Light-activated gene expression directs segregation of co-cultured cells in vitro

Article abstract of DOI:10.1021/cb9002305

Light-directed gene patterning methods have been described as a means to regulate gene expression in a spatially and temporally controlled manner. Several methods have been reported that use photocaged forms of small molecule effectors to control ligand-dependent transcription factors. Whereas these methods offer many advantages including high specificity and transient lightsensitivity, the free diffusion of the uncaged effector can limit both the magnitude and resolution of localized gene induction. Methods to date have been limited by the small fraction of irradiated cells that have expression levels significantly above uninduced background and have not been shown to affect a defined biological response. The tetracycline-dependent transactivator/ transrepressor system, RetroTET-ART, combined with a photocaged form of doxycycline (NvOC-Dox) can be used to form photolithographic patterns of induced expression wherein up to 85% of the patterned cells show expression levels above uninduced regions. The efficiency and inducibility of the RetroTET-ART system allows one to quantitatively measure the limits of resolution and the relative induction levels mediated by a small molecule photocaged effector for the first time. Well-defined patterns of reporter genes were reproducibly formed within 6-36 h with feature sizes as small as 300 m. After photo-patterning, NvOC-Dox can be rapidly removed, rendering cells photoinsensitive and allowing one to monitor GFP product formation in real time. Patterned co-expression of the cell surface ligand ephrin A5 on cell monolayers creates well-defined patterns that are sufficient to direct and segregate cocultured cells via either attractive or repulsive signaling cues. The ability to direct the arrangement of cells on living cell monolayers through the action of light may serve as a model system for engineering artificial tissues.

Full text of DOI:10.1021/cb9002305

ARTICLE
Light-Activated Gene Expression Directs
Segregation of Co-cultured Cells in Vitro
Daniel J. Sauers, Murali K. Temburni†, John B. Biggins, Luke M. Ceo, Deni S. Galileo, and John T. Koh*
Departments of Chemistry and Biochemistry and Biological Sciences, University of Delaware, Newark, Delaware 19716,
†Current address: Department of Biology, Georgian Court University, Lakewood, NJ 08701.
recise spatial and temporal control of gene ex-
pression is essential to the organization of com-
plex tissues, including pattern formation during
ABSTRACT Light-directed gene patterning methods have been described as a
means to regulate gene expression in a spatially and temporally controlled man-
ner. Several methods have been reported that use photocaged forms of small mol-
ecule effectors to control ligand-dependent transcription factors. Whereas these
methods offer many advantages including high specificity and transient light-
sensitivity, the free diffusion of the uncaged effector can limit both the magni-
tude and resolution of localized gene induction. Methods to date have been lim-
ited by the small fraction of irradiated cells that have expression levels significantly
above uninduced background and have not been shown to affect a defined biologi-
cal response. The tetracycline-dependent transactivator/transrepressor system,
RetroTET-ART, combined with a photocaged form of doxycycline (NvOC-Dox) can
be used to form photolithographic patterns of induced expression wherein up to
85% of the patterned cells show expression levels above uninduced regions. The
efficiency and inducibility of the RetroTET-ART system allows one to quantitatively
measure the limits of resolution and the relative induction levels mediated by a
small molecule photocaged effector for the first time. Well-defined patterns of re-
porter genes were reproducibly formed within 6Ϫ36 h with feature sizes as small
as 300 ␮m. After photo-patterning, NvOC-Dox can be rapidly removed, rendering
cells photoinsensitive and allowing one to monitor GFP product formation in real
time. Patterned co-expression of the cell surface ligand ephrin A5 on cell monolay-
ers creates well-defined patterns that are sufficient to direct and segregate co-
cultured cells via either attractive or repulsive signaling cues. The ability to direct
the arrangement of cells on living cell monolayers through the action of light may
serve as a model system for engineering artificial tissues.
P
tissue development (1, 2). The study of guidance cues
and other forms of intercellular communication that rely
on the coordinated expression of genes in multicellular
patterns can be hindered by the limited methods avail-
able to precisely control patterned gene expression in
vivo or in vitro. Recreating such phenomenon in in vitro
cell culture can provide important validation in con-
trolled systems for the roles of various attractive and re-
pulsive cues in normal development and cancer. The
precise spatial and temporal control of expression of cell
signaling molecules in vitro can serve as the basis for
creating more advanced artificial tissues in vitro. Sev-
eral methods have been developed to control spatio-
temporal gene expression by placing genes of interest
under the control of light using photocaged molecules
that regulate gene expression only when converted into
a biologically active form though the action of light
(3−13). Many of these systems utilize photocaged
biopolymers that are generally difficult to deliver into
cells and are typically introduced by microinjection into
oocytes, resulting in an organism that subsequently re-
mains light-sensitive (3, 5, 14−16). Photocaged ligands
of nuclear hormone receptors or ligand-dependent re-
pressors/activators readily enter cells and render sys-
tems photosensitive only upon treatment with caged ef-
fectors (4, 6, 17−22). However, once uncaged, small
molecule ligands can readily diffuse away from the site
of uncaging, potentially limiting the efficiency and mag-
nitude of response, which limits the spatial resolution of
the induced pattern of protein expression (19). Recent
reports of light-directed gene patterning by small mol-
ecule effectors show that only a small fraction of the ir-
radiated cells have expression levels significantly above
*Corresponding author,
johnkoh@udel.edu.
