A caged doxycycline analogue for photoactivated gene expression

Article abstract of DOI:10.1002/anie.200503339

High potential: High-resolution transgene expression in two biological applications demonstrates the potential of a photoactivated gene-expression technique. Irradiation of one half of a transgenic tobacco leaf produced a sharp boundary of reporter gene e

Full text of DOI:10.1002/anie.200503339

Angewandte
Chemie
Gene Function
rely on specific expression patterns enabled by endogenous
promoters. To improve the resolution of transgene expres-
sion, we synthesized a photosensitive (“caged”) doxycycline
analogue for precise light-controlled activation of genes based
on the “Tet-on” (Tet = tetracycline) system.[1] Because of the
ease and precision with which light can be manipulated, this
approach should make it possible to target subsets of cells for
transgene expression which can range from single cells and
tissue patches to whole organs.[2]
DOI: 10.1002/anie.200503339
A Caged Doxycycline Analogue for
Photoactivated Gene Expression**
Sidney B. Cambridge,* Daniel Geissler, Sandro Keller,
and Beate Cürten
To implement a photoactivated gene-expression system
that is generally applicable in any organism at any stage, we
based our approach on a conditional gene-expression para-
digm that uses a small, membrane-permeant molecule for
induction. The most prominent inducible system by far is the
Tet system, which is commonly used for the induction of
Conditional gene-expression paradigms are crucial tools for
the study of genes and gene functions. However, none of the
currently available paradigms permits transgene expression
with high spatial and temporal resolution; rather, they usually
Scheme 1. Synthesis and structure of caged doxycyclines 3a and 3b. The phenolic b-diketone system (ketone enol form) is highlighted in red in
doxycycline, and the caging groupin blue in 3a/3b. DMF=N,N-dimethylformamide.
transgenes in cell culture, tissues, and whole organisms.[1] The
most potent analogue for the Tet system is doxycycline, which
can bind to a modified and mutated version of the Tet
[*] Dr. S. B. Cambridge, D. Geissler
Max-Planck-Institut für Neurobiologie
Am Klopferspitz 18, 82152 München-Planegg (Germany)
repressor fused to a transcriptional activation domain (rtTA)
and induce transgenes under the control of the Tet pro-
moter.[3] Because doxycycline (Scheme 1) has several func-
tional groups that may be derivatized, it was necessary to
specifically target a group that is essential for transcriptional
activity and block this activity by “caging” with a photo-
sensitive protecting group. It was previously shown that the
phenolic b-diketone system (highlighted in red, Scheme 1) is
important for the formation of the doxycycline–magnesium
complex, which binds to the rtTA protein with high affinity.[4]
We therefore attempted to generate a 1-(4,5-dimethoxy-2-
nitrophenyl)ethyl (DMNPE) ether of doxycycline (DMNPE-
caged doxycycline) from the commercially available doxycy-
cline hyclate (hydrochloride hemiethanolate hemihydrate)
with the DMNPE moiety attached at the phenolic b-diketone
system.
Fax: (+49)89-899-50043
E-mail: sidney@neuro.mpg.de
S. Keller
Leibniz Institute of Molecular Pharmacology (FMP)
Robert-Rössle-Strasse 10
13125 Berlin (Germany)
Dr. B. Cürten
CarboGen AG
Neulandweg 5
5502 Hunzenschwil (Switzerland)
[**] We thank T. Bonhoeffer for continued support of this project; P.
Schmieder (FMP) and E. Krause (FMP) for the NMR spectroscopic
and mass-spectrometric analysis, respectively; R. Scholz for help
with the photoactivation of the tobacco leafs (in the laboratory of C.
Gatz); B. Schwappach for providing the Chinese hamster ovary
cells; and T. Bonhoeffer, V. Hagen, and V. Stein for critical reading of
the manuscript. This study was supported by the Max-Planck
society; a DFG fellowship(B.C.); a Volkswagen-Stiftung grant, a
Minorities in Neuroscience FellowshipProgram, and a Schloess-
mann Fellowship(S.B.C.); and the European Commission (grant
no. QLK3-CT-2002-01989; S.K.).
Doxycycline hydrochloride was neutralized with KOH
and then treated with an approximate fivefold excess of
freshly prepared 1-(4,5-dimethoxy-2-nitrophenyl)diazo-
ethane (1) obtained from 4,5-dimethoxy-2-nitroacetophe-
none hydrazone (2) by oxidation with MnO2. The alkylation
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2006, 45, 2229 –2231
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2229

