Tricyclic azepine derivatives: Pyrimido[4,5-b]-1,4-benzoxazepines as a novel class of epidermal growth factor receptor kinase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2005.12.018

A novel class of pyrimido[4,5-b]-1,4-benzoxazepines is described as inhibitors of epidermal growth factor receptor (EGFR) tyrosine kinase. Two compounds display potent EGFR inhibitory activity of less than 1 μM in cellular phosphorylation assays (IC₅₀ 0.47-0.69 μM) and are highly selective against a small kinase panel. Such compounds demonstrate anti-EGFR activity within a class that is different from any known EGFR inhibitor scaffolds. They also provide a basis for the design of kinase inhibitors with the desired selectivity profile.

Full text of DOI:10.1016/j.bmcl.2005.12.018

Bioorganic & Medicinal Chemistry Letters 16 (2006) 1643–1646
Tricyclic azepine derivatives: Pyrimido[4,5-b]-1,4-benzoxazepines
as a novel class of epidermal growth factor
receptor kinase inhibitors
Leon Smith, II,a, Evgueni L. Piatnitski,a,* Alexander S. Kiselyov,a,à Xiaohu Ouyang,a
Xiaoling Chen,a Sabina Burdzovic-Wizemann,a Yongjiang Xu,a Ying Wang,b
Robin L. Rosler,b Sheetal N. Patel,b Hui-Hsien Chiang,b Daniel L. Milligan,a
John Columbus,a Wai C. Wong,a Jacqueline F. Doodyb and Yaron R. Hadarib
aDepartment of Chemistry, ImClone Systems, 180 Varick Street, New York, NY 10014, USA
bDepartment of Protein Science, ImClone Systems, 180 Varick Street, New York, NY 10014, USA
Received 3 November 2005; revised 4 December 2005; accepted 6 December 2005
Available online 18 January 2006
Abstract—A novel class of pyrimido[4,5-b]-1,4-benzoxazepines is described as inhibitors of epidermal growth factor receptor
(EGFR) tyrosine kinase. Two compounds display potent EGFR inhibitory activity of less than 1 lM in cellular phosphorylation
assays (IC50 0.47–0.69 lM) and are highly selective against a small kinase panel. Such compounds demonstrate anti-EGFR activity
within a class that is different from any known EGFR inhibitor scaffolds. They also provide a basis for the design of kinase inhib-
itors with the desired selectivity profile.
ꢀ 2005 Elsevier Ltd. All rights reserved.
Growth factor interactions with cell surface receptors in-
duce the activation of signaling pathways that mediate a
wide range of responses such as proliferation, survival,
differentiation, and metabolism. The epidermal growth
factor receptor (EGFR) has served as a paradigm in
the study of how these tyrosine kinase receptors func-
tion and the signal transduction pathways involved.
EGFR belongs to the ErbB family that includes EGFR
(ErbB1), HER2/neu (ErbB2), HER3 (ErbB3), and
HER4 (ErbB4).1 These receptors regulate intracellular
signaling pathways mediating cell proliferation, differen-
tiation, migration, survival, and adhesion.2 In oncogen-
esis, EGFR overexpression is a common feature in
tumors of epithelial origin, including cancers of the lung,
breast, head and neck, colon, and bladder. Moreover,
patients expressing high levels of EGFR usually have
poor prognosis.3 Finally, studies have implicated EGFR
in the development of several human tumors, thus estab-
lishing EGFR as a target for cancer therapy.1 Therefore,
a strategy to inhibit and suppress the unregulated activa-
tion of the tyrosine kinase domain of EGFR by target-
ing the ATP-binding pocket of EGFR with a small
molecule would be beneficial.
The work presented below addresses the ongoing need
to develop kinase inhibitors which have high activity
and efficacy while demonstrating improved safety pro-
files. The majority of EGFR inhibitors reported in the
literature belong to the quinazoline class as exemplified
by Iressa (1a),4 which has recently been approved for the
treatment of non-small cell lung cancer, but has been
shown to have major toxicity side-effects in a broad
patient population.5 An alternative compound series
would potentially have a different pharmacokinetic
and toxicity profile as well as kinase selectivity. In this
paper, we report our initial progress toward optimizing
a novel kinase inhibitor template. Specifically, we
describe preliminary structure–activity relationships of
pyrimido-1,4-benzoxazepines4,5 as EGFR tyrosine
kinase inhibitors (Fig. 1).
