The advantage of covalent capture in the combinatorial screening of a dynamic library for the detection of weak interactions

Article abstract of DOI:10.1002/ejoc.200901516

In this paper we address the advantage of screening a dynamic library by covalent capture in comparison with an approach in which the target is not covalently bound to the molecular receptor. The aim is the selection of recognition units for the binding of an anion (or polyanion) by relatively weak binding interactions, a situation typically found in supramolecular chemistry. To compare the two approaches, two model systems have been studied both based on the functionalization of a molecular platform, by reversible imine formation. In the case of the noncovalently bound substrate, the platform P1 is a trisubstituted benzene unit, 2,4,6-trimethylbenzene-1,3,5-tricarbaldehyde, to select: three recognition arms for the binding of the trisodium salt of benzene-1,3,5tricarboxylate. For the covalent-capture-based approach the platforms P2 and P4 are benzene derivatives with a tethered phosphonate target (tetrabutylammonium 2-formylphenyl ethylphosphonate) for the selection of a single recognition unit. The library of recognition elements comprises phenyl and ammonium-functionalized amines. We show that the selection of recognition units for the binding of the substrate with weak to medium, binding constants may encounter, by using a noncovalently bound substrate, serious problems. This is because the best conditions for the amplification of the library, that is, a large excess of variable recognition elements and target, lead also to competitive binding of the elements not bound to the platform, with the target. This may result in negligible amplification of the best-fit members of the library. In contrast, upon tethering the target to the platform and using the covalent-capture strategy for the selection of the recognition elements, significant amplification is observed, even for systems with much lower binding constants. Although competition with excess recognition units may also become an issue in the case of the tethered target, there is a way to overcome the problem by working at low concentrations.

Full text of DOI:10.1002/ejoc.200901516

FULL PAPER
DOI: 10.1002/ejoc.200901516
The Advantage of Covalent Capture in the Combinatorial Screening of a
Dynamic Library for the Detection of Weak Interactions
Marco Martin,[a] Giulio Gasparini,[a] Matteo Graziani,[a] Leonard J. Prins,*[a] and
Paolo Scrimin*[a]
Keywords: Combinatorial chemistry / Covalent capture / Chemical library / Molecular recognition / Amines
In this paper we address the advantage of screening a dy-
namic library by covalent capture in comparison with an ap-
and ammonium-functionalized amines. We show that the se-
lection of recognition units for the binding of the substrate
proach in which the target is not covalently bound to the mo- with weak to medium binding constants may encounter, by
lecular receptor. The aim is the selection of recognition units
for the binding of an anion (or polyanion) by relatively weak
binding interactions, a situation typically found in supra-
molecular chemistry. To compare the two approaches, two
model systems have been studied both based on the func-
tionalization of a molecular platform by reversible imine for-
mation. In the case of the noncovalently bound substrate, the
platform P1 is a trisubstituted benzene unit, 2,4,6-trimeth-
ylbenzene-1,3,5-tricarbaldehyde, to select three recognition
arms for the binding of the trisodium salt of benzene-1,3,5-
tricarboxylate. For the covalent-capture-based approach the
platforms P2 and P4 are benzene derivatives with a tethered
phosphonate target (tetrabutylammonium 2-formylphenyl
ethylphosphonate) for the selection of a single recognition
unit. The library of recognition elements comprises phenyl-
using a noncovalently bound substrate, serious problems.
This is because the best conditions for the amplification of
the library, that is, a large excess of variable recognition ele-
ments and target, lead also to competitive binding of the ele-
ments not bound to the platform with the target. This may
result in negligible amplification of the best-fit members of
the library. In contrast, upon tethering the target to the plat-
form and using the covalent-capture strategy for the selec-
tion of the recognition elements, significant amplification is
observed, even for systems with much lower binding con-
stants. Although competition with excess recognition units
may also become an issue in the case of the tethered target,
there is a way to overcome the problem by working at low
concentrations.
of the library will result in the shift of the equilibrium with
a change in the composition of the library. This may lead
to the selection of the member of the DCL with the highest
affinity for the target. If this occurs, the concentration of
this member is “amplified”. The implication for drug[4] or
enzyme-inhibitor discovery,[5] just to give two examples, is
obvious. Nonetheless, the fundamental limitations of DCC
have been addressed in a series of theoretical contributions,
mainly by Severin[6] and Sanders[7] and their co-workers,
but also by others.[8] Under certain conditions,[6b] and this
has also been verified experimentally,[9] the addition of a
target to a DCL does not necessarily lead to the amplifi-
cation of the member of the DCL with the highest affinity
for the target. Rules of thumb for a successful selection of
library members with the highest affinity for the target have
been suggested.[5–8] Although spectacular examples have
demonstrated the possibility of isolating, also quantita-
tively, high-affinity artificial receptors from a DCL,[10] it
seems that the application of DCC for detecting weak to
medium binding events is much more cumbersome.[7d]
To make DCC attractive in the weak to medium binding
regime, we recently suggested[11] covalent capture as a
promising strategic improvement for the selection of the re-
cognition unit. Briefly, this implies that the target is cova-
Introduction
Since its first introduction, dynamic combinatorial chem-
istry (DCC)[1,2] has shown its potential as a way to obtain
collections of molecules without the hassle and complica-
tions related to the synthesis and deconvolution require-
ments of standard combinatorial chemistry. The possibility
of screening in situ the dynamic combinatorial library
(DCL) of molecules obtained by the spontaneous selection
of the most suitable receptor for a specific molecular target
has increased the interest in the study of these systems.
Briefly, a DCL is a collection of molecules formed by a
pool of common building blocks held together by reversible
bonds.[3] Because of the reversibility in the bond formation
the composition of a DCL is governed by the thermo-
dynamic equilibrium between its components. Any alter-
ation of this equilibrium such as the addition of a molecule
(target) that binds selectively to one (or more) member(s)
[a] Department of Chemical Sciences, University of Padova,
35131 Padova, Italy
Fax: +39-049-8275239
E-mail: paolo.scrimin@unipd.it
leonard.prins@unipd.it
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejoc.200901516.
3858
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 3858–3866

Combinatorial Screening for the Detection of Weak Interactions
lently bound to the receptor, which then “captures” molecu-
lar units that are able to interact with the target. Covalent
capture,[12] in spite of the popularity it has gained recently,
is not a new strategy as it is quite often used by natural
systems for increasing the strength and selectivity of the
interaction between them.[13] The conceptual difference be-
tween the covalent-capture approach and the traditional
one in DCC is shown in Figure 1 for the recognition of a
target T by a collection of potential recognition units that
are assembled into a common platform P. In the standard
DCC approach the library of receptors is first formed and,
subsequently, they interact with the target to form the com-
plex. In the specific case of Figure 1, for the sake of sim-
plicity, only one of the members of the library binds to the
target with a sizeable binding constant. The final outcome
of the selection is the identification of a molecular receptor
comprising the platform and the best-fit recognition units.
