Chemistry & Biology
Knowledge-Based Design of a Biosensor
to Quantify Localized ERK Activation
in Living Cells
Lutz Kummer,1Chia-Wen Hsu,2Onur Dagliyan,3Christopher MacNevin,2Melanie Kaufholz,4Bastian Zimmermann,4
Nikolay V. Dokholyan,3Klaus M. Hahn,2and Andreas Plu ¨ckthun1,*
1Department of Biochemistry, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland
2Department of Pharmacology
3Department of Biochemistry and Biophysics
School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA
4Biaffin GmbH & Co KG, Heinrich-Plett-Strasse 40, 34132 Kassel, Germany
Investigation of protein activation in living cells is
enced by the full complement of upstream regulators
they experience. Here, we describe the generation
of a biosensor based on the DARPin binding scaf-
fold suited for intracellular applications. Combining
library selection and knowledge-based design, we
created an ERK activity biosensor by derivatizing a
DARPin specific for phosphorylated ERK with a sol-
vatochromatic merocyanine dye, whose fluores-
cence increases upon pERK binding. The biosensor
specifically responded to pERK2, recognized by its
conformation, but not to ERK2 or other closely
related mitogen-activated kinases tested. Activated
endogenous ERK was visualized in mouse embryo
nucleus, perinuclear regions, and especially the
nucleoli. The DARPin-based biosensor will serve as
a useful tool for studying biological functions of
ERK in vitro and in vivo.
Traditional methods for studying signal transduction cascades
are based on biochemical methods, which have provided valu-
However, these methods can only probe interactions with spe-
cific proteins; they do not report activation of proteins in their
native environment within living cells, and ignore transport pro-
cesses and diffusion. Fluorescent biosensors based on affinity
tifying protein activity in vivo (Gulyani et al., 2011; Nalbant et al.,
2004; MacNevin et al., 2013). At present, specific binding probes
areoftenproteins, in mostcases antibodies.Althoughantibodies
tantly, their reliance on disulfide bonds hampers their use in the
reducing cytoplasmic milieu when expressed as intrabodies.
These problems led to the development of alternative families
of target-binding proteins based on stable polypeptide scaffolds
devoid of cysteine residues and disulfide bonds, thus ideally
suited for applications in reducing cellular environments (Binz
et al., 2005). As a prominent example, designed ankyrin repeat
proteins (DARPins) possess remarkable biophysical properties,
which are more favorable than those of antibody fragments for
their use in the design of biosensors (Brient-Litzler et al., 2010).
DARPins are based on domains consisting of ankyrin repeats
that are present in a great number of proteins across all phyla
and are involved in specific recognition between proteins
(Mosavi et al., 2004). A consensus design-based approach
was used to generate combinatorial libraries of DARPins by
sequence and structure analyses (Binz et al., 2003). DARPins
consist of 33-amino acid-long, consecutive homologous struc-
tural modules with fixed framework and variable potential inter-
action residues, which stack together to form elongated protein
domains (Binz et al., 2003). Specific high-affinity binders derived
from DARPin libraries can be generated against virtually any pro-
tein antigen by in vitro selection (Binz et al., 2004; Boersma and
Plu ¨ckthun, 2011; Kawe et al., 2006; Zahnd et al., 2006) and can
serve as the basis for the design of biosensors using fluores-
cence readouts, such as BRET (Kummer et al., 2012), or via
the attachment of environmentally sensitive dyes (Brient-Litzler
et al., 2010). Importantly, the defined interaction surface and
the uniformity of the DARPin scaffold simplify the sensor design
through knowledge-guided attachment of fluorophores, thus
minimizing previously required extensive optimization steps in
order to yield functional biosensors (Brient-Litzler et al., 2010;
Miranda et al., 2011; Nalbant et al., 2004).
binding to the respective target by attachment of a bright solva-
tochromatic fluorophore, which has emissive properties that
are dependent on the solvent environment. When positioned
appropriately in the binding protein, the exposure of the dye to
a hydrophobic environment, which forms upon target binding,
within the new protein-protein interaction interface causes a
change in fluorescence intensity and/or lmax. Specifically, we
Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 847
have previously described a set of highly fluorescent fluoro-
phores of the merocyanine family, which have been optimized
to be part of protein-based biosensors in living cells (Gulyani
et al., 2011; Nalbant et al., 2004; Toutchkine et al., 2003, 2007a,
2007b). The dyes can be excited at long wavelengths
cence. In addition, their bright fluorescence in hydrophobic envi-
of low concentrations of biosensor for the detection of endoge-
nous, unaltered target proteins. Both properties, brightness and
long wavelength, guarantee sensitive detection and use of low
anisms. Here we used a DARPin-based biosensor to study pat-
terns of extracellular signal-regulated kinase (ERK) activity in
to mapping ERK function without perturbing cell physiology.
ERK belongs to the family of mitogen-activated protein ki-
nases (MAPKs), a class of serine/threonine kinases that includes
the ERK, c-Jun N-terminal kinases (JNKs), and p38 subfamilies
(Chen et al., 2001). MAPKs regulate several physiological pro-
cesses and play a role in pathological phenomena, including
inflammation, apoptotic cell death, oncogenic transformation,
tumor cell invasion, and metastasis (Pearson et al., 2001). They
are part of a three-tiered phospho-relay cascade consisting of
a MAPK, a MAPK kinase (MAPKK), and a MAPKK kinase
(MAPKKK). Members of the ERK subfamily respond to stimuli
that induce cell proliferation and differentiation (Chen et al.,
2001). Depending on the cell type and the primary stimuli, ERK
activity spans different subcellular compartments (Chen et al.,
growth factor (EGF) leads to a significant accumulation of acti-
vated ERK in the nucleus, which is essential for morphological
transformation (Cowley et al., 1994). In contrast, EGF stimulation
in PC12 cells triggers cytoplasmic ERK activity only, resulting in
cellular proliferation, whereas neural growth factor causes
nuclear localization, which is essential for neuronal differentia-
tion (Marshall, 1995). Because ERK is expressed to varying
extents in all tissues responding to different stimuli, a biosensor
that enables visualization of activation patterns in vivo would be
valuable to dissect the multiple outcomes of ERK signaling. In
contrast to existing ERK activity reporters (Fujioka et al., 2006;
Green and Alberola-Ila, 2005; Harvey et al., 2008; Sato et al.,
2007), the dye-conjugated DARPin biosensor directly targets
the active ERK conformation, rather than relying on the phos-
phorylation of a diffusible substrate susceptible to modification
by both kinases and phosphatases.
In our study, we present a biosensor based on the DARPin
scaffold that responds specifically to active doubly phosphory-
latedERK(pERK). Bya structure-guided approach,solvatochro-
matic dyes were tested at different positions adjacent to the
DARPin binding interface. We identified sites for which dye
attachment resulted in a specific increase of fluorescence
through bound pERK2. Structural modeling of the biosensor in
complex with the target confirmed our design approach and re-
tion, the created biosensor reported selectively on the changing
dynamics and location of ERK activity upon stimulation or inhibi-
tion of ERK signaling in living cells.
Design of DARPin Conjugates
here are based on DARPin pE59, which has been selected by
ribosome display to bind selectively to phosphorylated ERK2
(pERK2) (Kummer et al., 2012). pE59 reliably detects pERK,
but does not bind to any other closely related member of the
MAPK family in either phosphorylated or nonphosphorylated
form, as shown both in vitro and in cellular assays (Kummer
et al., 2012). It also does not bind to the unphosphorylated
form of ERK, for which a reciprocally specific binder, E40, has
been developed as well (Kummer et al., 2012).
