Discovery of 5,6,7,8-tetrahydropyrido[3,4-d]pyrimidine inhibitors of Erk2

Article abstract of DOI:10.1016/j.bmcl.2014.04.068

The discovery and optimization of a series of tetrahydropyridopyrimidine based extracellular signal-regulated kinase (Erks) inhibitors discovered via HTS and structure based drug design is reported. The compounds demonstrate potent and selective inhibition of Erk2 and knockdown of phospho-RSK levels in HepG2 cells and tumor xenografts.

Full text of DOI:10.1016/j.bmcl.2014.04.068

Bioorganic & Medicinal Chemistry Letters 24 (2014) 2635–2639
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of 5,6,7,8-tetrahydropyrido[3,4-d]pyrimidine inhibitors
of Erk2
a
a
a
a
a
James F. Blake a, , John J. Gaudino , Jason De Meese , Peter Mohr , Mark Chicarelli , Hongqi Tian ,
Rustam Garrey a, Allen Thomas a, Christopher S. Siedem a, Michael B. Welch a, Gabrielle Kolakowski a,
Robert Kaus a, Michael Burkard a, Matthew Martinson a, Huifen Chen b, Brian Dean b, Danette A. Dudley b,
Stephen E. Gould b, Patricia Pacheco b, Sheerin Shahidi-Latham b, Weiru Wang b, Kristina West b,
Jianping Yin b, John Moffat b, Jacob B. Schwarz b
⇑
a Array BioPharma Inc, Boulder, CO 80301, USA
b Genentech Inc, 1 DNA Way, South San Francisco, CA 94080-4990, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The discovery and optimization of a series of tetrahydropyridopyrimidine based extracellular signal-reg-
ulated kinase (Erks) inhibitors discovered via HTS and structure based drug design is reported. The com-
pounds demonstrate potent and selective inhibition of Erk2 and knockdown of phospho-RSK levels in
HepG2 cells and tumor xenografts.
Received 11 March 2014
Revised 14 April 2014
Accepted 18 April 2014
Available online 29 April 2014
Ó 2014 Elsevier Ltd. All rights reserved.
Keywords:
Erk2 kinase inhibitors
Mek resistance
Structure based drug design
Inappropriate activation of the mitogen-activated protein
kinase (MAPK) signaling pathway is one of the key molecular driv-
ers in a large number of human cancers. Overexpression of growth
factor receptors and oncogenic Ras and Raf mutations, have been
found in cancers of the lung, breast, colon, pancreas, skin, thyroid
and blood.1–3 Deregulation of the MAPK pathway leads to over pro-
duction of phosphorylated extracellular-signal-regulated kinases,
Erk1/2 (pErk), which can then activate multiple cytosolic and
nuclear kinases, transcription factors and other proteins responsi-
ble for cell growth, proliferation, survival, angiogenesis and differ-
entiation.1,4–6
Given the highly specific nature of the MAPK/Erk signaling
pathway,7 and its prevalence in oncogenic progression, it is not
surprising that various components have proven to be attractive
targets for anti-cancer drug discovery. Additionally, it has recently
been shown that cells that have acquired resistance to Mek kinase
inhibitors retain sensitivity to Erk inhibitors.8 Thus there is consid-
erable potential for Erk inhibitors to both be efficacious and to
enhance/prolong efficacy in combination with inhibitors of
upstream pathway components in the clinical setting.
Herein we describe the discovery and optimization of a series of
selective Erk2 inhibitors that were found via combinatorial library
synthesis, high throughput screening (HTS), and structure based
drug design. Originally, compound 1 (Fig. 1) was synthesized as
part of a combinatorial library that targeted the ATP cleft of protein
kinases. The partial saturation of the bicyclic 6,6 framework and
incorporation of piperidino nitrogen provides a substantially lower
ClogP relative to the corresponding quinazoline core, which trans-
lates into significantly improved solubility properties. HTS screen-
ing identified 1 as a hit against Erk2 with an enzyme IC50 = 106 nM.
