Synthesis, Modeling, and Anti-Tubulin Activity of a D-Seco Paclitaxel Analogue

Article abstract of DOI:10.1021/ol036204c

(Equation presented) We have previously described a model of paclitaxel-microtubule binding that led to the prediction that analogues of paclitaxel lacking any D ring could stabilize microtubules as well as paclitaxel if the substituent present at C4 did

Full text of DOI:10.1021/ol036204c

ORGANIC
LETTERS
2004
Vol. 6, No. 4
461-464
Synthesis, Modeling, and Anti-Tubulin
Activity of a D-Seco Paclitaxel Analogue
,†
Luciano Barboni,* Guido Giarlo,† Massimo Ricciutelli,† Roberto Ballini,†
,‡
Gunda I. Georg,* David G. VanderVelde,‡ Richard H. Himes,§ Minmin Wang,|
,|
Ami Lakdawala,| and James P. Snyder*
Dipartimento di Scienze Chimiche, UniVersita` di Camerino, Via S. Agostino 1,
62032 Camerino (MC), Italy, Department of Medicinal Chemistry, UniVersity of
Kansas, Drug DiscoVery Program, Higuchi Biosciences Center, UniVersity of Kansas,
1251 Westcoe Hall DriVe, Lawrence, Kansas 66045, Department of Molecular
Biosciences, UniVersity of Kansas, Lawrence, Kansas 66045, and Department of
Chemistry, Emory UniVersity, 1515 Dickey DriVe, Atlanta, Georgia 30322
georg@ku.edu
Received November 10, 2003
ABSTRACT
We have previously described a model of paclitaxel−microtubule binding that led to the prediction that analogues of paclitaxel lacking any D
ring could stabilize microtubules as well as paclitaxel if the substituent present at C4 did not have unfavorable steric interactions with the
binding pocket. We report the synthesis of a 4-methyl paclitaxel analogue, compound 1, which bears this prediction out. Compound 1 is as
potent as paclitaxel at microtubule stabilization in vitro; however, it has only about one-four-hundredth the cytotoxicity of paclitaxel.
As part of our research into the oft-discussed role of the
oxetane ring on the biological activity of the anticancer agent
paclitaxel,1 we recently reported the synthesis of a series of
D-seco analogues, with an acetoxy group at C20 (see
compound 2).2 None of the compounds showed activity in
cytotoxicity and tubulin assembly assays. Following the
receptor models developed by one of us,3,4 we ascribed the
lack of activity to an unfavorable steric effect between the
protein and the C20 acetoxy group.2 Consistent with predic-
tions1 based on this viewpoint, Dubois and co-workers
demonstrated that a docetaxel analogue with only a carbocy-
clic ring (cyclopropyl) at this position (3), stabilizes micro-
tubules, with bioactivity in the microtubule assay being
reduced by only half relative to the parent compound.5 The
oxygen in the oxetane ring is therefore not a requirement
for bioactivity in these compounds. With the aim of testing
whether any ring is necessary at this position, we designed
and synthesized compound 1, in which C20 consists only of
a methyl group. The minireceptor model predicted that this
compound would stabilize microtubules (MTs), and indeed
it does: it is equipotent with paclitaxel. However, it has very
low activity in cytotoxicity assays, which is also the case
* Corresponding authors. (L.B.) Phone: +390737402240. Fax
+390737637345. (J.P.S.) Phone: 404-727-2415. Fax: 404-727-6586.
(G.I.G.) Phone: 785-864-4498. Fax: 785-864-5326.
† Universita` di Camerino.
‡ Department of Medicinal Chemistry and Drug Discovery Program,
Higuchi Biosciences Center, University of Kansas.
§ Department of Molecular Biosciences, University of Kansas.
| Emory University.
(1) Wang, M.; Cornett, B.; Nettles, J.; Liotta, D. C.; Snyder, J. P. J.
Org. Chem. 2000, 65, 1059-1068.
(2) Barboni, L.; Datta, A.; Dutta, D.; Georg, G. I.; Vander Velde, D. G.;
Himes, R. H.; Wang, M.; Snyder, J. P. J. Org. Chem. 2001, 66, 3321-
3329.
(3) Wang, M.; Xia, X.; Kim, Y.; Hwang, D.; Jansen, J. M.; Botta, M.;
Liotta, D. C.; Snyder, J. P. Org. Lett. 1999, 1, 43-46.
(4) Snyder, J. P.; Nettles, J. H.; Cornett, B.; Downing, K. N.; Nogales,
E. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 5312-5316.
(5) Dubois, J.; Thoret, S.; Gue´ritte, F.; Gue´nard, D. Tetrahedron Lett.
2000, 41, 3331-3334.
10.1021/ol036204c CCC: $27.50 © 2004 American Chemical Society
Published on Web 01/20/2004

