Layer Formations of Dipalmitoylphosphatidylcholine Liposomes in the

Structural and calorimetrical studies of the effect of different aminoglycosides on DPPC liposomes. Ágnes Oszlánczi , Attila Bóta , Gábor Czabai ,...
0 downloads 0 Views 203KB Size
Langmuir 1999, 15, 3101-3108

3101

Layer Formations of Dipalmitoylphosphatidylcholine Liposomes in the Pretransition Range Attila Bo´ta,*,† Tama´s Drucker,† Manfred Kriechbaum,‡ Zsolt Pa´lfia,§ and Ga´bor Re´z§ Institute of Biophysics and X-ray Structure Research, Austrian Academy of Sciences, Steyrergasse 17, A-8010 Graz, Austria, Institute of Physical Chemistry, Technical University of Budapest, Mu¨ egyetem rkp 3. H-1111 Budapest, Hungary, and Department of General Zoology, Lora´ nd Eo¨ tvo¨ s University, Puskin u. 3. H-1088 Budapest, Hungary Received April 29, 1998. In Final Form: February 2, 1999 Destroyed lamellar and chain packing arrangements of the dipalmitoylphosphocholine-water (30% w/w) system are formed in the pretransition range (31-35.5 °C) under quasi-equilibrium conditions observed by means of simultaneous small- and wide-angle X-ray scattering method. The peak profiles of the scattering curves detected in the pretransition range were modeled as the superpositions of the fitted profiles of the Lβ′ and Pβ′ phases corresponding to the one-dimensional layer arrangements and the subcells in the chain packing. On the basis of the fitted profiles the ratios of the phases can be given as a function of the temperature. It can be stated that (i) both the Lβ′ and Pβ′ phases are present in the temperature range of the pretransition and (ii) in the Pβ′ phase the layer arrangements and chain packing do not change to the same extent. The maximum loss in the layer arrangement was observed around 33 °C. The accumulation of the defect structures appears to be highly permanent as it was observed after sequential heat treatments, i.e., quenching to the temperature domain of the Lβ′ phase (28 °C) and then reheating above the pretransition range to the temperature domain of the Pβ′ phase (38 °C). A memory effect occurs in the sense that the destroyed structures are restored in the temperature range of phase Pβ′ after the formation of a largely reconstructed phase Lβ′. The destroyed layer structures formed in the pretransition and in the rippled phase temperature domains are not identical, as observed by freeze-fracture electron microscopy.

Introduction Phospholipids are the most frequent components of biological membranes.1 By dispersing these molecules in water, uni- or multilamellar vesicles are formed with concentric onionlike shells of alternating water and lipidbilayer regions.2 Due to their similarities the vesicles are considered as a model-system for living cell membranes.3 Depending on the temperature, the liposomes prepared from dipalmitoylphosphatidylcholine (DPPC) and water exhibit at least five different multilamellar structures in excess water.4-10 These structures are characterized by the specific lattice parameters of the periodical shell and by specific subcells in the chain packing.11-16 Accordingly, * To whom correspondence should be addressed. E-mail: [email protected]. † Technical University of Budapest. ‡ Austrian Academy of Sciences. § Lora ´ nd Eo¨tvo¨s University. (1) Cevc, G.; Marsh, D. Phospholipid bilayers. Physical principles and models; John Wiley & Sons: New York, 1987. (2) Liposomes-a practical approach; New, R. R. C., Ed.; IRL & Oxford University Press: Oxford, England, 1989. (3) Mouritsen, O. G.; Dammann, B.; Fogedby, H. C.; Ipsen, J. H.; Jeppesen, C.; Jørgensen, K.; Risbo, J.; Sabra, M. C.; Sperotto, M. M.; Zuckermann, M. J. Biophys. Chem. 1995, 55, 55. (4) Chapman, D.; Williams, R. M.; Ladbroke, B. D. Chem. Phys. Lipids 1967, 1, 445. (5) Tardieu, A.; Luzatti, V.; Reman, F. C. J. Mol. Biol. 1973, 75, 711. (6) Hinz, H.-J.; Sturtevant, J. M. J. Biol. Chem. 1972, 24, 6071. (7) Janiak, M. J.; Small, D. M.; Shipley, G. G. Biochemistry 1976, 15, 4575. (8) Lentz, B. R.; Freire, E.; Biltonen, R. L. Biochemistry 1978, 17, 4475. (9) Ruocco, M. J.; Shipley, G. G. Biochim. Biophys. Acta 1982, 691, 309. (10) Jørgensen, K. Biochim. Biophys. Acta 1995, 1240, 111. (11) Matuoka, S.; Yao, H.; Kato, S.; Hatta, I. Biophys. J. 1993, 64, 1456. (12) Brady, G. W.; Fein, D. B. Biochim. Biophys. Acta 1977, 464, 249.

