![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 53, Issue 5, 908-915, May 1998
School of Nursing (S.L.) and Department of Biochemistry, University of Mississippi Medical Center, Jackson, Mississippi 39216 (S.L., J.W.I., J.J.C.), and the Division de Cancerologie Experimentale, Centre de Recherche Pierre Fabre, 81106 Castres Cedex 06, France (B.T.H.)
 |
Summary |
|---|
Top
Summary
Introduction
Procedures
Results
Discussion
References
|
|---|
We present a comparison of the energetics of spiral formation for two
vinca alkaloids: a novel difluorinated vinorelbine derivative 20',20'-difluoro-3',4'-dihydrovinorelbine (F12158, or vinflunine) and
the parent compound, vinorelbine. Vinca alkaloids are antineoplastic agents that halt cell division at metaphase by inhibiting microtubule assembly and inducing tubulin self-association into spiral aggregates. The overall affinities for tubulin of vincristine, vinblastine, and
vinorelbine seem to correlate with their clinical doses, where vincristine with the highest overall affinity is used at the lowest doses. Doses of chemotherapeutic agents, however, also are determined by toxicities. In the physicochemical study described here, we used
sedimentation velocity to compare vinorelbine- and vinflunine-induced self-association of porcine brain tubulin in the presence of 50 µM GDP or 50 µM GTP. Vinflunine
demonstrates 3-16-fold lower overall affinity for tubulin and induces
smaller polymers compared with vinorelbine. Sedimentation velocity
provides the only direct evidence to date that vinflunine is a
tubulin-binding drug. Stopped-flow light scattering demonstrates the
shortest relaxation times for polymer redistribution for vinflunine
consistent with induction of the shortest spirals. Data collected at
5°, 15°, 25°, and 37° show increasing
20,w values with increasing temperature
and are consistent with an entropically driven process. These data are
entirely consistent with our hypothesis that vinflunine is likely to
result in reduced clinical neurotoxicity relative to vinorelbine,
vinblastine, and vincristine.
| |
Introduction |
|---|
|
Top
Summary
Introduction
Procedures
Results
Discussion
References
|
|---|
The
antineoplastic properties of vinca alkaloids arise from their
interaction with tubulin, the major component of microtubules in
mitotic spindles. These drugs diminish microtubule dynamics and
assembly, resulting in the arrest of cell division at metaphase. At
substoichiometric levels, in vitro, vinca alkaloids
stabilize microtubules, possibly by binding to microtubule ends and
inhibiting hydrolysis of GTP (Jordan and Wilson, 1990
; Jordan et
al., 1991; Toso et al., 1993). At higher
concentrations, microtubules depolymerize, and in the presence of MAPs
or at high magnesium concentrations (>2.5 mM), vinca
alkaloids induce large paracrystals made up of spiral helices of one or
two protofilaments (Fujiwara and Tilney, 1975; Amos et al.,
1984; Timasheff, 1986a, 1986b; Himes, 1991; Na and Nogales et
al., 1995).
Three vinca alkaloids that are currently important in anticancer
chemotherapy protocols, vincristine, vinblastine, and vinorelbine, are
used for treating a spectrum of solid and hematological tumors (Lobert
and Correia, 1992; Johnson et al., 1996). As antimitotic drugs, their affinity for tubulin has an impact on drug efficacy and
toxicity. The order of overall affinity for tubulin is vincristine > vinblastine > vinorelbine (Lobert et al., 1996).
This relative drug affinity for tubulin may be related to the clinical
doses used; generally, vincristine is administered at 2 mg/m2 and vinorelbine at 30 mg/m2 (Lobert and Correia, 1992; Chabner et
al., 1996). However, clinical doses also are based on
dose-limiting side effects. The clinical toxicity of vincristine is
mostly neurological, with predictable cumulative effects, whereas
vinorelbine is the least neurotoxic vinca alkaloid (Chabner et
al., 1996). Our hypothesis is that the overall drug affinity for
tubulin contributes to the severity of the neuropathies observed
clinically. In fact, this is currently the basis for clinical trials
with vinorelbine.
Vinca alkaloid binding is linked to tubulin self-association, and the
binding affinities can be determined by sedimentation velocity. These
data are best fit by an isodesmic ligand-mediated or ligand-mediated
plus ligand-facilitated model (Na and Timasheff, 1980, 1986a). The
major difference between the drugs is in
K2, the affinity of liganded heterodimers
for spirals, and in K3, the binding of drug
to polymers. All three drugs show an increase in maximum
20,w values with increasing
temperature, consistent with an entropically driven process. The
relative magnitudes of K2 were
independently confirmed by stopped-flow light scattering, in which
relaxation times, 
, are longest for vincristine polymers and
shortest for vinorelbine polymers (Lobert et al., 1996).
