Identification of 2-(4-pyridyl)thienopyridinones as GSK-3β inhibitors

Article abstract of DOI:10.1016/j.bmcl.2011.06.050

The discovery of a novel series of 2-(4-pyridyl)thienopyridinone GSK-3β inhibitors is reported. X-ray crystallography reveals its binding mode and enables rationalization of the SAR. The initial optimization of the template for improved cellular activity and predicted CNS penetration is also presented.

Full text of DOI:10.1016/j.bmcl.2011.06.050

Bioorganic & Medicinal Chemistry Letters 21 (2011) 4823–4827
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification of 2-(4-pyridyl)thienopyridinones as GSK-3b inhibitors
Gabriella Gentile a, , Giovanni Bernasconi a, , Alfonso Pozzan b, , Giancarlo Merlo a, Paola Marzorati a,
Paul Bamborough c, Benjamin Bax c, Angela Bridges c, Caroline Brough c, Paul Carter c, Geoffrey Cutler c,
Margarete Neu c, Mia Takada c
⇑
⇑
⇑
a Department of Medicinal Chemistry, Neuroscience Centre of Excellence of Drug Discovery, GlaxoSmithKline, Via A Fleming 4, 37135 Verona, Italy
b Computational and Structural Chemistry, GlaxoSmithKline, Via A Fleming 4, 37135 Verona, Italy
c Molecular Discovery Research, GlaxoSmithKline, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
The discovery of a novel series of 2-(4-pyridyl)thienopyridinone GSK-3b inhibitors is reported. X-ray
crystallography reveals its binding mode and enables rationalization of the SAR. The initial optimization
of the template for improved cellular activity and predicted CNS penetration is also presented.
Ó 2011 Elsevier Ltd. All rights reserved.
Received 3 May 2011
Revised 9 June 2011
Accepted 11 June 2011
Available online 29 June 2011
Keywords:
GSK-3 inhibitors
Glycogen synthase kinase-3 (GSK-3)
Lead optimization
Glycogen synthase kinase-3 (GSK-3) is
a
serine/threonine
part of an array targeted against protein kinases (Table 1). The
respectable potency of the series and its chemical novelty as a
kinase inhibitor scaffold made it an attractive template for further
study.
Compounds 1–5 showed moderate activity towards GSK-3b in
both the FP assay and a biochemical assay format.7 Interestingly,
some of the examples that were tested against the homologous ki-
nase CDK-2 showed moderate selectivity (e.g., 2, 3 and 5, Table 1).8
They were also tested in a cellular system measuring the phos-
phorylation of the GSK-3b substrate CRMP-2, showing weak but
encouraging weak activity.9
The initial data package also included full-curve inhibition
results against the other members of the GSK kinase screening pa-
nel. The selectivity of the thieno-pyridinone series is striking, mak-
ing them excellent starting points for optimization. For example,
compounds 2 and 4 show excellent selectivity against the panel
(Table 2).
Cross-screening of additional analogs within the initial set of
hits provided a valuable source of initial structure–activity rela-
tionships (SAR). For example, it was apparent that the pyridyl ring
was important for GSK-3b activity, as evidenced by the low activity
of the phenyl analogs 6–9 and the 3-pyridyl isomer 10 (FP
pIC50 <4.8). To rationalize this data, crystal structures were
obtained of three closely related compounds (Fig. 1) bound to
GSK-3b: 11 (2.4 Å), 12 (2.5 Å) and 13 (2.5 Å).10
kinase, which exists as two isoforms (
a and b) and is involved in
many cell functions.1 Its main role is the phosphorylation of glyco-
gen synthase (GS), therefore GSK-3 is implicated in type-2 diabe-
tes.2 In addition GSK-3 inhibition may be beneficial for the
treatment of neurodegenerative diseases, such as Alzheimers,3
and neurological diseases such as bipolar disorders.4 In particular
lithium is indicated as the preferential treatment for bipolar disor-
ders and the ability of this cation to inhibit GSK-3 has been pro-
posed as a potential therapeutic mechanism of action.5 Due to
the valuable therapeutic potential, identification of GSK-3 inhibi-
tors has become a focus of research for both academic centers
and pharmaceutical companies. The availability of crystal
structures of GSK-3b6 allows structure based lead optimization.
