Dynamic combinatorial chemistry with hydrazones: Cholate-based building blocks and libraries

Article abstract of DOI:10.1039/b917145b

We describe an efficient and general strategy for the synthesis of dimethyl acetal functionalised steroidal hydrazides based on the cholic acid skeleton with the aim of using these compounds as building blocks for dynamic combinatorial chemistry. Deprotection of the acetal protected building blocks with TFA leads to formation of libraries containing macrocyclic N-acyl hydrazone oligomers. The isolation of several of these, and their characterisation using NMR is described. The effects on the equilibrium library distribution by varying the substituents at C-7 and C-12, extending the side-chain with glycine, and inverting the configuration at C-3 are discussed. Finally, we report the exchange properties of these macrocycles and demonstrate new examples of proof-reading and self-sorting in dynamic combinatorial libraries. The Royal Society of Chemistry.

Full text of DOI:10.1039/b917145b

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Dynamic combinatorial chemistry with hydrazones: cholate-based building
blocks and libraries†
Mark G. Simpson,a Michael Pittelkow,*‡a Stephen P. Watsonb and Jeremy K. M. Sanders*a
Received 19th August 2009, Accepted 20th November 2009
First published as an Advance Article on the web 11th January 2010
DOI: 10.1039/b917145b
We describe an efficient and general strategy for the synthesis of dimethyl acetal functionalised steroidal
hydrazides based on the cholic acid skeleton with the aim of using these compounds as building blocks
for dynamic combinatorial chemistry. Deprotection of the acetal protected building blocks with TFA
leads to formation of libraries containing macrocyclic N-acyl hydrazone oligomers. The isolation of
several of these, and their characterisation using NMR is described. The effects on the equilibrium
library distribution by varying the substituents at C-7 and C-12, extending the side-chain with glycine,
and inverting the configuration at C-3 are discussed. Finally, we report the exchange properties of these
macrocycles and demonstrate new examples of proof-reading and self-sorting in dynamic
combinatorial libraries.
In our search for a mild alternative to transesterifcation, to
Introduction
permit the use of organic templates and foster weak non-covalent
The ability of a template molecule to direct the assembly of, or to
intermolecular interactions, we have been actively investigating the
select, its ‘ideal host’ by means of non-covalent interactions is fun-
trifluoroacetic acid catalysed exchange behaviour of macrocyclic
damental to the success of catalytic antibodies,1 ribozymes2 and
their DNA equivalents3 as well as certain imprinted polymers.4 In
terms of producing catalysts and molecular hosts capable of recog-
nition all have their attendant advantages and disadvantages.5
In the past decade, dynamic combinatorial chemistry has
emerged as a powerful new templating approach for the discovery
of unexpected supramolecular systems.6 In a dynamic combinato-
rial library (DCL), a series of receptors is allowed to interconvert
in solution through reversible bond-making and bond-breaking.
The reversibility gives the system the possibility to respond to the
addition of a template in such a way as to reveal stabilising non-
covalent intermolecular interactions. An equilibrium distribution
of receptors should be perturbed by the addition of a template that
binds selectively to, stabilises and hence favours the formation
of one particular receptor. The DCL can be screened in situ,
presenting a very attractive way of avoiding the laborious and
difficult task of designing molecular hosts or guests. Elegant
examples of template-directed assembly under thermodynamic
control have been reported, leading to the discovery of new
receptors and ligands utilising a variety of different reversible
chemistries.7–9
N-acyl hydrazones;11 the acid catalysed exchange of macrocyclic
imines was reported by Wild around the same time as this work
began.12 Historically, N-acyl hydrazones were for many years used
for isolating and characterizing liquid aldehydes and ketones. For
other applications, they have been largely neglected in the chemical
literature. This is surprising, given their appealing characteristics:
they are stable, easily accessible from carbonyl compounds and
hydrazides, andthey possess potentialpseudo-peptidic recognition
features (Fig. 1).13
The results reported herein build on our early work using
transesterification in DCLs, where we reported a series of
macrocyclic receptors based on cholate building blocks, whose
equilibria can be perturbed by the addition of alkali metal ions.10
aUniversity Chemical Laboratory, University of Cambridge, Lensfield
Road, Cambridge, CB2 1EW, United Kingdom. E-mail: jkms@cam.ac.uk;
Fax: +44 1223 336017; Tel: +44 1223 336411
bGlaxoSmithKline Medicines Research Centre, Gunnels Wood Road,
Stevenage, Hertfordshire SG1 2NY, United Kingdom
† Electronic supplementary information (ESI) available: Experimental and
analytical details for all compounds. See DOI: 10.1039/b917145b
Fig. 1 a) The reversible hydrazone reaction. b) General design of
cholate-derived building blocks.
