Dynamic covalent side-chain cross-links via intermolecular oxime or hydrazone formation from bifunctional peptides and simple organic linkers

Article abstract of DOI:10.1002/psc.2596

Peptide cyclization via chemoselective reactions between side chains has proven a useful strategy to control folded structure. We report here a method for the synthesis of side-chain to side-chain cyclic peptides based on the intermolecular reaction between a linear peptide functionalized with two aminooxy or hydrazide side chains and an organic dialdehyde linker. A family of oxime-based and hydrazone-based cyclic products is prepared in a modular and convergent fashion by combination of unprotected linear peptide precursors and various small molecule linkers in neutral aqueous buffer. The side-chain to side-chain linkages that result can alter peptide folding behavior. The dynamic covalent nature of the Schiff bases in the cyclic products can be utilized to create mixtures where product composition changes in response to experimental conditions. Thus, a linear peptide precursor can select one organic linker from a mixture, and a cyclic product can dynamically exchange the small molecule component of the macrocycle. Copyright 2014 European Peptide Society and John Wiley & Sons, Ltd. We report a method for the synthesis of side-chain to side-chain cyclic peptides based on the intermolecular reaction between a linear peptide with two aminooxy or hydrazide side chains and an organic dialdehyde linker. The dynamic covalent nature of the Schiff bases in the cyclic products enables the creation of mixtures where product composition changes in response to experimental conditions. Copyright

Full text of DOI:10.1002/psc.2596

Special Issue Article
Received: 26 August 2013
Revised: 14 November 2013
Accepted: 15 November 2013
Published online in Wiley Online Library: 9 January 2014
(wileyonlinelibrary.com) DOI 10.1002/psc.2596
Dynamic covalent side-chain cross-links
via intermolecular oxime or hydrazone
formation from bifunctional peptides
and simple organic linkers‡
Conor M. Haney and W. Seth Horne*
Peptide cyclization via chemoselective reactions between side chains has proven a useful strategy to control folded structure.
We report here a method for the synthesis of side-chain to side-chain cyclic peptides based on the intermolecular reaction
between a linear peptide functionalized with two aminooxy or hydrazide side chains and an organic dialdehyde linker. A
family of oxime-based and hydrazone-based cyclic products is prepared in a modular and convergent fashion by combination
of unprotected linear peptide precursors and various small molecule linkers in neutral aqueous buffer. The side-chain to
side-chain linkages that result can alter peptide folding behavior. The dynamic covalent nature of the Schiff bases in the cyclic
products can be utilized to create mixtures where product composition changes in response to experimental conditions. Thus,
a linear peptide precursor can select one organic linker from a mixture, and a cyclic product can dynamically exchange the
small molecule component of the macrocycle. Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd.
Supporting information may be found in the online version of this article at the publisher’s web site.
Keywords: cyclic peptides; peptide staples; Schiff bases; dynamic covalent chemistry
intermolecular cyclizations have been used to develop bioactive
agents [13–15].
Introduction
Peptides often lack well-defined folded structure when removed
from the context of a larger protein. In order to constrain short
sequences into compact folds, nature often employs disulfide
cross-links between cysteine side chains to form cyclic or polycy-
clic molecules [1]. Inspired by these natural architectures,
chemists have developed strategies for the synthesis of cyclic
peptide scaffolds with a particular structure and/or biological
function [2,3]. Most peptide side-chain to side-chain cross-linking
methods rely on the direct intramolecular reaction between two
compatible functional groups either on-resin during synthesis or
chemoselectively in an unprotected substrate. This intramolecu-
lar approach (Figure 1(A)) has led to cyclic oligomers with diverse
structures and biological activities based on linkages containing
disulfides [4], alkenes [5], lactams [6], oximes [7,8], thioethers
[9], 1,2,3-triazoles [10], and many other functional groups [2]. In
contrast to the intramolecular scheme, some methods for pep-
tide cyclization involve an intermolecular reaction between two
components: a linear peptide and a small organic molecule (Fig-
ure 1(B)). In this strategy, two identically functionalized side
chains in the peptide react with two compatible functional
groups in the linker to form two new bonds and a cyclic product.
After such a sequence, the small molecule component becomes
an integral part of the macrocycle, influencing the folding behav-
ior and function of the cyclic species. Examples of this
intermolecular approach include the reaction between peptide
thiols and linker alkyl halides to form thioethers [11] and the re-
action between peptide amines and linker carboxylic acid deriva-
tives to form amides [12]. Both thioether-based and amide-based
We have recently reported investigations on peptide cross-
linking based on oxime bonds [16,17], which can be formed
chemoselectively in unprotected peptides [18,19] and have the
potential for dynamic covalent exchange in water [20]. One
motivation for this work is the possibility of applying principles
of dynamic covalent chemistry [21,22] for the in situ selection of
peptide cross-links that promote a particular fold and biological
function (e.g. protein binding). In prior work, we showed that
oxime macrocycles formed via the intramolecular reaction
(Figure 1(A)) between aminooxy and aldehyde side chains can
influence folding in a simple α-helix [16] and a coiled-coil quater-
nary structure [17]. Although oxime formation triggered by in situ
unmasking of a side-chain aldehyde was able to kinetically trap a
particular folded structure [17], we saw limited evidence of
dynamic covalent exchange. In an effort to address this limitation
and further develop Schiff base cross-links as a general method
to control peptide folding, we report here a series of macrocyclic
oligomers prepared by an intermolecular strategy (Figure 1(B))
involving a peptide containing two nucleophilic aminooxy or
*
Correspondence to: W. Seth Horne, Department of Chemistry, University of
Pittsburgh, Pittsburgh, PA, USA. E-mail: horne@pitt.edu
‡
This article is published in Journal of Peptide Science as part of the Special Issue
devoted to contributions presented at the Chemical Protein Synthesis Meeting,
April 3-6, 2013, Vienna, edited by Christian Becker (University of Vienna, Austria).
