Total synthesis of a 28-member stereoisomer library of murisolins

Article abstract of DOI:10.1021/ja061801q

Total syntheses of two 16-member libraries of murisolin isomers are reported. In the first library, fluorous PMB (p-methoxybenzyl) groups encode configurations, and four mixtures of four dihydroxytetrahydrofurans are prepared by Shi epoxidation followed (

Full text of DOI:10.1021/ja061801q

Published on Web 06/28/2006
Total Synthesis of a 28-Member Stereoisomer Library of
Murisolins
Dennis P. Curran,* Qisheng Zhang, Cyrille Richard, Hejun Lu,
Venugopal Gudipati, and Craig S. Wilcox
Contribution from the Department of Chemistry, UniVersity of Pittsburgh,
Pittsburgh, PennsylVania 15260
Received March 23, 2006; E-mail: curran@pitt.edu
Abstract: Total syntheses of two 16-member libraries of murisolin isomers are reported. In the first library,
fluorous PMB (p-methoxybenzyl) groups encode configurations, and four mixtures of four dihydroxy-
tetrahydrofurans are prepared by Shi epoxidation followed (optionally) by Mitsunobu reaction. The mixtures
are coupled by Kocienski-Julia reaction with a single hydroxybutenolide followed by hydrogenation.
Demixing and detagging provide the 16 pure stereoisomers. In the second synthesis, a single mixture of
four fluorous-tagged dihydroxy-tetrahydrofurans is coupled with a four-compound mixture of hydroxy-
butenolides that bear derivatives of DMB (dimethoxybenzyl) groups with oligoethylene glycol (OEG) units
that encode the configurations at C4 and C34. The 16-compound mixture is subjected to hydrogenation,
double demixing, and detagging to provide the 16 isomerically pure murisolins. Twelve of these isomers
are new, while four match samples from the first library.
Introduction
A collection of some or all possible stereoisomers of a given
constitutional isomer of an organic compound is called a
stereoisomer library.1 In the field of natural products chemistry,
stereoisomer libraries are of high interest for studying stereo-
structure/activity relationships (SSAR). They can also be useful
for structure assignments of natural products whose stereo-
isomers can reasonably be expected to share similar or even
identical spectra. However, studies of stereoisomer libraries have
largely been limited to sugars and other fundamental bio-
molecule building blocks.2 This is because stereoisomer libraries
are too difficult to make by using current solution phase
synthetic technology.
With hundreds of members, the annonaceous acetogenins
comprise a large class of fatty acid derived natural products
that are isolated from members of the Annonaceae (custard-
apple) family.3 Most acetogenins sort into mono-THF or bis-
THF classes depending on whether they have one or two
dihydroxyalkyl-substituted tetrahydrofuran rings (Figure 1). The
two long chain alkyl substituents (RR or RR′) are of variable
lengths and are oxidized in assorted ways (alkenes, alcohols,
Figure 1. Dihydroxy mono- and bis-tetrahydrofuran classes of acetogenins.
ketones). There is often a butenolide ring terminating one chain,
while the other is terminated by a methyl group.
The interesting structures and potent antitumor or pesticidal
activities of acetogenins make them appealing synthetic targets.
(3) Reviews: (a) Zafra-Polo, M. C.; Gonza´lez, M. C.; Estornell, E.; Sahpaz,
S.; Cortes, D. Phytochemistry 1996, 42, 253-271. (b) Zafra-Polo, M. C.;
Figade`re, B.; Gallardo, T.; Tormo, J. R.; Cortes, D. Phytochemistry 1998,
48, 1087-1117. (c) Zeng, L.; Ye, Q.; Oberlies, N. H.; Shi, G.; Gu, Z.-M.;
He, K.; McLaughlin, J. L. Nat. Prod. Rep. 1996, 13, 275-306. (d) Cave´,
A.; Figade`re, B.; Laurens, A.; Cortes, D. In Progress in the Chemistry of
Organic Natural Products; Herz, W., Kirby, G. W., Moore, R. E., Seglich,
W., Tamm, Ch., Eds.; Springer-Verlag: Wien, 1997; Vol. 70, pp 81-288.
(e) Alali, F. Q.; Liu, X.-X.; McLaughlin, J. L. J. Nat. Prod. 1999, 62,
504-540. (f) Tormo, J. R.; Gallardo, T.; Gonza´lez, M. C.; Bermejo, A.;
Cabedo, N.; Andreu, I.; Estornell, E. Curr. Top. Phytochem. 1999, 2, 69-
90. (g) Bermejo, A.; Figade`re, B.; Zafra-Polo, M. C.; Barrachina, I.;
Estornell, E.; Cortes, D. Nat. Prod. Rep. 2005, 22, 269-303. (h) Ramirez,
E. A.; Hoye, T. R. In Studies in Natural Products Chemistry, Vol. 17;
Rahman, A., Ed.; Elsevier: Amsterdam, 1995; pp 251-282.
(1) Zhang, Q. S.; Lu, H. J.; Richard, C.; Curran, D. P. J. Am. Chem. Soc.
2004, 126, 36-37.
(2) For representative serial or solid-phase syntheses of multiple stereoisomers
of natural products or related compounds, see: (a) Busch, J.; Grether, Y.;
Ochs, D.; Sequin, U. J. Nat. Prod. 1998, 61, 591-597. (b) Larsson, M.;
Nguyen, B.-V.; Ho¨gberg, H.-E.; Hedenstro¨m, E. Eur. J. Org. Chem. 2001,
2, 353-363. (c) Takahashi, T.; Kusaka, S.; Doi, T.; Sunazuka, T.; Omura,
S. Angew. Chem., Int. Ed. 2003, 42, 5230-5234. (d) Tietze, L. F.;
Rackelmann, N.; Sekar, G. Angew. Chem., Int. Ed. 2003, 42, 4254-4257.
(e) Mulzer, J.; Kaselow, U.; Graske, K.-D.; Ku¨hne, H.; Sieg, A.; Martin,
H. J. Tetrahedron 2004, 60, 9599-9614. (f) Harrison, B. A.; Pasternak,
G. W.; Verdine, G. L. J. Med. Chem. 2003, 46, 677-680. (g) Crossman,
J.; Perkins, M. V. J. Org. Chem. 2006, 71, 177-124.
9
10.1021/ja061801q CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 9561-9573
9561

