Total synthesis and biological evaluation of the nakijiquinones

Article abstract of DOI:10.1021/ja011413i

The Her-2/Neu receptor tyrosine kinase is vastly overexpressed in about 30% of primary breast, ovary, and gastric carcinomas. The nakijiquinones are the only naturally occurring inhibitors of this important oncogene, and structural analogues of the nakijiquinones may display inhibitory properties toward other receptor tyrosine kinases involved in cell signaling and proliferation. Here, we describe the first enantioselective synthesis of the nakijiquinones. Key elements of the synthesis are (i) the reductive alkylation of a Wieland - Mieschertype enone with a tetramethoxyaryl bromide, (ii) the oxidative conversion of the aryl ring into a p-quinoid system, (iii) the regioselective saponification of one of the two vinylogous esters incorporated therein, and (iv) the selective introduction of different amino acids via nucleophilic conversion of the remaining vinylogous ester into the corresponding vinylogous amide. The correct stereochemistry and substitution patterns are completed by conversion of two keto groups into a methyl group and an endocyclic olefin via olefination/reduction and olefination/isomerization sequences, respectively. This synthesis route also gave access to analogues of nakijiquinone C with inverted configuration at C-2 or with an exocyclic instead of an endocyclic double bond. Investigation of the kinase-inhibiting properties of the synthesized derivatives revealed that the C-2 epimer 30 of nakijiquinone C is a potent and selective inhibitor of the KDR receptor, a receptor tyrosine kinase involved in tumor angiogenesis. Molecular modeling studies based on the crystal structure of KDR and a model of the ATP binding site built from a crystal structure of FGF-R revealed an insight into the structural basis for the difference in activity between the natural product nakijiquinone C and the C-2 epimer 30.

Full text of DOI:10.1021/ja011413i

11586
J. Am. Chem. Soc. 2001, 123, 11586-11593
Total Synthesis and Biological Evaluation of the Nakijiquinones
Petra Stahl,† Lars Kissau,† Ralph Mazitschek,‡ Axel Huwe,‡ Pascal Furet,§
Athanassios Giannis,*,‡ and Herbert Waldmann*,†
Contribution from the Department of Chemical Biology, Max-Planck-Institut fu¨r molekulare Physiologie,
Otto-Hahn-Strasse 11, 44227 Dortmund, Germany, Organische Chemie, UniVersita¨t Dortmund,
Fachbereich 3, 44221 Dortmund, Germany, Institut fu¨r Organische Chemie, UniVersita¨t Karlsruhe,
Richard-Willsta¨tter-Allee 2, 76128 Karlsruhe, Germany, and Oncology Research, NoVartis Pharma AG,
4002 Basel, Switzerland
ReceiVed June 8, 2001
Abstract: The Her-2/Neu receptor tyrosine kinase is vastly overexpressed in about 30% of primary breast,
ovary, and gastric carcinomas. The nakijiquinones are the only naturally occurring inhibitors of this important
oncogene, and structural analogues of the nakijiquinones may display inhibitory properties toward other receptor
tyrosine kinases involved in cell signaling and proliferation. Here, we describe the first enantioselective synthesis
of the nakijiquinones. Key elements of the synthesis are (i) the reductive alkylation of a Wieland-Miescher-
type enone with a tetramethoxyaryl bromide, (ii) the oxidative conversion of the aryl ring into a p-quinoid
system, (iii) the regioselective saponification of one of the two vinylogous esters incorporated therein, and
(iv) the selective introduction of different amino acids via nucleophilic conversion of the remaining vinylogous
ester into the corresponding vinylogous amide. The correct stereochemistry and substitution patterns are
completed by conversion of two keto groups into a methyl group and an endocyclic olefin via olefination/
reduction and olefination/isomerization sequences, respectively. This synthesis route also gave access to
analogues of nakijiquinone C with inverted configuration at C-2 or with an exocyclic instead of an endocyclic
double bond. Investigation of the kinase-inhibiting properties of the synthesized derivatives revealed that the
C-2 epimer 30 of nakijiquinone C is a potent and selective inhibitor of the KDR receptor, a receptor tyrosine
kinase involved in tumor angiogenesis. Molecular modeling studies based on the crystal structure of KDR and
a model of the ATP binding site built from a crystal structure of FGF-R revealed an insight into the structural
basis for the difference in activity between the natural product nakijiquinone C and the C-2 epimer 30.
Introduction
Among this class of natural products, the nakijiquinones (1)
are of particular relevance. These marine sesquiterpene quinones
display pronounced cytotoxicity against L 1210 murine leukemia
cells and KB human epidermoid carcinoma cells and, most
notably, were identified as the first naturally occurring inhibitors
of the Her-2/Neu (also called erbB-2) receptor tyrosine kinase
(RTK).1 Her-2/Neu is vastly overexpressed in about 30% of
primary breast, ovary, and gastric carcinomas.7,8 Amplification
is closely correlated with the clinical behavior of these
neoplasms, such that tumors with Her-2 amplification are more
aggressive and are associated with reduced patient survival.
Natural products embodying a Decalin-type core structure
and a quinoid or related aromatic side chain often are character-
ized by pronounced and manifold biological properties. For
instance, the marine sesquiterpene quinones nakijiquinone A-D
(1a-d),1 ilimaquinone (2),2 avarol (3),3 smenospongine (4),
smenospongidine (5), smenospongiarine (6),4 and mamanutha-
quinone (7)5 (Figure 1) display antimicrobial, antiviral, and
cytotoxic activities. Therefore, they offer promising opportunities
for the development of new compounds which may be employed
efficiently for the elucidation of intracellular events and/or the
development of new drugs. A prime example which clearly
shows the possible impact of this natural product-based approach
to study biological phenomena is the synthesis of ilimaquinone
and its use for the study of intracellular vesicular trafficking by
Snapper et al.6
Receptor tyrosine kinases are crucial for proliferation,
survival, and differentiation. They have been implicated in
several pathologies such as tumor growth and tumor angiogen-
esis. Compounds that can selectively block RTK activity are
of paramount importance for the development of new anticancer
drugs.8
* To whom correspondence should be addressed.
