- Email: [email protected]
Isothermal Titration Calorimetric Analysis of the Interaction between Cationic Lipids and Plasmid DNA
Archives of Biochemistry and Biophysics Vol. 386, No. 1, February 1, pp. 95–105, 2001 doi:10.1006/abbi.2000.2196, available online at http://www.idealibrary.com on
Isothermal Titration Calorimetric Analysis of the Interaction between Cationic Lipids and Plasmid DNA Brian A. Lobo,* Amber Davis,† Gary Koe,‡ Janet G. Smith,‡ and C. Russell Middaugh* ,1 *The Department of Pharmaceutical Chemistry, University of Kansas, 2095 Constant Avenue, Lawrence, Kansas 66047; †Haskell Indian Nations University, Lawrence, Kansas 66047; and ‡Valentis, Inc., 863A Mitten Road, Burlingame, California
Received August 4, 2000, and in revised form October 30, 2000
The effects of buffer and ionic strength upon the enthalpy of binding between plasmid DNA and a variety of cationic lipids used to enhance cellular transfection were studied using isothermal titration calorimetry at 25.0°C and pH 7.4. The cationic lipids DOTAP (1,2-dioleoyl-3-trimethyl ammonium propane), DDAB (dimethyl dioctadecyl ammonium bromide), DOTAP:cholesterol (1:1), and DDAB:cholesterol (1:1) bound endothermally to plasmid DNA with a negligible proton exchange with buffer. In contrast, DOTAP: DOPE (L-␣-dioleoyl phosphatidyl ethanolamine) (1:1) and DDAB:DOPE (1:1) liposomes displayed a negative enthalpy and a significant uptake of protons upon binding to plasmid DNA at neutral pH. These findings are most easily explained by a change in the apparent pK a of the amino group of DOPE upon binding. Complexes formed by reverse addition methods (DNA into lipid) produced different thermograms, sizes, zeta potentials, and aggregation behavior, suggesting that structurally different complexes were formed in each titration direction. Titrations performed in both directions in the presence of increasing ionic strength revealed a progressive decrease in the heat of binding and an increase in the lipid to DNA charge ratio at which aggregation occurred. The unfavorable binding enthalpy for the cationic lipids alone and with cholesterol implies an entropy-driven interaction, while the negative enthalpies observed with DOPE-containing lipid mixtures suggest an additional contribution from changes in protonation of DOPE. © 2001 Academic Press
Key Words: isothermal titration calorimetry; DNA; cationic lipids; enthalpy; ionic strength; helper lipid; proton linkage; nonviral gene therapy; lipoplexes.
1 To whom correspondence [email protected].
should
0003-9861/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.
be
addressed.
E-mail:
Cationic lipids currently offer one of the most promising approaches to facilitating the delivery and expression of DNA into cells with human clinical trials currently underway. The mechanism of association of cationic lipids in liposomal form with plasmid DNA is known to be complex and mediated by a variety of physical interactions including an ionic interaction between positively charged lipid head groups and negatively charged DNA phosphates (1), apolar interactions between lipid and DNA, repulsive forces between DNA molecules and between cationic lipid head groups (2), and hydration effects. These forces, combined with the structural properties of the plasmid DNA and liposomes (including lipid phase state) (2) and the ratio of the two components, can result in the formation of a variety of supramolecular structures variously characterized as spaghetti and meatballs (3), stacks of lamellar lipid sheets with intercalated DNA (4 – 6) and lipidcoated DNA strands arranged in an inverse hexagonal array (7, 8). The formation of these complexes is believed to occur in a highly cooperative manner mediated by a critical ratio of lipid to DNA that induces DNA collapse and lipid fusion (9, 10). Wong et al. (10), using a Bligh and Dyer monophase system, have estimated the dissociation constant of DODAC (dioleoyldimethyl ammonium chloride) 2/DNA complexes as less than 10 ⫺11 M, suggesting a very high affinity 2 Abbreviations used: DOTAP, 1,2-dioleoyl-3-trimethyl ammonium propane; DDAB, dimethyl dioctadecyl ammonium bromide; DOTAP: DOPE, L-␣-dioleoyl phosphatidyl ethanolamine; DODAC, dioleoyldimethyl ammonium chloride; CTAB, cetyl trimethyl ammonium bromide; DTAB, dodecyl trimethyl ammonium bromide; DOPC, 1,2-dioleoyl-sn-glycero-3-phosphocholine; DOPE, 1,2-dioleoyl phosphatidyl ethanolamine; ITC, isothermal titration calorimetry; TED, thermal electrical device; DLS, dynamic light scattering; PALS, phase analysis light scattering; TEA, triethanolamine; MWCO, molecular weight cut off; BES, 2-[bis(2-hydroxylethyl) amino]ethanesulfonic acid.
