Total synthesis and evaluation of cytostatin, its C10-C11 diastereomers, and additional key analogues: Impact on PP2A inhibition

Article abstract of DOI:10.1021/ja066477d

The total synthesis of cytostatin, an antitumor agent belonging to the fostriecin family of natural products, is described in full detail. The convergent approach relied on a key epoxide-opening reaction to join the two stereotriad units and a single-step

Full text of DOI:10.1021/ja066477d

Published on Web 12/07/2006
Total Synthesis and Evaluation of Cytostatin, Its C10-C11
Diastereomers, and Additional Key Analogues: Impact on
PP2A Inhibition
Brian G. Lawhorn,† Sobhana B. Boga,† Scott E. Wolkenberg,† David A. Colby,†
Carla-Maria Gauss,† Mark R. Swingle,‡ Lauren Amable,‡
Richard E. Honkanen,‡ and Dale L. Boger*,†
Contribution from the Department of Chemistry and The Skaggs Institute for Chemical Biology,
The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, and
Department of Biochemistry and Molecular Biology, UniVersity of South Alabama College of
Medicine, Mobile, Alabama 36688
Received September 13, 2006; E-mail: boger@scripps.edu
Abstract: The total synthesis of cytostatin, an antitumor agent belonging to the fostriecin family of natural
products, is described in full detail. The convergent approach relied on a key epoxide-opening reaction to
join the two stereotriad units and a single-step late-stage stereoselective installation of the sensitive (Z,Z,E)-
triene through a â-chelation-controlled nucleophilic addition. The synthetic route provided rapid access to
the C4-C6 stereoisomers of the cytostatin lactone, which were prepared and used to define the C4-C6
relative stereochemistry of the natural product. In addition to the natural product, each of the C10-C11
diastereomers of cytostatin was divergently prepared (11 steps from key convergence step) by this route
and used to unequivocally confirm the relative and absolute stereochemistry of cytostatin. Each of the
cytostatin diastereomers exhibited a reduced activity toward inhibition of PP2A (>100-fold), demonstrating
the importance of the presence and stereochemistry of the C10-methyl and C11-hydroxy groups for potent
PP2A inhibition. Extensions of the studies provided dephosphocytostatin, sulfocytostatin (a key analogue
related to the natural product sultriecin), 11-deshydroxycytostatin, and an analogue lacking the entire C12-
C18 (Z,Z,E)-triene segment, which were used to define the magnitude of the C9-phosphate (>4000-fold),
C11-alcohol (250-fold), and triene (220-fold) contribution to PP2A inhibition. A model of cytostatin bound
to the active site of PP2A is presented, compared to that of fostriecin, which is also presented in detail for
the first time, and used to provide insights into the role of the key substituents. Notably, the R,â-unsaturated
lactone of cytostatin, like that of fostriecin, is projected to serve as a key electrophile, providing a covalent
adduct with Cys269 unique to PP2A, contributing to its potency (g200-fold for fostriecin) and accounting
for its selectivity.
nM, PP2A/PP1 > 1000),5 and it shares the distinctive phosphate
monoester, Z,Z,E-triene, and R,â-unsaturated δ-lactone structural
Introduction
Cytostatin (1),1 an antitumor agent isolated from Streptomyces
sp. MJ654-Nf4 belonging to the fostriecin family of natural
products,2 displays potent cytotoxic activity toward leukemia
and melanoma cancer cell lines (IC50 ) 100 nM), induces
apoptosis,3 and inhibits lung tumor metastasis.4 Like the parent
of its class, fostriecin (2), cytostatin is a potent and selective
inhibitor of protein phosphatases 2A and 4 (PP2A, IC50 ) 210
units. Additional related natural products include sultriecin (3),
phospholine (4), the leustroducsins (5), and the phoslactomycins
(6) (Figure 1).6 The natural products 4-6 display different bio-
logical profiles than 1 and 2, as they are only weakly cytotoxic,
but they exhibit significant broad-spectrum antifungal activity.
On the basis of reported PP2A inhibition values, fostriecin (2,
IC50 ) 3.2 nM)7 is apparently considerably more potent than
† The Scripps Research Institute.
‡ University of South Alabama College of Medicine.
(1) (a) Amemiya, M.; Someno, T.; Sawa, R.; Naganawa, H.; Ishizuka, M.;
Takeuchi, T. J. Antibiot. 1994, 47, 541. (b) Amemiya, M.; Ueno, M.; Osono,
M.; Masuda, T.; Kinoshita, N.; Nishida, C.; Hamada, M.; Ishizuka, M.;
Takeuchi, T. J. Antibiot. 1994, 47, 536.
(5) Kawada, M.; Amemiya, M.; Ishizuka, M.; Takeuchi, T. Biochim. Biophys.
Acta 1999, 1452, 209.
(6) (a) Sultriecin: Ohkuma, H.; Naruse, N.; Nishiyama, Y.; Tsuno, T.; Hoshino,
Y.; Sawada, Y.; Konishi, M.; Oki, T. J. Antibiot. 1992, 45, 1239. (b)
Phospholine: Ozasa, T.; Tanaka, K.; Sasamata, M.; Kaniwa, H.; Shimizu,
M.; Matsumoto, H.; Iwanami, M. J. Antibiot. 1989, 42, 1339. (c)
Leustroducsins: Kohama, T.; Enokita, R.; Okazaki, T.; Miyaoka, H.;
Torikata, A.; Inukai, M.; Kaneko, I.; Kagasaki, T.; Sakaida, Y.; Satoh, A.;
Shiraishi, A. J. Antibiot. 1993, 46, 1503. Kohama, T.; Nakamura, T.;
Kinoshita, T.; Kaneko, I.; Shiraishi, A. J. Antibiot. 1993, 46, 1512. (d)
Phoslactomycins: Fushimi, S.; Nishikawa, S.; Shimazu, A.; Seto, H. J.
Antibiot. 1989, 42, 1019. (e) Phosphazomycins: Tomiya, T.; Uramoto, M.;
Isono, K. J. Antibiot. 1990, 43, 118.
(2) Lewy, D. S.; Gauss, C. M.; Soenen, D. R.; Boger, D. L. Curr. Med. Chem.
2002, 9, 2005.
(3) (a) Yamazaki, K.; Amemiya, M.; Ishizuka, M.; Takeuchi, T. J. Antibiot.
1995, 48, 1138. (b) Kawada, M.; Amemiya, M.; Ishizuka, M.; Takeuchi,
T. Jpn. J. Cancer Res. 1999, 90, 219.
(4) (a) Masuda, T.; Watanabe, S.; Amemiya, M.; Ishizuka, M.; Takeuchi, T.
J. Antibiot. 1995, 48, 528. (b) Kawada, M.; Kawatsu, M.; Masuda, T.; Ohba,
S.; Amemiya, M.; Kohama, T.; Ishizuka, M.; Takeuchi, T. Int. Immunop-
harm. 2003, 3, 179.
(7) Walsh, A. H.; Cheng, A.; Honkanen, R. E. FEBS Lett. 1997, 416, 230.
9
16720
J. AM. CHEM. SOC. 2006, 128, 16720-16732
10.1021/ja066477d CCC: $33.50 © 2006 American Chemical Society

Total Synthesis of Cytostatin and Key Analogues
A R T I C L E S
Figure 2. Key elements of fostriecin’s PP2A activity.16
have been conducted in a linear fashion and have relied on
cross-coupling strategies to install the triene in a multistep,
segmental manner that contrasts the approach to cytostatin
described herein.
In efforts aimed at identifying the features of fostriecin (2)
required for its potent and selective PP2A inhibition, we
prepared analogues of 2 that were used to demonstrate the
importance of the phosphate monoester and that defined the
role of the unsaturated lactone (Figure 2).16 In these studies,
we provided evidence that the unsaturated lactone serves as a
critical electrophile that reacts with C269 of PP2A that is not
present in PP1, accounting for the PP2A potency and selectivity
of 2,16 and this has now been confirmed in studies of PP2A
inhibition by phoslactomycin (6).17 Docking fostriecin into a
PP2A homology model2,18 revealed additional potential active-
site interactions, including a hydrogen bond between the C11-
hydroxy and Arg214 and the penetration of the triene unit into
a hydrophobic cleft, both of which are conserved binding
features of the nonselective PP1 inhibition pharmacophore.19
The importance of the extended hydrophobic segment for PP2A
binding has been confirmed with a cytostatin analogue lacking
part of the triene unit that displayed reduced potency.14 In
addition, the putative role of the C11-hydroxy is supported by
the lack of activity of both a fostriecin analogue and a cytostatin
analogue acetylated at C11.14,16 However, both of these acety-
lated analogues as well as that which probed the triene
contribution were otherwise altered in ways that might signifi-
cantly attenuate their PP2A inhibition, leaving the true magni-
tude of their contributions in question.
