A series of novel pyridine-bridged analogues of combretastatin-A4 (CA-4) were designed and synthesized. As expected, the 4-atom linker configuration retained little cytotoxicities in the compounds 2e, 3e, 3g, and 4i. Activities of the analogues with 3-atom linker varied widely depending on the phenyl ring substitutions, and the 3-atom linker containing nitrogen represents the more favorable linker structure. Among them, three analogues (4h, 4s, and 4t) potently inhibited cell survival and growth, arrested cell cycle, and blocked angiogenesis and vasculature formation in vivo in ways comparable to CA-4. The superposition of 4h and 4s in the colchicine-binding pocket of tubulin shows the binding posture of CA-4, 4h, and 4s are similar, as confirmed by the competitive binding assay where the ability of the ligands to replace tubulin-bound colchicine was measured. The binding data are consistent with the observed biological activities in antiproliferation and suppression of angiogenesis but are not predictive of their antitubulin polymerization activities.
A series of novel pyridine-bridged analogues of combretastatin-A4 (CA-4) were designed and synthesized. As expected, the 4-atom linker configuration retained little cytotoxicities in the compounds 2e, 3e, 3g, and 4i. Activities of the analogues with 3-atom linker varied widely depending on the phenyl ring substitutions, and the 3-atom linker containing nitrogen represents the more favorable linker structure. Among them, three analogues (4h, 4s, and 4t) potently inhibited cell survival and growth, arrested cell cycle, and blocked angiogenesis and vasculature formation in vivo in ways comparable to CA-4. The superposition of 4h and 4s in the colchicine-binding pocket of tubulin shows the binding posture of CA-4, 4h, and 4s are similar, as confirmed by the competitive binding assay where the ability of the ligands to replace tubulin-bound colchicine was measured. The binding data are consistent with the observed biological activities in antiproliferation and suppression of angiogenesis but are not predictive of their antitubulin polymerization activities.
Inhibition
of tubulin polymerization disrupts the formation of
tumor vasculature, making the microtubule cytoskeleton an effective
target for cancer chemotherapy.[1−3] Combretastatin-A4 (CA-4) is the
prototype of a large group of vascular disrupting agents that have
been designed, synthesized, and tested in various biological models
as potential therapeutic candidates for cancer treatment.[4,5] CA-4 binds to the colchicine binding site of tubulin to block microtubule
assembly, causing rapid vascular shutdown and cell death in the tumor.[6] The water-soluble phosphate prodrug form (CA-4P,
also known as fosbretabulin) is in phase II/III clinical trials either
alone or in combination with traditional chemotherapeutic agents or
with radiotherapy.[7−10] Meanwhile, over the past two decades, numerous novel derivatives
of CA-4 have been discovered to confer cytotoxic potency and antitubulin
activity that are comparable to CA-4, significantly expanding the
arsenal of vascular disrupting agents that could be further explored
for clinical applications. Despite the intense interest and the large
number of potent derivatives of CA-4 that have been discovered that
aimed at targeting the colchicine-binding site of tubulin, none of
these inhibitors has reached the clinical stage. Thus, challenges
remain in developing CA-4 analogues with improved pharmacological
properties for eventual acceptance in the clinic.Modifications
made on the two phenyl rings, for example, have led
to hundreds of active compounds that possess desirable cytotoxicity
while retaining varying degrees of antitubulin activities.[11] Most structural variations of the phenyl rings
involve different combinations of hydroxyl and methoxy substitutions.
These include various substituted phenyl rings[12] and other aromatic rings.[13] A
few reports have attempted to modify the trimethoxy ring with mixed
outcomes. For example, the m-methoxy group has been
substituted by a fluoride to yield a similarly potent compound.[14] In another example, when the trimethoxy ring
was replaced by a trimethyl ring,[15] the
cytotoxicity of the compound was significantly reduced but the antitubulin
activity was largely retained. This suggests that it might be possible
to achieve disruption of tumor vasculature with fewer cytotoxic side
effects.Modifications of the double bond have also led to diverse
structural
variations that remain viable as cytotoxic and antitubulin compounds.
The olefinic bond is believed to be critical in placing the two phenyl
rings at an appropriate distance and giving the molecule the right
dihedral angle to maximize the interaction with the target. As such,
replacement of the double bond by rings that facilitate a cis-locked configuration has proven to be effective in retaining
both cytotoxicity and antitubulin activity.[16−20] Indeed, this strategy has led to the discovery of
perhaps more active CA-4 analogues than any other types of structural
modifications. On the other hand, the observation that two-carbon
linkers are the optimal length of the bridge between the two phenyl
rings has somewhat limited exploration in this strategy of modifications
with some encouraging exceptions. For example, when the methylene
bridge is replaced by a carbonyl group, the resulting analogue, phenstatin,
actually retained much of the antitubulin activity.[21] Interestingly, increasing the bridge length to three carbons
such as a chalcone-like linker have been reported to strongly inhibit
tubulin polymerization as well as cell survival.[22] However, progress in this direction of structural modifications
has been limited.The current study was undertaken to investigate
the effect of a
novel variation of bridge length and structure on the anticancer activities
of resulting CA-4 analogues. Specifically we have designed, synthesized,
and evaluated a series of novel pyridine-linked CA-4 analogues (Figure 1) in which the distance between the two phenyl rings
is configured to be three or four atoms, i.e., meta- or para- to each
other. Pyridine has been introduced to replace the cis-double bond between the A ring and B ring.[23] However, it was found that the antitubulin activities were largely
lost in these pyridine-containing analogues. We show that cytotoxicity
and antitubulin activities comparable to CA-4 can be obtained when
the bridge length is fixed at three atoms (including the pyridinenitrogen) and substitutions on one or both of the phenyl rings are
optimized. Here we describe the synthesis of 34 pyridine-bridged CA-4
analogues that were tested for their ability to inhibit cancer cell
growth and proliferation by arresting cell cycles, their effect on
tubulin polymerization as well as their activities in blocking angiogenesis.