GlaxoSmithKline pursues kinase inhibitor discovery through a
‘‘systems-based’’ research strategy. One component of this is to
cross-screen compounds through a panel of kinase assays includ-
ing targets of interest and selectivity screens. A GSK-3b fluores-
cence polarization (FP) assay ran within this panel for several
years, so when novel GSK-3 inhibitors were sought the historical
screening data provided a rich source of potential lead compounds.
Among the molecules with GSK-3 activity were members of the
thieno-pyridinone series (e.g., 1–5), which had been prepared as
⇑
Corresponding authors. Tel.: +39 0458219355; fax: +39 0458218196 (G.G.,
G.B.); tel.: +39 3485926453; fax: +39 0458218196 (A.P.).
E-mail addresses: gabriella.gentile@aptuit.com, gabrygen@alice.it (G. Gentile),
gio.bernasconi@alice.it (G. Bernasconi), alfonso.pozzan@aptuit.com (A. Pozzan).
All three compounds showed the same binding mode within the
ATP-binding pocket. Two major interactions with the GSK-3b ATP
binding pocket were identified (Fig. 2) The first is a hydrogen bond
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.06.050

4824
G. Gentile et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4823–4827
Table 1
Enzymatic activity (GSK-3b), selectivity versus the homologous kinase (CDK-2)8 and cellular activity (CRMP-2) for hits 1–5
c
d
Compound
GSK-3b pIC50 FPa
GSK-3b pIC50 AlphaLISAb
CDK-2 pIC50
CRMP-2 pIC50
tPSA (Å2)11b
1
2
3
4
5
7.1
6.7
7.0
6.9
6.3
7.5
7.5
n.t.
7.3
7.2
n.t.
4.7
<4d
n.t.
4.8d
<4.0
5.8 (n = 16)
<4.5
4.8
5.9 (n = 4)
92
92
75
75
58
n.t. = not tested.
a
GSK-3b fluorescence polarization pIC50 (binding assay).7a
GSK-3b Alpha LISA pIC50 (enzymatic activity).7b
CDK-2 enzymatic activity assay.8
b
c
d
9
Cellular CRMP-2 Human Phosphorolation Inhibition pIC50
.
between the pyridyl nitrogen and the backbone NH atom of the
V135 residue at the hinge region, a conserved motif in kinase
inhibitors. However, the binding mode of this series is relatively
unusual in that the hinge makes no other hydrogen-bonding inter-
actions. The second hydrogen bond is formed between the pyridi-
none carbonyl moiety and two residual water molecules located at
the back of the pocket. The water molecules form a small network
and themselves hydrogen-bond to the backbone NH atoms of
D200, F201, and the acidic sidechain of E97. The crystal structure
is consistent with the SAR from the initial hits (Table 1), in which
disruption of the hinge hydrogen bond to V135 by removing or
moving the pyridyl nitrogen (6–10) leads to a large loss of potency.
The medicinal chemistry exploration of this series followed the
evidence highlighted by the crystallographic data. Firstly, the 4-
pyridyl hinge-binding group was exchanged for alternative N-con-
taining heterocyclic-2-substituents. The synthetic approach ap-
plied for this exploration is described in Scheme 1. Knoevenagel
condensation of 5-bromo-2-thiophenecarbaldehyde with malonic
acid afforded the alpha–beta unsaturated carboxylic acid interme-
diate that was then cyclized into the pyridone via an acyl azide for-
mation. Suzuki reaction was then applied to introduce the pyridine
ring in the C-2 position. Bromination of the resulting intermediate
gave the 7-bromo-thienopyridone derivative that was successfully
used for a large aryls and heteroaryls exploration in the C-7 posi-
tion, via Suzuki coupling. This synthetic route allowed the intro-
duction of heterocycles in the C-2 position and aryls and
heteroaryls in the C-7 position, proving to be very versatile for
an exhaustive SAR expansion.
Table 2
Kinase panel selectivity data
Compound
2
4
GSK-3b pIC50 (AlphaLISA)
Number of kinases tested
Number of kinases with pIC50 <6.0
Aurora A
7.5
51
50
7.3
45
42
6.0
5.9
O
O
O
N
N
N
N
N
N
O
S
S
Br
12
13
11
OH
Figure 1. From left to right, 11 (GSK-3b FP pIC50 = 6.9)b, 12 (GSK-3b AlphaLISA
pIC50 = 6.8)a, 13 (GSK-3b FP pIC50 = 6.9)a.
