Identification of a new chemical class of potent angiogenesis inhibitors based on conformational considerations and database searching

Article abstract of DOI:10.1016/S0960-894X(03)00626-7

The vascular endothelial growth factor (VEGF) tyrosine kinase receptors KDR and Flt-1 are targets of current interest in anticancer drug research. PTK787/ZK222584 is a potent inhibitor of these enzymes in clinical evaluation as an antiangiogenic agent. The development of a hypothesis concerning the bioactive conformation of this compound has led to the discovery of a new class of potent inhibitors of KDR and Flt-1, the anthranilamides. This could be achieved with a limited experimental effort, which only involved the testing of one archive compound and the synthesis and testing of one appropriate analogue.

Full text of DOI:10.1016/S0960-894X(03)00626-7

Bioorganic & Medicinal Chemistry Letters 13 (2003) 2967–2971
Identification of a New Chemical Class of Potent Angiogenesis
Inhibitors Based on Conformational Considerations and Database
Searching
Pascal Furet,a,* Guido Bold,a Francesco Hofmann,a Paul Manley,a
Thomas Meyera and Karl-Heinz Altmannb
aOncology Research, Novartis Pharma AG, CH-4002 Basel, Switzerland
bCorporate Research, Novartis Pharma AG, CH-4002 Basel, Switzerland
Received 22 January 2003; accepted 17 March 2003
Abstract—The vascular endothelial growth factor (VEGF) tyrosine kinase receptors KDR and Flt-1 are targets of current interest
in anticancer drug research. PTK787/ZK222584 is a potent inhibitor of these enzymes in clinical evaluation as an antiangiogenic
agent. The development of a hypothesis concerning the bioactive conformation of this compound has led to the discovery of a new
class of potent inhibitors of KDR and Flt-1, the anthranilamides. This could be achieved with a limited experimental effort, which
only involved the testing of one archive compound and the synthesis and testing of one appropriate analogue.
# 2003 Elsevier Ltd. All rights reserved.
Conformational Analysis of Compound 1
Introduction
Inhibition of tumor induced angiogenesis is a promising
strategy in anticancer drug research.1,2 Selective inhibi-
tion of the tyrosine kinase enzymatic activity of the two
vascular endothelial growth factor (VEGF) receptors
KDR and Flt-1 is an approach3ꢀ7 we are pursuing in
this area.
As 1 was identified by high-throughput screening and
not by structure-based design or pharmacophore mod-
eling, the structural and conformational determinants of
its kinase inhibitory activity were completely unknown
initially. As part of our efforts to elucidate these, we
undertook a conformational analysis of the inhibitor by
computational methods.11
In this respect, we have previously reported the dis-
covery of the anilinophthalazine compound PTK787/
ZK222584 (1), a potent and selective inhibitor of the
kinase activity of KDR and Flt-1, which is currently
undergoing clinical evaluation.8,9 More recently, we
have disclosed anthranilic acid amide derivatives, which
represent a novel chemical class of inhibitors of these
enzymes, with promising in vivo antitumor effects.10 In
this article, we describe how, following a line of reason-
ing based on a conformational analysis of 1, this new
class of angiogenesis inhibitors was discovered.
This analysis provided a set of 24 energy-minimized
conformations that could be classified in two subsets
corresponding, respectively, to an anti and a syn orien-
tation of the anilino N–H bond with respect to the C1–
N2 bond of the phthalazine nucleus as depicted in Fig-
ure 1. anti conformations were on average 3.0 kcal/mol
more stable than the syn conformations reflecting the
partial deconjugation of the aniline and phthalazine
moieties occurring in the latter because of steric hin-
drance between the chlorophenyl ring and the phthala-
zine benzenoid cycle. This was confirmed by ab initio
calculations on the smaller model compound lacking the
methylpyridyl part which indicated a destabilization of
the syn conformation by 3.1 kcal/mol compared to its
anti counterpart.
*Corresponding author. Tel.: +41-61-696-7990; fax: +41-61-696-
6246; e-mail: pascal.furet@pharma.novartis.com
0960-894X/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00626-7

2968
P. Furet et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2967–2971
Figure 3. Substructure query used in the database search.
