.15.
Solvent dependent binding preferences: For macrocycle 1 and 1:1 binding
of N-Boc-d,l-glutamate, binding in the two solvents CHCl3 and CH3CN
was computed as follows. Again for CHCl3, the GB/SA model was used
in a mixed mode MCMM/LMCS search. There were a number of differ-
ent conditions used to those for macrocycle 1 alone. Importantly, global
ligand translations and rotations were also included and were seen as cru-
cial to success of the conformational search. In total, 2 to 5 torsions were
adjusted each MCMM step, with a 1:1 ratio of MC torsional moves and/
or ligand translation to the lowmode moves. A MCSS usage-directed
structure selection option was used so that the least investigated structure
was used as the starting point for the next MC step. 10000 MCMM/
LMCS steps by using the MMFFs force field were performed with an
energy window of 25 kJmolꢀ1 above the global minimum used for saving
conformations. Computations were performed using MacroModel 8.1.
For binding in CH3CN, MD simulations using the Verlet algorithm were
performed in explicit solvent. Macrocycle 1 with bound enantiomer (l or
d N-Boc-glutamate) were introduced into the previously equilibrated (as
described above) cubic box containing 512 CH3CN molecules. Overlap-
ping solvent molecules and those giving rise to repulsive van der Waals
(vdW) interactions with the complex were then removed and short mini-
mizations (1000 steps, steepest descent algorithm) performed to remove
Computational details
Macrocycle solvent dependent conformational preferences: For macrocy-
cle 1 in CHCl3, to locate the lowest energy conformations, 5000 steps of
the mixed mode MCMM/LMCS conformational search algorithm were
performed. Within the MCMM/LMCS algorithm, a mixture of Monte
Carlo Multiple Mimima (MCMM) steps[23] or lowmode conformational
5686
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5674 – 5688

Enantioselective Receptors
FULL PAPER
any bad contacts. Minimizations were followed by NPT simulations (tem-
perature 298.15 K; pressure 1 atm) using PBC: 200 ps equilibration pre-
ceded 400 ps data collection phases using a timestep of 1 fs for both l
and d bound complexes. The OPLS-AA(2001) force field was used, with
a non-bond cutoff of 10 Š for van der Waals interactions and an Ewald
treatment for electrostatics.[28] For the binding enthalpies in CH3CN, DHL
can be calculated as follows:
[4] a) C. Miranda, F. Escartí, L. Lamarque, M. J. R. Yunta, P. Navarro,
E. García-EspaÇa, M. L. Jimeno, J. Am. Chem. Soc. 2004, 126, 823–
833; b) H. Ait-Haddou, S. L. Wiskur, V. M. Lynch, E. V. Anslyn, J.
Am. Chem. Soc. 2001, 123, 11296–11297; c) I. Alfonso, B. Dietrich,
F. Robolledo, V. Gotor, J.-M. Lehn, Helv. Chim. Acta 2001, 84, 280–
295; d) J. L. Sessler, A. Andrievsky, V. Krµl, V. Lynch, J. Am. Chem.
Soc. 1997, 119, 9385–9392.
[5] S. Rossi, G. M. Kyne, D. L. Turner, N. J. Wells, J. D. Kilburn, Angew.
Chem. 2002, 114, 4407–4409; Angew. Chem. Int. Ed. 2002, 41, 4233–
4236.
DHL ¼ HRꢀL ꢀ HR ꢀ HL
ð1Þ
[6] a) C. A. Hunter, D. H. Purvis, Angew. Chem. 1992, 104, 779–782;
Angew. Chem. Int. Ed. Engl. 1992, 31, 792–795; b) G. M. Kyne,
M. E. Light, M. B. Hursthouse, J. de Mendoza, J. D. Kilburn, J.
Chem. Soc. Perkin Trans. 1 2001, 1258–1263.
[7] S. Caddick, A. K. de K. Haynes, D. B. Judd, M. R. V. Williams, Tetra-
hedron Lett. 2000, 41, 3513–3516.
[8] P. Gꢂntert, C. Mumenthaler, K. Wꢂthrich, J. Mol. Biol. 1997, 273,
283–298; L. Brennan, D. L. Turner, A. C. Messias, M. L. Teodoro, J.