Molecular modeling and competitive binding assays were also performed
to better understand the structural requirements for the pyridine-linked
CA-4 analogues to retain antimitotic potency.
Figure 1
Structures of CA-4 and
three pyridine-linked analogues 4h, 4s,
and 4t.
Structures of CA-4 and
three pyridine-linked analogues 4h, 4s,
and 4t.
Results and Discussion
Chemistry
As shown in Scheme 1, the designed analogues were synthesized following
the general procedures
as detailed below and in the Experimental Section. Starting from commercially available dibromopyridines, palladium-catalyzed
monocoupling reactions gave mainly the products of 2 with
varying yields of the minor products of 3, some of which
were isolated and purified successfully. The products of 2 were further transformed to symmetrically or unsymmetrically diaryl-substituted
pyridines (3 and 4) by the application of
aforementioned sequential Suzuki coupling. The reaction of 4s with hydrogen chloride in ether afforded its chloride salt 5.
Scheme 1
Synthesis of Novel Pyridine-Bridged Analogues of Combrestastatin-A4
To
evaluate the cytotoxicities
of the pyridine-bridged combretastatin analogues, three humancancer
cell lines were cultured and treated with the pyridine-bridged combretastatin
analogues. Treatment at four different concentrations was used in
order to determine the IC50 values for each compound along
with the reference compound, CA-4. As shown in Table 1, when the two phenyl rings are para to each other on the
pyridine linker, separated by a 4-carbon bridge, the cytotoxicity
as measured by the antiproliferative activity is largely lost. Thus,
analogues 3e, 3g, and 4i all
showed high IC50 values, suggesting that such stretched-out
configuration of the whole molecule does not bind well in the purported
colchicine binding site to be effective. When the two phenyl rings
are 2,4 substitutions on the pyridine linker, making the bridge a
3-carbon length where the pyridinenitrogen is not included (3f, 3h), results are mixed, with 3f showing modest cytotoxicity while 3h shows none. However,
in the configuration where the 3-carbon linker contains the pyridinenitrogen, i.e., the two phenyl rings are now 2,6 meta to each other,
significant improvement in the antiproliferative activity is seen
in some analogues with further optimization of the phenyl ring substitutions.
Indeed, when separated by three carbons, ring substitutions appear
to be critical in conferring enhanced cytotoxicity. Trimethoxy substitution
on one phenyl ring generally affords few options of substitutions
on the other ring. For example, of the five analogues (4a, 4c, 4f, 4g, 4t) containing the trimethoxy phenyl group, 4a and 4f were inactive in all three cell lines, 4c was
inactive in A549 cells, and 4g was inactive in MDA-MB-231
cells. The analogue 4t that differed from CA-4 in just
the pyridine linker was found to be the only one in this group that
potently inhibited growth of all three cancer cell lines that were
tested.
Table 1
Growth Inhibition IC50 Values
of Synthetic Pyridine-Bridged CA-4 Analogues in Three Cancer Cell
Lines: MDA-MB-231, A549, and HeLa
growth inhibition IC50 (μM)
combretastatin
analogues
MDA-MB-231
A549
HeLa
2a
6.0 ± 3.5
9.0 ± 0.1
12.0 ± 0.2
2b
>25
>25
>50
2c
>100
>50
>50
2d
10.0 ± 0.1
2.0 ± 0.2
2.0 ± 0
2e
12.05 ± 0.02
>25
14.0 ± 0.3
2f
>100
>100
>25
3a
11.0 ± 1.8
>50
1.0 ± 0.2
3b
>50
>50
>25
3c
0.075 ± 0.001
0.079 ± 0.019
0.011 ± 0.001
3d
9.0 ± 1.0
>100
2.0 ± 0.4
3e
11.0 ± 0.4
>100
2.68 ± 0.08
3f
2.63 ± 0.15
7.86 ± 0.233
0.79 ± 0.02
3g
>100
>100
>50
3h
>50
>50
>25
4a
10.00 ± 1.56
>25
8.0 ± 0.5
4b
>50
>100
0.86 ± 0.01
4c
2.49 ± 0.0022
>50
0.026 ± 0.001
4d
>25
>25
>25
4e
13.0 ± 0.1
>25
>50
4f
>25
>50
>50
4g
>100
0.69 ± 0.05
4.43 ± 0.08
4h
0.0031 ± 0.0003
0.089 ± 0.008
0.0038 ± 0.0001
4i
>50
>50
>25
4j
>25
>25
>50
4k
17.2 ± 2.6
4.22 ± 0.58
>100
4l
>25
>25
>50
4m
0.034 ± 0.004
0.33 ± 0.01
0.034 ± 0.002
4n
>50
>50
>50
4o
>100
>100
>50
4p
9.02 ± 0.09
0.56 ± 0.01
0.067 ± 0.007
4q
>50
>50
>25
4r
>25
>50
>25
4s
0.0046 ± 0.0001
0.044 ± 0.002
0.0014 ± 0.0003
4t
0.069 ± 0.0028
2.64 ± 0.38
0.0047 ± 0.0009
5
0.0054 ± 0.0006
0.042 ± 0.008
0.0026 ± 0.0006
CA-4
0.0028 ± 0.0004
0.0038 ± 0.0003
0.0009 ± 0.0003
The most significant enhancement of antiproliferative
activity
was observed in the pyridine-linked analogues where one phenyl ring
accommodates the 2,4-dimethoxy substitutions. A symmetric analogue 3c with both phenyl ring adopting the 2,4-dimethoxy configuration
gave low nanomolar IC50 values in all three cancer cell
lines. However, when the dimethoxy substitution on one phenyl ring
is in any other position than 2,4 (4e, 3,4-dimethoxy; 4l, 3,5-dimethoxy; 4n, 2,3-dimethoxy; 4o, 2,6-dimethoxy), there is significant loss of the antiproliferative
activity with the only exception of 4m, having a 2,5-dimethoxyphenyl
ring. Interestingly, 3,4,5-trimethoxy substitution (4g) on the second phenyl ring produced modest antiproliferative activity
in A549 and HeLa cells, but 2,3,5-trimethoxy substitution (4q) abolished any such activity. When only monomethoxy substitution
is introduced into the second ring while maintaining the 2,4-dimethoxy
substitution on the first ring, 3-methoxy (4p), 4-methoxy
(4h), but not 2-methoxy (4r) substitution
affords nanomolar to micromolar cytotoxicity. In fact, 4h is one of the most potent analogue in the series. Finally, addition
of a 3-hydroxy substitution to the second ring of 4h leads
to yet the most potent analogue, 4s, that shows low nanomolar
potency in all three cell lines. Thus, one phenyl ring of 4s is identical to that of CA-4 with 3-hydroxy-4-methoxy substitutions
and the other phenyl ring contains 2,4-dimethoxy substitutions (Figure 1).