Furthermore, there is a decrease in with increasing tubulin
concentration, suggesting annealing of oligomers in addition to
association of liganded-heterodimers (Thusius et al., 1975;
Nogales et al., 1995; Lobert et al., 1996).
In the current work, we compared the energetics of tubulin spiral
formation in the presence of a novel difluorinated derivative of
vinorelbine, vinflunine (Fahy et al., 1997), with the parent compound using sedimentation velocity. We found that vinflunine induces
significantly smaller spirals than vinorelbine. The difference between
the two drugs is primarily in K2, the
affinity of liganded-heterodimers for spirals, and in
K3, the affinity of drug for polymers.
Van't Hoff analysis of data collected at 5°, 15°, and 25° shows
more positive 
H° and S° for vinflunine compared with
vinorelbine, consistent with an entropically driven process. The
increased H° corresponds to a less negative (less favorable)
G° for vinflunine. Stopped-flow light scattering was used to
investigate the mechanism of drug-induced spiral formation. Vinflunine
demonstrates shorter relaxation times compared with vinorelbine, and
the data are consistent with spiral formation occurring by the addition
of liganded heterodimers and the annealing of oligomers. These data
together suggest that vinflunine may demonstrate reduced neurotoxicity
relative to the other three vinca alkaloids. Our sedimentation velocity
studies are the only direct evidence that vinflunine is a
tubulin-binding drug and therefore contribute to understanding of its
potential as a clinically useful antimitotic agent.
| |
Experimental Procedures |
|---|
|
Top
Summary
Introduction
Procedures
Results
Discussion
References
|
|---|
Reagents. Deionized (Nanopure) water was used in all experiments. MgSO4, EGTA, GDP (type I), GTP (type II-S), and PIPES were purchased from Sigma Chemical (St. Louis, MO). Sephadex G-50 was from Pharmacia (Piscataway, NJ).
Tubulin purification.
Porcine brain tubulin (PC-tubulin)
free of MAPs was obtained by two cycles of warm/cold
polymerization/depolymerization followed by phosphocellulose
chromatography to separate tubulin from MAPs (Williams and Lee, 1982;
Correia et al., 1987). Protein concentrations were
determined spectrophotometrically (
278 = 1.2 liter/g·cm) (Detrich and Williams, 1978).
Sedimentation velocity experiments.
Vinorelbine- or
vinflunine-induced tubulin spiral formation was studied in the presence
of GDP or GTP by sedimentation velocity. Vinorelbine, as the tartrate,
was obtained from Pierre Fabre Medicaments (Gaillac, France), and
vinflunine sulfate was kindly provided by Dr. J. Fahy (Center de
Recherche Pierre Fabre, Castres, France). Tubulin samples (2 µM) were equilibrated in 10 mM PIPES, pH 6.9, 2 mM EGTA, 1 mM MgSO4,
and 50 µM GXP using spun columns (Lobert et
al., 1995). The free drug concentration (0.5-70 µM)
was obtained from the known drug concentration in the equilibration
buffer. For vinflunine, the extinction coefficients are
214, water = 61,650 M
1 cm1 and
268, water = 18,650 M1 cm1, and
for vinorelbine, they are 215, water = 57,800 M1
cm1 and 268, water = 17,150 M1
cm1 (Institut de Recherche Pierre Fabre). In
previous work (Lobert et al., 1996), stock clinical
solutions were purchased from Glaxo-Wellcome (Research Triangle Park,
NC). The stated concentration was interpreted as 10 mg/ml vinorelbine
ditartrate, although as noted recently, the solutions were actually 10 mg/ml vinorelbine. Thus, for comparison with the data collected here,
the vinorelbine concentrations from our previous work (Lobert et
al., 1996) were corrected by a factor of nearly 1.4. After
equilibration, the protein was brought to the desired final
concentration by dilution with the equilibration buffer. Sedimentation
studies were done in a Beckman Instruments (Palo Alto, CA) Optima XLA
analytical ultracentrifuge equipped with absorbance optics and an An60
Ti rotor. Samples were spun at 5°, 15°, 25°, or 37° at
appropriate speeds, and temperature was calibrated according to the
method of Liu and Stafford (1995). Velocity data were collected at 278 nm and at a spacing of 0.002 cm with one average in a continuous scan
mode. Data were analyzed using software (DCDT) provided by Dr. Walter
Stafford (Boston Biomedical Research Institute, Boston, MA) to generate
a distribution of sedimentation coefficients,
g(s) (Lobert et al., 1995).
Curve fitting of sedimentation velocity data.
