Discovery and SAR of 2-aminothiazole inhibitors of cyclin-dependent kinase 5/p25 as a potential treatment for Alzheimer's disease

Article abstract of DOI:10.1016/j.bmcl.2004.09.006

High-throughput screening with cyclin-dependent kinase 5 (cdk5)/p25 led to the discovery of N-(5-isopropyl-thiazol-2-yl)isobutyramide (1). This compound is an equipotent inhibitor of cdk5 and cyclin-dependent kinase 2 (cdk2)/cyclin E (IC₅₀ = ca. 320 nM). Parallel and directed synthesis techniques were utilized to explore the SAR of this series. Up to 60-fold improvements in potency at cdk5 and 12-fold selectivity over cdk2 were achieved.

Full text of DOI:10.1016/j.bmcl.2004.09.006

Bioorganic & Medicinal Chemistry Letters 14 (2004) 5521–5525
Discovery and SAR of 2-aminothiazole inhibitors of
cyclin-dependent kinase 5/p25 as a potential treatment for
AlzheimerÕs disease
Christopher J. Helal,* Mark A. Sanner, Christopher B. Cooper, Thomas Gant,
Mavis Adam, John C. Lucas, Zhijun Kang, Stanley Kupchinsky, Michael K. Ahlijanian,
Bonnie Tate, Frank S. Menniti, Kristin Kelly and Marcia Peterson
Neuroscience Medicinal Chemistry and Pharmacology, Pfizer Global Research and Development, Eastern Point Road, Groton,
CT 06340, USA
Received 9 July 2004; revised 2 September 2004; accepted 3 September 2004
Available online 29 September 2004
Abstract—High-throughput screening with cyclin-dependent kinase 5 (cdk5)/p25 led to the discovery of N-(5-isopropyl-thiazol-2-
yl)isobutyramide (1). This compound is an equipotent inhibitor of cdk5 and cyclin-dependent kinase 2 (cdk2)/cyclin E (IC50 = ca.
320nM). Parallel and directed synthesis techniques were utilized to explore the SAR of this series. Up to 60-fold improvements in
potency at cdk5 and 12-fold selectivity over cdk2 were achieved.
Ó 2004 Elsevier Ltd. All rights reserved.
Cyclin dependent kinases (cdk) play important roles in
cell cycling, and cdk inhibitors have been intensively re-
searched as therapeutic agents for the treatment of can-
cer.1 In addition, cdk5 and its noncyclin cofactor p35
phosphorylate the cytoskeletal stabilizing protein tau,
a critical component of cell structure regulation.2 The
discovery that the short-lived, membrane-bound p35
can be proteolytically cleaved by calpain to the more
stable, cytosolic form p25 led to the hypothesis that
cdk5/p25 plays an important role in AlzheimerÕs disease
etiology. The longer-lived cdk5/p25 complex is thought
to hyperphosphorylate tau, leading to the formation of
paired helical filaments and deposition of cytotoxic neu-
rofibrillary tangles.3 Inhibition of the anomalous cdk5/
p25 complex is, therefore, a viable target for treating
AlzheimerÕs disease by preventing tau hyperphosphoryl-
ation and neurofibrillary tangle formation. We have
developed a series of potent and selective 2-aminothiaz-
ole inhibitors of cdk5/p25 as potential therapeutic
agents for the treatment of AlzheimerÕs disease and
other neurodegenerative disorders.4,5
High-throughput screening with the cdk5/p25 complex
uncovered N-(5-isopropyl-thiazol-2-yl)isobutyramide
(1), as one of the most interesting hits due to its good
potency versus cdk5/p25 (IC50 = 321nM), low molecular
weight (MW = 212), and relatively straightforward
chemical modification (Fig. 1). In the presence of
increasing concentrations of ATP, 1 became less effec-
tive at inhibiting cdk5/p25, suggesting that this com-
pound is competitive at the ATP binding site.
Screening of 1 versus a panel of 24 diverse protein kin-
ases showed only weak inhibitory activity.6 Further test-
ing against other cyclin dependent kinases revealed that
1 was essentially equipotent at inhibiting cdk2/cyclin E
(IC50 = 318nM), a cancer target.1 Consequently, our
objectives were to improve cdk5 potency and minimize
cdk2 activity.
We focused our synthetic efforts on varying the amide
side chain and the 5-isopropyl group. Amides were pre-
pared in a high-speed parallel fashion by reacting the
2-amino-5-substituted thiazole with acid chlorides in
Keywords: 2-Aminothiazole; Cyclin-dependent kinase 5; p25; Cyclin-
dependent kinase 2; Cyclin E; AlzheimerÕs disease.
