A concise total synthesis of (+)-scholarisine a empowered by a unique C-H arylation

Article abstract of DOI:10.1021/ja406546k

The structurally unique akuammiline alkaloid (+)-scholarisine A was synthesized in 14 steps from a known enone (15 steps from commercial materials) through a route empowered by a unique C-H arylation reaction to forge its polycyclic core. Additional key steps include a pyrone Diels-Alder reaction and a radical cyclization/Keck allylation to fashion the core cage polycycle and one of the molecule's quaternary centers, as well as the use of a carefully positioned pendant hydroxyl group to facilitate the chemoselective reduction of an extremely unreactive lactam in the presence of a readily reduced lactone.

Full text of DOI:10.1021/ja406546k

Communication
pubs.acs.org/JACS
A Concise Total Synthesis of (+)-Scholarisine A Empowered by a
Unique C−H Arylation
Myles W. Smith and Scott A. Snyder*
Department of Chemistry, Columbia University, Havemeyer Hall, 3000 Broadway, New York, New York 10027, United States
S
* Supporting Information
Scheme 1. Unique Structure of Scholarisine A (1), Past
Synthetic Approaches, and a Distinct Retrosynthetic
Analysis of the Target Molecule
ABSTRACT: The structurally unique akuammiline alka-
loid (+)-scholarisine A was synthesized in 14 steps from a
known enone (15 steps from commercial materials)
through a route empowered by a unique C−H arylation
reaction to forge its polycyclic core. Additional key steps
include a pyrone Diels−Alder reaction and a radical
cyclization/Keck allylation to fashion the core cage
polycycle and one of the molecule’s quaternary centers,
as well as the use of a carefully positioned pendant
hydroxyl group to facilitate the chemoselective reduction
of an extremely unreactive lactam in the presence of a
readily reduced lactone.
onoterpenoid indole alkaloids have a long and storied
Mhistory in treating disease, elucidating core biochemical
pathways, and advancing the power of synthetic chemistry
through efforts to fashion and derivatize their structural
intricacies. One recent addition to the family is the akuammi-
line alkaloid scholarisine A (1, Scheme 1), a dense, polycyclic
compound isolated from Alstonia scholaris, a plant used in
traditional Chinese medicine to treat various respiratory
diseases.1 Its structure is quite unique within the class2 given
that it possesses an indolenine fused to a strained carbocyclic
cage, itself adorned with several tertiary and quaternary
stereocenters as well as potentially labile imine and bridging
lactone functional groups; its specific biological properties,
however, remain unexplored. Given this profile, it is
unsurprising that chemists have been drawn to its challenges
and its potential. In 2012, the Smith group developed an
elegant, enantioselective total synthesis of (+)-1 which
proceeded in 20 steps from a known compound (25 steps
from commercial materials).3 Key elements of their approach,
which constitutes the only complete solution to this target, are
shown in Scheme 1. In addition, Higuchi and co-workers
recently reported model studies toward the spiroindolenine
portion of 1 using an oxidative coupling to forge that domain.4
Herein, as part of a program interested in rapidly accessing
unique polycyclic frameworks,5 we report a distinct analysis of
the synthetic challenges posed by scholarisine A. That approach
sought to develop a unique C−H arylation to attach the
indolenine domain directly to a near fully functionalized core,
itself prepared rapidly and efficiently through the use of Diels−
Alder chemistry and a radical cyclization/capture cascade.