The
distributions of sedimentation coefficients,
g(s), were converted to weight average
20,w values by integration of the curves (
s *
g(s)ds/g(s)ds) using
Origin 3.5 (Microcal Software, Northampton, MA) and correction for
temperature and buffer components. The data were plotted as weight
average 20,w values versus free drug. Total protein concentration for each sample was determined from
g(s)ds. Sedimentation data were fit using the isodesmic ligand-mediated or the isodesmic ligand-mediated plus
ligand-facilitated model (also referred to as the combined model)
(Lobert et al., 1995). In these models,
K1 is the affinity of drug for tubulin heterodimers, K2 is the affinity of
liganded heterodimers for spiral polymers,
K3 is the affinity of drug for polymers,
and K4 is the association constant for
unliganded-tubulin heterodimers. Binding constants and error estimates
were obtained by fitting with the nonlinear least-squares program
Fitall (MTR Software, Toronto, Canada), modified to include the
appropriate fitting functions. We find that in the combined model fit,
it is necessary to constrain K4, the
association constant for unliganded tubulin heterodimers, to 1 × 104 M1 to
compare the binding affinities because only three binding constants are
independent in this scheme. As with any linear and nonlinear
least-squares procedures, the fitted parameters exhibit a certain
degree of cross-correlation, and the overall product, K1K2, is probably
better determined than the individual parameters, K1 and K2.
However, the results demonstrate that K1 is
nearly a constant for both drugs, 9.75 ± 2.62 × 104 M1
(average for the ligand-mediated fits at 5°, 15°, and 25°), and that the major difference between drugs and the major effect of GDP is
in the value of K2. This is compelling
evidence that we are correctly determining reliable parameters.
Stopped-flow light-scattering experiments.
Stopped-flow
rapid mixing experiments (manual stopped-flow apparatus; Hi-Tech,
Wiltshire, England) were carried out as described previously (Lobert
et al., 1996). Briefly, PC-tubulin (0.6-4 µM) was equilibrated, using spun columns, into 10 mM PIPES, pH
6.9, 2 mM EGTA, 1 mM
MgSO4, 50 µM GTP, and 70 µM vinorelbine or vinflunine. The final drug and tubulin
concentrations were 35 and 0.3-2 µM, respectively.
Samples were degassed for 1 hr at room temperature. Stopped-flow
rapid-mixing experiments at 25° were initiated by diluting tubulin
samples 1:1 with the same buffer without drug. Relaxation was monitored
using an SLM Aminco Bowman Series 2 Luminescence Spectrometer
(Rochester, NY) at 350 nm (90-degree scattering) over 10 min.
Relaxation times, , were determined using the following equation:
x = a0exp(
t/), where
a0 is the change in amplitude associated with over time, t (sec). We used 
2
determinations to select the best fits of the data using Origin 3.5.
| |
Results |
|---|
|
Top
Summary
Introduction
Procedures
Results
Discussion
References
|
|---|
Vinorelbine- and vinflunine-induced tubulin spiral formation.
The association of 2 µM PC-tubulin (10 mM
PIPES, pH 6.9, 1 mM MgSO4, 2 mM EGTA, 50 µM GXP) in the presence of
vinorelbine or vinflunine (0.5-70 µM) was studied with
the use of sedimentation velocity at 5°, 15°, 25°, and 37°.
Fig. 1 shows weight average sedimentation
coefficients, 20,w, plotted versus
drug concentrations for both drugs at each temperature. We found that
increasing the temperature resulted in larger
20,w values for vinorelbine and vinflunine, although the maximum s values in the presence of
vinflunine are considerably smaller than those for vinorelbine. For
example, at 37° in the presence of GDP (Fig. 1a), the maximum
20,w values for vinorelbine are
~18 S, whereas for vinflunine, the maximum values are near 11 S. Data
were fit with the ligand-mediated model or ligand-mediated plus
ligand-facilitated (combined) model to obtain equilibrium constants
(Tables 1 and
2). Fits with either the ligand-mediated
model or combined model were indistinguishable in terms of the standard
deviation of the fits. At 5°, 15°, and 25°, vinorelbine interacts
with tubulin with 3-16-fold higher overall affinity,
K1K2, than
vinflunine, depending on the model used to determine the binding
parameters (Tables 1 and 2). Thus, the modifications involving the C20'
difluorination (IUPAC numbering system is used throughout) and the lack
of the 3',4' double bond in vinflunine may result in this decrease in
its overall affinity for tubulin.
|
|
|
30°, the 20 S shoulder becomes significant,
indicating the formation of higher order aggregates, probably due to
tubulin denaturation. At low drug concentrations, there is a small but
significant portion of the unliganded tubulin at 37° forming larger
aggregates that sediment faster than the liganded tubulin spirals.
