- Email: [email protected]
Interaction of Linear Mono- and Diamines with Dimyristoylphosphatidylcholine and Dimyristoylphosphatidylglycerol Multilamellar Liposomes
Archives of Biochemistry and Biophysics Vol. 382, No. 2, October 15, pp. 224 –231, 2000 doi:10.1006/abbi.2000.2014, available online at http://www.idealibrary.com on
Interaction of Linear Mono- and Diamines with Dimyristoylphosphatidylcholine and Dimyristoylphosphatidylglycerol Multilamellar Liposomes Federico Momo,* ,† ,1 Sabrina Fabris,* and Roberto Stevanato* *Department of Physical Chemistry and †Istituto Nazionale Fisica della Materia, University of Venice, Dorsoduro 2137, 30123 Venice, Italy
Received March 28, 2000, and in revised form July 12, 2000
The effect of linear monoamines on dimyristoylphosphatidylglycerol and dimyristoylphosphatidylcholine multilamellar liposomes was studied as a function of their length and compared with the behavior of linear carboxylic acids. The role of the hydrophobic interactions was demonstrated and the free energy of the binding for each interacting carbon atom was determined. The thermotropic behavior of the liposomes was characterized by differential scanning calorimetry and it was shown that these molecules affect the temperature and the cooperativity of the gel to fluid state transition of the membrane differently. In particular, it appeared that membrane perturbation was maximum when the chain length of the amphipathic molecules ranged between 7 and 9 carbon atoms, with more pronounced effects in the case of monoamines. Molecules shorter than 3– 4 carbon atoms did not produce any observable change in the transition temperature. The study was extended to linear ␣,-diamines to investigate the amphipathic character of long diamines and to investigate the role of bridging bonds established with neighboring phospholipids. © 2000 Academic Press
Key Words: amines; phospholipids; liposomes.
The interaction of small amphipathic molecules with biomembranes is of wide interest (1–10) because these molecules may influence the cellular processes in a number of ways, all correlated to their lipid solubility and to the modifications they are able to produce in the lipid organization of the membrane bilayer such as the
induction of interdigitated phases or later phase separation phenomena (11, 12). It is well known, for example, that, in biomembranes, a largely cooperative behavior between proteins and their surroundings exists and that the protein-specific membrane functions can be affected by agents that do not bind specifically to proteins but, interacting with lipids, alter the bilayer structure. In this large class of compounds are included drugs such as anesthetics, narcotics, and antidepressants as well as cytotoxic molecules (13–16). Furthermore, -scission of alkoxyl radicals formed by autoxidation of polyunsaturated aliphatic chains may give rise to the formation of products of a lower molecular weight (17, 18) which, when inserted into the membrane, can damage its lipid organization. We investigated the partitioning between water and the lipid phase of linear monoamines in water dispersions of dimyristoylphosphatidylglycerol (DMPG) 2 and dimyristoylphosphatidylcholine (DMPC) multilamellar liposomes, and the results were compared with those obtained substituting amines with carboxylic acids to get information about the role of electrostatics and of hydrophobic interactions from the different charges of the molecules and lipids. The thermotropic behavior of the liposomes in the presence of amphipathic molecules was characterized by differential scanning calorimetry (DSC): depending on their length, these molecules affect the temperature and the cooperativity of the gel to fluid state transition of the membrane and the effects are much more pronounced in the case of monoamines. 2
1
To whom correspondence should be addressed. Fax: (41) 2578594. E-mail: [email protected]. 224
Abbreviations used: DMPG, dimyristoylphosphatidylglycerol; DMPC, dimyristoylphosphatidylcholine; DSC, differential scanning calorimetry; HRP, horseradish peroxidase. 0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
INTERACTION OF AMINES WITH MULTILAMELLAR LIPOSOMES
225
The study was extended to linear diamines and then to other polyamines, because it seemed natural to investigate the possibility that long diamines possess an amphipathic character, in contrast with the general behavior of polyamines, which are anchored at the surface and in no way penetrate the hydrophobic region of the membranes. DSC measurements confirmed this hypothesis for diamines longer than about nine carbon atoms and also showed the interesting dependence of the transition temperature of DMPG liposomes on the distance between amine groups of polyamines. MATERIALS AND METHODS Materials. Chemicals were analytical grade. All amines and carboxylic acids, 3,5-dichloro-2-hydroxybenzenesulfonic acid sodium salt, 4-aminoantipyrine, and the enzymes tyramine oxidase (tyramine:O 2 oxidoreductase, deaminating; EC 1.4.3.9), pig kidney amine oxidase (amine:O 2 oxidoreductase, deaminating; EC 1.4.3.6), and horseradish peroxidase (HRP) type II (donor: hydrogen peroxide oxidoreductase; EC 1.11.1.7) were purchased from Sigma Chemical Co. DMPG and DMPC, with certified purities of approximately 99% by TLC measurement, were also purchased from Sigma Chemical Co. Liposome preparation. Multilamellar vesicles were prepared following the method of Kusumi et al. (19). Phospholipids were dissolved in a 2:1 chloroform:methanol mixture, then dried with a stream of nitrogen gas, and kept under vacuum for at least 14 h. The dried lipids were suspended, when not otherwise specified, in a Hepes 0.1 M buffer, pH 7.2, solution containing the amines. The lipid dispersion, with a 101 mM final lipid concentration, was warmed at about 40°C, mixed repeatedly by a vortex for 30 s, and used as soon as obtained for DSC measurements. To determine the association constant, a 25.4 mM lipid dispersion was centrifuged at 12,000g for 15 min at 4°C and the supernatant carefully drained from the pellets. Amines and carboxylic acids were added to the buffer solution, in the phase of liposome formation, to avoid the evaporation of these volatile molecules under vacuum. Determination of the partition coefficients. The amount of free amines in the suspension of liposome-linked amines was determined by an amine oxidase enzymatic assay, according to an already applied procedure (20, 21). An aliquot of the supernatant was added to a buffered solution, pH 7.2, of 3 mM, 3,5-dichloro-2-hydroxybenzensulfonic acid sodium salt and 3 mM 4-aminoantipyrine containing 0.5 nmol of suitable amine oxidase and 1 nmol of HRP. The amine concentration was calculated from the absorbance value, recorded after 5 min of incubation at 37°C, and corrected by the blank. Measurements were carried out on a Beckman DU7 spectrophotometer equipped with a thermostatic quartz cell. Carboxylic acids were extracted from the supernatant, after acidification, with cyclohexane and the concentration was determined by GC–MS, using decanol as an internal standard. Mass spectra were obtained with a Shimadzu GC-17A gas chromatograph coupled with a Chrompack CP-Sil8C capillary column and interfaced with a Shimadzu QP-5000 mass spectrometer. DSC measurements. Differential scanning calorimetric measurements were performed on a Setaram DSC 92. About 50 mg of phospholipid dispersion was placed in an aluminum crucible. An identical crucible was filled with an equivalent weight of Hepes solution and placed in the reference cell. The temperature scanning rate was 0.5°C min ⫺1.
FIG. 1. (a) Fraction p of adsorbed molecules vs their acyl chain length n (in number of carbon atoms): (■) monoamines in DMPG liposomes; (F) monoamines in DMPC liposomes; (E) carboxylic acids in DMPC liposomes (pH 8). (b) Values of the corrected fraction p 0 ; symbols are the same as those in (a).
