An Angle on MK2 Inhibition—Optimization and Evaluation of Prevention of Activation Inhibitors

Article abstract of DOI:10.1002/cmdc.201900303

The mitogen-activated protein kinase p38α pathway has been an attractive target for the treatment of inflammatory conditions such as rheumatoid arthritis. While a number of p38α inhibitors have been taken to the clinic, they have been limited by their efficacy and toxicological profile. A lead identification program was initiated to selectively target prevention of activation (PoA) of mitogen-activated protein kinase-activated protein kinase 2 (MK2) rather than mitogen- and stress-activated protein kinase 1 (MSK1), both immediate downstream substrates of p38α, to improve the efficacy/safety profile over direct p38α inhibition. Starting with a series of pyrazole amide PoA MK2 inhibitor leads, and guided by structural chemistry and rational design, a highly selective imidazole 9 (2-(3′-(2-amino-2-oxoethyl)-[1,1′-biphenyl]-3-yl)-N-(5-(N,N-dimethylsulfamoyl)-2-methylphenyl)-1-propyl-1H-imidazole-5-carboxamide) and the orally bioavailable imidazole 18 (3-methyl-N-(2-methyl-5-sulfamoylphenyl)-2-(o-tolyl)imidazole-4-carboxamide) were discovered. The PoA concept was further evaluated by protein immunoblotting, which showed that the optimized PoA MK2 compounds, despite their biochemical selectivity against MSK1 phosphorylation, behaved similarly to p38 inhibitors in cellular signaling. This study highlights the importance of selective tool compounds in untangling complex signaling pathways, and although 9 and 18 were not differentiated from p38α inhibitors in a cellular context, they are still useful tools for further research directed to understand the role of MK2 in the p38α signaling pathway.

Full text of DOI:10.1002/cmdc.201900303

Accepted Article
Title: An angle on MK2 inhibition - Optimization and evaluation of
prevention of activation inhibitors
Authors: Mickael Tommy Mogemark, Ulf Hedström, Monica Norberg,
Emma Evertsson, Sarah R. Lever, Magnus Munck af
Rosenschöld, Hans Lönn, Peter Bold, Helena Käck,
Pia Berntsson, Johanna Vinblad, Jianming Liu, Anette
Welinder, Johan Karlsson, Arjan Snijder, Katerina Pardali, Ulf
Andersson, and Andrew M. Davis
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: ChemMedChem 10.1002/cmdc.201900303
Link to VoR: http://dx.doi.org/10.1002/cmdc.201900303
A Journal of
www.chemmedchem.org

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
An angle on MK2 inhibition - Optimization and evaluation of
prevention of activation inhibitors
Ulf Hedström,[b] Monica Norberg,[b] Emma Evertsson,[b] Sarah R. Lever,[b] Magnus Munck af
Rosenschöld,[b] Hans Lönn,[b] Peter Bold,[b] Helena Käck,[c] Pia Berntsson,[b] Johanna Vinblad,[b]
Jianming Liu,[c] Anette Welinder,[d] Johan Karlsson,[c] Arjan Snijder,[c] Katerina Pardali,*[b] Ulf
Andersson,[a] Andrew M. Davis,[b] Mickael Mogemark*[a]
Dedicated to Ulrika Svensson
[a]
[b]
Dr. Mickael Mogemark (https://orcid.org/0000-0002-7484-3719), Dr. Ulf Andersson
Clinical Pharmacology and Safety Sciences, R&D Biopharmaceuticals, AstraZeneca, Gothenburg, Sweden
Pepparedsleden 1, 431 83 Mölndal, Sweden
E-mail: mickael.mogemark@astrazeneca.com
Dr. Ulf Hedström, Dr. Monica Norberg, Dr. Emma Evertsson, Sarah R. Lever, Magnus Munck af Rosenschöld, Dr. Hans Lönn, Dr. Peter Bold, Pia
Berntsson, Johanna Vinblad, Dr. Katerina Pardali (https://orcid.org/0000-0002-4360-730X), Dr. Andrew M. Davis
Early Respiratory, Inflammation and Autoimmunity, R&D Biopharmaceuticals, AstraZeneca, Gothenburg, Sweden
Pepparedsleden 1, 431 83 Mölndal, Sweden
E-mail: : ulf.hedstrom@astrazeneca.com, emma.evertsson@astrazeneca.com, peter.bold@astrazeneca.com, pia.berntsson@astrazeneca.com,
katerina.pardali@astrazeneca.com, andy.davis@astrazeneca.com
[c]
[d]
Dr. Helena Käck, Dr. Jianming Liu, Johan Karlsson, Dr. Arjan Snijder
Discovery Sciences, R&D Biopharmaceuticals, AstraZeneca, Gothenburg, Sweden
Pepparedsleden 1, 431 83 Mölndal, Sweden
E-mail: helena.kack@astrazeneca.com, jianming.liu@astrazeneca.com
Anette Welinder
Pharmaceutical Technology and Development, Operations, AstraZeneca, Gothenburg, Sweden
Pepparedsleden 1, 431 83 Mölndal, Sweden
E-mail: anette.welinder@astrazeneca.com
1
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
Abstract: The mitogen-activated protein kinase p38α pathway has
been an attractive target for the treatment of inflammatory conditions
such as rheumatoid arthritis. While a number of p38α inhibitors have
been taken to the clinic they have been limited by their efficacy and
toxicological profile. A lead identification program was initiated to
selectively target prevention of activation (PoA) of mitogen-activated
protein kinase-activated protein kinase 2 (MK2) rather than mitogen-
and stress-activated protein kinase-1 (MSK1), both immediate
downstream substrates of p38α, to improve the efficacy/safety profile
over direct p38α inhibition. Starting with a series of pyrazole amide
PoA MK2 inhibitor leads, and guided by structural chemistry and
rational design, a highly selective imidazole 9 and the orally
bioavailable imidazole 18 were discovered. The PoA concept was
further evaluated by protein immunoblotting, which showed that the
optimized PoA MK2 compounds, despite their biochemical selectivity
against MSK1 phosphorylation, behaved similarly to p38 inhibitors in
cellular signaling. This study highlights the importance of selective tool
compounds in untangling complex signaling pathways, and while 9
and 18 were not differentiated from p38α inhibitors in a cellular context,
they still are useful tools for further research directed to understand
the role of MK2 in the p38α signaling pathway.