Because of the aggregation of unliganded tubulin, we used only data
from 3-70 µM in fitting the 37° data, rather than from
0.5-70 µM (Tables 1 and 2). Even with this editing, the
ligand-mediated models may not describe the data collected at 37°,
and the binding parameters may be unreliable. A more detailed study of
high temperature tubulin denaturation will be presented elsewhere
(Vulevic B, Lobert S, and Correia JJ, manuscript in preparation).
The K2app is the
drug-dependent indefinite self-association constant for tubulin spiral
formation (Na and Timasheff, 1985). Because self-association is linked
to drug binding, it is a measure of the extent of overall drug binding.
Fig. 2 shows plots of
K2app versus free drug
concentrations for the data collected in the presence of GDP (Fig. 2a)
or GTP (Fig. 2b) and fit with the ligand-mediated model. The
K2app was calculated from
individual binding affinities, K1 and
K2, as described in the legend. It is clear
that significantly more spiral formation occurs with vinorelbine than
with vinflunine at all temperatures studied. According to the Wyman
(1964) linkage theory, this means that more vinorelbine binds to
tubulin than vinflunine under all comparable conditions.
|
1 and 6.5 ± 2.7 × 104 M1,
respectively. Note that our previous value for vinorelbine under similar conditions, 5° and 25°, was 1.1 ± 0.3 × 105 M1 (Lobert
et al., 1996). The low degree of association with these drugs potentially makes parameter estimation difficult. To verify the
reliability of K1 values, we carried out
titrations of 4 µM tubulin with vinorelbine and
vinflunine in the presence of GTP at 25° (Table 1). At this higher
protein concentration, the extent of association is enhanced, and in
principle, parameter estimation should be improved. We obtained nearly
identical K1 values compared with the 2 µM tubulin data, 1.1 ± 0.3 × 105 M1 and
7.2 ± 3.4 × 104
M1, averaged over both models for
vinorelbine and vinflunine (Table 1). For the 2 µM
tubulin data at 25°, the K1 values were
1.2 ± 0.2 × 105
M1 and 6.9 ± 2.8 × 104 M1. Thus,
within the error of these measurements, there is no difference in
K1 values for these two drugs.
The difference between vinorelbine- and vinflunine-induced spiral
formation at 5°, 15°, 25°, and 37° is primarily in
K2, liganded heterodimers binding to
spirals, when data are fit with either model (Tables 1 and 2). For
vinorelbine data fit with the ligand-mediated model,
K2 is 4-14-fold larger than that for
vinflunine. This amounts to 0.82-1.56 kcal/mol enhancement for
vinorelbine compared with vinflunine. For the combined model, the
K2 enhancement by vinorelbine is smaller,
only 2-7-fold (0.41-1.15 kcal/mol), whereas the enhancement in
K3, drug binding to polymers, is 3-16-fold
(0.65-1.64 kcal/mol). Thus, spiral formation and drug binding to
spirals are significantly reduced in the presence of vinflunine
compared with vinorelbine.
GDP enhancement of vinorelbine- and vinflunine-induced tubulin
spiral formation.
GDP enhances vinorelbine- and vinflunine-induced
tubulin self-association (Fig. 1 and Table
3). The mean GDP enhancement in
K1K2 for the data
collected at 5°, 15°, and 25° is equal to 0.84 ± 0.20 kcal/mol for data fit with both models (Table
3). This agrees well with our previous
report of a mean GDP enhancement of 0.90 ± 0.17 kcal/mol for
vinblastine, vincristine, and vinorelbine (Lobert et al.,
1996). When data are fit with the ligand-mediated model, the effect of
GDP occurs primarily on K2, the affinity for association of liganded heterodimers, rather than on
K1, the affinity for drug binding to the
heterodimer. For the combined model, the GDP enhancement occurs
primarily in K2 and
K3. This is consistent with our
expectations of Wyman linkage, where drug binding enhances
self-association and self-association enhances drug binding (Lobert
et al., 1995).
|
Comparison of vinorelbine and vinflunine thermodynamic
parameters.
