A Concise Synthesis of Forskolin

Article abstract of DOI:10.1002/anie.201706809

A 24-step synthesis of (±)-forskolin is presented, which delivered hundred milligram quantities of this complex diterpene in one pass. Transformations key to our approach include: a) a strategic allylic transposition, b) stepwise assembly of a sterically encumbered isoxazole ring, and c) citric acid-modified Upjohn dihydroxylation of a resilient tetrasubstituted olefin. The developed route has exciting potential for the preparation of new forskolin analogues inaccessible by semisynthesis.

Full text of DOI:10.1002/anie.201706809

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: A concise synthesis of forskolin
Authors: Ondřej Hylse, Lukáš Maier, Roman Kučera, Tomáš Perečko,
Aneta Svobodová, Lukáš Kubala, Kamil Paruch, and Jakub
Svenda
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.201706809
Angew. Chem. 10.1002/ange.201706809
Link to VoR: http://dx.doi.org/10.1002/anie.201706809
http://dx.doi.org/10.1002/ange.201706809

10.1002/anie.201706809
Angewandte Chemie International Edition
COMMUNICATION
A Concise Synthesis of Forskolin
Ondřej Hylse, Lukáš Maier, Roman Kučera, Tomáš Perečko, Aneta Svobodová, Lukáš Kubala, Kamil
Paruch, and Jakub Švenda*
Abstract: We report a 24-step synthesis of ()-forskolin, which
delivered hundred-milligram quantities of this complex diterpene in
one pass. Transformations key to our approach include: (a) a strategic
allylic transposition, (b) stepwise assembly of a sterically encumbered
isoxazole ring, and (c) citric acid-modified Upjohn dihydroxylation of a
resilient tetrasubstituted olefin. We demonstrate that the developed
route has an exciting potential for the preparation of new forskolin
analogs inaccessible by semisynthesis.
Fully synthetic approaches often enable structural modifications
that would be difficult to address by semisynthesis. With its
densely grouped array of quaternary carbons and stereocenters,
forskolin (1) remains a daunting target, however. Three early
syntheses reported by the groups of Ziegler, Corey
(enantioselective), and Ikegami, published within one year of
each other (3140 step sequences),[6] remain landmark
achievements in the field. More recently, Lett and co-workers
disclosed a 32-step synthesis of racemic forskolin (1).[7] Formal
syntheses and numerous approaches toward forskolin (1) have
also been described in the literature.[8] Herein, we detail a 24-step
synthesis, which intercepted a late-stage intermediate of Lett[7b]
and enabled preparation of hundred-milligram quantities of ()-
forskolin [()-1]. The route gave also access to fully synthetic
forskolin analogs.
In our synthetic planning, we considered cis-decalin 3 (Scheme
1) as an early intermediate, because its ring systems map onto
the decalin substructure of forskolin (1) and it is readily prepared
on a multigram scale using a flexible DielsAlder strategy
developed by researchers at Hoffmann-La Roche (8 steps from
the commercially available materials).[9] Although decalin
intermediates of this type were contemplated already some 30
years ago,[10] their conversion into forskolin (1) eluded efficient
synthetic solution. As a strategic objective, we therefore targeted
concise construction of the -pyrone ring of forskolin (1) from cis-
decalin 3. An isoxazole-based route described below emerged as
attractive due to the minimum of redox and protecting group
manipulations required by such an approach.
Forskolin (1) is a complex natural product that has become a
standard research tool in biology.[1] This labdane diterpene was
shown to be an allosteric stimulator of cAMP-generating
membrane-bound adenylyl cyclases (ACs).[2] These enzymes are
involved in action of numerous small-molecule drugs, most
notably those affecting G-protein coupled receptors (GPCRs).
Direct targeting of ACs for therapeutic applications is complicated,
in part, by their structural similarity and the lack of isoform-
selective modulators.[3]
Reactivities of the two carbonyl groups in cis-decalin 3 differ
dramatically, as shown previously.[9b,11] Following precedent, a
highly selective addition of monolithium acetylide (2.0 equiv,
78 °C) occurred with convex-face approach of the reagent.[12]
Subsequent epimerization in the presence of aluminum oxide
(Brockmann grade IV, toluene, reflux) provided trans-decalin 4
(82% over 2 steps). We observed that the efficiency of the
epimerization was sensitive to the form of the reagent and,
Figure 1. Structure of forskolin (1) and drug NKH477 (2) derived through
semisynthesis.
Semisynthesis was explored extensively in a search of forskolin
analogs with improved properties[4] and these efforts led to the
identification of NKH477 (2, Figure 1), a heart failure drug
approved in Japan.[5]
strongly, to the substitution of the cis-decalin (see SI).[9b]
A
subsequent sequence of diastereoselective enone 1,2-reduction
(diisobutylaluminum hydride, 78 °C) and benzoylation
proceeded smoothly to give monoprotected allylic diol 5 in 74%
yield (2 steps).
At this stage, we performed an oxidative 1,3-transposition of the
tertiary allylic alcohol within 5 (excess pyridinium chlorochromate,
sodium acetate, 4Å molecular sieves),[13] followed by highly
diastereoselective 1,2-reduction of the resulting enone (6) using
diisobutylaluminum hydride (5.0 equiv); the latter conditions also
removed the benzoyl ester. The isolated 1,2-diol (not shown) was
protected as the corresponding acetonide (7, 61% over 3 steps).
This strategic sequence worked well and solved many of our initial
problems with installation of the syn-1,2-diol found in forskolin.
[a]
[b]
[c]
O. Hylse, Dr. L. Maier, R. Kučera, Dr. K. Paruch, Dr. J. Švenda
Department of Chemistry
Masaryk University
Kamenice 5, Brno, 625 00, Czech Republic
E-mail: svenda@chemi.muni.cz
O. Hylse, Dr. L. Maier, Dr. T. Perečko, A. Svobodová, Dr. L. Kubala,
Dr. K. Paruch, Dr. J. Švenda
International Clinical Research Center
St. Anne’s University Hospital
Pekařská 53, Brno, 656 91, Czech Republic
Dr. T. Perečko, A. Svobodová, Dr. L. Kubala
Institute of Biophysics
Academy of Sciences of the Czech Republic
Královopolská 135, Brno, 612 65, Czech Republic
This article is protected by copyright. All rights reserved.

