Discovery of novel isothiazole inhibitors of the TrkA kinase: Structure-activity relationship, computer modeling, optimization, and identification of highly potent antagonists

Article abstract of DOI:10.1016/j.bmcl.2006.04.003

The design, synthesis, and biological evaluation of potent inhibitors of the TrkA kinase is presented. A homology model is created to aid in the enhancement of potency and selectivity of isothiazole inhibitors found during a high-throughput screen. Three different syntheses are utilized to make diverse analogs within this series. Aminoheterocycles are found to be good urea surrogates, whereas bicyclic substituents on the C3 thio group were found to be extremely potent TrkA inhibitors in kinase and cell assays.

Full text of DOI:10.1016/j.bmcl.2006.04.003

Bioorganic & Medicinal Chemistry Letters 16 (2006) 3444–3448
Discovery of novel isothiazole inhibitors of the TrkA kinase:
Structure–activity relationship, computer modeling, optimization,
and identification of highly potent antagonists
Blaise Lippa,* Joel Morris, Matthew Corbett, Tricia A. Kwan, Mark C. Noe,
Sheri L. Snow, Thomas G. Gant, Melchiorra Mangiaracina, Heather A. Coffey,
Barbara Foster, Elisabeth A. Knauth and Matthew D. Wessel
Pfizer Inc., PGRD Groton, MS-8220-2203 Eastern Point Rd., Groton, CT 06340, USA
Received 25 February 2006; revised 31 March 2006; accepted 3 April 2006
Available online 24 April 2006
Abstract—The design, synthesis, and biological evaluation of potent inhibitors of the TrkA kinase is presented. A homology model
is created to aid in the enhancement of potency and selectivity of isothiazole inhibitors found during a high-throughput screen.
Three different syntheses are utilized to make diverse analogs within this series. Aminoheterocycles are found to be good urea
surrogates, whereas bicyclic substituents on the C3 thio group were found to be extremely potent TrkA inhibitors in kinase and cell
assays.
ꢀ 2006 Elsevier Ltd. All rights reserved.
Prostate and pancreatic cancers are two of the most
common forms of cancers in the US and will kill an
estimated 62,000 Americans in 2005.1 In the case of
pancreatic cancer, little progress has been made over
the past 50 years in the treatment of this particularly
painful disease, which has a 5-year survival rate of
4.6% from first diagnosis.1 Prostate cancer is the most
common non-cutaneous malignancy in men in the US,
and when tumors become resistant to anti-hormonal
therapy, other non-hormone related treatment options
are severely limited.2 Clearly, new therapies are urgently
needed to treat these diseases.
thought to play a key role in the progression of those
diseases. In the case of pancreatic cancer, increased
expression of TrkA also correlates with an increased
level of pain. Staurosporine Trk inhibitors from
Cephalon have showed excellent preclinical anti-tumor
efficacy6,7 and have entered human clinical trials.8
However, few Trk inhibitors are known and very few
show kinase selectivity.9–11
The therapeutic implications of an effective Trk inhibi-
tor may well go beyond cancer therapy. The link of
the Trks and their ligands with CNS related processes
has implicated this signaling cascade with a range of
other diseases, most notably pain.12 Because of the
therapeutic promise associated with inhibiting TrkA,
and the relative lack of potent and selective inhibitors,
a high-throughput screen was initiated which revealed
that several isothiazoles were effective in inhibiting the
TrkA kinase. Herein is reported the synthesis and
analysis of several series of potent and selective TrkA
kinase inhibitors.
TrkA is a part of a family of receptor tyrosine kinases
that also includes TrkB and TrkC. These kinases are
receptors for the neurotrophin family of ligands, which
includes nerve growth factor (NGF), brain-derived neu-
rotrophic factor (BDNF), and neurotrophins 3–5.3
Members of this family of ligands and receptors are
found to be up-regulated in a variety of human cancers,
particularly prostate4 and pancreatic cancers,5 and are
To better understand how the original lead isothiazole
inhibitor (1) would bind to TrkA, a homology model
of the kinase domain was constructed from insulin
receptor (IR-pdb code: 1IRK). The homology model
was built by first aligning the primary sequences of
Keywords: TrkA; Antagonist; Isothiazole; Homology model;
Synthesis.