Department of Chemistry, University of Pittsburgh, Pittsburgh, PA, USA
J. Pept. Sci. 2014; 20: 108–114
Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd.

DYNAMIC COVALENT PEPTIDE CROSS-LINKS
Figure 1. Strategies for peptide cyclization based on (A) intramolecular
reaction between two compatible side chains and (B) intermolecular
reaction between two identical side chains and a bifunctional small mol-
ecule linker.
hydrazide side chains and a dialdehyde organic linker. A family of
cyclic peptides of varying chemical structure and folding
propensity is readily synthesized by a convergent route and
shown capable of establishing dynamic covalent equilibria in
neutral aqueous buffer.
Scheme 1. Synthesis of protected hydrazide monomer 8. (a) tert-butyl
carbazate, DIEA, HOBt, and EDC · HCl; (b) TFA; (c) DIEA and 2-
chlorotritylchloride; (d) H2 and Pd/C; (e) Fmoc-OSu, DIEA, and TMS-Cl.
N′-ethylcarbodiimide hydrochloride (EDC · HCl) (2.11 g, 11.0 mmol,
1.1 equiv). The reaction was stirred at room temperature overnight
under nitrogen. The solution was then poured into a separatory fun-
nel, diluted with 150 ml ethyl acetate, and washed once each with
50 ml of 1 M hydrochloric acid, water, saturated sodium carbonate,
and brine. The organic layer was dried with anhydrous magnesium
sulfate and concentrated under reduced pressure. The crude oil
was purified by flash column chromatography (33% →66% ethyl ac-
etate in hexanes) to yield 6 as a white solid (2.34 g, 4.83 mmol, 48%
Materials and Methods
General
Solvents and reagents were purchased from commercial
suppliers and used as received unless otherwise specified. Flash
column chromatography was performed using Silicycle SiliaFlash
P60 (230–400 mesh) silica gel (SiliCycle, Quebec City, Quebec,
Canada). Optical rotations were measured on a Perkin-Elmer
241 digital polarimeter (PerkinElmer, Waltham, MA, USA) at ambi-
ent temperature using a sodium lamp. NMR spectra were
recorded on a Bruker Avance 400 MHz spectrometer (Bruker,
Billerica, MA, USA). MALDI-TOF mass spectra were collected on
a Voyager DE Pro instrument (Applied Biosystems, Foster City,
CA, USA). CD measurements were made on an Olis DSM17 CD
spectrometer (Olis, Bogart, GA, USA) using quartz cuvettes of
0.1 cm path length. Preparative and analytical HPLCs were
performed with Phenomenex Luna C18 columns (Phenomenex,
Torrance, CA, USA) and mobile phase composed of water/aceto-
nitrile containing 0.1% TFA. Typical gradients were 0–30% aceto-
nitrile over 40 min for preparative scales (15 ml/min flow rate), 5–
35% acetonitrile over 30 min for analytical (1 ml/min flow rate)
HPLCs of peptides 2 and 3 and their cyclized products, and either
10–30% (4a-c) acetonitrile or 10–35% (4d) acetonitrile over
30 min for peptide 4 and its cyclized products.
1
yield). [α]D =+1.8 (c=1.0, CHCl3). H NMR (400 MHz, CDCl3) δ =8.10
(1H, s), 7.34 (10H, m), 6.67 (1H, s), 5.88 (1H, d, J= 6.9 Hz), 5.16 (2H,
s), 5.09 (2H, s), 4.45 (1H, m), 2.26 (3H, m), 1.99 (1H, m), 1.44 (9H, s);
13C NMR (100 MHz, CDCl3) δ = 171.69, 156.42, 155.48, 136.06,
135.10, 128. 57, 128.47, 128.26, 128.15, 128.07, 81.67, 67.30, 67.07,
53.37, 29.88, 28.25, 28.05; high-resolution electrospray ionization
mass spectrometry (HRMS) m/z calculated for C25H32N3O7 [M + H]+:
486.2240; found: 486.2245.
(S)-Benzyl 2-(((benzyloxy)carbonyl)amino)-5-(2-((2-chlorophenyl)
diphenylmethyl)hydrazinyl)-5-oxopentanoate (7)
Compound
6 (2.19 g, 4.52 mmol) was dissolved in 20 ml
dichloromethane. TFA (5 ml) was added, and the reaction stirred
at room temperature for 2.5 h. The reaction was diluted with
50 ml dichloromethane, washed sequentially with 30 ml 5% w/v
sodium bicarbonate and 30 ml brine, then dried over anhydrous
magnesium sulfate, and concentrated. The resulting residue
was dissolved in 20 ml anhydrous dichloromethane under nitro-
gen, and DIEA (1.18 ml, 6.79 mmol, 1.5 equiv) was added. 2-
Chlorotrityl chloride (1.41 g, 4.52 mmol, 1 equiv) was dissolved
in 10 ml anhydrous dichloromethane and added slowly to the
reaction over 15 min. The reaction was stirred at room tempera-
ture overnight, diluted with 100 ml ethyl acetate, washed with
30 ml 10% w/v citric acid, and washed twice with 30 ml brine.
