Biophysical and Structural Properties of DNA·diC14-amidine Complexes

Cationic liposomes are used as vectors for gene delivery both in vitro and in vivo. Comprehension of both DNA/liposome interactions on a molecular level and a description of structural modifications involved, are prerequisites to an optimization of the transfection protocol and, thus, successful application in therapy. Formation and stability of a DNA/cationic liposome complex were investigated here at different DNA:lipid molar ratios (ρ). Isothermal titration calorimetry (ITC) of cationic liposomes with plasmid DNA was used to characterize the DNA-lipid interaction. Two processes were shown to be involved in the complex formation. A fast exothermic process was attributed to the electrostatic binding of DNA to the liposome surface. A subsequent slower endothermic reaction is likely to be caused by the fusion of the two components and their rearrangement into a new structure. Fluorescence and differential scanning calorimetry confirmed this interpretation. A kinetic model analyzes the ITC profile in terms of DNA/cationic liposome interactions.

Transfer of nucleic acids into cells requires the use of transfection agents such as virus-based vectors (1,2) or cationic liposomes (3)(4)(5)(6)(7)(8)(9)(10). Cationic liposomes offer several advantages over viral vectors, including the low immunogenic and inflammatory responses, the potential transfer of unlimited-size expression units, and the possibility for engineered cell-specific targeting. In vitro studies have established the efficiency of such DNA/cationic liposome complexes (4,6,8,10,11); in vivo, these systems are in most cases less effective for gene transfer than viral vectors and unstable in serum (12)(13)(14).
Transfection efficiency depends to a great extent on the cell lines or/and the lipid composition (15,16). Reasons for these discrepancies are largely unknown. One of them is our poor knowledge of the structure of the DNA/lipid complex and of its mode of entry into the cell. Extensive efforts have been made to characterize the DNA/cationic lipid complex (17) leading to different structural models (18,19). Gershon et al. (20) suggested that DNA is encapsulated inside large unilamellar liposomes. On the other hand, in the so-called "sandwich-like" structure observed by x-ray (21), freeze-fracture electron microscopy (22), or cryotransmission electron microscopy (23), DNA is adsorbed between liposome bilayers in an alternating flat lamellar packing (24). More recently, another model suggested that the liposomes are broken, resulting in DNA being coated by a cylindrical bilayer, as shown by freeze-fracture electron microscopy (25).
However, these structural studies provide no information about the nature of the interactions involved in the complex formation. The interactions between cationic liposomes of diC 14 -amidine (see Fig. 1) and a plasmid DNA, as well as the structural modifications involved in the complex formation, were investigated here at different DNA:lipid molar ratios (). This complex has been used successfully to transfect cells in vitro (10). Isothermal titration calorimetry (ITC) 1 was used to characterize the interactions involved in the complex formation. Fluorescence spectroscopy and differential scanning calorimetry (DSC) provided detailed information about the rearrangement of the two components during the complex formation.
We propose a kinetic model to analyze the ITC profile in terms of DNA and cationic liposome interactions.
Preparation of Liposomes-diC 14 -amidine was dissolved in chloroform, dried under a nitrogen stream, and stored overnight in a desiccator under vacuum. Liposomes were formed by addition of 10 mM Hepes buffer, pH 7.3, to the lipid film and mechanical mixing above the transition temperature. Prior to each experiment, the liposomal suspension was degassed under vacuum and vortexed for 10 min.
Isothermal Titration Calorimetry Measurements-ITC experiments * This work was supported by Fonds pour la Formation à la Recherche dans l'Industrie et dans l'Agriculture, Belgium and Vlaams Interuniversitair Instituut voor Biotechnologie, Belgium. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
At constant time intervals (every 7 or 6 min, respectively, for the high and low lipid concentration), aliquots of DNA plasmid solution were injected, via a 100-l rotating stirrer-syringe, into the sample cell (volume ϭ 1.33 ml) containing cationic liposomes of diC 14 -amidine in 10 mM Hepes (pH 7.3).
In control experiments, the DNA solution was injected into pure buffer. The heat of dilution was subtracted from the experimental curve in the final analysis.
Fusion Studies-DNA-induced fusion of diC 14 -amidine liposomes was monitored using pyrene-PC. Association of pyrene monomers ( ex ϭ 346 nm; em ϭ 376 nm) leads to the formation of so-called pyrene excimers ( ex ϭ 330 nm; em ϭ 477 nm). The excimer formation is proportional to the density of pyrene-PC (10% in moles) on the liposome surface. Therefore, any lipid mixing between labeled and unlabeled liposomes (present on a 20-fold excess) results in dilution of pyrene-PC, reflected by a decrease of the excimer fluorescence and a concomitant increase of the monomer fluorescence. Fluorescence was continuously recorded at the excimer maximum emission (477 nm). Labeled and unlabeled liposomes were mixed under stirring in a quartz fluorescence cuvette to a final diC 14 -amidine concentration of 177 M.
