MDR 1 activation is the predominant resistance mechanism selected by vinblastine in MES-SA cells

Single-step selection with vinblastine was performed in populations of the human sarcoma cell line MES-SA, to assess cellular mechanisms of resistance to the drug and mutation rates via fluctuation analysis. At a stringent selection with 20 nM vinblastine, resulting in 5–6 logs of cell killing, the mutation rate was 7 × 10–7per cell generation. Analysis of variance supported the hypothesis of spontaneous mutations conferring vinblastine resistance, rather than induction of adaptive response elements. Surviving clones displayed a stable multidrug resistance phenotype over a 3-month period. All propagated clones demonstrated high levels of resistance to vinblastine and paclitaxel, and lower cross-resistance to doxorubicin and etoposide. Activation of MDR 1 gene expression and P-glycoprotein function was demonstrable in all clones. No elevation was found in the expression of the mrp gene, the LRP-56 major vault protein and β-tubulin isotypes (M40, β4, 5β, and β9) in these mutants. We conclude that initial-step resistant mechanism in these vinblastine-selected mutants commonly arises from a stochastic mutation event with activation of the MDR 1 gene. © 2000 Cancer Research Campaign

P-glycoprotein (P-gp) is a multidrug efflux pump encoded by the MDR1 gene. P-gp is expressed in some normal and malignant tissues, and associated with clinical multidrug resistance (MDR) and poor prognosis in many cancer types (Ling, 1992;Gottesman, 1993;Sikic et al, 1997). Clinical trials using P-gp inhibitors such as the cyclosporin analogue valspodar (PSC 833) combined with chemotherapeutic agents are currently being carried out to reverse or prevent clinical drug resistance (Fisher and Sikic, 1995).
Vinca alkaloids such as vinblastine are potent anti-tumour agents and widely used in the chemotherapy of cancers. Cellular mechanisms of resistance to vinca alkaloids include expression of MDR1/P-gp, the mrp gene, alterations in tubulin or microtubuleassociated proteins, and alterations in the regulation of programmed cell death or apoptosis (Cabral and Barlow, 1989;Haber et al, 1995;Giannakakou et al, 1997;Dumontet and Sikic, 1999). Most cellular models of resistance to vinblastine were established by step-wise selection of cancer cells up to very high concentrations. Single-step selection and fluctuation analysis provide a useful genetic tool to investigate the frequency and the nature of various resistance mechanisms to cytotoxic agents (Luria and Delbrück, 1943;Goldie and Coldman, 1979;Kendal and Frost, 1988;Jaffrézou et al, 1994;Beketic-Oreskovic et al, 1995;Dumontet et al, 1996).
In previous reports, we have identified the stochastic nature of initial resistance mechanisms to etoposide, doxorubicin, paclitaxel, and doxorubicin in the presence of valspodar (PSC 833), in the human sarcoma cell line MES-SA Beketic-Oreskovic et al, 1995;Dumontet et al, 1996). In this study, we characterized the rate and mechanisms of resis-tant cells derived from vinblastine single-step selection. Acquired MDR1 expression in these single-step mutants was found in all resistant clones and was associated with the initial resistant phenotype in these clones.

Drugs and chemicals
Paclitaxel, etoposide, vinblastine, and doxorubicin were obtained from the drug repository of the National Cancer Institute, NIH (Bethesda, MD, USA

Cells and cell culture
The development, characterization, and culture of the human uterine sarcoma cell line MES-SA and its multistep-selected MDR variant Dx5 cells have been described (Harker et al, 1983;Harker and Sikic, 1985). MES-SA cells have a pseudodiploid karyotype of 45XX, which has remained stable over a 17-year period of cultivation. Monolayer cultures of MES-SA cells are not growth limited at low cell densities and have a plating efficiency greater than 85% (Chen GK and Sikic BI, unpublished data). Therefore, these human sarcoma cells are especially useful for fluctuation analysis.