LeGall, H. Santos, A. V. Xavier, J. Mol. Biol. 2000, 298, 61–82. Full
details of experimental methods are provided in ref. [5] and associat-
ed electronic Supporting Information.
where HR–L corresponds to the average enthalpy of the complex of the
macrocyclic receptor with the l enantiomer from the MD simulations,
and HR and HL correspond to the enthalpies for the isolated macrocycle
and l enantiomer conformations, respectively. An analogous expression
exists for DHD. However, as the isolated enantiomeric energies are the
same (HL = HD), MD simulations of the receptor bound l and d com-
plexes in CH3CN were sufficient to compute DDHL–D (298.15 K):
DDHLꢀD ¼ HRꢀL ꢀ HR-D
ð2Þ
[9] A. P. Bisson, C. A. Hunter, J. C. Morales, K. Young, Chem. Eur. J.
1998, 4, 845–851; all binding data from NMR titration experiments
are provided in the electronic Supporting Information.
[10] K. A. Connors, Binding Constants, Wiley, NewYork, 1987.
[11] a) L. Fielding, Tetrahedron 2000, 56, 6151–6170; b) C. S. Wilcox,
Frontiers in Supramolecular Organic Chemistry and Photochemistry
(Eds.: H.-J. Schneider, H. Dꢂrr), VCH, Weinheim, 1991, pp. 123–
143; c) G. Weber, S. R. Anderson, Biochemistry 1965, 4, 1942–1947.
[12] A. Cooper, Curr. Opin. Chem. Biol. 1999, 3, 557–563, and referen-
ces therein.
[13] For the specific application of isothermal calorimetry to carboxylate
recognition, see: a) B. R. Linton, M. S. Goodman, E. Fan, S. A. va-
n Arman, A. D. Hamilton, J. Org. Chem. 2001, 66, 7313–7319; see
also, for example: b) S. L. Wiskur, J. J. Lavigne, A. Metzger, S. L.
Tobey, V. Lynch, E. V. Anslyn, Chem. Eur. J. 2004, 10, 3792–3804;
c) F. P. Schmidtchen, Org. Lett. 2002, 4, 431–434; d) B. Linton, A. D.
Hamilton, Tetrahedron 1999, 55, 6027–6038.
[14] The ITC data was obtained using a MicroCal VP-ITC isothermal
calorimeter. Data was analysed by nonlinear least square fitting by
using the Origin software supplied by MicroCal: T. Wiseman, S. Wil-
liston, J. F. Brandts, L.-N. Lin, Anal. Biochem. 1989, 179, 131–137;
all binding data from ITC experiments are provided in the electron-
ic Supporting Information.
Individual HR–L (and HR–D) values were summed from the following en-
thalpic components:
HRꢀL ¼ ½Hintra þ Hnon-b þ HsolvꢂLꢀcomplex
ð3Þ
where Hintra and Hnon-b are the intramolecular and non-bond interaction
energies within the l (or d) complexes and Hsolv is the solvation energy
of the complex. Simulations were performed by using Impact 2.7.
Finally, to extract the most representative and predominant type of con-
formations from the MD trajectories, conformations were superimposed
and clustered into groups of similar conformations according to root
mean square (RMS) distances between heavy atoms. A representative
conformation was obtained for each cluster. The representative confor-
mation from the largest cluster is the most important, and is reported
and discussed in the results. Clustering calculations were performed using
NMRclust 1.2.[29]
Acknowledgements
[15] J. C. Adrian, C. S. Wilcox, J. Am. Chem. Soc. 1991, 113, 678–680.
[16] W. C. Still, A. Tempczyk, R. C. Cawley, T. Hendrickson, J. Am.
Chem. Soc. 1990, 112, 6127–6129; J. Weiser, P. S. Shenkin, W. C.
Still, J. Comput. Chem. 1999, 20, 217–230, 586–596; J. Weiser, A. A.
Weiser, P. S. Shenkin, W. C. Still, J. Comput. Chem. 1998, 19, 797–
808.
[17] Impact, version 2.7, Schrçdinger, LLC, NewYork, NY 2003.
[18] F. Mohamadi, N. G. J. Richards, W. C. Guida, R. Liskamp, M.
Lipton, C. Caufield, G. Chang, T. Hendrickson, W. C. Still, J.
Comput. Chem. 1990, 11, 440–467; MacroModel versions 8.0 and
8.1, Schrçdinger, LLC, NewYork, NY, 2003.
[19] T. R. Kelly, M. H. Kim, J. Am. Chem. Soc. 1994, 116, 7072–7080.
[20] a) R. K. Castellano, F. Diederich, E. A. Meyer, Angew. Chem. 2003,
115, 1244–1287; Angew. Chem. Int. Ed. 2003, 42, 1210–1250; b) M.