Communications
of doxycycline with 1 resulted in two stable diastereomers of
their hydrolytic stability in aqueous buffers. Incubation for
2 days at 378C did not lead to release of doxycycline, as
determined by HPLC analysis (data not shown). Pilot experi-
ments to assess the suitability of 3a and 3b for photoactivated
gene expression showed that 3a had only little background
activity, whereas 3b produced significant transgene expres-
sion even in the absence of irradiation. We currently do not
understand the reason for this difference especially as both
molecules are stable isomers. However, given that both
isomers displayed surprisingly different physicochemical
properties, differences in their biological properties were
therefore not unexpected. Consequently, 3a was used for all
further experiments.
DMNPE-caged doxycycline (3a and 3b) as main products
and a few unstable caged side products. Because the unstable
by-products complicated the separation of the stable isomers,
they were selectively hydrolyzed by brief treatment of the
reaction mixture with diluted trifluoroacetic acid (TFA) at
408C prior to preparative HPLC separation. The stable
diastereomers 3a and 3b were then isolated by reversed-
phase C-18 HPLC, whereby the yield of each isomer was
about 8%.
The structures of 3a and 3b were clarified by compre-
hensive NMR spectroscopic analysis (see Supporting Infor-
mation). Heteronuclear multiple-bond correlation (HMBC)
spectra of 3a and 3b showed cross-correlation between H1’
and C12 in both cases, and also between the three protons of
the methyl group at C1’ and C12 in the case of 3a.
Consequently, the DMNPE-caging group must have been
conjugated with C12, which is part of the phenolic b-diketone
system. This analysis strongly suggests that 3a and 3b are two
diastereomers which only differ in their configuration of the
asymmetric C1’ center of the DMNPE moiety. In addition, the
etherification of the C12 hydroxy group of doxycycline is
substantiated by a strong shift in the signals of C12 in the
13C NMR spectra of 3a and 3b by Dd = 15 and 18 ppm,
respectively, to higher fields relative to the chemical shift of
Caged doxycycline should ideally be membrane permeant
so that photoactivated gene expression is able to control
transgene expression in single cells. Although in many
applications doxycycline may be photoreleased extracellu-
larly to subsequently diffuse into the cytoplasm for transgene
induction, this scenario will probably not permit single-cell
resolution. The membrane permeability of 3a was therefore
assessed by isothermal titration calorimetry (ITC).[6,7]
A
comparison of ITC uptake and release protocols demon-
strated that 3a can quickly cross the membranes of large
unilamellar vesicles (LUVs) composed of the zwitterionic
phospholipid 1-palmitoyl-3-oleoyl-sn-glycero-2-phosphocho-
line (POPC).
Figure 1a shows a good simultaneous fit (red lines) to the
heats of the reaction Q obtained in uptake (diamonds) and
release (circles) experiments. This fit was based on the
1
C12 in doxycycline (see Supporting Information for the H
and 13C NMR data of doxycycline).
Both diastereomers are characterized by a notable long-
wavelength absorption maximum at around 345 nm (see
Table 1 and Supporting Information), which is caused by the
Table 1: Properties of 3a and 3b.
Caged
Absorbance Extinction
Photolysis
quantum
yield f
Solubility[a]
csat [mm]
doxycycline maximum[a] coefficient[a]
[b]
lmax [nm]
emax [mÀ1 cmÀ1
]
3a
3b
ꢀ347
10500
11400
0.013
0.075
380
770
345.5
[a] In HEPES buffer solution, pH 7.2. [b] In HEPES buffer solution
(pH 7.2)/acetonitrile (80:20, v/v).
Figure 1. ITC analysis of translocation of 3a across POPC membranes.
The heats of the reaction of water–membrane partitioning Q obtained
in uptake (diamonds) and release (circles) experiments are plotted
versus the injection number n. a) The fit (red lines) matches the
experimental data well (based on the assumption of complete mem-
brane permeation). b) A poor fit (blue lines) is obtained particularly for
the critical early injections (assuming no detectable membrane perme-
ation on the experimental timescale).
overlap of the long-wavelength absorption bands of the
phenolic b-diketone system in doxycycline and the DMNPE
chromophore. Irradiation of 3a and 3b between 300 and
400 nm in aqueous buffer solution led to the photorelease of
doxycycline, as confirmed by HPLC analysis (data not
shown), and the 4,5-dimethoxy-2-nitrosoacetophenone by-
product of the caging group, which did not show any harmful
effect in the biological experiments. Furthermore, 3a and 3b
displayed acceptable photochemical quantum yields
(Table 1).
assumptions of ideal mixing and full lipid accessibility of the
caged compound, that is, complete transbilayer equilibration
within the time needed for a single injection. In contrast,
based on the assumption that 3a cannot traverse membranes,
the fit in Figure 1b failed to describe the experimental data, as
there were significant deviations at low injection numbers n
(see Supporting Information for the ITC experimental data).
Repetition of the experiment again with 3a and once with a
mixture of 3a/3b produced almost identical results (data not
shown). The validity of using ITC to test for membrane
permeability was demonstrated in another set of experiments
Compounds 3a and 3b were relatively soluble in aqueous
2-[4-(2-hydroxyethyl)-1-piperazinyl]ethanesulfonic
acid
(HEPES) buffer solution at pH 7.2 (Table 1). This solubility
is essential for applications of caged doxycycline in sensitive
tissues, such as the nervous system, which do not tolerate
alcohols or other organic solvents as vehicles.[5] Thus,
concentrated stock solutions can be readily prepared and
maintained in water. In addition, 3a and 3b were tested for
2230 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 2229 –2231

