Photo-control of peptide conformation on a timescale of seconds with a conformationally constrained, blue-absorbing, photo-switchable linker

Article abstract of DOI:10.1039/b810533b

Azobenzene derivatives can be used to reversibly photo-regulate conformation and activity when introduced as intramolecular bridges in peptides and proteins. Here we report the design, synthesis, and characterization of an azobenzene derivative that absorbs between 400-450 nm in aqueous solution to produce the cis isomer, and relaxes back to the trans isomer with a half-life of a few seconds at room temperature. In the trans form, the linker can span a distance of approximately 25 A, so that it can bridge Cys residues spaced i, i + 15 in an α-helix. Switching of the azobenzene cross-linker from trans to cis causes a decrease in the helix content of peptides where the linker is attached via Cys residues spaced at i, i + 15 and i, i + 14 positions, no change in helix content with Cys residues spaced i, i + 11 and an increase in helix content in a peptide with Cys residues spaced at i, i + 7. In the presence of 10 mM reduced glutathione, the azobenzene cross-linker continued to photo-switch after 24 hours. This cross-linker design thus expands the possibilities for fast photo-control of peptide and protein structure in biochemical systems.

Full text of DOI:10.1039/b810533b

View Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Photo-control of peptide conformation on a timescale of seconds with a
conformationally constrained, blue-absorbing, photo-switchable linker
Andrew A. Beharry, Oleg Sadovski and G. Andrew Woolley*
Received 24th June 2008, Accepted 22nd August 2008
First published as an Advance Article on the web 24th September 2008
DOI: 10.1039/b810533b
Azobenzene derivatives can be used to reversibly photo-regulate conformation and activity when
introduced as intramolecular bridges in peptides and proteins. Here we report the design, synthesis, and
characterization of an azobenzene derivative that absorbs between 400–450 nm in aqueous solution to
produce the cis isomer, and relaxes back to the trans isomer with a half-life of a few seconds at room
˚
temperature. In the trans form, the linker can span a distance of approximately 25 A, so that it can
bridge Cys residues spaced i, i + 15 in an a-helix. Switching of the azobenzene cross-linker from trans
to cis causes a decrease in the helix content of peptides where the linker is attached via Cys residues
spaced at i, i + 15 and i, i + 14 positions, no change in helix content with Cys residues spaced i, i + 11
and an increase in helix content in a peptide with Cys residues spaced at i, i + 7. In the presence of
10 mM reduced glutathione, the azobenzene cross-linker continued to photo-switch after 24 hours. This
cross-linker design thus expands the possibilities for fast photo-control of peptide and protein structure
in biochemical systems.
the fraction of cis isomer can be changed from 5% to 80% (16-
Introduction
fold).12 Thermal isomerization, in contrast, yields >99.99% trans
isomer.7 Thus, to maximize the extent of photo-switching, it is
preferable to use thermal isomerization to reset the switch; in that
way, a much greater fold change in the cis isomer is possible. Since
many cellular signaling processes operate on the time frame of
seconds to minutes (e.g. ref. 13,14), the switch resetting process
should occur at least on this timescale. The half-life for thermal
cis-to-trans isomerization of unmodified azobenzene is on the
order of 3–4 days at room temperature; however, this can be
varied substantially by altering the nature of substituents on the
aromatic rings (e.g. ref. 15–18). For instance, an amino-substituted
azobenzene compound relaxes with a half-life of ~50 ms at room
temperature.19
Reversible optical control of protein structure offers the pos-
sibility of remote control of protein function in a variety of
environments.1–5 The azobenzene chromophore exhibits numerous
desirable properties for use as an optical switch including robust
reversibility and good quantum yields.6 In applying azobenzenes
to direct protein conformational change in biochemical systems,
a number of requirements must be met. These include (i) a
substantial structural change associated with isomerization that
can be coupled to protein conformational change, (ii) a suitable
wavelength and rate of thermal relaxation, (iii) stability in a cellular
environment.
The effectiveness of conformational control will depend to
a large degree on the end-to-end distance of the chromophore
and how this changes upon isomerization. Certain urea and
carbamate-substituted azobenzene derivatives, in particular, can
accommodate a wide range of end-to-end distances in their cis
isomers,7 a feature that may limit their ability to affect the
conformation of an attached peptide or protein.8 A minimum
number of bonds or a conformational constraint between the azo
moiety that isomerizes and the peptide/protein backbone is likely
to maximize the conformational switching effect.9,10
For unmodified azobenzene, the wavelength of maximal absorp-
tion of the more stable trans isomer occurs at 340 nm.6 Light of
this wavelength is scattered strongly by biological samples and
can be absorbed by cellular constituents (e.g. NADH). Longer
wavelengths lead to less light scattering and better penetration
in biological samples.11 Since the absorption spectra of trans and
cis forms of azobenzenes overlap extensively, irradiation typically
produces photostationary states of ~80% cis or ~95% trans.6,12
This fact limits the degree of photo-switching that is possible; e.g.
Finally, the photo-switch must be chemically stable under the
desired operating conditions. Whereas azobenzenes are generally
robust photo-switches, if they are to operate intracellularly, they
must be stable to the reducing environment created by the
presence of glutathione at millimolar concentrations.20,21 Such an
environment precludes the use of disulfide linkages for attaching
the photo-switch as was used previously.19 In addition, the N N
=
double bond itself may be sensitive to reduction in certain cases.22,23
With these considerations in mind, we designed the thiol-
reactive azobenzene-based photo-switch 5. We report here the
synthesis and characterization of this new photo-switch.
