Erland J.F. Demant

A model for computer simulation of P-glycoprotein and transmembrane ΔpH-mediated anthracycline transport in multidrug-resistant tumor cells

Anthracycline resistance in multidrug-resistant (MDR) tumor cells is due in part to a reduced cellular drug accumulation. Using a simple kinetic model and numerical computer simulations, we have analyzed mathematically the following possible mechanisms controlling fluxes of the membrane permeable anthracyclines in MDR cells: (1) active outward transport via a specific drug transporter (P-glycoprotein), (2) exocytotic drug export via the endosomal vesicle system, and (3) pH-gradients across the plasma membrane. Model calculations were based on morphometric and kinetic data previously presented in the literature for daunorubicin transport in wild-type Ehrlich ascites tumor cells (EHR2) and the corresponding daunorubicin (DNR)-resistant cell line EHR2/DNR+. The results confirm the possible importance of the cell-surface pH in controlling DNR accumulation in the cells. With P-glycoprotein as the main efflux pump, a catalytic constant of the protein greater than 40 mol DNR transported/mol protein per min is predicted by the model calculations. Changes in the drug binding affinity of P-glycoprotein (Km = 10(-9)-10(-6) M) is of little importance in influencing its effectiveness to reduce DNR accumulation, which could explain the broad substrate specificity of the MDR efflux pump system. The conditions to evaluate unidirectional fluxes of DNR across the plasma membrane in cells with active P-glycoprotein are defined. An efflux mechanism which relies solely on pH-dependent drug trapping in a pH 5 endosomal compartment by a simple diffusion process followed by exocytosis, appears inadequate to account for the high rate of DNR efflux in EHR2/DNR+ cells.

Transfer of ferritin-bound iron to adriamycin

Interactions of adriamycin with ferritin‐bound iron have been investigated. It is demonstrated (i) that adriamycin stimulates an iron‐dependent lipid peroxidation in submitochondrial particles in the presence of ferritin, and (ii) that incubation of adriamycin with ferritin results in a slow transfer of iron to adriamycin with formation of an adriamycin‐iron complex. The results are discussed in relation to the possible role for intracellular iron in adriamycin toxicity.

Stopped-flow kinetic analysis of long-chain fatty acid dissociation from bovine serum albumin

The kinetics of the interaction of long-chain fatty acids (referred to as fatty acids) with albumin is critical to understanding the role of albumin in fatty acid transport. In this study we have determined the kinetics of fatty acid dissociation from BSA and the BSA-related fatty acid probe BSA-HCA (BSA labelled with 7-hydroxycoumarin-4-acetic acid) by stopped-flow methods. Fatty acid-albumin complexes of a range of natural fatty acid types and albumin molecules (donors) were mixed with three fatty acid-binding acceptor proteins. Dissociation of fatty acids from the donor was monitored by either the time course of donor fluorescence/absorbance or the time course of acceptor fluorescence. The results of these measurements indicate that fatty acid dissociation from BSA as well as BSA-HCA is well described by a single exponential function over the entire range of fatty acid/albumin molar ratios used in these measurements, from 0.5:1 to 6:1. The observed rate constants (k(obs)) for the dissociation of each fatty acid type reveal little or no dependence on the initial fatty acid/albumin ratio. However, dissociation rates were dependent upon the type of fatty acid. In the case of native BSA with an initial fatty acid/BSA molar ratio of 3:1, the order of k(obs) values was stearic acid (1.5 s(-1)) < oleic acid < palmitic acid congruent with linoleic acid<arachidonic acid (8 s(-1)) at 37 degrees C. The corresponding values for BSA-HCA were about half the values for BSA. The results of this study show that the rate of fatty acid dissociation from native BSA is more than 10-fold faster than reported previously and that the off-rate constants for the five primary fatty acid-binding sites differ by less than a factor of 2. We conclude that for reported rates of fatty acid transport across cell membranes, dissociation of fatty acids from the fatty acid-BSA complexes used in the transport studies should not be rate-limiting.