Antimicrotubule Effects in HeLa Cells
To determine
the microtubule disrupting effects of the pyridine-linked analogues,
we selected 4h as a representative compound in a cell-based
phenotypic screening. HeLa and MDA-MB-231 cells were used to examine
the effect of 4h on the reorganization of microtubules
during mitosis. Cells were treated with either 4h or
CA-4 for 24 h, and microtubules were visualized by indirect immunofluorescence
techniques. An antibody for β-tubulin was used to visualize
interphase and mitotic microtubule structures. Vehicle-treated cells
exhibited a normal filamentous microtubule array, with microtubules
extending from the central regions of the cell to the cell periphery
(Figure 2A,D). Compound 4h caused
dramatic reduction of the interphase microtubule network, as shown
in Figure 2C in HeLa cells and in Figure 2F for MDA-MB-231 cells. As demonstrated in parts
B and E of Figure 2 for the two cell lines,
respectively, CA-4 induced very similar loss of the microtubule network.
Figure 2
Effects
of CA-4 and 4h on interphase microtubules.
MDA-MB-231 and HeLa cells were treated with vehicle (A,D) or 1 μM
compounds (B,C,E,F) as indicated for 24 h. Cells were then fixed and
microtubules visualized by indirect immunofluorescence techniques.
Normal interphase microtubules are visible in the control cells of
HeLa and MDA-MB-231. The loss of interphase microtubules induced by
CA-4 and 4h is shown in respective treatments as labeled.
Effects
of CA-4 and 4h on interphase microtubules.
MDA-MB-231 and HeLa cells were treated with vehicle (A,D) or 1 μM
compounds (B,C,E,F) as indicated for 24 h. Cells were then fixed and
microtubules visualized by indirect immunofluorescence techniques.
Normal interphase microtubules are visible in the control cells of
HeLa and MDA-MB-231. The loss of interphase microtubules induced by
CA-4 and 4h is shown in respective treatments as labeled.
Effect on Cell Cycle Arrest
To investigate the effect
of the pyridine-linked CA-4 analogues on cell cycle arrest, we used
flow cytometry to analyze the cell cycle distribution of HeLa cells
following treatment with 4h and 4s at 1
μM. Untreated cells were used as a negative control, and cells
treated with CA-4 were used as a positive control. As shown in Figure 3, the two most potent CA-4 analogues, 4h and 4s, were found to be as effective in arresting
the cell cycle at G2/M phase as CA-4. With the untreated cells, the
percentage of cells in the G0/G1 phase was at 61.6% with only 15.70%
in the G2/M phase. After treatment with 4h or 4s, the percentage of cells in the G2/M phase increased to 73.4% and
70.6%, respectively. These results compare favorably to 44.6% in the
G2/M phase for cells treated with CA-4.
Figure 3
Flow cytometric analysis
of cell cycle distributions of HeLa cells
treated with pyridine-bridged CA-4 analogues 4h and 4s.
Flow cytometric analysis
of cell cycle distributions of HeLa cells
treated with pyridine-bridged CA-4 analogues 4h and 4s.Similar results were
obtained when another cancer cell line, MDA-MB-231,
was used to test the effect of these compounds on cell cycle arrest.
At 1 μM concentration, 4h and 4s again
showed strong cell cycle inhibition. As indicated in Table 2, after treatment with 4h and 4s, 73.4% and 70.6% of HeLa cells were arrested in the G2/M
phase, respectively, whereas treatment with CA-4 resulted in 44.6%
G2/M phase cells. Similarly, the percentage of MDA-MB-231 cells in
the G2/M phase increased from 15.5% (control) to 57.8% (CA-4 treated),
70.5 (4h treated), and 54.1 (4s treated).
Table 2
HeLa and MDA-MB-231 Cell Cycle Distribution
for after Treatment with CA-4, 4h, and 4s
G0/G1 (%)
S (%)
G2/M (%)
HeLa
Cells
control
(−)
61.60
22.10
15.70
treated with CA-4
37.00
16.70
44.60
treated with 4h
10.30
15.40
73.40
treated with 4s
12.10
17.10
70.60
MDA-MB-231 Cells
control (−)
58.60
25.70
15.50
treated with CA-4
27.50
11.80
57.80
treated with 4h
13.10
16.40
70.50
treated with 4s
18.60
17.10
54.10
Inhibition of Tubulin Polymerization
in Vitro
The inhibition
of tubulin polymerization by the pyridine-linked CA-4 analogues 4h, 4s, and 4t were tested using
bovine brain tubulin. As shown in Figure 4,
incubation with either vehicle (DMSO), CA-4, 4h, 4s, or 4t resulted in various degrees of inhibition
of tubulin polymerization, depending on the compound and the dose.
