Theoretical investigation of the rubicordifolin cascade

Article abstract of DOI:10.1021/ol102157d

The results of a theoretical investigation on the complex cascade reaction leading to the natural product rubicordifolin are reported. These computations analyze the discrete transformations that are required during the conversion of the vinyl naphthoquinone starting material into the natural product, including two pseudopericyclic cyclizations as well as a diastereoselective, hetero-Diels-Alder reaction.

Full text of DOI:10.1021/ol102157d

ORGANIC
LETTERS
2
010
Vol. 12, No. 22
162-5165
Theoretical Investigation of the
Rubicordifolin Cascade
5
,
†
‡
Jean-Philip Lumb,* Jamin L. Krinsky, and Dirk Trauner
§
Department of Chemistry, UniVersity of California, Berkeley,
California 94720, United States
jplumb@stanford.edu
Received September 10, 2010
ABSTRACT
The results of a theoretical investigation on the complex cascade reaction leading to the natural product rubicordifolin are reported. These
computations analyze the discrete transformations that are required during the conversion of the vinyl naphthoquinone starting material into
the natural product, including two pseudopericyclic cyclizations as well as a diastereoselective, hetero-Diels-Alder reaction.
Over the past several years we have been interested in the
synthesis of an unusual class of pseudodimeric naphtho-
Scheme 1. Synthesis of Rubicordifolin (2) and Furomollugin (3)
quinones isolated from the medicinal plants Rubia cordifolia
1
,2
and R. oncotricha. Our syntheses have been based on
presumed biosynthetic pathways, guided largely by the
surprising fact that these highly oxidized natural products
are isolated in racemic form. Early on in the development
of this program, we reported that vinyl quinone 1 undergoes
a remarkable transformation in the presence of phenylboronic
acid at elevated temperatures, providing rubicordifolin (2)
and furomollugin (3) in yields of 45% and 36%, respectively
2
a
(
Scheme 1).
Initially, we proposed that an endo-selective, [4 + 2]
cycloaddition between o-quinone methide 4 and vinyl-
naphthofuran 5 could account for the formation of 2 as a
single diastereomer. However, certain aspects of this hy-
pothesis merited further investigation. These included the
precise manner in which 1 is converted into 4 and 5, as well
as an explanation for the exquisite diastereocontrol in the
cycloaddition.
3
†
With the aim of shedding light onto the mechanism of
this reaction we initiated a theoretical investigation into the
behavior of 1. Herein we report the results of these
Present address: Department of Chemistry, Stanford University.
Present address: Molecular Graphics and Computation Facility, Uni-
‡
versity of California, Berkeley.
§
Present address: Department of Chemistry, University of Munich.
(
1) For a review of bioactive natural products derived from Rubia see:
Chauhan, S.; Singh, R. Chem. BiodiVersity 2004, 1, 1241
2) (a) Lumb, J. P.; Trauner, D. J. Am. Chem. Soc. 2005, 127, 2870. (b)
Lumb, J. P.; Choong, K. C.; Trauner, D. J. Am. Chem. Soc. 2008, 130,
230
.
(3) Despite the absence of products arising from exo- or regioisomeric
transition states in the crude reaction mixture, their formation cannot be
unequivocally excluded in light of the incomplete mass balance of the
reaction.
(
9
.
10.1021/ol102157d  2010 American Chemical Society
Published on Web 10/25/2010

