Use of molecular recognition to drive chemical evolution. 1. Controlling the composition of an equilibrating mixture of simple arginine receptors

Full text of DOI:10.1021/ja961893r

J. Am. Chem. Soc. 1997, 119, 1147-1148
Use of Molecular Recognition To Drive Chemical
1147
Evolution. 1. Controlling the Composition of an
Equilibrating Mixture of Simple Arginine Receptors
Alexey V. Eliseev* and Marina I. Nelen
Department of Medicinal Chemistry, School of Pharmacy
State UniVersity of New York at Buffalo
Buffalo, New York 14260
ReceiVed June 5, 1996
ReVised Manuscript ReceiVed December 13, 1996
Molecular recognition plays a major role in biological
systems. During the last decade, considerable effort directed
toward modeling of biological non-covalent binding in chemical
systems resulted not only in the synthesis of numerous artificial
molecular receptors but also in the development of innovative
approaches to the generation of selective non-covalent binders.
Besides conventional organic synthesis, combinatorial chemistry
techniques have been applied to the creation of potential
molecular receptors.1,2 A merging of combinatorial chemistry
and DNA cloning techniques has recently been used for the
evolutionary selection/amplification of nucleic acids capable of
binding to biopolymers and low-molecular targets.3
Figure 1. Schematic representation of the experimental setup.
Scheme 1
We introduce here a new and general approach that involves
an automatic enrichment of a mixture of compounds (receptors)
in the component possessing the highest affinity to the particular
target ligand. This method is applicable to any systems in which
the components exist in a dynamic equilibrium. To the best of
our knowledge, this approach is the first application of the
Darwinian principles of mutation and survival (selection) to a
simple chemical system.4 The approach is illustrated by the
evolutionary-type formation of an anionic receptor for arginine.
For the generation of a simple pool of equilibrating receptors,
we have designed and synthesized5 the unsaturated dicarboxylate
1, which is capable of existing in three isomeric forms: trans,-
trans, cis,trans, and cis,cis (Scheme 1). The three isomers can
be interconverted by irradiation with UV light. According to
molecular modeling, the cis,cis isomer of 1 possesses an ideal
geometry for complexation with the guanidinium moiety of
arginine Via formation of two salt bridges (Scheme 1). The
cis,trans and, in particular, trans,trans isomers were expected
to show much lower affinity because of the longer distance
between the carboxylates.6
ylguanidinium hydrochloride were determined by the NMR
titrations with the isomeric mixture in methanol and ethanol.7
As expected from the electrostatic nature of binding, the
association in ethanol is characterized both by higher absolute
values of the binding constants (Kcis,cis ) 980 Vs 170 M-1 in
methanol) and by larger discrimination factors between the
isomers (Kcis,cis/Kcis,trans ) 6.5, Kcis,cis/Ktrans,trans >100 Vs Kcis,cis
/
Kcis,trans ) 2.4, Kcis,cis/Ktrans,trans g 17 in methanol).8
A solution of 1‚2Na in ethanol (2.6 × 10-4M, 7 mL) was
then subjected to circulation in the apparatus shown in Figure
1. Irradiation of the solution with a broad-band UV light
(mercury lamps in a Rayonet reactor) in a photochemical flow
cell (“mutation” chamber, Figure 1) led to the formation of a
distribution of isomers9 (cis,cis/cis,trans/trans,trans ) 3/28/69)
as shown in Figure 2a.10 The resulting mixture was then
pumped to the affinity column (“selection” chamber, Figure 1)
containing 2.8 mL of 1 × 10-2 M arginine immobilized on the
To verify our predictions of the relative affinities, the binding
constants of the disodium salts of all isomers of 1 to meth-
(1) (a) Still, W. C. Acc. Chem. Res. 1996, 29, 155-163. (b) Borchardt,
A.; Still, W. C. J. Am. Chem. Soc. 1994, 116, 373-374. (c) Boyce, R.; Li,
G.; Nestler, H. P.; Suenaga, T.; Still, C. W. J. Am. Chem. Soc. 1994, 116,
7955-7956.
(2) Goodman, M. S.; Jubian, V.; Linton, B.; Hamilton, A. D. J. Am.
Chem. Soc. 1995, 117, 11610-11611.
(7) A 5 × 10-4 M solution of 1‚2Na in CD3OD or CD3CD2OD was
irradiated by a mercury lamp in the NMR tube for 30 min. The 400 MHz
1H NMR spectrum of the resulting mixture contained sufficiently resolved
signals of all isomers of 1. The solution was then titrated with
methylguanidinium‚HCl (5 × 10-4 to 1.3 × 10-1 M), yielding the curves
of δ vs [titrant] (see Supporting Information). The individual association
constants for 1:1 complexes were then calculated from nonlinear least-square
curve fitting.