Results and discussion
Design and synthesis of the cross-linker
We wished to create a long-wavelength absorbing, conforma-
tionally constrained, cross-linker to permit effective coupling
of the geometrical change at the azo group with the peptide
backbone. Linkages of photo-switches to Cys residues (as opposed
to Lys (e.g. ref. 24)) ensure a minimum number of single bonds
Department of Chemistry, University of Toronto, 80 St. George St., Toronto,
ON, M5S 3H6, Canada. E-mail: awoolley@chem.utoronto.ca; Fax: (416)
978-0675; Tel: (416) 978-0675
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4323–4332 | 4323
©

View Online
(four rotatable sigma bonds) between the cross-linker and the
backbone. So that it might be stable in the reducing intracellular
environment, we wished the linker to produce thioether linkages,
rather than disulfides,19 with peptide Cys residues. Based on
previous studies,18,19 we expected that an azobenzene derivative
with a dialkylamino substituent in the para position would exhibit
a wavelength of maximal absorption substantially longer than
alkyl and amide substituted azobenzene derivatives, together with
a faster thermal relaxation rate. These considerations led us to the
design of 5, which was synthesized as outlined in Scheme 1.
Scheme 1
The yield of the first step was excellent, however, yields for step
two and three were poor (7%) and modest (33%), respectively.
Although protection of the amino group in (2) would be expected
to increase the yield in the second step, isolation of 4 by silica gel
chromatography was straightforward and the direct route avoided
two extra steps (protecting group introduction and removal).
The cross-linker (5) was then reacted with the peptides shown
in Fig. 1. The peptide sequences including the incorporation
of the non-natural Aib residue in peptide JRK-7 were designed
as described previously.7 Reaction with AB-15, FZ-14, FK-11W
and JRK-7, produces an intramolecular cross-link between Cys
residues spaced i, i + 15, i, i + 14, i, i + 11, and i, i + 7,
respectively. As a control, the cross-linker was also reacted with
two molecules of the small tripeptide glutathione, which is too
short to exhibit substantial secondary structure in solution. The
cross-linked peptides were purified by HPLC and characterized by
mass spectrometry.
Fig. 1 Primary sequences of the cross-linked peptides examined in this
study: (a) AB-15 (azo-linker 5 reacted via Cys residues spaced i, i + 15);
(b) FZ-14 (azo-linker 5 reacted via Cys residues spaced i, i + 14); (c)
FK-11W (azo-linker 5 reacted via Cys residues spaced i, i + 11); (d) JRK-7
(azo-linker 5 reacted via Cys residues spaced i, i + 7); (e) cross-linked
glutathione.
UV–Vis spectra: solvent and pH dependence
Fig. 2 shows a representative UV–Vis spectrum of the cross-
linker attached to a peptide (FK-11W) obtained in aqueous buffer
(pH 7.0). A broad maximum near 450 nm is observed that is shifted
to shorter wavelengths relative to the unconstrained diamino-
substituted azobenzene studied previously (lmax ~480 nm).19
Hallas et al.25,26 have studied the effects of cyclic vs. acyclic
alkylamino substituents on azobenzene dyes (e.g. compounds 6,
7) and reported a similar hypsochromic shift in the UV absorbance
of the cyclic compound (7).
Molecular modeling of 7 using the AM1 method shows that
the piperidine ring is twisted relative to the phenyl ring and
the nitrogen exhibits sp3 character (Fig. 3). This conformation
is preferred to reduce steric interactions between the equatorial
protons on the piperidine ring with the ortho protons on the phenyl
ring.25,27 Rotation of the piperidine ring reduces the extent of
p-overlap between the lone pair on the N atom and the adjacent
aromatic p-electrons thereby bringing about a hypsochromic shift
of lmax
.
25,26 A decreased resonance interaction between the N-atom
4324 | Org. Biomol. Chem., 2008, 6, 4323–4332
This journal is The Royal Society of Chemistry 2008
©

View Online
been ascribed to self-association of azo dyes,29 we found a linear
absorbance vs. concentration relationship over the concentration
ranges examined, no dependence of the spectra on temperature
(up to 80 ◦C in aqueous buffer), and no effect of cross-linked
peptide concentration on thermal relaxation kinetics (see below).
Fig. 4 shows UV spectra of the cross-linker in aqueous buffer
at a series of pH values from 1 to 7. At pH values of 5 and above,
only the single broad maximum near 450 nm was observed. As the
pH is lowered below 5, a second absorption band emerges near
600 nm. Based on previous studies with aminoazobenzenes,26,27 the
protonated cross-linker is expected to be in tautomeric equilibrium
between the ammonium and azonium species (Scheme 2); the latter
accounting for the absorbance near 600 nm.
Fig. 2 UV–Vis spectrum of the cross-linker attached to the peptide
FK-11W in 10 mM sodium phosphate buffer, pH 7.0, 22 ◦C.
Fig. 4 UV–Vis spectra of cross-linker (attached to glutathione) in
aqueous buffer (~15 mM) at a series of pH values.
Fig. 3 Models of compounds (6) (A) and (7) (B, C). Whereas rotation
of diethylamino groups in 6 can permit orbital overlap between the N
atom lone pair and the aromatic system (A), for 7, the piperidine ring
structure leads to steric overlap of ortho protons if the N-atom is fully
sp2 hybridized (steric clashes are indicated by red dashed lines) (B). Steric
overlap is relieved and the N atom becomes more sp3-like as the piperidine
ring is rotated relative to the aromatic system in the AM1 minimized
structure (C).
Protonated forms of the cross-linker do not appear to isomerize.
Thus, the effective pH range of this cross-linker is pH 5 and above.
Since most biological systems of interest operate in the region of
pH 7, this pH sensitivity is not likely to be problematic.
Photokinetic analysis: calculation of the absorption spectrum of
the cis isomer
and the aromatic system for the cyclic compound is supported by
13C-NMR shift data.28
The UV–Vis spectrum of the cis isomer could not be obtained
directly due to the fast rate of thermal cis-to-trans relaxation.
