A synthetic receptor motif designed for extended peptide conformations

Article abstract of DOI:10.1016/S0040-4020(00)00252-0

The intrinsic flexibility of short peptides makes them difficult targets for molecular recognition. Although in general too short to form stable secondary, structures, many are able to sample extended conformations. Using monomer cycloalkyl-1,2-trans-diamines alternated with cycloalkyl-1,2-trans- dicarboxylic acids, we have constructed a novel receptor motif to be complementary to extended tripeptides. Receptor design, synthesis, structural analysis and binding studies are presented. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0040-4020(00)00252-0

TETRAHEDRON
Pergamon
Tetrahedron 56 (2000) 3309±3318
A Synthetic Receptor Motif Designed for
Extended Peptide Conformations
*
Kevin Ryan, Leland J. Gershell and W. Clark Still
Department of Chemistry, Columbia University, Havemeyer Hall, New York, NY 10027, USA
Received 23 December 1999; accepted 22 March 2000
AbstractÐThe intrinsic ¯exibility of short peptides makes them dif®cult targets for molecular recognition. Although in general too short to
form stable secondary structures, many are able to sample extended conformations. Using monomer cycloalkyl-1,2-trans-diamines alter-
nated with cycloalkyl-1,2-trans-dicarboxylic acids, we have constructed a novel receptor motif to be complementary to extended tripeptides.
Receptor design, synthesis, structural analysis and binding studies are presented. q 2000 Elsevier Science Ltd. All rights reserved.
Introduction
regarding the three-dimensional structure of our receptors or
receptor±peptide complexes, we wondered if the predilec-
tion of some of our receptors for proline might be signi®-
cant. For example, does the conformational restriction of
proline contribute signi®cantly to the pre-organization of
the host±guest complex? While we design pre-organization
into our receptors, pre-organization in the target may also
enhance binding to the point where it is brought within the
limit of detection in our assay, provided there is at least a
modicum of hydrogen-bonding complementarity. An
equally well-matched tripeptide of greater ¯exibility might
fall short of this threshold. Another possibility is that a
tripeptide containing a proline-induced turn may be more
compact and therefore be more likely to match the size of a
receptor's binding pocket, thereby availing itself of a maxi-
mum number of hydrogen-bonds. Whatever the reason, this
trend has prompted us to take into consideration more
seriously the variable of peptide backbone conformational
preference when designing new molecules in our overall
pursuit of a collection of sequence-selective receptors that
covers the widest variety of sequences. Although we cannot
expect the tripeptide targets in our libraries to form well-
de®ned structures like a-helices and b-sheets, we can
expect many sequences to be capable of sampling extended
conformations, the antithesis of reverse turn structures. Here
we report the design and synthesis of a new receptor motif
intended to be complementary to the extended peptide
conformation, and its sequence-selective binding properties
as revealed through on-bead screening of an encoded tri-
peptide library.
The inherent ¯exibility of short peptides (2±15 amino acids)
makes them dif®cult targets for molecular recognition.
Arti®cial receptors designed to bind them face, in the case
of most sequences, a plethora of different unbound con-
formers at any given instant. A general rule in host/guest
chemistry is that a host must be pre-organized to bind a
target lest too much conformational entropy and enthalpy
be lost by the host (and the guest) upon binding to be
favorable.1 This means that, ideally and impossibly, a recep-
tor should be preorganized to bind many different peptide
conformers. A more realistic expectation is that a well
designed receptor should be able to bind a small family of
related conformers that are frequently sampled by the
unbound peptide. Despite this problem of target ¯exibility,
during the past few years we2 and others3,4 have made
progress in the sequence-speci®c recognition of short peptide
sequences in organic solvents and, to a lesser extent, in
water.5,6 Many of our own receptors have been constructed
from macrocycles designed to form pre-organized molecular
clefts lined with hydrogen-bond donors and acceptors.
In the process of screening our receptors against encoded
combinatorial tripeptide libraries, we have noted a modest
but recurring preference by some receptors for heterochiral
tripeptide sequences that contain proline (see Refs. 5, 7±10
for example). Proline is the amino acid most often
associated with b-turn structures in peptidesÐespecially
heterochiral peptides11Ðand its pyrrolidine ring makes it
the most conformationally restricted of the proteinogenic
amino acids. Although very little information is available
Results
Receptor design
Keywords: receptors; peptides; combinatorial chemistry; molecular recog-
nition.
* Corresponding author. Tel.: 11-212-854-2577; fax: 11-212-854-4231;
e-mail: clark@chem.columbia.edu
Our design is based on oligomers formed by the condensation
0040±4020/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020(00)00252-0

3310
K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
We veri®ed the conformational preference for the unbound
receptor motif by studying the crystal structure of
compound 1, a close analog of the receptor prototype.
Compound 1, a dicarboxylic acid-terminated version of a
1.5-mer receptor motif, crystallized in an extended confor-
mation that was virtually superimposable on an energy
minimized model of the receptor. As shown in Fig. 2, the
cyclohexyl rings de®ne a ¯at V-shaped surface. Cutting
through the middle of this surface is the hydrogen-bond
donor±acceptor array, which projects both above and
below the plane. In the lattice, each molecule is extensively
hydrogen bonded to others above and below the plane of the
cyclohexyl rings (not shown). We found this class of recep-
tors to be prone to insolubility and, when solubilized, to
non-crystalline aggregation in aprotic solvents. This
behavior may be due to the predisposition towards an
extended though imperfectly formed lattice driven by the
formation of hydrogen-bonds and permitted by the ¯at, rela-
tively rigid conformational preference of this diastereomer.
Figure 1.
of cycloalkyl-1,2-trans-dicarboxylic acids of one stereo-
chemistry with cycloalkyl-1,2-trans-diamines of the oppo-
site stereochemistry (Fig. 1). Molecular modeling12
indicated that oligomers constructed from these alternating
monomers should prefer to adopt a ¯at, extended conforma-
tion. As depicted in Fig. 1, the dipole±dipole directed align-
ment of the amide linkages presents an alternating array of
hydrogen-bond donors (amide N±H) and acceptors (amide
carbonyl) above and below the plane de®ned by the rings.