Received for review September 21, 2009
and accepted January 5, 2010.
Published online January 5, 2010
10.1021/cb9002305
© 2010 American Chemical Society
www.acschemicalbiology.org
VOL.5 NO.3
•
ACS CHEMICAL BIOLOGY
313

Figure 1. Light-directed gene patterning with photomasks using Retro TET-ART. a) Near-confluent monolayers of 3T3 cells
with gene of interest (GFP) under control of the Retro-Tet ART system. The transrepressor remains bound in the pres-
ence of NvOC-Dox. b) A photomask is used to selectively irradiate cells in a desired pattern. Liberated dox causes the re-
lease of the transrepressor and recruitment of the transactivator to the Tet operator. c) An actual pattern of induced gene
expression corresponding to the shape of the photomask (scale bar ؍ 1 mm).
uninduced background levels, which can significantly
Stable 3T3 cell lines that express tet-inducible GFP
limit potential applications (8, 23). Herein we report the (HRSp-GFP) under the control of the RetroTET-ART sys-
first efficient gene patterning system that enables one
to generate well-defined photolithographic patterns, to
evaluate quantitatively the spatial resolution and the
time course of photoactivated gene expression, and
demonstrate the physiological relevance of such pat-
terns by directing co-cultured cells through attractive
and repulsive cues.
tem were created by multiple rounds of infection, selec-
tion, and FACS sorting (25). Application of photocaged
doxycyline (nitroveratryloxydoxycycline, Nv-Dox) has
been described recently as a regulator of “Tet-on” de-
pendent transcription (18). Using the tet transactivator
RTAb alone, less than 30% of confluent 3T3 cells are ac-
tivated above uninduced levels upon treatment for 2 h
with Dox. By contrast, the full RetroTET-ART system,
which employs both a tetracycline (tet) dependent tran-
srepressor (tTRg) and a retro-tet transactivator (rtRAb),
was found to activate greater than 90% of cells above
uninduced levels under the same conditions (see Sup-
porting Information). We have prepared a photocaged
form of doxycycline (nitroveratryloxycarbonyl doxycycline,
or NvOC-Dox). Preparations of NvOC-Dox show no in-
crease in background GFP expression up to 29 ␮M. Brief
UVA exposure of cells containing 6 ␮M NvOC-Dox re-
RESULTS AND DISCUSSION
The first reported examples of light-activated gene ex-
pression by small molecule effectors used caged ago-
nists of nuclear/steroid hormone receptors (4, 6). After
surveying the estrogen receptor, thyroid hormone recep-
tor, and ecdysone receptor based systems (24), some
of which lacked efficiency and quick inducibility in in
vitro confluent cell monolayers, we found that in our
hands the tetracycline based RetroTET-ART (25) system
provided a robust response to transient ligand exposure sulted in levels of GFP expression comparable to direct
intrinsic to localized activation by effectors that can dif-
fuse from the site of uncaging. Contact-inhibited cells
(NIH 3T3) that stop dividing once confluent to form
well-defined monolayers were selected for patterning
experiments as they form stable patterned monolayers
appropriate for studying intercellular interactions of co-
cultured cells in vitro.
treatment with 9 ␮M dox for 2 h. No change in fluores-
cence was observed after equivalent irradiation in the ab-
sence of the caged ligand (see Supporting Information).
Time-Dependence of Light-Activated GFP Patterning.
Spatially confined patterns of UVA (330Ϫ380 nm) irra-
diation were generated using photomasks of UV-
absorbing polyester filters placed directly on the bot-
314
VOL.5 NO.3
•
313–320
•
2010
SAUERS ET AL.
www.acschemicalbiology.org

ARTICLE
Figure 2. a) Time lapse images of light-directed gene patterning of GFP. A near confluent monolayer of HRSp-GFP 3T3
cells was treated with 6 ␮M NvOC-Dox and irradiated for a total of 120 s with UVA light through a 344 ␮m photomask.
