Elucidating target specificity of the taccalonolide covalent microtubule stabilizers employing a combinatorial chemical approach

Article abstract of DOI:10.1038/s41467-019-14277-w

The taccalonolide microtubule stabilizers covalently bind β-tubulin and overcome clinically relevant taxane resistance mechanisms. Evaluations of the target specificity and detailed drug–target interactions of taccalonolides, however, have been limited in part by their irreversible target engagement. In this study, we report the synthesis of fluorogenic taccalonolide probes that maintain the native biological properties of the potent taccalonolide, AJ. These carefully optimized, cell-permeable probes outperform commercial taxane-based probes and enable direct visualization of taccalonolides in both live and fixed cells with dramatic microtubule colocalization. The specificity of taccalonolide binding to β-tubulin is demonstrated by immunoblotting, which allows for determination of the relative contribution of key tubulin residues and taccalonolide moieties for drug–target interactions by activity-based protein profiling utilizing site-directed mutagenesis and computational modeling. This combinatorial approach provides a generally applicable strategy for investigating the binding specificity and molecular interactions of covalent binding drugs in a cellular environment.

Full text of DOI:10.1038/s41467-019-14277-w

A
R
T
I
C
L
E
https://doi.org/10.1038/s41467-019-14277-w
OPEN
E
l
u
v
m
u
c
a
i
l
d
e
a
n
n
t
t
a
a
i
n
m
g
t
a
r
l
r
o
g
t
e
u
t
b
e
,
s
p
e
e
c
s
l
i
fi
c
b
p
i
i
t
l
p
r
y
o
f
s
c
&
t
h
e
e
m
t
p
a
c
c
y
a
l
o
g
n
a
o
l
i
d
e
c
c
o
o
i
i
c
a
u
l
t
a
a
i
z
e
r
l
o
i
n
b
i
t
o
r
c
h
m
i
c
a
R
r
o
a
h
1
,
2
,
6
3
,
6
4
3
,
5
L
i
n
D
*
,
S
m
a
n
t
h
a
S
.
Y
e
e
K
a
r
t
h
i
k
a
m
a
c
h
a
n
d
a
n
A
p
r
i
l
L
.
R
i
s
i
n
g
e
r
*
T
h
e
t
a
c
c
a
l
o
n
o
l
i
d
e
m
i
c
r
o
t
u
b
u
l
e
s
ta
b
i
l
i
z
e
r
s
c
o
v
a
l
e
n
t
l
y
b
i
n
d
β
-
t
u
bu
l
i
n
a
n
d
o
v
e
rc
o
m
e
c
l
i
n
i
c
a
l
l
y
r
e
l
ev
a
n
t
t
a
x
a
n
e
r
e
s
i
s
t
a
n
ce
m
e
c
h
a
n
i
s
m
s.
E
v
a
l
u
a
t
i
o
n
s
o
f
t
h
e
t
a
r
g
e
t
s
p
e
c
i
fi
c
i
t
y
a
n
d
d
e
t
a
i
l
e
d
d
r
u
g
–
t
a
r
g
e
t
i
n
t
e
r
a
c
t
i
o
n
s
o
f
t
a
c
c
a
l
o
n
o
l
i
d
e
s
,
h
e
o
w
e
v
e
r
,
h
a
v
e
b
e
e
n
l
i
m
i
t
e
d
i
n
p
a
r
t
b
y
t
h
e
i
r
i
r
r
e
-
e
e
v
e
r
s
i
b
l
e
t
a
r
g
e
t
e
n
g
a
g
e
m
e
n
t
.
I
n
t
h
i
s
s
t
u
d
y
,
w
r
e
p
o
r
t
t
h
e
s
f
y
n
t
h
e
s
i
s
o
f
fl
u
o
r
o
g
e
n
i
c
t
a
c
c
a
l
o
n
o
l
i
d
p
r
o
b
e
s
t
h
a
t
m
a
i
n
t
a
i
n
c
t
h
e
n
a
t
i
v
e
b
i
o
l
o
g
i
c
a
l
p
r
o
p
e
r
t
i
e
s
o
t
h
e
p
o
t
e
n
t
t
a
c
c
a
l
o
n
o
l
i
d
e
,
A
J
.
T
h
e
s
c
a
r
e
f
u
l
l
y
o
p
t
i
m
i
z
e
d
,
e
l
l
-
p
e
r
m
e
a
b
l
e
p
r
o
b
e
s
o
u
t
p
e
r
f
o
r
m
c
o
m
m
e
r
c
i
a
l
t
a
x
a
n
e
-
b
a
s
e
d
p
h
ro
b
e
s
a
n
d
c
e
n
a
b
l
e
d
i
r
e
c
t
v
i
s
u
a
l
i
z
a
t
i
o
n
o
f
t
a
c
c
a
l
o
n
o
l
i
d
e
s
i
n
b
o
t
h
l
i
v
d
n
e
a
n
d
fi
x
e
d
c
e
l
l
s
w
i
t
d
r
a
m
a
t
i
m
i
c
r
o
t
u
b
u
l
e
c
o
l
o
c
a
l
i
z
a
t
i
o
n
.
T
h
e
s
p
e
c
i
fi
c
i
t
y
o
f
t
a
c
ca
l
o
n
o
l
i
e
b
i
n
d
i
n
g
t
o
β
-
t
u
b
u
l
i
n
i
s
d
n
e
m
o
n
-
s
t
r
a
t
e
d
b
y
i
m
m
u
s
n
o
b
l
o
t
t
i
n
g
,
w
h
i
c
h
a
e
l
l
o
w
s
f
o
r
d
e
t
e
r
m
i
n
a
t
i
o
o
f
t
h
e
r
e
l
a
t
i
v
e
c
o
n
t
r
i
b
u
t
i
o
o
f
k
e
y
t
u
b
u
l
i
n
r
e
s
i
d
u
e
a
n
d
t
a
c
c
a
l
o
n
o
l
i
d
m
o
i
e
t
i
e
s
f
o
r
d
r
u
g
–
t
a
r
g
e
t
i
n
t
e
r
a
c
t
i
o
n
s
b
y
a
c
t
.
i
v
i
t
y
-
b
a
s
e
d
p
b
r
o
t
e
i
n
p
r
o fi
l
i
n
g
u
t
i
l
i
z
i
n
g
s
i
t
e
-
d
i
r
e
ct
e
d
m
u
t
a
g
e
n
e
s
i
s
a
n
d
c
o
m
p
u
t
a
t
i
o
n
a
l
m
o
d
e
l
i
n
g
T
h
i
s
c
o
m
-
i
n
a
t
o
r
i
a
l
a
p
p
r
o
a
c
h
p
r
o
v
i
d
e
s
a
g
e
n
e
r
a
l
l
y
a
p
p
l
i
c
a
b
l
e
s
t
r
a
t
e
g
y
f
o
i
r
i
n
v
e
s
t
i
g
a
t
i
n
g
t
h
e
b
i
n
d
i
n
g
s
p
e
c
i
fi
c
i
t
y
a
n
d
m
o
l
e
c
u
l
a
r
i
n
t
e
r
a
c
t
i
o
n
s
o
f
c
o
v
a
l
e
n
t
b
i
n
d
i
n
g
d
r
u
g
s
n
a
c
e
l
l
u
l
a
r
e
n
v
i
r
o
n
m
e
n
t
.
1 Department of Chemistry and Biochemistry, The University of Oklahoma, Norman, OK, USA. 2 Institute for Natural Products Applications and Research
Technologies, The University of Oklahoma, Norman, OK, USA. 3 Department of Pharmacology, The University of Texas Health Science Center, San Antonio,
TX, USA. 4 Department of Medicine, Division of Nephrology, The University of Texas Health Science Center, San Antonio, TX, USA. 5 Mays Cancer Center, The
University of Texas Health Science Center, San Antonio, TX, USA. 6These authors contributed equally: Lin Du, Samantha S. Yee. *email: Lin.Du-1@ou.edu;
risingera@uthscsa.edu
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s
1

A
R
T
I
C
L
E
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
any successful therapeutics, including aspirin, β-lactam covalent binding of the taccalonolides to β-tubulin and evaluate
antibiotics, esomeprazole (Nexium), and clopidogrel key β-tubulin residues and taccalonolide moieties that mediate
(Plavix)1,2 bind covalently to their drug targets. How- taccalonolide-tubulin binding. Flu-tacca-7 (11) represents a class
M
ever, the irreversible nature of their binding prompts safety of irreversible microtubule labeling probes that are superior to
concerns due to potential off-target reactivity and unanticipated commercially available options, providing a valuable tool for
side effects. Therefore, one of the most critical steps in the cellular evaluations of this important drug target.
covalent drug discovery process is the effective evaluation of their
target specificity and assessment of useful derisking strategies3,4
.
The development of modern ‘targeted covalent inhibitors’
(TCIs)5,6 has led to significant progress including the successful
launch of several preclinical and clinical studies for covalent
EGFR inhibitors, such as the FDA approved afatinib (Giltrif) and
osimertinib (Tagrisso), which exhibited promising therapeutic
effects against resistant cancer models expressing EGFR muta-
tions7–9. The systematic studies of TCIs have also revealed that
the safety of covalent drugs needs to be evaluated on a case-by-
case basis and the complexity of the covalent systems often urge
innovative approaches10–12 as necessary complements to con-
ventional preclinical and clinical studies.
The taxane class of microtubule stabilizers is a mainstay in the
clinical treatment of solid tumors even in the era of targeted
therapy and immunotherapy13–16. However, a major limitation of
the taxanes is acquired drug resistance. The taccalonolides are a
class of microtubule stabilizers that covalently bind β-tubulin17,18
and effectively circumvent clinically relevant models of resistance
to taxanes both in vitro and in vivo19–21. Despite their promising
therapeutic potential, the covalent nature of taccalonolide binding
has hampered our ability to perform detailed binding studies
using conventional approaches22,23. Consequently, it urged us to
develop a functional and rigorous activity-based approach to
elucidate the target specificity and drug–target interactions of the
taccalonolides. Here we describe the synthesis and optimization
of a fluorogenic taccalonolide probe, Flu-tacca-7 (11). This stable,
cell-permeable probe is used for activity-based protein profiling
(ABPP) in human cancer cell lines to confirm the specificity of
Results
Optimization of fluorogenic taccalonolide-based probes. While
we were exploring strategies for generating an optimal
taccalonolide-based chemical probe, Wang et al. reported the
crystal structure (PDB ID: 5EZY) of tubulin complexed with
taccalonolide AJ (2) (Fig. 1a, c)18. The authors proposed an
unusual reaction mechanism intending to explain the covalent
bond formation between the 22,23-epoxy moiety of 2 and β-
tubulin D226. However, on account of the existing conflicts in the
literature about the absolute configuration of the 22,23-epoxy
group21,24, we unambiguously verified the 22 R,23 R configura-
tion of 2 by single-crystal X-ray diffraction analysis (Fig. 1b)25.
Therefore, the opening of the 22,23-epoxy group of 2 is likely
facilitated via direct nucleophilic attack by the carboxylate of β-
tubulin D226 (Fig. 1d)26. This epoxide opening mechanism was
supported by covalent docking of 2 into β-tubulin using CovDock
affording a lowest-energy docking model that perfectly matched
the 5EZY crystal structure (RMSD = 0.221, Fig. 1c)27. Analysis of
the docking data also disclosed that several other key β-tubulin
residues (e.g. K19, H229, R278, L217, L219, and T223) were likely
to play important roles in mediating the binding affinity of 2
(Fig. 1e). But more importantly, the careful examination of the
chemical environment in the binding pocket revealed that the C-6
ketone group of 2 was positioned relatively remote from all β-
tubulin residues and was not involved in any inter- and intra-
molecular interactions (Fig. 1e). Thus, the taccalonolide C-6
position was identified as an optimal site for linker/payload
a
b
c
22
H
OAc
H
O
AcO
OAc
O
23
H
O
OH
O
H
O
H
R
OH
6
H
Taccalonolide AF (1): R = OAc
Taccalonolide AJ (2): R = OH
X-ray structure of 2
e
d
O
H
H
H
H
O
H
20
HO
27
O
O
O
O
H
O
O
22
D226
O
23
OH
D226
H
H
O
O
O
O
O
O
O
O
OH
OH
O
O
OH
2
Tubulin-taccalonolide complex
O
H-bond
( 2
Fig. 1 The taccalonolide microtubule stabilizers covalently bind β-tubulin. a
S
t
r
u
c
t
u
r
e
s
o
f
t
a
c
c
a
l
o
n
o
l
i
d
e
s
A
F
(
1
)
a
n
d
A
J
)
s
h
o
w
i
n
g
v
e
r
i
fi
e
d
a
b
s
o
l
u
t
e
c
o
r
o
o
n
fi
g
u
r
c
g
r
e
k
a
t
(
i
o
P
n
D
T
e
o
B
f
I
t
D
p
h
:
r
u
e
5
o
c
2
2
Y
s
r
,
2
)
e
3
-
o
e
f 2
p
o
x
(
t
h
y
m
o
b
m
s
i
e
o
e
t
u
c
y
n
h
u
. b
d
T
h
e
-
O
R
b
T
E
P
d
w
r
a
a
s
w
i
u
n
p
d β
3
g
o
f
t
h
p
u
e
o
l
s
i
n
d
D
g
w
2
a
l
e
i
2
r
|
-
t
6
e
c
h
. e
r
y
a
s
m
t
a
o
h
l
X
-
r
l
e
a
o
y
.
y
f 2
s
t
r
(
u
u
b
l
c
t
u
e
r
H
r
)
e
e
d
s
o
o
f
2
(
d
s
C
C
t
a
u
D
o β
t
C
I
D
u
d
:
b
i
1
9
0
n
t
7
(
h
7
R
e
9
M
b
a
0
S
i
)
. c
D =
T
h
0
a
3
e
p
2
n
a
u
b
l
g
i
s
e
o
t
h
n
f 2
a
e
e
d
c
r
y
d
t h e
a
s
t
b
a
y
l
s
t
u
c
t
u
E
Z
r
e
d
)
t
o
β
t
u
u
l
i
n
s
n
f
e
r
i
m
s
e
d
e
l
u
c
k
e
i
n
-
t
u
t
l
i
.
2
1
)
r
a
t
e
C
d
v
D
o
n
v
. d
m
d
h
e
p
o
d
o
f
r
e
a
.
c
c
T
k
i
o
n
a
n
i
s
m
b
e
t
w
e
r
e
a
n 2
d
a
-
t
u
b
f
i
n
T
e
k
t
u
b
i
n
i
d
u
e
t
h
m
b
e
r
a
e
n
n
d
i
n
g
2
f fi
w
i
s
t
y
i
n
d
c
k
i
o
t
d
h
l
a
s
c
t
r
t
u
e
f c
d
e
r
e
i
d
e
s
w
n
i
t
h
.
i
n
a
i
u
s
o
.
5
Å
r
o
m
2
d
i
s
p
l
a
y
e
d
T
h
e
k
e
y
O
m
o
l
e
c
l
e
i
d
g
i
g
2
n
d
T
2
r
e
i
n
e
s
i
t
2
i
m
p
r
o
e
e
c
u
r
a
c
y
o
o
i
n
g
e
x
p
e
r
i
m
e
t
s
2
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s

