Studies in polyphenol chemistry and bioactivity. 1. Preparation of building blocks from (+)-catechin. Procyanidin formation. Synthesis of the cancer cell growth inhibitor, 3-O-galloyl-(2R,3R)-epicatechin-4β,8-[3-O-galloyl-(2R,3R)-epicatechin]

Article abstract of DOI:10.1021/ja993020d

A project has been initiated to synthesize proanthocyanidin oligomers found in cocoa. Natural, readily available (+)-catechin was transformed into 5,7,3′,4′-tetra-O-benzyl-(-)-epicatechin (14) by (a) benzylation of the phenolic oxygens; (b) oxidation of the 3-alcohol to ketone by the Dess-Martin periodinane; and (c) reduction with lithium tri-sec-butylborohydride (L-Selectride) in the presence of LiBr. The additive diminishes the extent of ketone enolization while maintaining a stereoselectivity of ≥ 200:1. Oxidation of 14 with DDQ was performed best from the standpoint of product purification if ethylene glycol was used as the nucleophilic trapping agent. The resulting ether 19 was condensed with 14 using TiCl₄ to give a good yield of benzyl-protected epicatechin-4β,8-epicatechin (octa-O-benzylprocyanidin B₂, 20) as the sole dimeric product. Hydrogenolysis of 20 yielded procyanidin B₂ in the first enantiospecific synthesis of this natural product which employs protected intermediates and thereby allows the necessary product separation after the condensation step to be performed on nonpolar, nonsensitive intermediates. Acylation of 20 with tri-O-benzylgalloyl chloride followed by hydrogenolysis gave access to the title bis-gallate (24). This constitutes the first synthesis of this natural product, a compound notable for its PKC-inhibitory and anticancer activity.

Full text of DOI:10.1021/ja993020d

J. Am. Chem. Soc. 1999, 121, 12073-12081
12073
Studies in Polyphenol Chemistry and Bioactivity. 1. Preparation of
Building Blocks from (+)-Catechin. Procyanidin Formation. Synthesis
of the Cancer Cell Growth Inhibitor, 3-O-Galloyl-(2R,3R)-epicatechin-
4â,8-[3-O-galloyl-(2R,3R)-epicatechin]
Werner Tu1ckmantel,*,† Alan P. Kozikowski,*,† and Leo J. Romanczyk, Jr.‡
Contribution from Georgetown UniVersity Medical Center, Institute for CognitiVe and Computational
Sciences, Drug DiscoVery Program, 3900 ReserVoir Rd. NW, Washington, D.C. 20007, and M&M Mars,
High Street, Hackettstown, New Jersey 07840
ReceiVed August 19, 1999
Abstract: A project has been initiated to synthesize proanthocyanidin oligomers found in cocoa. Natural, readily
available (+)-catechin was transformed into 5,7,3′,4′-tetra-O-benzyl-(-)-epicatechin (14) by (a) benzylation
of the phenolic oxygens; (b) oxidation of the 3-alcohol to ketone by the Dess-Martin periodinane; and (c)
reduction with lithium tri-sec-butylborohydride (L-Selectride) in the presence of LiBr. The additive diminishes
the extent of ketone enolization while maintaining a stereoselectivity of g200:1. Oxidation of 14 with DDQ
was performed best from the standpoint of product purification if ethylene glycol was used as the nucleophilic
trapping agent. The resulting ether 19 was condensed with 14 using TiCl4 to give a good yield of benzyl-
protected epicatechin-4â,8-epicatechin (octa-O-benzylprocyanidin B2, 20) as the sole dimeric product.
Hydrogenolysis of 20 yielded procyanidin B2 in the first enantiospecific synthesis of this natural product which
employs protected intermediates and thereby allows the necessary product separation after the condensation
step to be performed on nonpolar, nonsensitive intermediates. Acylation of 20 with tri-O-benzylgalloyl chloride
followed by hydrogenolysis gave access to the title bis-gallate (24). This constitutes the first synthesis of this
natural product, a compound notable for its PKC-inhibitory and anticancer activity.
Introduction
diet2 and are found in traditional herbal medicines.3 As a
consequence of their multiple polar functionality, they interact
strongly but often unselectively with proteins, frequently result-
ing in the precipitation of insoluble protein-polyphenol com-
plexes.4 This reaction is the basis of the tanning process of
leather, for which reason certain classes of polyphenols are
commonly referred to as tannins. In addition, polyphenols
readily react with one-electron oxidants, resulting in powerful
free-radical scavenging (antioxidant) activity,5 and they complex
Fe2+ which is an initiator of radical formation.6 On the other
hand, certain polyphenols may under certain conditions exhibit
the opposite (prooxidant) effect.7 Cell damage by free radicals
is an important cause of various diseases. While protein binding
and radical scavenging are widespread properties among
Polyphenols are an important class of natural products.1 They
are ubiquitous in the plant kingdom and thus enter the human
† Georgetown University Medical Center.
‡ M&M Mars.
(1) (a) Freudenberg, K.; Weinges, K. Fortschr. Chem. Org. Naturstoffe
1958, 16, 1. (b) Weinges, K.; Ba¨hr, W.; Ebert, W.; Go¨ritz, K.; Marx, H.-D.
Fortschr. Chem. Org. Naturstoffe/Prog. Chem. Org. Nat. Prod. 1969, 27,
158. (c) The Chemistry of FlaVonoid Compounds; Geissman, T. A., Ed.;
Macmillan: New York, 1962. (d) Haslam, E. Chemistry of Vegetable
Tannins; Academic Press: London, New York, 1966. (e) Haslam, E. Plant
Polyphenols: Vegetable Tannins ReVisited; Cambridge University Press:
Cambridge, 1989. (f) Haslam, E. Rec. AdV. Phytochem. 1979, 12, 475. (g)
Haslam, E.; Cai, Y. Nat. Prod. Rep. 1994, 41. (h) Haslam, E. Practical
Polyphenolics-From Structure to Molecular Recognition and Physiological
Action; Cambridge University Press: Cambridge, 1998. (i) The FlaVonoids;
Harborne, J. B., Mabry, T. J., Mabry, H., Eds.; Chapman & Hall: London,
1975. (j) The FlaVonoids: AdVances in Research; Harborne, J. B., Mabry,
T. J., Eds.; Chapman & Hall: London, New York, 1982. (k) The FlaVonoids:
AdVances in Research Since 1980; Harborne, J. B., Ed.; Chapman & Hall:
New York, 1988. (l) The FlaVonoids: AdVances in Research Since 1986;
Harborne, J. B., Ed.; Chapman & Hall: London etc., 1994. (m) Plant
Phenolics (Methods in Plant Biochemistry, Vol. 1); Harborne, J. B., Ed.;
Academic Press: London, 1989. (n) Farkas, M.; Gabor, M.; Kallay, F.
FlaVonoids and BioflaVonoids. Current Research Trends; Elsevier: New
York, 1977. (o) Wollenweber, E.; Dietz, V. H. Phytochemistry 1981, 20,
869. (p) Okuda, T.; Yoshida, T.; Hatano, T. J. Nat. Prod. 1989, 52, 1. (q)
Nonaka, G. Pure Appl. Chem. 1989, 61, 357. (r) Okuda, T.; Yoshida, T.;
Hatano, T. Phytochemistry 1993, 32, 507. (s) Okuda, T.; Yoshida, T.;
Hatano, T. Fortschr. Chem. Org. Naturstoffe/Prog. Chem. Org. Nat. Prod.
1995, 66, 1. (t) Chemistry and Significance of Condensed Tannins;
Hemingway, R. W., Karchesy, J. J., Eds.; Plenum: New York, 1989. (u)
Plant Polyphenols. Synthesis, Properties, Significance; Hemingway, R. W.,
Laks, P. E., Eds.; Plenum: New York, London, 1992. (v) Ferreira, D.;
Steynberg, J. P.; Roux, D. G.; Brandt, E. V. Tetrahedron 1992, 48, 1743.
(w) Ferreira, D.; Bekker, R. Nat. Prod. Rep. 1996, 13, 411. (x) Phenolic
Metabolism in Plants; Stafford, H. A., Ibrahim, R. K., Eds.; Plenum: New
York, 1992. (y) Quideau, S.; Feldman, K. S. Chem. ReV. 1996, 96, 475.
(2) (a) Macheix, J.-J.; Fleuriet, A. Fruit Phenolics; CRC Press: Boca
Raton, FL, 1990. (b) Robards, K.; Antolovich, M. Analyst 1997, 122, 11R.
(3) (a) Reference 1e, pp 155-161. (b) Haslam, E.; Lelley, T. H.; Cai,
Y.; Martin, R.; Magnolato, D. Planta Med. 1989, 55, 1. (c) Haslam, E. J.
Nat. Prod. 1996, 59, 205. (d) Pietta, P. In FlaVonoids in Health and Disease;
Rice-Evans, C. A., Packer, L., Eds.; Marcel Dekker: New York, Basel,
Hong Kong, 1998; pp 61-110. (e) Tang, W.; Eisenbrand, G. Chinese Drugs
of Plant Origin. Chemistry, Pharmacology, and Use in Traditional and
Modern Medicine; Springer: Berlin etc., 1992; especially Nos. 5, 42, 44,
62, 63, 74, 75, 88, 107, 112-114. (f) Hsu, H.-Y.; Chen, Y.-P.; Hong, M.