Luria-Delbrück fluctuation analysis
Subclonal MES-SA cells were expanded for fluctuation analysis. Eight 25 cm 2 tissue culture flasks (Corning Glass Works, Corning, MDR1 activation is the predominant resistance mechanism selected by vinblastine in MES-SA cells NY, USA) were seeded with MES-SA cells at a low density (1000 cells per flask) and allowed to grow to near confluence (2 × 10 6 cells). The cells from each flask were seeded into separate 96-well plates (1.7 × 10 6 cells per plate). After 4 h of incubation to ensure active growth, cells were treated with 20 nM vinblastine. Preliminary experiments had demonstrated that this concentration of vinblastine resulted in approximately 5 logs of cell killing in MES-SA cells. The drug-containing medium was changed every other day for 14 days, and then replaced by drug-free medium. Surviving colonies were allowed to grow for another 2 weeks and were then individually harvested and propagated in drug-free medium for further studies. The mutation rate was calculated according to the method of Catcheside (Catcheside, 1951), from the equation: µ = 2 ln 2(r 2 /N 2 -r 1 /N 1 )/g where µ is the mutation rate per cell generation, r represents the numbers of resistant colonies at times 1 and 2, N is the initial cell number adjusted for plating efficiency, and g is the number of cell generations.
In a control experiment, bulk populations of MES-SA cells without expansion of the populations prior to drug exposure were treated directly with vinblastine. In this experiment, six 96-well plates were seeded with 1.7 × 10 6 cells per plate from a single population of 1.0 × 10 7 cells and received identical exposure to vinblastine as above.

Cytotoxicity assays and modulation of MDR by P-gp inhibitors
Cytotoxicity assays and assessment of reversal of the MDR phenotype were performed using the 72 h MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] colourimetric assay as previously described . The IC 50 (drug concentration resulting in 50% inhibition of MTT dye formation compared with the controls) was determined directly from semilogarithmic dose-response curves.

Cell growth rate
The doubling times of MES-SA and the isolated clones were determined by seeding cells in 96-well plates and incubating at 37°C for 4, 24, 48, and 72 h. The cellular growth rate was estimated by the MTT colourimetric method. Population doubling time (DTs) was calculated as previously described .

Flow cytometric analysis of P-gp expression
Retention of rhodamine 123 (Rh-123) was determined by a FACS-II™ dual laser flow cytometer (Becton-Dickinson Corp, Mountain View, CA, USA). Double labeling was performed with Rh-123 and the monoclonal antibody UIC2 (Immunotech Corp, Westbrook, ME, USA) . Cells were incubated with or without 2 µM valspodar at 37°C, and stained with Rh-123 (0.1 µg ml -1 ) for 45 min, followed by 20 min drug efflux in the presence or absence of 2 µM valspodar.

Western blotting and immunohistochemistry
The light-enhanced chemiluminescence Western blot protocol (Amersham Life Science) was used for the detection of P-gp. Total cell lysates were used for P-gp immunoblotting with the monoclonal antibody C219 (Signet Inc, Dedham, MA, USA) Beketic-Oreskovic et al, 1995). P-gp and LRP (p110) expression were also assessed by immunocytochemistry, utilizing the UIC2 and LRP-56 monoclonal antibodies (Caltag Laboratories Inc, San Francisco, CA, USA) Beketic-Oreskovic et al, 1995).

Drug accumulation
and [ 3 H]-etoposide accumulations in the presence or absence of the P-gp inhibitors valspodar and verapamil were quantitated using radiolabeled drugs, and normalized to whole cellular protein content (Beketic-Oreskovic et al, 1995;Chen et al, 1997).

Amplimers and RT-PCR
The amplimers and methodology for RT-PCR of MDR1, mrp, βtubulin isotypes, and rRNA have been previously documented (Beketic-Oreskovic et al, 1995;Dumontet et al, 1996;Chen et al, 1997). Table 1 presents the results of the fluctuation analysis experiment in which populations of MES-SA sarcoma cells were selected with vinblastine in a single step. The mean number of colonies surviving drug exposure in the fluctuation test group was 14 per cell population, each propagated from an initial seeding of 10 3 drug-sensitive cells, with a statistical variance of 107 from the mean. In contrast, the mean number of colonies surviving drug exposure among the control populations (derived from bulk cultures of cells) was 16 per population, with a variance of 37 (Table 1). The larger ratio of variance relative to the mean number of survivors in the fluctuation group (7.6 vs 2.3) supports the hypothesis of spontaneous mutations conferring resistance to vinblastine in these cells, rather than an effect of induction of resistance by drug exposure. The mutation rate for vinblastine resistance in this single-step selection was estimated to be 7 × 10 -7 per cell per generation according to the method of Catcheside (1951) (Table 1).

