10300

J. Phys. Chem. B 1999, 103, 10300-10310

Thermotropic Phase Behavior of Cationic Lipid-DNA Complexes Compared to Binary Lipid Mixtures Roman Zantl,† Laura Baicu,† Franck Artzner,†,§ Irene Sprenger,† Gert Rapp,‡ and Joachim O. Ra1 dler*,† Lehrstuhl fu¨ r Biophysik, E22, Technische UniVersita¨ t Mu¨ nchen, James-Franck-Str. 1, D-85748 Garching, Germany, and European Molecular Biology Laboratory, EMBL c/o DESY, Notkestrasse 85, D-22603, Hamburg, Germany ReceiVed: May 18, 1999; In Final Form: August 16, 1999

The thermotropic phase behavior of zwitterionic/cationic binary lipid mixtures is investigated and compared to its corresponding lipidic phase diagram of mixtures complexed with DNA. We focus on isoelectric cationic lipid-DNA condensates where the number of cationic lipids equals the number of phosphate groups on the DNA. Using differential scanning calorimetry, X-ray scattering, freeze fracture electron microscopy, and film balance, we studied mixtures of di-myristoyl-phosphatidyl-choline (DMPC) and the cationic lipid, dimyristoyl-tri-methyl-ammonium-propane (DMTAP). The lipid phase diagram shows the well-known LR, Lβ′, and Pβ′ ripple phase with peritectic behavior at a low molar fraction of cationic lipid, χTAP < 0.12. Beyond χTAP ) 0.8 crystalline phases appear. A systematic variation in the hydrocarbon chain tilt in the prevailing Lβ′ phase is measured by wide-angle X-ray scattering. Most importantly, the Lβ′ phase shows strong nonideal mixing with an azeotropic point at about 1:1 molar stoichiometry. This finding is related to the reduced headgroup area for equimolar mixtures found in monolayer pressure-area isotherms. The intercalation of DNA in cationic lipid-DNA complexes affects the lipid-phase behavior 2-fold: (i) the chain-melting transition temperature shifts to higher temperatures and (ii) a demixing gap with coexistence of lipid vesicles and lipidDNA complexes arises at a low cationic fraction, χTAP < 0.25. In agreement with experiments we present a thermodynamic model that describes the shift of the melting transition temperatures by DNA-induced electrostatic screening of the cationic membrane.

Introduction Cationic liposomes have found widespread application in molecular cell biology as transfection agents. When a protocol, known as lipofection, is used, plasmid DNA is delivered into eucaryotic cells by means of lipid mixtures of usually zwitterionic and cationic lipids. Cationic liposomes aggregate with DNA and enhance the gene delivery due to electrostatic attraction to the cell surface.1 During the last few years cationic lipids have also been tested in clinical trials as possible candidates for gene transfer in human gene therapy. For this purpose many new cationic amphiphiles have been synthesized and screened for their efficacy (for reviews see refs 2-4). However, at present there is little understanding, and much less ability to predict, gene-transfer efficiencies on a molecular level. It became apparent that, besides cytotoxicity and cell-specific interactions, also nonspecific physico-chemical properties of the lipid-DNA constructs play an important part in gene delivery. This concerns the structure, stability, and surface properties of the gene delivery complexes. The aggregates that form when cationic liposomes and plasmid DNA are initially mixed are condensed lipid-DNA composite phases with unique liquid-crystalline structures. When small-angle X-ray scattering (SAXS) was used, lamellar and hexagonal mesomorphic phases * To whom correspondence should be addressed. Phone: (49) 89 2891 2539. Fax: (49) 89 2891 2469. E-mail: [email protected]. † Technische Universita ¨ t Mu¨nchen. ‡ EMBL. § Permanent address: Faculte ´ de Pharmacie, UMR8612, CNRS, 5 rue J.B. Clement, 92296 Chatenay-Malabry Cedex, France.

with alternating DNA and lipid layers were discovered.3,5-7 Recent theoretical work is able to model lipid-DNA complexes within the framework of well-known lipid-phase polymorphism.8,9 The present work is an attempt to contribute to the understanding of the thermotropic phase behavior of mixed phospholipid-cationic liposomes and their corresponding DNAcomplexed mesomorphic aggregates. Specifically, we ask the question, what are the corresponding lipid phases in condensed lipid-DNA aggregates and what can we learn from changes in the lipid-phase behavior upon complexation with DNA? To this end, we investigate cationic lipids with saturated alkyl chains. We choose DMPC/DMTAP as a model phospholipid-cationic lipid (PL/CL) system. From the large number of available bilayer-forming cationic amphiphiles, or cationic lipids, DMTAP may be considered as one of the most natural cationic derivatives of DMPC. As seen in the chemical structure shown in Figure 1. The zwitterionic phosphatidyl-choline is simply replaced by propane-trimethylamine. The latter is a fully dissociated cationic quaternary amine. DMPC/DMTAP mixtures form lamellar lipid-DNA complexes analogous to the previously studied unsaturated DOPC/ DOTAP-DNA complexes.5,10 In DMPC/DMTAP-DNA complexes a low-temperature lamellar lipid gel phase and a hightemperature fluid phase exist as evidenced by recent small- and wide-angle X-ray scattering (SAXS and WAXS) experiments.11 Here, we present, for the first time, the full thermotropic phase diagram of this cationic lipid-DNA complex. The study is,

10.1021/jp991596j CCC: $18.00 © 1999 American Chemical Society Published on Web 10/30/1999

Behavior of Lipid-DNA and Binary Lipid Mixtures

J. Phys. Chem. B, Vol. 103, No. 46, 1999 10301 characterize each phase by means of electron microscopy, SAXS, and WAXS. In the discussion section the thermodynamic features of the cationic lipid-phase diagrams are analyzed. In particular, we recall the regular solution theory of azeotropic lipid mixtures to explain the elevated transition temperatures around equimolarity, and focus on the shift in the transition temperature between the pure lipid phases and the DNA complexed phases. We present a modified theory based on the work by Tra¨uble and Eibl,16 which is found to describe the experimentally determined shifts quantitatively. Finally, an Appendix is added to describe in detail the derivation of the C phase by analysis of the powder line shapes chain tilt of the Lβ′ of the WAXS experiments. Materials and Methods

Figure 1. (a) Structures of di-myristoyl-phosphatidyl-choline (DMPC) showing a zwitterionic headgroup and di-myristoyl-tri-methyl-ammonium-propane (DMTAP) with a quaternary amine group. (b) Schematic picture of the self-assembled supramolecular arrangement C in DMPC/DMTAP-DNA complexes. Both the Lβ′ gel state and LCR fluid phase exhibit intercalated lamellar order.

