Structure-based design of thienobenzoxepin inhibitors of PI3-kinase

Article abstract of DOI:10.1016/j.bmcl.2011.04.124

Starting from thienobenzopyran HTS hit 1, co-crystallization, molecular modeling and metabolic analysis were used to design potent and metabolically stable inhibitors of PI3-kinase. Compound 15 demonstrated PI3K pathway suppression in a mouse MCF7 xenogra

Full text of DOI:10.1016/j.bmcl.2011.04.124

Bioorganic & Medicinal Chemistry Letters 21 (2011) 4054–4058
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Structure-based design of thienobenzoxepin inhibitors of PI3-kinase
Steven T. Staben a, , Michael Siu a, Richard Goldsmith a, Alan G. Olivero a, Steven Do a, Daniel J. Burdick a,
⇑
Timothy P. Heffron a, Jenna Dotson a, Daniel P. Sutherlin a, Bing-Yan Zhu a, Vickie Tsui a, Hoa Le d, Leslie Lee b,
John Lesnick c, Cristina Lewis c, Jeremy M. Murray e, Jim Nonomiya c, Jodie Pang d, Wei Wei Prior b,
Laurent Salphati d, Lionel Rouge e, Deepak Sampath b, Steve Sideris c, Christian Wiesmann e, Ping Wu e
a Discovery Chemistry, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
b Small Molecule Translational Oncology, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
c Biochemical Pharmacology, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
d Drug Metabolism and Pharmacokinetics, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
e Structural Biology, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Starting from thienobenzopyran HTS hit 1, co-crystallization, molecular modeling and metabolic analysis
were used to design potent and metabolically stable inhibitors of PI3-kinase. Compound 15 demonstrated
PI3K pathway suppression in a mouse MCF7 xenograft model.
Received 5 March 2011
Revised 22 April 2011
Accepted 26 April 2011
Available online 13 May 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Structure-based design
PI3-kinase
Increased signaling resulting from somatic mutation of the
p110 subunit of PI3K and the lipid phosphatase PTEN provides evi-
O
N
a
dence for the importance of the PI3K signalling pathway in a variety
of human cancers.1 Previous reports from our group have detailed
the optimization of 4-morpholinothienopyrimidine PI3-kinase
inhibitors including GNE-493,2 GNE-477,3 and GDC-09414 (cur-
rently in clinical studies). Our continued effort toward the discovery
and development of inhibitors of the PI3K pathway for oncology
applications led to the identification of benzopyran 1 from a high-
1
(R = A, R2 = H, R3 = C)
GDC-0941
GNE-493
S
(R1 = B, R2 = H, R3 = D)
GNE-477 (R1 = A, R2 = Me, R3 = D)
N
R1
N
R3
R2
O
HO
N
Me
NSO2Me
NH
Me
throughput screen measuring single point inhibition of PI3K
a
A
B
M).5 Although a poor starting
S
F
(Fig. 1, confirmed PI3K IC50 = 0.254
a
l
N
O
point from the standpoint of potency relative to physicochemical
properties (c Log P = 4.1, LE = 0.39, LipE = 2.5, MW = 356),6 struc-
tural novelty of 1 made it a good candidate for further optimization.
A positional scan of aryl-substituents on the aniline amide re-
vealed the importance of o-halogens for inhibitory activity (Table
N
N
Me
NH2
N
C
D
HTS hit 1
: PI3Kα IC50 = 0.254 μM
1). Co-crystallization of 1f in PI3K
c suggested the ortho-halogen
Figure 1. Previously disclosed thienopyrimidine inhibitors of PI3-Kinase from
Genentech and thienobenzopyran HTS hit 1.
is located in a hydrophobic pocket defined by Ile831, Ile879 and
Lys833 (Fig. 2).7 It also indicated that the N-methyl aniline amide
bound in a cis-fashion.8 Analogs in which the 2-haloaryl ring was
replaced with branched alkyl substituents (e.g., 2 and 3) were sig-
nificantly less potent, perhaps because conformational preference
did not bias filling this hydrophobic pocket.