This arrangement is reminiscent of the peptide backbone of
the antibiotic vancomycin, which binds the (D)Ala±(D)Ala
dipeptide in an extended conformation.13 Models show that
this diastereomer (or its enantiomer) favors the extended
structure while other all-trans diastereomers do not. Models
also show that the distance between alternating hydrogen-
bond donor and acceptor is such that the array is comple-
mentary to that of a peptide in the extended conformation,
with the fragment shown large enough to span a tripeptide
sequence.
An important feature of the receptor design that was veri®ed
by the crystal structure is that the receptor has two discrete
faces, each presenting an alternating hydrogen-bond donor±
acceptor array. The extended peptide conformation itself is
in this respect similar as it has two discrete edges of alter-
nating hydrogen-bond donors and acceptors which, for
example, allow the extended conformation to assemble
into b-sheets of more than two strands in proteins. In an
effort to recognize simultaneously both edges of an
extended peptide conformation, we elaborated the design
by linking two receptors with an aromatic spacer to form
a bidentate version of the receptor. Our intention was to
allow two receptors to converge on, and hydrogen bond
to, a single extended peptide. Three such compounds, 5, 6,
and 7, are shown in Fig. 3. The sulfonamide linkage was
Figure 2.

K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
3311
Figure 3.
selected for the tetrahedral con®guration about the sulfur
atom. Modeling studies indicate that this, and not the planar
carboxamide linkage of analog 8, allows the two receptor
arms to align in a parallel manner to form a cavity in which
an extended peptide might bind. The third side of the cavity
may be de®ned either by the edge or more likely the face of
the aromatic ring.14 In addition to joining the two receptors
and allowing a convergent conformer, the linkers were
selected for their size and rigidity. Conformational search-
ing in the absence of a guest indicated that the aromatic
spacers in compounds 5 and 6, and to a lesser extent 7,
should prevent intramolecular hydrogen bonding
between the two receptor arms, thereby leaving them
available and far enough apart to accommodate a tripeptide
guest.
also found to be less prone to insolubility and aggregation.
In designing receptors for a general backbone conformation
we have avoided the more dif®cult task of trying to design
sequence speci®city. By using a combinatorial library of
tripeptides we initially leave this question to the assay. In
the event that a receptor shows a preference for a certain
sequence or family of sequences, study of these new leads
can be used to design the next generation of receptors based
on this novel motif. Compounds 9±11 are partial exceptions
to this because, by virtue of their sulfonic acid group, they
are expected to show some preference for tripeptides
containing positively charged side chains.
Receptor synthesis
As shown in Scheme 1, the core of the receptor structure
was made by condensing the appropriate cycloalkyl
diamine, either 13, 16 or 24 with a singly activated form
of the diacid, either anhydride 12, penta¯uorophenol ester
18 or acid ¯uorides 27 and 28, respectively. The chiral
cycloalkyl diamines were either purchased (13) or made
from the known compounds 1415 or 20.16 To prepare 16,
the benzyl group of 14 was exchanged for the Teoc group
(2-(trimethylsilyl)ethoxycarbonyl) using an established
The one- and two-arm receptors synthesized for this study
are shown in Fig. 3. Disperse Red (DR) dye-conjugates were
made for testing against an encoded tripeptide library
displayed on polystyrene beads. A pyrrolidine-3S,4S-
diamine was used in place of the cyclohexyldiamine of 1
to allow attachment of the dye in receptors 3 and 4, and to
provide a site for elaboration into the bidentate form of
receptors 5±11. The cyclopentyl diamine compounds were

3312
K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
Scheme 1.
procedure.17 This permitted the subsequent catalytic hydro-
genation of the azide groups to yield diamine 16. Two
different diamino pyrrolidines were needed in order to
append the DR to different sites (receptor 2 vs. 3) on the
receptor core. In the case of receptor 2, the dye was attached
to the cyclohexyl rings using a tyramine (Tyr) linker made
benzene-1,3-dicarboxylic acid. The sulfonyl chlorides
were either purchased (7) or made according to standard
procedures.19,20 The single-arm receptor sulfonic acids 9,
10 and 11 were isolated from the coupling reactions during
silica gel chromatography.
from
a Mitsunobu coupling between Boc-protected
tyramine and DR to yield 17 after TFA deprotection and
neutralization. For receptor 3 and its diastereomer 4, the DR
was attached to the pyrrolidine ring nitrogen of diazide 22
using DR and carbonyl diimidazole.18 In this case, the
diazide was subsequently reduced with triphenylphosphine
(to preserve the diazo group) before coupling 24 to the acid
¯uorides 27 and 28.
