Azobenzene photoswitching without ultraviolet light

Article abstract of DOI:10.1021/ja209239m

Most azobenzene-based photoswitches use UV light for photoisomerization. This can limit their application in biological systems, where UV light can trigger unwanted responses, including cellular apoptosis. We have found that substitution of all four ortho positions with methoxy groups in an amidoazobenzene derivative leads to a substantial (～35 nm) red shift of the n-π* band of the trans isomer, separating it from the cis n-π* transition. This red shift makes trans-to-cis photoswitching possible using green light (530-560 nm). The cis state is thermally stable with a half-life of ～2.4 days in the dark in aqueous solution. Reverse (cis-to-trans) photoswitching can be accomplished with blue light (460 nm), so bidirectional photoswitching between thermally stable isomers is possible without using UV light at all.

Full text of DOI:10.1021/ja209239m

COMMUNICATION
pubs.acs.org/JACS
Azobenzene Photoswitching without Ultraviolet Light
Andrew A. Beharry, Oleg Sadovski, and G. Andrew Woolley*
Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, ON, Canada M5S 3H6
S Supporting Information
b
Chart 1. Structures of the Azobenzene Derivatives Studied
ABSTRACT: Most azobenzene-based photoswitches use
UV light for photoisomerization. This can limit their appli-
cation in biological systems, where UV light can trigger
unwanted responses, including cellular apoptosis. We have
found that substitution of all four ortho positions with
methoxy groups in an amidoazobenzene derivative leads
to a substantial (∼35 nm) red shift of the nꢀπ* band of the
trans isomer, separating it from the cis nꢀπ* transition. This
red shift makes trans-to-cis photoswitching possible using
green light (530ꢀ560 nm). The cis state is thermally stable
with a half-life of ∼2.4 days in the dark in aqueous solution.
Reverse (cis-to-trans) photoswitching can be accomplished
with blue light (460 nm), so bidirectional photoswitching
between thermally stable isomers is possible without using
UV light at all.
of thermal cis-to-trans relaxation.24ꢀ30 A short-lived cis isomer
means that an intense light source is required in order to maintain
a substantial fraction of the cis isomer, an undesirable limitation
in vivo. Also, while blue-absorbing 2,20-dimethylaminoazoben-
zenes were found to undergo thousands of cycles of photo-
switching, green-absorbing 2,20-pyrrolidino derivatives showed
photobleaching effects that limited the number of cycles to
<100.29 The mechanism of this photobleaching process is un-
known but may involve participation of the o-amino group in a
photo-oxidation process.31,32 To retain the longer cis half-lives
observed with ortho substitution and avoid photobleaching,
we explored the effect of a different type of electron-donating
group: azobenzene switches bearing o-methoxy groups on the
4,40-diacetamidoazobenzene scaffold were synthesized (Chart 1).33
While the πꢀπ* band of trans-2,20-dimethoxy-4,40-diacetami-
doazobenzene (2) showed the expected red shift [see the
Supporting Information (SI)], the πꢀπ* band of (3) was
significantly blue-shifted and weaker (λmax ≈ 338 nm, ε =
16 170 Mꢀ1 cmꢀ1) and the nꢀπ* band significantly red-shifted
and stronger (480 nm, ∼4030 Mꢀ1 cmꢀ1) relative to those of the
parent compound (1) in dimethyl sulfoxide (DMSO) at 25 °C
(Figure 1; also see the SI). In retrospect, this behavior may have
been partially predicted on the basis of studies reported by
Bunce, Zerner, and colleagues34 and Bisle and Rau35 on azoben-
zene derivatives bearing alkyl substituents at all four ortho
positions where (smaller) blue shifts of πꢀπ* bands and red-
shifted nꢀπ* bands were also seen (see below).