‡ Present address: University of Copenhagen, Department of Chem-
istry, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark; E-mail:
pittel@kiku.dk
The pedigree of bile acids as supramolecular building
blocks for recognition has been well established. Davis14 and
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 1173–1180 | 1173
©

View Article Online
Bonar-Law and Sanders15,16 have shown that these molecules can
be functionalised and cyclised to form rigid cavities, capable of
recognising carbohydrates, alkaloids and metal ions, either alone,
or as part of other supramolecular architectures. The purpose of
the work described here was to equip a range of functionalised bile
acids with a dimethyl acetal and a hydrazide and study both the
range and nature of the macrocyclic N-acyl hydrazones formed
upon deprotection with a view to thermodynamic templating.
This work builds on our earlier studies of steroid16 and amino
acid-based17 macrocyclic N-acyl hydrazones.
monohydrate, and a concentrated solution of 5 in MeOH or EtOH
at room temperature.20 The reaction typically required 48 h to
reach completion. Heating was avoided, due to the presence of
the aromatic esters at the 3- and 12-positions of the steroid, which
react under forcing conditions.
The data in Scheme 2 summarises the pyridine-substituted
steroidal hydrazides that have been prepared by the general
reaction sequence depicted in Scheme 1, substituting reagents
where appropriate. Thus, the synthesis of 9 required the use
of pyridine-3-carboxylic acid, while 10 and 11 were synthesised
from chenodeoxocholic acid (8). In order to investigate the effect
of changing the geometry of a dimethyl acetal functionalised
hydrazide on the range of macrocyclic N-acyl hydrazones formed
at equilibrium 3-carboxybenzaldehyde was substituted for 4-
carboxybenzaldehyde in the synthesis of 12.
Results and discussion
Synthesis
Lithocholate, deoxycholate, chenodeoxycholate and cholate-
based dimethyl acetal functionalised hydrazides were obtained
by applying, with modifications where appropriate, the synthetic
methodology shown in Scheme 1. Methyl deoxycholate (2) was
synthesised, using the method of Fieser, in 85% yield from
deoxycholic acid (1).18
Scheme 2 Summary of pyridine-appended building blocks. Reagents and
conditions as in Scheme 1.
A second series of steroidal hydrazides, depicted in Scheme 3,
does not possess pyridine-3 or 4-carboxylate substituents at the
7- or 12-positions of the steroid. The aim with this sequence
of molecules was to leave free hydroxyl groups, to potentially
enable recognition and binding of polyhydroxylated guests, such
as carbohydrates.21 Again 4-carboxybenzaldehyde was substituted
for 3-carboxybenzaldehyde in the synthesis of 17, 18 and 19.
As well as being derived from deoxycholic (15 and 18), and
chenodeoxycholic acid (16), one monomer (19) was synthesised
from cholic acid. The lithocholate derivatives (14 and 17) were to
be used as control molecules.
Scheme 1 Synthesis of building blocks exemplified by the deoxy se-
ries. Reagents and conditions: a) MeOH, AcCl, room temperature.
b) 3-Carboxybenzaldehyde, EDC, DMAP, CH2Cl2. c) pTsOH, MeOH.
d) Pyridine-4-carboxylic acid, EDC, DMAP, CH2Cl2. e) NH2NH2, MeOH,
room temperature.
Selective esterification of the diol (2) at the 3-position
was achieved by means of EDC mediated coupling of 3-
carboxybenzaldehyde in CH2Cl2, with a catalytic amount of
DMAP, to form the ester (3) in 85% yield.19 The aldehyde (3)
was protected as its dimethyl acetal (4) by means of a catalytic
amount of pTsOH in MeOH. Coupling of pyridine-4-carboxylic
acid at the 12-position of 4 to form the ester (5), in 90% yield,
was also achieved by means of EDC in CH2Cl2. This required
a stoichiometric amount of DMAP, however, compared to the
catalytic amount sufficient for coupling at the 3-position. Efficient
and selective hydrazinolysis of the methyl ester (5) to form the
hydrazide (6) in 90% yield required 30 equivalents of hydrazine
Scheme 3 Second series of building blocks. Reagents and conditions: See
Scheme 1.
To further investigate the effect of changing the geometry and
flexibility of a dimethyl acetal functionalised steroidal hydrazide
on the range of macrocyclic N-acyl hydrazones formed at equi-
librium the glycine extended steroidal hydrazide (21) shown in
1174 | Org. Biomol. Chem., 2010, 8, 1173–1180
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
The final variation in structure was to invert the configuration at
the 3-position of deoxycholic acid (Scheme 6). This was achieved
by the previously reported method by Fiegel et al.22 Mitsunobu
coupling of formic acid with the alcohol (2) using DIAD and PPh3
in THF led to the 3b formate ester in 85% yield.23
Scheme 4 Synthesis of extended steroid building blocks. Reagents and
conditions: a) Glycine methyl ester, EDC, DMAP, Et3N, room tem-
perature. b) 3-Carboxybenzaldehyde, EDC, DMAP, CH2Cl2. c) pTsOH,
MeOH. d) NH2NH2, MeOH, room temperature.
