Structure-based design of a novel synthetic spiroketal pyran as a pharmacophore for the marine natural product Spongistatin 1

Article abstract of DOI:10.1016/S0960-894X(00)00044-5

SPIKET-P, a novel synthetic spiroketal pyran, was rationally designed as a pharmacophore for the tubulin depolymerizing marine natural product Spongistatin 1. SPIKET-P was prepared from the commercially available benzyl (R)-(-)-glycidyl ether using a versatile 11-step synthetic scheme in a stereocontrolled fashion. At nanomolar concentrations, SPIKET-P caused tubulin depolymerization in cell-free turbidity assays and exhibited potent cytotoxic activity against cancer cells as evidenced by destruction of microtubule organization, and prevention of mitotic spindle formation in human breast cancer cells. (C) 2000 Elsevier Science Ltd. All rights reserved.

Full text of DOI:10.1016/S0960-894X(00)00044-5

Bioorganic & Medicinal Chemistry Letters 10 (2000) 541±545
Structure-Based Design of a Novel Synthetic Spiroketal Pyran as
a Pharmacophore for the Marine Natural Product Spongistatin 1
Fatih M. Uckun,* Chen Mao, Alexei O. Vassilev, He Huang and Shyi-Tai Jan
Drug Discovery Program and Parker Hughes Cancer Center, Parker Hughes Institute, St. Paul, MN 55113, USA
Received 6 December 1999; accepted 13 January 2000
AbstractÐSPIKET-P, a novel synthetic spiroketal pyran, was rationally designed as a pharmocophore for the tubulin depolymer-
izing marine natural product Spongistatin 1. SPIKET-P was prepared from the commercially available benzyl (R)-( )-glycidyl
ether using a versatile 11-step synthetic scheme in a stereocontrolled fashion. At nanomolar concentrations, SPIKET-P caused
tubulin depolymerization in cell-free turbidity assays and exhibited potent cytotoxic activity against cancer cells as evidenced by
destruction of microtubule organization, and prevention of mitotic spindle formation in human breast cancer cells. # 2000 Elsevier
Science Ltd. All rights reserved.
Microtubules play a pivotal role in mitotic spindle
assembly and cell division.1±5 These cytoskeletal ele-
ments are formed by the self-association of the ab
tubulin heterodimers.1±5 Recently, the structure of the
ab tubulin dimer was resolved by electron crystal-
lography of zinc-induced tubulin sheets.6 According to
the reported atomic model, each 46Â40Â65 A tubulin
monomer is made up of a 205 amino acid N-terminal
GTP/GDP binding domain with a Rossman fold topol-
ogy typical for nucleotide-binding proteins, a 180 amino
acid intermediate domain comprised of a mixed b sheet
and ®ve helices which contains the taxol binding site,
and a predominantly helical C-terminal domain impli-
cated in binding of microtubule-associated protein
(MAP) and motor proteins.2,5
model of the ab tubulin dimer for the identi®cation of
the potential tubulin binding sites for SP. Our inte-
grated e€ort to identify the SP binding site involved
`cavity searching', analysis of the binding environment,
docking procedures based on the electron crystal-
lographic coordinates of tubulin, and the calculation of
the binding constant by a modi®ed score function
(LUDI score function).
The large molecular volume of SP, which has a >10 A
length and >10 A width, prompted us to hypothesize
that it likely binds to a deep pocket on the surface of
tubulin, which provides a highly hydrophobic environ-
ment for molecular interactions. We examined the elec-
tron crystallographic structure of tubulin using graphics
programs including GRASP and INSIGHT II to iden-
tify a possible binding site for SP that would have
suitable dimensions and contain a cluster of hydro-
phobic residues near the protein surface. The systematic
search resulted in the discovery of a candidate binding
pocket on the tubulin surface that is large enough to
accommodate SP and provides an environment for
extensive hydrophobic interactions. The SP molecule
was then docked into this candidate binding site using
the Anity module within the Insight II program. The
initial coordinates of SP, representing several di€erent
conformations, were modeled and energy-minimized
using the DISCOVER program. The docking results,
which utilized di€erent conformations of SP, were
analyzed and ranked based on their interaction scores.
The SP model that was assigned the highest interaction
score and lowest potential energy was chosen for further
Spongistatins are macrocyclic lactone compounds con-
taining six pyran-type rings and four of the six pyran
rings are incorporated into two spiro[5.5]ketal moieties
(Fig. 1A).7 Spongistatin 1 (SP) (R=Cl; Fig. 1A) is a
potent tubulin depolymerizing natural product which
has been isolated from an Eastern Indian Ocean sponge
in the genus Hytrios.7 SP exhibited potent cytotoxicity
with subnanomolar IC50 values against the members of
the NCI panel of 60 human cancer cell lines.7 In a
rational drug design e€ort intended to determine the
minimal molecular architecture of the SP structure
necessary for biologic activity, we used the 3-D atomic
*Corresponding author. Tel.: +1-651-697-9228; fax: +1-651-697-
1057.
0960-894X/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(00)00044-5

