Introduction

Top Abstract Introduction Results Discussion References

Dolastatin 10 is a potently cytotoxic peptide, with four of its five subunits derived from modified amino acids. It was originally isolated from the sea hare Dolabella auricularia (Pettit et al., 1987), a shell-less mollusk, but it was subsequently totally synthesized (Pettit et al., 1989). Dolastatin 10 causes the accumulation of cells arrested in mitosis and the disappearance of cellular microtubules (Bai et al., 1990b, 1992). The apparent target is the protein tubulin, the major component of microtubules. Characteristics of its interaction with tubulin include a potent inhibition of assembly (IC₅₀ value, 1.2 µM), noncompetitive inhibition of the binding of vinblastine to tubulin, and inhibition of nucleotide exchange (Bai et al., 1990a,b). Studies with [³H]dolastatin 10 showed tight binding of the peptide to tubulin, but attempts to demonstrate formation of a tubulin-dolastatin 10 complex were unsuccessful. The smallest distinct species at low drug concentrations seemed to contain two -tubulin molecules and probably two dolastatin 10 molecules. At higher drug concentrations, massive polymers of aberrant morphology (probably tightly coiled spirals) form (Bai et al., 1995).

Dolastatin 10 inhibits the growth of various cancer cell lines at subnanomolar concentrations (Pettit et al., 1987; Bai et al., 1990b; Beckwith et al., 1993; Turner et al., 1998), and the drug is active against human sarcoma, melanoma, and ovarian xenografts in nude mice (Pettit et al., 1987; Waud et al., 1993; Turner et al., 1998). In phase I studies in human subjects, the dose-limiting toxicity of dolastatin 10 was against bone marrow (Bagniewski et al., 1997), and the drug is currently in phase II clinical trials.

In our previous comparative studies of dolastatin 10 and vinblastine, we found the drugs had similar potencies as inhibitors of tubulin assembly, but the peptide was 20- to 50-fold more potent than vinblastine as an inhibitor of cell growth. The experiments described in this study were undertaken to gain insight into possible reasons for this difference. The most obvious possibility is a major difference in cellular accumulation of the two drugs.

Vinblastine and other vinca alkaloids have been shown to accumulate in cells (Lengsfeld et al., 1982; Gout et al., 1984; Ferguson and Cass, 1985; Jordan et al., 1991; Van Belle et al., 1991; Pierre et al., 1992; Singer and Himes, 1992; Breier et al., 1994; Dhamodharan et al., 1995; El Hafny et al., 1997), and studies with both vincristine (Breier et al., 1994) and vinblastine (Ferguson and Cass, 1985; Van Belle et al., 1991) have indicated that accumulation of these drugs is by passive diffusion. Singer and Himes (1992) did note a close correlation among vinca alkaloids between efficiency of drug accumulation and degree of cytotoxicity. Moreover, Singer and Himes (1992) extrapolated their data to the IC₅₀ values, concluding that at these initial extracellular concentrations the intracellular drug concentrations should be far lower than the total cellular tubulin concentration.

Our strategy was to directly examine drug effects primarily at the IC₅₀ values. We have thus compared uptake and accumulation of [³H]vinblastine and [³H]dolastatin 10 at equitoxic concentrations. Efflux studies, however, were performed at the IC₅₀ value for vinblastine. We found a good correlation between drug potency and the extent of intracellular drug accumulation. Thus, a substantially higher proportion of dolastatin 10 compared with vinblastine accumulates in cells. This results primarily from apparent entrapment of the peptide within cells, whereas substantial efflux of vinblastine occurs.

