Syntheses of doubly linked proanthocyanidins using free flavan units as nucleophiles: Insight into the origin of the high regioselectivity of annulation

Article abstract of DOI:10.1039/c9ob01896d

A synthesis method of doubly linked flavan dimers is reported via the acid-promoted annulation reaction using nascent catechins, (+)-catechin or (-)-epicatechin, as a dianionic partner and an ethylenedioxy-bridged flavan as a dicationic partner. Procyanidins A1 and A2 were synthesized. On the high regioselectivity of the annulation reactions, model experiments and computational studies were carried out.

Full text of DOI:10.1039/c9ob01896d

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
COMMUNICATION
Syntheses of doubly linked proanthocyanidins
using free flavan units as nucleophiles: insight into
the origin of the high regioselectivity of
annulation†
Cite this: Org. Biomol. Chem., 2019,
7, 9129
1
Received 29th August 2019,
Accepted 27th September 2019
DOI: 10.1039/c9ob01896d
Vipul V. Betkekar, Mio Harachi, Keisuke Suzuki * and Ken Ohmori
*
rsc.li/obc
A synthesis method of doubly linked flavan dimers is reported via
Two notable features in substrates 3 and 4 are: (1) the dioxy
the acid-promoted annulation reaction using nascent catechins, derivative 3 is bromo-capped at its C8 position to tame the
4
(
+)-catechin or (−)-epicatechin, as a dianionic partner and an ethy- nucleophilicity for avoiding the potential self-reactions, and
lenedioxy-bridged flavan as a dicationic partner. Procyanidins A1 (2) the nucleophilic substrate 4 is selectively protected with the
5
and A2 were synthesized. On the high regioselectivity of the free C7-phenol, which was prepared by our de novo protocol.
annulation reactions, model experiments and computational
For expanding the scope of the approach, we decided to
study two substrate modifications. Firstly, we conceived of the
idea of using catechin and epicatechin as the commercially
available starting materials. Since the selective protection of
the phenols in those compounds seemed tedious, we envi-
saged to use them in their free forms, although the regio-
selectivity may be problematic, because the C6/C8 nucleophilic
sites in the A-ring have similar reactivities (blue and yellow
in I, Fig. 2).
Secondly, one could expect a higher nucleophilicity of I at
the A-ring moiety from the extra free phenol (red, Fig. 2) than
its protected congener II. Given the case, we came to a notion
that the C8-bromo capping as in the dioxy-flavan unit may not
be necessary.
studies were carried out.
Oligomeric proanthocyanidins (OPA) are widespread in nature,
1
featuring extreme molecular diversity. One of the elements
that generates their diversity is the connectivity of the mono-
meric flavan units. In addition to the OPAs connected through
a single interflavan bond, doubly linked OPAs have also been
identified featuring
a dioxabicyclo[3.3.1]nonane skeleton.
Representatives are procyanidin A1 (1), composed of epicate-
chin and catechin, and also procyanidin A2 (2), composed of
2
two epicatechins (Fig. 1).
Although potential biological activities of the doubly-linked
3
OPAs have been reported, further studies have been ham-
pered by scarcity of their samples. This situation could be
changed by organic synthesis, making pure samples available,
although several synthesis challenges need to be solved.
Previously, we reported the total synthesis of procyanidin
Some additional concerns were: (1) poor solubility of free
flavans in common organic solvents, and (2) high sensitivity of
6
free flavans toward oxidants and other conditions.
We report here realization of these ideas to open an
efficient access to doubly linked OPAs, as demonstrated by the
achievement of concise total syntheses of procyanidin A1 (1)
and A2 (2). On the high regioselectivity of the annulation reac-
4
a
A2 (2) (Scheme 1). The double linkage was formed by using
,4-ethylenedioxy flavan derivative 3 as an equivalent of dica-
2
5
tion synthon I and phenol 4 as that of a dianion synthon II.
Among the two oxy-leaving groups in 3, the C4 one is initially
activated by a Lewis acid, generating the C4-cation, that is cap-
tured by the C8 nucleophilic carbon center in 4 with a rigorous
regioselectivity to form intermediate III. The subsequent gene-
ration of the C2-cation in III allows the oxy-cyclization by the
C7-phenol to give the annulated product 5.
