Asymmetric cyclopentannelation: Camphor-derived auxiliary

Article abstract of DOI:10.1021/ja020591o

The scope of an enantioselective cyclopentannelation reaction that makes use of allenyl ether-derived nucleophiles has been probed. The enantioselectivity is induced by a traceless chiral auxiliary that is easily derived from camphor. It has been shown th

Full text of DOI:10.1021/ja020591o

Published on Web 08/01/2002
Asymmetric Cyclopentannelation: Camphor-Derived Auxiliary
Paul E. Harrington, Tsuyoshi Murai, Chester Chu, and Marcus A. Tius*
Contribution from the Department of Chemistry, 2545 The Mall, UniVersity of Hawaii,
Honolulu, Hawaii 96822, and The Cancer Research Center of Hawaii, 1236 Lauhala Street,
Honolulu, Hawaii 96813
Received April 25, 2002
Abstract: The scope of an enantioselective cyclopentannelation reaction that makes use of allenyl ether-
derived nucleophiles has been probed. The enantioselectivity is induced by a traceless chiral auxiliary that
is easily derived from camphor. It has been shown that for γ-substituted allene ethers that are axially chiral,
very high enantiomeric excesses of cyclopentenone products are observed in the matched cases. Two
fundamentally different mechanisms are observed, one for the cyclizations of allenyl ketones (see eq 7),
the other for the cyclizations of allenyl alcohols (see eq 11). The methodology is versatile, efficient, and
well-suited for applications in synthesis.
conrotatory ring closure to cyclic cation 3. Loss of residue R4
Introduction
as a stable cation in an irreversible step leads to the observed
product, cross-conjugated cyclopentenone 4.
Some years ago we described a variant of the classical
Nazarov cyclization in which an allene ether function partici-
pates in the conrotation.1 In our early efforts, we explored
versions of this reaction that provided racemic products, but as
the potential of the methodology became clearer to us, we
considered adaptations that might provide enantiomerically
enriched products.2,3 We recently reported the first total synthesis
of natural (22R,23R)-roseophilin through an application of the
enantioselective version of the cyclopentannelation reaction.4
There are several aspects of the reaction that are noteworthy,
the most startling of which is the ease with which the cyclization
takes place. For example, we have been unable to isolate
R-allenyl ketones such as 1 (eq 1) because they undergo
spontaneous cyclization during workup.5 This contrasts with the
behavior of divinyl ketones, the conventional substrates of the
Nazarov reaction,6 cyclization of which often requires exposure
to strong acid.7 Our mechanistic hypothesis is summarized in
eq 1. Reversible protonation of ketone 1 leads to pentadienyl
cation 2. This intermediate can undergo thermally allowed
There are several points to be noted concerning this process.
Cleavage of cation R4 from intermediate 3 must take place
rapidly, otherwise competing reaction pathways involving
rearrangements or proton loss or both from 3 will erode the
yield of 4. We have had our best successes when R4 has been
an alkoxyalkyl group, so that the cation +R4 is stabilized by
electron pair donation from the adjacent oxygen atom. The
stereodetermining step is the cyclization of 2 to 3. If protonation
of 1 is reversible and if the cyclization is slow, erosion of the
geometrical integrity of the trisubstituted alkene can take place.
Loss of the stereochemical information of the alkene results in
a nonstereoselective cyclization reaction. This suggests that the
mildest possible conditions for the cyclization may not be
optimal for maintaining the stereospecificity of the cyclization.
For maximal stereospecificity it is useful to have a high
concentration of strong acid, leading to a high concentration of
2, and a fast cyclization.
* To whom correspondence should be addressed. E-mail: tius@
gold.chem.hawaii.edu.
(1) For example, see: (a) Tius, M. A.; Astrab, D. P. Tetrahedron Lett. 1984,
25, 1539. (b) Tius, M. A.; Astrab, D. P.; Fauq, A. H.; Ousset, J.-B.; Trehan,
S. J. Am. Chem. Soc. 1986, 108, 3438. (c) Tius, M. A.; Astrab, D. P.
Tetrahedron Lett. 1989, 30, 2333. (d) Tius, M. A.; Zhou, X.-M. Tetrahedron
Lett. 1989, 30, 4629. (e) Tius, M. A.; Kwok, C.-K.; Gu, X.-q.; Zhao, C.
Synth. Commun. 1994, 24, 871. (f) Tius, M. A.; Drake, D. J. Tetrahedron
1996, 52, 14651. (g) Tius, M. A.; Busch-Petersen, J.; Yamashita, M.
Tetrahedron Lett. 1998, 39, 4219. (h) Tius, M. A.; Hu, H.; Kawakami, J.
