Optimization of 4,6-bis-anilino-1H-pyrrolo[2,3-d]pyrimidine IGF-1R tyrosine kinase inhibitors towards JNK selectivity

Article abstract of DOI:10.1016/j.bmcl.2008.11.077

The SAR of C5′ functional groups with terminal basic amines at the C6 aniline of 4,6-bis-anilino-1H-pyrrolo[2,3-d]pyrimidines is reported. Examples demonstrate potent inhibition of IGF-1R with 1000-fold selectivity over JNK1 and 3 in enzymatic assays.

Full text of DOI:10.1016/j.bmcl.2008.11.077

Bioorganic & Medicinal Chemistry Letters 19 (2009) 360–364
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Optimization of 4,6-bis-anilino-1H-pyrrolo[2,3-d]pyrimidine IGF-1R tyrosine
kinase inhibitors towards JNK selectivity
Stanley D. Chamberlain a, Anikó M. Redman a, Joseph W. Wilson a, Felix Deanda b, J. Brad Shotwell a,
Roseanne Gerding a, Huangshu Lei a, Bin Yang a, Kirk L. Stevens a, Anne M. Hassell a, Lisa M. Shewchuk b,
M. Anthony Leesnitzer a, Jeffery L. Smith a, Peter Sabbatini c, Charity Atkins c, Arthur Groy c,
Jason L. Rowand c, Rakesh Kumar c, Robert A. Mook Jr. a, Ganesh Moorthy c, Samarjit Patnaik a,
*
a GlaxoSmithKline, Oncology R&D, 5 Moore Drive, Research Triangle Park, NC 27709, USA
b GlaxoSmithKline, Computational and Structural Chemistry, 5 Moore Drive, Research Triangle Park, NC 27709, USA
c GlaxoSmithKline, Oncology R&D, 1250 S. Collegeville Road, Collegeville, PA 19426, USA
a r t i c l e i n f o
a b s t r a c t
The SAR of C50 functional groups with terminal basic amines at the C6 aniline of 4,6-bis-anilino-1H-pyr-
rolo[2,3-d]pyrimidines is reported. Examples demonstrate potent inhibition of IGF-1R with 1000-fold
selectivity over JNK1 and 3 in enzymatic assays.
Article history:
Received 8 October 2008
Revised 19 November 2008
Accepted 20 November 2008
Available online 24 November 2008
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords:
IGF-1R
IGF-IR
JNK
Pyrrolopyrimidine
Aberrant signaling through the insulin-like growth factor-1
receptor (IGF-1R) has been implicated as a key process in tumor
progression and resistance to cancer therapy.1 Several small
molecular inhibitors which inhibit this pathway via binding to
the intracellular kinase domain have been reported.2 In the preced-
ing communication, we reported a series of novel 4,6-bis-anilino-
1H-pyrrolo[2,3-d] pyrimidines with a variety of 40 cyclic tertiary
amines at the C6 aniline as potent inhibitors of the IGF-1R tyrosine
kinase.3 A C40 isopropyl piperazine 1 was particularly noteworthy
with an optimal potency and pharmacokinetic profile.4 In a preli-
minary panel of 57 kinases 1 showed IC50 values P100 nM
(P50-fold selectivity) against 52 kinases, with IC50 measurements
for ALK, IGF-1R, IR, JNK1, and JNK3 at 0.5 nM, 2.0 nM, 1.6 nM,
13 nM, and 100 nM, respectively. When the isopropyl piperazine
substituent was moved to the C50 position (2) the IGF-1R potency
was retained (Table 1). Interestingly, the presence of an additional
C40 methyl substituent (3) dropped the IGF-1R enzyme potency by
10-fold. More importantly, the C50 isopropyl piperazine 2 also
seemed to impart superior selectivity against JNK1 and 3 over its
C40 counterpart 1.5
are activated by environmental stress and proinflammatory cyto-
kines like tumor necrosis factor-
and interleukin-1.6 The JNKs
a
modulate the activity of multiple transcription factors which regu-
late diverse cellular functions such as apoptosis, survival, and pro-
tein degradation,7 and have also been implicated in oncogenic
transformation.8 They are also intimately involved in cross-talk
with other signaling networks including the NF-j
B pathway.9 In
light of the complicating activities associated with JNK inhibition
we felt that the selectivity demonstrated by pyrrolopyrimidines 2
and 3 was preferable in order to understand the phenotypic effects
of inhibiting the IGF-1R signaling.