Materials and Methods

Materials. [³H]dolastatin 10 (5.4 Ci/mmol) was prepared as described previously (Bai et al., 1995), and [³H]vinblastine (15.5 Ci/mmol) was obtained from Amersham (Arlington Heights, IL). The methanol solvent for both drugs was removed by evaporation at room temperature under a gentle stream of air, and the residues were dissolved in dimethyl sulfoxide (stock solutions at 10 µM). Monoclonal antibody C219 against P-glycoprotein p170 (p-gp 170) was generously provided by Dr. M. S. Poruchynsky, National Cancer Institute. Horseradish peroxidase-conjugated goat anti-mouse IgG was obtained from Pierce (Rockford, IL). Human Burkitt lymphoma CA46 cells were obtained from the American Type Culture Collection (Manassas, VA). These cells were used in all experiments, except as noted. A cellular volume of 0.651 pl (Magrath et al., 1980) was used to calculate intracellular concentrations, with no correction made for nuclear volume. For analysis of p-gp 170 content in CA46 cells, two additional Burkitt lines were used as controls. These were the parental EW-36 line (Cherney et al., 1997) and the derived multidrug-resistant (MDR) line EW-36VCR60 (Mickley et al., 1998). Both lines were generously provided by Dr. T. Fojo, National Cancer Institute.

Cell Culture and Drug Transport Studies. In all experiments, cells were grown in suspension culture in RPMI-1640 medium supplemented with 10% fetal calf serum, 2 µM L-glutamine, 10 µg/ml gentamicin sulfate, and 0.1% (v/v) dimethyl sulfoxide (solvent for all drugs). Incubation was at 37° in a humidified 5% CO₂ atmosphere. For determination of IC₅₀ values, an increase in cell number after 24 h was the parameter measured. In these studies, the initial inoculum was 3.5 to 4.0 × 10⁵ cells/ml. Cell were grown in 25-cm² flasks containing 5 ml of medium.

Cellular Tubulin Content. Cells were harvested by centrifugation, washed with PBS, and lysed for 30 min at 4° in a buffer containing 0.1 M 4-morpholineethanesulfonate (pH 6.6 with NaOH in 1 M stock solution), 20 µg/ml aprotinin, 10 µg/ml leupeptin, 100 µg/ml phenylmethylsulfonyl fluoride, and 1% (v/v) NP-40. Extracts were centrifuged in an Eppendorf 5417C centrifuge at 14,000 rpm for 1 min at room temperature. The protein content of the supernatants was determined by the Lowry procedure. Tubulin in these extracts was quantitated with a competitive enzyme-linked immunosorbent assay method (Thrower et al., 1991). Purified bovine brain tubulin (Hamel and Lin, 1984) was used to coat 96-well, high-binding capacity microplates and was also used as the tubulin standard. For the coating procedure, 100 µl of a tubulin solution at 6 µg/ml was placed in each well, and the plate was left at room temperature for 1 h. For the generation of the standard tubulin curve, a tubulin solution at 0.2 mg/ml was used.

Cellular p-gp 170 Content. The Western blot technique was used to measure cellular p-gp 170 levels. Cell extracts were prepared as described for measuring tubulin content, except that the lysis buffer was the P-MER buffer obtained from Pierce supplemented with the protease inhibitor cocktail "Complete" obtained from and used as recommended by Boehringer-Mannheim. Extract proteins (40 µg/lane) were electrophoresed on 8% acrylamide Tris-glycine polyacrylamide gels obtained from Novex (pH 8.6). Proteins were electrotransferred to a polyvinylidene difluoride membrane. The membrane was soaked for 1 h at room temperature in a solution containing Tris-buffered saline (25 mM Tris-HCl at pH 7.4, 137 mM NaCl, and 3 mM KCl; from Quality Biological), 0.1% Tween 20, and 5% dry milk (Solution A), followed by a brief wash in Tris-buffered saline containing 0.1% Tween 20 (Solution B). The membrane was incubated overnight at 4° in the primary C219 antibody directed against p-gp 170 (the stock antibody solution was diluted 1/5,000 in Solution A). The membrane was washed three times for 5 min in Solution B. The membrane was then incubated for 1 h at room temperature in Solution B containing horseradish peroxidase-conjugated secondary antibody (goat antimurine IgG) at a 1/50,000 dilution. The membrane was washed three times in Solution B for 5 min. The membrane was placed in chemiluminescent SuperSignal West Dura Extended Duration Signal solution for horseradish peroxidase from Pierce for 5 min at room temperature. The membrane was exposed for 30 sec to Biomax-ML film from Kodak, as recommended by Pierce.