Furthermore, we aimed at inhibiting the phosphorylation of MK2
by binding to the heterodimeric complex of p38/MK2 via a
prevention of activation (PoA) mechanism.[9],[10] Hence, the
inhibition should be substrate selective since it wouldn’t interfere
with the p38-MSK1 signaling as described previously.[10] High
throughput screening of the AstraZeneca compound collection
afforded two distinct lead series represented by hit compounds 1
(series 1) and 2 (series 2) , showing a promising separation of
p38-MSK1 substrate selectivity (Figure 2).[10] In this paper we
describe our efforts to further explore the pyrazole amide series
2, evolved from compound 2.
Figure 2. AstraZeneca high throughput screening hit compounds 1-2 and
compound 3 from Bristol-Myers Squibb targeting p38α.
Introduction
The mitogen-activated protein kinases (MAPK) p38α MAPK
(p38α) and MAPK-activated protein kinase 2 (MK2) are attractive
drug discovery targets for the treatment of inflammatory
conditions such as rheumatoid arthritis (RA).[1],[2],[3] Several p38α
inhibitors have been tested in the clinic for rheumatoid arthritis
including BIRB796 (BI),[4] VX-745[5] and VX-702 (Vertex)[6], but
were discontinued due to concerns over their toxicological profile
and/or lack of efficacy at the doses that could be used.[7]
Selectively targeting MK2 downstream of p38α has been
suggested to provide p38α-like efficacy with downregulation of
TNF-α production along with an improved safety profile.[3],[8] The
interest in selectively targeting MK2 also resides in the fact that
p38α not only activates MK2 but also mitogen- and stress-
activated protein kinase 1 (MSK1). The inhibition of MSK1 leads
to downregulation of the anti-inflammatory cytokines interleukin
(IL)-10 and IL-1 receptor antagonist (IL-1RA) and hence reduction
of the overall efficacy.[8] We hypothesized that a ligand selectively
binding to the heterodimeric complex of p38α/MK2 could block the
production of pro-inflammatory TNF-α, whilst maintaining the IL-
10 and IL-1RA levels (Figure 1).
Results and Discussion
The p38α inhibitor 3, previously disclosed by Das et al,[11] is
structurally similar to compound 2, but while 2 is 98x selective for
MK2 inhibition over MSK1, 3 is not selective (Figure 2). In order
to understand the selectivity profile of 2, it was crystallized in
complex with p38α/MK2, and its binding interactions compared to
3. The X-ray of 2 bound to p38α/MK2 (pdb code: 4THY) shows
that it can bind to the ATP binding pocket of p38α (Figure 3a). The
binding of 2 to the p38α/MK2 complex displays hydrogen bond
interactions between the sulfonamide oxygen and the Asp168
backbone NH, and between the pyrazole nitrogen and the hinge
residue Met109. The n-propyl substituent does not appear to
make any specific interaction and leaves room for further
modification of this substituent. The pyrazole phenyl group
reaches towards the MK2 protein surface in the protein complex
however, the distance is too long for direct interactions (closest
distance to Gln369 is 3.7 Å). The structurally analogous
compound 3 displays similar binding interactions to p38α (Figure
3b), but without the sulfonamide oxygen interaction to Asp168,
potentially responsible for driving the selectivity for PoA-MK2
versus PoA-MSK1. Compound 2 is a moderately potent inhibitor
of MK2 but with good selectivity towards MSK1 through its specific
interactions with the p38α/MK2 complex, but with low aqueous
solubility and low stability in human liver microsomes. Hence, our
design strategy was to improve the MK2-PoA potency and the
drug-like properties, while improving selectivity towards MSK1 by
building on the interactions made by 2 to the p38α/MK2 complex,
and further extending towards the MK2 protein surface.
Figure 1. Schematic representation of MAP kinase signaling network.
2
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
molecules towards the MK2 protein surface of the p38α/MK2
complex in an attempt to identify novel interactions (Figure 3a and
3b). With the X-ray structure of 2 bound to the p38α/MK2 complex
as starting point, compounds reaching towards the MK2 protein
surface were designed and synthesized. Structure based design
suggested that the terminal phenyl acetamide substituent had the
possibility to generate a hydrogen bond to the Asp366 from MK2
protein as can be seen in docking shown in figure 4, which lead
us to include this substituent in a library to explore SAR. As
predicted, these meta-substituted phenyl acetamides showed
promising activity and selectivity profiles. The primary amide 9, as
well as the secondary amide 10, showed slightly improved
potency compared to unsubstituted 4, but more importantly the
selectivity over MSK1-PoA increased more than 3-fold. However,
both compounds suffered from poor metabolic stability and low
solubility. Compound 9 was tested in a pan kinase selectivity
panel of 273 kinases and was not active against any kinase tested
at 1 µM (Table S1 in the Supporting Information). The left hand
R1 substituent was revisited to investigate if the structure–activity
relationship (SAR) for the R2 = phenyl acetamide subseries is
parallel to that of the R2 = H subseries. The secondary
sulfonamide 11 showed low nanomolar MK2-PoA potency, but
with decreased MSK1 selectivity compared to the tertiary
sulfonamide 9. A similar trend was seen for the primary
sulfonamide 12, which was 10-fold less selective against MSK1
compared to compound 10, although an improved potency was
obtained. The primary amide 13 showed excellent MK2-PoA
potency (IC50 = 2 nM), but displayed no selectivity over MSK1, in
comparison with 6. Compounds 11 and 13 suffered from low
solubility but showed improved metabolic stability in human liver
microsomes compared to compound 9.