The Chemical Constituents of Oriental Herbs; Oriental Healing Arts
Institute: Long Beach, CA, 1982/1985; Vol. 1, pp 374-575, Vol. 2, pp
227-367. (g) Okuda, T.; Yoshida, T.; Hatano, T. In Economic and
Medicinal Plant Research, Vol. 5: Plants and Traditional Medicines;
Wagner, H. S., Farnsworth, N. R., Eds.; Academic Press: London, 1991;
pp 129-165. (h) Okuda, T.; Yoshida, T.; Mori, K.; Hatano, T. Heterocycles
1981, 15, 1323. (i) Okuda, T.; Yoshida, T.; Hatano, T. Planta Med. 1989,
55, 117. (j) Okuda, T.; Yoshida, T.; Hatano, T. Basic Life Sci. 1992, 59,
539.
(4) (a) Hagerman, A. E. in ref 1t, pp 323-333. (b) Hagerman, A. E.;
Rice, M. E.; Richard, N. T. J. Agric. Food Chem. 1998, 46, 2590.
10.1021/ja993020d CCC: $18.00 © 1999 American Chemical Society
Published on Web 12/04/1999

12074 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Tu¨ckmantel et al.
neuroprotective,16 antiinflammatory,17 antibacterial,18 and hy-
potensive,19 and inhibit the growth of cancer cells.5b,20 Epigal-
locatechin gallate (5), a polyphenol abundant in green tea,
inhibits the growth of human PC-3 and LNCaP 104-R prostate
tumor cells and of human MCF-7 mammary cancer cells in nude
mice while the structurally related compounds, epicatechin
gallate (3) and epigallocatechin (4), are inactive.21 A number
of polyphenols and crude plant extracts containing such
compounds inhibit tumor promotion (or biochemical markers
thereof) by 12-O-tetradecanoylphorbol-13-acetate in mouse
epidermis in vivo.22 Additional mechanisms of action besides
radical scavenging or nonspecific protein binding are believed
to be operative in the anticancer activity of at least some
compounds. The antitumor polyphenol coriariin A stimulates
the secretion of interleukin-1â and tumor necrosis factor-R, and
the latter protein has been proposed to be responsible for the
observed activity.23 Epigallocatechin and epigallocatechin gallate
have been shown not only to inhibit leukemia cell growth but
to induce apoptosis.24 Epigallocatechin gallate inhibits urokinase,
an enzyme crucial for cancer growth.25 These and related
findings explain the increasing popularity of polyphenol prepa-
rations, such as extracts of green tea, grape seeds, and pine bark,
as dietary supplements, even though the investigation of
absorption and bioavailability has lagged epidemiological studies
Figure 1. Structures and atom numbering of catechin and related
natural products.
polyphenols, little information was available until recently as
to whether polyphenols also participate in selective interactions
with enzymes or receptors. It has now, however, been shown
that polyphenols are indeed capable of such selective interac-
tions. A variety of eukaryote protein kinases including PKC
are inhibited by Ribes nigrum condensed tannin with IC50 values
ranging from 9 nM to 16 µM, and the IC50 values for three out
of four kinases vary by an order of magnitude for different
condensed tannin preparations.8 Specific intermolecular NOE
cross-peaks are found in solutions containing both (+)-catechin
(1) (Figure 1) and any of several small peptides, indicating the
existence of well-defined binding interactions.9 Rough binding
data have been obtained for 20 phenolic compounds at 16
receptors, and activities ranging from a high level of binding
to no binding at all have been observed both for certain receptors
when combined with different polyphenols and for certain
polyphenols when combined with different receptors.10 Numer-
ous other biological activities have been reported for polyphe-
nols; for example, they inhibit viral reverse transcriptase,11
inhibit the replication of HIV 1 in vitro,12 reduce the risk of
heart disease,13 suppress ulcer formation,14 are antimutagenic,15
(14) Saito, M.; Hosoyama, H.; Ariga, T.; Kataoka, S.; Yamaji, N. J. Agric.
Food Chem. 1998, 46, 1460.
(15) (a) Liviero, L.; Puglisi, P. P.; Morazzoni, P.; Bombardelli, E.
Fitoterapia 1994, 65, 203. (b) Tamagawa, K.; Iizuka, T.; Kobori, M.;
Shinmoto, H.; Tsushida, T. Nippon Shokuhin Kagaku Kogaku Kaishi 1998,
45, 420; Chem. Abstr. 1998, 129, 215951. (c) Constable, A.; Varga, N.;
Richoz, J.; Stadler, R. H. Mutagenesis 1996, 11, 189.
(16) Inanami, O.; Watanabe, Y.; Syuto, B.; Nakano, M.; Kuwabara, M.
Free Rad. Res. 1998, 29, 359.
(17) (a) Hartisch, C.; Kolodziej, H.; von Bruchhausen, F. Planta Med.
1997, 63, 106. (b) Blazso, G.; Gabor, M.; Rohdewald, P. Pharmazie 1997,
52, 380.
(18) Scalbert, A. Phytochemistry 1991, 30, 3875.
(19) Kameda, K.; Takaku, T.; Okuda, H.; Kimura, Y.; Okuda, T.; Hatano,
T.; Agata, I.; Arichi, S. J. Nat. Prod. 1987, 50, 680.
(5) (a) FlaVonoids in Health and Disease; Rice-Evans, C. A., Packer,
L., Eds.; Marcel Dekker: New York, Basel, Hong Kong, 1998. (b) Phenolic
Compounds in Foods and Their Effects on Health, II: Antioxidants and
Cancer PreVention; ACS Symp. Ser. No. 507; Huang, M.-T., Ho, C.-T.,
Lee, C. Y., Eds.; American Chemical Society: Washington, DC, 1992. (c)
Plumb, G. W.; De Pascual-Teresa, S.; Santos-Buelga, C.; Cheynier, V.;
Williamson, G. Free Rad. Res. 1998, 29, 351. (d) Saint-Cricq de Gaulejac,
N.; Provost, C.; Vivas, N. J. Agric. Food Chem. 1999, 47, 425. (e) Gao,
D.; Tawa, R.; Masaki, H.; Okano, Y.; Sakurai, H. Chem. Pharm. Bull. 1998,
46, 1383. (f) Ghiselli, A.; Nardini, M.; Baldi, A.; Scaccini, C. J. Agric.
Food Chem. 1998, 46, 361. (g) Nakao, M.; Takio, S.; Ono, K. Phytochem-
istry 1998, 49, 2379. (h) Sawai, Y.; Sakata, K. J. Agric. Food Chem. 1998,
46, 111. (i) Benzie, I. F. F.; Szeto, Y. T. J. Agric. Food Chem. 1999, 47,
633.
(6) (a) Morel, I.; Cillard, P.; Cillard, J. in ref 5a, pp 163-177. (b)
Yoshino, M.; Murakami, K. Anal. Biochem. 1998, 257, 40.
(7) (a) Metodiewa, D.; Jaiswal, A. K.; Cenas, N.; Dickancaite, E.; Segura-
Aguilar, J. Free Rad. Biol. Med. 1999, 26, 107. (b) Alanko, J.; Riutta, A.;
Holm, P.; Mucha, I.; Vapaatalo, H.; Metsa-Ketela, T. Free Rad. Biol. Med.
1999, 26, 193.
(20) (a) Food Phytochemicals for Cancer PreVention, I: Fruits and
Vegetables; ACS Symp. Ser. No. 546; Huang, M.-T., Osawa, T.; Ho, C.-
T., Rosen, R. T., Eds.; American Chemical Society, Washington, DC, 1994.
(b) Food Phytochemicals for Cancer PreVention, II: Teas, Spices, and Herbs;
ACS Symp. Ser. No. 547; Ho, C.-T., Osawa, T., Huang, M.-T., Rosen, R.
T., Eds.; American Chemical Society, Washington, DC, 1994. (c) Kato,
R.; Nakadate, T.; Yamamoto, S.; Sugimura, T. Carcinogenesis 1983, 4,
1301. (d) Miyamoto, K.; Kishi, N.; Koshiura, R.; Yoshida, T.; Hatano, T.;
Okuda, T. Chem. Pharm. Bull. 1987, 35, 814. (e) Deschner, E. E.; Ruperto,
J.; Wong, G.; Newmark, H. L. Carcinogenesis 1991, 7, 1193. (f) Kashiwada,
Y.; Nonaka, G.; Nishioka, I.; Chang, J.-J.; Lee, K.-H. J. Nat. Prod. 1992,
55, 1033. (g) Miyamoto, K.; Murayama, T.; Nomura, M.; Hatano, T.;
Yoshida, T.; Furukawa, T.; Koshiura, R.; Okuda, T. Anticancer Res. 1993,
13, 37. (h) Kolodziej, H.; Haberland, C.; Woerdenbag, H. J.; Konings, A.