542
F. M. Uckun et al. / Bioorg. Med. Chem. Lett. 10 (2000) 541±545
Figure 1. Modeling design of SPIKET-P as a new pharmocophore of tubulin depolymerizing agents: A, structures of Spongistatin and SPIKET-P;
B: the space-®lling model of b tubulin shown with ball-and-stick models of GDP and taxol molecules binding to the di€erent sites as labeled.
Spongistatin binding pocket (SBP) is marked in orange color for all aromatic residues in a close range. The residues N101 (colored in red ) is located
near the GDP exchange site (see text); C, SPIKET-P as a Spongistatin pharmacophore. SPIKET-P interaction with the putative SBP illustrated in a
surface model of the SBP and stick-and-ball model for SPIKET-P.
calculations. The docking simulation results indicated
that the putative SP binding pocket which is located on
the surface of tubulin is approximately 8 A wideÂ18 A
longÂ11 A deep (Fig. 1B). This unique pocket consists
of an unusual cluster of 10 aromatic residues situated in
close proximity that includes Y108, W103, Y185, W407,
Y408, F399, F404, F395, F418 and H406 (Fig. 1C).
pocket contacts the longitudinal interdimer interface of
the microtubule. The binding of SP to this pocket may
therefore hinder interdimer interactions of tubulin and
contribute to the tubulin depolymerizing activity of SP.
Advanced modeling studies of the interactions of SP
with this putative SP binding pocket indicated that the
two spiroketal groups of SP are in close contact with
protein residues lining the binding pocket and may
therefore serve as the critical binding components of SP.
The identi®cation of the spiroketal subunits as the likely
tubulin binding elements of SP prompted the hypothesis
that a synthetic spiroketal pyran (SPIKET-P) repre-
senting these subunits (Fig. 1A) could serve as a phar-
macophore for SP. Docking studies indicated that,
when bound to tubulin, the spiroketal ring of SPIKET-
P would be sandwiched between aromatic residues F404
The proposed tubulin binding pocket for SP is in close
proximity to the GDP exchange site on the b subunit of
the tubulin heterodimer (Fig. 1B). This location for the
SP binding pocket would provide a cogent explanation
for the ability of SP to inhibit the displacement of the
bound GDP molecules from tubulin.8,9 Furthermore,
according to the recently published high-resolution
model of the microtubule structure,10 this binding
Scheme 1. Retro-synthetic analysis.

F. M. Uckun et al. / Bioorg. Med. Chem. Lett. 10 (2000) 541±545
543
and W407 in the binding pocket and would provide
favorable hydrophobic interactions and van der Waals
contacts with these residues (Fig. 1C). The spiroketal
group of SPIKET-P has a molecular surface of 218 A,2
75% of which would be covered by the aforementioned
two aromatic rings. Therefore, SPIKET-P was selected
as a synthetic target.
Retro-synthetic analysis (Scheme 1) was started by con-
verting the spiroketal group in SPIKET-P to the car-
bonyl and hydroxy groups in 1, which was further
converted to two segments 2 and 3. Both 2 and 3 could
be derived from the commercially available benzyl (R)-
( )-glycidyl ether 5. The synthesis was initiated by
opening the commercially available epoxide 5 using
vinylmagnesium bromide to obtain the alcohol 6 which
was protected as tert-butyldimethylsilyl ether to form 7
(Scheme 2). Hydroboration of the terminal ole®n in 7
yielded the primary alcohol as in 8, which was then
converted to the mesylate 4. The mesylate group in 4
was substituted by cyanide to form 9 which was further
converted to aldehyde 2 by DIBAL reduction followed
by acid catalyzed hydrolysis. Alternatively, the mesylate
group in 4 was substituted by bromide to form 3.
Compound 3 was coupled with 2 by ®rst converting 3 to
a Grignard reagent and then reacting it with aldehyde 2
Figure 2. Turbidity measurements. Compounds (in 1% DMSO) were
added to tubulin (1 mg/mL, 0.1 M MES, 1 mM EGTA, 0.5 mM MgCl2,
0.1 mM EDTA, 2.5 M glycerol, 1 mg/mL leupeptin, 1 mg/mL aprotinin,
pH 6.5) followed by stimulation of polymerization with 1 mM GTP at
2 min. Optical density was measured using a Becton Dickinson UV
spectrophotometer (350 nm) using a thermostated cuvette holder to keep
the reaction at 37 ꢀC. Readings obtained from the spectrophotometer
were standardized by subtracting the background absorbance of the
compound in water from the sample reading following drug addition.
Scheme 2. Stereocontrolled synthesis of the SP pharmacophore SPIKET-P: (a) vinylMgBr, CuBr, 2 h, 0 ꢀC; 81%; (b) TBDMSCl, imidazole,
DMAP, 1 h at 0 ꢀC, 3 h at rt; 98%; (c) (1) BH3-THF, (2) H2O2, NaOH; 70%; (d) MsCl, Et3N, 2 h, 0 ꢀC; 94%; (e) NaCN, DMSO, 15-Crown-5,
40 ꢀC, overnight; 95%; (f) DIBAL-H, 78 ꢀC 3 h, 10% tartaric acid; 93%; (g) LiBr, acetone, re¯ux, 0.5 h; 80%; (h) Mg/Et2O; 45%; (i) oxalyl
chloride, DMSO, Et3N, 78 ꢀC; 83%; (j) 5% HF/CH3CN, 0.5 h, rt; 100%; (k) LDBB, 0 ꢀC; 50%.