N
A
T
U
R
E
C
O
M
M
U
N
I
CA
T
I
O
NS
|
h
t
t
p
s
:
//d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
01
9
-
1
4
2
7
7
-
w
A R T I C L E
OAc
H
O
22
23
AcO
OAc
O
O
OAc
H
O
O
AcO
OAc
O
H
Flu-tacca-2 (4): R =
Flu-tacca-3 (5): R =
, R′ = F, n = 2
, R′ = H, n = 2
O
H
H
OH
O
O
H
H
OH
H
OH
OH
O
O
H
H
OH
H
OH
O
O
6
H
HN
(7): 22,23-ene precursor of 8
N
4
N
N
O
n
NH
O
O
O
O
Flu-tacca-5 (8): R =
Flu-tacca-6 (9): R =
, R′ = H, n = 2
, R′ = H, n = 1
R′
O
O
OR
R′
O
Flu-tacca-1 (3)
O
RO
COOH
OAc
H
O
O
OAc
H
O
AcO
OAc
OAc
O
O
AcO
OAc
O
AcO
OAc
H
H
O
O
H
H
H
H
O
OH
OH
O
H
H
O
H
H
H
OH
OH
OH
O
H
H
OH
OH
O
OH
H
H
OH
NH
HN
O
O
H
O
NH
O
NH
O
O
COOH
O
O
COOH
HO
O
O
O
O
HO
O
O
Flu-tacca-8 (12)
Flu-tacca-7 (11)
(10): 22,23-ene precursor of 11
Flu-tacca-4 (6)
Fig. 2 Structures of the semi-synthetic taccalonolide-based fluorescent probes 3–12.
2 –
T
h
e
t
a
c
c
a
l
o
n
o
l
i
d
e
p
r
o
b
e
s
s
y
n
t
h
e
s
i
z
e
d
a
n
d
e
v
a
l
u
a
t
e
d
i
n
t
h
i
s
s
t
u
d
y
a
r
e
s
h
o
w
n
.
S
y
n
t
h
e
t
i
c
s
c
h
e
m
e
s
a
n
d
m
e
t
h
o
d
s
c
a
n
b
e
f
o
u
n
d
i
n
S
u
p
p
l
e
m
e
n
t
a
r
y
F
i
g
s
.
8
9
6
a
n
d
t
h
e
S
u
p
p
l
e
m
e
n
t
a
r
y
M
e
t
h
o
d
s
,
r
e
s
p
e
c
t
i
v
e
l
y
.
conjugation to generate a stable taccalonolide probe that was convenient approach to convert taccalonolide B (13) to its C-6
likely to maintain the native biological properties of 2. amino analogue 14 through reductive amination (Fig. 3). The
Our strategy to functionally characterize taccalonolide-tubulin employment of the 4 Å molecular sieve as a dewatering agent in
binding using fluorescent taccalonolide probes is based on the the reaction played a vital role in suppressing the formation of the
well-established activity-based protein profiling (ABPP) C-6 hydroxy side product (<5% yield)28. With 14 in hand as a key
approach11, which facilitates determination of drug–target intermediate, we were able to generate a set of stable amide-based
interactions in a cellular context and is particularly suited to fluorescent/fluorogenic probes 4–12 employing varying linker
compounds that covalently bind their targets. Initial attempts to length, fluorescent moieties, and prodrug strategies (Fig. 2). The
generate a stable taccalonolide probe by modification of optimization of the taccalonolide probes was guided by evaluation
taccalonolide C-6 led to the synthesis of Flu-tacca-1 (3) (Fig. 2), of their biological properties and comparison with the untagged
a fluorescent probe that enabled direct visualization of the taccalonolide AJ (2) in a series of cellular and biochemical
taccalonolides in live cancer cells28. However, there were several experiments (Table 1, Figs. 4, 5).
disadvantages of this probe, including the lability of the ester-
The synthesis of the amide-based probes, Flu-tacca-2 (4) and
based linker, the weak micromolar cellular potency, poor Flu-tacca-3 (5), was inspired by the structure of the commercial
fluorescence properties due to the masked phenolic hydroxyl taxane-based probe, Tubulin Tracker Green (Thermo Fisher
group of the fluorescein moiety, and high background fluores- Scientific, Oregon Green™ 488 Taxol, Bis-Acetate). Specifically, a
cence that necessitated removal of excess probe from the media protected fluorescent moiety (Oregon Green 488 for 4 and
prior to imaging. These limitations urged us to generate fluorescein for 5, diacetyl form) was conjugated with 14 via a β-
additional taccalonolide probes that were more suitable for ABPP alanine linker. According to the manufacturer’s description29, the
studies. By means of a survey of various strategies to effectively diacetyl protection on Oregon Green is intended to quench
modify the C-6 position of the taccalonolides, we identified a fluorescence prior to intracellular hydrolysis of the acetyl groups
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s
3

A
R
T
I
C
L
E
N
A
T
U
R
E
C
O
M
M
U
N
ICA
T
I
O
N
S
|
h
t
t
p
s
:
/
/d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
OAc
H
OAc
H
AcO
OAc
AcO
OAc
O
O
DMDO
DCM
acetone
NH4OAc
NaCNBH3
MeOH
H
H
O
O
H
H
OH
10
11
OH
O
H
H
O
H
O
H
OH
OH
4 Å
OH
OH
H
molecular sieves
H
NH2
HATU
DCM
Taccalonolide B (13)
14 (6S : 6R = 3 : 1)
pyridine
O
O
HOOC
HOOC
HOOC
O
O
COOH
MeSO3H
Pivalic anhydride
O
O
O
+
O
O
O
2 equiv
HO
HO
O
O
OH
15
16
Fig. 3 Synthesis of Flu-tacca-7 (11).
T
h
e
s
y
n
t
h
e
t
i
c
s
c
h
e
m
e
f
o
r
t
h
e
g
e
n
e
r
a
t
i
o
n
o
f
t
h
e
m
o
s
t
p
o
t
e
n
t
d
i
p
i
v
a
l
o
y
l
-
pr
o
t
e
c
t
e
d
t
a
c
c
a
l
o
n
o
l
i
d
e
fl
u
o
r
e
s
c
e
n
t
p
r
o
b
e
.
D
e
t
a
i
l
e
d
m
e
t
h
o
d
s
c
a
n
b
e
f
o
u
n
d
i
n
S
u
p
p
l
e
m
e
n
t
a
r
y
M
e
t
h
o
d
s
.
Oregon Green is not an optimal strategy for generating stable,
cell-permeable taccalonolide probes for imaging applications.
As we were exploring efficient separation protocols for the
synthetic isomeric mixture 5(6)-carboxyfluorescein (15) (Fig. 3),
we noticed that the dipivaloyl protection of 15 to give 16 resulted
in a complete quenching of carboxyfluorescein fluorescence in
aqueous solutions and enabled baseline-separation of the two
isomers by HPLC using a preparative C18 column. Intrigued by
the quenched fluorescence of 5-carboxyfluorescein dipivalate
(16), we synthesized the taccalonolide probe Flu-tacca-5 (8), a
dipivaloyl-protected analogue of 5 and 6 (Fig. 2). As expected,
compound 8 exhibited exceptional hydrolytic stability in both
organic and aqueous solutions as well as in the cell culture
medium (Supplementary Fig. 1). Moreover, the dipivaloyl
protection effectively quenched the fluorescence of the extra-
cellular probe in cell culture media enabling comparable
intracellular staining in live cells before and after excess probe
was removed from the medium with virtually no background
fluorescence (Fig. 4a). Further intracellular localization studies
indicated 8 promoted cellular microtubule stabilization and
colocalized with β-tubulin immunofluorescence in both fixed
HCC1937 and HeLa cell lines (Fig. 4b). Despite the improved
visualization of the dipivaloyl probe in both live and fixed cells,
the micromolar antiproliferative potency of 8 (HeLa, GI50:
1.5 µM; SK-OV-3, GI50: 4.5 µM) as compared to the nanomolar
potency of untagged 2 (HeLa, GI50: 8.5 nM; SK-OV-3, GI50:
6.2 nM) prompted us to further optimize the intracellular potency
of dipivaloyl-protected probes.
Inspired by a recent publication showing improved antiproli-
ferative activities can be achieved for taxane-based probes by
replacing a β-alanine linker with a shorter glycine linker31, we
synthesized two dipivaloyl-protected taccalonolide probes,
including Flu-tacca-6 (9) utilizing a glycine linker and Flu-
tacca-7 (11) featuring direct conjugation of the fluorescein moiety
with the taccalonalide skeleton by an amide bond (Figs. 2, 3).
Indeed, a progressive shortening of the linker between the
taccalonolide and the dipivaloyl-protected fluorescein moiety
enhanced the antiproliferative potency of 9 and 11 as compared
to 8 against HeLa and SK-OV-3 cell lines (Table 1). In particular,
the direct drug-fluorophore conjugation in 11 led to GI50 values
of 30–50 nM, a 50-fold improvement in potency as compared to 8
and <10-fold difference as compared to the untagged 2 (Table 1).
Table 1 Concentrations (nM) of taccalonolide probes that
cause a 50% decrease in the proliferation (GI50) of HeLa or
SK-OV-3 cells.
Compound
HeLa
SK-OV-3
2
3
8
.
1
5
0
0
0
.
1
2
6
N
.
D
2
1
.
4
2
6
2
,
0
0
4
5
6
7
8
>
2
,
0
,
0
0
0
0
>
2
0
,
0
0
0
1
2
2
0
1
0
,
7
0
0
0
1
1
1, 8 0 0 2 , 1 0 0
0, 8 0 0 1 , 6 0 0
8
>
,
3
0
0
,
0
6
2
5
0
0
0
0
0
0
0
>
4
9
>
4
2
0
0
,
0
0
0
0
9
a
1
,
4
0
0
0
0
,
5
0
0
9
7
4
6
0
7
2
7
0 6 0
10
11
12
>
2
,
2
0
0
,
8
0
0
0
a
0
3
3
1
,
a
0
0
0
2
0
0
7
,
2
0
4
0
0
a
d
a
t
a
f
r
o
m
4
i
n
d
e
p
e
n
d
e
n
t
e
x
p
e
r
i
m
e
n
t
s
.
G
I
v
a
l
u
e
s
w
e
r
e
o
b
t
a
i
n
ed
f
r
o
m
t
h
r
e
e
i
n
d
e
p
e
n
d
e
n
t
e
x
p
er
i
m
e
n
t
s
(
u
n
l
e
s
s
o
t
h
e
r
w
i
s
e
n
o
t
e
d
)
e
c
a
e
c
h
5
0
p
e
r
f
o
r
m
e
d
i
n
t
r
i
p
l
i
c
a
t
e
a
n
d
p
re
s
e
n
t
e
d
a
s
m
e
a
n
S
E
M
.
S
o
u
r
c
e
d
a
t
a
a
r
e
p
r
o
v
id
e
d
a
s
a
S
o
u
r
D a t a fil e .
by esterases to decrease background fluorescence of any
unincorporated probe. However, the Oregon Green probe 4 was
rapidly hydrolyzed even in methanol solutions likely due to the
relatively low pKa of the Oregon Green moiety (Oregon Green,
pKa 4.8 and fluorescein, pKa 6.5, unprotected form)30 and did not
have antiproliferative potency up to a concentration of 20 µM. In
contrast, the fluorescein probe 5 showed improved stability in
organic solutions but was readily hydrolyzed to yield the
deprotected form 6 in a 50% methanol/PBS solution and in the
RPMI 1640 medium (Supplementary Fig. 1a–c, g, and h). Both 5
and 6 were over 125-fold less potent than the untagged 2 against
HeLa cervical, SK-OV-3 ovarian, and two triple-negative breast
(HCC1806 and HCC1937) human cancer cell lines as determined
by the concentration that caused a 50% inhibition of proliferation
as compared to vehicle treated controls (GI50) (Table 1 and
Supplementary Fig. 2). Furthermore, both 5 and 6 showed strong
extracellular fluorescence in the cell culture medium that
remained at a low level even after excess probe was removed
from the medium prior to imaging (Fig. 4a). Together these
results indicated that the diacetyl protection of fluorescein or
4
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s