At 1 μM, all pyridine-linked analogues (4h, 4s, and 4t) failed to inhibit tubulin polymerization.
Compared to vehicle, 4h at this concentration appeared
to be slightly stimulating tubulin polymerization. In contrast, CA-4
at 1 μM inhibited tubulin polymerization by 35%. When analogues
concentrations were increased to 10 μM, 4h remained
ineffective against tubulin polymerization, whereas 4s and 4t were seen to inhibit tubulin polymerization
by 57% and 32%, respectively. In comparison, CA-4 at 10 μM nearly
completely blocked tubulin polymerization (Figure 4). These data suggest that it is possible for the pyridine-linked
analogues to have potent cytotoxicity but very little antitubulin
polymerization activity, such as 4h.
Figure 4
Effects of pyridine-linked
CA-4 analogues on microtubule dynamics.
Polymerization of tubulin at 37 °C in the presence of paclitaxel
(10 μM), CA-4 (1 and 10 μM), 4h (1 and 10
μM), 4s (1 and 10 μM), and 4t (1 and 10 μM) and were monitored continuously by measuring
the absorbance at 340 nm over 60 min. The reaction was initiated by
the addition of tubulin to have a final concentration of 3.0 mg of
tubulin per mL of incubation solution.
Effects of pyridine-linked
CA-4 analogues on microtubule dynamics.
Polymerization of tubulin at 37 °C in the presence of paclitaxel
(10 μM), CA-4 (1 and 10 μM), 4h (1 and 10
μM), 4s (1 and 10 μM), and 4t (1 and 10 μM) and were monitored continuously by measuring
the absorbance at 340 nm over 60 min. The reaction was initiated by
the addition of tubulin to have a final concentration of 3.0 mg of
tubulin per mL of incubation solution.
Antiangiogenesis Assay Using Chick Embryo Chorioallantoic Membrane
(CAM)
The three most potent pyridine-linked combretastatin
analogues, 4h, 4s, and 4t,
were then tested for antiangiogenic and vasculature disrupting properties
using the CAM assay. In this test, the vascular system of a fertilized
chicken embryo is used as a model. Figure 5 demonstrates the effects of 4h, 4s, and 4t along with CA-4 on the development of embryonal blood vessels
compared to a negative control (PBS) and a positive control (10 ng/plug
basic fibroblast growth factor (bFGF) and 25 ng/plug vascular endothelial
growth factor (VEGF)). The antiangiogenesis activities of CA-4 and
the three pyridine-bridged analogues were determined by the suppression
of angiogenic action of BV (bFGF + VEGF) when the compound was added
to a collagen plug containing BV and placed on the chorioallantoic
membrane of 10-day old embryos for 3 days. All compounds, 4h, 4s, 4t, and CA-4, led to inhibition of
new vessel growth 4 days after treatment scored on a scale of 0–3
in a blinded manner. The negative control (vehicle) and the test compounds 4s and 4t scored significantly lower than the
positive control (BV). The CAM assay results suggest that the pyridine-bridged
analogues 4h, 4s, and 4t are
potent inhibitors of angiogenesis, with 4s being more
effective than CA-4 in suppressing new vessel growth.
Figure 5
Effect of 4h, 4s, 4t, and
CA-4 on angiogenesis in chick embryo. CA-4 and pyridine-bridged analogues 4h, 4s, and 4t potently inhibited
the induction of angiogenesis by bFGF and VEGF (BV). Angiogenesis
was scored by two scorers in a blinded fashion with similar results.
*, p < 0.05; **, p < 0.01,
***, p < 0.001.
Effect of 4h, 4s, 4t, and
CA-4 on angiogenesis in chick embryo. CA-4 and pyridine-bridged analogues 4h, 4s, and 4t potently inhibited
the induction of angiogenesis by bFGF and VEGF (BV). Angiogenesis
was scored by two scorers in a blinded fashion with similar results.
*, p < 0.05; **, p < 0.01,
***, p < 0.001.
Mouse Plasma Concentrations of 4h, 4s,
and 4t vs CA-4
To evaluate the possible effect
of the pyridine bridge on the analogues’s bioavailability,
we measured the plasma concentration levels of 4h, 4s, and 4t in comparison with CA-4 in mice that
were given a single dose of 5 mg/kg of the compound (Figure 6). After oral administration, blood samples were
collected from the orbital sinus of the mice at 1, 3, 6, and 24 h.
At 1 and 3 h, CA-4 reached plasma concentration of 2.3 and 2.1 ng/mL,
respectively. At 6 and 24 h, plasma concentration of CA-4 was no longer
detectable. Analogue 4h and 4t showed lower
peak concentrations than CA-4. Analogue 4s was detected
at the highest plasma concentration of 9.2 ng/mL at the first hour,
with its level dropping rapidly at 6 h and going below detection limit
at 24 h. These preliminary results indicate that the pyridine-bridged
analogues varied widely in their bioavailability, as reflected in
their respective plasma concentrations. It appears that 4s demonstrated the best bioavailability of the three analogues tested.
Figure 6
Plasma
concentrations of CA-4, 4h, 4s, and 4t in mice after the administration of a single
oral dose of 5 mg/kg. Four blood samples were collected, each at 1,
3, 6, and 24 h, respectively, after oral intake.
Plasma
concentrations of CA-4, 4h, 4s, and 4t in mice after the administration of a single
oral dose of 5 mg/kg. Four blood samples were collected, each at 1,
3, 6, and 24 h, respectively, after oral intake.