Inactivation of cytochrome c oxidase activity in mitochondrial membranes during redox cycling of doxorubicin

Interactions of doxorubicin (DX) with the cardiolipin-dependent cytochrome c oxidase have been examined by using pig heart submitochondrial particles (SMP). A progressive and irreversible loss of oxidase activity is demonstrated in 2 hr incubations of the SMP with 10-100 microM DX in air-equilibrated medium with excess NADH to support redox-cycling of the drug. This oxidative mechanism for oxidase inactivation occurs in connection with a peroxidation process in the bulk membrane lipid, and is independent on turnover of the enzyme. It is related in a complex manner to the electron flux in the respiratory chain with antioxidant properties, and is maximal at the high reduction level of respiratory chain Complex I obtained in the presence of rotenone. Reduction of DX per se plays a minor role, and trace concentrations of chelatable metal ions (iron) are required to catalyse the reaction. Iron in the iron storage protein ferritin is released by DX, and at physiological low O2 concentrations ([O2] less than 20 microM), this iron is a better promoter of oxidase inactivation than is endogenous iron in the SMP. Kinetic analysis of inactivation data indicates the interaction of DX with low affinity (Km 35-55 microM) binding sites in the SMP membranes. Overall, the results point to the possible role of ferritin-iron in the mechanism of DX mitochondrial toxicity and argue against site specific effects of the DX-reduction/oxidation cycle on the cytochrome c oxidase or on its essential phospholipid (cardiolipin) environment.

Binding of transferrin‐iron by adriamycin at acidic pH

It is shown that adriamycin is able to chelate iron released from iron‐loaded serum transferrin in the pH range from 6.5‐4.1. The kinetics of iron transfer to free adriamycin and to adriamycin covalently attached to the transferrin has been determined. The results show that adriamycin, if introduced into intracellular acidic compartments, could function as acceptor for transferrin‐iron.

Covalent modification of serum transferrin with phospholipid and incorporation into liposomal membranes.

A method is described for incorporation of water-soluble proteins into liposomal membranes using covalent protein-phospholipid conjugates in detergent solution. A disulfide derivative of phosphatidylethanolamine containing a reactive N-hydroxysuccinimide ester group is synthesized, and the derivative is reacted with serum transferrin in deoxycholate-containing buffer. Disulfide-linked transferrin-phosphatidylethanolamine conjugates containing up to 6 mol phospholipid/mol protein are prepared. The amphiphilic conjugates have solubility properties very similar to integral membrane proteins. The conjugates self-associate to form protein micelles of narrow size distribution (Stokes radii 6-7 nm), and in the presence of excess phospholipid (egg phosphatidylcholine), they readily incorporate into liposomal membranes upon removal of detergent. Stable incorporation into liposomes requires the introduction of two molecules of phosphatidylethanolamine into the transferrin. Using the disulfide linker to release transferrin from the liposomes, evidence is presented for a function of the phosphatidylethanolamine as an anchor-molecule into the liposomal lipid. Optimal conditions for preparation of homogeneous liposomes with diameters in the range 30-125 nm and with a varying content of transferrin are defined. The liposomes appear well suited for studies on liposome-cell membrane interactions.

Doxorubicin sensitivity pattern in a panel of small-cell lung-cancer cell lines: correlation to etoposide and vincristine sensitivity and inverse correlation to carmustine sensitivity

SummaryThe aim of our investigations is to evaluate whether the sensitivity patterns of small-cell lung-cancer (SCLC) cell lines in vitro can be used in evaluating new drugs and in selecting drugs for the optimization of combination therapy. In our attempts to obtain a panel of cell lines demonstrating differential patterns in sensitivity, we have developed three SCLC lines exhibiting different types of multidrug resistance (MDR). In the present investigations we compared the sensitivity patterns shown by five wild-type SCLC lines and three MDR lines in response to six different types of drugs: doxorubicin, cytarabine, carmustine, cisplatin, vincristine, and etoposide. In the wild-type SCLC cell lines, the range of variation in sensitivity to all drugs was within a factor of 10. Cell lines showing low sensitivity to doxorubicin also exhibited low sensitivity to etoposide and vincristine, and vice versa. In contrast, the pattern of sensitivity to carmustine was almost the opposite of that to doxorubicin. A tendency to an inverse relationship between doxorubicin and carmustine sensitivity was also observed when doxorubicin sensitivity was reduced in near stationary cells and in cells exposed to the metabolic inhibitor 2-deoxy-d-glucose. In agreement with the pattern observed for the wild-type lines, all of the MDR sublines demonstrated collateral sensitivity to carmustine. As to cytarabine, the wild-type lines expressed a sensitivity pattern similar to that shown in response to doxorubicin. Interestingly, the opposite pattern was found in the MDR lines, as all three demonstrated cytarabine hypersensitivity. The combination of alkylating agents and “MDR” drugs are of proven clinical benefit in the treatment of solid tumors, as is the combination of anthracycline and cytarabine in acute myeloid leukemia. The experimentally derived sensitivity data on cytarabine, alkylating agents, and MDR drugs (i.e., etoposide, doxorubicin, vincristine) thus resemble the clinical experience with these drugs, and we conclude that the use of a clonogenic asay on the described panel of SCLC cell lines can give valuable information for the selection of agents for combination therapy.