95
96
LOBO ET AL.
interaction. A further Scatchard and Hill analysis demonstrated a marked cooperativity to the interaction. The less complex but comparable interaction of cationic single-chain surfactants with DNA has been characterized as a biphasic cooperative process comprised of an initial electrostatic interaction followed by cooperative hydrophobic interactions between solvent exposed lipid tails (11–14). Titration calorimetry studies (11) of the interaction of the single-chain micelle forming lipid CTAB (cetyl trimethyl ammonium bromide) in monomer form with DNA resulted in the proposal of a two-step cooperative process with an initial binding of surfactant to DNA occurring with a binding constant of 1.5 ⫻ 10 3 M ⫺1 and a ⌬H of ⫺20 kJ/mol. This was followed by the cooperative binding of further surfactant molecules with a cooperative affinity constant of 8.7 ⫻ 10 4 M ⫺1 and a ⌬H of 3.3 kJ/mol. Patterkine and Ganesh (12) have detected cooperative binding based upon the UV absorbance melting curves of DNA bound to CTAB and DOTAP. A similar electrostatic/apolar process was described by Bathaie et al. (13) for the cationic surfactant DTAB (dodecyl trimethyl ammonium bromide). The role of hydrophobic effects involving surfactants and participation of apolar interactions in the extent of binding have also been investigated by Bhattacharya and Mandal (14). The purpose of this study is to provide a thermodynamic analysis of the interaction of two commonly used cationic lipids with supercoiled DNA and to evaluate the effects of commonly used transfection helper lipids such as DOPE and cholesterol that are known to enhance transfection efficiency. Titrations were performed in both the forward (lipid into DNA) and reverse (DNA into lipid) direction to evaluate the formation of complexes with either component in excess and in various buffers to detect any proton linkage in the interaction. The effect of charge shielding on the interacting ionic groups was evaluated by the addition of NaCl to the buffer in both titration directions. We also present data from analogous titrations using dynamic light scattering and phase analysis light scattering to monitor changes in complex size and zeta potential. These data are used to identify regions of colloidal instability that impact the interpretation of the calorimetric results. EXPERIMENTAL PROCEDURES
Materials Plasmid DNA. Purified supercoiled plasmid DNA was provided by Valentis, Inc. (Burlingame, CA). Examination of the integrity of these plasmids by 1% agarose gel electrophoresis with ethidium bromide staining revealed less than 5% of nicked and open-circular forms. The plasmid vectors used for this study were pMB113 (9.1 kbp) and pMB290 (5.1 kbp). Plasmid purification is described in Ref. (15).
Cationic and helper lipids. DOTAP (1,2-dioleoyl-3-trimethyl ammonium propane), DDAB (dioctadecyl dimethyl ammonium bromide), DOPE (1,2-dioleoyl phosphatidyl ethanolamine), DOPC (1,2dioleoyl-sn-glycero-3-phosphocholine), and cholesterol were obtained from Avanti Polar Lipids (Alabaster, AL) in chloroform solution and were used without further purification.
