Isolation of high-affinity trypsin inhibitors from a DNA-encoded chemical library

Article abstract of DOI:10.1002/anie.200700654

(Chemical Equation Presented) Maturing is easy to do: Annealing benzamidine-oligonucleotide conjugates with a library of DNA-encoded compounds allows the affinity capture of pharmacophores that are capable of binding to exosites adjacent to the primary substrate-binding pocket of the serine protease trypsin. Selected conjugates show an improvement in IC₅₀ values of several orders of magnitude compared with the starting benzamidine.

Full text of DOI:10.1002/anie.200700654

Angewandte
Chemie
DOI: 10.1002/anie.200700654
DNA-Encoded Library
Isolation of High-Affinity Trypsin Inhibitors from a DNA-Encoded
Chemical Library**
Samu Melkko, Yixin Zhang, Christoph E. Dumelin, Jörg Scheuermann, and Dario Neri*
The isolation of small organic binding molecules specific to
proteins of interest is a central problem in chemistry, biology,
and medicine. The use of DNA fragments as “barcodes” for
the identification of chemical compounds in a library[1]
represents an attractive option for the synthesis and screening
of large combinatorial libraries. Herein, we describe the use
of a DNA-encoded chemical library for the isolation of a
family of potent trypsin inhibitors with IC50 values in the
nanomolar range. The best inhibitor N-(4-carbamimidoylben-
zyl)-2-(3-(3-(3-iodophenyl)thioureido)phenyl)acetamide dis-
played a remarkable selectivity among closely related serine
proteases, exhibiting a 40- and 6500-fold lower potency
towards thrombin and factor Xa, respectively, compared
with the inhibition of trypsin.
to perform similar affinity-capture procedures for small
molecular entities followed by PCR-based amplification and
identification of preferred binding specificities.[4] Such DNA-
encoded chemical libraries may be constructed as single-
pharmacophore libraries (in which suitable chemical moieties
are added in a stepwise fashion to molecular scaffolds[5–8]) or
as dual-pharmacophore chemical libraries (also termed
encoded self-assembling chemical (ESAC) libraries[9,10]).
This second methodology is particularly suited for the
improvement of binding affinity and specificity, starting
from lead compounds of modest potency.
In this work, we used ESAC technology for the isolation
of potent inhibitors of bovine trypsin, starting from benzami-
dine, a trypsin inhibitor with an IC50 value in approximately
the 100 mm range (Figure 1a). Trypsin was chosen as a model
protease in view of the wealth of structural and functional
information available for this enzyme.[11,12] Furthermore,
several trypsin-like serine proteases represent targets of
considerable biological and pharmaceutical relevance.[13,14]
We conjugated 5-(4-carbamimidoylbenzylamino)-5-oxo-
pentanoic acid to the 3’ end of an amino-modified 24-mer
oligonucleotide (see Scheme 1 in the Supporting Informa-
tion). The resulting conjugate was allowed to anneal to the
cognate radiolabeled oligonucleotide and used in affinity-
capture assays on trypsin resins at different coating densities.
Figure 1b shows that the benzamidine–DNA conjugate was
well retained on affinity resins coated at 2.5 mgmLÀ1, whereas
retention efficiency decreased at lower coating densities,
which is similar to what we have previously reported for other
protein binders.[15,16] On the basis of these results, we decided
to pair the benzamidine–oligonucleotide conjugate to a
DNA-encoded library of 620 chemical compounds[15] (each
individually coupled to a distinctive oligonucleotide; Fig-
ure 1a) and to perform affinity-capture selections on trypsin
resins coated at 0.1 and 0.02 mgmLÀ1 density, which corre-
sponds to more stringent selection conditions.
Biochemical display technologies (such as phage display[2]
and ribosome display[3]) routinely allow the isolation of
polypeptides with high affinity to virtually any protein target
of interest, starting from judiciously designed, large libraries
of protein mutants. The success of these technologies is due
not only to plasticity of certain proteins (e.g., antibodies) for
the recognition of cognate antigens, but also to the fact that
libraries of large size can be subjected to affinity-capture
procedures, followed by the amplification and identification
of enriched library members by virtue of the physical linkage
between phenotype (i.e., the antigen binding property) and
genotype (i.e., the gene coding for the antibody).
In principle, large collections of chemical compounds,
individually coupled to unique oligonucleotides, may allow us
[*] Dr. Y. Zhang, C. E. Dumelin, Dr. J. Scheuermann, Prof. D. Neri
Department Chemie und Angewandte Biowissenschaften
ETH Zürich
Wolfgang-Pauli-Strasse 10, 8093 Zürich (Switzerland)
Fax: (+41)44-633-1358
E-mail: neri@pharma.ethz.ch
Dr. S. Melkko
Philochem AG
Figure 2a shows representative results of affinity selec-
tions performed with the 620 benzamidine derivatives, which
were obtained by PCR amplification of the code-containing
DNA moieties with a fluorescently labeled primer, followed
by hybridization to microarrays spotted in quadruplicate with
cognate oligonucleotides corresponding to the 620 different
codes. Analysis of the ratios of fluorescence signal intensities
for the library compounds after selection on trypsin and on a
resin without antigen (Figure 2b) revealed that compounds
73, 77, 79, 437, and 585 were preferentially enriched after
selections performed at both 0.1 and 0.02 mgmLÀ1 coating
densities (Figure 2c). Interestingly, three of the five most
promising compounds contained a phenylthiourea moiety,
which suggests the presence of an exosite compatible with
c/o ETH Zürich
Wolfgang-Pauli-Strasse 10, 8093 Zürich (Switzerland)
[**] We are grateful to Dr. Jens Sobek and Dr. Ralph Schlapbach
(Functional Genomics Center Zürich) for help with the spotting of
oligonucleotide chips, and we acknowledge financial support from
the ETH Zürich, the Bundesamt für Bildung und Wissenschaft for
the EU Project “STROMA”, the Swiss National Science Foundation,
Philogen SpA, and Philochem AG. The authors declare competing
financial interests. ESAC technology is covered by a patent
application, which was licensed from the ETH Zürich to Philogen
under a share of revenues agreement. D.N. owns shares of Philogen
and consults for this company. Since November 2006, S.M. is
employed by Philochem, a daughter company of Philogen.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2007, 46, 4671 –4674
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4671

