Novel C-5 aminomethyl pyrrolotriazine dual inhibitors of EGFR and HER2 protein tyrosine kinases

Article abstract of DOI:10.1016/j.bmcl.2007.02.050

Novel C-5 aminomethyl pyrrolotriazines were prepared and optimized for dual EGFR and HER2 protein tyrosine kinase inhibition. The homopiperazine, 1p, emerged as a key lead and it showed promising oral efficacy in EGFR and dual EGFR/HER2 driven human tumor xenograft models. It is hypothesized that the C-5 homopiperazine side chain binds in the ribose-phosphate portion of the ATP binding pocket.

Full text of DOI:10.1016/j.bmcl.2007.02.050

Bioorganic & Medicinal Chemistry Letters 17 (2007) 2828–2833
Novel C-5 aminomethyl pyrrolotriazine dual inhibitors of EGFR
and HER2 protein tyrosine kinases
Harold Mastalerz,a,* Ming Chang,b Ashvinikumar Gavai,b Walter Johnson,a
David Langley,a Francis Y. Lee,b Punit Marathe,b Arvind Mathur,b
Simone Oppenheimer,b James Tarrant,a John S. Tokarski,b Gregory D. Vite,b
Dolatrai M. Vyas,a Henry Wong,a Tai W. Wong,b Hongjian Zhangb and Guifen Zhanga
aDepartments of Oncology Chemistry, Discovery Biology, Metabolism and Pharmacokinetics, Computer Aided Drug Design,
and Chemical Synthesis, Bristol-Myers Squibb Pharmaceutical Research Institute, 5 Research Parkway,
Wallingford, CT 06492-1951, USA
bPO Box 4000, Princeton, NJ 08543-4000, USA
Received 30 January 2007; revised 16 February 2007; accepted 21 February 2007
Available online 27 February 2007
Abstract—Novel C-5 aminomethyl pyrrolotriazines were prepared and optimized for dual EGFR and HER2 protein tyrosine kinase
inhibition. The homopiperazine, 1p, emerged as a key lead and it showed promising oral efficacy in EGFR and dual EGFR/HER2
driven human tumor xenograft models. It is hypothesized that the C-5 homopiperazine side chain binds in the ribose–phosphate
portion of the ATP binding pocket.
ꢀ 2007 Elsevier Ltd. All rights reserved.
The epidermal growth factor receptor (EGFR, ErbB1 or
HER1) and the human epidermal growth factor recep-
tor 2 (HER2, ErbB2) are members of the ErbB family
of receptor tyrosine kinases that have been clinically
validated as targets for cancer therapy.1 Their frequent
co-expression in a variety of tumor types and their
capacity to form heterodimers with other members of
the ErbB family provide a strong rationale for simulta-
neously targeting both of these receptors.2 We have
reported pyrrolotriazine dual EGFR and HER2 kinase
inhibitors with solubilizing groups that are linked to
C-5 via a methylene ether, i.e. 1a (Fig. 1).3a Morpholine
analog 1b, which showed potent kinase inhibition and
good oral efficacy in EGFR and HER2 driven tumor
models, emerged as the lead compound from that study.
It was hypothesized that its C-5 solubilizing group ex-
tends into the ribose-phosphate portion of the ATP
binding pocket where it participates in multiple hydro-
gen bonding interactions. We examined other solubiliz-
ing groups and linkers at C-5 to further probe this
binding site and here report our initial results for ana-
logs with C-5 aminomethyl groups, that is, 1c.
To make C-5 aminomethyl analogs 1c, we examined the
reaction of 4-chloro-5-(bromomethyl)-pyrrolotriazine
23a with primary or secondary amines. This gave the
desired C-5 methylamines but together with significant
amounts of C-4 amine side products. The 5-methyl
N
F
N
HN
R
4
5
3
N
6
N
2
N
1
7
O
R''
R'O
N
R =
HN
O
R'''
1a
1b
1c
Keywords: Pyrrolotriazine; EGFR; HER2; Kinase inhibitor.
*
Figure 1. C-5 Substituted pyrrolotriazine dual EGFR and HER2
kinase inhibitors.
