3-Anilino-4-arylmaleimides: Potent and selective inhibitors of glycogen synthase kinase-3 (GSK-3)

Article abstract of DOI:10.1016/S0960-894X(00)00721-6

Potent 3-anilino-4-arylmaleimide glycogen synthase kinase-3 (GSK-3) inhibitors have been prepared using automated array methodology. A number of these are highly selective, having little inhibitory potency against more than 20 other protein kinases.

Full text of DOI:10.1016/S0960-894X(00)00721-6

Bioorganic & Medicinal Chemistry Letters 11 (2001) 635±639
3-Anilino-4-arylmaleimides: Potent and Selective Inhibitors of
Glycogen Synthase Kinase-3 (GSK-3)
David G. Smith,* Marianne Bu€et, Ashley E. Fenwick, David Haigh, Robert J. Ife,
Martin Saunders, Brian P. Slingsby, Rachel Stacey and Robert W. Ward
SmithKline Beecham Pharmaceuticals, New Frontiers Science Park, Third Avenue, Harlow, Essex, CM19 5AW, UK
Received 3 November 2000; revised 7 December 2000; accepted 18 December 2000
AbstractÐPotent 3-anilino-4-arylmaleimide glycogen synthase kinase-3 (GSK-3) inhibitors have been prepared using automated
array methodology. A number of these are highly selective, having little inhibitory potency against more than 20 other protein
kinases. # 2001 Elsevier Science Ltd. All rights reserved.
Glycogen synthase kinase-3 (GSK-3), so called because
it was one of originally ®ve kinases known to phos-
phorylate glycogen synthase (GS),1 is a serine/threonine
kinase which exists as two isoforms (a and b) with
molecular weights of 51 and 46 kDa, respectively.2 The
catalytic domains have ꢀ90% identity but GSK-3a has
an additional ꢀ60 amino acid residues N-terminal to
the kinase domain which is glycine rich. (Recently a
third isoform of MW 53 kDa has been discovered which
may be a GSK-3a variant formed by post-translational
processing.3) Di€erences in function between the iso-
forms have yet to be established. GSK-3 plays a key role
in a number of diverse cellular processes. For example,
insulin-stimulated glycogen synthesis in skeletal muscle
is mediated by GS which is considered to be the rate-
determining enzyme of this process.4 Phosphorylation
of GS by GSK-3 leads to its deactivation.5 In type 2
diabetics the rate of glycogen synthesis is impaired,6 the
activity of GS is reduced7 and GSK-3 expression and
activity are elevated and inversely correlated with
insulin-stimulated glucose disposal,3 thus implicating
elevated GSK-3 in the disease process. GSK-3 is also
known to phosphorylate the microtubule associated
protein tau in mammalian cells,8 hyperphosphorylation
of which is an early event in neurodegenerative con-
ditions such as Alzheimers disease. GSK-3 inhibition is
also implicated in neuronal cell survival.9
GSK-3b (albeit at low millimolar concentration) has
been proposed as a possible underlying therapeutic
mechanism of action.10 In support of this, valproate,
another drug commonly used for the treatment of
bipolar disorder, is also an inhibitor of GSK-3 (both
isoforms) at concentrations which have clinical e-
cacy.11 Lithium has been shown to protect cerebellar
granule neurons from death12 and chronic lithium
treatment has demonstrable ecacy in the middle
cerebral artery occlusion model of stroke.13 Lithium has
also been found to mimic some of the actions of insulin
by stimulating glycogen synthesis.14 Lithium exerts
other e€ects apart from weak inhibition of GSK-3, and
Jope15 has argued that it is the integration of these
multiple actions of lithium which contribute to its
bene®cial e€ects.
More potent and selective inhibitors of GSK-3 than
lithium would thus not only help to delineate the rele-
vance of this kinase in a variety of disease states, but
may also serve to ameliorate the condition should a
suitable therapeutic agent be obtained. E€ort towards
obtaining such compounds is reported herein.