Table 4 gives the
thermodynamic parameters obtained from Van't Hoff analysis of the data
collected at 5°, 15°, and 25°. As discussed, the equilibrium
constants determined from fits of the data at 37° are not reliable
and thus were not used in this analysis. The overall affinity,
K1K2, from
ligand-mediated fits was used to calculate G. Similar results were
obtained from combined model fits (data not shown). Plots of
lnK1K2 versus 1/T
were used to determine H (data not shown). The free energy, G°,
at 25°, for vinflunine binding in the presence of GTP was less
negative than that for vinorelbine binding: 14.2 and 15.2 kcal/mol,
respectively. In the presence of GDP, the values were 15.1 and
16.1, respectively. Thus, the smaller K2
and K1K2 values for
vinflunine-induced tubulin self-association are reflected in the less
negative G. The H° and S° values are consistent with an
entropically driven polymerization process, although the absolute
values are model dependent (Table 4). The unfavorable H° in
general is compensated by a positive S°, regardless of the model
used to fit the data. We showed previously that H° is larger for
drugs with smaller
K1K2 values (Lobert et al., 1996). Likewise in this case, H° is larger for
vinflunine compared with vinorelbine.
|
Stopped-flow light-scattering drug dilution.
To investigate
the kinetics of drug-induced tubulin association, we used rapid-mixing
stopped-flow light scattering (Fig. 3).
Data were best fit with single exponentials to obtain the relaxation
times given in Table 5. In this table, A1
represents the change in the light-scattering signal, and is the
relaxation time in seconds. It can be seen that the relaxation times
for vinflunine are slightly shorter than those for vinorelbine (range, 2.14-20.90 versus 10.68-25.25 sec, respectively). This is consistent with a model that involves a cascade of dissociation events from larger
to smaller polymers on dilution. The larger polymers for vinorelbine
versus vinflunine (Fig. 1) are consistent with these longer average
relaxation events for vinorelbine. Furthermore, the relaxation time
decreases with increasing tubulin concentration (Table 5). This has
been observed previously in the relaxation data for vinblastine (Lobert
et al., 1996) and in the synchrotron data of Nogales
et al. (1995). It suggests that in addition to liganded
heterodimers adding to or dissociating from polymers, oligomers can
self-associate or dissociate in a mechanism consistent with annealing
of spiral polymers.
|
|
2) were plotted against the final tubulin
concentrations (Table 6). Vinorelbine has
a smaller off rate, kd, than
vinflunine: 0.043 and 0.185 sec1, respectively.
However, the binding affinities, determined from the ratio
ka/kd,
are very similar (probably due to larger error in the light-scattering
method): 5.6 × 105
M1 and 5.4 × 10 5 M1 for
vinorelbine and vinflunine, respectively. Note that the
K2 values determined from
sedimentation velocity data at 25°, fit with the ligand-mediated
model, are reasonably consistent with this result: 1.1 × 106
M1 and 3.0 × 105
M1 for vinorelbine and
vinflunine, respectively. Thus, we conclude that the mechanism of
spiral formation for these two drugs is identical, meaning that spiral
formation occurs by addition of heterodimers and oligomers.
|
| |
Discussion |
|---|
|
Top
Summary
Introduction
Procedures
Results
Discussion
References
|
|---|
Mechanism of vinca alkaloid-induced tubulin association.
Stopped-flow light-scattering data demonstrated that vinorelbine,
vinblastine, and vincristine-induced tubulin self-association occurs by
a similar mechanism for all three drugs (Lobert et al., 1996). The relaxation data can be analyzed according to the theory developed by Thusius et al. (1975). They showed for
glutamate dehydrogenase that relaxation times decrease with increasing
protein concentration, indicating that self-association occurs by both association of single protein subunits to the ends of polymers and by
association of oligomers. We propose that for all four vinca alkaloids
investigated, vinorelbine, vinblastine, vincristine, and vinflunine,
oligomer annealing can occur, in addition to liganded heterodimers
adding to the ends of spirals. This is consistent with our proposed
mechanism of microtubule inhibition in which spirals anneal to the ends
of microtubules and suppress dynamics (Lobert et al., 1995,
1996, 1997). Within error, the binding affinity, Kapp, values for vinorelbine and vinflunine
determined by light-scattering kinetics are reasonably consistent with
our sedimentation velocity data. Note that if light-scattering data
collected here for vinorelbine are combined with previous
light-scattering data collected for vinorelbine (Lobert et
al., 1996), the estimated Kapp is
9.4 × 105
M1 rather than 5.6 × 105 M1 as
reported here in Table 6. These combined data give a value closer to
the K2 value estimated on the basis of
sedimentation velocity at 25° (Table 1;
K2 = 1.1 ± 0.8 × 106 M1).
These Kapp values correspond to a
G°, Gvinorelbine Gvinflunine, of 0.33 kcal/mol. In other words, with light scattering measurements, spiral
formation induced by vinorelbine is favored by 0.33 kcal/mol relative
to vinflunine.