We constructed a docking model of 2 in complex with the IGF-
1R kinase domain to probe its binding mode (Fig. 1). As expected,
most of the inhibitor was predicted to have similar interactions
with the enzyme as 1 except the C50 4-isopropyl piperazine.10
The latter lay at the entrance of the ATP-binding site directed to-
wards the aD helix where it was predicted to participate in an ionic
interaction with the carboxylate of Asp1056. This ionic interaction
forced the backbone NH of Asp1056 to make unfavorable van der
Waals contact with the ethylene portions of the piperazinyl moi-
ety. Additionally, the N-isopropyl made contact with the hydroxyl
of Ser 1059. Thus, any gain in potency resulting from the ionic
interaction might be offset or negated by unfavorable polar-nonpo-
lar interactions. This may explain why 1 and 2 are essentially equi-
potent on IGF-1R enzyme. A similar docking model of 3 was also
The Jun N-terminal kinases (JNKs) are mitogen activated protein
kinases that play a well established role in immune response. They
* Corresponding author. Tel.: +1 919 4836229; fax: +1 919 4836053.
E-mail address: samarjitpatnaik@gmail.com (S. Patnaik).
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.11.077

S. D. Chamberlain et al. / Bioorg. Med. Chem. Lett. 19 (2009) 360–364
361
ties.3 Listed in Table 2 are analogs that were synthesized and
screened in our study. Notably, all compounds maintained greater
than 75-fold selectivity for IGF-1R over JNK1 and 311 (not shown in
Table 2) except 7, which still had a better selectivity than 1. Most
analogs with a C40 chloro or methyl substituent (5, 7, 9) seemed to
reduce the IGF-1R cellular activity by 4- to 10-fold compared to
their unsubstituted counterparts (4, 6, 8). However, the C50 N,N-
dimethylglycinamide containing molecules (10–13) displayed an
unanticipated IGF-1R activity profile with a toleration of methyla-
tion at C40 (11). On the other hand, methylation of the amide nitro-
gen proved deleterious toward cell potency (12 and 13),
emphasizing that rotational restrictions around the amide bond
may interfere with a favorable disposition of the basic amine in
the enzyme pocket. Compounds 10 and 11, which contained the
N,N-dimethylglycinamide substituent, had the most promising po-
tency and selectivity profiles with >5-fold improvement over the
Table 1
IGF-1R, JNK1, and JNK3 potencies for 1–3 (values represent an average of P2
individual measurements)
F
NH2
N
F
NH2
N
F
NH2
3''
O
H
O
H
O
H
N
4
N
6
N
6
N
6
N
N
HN
4'
N
N
HN
N
HN
N
N
H
H
O
H
O
O
2'
5'
N
5'
N
4'
N
N
1
2
3
N
Compound
IGF-1R
enzyme
IC50 (nM)
JNK1 enzyme
IC50 (nM)
JNK3 enzyme
IC50 (nM)
Phospho IGF-1R
cellular IC50 (nM)
Table 2
IGF-1R enzyme and Delfia, JNK1 results for 4–13 (values represent an average of P2
individual measurements)
1
2
3
2
4
20
13
3162
3891
100
5011
6310
117
201
270
H
H
F
N
N
3''
3
O
H
constructed (not shown), yielding a binding mode that was nearly
identical to that of 2 with one significant difference. The orienta-
tion of the C50 piperazinyl was rotated further out-of-plane with
respect to the C6 aniline compared to that of 2, perhaps to mini-
mize steric clash with the C40 methyl. This change disrupted the io-
nic interaction between the inhibitor and Asp1056, which was
perhaps reflected in its reduced IGF-1R enzyme IC50 value of
20 nM.