An additional conserved feature of fostriecin, cytostatin, and
other PP inhibitors is a methyl group proximal to the acidic
metal binding moiety, which has been proposed to mimic the
methyl group of phosphothreonine.18 This is a particularly
compelling hypothesis, given that PP2A displays a substrate
preference for phosphothreonine peptides over phosphoserine
peptides. Although it is unclear whether the conserved methyl
group is beneficial for inhibition, it has been proposed that it
may contribute to PP binding either through hydrophobic contact
with a protein residue or by restricting the conformation of the
natural product to one favorable for PP binding.18 The stereo-
chemical integrity of fostriecin’s C8 center is crucial for PP2A
activity,9 but whether this is due to effects related to the C8-
methyl or C8-hydroxy or both has not yet been determined.
Figure 1. Cytostatin and related natural products.
cytostatin (1, IC50 ) 210 nM), phospholine (4, IC50 ) 5.8 µM),8
leustroducsin H (5, IC50 ) 130 nM),4b and the phoslactomycins
(6, IC50 ) 3.7-4.9 µM).8 However, no side-by-side comparisons
have been conducted between fostriecin and the other members
of its class, limiting interpretations of the potentially useful
structure-function information embedded in the natural prod-
ucts. Cross-assay comparisons are particularly interesting, since
IC50 values for fostriecin inhibition of PP2A are dependent not
only on the assay enzyme concentration but also on the
phosphorylated substrate used in the assay.7,9
In preceding efforts related to the fostriecin family of natural
products, we determined the stereochemical configuration of
fostriecin (2) and reported its first total synthesis.10 Subsequent
to this work, a number of additional synthetic efforts have been
described, including eight total or formal syntheses of 2, each
highlighting the utility of alternative asymmetric synthetic
methods for the construction of the four chiral centers of 2 in
a stereoselective manner.11 A total synthesis of leustroducsin B
(phospholine)12 and a recent total synthesis of phoslactomycin
B13 have also been reported. In 2002, Waldmann reported the
total synthesis of 1 and several analogues,14 and Marshall has
since disclosed an alternative route to an advanced intermediate
of the Waldmann synthesis.15 The reported syntheses of 1 and
2, including our own initial fostriecin total synthesis, typically
(8) Usui, T.; Marriott, G.; Inagaki, M.; Swarup, G.; Osada, H. J. Biochem.
(Tokyo) 1999, 125, 960.
(9) Maki, K.; Motoki, R.; Fujii, K.; Kanai, M.; Kobayashi, T.; Tamura, S.;
Shibasaki, M. J. Am. Chem. Soc. 2005, 127, 17111.
(10) (a) Boger, D. L.; Hikota, M.; Lewis, B. M. J. Org. Chem. 1997, 62, 1748.
(b) Boger, D. L.; Ichikawa, S.; Zhong, W. J. Am. Chem. Soc. 2001, 123,
4161.
(11) For reviews of fostriecin total syntheses see: (a) Shibasaki, M.; Kanai, M.
Heterocycles 2005, 66, 727 and ref 2. (b) Chavez, D. E.; Jacobsen, E. N.
Angew. Chem., Int. Ed. 2001, 40, 3667. (c) Reddy, Y. K.; Falck, J. R.
Org. Lett. 2002, 4, 969. (d) Miyashita, K.; Ikejiri, M.; Kawasaki, H.;
Maemura, S.; Imanishi, T. Chem. Commun. 2002, 742. Miyashita, K.;
Ikejiri, M.; Kawasaki, H.; Maemura, S.; Imanishi, T. J. Am. Chem. Soc.
2003, 125, 8238. (e) Esumi, T.; Okamoto, N.; Hatakeyama, S. Chem.
Commun. 2002, 3042. (f) Fujii, K.; Maki, K.; Kanai, M.; Shibasaki, M.
Org. Lett. 2003, 5, 733 and ref 9. (g) Trost, B. M.; Frederiksen, M. U.;
Papillon, J. P.; Harrington, P. E.; Shin, S.; Shireman, B. T. J. Am. Chem.
Soc. 2005, 127, 3666. (h) Yadav, J. S.; Prathap, I.; Tadi, B. P. Tetrahedron
Lett. 2006, 47, 3773.
(12) Shimada, K.; Kaburugi, Y.; Fukuyama, T. J. Am. Chem. Soc. 2003, 125,
4048.
(13) Wang, Y.; Takeyama, R.; Kobayashi, Y. Angew. Chem., Int. Ed. 2006, 45,
3320.
(16) Buck, S. B.; Hardouin, C.; Ichikawa, S.; Soenen, D. R.; Gauss, C. M.;
Hwang, I.; Swingle, M. R.; Bonness, K. M.; Honkanen, R. E.; Boger, D.
L. J. Am. Chem. Soc. 2003, 125, 15694.
(17) Teruya, T.; Simizu, S.; Kanoh, N.; Osada, H. FEBS Lett. 2005, 579, 2463.
(18) Gauss, C. M.; Sheppeck, J. E., II; Nairn, A. C.; Chamberlin, R. Bioorg.
Med. Chem. 1997, 5, 1751.
(19) Reviews: (a) Colby, D. A.; Chamberlin, A. R. Mini-ReV. Med. Chem. 2006,
6, 109. (b) Sheppeck, J. E., II; Gauss, C. M.; Chamberlin, A. R. Bioorg.
Med. Chem. 1997, 5, 1739.
(14) (a) Bialy, L.; Waldmann, H. Angew. Chem., Int. Ed. 2002, 41, 1748. (b)
Bialy, L.; Waldmann, H. Chem. Commun. 2003, 1872. (c) Bialy, L.;
Waldmann, H. Chem. Eur. J. 2004, 10, 2759. (d) Bialy, L.; Lopez-Canet,
M.; Waldmann, H. Synthesis 2002, 2096.
(15) Marshall, J. A.; Ellis, K. Tetrahedron Lett. 2004, 45, 1351.
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16721

A R T I C L E S
Lawhorn et al.
Scheme 1. Synthesis of 12-15
Figure 3. Stereochemical assignment for cytostatin.
Herein, we provide full details of a total synthesis of cytostatin
that is complementary to previous syntheses of 1 and 2 and by
design provides rapid access to all of the stereoisomers of
cytostatin.20 Accordingly, its extension to the preparation of each
of the three additional C10-C11 stereoisomers of cytostatin is
also described, and these compounds, in conjunction with four
diastereomeric lactone partial structures, were used to unequivo-
cally confirm the relative and absolute stereochemistry of
cytostatin. The comparative biological assessment of cytostatin
and its diastereomers is reported and demonstrates the impor-
tance of the conserved C10-methyl group and the C11-hydroxy
for potent PP2A inhibition. Further extensions of the approach
provided dephosphocytostatin (40), 11-deshydroxycytostatin
(78), and 72, lacking the entire C12-C18 triene subunit, and
their comparative biological evaluation provided direct assess-
ments of the contribution that the cytostatin C9-phosphate, C11-
alcohol, and C12-C18 terminal triene make toward protein
phosphatase inhibition. Removal of the C9-phosphate resulted
in a complete loss in PP2A inhibition (>104-fold loss in activity)
as anticipated, removal of the C11-alcohol resulted in a 250-
fold loss in PP2A inhibition, and removal of the terminal C12-
C18 triene resulted in a 220-fold loss in PP2A inhibition.
Insights into the origin of these contributions as well as the
behavior of the cytostatin diastereomers were retrospectively
derived from the modeling of its binding to a PP2A homology
model18 and comparison to our earlier fostriecin model.2,16
Finally, and as a consequence of cytostatin’s structural similarity
to sultriecin (3), the corresponding cytostatin sulfate was
prepared for preliminary comparison.
well-founded assumption of a similar cyclic structure for
cytostatin, its reported H10-H11 coupling constant (J10,11
)
9.4 Hz)1a indicated a 10,11-anti configuration for the natural
product. Alternatively, assumption of the adoption of one of
two possible hydrogen-bonded chair conformations for the C9-
C11 structure of cytostatin gives rise to a similar anti H10-
H11 coupling constant and led Waldmann14 to the same (10S)-
stereochemical assignment (Figure 3C).