Figure 1. Two possible orientations of the aniline moiety with respect
to the phthalazine nucleus in 1.
tein and desolvating hydrophobic groups for con-
formational adaptation.
Implication for the active conformation
This led us to assume that the KDR/Flt-1 inhibitory
conformation of 1 is of the anti type with the con-
sequence that the anilino NH group and the phthalazine
N2 atom do not correspond to the hydrogen bond
donor–acceptor system interacting with the hinge region
present in many kinase inhibitors. This hypothesis
prompted us to further investigate the role played by the
different parts of the inhibitor in its KDR/Flt-1 inhibi-
tory activity. We were particularly interested to investi-
gate the importance of the phthalazine nitrogen atoms.
Database searching. The numerous available X-ray
crystal structures of inhibitor–kinase complexes show
the conservation of a hydrogen bonding interaction
between the inhibitor and the backbone of the protein
in the so called ‘hinge’ region.12,13 The hinge region
corresponds to the amino acid stretch that connects the
N- and C-terminal domains of the kinase whose inter-
face forms the ATP (cofactor) binding site. Inhibitors
often achieve hydrogen bonding to the hinge region by
means of a bidentate donor–acceptor system present in
their structure.14ꢀ16 One can recognize such a hydrogen
bond donor–acceptor system in 1 (the anilino NH group
and the phthalazine N2 atom) when the inhibitor is
represented in a syn conformation (see Fig. 1). How-
ever, as shown by our conformational analysis, the syn
conformations are of high energy. 1 is thus unlikely to
adopt a syn conformation upon binding to an enzyme
pocket. A potent bioactive compound is not expected to
bind to a protein in a conformation that is significantly
less stable than its predominant conformation in solu-
tion.17 To be highly active, a compound must avoid
wasting a large part of the free energy stabilization
gained by forming favorable interactions with the pro-
To this end, we conducted a series of substructure sear-
ches in the Novartis collection of compounds with the
aim of retrieving analogues or mimetics of 1 lacking one
or two of the phthalazine nitrogen atoms for subsequent
KDR/Flt-1 testing. In one of these searches, we rea-
soned that an anthranilic acid amide moiety presenting
a strong intramolecular hydrogen bond between the
amine and keto functionalities18 could mimic the
phthtalazine ring of 1 by formation of a pseudo six-
membered ring. As illustrated in Figure 2 by van der
Waals surface/electrostatic potential representations,19
such a moiety can mimic the phthalazine ring in all
respects but the possibility of potential N3 hydrogen
bonding. Thus, with this mimic we could probe the
importance of the phthalazine N3 atom for KDR/Flt-1
inhibitory activity. The Novartis compound collection
was consequently searched using the substructure query
shown in Figure 3. The search returned a single com-
pound available for testing: 2. Interestingly, this com-
pound turned out to inhibit KDR and Flt-1 with IC50
values of 3.7 and 23 mM, respectively, as reported in
Table 1.20
Table 1. IC50 (mM) valuesa of compounds 1–3 in kinase assays
Kinase
1b
2
3
KDR
Flt-1
c-Kit
EGF-R
CDK-1
c-Met
IGF-1R
c-Src
0.037
0.077
0.73
3.7
23
0.020
0.18
0.24
7.3
—
—
—
—
—
—
—
>10
>10
>10
>10
>10
>10
>10
>10
>10
7.0
>10
Figure 2. Van der Waals surface/molecular electrostatic potential
representations of the phthalazine moiety of 1 (left) and its anthranilic
acid amide mimetic (right). The potential isovalue contour lines
appear in red for positive values and in blue for negative ones.
PKA
aThe data represent averages of at least three determinations.
bData from ref 9.