The organic layer was dried over anhydrous magnesium sulfate
and concentrated. The crude oil was purified by flash column
chromatography (33% → 50% ethyl acetate in hexanes) to yield
product 7 as a yellow solid with a small amount of residual ethyl
acetate (1.83 g, 2.76 mmol, 61% yield). NMR spectra recorded in
chloroform showed two conformers in slow exchange on the
NMR time scale. 1H NMR is reported in DMSO, where only a single
set of peaks was observed, and 13C NMR is reported in CDCl3 with
all signals reported. [α]D = À1.2 (c = 1.0, CHCl3). 1H NMR, (400 MHz,
DMSO): δ = 9.01 (1H, d, J = 8.3 Hz), 7.79 (1H, d, J = 7.4 Hz), 7.65 (1H,
Amino Acid Monomer Synthesis
Oxime-functionalized amino acid residue X was synthesized
in protected form as previously described [16]. Hydrazide-
functionalized residue Z was incorporated into peptides using
an Fmoc/2-Cl-Trt protected monomer, which was prepared
from Cbz-L-Glu-OBn (5) by adaptation of published methods
(Scheme 1) [23].
(S)-tert-Butyl 2-(5-(benzyloxy)-4-(((benzyloxy)carbonyl)amino)-5
oxopentanoyl)hydrazinecarboxylate (6)
Cbz-L-Glu-OBn (5, 3.71 g, 10.0mmol) was dissolved in 50 ml anhy-
drous dichloromethane under nitrogen. N,N-diisopropylethylamine
(DIEA, 5.22 ml, 30.0 mmol, 3 equiv) was added, followed by tert-butyl
carabazate (1.45 g, 11.0 mol, 1.1 equiv), 1-hydroxybenzotriazole (HOBt)
hydrate (1.68 g, 11.0 mmol, 1.1 equiv), and N-(3-dimethylaminopropyl)-
J. Pept. Sci. 2014; 20: 108–114 Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jpepsci

HANEY AND HORNE
d, J = 7.7 Hz), 7.42 (4H, m), 7.35 (11H, m), 7.28 (8H, m), 7.19 (2H, m),
6.10 (1H, d, J = 8.3 Hz), 5.11 (2H, m), 5.02 (2H, m), 3.88 (1H, m), 1.88
(2H, t, J = 7.4 Hz), 1.56 (1H, m), 1.44 (1H, m); 13C NMR (100 MHz,
CDCl3) δ = 177.24, 171.52, 169.03, 156.19, 155.98, 141.83, 139.60,
138.04, 136.06, 135.03, 134.50, 133.41, 132.44, 132.04, 131.32,
129.68, 128.90, 128.68, 128.59, 128.52, 128.48, 128.43, 128.35,
128.30, 128.18, 128.14, 128.10, 128.04, 127.93, 127.70, 126.90,
126.27, 74.20, 73.68, 67.29, 67.05, 66.83, 53.84, 53.24, 30.42,
28.70, 26.41; HRMS m/z calculated for C39H37ClN3O5 [M + H]+:
662.2422; found: 662.2396.
were performed by treatment with 92.5: 3 : 3 : 1.5, TFA : H2O : EDT :
TIS. The cleaved peptide was precipitated into cold diethyl ether
and centrifuged, and the ether decanted. The pelleted crude pep-
tide was dissolved in 0.1% TFA in water for purification by prepara-
tive HPLC. Because of the formation of an unidentified contaminant
when left in acetonitrile-containing solution, fractions containing
the desired aminooxy or hydrazide-functionalized peptide were
frozen immediately for lyophilization [24]. All peptide samples used
for biophysical characterization and cross-linking reactions were
≥95% pure as judged by analytical HPLC. Analytical HPLCs and
MALDI-TOF MS of peptides 1–4 and the corresponding cross-linked
products are available in the Supporting Information.
(S)-2-((((9H-Fluoren-9-yl)methoxy)carbonyl)amino)-5-(2-((2-
chlorophenyl)diphenylmethyl)hydrazinyl)-5-oxopentanoic acid (8)
Compound 7 (1.71 g, 2.58 mmol) was dissolved in 20 ml anhy-
drous methanol under nitrogen. Palladium on carbon (0.171 g,
10 wt.%) was added. The atmosphere was purged, the vessel fit
with a hydrogen-filled balloon, and the resulting mixture stirred
overnight. After this time, the reaction was filtered through celite,
washed with methanol, and concentrated under reduced
pressure. The resulting white solid was dissolved in anhydrous
dichloromethane and placed under nitrogen atmosphere. DIEA
(1.79 ml, 10.32 mmol, 4 equiv) was added, followed by
trimethylsilyl chloride (0.655 ml, 5.16 mmol, 2 equiv). The reaction
was stirred for 15 min at room temperature. 9-Fluorenylmethyl
N-succinimidyl carbonate (1.04 g, 3.09 mmol, 1.2 equiv) was
added, and the reaction stirred at room temperature overnight.