The atomic structure of pE59 in complex with pERK2 was
determined (Protein Data Bank [PDB] ID code 3ZUV) (Kummer
et al., 2012). This allowed us to design and construct reagentless
fluorescent biosensors by a structure-guided integration of sol-
vatochromatic fluorophores into pE59, using a similar principle
as described recently (Brient-Litzler et al., 2010), but with mero-
cyanine dyes with excitation and emission maxima at a much
higher wavelength than used earlier. An ideal site for attachment
of a solvent-sensitive fluorophore must satisfy two criteria. First,
the coupling site should be proximal to the putative binding site
and thus offer a considerable change in the environment of the
dye in the free and complexed states of pE59. Second, the
attached fluorophore should neither disrupt binding between
pE59 and pERK2 nor change the binding specificity of the
DARPin significantly. By applying these criteria, potential
coupling sites for solvent-sensitive dyes of the merocyanine
family (Toutchkine et al., 2003, 2007a, 2007b) were identified
by calculating changes in the accessible surface area (DASA)
between the free and bound states of pE59 (Table 1). In total,
we identified six sites in DARPin pE59 that are located in the
neighborhood of pERK2 in the complex but are not involved in
the interaction (Figure 1; Table 1). Residues Leu53 and Ile119
were rejected, because attachment of merocyanine dyes at
these two positions probably interferes with binding to pERK2,
as judged by their location in the pE59-pERK2 binding interface
(Figure 1). Although no change of the ASA was calculated for
Gly91 upon binding of pERK2, its location appeared to be prom-
ising for a possible solvatochromatic effect on the attached fluo-
rophore, considering the molecular size of merocyanine dyes.
Table 1. Analysis of the Interface between DARPin pE59
DDG (kcal mol?1)b
aVariation ofaccessible surface areabetween the freeand pERK2-bound
states of pE59.
bDifference in DG of protein folding between DARPin pE59 and pE59
cysteine mutants, calculated by ROSETTA (Leaver-Fay et al., 2011).
cContacts between pE59 and pERK2 in the structure of the complex.
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
848 Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved
Production and Conjugation of DARPin-Based Sensors
The four residues Gly91, Gly113, Asn123, and Gly124 were
mutated individually into cysteine by site-directed mutagenesis
of the coding gene. Cysteine-containing
expressed in the cytoplasm of Escherichia coli and purified by
immobilized metal-ion-affinity chromatography (IMAC) through
their hexahistidine tag with yields comparable to wild-type
pE59 (pE59-wt) (?60–80 mg/l). Calculated differences of DG in
protein folding of pE59-wt and pE59 mutants using ROSETTA
(Leaver-Fayetal.,2011)suggested adestabilizing effect ofintro-
duced cysteines at the designated positions (DDG > 0 kcal
mol?1) (Table 1). Thus, pE59 mutants were analyzed by size-
exclusion chromatography under reducing conditions and by
ELISA to prove their structural integrity and functionality. All
tested DARPin mutants eluted at the same volume as pE59-wt
(Figure S1 available online). Furthermore, the pE59 mutants
bound pERK2 similarly to pE59-wt and retained their specificity
for the doubly phosphorylated form of ERK2 (Figure S2). We
conclude that the introduction of cysteines at the stated posi-
tions did not alter the favorable characteristics of pE59-wt.
In a next step, functional pE59 mutants were covalently deriv-
atized with a diverse set of merocyanine dyes bearing cysteine-
reactive iodoacetamide side chains (Toutchkine et al., 2003,
2007a, 2007b) (Figure S3). The products of the coupling reac-
tions were separated from the free fluorophore by size-exclusion
chromatography. Coupling yields were calculated from the
measured protein concentration and the absorbance spectrum
of the purified product. Dye:protein molar ratios were between
0.8 and 1.1 in all cases. Control reactions with pE59-wt, which
has no cysteine, yielded dye:protein ratios <0.09.
Properties of DARPin Conjugates
The fluorescence response of the conjugates was assayed in the
presence of ERK2 and pERK2. Three conjugates showed a sub-
stantial response: dye mero53 at position 91 (pE59-C91m53),
dye mero60 at position 123 (pE59-C123m60), and dye mero87
at position 123 (pE59-C123m87) (Figure 2A). The closely related
dyes mero53, mero60, and mero87 showed far stronger
responses than other tested fluorophores (mero61, mero62,
and mero199), but the optimal response for each set occurred
at different attachment positions, probably reflecting the size
and orientational preferences of the dye. The best-performing
dye attachment sites were both in the C-terminal part of DARPin
pE59, in the vicinity of the pE59-pERK2 interface (Kummer et al.,
2012). Position 91 is located in the loop connecting a helices 1
and 2 of internal repeat module 2, directly adjacent to residues
that are in contact with pERK2 and that were randomized
in the original design (Figure 1). Position 123 is part of a helix 1
in the C-terminal capping repeat, following residues involved in
the interaction with pERK2 (Figure 1). Of the three variants,
sensor pE59-C123m87 was highly specific for pERK2 and
showed promising signal brightness. In contrast, for the other
two conjugates, pE59-C91m53 and -C123m60, selectivity for
pERK2 over ERK2 was diminished (Figure 2A). To elucidate the
difference in selectivity of conjugate variants at position 123
bearing different merocyanine dyes, we analyzed pE59-C123m
variants qualitatively for pERK2 binding selectivity with ELISA
the most pronounced specificity, and to a lesser extent pE59-
C123m199. Selectivity in fluorescence response toward ERK2
and pERK2 was observed for variant pE59-C123m60, whereas
no significant selectivity was detected by measuring binding in
ELISA experiments (Figure 2A; Figure S4), but this can be
explained by the saturation of the ELISA under these conditions.
Surface plasmon resonance (SPR) measurements on a Bia-
core instrument were carried out with the most promising sensor
variant pE59-C123m87 to quantify a possible perturbation of
stant for pE59-C123m87 was 457 nM, compared to 117 nM of
native pE59-wt (Kummer et al., 2012), revealing a 4-fold
decrease of affinity for the given sensor variant upon incorpora-
tion of the cysteine and the dye (Figure S5). In accordance with
ELISA results (Figure S4), the pERK2 binding selectivity of
the conjugate was still remarkably high (selectivity pE59-
C123m87 >23), but decreased by a factor of 3 compared to
pE59-wt (selectivity pE59-wt >74; Kummer et al., 2012). Due to
referred to as pE59RFD biosensor henceforth.
To characterize the fluorescence properties of the pE59RFD
biosensor, we performed titration experiments with ERK2 and
pERK2 (Figure 2B). The conjugate only responded to increasing
amounts of pERK2. In contrast, no significant rise in fluores-
cence was detected for ERK2, even at high antigen concentra-
tions, which confirmed the sensor selectivity for the phosphory-
lated kinase form. As a negative control, we incorporated a
cysteine at the N terminus of pE59, a region that is not involved
in the interaction between pE59 and pERK2 and is therefore not
expected to offer a solvatochromatic effect upon target binding.
As expected, N-terminal mero87-labeled pE59 (pE59-ctrl)
showed no response to ERK2 or pERK2 (Figures 2B and 2C),
confirming that the detected pE59RFD response was indeed a
result of our design approach. Biosensor specificity was further
analyzed by fluorescence assays with purified active and inac-
tive forms of the MAPK ERK2, JNK1a1, JNK2a1, and p38a.