Figure 3a. Structure of compound 2 (cyan) binding to the p38α/MK2 complex.
Gln369, Glu368, Tyr367 and Asp366 (shown to the right in yellow) are part of
the MK2 chain of the complex. DFG loop residues Asp168, Phe169 and Gly170
are shown in dark green and the hinge residues Leu108 and Met109 are shown
as bold sticks. Putative hydrogen bonds are shown as dashed lines. (pdb code:
4TYH).
In summary, parallel SAR was observed for the two different sub-
series. Nonetheless, comparing compound 9 to the compound 4,
the combination of tertiary sulfonamide and primary m-
phenylacetamide substitutions afforded an apparent cooperative
effect to improve MK2 selectivity over MSK1. This selectivity data
supports the idea that substrate selectivity can be improved by
extending towards the MK2 surface in the protein dimer complex,
as suggested by docking studies.[12] These studies suggest the
formation of a possible hydrogen bond between the NH of the
phenylacetamide (R2-substituent) and Asp366 in the MK2 protein
(Figure 4).
Figure 3b. Comparison of compound 3 (orange) binding to the active site of
p38α (pdb code: 3OCG) and compound 2 (cyan) binding to the p38α/MK2
complex (p38α in green and MK2 in yellow) (pdb code: 4TYH).
Initial structure-activity relationship (SAR) investigation focused
on identifying alternatives to the pyrazole core. The imidazole-
based compound 4 (Table 1) offered an opportunity to maintain
the key interaction with Met109 while exploring lipophilicity-
dependent effects, e.g. solubility, and intrinsic clearance, and
thereby improve the overall pharmacokinetic profile. The potency
for the imidazole analogue 4 improved 2-fold in the MK2-PoA
assay compared to its pyrazole analogue 2. Despite equivalent
LogD, the solubility of imidazole 4 was improved 70-fold
compared to the corresponding pyrazole 2. The importance of the
sulfonamide in the aniline 5-position was confirmed by introducing
different polar groups, preserving the possibility for a hydrogen
bond interaction with Asp168 in the p38α/MK2 complex, as
previously shown in the pyrazole series.[10] Changing the tertiary
sulfonamide 4 to the primary sulfonamide 5 increased MK2-PoA
potency, but with an overall drop in the MK2-PoA/MSK1-PoA
selectivity. Replacing the sulfonamide with primary or secondary
amides 6 and 7, resulted in sub-nanomolar MK2-PoA potencies,
but also led to a complete loss of selectivity over MSK1. The
carboxylic acid 8 lost 6-fold in MK2-PoA potency with a 70-fold
gain of selectivity over MSK1-PoA compared to amide 6.
Unfortunately, the carboxylic acid analogue 8 did not inhibit TNF-
α production in lipopolysaccharide (LPS)-stimulated human
peripheral blood mononuclear cells (PBMC) (data not shown). To
further improve potency and selectivity, we extended the
Figure 4. Docking of compound 9 into the p38α/MK2 tertiary complex, showing
a potential H-bond between the terminal phenyl acetamide and Asp366 from the
MK2 protein.
3
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
This interaction could potentially explain the observed increase in
potency and selectivity. The sulfonamide substituent is remote
from the MK2 and p38α interface, but close to the DFG motif
which extends into the activation loop. Structural overlays
between free and MK2-bound p38 suggest that the conformation
of the p38 activation loop may be affected by interaction to MK2,
and thus interactions with the DFG motif may have a significance
for selectivity. However, we cannot yet explain the influence of the
sulfonamide on selectivity in any details from the crystal structures
or the docking studies.
subseries (most compounds showed both poor metabolic stability
and solubility), we focused our attention on the core structure
binding interactions and on modifications of the n-propyl chain in
an attempt to improve LogD, solubility and metabolic stability. As
was mentioned previously, the crystal structure of compound 2 in
the p38α/MK2 complex (Figure 3), shows no apparent interaction
between the n-propyl chain in compound 3 and the protein. We
speculated that the propyl-chain is important for the conformation
of the molecule, more precisely the torsional angle between the
heteroaromatic core and the phenyl ring. This hypothesis
prompted us to explore the conformational relationship, whilst
trying to reduce size and lipophilicity of the propyl chain.
To improve the overall pharmacokinetic profile with more ligand
efficient compounds than found for the m-phenylacetamide
4
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
Table 1. Activity data for SAR exploration.
Compound
R1
R2
H
MK2-PoA IC50 (nM)[a]
105
MSK1 selectivity[b]
98
Octanol LogD[c]
3.6†
Solubility (μM)[c]
HLM CLint (μl/min/mg)[c],[d]
2[e]
Me2NO2S
0.6†
42
62†
4
5
Me2NO2S
H2NO2S
H2NOC
H
H
H
H
H
59†
19
53
10
3.5†
2.7
2.7
3.5
0.4
3.8
4.5
3.5
3.3
3.1
110
26
280
6
14
0.4
<0.5
79
37
8
19
7
cPrHNOC
HO2C
3.4
85‡
16†
17‡
6.3†
3.7‡
2.2†
<3
8
820
50
9
Me2NO2S
Me2NO2S
270
150
95
0.5
>300
250
95
10
11
12
13
1
5
8
MeHNO2
S
H2NO2S
H2NOC
15
49
1.5
0.5
5.4
[a] Values are means of n≥3, except †n=2 and ‡n=1. [b] MSK1-PoA IC50/MK2-PoA IC50. [c] Values represent n=1, except †n=2. [d] Intrinsic clearance (CLint) in human
liver microsomes (HLM). [e] Pyrazole heteroaromatic core.
5
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
Table 2. DA is the calculated optimal dihedral angle between the imidazole and
phenyl rings. ΔΕ is the energy (kJ/mol) required to force the dihedral angle to
60°. Substituents on the third, more distant ring were excluded from the
calculations, since they are not expected to contribute to the dihedral angle.