† Max-Planck-Institute and University of Dortmund.
‡ University of Karlsruhe.
(6) (a) Radeke, H. S.; Digits, C. A.; Brumo, S. D.; Snapper, M. L. J.
Org. Chem. 1997, 62, 2823. (b) Radeke, H. S.; Digits, C. A.; Casaubon, R.
L.; Snapper, M. L. Chem. Biol. 1999, 6, 639.
§ Novartis Pharma AG.
(1) (a) Kobayashi, J.; Madono, T.; Shigemori, H. Tetrahedron 1995, 51,
10867. (b) Shigemori, H.; Madono, T.; Sasaki, T.; Mikami, Y.; Kobayashi,
J. Tetrahedron 1994, 50, 8347.
(2) Luibrandt, R. T.; Erdman, T. R.; Vollmer, J. J.; Scheuer, P. J.; Finer,
J.; Clardy, J. C. Tetrahedron 1979, 35, 609.
(3) Minale, L.; Riccio, R.; Sadano, G. Tetrahedron Lett. 1974, 3401.
(4) Kondracki, M. L.; Guyot, M. Tetrahedron 1989, 45, 1995.
(5) Swersey, J. C.; Barrows, L. R.; Ireland, C. M. Tetrahedron Lett. 1991,
32, 6687.
(7) (a) Protein Phosphorylation; Marks, F., Ed.; VCH: Weinheim, 1996.
(b) Protein Kinases; Woodgett, J. R., Ed.; Oxford University Press: Oxford,
1994. (c) Cooper, G. F. Oncogenes, 2nd ed.; Jones and Bartlett: Boston,
1995. (d) Wagener, C. Molekulare Onkologie; Thieme: Stuttgart, 1999.
(8) (a) Slamon, D. J.; Clark, G. M.; Wong, S. G.; Levin, W. J.; Ullrich,
A.; McGuire, W. L. Science 1987, 235, 177. (b) Hudziak, R. M.; Ullrich,
A. J. Biol. Chem. 1991, 266, 24109. (c) Dobrusin, E. M.; Fry, D. M. Annu.
Rep. Med. Chem. 1992, 27, 169.
10.1021/ja011413i CCC: $20.00 © 2001 American Chemical Society
Published on Web 11/01/2001

Synthesis and EValuation of the Nakijiquinones
J. Am. Chem. Soc., Vol. 123, No. 47, 2001 11587
Scheme 1. Retrosynthetic Analysis of the Nakijiquinones
planned to generate the selectively functionalized quinoid system
by oxidation of tetramethoxy-substituted aromatic precursor 9
to a 1,4-dicarbonyl compound and subsequent selective saponi-
fication of one of the two vinylogous methyl esters generated
thereby. Tracing the quinoid system back to a stable aromatic
compound was necessary, because we resorted to the reductive
alkylation of R,â-unsaturated Wieland-Miescher-type enone 11
with a benzyl halide 12 for the synthesis of the diterpene
structure and its coupling with the alkoxy-substituted aromatic
compound. This strategy had already proven to be useful in the
construction of related natural products.13 This synthesis plan
called for an intermediary keto group at C-2 which later on
would be converted to the required methyl group by an
olefination and subsequent isomerization. Finally, we envisaged
generating the endocyclic 5,6-alkene by olefination of a ketone
at C-5 and subsequent isomerization of the generated exocyclic
double bond. Thus, compound 10 was aimed at as an intermedi-
ate.
Figure 1. Structures of the nakijiquinones and related natural products.
During the last years, many small molecule inhibitors of
RTKs have been discovered, and some of them have already
entered clinical trails (see in a following section).
In addition to the opportunities for the treatment of cancer
and other diseases opened up by RTK inhibitors, such drugs
are also likely to be useful in dissecting signaling pathways.9-11
In this paper, we describe the first enantioselective total
synthesis of the nakijiquinones and closely related analogues
of these natural products and a first biological evaluation aimed
at the determination of their potency and their selectivity against
different tyrosine kinases.12
Synthesis of the Nakijiquinones. Benzyl halides (12) were
synthesized from catechol (13) as shown in Scheme 2; attempts
to synthesize these compounds from p-benzoquinone failed.
First, aromatic diol 13 was oxidized to the corresponding ortho-
quinone, which readily added sodium methoxide. Subsequent
reoxidation led to ortho-quinone (14), which upon treatment
with sulfuric acid in methanol rearranged to the corresponding
para-quinoid compound (15).14 Reduction of the carbonyl
groups proved unexpectedly difficult; that is, the use of CrCl2,
NaHSO3, B2H6, ascorbic acid, and TiCl3 failed. However, upon
treatment with NaBH4 in ethanol, the carbonyl groups were
cleanly reduced. The resulting diol (16) was unstable and,
therefore, directly converted into tetramethoxybenzene (17).15
To vary size and quality of the leaving group in the planned
reductive alkylation, benzyl halides 12a-c were synthesized.
