Total synthesis of (+)-isomigrastatin

Article abstract of DOI:10.1002/anie.200701837

(Chemical Equation Presented) Marginally stable natural products: The asymmetric total synthesis of the hydrolytically and thermally labile natural product (+)-isomigrastatin was demonstrated. The thermodynamic instability of a 2E-configured double bond in the context of this 12-membered macrolide was further demonstrated by phosphine-catalyzed isomerization to the 2Z configuration.

Full text of DOI:10.1002/anie.200701837

Communications
DOI: 10.1002/anie.200701837
Natural Product Synthesis
Total Synthesis of (+)-Isomigrastatin**
Isaac J. Krauss, Mihirbaran Mandal, and Samuel J. Danishefsky*
[
1]
Isomigrastatin (1), recently isolated from S. platensis, has
been a target for total synthesis in our laboratory. It is
[
1,2]
structurally related to migrastatin (2).
The ability of
migrastatins to retard cell migration might conceivably be
exploitable in inhibiting the progression of clinically manage-
able tumors to less containable metastatic states. Earlier, we
[
3a,b]
had described a successful inaugural total synthesis of 2
and how that effort had paved the way for a program in
[
4b,c]
diverted total synthesis (DTS).
DTS in turn enabled the
discovery of migrastatin-inspired structures which are con-
siderably more potent than 2 in suppressing colonization of
[
4]
tumors in vivo.
Our interest in isomigrastatin was further enhanced by its
chemical vulnerability. Thus, Shen and co-workers demon-
[
5a,b]
strated
that the ester (lactonic) linkage in 1 is rapidly
hydrolyzed, thereby leading to a family of previously known
biologically active systems, that is, the dorrigocins (Figure 1,
see bold black and gray arrows). Moreover, as very elegantly
shown, compound 1 undergoes extremely ready solvolytic and
thermally induced rearrangement to migrastatin itself
Figure 1. Isomigrastatin rearrangements (see text for details).
(
Figure 1, see thin arrows). Indeed, 2 may actually not be a
primary metabolite. Rather, it might well arise through a
biosynthesis pathway leading to isomigrastatin (1), which, in
turn, is transformed to 2 under isolation conditions.
Presumably, the ready transformation of 1 to 2 reflects the
fact that the two E-configured double bonds in the macro-
lactone are better accommodated in the context of a 14-
membered (cf. 2) rather than a 12-membered (cf. 1) macro-
lide. The pursuit of the total synthesis of compounds at the
margins of viability can be conducive to the discovery of new
[
6]
chemistry. Despite the deceptively simple allylic relation-
ship between 1 and 2, the present target presents a much
greater challenge and the route required to reach 1 had very
little homology with that used for 2.
As the lability of 1 probably arises from the strain inherent
in its trans–trans dienolide core, a responsive total synthesis
strategy (Figure 2) would seek to delay implementation of the
two E-configured double bonds until the latest possible stages
of the effort. Another area of challenge in reaching 1 by total
chemical synthesis is in the C11–C14 region. Of course, we
hoped to take advantage of lessons learned from our earlier
[
*] Dr. I. J. Krauss, Prof. S. J. Danishefsky
The Laboratory for BioorganicChemistry
Sloan-Kettering Institute for Cancer Research
1275 York Avenue, New York, NY 10021 (USA)
3
total synthesis of 2. However, the sp -configurational infor-
mation in migrastatin (2), from C11 outward, is vested solely
Fax: (+1)212-772-8691
E-mail: s-danishefsky@ski.mskcc.org
Prof. S. J. Danishefsky
The Department of Chemistry
Columbia University
Havemeyer Hall, New York, NY 10027 (USA)
Dr. M. Mandal
Schering-Plough Research Institute
2015 Galloping Hill Road, Kenilworth, NJ 07033 (USA)
[
**] Support for this work was provided by the National Institutes of
Health (5 R01 CA103823) I.J.K. is grateful for an NIH postdoctoral
fellowship (5 F32 AI063976). M.M. is grateful for a US Army grant
(
DAMD 17–03-1-0444). The authors thank Prof. Ben Shen and
members of his group for helpful discussions, spectra, and an
authenticsample for co mparison. We also thank Dr. Luis Todaro,
who is supported by the NIH-RCMI (RR-03037), for X-ray crystal-
lographicwork.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Figure 2. Syntheticstrategy to 1 (P: protecting group).