K.; Busch-Petersen, J. J. Org. Chem. 1998, 63, 5971. (i) Tius, M. A.; Chu,
C. C.; Nieves-Colberg, R. Tetrahedron Lett. 2001, 42, 2419.
(2) Hu, H.; Smith, D.; Cramer, R. E.; Tius, M. A. J. Am. Chem. Soc. 1999,
121, 9895.
(3) Harrington, P. E.; Tius, M. A. Org. Lett. 2000, 2, 2447.
(4) Harrington, P. E.; Tius, M. A. J. Am. Chem. Soc. 2001, 123, 8509.
(5) Hashmi and co-workers have noted similar reactivity in a related system:
Hashmi, A. S. K.; Bats, J. W.; Choi, J.-H.; Scharz, L. Tetrahedron Lett.
1998, 39, 7491.
(6) (a) Jacobi, P. A.; Armacost, L. M.; Kravitz, J. I.; Martinelli, M. J.; Selnick,
H. G. Tetrahedron Lett. 1988, 29, 6865. (b) Bender, J. A.; Arif, A. M.;
West, F. G. J. Am. Chem. Soc. 1999, 121, 7443. (c) Denmark, S. E.;
Wallace, M. A.; Walker, C. B., Jr. J. Org. Chem. 1990, 55, 5543.
(7) For discussions of the Nazarov reaction, see: Dolbier, W. R., Jr.; Koroniak,
H.; Houk, K. N.; Shen, C. Acc. Chem. Res. 1996, 29, 471 and Habermas,
K. L.; Denmark, S.; Jones, T. K. In Organic Reactions; Paquette, L. A.,
Ed.; John Wiley & Sons: New York, 1994; Vol. 45, pp 1-158.
9
10.1021/ja020591o CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 10091-10100
10091

A R T I C L E S
Harrington et al.
In cases in which R1 * H, the allene function is stereogenic.
Consideration of the two modes of conrotation that are available
to 2 suggests that the preferred mode, as indicated in eq 1, takes
place so as to move R1 away from the steric bulk of R2, leading
to the Z exocyclic double bond geometry in 3 and 4. In
preliminary work we have shown that this preference is
amplified when R1 is a large group, e.g., tert-butyl. By
controlling the direction of conrotation, one can transfer the axial
chirality of the allene function in 1 to the tetrahedral ring carbon
atom in 4.2
It is useful to consider two cases for performing the
cyclopentannelation reaction of eq 1 enantioselectively. First,
in the case in which R1 * H, control of the axial chirality of
the allene ether function in 1 results in control of absolute
stereochemistry in 4, as indicated in the preceding paragraph.
There is as yet no convenient method for preparing nonracemic
allene ethers.8 Although Hoppe and co-workers have demon-
strated an elegant synthesis of enantiomerically enriched axially
chiral allenyl carbamates, it is not clear whether their method
can be adapted to the synthesis of allene ethers.9 The allenyl
carbamates have been explored in the cyclopentannelation
reaction, and it appears that they react largely through a different
manifold.10 Second is the case in which R1 ) H, in which the
allene function in 1 is not stereogenic. Here there are two options
for controlling the absolute stereochemistry of product 4. One
can either use a chiral auxiliary as a control element, or one
can consider using a chiral Lewis acid catalyst in place of proton
in the first step of eq 1. The ease with which the cyclopentan-
nelation process always proceeds, even during workup with
aqueous KH2PO4 in all cases, suggests that a better application
of the chiral Lewis acid approach will be to cyclizations that
are slower than the one in eq 1. This makes the chiral auxiliary
approach the most attractive avenue for investigation.
ings that limited their utility. First, the nucleophilicity of
allenyllithium reagent 6 was limited. It was necessary to add
several equivalents of LiCl to the reaction mixtures in order
for the addition to morpholino enamides (e.g., 7) to take place
efficiently.11 Second, whereas the ee of products was good on
modest scale (up to 0.2 mmol), we discovered that marked
erosion of the ee took place upon scaling up of the reaction to
4 mmol. The preparation of the sugar-derived auxiliaries was
also tedious on large scale. For these reasons we developed a
second-generation chiral auxiliary based on camphor which was
not subject to the three limitations cited above. Camphor is an
attractive starting point for the synthesis of chiral auxiliaries
since both enantiomeric forms are available and are reasonably
cheap. Equation 3 summarizes our first result with the camphor-
derived auxiliary 9.4 Addition of 9 to morpholino enamide 7,
followed by treatment with HCl in a mixture of hexafluoroiso-
propanol (HFIP)12 and trifluoroethanol (TFE) at -78 °C led to
cyclopentenone ent-8 in 78% yield in 86% ee on a 1 g scale.
Cyclopentenone ent-8 was the key intermediate in our synthesis
of natural roseophilin. Following this early demonstration of
the utility of 9 in total synthesis, we wanted to determine the
scope and the limits of applicability of 9 to the enantioselective
construction of cross-conjugated cyclopentenones. We also
wanted to explore the effects on yield and enantioselectivity of
introducing substituents at the distal carbon atom of the allene
function in 9. We describe our findings in what follows.