544
F. M. Uckun et al. / Bioorg. Med. Chem. Lett. 10 (2000) 541±545
Figure 3. E€ects of SPIKET-P1 on mitotic spindle assembly in human breast cancer cells. BT-20 cells in log phase were trypsinized and seeded onto
sterile 22 mm2 coverslips in six-well plates. Cells on coverslips were returned to the incubator for 24 h prior to treatment. The following day
SPIKET-P (or Spongistatin 1) was added to yield a ®nal concentration of 10 nM. Final DMSO concentration was 0.025%. Cells were returned to a
37 ꢀC incubator for 24 h. Cells on coverslips were extracted prior to ®xation by a 10 min treatment with 4% CHAPS detergent in 1.0 PHEM bu€er
(60 mM Pipes, 25 mM Hepes, 10 mM EGTA, 4 mM MgCl2, pH 6.95). Cells were then ®xed for 10 min in 1.0 PHEM bu€er+0.15% glutar-
aldehyde and 2% formaldehyde. This was followed by a 15 min incubation in phosphate bu€ered saline+0.1% Triton X-100(PBS+Tx). Coverslips
were next incubated for 40 min at 37 ꢀC with a primary antibody recognizing a-tubulin (Sigma, St Louis, MO) in a humidi®ed chamber. Coverslips
were washed for 15 min in PBS-Tx followed by a 40 min incubation with a ¯uorescently labeled secondary antibody (Jackson Immunoresearch,
West Grove, PA). The coverslips were again rinsed in PBS-Tx and incubated with 5 mM Dapi (Sigma, St Louis, MO) for 20 min to label the DNA.
Coverslips were immediately inverted onto slides in Vectashield (Vector Labs, Burlingame, NH) to prevent photobleaching, sealed with nail polish
and stored at 4 ꢀC. Slides were examined using a Nikon TE200 inverted microscope equipped for epi¯uoresence and with high numerical aperture
objectives. Digital images were collected with a Princeton Instruments Micromax digital camera using Metamorph software (Universal Imaging,
West Chester, PA). Digital data was prepared for publication using Photoshop (Adobe Systems, Mountain View, CA) software and printed on a
Pictrography printer (Fuji Photo, Elmsford, NY).
Figure 4. SPIKET-P e€ects on MDA-MB-231 breast tumor cells. Vehicle-treated control cells are adherent with a well organized microtubule
cytoskeleton (A, green). In contrast, cells treated with 500 nM SPIKET-P (B) for 24 h have fewer, less organized microtubules. Additionally,
membrane blebbing and nuclear fragmentation is observed. Large vacuoles are present within treated cells. Green=microtubules; blue=DNA;
bar=20 mm.
to form 10 (Scheme 2). Swern oxidation converted 10 to
1. Deprotection of the two tert-butyldimethylsilyl pro-
tected hydroxyl groups in 1 followed by acid catalyzed
acetal formation to give the immediate benzyl-protected
precursor of SPIKET-P, 11, was carried out by treating
1 with 5% HF in acetonitrile at room temperature for
30 min. After the unsuccessful attempt of removing the
two benzylic protecting groups in 11 by platinum cata-
lyzed hydrogenation, SPIKET-P was obtained by treating
11 with lithium 4,40-di-tert-butylbiphenylide (LDBB).11