Department of Chemistry, Tokyo Institute of Technology, 2-12-1, O-okayama, Meguro-
ku, Tokyo 152-8551, Japan. E-mail: ksuzuki@chem.titech.ac.jp,
kohmori@chem.titech.ac.jp
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/
c9ob01896d
Fig. 1 Doubly-linked proanthocyanidin dimers.
Org. Biomol. Chem., 2019, 17, 9129–9134 | 9129
This journal is © The Royal Society of Chemistry 2019

View Article Online
Communication
Organic & Biomolecular Chemistry
Table 1 Model study
Run
Acid
Solvent
Conditions
11/%
12/%
1
2
3
4
5
PPTS
PPTS
PPTS
CSA
CSA
CSA
EtOH
EtOH
EtOH
EtOH
RT, 24 h
—
49
50
65
84
98
—
—
24
14
7
a
Reflux, 5 min
Reflux, 0.5 h
RT, 24 h
RT, 24 h
RT, 24 h
b
d
EtOH, dioxaned
c
6
EtOH, dioxane
1
a
Compound 13 was obtained in 12% yield, and compound 9 was
b
recovered in 36% yield. Compound 13 was obtained in 16% yield.
Scheme 1 Catechin annulation.
c
d
Molar ratio of 9/10 was 1.0/1.5. v/v = 1.
tive was to find out suitable reaction media, because the two
reaction partners have quite different solubility trends. While
dioxy-flavan unit 9 is lipophilic, free catechins are poorly
soluble in common non-protic solvents. Thus, we started with
ethanol as the medium to test the model reaction, and the
results are summarized in Table 1.
Fig. 2 Nucleophilic flavan units.
To a mixture of dioxy-flavan unit 9 and 10 (1 : 1 molar ratio)
in ethanol was added 2 equivalents of pyridinium p-toluene-
8
sulfonate (PPTS). At room temperature, no reaction occurred
tions, model experiments and computational studies have
been carried out.
even after 24 h, resulting in a complete recovery of the starting
materials (entry 1). However, upon refluxing, the reaction pro-
ceeded rapidly as suggested by TLC-1 after 5 min (entry 2).
Chromatographic purification gave the desired product 11 in
9% yield, and the starting material 9 was recovered in 36%
yield. In addition, a non-polar product was also isolated,
Scheme 2 shows the preparation of the top flavan unit 9.
1
For the ease of the H NMR spectra analysis, all hydroxy
groups in (−)-epicatechin (6) were protected as d -benzyl (Bn*)
7
4
b,d
7
4
ethers.
According to Kawamoto’s protocol, pentaacetate 7
was subjected to in situ hydrolysis and concurrent benzylation
of the resultant phenolates, giving ether 8. Heating of 8, ethyl-
ene glycol (1.5 equiv.) and DDQ (4.0 equiv.) in refluxing
CH Cl gave dioxy-flavan 9, ready for the annulation study.
9
which proved to be 2,4-diethoxy derivative 13 (12% yield). We
reasoned that this is a solvolytic product, which could be reac-
tivated to serve as a dication equivalent to take part in the
annulation.
2
2
As a feasibility study, we examined the reaction of dioxy-
flavan unit 9 and phloroglucinol (10) as a nucleophile for mod-
elling the reactivity of the A-ring in free flavan units. An objec-
Upon conducting the same reaction for a slightly longer
Scheme 2 Synthesis of top flavan unit 9.
time (30 min), dioxy-flavan unit 9 was fully consumed (TLC-2),
9130 | Org. Biomol. Chem., 2019, 17, 9129–9134
This journal is © The Royal Society of Chemistry 2019

View Article Online
Organic & Biomolecular Chemistry
Communication
1
3
giving the desired product 11 in 50% yield (entry 3). The 2,4- (C2 → O → C7, C4 → C8). The reaction predominantly
diethoxy derivative 13 was no longer obtained. However, a side occurred at the C8 and C7–O positions in 6.
