Photocontrol of protein activity in cultured cells and zebrafish with one- and two-photon illumination

Article abstract of DOI:10.1002/cbic.201000008

We have implemented a noninvasive optical method for the fast control of protein activity in a live zebrafish embryo. It relies on releasing a protein fused to a modified estrogen receptor ligand binding domain from its complex with cytoplasmic chaperones, upon the local photoactivation of a nonendogenous caged inducer. Molecular dynamics simulations were used to design cyclofen-OH, a photochemically stable inducer of the receptor specific for 4-hydroxy-tamoxifen (ER^T2). Cyclofen-OH was easily synthesized in two steps with good yields. At submicromolar concentrations, it activates proteins fused to the ER^T2 receptor. This was shown in cultured cells and in zebrafish embryos through emission properties and subcellular localization of properly engineered fluorescent proteins. Cyclofen-OH was successfully caged with various photolabile protecting groups. One particular caged compound was efficient in photoinducing the nuclear translocation of fluorescent proteins either globally (with 365 nm UV illumination) or locally (with a focused UV laser or with two-photon illumination at 750 nm). The present method for photocontrol of protein activity could be used more generally to investigate important physiological processes (e.g., in embryogenesis, organ regeneration and carcinogenesis) with high spatiotemporal resolution.

Full text of DOI:10.1002/cbic.201000008

DOI: 10.1002/cbic.201000008
Photocontrol of Protein Activity in Cultured Cells and
Zebrafish with One- and Two-Photon Illumination
Deepak Kumar Sinha,[a] Pierre Neveu,[a, b, i] Nathalie Gagey,[b] Isabelle Aujard,[b]
Chouaha Benbrahim-Bouzidi,[b] Thomas Le Saux,[b] Christine Rampon,[c, d] Carole Gauron,[c]
Bernard Goetz,[b] Sylvie Dubruille,[e] Marc Baaden,[f] Michel Volovitch,[c, g]
David Bensimon,*[a, h] Sophie Vriz,*[c, d] and Ludovic Jullien*[b]
We have implemented a noninvasive optical method for the
fast control of protein activity in a live zebrafish embryo. It
relies on releasing a protein fused to a modified estrogen re-
ceptor ligand binding domain from its complex with cytoplas-
mic chaperones, upon the local photoactivation of a nonen-
dogenous caged inducer. Molecular dynamics simulations were
used to design cyclofen-OH, a photochemically stable inducer
of the receptor specific for 4-hydroxy-tamoxifen (ERT2). Cyclo-
fen-OH was easily synthesized in two steps with good yields.
At submicromolar concentrations, it activates proteins fused to
the ERT2 receptor. This was shown in cultured cells and in
zebrafish embryos through emission properties and subcellular
localization of properly engineered fluorescent proteins. Cyclo-
fen-OH was successfully caged with various photolabile pro-
tecting groups. One particular caged compound was efficient
in photoinducing the nuclear translocation of fluorescent pro-
teins either globally (with 365 nm UV illumination) or locally
(with a focused UV laser or with two-photon illumination at
750 nm). The present method for photocontrol of protein ac-
tivity could be used more generally to investigate important
physiological processes (e.g., in embryogenesis, organ regener-
ation and carcinogenesis) with high spatiotemporal resolution.
[a] Dr. D. K. Sinha, Dr. P. Neveu, Dr. D. Bensimon
Ecole Normale Supꢀrieure
Introduction
Dꢀpartement de Physique and Dꢀpartement de Biologie
Laboratoire de Physique Statistique UMR CNRS-ENS 8550
24 rue Lhomond, 75231 Paris (France)
Cells respond to external signals by modifying their internal
state and their environment. In multicellular organisms in par-
ticular, cellular differentiation and intracellular signaling are
essential for the coordinated development of the organism.[1]
Revealing and understanding the spatiotemporal dynamics of
these complex interaction networks is a major goal in biology.
While some of the most important players of these networks
have been identified, much less is known of the quantitative
rules that govern their interaction with one another and with
other cellular components (affinities, rate constants, strength
of nonlinearities, such as feedback or feed-forward loops, etc.).
Investigating these interactions, which is a prerequisite for
understanding or modeling them, requires the development of
means to control or interfere spatially and temporally with
these processes.
E-mail: david.bensimon@ens.fr
[b] Dr. P. Neveu, Dr. N. Gagey, Dr. I. Aujard, C. Benbrahim-Bouzidi,
Dr. T. Le Saux, B. Goetz, Prof. Dr. L. Jullien
Ecole Normale Supꢀrieure, Dꢀpartement de Chimie
UMR CNRS-ENS-UPMC Paris 06 8640 PASTEUR
24 rue Lhomond, 75231 Paris (France)
E-mail: ludovic.jullien@ens.fr
[c] Dr. C. Rampon, C. Gauron, Prof. Dr. M. Volovitch, Prof. Dr. S. Vriz
Collꢁge de France, UMR 8542 CNRS-ENS-CDF
11 place M. Berthelot, 75231 Paris Cedex 05 (France)
E-mail: vriz@univ-paris-diderot.fr
[d] Dr. C. Rampon, Prof. Dr. S. Vriz
Universitꢀ Paris Diderot-Paris 7
80, rue du Gꢀnꢀral Leclerc Bat G. Pincus
94276 Le Kremlin-BicÞtre Cedex (France)
[e] S. Dubruille
To address these issues, various approaches have been intro-
duced to control protein activity. A first strategy relies on
tuning protein concentration by controlling gene expression
or messenger RNA translation. This goal can be achieved with
conditional gene expression systems[2] or by using antisense
oligonucleotides.[3] However, such a control introduces delays
associated with mRNA or protein syntheses that prevent inter-
ference with protein patterns at the time scale of fast biologi-
cal processes, such as phosphorylation.[4] A second strategy
avoids this drawback by directly acting at the protein level:
the activity of the protein of interest is restored with an appro-
priate stimulus. The fast spatiotemporal dynamics of photoacti-
vation methods have proved particularly attractive. In favora-
ble cases, photoactivatable substrates can be used to alter the
Institut Curie, UMR 176 Institut Curie–CNRS
26, rue d’Ulm, 75005 Paris (France)
[f] Dr. M. Baaden
Institut de Biologie Physico-Chimique
Laboratoire de Biochimie Thꢀorique, CNRS UPR 9080
13, rue Pierre et Marie Curie, 75005 Paris (France)
[g] Prof. Dr. M. Volovitch
Ecole Normale Supꢀrieure, Dꢀpartement de Biologie
46, rue d’Ulm, 75231 Paris Cedex 05 (France)
[h] Dr. D. Bensimon
Department of Chemistry and Biochemistry, UCLA, Los Angeles (USA)
[i] Dr. P. Neveu
Kavli Institute for Theoretical Physics
University of California at Santa Barbara, Santa Barbara CA 93106 (USA)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201000008.
ChemBioChem 2010, 11, 653 – 663
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
653

D. Bensimon, S. Vriz, L. Jullien et al.
function of native or engineered proteins.[5,6] Direct caging of
peptides and proteins has been reported by many groups.[7]
However, the caged precursors of these macromolecules must
be injected into cells and this limits their range of access. Ge-
netically encoded photoactivatable proteins do not suffer from
this drawback in transgenic organisms. They can be designed
to intrinsically bear the photoactive trigger.[8] Alternatively,
they can contain a nonphotoactive site, which can be activated
by a small permeant caged lipophilic molecule, such as those
derived from estradiol,[9] ecdysone,[10] 4-hydroxy tamoxifen,[11,12]
or doxycycline.[13] This method has been successfully imple-
mented to photocontrol gene expression in eukaryotic systems
with one- and two-photon excitation.[9,10,12,13]
large-scale genetic screens and its genomic sequence is now
available and provides valuable tools.
The experimental protocol for the present photocontrol of
protein activity in zebrafish is straightforward. First, an animal
expressing the appropriate fusion protein is engineered (pref-
erably permanently by incorporating the appropriate trans-
gene in its genome or alternatively transiently by injecting its
mRNA at the zygote stage). The embryo is incubated in an
aqueous solution of the caged inducer that penetrates the
whole organism. At a chosen stage of development, it is trans-
ferred to an illumination chamber where the inducer is un-
caged in the targeted cell(s) of the animal. The released ligand
then activates the protein fused to its receptor. In the present
study, we report photocontrol over nuclear translocation of
GFP–nls-ERT2 and mCherry–nls-ERT2, two fluorescent proteins
linked to the ERT2 receptor by a nuclear localization signal.