N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
h
t
t
p
s
:
//d
o
i
.
o
r
g
/
1
0
.
1
0
38
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
A
R
T
I
C
L
E
a
b
Before wash
After wash
β-Tubulin
8
6
5
8
HCC1937
HeLa
c
e
OAc
H
O
AcO
OAc
O
Vehicle
8
9
11
H
O
H
OH
O
H
H
OH
OH
O
H
HN
O
O
O
O
O
Flu-tacca-7 (11)
O
O
OAc
AcO
O
O
d
Vehicle
8
9
11
OAc
H
H
O
H
OH
O
H
H
OH
OH
H
NH
O
COO
O
O
O
Flu-tacca-8 (12)
Fig. 4 Optimization of taccalonolide-fluorescein probes. a
5 µ
h β
V
i
f
s
o
u
r
u
a
e
o
a
r
r
e
l
i
(
r
s
t
o
z
l
e
i
o
w
t
5
a
e
s
n
t
f
c
g
i
s
d
i
t
o
)
e
l
n
o
n
o
r
f
u
n
p
r
a
r
o
(
g
s
t
r
e
e
c
h
t
e
d
( 6
)
d
d
i
a
l
F
,
a
c
m
C
t
e
t
c
y
o
1
l
-
n
p
t
3
r
a
7
o
i
a
a
t
n
e
i
n
d
o
t
c
n
d
d
n
h
)
t
g
e
d
p
e
( 5
)
e
c
,
a
w
n
a
d
s
p
r
i
v
e
e
a
m
r
h
l
o
o
y
v
l
e
-
p
d
M
n
d
. e
g
r
o
a
p
o
t
n
e
d
c
t
r
b
e
e
d
p
a
n
e
-
e
e
( 8
)
e
i
n
u
s
-
p
d
r
b
o
y
n
r
d
u
c
b
e
f
o
s
i
s
6
n
h
l
i
v
e
S
m
K
e
-
d
O
V
m
-
3
. b
c
e
C
l
l
o
s
-
5
h
a
f
z
t
a
e
t
r
i
t
h
e
o
a
f 8
d
d
(
i
g
t
i
r
o
e
n
e
o
f
M
p
r
u
o
b
b
u
e
e
n
i
i
t
m
h
e
m
r
b
e
a
f
t
e
i
g
t
)
m
e
i
u
r
o
b
l
a
c
r
e
i
u
l
o
ca
l
i
o
n
n
)
w
i
t
-
t
l
i
u
n
ofl
c
e
(
o
r
s
.
w
n
a
n
e
h
i
)
(
s
b
e
a
i
n
fi
x
e
H
h
g
C
)
a
9
e
q
H
d
a
e fi
,
r
La
e
l
l
s
a
f
t
3
i
m
1
0 µ
r
o
e
a
d
d
t
i
o
f
c
r
l
h
a
a
n
. c
t
c
,
a
h
c
n
d
P
n
a
i
c
g
v
e
e
a
n
l
o
y
a
l
-
t
p
r
n
e
o
t
o
t
e
f
l
c
t
e
d
t
a
c
c
a
o
e
l
r
o
0
r
n
o
.
l
i
d
e
- fl
5 µM d
e fl
u
o
a
o
r
n
r
e
d
e
s
c
i
c
e
n
e
i
c
n
u
n
p
b
r
a
o
b
e
s
f
p
w
i
t
5
e
h
d
e
c
re
i
n
k
e
n
n
a
r
g
g
i
z
l
e
f
t
t
o
r
g
w
r
e
e
t
o
S
K
-
O
V
s
t
c
s
-
s
u
c
o
d
m
o
w
t
a
v
o
a
n
A
e
r
s
i
c
a
c
s
h fl
e
e
e
l
s
c
o
t
r
i
o
5
µ
M
c
t
e
d
o
o
r
h
p
r
i
o
p
t
m
a
a
a
m
g
i
T
e
a
r
m
i
p
e
h
o
s
i
m
e
c
g
c
a
u
e
a
t
i
s
-
a
s
i
t
i
f
( 11
r
n
d
e
p
d
r
o
c
e
n
g
n
i
t
i
w
o
t
l
r
e
f
o
r
e
e
u
i
l
m
t
o
d
i
r
c
y
c
o
e
m
p
a
r
e
l
a
t
i
v
u
s
c
e
p
r
e
r
t
i
s
(
t
o
o
r
)
t
l
o
t
h
g
r
e
t
fi
e
d
l
u
i
m
a
c
d
o
l
(
b
o
t
o
r
)
n
a
w
a
o
a
r
-
o
8
g
i
a
b
e
l
i
n
g
s
y
s
t
e
m
f
o
r
c
e
l
l
u
l
a
r
t
u
b
u
l
i
n
e
m
p
l
o
y
i
n
g
t
h
e
d
i
p
i
v
a
l
o
y
l
-
p
r
c
e
0 µ
c
c
l
o
o
l
l
i
d
b
.
e
S
u
-
t
a
7
w
o
h
i
c
h
e
i
s
l
e
a
v
S
e
o
o
e
n
r
F
l
u
t
a
c
p
o
c
e
l
l
u
l
a
r
e
n
t
r
y
w
h
e
r
i
t
c
a
n
b
i
n
d
c
e
l
l
u
l
a
r
m
i
c
r
o
t
u
b
l
e
s
.
S
c
a
l
e
b
a
s =
f
o
r
l
i
m
g
o
r
c
e
a
e
p
v
i
d
d
a
a
u
r
c
e
D
a
t
a fi
.
Furthermore, the potent probe 11 had significantly improved
To determine whether the differences in cellular potency
intracellular fluorescence brightness as compared to 8 and 9 when among the taccalonolide probes correlated with target engage-
used at equimolar concentrations and imaged under identical ment, we evaluated the potency and efficacy of the unprotected
acquisition and visualization conditions (Fig. 4c, d and Supple- probes 6 and 12 and dipivaloyl-protected probes 8 and 11 as
mentary Fig. 3). Together, our data demonstrate that Flu-tacca-7 compared to 2 in a biochemical tubulin polymerization assay. The
(11) represents a cell-permeable, fluorogenic probe that combines untagged taccalonolide AJ (2) promoted the polymerization of
the potent antiproliferative activities of taccalonolide AJ with purified tubulin (20 µM) in a concentration-dependent manner
excellent fluorescence properties, including the complete quench- over a range of 5–20 µM (Fig. 5a), as previously described17.
ing of fluorescence until the dipivaloyl moieties are cleaved, likely Unexpectedly, the most potent taccalonolide probe in cellular
by intracellular esterases (Fig. 4e), for the imaging of tubulin in assays, 11, only slightly enhanced microtubule polymerization at
both live and fixed cells.
the highest tested concentration (20 µM) (Fig. 5a). In contrast, 12,
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
ic
a
t
i
o
n
s
5