). Over a 2-hr period, vincristine and vinblastine continue to inhibit alkylation of sulfhydryls, whereas
in the presence of vinorelbine or vinflunine, alkylation is not
inhibited. If tubulin in spirals is a poor substrate for alkylation,
then the overall affinity for spiral formation should be a measure of
inhibition of alkylation. Furthermore, smaller spirals will exchange
tubulin heterodimers more readily. Relaxation times for
vinorelbine-induced spirals are 2-3-fold shorter than those for
vinblastine-induced spirals (Lobert et al., 1996).
Relaxation times for vincristine-induced spirals are several-fold
longer than those for vinblastine-induced spirals. We report here that exchange of liganded heterodimers in vinflunine-induced tubulin spirals
is even more rapid than that in vinorelbine-induced spirals. Thus, in
the presence of either vinorelbine or vinflunine, tubulin heterodimers
are more readily exposed to alkylation than when vinblastine or
vincristine is present. This differential exposure of sulfhydryls can
occur even though the conformations of the drug-induced spirals are
similar.
Analysis of the sedimentation velocity data in terms of overall binding
affinity, K1K2, has
shown that the order of vincristine > vinblastine > vinorelbine (Lobert et al., 1996). For data fit with the
ligand-mediated model, the difference
Gvincristine Gvinblastine, or G°, is 0.68 kcal/mol.
Comparing the same model fits for vinblastine minus the data reported
here for vinorelbine, we find that
Gvinblastine Gvinorelbine, G°, is 0.88 kcal/mol. The difference between vinorelbine and vinflunine,
Gvinorelbine Gvinflunine, amounts to a G° value of
1.00 kcal/mol. Data fit with the combined model show the same trends,
but the magnitudes of the differences are somewhat smaller. Thus,
vinorelbine-induced tubulin self-association is favored relative to
vinflunine as demonstrated by sedimentation velocity data and supported
by the kinetic data.
When individual parameters are examined, we find that the binding
affinity of vinorelbine, vinblastine, or vincristine for tubulin
heterodimers, K1, is identical within error
for all three drugs, 1.6 ± 0.3 × 105
M1 at 25° (Lobert et
al., 1996). For the vinorelbine and vinflunine data reported here
at 25°, the mean K1 value is 9.6 ± 2.7 × 104
M1. Thus, there is no significant
difference in the affinity of tubulin heterodimers for any of the vinca
alkaloids studied. As found with vinorelbine, vinblastine, and
vincristine, the major differences between vinorelbine and vinflunine
are in the affinity of liganded heterodimers for polymers,
K2, and in the affinity of drug for
polymers, K3.
GDP enhances vincristine-, vinblastine-, and vinorelbine-induced
tubulin self-assembly relative to GTP by 0.90 ± 0.17 kcal/mol (Lobert et al., 1996). In the work described here, we find
the same GDP enhancement of vinorelbine- and vinflunine-induced tubulin spiral formation, 0.84 ± 0.20 kcal/mol. The GDP enhancement
occurs primarily in K2 and
K3. It seems that GDP enhancement must be due to an intrinsic allosteric feature of the tubulin structure. The
GDP enhancement of vinca alkaloid-induced tubulin self-association observed now with four vinca alkaloids, vincristine, vinblastine, vinorelbine, and vinflunine suggests that spirals may propagate into
the microtubule core by a zipper-like or protofilament-peeling mechanism (Himes, 1991; Lobert et al., 1995). These results
indicate that a destabilizing effect on the GDP core of microtubules is a common mechanism in vinca alkaloid chemotherapeutics. Our kinetic results indicate that spirals or liganded heterodimers could associate with microtubule ends, suppressing dynamics. All four vinca alkaloids have the same energetic potential for propagation of spirals into the
microtubule core and disrupting lateral interactions. Thus, a
nucleotide-dependent allosteric effect is exhibited by all vinca alkaloids, such that drug binding induces the same cooperative depolymerization of mitotic spindles.
Implications for in vivo chemotherapeutic
effects.
Biosynthesis of vinca alkaloids involves a coupling
reaction of the two precursors, catharanthine and vindoline. During
this reaction, catharanthine undergoes a rearrangement to the
cleavamine skeleton. This structure is not directly comparable to that
of catharanthine. However, Prakash and Timasheff (1991) found that catharanthine will inhibit microtubule polymerization in
vitro and induce tubulin self-association, although the magnitude
of the effect is less than the "dimeric" vinblastine or
vincristine. Borman and Kuehne (1989) showed that modification of
vinblastine at the C4' position alters the drug activity in
vitro and in vivo. Vinorelbine is a vinblastine
derivative modified at C4' with an eight- instead of the nine-membered
C' ring of the natural compounds. It was synthesized in the late 1970s
(Mangeney et al., 1979) and shown to have a different
spectrum of clinical activity from vincristine and vinblastine and an
improved toxicity profile in cancer patients (Johnson et
al., 1996). Our biophysical studies demonstrate that vinorelbine
binds to tubulin with much weaker overall affinity, K1K2, than
vinblastine or vincristine, resulting in the formation of smaller
spirals. The weaker binding is not in the drug binding to tubulin
heterodimers but rather in the affinity of liganded heterodimers for
spiral polymers. This results in the formation of smaller spirals. The
results presented here demonstrate that vinflunine is a tubulin-binding
drug, inducing smaller spirals than vinorelbine. These data suggest
that like other vinca alkaloids, the antineoplastic effects of
vinflunine are a direct result of its interaction with mitotic
spindles, most likely in the form of small liganded spiral polymers. We
want to stress that this a novel way of thinking about the mode of
action of this class of antineoplastic agents.