Time-lapse images acquired over 36 h showed the emergence of a pattern of GFP expression in the region of irradiation
(right), while the phase contrast images (left) show no change in cell morphology due to the UVA exposure (scale bar ؍
500 ␮m). b) The induced and uninduced regions of the pattern used to calculate the average relative increase in GFP fluo-
rescence (red ؍ uninduced, blue ؍ induced) (scale bar ؍ 250 ␮m). c) Time-dependent induction of relative fluorescence
in induced versus uninduced regions relative to untransfected 3T3 cells approached 5-fold in 26 h.
tom of tissue culture dishes using a Nikon TE-2000E
equipped with a 100 W Hg-Arc lamp through a Nikon
cal changes could be seen as the result of the UV expo-
sure in the phase contrast images (Figure 2, panel a,
4X objective using a Nikon UV-1 filter (see Supporting In- left). Using a photomask that was 244 ␮m in width, a
formation). HRSP-GFP 3T3 cells were grown to near con- patterned stripe of GFP expression roughly 570 ␮m be-
fluence and treated with media containing 6 ␮M of gan to appear as early as 4–6 hours and approached a
NvOC-Dox 30 min prior to irradiation through the photo- maximum after 26 hours (Figure 2, panel a, right and
mask (Figure 1). After irradiation, the media was ex-
changed with ligand-free media. No longer light-
responsive, the cells could be monitored using both
brightfield and fluorescence imaging without altering
the pattern of GFP expression (see Supporting
Information). To obtain increased sensitivity and higher
resolution, composite images (40Ϫ60 megapixel)
were compiled by tiling a series of high magnification
images.
Time-lapse images taken over 36 h showed the emer-
gence of a clearly differentiated pattern of GFP expres-
sion at 4Ϫ6 h and approached a maximum intensity af-
ter 26 h (Figure 2, panel a). Photorelease of doxycycline
is many orders of magnitude faster than the process of
transcription/translation and the relatively slow forma-
tion of the GFP chromophore (26). Thus the time-
dependent induction of observed GFP fluorescence is
similar to that previously reported after adding doxycy-
Supporting Information). Patterns made in this manner
were still visible 4 days after irradiation. Under the de-
scribed experimental conditions, 70Ϫ85% of the cells
within the photoexposed region were clearly induced
above background, allowing us to clearly define the
boundaries of the patterned response and measure the
relative induction of patterned response. Relative induc-
tion of GFP expression was determined by comparing
the average fluorescence intensity of induced and unin-
duced regions of the pattern that are separated by only
65 ␮m (Figure 2, panel c). The time-dependent expres-
sion profile shows that the relative level of GFP expres-
sion was 4.9-fold higher in the patterned area than in
the unirradiated areas, which remained essentially un-
changed. Despite the transient exposure to uncaged
doxycycline, this represents approximately 50% of the
induction observed when similar cells were treated with
9 ␮M of dox for 2 h. Using contact-inhibited cell mono-
cline to HRSP-GFP-expressing cells (25). No morphologi- layers and photomasks, we have been able to achieve
www.acschemicalbiology.org
VOL.5 NO.3
•
313–320
•
2010
315

images matched the regions observed by direct
GFP fluorescence, thus supporting that the GFP fluo-
rescence images were the result of an increase in
GFP expression and not from alteration of the inher-
ent fluorescent properties of the irradiated cells (see
Supporting Information).
The high percentage of responsive cells observed
by this method allows one to form patterns with
well-defined boundaries and to test the limits of
resolution attainable by a soluble effector. Whereas
regions of induced expression were obtained in ev-
ery experiment, sharp patterns closely matching that
of the photomask were obtained only 20Ϫ30% of
the time. Distorted or more poorly defined patterns
having regions of low induction were observed in the
remaining cases (see Supporting Information). Pat-
tern distortion can be caused by inadvertently dis-
turbing the culture just after irradiation. Regions of
poor induction or irregularly shaped patterns are cir-
cumstantially linked to uneven cell plating (i.e., lo-
cal degree of confluency) of the cell monolayer; con-
fluent cells appear to be less responsive to transient
exposure to uncaged ligand. Within these limita-
tions, the same photomask could be used to repro-
duce nearly identical patterns multiple times (Figure 3,
Figure 3. Resolution, reproducibility, and intensity of photoinduced patterns of GFP ex-
pression. a) The limits of resolution of photoinduced patterns of gene expression us-
ing a photomask of decreasing line widths (shown in blue) and the resulting patterns
of expression shown as a line scan (average vertical) intensity profile. The smallest re-
producible pattern was a 304 ␮m wide stripe from a 195 ␮m wide template. b) Nearly
identical images can be reliably reproduced using the same photomask (scale bar ؍
500 ␮m). With a photomask template of 970 ␮m ؋ 3404 ␮m, the resulting patterns
were, left ؍ 1275 ␮m ؋ 3775 ␮m, middle ؍ 1340 ␮m ؋ 3720 ␮m, and right ؍
1230 ␮m ؋ 3700 ␮m. c) A 3D surface plot of intensity for a patterned triangle of GFP
expression about 4 mm per side that shows the relative induction with 8-bit image in-
tensity values (scale bar ؍ 500 ␮m).
highly induced and resolved patterns of gene expres-
sion that were sustained for many days.