Therefore, the absorption spectrum of the cis isomer was cal-
culated using the photokinetic method previously employed for
diamidoazobenzenes.30 Photokinetic curves were obtained upon
irradiation of cross-linked AB-15 in 50% methanol at 2 ◦C as
described in the Experimental section. These conditions were
The UV–Vis spectrum of the cross-linker changes somewhat
in shape and intensity depending on which peptide it is attached
to. Addition of co-solvents (methanol/propanol/acetonitrile) is
also observed to change the shape and intensity of the spectrum
(cf. Fig. 2 vs. Fig 5). Although such changes have previously
Scheme 2
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4323–4332 | 4325
©

View Online
chosen to maximize the % cis isomer that formed and to permit
comparison with CD obtained under the same solvent conditions
(see below). Fig. 5 shows the calculated spectrum of the pure cis
isomer, together with the directly measured spectrum of the trans
isomer. The cis spectrum shows a minimum at 400 nm, where the
spectrum of the trans isomer is at a maximum. The wavelength
of 400 nm is thus optimal for irradiation when the maximum %
cis is desired. Based on the estimated trans → cis and cis → trans
quantum yields obtained from the photokinetic analysis (approx.
0.1 and 0.5 respectively), a maximum of ~50% cis isomer is possible
with 400 nm radiation of sufficient intensity.
fit to exponential decay curves and time constants calculated. The
data are collected in Table 1.
In aqueous solution, half-lives on the order of a few seconds
were obtained (Table 1). Depending on which peptide the cross-
linker is attached to, the observed half-life varies by approximately
a factor of 20. For X-JRK-7, a slow and a fast process were
observed, suggesting the presence of more than one cis-state
conformation. Multiple cis-state conformations have previously
been observed by Renner et al. in studies with a cyclic azobenzene-
containing peptide.31 In the present case, it is possible that these
correspond to different twist senses of the azo group relative to
the peptide.32 Half-lives did not appear to be sensitive to dilution,
indicating that multiple time constants were not a result of self-
association of peptides. With increasing percentages of methanol,
the half-lives of the cross-linked peptides increased and mono-
exponential relaxation behavior was observed (Table 1). Decreased
solvent polarity may slow relaxation by destabilizing the transition
state for isomerization, which has been proposed to have dipolar
character similar to that in the excited state.33 Thus, the local
environment created around the photo-switch when attached to
a peptide or protein is an important factor in determining the
timescale of switching, a finding consistent with the observed
variability of half-lives for different peptides (Table 1). Overall,
the half-lives observed are somewhat longer than that observed for
the acyclic dialkyl amino cross-linker studied previously19 (where
a ~50 ms half-life was observed in water and 1–20 s in 25–50%
MeOH), consistent with a lesser degree of delocalization of the
system (as discussed above). A half-life of ~1 s is nevertheless
sufficiently rapid that ~100% trans isomer can be obtained, ~10–
20 s after turning off irradiation, a timescale fast enough to permit
probing of a wide variety of biochemical processes.
Fig. 5 The calculated absorption spectrum of the cis isomer (dashed line)
of cross-linked AB-15 in 50% methanol in 10 mM sodium phosphate buffer
pH 7.0. The spectrum of the trans isomer (solid line) in the same solvent
is shown for reference.
Effect of glutathione on photo-switching
Kinetics of thermal relaxation
In order to be of use for intracellular applications, photochemical
switches must be stable to the reducing environment inside the
cell. Typically, this environment is maintained by the small
tripeptide glutathione present in its reduced form in ~1 mM
concentrations.20,21 Cross-linkers bearing disulfide linkages will be
reductively cleaved rapidly under such conditions.19 Glutathione
is also known to reduce azo groups under certain conditions.22,23
Moroder and colleagues have recently demonstrated that another
azobenzene based peptide photo-switch was susceptible to re-
duction by glutathione.23 The mechanism of reduction appears
to involve attack of the thiol group of glutathione on the azo
double bond to form a sulfenyl hydrazide adduct. This species
If the cross-linker can relax to the trans isomer rapidly via a thermal
process, essentially 100% trans isomer can be quickly obtained in
the dark, thereby maximizing the change in concentration of cis
isomer upon switching (i.e. <0.01% cis to 50% cis upon irradiation
at 400 nm).
Previously, we found the dialkylamino-substituted azobenzene
to be sensitive to changes in solvent polarity.19 We therefore
wished to test the effect of methanol on the rate of thermal
relaxation. cis Isomers were produced by exposure to 400 nm
light. Thermal relaxation to the trans isomer was measured by
monitoring recovery of 450 nm absorbance over time. Data were
Table 1 Thermal relaxation half-lives (s) of cross-linked peptides
In 10 mM phosphate buffer, pH 7.0
+25% MeOH
2 ◦
+50% MeOH
2 ◦
Peptide
2 ◦
C
25 ◦
C
C
25 ◦
C
C
25 ◦C
XAB15
XFZ14
XFK11
XJRK7
XGSH
4.0
7.0
5.0
1.8
3.8
1.7
9.0
8.2
22
84
2.6
4.0
5.8
7.7
19
1.0
34
8
18
40
197
6.7
53
165
700
22
6(0.9)/91(0.1)a
0.5
2.3(0.74)/83(0.26)a
—
a X-JRK-7 decays in phosphate buffer were fit to double exponential functions. Fast and slow time constants are shown with the fractional contribution
of each process in brackets.
4326 | Org. Biomol. Chem., 2008, 6, 4323–4332
This journal is
The Royal Society of Chemistry 2008
©

View Online
can react with a second molecule of glutathione to produce
oxidized glutathione and the reduced hydrazo compound as
products.22,23 Alternatively, the sulfenyl hydrazide adduct may
dissociate to reform the initial azo species. The rates of these
reactions willdependonthe glutathione concentration, the peptide
concentration, as well as the redox potential of the cross-linker.