Methods Liposome preparation. Liposomes were prepared by a film extrusion method. A solution of lipid in chloroform, typically 25 to 50 mg/ml, was evaporated with nitrogen gas to form a dry lipid film inside a 1.7-ml glass vial. For preparations that contained more than one lipid, each lipid was dried and weighed in a separate vial. The amount of each lipid was adjusted to stoichiometric ratios and then redissolved in chloroform and combined into a single vial. Residual chloroform was removed by vacuum desiccation for 3 h. Buffer (10 mM TEA, Hepes, BES, or Tris, pH 7.4, with either 0, 150, 250, or 400 mM NaCl) was added to the dried lipid based upon the lipid weight to achieve the desired lipid concentration. The vial contents were vortexed for 1 to 2 min to completely remove lipid from the sides of the vial and were allowed to hydrate for at least 30 min. The lipid suspensions were then extruded 11 times through a 100-nm polycarbonate filter in 1-ml aliquots, using an Avanti mini-extruder (Avanti Polar Lipids, AL). The vortexing, hydration, and extrusion steps were all performed at 55– 65°C for lipid suspensions containing DDAB and at room temperature for DOTAP solutions. Liposome suspensions were used within 4 days of manufacture. Plasmid DNA preparation. Purified DNA stock solutions were diluted to the given concentrations using 10 mM Tris, pH 7.4, containing 0, 150, 250, or 400 mM NaCl. For titrations in Hepes, BES, or TEA buffer, DNA solutions were dialyzed using a Pierce 10,000 MWCO dialysis cassette (Pierce, Rockford, IL) in 2.0 L of buffer overnight at 5°C, with the dialysate retained for blank titrations and liposome hydration. The concentrations of DNA solutions used for calorimetric and light scattering titrations were determined by UVvisible spectrophotometry using a Hewlett–Packard Model 8453 UVvisible spectrophotometer and an extinction coefficient of 0.02 AU ml cm ⫺1 mg ⫺1 at 260 nm. Two to three aliquots were taken from each DNA solution (after degassing) and were diluted to 25 g/ml with buffer. The average of three measurements was taken for each diluted aliquot. An average molecular weight per nucleotide of 324.5 g/mol PO 423 was used for concentration calculation. Isothermal titration calorimetry (ITC). All calorimetric titrations were performed using a CSC Model 4200 isothermal titration calorimeter (Calorimetry Sciences Corp., Utah) at 25.0°C. The thermal electrical device (TED) of the calorimeter was calibrated at 25.0°C using 10 electrical pulses of 500 J, with both cells containing 1 ml purified water and the sample cell stirrer set at 75 rpm. All DNA and blank titrations were performed with full sample and reference cells, with the sample cell stirrer set at 297 rpm. The reference cell contained purified water and DNA and liposome solutions were degassed before use. CSC Data Collection or ITCRun software (Calorimetry Sciences Corp., Utah) were used for instrument control and for the collection of raw data (heat flow vs time). Raw titration data were integrated using CSC Dataworks or Bindworks 3.0 software. Titrations consisted of a 300- to 1200-s equilibration time with 5 min between injections. Three titration programs were utilized for the forward (lipid into DNA) titrations: 25 ⫻ 10, 40 ⫻ 5, and 75 ⫻ 1 l. Cationic lipid and DNA concentrations were 3.72 and 0.31 mM, respectively, for the 25 ⫻ 10 and 40 ⫻ 5 l programs and 26.1 mM lipid and 0.92 mM DNA for the 75 ⫻ 1-l scheme. Lipid concentration was doubled in the syringe and the injection volume halved for the forward titrations in 250 and 400 mM NaCl to obtain higher final charge ratios with an equivalent amount of lipid per injection. Reverse titrations of DNA into cationic lipid were conducted with a 40 ⫻ 3-l
ITC OF CATIONIC LIPIDS AND PLASMID DNA titration program (2.4 l delivered) using the following solutions: DNA (6.5 mM) into DDAB (1.6 mM) and DNA (7.1 mM) into DOTAP (1.1 mM). A 250-l syringe was used for all titrations, except for the 75 ⫻ 1-l titrations, where a 100-l syringe was used. Dilution heats were frequently obscured by an increase in thermal noise due to adherence of aggregated complexes to the stirrer and sides of the sample cell. This resulted in heats after saturation that were not reproducible and were dissimilar in magnitude to dilution heats measured from blank titrations. Therefore, dilution heats of liposomes (for lipid into DNA titrations) or DNA (for DNA into lipid titrations) were performed by injecting liposomes or DNA, respectively, into buffer and were run at least in duplicate. The heats of dilution were calculated as an