A R T I C L E S
Curran et al.
establish the absolute configuration. By using similar methods,
the structure of murisolin was refined to 1.147b and the structure
of murisolin A was narrowed to two candidates, 1.10 and 1.13.
These compounds show potent (sometimes extremely potent)
cytotoxic activities against several human cancer cell lines and
act by inhibition of complex I (NADH/ubiquinone oxidoreduc-
tase) in mitochondrial electron transport systems.3,7,8
Commensurate with the structural and biological interest,
there has been much synthetic work directed toward members
of both the mono-THF and bis-THF classes of acetogenins.3,9
Among others, the approaches of Sinha10 and Tanaka11 are well
designed for flexibility in stereoisomer synthesis. Through an
efficient plan that diverges from common intermediates as late
as possible to minimize work, the Sinha group has synthesized
36 isomers of a bis-THF-lactone that are suitably functionalized
to make all 64 isomers of the bis-dihydroxy-THF fragment of
these acetogenins.10 But the inexorable consequences of diver-
gence catch up sooner or later; many reactions will be needed
to convert the 36 intermediates in a serial or parallel fashion to
the 64 possible acetogenin products.
Roughly concurrent with our initial report1 of the synthesis
of a 16-member library including all the proposed structures of
the murisolins, Tanaka reported the synthesis of murisolin.12a
More recently, he has also described the synthesis of the
assigned structure of 16,19-cis-murisolin.12b
Figure 2. Structures of the (4R,34S)-murisolin family of acetogenins with
CIP designations of configurations for C15,16,19,20.
We have recently introduced solution phase mixture synthesis
with separation tagging as a means to leverage individual, serial,
and parallel syntheses of organic molecules. In the first approach
called “fluorous mixture synthesis” (FMS),13,14 incrementally
larger fluorous tags (Rf ) -(CF2)nCF3) were used to encode
either configuration or substituent information.15 This tagging
allows syntheses to be conducted on mixtures yet still provides
individual pure compounds of unambiguous structure in the end
because the mixtures can be separated (and hence decoded) by
But assigning stereostructures in acetogenins can be difficult,
and to date there are very few X-ray crystal structures reported
for these waxy compounds.3d While solving connectivities and
assigning stereostructures of isolated THF and hydroxybuten-
olide fragments are problems amendable to modern spectro-
scopic techniques,4 it can be very difficult to deduce (1) what
the configurations of isolated fragments are relative to each other
and (2) which long alkyl chain (RR or RR′) is on which hydroxy-
bearing carbon of the dihydroxy-THF ring. The elegant assign-
ments of uvaricin, bullaticin, and related bis-THF acetogenins
by McLaughlin, Hoye and co-workers show that rigorous
structure proofs by appropriate derivatizations are possible, but
substantial effort is involved.5
(8) The following papers report isolation and/or biological testing of muriso-
lin: (a) Tormo, J. R.; Royo, I.; Gallardo, T.; Zafra-Polo, M. C.; Hernandez,
P.; Cortes, D.; Pelaez, F. Oncology Res. 2003, 14, 147-154. (b) Mootoo,
B. S.; Ali, A.; Khan, A.; Reynolds, W. F.; McLean, S. J. Nat. Prod. 2000,
63, 807-811. (c) Tormo, J. R.; Gallardo, T.; Arago, R.; Cortes, D.;
Estornell, E. Chem.-Biol. Interact. 1999, 122, 171-183. (d) Jiang, Z.; Yu,
D.-Q. J. Nat. Prod. 1997, 60, 122-125. (e) Liaw, C.-C.; Chang, F.-R.;
Chen, S.-L.; Wu, C.-C.; Lee, K.-H.; Wu, Y.-C. Bioorg. Med. Chem. 2005,
13, 4767-4776.
(9) (a) Figade`re, B. Acc. Chem. Res. 1995, 28, 359-365. (b) Hoppe, R.; Scharf,
H. D. Synthesis 1995, 1447-1464. (c) Figade`re, B.; Cave´, A. In Studies in
Natural Products Chemistry; Atta-ur-Rahman, Ed.; Elsevier: Amsterdam,
1996; Vol. 18, pp 193-227. (d) Marshall, J. A.; Hinkle K. W.; Hagedorn,
C. E. Isr. J. Chem. 1997, 37, 97-107. (e) Casiraghi, G.; Zanardi, F.;
Battistini, L.; Rassu, G. Chemtracts: Org. Chem. 1998, 11, 803-827.
(10) (a) Sinha, S. C.; Sinha, A.; Yazbak, A.; Keinan, E. J. Org. Chem. 1996,
61, 7640-7641. (b) Avedissian, H.; Sinha, S. C.; Yazbak, A.; Sinha, A.;
Neogi, P.; Sinha, C.; Keinan, E. J. Org. Chem. 2000, 65, 6035-6051. (c)
Das, S.; Li, L.-S.; Abraham, S.; Chen, Z.; Sinha, S. C. J. Org. Chem. 2005,
70, 5922-5931.
(11) (a) Maezaki, N.; Kojima, N.; Tominaga, H.; Yanai, M.; Tanaka, T. Org.
Lett. 2003, 5, 1411-1414. (b) Kojima, N.; Maezaki, N.; Tominaga, H.;
Asai, M.; Yanai, M.; Tanaka, T. Chem.sEur. J. 2003, 9, 4980-4990. (c)
Kojima, N.; Maezaki, N.; Tominaga, H.; Yanai, M.; Urabe, D.; Tanaka, T.
Chem.sEur. J. 2004, 10, 672-680. (d) Watanabe, T.; Tanaka, Y.; Shoda,
R.; Sakamoto, R.; Kamikawa, K.; Uemura, M. J. Org. Chem. 2004, 69,
4152-4158. (e) Makabe, H.; Hattori, Y.; Kimura, Y.; Konno, H.; Abe,
M.; Miyoshi, H.; Tanaka, A.; Oritani, T. Tetrahedron 2004, 60, 10651-
10657.
The murisolins are an important group of three mono-THF
acetogenins (Figure 2) whose structures and the attendant
assignment problems are representative of many acetogenins.
Murisolin was isolated from the seeds of Annona muricata by
Cave´ and co-workers in 1990.6 These workers assigned the
1
relative configuration of the dihydroxy-THF ring by H NMR
spectroscopy studies. Five years later, McLaughlin again isolated
murisolin, this time from Asmina triloba, along with 16,19-cis-
murisolin and murisolin A.7,8 The structure of 16,19-cis-
1
murisolin was assigned as 1.16 by a combination of H NMR
studies on the natural product to establish the relative config-
uration of the dihydroxy-THF ring and Mosher ester studies to
(4) Dihydroxy-mono-THF rings: (a) Gale, J. B.; Yu, J.-G.; Khare, A.; Hu, X.
E.; Ho, D. K.; Cassady, J. M. Tetrahedron Lett. 1993, 34, 5851-5854.
Dihydroxy-bis-THF rings: (b) Hoye, T. R.; Suhadolnik, J. C. J. Am. Chem.
Soc. 1987, 109, 4402-4403. (c) Hoye, T. R.; Zhuang, Z. J. Org. Chem.
1988, 53, 5578-5580.
(12) (a) Maezaki, N.; Tominaga, H.; Kojima, N.; Yanai, M.; Urabe, D.; Ueki,
R.; Tanaka, T.; Yamori, T. Chem. Commun. 2004, 406-407. (b) Maezaki,
N.; Tominaga, H.; Kojima, N.; Yanai, M.; Urabe, D.; Tanaka, T. Chem.s
Eur. J. 2005, 11, 6237-6245.
(5) Rieser, M. J.; Hui, Y.; Rupprecht, J. K.; Kozlowski, J. F.; Wood, K. V.;
McLaughlin, J. L.; Hansen, P. R.; Zhuang, Z.; Hoye, T. R. J. Am. Chem.
Soc. 1992, 114, 10203-10213.
(13) Luo, Z. Y.; Zhang, Q. S.; Oderaotoshi, Y.; Curran, D. P. Science 2001,
291, 1766-1769.
(6) Myint, S. H.; Laurens, A.; Hocquemiller, R.; Cave´, A.; Davoust, D.; Cortes,
D. Heterocycles 1990, 31, 861-867.
(14) Short reviews: (a) Zhang, W. ArkiVoc 2004, 101-109. (b) Curran, D. P.
In The Handbook of Fluorous Chemistry; Gladysz, J. A., Curran, D. P.,
Horva´th, I. T., Eds.; Wiley-VCH: Weinheim, 2004; pp 128-156.
(7) Woo, M. H.; Zeng, L.; Ye, Q.; Gu, Z.-M.; Zhao, G.-X.; McLaughlin, J. L.
Bioorg. Med. Chem. Lett. 1995, 5, 1135-1140.
9
9562 J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006