To decipher the C4-C6 relative stereochemistry, four dia-
stereomeric lactones (C1-C7 partial structures) were synthe-
sized for spectral comparison (Scheme 1). Construction of
lactone 12 began with silylation (TBDPSCl, imidazole, DMF,
25 °C, 2 h, 99%) of methyl (S)-3-hydroxy-2-methylpropionate
(7), followed by conversion of the ester to the corresponding
aldehyde 9 (DIBAL-H, toluene, -78 °C, 1 h; TPAP, NMO,
CH2Cl2, 0 °C, 30 min, 82%, two steps).21 Crotylation of 9 (cis-
2-butenyldiisopinocampheylborane, THF, ether, -78 °C, 12 h;
NaOH, H2O2, 70 °C, 5 h, 63%, 8:1 dr)22 gave alcohol 10, which
was converted to the R,â-unsaturated lactone 12 via acylation
(acryloyl chloride, i-Pr2NEt, CH2Cl2, 0 °C, 2 h, 91%) and
subsequent ring-closing metathesis (Grubbs’s I catalyst, CH2-
Cl2, 40 °C, 12 h, 89%).23 This sequence was used to access
lactones 13, 14, and 15 in a stereodivergent manner by treating
(R)- or (S)-9 with the appropriate crotylboration reagent,
followed by acylation and ring-closing metathesis (Scheme 1).
Upon examination of the 1H NMR spectra of 12-15, it
became apparent that the 4,5-syn and 4,5-anti configurations
exhibit distinct coupling constants (Figure 4).24 Thus, the
coupling constants observed for cytostatin were indicative of a
4,5-syn relative configuration.25 Of the two 4,5-syn lactones,
the 1H NMR spectrum of 12 was closer to that of cytostatin, as
Chemistry
Stereochemical Assignment. The structure of cytostatin was
disclosed without a definition of its relative or absolute stereo-
chemistry, prompting us to secure a stereochemical assignment
prior to initiating the synthetic work. In previous efforts, we
defined the (5S,9S,11S)-stereochemistry for fostriecin,10a and this
assignment was extended to cytostatin on the basis of its
structural and functional similarity (Figure 3A). In these studies,
we observed an intramolecular hydrogen bond between the
C9-phosphate and C11-hydroxy group of fostriecin and dem-
onstrated that the resulting cyclic structure exists in a rigid
twist-boat conformation (Figure 3B) that gives rise to distinct
1H-1H coupling constants between H11 and H10a (syn) or
H10b (anti) (J10a,11 ) 3.7 Hz, J10b,11 ) 9.6 Hz). Based on a
(21) Ley, S. V.; Anthony, N. J.; Armstrong, A.; Brasca, M. G.; Clarke, T.;
Culshaw, D.; Greck, C.; Grice, P.; Jones, A. B.; Lygo, B.; Madin, A.;
Sheppard, R. N.; Slawin, A.; Williams, D. J. Tetrahedron 1989, 45, 7161.
(22) Brown, H. C.; Bhat, K. S.; Randad, R. S. J. Org. Chem. 1989, 54, 1570.
(23) Grubbs, R. H.; Chang, S. Tetrahedron 1998, 54, 4413.
(20) Lawhorn, B. G.; Boga, S. B.; Wolkenberg, S. E.; Boger, D. L. Heterocycles
2006, 70, in press.
9
16722 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Total Synthesis of Cytostatin and Key Analogues
A R T I C L E S
Figure 5. Synthetic plan for cytostatin.
Scheme 2. Synthesis of C1-C7
rapid access to the C10-C11 stereoisomers of 1 that could be
used to define the role of the C11-hydroxy and C10-methyl
group in PP2A binding. The convergent approach relied on
installation of the triene at a late stage and in a single step by
addition of 18 to a C11 aldehyde, enlisting substrate control to
set the C11 stereochemistry. Assembly of the C7-C8 bond by
coupling a cuprate derived from iodide 16 with epoxide 17
isolates the two stereotriads of cytostatin, allowing for inde-
pendent adjustment of their stereochemistry. In addition, this
bond construction provided the opportunity to utilize lactones
12-15 as intermediates in the total synthesis of the natural
product and its C4-C6 diastereomers. A Sharpless epoxidation
served to set the C9 and C10 stereochemistry, and the triene
was synthesized in a short, stereospecific approach relying on
an electrocyclic ring opening to set the geometry of the three
olefins (Figure 5).
Synthesis of C1-C7. Lactone 12, developed during our work
to secure a stereochemical assignment for cytostatin, served as
the starting point for synthesis of the C1-C7 fragment of
cytostatin (Scheme 2). The electrophilic R,â-unsaturated lactone
was masked by carbonyl reduction (DIBAL-H, CH2Cl2, -78
°C, 30 min) and conversion of the intermediate lactol to the
methyl acetal 19 (PPTS, MeOH, 25 °C, 10 min, 82%, two steps).
Acetal 19 formed as a single isomer26 under these conditions
but equilibrated to a 2:1 mixture of anomers when subjected to
MeOH/PPTS treatment for longer reaction times. Although the
C1 stereocenter would eventually be removed, selective forma-
tion of (1R)-19 allowed us to carry forward a single diastere-
omer, simplifying the isolation and characterization of inter-
mediates in the synthesis. Interestingly, the alternative and more
direct generation of 19 employing the corresponding methoxy
acetal precursor in the Grubbs’s metathesis cyclization proved
less satisfactory, with the substrate being more challenging to
effectively prepare from 10 (30-50%) and the product being a
mixture of C1 diastereomers. After desilylation of 19 (Bu4NF,
THF, 25 °C, 12 h) to give 20, efforts to directly convert the
alcohol to the corresponding halide proved problematic, as 20
either remained unchanged (PPh3, NIS; PPh3, CBr4) or decom-
posed (PPh3, I2) under standard conditions. However, 20 could
Figure 4. 1H NMR comparison for lactones 12-15.
exemplified by an average absolute difference in chemical shift
values (for H2-H6) of only 0.08 ppm, suggesting a 5,6-anti
relative configuration for the natural product. Most notably, the
chemical shifts (CD3OD) of 13 for H4, H5, 4-CH3 (δ 1.18),
and especially the key 6-CH3 (δ 0.88) were considerably distinct
from those of 1 (4-CH3 at δ 1.00, 6-CH3 at δ 0.98) and 12
(4-CH3 at δ 1.08, 6-CH3 at δ 0.99). On the basis of this analysis,
the (4S,5S,6S,9S,10S,11S)-configuration was assigned to cy-
tostatin. Independent of our efforts, Waldmann14 conducted and
has disclosed an analogous set of studies using a different
lactone series and a similar retrospective interpretation of the
cytostatin spectroscopic properties to arrive at the same stere-
ochemical assignments for which our relative and absolute
stereochemical assignments of the fostriecin distal C5 and C9/
C11 centers served as the foundation.10
In digesting the conformational features embedded in this
segment of cytostatin, we came to recognize that the distin-
guishing H5-H6 coupling constant was also diagnostic of the
side chain adopting a single, preferred conformation (Figure
4B). This single side-chain orientation is adopted to avoid the
syn pentane interactions of the two alternatives (∆E ≈ 3.7 kcal/
mol). In addition to accounting for the large H5-H6 coupling
constant (J ) 9.5-10.2 Hz) characteristic of their anti relation-
ship in this conformation, this also suggests that the role of the
cytostatin C4- and C6-methyl groups is to confine the side chain
to a single orientation, facilitating binding to PP2A, a potential
role that is reinforced in our retrospective modeling of the
cytostatin binding to a PP2A homology model.
Synthetic Plan. We designed a synthetic route to cytostatin
that would confirm its stereochemical assignment and provide
(24) Similar coupling patterns have been reported for other 4,5-disubstituted-
R,â-unsaturated lactones: (a) Ley, S. V.; Armstrong, A.; Diez-Martin, D.;
Ford, M. J.; Grice, P.; Knight, J. G.; Kolb, H. C.; Madin, A.; Marby, C.
A.; Mukherjee, S.; Shaw, A. N.; Slawin, A. M.; Vile, S.; White, A. D.;
Williams, D. J.; Woods, M. J. Chem. Soc., Perkin 1 1991, 667. (b) Keck,
G. E.; Li, X.-Y.; Knutson, C. E. Org. Lett. 1999, 1, 411. (c) Marshall, J.
A.; Adams, N. D. J. Org. Chem. 1999, 64, 5201. (d) Diez-Margin, D.;
Kotecha, N. R.; Ley, S. V.; Mantegani, S.; Menendez, J. C.; Organ, H.
M.; White, A. D. Tetrahedron 1992, 48, 7899.