Molecular Modeling
To elucidate the binding mode of
pyridine-linked combretastatin analogues, we postulated that the analogues
have the same binding site as colchicine and CA-4. We have visually
analyzed 20 top scoring postures for each compound, and most of the
top scoring poses of each compound changed very little. The Surflex
docking scores were 7.35 for CA-4, 7.79 for 4h, 7.53
for 4s and 5.75 for 4t, where higher scores
indicate greater binding affinity. The order of the docking scores
appears to be consistent with the IC50 values of the compounds
for growth inhibition of MDA-MB-231humanbreast cancer cells. Here
we examine the docking details of three most potent pyridine-linked
analogues, 4h, 4s, and 4t (IC50 values are 2.75 nM for CA-4, 3.13 nM for 4h, 4.56 nM for 4s, and 68.7 nM for 4t in
MDA-MB-231 cells) as compared to CA-4. The structures of these compounds
are shown in Figure 1.The binding modes
of these compounds in the colchicine-binding site of tubulin are depicted
in Figure 7. The binding postures of CA-4, 4h, and 4s (Figure 7A–C)
in the binding pockets are similar. CA-4 formed a hydrogen bond with
the backbone carbonyl oxygen of Val238, which was absent when 4h and 4s were bound. The close proximity of
the functional groups of CA-4 (Figure 7A) to
the polar amino acids Asn258, Lys352, Met259, and Cys241 suggests
a likely stronger electrostatic interaction with the protein. In addition,
the hydrophobic moiety of the CA-4 is well embedded in a pocket interacting
with several hydrophobic residues making CA-4 bind tightly to tubulin.
Despite the absence of one hydroxyl group in 4h, its
binding pose is essentially superimposable to 4s, and
they both bind in a fashion similar to CA-4. The close proximity of
the functional groups of 4h (Figure 7B) to the side chain polar residues Asn258, Met259, Tyr202,
Cys241, and Asn167 allows stronger interactions with the protein.
The additional hydroxyl group in 4s (Figure 7C) seems to be stabilized by its interaction with
the side chain of Lys352. Here again the hydrophobic moieties of these
compounds are well shielded by several hydrophobic residues in the
binding pocket. The binding posture of 4t, however, is
slightly different from CA-4, 4h, and 4s. The three bulky methoxy groups on the para and meta positions of
one of the phenyl rings of 4t present steric hindrance
to occupying the exact colchicine binding site, causing it to bind
to a better position in the nearby colchicine binding site. Even though
CA-4 also has three methoxy groups on one phenyl ring, the cis conformation
of CA-4 makes it more compact than 4t, thus it is less
impeded by steric hindrance to binding into the colchicine binding
site. The functional groups of 4t (Figure 7D) are in close proximity to the side chains of Asn349 and
Asn258, and the hydrophobic moieties interact with the neighboring
hydrophobic side chains. The overall interactions between 4t and tubulin are weaker than CA-4, 4h, and 4s.
Figure 7
Molecular modeling of CA-4 and three pyridine-linked analogues
in complex with tubulin. Shown is the proposed binding mode and interaction
between tubulin and selected compounds, (A) CA-4, (B) 4h, (C) 4s, and (D) 4t. The compounds and
important amino acids in the binding pockets are shown in stick model,
whereas tubulin is depicted in the ribbon model.
Molecular modeling of CA-4 and three pyridine-linked analogues
in complex with tubulin. Shown is the proposed binding mode and interaction
between tubulin and selected compounds, (A) CA-4, (B) 4h, (C) 4s, and (D) 4t. The compounds and
important amino acids in the binding pockets are shown in stick model,
whereas tubulin is depicted in the ribbon model.The superposition of all four compounds in the binding pocket
is
depicted in Figure 8. It clearly shows the
binding posture of CA-4, 4h, and 4s are
similar, whereas the binding location of 4t is slightly
different. The free energy of binding of these compounds with the
receptor protein tubulin was calculated by employing the MMGB/SA approach.
The binding energies are −77.46, −71.24, −66.72,
and −49.54 kcal/mol for CA-4, 4h, 4s, and 4t, respectively. The binding energies correlate
with the experimental IC50 values of these compounds in
MDA-MB-231humanbreast cancer cells. Both the docking scores and
the free energy of binding show weaker binding of 4t,
which is in agreement with their observed cytotoxic potencies (IC50: 4h < 4s < 4t) but not the experimentally measured antitubulin polymerization
activities where 4h was not active in inhibiting tubulin
polymerization and 4s and 4t were minimally
active. This inconsistency suggests that cytotoxicity of the pyridine-linked
CA-4 analogues could be separated from their antitubulin polymerization
activities.
Figure 8
Superposition of CA-4, 4h, 4s, and 4t in the colchicine-binding pocket of tubulin in surface
representation. The compounds are shown in stick model with carbon
atoms in dark-green (CA-4), light-green (4h), orange
(4s), or purple (4t).
Superposition of CA-4, 4h, 4s, and 4t in the colchicine-binding pocket of tubulin in surface
representation. The compounds are shown in stick model with carbon
atoms in dark-green (CA-4), light-green (4h), orange
(4s), or purple (4t).
Competitive Binding to Colchicine Binding Site of Tubulin
To determine if the CA-4 analogues indeed bind to the colchicine-binding
site of tubulin, we next performed a binding experiment based on an
HPLC-MS based method.[26] In this method,
colchicine was incubated with tubulin in the absence or presence of
various concentrations of the test compounds, CA-4, 4h, 4s, and 4t in the incubation buffer (80
mM PIPES, 2.0 mM MgCl2, 0.5 mM EGTA, pH 6.9) at 37 °C for 1 h. After incubation, the concentration of the
displaced colchicine was measured by HPLC-MS/MS as previously described.[26] The ability of the analogue to inhibit the binding
of colchicine was expressed as a percentage of control binding in
the absence of any competitor. As shown in Figure 9, CA-4 demonstrated the highest relative binding capacity,
reaching 78% of colchicine binding. All three analogues showed similar,
but slightly weaker binding affinity than CA-4. The binding results
are largely consistent with the molecular modeling predictions, with
minor differences in the order of binding scores where 4h and 4s scored higher than 4t.