Synthesis of a Murisolin Stereoisomer Library
A R T I C L E S
Figure 3. Strategy for murisolin fluorous mixture synthesis.
fluorous chromatography.16 This type of tag-based separation
with decoding is called “demixing” or “sorting”. More recently,
Wilcox and Turkyilmaz have introduced oligoethylene glycol
mixture synthesis (OEGMS).17 Here, it is the OEG unit (CH2-
CH2O) that is incrementally varied for coding and separation.
We have previously communicated the synthesis of a 16-
member stereoisomer library of murisolins by using fluorous
mixture synthesis,1 and we describe herein the full details of
this work. The fluorous mixture synthesis provided four mixtures
of four-tagged murisolins, which were demixed and detagged
to provide the final target compounds. We have also com-
municated the union of fluorous mixture synthesis and oligo-
ethylene glycol mixture synthesis to provide a second 16-
member library of doubly tagged murisolins, this time as a single
mixture.18 Double demixing and detagging then provided 4
previously made and 12 new murisolin isomers. We describe
herein the full details of the fluorous and double mixture
components of this work.
Having access to 24 of the 32 possible diastereomers provides
a complete picture of spectroscopic similarities and differences
among the murisolin isomers. In a forthcoming paper,19 we will
provide a detailed comparison of these spectra and other
characterization data. This comparison answers existing ques-
tions and raises new ones. Briefly, we rigorously show that the
structure of murisolin (1.14) is correct but suggest that the
structure of 16,19-cis-murisolin (1.16) is probably wrong and
provide a likely alternative (1.8). We also suggest structure 1.10
for murisolin A.
Results and Discussion
Fluorous Mixture Synthesis, Strategy: Murisolin has six
stereocenters, so its complete stereoisomer library has 64
members grouped as 32 pairs of enantiomers. In the first of
two mixture synthesis exercises, we targeted all 16 possible
stereoisomers of murisolin at the dihydroxy-THF portion of the
molecule with the stereocenters fixed in the hydroxybutenolide
fragment as 4R,34S. This library was prepared by a “mix/split”
strategy, as summarized in Figure 3. The long alkyl chain
between the two rings of protected murisolins 2{1-4,a-d}20
was dissected between C11 and C12 to provide a single
hydroxybutenolide 3 and 16 protected dihydroxy-THF fragments
4. The butenolide 3 bears an aldehyde, and the dihydroxy-THF
fragments 4 bear phenylsulfonyl tetrazoles for union by a
Kocienski-Julia coupling21 followed by hydrogenation.
The mix/split plan calls for the preparation of four mixtures
of protected murisolins 2 (by 2 splits), each mixture with one
of the four possible sets of configurations at C15 and C16. In
turn, each of these mixture samples comprises four compounds
with all possible configurations at C19 and C20 coded to
fluorous tags. Compounds that are otherwise stereoisomeric but
are not isomers because of differing Rf groups are called
quasienantiomers or quasidiastereomers or, more generally,
quasi-isomers.22 The four final mixtures of quasidiastereomers
are designated as Series 1-4 according to their configurations
at C15/C16, as shown in Figure 3. In this plan, the configuration
of each isomer is uniquely encoded by the mixture that it is in
and by the fluorous tag that it bears. Demixing of the four
(15) (a) Curran, D. P.; Furukawa, T. Org. Lett. 2002, 4, 2233-2235. (b) Zhang,
W.; Luo, Z.; Chen, C. H. T.; Curran, D. P. J. Am. Chem. Soc. 2002, 124,
10443-10450. (c) Zhang, Q. Ph.D. Thesis, University of Pittsburgh, 2003.
(d) Dandapani, S.; Jeske, M.; Curran, D. P. Proc. Nat. Acad. Sci. U.S.A.
2004, 101, 12008-12012. (e) Manku, S.; Curran, D. P. J. Comb. Chem.
2005, 7, 63-68. (f) Manku, S.; Curran, D. P. J. Org. Chem. 2005, 70,
4470-4473.
(20) (a) Fluorous- or OEG-tagged compounds are indicated by brackets after
numbers. In the fluorous mixture synthesis, the italic number indicates the
series (spatial encoding), and the bold letter indicates the fluorous PMB
tag. For example, 4{2,b} is compound 4 in Series 2 with fluorous tag b,
whereas 4{2,a-d} is a mixture of four compounds 4 in Series 2 with all
four fluorous tags a-d. In the double mixture synthesis, the plain text
number in the bracket is the OEG tag, and the bold letter is the fluorous
tag. (b) Individual stereoisomers of murisolin are indicated by numbers
after a decimal point (1.1...1.28). The order of the numbers after the decimal
follows relative configurations and not tags. Table 1 shows the complete
numbering and tagging scheme of both libraries of murisolins.
(16) Curran, D. P. In The Handbook of Fluorous Chemistry; Gladysz, J. A.,
Curran, D. P., Horva´th, I. T., Eds.; Wiley-VCH: Weinheim, 2004; pp 101-
127.
(17) Wilcox, C. S.; Turkyilmaz, S. Tetrahedron Lett. 2005, 46, 1827-1829.
(18) Wilcox, C. S.; Gudipati, V.; Lu, H.; Turkyilmaz, S.; Curran, D. P. Angew.
Chem., Int. Ed. 2005, 44, 6938-6940.
(19) Curran, D. P.; Zhang, Q.; Lu, H.; Gudipati, V. J. Am. Chem. Soc. 2006, in
press.
(21) Blakemore, P. R. J. Chem. Soc., Perkin Trans. 1 2002, 2563-2585.
(22) Zhang, Q. S.; Curran, D. P. Chem.sEur. J. 2005, 11, 4866-4880.
9
J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006 9563

A R T I C L E S
Curran et al.
Scheme 2. Synthesis of Fluorous-tagged Precursors 6a-d
Scheme 1. Synthesis of Hydroxybutenolide (4R,34S)-3
The enolate generated by treating lactone 928 with NaHMDS
was alkylated with the iodide 8 to afford 10 in 75% yield. The
sulfide 10 was directly converted to butenolide 11 in 70% yield
by oxidation with mCPBA and thermal elimination. Hydro-
boration/oxidation of the terminal olefin gave the primary
alcohol 12 in 70% yield. Finally, oxidation of 12 with PCC29
then gave aldehyde (4R,34S)-3 in 70% yield.
The minor (4S) enantiomer of alcohol 7 translated into a
diastereomeric impurity in hydroxylactone 3, which was an
inseparable ∼9/1 mixture of epimers as assayed by 13C NMR
spectroscopy. The spectra of epimers at C4 for 3 and subsequent
intermediates are very similar, so the diastereomeric impurity
was carried through the rest of the synthesis. This resulted in
minor diastereomeric impurities in each of the 16 final muriso-
lins. These were removed after detagging, as described in the
postmix section.
mixtures of 2 just prior to deprotection should unambiguously
provide all 16 stereoisomers 2{1,a} through 2{4,d}.
The plan calls for preparation of the four dihydroxy-THF
mixtures 4{1-4,a-d} from one four-compound mixture of
alkenes 5{a-d}. Splitting of this mixture in half and reagent-
controlled Shi epoxidation23 of 5{a-d} of each half with
enantiomeric fructose-derived ketones followed by closure of
the THF ring should provide the two C15/C16 erythro isomer
mixtures (Series 1 and 3). Division of these and Mitsunobu
reaction on half of each mixture should then provide the C15/
C16 threo isomer mixtures (Series 2 and 4). Finally, the
underlying individual component precursors 6a-d of mixture
5{a-d} are made by Brown allylboration reactions,24 followed
by fluorous tagging.
The four fluorous-tagged homoallyl alcohols 6 were made
by a Brown allylboration route,30 as shown in Scheme 2. The
MEM (methoxyethoxymethyl) ether of allyl alcohol was depro-
tonated by sec-BuLi to form the corresponding allylic anion,
which subsequently reacted with (+)-B-methoxydiisopino-
campheylborane ((+)-Ipc2BOMe) in the presence of BF3•Et2O
to generate the intermediate allylborane (not shown). Addition
of tridecanal (C12H25CHO) then provided the homoallylic
alcohol (19S,20S)-13 in 81% yield. The reaction was completely
Fluorous Mixture Synthesis, Premix Stage: The premix
stage involves synthesis of all the individual components of 3
and 6 that are needed for the mixture synthesis plan outlined in
Figure 3. Hydroxybutenolide (4R,34S)-3 is a known compound25
that was synthesized as summarized in Scheme 1. Sharpless
dihydroxylation26 of 1,8-nonadiene with AD-mix-â provided the
diol 7 in 60% yield with an enantiomer ratio of about 92/8.
The primary hydroxy group was selectively activated as a
tosylate in 85% yield by treatment with p-TsCl/Bu2SnO.27 The
secondary hydroxy group was protected as the TBS ether (85%
yield), and the tosylate was displaced by iodide to provide 8 in
90% yield.
(26) Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem. ReV. 1994,
94, 2483-2547.
(27) Martinelli, M. J.; Nayyar, N. K.; Moher, E. D.; Dhoket, U. P.; Pawlak, J.
M.; Vaidyanathan, R. Org. Lett. 1999, 1, 449-450.
(28) Sinha, S. C.; Sinha-Bagchi, A.; Yazbak, A.; Keinan, E. Tetrahedron Lett.
1995, 36, 9256-9260.
(23) (a) Shi, Y. Acc. Chem. Res. 2004, 37, 488-496. (b) Frohn, M.; Shi, Y.
Synthesis 2000, 1979-2000.
(29) Martin, T.; Soler, M. A.; Betancort, J. M.; Martin, V. S. J. Org. Chem.
1997, 62, 1570-1571.
(24) Brown, H. C.; Ramachandran, P. V. In AdVances in Asymmetric Synthesis;
Hassner, A., Ed.; Jai Press: Greenwich, 1995; Vol. 1, pp 147-210.
(25) Sinha and Keinan started from 1,9-decadiene and used a similar series of
steps terminating in oxidative cleavage to generate aldehyde 3. In contrast,
we used 1,8-nonadiene and generated the aldehyde by hydroboration and
oxidation.
(30) (a) Brown, H. C.; Jadhav, P. K.; Bhat, K. S. J. Am. Chem. Soc. 1988, 110,
1535-1538. (b) Jadhav, P. K.; Bhat, K. S.; Perumal, P. T.; Brown, H. C.
J. Org. Chem. 1986, 51, 432-439. (c) Nicolaou, K. C.; Groneberg, R. D.;
Miyazaki, T.; Stylianides, N. A.; Schulze, T. J.; Stahl, W. J. Am. Chem.
Soc. 1990, 112, 8193-8195.
9
9564 J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006

