Restoration of Microtubule Interaction and Cytotoxicity in D-seco Taxanes upon Incorporation of 20-Hydroxymethyl-4-allyloxy Groups

Article abstract of DOI:10.1021/acs.orglett.5b03119

To probe the exact role of the oxetane D ring in both tubulin binding and cytotoxicity of taxanes, novel D-seco taxanes bearing a C4 ether substituent have been prepared from paclitaxel 1a. Among them, 20-hydroxymethyl-4-allyloxy D-seco taxane 5e is the m

Full text of DOI:10.1021/acs.orglett.5b03119

Letter
pubs.acs.org/OrgLett
Restoration of Microtubule Interaction and Cytotoxicity in D-seco
Taxanes upon Incorporation of 20-Hydroxymethyl-4-allyloxy Groups
Shao-Rong Wang,† Chun-Gang Yang,† Pedro A. Sanchez-Murcia,‡ James P. Snyder,§ Ning Yan,†
́
Gonzalo Saez-Calvo,∥ Jose Fernando Díaz,∥ Federico Gago,‡ and Wei-Shuo Fang*,†
́
́
†State Key Laboratory of Bioactive Substances and Functions of Natural Medicines, Institute of Materia Medica, Chinese Academy of
Medical Sciences & Peking Union Medical College, 2A Nanwei Road, Beijing 100050, China
‡
́
Area de Farmacología, Departamento de Ciencias Biomed
ica del CSIC, E-28871 Madrid, Spain
§Department of Chemistry, Emory University, 1515 Dickey Drive, Atlanta, Georgia 30322, United States
∥Centro de Investigaciones Biolog
icas, Consejo Superior de Investigaciones Científicas, Ramiro de Maeztu 9, 28040 Madrid, Spain
́ ́ ́
icas, Universidad de Alcala, E-28871 Alcala de Henares, Unidad Asociada
al Instituto de Química Med
́
́
S
* Supporting Information
ABSTRACT: To probe the exact role of the oxetane D ring in
both tubulin binding and cytotoxicity of taxanes, novel D-seco
taxanes bearing a C4 ether substituent have been prepared
from paclitaxel 1a. Among them, 20-hydroxymethyl-4-allyloxy
D-seco taxane 5e is the most active in both tubulin and
cytotoxicity assays. It is only slightly less potent than 1a on
tubulin polymerization promotion in vitro and the most
cytotoxic among all D-seco taxanes known to date. The reason for the loss and restoration of bioactivity for these D-seco taxanes is
also discussed with the assistance of NMR and molecular modeling studies. From these results, we draw a conclusion that the
intact D ring of taxanes is not strictly necessary for their binding to tubulin and cytotoxic effects.
axane antitumor drugs, such as paclitaxel (Taxol) 1a and
Tdocetaxel 1b, have been widely used in clinics for more
than two decades. Although numerous medicinal chemistry
studies have been reported, there are still some questions to be
answered regarding the ligand−protein interaction for this
group of molecules. A better understanding of the interaction of
taxanes with their targets, i.e. β-tubulin in microtubule (MT), in
atomic detail, may give rise to the structure-based design of a
new generation of taxanes with improved activity.
deoxydocetaxel 21 and the 4-methyl D-seco analog 32 are almost
as potent as their parent compounds in the tubulin/MT assays.
Based on the above-mentioned studies, it was proposed that the
D ring is not critical for tubulin binding if the overall
conformation of the core structure and the location of the C4
substituent relative to that at C13 are preserved. When the
structure of the complex between 1a and Zn2+-stabilized
tubulin sheets was solved at 3.5 Å resolution,3 the oxygen atom
in the D ring was shown to establish a H-bond with Thr276 of
β-tubulin. Nonetheless, its removal (as in 2 and 3) was not
followed by a dramatic decrease in the power to promote
tubulin polymerization or to inhibit MT disassembly. In
contrast, a very significant drop in cytotoxicity (2−3 orders
of magnitude) was observed for most D-seco taxanes. Even the
most cytotoxic one, compound 2, is still 100-fold weaker than
its counterpart 1b.4 Although it was suggested that this
discrepancy may originate from cell membrane permeability
and stability, no evidence was available to prove these
suspicions. Therefore, it is reasonable to propose that the C5
oxygen atom may play a critical role for taxanes exerting
cytotoxicity. In fact, the D-seco derivative 4, on which an oxygen
atom was reinstalled at the C5 position, was synthesized to
prove this hypothesis. However, though it is only 3.5-fold
weaker than 1a for the MT disassembly inhibition, it is unable
to rescue the lost cytotoxicity (1000-fold weaker than 1b).4
One remaining unanswered question is the exact role of the
oxetane D ring. This unique four-membered ring was initially
believed to be essential for the bioactivity of 1a. However, more
recent studies cast doubts on this belief. For example, 5(20)-
Received: October 28, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.5b03119
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
To determine the real contribution of the D ring to the
bioactivities of taxanes, some C4 and C20 functionalized D-seco
analogs 5a−e were designed and synthesized. At the very
beginning, reinstatement of an acetoxy group to C4 was
attempted, but was found unfeasible due to undesired acyl
migration between the C4 and C20 hydroxyls. As the published
structure−activity relationships (SAR) on C4 in taxanes
revealed that the cytotoxicity of 4-methoxy-4-deacetyl paclitaxel
1c was only slightly weaker than that of 1a,5 the D-seco taxanes
5a−d bearing 4-OMe were prepared as surrogates at the
beginning.