	Results

Top Abstract Introduction Results Discussion References

Inhibition of CA46 Burkitt Lymphoma Cell Growth. In our initial studies we determined effects of the radiolabeled preparations of dolastatin 10 and vinblastine on the growth of CA46 Burkitt lymphoma cells. The [³H]dolastatin 10 was 20-fold more potent than the [³H]vinblastine. IC₅₀ values for the two drugs were 50 ± 10 (S.D.; n = 4) pM and 1.0 ± 0.3 (S.D.; n = 2) nM, respectively, for inhibition of growth after a 24-h incubation. The values obtained for the nonradiolabeled drugs were identical, within experimental error.

Cellular Accumulation of [³H]dolastatin 10 and [³H]vinblastine. Our initial drug uptake study was measurement of intracellular drug content after 24 h of growth at the IC₅₀ values (Table 1). Despite the 20-fold greater addition of vinblastine to the medium, virtually identical amounts of [³H]dolastatin 10 and [³H]vinblastine accumulated in the cells. Dolastatin 10 accumulated 19-fold more extensively than vinblastine, with only 3% of added vinblastine entering cells compared with 68% of the added dolastatin 10. This finding supports the idea that both drugs have the same intracellular target, assumedly tubulin (see Discussion), consistent with abundant morphological (mitotic arrest, disappearance of microtubules) and biochemical (effects on tubulin assembly etc.) observations.

Uptake Kinetics. On observing such extensive uptake of dolastatin 10 by the Burkitt cells, we hypothesized that it must have resulted from an active transport process, especially considering that the initial extracellular concentration was so low (50 pM). We also thought that the uptake of dolastatin 10 would be rapid. Studies of uptake kinetics and with cells depleted of ATP disproved both assumptions.

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Uptake of dolastatin 10 () and vinblastine () over 24 h in Burkitt cells treated at the IC50 value for each drug. The initial media concentrations were 50 pM for [3H]dolastatin 10 and 1.0 nM for [3H]vinblastine. The dolastatin 10 experiment was performed twice and the vinblastine experiment three times (standard deviations indicated).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Uptake of dolastatin 10 () and vinblastine () over 2 h in Burkitt cells treated at the IC50 value for each drug. The experiment was performed twice (standard deviations indicated).

Dolastatin 10 Uptake Is Not an Active Transport Process. Because the initial rate of dolastatin 10 uptake versus that of vinblastine seemed to reflect primarily the initial extracellular drug concentration, it seemed unlikely that an active transport process accounted for the more extensive intracellular accumulation of the peptide. Additional confirmation was obtained by examination of dolastatin 10 and vinblastine uptake at 1.0 nM and 50 pM, respectively, for comparison with the previous studies with dolastatin 10 at 50 pM and vinblastine at 1.0 nM. The data at the two concentrations of both drugs are presented in Fig. 3 as percentage of the added drug that enters cells over the first 2 h of incubation. For the first 15- to 20-min equivalent percentages of both drugs at both concentrations entered the cells, consistent with a diffusion-controlled mechanism. After 20 min, however, net additional accumulation of vinblastine was minimal, whereas that of dolastatin 10, at both concentrations, continued at virtually the same rate until the last time point.

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Proportion of dolastatin 10 and vinblastine taken up over 2 h in Burkitt cells relative to the initial drug concentration in the medium. Symbols as follows: , 50 pM [3H]dolastatin 10; , 1.0 nM [3H]dolastatin 10; , 50 pM [3H]vinblastine; , 1.0 nM [3H]vinblastine. The experiments were performed twice, and the 50 pM dolastatin 10 data and 1.0 nM vinblastine data are the same data presented in Fig. 2.

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Initial uptake rates of [3H]dolastatin 10 () and [3H]vinblastine () by Burkitt cells as a function of initial drug concentration in the medium. For each data point the indicated drug concentration was added to cells suspended in medium and rapidly mixed without disturbing the silicone oil cushion, and the tube was centrifuged. Cells and drug in the medium were in contact for 60-90 sec. The experiment was performed three times (standard deviations indicated).