, 1987; Gout et
al., 1984; Ferguson and Cass, 1985; Mullin et al., 1985; Sirotnak et al., 1986). It is very likely that
drug-induced tubulin spiraling contributes to drug retention and
cytotoxicity. Larger spirals should result in longer intracellular drug
retention, as found with vincristine compared with vinblastine
(Houghton et al., 1987). We showed that for vinorelbine,
vinblastine, and vincristine, drug spiraling correlates with clinical
doses and toxicity (Lobert et al., 1996). Most recently,
vinflunine, a difluorinated derivative of vinorelbine, modified
uniquely at the C20' position with a reduction of the 3',4' double bond
on the cleavamine moiety (Fahy et al., 1997) has
demonstrated preclinical antineoplastic activity. Definite in
vivo antitumor activity has been demonstrated (Kruczynski et
al., 1998b) and in vitro mitotic arresting and tubulin
interacting properties have been identified (Barret et al.,
1997). Specifically, the four vincas differentially interfere with the
binding of tritiated vincas to tubulin as follows: vincristine > vinblastine > vinorelbine > vinflunine. Consistent with
these findings, we have shown in the studies described here that
vinflunine binds with several-fold lower overall affinity to tubulin
than vinorelbine and, consequently, induces smaller spirals with a more
rapid relaxation time. Thus, according to our hypothesis, we predict
that vinflunine will have reduced toxicity compared with other vinca
alkaloids.
There are conflicting reports comparing in vitro drug
binding affinities and in vivo cytotoxicity. Plasma drug
levels at steady state reach nanomolar concentrations. Furthermore,
relative to extracellular levels, vinca alkaloids accumulate in cells
several-fold to 100-fold depending on the cell type (Gout et
al., 1984; Jordan et al., 1991). Intracellular drug
concentrations remain substoichiometric relative to the intracellular
tubulin, which is estimated to be near 20 µM (Jordan
et al., 1991). One study indicated that although vindesine
binds to tubulin with lower affinity than vincristine, it has greater
efficacy when B16 melanoma cells are exposed for 2 hr and transplanted
into mice (Jordan et al., 1985). A later study demonstrated
comparable cytotoxic effects in B16 melanoma cells for vincristine and
vindesine after 43-hr exposure (Singer and Himes, 1992). In the same
study, vinepidine was found to be less cytotoxic than vincristine,
although its Kapp value was larger. However, vinepidine uptake into cells was by far the lowest after 90-min exposure. Clearly, duration of exposure of cells to vincas is an
important contributor to cytotoxicity. It should be noted that in
cancer patients, the plasma terminal half-lives of these drugs are
1-3.5 days. Therefore, tissues are exposed to drugs over a period of
several days. In fact, in animal studies, vinepidine plasma levels have
not reached plateau even at 72 hr, and its cytotoxicity falls between
vinblastine and vincristine (Mullin et al., 1985). We now
have data showing that the overall affinity of vinepidine for tubulin
falls between vincristine and vinblastine (Lobert and Correia, 1997).
Our studies reported here suggest that the formation of larger spirals
results in relatively longer retention in cells and tissues and greater
potential for toxicity, as observed with vincristine and vinepidine.
This does not exclude the importance of other parameters that
contribute to pharmacokinetics (e.g., lipid solubility and drug uptake
into tissues and cells). Additional factors that lead to prolonging the
drug terminal half-life may also contribute to toxicity. We assert that
all of these factors are important for a quantitative understanding of
the time course for cellular and clinical toxicity.
| |
Acknowledgments |
|---|
We wish to thank Drs. Jacques Fahy and Jean-Marc Barret (Center de Recherche Pierre Fabre, Castres, France) for critical reading and helpful discussions during the preparation of this manuscript. We are grateful to the University of Mississippi Analytical Ultracentrifuge Facility and the Institut de Recherche Pierre Fabre for their support. Finally, we thank Pelahatchie Country Meat Packers for providing pig heads for tubulin purification. This is UMC AUF publication 0013.
| |
Footnotes |
|---|
Received November 11, 1997; Accepted February 2, 1998
This work supported by National Institutes of Health Grant NR00056 (S.L.) and by Institut de Recherche Pierre Fabre.