The liposome fusion triggered by DNA was quantified by considering the 0% fusion level as corresponding to the initial excimer fluorescence (without DNA) and the 100% level, to the fluorescence measured after addition of an excess of Triton X-100 (0.2% v/v final) to maximize pyrene-PC dispersion. Each sample was incubated for 5 min at the experimental temperature (28°C) before starting the experiment. DNA was added every 100 s to allow the stabilization of the fluorescence signal.
Differential Scanning Calorimetry Measurements-DSC measurements were performed on a MC-2 ultrasensitive differential scanning calorimeter (MicroCal Inc.) using twin 1.2-ml total filled cells. The samples were scanned from 10°C to 40°C at 1°C/min. All samples were degassed before measurements. The experimental data were processed using Origin software from MicroCal.
The fluorescence experiments were carried out on an SLM-8000 spectrofluorimeter, at 37°C, using an excitation wavelength of 492 nm and an emission wavelength of 520 nm. Each measurement was performed at least in duplicate with both liposomes and plasmid DNA from different batches. A liposome solution diluted in Hepes/NaCl buffer (pH 7.3) (final sample volume: 1 ml) was used as "zero" fluorescence (I 0 ). Maximal release (I f ) was obtained after addition of 10 l of TX-100. The release measured at each ratio was calculated as x ϭ [(I t Ϫ I 0 )/(I f Ϫ I 0 )] ϫ 100, where I t is the fluorescence measured at time t at the plateau for each ratio.

RESULTS
The overall interaction process between plasmid DNA and cationic diC 14 -amidine liposomes was investigated using high sensitivity isothermal titration calorimetry (ITC) (26).
Plasmid DNA was injected into the cell containing unilamellar cationic liposomes of diC 14 -amidine (Fig. 1). The lipid concentration was always kept well above the critical micellar concentration (CMC ϭ 3.9 ϫ 10 Ϫ7 M) during the whole titration, making the contribution of free lipid negligible.
At high lipid concentration (above 10 mM), a fast exothermic process ( Fig. 2A, triangles) and a concomitant endothermic one ( Fig. 2A, circles) were observed after DNA addition. The fast exothermic process (equilibrium was reached within 1 min) reflects the electrostatic binding of the plasmid DNA to the positive liposome surface as observed in most binding processes between charged molecules (27,28). The accompanying endothermic process is about six times slower (see inset in Fig. 2A) than the exothermic one and suggests a rearrangement of the two components. The curve in Fig. 2A clearly shows that this endothermic process is cooperative as reflected by the exponential increase of the signal. In a noncooperative binding mechanism, the enthalpy peak would be constant (28,29).
At low lipid concentrations, the exothermic process could not be detected anymore, but the endothermic one was still observed (Fig. 2B). No heat effect was observed above Х 0.6 ( Fig.  2, A and B). At this ratio, DNA complexation is maximal. At higher DNA:lipid ratios, free DNA was indeed detected in the supernatant of centrifuged complexes (data not shown). Although formation of DNA-lipid complex is, classically, a onestep process, transfection experiments demonstrated that the transfection efficiency associated to the complex was not significantly different when it was formed, as in the ITC experiment, by a stepwise addition (data not shown).
The ITC data provided, however, little insight about the endothermic reaction that involves probably major internal reorganization of the DNA/lipid complex. It has been reported that divalent cations such as Ca 2ϩ (30) and polycationic amino acids (31) induce the fusion of anionic liposomes. The mechanism whereby these cations can induce fusion arise primarily from neutralization of the surface charge of the anionic lipids. Similarly, multivalent anions such as oligonucleotides or DNA can trigger the fusion of cationic liposomes (20,32,33). The endothermic reaction here observed could therefore be assigned to a DNA-induced lipid fusion.
A fluorescent lipid (pyrene-DPPC) was inserted into the cationic liposomes to detect a possible lipid mixing at different DNA:cationic lipid molar ratios (34). Insertion of pyrene-DPPC did not affect significantly their transfection efficiency (data not shown). As illustrated in Fig. 3A,DNA caused a substantial decrease of the excimer fluorescence within 10 s after DNA injection, reflecting a rapid and accelerating fusion (Fig. 3B). Flocculation was observed above a 0.6 molar ratio ( Ն 0.6). A similar phenomenon has been reported previously (35)(36)(37). DNA induces fusion of cationic liposomes and leads to larger structures with increasing . The final DNA:lipid ratio at which cationic liposome fusion was observed, did not depend on the addition protocol (stepwise addition or one single addition) (data not shown).
DNA ability to induce lipid mixing is probably related to its capacity of destabilizing the lipid bilayer organization (38). Differential scanning calorimetry (DSC) revealed a gel-liquid transition at 23.0 Ϯ 0.1°C and an enthalpy of 7548 Ϯ 293 kcal/mol of lipid for diC 14 -amidine cationic liposomes in 10 mM Hepes. At low DNA:lipid molar ratio (e.g. ϭ 0.03; ϭ 0.21), neither the transition temperature nor the enthalpy were affected (Fig. 4, A and B). At higher DNA:lipid molar ratios, DNA destabilizes the liposomal bilayer in a way reminiscent of that FIG. 1. Structure of N-t-butyl-N-tetradecyl-3-tetradecylamino-propionamidine or diC 14 -amidine.
described for cholesterol: no significant shift of the transition temperature and a concomitant decrease of the enthalpy. Destabilization was completed at a 1.0 molar ratio (Fig. 4B), suggesting that the lipid structure has a new organization.