zobenzene compounds have been used for optical control of
A
conformation in diverse settings.1ꢀ4 In biological systems,
optical manipulation of biomolecular structure is a powerful
approach for deciphering cellular roles of biomolecules with high
spatiotemporal resolution.5ꢀ13 At equilibrium in the dark, azo-
benzene exists in the more stable trans conformation. Irradiation
with UV light produces a large fraction of the cis isomer, which
can revert back to the trans state thermally or upon irradiation
with blue light. Azobenzene has been employed as a photoswitch
by coupling the photoisomerization event to a change in the
structure/function of numerous biomolecular targets, including
peptides, proteins, and nucleic acids in vitro and ion channels
and receptors in vivo.3,14,15 Most of the azobenzene-modified
biomolecules developed to date require the use of UV light
for photoisomerization. This can limit the use of this molecular
switch in vivo. For example, UV light is strongly scattered,
making penetration of cells and tissues difficult.16 Furthermore,
UV light can trigger apoptotic events or other responses, compli-
cating studies of cells in culture and of model organisms.17ꢀ22
Azobenzene derivatives where photoisomerization can occur
entirely in the visible region would therefore be desirable for
in vivo applications. One approach for achieving long-wavelength
switching is to couple azobenzene derivatives to upconverting
nanophosphors.23 The size of these particles, however, may cause
interference with biological targets in vivo.
The spectrum of cis-(3) in DMSO was then determined by
producing a mixture of cis and trans isomers by UV irradiation
and determining the cis content by quantitative NMR methods
(see the SI). Remarkably, the spectrum of the cis isomer showed
a large (36 nm) blue shift of the nꢀπ* band relative to the trans
isomer and an obvious color change (Figure 1). For most
azobenzenes, including the parent compound, the trans and cis
Azobenzene-based photoswitchable cross-linkers and teth-
ered ligands with a 4,40-diamidoazobenzene core (1) have been
used extensively for the photocontrol of biomolecule structure
and function.3 The incorporation of electron-donating groups
ortho or para to the azo moiety can dramatically red-shift the
photoswitching wavelength but also markedly increases the rate
Received: September 30, 2011
Published: November 14, 2011
r
2011 American Chemical Society
19684
dx.doi.org/10.1021/ja209239m J. Am. Chem. Soc. 2011, 133, 19684–19687
|

Journal of the American Chemical Society
COMMUNICATION
Figure 2. Multiple rounds of photoswitching of (3) with alternating
green (530 nm) and blue (450ꢀ460 nm) light in aqueous solution at
25 °C. The gap on the time axis corresponds to 1 h of continuous high-
intensity (70 mW/cm2) irradiation.
Figure 1. Photoisomerization of (3) measured in DMSO at 25 °C.
isomers have nꢀπ* bands at similar wavelengths.36 To our
knowledge, an nꢀπ* shift larger than this has been seen only
with the bridged azobenzene derivative recently reported by
Siewertsen et al.,36 where the nꢀπ* band of the trans isomer
occurs at 490 nm and the cis isomer at 404 nm (see below). The
shifts in the nꢀπ* transitions between cis and trans isomers in
the tetra-o-alkyl-substituted azo derivatives studied by Bunce,
Zerner and colleagues34 were substantially smaller (4ꢀ9 nm).
This observation of a separation of the trans and cis nꢀπ* bands
in (3) indicated that trans-to-cis isomerization should be possible
not only with UV light (excitation in the trans πꢀπ* band) but
also with green light (excitation in the trans nꢀπ* band), a
wavelength that would normally lead to photoisomerization in
the opposite direction. Indeed, irradiation of trans-(3) with green
light (530ꢀ560 nm) was found to produce a mixture with an
∼80% fraction of the cis isomer (Figure 1), which is comparable
to the fraction produced using UV light with the parent
4,40-diamidoazobenzene (1). Photochemical cis-to-trans isomer-
ization of (3) could be then induced by excitation of the cis nꢀπ*
band using blue light. Irradiation with a 450ꢀ460 nm light-
emitting diode resulted in a large fraction of the trans isomer
(∼85%) (Figure 1). Similar photoswitching wavelengths and
isomer yields were also observed in other nonpolar solvents such
as acetonitrile, dioxane, and dichloromethane (see the SI).