Compounds with replacement 2-hetercycles (Fig. 3) included
14, in which a methyl substituent alpha to the pyridine nitrogen
was introduced. Alternative N-containing heterocycles prece-
dented as kinase hinge-binding groups such as the 7-aza-indole
(15) and pyrazole (16) were also prepared. Potency and selectivity
of compounds 14–16 towards GSK-3b and CDK-2 are listed in Table
3 compared with baseline compounds 1 and 5.
This exploration confirmed that the unsubstituted 4-pyridine
was the preferred heterocycle and that the introduction of a
methyl group at the alpha position of the pyridine was not toler-
ated. This is consistent with the crystal structure (Fig. 2), which
predicts that the methyl group would sterically prevent the close
association between the pyridine and the hinge needed to form a
hydrogen bond. Introduction of alternative heterocycles in the C-
2 position (15 and 16) led to a slight increase in potency against
GSK-3b, consistent with the donation of additional hydrogen-
bonding interactions from the NH groups of these heterocycles to
the hinge region. However, the hydrogen-bond accepting groups
are highly conserved between most protein kinases, in view of
which it is not surprising that the selectivity profile of these com-
pounds worsened. For example, 16 gained appreciable CDK-2
activity compared to 5. Because of this these compounds were
not pursued for GSK-3b.
Figure 2. Superimposition of the three closely related compounds 11 (magenta), 12
(orange) and 13 (green), cocrystalized in the ATP binding pocket of the GSK-3b
enzyme. Dotted lines indicated hydrogen bonds with V135 (kinase hinge region)
and one of two water molecules situated in a deep interior pocket.

G. Gentile et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4823–4827
4825
O
(d)
S
(b)
(a)
(c)
N
N3OC
HOOC
OHC
Br
Br
Br
Br
S
S
S
S
O
N
O
O
N
(f)
S
(e)
N
N
N
N
S
Br
R
Scheme 1. Synthesis of thieno-pyridinone derivatives. Reagents and conditions: (a) malonic acid, piperidine, pyridine, reflux (41%); (b) DPPA, TEA, toluene, rt (70%); (c) Ph2O,
250 °C (10%); (d) 4-pyridinylboronic acid, Pd(PPh3)4, aq. Na2CO3, DMF, microwave irradiation, 160 °C, 30 min (86%); (e) NBS, DMSO, rt (60%); (f) Aryl(heteroaryl)boronic acid,
Pd(PPh3)4, aq. Na2CO3, DMF, microwave irradiation.
O
O
O
We next turned our attention to the second hydrogen-bonding
interaction identified from the crystal structures, and investigated
the role of the pyridinone amide moiety. N-Me and O–Me analogs
of the baseline compound 5 were synthesized (17 and 18, Fig. 4).
Results of these modifications are reported in Table 4. N-methyl
substitution proved to be detrimental for activity while the O–Me
N
N
N
N
N
N
N
S
S
S
N
O
O
S
O
O
S
N
N
N
O
16
14
15
Figure 3. Exploration of RHS heterocycles.
Table 3
Activity and selectivity data for compounds 14–16 compared to pyridine analog
compounds 1 and 5
Compound GSK-3b pIC50a CDK-2 pIC50
CRMP-2 pIC50
tPSA (Å2)11b
8
9
1
7.5
5.6d
7.7
7.3
7.9
n.t.
n.t.
5.5
4.8
6.6
<4.0
<4.0
<4.0
5.9
92
92
108
58
14
15
5
16
6.4
74
a
Figure 5. X-ray (yellow) versus docked (green) pose for 13. RMS = 0.56 Å.
GSK-3b AlphaLISA pIC50 (enzymatic activity), except 14 (GSK-3b FP binding
pIC50).7
O
O
O
N
N
N
N
N
N
S
S
S
N
N
N
O
O
O
17
5
18
Figure 4. N-methyl and O-methyl analogs of compound 5.
Table 4
Activity and selectivity data for compounds 17 and 18 compared to baseline
derivative 5
Compounds GSK-3b pIC50a CDK-2 pIC50
CRMP-2 pIC50
tPSA (Å2)11b
8
9
5
17
18
7.3
6.6
5.0
4.8
<4.8
<4.8
5.9
<4.0
<4.0
58
47
47
Figure 6. The space covered by the entire set of docked molecules (517) is shown in
blue. Red points represent the 27 molecules selected from 21 different clusters. Top
and bottom five actives are labeled according to Table 5.
a
GSK-3b AlphaLISA pIC50 (enzymatic activity).