R
H
DA
1
R
H
DA
37
 (KJ/mol)
 (KJ/mol)
14
4.4
Me
Et
7
9
7
Me
Et
64
0.8
2.7
6.8
5.8
7.7
8.4
18
20
38
126
20
71
Figure 5. X-ray structure of compound 2 in the conformation that is bound to
the p38/MK2 complex, showing the dihedral angle between the pyrazole and
phenyl.
Pr
7
Pr
72
iPr
tBu
Ph
3
iPr
tBu
Ph
126
97
11
8
126
The crystal structure of compound 2 bound to the p38α/MK2
complex shows a dihedral angle of ~60° between the pyrazole
and the phenyl rings (Figure 5). For comparison, the crystal
structure of the analogue p38α compound 3 (pdb code: 3OCG)
shows a slightly larger dihedral angle of 66°. These measured
values are significantly larger than the energy optimized dihedral
angle of 20°, calculated using quantum mechanical geometry
optimization (B3LYP/6-31G**)[13] for compound 2 and for the
analogous compound 5. This suggests that the binding to the
protein increased the dihedral angle by 40°, corresponding to an
energy penalty of about 7 kJ/mol. The dihedral angle of compound
3, bearing an amine in 5-position on the imidazole, was also
calculated. The angle was 53° with a 1 kJ/mol energy penalty to
increase it to 60°. To identify substituents forcing a dihedral angle
closer to the optimal angle of 60°, and thus maximize the binding
free energy, we calculated the conformation of a molecule set.
The set included simplified forms of N-methylated and
unsubstituted imidazoles with different substituents in the 2-
position of the adjacent phenyl (Table 2). The N-methylimidazole
18 with a methyl in the 2-positon on the phenyl provided a
calculated dihedral angle of 64°. The energy cost of forcing the
conformation to a dihedral angle of 60° was calculated to be 0.8
kJ/mol. A small set of compounds were subsequently synthesized
including the non-methylated imidazoles as a control. To reduce
the LogD and optimize bioavailability, we decided to focus on the
primary sulfonamide on the aniline in this study.
Table 3. Activity data for exploration of dihedral angle.
Compound
R1
R2
MK2-PoA IC50
(nM)[a]
MSK1
selectivity[b]
5
nPr
Me
Me
H
H
19
61‡
3‡
10
5.8
22
17
18
19
20
21
3[c]
iPr
Me
Me
Et
>990†
>990‡
Not active†
1
>3.6
>1.5
-
H
H
Ph
< 0.6
[a] Values represent n≥3, except †n=2 and ‡n=1. [b] MSK1-PoA IC50/MK2-PoA
IC50. [c] Bristol-Myers Squibb Research and Development pyrazole core.
The described optimization of the dihedral angle focused on
potency and physicochemical properties of the compounds, not
taking the desired selectivity towards MSK1-PoA into account.
The resulting compound 18 thus displayed excellent 3 nM potency
for MK2-PoA but with a modest 22-fold selectivity. It also showed
improved potency in the human PBMC cell assay (Table 4)
compared to compound 9 and the lower LogD of 2.2 compared to
3.8 improved its properties for oral administration. Solubility was
increased from 0.5 to 784 µM while clearance in human liver
microsomes was reduced from >300 to <3 μl/min/mg. It was
permeable with an apparent permeability value of 1.4 x 10-6 cm/s
measured in Caco-2 cells, but exhibited an efflux ratio of 24, which
might limit its absorption at low doses. However, despite
promising in vitro data, compound 18 was cleared efficiently in rat
with a clearance value of 37.4 ml/min/kg which corresponds to
51% of the liver blood flow (Figure S1 in the Supporting
Information). To avoid more demanding in vivo studies, the in
vitro-in vivo disconnect was not further pursued for these tool
compounds at this stage of the project. Furthermore, compound
18 was inactive in a human ether-a-go-go-related gene (hERG)
assay and in cytochrome P450 enzyme inhibition assays (5
isoforms) (data not shown). It was also tested in a pan-kinase
selectivity panel of 130 kinases at 1 μM and showed excellent pan
kinase selectivity (<31% inhibition against all kinases, except for
p38α that showed 94% inhibition) (Table S2 in the Supporting
Information).
In agreement with the calculations, compound 18 with a
calculated dihedral angle of 64°, resulted in a 5-fold increase in
potency compared to 5 (Table 3). In alignment with our predictions,
the isopropyl substituted analogue 17 (126° dihedral angle)
decreased the potency 20-fold in comparison to 18. Three
matching compounds with an unsubstituted imidazole were also
prepared, 19, 20 and 21. All showed a complete loss in potency,
which could be explained by the non-optimal dihedral angle
varying between 7° and 20° (Table 2).
6
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
Table 4. Biochemical and cellular potency profiles
Compound
MK2-PoA IC50
(nM)[a]
MSK1
selectivity[b]
MK2 IC50 (nM)[c]
p38 IC50
Human PBMC
TNF-α IC50 (nM)[a]
Human PBMC
IL-10 IC50 (nM)[a]
IL-10 selectivity[d]
(nM)[c]
5
19
16†
3‡
10
270
22
< 0.6
0.8
0.5
-
Not active
290
4600
240†
1009
119†
1†
1191†
1874
390†
6.7†
5
9
Not active
1.9
3.2
6.7
3.3
3.4
1.6
0.6
1.5
18
Not active
83
3
1
2†
10‡
15
Not active
-
PH-797804
Losmapimod
PF-3644022
22
-
6
20
Not active
-
8
27
Not active
Not active
18500‡
10
3
Not active
Not active
Not active
1128
4540†
55†
1774
2583†
8485†
-
23
-
4
[a] Values are means of n≥3, except †n=2 and ‡n=1. [b] MSK1-PoA IC50 /MK2-PoA IC50. [c] Values represent n=1. [d] IL-10 IC50/TNF-α IC50
.
Despite good in vitro MSK1 selectivity, when compounds 5, 9 and
18 were assessed in human PBMC after LPS stimulation, we
were unable to detect the inhibition of TNF-α levels along with the
preservation of IL-10 levels (Table 4).
phosphorylation (Figure 6c). Compound 5 did not upregulate
TRIM28 phosphorylation while compound 9 appeared to induce it.