(8) The relatively low affinity difference between the cis,cis and cis,-
trans isomers is apparently due to an appreciable contribution of the second
COO- group to binding in the latter case. This contribution drops
substantially in the trans,trans isomer, as expected from a nonlinear distance
dependence of ionic interactions.
(3) (a) Famulok, M.; Szostak, J. W. Angew. Chem., Int. Ed. Engl. 1992,
31, 979-988. Famulok, M. J. Am. Chem. Soc. 1994, 116, 1698-1706 and
references therein.
(4) As opposed to “self-replicating systems” (for reviews, see: (a) Orgel,
L. E. Nature 1992, 358, 203-209. (b) Orgel, L. E. Acc. Chem. Res. 1995,
28, 109-118), our method leads to a thermodynamic rather than kinetic
preference in the selection of particular structure. The principle of mutation
(variation) has not been explicitly used either in self-replicating systems or
in the DNA evolution techniques (ref 3).
(5) The synthesis of 1 involved the acetylation of diphenylmethane (AcCl,
AlCl3 in CS2) to give bis(4-acetylphenyl)methane in 75% yield. The latter
was reacted with 2.2 equiv of methyl(diethylphosphono)acetate in a
Wadsworth-Emmons reaction (NaH, DMF, 10-25 °C, 72 h) to afford the
dimethyl ester of 1 (45% after column chromatography), which was then
hydrolyzed by NaOH in MeOH/water and isolated as the trans,trans diacid
of 1 (85%). 1H NMR (300 MHz, DMSO-d6): 12.1 s, 2H; 7.48 d, 4H, 7.27
d, 4H, 6.09 s, 2H; 3.98 s, 2H; 2.46 s, 2H.
(9) The HPLC analysis was performed on a Beckman Gold chromato-
graph (5 cm ODS column; 70% methanol in 0.15% aqueous H3PO4 at 0.5
mL/min), determining each isomer of 1 at its maximum absorption
wavelength (λmax 262 nm (ꢀ ) 33 400), 275 nm (ꢀ ) 30 000), and 282 nm
(ꢀ ) 26 700) for cis,cis, cis,trans, and trans,trans isomers, respectively).
Assignment/integration of the peaks was confirmed by comparison with
1
(6) In the minimized structures of 1 docked with the guanidinium ion
(Tripos software package SYBYL), the distances between the carboxylate
oxygens and guanidinium nitrogens were in the ranges 2.5-2.6, 3.3-5.6,
and 4.6-7.0 Å in the cis,cis, cis,trans, and trans,trans isomers, respectively.
the H NMR of the mixtures.
(10) Since the solution of 1 was irradiated during the limited time of
passing the flow cell, the equilibrium photostationary distribution of isomers
was not reached in the isomerization cycle.
S0002-7863(96)01893-8 CCC: $14.00 © 1997 American Chemical Society

1148 J. Am. Chem. Soc., Vol. 119, No. 5, 1997
Communications to the Editor
removed from the column contained no detectable amount of
any side products, all of which were found to remain in the
circulating solution. These results show that the circulation
experiment leads to (a) amplification of the total amount of the
effective binder in the system, (b) considerable selectivity of
the isomerization-selection process (e.g., changing the cis,cis/
cis,trans ratio from 1.7 in the photostationary state to 6.5 in
the selected subset), and (c) the possibility to isolate only those
components of the pool which possess the desired binding
affinity separating them from the side products.
A control circulation experiment performed with the sorbent
unmodified with arginine (acetylated aminopropyl silica gel)
led to only 12% of the isomers having accumulated on the
column after 30 cycles, their ratio (cis,cis/cis,trans/trans,trans
) 55/31/14) being close to the photostationary distribution in
solution. Performing the 30 cycles with the same sorbent but
without irradiation resulted in the column distribution cis,cis/
cis,trans/trans,trans ) 5/20/75 (12% total), probably due to a
minor thermal isomerization. Conclusively, the result of the
main circulation (evolution) experiment can only be assigned
to the gradual amplification of the effective isomer in the system
driven by its affinity to the immobilized target compound.