Results and Discussion
Side-chain protected library
The receptor±dye conjugates were screened against two
encoded, N-acetylated tripeptide librariesÐone with the
amino acid side chains protected and one with the side
chains deprotected. The results are shown in Table 1. In
the case of the side-chain protected library assayed in 5%
N-methylpyrrolidinone (NMP)/chloroform, the basic recep-
tor motif as represented by compounds 2 and 3 showed
sequence selective binding of the N-acetylated dipeptide
In order to make the two-arm receptors 5±8, the Teoc group
of receptor 2 was removed to yield 19, which was then
coupled to the various aromatic disulfonyl chlorides or, in
the case of receptor 8, the dipenta¯uorophenol ester of

K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
3313
Table 1. Tripeptide library binding results (notes: general conditions: 5% NMP/buffered chloroform, .48 h, rt.; library structure: Ac-AA1-AA2-AA3-
NH(CH2)5CONH-polystyrene)
Cpd
Conc. mM
Lib. copies
Side-chain protectedh
Side-chain deprotected
AA3
AA2
AA1
Freqe
Selectivityb
AA3
AA2
AA1
Freqe
Selectivityb
2
99
30
3
3
l-Asn
l-Asn
d-Ser
d-Ser
d-Asn
l-Asn
l-Asn
l-Asn
l-Asn
l-Asn
l-Asn
l-Asn
l-Asn
l-Asn
l-Asn
l-Asn
d-Val
d-Val
d-Val
d-Val
d-Gln
d-Val
d-Val
d-Val
d-Val
d-Val
d-Val
d-Val
d-Val
d-Val
d-Val
d-Val
l-Ala
d-Ala
d-Ala
d-Val
d-Val
d-Ala
d-Ser
d-Ala
d-Ser
l-Ala
l-Val
d-Ala
d-Val
d-Val
d-Ser
d-Ala
4/4
3/6
1/6
1/6
1/6
5/8
3/8
2/5
2/5
1/5
2/4
1/4
1/4
2/2
2/2
2/2
E
E
l-Val
d-Val
Gly
d-Ala
l-Ser
N/A
l-Gln
d-Lys
d-Lys
d-Lys
d-Val
l-Gln
d-Ala
d-Ala
d-Val
d-Val
2/2
1/4
1/4
1/4
1/4
G
G
2a
3
3
65
50
6
3
G
G
N/A
f
5
6
25
30
6
3
G
E
l-Val
l-Val
d-Val
l-Ser
l-Val
l-Val
l-Val
l-Ser
l-Ser
l-Val
l-Val
l-Val
l-Val
d-Lys
d-Val
d-Ala
l-Val
N/A
l-Gln
l-Gln
d-Gln
d-Val
l-Gln
l-Gln
l-Gln
d-Val
d-Val
l-Gln
l-Gln
l-Gln
l-Gln
d-Val
d-Val
d-Val
l-Gln
l-Gln
d-Ser
l-Ala
d-Val
l-Gln
l-Ser
d-Ala
3/6
1/6
1/6
1/6
1/5
1/5
1/5
1/5
1/5
5/8
2/8
1/8
2/2
2/5
1/5
1/5
1/5
G
G
6
7
20
25
3
6
l-Asn
l-Asn
l-Asn
d-Val
d-Val
d-Val
d-Ala
l-Val
3/5
1/5
1/5
G
c
d-Val
c
l-Val
l-Lys
l-Asn
4/4
M
l-Gln
d-Ala
d-Ser
l-Gln
l-Val
d-Lys
d-Lys
l-Gln
M
8
10
30
3
6
l-Asn
l-Asn
l-Asn
d-Gln
d-Pro
l-Asn
l-Asn
l-Asn
l-Asn
l-Asn
l-Ala
d-Val
d-Val
d-Val
l-Asn
l-Asn
d-Val
d-Val
d-Val
d-Val
d-Val
l-Val
d-Ala
l-Ala
d-Ala
d-Val
d-Val
d-Ala
d-Lys
d-Ser
l-Ser
l-Val
d-Gln
2/2
2/11
2/11
4/11
3/11
2/5
1/5
1/5
1/5
2/6
E
P
G
P
9d
10
11
20
25
3
6
E
M
d-Ala
d-Ser
d-Lys
d-Val
d-Val
d-Val
d-Lys
d-Lys
d-Val
4/8
2/8
2/8
Pg
4/6
a ,1% NMP.
b EˆExcellent (decoded beads were the only beads with red color. Other beads had no signi®cant color) GˆGood (decoded beads were the only beads with
red color. 5±15% of other beads had very light orange color) MˆModerate (decoded beads were the only beads with red color. More than 15% of other
beads had light to moderate orange color) PˆPoor (.10% beads had red color).
c Unreadable code.
d A representative portion (50%) of the darkest beads were sampled. Many other red and orange beads present.
e Number of times this sequence was found/total number of darkest red beads.
f Beads from the deprotected library induced precipitation of this receptor.
g Due to poor selectivity, only 30% of the darkest beads were picked and decoded.
h Amino acid protecting groups: Ser(OtBu), Lys(Boc), Gln(Trt), Asn(Trt); N/Aˆnot assayed.
(l)Asn(trt)-(d)Val, with a preference for any of four amino
acidsÐ(d/l)Ala, (d)Val or (d)Ser(OtBu)Ðin the third (C-
terminus) position. The degree of selectivity was good to
excellent, as judged by the small number of red beads and
the near absence of color in the remaining, non-red beads at
the equilibrium concentrations tested, 30±100 mM. The
core of the receptor itself appeared to be responsible for
binding since changing the site of the attachment of the
dye (compound 2 vs. 3) had little in¯uence on the sequences
selected. Furthermore, the all S,S-trans diastereomer 4
showed no binding at concentrations up to 100 mM. Thus,
this new receptor motif, despite its low core molecular
weight (,500), ef®ciently selected a family of four tripep-
tides out of a collection of 153ˆ3,375.21
if only for steric reasons. As shown in Table 1, with the
exception of 7, this was clearly not the case. Instead, the
same (l)Asn(trt)-(d)Val dipeptide was selected, again with
good to excellent selectivity, and with an almost identical
preference for the same third (C-terminal) amino acid.
Although we cannot rule out the possibility that 2 and 3
might bind with a stoichiometry of 2:1 receptor:tripeptide
and thus mimic a bidentate form, it seems unlikely at micro-
molar concentrations. Instead, it is possible that conformers
where the two receptor arms of 5 and 6 are aligned parallel
(to form a cleft) are not favored and that the receptor arms
act independently. Results with receptor 8 support this.
Here, the tetrahedral sulfonamide linkages have been
replaced by planar carboxamides, a change that molecular
models indicated would greatly disfavor cleft formation,
insuring that the arms would act independently. As
expected, receptor 8 bound the same sequence as the single
We had anticipated that the linkers of the bidentate receptors
5±8 would in¯uence the sequence speci®city of the receptor,

3314
K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
Figure 4.
arm receptors. Further evidence that the receptor arms in the
bidentate receptor act independently was provided by the
single arm receptor sulfonic acids, 9±11. These receptors,
although in general less selective, strongly favor the
(l)Asn(trt)-(d)Val dipeptide. Compound 9 was unique in
that it permitted a frame-shift of the dipeptide to the AA2-
AA3 position. Nevertheless, the possibility that a tripeptide
could be sandwiched in between the two arms simul-
taneously without changing the sequence speci®city cannot
be completely ruled out by these results and remains an
interesting, if unlikely, possibility. Nor can we completely
rule out the possibility that two arms from different
bidentate receptors converge on a single peptide.
in the speci®c sequence context of (d)Lys-(d)Val or
(d)Val-(d)Lys. We thus conclude that in combination with
a strong determinant of speci®city such as the sulfonate
anion, the core receptor's sequence speci®city can be modu-
lated. It should be noted that the (l)Val-(l)Gln-(l)Gln
sequence appeared once with sulfonic acid 9. It is possible
that this sequence is still able to bind to the receptor arm but
that the (d)Lys-(d)Val or (d)Val±(d)Lys dipeptides, whose
beads were of a deeper red than those selected by 2, 6, 7 and
8, bound more strongly and were picked and decoded
preferentially.