W. T. Phytother. Res. 1995, 9, 410. (i) Hoffman, R. Biochem. Biophys.
Res. Commun. 1995, 211, 600. (j) Murakami, A.; Ohigashi, H.; Koshimizu,
K. Biosci., Biotechnol., Biochem. 1996, 60, 1. (k) Valcic, S.; Timmermann,
B. N.; Alberts, D. S.; Wachter, G. A.; Krutzsch, M.; Wymer, J.; Guillen, J.
M. Anticancer Drugs 1996, 7, 461. (l) Matsumoto, N.; Kohri, T.; Okushio,
K.; Hara, Y. Jpn. J. Cancer Res. 1996, 87, 1034. (m) Gensler, H. L.;
Timmermann, B. N.; Valcic, S.; Wachter, G. A.; Dorr, R.; Dvorakova, K.;
Alberts, D. S. Nutr. Cancer 1996, 26, 325. (n) Kohlmeier, L.; Weterings,
K. G.; Steck, S.; Kok, F. J. Nutr. Cancer 1997, 27, 1. (o) So, F. V.; Guthrie,
N.; Chambers, A. F.; Carroll, K. K. Cancer Lett. 1997, 112, 127.
(21) Liao, S.; Umekita, Y.; Guo, J.; Kokontis, J. M.; Hiipakka, R. A.
Cancer Lett. 1995, 96, 239.
(8) Wang, B. H.; Foo, L. Y.; Polya, G. M. Phytochemistry 1996, 43,
359.
(9) Hatano, T.; Hemingway, R. W. J. Chem. Soc., Chem. Commun. 1996,
2537.
(10) Zhu, M.; Phillipson, J. D.; Greengrass, P. M.; Bowery, N. E.; Cai,
Y. Phytochemistry 1997, 44, 441.
(11) (a) Kakiuchi, N.; Hattori, M.; Namba, T.; Nishizawa, M.; Yamagishi,
T.; Okuda, T. J. Nat. Prod. 1985, 48, 614. (b) Kakiuchi, N.; Kusumoto, I.
T.; Hattori, M.; Namba, T.; Hatano, T.; Okuda, T. Phytother. Res. 1991, 5,
270. (c) Chu, S. C.; Hsieh, Y.-S.; Lin, J.-Y. J. Nat. Prod. 1992, 55, 179.
(12) Shahat, A. A.; Ismail, S. I.; Hammouda, F. M.; Azzam, S. A.;
Lemie`re, G.; De Bruyne, T.; De Swaef, S.; Pieters, L.; Vlietinck, A.
Phytomedicine 1998, 5, 133.
(13) (a) Samman, S.; Lyons Wall, P. M.; Cook, N. C. in ref 5a, pp 469-
481. (b) Aviram, M.; Fuhrman, B. Atherosclerosis 1998, 137, S45. (c)
Yamakoshi, J.; Kataoka, S.; Koga, T.; Ariga, T. Atherosclerosis 1999, 142,
139.
(22) (a) Gali, H. U.; Perchellet, E. M.; Gao, X. M.; Bottari, V.; Perchellet,
J. P. Anticancer Res. 1993, 13, 915. (b) Gali, H. U.; Perchellet, E. M.;
Gao, X. M.; Laks, P. E.; Perchellet, J. P. Cancer Lett. 1993, 72, 149. (c)
Gali, H. U.; Perchellet, E. M.; Gao, X. M.; Karchesy, J. J.; Perchellet, J. P.
Planta Med. 1994, 60, 235.
(23) Feldman, K. S.; Sarasrabudhe, K.; Smith, R. S.; Scheuchenzuber,
W. J. Bioorg. Med. Chem. Lett. 1999, 9, 985.
(24) Hibasami, H.; Achiwa, Y.; Fujikawa, T.; Komiya, T. Anticancer
Res. 1996, 16, 1943.
(25) Jankun, J.; Selman, S. H.; Swiercz, R.; Skrzypczak-Jankun, E. Nature
1997, 387, 561.

Studies in Polyphenol Chemistry and BioactiVity
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 12075
and in vitro experiments,26 and the economic importance of
polyphenols for health-related products can no longer be
dismissed.
material purified from cocoa and to determine structure-activity
relationships in various in vitro and ultimately in vivo models
of anticancer activity.
One major class of polyphenolic natural products, the
hydrolyzable tannins, are derived from carbohydrates by es-
terification with tannic acid or more complex polyhydroxylated
aromatic carboxylic acids. The second important group, and the
one on which our current interest is focused, is derived from
flavan-3-ols bearing one or several additional hydroxyl groups
on their aromatic rings. Eleven different hydroxylation patterns
of the A and B rings have been found in nature.27 Flavan-3-ols
are biosynthetically derived from (2S)-phenylalanine via flavan-
3,4-diols.28 These latter intermediates readily form a highly
stabilized carbenium ion (or quinone methide)29 in position 4
which attacks the A ring of a flavan-3-ol in what is essentially
a Friedel-Crafts alkylation process, forming an interflavan
bond.30 This process can be repeated once or several times,
resulting in chain-type oligomers which together with the dimers
are known as nonhydrolyzable tannins, condensed tannins, or
proanthocyanidins. The structural complexity of these com-
pounds rapidly increases with their chain length as a conse-
quence of different hydroxylation patterns and C-3 stereochem-
istry in the monomer units and different regio- and stereochem-
istries of the interflavan linkages, as well as additional structural
modifications. In addition, chain branching may occur by
alkylation of a monomer unit in both its 6- and 8-positions.
The isolation of pure compounds from natural sources thus
becomes increasingly difficult with increasing degree of oligo-
merization. Degradation by thiolysis31 permits identification of
the underlying building blocks but the task of elucidating the
position and stereochemistry of the interflavan linkages is
nontrivial. Both of these factors have resulted in few defined
oligomers above the tetramer level being described in the
literature. Proanthocyanidins and their parent monomers also
occur naturally in the form of a variety of derivatives, for
example, glycosides or esters with hydroxylated aromatic
carboxylic acids, such as gallic or hexahydroxydiphenic acid.
An example is known in which a gallate derivative exhibits
anticancer activity not found in the parent polyphenol.21
The synthetic challenge posed by proanthocyanidins is related
to the difficulty in controlling the interflavan regio- and
stereochemistry, as well as the sensitivity of the nonprotected
compounds to acids,36,37 bases,1v,36,38 and oxidizing agents.39 The
condensation between flavan-3-ols and 4-substituted, electro-
philic flavans has traditionally been performed without the use
of phenol protecting groups in a mildly acidic medium1v,31b,38,40
or recently with AgBF4 for SBn as the 4-substituent.41 The
products are mixtures of regio- and sometimes stereoisomers,
as well as higher oligomers despite the application of an excess
of the nucleophilic building block. They have usually been
separated by gel chromatography on Sephadex LH-20,31a
a
tedious process that requires a considerable investment of time
to develop for each particular separation task because of the
nonavailability of fast analytical tools such as HPLC columns
or TLC plates for this adsorbent. In addition, optically pure,
nonprotected 4-substituted catechins and epicatechins are not
readily available, being prepared by reduction of the expensive
natural product, (+)-taxifolin (the 4-ketone),42 or by in situ
degradation38a,40a,b or thiolytic degradation40a of natural proan-
thocyanidin oligomer fractions for which commercial sources
are difficult to identify or nonexisting.
It is therefore not surprising that efforts were undertaken prior
to our studies to address the possibility of building oligomeric
proanthocyanidins from protected building blocks. As an
additional incentive, protection of the phenolic but not of the
alcoholic hydroxyls would permit the regioselective elaboration
of derivatives such as 3-esters and -glycosides, as has been done
in the case of catechin using acetyl protecting groups.43 An
(34) (a) Forsyth, W. G. C.; Roberts, J. B. Biochem. J. 1960, 74, 374. (b)
Quesnel, V. C. Phytochemistry 1968, 7, 1583. (c) Jalal, M. A. F.; Collin,
H. A. Phytochemistry 1977, 16, 1377. (d) Porter, L. J.; Ma, Z.; Chan, B.
G. Phytochemistry 1991, 30, 1657. (e) Lee, M. C.; Chang, Y. Y.; Yim, M.
H. Han’guk Nonghwa Hakhoechi 1998, 41, 110-7; Chem. Abstr. 1998,
129, 207072. (f) Sanbongi, C.; Osakabe, N.; Natsume, M.; Takizawa, T.;
Gomi, S.; Osawa, T. J. Agric. Food Chem. 1998, 46, 454. (g) Hammerstone,
J. F.; Lazarus, S. A.; Mitchell, A. E.; Rucker, R.; Schmitz, H. H. J. Agric.
Food Chem. 1999, 47, 490.
Among the proanthocyanidins, two subtypes, the procyanidins
(5,7,3′,4′-hydroxylation) and prodelphinidins (5,7,3′,4′,5′-hy-
droxylation), are widespread in human foodstuffs,2,31a,32 e.g.,
cocoa.33-35 Cocoa procyanidins consist predominantly of epi-
catechin building blocks,34d and oligomers up to the size of the
decamer have been identified.34g From the pentamer on, these
oligomers exhibit growth inhibitory activity against various
cancer cell lines.35 To confirm the structures assigned to these
compounds, we have embarked on a program to synthesize
epicatechin oligomers of defined structure for comparison with
(35) (a) Romanczyk, L. J., Jr.; Hammerstone, J. F., Jr.; Buck, M. M.
