Isothermal titration calorimetry (ITC) was applied to investigate the interaction of drugs with liposomes. Two types of titration are possible. One type is when the liposome suspension in the cell is titrated by aliquots of drug solution, and the other is when the drug and liposome solutions take the opposite roles. In this paper, we employed the latter type because the disturbance of liposomes may be minimal in this titration type. We derived an equation in which the accumulated heat-flow is expressed as a function of the added lipid concentration. In the derivation, the uniform binding model was used although there may be various binding sites. This equation contains a parameter n, the number of binding sites per lipid molecule. In addition, we derive the relation between the dissociation constant (Kd), partition coefficient (Pm) and n. Binding parameters such as Kd, n, the Gibbs energy change, enthalpy change and entropy change were estimated for ANS (1-anilino-8-naphtarenesulfonate), TPB (tetraphenylborate), amlodipine, nifedipine, amitriptyline, nortriptyline, imipramine, desipramine, propranolol, chlorpromazine, promethazine, miconazole, indomethacin, diclofenac and diflunisal. For some drugs, the enthalpy change was the major binding affinity instead of the classical hydrophobic interaction in which entropy takes the essential role. We proved an approximate rule that for drugs with smaller n (the number of binding sites per lipid molecule), the entropy change contributes more than the enthalpy change.
Because drug targets usually exist inside the cell, the drugs must cross various cellular barriers by passive and/or transporter-mediated transfer to elicit their pharmacological and therapeutic effects. Transport pathways across the intestinal mucosa are, for example, passive diffusion via a paracellular route, passive diffusion via a transcellular route and a protein-mediated pathway via influx and efflux transporters. Although recently many drugs were proven to be transferred via a variety of transporters [1-3], many drugs are permeated by passive diffusion. The properties of drugs that affect the passive permeation of drugs through biological membranes are lipophilicity, charge, size and hydrogen bonding ability, and the most important property among these may be lipophilicity [
Various methods for the measurement of the liposome/water partition have been developed including ultrafiltration [
In this paper, we also employed ITC to study the binding of various drugs to liposomes. Two different types of experiments are possible regarding the relative roles of the drugs and the liposome: One type is when aliquots of a drug solution are injected into a liposome suspension (type-A titration), and the other is the opposite in which aliquots of a liposome suspension are injected into a drug solution (type-B titration). In type-A titration, the ratio of the number of bound drugs to the number of lipid molecules becomes large especially near the final stage of titration, which may disturb the liposome structure or change the electrical interaction. If we stop the titration at an early stage to avoid the disturbance and the added chemicals all bind to the liposome, we may estimate ΔH but not the dissociation constant due to the lack of information about the binding saturation. Then, we employed type-B titration in which near the final stage of titration, a relatively small number of drug molecules are adsorbed onto a relatively large number of liposomes so that the membrane disturbance is minimal. We derived an equation in which the accumulated heat flow is expressed in terms of the concentration of injected lipid. The observed data were well fitted by this equation to estimate the binding parameters. According to our equation, n, the number of binding sites per lipid molecule, can be estimated, and using this parameter, the partition coefficients of 15 drugs were calculated from the respective dissociation constants. Some drugs showed enthalpy-driven binding instead of the classical hydrophobic interaction in which the entropy change is a major driving force. We proved an approximate rule that for drugs with smaller n (binding involving larger lipid molecules), the entropy changes contribute more than the enthalpy changes. The value of n may be useful for future molecular consideration of the binding.
L-α-lecithin (egg yolk phosphatidylcholine, abbreviated as PC hereafter) was obtained from Avanti Polar Lipids (Birmingham, AL). Tetraphenylborate (TPB) was purchased from Dojindo Laboratory (Kumamoto, Japan), and other chemicals of the highest purity available were from Wako Pure Chemicals (Osaka, Japan). All chemicals were used without further purification.