Results and Discussion
A limited series of experiments was performed first to
optimize yield and optical purity for the reaction of eq 3. The
results of this study are summarized in Table 1. It is easy to
overinterpret these results, but it can be stated with confidence
that use of the polar solvent HFIP led to the best enantiomeric
excesses of product and that lower temperature was better than
higher temperature. The yield of ent-8 in Table 1, entry 5, is
given as 60%. This is for a reaction that was conducted with
0.2 mmol of 7. On 4 mmol scale, the yield of ent-8 improved
to 78%, whereas the ee was unaffected. Since HFIP has a high
melting point, it was advantageous to use a mixture with TFE
so that the cyclization reaction could be conducted at -78 °C.
Only two experiments involved the use of Lewis acid catalysts
Our first investigations of chiral auxiliaries led us to consider
pyranose derivatives, since they are cheap and readily available.3
Furthermore, loss of the sugar residue as a stable oxo cation
(cf. +R4, eq 1) can take place following cyclization. Although
the reactions of allenes substituted with these auxiliaries led to
cyclopentenones in good yield and in good enantiomeric excess
(up to 82% ee, eq 2), in many cases, there were two shortcom-
(8) However, see: Alexakis, A.; Mangeney, P.; Normant, J. F. Tetrahedron
Lett. 1985, 26, 4197.
(9) Schultz-Fademrecht, C.; Wibbeling, B.; Fro¨lich, R.; Hoppe, D. Org. Lett.
2001, 3, 1221 and references therein.
(10) Schultz-Fademrecht, C.; Tius, M. A.; Grimme, S.; Wibbeling, B.; Hoppe,
D. Angew. Chem., Int. Ed. 2002, 41, 1532.
(11) Seebach, D. Angew. Chem., Int. Ed. Engl. 1988, 27, 1624.
(12) Ichikawa, J.; Fujiwara, M.; Okauchi, T.; Minami, T. Synlett 1998, 927.
9
10092 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002

Asymmetric Cyclopentannelation
A R T I C L E S
Table 1. Optimization of Yield and Ee of Ent-8a
inadvertently produce samples of 11 that are highly enriched
(>96% ee) in the major enantiomer by allowing crystallization
to take place during workup or purification.
entry
solventb
ethanol
HFIP
HFIP
HFIP/TFE (1/1)
HFIP/TFE (1/1)
MeNO2
HFIP
THF
acid(equiv)/temp(°C)
% yield (ee) of ent-8
1
2
3
4
5c
6
7
8
9
HCl(11)/-78
HCl(47)/0
56 (64)
59 (80)
36 (84)
53 (85)
60 (86)
64 (64)
67 (64)
58 (66)
67 (69)
HCl(41)/-40
HCl(44)/-78
HCl(39)/-78
HCl(44)/-40
TFA(39)/-40
ZnCl2(2)/-78
FeCl3(2)/-78
The results that are summarized in Table 2 allow several
generalizations to be made. Entries 1-4 indicate that aryl
substitution at one or at both C-3 and C-4 of the cyclopentenone
is tolerated. The difference in the optical purity of 13 and 15
follows the same trend that we described for the D-glucose-
derived auxiliary and in all likelihood has the same origin.3 Since
the only difference in the two substrates 12 and 14 is the
presence in 14 of the 4-methoxy group on the aromatic ring,
one must attribute the erosion in the ee of 15 relative to that of
13 to its presence. The additional stabilization of the protonated
ring-opened intermediate (i.e., compound 2 in eq 1) by electron
pair donation by the oxygen atom through the phenyl ring
renders the cyclization reaction reversible. This, in turn, can
compromise the stereospecificity by allowing competing cy-
clization processes to take place, possibly involving the geo-
metrical isomer of 2 (eq 1, R2 and R3 anti).
Entries 5, 6, and 13 of Table 2 show that halogen is also
tolerated. The halogenated cyclopentenones 19 and 21 did not
separate on the Chiralcel-OD or Chiralcel-OJ columns due to
extensive tailing of the peaks; however, derivatizing the
hydroxyl made it possible to separate the enantiomers. In the
case of 19, exposure to trimethyloxonium tetrafluoroborate and
Hu¨nig’s base led to rapid (<5 min) conversion to the methyl
ether derivative for which baseline separation of the enantiomers
took place. In the case of 21, the derived benzoate was used
for the determination of optical purity by HPLC. Similar
difficulties were encountered during the analysis of bromocy-
clopentenone 35 (entry 13). In this case, conversion of 35 to
the acetate allowed the separation of enantiomers to take place.