overall average if constant with injection number and were subtracted from binding heats up to the point of aggregation. If the dilution heats were not constant with injection number, the average dilution heat per injection among replicates was calculated and subtracted on a one to one injection basis up to the point of aggregation. Dynamic light scattering (DLS). Dynamic light scattering titrations were performed with a Brookhaven Model BI200SM apparatus (Brookhaven, NY). Light scattering was detected at 90° employing a 50-mW helium/neon laser at a wavelength of 532 nm through a 200-m pinhole. Each forward titration employed four cuvettes: three contained 1.3 ml of DNA at 0.31 mM and one contained an equal volume of Tris buffer. DNA and buffer solutions were filtered at least six times through a 0.45-m Acrodisc filter and then added directly into the cuvettes based upon the equivalent weight of sample required. Into each cuvette was titrated 10 l of a 3.7 mM solution of cationic lipid with a pipet via a plastic tube passing through the cap. Each sample was stirred at approximately 100 rpm with a 10 ⫻ 3-mm magnetic stir bar, except during the measurement of particle size. Titration intervals were approximately 20 to 40 min in duration, since all four samples were titrated concurrently. Reverse titrations were conducted in a similar manner using three cuvettes containing 1.3 ml of cationic liposomes at 0.22 mM and one sample containing 1.3 ml of Tris buffer. Varying volumes of a 0.92 mM DNA solution were added to each sample to achieve evenly spaced (⫾) ratios. At every time point, each sample was measured in triplicate over a 1-min duration. Data points are plotted as the means ⫾ sem of the three samples. A cumulant method was utilized for analysis of the autocorrelation function, although the mean intensity-weighted diameter obtained from a bimodal distribution analysis (nonnegatively constrained least squares) of the data produced quite similar results. Phase analysis light scattering (PALS). Measurements of the zeta potentials of lipid–DNA complexes during forward and reverse titrations were made with a Brookhaven Zeta-PALS instrument (Brookhaven, NY). Samples were examined in disposable polystyrene cuvettes into which an electric probe was inserted. A 1-mm-diameter hole was drilled into the side of the cuvette above the level of the electrodes through which titrant was delivered via a Hamilton syringe at approximately 10-min intervals. Samples were stirred with an 8 ⫻ 1.5-mm magnetic stir bar. Forward titrations consisted of 13 injections of 10 l cationic lipid (14.2 mM) into 1.3 ml of DNA (0.74 mM). Reverse PALS titrations were conducted using the same DNA and lipid concentrations used for the reverse DLS titrations—1.3 ml of cationic lipid at 0.21 mM and a 0.92 mM DNA solution. Varying volumes of the DNA solution were added to achieve evenly spaced (⫾) ratios. Data points are plotted as the means ⫾ sem of at least three samples. The PALS measurement at each titration point consisted of five runs at 15 cycles (of the applied electric field) for each sample. The Smulchowski model was used for the calculation of zeta potential from the measured electrophoretic mobility.
97
FIG. 1. Effect of titration direction on the ITC thermograms of DOTAP and DNA. (A) 25 ⫻ 10-l titration of DOTAP liposomes (3.7 mM) titrated into plasmid DNA (0.31 mM) or 10 mM Tris buffer, pH 7.4, at 25.0°C. (B) Plasmid DNA (7.1 mM) into DOTAP liposomes (1.1 mM) using 40 injections of 2.4 l.
RESULTS
Titration of Cationic Lipid into DNA ITC titrations. A representative thermogram of a calorimetric titration of the cationic lipid DOTAP (3.72 mM) into plasmid DNA (0.31 mM) at 25.0°C using a program of 25 injections of 10 l (25 ⫻ 10 l) is presented in Fig. 1A. This lipid displays endothermic heats of binding that are relatively constant with injection number until the point of colloidal instability (see below) that results in complex aggregation and prevents the binding of further lipid. The aggregation of complexes in the cell causes an increase in the viscous heat of mixing (due to the adhesion of complexes to the stirrer) that creates a rapid upward (exothermic) slope in the baseline of the titration. This aggregation event occurs at charge ratios (positive to negative) in slight excess of unity. Titrations of DDAB into DNA under identical conditions gave results comparable to those of DOTAP, but did not display an exothermic event upon aggregation. Due to the aggregation of the complexes, an equilibrium saturation region is not present in these titra-
98
LOBO ET AL.