Keywords: Epidermal growth factor receptor; Kinase inhibitor; Ben-
zoxazepine; Kinase selectivity.
*
Corresponding author. Tel.: +1 718 473 5557; e-mail:
piatnits@gmail.com
Present address: Bristol-Myers Squibb, P.O. Box 5400, Princeton,
NJ 08543, USA.
à
Present address: Chemical Diversity, 1155 Sorrento Valley Road,
Suite 5, San Diego, CA 92121, USA.
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.12.018

1644
L. Smith, II et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1643–1646
improved activity over para analogs. Meta-bromo
R
F
(2i–2m) and meta-bromo-para-fluoro (2n–2r) com-
pounds showed enzymatic activity reaching 0.9 lM
and 15- to 30-fold increase in cellular potency when
compared to the initial HTS hit (2a). However, some
large groups such as a hydrophobic phenyl group are
not well tolerated at the meta position on the aniline
ring (see examples 2s, 2t).
NH
N
H
N
Cl
NH
N
O
HN
R
O
O
N
N
N
N
Br
O
N
(2a)
(1a)
(1)
Figure 1. Quinazoline general scaffold (1), Iressa (1a), and benzoxaze-
pine HTS hit (2a).
Another region that is amendable to structure–activity
relationship (SAR) studies is the benzene ring adjacent
to the seven-membered oxazepine ring. The availability
of a substituted hydroxyl aldehyde played a key role in
our initial SAR effort. We observed that, in general,
having a 7-methoxy, 8-methoxy or 8-halogen substituent
(2i–2r) was advantageous over a 9-methoxy substituent
(2u). Analogs substituted with other functional groups
such as carboxylic acid, amine, amide, and sulfonamide
groups (2v–2ab) were, on average, less active or equipo-
tent in enzymatic assay to 7-methoxy, 8-methoxy or
8-halogen-substituted compounds (2i–2r). The unsubsti-
tuted benzoxazepines (2ae) and (2af) demonstrated
potency similar to that of 7- or 8-substituted derivatives
(including methoxy analogs), suggesting that these two
positions are amenable to further modifications. This
observation indicates that both 7-methoxy and 8-meth-
oxy groups are favorable for the inhibition of kinase
activity.12 With the indication that two methoxy groups
on the benzene ring could improve activity, several
bis-methoxy analogs were prepared (Table 1). This
modification proved to be very fruitful, resulting in com-
pounds with 40- to 60-fold improvement in cellular
potency (2ag, 2ah). However, the 7,9-dimethoxy analog
(2ai) did not retain the activity seen with the 7,8-disub-
stituted compounds (2ag and 2ah). The cellular activity
of (2ag) proved to be superior to all previously prepared
benzoxazepine compounds. With this lead compound in
hand, future efforts will be directed toward refining the
benzoxazepine template.