Entries 7-12 summarize the synthesis of alkyl substituted
cyclopentenones. In this series there is an improvement in the
ee of product with increasing branching at C-4, as a comparison
of the optical purities of 23, 25 and 27 reveals. It would appear
that there is little to be gained in terms of optical purity in going
from isopropyl (see 27) or from cyclohexyl (see 31) to a C-4
tert-butyl group (see 29).
CH2CI2
a
b
All reactions were performed in THF with 1.8 equiv of 9. The
volume of the quenching solvent was equal to the volume of the reaction
solvent, except for entry 5. The volume of HFIP/TFE was double the
volume of the reaction solvent.
c
(entries 8 and 9). This is an area that we must examine in greater
detail in the future.
The scope of the reaction of eq 3 was probed with the series
of enamides shown in Table 2.13 The yields of cyclic products
varied from 33% (Table 2, entry 17) to 84% (entries 3 and 10),
whereas the enantiomeric excess of products varied between
55% (entry 1) and 87% (entry 10). The absolute configuration
of the products shown in Table 2 has been assigned by analogy
with the result of eq 3. Since we converted ent-8 to natural
roseophilin4 and since Boger prepared the enantiomer of
roseophilin by an independent method, the assignment of
absolute stereochemistry in the case of ent-8 is secure.14 The
stereochemistry of 33 has also been determined rigorously, by
correlating it to a compound whose absolute stereostructure we
determined crystallographically.2 The fact that the major enan-
tiomer in all cases cited in Table 2, with the exception of entry
1, is also the less mobile of the two on the Chiralcel-OD HPLC
column is reassuring but does not constitute proof of stereo-
chemistry. It should therefore be stated clearly that the stere-
ochemistry of the products shown in Table 2 must be considered
tentative until confirmed through independent methods. This is
especially true for cyclopentenone 11. This compound is an
outlier for two reasons. First, because of the reversal in
chromatographic mobility of the major enantiomer, mentioned
above, and second, because it had the lowest ee of all cases in
Table 2. It is certainly conceivable that cyclization of the allene
adduct of enamide 10 takes place according to a mechanistically
distinct pathway. Compound 11 also illustrates another potential
pitfall, as it is exceptionally crystalline. It is quite easy to
Entries 14-17 show that substitution by trialkylsilyl at C-3
is also tolerated. Although the enantiomeric excesses of the
silylated products are in all cases good, the same is not true for
the isolated yield of products. The low yield of 37 is due to
protiodesilylation that takes place during exposure to concen-
trated strong acid. The difference in the yields of 41 and 43
(entries 16 and 17) is at first puzzling, since one would
reasonably expect a higher yield for 43, which is more robust
in the presence of acid. We believe that the discrepancy is due
to our failure to control contact times with acid during the
execution of the early parts of this work. We have since
modified our procedure such that the reactions are quenched at
(13) The addition-cyclization reaction sequence was repeated for each of the
20 enamides in Table 2 using allenyllithium species 50 to secure a sample
of the racemic product. Since the determination of optical purity of the
cyclopentenone product in each case was performed by chiral HPLC, it is
essential that a sample of the racemate be available. In the absence of a
chromatogram of the racemate as a reference, it is impossible to determine
whether separation of the enantiomers from the enantioselective reaction
has been accomplished. The measured peak areas for each of the
enantiomers in the racemic samples never deviated from 50% by more
than 1%. Therefore, one can conservatively estimate that the enantiomeric
excesses that are reported in Table 1 are accurate to within (2%.
(14) Boger’s result is essential, as there had been no independent confirmation
of the absolute stereochemical assignment of roseophilin prior to our total
synthesis. Boger, D. L.; Hong, J. J. Am. Chem. Soc. 2001, 123, 8515.
9
J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002 10093

A R T I C L E S
Harrington et al.
Table 2. Cyclizations with Terminally Unsubstituted Allenea
9
10094 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002

Asymmetric Cyclopentannelation
A R T I C L E S
Table 2 (Continued)
a
b
All reactions were performed according to the optimal conditions shown in entry 5 of Table 1. Determined by chiral HPLC of the derived methyl
c
d
e
ether. Determined by chiral HPLC of the derived benzoate. Determined by chiral HPLC of the derived acetate. Low yield due to protiodesilylation.
low temperature with bicarbonate. With these modified proce-
dures it should be possible to preserve the trialkylsilyl group in
the product.