*
Corresponding author. Tel.: +1 860 715 4660; fax: +1 860 686
6258; e-mail: blaise.lippa@pfizer.com
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.04.003

B. Lippa et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3444–3448
3445
addition of p-methoxybenzylthiol with malonitrile. This
intermediate is then reacted with PhOCONCS followed
by ring closure with iodine to result in the isothiazole 3.
Thiol deprotection is accomplished with mercury ace-
tate, TFA, and hydrogen sulfide. The cyanide is then
hydrolyzed with sulfuric acid to provide the key interme-
diate 4. Functionalization of the thiol is accomplished
under mildly basic conditions with benzyl halides, and
urea formation occurs through the addition of amines
to result in final analogs 5a–h (Table 1).
H2N
O
5
H
H
Cl
3
N
S
N
S
N
O
1
Figure 1. Lead isothiazole inhibitor of the TrkA kinase, found through
a high-throughput screen.
TrkA with IR, and then constructing the 3D structure of
TrkA based on the 3D structure of IR, using the homol-
ogy model interface within the MOE software suite.13
Figure 2 depicts the active site of TrkA with 1 in a puta-
tive binding conformation. The unsubstituted carbox-
amide of 1 makes two key hydrogen bonds with the
same residues of the protein backbone where ATP
would normally interact. This proposed binding mode
is supported by previous SAR studies that have shown
that replacements for this carboxamide such as CN or
tetrazole, or even a methylcarboxamide are universally
not tolerated. This carboxamide also appears to rigidify
the inhibitor through an intramolecular hydrogen bond
with the internal nitrogen of the urea. The urea appears
to be pointed toward a solvent exposed region, whereas
the p-chlorobenzyl substituent is directed into a portion
of the protein where the sugar of ATP normally binds.
These orientations are in accord with SAR observations
at C5 that lipophilic urea substituents have decreased
potency.
The original lead 1 was shown to be a potent inhibitor in
the TrkA kinase assay14 and also relatively potent in the
TrkA cell assay.15 Unfortunately, it also had some ki-
nase selectivity issues, particularly the VEGFR-2 ki-
nase16 where it was only 4· selective. Substitution on
the C3 benzylic carbon produced remarkable effects in
potency and particularly VEGFR-2 kinase selectivity.
Substitution with methyl or ethyl at this position re-
tained kinase potency and enhanced cellular TrkA
potency. Substitution with i-Pr was tolerated, whereas
n-Pr, CF3, and Ph substitution resulted in a loss of
potency. Importantly, all substituents at this position
showed a large increase in VEGFR-2 kinase selectivity
from 4· (R = H, 1) to as high as 111· when R = Et
(5b). For synthetic ease, all compounds were originally
synthesized as racemates. However, it was possible to
utilize chiral chromatography to separate enantiomers.
Thus, 5b was separated to provide the R enantiomer
5c and the S enantiomer 5d. The R enantiomer (5c)
was determined to be >10· more potent than the S
enantiomer, and VEGFR-2 kinase selectivity was in-
creased to 1300·. Presumably, these benzylic substitu-
ents are filling the lipophilic pocket on TrkA, as
predicted (Fig. 1). The reason for increased selectivity
over VEGFR-2 appears to be caused by a difference in
this lipophilic pocket between VEGFR-2 and TrkA. It
is not entirely clear, but the pocket in VEGFR-2 is com-
posed mainly of aliphatic residues (Val, Ile, and Leu). In
the TrkA pocket, at least three Phe residues exist in
place of these aliphatic VEGFR-2 residues. This is
undoubtedly causing a change in the size, shape, and
nature of this pocket between the two proteins, thus
potentially leading to the selectivity trends observed.
As previously mentioned, the C4 cyanide analog 6
showed essentially no TrkA potency, presumably due
to its lack of the key carboxamide hydrogen bond do-
nor–acceptor motif.