These results suggest that both compounds behave as p38-like
inhibitors despite their biochemical selectivity against MSK1
phosphorylation.
In order to understand the effect of p38 inhibition, PoA MK2
inhibition and direct MK2 inhibition, we used protein
immunoblotting to explore activation of pathways around p38
signaling (Figure 1). We analyzed phosphorylation of HSP27,
TRIM28 and MSK1 as they are activated downstream of
p38.[14],[15],[16] After stimulating human PBMC with LPS in the
presence of the designated compounds at IC80 and 5 x IC80
concentrations of TNF-α inhibition, we observed that the known
p38 inhibitors PH-797804,[17] Losmapimod[18] and 3 (Figure 6a)
had a profile that was distinct from that of the MK2 active site
inhibitors PF-3644022,[19] 22[20] and 23[21] (Figure 6b). The p38
inhibitors all inhibited p38 phosphorylation, completely abolished
HSP27 phosphorylation and upregulated phosphorylation of
ERK1/2 (Figure 6a). In contrast, the MK2 active site inhibitors PF-
3644022, 22 and 23 upregulated p38 phosphorylation and
partially inhibited HSP27 phosphorylation, while ERK1/2
phosphorylation was not induced (Figure 6b). The upregulation of
p38 phosphorylation by MK2 inhibitors is in agreement with
previous findings that suggest that inhibition of MK2 maintains
p38 phosphorylation.[22] The mechanism for this relates to the
ability of activated MK2 to regulate the export of p38 from the
nucleus, which results in the dephosphorylation of p38 by
cytosolic phosphatases.[23] When we investigated compounds 5
and 9 to understand if they behaved in a p38-like or MK2-like
fashion, we observed that the compounds exhibited significant
reduction
of
HSP27
phosphorylation,
upregulated
phosphorylation of ERK1/2, and no upregulation of p38 or MSK1
7
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
IC80 from TNF-α assay, C2: 5xIC80 from TNF-α assay, C3: 4xIC80 from TNF-
α assay. Phospho-p38, phospho-HSP27 and phospho-ERK1/2 are highlighted
in red. The phosphorylated amino acid residues targeted by the anti-phospho
primary antibodies are indicated. Representative immunoblots are shown for
n=2 PBMC donors.
Figure 6. Immunoblots against proteins associated with p38 signaling in LPS-
stimulated human PBMC in the presence of a) p38 inhibitors: PH-797804,
losmapimod and compound 3; b) MK2 inhibitors: PF-3644022, compound 23
and compound 22; c) MK2 PoA inhibitors: compound 5 and compound 9.
Compound concentrations were chosen based on potencies observed for
inhibition of TNF-α in LPS-stimulated human PBMC. C0: no compound, C1:
8
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
However, our data for PF-3644022 (Table 4) show that direct
inhibition of MK2 also fails to preserve the anti-inflammatory IL-10
in human PBMC while inhibiting pro-inflammatory TNF-α
production. Therefore, sequestration of p38 cannot be the only
explanation for this and our data contradict previous literature
reports.[2] More recent scientific reports[25],[26],[27],[28],[29],[30],[31]
demonstrate that IL-10 is directly regulated by MK2 in addition to
MSK1. As such, our data invalidate MK2 as a target for selective
inhibition of inflammation.
Further analysis of a number of compounds in the same series
demonstrated that there is a disconnect between the in vitro PoA
MK2 and MSK1 selectivity and the human PBMC data
determining the downstream TNF-α and IL-10 release,
respectively (Figure 7).
Conclusions
In summary, by replacing the core pyrazole of compound 2 with
imidazole, potency and solubility were improved within the series
of compounds. The presence of dialkyl sulfonamide was essential
to balance MK2 potency and selectivity, through its interaction
with Asp168 in the p38 part of the p38/MK2 complex. We
discovered the highly selective compound 9 with a more than 200-
fold selectivity of MK2-PoA over MSK1-PoA by extending towards
the MK2 interface of the binding pocket. This unfortunately at the
cost of oral drug properties, and thus our focus moved to other
parts of the molecule. Supported by analysis of the central
torsional angle, compound 18 was discovered, representing a
novel series with increased potency and favorable physical
chemistry properties. When we attempted to confirm this
selectivity in the human PBMC cell assay, we found that selective
inhibition of the MK2 activation did not preserve anti-inflammatory
IL-10 while inhibiting pro-inflammatory TNF-α production. Using
protein immunoblotting to explore the pathways around p38
confirmed that the majority of PoA MK2 compounds, despite their
biochemical selectivity against MSK1 phosphorylation, behaved
similarly to p38 inhibitors. This paper serves as a useful reminder
that selectivity, as viewed from isolated enzyme assays, can be
misleading, and selectivity should be viewed from the cellular
context where proteins are in their native context and complete
signaling pathways are intact. These data, along with the fact that
selective inhibition of the MK2 activation did not preserve anti-
inflammatory IL-10 production, led us to discontinue the project.
Despite this, compounds 9 and 18 are useful tools for further
research directed to understand the role of MK2 in the p38
signaling pathway.
Figure 7. Plots of human PBMC IL-10 pIC50 vs. human PBMC TNF-α pIC50
and MSK1-prevention of activation (PoA) pIC50 vs. MK2-PoA pIC50. The line
represents y = x. Shown are ATP competitive p38 inhibitors (yellow) and MK2
PoA inhibitors (blue).
This is an intriguing observation that supports a previous
publication where substrate selective MK2 inhibitors were
proposed to behave as p38 inhibitors in a cellular context due to
differences in protein expression (MK2>p38),[24] causing all
active p38 to be sequestered in an inactive complex with MK2
9
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
Experimental Section
resuspended in 25 ml R10 buffer, diluted 1:10 in Trypan blue, and
counted in a Bürker chamber. The cells were diluted to 1.052 x
106 cells /ml in prewarmed R10 buffer (200,000 cells /190 µl) and
kept at 37°C. 1 µl compound/well was placed in 96-well round
bottom culture plates (Costar 96 well Cell Culture Cluster, cat.