Since biological applications require the use of the photo-
switch in aqueous solution, the photoswitching properties were
also determined in 25 mM phosphate buffer (pH 7.0). While the
parent compound (1) and the 2,20-derivative (2) are only
sparingly soluble in water, (3) is very soluble. Interestingly, in
aqueous solutions, the position of the πꢀπ* band of trans-(3)
was found to be affected by temperature. At higher temperatures,
the πꢀπ* band was blue-shifted relative to that of (1), as in
DMSO. Conversely, lowering the temperature to 4 °C caused a
red shift (λmax ≈ 380 nm) and a hyperchromic effect. The
spectrum of the cis isomer did not appear to be significantly
affected by temperature; as a result, there was a smaller separation
of the trans and cis nꢀπ* bands in aqueous solution. Never-
theless, irradiation with green wavelengths (530ꢀ560 nm) led to
production of a large fraction of the cis isomer (∼70%) that
could be switched back to the trans form with blue light (∼80%)
at 25 °C (see the SI).
Thermal cis-to-trans isomerization was followed by acquiring
UVꢀvis spectra as a function of time using a light source
too weak to cause significant photoisomerization. Half-lives are
collected in Table S1 in the SI. Notably, the half-life of (3)
is substantially longer than that of the parent compound (1)
(53 h vs ∼12 min in aqueous solution37). Decreasing the solvent
polarity further extended the lifetime of the cis isomer, an effect
that has been attributed to a dipolar transition state.24,28 A half-
life of 2.4 days means that both the cis and trans isomers can be
rapidly photogenerated and maintained in the dark for the
duration of most biological process where spatiotemporal con-
trol is of interest.
The pH dependence of the UVꢀvis absorbance spectra in
water was also investigated. The apparent pKa (∼3.8) for the
protonated (3) species was higher than the pKa for the unsub-
stituted compound (1) (see the SI), but the spectra were
essentially unaffected by pH above pH 5.0. Thus, protonation
of the photoswitch should not be a problem for biomolecular
targets operating near pH 7.
Multiple rounds of photoisomerization, and even constant
high-intensity irradiation with green light (70 mW/cm2) for
∼1 h, gave no evidence of photobleaching or photo-oxidation
(Figure 2). Conversely, (3) was found to be more sensitive to
reduction than the parent compound. Whereas (1) is stable for
extended periods in 1 mM reduced glutathione,38 (3) was
bleached with a half-life of ∼1.5 h (see the SI). This may limit
the application of (3) to more oxidizing environments in vivo.39
In an effort to understand the structural basis for the spectro-
scopic behavior of (3), we carried out computational modeling of
the compound in its cis and trans forms. We employed density
functional theory (DFT) using the B3LYP hybrid functional40
and the 6-31G* basis set since this method successfully repro-
duces the relative energies of the isomers as well as the spectra of
many azobenzene derivatives, including the parent compound
and bridged azobenzenes.41ꢀ44 A conformational search was
carried out as described in the SI; the lowest-energy conformers
for trans- and cis-(3) are shown in Figure 3. The trans isomer is
highly twisted in comparison with the parent compound (1),
which exhibits a nearly planar geometry (see the SI). The
2,20-dimethoxy species (2) can also adopt a planar structure
by having the substituents on opposite sides of the azo moiety
19685
dx.doi.org/10.1021/ja209239m |J. Am. Chem. Soc. 2011, 133, 19684–19687

Journal of the American Chemical Society
COMMUNICATION
the cis and trans nꢀπ* bands makes possible efficient switching
using nꢀπ* excitation. While for the bridged azobenzene near-
UV light at 385 nm was still required for optimal cis-to-trans
switching and green light (520 nm) for trans-to-cis, in the present
work, blue light (460 nm) and green light (530ꢀ560 nm) are
optimal for cis-to-trans and trans-to-cis switching, respectively.
In the bridged azobenzene case, however, the magnitude of
the nꢀπ* separation is larger, so more complete switching is
possible. Although incomplete photoconversion of cis and trans
isomers can enable photocontrol of biomolecule function in
specific cases,30,47 more complete switching is generally prefer-
able. By increasing the separation of the nꢀπ* bands, more
complete switching should be possible. A mechanism of selective
destabilization of the trans-isomer nꢀπ* HOMO by ortho
substituents suggests an approach to the design of azobenzene
derivatives with even better (i.e., more complete) switching
behavior that do not require the use of UV light. Further
manipulation of the ring substituents may also permit the
sensitivity to reduction by glutathione to be mitigated.38
The synthesis of (3) is straightforward, and derivatives could
be readily linked to other molecules via the p-amido substituents.
