Monitoring protein kinases in cellular media with highly selective chimeric reporters

Full text of DOI:10.1002/anie.200902374

Communications
DOI: 10.1002/anie.200902374
Lysate Assays
Monitoring Protein Kinases in Cellular Media with Highly Selective
Chimeric Reporters**
´
Elvedin Lukovic, Elizabeth Vogel Taylor, and Barbara Imperiali*
Protein kinases are important regulators of cellular function,
and the dynamics of their activities are critical indicators of
the health or pathology of living systems.[1,2] In particular,
extracellular signal regulated kinases 1 and 2 (ERK1/2) play a
pivotal role in the mitogen-activated protein kinase (MAPK)
signaling pathway responsible for regulated cell survival and
proliferation.[3] The centrality of these enzymes in normal and
diseased cell states underscores the need for high-throughput,
selective, and sensitive methods that accurately and directly
diagnose kinase activities. The benchmark phosphorylation
assays for ERK1/2 rely on transfer of radioactive g-phosphate
of [g-32P]ATP to peptide or protein substrates.[4] While
broadly employed, this approach has limitations, including
the discontinuous nature of the radioactive assay and the non-
native ATP concentrations that are utilized. Alternatively, for
cellular imaging, genetically encoded sensors that rely on
phosphorylation-based changes in fluorescence resonance
energy transfer (FRET) between fluorescent protein pairs[5,6]
have been constructed for several kinases, including ERK1/
2.[7–10] These sensors are powerful because they can be
expressed in cells; however, they cannot be used for high-
throughput screening of recombinant enzymes and unfractio-
nated cell lysates owing to the very limited fluorescence
changes that accompany phosphorylation. As a complemen-
tary approach, probes based on small, organic fluorophores
with direct readouts[6,11] can give sensitive and robust signals
under physiological conditions and are thus amenable to high-
throughput applications. For example, we have incorporated a
sulfonamido-oxine (Sox) chromophore into peptides[12,13] to
Figure 1. Sox-based chemosensors. a) The ERK1/2 probe utilizes the
PNT domain from Ets-1 for specific binding to the enzyme and senses
phosphorylation using Sox-dependent CHEF (lex =360 nm,
lem =485 nm). b) The Sox-PNT sensor is synthesized by native chem-
ical ligation (NCL). TEV=tomato etch virus protease; Bn=benzyl.
report phosphorylation by chelation-enhanced fluorescence
(CHEF; Figure 1a). The weak binding affinity of the unphos-
phorylated substrate for Mg2+ increases significantly upon
phosphorylation, resulting in robust (2- to 12-fold) fluores-
cence enhancements. This versatile peptide-based sensor
design has been applied to monitor the activity of numerous
Ser/Thr and Tyr kinases both in vitro[13] and in cell lysates.[12]
With more than 500 different kinases encoded in the
human genome, sensor selectivity becomes paramount,
particularly when enzymes are studied under conditions that
resemble their native environments, such as in unfractionated
cell lysates or live cells.[6] While many protein kinases exploit
linear recognition motifs comprising four to eight residues
that are proximal to the phosphorylation site to drive
specificity, a number of physiologically important kinases,
including ERK1/2 and other MAPKs, phosphorylate sub-
strates with short and ubiquitous consensus sequences. For
these enzymes, specificity is derived from extended recog-
nition elements that include protein–protein interactions
´
[*] E. Lukovic, Dr. E. Vogel Taylor, Prof. B. Imperiali
Departments of Chemistry and Biology
Massachusetts Institute of Technology
77 Massachusetts Ave., Cambridge, MA (USA)
Fax: (+1)617-452-2799
E-mail: imper@mit.edu
Homepage: http://web.mit.edu/imperiali/Home.html
[**] We thank Prof. K. A. Dalby for valuable advice and for providing
DNA of the PNT domain and the PEA-15 protein, and Dr. Kevin
Janes for helpful discussions. This research was supported by the
NIH Cell Migration Consortium (GM064346). The Biophysical
Instrumentation Facility for the Study of Complex Macromolecular
Systems (NSF-0070319) is also gratefully acknowledged.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200902374.