View larger version (14K): [in this window] [in a new window]   Fig. 5.   ATP depletion has negligible effects on dolastatin 10 and vinblastine uptake by Burkitt cells. The medium contained the following: , 1.0 nM [3H]dolastatin 10; , 1.0 nM [3H]vinblastine; , 1.0 nM [3H]dolastatin 10, NaN3, and 2-deoxy-D-glucose; or , 1.0 nM [3H]vinblastine, NaN3, and 2-deoxy-D-glucose. The data without NaN3 and 2-deoxy-D-glucose are the same as shown in Figs. 2 (vinblastine) and 3 (dolastatin 10), and the experiment with NaN3 and 2-deoxy-D-glucose was performed three times. Standard deviations are indicated for all data.

Efflux Kinetics. Cells were pretreated for 1 h with 1.0 nM [³H]dolastatin 10 or [³H]vinblastine, and drug efflux was followed (Fig. 6). The higher dolastatin 10 concentration was used to obtain sufficient radiolabel to follow in the efflux phase of the experiment, and we preferred a short treatment time to minimize effects because of poor viability of the cells. No efflux of dolastatin 10 was detected over the 2 h of the experiment, whereas almost 90% of the intracellular vinblastine was lost over the same period. After an initial lag of approximately 10 min, vinblastine efflux seemed to follow first-order kinetics for approximately 1 h, with a half-life of approximately 10 min. A similar rapid efflux of vinblastine has been reported for many cell lines (Lengsfeld et al., 1982; Gout et al., 1984; Ferguson and Cass, 1985; Pierre et al., 1992; El Hafny et al., 1997).

View larger version (13K): [in this window] [in a new window]   Fig. 6.   Dolastatin 10, but not vinblastine, is tenaciously retained by Burkitt cells. Cells were preloaded with radiolabeled drug during a 1-h incubation with either 1.0 nM [3H]dolastatin 10 () or 1.0 nM [3H]vinblastine (). Cells were harvested by centrifugation and resuspended in media without drug. At the indicated times, cells in aliquots of each reaction mixture were centrifuged through oil. The experiment was performed three times (standard deviations indicated).

View larger version (22K): [in this window] [in a new window]   Fig. 7.   Expression of p-gp 170 in CA46, EW-36, and EW-36VCR60 Burkitt cell lines. Equal amounts (40 µg) of total cell extract from each cell line was loaded onto the polyacrylamide gel. Western blot analysis for p-gp 170 was performed with the monoclonal antibody C219 as described in detail in the text.

	Discussion

Top Abstract Introduction Results Discussion References

We found that dolastatin 10 accumulated 15- to 19-fold more extensively than vinblastine at equitoxic concentrations (the IC₅₀ values) in Burkitt lymphoma cells and therefore conclude that the 20-fold greater cytotoxicity of dolastatin 10 results from its more extensive intracellular accumulation. Because ATP depletion did not greatly affect uptake of either drug, their entry into cells probably results from passive diffusion through the plasma membrane. Supporting this, we found that the octanol/water partition coefficients for dolastatin 10 (1.4) and vinblastine (1.3) at physiological pH were nearly identical.

The major difference we observed between the drugs was a rapid efflux of vinblastine (Lengsfeld et al., 1982; Gout et al., 1984; Ferguson and Cass, 1985; Pierre et al., 1992; El Hafny et al., 1997) in contrast to no detectable efflux of dolastatin 10. Vinblastine efflux, additionally, was reduced under conditions of ATP depletion, suggesting involvement of an active mechanism. Verapamil had little effect on intracellular vinblastine, and p-gp 170 was not detected in the CA46 cells. Therefore, another unknown transport mechanism seems to be responsible for the apparent active efflux of vinblastine that we observed.