We decided to further evaluate the potential of the unique
interaction with Asp1056 by substituting a variety of functional
groups at C50 with terminal amines. We were particularly inter-
ested in addressing a hypothesis that such substituents might lead
to significant reductions in the inhibitory activity against JNK1 and
3 as evidenced by 2 and 3. We decided, at the outset, to retain the
optimal C20-methoxy aniline at C6 and the C300-fluoroanthranila-
mide at C4 to maintain selectivity and pharmacokinetic proper-
4
N
6
H
1
N
N
1'
N
H
O
R1
5'
4'
R2
Compound R1
R2
IGF-1R
enzyme
Phospho IGF-1R JNK1 enzyme
cellular
IC50 (nM)
IC50 (nM) IC50 (nM)
N
N
O
4
5
6
7
H
Cl
H
8
13
13
81
895
212
840
631
1258
5011
501
O
N
CH3 13
N
N
8
H
6
25
1
929
5366
28
1258
3981
O
O
N
N
9
Cl
H
O
10
11
12
3162
N
H
O
O
CH3
H
0.5
38
1995
N
N
N
H
50
377
10,000
N
CH3
O
Figure 1. Structural model of IGF-1R (carbon atoms in gray) in complex with 2
(carbon atoms in green). The 4-isopropyl piperazine’s methyl groups in 2 are
overlapped in this pose. Intermolecular H-bond and ionic interactions are
highlighted with yellow lines.
N
13
CH3 63
1728
10,000
N
CH3

362
S. D. Chamberlain et al. / Bioorg. Med. Chem. Lett. 19 (2009) 360–364
IGF-1R cellular potency of 2. The toleration of the C40 methyl by 11
prompted a further exploration of analogs where the C50 amide
nitrogen was constrained back onto C40. To that end we synthe-
sized indoline and tetrahydroquinoline precursors, as illustrated
in Scheme 1, and concentrated on amides with pendant amines
to expand on the SAR demonstrated by 10 and 11.
Regioselective nitration of commercially available 14 with ace-
tic anhydride/nitric acid followed by deacetylation provided multi-
gram quantities of 15. This pivotal intermediate provided an easy
access to multiple analogs with the indoline substructure. To that
end 15 was acylated with bromoacetyl bromide, and subjected to
bromide displacement with various cyclic and acyclic secondary
amines. Acylation of 15 with acryloyl chloride followed by amine
addition provided the corresponding one carbon homologated ana-
logs. Final reduction of the nitro functionality provided anilines 16
which could be installed into the pyrrolopyrimidine core via SNAr
chemistry described in detail in a previous communication.12 Syn-
thesis of the corresponding ring-expanded tetrahydroquinolines
started with nitration of lactam 17.13 Subsequent borane reduction
provided the intermediate tetrahydroquinoline 18 which was
derivatized to 19.
Table 3 lists the biological results from our systematic evalua-
tion of nitrogen substituents in the constrained ring systems. The
acetyl group of indoline 20 yielded a modest inhibitor in the IGF-
1R cellular mechanistic assay. Although the attachment of a pri-
mary amine (21) worsened cell activity, the selectivity over JNK1
improved.11 However, single digit nanomolar IGF-1R potencies
and three orders of magnitude selectivity over JNK1 were obtained
with the methyl amine in 22, small acyclic secondary amines (23,
24, 26), and small cyclic amines (27, 28). Reduction in amine basi-
city lowered either cell potency (25) or selectivity (30). Com-
pounds with an ethylene between the amide carbonyl and amine
(32–34), in general, exhibited reduced cellular potency than their
non-homologated congeners. Removal of the amide carbonyl from
23 in 35 lowered IGF-1R potency and selectivity against JNK1. The
SAR observed in the indoline series was reflected in the tetrahydro-
quinolines (36–39). A simple acetyl on the ring nitrogen (36) had
modest cell potency which was rescued by the dimethylated amine
or piperidinyl analogs 37 and 38, respectively. These compounds
had the most favorable activity and selectivity parameters. The
presence of an extra carbon atom between the amide and amine
in 39 reduced IGF-1R cell potency. Thus a variety of indolines
(22–24, 26–28) and tetrahydroquinolines (37, 38) had a similar po-
tency and selectivity profile as the unconstrained analogs 10 and
11.