A
R
T
I
C
L
E
N A T U R E C O M MU N ICA T IO N S | h t t p s : / /d o i .o r g / 1 0 .1 0 3 8 / s4 1 4 6 7 - 0 1 9 -14 2 7 7 - w
a
c
d
0.10
0.05
0.00
Vehicle
5 µM 2
10 µM 2
20 µM 2
20 µM 8
20 µM 11
0
20
40
60
Time (min)
b
e
f
0.10
0.05
0.00
Vehicle
1 µM 2
1 µM 6
1 µM 11
1 µM 12
0
20
40
60
Time (min)
Fig. 5 Taccalonolide probes engage additional β-tubulin contacts. a, b
E
f
f
n
i
e
d
n
o
r
e
c
e
t
t
n
r
s
o
f
5–
f
r
e
2
o
a
h
d
e
0 μ
n
c
e
s
t
a
M
f
l
c
c
)
a
e
l
o
n
e
o
e
e
t
o
c
y
l
t
l
i
d
s
p
r
e
o
r
o
A
f 2
J
(
2
)
t
g
(
a
n
u
r
a
Z
s
d
l
o
n
t
a
c
p
s
)
c
o
i
o
a
l
o
m
n
e
o
r
l
i
i
e
p
d
z
d
u
e
a
)
r
o β
w
-
fl
u
n
n
e
o
w
d 11
)
u
d
r
e
i
s
c
e
v
r
m
) c
g
i
e
n
h
e
o
p
r
l
)
o
e
b
i
d
o
h
a
e
n
s
b
( 6, 8, 11
,
12)
o
n
t
h
e
p
o
l
y
m
e
r
i
z
a
t
i
o
n
o
f
p
u
r
ifie
d
p
o
r
c
i
n
e
t
u
b
u
l
i
n
.
a
C
o
n
c
e
n
t
r
a
t
i
o
n
-
d
e
p
e
t
(
f
f
o
n
u
b
i
n
l
y
t
i
o
t
h
(
i
c
l
a
c
k
a
n
d
2
i
n
s
h
a
d
e
s
o
f
b
l
u
e
f
r
o
m
l
i
g
h
t
e
s
t
t
o
d
a
r
k
e
s
t
a
s
c
o
n
c
e
n
t
r
a
t
i
o
n
i
n
c
r
e
a
s
e
s
.
T
h
e
e
p
o
c
o
H
d
u
c
t
i
t
t
n
o
p
o
u
d
i
v
a
l
o
t
e
c
t
i
n
g
u
g
Y
p
e
( 2
y
n 8
r 12
b
(
r
a
g
e
n
i
m
i
n
i
s
h
t
h
i
s
a
c
l
t
u
u
i
v
e
b
e
i
)
t
y
o
e
r 11
n
v
e
n
(
c
.
a
t
e
e
e
d
u
q
n
u
)
i
m
o
t
c
a
l
h
a
e
a
a
r
c
r
n
o
a
o
e
n
b
c
e
n
y
t
r
t
r
a
o
o
e
t
p
b
i
o
o
e
n
l
s
a
s
y
w
i
t
r
h
t
u
p
b
u
u
r
l
i
n
(
2
0
μ
M
)
.
b
T
h
C
e
e
d
r
t
e
t
e
d
o
c
a
n
b
o
l
i
d
p
b
e
s 6
t
o
r
(
d
e
p
l
a
b
r
u
m
e
l
r
e
p
t
a
s
e
n
t
t
h
a
n
2
(
b
g
r
i
n
i
i
l
i
t
m
e
i
z
e
i fi
( 12) e
e
d
t
u
b
u
l
i
n
. c–f
(11) f
a fi
o
m
p
a
t
i
s
n
f
o
p
a
l
i
s
h
d
b
s
t
r
u
c
u
r
e
5
E
o
a
u
s
n
y
t
-
t
i
n
w
i
t
m
o
d
e
l
o
f
β
-
t
u
c
l
i
d
e
o
k
e
t
o
t
a
c
l
o
l
i
d
e
p
(6) d
s
,
,
a
n
d
.
T
h
s
e
l
e
c
d
-
b
n
s
a
l
r
i
d
g
e
s
a
r
e
d
i
p
l
a
e
d
l
l
o
a
n
a
e
n
t
a
d
h
e
d
l
i
n
e
s
,
r
e
s
p
t
i
v
l
y
S
o
r
c
e
d
t
a
r
p
r
o
v
i
d
d
a
S
o
u
r
c
e
D
a
t
l
e
.
the deprotected analogue of 11, was more potent than 11 or 2 in binding model for 11 clearly showed that the two dipivaloyl
this biochemical assay (Fig. 5b). A more dramatic case was protecting groups hampered the ability of the fluorescein moiety
observed for the dipivaloyl-protected 8, which did not promote to be positioned into this additional binding pocket and the
polymerization even at 20 µM, as compared to its deprotected taccalonolide core structure was only correctly positioned in 2 of
analogue 6, which promoted robust polymerization at 1 µM that the 10 low-energy poses (Fig. 5f). In the other 8 low-energy poses
was equivalent to the effect of 20 µM of 2 (Fig. 5a, b). Due to the for 11, the taccalonolide core structures were “forced” into the
covalent nature of the interaction of the taccalonolide probes with binding pocket but the structures were significantly rotated
tubulin, we were able to determine that the time course of binding implying those binding models were not reliable and suggesting
of 12 to purified tubulin correlated with microtubule polymer- that 11 was poorly bound to β-tubulin. Similarly, the taccalono-
ization at both 1 and 20 µM (Supplementary Fig. 4). This lide core structure of 8 failed to correctly fit into the taccalonolide
confirms that the lag time associated with tubulin polymerization binding pocket in any of the 10 low-energy β-tubulin binding
seen with the taccalonolides (which is distinct from the models consistent with the inability of 8 to enhance tubulin
immediate polymerization observed with the taxanes) is asso- polymerization in biochemical assays as compared to vehicle
ciated with slow initial binding of the drug that then increases controls (Fig. 5a). Thus, the analysis of taccalonolide probe
rapidly concomitant with microtubule nucleation32.
binding based on covalent docking data was qualitatively
Covalent docking using CovDock was employed to rationalize consistent with the biochemical tubulin polymerization assay.
the increased potency of the taccalonolide probes 6 and 12 for Although more comprehensive computational and experimental
biochemical tubulin polymerization as compared to the unmo- analyses would be required to provide a more complete
dified taccalonolide 2. The top 10 low-energy poses (ligand- understanding of the mode of action for the taccalonolide probes,
receptor binding models) were generated for 6, 8, 11 and 12, the current binding analysis provides guidance for future efforts
respectively, that were docked into the optimized 5EZY structure. to generate taccalonolide analogues with improved binding
The representative lowest-energy pose obtained from each affinity to tubulin by engaging an additional binding pocket
docking experiment was displayed (Fig. 5c–f). The taccalonolide adjacent to the M-loop.
core structure of 12 (Fig. 5e) was correctly positioned into the
The finding that the dipivaloyl-protected fluorogenic probes 11
taccalonolide binding pocket (based on the 5EZY crystal structure and 8 are unable to directly interact with and polymerize tubulin
of 2, Fig. 5c) for each of the top 10 low-energy β-tubulin binding or exhibit fluorescence in the medium, but effectively stabilize
models. Interestingly, the fluorescein moiety of 12 occupied an microtubules and exhibit fluorescence in cellular assays suggests
adjacent binding pocket on β-tubulin close to the M-loop that upon cellular entry these probes are activated via the
affording additional interactions with β-tubulin residues via hydrolysis of the dipivaloyl groups, likely by cellular esterases,
hydrophobic interactions, H-bonds, and/or salt bridges (Fig. 5e). (Fig. 4e). Indeed, when HCC1937 cells were treated with the
A similar binding mode was predicted for 6 (Fig. 5d), suggesting dipivaloyl protected 22,23-ene analogue (10) of Flu-tacca-7 (11)
that the enhanced ability of 6 and 12 to promote microtubule to prevent irreversible binding to its target, its deprotected,
stabilization in biochemical assays could be attributed to fluorescent form 28 was able to be detected from cellular lysates
improved binding affinity to β-tubulin afforded by these by LCMS analysis (Supplementary Fig. 5). Thus, the dipivaloyl
additional contacts. In contrast, the lowest-energy β-tubulin protective groups serve two distinct roles for the taccalonolide
6
N
A
T
U
R
E
C
O
M
M
UN
I
C
A
TI
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s

N
A
T
U
R
E
C
O
M
M
U
N
I
CA
T
I
O
N
S
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
38
/
s
4
1
4
6
7
-
01
9
-
1
4
2
7
7
-
w
A R T I C L E
probes: effectively quenching extracellular fluorescence and epoxide-dependent cellular microtubule stabilization and colo-
retaining the taccalonolides in a binding-deficient form prior to calization was also observed by comparing the localization of
entry into cells where fluorescence and tubulin binding can 22,23-epoxide and 22,23-ene probes in both live and fixed cells
both occur.
(Supplementary Fig. 6a, b). On account of the covalent nature of
the taccalonolide interaction with tubulin17, the protein binding
profiles of 10 and 11 were also evaluated by immunoblotting
Target specificity of the taccalonolides. The generation of under denaturing and reducing conditions that would disrupt
the potent fluorogenic taccalonolide probe Flu-tacca-7 (11) pro- non-covalent interactions using an anti-fluorescein antibody.
vides the opportunity to evaluate the covalent binding specificity While the cells treated with 10 or 11 showed equivalent expres-
of the taccalonolides in cell-based assays and fully define the sion of β-tubulin (Fig. 6b), a single 50 kDa fluorescein-containing
key structural elements of both taccalonolides and β-tubulin band was readily identified in cells treated with 11, but not 10
that are essential for drug binding. The 22,23-epoxide moiety has (Fig. 6c). The results demonstrated that the 22,23-epoxy group
been suggested to be critical for the antiproliferative and micro- was essential for the taccalonolide probes to efficiently form a
tubule stabilizing effects of the taccalonolides18,24,33. Thus, we covalent bond with its major cellular target of β-tubulin in the
compared the antiproliferative activities, cellular localization, and cellular environment, demonstrating the specificity of this inter-
proteome reactivity profiles of Flu-tacca-7 (11) and its 22,23-ene action. Similar results were obtained from the comparison of Flu-
analogue 10. The single replacement of 22,23-epoxide in 11 by tacca-5 (8) and its 22,23-ene analogue 7 (Supplementary Fig. 6c,
22,23-ene in 10 completely abrogated the antiproliferative activ- d). Together, these data strongly demonstrate that the 22,23-
ities in all four human cancer cell lines (i.e. HeLa, SK-OV-3, epoxy moiety is critical for the covalent reaction of the taccalo-
HCC1806, and HCC1937) up to 20 μM (Table 1 and Supple- nolides with β-tubulin.
mentary Fig. 2). Additionally, 11 promoted a concentration-
Inspired by the ability to detect the cellular binding of Flu-tacca-
dependent (0.05–5 μM) increase in the polymerization of cellular 7 (11) to endogenous β-tubulin by immunoblot, we engineered a
tubulin, as detected by immunofluorescence and confocal ima- system to evaluate the relative contribution of individual β-tubulin
ging (Fig. 6a, red), in line with its antiproliferative potency in amino acid residues to taccalonolide binding by performing site-
HCC1937 cells (Supplementary Fig. 2). The intrinsic fluorescence directed mutagenesis on an ectopically expressed β-tubulin
of 11 (Fig. 6a, green) colocalized completely with the cellular construct tagged with GFP at the C-terminus to distinguish it by
microtubules. In contrast, the 22,23-ene analogue 10 failed to size on an immunoblot. In order to test the feasibility of our
either promote cellular tubulin polymerization (Fig. 6a, red) or approach, we first mutagenized the D226 residue of β-tubulin to
colocalize with microtubules (Fig. 6a, green) at 5 μM. This 22,23- either an asparagine or alanine (Supplementary Table 9). HeLa
a
0.05 μM 11
0.5 μM 11
5 μM 11
5 μM 10
Vehicle
b
c
d
e
–
10 11
–
10 11
50
50
80
GFP-tubulin
Tubulin
80
50
50
11
+
+
+
–
11
+
+
+
–
β-Tubulin
Fluorescein
β-Tubulin
Fluorescein
Fig. 6 The taccalonolide epoxide and β-tubulin D226 are critical for covalent binding. a
H
C
o
C
1
9
s
n
3
(
d
7
g
s
c
e
e
b
s
l
n
j
t
l
s
)
c
w
w
e
r
h β
e
t
-
o
r
t
i
h
e
u
m
a
b
m
t
e
u
u
d
d
w
i
t
h
m
o
0
u
.
n
n
u
0
ofl
g
t
5
–
5
µ
M
s
a
t
t
a
n
c
c
a
c
e
n
t
a
l
(
-β
o
r
n
e
-
o
d
l
)
i
d
w
u
w
a
e
p
r
o
b
e
s
w
i
t
h
( 11)
o
r
w
i
t
h
o
u
t
(
1
0
)
t
h
e
2
2
,
2
3
-
e
p
o
x
i
d
e
f
o
r
2
4
h
.
C
o
l
o
c
a
l
i
z
a
t
i
o
n
o
f
t
a
c
c
a
l
o
n
o
li
d
e
p
r
b
e
r
e
i
t
l
i
n
i
m
u
o
r
e
g
s
c
e
a
s
e
a
v
n
a
l
b
u
o
a
d
t
e
y b
f
d
b
o
r
y
r
8
c
a
h
o
n
.
n
a
C
f
o
n
l
c
a
-fl
l
i
u
s
m
o
a
g
s
s
i
n
c
w
g
e
e
.
b
,
c
H
C
C
1
9
3
7
c
e
l
H
m
l
s
e
m
t
L
r
a
u
e
a
c
o
t
e
e
d
s
w
w
i
e
n
t
r
h
e
g
5 µM 10
o
r
1
1
f
o
h
n
r
G
6
F
h
P
w
e
g
r
g
g d
e
e
l
y
s
e
d
a
u
n
e
t
e
d
s
e
t
w
s
n
o
b
l
d
t
t
i
u
s
i
n
n
t
i
t
u
b
l
i
n
h
d
t
i
t
i
r
e
i
n
a
n
t
ib
o
d
y c. d, e
l
l
t
r
a
n
s
f
e
c
t
e
d
w
i
t
-
t
a
d
T
U
h
B
B
1
c
o
r
u
ct
i
t
i
n
i
c
a
t
e
o
m
a
t
i
o
n
h
e
n
(
r
e
a
k
t
e
d
a
i
n
t
1
µ
M
1
1
o
e
l
l
y
a
t
e
r
e
h
a
r
v
e
s
t
e
d
f
o
r
i
n
b
l
o
tt
i
.
β
-
t
u
b
u
l
i
n
i
m
m
u
o
b
l
o
t
t
i
n
s
o
w
e
d
t
h
e
e
x
p
r
s
i
o
n
o
f
e
n
d
g
e
n
o
u
s
t
u
b
u
l
i
n
5
0
D
)
t
h
e
G
F
P
-
t
u
b
u
l
i
n
c
o
n
s
t
r
u
c
t
s
(
7
7
k
D
a
)
. e
A
n
a
n
t
i
-
fl
u
o
r
e
s
c
e
in
a
n
t
i
b
o
d
y
w
a
s
u
s
e
d
t
o
d
e
t
e
c
t 11
b
o
u
n
d
t
o
e
n
d
o
g
e
n
o
u
s
l
y
e
x
p
r
e
s
s
e
d β
-
t
u
b
u
l
i
n
(
l
o
w
e
r
b
a
n
d,
5
0
k
D
a
)
a
n
d
t
h
e
G
F
P
-
t
u
b
u
l
in
c
o
n
s
t
r
u
c
t
s
(
u
p
p
e
r
b
a
n
d
,
7
7
k
D
a
)
.
S
c
a
l
e
b
a
r
s =
1
0
µ
m
f
o
r
a
l
l
i
m
a
ge
s
.
S
o
u
r
c
e
d
a
t
a
a
r
e
p
r
o
v
i
d
ed
a
s
a
S
o
u
r
c
e
D
a
t
a fi
l
e
.
N
A
T
U
R
E
C
OM
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
ww
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
ic
a
t
i
o
n
s
7