The solvent dependence observed may cause the cis/trans ratio
at the photostationary state to be dependent on the local
environment of the target molecule incorporation site, although
effective switching is still possible under fully hydrated condi-
tions. Slow thermal relaxation means only brief pulses of low-
intensity light can be used for photoswitching in either direction.
In this manner, a biological target could be maintained in either
an “off” or “on” state as required and could be interconverted
with brief light pulses. Related tetra-o-methoxy-substituted deri-
vatives could also find applications in materials science and
supramolecular chemistry, where reversible conformational con-
trol using visible light is desired.23
Figure 3. Calculated structures of trans- and cis-(3). The absolute value
of the HOMO of each isomer is mapped onto the bond density surface
(blue is the maximum value of the HOMO).
(see the SI). This arrangement is seen in X-ray crystal structures
of related 2,20-alkoxy-substituted azobenzenes.45,46 The calcu-
lated energies of the isomers of (3) were sensitive to the
incorporation of a solvation term. In water and DMSO, the trans
isomer was calculated to be 8ꢀ9 kcal/mol more stable than the
cis form. This is consistent with the cis isomer being undetectable
by NMR spectroscopy in a dark-adapted sample of trans-(3) (see
the SI).
Time-dependent DFT calculations reproduced the blue shift
of the πꢀπ* band and the red shift of the nꢀπ* band of trans-(3)
relative to the parent compound (1) (see the SI). Moreover, the
large separation of the nꢀπ* bands of the trans and cis isomers
was also predicted, with the trans isomer having λmax = 520 nm
(slightly higher energy conformers had λmax = 505ꢀ515 nm) and
the cis isomer having λmax near 460 nm (a separation of
40ꢀ60 nm; the experimental separation was 36 nm).
’ ASSOCIATED CONTENT
S
Supporting Information. Syntheticprocedures and UVꢀ
b
vis spectra in other solvents; pH and temperature dependence of
the spectra, and calculated structures and spectra of compounds
(1)ꢀ(3). This material is available free of charge via the Internet
at http://pubs.acs.org.
A possible origin of this separation of the trans and cis nꢀπ*
bands, originally suggested by Bunce and Zerner,34 can be appre-
ciated by comparing the absolute value of the highest occupied
molecular orbital (HOMO) of each isomer mapped onto the bond
density surface (Figure 3) The HOMO is centered on the lone
pairs of the azoN atoms in each case. However, for the trans isomer,
the HOMO is close to the electron-rich oxygen atoms of the
methoxy groups, which may be expected to raise its relative energy.
Isomerization to the cis form relieves this interaction, and the azo
N atom lone pairs are exposed to solvent in the same manner as in
the parent compound. Indeed, the nꢀπ* transition of the cis
isomer is similar in energy to that of the parent compound.
As noted above, Siewertsen et al.36 reported a bridged
azobenzene derivative in which a large separation of nꢀπ* bands
was also observed. In that case, the nꢀπ* transition of the trans
isomer was red-shifted, but the nꢀπ* transition of the cis isomer
was also significantly blue-shifted relative to unmodified azoben-
zene (the corresponding parent compound), indicating that the
origin of the nꢀπ* separation is different. Moreover, in that case,
the cis isomer was the more stable isomer by 6ꢀ7 kcal/mol. In
both the bridged azobenzene case reported by Siewertesen and
the tetra-o-methoxy case described here, the large separation of
’ AUTHOR INFORMATION
Corresponding Author
awoolley@chem.utoronto.ca
’ ACKNOWLEDGMENT
This research was supported by the Natural Sciences and
Engineering Research Council of Canada. A.A.B. was supported
by an Ontario Graduate Scholarship.