Light-activated gene patterning using small molecule panel b).
effectors can provide resolved patterns of inducible
gene expression in vitro. The caged effector NvOC-Dox
The feature sizes that can be created by methods
that employ small molecule effectors are limited by li-
diffuses readily in and out of cells to render cellular sys- gand diffusion, which requires a balance between gen-
tems transiently light-sensitive. After local uncaging,
the small molecule ligand diffuses away from the pat-
terned region, limiting the duration and potentially the
intensity of gene response. For tightly binding ligandϪ
receptor complexes, the duration of transcription re-
erating a sufficiently high local concentration of effector
to trigger a transcriptional response while limiting the
amount of effector that diffuses into adjacent regions of
cells. The lower limit of feature sizes that can be pat-
terned by this system may be approximated using a
sponse is much longer due to the slow off-rate of the li- photomask composed of multiple stripes of decreasing
gand from the ligandϪreceptor complex and the half-life widths. A 195 ␮m wide photomasked stripe formed a
of the reporter protein GFP. In our studies, light-induced 304 ␮m strip of GFP-expressing cells, and although a
GFP expression continued to rise up to 26 h after photo- photomask strip of 101 ␮m did provide some cells with
patterning, long after effective concentrations of free li-
gand are lost to diffusion.
Pattern Reproducibility and Resolution. The NvOC-
Dox/RetroTet-ART system was capable of creating pat-
clearly induced expression, induction was inconsistent
and failed to form a well-defined pattern (Figure 3,
panel a). The pattern of induced gene expression was al-
ways larger than the size of the photomask, due in part
terns of many shapes and sizes. In all cases, the photoir- to optical diffraction and the diffusion of uncaged doxy-
radiated regions, though clearly visible by fluorescence cycline into adjacent cells. The larger size of the GFP-
(GFP) imaging, were indiscernible by phase contrast im- expressing cells does not appear to be due to cell migra-
aging. As an additional assay of induced gene expres-
sion, patterned cell cultures were immunostained with
tion as time lapse image analysis shows that cell
migrate on average only 13 Ϯ 9 ␮m over the experi-
anti-GFP antibodies. The resulting immunofluorescence ment in these contact-inhibited cell monolayers. The
316
VOL.5 NO.3
•
313–320
•
2010
SAUERS ET AL.
www.acschemicalbiology.org

ARTICLE
photomask method can create patterns to the reso-
lution limits of the NvDox/RetroTet-ART system, is
operationally simple, and can be adapted to any
common laboratory microscope.
Ligand diffusion does limit pattern resolution.
Whereas sufficient uncaged ligand is formed to cre-
ate patterns with 50% of the maximum inducible
expression, narrow features do not give rise to sig-
nificant expression, presumably because the un-
caged ligand is more rapidly diluted through diffu-
sion. Although it is possible to obtain patterns from
smaller features using a higher initial concentra-
tion of ligand, this might also cause gene activa-
tion in neighboring regions to increase. Ultimately,
the resolution of the patterns that can be achieved
may depend on the efficiency of ligand caging,
which limits the concentration of caged ligand that
can be used as well as the diffusion rate of the li-
gand out of the cell and its on/off rates with the
receptor.
Multigene Patterning and Segregation of Co-
cultured Cells. Whereas differences in fluores-
cence intensity between on and off pattern re-
gions of cell monolayer are easily discernible by
imaging, it remains to be demonstrated if these
differences in light-directed gene expression are
sufficient to cause a physiologically relevant ef-
fect. An important application of light-activated gene
expression can be to direct multicellular
organization to form artificial tissues. The membrane-
bound ephrin ligands and their Eph tyrosine kinase re-
ceptors help mediate cellϪcell interactions of many
different developmental processes (27−29). While the
interaction between the EphA7 receptor and the li-
gand ephrin A5 generally mediates a repulsive signal
between cells (30−32), the naturally occurring splice
variant, EphA7-T1, which lacks its tyrosine kinase sig-
naling domain (33), provides attractive interactions
with ephrin A5 that aid in the proper fusion of cranial
neural folds at the dorsal midline during mouse brain
development (34, 35). The opposing effects of EphA7
variants toward ephrin A5-expressing cells can be
used to direct cells toward or away from regions of in-
duced ephrin A5 expression.
Figure 4. Coincidental expression of GFP and ephrin A5 expression using the aphthovirus
2A polyprotein cleavage sequence. a) Pattern of increased ephrin A5 expression, visual-
ized by GFP fluorescence and b) the phase image of the same cells. c) The pattern was
stained with an antiephrin A5 antibody. d) The resulting red pattern is nearly identical to
that of the GFP as seen in an image overlay. e) Enlargement of image d confirms proper
2A cleavage, with surface staining of the ephrin A5 and cytoplamic GFP. All scale bars ؍
500 ␮m.
time, the ephrin A5 cDNA was subcloned downstream of
GFP in HRSp-GFP linked in-frame via the aphthovirus 2A
polyprotein sequence. The 2A sequence is a short amino
acid sequence containing a terminal P-G-P that leads to
intraribosomal cleavage of the nascent polypeptide be-
fore the terminal proline, thus providing an efficient
means of making multicistronic vectors that result in a
1:1 expression of multiple proteins (36−38). In response
to Tet, the HRSp-GFP-2A-ephrinA5 vector provides
equimolar production of GFP that is coincidental with eph-
rin A5 (Figure 4 and Supporting Information).