The novel receptor motif we have designed and tested here
was made from building blocks that resemble the R,R-trans-
cycloalkyl b-amino acids recently employed in the
construction of well-de®ned oligomeric helices, or
foldamers (Fig. 4).22,23 A comparison of the two structures
offers an interesting example of how small changes at the
monomer level can bring about large changes at the
oligomer level. Whereas the foldamer researchers used
homochiral b-amino acids, we used two different alternating
monomersÐan R,R-diacid (e.g. 12) and an S,S-diamine
(e.g. 16). Both designs exploited the conformational restric-
tion of the 1,2-trans-disubstituted cycloalkyl ring system to
construct a relatively rigid, well-de®ned structure. In our
design, the conformational restriction is enforced by the
cycloalkyl rings and by the up±down alignment of the
amide dipoles. Strong intramolecular hydrogen-bonds are
prevented and the amide groups are left free to hydrogen-
bond intermolecularly with a guest tripeptide, and to do so
on either of two discrete receptor faces. The foldamers, on
the other hand, offer an impressive example where intra-
molecular hydrogen-bonds are used to stabilize a designed
secondary structure, leaving them in this case unavailable
for molecular recognition. Taken together, the two
examples demonstrate how a careful consideration of mono-
mer stereochemistry, conformational restriction and
hydrogen-bonding can lead to the successful design of
oligomers having a de®ned shape and consequently well-
suited to a particular goalÐeither molecular recognition or
de novo secondary structure design.
In the meta substituted disulfonamide 7, models show that
interaction between the arms in the unbound form is more
favorable than in the other bidentate receptors. The aligned
but partly intramolecularly hydrogen-bonded arms may
present a very different binding surface and thus explain
in part the selection by this compound of a quite different
tripeptide sequence: (l)Val±(l)Lys(Boc)±(l)Asn(trt).
Side chain deprotected library
The af®nity of the new receptor motif for (l)Asn(trt)-(d)Val
was lost upon removal of the amino acid protecting groups,
as shown in the results for the side-chain deprotected library
in Table 1. Binding experiments revealed that in the depro-
tected library this sequence was replaced largely by the
homochiral sequence (l)Val-(l)Gln, with a strong prefer-
ence for a second (l)Gln in the third position. Once again,
the preference of the core receptor 2 extended to the
bidentate receptors, indicating that a functional cleft is
either not formed or does not change the sequence speci®-
city found for the isolated receptor arm. Contrary to the
results with the protected library, this speci®city even held
for the meta-substituted disulfonamide 7. In general, the
sequence selectivity was slightly less impressive against
the deprotected library as more beads had a higher back-
ground orange color than in the protected library under the
same conditions.
As shown in Fig. 1, the receptor prototype was initially
designed to be complementary to a peptide in an extended
conformation. However, efforts to co-crystallize several
receptor analogs with cognate tripeptides were not success-
ful, and the poor solubility of the receptor core prevented
detailed NMR studies. We have therefore not yet been
able to ascertain whether or not the peptides are indeed
bound in an extended conformation. The excellent
sequence-selectivity, which may in fact argue against
The single arm receptor sulfonic acids 9±11 bear a single
negative charge under the conditions of our assay. Although
this proved to have little effect when tested against the
protected library, we expected these receptors to show a
preference for lysine-containing peptides when tested
against the deprotected library. This was indeed found to
be the case for the two sulfonic acid receptors tested, 9 and
11. However, rather than binding lysine-containing peptides
indiscriminately, these receptors only bound that amino acid

K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
3315
speci®c binding to a conformation accessible to, and
frequently sampled by, many sequences, was somewhat
surprising given the low molecular weight of the receptor
and the clear lack of a binding cleft at least in the single arm
cases. Nevertheless, most receptor analogs displayed good
to excellent selectivity and, within each library, very similar
speci®city. The single arm receptor sulfonic acids 9 and 11
comprised a special subset where it was shown that the
speci®city could be altered with only a small loss in selec-
tivity by appending a functional group (here, sulfonate)
possessing distinct amino acid side chain preference
(unprotected lysine). This latter result is promising
because it leads to the possibility of using this novel recep-
tor motif in a modular, combinatorial approach in the search
for receptors with tunable speci®city for other tripeptide
sequences.
hexanes, vol/vol) to yield 15 (747 mg, 76%): IR (neat): nmax
2955, 2895, 2102, 1694, 1428, 1351, 1250, 1178,
1110 cm21 1H NMR (400 MHz, CDCl3, no TMS) d
;
[ppm] 4.19 (m, 2H), 3.96 (br s, 2H), 3.68 (m, 2H), 3.45
(m, 2H), 0.98 (m, 2H), 0.02 (s, 9H); Chemical ionization
(CI) (NH3)-LRMS, calcd for C10H19N7O2Si (M1NH3)
314.42, found 315.
