Conformational locking through allylic strain as a device for stereocontrol - Total synthesis of grandisine A

Article abstract of DOI:10.1002/anie.200703245

(Chemical Equation Presented) Controlled fusion: The use of a seco ring D precursor to control both the stereochemistry of a critical cycloaddition with acetaldehyde and the direction of protonation of the resultant silyl enol ether cycloadduct has led to the total synthesis of grandisine A, which shows promising selectivity for binding to the δ-opioid receptor. The D ring is installed in the final stages of the synthesis (see scheme, LA = Lewis acid).

Full text of DOI:10.1002/anie.200703245

Angewandte
Chemie
DOI: 10.1002/anie.200703245
Natural Products
Conformational Locking through Allylic Strain as a Device for
Stereocontrol—Total Synthesis of Grandisine A**
David J. Maloney and Samuel J. Danishefsky*
Carroll et al. have described the isolation and structural
assignment of a family of indolizidine alkaloids known as
grandisines from Elaeocarpus grandis.[1] From a biological
perspective, these small-molecule natural products (SMNP)
are of interest since they bind to the opioid receptors with
apparent selectivity.[2] On the basis of earlier studies that
addressed the consequences of binding to opioid receptors, it
is thought that agents with specific affinity for the d receptor
may well bring about modulation and reduction of pain while
minimizing common side effects such as nausea and hyper-
tension.[3] Although the binding affinity of various grandisines
to the d-opioid receptor does not yet warrant their develop-
ment as pharmaceuticals, their receptor selectivity is encour-
aging. To pursue such potentialities in the context of a
medicinal chemistry effort would require the development of
a reasonably concise and efficient total synthesis of a
particular grandisine, which would provide adequate material
for a collaborative exploratory program with specialist
laboratories. In this connection, we identified grandisine A
(1)[1a] as our target molecule.
bone stereoconnectivity at carbon atoms 7, 8, and 9. The syn-
syn relationship of the protons at these stereogenic centers
enforces a hemisphere-like presentation of rings B, C, and D,
thus resulting in a significant abutment of loci in rings B and D
(Scheme 1). Furthermore, if pyramidalization of the lone pair
of electrons on the bridgehead indolizidine nitrogen atom
tended to occur in the direction of a cisoid C:D bridgehead
junction, as expected, the B:D abutment could be even more
severe. Hence, stereocontrol would have to be manifested at
the kinetic level. Thermodynamic equilibration of a tricyclic
intermediate through a C10 ketone or vinylogously through
the C14 ketone of the fully mature grandisine would most
likely lead to an undesired B:C trans fusion.
From the outset, our retrosynthetic analysis favored
introducing the A ring of grandisine A by merging, in some
fashion, a single antipode of a 3-hydroxy-butyrate moiety 5 (X
is undefined) with a suitable tetrahydro-cis-fused pyranone,
which we represent for discussion purposes as 4 (Scheme 1). It
goes without saying that, in the merger of 4 and 5, the resident
sites of stereogenicity would be matched appropriately for
progression to 1. In the light of our concerns discussed above,
we are deliberately vague as to the form in which the future
D ring is presented in the step in which 4 and 5 merge. Indeed,
as will be seen, the proper timing in the fashioning of the
D ring is critical.
Aside from the uncertainties surrounding the D ring,
there was concern about the viability of the coupling step
itself. Insofar as it would involve the use of a C11-based
carbanion, there was a possibility of b elimination of the
oxygen atom of the ether-like ring, thus potentially compro-
mising the configurational robustness at C12 and even
threatening the feasibility of the central idea. Moreover,
there was concern that a base-induced merger of 4 and 5
might serve to equilibrate C8, which could undermine the
viability of a B:C cis fusion (see above).
This selection was influenced, to no small extent, by a
perception that it would be the most challenging of the
grandisines from the perspective of a stereoselective total
synthesis. It seemed that if a successful program to reach 1
could be realized, there would be a broader menu of options
for gaining access to other members of the grandisine family.
What makes the grandisine A congener unique is its back-
[*] Prof. S. J. Danishefsky
Laboratory of Bioorganic Chemistry
Memorial Sloan-Kettering Cancer Center
1275 York Avenue, Box 106
New York, NY 10021 (USA)
Fax: (+1)212-772-8691
E-mail: danishes@mskcc.org
and
Recognizing the B ring tetrahydropyranone substructural
motif in 4, we sought to exploit a Lewis acid catalyzed diene–
aldehyde cyclocondensation (LACDAC)[4] of an unusual sort.
