Bioorthogonal imaging of Aurora Kinase A in live cells

Article abstract of DOI:10.1002/anie.201200994

In living color: Aurora kinase A (AKA) was imaged in live cells using a bioorthogonal two-step reaction with a small molecule AKA inhibitor (see scheme) and a fluorescent reporter. The fluorescent molecule was localized to spindle poles and microtubules d

Full text of DOI:10.1002/anie.201200994

.
Angewandte
Communications
DOI: 10.1002/anie.201200994
Protein Labeling
Bioorthogonal Imaging of Aurora Kinase A in Live
Cells**
Katherine S. Yang, Ghyslain Budin, Thomas Reiner, Claudio Vinegoni, and
Ralph Weissleder*
Angewandte
Chemie
6598
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 6598 –6603

Angewandte
Chemie
Cell division (mitosis) is a tightly controlled process that is
coordinated and regulated by a network of proteins localized
in the nucleus. The key stages of mitosis are centrosome
maturation, chromosome condensation, nuclear envelope
breakdown, centrosome separation, bipolar spindle forma-
tion, chromosome separation, and finally cytokinesis.[1]
Aurora kinase A (AKA) belongs to the Aurora kinase
family of serine/threonine kinases, which have been shown
to play critical roles in mitotic progression.[1–4] During mitosis,
AKA localizes to centrosomes during late S to early G2
phase. As the cell proceeds to metaphase, AKA localizes to
the microtubules and near the spindle poles, where it remains
until anaphase when it migrates to some extent to the spindle
midzone. Finally, during cytokinesis, AKA localizes to the
midbody.[1,3–5] Whilst localized to these specific cellular
regions, AKA interacts with and phosphorylates several
intracellular targets, including p53, MBD3, and BRCA1,
each of which are critical mediators of malignant trans-
formation.[1,5] The unique stage-specific nuclear and intra-
cellular locations of AKA during mitosis thus make it an
interesting imaging target.
AKA overexpression has been reported in the majority of
epithelial cancers[6] and has been shown to cause defects in
cell division, such as centrosome amplification and abnormal
spindle formation.[7] An imbalance of AKA during mitosis
may also alter the normal interactions between AKA and
tumor suppressors such as p53, possibly promoting cell
viability and tumorigenesis.[7] Because of the link between
AKA overexpression and tumorigenesis, the use of AKA
small molecule inhibitors as potential therapeutics for cancer
is now an area of great interest.[6] For example, 4-{[9-chloro-7-
(2,6-difluorophenyl)-5H-pyrimido[5,4-d][2]benzazepin-2-yl}-
amino]benzoic acid (MLN8054) is an AKA-specific inhibitor
that has about 100-fold greater selectivity for AKA over other
kinases, and has been shown to result in mitotic arrest and
apoptosis.[8]
imaging agents has impaired our ability to visualize and
quantify AKA expression directly, either in panels of live cells
or in patient-derived primary cancer cells (e.g. circulating
tumor cells or from fine needle aspirates). Consequently,
owing to the challenge of designing such agents, imaging
probes with appropriate AKA selectivity and kinetics thus
remain extremely limited. Antibodies, for example, are
largely ineffective since they rarely have adequate cell
permeability. Other techniques that have been used to
image AKA include the use of fixed cells and AKA-specific
antibodies, or exogenous expression of fluorescently tagged
AKA. Neither technique, however, enables accurate visual-
ization of AKA in its endogenous state. The ability to
visualize AKA within live cells, or even by way of whole body
imaging, would not only provide new insights into the biology
of AKA and its ideal pharmacological inhibition, but could
also improve testing of the clinical efficacy of new small
molecule AKA inhibitors. Herein, in a bid to address this
need, we sought to develop small molecule AKA imaging
agents using MLN8054 as a scaffold, together with a bio-
orthogonal two-step labeling method.[9,10] We initially modi-
fied the carboxylic acid of our MLN8054 precursor with trans-
cyclooctene (TCO). The resulting compound (MLN8054-
TCO; 13) was cell-membrane permeable and, upon binding
its target, could be visualized using a bioorthogonal tetrazine
(Tz)-labeled fluorophore.[11] We thus report the first use of
MLN8054-TCO (13) as an imaging agent for AKA in live
cells and as a tool for quantifying cellular AKA expression.