In the present study, we retained the principle of a small
lipophilic caged inducer to photoactivate properly engineered
proteins, in vivo. We adopted a steroid-related inducer since
various proteins (e.g., Engrailed, Otx2, GaL4, p53, kinases, such
as Raf-1, Cre and Flp recombinases) fused to a steroid receptor
have been shown to be activated by binding of an appropriate Results and Discussion
ligand (Scheme 1).[14] In its absence, the receptor forms a cyto-
Cyclofen-OH synthesis
Since we observed that tamoxifen-OH was susceptible to pho-
toisomerization and photodegradation upon UV illumination at
uncaging time scales (see the Supporting Information) we
looked for another inducer structurally related to tamoxifen-
OH, but which would neither isomerize nor degrade upon UV
illumination. In view of its synthetic accessibility, we were
attracted by the core motif of the estrogen, cyclofenil (the bis-
(acetate) of the diphenol 1 in Scheme 2).[23] In particular we
wondered whether 4-hydroxy-cyclofen (cyclofen-OH or Ind;
Scheme 1) would be as active in binding the estrogen receptor
as tamoxifen-OH.
Scheme 1. A) The strategy to photoactivate properly engineered proteins. A
protein fused to the ERT2 receptor (Prot; here a fluorescent protein linked to
the receptor by a peptide acting as a nuclear localization signal) is inactivat-
ed by its assembly with a chaperone complex (CC). Upon photoactivation of
a caged precursor (cInd), a nonendogenous inducer (Ind) is released and
binds to the ERT2 moiety. The concomitant conformational change of the re-
ceptor causes assembly disruption and activates the ERT2-fused protein. Ex-
amples of inducers: B) Ind: 4-hydroxy-tamoxifen (tamoxifen-OH), and C) 4-
hydroxy-cyclofen (cyclofen-OH).
To obtain some insight about the interaction of the cyclo-
fen-OH putative inducer with the modified estrogen receptor
(ERT2) we performed several molecular dynamics (MD) simula-
tions using tamoxifen-OH as a reference. In the absence of any
crystal structure of the tamoxifen-OH–ERT2 complex, we ana-
lyzed the interaction of both cyclofen-OH and tamoxifen-OH
with the steroid receptor ERa mutated to adopt the ERT2 se-
quence (Gly400Val, Met543Ala, Leu544Ala) starting from the
2.0 ꢁ crystal structure of the ERa ligand binding domain com-
plexed with lasofoxifene,[24] a compound structurally closely
related to tamoxifen (and tamoxifen-OH). We investigated the
interaction of tamoxifen-OH and of cyclofen-OH with the refer-
ence ligand binding domain (simulations T1 and C1, respec-
tively). We also analyzed the interaction of cyclofen-OH with
the Asp351Glu mutant, in order to assess whether a slightly
longer acidic side chain at position 351 might benefit binding
(simulation C2).
plasmic assembly with a chaperone complex, which inactivates
the fusion protein.[15] Its function is restored in the presence of
the steroid ligand, which binds to the receptor and disrupts
the complex. Like Link et al.,[12] we chose to photocontrol the
activity of a target protein by fusing it to the extensively used
modified estrogen receptor ligand binding domain, ERT2, which
is specific for the nonendogenous 4-hydroxy-tamoxifen inducer
(tamoxifen-OH; Scheme 1).[16]
When aiming to photocontrol the activity of proteins in a
live animal, the zebrafish (Danio rerio) is a system of choice. Its
transparency and the existence of lines without pigments, its
small size, easy maintenance and transgenesis, abundant prog-
eny and rapid development allow real-time observations under
the microscope. In fact, zebrafish has become a popular verte-
brate model for developmental studies[17,18] and the investiga-
tion of human pathologies.[19,20–22] It was successfully used for
All MD simulations lead to stable ligand binding domain–
ligand complexes. Figure 1 shows representative snapshots of
the three simulations. The ligands are firmly inserted in their
binding pockets and are in very similar orientations for T1, C1
and C2. In particular, in the detailed view of the C1 binding
pocket and ligand orientation shown in Figure 1C, the cyclo-
fen-OH ligand positions itself in a similar way to other known
ERa ligands.[24] Furthermore, the stabilization of each ligand in
654
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 653 – 663

Photocontrol of Protein Activity in Cells and Zebrafish
Scheme 2. Syntheses of 4-hydroxy-cyclofen (Ind) and its caged derivatives: a) cyclohexanone, TiCl4, Zn, THF, reflux, 2 h; b) Cl(CH2)2N(CH3)2·HCl, K2CO3, acetone/
H2O, reflux, 18 h; c) caging alcohol, P(Ph)3, diisopropylazo dicarboxylate, THF, sonication, 20 min.
yield)[25] and the resulting diphe-
nol was subsequently mono-
alkylated with 2-(dimethylami-
no)ethyl chloride hydrochloride
(40% yield) to provide the tar-
geted compound. In particular,
cyclofen-OH synthesis does not
require any delicate separation
after monoalkylation since the
intermediate 1 is symmetrical.
This is in contrast to tamoxifen-
OH synthesis, which involves
either the separation of Z and E
stereoisomers or the perfor-
mance of stereoselective reac-
tions.[26]
Figure 1. MD simulation snapshots. A) Representative structure of simulation C1 highlighting the position of the
cyclofen-OH ligand within the ligand binding domain. The cyclofen-OH ligand remains solvent accessible through
two accessibility sites: a large cleft, which could serve ligand binding and unbinding, and a channel that might
enable solvent to hydrate the binding site; B) superposition of representative ligand binding domain–ligand com-
Validation of the cyclofen-OH
inducer
plexes from simulations T1 (black), C1 (blue) and C2 (red). The actual values of the root mean square displace-
ments (RMSD) of the protein stabilized beyond 20 ns are low, with C1 showing the least deviation (1.4 ꢁ) and the
reference simulation T1 was around 2 ꢁ RMSD. The mutant C2 shows a slightly higher RMSD (2.9 ꢁ) induced by
the introduction of a longer side chain at position 351; C) zoom on the ligand binding site with key interacting
residues (Arg394, Glu353, Asp351, His524, and Leu525).
First, we tested cyclofen-OH in
CV1 cells transfected with a plas-
mid expressing the gfp–nls-ERT2
fusion gene. When observed by
epifluorescence microscopy 24 h
complex was assessed by its contacts with the receptor (see
the Supporting Information). All structural analysis, observed
binding mode, number of contacts and hydrogen bonds ex-
tracted from MD simulations eventually led us to adopt cyclo-
fen-OH as a putative ERT2 inducer since the ERT2–cyclofen-OH
complex should exhibit comparable binding properties with
respect to the reference tamoxifen-OH complex.
later, these cells displayed a weak cytoplasmic fluorescence
background with occasional nuclear fluorescence (Figure 2A).
Addition of cyclofen-OH (or tamoxifen-OH) to the cell culture
medium resulted in the disappearance of cytoplasmic fluores-
cence and a strong increase in nuclear fluorescence without al-
teration of the cell morphology (Figure 2B). As expected, the
release upon ligand binding of GFP–nls-ERT2 from its cytoplas-
mic chaperone complex permitted its nuclear translocation.
We measured the cumulative distribution of nuclear fluores-
cence intensities (Figure 2C) from which we extracted the
average values (Figure 2D; see the Experimental Section). In
the absence of ligand, that probability differed significantly
from zero only at low fluorescence intensities. Upon addition
Cyclofen-OH synthesis
Cyclofen-OH (Ind) was easily synthesized in two steps
(Scheme 2): 4,4’-hydroxybenzophenone and cyclohexanone
were first coupled under McMurry reaction conditions (85%
ChemBioChem 2010, 11, 653 – 663
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
655

D. Bensimon, S. Vriz, L. Jullien et al.
of cyclofen-OH, the fraction of
cells displaying strong nuclear
fluorescence increased with the
concentration of ligand and
yielded a larger average value.
Noticeably, the data revealed
that cyclofen-OH and tamoxifen-
OH induced similar fluorescence
distributions at 3 mm of ligand
(Figure 2C), a similarity expected
from the MD simulations of their
binding to the ERT2 receptor.
Thus, it appears that cyclofen-
OH is nontoxic and as efficient
as tamoxifen-OH in binding to
the ERT2 receptor of the GFP–nls-
ERT2 fusion protein and in acti-
vating its nuclear translocation.
We then checked whether ze-
brafish embryos developed nor-
mally when incubated in various
concentrations of cyclofen-OH
(up to 5 mm) and displayed a
similar response to cell cultures
when GFP–nls-ERT2 (or mCherry–
nls-ERT2) mRNA was injected (at
the one-cell stage). In the ab-
sence of ligand, the embryos dis-
played a very weak, overall fluo-
rescence signal at 30 h postferti-
lization (hpf; Figure 3A and at
dome stage in Figure S3A in the
Supporting Information). Upon
incubation in a medium contain-
ing the ligand, the percentage of
positive embryos (defined as
those exhibiting nuclear fluores-
cence at 24 hpf) increased (Fig-
ure 3B and Figure S3B in the
Figure 2. Induction of nuclear translocation of GFP–nls-ERT2 by addition or photorelease of cyclofen-OH in CV1
cells transfected with gfp–nls-ERT2 plasmid. GFP fluorescence image of CV1 cells 24 h after transfection and further
incubation with: A) 0, or B) 3 mm Ind for 1 h (identical display range in A and B). C) Cumulative distribution after
~
&
*
*
0.0 ( ), 0.3 ( ), 3.0 ( ) and 5.0 mm (<) Ind, or 3.0 mm ( ) tamoxifen-OH treatment. D) Average value of the mean
nuclear intensity; error bars represent standard errors of the mean. GFP fluorescence image of CV1 cells imaged
24 h after transfection and further incubation in 6 mm cInd for 30 min, E) without, and F) with global 365 nm UV il-
lumination; scale bar: 10 mm.
