Tuning linear copolymers into protein-specific hosts

Article abstract of DOI:10.1002/anie.200601161

Being picky: A new concept for selective protein recognition by artificial receptor molecules involves statistical copolymerization of simple binding monomers in defined stoichiometric ratios. As an example of its efficiency, an arginine-rich lysozyme is recognized by a bisphosphonate-rich copolymer with a remarkable K_d value of 25 nM, 100-times superior to that of cytochrome c, which is comparable in size and isoelectric point. (Chemical Equation Presented).

Full text of DOI:10.1002/anie.200601161

Communications
Selective Protein Recognition
DOI: 10.1002/anie.200601161
Tuning Linear Copolymers into Protein-Specific
Hosts**
Sebastian J. Koch, Christian Renner, Xiulan Xie, and
Thomas Schrader*
The selective protein recognition by synthetic receptor
molecules remains an important challenge for supramolecular
and bioorganic chemistry. To distinguish between proteins of
similar sizes and pI values (pI = isoelectric point), but of
different biological function, an artificial host must be able to
recognize the pattern of amino acid residues, that is, the
topology, polarity, as well as the electrostatic potential of a
protein surface. From protein–protein interactions, we can
learn that nature prefers large contact areas (approximately
1500 Š2) with an unusual amount of arginine and aromatic
amino acid residues involved in polar interactions (Coulomb,
p-cation). These interactions are complemented and enforced
by hydrophobic attraction between approaching nonpolar
domains or patches.[1] This combination is partly reflected in
recent attempts to produce artificial protein binders that
include secondary-structure mimetics,[2] multiplication of
single specific[3] or unspecific binding sites,[4] evolutive
optimization of biomacromolecular scaffolds,[5] and molecular
[*] S. J. Koch, C. Renner, X. Xie, Prof. T. Schrader
Philipps-Universität Marburg
Fachbereich Chemie
Hans-Meerwein-Strasse, 35032 Marburg (Germany)
Fax: (+49)6421-28-25544
E-mail: schradet@staff.uni-marburg.de
[**] We thank Dr. J. Piehler (Biocenter Frankfurt University) for the
opportunity to perform RIfS measurements in his group. C.R. is also
indebted to the Deutsche Forschungsgemeinschaft for funding. We
gratefully acknowledge generous support from Prof. Hynes
(National University of Ireland, Galway) by providing his
WinEQNMR program to evaluate 2:1 complexation events. The
disclosure of preliminary results from a cell-wall degradation assay
with lysozyme by Kirstin Wenck is also gratefully acknowledged.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
6352
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 6352 –6355

