Covalent Lipid Pocket Ligands Targeting p38α MAPK Mutants

Article abstract of DOI:10.1002/anie.201706345

A chemical genetic approach is presented to covalently target a unique lipid binding pocket in the protein kinase p38α, whose function is not yet known. Based on a series of cocrystal structures, a library of 2-arylquinazolines that were decorated with electrophiles were designed and synthesized to covalently target tailored p38α mutants containing artificially introduced cysteine residues. Matching protein–ligand pairs were identified by MS analysis and further validated by MS/MS studies and protein crystallography. The covalent ligands that emerged from this approach showed excellent selectivity towards a single p38α mutant and will be applicable as suitable probes in future studies of biological systems to dissect the function of the lipid pocket by means of pharmacological perturbations.

Full text of DOI:10.1002/anie.201706345

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
International Edition Chemie
www.angewandte.org
Accepted Article
Title: Covalent Lipid Pocket Ligands Targeting p38α MAPK Mutants
Authors: Daniel Rauh, Mike Bührmann, Julia Hardick, Jörn Weisner,
and Lena Quambusch
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.201706345
Angew. Chem. 10.1002/ange.201706345
Link to VoR: http://dx.doi.org/10.1002/anie.201706345
http://dx.doi.org/10.1002/ange.201706345

Angewandte Chemie International Edition
10.1002/anie.201706345
COMMUNICATION
Covalent Lipid Pocket Ligands Targeting p38α MAPK Mutants
Mike Bührmann, Julia Hardick, Jörn Weisner, Lena Quambusch, Daniel Rauh*[a]
Abstract: We present a chemical genetic approach to covalently
target a unique lipid binding pocket (LP) in the protein kinase p38α.
To date, the function of the LP is not known. Based on a series of
co-crystal structures, we designed and synthesized a library of
probes are utilized in these approaches to dissect biological
processes.[
14,15,16]
For example, Shokat and co-workers used the
bump-and-hole approach to highlight the application of molecular
probes in conjugation with individually tailored kinases.[
Engineered proteins with functionally silent mutations introduced
close to a ligand binding site, e.g., the ATP binding pocket in
kinases, have led to artificial mutant-selective ligand sensitivity
and affinity, thus allowing biochemical characterization of the
interaction and affording drug target validation. Here, we describe
the enhancement of recently reported lipid pocket ligands
(LiPoLis) of p38α MAPK as putative probe molecules for the
poorly understood biological function of this kinase’s lipophilic
binding site by means of a chemical genetic approach (Bührmann
et al., submitted). For these studies, we designed mutant p38α
variants carrying cysteine residues to serve as anchor points to
form covalent bonds with the LiPoLis. Characterization by MS and
tandem mass spectrometry (MS/MS) analysis as well as co-
crystallization experiments revealed specific protein-ligand pairs.
17,18]
2-arylquinazolines decorated with electrophiles to covalently target
tailored p38α mutants containing artificially introduced cysteines.
Matching protein-ligand pairs were identified by MS analysis and
further validated by MS/MS studies and protein crystallography. The
covalent ligands that derived from this approach showed excellent
selectivity towards a single p38α mutant each and will be applicable
for future studies as suitable probes in biological systems to dissect
the function of the lipid pocket by means of pharmacological
perturbations.
Over the last two decades, protein kinases have emerged as
a key research topic in medicinal chemistry and chemical
biology.[1] Taking into account the role of protein kinases as the
central elements in signal transduction networks, the enzymatic
phosphotransfer reaction and its modulation have drawn
The p38α lipid pocket is located in the C-terminal sub-domain
of the kinase and several angstroms distant from the enzyme’s
[
1]
particular attention. Dysregulation of cellular signaling pathways
by kinase malfunction can lead directly to kinase-related diseases, active site. In the dormant state, the LP is present in a closed
explaining the clinical relevance and need for specific inhibitor
development and optimization.[2,3] Yet, the function of protein
kinases in the physiological context is complex and defined by far
more than only the catalytic transfer of phosphate groups. So-
called scaffolding functions are characterized by the interplay with
other molecular partners in the form of protein-target interactions
or the assembly of multi enzyme complexes.[4] Thus, kinases can
affect designated substrates with respect to their localization or
regulation, instead by alteration of their activity state by
phosphorylation. The p38 MAPK (mitogen-activated protein
kinase), which is generally associated with inflammatory diseases
in response to stress stimuli, has been shown to act via a kinase-
independent effect on DYRK1B (minibrain-related kinase),
preventing its function as a transcriptional activator by association
with upstream activators and cofactors.[5] Investigations on the
role of kinases in the intricate networks of protein-substrate
interactions can extend our understanding of disease origins,
molecular target identification, and drug discovery. The
emergence of proteomic methodologies has produced seminal
contributions to this field.[6,7,8] Interactome analyses have been
successfully pursued employing techniques such as pull-down
assays from cell lysates and immunoblotting or mass
spectrometric (MS) analysis[9,10,11], as well as functional
examination by live-cell imaging microscopy applications.[12,13]
Frequently, small molecules serving as highly selective chemical
conformation and its accessibility is controlled by the side chain
of Trp197 which blocks the opening of the site; upon ligand
binding, the tryptophan side chain flips 90°.[19] We recently
synthesized a diverse set of 2-arylquinazoline LiPoLis, but their
low binding affinity rendered them unsuitable as molecular probes.