U.S. Patent 5,554,645, Sept 10, 1996. (b) Romanczyk, L. J., Jr.; Hammer-
stone, J. F., Jr.; Buck, M. M. U.S. Patent 5,712,305, Jan 27, 1998. (c)
Romanczyk, L. J., Jr.; Hammerstone, J. F., Jr.; Buck, M. M.; Post, L. S.;
Cipolla, G. G.; McClelland, C. A.; Mundt, J. A.; Schmitz, H. H. PCT Int.
Appl. WO 9736,597, Oct 9, 1997; Chem. Abstr. 1997, 127, 322787. (d)
Romanczyk, L. J., Jr.; Hammerstone, J. F., Jr.; Buck, M. M.; Post, L. S.;
Cipolla, G. G.; McClelland, C. A.; Mundt, J. A.; Schmitz, H. H. PCT Int.
Appl. WO 9736,497, Oct 9, 1997; Chem. Abstr. 1997, 127, 330680.
(36) Hemingway, R. W. In ref 1t, pp 265-283.
(37) Prieur, C.; Rigaud, J.; Cheynier, V.; Moutounet, M. Phytochemistry
1994, 36, 781.
(38) (a) Foo, L. Y.; Porter, L. J. J. Chem. Soc., Perkin Trans. 1 1983,
1535. (b) Ferreira, D.; Steynberg, J. P.; Burger, J. F. W.; Bezuidenhoudt,
B. C. B. in ref 1x, pp 255-295.
(26) Miyazawa, T.; Nakagawa, K.; Kudo, M.; Muraishi, K.; Someya,
K. J. Agric. Food Chem. 1999, 47, 1083 and references quoted therein.
(27) Porter, L. J. in ref 1k, pp 21-62.
(39) (a) Laks, P. E. In ref 1t, pp 249-263. (b) de Freitas, V. A. P.;
Glories, Y.; Laguerre, M. J. Agric. Food Chem. 1998, 46, 376.
(40) (a) Fletcher, A.; Porter, L. J.; Haslam, E.; Gupta, R. K. J. Chem.
Soc., Perkin Trans. 1 1977, 1628. (b) Botha, J. J.; Ferreira, D.; Roux, D.
G. J. Chem. Soc., Perkin Trans. 1 1981, 1235. (c) Roux, D. G.; Ferreira,
D. Fortschr. Chem. Org. Naturst. 1982, 47-76. (d) Delcour, J. A.; Ferreira,
D.; Roux, D. G. J. Chem. Soc., Perkin Trans. 1 1983, 1711. (e) Fonknechten,
G.; Moll, M.; Cagniant, D.; Kirsch, G.; Muller, J. F. J. Inst. Brewing 1983,
89, 424. (f) Steenkamp, J. A.; Ferreira, D.; Roux, D. G.; Hull, W. E. J.
Chem. Soc., Perkin Trans. 1 1983, 23. (g) Delcour, J. A.; Vercruysse, S.
A. R. J. Inst. Brew. 1986, 92, 244. (h) Delcour, J. A. Farm. Tijdschr. Belg.
1989, 66, 7.
(28) (a) Jacques, D.; Opie, C. T.; Porter, L. J.; Haslam, E. J. Chem. Soc.,
Perkin Trans. 1 1977, 1637. (b) Platt, R. V.; Opie, C. T.; Haslam, E.
Phytochemistry 1984, 23, 2211.
(29) Hemingway, R. W.; Foo, L. Y. J. Chem. Soc., Chem. Commun.
1983, 1035.
(30) (a) Creasy, L. L.; Swain, T. Nature 1965, 208, 151. (b) Stafford,
H. A. Phytochemistry 1983, 22, 2643.
(31) (a) Thompson, R. S.; Jacques, D.; Haslam, E.; Tanner, R. J. N. J.
Chem. Soc., Perkin Trans. 1 1972, 1387. (b) Hemingway, R.; Foo, L.; Porter,
L. J. Chem. Soc., Perkin Trans. 1 1982, 1209. (c) Rigaud, J.; Perez-Ilzarbe,
J.; Ricardo da Silva, J. M.; Cheynier, V. J. Chromatogr. 1991, 540, 401.
(d) McGraw, G. W.; Steynberg, J. P.; Hemingway, R. W. Tetrahedron Lett.
1993, 34, 987.
(41) Steynberg, P. J.; Nel, R. J. J.; van Rensburg, H.; Bezuidenhoudt,
B. C. B.; Ferreira, D. Tetrahedron 1998, 54, 8153.
(32) See Supporting Information for some details.
(33) Rigaud, J.; Escribano-Bailon, M. T.; Prieur, C.; Souquet, J.-M.;
Cheynier, V. J. Chromatogr. 1993, 654, 255.
(42) (a) Swain, T. Chem. Ind. 1954, 1144. (b) Porter, L. J.; Foo, L. Y.
Phytochemistry 1982, 21, 2947.
(43) Weinges, K.; Seiler, D. Liebigs Ann. Chem. 1968, 714, 193.

12076 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Tu¨ckmantel et al.
interesting approach has been reported in which the 8-bromo
derivative of 3-O-benzyl-5,7,3′,4′-tetra-O-methylcatechin was
subjected to halogen-lithium exchange and reacted with an
O-methylated 4-ketone,44 thus ensuring complete regiocontrol,
but the choice of methyl as O-blocking group in this case
precludes the final obtention of the free dimer. The remaining
published work has made use of the above-described electro-
philic substitution process with inclusion of phenol protecting
groups on one or both of the reaction partners.45,46 In fact, all
of the key reactionssprotection, activation of the 4-position,
interflavan bond formation, and deprotectionshave been dem-
onstrated in the literature, as will be discussed below. To the
best of our knowledge, however, none of this work has resulted
in the synthesis of nonprotected, natural proanthocyanidins from
stereochemically pure, readily available precursors.46 We were
attracted to this approach by the ready availability of (-)-
epicatechin and, even more so, (+)-catechin as starting materials,
and by the prospect of being able to perform the foreseeable
product separations after the condensation step on nonpolar
intermediates which are more comfortable to handle for the
typical organic chemist than the polar, sensitive final products.
Scheme 1
Results and Discussion
Phenol Protection. Our previous experience with the syn-
thesis of polar organic compounds, such as amino acids and
inositol phosphates, suggested the use of benzyl as the phenol
protecting group, which can be removed in the final step by a
hydrogenolysis reaction that is usually clean and efficient. This
choice is corroborated by a literature report indicating that
several other commonly used phenol protecting groups (meth-
oxymethyl, benzyloxymethyl, tosyl, benzenesulfonyl) could not
be removed from the final products in a similar scheme.45a
Whereas the phenolic hydroxyls of catechin are readily O-
methylated (Me2SO4, K2CO3, acetone, reflux),47 the analogous
alkylation with benzyl bromide was first reported in the literature
to give only traces of the desired product.43 Others have
subsequently improved the yield to 45% using modified
conditions (BnBr, K2CO3, DMF45c or NaH, BnBr, DMF48)
(Scheme 1). As in the related case of phloroglucinol,45a
C-benzylation competes with O-benzylation,43 and under dif-
ferent conditions, mono- and di-C-benzyl derivatives with free
hydroxyl groups have been prepared from catechin by direct
alkylation with benzyl bromide.49 In view of the large amount
of byproducts, we considered it prudent to let a rough separation
by column chromatography precede the crystallization step, and
we have consistently obtained yields in the 20% range, and
occasionally somewhat higher, of 5,7,3′,4′-tetra-O-benzylcat-
echin (6), which was at least 97% pure by HPLC.50 While the
percentage of yield is far from impressive, the starting material
is available at a moderate price, and 35 g batches of 6 are readily
prepared in a 2 L flask.
The analogous benzylation of epicatechin (2) was less
successful, giving a product of lower purity even after recrys-
tallization. Since epicatechin is, in addition, considerably more
expensive than catechin, it was decided to start from catechin
and to invert the stereochemistry at C-3 subsequently.
Catechin is known to undergo quite readily a base-catalyzed
epimerization at C-2 to form ent-epicatechin through reversible
opening of ring C via a B-ring quinone methide intermediate.38a
No diastereomeric material was seen by NMR in our crystallized
tetra-O-benzylcatechin. We went farther and also checked for
the presence of the enantiomer by forming diastereomeric esters
with both Mosher acid chlorides (see the Supporting Informa-
tion). While both esters exhibited sharp and distinct signals in
(44) Weinges, K.; Perner, J.; Marx, H.-D. Chem. Ber. 1970, 103, 2344.
(45) (a) Kawamoto, H.; Nakatsubo, F.; Murakami, K. J. Wood Chem.
Technol. 1989, 9, 35. (b) Kawamoto, H.; Nakatsubo, F.; Murakami, K. J.
Wood Chem. Technol. 1990, 10, 59. (c) Kawamoto, H.; Nakatsubo, F.;
Murakami, K. Mokuzai Gakkaishi 1991, 37, 488; Chem. Abstr. 1991, 115,
279643y. (d) Kawamoto, H.; Nakatsubo, F.; Murakami, K. Mokuzai
Gakkaishi 1991, 37, 741; Chem. Abstr. 1992, 117, 69641m.