This crystal structure also provided additional paths for potency
improvement. The amide carbonyl of 1f (Fig. 2) is positioned with-
in water mediated H-bond distance to Tyr867 while the benzopy-
ran oxygen forms a key, but suboptimal, H-bond with the hinge
residue (3.6 Å). We anticipated that pyran ring-expansion would
provide an improved interaction with the hinge as well as remove
⇑
Corresponding author.
E-mail address: stevents@gene.com (S.T. Staben).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.04.124

S. T. Staben et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4054–4058
4055
Table 1
O
O
OH
c, d
Effect of amide substitution on P13aa IC50
a, b
( )n
n = 2
n = 3
n = 3
O
O
O
O
( )n
( )n
e-g
S
O
1f, 4, 5
S
n = 1
n = 2
n = 3
R
N
S
O
Me
O
N
MeO
Me
Compound
R
PI3Ka IC50 (lM)
Cl
1a
1b
1c
1d
1e
1f
1g
2
2-F-C6H4
3-F-C6H4
4-F-C6H4
2-Me–C6H4
0.254
ꢀ1.0
ꢀ1.0
0.699
0.673
0.108
>10
n
PI3Kα IC50 (μM) cLogP LipE
1f
1
2
3
0.108
0.024
>10
4.6
4.6
5.1
2.4
3
4
5
2-OMe–C6H4
2-Cl–C6H4
2-Pyridyl
iPr
--
Figure 3. Synthesis and biochemical potency of ring-expanded analogs.9 (a) base,
BrCH2CH2Br; (b) NaH; (c) POCl3, DMF; (d) HSCH2CO2Me, DMF; (e) LiOH, THF/H2O;
(f) SOCl2; (e) 2-Cl–N–Me–aniline, DIPEA, CH2Cl2. See Ref. 5 for a description of the
assay conditions.
5.2
17
3
iBu
a
See Ref. 5 for a description of the assay conditions.
Figure 4. Overlay of X-ray structures of benzoxepin 6 (orange) and benzopyran 1f
(blue) in PI3Kc (PDB 3R7R). Solvent exposed region omitted for clarity.
Figure 2. X-ray structure of 1f soaked into PI3Kc (PDB 3R7Q).
a benzylic ether methylene unit that could be prone to metabolic
oxidation. The synthesis of representative benzopyran and ring-ex-
panded analogs is presented in Figure 3. Alkylation of 2-hydroxy-
acetophenone with dibromoethane followed by sodium hydride
promoted cyclization gave the benzoxocanone. A two step thio-
phene synthesis proceeding through the b-chloroenal gave a thie-
no-benzoxepin (n = 2) and thienobenzoxocane (n = 3). Amide bond
formation was accomplished through intermediacy of acid chlo-
rides to give final compounds 1f, 4 and 5.
Unfortunately, benzoxepin 4 was cleared at a rate greater than
hepatic blood following intravenous administration to rat
(Clp = 60 mL/min/kg). Our strategy became potency improvement
with concomitant decreasing lipophilicity (progressive design dri-
ven by improving LipE).11 We hypothesized that judicious substi-
tution at the 8-position of the benzoxepin would improve
potency and decrease lipophilicity while simultaneously hindering
possible sites of metabolism.
Notably, improved potency against PI3Ka was achieved with N-
Although benzoxocane 5 displayed little inhibition of PI3Ka,
acyl, N-carbamoyl, carbamide, and 4-pyrazolyl substitution (6–9,
11 in Table 2). Tertiary amides, that is, 10, were less tolerated
and generally were considerably less potent. Unfortunately, these
lipophilicity-reducing modifications did not substantially improve
clearance in rat.