tions and an affinity constant cannot be obtained. Although it is possible to fit the binding isotherms to a single or a two independent site binding model, the fitting of these data that include extensive aggregation phenomena to equilibrium models is not appropriate. Furthermore, a true sigmoidal shape is not obtained even when the aggregation process occurs over a large number of injections, as seen with a titration protocol of 75 ⫻ 1-l injections of lipid into DNA (data not shown). Since binding heats are relatively constant with injection number, however, an apparent molar enthalpy of binding (⌬H app) can be calculated independently, assuming that all the lipid initially binds. The average corrected heats from the second through sixth injections for each titration were chosen as representative values for full ligand binding and divided by the moles of lipid per injection. Heats in this region of the isotherm should also contain the minimum thermal contribution from the aggregation process. The apparent enthalpies of binding in Tris buffer for DOTAP into DNA were 6.6 ⫾ 0.0 kJ/mol (25 ⫻ 1-l titration), 6.4 ⫾ 0.0 kJ/mol (40 ⫻ 5-l titration), and 7.3 ⫾ 0.1 kJ/mol (75 ⫻ 1-l titration). Apparent enthalpies for DDAB into DNA titrations were 5.6 ⫾ 0.3 kJ/mol (25 ⫻ 10-l titration), 7.2 ⫾ 0.2 kJ/mol (40 ⫻ 5-l titration), and 7.1 ⫾ 0.5 kJ/mol (75 ⫻ 1-l titration). Dynamic light scattering titrations. Light scattering titrations were designed to simulate the conditions of the 25 ⫻ 10-l forward ITC titrations and employed the same concentrations of DNA and lipid. The time period of complex formation, however, was much longer during the DLS titration procedure since the time between titrations was approximately 40 min, compared to the 5-min periods used during the calorimetry experiments. Changes in the mean diameter of the complexes during the titration of cationic lipid formulations into plasmid DNA are shown in Fig. 2A. The farthest point to the right on each titration curve represents the last measurement before aggregation and precipitation occur, prohibiting further measurement. With successive addition of lipid, the complexes were relatively constant in size until the point of colloidal instability. The mean diameters of free DOTAP and DDAB liposomes were 112 ⫾ 1 and 115 ⫾ 0.2 nm, respectively. The maximum (⫾) charge ratios that could be achieved before aggregation were similar to the ratios of cationic lipid to DNA at which the ITC data suggested aggregation. The change in polydispersity index of complexes formed from these forward titrations is shown in Fig. 2B. The DNA alone appears rather polydisperse which could, in part, reflect contributions to the autocorrelation function from internal segmental motions of the plasmid (16). Thus, the observed decreases in polydis-
FIG. 2. Quadratic cumulant diameter (A) and polydispersity (B) vs complex charge ratio (⫾) of DNA/lipid complexes determined by dynamic light scattering. Each data point represents the mean ⫾ sem of three samples of complexes formed by sequential titration of a cationic liposome formulation (3.7 mM) into plasmid DNA (0.31 mM) in 10 mM Tris, pH 7.4. (䊐) DOTAP alone, (E) DDAB alone, (⌬) DOTAP:DOPE (1:1). (C) Effect of (⫾) charge ratio on the zeta potential of DNA/lipid complexes determined by PALS titrations of cationic lipid formulations (14.2 mM) into plasmid DNA (0.74 mM) and 10 mM Tris buffer, pH 7.4. (■) DOTAP into DNA, (F) DDAB into DNA, (Œ) DOTAP:DOPE (1:1) into DNA, (䊐) DOTAP into buffer, (E) DDAB into buffer, (‚) DOTAP:DOPE (1:1) into buffer.