Our retrosynthetic analysis of 1 is delineated in the lower
part of Scheme 1. Describing its logic in the forward sense, we
were guided by the general notion that the target’s core
Received: July 1, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja406546k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
a
carbocyclic substructure could potentially be fashioned rapidly
if we could access a material such as 10 with an
oxabicyclo[2.2.2]octane. The key event for its preparation
was envisioned to be an endo- and diastereoselective pyrone
Diels−Alder reaction,6 setting the stage for a tandem 6-exo-trig
radical cyclization/Keck allylation7 to then access 9. This
cascade operation was expected to form the quaternary center
with high diastereoselection via exo-face delivery of the allyl
group to an intermediate α-carbonyl radical. From here,
completion of the last ring of the caged core through
epimerization of the amine-bearing stereocenter and lactamiza-
tion would enable attempts to directly incorporate the
indolenine domain via imine formation (to form 7) followed
by a novel, late-stage tertiary C−H arylation reaction. Although
such processes are known in specific contexts to occur with
activated, enolizable carbons, to the best of our knowledge no
examples involving nonenolizable,8 and potentially labile,
imines are known.9 Moreover, as an additional challenge,
positional selectivity between two different tertiary centers (the
starred carbons within 7) would be required, assuming that the
initial imine stereochemistry would not be product-determin-
ing.10 However, if these variables could be addressed
successfully, then (+)-scholarisine A (1) could result from a
final chemoselective reduction of a hindered lactam in the
presence of a fairly accessible lactone. To aid in that task, the
functionality of the strategically positioned allyl side chain could
potentially prove useful.
Scheme 2. Preparation of the Core of Scholarisine A
As shown in Scheme 2, our synthetic efforts began with a
Diels−Alder reaction between dienophile 1111 (available in
decagram quantities in three steps from N-Boc-D-Serine, see
Supporting Information) and a slight excess (1.2 to 1.8 equiv)
of methyl coumalate (12) at 100 °C in toluene for 33 h. These
conditions afforded the desired product (13) alongside a minor
amount of its separable diastereomer 14 and another very
minor component of unknown structure (3:1:0.1), in 83%
combined yield. A priori, this cycloaddition would seem to be
unfavorable, given the electron-deficient nature of both
components; as such, it is notable that it proceeds with such
efficiency.12 Moreover, the protecting groups on the serine-
derived dienophile had a significant effect on both the yield and
diastereoselectivity of the reaction; for instance, the corre-
sponding O-TBS-protected congener of 11 afforded Diels−
Alder adducts in 47% yield and a dr of 1.7:1:0.1.
a
Reagents and conditions: (a) 12 (1.8 equiv), toluene, 100 °C, 33 h,
83% (13:14:unk = 3.0:1.0:0.1); (b) TFA (6% v/v), CHCl3, 0 °C, 12 h,
83%; (c) Ph3P (1.5 equiv), imidazole (1.5 equiv), Br2 (1.5 equiv);
CH2Cl2, 0 °C, 40 min, 86%; (d) allyltributylstannane (3.0 equiv), Et3B
(5 × 0.2 equiv), benzene, air, 75 °C, 5 h, 59%; (e) TMG (1.1 equiv),
TEMPO (1.1 equiv), air, THF, 50 °C, 12 h, 68%; (f) NaBH3CN (5.0
equiv), TFA (17.6 equiv), CH2Cl2, 0 °C, 40 min; (g) EtOAc, 80 °C, 2
h, 91% over two steps; (h) IBX (3.5 equiv), EtOAc, 80 °C, 12 h,
>85%. TFA = trifluoroacetic acid, TMG = tetramethylguanidine,
TEMPO = 2,2,6,6-tetramethyl-1-piperidinyloxy, free radical, IBX = o-
iodoxybenzoic acid.
With 13 in hand, we next set about transforming its
protected alcohol into an appropriate radical cyclization
precursor. That task was accomplished using TFA in
noticed that adventitious oxygen could readily oxidize this
compound to its corresponding unsaturated congener 17,
presumably due to the ease of forming a captodative radical
intermediate; an optimized process using TMG and TEMPO in
THF under air gave enone 17 in 68% yield.15 This outcome
presented the possibility of a deprotection/exo-face reduction/
lactamization sequence. Gratifyingly, treatment of a CH2Cl2
solution of 17 with NaBH3CN and TFA at 0 °C followed by
heating in EtOAc afforded the desired caged compound in
excellent yield (91%) with concomitant reduction of the ketone
to the endo alcohol (structure not shown). Although the latter
event was undesired, control experiments showed that ketone
reduction was necessary to prevent pyrazine dimer formation.