(26) The trans-C1 stereochemistry was assigned on the basis of H3-H4 coupling
(J ) 6.0 Hz). Similar trans-1,4-disubstituted dihydropyrans display an H3-
H4 coupling (J ) 5.7 Hz) distinct from that of cis-1,4-disubstituted
dihydropyrans (J ) 1.9 Hz). See: Valverde, S.; Bernabe, M.; Garcia-Ochoa,
S.; Gomez, A. M. J. Org. Chem. 1990, 55, 2294.
(25) The H4-H5 coupling constant was initially reported as J4,5 ) 10.4 Hz,1a
but our measurement on natural cytostatin indicates J4,5 ) 2.7 Hz, consistent
with Waldmann’s report of this value.14d
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16723

A R T I C L E S
Lawhorn et al.
Scheme 3. Synthesis of C8-C11
Scheme 4. Synthesis of C12-C18
Scheme 5. Synthesis of Cytostatin
be converted to iodide 16 through a two-step process of
tosylation (p-TsCl, NaH, benzene, 25 °C, 4 h, 81%, two steps)
followed by iodide displacement (NaI, acetone, 56 °C, 12 h,
90%). The C1-C7 precursor 16 contains the C4-C6 stereotriad
of cytostatin, installed in a stereochemically tunable sequence
through Brown crotylboration of chiral building block 9.
Synthesis of C8-C11. Assembly of the C8-C11 epoxide
unit (Scheme 3) began with 22, which was prepared on a
multigram scale by a known procedure27 that relies on a
Sharpless asymmetric epoxidation reaction to set the chirality
of the epoxy alcohol. Epoxide opening using a procedure
developed by Tius (MeMgI, CuI, Et2O, THF, -40 °C, 3 h,
69%)28 provided a 3:1 mixture of 23 and its regioisomer (1,2-
diol), which was removed by periodate treatment (NaIO4, H2O,
Et2O, 25 °C, 1.5 h, 91%) followed by chromatography. After
debenzylation (H2, Pd/C, MeOH, 25 °C, 2 h, 99%), triol 2429
was transformed to epoxide 17 in a sequence that mirrors a
reported conversion of epi-(3R)-24 into its corresponding
epoxide.30 Thus, 1,2-diol protection (3,3-dimethoxypentane,
p-TsOH, DMF, 25 °C, 12 h, 68%) followed by PMB ether
formation (NaH, PMBCl, Bu4NI, THF, 0 to 20 °C, 20 h, 80%)
gave 26. After acetal removal (p-TsOH, MeOH, 25 °C, 7 h,
89%), diol 27 was transformed to epoxide 17 through tosylation
of the primary alcohol, followed by its intramolecular displace-
ment (NaH, p-Ts-imidazole, THF, -78 to 25 °C, 4 h, 82%).
The C8-C11 unit bears the C9 and C10 stereocenters of
cytostatin that contain the subtituents proposed to affect PP2A
active-site binding. Alternative C9-C10 stereochemical con-
figurations of cytostatin are accessible by transforming the
stereoisomers of triol 24 to their corresponding epoxides through
this simple and reliable synthesis.
(PPh3)4, ether, 0 °C, 15 min, 95% conversion)33 gave 18
stereoselectively. Isolation of bromide 18 was complicated by
its volatility and tendency to polymerize during distillation or
concentration. However, chromatographic purification (silica gel,
pentane) followed by careful concentration in the presence of
Et3N to prevent trace acid-induced polymerization allowed
isolation of 18 in superb yields (73%), considering its lability.
Bromide 18 proved more problematic to store, but it was readily
generated immediately prior to use from the dibromide 29,
which was stable for months when stored as a refrigerated
benzene solution. The C12-C18 precursor 18 contains the
(Z,Z,E)-conjugated triene, ready for incorporation into cytostatin
as a single unit.
Synthesis of C12-C18. Crucial to the implementation of
the convergent synthetic plan set forth herein was the ability to
supply and successfully manipulate the potentially labile bro-
motriene 18. In a modification of Taylor’s synthesis of Z,E-
dienals,31 addition of MeLi (THF, -78 °C, 2 h) to pyrylium
tetrafluoroborate, followed by electrocyclic ring opening, pro-
vided 28 in 66% as a single isomer (Scheme 4). Transformation
of 28 to dibromoolefin 29 (CBr4, PPh3, Et3N, CH2Cl2, 0 °C, 10
min, 97%)32 and selective (E)-bromide reduction (Bu3SnH, Pd-
Synthesis of Cytostatin. Assembly of 1 (Scheme 5) was
initiated by converting iodide 16 into the corresponding cuprate
(t-BuLi, ether, -78 °C, 5 min; then (2-Th)CuCNLi, THF, -78
to 0 °C, 5 min),34 followed by addition of 17 at 0 °C, to give
30 in 84%. The coupling of 16 and 17 was slow when alternative
metalated forms of 16 were used (i.e., R2CuLi, RMgI/cat. CuI,
R2CuCNLi2, RLi/BF3), and the success of the transformation
was dependent on the use of the higher order cuprate at 0 °C
and at concentrations of 50 mM or greater. Acetal formation
(ethyl vinyl ether, PPTS, CH2Cl2, 25 °C, 30 min, 90%) gave
31 as a 1.2:1 mixture of diastereomers which, for ease of
isolation and characterization, were separated after PMB
(27) Mori, K.; Seu, Y. B. Tetrahedron 1988, 44, 1035.
(28) Tius, M. A.; Fauq, A. H. J. Org. Chem. 1983, 48, 4131.
(29) (a) Mori, K.; Nomi, H.; Chuman, T.; Kohno, M.; Kato, K.; Noguchi, M.
Tetrahedron 1982, 38, 3705. (b) Shimizu, Y.; Kiyota, H.; Oritani, T.
Tetrahedron Lett. 2000, 41, 3141. (c) Kobayashi, Y.; Tan, C.; Kishi, Y. J.
Am. Chem. Soc. 2001, 123, 2076.
(30) Sabes, S. F.; Urbanek, R. A.; Forsyth, C. J. J. Am. Chem. Soc. 1998, 120,
2534.
(31) (a) Belosludtsev, Y. Y.; Borer, B. C.; Taylor, R. J. K. Synthesis 1991, 320.
(b) Furber, M.; Herbert, J. M.; Taylor, R. J. K. J. Chem. Soc., Perkin Trans.
1 1989, 683.
(32) (a) Corey, E. J.; Fuchs, P. L. Tetrahedron Lett. 1972, 3769. (b) Ramirez,
F.; Desai, N. B.; McKelvie, N. J. Am. Chem. Soc. 1962, 84, 1745.
(33) Uenishi, J.; Kawahama, R.; Yonemitsu, O.; Tsuji, J. J. Org. Chem. 1998,
63, 8965.
(34) Lipshutz, B. H.; Moretti, R.; Crow, R. Organic Syntheses; Wiley: New
York, 1993; Collect. Vol. VIII, p 33.
9
16724 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Total Synthesis of Cytostatin and Key Analogues
A R T I C L E S
Figure 6. Diastereoselectivity of the triene addition.
removal (DDQ, CH2Cl2, H2O, 25 °C, 30 min, 85%) and carried
through the subsequent four steps of the synthesis separately.
Oxidation of 32 (DMP, CH2Cl2, 25 °C, 15 min, 91%)35 provided
aldehyde 33, setting the stage for installation of the triene.
Bromine/lithium exchange on 18, followed by addition of 33,
gave 34 as a 2:1 mixture of diastereomers, favoring the
undesired 10,11-syn isomer (Felkin product). Following Still’s
precedent,36 the diastereoselectivity of the addition could be
modulated by using a copper-based nucleophile, which promotes
addition to a chelated â-alkoxy aldehyde. Thus, conversion of
18 to its Bu3P-stabilized cuprate (t-BuLi, ether, -78 °C, 1.5 h;
CuI-PBu3, 10 min) prior to addition of 33 (-78 °C, 30 min,
76%) produced 34 as a pair of chromatographically separable
diastereomers in a ratio of 7:1, favoring the desired 10,11-anti
product. As anticipated, the diastereoselectivity of the triene
addition was also highly dependent on the C9 protecting group
(Figure 6). When the nonchelating TES group was used instead
of EE, the Felkin product (syn) predominated when either the
organolithium or organocuprate nucleophile was added to the
aldehyde. This observation is consistent with the Felkin versus
chelation model for the diastereoselective addition and proved
useful for the synthesis of the cytostatin stereoisomers. Assign-
ment of the C10-C11 relative stereochemistry of anti-34 was
initially based on its H10-H11 coupling constant (J ) 9.6 Hz),
which was indicative of a 10,11-anti configuration and consistent
with a C9-alkoxy/C11-hydroxy hydrogen-bonded cyclic struc-
ture (Figure 3), and this stereochemical assignment was later
confirmed through analysis of the 9,11-acetonide derived from
a more advanced synthetic intermediate. The minor isomer (syn-
34) from the â-chelation-controlled triene addition exhibited an
H10-H11 coupling (J ) 3.6 Hz) expected to arise from a 10,-
11-syn configuration.