Synthesis of a Murisolin Stereoisomer Library
A R T I C L E S
Scheme 3. Synthesis of Shi Epoxidation Precursor Mixture
5{a-d}
Scheme 4. Shi Epoxidations of Series 1 and Series 3 Mixtures
diastereoselective (>98/2) and the enantiomeric excess of 13
was 95% as assayed by Mosher ester analysis. The configuration
of the hydroxyl group in (19S,20S)-13 was then inverted by a
Mitsunobu reaction followed by the hydrolysis of the inter-
mediate ester to provide the inverted alcohol (19S,20R)-14 in
70% yield. Similarly, the reaction of the allylborane derived
from (-)-Ipc2BOMe with tridecanal provided the enantiomer
(19R,20R)-13 (not shown). Another Mitsunobu-hydrolysis
sequence provided (19R,20S)-14 in 70% yield.
The four stereoisomeric alcohols were then selectively tagged
with four fluorous PMB bromides (BrCH2C6H4-p-O(CH2)3Rf),
which were synthesized from the corresponding alcohols.31 The
perfluorocarbon units (Rf) in the four fluorous tags are C2F5,
C4F9, C6F13, and C8F17. The alcohol (19S,20R)-14 was depro-
tonated with potassium tert-butoxide (t-BuOK) at -30 °C in
THF and reacted with fluorous PMB bromide bearing the C2F5
group to provide (19S,20R)-6a in 84% yield. Likewise, PMBF
ethers (19S,20S)-6b (Rf ) C4F9), (19R,20S)-6c (Rf ) C6F13),
and (19R,20R)-6d (Rf ) C8F17) were prepared. The structures
for these tagged precursors are shown in the lower part of
Scheme 2.
This was reacted with acetyl chloride to provide the key mixture
5{a-d} in 96% yield.
The THF ring and the stereocenters at C15 and C16 were
introduced along with the first split, as shown in Scheme 4.
The epoxidation of 5{a-d} by using the ketone 19 derived from
D-fructose followed the procedure reported by Shi and co-
workers.35 A stoichiometric amount of ketone 19 was used to
maximize the conversion and facial selectivity. The Series 1
epoxide mixture 20{1,a-d} was obtained in 81% yield together
with some recovered starting material 5{a-d}. The starting
material 5 was separated from the product 20 (81%) by flash
chromatography but was only partly separated from the ketone
19. The recovered alkene 5{a-d} containing some residual
ketone was subjected to the same reaction conditions to obtain
10% more epoxide 20{1,a-d} to give an overall yield of 91%.
The removal of the MEM group and the epoxide opening
were accomplished together by treating 20{1,a-d} with cam-
Fluorous Mixture Synthesis; Mixture Stage: The four-
tagged stereoisomers 6 were weighed and mixed to generate
the starting mixture 6{a-d} (Scheme 3). The molar ratio of
6a:b:c:d was approximately 1:1.1:1.2:1.2. Hydroboration of the
mixture 6{a-d} with BH3/THF in THF for 12 h followed by
treatment with hydrogen peroxide in aqueous NaOH for 2 h
afforded the desired primary alcohol 15{a-d} in 64% yield
(Scheme 3). The conversion of 15{a-d} to the corresponding
iodide 16{a-d} was accomplished by treatment with tri-
phenylphosphine and iodine in toluene.
In the key Negishi coupling,32 the zinc reagent generated by
(31) The fluorous PMB alcohols and fluorous silica products were purchased
from Fluorous Technologies, Inc. DPC holds an equity interest in this
company.
treating the iodide mixture 16{a-d} with t-BuLi in the presence
33
of dry ZnCl2 was transferred to a mixture of vinyl iodide 17
(32) Knochel, P.; Almena Perea, J. J.; Jones, P. Tetrahedron 1998, 54, 8275-
and Pd(PPh3)4 in THF.34 Following rapid flash chromatography,
18{a-d} was isolated in reproducible 52-61% yields. The
product of reductive deiodination of 16{a-d} (not shown) was
also isolated in trace amounts. The TBS protecting group in
18{a-d} was removed by TBAF to afford a primary alcohol.
8319.
(33) Smith, A. B.; LaMarche, M. J.; Falcone-Hindley, M. Org. Lett. 2001, 3,
695-698.
(34) The coupling of the acetate analogue of 17 to directly provide 5{a-d}
was also successful, but the resulting mixture contained impurities that were
difficult to remove by chromatography.
(35) Smith, A. B.; Brandt, B. M. Org. Lett. 2001, 3, 1685-1688.
9
J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006 9565

A R T I C L E S
Curran et al.
5{c}. A parallel relationship holds for the other two compounds
in 20{1,a-d} and for the four compounds in 20{3,a-d}.
We conducted a thorough analysis of the three epoxide
mixtures of 20 by LC NMR spectroscopy. Small samples of
each mixture of 20 (∼1 mg) were demixed on a FluoroFlash
PF-C8 column31 by eluting with a gradient of 80% CH3CN/
20% D2O up to 100% CH3CN over 30 min. Four peaks were
observed in the chromatogram of each sample due to the
expected separation based on the fluorous tag. Center cuts from
each of the peaks were sequentially pumped into the probe of
1
a 600 MHz NMR spectrometer, and H NMR spectra were
recorded by using standard techniques. The whole experiment,
including LC separation and subsequent collection of four FIDs,
took about 4 h to complete.
Expansions of the resulting 12 spectra in the region 4.20-
4.90 ppm are shown in Figure 5. This region of the spectrum
proved to be the most diagnostic, even though the resonances
arise from protecting group protons: the downfield peak(s) are
due to the benzylic methylene group (ArCH2O) of the PMBF
group, while the upfield peaks are due to the acetal methylene
group (OCH2O) of the MEM group. Spectra are listed in order
of compound elution from bottom (fastest eluting C2F5 tag) to
top (slowest eluting C8F17 tag).
These spectra validate the proposed identity of quasi-
enantiomers and the indexing in several ways. First, each control
spectrum in the (nonselective) mCPBA set is the sum of the
flanking two spectra in the Shi epoxidation sets. Consider the
spectra in line d (C8F17 tag). Compounds 20{1,d} (left spectrum)
and 20{3,d} (right spectrum) are diastereomers because they
were produced by epoxidation of the same alkene with D-
fructose- and L-fructose-derived ketones. Accordingly, their
spectra are different. The center spectrum from the nonselective
mCPBA oxidation should contain resonances from both of these
compounds, and indeed it is the sum of its flanking spectra.
The same analysis holds for the other three groups of three
spectra. So the Series 3 product 20{3,d} provides a control for
the true minor product of a Series 1 20{1,d} and vice versa.
However, one need not look to Series 3 for the controls because
the quasienantiomer of the minor product of 20{1,d} is indexed
in the library as the major product of 20{1,b}. Summing left
and right spectra b and d (or a and c) again provides the control
spectra in the center. Finally, the quasienantiomer of each major
product of Series 1 is present as one of the major products of
Series 3. For example, the quasienantiomer of 20{1,d} is
20{3,b}, and the corresponding spectra are identical. This type
of analysis holds true for the spectrum of every major product,
when compared with the appropriate quasienantiomer from its
own mixture or the true enantiomer from the other mixture.
Figure 4. Selected quasienantiomer relationships in the Shi epoxidation
product 20{1,a-d}.
phorsulfonic acid (CSA) in CH2Cl2 at room temperature for 2
h to give the substituted-THF mixture 21{1,a-d} in 81% yield.
Concurrently, we used the L-fructose ketone (ent-19) for the
epoxidation of 5{a-d} to generate the Series 3 mixture 20{3,a-
d} in 85% yield. Deprotective cyclization of this mixture by
treatment with CSA then afforded 21{3,a-d} in 83% yield.
We also treated 5{a-d} with mCPBA to obtain a mixture of
Series 1 and 3 mixtures 20{1,3,a-d} for reference.
The Shi epoxidation creates two new stereocenters. While
good selectivities are observed in related epoxidations of achiral
substrates, precedent23,33 suggests that it is unlikely that the
diastereoselectivity of this reaction with any of the four
substrates in 5{a-d} will approach 100%. Accordingly, both
mixtures of 20 are expected to contain four major and four minor
diastereomeric epoxides.
To assess the stereoselectivity of the Shi epoxidation, we
capitalized on a key “self-indexing” feature of fluorous mixture
synthesis. Each mixture is self-indexed because the expected
minor product of every substrate in the Shi epoxidation is
represented by the quasienantiomer of one of the other major
products. The quasienantiomers differ only in the number of
CF2 groups in the fluorous PMB group, and they are expected
to exhibit substantially identical spectra, even under high-
resolution conditions.22
Representative quasienantiomer relationships in the self-
indexed Series 1 20{1,a-d} mixture are shown in Figure 4.
Shi epoxidation of 5{a} with 19 is expected to give 20{1,a} as
the major product and 20{3,a} as the minor product (C2F5
series). Likewise, 5{c} gives 20{1,c} and 20{3,c} (C6F13 series).
The major product 20{1,a} of the first epoxidation is the
quasienantiomer of the minor product 20{3,c} of the second
and vice versa. Thus, to identify the minor product in the
epoxidation in a spectrum of 5{a}, we need only to look for
peaks identical to those of the major product in a spectrum of
Although these LC NMR experiments validate the principle
of self-indexing and prove the success of the early steps of the
mixture synthesis, they also leave unresolved the level of
stereoselectivity. While all of the epoxidations are clearly
stereoselective, the spectral resolution is not sufficient in any
case to quantify (or even estimate) the level of stereoselectivity.
This may be due to the inherently low resolution of the LC
NMR experiment (dilute samples, protic solvent).
Beyond resolution, there are two other problems in quantify-
ing selectivity. First, because the quasi-isomer separations are
so large, it is essential to use gradient conditions for the LC.
This results in recording successive spectra in different solvents.
9
9566 J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006