A
R
T
I
C
L
E
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
a
SER 277
LEU 217
LEU 219
O
O
O
GLN 282
O
ASP 226
O
O
11
O
TYR 283
22
C
A
OH
D
E
B
GLY 225
LEU 230
O
15
F
O
25
26
HO
OH
N
HO
H
O
–
O
HIE 229
GLY 370
LYS 19
O
LYS 372
ARG 369
O
O
H-bond
Salt bridge
–
O
b
c
GFP-tubulin
GFP-tubulin
β-Tubulin (50)
β-Tubulin (50)
d
e
2.0
1.5
1.0
0.5
0.0
2.0
1.5
1.0
0.5
0.0
*
**
***
****
****
Fig. 7 Systematic evaluation of β-tubulin residues that mediate taccalonolide binding. a
T
Å
h
e
o
k
m
e
y
t
t
h
u
e
b
p
u
r
l
o
i
n
b
r
e
e
a
s
r
i
e
d
u
d
e
s
p
t
l
h
a
a
y
t
e
m
e
n
d
t
i
a
h
t
e
e
t
h
e
b
i
a
n
p
d
h
i
n
a
g
n
a
d
f fi
h
n
y
i
d
t
y
o
p
f 12
i
i
n
c
t
he
d
o
c
k
i
n
g
m
o
d
e
l
s
t
r
u
c
t
u
r
e
g
e
n
e
r
a
t
e
d
b
y
C
o
v
D
o
c
k
.
T
h
e
r
e
s
i
d
u
e
s
w
i
t
h
i
n
t
h
e
r
a
d
i
u
s
o
f
3
.
5
f
r
i
s
d
i
l
e
f
t
g
r
r
o
h
o
b
i
m
d
t
e
n
t
u
t
e
c
e
t
e
r
a
c
e
a
t
t 11
x
c
t
i
n
o
s
n
s
a
e
r
n
e
d
e
a
p
t
e
h
i
c
e
d
t
d
e
w
o
d
b
y
g
r
e
e
n
d
a
s
r
l
s
h
e
8
d
h
l
.
o
t
i
n
P
e
s
i
b
b
p
e
n
w
n
e
a
p
n
t
h
e
d
e
a
,
b
e
2
r
i
g
d
h
c
t
g
r
a
p
w
n
D
h
e
d
a
. b, c
H
e
L
a
-
d
a
n
t
o
c
d
e
s
u
n
l
l
u
l
s
b
w
e
r
t
n
u
m
p
o
e
e
s
b
t
d
r
t
u
l
a
o
c
n
s
f
e
m
(
o
t
e
c
t
u
p
e
n
p
e
w
e
t
d
o
e
r
i
r
h
w
i
o
t
h
G
F
P
. b
7
k
W
-
t
a
A
D
g
n
g
e
a
)
d
n
T
U
B
u
B
1
e
u
c
s
o
c
n
f
o
a
o
n
e
i
s
t
r
a
m
e
a
T
u
n
c
t
u
b
t
i
n
i
a
k
c
s
b
o
n
v
e
t
w
i
t
y
h
w
i
n
a
g
f 11
d
s
i
c
u
h
a
s
o
t
t
e
e
d
w
d
i
o
t
h
t
r
e
i
t
h
1 µM 11
s β
f
o
r
o
-
t
r
t
e
e
l
l
s
r
e
l
y
s
e
d
j
e
c
o
t
r
e
i
m
b
l
t
t
i
n
g
t
i
- fl
. c β
. d, e
o
r
i
n
o
d
t
o
d
e
h
a
b
o
i
u
n
d
o
y
e
t
o
n
e
r
g
F
e
i
e
n
o
u
-
t
u
b
u
n
u
i
n
(
l
w
s
e
(
r
r
u
n
5
0
k
,
D
7
a
)
a
k
e
i
t
h
e
G
F
P
n
u
u
b
i
n
c
t
r
t
s
(
d
t
u
w
o
x
r
b
a
n
d
p
n
d
,
7
0
(
O
k
D
T
e
a
-
R
a
N
t
u
b
t
l
i
m
b
l
o
t
t
i
n
s
e
p
c
r
e
o
e
d
s
s
o
n
f
e
t
x
G
s
e
P
-
t
u
b
u
l
u
o
l
i
n
c
o
t
a
a
r
u
c
t d
e
r
a
n
d
o
A
c
7
y
)
a
n
d
e
g
e
o
u
s
u
a
n
i
n
l
b
a
n
t
e
,
5
a
)
a
i
o
o
h
A
n
t
h
d
i
n
g
o
e
o
h
e
t
d
e
p
p
i
c
a
n
c
l
l
p
e
p
e
s
r
s
d
t
n
s
u
b
li
n
h
i
m
t
n
n
n
o
T
b
d
o
g
n
u
,
h
s
l
x
p
r
e
s
s
e
d
t
b
u
l
i
n e
m n =
ifi
n
o
l
i
z
e
a
e
n
r
l
d
y
e
)
-
u
.
w
D
a
y
t
t
a
r
e
s
w
n
t
n
d
i
s
u
p
e
r
a
g
p
d
S E M f or
-
n =
w
3
i
n
e
d
n
t
e
x
m
e
t
s
a
t
s
e
r
t
h
W
e
a
t
d
e
D
e
2
6
w
o
h
n
c
h
a
r
e
f
r
o
5
c
i
n
d
e
m
d
p
p
i
m
t
s
.
n
a
A
O
d
V
i
y
’
s
o
s
t
g
h
b
o
u
c
t
w
e
a
s
t
s
e
r
e
u
s
t
o
a
l
c
u
l
a
t
t
a
t
i
t
i
c
g
n
i
fi
c
a
c
e
n
e
a
c
d
i
t
i
o
n
.
S
i
g
n
c
a
n
e
a
s
c
e
d
t
o
e
R
2
7
8
A
m
t
a
n
t
h
a
t
d
m
a
b
i
n
i
n
t
n
o
t
u
s
e
d
i
n
d
a
t
a
n
o
r
m
a
l
i
z
a
t
io
n
i
s
s
h
o
wn
:
*
p
<
0
.
0
5
,
*
*
p
<
0
.
01
,
*
*
*
p
<
0
.0
0
1
,
*
*
*
*
p
<
0
.
0
0
0
1
.
S
o
u
r
c
e
d
a
t
a
a
r
e
p
r
o
v
i
d
e
d
a
s
a
S
o
u
r
c
e
D
a
t
a
fi
l
e
.
cells that expressed wild type or mutant GFP-tubulin constructs the prediction of key interacting residues based on covalent
were treated with 11 at 1 µM for 8 h followed by immunoblotting docking to the 5EZY crystallographic structure qualitatively
using anti-β-tubulin and anti-fluorescein antibodies. In the β- matched the molecular dynamics simulation results of 2 using the
tubulin immunoblot, bands were detected for each of the GFP- cryo-EM structure of a mammalian microtubule26. Thus, besides
tagged forms of β-tubulin (wild type, D226N, or D226A) at 77 kDa D226, 6 β-tubulin residues (i.e., K19, H229, R278, L217, L219,
that were distinct from endogenous β-tubulin (50 kDa) (Fig. 6d). A and T223) were selected for mutagenesis with similar procedures,
fluorescein immunoblot of these same samples demonstrated that as described above (Supplementary Table 9) followed by
11 was able to interact with the ectopically expressed wild type immunoblotting to determine their impact on taccalonolide
form of GFP-tubulin, but not either D226 mutant, although it binding. It is important to note that although some of these
interacted with the endogenously expressed tubulin for all cases as mutants were expressed at lower levels than wild type GFP-
an internal control (Fig. 6e). These results confirm the critical role tubulin (Fig. 7b), they were each able to be incorporated into
of the 22,23-epoxide and β-tubulin D226 in taccalonolide-tubulin microtubules that were effectively stabilized by the addition of
binding18.
untagged taccalonolide AJ as determined by visualization of the
Encouraged by the results of the D226 mutagenesis, additional GFP-tag (Supplementary Fig. 7). The binding ratio of the probe
β-tubulin mutants were constructed in a similar fashion. The 11 to each GFP-tubulin mutant or endogenous β-tubulin was
analysis of the 5EZY crystal structure18 and the covalent docking quantifed by a ratio of fluorescein signal (Fig. 7b) compared to
models of both 2 (Fig. 1e) and 12 (Fig. 7a) revealed a potential for the β-tubulin signal (Fig. 7c) for each band. The binding ratio for
interaction of the taccalonolide skeleton with 6 residues via H- each mutant was normalized to that of wild type β-tubulin to
bonds and/or salt bridges and hydrophobic interactions (i.e. determine the relative significance of each tested β-tubulin
H229, R278, K19, Q282, T223, and K372) and 9 additional β- residue for probe binding to β-tubulin (Fig. 7d).
tubulin residues via only hydrophobic interactions (i.e. L217,
For all of the samples, the probe bound to the endogenous β-
R369, L219, L230, L371, Y283, G225, G370, and S277). Overall, tubulin at a similar level (Fig. 7e) indicating the mutagenic
8
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s