Corresponding author. Tel.: +1 203 677 7773; e-mail: harold.
mastalerz@bms.com
0960-894X/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.02.050

H. Mastalerz et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2828–2833
2829
sulfide, 3, was found to form selectively on reaction of 2
with thiophenol and so it was examined as an intermedi-
ate to make aminomethyl analogs 1c (Scheme 1).4 Reac-
tion of 3 with 1-(3-fluorobenzyl)-1H-indazol-5-amine
followed by oxidation gave the stable sulfoxide, 4. On
being heated neat with an excess of a primary or a sec-
ondary amine, it gave the C-5 aminomethyl product,
1c, presumably via cationic intermediate 5. If necessary,
this coupling was followed by protecting group removal
and the overall procedure was used to make most of the
analogs listed in Table 1. 5-Methyl-pyrrolotriazines with
acetate or chloride as leaving groups were also used to
prepare C-5 aminomethyl analogs (see Schemes 2 and 3).
For coupling reactions where the amine was difficult
to come by, smaller amounts of the amine were em-
ployed and dimethyl sulfoxide was used as the solvent.
The piperidine ester and acid, 1j and 1i, were obtained
by heating 4 with methyl piperidine-4-carboxylate. This
gave a mixture of the ester and the acid which were sep-
arated by preparative HPLC. Sulfonamide 1r was pre-
pared by reacting 1p with methansulfonyl chloride in
the presence of base. The 3-hydroxyl-homopiperazines,
1t and 1u, were made by heating 4 with excess 3-hydro-
xy-homopiperazine5 and separating the enantiomers by
preparative chiral HPLC. Ketone 1v was prepared by
perruthenate oxidation of a mixture of alcohols 1t and
1u in acetonitrile.6
N
F
N
O
S
Cl
N
R
HN
Ph
b, c
N
N
N
N
N
2 R = Br
3 R = SPh
4
a
N
F
d
N
1c
HN
N
N
N
5
Scheme 1. Reagents and conditions: (a) PhSH (1.0 equiv), (i-Pr)2EtN
(1.0 equiv), CH2Cl2, À20 ꢁC, 3 h; (b) 1-(3-fluorobenzyl)-1H-indazol-5-
amine (0.9 equiv), 1,2-dichloroethane/ n-BuOH (1:1), 85 ꢁC, 2.5 h; (c)
m-chloroperbenzoic acid (1 equiv), CHCl3, 0 ꢁC, 1 h , 55% overall; (d)
amine (about 30 equivalents), 135 ꢁC, sealed tube, overnight or amine
(about 5 equiv), dimethyl sulfoxide, 140 ꢁC, sealed tube, overnight.
Table 1. Structure–activity relationship for the C-5 aminomethyl analogs
N
F
N
HN
R
N
N
N
Metabolic stabilitye
0.14
4-h Plasma levelf (nM)
4100
a
Compound
R
H
IC50 (lM)
N87b
HER2
0.088
EGFR
0.081
HT29c or A2780d
>10c
1d
0.38
H
N
1e
1f
1g
1h
1i
0.047
0.11
0.36
0.063
0.11
>1
0.045
0.11
0.50
0.068
0.22
>1
1.95
1.2
nd
nd
nd
HO
HO
nd
nd
nd
N
nd
nd
nd
nd
N
0.14
>5
nd
0.29
0.23
nd
250
600
HO
N
>10c
>10c
HO2C
N
1j
>5
27
MeO2C
N
(continued on next page)

2830
H. Mastalerz et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2828–2833
Table 1 (continued)
a
Compound
R
IC50 (lM)
Metabolic stabilitye
4-h Plasma levelf (nM)
HER2
0.31
EGFR
0.14
N87b
HT29c or A2780d
nd
1k
1l
0.40
0.73
0.49
0.52
nd
nd
nd
H2NCO
HN
N
0.081
0.12
0.090
0.17
0.35
0.71
0.033
0.13
0.21
0.24
0.026
nd
0.00
nd
6300
480
nd
N
1m
1n
1o
1p
1q
1r
1s
1t
nd
N
N
O
0.20
>10c
nd
nd
HN
N
0.61
nd
nd
AcN
HN
N
0.027
0.12
0.11
0.62
0.33
0.20
0.16
3.6c
7.9c
>10c
>10c
>10c
0.05
0.21
nd
1300
93
N
N
N
Me
N
0.12
0
N
MeSO2
O
0.075
0.016
0.27
0.12
44
N
HN
HO
1506
N
N
HN
HO
1u
0.076
0.20
0.95
>10c
0.12
880
HN
O
1v
0.039
0.17
0.078
0.19
0.13
0.65
3.7
>10c
>5d
nd
nd
nd
nd
N
N
HN
HN
11g
14h
nd
0.074
0.68
>2.1d
0.00
N
HN
a IC50 values are reported as means of at least three determinations. Variability around the mean value was <15% for the enzymatic assays and <25%
for cell proliferation assays.
