Identification of a potent heterocyclic ligand to somatostatin receptor subtype 5 by the synthesis and screening of β-turn mimetic libraries

Article abstract of DOI:10.1021/ja983742p

Methods for the parallel synthesis of medium-ring heterocyclic β-turn mimetics (1) are described, which enable the rapid preparation of libraries of mimetics for the identification of small molecule ligands to receptors. The generality of this method was first demonstrated by the rapid and high-yielding synthesis of mimetics 1a-g that display a wide range of proteinogenic amino acid side chains. Small molecule heterocyclic mimetics of the medicinally important peptide somatostatin were then identified by the synthesis of a focused library of β-turn mimetics based upon the crucial Trp-Lys motif found in the turn region of somatostatin. Screening the library against a panel of the five cloned human somatostatin receptors (hSST₁-hSST₅) resulted in the identification of a potent small molecule ligand (1h) with selectivity toward hSST₅, as well as potent ligands toward receptor subtypes 1-4. Furthermore, structure-activity relationships were used to establish the importance of each of the three diversity inputs present in compound 1h. This investigation represents the first successful identification of potent, small molecule ligands through the synthesis and evaluation of a focused library of turn mimetics (for one example of a successful screening effort of a nonfocused β-turn library, see: Souers, A. J.; Virgilio, A. A.; Schuerer, S.; Ellman, J. A.; Kogan, T. P.; West, H. E.; Ankener, W.; Vanderslice, P. Bioorg. Med. Chem. Lett. 1998, 8, 2297-2302). The results of the library screening revealed unexpected stereochemical and functional group preferences, reinforcing the critical importance of synthesizing and evaluating collections of mimetics as opposed to traditional iterative synthesis and evaluation approaches. The ability to prepare libraries of heterocyclic turn mimetics that display three different side-chain inputs with multiple distinct side-chain orientations should enable the rapid identification of small molecule heterocyclic ligands to a large number of receptor targets.

Full text of DOI:10.1021/ja983742p

J. Am. Chem. Soc. 1999, 121, 1817-1825
1817
Identification of a Potent Heterocyclic Ligand To Somatostatin
Receptor Subtype 5 by the Synthesis and Screening of â-Turn
Mimetic Libraries
Andrew J. Souers,† Alex A. Virgilio,† Åsa Rosenquist,† Wasyl Fenuik,‡ and
Jonathan A. Ellman*,†
Contribution from the Department of Chemistry, UniVersity of California, Berkeley, California 94720, and
Glaxo Wellcome Cambridge Chemistry Laboratory, UniVersity Chemical Laboratory, Lensfield Road,
Cambridge, CB2 1EW, UK
ReceiVed October 26, 1998
Abstract: Methods for the parallel synthesis of medium-ring heterocyclic â-turn mimetics (1) are described,
which enable the rapid preparation of libraries of mimetics for the identification of small molecule ligands to
receptors. The generality of this method was first demonstrated by the rapid and high-yielding synthesis of
mimetics 1a-g that display a wide range of proteinogenic amino acid side chains. Small molecule heterocyclic
mimetics of the medicinally important peptide somatostatin were then identified by the synthesis of a focused
library of â-turn mimetics based upon the crucial Trp-Lys motif found in the turn region of somatostatin.
Screening the library against a panel of the five cloned human somatostatin receptors (hSST1-hSST5) resulted
in the identification of a potent small molecule ligand (1h) with selectivity toward hSST5, as well as potent
ligands toward receptor subtypes 1-4. Furthermore, structure-activity relationships were used to establish
the importance of each of the three diversity inputs present in compound 1h. This investigation represents the
first successful identification of potent, small molecule ligands through the synthesis and evaluation of a focused
library of turn mimetics (for one example of a successful screening effort of a nonfocused â-turn library, see:
Souers, A. J.; Virgilio, A. A.; Schu¨rer, S.; Ellman, J. A.; Kogan, T. P.; West, H. E.; Ankener, W.; Vanderslice,
P. Bioorg. Med. Chem. Lett. 1998, 8, 2297-2302). The results of the library screening revealed unexpected
stereochemical and functional group preferences, reinforcing the critical importance of synthesizing and
evaluating collections of mimetics as opposed to traditional iterative synthesis and evaluation approaches. The
ability to prepare libraries of heterocyclic turn mimetics that display three different side-chain inputs with
multiple distinct side-chain orientations should enable the rapid identification of small molecule heterocyclic
ligands to a large number of receptor targets.
Introduction
protein recognition. However, despite considerable effort, very
few bioactive small molecule â-turn mimetics have been
developed. The diversity of â-turn structures found in peptides
and proteins severely complicates the determination of the key
side chains and the correct display of those side chains in the
peptide’s bioactive conformation.5a,b
We recently reported a preliminary paper on synthesis
methods to prepare mimetics6 based upon the turn structure
whereby three amino acid side chains are displayed on a
medium-ring heterocyclic scaffold 1 (Figure 1).
A key feature in the design strategy is the capability of
performing rapid, parallel synthesis of multiple derivatives in
order to overcome the limitations in defining the exact bioactive
conformation of the target peptide.7a,b The synthetic sequence
enables the incorporation of all of the functionality present in
the side chains of the proteinogenic amino acids. In addition,
to reflect the diversity of side-chain orientations found in
Although peptides play an essential role in wide-ranging
biological processes, their use as biological probes and thera-
peutic agents has been complicated by poor receptor subtype
selectivity, poor biostability, or unfavorable absorption proper-
ties.2 The design of high affinity and specific constrained
peptidomimetic ligands has therefore been an extremely im-
portant endeavor in the fields of medicinal chemistry and
molecular recognition. In a successfully designed peptidomi-
metic, the essential amino acid side chains of the corresponding
peptide are displayed on an alternative scaffold such that the
spatial orientation of the side chains corresponds to the display
in the bioactive conformation of the peptide.3 The design and
synthesis of peptidomimetic scaffolds based on the â-turn
structure4a-c has received the most attention due to the
importance of this secondary structural motif in peptide and
† University of California.
‡ Glaxo Wellcome Cambridge Chemistry Laboratory.
(1) For one example of a successful screening effort of a nonfocused
â-turn library, see: Souers, A. J; Virgilio, A. A.; Schu¨rer, S.; Ellman, J.
A.; Kogan, T. P.; West, H. E.; Ankener, W.; Vanderslice, P. Bioorg. Med.
Chem. Lett. 1998, 8, 2297-2302.
(4) (a) Ball, J. B., Alewood, P. F. J. Mol. Recognit. 1990, 3, 55-63. (b)
Kahn, M. Synlett 1993, 11, 821-826. (c) Kitagawa, O.; Vander Velde, D.;
Dutta, D.; Morton, M.; Takusagawa, F.; Aube, J. J. Am. Chem. Soc. 1995,
117, 5169-5178.
(5) (a) Wilmot, C. M.; Thornton, J. M. J. Mol. Biol. 1988, 203, 221-
232. (b) Ball, J. B.; Hughes, R. A.; Alewood, P. F.; Andrews, P. R.
Tetrahedron 1993, 17, 3467-3478.
(6) Virgilio, A. A.; Schu¨rer, S. C.; Ellman, J. A. Tetrahedron Lett. 1996,
37, 6961-6964.
(2) Sawyer, T. K. Structure-Based Design: Diseases, Targets, Techniques
and DeVelopments; Veerapandian. P., Ed.; Marcel Dekker: New York, 1997;
pp 559-634.
(3) Gante, J. Angew. Chem., Int. Ed. Engl. 1994, 33, 1699-1720.
10.1021/ja983742p CCC: $18.00 © 1999 American Chemical Society
Published on Web 02/20/1999

1818 J. Am. Chem. Soc., Vol. 121, No. 9, 1999
Souers et al.
Figure 1. Heterocyclic â-turn mimetic displaying relevant side chains.
Figure 3. Figure 3. A potent nonpeptidal hSST5 ligand.
Peptide ligands to hSST5 inhibit insulin release,14 and hSST5
has been implicated as the receptor responsible for the antipro-
liferative effects15 of somatostatin analogues in normal as well
as tumoral cells. Prior design efforts toward small molecule
mimetics of somatostatin have produced low micromolar
mimetics based upon the display of side chains upon a
benzodiazepine scaffold,16 and through extensive efforts, non-
selective high nanomolar ligands based upon the display of side
chains upon sugar scaffolds.17a,b
Within the past year, Merck18a,b,c has identified subnanomolar
ligands to the somatostatin receptors and Novo Nordisk19 has
identified low nanomolar ligands to hSST4 by employing
massive screening and iterative optimization efforts. Although
these screening efforts were ultimately successful, a general
strategy for the identification of ligands to receptors through
the design, synthesis, and screening of a more focused collection
of peptidomimetics has the potential to be considerably more
efficient.