3799) and 190 ml cell solution was added and mixed at 300 rpm
for 1 min, followed by preincubation for 45 minutes at 37°C, 5%
CO2. The experiment was initiated by the addition of 10 µl of 5
ng/ml LPS (from E.coli serotype 0127:B8, cat. L-4516, batch
109K4063) per well and incubated for 18 hours at 37°C, 5% CO2.
After the incubation the plate was centrifuged at 1200 rpm at 4°C
and 160 µl cell-free supernatant was transferred to a V-bottomed
96-well plate (Greiner 96-well plates, cat. 651201) for cytokine
analysis.
PoA MK2 enzyme assay
The kinase reaction takes place as a homogeneous assay in
solution with purified recombinant proteins p-p38a and MK2. In
this assay, MK2 is phosphorylated by active kinase p-p38a in vitro,
phosphorylated MK2 is then captured by using NeutrAvidin
coated ELISA plate and detected with a specific antibody directed
towards the phosphorylated site (Thr222) of MK2. Compounds
that prevent MK2 from phosphorylation by p-p38a lead to a
decreased/no signal in an ELISA assay. Compounds are tested
in IC50 format (10 concentrations: 33.3 µM to 1.7 nM).
Cytokine V-plex MSD analysis
Recombinant human MK2 was expressed in E. coli and purified
in house Construct: His6-Avi(biotin)-hMK2[46-400], and was
prepared as a 20 nM solution in assay buffer
MSD kit with 96-well plates spotted with specific anti-cytokine
capture antibodies, custom made by Meso Scale Discovery, was
used. A combined working stock solution at 1 µg/ml of each
cytokine was prepared. An 8-point cytokine calibration curve of
10000, 2500, 625, 156, 39, 9.8, 2.4 pg/ml and blank sample was
prepared. For the assay the samples were thawed on ice. We
added 25 µl/well calibrator or sample into each well. The plate
was incubated at room temperature with shaking (600 rpm) for 2
hours. The plate was washed 3 x 300 ml with PBS-0.05% Tween-
20 (PBS-Tween tablets cat. 09 9410-100, Medicago). For each
plate used, we diluted a 60 µl aliquot of the stock detection
antibody mix into 2.94 ml of Diluent 100 and this reagent was kept
in the dark. We dispensed 25 µl/well detection antibody solution
and incubated at room temperature with shaking for a further 2
hours. The plate was washed with 3 x 300 ml PBS-0.05% Tween-
20. The read buffer was diluted 2-fold in deionized water to make
a final concentration of 2x read buffer T, by using 10 ml of stock
read buffer T (4x) and 10 ml deionized water for each plate and
dispensed 150 µl/well 2x read buffer T and the plate was read on
an MSD instrument.
PoA MSK1 enzyme assay
The kinase reaction takes place as a homogeneous assay in
solution with purified recombinant proteins p-p38a and MSK1. In
this assay, MSK1 is phosphorylated by active kinase p-p38a in
vitro, phosphorylated MSK1 is then captured on ELISA plate by
passive binding and detected with a specific antibody directed
towards the phosphorylated site (Thr581) of MSK1 as illustrated
in Figure 2. Compounds that prevent MSK1 from phosphorylation
by p-p38α lead to a decreased/no signal in ELISA assay
Compounds are tested in IC50 format (10 concentrations: 100 µM
to 5.1 nM).
Human PBMC assay
Human PBMC were prepared from 150 ml of fresh whole blood
per donor from healthy subjects. Informed consent was acquired
from all subjects. In brief, fresh whole blood was layered on top of
15 ml Ficoll Paque PLUS (cat. 17-1440-03, Amersham) and
centrifuged at 1600 rpm without brake for 35 mins at room
temperature. The PBMC layer (second from top below the plasma
layer) was collected by pipette and washed with at least 3
volumes R10 medium (RPMI 1640 w GlutaMax I + 25mM HEPES
(500 ml) (cat. 72400-021, GIBCO) supplemented with heat
inactivated FBS 10% (50 ml) (cat. 10438-026, GIBCO) and PEST
(Penicillin+Steptomycin), (5 ml) (cat. 15070-063, GIBCO) and
centrifuged at 1300 rpm for 10 minutes at room temperature. The
supernatant was removed and the pellet resuspended in 1 ml
phosphate buffered saline (PBS). Two ml of water was added to
lyse red blood cells with gentle agitation by pipetting up and down
for 30 seconds. The mixture was quenched by the addition of 2
ml 1.8% NaCl (sterile-filtered) and then further R10 buffer was
added to fill the 50 ml Falcon tubes, which where centrifuged
again. The cells were resuspended in fresh buffer and cells were
pooled. The wash step was repeated. The cells were
Calculation of results
Calculations of results were done in Excel. XLfit was used to fit
the calibrator data to a standard curve, from which the
concentration of each cytokine was calculated using the following
equation: % Inhibition = 100 x [(max-X)/(max-min)], where X =
cytokine concentration for each compound concentration, Max =
average value cytokine conc pos controls (+ LPS), Min = average
value cytokine conc neg controls (- LPS). We used XLfit to
calculate the results, inhibition method 203 is recommended.
Active compounds were defined as compounds with ≥30%
inhibition. Incomplete curves with inhibition >50% were
extrapolated by locking the max value to 100.