*
Corresponding author. Tel.: +1 860 715 5064; fax: +1 860 686
5291; e-mail: chris_j_helal@groton.pfizer.com
Figure 1. High-throughput screening lead 1.
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.09.006

5522
C. J. Helal et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5521–5525
Scheme 1. (a) (i) R0C(O)Cl (1.2equiv), morpholinemethyl resin
(4.5equiv), 1,2-dichloroethane, (ii) aminomethyl resin (3.75equiv);
(b) R0CO2H (1.2equiv), Et3N, (n-PrP(O)O)3 (1.2equiv), ethyl acetate,
23°C, 40–85% yield; (c) PhOC(O)Cl (1equiv), (i-Pr)2NEt, CH2Cl2,
ꢀ78 to 0°C, 80% yield; (d) R0NH2 (1.2equiv), 1,4-dioxane, 100°C, 30–
85% yield.
Scheme 3. (a) (i) n-BuLi (2equiv), THF, ꢀ78°C, 1h, (ii) Me3SiCl
(2equiv), ꢀ78 to ꢀ20 to ꢀ78°C, (iii) n-BuLi (1equiv), ꢀ78°C, 10min;
(b) RRC@O, 33–71% yield; (c) Et3SiH, CF3CO2H, 63–91% yield; (d)
H2 (50psi), Pd(OH)2 carbon, CF3CO2H, 62–91% yield; (e) R0CO2H
(1.2equiv), Et3N, (n-PrP(O)O)3 (1.2equiv), ethyl acetate, 23°C, 40–
85% yield; (f) (MeS)2, 74% yield.
Scheme 2. (a) 5,5-dibromobarbituric acid (0.5equiv), Et2O, 23°C, 18h,
filter, concentrate in vacuo, add pentane, filter, concentrate in vacuo,
46% yield; (b) thiourea (1equiv), ethanol, reflux, 18h, 58% yield.
the presence of a resin-bound morpholine base followed
by addition of an aminomethyl resin to scavenge excess
acid chloride (Scheme 1). Amides were also prepared
utilizing the tripropylphosphonic anhydride mediated
coupling of carboxylic acids with the 2-aminothiazole
core. Ureas were readily synthesized in a high-speed par-
allel format by conversion of a 2-aminothiazole to the
corresponding phenyl carbamate, which was reacted
with an amine in 1,4-dioxane at reflux.
Scheme 4. (a) HBF4 (5equiv), SM, H2O, rt, Cu(s) (1.6equiv), NaNO2,
(15equiv), H2O, 0°C 7%; (b) Me2NH2Cl, Et3N, DMSO, 60°C, 60%;
(c) H2, Pd/C, EtOAc, 77%; (d) PhCH2CO2H, (n-PrP(O)O)3, Et3N,
EtOAc, 23°C, 40%.
compounds with IC50 < 1lM, the IC50s reported are
the mean of at least two independent measurements with
standard error calculated. Each IC50 was determined
from a six-point dose–response curve run in triplicate.
2-Amino-5-substituted thiazoles were prepared by three
methods. For simple 5-alkyl analogs, commercially
available aldehydes were converted into a-bromoalde-
hydes and condensed with thiourea (Scheme 2).
Variation of the amide side chain of 1 with high speed
and traditional chemistries allowed us to rapidly explore
the first arm of the pharmacophore (Table 1). Relative
to the parent isopropyl (1), benzamide (2), phenyl carb-
amate (3), and phenyloxylate substitution (4) all resulted
in decreased potency. However, phenyl urea (5) and
phenylacetamide (6) provided the first analogs that dem-
onstrated improved activity. We observed little to no
change in the selectivity of these analogs versus cdk2.
The second method utilized KatritzkyÕs in situ silylation/
metallation of 2-aminothiazole followed by trapping
with an electrophile (Scheme 3).7 When trapping with
an aldehyde or ketone, the resultant carbinol was elimi-
nated to give the olefin using trifluoroacetic acid. The
carbinol was reduced via ionic hydrogenolysis (trifluoro-
acetic acid, triethylsilane) or using acid-catalyzed elimi-
nation/hydrogenation (trifluoroacetic acid, palladium
hydroxide on carbon, 50psi hydrogen).