Synthesis of a Murisolin Stereoisomer Library
A R T I C L E S
1
Figure 5. Expansions of the LC NMR spectra (600 MHz, H, CH3CN/D2O) of epoxide mixtures 20, Series 1, Series 3, and Series 1,3 mixture.
So solvent-dependent chemical shifts are of concern, although
we did not observe any evidence of this in the spectra in Figure
5. Second, and more seriously, only the sample from the heart
of the LC peak is injected into the NMR probe. The broader
the peak, the smaller the amount of sample that is injected and
the more that is omitted. Because it is conceivable that
quasidiastereomers bearing the same fluorous tag could be
separated by fluorous HPLC, there is concern that the NMR
spectra that are recorded from an LC experiment might not be
representative of the bulk sample in that peak. A minor isomer
could precede the main component or lag behind it.
To obtain quantitative facial selectivities for the Shi epoxi-
dation and to further support the above analysis, we demixed
by fluorous HPLC a small quantity of mixture 21{1,a-d}
following closure of the THF ring. The four individual fractions
were concentrated, and 600 MHz 1H NMR spectra of the
resulting products were recorded in CDCl3. The whole experi-
ment takes about the same time as an LC NMR experiment
(∼4 h) but provides much better quality spectra.
Expansion of these spectra showing the diagnostic resonances
of the methylene group of the PMBF group (ArCH2O) (4.45-
4.75 ppm) are reproduced in Figure 6. As expected, the reso-
nances of each minor isomer are identical to those of its corre-
sponding quasienantiomer. For example, compare the boxed
minor resonances in 21{1,d} with the boxed major resonances
in 21{1,b}. These are identical because the minor product mixed
with 21{1,d} is the quasienantiomer of the major product
21{1,b}. The reverse is also true, and the same pair of relation-
ships hold with the a/c compounds. This proves that these minor
resonances arise from the Shi epoxidation reaction and allows
integration of the 21{1, a-d} spectra to provide the following
stereoselectivities: a, 93/7; b, 92/8; c, 93/7; d, 92/8. A
comparable experiment was conducted with the Series 3 mixture
21{3,a-d}, and similar stereoselectivities were observed. These
results show that, for all eight substrates in the two mixtures,
the configurations of the precursors are inconsequential; the Shi
epoxidation occurs with good reagent control (∼93/7).
Figure 6. Determining stereoselectivities by quasienantiomer analysis:
Expansions of the 600 MHz H NMR spectra (CDCl3) of the components
of THF mixture 21{1,a-d}.
1
(92%). This provides the Series 1 primary alcohol 22{1,a-d},
ready for the final stage of the synthesis. In turn, the configu-
ration at C15 of the “Series 2” portion of 21{1,a-d} was first
inverted by a Mitsunobu reaction with p-nitrobenzoic acid
(pNBA), then the acetate and the benzoate were cleaved with
NaOH and the resulting diol was protected as a bis-TBS ether
(62% overall). Selective cleavage of the primary TBS group
with HF/pyridine produced Series 2 alcohol 22{2,a-d}. Like-
wise, enantiomeric Series 3 and 4 alcohols 22{3,a-d} (no
inversion at C15) and 22{4,a-d} (inversion at C15) were
prepared by the same set of reactions starting from L-fructose-
derived 21{3,a-d}.
The completion of the mixture reaction sequence is illustrated
with Series 1 compounds in Scheme 6. Alcohol 22{1,a-d} was
converted to the corresponding sulfide by a Mitsunobu reaction
with 1-phenyl-5-thiotetrazole (PTSH). The so formed sulfide
was oxidized to the sulfone mixture 4{1,a-d} in 88% yield
by treatment with mCPBA. Deprotonation of 4{1,a-d} by
NaHMDS generated the corresponding anion, which reacted
with the aldehyde (4R,34S)-3 to afford an alkene in 82% yield.
We continued the mixture synthesis with alcohol 21{1,a-
d} emanating from the D-fructose ketone epoxidation and ring
closing reaction by first dividing it into two portions (Scheme
5). The “Series 1” portion was protected as the TBS ether (81%)
and the acetate was removed by reduction with DIBAL-H
9
J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006 9567

A R T I C L E S
Curran et al.
Scheme 5. Synthesis of Series 1-4 THF Mixtures 22
Figure 7. 19F NMR spectrum (CDCl3) of 6{a-d}
Likewise, purification of mixtures is important, since it is
impractical to take crude products ahead with no purification
until the demixing.
To remove reaction byproducts, most of the crude product
mixtures were purified by flash chromatography on silica gel.
The yields reported were calculated based on the average
molecular weight of the mixture. Unlike past work, where we
often observed a single spot for four to seven compound
mixtures,15 most of the mixtures in this synthesis showed two
closely spaced spots in a TLC on standard silica gel, and some
showed three and even four. This indicates that the fluorous-
tagged quasi-isomers can be separated on silica gel and,
therefore, that selective loss of components in the mixture during
the purification process is possible. Either HPLC on an analytical
FluoroFlash column (4.6 mm × 250 mm) or 19F NMR
spectroscopy were used to analyze chromatography fractions
both to ensure that there was no loss of components in the
mixture either during the reaction or during the purification and
to prevent the remixing of impurities that were separated during
the flash chromatographies. Any fractions containing uninten-
tionally separated target products (by LC-MS analysis) were
remixed with the main fraction prior to the next reaction step.
In characterizing mixtures either before or after purification,
we needed to know whether the underlying components had
the expected structures, whether there were impurities present
(and what they were, if possible), and whether the mixture
components were present in the expected ratios.
Scheme 6. Synthesis of Final Fluorous-Tagged Mixtures 2
LC-MS analysis with a fluorous HPLC column was on the
front line for characterization because it potentially answers all
three questions by rapidly demixing mixtures, providing com-
ponent ratios by integration, and providing structural information
(molecular weight) all at once. Some crude and all purified
product mixtures showed four major peaks with expected masses
in an HPLC chromatogram on a FluoroFlash column. Obtaining
this kind of “four-peak” chromatogram was the green light for
taking a mixture to the next step. Additional minor peaks (for
example, unreacted starting materials) were sometimes observed
in pairs with the target peaks. In these cases, we either optimized
the reaction conditions further or chromatographed the mixture
prior to scaling up. In this way, we avoided the problems of
accumulated impurities that are a crucial concern in any
multistep mixture synthesis.
The approximate ratios of each component in a mixture can
quickly be obtained without demixing by 19F NMR spectros-
copy. As an example, the CF2 region of the 19F NMR spectrum
(CDCl3) of 6{a-d} is shown in Figure 7. The resonance at
-123.4 ppm is a CF2 resonance that is unique to the C4F9-tag,
while the resonance at -117.3 ppm is a CF2 group that is
characteristic of the C2F5-tag. The integration ratio of these two
resonances thus represents the ratio of the component with the
C4F9-tag to that with the C2F5-tag. The two closely spaced
resonances at -122.4 ppm correspond to one CF2 group from
This alkene was then selectively hydrogenated with Wilkinson’s
catalyst to give the final Series 1 mixture 2{1,a-d} in 89%
yield.
Series 2-4 intermediates 22 were converted to the analogous
intermediates 2 by the same sequence of steps as Series 1.
Isolated yields were comparable across the four series, and full
details are provided in the Supporting Information. Overall, the
fluorous mixture synthesis phase (Schemes 3-6) required 39
reaction steps to produce all four final mixtures 2. Conducting
the identical sequence of reactions on individual compounds
would have required 156 reaction steps.
Fluorous Mixture Synthesis, Characterization of Inter-
mediate Mixtures: In multistep mixture synthesis, it is not
possible to move ahead blindly by assuming that reactions have
worked. Instead, characterization of mixtures and their underly-
ing components is a crucial element in identifying and optimiz-
ing reaction conditions that provide suitable yields and purities.
9
9568 J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006

