Noncovalent template-assisted mimicry of multiloop protein surfaces: Assembling discontinuous and functional domains

Article abstract of DOI:10.1021/ja305360q

We report here a novel noncovalent synthetic strategy for template-assembled de novo protein design. In this approach, a peptide was first conjugated with two oligoguanosine strands via click chemistry and the conjugates were then self-assembled in the presence of metal ions. G-quadruplex formation directs two peptide strands to assemble on one face of the scaffold and form an adjacent two loop surface. This approach can be used to rapidly prepare multiple two-loop structures with both homo- and heterosequences.

Full text of DOI:10.1021/ja305360q

Communication
pubs.acs.org/JACS
Noncovalent Template-Assisted Mimicry of Multiloop Protein
Surfaces: Assembling Discontinuous and Functional Domains
Partha S. Ghosh† and Andrew D. Hamilton*,†,‡
†Department of Chemistry, Yale University, New Haven, Connecticut 06520-8107, United States
‡Department of Chemistry, Oxford University, Oxford, U.K. OX13TA
S
* Supporting Information
ABSTRACT: We report here a novel noncovalent
synthetic strategy for template-assembled de novo protein
design. In this approach, a peptide was first conjugated
with two oligoguanosine strands via click chemistry and
the conjugates were then self-assembled in the presence of
metal ions. G-quadruplex formation directs two peptide
strands to assemble on one face of the scaffold and form an
adjacent two loop surface. This approach can be used to
rapidly prepare multiple two-loop structures with both
homo- and heterosequences.
here is intense current interest in the identification of
Tmolecules that bind to protein targets with affinity and
specificity similar to antibodies.1 In considering different
strategies to this end, much can be learned from the
mechanisms by which antibodies achieve both high variability
and complementarity toward their targets. Critical for this are
the nature and positioning of multiple peptide loop domains
(the hypervariable loops) that comprise the complementary
determining region (CDR) and are held in place by covalent
(protein sequence) and noncovalent (protein folding and
association) bonds within the immunoglobulin light and heavy
Figure 1. (a) An example of antigen−antibody interactions: gp120
(HIV envelope protein; orange)-VRC01 (antibody for gp120; VH in
blue, VL in magenta), pdb 3NGB. (b) Covalent (i) and noncovalent
(ii) strategies for mimicking multiloop surfaces. (c) Display of two
constrained peptides on one face (5′) of G-quadruplex.
chains. In antibody−antigen recognition, several loops in the
CDR interact with discontinuous epitopes of an antigen (Figure
1a).
In recent years, several approaches have been taken to the
identification of synthetic scaffolds that display multiple peptide
loops with structural control over their position and potential
recognition properties, in direct analogy to antibodies. In a
seminal study, Mutter has developed a template assembled
synthetic protein approach (TASP) in which as many as three
peptide loops are covalently linked across a macrocyclic
oligopeptide template (Figure 1b(i)).2 This strategy has been
extended by Fairlie to other macrocyclic scaffolds including
linked oxazoles and thiazoles.3 In an alternative approach, we4
and others5 have shown that protein binding agents can be
prepared through the covalent attachment of several preformed
synthetic peptide loops to a macrocyclic scaffold that is itself
covalently or noncovalently formed. Recently in an important
development, Heinis and Winter have created phage display
libraries of double peptide loop derivatives (bicycles) through
the triple alkylation of three defined Cys residues within the
randomized peptide sequence.6 A related strategy based on
Cys-containing synthetic peptide libraries has been reported by
Timmerman for the preparation of loop structures.7
Each of these strategies involves covalently constraining the
chosen peptide sequence through a series of synthetic
modifications and thus lacks the direct and powerful simplicity
of the antibody whereby the loops form through noncovalently
stabilized protein folding. We sought a synthetic solution in
which functionalization of a peptide at either end with
supramolecular capping groups, that can themselves non-
covalently interact intra- and intermolecularly, might lead to the
self-assembly of multiple adjacent peptide loops (as in Figure
1b(ii)). In this way, a noncovalent template would direct
oriented assembly of peptide fragments, simultaneously
enforcing their conformational restrictions. A suitable supra-
molecular fragment for directed folding is represented by a
strand of oligonucleotides, which depending on their sequence
Received: June 3, 2012
Published: July 27, 2012
© 2012 American Chemical Society
13208
dx.doi.org/10.1021/ja305360q | J. Am. Chem. Soc. 2012, 134, 13208−13211