however, limited to the quasi-binary-phase diagram of isoelectric complexes, where the amount of DNA is fixed by the mole fraction of cationic lipid. For simplicity the DNA-bound binary phases are indexed with an additional “C”, for “condensed” or “complexed”. In the literature few purely lipidic phospholipid/ cationic lipid mixtures have been studied. In a seminal paper, Silvius investigated the phase behavior of zwitterionic and anionic lipids mixed with cationic amphiphiles and found that DPPC/DHDAC exhibits strongly nonideal mixing with pronounced maxima in the solidus and liquidus lines around equimolar mole fractions.12 Similar maxima were also found in other binary cationic mixtures.13-15 Our findings will be consistent with these phase diagrams. Moreover, we show that the complexation with DNA leads to surprisingly small effects on the overall phase behavior. The effects are 2-fold: first, the main transition shifts to higher temperatures and second a demixing gap appears for small mole fractions of cationic lipid. We relate this phase behavior to structural investigations, in C particular, of the Lβ′ phase, that reveals a continuous variation in tilt. Eventually, an untilted LCβ phase is located at χTAP ) 0.25, remarkably close to the demixing gap. The article is structured as follows. First, we present the calorimetric studies of binary DMPC/DMTAP mixtures. From the signatures of the differential scanning calorimetry (DSC) scans we derive the binary-phase diagram of DMPC/DMTAP. In parallel, we present the calorimetric measurements of isoelectric DMPC/DMTAP-DNA complexes and the corresponding pseudo-binary-phase diagram of the latter. We then

Materials and Sample Preparation. Di-myristoyl-phosphatidyl-choline (DMPC) and di-myristoyl-tri-methyl-ammoniumpropane (DMTAP) were purchased from Avanti Polar Lipids Inc. (Alabaster, AL) in chloroform. The lipids were mixed in the required ratios, dried under nitrogen flow, and subsequently stored in vacuum for more than 12 h. Multilamellar vesicles (MLVs) were prepared at a concentration of 5.0 mg/mL using Millipore water and vortex mixing every 20 min for 2 h at 60 °C. The resulting suspension was diluted to 2.5 mg/mL lipid concentration and divided into two parts. One part was used as pure lipid sample. The second half was complexed with an isoelectric amount of calf thymus DNA, i.e., a charge-neutral ratio (1:1) of DNA-phosphate groups to DMTAP headgroup. All samples were stored at 4 °C for 2 weeks to ensure thermodynamic equilibrium. Calf thymus DNA sodium salt (Sigma Aldrich Chemie GmbH, Deisenhofen, Germany) was dissolved in Millipore water and the concentration determined by 260/280 UV absorption. SUVs (small unilamellar vesicles) of DMPC and DMTAP were prepared by solvent evaporation under nitrogen flow, vacuum drying, resuspension in Millipore water at 70 °C for 4 h and 10 min of sonication. For X-ray measurements the lipid vesicle solutions containing a concentration of ca. 150 mg/mL were filled into quartz capillaries (Hilgenberg, Malsfeld, Germany). The lipid-DNA complexes were prepared by adding a 50 mg/mL vesicle suspension onto the freeze-dried DNA inside the quartz capillaries. The samples were annealed several times between 10 and 60 °C and allowed to equilibrate for more than 7 days. Electron Microscopy. A small amount (1 µL) of the sample was squeezed between two Balzers gold holders (sandwich technique) and then quickly frozen in nitrogen-cooled propane at room temperature (freezing velocity, 1000-3000 °C/s) following standard protocols.17,18 The freeze-fracture procedure was carried out in a Balzers HVA BAF 400 T at a pressure of 10-6 mbar and a temperature of approximately -170 °C. The shadowing was done with electron beam evaporators; the one, Pt/C, at an angle of 45° and the other, C, horizontally to stabilize the metal film.19,20 The replica was floated on a water-acetone surface (100:5). No further cleaning was necessary. The replica was picked up with 400-mesh copper grids. Pictures were taken with a Philips CM 100 at 100 kV with Agfa Scientia electron microscopy films. Calorimetry. DSC measurements were performed using a MC-2 differential scanning calorimeter (MicroCal, Northampton, MA) at heating rates of 20 °C/h. Tests with slower scanning rates obtained transition temperatures in accordance of 0.1 °C and transition peaks with widths deviating less than 5%. All samples contained a lipid concentration of 2.5 mg/mL and DNA concentrations between 0 and 1.35 mg/mL. Measuring and

10302 J. Phys. Chem. B, Vol. 103, No. 46, 1999

Zantl et al.

reference cells were held under 3 bar using pressurized nitrogen. The solidus (respectively, liquidus) temperature was determined by the intersection of a tangent at the low- (respectively, high-) temperature inflection point and the base line. Tm was determined as the maximum of the cP curves. The transition enthalpies were determined by numerical integration of the base line-corrected DSC curves. The estimated error of the transition enthalpy for vesicle solutions is about 6%, taking a 0.2 °C temperature precision and a weighing failure of 5%. Problems occurred because of the size (≈1 mm) and stickiness of lipidDNA aggregates. An exact control of the total amount of complexes transferred inside the DSC cell proved to be impossible. The reproducibility of DSC data was scrutinized for selected molar ratios by scanning different preparations more than five times. X-ray Diffraction. SAXS and WAXS was carried out in the EMBL synchrotron-beamline X13. A detailed description of the setup used is given by Rapp et al.21 The resolution in the reciprocal space was better than 0.002 Å-1 for the SAXS and 0.0055 Å-1 for the WAXS as measured by the HWHM of the lamellar peaks of silver behenate at q ) 0.108 Å-1 and of parabromobenzoe acid at q ) 1.34 Å-1. The investigated q range for SAXS was 0.03-0.35 Å-1 and for WAXS 1.0-1.5 Å-1. The temperature was controlled within 1 °C by a peltier-water bath setup. Film Balance. A Langmuir-type film balance was used as described in ref 22. The lipid was applied from a 10 mg/mL chloroform stock solution. The maximal monolayer area was 900 cm2 with a maximal compression ratio of 3:1. Typical sweeping rates were about 100 µm/s. All scans were performed on a Millipore subphase with the temperature stabilized at 20 °C. Results Differential Scanning Calorimetry. Baseline-corrected DSC scans of DMPC/DMTAP MLVs and DMPC/DMTAP-DNA complexes are shown in Figure 2a. The data are stacked as a function of the increasing mole fraction of cationic lipid, χTAP. The bottom scan corresponds to pure DMPC lipid with the wellknown gel-to-liquid crystalline transition at Tm ) 24.5 °C. With increasing cationic mole fractions, this main transition shifts to higher transition temperatures and exhibits a high-temperature shoulder. At χTAP ) 0.45 the transition temperature reaches a maximum and decreases from there to about 20 °C. Samples with very high DMTAP content exhibit a pronounced hysteresis in the heat-cooling cycles and dependence on sample history and preparation. The hatched triangle in the lipid phase diagram in Figure 3a indicates the region where enthalpic transitions were detected by DSC. Furthermore the DSC scans exhibit a small enthalpic pretransition at low TAP concentrations. The pretransition peaks are enlarged in Figure 2b. The transition temperature increases with the cationic mole fraction and merges P ) 0.18. The transition with the main transition at χTAP enthalpies of the endothermic transitions are summarized in Table 1. Note that the values for the enthalpies of the lipidDNA complexes in italic letters are lower estimates because in these cases loss of material could not be avoided during the transfer into the calorimeter measuring cell. In the case of the cationic lipid-DNA complexes we find similar behavior of the main transition, which also increases with increasing χTAP. However, at low DMTAP concentrations (Figure 2b), in contrast to the pure binary lipid mixtures we observe a second transition at about 40 °C besides the pretransition and the main transition. The area of the high-