The amino-substituted cross-linker studied here is expected to be
less sensitive to reduction than the compound studied by Moroder
and colleagues.23
The effect of glutathione on the photo-switching of the cross-
linker attached to the peptide JRK-7 was studied in aqueous buffer
pH 7.0 at 37 ◦C with 10 mM glutathione to mimic the intracellular
reducing environment. This concentration of glutathione is at the
high end of that normally found in cells.20,21 This peptide was
chosen because it had a relaxation rate slow enough to be measured
conveniently under these solution conditions (Table 1). At 37 ◦C,
X-JRK-7 displays a double exponential thermal relaxation process
with half-lives of 1 s (55% of total decay) and 20 s (45%). The
peptide was incubated with glutathione in the dark for 24 hours at
37 ◦C, to mimic conditions employed for loading cells in culture.
Fig. 6 shows thermal relaxation and photoisomerization of cross-
linked JRK-7 as monitored by measuring absorbance vs. time
for 3 h after the 24 h dark incubation. Photo-switching appeared
to be essentially unaffected (1 s (50% of total decay) and 22 s
(50%)) (Fig. 6), probably as a result of the more electron-rich
character of this azobenzene derivative compared to those studied
previously.10,23 The persistence of photo-switching for hours, even
in the presence of 10 mM glutathione, indicates that the cross-
linker should be functional in typical intracellular environments.
Conformational analysis of the cross-linker and effects
of photoisomerization
Fig. 7 (a) Probability distributions of distances between sulfur atoms in
models of the cross-linker derived from the molecular dynamics simula-
tions using the Amber forcefield (shaded areas), as well as conformational
searches using the AM1 method (histograms) (cis (red), trans (blue)).
Distances between S atoms within Cys residues spaced i, i + x in ideal
a-helices are also indicated with boxes. The centre of the box is placed
at the most probable distance and the box width indicates the range of
distances possible, considering Cys side chain rotamers. (b) Representative
structures and S–S distances observed (trans—top, cis—bottom).
To estimate the effects of the cross-linker on the conformational
properties of attached peptides or proteins, we first examined the
conformational properties of the cross-linker itself in cis and trans
conformations. The range of end-to-end distances exhibited by
the cross-linker provides a good first approximation for predicting
its effect on conformational ensembles of helical peptides.8,34
Simplified models of 5 after reaction with Cys residues of a peptide
(where the b position of a cysteine side-chain is represented by
a methyl group) in cis and trans conformation were built using
HyperChem.
conformational energies, the trans isomer has a most probable
˚
˚
end-to-end distance of ~21–23 A (and a range from 16 to 26 A).
The AM1 method tends to produce more sp3-like tetrahedral
geometry for the nitrogen atoms attached to the para positions
of the azobenzene group when compared to the Amber forcefield,
resulting in somewhat shorter end-to-end distances. Substantial
Fig. 7 shows histograms of the end-to-end (S–S) distances for
cis and trans isomers, together with some representative structures.
Depending on whether the Amber molecular mechanics forcefield
or the semiempirical AM1 method is employed for calculating
Fig. 6 Photo-switching of ~5 mM X-JRK-7 in phosphate buffer, pH 7.0, 37 ◦C with 10 mM reduced glutathione, as monitored by absorbance changes
at 400 nm over the course of 3 h. Irradiation at 400 nm was present during the periods indicated by solid bars.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4323–4332 | 4327
©

View Online
sp3 character for these atoms is consistent with the spectral data
JRK-7 is the least helical in the dark, since the trans form of the
linker is not compatible with the helical structure in this peptide.
Based on observed effects with previous cross-linkers, one
would predict that the trans form of the cross-linker (5) would
be compatible with a helical conformation of AB-15 and FZ-14
(Fig. 7) and that isomerization from trans to cis would lead to a
decrease in helical content. Conversely, a trans conformation of
the cross-linker should be incompatible with the standard helix
formation for JRK-7, but isomerization from trans to cis should
make helix formation more likely. For FK-11W the trans and cis
forms of the cross-linker appear to be about equally compatible
with helical structure.
In order to test these predictions, we performed CD studies of
the cross-linked peptides JRK-7, FK-11W, FZ-14 and AB-15 in
pH 7.0 phosphate buffer–methanol (50 : 50), as described in the
Experimental section. Although the co-solvent is likely to affect
the % helix content, the same overall conformational behaviour
is expected as in aqueous solution.35 Under these mixed solvent
conditions, the cross-linked peptides isomerize slowly enough to
monitor using a conventional CD instrument. Photoisomerization
produced ~45% of the cis isomer under these conditions, as
determined from UV–Vis experiments. Fig. 10 shows the effects
of photoisomerization on peptide conformation. As expected,
relaxation from cis to trans after irradiation increases helicity for
the cross-linked AB-15 and FZ-14 peptides, as evidenced by a
stronger negative CD signal at 225 nm. Conversely, relaxation
from cis to trans after irradiation decreases helicity for the JRK-
7 cross-linked peptide, as evidenced by a weaker negative CD
signal. No change in CD signal at 225 nm was observed for cross-
linked FK-11W upon photoisomerization. The half-lives of cis
forms of cross-linked peptides observed in the CD experiments
are comparable to the half-lives estimated from the UV data in
Table 1.
described above. The cis isomer shows a broader distribution of
˚
end-to-end distances with a most probable distance of ~16 A, but
˚
is compatible with distances from ~4 to ~21 A.