Figure 7. Confirmation of the C9-C11 stereochemistry.
removal (Et3N, CH3CN, 25 °C, 17 h, 99%) provided 1, identical
to a sample of natural cytostatin (1H NMR, TLC, HPLC,
HRMS).
At this stage, the C9-C11 relative stereochemistry of
synthetic cytostatin was confirmed by examining the spectral
properties of acetonide 41, derived from intermediate 37.
Cleavage of the silyl ether of 37 (HF-pyr, THF, pyridine, 25
°C, 2 h, 99%) provided dephosphocytostatin (40), which was
subjected to acetal formation (2,2-dimethoxypropane, p-TsOH,
THF, 25 °C, 1 h, 99%) to deliver 41 in excellent yield. The
acetonide methyl groups (δ 24.6, 25.1) and ketal carbon (δ
101.2) of 41 exhibited 13C NMR chemical shifts (CD3CN, 150
MHz) characteristic of anti-1,3-diol acetonides (δ 23.6-25.6
and 100.2-101.0) and distinct from those observed for syn-
1,3-diol acetonides (δ 18.6-19.9/29.8-30.2, and 98.0-99.3).37
Furthermore, the H10-H11 coupling constant (J ) 8.4 Hz)
confirmed the expected 10,11-anti stereochemistry. Dephos-
phocytostatin (40) displayed spectral properties similar to those
of 41, and particularly striking are its H10-H11 coupling
constant (J ) 8.4 Hz) and C10-methyl group 13C NMR chemical
shift (δ 10.8, CD3CN, 150 MHz), both of which are indistin-
guishable from those of 41. This comparison suggests that the
C9-C11 hydrogen-bonded cyclic structure of 40 (and 1) exists
in the twist-boat conformation observed for 41 and fostriecin
(Figure 3B) rather than the alternative chair conformation
(Figure 3C) suggested by Waldmann. The latter places the C10-
methyl group in an equatorial position that would result in a
downfield shifted 13C NMR resonance for the C10-methyl
instead of the identical signal that is observed.
Synthesis of Cytostatin Diastereomers. The completion of
the total synthesis of cytostatin set the stage for preparation of
the C10-C11 diastereomers (42-44, Figure 8) that were used
to confirm the C9-C11 stereochemical assignments and to
further delineate the cytostatin-PP2A interaction. epi-(11R)-
Cytostatin (42) was synthesized by advancing the 10,11-syn
isomer of 34 through the final six steps of the cytostatin
synthesis, which occurred without incident and was conducted
without significant optimization of the individual steps (Scheme
6). As noted above, syn-34 was the major product of the addition
of triene 18 to aldehyde 33 when the lithium derivative of 18
was used as the nucleophile (Figure 6), and this procedural
adjustment allowed accumulation of sufficient material to
produce 42.
Following silylation (TBSCl, imidazole, DMF, 25 °C, 2 h,
95%) of alcohol 34, treatment of 35 with dilute HCl for short
reaction times (0.02 N HCl, acetone, water, 25 °C, 10 min, 85%)
induced simultaneous C1 and C9 acetal hydrolysis without
affecting the silyl ether, whereas more conventional acidic
deprotection methods (0.2 N HCl-acetone; 80% aqueous
HOAc; 90% TFA-H2O) led to competitive global deprotection,
even when conducted at low temperatures. In this context, the
use of the C9 EE protecting group proved valuable, as it not
only served to direct the formation of the C11 stereocenter
(through chelation) but also could be effectively removed in
the presence of the labile silyl ether and conjugated triene.
Selective oxidation (Ag2CO3-Celite, benzene, 80 °C, 2 h, 80%)
of lactol 36 produced 37, which was phosphorylated (i-Pr2NP-
(OFm)2, tetrazole, CH3CN, CH2Cl2, 25 °C, 15 min; H2O2, 10
min, 82%) to give 38 using a modification of the protocol (H2O2
vs mCPBA) introduced by Waldmann14c albeit with an early
synthetic intermediate in their approach. Desilylation (HF-pyr,
THF, pyridine, 25 °C, 4 h, 85%) followed by fluorenylmethyl
(37) (a) Rychnovsky, S. D.; Skalitzky, D. J. Tetrahedron Lett. 1990, 31, 945.
(b) Evans, D. A.; Rieger, D. L.; Gage, J. R. Tetrahedron Lett. 1990, 49,
7099.
(35) Dess, D. B.; Martin, J. C. J. Am. Chem. Soc. 1991, 113, 7277.
(36) Still, W. C.; Schneider, J. A. Tetrahedron Lett. 1980, 21, 1035.
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16725

A R T I C L E S
Lawhorn et al.
Scheme 7. Synthesis of 43 and 44
Figure 8. Structures of cytostatin C10-C11 diastereomers.
Scheme 6. Synthesis of 42
The synthesis of cytostatin isomers 43 and 44 bearing the
(10R)-configuration was achieved by utilizing the known
epoxide 5030 in place of 17 in the synthesis (Scheme 7). A key
element of the derivation of the cytostatin synthesis in which
the two stereotriads are independently created and joined in a
convergent epoxide opening, followed by a single-step, late-
stage addition of the intact triene with diastereocontrolled
introduction of the C11 center, is that the approach provides
rapid, divergent access (11 steps from convergence point) to
each C9-C11 diastereomer. Thus, coupling of 16 and 50
proceeded as expected under the previously optimized conditions
(t-BuLi, ether, -78 °C, 5 min; then (2-Th)CuCNLi, THF, -78
to 0 °C, 5 min; 50, ether, 0 °C, 1 h, 72%) to provide 51 in
superb yield. A C9 TES group was selected in preference to
EE for synthesis of the (10R) series since â-chelation was not
required for installation of the natural (11S)-configuration and
the silyl ether formation does not produce additional diastere-
omers. Thus, through a sequence of silylation (TESCl, imida-
zole, DMF, 25 °C, 1 h, 87%), PMB removal (DDQ, CH2Cl2,
H2O, 25 °C, 15 min, 72%), and oxidation (DMP, CH2Cl2, 25
°C, 10 min, 90%),35 51 was converted to aldehyde 54. Addition
of 54 to the lithium derivative of 18 (t-BuLi, ether, -78 °C,
1.5 h; 54, -78 °C, 1 h, 70%) produced 55 as a 2:1 mixture of
diastereomers, favoring the Felkin 10,11-syn product (Figure
6). The expected stereochemical assignments were confirmed
on the basis of the H10-H11 coupling for 55 (J10,11 ) 9.0 Hz
for anti, J10,11 ) 2.4 Hz for syn). After chromatographic
separation, syn-55 and anti-55 were converted to 43 and 44 by
way of the process developed for the synthesis of 1 (Scheme
7). Of note, the TES group was also cleaved at a useful rate
under the conditions (0.02 N HCl, acetone, water, 25 °C, 10
min, 61%) developed for removal of the EE group in the
presence of the triene and labile allylic TBS ether. Thus, the
utility of our convergent total synthesis of cytostatin is high-
lighted by the fact that only 23 additional steps were required
to prepare its entire complement of C10-C11 diastereomers.
Confirmation of Stereochemical Assignment. The stereo-
chemistry of the C9-C11 portion of cytostatin has thus far been
assigned on the basis of a key coupling constant (J10,11 ) 9.4
Hz) that suggests a 10,11-anti relative stereochemistry and the
assumption that cytostatin and fostriecin have identical absolute
configurations at shared chiral centers (C9 and C11). While this
is sound reasoning and synthetic 1 proved spectroscopically
indistinguishable from natural cytostatin,14,20 the magnitude of
the optical rotation of synthetic 1 both reported by Waldmann14c
9
16726 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Total Synthesis of Cytostatin and Key Analogues
A R T I C L E S
Scheme 8. Synthesis of Cytostatin Analogues
Waldmann to impurities in his natural sample that resulted from
long-term storage.14c
Synthesis of Additional Key Analogues. In addition to the
C10-C11 diastereomers, our synthetic route provided access
to several other analogues useful for defining the impact of the
C9-C11 substituents and the triene unit on PP2A inhibition.