Thus, one of the double bonds in the diene would be
contained in a suitably substituted D3(4) tetrahydropyridine
ring, the C3 atom of the piperidine ring would be joined to an
a-oxygenated vinyl group (see diene 2), and the heterodie-
nophile would be acetaldehyde. Somehow, kinetic protona-
tion of the enol derivative in cycloadduct 3 would have to
occur from the b face to afford the cis B:C junction (see 4).
The results from some initial experiments are important in
appreciating the strategy that was adopted and implemented.
First, as has been reported, the LACDAC reaction of 6 and
acetaldehyde yielded 7, in which cycloaddition of the
acetaldehyde had occurred in an anti fashion to the existing
Department of Chemistry
Columbia University
3000 Broadway
New York, NY 10027 (USA)
Dr. D. J. Maloney
Laboratory of Bioorganic Chemistry
Memorial Sloan-Kettering Cancer Center
1275 York Avenue
New York, NY 10021 (USA)
[**] This work was supported by the National Institutes of Health
(grant CA103823). We are especially thankful to Julie Grinstead and
Rebecca Lambert for assistance in the preparation of the manu-
script and Dr. Brendan Crowley, Dr. Jeremy May, and Dr. Emile
Velthuisen for helpful discussions during the course of this work.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2007, 46, 7789 –7792
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7789

Communications
to provide ketone 11. Finally, silylation of the methyl
ketone gave the required enol 12. The setting for
studying the fateful LACDAC reaction was now in
place.
Cyclocondensation of diene 12 with acetaldehyde
was conducted as shown. Under the conditions of our
experiment (Scheme 2), we were unable to detect any
reaction intermediates en route to 14. On the basis of
our observations, we would tend to characterize the
mechanism of this particular LACDAC reaction as a
cycloaddition rather than the other extreme possibil-
ity described many years ago: that is, aldolization
followed by heterocyclization.[9] Deprotection of the
silyl enol ether of crude 14 gave rise to the desired 15,
as essentially the only product (in 74% yield over two
steps). In addition to achieving the desired face
selectivity corresponding to axial addition during the
ꢀ
formation of the C7 O bond of 14, the configuration
at C12 was seen to be under tight kinetic control—
corresponding formally to endo addition with respect
to the methyl group, as proposed in our early
investigation two decades ago.[9] We attribute this
high selectivity to the more concerted-like cyclo-
addition mechanism in which the endo preference for
disposition of the methyl group in alignment 13 may
well reflect an exo preference for the catalytic system
activating the carbonyl group of the acetaldehyde.
The combination of BF3·OEt2 with the aldehydo-
carbonyl group could well be quite sterically demanding and
might, therefore, preferentially be integrated into ensemble
13 in an exo sense. This would direct the methyl group to be
incorporated in an endo manner. Moreover, the imposed
conformational lock occasioned by the NCbz group readily
accounts for the b-face protonation in the conversion of 14
into 15. This arrangement corresponds to a stereo-electroni-
Scheme 1. Strategy toward grandisine A. PG=protecting group, TIPS=triisopro-
pylsilyl, TPAP=tetrapropylammonium perruthenate, NMO=N-methylmorpho-
line-N-oxide.
five-membered precursor to the D ring.[5] While stereomodi-
fication at C12 proved to be possible, we were never able to
achieve the required correction of the configuration at C9
[Scheme 1, Eq. (1)].
Another influential finding resulted from hydrogenation
of 8. Oxidation of the C10 alcohol of the resultant crude
tetrahydroproduct afforded only a trans-fused B:C product
(9) [Scheme 1, Eq. (2)]. This result again demonstrates the
difficulties of generating the required B:C cis fusion with an
intact five-membered D ring.
From these hard-won lessons, an exciting idea for reaching
grandisine A presented itself. The cyclocondensation reaction
with acetaldehyde would be conducted on a seco version of
the D ring; that is, on 12 (Scheme 2). We envisioned that the
planar Cbz group on the nitrogen atom as well as the
siloxyvinyl group at C8 would tend to impose a strong
pseudoaxial bias upon the vinyl function at C9 (to minimize
A(1,3)- and A(1,2)-type abutments (see 13 in Scheme 2).[6] It was
also anticipated that, even with the vinyl group in an axial-like
orientation, the LACDAC reaction would occur from the
ꢀ
a face, with formation of the incipient C7 O bond in an axial
sense with respect to the B ring.