Send reprint requests to: Dr. Sharon Lobert, University of Mississippi Medical Center, School of Nursing and Department of Biochemistry, 2500 N. State St., Jackson, MS 39216. E-mail: slobert{at}fiona.umsmed.edu.
| |
Abbreviations |
|---|
EGTA, ethylene glycol bis(
-aminoethyl
ether)-N,N,N',N'-tetraacetic
acid;
GXP, GTP or GDP;
MAP, microtubule associated protein;
PC-tubulin, MAP-free phosphocellulose purified tubulin;
PIPES, piperazine-N,N'-bis(2-ethanesulfonic
acid).
| |
References |
|---|
|
Top
Summary
Introduction
Procedures
Results
Discussion
References
|
|---|
dimer of tubulin from bovine brain.
Biochemistry
17:
3900-3907[Medline].
growth-inhibitory effects correlate with their retention by cultured Nb2 node lymphoma cells.
Eur J Cancer
34:
245-248.-class III tubulin: the role of nucleotides and -tubulin isotypes.
Biochemistry
34:
8050-8060[Medline].This article has been cited by other articles:
|
|
 |
|
  J. Bennouna, J.-P. Delord, M. Campone, and L. Nguyen Vinflunine: A New Microtubule Inhibitor Agent Clin. Cancer Res., March 15, 2008; 14(6): 1625 - 1632. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. C. Talbot, J. Margery, G. Dabouis, G. Dark, H. Taylor, H. Boussemart, V. Cadic, M.-C. Pinel, A. Riviere, L. Ollivier, et al. Phase II Study of Vinflunine in Malignant Pleural Mesothelioma J. Clin. Oncol., October 20, 2007; 25(30): 4751 - 4756. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Pourroy, M. Carre, S. Honore, V. Bourgarel-Rey, A. Kruczynski, C. Briand, and D. Braguer Low Concentrations of Vinflunine Induce Apoptosis in Human SK-N-SH Neuroblastoma Cells through a Postmitotic G1 Arrest and a Mitochondrial Pathway Mol. Pharmacol., September 1, 2004; 66(3): 580 - 591. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. A. Horner, S. Gould, J. P. Noakes, N. J. Rattray, S. L. Allen, E. Zotova, and J. C. Arezzo Lack of neurotoxicity of the vascular targeting agent ZD6126 following repeated i.v. dosing in the rat Mol. Cancer Ther., July 1, 2004; 3(7): 783 - 791. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Okouneva, B. T. Hill, L. Wilson, and M. A. Jordan The Effects of Vinflunine, Vinorelbine, and Vinblastine on Centromere Dynamics Mol. Cancer Ther., May 1, 2003; 2(5): 427 - 436. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Bennouna, P. Fumoleau, J.-P. Armand, E. Raymond, M. Campone, F.-M. Delgado, C. Puozzo, and M. Marty Phase I and pharmacokinetic study of the new vinca alkaloid vinflunine administered as a 10-min infusion every 3 weeks in patients with advanced solid tumours Ann. Onc., April 1, 2003; 14(4): 630 - 637. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. K. Ngan, K. Bellman, B. T. Hill, L. Wilson, and M. A. Jordan Mechanism of Mitotic Block and Inhibition of Cell Proliferation by the Semisynthetic Vinca Alkaloids Vinorelbine and Its Newer Derivative Vinflunine Mol. Pharmacol., July 1, 2001; 60(1): 225 - 232. [Abstract] [Full Text]
|
|
|
|
 |
|
 V. K. Ngan, K. Bellman, D. Panda, B. T. Hill, M. A. Jordan, and L. Wilson Novel Actions of the Antitumor Drugs Vinflunine and Vinorelbine on Microtubules Cancer Res., September 1, 2000; 60(18): 5045 - 5051. [Abstract] [Full Text]
|
|
|
|
|
|
S. Lobert, J. W. Ingram, and J. J. Correia Additivity of Dilantin and Vinblastine Inhibitory Effects on Microtubule Assembly Cancer Res., October 1, 1999; 59(19): 4816 - 4822. [Abstract] [Full Text] [PDF]
|
|
, 
) or vinflunine
(
, 
) at 37° (A), 25° (B), 15° (C), or 5° (D). Tubulin in
all experiments was 2 µM in the presence of 50 µM GTP (open symbols) or GDP
(closed symbols). Solid lines, fits using
the ligand-mediated model. The equilibrium constants obtained from
these fits are given in Tables 