The different MAPK family members have marked sequence
and structure homologies, with a sequence identity of >40%
over the highly conserved catalytic core. In agreement with the
Figure 1. Sites for Dye Labeling in the Structure of DARPin pE59
Ribbon diagram of pERK2-specific DARPin pE59 (gray, based on pERK2-
pE59 complex structure PDB ID code 3ZUV). Residues Gly91, Gly113,
Asn123, and Gly124, which were replaced by cysteine for site-specific dye
attachment, are shown in space-filling representation (red). Asn123 marks the
position of dye attachment in the final pE59RFD biosensor. Residues in con-
tact with pERK2 (yellow) in the complex structure are highlighted in green. See
also Figures S1 and S2.
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 849
binding specificity previously observed for pE59-wt (Kummer
et al., 2012), pE59RFD was highly specific for its cognate target
pERK2 and did not interact with any other MAPK or unphos-
phorylated ERK2 (Figure 2C). Again, the simultaneously assayed
sensor-incompetent pE59-ctrl showed no marked response to
any tested MAPK either (Figure 2C).
To address applications in living cells, we assessed possible
interfering effects from cellular components on the sensor
response. pE59RFD selectivity was analyzed by mixing the
ERK2 or pERK2 antigen with human embryonic kidney (HEK)
293T cell lysate. The lysate derived from nonstimulated cells
resulted in a low level of endogenous pERK2, as confirmed in
tivity remained remarkably high in the crude lysate and the
response range was only slightly diminished compared to buffer
samples (Figure 2D). Furthermore, the specificity of the sensor
EpE82, an ERK/pERK binder that had been selected in a previ-
ous study (Kummer et al., 2012). DARPin EpE82 binds to the
very same region of pERK2 as pE59, but with a considerably
higher affinity, and is thus ideally suited for complete inhibition
response of pE59RFD was completely abolished in samples
of background fluorescence in cell lysate samples with or
without EpE82 (Figure 2D, lysate) when compared to cell lysis
buffer samples (Figure 2D, black and white bar), indicating an
interfering effect from other cellular components. Nonetheless,
the dynamic range of pE59RFD remained substantially high.
Modeling of the pERK2-pE59RFD Interface
To understand the molecular mechanism of specificity between
the sensor and pERK2, we modeled the pE59-pERK2 complex
with the sensor-active dye mero87 and sensor-inactive dye
mero53 attached that are chemically similar. We used the crystal
structure of pE59-pERK2 (PDB ID code 3ZUV) and applied a
step-well energy potential as a constraint to model a covalent
bond between C123 and the dye, and between phosphate
groups and threonine/tyrosine. We used this model as a starting
conformation for discrete molecular dynamics simulations (Dag-
on the number of contacts between the dye and pERK2, mero87
interacts with pERK2 more frequently than mero53, explaining
the different fluorescence changes of these two dyes in our
experiments (Figures 2A and 3A). Furthermore, interaction
energy histograms calculated by MedusaScore (Yin et al.,
2008), which measures the interface energy between pERK2
and pE59, revealed an additional population of low-energy con-
complex of pE59-C123m53-pERK2 (Figure 3B). We also applied
a Go-like model (Dagliyan et al., 2011) to pE59-C123m87 to
increase the sampling of dye conformations. Based on the
obtained representative conformations of pE59-C123m87, the
pERK2 activation loop is shifted toward the mero87 docking
site (Figure 3C), with the dye positioned along the substrate-
binding groove of pERK2 keeping the merocyanine ring system
near the pE59-C123m87-pERK2 interface (Figure 3D). All these
results suggest that mero87 attachment to C123 is thermody-
namically more favorable than the attachment of mero53.
Quantifying the Localized Activation of ERK2
in Live Cells
The pE59RFD biosensor was tested in NIH 3T3 mouse embryo
fibroblasts (MEFs) stably expressing YPet, a yellow fluorescent
protein derivative (Nguyen and Daugherty, 2005). The introduc-
tion of a second fluorophore enabled ratiometric imaging to
Figure 2. Characterization of RFD Bio-
Sensor responses are shown as ratiometric
fluorescence emission of measurements with
(RFDkinase) or without (RFDbg) kinase present. All
experiments were performed in triplicate; error
bars reflect the standard deviation.
(A) Ratiometric sensor response (RFDkinase:RFDbg)
ofthreeDARPin-mero conjugates when incubated
with unphosphorylated or phosphorylated ERK2
(B) Titration of the sensor variant pE59-C123m87
and pE59-ctrl with increasing concentrations of
pERK2 and ERK2. For the control sensor, the
dye was attached to the N terminus of DARPin
pE59, which is not involved in the interac-
tion with pERK2. Closed circles, sensor/pERK2;
open circles, sensor/ERK2; closed triangles,
(C) The fluorescence response of the biosensor
and nonresponsive control was tested with
the MAPKs ERK2, JNK1a1, JNK2a1, and p38a
in the unphosphorylated and phosphorylated
(D) Selectivity and specificity of the pE59-C123m87 sensor variant in lysis buffer alone and HEK 293T cell lysate (2 mg/ml total protein) (lysate). The fluorescence
response to pERK2 was competed with DARPin EpE82, which binds to the same region of pERK2 as unmodified pE59.
See also Figures S3–S5.
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
850 Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved
correct for effects of cell thickness, uneven illumination, and
other factors that could affect dye brightness (Bright et al.,
1989; Gulyani et al., 2011; Hodgson et al., 2010; Nalbant et al.,
2004). ERK activity was indicated by an increase in the dye:YPet
emission ratio. The biosensor was loaded into living fibroblasts
through microinjection or bead loading and used to map local-
ized ERK activation (Figure 4). The fluorescence ratio showed a
nearly 3-fold elevation of ERK activity in the nucleus relative to
the cytoplasm, consistent with previous studies of cells in serum
(Chen et al., 1992; Gonzalez et al., 1993).
Movies S1 and S2 show the constitutive motility of MEFs. The
enable visualization of cytoplasmic activation), activity in the
perinuclear region, and changing activation in ruffles and mem-
brane protrusions near the cell periphery.
Control studies demonstrated that the observed ERK activa-
tion was not due to spurious interactions of the dye with other
cellular constituents; the use of a control biosensor with muta-
tionsoftwocrucial contactpositions intheDARPin pE59binding
interface (DARPin-pE59 with D46A/R90A), as guided by the
pE59-pERK2 complex structure (Kummer et al., 2012), greatly
diminished activation (Figure 4). Furthermore, treatment of cells
with the ERK pathway inhibitor U0126, which inhibits the
upstreamkinase MEK1/2 (Favata et al.,1998), also reduced acti-
vation (Figure 4), consistent with our other results indicating that
Figures S4 and S5).
Surprisingly, even higher activity was observed specifically in
nucleoli. This had not been directly observed previously, empha-
sizing the utility of the biosensor. ERK has been implicated in the
induction of rRNA synthesis (Zhao et al., 2003), for which a local-
ization in the nucleolus might indeed be expected.
Because ERK activity is obtained from the ratio of sensor
signal over YPet signal, it was important to verify that YPet is
not excluded from the nucleolar regions, nor that the DARPin
sensor is enriched in the nucleolar regions. It can be seen that
the increase in brightness of the DARPin sensor in nucleoli
compared to the nuclear region is much stronger (measured as
Figure 3. Modeling of pERK2/pE59 with
(A) Distribution of the number of contacts between
pERK2 and mero53 (black) and mero87 (red) (p <
10?16). This number is defined as distances closer
than 5.5 A˚between atoms of the dye and the
DARPin throughout the trajectories.