Chloromethylation was achieved in 71% by treatment with
Results and Discussion
Retrosynthetic Considerations. The nakijiquinones embody
three basic structural elements, an amino acid, a central
p-quinoid unit, and a diterpenoid system. To reach a high degree
of convergency, the nakijiquinones were first dissected in a
retrosynthetic sense into isospongiaquinone (8), an interesting
natural product in its own right (vide infra), and an amino acid
that can be introduced in the last step by conjugate addition to
the vinylogous methyl ester present1 (Scheme 1). It was further
(9) (a) Daub, H.; Weiss, F. U.; Wallasch, A.; Ullrich, A. Nature 1996,
379, 557. (b) Novogrodsky, A.; Vanichin, A.; Patya, M.; Gazit, A.; Osherov,
N.; Levitzki, A. Science 1994, 264, 1319. (c) Meydan, N.; Gru¨nberger, T.;
Dadi, H.; Shahar, M.; Arpaia, E.; Lapidot, Z.; Leeder, J. S.; Freedman, M.;
Cohen, A.; Gazit, A.; Levitzki, A.; Raiffman, C. M. Nature 1996, 379,
645. (d) Dudley, D. T.; Pong, L.; Decker, S. J.; Bridges, A. J.; Saltiel, A.
R. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 7686. (e) Lepley, R. A.;
Kirkpatrick, F. A. Arch. Biochem. Biophys. 996, 331, 141. (f) Buchdunger,
E.; Zimmermann, J.; Mett, H.; Meyer, T.; Myller, M.; Regenass, U.; Lydon,
N. B. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 2558.
(10) Hinterding, K.; Alonzo-D´ıas, D.; Waldmann, H. Angew. Chem. 1998,
110, 717. Angew. Chem., Int. Ed. 1998, 37, 2558.
(11) Levitzki, A.; Gazit, A. Science 1995, 267, 1782.
(12) Part of this work was published in preliminary form: Stahl, P.;
Waldmann, H. Angew. Chem. 1999, 111, 3935. Angew. Chem., Int. Ed.
1999, 38, 3710.
16
1-chloro-4-(chloromethoxy)butane in the presence of SnCl4
in THF/cyclohexane; in neat cyclohexane, the reaction did not
(13) (a) Sarma, A. S.; Chattopadhyay, P. J. Org. Chem. 1982, 47, 1727.
(b) An, J.; Wiemer, D. F. J. Org. Chem. 1996, 61, 8775. (c) Brunner, C.;
Radeke, H. S.; Tallarico, J. A.; Snapper, M. L. J. Org. Chem. 1995, 60,
1114. (d) Locke, E. P.; Hecht, S. M. Chem. Commun. 1996, 2717.
(14) Wanzlick, H.-W.; Jahnke, U. Chem. Ber. 1968, 101, 3744.
(15) Benington, F.; Morin, R. D.; Clark, L. C. J. Org. Chem. 1955, 20,
102.

11588 J. Am. Chem. Soc., Vol. 123, No. 47, 2001
Scheme 2. Synthesis of Benzyl Halides 12a-c
Stahl et al.
Scheme 3. Synthesis of Advanced Intermediate 22a
a PbO2, NaOMe, MeOH, 20-22 °C, 65%. b H2SO4, MeOH, 96%.
c NaBH4, EtOH, 0 °C. d (CH3O)2SO2, NaOH, NaHSO3, EtOH, reflux,
75% (2 steps). e Cl(CH2)4OCH2Cl, SnCl4, THF, cyclohexene, 71% 12a.
f (HCHO)n, HBr/CH3COOH, 74% 12b. g NaI, acetone, 94%.
occur. Bromomethyl derivative 12b was obtained by treatment
with 1.1 equiv HBr in acetic acid in the presence of 2 equiv of
paraformaldehyde.17 Benzyl iodide 12c was accessible from
either chloride 12a or bromide 12b via Finkelstein reaction. R,â-
Unsaturated ketone 18 was synthesized as described via
Robinson annelation employing L-phenylalanine as source of
chirality.18 After recrystallization from n-hexane/ethyl acetate,
the diketone was obtained in 98% ee (the ee value was
determined by means of NMR spectroscopy after reduction of
the keto group at C-5 with NaBH4 and esterification of the
resulting alcohol with (R)-Mosher acid).
Selective monoprotection of the C-5 keto group was achieved
in high yield by transacetalization with ethyl-2-methyl-1,3-
dioxolane.19 Under these conditions, only the more electrophilic
keto group was attacked, whereas direct ketalization with
ethylene glycol led to protection of both carbonyl groups.
The decisive reductive coupling between benzyl halides 12
and enone 11 turned out to be particularly challenging. Such
couplings employing less-substituted benzylic halides have been
described in the literature13a,b,d and proved to be reproducible
in our hands. However, the reported reaction conditions could
not be employed successfully to the synthesis of compound 10
(Scheme 3). In initial experiments, none of the benzyl halides
12 reacted with the radical anion generated from enone 11 by
treatment with lithium in ammonia at -78 °C. Alternatively,
we tried to effect the coupling by synthesis of the silylenol ether
via treatment with methyllithium20 or via direct generation of
the enolate by treatment of the R,â-unsaturated ketone with
L-selectride.20 However, only addition of MeLi to the carbonyl
group or reduction of the ketone was observed. Because of the
failure of these attempts, we carefully reexamined the conditions
for the originally considered reductive alkylation procedure, and
finally, conditions were found under which the desired coupling
product was obtained in 76% yield. The reaction is run best in
a MED, ethylene glycol, p-TsOH, CH2Cl2, 84%. b Li, NH3, THF,
H2O -78 °C; -33 °C, 45 min; -78 °C 12b, 2h then reflux, 76%.
c KOtBu, CH3PPH3Br, toluene, reflux, 98% 19. d 5% HCl, THF, rt,
quant, then 10%Pd/C, H2, Et3N, MeOH, 12h, rt, 92% (97:3 (20a:20b)).
e KOtBu, CH3PPH3Br, toluene, reflux, 98% 21. f RhCl3, CHCl3/EtOH,
2 d, reflux, 95% 9. g AgO, HNO3, dioxane, rt, 22% 22a, 60% 22b,
9.5% 8. h H2SO4, MeOH, 96%.
such a way that 30 equiv of lithium dust are first added to liquid
ammonia at -78 °C. Then, a solution of enone 11 in THF
containing traces of water (1:0.018) is added slowly over 30
min. The resulting solution is refluxed at -33 °C for 45 min.