persity of the lipid/DNA complex may result to some extent from decreases in these internal DNA motions upon lipid binding. The very low and constant polydis-
ITC OF CATIONIC LIPIDS AND PLASMID DNA
persity up to the point of aggregation of the DOTAP and DDAB complexes suggests that these complexes are not only very homogeneous but also precipitate without a gradual increase in size distribution. Phase analysis light scattering titrations. The forward titrations were repeated manually with PALS, which employed fewer titration steps than the ITC titrations to minimize the plating of the electrodes during the titration process. Titration additions were spaced approximately 10 min apart. As illustrated in Fig. 2C, the titration of DOTAP and DDAB liposomes into plasmid DNA all produced an initial increase in negative zeta potential at low charge ratios followed by a period of constant potential. Titrations of DOTAP and DDAB into DNA produced visual aggregation at a (⫾) charge ratio of 1.1:1. The measured zeta potential of DNA alone (⫺20 mV) was artifactually low due to the low scattering intensity of DNA in solution under these conditions. A more meaningful value of ⫺40 mV was obtained with the blank reverse titration of DNA into buffer (Fig. 4C). The increase in negative surface potential of the DNA from ⫺40 mV to ⫺55 to ⫺60 mV after the first titration was probably due to the initial compaction of DNA (on the surface of the liposomes). A similar study performed by Eastman et al. (17) of the zeta potential of 1,2-dimyristyloxypropyl-3-dimethyl-hydroxylethyl ammonium bromide/DNA complexes prepared as individual samples showed results very similar to those shown in Fig. 2C. Titration of DNA into Cationic Lipid Reverse ITC titrations. To examine regions of high lipid to DNA charge ratios, titrations were performed in the reverse direction (DNA into lipid) to provide some thermodynamic insight into the nature of the complexes formed on the other side of the charge neutrality boundary. A representative thermogram of a reverse titration of plasmid DNA (7.1 mM) into DOTAP (1.1 mM) using 40 ⫻ 2.4-l injections is shown in Fig. 1B. In contrast to the forward titrations, which produced complexes that aggregated at charge ratios near unity, these reverse titrations produced complexes that aggregated at (⫿) ratios of 0.33 ⫾ 0.02 for DOTAP and 0.29 ⫾ 0.01 for DDAB (which correspond to (⫾) ratios of 3.1 ⫾ 0.2 and 3.5 ⫾ 0.1, respectively). Although the aggregation event manifests itself in a more subtle manner with a reverse titration, it still prohibits the application of equilibrium binding site models to the data. Therefore, apparent molar enthalpies for reverse titrations were calculated as the average heat per injection for the second through sixth injections (which are relatively constant in magnitude) divided by the moles of DNA per injection. The appar-
99
ent enthalpy of binding for DNA into DOTAP is 5.4 ⫾ 0.4 kJ/mol and for DNA into DDAB 4.8 ⫾ 0.2 kJ/mol. The calculated enthalpies for the reverse titrations were lower than those obtained from any of the forward titration programs for both cationic lipids. This difference may be at least partially due to the higher concentrations of DNA and lipid required for the reverse titrations. Reverse dynamic light scattering titrations. Reverse DLS titrations were also performed at concentrations of DNA and lipid similar to those used during the calorimetric titrations, but 20 min between injections were required. Increasing amounts of DNA were titrated into lipid formulations at approximately evenly spaced (⫾) charge ratios. The quadratic cumulant diameter and polydispersity index of these complexes as a function of charge ratio are shown in Figs. 3A and 3B, respectively. Titrations were performed from right to left along the x axis and the last titration point before gross aggregation was observed is shown as the leftmost data point. All titration curves showed a continual increase in size with DNA addition, suggesting that complex formation and self-association occur over a broad range of charge ratios prior to gross colloidal instability. Reverse ITC titrations also manifest a greater total (⫾) charge ratio before aggregation for DDAB than DOTAP, although this value is different for the two methods. This could be due to differences in mixing speed and DNA addition rate between the two methods. The changes in complex polydispersity for the reverse DLS titrations with DOTAP, DDAB, and DOTAP:DOPE (1:1) are shown in Fig. 3B. All titrations show an increase in complex polydispersity as the DNA is added to the cationic lipid. The lipid vesicles alone were very monodisperse (polydispersities of DDAB, DOTAP, and DOTAP:DOPE (1:1) liposomes were 0.108 ⫾ 0.016, 0.105 ⫾ 0.007, and 0.111 ⫾ 0.009, respectively). Finally, it should be noted that a linear increase in scattering intensity is seen in all forward and reverse titrations with the sequential addition of the lipid or DNA (not illustrated). Reverse PALS titrations. Reverse PALS titrations for DNA into DDAB or DOTAP liposomes are shown in Fig. 3C and were performed from right to left along each curve. These titrations were similar in shape to the forward titrations, with the exception of the point of inflection from positive to negative surface potential. The titration points before saturation (DNA with an excess of cationic lipid) are similar in zeta potential to free cationic liposomes (⫹40 to ⫹50 mV). The point of visual aggregation occurred at a charge ratio of 4.0 for DDAB and at 2.5–3.0 for DOTAP, which again slightly preceded the point at which surface potential shifted from positive to negative. The charge ratios of com-
100
LOBO ET AL.