Cell monolayers expressing spatially discrete photo-
patterned ephrin A5/GFP were evaluated for their abil-
ity to direct adhesion of HEK293T cells expressing full
length or truncated forms of the EphA7 receptor.
EphA7-T1-expressing 293T cells preferentially adhere
to the regions of the 3T3 monolayers expressing in-
duced ephrin A5 (Figure 5, panels aϪc). Under identi-
cal conditions, no adhesion is observed when EphA7-
T1-expressing 293T cells are added to untransfected
3T3 monolayers. The selective adhesion of the 293T
A new 3T3 cell line was created that provides tet induc-
ible expression of ephrin A5 under control of the RetroTet
ART transrepressor/retrotransactivator. In order to di-
rectly visualize the regions of ephrin expression in real
www.acschemicalbiology.org
VOL.5 NO.3
•
313–320
•
2010
317

rescence intensity (Figure 5, panel c and Support-
ing Information).
Cells expressing the full length EphA7 were also
created, labeled with DiI and co-cultured with 3T3
cell monolayers expressing freshly patterned eph-
rin A5 (Figure 5, panels dϪf). In contrast to cells ex-
pressing the truncated EphA7-T1, cells expressing
full-length EphA7 were repelled by the regions of
photoinduced ephrin A5 expression and were found
3.1-fold less frequently in the triangular patterned
region than outside the region (Figure 4, panel e).
Note, however, that in a few spots, EphA7-
Figure 5. Segregation of EphA7-T1-expressing (a؊c) and EphA7-expressing (d؊f) 293T
cells by cell monolayers with patterned ephrin A5/GFP expression. a) Pattern of induced
ephrin A5 expression visualized by co-translated GFP fluorescence. b) Phase contrast
image of EphA7-T1-expressing 293T cells adhering to a 3T3 monolayer with patterned
ephrin A5 expression. c) Fluorescence image of DiI-labeled EphA7-T1-expressing 293T
cells, showing preferential adhesion to the region of induced ephrin A5. d) DiI-labeled
EphA7-expressing 293T cells co-cultured for 48 h on ephrin A7 patterned 3T3 monolay-
ers. e) EphA7 293T cells (red) merged with pattern of ephrin A5 expression visualized by
co-translated GFP. f) Higher magnification of boundary region. Scale bars for a, b, and c
؍ 500 ␮m. Scale bar for d ؍ 250 ␮m.
expressing 293T cells were able to invade regions
of patterned ephrin A5 to form well segregated clus-
ters, a phenomenon presumably created by cells
adhering to the substrate through gaps in the mono-
layer that grew to displace the monolayer (Figure 5,
panels and f). Excluding these well-defined local-
ized clusters, the density of EphA7-expressing cells
was 105-times lower in the patterned region than in
the adjacent nonpatterned region (Supporting Informa-
tion). Together these studies demonstrate that multi-
gene patterned monolayers can be used to segregate
and direct co-cultured cells using both attractive and re-
pulsive guidance cues in vitro, providing a means to
control the arrangement of cells in three dimensions.
Such systems form the basis for creating complex artifi-
cial tissues in vitro and provide new tools for studying
phenomena such as differentiation and development
that are dependent on coordinated expression of genes
cells to the region of patterned ephrin A5/GFP expres-
sion was directly visible in the phase contrast images
(Figure 5, panel a). Fluorescence image analysis of the
fluorescent vital dye (DiI)-labeled EphA7-T1-expressing
293T cells combined with the co-translated expression
of GFP allows one to monitor the pattern and the co-
cultured cells in real time. The attractive interactions be-
tween ephrin A5 patterned on the monolayer and the
full length EphA7 expressed in the labled co-cultured
293T cells provided a well-defined pattern representing
a 97% increase in 293T cells in the regions of ephrin A5
expression measured by the increase in integrated fluo- in multicellular patterns.
