1-Aryl-3,4-dihydroisoquinoline inhibitors of JNK3

Article abstract of DOI:10.1016/j.bmcl.2009.02.098

A series of 1-aryl-3,4-dihydroisoquinoline inhibitors of JNK3 are described. Compounds 20 and 24 are the most potent inhibitors (pIC50 7.3 and 6.9, respectively in a radiometric filter binding assay), with 10- and 1000-fold selectivity over JNK2 and JNK1,

Full text of DOI:10.1016/j.bmcl.2009.02.098

Bioorganic & Medicinal Chemistry Letters 19 (2009) 2230–2234
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
1-Aryl-3,4-dihydroisoquinoline inhibitors of JNK3
b
b
b
a
John A. Christopher a, , Francis L. Atkinson , Benjamin D. Bax , Murray J. B. Brown , Aurélie C. Champigny ,
*
Tsu Tshen Chuang b, Emma J. Jones b, Julie E. Mosley b, James R. Musgrave b,
a GlaxoSmithKline R&D, Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, UK
b GlaxoSmithKline R&D, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of 1-aryl-3,4-dihydroisoquinoline inhibitors of JNK3 are described. Compounds 20 and 24 are the
most potent inhibitors (pIC50 7.3 and 6.9, respectively in a radiometric filter binding assay), with 10- and
1000-fold selectivity over JNK2 and JNK1, respectively, and selectivity within the wider mitogen-acti-
vated protein kinase (MAPK) family against p38a and ERK2. X-ray crystallography of 16 reveals a highly
unusual binding mode where an H-bond acceptor interaction with the hinge region is made by a chloro
Received 28 December 2008
Revised 24 February 2009
Accepted 25 February 2009
Available online 28 February 2009
substituent.
Keywords:
Kinase
Inhibitor
JNK
Ó 2009 Elsevier Ltd. All rights reserved.
JNK3
MAPK
Selective
Protein kinases catalyse the phosphorylation of tyrosine and
serine/threonine residues in proteins involved in the regulation
of diverse cellular functions. Aberrant kinase activity is implicated
in many diseases, which makes the inhibition of kinases an attrac-
tive target for the pharmaceutical industry.1
side effect risks and will aid the further understanding of the roles
of the individual JNK kinases. Only a few cases of inhibitors with
significant selectivity within the JNK family have been published,
recent examples being 1–3, Table 1.10–12 JNK inhibitors have been
recently reviewed.13
The mitogen-activated protein kinase (MAPK) pathways com-
prise a major signaling system used to transduce extracellular sig-
nals to intracellular responses. The MAPK family of serine-
threonine protein kinases consists of the extracellular-regulated
protein kinases (ERK), p38 mitogen-activated kinases (p38 MAPK)
and the c-Jun N-terminal kinases (JNKs).2
c-Jun N-terminal kinases are implicated in several disease areas,
including neurodegeneration, rheumatoid arthritis, inflammatory
disorders, cancer and diabetes.3,4 They differ in their tissue expres-
sion profile and functions, with JNK1 and JNK2 being widely ex-
pressed, whereas JNK3 is expressed predominantly in the brain
and at lower levels in the heart and testes.5,6 JNK3 appears to play
important roles in the brain to mediate neurodegeneration, such as
beta amyloid processing, Tau phosphorylation and neuronal apop-
tosis in Alzheimer’s Disease, as well as the mediation of neurotox-
icity in a rodent model of Parkinson’s Disease.7–9 Identifying potent
inhibitors of JNK3, with selectivity within the wider MAPK family,
may contribute towards neuroprotection therapies with reduced
MeO
N
MeO
4
Cl
JNK3 pIC50 5.0
A screening exercise to identify new JNK3 inhibitors identified
dihydroisoquinoline 4 as a weakly potent (pIC50 5.0) inhibitor.