(3S,4S)-Diamino-(N-(trimethylsilyl)ethoxycarbonyl)pyr-
rolidine (16). 37 mg of 10% Pd/C was suspended in an
ethanol solution (1.5 ml) of 15 (397 mg, 1.33 mmol) under
argon. The argon was replaced by a hydrogen balloon and
the reaction was stirred for 18 h. The mixture was ®ltered
through celite and puri®ed by silica gel ¯ash chromato-
graphy (49.5% methanol/49.5% ethyl acetate/1% triethyl-
amine (TEA) to yield 16 (224 mg, 69%): IR (thin ®lm): nmax
1
3347 (br), 2952, 2896, 1679, 1434, 1359, 1249 cm21; H
NMR (400 MHz, CDCl3, no TMS) d [ppm] 4.15 (m, 2H),
3.70 (m, 2H), 3.04 (m, 4H), 1.49 (br s, 5.4H, including 4
amino H), 0.98 (m, 2H), 0.01 (s, 9H); CI(NH3)-LRMS, calcd
for C10H23N3O2Si, 245.39, found 246.
Experimental
General methods
Penta¯uorophenol ester (18). To a stirred solution of 12
(123 mg, 0.8 mmol) in dichloromethane (DCM, 2 ml) at 08C
was added dropwise a solution of 17 (346 mg, 0.8 mmol,
5 ml DCM) and TEA (0.11 ml, 0.8 mmol) over a period of
20 m. After 2 h the solvent was evaporated in vacuo and the
crude product was puri®ed by ¯ash chromatography
(gradient of 1.5±5% methanol/DCM/1 drop glacial acetic
acid per 100 ml solvent). Fractions containing the product
(Rfˆ0.4 in 5% MeOH/0.5% HOAc/DCM) were collected
and evaporated in vacuo. The residue (453 mg, 0.76
mmol) was dissolved in DCM (8 ml) along with 139 mg
penta¯uorophenol (0.76 mmol). To this was added a 4 ml
DCM solution of 1-(3-dimethylaminopropyl)-3-ethylcarbo-
diimide hydrochloride (EDC, 160 mg, 0.83 mmol, 1.1
equiv.). After 4 h the reaction was evaporated in vacuo
and puri®ed by silica gel ¯ash chromatography (3%
acetone/DCM) to yield 18 (437 mg 73%): IR (thin ®lm):
nmax 2931, 2861, 1780, 1650, 1600, 1518, 1338,
Unless otherwise indicated, all reagents and solvents were
purchased from Aldrich or Fluka and were of the highest
available purity and dryness. Thin layer chromatography
was done on pre-coated Silica Gel 60 plates (EM Science),
and solvent system A denotes: 8 benzene/8 chloroform/3
MeOH/0.3 water, vol/vol.
(1R,2R)-(2)-1,2-Cyclohexane dicarboxylic acid anhydride
(12). The racemic diacid was purchased from Aldrich and
resolved according to the literature procedure.24 Brie¯y, the
diacid was mixed with a molar equivalent of (R)-(1)-a-
methylbenzylamine in hot n-propanol and repeatedly
recrystallized from that solvent. After separation from the
amine by extraction, the anhydride was formed in THF by
the addition of 1.0 equiv. of dicyclohexyl carbodiimide
(DCC) to the isolated free diacid.25 A portion of the chiral
diacid precursor used initially in these studies was provided
1
as a gift from Merck±Schuchardt. The H NMR and the
speci®c rotation of the resolved material were identical to
the sample provided by Merck±Schuchardt.
1
1134 cm21; H NMR (400 MHz, CDCl3) d [ppm] 8.33 (d,
2H), 7.90, (dd, 4H), 7.08 (d, 2H), 6.81 (dd, 4H), 5.48 (br t,
1H), 4.16 (m 2H), 3.82 (m, 2H), 3.61 (q, 2H), 3.46 (m, 2H),
3.06 (dt, 1H), 2.72 (ddd, 2H), 2.38 (dt, 1H), 2.29 (br d, 1H),
1.83 (br m, 3H), 1.49 (m, 4H), 1.29 (t, 3H); FAB-LRMS
calcd for C38H36F5N5O6 753.71, found 754.
Dicarboxylic acid (1). To a stirred solution of 13 (29 mg,
0.25 mmol) and triethylamine (0.105 ml, 0.76 mmol) in
DMF (0.4 ml) under argon was added 12 (78 mg,
0.51 mmol) in a mixture of DCM (0.4 ml) and DMF
(0.4 ml). The resulting suspension was stirred overnight
and the solvents were evaporated under reduced pressure.
The residue was dissolved in saturated sodium bicarbonate
solution and washed with chloroform. Concentrated HCl
was added dropwise to precipitate the product, which was
then washed with water and dried under vacuum to yield
65 mg of a white solid (61%). Crystals for X-ray analysis
were prepared from DMSO by vapor diffusion with water.
Teoc-protected one-arm receptor (2). A solution of 18
(216 mg, 0.29 mmol) in dry DMF (2 ml) was added drop-
wise under argon over a 2 min period to a DMF (0.4 ml)
solution containing 16 (36.5 mg, 0.14 mmol) and
diisopropylethyl amine (DIEA, 0.052 ml, 0.29 mmol). The
reaction was followed by TLC (5% MeOH/95% chloro-
form/0.1% acetic acid), which, in addition to the product,
showed formation of the intramolecularly cyclized imide
product from 18 (Rfˆ0.9) and, initially, the mono-
substituted product (Rfˆ0.1). After stirring for 36 h the
precipitated product (Rfˆ0.4) was ®ltered off, washed
with DMF and dried in vacuo to yield a dark red solid, 2
(150 mg, 75%). Flash chromatography of the ®ltrate residue
(3S,4S)-Diazido-(N-(trimethylsilyl)ethoxycarbonyl)-
pyrrolidine (15) (Ref. 17). To a stirred, chilled (2608C)
solution of 14 (0.8 g, 3.28 mmol) in THF (6.5 ml) was
added dropwise a solution of Teoc-Cl (1.78 g, 9.9 mmol)
over a period of 20 min. The reaction was allowed to
warm to rt overnight. After 24 h the mixture was puri®ed
by silica gel ¯ash chromatography (10% ethyl acetate/90%
1
yielded an additional 14 mg: H NMR (400 MHz, DMSO-
d6) d [ppm] 8.66 (d, 4H), 7.95 (m, 6H), 7.84 (d, 4H), 7.71 (br
dd, 2H), 7.05 (m, 4H), 6.92 (d, 4H), 6.82 (d, 4H), 4.22 (br

3316
K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
dd, 4H), 3.95 (m, 2H), 3.90 (m, 2H), 3.83 (br dd, 4H), 3.59
(q, 4H), 3.46 (m 2H), 3.08 (br dd, 4H), 3.02 (m, 2H), 2.58
(m, 4H), 2.30 (br m, 4H), 1.54 (br s, 6H), 1.20 (t1m, 16H),
0.75 (dd, 2H), 20.12 (s, 9H); FAB-HRMS, calcd for
C74H93N13O12S 1383.6836, found 1383.6884.