In the covalent-capture strategy, for the purpose of selec-
tion, the target is covalently and irreversibly bound to the
platform, and the recognition units that better interact with
it are covalently captured by the platform. In this case the
selected library component is the molecular receptor with
the substrate covalently bound to it.
Figure 1. Schematic representation of the selection process of a re-
cognition unit for a target T by using an assembling platform P.
(a) Conventional DCL and (b) dynamic covalent capture. For the
sake of simplicity it is assumed that only one of the library mem-
bers binds to the target.
The covalent-capture process may be reversible or
irreversible. In the first case, also referred to as dynamic
covalent capture, the library composition will reflect the
thermodynamic stability of the DCL members, whereas in
the second case this is only true if the stabilities of the tran-
sition states towards product formation parallel those of the
products. In this work we wish to address the advantage of
Results and Discussion
The Libraries Considered
The two libraries were designed to select recognition
dynamic covalent capture with respect to traditional (i.e., units for an anionic host by using imine formation as the
the target noncovalently bound to the receptor) dynamic reversible reaction.[14] Accordingly, the libraries considered
combinatorial selection in detecting relatively weak binding were all under thermodynamic control. Platform P1 used in
events. This has been achieved by studying two model sys- the classic DCC approach for selecting three recognition
tems both based on the functionalization of a platform, a arms for an anion is 2,4,6-trimethylbenzene-1,3,5-tricarbal-
situation similar to that shown in Figure 1.
dehyde, whereas platform P2 is tetrabutylammonium 2-for-
The platform is not involved in the molecular recognition mylphenyl ethylphosphonate for the selection of a single re-
process but allows the connection between the recognition cognition arm for the attached anionic phosphonate. It is
elements and, in the case of covalent capture, also of the well documented that the three substituents at alternating
target. The presence of such a common building block is positions of a hexasubstituted benzene point in the same
an advantage, particularly if this is the limiting species in direction, either above or below the benzene plane.[15,16]
terms of concentration, because it minimizes selection prob- Hence, the three aldehyde substituents upon imine forma-
lems as demonstrated by Severin.[6b]
tion will present three arms protruding from the same side
Scheme 1. Synthesis of platforms P1–P4. Reagents and conditions: (i) HBr (33% in CH3COOH), (CH2O)n, ZnBr2, 90–100 °C, 93%; (ii)
[(CH3)2CNO2]–Na+, EtOH, 68%; (iii) NaH, CH3CH2P(=O)Cl2, NaHCO3, 12% (R = tBu), 15% (R = H); (iv) TBA acetate.
Eur. J. Org. Chem. 2010, 3858–3866
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
3859

M. Martin, G. Gasparini, M. Graziani, L. J. Prins, P. Scrimin
FULL PAPER
of the benzene ring, thus providing convergent recognition ently bound substrate approach in the selection of recogni-
elements for the binding of the target.[17] The synthetic stra- tion units for the binding of an anion, we limited the
tegies used to synthesize the two platforms are shown in number of the components of the libraries to the minimum
Scheme 1.
required, that is, two different types of amines, one bearing
The two selected platforms are quite different, not only a phenyl group, which is unable to interact with an anion,
in terms of their structure, but also for the number of re- and one comprising an ammonium ion supposed to interact
cognition elements that each of them presents. Indeed, one with the target through a charge–charge interaction. Thus,
may argue that platform P1 is greatly favored over platform the three-armed platform P1 leads to a four-membered li-
P2 in the binding of a target due to the presence of three brary and the single-armed platform P2 to only a two-mem-
converging recognition units compared with the single unit bered library. Charge–charge interactions do not have geo-
present in P2. However, one should consider that, having metrical requirements[18] unlike, for instance, hydrogen-
already connected the target anion (the phosphonate) in the bonding, and this removes most conformational issues in
case of P2, this amounts to an infinite binding constant the recognition process. They are, however, sensitive to dis-
for the binding of the target to the platform, a situation tance.
impossible to achieve in the case of P1. This is an important
For this reason we have also explored platform P4 con-
and critical factor that favors platform P2 over P1 in the taining a bulky tert-butyl group in the ortho position with
detection of binding events. This difference makes it im- respect to the phosphonate. This should force the phos-
possible to compare the composition of the library with P2 phonate group towards the ammonium group and thus
in the absence of the target. For this reason platform P3, strengthen the charge–charge interaction. The different li-
which bears a methoxy group in place of the phosphonate, braries obtained by using the three platforms are presented
was also considered.
in Figure 2, and the synthetic routes for the preparation of
The product distribution of the library obtained by reac- the platforms are shown in Scheme 1. It is crucial to under-
tion with P2 has been in all cases compared with that ob- stand that, whereas in the case of the conventional DCC
tained with P3 to normalize the results for any intrinsic approach based on platform P1 the selected library compo-
difference of stability of the formed imines not related to nent will be a molecular receptor for the target, with plat-
the stabilization due to the interaction with the phos- forms P2 and P4 the selection process will provide an indi-
phonate. In the case of platform P1 a comparison has been cation of the best recognition unit for the target, and the
made between the libraries obtained in the presence and pinpointed library member is not a molecular receptor. To
absence of the added target anion.
obtain a molecular receptor the selected recognition ele-
Because our aim was to prove the advantage of the coval- ments should be used for the functionalization of a plat-
ent-capture strategy in DCC compared with the noncoval- form like P1.
Figure 2. Library components obtained with platforms P1–P4 by using amines A–C.[19]
3860
www.eurjoc.org
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 3858–3866

Combinatorial Screening for the Detection of Weak Interactions
Figure 3. Calculated amplifications of the library components P1A3 (ᮀ), P1A2B (᭺), P1AB2 (᭹), and P1B3 (᭿) as a function of the
amount of T added for different ratios of [amines]/[P1] (3, 4, 6, and 15 for a–d, respectively). Conditions: [P1] = 1ϫ10–2 . Binding
constants for the binding of the target to the library members are 10, 100, and 1000 –1 for P1A2B, P1AB2, and P1B3, respectively.
Amplification factors are calculated as the ratio between the concentration of a species in the presence and absence of the target under
the same conditions.