Scheme 3. Synthesis of Compound 5b
Scheme 4. Synthesis of Compound 5c
Initially, we attempted to introduce the 4-OMe prior to the
oxetane ring opening. However, this changed the conformation
of the C ring (data not shown), thereby preventing the
elimination of the 5-I. Hence, based on the literature protocol,6
intermediate 6 was synthesized from 1a for the follow-up
studies. After the protection of the 20-OH with a triethylsilyl
(TES) group, the elimination of the iodo atom in compound 7
was realized by using the LiCl/NaHCO3 system in DMF, to
give the desired product in high yield (91%) (Scheme 1)
Hydrogenation over Pd/C gave both the desired product 15
and the cyclopropyl analog 16. Referring to the literature,8 the
formation of 16 might be through a palladium-catalyzed
intramolecular cyclodehalogenation. Synthesis of the known
compound 17 from 15 also failed using the reported
conditions,2 but could be realized using the previously
mentioned LiCl/NaHCO3 conditions. Compound 5c was
prepared by a procedure similar to that used for the preparation
of 5a. In addition, the known compound 1c was also prepared
as a control for the bioassays according to the literature
procedure.4
However, targeted compound 5d was not obtained through
either the oxidation of the 20-OH of 5a or other synthetic
strategies because the 4-formyl compound is not stable even
when stored below 0 °C. From our experience in the
preparation of 4-formyl analogs, the aldehyde group is prone
to the attack of the C2 free hydroxyl group, either upon
formation by hydrolysis of the attached acyl group or after
migration of the acyl group from C2 to C1.
Scheme 1. Synthesis of Compound 8
according to the similar synthetic strategy developed in our
lab.7 The selection of the protective group at C20 was
important, since the elimination could not occur if 20-OH was
masked with other groups, such as p-toluenesulfonyl or
methylthiomethyl.
The methylation of 4-OH on 8 seemed more difficult than
that in the preparation of 1c. After many trials, the most
optimal conditions were found using the NaHMDS and
MeOTf system in THF (Scheme 2) to afford 9 in 88% yield
As shown in Table 1, the tubulin binding affinity and
cytotoxicity toward A2780 cells of all 4-OMe D-seco
Scheme 2. Synthesis of Compound 5a
Table 1. Tubulin Binding Affinity and Cytotoxicity of the D-
seco Analogs
tubulin binding
cytotoxicity (IC50, nM)
A2780
compound
(Kb 35 °C, 107 M−1
)
1a
1c
1d
5a
5b
5c
5e
1.43 0.17
0.38 0.07
2.4
80
(based on 65% conversion). The C3′ amide was also
methylated yielding 10 if the reaction conditions were not
carefully controlled. After the opening of the 1,2-carbonate with
PhLi and subsequent desilylation, D-seco taxane 5a was
smoothly obtained in 71% yield after two steps. In contrast,
it was found that the removal of the 2′-O-TBS in 10 was
unexpectedly difficult due to the steric effect introduced by the
3′-N methyl (data not shown).
2.20 0.06
2.5
0.031 0.007
0.0062 0.0065
0.0082 0.0015
0.38 0.02
14 885
25 180
4350
42.9
Compound 5b was prepared from 9, as illustrated in Scheme
3. After selective deprotection of the 20-O-TES group, 11 was
treated with 2,6-di-tert-butyl-4-methylpyridine and MeOTf to
afford 12, which was converted to the desired product 5b over
two steps in moderate yield (35%).