The mixture was diluted with 150 ml ethyl acetate and washed
twice with 50 ml brine, and the organics were dried over magne-
sium sulfate and concentrated. Flash column chromatography
(50% → 100% ethyl acetate in hexanes) of the crude oil yielded
compound 8 as a white solid (518.8 mg, 0.79 mmol, 30% yield).
[α]D = +1.9 (c = 1, CHCl3). 1H NMR (400 MHz, DMSO): δ = 12.58
(1H, s), 9.04 (1H, d, J = 8.2 Hz), 7.89 (2H, d, J = 7.5 Hz), 7.80 (1H, d,
J = 7.8 Hz), 7.70 (2H, d, J = 7.4 Hz), 7.49 (1H, d, J = 7.8 Hz), 7.41
(4H, m), 7.31 (10H, m), 7.19 (2H, m), 6.11 (1H, m), 5.00 (1H, s),
4.24 (2H, m), 4.20 (1H, m), 3.72 (1H, m), 1.85 (2H, t, J = 7.8 Hz),
1.54 (1H, m), 1.42 (1H, m); 13C NMR (100 MHz, CDCl3): δ = 170.40,
156.26, 143.77, 143.56, 141.59, 141.23, 139.39, 134.46, 132.11,
131.25, 129.02, 128.71, 128.32, 128.00, 127.71, 127.03, 126.38,
125.04, 119.95, 73.70, 66.99, 60.42, 53.54, 53.06, 47.05, 30.47,
28.80, 28.29; HRMS m/z calculated for C39H33ClN3O5 [M À H]À =
658.2109; found = 658.2137.
Cross-link Formation
Stock solutions of peptides 2–4 were prepared in water and
concentration determined by UV–vis (Tyr absorbance at 276 nm,
ε
276 = 1450 MÀ1 cmÀ1) [25]. Stock solutions of organic linkers a–d
were also prepared in water. For solutions of a–c, concentration
was determined by UV–vis (a, ε300 = 255 MÀ1 cmÀ1 for the corre-
sponding cyclic hydrate; b, ε292 = 1120 MÀ1 cmÀ1; c, ε296 = 1950 MÀ1
cmÀ1) [26]. ortho-Phthaldialdehyde (a) was allowed to equilibrate
in water for at least 30 min prior to UV–vis measurement to ensure
complete conversion to the cyclic hydrate [26]. The concentration
of the stock solution of d was determined by simple mass to
volume. Cross-linking reactions were performed in 0.1 M
phosphate buffer at pH 7, with 50 μM peptide and 4 equiv
(200 μM concentration) of linker. The benzene-based dialdehydes
a–c were well soluble in water and were delivered to the reaction
from stock solutions of 500–700 μM concentration. Biphenyl-based
linker d is less soluble in water and consequently was added to the
reaction from a stock solution just below its solubility limit in water
(~300 μM). Following addition of the linker, aliquots were removed
and analyzed by HPLC and MALDI-TOF MS of collected eluent.
Employing the previous general conditions, cyclization reactions
were performed using various combinations of peptides 2–4 and
linkers a–d to prepare cyclic species 2a–2d, 3b, 3d, and 4a–4d. In
all reactions of the aminooxy-functionalized peptides 2 and 3,
complete consumption of starting material was observed in
1–2 h, with the appearance of between one and four separable
HPLC peaks with masses corresponding to cyclized peptide [Figure
S1 and Table S1 (see Supporting Information)]. Cross-link formation
with peptide 4 yielded complete cyclization with linkers a–c and a
mixture of cyclic and linear peptide with linker d.
Peptide Synthesis
Peptides 1–4 were prepared by manual Fmoc solid-phase
peptide synthesis on NovaPEG Rink amide resin (EMD Chemicals,
Gibbstown, NJ, USA) with microwave heating (MARS microwave
reactor, CEM, Matthews, NC, USA). Coupling reactions were
performed in N-methyl-2-pyrrolidinone (NMP) using 4 equiv of
amino acid, 4 equiv of activating reagent (HCTU for natural α-
amino acids; PyBOP for unnatural residues X and Z) and 6 equiv
of DIEA. The microwave coupling program consisted of a 2-min
ramp to 70 °C and a 4-min hold at 70 °C. Removal of the Fmoc
protecting group was achieved by treatment with 20% 4-
methylpiperidine in DMF. The microwave deprotection program
consisted of a 2-min ramp to 80 °C and a 2-min hold at 80 °C.
The resin was washed three times with DMF after each coupling
and deprotection. Once the sequence was completed, the N-ter-
minus was acetylated by treatment with 8 : 2 : 1, v : v : v, DMF :
DIEA : Ac2O, for 10–15 min at room temperature. Following wash-
ing and drying of the resin, cleavage and side-chain deprotection
CD Spectroscopy
Samples of linear peptides 1–4 were prepared in 0.1 M
phosphate buffer at a concentration of 50 μM peptide. Samples
of cyclized peptides 2a–2d, 3b, 3d, and 4a–4d were prepared
in 0.1 M phosphate buffer as described earlier (from 50 μM pep-
tide, 200 μM linker) and analyzed directly after completion of
the reaction as determined by HPLC. CD scans were performed
from 200 to 260 nm with 1-nm step size, 2-nm bandwidth, and
5-s integration time at 20 °C. Raw CD data were smoothed by
the Savitzky–Golay method as implemented in GraphPad Prism
(GraphPad Software, La Jolla, CA, USA).