In addition to biochemical potency in an Erk2 kinase assay, com-
pounds were tested for their ability to inhibit phosphorylation of
a downstream Erk1/2 substrate, p90Rsk in a cellular context. In
vitro cellular potency of 1 was assessed by measurement of the
Erk-dependent phosphorylation of p90Rsk(Ser380) levels in the
HepG2 cell line (cell IC50 = 4.4
l
M).9
O
N
H
N
N
HN
N
O
1
⇑
Corresponding author. Tel.: +1 303 386 1262; fax: +1 303 381 6652.
E-mail address: jblake@arraybiopharma.com (J.F. Blake).
Figure 1. HTS hit compound 1 from combinatorial library.
http://dx.doi.org/10.1016/j.bmcl.2014.04.068
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

2636
J. F. Blake et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2635–2639
5. A final diversification step employed standard unsymmetrical
urea synthesis methodology to give a set of analogs 6.
Small hydrophobic substitution proved to afford modest
potency increases while maintaining good physical properties.
Both the 3-F and 3-Cl substitutions of the benzyl ring were nearly
equipotent (Table 1) as were the 4-Cl and 4-F compounds. The 3-Cl,
4-F substitution yielded a slightly more cell potent lead and
emerged as our starting point for further optimization efforts. This
aromatic substitution pattern was also described in previous SAR
for Erk2 inhibitors engaged in a similar interaction with the P-loop
of the kinase.10
Results from docking studies and SAR from previous efforts10
suggested that substitution of the benzylic carbon could access
nearby residues such as Ser153, Asn154, Cys166, and Asp167 in a
stereospecific fashion. The initial SAR in this region focused on
exploring primarily polar substituents. As can be seen, the incorpo-
ration of simple alcohols at the benzyl carbon (14–17) or amines
(20–21) improves both the enzyme and cell potency (Table 1). Both
the primary and secondary amines were well tolerated. Testing in a
limited panel of 56 kinases at 100 nM concentration revealed that
compound 14 inhibited (greater than 50% inhibition) Cdk2/Cyclin
A, and Gsk3b. In vitro ADME revealed compound 14 possessed
somewhat high predicted hepatic clearance in human and rat
microsomes (CL = 9, and 46 mL/min/kg, respectively).
Substitution of the pyrimidine via the amino linker afforded an
additional opportunity to improve the potency. The results of this
effort are presented in Table 2. The 4-THP analog (28) proved to be
the most potent in this series of compounds. The Cdk2 enzyme
IC50 = 16 nM for compound 28. Given that the most potent com-
pounds have exceeded the sensitivity limit of the Erk2 kinase assay
([Erk2] = 2.5 nM in the assay) the apparent selectivity ratios may
be significantly underestimated. Testing against a panel of 56
kinases at 100 nM concentration showed that compound 28 inhib-
ited (greater than 50% inhibition) Cdk2/Cyclin A, Cdk5, Hipk4, and
Figure 2. Compound 1 docked into the X-ray structure of Erk2.
Docking of compound 1 into an X-ray crystal structure of Erk2
suggested that the tetrahydropyridopyrimidine ring occupies the
hydrophobic cavity near the hinge region (illustrated in Fig. 2).
The pyrimidine ring accepts a hydrogen bond from the backbone
NH of Met108 while the amino group at the 2-position donates
an H-bond to the carbonyl oxygen of the same residue. Addition-
ally, the urea oxygen of 1 accepts two hydrogen bonds, one from
Lys54 and one from a bound water molecule that is complexed
to Gln105, Glu71, and Asp167. The latter H-bond is a key feature
of our strategy to achieve selectivity against other kinases includ-
ing members of the closely related CMGC family of kinases. Based
on the structure of 1 docked into Erk2, we prepared a series of
substituted benzyl analogs to probe the SAR of the nascent hit.
