Probing selectivity in recognition-mediated dynamic covalent processes

Article abstract of DOI:10.1021/ol061177m

Two simple recognition-mediated dynamic Diels-Alder systems are used to probe the role of kinetics and thermodynamics in determining the equilibrium position in exchanging libraries and the time taken to reach that equilibrium. The selectivity expressed by recognition-driven dynamic processes is demonstrated to be less than the free-energy difference between the components as a result of compensatory effects arising from the extent of conversion to products within the library.

Full text of DOI:10.1021/ol061177m

ORGANIC
LETTERS
2006
Vol. 8, No. 17
3651-3654
Probing Selectivity in
Recognition-Mediated Dynamic Covalent
Processes
Raphael M. Bennes and Douglas Philp*
EaStCHEM and Centre for Biomolecular Sciences, School of Chemistry, UniVersity of
St. Andrews, North Haugh, St. Andrews, Fife KY16 9ST, United Kingdom
d.philp@st-andrews.ac.uk
Received May 12, 2006
ABSTRACT
Two simple recognition-mediated dynamic Diels−Alder systems are used to probe the role of kinetics and thermodynamics in determining the
equilibrium position in exchanging libraries and the time taken to reach that equilibrium. The selectivity expressed by recognition-driven
dynamic processes is demonstrated to be less than the free-energy difference between the components as a result of compensatory effects
arising from the extent of conversion to products within the library.
In contrast to combinatorial chemistry, which allows the
efficient and rapid syntheses of thousands of molecules
through kinetically controlled, irreversible processes, the
construction of dynamic combinatorial librariessa nascent
field1 in supramolecular chemistrysrelies on the formation
of structures that are generated by thermodynamically
controlled reversible covalent bonds. Conceptually, two
approaches may be considered depending on whether a
receptor or a substrate acts as a template for the assembly
of the other partner: “casting” consists of the receptor-
induced assembly of a substrate that fits the receptor.
Conversely, “molding” relies on the ability of the substrates
to induce the assembly of a receptor that binds or fits the
substrate optimally. Because all covalent bonds in the system
form under thermodynamic control, there is an element of
proofreading and repair of incorrect steric and electronic fits
with the template. A range of reversible chemical transfor-
mations2 (e.g., imine formation, transacylation, olefin me-
tathesis, or (retro) Diels-Alder reactions) have been em-
ployed in the construction of dynamic selection systems. In
most examples, several components which will ultimately
form the receptor are mixed with the template and the system
is allowed to reach equilibrium. The process, in theory,
generates the most thermodynamically stable complex.
(2) (a) Hutin, M.; Frantz, R.; Nitschke, J. R. Chem.-Eur. J. 2006, 12,
4077-4082. (b) Hutin, M.; Schalley, C. A.; Bernardinelli, G.; Nitschke, J.
R. Chem.-Eur. J. 2006, 12, 4069-4076. (c) Schultz, D.; Nitschke, J. R.
Angew. Chem., Int. Ed. 2006, 45, 2453-2456. (d) Giuseppone, N.; Lehn,
J.-M. Chem.-Eur. J. 2006, 12, 1715-1722. (e) Cantrill, S. J.; Chichak, K.
S.; Peters, A. J.; Stoddart, J. F. Acc. Chem. Res. 2005, 38, 1-9. (f) Chichak,
K. S.; Cantrill, S. J.; Pease, A. R.; Chiu, S. H.; Cave, G. M. W.; Atwood,
J. L.; Stoddart, J. F. Science 2004, 304, 1308-1312. (g) Boul, P. J.;
Reutenauer, P.; Lehn, J.-M. Org. Lett. 2005, 7, 15-18. (h) Ramstrom, O.;
Lehn, J.-M. ChemBioChem 2000, 1, 41-48. (i) Otto, S.; Furlan, R. L. E.;
Sanders, J. K. M. J. Am. Chem. Soc. 2000, 122, 12063-12064. (j)
Nakazawa, I.; Suda, S.; Masuda, M.; Asai, M.; Shimizu, T. Chem. Commun.
