selectivity is expected to be high, leading to the C(3)-C(15)
cis stereoisomer. The calculations also indicate a preference
for an equatorial orientation of the substituent at N(4) with
respect to the forming D ring (8a and 8c vs 8b and 8d).
However, there is negligible endo/exo selectivity predicted
for TSs with an equatorial N(4) substituent (8a vs 8c). In
contrast, a significant preference for the endo pathway is
predicted among TSs with an axial N(4) substituent (compare
8b and 8d).
The synthesis began with the preparation of N-acetyl
tetrahydro-â-carboline 6 via the acyl-Pictet-Spengler
reaction.5a Condensation of tryptamine (9) with aldehyde
10,18 and treatment of the resulting imine 7 with acetyl
chloride and 2,6-lutidine in the presence of thiourea catalyst
11 (10 mol %) afforded 6 in 81% yield and 94% ee on gram
scale. Deacetylation of the amide was then accomplished by
treatment of 6 with lithium amidotrihydroborate,19 providing
enantioenriched tetrahydro â-carboline 12 in 74% yield.
The C(17)-C(21) diene side-chain was then installed in
one step via a reductive amination. Thus, treatment of 12
with aldehyde 13,20 HOBz, and NaBH3CN in benzene
afforded amine 14 in 55% yield. Protection of the indole
nitrogen by treatment of 14 with Cbz-Cl and KHMDS
afforded the corresponding N-Cbz indole in 92% yield.
Subsequent removal of the TBDPS group with TBAF gave
the corresponding alcohol in 85% yield.21 Oxidation of this
alcohol with SO3‚pyridine22 and treatment of the resulting
aldehyde with Ph3PdCHCO2Me provided IMDA substrate
4b in 79% yield over the two steps.
In contrast, computations of the IMDA reaction of model
substrate 5 predict a modest preference for boatlike rather
than chairlike TSs (Figure 2), leading to cycloadducts with
The IMDA reaction of triene 4b promoted by 4 equiv of
Sc(OTf)3 in CH3CN proceeded with unexpectedly high
selectivity, affording the cycloadduct 15 as a single diaste-
reomer in 87% yield. The relative configuration of 15 at C(3),
C(15), and C(20) is consistent with reaction through a cis,
endo TS analogous to 8c or 8d (Figure 1). Removal of both
the N-Cbz and C(17)-OBz protecting groups was then
accomplished by exposure to Cs2CO3 in MeOH/THF, giving
the corresponding alcohol in 80% yield. Finally, hydrogena-
tion of the C(18)-C(19) olefin yielded (+)-yohimbine (1)
in quantitative yield. Synthetic (+)-1 was identified by
Figure 2. Lowest energy IMDA transition structures using model
substrate 5.
C(3)-C(15) trans configuration. Endo TSs, which generate
a trans ring fusion, are significantly preferred over exo TSs
(see Supporting Information), presumably due to a preference
for the amide diene to adopt a C(20)-C(21) s-cis conforma-
tion.15 These calculations are consistent with experimental
results reported previously with related substrates.8,16
Despite the poor diastereoselectivity predicted for the
thermal cyclization of 4a (Figure 1, TSs 8a vs 8c), we
decided to investigate substrates akin to 4 for the synthesis
of (+)-yohimbine. The C(3) stereogenic center could be set
via a catalytic asymmetric acyl-Pictet-Spengler condensa-
tion, and this would in turn serve to set the configuration at
C(15) via TS 8a or 8c in the IMDA reaction. Control over
the C(20) stereocenter depends on the endo/exo selectivity
in the IMDA reaction, and we reasoned that this might in
principle be achieved by a chiral catalyst.17
1
comparison to a sample of natural (+)-1 by H NMR, 13C
NMR, and IR spectroscopy, as well as by high-resolution
MS and optical rotation.