there are four transitions between the structural states. The pretransition occurs between the gel (Lβ′) and the rippled gel (Pβ′) phases. These parent phases have a characteristic one-dimensional lattice. Moreover, the rippled gel phase possesses long range, two-dimensional periodic structures, and it has a lower symmetry in the one-dimensional lattice than the nonrippled gel phase. In the temperature domain of the rippled gel phase, there is a metastable rippled phase Pβ′(mst) occurring during slow cooling from the temperature domain of the LR phase.17 Up to now considerable uncertainty has existed in the literature on the structural mechanisms underlying the pretransition (Lβ′ T Pβ′) of hydrated dipalmitoylphosphatidylcholine.7,18-23 The physicochemical basis for a decisive interpretation appears to be not yet sufficiently settled, in terms of the thermodynamic and kinetic determinants. One of the reasons for this ambiguity certainly lies in the fact that this particular phase transition is relatively slow and is accompanied by only minor changes in ∆H and ∆V. Furthermore, the kinetics are quite different in heating and cooling directions and are strongly dependent on whether the transition process is near to or far away from the equilibrium.17,24-29 This renders it highly difficult to (13) Stamatoff, J.; Feuer, B.; Guggenheim, H. J.; Tellez, G.; Yamane, T. Biophys. J. 1982, 38, 217. (14) Kriechbaum, M.; Laggner, P. Prog. Surf. Sci. 1996, 51, 233. (15) Rappolt, M.; Rapp, G. Eur. Biophys. J. 1996, 24, 381. (16) Quinn, P. J.; Takahashi, H.; Hatta, I. Biophys. J. 1995, 68, 1374. (17) Tenchov, B. G. ; Yao, H.; Hatta, I. Biophys. J. 1989, 56, 757. (18) Marsh, D.; Watts, A.; Knowles, P. F. Biochim. Biophys. Acta 1977, 465, 500. (19) Hawton, M. H.; Doane, J. W. Biophys. J. 1987, 52, 401. (20) Yang, C. P.; Nagle, J. F. Phys. Rev. A 1988, 37, 3993. (21) Wack, D. C.; Webb, W. W. Phys. Rev. A 1988, 40, 2712. (22) Laggner, P.; Kriechbaum, M. Chem. Phys. Lipids 1991, 57, 121. (23) Tu, K.; Tobias, D. J.; Blasie, J. K.; Klein, M. L. Biophys. J. 1996, 70, 595.

10.1021/la980507b CCC: $18.00 © 1999 American Chemical Society Published on Web 04/08/1999