4826
G. Gentile et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4823–4827
derivative turned out to be much less active, confirming the
requirement of the NH-pyridinone for activity. This can be ratio-
nalized from the crystal structure (Fig. 2), since these changes
would interfere with the formation of the water-mediated hydro-
gen bond by the pyridinone group.
The desired outcome of the final phase of the work reported
here was a GSK-3b inhibitor useful for the treatment of bipolar dis-
order. For this, CNS penetration would be essential. It is generally
accepted that compounds with polar surface area (tPSA) greater
than 60-70 Å2 are too polar to pass the blood–brain barrier.11 In
addition to CNS penetration, low tPSA has often been found to be
beneficial for cellular activity in various cases.12 Because GSK-3b
is an intracellular target, reduced tPSA would deliver ‘‘dual’’ posi-
tive effects for this series of molecules. The pyridinone group
makes a substantial contribution to the tPSA of 5 (58 Å2, compared
to 47 Å2 for 17 and 18) but, as outlined above, the initial explora-
tion suggested that modifications of the pyridinone group were
likely to be detrimental to GSK-3b activity. Regarding the hinge-
binding 4-pyridyl group, not only is this necessary for potency
and selectivity, but other hinge-binding groups able to replace it,
contain additional heteroatoms (e.g., 15 and 16). These have in-
creased tPSA (Table 3) with corresponding lower chance of CNS
permeability. Keeping a single hydrogen bonding with the very
conserved hinge binding region like in the case of the 4-pyridyl
group could also represent an advantage in terms of selectivity ver-
sus bidentate groups (e.g., 5 vs 16). Therefore, to exploit the attrac-
tive features of the 2-(4-pyridyl) thienopyridinone core, a structure
based array at the C-7 position was prepared. One major goal of
this was to modify the physico-chemical properties to yield mole-
cules with improved cellular permeability and with properties pre-
Table 5
Activity and selectivity data for the top 5 (19–23) and bottom 5 (24–28) compounds in the array ranked by GSK3-b activity
O
N
N
S
R
8
9
R
Top 5
GSK3-b pIC507a
CDK-2 pIC50
6.7
CRMP-2 pIC50
7.1
tPSA (Å2)11b
*
F
19
8.3
70
N
*
20
21
8.2
8.0
5.8
6.2
6.2
64
70
O
O
*
5.4a
N
*
N
22
23
8.0
7.9
<4.8
6.8
5.4
6.5
59
70
Cl
N
*
S
Bottom 5
*
24
25
7.0
6.9
5.3
5.2
5.4
5.5
55
68
O
*
N
O
*
O
26
6.8
5.3
5.3
55
F
O
*
27
28
6.5
6.5
<4.8a
4.9a
5.2
5.1
66
59
N
*
N
a
n P2. All the other values are mean of n = 1.

G. Gentile et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4823–4827
4827
dicted to permit CNS permeability (e.g., by lowering their tPSA as
outlined above). This approach also offered the ability to explore
a different region of the GSK-3b ATP binding pocket, towards the
exterior of the binding site and the sugar subpocket. This region
is less highly conserved than the hinge region, thereby affording
the prospect of increasing GSK-3b potency without adversely
affecting selectivity towards homologous kinases.
properties suitable for CNS permeability. These compounds repre-
sent a promising new lead series for developing GSK-3b inhibitors
for the treatment of neurodegenerative and neurological disorders
such as Alzheimer’s disease and bipolar disorder.
References and notes
1. Cohen, P.; Frame, S. Nat. Rev. Mol. Cell. Biol. 2001, 2, 769.
The array was prepared following the chemical route described
in Scheme 1. Boronic acids and boronates from the ACD catalogue13
and a collection of in-house reagents14 were selected to meet the
objectives outlined above, as follows. 2780 potential reagents were
enumerated in silico with the 2-(4-pyridine) thienopyridinone
using the Daylight reaction toolkit.15 MW, tPSA, calculated MCDK2
permeability and predicted Brain Tissue Binding (BTB)16 were com-
puted for the products. Following this step cut-offs of MW 6 00,
tPSA 675, calculated MCDK2 = high and BTB = low/medium were
applied. The rational for applying physchem and calculated ADME
properties were based to ensure on top of the CNS and cellular activ-
ity properties (as mentioned before), an acceptable developability
profile for future lead optimization efforts. These filters reduced
the initial set by 81% to 517 products with predicted CNS-permeable
characteristics. In order to characterize such molecules with respect
to the GSK-3b ATP binding site, the entire set containing the 517
compounds was docked into the 3D structure of the protein target
using the GOLD program.17 Preliminary docking experiments were
carried out on compound 13 to check if GOLD was able to reproduce
the X-Ray poses (Fig. 5). This was achieved by using GOLD standard
configurations and the Chemscore fitness function.17
Docking experiments were carried out on the entire GSK-3b X-
ray structure with the docking center defined by a list of 15 resi-
dues18,19 located around the binding pocket. As the primary goal
of this array was to expand the SAR on this series by exploring
the C-7 position of the 2-(4-pyridyl)thienopyridinone core, a deci-
sion was made to use two key components of the Chemscore fit-
ness function, Shbond and Slipo, corresponding to the scores for the
hydrogen bonding and liphophilic contribution respectively and
then select a smaller sub-set molecules covering as much as possi-
ble of the diversity defined by this two descriptors. Following this
strategy, 21 clusters were computed using a k-means clustering
method followed by a selection of 27 molecules with 21 coming
from each one of the different clusters20 (Fig. 6).