Release of fluorescent dextran-FITC with a Stokes diameter of 60 Å and encapsulated into the diC 14 -amidine liposomes confirms the DNA-induced destabilization of the liposomal bi-layers (Fig. 5A). A 80% release was recorded at a DNA:lipid molar ratio of 0.6 ( Fig. 5B), supporting the hypothesis that DNA triggers the disruption of the liposomal membrane. DISCUSSION Our data suggest that the DNA/diC 14 -amidine liposome complex formation proceeds in several steps: where D is DNA, L diC 14 -amidine liposomes, C a soluble DNAlipid complex, and finally, F represents the fused complex.
Step 1-The primary driving force of the complex (C) formation is the electrostatic interaction between the negative phosphate groups of DNA (D) and the positive-charged groups of the diC 14 -amidine liposomes (L). A single plasmid DNA (5.4 kbp) binds several positively charged diC 14 -amidine liposomes (60-nm diameter, as determined by laser light scattering) (data not shown). This process is rapid and exothermic as verified by isothermal titration calorimetry at high total lipid concentration. It is assumed that the complex dissociates with a k 2 rate constant. Step 2-Charge neutralization abolishes repulsion between cationic liposomes (32). In addition, the bilayer organization is strongly destabilized (as illustrated in FITC-dextran release analysis (Fig. 5) and in DSC profiles (Fig. 4)) and undergoes a slow entropy-driven endothermic fusion process. Maximal fusion is reached at Ͼ 0.6.
These two steps can be described by the following set of differential equations: , and [F] as molar concentrations of D, L, C, F, and k 1 , k 2 , and k 3 the corresponding phenomenological (macroscopic) rate constants. k 1 described a second order process; k 2 and k 3 , a first order mechanism. Numerical solutions of the concentration profiles D(t), L(t), C(t), and F(t) were calculated for a set of different values of rate constants, applying the Runge Kutta method for differential equations. Using the initial concentrations and time scale described in Fig. 2, A and B, a realistic set of estimated rate constants was derived: k 1 ϭ 35,000 s Ϫ1 ⅐ M Ϫ1 , k 2 ϭ 8.5 10 Ϫ3 s Ϫ1 , and k 3 ϭ 1.7 10 Ϫ1 s Ϫ1 . Fig. 6 (A and B)  The L concentration is much higher than that of D and products C and F and was thus considered as constant and not represented. At low lipid concentration, formation of the complex C is the limiting factor (k 1 is rate-limiting). As soon as it is formed, C is further converted into larger fused particles (F); both processes occur at almost the same rate (Fig. 6B). C does not accumulate. This explains why only the endothermic contribution was detectable on the calorimetric pattern (Fig. 2B). At high lipid concentration (Fig. 6A), the reaction proceeds similarly but faster due to high concentration, but interestingly, C accumulates during the first minute of the process. It is precisely within this period of time that the fast exothermic process was observed experimentally ( Fig. 2A) in the ITC pattern. In this situation, the k 3 is rate-limiting.
To summarize, this model allows to analyze the structural changes observed around a critical DNA:lipid ratio in terms of interactions and complex formation. The kinetic model does not take into account the transformation of the fused complex into larger structures (flocculation). The main reason is that above a 0.6 ratio, no heat changes were detectable in the titration calorimetry profile. DSC (Fig. 4) and dextran-FITC release experiments (Fig. 5) demonstrated, however, a complete destabilization of the lipid bilayer above the 0.6 ratio. Close to a 0.6 DNA:lipid molar ratio, further addition of DNA fully destabilizes the lattice due to compensation of the lipid-positive charges. As suggested by Dü zgü nes et al. (32) to explain polyanion-induced fusion of dioleyloxypropyltrimethylammonium (DOTMA) liposomes, water molecules are expelled from the liposomal surface during the complex formation, making it more hydrophobic. Consecutively, the intervesicular electrostatic repulsions are reduced, leading to a collapse of the complex into larger macroscopic aggregates (37,39). Those large particles start to sediment, as reflected by the white flocculating aspect of the solution.
Those results could corroborate the "rod-shaped" structural model (25) with a breakup of the liposomes and the coating of the DNA by cylindrical bilayers. However, they are not in agreement with either the Gershon model (20) or the "sandwich-like" model (22), both of which support persistence of liposomal structures after complex formation.
To relate different biophysical characterization of DNA/ diC 14 -amidine complexes with their transfection properties, we have compared the transfection efficiency on CHO cells at various DNA:lipid molar ratios. The resulting profile (Fig. 7) shows clearly that transfection increases strongly near the critical 0.6 molar DNA:lipid ratio, is maximal around 0.8, and then decreases at 1.0. This profile is reminiscent of the modification of the biophysical parameters at various DNA:lipid ratios. There is an obvious parallel between the results of our biophysical experiments and the transfection properties of the DNA/diC 14 -amidine complexes, showing that maximal transfection activity occurs in a range where liposome destabilization through DNA is close to maximal.