The C20 methyl analog 5c was synthesized from 6 (Scheme
4) by a modified procedure of that reported.2 In accordance
with the literature, catalytic hydrogenation of di-iodo D-seco
taxane 14 using PtO2 as the catalyst gave no reaction.
compounds 5a−c were found to be much lower (12−61-fold
and 2−3 orders of magnitude, respectively) than those of 1c.
Of note, these decreases are significantly higher than those
brought about by replacement of the 4-OAc in 1a with the 4-
OMe in 1c (4- and 33-fold, respectively, relative to 1a).
To understand the poor activity of 5a−c, we first focused on
their NMR parameters, which reflect the possible conforma-
tional changes in the core structure and thus the orientation of
C13 and C4 side chains. It must be borne in mind that opening
B
DOI: 10.1021/acs.orglett.5b03119
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
of the D ring brings about a change of puckering in the C ring
(Figure S1). The major differences between 1c and 5a−c, as
revealed by the NMR data (Table S1), are the chemical shifts of
protons 14α and 14β, which appear >0.2 ppm upfield for the
former and >0.1 ppm downfield for the latter, as well as the
J13,14α values (difference >1 Hz). The latter indicate that the
H13−C13−C14−H14α dihedral angle related to the puckering
of the A ring is significantly different in 5a−c from that in 1a.
Indeed, in the quantum mechanically optimized structures, this
angle in the 5a−c series is consistently smaller than that in 1a
and this slightly affects the relative orientation of the C13 side
chain. Support for these refined geometries is provided by the
good agreement between the J13,14α values computed using the
Karplus equation and the experimental measurements (Table
S2).
analog 1d with the intact oxetane D ring was also prepared as a
reference compound for the bioassays (Scheme S1).
We also tried to oxidize the 20-OH of 6 before the
elimination of the 5-I to shorten the synthetic steps. However,
after 24 was treated under various elimination conditions, only
an unexpected product 25 was obtained near-quantitatively
(Scheme 6). This observation suggested that an intramolecular
Scheme 6. Synthesis of Unexpected Compound 25
In addition, the divergence between tubulin binding affinity
and cytotoxicity, as that of 2 and 3, still remained for D-seco
taxanes 5a−c. While 5a was about 46 times less active than 1a
in the tubulin binding assay, it exhibited more than 6200 times
less cytotoxicity against A2780 cells. This suggests that the
intact oxetane ring is important for the cytotoxicity of taxanes
as previously thought. In light of this, we then asked whether
we could further potentiate both the binding affinity and the
cytotoxicity through modification of the C4 substituent.
The results of C4 extended aliphatic esters and carbonates
possessing comparable or superior activity in tubulin polymer-
ization and cytotoxicity assays9 demonstrated that extension of
the C4 substituent may lead to improved tubulin binding and
cell killing activities. For this reason, the 4-allyloxy D-seco
taxane 5e was designed for revising the conformation of the
entire molecule and strengthening the hydrophobic interaction
of the C4 substituent within the binding pocket.
reaction precedes the intermolecular reaction in the presence of
a catalytic base or acid. In addition, after the C4 hydroxyl of 24
had been substituted, the eliminated product could be obtained
by LiCl/NaHCO3 methods, but accompanied by a C5
chlorinated byproduct (1:1) (data not shown). This indicated
that reducing the hindrance at C20 could compromise the
partial untoward effect from the C4 substitution. From these
results, the plausible mechanisms for the occurrence of 25 are
proposed (Scheme S2).
The D-seco taxane 5e exhibited a binding affinity similar to
that of 1c and was even 2-fold more cytotoxic against A2780
cells than 1c (Table 1). Therefore, it was much more potent
than its 4-methoxy counterpart 5a in both tubulin binding (12-
fold) and cytotoxicity assay (350-fold). It is also noteworthy
that the resistant/sensitive (R/S) ratio of cytotoxicity for 5e is
lower than that of 1a on both A2780AD/A2780 and HeLa/
βIII/HeLa cell lines, although the absolute potency is still much
weaker than that of 1a (Table S3).