Linker Mixing Experiments
We carried out several reactions in which a single peptide was
mixed with two different organic linkers, with either simultaneous
wileyonlinelibrary.com/journal/jpepsci Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd. J. Pept. Sci. 2014; 20: 108–114

DYNAMIC COVALENT PEPTIDE CROSS-LINKS
or sequential addition of the small molecules. For simultaneous
addition experiments, a sample of peptide 2, 3, or 4 was prepared,
and linkers b and d added together to create a solution of 50μM
peptide, 150μM b, and 150 μM d in 0.1 M phosphate buffer at
pH 7. For sequential addition experiments, a sample of cyclic pep-
tide 2d was first prepared in 0.1M phosphate buffer from peptide
2 (50 μM) and linker d (200 μM). After complete conversion to 2d
was confirmed by HPLC, linker b (4equiv) was added from an
aqueous stock solution. An identical experiment was carried out
using peptide 4 and linkers b and d. The sequential addition of
linker b diluted the overall concentration of peptide and linkers
slightly versus the simultaneous addition protocol. All the
aforementioned reactions were monitored by HPLC and MALDI-
TOF MS of collected eluent.
analogue of 2 where the nucleophilic side chains have been
replaced with hydrazides by incorporation of residue Z. Fmoc-
protected variants of amino acid residues X [16] and Z (Scheme 1)
were synthesized and applied in standard Fmoc solid-phase
protocols to prepare peptides 2–4. Nucleophilic side chains
in X and Z were masked with 2-chlorotrityl groups during syn-
thesis. Following purification, peptides 2–4 were combined
with linkers a–d in a series of reactions toward various cyclic
products, detailed later. In the nomenclature we employ for
the cyclic peptides, the number corresponds to the peptide
component, and the letter to the organic linker component
(i.e. peptide 2 reacts with linker b to form cyclic product 2b).
We examined ten different binary combinations of peptides
2–4 and linkers a–d to prepare cyclic species 2a–2d, 3b, 3d,
and 4a–4d (Figure 2(B)). Each reaction consisted of 50 μM pep-
tide with 4 equiv of organic linker in pH 7 phosphate buffer and
was monitored by HPLC and mass spectrometry (Figure S1). In
most cases, multiple peaks were observed in the chromatogram
with mass corresponding to product. Each new oxime bond can
be formed with E or Z stereochemical configuration, so cyclic
products containing two oximes can exist as one of four possible
stereoisomers. Our prior work with oxime-containing cyclic
peptides shows that these isomers are typically separable by
HPLC and interconvert slowly under the conditions of the reac-
tion [16]. In reactions of linkers a–d with peptide 2, which has
an i → i + 4 spacing of X residues, the starting peptide was
completely consumed within 2 h to generate cyclic products
2a–2d cleanly. Peptide 3, with X residues spaced farther in
sequence, could be efficiently cross-linked by both small
benzene-based core (b) and larger biphenyl linker (d). Changing
the identity of the nucleophilic side chains to hydrazides (peptide
4) had minimal effect on the efficiency of the cyclization, except
for reaction with biphenyl linker d, which generated a mixture of
linear starting material and cyclic product. The observation that
the ratio of linear precursor 4 to cyclic product 4d did not change
over time suggests that an equilibrium has been established
between the two species. In none of the above reactions was
product observed arising from the formation of two oxime or
hydrazone bonds to the linker utilized from the nucleophilic res-
idues on two independent peptide chains. We attribute this to
the favorable kinetics of the intramolecular cyclization following
the first condensation reaction with linker versus competing
pathways that would cross-link two peptide chains.
Cross-link Exchange with O-Methylhydroxylamine
A sample of cyclic peptide 2b was prepared in 0.1 M phosphate
buffer from peptide 2 (50 μM) and linker b (200 μM). Following
complete conversion, the solution was acidified to pH 4 with
1 M HCl; then O-methylhydroxylamine (10 μl from a 1 M stock in
water) and aniline (10 μl from a 0.1 M stock in water) were added
to yield a sample composed of ~42 μM peptide 2b, residual linker
b (~160 μM), 1000 equiv of O-methylhydroxylamine, and 100
equiv aniline. The reaction was monitored by analytical HPLC at
time points from 1 to 5 h. An identical experiment was performed
using peptide 4 and linker b.
Results and Discussion
We used a 17-residue alanine-rich peptide (1) [27] and three
analogues (2–4) (Figure 2(A)) to explore the synthesis, folding,
and exchange behavior of cyclic peptides prepared via
intermolecular Schiff base formation. Based on peptide 1, sequence
variants 2–4 were designed, synthesized, and applied in a series of
experiments involving four commercially available dialdehyde
small molecules (a–d). In these studies, we sought to examine
how the spacing of the residues, the nature of the Schiff base
(oxime or hydrazone), and the structure of the linker influenced
the synthesis and folding propensity of the cyclic peptide. Peptides
2 and 3 contain two X residues with aminooxy-functionalized side
chains arranged in an i → i + 4 or i → i + 11 spacing. Peptide 4 is an
Figure 2. (A) Chemical structures of peptides 1–4, residues X and Z, and linkers a–d. (B) Peptides 2–4 were used in various combinations with linkers a–
d to generate a family of cyclic products (2a–2d, 3b, 3d, and 4a–4d).