Nishio, M. Hirota, Y. Umezawa, The CH/p Interaction, Wiley, New
York, 1998; c) E. Kim, S. Paliwal, C. S. Wilcox, J. Am. Chem. Soc.
1998, 120, 11192–11193; d) S. Paliwal, S. Geib, C. S. Wilcox, J. Am.
Chem. Soc. 1994, 116, 4497–4498.
[21] D. H. Williams, E. Stephens, D. P. OꢁBrien, M. Zhou, Angew. Chem.
2004, 116, 6760–6782; Angew. Chem. Int. Ed. 2004, 43, 6596–6616.
[22] a) K. Kavallieratos, C. M. Bertao, R. H. Crabtree, J. Org. Chem.
1999, 64, 1675–1683. See also: b) C. Schmuck, U. Machon, Chem.
Eur. J. 2005, 11, 1109–1118.
We thank the Commission of the European Union (TMR Network grant
“Enantioselective Separations” ERB FMRX-CT-98-0233) for financial
support, postgraduate fellowships (S.R. and A.R.) and postdoctoral fel-
lowships (J.M.H.). We are also grateful to Emilio Gallicchio for helpful
discussions and support with the Impact computations, and also to Schrç-
dinger Inc. for technical support.
[1] See for example, a) J. H. Hartley, T. D. James, C. J. Ward, J. Chem.
Soc. Perkin Trans. 1 2000, 3155–3184; b) M. W. Peczuh, A. D. Ham-
ilton, Chem. Rev. 2000, 100, 2479–2494; c) A. P. Davis, R. S. Ware-
ham, Angew. Chem. 1999, 111, 3160–3179; Angew. Chem. Int. Ed.
1999, 38, 2979–2996.
[2] C. S. Wilcox, Chem. Soc. Rev. 1993, 22, 383–395.
[3] For a review, see: a) R. J. Fitzmaurice, G. M. Kyne, D. Douheret,
J. D. Kilburn, J. Chem. Soc. Perkin Trans. 1 2002, 841–864; for
recent examples, see: b) C. Schmuck, M. Schwegmann, J. Am.
Chem. Soc. 2005, 127, 3373–3379; c) Y.-S. Zheng, C. Zhang, Org.
Lett. 2004, 6, 1189–1192; d) J. V. Hernµndez, A. I. Oliva, L. Simón,
F. M. Muniz, A. A. Mateos, J. R. Móran, J. Org. Chem. 2003, 68,
7513–7516; e) J. S. Disch, R. J. Staples, T. E. Concolino, T. E. Haas,
E. V. Rybak-Akimova, Inorg. Chem. 2003, 42, 6749–6763; f) J.
Raker, T. E. Glass, J. Org. Chem. 2002, 67, 6113–6116.
[23] G. Chang, W. C. Guida, W. C. Still, J. Am. Chem. Soc. 1989, 111,
4379–4386; M. Saunders, K. N. Houk, Y. D. Wu, W. C. Still, M.
Chem. Eur. J. 2005, 11, 5674 – 5688
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5687

J. D. Kilburn et al.
Lipton, G. Chang, W. C. Guida, J. Am. Chem. Soc. 1990, 112, 1419–
1427.
[27] www.chemfinder.com.
[28] M. P. Allen, D. J. Tildesley, Computer Simulation of Liquids, Claren-
don Press, NewYork, 1988; S. W. de Leeuw, J. W. Perram, E. R.
Smith, Proc. R. Soc. London Ser. A 1980, A373, 27–56.
[29] L. A. Kelley, S. P. Gardner, M. J. Sutcliffe, Protein Eng. 1996, 9,
1063–1065; http://neon.chem.le.ac.uk/nmr.clust/.
[24] I. Kolossvµry, W. C. Guida, J. Am. Chem. Soc. 1996, 118, 5011–5019.
[25] T. A. Halgren, J. Comput. Chem. 1996, 17, 490–519, 520–552, 553–
586, 616–641; T. A. Halgren, R. B. Nachbar, J. Comput. Chem. 1996,
17, 587–615; T. A. Halgren, J. Comput. Chem. 1999, 20, 720–729,
730–748.
[26] W. L. Jorgensen, D. S. Maxwell, J. Tirado-Rives, J. Am. Chem. Soc.
1996, 118, 11225–11236.
Received: April 20, 2005
Published online: July 20, 2005
5688
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5674 – 5688