It is important that the solution remains blue during this time,
indicating the formation and existence of the enolate. The
solution is then cooled to -78 °C, and THF is added to
guarantee that the enolate remains in solution (final ratio NH3/
THF ) 1.0:0.7). Then, immediately, a solution of 12 equiv of
benzyl bromide 12b in THF is added, and the solution is
refluxed for 2 h (final ratio THF/NH3/H2O ) 1:1.5:0.005). To
achieve a high yield, it is important to ensure that the solution
does not contain excess solvated electrons after addition of the
benzyl halide is complete, because they would induce dimer-
ization of the benzyl bromide. If required, surplus electrons can
be trapped by addition of isoprene. If the benzyl halide is added
to the reaction mixture at -33 °C, the amount of byproducts
formed increases.
(16) (a) Olah, G. A.; Beal, D. A.; Olah, J. A. J. Org. Chem. 1976, 41,
1627. (b) Olah, G. A.; Beal, D. A.; Yu, S. H.; Olah, J. A. Synthesis 1974,
560.
(17) Van der Made, A. W.; Van der Made, R. H. J. Org. Chem. 1993,
58, 1262. For a further preparation of benzyl bromide 12b, see ref 24.
(18) Hagiwara, H.; Uda, H. J. Org. Chem. 1988, 53, 2308.
(19) Wijnberg, J. B. P.; Jenniskens, L. H. D.; Brunekreef, G. A.; de Groot,
A. J. Org. Chem. 1990, 55, 941.
(20) (a) Smitth, A. B., III; Mewshaw, R. J. Org. Chem. 1984, 49, 3685.
(b) Knight, S. D.; Overman, L. E.; Pairaudeau, G. J. J. Am. Chem. Soc.
1993, 115, 9293-9294.
With an efficient process for the synthesis of R,R-dialkylated
ketone 10 developed, the generation of the stereocenter at C-2

Synthesis and EValuation of the Nakijiquinones
J. Am. Chem. Soc., Vol. 123, No. 47, 2001 11589
Scheme 4. Synthesis of Nakijiquinone 1c
was approached. The conversion of the keto group present in
10 into the corresponding exo-methylene compound 19 was
unexpectedly difficult. Several established methylenation re-
22
agents such as the Tebbe reagent21 and Cp2TiMe2 failed
completely, and under various conditions, for example, employ-
ing BuLi in dioxane or DME or NaH in DMSO, the Wittig
reaction either yielded no product at all or only an unsatisfactory
low amount of the desired olefin. Also, attempts to add MeLi
to the keto group followed by deoxygenation of the tertiary
alcohol in a Barton protocol failed. Finally, the use of 9 equiv
of H3CPPh3Br together with 9 equiv of KOtBu in refluxing
toluene23 turned out to be the method of choice, yielding
exocyclic olefin 19 in nearly quantitative yield. Disappointingly,
reduction of olefin 19 with PtO2 proceeded without recordable
stereoselectivity. We felt that this might be due to an unfavorable
conformation of the bicyclic system and envisaged that removal
of the acetal protecting group might improve the situation. This
turned out to be the case. After cleavage of the acetal, the
stereoselectivity of the reduction with PtO2 in CH2Cl2 was raised
to 80:20, and use of Pd/C as catalyst in NEt3 as solvent yielded
ketone 3 in 92% as a 97:3 mixture of epimers at C-2 (Scheme
3). Diastereomers 20a and 20b are readily separated by
chromatography. As planned, the ketone at C-5 was converted
in very high overall yield to endocyclic alkene 9 by olefination
(for which the conditions described previously once more proved
to be very efficient) and subsequent rhodium-catalyzed isomer-
ization of the exocyclic to the endocyclic double bond.
The next two steps in the synthetic sequence, that is, the
oxidation of the tetramethoxyphenyl ring to the para-quinoid
system and the subsequent selective removal of the correct
O-methyl group, posed major problems. Oxidation of the
aromatic system was initially attempted by applying CrO3 or
cerium ammonium nitrate which had proven to be useful
reagents in related studies.6,13c,24 However, oxidized intermediate
22a was formed only in low yield or not at all. After substantial
variation of the reaction conditions, the use of AgO and catalytic
amounts of HNO3 at room temperature in dioxane as solvent
finally evolved as the method of choice.25 Under these condi-
tions, p-diketone 22a and regioisomeric o-diketone 22b were
obtained in a total yield of 82% as an isomeric mixture. The
undesired o-diketone 22b was readily rearranged to 22a by
treatment with acid.
Initial attempts directed toward the regioselective removal
of the correct O-methyl group from 22a concentrated on the
use of different Lewis acids. From the synthesis of maesanin
and related compounds, it is known that such quinoid systems
can be regioselectively deprotected by means of HClO4 or
BCl3.26 However, application of this method to 22a resulted in
decomposition of the starting material. If milder methods such
as the use of TMSI or AlCl3 were used, the starting material
was reisolated. Also, application of alternative techniques for
the cleavage of methyl ethers (e.g., LiI in pyridine, or BF3‚
Et2O together with EtSH or NaSEt) only resulted in reisolation
or decomposition of the starting material. Furthermore, attempts
to cleave a methyl ether from tetramethoxyaryl intermediate 20a,
that is, prior to oxidation to the quinoid system, were frustrated
b
a KOH, MeOH, H2O, rt, 71% 8, 11% 8a. D- or L-Amino acid,
NaHCO3, EtOH, 30 °C or 40 °C, 24 h.
by Lewis acid mediated rearrangement of the terpene framework
and led to the formation of undesired compounds 23a and 23b
(Scheme 4).