The liposomes of PC were prepared using the standard extrusion method: Lipids were dissolved in a round flask with chloroform, and the solvent was evaporated by a rotary evaporator under nitrogen to form the thin lipid film. To remove the solvent completely, the lipid film was put in a vacuum overnight. A buffer solution was added to the flask so that the lipid concentration became a determined value. The buffer contained 150 mM NaCl, 10 mM HEPES (4-(2- hydroxyethyl)-1-piperazineethan-esulfonic acid) and 0.1 mM EDTA (2-({2-[bis(carbomethyl)amino]ethyl} (carboxymethyl)amino)acetic acid), and the pH was adjusted to 7.5. This buffer solution is called Buffer-H in this paper. The multilamellar vesicles were formed by mechanical vibration at room temperature in an N2 atmosphere. To achieve a certain size distribution, the vesicle suspension was extruded through polycarbonate membranes by N2 gas of 2000 kPa. Extrusion was performed 5 times using a filter of 200 nm pore size, followed by 5 extrusions using a filter of 100 nm pore size. The lipid concentration was determined by phosphorus content [26, 27].
The experiments were performed with an isothermal titration calorimeter (VP-ITC MicroCalorimeter, MicrocalTM, Northampton MA). The temperature was 25˚C except in the van’t Hoff experiment. The cell (1.4 mL) contained the solution of drugs, into which the liposome suspension (typically 41 mM of PC) was injected unless otherwise noted. The drug and liposome were separately dissolved in Buffer-H. Both solutions were degassed using Thermo Vac Sample Degassing (MicrocalTM Inc., Northampton, MA). The combination of the drug concentration in the cell and the volume of aliquots of the injected liposome suspension at each titration were optimized after preliminary experiments and written in text. The drug solution in the cell was stirred at 300 rpm. The period between two successive titrations was typically 240 sec. The heats of dilution were small compared with the binding interaction and corrected. The analysis was performed using an equation derived in this paper, which will be described below. The fitting calculation was performed with Origin software (MicrocalTM Inc., Northampton, MA).
As described in the Introduction, there are two possible positions of the liposome suspension and drug solution, which are called type-A and type-B titration. Ikonen et al. [
The liposome suspension (41 mM lipid concentration) was injected into ANS or TPB solutions in the cell whose concentrations were varied as shown in the table. The injected volume of each titration was 3 mL for TPB and 8 mL for ANS, and the titration continued until no heat was produced (except the dilution heat). The buffer solution was Buffer-H (pH 7.5). The temperature was 25˚C. The listed values were averaged and deviated for 5 independent experiments.
method, but this method has a drawback. Near the endpoint of titration, the maximum binding of drugs to liposomes is attained. Near this point, many drug molecules bind to the liposomal membrane so that the membrane is disturbed from the natural state. However, for type-B titration, more liposome vesicles are present in comparison with the drug molecules (except for the initial stage) so that the membrane is not disturbed. Therefore, we employed type-B titration here. This method has been used by previous authors [17-19].
This estimation of ΔH was applied to the binding of TPB and ANS. The PC liposome (41 mM of lipid) was injected with 3 µL of each titration. We set the TPB con-
centrations in the cell to 10, 25 and 35 µM to ascertain the independence of the drug concentration in the cell. The estimated values were 9.4 kcal/mol for 10 µM, 9.3 kcal/mol for 25 µM and 9.0 kcal/mol for 35 µM, which reveals that the constant is independent of the drug concentration. Similar experiments were performed for ANS, and the values were 7.8, 7.5 and 7.6 kcal/mol, respecttively, for 10, 25 and 35 µM in the cell. Although the estimated ΔH values seem constant, the “true” values might be slightly different from the estimated values. Because of the presence of free drug molecules in excess of liposomes, the “true” value might be larger than the estimated value.
The heat flows at every titration were accumulated or integrated from the first injection, and the accumulated values (ΔQ) were plotted against the concentration of the titrant (lipid) in the cell, CL. One typical result is shown in
We assume the following equilibrium equation holds:
In this equation, = the dissociation constant, = the concentration of unbound sites on the liposome, = the concentration of free chemicals in the aqueous solution, and = the concentration of bound chemicals. Although there may be various binding sites on the liposomal membrane with different dissociation constants, we assume that there is only one type of binding site. Then, the value in Eq.1 may be an averaged value of the various binding sites. denotes the total lipid concentration that has been injected in the cell, and is the number of binding sites per lipid molecule. The mass balance equation of the binding sites is:
Elimination of from Eqs.1 and 2 and rearrangement yield:
The total concentration of the chemicals in the cell, , is represented by:
The accumulated heat flow, ΔQ, should be represented by:
Here, represents the cell volume (1.4 mL).