2000, 881-882. (k) Ipaktschi, J.; Hosseinzadeh, R.; Schlaf, P. Angew.
Chem., Int. Ed. 1999, 38, 1658-1660. (l) Rowan, S. J.; Sanders, J. K. M.
J. Org. Chem. 1998, 63, 1536-1546. (m) Brady, P. A.; Sanders, J. K. M.
Chem. Soc. ReV. 1997, 26, 327-336.
(1) (a) Rowan, S. J.; Cantrill, S. J.; Cousins, G. R. L.; Sanders, J. K. M.;
Stoddart, J. F. Angew. Chem., Int. Ed. 2002, 41, 898-952. (b) Eliseev, A.
V. Drug DiscoVery Today 2004, 9, 348-349. (c) Otto, S.; Furlan, R. L. E.;
Sanders, J. K. M. Curr. Opin. Chem. Biol. 2002, 6, 321-327. (d) Otto, S.;
Furlan, R. L. E.; Sanders, J. K. M. Drug DiscoVery Today 2002, 7, 117-
125. (e) Ramstrom, O.; Lehn, J.-M. Nat. ReV. Drug DiscoVery 2002, 1,
26-36. (f) Eliseev, A. V.; Lehn, J.-M. Science 2001, 291, 2331.
10.1021/ol061177m CCC: $33.50
© 2006 American Chemical Society
Published on Web 07/26/2006

Considerable effort3 has been directed at understanding the
limits of this selection process. Because the entire process
is under thermodynamic control, one might expect that the
energy difference between the most stable and the next most
stable species in the equilibrating mixture will determine the
observed selectivity.
Here, we demonstrate using a very simple model system,
based on a reversible Diels-Alder reaction, that compensa-
tion effects exist in dynamic systems which serve to limit
selectivity under certain conditions. Additionally, we dem-
onstrate that kinetic effects are also a side effect of these
recognition-mediated selection processes, and these may
affect the time taken to reach equilibrium.
stabilization of the product ground state. We realized, by
extending our studies to encompass the related dienophiles
2 and 4, that these systems present an opportunity to assess
quantitatively the role of kinetic and thermodynamic effects
within a small dynamic library of cycloadducts. In particular,
we wished to explore whether the differences in stability
between the cycloadducts are expressed fully in an exchang-
ing library and whether recognition-induced rate acceleration
plays any role in the equilibration processes of the library.
Initially, we determined the equilibrium positions for each
of the reactions that form the cycloadducts 5-7 by measuring
1
the kinetics of the reactions by H NMR spectroscopy in
CDCl3 at 50 °C. These rate data were fitted to appropriate
kinetic models, and forward and reverse rate constants were
extracted (see Supporting Information). For comparison
purposes, the control reaction described previously6 was used
as a model of the situation in which recognition has no
influence on the reaction.
The equilibrium constants for each of the reactions and
the corresponding relative free energies are shown in Figure
1. Clearly, all three systems exhibit a greater extent of
Previously, we described4 the recognition-mediated reac-
tion between the diene 1 and the dienophile 3. In polar
solvents, such as DMSO, this reaction5 does not proceed in
the forward direction to any significant extent as a result of
the unfavorable loss of conjugation in the diene upon
reaction. In nonpolar solvents, such as CHCl3, the reaction
proceeds rapidly to around 50% conversion. The interaction
between the amidopyridine ring and the carboxylic acid in
the product 6 is responsible for this difference in behaviors
the formation of two hydrogen bonds between these groups
in CHCl3 stabilizes the product ground state significantly
rendering the reaction favorable in terms of free energy.