5
576
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 5576 –5579

Angewandte
Chemie
in the vicinal C13–C14 relationship. Accordingly, the migras-
[
3]
tatin experience does not inform as to the means for
management of the relatively remote C11–C14 stereocon-
nectivity in 1 flanking the C12–C13 E-configured double
bond. In Figure 2 is adumbrated the logic we hoped to employ
in organizing the stereochemistry and functionality of the
isomigrastatin, beyond C11. In the key stereo-defining step
(
see structure 3), chirality information from a C11–C12
oxirane and possibly from C15 would hopefully govern the
stereochemical sense of nucleophilic attack at C14, thereby
establishing the C11–C14 stereoconnectivity. With its stereo-
guidance role (if any) accomplished, the C15 oxygen atom
would eventually be converted to the ketone level (see
structure 5) in the context of paving the way for the ring-
closure phase of the synthesis to generate the dienolide en
route to 1.
Scheme 2. Fragment coupling. Reagents and conditions: a) LiBH4,
THF/H O; b) Ac O, pyridine; c) MOMCl, iPr NEt; d) K CO , MeOH;
2
2
2
2
3
e) (COCl) /DMSO, ꢀ788C, then iPr NEt; f) 15, DMSO/CHCl , room
2
2
3
temperature, 18 h. MOM: methoxymethyl.
The initial steps of the venture, drawn from our migras-
[
3a]
tatin synthesis occurred smoothly (Scheme 1). LACDAC
reaction of 6 and 7 led to 8 and thence to 9. Following Luche
[
7]
[8]
reduction and aqueous Ferrier rearrangement, as shown,
1
4
5
0 was in hand. Epoxidation of 10 afforded the a-oxirane 11 in
3% yield (after recrystallization) along with approximately
[
9]
% of the b-oxiranyl isomer (not shown). In a route
featuring maximum convergence, the known Wittig reagent
[
10]
[11]
1
3
served to alkylidenate the known aldehyde 12.
Hydrogenation of 14 afforded the appropriately functional-
ized Wittig reagent 15.
Scheme 3. Matched–mismatched S 2’ epoxide opening. Reagents and
N
In principle, coupling of 11 with 15 would be ultimately
convergent in reaching the required series. In practice, this
conditions: a) 1.5 equiv (S)-Me-CBS, 1.1 equiv BH -Me S, then TBSCl,
3
2
DMAP, Et N; b) MeCuCNLi, Et O, ꢀ208C!RT (see Ref. [14]). (S)-Me-
3
2
[
12]
union could not be effected in our hands, thus obliging us to
proceed in a more circumspect manner. Reduction of the
lactol arrangement in 11 was accomplished (Scheme 2) with
lithium borohydride, leading to alcohol 16. This compound
could be converted into its C9 MOM ether, 17. Oxidation
followed by coupling with phosphorane 15 afforded the
required aldehyde 18.
CBS: (S)-methyl oxazaborolidine; TBS: tert-butyldimethylsilyl; DMAP:
-dimethylaminopyridine.
4
corresponding mixture of TBS ethers 19 and 20 (Scheme 3).
