A novel and rapid encoding method based on mass spectrometry for "one-bead-one-compound" small molecule combinatorial libraries

Article abstract of DOI:10.1021/ja034539j

A novel and efficient encoding method based on mass spectrometry for "one-bead-one-compound" small molecule combinatorial libraries has been developed. The topologically segregated bifunctional resin beads with orthogonal protecting groups in the outer an

Full text of DOI:10.1021/ja034539j

Published on Web 04/29/2003
A Novel and Rapid Encoding Method Based on Mass
Spectrometry for “One-Bead-One-Compound” Small Molecule
Combinatorial Libraries
Aimin Song,† Jinhua Zhang,‡ Carlito B. Lebrilla,‡ and Kit S. Lam*,†
Contribution from the DiVision of Hematology and Oncology, Department of Internal Medicine,
UC DaVis Cancer Center, 4501 X Street, Sacramento, California 95817 and Department of
Chemistry, UniVersity of California, DaVis, California 95616
Received February 6, 2003; E-mail: kit.lam@ucdmc.ucdavis.edu
Abstract: A novel and efficient encoding method based on mass spectrometry for “one-bead-one-compound”
small molecule combinatorial libraries has been developed. The topologically segregated bifunctional resin
beads with orthogonal protecting groups in the outer and inner regions are first prepared according to our
previously published procedure. Prior to library synthesis, the inner core of each bead is derivatized with
3-4 different coding blocks on a cleavable linker. Each functional group on the scaffold is encoded by an
individual coding block containing a functional group with the same chemical reactivity. During the library
synthesis, the same chemical reactions take place on the scaffold (outer layer of the bead) and coding
blocks (inner core of the bead) concurrently. After screening, the coding tags in the positive beads are
released, followed by molecular mass determination using matrix-assisted laser desorption ionization Fourier
transform mass spectrometry. The chemical structure of library compounds can be readily identified
according to the molecular masses of the coding tags. The feasibility and efficiency of this approach were
demonstrated by the synthesis and screening of a model small molecule library containing 84 672 member
compounds, with a model receptor, streptavidin. Streptavidin binding ligands with structural similarity (17)
were identified. The decoding results were clear and unambiguous.
Introduction
bead in addition to the library component. This tag defines the
chemical history of any particular bead and hence the structure
Combinatorial chemistry has become an essential component
of the drug discovery process during the past decade. In 1991,
we first introduced the “one-bead-one-compound” (OBOC)
combinatorial library method.1 Using a “split-mix” synthesis
procedure,1,2 peptide or chemical libraries can be generated such
that each bead displays only one compound entity. With an on-
bead screening assay, literally millions of compound beads can
be screened against specific molecular targets in a few days.1
Individual positive beads can then be isolated for structure
determination. This approach has been successfully applied to
the identification of ligands for a large number of biological
targets.3 Peptide beads can easily be microsequenced with
Edman degradation using an automatic protein sequencer. In
contrast, structure determination of small molecule beads is
challenging. To solve this problem, various indirect encoding
methods have been developed and the subject has been
reviewed.3,4 In most cases, a coding tag is synthesized on each
of the compound it supports. It is released from the bead after
biological screening and analyzed by a highly sensitive analytical
technique. For example, electron capture capillary gas chroma-
tography has been successfully used for the detection of volatile
halocarbon tags released from the beads via photolytic5 or
oxidative6 cleavage; reverse-phase HPLC with fluorescence
detection has been used to determine the dansylated secondary
amine tags released by mineral acid hydrolysis,7 and automatic
microsequencing has been used to decode the peptide tags.8,9
Several physical encoding methods have also been described,
but they are either not possible with highly diverse libraries
(e.g., 250 000 compounds)10 or too big to be enclosed in a single
80 µm bead.11
Mass spectrometry (MS) has been widely used in the analysis
of combinatorial libraries due to its intrinsic sensitivity, speed
(5) 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. U.S.A.
1993, 90, 10922-10926.
(6) Nestler, H. P.; Bartlett, P. A.; Still, W. C. J. Org. Chem. 1994, 59, 4723-
4724.
† Department of Internal Medicine.
‡ Department of Chemistry.
(1) Lam, K. S.; Salmon, S. E.; Hersh, E. M.; Hruby, V. J.; Kazmierski, W.
M.; Knapp, R. J. Nature 1991, 354, 82-84.
(7) (a) Ni, Z. J.; Maclean, D.; Holmes, C. P.; Murphy, M. M.; Ruhland, B.;
Jacobs, J. W.; Gordon, E. M.; Gallop, M. A. J. Med. Chem. 1996, 39,
1601-1608. (b) Murphy, M. M.; Schullek, J. R.; Gordon, E. M.; Gallop,
M. A. J. Am. Chem. Soc. 1995, 117, 7029-7030.
(2) (a) Houghten, R. A.; Pinilla, C.; Blondelle, S. E.; Appel, J. R.; Dooley, C.
T.; Cuervo, J. H. Nature 1991, 354, 84-86. (b) Furka, A.; Sebestye´n, F.;
Asgedom, M.; Dibo´, G. Int. J. Pept. Protein Res. 1991, 37, 487-493.
(3) Lam, K. S.; Krchnak, V.; Lebl, M. Chem. ReV. 1997, 97, 411-448.
(4) (a) Czarnik, A. W. Curr. Opin. Chem. Biol. 1997, 1, 60-66. (b) Xiao, X.
Y. Front. Biotechnol. Pharm. 2000, 1, 114-149. (c) Barnes, C.; Bala-
subramanian, S. Curr. Opin. Chem. Biol. 2000, 4, 346-350. (d) Affleck,
R. L. Curr. Opin. Chem. Biol. 2001, 5, 257-263.
(8) (a) Nikolaiev, V.; Stierandova´, A.; Krchna´k, V.; Seligmann, B.; Lam, K.
S.; Salmon, S. E.; Lebl, M. Pept. Res. 1993, 6, 161-170. (b) Kerr, J. M.;
Banville, S. C.; Zuckermann, R. N. J. Am. Chem. Soc. 1993, 115, 2529-
2531.
(9) Liu, R.; Marik, J.; Lam, K. S. J. Am. Chem. Soc. 2002, 124, 7678-7680.
9
6180
J. AM. CHEM. SOC. 2003, 125, 6180-6188
10.1021/ja034539j CCC: $25.00 © 2003 American Chemical Society

MS Encoding Method for Small Molecule Libraries
A R T I C L E S
Scheme 1. General Synthetic and Encoding/Decoding Strategy of
Encoded Libraries
of analysis, specificity of detection, and automation capability.
Sequencing of peptides by MS has been developed,12,13 and
databases for MS sequencing of peptide libraries are com-
mercially available. Because simultaneous cleavage and ioniza-
tion occur under laser irradiation, peptides covalently attached
to a single polymeric bead by a photosensitive linker can be
directly sequenced by matrix-assisted laser desorption ionization
(MALDI) MS.13 Structure-indicating fragments can be readily
obtained by collision-induced dissociation (CID) or tandem MS
(MS/MS). The single-bead analysis approach for peptides has
also been applied to peptidomimetic and small molecule
compound libraries.14 For small libraries, the component on a
single bead can be directly identified with the exact molecular
mass and analysis of fragments. However, direct structure
identification of a compound on a single bead isolated from a
large diverse library of over 100 000 compounds will be very
difficult if not impossible. This is because compounds from such
combinatorial libraries often have molecular masses differing
by less than 200 Da, and some of them are poorly ionizable or
nonionizable. Some indirect MS-based encoding/decoding strat-
egies have been described. For example, stable isotopes (13C
or deuterium) have been incorporated to the coding tags to
generate different isotopic patterns for structure characteriza-
tion.15 Another elegant approach, but one limited to libraries of
ionizable compounds with repeating subunits, such as peptides,
has been developed by Youngquist et al.16 A small amount of
a termination reagent is added at each synthetic step to generate
a series of sequence-specific terminated products. Analysis of
such products generated from one single bead, with MALDI-
MS, enable one to reconstruct the complete peptide sequence.