10.1002/anie.201706809
Angewandte Chemie International Edition
COMMUNICATION
With the alkyne-containing intermediate 7 available in multigram
quantities (>4 g), we approached construction of the isoxazole
ring with [3+2] cycloaddition chemistry.[14] Exposing the alkyne (7)
to excess acetonitrile oxide (3.0 equiv, generated in situ from
freshly prepared N-hydroxyacetimidoyl chloride) led to a single
isomer of alkynyl oxime 8 (79% isolated yield) as the principal
component of the product mixture (ca. 5% of the desired isoxazole
detected by 1H NMR). This highly unusual outcome corresponds,
formally, to an interrupted cycloaddition process.[15] Alkynyl
oximes can be isomerized to isoxazoles however and exposure
of 8 to catalytic quantities of gold(III) chloride[16] (15 mol%, 1,2-
dichloroethane, 45 °C) induced efficient cyclization to isoxazole 9
(83%). A simpler and even more efficient protocol involved
colleagues[19] was uniquely effective and did provide the
dihydroxylation product (triol 11) in acceptable yield (64% after 3
cycles, gram scale), with good mass recovery of the unreacted
material.[20] An X-ray crystal structure of 11 unambiguously
confirmed the relative stereochemistry.
Selective protection of the 1,3-diol function in 11 (p-
toluenesulfonic acid, 2,2-dimethoxypropane) and hydrogenolysis
of the isoxazole ring (Raney-Ni, H2) gave cleanly -amino enone
12 (88% over 2 steps). Exposure of 12 to an anhydrous solution
of hydrochloric acid (2.0 equiv, dioxanetoluene, 23 °C) resulted
in an intramolecular cyclization to known -pyrone 13[7b] (84%). By
subtle but important modification of the previously published
three-step sequence (see SI),[6d,7b] we converted the -pyrone (13)
treatment of 8 with excess tetrabutylammonium fluoride (4.0 equiv, into ()-forskolin [()-1, >100 mg prepared in one pass] that was
23 oC), conditions that triggered not only formation of the
isoxazole but also the desired tert-butyldimethylsilyl ether
cleavage (product 10, 91%). Overall, the two-step sequence
(alkyne 7  isoxazole 10) provided a highly effective solution to
the difficult problem of crowded isoxazole assembly (see
discussion below).[17]
The projected dihydroxylation of tetrasubstituted alkene 10 was
expected to be challenging from the start but was pursued
because of reasonable precedent in the forskolin literature.[6b,6h,18]
We learned only late in the synthesis that none of these previously
described conditions worked on our substrate (intermediate 10).
After a screening of conditions (see SI), we found that the citric
acid-modified Upjohn protocol described by Sharpless and
structurally identical to the natural product (see SI).
It is worth mentioning that the dense array of substituents and
functional groups found in forskolin, and many of its synthetic
precursors, often complicated seemingly straightforward
chemistry operations.[21] For example, application of the standard
dipolar cycloaddition methodology for construction of the
isoxazole system was initially challenging. Numerous advanced
alkyne intermediates (e.g. 1619, Scheme 2) were virtually
unreactive in the cycloaddition process with acetonitrile oxide.[22]
We attribute this behavior primarily to the sterically hindered
environment around the alkyne. The single example of a
successful and high yielding cycloaddition we observed with
oxacyclic alkyne 14[9b] supports this idea (Scheme 2, top).
Scheme 1. Synthetic pathway to ()-forskolin [()-1] featuring stepwise construction of a crowded isoxazole intermediate.
This article is protected by copyright. All rights reserved.