3102 Langmuir, Vol. 15, No. 9, 1999

combine the results from different techniques with their different time scales and perturbation energies.17,27,30 In the transition temperature domain fluctuations increase significantly due to the weak first-order nature of the pretransition and states are formed with destruction of the one-dimensional layer arrangements.22,31 The defect structures appear to be highly permanent as was observed in sequential measurements.32,33 The goal of the present research was to reveal the features of the layer arrangements focusing on the slow structural development of the gel and rippled gel phases under different conditions of long-term controlled annealing. Materials and Methods Synthetic 1,2-dipalmitoyl-sn-3-glycerophosphocholine (DPPC) with a purity higher than 99% was purchased from Avanti Polar Lipids (Alabaster, AL) and used without further purification. The high purity of the lipid was confirmed by thin-layer chromatography. Deionized, triple quartz-distilled water was added to the dry lipid powder under a nitrogen gas atmosphere to yield a lipid concentration of 30% w/w. The mixture was kept at 50 °C for about 10 h and vortexed frequently. After incubation the sample was quenched to 4 °C, then reheated to 50 °C again, and vortexed intensively. The process was repeated 10 times for homogeneous hydration. The sample was stored at 4 °C. For X-ray measurements the lipid dispersion was transferred into thin-walled quartz capillaries (Hilgenberg, Germany) with a diameter of 1 mm. To remove air bubbles the capillaries were centrifuged for 10 min at 1000g at room temperature. The capillaries were sealed with a two-component synthetic resin and transferred into metal capillary holders placed into an aluminum block. This block was positioned into the beamline directly and was used as a thermal gradient incubator for controlled annealing at different temperatures around the pretransition. Both ends of the block were held at the desired temperatures by two thermostats. The block was moved across the beamline to make exposures on each sample. The actual temperatures were constant within less than 0.05 °C as controlled by a thermocouple. The small-angle X-ray (SAX) camera was supplied with a twodimensional position-sensitive detector (ITL Ltd., Mastings, U.K.) for the exposure of the SAX scattering and a linear positionsensitive detector (PSD 50-M, M. Braun, Garching, Germany) for the detection of the wide-angle X-ray (WAX) scattering. The camera was operated at the point-focus window of a rotating copper anode (Rigaku Denki, Japan) used typically at 6 kW. The beamline was collimated by a pinhole. A Ni filter was used to eliminate Kβ radiation. Each X-ray exposure took 1000 s. The scattering data were collected in the small angle range from 1/250 to 1/10 Å-1 and in the wide angle range from 1/4.5 to 1/3.5 Å-1. For the control study, the capillaries were heated to 28 °C at a rate of 5 °C/min; then, after 3 days of incubation at this temperature, they were measured by SAXS. After measurement, they were heated to 38 °C, incubated at this temperature for 3 days, and measured by SAXS. These control measurements (24) Tsuchida, K.; Hatta, I.; Imaizumi I. S.; Ohki, K.; Nozowa, Y. Biochim. Biophys. Acta 1985, 812, 249. (25) Tsuchida, K.; Ohki, K.; Sekiya, T.; Nozawa, Y.; Hatta, I. Biochim. Biophys. Acta 1987, 898, 53. (26) Tsuchida, K.; Hatta, I. Biochim. Biophys. Acta 1988, 945, 73. (27) Caffrey, M.; Fanger, G.; Magin, R. L.; Zhang, J. Biophys. J. 1990, 58, 677. (28) Tristram-Nagle, S.; Wiener, M. C.; Yang, C.-P.; Nagle, J. F. Biochemistry 1987, 26, 4288. (29) Kodama, M.; Kuwabara, M.; Seki, S. Biochim. Biophys. Acta 1982, 689, 567. (30) Laggner, P.; Kriechbaum, M.; Rapp, G.; Hendrix, J. 2nd Euro. Conf. Prog. in X-ray Sync. Radiat. Res. 1990, 995. (31) Bo´ta, A.; Kriechbaum, M.; Laggner, P. ACH-Models Chem. 1997, 134, 299. (32) Laggner, P.; Kriechbaum, M.; Bo´ta, A.; Rapp, G. In Synchrotron Radiation in the Biosciences; Chance, B., et al., Eds.; Clarendon Press: New York, 1994; p 204. (33) Bo´ta, A.; Kriechbaum, M. Colloids Surf. A 1998, 141, 441.

Bo´ ta et al. served to adjust the position of each capillary in the aluminum block, to check the filling and sealing of the capillaries. X-ray absorption was measured to state whether the diameter of the capillaries was within the range of 1 ( 0.05 mm. Capillaries outside this range were discarded. This way, the SAXS intensities of the Lβ′ and Pβ′ phases were within a range of 3% at the maximum value of the first Bragg reflection. The regularity of the layer structures of the Lβ′ and Pβ′ phases was checked by measuring the fwhm (full width at half-maximum) by SAXS. In the angular small-angle scattering curves obtained from the two-dimensional X-ray pictures the first Bragg profiles of the one-dimensional layer arrangements were at the positions of 1/63.5 and 1/70.5 Å-1, corresponding to the reported periodicities in the Lβ′ and Pβ′ phases, respectively. The ripple period of the Pβ′ phase appears in the form of a weak diffuse peak at 1/120 Å-1. In the wide angle range the characteristic chain packings were observed at the peak positions of 1/4.18 (with a satellite peak at 1/4.08) and 1/4.15 Å-1 corresponding to the hybrid and hexagonal subcells of the Lβ′ and Pβ′ phases, respectively. For data evaluation the SAXS and WAXS-profiles were fitted by Lorentzian functions

I(s) ) A/[1 + b ((s - speak position)/fwhm)2] + C

(1)