Scheme 6 Synthesis of steroid building block with inverted stereochem-
istry. Reagents and conditions: a) MeOH, AcCl, room temperature.
b) Formic acid, DIAD, PPh3, THF. c) Na, MeOH, room temperature.
d) 3-Carboxybenzaldehyde, EDC, DMAP, CH2Cl2. e) pTsOH, MeOH.
f) Pyridine-carboxylic acid, EDC, DMAP, CH2Cl2. g) NH2NH2, MeOH,
room temperature.
Scheme 4 was synthesised. The methyl ester (20) was obtained
in 95% yield following EDC mediated coupling of glycine methyl
ester hydrochloride with 1 in CH2Cl2, using a catalytic amount
of DMAP and Et3N. The alcohol (20) was selectively esterified
at the 3-position with 3-carboxybenzaldehyde by means of EDC
and a catalytic amount of DMAP in CH2Cl2. The aldehyde was
protected as the dimethyl acetal by treatment with a catalytic
amount of pTsOH in MeOH. Finally hydrazinolysis of the methyl
ester with 20 equivalents of hydrazine monohydrate in MeOH,
afforded the hydrazide (21) in 56% overall yield. The pyridine-
appended extended building block (23), was prepared in 43%
overall yield from the carboxylic acid (1) by the closely-related
route summarized in Scheme 5.
Removal of the formate group was achieved using a freshly
prepared solution of sodium methoxide to afford the alcohol with
the stereocentre reversed (24) in 75% yield from the carboxylic
acid (1). The subsequent steps were identical to those employed in
the synthesis of the hydrazide (6) leading to the hydrazide (25) in
25% overall yield from the ester (2).
Cyclisation studies
In order to favour the formation of macrocyclic N-acyl hydrazones
over linear polymers, deprotection of the dimethyl acetal function-
alised hydrazides was conducted under high dilution conditions.24
Thus, a 5 mM solution of the monomer in CH2Cl2, CHCl3 or
DMSO was treated with 5% TFA v/v at room temperature. The
acid serves to remove the dimethyl acetal protecting group and
also to catalyse formation of the macrocyclic N-acyl hydrazones.
In CH2Cl2 and CHCl3, deprotection of the dimethyl acetal and
cyclisation to form the N-acyl hydrazones were complete in a
matter of minutes, while the reaction in DMSO had to be stirred
at room temperature for 24 h before all of the starting material had
been consumed. For the purposes of recognition and isolation of
macrocycles we were more interested in, and so focused upon,
deprotection in the chlorinated solvents.
Analysis of the reaction mixtures, directly by reverse phase
HPLC and ESI-MS, indicated that the cyclic N-acyl hydrazone
dimers were the dominant products (>95% yield) from the de-
protection of the meta substituted dimethyl acetal functionalised
hydrazides (6, 9, 10, 11, 14, 15 and 16). Only trace amounts
of higher macrocycles, trimer and tetramer, could be detected
by ESI-MS. Increasing the concentration of the dimethyl acetal
functionalised steroidal hydrazide from 5 to 20 mM did not affect
Scheme 5 Synthesis of extended steroid building blocks. Reagents and
conditions: a) Benzyl alcohol, EDC, DMAP. b) 3-Carboxybenzaldehyde,
EDC, DMAP, CH2Cl2. c) pTSOH, MeOH. d) Pyridine-4-carboxylic acid,
EDC, DMAP, CH2Cl2. e) Pd/C 10%, H2. f) Glycine methyl ester, EDC,
DMAP, Et3N. g) NH2NH2, MeOH, room temperature.
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 1173–1180 | 1175
©

View Article Online
this distribution to any significant extent, suggesting that the shape
of these monomers is such that they are strongly predisposed
towards the formation of cyclic dimer.25
The reaction mixtures derived from the hydrazides 15 and 6
were neutralised with Et3N and subjected to preparative thin layer
chromatography. The pure N-acyl hydrazone macrocyclic dimers
26 and 27 (Fig. 2 and 3) were isolated in high yields (~90%), and
were soluble in acid free CHCl3 to a concentration of 20 mM.26
N-acyl hydrazone resonance, at 9.34 ppm, and the azamethine
proton, Hm, at 7.55 ppm. Further nOe cross-peaks could be
observed between Hm and Hh, at 8.08 ppm. This strongly supports
a cis E geometry about the N-acyl hydrazone. That the hydrazone
NH, is not directed into the cavity would seem to be supported
by the lack of any nOe cross-peaks between the N-acyl hydrazone
NH and the steroid 22-H or 23-H resonances.
While the 1H NMR spectrum of 27 in CDCl3 was also simple,
it differed from that of 26 in two noticeable respects (Fig. 3).
