Design and synthesis of a novel pyrrolidinyl pyrido pyrimidinone derivative as a potent inhibitor of PI3Kα and mTOR

Article abstract of DOI:10.1016/j.bmcl.2012.05.100

Lead optimization efforts that employed structure base drug design and physicochemical property based optimization leading to the discovery of a novel series of 4-methylpyrido pyrimidinone (MPP) are discussed. Synthesis and profile of 1, a PI3Kα/mTOR dual inhibitor, is highlighted.

Full text of DOI:10.1016/j.bmcl.2012.05.100

Bioorganic & Medicinal Chemistry Letters 22 (2012) 5098–5103
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design and synthesis of a novel pyrrolidinyl pyrido pyrimidinone derivative
as a potent inhibitor of PI3Ka and mTOR
Phuong T. Le a, , Hengmiao Cheng a, Sacha Ninkovic a, Michael Plewe a, Xiaojun Huang a, Hai Wang a,
Shubha Bagrodia b, Shaoxian Sun b, Daniel R. Knighton a, Caroline M. LaFleur Rogers a, Andrew Pannifer d,
Samantha Greasley a, Deepak Dalvie c, Eric Zhang c
⇑
a Cancer Chemistry, Pfizer Worldwide Research & Development, La Jolla Laboratories, Pfizer Inc., La Jolla, CA 92121, United States
b Oncology Research Unit, Pfizer Worldwide Research & Development, La Jolla Laboratories, Pfizer Inc., La Jolla, CA 92121, United States
c PDM, Pfizer Worldwide Research & Development, La Jolla Laboratories, Pfizer Inc., La Jolla, CA 92121, United States
d Pfizer Worldwide Research & Development, Sandwich Laboratories, Pfizer Inc., UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Lead optimization efforts that employed structure base drug design and physicochemical property based
optimization leading to the discovery of a novel series of 4-methylpyrido pyrimidinone (MPP) are dis-
Received 4 April 2012
Revised 22 May 2012
Accepted 29 May 2012
Available online 6 June 2012
cussed. Synthesis and profile of 1, a PI3K
a/mTOR dual inhibitor, is highlighted.
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
PI3K
a
mTOR
Cancer
Kinase inhibitor
Dual inhibitor
Methylpyrido pyrimidinone
Physicochemical property based
optimization
Cancer remains an area with high unmet medical need. In re-
cent years, there has been a surge in treating cancer via a targeted
approach of inhibiting protein kinases. A number of these protein
kinase inhibitors including receptor tyrosine kinase inhibitors gef-
itinib (IressaÒ), erlotinib (TarcevaÒ), and multiple kinase inhibitor
sunitinib (SutentÒ) have been successful entries into the market.1
Our work in this area was to find a clinical candidate that would
class IV PI3K kinase and regulates cellular growth and prolifera-
tion.5 mTOR exists in two different complexes: mTORC1, a rapamy-
cin sensitive complex signaling to SK6 and 4E-BP1, and mTORC2, a
rapamycin insensitive complex signaling to AKT.6 Because the
kinase domain of mTOR is homologous to the p110
a subunit of
PI3K , a compound that inhibits both PI3K and mTOR can be
a
a
designed. This compound then can offer great potential in treating
cancer. Several PI3K/mTOR dual inhibitors, including Pfizer’s
PF-04691502 (Fig. 1),7 are currently in clinical trials for the
inhibit both phosphatidylinositol 3-kinase alpha (PI3Ka) and
mammalian target of rapamycin (mTOR). PI3Ks are members of
the intracellular lipid kinases that regulate cell metabolism, sur-
vival, growth and profileration.2 There are 3 classes in the PI3K
family and within each class different isoforms exist. Class I is
divided into sub families A & B. Growth factor receptor tyrosine
kinases activate Class IA PI3Ks, and G-protein coupled receptors
activate Class IB. Class IA PI3K comprises heterodimers of p85 reg-
ulatory subunit and p110 catalytic subunit. The PI3KCA (phospha-
tidylinositol-3-kinase, catalytic), is a-polypeptide gene which
N
O
NH
N
N
N
H2N
N
N
O
O
H2N
N
N
N
O
encodes p110a subunit. Many human cancers, for example, cervi-
PF-4691502
cal, head, breast, colon, carry mutations in PI3KCA.3,4 mTOR is a
1
HO
⇑
Corresponding author.