Figure 9
Competitive
mass spectrometry-based binding assay for characterization
of CA-4, 4h, 4s, and 4t binding
to the colchicine binding site of tubulin.
Competitive
mass spectrometry-based binding assay for characterization
of CA-4, 4h, 4s, and 4t binding
to the colchicine binding site of tubulin.
Conclusion
We have designed and synthesized a series
of combretastatin analogues
with a pyridine linker that exceeds the standard two-atom linker length.
Structurally, these compounds represented nonisomerizable analogues
of CA-4 with three linker configurations: a 3-atom linker containing
the pyridinenitrogen, a 3-atom linker without the nitrogen, and a
4-atom linker with the two phenyl rings para to each other on pyridine.
As expected, the 4-atom linker configuration retained little cytotoxicities
in the compounds (2e, 3e, 3g, and 4i). Activities of the analogues with 3-atom linkers
varied widely depending on the phenyl ring substitutions, but the
most potent compounds are exclusively with the nitrogen-containing
linker configuration. The progressive improvement of activity from 3e to 3f to 4c demonstrates, at
least qualitatively, that the 3-atom linker containing nitrogen represents
the more favorable linker structure. More importantly, this novel
3-atom pyridine linker led us to discover at three analogues (4h, 4s, and 4t) that potently inhibited
cell survival and growth, arrested cell cycle, and blocked angiogenesis
and vasculature formation in vivo in ways comparable to CA-4. However,
unlike CA-4, 4h was inactive in inhibiting tubulin polymerization
and 4s and 4t were significantly weaker
than CA-4 in blocking antitubulin polymerization. Molecular modeling
analysis indicates all three analogues bind to the colchicine site
of tubulin with 4h and 4s, showing greater
binding affinity. Competitive binding assay confirmed that the pyridine-linked
analogues bind to the colchicine site with affinities similar to CA-4.
Experimental Section
General Chemistry
All reagents and solvents were purchased
from CombiPhos Catalyst, Combi-Block, AK Scientific, Sigma-Aldrich
Chemical Co., Acros, and Pharmco-AAPER and were used as received. 1H and 13C NMR spectra were obtained on a Bruker-300
NMR spectrometer. Chemical shifts are reported as parts per million
(ppm) relative to TMS. Mass spectral data were collected on an Agilent
2010 instrument, and HRMS spectra data were collected on a Thermo
LTQ Orbitrap-XL mass spectrometer in positive ion modes. Unless specified
otherwise, all tested compounds were confirmed to be >95% pure
by
HPLC and GC-MS. Melting points were obtained on a Haake Buchler melting
point apparatus and are uncorrected.
General Procedure for Synthesis
of Compounds 2 and 3
The mixture
of the dibromopyridines 1a–c (11.8
g, 0.05 mol), the phenyl boronic acids
(0.075 mol), sodium carbonate (15.9 g, 0.15 mol), and PdCl2(dppf) (0.40 g, 0.5 mmol) in toluene–ethanol (4:1, 100 mL)
was stirred at 80 °C until completion of reaction (about 3–4
days). After cooling down to room temperature, the reaction mixture
was filtered through Celite and the filtrate was concentrated. The
residue crude was purified to afford products 2 and 3 (minor) by flash column chromatography on silica gel with
hexane–ethyl acetate (9:1) as eluant.
The mixture of 2-bromo-6-(4-methoxyphenyl)pyridine
(2d, 1.3 g, 5 mmol), 4-methoxyphenyl boronic acids (1.1
g, 7.5 mol),
sodium carbonate (1.6 g, 0.015 mol), and PdCl2(dppf) (0.1
g, 0.1 mmol) in toluene–ethanol (4:1, 10 mL) was stirred overnight
at 120 °C. After cooling down to room temperature, the reaction
mixture was filtered through Celite followed by concentration of the
filtrate. The residue crude was purified to afford product 3d by flash column chromatography on silica gel with hexane–ethyl
acetate (9:1) as eluant.
The mixture of 2,6-dibromopyridines 1a (5.9 g, 0.025
mol), 2,4-dimethoxyphenyl boronic acids (13.7 g, 0.075 mol), sodium
carbonate (15.9 g, 0.15 mol), and PdCl2(dppf) (0.20 g,
0.25 mmol) in toluene–ethanol (4:1, 50 mL) was stirred overnight
at 120 °C. After cooling down to room temperature, the reaction
mixture was filtered through Celite and the filtrate was concentrated.
The residue crude was purified to afford product 3c by
flash column chromatography on silica gel with hexane–ethyl
acetate (9:1) as eluant.
The mixture of
the monobromopyridines 2a–f (3 mmol),
the phenyl boronic acids (4 mmol), sodium carbonate
(0.5 g, 5 mmol), and PdCl2(dppf) (0.1 g, 0.1 mmol) in toluene–ethanol
(4:1, 100 mL) was stirred overnight at 120 °C. After cooling
down to room temperature, the reaction was quenched with saturated
ammonium chloride solution; the organic layer was separated and extracted
with ethyl acetate. The combined organic solution was dried over MgSO4, filtered, and concentrated. The residue crude was purified
to afford products 4 by flash column chromatography on
silica gel with hexane–ethyl acetate as eluant.