High throughput screening of the SmithKline Beecham
compound collection against rabbit GSK-3a16 identi®ed
a number of maleimide-containing structures as potent
inhibitors. Bisindole maleimides of generic structure 1
were originally synthesised as selective protein kinase C
(PKC) inhibitors17 but, since the time of the original
publications, inhibition of other kinases has been
reported.18 The ability of this type of compound to
inhibit GSK-3 has recently been established19 but con-
cerns about a possible general selectivity issue with this
Bipolar disorder (manic depression) is commonly trea-
ted with lithium and the ability of this cation to inhibit
*Corresponding author. Tel.: +44-1279-627810; fax: +44-1279-
627841; e-mail: dave_g_smith@sbphrd.com
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(00)00721-6

636
D. G. Smith et al. / Bioorg. Med. Chem. Lett. 11 (2001) 635±639
class of compound (see Davies et al.19c) precluded
development as a template of interest for our studies. By
contrast, the lead anilinomaleimide 2 (IC50 910 nM,
rabbit GSK-3a) had no published precedent for kinase
inhibition and hence was considered a template worthy
of further exploration. (Recently, however, 3-anilino-4-
indole-maleimides have been disclosed as inhibitors of
PKCb.20)
of generally >95% purity.24 However, the amine dis-
placement reaction was unsatisfactory under these con-
ditions in those cases where Rand R were both ortho
1
substituted or where R1 was a strong ortho or para
electron withdrawing group. In these cases higher yields
and/or cleaner reactions were obtained using 4-aryl-3-
chloromaleimide and aniline in N-methylpyrrolidinone
at 150±200 ꢁC for 1±2 h. Whilst the conditions for
robotic synthesis were generally satisfactory in the case
of anilines, it was found that a large number of products
derived from primary amines hydrolysed to the hydroxy
maleimide 4 during the acidic work up, presumably as a
consequence of the greater basicity of the amine and
thus its more facile protonation and better leaving
group ability. This was easily overcome by utilising
conventional manual work up procedures with con-
comitant shorter reaction times.
Enzymology and SAR
Selected compounds from this series were competitive
with ATP25 when evaluated against recombinant human
GSK-3a (hGSK-3a).26 Compounds were routinely
evaluated against both hGSK-3a and b to establish
IC50s.27 In no case was selectivity demonstrated for
either enzyme, re¯ecting the homology of the ATP
binding site in these isoforms (53 residues immediately
surrounding the ATP binding site of hGSK-3 are iden-
tical apart from a Glu195(a) to Asp132(b) modi®cation).
The medicinal chemistry strategy was to react an initial
set of chloromaleimides (5, R=H, 4-Cl, and 4-OMe)
with a diverse set of anilines in a 3 (chloromaleimide)
Â16 (aniline) two-dimensional array. This gave a num-
ber of compounds with submicromolar potency which
were then elaborated in a variety of array formats via a
combination of rational design, additivity of sub-
stituents, and further exploitation of diversity to opti-
mise potency. Aryl substituents that were found to
confer sub 100 nM potency were subsequently investi-
gated by one-dimensional arrays to de®ne the limits of
the activity boundary. Similarly aniline substituents that
were routinely found to confer good potency were used
in arrays with newly synthesised chloromaleimide inter-
mediates. Use of these iterative processes rapidly led to
identi®cation of compounds with IC50<50 nM.
Chemistry
The synthetic route to anilinomaleimides of general
structure 6 is shown in Scheme 1. Base-induced con-
densation of an appropriate arylacetamide 3 with dime-
thyl oxalate gave the 3-hydroxy-4-arylmaleimides 421 in
53±100% yield which, after treatment with phosphorus
oxychloride in N,N-dimethylaniline22 or with oxalyl
chloride,23 provided the chlorinated derivatives 5 (15±
66%). Treatment of these with anilines furnished the
required anilinomaleimides 6.22
Ready access to arylacetamides led to the generation of
a number of chloromaleimides which could then be
combined array fashion with a diverse range of com-
mercially available anilines and amines. It quickly
became clear that the ®nal step of this scheme was
ideally suited to robotic synthesis, and conditions were
devised for the displacement such that compounds were
routinely prepared in one- or two-dimensional, 48-
membered arrays (or subsets thereof) using a Bohdan
RAM
Automatic
Synthesiser
(chloromaleimide
(0.5 mmol), aniline (2.5 equiv), methanol (2 mL), 65 ꢁC,
72 h followed by addition of ethyl acetate (6 mL) and
washing with HCl (2 M) and brine). Parallel chromato-
graphy (0±10% ether or 0±5% methanol in dichoro-
methane on silica cartridges) using Sympur (Anachem)
or Quad3 (Biotage) instrumentation provided material
Those substituents in the aryl and anilino rings that
frequently conferred potency are shown as a small sub-
set of data obtained (Table 1). Parent compounds (i.e.,
either R=H or R1=H ) are shown, where prepared, to
establish baseline data. Interestingly, both electron-
withdrawing and electron-donating groups in the non-
Scheme 1. Preparation of 3-anilino-4-arylmaleimides. Reagents: (a) (MeO2C)2, KOBut, DMF; (b) (COCl)2, DCM or POCl3, PhNMe2; (c) R1-PhNH2,
MeOH or N-methylpyrrolidinone.