product was obtained, which was identified as the bis-annu-
lated product 12 in 24% yield, arising from the iterated annu- polar component (R
lation reaction of the product 11 with 9 and/or 13. two compounds. Further purification by another preparative
To suppress this second annulation, the reaction conditions TLC [CH Cl , EtOAc (7/3), two runs] gave 15 (R = 0.28, 5%
were screened. When PPTS was replaced by camphorsulfonic yield) and 16 (R = 0.33, 4% yield). The less polar isomer was
The TLC purification also gave a small amount of a less
f
= 0.3), which proved to be a mixture of
2
2
f
f
acid (CSA), a stronger acid, the reaction proceeded at room assigned as 16 (C2 → O → C5, C4 → C6) by diagnostic NOEs
temperature (entry 4). The yield of the desired product 11 was observed between H4/C7-phenol and H8/C7-phenol. On the
improved to 65%, although the formation of 12 still persisted. other hand, the more polar isomer 15, having a connectivity as
During this reaction, we noted that the reaction mixture became C2 → O → C7, C4 → C6, by the fact an NOE correlation was
turbid, for which we suspected that the 2,4-diethoxy intermedi- solely observed between H4/C7-phenol, but not between H8/
1
0,11
13
ate 13 was poorly soluble in ethanol at room temperature.
In contrast, the use of 1,4-dioxane as a co-solvent (EtOH,
,4-dioxane v/v = 1) led to a clear solution, improving the ratio of benzyl groups in 14 were removed by hydrogenolysis [H
C7-phenol.
By following our previously established conditions, all the
(1
2
1 over 12 (12 : 1) (entry 5). Increased loading of 10 (1.5 equiv.) atm), ASCA-2®, MeOH, THF, H O (2/2/1), 3 h]. Anaerobic fil-
4
1
2
1
4
1
significantly improved the yield of 11 (98% yield) with a minimal tration (argon) and removal of volatile materials followed by
1
2
formation of the bis-annulated product 12 (entry 6).
lyophilization gave procyanidin A2 (2) (almost pure as assessed
1
Notably, in all the reactions stated above, we never observed by H NMR). Further purification by reverse phase preparative
any self-reaction product of the dioxy-flavan unit 9. This prom- HPLC [Mightysil RP-GP II, 20 mm φ × 250 mm, MeOH, H O
2
ising data supported our expectancy that the non-bromo (35/65) containing 0.1% TFA] and lyophilization gave 2 as a
1
13
capped substrate 9 would work, and we hoped it would apply snow-white amorphous solid. The spectral data ( H and
for real systems as well. NMR, IR, HRMS) of the final product 2 were in full agreement
With these promising results in hand, we turned our atten- with our previously reported data of the natural product.
C
4
a
tion to the annulation of dioxy-flavan unit 9 with (−)-epicate- Similarly, the deprotection of other regioisomeric products 15
chin (6). In a solvent mixture of ethanol and 1,4-dioxane (v/v = and 16 was carried out under hydrogenolytic conditions,
1
5
1
), free flavan 6 (1.5 equiv.) was allowed to react with dioxy- giving proanthocyanidin A6 (17) and proanthocyanidin A7
1
5a,16
flavan 9 in the presence of CSA (Scheme 3). After separation by (18)
in 82% and 86% yield, respectively.
preparative TLC [CH Cl , EtOAc (7/3)], the major annulated
To test the generality of the present annulation method, we
2
2
f
product 14 (R = 0.2) was obtained in 80% yield. The connec- next examined the synthesis of procyanidin A1 (1), a hetero-
tivity of 14 was proven by a diagnostic HMBC correlation dimer composed of (−)-epicatechin (6) as an upper flavan unit
between H4 and C8, confirming it as the C8 linked isomer and (+)-catechin (19) as a lower flavan unit (Scheme 4).
By the protocol described in Scheme 3, stirring a solution
of flavan 9 with 19 in the presence of CSA in a mixed solvent
Scheme 3 Annulation with (−)-epicatechin and synthesis of procyani-
din A2.