These hydrogen bonds can also serve to preorganize the
reagents prior to reaction, thus imparting a strong influence
on the kinetics of the reaction. Kinetic studies on the
conversion of 1 and 3-6 revealed that, although the reaction
was accelerated around 8-fold, the dominant effect was the
Figure 1. Relative rate constants, energies, and equilibrium
constants for the conversion of diene 1 and malelimides 2-4 to
cycloadducts 5-7, respectively, in CDCl3 at 50 °C. Relative
energies are in kilojoules per mole. NR represents a control reaction
where there is no molecular recognition present. rfrc represents
the relative forward rate constant for the formation of each
cycloadduct and is expressed as kNR(forward) ) 1.
reaction than the control (Figure 1, NR), with cycloadduct
6 being the most stable. The two-carbon spacer in 6 is exactly
the correct length to permit the formation of two stabilizing
hydrogen bonds between the carboxylic acid and amidopy-
ridine. Molecular mechanics calculations indicate that, in the
case of 5, the one-carbon spacer is too short to permit the
formation of two hydrogen bonds and that, in 7, the three-
carbon spacer is too long. Clearly, the recognition present
in these systems is responsible for the relative stabilities of
the products.
(3) (a) Saur, I.; Scopelliti, R.; Severin, K. Chem.-Eur. J. 2006, 12, 1058-
1066. (b) Buryak, A.; Severin, K. Angew. Chem., Int. Ed. 2005, 44, 7935-
7938. (c) Severin, K. Chem.-Eur. J. 2004, 10, 2565-2580. (d) Vial, L.;
Sanders, J. K. M.; Otto, S. New J. Chem. 2005, 29, 1001-1003. (e) Corbett,
P. T.; Sanders, J. K. M.; Otto, S. J. Am. Chem. Soc. 2005, 127, 9390-
9392. (f) Corbett, P. T.; Tong, L. H.; Sanders, J. K. M.; Otto, S. J. Am.
Chem. Soc. 2005, 127, 8902-8903.
Examination (Figure 1) of the relative rate constants for
the forward reaction, i.e., the reaction of the diene and
dienophile to form the appropriate cycloadduct, reveals a
significant kinetic effect. Reducing the tether length from
(4) Bennes, R. M.; Philp, D.; Spencer, N.; Kariuki, B. M.; Harris, K. D.
M. Org. Lett. 1999, 1, 1087-1090.
(6) Given the uniformly low conversions in the absence of significant
recognition-mediated stabilization of the product ground state, the extent
of the control reaction is difficult to measure. To minimize errors, the value
of the free energy and the rate constant used to derive the relative rate data
in Figure 1 are averages from data derived from reactions between
2-phenylfuran and three different maleimides which bear no recognition
sites. (See ref 4.)
(5) The reaction medium is important in Diels-Alder reactions generally.
(See, for example: Grieco, P. A.; Kauffman, M. D. J. Org. Chem. 1999,
64, 6041-6048.) The influence of solvent polarity on the recognition effects
exposed in this study is potentially a means of fine-tuning the selectivity
of these dynamic reactions, and these effects are currently under investigation
in our laboratory.
3652
Org. Lett., Vol. 8, No. 17, 2006

three carbons to one carbon should result in an increase in
rate on purely entropic grounds7 as the number of rotors in
the system is decreased. The fact that the observed rate
enhancement is maximal for the reaction between 1 and 2
suggests that the single-carbon spacer allows the easiest
access to the Diels-Alder transition state. However, the
single-carbon spacer does not provide sufficient flexibility
to permit the retention of two stabilizing hydrogen bonds in
the cycloadduct 5. Therefore, although the forward reaction
proceeds faster, the product 5 is less stable ensuring that the
reverse reaction is also fast. The overall outcome is a
relatively low Keq. When the spacer in the dienophile is
increased to two carbons in 3, the transition state stabilization
is diminished, but the increased flexibility in the spacer
ensures the retention of two stabilizing hydrogen bonds in
the cycloadduct 6. Therefore, although the forward reaction
proceeds somewhat slower, the product is much more stable
ensuring that the reverse reaction is also slow. The overall
outcome is that K6 > K5. The three-carbon spacer present in
4 (and, hence, 7) is too long to permit either transition state
stabilization or retention of hydrogen bonds in the cycload-
duct.