Although the high C(3)-C(15) cis selectivity of the IMDA
reaction could be anticipated, the high endo/exo selectivity
was not predicted from our computations. The high dr is
not entirely attributable to the presence of a Lewis acid, as
thermally induced cyclization of 4b (70 °C, benzene)
provided a 6:1 mixture of endo/exo cyclodducts (∆∆Gq )
1.2 kcal/mol). Interestingly, N-unprotected analogues (i.e.,
N-H indoles) were found to undergo poorly selective
thermally induced cyclization (2-3:1 endo/exo selectivity
at 23 °C, ∆∆Gq ) 0.4-0.6 kcal/mol). To better understand
the role of the indole protecting group in enhancing diaste-
(13) B3LYP has been shown to reproduce experimentally observed
kinetic isotope effects in Diels-Alder reactions, suggesting that this method
yields accurate TS geometries: (a) Beno, B. R.; Houk, K. N.; Singleton,
D. A. J. Am. Chem. Soc. 1996, 118, 9984-9985. Although B3LYP usually
yields accurate endo/exo and dienophile facial selectivities, the MP2 method
has been suggested to be more accurate for calculating endo/exo selectivities
in some cases: (b) Bakalova, S. M.; Santos, A. G. J. Org. Chem. 2004, 69,
8475-8481. However, MP2 Diels-Alder TS geometries have been shown
to be significantly distorted from B3LYP TS geometries. Using substrate
4a, MP2 predicts dienophile facial selectivities that are inconsistent with
experiment, whereas B3LYP predicts a significant preference for the
experimentally observed isomer (Vide infra). A comparison of results using
B3LYP and MP2 with various basis sets is provided in the Supporting
Information.
(14) Calculations were carried out using Gaussian 98: Frisch, M. J.; et.
al. Gaussian 98; Gaussian, Inc.: Pittsburgh, PA, 2002.
(15) For discussions of conformational requirements of carbonyl dienes
in IMDA reactions, see: (a) Boeckman, R. K.; Demko, D. M. J. Org. Chem.
1982, 47, 1789-1792. (b) Martin, S. F.; Williamson, S. A.; Gist, R. P.;
Smith, K. M. J. Org. Chem. 1983, 48, 5170-5180.
(17) For previous efforts from this laboratory directed toward natural
product synthesis employing catalyst-controlled diastereoselective Diels-
Alder reactions, see: (a) Joly, G. D.; Jacobsen, E. N. Org. Lett. 2002, 4,
1795-1798. (b) Chavez, D. E.; Jacobsen, E. N. Org. Lett. 2003, 5, 2563-
2565. (c) Boezio, A. A.; Jarvo, E. R.; Lawrence, B. M.; Jacobsen, E. N.
Angew. Chem., Int. Ed. 2005, 44, 6046-6050. (d) Balskus, E. P.; Jacobsen,
E. N. Science 2007, 317, 1736-1740.
(18) Smith, A. B., III; Safonov, I. G.; Corbett, R. M. J. Am. Chem. Soc.
2002, 124, 11102-11113.
(19) Myers, A. G.; Yang, B. H.; Chen, H.; McKinstry, L.; Kopecky, D.
J.; Gleason, J. L. J. Am. Chem. Soc. 1997, 119, 6496-6511. A modified
version of the reported procedure was used. See the Supporting Information
for details.
(20) Becher, J. Org. Synth. 1980, 59, 79-83.
(16) For other discussions of boatlike transition structures in IMDA
reactions, see: (a) Coe, J. W.; Roush, W. R. J. Org. Chem. 1989, 54, 915-
930. (b) Tantillo, D. J.; Houk, K. N.; Jung, M. E. J. Org. Chem. 2001, 66,
1938-1940.
(21) Trifluoroethanol was used as solvent for this transformation to
suppress transfer of the N-carbobenzyloxy group to the liberated alcohol.
(22) Parikh, J. R.; Doering, W. v. E. J. Am. Chem. Soc. 1967, 89, 5505-
5507.
Org. Lett., Vol. 10, No. 5, 2008
747