’ REFERENCES
(1) Yamada, M.; Kondo, M.; Mamiya, J.; Yu, Y.; Kinoshita, M.;
Barrett, C. J.; Ikeda, T. Angew. Chem., Int. Ed. 2008, 47, 4986.
(2) Russew, M. M.; Hecht, S. Adv. Mater. 2010, 22, 3348.
(3) Beharry, A. A.; Woolley, G. A. Chem. Soc. Rev. 2011, 40, 4422.
(4) Kusebauch, U.; Cadamuro, S. A.; Musiol, H. J.; Lenz, M. O.;
Wachtveitl, J.; Moroder, L.; Renner, C. Angew. Chem., Int. Ed. 2006,
45, 7015.
19686
dx.doi.org/10.1021/ja209239m |J. Am. Chem. Soc. 2011, 133, 19684–19687

Journal of the American Chemical Society
COMMUNICATION
(5) Fenno, L.; Yizhar, O.; Deisseroth, K. Annu. Rev. Neurosci. 2011,
34, 389.
(39) Moriarty-Craige, S. E.; Jones, D. P. Annu. Rev. Nutr. 2004,
24, 481.
(6) Szobota, S.; Isacoff, E. Y. Annu. Rev. Biophys. Biomol. Struct. 2010,
39, 329.
(40) Becke, A. D. J. Chem. Phys. 1997, 98, 5648.
(41) Chen, P. C.; Chieh, Y. C.; Wu, J. C. J. Mol. Struct. 2005, 715,
(7) Heckel, A.; Mayer, G. In The Chemical Biology of Nucleic Acids;
Mayer, G., Ed.; Wiley: Chichester, U.K., 2010; pp 279ꢀ306.
(8) Moglich, A.; Moffat, K. Photochem. Photobiol. Sci. 2011, 9, 1286.
(9) Deiters, A.; Garner, R. A.; Lusic, H.; Govan, J. M.; Dush, M.;
Nascone-Yoder, N. M.; Yoder, J. A. J. Am. Chem. Soc. 2011, 132, 15644.
(10) Volgraf, M.; Gorostiza, P.; Numano, R.; Kramer, R. H.; Isacoff,
E. Y.; Trauner, D. Nat. Chem. Biol. 2006, 2, 47.
(11) Wyart, C.; Del Bene, F.; Warp, E.; Scott, E. K.; Trauner, D.;
Baier, H.; Isacoff, E. Y. Nature 2009, 461, 407.
(12) Banghart, M. R.; Mourot, A.; Fortin, D. L.; Yao, J. Z.; Kramer,
R. H.; Trauner, D. Angew. Chem., Int. Ed. 2009, 48, 9097.
(13) Kramer, R. H.; Fortin, D. L.; Trauner, D. Curr. Opin. Neurobiol.
2009, 19, 544.
183.
(42) Nakamura, S.; Guillaumont, D. Dyes Pigm. 2000, 46, 85.
(43) Fang, W. H.; Liu, L. H.; Yuan, S. A.; Zhang, Y. J. Phys. Chem. A
2011, 115, 10027.
(44) Siewertsen, R.; Schonborn, J. B.; Hartke, B.; Renth, F.; Temps,
F. Phys. Chem. Chem. Phys. 2011, 13, 1054.
(45) Hummel, M.; Wurst, K.; Laus, G.; Schottenberger, H. Z.
Kristallogr.—New Cryst. Struct. 2009, 224, 303.
(46) Nagamani, S. A.; Norikane, Y.; Tamaoki, N. J. Org. Chem. 2005,
70, 9304.
(47) Banghart, M. R.; Volgraf, M.; Trauner, D. Biochemistry 2006,
45, 15129.
(14) Gorostiza, P.; Isacoff, E. Y. Science 2008, 322, 395.
(15) Hoppmann, C.; Schmieder, P.; Domaing, P.; Vogelreiter, G.;
Eichhorst, J.; Wiesner, B.; Morano, I.; Ruck-Braun, K.; Beyermann, M.
Angew. Chem., Int. Ed. 2011, 50, 7699.