E-mail address: phuong.le@pfizer.com (P.T. Le).
Figure 1. Compound 1 and Pfizer’s clinical candidate PF-4691502.
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.05.100

P. T. Le et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5098–5103
5099
Table 1
In vitro activities and physical properties of MPP analogs
d
Compd
Structure
cLogDa
MW
351
mTOR Kib (nM)
Mouse PI3K
1.56
a
Kic (nM)
S437pAKT IC50 (nM)
LipEe
5.16
HLM ERf
0.26
Sol/l
Mg
N
O
N
2
2.63
142
16.1
3.94
H2N
N
N
N
O
O
NH
N
N
12
2.11
310
18.6
0.436
2.61
6.48
0.62
11.9
H2N
N
a
logD prediction at pH 7.4 based on a model built by using Analiza experimental data.11
For a complete description of these assays, please see Ref. 7.
LipE = pIC50 À clogD.
b,c,d
e
f
HLM ER: Human liver microsome extraction ratio (
Kinetic apparent solubility at pH 6.5 from Analiza.
lL/min/million).
g
O
O
R2 NH2
Br
O
7
N
O
N
N
4
N
c
R1
N
8
NH
R2
R1
N
6
NH
R2
b
a
R1
N
5
NH
R2
R1
N
3
Cl
4
R3
Br
N
N
N
2
f
e
d
R1
N
9
N
O
R1
N
10
N
O
R1
N
N
O
8
R2
R2
R2
11
Scheme 1. The synthesis of MPP series exemplified as in 11. Reagents and conditions: (a) DIPEA, K2CO3, DMAC; (b) NBS, CHCl3; (c) tri-o-tolylphosphine, Pd(OAc)2, Et3N; (d)
DBU, KOtBu, DMA; (e) NBS, DMF; (f) R3-boronic acid, K2CO3, PdCl2(PPh3)2, DMF, H2O.
treatment of cancer. In this Letter, the design and synthesis of a no-
vel series of PI3K/mTOR dual inhibitors, exemplified by 1, will be
discussed.
removing potentially inefficient lipophilicity was applied to follow
up 2. Inspection of the co crystal structure of the MPP derivative
with PI3Kc indicated that the methoxypyridine ring nitrogen
High throughput screening (HTS) followed by hit to lead devel-
opment yielded 27 a 4-methylpyrido pyrimidinone (MPP) deriva-
tive, as an interesting lead. Compound 2 exhibited potent Ki
formed a key hydrogen bond with a conserved water molecule in
the kinase selectivity pocket, and the SAR suggested that the
hydrogen bond contributed significantly to the binding affinity.
Replacement of the methoxypyridine with a pyrazole would both
lower clogD and maintain the hydrogen bond with the conserved
water molecule. Hence, 12 was designed. Indeed, 12 boosted LipE
(1.6 nM) against mouse PI3Ka, a surrogate of human PI3Ka used
in the in vitro assay, and moderate Ki (142 nM) in the biochemical
assay using human mTOR kinase domain (Table 1). In the BT20 cell
assay, measuring inhibition of AKT phosphorylation at S473, com-
pound 2 demonstrated excellent cellular potency with an IC50 of
16 nM. Illustrated in Scheme 1 is the general synthetic route for
preparing MPP derivatives.