Simulations for Recognition Site Selection by Using
Conventional DCC
This is in full accord with the calculations performed by
Severin using a similar, even though simpler model.[6b] In
addition, these simulations confirm the advantage of having
a common building block present at limiting concentrations
(here P1). When no excess of amines (Figure 3a) is used,
the expected amplification of P1B3 induces an amplification
nearly as strong of the nonreceptor P1A3. Under these con-
ditions, the final library composition does not reflect at all
the attributed binding constants. On the other hand, a true
reflection of the binding constants in the final library com-
position is observed when the platform P1 is present at very
low relative concentrations (Figure 3d).
From these simulations we may conclude that in real ex-
periments we should work under conditions of a large ex-
cess of amines and target with respect to the platform. We
should also select a target anion with a very large binding
constant with the best receptor to maximize its amplifi-
cation in the DCL.[24]
By mixing the two amines A[19] and B with platform P1
the four products P1A3, P1A2B, P1AB2, and P1B3 are
formed (see Figure 2). On a statistical basis, assuming an
identical stability of the imines derived from A and B, the
relative ratio of these four products should be 1:3:3:1. As will
be shown below, however, amine A forms a more stable imine
and, consequently, library members containing A are favored
over those containing B. Addition of a generic target anion T
should amplify library members forming stronger complexes
with T, that is, P1B3, P1AB2, and P1A2B, in decreasing or-
der. In the solvent used for our studies (CD3OD) the binding
constants between an ammonium and a carboxylate or phos-
phonate anion are rather small, typically around 20–
30 –1.[20] Simulations of the library composition were per-
formed by assuming that each arm containing an ammo-
nium cation contributes one order of magnitude to the bind-
ing constant.[21] Accordingly, binding constants of 10, 100,
and 1000 –1 for the binding of T by the three library mem-
bers P1A2B, P1AB2, and P1B3, respectively, were consid-
ered.[22] The expected amplifications for this system were cal-
Experiments Using a Conventional DCL
On the basis of the above simulations we were first inter-
culated as a function of the ratio [amines]total/[P1] and the ested in selecting a target anion showing a sizeable binding
amount of target T added (Figure 3). constant with receptor P1B3, the most promising candidate.
These calculations indicate that maximum amplification The following anions were considered: The single-charged
should be observed when large excesses of amines and tar- sodium bis(p-nitrophenyl) phosphate and sodium benzoate,
get are used (see Figure 3d for high [amines]/[P1] ratios).[23] the double-charged disodium p-nitrophenyl phosphate and
Eur. J. Org. Chem. 2010, 3858–3866
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
3861

M. Martin, G. Gasparini, M. Graziani, L. J. Prins, P. Scrimin
FULL PAPER
disodium benzene-1,3-dicarboxylate, and the triple-charged in the absence of acid resulted in the relatively fast forma-
trisodium benzene-1,3,5-tricarboxylate (the trisodium salt tion of P1A3 (24 h) followed by a very slow equilibration to
of trimesic acid). Binding constants for the 1:1 complexes the final composition 4.6:9.3:5.4:1 of P1A3 (2.25ϫ10–3 ),
were determined in [D4]methanol by 1H NMR spectroscopy P1A2B (4.6ϫ10–3 ), P1AB2 (2.6ϫ10–3 ), and P1B3
and are reported in Table 1. As discussed in the Supporting (4.9ϫ10–4 ), respectively, in around 60 d. To avoid this
Information, analysis of the data reveals that different bind- very unpractical situation we first prepared P1B3 by adding
ing stoichiometries do not occur under the experimental only B to P1 and subsequently added A. Following this
conditions.
protocol the same equilibrium composition was reached in
6 d (see the Supporting Information).[25]
Table 1. Binding constants Kb of receptor P1B3 with different tar-
get anions T determined by 1H NMR spectroscopy in [D4]meth-
anol.
The next crucial experiment was the equilibration of the
library by using the trianionic target to determine the extent
of amplification of the library members containing the am-
monium group. The addition of the trisodium salt of trim-
esic acid to the three reagents (P1, A, and B) sped up the
equilibration process that nevertheless remained too slow
for practical applications. Accordingly, as done before, we
started from preformed P1B3 and equilibrated the DCL by
addition of A and T. Regrettably we were unable to use an
excess of T because of solubility problems. Thus, we could
not perform our experiment under the best conditions for
maximum amplification. Under the conditions [P1]o = [T]o
= 1ϫ10–2  and [A]o = [B]o = 4.4ϫ10–2  the equilibrium
mixture had the composition 4.5:7.6:3.1:1 of P1A3
(2.6ϫ10–3 ), P1A2B (4.4ϫ10–3 ), P1AB2 (1.8ϫ10–3 ),
and P1B3 (5.8ϫ10–4 ), respectively. For P1B3 the amplifi-
cation observed was only 1.2-fold greater than that ob-
served in the absence of the target anion. Our expectation
was of a 3.1-fold amplification for P1B3 (see the Supporting
Information). A summary of the observed and expected
amplifications is reported in Table 2.
T
Kb [–1]
Sodium benzoate
43Ϯ8
61Ϯ9
681Ϯ41
650Ϯ120
2762Ϯ519
Sodium bis(p-nitrophenyl) phosphate
Disodium benzene-1,3-dicarboxylate
Disodium p-nitrophenyl phosphate
Trisodium benzene-1,3,5-tricarboxylate
Figure 4 shows the experimental data for the binding of
the trisodium salt of trimesic acid. The good fitting of the
data, performed by assuming a 1:1 complex, confirms (see
also the Supporting Information) this binding stoichiome-
try.
Table 2. Expected and observed amplifications of the different li-
brary members as a consequence of the binding of trimesic acid to
the trisodium salt.
Figure 4. Binding isotherm of the trisodium salt of trimesic acid to
P1B3 (1ϫ10–3 ). The curve represents the best fitting of the data
points with a 1:1 binding stoichiometry.
Library member
P1A3
P1A2B
P1AB2 P1B3
Expected amplification[a]
0.6
1.2
0.5
1.0
1.0
0.7
3.1
1.2
Observed amplification[b]
Analysis of the data in Table 1 shows that the largest
binding constant is observed between P1B3 and the sodium
salt of trimesic acid, characterized by the number of charges
matching that of the receptor. For each decrease in charge
[a] See the Supporting Information for details. [b] See text and Fig-
ure 3c for details.