D. G. Smith et al. / Bioorg. Med. Chem. Lett. 11 (2001) 635±639
Table 1. IC50 values for 3-anilino-4-arylmaleimide inhibitors of hGSK-3aa
637
aIC50 (nM) versus hGSK-3a, see ref 27 for assay details. Results are a mean of two determinations run in duplicate (n=4) and are given as
meanÆSEM. The compound where R=2-OMe and R1=3-Cl-4-OH was used as an internal standard across all assay plates and is a mean of
28 determinations.
anilino aryl ring resulted in highly potent compounds
suggesting that either there is not an overriding
electronic component to the activity conferred by this
moiety or, that if there is, then the binding mode of
those particular compounds is likely to be di€erent.
Where comparisons allow, potency is always enhanced
over parent by the presence of a 3-carboxyl substituent
in the anilino moiety, suggesting a favourable interac-
tion of carboxylate with the protein. Potency is always
augmented by the presence of an additional 4-chloro
substituent. A similar increase of potency is observed by
the addition of 3-chloro or 3,5-dichloro substituents to
those compounds containing a 4-hydroxyaniline moiety,
suggesting that the greater acidity of the phenol, or the
above acid, may result in the generation of a stronger
H-bond interaction, or that potency is gained from the
additional lipophilicity. Combination of the favoured
aryl substituents with those aniline moieties which con-
sistently gave good potency resulted in some of the most
potent compounds to be obtained from this series
(Table 1, highlighted).
Potency of the three methylated compounds 9a±c (5
+MeI/K2CO3 prior to coupling with aniline, or use of
the N-methylated analogue of 3 in the reaction
sequence) is substantially reduced (9a, 9b IC50>5 mM;
9c IC50>3 mM) relative to the unsubstituted maleimide,
implying that the maleimide NH is essential for binding.
By analogy with published X-ray structures of indolo-
carbazole kinase inhibitors such as staurosporine28 it is
likely that the anilinomaleimide adopts a binding mode
where the maleimide proton and the adjacent carbonyl
oxygen form a donor±acceptor pair with the CˆO (n)
and NH (n+2) of three amino acid residues in the hinge
region of the ATP site of the kinase. Molecular model-
ling studies of compounds from this series using
an homology model of GSK-3 derived from cyclin
dependent kinase-2 (CDK-2) will be discussed in future
publications.
Selectivity
Selectivity of kinase inhibition is critical for pathway
analysis in cellular systems and in vivo. For this reason
some of the above compounds were pro®led against a
panel of more than 20 kinases (including GSK-3b),
Table 2.29 From the table it is clear that representative
examples from the anilinomaleimide series are selective
inhibitors of GSK-3 (see footnote to Table 2 for com-
pound descriptors). For the compounds shown, inhibi-
tion of the majority of these kinases was <50% at
10 mM in contrast to their potency as inhibitors of GSK-
3b. Signi®cantly, since a number of recent publications30
describe potent inhibitors of CDK-2 which also inhibit
GSK-3, two of the most potent GSK-3 inhibitors
Comparison of N-methylanilinomaleimides 7 with the
anilino analogues 6 shows a marked reduction in
potency of the former compounds. In the absence of
information to the contrary, this might suggest that the
aniline proton is critical for binding. However, the
indolines 8 show potency which, in some cases, is simi-
lar to that of the parent aniline, demonstrating that for
these compounds at least, an N±H interaction with the
protein appears not to be critical. Increased conforma-
tional restraint of the indoline may allow a binding
mode which is unavailable to the N-methylaniline
derived compounds.

638
D. G. Smith et al. / Bioorg. Med. Chem. Lett. 11 (2001) 635±639
Table 2. Selectivity of 3-anilino-4-arylmaleimide inhibitors for GSK-3ba
aValues are %I @ 10 mM using 100 mM ATP (see ref 19c for kinases used and assay details). ND% I at 10 mM versus this kinase not determined.
* 6a R=2-NO2, R1=3-Cl-4-OH; 6b R=4-OMe, R1=3-Cl; 6c R=3-NO2, R1=4-Cl-3-CO2H; 6d R=2-NO2, R1=4-Cl-3-CO2H; 8a R=2-OMe.
described here (6c and 6d, IC50 versus GSK-3a 26 and
28 nM, respectively) are not inhibitors of CDK-2.
DePaoli-Roach, A. A.; Roach, P. J. Arch. Biochem. Biophys.