Finally, the problem could be solved by following a different
train of thought. Compound 22a can not only be regarded as a
double methyl ether but also as a double vinylogous methyl
ester which should be cleavable under basic conditions. Gratify-
ingly, this reasoning proved to be correct, and treatment of 22a
with 1 N KOH resulted in the formation of selectively
deprotected vinylogous acid 8 in 71% yield. In addition,
regioisomeric saponification product 8a was formed in 11%
yield. Presumably, the reaction proceeds via conjugate addition
of hydroxide to one of the vinylogous esters and elimination of
methanol. The preferred formation of the correct regioisomer
under these conditions is not readily explained, in particular
because variations of the reaction conditions, that is, use of pure
methanol instead of H2O/methanol or use of LiOH instead of
KOH, lead to the preferred formation of undesired isomer 8a.
We assume that attack of hydroxide on both possible positions
is reversible and that the two enolate intermediates are in
equilibrium with each other. Obviously, the enolate leading to
the desired product is the thermodynamically more stable isomer
and is formed preferably under the particular reaction conditions
chosen.
(21) (a) Tebbe, F. N.; Parshall, G. W.; Reddy, G. S. J. Am. Chem. Soc.
1978, 100, 3611. (b) Pine, S. H.; Kim, G.; Lee, V. Org. Synth. 1990, 65,
72. (c) Pine, S. H.; Shen, G. S.; Hoang, H. Synthesis 1991, 165. (d) Cannizzo,
L. F.; Grubbs, R. H. J. Org. Chem. 1985, 50, 2386.
(22) Petasis, N. A.; Bzowej, E. I. J. Am. Chem. Soc. 1990, 112, 6392.
(23) Fitzer, L.; Quadbeck, U. Synth. Commun. 1985, 15, 855.
(24) Poigny, S.; Guyot, M.; Samadi, M. J. Org. Chem. 1998, 63, 5890.
(25) Snyder, C. S.; Rapoport, H. J. Am. Chem. Soc. 1972, 94, 227.
(26) (a) Kubo, I.; Kim, M.; Ganjian, I. Tetrahedron 1987, 43, 2653. (b)
Kubo, I.; Kamikawa, T.; Miura, I. Tetrahedron Lett. 1983, 24, 3825.
Compound 8 is identical to isospongiaquinone, a natural
product isolated from Stelospongia conulata.27 Specific rotation,
(27) Kazlauskas, R.; Murphy, P. T.; Warren, R. G.; Wells, R. J.; Blount,
J. F. Aust. J. Chem. 1978, 31, 2685.

11590 J. Am. Chem. Soc., Vol. 123, No. 47, 2001
Stahl et al.
Scheme 5. Synthesis of C2 Epimeric Nakijiquinone C
IR, and NMR spectroscopic data of synthetic 8 were in full
accord with the data reported for the natural product, thereby
proving the absolute configuration of the synthesized compound.
Isospongiaquinone 8 may be employed as a central interme-
diate for the synthesis of all nakijiquinones. Treatment of this
compound with amino acids in ethanol at 30 °C or 40 °C (see
the Supporting Information) in the presence of NaHCO3 results
in the conversion of the remaining vinylogous ester in 8 into
the corresponding vinylogous amide.1 We have verified this
finding and converted isospongiaquinone (8) into nakijiquinones
A, C and D (1a,c,d) by treatment with glycine, L-serine and
L-threonine, respectively (Scheme 4). In addition, D-threonine
and D-valine were introduced to give nakijiquinones 24 and 25.
The IR and NMR spectroscopic data recorded for the synthetic
samples matched the data published.
Analogue 30
a KOtBu, CH3PPH3Br, toluene, reflux, 98% 26. b RhCl3, CHCl3/
EtOH, 2d, reflux, 95% 27. c AgO, HNO3, dioxane, rt, 33% 28a, 27%
28b, 23% 29. d 28b H2SO4, MeOH, 96% 28a.tfnte KOH, MeOH, H2O,
f
rt, 59% 29. L-Serine, NaHCO3, EtOH, 40 °C, 24 h, 32%.
Scheme 6. Synthesis of Nakijiquinone C Analogue 32
Bearing an Exocyclic Double Bond
Unexpectedly, however, the values determined for the specific
rotation of the synthetic nakijiquinones differed markedly from
the data published for the natural products (see the Supporting
Information). Most striking was the fact that in the case of
nakijiquinones A, C, and D and D-threonine derivative 24 not
only the value but, in particular, the direction of the specific
rotation was in fact opposite to the reported data, raising the
question of whether epimerization might have occurred in the
last step of the synthesis.
But as described above, the spectroscopic data recorded for
the synthetic nakijiquinones are in full accord with the published
values for the natural nakijiquinones. In addition, for example,
the diastereomer formed from isospongiaquinone and D-serine
(i.e., the possible epimerization product analogous to nakiji-
quinone C) shows a specific rotation that markedly differs from
the value for the synthetic sample.1 Furthermore, all data
(including the specific rotation) recorded for isospongiaquinone
(8) synthesized as described here are in full accord with the
values published for this natural product. Thus, if the absolute
configuration of 8 is correct, nakijiquinones 1a, c, d, 24, and
25 must have been formed with the correct absolute configu-
ration as well.28
Synthesis of Nakijiquinone Analogues. For the study of the
relationship between structure and biological activity of the
nakijiquinones, variation of the substitution and the stereochem-
istry of the trans Decaline-type core structure of the natural
products is of great importance. Thus, for initial studies of this
aspect, we have synthesized nakijiquinone C analogues 30 and
32 with altered configuration at C-2 and with an exocyclic
instead of an endocyclic double bond, respectively. For this
purpose, the reaction conditions optimized for the synthesis of
the nakijiquinones could be applied directly.