High-throughput screening identified compound (2a) as
a small molecule inhibitor of EGFR with enzymatic
IC50 of 6 lM.6 The activity of (2a) in a cellular phos-
phorylation assay was shown to be weakly inhibitory
with an IC50 equal to 30 lM.7 The tricyclic structure
(2a) provides a novel framework to perform SAR stud-
ies. Analysis of the literature suggested the tricyclic core
of (2a) may play a vital role in positioning substituents
in the correct orientation (1).8 Hence, we kept the ben-
zoxazepine core unchanged while varying substituents
around it. Efforts therefore focused on substitution of
both the amine group on the pyrimidine ring and sub-
stituents on the benzoxazepine core benzene ring of
(2a) while retaining the seven-membered ring structural
feature. Access to such compounds was via the general
route summarized in Scheme 1, which is a modification
of previously reported procedures.9 5-Amino-4,6-dichlo-
ropyrimidine (3) was reacted with an aniline (4) resulting
in an intermediate (5) with moderate to good yields. A
variety of hydroxybenzaldehydes (6) were reacted with
(5) to provide benzoxazepine precursors (7). Reduction
of (7) led to the final product (2).10 We have prepared
a variety of benzoxazepine analogs (2) with some repre-
sentatives disclosed in Table 1.11
To explore how a substitution on the aniline phenyl
ring would affect activity, a variety of groups were
examined (Table 1). Noteworthy was the original
HTS hit (2a) and its direct analogs (2b, 2c) that possess
a benzylic amine rather than an anilino moiety seen in
more active analogs in Table 1. The unsubstituted com-
pounds (2d, 2e, and 2f) showed little activity, similar to
that seen in the original hit (2a), while several examples
of para substitution (2g, 2h) likewise exhibited discour-
aging results. In contrast, compounds with meta substi-
tution in the aniline ring demonstrated a dramatically
The synthetic route presented in Scheme 1 creates an
opportunity to examine some structural aspects of the
benzoxazepine core in EGFR inhibition. Intermediates
(7) were also isolated and tested for activity. The results
of enzymatic and cellular testing showed that derivatives
of the template (7) were virtually inactive in our
screens.8 The dramatic difference in activity between
dihydrobenzoxazepine analogs (2) and their precursors
(7) can be rationalized by looking at the overall geome-
try of the molecule. Compounds of general structure (7)
possess a rigidified tricylic core. Final products (2)
exhibit a less rigid structure that, apparently, is required
for the binding to the EGFR kinase ATP pocket. This
conclusion was further confirmed by the synthesis and
biological evaluation of compound (11) which contained
a benzoxazepine core having been rigidified by an amide
group (Scheme 2). The amide analog (11) showed no
activity when tested in both enzymatic and cellular as-
says compared to the benzoxazepine derivative (2ae).
The flexible benzoxazepine core structure desirable for
activity in our compound class is contrary to the quinaz-
oline template where a flat aromatic bicycle is generally
required for activity.12
OHC
R1
(6)
R2
HO
Cl
R1
NH
NH2
Cl
NH2
Cl
a
N
N
b
+
N
N
NH2
(3)
(5)
(4)
R1
R1
NH
N
NH
H
N
c
N
O
N
N
N
O
R2
R2
(7)
(2)
Scheme 1. Reagents and conditions: (a) DMF, 80 ꢁC; (b) NaH, DMF,
80 ꢁC; (c) LiBH4, THF.

L. Smith, II et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1643–1646
Table 1. Enzymatic and cellular activities of (2a–2ai) toward EGFR kinase6,7
1645
R1
NH
H
N
N
9
8
R 2
N
O
6
7
Compound
R1
R2
Enzymatic, IC50 (lM)a
Cellular phosphorylation, IC50 (lM)a
2a
2b
2c
CH2 Ph
CH2 Ph
8-Br
8-F
6.0
8.9
2.0
10.0
7.0
7.5
>10
5.8
1.0
3.4
1.5
3.0
3.1
2.0
1.0
1.0
1.0
0.9
2.0
>10
>10
>10
1.0
4.4
1.7
3.7
4.1
>10
5.7
4.9
3.4
1.1
0.34
1.2
6.3
30
>30
CH2 Ph
Ph
8-OCH3
8-Br
8-F
8.5
2d
2e
nt
nt
nt
nt
nt
Ph
Ph
2f
8-OCH3
8-F
2g
2h
2i
p-Cl Ph
p-Cl Ph
8-OCH3
8-Br
8-Cl
m-Br Ph
m-Br Ph
7.2
>30
6.9
2j
2k
2l
m-Br Ph
m-Br Ph
m-Br Ph
8-F
7-OCH3
8-OCH3
8-Br
1.9
1.0
8.4
4.3
7.2
8.9
2.4
9.9
9.8
2m
2n
2o
2p
2q
2r
m-Cl, p-F Ph
m-Cl, p-F Ph
m-Cl, p-F Ph
m-Cl, p-F Ph
m-Cl, p-F Ph
m-C6H5 Ph
m-C6H5 Ph
m-Br Ph
8-Cl
8-F
7-OCH3
8-OCH3
8-F
2s
2t
8-OCH3
9-OCH3
8-OCF3
8-COOH
8-COOH
8-NH2
8-NH2
8-NHSO2CH3
8-NHSO2CH3
2u
2v
nt
nt
m-Br Ph
m-Br Ph
2w
2z
>30
>30
>30
m-Cl, p-F Ph
m-Br Ph
2y
2z
m-Cl, p-F Ph
m-Br Ph
9.1
7.0
2aa
2ab
2ac
2ad
2ae
2af
2ag
2ah
2ai
m-Cl, p-F Ph
m-Br Ph
nt
8-NHCOCH@CH2
8-NHCOCH@CH2
H
5.4
8.9
m-Cl, p-F Ph
m-Br Ph
8.1
2.7
m-Cl, p-F Ph
m-Br Ph
H
7,8-(OCH3)2
7,8-(OCH3)2
7,9-(OCH3)2
0.47
0.69
m-Cl, p-F Ph
m-Br Ph
>30
nt, not tested; enzymatic activity is too low to be tested in cellular assay.
