Quinazolines as cyclin dependent kinase inhibitors

Article abstract of DOI:10.1016/S0960-894X(01)00185-8

Quinazolines have been identified as inhibitors of CDK4/D1 and CDK2/E. Aspects of the SAR were investigated using solution-phase, parallel synthesis. An X-ray crystal structure was obtained of quinazoline 51 bound in CDK2 and key interactions within the ATP binding pocket are defined.

Full text of DOI:10.1016/S0960-894X(01)00185-8

Bioorganic & Medicinal Chemistry Letters 11 (2001) 1157–1160
Quinazolines as Cyclin Dependent Kinase Inhibitors
Thais M. Sielecki,a,* Tricia L. Johnson,a Jie Liu,a Jodi K. Muckelbauer,a
Robert H. Grafstrom,a Sarah Cox,a John Boylan,a Catherine R. Burton,a Haiying Chen,a
Angela Smallwood,a Chong-Hwan Chang,a Michael Boisclair,b Pamela A. Benfield,a
George L. Trainora and Steven P. Seitza
aThe DuPont Pharmaceuticals Company, Wilmington, DE 19880-0500, USA
bMitotix Inc., Cambridge, MA02139, USA
Received 22 December 2000; accepted 2 March 2001
Abstract—Quinazolines have been identified as inhibitors of CDK4/D1 and CDK2/E. Aspects of the SAR were investigated using
solution-phase, parallel synthesis. An X-ray crystal structure was obtained of quinazoline 51 bound in CDK2 and key interactions
within the ATP binding pocket are defined. # 2001 DuPont Pharmaceuticals Company. Published by Elsevier Science Ltd. All
rights reserved.
Interruption of the cell cycle is one approach in the
treatment of proliferative diseases. The phases of the
cell cycle are driven by cyclin-dependent kinases1
(CDK), serine- and threonine-specific kinases, which act
to modulate levels of protein phosphorylation using
adenosine triphosphate (ATP) as a phosphate donor.
Several reviews of the biology of the CDK family of
kinases have been published.2
In our search for potent and selective CDK inhibitors
we explored the impact that R2, R4, and R6 substitution
has on the in vitro CDK inhibition of this series. An
X-ray crystal structure of 51 revealed key binding
elements.
The R2 substituents were installed using one of three
routes. The dichloroquinazoline 2 was reacted with an
appropriate amine in the presence of triethyl amine in
tetrahydrofuran to give the R4 substituted chloride 3,
which was then exposed to aniline at 100 ꢀC overnight
to yield the R2 anilino substituted analogues (Scheme
1).
Several cores have been reported as potent CDK inhi-
bitors including flavones (flavopiridol),3 azepines (paul-
lones),4 and purines.5 Here, we report on the use of
quinazolines as CDK inhibitors. Quinazoline 1 is a
CDK inhibitor found through screening which shows
modest potency for CDK4/D1 (IC50=24 mM), CDK2/E
(IC50=13.5 mM) and CDK1/B(also known as CDC2/B)
(IC50=35 mM). The compound has modest activity in
an HCT 116 cancer cell line (IC50=15 mM) but did not
inhibit c-Abl (a tryosine kinase), PKC, or PKA. This
quinazoline was determined to be ATP competitive
similar to quinazolines found to be inhibitors of tyr-
osine kinases.6
When R2 is a trichloromethyl moiety it was installed via
reaction of aniline 5 with trichloroacetonitrile in HCl
saturated dioxane. The resultant chloride was then
reacted with an appropriate amine to give the R4 amino
substituted 6 (Scheme 2).
Alternatively, when R2 is a trifluoromethyl or penta-
fluoroethyl group, the amide 7 was cyclized to the qui-
nazolinone 8 by reacting with an appropriate ester in
*Corresponding author. Tel.:+1-302-695-8051; fax: +1-302-695-
1502; e-mail: thais.m.sielecki@dupontpharma.com
Scheme 1. (a) H2NR, Et3N, THF, overnight; (b) H2NPh, 100–115 ꢀC,
overnight.