Table 3
IGF-1R enzyme/Delfia, JNK1 results for 20–39 (values represent an average of P2
individual measurements)
Compound
R
IGF-1R
enzyme
IC50 (nM) IC50 (nM)
Phospho
IGF-1R cellular enzyme
JNK1
H
H
2''
F
N
N
N
3''
IC50 (nM)
O
H
N
H
N
N
H
O
O
O
O
20
21
22
23
O
H
6
4
2
1
278
765
79
126
1000
1995
1585
N
R
R
R
R
O
O
O
NH2
N
N
N
H
N
56
N
N
O
O
24
25
O
O
3
19
3162
794
N
N
R
R
O
N
50
877
O
O
26
27
O
O
2
1
85
24
1000
1259
N
N
N
N
N
R
R
A better understanding of the SAR depicted in Table 3 was
achieved when a co-crystal structure of 40 (Fig. 2A, structure in
white) in complex with an Insulin Receptor (IR) mutant (C981S,
D1132N) was solved, revealing the inhibitor’s binding mode.14,15
NO2
NH2
O
O
O
O
O
O
a, b
c, d
X
O
O
28
29
30
O
O
O
1
32
13
128
7762
72
1995
631
N
N
n
N
N
N
NH
N
R
R
R
14
17
15
16
NH2
n = 1, 2
X = O, N-iPr, CH2
NO2
N
O
O
O
N
O
X
e, f
c, d
N
NH
NH
N
n
O
18
19
O
126
N
Scheme 1. Reagents and conditions: (a) Ac2O, HNO3, 65%; (b) HCl, MeOH, reflux,
94%; (c) BrCH2COCl, K2CO3, THF or acryloyl chloride, Et3N, THF, then amine, 40–90%;
(d) Pd/C, H2, 65–95%; (e) NaNO2, TFA; 70%; (f) BH3ꢀDMS, THF, 70%.

S. D. Chamberlain et al. / Bioorg. Med. Chem. Lett. 19 (2009) 360–364
363
Table 3 (continued)
Compound
R
IGF-1R Phospho
enzyme IGF-1R
JNK1
enzyme
IC50 (nM)
H
H
2''
F
N
N
N
3''
IC50
cellular
O
H
(nM)
IC50 (nM)
N
H
N
N
H
O
N
31
O
32
8
376
191
298
651
794
1984
16
1585
794
N
N
N
R
O
O
32
N
N
N
R
R
R
O
O
O
N
N
33
6
251
O
N
O
34
4
1585
100
O
35
40
25
1
N
R
N
N
O
O
O
O
O
36
H
7943
6310
2512
7943
Figure 2. (A) Co-crystal structure of IR Double Mutant (C981S, D1132N, carbon
atoms in gray) in complex with 40 (carbon atoms in green). Intermolecular H-bond
interactions are highlighted with yellow lines. (B) Structural model of JNK1 (carbon
atoms in gray) in complex with 40. Intermolecular H-bond and ionic interactions
are highlighted with yellow lines.
R
O
37
N
N
R
R
R
N
N
N
Table 4
In vivo rat DMPK Properties for selected analogs (values represent the average of three
animals)
O
Compound
Dose
(mg/kg)
CL
po DNAUC (0–8 h)
(ng h/mL/mg/kg)
po Cmax
(ng/mL)
% F
38
1
11
(mL/min/kg)
iv
—
2.5
2.6
2.5
po
2
4
23
40
9.0
—
112
48
133
85
190
210
233
192
455
365
—
O
13.0
13.0
10.0
60
61
98
N
39
10
209
73
corresponded to Asp1056 in IGF-1R which formed an ionic interac-
tion with the ionizable isopropyl amine in 2 (see structural model
in Fig. 1). This interaction of the N,N-dimethyl glycinamide side
chain with the aspartate residue may explain why the carbonyl
group, its distance from the ionizable amine, and the nature of
the amine are important in maintaining a certain degree of activity
against IGF-1R. Modeling efforts were initiated to understand the
selectivity against JNK1 and 3. To that end a sequence alignment
of IGF-1R, IR, JNK1, and JNK3 was generated, which was then used
to superimpose X-ray structures of all four protein kinases, and a
plausible explanation was identified. If we assume a similar bind-
ing mode for 40 in JNK1 and 3,18 we first observed that an aspara-
gine (Asn114 in JNK1, Asn152 in JNK3) corresponded to the
Compounds 40 and 23 are closely related analogs differing by only
a single methyl on the C200 carboxamide at the C4 aniline.16 More-