MDR phenotype and its reversal
A total of 13 clones which survived vinblastine exposure were isolated, propagated and tested for resistance to the selecting agent as well as to doxorubicin, paclitaxel and etoposide. These clones included at least one from each of the eight original populations. There was no detectable alteration in proliferation rates among these clones compared to the parental cell line, which has a generation time of approximately 22 h (data not shown). All of the propagated clones demonstrated an MDR phenotype ( Table 2), typical of MDR1 expression. Thus, all clones showed marked crossresistance to paclitaxel (ranging from 50-to 200-fold). Crossresistance to paclitaxel was usually greater than to the selecting agent vinblastine, and lower to doxorubicin and etoposide (Table  2). Five other clones were propagated (VL20-1.2, VL20-4.2, VL20-5.2, VL20-6.2, and VL20-8.2) and shown to have a similar drug resistance phenotype to that of their related clones in Table 2 (data not shown). The resistance to vinblastine and paclitaxel in all propagated clones was completely reversed by 2 µM valspodar. All propagated clones maintained stable resistance to vinblastine and paclitaxel over a 3-month period.

Rh-123 retention and potentiation by valspodar
The intracellular accumulation of the fluorescent P-gp substrate Rh-123 was examined in eight resistant clones. All clones demonstrated significantly decreased accumulation of Rh-123 (Figure 1), with Rh-123 retention ranging from 14% to 56% compared to the parental MES-SA cells (100%). Valspodar at 2 µM restored the intracellular accumulation of Rh-123 to levels similar to the parental cells, which do not express P-gp (Figure 1).

Accumulation of [ 3 H]-vinblastine, [ 3 H]-paclitaxel, and [ 3 H]-etoposide
The vinblastine selected clones demonstrated markedly reduced intracellular accumulation of [ 3 H]-vinblastine and [ 3 H]-paclitaxel relative to the wild-type MES-SA cells. Furthermore, this decreased drug accumulation was reversed by the MDR modulators valspodar and verapamil. In these experiments, 2 µM valspodar completely sensitized the vinblastine-resistant phenotype to the parental MES-SA cell level (Figure 2). Valspodar at 2 µM was more effective as a modulator in these cells than 6 µM verapamil, which only partially reversed the accumulation defects for both vinblastine and paclitaxel (Figure 3). One clone showed resistance to modulation by verapamil of the accumulation of vinblastine and paclitaxel (Figure 3). In general, these results suggest that a typical drug efflux mechanism associated with reduced accumulation of total intracellular drugs was responsible for cellular resistance to vinblastine in these clones.

Expression of MDR1 mRNA and P-gp
All examined clones expressed MDR1 mRNA transcripts and functional P-glycoprotein, compared to the parental MES-SA cells which do not express MDR1 ( Figure 5A). P-gp expression and function were analysed by double labeling of cells using the monoclonal antibody UIC2 combined with Rh-123 accumulation. Figure 4 demonstrates the association between P-gp expression  The difference in the two groups was that a small number of cells (10 3 ) were seeded and grown separately in the fluctuation group, according to Luria and Delbrück (1943), whereas the controls utilized bulk populations of cells (1.7 × 10 6 ) from mass cultures, prior to selection with 20 nM vinblastine; b calculated according to Catcheside (1951).  and decreased retention of Rh-123 in the resistant clones compared to parental MES-SA cells. There was a highly significant correlation of P-gp/UIC2 expression with decreased Rh-123 accumulation among the nine tested clones ( Figure 4D, r = 0.84, P < 0.002). Expression of P-gp was confirmed in these clones by Western blotting using the monoclonal antibody C219 ( Figure  5B).

LRP and mrp expression
Expression of the major vault protein LRP was assessed by immunocytochemistry, utilizing the LRP-56 monoclonal antibody. The parental, drug-sensitive MES-SA cells stained positively for LRP (p110). Vinblastine-selected variants had a similar degree of LRP-56 staining compared to parental MES-SA cells (data not shown). None of the vinblastine-selected clones had an increase in mrp expression (Figure 6) or altered β-tubulin isotypes (M40, β4, 5β, and β9), assessed by RT-PCR (data not shown).