Table 1. SAR of amide sidechain
For the synthesis of a 5-dimethylamino analog, 5-bromo-
2-aminothiazole was first converted to 5-bromo-2-nitro-
thiazole under Sandmeyer conditions. Nucleophilic
aromatic substitution of bromide with dimethylamine
followed by catalytic hydrogenation and acylation
afforded the desired compound (Scheme 4).
Compd
R
Cdk5 IC50
(nM)
Cdk2 IC50
(nM)
Select
(k2/k5)
1
2
3
4
5
6
i-Pr
Ph
321 18
1150
1600
318 18
1020
980
1
0.9
0.6
2.1
2.6
1.5
Kinase inhibition was measured by the use of scintilla-
tion proximity assays (SPA).8 The ATP final concentra-
tion used was the same in both the cdk5/p25 and cdk2/
cyclin E assay. The Km of ATP was measured to be
12lM for cdk5/p25 and 19lM for cdk2/cyclin E. For
OPh
C(@O)Ph
NHPh
CH2Ph
4580
102 19
64 12
9770
265 46
98 16

C. J. Helal et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5521–5525
5523
Table 2. SAR of 5-position
Compd
R
Cdk5 IC50 (nM)
Cdk2 IC50 (nM)
Select (k2/k5)
6
i-Pr
Et
64 12
621 212
98 16
412 140
74 11
1.5
0.7
3.0
2.4
2.6
7
8
c-Butyl
c-Pentyl
c-Hexyl
25
64
1
2
9
10
155 18
997 275
390 133
11
6530
6510
1.0
12
13
14
15
16
17
n-Pentyl
Ph
3270
1570
7670
4120
2000
4600
1.2
1.3
0.6
1.2
0.5
1.2
Acetyl
MeS
Br
91
900
4
105
480
3
Me2N
3990
4810
Investigation of the 5-position SAR was carried out with
the 2-phenylacetamide as a constant (Table 2). The ethyl
group (7) was 10-fold less potent than the parent isopro-
pyl (6). The cyclobutyl (8) improved potency by ca.
3-fold relative to isopropyl and cyclopentyl (9) was
equipotent with isopropyl. Cyclohexyl (10), however,
dropped off noticeably in potency, as did other groups
such as cyclopentenyl (11), n-pentyl (12), phenyl (13),
and acetyl (14). Direct heteroatom attachment afforded
analogs with a range of activities. Thiomethyl (15) pro-
vided potent inhibitory activity, whereas bromo was
considerably weaker (16). Dimethylamino (17), which
is sterically similar to isopropyl but more polar, also lost
considerable potency.
The 8-isoquinolyl derivative (30) was equally selective
versus cdk2 and had slightly improved cdk5 IC50. A
number of other fused heteroaryl derivatives were syn-
thesized (e.g., 31,32), but all had noticeably decreased
cdk5 activity.9
Table 3. SAR of aryl/heteroaryl acetamides
Compd
Ar
cdk5 IC50
(nM)
cdk2 IC50
(nM)
Select
(k2/k5)
8
25
1
74 11
3.0
4.7
Attention was then turned to optimization of the aryl-
acetamide. The incorporation of fused heteroaryl groups
led to improved potency (Table 3). The 5-benzimidazole
substituent (18) showed excellent cdk5 activity and pro-
vided the first signs of selectivity versus cdk2 (fivefold).
Other fused heteroaryl analogs with b-type substitution
such as 6-quinolyl (19) and 6-benzthiazole (20) were
more potent than the parent phenyl compound but did
not display the desired cdk2 selectivity. Relative to b-
type substitution, fused heteroaryl analogs with a-type
substitution were somewhat less potent (21–24). The 5-
isoquinolyl derivative (21), however, did show 5-fold
selectivity versus cdk2.
18
19
20
7
1
33
26
16
6
2
1
10
12
1
4
2.6
1.3
21
22
23
24
13
16
13
48
4
1
4
5
70 11
5.4
2.5
2.2
2.1
The improved potency and selectivity observed in the
fused heteroarylacetamide analogs led us to study these
same SAR trends in the corresponding ureas (Table 4).
The phenyl urea (25) was ca. 7-fold less potent than
the corresponding phenylacetamide (8) with little cdk2
selectivity. The incorporation of quinolines and isoquin-
olines with b-type substitution (26,27) afforded im-
proved potency relative to the parent phenyl urea and
improved cdk2 selectivity (ca. 4-fold). Compounds with
a-type substitution such as 5-quinolyl (28) gave even
better cdk5 activity. Changing the position of the
quinolyl nitrogen (29) improved cdk5 potency and for
the first time afforded >10-fold selectivity versus cdk2.