for cyclopropyl 3.6 The closely related compound that lacks
the 20-methyl group has dramatically reduced microtubule-
stabilizing activity as well as low cytotoxicity.7
Scheme 1a
a Reagents and conditions: (a) (But)4NI, MgBr2‚OEt2, CH2Cl2,
rt, 8 h (87%); (b) H2, PtO2, EtOAc, rt, 48 h (84%); (c) NaBH3CN,
DMPU, 70 °C, 15 h (73%); (d) Ac2O, DMAP, toluene, 80 °C, 48
h (60%).
using higher H2 pressure were unsuccessful, we treated the
5R iodine analogue 6 with NaBH3CN in DMPU10 and
obtained the dehydrohalogenation product 7 (73%).
Acetylation of 7 gave 8 (60%), which was treated with
an excess of phenyllithium. After workup, the reaction
mixture, consisting of multiple products that were not further
characterized, was treated with HF-pyridine, affording
compound 1 in 3% yield (Scheme 2).
Scheme 2a
Our strategy for the synthesis of 1 involves the introduction
of an easily removable group at C20 by nucleophilic ring
opening of the C4-C20 oxirane intermediate 4, which was
prepared from paclitaxel using a reported protocol.8 For this
purpose, the introduction of a halide at C20 as reported by
Dubois and co-workers5 was selected as the most direct
procedure, since halogen atoms are easily removed by
hydrogenolysis.9 Accordingly, compound 4 was treated with
(But)4NI in the presence of a Lewis acid (MgBr2‚OEt2) to
form the 4-hydroxy-20-iodo analogue 55 (87%, Scheme 1).
Hydrogenolysis of the C20 and C5 iodo groups was first
attempted using a Pd/C catalyst,9 but only starting material
was recovered. After some experimentation, it was found
that treatment of 5 with H2 (balloon pressure) and Adams
catalyst (PtO2) gave compound 6 (79%), in which only the
primary halogen was removed. This unusual pattern of
reactivity may be due to steric hindrance at C5 by the
substituents at C4. Since attempts to remove both iodines
a Reagents and conditions: (a) PhLi, THF, -78 °C, 3 h; (b) HF/
py, THF, 0 °C, 4 h (3%).
Our previous work on D-seco compounds led to the
conclusion that the “conformational lock” on paclitaxel,
originally ascribed to the oxetane ring, was more likely due
to the presence of the 4-acetyl group.2 Removal of the
4-acetyl leads to changes in the shape of the diterpene, as
reflected in some key J-couplings, and also to loss of
microtubule-stabilizing activity. It is instructive to compare
(6) Dubois, J.; Gue´ritte, F. Private communucation.
(7) Deka, V.; Dubois, J.; Thoret, S.; Gue´ritte, F.; Gue´nard, D. Org Lett.
2003, ASAP.
(8) Gunatilaka, A. A. L.; Ramdayal, F. D.; Sarragiotto, M. H.; Kingston,
D. G. I.; Sackett, D. L.; Hamel, E. J. Org. Chem. 1999, 64, 2694-2703.
(9) Pinder, A. R. Synthesis 1980, 425-452.
(10) Hutchins, R. O.; Kandasamy, D.; Maryanoff, C. A.; Masilamani,
D.; Maryanoff, B. E. J. Org. Chem. 1977, 42, 82-91.
462
Org. Lett., Vol. 6, No. 4, 2004