Figure 2. (a) Calorimetric heating scans of DMPC/DMTAP liposome suspensions (left) and DMPC/DMTAP/DNA complexes (right). The molar fraction of cationic lipid, χTAP, increases from bottom to top. All peaks are normalized to the same amplitude for clarity. Samples were prepared and equilibrated in Millipore water. (b) Magnified selection of calorimetric scans displaying the evolution of the pretransition at low DMTAP mole fraction. The purely lipidic mixtures (left) show peritectic behavior, while cationic lipid-DNA complexes (right) exhibit a demixing gap.

temperature transition peak increases at the expense of the area of the low-temperature peak. Both transitions do not significantly shift their position within this coexistence regime, χTAP < 0.32. The shape of the main transition peak at χTAP > 0.32 is, furthermore, strongly asymmetric. Beyond the composition χTAP ) 0.45 the transition temperature decreases monotonically. For mole fractions of 0.73 < χTAP < 1.0 we find a new lowtemperature transition with increasing enthalpy but constant transition temperature of about 20 °C. Construction of the Phase Diagrams. On the basis of the calorimetric data, we construct the phase diagrams as shown in Figure 3a,b. The DMPC/DMTAP mixtures are presented as purely binary systems given that the pressure is constant and that excess water can be neglected as a third component. In the case of the ternary DMPC/DMTAP/DNA complexes we present a two-dimensional cut through the truly three-dimensional phase diagram, which corresponds to stoichiometrically charged neutral complexes with a fixed DNA-to-cation ratio. The latter

Behavior of Lipid-DNA and Binary Lipid Mixtures

Figure 3. (a) Phase diagram of the DMPC/DMTAP system derived from calorimetry and SAXS. The highest transition temperature is reached at mole fraction cationic lipid, χmax TAP ) 0.45. Here, the coexistence region is of minimal width. At low DMTAP concentrations the phase diagram shows peritectic behavior of the adjacent Lβ′ and Pβ′ gel phases. At high DMTAP concentrations there are two distinguishable solid phases, S1 and S2. In the hedged region DSC scans were not reproducible. (b) In comparison, the pseudobinary thermotropic phase diagram DMPC/DMTAP mixtures complexed with an isoelectric amount of calf thymus DNA. The transition temperature maximum is max Tmax m ) 41.6 °C at about χTAP ) 0.42. At low χTAP values a demixing gap with coexisting vesicles and lipid-DNA complexes. The “stretched” peritectic behavior of the high chain melting transition in this region indicates the existence of DMTAP-depleted vesicles and DMTAPenriched lipid-DNA complexes. On the right side of the diagram simple C eutectic behavior of the complexed Lβ′ and one crystalline DMTAPc DNA S phase is observed.

TABLE 1: Summary of Transition Temperatures and Enthalpies for the Lipidic and Lipid-DNA Mixtures DMPC/DMTAP MLVs χTAP

Tm (°C)

Tpre (°C)

0.00 0.04 0.09 0.18 0.27 0.35 0.43 0.51 0.53 0.59 0.67 0.76 0.84 0.92 1.00

24.5 25.8 26.8 31.3 34.6 36.3 37.1 36.7 36.9 36.1 34.6 32.6 22.4 20.7 18.9

12.7 19.0 20.7 28.2 33.8

∆H (kcal/mol)

6.2 6.5 6.8 8.9 6.6 6.9 5.4 6.8 6.1 7.1 7.3 16.6 6.9 17.0 ca. 5 18.2 ca. 3 17.6

DMPC/DMTAP-DNA complexes Tm (°C)

Tpre (°C)

∆H (kcal/mol)

24.45 32.25 37.35 39.35 40.55 41.15 41.85 40.25 39.25 39.35 36.65 34.45 29.05 25.46

12.65 19.85 25.85 27.45

6.2 >5.9 >4.1/0.7 >1.9/2.4 >0.5/2.5 >1.9 >1.6 >2.0 >5.0 >1.7 >3.9 >0.2/3.2 >2.2/2.4 >3.1/0.1 >8.3

23.25 23.55 20.95. 22.45

can be called a pseudo-binary lipid-phase diagram of DNAcomplexed cationic lipid mixtures. For this reason we simply

J. Phys. Chem. B, Vol. 103, No. 46, 1999 10303 furnish the lipid-phases denotation with an additional index “C” for “complexed” or “condensed”. Solidus and liquidus lines for each phase diagram were constructed from the intersections of lateral peak tangents and the base line of the DSC scans respectively. To illustrate true broadening of the transition in the coexistence regions, the solidus and liquidus lines were corrected by the width of pure DMPC as suggested by Mabrey et al.23 The dominant phases in the binary DMPC/DMTAP phase diagram are the Lβ′ gel phase and the liquid-crystalline LR phase. Both phases are separated by narrow coexistence regions (drawn in gray) in the range 0.25 < χTAP < 0.75. Most striking is the fact that the chain-melting transition temperature reaches a pronounced maximum at about χmax TAP ) 0.45. At low cationic lipid mole fractions we find a Pβ′ ripple phase. The ripple phase P ) exhibits peritectic behavior with a peritectic point at χTAP 0.18. At this point the LR, Lβ′, and additionally the ripple phase (Pβ′) coexist. The pretransition temperature, Tpre, given by the maxima of the low-temperature peaks in Figure 2b increases from Tpre ) 12.5 °C of DMPC to TP ) 30 °C. At χTAP > 0.75 two different crystalline phases, S1 and S2, are found from two enthalpic transitions in the DSC spectra, which we will corroborate later in this article by two distinguishable WAXS reflexes. X-ray patterns will also reveal the coexistence of two Lβ′ WAXS peaks for samples containing more than 91% DMPC. Consequently, the low-temperature enthalpic transition detected by DSC is interpreted as the transition from S1 to S2, the middletemperature peak as the transition from S2 to LR, and the hightemperature peak as the transition from Lβ to LR. The lateralphase separation of the solid phases and the miscibility of the high-temperature phases correspond to eutectic behavior. In Figure 3b the phase diagram of the cationic lipid-DNA mixtures is shown. Again, the diagram is dominated by the C and LCR phases. The most remarkable differcorresponding Lβ′ ence is a demixing gap at low mole fractions of cationic lipid where excess lipid mixtures coexist with CL-DNA. DSC detects the vesicle pretransition and main transition and at higher C -LCR transition of the complexes. The temperatures the Lβ′ phase behavior of the lipid vesicles is very similar to that of pure lipid mixtures, but shifted toward that of lower TAP concentration inside the lipid vesicles. The majority of the cationic lipids are bound in CL-DNA complexes because of electrostatic effects. The high χTAP values in the complexes also lead to a relatively high transition temperature. With increasing χTAP the area of the transition peak increases at the expense of the pure vesicle peak because of the growing amount of CLDNA. We point out that the exact composition in the left-hand corner of the phase diagram (Figure 3b) cannot be derived by a simple lever rule as in the binary case. The coexistence of lipid-DNA complexes and binary excess liposomes is possible because of the additional degree of freedom in the ternary system. Samples containing more than 20% of DMTAP form only lipid-DNA complexes of one kind. In this case the lipidphase behavior in the DNA-complexed form is pseudo-binary. The first DSC-detectable transition of CL-DNA (χTAP ) 0.09) shows already a relatively high Tm of 39 °C that is slowly ) increasing to the maximum transition temperature of Tmax m max ) 0.4. 41.60 °C at the corresponding TAP mole fraction TTAP Tmax m of CL-DNA is consequently 5 °C higher than that for the pure lipid vesicles shown in Figure 3a. The thermotropic behavior of the CL-DNA at χTAP > 0.75 shows behavior similar to the lipid vesicles. The chain-melting temperature, Tm, of the complexes is close to that of binary lipid vesicles of corresponding composition. However, thermodynamically and structurally, only one solid phase, S, is observed with the onset