Notably, this docked structure suggested that there was
an un-utilized lipophilic pocket in the vicinity of the ben-
zylic carbon. Thus, analogs were designed to optimally
fill this space. In addition, the conformation of the urea
suggested that it could be replaced by other functional
groups, such as aminoheterocycles.
The initial synthesis of thiosubstituted isothiazole ureas
is shown in Scheme 1. Compound 2 is made through the
Based on the docked structure of 1, aminoheterocyclic
replacements for the urea were then targeted to further
elucidate the binding mode of inhibitors and to improve
ADME properties. Scheme 2 depicts a first generation
synthesis that allowed late-stage incorporation of di-
verse C5 aminoheterocycles, while conserving the most
potent substituents at the other isothiazole positions.
Compound 717 was first hydrolyzed to the methyl ester
and both sulfurs were oxidized to sulfones. It has been
shown previously that aromatic methylsulfones are
excellent leaving groups when participating in nucleo-
philic aromatic substitutions, and that the C5 sulfone
is more reactive than the C3 sulfone.17 Thus, displace-
ment of the C5 methylsulfone of 8 with ammonia in
Figure 2. Compound 1 docked into a homology model of TrkA.

3446
B. Lippa et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3444–3448
O
H2N
S 3
R2
H2N
HS
O
CN
MeO
H
CN
NH
H
MeO
H
H
N
S
N
OPh
R1
S
N
N
i,ii
OPh
v,vi
iii,iv
R3
N S 5
S
N
O
N S
O
O
2
3
4
5
Scheme 1. The synthesis of isothiazole analogs. Reagents: (i) PhOCONCS, 70%; (ii) I2, pyridine, 91%; (iii) Hg(OAc)2, H2S, TFA, 95%; (iv) H2SO4,
85%; (v) R1R2CHX, (i-Pr)2NEt; (vi) R3NH2.
Table 1. TrkA kinase and cell inhibiton, and VEGFR-2 kinase selectivity for isothiazoles 1, 5a–h, and 6
X
H
H
Cl
N
S
N
N S
O
R
Ra
VEGFR-2 kinase selectivityb
b
TrkA kinase inhibition IC50 (nM)
b
TrkA cell inhibition IC50 (nM)
Compound
X
1
CONH2
CONH2
CONH2
CONH2
CONH2
CONH2
CONH2
CONH2
CONH2
CN
H
7
3
300
96
4·
5a
5b
5c
5d
5e
5f
5g
5h
6
Me
Et
57·
111·
1300·
10·
188·
71·
NT
9
70
46
Etc
Etd
i-Pr
n-Pr
CF3
Ph
4
52
978
290
600
2870
NT
NT
16
35
110
2600
>5000
NT
NT
H
(NT, not tested).
a Compounds were tested as racemic mixtures, unless otherwise noted.
b Values are means of at least two experiments, assay error is <2·.
c R enantiomer.
d S enantiomer.
MeO
MeO2S
O
MeO
S
O
H2N
S
O
CN
H
Cl
Cl
vii, viii
i,ii
SMe
iii-vi
MeS
SO2Me
NH2
N
Het
N S
N S
N S
N S
Scheme 2. The synthesis of aminoheterocycle-substituted isothiazoles. Reagents: (i) H2SO4, MeOH, 93%; (ii) oxone, 94%; (iii) NH3, THF, 75%; (iv)
NaH, Boc2O, 82%; (v) p-ClBnSH, n-BuLi, 60%; (vi) TFA, CH2Cl2, 86%; (vii) HetArBr, Pd2(dba)3, rac-BINAP, Cs2CO3, Tol; (viii) Me3Al, NH4Cl,
Tol.
THF, Boc protection, and then displacement of the C3
sulfone with p-chlorobenzylthiol results in compound 9
in good yield after Boc deprotection. Notably, the
ammonia displacement must be done in THF; early at-
tempts with methanol resulted in bis addition of ammo-
nia to two isothiazoles, both at the C5 position.