a AgO, HNO3, dioxane, rt, 26% 31a, 40% 31b, 4% 2. b 31b H2SO4,
d
MeOH, 96% 31a. c KOH, MeOH, H2O, rt, 72% 2. L-Serine, NaHCO3,
EtOH, 40 °C, 24 h, 30%.
nakijiquinone C (30). Nakijiquinone C analogue 32 was obtained
accordingly from intermediate 21 formed in the synthesis of
the natural product by olefination of the keto group at C-5 (see
above). Thus, the tetramethoxyaryl ring embedded in 21 was
oxidized to a mixture of the para- and the ortho-quinoid
derivative, and the ortho-quinone was rearranged to the para-
quinone under acidic conditions (Scheme 6). Regioselective
saponification of one vinylogous ester yielded ilimaquinone (2)
which finally was treated with L-serine to give nakijiquinone C
analogue 32.
Determination of the Biological Activity. To obtain a
preliminary picture of the biochemical properties of the nakiji-
quinones and related compounds obtained in the synthesis effort
detailed above, we investigated their ability to inhibit several
different receptor tyrosine kinases.
To this end, apart from Her-2/Neu (vide supra), EGFR (ErbB-
1), IGF1R, VEGFR2 (KDR), and VEGFR3 (flt-4) were selected
to cover a broad spectrum of tyrosine kinases.
As shown in Scheme 5, C-2 epimeric nakijiquinone C
analogue 30 was obtained from tetramethoxyaryl intermediate
20b obtained as the minor diastereomer from the olefination/
reduction of alkylation product 10 (see Scheme 3). To this end,
ketone 20b was converted into exocyclic olefin 26 which was
isomerized to endocyclic alkene 27. Oxidation of the aromatic
ring gave para-quinoid intermediate 28a and ortho-quinone 28b
in a nearly equal ratio as well as a substantial amount of the
desired selectively unmasked C-2 epimer of isospongiaquinone
(29). Ortho-quinone 28b was rearranged to para-quinone 28a
by treatment with acid. Regioselective saponification of one
vinylogous ester yielded desired intermediate 29. Finally,
vinylogous ester 29 was converted into the C-2 epimer of
The EGFR (epidermal growth factor receptor, ErbB-1), which
is closely related to Her-2/Neu, was one of the first tyrosine
kinases described. It has been implicated in human tumorigen-
esis, for example, of glioblastoma as well as in numerous tumors
of epithelial origin, including breast and esophageal tumors.29
The insulin-like growth factor 1 receptor (IGF1R) exerts
mitogenic, cell survival, and insulin-like activities by binding
its ligands IGF1 and IGF2. It is involved in postnatal growth
physiology and has been shown to be connected to proliferative
disorders such as breast cancer.30
VEGFR2 and VEGFR3 are both receptors for the vascular
endothelial growth factor (VEGF) family. The VEGFRs are
predominantly expressed on endothelial cells. Whereas VEGFR2
(28) The direction of the specific rotation reported for nakijiquinone C
isolated from natural sources is indeed incorrect. Reexamination of the
original sample yielded a value of [R]D ) +138° (c ) 0.1, EtOH):
Kobayashi, J. Hokkaido University, Sapporo, Japan. Personal communica-
tion.
(29) Heldin, C.; Ro¨nnstrand L. In Oncogenes and Tumor Suppressors;
Peters, G., Vousden, K., Eds. Oxford University Press: New York, 1997,
p 62.
(30) Ellis, M. J.; Jenkins, S.; Hanfelt, J.; Redington, M. E.; Taylor, M.;
Leek, R.; Siddle, K.; Harris, A. Breast Cancer Res. Treat. 1998, 52, 175.
20

Synthesis and EValuation of the Nakijiquinones
J. Am. Chem. Soc., Vol. 123, No. 47, 2001 11591
Figure 3. Proposed binding mode of 30 into the active site of KDR.
Figure 2. Dose dependent inhibition of KDR kinase activity in vitro
by 30. Each point represents the mean ( standard deviation of 4
independent reading points.
finding suggests that the stereochemistry of the terpenoid core
structure is an important determinant of the tyrosine kinase
inhibiting activity of the nakijiquinones. In particular, it appears
to determine the selectivity for individual kinases, because
nakijiquinone C is a selective Her-2/Neu inhibitor1 and 2-epi-
nakijiquinone C selectively targets VEGFR2.
is responsible for endothelial cell proliferation and blood vessel
permeability, the VEGF receptor 3 seems to be critical for
lymphatic vessel development. Both receptor tyrosine kinases
are essential for tumor angiogenesis and lymphangiogenesis,
respectively.31-33 The restrictive expression of VEGFR on
endothelial cells predestines them as targets for selective
therapeutic intervention.
Briefly, in the assay, the kinase-catalyzed phosphorylation
of poly(Glu-Tyr) in the presence of varying concentrations of
inhibitor was determined. The kinases were employed as fusion
proteins of glutathione-S-transferase (GST) and the respective
kinase domain. Kinase activity was determined by means of an
anti-phosphotyrosine antibody conjugated to horseradish per-
oxidase (POD). The chemiluminescence caused by the reaction
catalyzed by POD immobilized after antibody binding to
phosphotyrosine residues was measured (see the Supporting
Information for details).
Fourteen of the synthesized compounds were investigated as
possible inhibitors for the tyrosine kinases mentioned above.
This selected group included nakijiquinones 1a, 1c, 1d, 24, 25,
quinoid compounds 8, 8a, 22a, and 22b (having the same
absolute configuration and the endocyclic double bond as the
terpenoid nakijiquinone core), quinoid nakijiquinone analogues
2 and 32 (with an exocyclic double bond), tetramethoxyaryl
intermediates 9 and 21 (with exo- and endocyclic double bond
and the stereochemistry found in the natural product), and,
finally, compound 30 (the C-2 epimer of nakijiquinone C).