b Cell line N87 is a human gastric carcinoma that overexpresses both EGFR and HER2.
c,d
Cell lines HT29 (human colon carcinoma) and A2780 (human ovarian carcinoma) do not express EGFR or HER2.
e Rate of metabolism (mmol minÀ1 mg proteinÀ1) by mouse liver microsomes after a 10-min incubation at 3 lM in the presence of NADPH.
f Averaged 4 h plasma levels in three male Balb/C mice for compounds administered orally at 50 mpk in Tween 80/PEG400/water = 10:40:50.
g The 6-methoxy analog of 1p.
h The 6-methyl analog of 1p.
Enantiomer 1t was independently made from N-(2-ami-
noethyl)-2-nitrobenzene-sulfonamide 6 as outlined in
Scheme 2.4 Reaction of 6 with (S)-epichlorohydrin in
Deprotection with thiophenol9 gave material that was
identical with 1t.
7
the presence of MgSO4 gave chlorohydrin 7 which
The 6-methoxy analog of homopiperazine 1p, 11, was
prepared by heating acetate 103a with excess homopiper-
azine (Scheme 3). To make the 6-methyl analog, 14, es-
ter 1210 was first reduced to alcohol 13. This was treated
cyclized to homopiperazine 8 on heating with cesium
carbonate.8 The monoprotected homopiperazine was
linked to the pyrrolotriazine by heating it with acetate 9.

H. Mastalerz et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2828–2833
2831
Cl
Compound 1e showed a profile that was typical of sec-
ondary amine analogs: kinase inhibition that was com-
parable to that of the 5-methyl parent compound, 1d,
but weaker antiproliferative activity. Tertiary amine
analogs, for example 1f and 1g, were generally less po-
tent but the hydroxypiperidine, 1h, was remarkable in
that it showed good potency in both the kinase and cell
assays. The piperidine acid, ester, and amide analogs, 1i,
1j, and 1k, were less potent while the piperazine, 1l,
showed good kinase inhibition but not antiproliferative
activity. The latter did however demonstrate better met-
abolic stability in the presence of mouse liver micro-
somes and mouse oral exposure than the
hydroxypiperidine 1h. This was in line with previous
observations3a that pyrrolotriazines with basic primary
or secondary amine substituents showed better meta-
bolic stability. N-Methyl piperazine 1m and amide ana-
logs, 1n and 1o, were less potent, indicating that the
basic secondary amino group on the piperazine side
chain was required for potent kinase inhibition. Homo-
piperazine 1p showed both potent kinase inhibition and
antiproliferative activity but did not significantly inhibit
proliferation of the HT29 cell line. Its N-methyl deriva-
tive, 1q, sulfonamide, 1r, and amide analog, 1s, showed
reduced potency and oral exposure. The (R)-3-hydroxy-
homopiperazine, 1t, showed a profile similar to that of
1p, while its enantiomer, 1u, was less potent. Ketone
1v and the 6-methoxy and 6-methyl analogs, 11 and
14, were also less potent. The latter results indicate that
additional substitution at C-6 is less tolerated.