Synthesis of a Murisolin Stereoisomer Library
A R T I C L E S
Figure 9. Estimated isomeric purity of the final tagged compounds 2.
Scheme 7. A Representative Deprotection
Figure 8. A typical HPLC chromatogram from the preparative demixing
of 2{1,a-d}. FluoroFlash-PF8 column, CH3CN/THF gradient (100% CH3-
CN to 90% CH3CN/10% THF over 10 min, then to 60% CH3CN/40% THF
over 20 min).
the C6F13-tag and one from the C8F17-tag. The integration of
this resonance can therefore be used to obtain the ratio of the
sum of the C6F13-tagged component and the C8F17-tagged
component to either the C2F5- or C4F9-tagged component. The
ratio of the C6F13-tagged component to the C8F17-tagged one
can be analyzed as described previously;36 however, simple
inspection of the two adjacent peak heights usually suffices.
stereocenters in 2 along with the approximate stereoselectivities.
If no diastereoisomer separation has occurred during the
synthesis, then the expected diastereopurity of each of the 16
products is about 80%. In practice, the expected levels of each
minor isomer are small enough that they are not easy to quantify
in a sample; but the potential cumulative effect is considerable.
The only point at which we carefully analyzed for minor
products during the mixture synthesis was the Shi epoxidation,
and we know that preparative flash chromatography of epoxide
20 or THF 21 did not remove the minor diastereomers from
this reaction.
The preparative demixing provided the first clear opportunity
to increase the diastereopurity by taking advantage of secondary
separations on fluorous chromatography. We used the tagging
scheme based on perfluoroethylene (CF2CF2) increments to
provide wide separation on the fluorous HPLC such that a minor
diastereomer that managed to separate from its major dia-
stereomer with the same tag would not run with the prior or
subsequent tagged isomer. In the preparative demixing, each
main peak was preceded by a very small peak (∼3%). In
addition to cutting out this unknown impurity and other obvious
minor peaks during collection in preparative runs, we were
conservative in shaving off both the early and late portions of
each main HPLC peak in the hopes of further removing minor
isomers.
Series 2-4 mixtures 2 were preparatively demixed in a
similar fashion to provide the other 12 tagged murisolin isomers.
These 16 compounds were then characterized by the standard
battery of small molecule techniques (see Supporting Informa-
tion) to ensure identity. We postponed further assessment of
diastereopurity until after the detagging.
Fluorous Mixture Synthesis, Postmix: The demixing con-
cluded the mixture synthesis phase by producing 16 individual
tagged murisolins 2, ready for detagging and final purification.
Samples were detagged in four groups of four by initial exposure
to aqueous HF in acetonitrile/THF (Scheme 7). The resulting
16 products were not characterized, though TLC analysis
showed two spots that were suggestive of the expected product
resulting from desilylation and the product resulting from further
1
We recorded H NMR spectra of every mixture, and these
“mixture spectra” sometimes provided useful information. For
example, it is very easy to follow the appearance or disappear-
ance of functional groups such as aldehydes with characteristic
resonances. But the mixture spectra do not yield the usual level
of information since they are the sum of four overlapping spectra
of the components. An LC NMR experiment, however, provides
the spectrum of individual components and is thus a powerful
tool that we routinely used in this work. The use of LC NMR
to validate quasienantiomer analysis is described above in Figure
6. More typically, we used this experiment to simply prove that
each underlying precursor in a mixture had been converted to
the target product. This use is illustrated in more detail in the
Supporting Information.
In short, when used judiciously, existing spectroscopic and
chromatographic techniques are capable of filling all separation
and characterization requirements for multistep fluorous mixture
synthesis.
Fluorous Mixture Synthesis, Demixing: Each of the four
final mixtures 2 of fluorous-tagged compounds was then
preparatively demixed to provide its four underlying components
in pure form. The demixing of Series 1 mixture 2{1,a-d} is
representative. The mixture (302 mg) was preparatively sepa-
rated over a FluoroFlash PF-8 column (20 mm × 250 mm)
under gradient conditions. A typical chromatogram for a
preparative run is shown in Figure 8. About 40 mg of the
mixture was injected each time. After eight injections, the
amounts of pure 2{1,a}, 2{1,b}, 2{1,c}, and 2{1,d} obtained
were 31.0, 49.9, 54.5, and 60.2 mg, respectively. The molar
ratio of the pure components is about 1:1.5:1.5:1.5, and the
overall recovery of the preparative separation was 65%.
The moderate (65%) recovery from the preparative demixing
is not unexpected because the underlying tagged components
at this point are not likely to be diastereomerically pure. Figure
9 summarizes the reactions that were used to introduce the
(36) Zhang, Q. S.; Rivkin, A.; Curran, D. P. J. Am. Chem. Soc. 2002, 124,
5774-5781.
9
J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006 9569

A R T I C L E S
Curran et al.
cleavage of the PMBF group. To complete the PMBF removal,
the crude products were taken up in wet dichloromethane, and
DDQ was added. After 1 h, the reaction mixture was worked
up, and the crude products were purified by flash chromatog-
raphy to give the 16 target murisolins as white waxes in yields
ranging from 62 to 71%.
At this stage, all 16 isomers of the murisolins 1 were carefully
characterized by a set of 1D (1H and 13C) and 2D (1H/1H COSY,
HMQC, HMBC) experiments. While the 1H NMR spectra were
clean, each 13C NMR spectrum exhibited two minor peaks very
close to the major peaks at about 152 and 70 ppm. Integrations
of appropriate resonances in each 13C NMR spectrum showed
that the ratio of major to minor products was about 15/1. This
minor product cannot be another diastereomer in the library (that
is, isomers at C15, C16, C19, or C20), since none of the 16
members of the stereoisomer library has these resonances.
In carefully checking intermediate purity, we located the
impurity in the starting TBS-protected hydroxybutenolide
(4R,34S)-12. This exhibits major resonances at 151.7 and 70.1
ppm, while there is a small impurity with resonances at 151.9
and 69.8 ppm. The chemical shifts and ratios (∼15/1) of these
resonances correspond well to the minor resonances in the final
product. Accordingly, the impurity is the epimer at C4 that
resulted from the Sharpless asymmetric dihydroxylation.
Figure 10. Strategy for the double mixture synthesis with oligoethylene
glycol tags and fluorous tags.
the single largest impurity (C4 epimer) was removed by
Chiralcel chromatography. Each of the 16 murisolins 1 has a
different retention time on the Chiralcel OD column, so
hindsight suggests that the reversed-phase purification was not
needed and that a single Chiralcel purification would have given
similar results.
We conclude from the HPLC studies that each of the 16
samples in the first-generation murisolin library is substantially
(g98%) isomerically pure. This conclusion is important not only
because each isomer was accompanied by three other quasi-
isomers during the fluorous mixture synthesis but also because
a number of the isomers have identical NMR spectra, even at
very high magnetic field.
We conducted LC-MS analyses of each sample on a standard
reversed-phase column (Xterra C18), and apparent purities
ranged from 90 to 97%. Minor peaks were present in each
chromatogram, and all peaks exhibited the same molecular ion
(m/e ) 581), suggesting that all the compounds detected in this
analysis are diastereomers. However, the analytical LC-MS
traces did not reflect our expectations from the 13C NMR spectra.
Instead of exhibiting one major impurity in a ratio to the major
product of 15/1, each chromatogram exhibited two or more trace
products whose amounts summed to 3-9%.
To remove these trace products, each sample was purified
by preparative reversed-phase HPLC on a Symmetry C18
column. However, 13C NMR spectra of all these samples still
showed the two extraneous resonances at 151.9 and 69.8 ppm,
so the samples were not yet pure. Analyses of these samples
on a Chiralcel OD column showed purities of 90-96%. This
time, each major peak in the chromatogram was accompanied
by a single minor peak. Thus, the reversed-phase purification
succeeded in removing small amounts of diastereomers that we
Double Mixture Synthesis: One of the limitations with the
fluorous mixture synthesis is that 16 fluorous tags are not
currently available, so the 16 isomers were made in four
mixtures of four compounds. Although this results in significant
savings in effort, we still sought a means to make 16 isomers
in a single reaction vessel. Toward this end, we have recently
introduced “double fluorous/oligoethylene glycol mixture syn-
thesis”, which is in essence the product of the two single mixture
synthesis methods.18 In other words, reaction of one mixture of
four fluorous-tagged isomers with another mixture of four
oligoethylene glycol-tagged isomers will give a new 16-
compound mixture in which each of the 16 isomers is separable
and identifiable by its unique combination of the two tags.
The plan for the double mixture synthesis is shown in Figure
10. The tagging strategy calls for the coupling of a four-
compound fluorous-tagged dihydroxy-THF fragment 25{a-d}20
with a four-compound OEG-tagged hydroxybutenolide fragment
24{1-4}. After one hydrogenation reaction to saturate the
alkene, the resulting 16-component mixture of doubly tagged
murisolins 23{1-4,a-d} is ready for double demixing and
detagging. Oligoethylene glycol homologues of the popular
dimethoxybenzyl group were selected for protection of the C4
hydroxy group of 24 because the needed reagents are readily
available and because we expected that the DMBOEG group
could be removed under the same conditions as those for the
PMBF group. Minimizing protecting group removal steps is
1
could not detect by H or 13C NMR analysis, but it did not
remove the major impurity (C4 epimer). This was finally
removed when each sample was purified by semipreparative
HPLC over a Chiralcel OD column to give the 16 pure murisolin
stereoisomers. These products did not exhibit the minor peaks
adjacent to the major ones at 151.7 and 70.1 ppm in their 13C
NMR spectra, and there were no significant impurities in either
the reversed-phase or Chiralcel OD phase analytical HPLC
chromatograms.
Overall, the HPLC results are qualitatively consistent with
the stereochemical analysis in Figure 9. Even though intermedi-
ate mixtures were purified by flash chromatography at several
stages during the fluorous mixture synthesis, these purifications
did not significantly enrich the diastereopurity of the intermedi-
ates. Crude samples after detagging were probably 80-90%
isomerically pure. Some minor diastereomers were removed in
the initial chromatography over reversed-phase silica gel, and
9
9570 J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006