Searching of the GlaxoSmithKline corporate collection identi-
fied numerous closely related analogues. A selection was screened
in JNK3 and p38
a fluorescence anisotropy kinase binding assays;
data are summarised in Table 2.14,15
This initial iteration of analog screening revealed several inter-
esting aspects to the series. Firstly, numerous changes to R1 with
methoxy R6 and R7 groups failed to increase JNK3 potency; at best
the moderate activity of the initial hit 4 was retained. Changes to
the R6 and R7 positions had a moderate impact; compounds 11
and 13 where the 7-position methoxy group was replaced by
chloro yielded slight increases in JNK3 inhibitory activity over
the initial hit, albeit with different R1 substituents to 4. Interest-
ingly, the analogues where the positions of the methoxy and chloro
* Corresponding author. Tel.: +44 0 1438 763578.
E-mail address: john@christopher257.fsnet.co.uk (J.A. Christopher).
Present address: Medimmune Cambridge, Milstein Building, Granta Park, Cam-
bridge, CB21 6GH, UK.
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.02.098

J. A. Christopher et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2230–2234
2231
Table 1
Selective JNK inhibitors 1–3, values in pIC50
Cl
H
H
N
H
N
N
N
N
N
O
1
2
H
N
H
N
O
O
N
H
CN
H
N
3
S
N
O
O
Compound
JNK1
JNK2
JNK3
p38a
Erk-2
110
211
312
7.1
6.4
<5.0
NR
NR
6.5
8.9
8.2
6.6
8.4
6.7
<4.8
NR
NR
<5.0
NR: Not reported.
Figure 1. Compound 11 docked into the active site of JNK3.16
substituents are interchanged were inactive at JNK3 (compare 11
with 12, 13 with 14). No inhibition of p38 was observed.
a
Encouraged by the moderate JNK3 activity and the initial indi-
cations of selectivity within the MAPK family, representative mol-
ecules such as 11 were docked into the active site of JNK3, Figure
1.16 Analysis of the dockings suggested the conserved hydrogen-
bond from the hinge backbone amide (Met149 in JNK3) was ac-
cepted by the dihydroisoquinoline nitrogen, with the R1 aromatic
substituent occupying the outer lipophilic region.19 This motif is
reminiscent of that seen in the crystal structure of close analogues
of 1, where the conserved H-bond is accepted by the indazole N2.10
By contrast, in the dihydroisoquinoline dockings the inner H-bond
to the backbone carbonyl (Glu147 in JNK3) is lost, whereas in 1 it is
donated by the indazole N1. However, this donor interaction,
whilst common, is by no means entirely conserved, and the ab-
sence was not taken to contradict the binding mode hypothesis.
The substituents at R6 and R7 make few interactions with the site,
which suggested that there might be significant opportunities for
improving this template.
within the wider JNK family, with the aim of improving the JNK3
potency and gaining a clearer understanding of the structure–
activity relationships. Screening of further analogs of the series
from the GlaxoSmithKline corporate collection was complemented
by synthesis of key molecules, focusing on the R6 and R7 positions.
The synthetic route was straightforward; appropriately substi-
tuted phenylethylamines 15 were acylated and then cyclised with
phosphorous oxychloride, as depicted in Scheme 1. Data for repre-
sentative compounds from this exercise are summarized in Table
3. The most active JNK3 inhibitors were those bearing methoxy
substitution at R6 and chloro substitution at R7, analogues where
these two substituents were interchanged were at best weakly ac-
tive at JNK3 (compare 16, 20, and 24 with 17, 21 and 9 (Table 2),
respectively). Gratifyingly, selectivity over p38a was maintained.
Mono- or di-substituted phenyl groups at R1 were preferred for
JNK3, amongst the most active being meta-bromo or chloro com-
pounds 16 and 24. Comparison of these with the unsubstituted
analogue 22 or the meta-fluoro variant 11 indicated that the bulk
and lipophilicity imparted by the larger halogens provided favour-
able interactions in the JNK3 ATP binding active site. Other R1
groups, for example benzyl or alkyl, (27, 28) were not tolerated.