J. Pept. Sci. 2014; 20: 108–114 Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jpepsci

HANEY AND HORNE
One motivation for the synthesis of side-chain cross-linked
peptides is to specify folding behavior in an otherwise poorly
structured chain. Although the focus of the present study is on
synthesis and dynamic exchange of cyclic peptides containing
Schiff base cross-links, we carried out some experiments to exam-
ine how residue spacing and the structure of the linker
influenced folding behavior. Peptide 1, on which the other se-
quences here are based, is partially α-helical in neutral aqueous
solution [27]. In order to examine the impact of various cross-
links on secondary structure, we compared the CD spectrum of
1 with those of linear precursors 2–4 and all the cyclic species
described earlier [Figures 3 and S2 (see Supporting Information)
and Table 1]. Control peptide 1 exhibits modest helicity by CD
as indicated by minima at 208 and 222 nm. Replacement of two
alanine residues with unnatural building blocks X or Z has little
to no impact on the secondary structure content of the
sequence, as seen in the data for linear peptides 2–4. Cyclic
peptides 2a–2d, all based on parent sequence 2, showed
differences in helical content based on the identity of the organic
linker. Cyclic peptide 2b, based on the meta-substituted
dialdehyde linker, showed the greatest helicity among the series.
In contrast to peptide 2 (i → i + 4 spacing of X residues), cycliza-
tion of peptide 3 (i → i + 11 spacing of X residues) had no effect
on the CD signature, regardless of the linker used. This observa-
tion was somewhat surprising, and its origin is not clear. The ef-
fect of linker identity on folding was still apparent but less
pronounced when the aminooxy groups in 2 were changed to
hydrazides (peptide 4); however, 2a and 4a showed similar
decreases helicity compared with their corresponding linear
precursors. Overall, these results suggest that intermolecular
Schiff base cross-links can influence folding; however, placement
of reactive residues and selection of the dialdehyde linker must
be guided by the folded state that is being targeted.
We undertook several experiments to explore the capacity of
the cyclic peptides formed through intermolecular Schiff base
cross-links to participate in dynamic covalent equilibria. In previ-
ous work with peptides cyclized via intramolecular oxime forma-
tion (aminooxy group on one side chain, aldehyde on another),
we observed that the cyclic product is quite resistant to ring
opening by an aminooxy-functionalized small molecule [16]. We
carried out similar competition reactions in the intermolecular
system (Figure 4(A)) by subjecting preformed samples of cyclic
peptides 2b and 4b to a large excess of O-methylhydroxylamine
under conditions known to favor Schiff base exchange (acidic pH,
aniline as a nucleophilic catalyst) [28]. Cyclic peptide 2b remains
unconsumed over several hours, although the ratio of oxime ste-
reoisomers changes (Figure S3). Hydrazones are known to be less
thermodynamically stable than oximes [29]. Thus, we hypothe-
sized that the exchange of the two hydrazones in 4b for two ox-
imes after ring opening to linear precursor 4 and consumption of
the liberated linker b by O-methylhydroxylamine would provide a
driving force for the reaction. The data support this hypothesis
(Figure 4(B)), as cyclic peptide 4b was almost quantitatively
converted back to linear precursor 4 when subjected to the same
conditions where 2b showed no reaction.
The ability to ring open cyclic peptides as described earlier is
perhaps interesting as a proof of concept for dynamic covalent
exchange, but a more compelling question is whether similar
Figure 3. CD scans of peptides 1, 2, 2a, and 2b at 50 μM concentration
in 0.1 M phosphate buffer, pH 7 at 20 °C.
Table 1. CD data for peptides 1–4 and cyclic derivativesa
Peptide
[θ]222 (deg cm2 dmolÀ1 cmÀ1
)
1
À14 300
À13 300
À6 500
2
2a
2b
2c
2d
3
À20 300
À16 900
À8 300
À13 000
À14 300
À14 000
À10 600
À5 500
3b
3d
4
4a
4b
4c
4d
À11 900
À11 200
À11 700
Figure 4. (A) The equilibrium established when peptide 2b or 4b is
treated with O-methylhydroxylamine and aniline at pH 4. (B) HPLC chro-
matogram of cyclic peptide 4b formed from peptide 4 and linker b at
pH 7 before (black) and 1 h after (blue) acidification and addition of
1000 equiv of O-methylhydroxylamine and 100 equiv of aniline.
aMeasurements made on samples of 50 μM peptide in 0.1 M
phosphate buffer, pH 7 at 20 °C.
wileyonlinelibrary.com/journal/jpepsci Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd. J. Pept. Sci. 2014; 20: 108–114

DYNAMIC COVALENT PEPTIDE CROSS-LINKS
principles can be applied to select from linkers in a dynamic
mixture of cyclic products. In order to investigate this issue, we
first carried out competition experiments in which a single linear
peptide was placed in solution with two different linkers
(Figure 5). A sample of peptide 2 was prepared, and 3 equiv of
linkers b and d were added simultaneously. In the resulting mix-
ture, cyclic peptide 2b predominates, and product distribution is
virtually identical whether linker b is added alone (Figure 5(B)) or
linkers b and d are added together (Figure 5(C)). A small amount of
cyclic peptide 2d is observed at early time points but converts
entirely to 2b by 4 h (Figure 5(C)). For linear peptides 3 and 4, sim-
ilar experiments involving simultaneous addition of linkers b and d
yield both possible products in similar quantities [Figure S4 (see
Supporting Information)]. These results suggest that bifunctional
peptides like 2–4 can select from a mixture of dialdehyde linkers
based on differences in the thermodynamic stability of the cyclic
products formed.