Daunorubicin and vincristine binding to plasma membrane vesicles from daunorubicin-resistant and wild type ehrlich ascites tumor cells

Tumor cell resistance to anthracyclines, epipodophyllotoxins and vinca alkaloids, called multi-drug resistance (MDR) is intimately linked to changes in the plasma membrane which facilitate an increased energy dependent drug extrusion in the resistant cell compared to the wild type cell. Isolated plasma membrane vesicles from wild type Ehrlich ascites tumor cells (EHR2) and the daunorubicin (DNR) resistant subline EHR2/DNR+ were utilised to study binding and possible transport of DNR and vincristine (VCR). A significant ATP enhanced increase in VCR binding to vesicles from EHR2/DNR+ compared to EHR2 was demonstrated. Furthermore, an increase in ATP enhanced VCR binding in proportion to content of the MDR associated P-glycoprotein was seen in plasma membrane vesicles prepared from various benign human endocrine tumors. VCR binding to EHR2/DNR+ vesicles was inhibited by other vinca alkaloids greater than actinomycin D greater than colchicine greater than anthracyclines, with 35-75 microM concentrations of anthracyclines needed for 50% inhibition. VCR binding to EHR2/DNR+ vesicles was pH and temperature dependent with an activation energy of -30 kJ/mol and was decreased by replacement of Na+ with K+ and by addition of Ca2+. Preincubation of vesicles with monoclonal antibody against the C terminal of P-glycoprotein had no effect on VCR binding and osmolality tests failed to show genuine transmembranal transport of VCR. DNR binding was similar in plasma membrane vesicles from both cell lines, and showed none of the characteristics mentioned for VCR. Furthermore, a radiolabeled N-hydroxysuccinimide ester derivative of doxorubicin, which inhibited VCR binding to EHR2/DNR+ membranes to an even greater extent than doxorubicin, labeled plasma membrane proteins from EHR2 and EHR2/DNR+ identically and did not demonstrate any binding to P-glycoprotein. Therefore, even though the study confirms the close link between vinca alkaloid binding and P-glycoprotein, it could not detect a similar association between anthracyclines and P-glycoprotein thus attesting to the complexity of the MDR phenotype.

Doxorubicin induced alterations in lipid metabolism of cultured myocardial cells.

Doxorubicin (DX) was found to inhibit the incorporation of [1-14C]linoleic acid and [1(3)-3H]glycerol into the major membrane phosphoglycerides, phosphatidylcholine and phosphatidylethanolamine of cultured myocardial cells in a dose-dependent manner (0.16-16 microM). It is suggested that DX affects de novo biosynthesis of these lipids. In contrast, DX-treatment of the cells stimulated incorporation of [1-14C]linoleic acid into triacylglycerol. The effects of DX on lipid metabolism were only demonstrable 20-24 hr after a 1 hr exposure of the cells to the drug indicating that DX exerts little or no direct effect on the enzymes participating in lipid synthesis and that the alterations in lipid metabolism induced by DX probably are secondary to inhibition of protein synthesis and progressive cell injury. Extensive peroxidative decomposition of membrane lipids appeared not to take place in the DX-treated cells as judged from fatty acid analysis of total membrane phosphoglyceride.

Characterization of the cooperative cross-linking of doxorubicin N-hydroxysuccinimide ester derivatives to water soluble proteins

Protein-anthracycline interactions have been examined by using reactive N-hydroxysuccinimide ester derivatives of doxorubicin. These compounds cross-link to lysine epsilon-amino groups with high efficiency and offer the possibility for structural studies of protein-anthracycline complex formation by using gel filtration, ultracentrifugation and spectrophotometric methods. The results are in accordance with association of anthracycline to the hydrophobic ligand binding cavities of serum albumin. The results for proteins not having hydrophobic domains (IgG, serum transferrin, lactotransferrin, ovotransferrin) suggest that complex formation is cooperative and involves two steps: initial self-association of anthracycline into aggregated structures and subsequent binding of protein at the aggregate surface. With serum transferrin, anthracycline self-association makes possible the assembly of stable nanometer-sized protein-anthracycline particles held together by non-covalent bonds. This reaction, which is highly reproducible and efficient, may have applications in the field of development of anthracycline carrier systems.