Thus, the N-acyl hydrazone NH resonance of 27 was observed
at 10.85 ppm, compared to 9.34 ppm in 26, and the azamethine
proton, Hq in 27 appeared at 8.62 ppm, compared to 7.55 ppm in
26. The fact that the presence of the pyridine-4-carboxylate at the
12-position of the steroid could stabilise an otherwise disfavoured
conformation was intriguing: the NOESY spectrum revealed
no nOe cross-peak between the N-acyl hydrazone resonance at
10.85 ppm and the azamethine proton, Hq, at 8.62 ppm, but a
cross-peak was observed between Hq and Hh, at 7.87 ppm and
also between the N-acyl hydrazone NH and the 23-H steroid
resonances, suggesting the presence of a trans E N-acyl hydrazone
geometry. A weak nOe cross-peak could also be observed between
the N-acyl hydrazone NH and the a-pyridyl proton of the pyridine
substituent, at 8.65 ppm. This would seem to confirm that
the pyridine substituent stabilises the trans E N-acyl hydrazone
conformation. The NOESY spectrum also contains exchange
peaks, indicating an alternative conformation populated to a lesser
extent. A CPK model of 27 suggests that a pair of intramolecular
hydrogen bonds, between the pyridine substituents and N-acyl
hydrazone trans NH could account for the stabilisation of this
conformation, hence the downfield shift (Dd = -1.5 ppm) of this
proton, compared to the N-acyl hydrazone cis NH in 26.
Fig. 2 Structure of cyclic dimer 26.
The 1H NMR spectrum of 26 (Fig. 2) in CDCl3 was surprisingly
simple, given that one might expect conformational or geomet-
rical isomerism about the N-acyl hydrazones. That cyclisation
had indeed taken place could be readily ascertained from the
disappearance of the methoxy methyl groups of 15 at 3.34 ppm,
the dimethyl acetal CH singlet at 5.43 ppm, the hydrazide
NH resonance, at 6.84 ppm, and hydrazide NH2 resonance, at
3.90 ppm, as well as the appearance of new signals in 26 due to the
azamethine proton, Hm at 7.55 ppm and the N-acyl hydrazone
NH signal at 9.34 ppm. Other shifts, could also be observed
in the aromatic and steroid regions of 26 compared with 15.
Particularly noticeable was the emergence and separation of the
two diastereotopic 23-H protons from the forest of steroid peaks
to 2.57 and 2.96 ppm.
N-Acyl hydrazones can exist in several possible arrangements
and geometrical isomers: cis E, cis Z, trans E or trans Z.27 The
formation of rigid macrocycles, especially under equilibrating
conditions, might be expected to favour the most stable geometry
or arrangement; in this case probably an E arrangement about the
N-acyl hydrazone NH. The NOESY spectrum of 26 in CDCl3 at
room temperature revealed strong nOe cross-peaks between the
In order for a template to perturb the equilibrium of a DCL,
a range of host structures must be present for templates to
distinguish between. DCLs containing 26 and 27 do not satisfy this
requirement due to their low diversity: distribution of macrocycles
was needed. A relatively simple way to change the geometry of the
monomers was to switch the aldehyde from meta to para.
When a 5 mM solution of the hydrazide building block (12) in
CH2Cl2 was deprotected using 5% TFA v/v at room temperature,
direct analysis of the reaction mixture by HPLC and ESI-MS
revealed the presence of the cyclic dimer (28, Fig. 4) and the
cyclic trimer (29, Fig. 5), in a 3 : 1 ratio after equilibrium was
reached, typically in 4–6 h. No 12 or linear, uncyclised species
were observed. Increasing the concentration of 12 led to increasing
amounts of trimer and above 50 mM some cyclic tetramer, as
expected.28 This was an encouraging result and interesting because
the overall flexibility of the molecule had not been changed,
in the sense of increasing the number of rotatable bonds, merely
the direction in space of the aldehyde. The cyclic dimer (28) and
the cyclic trimer (29) were readily separated by preparative thin
layer chromatography to afford reasonable quantities of the pure
macrocycles.
The 1H NMR spectrum of the cyclic dimer (28, Fig. 4) in CDCl3
was quite different from that of its regioisomer (27, Fig. 3). As
well as the usual downfield shifts of the 23-H steroid protons in
28 compared to 12, a considerable upfield shift (Dd = -0.54 ppm)
of the a-pyridyl and b-pyridyl (Dd = -0.20 ppm) protons of 28
compared to 12 as well as the Hd,e proton was observed. This
should be compared with an upfield shift of Dd = -0.17 ppm for
Fig. 3 Structure of cyclic dimer 27.