2. Nikoulina, S. E.; Ciararli, T. P.; Mudaliar, S.; Carter, L.; Jonson, K.; Henry, R. R.
Diabetes 2002, 51, 2190.
3. Planel, E.; Sun, X.; Takashima, A. Drug Dev. Res. 2002, 56, 491.
4. Phiel, C. J.; Klein, P. S. Annu. Rev. Pharmacol. Toxicol. 2001, 41, 789.
5. Klein, P. S.; Melton, D. A. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 8455.
6. (a) Aoki, M.; Iwamoto-Sugai, M.; Sugiura, I.; Sasaki, C.; Hasegawa, T.; Okumura,
C.; Sugio, S.; Kohono, T.; Matsuzaki, T. Acta Crystallogr. 2000, D56, 1464; (b) Bax,
B.; Carter, P. S.; Lewis, C.; Guy, A. R.; Bridges, A.; Tanner, R.; Pettman, G.;
Mannix, C.; Culbert, A. A.; Brown, M. J. B.; Smith, D. G.; Reith, A. D. Structure
(London) 2001, 9, 1143; (c) Bertrand, J. A.; Thieffine, S.; Vulpetti, A.; Cristiani, C.;
Valsasina, B.; Knapp, S.; Kalisz, H. M.; Flocco, M. J. Mol. Biol. 2003, 333, 393.
7. (a) The GSK-3b fluorescence polarisation assay was run using a human truncate
of GSK-3b expressed in baculovirus using
a proprietary GlaxoSmithKline
fluoroligand. Data was read on a Perkin Elmer Envision or Analyst and curves
analysed using IDBS ActivityBase software to produce an IC50 or pIC50 value.;
(b) The GSK-3b AlphaLISA assay used human full length His tagged GSK-3b
expressed in baculovirus. The assay was run utilising Perkin Elmer AlphaLISA
kits. Data was read on a Perkin Elmer Envision and curves analysed using IDBS
ActivityBase software to produce an IC50 or pIC50 value.
8. The CDK-2 33-P Leadseeker assay was performed using recombinant CDK-2
expressed in baculovirus. The assay used ATP (gamma-33P) from Sigma Aldrich
and Streptavidin coupled polystryrene SPA imaging beads (Leadseeker beads)
from GE Healthcare. Data was read on a Perkin Elmer Viewlux and curves
analysed using IDBS ActivityBase software to produce an IC50 or pIC50 value.
9. The CRMP-2 AlphaLISA assay was run using recombinant human CRMP-2
substrate, flag-tagged and over expressed in HEK-MSRII cells. The assay was
run using Perkin Elmer AlphaLISA kits. Data was read on a Perkin Elmer
Envision and curves analysed using IDBS ActivityBase software to produce an
IC50 or pIC50 value.
10. Truncated GSK-3b (27-393) was co expressed with FRATtide (197-226) by dual
baculovirus infection of Sf-9 cells (using a MOI of 3 for each virus), the protein
was purified using the method of Bax et.al., Structure (London) 2001, 9, 1143.
(see Ref. 6 for details).