where fwhm is the full width at half-maximum and A, b, and C are constant, C corresponds to the background,34 s is the absolute value of the scattering vector defined as s ) (2 θsin)/λ, 2θ is the scattering angle, and λ is the wavelength (1.542 Å). For freeze-fracture the lipid dispersion was prepared in a quantity of about 200 mg and stored in small covered glass vials. The same thermal gradient block was used for the incubation as for the X-ray measurements. The gold specimen holders used in freeze-fracture were preincubated at the same temperatures as the samples. Droplets of 1-2 µL were pipetted onto the gold holders which were then immediately plunged into partially solidified Freon for 20 s freezing and placed and stored in liquid nitrogen. Fracturing was carried out at -100 °C in a Balzers freeze-fracture device (Balzers AG, Vaduz, Liechtenstein). The fractured faces were etched for 30 s at -110 °C. The replicas, prepared by platinum-carbon shadowing, were cleaned with a solution of hypochlorous acid and washed on distilled water. From pure water, the replicas were picked up on 200 mesh copper grids. The electron micrographs were taken in a JEOL JEM-100 CX II electron microscope (Japan). For the evaluation of the micrographs optical density-based image analysis was used (Fenestra Vision, Kinetic Imaging Inc., Liverpool, U.K.). The particle size measurements (ZetaPlus Particle Sizing Device, Brookhaven Inst. Corp.) yielded an average diameter of about 0.84 µm in diluted liposome dispersion. Just then, we have found liposomes with an average diameter of 1-2 µm (with a maximum diameter of 6 µm) by the freeze-fracture method at a concentration of 30% w/w. The typical shape of liposomes was spheric, but in some cases the external shells formed irregular sheets.

Results Prehistory in the Pretransition Range. In the diffractograms of the rippled (Pβ′) and nonrippled (Lβ′) gel and the liquid crystalline (LR) phases, the fully hydrated DPPC/water liposomes show radially symmetric smallangle X-ray patterns in the temperature range of 28-45 °C characteristic of one-dimensionally periodic, unoriented systems. The X-ray pictures show further intermediate states with a different one-dimensional lamellar arrangement in the pretransition range. The patterns show a broadening and decrease of the small angle diffraction peaks indicating a drastic loss in the long-range correlation between the bilayer lamellae. Similar features have been reported for slow scan rate, so-called quasi equilibrium experiments.27 After 2 days of incubation the different changes of the Bragg profiles demonstrate strong thermal (34) Rappolt, M.; Rapp, G. Ber. Bunsen-Ges. Phys. Chem. 1996, 100, 1153.

Layer Formations of Liposomes

Figure 1. SAXS and WAXS curves detected in the pretransition range after 2 days incubation at the denoted temperatures. The investigated states are originated from different temperatures corresponding to LR, Pβ′, and Lβ′ phases, i.e., after quenching from 45 °C (upper curve in group of three), after quenching from 38 °C (middle curve in group of three), and after heating from 28 °C (lower curve in group of three). The characteristic peaks of the reference gel phases are at the denoted positions as follows: (i) in the small angle regime, sP ) 1/70.5 Å-1, sL ) 1/63.5 Å-1; (ii) in the wide angle regime, sh1 ) 1/4.18 Å-1, sh2 ) 1/4.08 Å-1 for the hybrid subcell and shex ) 4.15 Å-1 for the hexagonal subcell.

prehistory in the pretransition range as the diffractograms show qualitative differences depending on the starting temperatures of 28 (Lβ′), 38 (Pβ′) and 45 °C (LR), respectively, i.e., below or above the pretransition temperature range, as shown in Figure 1. Only a continuous change of Bragg profiles can be observed in the curves of the parent phases. The peak-centers shift from the position of the Lβ′ to the position of Pβ′. The profiles that belong to heating are sharper than those belonging to cooling. A more drastic change of profiles appears in states which originate from the liquid crystal phase. The metastable rippled phase Pβ′(mst) with drastically destroyed lamellar arrangements appears to be a well reconstructed Pβ′ phase at 33.3 °C and exhibits a noncorrelated layer structure at 32.2 °C (upper curves at 34.5, 33.3, and 32.2 °C in Figure 1a). The wide-angle scattering profiles of the subcells in the chain region, monitored simultaneously by SAXS, reveal that the changes and deformation occur in both lamellar and chain packing arrangements as can be seen in Figure 1b. For the metastable Pβ′ phase the destruction in the chain packing is not so dramatic as that in the lamellar arrangements. The pretransition exhibits about 1 order of magnitude faster relaxation in the heating than in the cooling direction; therefore, the scanning measurements described in the literature were performed in the heating direction. Nevertheless, the slow scan rates applied in many reported experiments prove not to be slow enough to get the final states in the transition region. In these cases the transition point appears at a higher temperature value. For example by the DSC method it was found that the transition point is at about 35.2 ( 0.3 °C (at 0.25 °C/min8 and 0.5 °C/min17 heating scan rates). For time-resolved SAXS measurements it has been reported that the transition point falls

Langmuir, Vol. 15, No. 9, 1999 3103

Figure 2. Change of SAXS and WAXS profiles after long incubation in the pretransition range at the denoted temperatures. The curves were detected after 1 day of incubation (lower curves in pair of curves) and 5 days of incubation (upper curve in pair of curve). The peak positions of the reference gel phases are also denoted.