10.1002/anie.201706809
Angewandte Chemie International Edition
COMMUNICATION
The other enantiomer, 1(ent-1), showed no stimulatory effects up
to 100 M and behaved, in this sense, analogously to 1,9-
dideoxyforskolin (20, commonly used negative control). It is worth
noting that activity of the unnatural enantiomer of forskolin had not
been examined previously.
To demonstrate the potential of our route in modifying positions of
forskolin that would be difficult or impossible to address by
semisynthesis, we have assembled two fully synthetic analogs 22
and 23 (see SI). While neither of these two analogs was more
active than natural forskolin in the membrane assay (see Figure
2), our preliminary data in HEK293 cells with individually
overexpressed AC isoforms indicate that the level of stimulation
varies significantly (including stimulation equipotent to forskolin)
depending on the specific isoform (data not shown). Improved
isoform selectivity of analogs resulting from structural changes at
new positions of forskolin has important implications.
In summary, we have developed a concise synthetic route to ()-
forskolin [()-1] that produced hundred-milligram quantities of this
complex target in one pass. Transformations key to the success
Scheme 2. A subset of terminal alkynes examined as substrates in the [3+2]
dipolar cycloaddition with acetonitrile oxide.
of our approach include: (i)
a
strategic oxidative
transpositionreduction sequence to setup the 1,2-diol function of
forskolin, (ii) stepwise construction of a densely substituted
isoxazole intermediate, and (iii) citric acid-modified Upjohn
dihydroxylation of a hindered tetrasubstituted olefin. In contrast to
the natural product, the unnatural enantiomer of forskolin [1(ent-
1)] was found to not have stimulatory activity toward ACs below
100 M. We have also demonstrated that our route provides
access to fully synthetic forskolin analogs modified at new
positions. The ongoing studies seek to explore this potential in
combination with biological profiling of analogs against a panel of
human AC isoforms.
In contrast stands the unusual reactivity of alkyne 7 (predominant
formation of alkynyl oxime, ≤5% of isoxazole) exploited in our
synthesis of forskolin (see Scheme 1), which may stem from
presence of the enyne unit within the substrate.[15] It is interesting
to add that electronic characteristics of the nitrile oxide can alter
the relative product distribution (see SI).
The enantiomers of synthetic forskolin were obtained separately
through efficient resolution of the racemic material either by semi-
preparative HPLC  first-eluting enantiomer: 1(ent-1), second-
eluting enantiomer: 1(ent-2)
 or through diastereomeric
camphanyl esters (see SI). 1(ent-2) stimulated cAMP production
with potency equal to that of the natural forskolin as determined
in the cAMP-based membrane assay (Figure 2).
Acknowledgements
This work was supported by the Czech Science Foundation
(GJ15-10504Y). The authors acknowledge project no. LQ1605
from the National Program of Sustainability II (MEYS CR) and
project CZ-OPENSCREEN: National Infrastructure for Chemical
Biology (LM2015063). JŠ and KP are grateful to Isabel & Alfred
Bader Fund, a Bader Philanthropy. We thank Dr. Marek Nečas
and the CF X-ray diffraction and Bio-SAXS facility supported by
the CIISB research infrastructure (LM2015043 funded by MEYS
CR) for obtaining the X-ray data. We thank Dr. Andrea Vítečková
Wünschová and Petra Daďová for their technical help with the AC
membrane assay.
[1] a) P. A. Insel, R. S. Ostrom, Cell. Mol. Neurobiol. 2003, 23, 305314; b) R.
H. Alasbahi, M. F. Melzig, Pharmazie 2012, 67, 513.
[2] a) K. B. Seamon, W. Padgett, J. W. Daly, Proc. Natl. Acad. Sci. USA 1981,
78, 33633367; b) C. W. Dessauer, T. T. Scully, A. G. Gilman, J. Biol. Chem.
1997, 272, 2227222277; c) G. Zhang, Y. Liu, A. E. Ruoho, J. H. Hurley,
Nature 1997, 386, 247253; d) J. J. G. Tesmer, R. K. Sunahara, A. G. Gilman,
S. R. Sprang, Science 1997, 278, 19071916.
[3] a) S. Pierre, T. Eschenhagen, G. Geisslinger, K. Scholich, Nat. Rev. Drug
Discov. 2009, 8, 321335; b) B. Pavan, C. Biondi, A. Dalpiaz, Drug. Disc.
Today 2009, 14, 982991; c) R. Seifert, K. Y. Beste, Sci. Signal. 2012, 5,
14; d) C. S. Brand, H. J. Hocker, A. A. Gorfe, C. N. Cavasotto, C. W.
Dessauer, J. Pharmacol. Exp. Ther. 2013, 347, 265275; e) C. W. Dessauer,
V. J. Watts, R. S. Ostrom, M. Conti, S. Dove, R. Seifert, Pharmacol. Rev.
2017, 69, 93139.
Figure 2. Modulation of activity of ACs by forskolin (1), 1,9-dideoxyforskolin (20),
7-deacetylforskolin (21), and fully synthetic analogs 22 and 23. The AC activity
was determined in a cAMP-based membrane assay (wild-type HEK293 cells-
derived) as described in the Supporting Information with increasing
concentrations of the diterpenes (0.004–400 µM). Data shown are means from
35 independent experiments. AC activity was normalized relative to the
maximum stimulation observed with 120 µM of natural forskolin.
This article is protected by copyright. All rights reserved.