in anhydrous methylene chloride (1.5 mL) at 0 °C was added di-
isopropylethyl amine (52.3 ␮L, 0.61 mmol). After 5 min, a solu-
tion of nitroveratryl chloroformate (80.75 mg, 0.29 mmol) in an-
hydrous dichloromethane (1.5 mL) was added via canula. The
reaction mixture was allowed to gradually warm to RT and was
stirred for 15 h. The crude mixture was chromatographed on
silica gel using a gradient (0Ϫ5%) of methanol in dichlorometh-
ane to afford 33.7 mg (0.049 mmol, 17%) of NvOC-Dox. 1H
NMR (DMSO-d6, 600 MHz, ␦): 1.52 (d, J ϭ 6.70 Hz, 3H, 6MeH),
2.43 (s, 6H, C-4 N(CH3)2), 2.56 (m, 1H, 5aH), 2.84 (m, 1H, 6H),
3.85 (s, 3H, 8==MeH), 3.95 (s, 3H, 7==MeH), 4.24 (br s, 1H, 4a),
5.46 (d, J ϭ 14.6 Hz, 1H, 1==aH), 5.57 (d, J ϭ 14.5 Hz, 1H, 1==bH),
6.84 (d, J ϭ 8.2 Hz, 1H, 9H), 6.95 (d, J ϭ 8.0 Hz, 1H, 7H), 7.16
(s, 1H, 6==H), 7.51 (t, J ϭ 8.1 Hz, 1H, 8H), 7.70 (s, 1H, 9==H), 9.12
(br s, 2H, C-2 CONH2), 11.65 (br s, 1H, C-10 OH), 14.97 (br s,
1H, OH). 13C NMR (DMSO-d6, 400 MHz, ␦): 194.5, 192.9, 183.2,
174.19, 173.18, 161.40, 153.51, 148.21, 147.86, 138.90,
138.64, 134.18, 127.09, 117.5, 116.56, 115.74, 111.74,
108.27, 107.8, 91.5, 72.38, 68.11, 65.92, 65.61, 56.28, 55.98,
METHODS
Plasmid Description and Cell Line Creation. HRSP-GFP-2A-
ephrin A5 was created by amplifying the ephrin A5 gene with a
forward primer containing a 2A sequence (gag ggc cgc ggc tcc ctc
ctc acc tgc ggc gac gtc gag gag aac ccc ggc ccc or E G R G S L L
T C G D V E E N P G P) in frame. This fragment was ligated into the
AgeI and ClaI sites of HRSpuro-GFP (36−38). A Western blot con-
firmed that the proper transcriptional cleavage was occurring at
the 2A sequence, and immunostaining supported the surface
expression of ephrin A5 with intracellular expression of GFP see
(Supporting Information). Both HRSp-GFP amd HRSP-GFP-2A-
ephrin A5 plasmids were used to create stable NIH 3T3 cell lines
using the calcium phosphate method. Doxycycline-inducible
cell lines were created for each by viral infection with the trans-
activator (RTAb) and transrepressor (RTRg) of the RetroTet-ART in-
ducible system as previously described (36−38).
Synthesis and Purification of NvOC-Dox. To a solution of doxy-
cycline hyclate (150 mg, 0.29 mmol) and 4 Å molecular sieves
318
VOL.5 NO.3
•
313–320
•
2010
SAUERS ET AL.
www.acschemicalbiology.org

ARTICLE
48.60, 45.40, 42.28, 38.52, 38.12, 37.10, 16.16. HRMS (ESI)
[Mϩ1] calcd for C32H34N3O14 684.20400; found 684.20400.
Photoinduced Patterning Procedure. 3T3 cells were seeded
into individual 60 mm plastic dishes (Falcon) such that they
would form a nearly confluent monolayer in 48 h. Successful
photo-patterning was obtained when the monolayer was closely
and evenly spaced, while maintaining a “spiky” appearance
(see Supporting Information). Under red light conditions, media
containing the appropriate amount of NvOC-Dox (up to 6 ␮M for
HRSp-GFP and up to 18 ␮M for HRSp-GFP-2A-ephrin A5) was
added to the cells and allowed to incubate for 30 min prior to ir-
radiation. Photomasks were made using Roscolux polyester
color gel no. 27 (medium red), due to its 0% transmittance of
light above 400 nm (400Ϫ600 nm) (see Supporting Informa-
tion). The desired photomask was then fixed to the bottom of
the cell plate and irradiated using a 100 W Hg-Arc lamp through
a Nikon 4X objective using a Nikon UV-1 filter (excitation
330Ϫ380 nm), in a Nikon TE-2000E epifluorescene microscope.
The duration of irradiation varied from a low of two 30 s expo-
sures, separated by 4 min, to a high of two 120 s exposures.
Whereas expression patterns created using the shortest irradia-
tion time were sometimes visible, the majority of cells in the pat-
terned area remained uninduced. The expression patterns with
the highest percentage of induced cells within the targeted re-
gion (Ͼ90%) were created using the longest exposure time of 4
min. The position of the photomask relative to the cells was ob-
tained from the x and y coordinates from the programmable mi-
croscope stage (Prior Proscan II). After irradiation,the cells were
washed 3 times with fresh medium and the patterns typically im-
aged Ն20 h later.
Image Preparation and Analysis. All images were acquired us-
ing an automated time-lapse microscopy system composed of
a Nikon TE-2000E epifluorescence microscope equipped with a
Prior Proscan II motorized stage and controller, a 100 W Hg-Arc
lamp, and a Photometrics CoolSNAP ES CCD camera using Meta-
morph software, which is described in detail in Fotos et al. (39)
The filters used were a Nikon UV-1 for UV irradiation, a Nikon EN
GFP HQ for GFP excitation, and a Nikon G-1B for red fluorescent
imaging. In order to maximize sensitivity and resolution, mul-
tiple individual images taken using the 10X objective were tiled
together into a single composite image. The benefits of tiling
multiple images together were improved cellular resolution, a
higher dynamic range of GFP intensity levels on the images, and
the ability to visualize larger induced patterns of expression
(up to 6 mm). The process of tiling together the numerous indi-
vidual images into a single larger image was efficiently accom-
plished using a number of custom-made actions for Adobe Pho-
toshop CS2 (see Supporting Information). The phase images
taken at each position were used to align individual positions.