Standard Pd-mediated aminations then are successful
in creating the amino-heterocycle bond,18 and the ana-
log is completed through a Weinreb amidation to give
10.
of the p-methoxybenzyl group (Scheme 3). Compound
11 was reacted with a heteroaryl isothiocyanate followed
by ring closure with iodine to yield compound 12. The
cyanide was then hydrolyzed with sulfuric acid, and
the crude product was subjected to relatively mild
deprotection conditions: 10% TFA in dichloromethane
in the presence of triethylsilane. This procedure cleanly
deprotected the C3 thiol in 73% yield (two steps).
Notably, it was found that the triethylsilane is
absolutely required to scavenge the benzyl cation as it
is formed and to prevent the reversibility of the
deprotection reaction. The thiol can then be capped with
a variety of electrophiles under standard conditions to
examine the effects of the C3 substituent.
While the original route to the aminoheterocycle-substi-
tuted isothiazoles allowed late-stage examination of the
aminoheterocycle portion of analogs, it did not allow
late-stage modulation of the C3 thio substituent. Thus,
a second synthetic method was created to allow facile
variation of the C3 thio substituent with a fixed amino-
heterocycle at C5. In designing such a scheme, there was
a desire to avoid the harsh C3 sulfur deprotection con-
ditions used previously in Scheme 1. Thus, a 2,4,6-tri-
methoxybenzyl-protecting group was utilized in place
Table 2 shows TrkA kinase and cell potency for a range
of C5 aminoheterocycles. Notably, a diverse set of
aminoheterocycles are shown to be adequate urea surro-
gates. Surprisingly, all aminoheterocycles revealed simi-
lar TrkA kinase and cell potency. Variation of the C3
sulfur substituent in this series showed that the replace-

B. Lippa et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3444–3448
3447
H2N
S 3
R2
O
OMe CN
MeO
OMe
S
H2N
HS
O
H
CN
MeO
H
H
S
N
i,ii
iii,iv
v
N
R1
N
Het
Het
N S 5
N S
Het
OMe NH
11
N S
13
OMe
12
14
Scheme 3. Alternate synthesis of aminoheterocycle-substituted isothiazoles. Reagents: (i) (Het)NCS, EtOAc, 60%; (ii) I2, pyridine, 81%; (iii) H2SO4;
(iv) Et3SiH, TFA, CH2Cl2, 73%—two steps; (v) R1R2CHX, (i-Pr)2NEt, DMF.
Table 2. TrkA kinase and cell inhibition for aminoheterocycle isothiazole analogs 10a–g and 14a–c
O
H2N
S
H
N
R2
R1
S
N
R1
R2
a
TrkA kinase inhibition IC50 (nM)
a
TrkA cell inhibition IC50 (nM)
Compound
10a
10b
10c
14a
14b
10d
14c
10e
p-Chlorobenzyl
p-Chlorobenzyl
p-Chlorobenzyl
CH(Me)(p-chlorobenzyl)
CH2(cyclohexyl)
p-Chlorobenzyl
o-Chlorobenzyl
p-Chlorobenzyl
H
>5000
33
19
NT
2-pyr
3-pyr
3-pyr
3-pyr
4-pyr
4-pyr
4-pyrimidine
756
581
NT
29
610
45
>1000
344
NT
28
47
NT
10f
p-Chlorobenzyl
179
84
>1000
NT
N
O
O
10g
p-Chlorobenzyl
N
H
(NT, not tested).
a Values are means of at least two experiments, assay error is <2·.
ment of the phenyl ring with a saturated cyclohexyl
group was not tolerated (14b), whereas an o-chloroben-
zyl group was well tolerated (14c).