To explain these results it occurred to us that, under
unit in the target anion, regardless of its nature, there is these conditions, the excess of the ammonium-function-
roughly a decrease in the binding constant of one order alized amine B may actually compete with the three poten-
of magnitude. Thus, the experimental results support the tial receptors of the library for the binding of the target. To
assumptions we have made in the calculations reported verify whether this was a reasonable argument we re-ran
above. On the basis of these results, we selected the triso- simulations of the system introducing this time also a bind-
dium salt of trimesic acid as our anionic target.
ing event between the free amine B and the target. These
Next we determined the equilibrium composition of the simulations are shown in Figure 5 and indicate that, al-
four-membered library to check whether it deviates from though substantial amplification of P1B3 should be ob-
the statistical one (1:3:3:1 for P1A3, P1A2B, P1AB2, and served in the case of negligible binding of B to T (top line
P1B3) in case the four library members are not equally of Figure 5) in the presence of a large excess of B, the ampli-
stable. The experiments were performed in [D4]methanol by fication of PB3 diminishes significantly as a function of the
1
using H NMR spectroscopy. It was immediately clear that strength of its interaction with T. In the extreme case of a
not only were the library members containing the phenyl binding constant for the complex B·T not much lower than
substituent A more stable, but they were also formed much that for complex P1B3·T, even deamplification of the latter
faster than those containing the ammonium substituent B. may be observed (lower curve of Figure 5). To verify
Thus, the mixing of the three reagents under the conditions whether B was actually a competitor of P1B3 for the bind-
[P1]o = 1ϫ10–2  and [A]o = [B]o = 4.4ϫ10–2  at 65 °C ing of T we monitored the formation of the complex be-
3862
www.eurjoc.org
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 3858–3866

Combinatorial Screening for the Detection of Weak Interactions
1
tween T and P1B3 by H NMR spectroscopy and, after its mation between the library components and the target,
formation, added increasing amounts of B. As shown in even when they are characterized by much lower binding
Figure 6, B competes with P1B3 for the binding of the tar- constants.
get, and a 20-fold excess of B almost totally suppresses the
Although of little relevance in this work, one may argue
formation of the complex P1B3·T. The apparent binding that the chosen receptor has a considerable degree of con-
constant for the formation of the B·T complex is formational freedom such that the entropic gain in the
460Ϯ50 –1. Clearly, if free B is involved in T complex- binding of the target by P1B3 compared with the free
ation, this amine is subtracted from the library equilibrium, amines is not high enough to offset their competition. Nev-
which results in an advantage for the library members con- ertheless, the results indicate the serious limitations of a
taining A, as experimentally observed.
conventional DCL approach in the selection of a target par-
ticularly under conditions of not particularly high binding
constants, which represent the vast majority of the cases
one meets in real selection experiments by using synthetic
molecular receptors.
DCL with a Platform-Bound Target: Dynamic Covalent
Capture
A possible way to avoid the above problem is to anchor
the target to the platform used for the selection of the re-
cognition element. In this way there is no unbound target
in solution. The binding of the recognition element to the
target results in its covalent capture by the platform (Fig-
ure 1b). Platforms P2–P4 upon mixing with amines B and
C give two products for each platform: Comparison of the
composition of the mixtures P2B, P2C and P4B, P4C with
reference mixture P3B, P3C gives information on the ampli-
fications obtained resulting from attractive interactions be-
tween the ammonium and phosphonate group. By using a
five-fold excess of each amine with respect to the plat-
form[26] we first determined that P3B and P3C are formed
in 1:1 ratio, which indicates that in this case the two library
members are equally stable. Next, the ratios of the two
imines formed with platforms P2 and P4 were determined.
Products P2B and P2C were formed in a 2.1:1 ratio,
whereas P4B and P4C are present in a 9:1 ratio. This
amounts to a 2.1-fold amplification in favor of the ammo-
nium derivative with platform P2 and a nine-fold amplifi-
cation with platform P4 containing the tert-butyl substitu-
ent. The latter, in particular, is an impressive result con-
sidering that the noncovalent interaction is simply that be-
tween a phosphonate and ammonium group, which under
these conditions amounts to a binding constant of only
58Ϯ19 –1. Note that in the case of the conventional DCL
discussed above, even taking advantage of a binding con-
stant around 50-fold larger, we were unable to detect signifi-
cant amplification. Thus, the dynamic covalent-capture ap-
Figure 5. Calculated amplification of P1B3 as a function of the
amount of amines A and B added for increasing binding affinity
between [T] and [B] [KTB = 0 (᭿), 10 (ᮀ), 100 (᭹), or 1000 –1
(᭺)]. Conditions: [P1] = [T] = 1ϫ10–2 . The binding constant for
the P1B3·T complex was fixed at 2.4ϫ103 –1.
Figure 6. The upward curve represents the binding of the trisodium
1
salt of trimesic acid (T) to P1B3 as monitored by H NMR spec-
troscopy ([D4]methanol) keeping constant [P1B3] = 2.9ϫ10–3 .
The downward curve represents the disappearance of the complex
P1B3·T upon addition of increasing amounts of free B. Note that
the graph shows two different experiments.
Together, these results indicate the difficulty of applying proach resulting in intramolecular trapping for the forma-
DCC to the detection of weak to medium binding events. tion of a DCL is significantly more sensitive in detecting
On one hand, as supported by our simulations and those weak interactions. The results obtained with platform P4
carried out by others,[6b] the use of a scaffold as a common indicate how costly conformational freedom is in these sys-
building block in limiting concentrations is one of the most tems: The introduction of the bulky tert-butyl group at the
promising ways to magnify the library response to a target. ortho position relative to the phosphonate results in a more
Amplification is maximized when large excesses of the other than four-fold increase in the amplification without chang-
library components and the target are used. On the other ing the functional groups directly involved in the recogni-
hand, such large excesses induce competitive complex for- tion process.
Eur. J. Org. Chem. 2010, 3858–3866
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
3863

M. Martin, G. Gasparini, M. Graziani, L. J. Prins, P. Scrimin
FULL PAPER
As discussed before, competition between the free amine lecular receptor by independent optimization of its several
B and receptor P1B3 for the binding of target T is the main fragments. Work in this direction is currently being pursued
reason for the absence of observed amplification in a con- in our laboratory.[30]
ventional DCL. Although much less sensitive, dynamic co-
valent capture is also subject to the same phenomenon.[27]
Experimental Section
As we have shown previously[27] (and we refer the reader to
the simulations reported in our previous paper) using a re-
lated system based on hydrazones rather than imines, a
large excess of hydrazides effectively caused a drop in am-
plification. This is because, when present in excess, the re-
cognition element may bind intermolecularly to the plat-
form-bound phosphonate, a process that competes with the
intramolecular interaction. Nonetheless, and this is the true
advantage of dynamic covalent capture, the intermolecular
competition can be effectively suppressed by working at low
concentrations, which does not affect the intramolecular
binding event. For the conventional DCL this is not an op-
tion because low concentrations affect all binding events.