(46) A partially racemic flavandiol building block analogous to 19 but
with 2,3-trans stereochemistry has been obtained by benzylation of (+)-
taxifolin followed by NaBH4 reduction, and condensed with 6 and 14:
Pierre, M.-C.; Che`ze, C.; Vercauteren, J. Tetrahedron Lett. 1997, 38, 5639.
The partial racemization is a consequence of reversible O-C2 cleavage
with formation of an unsaturated system that may be a quinone methide as
assumed by these authors, but which could also exist as an enone tautomer.
It should be noted that catechin does not epimerize under these benzylation
conditions (ref 45c), indicating that some mode of stabilization of the
reactive intermediate is of importance here. Also, the concomitant inversion
at C-3 together with that at C-2 necessitates that C-3 becomes olefinic at
some point during the course of the reaction, if not during the ring cleavage
step itself, then in a preceding or subsequent enolization event.
1
their 19F and H NMR spectra, no cross-contamination could
be detected. We thus conclude that our starting material is of
99% ee or better. It appears that quinone methide formation is
not important under aprotic conditions and/or in the presence
of a sufficiently reactive electrophile.
We have repeatedly made use in this investigation of the
8-bromo derivative 7 which is readily prepared in 99% regio-
selectivity (1H NMR) and high yield by reacting 6 with 1 equiv
of recrystallized NBS49 in CH2Cl2 at low temperature. This
(47) (a) Kostanecki, St. v.; Tambor, J. Ber. Dtsch. Chem. Ges. 1902,
35, 1867. (b) Mehta, P. P.; Whalley, W. B. J. Chem. Soc. 1963, 5327.
(48) Miura, S.; Midorikawa, T.; Awata, N. Radioisotopes 1983, 32, 225;
Chem. Abstr. 1983, 99, 158183r.
(49) Ballenegger, M. E.; Rimbault, C. G.; Albert, A. I.; Weith, A. J.;
Courbat, P.; Tyson, R. G.; Palmer, D. R.; Thompson, D. G. Eur. Pat.
0096007, July 29, 1987; Chem. Abstr. 1984, 100, 209512w.
(50) The tetrabenzyl ethers of 6- and 8-benzyl- and 6,8-dibenzylcatechin
were identified in the mother liquor, and tribenzyl ethers with and without
a C-benzyl group were isolated from more polar fractions eluted after the
tetrabenzyl ethers: Kozikowski, A. P.; Tu¨ckmantel, W. Unpublished.

Studies in Polyphenol Chemistry and BioactiVity
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 12077
method is superior to the one using pyridinium tribromide51,52
which in our hands gave impure reaction mixtures containing
small amounts of compounds in which benzyl groups appear
to have been oxidized to benzoate.
Scheme 2
Initial Attempts at an Inherently Regioselective Approach.
Our initial synthetic plan did not employ the well-documented
Lewis acid-mediated aromatic alkylation by 4-hydroxyflavans,
the regioselectivity of which could not be predicted with
certainty, but instead attempted to impose regioselectivity by
using a 6- or 8-lithiated protected flavan as the nucleophile, as
the Weinges group had done before. The electrophilic compo-
nent could then, of course, not be an alcohol. Weinges et al.
used a ketone but then an additional hydroxyl group originated
in position 4 which had subsequently to be removed. We
envisaged instead the use of the epoxide 9 (or a cyclic ester of
an analogous diol) (Scheme 1). Epoxides similar to 9 but having
a lower degree of oxygenation in their A ring have been prepared
or invoked as reaction intermediates.53 Olefin 8 could serve as
a precursor to 9. The compound is known but is formed as a
racemate on treatment of the mesylate of 6 with base.45c During
first attempts at inverting the C-3 stereochemistry of 7, we
subjected the compound to Mitsunobu conditions54 and obtained
the olefin 8 in 87% yield. Omission of the acid resulted in a
70% yield of 8 which exhibited an [R]D of -85°. To our
disappointment, however, a subsequent preparation gave a
product with an [R]D of only -61°, a clear indication that
racemization still occurred to a significant degree even under
these very mild and essentially neutral conditions. This finding
disqualified 8 as an intermediate in a controlled procyanidin
synthesis and prompted us to return to the conventional
electrophilic substitution approach. In hindsight, the lability of
8 is not really surprising since a highly stabilized, achiral
carbenium ion 11 may form under acid or surface catalysis.
Equally, the envisaged epoxide intermediate 9 might well prove
to be unattainable on any route55 because of a collusion toward
its destabilization of epoxide ring strain and carbenium ion
stabilization by the three oxygen substituents on ring A.
room temperature) or to N-methylmorpholine-N-oxide/catalytic
tetrapropylammonium perruthenate (MS 4 Å, CH2Cl2, room
temperature). On the other hand, reaction of 6 or 7 with the
Dess-Martin periodinane57 in moist CH2Cl258 gave reproducibly
yields of 90% or better of the corresponding ketones 12, 13
(Scheme 2). The halogen-free ketone 12 crystallizes readily from
CHCl3/Et2O and can be stored in crystalline form at -20 °C
for extended time periods.
The reduction of 12 has previously been performed with
NaBD4 (CHCl3/MeOH, room temperature) to give a 67:33 ratio
of epicatechin and catechin stereoisomers 6 and 14, respec-
tively.48 We found that DIBAL-H (applied to 13) gives an
improved but still unsatisfactory stereoselection (Table 1). A
bulky hydride source could be expected to exhibit a better
selectivity toward attack from the less hindered â face. Gratify-
ingly, the use of lithium tri-sec-butylborohydride (L-Selectride)
in THF resulted in the exclusive (g200:1, 1H NMR) formation
of the desired 3R-isomer. The yield, however, was only
moderate. We noticed that the thin layer chromatogram of the
reaction mixture did not further change upon addition of an
excess of the reagent, with some starting material apparently
remaining unreacted. This suggested that a fraction of the
starting material underwent enolization rather than reduction,
with the enolate being hydrolyzed back to the ketone by silica
gel. Upon alkaline H2O2 workup, however, the enolate was
destroyed, and the 3R-alcohol 15 was obtained as the sole
product.59 The situation is reminiscent of the addition of
Grignard and organolithium reagents to ketones, in which case
enolization can usually be suppressed or at least diminished by
transmetalation to an organocerium reagent with anhydrous
CeCl3.60 The following experiments were performed on the
bromine-free ketone 12. When CeCl3 (1 equiv in relation to
L-Selectride) was included in the reaction mixture, an improved
yield of the 3R-alcohol 14 was obtained without sacrifice in
stereoselectivity. Unfortunately, when the amount of reagent
was reduced from 3 to 1.5 equiv in a larger scale experiment,
Alcohol Inversion. Since the Mitsunobu reaction failed to
effect the required inversion at C-3, attention was turned to an
oxidation-reduction sequence. The low-yielding oxidation of
6 with DMSO/Ac2O has been described48 but gave at best traces
of the ketone in our hands. Others have reported that the
oxidation of hepta-O-methylprocyanidin A2 gave a moderate
yield of a monoketone under modified Oppenauer conditions
(fluorenone, KOCMe3), but they obtained complex mixtures
with CrO3 and with activated DMSO.56 We recovered starting
material upon exposure of 7 to pyridinium dichromate (DMF,
(51) Engel, D. W.; Hattingh, M.; Hundt, H. K. L.; Roux, D. G. J. Chem.
Soc., Chem. Commun. 1978, 695.
(52) Hundt, H. K. L.; Roux, D. G. J. Chem. Soc., Perkin Trans. 1 1981,
1227.
(53) Clark-Lewis, J. W.; McGarry, E. J.; Ilsley, A. H. Aust. J. Chem.
1974, 27, 865 and references quoted therein.
(54) Martin, S. F.; Dodge, J. A. Tetrahedron Lett. 1991, 32, 3017.
(55) Dimethyldioxirane, the reagent most likely to epoxidize 8 without
subsequent epoxide opening, can be expected to attack predominantly from
the less hindered face, resulting in 3â,4â, i.e., catechin, stereochemistry.
(56) Nonaka, G.; Morimoto, S.; Kinjo, J.; Nohara, T.; Nishioka, I. Chem.
Pharm. Bull. 1987, 35, 149.
(57) (a) Dess, D. B.; Martin, J. C. J. Org. Chem. 1983, 48, 4155. (b)
Ireland, R. E.; Liu, L. J. Org. Chem. 1993, 58, 2899.
(58) Meyer, S. D.; Schreiber, S. L. J. Org. Chem. 1994, 59, 7549.
(59) For the preparation of 15 on this route or alternatively by bromination
of 14 with NBS, see the Supporting Information.
(60) Imamoto, T.; Kusumoto, T.; Tawarayama, Y.; Sugiura, Y.; Mita,
T.; Hatanaka, Y.; Yokoyama, M. J. Org. Chem. 1984, 49, 3904.

12078 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Tu¨ckmantel et al.