0960-894X/01/$ - see front matter # 2001 DuPont Pharmaceuticals Company. Published by Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00185-8

1158
T. M. Sielecki et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1157–1160
the presence of NaOEt in ethanol. With the R2 group
installed the quinazolinone was converted to the chlo-
ride 9, which was then reacted with the appropriate
amine to give the R4 amino substituted 10 (Scheme 3).
similar profile to 13 but with greater potency for CDK2/
E (IC50=1.53 mM).
With the R2 SAR defined, we explored the 4-position
(Table 2). Branched alkyl groups such as t-butyl (14)
and t-amyl (22) are preferred. Longer groups such as n-
butyl (inactive at 50 mM) are inactive as are groups
added to increase hydrogen bonding such as sulfona-
mides (23) and 2-methylpentylamine (24). Benzyl sub-
stitution was modestly effective with electron
withdrawing substitution on the benzyl moiety render-
ing the compounds more potent for CDK2/E (28, 29).
Interestingly, the binding pocket distinguishes between
the R and S isomers of the 2-methylbenzylamine 30.
The R-isomer 31 has a 10-fold increased potency over
the S-isomer 32 in the CDK2/E assay. Substitution on
the aryl ring of the benzyl group did not improve the
potency of this substituent as seen for 33. The gem
dimethyl 34 showed activity averaging between 32 and
31. Overall, the data suggest that there is a size constraint
to the binding pocket in the area of the 4-position.
In evaluating the R4 position a parallel synthesis
approach was employed (Scheme 3). In a solution-
phase, parallel synthesis format, chloride 9 (R6=H) was
allowed to react with the appropriate primary or sec-
ondary amine in THF at room temperature overnight.
The reactions were then quenched with (amino-
methyl)polystyrene resin followed by DOWEX-50W-H
to remove residual starting materials. Following this
protocol, a total of 139 amines were reacted with 9. This
resulted in 124 evaluable compounds with acceptable
mass spectral analyses and greater than 85% purity by
HPLC analysis. Compounds that had >50% inhibition
in the enzyme assays were resynthesized as discretes
following the procedures outlined above.
Compounds with substitution at the 6-position were
synthesized following Scheme 3, starting with the
appropriately substituted aryl moiety. When R6 was an
aryl group, Suzuki chemistry was employed (Scheme 4).
Aromatic ring substitution at the 6-position was also
investigated (Table 3). Improved potency was observed
It was shown that small, electron withdrawing, non-
ionizable substituents such as trifluoromethyl and tri-
chloromethyl are preferred at R2 (Table 1). Compound
13 (CDK2/E IC50=6.1 mM) showed approximately 4-
fold selectivity for CDK2/E as compared to CDK2/A
and CDK1/Band 2-fold selectivity over CDK4/D. The
compound was shown to be selective against PKC and
PKA. Cell cycle effects were consistent with CDK2/
CDK4 inhibition (reversible G1 arrest in fibroblasts).
Analogue 14, the trifluoro version of 13, showed a
Table 1. Quinazoline substitution at the 2-position (R4=NHt-Bu,
R6=H)a
Compounds
R2
CDK4 IC50
(mM)
CDK2 IC50
(mM)
13b
14
15
16
17
18
19
20
21
CCl3
CF3
CF2CF3
Ph
CO2Et
CONH2
CO2H
H
11.9ꢁ0.0
14.3ꢁ3.4
>20.0
6.1ꢁ0.0
1.5ꢁ0.5
11.9ꢁ4.5
—
>5.8ꢁ0.0
164.5ꢁ1.5
>815
>361
168ꢁ13.0
>490
>408
>248
>171
>50
>34
NHPh
aN=2.