Figure 3. Induction of nuclear translocation of GFP–nls-ERT2 by addition of cyclofen-OH in wild-type zebrafish em-
Supporting Information) reach- bryos injected with gfp–nls-ERT2 mRNA (100 ngmLꢀ1) at the one-cell stage. GFP fluorescence image of embryos at
24–30 hpf after incubation with: A) 0, or B) 3 mm Ind. Note that the camera sensitivity was intensified by a factor
ing 50% at an inducer concen-
of 6 in A) with respect to B). C) Dependence of the efficiency of nuclear translocation of GFP–nls-ERT2 on inducer
p
tration, C1/2 =0.2 mm (Figure 3C
and Table S2 in the Supporting
Information).
concentration. Error bars are statistical errors estimated as (p(1ꢀp)/n), where p is the percentage of embryos
exhibiting a positive phenotype, and n is the total number of embryos investigated. The solid line is a guide for
the eyes; scale bar: 100 mm.
Caged cyclofen-OH: syntheses
and in vitro photochemical properties
the temporal dependence of the concentration of photore-
leased cyclofen-OH. With UV illumination at 365 nm, we ob-
served that the inducer Ind could be photoreleased quantita-
tively from its three caged precursors cInd, c’Ind, and c“Ind
with characteristic times of 270, 120, and 620 s, respectively,
corresponding to uncaging cross-sections with one-photon ex-
citation at 365 nm of 22, 47, and 10mꢀ1 cmꢀ1 (see the Support-
ing Information). We found that the cInd and c”Ind caged pre-
cursors were inert in the dark and led to photocontrol in ze-
brafish embryos upon UV illumination. In contrast, c’Ind led to
protein activation in zebrafish embryos, even in the absence of
Following the validation of cyclofen-OH as an efficient ERT2
ligand and its possible use to control protein activity in cell
cultures and zebrafish embryos, we caged it with 4,5-di-
methoxy-2-nitrobenzyl alcohol, 6-bromo-7-hydroxy-4-hydroxy-
methyl-coumarin[27] and 7-dimethyl-amino-4-hydroxymethyl-
coumarin[28] under Mitsonobu conditions to give cInd, c’Ind,
and c“Ind with 64, 45, and 32% yields, respectively (Scheme 2).
We characterized the uncaging of these compounds, in
vitro, using HPLC coupled to mass spectrometry to measure
656
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 653 – 663

Photocontrol of Protein Activity in Cells and Zebrafish
illumination. In the following we will focus on the easily acces-
sible cInd. We showed that cyclofen-OH could be released
from cInd with two-photon illumination at 750 nm with an
uncaging cross-section equal to 4 mGM (see the Supporting In-
formation) in agreement with values reported for the photola-
bile 4,5-dimethoxy-2-nitrobenzyl protecting group.[27,29] Al-
though modest, this uncaging cross-section is sufficient to ach-
ieve cInd uncaging at the 10 s time scale in a cell volume at
nondetrimental laser powers (10–20 mW with our 40ꢂ micro-
scope objective; see the Supporting Information).
nuclear translocation time tint =(1000ꢁ300) s. In contrast, at
the same time scale, the fluorescence level remains fairly con-
stant in the reference cell as well as in a control experiment in
which a cell not incubated with cInd was submitted to the
same laser exposure (wavelength, power and duration) as in
the targeted cell in Figure 4A.
We eventually studied the kinetics of translocation from the
fluorescence correlation spectroscopy (FCS) signal of the cyto-
plasmic GFP–nls-ERT2. We recorded time-series of FCS curves
following uncaging of cInd (6 mm) in targeted cells (Figure S6
in the Supporting Information). The FCS curves have been
globally analyzed as a function of the time (t) after cyclofen-
OH release by adopting a model in which the illumination
volume is assumed to contain two fluorescent species with dif-
fusion constant Di (i=1, 2): a GFP–nls-ERT2 unbound state 1
and a chaperone bound complex 2.[31] The FCS curve can then
be written as [Eq. (1)]:
Caged cyclofen-OH induced protein photoactivation in cells
In the absence of UV illumination, the fluorescence of CV1 cells
transfected with a plasmid carrying a gfp–nls-ERT2 fusion gene
incubated in 6 mm cInd for 30 min was essentially the same as
that observed in cells incubated without inducer (compare Fig-
ure 2A and E). This result shows that the caged ligand cInd
was inactive and nontoxic. On the other hand, when cInd was
illuminated by UV for a duration similar to its uncaging time,
the cells displayed the characteristic nuclear fluorescence they
show in the presence of ligand (compare Figure 2B and F).
We also used short light pulses emitted by a 365 nm UV
laser or by a Ti-Sa laser at 750 nm in a nondetrimental[30]
power regime (5 s at 5 mW and 10 s at 10–20 mW at the
sample) to photorelease Ind and analyze the translocation
kinetics of GFP–nls-ERT2 in a single cell with one- and two-
photon excitation, respectively. In all cases, the cytoplasm and
the nucleus of the targeted illuminated cell exhibited much
faster brightness changes than surrounding cells: they become
dimmer and brighter, respectively. This behavior is evidenced
in Figure 4A and B by comparing the fluorescence levels in the
cytoplasm and in the nucleus of the targeted cell by using the
nearby unilluminated cell as a reference. Figure 4C shows that
the average nuclear fluorescence intensity of a targeted cell in-
creases exponentially with time, leading to an estimate of the
ꢀ
ꢁ
2
X
QiNi
2
ð1Þ
GðtÞ ¼
P
Giðt;Ni;tiÞ
2
i¼1
i¼1 QiNi
with [Eq. (2)]:
ꢀ
ꢁꢀ1
ꢀ
ꢁꢀ
1
2
1
t
t
Giðt;Ni;tiÞ ¼
1 þ
1 þ w2
ð2Þ
ti
ti
Ni
where Ni is the average number of molecules in state i, con-
tained in the illumination volume, Qi and ti (=wxy2/4Di) their
brightness and diffusion time through the beam waist (wxy)
and w=wz/wxy the aspect ratio of the illuminated volume. Sat-
isfactory fits were obtained with various (t1,t2) combinations
without significantly altering characteristic decay times. To
reduce uncertainties, we fixed ti =1.7 ms, a value contained in
the range allowed by the previous fits and derived for GFP–
nls-ERT2 from the GFP diffusion coefficient (25 mm2 sꢀ1 [32]
by
)
taking into account the difference of molecular weights
(66.6 kDa for GFP–nls-ERT2 and 27 kDa for GFP).[33] Then we per-
formed a satisfactory global fit of the whole series of FCS auto-
correlation curves (Figure S6 in the Supporting Information).
From these fits the values of [Eq. (3)]:
Q2NiðtÞ
Q1N1ðtÞi þ Q2N2ðtÞ
Giðt ¼ 0; tÞ ¼ ꢂ
ꢃ
ð3Þ
2
at various times (t) were determined. It turns out that the ratio
G1(0,t)/G2(0,t) is constant (Figure 5A), which implies that the
ratio of bound to unbound species is also constant. We thus
deduced that the equilibrium between these two states occurs
on a time scale faster than the one associated to nuclear trans-
location. We correspondingly observed similar characteristic
times associated to the exponential increase of Gi(0,t) (Fig-
ure 5B and C). From G1(0,t)+G2(0,t), which is inversely propor-
tional to the number of cytoplasmic GFP–nls-ERT2 molecules,
we eventually deduced that nuclear translocation happens on
a timescale tn =(450ꢁ20) s (Figure 5D) that is in the same
order of magnitude as the value deduced from the increase in
nuclear fluorescence and in line with published estimates.[34]
Figure 4. Selective nuclear GFP–nls-ERT2 translocation in one CV1 cell (indi-
cated by an arrow in A) upon two-photon illumination (750 nm, 10 mW for
10 s). Fluorescence images of the targeted cell and a nearby unilluminated
cell: A) 0, and B) 60 min after illumination. C) Time evolution of the mean nu-
*
*
)
clear fluorescence intensity in the targeted cell ( ) and the reference cell (
shown in A) and B). The stars depict the corresponding evolution for a cInd
nonincubated control cells, which were illuminated similarly to the targeted
cell in A). The CV1 cells were transfected with a gfp–nls-ERT2 plasmid, further
incubated in 6 mm cInd for 30 min and washed before illumination; scale
bar: 10 mm.