the possibility of a dynamic equilibrium. This result was some-
what surprising given the fact that reasonable rates of oxime
exchange in water typically require low pH or nucleophilic cata-
lysts [28,30], whereas the reaction between 2, b, and d was car-
ried out in simple pH 7 phosphate. In order to probe this
behavior further, we carried out a variant of the competition
experiment in which linkers were added sequentially instead of
simultaneously (Figure 5(D)). A solution of peptide 2 was treated
with linker d, and the reaction allowed to proceed until complete
conversion to 2d was observed. An excess of linker b was then
added, and the mixture monitored over time. Over the course
of 5 h, analytical HPLC showed conversion of the initially formed
2d to the presumably more stable product 2b (Figure 5(D)). Thus,
the same end state is achieved whether the linkers b and d are
added simultaneously (Figure 5(C)) or sequentially (Figure 5(D))
to linear peptide precursor 2. This result provides strong support
for the hypothesis that the organic linker component in the cyclic
species can be swapped out in a dynamic covalent manner. The
oxime exchange reaction proceeded faster with added aniline as
a nucleophilic catalyst (completion within 2 vs 5 h). Nevertheless,
the reaction in simple pH 7 phosphate proceeds on a time scale
that would be appropriate for dynamic combinatorial selection
in systems involving peptide folding and/or binding. The ex-
change reaction was also accelerated by changing Schiff bases
involved from oximes to hydrazones, consistent with the known
differences in exchange kinetics between these functional
groups [29]. Thus, addition of linker b to a preformed sample of
cyclic peptide 4d leads to a mixture of 4b and 4d [Figure S5
(see Supporting Information)]. Unlike the behavior in the oxime-
based system, the ratio of 4, 4b, and 4d is different when the
linkers are added simultaneously versus when they are added se-
quentially, perhaps because of small differences in peptide and
linker concentration in the experimental procedures (see Section
on Materials and Methods). This result demonstrates that while
the hydrazone cross-links undergo more rapid exchange, the
equilibrium established is more complex and does not exclu-
sively favor a single product.
The previous observation that the composition of cyclic prod-
ucts in a mixture of 2b and 2d changed over time supported
Conclusions
In summary, we have described here a method for the synthesis
of side-chain to side-chain cyclic peptides based on the
intermolecular reaction between a linear peptide precursor with
two aminooxy or hydrazide side chains and a small molecule
dialdehyde linker. The resulting oxime-based or hydrazone-based
cyclic products are formed rapidly in neutral aqueous solution.
The synthetic approach is modular and convergent. Unprotected
linear peptide precursors can be used in combination with
diverse readily available dialdehydes to create libraries of cyclic
species. Cyclization impacts the folded structure of the peptides,
with differences observed based on residue spacing, cross-link
chemistry, and organic linker identity. The Schiff bases in the
cyclic products can be exploited to create dynamic covalent equi-
libria. Of particular note are the observations that a linear peptide
can select one linker from a mixture and that a preformed cyclic
species can dynamically swap out the small molecule component
of the linker to form the most stable product. The finding that dif-
ferent oxime cross-linked products can exchange at a reasonable
rate in simple pH 7 aqueous buffer recommends the future appli-
cation of the intermolecular cyclization strategy we describe for
the generation of peptide-based dynamic covalent systems.
Figure 5. (A) Equilibrium established when peptide 2, 3, or 4 is reacted
with a mixture of linkers b and d. (B) HPLC chromatogram of peptide 2
before (black) and 1 h after (blue) addition of linker b alone. (C) HPLC
chromatogram of peptide 2 at indicated time points after treatment with
a mixture of linkers b and d added simultaneously. (D) HPLC chromato-
gram of a mixture of peptide 2 and linker d after complete conversion
to 2d (black) and the same solution 5 h after addition of linker b (blue).
All reactions were carried out in 0.1 M phosphate buffer at pH 7.
J. Pept. Sci. 2014; 20: 108–114 Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jpepsci

HANEY AND HORNE
15 Jo H, Meinhardt N, Wu Y, Kulkarni S, Hu X, Low KE, Davies PL, DeGrado
WF, Greenbaum DC. Development of α-helical calpain probes by
mimicking a natural protein–protein interaction. J. Am. Chem. Soc.
2012; 134: 17704–17713.
Acknowledgement
Funding for this work was provided by the University of Pittsburgh.
16 Haney CM, Loch MT, Horne WS. Promoting peptide α-helix formation
with dynamic covalent oxime side-chain cross-links. Chem. Commun.
2011; 47: 10915–10917.
17 Haney CM, Horne WS. Oxime side-chain cross-links in an α-helical coiled-
coil protein: structure, thermodynamics, and folding-templated synthe-
sis of bicyclic species. Chem. Eur. J. 2013; 19: 11342–11351.
18 Rose K. Facile synthesis of homogeneous artificial proteins. J. Am.
Chem. Soc. 1994; 116: 30–33.
References
1 Daly NL, Craik DJ. Bioactive cystine knot proteins. Curr. Opin. Chem.
Biol. 2011; 15: 362–368.
2 White CJ, Yudin AK. Contemporary strategies for peptide
macrocyclization. Nat. Chem. 2011; 3: 509–524.