To design the modification of MLN8054 with TCO, we
used the crystal structure of AKA in complex with MLN8054
(PDB ID 2X81).[12] Analysis indicated that modification of
the carboxylic acid group would be the most viable side of the
molecule for derivatization since it extends toward the
solvent. MLN8054 was synthesized as described previously
with a minor modification (Scheme 1).[13] Briefly, ortholithia-
tion of the 4-chloro-N-Boc-aniline with tert-butyl lithium,
followed by addition of 2,6-difluorobenzoyl chloride, yielded
the benzophenone derivative 1. After aniline deprotection
and conversion of the amine 2 into an iodide 3, a Sonogashira
coupling was performed with an N-Boc propargylamine
(prepared following a described procedure)[13] affording the
alkyne 4. The amine group of 4 was then deprotected under
acidic conditions, followed by treatment with base, to allow
the direct intramolecular cyclization to form the azepinone
ring 5. Compound 6 was formed by treating 5 with N,N-
dimethylformamide dimethyl acetal in refluxing toluene. The
pyrimidinoazepine 7 (MLN8054) was prepared by treating
the enaminone 6 with 4-guanidinobenzoic acid in refluxing
ethanol in the presence of potassium carbonate. The carbox-
ylic acid 7 was then coupled to a piperidine derivative 10
(prepared in three steps from the 4-(N-Boc-amino)piperi-
dine) affording compound 11 in 79% yield. Finally, the amino
protective group was removed and the resulting primary
amine was treated with TCO-NHS, furnishing MLN8054-
TCO (13) with an overall yield of 6% in ten steps (Scheme 1).
Cycloaddition of MLN8054-TCO (13) to Texas Red-tetrazine
(TR-Tz) was initially investigated by mixing the two com-
pounds (0.25 mm), stirring for two minutes, and analyzing the
products by HPLC-MS. These spectra confirmed the quanti-
Despite interest in the involvement of AKA in cancer and
the development of several AKA inhibitors, lack of specific
[*] Dr. K. S. Yang,[+] Dr. G. Budin,[+] Dr. T. Reiner, Dr. C. Vinegoni,
Prof. R. Weissleder
Center for Systems Biology, Massachusetts General Hospital
185 Cambridge Street, Boston, MA 02114 (USA)
E-mail: rweissleder@mgh.harvard.edu
Prof. R. Weissleder
Harvard Medical School
200 Longwood Avenue, Boston, MA 02115 (USA)
[+] These authors contributed equally to this work.
[**] This work was supported by the National Institutes of Health (NIH)
grant number RO1EB010011 and P50CA086355, K.Y. was supported
by an NIH grant T32-CA079443 and T.R. was supported by a grant
from the German Academy of Sciences Leopoldina (LPDS 2009-24).
We thank Prof. Peter Sorger and Dr. Robert Yang for assistance with
cellWoRx and ImageRail and Paolo Fumene Feruglio for assistance
with image processing. We also thank Dr. Jonathan Carlson, Dr.
Greg Thurber, and Dr. Carlos Tassa for helpful input and discussions
about the manuscript.
Supporting information for this article (experimental details) is
available on the WWW under http://dx.doi.org/10.1002/anie.
201200994.
Angew. Chem. Int. Ed. 2012, 51, 6598 –6603
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6599

.
Angewandte
Communications
Scheme 1. Synthetic scheme of MLN8054-TCO (13). Boc=tert-butyloxycarbonyl; THF=tetrahydrofuran; TFA=trifluoroacetic acid; DCM=di-
chloromethane; Ms=methanesulfonyl; TEA=triethanolamine; DMF=dimethylformamide; HBTU=O-(benzotriazol-1-yl)-N,N,N’,N’-tetramethy-
luronium hexafluorophosphate; DIPEA=N,N-diisopropylethylamine; NHS=N-hydroxysuccinimide.
tative conversion of TR-Tz into the cycloaddition product
MLN8054-TR (Figure S1).