Scheme 2. Synthesis of 15. Reagents and conditions: a) CuI
(20 mol%), Me2S (10% v/v), vinyl-MgBr (1.0m THF), THF, ꢀ208C,
then acetaldehyde; b) TESCl, imidazole, CH2Cl2, 76% over 2 steps;
c) NaBH4, MeOH, 08C, 100%; d) MsCl, NEt3, CH2Cl2, 08C; e) cat. 10-
CSA, MeOH, 258C; f) oxalyl chloride, DMSO, (iPr)2NEt, CH2Cl2,
ꢀ788C, then DBU, reflux, 4 h, 66% over 4 steps; g) TIPSOTf, 2,6-
lutidine, CH2Cl2, 08C!258C, 97%; h) BF3·OEt2, acetaldehyde, Et2O,
ꢀ788C; i) TBAF, AcOH, THF, 258C, 74% over 2-steps.
A straightforward synthesis of 12, as shown in Scheme 2,
enabled the experimental evaluation of these concepts. The
route started from the readily available dihydropyridone 10.[7]
Nucleophilic vinylation followed by trapping with acetalde-
hyde led to diastereomeric aldol products. Progression from
the aldol adducts involved protection of the epimeric alcohols
as silyl ethers, reduction of the keto moiety, and activation of
the resultant alcohols as their mesylates.[8] Deprotection of
Cbz=benzyloxycarbonyl, TES=triethylsilyl, Ms=methanesulfonyl,
CSA=camphorsulfonic acid, DBU=1,8-diazabicyclo[5.4.0]undec-7-ene,
Tf =trifluoromethanesulfonyl, TBAF=tetrabutylammonium fluoride,
Bn=benzyl, LA=Lewis acid.
the side-chain alcohols and oxidation to the corresponding
ketone set the stage for b elimination of the mesyloxy group
7790 www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 7789 –7792

Angewandte
Chemie
cally preferred axial attack. In any case, the required array of
stereogenic centers encompassing carbon atoms 12, 7, 8, and 9
had been installed in 15.
magnitude of the optical rotation data [found: [a]D26
=
+ 180.1 cm3 gꢀ1 dmꢀ1 (c = 0.1 gcmꢀ3 in CH2Cl2), previously
reported: [a]2D3 = + 38.5 cm3 gꢀ1 dmꢀ1 (c = 0.1 gcmꢀ3 in
CH2Cl2)] to the synthetic material having a much higher
level of purity.[13,14] It is clear that the stereostructure of
grandisine A had been properly assigned and that its inau-
gural total synthesis was now complete.
With the desired rac-15 in hand, we addressed the
concluding phases of the synthesis. The next stage would
require annealing the A ring onto the pyranone. To avoid
issues associated with the coupling of a racemic material with
an enantiomerically defined entity, we readily obtained
enantiopure (+)-15 by resolution of its racemic precursor by
HPLC on a chiral support.[10] Next, the lithium enolate of 15
was coupled to (R)-3-(triethylsilyloxy)butanal (16)[11] medi-
ated by anhydrous zinc chloride (Scheme 3). The resultant b-
With the total synthesis of grandisine A accomplished,
and with intermediates en route to the natural products in
hand, we were in a position to evaluate some key thermody-
namic relationships. In this connection, we examined 15.
Equilibration of this compound with a base led to a 1:3 ratio
of the starting cis-fused 15 and trans-fused 21 (Scheme 4).
Scheme 3. Synthesis of grandisine A (1). Reagents and conditions:
a) LiHMDS, ZnCl2, THF, ꢀ788C, then 16, ꢀ788C!ꢀ508C, 3.5 h;
b) Dess–Martin periodinane, CH2Cl2; c) TFA, CH2Cl2, 73% over 3
steps; d) O3, MeOH, Sudan III (indicator), ꢀ788C, then Me2S,
ꢀ788C!258C; e) methyl (triphenylphosphoranylidene)acetate, ben-
zene, 608C!408C, 9.5 h, 80% over 2 steps; f) 10% Pd/C, H2 (1 atm),
MeOH; g) PhMe, reflux, 24 h, 98% over 2 steps; h) Lawesson’s
reagent, PhMe, 658C, 98%; i) Raney nickel (washed), THF, 258C, 94%.