P. Furet et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2967–2971
2969
When tested, 3 showed a dramatic increase in KDR/Flt-
1 inhibitory activity relative to 2, resulting in a com-
pound with the same level of potency and selectivity22 as
1 (Table 1), thus validating the mimetism concept. With
3, we had an entry in a new class of potent inhibitors of
KDR and Flt-1. The high activity of this compound
also shed some light on the structural determinants of
the potency of 1. As discussed above, it meant that the
phthalazine N3 atom is probably not engaged in
hydrogen bonding with KDR or Flt-1 and therefore a
polar atom at this position might not be required for
activity. Moreover, it gave support to the hypothesis
that the active conformation of 1 is of the anti type. An
active syn conformation for 1 would imply an active
conformation of 3 in which the amide group adopts an
improbable high energy cis conformation. An overlay of
the calculated lowest energy conformational minima of
1 and 3 is shown in Figure 4.
Chemistry
Figure 4. Superposition of the calculated lowest-energy conformations
of 1 and 3.
The synthesis of compound 3 is summarized in Scheme
1.23 Anthranilic acid was converted to its BOC-pro-
tected derivative 4,24 which upon HBTU activation was
reacted with p-chloro aniline to provide anilide 5.25
Following removal of the BOC-protecting group with
4 N HCl/dioxane, reductive amination of the resulting
amine 626 with pyridine-4-carbaldehyde in the presence
of NaCNBH3 gave the desired compound 327 as white
crystals in 33% yield.
Conclusion
As the result of an investigation aimed at understanding
the conformational and structural basis of the KDR/
Flt-1 inhibitory activity of 1, we identified a new class of
inhibitors of these enzymes that possess potent anti-
angiogenic and antitumor properties.10 In the era of
high throughput approaches to drug discovery, where it
is believed that the screening or synthesis of very large
numbers of compounds are required, the work reported
here illustrates that large scale experimental effort is not
the only way to tackle the difficult problem of identify-
ing active compounds of pharmacological interest. With
a structure-based reasoning that led to the testing of one
archive compound and the synthesis of only one ana-
logue, we were able to discover a novel inhibitor che-
motype for important target enzymes in anticancer drug
research.
Scheme 1. (i) BOC2O, NEt3, DMF, rt, 16 h, 59%; (ii) 4-chloro aniline,
HBTU, NMM, DMF, rt, 16 h, 35%; (iii) 4 N HCl/dioxane/MeOH
7.5/1, rt, 3.5 h, 98%; (iv) pyridine-4-carbaldehyde, NaCNBH3,
MeOH/AcOH 100/1, rt, 2 h, 33%.
Synthesis of a potent mimetic. The finding that 2 pos-
sessed significant KDR and Flt-1 inhibitory activity
prompted us to synthesize 3, the exact anthranilic amide
mimetic of 1 in which the para-fluoro and ortho-fluoro
substituted phenyl rings of 2 are replaced by a 4-pyridyl
moiety and a para-chloro substituted phenyl, respec-
tively. In the series of 1, a para-chloro substituent on the
phenyl ring of the aniline moiety is clearly beneficial.21
References and Notes
1. Matter, A. Drug Discov. Today 2001, 6, 1005.
2. Kerbel, R. S. Carcinogenesis 2000, 21, 505.
3. Sun, L.; McMahon, G. Drug Discov. Today 2000, 5, 344.
4. Manley, P. W.; Furet, P. In Anticancer Agents: Frontiers in
Cancer Chemotherapy; Ojima, I., Vite, G., Altmann, K., Eds.;
ACS Symposium Series 796; American Chemical Society:
Washington, DC, 2001; p 282.
5. Veikkola, T.; Karkkainen, M.; Claesson-Welsh, L.; Alitalo,
K. Cancer Res. 2000, 60, 203.

2970
P. Furet et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2967–2971
6. Boyer, S. J. Curr. Top. Med. Chem. 2002, 2, 973.
7. Adams, J.; Huang, P.; Patrick, D. Curr. Opin. Chem. Biol.
2002, 6, 486.
8. Bold, G.; Altmann, K.-H.; Frei, J.; Lang, M.; Manley,
P. W.; Traxler, P.; Wietfeld, B.; Brueggen, J.; Buchdunger, E.;
Cozens, R.; Ferrari, S.; Furet, P.; Hofmann, F.; Martiny-
Baron, G.; Mestan, J.; Roesel, J.; Sills, M.; Stover, D.; Ace-
moglu, F.; Boss, E.; Emmenegger, R.; Laesser, L.; Masso, E.;
Roth, R.; Schlachter, C.; Vetterli, W.; Wyss, D.; Wood, J. M.