To examine how to maximize the potency of these ligands, we
used the aforementioned mutated kinases carrying cysteine
substitutions at sites within close proximity to the LP, with the
intention to modify the structure of the LiPoLis to enable the
formation of covalent bonds with the introduced Cys residues.
(Fig. 1A). Based on co-crystal structures from previous studies,
we identified amino acids that could be replaced by cysteines and
which were located near possible derivatization sites for
introducing electrophiles into the LiPoLi scaffold.[20] Hence, we
considered solvent-exposed positions on the quinazoline scaffold
directed outwards from the pocket as these sites are unlikely to
induce steric collisions and may conveniently address the artificial
cysteines that could be introduced at the periphery of the LP of
the kinase. Amino acid side chains that meet these criteria were
selected for site-directed mutagenesis to cysteine residues,
namely Lys249, Ser251, Ser252 and Asp294. We then prepared
a panel of four different mutant constructs, each carrying one of
the designated cysteines (K249C, S251C, S252C and D294C)
(Fig. 1B). After expression and purification, all proteins were
tested for catalytic integrity by determination of ATP- and
m
substrate-K as kinetic parameters using an activity-based assay,
[
a]
Mike Bührmann, Julia Hardick, Jörn Weisner, Lena Quambusch,
Prof. Dr. Daniel Rauh
Faculty of Chemistry and Chemical Biology
TU Dortmund University
Otto-Hahn-Straße 4a, 44227 Dortmund, Germany
E-mail: daniel.rauh@tu-dortmund.de
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.201706345
COMMUNICATION
Figure 1. Schematic illustration of a chemical genetic approach towards covalent LiPoLis. A) Illustration of the p38α kinase domain highlighting the LP (orange)
in distance to the ATP binding site (bound ATP in yellow). Rearrangement of Trp197 opens LP upon binding of reversible LiPoLis. Introduction of artificial
cysteine side chains in the LPs periphery and modification of LiPoLis with electrophiles should lead to covalent bond formation to maximize target residence
time; (PDB codes: 1P38, 1ATP, 4DLJ). B) Overview of the LP and key interactions with the LiPoLi scaffold. Chosen amino acids for mutation to cysteines are
shown as yellow spheres and solvent-exposed compound positions are depicted with arrows (7-/8-amine: blue, 2-phenyl: red) (PDB code: 4DLJ). The design
of electrophile-decorated LiPoLis was based on the location of the introduced cysteines and the chosen derivatization sites.
showing comparable results for the WT and the generated
mutants (Fig. S1, Tab. S1).
however, these compounds did not directly react in a subsequent
Grignard reaction. Therefore, 14a,b were reduced to analogous
aldehydes 15a,b which readily underwent conversion in both the
Grignard reaction to precursors of LiPoLis 1e,f and 2e,f and the
Claisen-Schmidt aldol condensation with acetone[22] to afford the
covalent ligands 1d and 2d. Oxidation of 17a,b and 18a,b yielded
final products 1e,f and 2e,f (Scheme S2). The syntheses of
LiPoLis carrying the electrophile at the 2-phenyl ring of the
quinazoline required individual routes and starting materials for
each compound (Scheme S3-S5). In general, we followed
procedures similar to those shown in Scheme S1, but using
various benzaldehydes to form the derivatized 2-aryl quinazoline
scaffolds 20, 23, and 26. The different introduced functionalities
on the 2-phenyl ring served as precursors for the generation of
the corresponding Michael acceptor systems. Nitrile 22 was
directly converted in a Grignard reaction to give probe molecule
Electrophiles were introduced at the 7- and 8-position on the
quinazoline scaffold and by meta-/para-substitution of the
2-phenyl moiety of the LiPoLis (Fig. 1B). Generation of the
designed 7-/8-N-acyl amides (1a-c and 2a-c) and 7-/8-alkyl
ketones (1d-f and 2d-f) were each realized by a common
synthesis route (Schemes S1, S2). In the first group of
compounds, the quinazoline scaffold was built up starting with the
condensation of nitroanthranilic acids 5a,b in the presence of
formamide
Hexachlorocyclophosphazene (HCCP)-mediated activation of the
-hydroxyl group and subsequent substitution[21] with
and
benzoic
anhydride.
