Introduction

Top Summary Introduction Materials & Methods Results Discussion References

MDR is a form of resistance to natural product derived anticancer agents, such as the anthracyclines, vinca alkaloids, and epipodophyllotoxins, characterized by an increased ATP-dependent efflux of the cytotoxic agent over the cellular plasma membrane. Two plasma membrane drug transporter proteins, Pgp and MRP, have been cloned and shown by transfection to induce the MDR phenotype to tumor cells (Bradley et al., 1988; Broxterman et al., 1995a; Loe et al., 1996a). Both proteins create a concentration gradient of the anthracycline DNR over the cellular plasma membrane (Zaman et al., 1994; Frézard and Garnier-Suillerot, 1991a, 1991b), resulting in a lowered intracellular drug concentration. Thus MDR produced by overexpression of MRP seems to be phenotypically similar to that caused by Pgp. However, major differences exist with regard to the substrates handled by these proteins. Although the chemical requirements for a compound to be a substrate for Pgp are not clear (Pearce et al., 1989), its substrate specificity seems to be restricted to neutral or cationic molecules. In contrast, MRP has been identified as a transporter of organic anions, such as the glutathione conjugates leukotriene C₄ (Loe et al., 1996a; Leier et al., 1994), dinitrophenylglutathione (Heijn et al., 1997), and the fluorescent dye calcein (Feller et al., 1995). In addition, the MRP-mediated efflux of these anionic species, as well as the classical MDR drugs, but not the Pgp-mediated drug efflux, is inhibited by drugs known to affect organic anion transport, like probenecid (Feller et al., 1995) or the anionic quinoline derivative MK571 (Gekeler et al., 1995). A further distinction between the transport properties of both proteins is that the efflux of the typical MDR drugs, such as DNR by MRP, seems to be dependent on the intracellular glutathione levels (Versantvoort et al., 1995). This was concluded from the demonstration that MRP-mediated cellular DNR (Versantvoort et al., 1995) or epirubicin (Davey et al., 1995) efflux is inhibited by lowering the cellular glutathione levels with DL-buthionine-(S,R)-sulfoximine, which can be reversed by repletion of cellular glutathione levels by treatment of the cells with glutathione ethyl ester (Versantvoort et al., 1995). These and other data (Loe et al., 1996b) have led to the speculation that MRP might be identical to the glutathione-conjugate (or GS-X) pump and would export drugs such as the anthracyclines only after they were chemically transformed to negatively charged glutathione conjugates (Ishikawa et al., 1995). The rapid efflux of drugs such as DNR (Versantvoort et al., 1994) with no detectable concomitant increase in glutathione efflux (Versantvoort et al., 1995), as well as the demonstration of ATP-dependent uptake of DNR or etoposide in MRP-expressing inside-out vesicles (Paul et al., 1996), and the inhibition of MRP-mediated dinitrophenyl-glutathione transport by anthracyclines in such a system (Heijn et al., 1997) have, however, challenged the hypothesis that MRP would exclusively transport the involved cytostatic drugs after metabolic conversion to conjugates (Broxterman et al., 1996).

Previous studies by our group have been directed at the question of the substrate specificity of Pgp-mediated drug efflux to contribute to a rational drug design for the treatment of drug resistant tumors (Mankhetkorn et al., 1996; Garnier-Suillerot, 1995; Priebe, 1995). With that purpose, we studied the ability of various chemical classes of molecules to be recognized by Pgp (Borrel et al., 1995; Borrel et al., 1994a) and the kinetics of Pgp-mediated efflux of a series of anthracyclines (Frézard and Garnier-Suillerot, 1991b; Mankhetkorn et al., 1996; Pereira et al., 1994; Borrel et al., 1994b). In the present work, we have systematically measured efflux parameters of DNR and three of its sugar-modified analogs and two DOX analogs with different pK_a values and lipophilicity in MRP-overexpressing small cell lung cancer cells and compared those with Pgp-mediated efflux parameters. Our data show that despite the different substrate profile of Pgp and MRP, the efflux properties of these anthracyclines are very similar.

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Cell culture

K562 leukemia cells and the Pgp-expressing K562/ADR cells (Mankhetkorn et al., 1996), as well as GLC₄ and the MRP-expressing GLC₄/ADR cells (Zijlstra et al., 1987), were cultured in RPMI 1640 (Sigma Chemical, St. Louis, MO) medium supplemented with 10% fetal calf serum (Biomedia, Boussens, France) at 37° in a humidified incubator with 5% CO₂. The resistant K562/ADR and GLC₄/ADR cells were cultured with 400 nM or 1.2 µM DOX, respectively, until 1-4 wk before experiments. Cell cultures used for experiments were split 1:2 1 day before use to assure logarithmic growth.