For the trans form particularly, these end-to-end distances are
significantly longer than for the diamino- and diamido-substituted
azobenzenes studied previously so that this cross-linker can be
used to span Cys residues more distant from one another in a
peptide. Although the cyclic alkyl moiety limits the flexibility of
the cross-linker compared to an analogous acyclic version, the
molecular modeling suggests that there is still substantial flexibility
associated with the cycloalkane rings so that a large range of end-
to-end distances is still possible, especially for the cis isomer. An
even more rigid derivative will perhaps be required to maximize the
difference in end-to-end distance between cis and trans isomers.
Fig. 7(a) shows S–S spacing for Cys residues spaced i, i + 15, i, i
+ 14, i, i + 11, and i, i + 7, in an ideal a-helix. The trans distribution
of the cross-linker fits best with the a-helical conformation of AB-
15 (i, i + 15 ) and FZ-14 (i, i + 14) and least well with the JRK-7 (i,
i + 7) peptide. Molecular modeling confirms the compatibility of
the trans form of the cross-linker with a helical structure of FZ-14
and of the cis form of the cross-linker with a helical structure for
JRK-7 (Fig. 8).
Fig. 8 Molecular models of (A) FZ-14 in an a-helical conformation with
the trans form of the cross-linker attached; (B) JRK-7 in an a-helical
conformation with the cis form of the cross-linker attached.
For cross-linked JRK-7, the half-life for thermal relaxation in
the absence of methanol co-solvent is slow enough to permit CD
kinetic measurements; these data are shown in Fig. 11. At low
temperatures, the relaxation is dominated by a single decay process
with a time constant near 5 s seen in both the UV data (Table 1) and
the kinetic CD data (Fig. 11). Although the overall helix content
is decreased in the absence of the co-solvent, the same trend is
observed, i.e. cis-to-trans thermal relaxation leads to a decrease
in helix content as evidenced by a weaker negative CD signal at
225 nm.
Fig. 9 shows CD spectra of uncross-linked and dark-adapted
cross-linked peptides in aqueous buffer pH 7.0 at room tem-
perature. As expected, cross-linked AB-15 and FZ-14 show the
strongest helicity since the trans form of the linker is compatible
with the helical structure in these peptides. In contrast, cross-linked
Summary
We have designed and synthesized a thiol-reactive azobenzene
cross-linker that undergoes trans-to-cis photoisomerization upon
irradiation at 400–450 nm and relaxes thermally with a half-life
of about 1 s in aqueous solution. The cross-linker can be used
to reversibly control the helical content of attached peptides. Its
effects on helix content can be predicted by comparing the range
of S–S distances in the intrinsic conformational ensemble of the
peptide with the range compatible with the chemical structure of
the linker. The trans form of the cross-linker is compatible with
Cys residues spaced i, i + 15 and i, i + 14 in an a-helix, whereas
the cis form of the cross-linker is compatible with Cys residues
spaced i, i + 7. Photo-switching is maintained for many hours
under reducing conditions similar to those found inside cells.
Fig. 9 CD spectra of uncross-linked (A) and dark-adapted cross-linked
(B) peptides in aqueous phosphate buffer pH 7.0 at room temperature.
AB-15 (—), FZ-14(– – –), FK-11W (– - - –), JRK-7 (– - –).
4328 | Org. Biomol. Chem., 2008, 6, 4323–4332
This journal is
The Royal Society of Chemistry 2008
©

View Online
Fig. 10 Time-dependent circular dichroism (CD) signals at 225 nm after irradiation of (A) cross-linked AB-15 (~17 mM, 1 cm cell, 6 ◦C (left) 25 ◦C
(right)); (B) cross-linked FZ-14 (~12 mM, 1 cm cell, 6 ◦C (left) 25 ◦C (right)); (C) cross-linked FK-11W (~17 mM, 1 cm cell, 6 ◦C (left) 25 ◦C (right)); (D)
cross-linked JRK-7 (~37 mM, 1 cm cell, 6 ◦C (left) 25 ◦C (right)); solvent conditions: 50% methanol in phosphate buffer, pH 7.0.
Experimental
All chemicals were purchased from Sigma-Aldrich Chemical Co.,
1
except if specified otherwise. All H and 13C NMR spectra were
recorded using a Varian Unity 400 or a Varian Gemini 300
spectrometer. High resolution mass spectra were obtained either
by electron impact (EI) or electrospray (ESI) ionization. Peptide
mass spectra were obtained by MALDI ionization.
Synthesis of the cross-linker
Bis-(2-chloroethyl)-amine hydrochloride (2). 5 g (0.048 mol) of
diethanolamine (1) were dissolved in 20 mL of chloroform and
added dropwise with stirring to a solution of 14.15 g (0.11 mol) of
thionyl chloride in 15 mL of chloroform. When all the amine had
Fig. 11 Time-dependent circular dichroism (CD) signals at 225 nm after
irradiation of cross-linked JRK-7 (~58 mM, 1 cm cell, 6 ◦C); in phosphate
buffer, pH 7.0.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4323–4332 | 4329
©

View Online
been added, the volatile species were distilled off and the remaining
material was recrystallized from acetone to give 2 (yield: 90%).
reduced state. The solution was stirred and 350 mL of a 5.7 mM
solution of the linker in DMSO were added dropwise. Although
DMSO is a mild oxidant, no effect on cross-linking was observed.
The mixture was heated to 40 ◦C and left to stir for 24 h, protected
from light. The modified peptides were purified by HPLC (SB-
C18 column, as described above) using a linear gradient of
5–57% acetonitrile–H2O (+0.1% trifluoroacetic acid) over the
course of 31 min (elution at 50% acetonitrile). The modified
peptide compositions were confirmed by MALDI-MS [M+]:
calc’d for C110H166N38O31S2 (cross-linked FZ-14) = 2563.8 Da;
obs’d = 2564.1 Da; calc’d for C110H166N38O31S2 (cross-linked
AB-15) obs’d = 2562.8 Da.