(3S,4S)-N-Boc-3,4-diazidopyrrolidine (21). To a solution
of (3R,4R)-N-Boc-pyrrolidinediol dimesylate 20 (8.60 g,
23.93 mmol) in DMF (200 ml) was added NaN3 (16.8 g,
0.258 mol). The mixture was heated to 908C for 24 h, then
evaporated to dryness in vacuo. The residue was diluted
with EtOAc (100 ml) and washed with water (3£20 ml).
The aqueous portion was back-extracted with EtOAc
(1£20 ml). The organic portions were combined, washed
with brine (1£10 ml), dried (MgSO4) and concentrated.
The residue was ®ltered through a plug of silica gel (20%
EtOAc/hexanes) and concentrated to give 21 as a yellow oil
(4.12 g, 68%). TLC Rfˆ0.81 (25% EtOAc/hexanes); IR
Deprotected one-arm receptor (19). Tri¯uoroacetic acid
(TFA, 0.2 ml) was added to a stirred suspension of 2 in
DCM (2 ml). The reaction was followed by TLC (solvent
system A, Rfˆ0.45). After 15 h the solvent (and excess
TFA) was evaporated under reduced pressure to yield a
dark red solid, 19, TFA salt (134 mg, 95 %): IR (KBr):
nmax 3286, 2934, 2861, 1654, 1612, 1552, 1510, 1437,
1
(®lm) nmax 2970, 2097, 1695 cm21; H NMR (400 MHz,
1395, 1339 cm21
;
1H NMR (400 MHz, DMSO-d6) d
CDCl3) d 3.94 (m, 2H), 3.65 (m, 2H), 3.34 (m, 2H), 1.45
(s, 9H); LRMS (CI) Calcd for C9H15N7O2 253.26, found
254.
[ppm] 8.78 (br m, 2H), 8.35 (d, 4H), 7.91 (d, 4H), 7.85 (d,
4H), 7.70 (dd, 2H), 7.05 (d, 4H), 6.95 (d, 4H), 6.82 (d, 4H),
4.22 (dd, 4H), 4.09 (m, 2H), 3.86 (m, 4H), 3.59 (q, 4H), 3.38
(br m, 2H), 3.09 (m, 4H), 2.82 (br m, 2H), 2.55 (m, 4H), 2.29
(br m, 4H), 1.68 (br d, 6H), 1.18 (t1m, 16H); FAB-LRMS,
calcd for C68H81N13O10 (free base) 1240.5, found 1242.
(3S,4S)-3,4-Diazidopyrrolidine, TFA salt (22). To a stir-
red solution of 21 (1.00 g, 3.95 mmol) in DCM (25 ml) was
added tri¯uoroacetic acid (2.5 ml). After TLC indicated
complete conversion, the solvent and excess TFA were
removed in vacuo, to yield 21 (1.05 g, quant.), which was
used without further puri®cation. TLC Rfˆ0.4 (10% MeOH/
General sulfonamide coupling procedure (e.g. 5 and 9).
To a stirred solution of 19 (12.3 mg, 9 mmol) in N-methyl-
pyrrolidinone (NMP, 0.4 ml) was added DIEA (5 ml,
27 mmol), followed by naphthyl-2,7-disulfonyl chloride
(1.5 mg, 4.5 mmol) in NMP (20 ml). The reaction was
followed by TLC (solvent system A). After 9 h, additional
disulfonyl chloride was added (0.49 mg, 0.3 mmol, in 11 ml
DMF). After stirring for a total of 45 h, enough silica gel to
absorb all the reaction solvent was added. The entire reac-
tion mixture was transferred to the top of a chloroform/silica
gel ¯ash chromatography column. The products were eluted
with a 0±5% MeOH gradient in chloroform. (Note: The
poor solubility of 19 and the products necessitated the
use of the polar, high-boiling solvent NMP. In some cases
the NMP co-eluted with the puri®ed product and could not
be removed, e.g. 5. The sulfonyl chlorides also slowly
decomposed in this solvent.)
1
DCM); H NMR (400 MHz, DMSO-d6) d 9.43 (br s, 2H),
4.53 (dd, 2H), 3.39 (dd, 2H), 3.23 (dt, 2H).
Disperse red-pyrrolidine diazide (23). To a solution of
disperse red (300 mg, 0.95 mmol) in THF (15 ml) was
added 1,10-carbonyldiimidazole (189 mg, 1.15 mmol).
After 6 h a solution of 22 (TFA salt, 400 mg, 1.50 mmol)
and diisopropylethylamine (0.30 ml, 1.72 mmol) in THF
(2 ml) was added. The mixture was brought to re¯ux over-
night. The crude mixture was evaporated to dryness in
vacuo and puri®ed by silica gel chromatography (40%
EtOAc/hexanes) to give 23 as a red powder (461 mg,
0.934 mmol, 98%]. TLC Rfˆ0.13 (25% EtOAc/hexanes);
IR (KBr) nmax 2100, 1694, 1604, 1516, 1427, 1390, 1338,
1142, 1106 cm21; 1H NMR (400 MHz, DMSO-d6) d 8.34 (d,
2H), 7.91 (d, 2H), 7.83 (d, 2H), 6.91 (d, 2H), 4.30 (m, 2H),
4.21 (t, 2H), 3.71 (t, 2H), 3.50 (m, 4H), 3.24 (m, 2H), 1.15 (t,
3H); LRMS (CI) calcd for C21H23N11O4 493.48, found 494.