1993, 304, 219. (d) Eldar-Finkelmann, H.; Argast, G. M.;
Foord, O.; Fischer, E. H.; Krebs, E. G. Proc. Natl. Acad. Sci.
U.S.A. 1996, 93, 10228.
6. DeFronzo, R. A.; Bonadonna, R. C.; Ferrannini, E. Dia-
betes Care 1992, 15, 318.
Conclusion
7. (a) Thorburn, A. W.; Gumbiner, B.; Bulacan, F.; Wallace,
P.; Henry, R. R. J. Clin. Invest. 1990, 85, 522. (b) Beck-Niel-
sen, H.; Vaag, A.; Damsbo, P.; Handberg, A.; Nielsen, O. H.;
Henriksen, J. E.; Thye-Ronn, P. Diabetes Care 1992, 15, 418.
(c) Bogardus, C.; Lillioja, S.; Stone, K.; Mott, D. J. Clin.
Invest. 1984, 73, 1185.
8. Lovestone, S.; Reynolds, C. H.; Latimer, D.; Davis, D. R.;
Anderton, B. H.; Gallo, J.-M.; Hanger, D.; Mulot, S.; Mar-
quardt, B. Curr. Biol. 1994, 4, 1077.
9. Pap, M.; Cooper, G. M. J. Biol. Chem. 1998, 273, 19929.
10. Klein, P. S.; Melton, D. A. Proc. Natl. Acad. Sci. U.S.A.
1996, 93, 8455.
11. Chen, G.; Huang, L.-D.; Jiang, Y.-M.; Manji, H. K. J.
Neurochem. 1999, 72, 1327.
An ecient, automated synthetic approach to deriva-
tives of a novel kinase inhibitor template is described
and compounds presented here represent some of the
most potent and selective GSK-3 inhibitors reported to
date. The high potency and good selectivity of com-
pounds from Table 2 should allow their use as tool
compounds to aid resolution of the often complex nat-
ure of those signalling pathways where GSK-3 is
believed to be implicated. The cellular activity of one
such compound (6a, SB-415286) has already been
described.25,31
12. D'Mello, S.; Anelli, R.; Calissano, P. Exp. Cell Res. 1994,
211, 332.
13. Nonaka, S.; Chuang, D.-M. Neuroreport 1998, 9, 2081.
14. Bosch, F.; Gomez-Foix, A. M.; Arino, J.; Guinovart, J. J.
J. Biol. Chem. 1986, 261, 16927.
15. Jope, R. S. Molecular Psychiatry 1999, 4, 117.
16. Rabbit GSK-3a is 95% homologous to human GSK-3a
and is >99% identical in the kinase domain. Material was
purchased from The Division of Signal Transduction Therapy,
Department of Biochemistry, The University of Dundee,
Dundee, DD1 5EH, Scotland, UK.
17. Davis, P. D.; Hill, C. H.; Lawton, G.; Nixon, J. S.; Wilk-
inson, S. E.; Hurst, S. A.; Keech, E.; Turner, S. E. J. Med.
Chem. 1992, 35, 177.
Acknowledgements
The authors acknowledge the assistance of SmithKline
Beecham colleagues from the Department of Bio-
technology and Genetics for the recombinant human
enzymes, the Departments of Vascular Biology and
Neurosciences for helpful discussions, Molecular
Screening Technologies for assay of the compounds and
the Division of Signal Transduction Therapy, Depart-
ment of Biochemistry, The University of Dundee for
selectivity screening.
18. (a) Alessi, D. R. FEBS Lett. 1997, 402, 121. (b) Deak, M.;
Clifton, A. D.; Lucocq, J. M.; Alessi, D. R. EMBO J. 1998,
17, 4426.
19. (a) PCT International Application WO 97/41854, 1997.
(b) Hers, I.; Tavare, J. M.; Denton, R. M. FEBS Lett. 1999,
460, 433. (c) Davies, S. P.; Reddy, H.; Caivano, M.; Cohen, P.
Biochem. J. 2000, 351, 95.
20. PCT International Application WO 00/06564, 2000.
21. Rooney, C. S.; Randall, W. C.; Streeter, K. B.; Zeigler, C.;
Cragoe, E. J., Jr.; Schwam, H.; Michelson, S. R.; Williams,
H. W. R.; Eichler, E.; Duggan, D. E.; Ulm, E. H.; Noll, R. M.
J. Med. Chem. 1983, 26, 700.
22. Wiley, R. H.; Slaymaker, S. C. J. Am. Chem. Soc. 1958,
80, 1385.