10304 J. Phys. Chem. B, Vol. 103, No. 46, 1999

Zantl et al.

Figure 4. Freeze fracture electron micrographs of binary D1MPC/DMTAP vesicles. (a) The ripple phase of pure DMPC vesicles compared to (b) vesicles containing 8% DMTAP (the insets are magnified 5×). The ripple repeat distance is about 20 nm in both cases. (c) Vesicles in the Lβ′ gel phase. (d) Vesicles quenched from the Lβ′-LR transition temperature, Tm ) 36.5 °C. (e) Pure DMTAP vesicles in the solid state and (f) in the liquid crystalline LR state. C of the Lβ′ -S coexistence region being shifted to higher χTAP values. Interestingly, no hysteresis due to sample history is found anymore for complexes at low temperatures. Freeze-Fracture Electron Microscopy. The different phases of binary lipid vesicles are visualized by freeze-fracture electron micrographs. The rippled surface of a multilamellar vesicle in the ripple phase at 21.6 °C is clearly seen for pure DMPC

vesicles (Figure 4a) and vesicles containing 8% DMTAP (Figure 4b), χTAP ) 0.08. The ripple wavelengths, λ ) 20.7 and 20.4 nm, respectively, are similar. At χTAP ) 0.48 no ripples were found at any temperature on these vesicles. In the gel Lβ′ phase (T ) 34 °C) the vesicle surface is totally smooth (Figure 4c). In the middle of the Lβ′-LR phase coexistence region at Tm ) 36.5 °C the vesicle surface exhibits a surface relief which might

Behavior of Lipid-DNA and Binary Lipid Mixtures

J. Phys. Chem. B, Vol. 103, No. 46, 1999 10305

Figure 5. Freeze fracture electron micrograph of isoelectric, lamellar cationic lipid-DNA complexes containing a cationic mole fraction, χTAP ) 0.5, quenched from 20 °C. The extended layered structure is in agreement with the structure derived from X-ray scattering shown in Figure 1b. The DNA strands cannot be resolved by freeze fracture electron microscopy.

correspond to membrane deformations caused by the coexistence of Lβ′ and LR domains (Figure 4d).24 Pure DMTAP vesicles quenched from 26 °C show a characteristic surface structure as well as smooth regions on the surface (Figure 4e). Possibly C and the solid phase S is lateral phase separation of Lβ′ observed. In Figure 4f, finally, it is shown that DMTAP vesicles have smooth surfaces at 36 °C corresponding to the liquidcrystalline phase LR. The fluid DMTAP vesicles with diameters smaller than 2 µm are somewhat smaller than vesicles of DMPC-DMTAP mixtures, where diameters up to 30 µm are found. In the case of lipid-DNA complexes electron microscopy reveals the existence of condensed multilamellar aggregates. As an example a freeze-fracture micrograph of aggregates with equimolar lipid composition is shown in Figure 5. The sample was quenched from 20 °C to room temperature, corresponding C to the Lβ′ state of the membranes. More than a hundred membranes in parallel order can be counted in one stack. However, a parallel arrangement of the DNA rods, as detected by SAXS, is not observable. The expected repeat distance of about 4 nm is indeed too small to be resolved by freeze fracture. Wide-Angle X-ray Scattering of the Gel Phases. Figure 6 shows wide-angle X-ray diffraction data of purely lipidic multilamellar vesicles (MLVs) and lipid-DNA complexes. The well-known wide-angle reflections of the LR, Pβ′, and Lβ′ phase of pure DMPC is given as a reference on the left side of the panel. The DMPC Lβ′ peak shows a sharp, resolution-limited, peak superimposed with a broadened peak at higher q values.25 The latter arises from powder averaging of the (1,1) reflection of a distorted hexagonal lattice with chains tilted toward the next neighbors. We compare the line shape of the WAXS peak of the binary DMPC/DMTAP mixtures. The latter exhibit a broadened reflection centered around 1.48 Å-1 at low temperatures. The broadening is characteristic for a Lβ′ phase with hydrocarbon chain tilt between the nearest neighbors.26,27 In particular we can exclude the possibility of a rippled Pβ′

Figure 6. (a) Wide-angle X-ray scattering (WAXS) of DMPC multilamellar vesicles (MLVs) in the Lβ′, Pβ′, and LR phase. (b) WAXS signal of the low-temperature Lβ′ phase for binary DMPC/DMTAP mixtures with varying mole fractions of cationic lipid. (c) DMTAP shows two different solid phases, S1 and S2, at low temperatures and a fluid LR phase at high temperatures. (d) In comparison the WAXS signal of lipid-DNA complexes for different mole fractions of DMTAP and (e) DMTAP/DNA complexes is shown. Note that the shape of the WAXS reflexes are indicative of different tilts of the hydrocarbon chains with respect to the bilayer normal as described in detail in the Appendix.

10306 J. Phys. Chem. B, Vol. 103, No. 46, 1999

Zantl et al.

Figure 7. Small-angle X-ray scattering at the chain-melting transition. C The coexistence of Lβ′ and LCβ is seen by two sets of integral Bragg reflections according to two lamellar spacings. (inset) Fractional conversion shown by the fraction gel phase, fraction liquid crystalline phase as a function of temperature calculated from the normalized C second Bragg reflection (left inset) and from the Lβ′ -WAXS-peak intensity (right inset).