plexes within the ⫾20-mV region correspond reasonably well with the onset of aggregation seen for DDAB and DOTAP with the reverse ITC titrations (3.1 ⫾ 0.2 for DOTAP and 3.5 ⫾ 0.1 for DDAB). Effect of Salt Concentration
FIG. 3. (A, B) Effect of charge ratio (⫾) on the quadratic cumulant diameter (A) and polydispersity (B) of DNA/lipid complexes determined by DLS. Each data point represents the mean ⫾ sem of three samples of complexes formed by sequential titration of plasmid DNA (0.92 mM) into cationic lipid formulations (0.22 mM). Titrations were performed from right to left along the x axis. (䊐) DOTAP alone, (E) DDAB alone, (‚) DOTAP:DOPE (1:1). (C) Effect of (⫾) ratio on the zeta potential of lipid/DNA complexes from PALS titrations of plasmid DNA (0.92 mM) into cationic lipid formulations (0.21 mM) and Tris buffer. Titrations were performed from right to left along the x axis. (}) DNA into DOTAP, (■) DNA into DDAB, (‚) DNA into Tris buffer.
Forward and reverse titrations of DOTAP and DNA were repeated in the presence of high concentrations of sodium chloride to modify any ionic component of the interaction and to determine related effects upon the enthalpy of binding. Titrations were performed using identical experimental conditions employed in the absence of salt except where noted. As shown in Fig. 4A, an increase in NaCl concentration to 400 mM results in a dramatic decrease in the binding heats, accompanied by an increase in the amount of lipid required to induce colloidal instability. Liposome stability in this range of salt concentrations was not compromised since DOTAP liposomes in 400 mM NaCl were both stable and possessed only a slightly higher mean diameter than in the absence of salt (154 ⫾ 2 nm vs 112 ⫾ 1 nm). DOTAP liposomes were unstable, however, above 400 mM NaCl. Samples removed from the calorimeter after completion of the salt titrations appeared as diffuse, cloudy solutions, compared to the more opaque particulate suspension seen in the absence of salt. This suggests that gross aggregation is somewhat inhibited when salt is present above 150 mM. The lack of an exothermic heat event after saturation during these salt titrations could also be due to this inhibition of gross aggregation. Furthermore, the magnitude of the baseline shift that coincides with aggregation was progressively reduced with increasing salt (data not shown). There was no obvious evidence for any type of salt-dependent binding saturation event since the loss of binding heat occurred within one to two injections at all concentrations of added salt. Similar results were seen when these titrations were repeated using DLS and PALS, as shown in Figs. 4B and 4C. DNA alone appeared more polydisperse with added salt, while complex polydispersity increased slightly with higher salt concentration (from 0.1 to 0.2; not shown). Complex polydispersity, however, remained constant with injection number within each titration (not shown). Although the zeta potential of DNA alone was significantly reduced in the presence of increasing salt, the zeta potential of the complexes formed under these conditions remained at ⫺40 mV at all salt concentrations. Reverse titrations of DNA into DOTAP in the presence of 150, 250, and 400 mM NaCl also display major effects of salt concentration, as shown in Figs. 5A through 5C. Salt addition completely eliminates the progressive increase in binding heats prior to the saturation seen in the absence of salt, suggesting that this
ITC OF CATIONIC LIPIDS AND PLASMID DNA