Scheme 2. (a) CCl3CN, dioxane, HCl (g); (b) H2NR, Et3N, THF,
overnight.
bCDC2/BIC 50=23.00ꢁ11.00 mM, cdk2/A IC50=25.80ꢁ24.00 mM.
Table 2. Quinazoline substitution at the 4-position (R2=CF3,
R6=H)
Compounds
R4
CDK4 IC50 CDK2 IC50
(mM)
(mM)
14
22
23
24
25
26
27
28
29
30
31
32
33
34
NHC(CH3)3
NHC(CH3)2CH2CH3
NHSO2CH(CH3)2
14.3ꢁ3.4
16.8ꢁ2.7
>157
1.5ꢁ0.5
1.4ꢁ0.1
>31.0
NHCH(CH3)CH2CH2CH3 119.0ꢁ12.0 245.0ꢁ49.0
Scheme 3. (a) RCO2Et, NaOEt, EtOH; (b) POCl3/reflux or POCl3,
CH3CN, N,N-dimethylaniline; (c) NHR0, THF, rt to reflux; (d) (i)
NHR1R2, THF, rt overnight; (ii) resin quench with (aminomethyl)-
polystyrene followed by DOWEX-50W-H.
3-Hydroxypiperidine
>84
32.4ꢁ2.0
14.8ꢁ0.2
2.3ꢁ0.5
15.6ꢁ1.2
8.9ꢁ1.8
6.8
2-Hydroxymethylpyrolidine 63.0ꢁ0.0
NHC(CH3)2CH2OH
NHCH2(30-F-Ph)
15.0ꢁ0.0
>39
>19
59
NHCH2(30-F,50-F-Ph)
(R,S)-NHCH(CH3)Ph
(R)-NHCH(CH3)Ph
(S)-NHCH(CH3)Ph
(R,S)-NHCH(CH3)(40-F-Ph)
NHC(CH3)2Ph
39.0ꢁ1.4
3.2ꢁ0.8
40a
31.0
16.5ꢁ3.5
14.0ꢁ1.4
35a
>151.0
Scheme 4. (a) RB(OH)2, 2 M Na2CO3, TBABr, EtOH, Pd(PPh3)4,
toluene, reflux overnight.
a% activity at 50 mM.

T. M. Sielecki et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1157–1160
1159
for R6 trifluoromethyl (37) and R6 heteroaryl (38–41).
Compound 37 was also notable for its approximately
50-fold selectivity for CDK2/E over CDK4/D and
CDK6/D while maintaining a modest 2-fold selectivity
with respect to CDK2/A.
increasing activity 9-fold as compared to 13 in the
CDK2/E assay.
Compounds were evaluated for their ability to inhibit
cell growth (Table 5). Quinazoline 36 showed the best
translation into an HCT 116 cancer cell line, having low
micromolar (IC50=5.72 mM) growth inhibition. Inter-
estingly, normal human fibroblasts (AG1523)7 tested
under identical assay conditions were relatively insensi-
tive to the inhibitors.
A number of R6 substituted phenyl groups were exam-
ined to determine if additional interaction with the
ATP binding pocket could be exploited (Table 4). The
3-amino substituted analogue (51) was optimal,
The crystal structure of human CDK2 complexed with
51 was determined to 2.0 A resolution. The inhibitor
binds into the ATP binding pocket located in the deep
cleft between the upper and lower domains of CDK2.8
Binding of 51 does not cause significant conformational
changes in backbone atom positions as compared to the
CDK2 apo enzyme structure8 (0.72 A rmsd). However,
significant movements of side chains were observed for
Lys33 (3.9 A) and Lys89 (5.7 A). These side-chain
movements occurred to accommodate inhibitor binding.
Residues within 3.5 A of the inhibitor include Lys33,
Phe80, Leu83, His84, Asp86, Lys89, and Asp145. The
CDK2/51 electron density maps suggest an indirect
hydrogen bond between N1 and the backbone oxygen
of Glu81 via a water molecule (Fig. 1). An additional
hydrogen bond is observed between the side-chain oxy-
gen of Asp86 and the N on R6. There is a stacking
interaction between the trifluoro group and Phe80.