Figure 8. (a) Pressure-area diagram of monolayers of pure DMPC, DMTAP, and an equimolar mixture of DMPC/DMTAP at the airwater interface. The inset shows the average headgroup area as a function of cationic mole fractions for two constant pressures corresponding to the crystalline and liquid expanded phases, respectively. (b) Schematic drawing of the headgroup arrangement in the fluid and frozen state at 1:1 stoichiometry. In the gel state the quaternary amine (TAP) and phosphate group get into closer proximity, which might be responsible for the overall attractive headgroup interaction.

configuration because the electron microscopy study clearly showed a flat membrane. In the Appendix we describe in detail how the wide-angle reflections of Lβ′ phases characterize different structures, depending on the tilt and tilt direction of the chains with respect to the in-plane orientation. In our data we see a slight tendency of increasing tilt angle with increasing mole fraction, χTAP, cationic lipid. The same binary lipid mixtures complexed with DNA exhibit similar wide-angle diffraction. The line shape is in agreement C with a tilted Lβ′ phase, where the tilt direction points again between next neighbors. However, the tilt angle increases with increasing molar fraction, χTAP, as can be seen from the apparent broadening of the width of the reflection. In fact, the chain tilt seems to almost vanish close to χTAP ) 0.25, where the width is as narrow as expected for an untilted Lβ phase. As a general observation, we find that the tilt of all binary lipid mixtures decreases with the addition of DNA. On the right-hand side of Figure 6 two solid phases, S1 and S2, and the liquid-crystalline LR phase of pure DMTAP are shown. In the case of pure DMTAP the complexation with DNA leads to a new solid phase, S, that is slightly different from the S1 and S2. Also, WAXS showed a strong thermotropic polymorphism of DMPC-DMTAP vesicles as well as for the lipidDNA complexes at high χTAP, i.e., the hatched region in the phase diagrams shown in Figure 3. Fast cooling of the fluid phases revealed retarded crystallization and consequently supercooled LR and metastable Lβ phases. This explains the problem of reproducibility of the DSC-derived phase transition temperatures in this region of the phase diagram. Small-Angle X-ray Scattering at the Chain-Melting Transition. Figure 7 shows the SAXS signal of lipid-DNA

complexes at the chain-melting transition. The top scan corresponds to the liquid-crystalline LCR state and the bottom scan to C gel state as sketched schematically in Figure 1b. CLthe Lβ′ DNA in the LCR phase consists of lipid bilayers in the liquidcrystalline state with intercalated parallel-ordered DNA double C helices.5 In the case of the Lβ′ phase the lipid membrane is gellike with a charge-density-dependent chain tilt angle.11 Both scans show a set of equally spaced Bragg reflections corresponding to the lamellar stacking of the lipid membrane. In both scans an additional broad peak is seen, which corresponds to the lattice constant of the intercalated DNA lattice. The DNA C gel lattice constant of the LCR phase is larger than that in the Lβ′ state, while the interlamellar distance is larger in the latter. The coexistence of two phases is clearly seen in the temperature interval around Tm. While the position of the lamellar Bragg reflections remains constant, the amplitudes increase and decrease, respectively. The melting transition was found to be fully reversible in all cases. SAXS may be employed to determine the relative fractions C phases and allow one to determine the solidus of the LCR and Lβ′ and liquidus temperatures. The normalized intensities of the C second Bragg peaks (wide-angle Lβ′ peaks) are plotted versus temperature in the left (right) inset of Figure 7. Headgroup Area Studied by Film Balance. In an attempt to understand the TAP-PC headgroup interaction, we examined a series of pressure-area isotherms of DMPC/DMTAP monolayers using a standard Langmuir film balance. Figure 8a shows the pressure-area diagrams of pure DMPC, pure DMTAP, and an equimolar mixture of both lipids. We plotted the headgroup area, aH, at two constant pressures corresponding to the liquid-

Behavior of Lipid-DNA and Binary Lipid Mixtures

J. Phys. Chem. B, Vol. 103, No. 46, 1999 10307

expanded and -crystalline phase (Figure 8 inset) as a function of the cationic mole fraction. In the liquid-expanded phase, the area decreases monotonically from the larger PC headgroup size down to the TAP headgroup. In contrast, in the ordered phase the headgroup shows a minimum at about the equimolar ratio. Discussion Nonideal Mixing of Cationic-Zwitterionic Lipid Mixtures. The most striking feature in the binary-phase diagram of DMPC/DMTAP (Figure 3a) is the maximum in the gel-toliquid crystalline transition temperature at a molar ratio of max ) 0.45. Similar nonideal mixing behavior was previously TTAP reported for related cationic amphiphiles.12,13 This phase behavior of cationic-zwitterionic alloys is particularily distinct from anionic-zwitterionic systems such as DMPC/DMPS or DMPC/DMPG, which show almost ideal mixing.28 A thermodynamic description of nonideal mixing is given qualitatively by simple regular solution theory. For a regular solution the excess free energy is written

gE ) ΩXAXB

(1)

where Ω is a parameter independent of temperature and composition and XA and XB the molar fractions of components A and B. To obtain phase diagrams like Figure 3a we have to assume an attractive intermolecular interaction, Ω, which is stronger in the gel state compared to that in the fluid state. As suggested by Silvius, the elevation of the main phase transition in these mixtures is largely due to electrostatic- rather than hydrogen-bonding interactions.12 Furthermore, the maximum around equimolar composition indicates the formation of interacting units with 1:1 stoichiometry. The latter are often called complexes in the literature. We avoid the term complex, which here is reserved for lipid-DNA aggregates, but ask the question, if stoichiometric units do exist in the lipidic binary system. In alloy systems maxima in binary-phase diagrams are nearly always associated with the existence of an intermetallic phase.29 However, our electron microscopy studies as well as the SAX and WAX scattering data provide no indication of a distinct stoichiometric-phase and macroscopic-phase separation. Nonetheless, it is most likely that stoichiometric units, such as PC-TAP pairs, do form. These congruently melt at the gelto-liquid crystalline transition. We found that the headgroup areas of the binary mixtures at the air-water interface exhibit a minimum at a 1:1 PC-TAP ratio (Figure 8a). A molecular interpretation is given in Figure 8b. The small headgroup of the cationic amphiphiles are able to add onto the phosphate group and hence the PC-TAP headgroup interaction may be attractive. Alternating PC and TAP groups are able to pack tightly and will lead to a reduction of the average headgroup area. In fact, it was shown by NMR measurements that cationic amphiphiles can change the orientation of the choline dipole by more than 30° toward the water phase.15 The rotation of the N+ end of the choline group further out into the bulk water will reduce the electrostatic self-energy of this group. TAPPC units can be thought to exhibit a chain-packing density resembling that of phosphatidyl-ethanolamines. Recent Monte Carlo simulations of TAP-PC headgroup arrangements provide evidence for the existence of stoichiometric domains that can be imagined as clusters with increased TAP-PC correlations mixed into an otherwise disordered PC-TAP gel phase.30 Our strongest experimental arguments for the existence of such stoichiometric units are, firstly, the narrowing of the transition width compared to the pure DMPC and, secondly, the decrease