Current chemical encoding methods have played an important
role in the advancement of OBOC combinatorial chemistry.
However, those methods often require orthogonal chemistries
for tagging and, therefore, additional synthetic steps. For
example, in the halocarbon encoding method developed by Still
et al.,5,6 an additional 16-24 h are needed to encode each
building block. In addition to the increased time and cost, the
tagging molecules themselves potentially could interfere during
the screening. To address these problems, our group has
developed a new peptide-based encoding method that enables
us to topologically segregate the testing compounds from the
coding tags.9 The resin beads are first derivatized with orthogo-
nal protecting groups in the outer and inner regions separately.
A coding tag precursor consisting of a sequence of R-amino
acids, of which the side chains can be derivatized, is then
constructed in the interior of the beads. During the library
synthesis, building blocks are coupled to the outer scaffold and
the side chains of the inner coding peptide simultaneously. In
this way, the extra synthetic steps for coding tags are eliminated
by combining them with the library synthesis. After biological
screening, the structures of active compounds can be easily
determined by direct sequencing of the coding peptides with
Edman degradation. The utility and reliability of this encoding
method has been confirmed. However, like other encoding
strategies, this method has its limitations. First, this decoding
method is based on Edman degradation and, therefore, is slow
and expensive. Second, building blocks have to be carefully
chosen to avoid retention time overlap of their amino acid
derivatives during sequencing. Third, unless one wants to
synthesize trifunctional R-amino acids with unusual side chains,
choices for scaffolds are limited to those having the same
functional groups as the side chains of commercially available
trifunctional amino acids. Most recently, we described an
improvement of the previously mentioned peptide encoding
method by using mass spectrometry to decode the peptide tag.17
This new development alleviates the first two limitations
outlined previously. However, the analysis of the decoding
results could be complicated, and the method has not yet been
routinely applied to our library screening efforts.
In an effort to continue to improve the OBOC technology,
we have recently developed a new and highly efficient MS-
based encoding strategy that is particularly suited to the small
organic molecule libraries (Scheme 1). In this encoding system,
the topologically segregated bifunctional resin beads are used
to segregate the library compounds (outer layer) from coding
tags (inner core). Each functional group (X, Y, and Z) on the
scaffold (S) is encoded by an individual coding block (C1, C2,
and C3) containing a functional group (X′, Y′ and Z′) that has
the same chemical reactivity as the one on the scaffold. The
chemical reactions for library synthesis occur on the scaffold
as well as on inner coding blocks at the same time to form the
coding tags. After screening, the coding tags in the positive
beads are released by chemical cleavage and characterized by
(10) (a) Vaino, A. R.; Janda, K. D. Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
7692-7696. (b) Guiles, J. W.; Lanter, C. L.; Rivero, R. A. Angew. Chem.,
Int. Ed. 1998, 37, 926-928.
(11) (a) Nicolaou, K. C.; Xiao, X. Y.; Parandoosh, Z.; Senyei, A.; Nova, M. P.
Angew. Chem., Int. Ed. Engl. 1995, 34, 2289-2291. (b) Moran, E. J.;
Sarshar, S.; Cargill, J. F.; Shahbaz, M. M.; Lio, A.; Mjalli, A. M. M.;
Armstrong, R. W. J. Am. Chem. Soc. 1995, 117, 10787-10788. (c) Li,
W.; Czarnik, A. W.; Lillig, J.; Xiao, X. Y. J. Comb. Chem. 2000, 2, 224-
227.
(12) (a) Biemann, K.; Martin, S. A. Mass Spectrom. ReV. 1987, 6, 1-76. (b)
Egner, B. J.; Langley, G. J.; Bradley, M. J. Org. Chem. 1995, 60, 2652-
2653.
(13) (a) Siuzdak, G.; Lewis, J. K. Biotechnol. Bioeng. 1998, 61, 127-134. (b)
Fitzgerald, M. C.; Harris, K.; Shevlin, C. G.; Siuzdak, G. Bioorg. Med.
Chem. Lett. 1996, 6, 979-982.
(14) (a) Haskins, N. J.; Hunter, D. J.; Organ, A. J.; Rahman, S. S.; Thom, C.
Rapid Commun. Mass Spectrom. 1995, 9, 1437-1440. (b) Lorthioir, O.;
Carr, R. A. E.; Congreve, M. S.; Geysen, M. H.; Kay, C.; Marshall, P.;
McKeown, S. C.; Parr, N. J.; Scicinski, J. J.; Watson, S. P. Anal. Chem.
2001, 73, 963-970.
(15) (a) Geysen, H. M.; Wagner, C. D.; Bodnar, W. M.; Markworth, C. J.; Parke,
G. J.; Schoenen, F. J.; Wagner, D. S.; Kinder, D. S. Chem. Biol. 1996, 3,
679-688. (b) Wagner, D. S.; Markworth, D. J.; Wagner, C. D.; Schoenen,
F. J.; Rewerts, C. E.; Kay, B. K.; Geysen, H. M. Comb. Chem. High
Throughput Screening 1998, 1, 143-153. (c) Lane, S. J.; Pipe, A. Rapid
Commun. Mass Spectrom. 2000, 14, 782-793.
(16) (a) Youngquist, R. S.; Fuentes, G. R.; Lacey, M. P.; Keough, T. Rapid
Commun. Mass Spectrom. 1994, 8, 77-81. (b) Youngquist, R. S.; Fuentes,
G. R.; Lacey, M. P.; Keough, T. J. Am. Chem. Soc. 1995, 117, 3900-
3906.
(17) Franz, A. H.; Liu, R.; Song, A.; Lam, K. S.; Lebrilla, C. B. J. Comb. Chem.
2003, 5, 125-137.
9
J. AM. CHEM. SOC. VOL. 125, NO. 20, 2003 6181

A R T I C L E S
Song et al.
Scheme 2. Synthesis and Cleavage of the Linker 1a
MS. The structures of active compounds can be readily identified
according to the exact molecular masses of coding tags. To
demonstrate the feasibility and efficiency of this encoding
method, a model small molecule library containing 84 672
members is synthesized and screened against streptavidin.
Results and Discussion
General Encoding Strategy. The general synthetic and
encoding strategy for our encoded libraries is illustrated in
Scheme 1. The resin beads are first derivatized with orthogonal
protecting groups in the outer (P, 60%) and inner (P′, 40%)
regions separately using our published procedure.9 A cleavable
linker, which can facilitate the mass determination of coding
compounds, is then built in the interior of the bead. Coding
blocks (C1, C2, and C3) chosen according to the functional
groups (X, Y, and Z) on the scaffold (S) are then anchored to
the linker. Each coding block contains only one functional group,
which has the same chemical reactivity as the one on the
scaffold. The scaffold for library synthesis is then coupled to
the outer layer of the beads. During the library synthesis, the
same building blocks (B1, B2, and B3) are coupled to the scaffold
and inner coding blocks concurrently. After screening, the
positive beads are isolated and treated with cleavage reagents.
Only the inner coding tags are released from the beads. The
cleavage solution containing coding tags is then analyzed with
high resolution MS. According to the molecular masses of
coding tags, the chemical structure of the library compound on
the outer layer of the bead can be readily identified.