These interactions are unique to this series and repre-
sent an alternate binding mode to that reported for a
series of 4-anilinoquinazolines.13
Table 3. Quinazoline substitution at the 6-position (R4=NHt-Bu)
Compounds
R2
R6,7,8
CDK4 IC50
(mM)
CDK2 IC50
(mM)
35
36
37a
38
39
40
41
CF3
CF3
CF3
OH
NHCOH
CF3
18.00ꢁ0.00
1.150ꢁ0.210
14.00ꢁ0.000 0.925ꢁ0.0071
>37.00
0.790ꢁ0.000
CF3 (3-Thiophene) 3.550ꢁ0.210 0.545ꢁ0.150
CF3 (2-Furyl) >3.00
1.20ꢁ0.140
CF3 (2-Thiophene)
6.40ꢁ0.990 1.180ꢁ0.550
CF3 (3-Pyridyl) 1.20ꢁ0.140
8.30ꢁ0.00
aCDK1/BIC
A IC50=2.05ꢁ0.21 mM.
>0.74 mM, CDK6/D2 IC50=56.00ꢁ1.40 mM, CDK2/
50
Table 4. 6-Substituted phenyl quinazolines (R2=CF3, R4=NHt-Bu)
Compounds
R6
CDK4 IC50
(mM)
CDK2 IC50
(mM)
42
43
44
45
46
47
48
49
50
51
52
53
Ph
7.7ꢁ1.0
>2.7
>27.0
11.0ꢁ0.0
>6.7
12.0ꢁ2.8
8.3ꢁ0.0
>26.0
>2.4
1.8ꢁ0.1
1.0ꢁ0.1
10.0ꢁ2.8
0.7ꢁ0.1
3.3ꢁ0.3
1.0ꢁ0.1
1.2ꢁ0.1
2.6ꢁ0.9
>12.0
2-MeOPh
4-MeOPh
4-HOPh
3-MeOPh
3-HOPh
3-PyridylPh
4-ClPh
2,4-DichloroPh
3-NH2Ph
4-NH2Ph
3-CONH2Ph
Employing solution-phase, parallel synthesis, a series of
4-aminosubstituted quinazoline CDK inhibitors was
obtained. In interactions with CDK2/E complex, the 4-
position of the quinazoline ring prefers small, branched
alkyl groups and benzyl groups. This series of 4-amino-
quinazolines has 5–20 times greater affinities for CDK2
as compared to CDK4 complexes. The 2-position of the
quinazoline prefers the trifluoromethyl group while the
6-position is sensitive to both size and electronics as
exemplified by 38 (CDK2/cyclin E IC50=0.54 mM) and
51 (CDK2/cyclin E IC50=0.65 mM). The effects on the
cell cycle of representative compound 13 were consistent
with CDK2/CDK4 inhibition. The crystal structure of
51 with CDK2 indicates that it binds in the ATP pocket
of CDK2 and has an indirect hydrogen bond backbone
oxygen of Glu81 via a water molecule. Overall, the qui-
nazoline series represents a novel core in the inhibition
of cyclic-dependent kinases with submicromolar
potency.
>2.1
>1.1
>2.5
0.6ꢁ0.1
1.1ꢁ0.1
1.3ꢁ 0.1
Table 5. Quinazolines dosed in HCT116 cell assay
Compounds
SRBHCT 116 IC 50
(mM)
SRBAG1523 IC 50
(mM)
13
14
35
36
37
38
25.00ꢁ1.40
20.60ꢁ4.80
7.30ꢁ0.64
5.72ꢁ0.44
18.60ꢁ4.30
18.40ꢁ1.10
—
>74.00ꢁ0.00
>18.00
—
>59.00
>28.00ꢁ0.00
Figure 1. Crystal structure of CDK2 complexed with 51.9

1160
T. M. Sielecki et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1157–1160
vitamins. Cultures were weaned onto 10% FBS over a two
week period before experiments.