Scheme 4 Annulation with (+)-catechin and synthesis of procyanidin A1.
Org. Biomol. Chem., 2019, 17, 9129–9134 | 9131
This journal is © The Royal Society of Chemistry 2019

View Article Online
Communication
Organic & Biomolecular Chemistry
of ethanol and 1,4-dioxane (v/v = 1) smoothly gave the annula- nucleophile 28, removing the B-ring and the C3 hydroxy group
tion product 20 as the major product in 84% yield along with a from a general flavan unit. Thus, the model substrate 28 was
5
mixture of two minor regioisomers 21 and 22 in 8% combined synthesized based on our reported method (Scheme 5). The
yield. The connectivity (2β → O → 7, 4β → 8) in 20 was verified regioselective lithiation of 25 (n-BuLi, THF, −78 °C, 1 h) and
1
3
by a diagnostic HMBC correlation between H4 and C8.
Further separation of the minor regioisomers by preparative BF
TLC [CH Cl , EtOAc (7/3), two runs] gave 21 and 22 in 3% yield with NaH (DMF, 80 °C, 2 h), cleanly affording pyran 27 in 72%
subsequent reaction with oxetane in the presence of
1
8
3
·OEt
2
gave alcohol 26, which was cyclized by treatment
2
2
each, which were characterized by NOE studies.
yield. The simultaneous removal of benzyl groups by hydroge-
(1 nolysis [H (1 atm), ASCA-2®] gave phenol 28.
Having model nucleophile 28 in hand, we conducted the
din A1 (1). All physical data of the synthetic compound fully annulation with di-oxy flavan 9 under previously optimized
Finally, dimer 20 was subjected to hydrogenolysis [H
2
2
2
atm), ASCA-2®, MeOH, THF, H O (2/2/1), 3 h], giving procyani-
2
a
matched with those of the reported natural product.
conditions. The reaction proceeded smoothly, giving the
Moreover, other regioisomers 21 and 22 were deprotected by anticipated C8-regioisomer 29 as a major product (86% yield),
the same hydrogenolysis, giving a natural proanthocyanidin along with the C6-regioisomers, 30 and 31 in 4% yield, respect-
1
7
13
2
3
(no trivial name) and its regioisomer 24 in 89% and 94% ively (Scheme 6). This observation concluded that the pres-
yield, respectively. The latter isomer 24 is a new compound, ence of an aryl ring (catechol) at C2 and the hydroxy group at
which has not yet been isolated from nature. C3 on the pyran ring has essentially no effect on the C8/C6-
The OPA syntheses described above feature high regio- regioselectivity in the dimerization.
selectivity of the annulation, which is outlined in Fig. 3.
The question was now focused to the relative reactivities of
Formation of the major product, 14 or 20, could be traced the C8 and the C6 centers toward electrophiles. The high regio-
back to the predominant formation of the C4–C8 bond selectivities of the annulation originate from the site selectivity
between two flavan units I and II, forming the major inter- of the nucleophilic free catechin unit, C8 > C6, implying the
mediate III, which unequivocally undergoes oxycyclization by relevance to the relative natural abundance of the C8-linked
1
9
the C7-phenol, giving the major product V. The minor pathway oligomers in comparison with the C6-linked congeners.
starts with the C6 center in II to generate intermediate IV, To address the origin of C8/C6-regioselectivity, the HOMO
which entails two optional oxycyclizations to give almost equal coefficients of 28 at the C6 and the C8 positions were esti-
amounts of the cyclized products VI and VII in an essentially mated by DFT calculation, which was not fruitful, as no
non-selective manner.
obvious difference was observed. By contrast, DFT calculation
Concerning the mechanistic basis for the regioselectivity at
the key annulation stage, we raised a question on the origin of
the C8/C6-regioselectivity. To address the preference of the C8
position over the C6 position in the initial C–C bond for-
mation, we carried out a model study with a “simplified”
Scheme 5 Synthesis of model substrate 28.
Fig. 3 Rationale for the regioselective annulation.
Scheme 6 Annulation with model nucleophilic partner 28.