Compound 2 demonstrated excellent kinome selectivity, arising
from the tight binding of the methyl at the 4 position of the MPP
core in a small, unique hydrophobic pocket near the hinge region
of PI3K and mTOR. The co crystal structure of a MPP derivative
analogous to 2 bound in PI3Kc8 was described comprehensively
by Cheng et al.7 For compound design purposes, we used this struc-
ture as a model for PI3K
not available. Although 2 had good PI3K
ited moderate mTOR activity, poor solubility (3.9
ate lipophilic efficiency (LipE) of 5.16 based on PI3K
a
because crystal structures of PI3K
enzyme activity, it exhib-
M), and moder-
cell IC50
a were
a
l
a
(Table 1). In general, a better LipE value leads to better in vivo
properties through a combination of higher in vitro potency and
lower lipophilicity.9 Because LipE = pIC50 À clogD, both higher cell
potency and lower clogD will lead to an increased LipE. Addition-
ally, lowering clogD will also yield increased solubility.10 Conse-
quently, the medicinal chemistry strategy of lowering clogD by
Figure 2. Co crystal structure of 12 with PI3Kc.

5100
P. T. Le et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5098–5103
by more than one unit and improved solubility. Its PI3K
was improved by more than fivefold from 16 to 2.6 nM and its
mTOR Ki activity also improved by sevenfold from 142 to 19 nM.
a
potency
derivatives while maintaining low MW, low clogD and good LipE,
several compounds, listed in Table 3, with N8–N bond instead of
N8–C bond, were synthesized. Compound 18, with a bulky diethyl-
Co crystal structure of 12 with PI3K
c
protein confirmed the hydro-
amine side chain at N8 position, is the least active, with PI3Ka Ki of
gen bond between pyrazole ring nitrogen and the conserved water
molecule (Fig. 2). The cyclopentyl group also formed hydrophobic
interactions with the protein. Because the mTOR crystal structure
is not available, it is challenging to rationalize the increase in
mTOR potency for 12 at the molecular level. However, 12 also
exhibited increased human liver microsome (HLM) clearance, with
extraction ratio (ER) of 0.6, compared with ER of 0.26 for 2.
General medicinal chemistry design principles for improving
in vitro metabolic stability guided us to design analogs with lower
clogP (or clogD).10 Following the synthetic route described in
Scheme 1, a small series of analogs with lower clogD, were de-
signed to help us determining the best R3 group for subsequent
optimization effort. In a head to head comparison between the
methoxypyridine and pyrazole R3 groups, 13 versus 14 and 15 ver-
sus 16, both LipE and HLM ER ratio were better for the pyrazole
derivatives than for the methoxypyridine derivatives. Interestingly,
replacing the cyclopropyl group in 16 with an isopropyl group, as
in 17, caused an increase in the in vitro potency, while maintaining
HLM ER of 16. When 17 was studied in rat liver microsomal stabil-
138 nM, mTOR Ki of 663 nM, and cell IC50 of 2260 nM. Replacing
the diethyl amine with isobutyl amine as in 19, results in a moder-
ate increase in potency. Compound 1, with a pyrrolidine at N8
position, demonstrated an excellent overall profile: clogD of 1.6,
potent Ki against both PI3Ka (12.5 nM) and mTOR (10.6 nM), PI3Ka
cell IC50 of 7 nM, and a very high LipE (6.55). The measured pKa of 1
is 3.7. Compared to 12, compound 1 exhibited similar LipE how-
ever with a threefold increase in solubility and a twofold more
stable in in vitro HLM clearance. Compared to the original lead 2,
compound 1 showed a better LipE and exhibited dual inhibitor pro-
file with better solubility (Table 3). Before designing 1, we per-
formed a literature search on N8 aminoalkyl substituted pyrido
pyrimidinones and found very few disclosures of this structural
class. We were aware that such an unprecedented motif could
potentially be associated with metabolic instability and/or reactive
metabolites. However, given the novelty of the species, we decided
to make the compound and evaluate it. We performed a P450 cat-
alyzed metabolic study on 1 using human liver microsome with
and without GSH and found no GSH adduct. The major metabolite
formed was identified as compound 35 (MW = 343). The proposed
pathways to form 35 were depicted in Scheme 2.
