Powerful protein binders from designed polypeptides and small organic moleculesa-A general concept for protein recognition

Article abstract of DOI:10.1002/anie.201005059

High-affinity binders for the C-reactive protein (CRP), with dissociation constants in the pM to nM range and selectivities in human serum comparable to those of antibodies, were obtained by conjugation of 16 designed polypeptides to phosphocholine, a sma

Full text of DOI:10.1002/anie.201005059

DOI: 10.1002/anie.201005059
Small-Molecule Binders
Powerful Protein Binders from Designed Polypeptides and Small
Organic Molecules—A General Concept for Protein Recognition**
Lotta T. Tegler, Guillaume Nonglaton, Frank Bꢀttner, Karin Caldwell, Tony Christopeit,
U. Helena Danielson, Karin Fromell, Thomas Gossas, Anders Larsson, Paola Longati,
Thomas Norberg, Ramesh Ramapanicker, Johan Rydberg, and Lars Baltzer*
Drug development, in vitro and in vivo diagnostics, industrial
protein purification, innumerable bioanalytical research ap-
plications, and many other areas of biotechnology are
critically dependent on access to molecules that recognize
and bind proteins specifically and with high affinity in
complex biological media. While antibody technologies and
organic synthesis remain invaluable tools for biotechnology
and biomedicine, the quest continues for robust and efficient
binder molecules that expand chemical diversity and are less
costly and time consuming to develop. Currently, protein
binders are almost exclusively of a purely biological origin,
that is, they are antibodies, engineered proteins, and aptam-
ers, or purely synthetic, for example, small organic pharma-
ceuticals. Here we report on a new concept for protein
recognition based on a set of designed polypeptides con-
jugated to small organic molecules. The resulting hybrid
molecules bind proteins with specificities and affinities that
compare well to antibodies, while being by comparison easy
to prepare and more than an order of magnitude smaller.
At the core of the technology are polypeptides not
developed specifically for a given protein but selected from a
designed set of 42-residue sequences of general applicability.
An analogy may be found in nature, where a small subset of
amino acids at a protein–protein interface, the so called hot
spots, dominate the interaction between proteins, whereas the
remaining amino acid residues at the protein–protein inter-
face can be mutated extensively without greatly reducing the
affinity or specificity.[1–3] The small organic molecules, or “hot-
spot mimics”, need only to bind with micromolar dissociation
constants for the polypeptide conjugates to have KD values in
the nanomolar to picomolar range. The technology enables
chemists with access to small-molecule binders that would
normally be considered “failed”, “fragments”, or “at an early
stage of development” to prepare powerful binders with
relative ease and use them for target validation, biomarker
identification, protein purification, diagnostics, as pharma-
ceuticals, etc. The concept is demonstrated here for the C-
reactive protein (CRP), a protein of interest as a diagnostic
biomarker.[4] CRP plays a key role in inflammation and the so
called high-sensitivity CRP test is used as a cardiovascular
risk marker.[5]
[*] Dr. L. T. Tegler, Dr. G. Nonglaton,[+] Dr. F. Bꢀttner,[#] T. Christopeit,
Prof. U. H. Danielson, Dr. T. Gossas, Dr. P. Longati,[$]
Prof. T. Norberg, Prof. R. Ramapanicker,[++] Dr. J. Rydberg,
Prof. L. Baltzer
Department of Biochemistry and Organic Chemistry
Uppsala University
P.O. Box 576, 75123 Uppsala (Sweden)
Fax: (+46)18-471-3818
E-mail: lars.baltzer@biorg.uu.se
Homepage: http://www.biorg.uu.se
Prof. K. Caldwell, Dr. K. Fromell
A 16-membered set of 42-residue polypeptides was
designed to serve as scaffolds to which small molecules
could be attached to form binder candidates (Figure 1). The
polypeptides have no natural origin and were designed
de novo to have some propensity for folding into helix-loop-
helix motifs.[6] The underlying principle was that hydrophobic
interactions would provide the binding energy and charged
residues the selectivity in protein binding. Helices were
designed to be amphiphilic with charged residues introduced
in four different combinations to give total charges of À7, À4,
À1, and + 2, with the hydrophobic residues unaltered
throughout the set. The site of incorporation of the small-
molecule ligand was also varied and positioned at the early
part of helix I (position 8), the latter part of helix I (posi-
tion 17), the loop region (position 22), and the middle of
helix II (position 34). Only modest amounts of extra binding
energy are required for enhanced affinities on the order of 3–
4 orders of magnitude over that of the small molecule, as the
entropic cost for binding has been reduced as a consequence
of conjugation of the polypeptide to the small molecule.