JNK1, JNK2 and JNK3 inhibitory activities of key compounds 20
and 24 were assessed in analogous radiometric filter binding as-
says, Table 4.20 Erk-2 data were also obtained, which together with
With this binding hypothesis in hand, we initiated an explora-
tion of the template to understand the profile of the molecules
Table 2
JNK3 and p38
a inhibition (fluorescence anisotropy kinase binding assay) of
compounds 5–14, values in pIC50
R6
R7
the p38
a data already in hand from a fluorescence anisotropy
N
R1
binding assay provides a full picture of selectivity within the wider
MAPK family.20 The data revealed a highly encouraging selectivity
profile; 20 and 24 have approximately 10 and 1000-fold selectivity
for JNK3 over JNK2 and JNK1, respectively. The JNK family selectiv-
Compound
R1
R6
R7
JNK3
p38a
5
6
7
8
9
10
11
12
13
14
3,4-Cl2-Phenyl
Phenyl
OMe
OMe
OMe
OMe
Cl
H
OMe
Cl
OMe
OMe
OMe
OMe
OMe
H
Cl
OMe
Cl
5.2
<4.8
<4.8a
<4.8
<4.8
<4.8a
<4.8
<4.8
<4.8a
<4.8
<4.8a
<4.8
<4.8
<4.8
4.9
<4.8
5.4
<4.8
5.3
<4.8
4-Cl-Phenyl
3-MeO-Phenyl
3-Cl-Phenyl
3-Cl-Phenyl
3-F-Phenyl
3-F-Phenyl
4-F-Phenyl
4-F-Phenyl
R6
R7
R6
R7
(a), (b)
NH2
N
R1
15
OMe
Cl
Scheme 1. Synthesis of dihydroisoquinolines. Reagents and conditions: (a) R1-
COCl, CH2Cl2, 5% (aq) NaOH, rt; (b) POCl3, CH2Cl2, reflux, or microwave heating,
120 °C.
OMe
TR-FRET data.15
a

2232
J. A. Christopher et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2230–2234
Table 3
JNK3 and p38
a inhibition (fluorescence anisotropy kinase binding assay) of
compounds 16–28, values in pIC50
R6
R7
N
R1
Compound
R1
R6
R7
JNK3
p38a
16
17
18
19
20
21
22
23
24
25
26
27
28
3-Br-Phenyl
3-Br-Phenyl
2-Napthyl
2-Napthyl
3,4-Cl2Phenyl
3,4-Cl2Phenyl
Phenyl
Phenyl
3-Cl-Phenyl
3-Me-Phenyl
OMe
Cl
OMe
Cl
OMe
Cl
OMe
Cl
OMe
OMe
OMe
OMe
OMe
Cl
OMe
Cl
OMe
Cl
OMe
Cl
OMe
Cl
Cl
Cl
Cl
6.4
5.0
6.0
<4.8
6.5
4.8
5.2
<4.8
6.6
5.7
4.9
<4.8a
<4.8
<4.8a
5.1
<4.8a
<4.8
<4.8a
5.1
<4.8
4.9
2-F,3-Cl-Phenyl
-CH2Phenyl
Ethyl
6.2
<4.8
<4.8
<4.8
<4.8
Cl
a
TR-FRET data.15
Table 4
JNK1, JNK2, JNK3, Erk-2 and p38
Figure 2. Crystal structure of compound 16 in the JNK3 active site.21
a
inhibition of compounds 20 and 24, values in pIC50
Compound
JNK1a
JNK2a
JNK3a
Erk-2a
p38ab
20
24
4.0
<4.0
6.1
5.9
7.3
6.9
<4.0
<4.0
5.1
5.1
A second interaction made by the chlorine is a contact (3.4 Å)
with the backbone carbonyl of Glu147; this ‘inner’ carbonyl ac-
cepts an H-bond from ATP and is commonly utilized by kinase
inhibitors.17,19 Attractive interactions between halogens, here act-
ing as a Lewis acid, and Lewis bases such as oxygen are termed
‘halogen bonds’; they are less well known than hydrogen bonds,
but have attracted considerable attention in recent years. Small-
molecule crystal-structure database surveys, ab initio calculations
and gas-phase studies have shown beyond doubt that such interac-
tions can be attractive and of non-negligible strength.28–32 In addi-
tion, recent studies have shown that such interactions occur in
protein–ligand crystal structures.32,33
a
Radiometric filter binding assay.20
Fluorescence anisotropy kinase binding assay.15
b
ity is complemented by selectivity within the wider MAPK family
over Erk-2 and p38
respectively.