Having established the basic thermodynamic parameters
for these systems, we next studied a reaction in which the
maleimides 2 and 3 compete for the diene 1. We expected
that the recognition-mediated rate acceleration observed in
the reaction between 1 and 2 should result in 5 being the
kinetic product, whereas the greater stability of 6 should
result in it being the thermodynamic product of the reaction.
This expectation was confirmed (Figure 2) by the results of
a kinetic experiment performed in CDCl3 at 50 °C at starting
concentrations of [1] ) [2] ) [3] ) 25 mM. Although 5 is,
indeed, formed faster than 6, the greater stability of 6 ensures
that, after 180 min, it becomes the major cycloadduct in
solution. Simulation of the experimental data using the
kinetic parameters derived from our initial experiments
affords a pleasingly good fit (solid lines, Figure 2) to the
experimental data. At equilibrium (t ) 40 h), the observed
selectivity (ratio [6]/[5]) is 3.55 in favor of 6.
Given the success of the three-component system, we next
studied the system in which all four components, the diene
1 and the three maleimides 2-4, were present. Kinetic
experiments were performed in a manner identical to that
described above in CDCl3 at 50 °C at starting concentrations
of [1] ) [2] ) [3] ) [4] ) 25 mM. In this situation,
cycloadduct 5 is once again formed most rapidly, but after
3 h, 6, as before, becomes the dominant species in solution.
In this system, at equilibrium (t ) 40 h), the observed
selectivities are: [6]:[5] ) 3.49, [6]:[7] ) 35.7, and [5]:[7]
) 10.2.
Figure 2. Progress of the three-component dynamic system as a
function of time. Experimental conditions: [1] ) [2] ) [3] ) 25
mM; CDCl3; 50 °C. (a) Experimental concentrations of 5 and 6 as
a function of time. Experimental points are the filled circles, and
the solid lines represent the simulated responses. (b) Selectivity of
[6]/[5] as a function of time. Experimental points are the filled
squares and the solid line represents the simulated response.
The selectivities observed in both the three-component and
the four-component systems are below those expected purely
on the basis of the free-energy differences between the
isolated systems. This effect is to be expected because the
expression for the final ratio of 6/5, for example, depends
not only on K5 and K6 but also on the equilibrium concentra-
tions of 2 and 3. On the basis of this information, we
calculated the response of both the two-component and the
three-component systems to differing starting concentrations
of diene 1. The results are shown in Figure 3a.
These response curVes suggest that maximum selectiVity
is achieVed only at low conVersionssat infinitely low
conversion, [6]/[5] ≈ K6/K5. Derivation of the equation for
the equilibrium constant for this competitive process reveals
the reason for the observed drop in selectivity. The overall
equilibrium constant Keq depends not only on the K6/K5 ratio
but also on the ratio of the remaining starting materials
remaining at equilibrium, i.e., on [3]/[2]. This dependence
on the equilibrium concentrations of the starting materials
results in a compensatory effect in the final position of the
equilibrium. At equilibrium, a system with high selectivity
(7) The association of the diene and the dienophile renders the reaction
between them pseudointramolecular. Rate enhancement is therefore expected
on purely entropic grounds. See: Page, M. I.; Jencks, W. P. Proc. Natl.
Acad. Sci. U.S.A. 1971, 68, 1678-1683. The calculated rotor increments
on moving from 2 to 3 (23 eu) and from 3 to 4 (20 eu) are in the range
expected from this model. An alternative view is that, with suitable spacer
design, the association of the diene and the dienophile results in a high
population of near attack conformations (NACs) for the reaction between
them. See: Lightstone, F. C.; Bruice, T. C. Acc. Chem. Res. 1999, 32, 127-
136. In this scenario, it is likely that the [1‚2] complex has the highest
proportion of NACs and is, therefore, accelerated most strongly.