TAP < 1/2. (aPC of the headgroup area a(1:1) H H + aH ) in the crystalline phase of monolayers. Shift in the Transition Temperature for Lipid-DNA Complexes. Condensation of cationic lipid vesicles and DNA results in a substantial increase of the lipid chain-melting transition temperature. The first quantitative study of the melting transition in charged binary lipid membranes was carried out by Tra¨uble and Eibl.16 In general, charging the membrane lowers the lipid chain-melting transition temperature. The decrease in transition temperature, ∆Tm, with increasing surface charge arises from an additional electrostatic free energy change, ∆Gel, due to the change in surface area associated with the gel-toliquid crystalline-phase transition. The lateral electrostatic pressure, Πel, favors an increase in the lipid headgroup area, A. The transition temperature of a charged versus neutral lipid membrane shifts by

∆Tm )

∆Gel Πel∆A ) ∆S* ∆S*

(2)

where ∆S* denotes the transition entropy of the chain melting. Assuming full dissociation of the cationic lipid and large surface potential eΨ . kT, the electrostatic pressure can be calculated using classical Gouy-Chapman theory:16,31

∆Tm )

2RT ∆A - (200RTc)1/2(40RT/e∆S*)∆A ∆S* A

(3)

The first term is the maximum electrostatic shift due to the introduction of surface charge. The second term comprises the entropic contribution of the diffuse layer of counterions. The latter screens the electrostatic effect in the presence of salt. For the lipids investigated here, we may insert ∆A/A ) 0.19 as measured from the repeat distance of the intercalated DNA lattice DMPC/DMTAP-DNA11 and ∆S* ) 2 × 1022 kT/mol °C as known for pure DMPC.31 Equation 3 enables us to predict the temperature shift in the presence of salt. For example, in the case of equimolar mixtures of DMPC/DMTAP we measured ∆Tm ) 2.6 °C for 10 mM NaCl and ∆Tm ) 4.3 °C for 100 mM NaCl, in good agreement with eq 3 (see Figure 9a). However, if we neutralize cationic liposomes by an isoelectric quantity of DNA, one might expect that the total shift in the transition temperature is simply given by the first term of eq 3, yielding ∆Tm ) 10.8 °C. This means that we should find a transition temperature Tmax m + ∆Tm ) 47.8 °C, which is higher than that of the experiment shown in Figure 9a. To correctly describe the shift in the transition temperature of DMPC/DMTAP in the presence of intercalated DNA, we will proceed analogously to Tra¨uble and Eibl. In brief, we will show that the adsorbed DNA acts like a condensed twodimensional counterion lattice, which contributes to the change in free energy by its electrostatic pressure. We take advantage of the fact that the structure and molecular arrangment of the lipid/DNA complexes are known from previous X-ray studies.5,11 DNA strands are intercalated between lamellar galleries of lipid bilayers as shown in Figure 1b. In isoelectric complexes the spacing between adjacent DNA strands, d, is fixed by the charge density of the membrane. An expansion around the spacing, d1/2, for χTAP ) 0.5 yields an approximately reciprocal dependence on the mole fraction of the cationic lipid:11,32

d≈

d1/2 2χTAP

(4)

10308 J. Phys. Chem. B, Vol. 103, No. 46, 1999

Zantl et al.

Figure 9. (a) Comparison of the DSC main transition peaks of an equimolar mixture of DMPC/DMTAP with and without DNA. A similar shift to higher transition temperature as induced by DNA is observed by the addition of salt (- - -). (b) The shift in the chain-melting transition temperature of DMPC/DMTAP mixtures due to DNA condensation. The straight line in the middle region indicates the theoretical prediction according to eq 8.

In addition, the DNA lattice exhibits a lateral compressibility modulus, BDNA el , which is determined by the electrostatic repulsion between the strands:33,34

BDNA ) el

λ2 6πd

(5)

Here,  ) 0r denotes the dielectric constant of water and λ ) e/1.7 Å, the line charge density of phosphate groups along the DNA. The two-dimensional compressional modulus has units of energy/area and was measured by analysis of the thermal diffuse scattering of the DNA X-ray reflection.33,34 At the chain-melting transition the lipid membrane expands laterally and the DNA lattice spacing increases with it. The gain in electrostatic free energy of the DNA lattice is approximated by ∆F/A ) ∆[1/2B(d/d0)2] ≈ B(∆d/d). Hence, the shift in transition temperature is obtained using the first term in eqs 3 and 5, accounting for the fact that the intercalated DNA lattice (i) fully compensates the surface charge and (ii) has a finite lateral electrostatic compressibility.

∆Tm )

A DNA∆d 2RT ∆A B ∆S* A ∆S* el d

(6)

Equation 6 compares well with experiment as shown in Figure 9b. Because of the dependence of BDNA on χTAP via eqs 4 and el

5, the temperature shift, ∆Tm ) TA - TBχTAP, is a linear function of χTAP. For d1/2 ) 36 we obtain B ) 2 × 10-2 kT/Å2 from eq 5 and yield TA ) 10.8 °C and TB ) 7.6 °C, assuming ∆A/A ) ∆d/d ) 0.19 and ∆S* ) 2 × 1022 kT/mol °C as used before. The experimental results suggest TA ) 11 °C and TB ) 13 °C in the middle χTAP regime. This experimental result is matched by the theory with even better precision, if we take the experimentally determined value for the compressional modulus, B ) 2.8 × 10-2 kT/Å2 at d1/2 ) 36.33,34 As seen in Figures 9b and 3b the theory will not be applicable at a low cationic mole fraction, where we found a demixing gap, as well as at the far right side of the phase diagram, where the existence of ill-defined dehydrated solid phases impairs the comparison. The agreement of our simple theory is remarkable considering the apparent complexity of the system. However, the lipidDNA complexes are special in that all counterions are released from the complex and the DNA effectively plays the role of a two-dimensional layer of rodlike counterions. Secondly, the DNA strands are separated from the bilayer by a thin hydration layer, approximately 2 Å according to the X-ray results, and are well-separated from each other because of the constraint to preserve local charge neutrality. Hence, no short-range forces become effective. Finally, the effect of monovalent salt on our particular binary lipid system is well-described by the GouyChapman model, which is a prerequisite that the electrostatic theory can be extended to the case of an adsorbed DNA lattice. In this context it is also interessting to note that DMPE exhibits a chain-melting temperature, Tm ) 48 °C, that is comparable to a completely neutralized equimolar PC-TAP mixture, as described by the first term in eq 3. This indicates that the state of hydration and chain packing of PC-TAP units is similar to that of DMPE, which is very intuitive, when looking at the molecular packing model given in Figure 8b. The total enthalpy change of the melting transition due to DNA complexation could not be measured accurately because of problems with material transfer. However, as indicated in Table 1 and Figure 9a, the enthalpy change decreases from 6 kcal/mol to about 3-5 kcal/ mol. Similar enthalpy changes are known for negatively charged membranes with adsorbed proteins like cytochrome c.35 Note that here the change of the enthalpy change is discussed. A more important quantity in this context might be the enthalpy of lipid-DNA complex formation. The latter has been measured, for example, for CTAB and DNA using titration calorimetry.36 Finally, we like to comment on the distinct shape of the DSC peak of DNA-lipid complexes compared to the sharp peak of equimolar lipidic mixtures (compare Figure 9a). It is reasonable to assume that DNA-induced demixing is responsible for the observed broadening. Lateral redistribution of cationic lipids in the presence of DNA was shown to be expected by recent theoretical studies.8,9 Conclusion The phase behavior of the zwitterionic-cationic mixture DMPC/DMTAP was studied with focus on the thermodynamic effect of DNA complexation. We found strongly nonideal mixing expressed in an elevated azeotropic point at almost equimolar mole fractions. The formation of stoichiometric PCTAP units is assumed on the basis of the observed reduction in the PC-TAP headgroup area in lipid monolayers as well as the higher cooperativity of the main-phase transition for equimolar mixtures. The complexation of DMPC/DMTAP liposomes with DNA results in the formation of condensed lipid-DNA composite aggregates. These aggregates have a lamellar liquid-crystalline structure, which retains in the gel and