Dolastatin 10 and vinblastine, at their IC₅₀ values, reached similar intracellular concentrations after 24 h, supporting the conclusion from morphological and biochemical findings that their intracellular target is tubulin. To define this more quantitatively, we measured total cellular protein in the Burkitt cells, obtaining a value of 62 pg of protein/cell, and determined that tubulin was 5.0% of cellular protein. Moreover, there was no decline in tubulin levels during 24 h of treatment with either vinblastine, in agreement with Jordan et al. (1991), or dolastatin 10 at the IC₅₀ values. Taking cell volume as 0.651 pl (Magrath et al., 1980) and ignoring the nucleus, CA46 cells contain 48 µM total tubulin, so that approximately one molecule of dolastatin 10 or vinblastine for 500-1000 tubulin -dimers inhibits cell growth by 50%. Expressed another way, assuming 2/3 of tubulin is in polymer, the average microtubule is 5 µm in length (the cell radius), and each microtubule is anchored at the microtubule-organizing center, the concentration of microtubule (+)-ends in the untreated cell would be ~4 nM and tubulin at (+)-ends (13 protofilaments) ~50 nM. This is similar to intracellular concentrations of the drugs (50-100 nM) that we observed at the IC₅₀ values and consistent with the hypothesis that interference with dynamic instability of microtubules rather than disassembly is primarily responsible for the inhibitory effects of microtubule depolymerizing agents on cell growth (Wilson and Jordan, 1995). Furthermore, effects on dynamic instability by vinblastine in biochemical systems (Jordan and Wilson, 1990) and cells (Dhamodharan et al., 1995) have been observed at concentrations similar to the intracellular concentration we obtained at the IC₅₀ value for CA46 cells.

We find vinblastine to be more potent than did Singer and Himes (1992), but our vinblastine:tubulin ratio is almost identical with that obtained from the data of Jordan et al. (1991). Singer and Himes (1992) derived an intracellular vinblastine concentration in B16 melanoma cells at the IC₅₀ value from the accumulation factor determined at 1 µM drug (initial extracellular concentration) and concluded that at the IC₅₀ value the B16 melanoma cells would have a vinblastine:tubulin ratio of about 1:150-200. Jordan et al. (1991) reported that the tubulin concentration of HeLa cells was 2.0 mg/ml and that near the IC₅₀ value the intracellular vinblastine concentration was 38 nM, yielding a drug:tubulin ratio of 1:500.

Our comparison of dolastatin 10 and vinblastine suggests that equitoxic concentrations of microtubule assembly inhibitors in the culture medium lead to intracellular drug concentrations that have equivalent effects on the microtubule cytoskeleton and, consequently, on the fate of the cell. The study of Jordan et al. (1991) in HeLa cells with vinblastine and vincristine supports this concept. Vincristine and vinblastine do not show major differences in their interactions with tubulin (Himes, 1991), and, at initial extracellular drug concentrations close to the IC₅₀ and IC₉₀ values, intracellular drug concentrations were the same (30-40 and 150-160 nM, respectively). [The actual data (Jordan et al., 1991) follows. For vinblastine, the IC₅₀ value was 0.45 nM and the extrapolated IC₉₀ value was about 3 nM; for vincristine, 1.8 nM and about 7 nM, respectively. Jordan et al. (1991) reported 38 nM intracellular vinblastine with an initial extracellular concentration of 0.6 nM, and 33 nM intracellular vincristine with 1.8 nM extracellular drug. With 2.6 nM extracellular vinblastine, the intracellular concentration was 152 nM; with 10 nM extracellular vincristine, the intracellular drug was 163 nM.]