Figure 2. The components of the â-turn mimetic 1.
proteins, all four possible stereoisomers are accessible at the i
+ 1 and i + 2 positions. Side-chain display may also be varied
by introducing backbone elements to change the ring size and
vectoral display of the amine and thiol functionality. As shown
in Figure 2, all of the side-chain precursors are accessed from
readily available building blocks: the i + 3 side chain is derived
from primary amines, the i + 2 side chain from Fmoc amino
acids, and the i + 1 side chain from R-bromo acids, which can
be prepared in highly enantioenriched form in a single step from
the corresponding side chain protected amino acids.8
To establish the ability of heterocycle 1 to serve as a highly
effective mimetic of peptide and protein structure, we have
focused on the development of potent ligands to the somatostatin
receptors. To date, five human somatostatin receptor subtypes
(hSST1-hSST5) have been identified and cloned with each
subtype having distinct expression patterns and activities.9
Somatostatin acts on these receptor subtypes to inhibit the
release of a number of pancreatic pituitary and gastrointestinal
hormones including glucagon, growth hormone, and insulin.
Somatostatin is also present in the nervous system,10 where it
is believed to mediate several processes including cognition and
locomoter activity. Due to the wide-ranging biological effects
of somatostatin, numerous peptide analogues have been prepared
to improve specificity and biostability.11 These efforts culmi-
nated in the development of several potent and specific cyclic
peptide analogues with a â-turn structure as the key binding
determinant. These analogues are best exemplified by the cyclic
octapeptide octreotide,12 which is an approved drug for the
treatment of carcinomas, acromegaly, and gastro-entero-
pancreatic tumors. Unfortunately, octreotide has poor oral
bioavailability and must be administered by subcutaneous or
intraveneous injection.13 Clearly, potent small molecule mimetics
of somatostatin would be highly desirable.
Herein, we report for the first time the detailed experimental
procedures for the synthesis of â-turn mimetics 1 and the devel-
opment of heterocycle 1h (MW 521) that is a potent small mo-
lecule ligand of the hSST5 receptor (IC50 ) 87 nM) (Figure 3).
Results and Discussion
â-Turn Mimetic Synthesis. The final step in the synthesis
of â turn mimetic 1 would necessarily be macrocyclization to
provide the desired strained medium-ring hetererocycle. Either
(13) Rivier, J.; Brown, M.; Vale, W. Biochem. Biophys. Res. Commun.
1975, 65, 746-751.
(14) Coy, D. H.; Rossowski, W. J. Ciba Found. Symp. 1995, 190, 240-
252.
(15) Cordelier, P.; Esteve, J. P.; Bousquet, C.; Delesque, N.; O’Carroll,
A.; Schally, A. V.; Vaysse, N.; Susini, C.; Buscail, L. Proc. Natl. Acad.
Sci. 1997, 94, 9343-9348.
(16) Papageorgiou, C.; Borer, X. A. Bioorg. Med. Chem. Lett. 1996, 6,
267-272.
(17) (a) Hirschmann, R.; Nicolaou, K. C.; Pietranico, S.; Salvino, J.;
Leahy, E. M.; Sprengeler, P. A.; Furst, G.; Smith, A. B., III, Strader, C.
D.; Cascieri, M. A.; Candelore, M. R.; Donaldson, C.; Valle, W.; Maechler,
L. J. Am. Chem. Soc. 1992, 114, 9217-9218. (b) Hirschmann, R.; Nicolaou,
K. C.; Pietranico, S.; Leahy, E. M.; Salvino, J.; Arison, B.; Cichy, M. A.;
Spoors, P. G.; Shakespeare, W. C.; Sprengeler, P. A.; Hamley, P.; Smith,
A. B., III; Reisine, T.; Raynor, K.; Maechler, L.; Donaldson, C.; Vale, W.;
Freidinger, R. M.; Cascieri, M. R.; Strader, C. D. J. Am. Chem. Soc. 1993,
115, 12550-12568.
(7) (a) Hirschmann and Smith have highlighted the central importance
of being able to prepare libraries of peptidomimetics that they have
developed based upon sugar scaffolds, but accounts of library synthesis
have not been reported. Hirschmann, R.; Hynes, J., Jr.; Cichy-Knight, M.
A.; van Rijn, R.; Sprengeler, P.A.; Spoors, P. G.; Shakespeare, W. C.;
Pietranico-Cole, S.; Barbosa, J.; Liu, J.; Yao, W.; Rohrer, S.; Smith, A. B.,
III. J. Med. Chem. 1998, 41, 1382-1391. (b) Feng, Y.; Wang, Z.; Jin, S.;
Burgess, K. J. Am. Chem. Soc. 1998, 121, 10768-10769.
(18) (a) Yang, L.; Guo, L.; Pasternak, A.; Mosley, R.; Rohrer, S.; Birzin,
E.; Foor, F.; Cheng, K.; Schaeffer, J.; Patchett, A. A. J. Med. Chem. 1998,
41, 2175-2179. (b) Yang, L.; Berk, S. C.; Rohrer, S. P.; Mosley, R. T.;
Guo, L.; Underwood, D. J.; Arison, B. H.; Birzin, E. T.; Hayes, E. C.;
Mitra, S. W.; Parmar, R. M.; Cheng, K.; Wu, T.; Butler, B. S.; Foor, F.;
Paternak, A.; Pan, Y.; Silva, M.; Freidinger, R. M.; Smith, R. G.; Chapman,
K.; Schaeffer, J. M.; Patchett, A. A. Proc. Natl. Acad. Sci. U.S.A. 1998,
95, 10836-10841. (c) Rohrer, S. P.; Birzin, E. T.; Mosley, R. T.; Berk, S.
C.; Hutchins, S. M.; Shen, D.; Xiong, Y.; Hayes, E. C.; Parmar, R. M.;
Foor, F.; Mitra, S. W.; Degrado, S. J.; Shu, M.; Klopp, J. M.; Cai, S.; Blake,
A.; Chan, W.; Pasternak, A.; Yang, L.; Smith, R. G.; Chapman, K. T.;
Schaeffer, J. M. Science 1998, 282, 737-740.
(8) Souers, A. J.; Schu¨rer, S. C.; Kwack, H. H.; Virgilio, A. A.; Ellman,
J. A. Synthesis 1998, in press.
(9) Reisine, T.; Bell, G. I. Neuroscience 1995, 67, 777-790.
(10) Patel, Y. C. J. Endocrinol. InVest. 1997, 20, 348-367.
(11) Lamberts, S. W. J.; Krenning, E. P.; Reubi, J. C. Endocrine ReV.
1991, 12, 450-482.
(12) Bauer, W.; Briner, U.; Doepfner, W.; Haller, R.; Huguenin, R.;
Marbach, P.; Petcher, T. J.; Pless, J. Life Sci. 1982, 31, 1133-1140.
(19) Ankersen, M.; Crider, M.; Liu, S.; Ho, B.; Andersen, H. S.; Stidsen,
C. J. Am. Chem. Soc. 1998, 120, 1368-1373.

Synthesis and Screening of â-Turn Mimetic Libraries
J. Am. Chem. Soc., Vol. 121, No. 9, 1999 1819
Scheme 1a
Figure 4. Retrosynthesis of turn mimetic 1.
macrolactamization or intramolecular thioalkylation could be
considered. Indeed, due to the limited precedent for preparing
strained nine-membered ring lactams through macrocyclization,
we had initially explored both strategies toward the synthesis
of turn mimetics.31a In this initial work, macrolactamization
proceeded in poor yields and resulted in considerable quantities
of dimer under a variety of reaction conditions. In contrast,
thioalkylation proceeded in high yields with minimal dimer or
oligomer formation and therefore was selected to prepare
heterocycle 1 (Figure 4). We chose to prepare cyclization
precursor 7 by solid-phase synthesis to enable an efficient
parallel synthesis strategy. The mercaptyl functionality in
precursor 7 would serve as the ideal support attachment site,
since it would be unmasked for final cyclization upon release
from support. Support-bound intermediate 6 would be prepared
from secondary amine 5 by sequential coupling of an amino
acid to incorporate the i + 2 side chain followed by an R-bromo
acid to incorporate the i + 1 side chain. The i +3 side chain
would be introduced by amine displacement upon the activated
support-bound backbone 4. We chose to attach the activated
hydroxyalkylthiol backbone element to the solid support through
a disulfide linkage rather than a thioether linkage to minimize
competitive intramolecular displacement, since disulfides are
much poorer nucleophiles than thioethers.
a (a) PyBOP, HOBt, i-Pr2EtN, aminomethylpolystyrene. (b) NaOMe,
3:1 THF/MeOH. (c) 2-benzothiazolyl 2-methanesulfonoxyalkyl disul-
fide (3a or b).