[
15]
In the event, both components of the mixture underwent
clean S 2’ displacement with lithium cyanomethylcuprate
N
[
13]
Precedent from the pioneering work of Marshall et al.
with high selectivity for attack anti to the C15 OTBS group as
implied in Figure 2, conformer 3. Thus, unlike the case of
Marshall et al., the configuration at C15 is apparently a very
important stereochemical marker in determining the sense of
could have been taken to suggest S 2’ attack (Scheme 3) by a
N
first-order cyanocuprate at C14 would occur anti to the
epoxide, regardless of the configuration of the proximal C15
OTBS group. Enone 18 was reduced with sodium borohy-
dride to give a 1:1.2 mixture of epimeric allylic alcohols. This
step was followed by protection, as shown, to afford the
[
13]
the S 2’ substitution. This connectivity provided a strong
N
incentive to reach a particular (in this case R) C15 epimer
with stereocontrol. As indicated above, metal borohydride
Scheme 1. Fragment synthesis. Reagents and conditions: a) NaBH , CeCl -H O, MeOH, ꢀ20!08C, 2 h; b) CSA, THF/H O 9:1, reflux, 3 h;
4
3
2
2
c) mCPBA, K CO , CH Cl , 08C, 24 h; d) DMSO, room temperature; e) H , Pd/C, THF/MeOH, 16 h. TMS: trimethylsilyl; CSA: camphorsulfonic
2
3
2
2
2
acid; mCPBA: m-chloroperbenzoic acid; DMSO; dimethyl sulfoxide.
Angew. Chem. Int. Ed. 2007, 46, 5576 –5579
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5577

Communications
reduction drawing upon resident stereochemical biases had
led to a mixture of C15 epimers. Fortunately for our purpose,
agreed, within error, with that of the natural sample ([a] = +
D
170, CHCl , c = 0.18), confirming the absolute configuration
3
[
15]
the powerful reagent-controlled technology of Corey et al.
provided the required solution. In our case, use of the (S)-Me-
as drawn.
[
21]
Action of trimethylphosphine (Scheme 5) on synthetic
1 resulted in its complete conversion into 27, showing
rigorously that, as expected, the Z-configured 2,3 double-
[
15b]
CBS Corey catalyst
in the reduction of 18 afforded 19 with
good R selectivity (approximately 10:1).
Following addition of lithium cyanomethylcuprate to the
enriched mixture, purification was possible, providing 21 in
8
0% yield. After extensive exploration of macroannulation
strategies, it was determined that successful olefin metathesis
macrocyclization hinged on the absence of a double bond at
[
16]
C2–C3.
Acylation of the C11 alcohol 21 with racemic
[
17]
selenoacid 23 proceeded with substantial kinetic resolution,
providing 24 as an approximately 8:1 mixture of inseparable
C2 epimers (Scheme 4). After removal of the TBS protecting
Scheme 5. Isomerization of 1 to the 2,3-cis isomer.
[
22]
bond linkage is more stable than the E variant.
This
stability order raises the question as to what factors are
responsible for selective E-olefin formation (in 26!1). We
would argue that the preference for a pro-E over a pro-Z syn
selenoxide transition state must be governed by minimization
of vicinal nonbonded (van der Waals) interactions. Appa-
rently, the transannular steric strain factors which eventually
make the E isomer less stable than the Z isomer are not felt at
[
23]
this early point along the reaction coordinate.
In summary, the inaugural total synthesis of the very
elusive isomigrastatin, while still a work in progress from a
process perspective (see the low yield of 26), illustrates
3
important principles in the control of sp -level stereochem-
istry and provides several pleasing examples of convergence.
The properties of the broad family of migrastatins continue to
stimulate research in our laboratory.
Received: April 25, 2007
Published online: June 21, 2007
Keywords: inhibitors · isomerization · metathesis ·
natural products · total synthesis
Scheme 4. Synthesis of (+)-isomigrastatin (1). Reagents and condi-
tions: a) (ꢁ)-23 (excess), EDCI, DMAP, (3 equiv each) CH Cl , 08C;
.
2
2
b) pyridine-HF (1.1:1 mol/mol), 408C; c) (COCl) , DMSO, ꢀ788C,
2
then iPr NEt; d) Me BBr, iPr NEt, ꢀ788C, then THF/NaHCO (aq);
2
2
2
3
e) 20 mol% Grubbs’ second-generation catalyst, toluene, 1108C,
2
min; f) mCPBA, ꢀ788C, then iPr NEt, ꢀ788C!RT. EDCI: 1-ethyl-3-
[1] E. J. Woo, C. M. Starks, J. R. Carney, R. Arslanian, L. Cadapan,
2
(
3’-dimethylaminopropyl)carbodiimide.