Journal of the American Chemical Society
Communication
can form a range of different aggregate structures and
stoichiometries.8 In this paper, we show a first example of
this strategy in which two oligoguanosine strands are attached
to both N- and C-termini of a peptide through their 5′ ends.
Upon addition of metal ions, formation of parallel G-
quadruplexes should take place spontaneously in aqueous
media via synergistic H-bonding, metal−ligand and aromatic
stacking interactions (Figure 1c). The oriented assembly will
thus be induced to display two loops from the same face of the
template. The noncovalent synthetic strategy can be further
exploited to create a combination of multiloop structures.
The key challenge is the ready preparation of linear
oligomeric components composed of oligoG−peptide−oligoG
strands. The chemistry of conventional peptide solid-phase
synthesis is incompatible with that of oligonucleotides, which
makes difficult the linear synthesis of peptide−oligonucleotide
conjugates (POCs) on a single support. Instead a fragment
conjugation approach was developed involving copper(I)-
catalyzed azide−alkyne [3 + 2] cycloaddition (CuAAC) as a
ligation method. The CuAAC proceeds rapidly with high yield
in an aqueous medium.9 This bioorthogonal click reaction can
be exploited for any peptide sequence requiring no protecting
groups on either the loop or the template. The optimal strategy
involved reacting two alkyne functionalized oligonucleotides
with a peptide strand containing two azide groups near the C-
and N-termini.
(>O2) to give sufficient mass difference between the conjugates
(O1−P1−O1 ≫ O2−P2−O2, vide infra). The desired double
click product was not obtained using CuSO4/sodium ascorbate
(NaAsc) mixture with 1 equiv Cu/azide, as reported.10
However, using tris(3-hydroxypropyltriazolylmethyl)amine
(THPTA), a water-soluble Cu(I)-stabilizing ligand, efficient
conjugation was achieved {CuSO4/NaAsc/THPTA (1:10:5; 10
equiv Cu/azide) in 2:1 TEAA/tBuOH (TEAA = triethylamine
acetate buffer, 200 mM, pH 7.2)}.11 The POCs were purified
by RP-HPLC and characterized by MALDI-TOF.
The POCs were then self-assembled in metal ion (K+ or
Na+) containing buffer (10 mM Tris-HCl, 80 mM salt, pH 7.4)
by first heating at 90 °C for 10 min, followed by slow cooling to
room temperature and finally incubation at 4 °C for an
additional 48 h.
The structure of the annealed POCs was characterized by
nondenaturing polyacrylamide gel electrophoresis (PAGE) and
circular dichroism (CD). As depicted in Figure 2a, in the
presence of metal ions, the POCs could reasonably assemble
We first synthesized Fmoc-protected azido amino acids
(AA_N3s) from the corresponding amino acids by Cu(II)-
catalyzed diazo transfer reaction (Scheme 1a). These were then
Scheme 1. Synthesis of OligoG−Peptide−OligoG
Conjugates: (a) Fmoc-azide Amino Acids (AA_N3s) via
diazo transfer reaction, (b) peptides with AA_N3s near the
Termini by SPPS and Peptide−Oligo Conjugation by
a
CuAAC
a
(i) NaHCO3, CuSO4, H2O/MeOH/CH2Cl2, rt, ON; (ii) (a) HOBt,
HBTU, DIPEA, DMF, rt, 1 h; (b) piperidine, DMF, rt, 15 min; (iii)
H2O, rt, 4 h (iv) 1:10:5 CuSO4:NaAsc:THPTA, 2:1 TEAA:tBuOH, rt,
3 h.
incorporated into model peptides (P1 and P2) by Fmoc-based
solid-phase chemistry. The peptides differ in molecular weight,
location of the ligation points (AA_N3s) and the distance
between two N3-AAs. P1 contains aspartic acid and alanine
with the Lys_N3s separated by six residues. P2 consists of five
aspartic acid and glycine residues with two Dap_N3s at the
termini.
Figure 2. (a) Alternative folded structures of self-assembled POCs. (b)
PAGE images showing POCs are forming dimeric complexes
according to their electromobility shifts. (c and d) CD spectra of
the POC complexes and known parallel and antiparallel G-
quadruplexes. (e) Kinetics of G-quadruplex formation of POC1 and
TG4T2. (f) Dissociation kinetics of the complexes at their T1/2, inset
shows their melting profile.
The higher molecular weight peptide P1 (>P2) was
conjugated with higher molecular weight oligonucleotide O1
13209
dx.doi.org/10.1021/ja305360q | J. Am. Chem. Soc. 2012, 134, 13208−13211