1 to paclitaxel and to another compound we have previously
analyzed, 20-acetoxy-4-deacetyl-5-epi-20O-secopaclitaxel
(9), which lacks both the oxetane ring and the 4-acetate.11
Two important descriptors of the overall shape of the
diterpene are the coupling constant between H2 and H3,
which serves as a reporter of the dihedral angle along the
bottom portion of the B ring, and the coupling between H13
and H14R and 14â, which serves a similar function for the
A ring. In paclitaxel, J2-3 is 7.2 Hz; in 1, it is 7.4 Hz
(identical within experimental error). In paclitaxel and in 1,
H13 has nearly equal couplings to both protons at C14, and
the signal is a broad triplet. In 9, J2-3 is 5.3 Hz and H13 is
a doublet of doublets with J-couplings of 4.4 and 10.3 Hz.
Thus, in paclitaxel, the A ring sustains a boat conformation,
while in 9, it adopts an envelope conformation. The main
point of difference between 1 and paclitaxel is the 5-6
double bond, which leads to a flattening of the normal chair
conformation of the C ring and may also provide some
conformational rigidity. In light of the overall conformational
similarity to paclitaxel, this is evidently a local perturbation.
We have also previously reported on solvent-dependent
NOE and J-couplings for paclitaxel, which were interpreted
in terms of a fast conformational equilibrium between
multiple families of conformations, two “collapsed” and one
extended.12 The solution NMR data indicate that the same
is true for compound 1. In chloroform, the J2′-3′ coupling
constant is 2.4 Hz, very similar to that of paclitaxel (2.6 Hz).
In DMSO, J2′-3′ increases to 6.3 Hz (for paclitaxel, 7.0 Hz),
indicating an increase in the population of a “polar collapsed”
conformation.
The D-seco compound 1 is equipotent with paclitaxel in
promoting the polymerization of tubulin to MTs in our in
vitro assay.13 Electron microscopy was used to verify that
the products of the assembly reaction were microtubules.
However, in a cytotoxicity assay employing the MCF-7 cell
line, it had about 0.25% of the activity of paclitaxel. (The
ED50 for paclitaxel in these actively dividing cells is 4 nM;
for 1, the ED50 was 1.6 µM.) The docetaxel cyclopropyl
analogue 3, which similarly lacks an oxygen in the D ring,
also loses much more cytotoxicity than microtubule binding.
In a KB cell line, 3 was found to be about 1% as cytotoxic
as paclitaxel.6
Figure 1. (a) Superposition of paclitaxel (blue) and D-seco 1
(orange) in the electron crystallographic paclitaxel/tublin model4
resulting from flexible docking of 1 into the binding site. The C20
carbons of the two analogues are juxtaposed and in close contact
with Leu215. (b) Overlap of paclitaxel, 1, and 3 (green) in the
subsite around ring-D.
Two computational approaches have led to the prediction
that the D-seco analogue 1 is expected to be highly active
in the microtubule stabilization assay. First, we performed a
flexible docking experiment with 1 in the paclitaxel-tubulin
model derived from electron crystallography (EC).3 The
DOCK procedure14 delivered a binding mode nearly identical
to that of paclitaxel as pictured in Figure 1. Bound to the
protein, the molecule adopts the T-conformation in common
with paclitaxel and analogues.3 Not surprisingly, the rigid
tricyclic A-C ring core facilitates the alignment of the C20
methyl group in 1 and the corresponding methylene carbon
in the oxetane D-ring. As previously noted,1 the hydrogens
on both C20 carbons appear to experience close approach
with the methyls of Leu215 (r(H- - -H) ) 2.6-3.1 Å). While
the binding forms of 1 and paclitaxel are not those observed
as major contributors to the solution conformation, the
geometric similarity in solution as documented in the
previous section is matched by the similarities in the model
binding sites. In an earlier study, we observed the T-
conformation as a low-population (<10%) torsional isomer
in CDCl3.12a Given the overall resemblance of paclitaxel and
1 in chloroform, it can be safely projected that D-seco 1
enjoys low-population residence in the T-shape in condensed
media as well.
(11) Boge, T. C.; Hepperle, M.; Vander Velde, D. G.; Gunn, C. W.;
Grunewald, G. L.; Georg, G. I. Bioorg. Med. Chem. Lett. 1999, 9, 3041-
3046.
(12) (a) Snyder, J. P.; Nevins, N.; Cicero, D. O,; Jansen, J. J. Am. Chem.
Soc. 2000, 122, 724-725. (b) Vander Velde, D. G.; Georg, G. I.; Grunewald,
G. L.; Gunn, C. W.; Mitscher, L. A. J. Am. Chem. Soc. 1993, 113, 11650-
11651.
(13) Liu, Y.; Ali, S. Y.; Boge, T. C.; Georg, G. I.; Victory, S.; Zygmunt,
J.; Marquez, R. T.; Himes, R. H. Combi. Chem. High Thr. Scr. 2002, 5,
39-48.
(14) Ewing, T. J. A.; Kuntz, E. E. J. Comput. Chem. 1997, 18, 1175-
1189; cf. http://www.cmpharm.ucsf.edu/kuntz/kuntz.html.
Org. Lett., Vol. 6, No. 4, 2004
463