We use matrix-assisted laser desorption/ionization Fourier
transform mass spectrometry (MALDI-FTMS) for decoding due
to its high resolution, high accuracy, and high sensitivity. Since
we need only the molecular masses of coding tags to identify
the structure of a library compound, a very small amount
(picomole or even femtomole) of coding compounds is enough
for MALDI-FTMS detection. Considering a library based on a
scaffold with four diversities, if we use 100 different building
blocks in each synthetic step, we will get a library containing
1004 compounds. The total number of coding tags for such a
library is only 400. MALDI-FTMS, with an accuracy down to
a few ppm, can easily discriminate all the 400 different coding
tags as long as they are not isomers, which can be easily avoided
by careful selection of building blocks and coding blocks.
Because each coding tag constitutes only about a 10% equivalent
of the whole bead, and they all reside in the bead interior,
interference by the coding tags during biological screening
should be minimal.
a Reagents and conditions: (i) 3 equiv of Fmoc-Met-OH, DIC, and HOBt
in DMF, rt, 1 h; (ii) 20% piperidine in DMF, rt, 30 min; (iii) 3 equiv of
Fmoc-Phe-OH, DIC, and HOBt in DMF, rt, 1h; (iv) 3 equiv of Fmoc-
NH(CH2CH2O)2(CH2)2NHCO(CH2)2COOH, DIC, and HOBt in DMF, rt,
3h; (v) 0.25 M CNBr in 70% formic acid, rt, overnight.
peptide-like linker (Scheme 2, 1) that meets these four criteria
has been designed and synthesized on solid phase using the
standard Fmoc chemistry.19 In principle, any chemically cleav-
able or photosensitive linkers can be used as the cleavable part
as long as they are compatible with the library synthesis and
screening. Methionine is chosen in our study because its
cleavage by cyanogen bromide (CNBr) is clean and specific
and the final homoserine lactone product20 is chemically stable.
This cleavage method has been successfully applied to single-
bead analysis of peptides.16 Two phenylalanines are coupled to
the methionine to increase the molecular weight of the linker.
Finally, a linear hydrophilic molecule is introduced to the linker
to enhance solubility of the coding tag in the extraction solvent
(50% acetonitrile/water). The whole linker has excellent chemi-
cal stability and is very suitable for MALDI-FTMS detection.
The oxygen atoms, the amide bonds, and the side chain of
phenylalanines in the linker allow efficient formation of
primarily sodiated species and, therefore, provide efficient
ionization.
To evaluate the utility of the linker in encoding, we
synthesized five different compounds (Scheme 3) in the inner
core of topologically segregated bifunctional TentaGel beads
(inner core:outer layer ) 2:3) via the linker shown in Scheme
2. MS analysis of a single bead from this synthesis revealed
the expected five mass signals (Figure 1). The result is clear
and unambiguous, indicating that the linker does meet the
requirements for the proposed encoding method.
Selection and Relative Reactivity of Coding Blocks. Coding
blocks are chosen according to the functional groups on the
scaffold for encoded library synthesis. Each of the coding blocks
should have an identical or related functional group on the
scaffold and a carboxyl group through which the coding block
is anchored to the linker. Due to the commercial availability of
thousands of organic acids, one can easily find coding blocks
for almost any selected scaffold. An ideal coding block should
react with the corresponding building block to generate a stable
product that is inert to subsequent chemical reactions. For
instance, proline could be used to encode reductive alkylation
Linker Design. An ionization linker has been used to enhance
ionization of poorly ionizable or nonionizable molecules.18 The
linker also provides a mass shift which overcomes signal overlap
with matrix molecules. To effectively decode each bead with
mass spectrometry, the linker should meet the following four
criteria. First, the linker must be inert to the chemical reactions
for library synthesis and stable under the conditions used for
various biological screenings. Second, the linker should be
highly sensitive to the ionization method so that the final coding
tags with different structures can be readily detected. Third, its
cleavage must be clean and efficient. Fourth, the linker should
have excellent solubility in the extraction solvent. A simple
(18) Carrasco, M. R.; Fitzgerald, M. C.; Oda, Y.; Kent, S. B. H. Tetrahedron
Lett. 1997, 38, 6331-6334.
(19) Fields, G. B.; Noble, R. L. Int. J. Pept. Protein Res. 1990, 35, 161-214.
(20) Gross, E.; Witkop, B. J. Biol. Chem. 1962, 237, 1856-1860.
9
6182 J. AM. CHEM. SOC. VOL. 125, NO. 20, 2003

MS Encoding Method for Small Molecule Libraries
A R T I C L E S
Scheme 3. Synthesis of Five Compounds on the Linker in the
Inner Core of Topologically Segregated Bifunctional TentaGel
Beads (Inner Core:Outer Layer ) 2:3)a
Scheme 4. Synthetic and Encoding Reactions of a Model
Compounda
a Reagents and conditions: (i) 20% piperidine in DMF, rt, 30 min; (ii)
2 equiv of Trt-Gly-OH, benzoic acid, Fmoc-nipecotic acid, 4-nitropheny-
lacetic acid, and 4-(chloromethyl)benzoic acid, 10 equiv of DIC and HOBt
in DMF, rt, 1 h; (iii) 1% TFA and 5% TIS in DCM, rt, 2 min × 4; (iv)
0.25 M piperidine in 5% DIEA/DMF, rt, overnight; (v) 2 M SnCl2‚2H2O
in DMF, rt, 6h; (vi) 20 equiv of Ac2O, 5 equiv of DIEA in DCM, rt,
overnight.
a Reagents and conditions: (i) 20% piperidine in DMF, rt, 30 min; (ii)
a mixture of Trt-Gly-OH (2.01 equiv), N-phthaloylglycine (0.34 equiv),
and 4-nitrophenylacetic acid (3.45 equiv), HOBt (6 equiv), and DIC (6
equiv) in DMF, rt, 2 h; (iii) 1% TFA and 5% TIS in DCM, 2 min × 4; (iv)
5 equiv of benzoic acid, HOBt, and DIC in DMF, rt, 2 h; (v) 50% TFA in
DCM, rt, 30 min; (vi) 5 equiv of Boc-Phe-OH, HOBt, and DIC in DMF,
rt, 2 h; (vii) 50% TFA in DCM, rt, 30 min; (viii) 5 equiv of 4-fluoro-3-
nitrobenzoic acid, HOBt, and DIC in DMF, rt, 2 h; (ix) 0.25 M propylamine
in 5% DIEA/DMF, rt, overnight; (x) 2 M SnCl2‚2H2O in DMF, 3 h × 2;
(xi) 20 equiv of chloroacetic anhydride and 4 equiv of DIEA in DCM, rt,
overnight; (xii) 10% DIEA in DMF, rt, 6h; (xiii) 10% piperidine and 10%
DIEA in DMF, rt, overnight.
model scaffold (Scheme 4). The carboxyl group of the scaffold
is usually connected to the solid support via an amide or an
ester bond. If an additional diversity is desired, an amino acid
may be introduced prior to the coupling of the scaffold. The
(21) (a) Lee, J.; Murray, W. V.; Rivero, R. A. J. Org. Chem. 1997, 62, 3874-
3879. (b) Lee, J.; Gauthier, D.; Rivero, R. A. Tetrahedron Lett. 1998, 39,
201-204. (c) Schwarz, M. K.; Tumelty, D.; Gallop, M. A. Tetrahedron
Lett. 1998, 39, 8397-8400. (d) Mayer, J. P.; Lewis, G. S.; McGee, C.;
Bankaitis-Davis, D. Tetrahedron Lett. 1998, 39, 6655-6658. (e) Tumelty,
D.; Schwarz, M. K.; Needels, M. C. Tetrahedron Lett. 1998, 39, 7467-
7470. (f) Wijkmans, J. C. H. M.; Culshaw, A. J.; Baxter, A. D. Mol.