ChemBioChem 2010, 11, 653 – 663
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
657

D. Bensimon, S. Vriz, L. Jullien et al.
Upon uncaging of cInd by UV illumination of the whole
embryo at 4–5 hpf, a global nuclear translocation of the GFP-
fused protein was observed, in line with the cell culture data
(Figure 6B). We studied the dependence of GFP nuclear trans-
location on the duration of UV illumination in more detail
(Table S3 and Figure S5 in the Supporting Information). As
anticipated, the translocation yield increased with illumination
duration. Using the dependence observed in Figure 3C for cali-
bration, the data also suggested that the concentration of cInd
in the embryo was in the same range as its concentration in
the incubation solution (see the Supporting Information).
Then, to verify that the photocontrol of protein activity did
locally occur around targeted illuminated cells, we performed
a colocalization experiment involving two caged compounds
sharing the same caging group: cInd and caged fluorescein
dextran (cFd). We used a UV laser (5 mW) to perform a local
photoactivation (for 5 s) of cInd and cFd in wild-type embryos
injected at the one-cell stage with cFd and gfp–nls-ERT2 mRNA
and incubated in embryo medium supplemented with 3 mm
cInd. From the different emission spectra of GFP and fluores-
cein, their contributions to the fluorescence signal were sepa-
rated by analyzing images recorded at different wavelengths.
Figure 7A–C displays images of a zebrafish embryo (60 min
after illumination) that were obtained from recording the emis-
sion from GFP, fluorescein, and both fluorophores, respectively.
One notices the cytoplasmic localization of fluorescein in the
group of cells in which the GFP–nls-ERT2 signal is predominant-
ly localized in the nucleus.
Figure 5. Analysis of the FCS autocorrelation curves of cytoplasmic GFP in a
CV1 cell transfected with a gfp–nls-ERT2 plasmid, incubated in 6 mm cInd for
30 min and illuminated for 5 s with a UV laser (375 nm, 5 mW; see also Fig-
ure S6 in the Supporting Information). Time dependence of: A) G1(t=0,t)/
G2(t=0,t), B) G1(t=0,t), C) G2(t=0,t), and D) 1/[G1(t=0,t)+G2(t=0,t)] extracted
from a two-species model with t1 =1.7 ms and t2 =240 ms. Dots: experi-
mental points; solid line: fit to an exponential decrease of the cytoplasmic
species (due to nuclear translocation).
Caged cyclofen-OH-induced protein photoactivation in
zebrafish embryos
We then investigated the photoinduction of nuclear transloca-
tion in live zebrafish embryos. Wild-type embryos were inject-
ed with gfp–nls-ERT2 mRNA at the one-cell stage and incubated
(at 1 to 2 hpf) for 90 min in embryo medium supplemented
with 3 mm cInd. As observed in cell cultures, cInd was inactive
and nontoxic. In comparison with embryos incubated in regu-
lar medium, it did not induce a significantly larger GFP nuclear
translocation and mortality (Figure 6A and Table S3 in the Sup-
porting Information).
Figure 7. Images resulting from local photoactivation of GFP–nls-ERT2 in a
wild-type embryo injected with gfp–nls-ERT2 mRNA and caged fluorescein
dextran at the one-cell stage and conditioned as in Figure 6A and B. Emis-
sion from: A) GFP, B) fluorescein, and C) both fluorophores; scale bar:
100 mm.
We also performed a second colocalization experiment with
two-photon illumination. Our aim here was to demonstrate
the colocalization of the two fluorescent proteins, GFP–nls-ERT2
and mCherry–nls-ERT2, the latter acts as a reporter of the illumi-
nated cell. Figure 8A–C displays the distribution of fluores-
cence intensity in a wild-type embryo injected with gfp–nls-
ERT2 and mCherry–nls-ERT2 mRNA at the one-cell stage. After in-
cubation from the dome stage to 24 hpf in embryo medium
supplemented with 3 mm cInd and subsequent washing, the
embryo was illuminated at 24 hpf in the tail for 10 s with a Ti-
Sa laser (750 nm, 10 mW). After 60 min, we observed a marked
increase in fluorescence from a single nucleus in both green
and red fluorescence channels, indicating that nuclear translo-
cation of both GFP–nls-ERT2 and mCherry–nls-ERT2 occurred in
Figure 6. Induction of nuclear translocation of GFP–nls-ERT2 by photorelease
of cyclofen-OH in wild-type zebrafish embryos injected with gfp–nls-ERT2
mRNA (100 ngmLꢀ1) at the one-cell stage. Confocal GFP fluorescence image
of embryos at 4 hpf (following incubation at 2 hpf for 90 min in 3 mm cInd
and washing): A) without illumination, and B) 30 min after illumination with
UV light; scale bar: 100 mm.
658
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 653 – 663

Photocontrol of Protein Activity in Cells and Zebrafish
is compatible with a wide variety of proteins and could open
up opportunities for the local spatiotemporal investigation of
developmental pathways,[35,36] the identification of stem cells
and the study of cancer in a live organism.
Figure 8. Images resulting from local photoactivation (750 nm, 10 mW, 10 s)
of GFP–nls-ERT2 and mCherry–nls-ERT2 in the tail at 24 hpf in a wild-type
embryo. A) GFP, and B) mCherry, and C) their superposition evidences the
colocalization of these proteins after photoactivation; D) the images are in-
tensity coded; scale bar: 100 mm.
Experimental Section
Molecular dynamics simulation: MD simulations were run for
30 ns with the YASARA program.[37] Force-field parameterization for
tamoxifen-OH and cyclofen-OH ligands was carried out by using
the AutoSMILES procedure, otherwise the AMBER99 forcefield was
used.[38] All systems were solvated with 7936 explicit TIP3P water
molecules and 20Na+ and 22Clꢀ counterions were added as back-
ground salt and to preserve overall electrical neutrality. Each
system was energy minimized by using the steepest descent
method to relax any steric conflicts before beginning the simula-
tions. Simulations were carried out with periodic boundary condi-
tions. Long-range electrostatic interactions were calculated by
using PME with a direct-space cut-off of 7.86 ꢁ. All simulations
were performed by using an NVT ensemble at 298 K. A 2 fs/1 fs
double-integration time step was used. Graphics were prepared
with VMD.[39] Standard conformational analysis was carried out by
using YASARA, Gromacs tools[40] and locally written code. Statistical
and data analysis was performed by using the R statistical software
package[41] and Xmgrace.[42]
a single cell of the embryo. To rule out the possibility that ob-
serving a bright spot was due to statistical variation rather
than photoactivation, we performed a statistical analysis on
the distributions of fluorescence intensity from GFP–nls-ERT2
and mCherry–nls-ERT2 in the images of the fish tail displayed in
Figure 8A and B (see the Supporting Information). As shown in
Figure S7 in the Supporting Information, both normalized his-
tograms of fluorescence levels exhibit a smooth distribution at
low fluorescence intensity, corresponding to the population of
background cells and therefore acting as an internal control,
and a spike at much higher intensity associated to the bright
cell. From the intensity distributions, we computed their mean
(mg,mr) and standard deviations (sg,sr) and noticed that the in-
tensity of the green and red fluorescence signal in the bright
cell was observed at 16.7sg and 10.6sr, respectively, from the
distribution means. From these observations, we estimate that
the probability of observing a random fluctuation of fluores-
cence in both channels is smaller than exp(ꢀ50)~10ꢀ22, which
is negligibly small, even after multiplying by the number of
cells in the tail. We thus deduce that this highly localized fluo-
rescence of both GFP–nls-ERT2 and mCherry–nls-ERT2 in the nu-
cleus of a single cell results from the local two-photon uncag-
ing of cInd, which was indeed performed in the area where
we observed this increased localized fluorescence 60 min later
under a confocal microscope.
Syntheses
General: 4,5-Dimethoxy-2-nitrobenzyl alcohol was purchased from
Sigma. 6-Bromo-7-hydroxy-4-hydroxymethyl-coumarin and 7-dime-
thylamino-4-hydroxymethylcoumarin were synthesized according
to published procedures.[43,44] Commercially available reagents
were used as obtained. Microanalyses were performed by the Ser-
vice de Microanalyses de Gif sur Yvette. Melting points were deter-
mined with a Bꢃchi 510 capillary apparatus. The 1H and 13C NMR
spectra were recorded at room temperature on Bruker AM 250 or
400 spectrometers; chemical shifts are reported in ppm with pro-
tonated solvent as internal reference (1H, CHCl3 in CDCl3 7.26 ppm,
CHD2OD in CD3OD 3.31 ppm, CHD2SOCD3 in CD3SOCD3 2.49 ppm,
CHD2COCD3 in CD3COCD3 2.05 ppm; 13C, 13CDCl3 in CDCl3 77.0,
13CD3OD in CD3OD 49.1 ppm, 13CD3SOCD3 in CD3SOCD3 39.7 ppm,
13CD3COCD3 in CD3COCD3 29.9 ppm); coupling constants (J) are
given in Hz. Mass spectra (chemical ionization and high resolution
with NH3 or CH4) were performed by the Service de Spectromꢄtrie
de Masse de l’ENS (Paris). Column chromatography was performed
on silica gel 60 (0.040–0.063 nm; Merck). Analytical thin-layer chro-
matography (TLC) was conducted on Merck silica gel 60 F254 pre-
coated plates. HPLC analyses and purifications of the final caged
species were performed on a Waters system incorporating a
Wdelta 600 pump with a PDA 996 UV detector working at 245 nm
(columns: analytical: X-Terra Waters MS C18, 150ꢂ4.6 mm, 5 mm,
1 mLminꢀ1 flow; preparative: X-Terra Waters Prep MS C18, 150ꢂ
19 mm, 5 mm, 12 mLminꢀ1 flow; elution with solution A: water
with 0.05% formic acid; and solution B: acetonitrile with 0.05%
formic acid).