6828
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 6828 –6831

Angewandte
Chemie
distal to the phosphorylation site.[14] For example, ERK1 and
2 phosphorylate the transcription factor Ets-1 at Thr38 within
a short ThrPro (TP) consensus motif.[15] Since this short
sequence would be the target of multiple kinases, ERK
recognition of Ets-1 depends an adjoining N-terminal pointed
(PNT) domain[16] to dock the substrate specifically to ERK1/
2, which engages the phosphorylation machinery.[17] With this
docking-domain strategy, PNT-based substrates demonstrate
good affinity for ERK1/2 (KM ꢀ 6–9 mm),[16] in stark contrast
to short peptide substrates derived only from the TP sequence
(KM > 200 mm).[16,18] Owing to the size of the PNT domain
(11 kDa), these docking-domain interactions cannot be
exploited with the types of synthetic peptide-based sensors
that have been reported. However, in light of its importance
in cellular homeostasis and the prominence of ERK disregu-
lation in cancer, we set out to construct a chimeric sensor for
ERK1/2 that combines the advantageous reporting properties
of the Sox fluorophore with the outstanding specificity
provided by the native protein domain-based recognition
(Figure 1a). Most importantly, to ensure facile and high-
throughput analysis of ERK1/2 activity, our ultimate require-
ment is that the probe be highly selective in cell lysates where
it would be exposed to hundreds of other active kinases.
Herein we describe the semisynthesis of a chimeric Sox-
based ERK1/2 sensor through a key native chemical ligation
(NCL) reaction that efficiently conjugates the recombinat
PNT domain of Ets-1 to a synthetic ERK1/2 consensus
sequence including the Sox sensing module. The extended
PNT recognition element confers the ERK1/2 sensor with
excellent selectivity, as demonstrated by comparative quanti-
tative analyses with a panel of related recombinant enzymes
and in unfractionated lysates from four different cell lines.
Most importantly, the docking-domain-based sensor design
should be generally applicable to the development of
selective sensors for other medically important kinases.
The new sensor was assembled as illustrated in Figure 1,
using NCL[19] to ligate the synthetic Sox-containing peptide
thioester with the expressed PNT domain, comprising Ets-1
residues 46–138 (Figure 1b). The peptide thioester was
synthesized using Fmoc-based solid-phase peptide synthesis
(SPPS; Fmoc = fluorenylmethyloxycarbonyl) on highly acid-
labile TGT resin, with subsequent off-bead thioesterification
of the protected peptide.[20] The Sox chromophore was
introduced as the amino acid C-Sox.[13] An optimized
phosphorylation motif based on the ERK2 phosphorylation
sequence within the myelin basic protein (MBP;[18]
TPGGRR) was used in place of the phosphorylated region
of Ets-1 (TPSSKE). When the Sox chromophore was
incorporated into these short peptides, preliminary studies
indicated that the MBP-based sequence had better fluores-
cent properties (Table S1 in the Supporting Information). The
distance between the TP recognition sequence and the PNT
domain in the wild-type protein was preserved in the sensor
(Table S2 in the Supporting Information). This design intro-
duced residue replacements in the unstructured N-terminal
region of Ets-1, thereby minimizing perturbations to the
overall secondary structure. Additionally, the C-terminal
residue, Met44, was changed to Gly to eliminate the
possibility of epimerization during thioesterification and to
increase ligation efficiency.[21] The expressed C-terminal
fragment of the sensor, GST-PNT, was proteolyzed to reveal
Cys-PNT. After ligation of Cys-PNT to the peptide thioester
in nondenaturing conditions, the ERK1/2 probe Sox-PNTwas
isolated in good yield (24%; the accurate mass of the isolated
material was based on the molar absorptivity emax of the Sox
chromophore; see the Supporting Information). The corre-
sponding phosphoprotein (pThr38) pSox-PNT was con-
structed using analogous methods.