DiVersity 1998, 3, 117-120. (g) Morales, G. A.; Corbett, J. W.; DeGrado,
W. F. J. Org. Chem. 1998, 63, 1172-1177. (h) Stephensen, H.; Zaragoza,
F. Tetrahedron Lett. 1999, 40, 5799-5802. (i) Schwarz, M. K.; Tumelty,
D.; Gallop, M. A. J. Org. Chem. 1999, 64, 2219-2231. (j) Lee, J.; Gauthier,
D.; Rivero, R. A. J. Org. Chem. 1999, 64, 3060-3065. (k) Smith, J. M.;
Gard, J.; Cummings, W.; Kanizsai, A.; Krchna´k, V. J. Comb. Chem. 1999,
1, 368-370. (l) Kilburn, J. P.; Lau, J.; Jones, R. C. F. Tetrahedron Lett.
2000, 41, 5419-5421. (m) Wu, Z.; Rea, P.; Wickham, G. Tetrahedron
Lett. 2000, 41, 9871-9874. (n) Nefzi, A.; Giulianotti, M. A.; Houghten,
R. A. Tetrahedron Lett. 2000, 41, 2283-2287. (o) Yokum, T. S.; Alsina,
J.; Barany, G. J. Comb. Chem. 2000, 2, 282-292. (p) Yeh, C. M.; Tung,
C. L.; Sun, C. M. J. Comb. Chem. 2000, 2, 341-348. (q) Krchna´k, V.;
Smith, J.; Va´gner, J. Tetrahedron Lett. 2001, 42, 1627-1630.
Figure 1. MALDI-FTMS spectrum of five tags in the 40% inner region
of a single TentaGel bead.
on the scaffold to form a stable tertiary amine, while glycine is
not a good choice unless the reductive alkylation is the last
synthetic step. This is because the secondary amine generated
is still reactive to acylation or reductive alkylation.
A number of small molecule libraries based on 4-fluoro-3-
nitrobenzoic acid scaffold have been reported by several
groups.21,22 The ortho-nitro fluoride in this molecule has been
shown to undergo facile aromatic nucleophilic substitution with
sulfur or nitrogen nucleophiles followed by the reduction of
the nitro group. To illustrate the selection of coding blocks in
our encoding method, we use 4-fluoro-3-nitrobenzoic acid as a
(22) Zaragoza, F.; Stephensen, H. J. Org. Chem. 1999, 64, 2555-2557.
9
J. AM. CHEM. SOC. VOL. 125, NO. 20, 2003 6183

A R T I C L E S
Song et al.
Table 1. Relative Coupling Efficiency of Some Coding Blocks
relative
coding blocks
benzoic acid
acrylic acid
bromoacetic acid
4-bromomethylphenylacetic acid
4-(chloromethyl)benzoic acid
4-maleimidobutyric acid
N-phthaloylglycine
N-tritylglycine (Trt-Gly-OH)
N-Fmoc-piperidinecarboxylic acid
3-nitrophenylacetic acid
reactivity
encoded reaction
1.00
1.79
17.76
1.39
1.81
0.47
5.71
1.55
0.93
2.75
Michael addition
nucleophilic substitution
nucleophilic substitution
nucleophilic substitution
nucleophilic substitution
nucleophilic substitution
amino acid coupling
reductive alkylation
aromatic reduction
and acylation
4-nitrophenylacetic acid
0.70
aromatic reduction
and acylation
amino acid can easily be encoded by a precoupled carboxylic
acid in the interior of the bead. Either nitrobenzoic acid or
nitrophenylactetic acid is a good choice to encode the nitro
group. However, we cannot use an ortho-nitro compound to
encode the ortho-nitro fluoride, since the nitro group in the
coding block will be reactive to the subsequent reactions.
Alternately, we might select bromoacetic acid, 4-bromometh-
ylphenylacetic acid, or 4-(chloromethyl)benzoic acid as the
coding block if the building block is going to be a sulfur
nucleophile or a secondary amine.23 In this case, the reaction
product will be a stable thioether or tertiary amine. Those acids
can be also used to encode Michael addition, since the same
nucleophiles are used as the reactant. For primary amine building
blocks, which react with the previously mentioned coding blocks
to generate a secondary amine, 4-maleimidobutanoic acid or
N-phthaloylglycine can be chosen because the reaction generates
a stable diamide compound (Scheme 4).24
The loading of each of the coding blocks should be similar
so that comparable signal intensity of each decoding peak can
be obtained during MS analysis. This can be achieved by
adjusting the concentration of the coding blocks used in the
ligation, according to the inverse of their relative coupling
efficiency. The relative coupling efficiency of 10 coding blocks
were determined using benzoic acid as the standard reference
(Table 1). Equal equivalents of a coding block and benzoic acid
were coupled to the linker on TentaGel. After cleavage and MS
analysis, the relative coupling efficiency of the coding block is
expressed as the signal intensity ratio of the coding block to
the reference benzoic acid.
A model compound (7, Scheme 4) from a published small
molecule library using 4-fluoro-3-nitrobenzoic acid scaffold was
synthesized and encoded based on our selection of coding
blocks.22 The synthetic and encoding reactions are shown in
Scheme 4. The first diversity, that is, the amino acid, was
encoded by a precoupled acid on the Trt-Gly-OH (8, Scheme
4). N-phthaloylglycine and 4-nitrophenylacetic acid were used
to encode the ortho-nitro fluoride and the nitro group on the
scaffold, respectively. The resin beads containing a presynthe-
sized linker in the inner core were divided into two portions.
One portion of the resin was treated with TFA to remove an
outside Boc protecting group, followed by coupling with Boc-
Met-OH. As a result, both the outer and inner layers of these
Figure 2. MALDI-FTMS spectra of single-bead analysis for the resin
containing library compound 7 and coding tags 8-10: (a) outside-
noncleavable bead; (b) outside-cleavable bead.
beads contained a cleavable linker. In contrast, the remaining
portion of the resin contained a cleavable linker in the inner
core only. The inside Fmoc group of both portions of resin was
then removed. A mixture of Trt-Gly-OH, N-phthaloylglycine,
and 4-nitrophenylacetic acid, whose concentrations have been
adjusted according to their relative reactivity (Table 1), was
coupled to the interior of the resin. The Trt (trityl) protecting
group on Trt-Gly-OH was removed using 1% TFA (trifluoro-
acetic acid), and benzoic acid was coupled to glycine to encode
the outside phenylalanine. After Boc deprotection and Boc-Phe-
OH coupling, the scaffold was then coupled to outside pheny-
lalanine. Propylamine reacted with inside N-phthaloylglycine
to form a stable amide bond (9) when replacing the fluoride on
the scaffold. Both of the outside and inside nitro groups were
reduced with tin(II) chloride, followed by acylation with
chloroacetic anhydride. The tetrahydroquinoxalin ring was then
formed by treating the resin with a base. Finally, substitution
of the remaining chloride with piperidine generated library
compound 7 and coding tag 10. A single bead from both portions
of resin was then treated with cyanogen bromide and analyzed
with MALDI-FTMS (Figure 2).