From Eqs. 3-5, we obtained:
In this equation, is expressed as a function of. By a non-linear least square method, we can estimate the values of, and n. Note that is given. As described below, the value of n, the number of binding sites per lipid molecule, is essential for the estimation of the partition coefficient (see Eq.9). The comparison of Eq.6 with the experimental data is shown in
When the temperature was changed from 10˚C to 40˚C, ANS of 35 µM was titrated with 8 µL of 41 mM PC liposome, and and ΔH were estimated at various temperatures. The results are shown in panel A of
For the discussion of the interaction of drugs with the liposomal membrane, partition coefficients are frequently used instead of Kd. Therefore, we consider the evaluation of the partition coefficient of the drug/liposome, from. Because is the concentration of the bound drug in the cell (the volume of which is (1.4 mL)), the total amount of the bound drugis, and the concentration within the liposomal membrane is repressented by, where represents the lipid volume in the cell and equals Here a is the lipid volume per mole of lipid, and 0.755 mL/mmol is reported (27). Therefore, the partition coefficient should be represented by:
Because the partition coefficient is considered at the concentration range where the binding is proportional to the concentration, i.e., where the approximation holds, Eq.3 can be approximated to:
Inserting Eq.8 into Eq.7, we obtain the following simple equation:
Note that is dimensionless, and thus, the partition coefficient derived is also dimensionless. Eq.9 contains n, the number of binding sites per lipid molecule.Then, is not proportional to, the binding constant. For example, as shown below in
The partition coefficients were determined for various drugs, and the results are listed in
The drug concentrations in the cell were 30 µM except for TPB and ANS. Each injection volume was 5 µL of 41 mM lipid suspension. The buffer was Buffer-H (pH 7.5), and the temperature was 25˚C. For TPB and ANS, the averaged values of
azepine. The ITC method has a shortcoming in that it is not applicable to chemicals with small enthalpy changes of binding.
In this table, Pm(ref) represents the values taken from the references in which the partition coefficients to liposomal membranes were experimentally determined, while the logPoct values (the partition coefficients of octanal/water) are obtained from experimental values written in the DrugBank web site [
The thermodynamic quantities of the binding, such as the Gibbs energy change, and the entropy change, were calculated by the thermodynamic equations.
The data listed in
As shown in the Appendix, desipramine and nortrip-
tyline are demethylated derivatives of imipramine and amitriptyline, respectively. For these two combinations, the demethylated compounds with the larger Kd values have larger ΔH absolute values, smaller TΔS absolute values and smaller n. Interestingly, only one replacement of CH3 with H influences the thermodynamic quantities. Amlodipine and nifedipine are homologs. As shown in
In this paper, we applied ITC to investigate the interaction between liposomes and drugs, and the liposome suspension was injected into the drug solution (type-B titration). In this method, the liposomes were present in excess amounts of the drug except in the early stage. Then, the disturbance of liposomes may be small. Data were analyzed with Eq.6, and this method gave the ordinary binding parameters such as Kd(ΔG) and ΔH, in addition to n, the number of binding sites per lipid molecule. The partition coefficients were calculated using n and agreed well with those reported values of liposomes or the octanol/water partition coefficients, except for some drugs (amlodipine, nifedipine, miconazole and diflunisal). For many drugs, the binding is entropydriven according to the hydrophobic interaction, but for some (TPB, ANS, amlodipine, desipramine and diflunisal), the enthalpy is the major driving force (non-classical hydrophobic interaction [
ITC, isothermal titration calorimetry; Kd; dissociation constant; n, the number of binding sites per lipid molecule; Pm, partition coefficient of drug/liposome; ANS, 1-anilino-8-naphtarensulfonate; TPB, tetraphenylborate; type-B titration; aliquots of a liposome suspension are injected into a drug solution; PC, egg yolk phosphatidylcholine; Buffer-H, a buffer solution containing 150 mM NaCl, 10 mM HEPES (pH 7.5) and 0.1 mM EDTA; CL, lipid concentration in the cell during a titration; Ct total drug concentration (bound and free) in the cell; Poct, partition coefficient of octanol/water.
Chemical structures used in this experiment.
Tetraphenylborate (TPB) 1-Anilino-8-naphtarenesulfonate (ANS)
Amlodipine Nifedipine
Amitriptyline Nortriptyline
Imipramine Desipramine
Propranolol Chlorpromazine
Promethazine Indomethacin
Diclofenac sodium Diflunisal
Miconazole