a The compounds showing greater than 5 lM enzymatic activity were not generally tested in cellular assay.
IGF1R, c-Met, Flt3, FGFR1, and c-Kit). We were
Br
O
OH
O
delighted to discover that most of the compounds in Ta-
ble 1 are highly selective for the EGFR kinase within
this cross-screening panel when tested at 10 lM com-
pound concentration. In particular, (2ag) selectively
inhibited EGFR over HER2, a closely related kinase
family member.12,13
(9)
Cl
Br
NH
N
Ph
NH2
NH2
Cl
NH2
Cl
N
a
N
b, c
N
(3)
(8)
Br
NH
H
Some compounds reported in this study display little
correlation between enzymatic activity and cellular
activity. A possible explanation for this discrepancy
could be poor solubility of test compounds at the cellu-
lar assay conditions. Thus, attaching solubilizing groups
at various positions (compare to quinazoline (1a)) could
improve both solubility and possibly change the selectiv-
ity profile to pick up cross-reactivity with some thera-
peutically relevant kinases. Our on-going efforts to
optimize the compound properties and to demonstrate
in vivo potency will be reported in due course. In
N
N
d
Br
NH
H
O
O
O
N
N
Cl
N
N
O
(10)
(11)
Scheme 2. Reagents and conditions: (a) DMF, 80 ꢁC; (b) oxalyl
chloride, DMF, THF, 0 ꢁC; (c) Et3N, THF; (d) HBr, AcOH, 80 ꢁC.
The compounds described in this manuscript were cross-
screened against a panel of tyrosine and serine/threonine
kinases (HER2, KDR, CDK2, Raf1, Flt1, InR, Akt,

1646
L. Smith, II et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1643–1646
Na2CO3 (0.2 M, 16 mL) and NaHCO3 (0.2 M, 34 mL),
and the pH was adjusted to 9.6. Prior to adding cell lysates
to the wells, the plates were washed 3 times with
PBS + 0.1% Tween 20 and blocked by 3% BSA in PBS
(200 lL for 1 h incubation). 80 lL of cell lysates was
transferred to the coated wells and incubated for 1 h at
4 ꢁC. After incubation, the plates were washed three times
with PBS + 0.1% Tween 20. To detect autophosphorylated
EGFR (tyrosine residues), 100 lL of anti-phosphotyrosine
antibodies (RC20:HRP, Transduction Laboratories) was
added per well (final concentration 0.5 lg/mL in PBS) and
incubated for 1 h. The plates were then washed six times
with PBS + 0.1% Tween 20. Enzymatic activity of HRP
was detected by adding 50 lL/well of equal amounts of the
Kirkegaard & Perry Laboratories (KPL) substrate A and
substrate B (KPL cat. #54-61-0). The reaction was
stopped by adding 0.1 N H2SO4 (50 lL/well) and absor-
bance was detected at 450 nm.
summary, the data presented here describe optimization
efforts of a novel series of EGFR tyrosine kinase inhib-
itors with good cellular potency (improved IC50 from
double-digit micromolar to nanomolar range) and
excellent selectivity against related kinases.13 As such,
benzoxazepine derivatives should serve as a useful
chemotype for kinase inhibitors and may provide a basis
for the design of novel therapeutic agents.14
References and notes
1. Blume-Jensen, P.; Hunter, T. Nature 2001, 411, 355.
2. (a) Mendelsohn, J.; Baselga, J. Oncogene 2000, 19, 6550;
(b) de Bono, J. S.; Rowinsky, E. K. Trends Mol. Med.