Conclusions
We have shown that photoreleasing cyclofen-OH from a caged
precursor is an efficient strategy to restore the function of a
protein fused to the ERT2 receptor and investigated its dynami-
cal effects in cultured cells and in live zebrafish embryos. Co-
localization experiments, with both UV and two-photon excita-
tions, suggest that the light-induced activation of proteins can
be restricted to a small group of targeted cells. This observa-
tion is significant when evaluating the feasibility of the photo-
control of protein activity down to the single cell level by
using the present approach. From the data of the cyclofen-OH
dose-response curve, the kinetics of uncaging and an estimate
of the cyclofen-OH concentration within the embryo, one can
set the two-photon illumination conditions to deliver and tune
the concentration of cyclofen-OH in the targeted cell to a level
slightly below the saturation of its receptor. Any leakage of the
inducer in the extracellular medium and the neighboring cells
would then be at too low a concentration to turn on protein
activity in those cells. The present noninvasive optical method
4-(Cyclohexylidene(4-hydroxyphenyl)methyl)phenol (1):[25] Titanium
chloride (6.2 mL, 56 mmol) was added dropwise under argon to a
stirred suspension of zinc powder (8.20 g, 126 mmol) in dry tetra-
hydrofuran (80 mL) at ꢀ108C. When the addition was complete,
the mixture was warmed to room temperature and then refluxed
for 2 h. A solution of 4,4’-hydroxybenzophenone (2.0 g, 9 mmol)
and cyclohexanone (4 mL, 38 mmol) in dry tetrahydrofuran
(120 mL) was added to the cooled suspension of the titanium re-
ChemBioChem 2010, 11, 653 – 663
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
659

D. Bensimon, S. Vriz, L. Jullien et al.
agent at 08C and the mixture was refluxed for 2 h. After being
cooled to room temperature, the reaction mixture was quenched
with 10% (w/v) aqueous potassium carbonate (30 mL), filtered and
extracted with ethyl acetate. The organic layer was washed with
brine, dried over MgSO4, and concentrated in vacuo. Flash column
chromatography (cyclohexane/ethyl acetate, 3:1, v/v) afforded 1 as
a white powder (2.15 g, 85%). 1H NMR (250 MHz, [D6]DMSO): d=
9.28 (s, 2H), 6.82 (d, J=8.4 Hz, 4H), 6.65 (d, J=8.4 Hz, 4H), 2.14 (m,
4H), 1.52 (m, 6H); 13C NMR (100 MHz, [D6]DMSO): d=155.5, 136.2,
133.8, 133.6, 130.3, 114.6, 31.9, 28.1, 26.3.
vent was removed in vacuo. Purification by column chromatogra-
phy with dichloromethane/methanol (9:1, v/v) as eluent yielded
1
c’Ind (60 mg, 30%). H NMR (250 MHz, [D6]DMSO): d=8.00 (s, 1H),
7.09–6.75 (m, 9H), 6.39 (s, 1H), 5.33 (s, 2H), 4.06 (t, J=5.6 Hz, 2H),
2.75 (t, J=5.6 Hz, 2H), 2.32 (s, 6H), 2.17 (m, 4H), 1.55 (s, 6H);
13C NMR (100 MHz, CDCl3): d=160.2, 157.0, 156.2, 155.8, 154.1,
149.1, 138.7, 137.3, 135.6, 133.0, 131.0, 130.7, 127.5, 114.1, 113.8,
112.6, 111.9, 108.4, 104.0, 65.7, 65.5, 58.1, 45.7, 32.4, 28.6, 26.7; MS
(CI, NH3): m/z 606.604 (calcd average mass for C33H3581BrNO5:
606.19, C33H3579BrNO5: 604.16); MS (CI, NH3, HR): m/z 606.1677
(calcd average mass for C33H3581BrNO5: 606.1684), 604.1693 (calcd
average mass for C33H3579BrNO5: 604.1699). After preparative HPLC
purification (elution profile: 0–5 min: 50% A and 50% B; 5–15 min:
10% A and 90% B), c’Ind was shown by analytical HPLC to contain
less than 2% residual of Ind.
4-((4-(2-(Dimethylamino)ethoxy)phenyl)(cyclohexylidene)methyl)-
phenol (Ind; 4-hydroxy-cyclofen): 2-(Dimethylamino)ethylchloride
hydrochloride (256 mg, 1.78 mmol) was dissolved in a solution of
acetone/water (19:1 v/v, 40 mL) and treated with potassium car-
bonate (600 mg, 4.4 mmol). The mixture was stirred at 08C for
30 min. Compound 1 (500 mg, 1.78 mmol) was dissolved in the
above solution at 08C and potassium carbonate (580 mg,
4.2 mmol) was added. The mixture was refluxed for 18 h. The
solids were filtered off, and the filtrate was concentrated in vacuo.
The crude product was purified by flash column chromatography
(dichloromethane/methanol, 9:1, v/v) to afford Ind as a white
powder (250 mg, 40%). m.p.: 180–1818C (methanol); 1H NMR
(250 MHz, CDCl3): d=6.95–6.89 (m, 4H), 6.70 (AA’XX’, J=8.5 Hz,
2H), 6.59 (AA’XX’, J=8.5 Hz, 2H), 4.04 (t, J=5.5 Hz, 2H), 2.82 (t, J=
5.5 Hz, 2H), 2.40 (s, 6H), 2.25–2.21 (m, 4H), 1.57 (m, 6H); 13C NMR
(100 MHz, CDCl3): d=156.6, 154.7, 137.9, 136.1, 135.0, 133.5, 131.0,
130.8, 115.0, 113.5, 64.6, 58.1, 45.3, 45.2, 32.5, 32.4, 28.6, 26.8; MS
(CI, CH4): m/z 352 (calcd average mass for C23H29NO2: 351.22); MS
(CI, CH4, HR): m/z 352.2271 (calcd average mass for C23H30NO2:
352.2277); elemental analysis calcd (%) for C23H29NO2·0.5H2O
(360.49): C 76.63, H 8.39, N 3.89; found: C 76.95, H 8.09, N 3.87.
4-((4-((4-(2-(Dimethylamino)ethoxy)phenyl)(cyclohexylidene)methyl)-
phenoxy)methyl)-7-dimethylaminocoumarin (c“Ind): This was ob-
tained in the same way as cInd by using Ind (200 mg, 0.57 mmol),
7-(dimethylamino)-4-(hydroxymethyl)-2H-chromen-2-one[44]
(125 mg, 0.57 mmol), triphenylphosphine (150 mg, 0.57 mmol), di-
isopropylazodicarboxylate (0.112 mL, 0.57 mmol) and tetrahydrofur-
an (0.3 mL). Purification by column chromatography with dichloro-
methane/methanol (9:1, v/v) as eluent, yielded c’’Ind (100 mg,
1
32%). H NMR (250 MHz, CDCl3): d=7.38 (d, J=8.9 Hz, 1H), 7.04–
6.74 (m, 8H), 6.61 (dd, J=2.4, 8.9 Hz, 1H), 6.53 (d, J=2.4 Hz, 1H),
6.32 (s, 1H), 5.12 (s, 2H), 4.15 (t, J=5.2 Hz, 2H), 3.05 (s, 6H), 2.91 (t,
J=5.2 Hz, 2H), 2.48 (s, 6H), 2.22 (m, 4H), 1.57 (m, 6H); 13C NMR
(63 MHz, CDCl3): d=161.9, 157.0, 156.2, 156.0, 152.8, 150.5, 138.6,
137.0, 135.8, 133.2, 131.0, 130.9, 124.3, 114.1, 113.9, 108.9, 107.7,
106.8, 98.4, 65.9, 65.7, 58.2, 45.7, 40.1, 32.5, 28.6, 26.8; MS (CI, CH4):
m/z 553 (calcd average mass for C35H41N2O4: 553.3; MS (CI, CH4,
HR): m/z 553.3065 (calcd average mass for C35H41N2O4: 553.3066).
After preparative HPLC purification (elution profile: 0–50 min: 15%
A and 85% B), c”Ind was shown by analytical HPLC to contain less
than 2% residual of Ind.
2-(4-((4-(4,5-Dimethoxy-2-nitrobenzyloxy)phenyl)cyclohexylidene)me-
thyl)phenoxy)-N,N-dimethylethanamine (cInd):[45] A solution of Ind
(180 mg, 0.5 mmol), 4,5-dimethoxy-2-nitrobenzyl alcohol (115 mg,
0.54 mmol) and triphenylphosphine (140 mg, 0.54 mmol) in tetra-
hydrofuran (0.25 mL) was sonicated for several minutes to allow
for mixing. Diisopropylazodicarboxylate (0.106 mL, 0.54 mmol) was
added dropwise to the resulting viscous solution during sonication.