10
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
Protein immunoblot analysis
Signaling Technology, Boston, MA, USA), 1:1000. After washing
of the membranes 3 times in TBS-Tween, they were incubated
with an HRP-conjugated anti-rabbit IgG antibody at RT for 1 hour
with constant agitation. Following 3 more washes in TBS-Tween,
the chemiluminescence was developed in a freshly made solution
containing 2.5 mM luminol, 200 µM coumaric acid, 0.018%
hydrogen peroxide and 100 mM Tris (pH 8.6), and the signal was
transferred to an X-ray film. The membranes were stripped of all
antibodies by incubation in a solution containing 2% SDS, 100
mM 2-mercaptoethanol and 62.5 mM Tris (pH 6.7) at 60°C for 30
minutes. After 2 washes in distilled water and 3 washes in TBS-
Tween, the membranes were incubated with the following
antibodies at 4°C overnight with constant agitation:
p38α/MAPK14 antibody (cat. AF8691, R&D Systems,
Minneapolis, MN, USA), 1:2000; MSK1 antibody (cat. 3489, Cell
Signaling Technology, Boston, MA, USA), 1:2500; HSP27
antibody (cat. 2402, Signaling Technology, Boston, MA, USA),
1:2000; TRIM28 antibody (cat. ab109287, Abcam, Cambridge,
UK), 1:10000; ERK (p44/p42) antibody (cat. 4695, Cell Signaling
Technology, Boston, MA, USA), 1:1000. As described above, the
membranes were again washed, probed with secondary HRP-
conjugated antibodies (anti-mouse IgG for the HRP27 antibody
and anti-rabbit IgG for the other antibodies) and the
chemiluminescence was developed.
8 x 106 PBMC isolated from freshly drawn blood from healthy
volunteers were placed in 35 mm dishes (Corning) in RPMI1640
medium (cat. # 11875-093, Thermo Fisher, Waltham, MA, USA)
supplemented with 10% heat inactivated fetal bovine serum (cat.
10082-147, Thermo Fisher, Waltham, MA, USA), incubated with
the designated compounds or DMSO for 45 min, followed by an
incubation with 0.5 ng/ml LPS (cat. L4516 Sigma Aldrich, St. Louis,
MO, USA) for 30 min. Both incubations were done at 37˚C in a
5% CO2 incubator. The cells were then centrifuged at 1000 g and
4°C for 2 minutes and lysed on ice with lysis buffer (20 mM Tris
(pH 7.2), 1 mM EDTA, 1 mM EGTA, 5 mM sodium pyrophosphate,
1 mM sodium orthophosphate, 50 mM sodium fluoride, 270 mM
sucrose and 1% Triton X-100) supplemented with Complete Mini
Protease Inhibitor Cocktail tablets (cat. 11836153001, Roche,
Switzerland). The lysates were centrifuged at 21 000 g at 4°C for
10 min and protein concentrations in the supernatants were
measured using a BCA protein assay kit (cat. 23225, Thermo
Fisher). The proteins were then separated by SDS-PAGE on
NuPAGE Novex 4-12% Bis-Tris gels (cat. NP0321BOX, Thermo
Fisher) and transferred to PVDF membranes using the iBlot
system (Thermo Fisher). Following blocking in TBS-Tween with
5% BSA, the membranes were incubated with the following
antibodies at 4°C overnight with constant agitation: phospho-p38
(pT180/pY182) antibody (cat. 4511, Cell Signaling Technology,
Boston, MA, USA), 1:2500; phospho-MSK1 (pS376) antibody (cat.
9591, Cell Signaling Technology, Boston, MA, USA), 1:500;
phospho-HSP27 (pS78) antibody (cat. 1543-1, Epitomics,
Burlingame, CA, USA), 1:2000; phospho-TRIM28 (pS473)
antibody (cat. ab109545, Abcam, Cambridge, UK), 1:20000;
phospho-ERK (p44/p42) (pT202/pY204) antibody (cat. 4376, Cell
Supporting Information
The supporting information contains synthetic procedures and
analytical characterization for all new compounds described in
this paper, rat pharmacokinetic data for 18 and pan-kinase
selectivity data for 9 and 18.
11
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
Keywords: kinase inhibitor • mode of action • prevention of activation • ligand design • drug discovery
Table of Contents
The p38α pathway is an attractive target for inflammation research. Cellular signaling downstream of p38α was probed using
optimized pyrazole amide MK2 prevention of activation (PoA) inhibitors. Despite biochemical selectivity blocking p38 phosphorylation
of MK2 vs MSK1, in cells PoA MK2 and p38 inhibitors behaved similarly. This study highlights the importance of selective tool
compounds in untangling complex cellular signaling pathways.
References
1 S. Duraisamy, M. Bajpai, U. Bughani, S.G. Dastidar, A. Ray, P. Chopra, Expert Opin. Ther. Targets 2008, 12, 921 – 936.
2 M. Gaestel, A. Kotlyarov, M. Kracht, Nat. Rev. Drug Discovery 2009, 8, 480 – 499.
3 M. Fiore, S. Forli, F. Manetti, J. Med. Chem. 2016, 59, 3609 − 3634.
4 J. Regan, S. Breitfelder, P. Cirillo, T. Gilmore, A.G. Graham, E. Hickey, B. Klaus, J. Madwed, M. Moriak, N. Moss, C. Pargellis, S. Pav, A. Proto, A. Swinamer,
L. Tong, C. Torcellini, J. Med. Chem. 2002, 45, 2994 – 3008.
5 J.P. Duffy, E.M. Harrington, F.G. Salituro, J.E. Cochran, J. Green, H. Gao, G.W. Bemis, G. Evindar, V.P. Galullo, P.J. Ford, U.A. Germann, K.P. Wilson, S.F.
Bellon, G. Chen, P. Taslimi, P. Jones, C. Huang, S. Pazhanisamy, Y. Wang, M.A. Murcko, M.S.S. Su, ACS Med. Chem. Lett. 2011, 2, 758 – 763.
6 C. Ding, Curr. Opin. Investig. Drugs 2006, 7, 1020 − 1025.
7 M.C. Genovese, Arthritis Rheum. 2009, 60, 317 – 320.
8 J.C.S. Arthur, S.C. Ley, Nat. Rev. Immunol. 2013, 13, 679 – 692.
9 J.G. Cumming, J.E. Debreczeni, F. Edfeldt, E. Evertsson, M. Harrison, G.A. Holdgate, M.J. James, S.G. Lamont, K. Oldham, J.E. Sullivan, S.J. Wells, J. Med.