OH
a
H
N
H
N
S
NH2
S
N
H
O
O
O2N
O
O
O2N
6
7
OH
b
N
NH
S
O
O2N
O
N
F
N
8
HN
AcO
c
d
N
9
N
N
1t
Scheme 2. Reagents and conditions: (a) S-(+)-epichlorhydrin (0.5
equiv), MgSO4 (0.5 equiv), MeOH, rt, 8 h , 41%; (b) Cs2CO3 (3 equiv),
acetonitrile, 65 ꢁC, 6.5 h, 31%; (c) 8 (1 equiv), Et(i-Pr)2N (1.5 equiv),
CH3CN, pressure vessel, 102 ꢁC, 17 h, 85%; (d) PhSH (2 equiv),
K2CO3 (5 equiv), DMF, rt, 50 min, 92%.
with thionyl chloride to give a chloromethyl intermedi-
ate that was reacted directly with homopiperazine to
give 14.
Aminomethyl analogs that showed promising EGFR
and HER2 kinase inhibition were screened for their anti-
proliferative activity against the N87 cell line. This is a
human gastric carcinoma that overexpresses both
EGFR and HER2.11,12 Analogs were also screened in
HT29 (human colon carcinoma) or A2780 (human ovar-
ian carcinoma) cell proliferation assays. These are cell
lines that do not depend on EGFR or HER2 signaling
and provided a check for off-target antiproliferative ef-
fects. The SAR data are summarized in Table 1. All of
the analogs carried the C-4 m-fluorobenzylindazolylami-
no side chain since it was previously found to provide
optimal dual EGFR and HER2 kinase inhibition.3a,b
Figure 2 shows a potential binding mode for piperazine
1p in the ATP binding site that was modeled using the
published X-ray structure of the complex between lapat-
inib and EGFR kinase.13 The docked poses were energy-
minimized in Maestro14 using the OPLS-AA force
field15 and the GBSA continuum model16, an implicit
solvation model. The pyrrolotriazine core is oriented
N
F
N
HN
R
N
R'
N
N
a
11 R' = OMe
R =
10 R' = OMe
R = CH2OAc
N
NH
c
b
14 R' = Me
R =
12 R' = Me
R = CO2Me
13 R' = Me
R = CH2OH
N
NH
Figure 2. Predicted binding mode of compound 1p modeled in the
X-ray structure of the lapatinib/EGFR kinase complex, see Ref. 13.
The C-5 side chain extends into the ribose phosphate binding region
where the protonated azepine NH may form hydrogen bonds with
Asp831, Asn818 and/or Arg817. Image created with Pymol from
DeLano Scientific LLC, San Carlos, CA, USA. http://www.pymol.org
Scheme 3. Reagents and conditions: (a) homopiperazine (10 equiv),
CH3CN, microwave, 170 ꢁC, 30 min, 61%; (b) LiAlH4 (3 equiv), THF,
6 h, 46%; (c) SOCl2 (1.2 equiv), CH2Cl2, 1 h; then homopiperazine (2
equiv), N-methylmorpholine (2 equiv), CH2Cl2, rt, 20 h, 21%.

2832
H. Mastalerz et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2828–2833
Table 2. Pharmacokinetic parametersa for 1p in mice
Cmax (nM) po
Tmax (h) po
AUC
48913
(nM h) po
t1/2 (h) iv
MRT (h) iv
4.0
Cl (mL/min/kg) iv
39
Vss (L/kg) iv
Fpo (%)
77
0–8 h
7951
3
3.7
9.5
a pK parameters were obtained from composite plasma concentration–time profiles with an average of three male Balb/C mice at each time point.
Mice were dosed at 90 (po) and 10 (iv) mg kgÀ1 using Tween 80/PEG400/water (10:40:50, v/v) as the vehicle.
in the ATP binding site such that there is a hydrogen
bond between N-1 and the hinge region Met769 NH,
and the C-4 benzyl indazole group extends back into a
deep hydrophobic pocket formed partially by the
alpha-C-helix. The C-5 substituent extends out into the
ribose phosphate pocket where the protonated homo-
piperazine NH can hydrogen bond with the side chains
of Asp831, Asn818 and/or the backbone carbonyl oxy-
gen of Arg817. In this model, there is also an intramo-
lecular hydrogen bond between the C-4 aniline NH
and the hompiperazine tertiary nitrogen atom. This
binding model is similar to that previously suggested3a
for morpholine analog 1b and has been hypothesized17
for quinazoline EGFR and HER2 kinase inhibitors with
solubilizing side chains at C-5. It also offers an explana-
tion for why (R)-3-hydroxy-homopiperazine 1t is more
potent than its S-enantiomer, 1u. The hydroxy group
of the former homopiperazine is better situated to
hydrogen bond with Asp831 and/or the conserved
Lys745 that are normally involved in binding the phos-
phate group of ATP.