Initial spectroscopic studies with Sox-PNT and pSox-PNT
revealed a robust threefold enhancement in fluorescence
upon phosphorylation (Figures S2 and S3 in the Supporting
Information). Moreover, similar to Sox-based peptide sen-
sors,[13] Sox-PNT was found to have an excellent Z’ factor
value (0.81), which is a statistical quality parameter used to
evaluate and validate performance of assays, with useful
ranges being 0.5–1.[22] Subsequent in vitro assays determined
Sox-PNT to be an efficient substrate for ERK2 when
compared with the corresponding Sox peptide (Ac-VP-
CSox-LTPGGRRG-OH; Figure 2a and Figure S3 in the
Figure 2. In vitro characterization of Sox-PNT. a) The efficiency of
phosphorylation by recombinant ERK2 (11 ng) of the Sox-PNT probe
was compared to that of the Sox peptide under identical conditions.
b) The kinetic parameters for Sox-PNT were obtained with ERK2
(10 ng) from a direct fit of n vs. [S] plots using the Briggs–Haldane
equation. Plotted values indicate the mean standard error of measure-
ment (Æs.e.m.) for triplicate measurements. c) Promiscuity of Sox-
PNT (5 mm) was tested with a panel of related kinases at 15 nm (black
bars) and 150 nm (white bars) of each enzyme. Inset: a representative
plot of the change in the fluorescence signal over time obtained with
15 nm enzyme in the fluorescence plate reader. Plotted values for
15 nm enzyme indicate the mean Æs.e.m. for triplicate measure-
ments.
Angew. Chem. Int. Ed. 2009, 48, 6828 –6831
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6829

Communications
Supporting
Information).
Furthermore, Sox-PNT dem-
onstrated a similar KM value
(14.9 mm) and a slightly lower
catalytic efficiency (kcat/KM =
47 minÀ1 mmÀ1) compared to
the wild-type PNT (KM =
9 mm,
kcat/KM =
132 minÀ1 mmÀ1),[23] thus indi-
cating that use of the MBP-
derived TP sequence had
minimal effect relative to the
native protein (Figure 2b). In
contrast,
the
MBPtide
(APRTPGGRR), the basis
of the Sox-PNT phosphoryla-
tion sequence, was reported
to have substantially poorer
kinetic parameters (KM =
2 mm,
kcat/KM =
[18]
0.11 minÀ1 mmÀ1
)
for
ERK2, underscoring the
importance of the PNT
domain in substrate kinetics.
Finally, Sox-PNT exhibited
high selectivity for ERK1/2
when compared to related
kinases from the JNK, p38,
and CDK families (Fig-
ure 2c).
To demonstrate that the
preference of Sox-PNT for
ERK1/2 can translate to com-
plex media, the probe was
exposed to a panel of unfrac-
tionated cell lysates that con-
tained varying levels of active
ERK1/2. The activity of cel-
lular ERK1/2 is linked to its
phosphorylation state, which
is modulated by the epider-
mal growth factor (EGF) sig-
Figure 3. Specificity of the Sox-PNT sensor toward ERK1/2 in unfractionated cell lysates. a) The EGF
signalling pathway results in stimulation of ERK1/2 activity. U0126, an inhibitor of an upstream kinase
(MEK1/2) and PEA-15, a direct inhibitor of ERK1/2, can regulate the activity of ERK1/2. b) Sox-PNT (5 mm)
was used to measure enzyme activity in 40 mg untreated lysates (red bars), EGF-stimulated lysates (blue
bars), or U0126-treated and then EGF-stimulated lysates (black bars). Inset: western blot for pERK1/2 (top)
and ERK1/2 (bottom). c) The ability of Sox-PNT (black bars, 5 mm) to report the different phosphorylation
states of ERK1/2 in HeLa lysates (40 mg) was directly compared to the Sox peptide (white bars, 5 mm). d) To
obtain the IC50 value for inhibition of ERK1/2 by PEA-15, EGF-stimulated NIH-3T3 lysates (40 mg) were
treated with various concentrations of PEA-15, and enzyme activity was measured with Sox-PNT (5 mm).
e) Kinase activity was measured with Sox-PNT (5 mm) from an EGF-stimulated HeLa lysate (40 mg) before
(input) and after immunodepletion of this lysate with anti-ERK1/2 or naꢀve rabbit IgG. Inset: western blot
for ERK1/2 in the measured samples. The top band in all western blots is ERK1 (44 kDa) and the bottom
band ERK2 (42 kDa). Plotted values indicate the mean Æs.e.m. for triplicate measurements.