Table 1. Effect of Additives on the Reduction of Ketones 12 and 13a
scale
(mmol)
crude yield
(%)
recrystd yield
(%)
compd
reagent/equiv
additive/equiv
none
none
ZnCl2/1.5
ZnCl2/3
CeCl3/3.1
CeCl3/1.5
none
cis/transb
ketone/alcoholc
13
DIBAH/3
0.06
0.82
0.09
0.08
0.07
4.5
0.06
0.07
0.13
0.10
0.09
77
88:12
g200:1
94:6
83d
55
80
96
77
LiBH(s-Bu)3/3
LiBH(s-Bu)3/3
LiBH(s-Bu)3/3
LiBH(s-Bu)3/3
LiBH(s-Bu)3/1.5
LiBHEt3/3
LiBH(s-Bu)3/1.5
LiBH(s-Bu)3/1.5
LiBH(s-Bu)3/1.5
LiBH(s-Bu)3/1.5
LiBH(s-Bu)3/1.3
LiBH(s-Bu)3/1.5
47
10
53:47
g200:1
g200:1
83:17
g200:1
g200:1
g200:1
g200:1
g200:1
200:1
56
68
none
19:81
19:81
5:95
CeCl3/1.5
CeCl3/1.5, LiBr/6
LiBr/6
LiBr/5.2
n-Bu4NCl/3
8:92
81
0.11
81
a All reactions in THF (DIBAH reduction in CH2Cl2/hexane) at -78 °C. Borohydride reductions were followed by an oxidative workup with
1
1
NaOH/H2O2. b By H NMR. c For entries in this column, the oxidative workup was omitted. d Besides 9% of unreacted 13 (by H NMR).
the yield returned to near its original value. One factor that may
be involved in this decrease is the presence of residual water in
supposedly anhydrous CeCl3 which may destroy part of the
reducing agent. We also noticed that all or part of the CeCl3
remained undissolved throughout the course of the reaction, thus
challenging the assumption of the formation of a new reagent
on which this experiment was based in the first place. Substitut-
ing soluble ZnCl2 for CeCl3, a steep improvement in yield
resulted as the amount of additive was increased from 0.5 to 1
equiv relative to the hydride, but the loss in stereoselectivity
was equally pronounced. An influence of additives (ZnI2,
MgBr2, Ti(OiPr)4, or TiCl(OiPr)3) on the stereoselectivity of an
L-Selectride reduction has previously been observed.61 We
eventually also replaced L-Selectride by the less bulky lithium
triethylborohydride, reasoning that lowered steric hindrance
might favor reduction over enolization, and observed indeed a
marginally improved yield, but again accompanied by a serious
loss in stereoselectivity.
At this point, we wondered whether CeCl3 could be solubi-
lized in THF through addition of lithium halides in the same
manner as observed for some transition metal halides. In this
series of experiments, the oxidative workup was omitted,
allowing the ratio of alcohol and unreacted ketone to be
determined by NMR, although the presence of boron impurities
in the crude products prevented the determination of yields.
When 12 was reduced with 1.5 equiv of L-Selectride, a 4:1 ratio
of alcohol and starting ketone resulted. Inclusion of 1.5 equiv
of CeCl3 and 6 equiv of anhydrous LiBr improved this ratio to
20:1. By visual appearance, we nevertheless observed that no
or little CeCl3 had dissolved. Concluding that at least some of
the effect was due to LiBr rather than CeCl3, we used 6 equiv
of LiBr alone as the additive and obtained an alcohol/ketone
ratio of 11:1. This protocol was scaled up including an oxidative
workup and afforded reproducibly 5,7,3′,4′-tetra-O-benzylepi-
catechin (14) in a reasonable 81-82% yield on an up to 77
mmol scale. That no racemization had occurred at the ketone
stage was demonstrated by the preparation of both Mosher esters
from 14 which showed no cross-contamination, and we estimate
the chemical and optical purity of our material to be at least
98%.
that the addition of isopropylmagnesium bromide to an aldehyde
susceptible to enolization was greatly improved upon adding
an equimolar amount (with respect to the Grignard reagent) of
hexaethylguanidinium bromide.62 These authors quoted older
work63 that reported an improvement in the ratio of direct
addition vs â-hydride transfer in the reaction of diisopropyl
ketone with n-PrMgBr. Among the successful additives were
LiBr and LiClO4 as well as several tetra-n-butylammonium salts,
most notably the chloride. When 12 was reduced with 1.5 equiv
of L-Selectride in the presence of 3 equiv of dried n-Bu4NCl,
14 was formed in 81% yield and a stereoselectivity of 200:1.
This result demonstrates that the Lewis acidity of LiBr is not
the deciding factor in modifying the outcome of the reaction,
but what exactly the salt additive does is not clear. A recent
investigation64 concludes that the much-used ionic reaction
medium LiClO4/ether functions as a weak Lewis acid in its
acceleration of Diels-Alder reactions, but this finding does not
have to be applicable to other types of reactions or to different
ionic solutions. From the preparative point of view, it appears
that the use of nontransmetalating salt additives in hydride
reductions would deserve further investigation.
Oxidation at C-4. Three reagents have been employed in
the literature to oxidize O-alkylated catechins and epicatechins,
namely K2S2O8/catalytic CuSO4,65 lead(IV) acetate,45c,66 and
DDQ.67 Using the first-listed reagent, we observed that the
starting material was not or little soluble in the reaction medium,
aqueous CH3CN, and that a complex mixture of unidentified
products was eventually formed. Reaction of 15 or its 3-O-acetyl
derivative with lead tetraacetate gave, after acetate saponification
(NaOMe, MeOH/THF), meager yields of impure 3,4-diol. On
the other hand, a much cleaner reaction was observed between
14 and DDQ in CHCl3/MeOH. Over several hours at room
temperature, a polar product emerged while a TLC spot with
the same mobility as the starting material persisted even after
9 h. Isolation of both components gave a 21% yield of the
(62) Schlama, T.; Baati, R.; Gouverneur, V.; Valleix, A.; Falck, J. R.;
Mioskowski, C. Angew. Chem., Int. Ed. Engl. 1998, 37, 2085.
(63) Chastrette, M.; Amouroux, R. Bull. Soc. Chim. Fr. 1970, 4348.
(64) Springer, G.; Elam, C.; Edwards, A.; Bowe, C.; Boyles, D.;
Bartmess, J.; Chandler, M.; West, K.; Williams, J.; Green, J.; Pagni, R.
M.; Kabalka, J. W. J. Org. Chem. 1999, 64, 2202.
We are not aware of previous observations of an effect of
nontransmetalating salt additives on the product distribution in
borohydride reductions of ketones but there are reports in the
literature of such effects on the related addition of Grignard
reagents to aldehydes and ketones. It has recently been shown
(65) (a) Bhatt, M. V.; Perumal, P. T. Tetrahedron Lett. 1981, 22, 2605.
(b) Mouton, C. H. L.; Steenkamp, J. A.; Young, D. A.; Bezuidenhoudt, B.
C. B.; Ferreira, D. Tetrahedron 1990, 46, 6885.
(66) (a) Bokadia, M. M.; Brown, B. R.; Cummings, W. J. Chem. Soc.
1960, 3308. (b) Betts, M. J.; Brown, B. R.; Shaw, M. R. J. Chem. Soc. (C)
1969, 1178.
(67) (a) Steenkamp, J. A.; Ferreira, D.; Roux, D. G. Tetrahedron Lett.
1985, 26, 3045. (b) Steenkamp, J. A.; Mouton, C. H. L.; Ferreira, D.
Tetrahedron 1991, 47, 6705.
(61) Tanis, S. P.; Chuang, Y.-H.; Head, D. B. J. Org. Chem. 1988, 53,
4929.

Studies in Polyphenol Chemistry and BioactiVity
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 12079
4-methoxy derivative 16a, which had the same chromatographic
mobility as the starting material under a variety of conditions,
and 29% of a single, polar 2,4-dimethoxy derivative 17a of
unknown stereochemistry. Since the presence of bis-oxidized
material indicated that the reaction should be terminated
promptly after consumption of the starting material, and our
inability to monitor the initial oxidation of 14 to 16a rendered
this task difficult, 2-propanol was substituted for MeOH in the
hope that this would result in a lower polarity of the 4-alkoxy
derivative. This measure fulfilled its purpose, and after comple-
tion of the reaction, an 87% crude yield of the oxidation product
was obtained. This material, however, proved very impure. An
analogous result was obtained using the 3-acetate as starting
material. In this case, we obtained small amounts of pure or
enriched byproducts by preparative HPLC, and identified these
as the 2,4-diisopropoxy-, the 2-isopropoxy-, and the 4-ethoxy
derivatives, 17c, 18c, and 16b, respectively. The latter undoubt-
edly arose from the ethanol present in chloroform as a stabilizer,
and CHCl3 was henceforth replaced with CH2Cl2. Coming to
the conclusion that reducing the polarity of the alcohol nucleo-
phile had made things worse by rendering the dialkoxylation
product, the formation of a small amount of which apparently
cannot be avoided even if the reaction is stopped in a timely
manner, barely separable from the desired product, the opposite
strategy was pursued next. Exposure of 14 to DDQ in THF/
water 10:1 (THF was used here to obtain a homogeneous
solution) gave 33% of the 3,4-diol 16d together with 44% of
recovered 14, both somewhat impure. Better results were
obtained using ethylene glycol in CH2Cl268 as the nucleophile,
giving a 52% isolated yield of 19 on a 20 mmol scale. All
impurities detected by us69 including the presumed 2-alkoxylated
regioisomer are readily removed by column chromatography,
and the product itself could be crystallized and is in this form
stable to extended storage at -20 °C.70
Scheme 3
We leave the question of stereochemistry at C-4 open since
J3,4 is not a realiable indicator of stereochemistry in the
epicatechin series (see below). Specifically, the small, unre-
solved coupling between 2-H and 3-H requires that these two
protons be not far from orthogonal, leaving for the dihedral
angles between 3-H and 4-H in the 4R and 4â stereoisomers
relatively small but still overlapping ranges. Mechanistic argu-
ments favor 4â-stereochemistry as discussed below for the
epicatechin 4,8-dimer.