Intermediate 37, ideally suited for the preparation of phosphate
replacements, provided not only dephosphocytostatin (40) but
also, by sulfation (SO3-pyridine, THF, 25 °C, 30 min, 71%)
followed by desilylation (HF-pyr, THF, pyridine, 25 °C, 1.5
h, 83%), 67, the sulfated version of cytostatin (Scheme 8).
Conversion of acetal 30 to its corresponding lactone through
hydrolysis (0.02 N HCl, acetone, water, 25 °C, 10 min, 99%)
and oxidation (Ag2CO3-Celite, benzene, 80 °C, 2 h, 96%) gave
68. Phosphorylation (i-Pr2NP(OFm)2, tetrazole, CH3CN, CH2-
Cl2, 25 °C, 15 min; H2O2, 10 min, 88%) of 68 was followed by
Et3N treatment (Et3N, CH3CN, 25 °C, 20 h, 88%) to provide
the phosphate 70 (Scheme 8). Alternatively, cleavage of the
PMB group of 69 (DDQ, CH2Cl2, H2O, 25 °C, 30 min, 70%),
followed by Et3N treatment (Et3N, CH3CN, 25 °C, 17 h, 99%),
produced 72, a cytostatin C1-C11 partial structure that lacks
the C12-C18 triene unit.
To provide a means to quantify the impact of the C11-
hydroxy on PP2A binding, 11-deshydroxycytostatin (78) was
constructed through a late-stage modification of the cytostatin
total synthesis (Scheme 9). Alcohol 32 was transformed to iodide
73 (NIS, PPh3, imidazole, THF, 25 °C, 5 min, 74%), a suitable
coupling partner for bromotriene 18 at the correct oxidation state
to deliver 78. The union of 18 and 73 was achieved through
alkylation of the cuprate derived from 18 (t-BuLi, ether, -78
°C, 1 h; CuI-PBu3, 10 min; 73, 0 °C, 1 h, 63%) to provide 74
in good yield with only minimal efforts at optimization. Notably,
this cuprate coupling represents a nice example of an sp2-sp3
organometallic coupling reaction enlisting an unactivated sp3
halide complementary to the more contemporary Stille, Suzuki,
and Negishi reactions of vinyl stannanes, boronic acids, and
organic zinc reagents currently being explored with sp3 halides.
Acetal hydrolysis (0.02 N HCl, acetone, water, 25 °C, 10 min,
77%), followed by lactol oxidation (Ag2CO3-Celite, benzene,
80 °C, 30 min, 76%), gave lactone 76, which after phospho-
1
Figure 9. Partial H NMR comparison for natural cytostatin, 1, and 42-
44.
([R]20D +46 (c 0.27, MeOH)) and found herein ([R]25D +45 (c
0.07, MeOH)) did not match that measured for natural cytostatin
([R]25 +20 (c 0.27, MeOH)) by Waldmann on an aged
D
authentic sample,14c leaving room for doubt about the identity
of natural cytostatin. Visual comparison of the 1H NMR spectra
of the four C10-C11 diastereomers (1, 42, 43, 44, Figure 9)
with the spectrum of natural cytostatin demonstrates that
synthetic 1 matches natural cytostatin identically, while 42-
44 display substantially different resonances arising from C11-
H, C9-H, and C10-Me. In general, the C9 signals for the
10,11-syn isomers (42, 43) were farther upfield than for the
10,11-anti isomers (1, 44), while the C11 signals in the syn
compounds were farther downfield than those of the anti
isomers. The 10,11-anti isomer 44 most closely resembles 1
spectrally, but the large difference in the shifts of their H11
protons (∆δ ) 0.22 ppm) clearly distinguishes the two. The
1
sensitivity of the H NMR signals to stereochemical changes
in the C9-C11 region and the magnitude of the H10-H11
coupling for each diastereomer can be best interpreted by their
adoption of a C9-phosphate/C11-hydroxy hydrogen-bonded
twist-boat conformation. These results also remove any doubt
about the 9,10-relative configuration, since similarly large
spectral changes would be observed if the C9-C11 cyclic
structure were altered through that locus. In total, the dissonance
of the alternative C10-C11 diastereomers with cytostatin, the
comparison of the C1-C7 lactone partial structures with
cytostatin, the spectral identity of synthetic 1 with natural
cytostatin, the biological profile of synthetic 1, and the fact that
synthetic 1 shares the same sign of optical rotation with natural
cytostatin authenticate the (4S,5S,6S,9S,10S,11S) assignment.
The discrepancy in the magnitude of the optical rotation of
synthetic 1 and natural cytostatin has been attributed by
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16727

A R T I C L E S
Lawhorn et al.
Scheme 9. Synthesis of 78
Figure 10. Protein phosphatase and cytotoxic activity (IC50, µM).
both assays, fostriecin was found to be more potent than
cytostatin (3-11 fold), but the difference was much more
significant (11-fold) in the assay using a natural substrate
(phosphohistone), as shown in Figure 10. Additionally, both
fostriecin (50-fold) and cytostatin (15-fold) as well as the
subsequent analogue series (typically 2-7-fold) exhibited more
potent activity against PP2A when a natural (phosphohistone)
versus artificial (DiFMUP) substrate was employed. Both results
are summarized in Figure 10, and each assay provided quali-
tatively identical trends in relative activity for the series
examined, but those derived from the phosphohistone assay
proved easier to distinguish and are probably the most relevant
to discuss. Here, cytostatin proved to be ca. 10-fold less potent
than fostriecin, and both failed to inhibit PP1 or PP5, illustrating
their exquisite selectivity (>105-fold for fostriecin and >103-
104-fold for cytostatin).
C9-Phosphate Modifications. Two key analogues of cy-
tostatin containing modifications to the C9-phosphate were
prepared for examination. The first, dephosphocytostatin (40),
proved inactive (IC50 >100 µM), as anticipated, and analogous
to observations made with dephosphofostriecin,16 illustrating that
the phosphate is an essential component of the PP pharmacoph-
ore. The second, sulfocytostatin (67), replaced the phosphate
with a sulfate, providing a key analogue that is nearly identical
in structure to sultriecin (3), a naturally occurring antitumor
agent.6a Given the structural similarity of sultriecin to cytostatin
and fostriecin, it may have been reasonable to suggest that its
antitumor activity arises from a mechanism operative for 1 and
2 and that it would inhibit PP2A, although this has yet to be
examined. Accordingly, the assessment of sulfocytostatin (67)
was conducted not only to probe the importance of the C9-
phosphate but also in anticipation that it would provide insight
into the mechanism of action of sultriecin itself. Remarkably,
sulfocytostatin was inactive against all PP’s examined but did
exhibit cytotoxic acitivity (L1210 IC50 ) 7 µM) at a level
consistent with that observed for sultriecin (IC50 ) 0.75-9
µM).6a In addition to illustrating again the essential role of the
C9-phosphate, this result would suggest that the antitumor
activity of sultriecin does not arise from PP inhibition.
rylation (i-Pr2NP(OFm)2, tetrazole, CH2Cl2, CH3CN, 25 °C, 30
min; H2O2, 5 min, 65%) and Et3N treatment (Et3N, CH3CN, 25
°C, 17 h, 79%) provided 78.
Biological Evaluation
Assay Substrate and Cross-Assay Comparisons: Cytosta-
tin versus Fostriecin. One of the first questions addressed was
the relative PP2A inhibition activity of cytostatin versus
fostriecin. Fostriecin reportedly exhibits potent PP2A inhibition
when enlisting phosphorylase a (IC50 ) 1-3 nM, rabbit),7,16
glycogen phosphorylase (IC50 ) 1.5 nM, bovine),38 or phos-
phohistone (IC50 ) 2-5.6 nM, rabbit and human)9,39 as the
substrate in assays containing roughly the same amounts of
enzyme; it is slightly less effective when enlisting phospho-
peptide KR as the substrate (IC50 ) 40 nM, human)40 and is
substantially less effective when enlisting commercial phos-
phopeptide, phosphocasein, and p-nitrophenylphosphate as
substrates (IC50 ) 5-40 µM, human).9 In contrast, albeit in a
much more limited series of comparisons, cytostatin displays a
narrower and less potent range of PP2A inhibition, depending
on the substrate enlisted, with reports of IC50’s of 210 nM (p-
nitrophenylphosphate, human at 0.05 unit),5 33 nM (p-nitro-
phenylphosphate, bovine at 0.025 unit),14 and 100 nM (phos-
phorylase a, rabbit at 0.01 unit).41 It is clearly less potent than
fostriecin using phosphorylase a, but it is less sensitive to the
substrate choice in the assay and remains active when enlisting
p-nitrophenylphosphate as the substrate. Consequently, we
conducted side-by-side comparisons of cytostatin and fostriecin,
enlisting both phosphohistone (a natural substrate) and 6,8-
difluoro-4-methylumbelliferyl phosphate (DiFMUP, a simplified
artificial substrate)42 as substrates for bovine brain PP2A.7 In
(38) Hastie, C. J.; Cohen, P. T. W. FEBS Lett. 1998, 431, 357.