Illustrated in Scheme 3 is the synthetic route for preparing 1,
which commenced with commercially available amino pyrimidine
20. Bromination of 20 with bromine afforded 21 in 93% yield. The
2-amine group in 21 was protected by converting to the dimethyl
pyrrole derivative 23, followed by hydrazine displacement in
30 min to generate compound 24 in 74% yield. This intermediate
set the stage for the N–N bicyclic ring formation. The bicyclic
ity assay, it was highly cleared at 349 ll/min/mg. Although these
compounds with very low molecular weight (MW) exhibited
attractive LipE, they did not have the desired overall attributes to
warrant progression to in vivo studies (Table 2).
Nevertheless, we were encouraged by the data obtained from
these analogs which suggested that MPP derivatives with low
MW and a pyrazole head piece can achieve good in vitro potency
against both PI3Ka and mTOR, and desirable metabolic stability
in the HLM assay. In order to design structurally novel MPP
Table 2
In vitro activities and physical properties of the substituted pyrazole and N-methoxyl pyridine molecules
Compd
Structure
cLogD
MW
mTOR Ki (nM)
Mouse PI3K
a
Ki (nM)
S437pAKT IC50 (nM)
LipE
5.27
HLM ER
0.83
Sol/lM
N
O
N
13
2.31
337
76.4
3.70
26.5
11.1
H2N
N
N
O
NH
N
N
14
15
2.02
2.24
296
323
51.2
7.00
16.7
5.76
4.51
0.66
0.59
13.0
16.2
H2N
O
N
N
N
O
N
N
N
ND
26.0
176
H2N
H2N
H2N
N
N
N
N
N
N
O
O
O
NH
N
16
17
1.56
1.88
282
284
190
36.0
13.5
399
4.83
6.14
0.26
0.25
16.7
43.4
NH
N
23.5
9.64

P. T. Le et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5098–5103
5101
Table 3
In vitro activities of the MPP analogs
Compd
Structure
cLogD
MW
313
mTOR Ki (nM)
Mouse PI3K
138
a
Ki (nM)
S437pAKT IC50 (nM)
LipE
4.0
HLM ER
NA
Sol/
lM
NH
N
N
18
1.66
663
2260
434
H2N
N
N
N
O
NH
N
N
N
19
1.14
313
311
352
370
123
10.6
ND
827
12.5
10.0
2.00
867
4.9
0.6
NA
H2N
N
N
HN
O
NH
N
1
1.6
7.01
75.7
36.6
6.55
5.03
5.3
0.35
43.1
H2N
N
N
N
O
N
O
N
N
33
34
2.12
2.14
ND
ND
O
H2N
N
N
N
N
N
O
F
272
0.27
44.0
O
H2N
N
N
NH
N
NH
NH
N
N
N
N
N
H2N
N
N
N
O
H2O
- H2O
H2N
N
N
N
O
H2N
N
N
N
O
OH
[O]
[O]
1
NH
N
NH
N
N
N
H2N
N
N
NH
O
O
H2N
N
N
N
O
O
[O]
N
[O]
NH
N
H2N
N
N
O
NH
CO2H
35
Scheme 2. P450 catalyzed metabolism of 1 in human liver microsome.