Energies of 20–30 kJmolÀ1 correspond to hydrophobic inter-
Department of Physical and Analytical Chemistry
Uppsala University (Sweden)
Prof. A. Larsson
Department of Medical Sciences, Akademiska sjukhuset
75185 Uppsala (Sweden)
[+] Current address: Dꢁpartement micro-Technologie pour la Biologie
et la Santꢁ CEA-LETI, Minatec, Grenoble (France)
++] Current address: Department of Chemistry, Indian Institute of
[
Technology, Kanpur (India)
[$] Current address: Department of Clinical Science, Intervention and
Technology, Novum Centrum, Karolinska University Hospital,
Huddinge (Sweden)
[#] Current address: Leipziger Arzneimittelwerk, Leipzig (Germany)
[**] We are indebted to the European Union for financial support
through the IP CARE-MAN (contract no. NMP4-CT-2006-017333),
to the Knut and Alice Wallenberg Foundation, and to UppsalaBio
through its Bio-X project. We also thank the Swedish Research
Council for financial support through a grant to L.B. We are indebted
to Greta Hulting for technical assistance and Dr. Jennifer Neff of
Allvivo LLC for the generous gift of Pluronic-F108 PDF.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201005059.
Angew. Chem. Int. Ed. 2011, 50, 1823 –1827
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1823

Communications
spacer
to
form
PC6
(Figure 2). The strategy to
link PC through the trime-
thylammonium group was
based on the expectation
that important interactions
of the phosphate group
would be weakened if it was
chemically modified.[10] The
4-nitrophenyl ester in PC6
was chosen to enable incor-
poration of the spacer-modi-
fied phosphocholine into the
polypeptides by forming an
amide linkage at the side
chain of a lysine residue in a
one-step reaction in DMSO
solution. Only one free lysine
residue was accessible in each
sequence. For
a
detailed
description of the synthesis
of PC6 and the conjugation
reaction see the Supporting
Information.
PC6 was treated with each
member of the set of 16
polypeptides to form 16
binder candidates and each
one was evaluated with
regards to affinity and selec-
tivity. The affinity for CRP
was estimated in a screening
Figure 1. Polypeptide sequence library and graphical representation of the hit profile (red squares). The 16-
membered set of 42-residue sequences varies in terms of charge and the site of incorporation of PC-6. To
illustrate the variation pattern, Asp and Glu residues are color coded red, Arg residues are blue, linkage
sites (L) for PC-6 are yellow, and linkage sites for the chromophore (C) are violet. All N terminals are
acetylated.
actions involving, for example, 2–4 leucin side chains in the
polypeptide. A principle behind the design is that if a number
of residues are in position to bind, some of them are likely to
find productive binding sites, although it is not possible to
predict which ones. Charge–charge interactions are individ-
ually weak, but may contribute through cooperativity. Helices
are expected to present binding residues better than unor-
dered polypeptides for reasons of preorganization.
experiment based on fluorescence intensity using a microtiter
plate reader. A 500 nm solution of polypeptide conjugate in
50 mm tris(hydroxymethyl)aminomethane (Tris) buffer at
pH 7 containing 5 mm CaCl2 and 150 mm NaCl was titrated
in three steps with CRP. All polypeptides carried 7-methox-
ycoumarin chromophore probes and the intensity of the
fluorescence emission was found to be affected by binding to
CRP. Purified human CRP was added from a stock solution at
a concentration of 2.3 mgmLÀ1, to give total CRP concen-
trations of 500 nm, 1000 nm, and 1500 nm. Each measurement
was carried out in triplicate and compared to a negative
control, where the binder alone was titrated with buffer. The
purpose of titrating the binder in three steps was to determine
within experimental accuracy whether binding to the poly-
peptide conjugate was saturated at a 500 nm concentration of
CRP, in which case an apparent KD value of 10 nm or lower
could be estimated assuming an experimental error of 10% or
less (see the Supporting Information). Binders for which
changes in the fluorescence intensity at 410 nm were observed
after the addition of one equivalent of CRP, but no significant
further changes after two and three additions were considered
hits. There were 11 hits in the set of 16 binder candidates
(Figure 1), although the evaluation is somewhat subjective,
since the overall change in fluorescence intensity is often
small and difficult to evaluate. Polypeptides with a total
charge of + 2 or À1 gave the most hits, which is not surprising
in view of the fact that CRP is an acidic protein with a protein
Structurally, the polypeptides show the hallmarks of
1
molten globules. The H NMR spectrum of the sequence 4-
C10L17PC6 is characterized by broadened resonances and
poor shift dispersion. The CD signature shows a high content
of secondary structure, with a mean residue ellipticity Q222
=
À20000 degcm2 dmolÀ1 (see the Supporting Information), but
there is no structural evidence yet to show that the polypep-
tide adopts a helical conformation in a complex with CRP.