(13) in growth medium (1.5 mm, 0.1% DMSO). Excess probe
was washed out and TR-Tz was added in the growth medium
for 20 minutes (1 mm, 0.1% DMSO). Cells were fixed for
DNA staining with Hoechst dye and imaged with a DeltaVi-
sion deconvolution microscope (Figure 1). GFP-AKA local-
ized to the microtubules near the spindle poles during
metaphase, as previously reported[3] (Figure 1a). As pre-
dicted, we observed a very similar microtubule localization of
the small-molecule MLN8054-TCO (13), as revealed by
reaction with TR-Tz (Figure 1b). Subsequent merging of
the GFP-AKA and MLN8053-TCO (13) images, in addition
to Hoechst staining (Figure 1c), revealed significant colocal-
ization of the GFP and Texas Red signals (yellow, Figure 1d).
Control experiments verified that neither TR-Tz nor the pre-
treated compound MLN8054-TR exhibit significant fluores-
cent staining in PANC-1 tumor cells (Figure S3). These
experiments not only demonstrate the low background
signal of the fluorophore but also highlight the advantage of
the bioorthogonal approach over a directly conjugated drug–
fluorophore pair (Figure S3). When unlabeled MLN8054 (7)
was used, no significant staining was observed with TR-Tz
(Figure S3B). Given the unique spatial localization of AKA
during mitosis, we next determined whether it would be
possible to image its distribution using the bioorthogonal
approach. Accordingly, after treating PANC-1 cells as above,
The inhibitory effects of MLN8054 (7), MLN8054-TCO
(13), TR-Tz, and the pre-treated MLN8054-TR on AKAwere
evaluated using a kinase activity assay (see Supporting
Information). The IC50 value of MLN8054 (7) was 1.5 nm.
Analysis of MLN8054-TCO (13) resulted, as expected, in
a lower, but still acceptable, IC50 value of 61.8 nm (Figure S2
and Table S1). In contrast, the pre-treated fluorescent
MLN8054-TR had an affinity of only 437.3 nm (Figure S2
and Table S1), which supported the use of the two-step
labeling method for imaging. Given the reasonable enzyme
affinity achieved with MLN8054-TCO (13), we sought to
image AKA within cells. This imaging was performed in two
separate experiments, in which: 1) AKA was tagged in live
cells with MLN8054-TCO (13) followed by TR-Tz treatment,
and then imaged in fixed cells, and 2) AKA was tagged in live
cells with MLN8054-TCO (13) followed by treatment with
a membrane-permeable carboxyfluorescein diacetate-Tz, and
then imaged in live cells.
To investigate colocalization between AKA and the
bioorthogonal small molecule in fixed cells, we designed
pancreatic cancer cells (PANC-1) to express a green fluo-
rescent protein (GFP)/AKA fusion protein (GFP-AKA).
Cells were first incubated for 20 minutes with MLN8054-TCO
6600 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 6598 –6603

Angewandte
Chemie
Figure 1. Colocalization of MLN8054-TCO (13)/TR-Tz with GFP-AKA.
PANC-1 Tet-On cells were incubated for 20 min with 1.5 mm MLN8054-
TCO (13), washed, and incubated for 20 min with 1 mm Texas Red-Tz
for bioorthogonal reaction inside the cells. 40ꢀ images collected by
deconvolution microscopy. a) GFP-AKA localization in PANC-1 cells,
b) MLN8054-TCO (13)/TR-Tz staining, c) Hoechst staining of fixed
nuclei, d) merge. Scale bar: 10 mm.