N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
NS
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
10
.
1
0
38
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
A R T I C L E
manipulations did not affect the extent of the intrinsic binding of β-tubulin residues (e.g. K19 and L219) are critical for the correct
the taccalonolide to β-tubulin. Further analysis of the relative positioning of the taccalonolide core structure into its binding
binding ratios for GFP-tubulin (Fig. 7d) clearly showed that the pocket to facilitate the covalent reaction between D226 and the
mutation of different β-tubulin residues distinctly affected the 22,23-epoxide. It is also reasonable to hypothesize that other β-
binding of the probe to β-tubulin. The extent by which different tubulin residues, including L217, H229, and R278, that interact
β-tubulin residues affected taccalonolide binding correlated with with the taccalonolides at sites relatively remote from the site of
the type of interaction (Fig. 7a and Supplementary Fig. 8), as well the covalent interaction (Fig. 7a) are less important for covalent
as the distance of the residue to the covalent binding site (C-22 of binding but may play an important role in mediating the post-
11) (Supplementary Table 11). Specifically, the β-tubulin residues reaction stability of the taccalonolide-microtubule complex. Thus,
K19, H229, and R278 were predicted to form both H-bonds and these findings that empirically establish the critical β-tubulin
hydrophobic interactions with specific moieties on the probe residues and the potential pharmacophore of taccalonolides may
(Fig. 7a), which were progressively distanced from the site of provide important guidance for the rational design and genera-
covalent binding (C-22 on probe ring E) (Supplementary Fig. 8a tion of ‘drug-like’ taccalonolide analogues via strategies such as
and Supplementary Table 11). The K19 side chain strongly semi-synthesis and structural simplification34.
interacted with several moieties on ring F (i.e. 25-OH, 26-CO,
and 28-Me) which were all close to the covalent binding site
(shortest distance, 3.3 Å from C-26 to C-22) (Fig. 7a). Accord-
ingly, the K19A mutation almost completely abrogated the probe
binding to a similar extent as the D226 mutations (Figs. 6e, 7b, d).
In contrast, the H229 side chain mainly interacted with the
moieties on rings B-D which were relatively remote from the
covalent binding site (shortest distance, 15-OH, 5.2 Å to C-22)
(Fig. 7a). Thus, the H229A mutation only moderately inhibited
binding (Fig. 7b, d). This observation was consistent with
previous SAR studies demonstrating that 25-OH esterification
of the 22,23-epoxy taccalonolides dramatically suppressed anti-
proliferative potency while modification of 15-OH only slightly
affected potency21. Although R278 was predicted to interact with
both the taccalonolide core structure (e.g. 11-Me on ring B and
11-Ac on ring C) and the fluorescein moiety via hydrophobic
interactions, two H-bonds, and a salt bridge (Fig. 7a), the R278A
mutation did not affect probe binding most likely due to the
remote location of the interacting sites from the covalent binding
site (e.g., 11-Ac, 8.8 Å to C-22). A similar trend was observed for
L217 and L219, which were predicted to interact with the
taccalonolide core structure via hydrophobic interactions (Sup-
plementary Fig. 8c, d). Although both L217 and L219 strongly
interacted with 11-Ac and 12-Ac on taccalonolide ring C,
only L219 showed hydrophobic interactions with 21-Me that
collocated with the binding site on ring E. Therefore, the L219A
mutation significantly suppressed the probe binding, while the
L217A mutation only exhibited a weak inhibitory effect (Fig. 7a,
d). An exception was observed for T223, which was predicted to
interact with 26-CO on ring F through a H-bond-connected H2O
bridge in the 5EZY crystal structure (Figs. 1e and 7a). The T223
hydroxyl was also predicted to play an important role in fixing
the carboxylate of D226 to facilitate the covalent reaction with
the 22,23-epoxide on the basis of the molecular dynamics
simulation of 226. Interestingly, the T223A mutation only had a
moderate effect on binding (Fig. 7a, d) despite the predicted
importance of this residue for taccalonolide binding. To gain
additional insight into the impact of the T223A and
H229A mutants on the rate of taccalonolide binding, we
evaluated the binding of the probe 11 at 1, 2, 4, and 8 h. The
extent of binding of 11 to endogenous tubulin was minimal at 1 h
with increased binding observed at 2–4 h regardless of what form
of tubulin (mutant or wild type) was ectopically expressed
(Supplementary Fig. 9). Binding to ectopic wild type tubulin was
detectable at 4–8 h with the T223A and H229A mutants showing
delays in the rate and extent of binding with a more pronounced
effect for the H229A mutant, consistent with Fig. 7 (Supplemen-
tary Fig. 9).
Flu-tacca-7 as a superior tubulin probe. The Flu-tacca probes
represent a class of irreversible, fluorogenic microtubule probes
that can be utilized for cellular microtubule imaging and binding
studies under conditions that have conventionally been unfa-
vorable for the use of non-covalent probes. To demonstrate the
utility of our cell-permeable probes for cellular imaging applica-
tions, we compared the cellular microtubule staining activities of
Flu-tacca-7 (11) to Tubulin Tracker Green (Thermo Fisher Sci-
entific, Oregon Green™ 488 Taxol, Bis-Acetate) and the far-red
probe siR-Tubulin (Cytoskeleton), two commercial taxane-based
probes commonly used for microtubule imaging in live cells.
Tubulin Tracker Green was not amenable to visualization prior to
washing excess probe from the media (Fig. 8a) and pluronic F-
127 was necessary to facilitate drug loading into cells and decrease
background fluorescence (Supplementary Fig. 10). In contrast,
Flu-tacca-7 was effectively visualized without washing or the use
of additional reagents to facilitate cell loading providing com-
parable convenience to the far-red probe siR-Tubulin (Fig. 8a).
However, the intracellular staining of Flu-tacca-7 was superior to
both Tubulin Tracker Green and siR-Tubulin when microtubules
were depolymerized under chilled conditions or in cells expres-
sing drug efflux transporters (Fig. 8). Brief chilling was sufficient
to destabilize microtubules in the presence of the commercial
taxane probes, resulting in loss of probe signal, which could only
be detected slightly over background when images were hyper-
contrasted (Supplementary Fig. 10). In contrast, visualization of
the taccalonolide probe 11 was retained at a similar level before
and after chilling due to the covalent and irreversible nature of its
binding to β-tubulin (Fig. 8a). The tubulin labeling efficacy of 11,
Tubulin Tracker Green, and siR-Tubulin were also compared in
SK-OV-3-MDR1-M6/6 ovarian cancer cells, which express 1000-
fold increased levels of the P-glycoprotein drug efflux pump
MDR-1 as compared to the parental SK-OV-3 cell line (Fig. 8b
and Supplementary Fig. 11). While 11 was effectively visualized in
SK-OV-3-MDR-1-M6/6 cells, cellular staining of either Tubulin
Tracker Green or siR-Tubulin was barely detectable above
background in these multi-drug resistant cells (Fig. 8b). This
observation is consistent with the fact that siR-Tubulin is
recommended to be used in combination with verapamil to block
probe efflux as both it and Tubulin Tracker Green are P-
glycoprotein substrates. Additionally, 11 retained potency, effi-
cacy, and microtubule binding in a βIII-tubulin expressing HeLa
cell line (Supplementary Fig. 12), representing another clinically
relevant form of taxane drug resistance. Therefore, our Flu-tacca
irreversible microtubule probes are superior to commercial
taxane-based probes, combining a high degree of specificity,
brightness, and convenience with their ability to be used
under conventionally unfavorable conditions (i.e., chilling and
Together, the analysis above suggests that the structures of the
taccalonolide rings E and F and their direct interactions with
N
A
T
U
R
E
C
OM
M
U
N
I
C
A
T
I
ON
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
re
c
o
m
m
u
n
i
c
a
t
i
o
n
s
9

A
R
T
I
C
L
E
N A T U R E C O M MU N ICA T IO N S | h t t p s : / /d o i .o r g / 1 0 .1 0 3 8 / s4 1 4 6 7 - 0 1 9 -14 2 7 7 - w
a
b
SK-OV-3
After wash
SK-OV-3-MDR-1-M6/6
Before wash
Chilled
Brightfield
Before wash
After wash
Brightfield
11
PTX-OG
siR-Tubulin
Fig. 8 Taccalonolide probes compared to commercial taxane-based probes. a
S
K
-
O
V
-
3
c
h
b
u
o
t
m
a
o
n
m
o
)
v
a
r
e
d
f
d
r
i
a
5
n
h
c
a
a
n
s
n
d
w
0
m
e
c
i
e
m
e
r
c
e
e
c
l
l
d
s
w
e
f
r
o
a
e
r
t −
h
d
a
t
e
r
t
e
a
a
0
b
r
u
t
e
d
d
C
f
c
e
w
m
i
t
e
r
5
h
d
2
0
.
5 µ
M
o
f 11
,
T
u
bu
l
i
n
T
r
a
c
k
e
r
G
r
e
e
n
(
PT
X
-
O
G
;
T
h
e
r
m
o
F
i
s
h
e
r
;
m
i
d
d
l
e
)
,
o
r
s
i
R
-
T
u
b
u
l
i
n
(
C
y
t
o
s
k
e
l
e
t
o
n
,
I
n
;
t
t
f
o
a
g
b
e
r
e
te
i
a
w
a
s
r
e
m
o
v
e
e
d
d
(
b
h
7
e
i
°
a
f
l
C
g
o
r
d
e
w
a
s
h
K
)
,
-
a
O
f
V
t
e
-
a
r
3
t
-
e
h
M
r
e
D
w
e
t
r
R
e
1
s
a
a
t
e
d
6
g
g
m
e
h
h
c
d
u
e
o
i
a
m
s
u
w
a
e
n
e
n
r
e
o
i
r
e
r
a
e
p
l
a
n
e
c
c
e
c
c
d
a
q
b
y
e
f
r
r
e
s
h
m
e
e
x
d
p
v
a
i
r
u
e
m
(
a
g
z
0 µ
f
t
e
P
r
-
o
m
w
a
c
c
o
s
h
)
p
d
a
,
r
i
l
o
r
a
f
t
e
r
h
e
t
r
e
a
t
d
c
e
l
l
r
e
M
h
i
l
l
e
d
2
o
o
o
°
e
a
c
f
o
0
m
i
n
a
nd fi
x
(
c
l
e
)
. b
S
e
s
-
M
/
6
v
a
i
a
n
c
c
e
l
l
s
s
s
i
n
g
l
y
o
o
t
e
i
n
w
w
e
r
e
t
r
u
u
e
a
te
w
r
t
i
a
a
t
h
.
5
µ
o
f
b
e
a
c
p
r
f
o
r
h
o
a n d i m ag e d a t
b
3
b
e
f
o
r
e
a
n
d
f
t
a
h
r
i
n
.
T
a
m
m
r
g
a
a
h fi
u
i
s
i
t
i
o
S
n
c
a
n
d
i
s
u
a
l
i
a
t
i
n
o
n
t
i
o
n
s
g
e
.
r
e
s
e
d
e
o
l
l
i
r
a
p
g
r
e
s
o
t
a
i
n
s
e
e
r
h
D
p
a
r
e
l
.
B
r
i
g
h
t fie
l
d
i
m
e
s
a
r
h
o
w
n
i
n
t
h
f
r
i
h
t
l
m
f
o
g
u
r
e
.
a
l
e
b
r
s
=
5
f
r
l
i
m
a
e
s
S
o
r
c
a
a
o
v
i
d
e
d
a
S
t
a fi
e
.
drug-resistant cells) without the need for additional pharmaco- and antibody-drug conjugates) and paves the way for further
logical manipulations.
development of the taccalonolides as microtubule stabilizers for
cancer therapy. A further evaluation of the critical taccalonolide
moieties and β-tubulin residues that mediate binding and biolo-
gical activity will continue to provide mechanistic insight to
optimize our models of taccalonolide binding and design more
optimal semi-synthetic and possibly fully-synthetic taccalonolide-
like compounds that could have efficacy in drug resistant models
due to their irreversible target binding. Additionally, the fluoro-
genic taccalonolide-based probes represent highly specific, irre-
versible tubulin-labeling probes that have superior utilities as
compared to commercial taxane-based probes.
Discussion
While three distinct classes of covalent microtubule stabilizing
agents (i.e., the taccalonolides, cyclostreptin, and dactylolide/
zampanolide) have been evaluated for their potential as cancer
therapeutics that circumvent taxane associated drug resistance in
biochemical and cellular assays32, the taccalonolides are the only
class that has demonstrated in vivo efficacy in both drug sensitive
and resistant tumor models17,20,21,25. Here we describe the
development of a combinatorial chemical proteomics approach
that enabled confirmation of the target specificity of taccalonolide
binding to β-tubulin in cellular assays and identification of key
structural features of both the taccalonolides and β-tubulin that
are critical for drug–target binding. Our combinatorial strategy to
empirically establish the key taccalonolide binding residues in
cell-based assays should be broadly useful for detailed ligand-
protein binding studies and structure-based optimization of quite
a number of existing drugs and/or drug candidates that covalently
modify their targets.
Methods
Synthesis of taccalonolide probes. The detailed procedures for the synthesis of
taccalonolide probes 3–12 are described in Supplementary Methods and reaction
schemes provided in Supplementary Figs. 82–96.
Computational modeling. The computational modeling experiments were con-
ducted using the Schrödinger Small-Molecule Drug Discovery Suite (2018-4). The
crystal structure 5EZY was downloaded from PDB and optimized using the Protein
Preparation Wizard following the standard protocol36. Briefly, the structure was
first preprocessed by assigning bond orders, adding hydrogen atoms, creating zero-
bond orders to metals, and creating disulfide bonds between two sulfur atoms
within 3.2 Å from each other. Prime was used to predict and fill in the missing side
chains. Water molecules beyond 5 Å from het groups were removed. To further
While the immunofluorescence and immunoblotting with the
Flu-tacca probes clearly demonstrate that the predominant cel-
lular target of the taccalonolides is indeed a covalent interaction
with β-tubulin, we cannot rule out the possibility of other minor optimize the model, the H-bond assignment was optimized by sampling water
orientations and using PROPKA to assign protonation states of side chains at pH
7.0. All Asp, Glu, Arg, and Lys residues were left in their charged state and the
proper His tautomer was also manually selected to maximize hydrogen bonding.
covalent or non-covalent targets. One important consideration is
the finding that taccalonolides lacking the critical covalent
binding moiety, the 22,23-epoxide, show no antiproliferative,
cytotoxic, or microtubule bundling activities in cells and do not
directly interact with tubulin in biochemical assays. The lack of
any detectable target engagement or bioactivity of probes lacking
this epoxide in the current study lends further support to the
specificity of this interaction.
An unanticipated finding from the generation of the Flu-tacca
probes in this study is the identification of a strategy to improve
the binding affinity and microtubule stabilizing potency of the
taccalonolides by targeting a binding pocket nearby the taccalo-
nolide binding site that engages the M-loop of β-tubulin, which is
integral in pharmacological microtubule stabilization18,35. These
findings provide an opportunity to develop additional ‘drug-like’
taccalonolide analogues with sub-nanomolar cellular potency,
Next, a brief relaxation was performed on the structure. This is a two-part pro-
cedure that consists of optimizing hydroxyl and thiol torsions in the first stage
followed by an all-atom constrained minimization in the second stage to relieve
clashes. The minimization was terminated when the RMSD reached a maximum
value of 0.30 Å. The optimized protein structure was simplified by only retaining
chain B comprising the taccalonolide AJ-β-tubulin complex and removing the
other 5 chains. All water molecules were removed except for the one that formed
H-bonds between β-tubulin T223 and the 26-carbonyl group of taccalonolide AJ.
The ligand structures were optimized using the Ligand Preparation Wizard
(LigPrep)36. Briefly, energy minimization of the ligands was conducted using the
OPLS3 force field. The ionization states were generated at pH 7.0 using Epik and
the dominating tautomer of each ligand was retained for docking experiments.
Further covalent docking experiments were performed using CovDock27. In the
CovDock wizard, D226 was selected as the reactive residue. The docking box was
centered on the coordinates X 3.2/Y −63.5/Z 22.6 in the length of 20 Å. The
covalent reaction was defined as an epoxide opening reaction that was constrained
to take place only on C-22 of the ligands. The ‘thorough’ docking mode was used
which is desired for targeted drug delivery strategies (i.e., peptide- for the optimization of poses. The cutoff of 2.5 kcal/mol was applied for retaining
10
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
w
w
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s