The results from outside-noncleavable (Figure 2a) and
outside-cleavable (Figure 2b) beads are consistent. The signals
of three coding tags are clearly seen in both cases, while that
of the library compound 7 appears in Figure 2b. The molecular
mass 590.30 corresponds to the protonated product of library
compound 7. It should be noted that the signal of the library
compound is much weaker than those of coding tags, even
though its concentration is much higher. This means the library
compound is not as sensitive as the coding tags to the MALDI-
FTMS method.
(23) Dankwardt, S. M.; Newman, S. R.; Krstenansky, J. L. Tetrahedron Lett.
1995, 36, 4923-4926.
(24) Yuan, C.; Williams, R. M. J. Am. Chem. Soc. 1997, 119, 11777-11784.
Synthesis and Screening of a Model Encoded Library. To
demonstrate the utility and efficiency of our encoding method,
9
6184 J. AM. CHEM. SOC. VOL. 125, NO. 20, 2003

MS Encoding Method for Small Molecule Libraries
A R T I C L E S
Scheme 5. Synthetic and Encoding Reactions of a Model Small
Molecule Librarya
Figure 3. MALDI-FTMS spectrum of single-bead analysis for the resin
containing library compound 11 and coding tags 12-14. As in Figure 2b,
this model compound was synthesized on beads with a cleavable linker on
both the outer layer and inner core.
reacted with the aryl amino groups on both the scaffold and
the coding block. The Boc and acid-labile side-chain protecting
groups of amino acids were removed by treatment with TFA
after library synthesis. A randomly selected model compound
(Scheme 5, 11) from this library was synthesized and encoded
prior to the library synthesis. The model compound was linked
to the solid support via methionine to make it releasable by
CNBr. The decoding result is shown in Figure 3. The obtained
molecular masses of library compound 11 and three coding tags
are consistent with the calculated values, indicating the structures
of library compounds can be reliably decoded.
In the library synthesis, we selected 42 secondary amines,
42 carboxylic acids or Boc-protected amino acids, and 48
carboxylic acids (anhydrides, acyl chlorides, or sulfonyl chlo-
rides) as the first (BB1), second (BB2), and third (BB3) building
blocks, respectively. The molecular weights of coding tags were
calculated prior to library synthesis to avoid any ambiguity in
the final decoding. After the library synthesis was completed,
50 resin beads were randomly picked for single-bead analysis.
MALDI-FTMS was used to unambiguously decode 46 of them.
No signals were obtained from four samples, probably due to
the loss of the beads during sample transfer.
This 84 672-member library (42 × 42 × 48) was screened
against streptavidin at an extremely dilute streptavidin-alkaline
phosphatase conjugate concentration (1:100 000 or 50 pM) using
an enzyme-linked colorimetric assay.9,25 Positive beads (20)
were isolated, and 18 of them were successfully analyzed with
MALDI-FTMS. Of these 18 beads, 2 yielded identical decoding
results (Table 2, entry 1). A typical decoding spectrum is shown
in Figure 4. Most of the streptavidin ligands shown in Table 2
share structural similarities. At the position of the first building
block, morpholine was found 7 times, and thiomorpholine was
found 3 times. The second building block was quite variable,
but a neutral amino acid appeared to be preferable. At the
position of the third building block, 4-pyridinecarboxylic acid
was seen frequently (8 times). 4-Pyridinecarboxylic acid has
been found to be important for streptavidin binding in our
previous work.9 Its structural analogues, 3-pyridinepropanoic
acid and pyrazinecarboxylic acid, were also observed in some
of the ligands. The streptavidin binding affinity of these ligands
was confirmed by resynthesis of these compounds on regular
TentaGel resin and rescreening against streptavidin using the
same method. Two of these compound beads (Scheme 6,
a Reagents and conditions: (i) 20% piperidine in DMF, rt, 30 min; (ii)
a mixture of 4-(chloromethyl)benzoic acid (1.18 equiv), N-Fmoc-3-
piperidinecarboxylic acid (2.45 equiv), 4-nitrophenylacetic acid (2.37 equiv),
HOBt (6 equiv), and DIC (6 equiv) in DMF, rt, 2 h; (iii) 50% TFA in
DCM, rt, 30 min; (iv) 5 equiv of (R,S)-N-Fmoc-â-amino-5-fluoro-2-
nitrobenzenepropanoic acid, HOBt and DIC in DMF, rt, 5 h; (v) 1 M
secondary amine in 5% DBU/DMF, rt, overnight; (vi) 10 equiv of carboxylic
acid, HOBt and DIC in DMF, rt, 4 h; (vii) 2 M SnCl2‚2H2O in DMF, 3 h
× 2; (viii) 30 equiv of carboxylic acid, 30 equiv of DIC, and 6 equiv of
DIEA in DCM, rt, 12 h × 2.
a simple encoded small molecule library was synthesized and
screened against streptavidin using (R,S)-N-Fmoc-â-amino-5-
fluoro-2-nitrobenzenepropanoic acid as the scaffold (Scheme
5). 4-(Chloromethyl)benzoic acid, N-Fmoc-3-piperidinecarboxy-
lic acid, and 4-nitrophenylacetic acid were chosen to encode
the para-nitro fluoride, Fmoc-protected amino group, and nitro
group on the scaffold, respectively. At the beginning of library
synthesis, a secondary amine was used as the first building block
to replace both the para-nitro fluoride on the scaffold and the
chloride of the coding block 4-(chloromethyl)benzoic acid. The
Fmoc protecting group was removed in this step simultaneously.
A carboxylic acid or a Boc-protected amino acid was then
coupled to the amino group on the scaffold as well as the coding
block 3-piperidinecarboxylic acid, followed by reduction of nitro
groups with tin(II) chloride. In the last synthetic step, a
carboxylic acid (anhydride, acyl chloride, or sulfonyl chloride)
(25) Lam, K. S.; Lebl, M. ImmunoMethods 1992, 1, 11-15.
9
J. AM. CHEM. SOC. VOL. 125, NO. 20, 2003 6185

A R T I C L E S
Song et al.
Table 2. Decoding Results of Positive Beads Obtained from Streptavidin Binding Assay
1
2
3
4
entry
BB1 (R and R )
BB2 (R )
BB3 (R )
1a
2
3
4
5
6
7
8
9
morpholine
morpholine
morpholine
morpholine
morpholine
morpholine
thiomorpholine
thiomorpholine
thiomorpholine
ethyl 1-piperazinecarboxylate
Ala
Ala
Gly
Gly
Ser
Gly
acetic acid
Ala
Ala
(S)-R-amino-
4-pyridinecarboxylic acid
3-benzoyl-2-pyridinecarboxylic acid
pyrazinecarboxylic acid
4-oxo-benzenebutanoic acid
4-oxo-benzenebutanoic acid
3,5-dimethoxybenzoic acid
4-pyridinecarboxylic acid
4-pyridinecarboxylic acid
[1,1′-biphenyl]-4-carboxylic acid
4-pyridinecarboxylic acid
10
cyclohexaneacetic acid
11
12
13
14
15
16
17
ethyl 1-piperazinecarboxylate
ethyl 1-piperazinecarboxylate
1-(2-pyridyl)piperazine
1-(2-cyanophenyl)piperazine
decahydroquinoline
Leu
Ser
Gly
Leu
Gly
acetic acid
Ser
4-pyridinecarboxylic acid
3-pyridinepropanoic acid
4-pyridinecarboxylic acid
4-pyridinecarboxylic acid
4-pyridinecarboxylic acid
4-oxo-benzenebutanoic acid
3-pyridinepropanoic acid
1-(2-furanylcarbonyl)piperazine
1,4-dioxa-8-azaspiro[4.5]decane
a The structure was found twice.
segregation of coding tags from the library compound avoids
interference by the coding tag during biological screening. To
save time, the linkers and some of the coding blocks can be
coupled to a large batch of resin in advance, and only a portion
(e.g., 10 mL of 90 µm settled beads, which is equivalent to 7.5
million beads) is needed for the synthesis of each library.