Scheme 2a
Linker 2 was first prepared by coupling S-acetyl 2-mercapto-
2-methyl propionic acid20 to aminomethylated polystyrene resin
(Scheme 1). The gem-dimethyl substituents adjacent to the sulfur
atom are necessary to improve the stability of the eventual
disulfide bond. Absence of the gem-dimethyl substituents, such
as with the analogous 2-mercaptoacetic acid derived linker,
results in reduced yields presumably due to the lability of the
disulfide bond to amine bases under the forcing conditions used
for amine displacement.21a,b
Next, methods for attaching the backbone hydroxyalkylthiol
to the solid support and activating the hydroxyl for nucleophilic
displacement were needed. Methanolysis of thioester 2 followed
by disulfide interchange with the methanesulfonoxyalkanethiol
backbone component activated as a 2-benzothiazolyl (Bt) mixed
disulfide22 3 provided mesyl ester 4 (Scheme 1). Completion
of the disulfide interchange was monitored by the disappearance
of free thiol on the support using Ellman’s reagent.23 Although
the backbone component could also be loaded onto the support
as the free alcohol followed by conversion to the mesylate with
mesyl anhydride and 2,6-lutidine in CH2Cl2, lower yields of
the final mimetics were observed.
a (a) H2NCH2Ri+3, NMP, 55 °C. (b) Fmoc-Amino Acid, HATU,
i-Pr2EtN, DMF. (c) piperidine (20%), DMF. (d) R-bromo acid, DICI,
HOAt, DMF.
The support-bound mesylate 4 was then treated with a
concentrated solution of the appropriate primary amine to
introduce the i + 3 side chain to give 5. The i + 2 side-chain
was then incorporated by coupling an Fmoc-protected amino
acid. Hydroxyazabenzotriazole based coupling agents were
necessary in order to achieve complete coupling with the
relatively hindered support-bound secondary amine 5. Standard
Fmoc deprotection and subsequent coupling with an R-bromo
acid to incorporate the i +1 side chain provided the support-
bound cyclization precursor 6 (Scheme 2).
A key challenge in the synthesis of mimetic 1 was to develop
appropriate support release and cyclization steps for the parallel
synthesis of the pure mimetics. We initially used the volatile
reagents triethyl phosphine and N-methyl morpholine (NMM)
to reduce the support-bound disulfide and drive the cyclization,
(20) Mery, J.; Brugidou, J.; Derancourt, J. J. Pept. Res. 1992, 5, 223-
240.
(21) (a) Mery, J.; Granier, C.; Juin, M.; Brugidou, J. Int. J. Pept. Protein
Res. 1993, 42, 44-52. (b) Greene, T. W.; Wuts, P. G. M. ProtectiVe Groups
in Organic Synthesis; John Wiley & Sons: New York, 1991; pp 438-440.
(22) Brzezinska, E.; Ternay, A. L. J. Org. Chem. 1994, 59, 8239-8244.
(23) Ellman, G. L. Arch. Biochem. Biophys. 1959, 82, 70-77.

1820 J. Am. Chem. Soc., Vol. 121, No. 9, 1999
Souers et al.
Table 1. Synthesis of â-turn Mimeticsa
Scheme 3a
a (a) TCEP. (b) 1% divinylbenzene-polystyrene-N,N,N,N-tetrameth-
ylguanidine resin.
respectively, since they could be removed by concentration in
vacuo. However, while the volatile reagents could be removed,
the triethyl phosphine oxide and N-methyl morpholine hydro-
bromide byproducts contaminated the final turn mimetics after
concentration. These contaminants complicated analytical char-
acterization of the compounds and had the potential to interfere
with biological evaluation. We therefore developed a solid-phase
extraction approach for the rapid removal of HBr and phosphine
impurities.
a The stereochemical configuration of the R-carbon at the i + 1 site
is R and at the i + 2 site is S unless otherwise specified. The purity of
the crude product directly off of the support was determined by reverse
phase HPLC, 60-100% CH3OH in 0.1% aqueous TFA as monitored
at 220, 254, or 280 nM.
A support-bound amine could easily be used in place of a
tertiary amine to scavenge HBr; however, a suitable support-
bound reagent could not be identified that would scavenge the
triethylphosphine oxide. We therefore employed the modified
phosphine reagent tris-(2-carboxyethyl)phosphine (TCEP) to
reduce the disulfide. Both excess TCEP and the oxide should
be scavenged by the same support-bound amine that was used
to scavenge HBr.24 Support-bound N,N,N,N-tetramethylguani-
dine (TMG) proved to be the optimal scavenging reagent
because it served both as an effective solid-phase extraction
resin and as a general base to promote thioether formation to
yield cyclic product 1. Cyclization with the support-bound
guanidine proceeded to completion within 24 h at room
temperature. Furthermore, the formation of cyclic dimer or
oligomers was not observed when the cyclization reaction was
performed at 1 mM substrate concentrations. Filtration to remove
the resin followed by concentration in vacuo yielded the crude
product 1 uncontaminated with reagent byproducts (Scheme 3).
The preliminary publication of this work marked the first use
of solid-phase synthesis and solid-phase extraction in the same
synthesis sequence and represented one of the first applications
of solid-phase extraction methods for the synthesis of compound
libraries.6
during the synthesis sequence cannot be ruled out. The suc-
cessful introduction of the functionalized lysine, aspartic acid,
and tryptophan side chains, nine- and ten-membered ring sizes,
and both stereochemistries at the i + 1 and i + 2 sites clearly
demonstrates the broad generality of the synthesis sequence.
Design and Synthesis of Mimetic Library Targeting
Somatostatin. A focused library of somatostatin mimetics
(Figure 5) was designed based upon prior, extensive studies of
peptide analogues of somatostatin.25a-c Other active cyclic
peptides include a retro-enantio analogue26a,b of somatostatin
wherein both the order and stereochemistry of the central Trp
and Lys residues are reversed. These examples indicated that
several alternative displays of the tryptophan and lysine side
chains could potentially provide active compounds. For this
reason all eight possible combinations of the Trp and Lys side
chains of both stereochemistries were included at the i + 1 and
i + 2 positions. Twenty-two different amines were introduced
at the i + 3 position. Eight of the amines were chosen for their
their similarity to the i + 3 (Phe and Tyr) as well as the i (Thr
and Val) residues present in numerous somatostatin analogues.
The remaining 14 amines were selected to maximize the
diversity of functionality at this site employing a heirarchical,
2D structure key-based clustering algorithm developed by MDL
Information Systems.27
Seven mimetics 1(a-g) were prepared to demonstrate the
generality of the synthesis sequence. High levels of purity
1
(>90%) as determined by HPLC and H NMR were observed
for each of the crude mimetics obtained directly from the resin.
In addition, the average overall yield for the chromatographed
and analytically pure material was 59% based upon the expected
loading of 2 (Table 1). The only identifiable side-products
(e5%) were the minor diastereomers corresponding to inversion
at the R-carbon at the i + 1 position. This most likely results
from the initial optical purity of the R-bromo acids that supply
the i + 1 side chain (generally ∼95% ee as determined by chiral
GC or HPLC analysis), although a small amount of racemization
(25) (a) Brady, S. F., Paleveda, W. J.; Arison, B. H.; Saperstein, R.;
Brady, E. J.; Raynor, K.; Reisine, T.; Veber, D. F.; Freidinger, R. M.
Tetrahedron 1993, 49, 3449-3466. (b) Veber, D. F.; Saperstein, R.; Nutt,
R. F.; Freidinger, R. M.; Brady, S. F.; Curley, P.; Perlow, D. S.; Paleveda,
W. J.; Colton, C. D.; Zacchei, A. G.; Tocco, D. J.; Hoff, D. R.; Vandlen,
R. L.; Gerich, J. E.; Hall, L.; Mandarino, L.; Cordes, E. H.; Anderson, P.
S.; Hirschmann, R. Life Sci. 1984, 34, 1371-1378. (c) Hocart, S. J.; Reddy,
V.; Murphy, W. A.; Coy, D. H. J. Med. Chem. 1995, 38, 1974-1989.