J. Med. Chem. 2000, 43, 2310.
9. Bold, G.; Frei, J.; Furet, P.; Manley, P. W.; Bruggen, J;
Cozens, R.; Ferrari, S.; Hofmann, F.; Martiny-Baron, G.;
Mestan, J.; Meyer, T.; Wood, J. M. Drugs Future 2002, 27, 43.
10. Manley, P. W.; Furet, P.; Bold, G.; Bruggen, J.; Mestan,
J.; Meyer, T.; Schnell, C. R.; Wood, J.; Haberey, M.; Huth,
A.; Kruger, M.; Menrad, A.; Ottow, E.; Seidelmann, D.; Sie-
meister, G.; Thierauch, K.-H. J. Med. Chem. 2002, 45, 5687.
11. The conformational analysis of 1 was performed in Mac-
romodel30 using the Monte-Carlo/energy minimization
method (automatic set up-standard protocol) and the
AMBER* force field in conjunction with the GB/SA water
solvation model. The ab initio calculations (restricted Har-
tree–Fock) were performed in Gaussian 92 using the 3-21G
basis set and full geometry optimization.
12. Toledo, L. M.; Lydon, N. B.; Elbaum, D. Curr. Med.
Chem. 1999, 6, 775.
13. Traxler, P.; Furet, P. Pharmacol. Ther. 1999, 82, 195.
14. Furet, P.; Meyer, T.; Strauss, A.; Raccuglia, S.; Rondeau,
J. M. Bioorg. Med. Chem. Lett. 2002, 12, 221.
15. Zhu, X.; Kim, J. L.; Newcomb, J. R.; Rose, P. E.; Stover,
D. R.; Toledo, L. M.; Zhao, H.; Morgenstern, K. A. Structure
1999, 7, 651.
16. Mohammadi, M.; Froum, S.; Hamby, J. M.; Schroeder,
M. C.; Panek, R. L.; Lu, G. H.; Eliseenkova, A. V.; Green, D.;
Schlessinger, J.; Hubbard, S. R. EMBO J. 1998, 17, 5896.
17. Bostrom, J.; Norrby, P.-O.; Liljefors, T. J. Comput.-Aided
Mol. Des. 1998, 12, 383.
18. A search of the Cambridge crystallographic database
using the anthralinic acid amide moiety shown in Figure 3 as
substructure query returned 14 X-ray crystal structures of
small organic molecules presenting this motif. In all of them,
an intramolecular hydrogen bond between the amine and keto
functionalities was present. Representative structures of this
type have the following entry codes in the database: YEG-
WUC, YEGWOW, TIDNAV and MAMBIL. Ab initio cal-
culations using the 3–21 G basis set indicate that the global
conformational minimum of the molecular fragment shown on
the right hand side of Figure 2 is the conformer presenting the
intramolecular hydrogen bond. The next conformational
minimum is less stable by around 5 kcal/mol.
19. In Figure 2, the molecular electrostatic potential displayed
as isovalue contour lines on the van der Waals surface31 was
calculated using the Coulomb formula with atomic point
charges. These were obtained in MOPAC (version 6.0)32 by
fitting the molecular electrostatic potential calculated from a
wavefunction in the MNDO approximation.
20. Enzyme Inhibition Assays. For the tyrosine kinases, inhi-
bition assays were performed as filter binding assays using
recombinant GST-fused kinase domains of the receptors
expressed in baculovirus and purified over glutathione
sepharose. [33P]-ATP was used as the phosphate donor and the
polyGluTyr (4:1) peptide was used as the acceptor (for details
see ref 8). The CDK1 and PKA assays are described in refs 28
and 29, respectively. Assays were performed under conditions
optimized for each kinase and with ATP concentrations simi-
lar to the Km of the respective enzyme towards ATP: 8.0 mM
(KDR, Flt-1), 1.0 mM (c-Kit, c-Met), 2.0 mM (EGF-R), 20 mM
(c-Src), 30 mM (IGF-1R), 7.5 mM (CDK1) and 50 mM (PKA).