hydrazine 26 was prepared by alkylating 24 with 1,4-dibromobu-
tane 25. To build the third ring via Heck reaction, amidation of
26 with acryloyl chloride afforded 27 in 99% yield. Several
conditions (e.g., Pd(PPh3)4, with different bases (e.g., TEA, DIPEA,
KOAc, Ag2CO3) were screened for the intramolecular Heck reaction
of 27 to find the best conditions. And the use of Ag2CO3 as the base

5102
P. T. Le et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5098–5103
O
Br
Cl
a
+
Br
Cl
N
b
c
N
N
N
N
23
H2N
N
Cl
H2N
N
O
22
21
20
Br
N
N
Br
e
N
d
Cl
Br
N
N
N
26
NH
N
Br
N
N
24
NH
NH2
O
25
Br
N
N
N
g
f
N
N
N
N
O
N
29
N
N
O
H2N
N
30
N
N
O
N
N
27
N
N
O
28
NH
N
Br
O
N
i
N
h
O
N
N
O
H2N
N
31
N
N
O
B
H2N
N
N
N
O
O
1
32
Scheme 3. Synthesis of 1. Reagents and conditions: (a) Br2, CH2Cl2, RT (93%); (b) p-TsOH, toluene, reflux 24 h (67%); (c) NH2NH2, DIPEA, DMAC, microwave 100 °C, 30 min
(74%); (d) DIPEA, DMAC, 60 °C, over night (64%); (e) DIPEA, CH2Cl2, RT (99%); (f) Pd(PPh3)4, Ag2CO3, 70 °C, THF, 3 h (30%); (g) NH2OHÁHCl, EtOH, H2O (70%); (h) Br2, TEA, CCl4,
DMF, RT (60%); (i) PdCl2(PPh3)2, P(C(CH3)3)3BF4, CsF, DMAC, H2O, 125 °C microwave (40%).
was the key to achieve 30% isolated yield for 28.12 In this reaction,
the exo intermediate 29 was not observed and isolated. Deprotec-
tion of the dimethyl pyrole followed by bromination gave 31. Final-
ly, Suzuki coupling of 31 with boronic ester 32 at 125 °C under
microwave conditions gave the desired product 1 in 40% yield.
To determine the SAR in the kinase selectivity pocket, analogs of
1 with different head groups including methoxypyridine and 3-flu-
oro-methoxypyridine were prepared following the above described
synthetic route. The data for these compounds is summarized in
Table 3. With 3-fluoromethoxypyridine as the head piece, com-
methoxyl pyridine group with the pyrazole generally gives higher
LipE inhibitors.
The SAR discussed in this Letter is summarized by the LipE plot
in Figure 3. Starting from 2, with a LipE of 5.16, by reducing MW
and strategically lowering clogD by removing lipophilicity and
incorporating structural novelty, compound 1, with higher LipE
(6.55) and greater solubility, was discovered.
Co crystal structure of 34 bound with PI3Kc was determined
(Fig. 4).13 The aminopyrimidine formed key hydrogen bonds with
the hinge residue Val 882. 3-fluoromethoxypyridine bound in the
selectivity pocket, and the pyridine ring nitrogen formed a hydro-
gen bond with the conserved water molecule. From the cocrystal
structure, the pyrrolidine ring bound in a similar way to how the
cyclopentyl ring in 12 bound in the ribose binding pocket. This
could explain why the LipE of 12 and 1 are very similar. The benefit
from lower clogD was seen mostly in solubility and metabolic
stability.
pound 34 is a PI3Ka selective inhibitor. Both methoxy pyridine 33
and fluoro methoxy pyridine 34 have lower LipE value than 1 (Table
3). This is consistent with the earlier finding that replacing the
Figure 3. LipE plot of clogD versus p(IC50) of PIK
a.
Figure 4. Co crystal structure of 34 bound to PI3Kc.

P. T. Le et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5098–5103
5103
Table 4
References and notes
Rat pharmacokinectic profile of 1
1. Cui, J. Expert. Opin. Ther. Patents 2007, 17, 1035.
Compd
IV (2 mg/kg)
Oral (5 mg/kg)
2. (a) Engelman, J. A.; Luo, J.; Cantley, L. C. Nature 2006, 7, 606; (b) Fruman, D. A.;
Myers, R. E.; Cantley, L. C. Nat. Rev. Genet. 1998, 67, 481; (c) Katso, R.;
Okkenhaug, K.; Ahmadi, K.; White, S.; Timms, J.; Waterfield, M. D. Annu. Rev.