4
calculations which, in conjunction with experimental find-
energy barrier of 32.0(29.4) kcal/mol. In transition state TS1-
9, significant twisting about the C -C double bond places
a formally empty p-orbital at C in the best natural bond
ings, have led us to revise our proposed mechanism for the
formation of 2. The relatively new M06-2X functional was
used for geometry optimizations and frequency calculations,
in conjunction with the 6-311+G(d,p) basis set. When
3
4
5
4
8
9
orbital (NBO) resonance form. According to a NLMO
6
analysis, the C
1
-O
1
π-bond donates 2.7% of its electron pair
possible, the energies of optimized structures were recalcu-
lated by using the double hybrid functional B2PLYP-D,
to C , while an O
4
1
lone pair donates 7.9% of its electron
7
density. This suggests that the transformation of 1 to 9 is
more accurately described as a pseudopericyclic cyclization,
by the arrow pushing mechanism illustrated in Scheme 3,
which includes a second-order perturbative component and
an empirical dispersion correction; results for this method
are given in parentheses. Reported energies for both methods
were corrected for zero-point energy calculated at the
M06-2X level, and isomeric sets were constructed so that
energies could be directly compared to that of compound 1
1
0
than as a classical Nazarov-type cyclization.
Once formed, zwitterion 9 can undergo either of two
irreversible transformations (Scheme 3, paths B and C). In
path B, a presumably intermolecular proton transfer generates
(
H
1 is defined as two molecules of 1 and one molecule of
O). All values are reported in kcal/mol.
6
, which is 30.8(23.5) kcal/mol downhill relative to s-cis-1
2
(the kinetics of this step were not modeled). In path C, a
retro-hydroxyalkylation initially affords ion-pair 10. Fol-
lowing proton transfer, this path yields furomollugin (3) and
one molecule of acetone, which are favored enthalpically
over s-cis-1 by 18.7(13.6) kcal/mol. The energy of activation
Scheme 2. Reactivity of Vinylnaphthoquinone 1
E(act)9-10 for the formation of ion pair 10 was estimated to
be 18.0(17.9) kcal/mol by optimizing the reactant 9 and the
dissociation transition state using a PCM water solvation
1
1
model. We assume that the relative energy barriers of
pathways B and C are energetically similar given that
formation of 3 and 6 appears to be irreversible, and that these
two products are formed in near equimolar quantities when
starting from vinylquinone 1 (Scheme 2).
While formation of 9 by this mechanism is predicted to
be feasible at the elevated temperatures required to generate
rubicordifolin (2), transition state TS1-9 is prohibitively high
in energy to explain the ambient-temperature reactivity of 1
12
in THF (Scheme 2). The presence of adventitious acid was
1
3,14
not ruled out in the latter case,
and thus the above
calculations were repeated with protonation at the quinone
carbonyl. Under acid catalysis, the activation energy required
for the formation of 9 decreases to 7.8(6.0) kcal/mol, while
First we discuss the proposed formation of dienophile 5
15
and furomollugin (3). Vinylnaphthoquinone 1 underwent
spontaneous cyclization to provide a 1:1 mixture of 3 and 6
when left as a dilute solution in THF for extended periods
E(act)9-10 decreases to 11.2(11.1) kcal/mol (with a THF PCM
solvent correction). This is consistent with the experimental
2
a
at room temperature (Scheme 2). All attempts to improve
upon the selectivity or yield of this process by changing
solvent, temperature, or pH resulted in either decreased yields
of 3 and 6 or intractable mixtures. Importantly, exposure of
observation that the rate of formation of 3 and 6 is greatly
14
increased in the presence of either Brønsted or Lewis acids.
(
(
8) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88, 899.
1
to a variety of Lewis acids, including complexes of Mg,
9) For Natural Localized Molecular Orbital (NLMO) analysis see: (a)
Ti, Al, or Sn, consistently resulted in near instantaneous
formation of 3 and 6, albeit in low isolated yields. The sole
Reed, A. E.; Weinhold, F. J. Chem. Phys. 1985, 83, 1736. (b) H u¨ bler, K.;
Hunt, P. A.; Maddock, S. M.; Rickard, C. E. F.; Roper, W. R.; Salter, D. M.;
Schwerdtfeger, P. Organometallics 1997, 16, 5076.
exception was the action of Sc(OTf)
3
in CH
3
CN, which
(10) For a recent computational analysis of pseudopericyclic cyclizations
see: Duncan, J. A.; Calkins, D. E. G.; Chavarha, M. J. Am. Chem. Soc.
afforded dimer 8 in 60% yield. These results are significant
when discussing the role that phenylboronic acid plays in
the synthesis of 2 (vide infra).
2
008, 130, 6740, and references cited therein.
(
11) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. ReV. 2005, 105, 2999.
Only the activation energy is given here because the T.S. energy is not
directly comparable with the other (gas-phase) results.
Computational results suggest that the formation of 3 and
(
12) Since the reaction mixture was exposed to ambient light, a radical
6
occurs via a reversible cyclization of s-trans-1 to form
mechanism could account for the room temperature cyclization of 1. For a
related, photoinduced cyclization of vinyl quinones see: Iwamoto, H.;
Takuwa, A.; Hamada, K.; Fujiwara, R. J. Chem. Soc., Perkin Trans. 1 1999,
zwitterion 9 (Scheme 3, path A), passing over a sizable
5, 575. A light-induced radical mechanism in the formation of rubicordifolin
(
4) Frisch, M. J., et al. Gaussian 09, Revision A.2, Gaussian, Inc.,
seems unlikely, however, since the reaction vessel was thoroughly wrapped
in aluminum foil prior to thermolysis.
Wallingford, CT, 2009. Complete citation in the Supporting Information.
(
(
5) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215.
6) (a) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem.
(13) Butyric acid is a known impurity of “aged” THF: Coetzee, J. F.;
Chang, T.-H. Pure Appl. Chem. 1985, 57, 633, and references therein
(14) For a related room temperature, acid mediated cyclization of vinyl-
p-quinones see: Taing, M.; Moore, H. W. J. Org. Chem. 1996, 61, 329
(15) See the Supporting Information for a complete analysis of proto-
nated 1.
.
Phys. 1980, 72, 650. (b) McLean, A. D.; Chandler, G. S. J. Chem. Phys.
980, 72, 5639.
7) (a) Grimme, S. J. Chem. Phys. 2006, 124, 034108. (b) Schwabe, T.;
1
.
(
Grimme, S. Phys. Chem. Chem. Phys. 2007, 9, 3397.
Org. Lett., Vol. 12, No. 22, 2010
5163