21. A comparison of he KDR inhibitory activity of the first
two compounds reported in Table 3 of ref 8 shows the bene-
ficial effect of this chlorine atom.
22. 3 like 1 is a potent inhibitor for KDR, Flt-1 and kinases of
the PDGF-R family like c-Kit. It is inactive on kinases of the
AGC family (PKA), cyclin-dependent kinase family (CDK1)
and other tyrosine kinase families like ErbB (EGFR) and
c-Src.
23. The following abbreviations are used in the descriptions of
experimental chemistry procedures: AcOEt, ethyl acetate;
AcOH, acetic acid; BOC2O, di-tert-butyl-dicarbonate; DIPE,
di-iso-propylether; DMF, N,N-dimethylformamide; FC, flash
chromatography; HBTU, (O-benzotriazol-1-yl)-N,N,N0,N0-
tetramethyluronium-hexafluorophosphate; MeOH, methanol;
Mp, melting point; NMM, N-methyl morpholine; i-PrOH, iso-
propanol; rt, room temperature. All starting materials were
purchased from commercial suppliers and used without fur-
ther purification. Solvents were of reagent grade purity and
used as purchased without any additional distillation. FC was
performed on Merck silica gel 60 (0.043–0.060 mm). NMR
spectra were recorded on a 200-MHz Varian Gemini-200
instrument. Electrospray mass spectra were obtained with a
Fisons Instruments VG Platform II. Melting points were
recorded on a Buchi model 535 melting point apparatus and
are uncorrected. Elemental analyses were performed at Solvias
AG, Basel, Switzerland.
24. Synthesis of 4, 2-tert-Butoxycarbonylamino-benzoic acid.
To a solution of 40 g (0.292 mol) of anthranilic acid in 1 L of
DMF were added 86.6 g (0.397 mol) of BOC2O and the reac-
tion mixture was stirred at room temperature for 16 h. The
solvent was then evaporated, the residue re-dissolved in
CH2Cl2 and the solution extracted with 10% citric acid and
saline. The combined aqueous extracts were twice re-extracted
with CH2Cl2, the organic extracts combined, dried over
Na2SO4, and the solvent evaporated. The resulting yellow oil
was re-dissolved in CH2Cl2 and the product precipitated with
DIPE. Reprecipitation from CH2Cl2/DIPE gave 33.6
g
(48.5%) of the title compound as white crystals. From the
mother liquor two additional crops of 3.2 g (white crystals)
and 3.9 g (light yellow crystals), respectively, were 155–156 ꢁC.
C, H, N (237.26): C 60.71%, H 6.42%, N 5.88%, O 26.72%
1
(calcd C 60.75%, H 6.37%, N 5.90%, O 26.97%). H NMR
(200 MHz, CDCl3) d 10.0 (s, 1H); 8.45 (d, 1H); 8.11 (dd, 1H);
7.53 (dt, 1H); 7.04 (dt, 1H); 1.55 (s, 9H).
25. Synthesis of 5, [2-(4-chloro-phenylcarbamoyl)-phenyl]-
carbamic acid-tert-butyl ester. To a solution of 33.6 g of 4
(0.142 mol) and 36.2 g (0.283 mol) of p-chloro aniline in 660
mL of DMF were added 39.2 mL (0.355 mol) of NMM and
64.3 g (0.170 mol) of HBTU and the reaction mixture was
stirred at room temperature for 16 h. The solvent was then
evaporated, the residue re-dissolved in CH2Cl2 and the solu-
tion extracted with 10% citric acid, satd aq NaHCO3, and
saline. The aqueous extracts were re-extracted with CH2Cl2,
the organic extracts combined, dried over Na2SO4, and the
solvent evaporated. Treatment of the resulting orange-colored
oil with CH2Cl2 resulted in formation of an orange-colored
precipitate, which was discarded. Evaporation of the filtrate,
dissolution of the residue in CH2Cl2 and precipitation with
ꢁ
DIPE gave 16.9 g (35%) of 5 as white crystals. Mp ꢂ215 C
(dec.). C, H, N (337.81): C 62.23%, H 5.55%, N 8.13%, Cl
10.12%, O 14.23% (calcd C 62.34%, H 5.52%, N 8.08%, Cl,
1
10.22%, O 13.84%). H NMR (200 MHz, DMSO-d6) d 10.5
(s, 1H); 9.85 (s, 1H); 8.08 (d, 1H); 7.77 (dt, 1H); 7.75 (d, 2H);
7.52 (dt, 1H); 7.42 (d, 2H); 7.1 (t, 1H); 1.45 (s, 9H). EI–MS:
247 (100% [MꢀBOC]).