Figure 3. Ratio of the parent phases in the fitted SAXS- and WAXS-peak profiles monitored in the pretransition range (Lβ′ phase, triangles; Pβ′ phase, circles). The data are determined by SAXS (open symbols) and by WAXS (bold symbols) on three independent series of measurements including new sample preparations.

between 32 and 33.5 °C.27 On the basis of the shift of the diffraction profiles detected after 2 days of incubation and shown in Figure 1a, it can be stated that the pretransition is in progress at 31 °C. Following the time relaxation of the layer and chain packing some significant differences in the Bragg profiles monitored after 1 and 5 days of incubation can be observed only in a narrow temperature interval between about 31.0 and 33.3 °C as shown in parts a and b of Figure 2 (cooling from 38 °C). Longer incubation does not yield any significant changes in the profiles, suggesting that the transition point (Tp) is in the vicinity of 32 °C. To quantify the transitional states we have analyzed the WAXS and first Bragg SAXS peaks. The peak profiles were modeled as superpositions of the fitted profiles of the parent phases. Thereby, the normalized amplitudes of the fitted profiles give the ratios of the phases which are present during the transition, and the ratios can be given as a function of the temperature (Figure 3). These

3104 Langmuir, Vol. 15, No. 9, 1999

Bo´ ta et al.

data evaluation has revealed that the two parent phases are present in a wide temperature range. At 31.4 ( 0.2 °C the layer arrangement corresponding to the Lβ′ phase is present in 50%, but the ratio corresponding to the Pβ′ phase appears only at a higher temperature (at about 33 °C). The sum of the ratios is less than 100%, which demonstrates that in the transition range a destruction of the one-dimensional arrangements occurred. In chain packing a simultaneous change can be observed, as the sum of the ratios is close to 100% in the whole pretransition range. Here, 50% of both phases are present at the same temperature of 31.3 ( 0.2 °C. In both the small- and wideangle regime the width of each parent phase profile is larger than that present in the reference states at 28 and 38 °C, respectively, showing that smaller domains exist in the whole pretransition range than in the fully developed states. Description of the Transition States by a Centrosymmetrical Shell Model. The scattering function of an unoriented, random system can be generally calculated as

I(s) )

∑|S(s)|2

(2)

Figure 4. Modeled SAXS profiles of the two parent phases (Pβ′, upper curve; Lβ′, lower curve) and of two transitional states with different number of shells of each phase (1-5, a; 2-7, b).

where S(s) is the structure factor and the sum corresponds to the number of scattering units. Assuming that the liposomes are spherical multilamellar units, with radius r (with internal radius R) and each shell is built up from different subshells corresponding to alternating water and lipid bilayer regions, the structure factor is given by i)k

S(R,k,s) ) const

(F(r) - Fwater)[(sin(2πsr) ∑ i)1 2πsr cos(2πsr))/(2πs)3] (3)

where k is the number of subshells in each liposome. A stepped scattering length density was used as a function of the radius. The repeat distance consists of one water layer, two head layers of lipid and two hydrocarbon chain layers. The scattering length density data have been given in the literature;35 nevertheless, we have modified those, considering the ratios of the Bragg intensities of the parent phases at different order detected for the reference states. The calculation in eq 2 was taken over different R, assuming a Gaussian distribution of particle size of the liposomes, with Rmin ) 150 Å and Rmax ) 300 Å. Moreover, we have used another condition, namely, in the liposome the number of the shells, n was changed randomly between 50 and 100. We have intended to calculate the shape of scattering curves in arbitrary units, therefore in eq 3 the constant was taken into account with a value of 1. Using the shell model, it can be assumed that the Bragg peak is very sharp already for n ) 50. The characteristic shape of the parent phases can be reconstructed only if some fluctuations of the scattering length density are considered. We used random, independent fluctuations over the width of each subshell. The frequency of the width was calculated by a Gaussian distribution within one standard deviation range. It was found that the scattering profiles of the parent phases could be well modeled when the relative standard deviation related to the repeat distance was 15% for Lβ′ and 23% for Pβ′, respectively. The transitional states can be described by assuming that the two parent phases with different number of shells coexist in each liposome. In this case it was necessary to apply (35) Takeda, T.; Akabori, K.; Toyoshima, Y.; Komura, S.; Takebe, Y. Jpn. J. Appl. Phys. 1987, 1791.

