Thermodynamics and kinetics of P-glycoprotein-substrate interactions

P-glycoprotein (Pgp, ABCB1) is a transmembrane protein, which extrudes a large number
of structurally diverse compounds out of the cell membrane at the expense of ATP
hydrolysis. The overexpression of Pgp strongly contributes to multidrug resistance, which
hampers the chemotherapy of cancer and some other drug-treatable diseases. Therefore, the
general aim of this thesis was to quantitatively characterize the thermodynamics and the
kinetics of Pgp-substrate interactions. Specific emphasis was placed on the understanding
the influence of the lipid bilayer since Pgp binds its substrates from the cytosolic leaflet of
the membrane.
Drug binding to Pgp can be divided into two steps. The first step is the drug
partitioning into the lipid bilayer described by the lipid-water partition coefficient, Klw, and
the second step is the drug binding to Pgp from the lipid phase depicted by the binding
constant of drug to Pgp from the lipid phase, Ktl. The binding constant of the drug from the
aqueous phase to Pgp, Ktw, is thus the product of these two binding constants. Since the
separation of drug binding to the lipid bilayer and to Pgp from the lipid bilayer has not
been done satisfactorily a question mark was still hanging above the strength of the Pgpdrug
interactions. Therefore, the aim of the first part of this thesis was to quantify the
binding constant of drugs to Pgp from the lipid phase, Ktl and furthermore to test the
energetic soundness of the previously suggested hydrogen bond hypothesis. The lipidwater
partition coefficient, Klw, and the drug binding constant from water to Pgp, Ktw, were
determined independently for the 15 drugs and for the first time the binding constant of
drugs to Pgp from the lipid phase and the corresponding free energies 0
ΔGtl were reported.
We found that the free energy of lipid-water partitioning 0
lw ΔG is more negative than the
free energy of drug-binding to Pgp from the lipid phase 0
tl ΔG for all 15 drugs studied but
0
tl ΔG varied more strongly than 0
lw ΔG . This suggests that the drug interactions to Pgp are
weaker but more specific than to the lipid bilayer. Furthermore, the higher concentration of
drugs in the membrane than in the aqueous phase at the concentration of half-maximum
Pgp activity supports the relative weak interactions. The results are well explained with the
hydrogen bond hypothesis and they indicate the existence of large drug binding region(s) in
transmembrane domains of Pgp. In addition, we demonstrated that the binding constant of
drug to Pgp from the aqueous phase, Ktw, depends exponentially on the lateral packing
density of the membrane, πM. This finding can explain the diverse concentrations of half
maximum Pgp activity, K1, reported in literature for the same drug but determined in the
different membrane environments.
The common aim of the next three parts of this thesis was to elucidate Pgp activity
in the cellular ensemble as well as in an isolated environment. This was done by measuring
the Pgp activity for 15 structurally diverse Pgp substrates in living MDR1-transfected
mouse embryo fibroblasts by monitoring the extracellular acidification rate, ECAR, and in
plasma membrane vesicles formed from the same cells by monitoring the rate of ATP
hydrolysis. Because these two systems exhibit several differences such as the direction of
drug approaching Pgp, it was interesting to compare the kinetics of Pgp activity measured
in both systems. We found that the concentrations of half-maximum Pgp activation, K1, for
15 drugs were identical as long as the pH of the environment was identical and the possible
artifacts such as vesicle and the drug association in the phosphate release measurements
and cytotoxic effects of drugs in the ECAR measurements could be excluded. We,
therefore, concluded that whether the drugs approach Pgp from the extracellular or
cytosolic side of the membrane does not play a role as long as the drugs can deprotonate
and cross the membrane in their uncharged form. A reasonably linear correlation was also
found for the relative maximum Pgp activities, V1. The observed outliers were mainly due
to artifacts. In order to compare the absolute rate enhancements stimulated by various
drugs the turnover numbers were estimated. The drug-stimulated turnover numbers were in
good agreement if the Pgp activity in living cells was measured in the presence of
pyruvate. In the absence of pyruvate the turnover numbers in living cells is however
higher. Furthermore, it was found that the logarithm of maximum Pgp activity, lnV1,
decreases linearly with decreasing free energy of drug binding to Pgp from the aqueous
phase, 0
tw(1) ΔG . Thus, we concluded that the drug release to the extracellular leaflet of the
lipid bilayer is the rate-determining step for Pgp activity.
The previous studies about the influence of the metabolic state of the cells on Pgp
activity revealed that the verapamil-stimulated ECAR was linearly correlated with the basal
ECAR under the nutritional condition where the metabolic state of cells was low. However,
when the metabolic state of cells was high (in the presence of glucose) the drug-stimulated
ECAR flattened out. The increase in the drug-stimulated ECAR was suppressed compared
to the increase in the basal ECAR. To better understand the influence of the metabolic state
of the cell on the basal and the drug-stimulated ECAR we further studied the metabolic rate
of mouse embryo fibroblasts by measuring the glucose consumption and the lactate
production rate by means of 13C-NMR spectroscopy. We found that the relative verapamil-stimulated enhancement in the metabolic rate of NIH-MDR1-G185 cells was in good
agreement with the values obtained from the ECAR measurements upon longtime
verapamil stimulation. However, the 13C-NMR measurements revealed that the absolute
metabolic rate of wild-type and MDR1-transfected cells in the absence of drugs were about
three-fold higher than those observed by monitoring the ECAR of the same cells. Also the
absolute metabolic rate of MDR1-transfected cells stimulated with verapamil was distinctly
higher. Furthermore, it was demonstrated by separating the intra- and the extracellular
lactate NMR signals with a novel shift reagent that the lactate concentrations inside and
outside of MDR1-transfected cells were similar. These results support the assumption that
the flattening observed between the basal and the verapamil-stimulated ECAR under
conditions of high ECARs, is due to the suppression of glycolysis. The experiments further
indicate that it is not due to the saturation of monocarboxylate transporters.