Figure 2. Expression of AKA in ten cell lines measured using
MLN8054-TCO (13). Cell lines were plated in duplicate in 96-well
plates 48 hr prior to assay. Cells were first incubated with 1.5 mm
MLN8054-TCO (13) for 20 min (0.1% DMSO). After washing, cells
were incubated with 1 mm TR-Tz (0.1% DMSO). Following washing,
cells were fixed and processed for staining with an a-AKA antibody
and Hoechst. Integrated fluorescence intensity was measured by
imaging at 10ꢀ with a cellWoRX microscope (Applied Precision).
a) Percent maximum signal of the integrated intensity for each cell line
was calculated for the TRITC channel (MLN8054-TCO (13)/TR-Tz) and
the FITC channel (AKA antibody) using ImageRail software.[14] b) Corre-
lation between the percent maximum signal for the AKA antibody (x-
axis) and MLN8054-TCO (13)/TR-Tz (y-axis). Data were fit by linear
regression using GraphPad (Prism), with an R2 =0.93. Data represent
at least two independent experiments.
we found equally good colocalization of the two dyes during
different stages of the G2/M phase (Figure S4). Interestingly,
GFP-AKA expression and MLN8054-TCO (13) staining were
both significantly reduced in interphase, a period when AKA
is not significantly expressed (Figure S4A).[2,3] The results
also demonstrated that bioorthogonal chemistry within live
cells is possible and is sufficiently fast and selective under
biological conditions.
We next determined whether the two-step AKA imaging
procedure could be applied to other cancer cell lines. To
compare small molecule AKA imaging to standard AKA
antibody staining (which requires fixed cells), we selected
eight cancer cell lines (HeLa, PANC-1, A549, SKOV-3,
MDA-MB-231, U-87 MG, SW480, and MIA PaCa-2) and two
normal cell lines (IMR-90 and HEK-293). Cells were plated in
duplicate in 96-well plates and were treated with MLN8054-
TCO (13) then TR-Tz, as described above. Each well was
imaged at 10ꢀ magnification to examine the entire population
of cells. The fluorescence intensity for both the AKA
antibody channel and the Texas Red channel were then
determined separately using ImageRail software.[14] As shown
in Figure 2a, antibody staining revealed that each normal and
cancer cell line expressed different levels of AKA. Similar
variations in AKA expression levels were likewise detected
following the small molecule two-step labeling procedure,
with good correlation (Figure 2a and b, R2 = 0.93).
(Figure S5). Rather, TR-Tz appears to enter live cells through
endocytosis. The bright, punctate staining emanating from the
endocytic vesicles likely served to mask the localization of
TR-Tz to MLN8054-TCO (13)/AKA, as observed following
fixation/permeabilization/washing in the fixed cell scenario
described above. To identify alternative fluorophores suitable
for live cell imaging, we screened a library of various Tz-
modified fluorophores to determine which might exhibit
diffusion-mediated cellular uptake. Ten dyes with an emission
wavelength in the red or green spectral region were chosen
(Figure S5). For this experiment, we also engineered a PANC-
1 cell line to stably express mCherry-AKA (RFP-AKA) so as
to visualize dyes emitting in the green spectral region. With
the notable exception of carboxyfluorescein diacetate-Tz
(CFDA-Tz), most fluorophore-Tz dyes tested showed lower
While the feasibility experiments described above were
performed using fixed cells, our ultimate goal was to develop
a method in which AKA could be imaged in live cells.
Unfortunately, cellular uptake of Texas Red (and its Tz
derivatives) was found to be limited in intact live cells
accumulation
in
live
cells.
Boron-dipyrromethene
(BODIPY)-Tz and nitrobenzoxadiazole (NBD)-Tz also
showed some colocalization with AKA but at the expense
of higher background signals; on account of this, they were
not pursued further (Figure S5).
Angew. Chem. Int. Ed. 2012, 51, 6598 –6603
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6601

.
Angewandte
Communications
CFDAwas measured at 526 nm (Figure 3a). Hydrolysis of the
acetate moieties was rapid, reaching a plateau after only
60 minutes of incubation with 20 mg of cell lysate. While
decreasing the quantity of cell lysate slowed the kinetics of
activation, the same signal plateau was reached after a few
hours. Without cell lysate, the dye remained in its native state.