Journal of the American Chemical Society
Communication
into two dimeric or one tetrameric forms. To assess the
molecularity of the association, we performed a PAGE assay
using short oligonucleotide duplexes (15-mer and 21-mer) as
molecular weight markers and known tetramers ((TG4T2)4/
K+) and (TG4T)4/K+) as a positive control. The gel mobility
shift of POCs/K+ indicated that the POCs assembled into a
bimolecular quadruplex structure (Figure 2b).
Dimeric assembly of the POCs can take two quadruplex
topologies depending on the orientation of the oligonucleotide
strands in a parallel or antiparallel direction (see Figure 2a). To
determine the orientation, we compared the circular dichroism
spectra of the assembled POCs with those of known parallel
and antiparallel quadruplexes ((TG4T2)4/K+) and
(TG4T4G4T)2/Na+, respectively).12 We observed the charac-
teristic peaks of a parallel DNA quadruplex for both POC1 and
2 in the presence of K+ (maxima at 263 and 203 nm and a
minimum at 242 nm, Figure 2c,d).
To better understand the assembly, we studied the kinetics of
POC association (at 4 °C). That POC1 formed the G-
quadruplex core faster than control TG4T2 is likely due to an
initial intramolecular association of two oligoguanosine strands
attached to the peptide followed by their bimolecular assembly
to form the quadruplex core (Figure 2e). We also studied
dissociation behavior of the assembled structures, in Na+-
solution as the K+-complexes did not denature even at high
temperature. The melting profile of the assembled structures
with POC1 was similar to those with (TG4T2)4/Na+ giving
further support for dimeric association of POC1 forming four
G-tetrads as opposed to tetrameric complexes with eight G-
tetrads and hence a higher expected melting temperature
(Figure 2f inset). By comparing the dissociation kinetics at a
particular temperature (60 °C), we found that the presence of
the loops slowed down the melting of the quadruplex core
(Figure 2f).
This approach can provide a direct route to combinatorial
libraries by multicomponent self-assembly.13 For example, by
combining two from a library of n POCs, comprising different
peptide sequences, it is possible to formulate a diverse
collection of double loop structures with n homo- and n(n −
1)/2 heterosequences (total n(n + 1)/2). To demonstrate this,
we co-incubated equimolar ratios of POC1 and POC2 in K+-
containing buffer. The annealed solution was then analyzed by
PAGE using the homoassemblies (POC12 and POC22) as a
reference. As shown in Figure 3b, three bands appeared during
co-assembly of the two POCs (middle lane), which correspond
to their statistical mixture (Figure 3c).
In conclusion, by using an improved peptide−oligonucleo-
tide click conjugation methodology, we have developed a self-
organizing structure that through G-quadruplex formation
positions two peptide loops on one surface, in direct analogy
to antibody binding domains. The multicomponent non-
covalent synthesis allows facile formation of homo- and
heterocombinations.
ASSOCIATED CONTENT
* Supporting Information
■
S
Synthetic method of POC, experimental protocols and