(26) (a) Pallai, P.; Struthers, S.; Goodman, M.; Rivier, J.; Vale, W.
Biopolymers 1983, 22, 2523-2538. (b) Murphy, W. A.; Heiman, M. L.;
Lance, V. A.; Mezo, I.; Coy, D. H. Biochem. Biophys. Res. Commun. 1985,
132, 922-928.
(24) Burns, J. A.; Butler, J. C.; Moran, J.; Whitesides, G. M. J. Org.
Chem. 1991, 56, 2648-2650.
(27) Muskal, S. Unpublished Results. MDL Information Systems, Inc.,
San Leandro, CA.

Synthesis and Screening of â-Turn Mimetic Libraries
J. Am. Chem. Soc., Vol. 121, No. 9, 1999 1821
Figure 5. Focused library of somatostatin mimetics.
in somatostatin. The screening data also clearly indicated that
the preferred stereochemistry at the i + 1 and i + 2 positions
is of D configuration (Table 1). While the selection of D-Trp is
not surprising given the large number of potent cyclic peptide
analogues of somatostatin that incorporate D-Trp at the i + 1
position,25a-c Lys of L-configuration is clearly preferred at the
i + 2 position in peptides.28 Additionally, large aromatic groups
are preferred at the i + 3 position of heterocycle 1. This result
is consistent with the clear requirement for hydrophobic contacts
in peptide analogues of somatostatin,29 although it would not
have been predicted based upon the preference for Val or Thr
at the i + 3 position of the peptide analogues. The unexpected
binding preference for D-Lys at the i + 2 position, which has
been observed once before in a cyclic peptide somatostatin
analogue,30 and the large aromatic groups at the i + 3 position
highlight the importance of evaluating multiple mimetics with
varied side-chain displays even when the structures are modeled
on an extensively studied peptide such as somatostatin.
Figure 6. Scaled-up somatostatin ligands.
Each compound was then screened against all five cloned
human somatostatin receptor subtypes. The ability of the
compounds to inhibit the binding of [125I]-Tyr11-SRIF to human
recombinant SST1, SST3, SST4, and SST5 and the binding of
The effect of stereochemistry on bioactivity is highlighted
in Table 2. Mimetic 1h exhibits remarkable potency toward
SST5, while the activity of i + 2 epimer 1j toward the same
subtype decreases by nearly 15-fold. Furthermore, the activity
of the S,S enantiomer 1i also shows a considerable decrease
toward hSST5. It is further interesting to note that heterocycle
1h also exhibits the highest level of selectivity, with its affinity
for hSST5 being greater than 10-fold over that of subtypes 2-4,
and nearly 10-fold over hSST1. Finally, heterocycles 1k and 1l
were found to be the most potent toward subtypes 2 and 3,
[
125I]-BIM-23027 to the human recombinant SST2 all expressed
in CHO-K1 cells was measured. The compounds that showed
the greatest inhibition of specific binding were then tested at
lower concentrations in order to more accurately define their
potency. The most active compounds targeting hSST2, hSST3,
and hSST5, 1k, 1l, and 1h, respectively, were prepared on a
larger scale for exact IC50 determinations on fully characterized
and purified material (Figure 6). The two stereoisomers, 1i and
1j, of the most potent ligand to hSST5, 1h, showed modest
activity in the library screen. These compounds were also
prepared on a large scale for exact IC50 determinations in order
to further assess the impact of stereochemistry on binding.
Intepretation of assay results. The binding data clearly
established the preference for Trp at the i + 1 position and Lys
at the i + 2 position, which corresponds to the sequence found
(28) Nutt, R. F.; Veber, D. F.; Curley, P. E.; Saperstein, R.; Hirschmann,
R. F. Int. J. Pept. Protein Res. 1983, 21, 66-73.
(29) Nutt, R. F.; Saperstein, R.; Veber, D. F. Peptides - Structure &
Function. Proceedings of the Eighth American Peptide Symposium; Hruby,
V. J., Rich, D. H., Eds.; Pierce Chemical Co.: Rockford, Il, 1983; pp 345-
348.
(30) Freidinger, R. M.; Veber, D. F. In Conformationally Directed Drug
Design: Vida, J. A., Gordon, M., Eds.; ACS Symposium Series 251;
American Chemical Society: Washington, DC, 1984; pp 169-187.

1822 J. Am. Chem. Soc., Vol. 121, No. 9, 1999
Souers et al.
Conclusion
The identification of peptidomimetics is an important goal
to many academic and industrial researchers. The design and
synthesis of peptidomimetic scaffolds based on the â-turn
structure has received the most attention due to the importance
of this secondary structural motif in peptide and protein
recognition. However, despite considerable effort, very few
bioactive small molecule â-turn mimetics have been developed
due to the difficulty of determining the identity and orientation
of key side chains.
Figure 7. Alanine scan derivatives of ligand 1h.
Table 2. IC50 Values (nM) for the Most Potent Somatostatin
Mimetics (values reflect assay results of purified material)
Herein we report the first successful application of a designed
combinatorial library of â-turn mimetics. The synthesis and
screening of a 172-member library of medium-ring heterocycles
incorporating the side-chain display observed in the critical
â-turn present in cyclic peptide analogues of somatostatin
resulted in the identification of an 87 nM ligand (1h) to hSST5.
Central to the success of these efforts was the cability of
preparing heterocycles displaying different side chains corre-
sponding to the i + 1, i + 2 and i + 3 positions of the â-turn,
in addition to different side chain stereochemistries. Unantici-
pated side chains and stereochemistries were identified in this
study demonstrating the critical importance of evaluating
collections of mimetics. Further optimization of specificity and
affinity will be attempted through the parallel synthesis and
evaluation of constrained mimetics. Finally, using the methods
described herein, a large library of turn mimetics that incorporate
all of the proteinogenic amino acid side chains at the i + 1, i +
2, and i + 3 positions has been prepared.31c Screening this library
should prove to be ideal for the identification of peptidomimetic
ligands to receptors where relatively little information is
available regarding receptor-peptide ligand interactions.1
ligand
hSST1
hSST2
hSST3
hSST4
hSST5
1h
1i
1j
1k
1l
501
2240
4467
1096
851
1585
3630
4571
158
3090
813
1202
1175
135
1047
363
3162
479
87
832
1202
589
252
245
120
respectively, with the latter compound also showing affinity for
three of the remaining four subtypes (hSST2, hSST4, and hSST5).
To further establish that each side chain plays an important
role in recognition, three derivatives of the most potent ligand
1h (IC50 ) 87 nM to hSST5) were prepared with each derivative
having one of the three side chains replaced with a methyl
substituent (Figure 7). None of the derivatives showed signifi-
cant binding to hSST5 (IC50 > 10 µM) unambiguously
establishing the importance of each side chain for binding.
Synthesis and Evaluation of an i + 3 Diversity Library.
A second series of mimetics was then prepared based upon the
structures of the most potent compounds targeting hSST2 and
hSST5. The i + 1 and i + 2 side chains were fixed as the Trp/
Lys side chains, respectively, both with D-stereochemistry
(Figure 8). To introduce the i + 3 side chain, 16 amines
containing aromatic functionality were selected based upon
structual similarity to those amines that resulted in heterocycles
of greatest activity from the first library. Biological evaluation
of this small collection of heterocycles established that a
majority of the compounds had submicromolar IC50 values.
However, none of the incorporated i + 3 side chains resulted
in improved activity toward either hSST2 or hSST5.