(B) Distribution of estimated binding energies
(MedusaScore) between pERK2 and pE59 in the
presence of mero53 (black) and mero87 (red) dyes
(p < 10?16).
(C) The activation loop with (orange) and without
along with the substrate-binding groove of pERK2.
about 2.3-fold by integration of the corre-
sponding areas) than the very slight
exclusion of YFP (about 0.85-fold) (Fig-
ure S7). To further demonstrate that the
increased brightness of the sensor moni-
tors a higher concentration of pERK, and not sensor, in the
nucleoli, the background fluorescence was also monitored in
the presence of the MEK1/2 inhibitor U0126. No enrichment of
the DARPin sensor can be seen (Figure S8). The same is true
for the weak-binding DARPin mutant with mutations in the inter-
face (DARPin-pE59 with D46A/R90A) (Figure S9).
In yet another control, mCerulean was directly linked to the
DARPin biosensor (Figure S10). In this case, the ratiometric
bution of the sensor and the cell thickness control. This control
also demonstrated higher ERK activation in the nucleolus.
Because the dye fluorescence was diminished by attachment
out ratio imaging using the separately expressed YPet.
growth factors have been unsuccessful so far. Preliminary
experiments revealed dynamic activation events in the cyto-
plasm of motile cells, but these were complex and less intense
than the events in the nucleus.
It is possible that the DARPin inhibits nuclear translocation of
ERK. Briefly, ERK lacks classical nuclear localization signal
sequences (reviewed in Roskoski, 2012). Tyr and Thr phosphor-
ylation in the ERK phosphorylation loop is believed to lead to
dissociation from scaffolding proteins, which hold ERK in the
cytoplasm. An energy-independent direct transport through
interaction with nuclear pore complex proteins appears to
involve a region of ERK partially overlapping with the DARPin
binding site. Another energy-dependent mechanism engaging
residues in the ERK kinase insertion domain, which dock to im-
portin a-7, may also be hindered by DARPin binding.
In vitro kinase assays revealed that pE59-wt sterically inhibits
manner (unpublished data). However, possible effects of RFD
biosensors on ERK signaling in living cells were minimized by
using the lowest possible biosensor concentrations in vivo. The
ing with a fast dissociation phase (Figure S5), favoring detection
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 851
rather than inhibition of pERK2 activity. Thus, the design of a
DARPin-based biosensor contrasts with that of an inhibitor, for
which a high-affinity binding is most effective (Parizek et al.,
2012), and DARPins with higher affinity for ERK have been
obtained as well (Kummer et al., 2012) but were not chosen for
the present study. Nonetheless, we cannot exclude that nuclear
localization is inhibited under these conditions, despite having
used a DARPin of low affinity and at low concentrations.
TheDARPinsensor clearly reacts to the activated kinase in the
nucleus. However, the fact that for ERK, unlike for most kinases,
nuclear transport is also controlled by residues in the kinase
domain makes it more challenging to construct an ERK sensor
that interferes with neither activity nor transport. Nonetheless,
this work clearly indicates the ability of the biosensor to report
dynamics and localization of ERK activation.
Fluorescent biosensors can provide new insights into localized
activation of proteins in living cells (Welch et al., 2011). Our
design of an ERK activity reporter differs significantly from previ-
ously reported approaches (Fujioka et al., 2006; Green and
Alberola-Ila, 2005; Harvey et al., 2008; Sato et al., 2007). We
directly sense the occurrence and localization of endogenous
activated ERK by targeting unique features of the active kinase
rescence resonance energy transfer (FRET)-based kinase sen-
sors rely on fluorescence readouts from kinase substrates,
whichareprone tomodification byother cellularenzymes,which
thuslead themto report the action of both kinases and phospha-
tases. Furthermore, activated sensors are subject to diffusion,
independent of the actual enzyme target to be monitored. Due
to the significantly lower brightness of FRET signals as
compared to directly excited dyes, protein dynamics at certain
cell positions remain elusive when examined with FRET-based
reporters. As an example, Src family kinase activity in protrusion
and retraction dynamics was not reported until examined with a
dye-labeled monobody biosensor (Gulyani et al., 2011), a design
approach similarly aiming for binding to the active enzyme, as in
the DARPin-based ERK activity sensor presented here. In the
case of ERK, however, the conformational changes upon activa-
tion are very subtle, requiring subtractive panning to generate
ERK- and pERK-specific binders (Kummer et al., 2012), whose
specificity was also studied by X-ray crystallography.
The development of dye-based biosensors derived from natu-
mization of fluorophore attachment for each individual reagent in
proposed recently, biosensors based on affinity reagents, which
canbetailored to virtuallyanytargetproteinofinterest,mayoffer
the opportunity to simplify biosensor development by targeting
primarily residues near the variable regions for attachment of
environmentally sensitive dyes (Brient-Litzler et al., 2010;
Gulyani et al., 2011; Miranda et al., 2011). In this regard, the
development of binding scaffolds (e.g., DARPins) devoid of cys-
teines (Binz et al., 2005; Boersma and Plu ¨ckthun, 2011), thus
facilitating site-specific dye attachment, potentially paves the
way for a generally applicable approach to biosensor design. If
structural data of the complex with the target antigen are avail-
able (Brient-Litzler et al., 2010; Kummer et al., 2012), one can
easily identify suitable positions for dye conjugation in selected
DARPin binders. Yet, even in the absence of any structural infor-
mation about the interaction interface, the conserved binding
mode of designed binding scaffolds enables sensor develop-
ment. As in the case of DARPins, the target binding site is
composed of a number of randomized residues embedded in a
constant polypeptide backbone, which guarantees conserved
biophysical properties (Binz et al., 2005). In most cases, only a
subset of the randomized residues is involved in binding to the
target protein. Consequently, suitable positions for dye attach-
ment are either variable positions not involved in binding as
determined by site-directed mutagenesis, but located in the
neighborhood of the antigen binding site, or are conserved
exposed positions adjacent to putative interaction residues
(Brient-Litzler et al., 2010; Gulyani et al., 2011; Kummer et al.,
2012; Miranda et al., 2011).
In our study, we based the design of an ERK activity sensor on
the previously selected DARPin binder pE59 that targets doubly
phosphorylated ERK2 selectively both in vitro and in cellular
Figure 4. Quantifying Activation of Endogenous ERK in Living Cells
(A) Ratio images of untreated (left) and U0126-pretreated (right) 3T3 mouse
embryonic fibroblasts. The scale bar represents 20 mm.
(B) Comparison of the average ratios of the biosensor in2% fetal bovine serum
(blue, 13 cells), pretreated with 10 mM U0126 (red, 46 cells), and the control
DARPin in 2% fetal bovine serum (green, 60 cells). Error bars reflect the
standard deviation of the average biosensor ratios for the different experi-
See also Figures S6–S10.