Nakijiquinones 1a, 1c, 1d, 24, and 25 are only poor inhibitors
or not inhibitors of the kinases investigated; that is, they do not
display substantial inhibition of the kinases in concentrations
lower that 30 µM.
Molecular Modeling Studies. Recent molecular modeling
studies performed on a homology model of the VEGFR2 (KDR)
structure34 provided valuable insights into the potential binding
modes of competitive inhibitors of ATP binding to this protein.
To further our understanding of the difference in activity
between the natural product nakijiquinone and C-2 epimer 30,
we undertook molecular modeling studies using the recently
published crystal structure of KDR35 and a model of the ATP
binding site based on a crystal structure of the FGF-R34 kinase
which has a high homology to KDR.36 The initial docking
studies performed with the GOLD software using the KDR
structure provided a good indication about potential binding
modes. However, these did not yield a satisfactory explanation
for the observed differences in activity.37 In contrast to the
structure of the FGF-R kinase which was used to construct an
ATP binding site model, the available KDR crystal structure is
not a complex with an inhibitor and, thus, contains a more open
active site.38 Using the FGF-R kinase derived model, it was
finally possible to determine a binding mode which is consistent
with the available SAR data. The proposed binding mode of
30 is shown in Figure 3.39
(34) Bold G.; Altmann, K.-H.; Frei, J.; Lang, M.; Manley, P. W.; Traxler,
P.; Wietfeld, B.; Bru¨ggen, J.; Buchdunger, E.; Cozens, R.; Ferrari, S.; Furet,
P.; Hofmann, F.; Martiny-Baron, G.; Mestan, J.; Ro¨sel, J.; Sills, M.; Stover,
D.; Acemoglu, F.; Boss, E.; Emmenegger, R.; La¨sser, L.; Masso, E.; Roth,
R.; Schlachter, C.; Vetterli, W.; Wyss, D.; Wood, J. M. J. Med. Chem.
2000, 43, 2310.
(35) McTigue M. A.; Wickersham, J. A.; Pinko, C.; Showalter, R. E.;
Parast, C. V.; Tempczyk-Russel, A.; Gehring, M. R.; Mroczkowski, B.;
Kann, C.-C.; Villafranca, J. E.; Appelt, K. Structure 1999, 7, 319.
(36) Hubbard S. R.; Till, J. H. Annu. ReV. Biochem. 2000, 69, 373.
(37) (a) GOLD is the result of a collaborative LINK project between
Sheffield University, Glaxo Wellcome, and the Cambridge Crystallographic
Data Centre (CCDC). (b) Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.;
Taylor, R. Development and Validation of a Genetic Algorithm for Flexible
Docking. J. Mol. Biol. 1997, 267, 727. (c) Jones, G.; Willett, P.; Glen, R.
C. Molecular Recognition of Receptor Sites Using a Genetic Algorithm
with a Description of Desolvation. J. Mol. Biol. 1995, 245, 43. (d) A
thorough discussion of the accuracy of GOLD predictions and of known
problems can be found at http://www.ccdc.cam.ac.uk/prods/gold/ccdc_test.
html.
The same is true for quinoid analogues 8, 8a, 22a, 22b, 2,
and 32 and tetramethoxyaryl compounds 9 and 21 which all
have the same arrangement of stereocenters as the nakijiqui-
nones.
However, 2-epi-nakijiquinone C (30) turned out to be a good
and selective inhibitor of VEGFR2 (KDR). It inhibits this
tyrosine kinase with an IC50 value of 21 µM (at a concentration
of 25 µM ATP) (Figures 2 and 3) and does not display
remarkable activity toward the other kinases investigated. This
(31) Carmeliet, P.; Jain, R. K. Nature 2000, 407, 249.
(32) Saaristo, A.; Partanen, T. A.; Arola, J.; Jussila, L.; Hytonen, M.;
Makitie, A.; Vento, S.; Kaipainen, A.; Malmberg, H.; Alitalo, K. Am. J.
Pathol. 2000, 157, 7.
(38) Mohammadi, M.; Schlessinger, J.; Hubbard, S. R. Cell 1996, 86,
577.
(39) Figures 3-5 were created using WitnotP. WitnotP is a molecular
modeling program developed by and licenced from Novartis AG, Basel.
For further information on WitnotP, contact A. Widmer, Novartis AG, Basel.
(33) Karkkainen, M. J.; Petrova, T. V. Oncogene 2000, 19, 5598.

11592 J. Am. Chem. Soc., Vol. 123, No. 47, 2001
Stahl et al.
Figure 4. Compound 30 within the active site of KDR. Hydrogen
bonds are indicated by green dotted lines. The hydrogen bonds to Cys
919 and Glu 917 are 2.0 and 2.7 Å long, respectively. Those hydrogen
bonds possibly formed with Leu 840 and Asn 923 measure 2.7 and
2.5 Å, respectively.
Figure 5. Nakijiquinone 1c within the active site of KDR. The clashing
groups are indicated by color: Nakijiquinone C-2 (magenta) and L1035
(red).
compounds 9 and 21 clashes with V848, which may also
contribute to their inactivity toward KDR.
The Decalin subunit of 30 fits into a hydrophobic pocket
formed by V916,40 V914, V899, L889, C1045, F1047, and the
hydrocarbon part of the side chain of K868 near the nucleotide
binding loop of KDR. In the activated form of the kinase, K868
is thought to form a salt brigde to E885, which ensures the
proper coordination for two of the phosphate oxygens of ATP.