Following a similar approach to synthesize a C,D-spiro
taxane in our lab,7 the allylation of 4-OH could only be realized
in the presence of (PPh3)4Pd and ally-tert-butylcarbonate, since
it was found difficult to occur by using similar conditions as
methylation (Scheme 5). However, from compound 8, the
Contrary to the previous thought that the intact oxetane D
ring is crucial for cytotoxicity,4 we found that 20-hydroxylation
combined with attachment of a hydrophobic chain of proper
length (e.g., allyloxy) to C4 not only enhances the binding
affinity and promotes tubulin polymerization (only 2.4-fold
weaker than 1a) but also increases the cytotoxicity to a great
extent (only 18-fold less cytotoxic than 1a).
Scheme 5. Synthesis of Compound 5e
The NMR data for 5e, when compared to the data for 1c,
showed the same tendency that was observed for 5a−c in the 4-
OMe D-seco series but to a lesser extent (Table S1). In fact, the
value of the H13−C13−C14−H14α dihedral angle for 5e
(141.4°, Table S2) was intermediate between those measured
for 1a (147.5°) and 5a−c (∼135°). However, 5e behaved as 1c
in molecular and cellular assays rather than as 5a−c did.
This contradiction could be interpreted by a molecular
modeling study. Inspection of the most up-to-date molecular
model of the complex of 1a with tubulin10 reveals a very tight
fit between the bulky drug and the taxane-binding site (Figure
1). At the bottom of this large pocket the oxygen of the oxetane
ring acts as a H-bond acceptor for the backbone NH of Thr276.
The 4-OMe D-seco 5a−c derivatives, which differ from 1a and
1b in the puckering of ring C, cannot establish this interaction.
But when the R1 substituent at C20 is a hydroxymethyl (as in
5a), this oxygen can presumably act as a H-bond donor for the
backbone CO of Thr276 instead. This interaction is reinforced
when an allyloxy group is present at C4 (as in 5e) because this
extended group, relative to a methoxy (as in 1c) or an acetoxy
(as in 1a), improves the van der Waals contacts with the phenyl
ring of Phe272. If it is absent, however, the strength of the H-
bond involving the 20-hydroxymethyl is not enough to
reaction occurred exclusively at the 3′-N amide (data not
shown). We also attempted the reaction from the C4 allyl
carbonate to avoid a side reaction, but the in situ conversion
rate from the carbonate to the allyl ether was not satisfactory.
Finally, alkylation of 4-OH only succeeded after the steric
hindrance at the C20 position was reduced by oxidation to an
aldehyde 20 with NMO/TPAP (yield 65%) or 2-iodoxybenzoic
acid (IBX) (yield 91%). Then, the C4 allyloxy analog 21 was
obtained in 95% yield without formation of the C3′-N allyl
byproduct. After the C20 aldehyde group in 21 was reduced by
NaBH4, the freshly generated hydroxyl was concealed by a TES
group. Eventually, the desired product 5e was obtained in 70%
yield over last two steps. In addition, the C4-allyloxy ether
C
DOI: 10.1021/acs.orglett.5b03119
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.or-
glett.5b03119.
Complete experimental procedures, product character-
ization, spectroscopic data for all new compounds, and
other experimental details (PDF)
AUTHOR INFORMATION
Corresponding Author
■
*E-mail: wfang@imm.ac.cn.
Notes
Figure 1. Taxane-binding site in β-tubulin showing bound 1a (C
atoms in yellow) and 5e (C atoms in green).
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
S.R.W. and W.S.F. thank the National Natural Science
Foundation of China (Grant Nos. 30930108 and 81202434)
for financial support. F.G. is grateful to the Spanish Ministerio
compensate for the worse fit and the lack of interaction with
NH of Thr276. Furthermore, if the substituent at C20 lacks the
OH, the binding affinity drops dramatically because neither a
methyl group (as in 5c) nor a methoxy group (as in 5b) can be
accommodated in the polar and relatively small subpocket that
lodges the oxetane of 1a in β-tubulin. The beneficial effect of
the allyloxy group over an acetoxy on C4 is also manifested in
taxane 1d, which displays greater affinity for tubulin and is
slightly more cytotoxic on HeLa cells than 1a (Table 1 and
Table S3).
́
de Economıa y Competitividad (SAF2012-39760-C02-02) and
Comunidad Automoma de Madrid (S2010-BMD-2457). We
́
also thank Dr. Ruth Matesanz (CSIC, Madrid, Spain) for her
work on the biological evaluation of partial compounds during
her visit to IMM.
REFERENCES
(1) Dubois, J.; Thoret, S.; Guer
2000, 41, 3331−3334.
■
́
itte, F.; Guen
́
ard, D. Tetrahedron Lett.