The MAPK selectivity of the template is matched by an encour-
aging wider selectivity profile. For example 20 and 24 were inac-
tive (pIC50 < 5.0) at numerous kinases, including B-Raf, c-Fms,
a, approximately 1000- and 100-fold,
CDK-2, EGFR, ErbB2, GSK3b, IKK-a, IKK-b, Lck, MLK3, PLK1, Rock1
and SGK1.
In the present structure the C–ClÁ Á ÁO angle is 163°, which is
consistent with the literature data showing that, in a halogen bond,
a linear geometry is preferred about the halogen atom.28–30 Calcu-
lations show that the electron density around the halogen is aniso-
tropic, meaning the halogen atom has a ‘crown’ of positive
electrostatic potential in the direction of the C–Hal bond, sur-
rounded by a band of negative electrostatic potential.24,32,34,35
The positive crown is responsible for ‘head-on’ interactions with
electron donors (i.e., halogen bonds), the negative band for ‘side-
on’ interactions with proton donors (i.e., hydrogen bonds).22,23,25,36
The polar interactions of the chlorine in this structure illustrate
this ‘amphiphilic’ nature of halogens, a property that has also been
observed in molecular crystals.24,37
Intrigued by the clear SAR trends that had emerged, we sought
to obtain a ligand-bound crystal structure to aid our understand-
ing. A crystal structure of compound 16 complexed with JNK3
was obtained to 2.4 Å resolution, which revealed a binding mode
very different to that expected from the docking experiments
(Fig. 2).21
Firstly, contact with the hinge is made by the chloro(metho-
xy)benzene moiety, not by the dihydroisoquinoline nitrogen. Sec-
ondly, the R1 aromatic substituent occupies an ‘induced-fit’ hydro-
phobic pocket at the rear of the active site, not the outer lipophilic
region.
The conserved hydrogen-bond with the hinge donor (the back-
bone NH of Met149) appears to be made by the R7 chlorine atom.19
Such hydrogen bonds with organic halogens as acceptors are rare:
however, recent studies utilizing small-molecule crystal-structure
database surveys or ab initio calculations have shown that such
interactions are genuine, if weaker than ‘normal’ H-bonds to oxy-
gen or nitrogen acceptors.22–24 These interactions have not yet
been systematically studied in protein–ligand crystal structures,
but have been shown to occur in the PDB.25 The ClÁ Á ÁH distance
here is 2.7 Å, the ClÁ Á ÁH–N angle 127° and the HÁ Á ÁCl–C angle
113°.26 These parameters are consistent with those observed in
the small-molecule database studies, placing the interaction to-
wards the weaker end of the scale in the classification of
Jeffrey.22,23,27
The C@OÁ Á ÁCl angle observed here is 139°, and inspection shows
that the chlorine lies close to the plane of the carbonyl. Both these
observations are consistent with the interaction being with an oxy-
gen lone pair; it should be noted that this preference is not neces-
sarily very strong.28
Whilst the polar interactions of the R7 chloro substituent with
the hinge region described above are interesting, the question of
how much they actually contribute to binding remains open. It is
likely that binding is driven primarily by contacts with hydropho-
bic residues (discussed further below), and it might be that the
contribution of the polar interactions made by the chlorine to the
affinity is minor. However, a precedent for the halogen-mediated
interaction of 16 with the hinge can be seen in the complex of
4,5,6,7-tetrabromobenzotriazole 29 with CDK2 (PDB 1P5E).36 Here,

J. A. Christopher et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2230–2234
2233
the bromine in the 5-position makes a halogen bond with the car-
bonyl oxygen of Glu81, which corresponds to Glu147 of JNK3. This
bromine also contacts the conserved donor residue in CDK2 (the
backbone NH of Leu83); although the geometry is poor, this might
constitute a weak H-bond, similar to that seen with 16.
the fact that this residue is peripheral, and thus would not be ex-
pected to significantly impact inhibitor binding.
The residues that comprise the induced-fit pocket in JNK3 are
largely conserved in p38
keeper, which is Thr106 in p38
tion of the structures of JNK3 with 16 and of comparable p38
a
, a notable exception being the gate-
a
. Despite these similarities, inspec-
a
Br
crystal structures suggests that there are considerable differences
in the shape of the pocket, due to, for example, differences in do-
main orientation.38 This, alongside the differing gatekeeper and
Br
Br
N
29
N
N
H
differences in the hinge region, means the lack of p38
can be easily understood.
a activity
Br
Replacement of the R7 chlorine of 16 with bromine, fluorine or
other groups capable of accepting a hydrogen bond are outside the
scope of this communication, but would serve to further probe the
importance for inhibition of this substituent.