Org. Lett., Vol. 8, No. 17, 2006
3653

described similar findings based on simulations of larger
libraries with different topologies. In summary, the higher
the selectivity for a given product, the more strongly that
selectivity will depend on the starting concentrations of the
components. This effect can be observed in the results for
our three-component system where compound 7 is disfavored
strongly. The selectivity of this product with respect to 6,
the strongly favored component, depends (Figure 3) signifi-
cantly on the starting concentration of the diene 1.
These effects can be demonstrated experimentally. We
repeated the selection experiments for both systems with
starting concentrations of the diene 1 of only 10 mM. Under
these conditions, the overall conversion within the system
will undoubtedly be lower. For the three-component system
at equilibrium (t ) 40 h), the observed selectivity is: [6]/[5])
) 3.83. For the four-component system at equilibrium (t )
40 h), the observed selectivities are: [6]/[5] ) 3.82, [6]/[7]
) 40.5, and [5]/[7] ) 10.6. All of these selectivities are
higher than those observed when the starting diene concen-
tration is 25 mM. Satisfyingly, the experimentally determined
values of the selectivities obtained in these experiments
match well (open squares, Figure 3b) with our predictions.
Although the two dynamic systems reported here are very
small in terms of the numbers of components and are limited
in terms of structural diversity, they afford some important
insights into the design and implementation of more complex
systems. At the level of thermodynamics, one can expect
less selectivity than the absolute energy differences between
isolated library components. This diminished expression of
free-energy differences is dependent on the starting condi-
tions and may result in the selection of more than one
structure by the dynamic library if the energy differences
between target components are relatively small. At a kinetic
level, preorganization of the precursor reagents is an
inevitable consequence of the molecular recognition, which
drives the selection process. This preorganization must have
an effect on how fast the dynamic system reaches equilib-
rium. In certain circumstances, this may significantly lengthen
the time taken to reach the final equilibrium position to an
unacceptable level. These observations should help instruct
the design of dynamic selection systems generally.
Figure 3. (a) Selectivity in the three-component system as a
function of the starting concentration of diene 1. The simulated
response is the solid line. Experimental points (open squares): A,
[1] ) [2] ) [3] ) 25 mM, and B, [1] ) 10 mM, [2] ) [3] ) 25
mM. (b) Selectivity in the four-component system as a function of
the starting concentration of diene 1. The simulated response is
the solid line. Experimental points (open squares): A, [1] ) [2] )
[3] ) [4] ) 25 mM, and B, [1] ) 10 mM, [2] ) [3] ) [4] ) 25
mM. All experiments were performed at 50 °C in CDCl3.
should have almost no starting material left which gives rise
to the favored product. However, the concentration of the
starting material which gives rise to the disfavored product
will still be high at equilibrium.
Therefore, in the example here, the higher the K6/K5 value,
the lower the [3]/[2] ratio and the bigger the compensation
effect at higher conversions. This effect is exacerbated when
the concentration of the common component is higher than
that of the starting materials. Severin3c and Sanders8 have
Acknowledgment. We thank the Engineering and Physi-
cal Sciences Research Council and the University of St.
Andrews for financial support.
Supporting Information Available: Details of the ex-
traction of kinetic and thermodynamic parameters by kinetic
simulation and fitting. This material is available free of
charge via the Internet at http://pubs.acs.org.
(8) (a) Corbett, P. T.; Otto, S.; Sanders, J. K. M. Org. Lett. 2004, 6,
1825-1827. (b) Corbett, P. T.; Otto, S.; Sanders, J. K. M. Chem.-Eur. J.
2004, 10, 3139-3143.
OL061177M
3654
Org. Lett., Vol. 8, No. 17, 2006