4
phenethylamine yielded 7a,b. After reduction of the nitro group,
the 7-/8-position was available for nucleophilic substitution with
corresponding acid chlorides to yield the final covalent LiPoLis 1a-
c and 2a-c (Scheme S1). Following the second route, quinazoline
scaffold formation was achieved by condensation of the
diesterified aminoterephthalic acid building blocks 11a,b and
ammonia. The resulting carboxamides 12a,b were converted to
the corresponding carbonitriles 13a,b, using TFA anhydride;
3. Amino component 25 underwent substitution with acryloyl
chloride to generate final compound 4a, and the aldehyde
intermediate 29 served as a substrate in a Claisen-Schmidt
condensation yielding LiPoLi 4b. Altogether, a focused library of
15 covalent LiPoLis and their reversible counterparts was
synthesized to address the four aforementioned p38α mutants
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.201706345
COMMUNICATION
Table 1. Structural identity of synthesized covalent LiPoLis and their
reversible counterparts.
(
Table 1). Verification of our approach via identification of suitable
protein-ligand combinations was subsequently conducted by
means of MS studies. We chose MS as an effective tool enabling
the detection of covalent adducts and analyzed whether the
covalent LiPoLis are capable of modifying the p38α mutants. This
MS-based screen allowed us to identify the best matching protein-
ligand pairs based on observed kinase labeling. Therefore,
protein samples were incubated with the corresponding LiPoLis
for 1 h on ice prior to subsequent MS measurements. The
resulting spectra were compared with the mass derived from the
spectrum of the apo protein and the resulting mass shifts were
calculated (Fig. S2).
Initial control experiments using the p38α wildtype variant
(WT) without any cysteine residues in proximity to the LP
established that the covalent LiPoLis did not modify the WT
protein, implying that any modification in the presence of the Cys
mutants must occur at the artificially introduced cysteine (Fig. 2A).
Additional control experiments with the reversible counterparts of
some of the synthesized ligands (1b,f and 2b,f) confirmed the lack
of covalent labeling of the proteins (data not shown). When
incubating the four mutant proteins with the covalent LiPoLis, a
mass shift compared to the apo-protein was observed in some
cases, and the observed mass shift corresponded to a single
modification each. Generally, the results showed a recognizable
pattern of kinase modification in terms of compound and cysteine
reactivity, respectively (Fig. S3).
Figure 2. A) Mass spectra of p38
͍
α WT in absence (DMSO control) and
presence of 4a and 4b after 1 h of incubation. B) Mass spectra of p38α
S251C and S252C in absence (DMSO control) and presence of 4a and 4b
after 1 h of incubation. C) MS/MS spectrum of p38α S251C after incubation
with 4a, tryptic digest and analysis of the resulting peptide fragments.
Intriguingly, we observed a relatively high reactivity in the
mutant kinase carrying Cys252 (p38α S252C), with conspicuous
labeling by the covalent LiPoLis. Certain positions in a protein are
prone to alter the amino acid side chain nucleophilicity; for
cysteine residues, the location at the N-terminus of proximal
α-helices has been reported to significantly decrease the thiol’s
To our satisfaction, we found a clean profile of adducts with
the 2-aryl-modified covalent LiPoLis (3 and 4a,b) towards the
panel of tested mutant proteins, as indicated by the significant
modification of p38α S251C and S252C while sparing WT and
both of the other mutants (K249C, D294C), thereby substantiating
our initial design (Fig. 2A). Incubation with 4a,b (both substituted
in the meta-position of the 2-phenyl ring) led to considerable
labeling of the aforementioned mutants whereas 3 (substituted in
pK
a
, leading to favored interactions and reactivity towards
electrophiles.[
23]
Reactivity of Cys252 could additionally be
promoted by its side chain’s rotamer orientation and accessibility,
facilitating covalent bond formation by approaching electrophiles.
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.201706345
COMMUNICATION
para-position) did not react with p38α S251C and S252C. The
covalent LiPoLis 4a,b were subsequently further characterized by
MS/MS measurements to verify the proposed labeling position,
i.e., the introduced cysteines. Protein-ligand samples were
analyzed after tryptic digestion, but inferior flight behavior of the
resulting peptide fragments in the mass spectrometer
complicated the analysis due to insufficient sequence coverage.