Angewandte
Chemie
which showed the differential temperature-dependent mem-
brane partitioning of sodium dodecyl sulfate (SDS; see
Supporting Information).[7] Taken together, the data strongly
suggest that the chemically modified doxycycline retained the
ability to passively cross membranes, which in turn should aid
the photoactivation of single cells.
To determine if caged doxycycline could be used for
photoactivated gene expression, two different expression
paradigms were tested. First, as a proof-of-principle experi-
ment, we employed stably transfected Chinese hamster ovary
(CHO) cells which express M2-rtTA[8] and contain a tetracy-
cline-dependent enhanced green fluorescent protein (EGFP)
construct. Incubation of these cells with unmodified doxycy-
cline led to widespread green-fluorescent-protein (GFP)
fluorescence (Figure 2a). Conversely, incubation with 3a did
precision in compact tissue as opposed to dispersed cells in
culture.
In conclusion, we demonstrated that transgene expression
can be accurately manipulated by simple irradiation with UV
light. The doses of UV light needed for induction were not
harmful, as no signs of cell damage were apparent (see
Supporting Information). Previously, small-molecule-based
photoactivated gene expression was demonstrated in cell
culture with caged tamoxifen by global irradiation of the
entire cell-culture medium.[10] Although this approach did not
yield any spatial resolution, a different approach by Lawrence
and co-workers in which caged ecdysone was applied
followed by local photoactivation indeed induced transgene
expression only in irradiated 293T cells.[11] The photoactivated
gene-expression system described herein is based on the
popular Tet system, which provides a rich infrastructure of
Tet-dependent transgenes in various organisms that range
from yeast to plants and mice. Thus, we have established a
paradigm for the photoactivation of genes with single-cell
resolution. We predict that this approach will be a tremen-
dously powerful tool for various research areas, including
single-cell lineage tracing during the development or imple-
mentation of a better cancer model by induction of oncogenes
in single cells surrounded by a wild-type background.
Received: September 20, 2005
Revised: January 2, 2006
Published online: February 28, 2006
Keywords: caged compounds · gene expression · membranes ·
.
photolysis
Figure 2. Spatially restricted photoactivated gene expression. a) CHO
cells (M2-rtTA: tetEGFP) incubated with unmodified doxycycline.
b) CHO cells incubated with 3a, no irradiation. c) Same dish as in
Figure 2b; irradiation-induced, widespread EGFP fluorescence. d) A
sharpboundary between irradiated and non-irradiated areas was
revealed by photoactivation of a GUS reporter gene in one half of
transgenic tobacco tissue.
[1] M. Gossen, A. L. Bonin, S. Freundlieb, H. Bujard, Curr. Opin.
Biotechnol. 1994, 5, 516 – 520.
[2] S. B. Cambridge, R. L. Davis, J. S. Minden, Science 1997, 277,
825 – 828.
[3] M. Gossen, S. Freundlieb, G. Bender, G. Muller, W. Hillen, H.
Bujard, Science 1995, 268, 1766 – 1769.
[4] W. Hinrichs, C. Kisker, M. Duvel, A. Muller, K. Tovar, W. Hillen,
W. Saenger, Science 1994, 264, 418 – 420.
[5] G. Lindner, G. Grosse, Z. Mikrosk.-Anat. Forsch. 1976, 90, 557 –
564.
[6] H. H. Heerklotz, H. Binder, R. M. Epand, Biophys. J. 1999, 76,
2606 – 2613.
[7] S. Keller, H. H. Heerklotz, A. Blume, J. Am. Chem. Soc. 2005,
128, 1279 – 1286.
[8] S. Urlinger, U. Baron, M. Thellmann, M. T. Hasan, H. Bujard, W.
Hillen, Proc. Natl. Acad. Sci. USA 2000, 97, 7963 – 7968.
[9] C. Gatz, C. Frohberg, R. Wendenburg, Plant J. 1992, 2, 397 – 404.
[10] Y. H. Shi, J. T. Koh, ChemBioChem 2004, 5, 788 – 796.
[11] W. Lin, C. Albanese, R. G. Pestell, D. S. Lawrence, Chem. Biol.
2002, 9, 1347 – 1353.
not produce any fluorescence, thus indicating that its tran-
scriptional activity was inhibited (Figure 2b). However,
irradiation of cells in the same dish with long-wavelength
UV light induced significant EGFP expression similar to
EGFP levels seen with unmodified doxycycline. These data
demonstrated that caged doxycycline can be used for photo-
activated gene expression as a tool for localized transgene
expression.
To test photoactivation in a three-dimensional tissue, a
second paradigm was employed using transgenic tobacco leafs
which harbored a quasi Tet-on system based on de-repression
of a constitutive cauliflower mosaic virus (CaMV) 35S
promoter which drives a b-glucuronidase (GUS) reporter
gene.[9] Thus, in the absence of doxycycline, the Tet repressor
prevents transcription from a modified CaMV promoter that
contains three repressor binding sites. Incubation of leaf
tissue with 3a followed by irradiation produced a sharp
boundary of GUS expression between the irradiated and non-
irradiated areas (Figure 2d). To our knowledge, this method is
the first that uses photoactivation based on an inducible gene
expression paradigm to direct transgene expression with high
Angew. Chem. Int. Ed. 2006, 45, 2229 –2231
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2231