The fluorescence images were then overlaid on top of the phase
images and merged together to create the final image of the in-
duced pattern of expression. All adjustments in brightness and
contrast were applied equally to all of the component images of
the final composite tiled image, thus preserving the signifi-
cance of the relative differences between uninduced and in-
duced regions of the image.
Supporting Information Available: This material is available
free of charge via the Internet.
REFERENCES
1. Gilbert, S. F. (2006) Developmental Biology, 8th ed., Sinauer Associ-
ates, Inc., Sunderland, MA.
2. Lewis, J. (2008) From signals to patterns: Space, time, and mathe-
matics in developmental biology, Science 322, 399–403.
3. Ando, H., Furuta, T., Tsien, R. Y., and Okamoto, H. (2001) Photo-
mediated gene activation using caged RNA/DNA in zebrafish em-
bryos, Nat. Genet. 28, 317–325.
4. Cruz, F. G., Koh, J. T., and Link, K. H. (2000) Light-activated gene ex-
pression, J. Am. Chem. Soc. 122, 8777–8778.
5. Haselton, F. R., Tseng, W. C., and Chang, M. S. (1997) Light acti-
vated protein expression using caged transfected plasmid i: deliv-
ery by liposomes to cultured retinal endothelium, Invest. Ophthal-
mol. Visual Sci. 38, 2082–2082.
6. Lin, W. Y., Albanese, C., Pestell, R. G., and Lawrence, D. S. (2002)
Spatially discrete, light-driven protein expression, Chem. Biol. 9,
1347–1353.
7. Cambridge, S. B., Davis, R. L., and Minden, J. S. (1997) Drosophila
mitotic domain boundaries as cell fate boundaries, Science 277,
825–828.
8. Cambridge, S. B., Geissler, D., Calegari, F., Anastassiadis, K., Hasan,
M. T., Stewart, A. F., Huttner, W. B., Hagen, V., and Bonhoeffer, T.
(2009) Doxycycline-dependent photoactivated gene expression in
eukaryotic systems, Nat. Methods 6, 527–U86.
9. Lee, H. M., Larson, D. R., and Lawrence, D. S. (2009) Illuminating the
chemistry of life: design, synthesis, and applications of “caged” and
related photoresponsive compounds, ACS Chem. Biol. 4, 409–
427.
10. Mayer, G., and Heckel, A. (2006) Biologically active molecules with
a “light switch”, Angew. Chem., Int. Ed. 45, 4900–4921.
11. Tang, X. J., and Dmochowski, I. J. (2007) Regulating gene expres-
sion with light-activated oligonucleotides, Mol. Biosyst. 3, 100–
110.
12. Young, D. D., and Deiters, A. (2007) Photochemical control of biolog-
ical processes, Org. Biomol. Chem. 5, 999–1005.
13. Young, D. D., Garner, R. A., Yoder, J. A., and Deiters, A. (2009) Light-
activation of gene function in mammalian cells via ribozymes,
Chem. Commun. 568–570.
14. Shestopalov, I. A., Sinha, S., and Chen, J. K. (2007) Light-controlled
gene silencing in zebrafish embryos, Nat. Chem. Biol. 3, 650–651.
15. Young, D. D., Lusic, H., Lively, M. O., Yoder, J. A., and Deiters, A.
(2008) Gene silencing in mammalian cells with light-activated anti-
sense agents, ChemBioChem 9, 2937–2940.
16. Shah, S., Rangarajan, S., and Friedman, S. H. (2005) Light-activated
RNA interference, Angew. Chem., Int. Ed 44, 1328–1332.
17. Biggins, J. B., Hashimoto, A., and Koh, J. T. (2007) Photocaged ago-
nist for an analogue-specific form of the vitamin D receptor, Chem-
BioChem 8, 799–803.
18. Cambridge, S. B., Geissler, D., Keller, S., and Curten, B. (2006) A
caged doxycycline analogue for photoactivated gene expression,
Angew. Chem., Int. Ed. 45, 2229–2231.
19. Link, K. H., Cruz, F. G., Ye, H.-F., O’Reilly, K., Dowdell, S., and Koh, J. T.
(2004) Photo-caged agonists of the nuclear receptors RAR␥ and TR␤
provide unique time-dependent gene expression profiles for light-
activated gene patterning, Bioorg. Med. Chem. 12, 5949–5959.
20. Link, K. H., Shi, Y. H., and Koh, J. T. (2005) Light activated recombi-
nation, J. Am. Chem. Soc. 127, 13088–13089.