40
26
6
4
100 11

5524
C. J. Helal et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5521–5525
Table 4. SAR of aryl/heteroaryl ureas
plified by 32. SAR patterns that emerged are (1) potency
and selectivity are increased by the emplacement
of fused heteroaryl rings in both acetamides and ureas
off of the 2-amino position and (2) a small lipophilic
pocket appears to exist in cdk5 near the thiazole 5-posi-
tion, with cyclobutyl providing optimal enzyme inter-
actions. These insights will be useful in further
chemistry efforts.
Compd
Ar
cdk5 IC50
(nM)
cdk2 IC50
(nM)
Select
(k2/k5)
25
26
183 89
44 11
290 50
167 27
1.6
3.8
References and notes
1. Nugiel, D. A.; Vidwans, A.; Etxkorn, A.-M.; Rossi, K. A.;
Benfield, P. A.; Burton, C. R.; Cox, S.; Doleniak, D.;
Seitz, S. J. Med. Chem. 2002, 45, 5224, and references cited
therein.
2. Lau, L.-F.; Schachter, J. B.; Seymour, P. A.; Sanner, M.
A. Curr. Topics Med. Chem. 2002, 2, 395.
27
28
44
13
1
1
176 37
4.0
3.3
43
4
3. Lau, L.-F.; Seymour, P. A.; Sanner, M. A.; Schachter, J.
B. J. Mol. Neurosci. 2002, 19, 267.
29
30
8
5
1
2
96 12
12.0
11.0
4. For recent articles describing 2-aminothiazole inhibitors
of cdk2 see: (a) Kim, K. S.; Kimball, S. D.; Misra, R. N.;
Rawlins, D. B.; Hunt, J. T.; Xiao, H.-Y.; Lu, S.; Qian, L.;
Han, W.-C.; Shan, W.; Mitt, T.; Cai, Z.-W.; Poss, M. A.;
Zhu, H.; Sack, J. S.; Tokarski, J. S.; Chang, C.-Y.;
Pavletich, N.; Kamath, A.; Humphreys, W. G.; Marathe,
P.; Bursuker, I.; Kellar, K. A.; Roongta, U.; Batorsky, R.;
Mulheron, J. G.; Bol, D.; Fairchild, C.; Lee, F. Y.;
Webster, K. R. J. Med. Chem. 2002, 45, 3905; (b) Misra,
R. N.; Xiao, H.-Y.; Kim, K. S.; Lu, S.; Han, W.-C.;
Barbosa, S. A.; Hunt, J. T.; Rawlins, D. B.; Shan, W.;
Ahmed, S. Z.; Qian, L.; Chen, B.-C.; Zhao, R.; Bednarz,
M. S.; Kellar, K. A.; Mulheron, J. G.; Batorsky, R.;
Roongta, U.; Kamath, A.; Marathe, P.; Ranadive, S. A.;
Sack, J. S.; Tokarski, J. S.; Pavletich, N. P.; Lee, F. Y. F.;
Webster, K. R.; Kimball, S. D. J. Med. Chem. 2004, 47,
1719; (c) Misra, R. N.; Xiao, H.; Williams, D. K.; Kim, K.
S.; Lu, S.; Keller, K. A.; Mulheron, J. G.; Batorsky, R.;
Tokarski, J. S.; Sack, J. S.; Kimball, S. D.; Lee, F. Y.;
Webster, K. R. Bioorg. Med. Chem. Lett. 2004, 14, 2973.
5. (a) Cooper, C. B.; Helal, C. J.; Sanner, M. A. EU Patent
EP-1256578-A1, 2002; (b) Sanner, M. A.; Helal, C. J.;
Cooper, C. B. US Patent U.S. 6,720,427 B2, 2004; (c)
Pevarello, P.; Amica, R.; Villa, M.; Salom, B.; Vulpetti,
A.; Varasi, M. U.S. Patent 372,832, 2000; (d) Pevarello, P.;
Amici, R.; Traquandi, G.; Villa, M.; Vulpetti, A.; Isacchi,
A. WO Patent 0026203, 2000.
55
5
31
32
190
41
8
5
650 72
106 38
3.4
2.6
Modeling of these analogs in the ATP binding region of
cdk5/p25 suggests hydrogen bonding between the ami-
nothiazole N–H and heterocyclic N with the carbonyl
and N–H of Cys83, respectively.10 Substituents at the
aminothiazole 5-position appear to make a hydrophobic
interaction with Phe80, which is in line with SAR
observed (Table 2). Aromatic amides (Table 3) and
ureas (Table 4) may be interacting with IleA10, a
possible reason for increased potency with these groups.