Receptor (5): 2.2 mg (including an unknown amount of
NMP) Rfˆ0.6 (10% MeOH/chloroform); FAB-LRMS,
calcd for C146H166N26O24S2 2733.2, found 2734.
Receptor (9): 5.5 mg (81%), Rfˆ0.1 (10% MeOH/chloro-
form); FAB-LRMS, calcd for C78H87N13O15S2 1510.74,
found, 1512.
Receptor (6): 5.0 mg (51%), Rfˆ0.74 (10% MeOH/
DCM); MALDI-TOF MS, calcd for C142H164N26O24S2
2683.11, found 2706.99 (M1Na).
Receptor (10): 2.1 mg (38%), Rfˆ0.4 (10% MeOH/
DCM); MALDI-TOF MS, calcd for C74H85N13O15S2
1459.57, found 1497.67 (M1K).
Receptor (7): 5.4 mg (including an unknown amount of
NMP) Rfˆ0.75 (eluted ®rst in 5% MeOH/chloroform
then in solvent system A: FAB-LRMS, calcd for
C142H164N26O24S2 2683.1, found 2684.
Disperse red-pyrrolidine diamine (24). A solution of 23
(40 mg, 0.08 mmol) and triphenylphosphine (50 mg,
0.19 mmol) in toluene (1.5 ml) was brought to re¯ux.
After 10 min, the reaction mixture was cooled to 508C,
and a solution of water (0.025 ml) in THF (0.3 ml) was
added. Re¯ux was resumed and continued overnight.
After evaporation of the solvents the crude product was
puri®ed by ¯ash chromatography (1% Et3N/10% MeOH/
DCM) to yield a red solid (34 mg, 0.077 mmol, 95%).
TLC Rfˆ0.29 (1% Et3N/10% MeOH/DCM); IR (KBr)
nmax 1509, 1438, 1193, 1120, 858, 785, 722, 695,
1
651 cm21; H NMR (400 MHz, DMSO-d6) d 8.35 (d, 2H),
7.93 (d, 2H), 7.84 (d, 2H), 6.93 (d, 2H), 4.17 (t, 2H), 3.70 (t,
2H), 3.52 (m, 4H), 2.93 (m, 4H), 1.16 (t, 3H); LRMS (CI)
calcd for C21H27N7O4 441.48, found 442.
Receptor (11): 6.2 mg (67%): Rfˆ0.35 (eluted as in 7);
FAB-HRMS, calcd for C74H86N13O15S2 (M11)
1460.5808, found 1460.5769.
(1R,2R)-Cyclohexanedicarboxylic acid mono-n-propyl-
amide (25). To a stirred solution of 12 (150 mg, 0.97
mmol) in DCM (10 ml) was added dropwise n-propylamine
(0.25 ml, 3.02 mmol). A white precipitate soon appeared
and the reaction mixture was stirred for 2 h, followed by
Receptor (8). Made similarly, but the dipenta¯uorophenol
ester of benzene-1,3-dicarboxylic acid was used in place
of the aromatic disulfonyl chlorides. 8.7 mg (90 %):
FAB-LRMS, calcd for C144H164N26O22 2611.0, found
2612.

K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
3317
acidi®cation with 1 M HCl (10 ml). The layers were sepa-
rated, and the aqueous phase was extracted with EtOAc
(3£5 ml). The organic portions were combined, washed
with brine (1£5 ml), dried (MgSO4), ®ltered, and concen-
trated to give 25 as an off-white solid (183 mg, 0.86 mmol,
88%). TLC Rfˆ0.5 (75% EtOAc/hexanes); IR (KBr) nmax
1H NMR (400 MHz, DMSO-d6) d 11.86 (br s, 1H), 7.69 (t,
1H), 2.93 (m, 2H), 2.41 (td, 1H), 2.28 (td, 1H), 1.91 (m, 1H),
1.76 (m, 1H), 1.68 (m, 2H), 1.34 (m, 2H), 1.20 (m, 4H), 0.80
(t, 3H); LRMS (CI) calcd for C11H19NO3 213.27, found 214.
receptor was estimated by taking an absorbance measure-
ment (chloroform, 476 nm, ext. coeff.<35,938 per chromo-
phore) of the solution after diluting the NMP to 0.5%, vol/
vol.
3309, 2936, 1694, 1645, 1549, 1276, 1261, 921, 681 cm21
;
Acknowledgements
This work was supported by Grant CHE97-07870 from the
National Science Foundation. Fellowships from the
National Institutes of Health (from the NIAID to K. R.
and from the Medical Scientist Training Program to
L. J. G.) are gratefully acknowledged. We thank Prof.
Gerard Parkin and Dr Tony Hascall for performing the
X-ray crystallographic analysis. We are grateful to Vinka
Parmakovich, Barbara Sporer and Yasuhiro Itagaki of the
Columbia University Mass Spectroscopy Facility. We also
thank M. David Weingarten for helpful discussions.
(1S,2S)-Cyclohexane-1,2-dicarboxylic acid mono-n-propyl-
amide (26). Prepared in a manner analogous to 25 from the
enantiomer of 12.
Acid ¯uoride (27). To a stirred solution of (1R,2R)-cyclo-
hexanedicarboxylic acid mono-n-propylamide 25 (50 mg,
0.23 mmol) and pyridine (23 ml, 0.28 mmol) in DCM
(2 ml) was added dropwise to a solution of cyanuric ¯uoride
(63 mg, 0.47 mmol) in DCM (1 ml). A white precipitate
formed during the addition. The resulting suspension was
stirred for 1.5 h, diluted with DCM (3 ml), washed with
water (2£1 ml), dried (MgSO4), ®ltered and concentrated
to afford the acid ¯uoride as a white crystalline solid
(49 mg, 0.23 mmol, 97%). This material was used immedi-
ately in the subsequent step. Acid ¯uoride 28 was prepared
analogously.
References
1. Cram, D. J. Angew. Chem., Int. Ed. Engl. 1988, 27, 1009.
2. Still, W. C. Acc. Chem. Res. 1996, 29 (3), 155.
3. Hossain, M. A.; Schneider, H. J. J. Am. Chem. Soc. 1998, 120
(43), 11208.