Condensation Step: Reaction of 14 with 19. A protocol
exists for the electrophilic alkylation of O-protected polyphenols
with flavan-3,4-diols45 and was found applicable to 19 as well
(Scheme 3). The nucleophilic component 14 was employed in
a 4-fold excess to decrease the extent of higher oligomer
formation that result from repeated attack of the electrophile
on already formed smaller oligomers. Nevertheless, when the
reactants were exposed to 2 equiv of TiCl4 in a solvent mixture
of CH2Cl2 and THF at room temperature, a mixture of a single
dimer 20, two trimers, and a tetramer were obtained in yields
of 53, 22, 4.5, and 1.5%, respectively. The trimers and the
tetramer are still under study and will be the subjects of a
separate publication. The formation of single stereoisomers in
the reaction of epicatechin-derived flavan-3,4-diols has been
commonly observed in the literature, and 4â (i.e., 3,4-trans)
stereochemistry has generally been assigned to these products.71
This makes sense from a mechanistic point of view if backside
shielding of the 4-carbenium ion by the 3R-hydroxyl is effective
during the alkylation reaction, and also represents a reaction
on the less congested face of the pyran ring opposite from the
2-aryl substituent. Characteristically, in the catechin series where
these two effects are opposed to each other, mixtures of 4R
and 4â stereoisomers may be obtained.45c We did, however,
feel disconcerted by the fact that the assignment of interflavan
stereochemistry in the literature is based on 1H NMR coupling
constants and chiroptical methods,72,73 both techniques of limited
reliability in conformationally flexible systems. Only if the
magnitude of J3,4 demonstrates a (nearly) trans-diaxial relation-
(68) We initially overlooked the fact that ethylene glycol is poorly soluble
in CH2Cl2, but the reaction works fine anyway.
(69) HPLC separation of the column forerun resulted in isolation or
enrichment of residual starting material and of byproducts to which the
following structures have tentatively been assigned: the 2-alkoxy derivative;
a product containing two epicatechin moieties linked by an ethylenedioxy
group, most likely both in their 4-positions; and a product of 2-fold oxidation
in positions 2,4 or 4,4 which are spanned by a single ethylenedioxy group.
(70) We also changed the workup protocol, which aims at reducing or
otherwise destroying residual DDQ and at neutralizing DDQH2, so as to
permit the isolation of prepurified product by filtration over a short silica
gel column that would be readily penetrated by the large amounts of DDQ
and DDQH2 present. The original procedure used an excess of NaBH4, a
rather hazardous operation on larger reaction scales. We were able to replace
this scavenging agent with Amberlite IRA-400 (OH- form) but remained
dissatisfied because of the large amount of resin needed. Eventually, it was
found that a slight excess of 4-(dimethylamino)pyridine (DMAP) serves
well for this purpose. The formation of solid, dark-colored molecular
complexes between DDQ and the three isomeric aminopyridines has been
reported: Tosi, G.; Bruni, P.; Cardellini, L. Gazz. Chim. Ital. 1984, 114,
125.
(71) According to ref 1t (p 92), only a single compound, a natural product,
with 2,3-cis-3,4-cis stereochemistry is known.
(72) For a characteristic example, see: Kolodziej, H. Phytochemistry
1986, 25, 1209.
(73) Reliance on circular dichroism as a source of errors is documented
in: Hundt, A. F.; Burger, J. F. W.; Steynberg, J. P.; Steenkamp, J. A.;
Ferreira, D. Tetrahedron Lett. 1990, 31, 5073.

12080 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Tu¨ckmantel et al.
ship between the coupling protons, as is often the case for
catechin derivatives, can a safe conclusion regarding the
interflavan stereochemistry be drawn. To complicate the situ-
ation, it has been shown that catechin units in proanthocyanidin
oligomers may adopt an alternative C-ring conformation that
results in small, unresolved J2,3 and J3,4 despite trans,trans
disposition of the substituents.74 For epicatechin derivatives, 3-H
and 4-H have an equatorial-axial or equatorial-equatorial
relationship, and these two alternatives cannot be distinguished
with certainty by coupling constants. Under these circumstances,
an X-ray structure analysis of 20 or a derivative thereof would
be highly desirable. We were, unfortunately, unable to crystallize
20 or several of its derivatives,75 and we are not aware of a
crystal structure analysis for any dimeric proanthocyanidin76
other than rigidified so-called A-type proanthocyanidins.77 As
a consequence, a traditional oxidative degradation approach has
been pursued and has resulted in the transformation of an
O-alkylated derivative of 21 into (R)-2,4-diphenylbutyric acid
in which the carboxyl group is derived from the A2 ring, and
the phenyl groups are derived from the A1 and B1 rings. This
work confirms the hitherto hypothetical â stereochemistry of
the interflavan linkage in 21 and will be published in due course.
The 4,8-regiochemistry of 20 follows from that of its depro-
tection product 21 (see below). Preferred attack of an electro-
phile at the 8- vs the 6-position of a tetra-O-alkylcatechin has
previously been demonstrated by X-ray structure analysis of
the reaction product, 8-bromo-5,7,3′,4′-tetra-O-methylcatechin.51
The inference of regiochemistry from the small chemical shift
differences between 6-H and 8-H in 8- and 6-substituted com-
pounds52,72,78 is, to the contrary, unreliable, and the assignment
of resonances to 6-H and 8-H has been subjected to revision.79,80
As there is little electronic difference between positions 6 and
8, their differential reactivity is apparently based on the lower
degree of steric hindrance exerted by the tied-back alkoxy group
of which the pyran oxygen is a part, compared to the flexible
alkoxy groups at C-5 and C-7. It therefore stands to reason that
this effect is even more pronounced in the O-benzyl series
compared with the O-methyl series. The alkylation of tetra-O-
methylcatechin with o- or p-hydroxybenzyl alcohol (smaller
electrophiles than 19) at elevated temperatures resulted, how-
ever, in a mixture of 6- and 8-substituted products.81 Others
have previously assigned 4,8-regiochemistry to condensation
products derived from 6 and 14.45c,46 While the exclusive
formation of the 4,8-dimer in this reaction is a welcome result
as far as the purification of the product is concerned, it also
raises the question of how to prepare the 4,6-dimer. In a first
unsuccessful attempt, 19 was reacted with 15, which is prevented
from reacting at the 8-position by its bromo substituent, but no
dimer was isolated.82 This result conforms to the preceding
reasoning but may also reflect the inductive deactivation of the
A ring by the Br substituent. That the situation may not be
entirely so simple is pointed out by the fact that two trimers
were obtained in the condensation reaction, one of which
therefore must violate the provided stereochemical or regio-
chemical rationale, or both; unless, of course, these compounds
turn out to be isolatable rotamers.
The issue of rotational isomerism around the interflavan bond
(or around several interflavan bonds in the case of higher
oligomers, with an associated exponential increase in the number
of possible rotamers)31a,40a,b,83 introduces a significant complica-
tion in working with these compounds. At present, there is no
indication (e.g., by HPLC) that 20 or its derivatives exist in the
form of rotamers isolatable on the laboratory time scale (if the
two trimers are not such a case). On the NMR time scale,
however, two sets of sharp signals for the two rotamers can be
distinguished. The 1H NMR spectra at 300 MHz of O-protected
dimers usually remain interpretable (outside their benzyl regions)
if a COSY experiment is enlisted for help, but for higher
oligomers, product characterization will predominantly have to
depend on HPLC and mass spectrometry.
The free epicatechin 4â,8-dimer (21, procyanidin B2)31a,38a,84,85
was uneventfully prepared from its benzylated precursor by
hydrogenolysis. Only one set each of 1H and 13C NMR signals
is observed, but the 1H signals are close to the coalescence point
at room temperature, and part of the 13C signals are strongly
broadened as well. The 4,8-regiochemistry of natural procya-
nidin B2 has recently been rigorously proven by 1H-13C NMR
correlation (HMBC).80 The decaacetate 22 derived from our
1
material exhibits H and 13C NMR spectra which are in good
agreement with those reported by these authors for the decaac-
etate of the natural product. Selected 1H NMR signals published
for 22 derived from natural material85a,86 agree also with ours,
and appropriate 13C NMR signals of our product agree with
selected data published earlier for the decaacetate of natural87
and synthetic material.38a
Galloylation of the Dimer and Biological Activity of the
Bisgallate. A recent report21 disclosing the anticancer activity
of epigallocatechin 3-gallate (5) in a system in which epigal-
(74) Balas, L.; Vercauteren, J.; Laguerre, M. Magn. Reson. Chem. 1995,
33, 85.