Because the molecular masses of coding tags can be accurately
determined in a single analysis step, no combination with other
techniques such as HPLC or MS/MS is needed for decoding.
We can now manually analyze 20-30 beads in an hour. We
envision that, with automation, over 500-1000 beads can be
analyzed in a single day. The sensitivity of this encoding method
is extremely high. We have been able to decode as small as a
10% fragment of a single 90 µm TentaGel bead with no
ambiguity. With this encoding approach, we may decrease the
size of the beads to 30 µm and still obtain more than adequate
mass signals for accurate decoding. However, further decreases
in bead size to less than 30 µm will make the library synthesis
and screening technically more difficult to handle. Work is
currently underway in our laboratory to fully automate the
sample preparation and the laser desorption ionization steps.
We are also developing software to facilitate both the selection
of building blocks and the decoding process, as well as archiving
the massive amounts of data that we expect to obtain.
Figure 4. MALDI-FTMS decoding spectrum of a positive bead from the
library screening against streptavidin.
Scheme 6. Chemical Structures of Two Streptavidin Binding
Ligands
compounds 15 and 16) stained very dark suggesting higher
Experimental Section
binding affinities than those of the others.
Material and General Methods. TentaGel S NH2 resin (90 µm,
0.26 mmol/g) was purchased from Rapp Polymere GmbH (Tu¨bingen,
Germany). All the calculations for synthesis were based on a substitution
of 0.26 mmol/g. (R,S)-N-Fmoc-â-amino-5-fluoro-2-nitrobenzenepro-
panoic acid was purchased from InnovaChem (Tucson, AZ). Alloc-
Osu (N-(allyloxycarbonyloxy)succinimide) was purchased from Senn
Chemicals (Dielsdorf, Switzerland). NR-Fmoc-protected amino acids
and HOBt (1-hydroxybenzotriazole) were purchased from GL Biochem
(Shanghai, China). NR-Boc-protected amino acids were purchased from
Chem-Impex International (Wood Dale, IL) and Advanced ChemTech
(Louisville, KY). Streptavidin-alkaline phosphatase conjugate was
purchased from Upstate Biotechnology (Lake Placid, NY). All solvents
and other chemical reagents were purchased from Aldrich (Milwaukee,
Conclusion
The encoding method described in this report is simple, fast,
and robust and can easily encode small molecule libraries with
over a million unique compounds. The extra encoding steps
necessary for many other chemical encoding techniques are
eliminated in this method, because the library synthesis steps
and the encoding steps are combined. One additional but very
important advantage of combining compound synthesis and
encoding steps is that the decoding information enables us to
estimate the quality of library compounds on an individual bead.
Since the coupling chemistry on the scaffold and the coding
block is either identical or very similar, the quality of a specific
coding tag on the decoding spectrum reflects the quality of the
corresponding coupling reaction on the scaffold. The topological
1
WI) and were analytical grade. H NMR and 13C NMR spectra were
recorded on a Bruker DRX 500 MHz spectrometer at 25 °C. A
commercial MALDI-FTMS instrument (IonSpec Corp., Irvine, CA),
equipped with an external MALDI source, a 4.7 T superconducting
9
6186 J. AM. CHEM. SOC. VOL. 125, NO. 20, 2003

MS Encoding Method for Small Molecule Libraries
A R T I C L E S
magnet, and an Nd:YAG laser operating at 355 nm, was used for
MALDI-FTMS analysis. All the experiments were carried out at room
temperature unless otherwise noted. For Fmoc deprotection, the resin
was treated with 20% piperidine/N,N-dimethylformamide (DMF) for
30 min and then washed thoroughly with DMF, methanol (MeOH),
and DMF 3 times each. For Boc deprotection, the resin was incubated
with 50% TFA/dichloromethane (DCM) for 30 min and then washed
thoroughly with DCM (twice), 2.5% N,N-diisopropylethylamine (DIEA)/
DCM (3 times), DCM (twice), MeOH (3 times), and DMF (3 times).
Synthesis of N-(9-Fluorenylmethoxycarbonyl)-4-{2-[2-(2-amino-
ethoxyl)ethoxy]ethylamino}-4-oxo-butanoic Acid (N-Fmoc-2,2′-(eth-
ylenedioxy)bis(ethylamine) Monosuccinamide, Fmoc-Ebes-OH).26
2,2′-(ethylenedioxy)bis(ethylamine) (1.46 mL, 10 mmol) was dissolved
in 50 mL acetonitrile. A solution of succinic anhydride (1.0 g, 10 mmol)
in 25 mL of acetonitrile was added dropwise under vigorous magnetic
stirring over 1 h. The stirring was stopped after 3 h. After the waxy
product settled, the organic solvent was decanted and discarded. The
product was redissolved in 100 mL of 50% acetonitrile/water and chilled
in an ice bath for 30 min. A solution of N-(9-fluorenylmethoxycarbo-
nyloxy)succinimide (Fmoc-OSu, 4.4 g, 13 mmol) in 25 mL of
acetonitrile was added dropwise under vigorous magnetic stirring over
1 h. The pH of the reaction mixture was maintained at 8-9 by adding
DIEA throughout the reaction. The reaction was allowed to proceed
overnight at room temperature. The solvents were removed in vacuo.
The residue was dissolved in 100 mL of 5% aqueous NaHCO3 solution
and washed with ethyl acetate. The aqueous phase was then acidified
with 1 M HCl to pH 2 and extracted with ethyl acetate (50 mL × 3).
The combined organic phase was washed with water and dried over
anhydrous MgSO4. The solution was concentrated to a small volume
and diluted with hexanes. A white solid was obtained with a yield of
72.6%. Its purity was determined to be 98% by HPLC analysis based
mL of DCM was added to the resin followed by a solution of Pd-
(PPh3)4 (75.1 mg, 0.065 mmol) in 12 mL of DCM.28 The mixture was
shaken in an argon atmosphere for 30 min. This process was repeated
once. The resin was washed with DCM, DMF, and DCM 3 times each.
A solution of di-tert-butyl dicarbonate (1.19 mL, 5.2 mmol) in 10 mL
of DCM was added to the resin, followed by the addition of DIEA
(226.4 µL, 1.3 mmol). The mixture was shaken until the ninhydrin
test was negative. The obtained outside-Boc-inside-Fmoc-linker-
bifunctional resin was washed with DCM, DMF, DCM and MeOH 3
times each and then dried in vacuo. The percentage of inner region
was determined to be 39% using quantitative UV absorption analysis
of the dibenzofulvene-piperidine adduct released by treatment with
piperidine.29
Determination of the Relative Reactivity of the Coding Blocks.
Fmoc-linker resin (20 mg, 0.0052 mmol) was swollen in DMF
overnight, followed by Fmoc deprotection. A mixture of the coding
block (0.0156 mmol), benzoic acid (1.91 mg, 0.0156 mmol), HOBt
(4.22 mg, 0.0312 mmol), DIC (4.9 µL, 0.0312 mmol), and 0.4 mL of
DMF was agitated for 30 min and then added to the resin. The reaction
mixture was agitated until the ninhydrin test was negative. The resin
was washed with DMF, DCM, and MeOH thoroughly. Beads (50) were
randomly picked and divided into five groups for cleavage and MALDI-
FTMS analysis.