The drug transport activity of Pgp is coupled to ATP hydrolysis but how these two
steps are connected is still a matter of debate. Therefore, the goal in the fifth part of this
thesis was to study the interplay between the transmembrane domains, TMDs, of Pgp,
which bind and release the drugs, and the nucleotide binding domains, NBDs, of Pgp,
which bind and hydrolyze ATP. Moreover, we aimed at getting additional information
about the influence of the lateral membrane packing density, πM, on the transition state
parameters of ATP hydrolysis by Pgp (the activation enthalpy ΔH‡ , the activation entropy
ΔS ‡ , and the free energy of activation ΔG‡ ) and, thus, on the mechanism of Pgp activity.
For this purpose we measured the temperature dependence of steady-state Pgp activity in
the plasma membrane vesicles of NIH-MDR1-G185 cells in the absence and presence of
several exogenous Pgp substrates and analyzed the transition state parameters for ATP
hydrolysis by Pgp (the activation enthalpy ΔH‡ , the activation entropy ΔS ‡ , and the free
energy of activation ΔG‡ ). We found that the activation enthalpy, ΔH ‡ , of ATP hydrolysis
by Pgp was relatively large for the basal and drug-stimulated Pgp activity and decreased in
the presence of all drugs investigated, even in the presence of drugs that reduce Pgp
activity. The free energy of activation, ΔG‡ , in contrary, decreased at low and increased at
high drug concentrations of verapamil and promazine, whereas it only increased for
PSC833. Thus, by plotting the free energy of activation, ΔG‡ , as function of drug
concentration revealed the mirror images of Pgp activation profiles, in good agreement
with the Eyring’s transition state theory. Furthermore, by comparing the transition state
parameters for basal Pgp in plasma membrane vesicles of NIH-MDR1-G185 cells with the
literature data measured in different membrane environments, we found that the activation
enthalpy, ΔH‡ , of ATP hydrolysis by Pgp decreases significantly with decreasing lateral
membrane packing density, πM, whereas the free energy of activation, ΔG‡ , decreased only
slightly. The Pgp activity cycle contains several steps and each step has its own free energy
of activation, ‡
i ΔG . The sum of these individual free energies is the free energy of
activation, ΔG‡ , obtained from the steady-state Pgp activity measurements. The
experimental data revealed that the different drugs affected the free energy of activation,
ΔG‡ , in characteristic manner. However, whether the rate-limiting step is the ATP
hydrolysis or the conformational change of Pgp could not be decided conclusively from the
present data and further experiments are needed.
In the last part of this thesis we aimed at identifying whether the dyes, methylene
blue, acridine orange, basic fuchsin, and ethyl eosin are substrates for Pgp. This is relevant
since they are widely used in histological studies. Furthermore, their applicability for the
treatment of cancer and microbial infections with photodynamic therapy has been tested. In
accordance with the established rules for the intrinsic Pgp substrates we predicted that
methylene blue, acridine orange and basic fuchsin are intrinsic substrates for Pgp whereas
ethyl eosin is not. The Pgp activity measurements in plasma membrane vesicles and in
living MDR1-transfected mouse embryo fibroblasts supported the prediction. Furthermore,
the Pgp activity measurements in living cells revealed that methylene blue, acridine orange
and ethyl eosin influenced the energy metabolism of wild-type and MDR1-transfected
mouse embryo fibroblasts in a complex manner. Thus, the evaluation of Pgp activity in
living cells gave further valuable information about the cytotoxicity of drugs.
In conclusion. We evaluated the thermodynamics and the kinetics of Pgp-substrate
interactions. Since Pgp binds its substrates from the lipid phase the influence of the lipid
bilayer on both of them was emphasized. The intrinsic drug binding constants, the binding
constants of drug to Pgp from lipid phase, Ktl, for the 15 Pgp-substrates were evaluated
quantitatively for the first time and revealed the weak interactions between Pgp and its
substrates. The thermodynamic evaluation further indicated that the hydrogen bonds
contribute significantly on the energetics of Pgp-substrate interactions supporting the
hydrogen bond hypothesis and the existence of large drug binding region(s). Furthermore,
it was shown that the concentration of half-maximum Pgp activity, K1, of a specific drug
depends on the lipid-water partition coefficient of the drug, Klw, which in turns depends on
the lateral packing density of the membrane, πM, and therefore is dependent on the lipid
membrane where Pgp activity is determined. The rate of Pgp activity evaluated in living
cells and in inverted plasma membrane vesicles revealed that Pgp activity is controlled by
the binding affinity of the drug to Pgp from the aqueous phase, 0
tw(1) ΔG . In living cells the
metabolic state of cells further affected the rate of Pgp activity. By evaluating the transition
state parameters from the temperature dependence of Pgp activity in the absence and
presence of Pgp substrates and comparing the data with the available literature values
suggested that the membrane packing density seems to slightly affect the basal Pgp
activity, whereas the effective drug transport rate seems not to be affected. Moreover, the
mode of Pgp action changed from entropy-driven to enthalpy-driven in the membrane with
decreasing membrane packing density suggesting that Pgp and the membrane form a
functional unity.