What accounts for the dramatic retention of dolastatin 10 as compared with vinblastine by CA46 cells? The difference in cellular accumulation can be explained in terms of the interactions of the two drugs with tubulin. Although their quantitative effects on assembly reactions are similar, there are major differences in their binding to tubulin and/or aberrant tubulin polymers. With both drugs it is difficult to document binding to the -tubulin dimer in the absence of an associated formation of spiral polymers. Both polymers and smaller oligomeric intermediates can be resolved from heterodimer on HPLC gel permeation columns, but preservation of vinblastine-induced structures requires drug in the column equilibration buffer (Singer et al., 1988). Preservation of dolastatin 10-induced structures does not require drug in the column buffer (Bai et al., 1995), and an unbound dolastatin 10 peak only appeared when drug:tubulin > 1.0. Moreover, the apparent K_a value obtained for dolastatin 10 (3.8 × 10⁷ M¹; Bai et al., 1995) is 21-fold higher than that obtained for vinblastine (1.8 × 10⁶ M¹; Safa et al., 1987) under comparable reaction conditions. [Literature values for the apparent K_a for vinblastine vary from 2 × 10⁴ to 6 × 10⁶ M¹, and the variation seems to depend primarily on reaction conditions (Himes, 1991).] This difference in apparent K_a values is almost identical with the difference in the accumulation factors obtained at 24 h.

The much more rapid uptake of vinblastine compared with dolastatin 10 suggested that cell kill with the vinca alkaloid would require a shorter treatment time, although vinblastine efflux after medium exchange or possible differences in minimum drug exposure time for onset of apoptosis (cf. Mooberry et al., 1997) could complicate data interpretation. Other lymphoma cell lines showed irreversible cytotoxicity after 4-8 h of exposure to dolastatin 10 (Beckwith et al., 1993), in good agreement with our observation that maximum uptake of the peptide at its IC₅₀ value occurred at 6 h. When we examined drug exposure time for the Burkitt cells at the IC₉₀ values (determined from the 24-h continuous exposure studies), we found that 4 h of exposure to dolastatin 10 and 24 h of exposure to vinblastine were required for maximum effects on subsequent cell growth (data not presented). Thus, differences in efflux and uptake rates between the two drugs, together with difference in the strength of their interactions with tubulin, probably play roles in the rapidity with which they kill cells. Our data suggest that for clinical studies dolastatin 10 would be optimally delivered as a slow infusion over several hours rather than by bolus administration, and that still longer infusion times would enhance the efficacy of vinblastine.

The depsipeptides cryptophycins 1 and 52 have many similarities to dolastatin 10 in their effects on tubulin, including formation of structures stable to gel filtration (Kerksiek et al., 1995; Bai et al., 1996; Smith and Zhang, 1996; Moore, 1997). For Burkitt CA46, EW-36, and EW-36VCR60 cells we obtained IC₅₀ values, respectively, of 20, 12, and 13 pM for cryptophycin 1 (unpublished observations). Cryptophycin 52 has been reported to be similar to cryptophycin 1 in its interactions with cells and tubulin (Moore, 1997). One would therefore predict similar findings to those reported here for dolastatin 10, and there have been two recent cellular uptake studies with cryptophycin 52 (Chen et al., 1998; Panda et al., 1998).

Chen et al. (1998), using a very high cryptophycin 52 concentration (1.5 µM) in a leukemia cell line, observed no detectable drug efflux (similar to our finding), maximal uptake by 20 min (versus the 6 h we found for dolastatin 10), and saturable uptake, suggesting an active transport process (opposite to our finding with dolastatin 10 concentrations up to 10 nM).

Panda et al. (1998), studying HeLa cells treated with 11 pM cryptophycin 52, the IC₅₀ value, found that the intracellular drug concentration at 20 h was 8 nM, with an accumulation factor of 730 (somewhat lower than we found with dolastatin 10). Because the 2.0 mg/ml tubulin content of HeLa cells (Jordan et al., 1991) is less than half that of Burkitt cells, the data of Panda et al. (1998) yield a drug:tubulin ratio of 1:2500 at the IC₅₀ value, 2.5 to 5-fold lower than our ratio for dolastatin 10 and 3- to 5-fold lower than our ratio for vinblastine. Panda et al. (1998) interpreted their results in terms of an exceptionally tight binding of cryptophycin 52 to microtubule ends with consequent inhibition of dynamic instability. A direct comparison of dolastatin 10 with cryptophycin 52 should provide additional insights into the mechanism of action of these potent peptide antimitotic drugs.