Drugs and chemicals. DNR and PIRA were kindly provided by Roger Bellon laboratory. OH-DOX, OH-DNR, eso-DNR, and Br-DNR were provided by author W.P. (Priebe, 1995). Stock solutions were prepared in water just before use. Concentrations were determined by diluting stock solutions in water to approximately 10 µM and using ₄₈₀ = 11,500/M/cm. Experiments were performed in HEPES/Na buffer solutions containing 20 mM HEPES plus 132 mM NaCl, 3.5 mM KCl, 1 mM CaCl₂ and 0.5 mM MgCl₂, pH 7.25, in the presence of 10 mM sodium azide. To synthesize ATP via the glycolysis, 5 mM glucose was added at the start of efflux experiments (see below).

Cellular drug accumulation. The rationale and validation of our experimental set-up for measuring the kinetics of active transport of anthracyclines from tumor cells has been extensively described and discussed before (Frézard and Garnier-Suillerot, 1991a, 1991b; Mankhetkorn et al., 1996; Borrel et al., 1995; Borrel et al., 1994a, 1994b; Pereira et al., 1994). It is based on a continuous spectrofluorometric monitoring (Perkin Elmer LS50B spectrofluorometer) of the decrease of the fluorescence signal of the anthracycline at 590 nm (_ex = 480 nm) after incubation with cells in a 1-cm quartz cuvette. The decrease of fluorescence that occurred during incubation with cells is caused by quenching of the fluorescence after intercalation of anthracycline between the base-pairs of DNA. We have previously shown that this method allows us to measure accurately the free cytosolic concentration of anthracyclines in steady state, their initial rates of uptake, and kinetics of active efflux (Frézard and Garnier-Suillerot, 1991a, 1991b; Mankhetkorn et al., 1996; Borrel et al., 1995; Borrel et al., 1994a, 1994b; Pereira et al., 1994).

Determination of the MRP or Pgp-mediated efflux of anthracycline derivatives. Cells (1 × 10⁶/ml; 2 ml per cuvette) are preincubated for 30 min in HEPES buffer with sodium azide but without glucose. Depletion of ATP in these cells was 90%, as checked with the luciferin-luciferase test (Kimmich et al., 1975). The cells remained viable throughout the experiment, as checked with trypan blue. After addition of anthracyclines, the decrease of the signal is followed until steady state is reached. Because the pH of the medium is chosen to equal the intracellular pH, at steady state the C_e is equal to the C_i. Then glucose is added, which leads to restoration of control ATP levels within 2 min and increase of the fluorescence signal because of the efflux of anthracycline. The ATP-dependent anthracycline efflux is determined from the slope of the tangent of the curve F = f(t), where F is the fluorescence intensity at the time of addition of glucose. Under these conditions, at the moment glucose is added, C_i = C_e and the passive influx and efflux are equal; therefore, the net initial efflux represents the MRP or Pgp-mediated active efflux only. Similar experiments were performed with the drug-sensitive cell lines and no efflux was measured (Fig. 2).

Mathematical calculations. The maximal efflux rate (V_M), apparent Michaelis-Menten constant (K_m) and cooperativity constant (n_H) for the transport of anthracyclines were computed by nonlinear regression analysis of V_a versus C_i data using the MacCurveFit program and assuming that the transport follows the Hill equation (Hill, 1985):

(1)

where n_H, the Hill coefficient, represents the cooperativity constant.

(2)

(3)

(4)

(5)

Hydrophobicity of the anthracycline derivatives. An estimation of the hydrophobicity of a compound is given by logP, where P is the partition coefficient in an n-octanol/water system. In a first approximation, the logP value of a compound can be estimated by adding the fi values of its fragments (Rekker, 1977). We have determined the variation of this calculated logP (c logP) of tested compounds using DNR as a reference compound and the equation c logP = log P₀ +fi, where P₀ is the partition coefficient for DNR in n-octanol/water and fi is the n-octanol/water fragmental constant of the fragments of a compound that differentiates it from DNR. The values are reported in Fig. 1.

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Structures, pKa and relative hydrophobicity values of the anthracyclines used. Variation of hydrophobicity evaluated as sum of fragmental values fi, referred to DNR (the calculated log P, c log p = log P0 + fi where log P0 is the partition coefficient for DNR in n-octanol/water).