Communications
Figure 1. a) Schematic representation of a 24-mer oligonucleotide (covalently conjugated to the weak trypsin inhibitor benzamidine) paired with a
library of 620 DNA-encoded chemical compounds, constituting a library of 620 DNA-encoded benzamidine derivatives. b) Benzamidine–
oligonucleotide conjugate and the corresponding negative-control oligonucleotide were paired with a cognate radiolabeled oligonucleotide and
subjected to an affinity-capture assay by using trypsin-coated sepharose beads of four different coating densities. The radioactivity of aliquots of
the input, the flow-through fractions of different washing steps, and of the remaining slurry were determined. Ve =elution volume.
affinity-capture radioactive
assays similar to the ones of
Figure 1b (data not shown).
Bidentate derivatives consist-
ing of benzamidine and of
compound 73 connected by
different linkers were synthe-
sized (see Scheme 2 in the
Supporting Information) and
tested in trypsin inhibition
assays with a fluorogenic sub-
strate (see Figure 1 in the
Supporting Information).
Scheme 1
shows
the
chemical structures and the
trypsin IC50 values for a series
of four parental benzamidine
derivatives (1–4), a derivate
of the selected compound 73
(5), six bidentate conjugates
of benzamidine and com-
pound 73 (6–8, 10–12), and,
as a negative control com-
pound, a bidentate conjugate
of benzamidine and com-
pound 322 (9). The most
active inhibitor 10 (N-(4-car-
bamimidoylbenzyl)-2-(3-(3-
(3-iodophenyl)thioureido)-
phenyl)acetamide) contained
a phenyl moiety in the linker
and exhibited an IC50 value of
98 nm. However, bidentate
ligands (e.g., 8) with flexible
Figure 2. a) Microarray decoding of ESAC affinity maturation selections against trypsin. The microarray
slides for selections with trypsin resin of two different coating densities and with negative-control resin are
shown. A section of the microarray is magnified, showing the preferential enrichment of the signal for
compound 73 when the selections were performed in the presence of trypsin. b) The graphs show ratios of
the average microarray signal intensities for individual compounds of selections to trypsin against the
negative-control selection. c) Structures of library compounds 73, 77, 79, 437, and 585. The signals of these
compounds were enriched in the selections to trypsin.
phenylthiourea binding adjacent to the S1 (and benzamidine)
binding pocket of trypsin. Compound 73 was chosen for
further characterization as it yielded the best results on
alkyl linkers also exhibited a dramatically improved affinity
(600 nm), compared with a set of parental benzamidine
derivatives (1–4), whose IC50 values were in the 11–220 mm
4672 www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 4671 –4674