N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
38
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
A R T I C L E
poses for further optimization in each cycle. The top 10 low-energy poses were
Tubulin polymerization assay. Biochemical tubulin polymerization assays were
performed using purified porcine brain tubulin (Cytoskeleton). In individual wells
of a 96-well plate, 1 µL of each ×100 drug stock was incubated with 20 µM porcine
brain tubulin in GPEM glycerol buffer (1 mM GTP, 10% glycerol, 80 mM PIPES
pH 6.9, 2 mM MgCl2 and 0.5 mM EGTA) in a final volume of 100 µL. Pure porcine
tubulin was prepared on ice at 4 °C to inhibit tubulin polymerization until the assay
was initiated, while the plate reader was pre-warmed to 37 °C. Tubulin poly-
merization was measured every minute for an hour by light scattering at 340 nm in
a Spectramax plate reader using SoftMax software (Molecular Devices). Light
scattering was normalized to the initial measurement for each well. For the probe-
tubulin binding assay, samples were prepared on ice in tubes instead of a 96 well
plate, and moved to a 37 °C heat block to initiate binding and polymerization. The
time zero (0’) sample consisted of tubulin polymerization buffer prior to addition
of taccalonolide probe. At each designated time point, 2 µL of the sample was
added to 50 µL NuPAGE sample buffer with 20% β-Mercaptoethanol and 10% of
the resulting sample was subjected to PAGE and immunoblotted for β-tubulin at
1:1000 (abcam, ab6046) or fluorescein at 1:500 (abcam, ab19491) with IRDye 680
or 800 goat anti-rabbit secondary antibodies at 1:10,000 (LI-COR Biosciences) and
imaged on an Odyssey FC (LI-COR Biosciences).
generated and retained for each docking experiment. All the 10 docking models
were visually checked for the binding interactions of the taccalonolide core
structure to filter out the inappropriate binding models with the taccalonolide core
structures that were significantly rotated or positioned outside the binding pocket.
The lowest-energy pose showing correct spatial arrangement of the taccalonolide
core structure was selected for analysis of the ligand-protein binding modes.
Cell lines. HCC1806 (CRL-2335) and HCC1937 (CRL-2336) human triple-
negative breast cancer cells, HeLa (CCL-2) cervical cancer cells and SK-OV-3
(HTB-77) ovarian cancer cells were obtained from ATCC (Manassas, VA) and
validated by STR profiling (Genetica). SK-OV-3 cells stably overexpressing Pgp by
adenoviral-mediated expression of MDR1 were obtained from Dr. Susan Kane and
subcloned by limiting dilution to isolate the single-cell clones utilized in these
studies as SK-OV-3-MDR-1-6/620. A single-cell clone from transfection of HeLa
cells with βIII-tubulin, designated wild type βIII, was constructed and obtained
from Dr. Richard Ludueña20. HCC1806 and HCC1937 cells were cultured in RPMI
1640 media (Corning) with 10% FBS (Cellgro) and 50 µg/mL gentamicin (Gibco).
HeLa, βIII-tubulin expressing HeLa, SK-OV-3 and SK-OV-3/MDR-1-6/6 cells were
grown in BME media with Earle’s salts (Gibco) with 10% FBS, 1× final 1% Glu-
taMax™ Supplement (Gibco), and 50 µg/mL gentamicin. The use of HeLa cells
allows for the direct comparison of the in vitro potency as compared to other
compounds of this class, which have been predominantly reported in this line and
compound potencies are consistent among three additional cancer cell lines. HeLa
cells were also used for ectopic expression of tubulin mutants due to the high
degree of transfectability of this cell line. Cells were tested for mycoplasma con-
tamination using the Mycoplasma Detection Kit-Quick Test (Cat: B39032, Lot:
JW004).
Site-directed mutagenesis. The QuikChange II XL Site-Directed Mutagenesis Kit
(Agilent Technologies) was used according to the manufacturer’s directions with
any changes noted. The template for mutagenesis was the human TUBB1 ORF
mammalian expression plasmid, C-GFPSpark tag from Sino Biological Inc.
(HG11626-ACG) using the primers listed in Supplementary Table 9. After Dpn I
digestion, amplification products were stored at 4 °C until transformation into
DH10B or XL10-Gold competent cells. DNA constructs were isolated using the
QIAGEN Plasmid Midi Kit and Thermo Scientific GeneJet Plasmid Mini Kit. DNA
concentrations were measured using a NanoDrop 2000 (Thermo Fisher Scientific).
All constructs were sequenced using GENEWIZ and sequences verified using
SnapGene.
Antiproliferative assay. The sulforhodamine B (SRB) assay was utilized to
examine the antiproliferative and cytotoxic effects of the compounds37,38
.
Activity-based protein profiling and immunoblotting. For cellular binding stu-
dies, HeLa cells were seeded to 80–90% confluence in 6-well dishes and transiently
transfected with the wild type or mutant tubulin constructs using Lipofectamine
3000 Transfection Reagent (Thermo Fisher Scientific) for 16-18 h. Media with the
lipofectamine reagent were removed from the wells and replaced with fresh BME
media for approximately 24 h prior to drug treatment. Cells were treated with 1 µM
11, 1 µM taccalonolide AJ or EtOH vehicle for 1–8 h, respectively. Cells were
collected by scraping with a cell lifter and lysed with cell extraction buffer (Invi-
trogen) supplemented with protease inhibitor cocktail (Sigma-Aldrich), 50 mM
NaF, 200 μM Na3VO4 (Sigma-Aldrich), and 1 mM phenylmethylsulfonyl fluoride
(PMSF) (Sigma-Aldrich). Protein concentration was determined by a Coomassie
Plus assay kit (Thermo Scientific), equal amounts of protein resolved by
SDS–PAGE on NuPage Bolt 10% Bis-Tris gels (Life Technologies), and transferred
to Immobilion-FL PVDF membranes (Millipore). Membranes were blocked in
Odyssey blocking buffer (LI-COR Biosciences, Lincoln, NE, USA) and probed with
anti-fluorescein at 1:500 (abcam, ab19491) or β-tubulin 1:1000 (abcam, ab6046)
with IRDye 680 or 800 goat anti-rabbit secondary antibodies at 1:10,000 (LI-COR
Biosciences, T8660) and imaged on an Odyssey FC (LI-COR Biosciences). βIII-
tubulin was detected using a monoclonal antibody produced in mouse (1:400)
(Sigma-Aldrich) clone SDL.3D10, ascites fluid.
Revert total protein staining was utilized to demonstrate relative equal total
protein for each lysate (LI-COR Biosciences). The relative binding ratio of mutants
was calculated as: (fluorescein signalmutant/tubulin signalmutant)/(fluorescein
signalwildtype/tubulinwildtype) and expressed as percent of the wildtype signal for each
independent experiment. For imaging studies, wild type or mutant TUBB1-GFP
constructs were transfected into HeLa cells (40,000 cells/well in a 96-well plate)
using Lipofectamine 3000 for 16–18 h prior to washing and replacing with fresh
media. After 7 h recovery, cells were imaged before and after treatment with vehicle
or 100 nM taccalonolide AJ for 22 h using the Operetta. Uncropped blots can be
found in the source data file.
Approximately 2000 cells per well (for SK-OV-3, HeLa and βIII-tubulin expressing
HeLa) or 4000 cells per well (for HCC1806 and HCC1937) were seeded in 96-well
plates. For each biological replicate, cells were treated in triplicate with each
concentration of compound or vehicle control for 48 h in a final volume of 200 µL.
The plates were fixed with 10% trichloroacetic acid for protein precipitation of
adherent cells and then washed with distilled water. In total 100 µL of SRB dye,
which binds protein stoichiometrically, was added and then unbound dye removed
with 1% acetic acid followed by the addition of 200 µL 10 mM Tris to to solubilize
the dye, which was quantified by absorbance at 560 nm. The percent growth of
treated cells relative to the density at the time of drug addition was calculated as
compared to vehicle treated cells. Concentration-response curves were generated
by non-linear regression analysis using Prism software 7.04 (GraphPad) and the
GI50 of each compound was calculated and defined as the concentration that
caused a 50% decrease in cellular proliferation in the 48 h of drug incubation in
comparison to vehicle control from 3 independent experiments.
Live cell fluorescence imaging and immunofluorescence. HCC1937, SK-OV-3,
and SK-OV-3/MDR-1-6/6 cells were plated in PerkinElmer cell carrier imaging 96-
well plates at a density of 8000–10,000 cells/well. HeLa and HeLa βIII-tubulin
overexpressing human cervical cancer cells were plated in PerkinElmer cell carrier
imaging 96-well plates at a density of 4000 cells/well. Cells were treated with vehicle
control or compounds at the indicated final concentration for each individual
experiment. Tubulin Tracker Green and siR-Tubulin stock solutions were prepared
in anhydrous DMSO (Sigma Aldrich) at concentrations of 2 mM or 1 mM,
respectively. Pluronic® F-127 (Invitrogen) was added at a 1:1 ratio from a 20%
(w/v) DMSO stock solution where indicated. For live cell imaging, cells were
imaged 5 h after treatment with compound and then washed with fresh media or
Hank’s Balanced Salt Solution (HBSS) (Sigma-Aldrich) supplemented with 2 mM
CaCl2 and 0.8 mM MgSO4 and imaged on the Operetta high content imager using
Harmony software (PerkinElmer). HeLa and HeLa βIII-tubulin overexpressing
human cervical cancer cells were treated with vehicle (ethanol), 0.5 µM or 5 µM of
probes respectively for 5 h treatment in HBSS. Wells were washed prior to fixing
with methanol. Images were taken with the Operetta at ×20. For colocalization
experiments, cells were fixed with methanol after treatment and subjected to
immunofluorescence for β-tubulin at 1:1000 (Sigma T-4026) with goat anti-mouse
IgG (H + L) cross-absorbed secondary antibody, Texas Red-X at 1:200 (Invitrogen
T-862), while the fluorescein-tagged taccalonolide was directly detected. Probe
treated SK-OV-3 cells were imaged in medium prior to wash or in PBS after
washing or chilling at −20 °C for 20 min and fixed with methanol. For confocal
imaging, HCC1937 cells were treated with 0.05–5 µM taccalonolide probes for
6–24 h on glass coverslips in a 6-well plate and fixed with methanol prior to
β-tubulin immunofluorescence at 1:1000 (Sigma T-4026) using goat anti-mouse
IgG (H + L) cross-absorbed secondary antibody, Texas Red-X at 1:200 (Invitrogen
T-862). Confocal images were acquired using a SP8 Leica DMi8 microscope using a
×63 oil objective. All images are representative of the phenotypes observed from
examining multiple fields from at least 2 independent experiments.
Cellular biotransformation assays. HCC1937 cells were grown to 90% confluence
then treated with vehicle (ethanol), or 1 µM 10 for 8 h in a total volume of 5 mL.
The media was harvested while the cell pellet was lysed by dounce homogenization
in a hypotonic buffer (1 mM EGTA and 1 mM MgSO4, pH 7) after 2 washes with
1× PBS. The cell lysates were extracted by ethyl acetate. The organic layers were
dried down and re-dissolved in MeOH for LCMS analysis.
qRT-PCR. RNA was isolated from SK-OV-3 and SK-OV-3/MDR-1-6/6 ovarian
cancer cells by Trizol and chloroform extraction. The RNA pellet was resuspended
in nuclease-free water and quantified using a Nanodrop 2000. RNA was converted
to cDNA with iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad) and
qRT–PCR was completed using iTaq Universal SYBr Green Supermix (Bio-Rad).
Human Pgp primers (Sigma-Aldrich) were generated based on previous reports39
as Pgp1: 5′- AAAGCGACTGAATGTTCAGTGG-3′ and Pgp2: 5′- AATAGATG
CCTTTCTGTGCCAG-3′ and specificity was confirmed using NCBI Primer-
BLAST. Human GAPDH primers: 5′- GCAAATTCCATGGCACCGT-3′ and
N
A
T
U
R
E
C
OM
M
U
N
I
C
A
T
I
ON
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
ww
.n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
re
c
o
m
m
u
n
i
c
a
t
i
o
n
s
11