Entries 18-20 show that the method can be used to prepare
fused carbocyclic rings, which adds to the versatility of the
process. The only fused heterocyclic ring that we examined,
dihydropyran 51, provided cyclic product of undetermined ee
in only 29% yield. It may be that the tetrahydropyranyl function
in 51 provides additional pathways for acid-catalyzed decom-
position of the adduct with 9.
from substituted allenyllithiums related to 9. Our first target
was methylallene 54 (eq 5). When the allene is stereogenic, as
it is in 54, one must consider matched and mismatched cases
(15) A limitation of all versions of the cyclopentannelation reaction is the
requirement of a non-hydrogen substituent at the R-carbon atom of the
acyclic intermediate (R3 * H in 1, eq 1). The reason for this is not
completely obvious. We have postulated that when R3 * H the reactive
U-shaped conformer of the pentadienyl cation 2 is favored for steric
reasons.1e We were therefore not surprised that the attempted cyclopen-
We examined a single example of a â,â-disubstituted mor-
pholino enamide (eq 4). The addition reaction of 52 with 9,
followed by exposure of the adduct to acid under the conven-
tional reaction conditions, produced cyclopentenone 53 in 14%
tannelation reaction of morpholino enamide i failed to produce the desired
product. The major products that were isolated from the reaction were
chloroenones ii and iii. It is likely that ii is derived from iii. Both these
products suggest that the high concentration of chloride during the
cyclization reaction may provide a competing reaction pathway that may
erode the yield of cyclic product in some cases.
yield, along with a large amount of recovered amide 52. The
low yield in this case is apparently a consequence of proton
transfer from the â-methyl group to 9. Although the ee of 53 is
not particularly high, this example is significant in that it
demonstrates that the asymmetric cyclopentannelation reaction
is suitable for the synthesis of cyclopentenones bearing qua-
ternary carbon atoms. The cyclization reactions of tetrasubsti-
tuted enamides and esters will be interesting to study, since they
potentially offer a solution to a difficult problem in synthesis.15
The limitations that have been cited above notwithstanding,
the asymmetric cyclopentannelations of 9 are preparatively
useful. Our next goal was to determine whether useful yields
of enantiomerically enriched cyclic products could be obtained
9
J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002 10095

A R T I C L E S
Harrington et al.
Table 3. Cyclizations of Allene Terminally Substituted with a
separately. When we initiated this phase of the work, we had
no indication of whether the auxiliary or whether the stereogenic
allene would exert the dominant influence on the absolute
stereochemical course of the cyclization. Certainly the alkyl
substituent on the allene significantly affects the stereochemical
course of the cyclization. For example, we had shown in earlier
work that in the absence of auxiliary, a n-butyl group on the
γ-carbon of an enantiomerically enriched allene leads to 84%
transfer of asymmetry during the cyclization.2 Butynyl ether
56 (eq 5) was prepared in high yield by trapping the corre-
sponding lithium acetylide with iodomethane. Exposure of 56
to n-butyllithium in THF at -78 °C, followed by quenching
the reaction mixture at -78 °C with a solution of tert-butyl
alcohol in THF led to a ca. 20/1 mixture of diastereomers 54
and 55. We were able to tentatively assign the stereochemistry
of the major diastereomer through two experiments. In the first
experiment, the 20/1 mixture of 54 and 55 was deprotonated
with n-butyllithium to produce lithioallene 62 and its diastere-
omer. These were allowed to react with morpholino enamide
12 (entry 2, Table 3), and cyclization was effected in the usual
way, with HCl in HFIP/TFE at -78 °C. This led to cyclic
product 58 in 67% yield and in 79% ee.
Methyl Groupa
Diastereomers 54 and 55 were not completely separable by
flash column chromatography, however, a ca. 3/4 mixture of
the two could be obtained by combining chromatography
fractions. The addition-cyclization sequence with enamide 12
was repeated with the mixture of allenes that was enriched with
the minor diastereomer. The yield of 58 from this experiment
was 67% (virtually unchanged), but the ee fell from 79 to 71%.
Since the reproducibility of the measurement of ee by chiral
HPLC is (2%,13 this is a significant difference, and the erosion
in the ee of 58 must be the result of a mismatched reaction
proceeding through 55. The axial chiralities of allenes 54 and
55 favor conrotation in opposite directions. Although we had
not predicted that the matched diastereomer 54 would predomi-
nate in the isomerization step (eq 5), this observation can be
understood as the result of two diastereoselective proton
transfers. Exposure of 56 to n-butyllithium results in the selective
abstraction of the pro-S propargylic proton, leading to internally
chelated structure 63 (eq 6). Removal of the pro-R proton would
have placed the 1-propynyl group in close proximity to the bulk
of the auxiliary and is apparently disfavored for this reason.
The protonation of 63 by tert-butyl alcohol must also take place
stereoselectively, as shown in eq 6. The anti stereoselectivity
that has been postulated for this process is plausible. The
hypothesis that is summarized in eq 6 also postulates a role for
the auxiliary that is consistent with its effect on the stereo-
chemical course of the cyclization. This is shown for the
matched case in eq 7. The adduct of lithioallene 62 and enamide
12, following acid-catalyzed collapse of the tetrahedral inter-
mediate, leads to 64. Electron pair donation by the pyranyl
oxygen atom of the auxiliary stabilizes the pentadienyl cation
and restricts it to the conformation indicated in the structure.