10.1002/anie.201706809
Angewandte Chemie International Edition
COMMUNICATION
[4] a) K. B. Seamon, J. W. Daly, H. Metzger, N. J. de Souza, J. Reden, J. Med.
Chem. 1983, 26, 436439; b) J. D. Robbins, D. L. Boring, W.-J. Tang, R.
Shank, K. B. Seamon, J. Med. Chem. 1996, 39, 27452752; c) T. Onda, Y.
Hashimoto, M. Nagai, H. Kuramochi, S. Saito, H. Yamazaki, Y. Toya, I. Sakai,
C. J. Homcy, K. Nishikawa, Y. Ishikawa, J. Biol. Chem. 2001, 276,
4778547793; d) C. Pinto, D. Papa, M. Hübner, T.-C. Mou, G. H. Lushington,
R. Seifert, J. Pharmacol. Exp. Ther. 2008, 325, 2736.
[23] We were unable to advance this early isoxazole intermediate (15) due to
the robust nature of the oxacycle.
[5] S. Hosoda, T. Motomiya, T. Katagiri, T. Takano, S. Sasayama, H. Toshima,
N. Ogawa, Jpn. J. Clin. Pharmacol. Ther. 1997, 28, 583602.
[6] a) F. E. Ziegler, B. H. Jaynes, M. T. Saindane, Tetrahedron Lett. 1985, 26,
3307–3310; b) F. E. Ziegler, B. H. Jaynes, M. T. Saindane, J. Am. Chem.
Soc. 1987, 109, 8115–8116; c) F. E. Ziegler, B. H. Jaynes, Tetrahedron Lett.
1987, 28, 2339–2342; d) F. E. Ziegler, B. H. Jaynes, Tetrahedron Lett. 1988,
29, 2031–2032; e) E. J. Corey, P. Da Silva Jardine, J. C. Rohloff, J. Am.
Chem. Soc. 1988, 110, 3672–3673; f) E. J. Corey, P. Da Silva Jardine, T.
Mohri, Tetrahedron Lett. 1988, 29, 6409–6412; g) E. J. Corey, P. Da Silva
Jardine, Tetrahedron Lett. 1989, 30, 7297–7300; h) S. Hashimoto, S. Sakata,
M. Sonegawa, S. Ikegami, J. Am. Chem. Soc. 1988, 110, 3670–3672.
[7] a) B. Delpech, D. Calvo, R. Lett, Tetrahedron Lett. 1996, 37, 1015–1018; b)
B. Delpech, D. Calvo, R. Lett, Tetrahedron Lett. 1996, 37, 1019–1022.
[8] a) For a review of the field until 1992, see: M. I. Colombo, J. Zinczuk, E. A.
Ruveda, Tetrahedron 1992, 48, 963–1037; b) J. J. Harnett, J. K. Sutherland,
H. Yang, J. Chem. Soc.-Perkin Trans. 1 1995, 1391–1395; c) M. Leclaire, R.
Levet, F. Péricaud, L. Ricard, J. Y. Lallemand, Tetrahedron 1996, 52, 7703–
7718; d) M. Leclaire, R. Levet, F. Ferreira, P. H. Ducrot, J. Y. Lallemand, L.
Ricard, Chem. Commun. 2000, 1737–1738; e) A. Anikin, M. Maslov, J. Sieler,
S. Blaurock, J. Baldamus, L. Hennig, M. Findeisen, G. Reinhardt, R. Oehme,
P. Welzel, Tetrahedron 2003, 59, 5295–5305 and references therein; f) H.
Ye, G. Deng, J. Liu, F. G. Qiu, Org. Lett. 2009, 11, 5442–5444.
[9] (a) F. Kienzle, J. Stadlwieser, I. Mergelsberg, Helv. Chim. Acta 1989, 72,
348–352; b) F. Kienzle, J. Stadlwieser, W. Rank, P. Schönholzer, Helv. Chim.
Acta 1990, 73, 1108–1139.
[10] G. Bold, S. Chao, R. Bhide, S.-H. Wu, D. V. Patel, C. J. Sih, C. Chidester,
Tetrahedron Lett. 1987, 28, 1973–1976.
[11] D. Liotta, M. Saindane, U. Sunay, W. C. L. Jamison, J. Grossman, P.