We hypothesized amide hydrolysis, para-oxidation, or de-
methylation of the electron-rich N-methyl-aniline in these analogs
might be responsible for the continually observed high clearance.12
A metabolite ID experiment in human hepatocytes for 9 indeed
identified significant hydrolysis of the aniline amide as well as an
benzoxepin 4 was ꢀ5-fold more potent than benzopyran 1f.10 This
potency difference was consistent across a comparison of direct
benzopyran/benzoxepin analogs. Notably, this change is calculated
to be lipophilicity neutral, and therefore provides an improvement
in lipophilic efficiency (LipE). Overlay of crystal structures of 1f and
6 (vide infra) in PI3Kc (Fig. 4) shows a shorter hydrogen bond and
better angle to hinge-residue Val882. Improved occupancy of the
binding pocket by an extra methylene unit could also be responsi-
ble for the potency increase against the PI3K isoforms.

4056
S. T. Staben et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4054–4058
Table 2
Benzoxepin 8- and 4-anilino substitution alters potency and clearance
R1
O
S
8
R2
X
O
N
Me
Cl
Compound
R1
H
R2
H
PI3K
16.3
a
Ki (nM)
Prolif. MCF7.1 IC50
2.9
(
lM)
Rat Clp (mL/min/kg)
60
Mouse Clp (mL/min/kg)
ND
c Log P
LipE
3.19
4
4.6
H
N
Me
6
H
1.8
0.807
72
ND
3.9
4.84
O
O
H
N
OMe
7
8
9
H
H
H
1.5
1.2
2.5
0.678
6.2
44
89
ND
ND
ND
ND
4.5
3.5
3.8
4.32
5.42
4.80
O
O
NH2
Me
1.1
N
H
O
10
H
31
>30
ND
ND
3.3
4.21
N
O
NH
N
11
H
—
0.6
12
ND
23
ND
ND
4.9
3.5
4.32
2.82
12a
H
473
ND
13a
—
65
3.3
ND
ND
3.7
3.49
NH
N
O
O
O
Me
Me
Me
14
2.5
0.719
28
24
3
5.60
N
N
H
H
O
Me
15
3.7
0.303
36
15
2.7
5.73
N
H
N
Me
X = C unless otherwise noted. a X = N. ND = not determined. Clearance values obtained from a discrete iv dose of 1 mg/kg or a cassette iv dose of 0.25 mg/kg. See Ref. 5 for a
description of the assay conditions.
oxidative metabolite resulting from de-methylation, aromatic oxi-
O
dation and glucoronindation of the aniline.13 We projected that
O
Me
steric and electronic modification of the aniline, reducing lipophil-
icity while blocking para-oxidation, would not only decrease clear-
ance but eliminate generation of these potentially toxic and
reactive species.14 Accordingly we synthesized 12 (c log P = 3.5),
with a 4-aminopyridine replacement for the aniline. As hypothe-
sized, 12 did display decreased clearance in rat (23 mL/min/kg),
N
H
1
, R = CONMe2, R2 = H
15
R2
S
16, R1 = F, R 2 = H
17, R1 = H, R2 = CONMe2
R1
O
N
Me
Cl
compound clogP Rat Clp
although lacked potency against PI3Ka which could not be suffi-
(mL/min/kg)
(mL/min/kg)
Mouse Clp
15
ciently recovered with the most potent 8-substitution (i.e., 13).
We next examined substitution on the aniline ring with electron-
withdrawing polar amide functionality (14, 15 Table 2). Gratify-
ingly, potency was maintained while lipophilicity was reduced
(lipE 15 = 5.73) and a corresponding decrease in rodent clearance
was observed. Interestingly, both the position and identity of the
aniline substituent affected rodent clearance (Fig. 5). For example,
transposition of the amide to the 5-position or replacement of the
amide with a halide did not improve metabolic stability (16, 17).
Interestingly, consistent decreased clearance with para-amide
substitution could be seen in vivo over a broad range of benzoxepin
inhibitors (Fig. 6). In addition to decreasing total clearance, the de-
creased lipophilicity with this substitution pattern (c log P = 2.7 for
15
16
17
2.7
4.0
2.7
36
153
66
182
55
Figure 5. Identity and position of aniline substitution affects rodent clearance.