9132 | Org. Biomol. Chem., 2019, 17, 9129–9134
This journal is © The Royal Society of Chemistry 2019

View Article Online
Organic & Biomolecular Chemistry
Communication
cyanidins. Model experiments and computational studies vali-
dated the high regiochemical selectivity in annulation reac-
tions. Biological studies using synthetic dimers are currently
in progress.
Fig. 4 DFT calculations on the Wheland intermediate.
Conflicts of interest
There are no conflicts to declare.
Acknowledgements
This work was supported by the JSPS KAKENHI Grant
Numbers JP16H06351, JP16H04107, JP18H04391 and Nagase
Science Technology Foundation.
Notes and references
Fig. 5 Origin of C8 regioselectivity.
1
(a) A. B. Bohm, Introduction to Flavonoids, Harwood
Academic, Amsterdam, 1998; (b) Plant Polyphenols 2:
Chemistry, Biology, Pharmacology, Ecology, ed. G. G. Gross,
R. W. Hemingway, T. Yoshida and S. J. Branham, Kluwer
Academic/Plenum, New York, 1999; (c) Flavonoids:
Chemistry, Biochemistry and Applications, ed. Ø. M.
Andersen and K. R. Markham, CRC Press/Taylor & Francis,
Boca Raton, 2006.
on the putative Wheland intermediates 32 and 33, generated
+
by the reaction on the model substrate 28 with “CH
3
”
suggested that the C8-methyl intermediate 32 is energetically
preferred over the C6-counterpart 33, in line with the preferred
C8 substitution rather than the C6 substitution (Fig. 4).
2
0a
To gain further insight, we addressed LUMO maps
the NBOs
and
2
0b
of 32 and 33 (Fig. 5). The LUMO maps identified
2 (a) K. Weinges, W. Kaltenhäuser, H.-D. Marx, E. Nader,
F. Nader, J. Perner and D. Seiler, Justus Liebigs Ann. Chem.,
1968, 711, 184; (b) G.-I. Nonaka, S. Morimoto, J.-I. Kinjo,
T. Nohara and I. Nishioka, Chem. Pharm. Bull., 1987, 35,
149; (c) K. Weinges, K. Göritz and F. Nader, Justus Liebigs
Ann. Chem., 1968, 715, 164; (d) D. Jacques, E. Haslam,
G. R. Bedford and D. Greatbanks, J. Chem. Soc., Perkin
Trans. 1, 1974, 2663.
3 (a) T. D. Bruyne, L. Pieters, H. Deelstra and A. Vlietinck,
Biochem. Syst. Ecol., 1999, 27, 445; (b) D.-Y. Xie and
R. A. Dixon, Phytochemistry, 2005, 66, 2127; (c) F. He,
Q.-H. Pan, Y. Shi and C.-Q. Duan, Molecules, 2008, 13, 3007;
(d) X. Xu, H. Xie, Y. Wang and X. Wei, J. Agric. Food Chem.,
2010, 58, 11667; (e) H. Zhang, Yerigui, Y. Yang and C. Ma,
J. Agric. Food Chem., 2013, 61, 8814.
4 (a) Y. Ito, K. Ohmori and K. Suzuki, Angew. Chem., Int. Ed.,
2014, 53, 10129; (b) Y. Noguchi, R. Takeda, K. Suzuki and
K. Ohmori, Org. Lett., 2018, 20, 2857; (c) K. Ohmori,
N. Ushimaru and K. Suzuki, Proc. Natl. Acad. Sci. U. S. A.,
2004, 101, 12002; (d) K. Ohmori, T. Shono, Y. Hatakoshi,
T. Yano and K. Suzuki, Angew. Chem., Int. Ed., 2011, 50,
4862.
C5 for 32 and C8a for 33 as the most electron-deficient sites,
which correspond to the para positions to the methyl-bonded
carbons. The NBO analyses showed both of these sites accept
significant electron donation from the adjacent oxygen atoms
(see C5–O5 for 32 and C8a–O1 for 33), which stand in contrast
to the weaker interactions of other C–O bonds.