over, like 23, 40 is a potent IGF-1R inhibitor (enzyme IC50 = 2 nM,
phospho IGF-1R cellular IC50 = 85 nM) which is also selective for
this kinase over JNK1/3 (enzyme IC50 = 5010/2000 nM).17 Given
the high sequence identity (ꢁ80%) between IGF-1R and IR in the ki-
nase domain, we reasoned that the binding mode of 40 in IGF-1R
would be nearly identical to that observed in IR. The crystal struc-
ture showed that 40 formed two favorable interactions with
Asp1083 (Fig. 2). One is an H-bond interaction of the amide car-
bonyl with the residue’s backbone NH; the second is an ionic inter-
action with the side chain carboxylate. Asp1083 in the IR mutant

364
S. D. Chamberlain et al. / Bioorg. Med. Chem. Lett. 19 (2009) 360–364
Asp1056 of IGF-1R (or Asp1083 for IR). We recognized that the
asparagine’s side chain carbonyl could form an H-bond with the
inhibitor’s protonated amine. However, as revealed by numerous
JNK X-ray structures, we found in our structural model (Fig. 2B)
that the carbonyl instead formed an internal H-bond interaction
with the enzyme’s backbone. In all cases, the amino group of the
asparagine was directed up towards the b1 strand, which would
place it in proximity to the inhibitor’s protonated amine, poten-
tially producing a non-favorable interaction. Thus the placement
of ionizable amines in the vicinity of the Asn leads to the observed
selectivity.
The in vivo rat pharmacokinetic properties of selected analogs
are listed in Table 4. While the lead piperazine 2 and the uncon-
strained analog 4 had reasonable oral exposure, the in vivo clear-
ance for 4 was quite high. Gratifyingly the indoline analogs 23
and 40 demonstrated better exposure (DNAUC) and higher maxi-
mum concentrations (Cmax) after oral administration. The excellent
oral bioavailability of 40, coupled with its potency and selectivity
profile, makes it a preferred candidate for advanced in vivo phar-
macodynamic studies.15,16
References and notes
1. For recent reviews see: (a) Hartog, H.; Wesseling, J.; Boezen, H. M.; van der
Graaf, W. T. A. Eur. J. Cancer 2007, 43, 1895; (b) Casa, A. J.; Dearth, R. K.;
Litzenburger, B. C.; Lee, A. V.; Cui, X. Front. Biosci. 2008, 13, 3906.
2. (a) Rodon, J.; DeSantos, V.; Ferry, R. J., Jr.; Kurzrork, R. Mol. Cancer Ther. 2008, 7,
2575; (b) Sarma, P. K. S.; Tandon, R.; Gupta, P.; Dastidar, S. G.; Ray, A.; Das, B.;
Cliffe, I. A. Expert Opin. Ther. Patents 2007, 17, 25.
3. Chamberlain, S. D.; Atkins, C.; Deanda, F.; Groy, A.; Kumar, R.; Leesnitzer, M. A.;
Lei, H.; Moorthy, G.; Patnaik, S.; Quinn, R.; Redman, A. M.; Rowland, J.;
Sabbatini, P.; Shewchuck, L.; Stevens, K. L.; Wilson, J. W.; Yang, B.; Shotwell, J. B.
Bioorg. Med. Chem. Lett. Preceding communication.
4. The IGF-1R enzyme and cellular assays have been described in the reference
section of the preceding communication: Chamberlain, S. D.; Atkins, C.;
Deanda, F.; Groy, A.; Kumar, R.; Leesnitzer, M. A.; Lei, H.; Moorthy, G.;
Patnaik, S.; Quinn, R.; Redman, A. M.; Rowland, J.; Sabbatini, P.; Shewchuck, L.;
Stevens, K. L.; Wilson, J. W.; Yang, B.; Shotwell, J. B. Bioorg. Med. Chem. Lett.
Preceding communication.
5. Description of the JNK1/3 enzyme assay protocols are provided in the
supplementary data section.