Synthesis of a Murisolin Stereoisomer Library
A R T I C L E S
Scheme 9. Double Mixture Synthesis and Demixing
Scheme 8. Synthesis of Fluorous-Tagged Mixture 25{a-d}
important because these occur during the postmix stage and
therefore must be conducted on 16 samples.
For the fluorous mixture 25 we choose to work in Series 4
(15R,16R), which by then we knew contained the natural product
murisolin 1.14. The OEG mixture 24 contains all four possible
isomers of the hydroxybutenolide, so the final stereoisomer
sublibrary will contain murisolin and three of the isomers that
are already in hand, along with 12 new isomers. The new set
of 12 isomers includes all possible stereoisomers of the natural
product murisolin at C4 and C34. Taken together, the two
libraries provide a total of 28 of the 64 possible isomers of
murisolin, with 24 of the 32 possible diastereomers.
The double mixture reaction sequence follows the established
route, but we made one important improvement in each
fragment. In the dihydroxy-THF fragment 25, we replaced the
TBS group on O15 with a standard PMB group, again to
minimize protecting group removal steps. In the synthesis of
the hydroxybutenolide fragment 24, we exchanged the Sharpless
AD reaction with a Jacobsen hydrolytic kinetic resolution37 of
an epoxide. This provided 24 in isomerically pure form and
accordingly eliminated the set of minor isomers at C4 that
needed to be removed during the postmix stage of the fluorous
mixture synthesis. The synthesis of aldehyde 24{1-4} was the
first example of an OEG mixture synthesis, and this work is
described elsewhere in full detail.38
The synthesis of the needed fluorous mixture 25{a-d} is
summarized in Scheme 8. Intermediate 21{3,a-d} (Scheme 4)
was inverted by Mitsunobu reaction (inverts Series 3 to Series
4) and the diol resulting after hydrolysis of both esters was
monosilylated with TBSCl to give 26{a-d}. The resulting
alcohol was reacted with PMBBr by a standard procedure, and
the TBS ether then was converted to the N-phenyltetrazolyl
sulfone 25{a-d} by the same sequence of steps used before.
The double mixture steps are summarized in Scheme 9.
Kocienski-Julia coupling of four-component fluorous-tagged
mixture 25{a-d} (0.28 mmol) with four-component OEG
mixture 24{1-4} (0.28 mmol) as usual provided a crude alkene
product 27{1-4,a-d}.
Fluorous HPLC analysis of this mixture occurred with
fluorous demixing but not OEG demixing to provide four pairs
of peaks, each pair with a ratio of about 85/15. Both fluorous
HPLC analysis and standard TLC analysis showed that the
minor peaks belonged to the starting sulfones in 25{a-d}.
Accordingly, the crude mixture 27 was purified by silica flash
chromatography by using a step gradient to sequentially elute
groups of molecules based on the OEG tags.38 First step elution
with 25% hexane/ethyl acetate provided unreacted sulfone
25{a-d} followed by a fraction of the four coupled products
bearing the n ) 1 OEG tag and all four possible fluorous tags
27{1,a-d}. The succeeding steps of the gradient provided the
other three four-compound mixtures 27 with OEG ) 2, 3, and
4. These four mixtures were then remixed to give a pure 16-
compound mixture 27{1-4,a-d}, which was analyzed by
fluorous HPLC. As expected, the set of four major peaks
resulting from the Julia products 27 remained, but the set of
four minor peaks resulting from the starting sulfone 25 were
absent. Figure 11 shows the HPLC chromatograms of 27{1-
4,a-d} before and after purification. The total yield of the
purified 16-compound mixture 27 obtained by a single flash
chromatography was 71%.
Thus for unknown reasons, the large scale, double mixture
coupling reaction did not result in complete consumption of
the sulfone 25{a-d}. Nonetheless, it was still possible to readily
remove the four unreacted sulfones from the 16-products 27
prior to the hydrogenation reaction. This reaction proceeded
smoothly on a 320 mg sample of 27{1-4,a-d} to provide
23{1-4,a-d} in crude quantitative yield, ready for double
demixing.
(37) (a) Schaus, S. E.; Brandes, B. D.; Larrow, J. F.; Tokunaga, M.; Hansen,
K. B.; Gould, A. E.; Furrow, M. E.; Jacobsen, E. N. J. Am. Chem. Soc.
2002, 124, 1307-1315. (b) Liu, P.; Jacobsen, E. N. J. Am. Chem. Soc.
2001, 123, 10772-10773.
(38) Gudipati, V.; Curran, D. P.; Wilcox, C. S. J. Org. Chem. 2006, ASAP
article.
Double demixing of a 16-compound mixture requires five
separations. A first separation against one class of tags provides
9
J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006 9571

A R T I C L E S
Curran et al.
This is in line with expectations, since the starting aldehyde 24
was made in diastereopure form by a different route.
Since the prior work suggested that the reversed-phase
chromatography was not needed, the products 1.13-1.28 were
purified by semipreparative HPLC over a Chiralcel OD column
to remove minor diastereomeric impurities. The purities of these
products under both reversed-phase and chiral phase conditions
were assessed. All compounds exhibited single peaks on the
chiral column, showing the success of the preparative separation.
13 of the samples also showed excellent purity in the reversed-
phase analysis (g95%). However, unexpectedly three of the
samples (see Supporting Information) showed significant im-
purities (5-15%), and accordingly these three were repurifired
by reversed-phase chromatography to provide all 16 pure
products.
Four of the products (1.13-1.16) were compared with the
corresponding products from the prior library, and the samples
were identical by spectroscopy and (more importantly) by co-
injection on the Chiralcel OD column. The other 12 compounds
are new and were fully characterized as usual.
Compound Characterization: Table 1 compiles information
about all 28 members of the stereoisomer library, including their
configurations, their tags in the single and/or double mixture
library, their retention times on a Chiralcel OD column, and
their optical rotations in MeOH (c ) 0.05-0.27). The retention
times on the chiral column vary over a wide range, and are
accordingly very useful for matching or differentiating isomers.
In contrast, the optical rotations are of little value because they
are very small and may be concentration dependent. All 28
compounds were fully characterized by the usual spectroscopic
techniques, and data are compiled in text format in the
Supporting Information.
Comparisons of spectra and other data to make murisolin
structure assignments are the topics of a forthcoming full
paper.19 Briefly, there is substantial spectral identity among the
murisolin isomers. Indeed, all 32 murisolin isomers exhibit one
of only six sets of 1H NMR spectra at 600 MHz; no isomer has
a unique spectrum. Nonetheless, we have rigorously confirmed
that the structure of murisolin 1.14 is correct but suggest that
the structure of 16,19-cis-murisolin 1.16 is probably incorrect
and that it may be 1.8 instead. We also suggest that murisolin
A may be 1.10.
Figure 11. Fluorous HPLC traces of crude (a) and purified (b) 31{1,a-
d} show that the unreacted sulfones 25{a-d} (minor peak set in (a) to the
left of major peak set) have been removed.
four mixtures of four compounds. In turn, these four mixtures
are each separated against the second class of tags to provide
the 16 pure products. While double demixings can be conducted
in either order,18 we decided here to conduct the OEG demixing
first because it is a simple flash chromatography that is easier
to accomplish than preparative HPLC on the larger initial
sample.
Mixture 23{1-4,a-d} was demixed over silica gel by a four-
step gradient (EtOAc in hexanes, 25%, 50%, 65%, 80%) to
provide four fractions of products bearing increasingly larger
OEG tags. Each fraction contained four compounds with the
four possible fluorous tags: 23{1,a-d, 52 mg; 23{2,a-d}, 66
mg; 23{3,a-d}, 55 mg; 23{4,a-d}, 50 mg 84% overall from
alkene 31). Each of the resulting four fractions was demixed in
the fluorous dimension on a FluoroFlash HPLC column by the
conditions described above. Sample sizes were about 20 mg
per injection, so each fraction required three injections for
complete demixing. Minor peaks were not collected, and the
leading and trailing edges of the major peaks were shaved off
to provide the maximum possible purity. This provided all 16
individual products 2.13-2.28 in a total recovery of 72% from
the purified mixture 23.
Conclusions
Despite their inherent value, stereoisomer libraries of complex
natural products are rarely made because too much effort is
involved in traditional “one compound at a time” approaches
to total synthesis. Fluorous mixture synthesis, oligoethylene
glycol mixture synthesis, and double mixture synthesis use
separation tags to leverage the power of traditional solution
phase methods by providing more compounds per unit of
synthetic effort. Standard small molecule separation and char-
acterization techniques can be used even though they are applied
to mixtures rather than single compounds. The separation tags
(fluorous or oligoethylene glycol) provide for orchestrated
demixing of the mixtures at any time. The syntheses of murisolin
isomers show that combining solution phase mixture synthesis
techniques with synthetic plans that minimize steps both with
efficient routes and with late splits allows for a substantial
increase in the number of stereoisomers of complex compounds
that can be made.
In the postmix stage, the normal PMB group, the fluorous
PMB group, and the OEG DMB group were simultaneously
removed by treatment of the compounds with 3.3 equiv of DDQ.
Samples were initially purified by preparative TLC to remove
1
DDQ remnants. Careful analysis of one of the products by H
NMR (500 MHz) and 13C NMR (125 MHz) showed traces of
impurities; however, resonances at 151.1 and 69.8 from the C4
epimer that were present in the first library were not observed.
9
9572 J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006