Acknowledgment: We thank Dr. Steve Bai for assistance
with 2-D NMR experiments, Professor Helen M. Blau (Stanford
University) for providing RetroTet-ART system plasmids, Profes-
sor Elena Pasquale (Burnham Institute) for the ephrin A5 expres-
sion plasmid and antibodies, and Prof. Jonas Frise´n (Karolinska
Institute) for providing EphA7 and EphA7-T1 expression plas-
mids. J.B.B. is a Robert Black Fellow of the Damon Runyon Can-
cer Research Foundation (DRG 1852-05). This work was sup-
ported by the National Institutes of Health (NINDS
21. Shi, Y. H., and Koh, J. T. (2004) Light-activated transcription and re-
pression by using photocaged serms, ChemBioChem 5, 788–796.
22. Young, D. D., and Deiters, A. (2007) Photochemical activation of pro-
tein expression in bacterial cells, Angew. Chem., Int. Ed. 46, 4290–
4292.
R01NS049523 and NIDDK R01DK54257).
www.acschemicalbiology.org
VOL.5 NO.3
•
313–320
•
2010
319

23. Lin, W., Albanese, C., Pestell, R. G., and Lawrence, D. S. (2002) Spa-
tially discrete, light-driven protein expression, Chem. Biol. 9, 1347–
1353.
24. Biggins, J. B., Sauers, D., Temberni, M., Koh, J. T.Unpublished re-
sults.
25. Rossi, F. M. V., Guicherit, O. M., Spicher, A., Kringstein, A. M., Fa-
tyol, K., Blakely, B. T., and Blau, H. M. (1998) Tetracycline-
regulatable factors with distinct dimerization domains allow revers-
ible growth inhibition by p16, Nat. Genet. 20, 389–393.
26. Reid, B. G., and Flynn, G. C. (1997) Chromophore formation in green
fluorescent protein, Biochemistry 36, 6786–6791.
27. Frisen, J., Holmberg, J., and Barbacid, M. (1999) Ephrins and their
Eph receptors: Multitalented directors of embryonic development,
EMBO J. 18, 5159–5165.
28. Frisen, J., Yates, P. A., McLaughlin, T., Friedman, G. C., O’Leary,
D. D. M., and Barbacid, M. (1998) Ephrin-A5 (AL-1/RAGS) is essen-
tial for proper retinal axon guidance and topographic mapping in
the mammalian visual system, Neuron 20, 235–243.
29. Holder, N., and Klein, R. (1999) Eph receptors and ephrins: Effec-
tors of morphogenesis, Development 126, 2033–2044.
30. Castellani, V., Yue, Y., Gao, P. P., Zhou, R., and Bolz, J. (1998) Dual
action of a ligand for Eph receptor tyrosine kinases on specific pop-
ulations of axons during the development of cortical circuits,
J. Neurosci. 18, 4663–4672.
31. Davy, A., Gale, N. W., Murray, E. W., Klinghoffer, R. A., Soriano, P.,
and Feuerstein, C. (1999) Compartmentalized signaling by GPI-
anchored ephrin-A5 requires the fyn tyrosine kinase to regulate cel-
lular adhesion, Genes Dev. 13, 3125–3135.
32. Pandey, A., Shao, H., Marks, R. M., Polverini, P. J., and Dixit, V. M.
(1995) Role of B61, the ligand for the Eck receptor tyrosine kinase,
in TNF-␣-induced angiogenesis, Science 268, 567–569.
33. Holmberg, J., Clarke, D. L., and Frisen, J. (2000) Regulation of repul-
sion versus adhesion by different splice forms of an Eph receptor,
Nature 408, 203–206.
34. Ciossek, T., Millauer, B., and Ullrich, A. (1995) Identification of alter-
natively spliced mrnas encoding variants of MDK1, a novel receptor
tyrosine kinase expressed in the murine nervous system, Onco-
gene 10, 97–108.
35. Valenzuela, D. M., et al. (1995) Identification of full-length and trun-
cated forms of Ehk-3a novel member of the Eph receptor tyrosine ki-
nase family, Oncogene 10, 1573–1580.
36. de Felipe, P., and Izquierdo, M. (2000) Tricistronic and tetracis-
tronic retroviral vectors for gene transfer, Hum. Gene Ther. 11,
1921–1931.
37. de Felipe, P., Martin, V., Cortes, M., Ryan, M., and Izquierdo, M.
(1999) Use of the 2A sequence from foot-and-mouth disease virus
in the generation of retroviral vectors for gene therapy, Gene Ther.
6, 198–208.
38. de Felipe, P., and Ryan, M. (2004) Targeting of proteins derived from
self-processing polyproteins containing multiple signal sequences,
Traffic 5, 616–626.
39. Fotos, J. S., Patel, V. P., Karin, N. J., Temburni, M. K., Koh, J. T., and
Galileo, D. S. (2006) Automated time-lapse microscopy and high-
resolution tracking of cell migration, Cytotechnology 51, 7–19.
320
VOL.5 NO.3
•
313–320
•
2010
SAUERS ET AL.
www.acschemicalbiology.org