The constitution of the full cage structure was secured by
extensive 2-D NMR studies as well as X-ray analysis of a
crystalline derivative. Reoxidation with IBX then completed the
synthesis of ketone 8 in >85% yield.
13
CHCl3 at 0 °C to remove the acetonide chemoselectively
followed by bromination under Appel-type conditions to afford
bromide 15 in 86% yield. With the stage now set for our key
cascade, we were pleased to find, after a screen of several
carbon-based radical traps, that following exposure of 15 to
allyltributylstannane and Et3B as a radical initiator14 under an
air atmosphere 16 could be obtained in 59% yield as a single
diastereomer possessing the desired stereochemistry at the C-
20 quaternary center (as confirmed by X-ray analysis).
Having reached this critical staging area, we next envisaged
deprotecting the nitrogen atom and effecting epimerization/
lactamization to access 8 directly through treatment with the
appropriate base. However, exploration of various conditions
along these lines revealed that the deprotected α-amino ketone
underwent facile dimerization and oxidation instead to give
pyrazine 18. Fortunately, a solution presented itself during
epimerization attempts on 16 using DBU as base when we
We were now poised to attempt the challenging late-stage
arylation reaction to append the indolenine domain onto our
B
dx.doi.org/10.1021/ja406546k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
a
Scheme 3. Completion of the Total Synthesis of (+)-Scholarisine A (1)
a
Reagents and conditions: (a) 2-iodoaniline (1.1 equiv), PPTS (0.09 equiv), PhMe, THF, 4 Å m.s., 90 °C, 18 h; (b) n-Bu3SnH (1.2 equiv), ACHN
(1.2 equiv), toluene, 110 °C, 4.5 h, 25% (3 steps), 22:23 = 3.0:1.0; (c) NIS (2.0 equiv), CH2Cl2, 23 °C, 20 h, 72% or IDSI (1.7 equiv), CH2Cl2, −20
°C, 45 min, 27%; (d) O3, CH2Cl2, MeOH, −78 °C, 2 min; Me2S (22 equiv), −78→23 °C; NaBH4 (2.6 equiv), EtOH, CH2Cl2, −40 → −10 °C, 1.5
h, 68%; (e) Lawesson’s Reagent (2.1 equiv), THF, toluene, 110 °C, 16 h, 47%; (f) Raney Ni, THF, 23 °C, 1 h, 86%; (g) PhIO (5.4 equiv), CH2Cl2,
23 °C, 98%. PPTS = pyridinium 4-toluenesulfonate, ACHN = 1,1′-azobis(cyclohexanecarbonitrile), NIS = N-iodosuccinimide, IDSI = (Et2SI)2Cl·
SbCl6.
caged framework. In the event, crude ketone 8 (Scheme 3) was
directly subjected to a condensation reaction with 2-iodoaniline
to provide the corresponding imine 7 as a 2.4−2.8:1 mixture of
geometric isomers (batch dependent).16 This material was then
treated without purification with n-Bu3SnH and 1,1′-azobis-
(cyclohexanecarbonitrile) to give the desired spiroindolenine
22 in 18% overall yield for the three steps starting from the
precursor to carbocycle 8. It also provided a small amount of
product (6%, three steps) resulting from functionalization of
the tertiary C−H bond at C-9 (i.e., 23), revealing that relatively
good positional selectivity could be achieved in this complex
context.17 Overall, the combined yield for the spiroindolenine
products over the three-step sequence is 62% per step.18
Several points about this critical sequence require mention.
First, we believe the annulation likely proceeds in the
mechanistic fashion shown via initial aryl radical formation,
1,5-H atom transfer and cyclization onto the aryl group, and
terminating oxidation of the resulting cyclohexadienyl radical.9b
Second, a transformation of this sort has not, to the best of our
knowledge, been applied to the synthesis of such units, much
less in a setting so complex.9a,b,19 Third, multiple attempts at
earlier aryl group installation, e.g. via Fischer indole synthesis or
enolate arylation on compounds similar to 13 (cf. Scheme 2) or
through aryl group incorporation into the original dienophile
(i.e., 11), proved fruitless.