The polypeptide scaffolds are not preorganized to fit a unique
surface epitope of a protein, but adapt to the protein surface
in the vicinity of the small-molecule binding site. The binder
for CRP was designed based on the crystal structure of the
protein complexed to its natural ligand, phosphocholine (PC,
Figure 2).[7] PC binds CRP with an affinity of approximately
6 mm[8,9] in the presence of two Ca2+ ions, which intercalate
between the phosphate group of PC and Asp60, Asn61, Glu
138, Gln139, Asp140, Glu147, and Gln150 in the PC binding
site. A spacer was introduced by replacing one of the methyl
groups of the trimethylammonium group by a functionalized
1824
www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 1823 –1827

is very high, and higher than that of the
phosphocholine residue by probably
three to four orders of magnitude. The
affinity of the polypeptide 4-C15L8
without PC attached could not be
measured under the conditions
used—no change in the fluorescence
was observed upon addition of 2 mm
CRP to a 500 nm solution of the
scaffold—and one can only estimate
that it has a dissociation constant that
is higher than 100 mm. PC was shown
to inhibit the binding of 4-C15L8
coupled to PC6, thus demonstrating
that the PC6 residue of the conjugate
binder targets the PC binding site of
CRP as designed (data not shown).
The overall affinity is thus due to
cooperativity between the PC binding
group, the polypeptide, and possibly
also the aliphatic spacer. A suitable
mechanistic model could not be iden-
tified by fitting equations representing
a variety of interaction models to the
SPR data, and thus kinetic parameters
and dissociation constants could not be
determined. The binders were ranked
by calculating from the data set their
concentration that provides half satu-
ration (B50 values; see the Supporting
information). The seven highest rank-
ing binders identified by SPR were all
in the group of ten identified by
fluorescence and only one of the ten
Figure 2. The crystal structure of CRP in a complex with PC and Ca2+ ions shows that PC binds to
each subunit of CRP in the presence of the Ca2+ ions. Modification of PC by introduction of a
spacer to form PC-6, followed by reaction with a polypeptide gives the binder molecule. A
chromophore is routinely introduced to enable determinations of the concentration and for
fluorescence titrations, but probably contributes little or nothing to affinity. C denotes coumarin,
but dansyl (D) is also used.
isoelectric point (PI) of 5.3. It was not possible to evaluate
which amino acids were critical for binding. We synthesized
an N-terminal 20-mer and an N-terminal 32-mer of 4-
C10L17PC6, and analysis showed that the full-length
sequence binds better than the truncated ones, and could
not in this case have been replaced by shorter ones (see the
Supporting Information). Multivalency is not the reason for
the high affinity towards CRP, as demonstrated by surface
plasmon resonance (SPR) and fluorescence titrations (see the
Supporting Information).
All binder candidates were also characterized by SPR
biosensor analysis (Biacore) to obtain a more detailed insight
into their interactions with CRP, as described previously for
PC.[9] The protein was immobilized on the chip under
standard conditions and the interactions of each binder with
CRP was studied in the presence of 10 mm CaCl2, with the
binder concentrations varied in the range from 0.5 to 1200 nm
(Figure 3). The results from the SPR analysis were in broad
agreement with those from the fluorescence-based screening,
as the strongest interactions with CRP were found among the
binder molecules with total charges of + 2 and À1. All binders
interacted reversibly with CRP, and several tight binders were
identified. The dissociation of 4-C10L17PC6 was extremely
slow and essentially irreversible. The exceptional stability of
the complex shows that the affinity of 4-C10L17PC6 for CRP
highest ranked binders according to SPR was not identified
on the basis of fluorescence.