Clearance values obtained from a discrete iv dose of 1 mg/kg or a cassette iv dose of
0.25 mg/kg.
15, vs 3.8 for 9 and 4.6 for 4) is likely responsible for a decreased cel-
lular/enzymatic potency shift observed with these analogs (Table 2).

S. T. Staben et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4054–4058
4057
120
100
80
60
40
20
0
120
100
GDC-0941
GNE-614
80
60
40
pAkt
pPRAS40
1 hr
pS6RP
pAkt
pPRAS40
6 hr
pS6RP
Rat Clp
(mL/min/kg
Figure 8. Comparison oral single dose PK/PD of 15 (GNE-614) and GDC-0941 at
50 mg/kg. Compound 15 shows significant pathway suppression in mice bearing
MCF7-neo/HER2 tumor xenografts indicated as pAKT, pPRAS40, pS6RP levels
O
R
relative to vehicle controls. At one hour [15]plasma = 7.38
6.72 M, [15]tumor = 1.08 M, [GDC-0941]tumor = 2.83 M; at 6 h [15]plasma
M, [GDC-0941]plasma = 4.50 M, [15]tumor = 2.24
M.14
l
M, [GDC-0941]plasma
=
=
S
l
l
l
X
11.10
l
l
lM, [GDC-0941]tumor
O
N
= 1.34
l
Me
20
Cl
series focused on improving bioavailability, removal of perceived
structural liabilities, and decreasing potency against the DNA-PK
counter-target. The results of these efforts will be disclosed in fu-
ture publications. These inhibitors also presented a novel vector
for exploring selectivity over PI3Kb.16
X ≠ CONR1R2
X = CONR1R2
Figure 6. Substructure versus Rat Clp. Rat Clp was measured by iv dose between
0.25 and 1 mg/kg solution. Some values are determined by cassette dosing.
Compound 16 (Clp = 153 mL/min/kg) omitted for clarity.
References and notes
in vitro profile for 15:
1. (a) Shayesteh, L.; Kuo, W. L.; Baldocchi, R.; Godfrey, T.; Collins, C.; Pinkel, D.;
Powell, B.; Mills, G. B.; Gray, J. W. Nat. Genet. 1999, 21, 99; (b) Samuels, Y.;
Wang, Z.; Bardelli, A.; Silliman, N.; Ptak, J.; Szabo, S.; Yan, H.; Gazdar, A.; Powell,
S. M.; Riggins, G. J.; Willson, J. K.; Markowitz, S.; Kinzler, K. W.; Vogelstein, B.;
Velculescu, V. E. Science 2004, 30, 554; (c) Parsons, D. W.; Wang, T. L.; Samuels,
Y.; Bardelli, A.; Cummins, J. M.; DeLong, L.; Silliman, N.; Ptak, J.; Szabo, S.;
Willson, J. K.; Markowitz, S.; Kinzler, K. W.; Vogelstein, B.; Lengauer, C.;
Velculescu, V. E. Nature 2005, 436, 792; (d) Chalhoub, N.; Baker, S. J. Annu. Rev.
Pathol. Mech. Dis. 2009, 4, 127; (e) Hennessy, B. T.; Smith, D. L.; Ram, P. T.; Lu, Y.;
Mills, G. B. Nat. Rev. Drug Disc. 2005, 4, 98; (f) Wee, S.; Lengauer, C.;
Wiederschain, D. Curr. Opin. Oncol. 2008, 20, 77; (g) Marone, R.; Cmiljanovic,
V.; Giese, B.; Wymann, M. P. Biochim. Biophys. Acta 2008, 111, 159; (h) Yap, T.
A.; Garrett, M. D.; Walton, M. I.; Raynaud, F.; de Bono, J. S.; Workman, P. Curr.
Opin. Pharmacol. 2008, 8, 393; (i) Crabbe, T.; Welham, M. J.; Ward, S. G. Trends
Biochem. Sci. 2007, 32, 460.