NIS.23 Although this intermediate (structure and absolute
configuration verified by X-ray) ultimately proved to be
unproductive in its own right, its synthesis prompted us to
explore the allyl group, or some group derived from it, as a
means to achieve chemoselective reduction of the lactam
through intramolecular assistance. Thus, ozonolytic cleavage of
the alkene and reduction of the resulting aldehyde to amino
alcohol 25 provided a potential handle in the form of a primary
alcohol. Pleasingly, treatment of 25 with Lawesson’s reagent at
110 °C effected thiation/cyclodehydration to deliver thioimi-
date 26. Without that handle, the lactam groups of 22, or its di-
and tetrahydro derivatives, did not undergo thionation under
identical (and even more vigorous) conditions.24 Subsequent
excision of the S atom within 26 using Raney Ni in THF at 23
°C for 1 h simultaneously formed the ethyl group and imine
function of 1 in the form of 27. A final oxidation using PhIO at
ambient temperature then gave (+)-scholarisine A (1) in near-
quantitative yield, the spectroscopic data of which matching
those of the naturally-derived material in all respects.
In conclusion, we have completed an enantioselective
synthesis of (+)-scholarisine A (1) in 14 steps from known
enone 11 (15 steps from commercial materials). This route
compares favorably in terms of step count to the inaugural
Smith synthesis3 of 1. Key discoveries include an efficient and
diastereoselective pyrone Diels−Alder reaction to rapidly form
the appropriately functionalized [2.2.2]-bicycle, a radical
cyclization/Keck allylation to concurrently forge the [3.3.1]-
bicycle and C-20 quaternary center, an indolenine annulation at
a nonenolizable tertiary center via a novel late-stage radical C−
H arylation, and the use of a pendant hydroxyl group to
facilitate the chemoselective reduction of an extremely
unreactive lactam. Current and future work is directed toward
applying the developed strategy to related targets, exploring the
scope of the C−H arylation step, and probing the biochemical
potential of scholarisine A itself.
From intermediate 22, only functional group interconver-
sions remained to reach the target, namely transformation of
the allyl group and the lactam to the ethyl group and imine,
respectively, of 1. While the first of these tasks could be readily
achieved in initial studies (not shown),20 the reduction of the
lactam in 22 (or its ethyl or propyl congener) proved extremely
difficult, with >20 methods (both chemoselective or non-
chemoselective) being unsuccessful.21 The key clue toward an
effective strategy came in the formation of iodoimidate 24
under iodocyclization conditions employing either IDSI22 or
C
dx.doi.org/10.1021/ja406546k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
°C) while its 2,6-dimethylaniline congener does so even more readily
(110 °C). We expected our 2-iodoaniline derived imine to isomerize at
ASSOCIATED CONTENT
* Supporting Information
Detailed experimental procedures, copies of all spectral data,
and full characterization. This material is available free of charge
via the Internet at http://pubs.acs.org.
■
S
an intermediate rate: (a) Wurmb-Gerlich, D.; Vogtle, F.; Mannschrek,
̈
A.; Staab, H. A. Liebigs Ann. Chem. 1967, 708, 36. In addition, any
residual acid present would be expected to facilitate the isomerization
process: (b) Jennings, W. B.; Al-Showiman, S.; Tolley, M. S.; Boyd, D.
R. J. Chem. Soc., Perkin Trans. 2 1975, 1535.
AUTHOR INFORMATION
Corresponding Author
sas2197@columbia.edu
Notes
(11) Ageno, G.; Banfi, L.; Cascio, G.; Guanti, G.; Manghisi, E.; Riva,
R.; Rocca, V. Tetrahedron 1995, 51, 8121. The Weinreb amide
precursor to 11 is commercially available (Oakwood Chemical,
product no. 048634).