The gatekeeper residue in Erk-2 is Gln103, which does not un-
dergo the rearrangement observed for Met146 in JNK3, and this ki-
nase is believed to be relatively insensitive to back-pocket binding
ligands.39 The lack of Erk-2 inhibitory activity of 20 and 24 was
thus expected.
In summary, 1-aryl-3,4-dihydroisoquinolines such as 20 and 24
represent a potent series of JNK3 inhibitors (pIC50 7.3 and 6.9,
respectively in a radiometric filter binding assay), with 10- and
1000-fold selectivity over JNK2 and JNK1, respectively, and an
encouraging wider kinase selectivity profile. The highly unusual
binding mode where the conserved H-bond acceptor interaction
with the hinge region is made by the R7 chloro substituent is a
key feature of the series, and with this structural information in
hand, optimization of the R1 and R6 substituents is a focus for fu-
ture investigations.
The methoxy substituent at R6 impinges on the outer lipophilic
region, and makes contacts with hydrophobic residues there (e.g.,
Ile70, Leu148 and Ala151).19 Further, it might modulate the elec-
tron density on the chlorobenzene moiety, and thus affect the polar
interactions discussed above. However, the lack of SAR around this
position means these effects require further study.
The isoquinoline moiety occupies the adenine-binding region,
and makes extensive contacts with the hydrophobic residues sur-
rounding it (e.g., Ile70, Val78, Ala91, Met 146, Val196, and
Leu206).19 Additionally, the isoquinoline nitrogen accepts a hydro-
gen-bond from a water molecule that is also H-bonded to the con-
served Lysine (Lys93).
The R1 substituent of 16 occupies a lipophilic pocket at the back
of the active site that is exposed by the rearrangement of the gate-
keeper residue Met146 and is formed mainly by the sidechains of
Met146, Ala91, Leu144, Ile124, Leu206 and Lys93 (the lipophilic
portion thereof).10–12
In addition to these non-polar contacts, the bromine atom of 16
is in contact with the backbone carbonyl oxygens of Leu144 (3.1 Å)
and Ala91 (3.3 Å). These interactions appear to be further examples
of the halogen bond; in contrast to the situation described above,
however, the halogen in both these cases lies approximately per-
pendicular to the carbonyl plane, suggesting that the interaction
Acknowledgments
Yvonne Joseph, Jeffery Smith, Laurie Gordon and the GSK
Screening and Compound Profiling group are gratefully acknowl-
edged for the generation of kinase inhibition data. Synthetic contri-
butions from K. Franzmann, F. Copp, S. Swanson and C. Foster are
gratefully acknowledged.
References and notes
1. Parang, K.; Sun, G. Curr. Opin. Drug Disc. Dev. 2004, 7, 617.
2. Herlaar, E.; Brown, Z. Mol. Med. Today 1999, 5, 439.
3. Manning, A. M.; Davis, R. J. Nat. Rev. Drug Disc. 2003, 2, 554.
4. Liu, G.; Rondinone, C. M. Curr. Opin. Invest. Drugs 2005, 6, 979.
5. Kuan, C.-Y.; Yang, D. D.; Roy, D. R. S.; Davis, R. J.; Rakic, P.; Flavell, R. A. Neuron
1999, 22, 667.
6. Martin, J. H.; Mohit, A. A.; Miller, C. A. Mol. Brain Res. 1996, 35, 47.
7. Resnick, L.; Fennell, M. Drug Discovery Today 2004, 9, 932.
8. Kimberley, W. T.; Zheng, J. B.; Town, T.; Flavell, R. A.; Selkoe, D. J. J. Neurosci.
2005, 25, 5533.
9. Pan, J.; Wang, G.; Yang, H.-Q.; Hong, Z.; Xiao, Q.; Ren, R.-J.; Zhou, H.-Y.; Bai, L.;
Chen, S.-D. Mol. Pharmacol. 2007, 72, 1607.