Figure 5. Characteristic experimental SAXS and WAXS curves of the long time incubated sample after sequential treatment: first at 33.2 °C, then after cooling to 28 °C, and finally after reheating to 38 °C. The curves of the parent phases are also shown. The ripple periodicity appears at about 1/120 Å in the Pβ′ phase (marked by an arrow).

a higher level of the fluctuations. For 40% relative standard deviation the effect of fluctuation can be seen in Figure 4. Besides the modeled scattering profiles of the parent phases, two transitional states are shown in which the parent phases are present in a ratio of 1:1. One consists of alternating “domains” with shell number between 2 and 7, the other one between 1 and 5. The curve of the former state has two peaks, the curve of the latter state has only one, and they are diffuse. This model calculation reveals that the fluctuation increased significantly in the vicinity of the pretransition, leading to the broadening and loss in intensity of the Bragg profiles. The time relaxation, observed in the transition range is coupled with the formation of “domains” which consist of shells small in number. Accumulation of Defects in the Thermally Adjacent Phases. A maximum loss in layer arrangement was observed at 33 ( 0.5 °C. Therefore the further change of defects was followed on the state which was formed in this temperature range after 5 days. After sequential treatments, i.e., cooling (at a rate of about 5 °C/min) to 28 °C and then after 2 days of incubation reheating (at a rate of about 5 °C/min) to 38 °C, the scattering profiles have

Layer Formations of Liposomes

Figure 6. Accumulation of defects formed in the pretransition range, after cooling to 28 °C and then reheating to 38 °C, illustrated by the change of the control parameter (the reciprocal value of the full width at half-maximum of the first Bragg peaks) (28 °C, triangles; 38 °C, circles).

evidenced differently reconstructed Lβ′ and Pβ′ phases (Figure 5). During earlier SAXS studies no Pβ′ phase was observed at 30 °C. Accordingly, without thermal prehistory, 28 ° was considered as a temperature where only the regular Lβ′ phase exists. The curve of the quenched state at 28 °C has a wider Bragg profile than that of the reference state. After the sample was reheated to 38 °C, the profile

Langmuir, Vol. 15, No. 9, 1999 3105

seems to be similar to that detected at 33.2 °C. This indicates that the layer structures maintain the defects once accumulated around the pretransition for prolonged times even after a double transition. After closer inspection, the fit provides more structural details in both the small and the wide angle regimes.33 The two states formed at 33.3 and 38 °C are not equivalent. Generally, the domain sizes are smaller in the state formed at 38 °C than those formed in the pretransition region, at 33.2 °C. Domains of the Lβ′ phase are also present at 38 °C with a slightly modulated periodicity. For the whole series, which had been kept in the transition range, the relaxation of layer structures was followed in the same way, i.e., by cooling to 28 °C and then reheating to 38 °C. The loss in the layer correlation is characterized by the reciprocal value of the full width at half-maximum (fwhm) of the first Bragg peaks and is shown in Figure 6. As in the case of the reference phase Lβ′, the reciprocal value of fwhm was found maximum, showing that phase Lβ′ has a higher degree of order than phase Pβ′; its value was used for the normalization of the other values. Hence, this parameter as a control parameter is arbitrarily set between 0 and 1. A system with no regular layer correlation would have the control parameter n ) 0, since its continuous small angle pattern would merge with the tangents under the Bragg peaks. At 28 °C the states that were previously annealed in the narrower transition range (between ca. 31-34 °C) did not exhibit a maximum presence of Lβ′. This irreversibility indicates a thermal prehistory which is connected

Figure 7. Freeze-fracture electron micrographs of dipalmitoylphosphatidylcholine liposome. Liposomes were quenched from (a) 28 ( 0.2 °C, (b) 38 ( 0.2 °C, and (c, d) 33 ( 0.2 °C. Previously the lipid dispersions were heated from room temperature (about 24 °C) to these temperatures and incubated for one week. The bar represents 0.25 µm in all cases. The direction of shadowing is marked by an arrow.

3106 Langmuir, Vol. 15, No. 9, 1999

Bo´ ta et al.