After sonicating for 20 min, the reaction mixture was concentrated
in vacuo and purified by column chromatography on silica gel
with dichloromethane/methanol (9:1, v/v) as eluent to give cInd
(180 mg, 64%); m.p.: 116–1178C (isopropylether, yellow crystals);
1H NMR (400 MHz, CDCl3): d=7.76 (s, 1H), 7.35 (s, 1H), 7.04
(AA’XX’, J=8.7 Hz, 2H), 7.00 (AA’XX’, J=8.7 Hz, 2H), 6.90 (AA’XX’,
J=8.7 Hz, 2H), 6.82 (AA’XX’, J=8.7 Hz, 2H), 5.47 (s, 2H), 4.03 (t, J=
5.8 Hz, 2H), 3.96 (s, 3H), 3.94 (s, 3H), 2.70 (t, J=5.8 Hz, 2H), 2.32 (s,
6H), 2.23 (m, 4H), 1.58 (m, 6H); 13C NMR (100 MHz, CDCl3): d=
157.0, 156.2, 153.8, 147.7, 138.9, 138.5, 136.8, 137.7, 133.2, 131.0,
130.7, 129.6, 114.3, 113.8, 109.4, 107.9, 67.0, 65.8, 58.3, 56.3, 56.3,
45.8, 32.4, 28.6, 26.8; elemental analysis calcd (%) for C32H38N2O6
(546.66): C 70.32, H 7.00, N 5.12; found: C 70.11, H 7.24, N 5.02.
After preparative HPLC purification (elution profile: 0–5 min: 50%
A and 50% B; 5–15 min: 10% A and 90% B), cInd was shown by
analytical HPLC to contain less than 2% residual of Ind.
HPLC coupled to mass spectrometry: High pressure liquid chro-
matography was carried out with an Accela system liquid chroma-
tograph (Thermo Finnigan, Les Ulis, France) equipped with a Hy-
persil gold column (1.9 mmꢂ2.1ꢂ50 mm) connected to a Thermo-
Finnigan TSQ Quantum Discovery Max triple quadrupole mass
spectrometer. Sample solution (5 mL) was injected in the chromato-
graphic column. For photodeprotection studies, the samples were
eluted in isocratic mode at a flow rate of 400 mLminꢀ1 with a
water/acetonitrile mixture (60–40%, v/v) containing 0.05% formic
acid. For the isomerization study, the separation of isomers was
carried out by gradient elution at a flow rate of 400 mLminꢀ1 from
5% to 95% water/methanol mixtures containing 0.05% formic acid
in 20 min. Between each injection, the column was equilibrated
with the mobile phase for 5 min. After separation, the analytes
were introduced in the mass spectrometer through a heated elec-
trospray ionization source (508C) operating in the positive mode.
The temperature of the capillary transfer was set at 2708C. Nitro-
gen was employed as nebulizing (35 psi) and auxiliary gas (30 arbi-
trary units). Argon was used as collision gas (1.0 milliTorr in Q2). 4-
Hydroxy-cyclofen (Ind) was observed (ion spray voltage of 3000 V)
in the single reaction monitoring (SRM) mode (m/z 352.2!72) by
using 30 V collision energy and 130 V tube lens. The stereoisomers
of 4-hydroxytamoxifen and the corresponding dehydrophenanth-
rens (ion spray voltage of 3000 V) were followed in the single ion
monitoring (m/z 388.3) mode by using 175 V tube lens. All the pos-
sible settings were optimized by repetitive injections of the analyte
in the chromatographic system. Instrument control and data col-
4-((4-((4-(2-(Dimethylamino)ethoxy)phenyl)(cyclohexylidene)methyl)-
phenoxy)methyl)-6-bromo-7-hydroxycoumarin (c’Ind): This was ob-
tained in the same way as cInd by using Ind (110 mg, 0.31 mmol),
6-bromo-7-methoxymethoxy-4-hydroxycoumarin[46]
(100 mg,
0.31 mmol), triphenylphosphine (86 mg, 0.33 mmol), diisopropyla-
zodicarboxylate (0.065 mL, 0.33 mmol) and tetrahydrofuran
(0.2 mL). After sonication, trifluoroacetic acid (2 mL) was added and
the mixture was stirred at room temperature for 15 min. The sol-
660
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 653 – 663

Photocontrol of Protein Activity in Cells and Zebrafish
lection were handled by a computer equipped with Xcalibur soft-
ware (version 2.0).
various ligands or their caged precursors, CV1 cells were incubated
at various concentrations of these substrates (0–5 mm) for different
durations (15–60 min) and fixed in PFA (4%) before imaging. In un-
caging assays of cInd, the molecule was added 24 hpt for 15 to
30 min before illumination. Expression of GFP–nls-ERT2 was assayed
24 h after transfection by imaging the GFP fluorescence.
Methods for experiments in cells and zebrafish embryos
Application of the inducers and their caged precursors: Tamoxifen-
OH, cyclofen-OH (Ind) and its various caged precursors (cInd, c’Ind,
c“Ind) were solubilized at 10 mm in DMSO; these stock solutions
were stored at ꢀ208C and proved stable for several months. Ali-
quots of those solutions were added to the aqueous solutions con-
taining the cells and the zebrafish embryos to reach the concentra-
tions indicated in the text. The dilution factor exceeded 103 and
we did not notice any effect when the same volume of pure
DMSO was added to the biological samples in control experiments.
After incubation with the caged precursors, the cells and the ze-
brafish embryos were washed with plain medium prior to illumina-
tion to avoid any possible interference, which could originate from
photoreleasing the inducer in the incubating solution.
Zebrafish embryo experiments: Wild-type zebrafish embryos were
injected at the one-cell stage with the appropriate mRNA synthe-
sized with an in vitro transcription kit (mMessage mMachine,
Ambion). They were subsequently dechorionated by pronase treat-
ment at dome stage prior to incubation in an aqueous solution of
the various substrates (up to 30 hpf with Ind, for 90 min with cInd
except for the experiment displayed in Figure 8, which was up to
24 hpf). Ind was photoreleased from cInd 4–5 hpf (except in
Figure 8, 24 hpf). Illuminated embryos were observed 30 min
(60 min in Figure 8) after illumination for GFP–nls-ERT2 or mCherry–
nls-ERT2 fluorescence. Embryos positive for either GFP or mCherry
nuclear translocation were scored under a microscope in a double-
blind protocol.
The gfp–nls-ERT2 and mCherry–nls-ERT2 coding plasmids: An inter-
mediate vector (pC5fiERT) was first constructed by inserting Cre-
ERT2 coding sequence (fragment XhoI–EcL136I from pCre-ERT2,[47]
a
Image acquisition and analysis: A fluorescence microscope (Olym-
pus BX51WI) equipped with a Luca CCD Camera (Andor technolo-
gies) was used for image acquisition of the cells (filters: U-MWIBA3
Olympus for GFP and U-MWIG3 Olympus for mCherry). In a given
series of experiments, all the conditions (EM gain, exposure dura-
tion, lamp power, etc.) were identical to allow for a comparison of
the observed fluorescence intensities. The images of embryos were
acquired by using confocal microscopes Leica TCS SP2 AOBS or
Zeiss Axiovert 200M LSM510-Meta. To obtain Figure 2C and D, we
analyzed the images recorded at all the Ind concentrations by ex-
tracting the average nuclear intensity (I) from each cell (at least
100 cells were analyzed at each Ind concentration) with a home-
developed LabView program by using IMAQ VISION (a rectangular
region of interest (roi) was manually drawn across each cell in each
image. The IMAQ AutoBThreshold vi (virtual instrument: LabView
program) was then used to compute the optimal threshold value
for each roi—with a single cell inside a roi—to segment the nucle-
ar region from the rest. We subsequently plotted the resulting his-
togram I/Imax after normalization relative to the intensity Imax of the
brightest cell nucleus across all the cells and all the Ind concentra-
tions. At a given Ind concentration, the cumulative distribution
function P(x>I/Imax), which represents the probability that the
random variable I/Imax takes on a value less than or equal to x, was
subsequently obtained by integrating the histogram from 0 to I/
Imax and normalizing such that P(x>0)=1 (Figure 2C). Averaging
the histogram was correspondingly performed to extract the aver-
aged values shown in Figure 2D. For the analysis of the pheno-
types resulting in Figure 3C and Figure S5 in the Supporting Infor-
mation, we considered as positive those embryos exhibiting a
spotty fluorescence associated to fluorescence localized in the
nuclei. For the analysis of localization and colocalization of GFP
and fluorescein displayed in Figure 7A–C, we performed multi-
channel fluorescence imaging (excitation: 488 nm; emission chan-
nels: 500–512, 512–522 and 530–600 nm). Using reference spectra
acquired from embryos labeled with GFP or fluorescein only, and
assuming the total measured fluorescence in any channel to linear-
ly reflect the individual contributions of the fluorophores, we used
a home-written Matlab code to retrieve the fractional intensities of
GFP and fluorescein.