A
R
T
I
C
L
E
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
5′- TCGCCCCACTTGATTTTGG-3′. Relative hPgp mRNA transcript levels were
evaluated and presented from two biologically independent experiments, each
performed in duplicate.
18. Wang, Y. et al. Mechanism of microtubule stabilization by taccalonolide AJ.
Nat. Commun. 8, 15787 (2017).
19. Tinley, T. L. et al. Taccalonolides E and A: Plant-derived steroids with
microtubule-stabilizing activity. Cancer Res. 63, 3211–3220 (2003).
20. Risinger, A. L. et al. The taccalonolides: microtubule stabilizers that
circumvent clinically relevant taxane resistance mechanisms. Cancer Res. 68,
8881–8888 (2008).
21. Ola, A. R. B. et al. Taccalonolide microtubule stabilizers generated using
semisynthesis define the effects of mono acyloxy moieties at C-7 or C-15 and
disubstitutions at C-7 and C-25. J. Nat. Prod. 81, 579–593 (2018).
22. de Jong, L. A. A., Uges, D. R. A., Franke, J. P. & Bischoff, R. Receptor-ligand
binding assays: technologies and applications. J. Chromatogr. B 829, 1–25
(2005).
Statistics. For binding studies, a one-way ANOVA with Tukey’s multiple com-
parison post-hoc test and adjustment for multiple comparisons were used to
determine statistical significance between each condition and significance of
mutants as compared to wild type depicted in the figure. Exact n values and
P values are in Supplementary Table 10. For qRT-PCR comparing human Pgp
(hPgp) expression in SK-OV-3 and SK-OV-3/MDR-1-6/6 cell lines a one-tailed
t-test was performed to give a P value of 0.0307.
Reporting summary. Further information on research design is available in
the Nature Research Reporting Summary linked to this article.
23. Pollard, T. D. A guide to simple and informative binding assays. Mol. Biol. Cell
21, 4061–4067 (2010).
24. Li, J. et al. Potent taccalonolides, AF and AJ, inform significant structure-
activity relationships and tubulin as the binding site of these microtubule
stabilizers. J. Am. Chem. Soc. 133, 19064–19067 (2011).
25. Risinger, A. L. et al. Pharmacokinetic analysis and in vivo antitumor efficacy
of taccalonolides AF and AJ. J. Nat. Prod. 80, 409–414 (2017).
26. Sanchez-Murcia, P. A., Mills, A., Cortes-Cabrera, A. & Gago, F. Unravelling
the covalent binding of zampanolide and taccalonolide AJ to a minimalist
representation of a human microtubule. J. Comput. Aided. Mol. Des. 33,
627–644 (2019).
27. Zhu, K. et al. Docking covalent inhibitors: a parameter free approach to pose
prediction and scoring. J. Chem. Inf. Model. 54, 1932–1940 (2014).
28. Du, L. et al. Identification of C-6 as a new site for linker conjugation to the
taccalonolide microtubule stabilizers. J. Nat. Prod. 82, 583–588 (2019).
29. ThermoFisher. www.thermofisher.com/order/catalog/product/T34075?
SID=srch-srp-T34075.
Data availability
The source data underlying Table 1, Figs. 4–8, and Supplementary Figs. 2–4, 6, 7, 9–12
are provided as a Source Data file. Chemical synthesis procedures and NMR data on all
compounds can be found in Supplementary Figs. 13–96. The x-ray crystallographic
coordinates for structures reported in this study have been deposited at the Cambridge
Crystallographic Data Centre (CDCC) under deposition number 1907790. These data
can be obtained free of charge from the Cambridge Crystallographic Data Centre via
www.ccdc.cam.ac.uk/data_request/cif. The data underlying the findings of this study can
be obtained from the corresponding authors upon reasonable request.
Received: 26 July 2019; Accepted: 19 December 2019;
30. Mottram, L. F., Boonyarattanakalin, S., Kovel, R. E. & Peterson, B. R. The
Pennsylvania green fluorophore: a hybrid of Oregon Green and Tokyo Green
for the construction of hydrophobic and pH-insensitive molecular probes.
Org. Lett. 8, 581–584 (2006).
31. Lee, M. M., Gao, Z. & Peterson, B. R. Synthesis of a fluorescent analogue of
paclitaxel that selectively binds microtubules and sensitively detects efflux by
P-Glycoprotein. Angew. Chem. Int. Ed. 56, 6927–6931 (2017).
32. Balaguer, F. A. et al. Crystal structure of the cyclostreptin-tubulin adduct:
implications for tubulin activation by taxane-site ligands. Int. J. Mol. Sci. 20,
E1392 (2019).
33. Peng, J., Risinger, A. L., Li, J. & Mooberry, S. L. Synthetic reactions with rare
taccalonolides reveal the value of C-22,23 epoxidation for microtubule
stabilizing potency. J. Med. Chem. 57, 6141–6149 (2014).
34. Wang, S. Z., Dong, G. Q. & Sheng, C. Q. Structural simplification of natural
products. Chem. Rev. 119, 4180–4220 (2019).
35. Xiao, H. et al. Insights into the mechanism of microtubule stabilization by
Taxol. Proc. Natl Acad. Sci. USA 103, 10166–10173 (2006).
36. Sastry, G. M., Adzhigirey, M., Day, T., Annabhimoju, R. & Sherman, W.
Protein and ligand preparation: parameters, protocols, and influence on
virtual screening enrichments. J. Comput. Aid. Mol. Des. 27, 221–234 (2013).
37. Skehan, P. et al. New colorimetric cytotoxicity assay for anticancer-drug
screening. J. Natl Cancer Inst. 82, 1107–1112 (1990).
38. Vichai, V. & Kirtikara, K. Sulforhodamine B colorimetric assay for cytotoxicity
screening. Nat. Protoc. 1, 1112–1116 (2006).
39. He, S. et al. P-Glycoprotein/MDR1 regulates pokemon gene transcription
through p53 expression in human breast cancer cells. Int. J. Mol. Sci. 11,
3309–3051 (2010).
References
1. Singh, J., Petter, R. C., Baillie, T. A. & Whitty, A. The resurgence of covalent
drugs. Nat. Rev. Drug. Discov. 10, 307–317 (2011).
2. Robertson, J. G. Mechanistic basis of enzyme-targeted drugs. Biochemistry 44,
5561–5571 (2005).
3. Bauer, R. A. Covalent inhibitors in drug discovery: from accidental discoveries
to avoided liabilities and designed therapies. Drug. Discov. Today 20,
1061–1073 (2015).
4. Johnson, D. S., Weerapana, E. & Cravatt, B. F. Strategies for discovering and
derisking covalent, irreversible enzyme inhibitors. Future Med. Chem. 2,
949–964 (2010).
5. Baillie, T. A. Targeted covalent inhibitors for drug design. Angew. Chem. Int.
Ed. Engl. 55, 13408–13421 (2016).
6. Lonsdale, R. & Ward, R. A. Structure-based design of targeted covalent
inhibitors. Chem. Soc. Rev. 47, 3816–3830 (2018).
7. Irie, H. et al. TAS0728, a covalent-binding, HER2-selective kinase inhibitor
shows potent antitumor activity in preclinical models. Mol. Cancer Ther. 18,
733–742 (2019).
8. Murakami, H. et al. Phase I study of TAS-121, a novel third-generation
epidermal growth factor receptor (EGFR) inhibitor, in patients with EGFR
mutation-positive non-small-cell lung cancer (NSCLC). Ann. Oncol. 28,
1207–1217 (2017).
9. Ito, K. et al. TAS-121, a selective mutant EGFR inhibitor, shows activity
against tumors expressing various EGFR mutations including T790M and
uncommon mutations G719X. Mol. Cancer Ther. 18, 920–928 (2019).
10. Ziegler, S., Pries, V., Hedberg, C. & Waldmann, H. Target identification for
small bioactive molecules: finding the needle in the haystack. Angew. Chem.
Int. Ed. 52, 2744–2792 (2013).
Acknowledgements
This work was funded by R01 CA219948 to A.L.R. and L.D. We would like to thank Dr.
Madesh Muniswamy (University of Texas Health Science Center San Antonio, TX) for
use of the confocal microscope. We would also like to thank Dr. Lily Q. Dong (University
of Texas Health Science Center San Antonio, TX) for the DH10B cells for site-directed
mutagenesis and centrifuges required for the DNA midipreps.
11. Bottcher, T., Pitscheider, M. & Sieber, S. A. Natural products and their
biological targets: proteomic and metabolomic labeling strategies. Angew.
Chem. Int. Ed. 49, 2680–2698 (2010).
12. Wright, M. H. & Sieber, S. A. Chemical proteomics approaches for identifying
the cellular targets of natural products. Nat. Prod. Rep. 33, 681–708 (2016).
13. Broggini-Tenzer, A. et al. Combined treatment strategies for microtubule
stabilizing agent-resistant tumors. J. Natl Cancer Inst. 107, dju504 (2015).
14. Jhaveri, K. et al. A phase I trial of ganetespib in combination with paclitaxel
and trastuzumab in patients with human epidermal growth factor receptor-2
(HER2)-positive metastatic breast cancer. Breast Cancer Res. 19, 89 (2017).
15. Rohena, C. C. & Mooberry, S. L. Recent progress with microtubule stabilizers:
new compounds, binding modes and cellular activities. Nat. Prod. Rep. 31,
335–355 (2014).
Author contributions
L.D. and A.L.R. conceived the project. L.D. performed the organic synthesis and com-
putational modeling. S.S.Y. conducted the biological assays with the exclusion of the
confocal imaging that was done by K.R. L.D. and A.L.R. wrote the manuscript.
16. Schmid, P. et al. Atezolizumab and Nab-paclitaxel in advanced triple-negative
breast cancer. N. Engl. J. Med. 379, 2108–2121 (2018).
17. Risinger, A. L. et al. Taccalonolide binding to tubulin imparts microtubule
stability and potent in vivo activity. Cancer Res. 73, 6780–6792 (2013).
Competing interests
L.D., A.L.R., and S.S.Y. are listed as inventors on a pending patent application pertaining
to the taccalonolide microtubule stabilizers. K.R. declares no competing interests.
12
N
A
T
U
R
E
C
O
M
M
UN
I
C
A
TI
O
N
S
|
(2 0 2 0 ) 1 1 : 6 5 4 | h t t p s : / / d o i . o r g / 1 0 . 1 0 3 8 / s 41 4 6 7 - 0 1 9 - 1 4 2 7 7 - w | w ww .n a t u r e . c o m / n a t u re c o m m u n i c a t i o n s

N
A
T
UR
E
C
O
M
M
U
N
I
CA
T
I
O
N
S
|
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
10
.
1
0
38
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
A R T I C L E
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/s41467-
Open Access This article is licensed under
Attribution 4.0 International License, which permits use, sharing,
a Creative Commons
019-14277-w.
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
Correspondence and requests for materials should be addressed to L.D. or A.L.R.
Peer review information Nature Communications thanks Blake Peterson, Federico Gago
and the other, anonymous, reviewer(s) for their contribution to the peer review of this
work. Peer reviewer reports are available.
Reprints and permission information is available at http://www.nature.com/reprints
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s) 2020
N
A
T
U
R
E
C
OM
M
U
N
I
C
A
T
I
O
N
S
|
(
2
0
2
0
)
1
1
:
6
5
4
|
h
t
t
p
s
:/
/
d
o
i
.
o
r
g
/
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
1
9
-
1
4
2
7
7
-
w
|
w
ww
.
n
a
t
u
r
e
.
c
o
m
/
n
a
t
u
r
e
c
o
m
m
u
n
ic
a
t
i
o
n
s
13