(12) Reported Diels−Alder reactions with methyl coumalate (12)
typically involve electron-rich or highly electron-deficient dienophiles.
See, for example: (a) Shimo, T.; Yoshimura, H.; Suishu, T.; Somekawa,
K. Nippon Kagaku Kaishi 1988, 1984 (maleimides). (b) Shimo, T.;
Nagahama, H.; Yozoru, K.; Somekawa, K. J. Heterocyclic Chem. 1992,
29, 801 (p-benzoquinones). (c) Bates, R. W.; Pratt, A. J.; Rendle, P.
M.; Thomson, W. T. Aust. J. Chem. 1998, 51, 383 (1,1-bis(methylthio)
ethene). (d) Chen, C.-H.; Liao, C.-C. Org. Lett. 2000, 2, 2049 (2-
methoxyfurans).
■
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The National Science Foundation (CHE-0619638) is thanked
for acquisition of an X-ray diffractometer. We thank Prof.
Gerard Parkin, Dr. Wesley Sattler, and Mr. Ahmed Al-Harbi for
performing crystallographic analyses. We thank Dr. John
Decatur and Dr. Yasuhiro Itagaki for NMR spectroscopic and
mass spectrometric assistance, respectively. Financial support
was provided by Columbia University, Bristol−Myers Squibb,
Eli Lilly, and Amgen.
(13) Yonezawa, Y.; Tani, N.; Shin, C.-G. Heterocycles 2005, 65, 95.
(14) Use of AIBN as a radical initiator proved to be ineffective. For
an example of another case where Et3B was uniquely effective as a
radical initiator, see: Reddy, L. R.; Fournier, J.-F.; Reddy, B. V. S.;
Corey, E. J. J. Am. Chem. Soc. 2005, 127, 8974.
(15) In the absence of TEMPO, the reaction proceeded in ca. 40%
yield. We suspect TEMPO is a more effective radical trap than O2,
leading to less decomposition/side reactions.
REFERENCES
■
(1) Cai, X.-H.; Tan, Q.-G.; Liu, Y.-P.; Feng, T.; Du, Z.-Z.; Li, W.-Q.;
Luo, X.-D. Org. Lett. 2008, 10, 577.
(16) Although this step might be envisioned to be difficult,
condensations of anilines and 2-adamantanone, a material that is
similar in structure to 8, are known. For a representative example, see:
Sasaki, T.; Eguchi, S.; Hirako, Y. Tetrahedron 1976, 32, 437.
(17) Given the starting ratio of imines, it is possible that the original
stereochemistry of that functional group was product-determining.
Since the initial isomers could not be readily isolated as pure
components or their stereochemistry assigned, it is difficult to draw a
conclusion in terms of whether isomerization is occurring prior to
cyclization.
(18) Aside from the formation of several unidentifiable products
generated in trace quantities, the only other characterizable species
isolated (≤5% over three steps) was that derived from trapping of the
intermediate cyclohexadienyl radical 21 by C6H10CN· (from ACHN).
(19) In model studies using the corresponding 2-adamantanone-
derived imine with two identical tertiary C−H bonds, yields of ca. 60%
could be obtained. Studies are continuing to explore the scope of this
transformation: Rageot, D.; Snyder, S. A., unpublished results.
(20) This transformation was achieved through ozonolysis, dithiolane
formation, and desulfurization with Raney Nickel.
(2) For selected examples of recent syntheses of other akuammiline
alkaloids, such as vincorine and aspidopylline A, see: (a) Zhang, M.;
Huang, X.; Shen, L.; Qin, Y. J. Am. Chem. Soc. 2009, 131, 6013. (b) Zu,
L.; Boal, B. W.; Garg, N. K. J. Am. Chem. Soc. 2011, 133, 8877. (c) Zi,
W.; Xie, W.; Ma, D. J. Am. Chem. Soc. 2012, 134, 9126. (d) Horning,
B. D.; MacMillan, D. W. C. J. Am. Chem. Soc. 2013, 135, 6442.
(3) (a) Adams, G. L.; Carroll, P. J.; Smith, A. B., III. J. Am. Chem. Soc.