Compound 1p was evaluated in EGFR/HER2 driven
N87 human gastric and EGFR driven GEO human co-
lon carcinoma xenograft models in athymic mice. In
both studies it was administered orally, twice a day
(BID), for 14 consecutive days. It was robustly active
in the N87 model yielding tumor growth inhibition of
97 to 121% over the dose-range of 90–180 mg/kg (Table
3). A clear maximum-tolerated dose (MTD) was reached
at the 180 mg/kg dose which was associated with
approximately 12% drug related mortality. In the
GEO model, 1p was tested at 90 and 135 mg/kg and
was active at both dose levels. The 4-hydroxy-homopip-
erazine, 1t, was also evaluated in vivo but it did not
demonstrate an advantage over 1p.
In summary, pyrrolotriazines with diamino solubilizing
groups that are tethered to C-5 showed potent inhibition
of both EGFR and HER2 kinases. Modeling studies
suggested that the solubilizing group can extend into
the ribose–phosphate binding region of the ATP binding
pocket where it can participate in multiple hydrogen
bonding interactions. The homopiperazine analog, 1p,
emerged as a key lead and it exhibited potent kinase
inhibition, antiproliferative activity, and oral efficacy
in tumor xenograft models. This encouraged a broader
examination of other C-5 amino and diamino solubiliz-
ing groups and the results will be described in future
publications.
The PK parameters for 1p in mice are summarized in
Table 2. Absorption of the compound after oral admin-
istration was excellent (77% bioavailable) and this was
consistent with its high permeability in Caco-2 cells.
After intravenous administration, 1p showed a moderate
plasma clearance of 39 mg/min/kg, with a half-life (t1/2
)
and mean residency time (MRT) of 3.7 and 4 h, respec-
tively. A relatively high steady-state volume of distribu-
tion (Vss) was observed in mice. The compound was
98.4% bound to mouse serum proteins, yielding a free
fraction of 1.6% in that species. The aqueous solubility
of 1p was pH-dependent, ranging from 9.7 mg/mL with
a solution pH of 3.9 for the HCl salt to 0.012 mg/mL in
a pH 7.5 phosphate buffer. The relatively low aqueous
solubility at neutral pH is similar to that of other ATP
competitive EGFR kinase inhibitors.18
References and notes
1. (a) Baselga, J.; Arteaga, C. L. J. Clin. Oncol. 2005, 23,
2445; (b) Kamath, S.; Buolamwini, J. K. Med. Res. Rev.
2006, 26, 569, and references therein.
2. Baselga, J. Ann. Oncol. 2002, 13, 8.
3. (a) Mastalerz, H.; Chang, M.; Chen, P.; Dextraze, P.;
Fink, B. E.; Gavai, A.; Goyal, B.; Han, W.-C.; Johnson,
W.; Langley, D.; Lee, F. Y.; Oppenheimer, S.; Ruediger,
E.; Tarrant, J.; Tokarski, J. S.; Vite, G. D.; Vyas, D. M.;
Wong, T. W.; Zhang, H.; Zhang, G. Bioorg. Med. Chem.
Lett. in press; Quinazolines with a C-4 benzylindazolyl-
amino side chain similarly show potent dual EGFR and
HER2 kinase inhibition, see: (b) Brignola, P. S.; Lackey,
K.; Kadwell, S. H.; Hoffman, C.; Horne, E.; Carter, H. L.;
Stuart, J. D.; Blackburn, K.; Moyer, M. B.; Alligood, K.
J.; Knight, W. B.; Wood, E. R. J. Biol. Chem. 2002, 277,
1576, and references therein.
4. For experimental details, see: Mastalerz, H; Zhang, G.;
Tarrant, J.; Vite, G. D.; US 6,908,916.
5. Sarri, W. S.; Raab, A. W.; King, S. W. J. Org. Chem.
1971, 36, 1711.