However, the corresponding peptide fragment of p38α S251C
(
ICSESAR) with LiPoLi 4a could be identified, showing mass
shifts that corresponded to the calculated peptide mass
consistent with the molecular weight of the covalent probe
Figure 3. Co-crystal structures of A) p38α S251C in complex with 4a
(green) and B) p38α S252C in complex with 4b (yellow), both showing a
covalent bond to the corresponding cysteine thiols. At resolutions of 2.0 Å
each; 2Fo-Fc maps contoured at 1.0σ. Water molecules are shown as red
spheres. Hydrogen-bond interactions of the ligands with the protein and
water molecules are illustrated by yellow dotted lines (PDB: 5O8U, 5O8V).
(
Fig. 2B). The spectrum clearly depicted the presence of the
+
modified peptide as y
6
-fragment (calc. for [y
6
+4a] : 1062.3 Da,
found: 1061.5 Da) in combination with unlabeled y
5
-fragment
+
(
549.3 Da), as well as the labeled b
2
- ([b
2
+4a] : 626.3 Da) and
+
b
3
-fragment ([b
3
+4a] : 713.3 Da) were detected, allowing the
distinct assignment of the labeling position at Cys252. This finding
underlines the specific labeling of the introduced cysteines by the
designed LiPoLis.
4a did selectively occur at the artificially introduced cysteine
residue at the LP. Ultimately, protein crystallography delivered the
final evidence that our chemical genetic approach had been
successful. Crystal structures of 4a and 4b in complexes with the
mutant kinases showed a clearly defined covalent bond to Cys251
and Cys252, respectively.
Finally, the results of the mass spectroscopy data were
confirmed by protein crystallization studies. We attempted
crystallization trials for each kinase variant with the covalently
bound LiPoLis that we synthesized. We found that two protein-
ligand combinations reliably gave crystals under the chosen
conditions and consequently, the co-crystal structures of 4a in
complex with p38α S251C and 4b in complex with p38α S252C
could successfully be solved (Fig. 3A,B; PDB: 5O8U, 5O8V).
By following the presented approach, we achieved maximized
ligand residence time at the LP, overcoming the affinity and
potency shortcomings of our first generation of LiPoLis. These
novel ligands represent the first of their kind to covalently target
the LP in mutant p38α MAPK, paving the way for future intensive
analyses of its cellular function. With the suitable tools in place,
we are currently pursuing biochemical cellular studies to elucidate
the biological role of the p38α MAPK via pharmacological
perturbations.
Both compounds adopt the proposed binding mode in the LP
by forming the characteristic interactions known from the co-
crystal structures of reversible LiPoLis (Bührmann et al.,
submitted). These interactions include a π-π-stacking interaction
between the Trp197 side chain and the quinazoline core, a direct
hydrogen bond between Asp194 and the 7-amine, the phenethyl
residue of the LiPoLi binding deeply into the LP as well as a
complex water-mediated hydrogen bond network surrounding the
periphery of the binding site. Most importantly, for 4a a well-
defined electron density between the introduced Michael acceptor
system and Cys251 clearly indicated a covalent bond at this
position (Fig. 3A). The carbonyl oxygen of the introduced
acrylamide group formed an additional interaction with the
backbone NH of Cys251. Similarly, in the second crystal structure,
distinct electron density was also observed between the former
electrophile of 4b and Cys252 (Fig. 3B). In this case, the newly
introduced carbonyl group of the alkyl ketone formed an additional
direct hydrogen bond to the backbone NH of Trp197.
Acknowledgements
Simone Eppmann and Andreas Arndt are thanked for protein
expression and purification. Dr. Petra Janning, Jens Warmers,
Andreas Brockmeyer and Malte Metz are thanked for support
during MS and MS/MS measurements.
The authors declare no competing financial interests.
[1]
P. Cohen, D. R. Alessi, ACS Chem. Biol. 2013, 8, 96-104.
J. M. Zhang, P. L. Yang, N. S. Gray, Nat. Rev. Cancer 2009, 9, 28-39.
P. Cohen, Nat. Rev. Drug Discovery 2002, 1, 309-315.
[
[
2]
3]
In summary, we presented a chemical genetic approach
towards the optimization of 2-arylquinazolines to serve as putative
probe molecules to dissect the biological function of the LP in
p38α MAPK. Starting from previous work on reversibly binding
LiPoLis and a diverse set of complex protein crystal structures,
we designed kinase mutants with cysteine residues near the LP
to be addressed by the corresponding electrophile-bearing,
covalently binding ligands. MS studies conducted with the
generated mutants in the presence of the newly synthesized
[4]
J. Rauch, N. Volinsky, D. Romano, W. Kolch, Cell Commun. Signaling
2011, 9, 23.