We anticipate that significant improvement in affinity and
specificity may yet be achieved. Evaluation of much larger
collections of amines at the i + 3 position could result in
improved binding to hSST2 or hSST5, while modifications that
enhance the rigidity of the heterocycle and of the side chain
display may also provide enhanced affinity and particularly
specificity. Multiple low-energy conformers of heterocycle 1h
are likely populated based upon molecular modeling and NMR
analysis of 1 and of earlier mimetic structures,6,31a-c and
introduction of constrained backbone elements or cyclic amino
acid derivatives at the i + 2 position will signicantly reduce
the conformational flexibility of heterocyle 1h. Finally, the i +
1 and i + 2 side chains can also be biased by introduction of
â-methyl-substituted side chains. Related approaches have
successfully been employed to greatly enhance the affinity and
specificity of peptide ligands.32a,b
Materials and Methods
Reagents and General Methods. Unless otherwise noted, all
reagents were obtained from commercial suppliers and used without
further purification. When used as a reaction solvent, CH2Cl2 was
distilled from CaH2, and THF and dioxane were distilled under N2 from
sodium benzophenone ketyl, all immediately prior to use. Deoxygen-
ation of solvents and reaction mixtures was achieved by bubbling N2
or Ar through them for 15-20 min. N-Fmoc-protected amino acids,
Merrifield chloromethyl resin, and aminomethyl polystyrene resin were
purchased from Calbiochem-Novabiochem Corp. (San Diego, CA) or
Bachem Bioscience Inc. (King of Prussia, PA). R-Bromo acids were
prepared in one step in optically enriched form acccording to the
recently published procedure.8
Chromatography was carried out using Merck 60 230-240 mesh
silica gel according to the procedure described by Still.33 Components
were visualized either by ultraviolet light or anisaldehyde staining.34
Melting points were determined in open Pyrex capillaries and are
uncorrected. NMR spectra were recorded on Bruker AMX 300, AMX
400, AM 400, or AM 500 machines. Samples for NMR analysis that
contained rotamers were heated to cause coalescence of the different
rotameric peaks. NMR chemical shifts are expressed in ppm downfield,
relative to internal solvent peaks. Coupling constants, J, are listed in
hertz. When 31P NMR spectra were obtained, the chemical shifts were
calibrated on tetramethyl phosphate as an internal standard at 3.08 ppm.
Reverse phase HPLC was performed on a DynamaxΠHPLC system
from Rainin using a Microsorb C18 column from Rainin and a flow
rate of 1 mL/min with methanol/water 0.1% TFA as the mobile phase.
MALDI-TOF mass spectrometry was performed on a Voyager Bio-
Spectrometry workstation. Infrared spectra were recorded on a Perkin-
Elmer 1600 FT-IR spectrophotometer, either taken as thin films on
NaCl plates or as KBr pellets. Elemental analyses were performed by
(31) (a) Virgilio, A. A.; Ellman, J. A. J. Am. Chem. Soc. 1994, 116,
11580-11581. (b) Virgilio, A. A.; Bray, A. A.; Zhang, W.; Trinh, L.;
Ellman, J. A. Tetrahedron 1997, 53, 6635-6644. (c) Virgilio, A. A. 1997.
The Design, Synthesis, and Evaluation of Libraries of â-Turn Mimetics.
Ph. D. Dissertation, University of California, Berkeley, CA.
(32) (a) Liao, S.; Shenderovich, M. D.; Zhang, Z.; Maletinska, L.;
Slaninova, J.; Hruby, V. J. J. Am. Chem. Soc. 1998, 120, 7393-7394. (b)
Haskell-Leuvano, C.; Toth, K.; Boteju, L.; Job, C.; Castrucci, A. M.; Hadley,
M. E.; Hruby, V. J. J. Med. Chem. 1997, 40, 2740-2749.
(33) Still, W. C.; Kahn, M.; Mitra, A. J. Org. Chem. 1978, 43, 2923.
(34) Gordon, A. J.; Ford, R. A. The Chemist’s Companion; John Wiley
& Sons: New York, 1972; pp 378-380.

Synthesis and Screening of â-Turn Mimetic Libraries
J. Am. Chem. Soc., Vol. 121, No. 9, 1999 1823
Figure 8. Ri+3 diversification library.
M-H-W Laboratories (Phoenix, AZ) or by the Microanalytical Labora-
tory at the University of California at Berkeley.
121.1, 122.0, 124.6, 126.2, 135.6, 154.7, 171.3. Anal. Calcd for C11H13-
NO3S4 C, 39.38; H, 3.91; N, 4.18. Found C, 39.36; H, 4.04; N, 3.96.
Tris(2-carboxyethyl)phosphine. The preparation of tris(2-carboxy-
ethyl)phosphine24 was altered so as to afford the neutral compound
rather than the hydrochloride salt. Tris(2-cyanoethyl)phosphine (92%
purity, 8% phosphine oxide) was purchased from Strem Chemicals.
Tris(2-cyanoethyl)phosphine (25 g, 119 mmol) was refluxed under a
N2 atmosphere in 85 mL of concentrated HCl (aqueous) for 3 h. The
solution was cooled to room temperature, and 100 mL of deoxygenated
water followed by 16.1 g (643 mmol) of NaOH in 100 mL of
deoxygenated water were added while a nitrogen atmosphere was
maintained. The solution was extracted with oxygen-free THF (3 ×
500 mL). The combined THF extracts were concentrated in vacuo to
obtain crystals which were dried under high vacuum to give 21 g (76
mmol, 64%) of 90% pure (10% phosphine oxide) tris(2-carboxyethyl)-
phosphine. 1H NMR (400 MHz, d6-DMSO) δ 1.60 (t, 6, J ) 7.9), 2.28
(dt, 6, J ) 9.4, 7.9), 9.44 (br s, 3). 13C NMR (400 MHz, d6-DMSO) δ
27.8 (d, J ) 13.0), 33.7 (d, J ) 15.5), 123.2 (d, J ) 10.8). 31P NMR
(400 MHz, d6-DMSO): δ -26.3.
Preparation of N,N,N,N-Tetramethylguanidine Resin. To 25 g
of 200-400 mesh Merrifield chloromethyl resin (1.0 mmol/g, 25 mmol)
was added a 1.0 M solution of tetramethylguanidine (25 mL, 250 mmol)
in CH2Cl2 (150 mL). Catalytic tetrabutylammonium iodide was added,
and the reaction slurry was heated to reflux for 14 h. The reaction
mixture was transferred to a fritted glass funnel, and the resin was rinsed
exhaustively with CH2Cl2 (2 × 200 mL), 3:1:1 THF/MeOH/1 N NaOH
(aqueous) (5 × 200 mL), THF (2 × 200 mL), and MeOH (2 × 200
mL). The resin was then dried in vacuo. No detectable amount of
chloride remained on the resin as indicated by the Beilstein test and
Vollhard titration.35 Combustion analysis was performed to determine
nitrogen content. Anal. Calcd N, 3.9 (0.93 mequiv/g). Found N, 2.2
(0.52 mequiv/g).
General Procedure for the Solid-Phase Synthesis of â-Turn
Mimetics 1. Aminomethyl-derivatized polystyrene resin is solvated in
DMF (10 mL/g of resin), and S-acetyl 2-mercapto-2-methyl propionic
acid (0.2 M), PyBOP (0.2 M), HOBt (0.2 M), and i-Pr2EtN (0.4 M)
are added. After 8 h, the solution is drained, and the resin is rinsed
with DMF (3×), CH2Cl2 (4×), and MeOH (2×), then dried in vacuo.
Resin 2 is then solvated in a 3:1 THF/MeOH mixture (10 mL/g of
resin) and purged with Ar (20 min). NaOMe is added to reach a final
concentration of 0.2 M, and the reaction vessel is stoppered and shaken
for 1 h. Excess AcOH is added to quench the reaction, and the resin is
isolated by filtration. The resin is rinsed with 3:1 THF/MeOH (2×)
and CH2Cl2 (3×). A 0.1 M solution of the benzothiazolyl-activated
disulfide mesylate 3a in CH2Cl2 is added to the resin, and the reaction
Preparation of Disulfide Methanesulfonate Linkers (3a,b). Modi-
fications were made to the procedure reported by Brzezinska and Ternay
for the preparation of 2-benzothiazolyl disulfides22 since the purity and
yield could not be reproduced in our hands. 2,2′-Dithiobis(benzothia-
zole) (73.2 g, 220 mmol) was placed in a 5-L three-necked round-
bottom flask and dissolved in 3 L of chloroform with heating and
mechanical stirring. After all of the 2,2′-dithiobis(benzothiazole) had
dissolved, the solution was allowed to cool to room temperature. Over
a period of 2 h â-mercaptoethanol (14 mL, 200 mmol) in 500 mL of
chloroform was then added with an addition funnel. The reaction
solution was stirred for an additional 2 h and then extracted with 5%
NaOH (aqueous) (2 × 300 mL) and brine (2 × 300 mL). The
chloroform layer was dried (MgSO4), and the solvent was removed by
concentration in vacuo. To remove unreacted 2,2′-dithiobis(benzothia-
zole), the crude product was suspended in 700 mL of methanol and
refluxed with stirring for 30 min. The mixture was slowly cooled to 4
°C and maintained at 4 °C overnight. The crystallized 2,2′-dithiobis-
(benzothiazole) was removed by suction filtration and rinsed with cold
methanol. The methanol solution was concentrated in vacuo to yield
41.2 g (170 mmol, 85%) of a bright yellow solid which was used
without further purification. Mp 65 °C [lit. mp 69-71 °C].22 1H NMR
(300 MHz, CDCl3) δ 3.07 (t, 2, J ) 5.4), 3.91 (dt, 2, J ) 6.8, 5.4),
4.46 (t, 1, 6.8), 7.34 (td, 1, J ) 8.0, 1.2), 7.45 (td, 1, J ) 8.0, 1.2),
7.77 (dd, 1, J ) 8.0, 1.2), 7.90 (dd, 1, J ) 8.0, 1.2).