Analogs of compound 1 were prepared according to the route
outlined in Scheme 1. Primary amines 2 were initially converted
to bis-CBz protected guanidines 3 using commercially available
(E)-benzyl (1H-pyrazol-1-yl) methylenedicarbamate. These pro-
tected guanidines could be purified via silica gel chromatography
if necessary. Deprotection gave the mono-substituted guanidines
4 which were then reacted with the enamino-ketone shown (pre-
pared in one step from commercially available Boc-piperidin-3-one
and 1-t-butoxy-N,N,N0,N0-tetramethylmethanediamine) to give, after
Boc deprotection, a set of 2-substituted piperidinoaminopyrimidines
Pkcl. Unfortunately compound 28 still possesses somewhat high
predicted hepatic clearance in human and rat microsomes
(CL = 11, and 31 mL/min/kg, respectively).
During our lead optimization efforts we obtained an X-ray
structure of 28 bound to Erk2 at 1.95 Å resolution (Fig. 3).12 The
crystal structure confirmed that the tetrahydropyridopyrimidine
core and amino substitution interacts via a pair of H-bonds to
the hinge region of Erk2 at Met108. The longer methyl hydroxy
side chain interacts with the carboxylate side chain of Asp167
(2.69 Å). As we posited above, a bound water molecule forms an
H-bond to the urea carbonyl of 28 (3.03 Å), and also interacts with
the side chain of Lys54 (2.82 Å). The 4-THP interacts with Lys114
via the ether oxygen (3.18 Å), though it appears that the bulk of
the potency gain from this substituent is derived from the hydro-
phobic interactions in this region. The 3-Cl,4-F phenyl group occu-
pies a small hydrophobic pocket under the P-loop that is formed
when Tyr36 is displaced toward the C-helix.
z
B
C
HN
HN
R1
a
b
NH2
z
CB
R1
N
3
2
In an effort to improve the pharmacokinetics of this series of
compounds further we decided to explore various substitutions
at the 2-position of the tetrahydropyridopyrimidine ring and ben-
zylic carbon of the phenyl ring. The results of this SAR are summa-
rized in Table 3. The 3,4-dichloro substitution gave rise to some of
the most potent inhibitors of Erk2. However, this substitution pat-
tern also tended to possess higher predicted human clearance,
lower fold selectivity versus Cdk2 and generally higher human
PPB. A key difference between the structures of Erk2 and Cdk2 in
the hinge region is the presence of the larger Phe82 of Cdk2 versus
Leu107 in Erk2. Targeting this residue with larger substitutions is
expected to improve the selectivity of this class of compounds fur-
ther. Larger groups at the 4-position of the benzyl ring tended to
increase the selectivity versus. Cdk2 (e.g., –methoxy, and –CF3).
Interestingly, the 3-Cl,4-CN substitution of compound 42 revealed
NH2
,
N
c
e
d
+
N
N
HN
R1
NH
oc
B
O
oc
B
O
4
N
N
f
H
N
R1
N
R1
NH
N
N
R2
N
H
N
H
O
6
5
Scheme 1. Reagents and conditions: (a) 2, (E)-benzyl (1H-pyrazol-1-yl)methylen-
edicarbamate, THF, rt, 16 h; (b) Pearlman’s Catalyst, H2 (1 atm), THF, EtOH, rt, 16 h;
(c) 1-t-butoxy-N,N,N0,N0-tetramethylmethanediamine, toluene, rt, 16 h; (d) EtOH,
NaOEt, 45 °C, 18 h; (e) 2N HCl in dioxane, CH2Cl2, MeOH, rt, 16 h; (f) R2NH2, DIEA,
CDI, CH2Cl2, 1 h then add 5, then 2 h, rt.