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
852 Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved
assays (Kummer et al., 2012) and discriminates against the non-
phosphorylated form, which is selectively recognized by DARPin
binder E40 (Kummer et al., 2012). The determined structure of
the pE59-pERK2 complex facilitated the search for suitable
positions to attach solvent-sensitive merocyanine dyes as
reporter fluorophores for pERK2 binding. Identified residues for
dye coupling were located exclusively at nonvariable framework
positions. Remarkably, the replacement of given amino acids
by cysteine did not change the conformational integrity of the
DARPin scaffold, even though some of them were originally gly-
cines and a destabilizing effect was predicted. This underlines
the general robustness of the DARPin framework to carry a
mutational load outside the randomized positions without any
need for later stability engineering. Therefore, DARPins appear
because the stability of typical library members would allow the
incorporation of dyes at a variety of positions, thereby increasing
the chances to generate biosensors with desired properties.
Our final ERK activity sensor, pE59RFD, was highly selective
for pERK2 in fluorescence assays over ERK2 (Figure 2). No
response for other MAPK family members was observed either,
highlighting that initial selections of specific DARPin binders can
truly yield functional biosensors with the required selectivity and
sensitivity. Because of therather smallstructural differencesthat
need to be discriminated, it is not surprising that fluorescence
properties and binding specificity were dependent on both the
positioning and the type of dye used for labeling. As shown
exemplarily for DARPin position 123, only conjugation with
mero87 resulted in preserved binding behavior and sufficient
by structural modeling (Figure 3). The dye itself thus appears to
introduce another element of selectivity.
Importantly for cellular applications, both a high selectivity of
binding and a bright fluorescence response are desirable. It is
required that nonphosphorylated ERK does not compete with
pERK, nor that ERK is sensed. Conversely, affinity to pERK
must not be so tight that the sensor acts as a potent inhibitor,
and the sensor should not be present in large stoichiometric
excess. Although the original DARPin pE59 sterically blocks
docking and consequently phosphorylation of F site-containing
pERK2 substrates (unpublished data), we used this DARPin spe-
cifically because of its moderate affinity combined with high
discrimination power. The affinity of pE59RFD for pERK2 (Fig-
ibility and specificity (Gulyani et al., 2011; Kraynov et al., 2000;
Nalbant et al., 2004; Pertz et al., 2006). In this respect, the
4-fold further decreased affinity of pE59RFD to pERK2, but with
retained high discrimination against nonphosphorylated ERK2,
in combination with the low sensor concentrations applied here,
in living cells. Nevertheless, it is important to note that essentially
all biosensors, including FRET-based sensors, perturb cell phys-
iology to some extent, thus requiring low sensor concentrations
and/or controlled experimental conditions to reach meaningful
biological conclusions (Gulyani et al., 2011; Harvey et al., 2008).
We used pE59RFD to study ERK activity in living cells (Fig-
ure 4), revealing localized activation in the nucleus. This was
consistent with previous studies, suggesting that natural ERK
signaling functions are not perturbed by pE59RFD: ERK is acti-
vated in the nucleus in response to mitogens in serum, with acti-
vation persistent as long as the mitogenic stimulus is present
(Lenormand et al., 1993). Importantly, the potent inhibitor
U0126 of the upstream kinase MEK1/2 (Favata et al., 1998) sup-
pressed activation almost completely (Figure 4). The DARPin-
based biosensor revealed ERK activation in the nucleolus, an
effect not previously directly observed, but consistent with the
involvement of ERK in the induction of rRNA synthesis (Zhao
et al., 2003).
We have demonstrated that previously determined design
principles (Brient-Litzler et al., 2010; Gulyani et al., 2011)
are applicable to produce an ERK activity biosensor based
on the designed ankyrin repeat protein (DARPin) framework.
In the present case, only the altered conformation of the
active kinase (Canagarajah et al., 1997; Kummer et al.,
2012) is being detected, pointing to a concept applicable to
changes in the cell. The conserved binding mode and
conformational integrity of alternative binding scaffolds,
such as DARPins, potentially simplify sensor design by tak-
ing advantage of a generalizable workflow to streamline
biosensor production. Combined with dyes of exceptional
sensitivity, biosensors may prove to become useful chemi-
cal tools in studying subtle changes of protein dynamics in
living cells. Due to their favorable biophysical properties
and stability, DARPin-based biosensors may also be useful
for recent advances in point-of-care immunoassays based
on microfluidics (Gervais and Delamarche, 2009) through
implementation of both selective capture and sensitive
detection of target antigens in one step. Thus, the tech-
nology based on long-wavelength solvatochromatic dyes
together with DARPins with high discrimination power
between conformers holds promise both for in vitro and
in vivo applications.
The crystal structure of the complex between DARPin pE59 and pERK2 (PDB
ID code 3ZUV) (Kummer et al., 2012) was used to calculate solvent-accessible
surface areas with the Proteins, Interfaces, Surfaces and Assemblies (PISA)
server at the European Bioinformatics Institute (http://www.ebi.ac.uk/pdbe/
prot_int/pistart.html) (Krissinel and Henrick, 2007). Contact residues were
identified with PISA and the CCP4 software suite (Winn et al., 2011).DG values
for the folding of DARPin pE59 and its cysteine variants were calculated with
ROSETTA (Leaver-Fay et al., 2011).
Experiments were performed according to standard protocols (Sambrook and
Russell, 2001). Enzymes and buffers were purchased from New England
BioLabs or Fermentas. Oligonucleotides were obtained from Microsynth.
Site-Directed Mutagenesis and Cloning
Changesof residues Gly91, Gly113, Asn123, and Gly124 to cysteine as well as
Asp46 and Arg90 to alanine in pE59 were made by mutagenesis of plasmid
pDST67_pE59 (pE59-wt) (Kummer et al., 2012) using the QuikChange II Site-
Directed Mutagenesis Kit (Stratagene). Mutagenic primers (Table S1) were
designed as proposed (Brient-Litzler et al., 2010) to avoid primer annealing
on nontargeted DNA segments due to the repetitive nature of the DARPin
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 853
nucleotide sequence. Mutated DARPin genes were cloned into pDST67 via
the BamHI/HindIII sites (Steiner et al., 2008), yielding pDST67_pE59-C91,
_pE59-C113, _pE59-C123, _pE59-C124, and _pE59-D46A_R90A (control
Foranonresponsive control, weincorporatedacysteine attheNterminusof
pE59, a region distant from the interaction interface, to avoid any solvatochro-
matic effect of attached dyes upon ERK binding. A two-residue insertion,
comprising glycine and cysteine, was cloned into pDST67 via EcoRI/BamHI
(oligonucleotides pDST67Cf and pDST67r; Table S1). The two residues were
inserted behind the N-terminal MRGSH6tag and directly preceding the BamHI
site, which constitutes the 50site for DARPin cloning. The modified vector was
named pDST67C. The DARPin pE59 gene was inserted into pDST67C via the
BamHI/HindIII sites, yielding pDST67C_pE59 (pE59-ctrl).
Protein Production, Purification, and Characterization
Biotinylated ERK2, pERK2, JNK1a1, pJNK1a1, JNK2a1, pJNK2a1, p38a, and
pp38a were produced and purified as outlined previously (Kummer et al.,
2012). DARPin pE59-wt and its mutant derivatives, all cloned into pDST67,
were expressed and purified as described (Dokholyan et al., 1998). DARPin
using their hexahistidine tag by IMAC. For cysteine-containing DARPins,
buffers were supplemented with 5 mM b-mercaptoethanol throughout the
purification process to prevent formation of intermolecular disulfide bonds.