	Results

Top Summary Introduction Materials & Methods Results Discussion References

The structures of the anthracycline derivatives used are shown in Fig. 1, together with the pK_a values of deprotonation of their amino group and their hydrophobicity estimated from their c logP values. Because the concentration of neutral form of the anthracyclines is in equilibrium between extracellular medium and cytosol when steady state is reached (the condition used to calculate C_i) the extracellular pH has to be equal to cytosolic pH to allow these calculations. We have chosen here an extracellular pH of 7.25, because the cytosolic pH in both the sensitive and the resistant K562 and GLC₄ cells was shown to be within the range of 7.2-7.3 (Frézard and Garnier-Suillerot, 1991a; Versantvoort et al., 1992).

Determination of the K_m, V_M, and n_H coefficients of anthracycline efflux

Typical examples of these experiments are shown in Fig. 2. The time required to reach the steady state of anthracycline accumulation in the energy-depleted cells varied between 20 min (for Br-DNR and PIRA) and 90 min (for DNR). At this time-point, C_i = C_e was calculated from the (nonquenched) fluorescence (Frézard and Garnier-Suillerot, 1991a, 1991b; Mankhetkorn et al., 1996) and V_a was calculated as C_T /F₀ · (dF/dt), where F₀ is the fluorescence of a C_T µM anthracycline solution and dF/dt the slope of the tangent to the curve F(t) after addition of glucose which initiated the active efflux component. The same experiment was performed using the parent cell lines, in which no active efflux could be detected (shown for DNR and PIRA in Fig. 2).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Incorporation of PIRA and DNR in GLC4/S and GLC4/ADR cells after ATP depletion by incubation with azide and determination of the active efflux rate (Va) after restoration of ATP synthesis by the addition of glucose (normalized fluorescence intensity).

The K_m, V_M, and n_H coefficients of the active transport were then calculated as described from the curves of V_a versus C_i. Fig. 3 shows two examples of the curves obtained for DNR and Br-DNR. The values of the three parameters for all anthracyclines are reported in Tables 1 and 2 for GLC4/ADR and K562/ADR cells, respectively. It seems that the apparent K_m values were within the rather narrow range of 0.4-2.4 µM for the GLC₄/ADR as well as K562/ADR cells, indicating a similar affinity of the anthracyclines for MRP and Pgp. However, the V_M values seemed to vary in a systematic way, dependent on the anthracycline and on the cell line. Fig. 4 shows a representation of V_M as a function of the percentage of neutral form of the anthracycline. In the case of K562/ADR, no trend could be seen, whereas for the GLC₄/ADR cells, an apparently negative correlation between V_M and the percentage of neutral form of the anthracycline was seen.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Kinetics of the Pgp-mediated efflux of DNR and Br-DNR plotted as a function of Ci. GLC4/ADR cells (1 × 106/ml) were incubated in the presence of various concentrations of drug ranging from 0 to 12 µM. Va and Ci were determined as described. Data are from 3-5 independent experiments on different days.

View larger version (35K): [in this window] [in a new window]   Fig. 4.   Relation between the maximum rate of efflux VM and the percentage of drug in the neutral form for GLC4/ADR and K562/ADR cells.

The mean active efflux coefficients K_a were calculated using eq. 5 and are shown in Tables 1 and 2. This parameter had a somewhat lower value for the two analogues in which the amino group had been replaced by a hydroxy group (OH-DOX and OH-DNR), particularly for the GLC₄/ADR cells. The efficiency of efflux of PIRA and Br-DNR seemed to be slightly higher than for DNR in the case of K562/ADR cells but not GLC₄/ADR cells.

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

Overexpression of Pgp or MRP in tumor cells, either resulting from selection with cytostatic agents or from transfection with the MDR1 or MRP gene leads to resistance to a spectrum of anticancer drugs, represented by some major clinically active drugs, such as DOX, DNR, vincristine, and etoposide. Resistance to the anticancer agents taxol and mitoxantrone is high in Pgp-overexpressing cells but less or absent in MRP-overexpressing cells (Zaman et al., 1994; Cole et al., 1994; Broxterman et al., 1995b). It has been established that both proteins belong to the superfamily of ATP-binding cassette transmembrane transporter proteins (Hughes, 1994), which have ATP-binding sites (Cornwell et al., 1987), ATPase activity (Hooijberg, 1997), and use the energy from ATP hydrolysis, probably to directly translocate their substrates (Broxterman and Pinedo, 1991).

Despite considerable overlap in substrate specificity between Pgp and MRP, major differences have now been established, mainly related to the charge of putative substrates. Whereas Pgp seems to transport neutral and positively charged substrates, MRP has clearly been shown to have the properties of an organic anion transporter (cMOAT or GS-X pump) (Loe et al., 1996a; Leier et al., 1994). Whereas the presence of a positive charge on such molecules as rhodamine 123 or SYTO 16 may be the reason for their highly efficient Pgp-mediated transport, it may preclude their efficient transport by MRP (Twentyman et al.; 1994; Broxterman et al., 1997).