Cell Dev. Biol. 2001, 17, 615.
Cl (ml/min/kg)
36.3
Vdss (L/kg)
T1/2 (hour)
Cmax (ng/ml)
F (%)
52
1
2.9
0.9
0.144
3. (a) Samuels, Y. et al Science 2004, 304, 554; (b) Bader, A. G.; Kang, S.; Zhao, L.;
Vogt, P. K. Nat. Rev. Cancer 2005, 5, 921.
4. (a) Isakoff, S. J. et al Cancer Res. 2005, 65, 10992; (b) Samuels, Y. et al Cancer Cell
2005, 7, 561.
5. (a) Vivanco, I.; Sawyers, C. L. Nat. Rev. Cancer 2002, 2, 489; (b) Strimpakos, A. S.;
Karapanagiotou, E. M.; Saif, W. W.; Syrigos, K. N. Cancer Treat. Rev. 2009, 35,
148.
6. Sarbassov, D. D.; Ali, S. M.; Sabatini, D. M. Curr. Opin. Cell Biol. 2005, 17, 596.
7. Cheng, H.; Bagrodia, S.; Bailey, S.; Edwards, M.; Hoffman, J.; Hu, Q.; Kania, R.;
Knighton, D. R.; Marx, M. A.; Ninkovic, S.; Sun, S.; Zhang, E. MedChemComm
2010, 1, 139.
Compound 1 exhibited excellent kinase selectivity from Pfizer’s
in-house kinase selectivity panel of 43 kinases. Compound 1 was
progressed to in vivo rat PK studies. Although 1 exhibited low
HLM ER, it was highly cleared in rat liver microsome (RLM) with
ER of 0.88. The rat PK profile of 1 is summarized in Table 4. It
exhibited moderate clearance, moderate Vdss, T1/2 of 0.9 h, and
good oral bioavailability.
In conclusion, we reported the design, synthesis and SAR of a
novel series of MPP derivatives. Through integration of SBDD and
physical properties based optimization, several parameters, most
notably LipE and solubility, were optimized. Compound 1 exhib-
8. We could not obtain crystal structure of PI3K
PI3K
a. Therefore we used the close-in
c
.
9. Edwards, M. E.; Price, D. A. In Annual Reports in Medicinal Chemistry; Macor, J. E.,
Desai, M. C., Eds.; Elsevier, 2010; Vol. 45, pp 380–390.
10. Johnson, T. W.; Dress, K. R.; Edwards, M. Bioorg. Med. Chem. Lett. 2009, 19, 5560.
11. In this protocol Analiza’s standard two phase system plates were used. Octanol
in equilibrium with universal buffer (0.01 M each of acetic, boric, and
phosphoric acids with 0.15 M NaCl) at pH 7.4 was used to prepare
partitioning plates for the assay. DMSO stock solutions were added to each
partitioning plate to a final concentration of 10% DMSO. The plates were sealed,
vortexed on Analiza’s specially designed deep well plate mixer, and
centrifuged to aid in phase settling. The assay was conducted on the ADW
workstation using chemiluminescent nitrogen detection. Concentration in the
phases is reported in ppm.
ited potent activity against PI3Ka and mTOR, high LipE, good ki-
nase selectivity and robust ADMET properties. In the in vivo rat
PK studies, 1 demonstrated good oral bioavailability.
Acknowledgments
12. Dounay, A. B.; Overman, L. E. Chem. Rev. 2003, 103, 2945.
13. The pdb codes for compound 12 (4FA6) and 34 (4FAD) were deposited with the
RCSB Protein Data Bank (PDB) database.
The authors would like to thank Drs. Simon Bailey and Martin
Edwards for helpful discussion.