S. Zavala, P. Licari, J. Antibiot. 2002, 55, 141.
[
2] a) K. Nakae, Y. Yoshimoto, T. Sawa, Y. Homma, M. Hamada, T.
Takeuchi, M. Imoto, J. Antibiot. 2000, 53, 1130; b) K. Nakae, Y.
Yoshimoto, M. Ueda, T. Sawa, Y. Takahashi, H. Naganawa, T.
Takeuchi, M. Imoto, J. Antibiot. 2000, 53, 1228; c) Y. Takemoto,
K. Nakae, M. Kawatani, Y. Takahashi, H. Naganawa, M. Imoto,
J. Antibiot. 2001, 54, 1104; d) H. Nakamura, Y. Takahashi, H.
Naganawa, K. Nakae, M. Imoto, M. Shiro, K. Matsumura, H.
Watanabe, T. Kitahara, J. Antibiot. 2002, 55, 442.
group, oxidation at C15, and MOM deprotection, ring-closing
metathesis of 25 afforded the desired E-configured cyclized
product 26 in 21% isolated yield, along with 36% of the C6–
[
18]
C7 Z isomer (not shown). At this stage, the C2 epimers of
6 could be separated. Gratifyingly, oxidative deselenation of
either C2 epimer afforded isomigrastatin (1) with very high
2
[
3] a) C. Gaul, J. T. Njardarson, S. J. Danishefsky, J. Am. Chem. Soc.
2003, 125, 6042 – 6043; b) C. Gaul, S. J. Danishefsky, Tetrahedron
Lett. 2002, 43, 9039 – 9042.
[
19] 1
13
selectivity for the E geometry at C2–C3.
H and C NMR
spectra of synthetic 1 were identical with those obtained from
[
20]
[4] a) D. Shan, L. Chen, J. T. Njardarson, C. Gaul, X. Ma, S. J.
a naturally derived sample of isomigrastatin.
Danishefsky, X.-Y. Huang, Proc. Natl. Acad. Sci. USA 2005, 102,
Moreover, synthetic and natural 1 exhibited identical
behavior on TLC and provided the same product distribution
under the hydrolytic conditions reported by Shen and co-
3772 – 3776; b) C. Gaul, J. T. Njardarson, D. Shan, D. C. Dorn,
K. D. Wu, W. P. Tong, X. Y. Huang, M. A. S. Moore, S. J.
Danishefsky, J. Am. Chem. Soc. 2004, 126, 11326 – 11337;
c) J. T. Gaul, C. Njardarson, D. Shan, X. Y. Huang, S. J.
Danishefsky, J. Am. Chem. Soc. 2004, 126, 1038 – 1040.
[
5a,b]
workers,
as monitored by LC/MS analysis. Finally, the
optical rotation of synthetic 1 ([a] = + 178, CHCl , c = 0.18)
D
3
5
578
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 5576 –5579

Angewandte
Chemie
[
5] a) J. H. Ju, S. K. Lim, H. Jiang, B. Shen, J. Am. Chem. Soc. 2005,
27, 1622 – 1623; b) J. H. Ju, S. K. Lim, H. Jiang, J. W. Seo, B.
[14] Compound 20 was subjected to the cuprate S 2’ reaction as a
N
1
mixture with 19, from which it was virtually inseparable.
Shen, J. Am. Chem. Soc. 2005, 127, 11930 – 11931; c) J. H. Ju,
S. K. Lim, H. Jiang, J. W. Seo, Y. Her, B. Shen, Org. Lett. 2006, 8,
However, the corresponding products 21 and 22 were easily
separated and isolated with a combined yield of 75%.
5
865 – 5868.