The selectivity of the tightest binder according to Biacore
ranking, 4-C10L17PC6, was investigated by attaching it to a
polystyrene (PS) nanoparticle through a disulfide bridge to
the Cys residue at position 24, and incubating it with patient
serum having a CRP concentration of 60 mgLÀ1 that was
obtained from the Uppsala University hospital. After incu-
bation of the nanoparticles with serum, the mixture was
centrifuged and the supernatant removed. The particles were
washed three times with buffer, treated with DTT to reduce
the disulfide bonds, and the resulting supernatant analyzed by
sodium dodecylsulfate polyacrylamide gel electrophoresis
(SDS-PAGE; Figure 4). As a control, an avian antibody (IgY)
directed against CRP was attached to nanoparticles and
treated the same way. Several bands were observed, most
likely because CRP binds not only the binder and the
antibody, but also several proteins in the complement
activation system, which are captured from the serum as
well. The critical observation is that both the 4-C10L17PC6
binder and the anti-CRP antibody extract the same pattern of
proteins, thereby giving rise to the same bands on the gel. The
synthetic binder is as selective as the antibody in human
serum.
Angew. Chem. Int. Ed. 2011, 50, 1823 –1827
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1825

Communications
Figure 3. Sensorgrams from SPR biosensor analysis of interactions between the 16 binder candidates and immobilized CRP. The binder 4-
C10L17PC6 dissociates extremely slowly. The concentrations of binders used were 1200, 800, 533, 355, 237, 158, 105, 70.2, 46.8, 31.2, 20.8, 13.8,
9.24, 6.61, 4.11, 2.74, 1.82, 1.21, 0.81, and 0.54 nm.
To further demonstrate the selectivity in biological fluids
we introduced tight binders in well-known bioanalytical test
formats (see the Supporting Information). Several of the
polypeptide conjugate binders provided excellent results in
measuring CRP in human serum in an enzyme-linked
immunosorbent assay (ELISA).
A nitrocellulose-based
assay showed that 4-C10L17PC6 compared well with the
anti-CRP antibody in a commercially available point of care
assay. A binder for CRP based on the polypeptide sequence 3-
D37L34 was recently shown to perform extremely well on
chip in a point of care assay.[11]
The small set of polypeptides required to develop
excellent binding performance contrasts sharply with the
huge libraries that need to be accessed to find specific high-
affinity antibodies, aptamers, or phage-display-generated
proteins. The specific binding of specific proteins is associated
with unique sequences in biomolecular recognition. The
introduction of a small organic residue with several functional
groups by chemical synthesis dramatically relaxes the need
for structural complexity in the polypeptide conjugates. The
molten globule-like character of the polypeptide conjugates
Figure 4. SDS-PAGE of extracts by 4-D10L17PC6 and an avian anti-
CRP antibody from human serum. Lanes from left to right: 1) anti-CRP
in human serum, 2) anti-CRP in CRP-free serum, 3) 4D10L17PC6 in
human serum, 4) 4D10L17 in CRP-free serum, 5) blank, PS beads in
human serum, 6) blank, PS beads in CRP-free serum, 7) HSA, 8) CRP,
9), 10) standard proteins (97, 66, 40, 30, 20.1, and 14.4 kDa),
11) prestained standard.
1826
www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 1823 –1827

contrast sharply with the high level of structural preorganiza-
tion of biomacromolecules.
synthetic binders for proteins that are robust and easy to
modify, and that furthermore can be generated from a
comparatively simple set of polypeptide scaffolds will have
a large impact on clinical diagnostics and drug development in
the near future.
High-resolution structural information is not yet avail-
able, but the competitive binding by PC demonstrates that the
PC binding site of CRP is targeted, and the variation in binder
affinities over many orders of magnitude is clear evidence
that the polypeptide sequence interacts with CRP. The lack of
structural definition in the molten globule-like conjugate Experimental Section
Polypeptides were synthesized by solid-phase peptide synthesis and
molecules suggests an “adapted fit” binding mechanism.