HMDS=hexamethyldisilazide, TFA=trifluoroacetic acid.
Scheme 4. Epimerization of grandisine A.
This ratio corresponds at least roughly to the thermodynamic
equilibrium under these conditions. This was established by
re-equilibration of the purified major compound 21, where a
3:1 ratio of 21:15 was again obtained. Hence, it does seem
that, at the bicyclic level, the trans-fused epimer is more stable
than the cis, although by only a relatively small differential
(approximately 1 kcalmolꢀ1). In addition to equilibrating 15
to 21, under these conditions, a minor amount of 22 (21:15:22
ca. 3:1:1) was also observed. It is interesting to note that 22
hydroxyketone 17 was oxidized, and subsequent acid-cata-
lyzed deprotection of the silyl group led to dehydrative
cyclization, thereby providing 18 as a single antipode. Thus, in
practice, reaching ring A by annulation of an appropriately
matched b-hydroxybutyrate derivative could be realized in
spite of the potential difficulties discussed previously (see 22
below).
ꢀ
formally corresponds to b elimination of the C7 O bond via
the C8 enolate of 15 or 21. This possibility had been raised
above in connection with the proposed generalized conden-
sation of 4 and 5, but, in fact, does not occur under the very
mild conditions used in the coupling of 15 and 16.
There now remained the requirement of installing the
D ring from its seco precursor. Clearly, we would be exploit-
ing the vinyl group in this regard. In practice, the vinyl group
was cleaved by ozonolysis and the resultant aldehyde was
homologated through a Wittig-type condensation with methyl
(triphenylphosphoranylidene)acetate to afford the a,b-unsat-
urated ester 19.[12] Concurrent reduction of the disubstituted
double bond and cleavage of the Cbz group followed by
heating led to lactamization, resulting in formation of 20. In
the last stage of the synthesis, Lawessonꢀs reagent served to
convert 20 into the requisite thiolactam. Following reduction
with Raney nickel under carefully defined conditions, this
intermediate was converted into grandisine A (1). The
spectroscopic properties of fully synthetic 1 were identical
to those previously reported and, in any case, are independ-
ently conclusive. We attribute the significant difference in the
In addition, we also treated grandisine A (1) under
equilibrating conditions. Here, there was a clean conversion
into 8-epi-grandisine (23). As such, our early conjecture that
8-epi-grandisine is likely to be far more stable than 1 turned
out to be correct. Thus, it is clear that the total synthesis of
grandisine had been accomplished through kinetic control to
maintain the cis B:C ring fusion in its less stable form.
In summary, the total synthesis of grandisine A has been
accomplished. The defining step in the total synthesis was a
LACDAC reaction which exhibited stereo-electronic control
ꢀ
that favored axial addition in the formation of the C7 O
bond. Under these circumstances, endo addition was con-
trolled through preferential presentation of the catalytic
domain of ensemble 13 to the less hindered exo face.[15] It
appears that the course of this reaction was governed by a
conformational lock imposed on the bicyclic 13, wherein the
Angew. Chem. Int. Ed. 2007, 46, 7789 –7792
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7791

Communications
1984, 106, 2456; d) S. Danishefsky, E. R. Larson, D. Askin, J.
Am. Chem. Soc. 1982, 104, 6457; e) E. R. Larson, S. J. Dani-
shefsky, J. Am. Chem. Soc. 1982, 104, 6458; f) E. R. Larson, S.
Danishefsky, Tetrahedron Lett. 1982, 23, 1975; g) S. Danishefsky,
J. F. Kerwin, S. Kobayashi, J. Am. Chem. Soc. 1982, 104, 358;
h) S. Danishefsky, N. Kato, D. Askin, J. F. Kerwin, J. Am. Chem.
Soc. 1982, 104, 360.
vinyl group was coaxed into a strongly preferred pseudoaxial
orientation. We believe that there are significant take-home
lessons to be garnered from this now straightforward route to
grandisine A. Moreover, the newfound accessibility of the
natural product in its appropriate enantiomeric form enables
a medicinal chemistry program to probe structure–activity
relationships in an attempt to upgrade its binding affinity to
the d-opioid receptor.
[10] The relative stereochemistry was determined by extensive 2D
NMR analysis; see the Supporting Information for a detailed
procedure of the HPLC (chiral support) separation of the
enantiomers.