Materials: Solvents and reagents were purchased from commercial
sources and used without further purification. Milli-Q water was
used throughout. The following buffers were used: 2-(Morpholino)-
ethanesulfonic acid (MES) (pH = 6.0–6.8) and 2-[4-(2-hydroxy-
ethyl)piperazin-1-yl]ethanesulfonic acid (pH = 7.0–8.0). Platform
P3 is commercially available. The synthesis of platform P2 has al-
ready been reported.[27] 1H NMR spectra were recorded with
Bruker AC 250 F or AM 300 spectrometers. ESI-MS data were re-
corded with an Agilent 1100 Series LC system.
1,3,5-Triethyl-2,4,6-triformylbenzene (P1):[31,32] 1,3,5-Triethylbenz-
ene (10.06 g, 60.8 mmol), acetic acid (100 mL), paraformaldehyde
(16.42 g, 547 mmol, 9 equiv.), a 33% (5.7 ) HBr solution in HOAc
(570 mmol, 100 mL, 9.4 equiv.), and ZnBr2 (41.08 g, 182 mmol,
3 equiv.) were introduced into a 250 mL pressure vial. The vial was
sealed and the mixture stirred at 90–100 °C for 24 h. The solution
turned reddish-brown in a few minutes, and TLC (EtOAc/petro-
leum ether, 1:20; Rf = 0.52) after 24 h revealed the disappearance
of the starting material. The solution was poured into water
(500 mL), and the solid precipitated was filtered and dried to give
26 g of crude 1,3,5-tris(bromomethyl)-2,4,6-triethylbenzene (97%
yield) that was pure enough for the subsequent reaction. M.p. 158–
Conclusions
The selection of recognition units for the binding of a
substrate (in our case an organic anion or polyanion) with
weak to medium binding constants by conventional DCC
(substrate noncovalently bound to the potential molecular
receptor) under experimental conditions typically encoun-
tered in the laboratory may come across serious problems.
1
160 °C. H NMR (250 MHz, CDCl3): δ = 1.34 (t, J = 7.68 Hz, 9
H, CH3), 2.94 (q, J = 7.68 Hz, 6 H, CH2), 4.58 (s, 6 H, CH2)
ppm. 13C NMR (63 MHz, CDCl3): δ = 15.56, 22.71, 28.55, 132.62,
This is because, even by using the common platform ap- 144.96 ppm. IR (KBr): ν = 2965, 1570, 1493, 1450, 1378, 1200,
˜
1060, 1037, 956, 760, 702, 582 cm–1. Subsequently, Na (3.21 g,
proach (Figure 1a), the conditions leading to amplification
139.7 mmol) was added to a 500 mL round-bottomed flask con-
of the library, that is, a large excess of variable recognition
taining dry ethanol (200 mL). After the disappearance of Na, 2-
elements and target, also lead to competitive binding of the
propyl nitrite (15.9 mL, 177 mmol, 6 equiv.) was added. After 1.5 h,
elements not bound to the platform with the target. This
1,3,5-tris(bromomethyl)-2,4,6-triethylbenzene (13 g, 29.48 mmol)
may result in negligible amplification of the best-fit mem-
was added to the milky solution formed. The solution was stirred
bers of the library. In contrast, when tethering the target to
for 24 h. After this period, a precipitate of NaBr had formed, and
the platform and using the covalent-capture strategy for the
TLC revealed the disappearance of the reagent. After evaporation
selection of the recognition element, significant amplifi-
cation is observed even with much lower binding constants.
of the solvent, the solid was taken up with diethyl ether and ex-
tracted with 10% NaOH (3ϫ 50 mL). The concentrated organic
Although competition with excess recognition units may solution gave 5.8 g of a crude product (80% yield) that was recrys-
1
tallized from ethanol. H NMR (250 MHz, CDCl3): δ = 1.30 (t, J
= 7.53 Hz, 9 H, CH3), 3.00 (q, J = 7.53 Hz, 6 H, CH2), 10.61 (s, 3
H, COH) ppm. 13C NMR (63 MHz, CDCl3): δ = 16.48, 22.56,
also become an issue in the case of the tethered target, there
is a way to overcome the problem by working at low con-
centration.[27] The intramolecular recognition process in-
volved in covalent capture is not affected by dilution,
whereas the intermolecular process involved in the standard
recognition process is greatly diminished. This indicates
that, whenever synthetically possible, tethering followed by
covalent capture is the strategy one should pursue whenever
a common platform-based library is used for the selection
of a recognition unit to be used in the synthesis of a molec-
ular receptor. There are, however, conditions in which the
tethering approach is not synthetically accessible and, ac-
cordingly, the system must be critically evaluated before
choosing the most appropriate strategy. Nevertheless, we
speculate that tethering followed by covalent capture of the
134.09, 149.30, 194.25 ppm. IR (KBr): ν = 2976, 2875, 1696, 1553,
˜
1458, 1421, 137 cm–1. MS (ESI): m/z = 247 [M + H]+.
General Procedure for the Synthesis of Homoimines P1A3 and P1B3:
Platform P1 (115 mg, 0.47 mmol) was introduced into a 5 mL vial
containing CHCl3 (3 mL). Amine A or B (3.6 equiv.) and molecular
sieves (4 Å) to remove the water formed were subsequently added
to the solution. The mixture was stirred at 50–60 °C for 16 h. After
this period, the solvent was evaporated and the excess A or B re-
moved by distillation under reduced pressure. The following prod-
ucts were obtained.
1
P1A3: H NMR (250 MHz, CDCl3): δ = 0.87 (t, J = 7.3 Hz, 9 H,
CH3), 2.47 (q, J = 7.3 Hz, 6 H, CH2), 3.05 (t, J = 7.2 Hz, 6 H,
PhCH2), 3.92 (dt, J = 1.1, 7.2 Hz, 6 H, NCH2), 8.48 (t, J = 1.1 Hz,
best-fit recognition unit may be used to build complex mo- 3 H, CH=N) ppm. C38H45N3 (543.79): calcd. C 84.28, H 8.16, N
7.56; found C 83.97, H 8.31, N 7.38.
lecular receptors with very high affinities for their target by
using a protocol reminiscent of that used in “fragment-
P1B3: 1H NMR (250 MHz, CD3OD): δ = 1.11 (t, J = 7.4 Hz, 9 H,
based drug discovery”,[28,29] that is, by building up the mo- CH3), 2.82 (q, J = 7.4 Hz, 6 H, CH2), 3.32 (s, 27 H, CH3), 3.77 (t,
3864
www.eurjoc.org
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 3858–3866

Combinatorial Screening for the Detection of Weak Interactions
J = 6.4 Hz, 6 H, NCH2), 4.21 (dt, J = 1.3, 6.4 Hz, 6 H, NCH2), 112.63, 67.92, 56.21, 55.87, 54.66 ppm. MS (ESI, MeOH): calcd.