10. Swahn, B.-M.; Huerta, F.; Kallin, E.; Malmström, J.; Weigelt, T.; Viklund, J.;
Womack, P.; Xue, Y.; Öhberg, L. Bioorg. Med. Chem. Lett. 2005, 15, 5095.
11. Swahn, B.-M.; Xue, Y.; Arzel, E.; Kallin, E.; Magnus, A.; Plobeck, N.; Viklund, J.
Bioorg. Med. Chem. Lett. 2006, 16, 1397.
12. Angell, R. M.; Atkinson, F. L.; Brown, M. J.; Chuang, T. T.; Christopher, J. A.;
Cichy-Knight, M.; Dunn, A. K.; Hightower, K. E.; Malkakorpi, S.; Musgrave, J. R.;
Neu, M.; Rowland, P.; Shea, R. L.; Smith, J. L.; Somers, D. O.; Thomas, S. A.;
Thompson, G.; Wang, R. Bioorg. Med. Chem. Lett. 2007, 17, 1296.
13. LoGrasso, P.; Kamenecka, T. Mini-Rev. Med. Chem. 2008, 8, 755.
14. pIC50 = ÀLog10 IC50; where the IC50 is the concentration of compound required
to inhibit the kinase activity by 50%. JNK3 inhibitory activity was determined
is with the p-electrons of the peptide bond instead of with an oxy-
gen lone pair.32
The importance of these halogen bonds can perhaps be gauged
by comparison of the JNK3 potency of compound 16 with that of 25
(Table 3). In the latter the bromine is replaced by a methyl group,
and the potency is reduced by ꢀ0.7 log units. In addition, the
chloro-analogue of 16 (compound 24), which can also presumably
make the halogen bonds seen in the crystal structure, shows activ-
ity almost a log unit better than 25. If the role of the halogen or
methyl in these compounds was purely to enhance the steric com-
plementarity of the R1 substituent to the hydrophobic pocket then
this difference might not be expected.
The induced-fit pocket contains the only active-site residue
that differs between JNK3 and JNK1, Leu144 in the former corre-
sponding to Ile106 in the latter. In common with the rationale
postulated for 3 we believe that the significant selectivity exhib-
ited by this series for JNK3 over JNK1 could result from the dif-
ference in the shape of the hydrophobic pocket introduced by
this change.12
This selectivity rationale assumes that the binding mode ob-
served in JNK3 is conserved in JNK1. Despite the relatively low
affinity of these compounds for JNK1 (Table 4), it proved possible
to obtain protein–ligand crystal structures of 24 and 26 complexed
with JNK1, which both showed the same binding mode as 16 in
JNK3.21
Only one residue differs in the active sites of JNK3 and JNK2;
Met115 in the former corresponding to Leu77 in the latter. The ser-
ies has only moderate selectivity for JNK3 over JNK2, indicative of
using
a fluorescence anisotropy kinase binding assay. The kinase, a
fluorescently labeled inhibitor and a variable concentration of test compound
are incubated together to reach thermodynamic equilibrium under conditions
such that in the absence of test compound the fluorescent inhibitor is
significantly (>50%) enzyme bound and in the presence of
a sufficient
concentration (>10 Â Ki, where Ki = dissociation constant for inhibitor
binding) of a potent inhibitor the anisotropy of the unbound fluorescent
ligand is measurably different from the bound value. Truncated human JNK3
was expressed in baculovirus as an N-terminal His (6)-tagged fusion protein.