IMAC fractions were analyzed by SDS-PAGE. Pure fractions were pooled
and stored in sodium phosphate buffer (pH 7.4) supplemented with 150 mM
NaCl, 1 mM EDTA, 10% (v/v) glycerol, and 2 mM DTT. Analytical gel filtration
was performed in PBS with 2 mM DTT as described (Binz et al., 2003) using
purified DARPin variants at 15 mM protein concentration to verify the mono-
meric state of cysteine-containing DARPins. Binding of DARPin variants to
kinases was analyzed by ELISA.
The merocyanine dye mero53 has been described previously (Toutchkine
et al., 2007b), whereas the dyes mero60, 61, 62, and 87 have been published
recently (MacNevin et al., 2013), and mero199 will be described elsewhere
(C.M., A. Toutchkine, C.-W. Hsu, L. Li, T.T. Davis, and K.M.H., unpublished
ing buffer (sodium phosphate buffer [pH 7.5], 150 mM NaCl, 1 mM EDTA) prior
to labeling with dyes using size-exclusion G-25 (GE Healthcare) columns.
Thiol-reactive merocyanine dyes (10–20 mM in DMSO) were added to 200 ml
DARPins (100–200 mM) in 10-fold excess with less than 10% DMSO in the final
reaction mixture. After incubation for 3 hr at room temperature, the reaction
was stopped by addition of 1 ml b-mercaptoethanol. The DARPin conjugates
were separated from unreacted fluorophore by G-25 size-exclusion chroma-
tography. Labeled proteins were subjected to SDS-PAGE and single fluores-
cent bands were observed with the molecular weights corresponding to DAR-
Pins. Control samples of free dye were visible at lower molecular weight than
protein. Labeling efficiencies were calculated by measuring protein and dye
concentrations in the purified conjugates. Dye concentrations were estimated
using absorbancespectroscopyof the dye at theirrespective absorption max-
ima after dissolving the labeled conjugates in DMSO. Protein concentrations
were determined with the Pierce 660 nm protein assay (Thermo Scientific) us-
ing known concentrations of DARPin pE59 as standards. Binding of labeled
conjugates to kinases was analyzed by fluorescence measurements and
ELISA experiments were performed in PBST (PBS [pH 7.4], 0.05% Tween 20)
supplemented with 1 mM DTT. Incubation with detection antibodies was
performed in PBST without DTT. Biotinylated ERK2 or biotinylated pERK2
(50 nM each) was immobilized in neutravidin-coated microtiter plates blocked
with BSA. Target immobilization and subsequent ELISA steps were per-
formed at 4?C. Purified DARPins or labeled DARPin conjugates (100 nM)
were applied to wells with or without immobilized ERK2 antigen for 1 hr.
DARPin binding was detected by an anti-RGS-His antibody (1:5,000;
QIAGEN), followed by a conjugate of an anti-mouse antibody with alkaline
phosphatase (1:10,000; Thermo Scientific). As substrate, p-nitrophenylphos-
phate was used (Sigma).
Dye-labeled DARPins (600 nM) in assay buffer (PBS [pH 7.4], 1 mM DTT) were
mixed 1:1 (v/v) with target pre-equilibrated in assay buffer (1 mM if not other-
wise indicated). Fluorescence spectra were acquired at 25?C using an Infinite
M1000 Pro plate reader (Tecan). Excitation was at 595 nm (mero53), 565 nm
(mero60), and 585 nm (mero87) for emission spectra. Emission was at
630 nm (mero53), 595 nm (mero60), and 615 nm (mero87) for excitation
spectra. One-point measurements were performed with the listed wavelength
parameters using a bandwidth of 5 nm for both excitation and emission. Titra-
tion of the fluorescence response of the sensor variant pE59-C123m87 and
pE59-ctrl was carried out with varying concentrations of ERK2 and pERK2
(final assay concentrations 0.025, 0.05, 0.1, 0.2, 0.4, 0.6, 1.0, 1.5, 3.0 mM).
For assaying the pE59-C123m87 response in cell lysate, HEK 293T cells
were lysed in cold lysis buffer (20 mM Tris [pH 7.7], 150 mM NaCl, 0.1 mM
inhibitor cocktail EDTA-free [Roche], 1% Triton X-100). The lysate was cleared
by centrifugation at 14,000 rpm for 15 min at 4?C. Total protein concentration
of the cleared lysate was assessed with the BCA protein assay reagent
(Thermo Scientific) using BSA as the standard. For fluorescence measure-
ments, sensor variant pE59-C123m87, ERK target (ERK2 or pERK2), and
EpE82 were pre-equilibrated with lysis buffer. The sensor DARPin, ERK anti-
gen, and lysate with or without EpE82 were mixed. For comparison, the
same experiment was carried out in lysis buffer without cell lysate. Further-
more, the sensor response in lysate without exogenously added ERK2 was
determined. The final protein concentrations in the described lysate assay
setups were 300 nM for pE59-C123m87, 300 nM for ERK2, 6 mM for EpE82,
and 2 mg/ml (total protein concentration) for the cell lysate.
Surface Plasmon Resonance
SPR was measured using a Biacore 2000 instrument (GE Healthcare). The
running buffer was 50 mM Tris (pH 7.4), 150 mM NaCl, 0.05 mM EDTA, and
0.005% Tween 20. Biotinylated ERK2 or pERK2 was immobilized on a strep-
tavidin SA chip (GE Healthcare). Kinetic interaction was measured at a flow
rate of 30 ml/min with injection of varying pE59RFD (pE59-C123m87) con-
centrations followed by an off-rate measurement. The sensorgrams were
corrected by subtracting the signal of an uncoated reference cell. Zero-con-
centration samples (blanks) were included for ‘‘double-referencing.’’ The
kinetic data were evaluated by fitting the equilibrium binding responses with
Scrubber2 (BioLogic Software).
Modeling and Molecular Dynamics of the ERK2-pE59RFD Complex
We mutated Asn123 of pE59 to cysteine using the pE59-pERK2 crystal struc-
a step-well potential that mimics a covalent bond (Ding et al., 2008). A similar
step-well potential was applied to the phosphate groups on tyrosine and thre-
onine in pERK2. For the investigation of the dynamics of the binding of mero53
and mero87 at the pE59-pERK2 interface, we used discrete molecular
et al., 1998) that has been used in different biological applications (Dagliyan
et al., 2011; Ding et al., 2010). After the minimization of the system with har-
monic constraints at high temperature (0.7 kcal [mol kB]?1) to eliminate steric
clashes, the systems were simulated at a temperature of 0.5 kcal (mol kB)?1
for 20 ns. Ten different simulation trajectories (with different initial velocities)
for each system were combined and the probability density was calculated
based on the number of contacts and MedusaScore (Yin et al., 2008), which
bonding, and electrostatic interactions. To measure the significance of two
distributions, we calculated p values using the Kolmogorov-Smirnov test. In
addition, a simulation with a Go-like model (Dagliyan et al., 2011) -constrained
pE59-C123m87 (pE59RFD), where native contacts were kept with another
step-well potential to increase the sampling of dye conformations, was
performed, and the minimum-energy conformation is represented.