Angewandte
Chemie
imprinting.[6] Herein, we introduce a new chemical concept
that involves statistical copolymerization of simple binding
monomers in defined stoichiometric ratios.[7]
Table 1: Modular set of methacrylamide-based co-monomers (n=1 or 7)
andpolymers 1–6 derived thereof.[a]
The underlying idea is the following: for a given protein
target, the chemist chooses from a preformed set of polymer-
izable binding sites those that match the majority of amino
acid residues dominating the surface character of the protein.
By statistical free-radical copolymerization, a flexible copo-
lymer is produced that is capable of performing an extensive
“induced-fit” procedure on top of the respective protein
surface, and therefore reaching maximal attractive noncova-
lent interactions. No cross-linking is employed to guarantee
perfect water solubility; no imprinting technique is used
because the linear polymer is expected to adapt its shape to
flat and rugged surface topologies. Thus, technical simplicity
is maintained to accomplish a complex task. This modular
approach deliberately avoids extensive design, requiring
exact knowledge of protein-surface topology from crystal
structures. Herein, we disclose preliminary, but highly prom-
ising, results.
Polymer Bisphosphonate Alcohol Dansyl Cyclohexyl Dodecyl
a
b,c
d
e
f
1
2
3
4
5
6
1
1
1
3
1
3
–
0.1
0.4
1
0.5
0.3
0.5
–
–
–
1
–
–
–
–
–
–
1
1
3[b]
1[c]
1[c]
1[c]
1[c]
According to the above-outlined concept, the whole
procedure can be subdivided into three main steps: 1) A set
of simple methacrylamide-based co-monomers is prepared;
this ensemble is by itself capable of recognizing all proteino-
genic basic, acidic, polar, aromatic, and nonpolar amino acid
residues. 2) From this set, the co-monomers that are comple-
mentary to the most characteristic protein surface residues
are copolymerized in an estimated appropriate stoichiometric
ratio, with simultaneous incorporation of a small amount of
polymerizable fluorophors (see Table 1). The new polymeric
materials are conveniently characterized by 1H and 31P NMR
spectroscopy, gel permeation chromatography, elemental
analysis, as well as fluorescence properties. 3) In the last
phase, affinities to very similar proteins are evaluated, at best
in mixtures.
To demonstrate the feasibility of this concept, we faced
the task of distinguishing between various basic proteins of
similar size and pI value. To this end, we prepared four
different binding monomers, tailored for basic, aliphatic, and
aromatic residues; an amino sugar based component was
added for increased water solubility. Further to the simple
polymers 1 and 2, four copolymers (3–6) were synthesized,
each of which contained three different binding monomers in
a 1:1:1 or 3:1:1 ratio and 10% of a dansyl fluorescence label.
Free-radical copolymerization was conducted in N,N-dime-
thylformamide (DMF) and initiated by 0.5–0.7 mol% of
azobisisobutyronitrile. Subsequent polymer-analogous reac-
tions with LiBr led to mild dealkylation of the bisphospho-
nates and afforded the final receptor polymers. Their
composition is depicted in Table 1. Molecular-weight deter-
minations for 1 and 2 in DMF and water produced compa-
rable results, excluding a potential ion-pair induced aggrega-
tion. For copolymer 5 with hydrophobic residues, very similar
hydrodynamic radii were calculated from NMR spectroscopic
diffusion experiments at varying concentrations (50 mm to
1.6 mm). These hydrodynamic radii also suggest monomeric
species as do the symmetric Mw/Mn curves. Moreover, this
copolymer displays a very compact nature—resembling a
protein—most likely resulting from burial of dodecyl tails in
[a] Each row indicates the relative ratio of co-monomers for a given
copolymer. [b] Alcohol monomer from 2-hydroxypropylamine b. [c] Alco-
hol monomer from 2-glucosamine c.
its inner core while exposing the polar bisphosphonate head
groups into the bulk solvent. Copolymerization parameters,
determined for 2, indicate a perfect statistical copolymeriza-
tion, ruling out the formation of hydrophobic blocks or
aggregated micelles.
In a systematic study, each fluorescent polymer was
titrated with the same series of neutral and basic proteins.[8]
In most cases, fluorescence emission intensities increased
markedly on protein binding, presumably owing to reorien-
tation of the polymerꢀs binding sites towards the protein
surface with concomitant deshielding of the fluorophor.
Unfortunately, the most basic protein histone H1formed
insoluble complexes even at 10 nm concentrations, precluding
the exact determination of its free binding energies. All of the
other examined protein complexes, however, furnished bind-
ing isotherms with a perfect fit by nonlinear regression
methods.[9] The respective association constants (Ka) are
summarized in Table 2; they cover a relatively broad range
from 104 mÀ1 up to over 107 mÀ1
.
It is instructive to compare Ka values for a given protein
among the various copolymers because they reveal the
contribution of additional noncovalent interactions, espe-
cially in relation to pure polybisphosphonates. Thus, bovine
serum albumin (BSA), a blood lipid carrier, is bound much
more strongly if the polymer (5) also contains long aliphatic
side chains for additional hydrophobic attraction (Figure 1).
Most likely these imitate the natural binding mode inside the
nonpolar cleft. Similarly, for all copolymers, the dansyl-
containing 3 displays superior affinity towards trypsin with its
unusually high content of aromatic residues (dissociation
constant, Kd = 170 nm).[10] Clearly, additional p stacking ren-
ders complex formation even 100-times more efficient than
the parent bisphosphonate copolymer 2. Experimentally
determined stoichiometries (Job plots) remain around 1:1
Angew. Chem. Int. Ed. 2006, 45, 6352 –6355
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6353