From the perspective of application, the binder concept offers
robust binders that do not denature and can be stored
lyophilized or in solution for years. They are chromato-
graphically pure, with little or no batch to batch variation. The
chemical nature and absence of a biological history of
designed binder molecules minimize the risk for undesirable
interactions with endogenous biomolecules that will interfere
with, for example, diagnostic tests and other applications.[12]
The set of polypeptides is not limited to forming binders for
CRP. We reported previously that the conjugation of the 42-
residue polypeptide KE2 to benzenesulfonamide, a known
inhibitor of human carbonic anhydrase II (HCAII) with a
KD value of 1.5 mm, afforded a hybrid molecule that binds
HCAII with a KD value of 20 nm in buffer.[13] A binder
molecule for HCAII, based on benzensulfonamide but
developed using the 16 polypeptides reported here, discrimi-
nated between HCAI and HCAII—two isoforms of human
carbonic anhydrase with 60% homology—by a factor of 30,
although the affinity for benzenesulfonamide is the same for
both proteins.[14] Several binders are under development using
the same set of polypeptides. A concept for enhancing affinity
was reported previously by King and Burgen, where 4-
substituted benzenesulfonamides were shown to bind more
strongly to human carbonic anhydrase II the longer the
aliphatic chain in the 4-position. The concept presented here
is very different. A wide range of affinities is obtained even
though the hydrophobic side chains are the same in all 16
polypeptides. Aliphatic side chains enhance affinity non-
specifically in a hydrophobic binding pocket, but the combi-
nation of hydrophobic substituents with charged amino acid
residues introduce enhanced affinity as well as selectivity.[15]
While the theory behind cooperativity in biomolecular
recognition is now well understood,[16] the quantification of
the level of cooperativity obtained for these binders requires
that the affinity of the polypeptide can be measured precisely
for the protein surface area in proximity to the small-
molecule binding site, which is unrealistic.
purified by HPLC. Affinities were determined by fluorescence
titration and surface plasmon resonance (Biacore). Selectivity was
evaluated by ELISA and by pull-down experiments followed by SDS-
PAGE. For full experimental detail see the Supporting Information.
Received: August 12, 2010
Revised: December 12, 2010
Published online: January 14, 2011
Keywords: bioorganic chemistry · C-reactive protein ·
.
phosphocholine · polypeptides · proteins
[1] W. L. DeLano, Curr. Opin. Struct. Biol. 2002, 12, 14 – 20.
[2] A. A. Kossiakoff, S. Koide, Curr. Opin. Struct. Biol. 2008, 18,
499 – 506.
[3] A. A. Bogan, K. S. Thorn, J. Mol. Biol. 1998, 280, 1 – 9.
[4] J. Westhuyzen, H. Healy, Ann. Clin. Lab. Sci. 2000, 30, 133 – 143.
[5] S. Verma, P. E. Szmitko, P. M. Ridker, Nat. Clin. Pract.
Cardiovasc. Med. 2005, 2, 29 – 36.
[6] W. F. DeGrado, C. M. Summa, V. Pavone, F. Nastri, A.
Lombardi, Annu. Rev. Biochem. 1999, 68, 779 – 819.
[7] D. Thompson, M. B. Pepys, S. P. Wood, Structure 1999, 7, 169 –
177. PDB ID: 1B09.
[8] J. K. Anderson, R. M. Stroud, J. E. Volanakis, Fed. Proc. 1978,
37, 1495 – 1499.
[9] T. Christopeit, T. Gossas, U. H. Danielson, Anal. Biochem. 2009,
391, 39 – 44.
[10] M. B. Pepys, G. M. Hirschfield, G. A. Tennant, J. R. Gallimore,
M. C. Kahan, V. Belotti, P. N. Hawkins, R. M. Myers, M. D.
Smith, A. Polara, A. J. Cobb, S. V. Ley, J. A. Aquilina, C. V.
Robinson, I. Sharif, G. A. Gray, C. A. Sabin, M. C. Jenvey, S. E.
Kolstoe, D. Thompson, S. P. Wood, Nature 2006, 440, 1217 – 1221.
[11] C. Albrecht, P. Fechner, D. Honcharenko, L. Baltzer, G.
Gauglitz, Biosens. Bioelectron. 2010, 25, 2302 – 2308.
[12] L. J. Kricka, Clin. Chem. 1999, 45, 942 – 956.
[13] T. Andersson, M. Lundquist, G. T. Dolphin, K. Enander, B.-H.
Jonsson, J. W. Nilsson, L. Baltzer, Chem. Biol. 2005, 12, 1245 –
1252.
[14] L. T. Tegler, K. Fromell, B.-H. Jonsson, J. Viljanen, C. Winander,
J. Carlsson, L. Baltzer, ChemBioChem 2010, DOI: 10.1002/
cbic.201000556.
[15] R. W. King, A. S. V. Burgen, Proc. R. Soc. London Ser. B 1976,
193, 107 – 125.
[16] A. Fersht, A Guide to Enzyme Catalysis and Protein Folding,
W. H. Freeman and Company, New York, 1999, chap. 11.
In conclusion, the problem of creating high-affinity bind-
ers for proteins has been reduced to that of finding small-
molecule binders that bind at a micromolar level, a problem
that is of considerably less complexity. We believe that
Angew. Chem. Int. Ed. 2011, 50, 1823 –1827
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1827