supplementary figures. This material is available free of charge
via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
andrew.hamilton@admin.ox.ac.uk
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was financially supported by the National Science
Foundation (CHE0750357) and Yale University. We also thank
Dr. Ian Jones for his assistance.
Figure 3a shows a molecular model of the double peptide
loop structure obtained by keeping the quadruplex core fixed
REFERENCES
■
(1) (a) Williams, B. A. R.; Diehnelt, C. W.; Belcher, P.; Greving, M.;
Woodbury, N. W.; Johnston, S. A.; Chaput, J. C. J. Am. Chem. Soc.
2009, 131, 17233−17241. (b) Villen, J.; Rodriguez-Mias, R. A.; Nunez,
J. I.; Giralt, E.; Sobrino, F.; Andreu, D. Chem. Biol. 2006, 13, 815−823.
(2) Mutter, M.; Dumy, P.; Garrouste, P.; Lehmann, C.; Mathieu, M.;
Peggion, C.; Peluso, S.; Razaname, A.; Tuchscherer, G. Angew. Chem.,
Int. Ed. 1996, 35, 1482−1485.
(3) Singh, Y.; Stoermer, M. J.; Lucke, A. J.; Guthrie, T.; Fairlie, D. P.
J. Am. Chem. Soc. 2005, 127, 6563−6572.
(4) (a) Hamuro, Y.; Calama, M. C.; Park, H. S.; Hamilton, A. D.
Angew. Chem., Int. Ed. 1997, 36, 2680−2683. (b) Park, H. S.; Lin, Q.;
Hamilton, A. D. J. Am. Chem. Soc. 1999, 121, 8−13. (c) Cai, J. F.;
Rosenzweig, B. A.; Hamilton, A. D. Chem.Eur. J. 2009, 15, 328−332.
(5) (a) Chamorro, C.; Kruijtzer, J. A. W.; Farsaraki, M.; Balzarini, J.;
Liskamp, R. M. J. Chem. Commun. 2009, 821−823. (b) Pfeifer, M. E.;
Robinson, J. A. Chem. Commun. 1998, 1977−1978.
Figure 3. (a) Molecular model of the assembled structures. (b) PAGE
of the complexes after coassembly of POC1 and POC2 in presence of
K+. (c) Schematic illustration of the structures formed (a statistical
mixture) during the coassembly.
(6) Heinis, C.; Rutherford, T.; Freund, S.; Winter, G. Nat. Chem. Biol.
2009, 5, 502−507.
(7) Timmerman, P.; Beld, J.; Puijk, W. C.; Meloen, R. H.
ChemBioChem 2005, 6, 821−824.
(8) Rosenzweig, B. A.; Ross, N. T.; Adler, M. J.; Hamilton, A. D. J.
Am. Chem. Soc. 2010, 132, 6749−6754.
(9) (a) Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int.
Ed. 2001, 40, 2004−2021. (b) Beatty, K. E.; Tirrell, D. A. Bioorg. Med.
Chem. Lett. 2008, 18, 5995−5999.
and then energy minimizing the peptide fragment. The result
shows a potential binding surface defined by the sequence and
conformations of the peptide loops as well as their relative
distance across the G-quadruplex surface.
(10) Gogoi, K.; Mane, M. V.; Kunte, S. S.; Kumar, V. A. Nucleic Acids
Res. 2007, 35, e139.
13210
dx.doi.org/10.1021/ja305360q | J. Am. Chem. Soc. 2012, 134, 13208−13211

Journal of the American Chemical Society
Communication
(11) (a) Hong, V.; Presolski, S. I.; Ma, C.; Finn, M. G. Angew. Chem.,
Int. Ed. 2009, 48, 9879−9883. (b) Ghosh, P. S.; Hamilton, A. D.
Chem.Eur. J. 2012, 18, 2361−2365.
(12) (a) Lu, M.; Guo, Q.; Kallenbach, N. R. Biochemistry 1992, 31,
2455−2459. (b) Miyoshi, D.; Nakao, A.; Sugimoto, N. Biochemistry
2002, 41, 15017−15024.
(13) Roy, L.; Case, M. A. J. Am. Chem. Soc. 2010, 132, 8894−8896.
13211
dx.doi.org/10.1021/ja305360q | J. Am. Chem. Soc. 2012, 134, 13208−13211