This enzyme (JNK3) was activated in 50 mM Tris/HCl pH 7.5, 0.1 mM EGTA,
0.1% beta-mercaptoethanol, 0.1 mM sodium vanadate, 10 mM magnesium
acetate, 0.1 mM ATP with 100 nM active MKK4 and MKK7beta at 30 °C for
30 min. Following activation, the JNK3 is purified by Ni-NTA agarose
chromatography. The JNK3 was then dialyzed into storage buffer (50 mM
Tris/HCl pH 7.5, 270 mM Sucrose, 150 mM NaCl, 0.1 mM EGTA, 0.1% beta-

2234
J. A. Christopher et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2230–2234
mercaptoethanol, 0.03% Brij-35, 1 mM benzamidine, 0.2 mM PMSF), snap
frozen in liquid nitrogen and stored at À70 °C. Inhibitor binding to JNK3 was
assessed by fluorescence anisotropy competitive binding assay. All
components are dissolved in buffer of composition 50 mM HEPES, pH 7.5,
1 mM CHAPS, 1 mM DTT, 10 mM MgCl2 with final concentrations of 10 nM
JNK3 and 2 nM of a fluorescently labeled inhibitor. This reaction mixture is
added to wells containing various concentrations of test compound (0.28 nM–
binding assay using N-terminal GST-tagged full-length human Erk-2, activated
with MEK1, with myelin basic protein as substrate, in the presence of 155
ATP.
lM
a
21. Truncated JNK3 (Jnk3t; residues 39–402)14 was purified following a five stage
process after lysis of E. coli cells expressing GST-JNK3t. [Glutathione Sepharose
(GSH), Thrombin cleavage, GSH, Source 15-Q anion exchange, SEC]. The protein
was supplied in 50 mM Tris/HCl pH 8.0, 150 mM NaCl post final stage Superdex
200 prep grade Size Exclusion column. Co-crystals of Jnk3t with 16 were grown
at 20 °C using the hanging drop method combined with micro-seeding. Protein
at 13 mg/mL, pre-incubated with 5 mM compound, was mixed with serial
dilutions of a Jnk3t seed stock (made in 25% peg 3350, 0.1 M sodium Hepes
pH7.5). Drops were then equilibrated over a reservoir containing 18% peg3350,
0.1 M sodium Hepes pH7.5) before freezing in mother liquor plus 15% glycerol.
A 2.4 Å dataset was collected from a single frozen crystal of Jnk3t/16 on a
Mar345 detector mounted on a micromax 007HF rotating anode generator. The
crystal structure was refined starting from the coordinates of another Jnk3t
complex crystal stucture (pdb code: 2O0U—ligand removed before start of
refinement). The deposition code for the Jnk3t/16 complex is 2waj. Crystal
structures of 24 and 26 with a similarly truncated version of Jnk1 are virtually
identical (Bax et al., unpublished results).
16.6 lM final) or DMSO vehicle (<3% final) in black 384-well microtitre plates
and equilibrated for 30–300 min at room temperature to reach equilibrium.
Fluorescence anisotropy is read in Molecular Devices Acquest (excitation
530 nm/emission 580 nm). The error of the assay is estimated as 0.2 log units,
based on the median standard deviation of all compounds which have been
tested more than eight times and have a mean pIC50 > 6.
15. p38a Fluorescence anisotropy kinase binding assay data were generated using
conditions previously described, see: Barker, M. D.; Hamblin, J. N.; Jones, K. L.;
Patel, V. K.; Swanson, S.; Walker, A. L. Int. Patent Appl. WO 05/073232.
Alternatively, where indicated, p38
time-resolved fluorescence resonance energy transfer (TR-FRET) assay:
Recombinant His-tagged human p38 was activated using 3 M unactivated
p38 incubated in 200 mM Hepes pH 7.4, 625 mM NaCl; 1 mM DTT with
a inhibitory activity was determined using
a
a
l
a
27 nM active MKK6 (Upstate). Biotinylated GST-ATF2 (residues 19–96, 400 nM
final), ATP (125 mM final) and MgCl2 (5 mM final) in assay buffer (40 mM
22. Brammer, L.; Bruton, E. A.; Sherwood, P. Cryst. Growth Des. 2001, 1, 277.
23. Sony, S. M. M.; Ponnuswamy, M. N. Bull. Chem. Soc. Jpn. 2006, 79, 1766.
24. Lu, Y.-X.; Zou, J.-W.; Wang, Y.-H.; Zhang, H.-X.; Yu, Q.-S.; Jiang, Y.-J. J. Mol.
Struct.: THEOCHEM 2006, 766, 119.
25. Panigrahi, S. K.; Desiraju, G. R. Proteins: Struct. Funct. Bioinform. 2007, 67, 128.
26. Note that the hydrogen atoms were not resolved in this structure: where
necessary for analysis of the H-bond geometries they were added in standard
positions using Maestro (Schrödinger Inc.).