26. Synthesis of 6, 2-amino-N-(4-chloro-phenyl)-benzamide-
hydrochloride. To a suspension of 16.5 g (47.6 mmol) of 5 in
660 mL of MeOH were added 40 mL of 4 N HCl/dioxane (N2-
atmosphere). After 30 min a clear yellow solution had formed
and after 3.5 h the mixture was concentrated to ca. 50% of its

P. Furet et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2967–2971
2971
original volume, resulting in the formation of a white suspen-
sion. 500 mL of DIPE were then added and after 30 min the
precipitate was isolated by filtration (white crystals). This
material was reprecipitated from MeOH/DIPE to yield 13.2 g
solvent evaporated. The residue was purified by FC in
CH2Cl2/MeOH 30/1 and then recrystallized from hot i-PrOH
to give 0.200 g (33%) of 3 as white crystals. Mp 134–139 ꢁC.
C, H, N (337.81): C 67.1%, H 4.7%, N 12.1%, O 4.9% (calcd
C 67.56%, H 4.77%, N 12.44%, O 4.74%). 1H NMR
(200 MHz, DMSO-d6) d 8.5 (dd, 2H), 7.90 (t, 1H), 7.78 (dd,
2H), 7.70 (dd, 1H), 7.95–7.80 (2ꢃd, 2H+2H), 7.25 (t, 1H),
6.65 (t, 1H), 6.55 (d, 1H),4.50 (d, 2H).
28. Rialet, V.; Meijer, L. Anticancer Res. 1991, 11, 1581.
29. Meyer, T.; Regenass, U.; Fabbro, D.; Alteri, E.; Roesel, J.;
Muller, M.; Caravatti, G.; Matter, A. Int. J. Cancer 1989, 43, 851.
30. Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Lis-
kamp, R.; Lipton, M.; Caufield, C.; Chang, G.; Hendrickson,
T.; Still, W. C. J. Comput. Chem. 1990, 11, 440.
1
(97.9%) of 6 as white crystals. H NMR (200 MHz, DMSO-
d6) d 10.5 (s, 1H), 7.8 (m, 3H), 7.65 (s, very broad 3H), 7.45
(m, 3H), 7.10 (d, 1H), 7.08 (t, 1H). EI–MS: 247, 249 ([M+H]).
27. Synthesis of 3, N-(4-Chloro-phenyl)-2-[(pyridin-4-ylme-
thyl)-amino]-benzamide. To a solution of 0.500 g (1.77 mmol)
of 6 and 0.227 g (2.12 mmol) of pyridine-4-carbaldehyde in 8
mL of MeOH were added 80 mL of AcOH and the mixture
was stirred at room temperature for 2 h (N2-atmosphere).
After this time NaCNBH3 (0.355 g, 5.65 mmol) was added in
several portions and stirring continued for additional 2 h. The
solution was then cooled to 0 ꢁC and 100 mL of satd aq
NaHCO3 were added. The aqueous solution was extracted
with AcOEt (3ꢃ) and the combined organic extracts subse-
quently washed with water (2ꢃ), dried over Na2SO4 and the
31. (a) ‘In-house’ addition to Macromodel: Bohacek, R. S.;
Guida, W. C. J. Mol. Graph 1989, 7, 113. (b) Eyraud, V.;
Dietrich, A., Novartis Pharma Inc. Unpublished results.
32. Stewart, J. J. P. J. Comput.-Aided Mol. Des. 1990, 1, 1.