Figure 8. Freeze-fracture electron micrographs of dipalmitoylphosphatidylcholine liposomes. Liposomes were quenched from 28 ( 0.2 °C, after 3 days of incubation at this temperature after the temperature shift from (a) 38 ( 0.2 °C and (b) 33 ( 0.2 °C, respectively and from 38 ( 0.2 °C, after the 3 days of incubation and the temperature shift from 28 and previously from 33 ( 0.2 °C (c,d,e,f).

with the destroyed lamellar structures. The effect of the prehistory was even more pronounced, when the samples were heated into the Pβ′ temperature domain. The control parameter of Pβ′ is much less than expected in the case of full reversibility. The recovery of the destroyed lamellar structures, present in the pretransition range, indicates a memory effect, which can be characterized by the difference of the control parameters in the Lβ′ and Pβ′ phases, respectively. This phenomenon was not observed in the case of a similar series of experiments with altered sequence of incubations, i.e., first heating to 38 °C and then cooling to 28 °C. In this case the results show minor

thermal prehistory and no indication of memory, suggesting that the conservation of defect structures is extended in the Pβ′ phase to a higher range than the in Lβ′ phase. We have observed that the extent of the memory effect strongly depends on the temperature to which the samples were cooled in the second step and also on the temperature to which the samples were heated in the third step. The memory effect was stronger when the change in the temperature during the heat treatments was smaller. So far we have found that the memory effect was the strongest in the sample which was kept at about 33 °C in the first step. Now we have reported results of

Layer Formations of Liposomes

the study in which the change in the temperature was 5 °C in both the cooling and heating directions. Freeze-Fracture. To follow the formation of the layer defects, scattering data themselves do not provide unambiguous interpretations. To overcome this problem the investigations were supplemented by freeze-fracture technique, which is an excellent method of direct visualization of irregular local structures.36-38 First we studied an intermediate and the two reference states. The lipid dispersions were heated from room temperature to three different temperatures: 28 ( 0.2, 33 ( 0.2, and 38 ( 0.2 °C, i.e., to the temperature domains of the Lβ′ phase, the pretransition regime, and the Pβ′ phase, respectively. The freeze-fracture faces of the structures formed after 5 days of incubation are shown in parts a-d of Figure 7. At 28 °C the typically large-size multilamellar bodies composed of closely packed parallel membrane layers with smooth surface appear (Figure 7a), which clearly exhibit the known morphology of the gel phase.25,39 The parallel membranes form large sheets, the plane of which, however, is gently and irregularly wrinkled at some places. The well-known characteristic rippled freezefracture pattern of the Pβ′ phase appears in Figure 7b. The total surface of the membranes is densely and regularly rippled with a periodicity of 149 ( 17 Å and exhibits a banded pattern with 3-fold symmetry.40,41 At 33 °C the multilamellar smooth surface of the Lβ′ phase and the rippled surface of the Pβ′ phase is also present, but the latter shows a less ordered pattern at this temperature than at 38 °C (Figure 7c,d). It is obvious that the parent phases still exist at this temperature. In general, the rippled system at 33 °C is significantly less abundant and less ordered than at higher temperature. The ripples run more-or-less parallel to form concentric or destroyed patterns. It was observed that the lateral extension of the domains with different parent phases was larger than 0.05 µm. Therefore, in the SAXS curves the broadening of the Bragg profiles can be ascribed rather to the regularity of the layer arrangement than to the lateral size of the domains. In the second part of our freeze-fracture study we attempted to reveal some visual information on the surface patterns in the states which proved to be very similar as judged by their scattering patterns. The effect of the heat treatment on the morphology formed in the intermediate state (33 °C) (INT) and in the reference rippled phase (38 °C) (REF) was also studied. Samples were rapidly quenched to 28 °C from these temperatures. After 3 days of incubation at 28 °C the smooth surface of the Lβ′ phase appeared (Figure 8a). Although the membrane faces are very similar in INT and REF, they weakly show a prehistory as the sample quenched from 33 °C contains strongly wrinkled surfaces in some places (Figure 8b). To follow the formation of the defects, the samples were reheated from 28 °C (after 3 days) to 38 °C. As the Pβ′ phase has several structural features, more details can be measured in its membrane face than in the case of the earlier discussed Lβ′ phase. This means that the rippled patterns could be described by some parametrization, but (36) Hope, M. J.; Wong, K. F.; Cullis, P. R. J. Electron Microsc. Technol. 1989, 13, 277. (37) Luna, J. E.; McConnel, H. M. Biochim. Biophys. Acta 1977, 466, 381. (38) Krbeck, R.; Gebhardt, C.; Gruler, H.; Sackman, E. Biochim. Biophys. Acta 1979, 554, 1. (39) Copeland, B. R.; McConnel, H. M. Biochim. Biophys. Acta 1980, 599, 95. (40) Stewart, T. P.; Hui, S. W.; Portis, A. R.; Papahadjopoulos, Jr. D. Biochim. Biophys. Acta 1979, 556, 1. (41) Seul, M.; Andelman, D. Science 1995, 267, 476.