Table 3. TrkA kinase and cell inhibiton for isothiazoles with sulfur
linked C3 bicycles 5i-o
H2N
S 3
O
H
H
N
N
While the aminoheterocyclic urea replacements did dem-
onstrate that the urea could be replaced with other func-
tionality, TrkA potency was not increased, and ADME
properties were not significantly improved. To further
enhance potency, attention turned back to the C3 substi-
tuent. Based on the knowledge that substitution of the
benzylic position was beneficial (Table 1), this structural
motif was retained, while cyclizing the benzylic substitu-
ent back onto the phenyl ring to create a bicyclic moiety,
thus further rigidifying the structure. Utilizing the syn-
thetic pathway of Scheme 1, several such analogs were
synthesized in the C5 urea series (Table 3). Compounds
5i and 5j largely retained potency when compared to the
acyclic ethyl derivative 5b. In these cases, the potency of
the methylated and unsubstituted urea were similar.
Enlarging the ring to a 6-membered size resulted in a
ꢀ10· decrease in potency (5k and 5l). However, further
ring enlargement to a 7-membered ring provided the
most potent TrkA inhibitors synthesized to date. An
approximately 10· improvement in potency from the
5-membered ring is realized and a <1 nM kinase potency
is achieved along with 17 nM cell potency (5m). Separa-
tion of these enantiomers again showed that the major-
ity of the potency resides in the R enantiomer (5n), and a
further increase in potency, now to 7 nM in the cell, is
X
N S 5
R
O
n
Compounda
X
n
R
TrkA kinase
inhibition
b
IC50 (nM)
TrkA cell
inhibition
b
IC50 (nM)
5i
Cl
Cl
H
H
H
H
H
1
1
2
2
3
3
3
H
10
8
110
118
5j
Me
H
5k
5l
105
111
<1
<1
91
1170
1020
17
Me
H
5m
5nc
5od
H
7
>1000
H
a Compounds were tested as racemic mixtures, unless otherwise noted.
b Values are means of at least two experiments, assay error is <2·.
c R enantiomer.
d S enantiomer.
achieved. It should also be noted that, although not test-
ed regularly, compounds in all series tended to inhibit
TrkB with similar potency. TrkC selectivity was not
determined.

3448
B. Lippa et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3444–3448
8. Marshall, J. L.; Kindler, H.; Deeken, J.; Bhargava, P.;
Vogelzang, N. J.; Rizvi, N.; Luhtala, T.; Boylan, S.;
Dordal, M.; Robertson, P.; Hawkins, M. J.; Ratain, M. J.
Invest. New Drugs 2005, 23, 31.
9. Gingrich, D. E.; Yang, S. X.; Gessner, G. W.; Angeles, T.
S.; Hudkins, R. L. J. Med. Chem. 2005, 48, 3776.
10. Wood, E. R.; Kuyper, L.; Petrov, K. G.; Hunter, R. N.,
III; Harris, P. A.; Lackey, K. Bioorg. Med. Chem. Lett.
2004, 14, 953.
11. Block, M. H.; Han, Y.; Josey, J. A.; Lee, J. W.; Scott, D.;
Wang, B.; Wang, H.; Wang, T.; Yu, D. W.O. Patent
2005049033.
12. Miranda, C.; Di Virgillio, M.; Selleri, S.; Zanotti, G.;
Pagliardini, S.; Pierotti, M. A.; Greco, A. J. Biol. Chem.
2002, 277, 6455.
In conclusion, a homology model was utilized to design
highly potent and selective inhibitors of the TrkA ki-
nase. These compounds were synthesized through three
novel synthetic routes, which allowed access to diverse
functionality. Aminoheterocycles were shown to be
good urea surrogates for inhibiting TrkA, whereas C3-
thiosubstituted bicycles were shown to be extremely po-
tent TrkA inhibitors, on par or better than all other
known TrkA inhibitors. These compounds should offer
the scientific community excellent tools for further
understanding the TrkA signaling cascade, and the
implications for inhibiting this pathway. Further in vivo
studies are needed to determine if these inhibitors will
have desirable anti-cancer therapeutic benefits.
13. Molecular Operating Environment, Chemical Comput-
ing Group, 1010 Sherbrooke Street West, Suite 910,
Montreal, Canada H3A 2R7; http://www.chemcomp.
com.