The fluorescent signal before and after activation was shown
to increase by a factor of approximately ten (Figure 3b).
We subsequently investigated the use of CFDA-Tz in the
two-step labeling procedure, using the red fluorescent protein
(RFP)/AKA fusion (RFP-AKA) PANC-1 cell line. Live cells
were imaged in a warm, humidified imaging chamber
equipped with a microscope. RFP-AKA was seen to localize
to centrosomes and microtubules/spindles during metaphase
of mitosis (Figure 4a). A strikingly similar localization was
observed for MLN8054-TCO (13)/CFDA-Tz (Figure 4b).
Merging these images together, along with the DRAQ5
DNA stain, again demonstrated a high degree of colocaliza-
Figure 3. Activation properties of CFDA-Tz. Left: Activation of CFDA
fluorescence by esterases in the cell lysate. a) CFDA-Tz dye (20 mm)
was incubated at room temperature with HT1080 cell lysate (0, 1, 2.5,
5, 10, or 20 mg). Fluorescence emission was recorded at 526 nm over
time; b) Emission spectrum of CFDA-Tz before and after activation by
lysate. CFDA-Tz dye (20 mm) was incubated with HT1080 cell lysate (0
or 10 mg) for 40 min at room temperature. The fluorescence emission
scan was then recorded.
Carboxyfluorescein diacetate (CFDA) is a nonfluorescent,
cell-permeable dye, activated by nonspecific cellular ester-
ases.[15] For these experiments, we chose to synthesize a Tz
conjugate of this dye and evaluate its activation properties in
cell lysates. CFDA-Tz was incubated with varying amounts of
HT1080 cell lysate, and the fluorescence from activated
Figure 4. Live cell colocalization of MLN8054-TCO (13)/CFDA-Tz in
PANC-1 Tet-On cells expressing RFP-AKA. PANC-1 cells were incubated
for 30 min with 125 nm MLN8054-TCO (13). Cells were washed and
incubated for 30 min with 187.5 nm CFDA-Tz for bioorthogonal reac-
tion inside living cells. After washing, cells were incubated for 10 min
with DRAQ5 to stain nuclei in live cells. 40ꢀ images with 1.6 zoom
were collected by deconvolution microscopy. a) PANC-1 cellular local-
ization of RFP-AKA, b) MLN8054-TCO (13)/CFDA-Tz staining,
c) DRAQ5 staining of nuclei, d) merge. Scale bar: 10 mm.
Figure 5. Localization of MLN8054-TCO (13)/CFDA-Tz in dividing
HT1080 H2B-Apple cells. HT1080 cells were incubated for 30 min with
125 nm MLN8054-TCO (13), washed and incubated for 30 min with
1 mm CFDA-Tz for bioorthogonal reaction inside living cells. Following
2 h of washing, live cells at different stages of mitosis were imaged at
40ꢀ with 1.6 zoom by deconvolution microscopy. Left column:
MLN8054-TCO (13)/CFDA-Tz, Middle column: H2B-Apple, Right
column: merge. a) Interphase, b) Prophase, c) Prometaphase, d) Meta-
phase, e) Anaphase, f) Telophase, g) Cytokinesis. Scale bar: 10 mm.
6602 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 6598 –6603

Angewandte
Chemie
tion (Figure 4c,d). Control experiments were performed
using an HT1080 fibrosarcoma cell line stably transfected
with H2B-Apple (a red fluorescent protein), to enable easy
identification of actively dividing cells. These experiments
served to further demonstrate that neither CFDA-Tz nor the
pre-treated compound MLN8054-CFDA exhibit significant
fluorescent staining in tumor cells (Figure S6). Labeling of
endogenous AKA in live cells with MLN8054-TCO (13)/
CFDA-Tz during different stages of the G2/M phase was also
examined using the HT1080 H2B-Apple cell line (Figure 5).