naling pathway (Figure 3a). Briefly, EGF interacts with EGF
receptors (EGFRs), which leads to activation of MEK1/2,
which in turn phosphorylates and activates ERK1/2. This
event can be regulated either by the upstream MEK1/2
inhibitor U0126 or by the direct ERK1/2 inhibitor PEA-15.[24]
Summarized in Figure 3b are the results of ERK1/2 activity
analyses on the crude lysates from four mammalian cell lines,
which reveal the selectivity of the sensor in these complex
media. In all cases, untreated lysates showed relatively low
basal activity, while displaying slight variation among differ-
ent cell lines, as expected. Upon EGF stimulation, there was a
four- to tenfold increase in activity.[25] To demonstrate that
Sox-PNT was specifically monitoring ERK1/2 activity, cells
were exposed to the inhibitor U0126 and subsequently EGF-
stimulated and lysed. Under these conditions, the ERK1/2
activity returned to nearly basal levels. Western blot analysis
was used to demonstrate that both stimulated and U0126
inhibitor-treated cells expressed ERK1/2 (Figure 3b), how-
ever, only EGF-stimulated samples showed enhanced levels
of activated ERK1/2, as evidenced by analysis with the
phospho-ERK1/2-specific antibody. In contrast, the Sox pep-
tide, lacking the PNT docking domain, showed promiscuous
activity and signaled phosphorylation that could not be
correlated with ERK1/2 activity (Figure 3c).
Further evidence that Sox-PNT is selectively modified by
ERK1/2 was obtained with PEA-15, a direct protein inhibitor
of ERK1/2. Titration of PEA-15 into EGF-stimulated NIH-
3T3 lysates resulted in a dose-dependent response with a half
inhibitory concentration of 40 nm and a Ki value (30 nm) that
reflected the reported Ki values (20 nm; Figure 3d and the
Supporting Information).[24]
To directly correlate the observed fluorescent signal to the
presence of ERK, we exposed our probe to ERK1/2-depleted
lysates. Indeed, immunodepletion of ERK1/2 from EGF-
6830 www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 6828 –6831

Angewandte
Chemie
stimulated HeLa lysate reduced activity by sevenfold com-
pared to the input lysate or the sample that had been depleted
with naꢀve rabbit immunoglobulin G (IgG, Figure 3e). This
finding indicates that the Sox-PNT signal is predominantly
due to the ERK1/2-mediated phosphorylation. Slight residual
activity in anti-ERK1/2 lysate can be attributed to incomplete
removal of the kinase, which is highly dependent on the
efficiency of antibody and kinase binding. Immunodepletion
was confirmed by western blot analysis with the immunode-
pleting antibody (Figure 3e, inset). Having validated the
selectivity of Sox-PNT for ERK1/2, the sensor can be used to
measure active ERK1/2 (13 ng) in EGF-stimulated lysate
(40 mg; Figure S5 in the Supporting Information), which will
be an important tool for quantifying ERK1/2 levels in tissue
samples.
The MAPK signaling pathways are composed of numer-
ous kinases intricately regulated by stress responses and
extracellular signals. Deconvoluting the specific functions of
individual enzymes has been challenging, partly owing to the
difficulty of creating probes that exclusively target a kinase of
interest. Herein we have presented a selective ERK1/2
activity chemosensor that comprises both a chemical sensing
motif and recombinant enzyme docking domain. Thereby, the
Sox-based kinase-sensing strategy has been extended beyond
the realm of enzymes that recognize linear peptide substrates.
Quantitative studies with the probe indicate that the PNT
domain confers exquisite selectivity toward ERK1/2, which
was impossible to achieve with simple peptide probes. More-
over, the docking-domain approach now allows us to target a
wider set of kinases (such as other members of the MAPK
family, JNK and p38) that have been elusive to date owing to
their complex substrate recognition mechanisms. The chi-
meric protein probe also offers distinct advantages for
solution-based analyses that can be carried out with simple
equipment. The reliable semisynthesis of multimilligram
(6 mg) quantities of Sox-PNT allows at least 5000 assays to
be performed in 384-well plates. Furthermore, Sox-based
sensors exhibit large dynamic ranges with excellent Z’ factor
values,[13] which are a measure of the fidelity of assay data.[22]
Currently, in light of the efficient semisynthesis, excellent
selectivity, and robustness in high-throughput analysis, the
Sox-PNT sensor can be broadly applied for quantifying
ERK1/2 activities in applications ranging from drug discovery
to diagnostics.