To the solution of 4s (0.3 g, 0.88 mmol) in THF (5 mL) was added HCl solution in ether
(0.44 mL, 0.88 mmol, 2 M). The resultant mixture was stirred at rt
overnight. The precipitate was filtered and washed with dichloromethane
to afford 5 (0.32 g) in yield of 97%. Solid, mp 140 °C
(d). 1H NMR (DMSO-d6, 300 Hz):
8.28 (1H, t, J = 7.8 Hz), 7.96 (1H, d, J = 7.8 Hz), 7.89 (1H, d, J = 7.8 Hz), 7.75 (1H,
d, J = 8.4 Hz), 7.45–7.42 (2H, m), 7.13 (1H,
d, J = 8.7 Hz), 3.92 (3H, s), 3.861 (3H, s), 3.855
(3H, s). 13C NMR (DMSO-d6,
75 Hz): 163.1, 158.6, 152.8, 151.6, 150.6, 147.1, 143.1, 132.5, 126.4,
124.1, 121.0, 119.9, 115.7, 114.8, 112.5, 106.4, 99.0, 56.3, 56.0,
55.9. MS-ESI: 338 (M+-Cl). HRMS (ESI(+)) calcd for C20H20NO4 (M – Cl): 338.1392; found
338.1387.
Antiproliferative Assays
Human breast
adenocarcinoma
(MDA-MB-231), humannonsmall cell lung carcinoma (A549), and humancervical carcinoma (HeLa) cells were grown in DMEM medium supplemented
with 115 units/mL of penicillin G, 115 μg/mL of streptomycin,
and 10% fetal bovine serum (all from Life Technologies, Grand Island,
NY). Cells were seeded in 96-well plates (5 × 103 cells/well)
containing 50 μL of growth medium for 24 h. After medium removal,
100 μL of fresh medium containing individual analogue compounds
at different concentrations was added to each well and incubated at
37 °C for 72 h. After 24 h of culture, the cells were supplemented
with 50 μL of analogue compounds dissolved in DMSO (less than
0.25% in each preparation). After 72 h of incubation, 20 μL
of resazurin was added for 2 h before recording fluorescence at 560
nm (excitation) and 590 nm (emission) using a Victor microtiter plate
fluorimeter (PerkinElmer, USA). The IC50 was defined as
the compound concentration required to inhibit cell proliferation
by 50% in comparison with cells treated with the maximum amount of
DMSO (0.25%) and considered as 100% viability.
Immunofluorescence
MDA-MB-231 and HeLa cells were grown
on a Laboratory-Tek chamber slide (VWR International, Radnor, PA)
and treated with vehicle (DMSO) or 1 μM CA-4 or 4h for 24 h to 4 days. At the completion of treatment, cells were fixed
with 3.7% formaldehyde and permeabilized with 0.1% Triton X-100 in
PBS for 4 min. The cells were first incubated for 1 h in a solution
of PBS containing 1% BSA and calf serum to block nonspecific antibody
binding. The cells were then incubated with the mouse antitubulin
antibody (1:200) (Life Technologies, Grand Island, NY), washed three
times in PBS containing 1% BSA, and incubated for 2 h at room temperature
with secondary goat antimouse antibody Alexa 488 labeled (1:200) for
tubulin staining. The chamber slides were examined and photographed
using a Nikon ES800 fluorescence microscope with a digital camera.
Tubulin Polymerization Assays
A tubulin polymerization
kit (Cytoskeleton, Denver, CO) was used to evaluate effect of the
pyridine-linked CA-4 analogues on tubulin assembly in vitro.[24,25] It is based on the principal that light is scattered by microtubules
to an extent that is proportional to the concentration of the microtubule
polymer. Compounds that interact with tubulin will alter the polymerization
of tubulin, and this can be detected using a spectrophotometer. The
absorbance at 340 nm at 37 °C is monitored. The experimental
procedure of the assay was performed as described in version 8.2 of
the tubulin polymerization assay kit manual. Varying concentrations
of compounds were preincubated with 10 μM bovine brain tubulin
in glutamate buffer at 30 °C and then cooled to 0 °C. After
the addition of 0.4 mM GTP, the mixtures were transferred to 0 °C
cuvettes in a recording spectrophotometer and warmed to 30 °C.
Tubulin assembly was monitored by measuring the optical density at
340 nm using a BioTek Synergy 4 multifunction microplate spectrophotometer.
Cytofluorimetric Analysis of Cell Cycle Distribution
Cells
treated with tested compounds for 24 h were washed once in
PBS and resuspended in 1 mL of 70% ice-cold ethanol and stored at –20 °C. Fixed cells were washed twice in PBS and
then treated with 1 mL of 0.1 mg/mL of RNase A solution at 37 °C
for 1 h. DNA was then stained with a PBS solution containing 0.1 mg/mL
propidium iodide for 30 min at room temperature in the dark. Cell
cycle analysis was determined with Accuri C6 (BD Biosciences, Mountain
View, CA).
Chick Embryo Chorioallantoic Membrane (CAM)
Assay
Fertilized
embryos were obtained from Charles River Laboratories and incubated
at 37.5 °C for 3 days, removed from their shell using a Dremel
tool and placed into a covered weighboat for 10 further days of incubation.
Solidified 30 μL onplants containing 2.1 mg/mL rat tail collagen
(BD Biosciences, Bedford, MA) and 10 ng/plug bFGF and 30 ng/plug VEGF
in the presence or absence of Ca-4 and 4s were placed
on the embryo chorioallantoic membrane (CAM) over two pieces of nylon
mesh approximately 0.5 cm2. Four collagen onplants were
added per egg on at least three separate eggs. After 4 days of incubation,
images were taken of each plug on surviving embryos using a mini-Vid
camera (LW Scientific, Lawrenceville, GA) and quantified in a masked
fashion on a scale from 0 to 3 with 0 representing no angiogenesis
and 3 representing extreme angiogenesis. Data from one scorer (confirmed
by a second scorer) are presented as the means ± standard errors
of the mean. Statistical significances were determined by one-way
analysis of variance (ANOVA) followed by Dunnett’s multiple
comparison test (GraphPad Prism, La Jolla, CA).