2-Benzothiazolyl 2-hydroxyethyl disulfide (32 g, 132 mmol) in 500
mL of methylene chloride was cooled to 0 °C, and 31 mL (263 mmol)
of 2,6-lutidine was added with mechanical stirring. Methanesulfonyl
anhydride (35.4 g, 197 mmol) was added in portions over 15 min. The
reaction mixture was then allowed to warm to room temperature while
stirring. When no starting material was left (0.3 h as determined by
TLC), the reaction mixture was washed with 1 N HCl (aqueous) (2 ×
500 mL), 1 N NaOH (aqueous) (2 × 500 mL), and brine (2 × 500
mL) and dried (MgSO4) and concentrated in vacuo. Purification by
column chromatography, eluting first with 7:1:2 hexanes/dichloroethane/
ethyl acetate, and then gradually increasing the dichloroethane and ethyl
acetate content, gave 37.6 g (117 mmol, 89%) of product 3a as white
crystals: 1H NMR (300 MHz, CDCl3) δ 3.04 (s, 3H), 3.29 (t, 2, J )
6.4), 4.55 (t, 2, J ) 6.4), 7.36 (td, 1, J ) 8.0, 1.2), 7.46 (td, 1, J ) 8.0,
1.2), 7.82 (dd, 1, J ) 8.0, 1.2), 7.90 (dd, 1, J ) 8.0, 1.2). 13C NMR
(300 MHz, CDCl3) δ 37.7, 38.0, 66.6, 121.2, 122.4, 125.0, 126.5, 135.9,
154.7, 170.2. Anal. Calcd for C10H11NO3S4 C, 37.36; H, 3.45; N, 4.36.
Found C, 37.50; H, 3.28; N, 4.12.
(3b): 81%. 1H NMR (400 MHz, CDCl3) δ 2.19 (tt, 2, J ) 7.0, 5.9),
2.98 (s, 3H), 3.04 (t, 2, J ) 7.0), 4.33 (t, 2, J ) 5.9), 7.31 (td, 1, J )
8.0, 1.2), 7.41 (td, 1, J ) 8.0, 1.2), 7.78 (dd, 1, J ) 8.0, 1.2), 7.84 (dd,
1, J ) 8.0, 1.2). 13C NMR (400 MHz, CDCl3) δ 28.3, 34.9, 37.2, 67.4,
(35) Vogel, A. I. Vogel’s Textbook of QuantitatiVe Inorganic Analysis,
4th ed.; Longmans: London, 1978; pp 342.

1824 J. Am. Chem. Soc., Vol. 121, No. 9, 1999
Souers et al.
vessel is stoppered under an Ar atmosphere. After 12 h, the resin is
isolated by filtration and rinsed with CH2Cl2 (5×) and MeOH (2×) to
afford 4. The mesylate 4 is treated with a 1.0 M solution of the desired
amine in NMP for 16 h at 50 °C. The resin is isolated by filtration and
rinsed with DMF (3×), CH2Cl2 (5×), and MeOH (2×) to give 5. Resin
5 is solvated in DMF (10 mL/g of resin) and the desired Fmoc-protected
amino acid (0.2 M), HATU (0.2 M), and i-Pr2EtN (0.4 M) are added.
After 8 h, the resin is collected by filtration and rinsed with DMF (3×),
CH2Cl2 (5×), and MeOH (2×). A solution of piperidine (20% v/v) in
DMF is used to deprotect the resin (20 min), and it is rinsed as before.
The deprotected resin is solvated in DMF (10 mL/g of resin) and the
desired R-bromo acid, HOAt (0.1 M), and DICI (0.1 M) are added.
After 4 h, the resin is isolated by filtration and rinsed as above to provide
6. The disulfide linkage of 6 to the solid support is reduced by treatment
with a 3.0 mM solution of TCEP in a 9:1 dioxane/H2O solution (1
mL/mmol of resin bound 6), which has been purged with Ar (20 min).
After 8 h, the solution of acyclic turn mimetic is transferred via a
filtration cannula and under Ar pressure to a flask containing ∼30 equiv
of support-bound tetramethyl guanidine. After the disappearance of all
of the acyclic material (<24 h as determined by HPLC), the solution
is filtered to remove the resin and the filtrate is concentrated in vacuo
to yield crude 1. The crude mimetics (1) were purified for analysis by
column chromatography on silica gel using EtOAc/hexane mixtures
as the eluant.
Determination of the Expected Loading of Linker (2). Amino-
methyl polystyrene resin (100 mg) was acylated with Fmoc-Aib-OH
(65 mg, 0.20 mmol) using PyBOP (104 mg, 0.20 mmol), HOBt (27
mg, 0.20 mmol), i-Pr2EtN (70 mL, 0.40 mmol) in 1 mL of DMF. The
conditions were identical to those used to load the linker precursor,
S-acetyl 2-mercapto-2-methyl propionic acid. Fmoc-Aib-OH should
couple to the aminomethyl resin similar to S-acetyl 2-mercapto-2-methyl
propionic acid since both have two geminal methyl substituents on the
R-carbon. The loading level of the resin (negative to bromophenol blue)
was then determined by spectrophotometric quantitation of the Fmoc
chromophore to be 0.50 ( 0.02 mequiv/g.36
1), 7.65 (m, 1), 8.01 (d, 1, J ) 8.1), 8.13 (m, 1). LRMS (MALDI-
TOF) mass calcd for C44H57N4O6S (MH+) 769.4. Found 769.4. Anal.
Calcd for C44H56N4O6S C, 68.72; H, 7.34; N, 7.29. Found C, 68.38; H,
7.21; N, 7.21.
(1 g): 1H NMR (d6-DMSO, 300 MHz, 75 °C) δ 0.20 (m, 2), 0.42
(m, 2), 0.89-0.96 (m, 1), 1.20-1.44 (m, 4), 1.36 (s, 9), 1.63 (s, 9),
1.60-1.81 (m, 2), 2.73-2.95 (m, 4), 3.03-3.26 (m, 4), 3.36-3.43 (m,
2), 3,84 (dd, 1, J ) 8.4, 5.3), 4.57 (m, 1), 6.32 (m, 1), 7.23 (t, 1, J )
6.9), 7.31 (t, 1, J ) 6.9), 7.42 (s, 1), 6.60 (d, 1, J ) 6.9), 7.60 (d, 1,
J ) 6.9), 8.02 (d, 1, J ) 8.1), 8.58 (d, 1, J ) 9.6). LRMS (MALDI-
TOF) mass calcd for C33H49N4O6S (MH+) 629.3. Found 629.6. Anal.
Calcd for C33H48N4O6S C, 63.03; H, 7.69; N, 8.91. Found C, 63.19; H,
7.46; N, 8.71.
Synthesis of the Focused Library of Somatostatin Mimetics. To
initiate library synthesis, the linker was coupled to large bead (40-60
mesh) aminomethyl resin and the support-bound thiol was generated
as described previously. 2-Benzothiazolylyl 2-(4,4′-dimethoxytrityloxy)-
ethyl disulfide (1.36 g, 2.50 mmol) dissolved in CH2Cl2 was then added
to 5.0 g of the support-bound thiol in a 100-mL round-bottom flask
and the reaction vessel stoppered under nitrogen and shaken for 21 h.