Communications
5
Table 2: Protein affinities to hosts 1–6, determined as K1:1 values by fluorescence titrations in 30 mm
relative polymer affinities (6: 4 ” 1 0; 8: 6 ”
phosphate buffer at pH 7.0.[a]
5
6
105; 7: 8 ” 1 0; 5: 2 ” 1 0).
In our systematic binding study for
representative basic proteins, we identified
among polymers 1–6 a candidate with pro-
nounced protein selectivity (Figure 2). Argi-
nine-rich lysozyme is recognized by
bisphosphonate-rich copolymer 6 with a
remarkable Kd value of 25 nm, which is
100-times superior to cytochrome c (compa-
rable in size and pI; Table 2). Close inspec-
tion of the electrostatic potential surface[13]
reveals numerous hydrophobic patches
between the Arg residues, ideal for van der
Waals interactions with the copolymerꢀs long
Polymer
Cytochrome c
Ka [mÀ1 Ratio[b]
Trypsin
Ka [mÀ1 Ratio[b]
Lysozyme
Ka [mÀ1 Ratio[b]
BSA
]
]
]
]
Ka [mÀ1
Ratio[b]
1
2
3
4
5
6
6”105
2”105
5”105
3”105
2”106
4”105
1:2.0
1:3.0
1.0:1
1.5:1
1.7:1
2.0:1
>105
4”104
6”106
–
–
>105
4”106
7”106
9”105
–
–
2”105
2”105
2”105
4”105
7”105
6”105
1:2.5
1:1.5
1:1.5
1.5:1
1.5:1
1.5:1
1:4.0
1.5:1
–
1:4.0
1:2.0
2.0:1
–
–
–
6”105
1.5:1
4”107
2.0:1
[a] Complex formation was assumedto occur without cooperativity andwith uniform binding constants
for each step. Standard deviations from the nonlinear regressions were calculated at 2–17%.
[b] Polymer/protein stoichiometry.
aliphatic tails. Compared to protein complexes of 1 or 2, the
glucosamine content in 3–6 leads to markedly improved water
solubilities.
Preliminary results from an enzyme assay measuring the
kinetics of bacterial cell wall degradation by lysozyme
(Micrococcus lysodeicticus) indicate a competitive inhibition
of enzyme activity by bisphosphonate-containing copolymers.
Intriguingly, the corresponding IC50 values follow the same
order as their binding constants determined by fluorescence
5
titration. Thus, polymers 4 (Ka = 9 ” 1 0 mÀ1), 3 (Ka = 7 ”
7
106 mÀ1), and 6 (Ka = 4 ” 1 0 mÀ1) produced IC50 values of
roughly 20, 10, and 1mm for enzymatic bacterial degradation,
indicating, that 2 equivalents of polymer 6 suffice to com-
pletely shut down lysozyme activity, whereas polymer 4 needs
about 50 equivalents for the same task ( ꢀ 1 mm protein
concentration).
Reflectometric interference spectroscopy (RIfS) meas-
urements record the change of optical thickness for com-
pound layers deposited successively on glass chips. A first
layer of cationic polyethyleneimine (PEI) was coated with
negatively charged copolymers 3 and 5, followed by non-
covalent immobilization of various proteins. The binding
constants and dissociation rates derived thereof confirm the
affinity and selectivity trends observed in solution (see
Supporting Information for details).[14] More importantly,
however, the amount of irreversible binding contributions
compared to the parent polybisphosphonates 1 and 2
increased drastically (Figure 3). As the latter rely mostly on
electrostatic interactions, this change strongly indicates
Figure 1. Top: Target optimization of homopolymer 1 into an efficient
BSA binder 5 by incorporation of dodecyl co-monomers for additional
hydrophobic interactions. Bottom: Crystal structure of human serum
[12]
albumin with complexedlipids.
Note the basic residues (blue) at the
entrance of the clefts, holding the myristic acid guest (black) in place
(see zoom).
throughout the whole series, indicating optimal
surface coverage of the protein.[11] To prove that
cooperative effects are negligible, two more
copolymers, 7 and 8, were synthesized that were
comprised of the same monomer composition but
different molecular weights (see Supporting Infor-
mation). Their bisphosphonate (BP) content (2:1:1
BP/dodecyl/sugar) lies between that of polymer 5
(1:1:1) and polymer 6 (3:1:1). This also implies that
the amount of hydrophobic dodecylmethacryla-
mide co-monomers steadily increases from 6 over 7
and 8 to 5. Intriguingly, the binding constants with
cytochrome c follow the very same trend, indicat-
ing that, in this series, the number of hydrophobic
Figure 2. Drastic selectivity increase from copolymer 2 to lysozyme binder 6.
interactions and not electrostatics determines the lys=lysozyme, cyt=cytochrome c, P=phosphonate.
6354
www.angewandte.org ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2006, 45, 6352 –6355