Phillips, J. Org. Chem. 1985, 50, 3241–3243.
[12] In agreement with the previous reports (see ref. 9b), attempted additions of
sp3 or sp2 organometallic nucleophiles to cis-decalin 3 were problematic.
[13] a) W. G. Dauben, D. M. Michno, J. Org. Chem. 1977, 42, 682–685; b) P.
Sundararaman, W. Herz, J. Org. Chem. 1977, 42, 813819; c) W. S. Cheung,
H. N. C. Wong, Tetrahedron Lett. 1998, 39, 6521–6524.
[14] a) A. Padwa, W. H. Pearson, Eds., Synthetic Applications of 1,3-Dipolar
Cycloaddition Chemistry Toward Heterocycles and Natural Products, John
Wiley & Sons, Inc., New York, USA, 2002; b) H. Feuer, K. Torssell, Eds.,
Nitrile Oxides, Nitrones, and Nitronates in Organic Synthesis: Novel
Strategies in Synthesis, Wiley-Interscience, Hoboken, N.J., 2008.
[15] P. Beltrame, P. Sartirana, C. Vintani, J. Chem. Soc. B-Phys. Org. 1971,
814817 and references therein.
[16] a) C. Praveen, A. Kalyanasundaram, P. T. Perumal, Synlett 2010, 777–781;
b) S. Murarka, A. Studer, Org. Lett. 2011, 13, 2746–2749.
[17] A more direct construction of the decalinisoxazole CC bond linkage using
metal-catalyzed cross-couplings proved difficult (unpublished results).
[18] a) S. Hashimoto, M. Sonegawa, S. Sakata, S. Ikegami, J. Chem. Soc.-
Chem. Commun. 1987, 24–25; b) K. C. Nicolaou, S. Kubota, W. S. Li, J.
Chem. Soc.-Chem. Commun. 1989, 512–514.
[19] P. Dupau, R. Epple, A. A. Thomas, V. V. Fokin, K. B. Sharpless, Adv. Synth.
Catal. 2002, 344, 421–433.
[20] tert-Butyldimethylsilyl ether 9 (see Scheme 1) was unreactive under these
conditions.
[21] For an unusual reactivity observed during our early studies, see: R. Kučera,
O. Hylse, M. Babiak, J. Švenda, Tetrahedron Lett. 2015, 56, 61716173.
[22] For an example of low-yielding isoxazoline formation using an oxime
precursor structurally related to 17, see: A. Barco, S. Benetti, G. Spalluto, A.
Casolari, G. P. Pollini, V. Zanirato, P. G. Baraldi, Il Farmaco 1991, 46, 1281–
1295.
This article is protected by copyright. All rights reserved.

10.1002/anie.201706809
Angewandte Chemie International Edition
COMMUNICATION
Text and graphics for the Table-of-Contents
Forskolin is a complex diterpene commonly used in biomedical research to raise levels of cAMP. Previous strategies to access forskolin
analogs with improved properties focused primarily on semisynthesis. To complement these approaches, we have streamlined the
chemical synthesis of forskolin and applied it to the preparation of fully synthetic forskolin analogs.
Keywords: forskolin, adenylyl cyclases, natural product synthesis, structural analogs
This article is protected by copyright. All rights reserved.