2. Sutherlin, D. P.; Sampath, D.; Berry, M.; Castanedo, G.; Chang, Z.; Chuckowree,
I.; Dotson, J.; Folkes, A.; Friedman, L.; Goldsmith, R.; Heffron, T.; Lee, L.; Lesnick,
J.; Lewis, C.; Mathieu, S.; Nonomiya, J.; Olivero, A.; Pang, J.; Prior, W. W.;
Salphati, L.; Sideris, S.; Tian, Q.; Tsui, V.; Wan, N. C.; Wang, S.; Wiesmann, C.;
Wong, S.; Zhu, B.-Y. J. Med. Chem. 2010, 53, 1086.
3. Heffron, T. P.; Berry, M.; Castanedo, G.; Chang, C.; Chuckowree, I.; Dotson, J.;
Folkes, A.; Gunzner, J.; Lesnick, J. D.; Lewis, C.; Mathieu, S.; Nonomiya, J.;
Olivero, A.; Pang, J.; Peterson, D.; Salphati, L.; Sampath, D.; Sideris, S.; Sutherlin,
D. P.; Tsui, V.; Wan, N. C.; Wang, S. M.; Wong, S.; Zhu, B. Y. Bioorg. Med. Chem.
Lett. 2010, 20, 2408.
PI3Kα IC50
PI3Kβ IC50
PI3Kδ IC50
PI3Kγ IC50
mTOR IC50
pAKT (PC3, S473):
Prolif EC50 (PC3):
Prolif EC50 (MCF7):
ppb (%, H/R/M/D/C):
:
4.6 nM
60 nM
:
:
1.7 nM
:
5.0 nM
:
530 nM
310 nM
550 nM
256 nM
97/97/98/93/91
PK profile for 15
species
Clp
(mL/min/kg)
F%
dose, formulation AUC
(μM h)
rat
mouse
dog
36
15
5
50
28
120
--
100 mg/kg, 60% PEG 39
100 mg/kg, MCT
2 mg/kg, 60% PEG
--
330
17
--
cyno
10
4. Folkes, A. J.; Ahmadi, K.; Alderton, W. K.; Aliz, S.; Baker, S. J.; Box, G.;
Chuckowree, I. S.; Clarke, P. A.; Depledge, P.; Eccles, S. A.; Friedman, L. S.; Hayes,
A.; Hancox, T. C.; Kugendradas, A.; Lensun, L.; Moore, P.; Olivero, A. G.; Pang, J.;
Patel, S.; Pergl-Wilson, G.; Raynaud, F. I.; Robson, A.; Saghir, N.; Salphati, L.;
Sohal, S.; Ultsch, M. H.; Valenti, M.; Wallweber, H. J. A.; Wan, N. C.; Wiesmann,
C.; Workman, P.; Zhyvoloup, A.; Zvelebil, M. J.; Shuttleworth, S. J. J. Med. Chem.
2008, 51, 5522.
Figure 7. In vitro and pharmacokinetic profile for 15 (GNE-614).
Compound 15 (GNE-614) was selected for further profiling
based on its acceptable cellular potency and in vitro metabolic sta-
bility. This inhibitor was active against all class I PI3-kinase iso-
forms, yet relatively inactive against mTOR (Fig. 7). Compound
15 displayed low to moderate clearance across species and moder-
ate to high bioavailability of suspension and solution oral doses. A
single dose PK/PD in nude mice bearing an MCF7-neo/HER2 tumor
xenograft demonstrated prolonged suppression of PI3K pathway
markers including pAKT, pPRAS40 and pS6RP (Fig. 8) comparable
to an equivalent dose of the clinical PI3-kinase inhibitor GDC-
0941 at a 6-h timepoint.15
In summary, structure-based design was used to optimize benz-
opyran HTS hit 1 to potent and metabolically stable 15. Impor-
tantly, 15 displayed significant suppression of PI3K-pathway
markers in vivo. Disappointingly, 15 was identified as being a
potent inhibitor of DNA-PK (IC50 = 6 nM). Future work within this
5. Enzymatic activity of the PI3K isoforms was measured using a competitive
displacement fluorescence polarization assay as described previously
(Sutherlin et al.).2 All IC50
s reported are geometric means of at least
duplicate measurements. Anti-proliferative activity in the MCF7.1 cell line
was measured following a 3-day incubation using Cell-titer Glo™. The MCF7.1
cell line is an in vivo selected line developed at Genentech and originally
derived from the parental human MCF7 breast cancer cell line (ATCC,
Manassas, VA). IC50s were determined using
geometric means of multiply replicates.