The challenge in gaining selectivity over cdk2 is clear,
considering that only two out of the 29 ATP binding
pocket residues are different. The two that are different,
Cys83 and Asp84 in cdk5 versus Leu83 and His84 in
cdk2, are located in the backbone hydrogen-bonding re-
gion, but the side chains do not point into the ATP bind-
ing pocket, reducing their ability to influence selective
binding of a substrate. The observed selectivity most
likely is the result of subtle differences in conformations
of conserved residues near the bound inhibitor. For
example, modeling of the selective analog 30 in cdk5/
p25, Lys89 is disposed to make a hydrogen bond with
the isoquinolyl nitrogen, whereas that same residue is
further away in cdk2 and cannot make the same inter-
action, possibly explaining the observed selectivity.
6. The kinase selectivity panel consisted of Akt-1, PKCa,
phosphorylase kinase, lck, p70 S6K, p38d, p38c, p38b,
p38, PRAK, ROKa, SGK, CKII, ChK1, PDK1, JNK,
GSK3b, MAPK, RSK2, MSK1, PKA, MAPKAP kinase-
2, AMPK, and MEK1. All assays were conducted in the
presence of 100lM ATP. Compound 1 showed less than
50% inhibition versus the kinases tested.
7. Katritzky, A. R.; Laurenzo, K. S.; Relyea, D. I. Can. J.
Chem. 1988, 66, 1617–1624.
8. See Ref. 5a for detailed cdk5/p25 and cdk2/cyclin E assay
methods.
9. 1H NMR and mass spectral data for representative
compounds:
Compound 1: 1H NMR (400MHz, CDCl3) d 7.02 (s,
1H), 3.13 (m, 1H), 2.71 (m, 1H), 1.31 (d, J = 6.8Hz, 6H),
1.27 (d, J = 6.8Hz, 6H); MS (AP/CI) 213.3 (M+H)+,
100%.
Compound 5: 1H NMR (400MHz, CDCl3) d 7.49 (d,
J = 7.7Hz, 1H), 7.3 (m, 3H), 7.09 (m, 1H), 7.02 (s, 1H), 3.1
(m, 1H), 1.32 (d, J = 6.8Hz, 6H); MS (AP/CI) 262.2
(M+H) +, 100%.
In summary, modifications to HTS hit 1 successfully
decreased the cdk5 IC50 to <10nM and improved cdk2
selectivity to >10-fold while keeping the molecular
weight of the target compounds below 350amu, exem-

C. J. Helal et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5521–5525
5525
Compound 6: 1H NMR (400MHz, CDCl3) d 7.3 (m,
5H), 7.02 (s, 1H), 3.79 (s, 2H), 3.08 (m, 1H), 1.29
(dd, J = 1.4, 6.8Hz, 6H); MS (AP/CI) 261.3 (M+H)+,
100%.
2.1 (m, 2H), 1.95 (m, 1H), 1.87 (m, 1H); MS (AP/CI) 313.0
(M+H)+, 100%.
Compound 30: 1H NMR (400MHz, CDCl3) d 9.54 (s,
1H), 8.50 (d, J = 5.4Hz, 1H), 8.27 (d, J = 6.6Hz), 7.61 (t,
J = 7.9Hz, 2H), 7.56 (d, J = 5.6Hz, 1H), 7.01, (s, 1H), 3.54
(m, 1H), 2.36 (m, 2H), 2.09 (m, 2H), 1.97 (m, 1H), 1.89 (m,
1H); MS (AP/CI) 325.0 (M+H)+, 30%; 323.2 (MꢀH)ꢀ,
100%.
Compound 8: 1H NMR (400MHz, CDCl3) d 7.3 (m, 5H),
7.02 (s, 1H), 3.81 (s, 2H), 2.37 (m, 2H), 2.13 (m, 2H), 2.0
(m, 2H); MS (AP/CI) 273.1 (M+H)+, 100%.
Compound 18: 1H NMR (400MHz, CDCl3) d 8.17 (s,
1H), 7.67 (s, 1H), 7.58 (m, 2H), 7.26 (dd, J = 1.4, 8.3Hz,
1H), 7.02 (s, 1H), 3.87 (s, 2H), 3.6 (m, 1H), 2.35 (m, 2H),
10. cdk5/p25 crystal structure ID code is 1h4L in the Protein
Data Bank.