J. F. Blake et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2635–2639
2637
Table 1
Tetrahydropyridopyrimidine Benzylic SAR
X
N
H
N
N
HN
N
Y
R1
O
Compound
R1
X
Y
Erk2 Inhibition IC50, nM
HepG2 Cell Inhibition IC50, nM
7
8
9
H
H
H
H
H
H
H
H
F
H
H
H
F
Cl
F
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
283
69
91
93
89
81
66
4
3
6
2
11
122
2
4
8
4384
6620
2138
2611
1856
1628
1216
160
88
200
108
368
2109
356
144
2757
917
Cl
H
H
Cl
F
F
F
F
F
F
F
F
F
F
F
10
11
12
13
14
15
16
17
18
19
20
21
22
23
CH2OH
CH(R-CH3)OH
CH(S-CH3)OH
CH(S-OH)CH2CH3
CH2OCH3
CH2OCH2CH2OCH3
CH2NH2
CH2NHCH3
CH2NHCH2CH2OH
CH2NHSO2CH3
7
performed in female NCR nude mice, male SD rats, male beagle
dogs, and male Cynomolgous monkeys. The results of these exper-
iments are summarized in Table 4.
Table 2
Tetrahydropyridopyrimidine Amine SAR
F
N
Given sufficient kinase selectivity and potency in inhibiting the
intracellular signaling activity of Erk1/2, we tested compounds for
their anti-proliferative activity against tumor cell lines carrying
Braf (V600E) and K-Ras activating mutations. It has been shown
that Mek inhibitors, which are expected to have the same mecha-
nism of action as Erk inhibitors, arrest proliferation at the G1
checkpoint.11 Proliferation of HCT116 (K-RasG12D) and A375 (B-
RafV600E) cell lines was inhibited by 38 with EC50s of 0.74 and
H
N
N
HN
R1
N
l
C
O
OH
Compound R1
Erk2 Inhibition HepG2 Cell
IC50, nM
Inhibition IC50, nM
24
25
26
27
28
29
30
31
32
33
34
35
36
37
CH3
CH2CH3
43
20
9
4
1
3
41
2
2255
753
2450
330
45
473
3009
198
348
13238
513
7438
247
0.39 lM, respectively. Figure 4 shows that the anti-proliferative
CH2CH2OH
Cyclopentyl
4-THP
(S)-2-Propan-1-ol
t-Butyl
(1S,3S)-3-Cyclopentanol
4-F phenyl
3,5-Dimethylisoxazol-4-yl 26
action of 38 on HCT116 cells was accompanied by an increase in
the G1 (2N-DNA) fraction and inhibition of Erk-mediated p90Rsk
Thr359/Ser363 phosphorylation.
Based upon its selectivity, overall physiochemical properties,
in vivo PK profile and moderate plasma protein binding (ca.
89.5% bound in mouse, and ca. 82.4% bound to human plasma
3
Isobutyl
Acetyl
10
93
6
1-Methyl-1H-pyrazol-4-yl
2-Methylpyridin-4-yl
7
445
a suprisingly significant drop in selectivity. Determination of the X-
ray structure of 42 bound to Cdk2 revealed that the cyano group
was able to accept an H-bond from the backbone NH of Tyr15,
located on the P-loop of Cdk2. Attempting to balance the potency
and PK properties of our compounds; we opted to revisit the 3-F,
4-Cl phenyl analog. Compound 38 proved to be potent with Erk2
IC50 = 2 nM, HepG2 cell IC50 = 25 nM and Cdk2 enzyme IC50 = 10
nM. The kinase selectivity profile of compound 38 was determined
against an expanded panel of 170 kinases; apart from Erk1 and
Erk2, only 2 other kinases, Cdk2/Cyclin A and PrkX, were inhibited
by greater than 75% at a test concentration of 100 nM (Supplemen-
tary material).