Received: July 19, 2007
Published online: September 17, 2007
[11] Prepared in two steps from commercially available ethyl (R)-3-
hydroxy butyrate: 1) TESCl, imidazole, CH2Cl2; 2) diisobutyla-
luminum hydride, CH2Cl2, ꢀ788C.
Keywords: allylic strain · cycloaddition · natural products ·
[12] A more direct route that involved a cross-metathesis of 18 with
methyl or ethyl acrylate was attempted; however, despite a
number of different catalysts and reaction conditions being used,
only starting material was recovered. Presumably this result is
due to the hindered nature of the vinyl group.
.
steric hindrance · total synthesis
[1] a) A. R. Carroll, G. Arumugan, R. J. Quinn, J. Redburn, G.
Guymer, P. Grimshaw, J. Org. Chem. 2005, 70, 1889; b) P. L.
Katavic, D. A. Venables, P. I. Forster, G. Guymer, A. R. Carroll,
J. Nat. Prod. 2006, 69, 1295.
[13] See the Supporting Information for the 1H and 13C spectra of the
authentic and synthetic natural products. The accuracy of our
polarimeter was verified with commercially available optically
pure compounds (R)-(+)-1,1’-bi-2-naphthol and (S)-(ꢀ)-4-
benzyl-2-oxazolidinone.
[2] Personal communication from Prof. Anthony Carroll (corre-
sponding author on Ref. [1a]).
[3] M. Williams, E. A. Kowaluk, S. P. Arneric, J. Med. Chem. 1999,
42, 1481.
[4] S. Danishefsky, Chemtracts: Org. Chem. 1989, 2, 273.
[14] At the time that we were planning the closing phases of the total
synthesis, the absolute configuration of naturally occurring
grandisine A was not known. By happenstance, we first used
the R form of 3-hydroxybutyrate, which in turn led to the
R aldehyde 16. Ultimately, this was coupled to both (+)-15 and
(ꢀ)-15. At that stage we had no way of knowing which one would
be opportune for merger with (R)-16 to attain the relative
configuration of grandisine A. As shown above, coupling with
(+)-15 led eventually to grandisine A (1). The same sequence of
steps starting with (ꢀ)-15 led to a final product whose 1H NMR
spectrum is similar, but clearly different from that of grandi-
sine A. This product is thus assigned the structure, 16-epi-ent-
grandisine A, the 1H and 13C spectra of which are provided in the
Supporting Information. Hence there can be no doubt that the
relative configuration of synthetic 1 corresponds to that of
natural grandisine A. It also seems likely (discrepancies in the
magnitude of dextrarotation notwithstanding) that the absolute
configuration of synthetic 1 also corresponds to that of natural
grandisine A.
[5] The exo:endo ratio for the LACDAC reaction of
6 and
acetaldehyde varied from about 2:1 to as high as 7:1; D. J.
Maloney, S. J. Danishefsky, Heterocycles 2007, 72, 167.
[6] a) F. Johnson, S. K. Malhotra, J. Am. Chem. Soc. 1965, 87, 5492;
b) F. Johnson, Chem. Rev. 1968, 68, 375; c) R. W. Hoffmann,
Chem. Rev. 1989, 89, 1841; d) C. E. Neipp, S. F. Martin, J. Org.
Chem. 2003, 68, 8867, and references therein.
[7] a) R. Sebesta, M. G. Pizzuti, A. J. Boersma, A. J. Minnaard, B. L.
Feringa, Chem. Commun. 2005, 1711; b) S. Knapp, C. Yang, S.
Pabbaraja, B. Rempel, S. Reid, S. G. Withers, J. Org. Chem. 2005,
70, 7715; c) R. B. Jagt, J. G. De Vries, B. L. Feringa, A. J.
Minnaard, Org. Lett. 2005, 7, 2433.
[8] Reactions after the aldol reaction were carried through as a
mixture of epimeric alcohols at C10 until 11.
[9] For a study and analysis of these mechanistic issues, see for
example a) S. Danishefsky, E. Larson, D. Askin, N. Kato, J. Am.
Chem. Soc. 1985, 107, 1246; b) S. J. Danishefsky, W. H. Pearson,
D. F. Harvey, J. Am. Chem. Soc. 1984, 106, 2455; c) S. J.
Danishefsky, W. H. Pearson, D. Harvey, J. Am. Chem. Soc.
[15] S. E. Denmark, B. R. Henke, E. Weber, J. Am. Chem. Soc. 1987,
109, 2512.
7792
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 7789 –7792