DISCUSSION
Fluctuation analysis provides a powerful approach to studying the nature and rate of acquired resistance to anticancer agents, although such studies are difficult and laborious in mammalian cells (Luria and Delbrück, 1943;Goldie and Coldman, 1979;Kendal and Frost, 1988;Jaffrézou et al, 1994;Beketic-Oreskovic et al, 1995;Dumontet et al, 1996). The human sarcoma cell line MES-SA is particularly suitable for such studies because of its high plating efficiency, relative drug sensitivity, and karyotypic and genetic stability over two decades of growth (Harker et al, 1983;Chen and Sikic, unpublished data). Previously, 1000 100 10 MES-SA VL20-2.1 VL20-4.1 VL20-6.1 VL20-4.2 VL20-6.2 VL20-7.1 VL20-8.1 VL20-8. we have shown that activation of the MDR1 gene in MES-SA sarcoma cells could be acquired by stochastic genetic events Dumontet et al, 1996). MDR1 activation was present in all 13 mutants selected with doxorubicin , and in four of nine (44%) selected with paclitaxel (Dumontet et al, 1996). It has been shown that multiple mechanisms are involved in cellular resistance to vinca alkaloids (Cabral and Barlow, 1989;Haber et al, 1995;Giannakakou et al, 1997;Dumontet and Sikic, 1999). In order to investigate the nature or origin of acquired resistance to vinca alkaloids in MES-SA cells, we used fluctuation analysis to examine the potential diversity of resistance mecha-nisms that may arise among eight parallel cell populations in a single-step exposure to vinblastine. We found that all eight mutants from separate parental cell populations acquired MDR1 expression and manifested the MDR phenotype characteristic of P-gp.
In theory, if resistance was acquired by induced events or directed mutations, the number of surviving colonies would be expected to have a Poisson distribution, with the variance close to the mean (Luria and Delbrück, 1943;Goldie and Coldman, 1979;Kendal and Frost, 1988). In this study, the variance in number of surviving colonies per plate was over 8-fold greater than the mean, indicating a spontaneous event rather than direct induction of MDR1 expression by drug exposure. These data are strong evidence that a random mutational mechanism caused the cellular resistance to vinblastine. The Luria-Delbrück test may be particularly applicable for analysis of mutations due to single-step irreversible events, rather than two-or multiple-step genetic changes such as gene amplification (Kimmel and Axelrod, 1994). The estimated rate of mutation (7 × 10 -7 per cell generation) conferring resistance to 20 nM vinblastine in MES-SA cells is similar to that which we observed for doxorubicin resistance (1.8 × 10 -6 ) in the same cells, and somewhat lower than that which has been described for some gene amplification events in other cell lines, 1-7 × 10 -5 per cell generation (Tlsty et al, 1989;Baker et al, 1974;Crawford et al, 1983;Cole et al, 1976). It should be pointed out that the spontaneous mutation rate is not always constant, but can be highly dependent on the parental cell line, drug concentration, and other experimental conditions (Boesen et al, 1994).
A significant level of MDR1 transcripts and P-gp expression were verified in all tested clones which were initially cross-resistant to the MDR-related drugs doxorubicin, vinblastine, paclitaxel, and etoposide. There was some variability in the pattern of resistance, which may reflect altered forms or post-translational regulation of P-gp in these cells or clonal variation in other genes. In addition to ruling out alterations in LRP and mrp expression, we found no obvious alterations in tubulin content (data not shown) or β-tubulin isotype expression in these mutants. Functional analysis  Table 2). Thus, these vinblastine-selected clones are phenotypically similar to the previously published MES-SA mutants selected by doxorubicin with activation of MDR1 . The mechanisms of acquired MDR related to activation of MDR1 are not well understood. The genetic mechanisms that regulate MDR1 expression acquired during drug selection may differ from those responsible for constitutive expression, and these single-step models may provide new insights into this question.
In conclusion, these data demonstrate that resistance to 20 nM vinblastine in the human sarcoma cell line MES-SA is not induced but arises spontaneously with an apparent mutation rate of 7 × 10 -7 per cell generation. Vinblastine exposure selected predominantly for MDR1-expressing variants. It is likely that these mutants arose from one major event leading to an activation of MDR1. The detailed nature of the MDR1 gene activation in these clones is being investigated.

Figure 6
Histogram of mrp mRNA levels in vinblastine-selected MDR mutants. Mrp mRNA expression in vinblastine-selected mutants was determined by RT-PCR, rRNA was used as the control gene to normalize expression as described. MES-SA and its MDR subline were used a controls.