NIH 3T3 MEFs stably expressing YPet, a yellow fluorescent protein variant
modified Eagle’s medium (Cellgro) supplemented with 10% fetal calf serum,
2 mM GlutaMAX (GIBCO, Life Technologies), 125 mg/l amphotericin
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
854 Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved
B(Sigma-Aldrich),and 100mg/mlpenicillin/streptomycin (Cellgro).Forimaging
experiments, MEFs were plated onto coverslips coated with 5 mg/ml fibro-
nectin (Sigma) overnight. Culturemedium wasexchanged for imaging medium
(Ham’s F-12K medium without phenol red [SAFC Biosciences] with 2% fetal
bovine serum, 15 mM HEPES, 2 mM GlutaMAX, and 125 mg/l penicillin/strep-
tomycin). The DARPin-based biosensor and the control biosensor were trans-
ferred into cells using a bead-loading approach (McNeil and Warder, 1987) or
microinjection (Gulyani et al., 2011). Cells were washed twice in PBS and then
to imaging. For inhibition of ERK signaling, cells were treated with 10 mM
Imaging was performed on an Olympus IX81 inverted microscope equipped
with ZDC focus drift compensator, a cooled digital 14-bit CCD camera (Cool-
snap ES-2; Roper Scientific), a 100 W mercury arc lamp, and MetaMorph
imaging software (Molecular Devices). Images were acquired using a
403 UPlanFLN 1.3 NA oil immersion objective and processed as described
previously (Hodgson et al., 2006). A multiband dichroic mirror (440/505/585,
89006; Chroma) was used with band-pass filters for YPet (ET500/20X and
ET535/30M; Chroma), mero87 (HQ572/35X and ET632/60M; Chroma), or
mCerulean (ET430/24X and ET470/24M). The average intensities of each
image in the cytoplasm, nucleus, or nucleolus were measured using the region
statistics function of MetaMorph with manual threshholding. Background,
shading correction, and ratioimagingwere carried outas previously described
(Hodgson et al., 2006). ERK activity was monitored as the ratio of dye:YPet
fluorescence (Figure S6). Using intensity thresholding operations (applying a
‘‘mask’’), the borders of the nucleus and cell edge were identified (Hodgson
et al., 2006, 2008).
For comparison of cells, scaling was normalized to be linear between 5%
and 95% of maximal activation values.
Supplemental Information includes ten figures, one table, and two movies and
can be found with this article online at http://dx.doi.org/10.1016/j.chembiol.
We thank A. Honegger for calculations of Gibbs free energies of protein
variants. L.K. was supported by a doctoral fellowship of the Ernst Schering
Foundation. We gratefully acknowledge funding from Schweizerische Natio-
nalfonds grant 3100A0B-128671, the Swiss National Center of Competence
in Research in Structural Biology, the PhosphoNetX Project in SystemsX (all
to A.P.), the European Union FP7 Collaborative Project AffinityProteome (con-
tract 222635) (to A.P., M.K., and B.Z.), and grant GM090317 from the National
Institutes of Health (to K.M.H.). A.P. is a founder and shareholder of Molecular
Partners, AG, which commercializes the DARPin technology.
Received: November 13, 2012
Revised: April 11, 2013
Accepted: April 23, 2013
Published: June 20, 2013
Binz, H.K., Stumpp, M.T., Forrer, P., Amstutz, P., and Plu ¨ckthun, A. (2003).
Designing repeat proteins: well-expressed, soluble and stable proteins from
combinatorial libraries of consensus ankyrin repeat proteins. J. Mol. Biol.
Binz, H.K., Amstutz, P., Kohl, A., Stumpp, M.T., Briand, C., Forrer, P., Gru ¨tter,
M.G., and Plu ¨ckthun, A. (2004). High-affinity binders selected from designed
ankyrin repeat protein libraries. Nat. Biotechnol. 22, 575–582.
Binz, H.K., Amstutz, P., and Plu ¨ckthun, A. (2005). Engineering novel binding
proteins from nonimmunoglobulin domains. Nat. Biotechnol. 23, 1257–1268.
Boersma, Y.L., and Plu ¨ckthun, A. (2011). DARPins and other repeat protein
scaffolds: advances in engineering and applications. Curr. Opin. Biotechnol.
Brient-Litzler, E., Plu ¨ckthun, A., and Bedouelle, H. (2010). Knowledge-based
design of reagentless fluorescent biosensors from a designed ankyrin repeat
protein. Protein Eng. Des. Sel. 23, 229–241.
ratio imaging microscopy. Methods Cell Biol. 30, 157–192.
Canagarajah, B.J., Khokhlatchev, A., Cobb, M.H., and Goldsmith, E.J. (1997).
Activation mechanism of the MAP kinase ERK2 by dual phosphorylation. Cell
Chen, R.H., Sarnecki, C., and Blenis, J. (1992). Nuclear localization and regu-
lation of erk- and rsk-encoded protein kinases. Mol. Cell. Biol. 12, 915–927.
Chen, Z., Gibson, T.B., Robinson, F., Silvestro, L., Pearson, G., Xu, B., Wright,
A., Vanderbilt, C., and Cobb, M.H. (2001). MAP kinases. Chem. Rev. 101,
Cowley, S., Paterson, H., Kemp, P., and Marshall, C.J. (1994). Activation of
MAP kinase kinase is necessary and sufficient for PC12 differentiation and
for transformation of NIH 3T3 cells. Cell 77, 841–852.
Structural and dynamic determinants of protein-peptide recognition. Structure
Ding, F., Tsao, D., Nie, H., and Dokholyan, N.V. (2008). Ab initio folding of
proteins with all-atom discrete molecular dynamics. Structure 16, 1010–1018.
Ding, F., Yin, S., and Dokholyan, N.V. (2010). Rapid flexible docking using a
stochastic rotamer library of ligands. J. Chem. Inf. Model. 50, 1623–1632.
Dokholyan, N.V., Buldyrev, S.V., Stanley, H.E., and Shakhnovich, E.I. (1998).
Discrete molecular dynamics studies of the folding of a protein-like model.
Fold. Des. 3, 577–587.
Favata, M.F., Horiuchi, K.Y., Manos, E.J., Daulerio, A.J., Stradley, D.A.,
Feeser, W.S., Van Dyk, D.E., Pitts, W.J., Earl, R.A., Hobbs, F., et al. (1998).
Identification of a novel inhibitor of mitogen-activated protein kinase kinase.
J. Biol. Chem. 273, 18623–18632.
Fujioka, A., Terai, K., Itoh, R.E., Aoki, K., Nakamura, T., Kuroda, S., Nishida, E.,
and Matsuda, M. (2006). Dynamics of the Ras/ERK MAPK cascade as moni-
tored by fluorescent probes. J. Biol. Chem. 281, 8917–8926.
Gervais, L., and Delamarche, E. (2009). Toward one-step point-of-care immu-
nodiagnostics using capillary-driven microfluidics and PDMS substrates. Lab
Chip 9, 3330–3337.
Gonzalez,F.A.,Seth,A.,Raden,D.L.,Bowman, D.S.,Fay,F.S., andDavis,R.J.
(1993). Serum-induced translocation of mitogen-activated protein kinase to
the cell surface ruffling membrane and the nucleus. J. Cell Biol. 122, 1089–
Green, H.M., and Alberola-Ila, J. (2005). Development of ERK activity sensor,
an in vitro, FRET-based sensor of extracellular regulated kinase activity. BMC
Chem. Biol. 5, 1.
Gulyani, A., Vitriol, E., Allen, R., Wu, J., Gremyachinskiy, D., Lewis, S., Dewar,
B., Graves, L.M., Kay, B.K., Kuhlman, B., et al. (2011). A biosensor generated
via high-throughput screening quantifies cell edge Src dynamics. Nat. Chem.
Biol. 7, 437–444.
Harvey, C.D., Ehrhardt, A.G., Cellurale, C., Zhong, H., Yasuda, R., Davis, R.J.,
and Svoboda, K. (2008). A genetically encoded fluorescent sensor of ERK
activity. Proc. Natl. Acad. Sci. USA 105, 19264–19269.