Bis-(4-piperazin-1-yl-phenyl)-diazene (4). A mixture of 2.5 g
(0.012 mol) 4,4¢-azobis-benzeneamine (3) (Alfa Aesar), 4.42 g
(0.025 mol) 2, 3.41 g (0.262 mol) potassium carbonate, 1.05 g
(0.007 mol) sodium iodide and xylene (100 mL) was heated at
140–145 ◦C for 1 h. The cooled reaction mixture was poured into
200 mL water and then extracted with ethyl acetate (5 ¥ 100 mL).
The combined organic extracts were dried (Na2SO4), filtered, and
the solvent was removed under reduced pressure to afford 1.8 g of
a dark oil. Column chromatography on silica using ethyl acetate–
1
methanol yielded 0.28 g of 4 (yield: 7%). H-NMR (400 MHz,
Intramolecular cross-linking of Cys residues in FK-11W was
performed as follows: 3.0 mg (1.5 mmol) of FK-11W and
0.43 mg (1.5 mmol) of tris(carboxyethyl)phosphine were dissolved
in 300 mL of 15 mM Tris·Cl buffer (pH 8) and incubated for 1 h at
room temperature to ensure cysteine residues were in their reduced
state. The solution was stirred and 700 mL of a 4.2 mM solution of
the linker in DMSO were added dropwise. The mixture was heated
to 40 ◦C and left to stir for 24 h, protected from light. The modified
peptide was purified by HPLC (SB-C18 column, as described
above) using a linear gradient of 5–80% acetonitrile–H2O (+0.1%
trifluoroacetic acid) over the course of 45 min (elution at 49%
acetonitrile). The modified peptide composition was confirmed
by MALDI-MS [M+]: calc’d for C104H156N36O29S2 = 2421.6 Da;
obs’d = 2421.1 Da.
DMSO-d6) d ppm 2.80–2.85 (m, 8H, 4 CH2), 3.15–3.25 (m, 8H, 4
CH2), 6.80 (s, 2H, 2NH), 7.00 (d, 9.4 Hz, 4H, 4CH), 7.70 (d, 9.4 Hz,
4H, 4CH). HRMS-ESI calc’d (MH+)(C20H27N6) 351.2291; obs’d
351.2278.
2-Chloro-1-[4-(4-{4-[4-(2-chloroacetyl)-piperazin-1-yl]phenyl-
azo}-phenyl)-piperazin-1-yl]-ethanone (5). To
a solution of
0.028 g (0.8 mmol) of 4 and 0.03 mL (0.28 mmol) of triethylamine
in 4 mL CH2Cl2 was slowly added 0.017 mL (0.21 mmol) of
chloroacetyl chloride. The reaction mixture was stirred overnight
and the solvent was removed under reduced pressure. The residue
was purified by column chromatography on silica (CH2Cl2–
1
EtOAc) to give 0.16 g of 5 (yield: 33%). H-NMR (300 MHz,
DMSO-d6) d ppm 3.30–3.35 (m, 4H), 3.40–3.45 (m, 4H), 3.60–
3.65 (m, 8H), 4.4 (s, 4H), 7.1 (d, 9.1 Hz, 4H), 7.7 (d, 9.1 Hz,
4H). HRMS-ESI calc’d (MH+)(C24H29Cl2N6O2) 503.1723; obs’d
503.1702.
Intramolecular cross-linking of Cys residues in JRK-7 was
performed as follows: 1.0 mg (0.6 mmol) of JRK-7 and 0.17 mg
(0.6 mmol) of tris(carboxyethyl)phosphine were dissolved in
150 mL of 50 mM Tris·Cl buffer (pH 8) and incubated for 1 h at
room temperature to ensure cysteine residues were in their reduced
state. The solution was stirred and 350 mL of a 3.4 mM solution
of the linker in DMSO were added dropwise. The mixture was
heated to 40 ◦C and left to stir for 24 h. The modified peptide was
purified by HPLC (SB-C18 column, as described above) using a
linear gradient of 5–57% acetonitrile–H2O (+0.1% trifluoroacetic
acid) over the course of 31 min (elution at 50% acetonitrile).
The modified peptide composition was confirmed by MALDI-
MS [M+]: calc’d for C89H140N30O25S2 = 2077.3 Da; obs’d =
2075.9 Da.
A glutathione derivative of the linker was prepared as follows:
100 mL of a 13 mM solution of glutathione in 50 mM Tris·Cl
buffer (pH 8) was stirred and 50 mL of an 8.5 mM solution of the
linker in DMSO were added. The mixture was heated to 40 ◦C
and left to stir for 24 h, protected from light. The glutathione
derivative was purified by HPLC (SB-C18 column, as described
above) using a linear gradient of 5–50% acetonitrile–H2O (+0.1%
trifluoroacetic acid) over the course of 27 min (elution at 45%
acetonitrile). The glutathione derivative molecular composition
was confirmed by MALDI-MS [M+]: calc’d for C44H60N12O14S2 =
1045.1 Da; obs’d = 1044.3 Da.
Peptide synthesis
Standard Fmoc-based solid-phase peptide synthesis was used
to prepare the peptides AB-15 (Ac-WGCAEAAAREAAAREA-
ACRQ-NH2), FZ-14 (Ac-WGACEAAAREAAAREAACRQ-
NH2) and FK-11W (Ac-WGEACAREAAAREAACRQ-NH2).
The peptide JRK-7 (Ac-EACARVAibAACEAAARQ-NH2) was
purchased from Jerini Biotools (Berlin, Germany). The peptides
AB-15, FZ-14 and FK-11 were constructed on PAL-resin (capacity
0.60 mmol g-1). Coupling used 4 equivalents HATU, 6 equivalents
DIPEA and 4 equivalents amino acid to give peptides AB-
15 (yield: 52%), FZ-14 (yield: 55%) and FK-11W (yield: 50%).