Langmuir, Vol. 15, No. 9, 1999 3107

Figure 9. Scheme of the characteristic quantities of the patterns observed in freeze-fracture electron micrographs at 38 °C, corresponding to the surface features of the Pβ′ phase.

the micrographs, especially in the case of the INT sample, revealed large areas without any recognizable pattern. In the pictures shown in parts c-f of Figure 8, the typical rippled structure of the Pβ′ phase is certainly present but several differences are visible in the appearance of the pattern. The regular 3-fold symmetry of the banded structure is not present in the sample that was originally kept in the pretransition region. Furthermore, the periodicity in these samples seems to be irregular. To characterize the different details of the patterns the following characteristic quantities were measured: the length of the periodicity of the ripples, the width of the band, and the number of the ripples in the half-band (from the ridge to the borderline of the band) (Figure 9). Comparing the characteristic quantities in Figures 7b and 8c-f it can be concluded that (i) the periodicity of the ripples is larger in the INT sample than it was observed and reported in the reference states (INT, 185.7 ( 51.4 Å, REF, 149.3 ( 17.3 Å) (ii) the width of the band shows a much higher fluctuation in the INT sample (the width and its standard deviations are 2706 ( 1387 Å in INT and 2634 ( 326 Å in REF, respectively), and (iii) the number of the ripples at the half band of INT is less than at REF (INT, 7.9 ( 5.1; REF, 10.7 ( 4.5). Conclusions The formation of the Pβ′ layer structure takes place at higher temperature than the changes of the subcells in the chain range. Moreover, this partially occurs with a minor change in enthalpy as it was observed after sequential DSC measurements27 and hence fluctuations significantly increase in the layer arrangements and defect structures are formed. The formation of the defect structure is facilitated, since their enthalpy of formation is compensated by the increase of the configurational entropy. The defect structures are manifested not only in the regularity of the one-dimensional arrangements but also in the creation of the layer curvatures.42,43 Thermodynamically, the curvatures are preferred formations in the Pβ′ phase but hindered in the Lβ′ phase. In the Pβ′ phase the spontaneous curvature results in a larger decrease in the total free energy than in the Lβ′ phase, leading to a different destruction of the regularity in the two phases.44 Hence, an asymmetry in the regularity is expected and observed on the two sides of Tp.31 On the other hand, the ability of the Pβ′ phase to keep the (42) Gruner, S. M. J. Phys. Chem. 1989, 93, 7562. (43) Charvolin, J. Mol. Cryst. Liq. Cryst. 1991, 198, 145. (44) Gebhardt, C.; Gruler, H.; Sackmann, E. Z. Naturforsch. 1977, 32c, 581.

3108 Langmuir, Vol. 15, No. 9, 1999

structural defects connected to the curvatures is higher than that of the Lβ′ phase. After quenching from the pretransition range to a temperature T < Tp, a number of curvatures may be retained and they form the boundaries of the grains of the Pβ′ phases, supposing a nucleation mechanism. These defects promote the formation of further irregularities in the temperature domain of the Pβ′ phase. Besides fluctuations, the effect of impurities always present in the system must be taken into consideration.45-47 The impurities tend to accumulate in the interfaces of the domains. Therefore the modulated (45) Ipsen, J. H.; Mouritsen, O. G.; Cruzeiro-Hansson, L. In Phase Transitions in Soft Condensed Matter; Riste, T., Sherrington, D., Eds.; NATO ASI Series B: Physics; Plenum Press: New York, 1989; p 283. (46) Lohner, K. Chem. Phys. Lipids 1991, 57, 341. (47) Jørgensen, K.; Ipsen, J. H.; Mouritsen, O. G.; Bennett, D.; Zuckermann, M. J. Biochim. Biophys. Acta 1991, 1067, 241.

Bo´ ta et al.

interfaces can significantly contribute to the formation of the defect structures, especially in the Pβ′ phase. The regions of accumulated impurities enmeshing not regularly in the whole liposome can reduce the size of the homogeneous ranges and hinder the formation of the longrange surface regularity of the Pβ′ phase. Acknowledgment. This work has been supported by an exchange fellowship of the Austrian Academy of Sciences, the Sze´chenyi-Istva´n-Scholarship of the Hungarian Academy of Sciences and the Hungarian Scientific Fund OTKA (T 014396). The first author gratefully acknowledges the support of Prof. Laggner. We thank Ms. T. Kiss, Ms. M. Kocsis, and Mr. M. Ko¨vi for technical assistance. LA980507B