RESULTS AND DISCUSSION
Monoamines and carboxylic acids. The fraction of molecules, amines, or acids, indicated as A in the following, adsorbed to the membrane is given by p ⫽ [A] lip/[A] total and can be expressed in terms of measured quantities by p ⫽ ([A] total ⫺ [A])/[A] total, where the subscript “lip” is used to differentiate between a molecule that is bound to the membrane and a molecule free in solution. The values of p, plotted in Fig. 1a, show a strong dependence on phospholipid charge and amine length; they are also compared with the analogue values, when available, for fatty acids. In the case of fatty acids, measuring p presented some experimental difficulties that limited the number of useful experimental points; in particular, in GC–MS measurements, it was impossible to separate fatty acids from DMPG phospholipid residues satisfactorily. The difference between the values of p due to the different charges of DMPC and DMPG liposomes is in part explained on electrostatics grounds: on the basis of the Gouy–Chapman theory, free amines accumulate at the negatively charged lipid surface according to the Boltzmann expression [A] s ⫽ [A]exp(⫺ze(0)/kT), where [A] is the free amine concentration in the bulk of the suspension, [A] s its concentration at the membrane surface, z the amine charge, and (0) the membrane potential near the surface. As a consequence, the mea-
226
MOMO, FABRIS, AND STEVANATO
FIG. 2. Free energy (in kcal mol ⫺1) of the binding of phospholipids with (■) monoamines in DMPG liposomes, (F) monoamines in DMPC liposomes, and (E) carboxylic acids in DMPC liposomes (pH 8) as a function of their length n.
sured effective partition coefficient p is enhanced with respect to the intrinsic partition coefficient p 0 , which is given by p 0 ⫽ [A] lip/([A] lip ⫹ [A]exp(⫺ze (0)/kT)). The values of p 0 , with T ⫽ 291 K, z ⫽ 1, and (0) ⫽ ⫺60 mV in the case of DMPG and (0) ⫽ 0 for DMPC, are plotted in Fig. 1b. The value of (0) ⫽ ⫺60 mV was experimentally determined and agrees with those reported in the literature (22). Since p 0 should be independent of the surface charge of the membrane, we have to conclude that electrostatics does not completely account for the different partitioning of amines in DMPG and DMPC.
FIG. 3. DSC profiles of the gel to fluid state transition of DMPC multilamellar liposomes in the presence of linear monoamines of length n (An). y axis, dH/dT in arbitrary units; x axis, temperature T in °C.
FIG. 4. DSC profiles of the gel to fluid state transition of DMPG multilamellar liposomes in the presence of linear monoamines.
The residual difference, anticipating the results of DSC experiments, is most likely explained by the different anchoring sites of amines in DMPG and DMPC membranes: at a parity of length, amines bound to DMPG reside deeper in the bilayer than in the case of DMPC and participate with more methylene groups in the hydrophobic interactions between the alkyl chains that govern the partitioning. The change of free energy involved in the transfer of a species from water to the lipid phase, following Peitzsch and McLaughlin (8) is given by ⌬G ⫽ ⫺RT ln(55.6K p /[PL]), where K p ⫽ [A] lip/[A] s and [PL] is the lipid concentration. When plotted against their length, the ⌬G values of amines are well described by two straight lines of nearly equal slope ⬇ ⫺0.5 kcal mol ⫺1 per carbon (Fig. 2). This value is in accordance with the quoted one (⬃⫺0.8 kcal mol ⫺1 per carbon) in (8) and represents the contribution to the total ⌬G of the weak hydrophobic bond established by each methylene group of the amine with the neighboring phospholipids. DSC measurements. Some of the DSC profiles of the gel to liquid crystal phase transition of DMPC and DMPG multilamellar liposomes in the presence of monoamines and carboxylic acids are shown in Figs. 3– 6, it being understood that the data not reported simply confirm what is discussed in the following; the concentration of phospholipids in the liposomes dispersion was 101 mM, and the concentration of amphipathic molecules was 20 mM. The transition temperatures as a function of the length n of the amine and acid chains are plotted in
INTERACTION OF AMINES WITH MULTILAMELLAR LIPOSOMES
FIG. 5. DSC profiles of the gel to fluid state transition of DMPC multilamellar liposomes in the presence of carboxylic acids (pH 6) of length n (Cn).
Fig. 7. The peak transition temperatures of carboxylic acids, both in DMPC and in DMPG, show a minimum at n ⫽ 7 (pH 6) and n ⫽ 9 (pH 8); amines have their minimum at n ⫽ 7 (DMPG, pH 7.2) and n ⫽ 9 (DMPC, pH 7.2). The experiments with carboxylic acids were
FIG. 6. DSC profiles of the gel to fluid state transition of DMPC multilamellar liposomes in the presence of carboxylic acids (pH 8).
227
FIG. 7. Phase transition temperatures of (a) DMPC liposomes and (b) DMPG liposomes vs the chain length n of (■) monoamines, (E) carboxylic acids, pH 6.0, and (‚) carboxylic acids, pH 8.0; (- - -) transition temperature of pure DMPC and DMPG liposomes. Temperatures are taken at the peak of the DSC profile.
performed at two distinct pH values, which roughly means introducing into the membrane almost totally protonated (pH 6) and unprotonated (pH 8) carboxylic acids (21), and it is well known that the two forms have different anchoring points in the polar region of the bilayers (23). Thus the curves in Fig. 7 can be interpreted by simply assuming that the minima of the transition temperature are determined by the same effective penetration depth of the amphipathic chains in the membrane, which differs from the length n of amines and acids because of the position of their polar heads. The anchoring site of carboxylic acids in the COOH form lies approximately at the level of the glycerol backbone of phospholipids (9, 23), while for the COO ⫺ form, it is shifted toward the membrane surface by about two carbon atoms. Amines show a similar behavior, depending on the phospholipid charge rather than the pH; this is probably due to a competition between the interaction of the positive amine group with the phosphate and with water: in the case of DMPC, the electrostatic interaction with the negative charge of the phosphate is partially reduced by the presence of the positive charge of the choline ammonium group, and the interaction with water is favored, allowing for a shift of the molecule toward the membrane surface. The relationship between the lowering, and subsequent increase, of the transition temperatures and the
228
MOMO, FABRIS, AND STEVANATO
FIG. 8. Phase transition temperatures of DMPC liposomes vs the chain length n of (■) 20 mM monoamines, (F) 10 mM monoamines, and (Œ) 5 mM monoamines; (E) interpolated points corresponding to equal effective concentrations of monoamines with respect to 10 mM nonylamine.
length of the molecules intercalated into the membrane hydrophobic region is immediate. In fact, the rigid lipid organization of the gel state cannot be maintained by the phospholipid chains because they must accommodate the relatively large (about twice the methylene groups in volume) terminal methyl groups of the dopant and because they are also forced to occupy the potential voids created beneath them. It can be reasonably expected that the perturbation is stronger when the methyl group is placed in the middle region of the bilayer, and similar effects were reported in mixed-chain phosphatidylcholines (24) and when bulky groups were placed in the membrane (13). On the contrary, above this critical length, the transition temperature rises again because the methyl group is pushed toward an intrinsically less ordered region and because the hydrophobic interaction between the adjacent chains is partially restored. Indeed, the transition temperature is raised above the unperturbed value when the length of the host molecules is comparable with that of phospholipids, thus suggesting a preferential interaction of long molecules with the gel state. The data used in the discussion were obtained at a relatively high, 20 mM, concentration of dopants to enhance the effects on the transition temperature; the DSC experiments were repeated at 10 and 5 mM amine concentration and the results, plotted in Fig. 8, suggest that even lower amine concentration could be effective. It is also clear that the amine concentration does not modify the way the transition temperature depends on the amine length, as, in the various curves, only the depth of the minimum, and not its position, is changed. In any case, it must be noted that the curves just discussed were obtained with equimolar concentrations of the species and that, due to their different partitioning, this does not correspond to equal effective concentrations of molecules in the membrane. To check this point, we sketched a curve of the transition tem-
peratures in Fig. 8 corresponding to a 10 mM concentration of nonylamine and to an equal effective concentration of the other amines. Taking into account the partition coefficients of the amines, the curve was roughly estimated by interpolation of the experimental data at different amine concentrations: as can be seen, by comparison with the analogue 10 mM equimolar data, the correction leads to a smoothing of the curve, but it does not change the validity of the above considerations about the minimum. Another point deserves further discussion: first, amines produce a much more marked decrease of the transition temperature than carboxylic acids do, and this cannot be a simple consequence of the partitioning because the partition coefficients p of amines in DMPG and DMPC are different whereas they are nearly equal for amines and carboxylic acids in DMPC. Second, the broadened DSC profiles of amines show an abrupt loss of cooperativity in the transition that is not revealed in the presence of fatty acids. We are then led to conclude that there are mechanisms, involving the polar region of the membranes, that affect phospholipid organization very efficaciously, in some sort of synergy with perturbations of the acyl chain packaging. We make the hypothesis that this mechanism can be identified in the conformational change of the polar heads of the phospholipids described in (25–28), when positively charged molecules, such as charged anesthetics, amphiphiles, and metal ions bind to membranes: in phosphatidylcholine liposomes, the P ⫺–N ⫹ dipole in the head group is oriented parallel to the bilayer surface, and the presence of adsorbed positive molecules causes a rotation that brings it closer to the normal. The same authors noted that the incorporation of these charged species induces only a small disordering of the hydrocarbon chains, but the rotation of the polar heads, and the consequent possible change of the prevalent interactions between phospholipid heads, can reasonably be considered to be responsible for the cooperativity loss and for the more marked effects on the transition temperature. Di- and polyamines. Polyamines bind to the liposome surface, most likely forming complexes with the phosphate groups of phospholipids; introducing the association constant K f, the equilibrium A ⫹ PL lip ^ A 䡠 PL lip between phospholipids (PL) and amines (A) is given by {A 䡠 PL} lip ⫽ K f[A]{PL} lip, where the braces refer to the surface concentrations, the square brackets refer to the volume concentrations, and K f is expressed in M ⫺1. Here we used the more simple equation [A 䡠 PL] ⫽ K f[A][PL], representing the equilibrium between homogeneous liquid phases; the limits of validity for this assumption have already been discussed in (21). The K f values of butylamine, representative of short monoamines, and of the naturally occurring pu-
INTERACTION OF AMINES WITH MULTILAMELLAR LIPOSOMES
229
TABLE I
K f Values of the Complexes between Protonated Mono-, Di-, and Tetramines and Differently Charged Multilamellar Liposomes a
BUA (⫹1) PUT (⫹2) SPD (⫹3) SPM (⫹4)
DMPC (0)
DMPG (⫺1)
11 ⫾ 2 6 ⫾1 3 ⫾1 0.6 ⫾ 0.1
36 ⫾ 2 (1.9 ⫾ 0.4) ⫻ 10 2 (1.1 ⫾ 0.3) ⫻ 10 3 (1.1 ⫾ 0.2) ⫻ 10 4
a K f values are expressed in M ⫺1. The nominal charges at physiological pH for phospholipids and amines are reported in parentheses.
trescine, spermidine, and spermine are reported in Table I; the binding of amines to zwitterionic DMPC is characterized by relatively low K f values, of the order of unity; instead, in the case of DMPG, we observe a strong dependence on polyamine charge z, as more than two orders of magnitude separate the K f values of monoamines from spermine. When plotted on a logarithmic scale as a function of z, the values are fitted with good agreement by a straight line (Fig. 9). The behavior is explained by assuming an intrinsic association constant K 0 nearly independent of z and an effective constant K f enhanced by the electrostatic attraction of the free amines to the negatively charged lipid surface according to the Boltzmann expression K f ⫽ K 0 exp(⫺ze(0)/kT). Linear interpolation of the data points in Fig. 9 allows us to calculate the surface potential (0) ⫽ ⫺60 mV of DMPG bilayers and the value of K 0 ⫽ 4.7 M ⫺1, which is in the same range, 2–10 M ⫺1, reported for the K 0 of spermine with other negatively charged liposomes such as phosphatidylinositol (29), phosphatidate (30), and phosphatidylserine (22). The K 0 value of amines bound to negatively charged phospholipid is of the same order of the K f ⫽ K 0 found for zwitterionic DMPC liposomes; this means that the accumulation of cations in the aqueous diffuse double layer adjacent to the membrane surface is by far the main factor modu-
FIG. 9. Natural logarithm of the association constant vs the charge z of amines.
FIG. 10. Association constant of diamines vs their length n with (■) DMPG and (E) DMPC multilamellar liposomes.
lating the interaction of all mono- and polyamines with charged and zwitterionic liposomes. The K f values were also determined (Fig. 10) for diamines as a function of their length in number of carbon atoms. Like polyamines, diamines bind to DMPC with very low values of K f; in the case of DMPG, all the values of diamines, from three to seven carbon atoms long, are about the same and lie in the 100 –200 M ⫺1 interval; higher values, out of the range, are featured by 1,2-diaminoethane and amines with nine or more carbon atoms. In the first case, the large K f of diaminoethane is attributable to its high charge density, as expected on the basis of the Boltzmann theory, when corrected for the size of ions, which is usually in its simplest formulation where ions are considered as undifferentiated point charges. The surprisingly large association constant of long diamines suggests some different binding mechanisms, which will be discussed together with the results of DSC measurements. The transition temperatures of DMPC and DMPG liposomes in the presence of diamines are displayed in Fig. 11 as a function of their length in number of carbon atoms. It is straightforward to see that short diamines raise the transition temperature of DMPG in
FIG. 11. Phase transition temperatures in the presence of linear diamines vs their acyl chain length n of (E) DMPC liposomes [( 䡠 䡠 䡠 ) transition temperature of pure DMPC] and (■) DMPG liposomes [(- - -) transition temperature of pure DMPG].
230
MOMO, FABRIS, AND STEVANATO TABLE II
Transition Temperatures of DMPG Liposomes in the Presence of Various Polyamines a Spermine
26.6°C (24.3°C)
Triethylenetetramine
27.8°C
Spermidine
25.2°C
Diethylenetriamine
26.5°C
a The transition temperature of pure DMPG is given in parentheses. Hydrogen atoms are omitted in the skeleton structures of the polyamines.
a marked way. All polyamines produce a similar effect, which reflects the ability of these molecules to directly influence the structural organization of lipids in the bilayer, both reducing the electrostatic repulsive forces and creating bridging bonds between neighboring phosphate groups. A substantially different mechanism must be hypothesized for diamines nine or more methylene groups long, for which the transition temperature is lowered. It can be identified, by analogy with amphipathic molecules, by the possibility that long diamines will penetrate the inner hydrophobic region of the membrane with their folded carbon chains, being anchored to the polar region near the surface through their positively charged ends. In this way, it is also possible to explain the large values of the association constant because the hydrophobic interaction between the hydrocarbon chains brings a new contribution to the binding. In the case of DMPC, the transition temperature is very slightly modified by short amines. This is not surprising because, as is evident from their low association constants, all amines only weakly interact with zwitterionic liposomes; in any case, a lowering produced by long diamines is still observable and confirms the hypothesis of the amphipathic character of long diamines. A consideration of the interaction of short diamines with DMPG liposomes is naturally suggested by the fact that the DMPG transition temperature increases monotonically as amines get shorter, with the maximum in correspondence of 1,2-diaminoethane. The experimental data seem to assign an important role to the bridging bonds which diamines can establish between phosphate groups; in fact, the observed trend of the transition temperatures is in accordance with the reasonable hypothesis that short bridging bonds can modify the lipid organization more effectively than long ones, giving origin to more rigid structures, characterized by higher transition temperatures.
The distance between amine groups is also critical in polyamines; in Table II we report the transition temperatures of DMPG liposomes in the presence of spermine, spermidine, triethylenetetramine, and diethylenetriamine: it can be seen immediately that the transition temperature is significantly raised when the hydrocarbon segments of spermine and spermidine are substituted by the shorter ethylene groups. CONCLUSIONS
Positively charged linear amines, when inserted into phospholipid bilayers, roughly repeat the results of fatty acids for the part concerning the interactions between the hydrocarbon chains. The partition coefficients, systematically determined for amines and carboxylic acids as a function of their length, show that (i) the binding is governed by hydrophobic interactions between the methylene groups of the chains; the partition constant K p increases exponentially with the length of amines and fatty acids but under a minimum length (⬃3 carbon atoms) of the chain, the hydrophobic contribution to the binding is negligible; and the free energy of the binding is ⫺0.5 kcal/mol for each interacting methylene group and (ii) electrostatics plays an important role because it determines the accumulation at the surface of negatively charged membranes of positively charged amines. Moreover, experiments showed that liposomes display roughly the same profile of the transition temperature in the presence of amines and acids. In both cases, the transition temperature is lowest when the amphipathic molecules are about one half of phospholipids in length, so that the disordering, of steric origin, induced in the acyl chains must also be similar. Instead, amines differ from fatty acids because they not only determine a more marked decrease in the transition temperature but also produce a loss of lipid
INTERACTION OF AMINES WITH MULTILAMELLAR LIPOSOMES
cooperativity that is not recognizable in fatty acids– lipid bilayer interactions. Somehow amines are more effective as membrane-perturbing agents, and one possible explanation for this is the induced tilt of the phospholipid polar heads. Polyamines do not partition in the lipid region and anchor themselves at the membrane surface; anyway we demonstrated that long diamines are able to introduce part of their hydrocarbon chain into the hydrophobic region of the bilayer. Shorter diamines, which do not feature any amphiphilic behavior, also affect the bilayer structure of negatively charged liposomes, raising the transition temperature. The strong dependence of the temperature on the length of the diamines makes it reasonable to assume that one important mechanism of interaction is the bridging bonds between phosphate groups of neighboring phospholipids. We extended this observation to polyamines, which are known to be important modulators of many processes in biomembranes also. REFERENCES 1. Jain, M. K., and Min Wu, N. (1977) Effect of small molecules on the dipalmitoyl lecithin liposomal bilayer. III: Phase transition in lipid bilayer. J. Membr. Biol. 34, 157–201. 2. Rooney, E. K., East, J. M., Jones, O. T., McWhirter, J., Simmonds, A. C., and Lee, A. G. (1983) Interaction of fatty acids with lipid bilayers. Biochim. Biophys. Acta 728, 159 –170. 3. Ellena, J. F., Archer, S. J., Dominey, R. N., Hill, B. D., and Cafiso, D. S. (1988) Localizing the nitroxide group of fatty acid and voltage-sensitive spin-labels in phospholipid bilayers. Biochim. Biophys. Acta 940, 63–70. 4. Fumero, J., Bammel, B. P., Hopkins, H. P., and Smith, J. C. (1988) The effect of potential sensitive molecular probes on the thermal phase transition in the dimyristoylphospatidylcholine preparations. Biochim. Biophys. Acta 944, 164 –176. 5. Hamilton, J. A., Fujito, D. T., and Hammer, C. F. (1991) Solubilization and localization of weakly polar lipids in unsonicated egg phosphatidylcholine: A 13C MAS NMR study. Biochemistry 30, 2894 –2902. 6. Seelig, J., and Ganz, P. (1991) Nonclassical hydrophobic effect in membrane binding equilibria. Biochemistry 30, 9354 –9359. 7. Abrams, F. S., Chattopadhyay, A., and London, E. (1992) Determination of the location of fluorescent probes attached to fatty acids using parallax analysis of fluorescence quenching: effect of carboxyl ionization state environment on depth. Biochemistry 31, 5322–5327. 8. Peitzsch, R. M., and McLaughlin, S. (1993) Binding of acylated peptides and fatty acids to phospholipid vesicles: Pertinence to myristoylated proteins. Biochemistry 32, 10436 –10443. 9. Richieri, G. V., Ogata, R. T., and Kleinfeld, A. M. (1993) Thermodynamics of fatty acids binding to fatty acids-binding proteins and fatty acids partition between water and membranes measured using the fluorescent probe ADIFAB. J. Biol. Chem. 270, 15076 –15084. 10. Wisniewka, A., Nishimoto, Y., Hyde, J. S., Kusumi, A., and Subczynski, W. K. (1996) Depth dependence of the perturbing effect of placing a bulky group (oxazolidine ring spin labels) in the membrane on the membrane phase transition. Biochim. Biophys. Acta 1278, 68 –72.
231
11. Simon, S. A., and McIntosh, T. J. (1984) Interdigitated hydrocarbon chain packing causes the biphasic transition behaviour in lipid/alcohol suspension. Biochim. Biophys. Acta 773, 169 – 172. 12. Rowe, E. S. (1987) Induction of lateral phase separation in binary lipid mixtures by alcohol. Biochemistry 26, 46 –51. 13. Baber, J., Ellena, J. F., and Cafiso, D. S. (1995) Distribution of general anaesthetics in phospholipid bilayers determined using 2 H NMR and 1H– 1H NOE spectroscopy. Biochemistry 34, 6533– 6539. 14. Miyazaky, J., Hideg, K., and Marsh, D. (1992) Interfacial ionization and partitioning of membrane-bound local anaesthetics. Biochim. Biophys. Acta 1103, 62– 68. 15. Anel, A., Richieri, G. V., and Kleinfeld, A. M. (1993) Membrane partition of fatty acids and inhibition of T cell function. Biochemistry 32, 530 –536. 16. Aranda, F. J., and Villalain, J. (1997) The interaction of abietic acid with phospholipid membranes. Biochim. Biophys. Acta 1327, 171–180. 17. Frankel, E. N. (1983) Volatile lipid oxidation products. Prog. Lipid Res. 22, 1–33. 18. Gardner, H. W. (1989) Oxygen radical chemistry of polyunsaturated fatty acids. Free Radic. Biol. Med. 7, 65– 86. 19. Kusumi, A., Subczynski, W. K., and Hyde, J. S. (1982) Oxygen transport parameter in membranes as deduced by saturation recovery measurements of spin–lattice relaxation times of spin labels. Proc. Natl. Acad. Sci. USA 79, 1854 –1858. 20. Momo, F., Wisniewska, A., and Stevanato, R. (1995) EPR study of spermine interaction with multilamellar phosphatidylcholine liposomes. Biochim. Biophys. Acta 1240, 89 –94. 21. Stevanato, R., Wisniewska, A., and Momo, F. (1997) Interaction of spermine with dimyristoylphosphatidylglycerol multilamellar liposomes. Arch. Biochem. Biophys. 346, 203–207. 22. Chung, L., Kaloyanides, G., McDaniel, R., McLaughlin, A., and Mclaughlin, S. (1985) Interaction of gentamicin and spermine with bilayer membranes containing negatively charged phospholipids. Biochemistry 24, 442– 452. 23. Sanson, A., Ptak, M., Rigaud, J. L., and Gary-Bobo, C. M. (1976) An ESR study of the anchoring of spin-labeled stearic acid in lecithin multilayers. Chem. Phys. Lipids 17, 435– 444. 24. Bultman, T., Lin, H., Wang, Z., and Huang, C. (1991) Thermotropic and mixing behaviour of mixed-chain phosphatidylcholines with molecular weights identical with that of L-␣-dipalmitoylphosphatidylcholine. Biochemistry 30, 7194 –7202. 25. Akutsu, H., and Seelig, J. (1981) Interaction of metal ions with phosphatidylcholine bilayer membranes. Biochemistry 20, 7366 –7373. 26. Scherer, P., and Seelig, J. (1989) Electric charge efects on phospholipid headgroups. Phosphatidylcholine in mixtures with cationic and anionic amphiphiles. Biochemistry 28, 7720 –7728. 27. Beschiaschvili, G., and Seelig, J. (1990) Melittin binding to mixed phosphatidylglycerol/phosphatidylcholine membranes. Biochemistry 29, 52–58. 28. Bechinger, B., and Seelig, J. (1991) Interaction of electric dipoles with phospholipid headgroups. A 2H and 31P NMR study of phloretin and phloretin analogues in phopshatidylcholine membranes. Biochemistry 30, 3923–3929. 29. Toner, M., Vaio, G., McLaughlin, A., and McLaughlin, S. (1988) Biochemistry 27, 7435–7443. 30. Meers, P., Hong, K., Bentz, J., and Papahadjopoulos, D. (1986) Biochemistry 25, 3109 –3118.
Interaction of Linear Mono- and Diamines with Dimyristoylphosphatidylcholine and Dimyristoylphosphatidylglycerol Multilamellar Liposomes Federico Momo,* ,† ,1 Sabrina Fabris,* and Roberto Stevanato* *Department of Physical Chemistry and †Istituto Nazionale Fisica della Materia, University of Venice, Dorsoduro 2137, 30123 Venice, Italy
Received March 28, 2000, and in revised form July 12, 2000
The effect of linear monoamines on dimyristoylphosphatidylglycerol and dimyristoylphosphatidylcholine multilamellar liposomes was studied as a function of their length and compared with the behavior of linear carboxylic acids. The role of the hydrophobic interactions was demonstrated and the free energy of the binding for each interacting carbon atom was determined. The thermotropic behavior of the liposomes was characterized by differential scanning calorimetry and it was shown that these molecules affect the temperature and the cooperativity of the gel to fluid state transition of the membrane differently. In particular, it appeared that membrane perturbation was maximum when the chain length of the amphipathic molecules ranged between 7 and 9 carbon atoms, with more pronounced effects in the case of monoamines. Molecules shorter than 3– 4 carbon atoms did not produce any observable change in the transition temperature. The study was extended to linear ␣,-diamines to investigate the amphipathic character of long diamines and to investigate the role of bridging bonds established with neighboring phospholipids. © 2000 Academic Press
Key Words: amines; phospholipids; liposomes.
The interaction of small amphipathic molecules with biomembranes is of wide interest (1–10) because these molecules may influence the cellular processes in a number of ways, all correlated to their lipid solubility and to the modifications they are able to produce in the lipid organization of the membrane bilayer such as the
induction of interdigitated phases or later phase separation phenomena (11, 12). It is well known, for example, that, in biomembranes, a largely cooperative behavior between proteins and their surroundings exists and that the protein-specific membrane functions can be affected by agents that do not bind specifically to proteins but, interacting with lipids, alter the bilayer structure. In this large class of compounds are included drugs such as anesthetics, narcotics, and antidepressants as well as cytotoxic molecules (13–16). Furthermore, -scission of alkoxyl radicals formed by autoxidation of polyunsaturated aliphatic chains may give rise to the formation of products of a lower molecular weight (17, 18) which, when inserted into the membrane, can damage its lipid organization. We investigated the partitioning between water and the lipid phase of linear monoamines in water dispersions of dimyristoylphosphatidylglycerol (DMPG) 2 and dimyristoylphosphatidylcholine (DMPC) multilamellar liposomes, and the results were compared with those obtained substituting amines with carboxylic acids to get information about the role of electrostatics and of hydrophobic interactions from the different charges of the molecules and lipids. The thermotropic behavior of the liposomes in the presence of amphipathic molecules was characterized by differential scanning calorimetry (DSC): depending on their length, these molecules affect the temperature and the cooperativity of the gel to fluid state transition of the membrane and the effects are much more pronounced in the case of monoamines. 2
1
To whom correspondence should be addressed. Fax: (41) 2578594. E-mail: [email protected]. 224
Abbreviations used: DMPG, dimyristoylphosphatidylglycerol; DMPC, dimyristoylphosphatidylcholine; DSC, differential scanning calorimetry; HRP, horseradish peroxidase. 0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
INTERACTION OF AMINES WITH MULTILAMELLAR LIPOSOMES
225
The study was extended to linear diamines and then to other polyamines, because it seemed natural to investigate the possibility that long diamines possess an amphipathic character, in contrast with the general behavior of polyamines, which are anchored at the surface and in no way penetrate the hydrophobic region of the membranes. DSC measurements confirmed this hypothesis for diamines longer than about nine carbon atoms and also showed the interesting dependence of the transition temperature of DMPG liposomes on the distance between amine groups of polyamines. MATERIALS AND METHODS Materials. Chemicals were analytical grade. All amines and carboxylic acids, 3,5-dichloro-2-hydroxybenzenesulfonic acid sodium salt, 4-aminoantipyrine, and the enzymes tyramine oxidase (tyramine:O 2 oxidoreductase, deaminating; EC 1.4.3.9), pig kidney amine oxidase (amine:O 2 oxidoreductase, deaminating; EC 1.4.3.6), and horseradish peroxidase (HRP) type II (donor: hydrogen peroxide oxidoreductase; EC 1.11.1.7) were purchased from Sigma Chemical Co. DMPG and DMPC, with certified purities of approximately 99% by TLC measurement, were also purchased from Sigma Chemical Co. Liposome preparation. Multilamellar vesicles were prepared following the method of Kusumi et al. (19). Phospholipids were dissolved in a 2:1 chloroform:methanol mixture, then dried with a stream of nitrogen gas, and kept under vacuum for at least 14 h. The dried lipids were suspended, when not otherwise specified, in a Hepes 0.1 M buffer, pH 7.2, solution containing the amines. The lipid dispersion, with a 101 mM final lipid concentration, was warmed at about 40°C, mixed repeatedly by a vortex for 30 s, and used as soon as obtained for DSC measurements. To determine the association constant, a 25.4 mM lipid dispersion was centrifuged at 12,000g for 15 min at 4°C and the supernatant carefully drained from the pellets. Amines and carboxylic acids were added to the buffer solution, in the phase of liposome formation, to avoid the evaporation of these volatile molecules under vacuum. Determination of the partition coefficients. The amount of free amines in the suspension of liposome-linked amines was determined by an amine oxidase enzymatic assay, according to an already applied procedure (20, 21). An aliquot of the supernatant was added to a buffered solution, pH 7.2, of 3 mM, 3,5-dichloro-2-hydroxybenzensulfonic acid sodium salt and 3 mM 4-aminoantipyrine containing 0.5 nmol of suitable amine oxidase and 1 nmol of HRP. The amine concentration was calculated from the absorbance value, recorded after 5 min of incubation at 37°C, and corrected by the blank. Measurements were carried out on a Beckman DU7 spectrophotometer equipped with a thermostatic quartz cell. Carboxylic acids were extracted from the supernatant, after acidification, with cyclohexane and the concentration was determined by GC–MS, using decanol as an internal standard. Mass spectra were obtained with a Shimadzu GC-17A gas chromatograph coupled with a Chrompack CP-Sil8C capillary column and interfaced with a Shimadzu QP-5000 mass spectrometer. DSC measurements. Differential scanning calorimetric measurements were performed on a Setaram DSC 92. About 50 mg of phospholipid dispersion was placed in an aluminum crucible. An identical crucible was filled with an equivalent weight of Hepes solution and placed in the reference cell. The temperature scanning rate was 0.5°C min ⫺1.
FIG. 1. (a) Fraction p of adsorbed molecules vs their acyl chain length n (in number of carbon atoms): (■) monoamines in DMPG liposomes; (F) monoamines in DMPC liposomes; (E) carboxylic acids in DMPC liposomes (pH 8). (b) Values of the corrected fraction p 0 ; symbols are the same as those in (a).
RESULTS AND DISCUSSION
Monoamines and carboxylic acids. The fraction of molecules, amines, or acids, indicated as A in the following, adsorbed to the membrane is given by p ⫽ [A] lip/[A] total and can be expressed in terms of measured quantities by p ⫽ ([A] total ⫺ [A])/[A] total, where the subscript “lip” is used to differentiate between a molecule that is bound to the membrane and a molecule free in solution. The values of p, plotted in Fig. 1a, show a strong dependence on phospholipid charge and amine length; they are also compared with the analogue values, when available, for fatty acids. In the case of fatty acids, measuring p presented some experimental difficulties that limited the number of useful experimental points; in particular, in GC–MS measurements, it was impossible to separate fatty acids from DMPG phospholipid residues satisfactorily. The difference between the values of p due to the different charges of DMPC and DMPG liposomes is in part explained on electrostatics grounds: on the basis of the Gouy–Chapman theory, free amines accumulate at the negatively charged lipid surface according to the Boltzmann expression [A] s ⫽ [A]exp(⫺ze(0)/kT), where [A] is the free amine concentration in the bulk of the suspension, [A] s its concentration at the membrane surface, z the amine charge, and (0) the membrane potential near the surface. As a consequence, the mea-
226
MOMO, FABRIS, AND STEVANATO
FIG. 2. Free energy (in kcal mol ⫺1) of the binding of phospholipids with (■) monoamines in DMPG liposomes, (F) monoamines in DMPC liposomes, and (E) carboxylic acids in DMPC liposomes (pH 8) as a function of their length n.
sured effective partition coefficient p is enhanced with respect to the intrinsic partition coefficient p 0 , which is given by p 0 ⫽ [A] lip/([A] lip ⫹ [A]exp(⫺ze (0)/kT)). The values of p 0 , with T ⫽ 291 K, z ⫽ 1, and (0) ⫽ ⫺60 mV in the case of DMPG and (0) ⫽ 0 for DMPC, are plotted in Fig. 1b. The value of (0) ⫽ ⫺60 mV was experimentally determined and agrees with those reported in the literature (22). Since p 0 should be independent of the surface charge of the membrane, we have to conclude that electrostatics does not completely account for the different partitioning of amines in DMPG and DMPC.
FIG. 3. DSC profiles of the gel to fluid state transition of DMPC multilamellar liposomes in the presence of linear monoamines of length n (An). y axis, dH/dT in arbitrary units; x axis, temperature T in °C.
FIG. 4. DSC profiles of the gel to fluid state transition of DMPG multilamellar liposomes in the presence of linear monoamines.
The residual difference, anticipating the results of DSC experiments, is most likely explained by the different anchoring sites of amines in DMPG and DMPC membranes: at a parity of length, amines bound to DMPG reside deeper in the bilayer than in the case of DMPC and participate with more methylene groups in the hydrophobic interactions between the alkyl chains that govern the partitioning. The change of free energy involved in the transfer of a species from water to the lipid phase, following Peitzsch and McLaughlin (8) is given by ⌬G ⫽ ⫺RT ln(55.6K p /[PL]), where K p ⫽ [A] lip/[A] s and [PL] is the lipid concentration. When plotted against their length, the ⌬G values of amines are well described by two straight lines of nearly equal slope ⬇ ⫺0.5 kcal mol ⫺1 per carbon (Fig. 2). This value is in accordance with the quoted one (⬃⫺0.8 kcal mol ⫺1 per carbon) in (8) and represents the contribution to the total ⌬G of the weak hydrophobic bond established by each methylene group of the amine with the neighboring phospholipids. DSC measurements. Some of the DSC profiles of the gel to liquid crystal phase transition of DMPC and DMPG multilamellar liposomes in the presence of monoamines and carboxylic acids are shown in Figs. 3– 6, it being understood that the data not reported simply confirm what is discussed in the following; the concentration of phospholipids in the liposomes dispersion was 101 mM, and the concentration of amphipathic molecules was 20 mM. The transition temperatures as a function of the length n of the amine and acid chains are plotted in
INTERACTION OF AMINES WITH MULTILAMELLAR LIPOSOMES
FIG. 5. DSC profiles of the gel to fluid state transition of DMPC multilamellar liposomes in the presence of carboxylic acids (pH 6) of length n (Cn).
Fig. 7. The peak transition temperatures of carboxylic acids, both in DMPC and in DMPG, show a minimum at n ⫽ 7 (pH 6) and n ⫽ 9 (pH 8); amines have their minimum at n ⫽ 7 (DMPG, pH 7.2) and n ⫽ 9 (DMPC, pH 7.2). The experiments with carboxylic acids were
FIG. 6. DSC profiles of the gel to fluid state transition of DMPC multilamellar liposomes in the presence of carboxylic acids (pH 8).
227
FIG. 7. Phase transition temperatures of (a) DMPC liposomes and (b) DMPG liposomes vs the chain length n of (■) monoamines, (E) carboxylic acids, pH 6.0, and (‚) carboxylic acids, pH 8.0; (- - -) transition temperature of pure DMPC and DMPG liposomes. Temperatures are taken at the peak of the DSC profile.
performed at two distinct pH values, which roughly means introducing into the membrane almost totally protonated (pH 6) and unprotonated (pH 8) carboxylic acids (21), and it is well known that the two forms have different anchoring points in the polar region of the bilayers (23). Thus the curves in Fig. 7 can be interpreted by simply assuming that the minima of the transition temperature are determined by the same effective penetration depth of the amphipathic chains in the membrane, which differs from the length n of amines and acids because of the position of their polar heads. The anchoring site of carboxylic acids in the COOH form lies approximately at the level of the glycerol backbone of phospholipids (9, 23), while for the COO ⫺ form, it is shifted toward the membrane surface by about two carbon atoms. Amines show a similar behavior, depending on the phospholipid charge rather than the pH; this is probably due to a competition between the interaction of the positive amine group with the phosphate and with water: in the case of DMPC, the electrostatic interaction with the negative charge of the phosphate is partially reduced by the presence of the positive charge of the choline ammonium group, and the interaction with water is favored, allowing for a shift of the molecule toward the membrane surface. The relationship between the lowering, and subsequent increase, of the transition temperatures and the
228
MOMO, FABRIS, AND STEVANATO
FIG. 8. Phase transition temperatures of DMPC liposomes vs the chain length n of (■) 20 mM monoamines, (F) 10 mM monoamines, and (Œ) 5 mM monoamines; (E) interpolated points corresponding to equal effective concentrations of monoamines with respect to 10 mM nonylamine.
length of the molecules intercalated into the membrane hydrophobic region is immediate. In fact, the rigid lipid organization of the gel state cannot be maintained by the phospholipid chains because they must accommodate the relatively large (about twice the methylene groups in volume) terminal methyl groups of the dopant and because they are also forced to occupy the potential voids created beneath them. It can be reasonably expected that the perturbation is stronger when the methyl group is placed in the middle region of the bilayer, and similar effects were reported in mixed-chain phosphatidylcholines (24) and when bulky groups were placed in the membrane (13). On the contrary, above this critical length, the transition temperature rises again because the methyl group is pushed toward an intrinsically less ordered region and because the hydrophobic interaction between the adjacent chains is partially restored. Indeed, the transition temperature is raised above the unperturbed value when the length of the host molecules is comparable with that of phospholipids, thus suggesting a preferential interaction of long molecules with the gel state. The data used in the discussion were obtained at a relatively high, 20 mM, concentration of dopants to enhance the effects on the transition temperature; the DSC experiments were repeated at 10 and 5 mM amine concentration and the results, plotted in Fig. 8, suggest that even lower amine concentration could be effective. It is also clear that the amine concentration does not modify the way the transition temperature depends on the amine length, as, in the various curves, only the depth of the minimum, and not its position, is changed. In any case, it must be noted that the curves just discussed were obtained with equimolar concentrations of the species and that, due to their different partitioning, this does not correspond to equal effective concentrations of molecules in the membrane. To check this point, we sketched a curve of the transition tem-
peratures in Fig. 8 corresponding to a 10 mM concentration of nonylamine and to an equal effective concentration of the other amines. Taking into account the partition coefficients of the amines, the curve was roughly estimated by interpolation of the experimental data at different amine concentrations: as can be seen, by comparison with the analogue 10 mM equimolar data, the correction leads to a smoothing of the curve, but it does not change the validity of the above considerations about the minimum. Another point deserves further discussion: first, amines produce a much more marked decrease of the transition temperature than carboxylic acids do, and this cannot be a simple consequence of the partitioning because the partition coefficients p of amines in DMPG and DMPC are different whereas they are nearly equal for amines and carboxylic acids in DMPC. Second, the broadened DSC profiles of amines show an abrupt loss of cooperativity in the transition that is not revealed in the presence of fatty acids. We are then led to conclude that there are mechanisms, involving the polar region of the membranes, that affect phospholipid organization very efficaciously, in some sort of synergy with perturbations of the acyl chain packaging. We make the hypothesis that this mechanism can be identified in the conformational change of the polar heads of the phospholipids described in (25–28), when positively charged molecules, such as charged anesthetics, amphiphiles, and metal ions bind to membranes: in phosphatidylcholine liposomes, the P ⫺–N ⫹ dipole in the head group is oriented parallel to the bilayer surface, and the presence of adsorbed positive molecules causes a rotation that brings it closer to the normal. The same authors noted that the incorporation of these charged species induces only a small disordering of the hydrocarbon chains, but the rotation of the polar heads, and the consequent possible change of the prevalent interactions between phospholipid heads, can reasonably be considered to be responsible for the cooperativity loss and for the more marked effects on the transition temperature. Di- and polyamines. Polyamines bind to the liposome surface, most likely forming complexes with the phosphate groups of phospholipids; introducing the association constant K f, the equilibrium A ⫹ PL lip ^ A 䡠 PL lip between phospholipids (PL) and amines (A) is given by {A 䡠 PL} lip ⫽ K f[A]{PL} lip, where the braces refer to the surface concentrations, the square brackets refer to the volume concentrations, and K f is expressed in M ⫺1. Here we used the more simple equation [A 䡠 PL] ⫽ K f[A][PL], representing the equilibrium between homogeneous liquid phases; the limits of validity for this assumption have already been discussed in (21). The K f values of butylamine, representative of short monoamines, and of the naturally occurring pu-
INTERACTION OF AMINES WITH MULTILAMELLAR LIPOSOMES
229
TABLE I
K f Values of the Complexes between Protonated Mono-, Di-, and Tetramines and Differently Charged Multilamellar Liposomes a
BUA (⫹1) PUT (⫹2) SPD (⫹3) SPM (⫹4)
DMPC (0)
DMPG (⫺1)
11 ⫾ 2 6 ⫾1 3 ⫾1 0.6 ⫾ 0.1
36 ⫾ 2 (1.9 ⫾ 0.4) ⫻ 10 2 (1.1 ⫾ 0.3) ⫻ 10 3 (1.1 ⫾ 0.2) ⫻ 10 4
a K f values are expressed in M ⫺1. The nominal charges at physiological pH for phospholipids and amines are reported in parentheses.
trescine, spermidine, and spermine are reported in Table I; the binding of amines to zwitterionic DMPC is characterized by relatively low K f values, of the order of unity; instead, in the case of DMPG, we observe a strong dependence on polyamine charge z, as more than two orders of magnitude separate the K f values of monoamines from spermine. When plotted on a logarithmic scale as a function of z, the values are fitted with good agreement by a straight line (Fig. 9). The behavior is explained by assuming an intrinsic association constant K 0 nearly independent of z and an effective constant K f enhanced by the electrostatic attraction of the free amines to the negatively charged lipid surface according to the Boltzmann expression K f ⫽ K 0 exp(⫺ze(0)/kT). Linear interpolation of the data points in Fig. 9 allows us to calculate the surface potential (0) ⫽ ⫺60 mV of DMPG bilayers and the value of K 0 ⫽ 4.7 M ⫺1, which is in the same range, 2–10 M ⫺1, reported for the K 0 of spermine with other negatively charged liposomes such as phosphatidylinositol (29), phosphatidate (30), and phosphatidylserine (22). The K 0 value of amines bound to negatively charged phospholipid is of the same order of the K f ⫽ K 0 found for zwitterionic DMPC liposomes; this means that the accumulation of cations in the aqueous diffuse double layer adjacent to the membrane surface is by far the main factor modu-
FIG. 9. Natural logarithm of the association constant vs the charge z of amines.
FIG. 10. Association constant of diamines vs their length n with (■) DMPG and (E) DMPC multilamellar liposomes.
lating the interaction of all mono- and polyamines with charged and zwitterionic liposomes. The K f values were also determined (Fig. 10) for diamines as a function of their length in number of carbon atoms. Like polyamines, diamines bind to DMPC with very low values of K f; in the case of DMPG, all the values of diamines, from three to seven carbon atoms long, are about the same and lie in the 100 –200 M ⫺1 interval; higher values, out of the range, are featured by 1,2-diaminoethane and amines with nine or more carbon atoms. In the first case, the large K f of diaminoethane is attributable to its high charge density, as expected on the basis of the Boltzmann theory, when corrected for the size of ions, which is usually in its simplest formulation where ions are considered as undifferentiated point charges. The surprisingly large association constant of long diamines suggests some different binding mechanisms, which will be discussed together with the results of DSC measurements. The transition temperatures of DMPC and DMPG liposomes in the presence of diamines are displayed in Fig. 11 as a function of their length in number of carbon atoms. It is straightforward to see that short diamines raise the transition temperature of DMPG in
FIG. 11. Phase transition temperatures in the presence of linear diamines vs their acyl chain length n of (E) DMPC liposomes [( 䡠 䡠 䡠 ) transition temperature of pure DMPC] and (■) DMPG liposomes [(- - -) transition temperature of pure DMPG].
230
MOMO, FABRIS, AND STEVANATO TABLE II
Transition Temperatures of DMPG Liposomes in the Presence of Various Polyamines a Spermine
26.6°C (24.3°C)
Triethylenetetramine
27.8°C
Spermidine
25.2°C
Diethylenetriamine
26.5°C
a The transition temperature of pure DMPG is given in parentheses. Hydrogen atoms are omitted in the skeleton structures of the polyamines.
a marked way. All polyamines produce a similar effect, which reflects the ability of these molecules to directly influence the structural organization of lipids in the bilayer, both reducing the electrostatic repulsive forces and creating bridging bonds between neighboring phosphate groups. A substantially different mechanism must be hypothesized for diamines nine or more methylene groups long, for which the transition temperature is lowered. It can be identified, by analogy with amphipathic molecules, by the possibility that long diamines will penetrate the inner hydrophobic region of the membrane with their folded carbon chains, being anchored to the polar region near the surface through their positively charged ends. In this way, it is also possible to explain the large values of the association constant because the hydrophobic interaction between the hydrocarbon chains brings a new contribution to the binding. In the case of DMPC, the transition temperature is very slightly modified by short amines. This is not surprising because, as is evident from their low association constants, all amines only weakly interact with zwitterionic liposomes; in any case, a lowering produced by long diamines is still observable and confirms the hypothesis of the amphipathic character of long diamines. A consideration of the interaction of short diamines with DMPG liposomes is naturally suggested by the fact that the DMPG transition temperature increases monotonically as amines get shorter, with the maximum in correspondence of 1,2-diaminoethane. The experimental data seem to assign an important role to the bridging bonds which diamines can establish between phosphate groups; in fact, the observed trend of the transition temperatures is in accordance with the reasonable hypothesis that short bridging bonds can modify the lipid organization more effectively than long ones, giving origin to more rigid structures, characterized by higher transition temperatures.
The distance between amine groups is also critical in polyamines; in Table II we report the transition temperatures of DMPG liposomes in the presence of spermine, spermidine, triethylenetetramine, and diethylenetriamine: it can be seen immediately that the transition temperature is significantly raised when the hydrocarbon segments of spermine and spermidine are substituted by the shorter ethylene groups. CONCLUSIONS
Positively charged linear amines, when inserted into phospholipid bilayers, roughly repeat the results of fatty acids for the part concerning the interactions between the hydrocarbon chains. The partition coefficients, systematically determined for amines and carboxylic acids as a function of their length, show that (i) the binding is governed by hydrophobic interactions between the methylene groups of the chains; the partition constant K p increases exponentially with the length of amines and fatty acids but under a minimum length (⬃3 carbon atoms) of the chain, the hydrophobic contribution to the binding is negligible; and the free energy of the binding is ⫺0.5 kcal/mol for each interacting methylene group and (ii) electrostatics plays an important role because it determines the accumulation at the surface of negatively charged membranes of positively charged amines. Moreover, experiments showed that liposomes display roughly the same profile of the transition temperature in the presence of amines and acids. In both cases, the transition temperature is lowest when the amphipathic molecules are about one half of phospholipids in length, so that the disordering, of steric origin, induced in the acyl chains must also be similar. Instead, amines differ from fatty acids because they not only determine a more marked decrease in the transition temperature but also produce a loss of lipid
INTERACTION OF AMINES WITH MULTILAMELLAR LIPOSOMES
cooperativity that is not recognizable in fatty acids– lipid bilayer interactions. Somehow amines are more effective as membrane-perturbing agents, and one possible explanation for this is the induced tilt of the phospholipid polar heads. Polyamines do not partition in the lipid region and anchor themselves at the membrane surface; anyway we demonstrated that long diamines are able to introduce part of their hydrocarbon chain into the hydrophobic region of the bilayer. Shorter diamines, which do not feature any amphiphilic behavior, also affect the bilayer structure of negatively charged liposomes, raising the transition temperature. The strong dependence of the temperature on the length of the diamines makes it reasonable to assume that one important mechanism of interaction is the bridging bonds between phosphate groups of neighboring phospholipids. We extended this observation to polyamines, which are known to be important modulators of many processes in biomembranes also. REFERENCES 1. Jain, M. K., and Min Wu, N. (1977) Effect of small molecules on the dipalmitoyl lecithin liposomal bilayer. III: Phase transition in lipid bilayer. J. Membr. Biol. 34, 157–201. 2. Rooney, E. K., East, J. M., Jones, O. T., McWhirter, J., Simmonds, A. C., and Lee, A. G. (1983) Interaction of fatty acids with lipid bilayers. Biochim. Biophys. Acta 728, 159 –170. 3. Ellena, J. F., Archer, S. J., Dominey, R. N., Hill, B. D., and Cafiso, D. S. (1988) Localizing the nitroxide group of fatty acid and voltage-sensitive spin-labels in phospholipid bilayers. Biochim. Biophys. Acta 940, 63–70. 4. Fumero, J., Bammel, B. P., Hopkins, H. P., and Smith, J. C. (1988) The effect of potential sensitive molecular probes on the thermal phase transition in the dimyristoylphospatidylcholine preparations. Biochim. Biophys. Acta 944, 164 –176. 5. Hamilton, J. A., Fujito, D. T., and Hammer, C. F. (1991) Solubilization and localization of weakly polar lipids in unsonicated egg phosphatidylcholine: A 13C MAS NMR study. Biochemistry 30, 2894 –2902. 6. Seelig, J., and Ganz, P. (1991) Nonclassical hydrophobic effect in membrane binding equilibria. Biochemistry 30, 9354 –9359. 7. Abrams, F. S., Chattopadhyay, A., and London, E. (1992) Determination of the location of fluorescent probes attached to fatty acids using parallax analysis of fluorescence quenching: effect of carboxyl ionization state environment on depth. Biochemistry 31, 5322–5327. 8. Peitzsch, R. M., and McLaughlin, S. (1993) Binding of acylated peptides and fatty acids to phospholipid vesicles: Pertinence to myristoylated proteins. Biochemistry 32, 10436 –10443. 9. Richieri, G. V., Ogata, R. T., and Kleinfeld, A. M. (1993) Thermodynamics of fatty acids binding to fatty acids-binding proteins and fatty acids partition between water and membranes measured using the fluorescent probe ADIFAB. J. Biol. Chem. 270, 15076 –15084. 10. Wisniewka, A., Nishimoto, Y., Hyde, J. S., Kusumi, A., and Subczynski, W. K. (1996) Depth dependence of the perturbing effect of placing a bulky group (oxazolidine ring spin labels) in the membrane on the membrane phase transition. Biochim. Biophys. Acta 1278, 68 –72.
231
11. Simon, S. A., and McIntosh, T. J. (1984) Interdigitated hydrocarbon chain packing causes the biphasic transition behaviour in lipid/alcohol suspension. Biochim. Biophys. Acta 773, 169 – 172. 12. Rowe, E. S. (1987) Induction of lateral phase separation in binary lipid mixtures by alcohol. Biochemistry 26, 46 –51. 13. Baber, J., Ellena, J. F., and Cafiso, D. S. (1995) Distribution of general anaesthetics in phospholipid bilayers determined using 2 H NMR and 1H– 1H NOE spectroscopy. Biochemistry 34, 6533– 6539. 14. Miyazaky, J., Hideg, K., and Marsh, D. (1992) Interfacial ionization and partitioning of membrane-bound local anaesthetics. Biochim. Biophys. Acta 1103, 62– 68. 15. Anel, A., Richieri, G. V., and Kleinfeld, A. M. (1993) Membrane partition of fatty acids and inhibition of T cell function. Biochemistry 32, 530 –536. 16. Aranda, F. J., and Villalain, J. (1997) The interaction of abietic acid with phospholipid membranes. Biochim. Biophys. Acta 1327, 171–180. 17. Frankel, E. N. (1983) Volatile lipid oxidation products. Prog. Lipid Res. 22, 1–33. 18. Gardner, H. W. (1989) Oxygen radical chemistry of polyunsaturated fatty acids. Free Radic. Biol. Med. 7, 65– 86. 19. Kusumi, A., Subczynski, W. K., and Hyde, J. S. (1982) Oxygen transport parameter in membranes as deduced by saturation recovery measurements of spin–lattice relaxation times of spin labels. Proc. Natl. Acad. Sci. USA 79, 1854 –1858. 20. Momo, F., Wisniewska, A., and Stevanato, R. (1995) EPR study of spermine interaction with multilamellar phosphatidylcholine liposomes. Biochim. Biophys. Acta 1240, 89 –94. 21. Stevanato, R., Wisniewska, A., and Momo, F. (1997) Interaction of spermine with dimyristoylphosphatidylglycerol multilamellar liposomes. Arch. Biochem. Biophys. 346, 203–207. 22. Chung, L., Kaloyanides, G., McDaniel, R., McLaughlin, A., and Mclaughlin, S. (1985) Interaction of gentamicin and spermine with bilayer membranes containing negatively charged phospholipids. Biochemistry 24, 442– 452. 23. Sanson, A., Ptak, M., Rigaud, J. L., and Gary-Bobo, C. M. (1976) An ESR study of the anchoring of spin-labeled stearic acid in lecithin multilayers. Chem. Phys. Lipids 17, 435– 444. 24. Bultman, T., Lin, H., Wang, Z., and Huang, C. (1991) Thermotropic and mixing behaviour of mixed-chain phosphatidylcholines with molecular weights identical with that of L-␣-dipalmitoylphosphatidylcholine. Biochemistry 30, 7194 –7202. 25. Akutsu, H., and Seelig, J. (1981) Interaction of metal ions with phosphatidylcholine bilayer membranes. Biochemistry 20, 7366 –7373. 26. Scherer, P., and Seelig, J. (1989) Electric charge efects on phospholipid headgroups. Phosphatidylcholine in mixtures with cationic and anionic amphiphiles. Biochemistry 28, 7720 –7728. 27. Beschiaschvili, G., and Seelig, J. (1990) Melittin binding to mixed phosphatidylglycerol/phosphatidylcholine membranes. Biochemistry 29, 52–58. 28. Bechinger, B., and Seelig, J. (1991) Interaction of electric dipoles with phospholipid headgroups. A 2H and 31P NMR study of phloretin and phloretin analogues in phopshatidylcholine membranes. Biochemistry 30, 3923–3929. 29. Toner, M., Vaio, G., McLaughlin, A., and McLaughlin, S. (1988) Biochemistry 27, 7435–7443. 30. Meers, P., Hong, K., Bentz, J., and Papahadjopoulos, D. (1986) Biochemistry 25, 3109 –3118.