Co-crystals of GSK-3b-FRATide with the compounds were grown at 20 °C using
the sitting drop method in 96-well Innovadyne SD-2 plates (‘MRC plates’),
using 80 lL of well solution and 120 or 100 nL of protein and 60 or 100 nL of
well solution (2 + 1 and 1 + 1 protein:well ratio), respectively. A 8 Â 12 grid
was designed to allow for the following range of conditions: 19–30% PEG3350,
0, 5% or 10% Glycerol, 0.1 M Buffer (BisTris pH 6.5–6.8, or Citrate pH 6.3–6.5),
0.2 M Ammonium Sulphate, and
0 or 0.1 M NaCl. The protein was pre-
incubated on ice for 1 h with 2 mM of compound (2% DMSO) and then
centrifuged at 13000g at 4 °C for 15 min before setting the Innovadyne tray
with the Mosquito liquid handling robot. Chunky crystals grew to full size
(100–400 lm diameter) within 1–2 weeks. Crystals were frozen in 30% PEG
Results from top and bottom five actives of this array are re-
ported in Table 5.
3350, 10% Glycerol, 0.1 M BisTris pH6.5 and 0.2 M Ammonium Sulphate,
containing 0.1 M compound (and 1% DMSO).
Data were collected from single frozen crystals on beam-line ID-23-1 at the
European synchrotron radiation facility (ESRF). Crystal structures of
compounds 11, 12 and 13 in complex with GSK-3b and Frattide were refined
from another structure determined in the same cell (from pdb code: 1gng). The
complex structures of 11 (2.37 Å), 12 (2.48 Å) and 13 (2.49 Å) with GSK-3 have
been deposited in the pdb with codes 3zrk, 3zrl, and 3zrm, respectively.
11. (a) Pajouhesh, H.; Lenz, G. R. NeuroRx. 2005, 2, 541; Topological polar surface
area (tPSA) was calculated by the method of (b) Ertl, P.; Rohde, B.; Selzer, P. J.
Med. Chem. 2000, 43, 3714.
The structure/property-based C-7 position expansion of the 2-
(4-pyridyl) thienopyridinone core led to the preparation of mole-
cules spanning almost 100-fold GSK3-b activity. Significantly, for
examples such as 19 and 20 it proved possible to reduce tPSA be-
low the threshold for CNS permeability while still maintaining
(ꢀsix-fold increase) the GSK-3b potency compared to the parent
molecules 1–5 (Table 1). For some compounds (e.g., 20–22), it also
proved possible to maintain a good window of selectivity against
CDK-2. Most encouragingly of all, cellular activity was generally
present on the whole series, with some compounds like 19 and
23 showing much improvement over 1–5. Taken together, the
overall improvement of the predicted CNS- and cell-permeability
properties while maintaining or improving the GSK-3b potency
demonstrated the effectiveness of the multi-objective design ap-
proach used to select the substituents. It also showed in an effi-
cient way the value of targeting the C7 position for modification,
providing useful molecules for future lead optimization efforts.
In conclusion, we have presented the discovery of a novel series
of potent inhibitors of GSK-3b from a kinase-targeted array. The
series is cell-permeable and generally quite selective. The binding
mode has been determined by X-ray crystallography and found to
be consistent with the SAR. In addition, examples have been pre-
pared that retain excellent activity with predicted physicochemical
12. Clark, D. E. Future Med. Chem. 2011, 3, 469.
13. Available Chemical Directory (ACD) database, 2009, www.symix.com.
14. An in-house ISIS based database of reagents purchased from various vendors
not included in standard reagents catalogues like ACD.
15. The Daylight toolkit is available from Daylight, 2009, www.daylight.com.
16. Calculated MCDK2 permeability and predicted Brain Tissue Binding models are
in-house models derived from internal GSK data. tPSA was computed using the
Ertl methods (a) Ertl, E.; Rohde, B.; Selzer, P. J. Med. Chem. 2000, 473, 3714
17. GOLD v. 3.2 (2007) is
www.ccdc.cam.ac.uk.
a docking program available from CCDC, http://
18. (a) The GOLD docking configuration was based on default settings. As scoring
function the CHEMSCORE was used, eight water molecules included in the
active side were allowed to toggle on/off as implemented in GOLD.; (b)
Verdonk, M. L.; Chessari, G.; Cole, J. C.; Hartshorn, M. J.; Murray, C. W.; Nissink,
J. W. M.; Taylor, R. D.; Taylor, R. J. Med. Chem. 2005, 48, 6504.
19. The amino acids defining the active site were: N64, D133, D200, C199, G63, I62,
L132, L188, K85, F67, Y134, V110, V135 and V70.
20. k-Means clustering was performed using the module available within Spotfire
Decision Site 8.2.1 (http:// spotfire.tibco.com/). The Euclidean distance was
used as similarity measure and the selection of the 27 molecules was based on
maximizing the number of clusters represented in the smaller set.