An aliquot of resin was exposed to Ellman’s reagent,23 and no color
change was observed, indicating that no free thiol remained in the solid
support. The resin was collected by filtration and rinsed with DMF,
CH2Cl2 (4×), and MeOH (2×), and dried in vacuo. The loading level
of the resin at this stage was determined by quantitation of the 4,4′-
dimethoxytrityl cation chromophore and found to be 0.19 ( 0.01
mequiv/g. The 4,4′-dimethoxytrityl ether was then cleaved by treatment
of the resin with 3% trichloroacetic acid in CH2Cl2 (2 × 3 min) and
then rinsed with CH2Cl2 (2×), MeOH, CH2Cl2 (2×), MeOH, and dried
over P2O5. The resin was then solvated in 30 mL of CH2Cl2 and 1.05
mL (9.0 mmol) of 2,6-lutidine followed by 1.05 g (6.0 mmol) of mesyl
anhydride were added while bubbling nitrogen through the 50-mL
peptide flask. The flask was stoppered and shaken for 1 h. The resin
was then rinsed with dry CH2Cl2 (4×), and the mesylation and rinsing
steps were repeated. To cap any nonmesylated alcohol, the resin was
solvated in 30 mL of CH2Cl2 and 1.57 mL (9.0 mmol) of i-Pr2EtN
followed by 427 µL of acetyl chloride were added. After 5 min, the
resin was rinsed with dry CH2Cl2 (4×) and dried in vacuo. Each of the
22 primary amines (2.0 mmol) were added to a sample vial and
dissolved in 2 mL of anhydrous NMP. Resin (200 mg each) was then
added to each vial and the vials immersed in a 50 °C oil bath for 12 h.
The reaction solvent was filtered away, and the resin batches were rinsed
with DMF (1 × 30 min) using a vacuum applied through a Teflon
cannula to remove solvent. The resin was then allowed to solvate in
2.0 mL/vial of 3:2 dichloroethane/DMF to attain an isopicnic slurry.
The resin was transferred as a slurry (210 µL/well) into the corre-
sponding wells of modified 8 × 12 pipet tip trays (Labsystems, Finntip
Universal, catalog No. 9400210). The pipet tip trays had been cut to
fit into the wells of a microtiter plate. Three holes had been drilled
(MC79 drill bit) in the bottom of each well to allow the passage of
solvent while retaining the 40-60 mesh beads. After transfer, the resin
beads were rinsed with DMF (1 × 10 min) and CH2Cl2 (3 × 10 min),
allowed to air-dry for 1 h, and rinsed with dry DMF (1 × 15 min).
The corresponding Fmoc amino acids (i + 2 position) and R-bromo
acids (i + 1 position) were then coupled, and the resin was rinsed as
described previously. A 300-mL mixture of 7:2:1 DMF/i-PrOH/H2O
and 5% v/v N-methylmorpholine was purged with Ar (20 min), and
7.5 mL of a 1.0 M solution of triethylphosphine in THF was added.
Each well of the two microtiter plates was treated with 750 µL of the
solution, and the beads were immersed in the microtiter plates. An
additional 400 µL of the cleavage solution was added through the top
of the modified pipet tip holders containing the resin beads. The
microtiter plate assemblies were placed in an oven under a nitrogen
atmosphere and heated to 40-45 °C for 16 h and then to 60 °C for 3.5
h. The modified pipet tip holders were elevated above the microtiter
plates in such a way that the solvent could drip into the microtiter wells
below for 10 min. Cleavage solvent was removed by concentration in
vacuo on a Jouan RC10 concentrator equipped with a microtiter-plate
rotor. Methanol (1 mL) was then added to each well, and the resulting
solutions were reconcentrated in vacuo. The side-chain protecting
groups were removed by addition of 0.4 mL of 18:1:1 TFA/dimethyl
(1a): 1H NMR (d6-DMSO, 300 MHz, 75 °C) δ 0.91 (m, 6), 1.17 (d,
3, J ) 6.6), 1.37 (m, 1), 1.59 (m, 1), 1.68 (m, 1), 2.85-3.37 (m, 7),
3.52 (q, 1, 6.6), 3.60 (m, 1), 4.52 (m, 1), 7.19 (t, 1, J ) 7.4), 7.31 (t,
2, J ) 7.4), 7.38 (d, 2, J ) 7.4), 8.48 (d, 1, J ) 9.5). LRMS (MALDI-
TOF) mass calcd for C19H29N2O2S2 (MH+) 381.2. Found 381.3. Anal.
Calcd for C19H28N2O2S2 C, 59.96; H, 7.42; N, 7.36. Found C, 59.99;
H, 7.56; N, 7.21.
(1b): 1H NMR (d6-DMSO, 300 MHz, 75 °C) δ 0.90 (d, 6, J )
6.2), 1.18-1.70 (m, 6), 2.70-3.57 (m, 8), 3.87 (m, 1), 4.61 (m, 1),
7.18 (t, 1, J ) 7.5), 7.31 (t, 2, J ) 7.5), 7.39 (d, 2, J ) 7.5), 8.00 (m,
1). LRMS (MALDI-TOF) mass calcd for C19H29N2O2S2 (MH+) 381.2.
Found 381.3. Anal. Calcd for C19H28N2O2S2 C, 59.96; H, 7.42; N, 7.36.
Found C, 59.71; H, 7.27; N, 7.17.
(1c): 1H NMR (d6-DMSO, 300 MHz, 75 °C) δ 0.77 (d, 3, J ) 6.7),
0.84 (d, 3, J ) 6.7), 1.17 (d, 3, J ) 6.6), 1.25-1.42 (m, 5), 1.37 (s, 9),
1.78-1.93 (m, 2), 2.71-2.78 (m, 1), 2.87-3.09 (m, 4), 3.18-3.41 (m,
3), 3.60 (q, 1, J ) 6.6), 4.52 (m, 1), 6.36 (m, 1), 8.58 (d, 1, J ) 9.4).
LRMS (MALDI-TOF) mass calcd for C20H37N3O4SNa (MNa+) 438.2.
Found 438.6. Anal. Calcd for C20H37N3O4S C, 57.80; H, 8.97; N, 10.11.
Found C, 57.72; H, 8.82; N, 9.91.
(1d): 1H NMR (d6-DMSO, 300 MHz, 75 °C) δ 0.77 (d, 3, J )
6.7), 0.86 (d, 3, J ) 6.7), 1.38 (s, 9), 1.90 (m, 1), 2.32 (m, 1), 2.55-
3.10 (m, 5), 3.24.3.40 (m, 4), 4.97 (m, 1), 8.60 (m, 1). Anal. Calcd for
C16H28N2O4S C, 55.79; H, 8.19; N, 8.13. Found C, 56.06; H, 8.38; N,
7.92.
(1e): 1H NMR (d6-DMSO, 400 MHz, 75 °C) δ 0.97 (d, 3, J ) 6.6),
1.03 (d, 3, J ) 6.6), 1.24 (d, 3, J ) 6.3), 1.94 (m, 1), 2.16 (m, 1),
2.38-2.50 (m, 3), 3.16 (m, 1), 3.33 (m, 1), 3.75 (m, 1), 3.78 (s, 3),
3.79 (s, 6), 4.40 (m, 1), 4.48 (m, 1), 5.27 (m, 1), 6.26 (s, 2), 8.06 (m,
1). LRMS (MALDI-TOF) mass calcd for C21H33N2O5S (MH+) 425.2.
Found 4252.
(1f): 1H NMR (d6-DMSO, 300 MHz, 75 °C) δ 0.1.14 (m, 2), 1.36
(s, 9), 1.63 (s, 9), 1.25-1.65 (m, 3), 1.78 (m, 1), 2.27 (m, 2), 4.10 (m,
1), 4.37 (m, 1), 4.74 (m, 1), 6.36 (m, 1), 7.13-7.34 (m, 12), 7.49 (m,
(36) Milligen Technical Note 3.10.

Synthesis and Screening of â-Turn Mimetic Libraries
J. Am. Chem. Soc., Vol. 121, No. 9, 1999 1825
sulfide/water to each well via Drummond manifold. After 0.5 h, the
cleavage solutions were reconcentrated in vacuo. DMSO (100 µL) was
added to each well, and the plates were stored at -20 °C. Some triethyl
phosphine oxide byproduct contaminated the wells of the resultant
library, but verification of the active wells with purified material
revealed that the contaminant had little or no effect on binding activity
in this competitive radioligand binding assay.
Seven compounds from each microtiter plate were picked at random
and analyzed by mass spectrometry in a matrix of R-cyano-4-
hydroxycinnamic acid on a Perspective Biosystems MALDI-TOF
instrument. Of the Fourteen 14 samples, 13 showed the proper molecular
ion.