Acknowledgements
8. (a) DeBondt, H. L.; Rosenblatt, J.; Jancarik, J.; Jones,
H. D.; Morgan, D. O.; Kim, S.-H. Nature 1993, 363, 595. (b)
Schulze-Gahmen, U.; DeBondt, H. L.; Kim, S.-H. J. Med.
Chem. 1996, 39, 4540.
The authors would like to thank Marc Arnone, Melissa
J. Ashbacher, Debra Doleniak, Laura Handel, Marv
Kendal, Lynn Leffet, Diane Sharp, Fariba Shoarinejad,
Lisa Sisk, Marge Stafford, and Ann Klemm for their
efforts in carrying out the in vitro assays.
9. Protein purification, crystallization, and structure determi-
nation: CDK2 protein was prepared and purified as descri-
bed10 with slight modifications including the addition of 10%
(v/v) glycerol during the SP-sepharose and ATP-agarose col-
umn steps. CDK2 protein was concentrated to 6 mg/mL using
a Collodion concentrator against 10 mM HEPES pH 7.4,
15 mM NaCl. Crystals were grown by vapor diffusion at 18 ꢀC
from sitting drops containing premixed and filtered (0.22 mm)
solutions of 3.0 mg/mL CDK2, 32.5 mM HEPES (pH 7.4),
11.3 mM NaCl, 12.5 mM ammonium acetate, 2 mM DTT, 2–
4% PEG 4000 against 100 mM HEPES (pH 7.4), 50 mM
ammonium acetate, 2 mM DTT, 4–14% PEG 4000. CDK2
crystals were transferred to a solution containing 10 mM
HEPES (pH 7.4), 15 mM NaCl, 1.0 mM 51, 5.0% DMSO and
soaked with inhibitor for 4 days. Crystals of CDK2/51 were
briefly transferred into cryo-protectant (10 mM HEPES pH
7.4, 15 mM NaCl, 25% MPD) and flash frozen in liquid
nitrogen in preparation for cryo-data collection. Diffraction
data were collected at ꢃ170 ꢀC at the DND-CAT beam line,
Advanced Photon Source, Argonne National Laboratories.
Data were processed and scaled with HKL.11 Crystals were
orthorhombic and belonged to the space group P212121 with
unit cell dimensions a=71.96 A, b=73.51 A, c=54.28 A,
a=b=g=90.0ꢀ. Data is 98% complete to 2.0 A resolution
with an overall R-merge of 8.0%. Initial rigid body refine-
ments12 of the apo CDK28a structure against the CDK2/51
data were unsuccessful (R-factor=47.0%). A translation
search12 found the highest peak in fractional coordinates at
x=0.47, y=0.019, z=0.000. Subsequent rigid body refine-
ments lowered the R-factor to 33.2%. The Fo-Fc electron
density map revealed the position of inhibitor in the ATP
binding pocket along with a water molecule sandwiched
between the inhibitor and backbone oxygen of Glu81. The
CDK2 + 51 + HOH structure (Fig. 1) was refined with
XPLOR12 to a final R-factor of 21.9% and an R-free of
27.4%.
References and Notes
1. Pines, J. Adv. Cancer Res. 1995, 66, 181.
2. Sielecki, T. M.; Boylan, J. F.; Benfield, P. A.; Trainor, G. L.
J. Med. Chem. 2000, 43, 1.
3. Senderowicz, A. M.; Headlee, D.; Stinson, S. F.; Lush,
R. M.; Kalil, N.; Villalba, L.; Hill, K.; Steinberg, S. M.; Figg,
W. D.; Tompkins, A.; Arbuck, S. G.; Sausville, E. A. J. Clin.
Oncol. 1998, 16, 2986.