Conrotation in 64 is biased to take place through a counter-
clockwise motion, as shown, for two reasons. First, counter-
clockwise rotation moves the phenyl ring in 64 away from the
steric bulk of the auxiliary. Second, conrotation in this direction
moves the methyl group on the allene away from the phenyl.
Under the reaction conditions cyclic cation 65 is probably
formed irreversibly. Loss of the chiral auxiliary as a stable cation
a
All reactions were performed according to the optimal conditions
shown in entry 5 of Table 1 starting with the ca. 20/1 mixture of 54 and
55. Determined by chiral HPLC of the derived benzoate.
b
66 leads to cyclopentenone 67 as the kinetic product, which
undergoes acid-catalyzed isomerization in situ to give E isomer
58 as the final product.16
The results of Table 3 show that the yield and ee of all cyclic
products were good. For the sake of convenience, isomerization
to the thermodynamically favored E isomers was allowed to
take place, and data have been recorded for the E isomers only.
With the exception of the result of entry 4, the ee values of all
products are as good or better than the corresponding reactions
of the same enamides with allenyllithium 9. This is what one
would predict for the products of the reaction in which auxiliary
and allene chirality are matched. These results also suggest that,
if one were to replace the methyl group in allene 54 by a larger
(16) Note that “clockwise” conrotation in 64 will produce a diastereomer of 67
in which the exocyclic double bond is E. Therefore, Z-to-E isomerization
of the product in situ may erode the observed ee.
9
10096 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002

Asymmetric Cyclopentannelation
A R T I C L E S
Table 4. Cyclizations of Allene Terminally Substituted with a
tert-Butyl Groupa
alkyl group, much higher enantioselectivities would be expected
for the cyclic products. This is exactly what we have observed.
a
All reactions were performed according to the optimal conditions
b
shown in entry 5 of Table 1. Determined by chiral HPLC of the derived
benzoate.
71 as the exclusive product. Because the signals in the 1H NMR
spectra at 300 MHz for the propargylic methylene and the
camphor-derived portions of 71 and 56 deviated by less than
0.1 ppm in all cases, the stereochemistry of 71 was assumed to
be as shown in eq 8. Isomerization of propargyl ether 71 to
allenyl ethers 72 and 73 took place under the same conditions
that were used for 56. The major allene diastereomer 72 was
isolated in 65% yield, along with 27% of a mixture of allene
73 and propargyl ether 71. The approximate ratio of 72 to 73
was 3/1. In contrast to 54 and 55, diastereomers 72 and 73 were
easily separable by flash column chromatography. That major
The γ-tert-butyl allene 72 was prepared according to the
procedure that is summarized in eq 8. Lactone 68 was reduced
to lactol 69. Alkylation of 69 with 2 equiv of propargylic
bromide 7017 under phase-transfer conditions produced R-isomer
(17) Brandsma, L. PreparatiVe Acetylenic Chemistry, 2nd ed; Elsevier: New
York, 1988; pp 248-249.
9
J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002 10097

A R T I C L E S
Harrington et al.
isomer 72 has the stereochemistry indicated in the structure
shown in eq 8 can be shown by the following two experiments
(eqs 9 and 10).
To probe the scope of this process a little further, the
experiments that are summarized in eq 11 were carried out.
Enamide 12 was allowed to react with methyllithium to give
ketone 82. Addition of the lithio allene derived from 72 led to
a mixture of diastereomeric tertiary alcohols 83 that were
separable by flash column chromatography. Pure samples of
the two diastereomers were cyclized independently. Cyclization
of the chromatographically less mobile diastereomer led to a
1:1 mixture of 84 and 85, whereas cyclization of the more
mobile diastereomer led almost exclusively to 84 along with a
small amount (<5%) of a Z diastereomer of unknown absolute
configuration.18 These results suggest that in the case of tertiary
alcohols such as 83, the direction of conrotation is strongly
influenced by the stereochemistry of the tertiary alcohol. This
is somewhat surprising, since it clearly shows that formation
of the carbon-carbon bond takes place at least in part before
the ionization of the tertiary, bis-allylic alcohol has occurred.10
For the cyclization of tertiary alcohols such as 83, boron
trifluoride etherate is usually the optimal catalyst;19 however,
we have not performed a completely comprehensive correlation
of catalysts, yields, and enantiomeric excesses for the variants
of the allene cyclopentannelation. At the present stage of
development of the methodology, the enantioselective cycliza-
tion of tertiary alcohols is not preparatively useful.20,21
Major allene diastereomer 72 was deprotonated, and the lithio
allene was allowed to react with morpholino enamide 22 (eq
9). Acid-catalyzed cyclization, as before, provided (Z)-cyclo-
pentenone 76 along with a small amount of E isomer 79.