The chair-twist conformation of the Decalin moiety displayed
in Figure 3 was found to be the most stable conformation in
the environment of the hydrophobic pocket of our model. The
L-serine moiety contributes a hydrogen bond to the backbone
NH of N923. The phenolic OH and the adjacent quinone
carbonyl group engage in a bidentate manner the backbone
carbonyl of E917 and the backbone NH of C919 in hydrogen
bonding interactions. Possibly another hydrogen bond forms
between the serine NH and the backbone carbonyl group of
L840, a residue belonging to the nucleotide binding loop.
Considering that 30 was found to be a competitive inhibitor of
ATP binding and that E917 and C919 correspond to the residues
of the hinge loop that are seen to make hydrogen bonds
with the purine moiety of ATP and with heterocyclic inhibitors
in the crystal structure of all protein kinases determined
thus far,34,41 this binding mode appears very reasonable (see
Figure 4).
Conclusion
In conclusion, we have developed the first enantioselective
route to the nakijiquinones, the only natural products known to
selectively inhibit the Her-2/Neu protooncogene. This synthetic
route gives access to various analogues of these interesting and
biologically relevant kinase inhibitors. Thereby, new opportuni-
ties for the development of selective inhibitors of the Her-2/
Neu and the KDR receptor tyrosine kinases and for the study
of the biological phenomena influenced by these enzymes may
be opened up. Compound 30 represents a promising starting
point for the development of more potent inhibitors for the KDR
receptor. Such a new starting point may be of particular
relevance, because other KDR inhibitors have already been
advanced into clinical studies. Thus, currently, a small molecule
inhibitor of KDR, SU5416, is under evaluation for the treatment
of different human cancers.41 It was shown that the antitumor
effect results from the inhibition of VEGF induced angiogenesis
and is not caused by any cytotoxicity.42
Furthermore, recently, some more potent inhibitors of the FGF
and KDR receptors with nanomolar affinities toward the isolated
kinases have been reported. The new KDR inhibitor identified
by us is significantly less potent than these compounds.
However, its underlying structure is not comparable with the
basic structure types of the other KDR inhibitors reported so
far.
While this pattern of hydrogen bonds can also be attained
by nakijiquinone 1c, the Decalin moiety of the natural product
cannot be fit into the hydrophobic pocket in the way described
for epimer 30. We attribute this to a possible clash of the
equatorial methyl group at C-2 with L1035 (see Figure 5).
The new KDR receptor inhibitor compound may, therefore,
open up the opportunity to develop a new approach for the
therapy of angiogenesis-dependent diseases, such as solid and
epithelial tumors, as well as for diabetic retinopathy and
rheumatoid arthritis.43,44
This model also allows for an explanation for the lack of
activity exhibited by compounds 1a, 1c, 1d, 2, 8, 8a, 9, 21,
22a, 22b, 24, 25, and 32. With the exception of 1a, 1c, 1d, and
32, these compounds cannot form the hydrogen bonds described
for 30 and exhibit the same stereochemistry on the Decalin
moiety as 1c. In addition, one of the methoxy groups in
(42) Fong, T. A.; Shawver, L. K.; Sun, L.; Tang, C.; App, H.; Powell,
T. J.; Kim, Y. H.; Schreck, R.; Wang, X.; Risau, W.; Ullrich, A.; Hirth, K.
P.; McMahon, G. Cancer Res. 1999, 59, 99.
(40) Residues were numbered according to the numbering proposed by
McTigue et al.35
(43) Hamby, J. M.; Showalter, H. D. Pharmacol. Ther. 1999, 82, 169.
(44) Sun, L.; McMahon, G. Drug DiscoVery Today 2000, 5, 344.
(45) Vriend, G. J. Mol. Graphics 1990, 8, 52-56.
(46) (a) 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. (b) Mohamadi, F.; Richards, N.
G.; Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.;
Hendrickson, T.; Still, W. C. J. Comput. Chem. 1990, 11, 440.
(47) Hanks, S. K.; Quinn, A. M. Methods Enzymol. 1991, 200, 38.
(41) As of December 2000; for current state of clinical trials, see
http://cancertrials.nci.nih.gov. SU5461 inhibits the isolated VEGF-R2 and
the FGF-R1 tyrosine kinases with IC50 values of 0.7 and 7 µM at an
effective ATP concentration of 10 µM, respectively. The PDGF-R is
inhibited with an IC50-value of 10.5 µM at 20 µM ATP. Sun, L.; Tran, N.;
Liang, C.; Tang, F.; Rice, A.; Schreck, R.; Waltz, K.; Shawver, L. K.;
McMahon, G.; Tang, C. J. Med. Chem. 1999, 42, 5120-5130.

Synthesis and EValuation of the Nakijiquinones
J. Am. Chem. Soc., Vol. 123, No. 47, 2001 11593
Acknowledgment. This research was supported by the Fonds
der Chemischen Industrie. R.M. and A.H. are grateful to the
Land Baden-Wu¨rttemberg for a scholarship (Landesgraduierten-
fo¨rderung). L.K. is grateful to the Studienstiftung des deutschen
Volkes and the Fonds der Chemischen Industrie for scholarships.
Supporting Information Available: Experimental proce-
dures and spectroscopic and analytical data for all compounds
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
(48) (a) McDonald, D. Q.; Still, W. C. Tetrahedron Lett. 1992, 33, 7743.
(b) Weiner, S. J.; Kollman, P.; Case, D. A.; Singh, U. C.; Ghio, C.; Alagona,
S.; Profeta, S.; Weiner, P. J. Am. Chem. Soc. 1984, 106, 765.
(49) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. J. Am.
Chem. Soc. 1990, 112, 6127.
(50) For further details on the construction of the VEGFR2 homology
model, please contact Dr. Pascal Furet at pascal.furet@pharma.novartis.com.
JA011413I