1176 | Org. Biomol. Chem., 2010, 8, 1173–1180
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
X-ray crystal structure of 28, obtained by slow evaporation of a
solution of 28 in CD2Cl2, revealed the geometry of the N-acyl
hydrazone to be cis E.16
1
The H NMR spectrum of the macrocyclic N-acyl hydrazone
trimer (29, Fig. 5) also derived from the hydrazide 12, differs from
that of the cyclic dimer (28) in all of the key resonances. The
azamethine proton, Hm at 7.86 ppm and the N-acyl hydrazone
NH are downfield shifted compared with 12. In this case the
NH proton is extremely downfield shifted to 11.10 ppm. Also
noticeable was the absence of the large upfield shift in the a and
b-pyridyl protons in 29, Dd = +0.09 ppm and Dd = +0.05 ppm
respectively, compared to 28. While the downfield shift of the N-
acyl hydrazone NH suggests the presence of hydrogen-bonding,
COSY and NOESY experiments on a solution of the cyclic trimer
(29) in CDCl3 support the presence of a cis E geometry. Thus,
an nOe cross-peak was observed between the N-acyl hydrazone
NH, at 11.10 ppm and the azamething proton, Hm, at 7.64 ppm.
Furthermore an nOe cross-peak was observed between Hm and
He, at 7.70 ppm. The reason for the downfield shift of the N-acyl
hydrazone NH was not clear, but it may be due to intermolecular
hydrogen-bonding between two, or more, different trimer units.29
If it were true that simply changing the geometry of the aldehyde
at the 3-position from meta to para were responsible for increasing
the stability of the cyclic trimer compared to the cyclic dimer one
might expect that deprotection of the para substituted monomer
18 would result in a similar product distribution to that obtained
from the deprotection of 12. This was not found, however, to be
the case: deprotection of a 5 mM solution of 18 in CH2Cl2 led
initially to a large distribution of cyclic products (dimer, trimer
and tetramer), by HPLC, but over a period of several hours the
amounts of trimer and tetramer decreased. At equilibrium the
ratio of dimer to trimer was found to be 8 : 1, with no significant
amounts of the tetramer. The dimer and trimer were separated
by preparative chromatography, allowing ESI-MS analysis and
Fig. 4 Structure of cyclic dimer 28.
1
confirmation of the HPLC assignments. The H NMR spectra
of the cyclic dimer and trimer could be recorded, but their poor
solubility in CDCl3 required the addition of d4-MeOH, depriving
us of the valuable structural information supplied by the N-acyl
hydrazone NH signal. Increasing the concentration of 18, from
5 mM to 10 mM, led to a cyclic dimer to trimer ratio of 6 : 1.
The results of deprotecting the para substituted dimethyl acetal
functionalised hydrazides, 17 and 19, seem to confirm that the
presence of pyridine-4-carboxylate substituent at the 12-position
of the steroid controls the distribution of the macrocycles formed
on deprotection and, in addition, increases the solubility of the
macrocycles in CDCl3. Thus, deprotection of a 5 mM solution of
17 in CH2Cl2, using the standard conditions of 5% TFA v/v led
to a single macrocyclic product, the dimer, by HPLC and ESI-
MS. The solubility of the neutral macrocycle, derived from 17, in
CDCl3 was initially good, but after several minutes a precipitate
formed, leading to broad signals in the 1H NMR spectrum.
The N-acyl hydrazones derived from 19 were designed to afford
a platform for templating experiments with polyhydroxylated
molecules in CHCl3. However, addition of TFA to a 5 mM solution
of 19 in CH2Cl2 and CHCl3, led to the formation of a precipitate
after an hour. Filtration of the precipitate and analysis of the
filtrate revealed the presence of only a single macrocycle, which
does not satisfy the diversity requirements for DCLs. Carrying
out the deprotection in DMSO did not lead to the formation of a
Fig. 5 Structure of cyclic trimer 29.
the a-pyridyl protons and Dd = 0 ppm for b-pyridyl protons of 27
compared to the hydrazide 6 and may be indicative of p–p stacking
of the two pyridine substituents in 28; this is less favourable in 27,
due to the meta substituted aldehyde substituent at the 3-position.
The lowfield azamethine and N-acyl hydrazone resonances of 28
compared to 27, suggest the N-acyl hydrazone adopts a cis E
geometry in 28.
The arrangement of the cyclic hydrazone was confirmed by the
NOESY spectrum of 28 in CDCl3 at room temperature. A strong
nOe cross-peak was observed between the N-acyl hydrazone NH,
at 9.17 ppm, and the azamethine proton, Hm, at 7.64 ppm. Also
an nOe cross-peak was observed between the azamethine proton,
Hm, and He, at 7.49 ppm. That the N-acyl hydrazone NH was
not directed into the cavity was deduced from the lack of any
significant nOe cross-peak between the N-acyl hydrazone NH and
the 23-H steroid protons. The change of geometry from meta
(27) to para (28) at the 3-position renders the trans E N-acyl
hydrazone geometry unfavourable compared to the cis E geometry.
The solution phase structure was also found in the solid state. An
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 1173–1180 | 1177
©

View Article Online
precipitate, however, still yielded only a single macrocyclic product
(dimer).
purposes of templating in chlorinated solvents we must therefore
address these design issues and incorporate these lessons into
future molecules.