(39) Cheng, A.; Balczon, R.; Zuo, Z.; Koons, J. S.; Walsh, A. H.; Honkanen,
R. E. Cancer Res. 1998, 58, 3611.
The C11-Alcohol. Probably one of the most ambiguous
features of the cytostatin and fostriecin structures has been the
importance and role of the C11-hydroxy group. Both we16 and
Waldmann14 have reported derivatives (-OAc) that were
inactive against PP2A but with analogues that were altered in
ways that also significantly attenuated their PP2A inhibition.
Thus, it was clear that the C11-alcohol plays an important role,
consistent with its conserved presence in all the natural products,
(40) Roberge, M.; Tudan, C.; Hung, S. M. F.; Harder, K. W.; Jirik, F. R.;
Anderson, H. Cancer Res. 1994, 54, 6115.
(41) Launey, T.; Endo, S.; Sakai, R.; Harano, J.; Ito, M. Proc. Natl. Acad. Sci.
U.S.A. 2004, 101, 676.
(42) Structure of DiFMUP:
9
16728 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Total Synthesis of Cytostatin and Key Analogues
A R T I C L E S
but the magnitude of its effect could not be established in these
studies. Consequently, 11-deshydroxycytostatin (78), lacking
only this key C11-alcohol, was prepared for direct comparative
examination (PP2A IC50 ) 6.9 µM), and it was found to be
250-fold less potent than cytostatin. Consistent with this impact,
the retrospective cytostatin-PP2A modeling as well as our
earlier fostriecin-PP2A model2 defined a key hydrogen bond
between the C11-hydroxy group and a conserved Arg214. This
hydrogen bond stabilizes a turn in the bound conformation of
1 or 2, orienting the terminal triene into a hydrophobic pocket,
where it π-stacks with the indole side chain of Trp200.
Notably, the activity of 70, which also lacks the C11-hydroxy
group and replaces the sensitive triene with a p-methoxybenzyl
ether, exhibited PP2A inhibition (IC50 ) 14 µM) only 2-fold
less potently than 11-deshydroxycytostatin, suggesting that the
PMB ether may serve as a simplified, stable replacement for
the labile triene. The retrospective model places this PMB
replacement directly over the Trp200 indole, benefiting from
the same π-stacking as the authentic C12-C18 triene.
The C12-C18 Triene. The examination of 72, lacking only
the entire C12-C18 segment of cytostatin, permitted a direct
assessment of the (Z,Z,E)-triene contribution to the magnitude
of PP2A inhibition potency and selectivity. Analogue 72
inhibited PP2A (IC50 ) 5.9 µM) but not PP1 or PP5, maintaining
selectivity for PP2A, but it was 220-fold less active than
cytostatin itself. This observation complements Waldmann’s
examination of a truncated C12-C13 alkyne and a C12-C13
(Z)-vinyl iodide, which were found to be 10-fold less active
than and equipotent with cytostatin, respectively, albeit in a
PP2A assay enlisting p-nitrophenylphosphate as substrate.14 Our
earlier model of fostriecin bound at the PP2A active site and
the retrospective modeling studies of cytostatin that followed
defined a beautiful π-stacking interaction of nearly all of the
triene with the indole side chain of Trp200 in a hydrophobic
cleft, providing a compelling rationale for the (Z,Z,E)-triene
enhancement of binding affinity and its unusual stereochemistry.
Moreover, this model suggests that the largest share of the
π-stacking interaction occurs with C13-C17, implying that the
terminal (Z,E)-diene is more engaged than the internal C12-
C13 (Z)-alkene.
followed did not reveal a conformational role for this substituent,
but it does fit snugly into a hydrophobic pocket defined by
Tyr127 and Tyr265 and shields the active site and C9-phosphate
from solvent access. Adoption of the unnatural (10R)-config-
uration with 43 precludes methyl binding in this pocket, but it
does not introduce destabilizing interactions, as the methyl group
now extends out of the active site into solvent-accessible space
and no longer shields the active site and C9-phosphate. Although
it is not easy to define a role for this substituent that might
account for a 780-fold increase in inhibitory activity, it is
consistent with its proposed role18 of mimicking a phospho-
threonine methyl group, paralleling the substrate preference of
PP2A for phosphothreonine peptides (5-100 fold) over phos-
phoserine peptides.47 In the retrospective modeling, the methyl
group of cytostatin resides close to the phosphate, tucked into
a hydrophobic pocket, whereas that of 43 is distal from the
phosphate and incapable of serving as such a threonine methyl
mimic. Moreover, in the side-by-side comparison with our
fostriecin-PP2A model, we came to recognize that the fostriecin
C8-methyl group and the cytostatin C10-methyl group extend
into the same enzyme pocket and that they are spacially much
closer to one another than might be intuitively expected, being
displaced by only 2.85 Å. For us, this provides a compelling
case for suggesting that the cytostatin C10-methyl group, like
the fostreicin C8-methyl group, serves as a mimic of the
phosphothreonine methyl group, accounting for its prominent
role.
Surprisingly and interestingly, the (10R,11R)-diastereomer 44,
in which the stereochemistry of both the C10-methyl group and
the C11-alcohol is inverted, was the most potent of the series
(IC50 ) 3.8 µM), being 140-fold less active than the natural
product. A cursory examination of the retrospective model did
not reveal insights into why 44 might be a 5-fold more effective
inhibitor than either 42 or 43, but the magnitude of the loss in
activity is still so significant as to reflect a substantially disrupted
binding.
The sensitivity of the cytostatin-PP2A interaction to the
stereochemical orientation of the C9-C11 segment underscores
the critical nature of the C9-C11 substituents and confirms that
the proximal methyl and C11-hydrogen-bonding groups are
important elements of the PP pharmacophore for cytostatin and
its congeners.
The C10/C11 Cytostatin Diastereomers 42-44. Each of
the cytostatin diastereomers 42-44 proved to be significantly
or substantially less potent than the natural product itself. This
satisfying result not only reinforces the natural product structural
assignment but also indicates that both the C10 and C11
substituents productively contribute to the properties of cytosta-
tin. The C11 epimer 42, in which only the stereochemistry of
the alcohol is inverted, proved to be 700-fold less potent against
PP2A than cytostatin itself. This C11 epimer proved to be even
less potent than 78, lacking the C11-alcohol, as well as 72,
lacking the entire triene segment, consistent with a substantially
disrupted binding that precludes effective accommodation of
both the inverted alcohol and the triene simultaneously.
Cytotoxic Activity. We also evaluated compounds 1, 2, 40,
72, and 78 for in vitro cytotoxic activity against the L1210 cell
line (Figure 10). In this functional assay, fostriecin (IC50 ) 0.3
µM) and cytostatin (IC50 ) 0.6 µM) exhibited comparable
activity, albeit with cytostatin being roughly 2-fold less potent.
Similarly, dephosphocytostatin (40, IC50 ) 8 µM) was 10-fold
less active than the natural product and comparable in activity
to dephosphofostriecin (IC50 ) 20 µM),16 both exhibiting
potencies that most consider as nearly inactive. Finally, both
(43) rhPP1R was expressed in E. coli with the pKK-223-2 vector (accession
no. M77749) and purified essentially as described: Zhang, L.; Lee, E. Y.
Biochemistry 1997, 36, 8209.
(44) (a) Honkanen, R. E.; Zwiller, J.; Moore, R. E.; Daily, S. L.; Khatra, B. S.;
Dukelow, M.; Boynton, A. L. J. Biol. Chem. 1990, 265, 19401. (b) Huang,
X.; Swingle, M. R.; Honkanen, R. E. Methods Enzymol. 2000, 315, 579.
(45) Swingle, M. R.; Honkanen, R. E.; Ciszak, E. M. J. Biol. Chem. 2004, 279,
33992.