Peptides were purified by HPLC on a semipreparative SB-C18
column (Zorbax, 9.4 mm ID ¥ 24 cm) using a linear gradient of 5–
70% acetonitrile–H2O (+0.1% trifluoroacetic acid) over the course
of 30 min: AB-15 and FZ-14 eluted at 48% acetonitrile; FK-11W
eluted at 47% acetonitrile; JRK-7 eluted at 50% acetonitrile. The
peptides’ molecular compositions were confirmed by MALDI-MS
[M+]: AB-15 (C86H140N32O29S2) calc’d: 2131.0 Da; obs’d: 2131.8
Da, FZ-14 (C86H140N32O29S2) calc’d: 2131.0 Da, obs’d: 2131.5 Da;
FK-11W (C80H130N30O27S2) calc’d: 1988.9 Da, obs’d: 1988.2 Da;
JRK-7 (C65H114N24O23S2) calc’d: 1644.8 Da, obs’d: 1644.4 Da.
Peptide cross-linking
UV–Vis spectra and photoisomerization
Intramolecular cross-linking of Cys residues in AB-15 and FZ-14
was performed as follows: 2.1 mg (1.0 mmol) of AB-15 or FZ-14
and 0.29 mg (1.0 mmol) of tris(carboxyethyl)phosphine were
dissolved in 150 mL of 50 mM Tris·Cl buffer (pH 8) and incubated
for 1 h at room temperature to ensure cysteine residues were in their
Ultraviolet absorbance spectra were obtained using either a
Perkin-Elmer Lambda 2 spectrophotometer or using a diode
array UV–Vis spectrophotometer (Ocean Optics Inc., USB4000)
coupled to a temperature controlled cuvette holder (Quantum
Northwest, Inc.). Measurements of thermal relaxation rates and
4330 | Org. Biomol. Chem., 2008, 6, 4323–4332
This journal is
The Royal Society of Chemistry 2008
©

View Online
photokinetic experiments used the latter arrangement. Irradiation
of the sample (at 90◦ to the light source and detector used for
the absorbance measurements) was carried out using a xenon
lamp (450 W, Osram 450XBO/2 OFR) coupled to a double
monochromator. Rates of thermal cis-to-trans isomerization were
measured for a series of temperatures by monitoring absorbance
at 450 nm after irradiation to convert a percentage of the solution
to the cis isomer. The light used for the absorbance measurement
was of sufficiently low intensity to cause negligible isomerization.
Solution conditions are described in the figure legends.
Molecular modeling
Models of cis and trans cross-linkers were built using HyperChem
(v.8, Hypercube Inc.), with the linker terminated with methyl
groups representing theb carbon of Cys in the cross-linked peptide,
and minimized using the Amber99 forcefield. Restraints were
added to the azo bond for the cis conformation (force constant
16). Molecular dynamics runs were performed in vacuo essentially
as described previously19 with a distant dependent dielectric, and
1–4 scale factors of 0.833 for electrostatic and 0.5 for van der
Waals interactions, a step size of 1 fs and 300 K as the simulation
temperature. Trajectories were analyzed to verify that numerous
torsion angle changes occurred for all single bonds during the
course of the simulation to ensure that conformational space was
adequately sampled. All points histograms were then produced for
the S–S distance during the full set of simulations for each isomer.
Additional conformational searches were performed using the
conformational search algorithm of HyperChem (v.8, Hypercube
Inc.) and the AM1 method. The ranges for acyclic torsions and
ring torsion flexing were 60 to 180◦ and 30 to 120◦, respectively.
A usage directed search algorithm was employed, with structures
A
molar extinction coefficient for the cross-linker of
16 800 M-1 cm-1 at 450 nm in pH 7.0 phosphate buffer was
determined by comparing absorbance at 450 nm vs. 280 nm for
the cross-linked peptide FK-11W. An extinction coefficient of
5625 M-1 cm-1 at 280 nm from the Trp residue at the N-terminus
of the cross-linked FK-11W was used. The UV–Vis spectrum of
cross-linked GSH (which does not have a Trp residue) was used
to correct for contributions of the cross-linker to the measured
absorbance at 280 nm.
˚
with non-H atom positions that varied by less than 0.25 A RMS or
Circular dichroism measurements
with torsion angle changes less than 5◦ considered as duplicates.
Structures less than 6 kcal mol-1 in energy above the lowest found
were retained.
Circular dichroism (CD) measurements were performed on a Jasco
Model J-710 spectropolarimeter. Tris(carboxyethyl)phosphine
(1 mM) was present in the uncross-linked peptide samples to
ensure that cysteine residues were in their reduced form. Peptide
concentrations were determined using a molar extinction coeffi-
cient for the cross-linker of 16 800 M-1 cm-1 at 450 nm in pH 7.0
phosphate buffer. For fast CD measurements, the instrument was
set to time-drive mode. Samples were dissolved in 10 mM pH 7.0
sodium phosphate buffer and/or 50% methanol in 10 mM pH 7.0
sodium phosphate buffer. Temperatures were measured using a
microprobe directly in the sample cell. The nitrogen flow rate
was increased to minimize condensation on the cuvette when the
temperature of the sample was 6 ◦C. Samples in cuvettes were
irradiated for 1 minute with a metal halide lamp (150 W, Osram
HQI-SE150/NDX (unfiltered), then measured immediately.
Acknowledgements
We are grateful to the Natural Sciences and Engineering Research
Council of Canada for financial support.