Next, the in vivo PK of 38 was evaluated in male CD-1 mice and
demonstrated that the clearance was moderate (CL = 33 mL/min/
kg) at 0.5 mg/kg iv, and oral bioavailability was acceptable
(F = 29%) at 2 mg/kg. More extensive PK evaluations were
Figure 3. X-ray structure of 28 bound to Erk2, solved at 1.95 Å resolution (PDB
code: 4O6E).12

2638
J. F. Blake et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2635–2639
Table 3
Tetrahydropyridopyrimidine SAR
X
N
H
N
N
HN
R2
N
Y
O
R1
Compounds R1
R2
X
Y
Erk2 Inhibition HepG2 Cell Inhibition Cdk2 Inhibition CLa, mL/min/
Human PPB%
Bound
IC50, nM
IC50, nM
IC50, nM
kg
Rat Human
38
39
40
CH2OH 4-THP
CH2OH 4-THP
Cl
Cl
F
Cl
2
1
2
25
17
21
10
11
69
23
34
52
8
11
15
82.4
89.7
Me
4-THP
ACH@CHANA
(indole)
41
42
43
44
45
46
47
48
49
50
51
Me
Et
Me
Me
Et
Et
Me
Et
4-THP
4-THP
Cl
CN
Cl
Cl
F
F
1
1
2
1
3
2
3
2
1
1
1
12
4
7
13
2
2
104
13
12
821
71
9
43
36
42
36
27
43
35
29
37
40
32
13
10
12
11
7
14
11
8
90.5
Cl
Cl
Cl
F
Cl
F
Cl
Cl
Cl
Cl
4S-(2R-CH2OH-THP)
(S)-2-propan-1-ol
(S)-2-propan-1-ol
(S)-2-propan-1-ol
(S)-2-propan-1-ol
(S)-2-propan-1-ol
4-(2-methyl-pyridyl)
21
51
23
110
14
23
35
43
97.0
77.4
97.5
98.7
91.9
99.4
98.2
99.0
Cl
CF3
OMe
Cl
Me
Me
Me
14
17
14
1-methyl-1H-pyrazol-4-yl Cl
1-methyl-1H-pyrazol-5-yl Cl
4
23
a
Hepatic CL predicted from the corresponding species microsomes.
protein) we elected to advance 38 into a dose escalation PK/PD
study in nude mice bearing subcutaneous human HCT116 colorec-
tal cancer xenografts (Fig. 5).
At the 1 and 8 h time points the plasma levels of 38 increased in
a linear fashion as the dose was increased from 75 to 150 mg/kg
once daily (QD), with little increase in exposure at the 150 mg/kg
twice daily (BID) dose. The corresponding pRsk knockdown at 1 h
was observed to be 75–80% starting at the 75 mg/kg dose. At the
8 h time point pRsk knockdown ranged from 20% (75 mg/kg) to
55% (150 mg/kg) QD dose. Based upon these results compound
38 was progressed into a proof-of-concept tumor growth inhibi-
tion (TGI) study.
Table 4
Key Pharmacokinetic parameters for 38
Species
Mouse
100
Rat
5
Dog
2
Cyno
3
Dose PO (mg/kg)
30
150
Multiple doses of 38 were administered BID by oral gavage in
nude mice bearing sub-cutaneous HCT116 colorectal cancer xeno-
grafts (Fig. 6). Tumor growth inhibition of 69%, 95% CI (30, 88), was
obtained with BID dosing of 38 for 28 days at 150 mg/kg. Thus,
compound 38 is efficacious against human HCT116 colorectal
Cmax
tmax (h)
t1/2 (h)
AUC0
(
l
M)
3.12
0.5
2.4
4.81
134
58
9.82
0.25
7.5
13.9
0.25
2.7
26.7
0.09
0.33
1.7
0.2
104
11
0.59
0.42
0.6
0.61
32
26
0.16
1.33
5.2
0.65
33
(l
M h)
30
to
1
CL (mL/min/kg) IVa
F (%)
73
64
19
a
IV clearance measured at 1 mg/kg.
