Z. Chen et al. / Bioorg. Med. Chem. Lett. 23 (2013) 6688–6694
6693
Figure 3. The modeled binding mode of benzosuberene 8 (rendered in tubes with orange carbon atoms) is shown with the binding site (Connolly surface in a) and compared
with the co-crystallized ligand DAMA-colchicine (rendered in tubes with purple carbon atoms) in b.5 The modeled binding mode of benzosuberene 8 seems to be buried
deeper with more van der Waals interactions with the b-subunit (colored light blue) and differs in its interactions with the T5 loop (rendered in light pink tube, shown
partially from Thr a179 onwards for clarity with the red dot indicating the truncation point) of the a-subunit when compared with DAMA-colchicine.
ing mode of 19b, the phenolic hydroxyl group accepts a hydrogen
bond from Val 181 of the T5 loop of the -subunit rather than
donating one to the main chain oxygen of Thr 179. There is signif-
a
a
a
icant overlap between the modeled binding modes of 25 and 19a.
Although the hydroxyethyl group at C8 may cause the B-ring to
pucker away from the T7 loop, two putative hydrogen bonds be-
tween the terminal hydroxyl group and the main chain of Asp
b251 and Leu b252 may partially compensate for the loss of any
favorable van der Waals interactions.
Subsequently, we tested whether structural variation on the B-
ring architecture would be tolerated.
Results from modeling studies of the diene analog 36 indicate
that it binds similar to the parent benzosuberene 8 (Fig. 5a) pre-
serving the van der Waals interactions of the trimethoxyphenyl
moiety and the hydrogen bond with Thr a179, but with a more ri-
gid B-ring that contributes favorably to its affinity. On the other
hand, the modeled binding mode of cyclic ether analog 37 (5S,8R
stereoisomer shown in Fig. 5b) diverges from that of its parent
and differs in its interactions with the binding site residues. The
cyclic ether functionality seems to perturb the orientation of the
trimethoxyphenyl moiety as well as the phenolic hydroxyl group
Figure 4. The modeled binding modes of C-8 analogs of benzosuberene, 19a
(rendered in tubes with green carbon atoms), 19b (rendered in tubes with light
green carbon atoms) and 25 (rendered in tubes with lavender carbon atoms) are
shown in the colchicine binding site and compared with the parent benzosuberene
8 (ball-and-stick rendition with orange carbon atoms). While all three C-8 analogs
have a different orientation of the B-ring compared to 8, only 25 seems to form
hydrogen bonds with the main chain of Asp b251 and Leu b252 of the T7 loop of the
b-subunit.
which now accepts a hydrogen bond from Val
of the -subunit rather than donating one to the main chain oxy-
gen of Thr 179. We also modeled the binding mode of the other
a181 of the T5 loop
a
a
stereoisomer of 37 (5R,8S, not shown in Fig. 5b) and observed that
the flipped cyclic ether functionality further perturbs the orienta-
tion of the trimethoxyphenyl moiety while maintaining the hydro-
gen bond with Thr
residues of the T7 loop.
a179 and van der Waals interactions with
the
a
-subunit. The T5 loop interacts differently with colchicine
181 donates a hydrogen
where the main chain nitrogen of Val
a
In summary, we have presented modeling studies that provide a
rationale for the binding mode of benzosuberenes, as represented
bond to the keto group of colchicine. To explore the structure–
activity relationship around the B-ring C8 of benzosuberene 8, sev-
eral analogs were synthesized. Based on the modeling studies,
these analogs would putatively interact with the non-polar resi-
dues like Leu b248 and Leu b252 of the T7 loop of the b-subunit
(Fig. 3). The modeled binding modes of three C8 analogs, namely
19a, 19b and 25 are compared with that of the parent benzosube-
rene 8 in Figure 4. The modeled binding mode of 19a indicates that
the methyl group at C8 may cause the B-ring to pucker away from
the T7 loop which in turn may reduce any favorable van der Waals
interactions. Even though the hydrogen-bond with the T5 loop of
by the parent compound 8, into the colchicine binding site of
a,b
tubulin. We have explored a small set of C-8 susbstituents on the
B-ring which led to analogs with reduced or no potency. We also
described two new benzosuberene analogs with novel structural
features on the B-ring, namely the diene analog 36 and the cyclic
ether analog 37. Both compounds show excellent cytotoxicity
potencies that are comparable to those of colchicine (1) and the
parent benzosuberene 8. Whereas the high potency for 36 could
be supported by modeling studies which predict no significant
departure from the favorable binding mode found for 8, the
potency of the racemic ether analog 37 came as a surprise. The
docking studies for this compound were not supportive of its
excellent potency, an observation that emphasizes a limitation of
the predictability of molecular modeling studies for colchicine
mimetics. In fact, a large set of structurally diverse compounds
are reported to bind into the colchicine binding site with excellent
the a-subunit is preserved, this binding mode of 19a loses affinity
because of the changed orientation of the B-ring and the trime-
thoxyphenyl moiety. The modeled binding mode of 19b indicates
that the ethyl group at C-8 significantly perturbs the orientation
of the entire scaffold of the parent benzosuberene 8. Not only does
the B-ring pucker away from the T7 loop, but also orients the ethyl
group unfavorably towards the solvent. Interestingly, in this bind-
Chen, Zecheng