2012, 134, 4037. (b) Adams, G. L.; Carroll, P. J.; Smith, A. B., III. J.
Am. Chem. Soc. 2013, 135, 519.
(4) Watanabe, T.; Kato, N.; Umezawa, N.; Higuchi, T. Chem.Eur. J.
2013, 19, 4255.
(5) Selected examples from our group include the total synthesis of
the terpenoid rippertenol and the polycycles of the securinega
alkaloids: (a) Snyder, S. A.; Wespe, D. A.; von Hof, J. M. J. Am. Chem.
Soc. 2011, 133, 8850. (b) ElSohly, A. M.; Wespe, D. A.; Poore, T. J.;
Snyder, S. A. Angew. Chem., Int. Ed. 2013, 52, 5789.
(6) For a review, see: Afarinkia, K.; Vinader, V.; Nelson, T. D.;
Posner, G. H. Tetrahedron 1992, 48, 9111.
(7) For an example of a tandem radical cyclization/Keck allylation,
see: Evans, P. A.; Manangan, T. Tetrahedron Lett. 1997, 38, 8165.
(8) The aza-enolate derived from 7 would be extremely strained,
formally containing an anti-Bredt olefin within a six-membered ring.
This outcome would preclude the use of methods that proceed
through enamine/aza-enolate intermediates.
(21) Selected reductions attempted: Tf2O/HSiEt3, Cp2ZrHCl,
LiAlH4, Red-Al, AlH3, AlH3·EtNMe2, BH3·THF, BH3·SMe2; reduction
of corresponding methyl imidate (Me3OBF4): LiAlH4, BH3·THF,
AlH3, NaBH4, NaBH4/HCl, Zn/AcOH, LiDBB, Ra-Ni. In all cases, no
reduction of the lactam (or its derivative) was observed.
(9) For radical-based H-atom transfer/cyclization approaches
towards oxindoles, see: (a) Beckwith, A. L. J.; Storey, J. M. D. J.
Chem. Soc., Chem. Commun. 1995, 977. (b) Beckwith, A. L. J.; Bowry,
V. W.; Bowman, W. R.; Mann, E.; Parr, J.; Storey, J. M. D. Angew.
Chem., Int. Ed. 2004, 43, 95. For selected Pd-catalyzed approaches to
indolines involving activation of nonenolizable C−H bonds, see:
(c) Watanabe, T.; Oishi, S.; Fujii, N.; Ohno, H. Org. Lett. 2008, 10,
1759 (primary C−H). (d) Nakanishi, M.; Katayev, D.; Besnard, C.;
(22) Snyder, S. A.; Treitler, D. S.; Brucks, A. P. J. Am. Chem. Soc.
2010, 132, 14303.
(23) Arunachalam, T.; Fan, H.; Pillai, K. M. R.; Ranganathan, R. S. J.
Org. Chem. 1995, 60, 4428.
(24) Nishio, T.; Sekiguchi, H. Tetrahedron 1999, 55, 5017. In the
present case, their proposed mechanism involving initial thiation of the
alcohol, thionation of the lactam, and cyclization with loss of H2S
seems unlikely given that 22 and its congeners lacking the alcohol
group did not undergo thionation.
Kundig, E. P. Angew. Chem., Int. Ed. 2011, 50, 7438 (secondary C−H).
̈
(e) Saget, T.; Perez, D.; Cramer, N. Org. Lett. 2013, 15, 1354 (tertiary
cyclopropyl C−H).
(10) At the planning stage, we recognized that if isomerization of the
imine geometric isomers was competitive (e.g., thermally) with the C−
H functionalization process, then the initial E/Z ratio may not be
important. Staab and co-workers have shown that the imine derived
from acetone and aniline isomerizes rapidly on the NMR time scale at
higher temperatures (E/Z-Me signal coalescence temperature = 126
D
dx.doi.org/10.1021/ja406546k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX