Ligand self-association at the surface of liposomes: A complication during equilibrium-binding studies

Ligand self-association at the surface of liposomes: A complication during equilibrium-binding studies

ANALYTICAL BIOCHEMISTRY 143, 135-140 (1984) Ligand Self-Association at the Surface of Liposomes: A Complication during Equilibrium-Binding Studies ...

507KB Sizes 0 Downloads 25 Views

ANALYTICAL

BIOCHEMISTRY

143, 135-140 (1984)

Ligand Self-Association at the Surface of Liposomes: A Complication during Equilibrium-Binding Studies THOMAS G. BURKE AND THOMAS R.TRITTON' Department

of Pharmacology,

Yale University

School

of Medicine,

New Haven,

Connecticut

06510

Received June 20, 1984 Daunomycin and carminomycin, two anthracycline antibiotics known to bind phospholipid bilayers, appear to self-associate at the surface of liposomes at high bound drug/lipid ratios (r). Pluorescence intensity, lifetime, and anisotropy measurements have been used to monitor the equilibrium binding of these drugs to small unilamellar solid-phase dipalmitoylphosphatidylcholine vesicles. Association of an anthracycline with excess liposome (low r) resulted in an increase in both the observed intensity and the fluorescence lifetime. At low vesicle concentrations (high r), a decrease in the total emission intensity was observed which was not paralleled by the excited-state lifetime. The data from these experiments are consistent with the formation of nonfluorescent anthracycline complexes at the surface of liposomes. Such ligand self-association is a potential complication in any studies on the interaction of amphipathic molecules with liposomes conducted at high r values. Because ligand self-association limits the collection of binding data over certain concentration ranges, this consequently results in greater uncertainty in the determination of the maximum value of r (n) in equilibrium binding studies. io 1984 Academic

Press. Inc.

KEY WORDS:

drug-membrane

fluorescence; anthracyclines; self-association: liposomes; equilibrium interactions.

Phospholipid vesicles or liposomes have been widely used as a model system for characterizing drug-membrane interactions. Equilibrium-binding studies are one such application. In such experiments, it is common for a fixed amount of drug to be titrated with increasing concentrations of sonicated liposome solution with the extent of binding being monitored by changes in an optical property of the drug such as absorbance, fluorescence intensity, or fluorescence anisotropy. The importance of covering a wide range of concentrations in determining the

binding;

number of binding sites per mole of receptor (n) from Scatchard plots has recently been emphasized (1). The anthracycline antibiotics have previously been shown to bind to dipalmitoylphosphatidylcholine (DPPC)* bilayers (2,3). While using fluorescence titration to study the interaction of two anthracyclines, daunomycin and carminomycin, with sonicated DPPC liposomes, we observed phenomena at low vesicle concentrations inconsistent with simple uncomplicated binding equilibria. Because daunomycin is known to self-associate in aqueous solution (4) we reasoned that the drug molecules might also aggregate on the surface of a phospholipid bilayer, even though the concentrations for free drug association are greater than those used in the present experiments. Our findings demonstrate, in the case of the anthracyclines and possibly other amphipaths (especially those known to self-associate in aqueous solution), the limitations of the liposome

’ To whom correspondence should be addressed. * Abbreviations used: DPPC, L-cudipalmitoylphosphatidylcholine; n, maximum number of binding sites per lipid; PBS, phosphate-buffered saline containing I37 mM NaCl, 3 mM KCl, 8 mM NqHPO,, and I mM KH2P04 (pH 7.4); &,,, phase transition temperature; a. anisotropy; Cr, free drug concentration; r, number of bound drug molecules per lipid; &,,, apparent binding constant; F, fluorescence intensity; 7, fluorescence lifetime. 135

0003-2697184 $3.00 Copyright 8 1984 by Academic Pws, Inc. All rights of reproduction in any form reserved.

136

BURKE

AND TRITTON

system in obtaining the necessary range of measurements for the unambiguous determination of n. We report here fluorescence experiments which strongly suggest that selfassociation of anthracyclines occurs on the surface of solid-phase DPPC liposomes, and demonstrate how such ligand self-association affects the analysis of the binding data. MATERIALS

AND METHODS

Chemicals. Daunomycin was the gift of Dr. Leonard H. Kedda of the Division of Cancer Treatment, National Cancer Institute. Carminomycin was provided by Bristol Laboratories, Syracuse, New York. Both of the anthracyclines were in their hydrochloride form and were used without further purification, since thin-layer chromatography analysis on silica gel, using a solvent system ofchloroform / methanol/water (40 : 10 : l), showed them to be >99% pure. Stock solutions of the anthracyclines were prepared in methanol and stored in the dark at -20°C. Drug concentrations were determined by absorbance measurements using the following wavelengths and extinction coefficients: daunomycin, 480 nm, 11,500 M-’ cm-’ (4); and carminomycin, 490 nm, 8200 M-’ cm-’ (5). L-cY-Dipalmitoylphosphatidylcholine (DPPC) and glycogen were obtained from Sigma Chemical Company and were used without further purification. Liposome preparation. Small unilamellar liposomes were prepared the day of an experiment by the method of Poste et al. (6). Lipid was weighed and suspended in phosphate-buffered saline (PBS), pH 7.4, at a concentration of 5 mg/ml. Lipid dispersions were prepared by vortex mixing for 5- 10 min above the Thl of the lipid. The lipid dispersion was then subject to ultrasound with a Laboratory Supplies Company (Hicksville, N. Y.) bath-type sonicator for 3-4 h until no further optical clearing of the solution was observed. TLC analysis on silica gel, using a solvent system of chloroform/methanol/acetic acid/water (25: 15:4:2), revealed

that no decomposition of the lipid had occurred during sonication. The sonicated lipid solution was annealed above the TM for 0.51 h and then cooled to ambient temperature prior to use. Differential scanning calorimetry studies of the sonicated DPPC vesicles showed a main phase transition centered at 36.9”C in agreement with results from Suurkuusk et al. (7). Fluorescence instrumentation. All fluorescence measurements were obtained using a SLM Model 4800 subnanosecond spectrofluorometer with a thermostated cuvette compartment. This instrument was interfaced to a Hewlett-Packard 9825 data processor. Excitation and emission spectra were recorded in the ratio mode using an excitation resolution of 4 nm and an emission resolution of 8 nm. Steady-state fluorescence intensity measurements were made in the ratio mode without polarizers. Steady-state anisotropy (a) measurements were determined with the instrument in the “T-format” for simultaneous measurement of two polarized intensities. The alignment of the polarizers was

routinely checked using a 2 mg/ml suspension of glycogen in water (anisotropy values of >0.99 were obtained). Fluorescence lifetimes were measured using the phase-modulation technique of Spencer and Weber (8). Lifetimes were determined by phase shift using exciting light modulated at 30 MHz. A glycogen scattering solution (25 mg/ml) was used as reference and the amount of scattered light was adjusted using neutral density filters (Schott) such that signals from both the scatterer and the sample were approximately equal. Intensity, anisotropy, and lifetime measurements were conducted using an excitation wavelength of 470 nm and slitwidths of 4 nm, two 500-nm short-wavepass filters (Melles Griot) in the excitation beam to prevent transmission of stray light from the excitation monochromator and a 550~nm long-wavepass filter (Schott) for the emission channel(s) in order to isolate fluorescence from scattered light. All experiments were conducted in l-cm quartz cuvettes. The

LIGAND

SELF-ASSOCIATION

AT THE SURFACE OF LIPOSOMES

background fluorescence and scatter from unlabeled lipids or from solvents was less than 1% of the total intensity for all of the measurements. Equilibrium-binding measurements. Samples were prepared by adding liposome solution to tubes containing identical amounts of a stock solution of drug in PBS such that the final drug concentration was 1 X 10e6 M. The samples were equilibrated for 1 h prior to measurement. Fluorescence anisotropy titration was used to determine the concentration of free and bound drug according to a = .f&F + f&B + f&s,

111 where a equals the measured anisotropy and ar, an and as are the anisotropies of free drug, bound drug, and scatterer, respectively. The terms fr, fn, and fs refer to the fraction of the total fluorescence signal which originates from free drug, bound drug, and background scatter, respectively, and fF + fn + fs = 1. In all cases,fs did not exceed 1% of the total signal. The value of as is dependent upon the turbidity of the sample and the instrumental conditions (i.e., excitation wavelength, emission filters) used for a particular measurement. Using the same experimental conditions, the value of as was evaluated by direct measurement of the scatter from a buffer solution or vesicle preparation without fluorophore. Alternatively, as was determined by increasing the amount of background scatter through dilution and monitoring the change in the measured anisotropy. Results from both approaches were in close agreement. The anisotropy value for free drug was determined from the simplified form of Eq. [l] by using a buffer blank to determine fs. The bound anisotropy value was determined by extrapolation to infinite total lipid concentration from a double-reciprocal plot of l/a vs l/total lipid concentration, using only the samples where >80% of the drug was bound. In determining the amount of bound drug during a titration, it was assumed that the termfsas increased linearly with increasing lipid concentration, and values for this term

137

at the minimum and maximum lipid concentration were determined for each experiment. Eq. [l] was solved for fr in terms of& and the absolute amount of each species was computed based upon the total drug concentration and the factor by which the fluorescence intensity of the drug increased upon binding. Analysis of binding data. Plots of Cflr vs Cr, where Cr represents the concentration of free drug and r represents the concentration of bound drug/total lipid concentration, were used to evaluate binding constants and site stoichiometries (9). For a single class of binding sites, or multiple classes of independent, noninteracting sites with identical binding constants, this plot gives a straight line with slope of l/n and y-intercept of l/(nK,,), where n is the maximum number of binding sites per lipid molecule and Kapp is the apparent binding constant. The best linearleast-squares fit of G/r vs Cr was obtained, from which n and Kapp were calculated. All calculations were carried out on a Northstar Horizon microcomputer equipped with a hardware floating-point processor. RESULTS

Figure 1 shows the fluorescence emission spectra of daunomycin free in solution and bound to solid-phase DPPC bilayers. Assuming a value of 3000 lipid molecules per liposome (lo), there was approximately one daunomycin molecule bound per liposome under these conditions. The increase in fluorescence intensity at wavelengths greater than 580 nm is taken to indicate the association of a monomeric daunomycin molecule with the bilayer. Also accompanying the binding of daunomycin to the DPPC bilayers was an increase in the fluorescence lifetime from 1.2 to 1.6 ns. The emission spectra of carminomycin occur in the same spectral region as that for daunomycin, and a similar increase in intensity and lifetime was observed upon binding. Daunomycin has been shown to self-associate in aqueous solution at concentrations

138

BURKE AND TRITTON

0

450

500 WAVELENGTH

550

600

(ml

1. Fluorescence excitation and emission spectra of daunomycin. The relative fluorescence intensity of 2 X 1O-6 M drug in PBS buffer, 23”C, is shown. The uncorrected excitation spectra of free drug (a) was recorded with &,,, = 592 nm. The uncorrected emission spectra for free drug (b) and for bound drug in a solution containing small unilamellar DPPC vesicles at a lipid concentration of 6.5 mM (c) were recorded with &,, = 470 nm. The lifetime of daunomycin increased from 1.2 to 1.6 ns upon binding. A further increase in the vesicle concentration did not result in changes in the observed intensity and lifetime, indicating essentially complete binding of the drug at these concentrations. FIG.

in excess of 1 X 10e5 M (4). Upon the selfassociation of daunomycin in solution, the apparent extinction coefficient was observed to decrease. Since the drug concentrations in our experiments are lo-fold less than that required for self-association in aqueous solution, and the fluorescence intensity of carminomycin and daunomycin in phosphatebuffered saline, pH 7.4, at 5°C was shown to vary linearly with concentration from 6 X lO-7 to 2 X lO-6 M (data not shown), we assume that our experiments are free of selfassociation of anthracycline in solution. The evidence for anthracycline self-association at the surface of liposomes comes from the fluorescence intensity titration of a fixed amount of daunomycin or carminomycin by DPPC liposomes at 5°C as shown in Fig. 2. F,, and 7. refer to the initial fluorescence intensity and lifetime of the drug in solution, respectively, such that the ratios F/F0 and r/r0 represent the relative change in intensity or lifetime with varying total lipid concentration or [DPPC]. Evident

for both drugs is the initial decrease in the relative intensity upon the introduction of DPPC liposomes into the solution. The relative intensities for daunomycin and carminomycin reach a minimum and then gradually increase and level off at values greater than unity with increasing lipid concentration. In striking contrast to the trend in the relative intensity data, the relative lifetime data for both drugs do not show a reduction with the addition of liposomes. Instead, the T/T~ values, increase gradually and finally level off with increasing liposome concentration. These results are consistent with the formation of nonfluorescent anthracycline aggregates at the surface of liposomes when the number of bound drug molecules per liposome is high (low vesicle concentrations). An alternative explanation would be that the liposomes promote the self-association of these compounds in solution, but this possibility is unlikely because of the dilute concentrations of liposome used. Iodide quenching experiments with daunomycin and carminomycin bound to excess solid-phase DPPC liposomes have shown that the fluorophores are accessible to membrane-impermeable iodide, indicating that the bound anthracyclines are located at the lipid-solvent interface (Burke and T&ton, unpublished results). Self-association of the anthracyclines at the bilayer appears to be the most plausible explanation for the observed reduction in fluorescence intensity since the apparent extinction coefficient of daunomycin was observed to decrease upon self-association in solution (4). In this manner, aggregation of drug molecules at the surface of liposomes with consequent self-quenching removes a fraction from observation. The observed fluorescence is from either free fluorophore or unaggregated bound fluorophore. Since the bound drug has a greater lifetime than that of free drug, the lifetime measurements should reflect only an increase in the relative value until the drug is completely bound, which was experimentally observed. The re-

LIGAND

.50

5

1.0 [DfPCl.

SELF-ASSOCIATION

2.0 mM

139

AT THE SURFACE OF LIPOSOMES

.5 I 0

2

.6

4 [DPPC].

mM

FIG. 2. Fluorescence intensity and lifetime titration of daunomycin (A) and carminomycin (B) binding to DPPC liposomes at 5°C. A fixed amount of drug (1 X low6 M) was titrated with solutions of increasing lipid concentration ([DPPC]). F. and r0 refer to the initial intensity and lifetime of the drug, respectively. F and T refer to the measured intensity and lifetime at a given lipid concentration. The r,, values for daunomycin and carminomycin were I .2 ns and 1.9 ns, respectively.

covery of the relative intensity to values section). The dashed line of Fig. 3 thereby represents a hypothetical curve which the greater than unity at high lipid concentrations is due to the redistribution of bound drug data might have followed had self-association from the nonfluorescent aggregated form to not occurred at high r values. the fluorescent monomeric form. In order to demonstrate how ligand self020 association is manifested in the binding data, we conducted binding studies of carminomycin to DPPC liposomes at 8°C. Titration by fluorescence anisotropy proved to be a 015 sensitive method because of the 5-fold increase in anisotropy which accompanied binding. The range of experimental binding data can be shown by the construction of a - ‘lo r vs log Cr plot, where Cr is the concentration of free fluorophore and r is the number of 005 drug molecules bound per total lipid ( 1). This type of plot is expected to yield an Sshaped curve, approaching n as log Cf approaches infinity with an inflection point at 1 1 -70 -6.0 -6.5 lt,,2. Figure 3 is such a plot for the carmilog ct nomycin-DPPC binding data. A deviation FIG. 3. A r vs log Cr plot for the equilibrium binding from the expected S-shaped curve occurs at of I x 10m6 M carminomycin to DPPC liposomes at high r values which we attribute to the self- 8°C. The fraction of bound drug was determined by association of some of the bound carminofluorescence anisotropy titration. The inset shows how mycin. The inset of Fig. 3 demonstrates the the relative fluorescence intensity varies with lipid conreduction in fluorescence intensity which ac- centration. F/F0 values less than I indicate that significant companies aggregation. Using the data from amounts of drug are self-associated at the liposome surface. The solid curve is based upon the data where the lower Cr region of the S-curve, a Cflr vs self-association is minimal. The dashed line represents a C, plot yielded n = 0.017 and K,,, = 2.2 hvoothetical curve that the data mav have followed had X lo4 M-’ (see the Materials and Methods self-association not occurred at high r.

140

BURKE

AND

DISCUSSION

This study has presented evidence that the anthracyclines daunomycin and carminomycin self-associate at the surface of liposomes at high r values. Other investigators have reported difficulties with ligand selfassociation while conducting equilibriumbinding studies at high r values (11,12). In both cases, however, the problem was attributed to ligand aggregation in solution. The anthracycline-binding data reported here is best explained by ligand self-association at the liposome surface. Such ligand self-association can be expected to be a problem in studies on the interaction of other amphipathic molecules with liposomes conducted at high r values, especially those amphipaths known from solution studies to have relatively high self-affinities. When this self-association occurs, the system is not amenable to the collection of binding data at high r values, and this limitation results in greater uncertainty in the determination of n, the number of binding sites per lipid molecule. ACKNOWLEDGMENTS We thank Marc Adler for many useful discussions concerning this work and especially for his assistance with computer graphics (13). This work was supported

TRITTON

by NIH (CA28852). T.R.T. is the recipient of a Research Career Development Award (CAO0684) and T.G.B. was supported by a NIH Training Grant (CAO9085).

REFERENCES

4. 5. 6. I. 8. 9.

Klotz, I. M. (1982) Science (Washington, D. C.) 217, 1247-1249. Rarczmar, G., and T&ton, T. R. (1979) Biochim. Biophys. Acta 557, 306-3 19. Goldman, R., Facchinetti, T., Bach, D., Raz, A., and Shinitzky, M. (1978) B&him. Biophys. Acta 512,254-269. Chaires, J. B., Dattagupta, N., and Crothers, D. M. (1982) Biochemistry 21, 3927-3932. Brazhnikova, M. G., Zbarsky, V. B., Ponomarenko, V. T., and Potapova, N. P. (1974) J. Antibiot. 27, 254-259. Poste, G. E., Papahadjopoulas, D., and Vail, W. J. ( 1976) Methods Cell Biol. 14, 33-7 1. Suurkuusk, J., Lentz, B. R., Barenholz, Y., Biltonen, R. L., and Thompson, T. E. (1976) Biochemistry 15, 1393-1401. Spencer, R. D., and Weber, G. (1969) Ann. N. Y. Acad. Sci. 158, 361-376. Rlotz, I. M., and Hunston, D. L. (1971) Biochemistry 10, 3065-3069.

10. Huang, C., and Mason, J. T. (1978) Proc. Natl. Acad. Sci. USA 75, 308-3 10. I 1. Witzke, N. M., and Bittman, R. (1984) Biochemistry 23, 1668-1674. 12. Kauffman, R. F., Chapman, C. J., and Pfeiffer, D. R. (1983) Biochemistry 22, 3985-3992. 13. Adler, M. (1982) Comput. Programs Biomed. 15, 133-140.