Synthesis of Compound 2-6. Outside-Boc-inside-Fmoc-linker-
bifunctional resin (20 mg, 0.0052 mmol) was swollen in DMF
overnight, followed by Fmoc deprotection. The resin was incubated
with a mixture of Trt-Gly-OH (3.30 mg, 0.0104 mmol), benzoic acid
(1.27 mg, 0.0104 mmol), N-Fmoc-3-piperidinecarboxylic acid (3.65
mg, 0.0104 mmol), 4-nitrophenylacetic acid (1.88 mg, 0.0104 mmol),
4-(chloromethyl)benzoic acid (1.77 mg, 0.0104 mmol), HOBt (5.62
mg, 0.0416 mmol), and DIC (6.5 µL, 0.0416 mmol) in 0.4 mL of DMF
until the ninhydrin test was negative. The resin was washed with DMF
(0.5 mL × 3), MeOH (0.5 mL × 3), and DCM (0.5 mL × 3) and then
treated with 1% TFA and 5% triisopropylsilane (TIS) in DCM (0.5
mL and 2 min, 4 times). After the resin was washed with DCM (0.5
mL × 3), MeOH (0.5 mL × 3) and 5% DIEA/DMF (0.5 mL × 3), it
was agitated with 0.25 M piperidine in 5% DIEA/DMF overnight. To
the resin thoroughly washed with DMF, MeOH, and DMF was added
0.5 mL of 2 M SnCl2‚2H2O in DMF. The mixture was shaken for 6 h,
followed by washing with DMF (0.5 mL × 3), DCM (0.5 mL × 3),
MeOH (0.5 mL × 3), and DCM (0.5 mL × 3). All the amino groups
were then blocked by incubation with acetic anhydride (9.8 µL, 0.104
mmol) and DIEA (3.6 µL, 0.0208 mmol) in 0.5 mL of DCM overnight.
The resin was washed with DMF, DCM, and MeOH thoroughly.
Synthesis of Model Compound 7 and Tags 8-10. Outside-Boc-
inside-Fmoc-linker-bifunctional resin (20 mg, 0.0052 mmol) was
swollen in DCM overnight, followed by Boc deprotection. A solution
of Boc-Met-OH (3.89 mg, 0.0156 mmol), HOBt (2.11 mg, 0.0156
mmol), and DIC (2.5 µL, 0.016 mmol) in DMF was added to the resin.
The mixture was agitated until the ninhydrin test was negative. The
resin was washed 5 times with DMF.
1
on the absorption at 254 nm. Mp 111-113 °C; H NMR (500 MHz,
DMSO-d6) δ/ppm 7.91 (s, 1H), 7.89 (d, J ) 7.5 Hz, 2H), 7.69 (d, J )
7.5 Hz, 2H), 7.41 (t, J ) 7.5 Hz, 2H), 7.33 (m, 2H), 6.28 (s, 1H), 4.29
(d, J ) 6.8 Hz, 2H), 4.21 (t, J ) 6.8 Hz, 1H), 3.5-3.6 (m, 4H), 3.3-
3.4 (m, 4H), 3.1-3.2 (m, 4H), 2.40 (t, J ) 7.0 Hz, 2H), 2.30 (t, J )
7.0 Hz, 2H); 13C NMR (125 MHz, DMSO-d6) δ/ppm 174.6, 171.9,
156.9, 144.6, 141.5, 128.3, 128.0, 125.9, 120.8, 70.2, 69.8, 66.0, 47.4,
40.8, 39.3, 30.8, 30.1; MALDI-FTMS (M + Na+) m/z calcd, 493.195;
found, 493.198.
Solid-Phase Synthesis of the Cleavable Linker 1. The sequence
Phe-Phe-Met was synthesized on TentaGel S NH2 resin using standard
Fmoc chemistry with 1,3-diisopropylcarbodiimide (DIC) and HOBt as
the activating system.19 A ninhydrin test was carried out to monitor
amino acid coupling and Fmoc deprotection.27 After Fmoc deprotection
of the last Phe, a mixture of Fmoc-Ebes-OH (3 equiv), HOBt (3 equiv),
and DIC (3 equiv) in DMF was added to the resin. The reaction mixture
was gently shaken overnight until the ninhydrin test was negative. The
obtained Fmoc-linker resin was washed with DMF, DCM, MeOH 3
times each and then dried in vacuo.
Preparation of Topologically Segregated Bifunctional TentaGel
Resin Beads with 60% Boc Outside and 40% Fmoc-Linker 1 Inside
(Outside-Boc-Inside-Fmoc-Linker-Bifunctional Resin). TentaGel S
NH2 resin beads (1.0 g, 0.26 mmol) were swollen in water for 48 h.
The water was drained, and a solution of Alloc-OSu (31.1 mg, 0.156
mmol) in a DCM/diethyl ether mixture (50 mL, v/v ) 55:45) was added
to the resin, followed by addition of DIEA (55 µL, 0.312 mmol). The
resulting mixture was shaken vigorously for 1 h. The resin was washed
3 times with DCM and 6 times with DMF. Fmoc-Linker 1 was then
built in the inner region of the resin beads using the previously
mentioned procedure. The resin was washed 3 times with DCM. In
the presence of argon, a solution of PhSiH3 (770 µL, 6.24 mmol) in 4
Outside-Boc-inside-Fmoc-linker-bifunctional resin (either with or
without outside methione, 20 mg, 0.0052 mmol) was swollen in DMF
overnight, followed by Fmoc deprotection. A mixture of Trt-Gly-OH
(3.32 mg, 0.0105 mmol), N-phthaloylglycine (0.36 mg, 0.001 77 mmol),
4-nitrophenylacetic acid (3.25 mg, 0.0179 mmol), HOBt (4.22 mg,
0.0312 mmol), DIC (4.9 µL, 0.0312 mmol), and 0.4 mL DMF was
agitated for 30 min and was added to the resin. The reaction mixture
was agitated until the ninhydrin test was negative. The resin was washed
with DMF (0.5 mL × 3), MeOH (0.5 mL × 3), and DCM (0.5 mL ×
3) and then treated with 1% TFA and 5% TIS in DCM (0.5 mL and 2
(28) Grieco, P.; Gitu, P. M.; Hruby, V. J. J. Pept. Res. 2001, 57, 250-256.
(29) Bennett, W. D., Christensen, J. W., Hamaker, L. K., Peterson, M. L.,
Rhodes, M. R., Saneii, H. H., Eds. In AdVanced ChemTech Handbook of
Combinatorial and Solid-Phase Organic Chemistry? A Guide to Principles,
Products and Protocols; Advanced ChemTech Inc.: Louisville, KY, 1998;
p 330.
(26) Zhao, Z. G.; Im, J. S.; Lam, K. S.; Lake, D. F. Bioconjugate Chem. 1999,
10, 424-430.
(27) Kaiser, E.; Colescott, R. L.; Bossinger, C. D.; Cook, P. I. Anal. Biochem.
1970, 34, 595-598.