In an attempt to understand the effect of flexibility on product
distribution further the meta substituted glycine extended dimethyl
acetal functionalised hydrazides (21 and 23) were studied. How-
ever, deprotecting 21 and 23 at a concentration of 5 mM in CH2Cl2
again led to the cyclic dimer as major product (>90%), with cyclic
trimer making up the remainder of the product. Increasing the
concentration of 21 and 23 to 20 mM did not increase the diversity
of reaction products significantly.
The final modification to the steroid structure, that is, inversion
at the 3-position of the steroid to form the meta substituted
dimethyl acetal functionalised hydrazide (25), was examined in
an attempt to increase the diversity of macrocycles at equilibrium.
In the initial distribution of products identified by HPLC and
ESI-MS a range of macrocyclic N-acyl hydrazones from cyclic
dimer to pentamer were observed. This, however, proved to be only
an ephemeral, kinetic, distribution and after 24 h the distribution
of products collapsed leaving the cyclic N-acyl hydrazone dimer
(30) as the only detectable product at equilibrium (Fig. 6).
Demonstration of reversibility and exchange experiments
In order to demonstrate that the distribution of macrocycles
is at equilibrium it was necessary to show that the formation
of macrocyclic steroidal N-acyl hydrazones is reversible. There
are two relatively simple experiments that accomplish this. The
most satisfying is to take a library solution, isolate one of the
macrocycles and show that this, under the reaction conditions,
gives rise to the original library distribution. Indeed, when a
2.5 mM solution of pure 28 in CH2Cl2 was treated with 5% TFA
v/v. HPLC and ESI-MS indicated the presence of both 28 and 29
in a 3 : 1 ratio after 8 h. When this experiment was repeated using
a 1.7 mM solution of pure 29 in CH2Cl2 a 3 : 1 ratio of 28 and
29 was observed by HPLC and ESI-MS after 6 h (Fig. 7). This
confirms both the reversibility of steroidal macrocyclic N-acyl
hydrazone formation and that the product distribution is under
thermodynamic control.
Fig. 6 Structure of cyclic dimer 30.
1
The H NMR spectrum of 30 (obtained by preparative thin
layer chromatography) was, as expected, quite different from its
stereoisomer (27) in that the N-acyl hydrazone NH resonance
appears at 9.58 ppm in 30 and the azamethine proton at 7.79 ppm,
compared with 10.85 ppm and 8.62 ppm for the same resonances
in 27. The cis E geometry of 30 was confirmed by NOESY: a strong
nOe cross peak was observed between the N-acyl hydrazone NH
resonance at 9.58 ppm and the azamethine proton at 7.79 ppm.
In addition a strong nOe cross-peak could be observed between
the azamethine proton and Hh, at 7.50 ppm. The most dramatic
difference between 27 and 30 is the large upfield shift of Hg
(Dd = +0.84 ppm) in 30, attributed to the close proximity of
the 12-pyridine substituent in the constrained macrocycle. There
does not appear to be any evidence of p–p stacking between the
two 12-pyridine substituents in 30.
The range of structural modifications made to the basic bile
acid core has taught us about the factors that influence the
distribution of macrocyclic N-acyl hydrazones at equilibrium.
Thus, a meta substituted aldehyde leads to predominant formation
of macrocyclic N-acyl hydrazone dimers at equilibrium, regardless
of the stereochemistry at the 3-positon and the presence of a
glycine extension. A para substituted aldehyde on the other hand
enables significant formation of macrocyclic trimer, with the caveat
that a pyridine substituent must be present at the 12-position. This
also has the benefit of increasing the solubility of the macrocycle
in CH2Cl2 and CHCl3. If we wish to use these macrocycles for the
Fig. 7 Demonstration of thermodynamic control.
Alternatively, reversibility can be demonstrated by an exchange
experiment using two macrocycles that may be distinguished
analytically. If the formation of two macrocyclic dimers is re-
versible they should give rise to a statistical mixture of homo
and heterodimers in the presence of acid. In order to test this, a
series of experiments was carried out using several dimethyl acetal
functionalised hydrazides. The first experiment involved mixing
the preformed macrocycles and the second experiment involved
deprotecting the two dimethyl acetal functionalised hydrazide
monomers in situ. These experiments were compared to see, if
any, differences between kinetic and thermodynamic distributions
of products.
We were particularly interested to see if the kinetic stability of
macrocyclic N-acyl hydrazone dimers would prohibit ring opening
and exchange, as might, perhaps, be expected for such favoured
structures. If kinetic stability were the underlying cause then one
might expect to see no mixing of preformed cyclic dimers.