8.90 (t, J = 1.3 Hz, 3 H, CH=N) ppm. C30H57Cl3N6 (608.18): calcd. for [M]+ 221.2, found 221.0. C13H21ClN2O (256.77): calcd. C 60.81,
C 59.25, H 9.45, N 13.82; found C 59.01, H 9.52, N 13.71.
H 8.24, N 10.91; found C 60.95, H 8.19, N 11.02.
2-tert-Butyl-6-formylphenyl Ethylphosphonate (P4): A solution of 2-
tert-butyl-6-formylphenol in dry THF (1.01 g, 5.61 mmol) was
slowly added to a suspension of NaH (11.2 mmol) under N2 at
0 °C. Immediately afterwards, ethylphosphonic dichloride (1.65 g,
11.2 mmol) was added and the solution allowed to warm to room
temperature. After 12 h, it was cooled again to 0 °C, and 12  HCl
(2 mL) was added, which resulted in the immediate precipitation
of a white solid. After filtration, the THF was evaporated and the
crude product taken up with chloroform and extracted first with a
pH = 8 aqueous solution of NaHCO3 and next with a pH = 1
solution of HCl. After drying, evaporation of the organic solvent
gave 624 mg of a greenish-yellow solid. This material was triturated
with ethanol and dried to give a white solid (590 mg). The above
acid (40.4 mg, 0.15 mmol) was converted into the tetrabutylammo-
nium salt by dissolving it in methanol and adding tetrabutylammo-
nium acetate (45.1 mg, 0.15 mmol) dissolved in water. The meth-
anol was evaporated, and the residual water was removed by lyoph-
ilization. A white solid (79 mg) was obtained. 1H NMR (250 MHz,
CD3OD): δ = 10.51 (s, 1 H, CHO), 7.65 (m, 3 H, Ar), 3.24 (m, 8
H, NCH2), 1.9–1.7 (m, 2 H, PCH2), 1.7–1.6 (m 8 H, CH2), 1.6–1.4
P3C: 1H NMR (300 MHz, CD3OD): δ = 8.76 (s, 1 H, CH=N),
7.90 (dd, J = 7.71, 1.74 Hz, 1 H, Ar), 7.4 (m, 1 H, Ar), 7.15–7.34
(m, 5 H, Ar), 6.82–6.93 (m, 2 H, Ar), 4.62 (s, 2 H, CH2Ar), 3.68
(s, 3 H, OCH3) ppm. 13C NMR (50 MHz, CD3OD): δ = 162.05,
141.68, 136.77, 134.21, 129.58, 129.35, 128.09, 125.00, 123.10,
121.79, 112.63, 56.35, 55.70 ppm. MS (ESI, CH3CN + 0.1%
HCOOH): calcd. for [M + H]+ 226.2, found 226.0. C15H15NO
(225.29): calcd. C 79.97, H 6.71, N 6.22; found C 79.85, H 6.86, N
6.14.
P4B·Bu4Cl: 1H NMR (250 MHz, CD3OD): δ = 9.26 (s, 1 H,
CH=N), 7.73 (dd, J = 7.70, 1.69 Hz, 1 H, Ar), 7.49 (dd, J = 7.82,
1.56 Hz, 1 H, Ar), 7.03 (t, J = 7.76 Hz, 1 H, Ar), 4.07 (m, 2 H,
CH2), 3.72 (t, J = 5.98 Hz, 2 H, CH2), 3.18–3.26 (m + s, 17 H,
CH2 and NCH3), 1.86 (dq, J = 17.53, 7.65 Hz, 2 H, CH2), 1.66 (m,
8 H, CH2), 1.47 (s, 9 H, tBu), 1.37–1.46 (m, 8 H, CH2), 1.28 (dt, J
= 18.96, 7.61 Hz, 3 H, CH3), 1.03 (t, J = 7.28 Hz, 12 H, CH3) ppm.
13C NMR (75 MHz, CD3OD): δ = 166.67, 133.49, 131.45, 130.96,
126.60, 123.92, 118.24, 67.90, 59.57, 55.40, 54.77, 36.06, 31.44,
24.81, 24.31, 20.73, 13.95, 8.28 ppm. 31P NMR (122 MHz,
(m, 8 H, CH2), 1.50 (s, 9 H, tBu), 1.4–1.2 (m, 3 H, CH3), 1.03 (t, CD3OD): δ = 30.05 ppm. MS (ESI, MeOH): calcd. for [M + H]+
J = 14.5 Hz, 12 H, CH3) ppm. 13C NMR (63 MHz, CD3CN): δ =
191.68, 156.20 (d, JP,C = 9.3 Hz) 143.54 (d, JP,C = 3.7 Hz), 133.24,
132.34 (d, JP,C = 2.3 Hz), 125.83, 122.59 (d, JP,C = 1.2 Hz), 59.19,
35.79, 31.27, 24.20, 21.96, 20.23, 13.72, 8.76 (d, JP,C = 6.3 Hz) ppm.
31P NMR (122 MHz, CDCl3): δ = 26.79 ppm. MS (ESI; MeOH,
negative mode): m/z = 269.2 [M – H]+.
355.1, found 355.1. C18H31N2O3P (354.43): calcd. C 61.00, H 8.82,
N 7.90; found C 60.87, H 8.93, N 7.81.
P4C: 1H NMR (250 MHz, CD3OD): δ = 9.38 (s, 1 H, CH=N),
7.73 (dd, J = 1.6, J = 7.7 Hz, 1 H, Ar), 7.49 (dd, J = 1.2, J =
7.5 Hz, 1 H, Ar), 7.23–7.40 (m, 5 H, Ar), 7.02 (t, J = 7.8 Hz, 1 H,
Ar), 4.57 (s, 2 H, CH2Ar), 3.19–3.26 (m, 8 H, CH2), 1.8–1.6 (m, 10
H, CH2), 1.49 (s, 9 H, tBu), 1.5–1.3 (m, 8 H, CH2), 1.24 (td, J =
18.68, 7.65 Hz, 3 H, CH3), 1.02 (t, J = 7.28 Hz, 12 H, CH3) ppm.