6. Griffith, W. P.; Ley, S. V. Aldrichimica Acta 1990, 23, 13.
7. Bergeron, R. J.; Ludin, C.; Muller, R.; Smith, R. E.;
Phanstiel, O. J. Org. Chem. 1997, 62, 3285.
8. Kan, T.; Kobayashi, H.; Fukuyama Synlett 2002, 697, The
diazapine 8 most likely forms via the 7-endo ring closure
Table 3. In vivo antitumor activity of orally administered 1p against
established N87 and GEO xenografts implanted subcutaneously in
athymic micea
Tumor model Dosea, mg kgÀ1 Schedule % TGIb Pc
N87
90
135
180
BID · 14
BID · 14
97
111
121
0.0037
0.0001
0.0018
GEO
90
135
60
78
0.0001
0.0018
a Vehicle: Tween 80/PEG400/water, 10:40:50.
b Percent tumor growth inhibition during treatment. Activity is defined
as %TGI P50%.
c Probability for median tumor weight at the end of drug treatment,
compared with controls.

H. Mastalerz et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2828–2833
2833
of an epoxide intermediate. Tlc monitoring of the reaction
indicated that 7 first converted to a less polar product,
presumably the epoxide, that then slowly reacted to form
8. The piperazine that would arise from 6-exo ring closure
of the epoxide intermediate may have also been formed
but was not isolated.
Horne, E.; Lackey, K.; Alligood, K. J.; Rusnak, D. W.;
Gilmer, T. M.; Shewchuk, L. Cancer Res. 2004, 64, 6652.
14. Schrodinger, L. L. C. 2003. 32nd Floor, Tower 45, 120
West forty-fifth street, New York, 10036.
15. Kaminski, G. A.; Friesner, R. A.; Tirado-Rives, J.;
Jorgensen, W. L. J. Phys. Chem. B 2001, 105, 6474.
16. Yu, Z.; Jacobson, M. P.; Friesner, R. A. J. Comput. Chem.
2006, 27, 72.
9. Fukuyama, T.; Jow, C.-K.; Cheung, M. Tetrahedron Lett.
1995, 36, 6373.
10. Mastalerz, H.; Gavai, A. V.; Fink, B.; Struzynski, C.;
Tarrant, J.; Vite, G. D.; Wong, T. W.; Zhang, G.; Vyas, D.
M. Can. J. Chem 2006, 84, 528.
11. Wong, T. W.; Lee, F. Y.; Yu, C.; Lou, F. R.; Oppenhei-
mer, S.; Zhang, H.; Smykla, R. A.; Mastalerz, H.; Fink, B.
E.; Hunt, J. T.; Gavai, A. V.; Vite, G. D. Clin. Cancer Res.
2006, 12, 6186.
12. Additional cell proliferation data (not shown) using the
HER2 overexpressing BT-474 human breast tumor cell
line and a constitutively activated Sal2 murine salivary
gland tumor that expresses a CD8HER2 fusion protein
was found to track with that observed for the N87 cell line.
13. Wood, E. R.; Truesdale, A. T.; McDonald, O. B.; Yuan,
D.; Hassell, A.; Dickerson, S. H.; Ellis, B.; Pennisi, C.;
17. (a) Ballard, P.; Bradbury, R. H.; Harris, C. S.; Hennequin,
L. F. A.; Hickinson, M.; Johnson, P. D.; Kettle, J. G.;
Kinowska, T.; Leach, A. G.; Morgentin, R.; Pass, M.;
Ogilvie, D. J.; Olivier, A.; Warin, N.; Williams, E. J.
Bioorg. Med. Chem. Lett. 2006, 16, 1633; (b) Ballard, P.;
Bradbury, C. S.; Hennequin, L. F. A.; Hickinson, M.;
Johnson, P. D.; Kettle, J. G.; Kinowska, T.; Morgentin,
R.; Ogilvie, D. J.; Olivier, A. Bioorg. Med. Chem. Lett.
2005, 15, 4226.
18. For example, it is reported that the quinazoline EGFR
inhibitor, gefitinib, is sparingly soluble at pH 1 but
practically insoluble above pH 7 with solubility dropping
sharply between pH 4 and pH 6: Florence, K. Y.
Physicians Desk Reference; Thomson Healthcare, 2006.