(16) Cheong, W. F.; Prahl, S. A.; Welch, A. J. IEEE J. Quantum
Electron. 1990, 26, 2166.
(17) Banerjee, G.; Gupta, N.; Kapoor, A.; Raman, G. Cancer Lett.
2005, 223, 275.
(18) Kamarajan, P.; Chao, C. C. Biosci. Rep. 2000, 20, 99.
(19) Cooper, S.; Ranger-Moore, J.; Bowden, T. G. Mol. Carcinog.
2005, 43, 108.
(20) Bachelor, M. A.; Bowden, G. T. Semin. Cancer Biol. 2004, 14,
131.
(21) Weger, B. D.; Sahinbas, M.; Otto, G. W.; Mracek, P.; Armant,
O.; Dolle, D.; Lahiri, K.; Vallone, D.; Ettwiller, L.; Geisler, R.; Foulkes,
N. S.; Dickmeis, T. PLoS One 2011, 6, No. e17080.
(22) Tamai, T. K.; Vardhanabhuti, V.; Foulkes, N. S.; Whitmore, D.
Curr. Biol. 2004, 14, R104.
(23) Wu, W.; Yao, L.; Yang, T.; Yin, R.; Li, F.; Yu, Y. J. Am. Chem. Soc.
2011, 133, 15810.
(24) Nishimura, N.; Sueyoshi, T.; Yamanaka, H.; Imai, E.; Yamamoto,
S.; Hasegawa, S. Bull. Chem. Soc. Jpn. 1976, 49, 1381.
(25) Le Fevre, R. J. W.; Northcott, J. J. Chem. Soc. 1953, 867.
(26) Mustroph, H. Dyes Pigm. 1991, 15, 129.
(27) Mustroph, H. Dyes Pigm. 1991, 16, 223.
(28) Dokic, J.; Gothe, M.; Wirth, J.; Peters, M. V.; Schwarz, J.; Hecht,
S.; Saalfrank, P. J. Phys. Chem. A 2009, 113, 6763.
(29) Sadovski, O.; Beharry, A. A.; Zhang, F.; Woolley, G. A. Angew.
Chem., Int. Ed. 2009, 48, 1484.
(30) Mourot, A.; Kienzler, M. A.; Banghart, M. R.; Fehrentz, T.;
Huber, F. M. E.; Stein, M.; Kramer, R. H.; Trauner, D. ACS Chem.
Neurosci. 2011, 2, 536.
(31) Galindo, C.; Jacques, P.; Kalt, A. J. Photochem. Photobiol., A
2000, 130, 35.
(32) Hihara, T.; Okada, Y.; Morita, Z. Dyes Pigm. 2007, 75, 225.
(33) (a) The synthesis of 2,20,6,60-tetramethoxyazobenzene [i.e., (3)
without the p-acetamido groups] has been reported, but the photo-
isomerization behavior apparently was not studied. See: Wheeler, O. H.;
Gonzalez, D. Tetrahedron 1964, 20, 189. (b) The photoisomerization of
2,20-dimethoxyazobenzene was studied by Tahara et al. and exhibits
some of the features observed here. See: Tahara, R.; Morozumi, T.;
Nakamura, H.; Shimomura, M. J. Phys. Chem. B 1997, 101, 7736.
(34) Forber, C. L.; Kelusky, E. C.; Bunce, N. J.; Zerner, M. C. J. Am.
Chem. Soc. 1985, 107, 5884.
(35) Bisle, H.; Rau, H. Chem. Phys. Lett. 1975, 31, 264.
(36) Siewertsen, R.; Neumann, H.; Buchheim-Stehn, B.; Herges, R.;
Nather, C.; Renth, F.; Temps, F. J. Am. Chem. Soc. 2009, 131, 15594.
(37) Borisenko, V.; Woolley, G. A. J. Photochem. Photobiol., A 2005,
173, 21.
(38) Beharry, A. A.; Wong, L.; Tropepe, V.; Woolley, G. A. Angew.
Chem., Int. Ed. 2011, 50, 1325.
19687
dx.doi.org/10.1021/ja209239m |J. Am. Chem. Soc. 2011, 133, 19684–19687