a
Scheme 3. Proposed Mechanism for the Formation of Furomollugin (3) and Coupling Partners 11 and 5
a
The various pathways accessible to vinylquinone 1 are denoted in red above the appropriate arrows. Arrows that are in blue represent irreversible steps
in the formation of furomollugin (3) and coupling partners 11 and 5.
The final step in the formation of dienophile 5 is the
dehydration of naphthofuran 6 (Scheme 3). Experimental and
computational results suggest that this process is reversible given
that calculated 6 is enthalpically favored over 5 (and one
Given the ease with which 11 is formed, we conclude that
its irreversible hydrolysis to generate 4 must either be slower
than the formation of 9 and 5 or inoperative in order to allow
1
8
for the formation of 9 and thus 5. Alternatively, 11 could
serve as a suitable diene for the key hetero-Diels-Alder
cycloaddition with 5, wherein hydrolysis of the resulting
2
molecule of H O) by 11.2(8.6) kcal/mol, and that exposing 6
to the reaction conditions does not produce 5. Since phenyl-
boronic acid is required for the formation of rubicoridfolin (2),
and since byproducts such as dimer 8 are absent from the crude
reaction mixture (Scheme 2), we believe that this reagent
facilitates the reversible dehydration of 6 under mild condi-
1
9
adduct would then generate 2 (vide infra).
Taken together, the results presented above suggest that
the formation of coupling partners 11 and 5 proceeds via
reversible cyclization of 1 to 11 concomitant with the rate-
limiting formation of intermediate 9. Intermolecular proton
transfer would then provide 6, which in the presence of a
16
tions.
Next, we turn our attention to the identity and formation
of the key heterodiene coupling partner. Initially we assumed
that phenylboronic acid, serving as a Lewis acid, facilitated
cyclization of 1 and subsequent demethylation of the
carbomethoxy group to generate coupling partner 4 (Scheme
2
0
suitable boron reagent transiently generates 5. The dehy-
dration of 6 is predicted to be significantly faster than the
1
6
conversion of 1 to 9, which is a requirement in order for
sufficient quantities of 11 to be present to react with 5.
With an understanding of the various equilibria accessible
to 1, we turned our attention to the key hetero-Diels-Alder
cycloaddition. As previously discussed, we initially proposed
that o-quinone methide 4 underwent an endo-selective cycload-
dition with vinyl naphthofuran 5 to directly afford rubicordifolin
2
a
1
). However, this hypothesis now seems unlikely, given
the propensity of 1 to rapidly cyclize to 3 and 6 in the
presence of Lewis acids.
On the other hand, the noncatalyzed cyclization of s-cis-1
to generate 11 is predicted to be endothermic by ∼2 kcal/
1
7
2a
mol and to pass over an energy barrier, TS1-11, of only
(2) (Scheme 1). While this Diels-Alder reaction is predicted
21
1
9.4(18.9) kcal/mol (Scheme 3, path D). Therefore, at
to be accessible under the reaction conditions, 11 is more
likely the [4 + 2] coupling partner based on the results presented
above. Accordingly, we evaluated an alternative sequence of
events in which the 11/5 cycloaddition precedes hydrolysis
(Scheme 4). In this case the pathway through endo-TS12 is
elevated temperatures 1 is predicted to exist in equilibrium
with 11. Interestingly, the propensity for 1 to cyclize to 9 vs
1
1 appears to be sensitive to pH. The noncatalyzed value of
TS1-11 is 12.6(10.5) kcal/mol lower than TS1-9, whereas
under acid catalysis, TS1-9 is favored by ∼2 kcal/mol. This
suggests that careful maintenance of a neutral pH is required
for the formation of 11 and thus the natural product.
(18) The corresponding carboxylic acid of 1 has been evaluated as a
potential reactive intermediate. See the Supporting Information for a detailed
discussion.
(
19) For a related cyclization/[4 + 2] cycloaddition cascade see the
(
16) The dehydration of 2-(benzofuran-2-yl)propan-2-ol mediated by
synthesis of torreyanic acid: Li, C.; Johnson, R. P.; Porco, J. A. J. Am.
Chem. Soc. 2003, 125, 5095.
phenylboronic acid was modeled by using a THF PCM solvent correction.
The activation energy is 20.5 kcal/mol and the transformation is endothermic
by 8.3 kcal/mol. See the Supporting Information for details.
(20) At elevated temperatures, phenylboronic acid is believed to exist
2
in equilibrium with phenyl boroxine and 3 equiv of H O: Chen, F.; Kina,
(
17) Note that the B2PLYP-D value of 8.0 kcal/mol suggests that the
A.; Hayashi, T. Org. Lett. 2006, 8, 341. For a review of boronic acids in
catalysis see: Ishihara, K.; Yamamoto, H. Eur. J. Org. Chem. 1999, 3, 527.
(21) See the Supporting Information for the 4/5 cycloaddition.
M06-2X value of 2 kcal/mol represents a lower limit in the energetic
difference between 1 and 11.
5164
Org. Lett., Vol. 12, No. 22, 2010