(81) McGraw, G. W.; Hemingway, R. W. J. Chem. Soc., Perkin Trans.
1 1982, 973. The authors assigned product regiochemistries solely based
on chemical shifts, rather than correlating their 8-substituted products with
8-bromotetra-O-methylcatechin the X-ray structure of which had been
reported by that time (ref 51).
(82) Unpublished experiment with A. Hoepping. The related use of
6-iodocatechin as a nucleophilic component in coupling reactions with the
aim of imposing regioselectivity has been described: Young, D. A.; Cronje´,
A.; Botes, A. L.; Ferreira, D.; Roux, D. J. J. Chem. Soc., Perkin Trans. 1
1985, 2521. Other authors, however, have subsequently challenged both
the regioisomeric nature of the starting material as well as its direct
involvement in the interflavan bond forming process: ref 79b.
(83) (a) Bergmann, W. R.; Barkley, M. D.; Hemingway, R. W.; Mattice,
W. L. J. Am. Chem. Soc. 1987, 109, 6614. (b) Hatano, T.; Hemingway, R.
W. J. Chem. Soc., Perkin Trans. 2 1997, 1035.
(84) Weinges, K.; Kaltenha¨user, W.; Marx, H.-D.; Nader, E.; Nader, F.;
Perner, J.; Seiler, D. Liebigs Ann. Chem. 1968, 711, 184.
(85) (a) Weinges, K.; Go¨ritz, K.; Nader, F. Liebigs Ann. Chem. 1968,
715, 164. (b) Hsu, F.-L.; Chen, H.-F. Planta Med. 1993, 59, 405. (c) The
enantiomer has also been isolated: delle Monache, F.; Ferrari, F.; Marini
Betto`lo, G. B. Gazz. Chim. Ital. 1971, 101, 387.
(75) A variety of esters and urethanes, as well as two monobromides,
the 3,3-dideoxy derivative, and the diketone, have been investigated by us.
One group of researchers reported crystallization of the peracetate (ref 34a)
but others described the compound as amorphous (ref 84).
(76) The free dimeric proanthocyanidin, fisetinidol-4,8-catechin (fise-
tinidol is the 5-deoxy analogue of catechin), has been crystallized, but an
X-ray diffraction analysis failed to yield its structure as a consequence of
conformational disorder: unpublished work quoted by R. W. Hemingway
in Polyphenols 96 (proceedings of the conference in Bordeaux, France, July
15-18, 1996); Vercauteren, J.; Che`ze, C.; Triaud, J., Eds.; Editions INRA,
Paris, 1998; pp 81-103.
(77) van Rooyen, P. H.; Redelinghuys, H. J. P. S. Afr. J. Chem. 1983,
36, 49.
(78) A summary of the truly bewildering sole reliance, until quite
recently, on questionable empirical techniques for regiochemical assignment
is given in: Balas, L.; Vercauteren, J. Magn. Reson. Chem. 1994, 32, 386.
(79) (a) Kiehlmann, E.; Tracey, A. S. Can. J. Chem. 1986, 64, 1998. (b)
Kiehlmann, E.; Lehto, N.; Cherniwchan, D. Can. J. Chem. 1988, 66, 2431.
(c) De Bruyne, T.; Pieters, L. A. C.; Dommisse, R. A.; Kolodziej, H.; Wray,
V.; Domke, T.; Vlietinck, A. J. Phytochemistry 1996 43, 265. (d)
Hemingway, R. W.; Tobiason, F. L.; McGraw, G. W.; Steynberg, J. P.
Magn. Reson. Chem. 1996, 34, 424.
(86) (a) Langhammer, L.; Rauwald, H.-W.; Schulze, G. Arch. Pharm.
(Weinheim) 1981, 114, 424. (b) Engelshowe, R. Planta Med. 1983, 49, 170.
(87) Porter, L. J.; Newman, R. H.; Foo, L. Y.; Wong, H. J. Chem. Soc.,
Perkin Trans. 1 1982, 1217.
(80) Khan, M. L.; Haslam, E.; Williamson, M. P. Magn. Reson. Chem.
1997, 35, 854.

Studies in Polyphenol Chemistry and BioactiVity
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 12081
Scheme 4
Compound 24 is a low-micromolar inhibitor of protein kinase
C91 and an inhibitor of ACE (angiotensin-converting enzyme).34e
It increases the lifespan of mice inoculated with sarcoma-180
and is more active in this model than 5 but less active than
several oligomeric ellagitannins.20d It also exhibits weak cyto-
toxicity against RPMI-7951 human melanoma cells while being
inactive against several other human cancer cell lines.20f
Recently, 24 has been shown to inhibit the growth of several
human breast cancer cell lines in a concentration range of 50-
100 mg/L,92 comparable to compound 5 which has generated
considerable attention because of its anticancer activity.
Conclusion and Perspectives
A convenient, partially optimized procedure has been devel-
oped for the preparation of the electrophilic building block 19
which permits the stereoselective assembly of epicatechin
oligomers in which all phenolic hydroxyl groups are protected
as benzyl ethers. The only dimer formed in the TiCl4-mediated
condensation reaction between 14 and 19 is the 4â,8-isomer.
Resubjection of smaller oligomers to the reaction conditions
should enable us to generate mixtures of larger oligomers, the
complexity of which is increased in a stepwise manner (all
products contain the common partial structure of the starting
oligomer) and thus hopefully will remain manageable with
present-day purification techniques. While the preparation of
these compounds and the rigorous elucidation of their structure
will require additional effort, a natural product exhibiting
significant activity against cancer cells has been obtained at this
early project stage by esterifying the alcoholic hydroxyl groups
of 20 with protected gallic acid, followed by hydrogenolysis.
locatechin (4) is inactive prompted us to synthesize the 3,3-
bisgallate 24 derived from 21, which has not previously been
obtained synthetically (Scheme 4). Tri-O-benzylgallic acid88 was
converted in situ into its chloride, and 20 was esterified with
this reagent in pyridine solution in the presence of DMAP,
furnishing the diester 23 in near-quantitative yield. Hydro-
genolysis of 23 over 20% Pd(OH)2/C gave the deprotected
bisgallate 24 in 90% yield as a hydrate without need for
chromatographic purification (5.3 equiv of water per bisgallate
molecule by combustion analysis of a batch lyophilized from
water). This compound has previously been isolated from natural
1
sources.34e,39b,89,90 Our H NMR spectrum, while not highly
characteristic because of line broadening, is in reasonable
agreement with that of ref 90a (which was acquired at 100
MHz), except for the omission without comment of the B1 and
B2 ring protons from the literature spectrum.
3-O-Galloylepicatechin is equally readily prepared from 14.
The bisgallate 24 has been isolated from grape seeds, wine,
rhubarb, and cocoa and thus is taken up with the human diet.
Acknowledgment. We are grateful to M&M Mars, Inc. for
funding of this project.
Supporting Information Available: Paragraphs on nomen-
clature and the occurrence of proanthocyanidins in human
foodstuffs; Experimental Section (PDF). This material is avail-
able free of charge via the Internet at http://pubs.acs.org.
(88) Cavallito, C. J.; Buck, J. S. J. Am. Chem. Soc. 1943, 65, 2140.
(89) Ricardo da Silva, J. M.; Rigaud, J.; Cheynier, V.; Cheminat, A.;
Moutounet, M. Phytochemistry 1991, 30, 1259.
JA993020D
(90) (a) Nonaka, G.; Miwa, N.; Nishioka, I. Phytochemistry 1982, 21,
429. (b) Nonaka, G.; Kawahara, O.; Nishioka, I. Chem. Pharm. Bull. 1983,
31, 3906. (c) Kashiwada, Y.; Nonaka, G.; Nishioka, I. Chem. Pharm. Bull.
1986, 34, 4083. (d) Kashiwada, Y.; Nonaka, G.; Nishioka, I. Chem. Pharm.
Bull. 1989, 37, 999. (e) Hashimoto, F.; Nonaka, G.; Nishioka, I. Chem.
Pharm. Bull. 1989, 37, 3255. (f) Hmamouchi, M.; Es-Safi, N.; Essassi, E.
M. Fitoterapia 1997, 68, 332. (g) de Freitas, V. A. P.; Glories, Y.;
Bourgeois, G.; Vitry, C. Phytochemistry 1998, 49, 1435. (h) The enantiomer
of 24 has also been isolated from natural sources: Geiss, F.; Heinrich, M.;
Hunkler, D.; Rimpler, H. Phytochemistry 1995, 39, 635.
(91) Kashiwada, Y.; Nonaka, G.; Nishioka, I.; Ballas, L. M.; Jiang, J.
B.; Janzen, W. P.; Lee, K.-H. Bioorg. Med. Chem. Lett. 1992, 2, 239.
(92) Romanczyk, L. J., Jr.; Kozikowski, A. P.; Tueckmantel, W.;
Lippman, M. E. PCT Int. Appl. WO 99/19,319, April 22, 1999. Briefly,
the observed cell growth (as percent of the growth of untreated cells) at
concentrations of 24 of 50 and 100 mg/L is 84 and 62% for MDA MB 231
cells, 33 and 8% for MDA MB 435 cells, 23 and 2% for MDA 231 cells,
and 55 and 7% for MCF-7 cells, respectively.