4. Lowik, D. P. W. M.; Mulders, S. J. E.; Cheng, Y.; Shao, Y.;
Liskamp, R. M. J. Tetrahedron Lett. 1996, 37, 8253.
5. Torneiro, M.; Still, W. C. J. Am. Chem. Soc. 1995, 117,
5887.
Receptor (3). The acid ¯uoride 27 was added to a stirred
solution of diamine 24 (10 mg, 0.023 mmol) and diiso-
propylethylamine (12 ml, 0.070 mmol) in DMF (0.5 ml).
While stirring overnight, a dark red precipitate formed.
TLC showed a new red-colored band with Rfˆ0.4 (10%
MeOH/DCM). The mixture was evaporated to dryness and
puri®ed by preparative TLC (5% MeOH/CHCl3) to yield a
red solid (5 mg, 6 mmol, 26%): FAB-HRMS, calcd for
C43H61N9O8K 870.4280, found 870.4265 (M1K).
6. Davies, M.; Bonnat, M.; Guillier, F.; Kilburn, J. D.; Bradley, M.
J. Org. Chem. 1998, 63, 8696.
7. Carrasco, M. R.; Still, W. C. Chem. Biol. 1995, 2, 205.
8. Maletic, M.; Wennemers, H.; McDonald, D. Q.; Breslow, R.;
Still, W. C. Angew. Chem., Int. Ed. Engl. 1996, 35 (13/14), 1490.
9. Wennemers, H.; Yoon, S. S.; Still, W. C. J. Org. Chem 1995,
60, 1108.
10. Shao, Y. F.; Still, W. C. J. Org. Chem. 1996, 61 (18), 6086.
11. Rose, G. D.; Gierasch, L. M.; Smith, J. A. Adv. Protein Chem.
1985, 37, 1.
12. Macromodelw V 6.0 Mohamadi, F.; Richards, N. G. J.; Guida,
W. C.; Liskamp, R.; Lipton, M.; Caul®eld, C.; Chang, G.;
Hendrickson, T.; Still, W. C. J. Comput. Chem. 1990, 11, 440.
Monte Carlo conformational searches were performed in chloro-
form using default settings. Structures were minimized using the
MM2 force®eld.
Receptor (4). Made in a manner analogous to 3, from
diamine 24 and acid ¯uoride 28. Red solid (7 mg, 8 mmol,
37%); TLC Rfˆ0.61 (10% MeOH/DCM); FAB-HRMS,
calcd for C43H61N9O8K 870.4280, found 870.4286 (M1K).
Binding studies
Stock solutions of the receptors were made in neat
N-methylpyrrolidinone (NMP) using gentle to moderate
heating and/or sonication where necessary. A weight of
approximately 3 or 6 copies of the library was shaken in a
receptor solution of 5% NMP/chloroform buffered with
0.5 mM trioctylamine/0.25 mM tri¯uoroacetic acid at
room temperature. After equilibration for at least 48 h the
beads were inspected for accumulation of the red receptor-
dye conjugate and the concentration was adjusted until,
upon further equilibration, a small group of beads (usually
3±12) were stained red or dark orange. Equilibration was
allowed to continue for at least 12 h after the number of red
beads appeared constant. Unless otherwise noted in Table 1,
all of the darkest beads were then picked and decoded by
capillary GC as previously described.26 Using this pro-
cedure, the amount of receptor held in the beads at equi-
librium is insigni®cant. The ®nal concentration of the
13. Loll, P. J.; Bevivino, A. E.; Korty, B. D.; Axelsen, P. H. J. Am.
Chem. Soc. 1997, 119, 1516.
14. Beddoes, R. L.; Dalton, L.; Joule, J. A.; Mills, O. S.; Street,
J. D.; Watt, C. I. F. J. Chem. Soc., Perkin Trans II 1986, 787.
15. Reddy, D. R.; Thornton, E. R. J. Chem. Soc., Chem. Commun.
1992, 172.
16. Nagel, U. Angew. Chem., Int. Ed. Engl. 1984, 23 (6), 435.
17. Campbell, A. L.; Pilipauskas, D. R.; Khanna, I. K.; Rhodes,
R. A. Tetrahedron Lett. 1987, 28 (21), 2331.
18. O'Malley, G. J.; Spahl, B.; Cherill, R. J.; Kosley, R. W. J. Org.
Chem. 1990, 55, 1102.
19. Naphthalene-2,7-disulfonyl chloride from the disodium salt:
Muth, F. Houbeyn-Weyll 1955, 565.
20. Benzene-1,4-disulfonyl chloride from the sodium salt using
oxalyl chloride/DMF, via diazotization of the 4-amino benzene-
sulfonic acid with SO2: Meerwein, H.; Dittmar, G.; Gollner, R.;
Hafner, K.; Mensch, F.; Steinfort, O. Chem. Ber. 1957, 90 (6), 851.

3318
K. Ryan et al. / Tetrahedron 56 (2000) 3309±3318
21. Yoon, S. S. Y.; Still, W. C. Tetrahedron 1995, 51 (2), 567 and
references therein. The library contained 153 potential sequences
of the structure: Ac-AA3-AA2-AA1-HN-polystyrene, where AAn
represents Gly or the d or l isomers of: Ala, Ser, Val, Pro, Asn,
Gln, or Lys.
23. Appella, D. H.; Christianson, L. A.; Klein, D. A.; Gellman,
S. H. Nature 1997, 387, 381.
24. French Patent, Rhone-Poulenc, 1974, No. 74 06885.
25. Holzapfel, C. W.; Pettit, G. R. J. Org. Chem. 1984, 50, 2323.
26. Ohlmeyer, M. H. J.; Swanson, R. N.; Dillard, L. W.; Reader,
J. C.; Asouline, G.; Kobayashi, R.; Wigler, M.; Still, W. C. Proc.
Natl. Acad. Sci., USA 1993, 90, 10922.
22. Appella, D. H.; Christianson, L. A.; Karle, I. L.; Powell, D. R.;
Gellman, S. H. J. Am. Chem. Soc. 1996, 118 (51), 13071.