Chem. 2015, 58, 278 − 293.
10 W. Davidson, L. Frego, G.W. Peet, R.R. Kroe, M.E. Labadia, S.M. Lukas, R.J. Snow, S. Jakes, C.A. Grygon, C. Pargellis, B.G. Werneburg, Biochemistry 2004,
43, 11658 – 11671.
11 J. Das, R.V. Moquin, A.J. Dyckman, T. Li, S. Pitt, R. Zhang, D.R. Shen, K.W. McIntyre, K. Gillooly, A.M. Doweyko, J.A. Newitt, J.S. Sack, H. Zhang, S. E.
Kiefer, K. Kish, M. McKinnon, J.C. Barrish, J.H. Dodd, G.L. Schieven, K. Leftheris, Bioorg. Med. Chem. Lett. 2010, 20, 6886 − 6889.
12 Small-Molecule Drug Discovery Suite 2014-2: Glide, version 6.3, Schrödinger, LLC, New York, NY, 2014.
13 Jaguar, version 8.4, Schrödinger, LLC, New York, NY, 2014.
14 F.M. Gurgis, W. Ziaziaris, L. Munoz, Mol. Pharmacol. 2014, 85, 345 – 356.
15 S. Kostenko, U. Moens, Cell. Mol. Life Sci. 2009, 66, 3289 – 3307.
16 K. Singh, M. Cassano, E. Planet, S. Sebastian, S.M. Jang, G. Sohi, H. Faralli, J. Choi, H.D. Youn, F.J. Dilworth, D. Trono, Genes Dev. 2015, 29, 513 – 525.
17 S.R. Selness, R.V. Devraj, B. Devadas, J.K. Walker, T.L. Boehm, R.C. Durley, H. Shieh, L. Xing, P.V. Rucker, K.D. Jerome, A.G. Benson, L.D. Marrufo, H.M.
Madsen, J. Hitchcock, T.J. Owen, L. Christie, M.A. Promo, M, B.S. Hickory, E. Alvira, W. Naing, R. Blevis-Bal, D. Messing, J. Yang, M.K. Mao, G. R. Yalamanchili,
R.V. Embse, J. Hirsch, J.; M. Saabye, S. Bonar, E. Webb, G. Anderson, J.B. Monahan, Bioorg. Med. Chem. Lett. 2011, 21, 4066 – 4071.
18 N.M. Aston, P. Bamborough, J.B. Buckton, C.D. Edwards, D.S. Holmes, K.L. Jones, V.K. Patel, P.A. Smee, D.O. Somers, G. Vitulli, A.L. Walker, J. Med.
Chem. 2009, 52, 6257 – 6269.
19 D.R. Anderson, M.J. Meyers, R.G. Kurumbail, N. Caspers, G.I. Poda, S.A. Long, B.S. Pierce, M.W. Mahoney, R.J. Mourey, M.D. Parikh, Bioorg. Med. Chem.
Lett. 2009, 19, 4882 – 4884.
12
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201900303
ChemMedChem
FULL PAPER
20 X. Huang, X. Zhu, X. Chen, W. Zhou, D. Xiao, S. Degrado, R. Aslanian, J. Fossetta, D. Lundell, D.; F. Tian, P. Trivedi, A. Palani, Bioorg. Med. Chem. Lett.
2012, 22, 65 – 70.
21 D. Xiao, A. Palani, X. Huang, M. Sofolarides, W. Zhou, X. Chen, R. Aslanian, Z. Guo, J. Fossetta, F. Tian, P. Trivedi, P. Spacciapoli, C.E. Whitehurst, D.
Lundell, Bioorg. Med. Chem. Lett. 2013, 23, 3262 – 3266.
22 X. Gong, X. Ming, P. Deng, Y. Jiang, J. Cell. Biochem. 2010, 110, 1420 – 1429.
23 R. Ben-Levy, S. Hooper, R. Wilson, H.F. Paterson, C.J. Marshall, Curr. Biol. 1998, 8, 1049 – 1057.
24 B.S. Hendriks, K.M. Seidl, J.R. Chabot, BMC Syst. Biol. 2010, 4, 1 – 17.
25 K. Soukup, A. Halfmann, M. Le Bras, E. Sahin, S. Vittori, F. Poyer, C. Schuh, R. Luger, B. Niederreiter, T. Haider, D. Stoiber, S. Blüml, G. Schabbauer, A.
Kotlyarov, M. Gaestel, T. Felzmann, A.M. Dohnal, J. Immunol. 2015, 195, 541 – 552.
26 C. Ehlting, M. Trilling, C. Tiedje, A. Zimmermann, V.T.K. Le, M. Gaestel, H. Hengel, D. Häussinger, J.G. Bode, Z. Gastroenterol. 2014, 52, 5 – 48.
27 C. Ehlting, N. Ronkina, O. Böhmer, U. Albrecht, K.A. Bode, K.S. Lang, A. Kotlyarov, D. Radzioch, M. Gaestel, D. Häussinger, J.G. Bode, J. Biol. Chem. 2011,
286, 24113 – 24124.
28 J. Fricke, P.L. Collins, L.Y. Koo, C.R. Brown, J. Virol. 2013, 87, 1333 – 47.
29 A.J. Guess, R. Ayoob, R.; Chanley, M.; Manley, J.; Agrawal, S.; Pengal, R.; Becknell, B.; Benndorf, R.; Smoyer, W.E.; Cajaiba, M.M.; Pyle, A.L.; Kopp, J.B.;
Ronkina, N.; Gaestel, M.; Benndorf, R.; Smoyer, W.E. PLoS One 2013, 8, e54239.
30 K.F. Mackenzie, M.J. Pattison, J.S.C. Arthur, Crit. Rev. Immunol. 2014, 34, 315 – 345.
31 C. Olaisen, R. Müller, A. Nedal, M. Otterlei, Cell. Signalling 2015, 27, 1478 – 1487.
13
This article is protected by copyright. All rights reserved.