3 Azzarito V, Long K, Murphy NS, Wilson AJ. Inhibition of α-helix
mediated protein-protein interactions using designed molecules.
Nat. Chem. 2013; 5: 161–173.
4 Jackson DY, King DS, Chmielewski J, Singh S, Schultz PG. General ap-
proach to the synthesis of short α-helical peptides. J. Am. Chem. Soc.
1991; 113: 9391–9392.
19 Canne LE, Ferredamare AR, Burley SK, Kent SBH. Total chemical
synthesis of a unique transcription factor-related protein: cMyc-Max.
J. Am. Chem. Soc. 1995; 117: 2998–3007.
20 Belowich ME, Stoddart JF. Dynamic imine chemistry. Chem. Soc. Rev.
2012; 41: 2003–2024.
5 Blackwell HE, Grubbs RH. Highly efficient synthesis of covalently cross-
linked peptide helices by ring-closing metathesis. Angew. Chem. Int.
Ed. 1998; 37: 3281–3284.
6 Taylor JW. The synthesis and study of side-chain lactam-bridged pep-
tides. Pept. Sci. 2002; 66: 49–75.
7 Pallin TD, Tam JP. Cyclisation of totally unprotected peptides in aque-
ous solution by oxime formation. J. Chem. Soc. Chem. Comun. 1995:
2021–2022.
8 Wahl F, Mutter M. Analogues of oxytocin with an oxime bridge using
chemoselectively addressable building blocks. Tetrahedron Lett. 1996;
37: 6861–6864.
9 Brunel FM, Dawson PE. Synthesis of constrained helical peptides by
thioether ligation: application to analogs of gp41. Chem. Commun.
2005: 2552–2554.
10 Kawamoto SA, Coleska A, Ran X, Yi H, Yang C-Y, Wang S. Design of
triazole-stapled BCL9 α-helical peptides to target the β-catenin/b-
cell CLL/lymphoma 9 (BCL9) protein–protein interaction. J. Med.
Chem. 2012; 55: 1137–1146.
11 Muppidi A, Wang Z, Li X, Chen J, Lin Q. Achieving cell penetration with
distance-matching cysteine cross-linkers: a facile route to cell-permeable
peptide dual inhibitors of Mdm2/Mdmx. Chem. Commun. 2011;
47: 9396–9398.
12 Fujimoto K, Kajino M, Inouye M. Development of a series of cross-
linking agents that effectively stabilize α-helical structures in various
short peptides. Chem. Eur. J. 2008; 14: 857–863.
21 Rowan SJ, Cantrill SJ, Cousins GRL, Sanders JKM, Stoddart JF. Dynamic
covalent chemistry. Angew. Chem. Int. Ed. 2002; 41: 898–952.
22 Corbett PT, Leclaire J, Vial L, West KR, Wietor JL, Sanders JKM, Otto S.
Dynamic combinatorial chemistry. Chem. Rev. 2006; 106: 3652–3711.
23 Dufau L, Marques AS, Fanelli RR, Kihal N, Vidu A, Milcent T, Soulier J-L,
Rodrigo J, Desvergne A, Leblanc K, Bernadat G, Crousse B, Reboud-
Ravaux M, Ongeri S. Carbonylhydrazide-based molecular tongs inhibit
wild-type and mutated HIV-1 protease dimerization. J. Med. Chem.
2012; 55: 6762–6775.
24 Vila-Perelló M, Gallego RG, Andreu D. A simple approach to well-
defined sugar-coated surfaces for interaction studies. ChemBioChem
2005; 6: 1831–1838.
25 Gill SC, Hippel PHv. Calculation of protein extinction coefficients from
amino acid sequence data. Anal. Biochem. 1989; 182: 319–326.
26 McDonald RS, Martin EV. The kinetics and equilibrium of the hydration
of phthalaldehyde. Can. J. Chem. 1979; 57: 506–516.
27 Marqusee S, Baldwin RL. Helix stabilization by Glu-…Lys+ salt bridges
in short peptides of de novo design. Proc. Natl. Acad. Sci. U. S. A. 1987;
84: 8898–8902.
28 Dirksen A, Dirksen S, Hackeng TM, Dawson PE. Nucleophilic catalysis
of hydrazone formation and transimination: implications for dynamic
covalent chemistry. J. Am. Chem. Soc. 2006; 128: 15602–15603.
29 Kalia J, Raines RT. Hydrolytic stability of hydrazones and oximes.
Angew. Chem. Int. Ed. 2008; 47: 7523–7526.
30 Dirksen A, Hackeng TM, Dawson PE. Nucleophilic catalysis of oxime
ligation. Angew. Chem. Int. Ed. 2006; 45: 7581–7584.
13 Zhang FZ, Sadovski O, Xin SJ, Woolley GA. Stabilization of folded pep-
tide and protein structures via distance matching with a long, rigid
cross-linker. J. Am. Chem. Soc. 2007; 129: 14154–14155.
14 Muppidi A, Doi K, Edwardraja S, Drake EJ, Gulick AM, Wang H-G, Lin Q.
Rational design of proteolytically stable, cell-permeable peptide-
based selective Mcl-1 inhibitors. J. Am. Chem. Soc. 2012; 134:
14734–14737.
Supporting Information
Supporting information may be found in the online version of
this article at the publisher’s web site.
wileyonlinelibrary.com/journal/jpepsci Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd. J. Pept. Sci. 2014; 20: 108–114