Keywords: enzymes · immunoassays · native chemical ligation ·
phosphorylation · protein kinase sensors
.
[1] G. Manning, D. B. Whyte, R. Martinez, T. Hunter, S. Sudarsa-
nam, Science 2002, 298, 1912.
[2] T. Hunter, Cell 2000, 100, 113.
[3] P. P. Roux, J. Blenis, Microbiol. Mol. Biol. Rev. 2004, 68, 320.
[4] K. A. Janes, J. G. Albeck, L. X. Peng, P. K. Sorger, D. A.
Lauffenburger, M. B. Yaffe, Mol. Cell. Proteomics 2003, 2, 463.
[5] Q. Ni, D. V. Titov, J. Zhang, Methods 2006, 40, 279.
[6] D. M. Rothman, M. D. Shults, B. Imperiali, Trends Cell Biol.
2005, 15, 502.
[7] H. M. Green, J. Alberola-Ila, BMC Chem. Biol. 2005, 5, 1.
[8] A. Fujioka, K. Terai, R. E. Itoh, K. Aoki, T. Nakamura, S.
Kuroda, E. Nishida, M. Matsuda, J. Biol. Chem. 2006, 281, 8917.
[9] M. Sato, Y. Kawai, Y. Umezawa, Anal. Chem. 2007, 79, 2570.
[10] C. D. Harvey, A. G. Ehrhardt, C. Cellurale, H. Zhong, R.
Yasuda, R. J. Davis, K. Svoboda, Proc. Natl. Acad. Sci. USA
2008, 105, 19264.
[11] V. Sharma, Q. Wang, D. S. Lawrence, Biochim. Biophys. Acta
Proteins Proteomics 2008, 1784, 94.
[12] M. D. Shults, K. A. Janes, D. A. Lauffenburger, B. Imperiali, Nat.
Methods 2005, 2, 277.
[13] E. Lukovic, J. A. Gonzalez-Vera, B. Imperiali, J. Am. Chem. Soc.
2008, 130, 12821.
[14] A. Remenyi, M. C. Good, W. A. Lim, Curr. Opin. Struct. Biol.
2006, 16, 676.
[15] C. E. Foulds, M. L. Nelson, A. G. Blaszczak, B. J. Graves, Mol.
Cell. Biol. 2004, 24, 10954.
[16] J. J. Seidel, B. J. Graves, Genes Dev. 2002, 16, 127.
[17] M. A. Rainey, K. Callaway, R. Barnes, B. Wilson, K. N. Dalby, J.
Am. Chem. Soc. 2005, 127, 10494.
[18] J. W. Haycock, J. Neurosci. Methods 2002, 116, 29.
[19] P. E. Dawson, T. W. Muir, I. Clark-Lewis, S. B. H. Kent, Science
1994, 266, 776.
[20] E. M. Vogel, B. Imperiali, Protein Sci. 2007, 16, 550.
[21] T. M. Hackeng, J. H. Griffin, P. E. Dawson, Proc. Natl. Acad. Sci.
USA 1999, 96, 10068.
[22] J. H. Zhang, T. D. Chung, K. R. Oldenburg, J. Biomol. Screening
1999, 4, 67.
[23] O. Abramczyk, M. A. Rainey, R. Barnes, L. Martin, K. N. Dalby,
Biochemistry 2007, 46, 9174.
[24] K. Callaway, O. Abramczyk, L. Martin, K. N. Dalby, Biochem-
istry 2007, 46, 9187.
[25] A. R. Asthagiri, A. F. Horwitz, D. A. Lauffenburger, Anal.
Biochem. 1999, 269, 342.
Received: May 4, 2009
Revised: June 22, 2009
Published online: August 13, 2009
Angew. Chem. Int. Ed. 2009, 48, 6828 –6831
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6831