Hodgson, L., Nalbant, P., Shen, F., and Hahn, K. (2006). Imaging and photo-
bleach correction of Mero-CBD, sensor of endogenous Cdc42 activation.
Methods Enzymol. 406, 140–156.
Hodgson, L., Pertz, O., and Hahn, K.M. (2008). Design and optimization of
genetically encoded fluorescent biosensors: GTPase biosensors. Methods
Cell Biol. 85, 63–81.
Hodgson, L., Shen, F., and Hahn, K. (2010). Biosensors for characterizing the
dynamics of Rho family GTPases in living cells. Curr. Protoc. Cell Biol. 46,
Kawe, M., Forrer, P., Amstutz, P., and Plu ¨ckthun, A. (2006). Isolation of intra-
cellular proteinase inhibitors derived from designed ankyrin repeat proteins
by genetic screening. J. Biol. Chem. 281, 40252–40263.
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 855
Kraynov, V.S., Chamberlain, C., Bokoch, G.M., Schwartz, M.A., Slabaugh, S., Download full-text
and Hahn, K.M. (2000). Localized Rac activation dynamics visualized in living
cells. Science 290, 333–337.
Krissinel, E., and Henrick, K. (2007). Inference of macromolecular assemblies
from crystalline state. J. Mol. Biol. 372, 774–797.
Kummer, L., Parizek, P., Rube, P., Millgramm, B., Prinz, A., Mittl, P.R.,
Kaufholz, M., Zimmermann, B., Herberg, F.W., and Plu ¨ckthun, A. (2012).
Structural and functional analysis of phosphorylation-specific binders of the
kinase ERK from designed ankyrin repeat protein libraries. Proc. Natl. Acad.
Sci. USA 109, E2248–E2257.
Leaver-Fay, A., Tyka, M., Lewis, S.M., Lange, O.F., Thompson, J., Jacak, R.,
Kaufman, K., Renfrew, P.D., Smith, C.A., Sheffler, W., et al. (2011).
ROSETTA3: an object-oriented software suite for the simulation and design
of macromolecules. Methods Enzymol. 487, 545–574.
Lenormand, P., Sardet, C., Page `s, G., L’Allemain, G., Brunet, A., and
Pouysse ´gur, J. (1993). Growth factors induce nuclear translocation of MAP
kinases (p42mapk and p44mapk) but not of their activator MAP kinase kinase
(p45mapkk) in fibroblasts. J. Cell Biol. 122, 1079–1088.
MacNevin, C.J., Gremyachinskiy, D., Hsu, C.W., Li, L., Rougie, M., Davis, T.T.,
and Hahn, K.M. (2013). Environment-sensing merocyanine dyes for live cell
imaging applications. Bioconjug. Chem. 24, 215–223.
Marshall, C.J. (1995). Specificity of receptor tyrosine kinase signaling: tran-
sient versus sustained extracellular signal-regulated kinase activation. Cell
McNeil, P.L., and Warder, E. (1987). Glass beads load macromolecules into
living cells. J. Cell Sci. 88, 669–678.
Miranda, F.F., Brient-Litzler, E., Zidane, N., Pecorari, F., and Bedouelle, H.
(2011). Reagentless fluorescent biosensors from artificial families of antigen
binding proteins. Biosens. Bioelectron. 26, 4184–4190.
Mosavi, L.K., Cammett, T.J., Desrosiers, D.C., and Peng, Z.Y. (2004). The
ankyrin repeat as molecular architecture for protein recognition. Protein Sci.
Nalbant, P., Hodgson, L., Kraynov, V., Toutchkine, A., and Hahn, K.M. (2004).
Activation of endogenous Cdc42 visualized in living cells. Science 305, 1615–
Nguyen, A.W., and Daugherty, P.S. (2005). Evolutionary optimization of fluo-
rescent proteins for intracellular FRET. Nat. Biotechnol. 23, 355–360.
Parizek, P., Kummer, L., Rube, P., Prinz, A., Herberg, F.W., and Plu ¨ckthun, A.
(2012). Designed ankyrin repeat proteins (DARPins) as novel isoform-specific
intracellular inhibitors of c-Jun N-terminal kinases. ACS Chem. Biol. 7, 1356–
K., and Cobb, M.H. (2001). Mitogen-activated protein (MAP) kinase pathways:
regulation and physiological functions. Endocr. Rev. 22, 153–183.
Pertz, O., Hodgson, L., Klemke, R.L., and Hahn, K.M. (2006). Spatiotemporal
dynamics of RhoA activity in migrating cells. Nature 440, 1069–1072.
Roskoski, R., Jr. (2012). ERK1/2 MAP kinases: structure, function, and regula-
tion. Pharmacol. Res. 66, 105–143.
Sambrook, J., and Russell, D.W. (2001). Molecular Cloning: A Laboratory
Manual (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press).
Sato, M., Kawai, Y.,and Umezawa, Y. (2007). Genetically encoded fluorescent
indicators to visualize protein phosphorylation by extracellular signal-regu-
lated kinase in single living cells. Anal. Chem. 79, 2570–2575.
Steiner, D., Forrer, P., and Plu ¨ckthun, A. (2008). Efficient selection of DARPins
with sub-nanomolar affinities using SRP phage display. J. Mol. Biol. 382,
Toutchkine, A., Kraynov, V., and Hahn, K. (2003). Solvent-sensitive dyes to
report protein conformational changes in living cells. J. Am. Chem. Soc. 125,
Toutchkine, A., Nguyen, D.V., and Hahn, K.M. (2007a). Merocyanine dyes with
improved photostability. Org. Lett. 9, 2775–2777.
Toutchkine, A., Nguyen, D.V., and Hahn, K.M. (2007b). Simple one-pot prep-
aration of water-soluble, cysteine-reactive cyanine and merocyanine dyes
for biological imaging. Bioconjug. Chem. 18, 1344–1348.
Welch, C.M., Elliott, H., Danuser, G., and Hahn, K.M. (2011). Imaging the coor-
dination of multiple signalling activities in living cells. Nat. Rev. Mol. Cell Biol.
Winn, M.D., Ballard, C.C., Cowtan, K.D., Dodson, E.J., Emsley, P., Evans,
P.R., Keegan, R.M., Krissinel, E.B., Leslie, A.G., McCoy, A., et al. (2011).
Overview of the CCP4 suite and current developments. Acta Crystallogr. D
Biol. Crystallogr. 67, 235–242.
Yin, S., Biedermannova, L., Vondrasek, J., and Dokholyan, N.V. (2008).
MedusaScore: an accurate force field-based scoring function for virtual drug
screening. J. Chem. Inf. Model. 48, 1656–1662.
Zahnd, C., Pecorari, F., Straumann, N., Wyler, E., and Plu ¨ckthun, A. (2006).
Selection and characterization of Her2 binding-designed ankyrin repeat
proteins. J. Biol. Chem. 281, 35167–35175.
Zhao, J., Yuan, X., Fro ¨din, M., and Grummt, I. (2003). ERK-dependent phos-
phorylation of the transcription initiation factor TIF-IA is required for RNA
polymerase I transcription and cell growth. Mol. Cell 11, 405–413.
Chemistry & Biology
DARPin-Based Biosensor for Monitoring ERK Activity
856 Chemistry & Biology 20, 847–856, June 20, 2013 ª2013 Elsevier Ltd All rights reserved