In conclusion, we have developed a new method of targeted
evolution of an equilibrating mixture toward the components
possessing the highest affinity to a given ligand. This method
offers promise for studies of molecular recognition, in that it
allows one to reveal the most effective binders among multiple
structures, including complex mixtures. In addition, the method
may also find synthetic utility for “one-reactor” syntheses of
compounds with desired binding properties. Other future
applications might include use of the technique to drive
combinatorial pools of compounds toward particular subsets of
structures.13 We are currently exploring the possible use of this
approach in systems containing reversibly forming positional
isomers (e.g., oligoaldehydes modified with functional amines
Via imine bonds, etc.).
Figure 2. Distribution of the isomers of 1 (%) (a) in solution, after
one irradiation cycle; (b) on the arginine-containing column, after 30
selection-isomerization cycles; and (c) in solution, in the equilibrium
photostationary state.
silica gel support.11 While passing through the column most
of the high-affinity cis,cis isomer was retained on the silica phase
as a complex with the attached arginine.
According to the experimental design, the photochemical
isomerization of 1 in the subsequent cycle should regenerate
the fraction of the high-affinity isomer in the mixture which is
in turn transferred to the selection site. Multiple selection-
isomerization (“mutation”) cycles should lead to increasing the
amount of the effective binder (the cis,cis isomer) accumulated
on the solid support. This method is therefore designed to
provide an automatic amplification of the high-affinity com-
ponent in a system existing in dynamic equilibrium.
After ca. 30 isomerization-selection cycles performed within
8 h, the concentration of 1 in the circulating solution decreased
to 11% of its original value and the distribution of isomers
changed to cis,cis/cis,trans/trans,trans ) 48/29/23. We then
terminated the experiment and treated the silica with several
portions of 1 M aqueous and aqueous/EtOH NaCl to dissociate
the isomeric forms of 1 from the immobilized arginine. HPLC
analysis of the resulting mixture yielded the distribution of
isomers cis,cis/cis,trans/trans,trans ) 85/13/2 (Figure 2b), the
total yield of 1 on the column being 54 ( 5% of the initial
amount.
Acknowledgment. This work was supported by the start-up grant
to A.V.E. from SUNY/Buffalo and the Petroleum Research Fund (Grant
No. 29948-G1). We thank Dr. D. G. Hangauer and Dr. A. J. Solo for
their comments on the manuscript and their continued encouragement
of this project.
For comparison, we performed a long irradiation (8 h) of the
solution of 1 which led to the equilibrium photostationary
distribution of isomers cis,cis/cis,trans/trans,trans ) 52/31/17
with the total yield 66% (Figure 2c) as well as to the formation
of some side products.12 In contrast, the enriched mixture
Supporting Information Available: 1H NMR titration curves of
all isomers of 1 with methylguanidinium in ethanol, HPLC chromato-
grams, and the scheme of the experimental setup (8 pages). See any
current masthead page for ordering and Internet access instructions.
(11) The concentration of immobilized arginine was chosen according
to the binding constant values to provide maximum binding of 1 (cis,cis)
(90% under given conditions) and moderate or weak binding of the other
two isomers (64% and 8% for cis,trans and trans,trans, respectively). The
excess of arginine over 1 was necessary to bind all portions of the effective
isomer forming in the subsequent cycles. N-R-t-BOC-arginine was attached
to (aminopropyl)silica (Wu, R.; Grossman, L. Methods Enzymol. 1987, 154,
299) by the DCC coupling. The residual amino groups on the support were
then acylated by Ac2O in DMF.
(12) HPLC showed formation of two peaks other than the isomers of 1
(see Supporting Information), attributable most likely to the condensation
products, similar to the well-studied photodimers of cinnamic acid (see,
for example: Nakamura, Y. J. Chem. Soc., Chem. Commun. 1982, 477.
D’Auria, M.; Vantaggi, A. Tetrahedron 1992, 48, 2523).
JA961893R
(13) The choice of terms “ligand” and “receptor” is rather arbitrary; the
method can likewise be applied to the equilibration of potential ligands,
e.g., enzyme inhibitors if an immobilized enzyme is used for selection.
Recent studies on the populations of effective binders in known combina-
torial libraries (e.g., Freier, S. M.; Konings, D. A. M.; Wyatt, J. R.; Ecker,
D. J. J. Med. Chem. 1995, 38, 344-352. Wilson-Lingardo, L.; Davis, P.
W.; Ecker, D. J.; Hebert, N.; Acevedo, O.; Sprankle, K. Brennan, T.;
Schwarsz, L.; Freier, S. M.; Wyatt, J. R. J. Med. Chem. 1996, 39, 2720)
show that small subsets of compounds exhibit considerably larger affinities
than the bulk of the components. Thus, our method applied to equilibrating
libraries would allow amplification of such subsets and facilitate further
identification/generation of the effective components.