microtubule-stabilizing agents by comparison to paclitaxel.
The second-generation minireceptor model yielded the same
result for 31 prior to its reported synthesis and testing,6 the
latter demonstrating the compound to be very similar to
paclitaxel in its microtubule binding. The origin of the
overestimation of the binding capacity of 1 is illustrated by
comparing Figures 1b and 2. Optimization of the structure
in the protein-truncated minireceptor causes the C20 methyl
in 1 to shift approximately 1 Å away from Leu215, thereby
reducing the steric compression implied by the full protein
model (Figure 1).
Contrary to long-held dogma, the results for cyclopropane
3 demonstrated for the first time that neither the intact
oxetane ring nor an oxygen on C-5 are necessary for taxane
efficacy with tubulin.1,5 The microtubule-stabilizing predic-
tion for D-seco 1, its synthesis, and subsequent demonstration
as a potent stabilizer of microtubles constitute another
important exception to the “oxetane rule.” Obviously, the D
ring is not necessary for maintaining the conformational
properties of the taxane diterpene so long as its rupture or
removal is appropriately compensated elsewhere in the
molecule. The situation is reminiscent of the unusual
bioactivity of C2-meta-azido baccatin, a molecule lacking
the otherwise “essential” C13 taxane side chain.18 Unlike
the latter case, however, the full paclitaxel microtubule
assembly activity is retained by D-ring-free 1.
The unexpectedly low cytotoxicity of 1 deserves comment.
Taxane cytotoxicity reflects other factors besides microtubule
binding, including water solubility, partitioning into cells,
and active efflux, among others. We have preliminary data
to suggest that 1 does indeed partition into cells about as
efficiently as paclitaxel, and in a future publication, we will
elaborate on its biological effects and its potential utility.
Figure 2. Superposition of paclitaxel (blue), D-seco 1 (orange),
and cyclopropyl 3 (green) in the third-generation minireceptor; the
B, C and D rings of the taxanes are illustrated along with two
minireceptor residues in the vicinity of ring-D.
The second assessment of the binding capacity of D-seco
1 took advantage of the 3D-QSAR minireceptor approach.
Prior to our work devoted to determination of the EC-derived
conformation of paclitaxel bound to tubulin,3 we developed
a second-generation minireceptor model based on the “non-
polar-collapsed” conformation of the ligand and a compatible
conformation of epothilone A.15 As a result of unsuccessful
attempts to accommodate the activity of various C20
oxygenated D-seco taxanes with this model,2 it became clear
that a minireceptor more faithful to the protein structure was
required. Consequently, a third-generation model based on
T-paclitaxel and the evolving D-seco SAR was subsequently
developed16 within the context of the PrGen package17 and
employed here and in the previous work2 to examine various
D-ring-ruptured paclitaxel analogues. Compound 1 surfaced
as a structure that was both synthetically accessible and
endowed with highly favorable predicted tubulin-polymer-
ization capacity. The D-ring region of our latest minireceptor
is portrayed in Figure 2 with the superimposed structures of
paclitaxel, 1, and 3 in their optimized locations.
Acknowledgment. This work was supported by NIH CA-
82801 (to G. I. G.). L.B. is grateful to Consiglio Nazionale
delle Ricerche, Italy, for a “Mobilita` di breve durata”
fellowship. J.P.S and A.L. are appreciative to Dennis Liotta
(Emory University) for encouragement and support. The
authors thank Franc¸oise Gue´ritte and Joe¨lle Dubois of the
ICSN, Gif-sur-Yvette, France, for helpful discussions and
permission to disclose unpublished results. We also thank
Jacquelyn Huff for her excellent technical assistance.
With paclitaxel estimated binding affinity as a yardstick
(8.9 × 10-6), the corresponding affinities of 1 and 3 were
calculated to be 2.1 × 10-7 and 4.5 × 10-7, respectively.
Both are predicted to be somewhat more effective as
Supporting Information Available: Experimental pro-
cedures and spectral data for compounds 1 and 5-8. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(15) Wang, M.; Xia, X.; Kim, Y.; Hwang, D.; Jansen, J. M.; Botta, M.;
Liotta, D. C.; Snyder, J. P. Org. Lett. 1999, 1, 43-46.
(16) Wang, M.; Lakdawala, A.; Snyder, J. P. Unpublished results.
(17) (a) Vedani, A.; Zbinden, P.; Snyder, J. P.; Greenidge, P. A. J. Am.
Chem. Soc. 1995, 117, 4987-4994. (b) Zbinden, P.; Dobler, M.; Folkers,
G.; Vedani, A. Quant. Struct.-Act. Relat. 1998, 17, 122-130.
OL036204C
(18) He, L.; Jagtap, P. G.; Kingston, D. G.; Shen, H. J.; Orr, G. A.;
Horwitz, S. B. Biochemistry 2000, 39, 3972-3978.
464
Org. Lett., Vol. 6, No. 4, 2004