[
[
[
[
5]
6]
7]
8]
S. Lim, Y. Zou, E. Friedman, J. Biol. Chem. 2002, 277, 49438-49445.
J. M. Gilmore, M. P. Washburn, J. Proteomics 2010, 73, 2078-2091.
R. Drissi, M. L. Dubois, F. M. Boisvert, FEBS J. 2013, 280, 5626-5634.
Y. M. Ye, E. C. Mar, S. X. Tong, S. Sammons, S. A. Fang, L. J. Anderson,
D. X. Wang, J. Virol. Methods 2010, 163, 87-95.
[
9]
J. R. Yates, C. I. Ruse, A. Nakorchevsky, Annu. Rev. Biomed. Eng. 2009,
11, 49-79.
[10] M. Vidal, M. E. Cusick, A. L. Barabasi, Cell 2011, 144, 986-998.
[11] D. I. Shirasaki, E. R. Greiner, I. Al-Ramahi, M. Gray, P. Boontheung, D.
H. Geschwind, J. Botas, G. Coppola, S. Horvath, J. A. Loo, X. W. Yang,
Neuron 2012, 75, 41-57.
LiPoLis revealed
a number of protein-ligand adducts, in
agreement with our design hypothesis. Furthermore, in MS/MS
studies we showed that the observed labeling of p38α S251C by
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.201706345
COMMUNICATION
[
12] G. Lukinavicius, K. Umezawa, N. Olivier, A. Honigmann, G. Y. Yang, T.
Plass, V. Mueller, L. Reymond, I. R. Correa, Z. G. Luo, C. Schultz, E. A.
Lemke, P. Heppenstall, C. Eggeling, S. Manley, K. Johnsson, Nat. Chem.
[18] J. A. Blair, D. Rauh, C. Kung, C. H. Yun, Q. W. Fan, H. Rode, C. Zhang,
M. J. Eck, W. A. Weiss, K. M. Shokat, Nat. Chem. Biol. 2007, 3, 229-238.
[19] J. J. P. Perry, R. M. Harris, D. Moiani, A. J. Olson, J. A. Tainer, J. Mol.
Biol. 2009, 391, 1-11.
2
013, 5, 132-139.
13] L. Dehmelt, P. I. H. Bastiaens, Nat. Rev. Mol. Cell Biol. 2010, 11, 440-
52.
[
[20] M. Getlik, J. R. Simard, M. Termathe, C. Grütter, M. Rabiller, W. A. van
Otterlo, D. Rauh, PLoS One 2012, 7, e39713.
4
[
[
[
[
14] J. Chan, S. C. Dodani, C. J. Chang, Nat. Chem. 2012, 4, 973-984.
15] S. K. Mamidyala, M. G. Finn, Chem. Soc. Rev. 2010, 39, 1252-1261.
16] A. W. Puri, M. Bogyo, ACS Chem. Biol. 2009, 4, 603-616.
[21] Z. L. Shen, X. F. He, J. L. Dai, W. M. Mo, B. X. Hu, N. Sun, X. Q. Hu,
Tetrahedron 2011, 67, 1665-1672.
[22] R. Mestres, Green Chem. 2004, 6, 583-603.
17] A. C. Bishop, O. Buzko, K. M. Shokat, Trends Cell Biol. 2001, 11, 167-
[23] T. Kortemme, T. E. Creighton, J. Mol. Biol. 1995, 253, 799-812.
172.
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.201706345
COMMUNICATION
COMMUNICATION
Mike Bührmann, Julia Hardick, Jörn
Weisner, Lena Quambusch, Daniel
Rauh*
Page No. – Page No.
Covalent Lipid Pocket Ligands
Targeting p38α MAPK Mutants
Here, we present a chemical genetic approach to covalently target a unique pocket
in the protein kinase p38α with yet unknown biological function. We designed and
synthesized a focused library of electrophile-bearing quinazolines and validated their
binding properties to mutant p38α by means of mass spectrometry and protein X-ray
crystallography. This approach successfully led to the identification of specific
protein-ligand pairs suitable for application future studies.
Keywords: ligand design • p38α MAPK • covalent probes • mass spectrometry • chemical genetic
This article is protected by copyright. All rights reserved.