1178 | Org. Biomol. Chem., 2010, 8, 1173–1180
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
When the monomers are deprotected in situ, however, one would
expect to see the heterodimer and for this to be a stable entity. With
monomers 6 and 14 the results of the two mixing experiments
were surprising. No mixing of the preformed macrocyclic N-acyl
hydrazones derived from 6 and 14 was observed by HPLC or ESI-
MS after a period of seven days. This was taken as evidence of the
kinetic stability of the dimers. When 6 and 14 were deprotected
together under the same conditions, the mixed dimer could be
observed by HPLC and ESI-MS, but over a period of 24 h
it disappeared (Fig. 8). This surprising example of self-sorting
prompted a further series of experiments using other macrocyclic
N-acyl hydrazone dimers.
of the cyclic dimer 27 and the macrocyclic dimer derived from
14 as indicated by their 1H NMR spectra. These geometrical
differences could be responsible for the slight differences of the
monomers, thus disfavouring the heterodimer as compared to the
homodimers. This reasoning is limited, however, to a consideration
of solutions of the neutral macrocycles and the exchange behaviour
is observed in the presence of TFA. One can envisage both kinetic
and thermodynamic grounds for the differentially substituted
monomers refusing to mix, but evidence in favour of either was
not available from these experiments.
A further example of self-sorting could be observed by mixing
the monomers 14 and 25. Upon addition of TFA (5% v/v) to
a 2.5 mM solution of 14 and 25 in CH2Cl2, analysis by HPLC
and ESI-MS indicated the formation of several mixed products
as well as the higher oligomers of monomer 25. Over a period of
24 h, however, these mixed and higher macrocycles were proof-
read, to leave just the two homodimers, derived from 14 and 25,
by HPLC and ESI-MS. The phenomenon of self-sorting seems to
be connected with the presence of a 4-pyridine substituent in the
12-position of the steroid.
Conclusion
We have synthesised a series of steroid based building blocks for
hydrazone based dynamic combinatorial chemistry containing
a hydrazide and a dimethylacetal protected aldehyde. Upon
treatment of a building block in organic solvents with TFA the
dominant products were cyclic dimers at thermodynamic equilib-
rium. Larger macrocycles were present as kinetic intermediates,
and macrocycles with different conformations tend to self-sort
rather than mix. The use of these macrocycles for the purpose
of templating in chlorinated solvents rely on careful design of
the building blocks. It may be important to avoid self-templating
motifs in the building blocks, however, this is an issue that can be
difficult to predict.
Fig. 8 Proof-reading in DCLs.
When equimolar 5 mM solutions of the two macrocyclic N-acyl
hydrazone dimers derived from 9 and 14, were mixed at room
temperature in the presence of 5% TFA v/v the formation of
heterodimer could be observed by both HPLC and ESI-MS over a
period of seven days. The slow nature of the reaction was, in part,
to be expected due to the constrained nature of the macrocycles
and the bias towards ring closure over a second ring opening.
Confirmation of this result was obtained by deprotecting 9 and
14 together at the same concentration. A statistical mixture of the
three possible products was observed and, significantly, did not
change over a period of seven days (Fig. 9). Further experiments
using the monomers 14 and 16 also indicated that exchange of
macrocyclic N-acyl hydrazones is a feasible process, occurring
over a period of several days. Mixing of 10 and 14 did not lead to
self-sorting.
Acknowledgements
We thank BBSRC and The Danish Natural Science Research
Council for financial support and GlaxoWellcome for a CASE
award (MGS).
Notes and references
1 N. R. Thomas, Nat. Prod. Rep., 1996, 13, 479–511.
2 J. R. Lorsch and J. W. Szostak, Acc. Chem. Res., 1996, 29, 103–110.
3 B. Seelig and A. Jaschke, Chem. Biol., 1999, 6, 167–176.
4 G. Wulff, Angew. Chem., Int. Ed. Engl., 1995, 34, 1812–1832.
5 P. A. Brady and J. K. M. Sanders, Chem. Soc. Rev., 1997, 26, 327–336.
6 (a) P. T. Corbett, J. Leclaire, L. Vial, J.-L. Wietor, K. R. West, J. K. M.
Sanders and S. Otto, Chem. Rev., 2006, 106, 3652–3711; (b) J.-M. Lehn,
Chem. Soc. Rev., 2007, 36, 151–160; (c) S. Ladame, Org. Biomol. Chem.,
2008, 6, 219–226; (d) R. F. Ludlow and S. Otto, Chem. Soc. Rev., 2008,
37, 101–108.
7 R. T. S. Lam, A. Belenguer, S. L. Roberts, C. Naumann, T. Jarrosson,
S. Otto and J. K. M. Sanders, Science, 2005, 308, 667–669.
8 J.-L. Wietor, G. D. Pantos and J. K. M. Sanders, Angew. Chem., Int.
Ed., 2008, 47, 2689–2692.
9 (a) L. Vial, R. F. Ludlow, J. Leclaire, R. Pe´rez-Ferna´ndez and S. Otto,
J. Am. Chem. Soc., 2006, 128, 10253–10247; (b) R. Pe´rez-Ferna´ndez,
M. Pittelkow, A. M. Belenguer, L. Lane, C. V. Robinson and J. K. M.
Sanders, Chem. Commun., 2009, 3708–3710.