In general, both opening of the oxetane D ring and
shortening of 4-OAc to 4-OMe decreased the interaction
with tubulin and also the cytotoxicity. The biological activity
was almost abolished in 4-OH and -H analogs.11 The
substituents attached to C4, on the other hand, extend deep
into the cavity and are directed toward the floor of the tubulin
binding pocket, resulting in either no contact with the floor or,
if long enough, a hydrophobic interaction that improves
binding provided there no steric clash. There is an exception
to the lack of C4 substituent contact as described above,
namely the encounter between the C4 unit and water
molecules that occupy the binding site in the absence of ligand.
Apart from the above-mentioned enthalpy considerations,
the enhanced bioactivity of 5e can also be explained through an
entropy-driven process. The taxane-tubulin binding pocket is an
expansive cavity capable of binding a wide variety of antitubulin
drugs12 as well as a pool of water molecules in the absence of
ligand. MT binding of such molecules can lead to either
favorable or unfavorable entropy contributions to ΔG, as
demonstrated by Buey et al.13 It can be hypothesized that the
improvement of the binding and cytotoxic action of compound
5e relative to the other D-seco compounds in Table 1 results
from penetration of the allyloxy group into the tubulin binding
pocket leading to the displacement of one or more tightly
bound water molecules. The accompanying favorable ligand
desolvation entropy can be enhanced by installation of C4
moieties that are able to drive out additional water molecules
and, thereby, improve the tubulin binding affinity and cytotoxic
potential of the D-seco class of agents.
(2) Barboni, L.; Giarlo, G.; Ricciutelli, M.; Ballini, R.; Georg, G. I.;
Vander Velde, D. G.; Himes, R. H.; Wang, M. M.; Lakdawala, A.;
Snyder, J. P. Org. Lett. 2004, 6, 461−464.
(3) Loewe, J.; Li, H.; Downing, K. H.; Nogales, E. J. Mol. Biol. 2001,
313, 1045−1057.
(4) Thoret, S.; Guer
8, 2301−2304.
́ ́
itte, F.; Guenard, D.; Dubois, J. Org. Lett. 2006,
(5) Chen, S. H. Tetrahedron Lett. 1996, 37, 3935−3938.
(6) 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.
́
(7) Wang, S. R.; Sanchez-Murica, P. A.; Gago, F.; Fang, W. S. Org.
Biomol. Chem. 2016, DOI: 10.1039/C5OB02131F.
(8) Wessel, H. P.; Viaud, M. C.; Gardon, V. Carbohydr. Res. 1993,
245, 233−244.
(9) (a) Georg, G. I.; Ali, S. M.; Boge, T. C.; Datta, A.; Falborg, L.
Tetrahedron Lett. 1994, 35, 8931−8934. (b) Chordia, M. D.; Yuan, H.;
Jagtap, P. G.; Kadow, J. F.; Long, B. H.; Fairchild, C. R.; Johnston, K.
A.; Kingston, D. G. I. Bioorg. Med. Chem. 2001, 9, 171−178.
(10) Alushin, G. M.; Lander, G. C.; Kellogg, E. H.; Zhang, R.; Baker,
D.; Nogales, E. Cell 2014, 157, 1117−1129.
(11) (a) Neidigh, K. A.; Gharpure, M. M.; Rimoldi, J. M.; Kingston,
D. G. I. Tetrahedron Lett. 1994, 35, 6839−6842. (b) Chordia, M. D.;
Chaudhary, A. G.; Kingston, D. G. I. Tetrahedron Lett. 1994, 35,
6843−6846.
(12) (a) Nettles, J. H.; Li, H.; Cornett, B.; Krahn, J. M.; Snyder, J. P.;
Downing, K. H. Science 2004, 305, 866−869. (b) Kingston, D. G. I.;
Snyder, J. P. Acc. Chem. Res. 2014, 47, 2682−2691. (c) Daly, E. M.;
Taylor, R. E. Curr. Chem. Biol. 2009, 3, 47−59.
(13) Buey, R. M.; Barasoain, I.; Jackson, E.; Meyer, A.; Giannakakou,
P.; Paterson, I.; Mooberry, S.; Andreu, J. M.; Díaz, J. F. Chem. Biol.
2005, 12, 1269−1279.
In conclusion, the present results demonstrate that an intact
D ring in taxanes is not strictly necessary for their interaction
with tubulin and for exerting their cytotoxic effects.
D
DOI: 10.1021/acs.orglett.5b03119
Org. Lett. XXXX, XXX, XXX−XXX