kind gift from Daniel Metzger and Pierre Chambon, Strasbourg,
France) into pcDNA5/FRT (Invitrogen, digested with XhoI and
EcoO109I, blunted with Klenow). We then prepared pC5fGFPERT,
which contained the gfp–nls-ERT2 coding sequence downstream of
the CMV and T7 promoters and upstream the bovine growth hor-
mone polyadenylation signal, by trimolecular ligation between the
NheI–BsrGI fragment from pEGFP-C1 (Invitrogen), pC5fiERT cleaved
by NheI and XhoI, and a BsrGI–XhoI adaptor (sense: 5’-GTA CAG
GAT CCC CAA GAA GAA GCG CAA GGT GGC GCC CGG GC-3’; anti-
sense: 5’-TCG AGC CCG GGC GCC ACC TTG CGC TTC TTC TTG GGG
ATC CT-3’). The pC5fGFPERT plasmid codes for a 567 amino acid
(aa) long fusion protein (1–238: eGFP[aa1–aa238]; 239–253: RIP-
SV40nls-APGLEP; 254–567: hESR1[aa282–aa595, with three muta-
tions G400V, M543A, L544A]). This construct was used for expres-
sion in CV1 cells, and a second plasmid was built by transferring
the NheI–PmeI fragment of pC5fGFPERT into pCS2+ (a kind gift
from Richard M. Harland, Berkeley, California, cleaved by XbaI and
SnaBI). The new construct, pCSGFPnERT, which contained the same
gfp–nls-ERT2 coding sequence downstream simian CMV and SP6
promoters and upstream the late SV40 polyadenylation signal, was
used to prepare synthetic RNA in vitro for injection into zebrafish
embryos. Plasmid pCSmChnERT was obtained by substituting
mCherry for the GFP sequence, through insertion of the BamHI–
SgrAI fragment of pREST-mCherry (a kind gift from Roger Y.
Tsien,[48] San Diego, California) into pCSnERT (a derivative of
pCSGFPnERT prepared by inserting an adaptor between BamHI
and SmaI sites; sense: 5’-GAT CCTA GGC TAG CAC CGG CGG CAT
GGA CGA GCT GCT GTA CAA ATC GAT CCC CAA GAA GAA GCG
CAA GGT GGC GCC C-3’, antisense: 5’-GGG CGC CAC CTT GCG CTT
CTT CTT GGG GAT CGA TTT GTA CAG CAG CTC GTC CAT GCC GCC
GGT GCT AGC CTA G-3’) cleaved by BamHI and SgrAI. The
pCSmChnERT plasmid codes for a 566 aa long fusion protein (1–
237: mCherry[aa1–aa237]; 238–252: SIP-SV40nls-APGLEP; 253–566:
hESR1[aa282–aa595, with three mutations G400V, M543A, L544A]).
All conditions for restriction digests, gel purification, ligation and
bacteria transformation were according to standard procedures.[49]
Complete plasmid sequences are available upon request.
Cell experiments: CV1 cells were plated on 35 mm Petri dishes in 1
to 2 mL of incubation medium (10% FBS in DMEM) at a density of
100–200 cells per mm2 and incubated at 378C and 5% CO2 for
24 h before transfection. CV1 cells were transfected with plasmid
(1 mg) by using Lipofectamin (Invitrogen). To assay the effect of the
Fluorescence correlation spectroscopy: FCS was used: 1) to charac-
terize the focal points of the various objectives by analyzing the
autocorrelation curves of 50 or 100 nm fluorescein solutions in
0.1m NaOH (Vexc, the illuminated volume, from the value of the
ChemBioChem 2010, 11, 653 – 663
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
661

D. Bensimon, S. Vriz, L. Jullien et al.
providing the original plasmid containing the ERT2 coding se-
quence. Gil Levkowitz is acknowledged for a generous gift of
caged fluorescein dextran. The authors declare no competing
financial interests.
autocorrelation function at zero time by using fluorescein concen-
tration and wxy, the beam waist, from the fluorescein diffusion time
by using its known diffusion coefficient);[50] 2) to analyze the kinet-
ics of the nuclear translocation of GFP–nls-ERT2. For FCS experi-
ments, illumination in the preceding microscope (Olympus
BX51WI) was provided either by a 488 nm laser (Ar-ion, Spectra
Physics) or by a mode-locked Ti-Sapphire laser (200 fs, 76 MHz,
750 nm; Mira, Coherent). The fluorescence photons were collected
through filters (U-MWIBA3 set; without excitation filter, emission
filter: BP460–495), dichroic mirrors (DM505 for 488 nm excitation
and 700 short pass; Olympus for 750 nm excitation wavelengths)
and optical fibers (FG200 LCR multimode fiber, Thorlabs) and were
detected with avalanche photodiodes (SPCM-AQR-14, Perkin–
Elmer) coupled to an ALV-6000 correlator (ALV GmbH). The incident
powers at the sample were measured with a NOVA II powermeter
(Laser Measurement Instruments). All the series of experiments re-
ported in the present work were performed in a regime of laser
powers (3 to 5 mW for 488 nm and 5 to 20 mW for 750 nm) in
which fluorescein exhibits a linear (with one-photon excitation) or
quadratic (with two-photon excitation) dependence of the intensi-
ty of fluorescence emission on the illumination power.
Keywords: cage compounds
· cells · gene expression ·
photochemical methods · protein modifications
[1] J. Gerhart, M. Kirschner, Cells, Embryos, and Evolution, Blackwell, Malden,
MA, 1997.
[2] A. D. S. Ryding, M. G. F. Sharp, J. J. Mullins, J. Endocrinol. 2001, 171, 1–
14.
[3] S. Karkare, D. Bhatnagar, Appl. Microbiol. Biotechnol. 2006, 71, 575–586.
[4] a) A. Fujioka, K. Terai, R. E. Itoh, K. Aoki, T. Nakamura, S. Kuroda, E. Nishi-
da, M. Matsuda, J. Biol. Chem. 2006, 281, 8917–8926; b) J. Dengjel, V.
Akimov, J. V. Olsen, J. Bunkenborg, M. Mann, B. Blagoev, J. S. Andersen,
Nat. Biotechnol. 2007, 25, 566–568.
[5] M. Volgraf, P. Gorostiza, R. Numano, R. H. Kramer, E. Y. Isacoff, D. Trauner,
Nat. Chem. Biol. 2006, 2, 47–52.
[6] P. Neveu, I. Aujard, C. Benbrahim, T. Le Saux, J.-F. Allemand, S. Vriz, D.
Bensimon, L. Jullien, Angew. Chem. 2008, 120, 3804–3806; Angew.
Chem. Int. Ed. 2008, 47, 3744–3746.
Illumination experiments
UV lamps: One-photon illumination experiments were performed
at 208C, with bench top UV lamps (365 nm, essentially a strong
line at 365 nm accompanied by a Gaussian spectral dispersion
around 350 nm with a 40 nm width at half height; Fisher Bioblock)
delivering typical 2.2ꢂ10ꢀ5 (4 W; in vitro experiments) and 4ꢂ10ꢀ5
(6 W; in vivo experiments) Einsteinminꢀ1 photon fluxes in the illu-
minated sample.[6] We found that when illuminated for up to 4 min
the embryos developed normally. We also verified that the byprod-
uct resulting from uncaging cInd did not induce any noticeable
morphological alterations in embryo development.
[7] a) Dynamic Studies in Biology (Eds.: M. Goeldner, R. Givens), Wiley-VCH,
Weinheim, 2005; b) G. Mayer, A. Heckel, Angew. Chem. 2006, 118, 5020–
5042; Angew. Chem. Int. Ed. 2006, 45, 4900–4921; c) G. C. R. Ellis-Davies,
Nat. Methods 2007, 4, 619–628; d) H.-M. Lee, D. R. Larson, D. S. Law-
rence, ACS Chem. Biol. 2009, 4, 409–427; e) A. Specht, F. Bolze, Z.
Omran, J.-F. Nicoud, M. Goeldner, HFSP J. 2009, 3, 255–264.
[8] a) F. Zhang, L.-P. Wang, M. Brauner, J. F. Liewald, K. Kay, N. Watzke, P. G.
Wood, E. Bamberg, G. Nagel, A. Gottschalk, K. Deisseroth, Nature 2007,
446, 633–641; b) Y. I. Wu, D. Frey, O. I. Lungu, A. Jaehrig, I. Schlichting,
B. Kuhlman, K. M. Hahn, Nature 2009, 461, 104–108.
[9] F. G. Cruz, J. T. Koh, K. H. Link, J. Am. Chem. Soc. 2000, 122, 8777–8778.
[10] W. Y. Lin, C. Albanese, R. G. Pestell, D. S. Lawrence, Chem. Biol. 2002, 9,
1347–1353.