(46) Gee, K. R.; Sun, W.-C.; Bhalgat, M. K.; Upson, R. H.; Klaubert, D. H.;
Latham, K. A.; Haugland, R. P. Anal. Biochem. 1999, 273, 41.
(47) Agostinis, P.; Goris, J.; Pinna, L. A.; Marchiori, F.; Perich, J. W.; Meyer,
H. E.; Merlevede, W. Eur. J. Biochem. 1990, 189, 235.
The C10 epimer 43, in which only the stereochemistry of
the methyl group is inverted, was even less active, being ca.
780-fold less potent than cytostatin. This result indicates a
prominent role for this substituent and its stereochemistry and
was especially surprising since the inversion of the C10 center
was expected to have the least perturbation on the conformation
of free cytostatin. The retrospective modeling studies that
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16729

A R T I C L E S
Lawhorn et al.
Figure 12. The C1-C8 region of fostriecin (A) and cytostatin (B) bound
to PP2A.
reversible, conjugate addition to the R,â-unsaturated lactone
(CysSH-C3 distance is 2.59 and 3.15 Å for fostriecin and
cytostatin, respectively), which is activated by an Arg268
hydrogen bond to the lactone carbonyl, accounting for the PP
selectivity and the enhanced potency (g200-fold for fostriecin
relative to 2,3-dihydrofostriecin16 and ca. 1000-fold for an
analogous pair of simplified cytostatin analogues14). This is
illustrated in Figure 12 for cytostatin, which also highlights a
key conformational feature imposed by the interacting C4- and
C6-methyl groups. The presence and stereochemistry of these
two substituents restrict the side chain to a single orientation
that avoids syn pentane interactions and facilitates active-site
binding (see also Figure 4). Fostriecin accomplishes the same
side-chain restriction, although with a slightly altered orientation,
through preferential adoption of a C5 H-eclipsed conformation
with the C6-C7 olefin.
The largest differences in the two models naturally emerge
in the C6-C8 region, with cytostatin adopting a gauche turn at
C7-C8, whereas fostriecin accomplishes this at C8-C9. This
latter turn in fostriecin is stablilized by a C8-OH hydrogen
bond to Asp57 and its cooperative binding to the proximal metal
cation, potentially displacing a reactive active-site water nu-
cleophile enlisted for phosphate hydrolysis. Moreover, this
places the fostriecin C8-methyl group in a hydrophobic pocket
potentially reserved for the methyl group of phosphothreonine
substrates. This suggests important roles for both the presence
and stereochemistry of each of the fostriecin C8 substituents,
and it is notable that Shibasaki has reported that C8-epi-
fostriecin is 2000-fold less potent than the natural product in
assays enlisting a natural substrate (phosphohistone).9
Figure 11. Models of fostriecin (A) and cytostatin (B) bound to PP2A.
72 and 78 were found to be roughly 10-fold less potent than
cytostatin, consistent with their reduced PP2A inhibition.
Models of Cytostatin and Fostriecin Bound at the PP2A
Active Site. Complementary to our own earlier modeling2,16 of
fostriecin bound to a homology model of PP2A,18 we examined
the binding of cytostatin at the PP2A active site. The former
model not only provided insights into the active-site binding
by fostriecin, identifying key features of the interaction, but also
defined the origin of the PP2A selectivity that stems from
reversible conjugate addition of the active-site Cys269 (unique
to PP2A and PP4 and absent in PP1, PP2B, PP5, and PP7) to
the R,â-unsaturated lactone. Although this covalent addition to
fostriecin or cytostatin has not been experimentally verified, it
was originally supported by the decreased affinity of 2,3-
dihydrofostriecin (g200-fold less potent than 2)16 and confirmed
with the more recent isolation and identification of the analogous
Cys269 adduct with phoslactomycin (6).17
Since the details of the fostriecin-PP2A model have not been
disclosed elsewhere, it is presented herein alongside the newly
generated cytostatin-PP2A model. Side-by-side overviews of
the two models are presented in Figure 11 with space-filling
representations of fostriecin and cytostatin. Both adopt nearly
identical bent conformations, with the C9-phosphate binding
and bridging both active-site metal cations. The lactone contacts
the â12-â13 loop stabilized by hydrophobic interactions with
the surrounding residues (Tyr267), and Cys269 is poised for
The backbones for the remainder of the bound cytostatin and
fostriecin are essentially identical, being anchored by the C9-
9
16730 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Total Synthesis of Cytostatin and Key Analogues
A R T I C L E S
Figure 13. The C9-C11 region of fostriecin (A) and cytostatin (B) bound
to PP2A.
phosphate binding and bridging of both active-site metal cations
and the C11-hydroxy group forming a key hydrogen bond with
the conserved Arg214 (Figure 13). The latter hydrogen bond
induces, or at least stabilizes, a turn in the bound conformations,
orienting the C12-C18 triene into a hydrophobic cleft, where
it stacks with the indole side chain of Trp200. Notably, the
unusual (Z,Z,E)-triene geometry allows it to π-stack across the
entire length of the indole, maximizing the intermolecular
binding stabilization (Figure 14). The importance of the
phosphate binding is clear from the reduced activity of dephos-
phofostriecin (>105-fold)16 and dephosphocytostatin (>103-
104 fold), whereas the importance of the latter two interactions
is clear from the evaluations herein of 11-deshydroxycytostatin
(250-fold) and 72, lacking the entire triene segment (220-fold).
No active-site interaction for the fostriecin terminal alcohol was
detected, consistent with its lack of impact on the PP2A
inhibition2,16 and its absence in all other related natural products,
including cytostatin. The largest and only other difference
between cytostatin and fostriecin throughout this region is the
cytostatin C10-methyl group. This methyl group resides close
to the phosphate, tucked into a hydrophobic pocket, potentially
mimicking the methyl group of a phosphothreonine substrate.
In comparing fostriecin and cytostatin models, the cytostatin
C10-methyl group was found to extend into the same hydro-
phobic pocket as the fostriecin C8-methyl group, and the two
are spacially closer to one another than one might suspect (2.85
Å) (Figure 13). The importance of the cytostatin C10-methyl
(780-fold reduction for epimer) and the analogous importance
of the fostriecin C8 substituents (2000-fold reduction for epimer)
make a compelling case for suggesting that these centers have
similar roles in their respective natural products, in which the
methyl groups mimic a phosphothreonine substrate methyl group
Figure 14. The C12-C18 region of fostriecin (A) and cytostatin (B and
C) bound to PP2A.
used by and sufficient for the enzyme to distinguish its substrates
from phosphoserine peptides.
Conclusion
Synthetic studies leading to a definitive assignment of the
relative and absolute stereochemistry of cytostatin and culminat-
ing in its total synthesis and that of each of its C10-C11
stereoisomers were conducted. Key elements of the strategy
include a convergent assembly of the natural product core that
permits the early-stage independent adjustment of the C4-C6
or C9-C10 stereochemistry, coupled with a late-stage single-
step installation of the sensitive (Z,Z,E)-triene, with a reaction
that allows substrate control of the C11 stereochemistry.
Extensions of the synthetic approach provided additional key
analogues, including dephosphocytostatin (40), sulfocytostatin
(67), 11-deshydroxycytostatin (78), and 72, lacking the entire
triene unit. Biological assessment of cytostatin, its diastereomers,
and the key analogues demonstrated the importance and
quantitated the magnitude of the presence and stereochemistry
of the C10/C11 substituents (>140-fold), the C9-phosphate
(>103-104-fold), C10-methyl (ca. 780-fold), C11-hydroxy (250-
fold), and the C12-C18 (Z,Z,E)-triene (220-fold) contributions
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16731

A R T I C L E S
Lawhorn et al.
to the potent PP2A inhibition. Retrospective modeling of
cytostatin within the active site of a PP2A catalytic subunit
homology model and its comparison with an analogous fos-
triecin-PP2A model provided key insights into the role of each
of the natural product segments, their substituents, and the
accompanying stereochemistry. Remarkably, each element of
the cytostatin structure plays a productive, and sometimes
surprisingly prominent, role in its PP2A inhibition. Finally,
indirect evidence is provided that sultriecin, despite its structural
similarity to cytostatin, does not apparently derive its cytotoxic
and antitumor activity through protein phosphatase inhibition.
Acknowledgment. We gratefully acknowledge the support
of the NIH (CA93456 and CA42056, D.L.B., L.A., and R.E.H.).
We thank Prof. Ishizuka of the Microbial Chemistry Research
Foundation for an authentic sample of cytostatin and Inkyu
Hwang for conducting the cytotoxic activity assays.
Supporting Information Available: Full experimental details.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA066477D
9
16732 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006