Scheme 4. Analysis of the Key, Hetero-Diels-Alder Cycloaddition
remarkably facile (Eact ) 4.6 kcal/mol) and favored over exo-
TS13 by 8.5 kcal/mol. Regioisomeric transition states TS14
and TS15 are disfavored by 11.1 and 14.0 kcal/mol, respec-
tively. Irreversible hydrolysis of adduct 16, either under the
reaction conditions or upon workup, would then lead to
rubicordifolin (2).
partners from two reservoirs, which are in turn separated by
the rate-determining step of the reaction.
Scheme 5. Overall Reaction Mechanism
It appears that a favorable π-stacking interaction between 11
and 5 in the [4 + 2] transition state plays a critical role in
determining the stereochemical outcome of the cycloaddition
(Scheme 4). In TS12 the two coupling partners are separated
by roughly 3.3 Å, which is well within the 3.6 Å threshold
22b
generally accepted for intermolecular interactions. Interest-
ingly, in preliminary studies with the B3LYP functional,
π-stacking was not observed, and at this level of theory exo-
5,22
TS13 was slightly favored over endo-TS12.
The relative
stereochemistry of 2 has now been confirmed by single-
crystal X-ray diffraction analysis (Scheme 4) and is consistent
with the relative stereochemistry resulting from endo-TS12.
Consequently, we conclude that π-stacking provides a
sufficient energetic stabilization to dictate the stereochemical
course of the cycloaddition, and thus the observed config-
uration of the natural product.
On the basis of the results presented herein, we propose a
revised mechanism that accounts for the remarkable forma-
tion of rubicordifolin (2) in a single synthetic operation
starting from naphthoquinone 1 (Scheme 5). Importantly, this
proposal does not require the orchestration of two, irrevers-
ible reaction steps in the formation of the natural product,
but instead, predicts the reversible formation of both coupling
Acknowledgement. We thank Pfizer Inc. for sponsoring an
ACS predoctoral fellowship for J.P.L. All computations were per-
formed at the UCB Mol. Graphics and Comp. Facility on equip-
ment purchased with funds from NSF Grant CHE-0840505.
Supporting Information Available: The synthetic pro-
cedure and spectroscopic data for compound 8, crystal-
lographic data for 2 (CIF), and computational data. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(
22) For examples of B3LYP acting poorly in modeling dispersion forces
ˇ
see: (a) Sponer, J.; Riley, K. E.; Hobza, P. Phys. Chem. Chem. Phys. 2008,
0, 2595. (b) Ujaque, G.; Lee, P. S.; Houk, K. N.; Hentemann, M. F.;
Danishefsky, S. J. Chem.sEur. J. 2002, 8, 3423. (c) Zhao, Y.; Truhlar,
1
D. G. J. Chem. Theory Comput. 2006, 2, 1009
.
OL102157D
Org. Lett., Vol. 12, No. 22, 2010
5165