In this article, we present data on the kinetics of ATP-dependent transport of a series of anthracyclines in the MRP-overexpressing GLC4/ADR cells and compare the kinetic parameters with new and previous data on Pgp-mediated transport. Basically, we blocked the energy-dependent transport of the anthracyclines by MRP and Pgp by depleting ATP in the cells until steady state was reached. After applying glucose, the initial efflux rate of the fluorescent anthracyclines can be measured, because the quenching by DNA is released (Frézard and Garnier-Suillerot, 1991b; Mankhetkorn et al., 1996; Garnier-Suillerot, 1995; Borrel et al., 1994a, 1994b; Pereira et al., 1994). This allows the accurate estimation of efflux kinetics from intact viable cells with the transporter proteins in their native membrane environment. The value we found for the V_M of active DNR transport in the GLC₄/ADR cells was 2 nM/sec, which corresponds to 1.2 million molecules of DNR per cell per second, which is in good agreement with the V_M for DNR transport from the GLC4/ADR cells, as has been determined previously by a radioactive method (200 pmol/10⁶ cells/min or about 2 million molecules DNR per cell per second) (Versantvoort et al., 1994). The parameter V_M contains both the turnover number of the relevant transporter protein as well as the density of transporter protein molecules on the cellular plasma membrane. Thus, in comparing the data recorded for the two drug-resistant cell lines in Tables 1 and 2, a difference in V_M for any substrate could arise from a difference in either the turnover number of the transporter or the density of the transporter molecules.

We show that the mean active efflux coefficient for the present series of anthracyclines is very similar for Pgp- and MRP-mediated transport, except for a tendency for a slightly higher k_a for PIRA and Br-DNR in the case of Pgp-mediated efflux. A further similarity between Pgp- and MRP-mediated anthracycline transport follows from what are apparently not simple hyperbolic curves of V_a versus C_i, with many Hill coefficients close to 2. This means that for both proteins, a positive cooperative transport of two molecules of anthracycline may be hypothesized, as was previously already suggested for Pgp using a different methodology (Spoelstra et al., 1992).

Substitution of a hydroxyl group for the amine group as in OH-DOX and OH-DNR did not seem to abolish its transport by Pgp (Borrel et al., 1994b) or MRP, but it did result in a similar reduction of the k_a. We have also seen that after depletion of cellular GSH by 20 hr of pretreatment of the GLC₄/ADR cells with 25 µM buthionine sulfoximine (Versantvoort et al., 1995), the efflux of OH-DOX or OH-DNR was abolished (not shown). Thus, the theory that the amino group in the sugar portion of doxorubicin might be an important but not essential part of the anthracycline molecule for recognition by Pgp (Priebe et al., 1993) also seems to hold for MRP. However, we also found a remarkable decrease in the V_M of the transport of both permanently neutral derivatives in the case of MRP-mediated transport only (Fig. 4). This would suggest that the presence of the amino group in the anthracycline structure is more important in the case of MRP-mediated transport. As pointed out before, other evidence is not in favor of efficient transport of permanently positively charged substrates by MRP.

On the other hand, because the transport by MRP of the two negatively charged substrates, oxidized glutathione (Heijn et al. 1997) and azidophenylglutathione (Shen et al., 1997), can be competitively inhibited by DNR, it has been hypothesized that the anthracycline binding site of MRP may be the same as or near the organic anion binding site (Heijn et al., 1997). Thus it may be suggested that the anthracycline and a negatively charged molecule interact in a (partly overlapping) binding site of MRP. Further indications for that suggestion are the GSH dependence of anthracycline and vinca alkaloid transport by MRP in intact cells (Versantvoort et al., 1995; Davey et al., 1995) and the finding that GSH can stimulate the ATP-dependent transport of the anticancer drug vincristine into inside-out vesicles of MRP-transfected cells (Loe et al., 1996b). Because GSH is present in the intact cells but not in some other reported vesicle experiments (Paul et al., 1996), it may be that GSH by a (cooperative) interaction decreases the K_M for the binding of DNR to its transporter site.

In conclusion, our data show that the kinetics of anthracycline transport by MRP are very similar to those for Pgp, and we are therefore in favor of the idea that MRP transports these molecules predominantly in their unmetabolized form. This does not exclude the idea that these species may be MRP substrates if some metabolism of anthracyclines occurs in tumor cells (e.g., to glucuronides). In this respect, it will be important to identify the anthracycline binding site(s) of MRP and their connection with the binding site for organic anions.