Acknowledgments
14. The kinase domain of the human TrkA receptor in
phosphorylation buffer with 0.5 lM ATP is incubated in
plates coated with PGT substrate. Compounds are incu-
bated for 30 min and the plates are washed. The amount
of phosphorylated PGT is quantitated by incubation with
HRP-conjugated PY-54 antibody, and developed with
ABTS substrate. Kwan, T. A.; LaGreca, S. D.; Lippa, B.
S.; Morris, J.; Wessel, M. D. W.O. Patent 2004011461.
15. PAE cells overexpressing TrkA were serum-deprived and
treated with compounds for 3 h followed by the addition
of 150 ng/ml NGF for a further 10 min to stimulate
phosphorylation. Cells were then lysed and lysates were
captured on Reacti-Bind goat anti-rabbit coated plates
that had been further coated with Santa Cruz anti-Trk (C-
14) sc-11. Phosphotyrosine levels were detected with the
anti-phospho-tyrosine antibody PY99. Kwan, T. A.;
LaGreca, S. D.; Lippa, B. S.; Morris, J.; Wessel, M. D.
W.O. Patent 2004011461.
16. VEGFR-2 activity was analyzed using a standard ELISA
in which purified VEGFR-2, in the presence of 10 lM
ATP, phosphorylated PGT bound to a 96-well plate in the
presence or absence of an inhibitor. A phosphorylated
substrate was detected using PY20 antibody with a TMB
colorimetric readout Beebe, J. S.; Jani, J. P.; Knauth, E.;
Goodwin, P.; Higdon, C.; Rossi, A. M.; Emerson, E.;
Finkelstein, M.; Floyd, E.; Harriman, S.; Atherton, J.;
Hillerman, S.; Soderstrom, C.; Kou, K.; Gant, T.; Noe,
M. C.; Foster, B.; Rastinejad, F.; Marx, M. A.; Schaeffer,
T.; Whalen, P. M.; Roberts, W. G. Cancer Res. 2003, 63,
7301.
Susan LaGreca, Nandini Patel, Goss S. Kauffman,
Richard Connell, Charles Boos, Michael Chen, William
Hungerford, Carl Thompson Jessica Frie, Michael
D’Occhio, Daniel Virtue, Sukhamaya Bain, and Jennifer
Achilles provided additional chemistry support. Addi-
tional biology support was received from Farzan
Rastinejad, Jitesh Jani, Larry Wagner, Kou Kou, Jean
Beebe, Kevin Coleman, and Andrew Garton. X-ray
crystallography to determine stereochemistry was
provided by Jon Bordner and Ivan Samardjiev.
References and notes
1. Ries, L. A. G.; Eisner, M. P.; Kosary, C. L.; Hankey, B. F.;
Miller, B. A.; Clegg, L.; Mariotto, A.; Feuer, E. J.; Edwards,
B. K. Eds., SEER Cancer Statistics Review, 1975–2002,
National Cancer Institute: Bethesda, MD. http://seer.can-
cer.gov/csr/1975_2002/, based on November 2004 SEER
data submission, posted to the SEER web site 2005.
2. Isaacs, W.; DeMarzo, A.; Nelson, W. G. Cancer Cell 2002,
2, 113.
3. Nakagawara, A. Cancer Lett. 2001, 169, 107.
4. Weeraratna, A. T.; Arnold, J. T.; George, D. J.; DeMarzo,
A.; Isaacs, J. T. The Prostate 2000, 45, 140.
5. Zhang, Y.; Dang, C.; Ma, Q.; Shimahara, Y. Oncol. Rep.
2005, 14, 161.
6. Ruggeri, B. A.; Miknyoczki, S. J.; Singh, J.; Hudkins, R.
L. Curr. Med. Chem. 1999, 6, 845.
7. George,D.J.;Dionne,C.A.;Jani,J.;Angeles,T.;Murakata,
C.; Lamb, J.; Isaacs, J. T. Cancer Res. 1999, 59, 2395.
17. Hatchard, W. R. J. Org. Chem. 1964, 29, 665.
18. Wolfe, J. P.; Wagaw, S.; Buchwald, S. L. J. Am. Chem.
Soc. 1996, 118, 7215.