Products 76 and 79 were isolated in 94 and 92% ee, respectively,
suggesting that the reaction of eq 9 corresponds to the matched
case. The experiment was repeated with a 9/1 mixture of allenes
enriched in minor diastereomer 73 (eq 10). Cyclopentenones
80 and 81 were isolated from this experiment in lower yield
and lower ee, 72 and 61% ee for 80 and 81, respectively. The
absolute stereochemistry at C-4 in products 80 and 81 is the
opposite of that of products 76 and 79. This proves that the
reaction of eq 10 corresponds to the mismatched case, that the
reaction of eq 9 corresponds to the matched case, and that the
effect of the allene stereochemistry in the case of 73 completely
overwhelms that of the auxiliary. By isomerizing 76 to 79 and
comparing the chiral HPLC chromatograms it was shown that
76 and 79 have the same stereochemistry at C-4. The same
experiment was done for 80 and 81. The minor peak in the
chromatogram of 79 was the major peak in the chromatogram
of 81.
As a final illustration of some of the applications to synthesis
that may follow, we performed the reactions that are summarized
in eq 12. Cyclopentenone 75 was converted to allyl ether 86 in
90% yield, and then heated to 190 °C in o-dichlorobenzene for
30 min. This led to R-diketone 87 in 70% isolated yield,22
isomerization of the exocyclic double bond having taken place
during the Claisen reaction. Attempts to catalyze the Claisen
Table 4 summarizes the results of the cyclizations of 72 with
representative morpholino enamides. In this work, the acid was
neutralized at low temperature by the addition of saturated
aqueous sodium bicarbonate. This modification of our normal
procedure allows the kinetic product bearing the Z geometry of
the exocyclic double bond to be isolated. Table 4 reflects this
modified procedure. The enantiomeric excesses of the products
are all excellent, and range from 92 to 96% ee.
23
24
process either with PdCl2(PhCN)2 or with Ho(fod)3 failed.
(18) The overall yield of 84 (95% ee) for the two steps from 82 was 67%. Z
diastereomer 85 was isomerized to ent-84 (93% ee) in 22% overall yield
from 82 by exposure to sunlight in benzene containing catalytic PhSSPh.
(19) See ref 13 in: Harrington, P. E.; Li. L.; Tius, M. A. J. Org. Chem. 1999,
64, 4025.
9
10098 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002

Asymmetric Cyclopentannelation
A R T I C L E S
Product 87 was isolated as a single diastereomer. The assignment
of stereochemistry of the quaternary carbon atom was made on
the basis of the positive nOe between the benzylic methine
proton and the allylic methylene protons. The positive nOe
between the tert-butyl group and the benzylic methine and
phenyl protons shows that the geometry of the exocyclic double
bond is E.
A question that we had early in our investigation was whether
the cyclization would prove to be suitable for the asymmetric
construction of quaternary carbon atoms â to the enolic hydroxyl
group. Our preliminary results (eq 4) suggest that this is the
case, however, improvement of the reactivity of the morpholino
enamide, the electrophilic partner in the addition step, will be
necessary in order to fully exploit the potential of this process.
One can use the method for the asymmetric construction of
quaternary carbon atoms through the Claisen reaction shown
in eq 12. Trost has demonstrated excellent transfer of asymmetry
in a related system.24
There appear to be (at least) two mechanistic pathways for
the cationic cyclization. In the case of the morpholino enamides,
the stereochemical outcome of the reactions is consistent with
a thermally allowed conrotation (eq 7). By contrast, enones
apparently belong to a different manifold of reactions in which
the stereochemistry of the tertiary alcohol adduct influences the
stereochemical course of the cyclization (eq 11). Since we now
appear to have an understanding of the mechanism of the
asymmetric cyclopentannelations, improvements in auxiliary
design can be contemplated.25
Conclusions
The scope of an asymmetric cyclopentannelation reaction that
we recently developed for a synthesis of natural roseophilin has
been defined. Yields and enantiomeric excesses of cyclic
products are generally good in the case of the terminally
unsubstituted allene (Table 2). Moreover, a wide variety of
carbon and heteroatomic substituents are tolerated, suggesting
broad synthetic utility. In the cases in which the allene bears a
terminal substituent one must consider matched and mismatched
cases, since allene chirality as well as chiral auxiliary affect
the stereochemical outcome of the cyclization. We were
fortunate that the isomerization of propargyl ether to allenyl
ether was controlled by the auxiliary so as to produce a
preponderance of matched products 54 and 72 (eqs 5 and 8).
The higher enantiomeric excesses of the products of Tables 3
and 4 versus those of Table 2 reflect the chiral auxiliary and
axial chirality of the allene working in concert. The effect of
the large tert-butyl substituent in allene ether 72 leads to the
highest enantiomeric excesses we have been able to achieve to
date (Table 4). A challenge for the future will be to replace the
tert-butyl group with an easily cleavable sterically demanding
group.