References
1 C. Renner and L Moroder, ChemBioChem, 2006, 7, 868.
2 G. A. Woolley, A. S. Jaikaran, M. Berezovski, J. P. Calarco, S. N. Krylov,
O. S. Smart and J. R. Kumita, Biochemistry, 2006, 45, 6075.
3 M. Volgraf, P. Gorostiza, R. Numano, R. H. Kramer, E. Y. Isacoff and
D. Trauner, Nat. Chem. Biol., 2006, 2, 47.
4 Y. Liu and D. Sen, J. Mol. Biol., 2004, 341, 887.
5 A. M. Caamano, M. E. Vazquez, J. Martinez-Costas, L. Castedo and
J. L. Mascarenas, Angew. Chem., Int. Ed., 2000, 39, 3104.
6 H. Rau, in Photochromism. Molecules and Systems, ed. H. Durr and
H. Bouas-Laurent, Elsevier, Amsterdam, 1990, pp. 165–192.
7 N. Pozhidaeva, M. E. Cormier, A. Chaudhari and G. A. Woolley,
Bioconjugate Chem., 2004, 15, 1297.
Photokinetic analysis
Photokinetic analyses were performed essentially as described
previously.30 A solution of cross-linked AB-15 (~15 mM) was
prepared in 50% methanol–50% phosphate buffer pH 7.0. Ab-
sorption spectra were recorded with the diode array UV–Vis
spectrophotometer described above, using slits of 16 nm and
16 nm. Two wavelengths of irradiation (l1 = 390 nm and l2 =
405 nm), both within the p–p* band of the chromophore, were
chosen for two sets of photokinetic measurements. In the first
set, the cross-linked peptide solution was irradiated with 390 nm
light and the absorbance at 390 nm and 450 nm were recorded.
In the second set, the cross-linked peptide solution was irradiated
with 405 nm light and the absorbance at 405 nm and 450 nm
were recorded. A total irradiation time of 80–130 s was used with
a 1 s sampling period. The intensity of the light was determined
directly in the working cell using an aqueous solution of potassium
ferrioxalate.36 The working cell was a quartz cuvette of 1 cm path
length with a stir bar containing 1 mL of the cross-linked peptide
solution. Data was analyzed as described previously.30
8 D. C. Burns, D. G. Flint, J. R. Kumita, H. J. Feldman, L. Serrano, Z.
Zhang, O. S. Smart and G. A. Woolley, Biochemistry, 2004, 43, 15329.
9 U. Kusebauch, S. A. Cadamuro, H. J. Musiol, M. O. Lenz, J. Wachtveitl,
L. Moroder and C. Renner, Angew. Chem., Int. Ed., 2006, 45, 7015.
10 R. F. Standaert and S. B. Park, J. Org. Chem., 2006, 71, 7952.
11 W. Cheong, S. Prahl and A. Welch, IEEE J. Quantum Electron., 1990,
26, 2166.
12 D. A. James, D. C. Burns and G. A. Woolley, Protein Eng., 2001, 14,
983.
13 D. M. Bers and T. Guo, Ann. N. Y. Acad. Sci., 2005, 1047, 86.
14 U. S. Bhalla, Curr. Opin. Genet. Dev., 2004, 14, 375.
15 E. Talaty and J. Fargo, J. Chem. Soc., 1967, 65.
16 R. Le Fevre and J. Northcott, J. Chem. Soc., 1953, 867.
17 C. Forber, E. Kelusky, N. Bunce and M. Zerner, J. Am. Chem. Soc.,
1985, 107, 5884.
18 N. Nishimura, T. Sueyoshi, H. Yamanaka, E. Imai, S. Yamamoto and
S. Hasegawa, Bull. Chem. Soc. Jpn., 1976, 49, 1381.
19 L. Chi, O. Sadovski and G. A. Woolley, Bioconjugate Chem., 2006, 17,
670.
20 H. R. Lopez-Mirabal and J. R. Winther, Biochim. Biophys. Acta, 2008,
1783, 629.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4323–4332 | 4331
©

View Online
21 H. Ostergaard, C. Tachibana and J. R. Winther, J. Cell Biol., 2004, 166,
29 J. Hamlin, D. Phillips and A. Whiting, Dyes Pigm., 1999, 41,
137.
337.
22 E. Kosower and H. Kanety-Londner, J. Am. Chem. Soc., 1976, 98,
30 V. Borisenko and G. A. Woolley, J. Photochem. Photobiol., A, 2005,
3001.
173, 21.
23 C. Boulegue, M. Loweneck, C. Renner and L. Moroder, ChemBioChem,
2007, 8, 591.
24 K. Fujimoto, M. Amano, Y. Horibe and M. Inouye, Org. Lett., 2006,
8, 285.
31 C. Renner, J. Cramer, R. Behrendt and L. Moroder, Biopolymers, 2000,
54, 501.
32 J. R. Kumita, O. S. Smart and G. A. Woolley, Proc. Natl. Acad. Sci.
U. S. A., 2000, 97, 3803.
25 G. Hallas, R. Marsden, J. Hepworth and D. Mason, J. Chem. Soc.,
Perkin Trans. 2, 1984, 149.
26 J. Hepworth, D. Mason, G. Hallas and R. Marsden, Dyes Pigm., 1985,
6, 389.
27 A. Bottini and C. Nash, J. Am. Chem. Soc., 1962, 84, 734.
28 C. Nash and G. Maciel, J. Phys. Chem., 1964, 68, 832.
33 N. Nishimura, T. Tanaka, M. Asano and Y. Sueishi, J. Chem. Soc.,
Perkin Trans. 2, 1986, 1839.
34 G. A. Woolley, Acc. Chem. Res., 2005, 38, 486.
35 J. E. Brown and W. A. Klee, Biochemistry, 1971, 10, 470.
36 S. Murov, I. Carmichael and G. Hug, Handbook of Photochemistry,
Marcel Dekker, Inc, New York, 1993, pp. 299–305.
4332 | Org. Biomol. Chem., 2008, 6, 4323–4332
This journal is
The Royal Society of Chemistry 2008
©