9
J. AM. CHEM. SOC. VOL. 125, NO. 20, 2003 6187

A R T I C L E S
Song et al.
min, 4 times). After the resin was washed with DCM (0.5 mL × 2),
2.5% DIEA/DCM (0.5 mL × 2), DCM (0.5 mL × 2), MeOH (0.5 mL
× 3), and DMF (0.5 mL × 3), it was agitated with a mixture of benzoic
acid (3.18 mg, 0.026 mmol), HOBt (3.52 mg, 0.026 mmol), DIC (4.1
µL, 0.026 mmol), and 0.4 mL of DMF for 2 h, followed by Boc
deprotection. A solution of 4-fluoro-3-nitrobenzoic acid (0.026 mmol),
HOBt (3.52 mg, 0.026 mmol), and DIC (4.1 µL, 0.026 mmol) in 0.4
mL of DMF was added to the resin. After the ninhydrin test was
negative, the resin was washed thoroughly with DMF, MeOH, and DMF
3 times each. A 0.5 mL aliquot of 0.25 M propylamine in 5% DIEA/
DMF was added to the resin, and the resulting mixture was agitated
overnight. The resin was washed extensively with DMF and then
agitated with 0.5 mL of 2 M SnCl2‚2H2O in DMF for 3 h. The reduction
was repeated. The resin was washed thoroughly with DMF, MeOH,
DMF, and DCM and then agitated with a solution of chloroacetic
anhydride (17.8 mg, 0.104 mmol) and DIEA (3.7 µL, 0.021 mmol) in
0.4 mL of DCM overnight. The solution was drained. After the resin
was washed with DCM (0.5 mL × 3), MeOH (0.5 mL × 3), and DMF
(0.5 mL × 3), it was treated with 10% DIEA/DMF for 6 h. The mixture
was filtered, and a solution of 10% piperidine and 10% DIEA in DMF
was added to the resin. The resulting mixture was shaken overnight.
The resin was washed with DMF, DCM, and MeOH 3 times each.
Synthesis of the Model Encoded Library. Outside-Boc-inside-
Fmoc-linker-bifunctional resin (0.5 g, 0.13 mmol) was swollen in DMF
overnight, followed by Fmoc deprotection. A mixture of 4-(chlorom-
ethyl)benzoic acid (26.17 mg, 0.1534 mmol), N-Fmoc-3-piperidinecar-
boxylic acid (113.92 mg, 0.3185 mmol), 4-nitrophenylacetic acid (55.81
mg, 0.3081 mmol), HOBt (105.46 mg, 0.78 mmol), DIC (122.1 µL,
0.78 mmol), and 10 mL of DMF was agitated for 30 min and then
added to the resin. The resulting mixture was agitated until the ninhydrin
test was negative. The resin was washed with DMF, MeOH, and DCM
3 times each, followed by Boc deprotection. A solution of (R,S)-N-
Fmoc-â-amino-5-fluoro-2-nitrobenzenepropanoic acid (272.66 mg, 0.65
mmol), HOBt (87.88 mg, 0.65 mmol), and DIC (101.8 µL, 0.65 mmol)
in 8 mL of DMF was added to the resin. The reaction mixture was
shaken until the ninhydrin test was negative. The resin was filtered
and washed with DMF, MeOH, and DMF 3 times each. The resin was
then split into 42 aliquots; to each of those a solution of one of 42
secondary amines in 5% 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU)/
DMF was added. The reaction was allowed to proceed overnight. The
resin beads were combined and washed with DMF, MeOH, and DMF
3 times each. The resin was split into 42 equal portions again. Each
one of the 42 carboxylic acids and NR-Boc-protected amino acids (0.031
mmol) was dissolved in a solution of HOBt (4.19 mg, 0.031 mmol) in
0.5 mL of DMF followed by addition of DIC (4.9 µL, 0.031 mmol).
The solutions were added to the 42 portions of resin individually. The
reaction mixtures were shaken for 4 h. The resin beads were combined
and washed with DMF, MeOH, and DMF 3 times each, followed by
incubation with 10 mL of 2 M SnCl2‚2H2O in DMF for 3 h. The
reduction was repeated. After washing thoroughly with DMF, DCM,
MeOH, and DCM 3 times each, the resin was split into 48 aliquots.
To each aliquot of resin, a solution of 1 of 40 carboxylic acids (0.081
mmol) and DIC (12.7 µL, 0.081 mmol) in 0.7 mL of DCM (or a
solution of 1 of 8 anhydrides, acyl chlorides, and sulfonyl chlorides
(0.041 mmol) in 0.7 mL of DCM) was added followed by addition of
DIEA (2.8 µL, 0.016 mmol). The reaction was carried out for 12 h
and repeated. Complete acylation was confirmed by a negative chloranil
test.30 The resin beads were combined, washed with DCM, DMF, DCM,
and MeOH 3 times each, and then dried in vacuo. The bead-supported
library was treated with a cleavage mixture consisting of TFA, TIS,
and water (v/v/v ) 95:2.5:2.5) for 2 h to remove the Boc and side-
chain protecting groups of amino acids. After the bead-supported library
was extensively washed with DCM, DMF, DCM, and MeOH, it was
stored in MeOH at 4 °C.
Library Screening. A 0.2 mL aliquot of the bead-supported library
was transferred into a 10 mL disposable polypropylene column with a
polyethylene frit. The beads were washed with water (4 mL × 10) and
then agitated with 4 mL of PBSTGNaN3 buffer (8.0 mM Na2HPO4,
1.5 mM KH2PO4, 137 mM NaCl, 2.7 mM KCl, pH 7.4, plus 0.1%
Tween-20, 0.1% gelatin, and 0.05% sodium azide) for 1 h. To the beads
washed with PBSTGNaN3 buffer (4 mL × 5) streptavidin-alkaline
phosphatase conjugate solution (2 mL) was added at a dilution of
1:100 000 (original concentration was 1 mg/mL) in PBSTGNaN3 buffer.
The mixture was incubated for 1 h. The beads were filtered and washed
with TBS buffer (2.5 mM Tris-HCl, 13.7 mM NaCl, 0.27 mM KCl,
pH 8.0, 4 mL), followed by washing with BCIP buffer (0.1 M Tris-
HCl, 0.1 M NaCl, 2.34 mM MgCl2, pH 8.8, 4 mL × 2). A solution of
the alkaline phosphatase substrate 5-bromo-4-chloro-3-indoyl phosphate
(BCIP) in BCIP buffer (1.65 mg/mL, 2 mL) was added to the resin to
develop color for 1 h. The enzymatic reaction was stopped by washing
with PBSTGNaN3 buffer (4 mL × 5) and water (4 mL × 5). The blue-
colored beads were retrieved and treated with 6.0 M guanidine-HCl
(pH 1.0) solution to strip the protein off the beads and then were washed
twice with water. The indigo dye was removed by incubating the beads
with acetone for 15 min. After the beads are washed with ethanol twice,
they are ready for cleavage.
Cleavage and MALDI-FTMS Analysis of a Single Bead. A single
resin bead was transferred to a 200 µL polypropylene microcentrifuge
tube in ethanol under a microscope. The solvent was evaporated in
vacuo, and 10 µL of 0.25 M CNBr in 70% formic acid was added.
The mixture was gently shaken overnight in the dark. The cleavage
was stopped by freezing and lyophilizing to dryness. The residue was
redissolved in 10 µL of 50% acetonitrile/water. The sample was applied
on the probe with 2 µL aliquots. Sodium dopant (0.01 M NaCl in 50%
ethanol/water, 1 µL) was added to the probe tip followed by matrix
solution (0.4 M of 2,5-dihydroxybenzoic acid in ethanol, 1 µL). Hot
air was used to quickly crystallize the sample on the probe.
Acknowledgment. This work was supported by NIH R33CA-
86364. The 500 MHz NMR spectrometer was purchased in part
with a grant from the National Sciences Foundation, NSF
9724412.
Supporting Information Available: Structures of building
blocks used in library synthesis and molecular weights of the
corresponding coding tags. This material is available free of
charge via the Internet at http://pubs.acs.org.
JA034539J
(30) Vojkovsky, T. Pept. Res. 1995, 8, 236-237.
9
6188 J. AM. CHEM. SOC. VOL. 125, NO. 20, 2003