Angewandte
Chemie
Schneider, J. Feigon, Z. Hostomsky, L. Gold, Biochemistry 1995,
34, 9599.
essential hydrophobic, van der Waals, or p-stacking inter-
actions between the aliphatic and aromatic co-monomer
residues and nonpolar protein surface areas. For a kinetically
stable protein immobilization, low off-rates (dissociation
rates) are mandatory; the new copolymers thus offer the
[6] Reconstructed liposomes: M. Santos, B. C. Roy, H. Goicoechea,
A. D. Campiglia, S. J. Mallik, J. Am. Chem. Soc. 2004, 126,
10738; CuII-imprinted silica: S. Mallik, S. D. Plunkett, P. K. Dhal,
R. D. Robinson, D. Pack, D. Shenk, F. H. Arnold, New J. Chem.
1994, 18, 299; saccharide/polymer coating: H. Q. Shi, W. B. Tsai,
M. D. Garrison, S. Ferrari, B. D. Ratner, Nature 1999, 398, 593.
[7] It was reported that merely controlling the amount and nature of
functional side chains yielded, in a combinatorial fashion, a
polymer with remarkable phosphatase properties, although
statistical distribution of catalytically active sites was allowed:
F. M. Menger, A. V. Eliseev, V. A. Migulin, J. Org. Chem. 1995,
60, 6666.
[8] For an account on the parent bisphosphonate homopolymers as
synthetic hosts for basic proteins, see: C. Renner, J. Piehler, T.
Schrader, J. Am. Chem. Soc. 2006, 128, 620.
[9] K. A. Connors, Binding Constants, Wiley, New York, 1987.
[10] Protein surface residue distributions were evaluated with the
PROVE program (PROtein Volume Evaluation): J. Pontius, J.
Richelle, S. J. Wodak, J. Mol. Biol. 1996, 264, 121.
Figure 3. RIfS curves for lysozyme recognition on a polymer-coatedPEI
surface (solidline: homopolymer 1; dashed line: dansyl copolymer 3).
Note the elevatedlevel of irreversible binding with 3. nd=optical
thickness.
[11] M. T. Blanda, J. H. Horner, M. Newcomb, J. Org. Chem. 1989,
54, 4626. Three independent titrations between polymer 1 and
cytochrome c furnished identical stoichiometry factors and the
following Ka values: a) 5.4 ” 105 mÀ1 Æ 3% (1:2); b) 5.5 ”
105 mÀ1 Æ 2% (1:2); 5.8 ” 105 mÀ1 Æ 9% (1:1.9).
unique option of mild noncovalent protein attachment to the
chip surface and its potential observation in a functional state
(e.g., in new enzyme-sensor systems).
[12] S. Curry, H. Mandelkow, P. Brick, N. Franks, Nat. Struct. Biol.
1998, 5, 827.
We are currently optimizing the above-introduced mod-
ular concept by creating diverse copolymer libraries with a
representative pool of binding monomers for all important
noncovalent interactions occurring in protein recognition
events. A fast RIfS protocol will be useful for efficient
screening. Optimized binding monomers for each class of
amino acid residues may pave the way towards the ultimate
goal—artificial antibodies.
[13] M. F. Sanner, J. C. Spehner, A. J. Olson, Biopolymers 1996, 38,
305.
[14] J. Piehler, A. Brecht, G. Gauglitz, M. Zerlin, C. Maul, R.
Thiericke, S. Grabley, Anal. Biochem. 1997, 249, 94.
Received: March 23, 2006
Revised: June 8, 2006
Published online: August 28, 2006
Keywords: copolymers · NMR spectroscopy ·
.
noncovalent interactions · protein recognition ·
reflectometric interference spectroscopy
[1] T. A. Larsen, A. J. Olsen, D. S. Goodsell, Structure 1998, 6, 421; J.
Janin, S. Miller, C. Chothia, J. Mol. Biol. 1988, 204, 1 55; A. A.
Bogan, K. S. Thorn, J. Mol. Biol. 1998, 280, 1; F. B. Sheinerman,
R. Norel, B. Honig, Curr. Opin. Struct. Biol. 2000, 10, 153.
[2] a-helices: H. Yin, G. I. Lee, K. A. Sedey, J. M. Rodriguez, H. G.
Wang, S. M. Sebti, A. D. Hamilton, J. Am. Chem. Soc. 2005, 127,
5463; b-turns: A. B. Smith, W. Y. Wang, P. A. Sprengeler, R.
Hirschmann, J. Am. Chem. Soc. 2000, 122, 11037.
[3] L. Baldini, A. J. Wilson, J. Hong, A. D. Hamilton, J. Am. Chem.
Soc. 2004, 126, 5656; E. A. Merritt, Z. Zhang, J. C. Pickens, M.
Ahn, W. G. J. Hol, E. Fan, J. Am. Chem. Soc. 2002, 124, 8818; A.
Ojida, Y. Miyahara, T. Kohira, I. Hamachi, Biopolymers 2004,
76, 177; Md. A. Fazal, B. C. Roy, S. Sun, S. Mallik, K. R. Rodgers,
J. Am. Chem. Soc. 2001, 123, 6283.
[4] Gold clusters: N. O. Fischer, A. Verma, C. M. Goodman, J. M.
Simard, V. M. Rotello, J. Am. Chem. Soc. 2003, 125, 13387;
postsynthetically modified polymers: L. E. Strong, L. L. Kies-
sling, J. Am. Chem. Soc. 1999, 121, 6193.
[5] Protein grafting: S. E. Rutledge, H. M. Volkman, A. Schepartz,
J. Am. Chem. Soc. 2003, 125, 14336; RNA aptamers: D. J.
Angew. Chem. Int. Ed. 2006, 45, 6352 –6355
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6355