Figure 4. Activity of 38 in HCT116 cells. (A) Inhibition of Erk-mediated p90Rsk
phosphorylation following treatment of cells for 2 h with the indicated concentra-
tions of 38. (B) Anti-proliferative activity as determined by high-content imaging
following 48 h treatment. (C) Accumulation of cells with 2N DNA content as
determined by high-content imaging following 48 h treatment.
Figure 5. PK/PD in HCT116 xenograft tumors for compound 38 at 1 and 8 h time
points for 75 mg/kg, 150 mg/kg, and 150 mg/kg BID in nude mice.

J. F. Blake et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2635–2639
2639
Additionally, the compounds display excellent knockdown of phos-
phoRsk levels in the HepG2 cell line and in human tumor xeno-
grafts. The more advanced compounds demonstrate excellent
selectivity combined with good ADME properties. While this pro-
file is encouraging, development as therapeutic agents will require
additional optimization efforts.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2014.
04.068.
References and notes
1. Dhillon, A. S.; Hagan, S.; Rath, O.; Kolch, W. Oncogene 2007, 26, 3279.
2. Platanias, L. C. Blood 2003, 101, 4667.
3. Hoshino, R.; Chatani, Y.; Yamori, T.; Tsuruo, T.; Oka, H.; Yoshida, O.; Shimada,
Y.; Ari-i, S.; Wada, H.; Fujimoto, J.; Kohno, M. Oncogene 1999, 18, 813.
4. Meloche, S.; Pouyssegur, J. Oncogene 2007, 26, 3227.
5. Hanahan, D.; Weinberg, R. A. Cell 2000, 100, 57.
6. Turjanski, A. G.; Vaque, J. P.; Gutkind, J. S. Oncogene 2007, 26, 3240.
7. Roberts, P. J.; Der, C. J. Oncogene 2007, 26, 3291.
8. Hatzivassiliou, G.; Liu, B.; O’Brien, C.; Spoerke, J. M.; Hoeflich, K. P.; Haverty, P.
M.; Soriano, R.; Forrest, W. F.; Heldens, S.; Chen, H.; Toy, K.; Ha, C.; Zhou, W.;
Song, K.; Friedman, L. S.; Amler, L. C.; Hampton, G. M.; Moffat, J.; Belvin, M.;
Lackner, M. R. Mol. Cancer Ther. 2012, 11, 1143.
Figure 6. HCT116 colorectal tumor xenograft study for compound 38 dosed at
150 mg/kg po BID. Changes in tumor volumes over time are depicted as cubic spline
fits generated via linear mixed effects analysis of log-transformed volumes.
9. Dalby, K. N.; Morrice, N.; Caudwell, F. B.; Avruch, J.; Cohen, P. J. Biol. Chem. 1998,
273, 1496.
cancer xenografts in vivo when dosed orally either QD or BID.
Overall body weight loss, including the vehicle control group,
was minimal (not statistically significant) in all groups. Doses of
50–150 mg/kg of 38 were generally well tolerated.
We have described the discovery and optimization of a series of
potent and selective Erk2 antagonists. The compounds display
potent enzyme and cell based inhibition of our primary targets.
10. Aronov, A. M.; Baker, C.; Bemis, G. W.; Cao, J.; Chen, G.; Ford, P. J.; Germann, U.
A.; Green, J.; Hale, M. R.; Jacobs, M.; Janetka, J. W.; Maltais, F.; Martinez-Botella,
G.; Namchuk, M. N.; Straub, J.; Tang, Q.; Xie, X. J. Med. Chem. 2007, 50, 1280.
11. Gysin, S.; Lee, S.-H.; Dean, N. M.; McMahon, M. Cancer Res. 2005, 65, 4870.
12. Atomic coordinates of the structure 28 bound to Erk2 were deposited with the
RCSB Protein Data Bank (PDB) under the accession code 4O6E. See
Supplementary material for methods.