4. Zaharevitz, D.; Kunick, C.; Schultz, C.; Meijer, L.; Leost,
M.; Gussio, R.; Senderowicz, A.; Lahusen, T.; Sausville, E.
Amer. Assoc. Cancer Res., April 10–14, 1999, Philadelphia,
PA.
5. (a) Vesely, J.; Havlicek, L.; Strnad, M.; Blow, J. J.;
Donella-Deana, A.; Pinna, L.; Letham, D. S.; Kato, J. Y.;
Detivaud, L.; Leclerc, S.; Meijer, L. Eur. J. Biochem. 1994,
224, 771. (b) Abraham, R. T.; Acquarone, M.; Andersen, A.;
Asensi, A.; Belle, R.; Berger, F.; Bergounioux, C.; Brunn, G.;
Buquet-Fagot, C.; Fagot, D.; Glab, N.; Goudeau, M.; Guer-
rier, P.; Houghton, P. J.; Hendriks, H.; Kloareg, B.; Lippai,
M.; Marie, D.; Maro, B.; Meijer, L.; Mester, J.; Mulner-Lor-
illon, O.; Poulet, S. A.; Schierenberg, E.; Schutte, B.; Vaulot,
D.; Verlhac, M. H. Biol. Cell 1995, 83, 105. (c) Schutte, B.;
Nieland, L.; van Engeland, M.; Henfling, M.; Meijer, L.;
Ramaekers, F. Exp. Cell. Res. 1997, 236, 4.
6. (a) Singh, P.; Kumar, R. J. Enzyme Inhib. 1998, 13, 125. (b)
Gibson, K. H.; Brundy, W.; Godfrey, A. A.; Woodburn, J. R.;
Ashton, S. E.; Curry, B. J.; Scarlett, L.; Barker, A. J.; Brown,
D. S. Bioorg. Med. Chem. Lett. 1997, 7, 2723. (c) Thompson,
A. M.; Murray, D. K.; Elliott, W. L.; Fry, D. W.; Nelson,
J. A.; Showalter, H. D.; Roberts, B. J.; Vincent, P. W.; Denny,
W. A. J. Med. Chem. 1997, 40, 3915. (d) Myers, M. R.; Setzer,
N. N.; Spada, A. P.; Persons, P. E.; Ly, C. Q.; Maguire, M. P.;
Zulli, A. L.; Cheney, D. L.; Zilberstein, A.; Johnson, S. E.;
Franks, C. F.; Mitchell, K. J. Bioorg. Med. Chem. Lett. 1997,
7, 421. (e) Myers, M. R.; Setzer, N. N.; Spada, A. P.; Zulli,
A. L.; Hsu, C. J.; Zilberstein, A.; Johnson, S. E.; Hook, L. E.;
Jacoski, M. V. Bioorg. Med. Chem. Lett. 1997, 7, 417.
7. Human foreskin fibroblasts (line AG01523) were purchased
from the N.I.A. Aging Cell Culture Repository, Coriell Insti-
tute for Medical Research (Camden, NJ). Cells were routinely
grown in Eagle’s MEM containing 20% FBS and a 2ꢂ con-
centration of essential and non-essential amino acids and
10. Rosenblatt, J.; DeBondt, H.; Jancarik, J.; Morgan, D. O.;
Kim, S.-H. J. Mol. Biol. 1993, 230, 1317.
11. Otwinowski, Z.; Minor, W. In Methods in Enzymology
Macromolecular Crystallography, part A; Carter, C. W., Jr.,
Sweet, R. M., Eds.; Academic: New York, 1997; Vol. 276, pp
307–326.
12. Brunger, A. T.; Kuriyan, J.; Karplus, M. Science 1987,
¨
234, 458.
13. Shewchuk, L.; Hassell, A.; Wisely, B.; Rocque, W.;
Holmes, W.; Veal, J.; Kuyper, L. F. J. Med. Chem. 2000, 43,
133.