a 4-parameter fit and are
6. c Log P determined using biobyteÒ software. LE = ligand efficiency =
D
G/Nnon-hydrogen atoms; lipE = lipophilic efficiency = pIC50 (PI3Ka) À c Log P.
7. The halogen does not have clear density in these structures, however, modeling
suggests it occupies this hydrophobic pocket (1f and 6).
8. 8 of 9 N-methylaniline amides in the Cambridge structural database display a
‘cis’ conformation (most containing ortho halogens). 8 of 17 N-methyl-N-
isopropyl amides are in the ‘trans’ orientation. For discussion of solid-state
structures of N-methylaniline amides see: (a) Itai, A.; Toriumi, Y.; Saito, S.;
Kagechika, H.; Shudo, K. J. Am. Chem. Soc. 1992, 114, 10649; (b) Toriumi, Y.;

4058
S. T. Staben et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4054–4058
Kasuya, A.; Itai, A. J. Org. Chem. 1990, 55, 259; (c) Kagechika, H.; Himi, T.;
Namakawa, K.; Kawachi, E.; Hashimoto, Y.; Shudo, K. J. Med. Chem. 1989, 32,
2292. and references therein.
12. MetasiteÒ metabolism prediction software predicted N-de-methylation and
para-aniline-oxidation of 4 to be major oxidative metabolites.
13. Cryopreserved human hepatocytes, [11]i = 5 lM. The major metabolite by UV
9. For a detailed description of synthetic work related to this manuscript please
see: Do, S.; Goldsmith, R.; Heffron, T.; Kolesnikov, A.; Staben, S.; Olivero, A.; Siu,
M.; Sutherlin, D. P.; Zhu, B.-Y.; Goldsmith, P.; Bayliss, T.; Folkes, A.; Pegg, N.
WO/2009/123971.
10. We also saw a qualitative improvement in chemical stability as benzopyran
analogs often decomposed after prolonged storage in DMSO.
11. Leeson, P. D.; Springthorpe, B. Nat. Rev. Drug Disc. 2007, 6, 881; For an example
of a lead optimization program analyzed by progressive lipE optimization see:
Ryckmans, T.; Edwards, M. P.; Horne, V. A.; Correia, A. M.; Owen, D. R.;
Thompson, L. R.; Tran, I.; Tutt, M. F.; Young, T. Bioorg. Med. Chem. Lett. 2009, 19,
4406.
254 was identified to be the amide hydrolysis product.
14. For information regarding reactivity and potential toxicity of metabolites of
anilines see: (a) Blagg, J. Annal. Rep. Med. Chem. 2006, 41, 353; (b) Hop, E. C. A.;
Kalgutkar, A. S.; Soglia, J. R. Annal. Rep. Med. Chem. 2006, 41, 370; (c) Kalgutkar,
A. S.; Dalvia, D. K.; O’Donnell, J. P.; Taylor, T. J.; Sahakian, D. C. Curr. Drug Met.
2002, 3, 379.
15. The delay in pathway knockdown for 15 versus GDC-0941 is likely due to
different rates of absorption. 15 dosed at 50 mg/kg in MCT vehicle in nude mice
had a Tmax ꢀ3 h. GDC-0941 dosed at 50 mg/kg in MCT vehicle in nude mice had
a Tmax <30 min. MTD for 15 was not determined in this study.
16. Heffron et al. in preparation.