Synthesis of a Murisolin Stereoisomer Library
A R T I C L E S
Table 1. Configurations, Tags, Optical Rotations, and Retention Times of the 28 Murisolin Isomers
e
f
number
abs configa
rel configb
FMS precc
FTag
DMS precd
OEG tag
F tag
RD
TR
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
RRSSRS
RRSSSS
RRSRSS
RRSRRS
RSSSRS
RSSSSS
RSSRSS
RSSRRS
RSRRSS
RSRRRS
RSRSRS
RSRSSS
RRRRSS
RRRRRS
RRRSRS
RRRSSS
RRRRSR
RRRRRR
RRRSRR
RRRSSR
SRRRSR
SRRRRR
SRRSRR
SRRSSR
SRRRSS
SRRRRS
SRRSRS
SRRSSS
aete
aett
aece
aect
atte
attt
atce
atct
aete
aett
aect
aece
atte
attt
atce
atct
stte
sttt
stce
stct
atte
attt
2{1,a}
2{1,b}
2{1,c}
2{1,d}
2{2,a}
2{2,b}
2{2,c}
2{2,d}
2{3,c}
2{3,d}
2{3,a}
2{3,b}
2{4,c}
2{4,d}
2{4,a}
2{4,b}
-
C2F5
C4F9
C6F13
C8F17
C2F5
C4F9
C6F13
C8F17
C6F13
C8F17
C2F5
C4F9
C6F13
C8F17
C2F5
C4F9
-
-
-
-
-
+1.1
27.8
20.3
20.6
18.0
22.6
20.7
14.0
15.9
25.3
22.1
21.6
14.3
-
-
+0.7
+9.9
+8.2
+1.9
+2.1
+7.5
+4.5
+13.5
+9.5
+3.1
+6.9
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
1.9
-
-
-
1.10
1.11
1.12
1.13
1.14
1.15
1.16
1.17
1.18
1.19
1.20
1.21h
1.22h
1.23h
1.24h
1.25
1.26
1.27
1.28
-
-
-
-
-
-
-
-
-
23{1,c}
23{1,d}
23{1,a}
23{1,b}
23{2,c}
23{2,d}
23{2,a}
23{2,b}
23{4,c}
23{4,d}
23{4,a}
23{4,b}
23{3,c}
23{3,d}
23{3,a}
23{3,b}
n ) 1
n ) 1
n ) 1
n ) 1
n ) 2
n ) 2
n ) 2
n ) 2
n ) 4
n ) 4
n ) 4
n ) 4
n ) 3
n ) 3
n ) 3
n ) 3
C6F13
C8F17
C2F5
C4F9
C6F13
C8F17
C2F5
C4F9
C6F13
C8F17
C2F5
C4F9
C6F13
C8F17
C2F5
C4F9
+14.0/+10.3
+16.1/+6.6
+3.0/+4.1
+8.1/+9.8
-2.5
21.6/22.0g
23.2/24.6g
15.9/16.0g
16.4/17.2g
23.8
26.3
17.5
17.9
21.9
25.1
15.6
16.3
22.1
-
-
+5.2
-
-
-5.0
-
-
+4.1
-
-
+4.3
-
-
-1.2
atce
atct
stte
sttt
stce
stct
-
-
+1.5
-
-
-0.3
-
-
+3.4
-
-
-1.9
24.5
15.8
16.6
-
-
-11.9
-1.4
-
-
b
a C4,15,16,19,20,34.
c s/a, syn/anti at C4,34, t/e, threo/erythro at C15,16 and C19,20; c/t, cis/trans at C16,19; {series number, fluorous tag}. d {OEG
tag, fluorous tag}. e In MeOH, c ) 0.05-0.27; see Supporting Information for concentrations. f Retention time in min on a Chiralcel OD column, 94/6,
hexane/2-propanol, 1.5 mL/min; the absolute retention. g The retention times correspond to samples run from the fluorous mixture library (first value) and
the double mixture library (second value) injected several months apart. The samples were identical on co-injection. h 1.21-1.24 are enantiomers of 1.5-
1.8.
The fluorous mixture synthesis and double mixture synthesis
described herein provided 32 murisolins, 28 of which are unique
(four compounds are the same in the two libraries). Among these
28, there are four pairs of enantiomers, and therefore we have
made 24 of the 32 possible diastereomers of murisolin. The
mixture synthesis plans provide for substantial savings in effort
over the same plans conducted in parallel or serial on individual
compounds; the fluorous mixture synthesis saves well over 100
chemical steps, while the double mixture synthesis saves about
60 steps.
The effect of leveraging of effort can also be seen by posing
the question: is it realistically possible to make all 64 isomers
of murisolin by using these techniques? The answer is certainly
“yes”. Indeed, if we had available to us the four-compound
mixture of hydroxybutenolides 25{a-d} at the end of the first
mixture synthesis, then we could have used this instead of the
single hydroxybutenolide 3 in the four final couplings to produce
four mixtures of 16 compounds instead of four mixtures of four
compounds. Double demixing and detagging of these four
mixtures would have provided all 64 isomers of murisolin. This
“paper synthesis” of the complete murisolin stereoisomer library
has the same number of premix and mixture synthesis steps as
the actual synthesis that we conducted in making the second
16-compound library; no new intermediates have to be made!
It requires only 15 extra demixings (4 × 5 demixings instead
of 1 × 5) and 48 extra detaggings to provide 64 products instead
of 16. The only trade off is sample size, since on the same scale
each of the 64 compounds would be produced in one-fourth of
the amount as 16. That we could potentially make 48 more
compounds with only a few more HPLC injections and
detaggings is a dramatic example of how tagging and splitting
can combine to provide a large stereoisomer library.
Although we have the capability, we do not currently plan
to make all 64 isomers of murisolin because we have already
obtained a complete picture of the spectroscopic behavior of
this family of compounds from the 28 isomers in hand, along
with derivatives. This is the first such complete picture to be
obtained for any acetogenin. A forthcoming paper19 analyzes
these data fully and discusses ramifications for stereostructure
assignments of murisolins, other acetogenins, and related
compounds with local symmetry and remote groups of stereo-
centers.
Acknowledgment. We thank the National Institutes of Health
(National Institute for General Medical Sciences) for funding
of this work. We also thank B. Figade`re and T. Tanaka for
providing samples, spectra, and helpful comments (BF). We
also thank Mr. Serhan Turklyilmaz for help preparing OEG
tagging reagents and Ms. F.-M. Lin and Dr. F. -T. Lin for
running LC NMR experiments. We dedicate this paper to Prof.
Barry M. Trost on the occasion of his 65th birthday.
Supporting Information Available: Contains full experi-
mental details and key characterization data of new compounds
(65 pages). This material is available free of charge via the
Internet at http://pubs.acs.org.
JA061801Q
9
J. AM. CHEM. SOC. VOL. 128, NO. 29, 2006 9573