Fig. 9 Demonstration of thermodynamic control, without proof-reading.
This behaviour was unexpected. There does not appear to be any
obvious reason why two steroids which both form macrocyclic
dimers should not mix together to form a heterodimer. On the
other hand there are minor differences in the conformations
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 1173–1180 | 1179
©

View Article Online
10 P. A. Brady and J. K. M. Sanders, J. Chem. Soc., Perkin Trans. 1, 1997,
3237–3253.
11 For an early example, see: G. R. L. Cousins, S.-A. Poulsen and J. K. M.
Sanders, Chem. Commun., 1999, 1575–1576.
12 D. C. R. Hockless, L. F. Lindoy, G. F. Swiggers and S. B. Wild, J. Chem.
Soc., Perkin Trans. 1, 1998, 117–122.
2007, 3092–3093; (i) S. M. Voshell, S. J. Lee and M. R. Gagne´, J. Am.
Chem. Soc., 2006, 128, 12422–12423.
18 L. F. Fieser and S. Rajagopalan, J. Am. Chem. Soc., 1950, 72, 5530–
5536.
19 J. C. Sheehan and S. L. Ledis, J. Am. Chem. Soc., 1973, 95, 875–
879.
13 L. Spialter, D. H. O’Brien, G. L. Unterreiner and W. A. Rush, J. Org.
Chem., 1965, 30, 3278–3283.
20 M. M. Lenman, A. Lewis and D. Gani, J. Chem. Soc., Perkin Trans. 1,
1997, 2297–2311.
14 A. P. Davis, Chem. Soc. Rev., 1993, 22, 243–253.
21 (a) K. M. Bhattarai, R. P. Bonar-Law, A. P. Davis and B. A. Murray,
J. Chem. Soc., Chem. Commun., 1992, 752–754; (b) R. P. Bonar-Law
and J. K. M. Sanders, J. Am. Chem. Soc., 1995, 117, 259–271.
22 D. Albert and M. Fiegel, Helv. Chim. Acta, 1997, 80, 2168–2181.
23 O. Mitsunobu, Synthesis, 1981, 1.
15 (a) R. P. Bonar-Law and J. K. M. Sanders, J. Chem. Soc., Perkin Trans.
1, 1995, 3085–3095; (b) R. P. Bonar-Law and J. K. M. Sanders, J. Am.
Chem. Soc., 1995, 117, 259–271.
16 M. G. Simpson, S. P. Watson, N. Feeder, J. E. Davies and J. K. M.
Sanders, Org. Lett., 2000, 2, 1435–1438.
17 (a) J. Liu, K. R. West, C. R. Bondy and J. K. M. Sanders, Org. Biomol.
Chem., 2007, 5, 778–786; (b) R. F. Ludlow, J. Liu, H. Li, S. L. Roberts,
J K. M. Sanders and S Otto, Angew. Chem., Int. Ed., 2007, 46, 5762–
5764; (c) S. L. Roberts, R. L. E. Furlan, G. R. L. Cousins and J. K. M.
Sanders, Chem. Commun., 2002, 938–939; (d) R. L. E. Furlan, Y.-F.
Ng, G. R. L. Cousins, J. E. Redman and J. K. M. Sanders, Tetrahedron,
2002, 58, 771–778; (e) R. L. E. Furlan, Y.-F. Ng, S. Otto and J. K. M.
Sanders, J. Am. Chem. Soc., 2001, 123, 8876–8877; (f) G. R. L. Cousins,
R. L. E. Furlan, Y.-F. Ng, J. E. Redman and J. K. M. Sanders, Angew.
Chem., Int. Ed., 2001, 40, 423–428; (g) R. L. E. Furlan, G. R. L. Cousins
and J. K. M. Sanders, Chem. Commun., 2000, 1761–1762; (h) F. Bulos,
S. L. Roberts, R. L. E. Furlan and J. K. M. Sanders, Chem. Commun.,
24 H. Jacobsen and W. H. Stockmayer, J. Chem. Phys., 1950, 18, 1600–
1606.
25 S. J. Rowan, P. A. Brady and J. K. M. Sanders, Angew. Chem., Int. Ed.
Engl., 1996, 35, 2143–2145.
26 The solubility of macrocyclic N-acyl hydrazones is often rather poor in
chlorinated solvents. This is especially so when the solution is neutral.
The addition of TFA often helps to keep the macrocycles in solution.
27 G. Palla, G. Predieri, P. Domiano, C. Vignali and W. Turner, Tetrahe-
dron, 1986, 42, 3649–3654.
28 G. Ercolani, L. Mandolini, P. Mencarelli and S. Roelens, J. Am. Chem.
Soc., 1993, 115, 3901.
29 R. P. Bonar-Law and J. K. M. Sanders, Tetrahedron Lett., 1993, 34,
1677–1680.
1180 | Org. Biomol. Chem., 2010, 8, 1173–1180
This journal is
The Royal Society of Chemistry 2010
©