27. An Introduction to Hydrogen Bonding; Jeffrey, G. A., Ed.; Oxford University Press:
New York, 1997.
28. Lommerse, J. P. M.; Stone, A. J.; Taylor, R.; Allen, F. H. J. Am. Chem. Soc. 1996, 118,
3108.
HEPES pH 7.4, 1 mM DTT) were added to wells containing 1
concentrations of compound or DMSO vehicle (3% final) in NUNC 384-well
black plates. The reaction was initiated by addition of p38 (100 pM final) to
give a total volume of 30 l. After 120 min incubation (rt), 15 l of 100 mM
EDTA pH 7.4 was added followed by detection reagent (15 l) in buffer
(100 mM HEPES pH 7.4, 150 mM NaCl, 0.1% w/v/BSA, 1 mM DTT) containing
antiphosphothreonine-ATF2-71 polyclonal antibody (Cell Signalling
ll of various
a
l
l
l
Technology, Beverly Massachusetts, MA) labelled with W-1024 Eu chelate
(Wallac OY, Turku, Finland), and APC-labelled streptavidin (Prozyme, San
Leandro, CA). After 60 min further incubation (rt) the ATF-2 phosphorylation
was measured using
Pangbourne, UK) as
a
Packard Discovery plate reader (Perkin–Elmer,
29. Ouvrard, C.; Le Questel, J.-Y.; Berthelot, M.; Laurence, C. Acta Crystallogr., Sect. B
2003, B59, 512.
a
ratio of specific 665 nm energy transfer signal to
reference Eu 620 nm signal.
30. Legon, A. C. Angew. Chem., Int. Ed. 1999, 38, 2686.
16. The public-domain JNK3 crystal structure (PDB accession code 1JNK) was used
31. Riley, K. E.; Merz, K. M. J. Phys. Chem. A 2007, 111, 1688.
32. Auffinger, P.; Hays, F. A.; Westhof, E.; Ho, P. S. Proc. Natl. Acad. Sci. U.S.A. 2004,
101, 16789.
for docking.17 ATP and water molecules were removed from the PDB file and
18
ligands were docked into the active site using GOLD
.
17. Xie, X.; Gu, Y.; Fox, T.; Coll, J. T.; Fleming, M. A.; Markland, W.; Caron, P. R.;
Wilson, K. P.; Su, M. S.-S. Structure 1998, 6, 983.
33. Voth, A. R.; Ho, P. S. Curr. Top. Med. Chem. 2007, 7, 1336.
34. Bosch, E.; Barnes, C. L. Cryst. Growth Des. 2002, 2, 299.
18. Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor, R. J. Mol. Biol. 1997, 267,
727.
35. Clark, T.; Hennemann, M.; Murray, J. S.; Politzer, P. J. Mol. Model. 2007, 13, 291.
36. PDB accession code 1P5E, see: De Moliner, E.; Brown, N. R.; Johnson, L. N. Eur. J.
Biochem. 2003, 270, 3174.
37. Lieberman, H. F.; Davey, R. J.; Newsham, D. M. T. Chem. Mater. 2000, 12, 490.
38. Scapin, G.; Patel, S. B.; Lisnock, J.; Becker, J. W.; LoGrasso, P. V. Chem. Biol. 2003,
10, 705.
39. Wang, Z.; Canagarajah, B. J.; Boehm, J. C.; Kassisà, S.; Cobb, M. H.; Young,
P. R.; Abdel-Meguid, S.; Adams, J. L.; Goldsmith, E. J. Structure 1998, 6,
1117.
19. Adams, J. L.; Veal, J.; Shewchuk, L. In Protein Crystallography in Drug Discovery;
Babine, R. E., Abdel-Meguid, S. S., Eds.; Wiley-VCH: Weinheim, 2004. Chapter 2.
20. Data in radiometric filter binding assays were obtained from Millipore using
the IC50 Profiler ExpressTM service, for further details see: www.millipore.com.
JNK data were obtained using N-terminal His-tagged full-length human
JNK1a1, JNK2a2 or JNK3 with ATF2 as substrate, in the presence of 45, 45, or
10 lM ATP, respectively. Erk-2 data were obtained in a radiometric filter