Journal of the American Chemical Society
Communication
dienamines 23a−c and the downstream cyclohexene structures
24a−f with complete diastereocontrol upon reaction with N-
phenylmaleimide or other dienophiles in good to excellent
yields. Moving away from simple but-3-enamides, indole
substrate 25a,b, where the β,γ-unsaturation is an integral part
of the heteroaromatic ring, also produced the desired
cycloadducts 26a,b. For ease of isolation, these were further
oxidized by addition of DDQ at the end of the reaction and
isolated as the aromatized β-carbolines 27a and 27b in 77%
and 89% yield, respectively. Finally, both amide functional
groups within succinamide 28 could be reduced to their
respective enamine intermediates, forming overall a symmetric
bisamino-diene species 29 that underwent cycloaddition to
furnish symmetric tetrasubstituted 30 as a single isomer.
Remarkably, during the course of this reaction, all six carbons
contained within the final cyclohexene product saw their
hybridization state change from sp3 to sp2 (or vice versa),
resulting in a relatively complex architecture arising in a single-
pot transformation from a simple building block.
Scheme 4. (a) Dehydrosecodine at the Center of the
Monoterpene Indole Alkaloid Biosynthesis; (b) A New
Total Synthesis of Catharanthine
To firmly establish this reductive dienamine generation
strategy in complex natural product total synthesis, we set our
sights on one of the most important yet elusive intermediates
in monoterpene indole alkaloid natural products chemistry,
dehydrosecodine (20). Since the pioneering studies of
Wenkert in 1962,17 Scott,18a and recently De Luca18b and
O’Connor,18c−e this functionally rich molecular entity has been
putatively identified as the common precursor to a wide variety
of skeletally varied Vinca, Iboga, and Aspidospema alkaloids.18f
Possessing a 1,2-dihydropyridine motif capable of meeting
either the electronic demands of a diene (normal electron
demand Diels−Alder cycloaddition toward catharanthine 3;
see Scheme 4a) or a dienophile (inverse electron demand
Diels−Alder cycloaddition toward tabersonine 31),19 dehy-
drosecodine (20) has remained elusive due to its high
reactivity and inherently redox-sensitive functionalities, in
particular 1,2-dihydropyridine and indole-2-acrylate.18e,20 Not
unsurprisingly, nature’s way has inspired the approaches of
many synthetic chemists over the years;21 in fact, more than
half of the total and formal syntheses of catharanthine
published to date have indeed relied on a Diels−Alder
approach to the isoquinuclidine core.21a−n Interestingly,
however, not one proceeded directly via dehydrosecodine.
This is partly due to the difficulty of accessing the 5-ethyl-
substituted 1,2-dihydropyridine motif (because of undesired
regioselectivity in the reduction of pyridinium ions; see
Scheme 1d), particularly in the presence of the sensitive/
reactive indole-2-acrylate fragment.20
Recognizing that our reductive strategy offers reliable
regiocontrol in 1,2-dihydropyridine synthesis, as well as
notable and well-documented chemoselectivity for the
reduction of the lactam carbonyl over other functional groups,
including alkenes, we set on a journey to access catharanthine
(3) via its elusive biosynthetic precursor dehydrosecodine
(20).
Our synthesis began with the formation of the α-substituted,
β,γ-unsaturated δ-lactam 35 in a two-step sequence from
commercially available starting materials (Scheme 4b). At high
temperatures, tryptamine (33) and dihydropyrone (34)
reacted to form the unsaturated lactam as a mixture of
constitutional isomers in 51% yield.22 Subsequent double
deprotonation of the mixture with 2 equiv of LDA and α-
alkylation with ethyl iodide resulted in the formation of desired
35 in 83% yield. After extensive investigations (see the
Stephenson’s photoredox-catalyzed C2-functionalization of
unprotected indoles,23 we were able to introduce a
phosphonoester group at the C2 position of indole 35,
resulting in isolation of 37 in 54% yield. The phosphonoester
37 could in turn be used to install the terminal methylene
group of 38 via the Rathke modification of the Horner−
Wadsworth−Emmons reaction by using paraformaldehyde, in
83% yield.24,25
Having established a four-step route to the precursor of
dehydrosecodine 20, the stage was set for the final reductive [4
+ 2] cycloaddition sequence. Pleasingly, upon submission of
38 to the newly developed reaction conditions, catharanthine
(3) was indeed produced, albeit in trace amounts as
determined by 1H NMR analysis of the crude reaction
mixture. Extensive optimization of the reductive activation
step led to an improved isolated yield (11%) of 3 when TMDS
was slowly added to a solution of precursor 38 and Vaska’s
complex, thus completing the fully biomimetic total synthesis
of the alkaloid and establishing the intermediacy of its evasive
and intriguing biosynthetic precursor, dehydrosecodine.
Efforts to isolate byproducts in the final reaction, to
understand the low mass return, were unfruitful. Consequently,
the reaction was performed in deuterated solvent in an NMR
tube, in the hope of observing transient species.26 Upon slow
addition of TMDS to a solution of 38 and Vaska’s complex in
d8-toluene, catharanthine was immediately produced in 15%
NMR yield, alongside reduced species 40 (85% NMR yield, as
a mixture of isomers at the dihydropyridine), arising from the
apparent hydridic reduction of the indole-2-acrylate in
dehydrosecodine (20) (Scheme 5).27 Attempted purification
via flash column chromatography on silica gel failed to provide
10831
J. Am. Chem. Soc. 2021, 143, 10828−10835