13C NMR (63 MHz, CD3OD): δ = 165.43, 143.52, 140.84, 131.44,
130.95, 130.92, 129.92, 129.25, 127.97, 126.76, 123.98, 65.57, 59.56,
36.16, 31.59, 24.80, 24.16, 20.73, 13.94, 8.28 ppm. 31P NMR
(122 MHz, CD3OD): δ = 26.27 ppm. MS (ESI, MeOH, negative
mode): calcd. for [M]– 358.3, found 358.0. C36H61N2O3P (600.86):
calcd. C 71.96, H 10.23, N 4.66; found C 71.77, H 10.38, N 4.53.
General Procedure for the Synthesis of Imines P2B, P2C, P3B, P3C,
P4B, and P4C: Equivalent amounts of platform and amine were
dissolved in CD3OD (600 µL) in a screw-cap NMR tube to form a
5 m solution. The tube was kept at 50 °C for 2 d, and the reaction
monitored by NMR spectroscopy (disappearance of the aldehyde
proton signal). After cooling, the solvent was evaporated to give
the desired compound.
P2B·Bu4Cl: 1H NMR (250 MHz, CD3OD): δ = 8.92 (s, 1 H,
CH=N), 7.92 (dd, J = 7.79, 1.44 Hz, 1 H, Ar), 7.01–7.58 (m, 3 H,
Ar), 4.10 (m, 2 H, NCH2), 3.70 (t, J = 6.00 Hz, 2 H, NCH2), 3.19–
3.26 (m, 17 H, CH2 and NCH3), 1.59–1.83 (m, 10 H, CH2), 1.41
(m, 8 H, CH2), 1.21 (dt, J = 18.87, 7.97 Hz, 3 H, CH3), 1.02 (t, J
= 7.28 Hz, 12 H, CH3) ppm. 13C NMR (63 MHz, CD3OD): δ =
163.06, 133.49, 130.35, 127.85, 124.48, 123.37, 122.90, 67.84, 59.50,
55.79, 54.68, 30.75, 24.79, 20.73, 13.97, 8.26 ppm. 31P NMR
(122 MHz, CD3OD): δ = 26.36 ppm. MS (ESI; MeOH): calcd. for
[M + H]+ 299.2, found 299.1; calcd. for [M + Na]+ 321.2, found
321.1.
General Procedure for the Combinatorial Screening by Using Plat-
form P1
In the Absence of the Target Anion: A solution of amines A and B
([A] = [B] + 3[P1B3]) was added to a solution of imine P1B3 in
CD3OD (2 mL) to make the final concentrations reported in the
experiments. This solution was kept at 65 °C until full equilibration
of the library was observed as evidenced by the absence of changes
1
in the H NMR spectrum.
In the Presence of the Target Anion: A mother solution of imine
P1B3 in CD3OD (2 mL) was split into two portions. Amines A, B
([A] = [B] + 3[P1B3]), and the target anion were added to one of
the two to make twice the final concentrations of these additives
reported in the experiments. The two solutions were then mixed in
different ratios, and the resulting solution was kept at 65 °C until
full equilibration of the library was observed, as evidenced by the
P2C: 1H NMR (250 MHz, CD3OD): δ = 8.98 (s, 1 H, CH=N),
7.88 (dd, J = 7.80, 1.24 Hz, 1 H, Ar), 7.6–7.1 (m, 8 H, Ar), 4.55
(s, 2 H, CH2Ar), 3.3–3.2 (m, 8 H, CH2), 1.8–1.6 (m, 10 H, CH2),
1.41 (m, 8 H, CH2), 1.3–1.1 (m, 3 H, CH3), 1.02 (t, J = 7.26 Hz,
12 H, CH3) ppm. 13C NMR (63 MHz, CD3OD): δ = 161.78,
140.66, 136.71, 133.24, 129.60, 129.35, 128.25, 124.22, 123.20,
122.40, 118.23, 65.99, 59.50, 50.01, 24.79, 23.06, 20.77, 13.98,
8.28 ppm. 31P NMR (122 MHz, CD3OD): δ = 26.65 ppm. MS
(ESI, CD3OD, negative mode): calcd. for [M]– 302.2, found 301.9.
1
absence of changes in the H NMR spectrum.
General Procedure for the Combinatorial Screening by Using Plat-
forms P2–P4: The two amines B and C (5 equiv.) were added to a
5 m solution (600 µL) of the appropriate platform in CD3OD in
P3B: 1H NMR (250 MHz, CD3OD): δ = 8.87 (s, 1 H, CH=N), 7.87
(dd, J = 1.7, 7.7 Hz, 1 H, Ar), 7.46 (m, 1 H, Ar), 7.1–6.9 (m, 2 H,
Ar), 4.1–4.0 (m, 2 H, CH2), 3.89 (s, 3 H, OCH3), 3.67 (t, J = 5.3 Hz, an NMR tube, and the solution was kept at 50 °C until full equili-
2 H, CH2), 3.26 (s, 9 H, N, CH3) ppm. 13C NMR (63 MHz, bration of the library was observed, as evidenced by the absence of
1
CD3OD): δ = 162.05, 141.68, 134.21, 128.09, 125.00, 121.79, changes in the H NMR spectrum.
Eur. J. Org. Chem. 2010, 3858–3866
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
3865

M. Martin, G. Gasparini, M. Graziani, L. J. Prins, P. Scrimin
FULL PAPER
E. V. Anslyn, J. Am. Chem. Soc. 2002, 124, 13731–13736; c)
L. A. Cabell, M. D. Best, J. J. Lavigne, S. E. Schneider, D. M.
Rerreault, M. Monahan, E. V. Anslyn, J. Chem. Soc. Perkin
Trans. 2 2001, 315–323; d) J. Chin, J. Oh, S. Y. Jon, S. H. Park,
C. Walsdorff, B. Stranix, A. Ghoussoub, S. J. Lee, H. J. Chung,
S.-M. Park, K. Kim, J. Am. Chem. Soc. 2002, 124, 5374–5379;
K. Niikura, A. Metzger, E. V. Anslyn, J. Am. Chem. Soc. 1998,
120, 8533–8534; e) A. C. McCleskey, M. J. Griffin, S. E.
Schneider, J. T. McDevitt, E. V. Anslyn, J. Am. Chem. Soc.
2003, 125, 1114–1115.
Supporting Information (see footnote on the first page of this arti-
cle): Library equilibration studies, additional simulations, and
binding studies of receptor P1B3 with different anions.