[15] a) E. J. Corey, R. K. Bakshi, S. Shibata, J. Am. Chem. Soc. 1987,
109, 5551 – 5553; b) E. J. Corey, C. J. Helal, Angew. Chem. 1998,
[
[
6] S. J. Danishefsky, J. Schkeryantz, Synlett 1995, 475 – 490.
7] A. L. Gemal, J. L. Luche, J. Am. Chem. Soc. 1981, 103, 5454 –
110, 2092 – 2118; Angew. Chem. Int. Ed. 1998, 37, 1986 – 2012;
c) E. J. Corey, C. K. Bakshi, S. Shibata, C. P. Chen, V. K. Singh, J.
Am. Chem. Soc. 1987, 109, 7925 – 7926.
5
459.
[
[
8] R. J. Ferrier, J. Chem. Soc. 1964, 5443.
9] A trace quantity of the a-oxiranyl b-anomeric lactol was also
observed (see Supporting Information).
10] D. B. Denney, J. Song, J. Org. Chem. 1964, 29, 495.
11] Y. Egawa, T. Okuda, M. Suzuki, Chem. Pharm. Bull. 1963, 11,
[
16] Numerous ring-closing metathesis strategies failed to give the
required (2,3),(6,7)-dienolide directly, whether closure was
attempted at the 2,3-or 6,7 p- ositions. This was likely due to
decomposition of this unstable motif under metathesis condi-
tions.
[
[
5
89.
[
[
17] See the Supporting Information for preparation of 24.
18] Integration of crude HNMR signals showed a 1.6:1 ratio
[
12] There is precedent for Wittig and Horner–Emmons couplings
directly with 2,3-epoxylactols: see J. E. Harvey, S. A. Raw,
R. J. K. Taylor, Org. Lett. 2004, 6, 2611 – 2614. The difficulty in
our case may have been due to the presence of an additional
methyl group adjacent to the anomeric carbon center.
13] a) J. A. Marshall, J. D. Trometer, B. E. Blough, T. D. Crute, J.
Org. Chem. 1988, 53, 4274 – 4282; b) J. A. Marshall, B. E.
Blough, J. Org. Chem. 1990, 55, 1540 – 1547; c) J. A. Marshall,
B. E. Blough, J. Org. Chem. 1991, 56, 2225 – 2234. A substantial
matched–mismatched effect is seen in Marshallꢀs example
1
favoring the Z geometry at C6–C7 for each C2 epimer. Future
efforts will be directed toward enhancement of yield and
selectivity. For a promising direction, see: A. Fꢁrstner, K.
Radkowski, C Wirtz, R. Goddard, C. W. Lehmann, R. Mynott, J.
Am. Chem. Soc. 2002, 124, 7061 – 7069.
[
[
[
19] The major C2 epimer eliminated along C2–C3 to give a
separable 9:1 mixture of E and Z products. The minor C2
epimer afforded exclusively the E isomer.
20] Natural material was graciously provided by B. Shen and co-
workers.
(
below). In the case of 20, the mismatch is enough to overturn
selectivity to a syn S 2’ result, possibly owing to the larger size of
the glutarimide appendage (R’, Scheme 3) relative to the methyl
group in Marshallꢀs system.
N
[21] Previous examples of phosphine-catalyzed E–Z equilibration:
a) C. Zhang, X. Lu, J. Org. Chem. 1995, 60, 2906 – 2908; b) S.
Ganguly, D. M. Roundhill, J. Chem. Soc. Chem. Commun. 1991,
639 – 640.
[
22] This isomerization could also be effected by excessive exposure
to silica gel or weak nucleophiles such as pyridine. The silica-
promoted rearrangement also afforded a substantial quantity of
migrastatin (2).
[
23] For a discussion of the Curtin–Hammet principle, which is
relevant to this discussion, see: J. I. Seeman, Pure Appl. Chem.
1987, 59, 1661 – 1672.
Angew. Chem. Int. Ed. 2007, 46, 5576 –5579
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5579