In this experiment, MLN8054-TCO (13)/CFDA-Tz showed
very good correlation with the expected localization of AKA
during mitosis. Overall, our results not only show that the
MLN8054-TCO (13) bioorthogonal two-step labeling proce-
dure is useful for labeling fixed cells, but also demonstrate
that this technique can be used to label targets for imaging in
living cells.
In conclusion, we show that endogenous AKA can be
imaged in live cells by way of a two-step bioorthogonal
reaction. As an imaging agent, MLN8054-TCO (13) in
combination with CFDA-Tz demonstrated superior selectiv-
ity, fast reaction kinetics, and a sufficient target-to-back-
ground ratio to enable AKA visualization. These results build
on previous studies that have used two-step bioorthogonal
imaging approaches to visualize both cell surface[16] and
cytoplasmic targets.[11,17,18] This technique is an important step
towards designing agents for imaging AKA at the whole-body
level, for example, for positron emission tomography or for
use with near-infrared optical imaging techniques. We believe
that the described strategy for bioorthogonal labeling of a cell
cycle-specific intracellular/nuclear target will likely enable
the labeling of other intracellular targets. Furthermore, we
expect that this approach will allow us to also examine how
labeled AKA drugs behave and interact not only in live cells
but also within the tumor microenvironment of mice.
Keywords: aurora kinase · cancer · cell cycle · cycloaddition ·
live cell imaging
.
[1] T. Marumoto, D. Zhang, H. Saya, Nat. Rev. Cancer 2005, 5, 42.
[2] G. Vader, S. M. Lens, Biochim. Biophys. Acta Rev. Cancer 2008,
1786, 60.
[3] M. Carmena, W. C. Earnshaw, Nat. Rev. Mol. Cell Biol. 2003, 4,
842.
[4] A. R. Barr, F. Gergely, J. Cell Sci. 2007, 120, 2987.
[5] R. D. Carvajal, A. Tse, G. K. Schwartz, Clin. Cancer Res. 2006,
12, 6869.
[6] S. M. Lens, E. E. Voest, R. H. Medema, Nat. Rev. Cancer 2010,
10, 825.
[7] J. Fu, M. Bian, Q. Jiang, C. Zhang, Mol. Cancer Res. 2007, 5, 1.
[8] M. G. Manfredi et al., Proc. Natl. Acad. Sci. USA 2007, 104,
4106.
[9] M. L. Blackman, M. Royzen, J. M. Fox, J. Am. Chem. Soc. 2008,
130, 13518.
[10] N. K. Devaraj, R. Weissleder, S. A. Hilderbrand, Bioconjugate
Chem. 2008, 19, 2297.
[11] N. K. Devaraj, S. Hilderbrand, R. Upadhyay, R. Mazitschek, R.
Weissleder, Angew. Chem. 2010, 122, 2931; Angew. Chem. Int.
Ed. 2010, 49, 2869.
[12] D. A. Sloane et al., ACS Chem. Biol. 2010, 5, 563.
[13] T. T. Denton, X. Zhang, J. R. Cashman, J. Med. Chem. 2005, 48,
224.
[14] B. L. Millard, M. Niepel, M. P. Menden, J. L. Muhlich, P. K.
Sorger, Nat. Methods 2011, 8, 487.
[15] B. Rotman, B. W. Papermaster, Proc. Natl. Acad. Sci. USA 1966,
55, 134.
[16] N. K. Devaraj, R. Upadhyay, J. B. Haun, S. A. Hilderbrand, R.
Weissleder, Angew. Chem. 2009, 121, 7147; Angew. Chem. Int.
Ed. 2009, 48, 7013.
[17] T. Reiner, S. Earley, A. Turetsky, R. Weissleder, ChemBioChem
2010, 11, 2374.
[18] G. Budin, K. S. Yang, T. Reiner, R. Weissleder, Angew. Chem.
2011, 123, 9550; Angew. Chem. Int. Ed. 2011, 50, 9378.
Received: February 6, 2012
Published online: May 29, 2012
Angew. Chem. Int. Ed. 2012, 51, 6598 –6603
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6603