Implication from these aspects is that both 32 and 33 are
best expressed by canonical structures I and II, in which one
could find a common local structure of a 1,3,5-trioxy-substi-
tuted pentadienyl cation, sharing a localized cation at a central
carbon center that is stabilized not only by the two CvC
bonds, but also by a strong n → p conjugation.
Although the electronic and orbital parameters are virtually
the same in I and II, we noticed a notable difference in the
molecular shape. Namely, the bicyclic skeleton in II is sizably
distorted, presumably for paying the cost of localizing a CvC
bond at the exo position (C4a–C5) to the pyran ring, while the
shape of I appears to be free from such constraints. We believe
this is the origin of the energy difference of 32 and 33, which
2
1
in turn would explain the C8/C6-regioslectivity.
In conclusion, we have demonstrated the direct syntheses
of the doubly linked OPAs, procyanidins A1 (1) and A2 (2) in
five steps from commercially available starting materials. This
approach significantly simplifies the synthetic procedure, side-
stepping the selective protection of synthetically indistinguish-
able phenols on the nucleophilic flavan unit, and it could be
employed for the synthesis of various doubly linked proantho-
5 S. Stadlbauer, K. Ohmori, F. Hattori and K. Suzuki, Chem.
Commun., 2012, 48, 8425.
6 (a) A. J. Birch, J. W. Clark-Lewis and A. V. Robertson,
J. Chem. Soc., 1957, 3586; (b) P. Kiatgrajai, J. D. Wellons,
L. Gollob and J. D. White, J. Org. Chem., 1982, 47, 2910;
(c) P. P. Mehta and W. E. Whalley, J. Chem. Soc., 1963, 5327;
This journal is © The Royal Society of Chemistry 2019
Org. Biomol. Chem., 2019, 17, 9129–9134 | 9133

View Article Online
Communication
Organic & Biomolecular Chemistry
(
d) N. Ishino, E. Yanase, S. Nakatsuka and W. E. Whalley, 14 N. E. CHEMCAT, Co.
Biosci. Biotechnol. Biochem., 2010, 74, 875.
H. Kawamoto, Nakatsubo and K. Murakami, Synth.
Commun., 1996, 26, 531.
15 (a) S. Morimoto, G.-I. Nonaka and I. Nishioka, Chem.
Pharm. Bull., 1987, 35, 4717; (b) K. Kamiya, C. Watanabe,
H. Endang, M. Umar and T. Satake, Chem. Pharm. Bull.,
2001, 49, 551.
16 L. J. Porter, Z. Ma and B. G. Chant, Phytochemistry, 1999,
51, 297.
7
8
9
M. Miyashita, A. Yoshikoshi and P. A. Grieco, J. Org. Chem.,
1
977, 42, 3772.
Obtained as a single diastereomer, albeit stereochemistries
unassigned.
0 A pure sample of diethoxy flavan 13 was prepared in 93%
17 H. Lou, Y. Yamazaki, T. Sasaki, M. Uchida, H. Tanaka and
S. Oka, Phytochemistry, 1999, 51, 297.
1
yield from 9 (CSA, EtOH, 1,4-dioxane, RT, 16 h). Indeed, it 18 M. Yamaguchi, Y. Nobayashi and I. Hirao, Tetrahedron Lett.,
was found to be sparingly soluble in ethanol, and soluble
in 1,4-dioxane.
1 Re-exposure of 13 with 10 to the same acidic conditions
1983, 24, 5121.
19 G. J. Tanner, K. T. Francki, S. Abrahams, J. M. Watson,
P. J. Larkin and A. R. Ashton, J. Biol. Chem., 2003, 278, 31647.
20 (a) Calculated by Spartan’18 program. For details, see the
ESI;†; (b) Calculated by Gaussian16 program. For details,
see the ESI.†
1
1
1
gave the annulation products 11 and 12.
2 Reversal of the ratio of 9/10 to (1.5 : 1) drastically increased
the formation of the bis-annulated product 12 over 11.
3 See the ESI for spectral details.†
21 Details will be reported in due course.
9134 | Org. Biomol. Chem., 2019, 17, 9129–9134
This journal is © The Royal Society of Chemistry 2019