UV laser and two-photon excitation: A 40ꢂ0.8 NA water immersion
objective (Olympus) was used to image the embryos on a CCD
camera (Andor Luca) and locate the focal spot of the UV laser/two-
photon excitation. For the UV illumination (375 nm, CW, from Crys-
tal Laser) a beam of 1 mm diameter was coupled to the micro-
scope without expansion. For two-photon illumination (200 fs,
76 MHz, 750 nm, provided by a mode-locked Ti-Sapphire laser,
Mira, Coherent) the beam was expanded to ~6 mm diameter to fill
the back aperture of the objective. The incident power at the
sample (~5 mW with the UV and ꢂ20 mW with the IR laser) was
measured with a NOVA II powermeter (Laser Measurement Instru-
ments).
[11] Y. Shi, J. T. Koh, ChemBioChem 2004, 5, 788–796.
[12] K. H. Link, Y. Shi, J. T. Koh, J. Am. Chem. Soc. 2005, 127, 13088–13089.
[13] S. B. Cambridge, D. Geissler, F. Calegari, K. Anastassiadis, M. T. Hasan,
A. F. Stewart, W. B. Huttner, V. Hagen, T. Bonhoeffer, Nat. Methods 2009,
6, 527–533.
[14] D. Picard, Curr. Opin. Biotechnol. 1994, 5, 511–515.
[15] a) J. Cheung, D. F. Smith, Mol. Endocrinol. 2000, 14, 939–946; b) D.
Picard, Trends Endocrinol. Metabol. 2006, 17, 229–235; c) W. B. Pratt, Y.
Morishima, Y. Osawa, J. Biol. Chem. 2008, 283, 22885–22889.
[16] a) R. Feil, J. Brocard, B. Mascrez, M. LeMeur, D. Metzger, P. Chambon,
Proc. Natl. Acad. Sci. USA 1996, 93, 10887–10890; b) J. Brocard, X.
Warot, O. Wendling, N. Messaddeq, J.-L. Vonesch, P. Chambon, D.
Metzger, Proc. Natl. Acad. Sci. USA 1997, 94, 14559–14563.
[17] R. T. Peterson, B. A. Link, J. E. Dowling, S. L. Schreiber, Proc. Natl. Acad.
Sci. USA 2000, 97, 12965–12969.
Acknowledgements
[18] P. J. Keller, A. D. Schmidt, J. Wittbrodt, E. H. K. Stelzer, Science 2008, 322,
1065–1069.
[19] J. T. Shin, M. M. C. Fishman, Annu. Rev. Genomics Hum. Genet. 2002, 3,
311–340.
[20] U. Langheinrich, BioEssays 2003, 25, 904–912.
This work has been supported by the ANR (PCV 2008, Proteo-
phane), the NABI CNRS–Weizmann Institute program (fellowship
to D.K.S.), and the Ministꢁre de la Recherche (fellowships to N.G.
and P.N.). P.N. is supported in part by the National Science Foun-
dation under grant no. PHY05-51164. D.B. acknowledges partial
support of a PUF ENS-UCLA grant. M.B. thanks ANR for support
(Project ANR-06-PCVI-0025). The authors thank Philippe Leclerc
and Benjamin Matthieu for their kind help with confocal imag-
ing. They are also thankful to Dr. Vincent Jullien and Prof. Gꢀrard
Pons at Service de Pharmacologie Clinique, Saint Vincent de Paul
Hospital Paris, for access to their HPLC-MS instrument. The au-
thors thank Dr. Daniel Metzger (IGBMC, Strasbourg, France) for
[21] G. E. Ackermann, B. H. Paw, Front. Biosci. 2003, 8, 1227–1253.
[22] J. F. Amatruda, J. L. Shepard, H. M. Stern, L. I. Zon, Cancer Cell 2002, 1,
229–231.
[23] C. B. Lunan, A. Klopper, Clin. Endocrinol. 1975, 4, 551–572.
[24] F. F. Vajdos, L. R. Hoth, K. F. Geoghegan, S. P. Simons, P. K. LeMotte, D. E.
Danley, M. J. Ammirati, J. Pandit, Protein Sci. 2007, 16, 897–905.
[25] M. M. Cid, J. A. Seijas, M. C. Villaverde, L. Castedo, Tetrahedron 1988, 44,
6197–6200.
[26] S. Gauthier, J. Mailhot, F. Labrie, J. Org. Chem. 1996, 61, 3890–3893.
[27] T. Furuta, S. S.-H. Wang, J. L. Dantzker, T. M. Dore, W. J. Bybee, E. M. Call-
away, W. Denk, R. Y. Tsien, Proc. Natl. Acad. Sci. USA 1999, 96, 1193–
1200.
662
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 653 – 663

Photocontrol of Protein Activity in Cells and Zebrafish
[28] V. R. Shembekar, Y. Chen, B. K. Carpenter, G. P. Hess, Biochemistry 2005,
44, 7107–7114.
[37] E. Krieger, T. Darden, S. B. Nabuurs, A. Finkelstein, G. Vriend, Proteins
Struct. Funct. Bioinf. 2004, 57, 678–683.
[38] J. Wang, P. Cieplak, P. A. Kollman, J. Comput. Chem. 2000, 21, 1049–
1074.
[39] W. Humphrey, A. Dalke, K. Schulten, J. Mol. Graph. 1996, 14, 33–38.
[40] GROMACS (http://www.gromacs.org).
[41] R Development Core Team, R: A Language and Environment for Statisti-
cal Computing, Vienna, Austria, 2007.
[42] Grace (http://plasma-gate.weizmann.ac.il/Grace/).
[43] H. J. Montgomery, B. Perdicakis, D. Fishlock, G. A. Lajoie, E. Jervis, J. G.
Guillemette, Bioorg. Med. Chem. 2002, 10, 1919–1927.
[44] T. Eckardt, V. Hagen, B. Schade, R. Schmidt, C. Schweitzer, J. Bendig, J.
Org. Chem. 2002, 67, 703–710.
[45] S. D. Lepore, Y. He, J. Org. Chem. 2003, 68, 8261–8263.
[46] A. Z. Suzuki, T. Watanabe, M. Kawamoto, K. Nishiyama, H. Yamashita, M.
Ishii, M. Iwamura, T. Furuta, Org. Lett. 2003, 5, 4867–4870.
[47] R. Feil, J. Wagner, D. Metzger, P. Chambon, Biochem. Biophys. Res.
Commun. 1997, 237, 752–757.
[29] a) E. B. Brown, J. B. Shear, S. R. Adams, R. Y. Tsien, W. W. Webb, Biophys.
J. 1999, 76, 489–499; b) I. Aujard, C. Benbrahim, M. Gouget, O. Ruel,
J. B. Baudin, P. Neveu, L. Jullien, Chem. Eur. J. 2006, 12, 6865–6879.
[30] We mentioned the possible photoinduced damage arising in biological
samples when two-photon illumination is used for photoactivation (see
page seven of the Supporting Information of ref. [6]). In the present
work, we paid attention to remain in the nondetrimental regime of
laser powers.
[31] In all investigated cases, a global fit of the FCS curves to a one-species
diffusion model was poor.
[32] Y. Chen, J. D. Mꢃller, Q. Q. Ruan, E. Gratton, Biophys. J. 2002, 82, 133–
144.
[33] From the diffusion times t1 =1.7 ms and t2 =240 ms, we deduced the
corresponding diffusion coefficients for the unbound and bound spe-
cies: D1 =18.5 mm2 sꢀ1 and D2 =0.1 mm2 sꢀ1. In particular, D2 is signifi-
cantly smaller than D1 as anticipated from the formation of a large
complex between GFP–nls-ERT2 and the chaperone complex.
[34] a) N. Shulga, P. Roberts, Z. Gu, L. Spitz, M. M. Tabb, M. Nomura, D. S.
Goldfarb, J. Cell. Biol. 1996, 135, 329–339; b) R. B. Kopito, M. Elbaum,
Proc. Natl. Acad. Sci. USA 2007, 104, 12743–12748.
[35] I. A. Shestopalov, J. K. Chen, Chem. Soc. Rev. 2008, 37, 1294–1307.
[36] P. Neveu, D. Sinha, P. Kettunen, S. Vriz, L. Jullien, D. Bensimon in Single
Molecule Spectroscopy in Chemistry, Physics and Biology (Eds.: A. Grꢅs-
lund, R. Rigler, J. Widengren), Springer, Heidelberg, 2009, pp. 305–316.
[48] N. C. Shaner, R. E. Campbell, P. A. Steinbach, B. N. G. Giepmans, A. E.
Palmer, R. Y. Tsien, Nat. Biotechnol. 2004, 22, 1567–1572.
[49] J. M. Urbach, T. Wei, N. Liberati, D. Grenfell-Lee, J. Villanueva, G. Wu,
F. M. Ausubel, Curr. Protoc. Mol. Biol. 2009, Unit 19.7.
[50] O. Krichevsky, G. Bonnet, Rep. Prog. Phys. 2002, 65, 251–297.
Received: January 5, 2010
Published online on February 24, 2010
ChemBioChem 2010, 11, 653 – 663
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
663