Experimental Section
Representative Procedure. Preparation of 29. To a solution of
allene H-9 (115 mg, 0.544 mmol) in THF (3 mL) at -78 °C was added
n-BuLi (225 µL, 2.46 M in hexanes, 0.554 mmol). After 30 min, a
solution of amide 28 (84 mg, 0.40 mmol) in THF (3 mL) at -78 °C
was added via cannula. The reaction mixture was warmed from -78
to -35 °C over 1 h, cooled to -78 °C, and quenched by rapid addition,
through a large bore cannula, to HCl in HFIP/TFE (generated by the
addition of 750 µL of acetyl chloride to a mixture of 3 mL of HFIP
and 3 mL of TFE) at -78 °C. The flask was removed from the cooling
bath, warmed to room temperature, and diluted with saturated NaHCO3,
pH 7 buffer, brine, and EtOAc. The aqueous phase was extracted with
EtOAc (3×), and the combined organic extracts were washed with brine
(1×) and dried over MgSO4. Purification by flash column chromatog-
raphy on silica gel (5-10% EtOAc in hexanes) gave cyclopentenone
29 (60 mg, 84% yield, 87% ee) as a white solid: mp 97-98 °C; Rf )
0.39 (20% EtOAc in hexanes); 1H NMR (300 MHz, CDCl3) δ 6.04 (s,
1H), 5.99 (s br, 1H), 5.31 (s, 1H), 2.95 (s br, 1H), 2.09 (d, J ) 1.0 Hz,
3H), 0.97 (s, 9H); 13C NMR (75 MHz, CDCl3) δ 189.5, 152.3, 143.5,
142.3, 116.3, 54.6, 35.1, 28.9, 15.7; IR (neat) 3310 (br), 2960, 1685,
1620, 1405, 1360, 1195, 1105 cm-1; EIMS m/z 125 (9), 124 (100), 95
(7); HREIMS calcd for C11H16O2 180.1150, found 180.1280; chiral
HPLC (5% 2-propanol in hexanes, Chiralcel OD, 250 mm × 10 mm,
254 nm, 1 mL/min) tR ) 22.1 min (major), tR ) 24.7 min (minor).
The selective isomerization of propargyl ethers to allene ethers
of defined axial stereochemistry has not been examined in detail.
It may be possible through manipulation of reaction conditions
to improve the diastereomeric ratio of products. This would be
significant, as there is currently no general method for the
preparation of axially chiral allenyl ethers.8
Acknowledgment. We thank the National Institutes of Health
(GM57873) for generous support, and Dr. Oliver Weichold for
the synthesis of the carboxylic acids that were used for the
(20) For example, cyclization of i, the adduct of 9 with enone 82, led to
cyclopentenone ii in good yield, however the ee was only 31%.
(25) A reviewer raised the question of potential racemization of the final
products. Circumstantial evidence strongly suggests that if any racemization
takes place, this is a very slow process under the reaction conditions. First,
our measurement of the enantiomeric excess of each of the reaction products
has been invariant, within the confidence limits of our measurement, over
two or more runs. As mentioned in ref 13 above, our uncertainty in the
HPLC measurements is no more than (2%. Second, since we did not take
great pains to reproduce the times of exposure of the intermediate products
to the strongly acidic conditions for the cyclization process, if acid-catalyzed
racemization were taking place at an appreciable rate, it is very likely that
we would have detected this in at least one of the examples. For example,
the reaction of eq 9 was warmed to room temperature in the presence of
HCl/HFIP/TFE, and product 76 was isolated with 94% ee. The same product
is reported in Table 4, entry 3, from a reaction that was quenched at -78
°C in which the ee of 76 was 92%. There is no difference in the
enantiomeric excess of the product from the two reactions, within the
uncertainty of the measurement.
(21) The results of eq 11 apparently contradict earlier observations made in our
group.2 It may be possible to reconcile the two sets of results if ionization
of the tertiary benzylic alcohol derived from (-)-16 (Scheme 3 in ref 2) is
complete prior to the development of the C3-C4 bond in (-)-17.
(22) Ponaras, A. A. Tetrahedron Lett. 1980, 21, 4803.
(23) Hiersemann, M. Synlett 1999, 1823.
(24) Trost, B. M.; Schroeder, G. M. J. Am. Chem. Soc. 2000, 122, 3785.
9
J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002 10099

A R T I C L E S
Harrington et al.
preparation of enamides 46 and 48. We thank Callery Chemical
Co. for a generous gift of potassium tert-butoxide.
33, 35, 37, 39, 41, 43, 45, 47, 49, 53, 56-61, and 74-78, (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
Supporting Information Available: Experimental procedures
and spectroscopic data for 54, 71, 72, 84, 86, 87, and ii.
Spectroscopic data for 11, 13, 15, 17, 19, 21, 23, 25, 27, 31,
JA020591O
9
10100 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002