6. Davis, R. J. Cell 2000, 103, 239.
7. (a) Bogoyevitch, M. A.; Kobe, B. Microbiol. Mol. Biol. Rev. 2006, 70, 1061; (b) Barr,
R. K.; Bogoyevitch, M. A. Int. J. Biochem. Cell Biol. 2001, 33, 1047.
8. Nateri, A. S.; Spencer-Dene, B.; Behrens, A. Nature 2005, 437, 281.
9. Kracht, M. Anti-Inflamm. Anti-Allergy Agents Med. Chem. 2007, 6, 71.
10. A description of the interactions of 1 with the kinase domain of IGF-1R and IR
has been presented in the preceding communication.
In summary, detailed investigation of C50 substituents of 4,6-
bis-anilino-1H-pyrrolo[2,3-d]pyrimidines revealed a set of com-
pounds with nanomolar inhibition of the IGF-1R tyrosine kinase
in enzymatic and cellular assays. An unprecedented interaction
of ionizable terminal amines at C50, sensitive to conformational
flexibility, with aspartate residue 1056 in IGF-1R was discov-
ered. The amide carbonyl of the N,N-dimethyl glycinamide ana-
logs participated in an additional H-bond with the backbone NH
of the aspartate residue. These basic amine containing func-
tional groups also imparted a 1000-fold selectivity over JNK
which was not attained by previously reported C40 substituted
analogs. Molecular modeling shed light on a possible explana-
tion for this selectivity. The examination of rodent pharmacoki-
netics of selected analogs provided promising compounds,
especially 23 and 40, which could be used orally for biological
characterization.
11. Trends in the data for JNK3 were very similar to that of JNK1. Only the JNK1
data is reported in Tables 2 and 3 for brevity.
12. Chamberlain, S.; Lei, H.; Patnaik, S.; Gerding, R.; Redman, A.; Stevens, K.;
Wilson, J.; Yang, B.; Shotwell, J.; Moorthy, G. J. Org. Chem. 2008, 73, 9511.
13. Iyobe, A.; Uchida, M.; Kamata, K.; Hotei, Y.; Kusama, H.; Harada, H. Chem.
Pharm. Bull. 2001, 49, 822.
14. IR protein was expressed and purified as previously described in Li, S.; Covino,
N. D.; Stein, E. G.; Till, J. H.; Hubbard, S. R. J. Biol. Chem. 2003, 278, 26007.
Protein at 10 mg/mL was complexed with a 3-fold molar excess of inhibitor for
1 h prior to crystallization. Crystals were grown by hanging drop vapor
diffusion at 22 °C from 0.1 M MOPS, pH 7.0, 1.0 M trisodium citrate. Crystals
were flash frozen in PFO prior to data collection. The structure was solved by
molecular replacement using PDB/1P14 as a starting model and refined to an R-
factor of 20% at 2.1 Å using REFMAC. Crystallographic data for the structure A
in Figure 2 have been deposited at PDB/3EKK.
15. The insulin receptor (IR) enzyme and phospho IR cellular IC50 values for 40
were 1.8 nM and 79 nM, respectively. The pyrrolopyrimidines described in this
communication were, to all intents and purposes, equipotent against IR. The
toxicity implications of not having selectivity toward IR have been addressed
with respect to compound 40 in an account that has been recently submitted
to Cancer Research. The article includes the measurement of metabolic
endpoints after oral administration of various doses of 40 in mice.
16. With similarly substituted C6 anilines the N-methyl C200 carboxamides
displayed the same SAR trends as the unsubstituted C200 carboxamides listed
in Tables 2 and 3.
Acknowledgments
The authors thank Julie Mosley, Giorgia Vicentini, and Emma
Jones for the crystal structure in the supporting information.18
17. Estimates of the impact of plasma protein binding on overall potency for 23
and 40 could be made by carrying out the phospho IGF-1R cellular assay (see
reference section in preceding communication) in the presence of 2% HSA and
0.1% AAG. Under these conditions, the observed IC50 values for 23 and 40 were
487 and 310 nM, respectively.
Supplementary data
18. There is a crystal structure of a 4,6-bis-anilino-1H-pyrrolo[2,3-d]pyrimidine
derivative in complex with JNK1 (PDB/3ELJ) which is consistent with the
structural model in Figure 2, B. Please see supplementary data.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2008.11.077.