Behavior of Lipid-DNA and Binary Lipid Mixtures

J. Phys. Chem. B, Vol. 103, No. 46, 1999 10309

Figure 10. Schematic presentation of the WAXS powder signal arising from hexagonal-packed alkyl chains. From top to bottom a real space picture of tilted rod lattices, the representation in Fourier space and the powder-averaged signal is shown. The WAXS peak exhibits characteristic line shapes depending on the tilt direction: (a) tilt toward next neighbors, LβI, (b) tilt between next neighbors, LβF, and (c) no tilt, Lβ.

liquid-crystalline phase. WAXS gave evidence for variations in the lipid chain tilt throughout the Lβ′ phase. The presence of DNA alters the lipidic thermotropic phase behavior 2-fold: we found (i) a systematic increase in the main transition temperature and (ii) a demixing gap at low TAP concentration. The shift in the transition temperature is in quantitative agreement with a modified Gouy-Chapman theory that treats the DNA as an adsorbed rod lattice. The results presented here should be more generally valid for other cationic lipids. Similar behavior could be expected, e.g., for DODAB, DOTMA, and double-chain amphiphiles with small cationic headgroups. For transfection applications saturated lipids are usually avoided because of the presence of lipid-phase transitions. However, in view of the present study, saturated cationic lipids might show an even broader spectrum of properties and need not be ruled out a priori. For example, isothermally induced transitions such as pH- or ionic-induced transitions could act synergetically in the course of gene delivery. Acknowledgment. We thank E. Sackmann for support and critical comments on the phase diagrams. We enjoyed helpful discussions with D. Pink, A. Tardieu, P. Garidel, and Th. Heimburg and furthermore benefited from a constructive referee report. F.A. gratefully acknowledges a European TMR postdoctoral scholarship. This work was supported by BMBF Grant 03-SA5TU1-0. Appendix: Tilt Direction Analysis by Wide-Angle X-ray Scattering In this Appendix we will shortly recall the reciprocal lattice of Lβ′ phases and discuss the analysis of the tilt direction in the Lβ′ phases from powder-averaged line shapes as proposed by Tardieu and Luzzatti26 and Levelut.27 In the chain frozen Lβ′ phase the hydrocarbon chains pack in an all-trans configuration in hexagonal, or rather distorted hexagonal, local order. These phases may have different tilt as

experimentally observed in X-ray scattering experiments by Smith et al. using oriented freely suspended DMPC films.37 Following the nomenclature of these authors, we denote the Lβ phases as hexatic phases with (a) tilt toward the nearest neighbor LβI or (b) tilt between next neighbors LβF. Intermediate directions are known as LβL phases.37,38 In the Lβ gel phase the chain director is normal to the bilayer as shown in Figure 10c. The electron density of the chain lattice can be described by a product of a hexagonal lattice of infinite rods and the form factor of the layer, i.e., the electronic density perpendicular to the bilayer. Thus, in reciprocal space the lattice is given by the convolution product of a hexagonal two-dimensional point lattice perpendicular to the rods and the rodlike form factor resulting from the Fourier transform of the bilayer. The three cases are shown in Figure 10. The shape of the rodlike form factor is related to the bilayers thickness. Here, it is approximated by a Gaussian profile. The WAXS line shape depends on the position of the rod center vs the equatorial plane (qx, qy) (Figure 10): (a) If the center is located in the equatorial plane, the powder averaging yields an asymmetric resolution-limited peak with a foot at wide angle, indicating that the rod is tangential to the q sphere. (b) If the center is out-of-plane, the powder averaging yields a symmetric broadened peak and its width is increasing with qz. In the Lβ gel phase (c), all the centers are located in-plane and the WAXS reflection is sharp. In the first Lβ′ gel phase, two centers are in-plane and four centers are out-of-plane so that the WAXS is the sum of a sharp peak and a broadened peak. In the second Lβ′ gel phase all the peaks are out-of-plane at qz and 2qz and the peak is the superposition of two broadened peaks. The line shapes of the WAXS peaks are clearly distinguishable, depending on the direction of the tilt. References and Notes (1) Felgner, P. L.; Gadek, T. R.; Holm, M.; Roman, R.; Chan, H. W.; Wenz, M.; Northrop, J. P.; Ringold, G. M.; Danielson, M. Lipofection: A highly efficient, lipid-mediated DNA-transfection procedure. Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 7413.