Angewandte
Chemie
Scheme 1. Structures of chemical compounds that were tested for their potency to inhibit trypsin. The determination of the IC50 values was
performed as described in the Supporting Information. For compound 5, no inhibition was detected at a compound concentration of 250 mm.
*
range. Bidentate conjugates of benzamidine and library
compounds, which had not been enriched in the affinity
selections, displayed IC50 values comparable with the one of
benzamidine (e.g., 12 mm for 9, a conjugate of benzamidine
and library compound 322).
The IC50 values of compound 10 towards factor Xa and
thrombin, two trypsin-like serine proteases, were 640 mm and
4 mm, which corresponds to a 6500- and 40-fold selectivity,
respectively, in favor of trypsin (data not shown).
The results presented herein show that 620 DNA-encoded
benzamidine derivatives could be rapidly prepared and
screened for improved trypsin binders. In principle, the
same methodology can be applied to the maturation of any
lead compound that can be coupled to an oligonucleotide, for
virtually any protein target of interest, without the need to
resynthesize the chemical library. After selection decoding
and a modular synthesis of bidentate ligands, selective
inhibitors with nanomolar potency to trypsin could be rapidly
identified. In further studies, it will be interesting to elucidate
the structural basis for the improvement of binding affinity
conferred by the selected phenylthiourea moiety. As for
phage-display technology, a judicious choice of suitable
coating density on affinity-capture support is crucial to the
success of selection methodologies with DNA-encoded chem-
ical libraries. Such investigations are facilitated by the
possibility of performing model selections with radiolabeled
DNA derivatives (Figure 1b).
We used oligonucleotide-based microarray technology for
the decoding of the selection experiments (Figure 2a), how-
ever, in principle other methodologies such as high-through-
put sequencing can be considered.[1,17,18] The site-specific
immobilization of amino-tagged oligonucleotides on epoxy-
silane-modified glass slides resulted in highly reproducible
fluorescence signal intensities (Figure 2a), which are required
for efficient decoding procedures. The essentially flat profile
for the ratios of fluorescence signal intensities corresponding
to selections performed in the presence or absence of target
antigen (Figure 2b) constitutes additional evidence for the
robustness of the decoding procedure, facilitating the iden-
tification of bidentate binders, which were preferentially
enriched with trypsin resin.
In summary, the results discussed here provide evidence
that ESAC technology can be used for the rapid affinity
maturation of lead protein binders[9] and stimulates research
efforts for the development of improved synthetic and
decoding methods related to DNA-encoded chemical libra-
ries.
Received: February 12, 2007
Published online: May 11, 2007
Angew. Chem. Int. Ed. 2007, 46, 4671 –4674
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
4673

Communications
[9] S. Melkko, J. Scheuermann, C. E. Dumelin, D. Neri, Nat.
Biotechnol. 2004, 22, 568.
[10] K. I. Sprinz, D. M. Tagore, A. D. Hamilton, Bioorg. Med. Chem.
Lett. 2005, 15, 3908.
[11] J. J. Perona, C. S. Craik, J. Biol. Chem. 1997, 272, 29987.
[12] T. A. Steitz, R. G. Shulman, Annu. Rev. Biophys. Bioeng. 1982,
11, 419.
[13] A. Schweinitz et al., J. Biol. Chem. 2004, 279, 33613; see the
Supporting Information.
[14] B. Turk, Nat. Rev. Drug Discovery 2006, 5, 785.
[15] C. E. Dumelin, J. Scheuermann, S. Melkko, D. Neri, Bioconju-
gate Chem. 2006, 17, 366.
[16] S. Melkko, C. E. Dumelin, J. Scheuermann, D. Neri, Chem. Biol.
2006, 13, 225.
[17] S. Melkko, J. Sobek, G. Guarda, J. Scheuermann, C. E. Dumelin,
D. Neri, Chimia 2005, 59, 798.
[18] M. Margulies et al., Nature 2005, 437, 376; see the Supporting
Information.
Keywords: benzamidines · compound libraries · self-assembly ·
supramolecular chemistry · trypsin
.
[1] J. Scheuermann, C. E. Dumelin, S. Melkko, D. Neri, J. Biotech-
nol. 2006, 126, 568.
[2] G. Winter, A. D. Griffiths, R. E. Hawkins, H. R. Hoogenboom,
Annu. Rev. Immunol. 1994, 12, 433.
[3] C. Schaffitzel, J. Hanes, L. Jermutus, A. Pluckthun, J. Immunol.
Methods 1999, 231, 119.
[4] S. Brenner, R. A. Lerner, Proc. Natl. Acad. Sci. USA 1992, 89,
5381.
[5] Z. J. Gartner, B. N. Tse, R. Grubina, J. B. Doyon, T. M. Snyder,
D. R. Liu, Science 2004, 305, 1601.
[6] D. R. Halpin, P. B. Harbury, PLoS Biol. 2004, 2, e174.
[7] D. R. Halpin, P. B. Harbury, PLoS Biol. 2004, 2, e173.
[8] D. R. Halpin, J. A. Lee, S. J. Wrenn, P. B. Harbury, PLoS Biol.
2004, 2, e175.
4674 www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 4671 –4674