Molecular
Modeling
All the docking studies were carried
out using Sybyl-X 1.3 on a linux workstation. The initial coordinates
for tubulin was taken from the crystal structure of tubulin in complex
with colchicine (PDB: 1SA0.pdb). After removing all the ions and substructures,
such as GDP and GTP present in the crystal structure, hydrogen atoms
were added to the protein system using Amber99 force field. Four representative
CA-4 analogues, including the parent compound CA-4, were selected
for the docking studies. The 3D structures of these selected compounds
were first built using Sybyl-X 1.3 sketch followed by energy minimization
using the MMFF94 force field and Gasteiger–Marsili charges.
We employed Powell’s method for optimizing the geometry with
a distance dependent dielectric constant and a termination energy
gradient of 0.05 kcal/mol. Structural resemblance of combretastatin
to colchicine indicates that combretastatin binds to tubulin at the
same site as colchicine and presumably has a similar mode of action.
Hence we used colchicine binding site for docking our selected compounds
to tubulin. We used the Surflex docking program to automatically dock
all the selected compounds into the binding pocket of tubulin. Surflex
is a fully automatic flexible molecular docking algorithm that combines
Hammerhead’s empirical scoring function with a molecular similarity
method to generate putative poses of ligand fragments. Docking process
was guided by the protomol, which is a computational representation
of the intended binding site. We used the position of the native ligand,
colchicine, in the crystal structure to generate the protomol. In
all the cases, we used the default screening protocol of Surflex that
employs Surflex’s ligand preminimization and postdocking all-atom
minimization. Standard parameters were used to estimate the binding
affinity characterized by the Surflex-Dock scores. In each docking
experiment, 20 poses were saved for each compound based on their total
docking score values. A higher score represents stronger binding affinity.
The optimal binding pose of the docked compounds was selected based
on the Surflex scores and visual inspection of the docked complexes.
The binding free energies of the complexes were calculated using the
MM/GBSA method. OPLS-AA force field and GB/SA continuum solvent model
were used to calculate the necessary energies of the complex.
Competitive
Binding Assay Using LC-MS/MS
A Thermo TSQ
Advantage instrument was used to determine the competitive binding
affinity of the pyridine-bridged CA-4 analogues by measuring the concentration
of colchicine displaced by the analogues. Briefly, colchicine (1.2
μM) was incubated with tubulin (1.3 mg/mL) in the incubation
buffer (80 mM PIPES, 2.0 mM MgCl2, 0.5 mM EGTA, pH 6.9)
at 37 °C for 1 h. Varying concentrations (0.1–125 μM)
of 4h, 4s, and 4t were used
to compete with colchicine originally bound to tubulin. After incubation,
the filtrate was obtained as previously described.[26] The ability of the analogue to inhibit the binding of colchicine
was expressed as a percentage of control binding in the absence of
any competitor.
Pharmacokinetic Study in Mice
Female
C57BL/6 mice were
used for the pharmacokinetic study of selected CA-4 analogues. Mice
were given oral gavage containing PBS and ethanol-dissolved CA-4, 4h, 4s, and 4t, at a single dose
of 5 mg/kg/mouse. After oral administration, blood samples were collected
from the orbital sinus of the mice at 1, 3, 6, and 24 h, with each
group of mice subjected to only one sampling. All procedures involving
these animals were conducted in compliance with State and Federal
laws, standards of the U.S. Department of Health and Human Services,
and guidelines established by Xavier University Animal Care and Use
Committee. The facilities and laboratory animals program of Xavier
University are accredited by the Association for the Assessment and
Accreditation of Laboratory Animal Care.
Authors: Keira Gaukroger; John A Hadfield; Nicholas J Lawrence; Steven Nolan; Alan T McGown Journal: Org Biomol Chem Date: 2003-09-07 Impact factor: 3.876
Authors: S Ducki; R Forrest; J A Hadfield; A Kendall; N J Lawrence; A T McGown; D Rennison Journal: Bioorg Med Chem Lett Date: 1998-05-05 Impact factor: 2.823
Authors: Afshin Dowlati; Kelly Robertson; Matthew Cooney; William P Petros; Michael Stratford; John Jesberger; Niusha Rafie; Beth Overmoyer; Vinit Makkar; Bruce Stambler; Anne Taylor; John Waas; Jonathan S Lewin; Keith R McCrae; Scot C Remick Journal: Cancer Res Date: 2002-06-15 Impact factor: 12.701
Authors: Gordon J S Rustin; Susan M Galbraith; Helen Anderson; Michael Stratford; Lisa K Folkes; Luiza Sena; Lindsey Gumbrell; Patricia M Price Journal: J Clin Oncol Date: 2003-06-13 Impact factor: 44.544
Authors: James P Stevenson; Mark Rosen; Weijing Sun; Maryann Gallagher; Daniel G Haller; David Vaughn; Bruce Giantonio; Ross Zimmer; William P Petros; Michael Stratford; David Chaplin; Scott L Young; Mitchell Schnall; Peter J O'Dwyer Journal: J Clin Oncol Date: 2003-12-01 Impact factor: 44.544
Authors: Zaki S Seddigi; M Shaheer Malik; A Prasanth Saraswati; Saleh A Ahmed; Ahmed O Babalghith; Hawazen A Lamfon; Ahmed Kamal Journal: Medchemcomm Date: 2017-07-04 Impact factor: 3.597
Authors: Daniel Tarade; Dennis Ma; Christopher Pignanelli; Fadi Mansour; Daniel Simard; Sean van den Berg; James Gauld; James McNulty; Siyaram Pandey Journal: PLoS One Date: 2017-03-02 Impact factor: 3.240
Figure 1
Scheme 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9