Acknowledgments
Support by the Ministry of University and Research (MUR) [con-
tract 2006039071 (PRIN 2006)] to P. S. is kindly acknowledged.
[18] H.-J. Schneider, A. Yatsimirsky, Principles and Methods in Sup-
ramolecular Chemistry, Wiley, New York, 2000.
[1] P. T. Corbett, J. Leclaire, L. Vial, K. R. West, J. L. Wietor,
J. K. M. Sanders, S. Otto, Chem. Rev. 2006, 106, 3652–3711.
[2] a) J. M. Lehn, Chem. Eur. J. 1999, 5, 2455–2463; b) S. Ladame,
Org. Biomol. Chem. 2008, 6, 219–226.
[3] a) S. J. Rowan, S. J. Cantrill, G. R. L. Cousins, J. K. M. San-
ders, J. F. Stoddart, Angew. Chem. Int. Ed. 2002, 41, 898–952;
b) S. Otto, R. L. E. Furlan, J. K. M. Sanders, Curr. Opin.
Chem. Biol. 2002, 6, 321–327.
[4] O. Ramstrom, J. M. Lehn, Nat. Rev. Drug Discov. 2002, 1, 26–
36.
[5] M. Hochgurtel, R. Biesinger, H. Kroth, D. Piecha, M. W. Hof-
mann, S. Krause, O. Schaaf, C. Nicolau, A. V. Eliseev, J. Med.
Chem. 2003, 46, 356–358.
[6] a) Z. Grote, R. Scopelliti, K. Severin, Angew. Chem. 2003, 115,
3951–3955; Angew. Chem. Int. Ed. 2003, 42, 3821–3825; b) K.
Severin, Chem. Eur. J. 2004, 10, 2565–2580.
[7] a) P. T. Corbett, S. Otto, J. K. M. Sanders, Chem. Eur. J. 2004,
10, 3139–3143; b) P. T. Corbett, S. Otto, J. K. M. Sanders, Org.
Lett. 2004, 6, 1825–1827; c) P. T. Corbett, J. K. M. Sanders, S.
Otto, J. Am. Chem. Soc. 2006, 128, 10253–10257; P. T. Corbett,
[19] Amine A instead of C has been used for platform P1 to mini-
mize steric hindrance during the recognition process.
[20] H.-J. Schneider, Angew. Chem. Int. Ed. 2009, 48, 3924–3977.
[21] Simulations were carried out with: P. Gans, A. Sabatini, A.
Vacca, Hyperquad Simulation and Speciation, version 3.2.24,
Protonic Software, UK, 2006. Details are given in the Support-
ing Information.
[22] For the consistency of this assumption, see also the binding
data reported in Table 1.
[23] This conclusion, valid for a platform-based library, cannot be
generalized (see ref.[9]).
[24] In a real system clearly one cannot know a priori what is the
best target.
[25] As is well known, acid catalysis does accelerate the equilibra-
tion process. However, we have noticed that the addition of
acid also alters the equilibrium composition,[14] especially in
the presence of a very low excess of amines. For this reason we
have preferred to run our experiment without the addition of
acid.
J. K. M. Sanders, S. Otto, Angew. Chem. 2007, 119, 9014–9017; [26] For an indirect analysis of such a DCC by using a stoichiomet-
Angew. Chem. Int. Ed. 2007, 46, 8858–8861; d) P. T. Corbett,
J. K. M. Sanders, S. Otto, Chem. Eur. J. 2008, 14, 2153–2166.
[8] a) J. S. Moore, N. W. Zimmerman, Org. Lett. 2000, 2, 915–918;
b) I. Huc, R. Nguyen, Comb. Chem. High Throughput Screen-
ing 2001, 4, 53–74.
[9] I. Saur, K. Severin, Chem. Commun. 2005, 1471–1473.
[10] L. Vial, R. F. Ludlow, J. Leclaire, R. Perez-Fernandez, S. Otto,
J. Am. Chem. Soc. 2006, 128, 10253–10257.
ric amount of recognition units, see: G. Gasparini, F. Bettin, P.
Scrimin, L. Prins, Angew. Chem. Int. Ed. 2009, 48, 4546–4550.
[27] G. Gasparini, M. Martin, L. J. Prins, P. Scrimin, Chem. Com-
mun. 2007, 1340–1342.
[28] a) T. Hesterkamp, M. Whittaker, Curr. Opin. Chem. Biol. 2008,
12, 260–268; b) P. J. Hajduk, J. Greer, Nat. Rev. Drug Discov.
2007, 6, 211–219; c) A. A. Alex, M. M. Fiocco, Curr. Top. Med.
Chem. 2007, 7, 1544–1567.
[11] G. Gasparini, L. J. Prins, P. Scrimin, Angew. Chem. Int. Ed.
2008, 47, 2475–2479.
[12] L. J. Prins, P. Scrimin, Angew. Chem. Int. Ed. 2009, 48, 2288–
2306.
[13] X. Y. Zhang, K. N. Houk, Acc. Chem. Res. 2005, 38, 379–385.
[14] N. Giuseppone, J.-M. Lehn, Chem. Eur. J. 2006, 12, 1715–1722.
[15] K. V. Kilway, J. S. Siegel, Tetrahedron 2001, 57, 3615–3627.
[16] K. J. Wallace, W. J. Belcher, D. R. Turner, K. F. Syed, J. W.
Steed, J. Am. Chem. Soc. 2003, 125, 9699–9715.
[17] For selected examples, see: a) T. Szabo, B. M. O’Leary, J. Re-
bek Jr., Angew. Chem. Int. Ed. 1998, 37, 3410–3413; b) M. Ko-
miyama, S. Kina, K. Matsamura, S. T. Sumaoka, V. M. Lynch,
[29] a) D. A. Erlanson, J. A. Wells, A. C. Braisted, Ann. Rev. Biphys.
Biomol. Struct. 2004, 33, 199–223; b) M. F. Scmidt, A. Isidro-
Llobet, M. Lisurek, A. El-Dahshan, J. Tan, R. Hilgenfeld, J.
Rademann, Angew. Chem. Int. Ed. 2008, 47, 3275–3278.
[30] G. Gasparini, M. Dal Molin, L. J. Prins Eur. J. Org. Chem.
2010, 2429–2440.
[31] A. Metzger, V. M. Lynch, E. V. Anslyn, Angew. Chem. Int. Ed.
Engl. 1997, 36, 862–865.
[32] C. Walsdorff, W. Saak, S. Pohl, J. Chem. Res. (S) 1996, 282–
283.
Received: December 30, 2009
Published Online: May 19, 2010
3866
www.eurjoc.org
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 3858–3866