2002, 8, 19.
3. Brabenber, J.; Danenberg, K. D.; Metzger, R.; Schneider,
P. M.; Park, J.; Salonga, D.; Holscher, D.; Danenberg, P.
V. Clin. Cancer Res. 2001, 7, 1850.
8. Tsou, H.-R.; Mamuya, N.; Johnson, B. D.; Reich, M. F.;
Gruber, B. C.; Ye, F.; Nilakantan, R.; Shen, R.; Disca-
fani, C.; DeBlanc, R.; Davis, R.; Koehn, F.; Greenberger,
L. M.; Wang, Y.-F.; Wissner, A. J. Med. Chem. 2001, 44,
2719.
9. (a) Levkovskaya, L. G.; Sazonov, N. V.; Grineva, N. A.;
Mamaeva, I. E.; Serochkina, L. A.; Safonova, T. S.
Khimiya Geterotsiklicheskikh Soedinenii 1985, 1, 122; (b)
Yang, J.; Che, X.; Dang, Q.; Wei, Z.; Gao, S.; Bai, X. Org.
Lett. 2005, 7(8), 1541.
4. Bridges, A. J. Chem. Rev. 2001, 101, 2541.
5. (a) Lee, M. W.; Seo, C. W.; Kim, S. W.; Yang, H. J.; Lee,
H. W.; Choi, J. H.; Moon, K. C.; Koh, J. K. Acta Derm.
Venereol. 2004, 84(1), 23; (b) Elkind, N. B.; Szentpetery,
Z.; Apati, A.; Ozvegy-Laczka, C.; Varady, G.; Ujhelly, O.;
Szabo, K.; Homolya, L.; Varadi, A.; Buday, L.; Keri, G.;
Nemet, K.; Sarkadi, B. Cancer Res. 2005, 65(5), 1770;
(c) Van Zandwijk, N. Br. J. Cancer 2003, 89(Suppl. 2), S9.
6. See Ref. 7 for EGFR enzymatic assay conditions.
7. EGFR cellular phosphorylation assay: DiFi cells were
plated in 96-well Costar plates (2.5 · 105 cells per well) and
incubated for 4 h. The cells were then starved in serum-
free medium overnight. The next morning test compounds
were added to individual wells. Following a 2 h incuba-
tion, cells were lysed in 100 lL lysis buffer (150 mM NaCl,
50 mM Hepes, 0.5% Triton X-100, 10 mM NaPPi, 50 mM
NaF, 1 mM Na3VO4, and protease inhibitors, pH 7.5) and
rocked for 1 h at 4 ꢁC. The autophosphorylation level of
EGFR was then analyzed by ELISA using anti-phospho-
tyrosine antibodies. ELISA: 96-well ELISA plates were
coated with 100 lL/well of 1 lg/mL anti-EGFR antibodies
(Erbitux) and incubated overnight at 4 ꢁC. The anti-
EGFR antibodies were prepared in a buffer made with
10. Smith, II, L.; Hadari, Y. R. PCT Int. Appl. 2005009384,
2005.
11. All final compounds were characterized by H NMR and
1
MS.
12. Gibson, K. H.; Grundy, 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(21),
2723.
13. Rewcastle, G. W.; Denny, W. A.; Bridges, A. J.; Zhou, H.;
Cody, D. R.; McMichael, A.; Fry, D. W. J. Med. Chem.
1995, 38, 3482.
14. Pan, W.; Liu, H.; Xu, Y.-J.; Chen, X.; Kim, K. H.;
Milligan, D. L.; Columbus, J.; Hadari, Y. R.; Kussi, P.;
Wong, W. C.; Labelle, M. Bioorg. Med. Chem. Lett. 2005,
15, 5474.