F. M. Uckun et al. / Bioorg. Med. Chem. Lett. 10 (2000) 541±545
545
We ®rst examined the ability of SPIKET-P to cause tubu-
References and Notes
1. Avila, J. Life Sci. 1992, 50, 327.
2. Hyams, J. S.; Lloyd, C. W. Microtubules; New York, 1994.
3. Hyman, A.; Karsenti, E. J. Cell Sci. 1998, 111, 2077.
4. Downing, K. H.; Nogales, E. Curr. Opin. Cell Biol. 1998, 10,
16.
5. Kozielski, F.; Arnal, I.; Wade, R. H. Curr. Biol. 1998, 8, 191.
6. Nogales, E.; Wolf, S. G.; Downing, K. H. Nature 1998, 391,
199.
7. Pettit, G. R.; Cichacz, Z. A.; Gao, F.; Herald, C. L.; Boyd,
M. R. J. Org. Chem. 1993, 58, 1302.
8. Bai, R.; Cichacz, Z. A.; Herald, C. L.; Pettit, G. R.; Hamel,
E. Molecular Pharmacology 1993, 44, 757.
9. Bai, R.; Taylor, G. F.; Cichacz, Z. A.; Herald, C. L.; Kepler,
J. A.; Pettit, G. R.; Hamel, E. Biochemistry 1995, 34, 9714.
10. Nogales, E.; Whittaker, M.; Milligan, R. A.; Downing, K.
H. Cell 1999, 96, 79.
11. Cohen, T.; Jeong, I.-H.; Mudryk, B.; Bhupathy, M.;
Awad, M. M. A. J. Org. Chem. 1990, 55, 1528.
12. Vassilev, A.; Ozer, Z.; Navara, C.; Mahajan, S.; Uckun, F.
M. J. Biol Chem. 1999, 274, 164616.
lin depolymerization or prevent tubulin polymerization.
Bovine brain tubulin (Sigma, St. Louis, MO) was used
in standard turbidity assays to test the e€ect of SPI-
KET-P on GTP-induced tubulin polymerization. As
shown in Figure 2, SPIKET-P caused partial depoly-
merization of tubulin and inhibited GTP-induced tubu-
lin polymerization. We next examined the e€ects of
SPIKET-P on human breast cancer cells (Fig. 3) using
confocal laser scanning microscopy.12 At low nanomo-
lar concentration, SPIKET-P abrogated cell division in
human breast cancer cells by preventing mitotic spindle
assembly. Whereas 2.3% of vehicle-treated control cells
were in mitosis at 24 h, none of the SPIKET-P treated
cells was. Confocal images of SPIKET-P treated breast
cancer cells were virtually identical to those of Spongis-
tatin 1 treated breast cancer cells. As shown in Figure 4,
breast cancer cells treated with higher concentrations of
SPIKET-P showed destruction of the microtubule
organization, membrane blebbing, and nuclear frag-
mentation consistent with apoptosis.
13. Characterization data: [(2R,8R)-8-(hydroxymethyl)-1,7-
1
dioxaspiro[5,5]undec-2-yl]methan-1-ol (SPIKET-P): H NMR
In summary, SPIKET-P, a new pharmocophore of the
tubulin depolymerizing marine natural product Spon-
gistatin 1, has been rationally designed based upon the
recently resolved electron crystallographic structure of
tubulin. SPIKET-P, which was synthesized in a stereo-
controlled fashion from the commercially available
benzyl (R)-( )-glycidyl ether in 11 steps,13 caused tubu-
lin depolymerization in turbidity assays, prevented
mitotic spindle formation, destroyed the microtubule
organization, and induced apoptosis in human breast
cancer cells.
(300 MHz, CDCl3) d 1.26±1.53 (m, 6H), 1.63 (m, 4H), 1.88 (m,
2H), 2.14 (bs, 2H), 3.52 (dd, J=11.0, 7.0 Hz, 2H), 3.62 (dd,
J=11.0, 3.5 Hz, 2H), 3.75 (m, 2H); 13C NMR (75 MHz,
CDCl3) d 96.0, 69.7, 66.2, 35.2, 26.3, 18.2; GC/MS m/z 216
(M+); IR (neat): 3373, 2939, 2871 cm 1; [a]2d2 59.4 (c 0.7,
CHCl3); X-ray: Space group: P21, unit cell: a=11.4510(15) A,
b=8.9628(11) A, c=22.942(3) A, a=90ꢀ, b=93.396(3)ꢀ,
g=90ꢀ, volume=2350.5(5) A3, Z=8, y range for data collec-
tion=1.78 to 28.32ꢀ (l=0.71073 A), total re¯ections col-
lected=12,806, independent re¯ections=9044, data/restraints/
parameters=9044/1/542, R1 [I >2s(I)]=0.139, wR2=0.187,
goodness of ®t on F 2=1.009.