(1l): 1H NMR (500 MHz, DMSO-d6, 75 °C) δ 1.10-1.60 (m, 5H),
1.90 (m, 1H), 2.62-4.21 (m, 10H), 4.53 (m, 1H), 4.75 (m, 1H), 5.95
(s, 2H), 6.72-6.87 (m, 3H), 6.97-7.19 (m, 3H), 7.32 (m, 1H), 7.55
(m, 1H), 8.21 (m, 1H), 10.65 (m, 1H). LRMS (MALDI-TOF) R-cyano-
4-hydroxycinnamic acid matrix mass calcd for C27H33N4O4S (MH+)
509.2, found 509.6.
(1m): 1H NMR (300 MHz, MeOH-d4, 55 °C) δ 1.22-1.29 (m, 4H),
1.41-1.52 (m, 2H), 2.22-2.37 (m, 2H), 2.72-2.79 (m, 3H), 2.89-
2.93 (m, 1H), 3.18-3.21 (M, 1H), 3.30-3.34 (m, 2H), 3.51-3.59 (m,
2H), 4.61-4.72 (m, 2H), 7.21-7.39 (m, 2H), 7.50 (s, 1H), 7.57-7.59
(m, 1H), 7.71-7.76 (m, 1H). HRMS (FAB) exact mass calcd for
C20H28N4O2S (MH+) 389.2009, found 389.2004.
Synthesis of â-turn Mimetics for Validation of Bioactivity. Five
of the most potent compounds from the directed library and three
alanine scan compounds were synthesized on a multimilligram scale
as described previously. These compounds were purified by silica
column chromatography and characterized by 1H NMR and low
resolution mass spectrometry MALDI-TOF or FAB. The characteriza-
tion data are listed below after the appropriate compound code.
(1h): 1H NMR (500 MHz, DMSO-d6, 75 °C) δ 1.20-1.45 (m, 3H),
1.60-1.70 (m, 2H), 1.92 (m, 1H), 2.62-3.60 (m, 5H), 3.98 (m, 1H),
4.11-4.21 (m, 2H), 4.61 (m, 1H), 4.79 (m, 1H), 5.38 (m, 1H), 5.68
(d, J ) 15.6 Hz, 1H), 6.97-7.19 (m, 3H), 7.30-7.59 (m, 6H), 7.84
(d, J ) 8.8, 1H), 7.92 (m, 1H), 8.06 (m, 1H), 8.29 (m, 1H), 10.60 (m,
1H). LRMS (MALDI-TOF) R-cyano-4-hydroxycinnamic acid matrix:
mass calcd for C30H35N4O2S (MH+) 515.2, found 515.3.
1
(1n): H NMR (300 MHz, MeOH-d4, 55 °C) δ 1.35 (d, J ) 6.5,
3H), 2.41-2.45 (m, 1H), 2.62-2.74 (m, 1H), 3.07-3.09 (m, 2H), 3.31-
3.42 (m, 2H), 3.92-3.95 (m, 1H), 4.13 (d, J ) 13.0, 1H), 4.91-4.95
(m, 1H), 5.68-5.73 (d, J ) 13.0, 1H), 7.18-7.22 (m, 3H), 7.25 (d, J
) 7.3, 1H), 7.38 (d, J ) 6.1, 2H), 7.49-7.47 (m, 2H), 7.63 (d, J )
7.3, 1H), 7.78-7.84 (m, 2H), 8.04 (d, J ) 7.7, 1H). HRMS (FAB)
exact mass calcd for C27H27N3O2S (MH+) 558.2413, found 558.2415.
(1o): 1H NMR (300 MHz, DMSO-d6, 75 °C) δ 1.18-1.22 (m, 1H),
1.29 (m, 3H), 1.41-1.49 (m, 2H), 1.51-1.67 (m, 2H), 2.58-2.62 (m,
2H), 2.73-2.68 (m, 1H), 2.91-2.97 (m, 2H), 3.04-3.13 (m, 1H), 3.49-
3.56 (m, 1H), 3.83-3.87 (m, 1H), 4.0 (d, J ) 14.4, 1H), 4.69-4.76
(m, 1H), 5.75 (d, J ) 14.5, 1H), 7.41-7.59 (m, 3H), 7.62-7.71 (m,
2H), 7.80-7.83 (m, 1H), 7.91-7.98 (m, 1H). HRMS (FAB) exact mass
calcd for C22H29N3O2S (MH+) 400.2045, found 400.2050.
(1i): 1H NMR (500 MHz, DMSO-d6, 75 °C) δ 1.20-1.45 (m, 3H),
1.60-1.70 (m, 2H), 1.92 (m, 1H), 2.62-3.60 (m, 5H), 3.98 (m, 1H),
4.11-4.21 (m, 2H), 4.61 (m, 1H), 4.79 (m, 1H), 5.38 (m, 1H), 5.68
(d, J ) 15.6 Hz, 1H), 6.97-7.19 (m, 3H), 7.30-7.59 (m, 6H), 7.84
(d, J ) 8.8, 1H), 7.92 (m, 1H), 8.06 (m, 1H), 8.29 (m, 1H), 10.60 (m,
1H). LRMS (MALDI-TOF) R-cyano-4-hydroxycinnamic acid matrix
mass calcd for C30H35N4O2S (MH+) 515.2, found 515.6.
Procedure for Radioligand Binding Assay. Cell membranes (2-
30 µg protein) were incubated with 0.03 nM [125I]-[Tyr11]-SRIF and
increasing concentrations of SRIF or SRIF analogues for 90 min at
room temperature.37 Nonspecific binding was defined with 1 µM SRIF.
The assay was terminated by rapid filtration through Whatman GF/C
glass fiber filters soaked in 0.5% polyethylenimine (PEI), followed by
3 × 3 mL washes of 50 mM Tris-HCl pH 7.4 containing 5 mg/mL
bovine serum albumin (BSA). Radioactivity in the filters was deter-
mined using a Canberra Packard Cobra II Auto-counter. Data were
analyzed by nonlinear regression using the curve-fitting programs
EBDA38 and LIGAND.39
(1j): 1H NMR (500 MHz, DMSO-d6, 75 °C) δ 1.32 (q, J ) 7.5 Hz,
1H), 1.47 (m, 1H), 1.57-1.69 (m, 2H), 1.92 (m, 1H), 2.75 (m, 3H),
2.83 (m, 2H), 3.00-3.10 (m, 1H), 3.25 (d, J ) 14.6, 8.9 Hz, 1H), 3.37
(m, 1H), 3.78 (dd, J ) 8.9,5.3 Hz, 1H), 4.67 (d, J ) 8.0, 1H), 4.73 (m,
1H), 5.22 (d, J ) 15.4, 1H), 6.96 (t, J ) 7.5, 1H), 7.06 (m, 2H), 7.32
(d, J ) 8.0, 1H), 7.38 (d, J ) 6.9 Hz, 1H), 7.45 (t, J ) 8.0 Hz, 1H),
7.50-7.54 (m, 3H), 7.85 (d, J ) 8.0 Hz, 1H), 7.93 (m, 1H), 8.03 (m,
1H), 8.62 (d, J ) 9.7, 1H), 10.64 (s, 1H). LRMS (MALDI-TOF)
R-cyano-4-hydroxycinnamic acid matrix mass calcd for C30H35N4O2S
(MH+) 515.2, found 515.4.
Acknowledgment. This work was supported by funding
from the National Institutes of Health (GM53696). Å.R. would
like to thank the Knut and Alice Wallenberg Foundation for
funding.
(1k): 1H NMR (500 MHz, DMSO-d6, 75 °C) δ 1.16-1.26 (m, 2H),
1.49-1.59 (m, 3H), 1.83 (m, 1H), 2.60-2.80 (m, 4H), 2.95-3.46 (m,
3H), 3.60 (m, 2H), 3.75 (m, 1H), 3.91 (m, 1H), 4.30 (m, 1H), 4.46 (m,
1H), 4.65 (m, 1H), 6.97 (t, J ) 7.5, 1H), 7.06 (t, J ) 7.5 Hz, 1H),
7.14 (m, 1H), 7.19 (m, 1H), 7.33 (m, 2H), 7.40 (m, 2H), 7.50-7.57
(m, 2H), 8.17 (m, 1H), 10.63 (s, 1H). HRMS (FAB) exact mass calcd
for C27H35N4O4S2 (MH+) 511.2201, found 511.2202.
JA983742P
(37) Castro, S. W.; Buell, G.; Feniuk, W.; Humphrey, P. P. A. Br. J.
Pharmacol. 1996, 117, 639-646.
(38) Mcpherson, G. A. J. Pharmacol. Methods 1985, 14, 213-228.
(39) Munson, P. J.; Rodbard, D. Anal. Biochem. 1980, 107, 220-239.