10310 J. Phys. Chem. B, Vol. 103, No. 46, 1999 (2) Behr, J.-B. Gene transfer with synthetic cationic amphiphiles: Prospects for gene therapy. Bioconjugate Chem. 1994, 5, 382. (3) Lasic, D. D. Liposomes in Gene DeliVery; CRC Press: Boca Raton, FL, 1997. (4) Miller, A. D. Cationic liposomes for gene therapy. Angew. Chem., Int. Ed. 1998, 37, 1769. (5) Ra¨dler, J. O.; Koltover, I.; Salditt, T.; Safinya, C. R. Structure of DNA-cationic liposome complexes: DNA intercalation in multi-lamellar membranes in distinct interhelical packing regimes. Science 1997, 275, 810. (6) Koltover, I.; Ra¨dler, J. O.; Salditt, T.; Safinya, C. R. The inverted hexagonal phase of DNA-cationic liposome complexes: Structure to gene release mechanism correlations. Science 1998, 281, 78. (7) Boukhnikachvili, T.; Aguerre-Chariol, O.; Airiau, M.; Lesieur, S.; Ollivon, M.; Vacus, J. Structure of in-serum transfecting DNA cationic lipid complexes. FEBS Lett. 1997, 409, 188. (8) Harries, D.; May, S.; Gelbart, W. M.; Ben-Shaul, A. Structure, stability and thermodynamics of lamellar DNA-lipid complexes. Biophys. J. 1998, 75, 159. (9) Bruinsma, R.; Mashl, J. Long-range electrostatic interaction in DNA-cationic lipid complexes. Europhys. Lett. 1998, 41, 165. (10) Artzner, F.; Zantl, R.; Rapp, G.; Ra¨dler, J. O. Observation of a rectangular columnar phase in condensed lamellar cationic lipid-DNA complexes. Phys. ReV. Lett. 1998, 81, 5015. (11) Zantl, R.; Artzner, F.; Rapp, G.; Ra¨dler, J. O. Thermotropic structural changes of saturated-cationic-lipid-DNA complexes. Europhys. Lett. 1998, 45, 90. (12) Silvius, J. R. Anomalous mixing of zwitterionic and anionic phospholipids with double-chain cationic amphiphiles in lipid bilayers. Biochim. Biophys. Acta 1991, 1070, 51. (13) Linseisen, F. M.; Bayerl, S.; Bayerl, T. 2H-NMR and DSC study of DPPC-DODAB mixtures. Chem. Phys. Lipids 1995, 83, 9. (14) Gaub, H.; Bu¨schel, R.; Ringsdorf, H.; Sackmann, E. Phase transitions, lateral phase separation and microstructure of model membranes composed of a polymerizable two-chain lipid and dimyristoylphosphatidylcholine. Chem. Phys. Lip. 1985, 37, 19. (15) Scherer, P. G.; Seelig, J. Electric charge effects on phospholipid headgroups. phosphatidylcholine in mixtures with cationic and anionic amphiphiles. Biochemistry 1989, 28, 7720. (16) Tra¨uble, H.; Eibl, H. Electrostatic effects on lipid phase transitions: membrane structure and ionic environment. Proc. Natl. Acad. Sci. U.S.A. 1974, 71, 214. (17) Rebhun, L. Freeze-substitution and freeze-drying. Princ. Tech. Electron Microsc. 1972, 28, 3. (18) Gulik-Krzywicki, T.; Costello, M. J. The use of low temperature X-ray diffraction to evaluate freezing methods in freeze-fracture electron microscopy. J. Microsc. 1978, 112, 103. (19) Zingsheim, H. P.; Abermann, R.; Bachmann, L. Apparatus for ultrashadowing of freeze-etched electron microscopic specimens. J. Phys. E: Sci. Instrum. 1970, 3, 39. (20) Zingsheim, H. P.; Abermann, R.; Bachmann, L. Shadow casting and heat damage. Proc. 7th Int. Cong. Electron Microsc. 1970, 71, 411.

Zantl et al. (21) Rapp, G.; Gabriel, A.; Dosire, M. H. J.; Koch, M. A dual detector single readout system for simultaneous small-(SAXS) and wide-angle X-ray (WAXS) scattering. Nuclear Instruments and Methods in Physics Research A, 1995, 357, 178. (22) Albrecht, O.; Gruler, H.; Sackmann, E. Pressure-composition phase diagram of cholesterol/lecithin, cholesterol/phosphhatic acid, and lecithin/ phosphatic acid mixed monolayers: A langmuir film balance study. J. Colloid Interface Sci. 1980, 81, 5015. (23) Mabrey, S.; Sturtevant, J. M. Investigation of phase transitions of lipids and lipid mixtures by sensitivity differential scanning calorimetry. Proc. Natl. Acad. Sci. U.S.A. 1976, 73, 1. (24) Lipowsky, R. Handbook of Biological Physics; North Holland Publishing Corp.: Amsterdam, 1995. (25) Sun, J.-W.; Suter, R. M.; Knewtson, M. A.; Worthington, C. R.; Tristram-Nagle, S.; Zhang, R.; Nagle, J. F. Order and disorder in fully hydrated unoriented bilayers of gel phase dipalmitoylphosphatidylcholine. Phys. ReV. E 1994, 49, 4665. (26) Tardieu, A.; Luzzati, V. Structure and polymorphism of the hydrocarbon chains of lipids: A study of lecithin-water phases. J. Mol. Biol. 1972, 75, 711. (27) Craievich, A. F.; Levelut, A. M.; Lambert, M.; Albon, N. Polymorphisme du dipalmitoyl 1,2 glyceride. J. Phys. 1978, 39, 377. (28) Hauser, H. Effect of inorganic cations on phase transitions. Chem. Phys. Lipids 1991, 57, 309. (29) Cahn, R. W.; Haasen, P. Physical Metallurgy, 3rd ed.; NorthHolland Physics Publishing: Amsterdam, 1983. (30) Tate, A.; Quinn, B.; Merkel, R.; Pink, D. To be published, 1999. (31) Ja¨hnig, F. Electrostatic free energy and shift of the phase transition for charged lipid membranes. Biophys. Chem. 1976, 4, 309. (32) Ra¨dler, J. O.; Koltover, I.; Jamieson, A.; Salditt, T.; Safinya, C. R. Structure and interfacial aspects of self-assembled cationic lipid-DNA gene carrier complexes. Langmuir 1998, 14, 4272. (33) Salditt, T.; Koltover, I.; Ra¨dler, J. O.; Safinya, C. R. Twodimensional smectic ordering of linear DNA chains in self-assembled DNAcationic liposome mixtures. Phys. ReV. Lett. 1997, 79, 2582. (34) Salditt, T.; Koltover, I.; Ra¨dler, J. O.; Safinya, C. R. Self-assembled DNA-cationic-lipid complexes: Two-dimensional smectic ordering, correlations, and interactions. Phys. ReV. E 1998, 58, 889. (35) Heimburg, R. L.; Biltonen, T. Thermotropic behavior of dimyristoylphosphatidylglycerol and its interaction with cytochrome c. Biochemistry 1994, 33, 9477. (36) Spink, C. H.; Chaires, J. B. Thermodynamics of the binding of a cationic lipid to DNA. J. Am. Chem. Soc. 1997, 119, 10920. (37) Smith, G. S.; Sirota, E. B.; Safinya, C. R.; Plano, R. J.; Clark, N. A. X-ray structural studies of freely suspended hydrated DMPC multimembrane films. J. Chem. Phys. 1989, 92, 4519. (38) Hentschel, M. P.; Rusticelli, F. Structure of the ripple phase Pβ′ in hydrated phosphatidylcholine multimembranes. Phys. ReV. Lett. 1991, 66, 903.