23. PCT International Application WO 97/34890, 1997.
24. All compounds had 1H NMRand mass spectra consistent
with their proposed structure.
References and Notes
1. Cohen, P.; Yellowlees, D.; Aitken, A.; Donella-Deana, A.;
Hemmings, B. A.; Parker, P. J. Eur. J. Biochem. 1982, 124, 21.
2. Woodgett, J. R. Trends Biochem. Sci. 1991, 16, 177.
3. Nikoulina, S. E.; Ciaraldi, T. P.; Mudaliar, S.; Mohideen,
P.; Carter, L.; Henry, R. R. Diabetes 2000, 49, 263.
4. (a) Leloir, L. F.; Olavaria, S. H.; Goldenberg, S. H.; Car-
minatti, H. Arch. Biochem. Biophys. 1959, 81, 508. (b) Man-
darino, L.; Wright, K.; Verity, L.; Nichols, J.; Bell, J.;
Kolterman, O.; Beck-Nielsen, H. J. Clin. Invest. 1987, 80, 655.
5. (a) Roach, P. FASEB J. 1990, 4, 2961. (b) Skurat, A. V.;
Roach, P. J. Biochem. J. 1996, 313, 45. (c) Zhang, W.;

D. G. Smith et al. / Bioorg. Med. Chem. Lett. 11 (2001) 635±639
639
25. Coghlan, M. P.; Culbert, A. A.; Cross, D. A. E.;
Corcoran, S. L.; Yates, J. W.; Pearce, N. J.; Rausch, O. L.;
Murphy, G. J.; Carter, P. S.; Roxbee-Cox, L.; Mills, D.;
Brown, M. J.; Haigh, D.; Ward, R. W.; Smith, D. G.;
Murray, K. J.; Reith, A. D.; Holder, J. C. Chem. Biol. 2000, 7,
793.
26. hGSK-3 isoforms with a HexaHis tag were expressed
using the baculovirus expression system and protein was
recovered from cells by homogenisation and puri®cation using
NiNTA Super¯ow (Qiagen).
generated for each compound by ®tting to a four parameter
model.
28. (a) Prade, L.; Engh, R. A.; Girod, A.; Kinzel, V.; Huber,
R.; Bossemeyer, D. Structure 1997, 5, 1627. (b) Lawrie, A. M.;
Noble, M. E.; Tunnah, P.; Brown, N. R.; Johnson, L. N.;
Endicott, J. A. Nat. Struct. Biol. 1997, 4, 796. (c) Zhu, X.;
Kim, J. L.; Rose, P. E.; Stover, D. R.; Toledo, L. M.; Zhao,
H.; Morgenstern, K. A. Structure 1999, 7, 651. (d) Lamers,
M. B.; Antson, A. A.; Hubbard, R. E.; Scott, R. K.; Williams,
D. H. J. Mol. Biol. 1999, 285, 713.
27. Using microtitre plates, GSK-3 was assayed in 50 mM
MOPS bu€er, pH 7.0 containing 5% glycerol, 0.01% Tween-
20, 7.5 mM 2-mercaptoethanol, 10 mM magnesium acetate,
8 mM substrate peptide (Biotin-KYRRAAVPPSPSLSRHSS-
PHQ(SP)EDEEE, where (SP) is a pre-phosphorylated serine)
and 10 mM [g-33P]-ATP. After incubation for 1 h at room
temperature, the reaction was stopped by addition of 50 mM
EDTA solution containing Streptavidin coated SPA beads
(Amersham) to give a ®nal 0.2 mg of beads per assay well in a
384 microtitre plate format. Following centrifugation, the
microtitre plates were counted in a Trilux 1450 microbeta
liquid scintillation counter (Wallac). IC50 values were
29. Compounds were assayed in the laboratories of the Divi-
sion of Signal Transduction Therapy, Department of Bio-
chemistry, The University of Dundee, Dundee, DD1 5EH,
Scotland, UK. Kinases are de®ned and assay details are given
in Davies et al. ref 19c above.
30. Leost, M.; Schultz, C.; Link, A.; Yong-Zhong, W.; Bier-
nat, J.; Mandelkow, E.-M.; Bibb, J. A.; Snyder, G. L.;
Greengard, P.; Zaharevitz, D. W.; Gussio, R.; Senderowicz,
A. M.; Sausville, E. A.; Kunick, C.; Meijer, L. Eur. J. Bio-
chem. 2000, 267, 5983 and references therein.
31. Cross, D. A. E.; Culbert, A. A.; Chalmers, K. A.; Facci,
L.; Skaper, S. D.; Reith, A. D. 2001, in press.