FIG. 4. Effect of NaCl concentration on the integrated heats of binding, complex diameter, and zeta potential of complexes formed by addition of DOTAP liposomes to DNA by ITC, DLS, and PALS. (A) Representative titrations of DOTAP into DNA in 10 mM Tris buffer, pH 7.4, at 25.0°C. (}) No added NaCl, (‚) 150 mM NaCl, (E) 250 mM NaCl, (⫻) 400 mM NaCl. Titrations consisted of 25 injections of 10 l of DOTAP (3.7 mM) into DNA (0.31 mM) for 0 and 150 mM NaCl and 50 injections of 5 l of DOTAP (7.4 mM) into DNA (0.31 mM) for 250 and 400 mM NaCl. Note: Lines between data points do not represent a binding model fit to the data. (B) Cumulant diameter of complexes formed by addition of DOTAP liposomes (3.72 mM) to DNA (0.31 mM) vs (⫾) charge ratio. (C) Zeta potential of DOTAP/DNA complexes formed by addition of DOTAP liposomes (14.2 mM) into DNA (0.74 mM) vs (⫾) charge ratio. Panels B and C: (}) no added NaCl, (Œ) 150 mM NaCl, (F) 250 mM NaCl. Open symbols represent titrations of DOTAP into each buffer alone.
increase in heat may be coincident with gross aggregation. Increasing salt concentrations also produce a reduction in the binding heat per injection and a decrease in the amount of DNA required to induce aggregation. In both directions, however, an increase in salt concentration results in an increase in the (⫾) charge ratio
101
FIG. 5. Effect of NaCl concentration on the integrated heats of binding, complex diameter, and zeta potential of complexes formed by addition of DNA into DOTAP liposomes by ITC, DLS, and PALS. (A) Representative ITC titrations of DNA into DOTAP in 10 mM Tris buffer, pH 7.4, at 25.0°C. (*) No added NaCl, (F) 250 mM NaCl, (E) 400 mM NaCl. Titrations consisted of 40 injections of 2.4 l of DNA (7.1 mM) into DOTAP (1.1 mM) for 0 mM NaCl or DOTAP (4.3 mM) for 250 and 400 mM NaCl. (B) Cumulant diameter of complexes formed by addition of DNA (1.4 mM) to DOTAP liposomes (0.21 mM) vs (⫾) charge ratio. (C) Zeta potential of DOTAP/DNA complexes formed by addition of DNA (0.94 mM) into DOTAP liposomes (0.21 mM) vs (⫾) charge ratio. Panels B and C: (}) no added NaCl, (Œ) 150 mM NaCl, (■) 250 mM NaCl. Open symbols represent titrations of DNA into each buffer alone.
102
LOBO ET AL.
that produces aggregation. Complexes formed by addition of DNA into lipid did not show a significant increase in polydispersity with higher concentrations of salt (not illustrated). Effect of Helper Lipid The neutral helper lipids DOPE and cholesterol were included at a 1:1 mole ratio with DOTAP or DDAB to determine if their presence would alter the thermodynamics of cationic lipid/DNA association. This ratio is one commonly employed to enhance the effectiveness of lipoplexes as transfection agents (18). Forward titrations were performed using a 25 ⫻ 10-l titration program and the same cationic lipid and DNA concentrations used in the pure cationic lipid studies. In all cases, the inclusion of DOPE or cholesterol at a 1:1 molar ratio with either DOTAP or DDAB did not significantly change the shape of the binding isotherm from that of the titration of pure cationic lipid into DNA. Binding heats were constant with injection number and were rapidly lost within a 1:1 to a 1.5:1 charge ratio (not shown). To determine the contribution of any proton-linked equilibria to the enthalpy of binding, the titration of each cationic lipid mixture into DNA was performed in a variety of buffers with varying enthalpies of ionization at a constant buffer concentration of 10 mM at pH 7.4. As shown by Baker and Murphy (19), the intercept of a plot of apparent binding enthalpy vs buffer ionization enthalpy under these conditions corresponds to the buffer-independent binding enthalpy (⌬H 0 ) and
TABLE I
Buffer-Independent Molar Enthalpies of Binding and Linked Protonation of Binding of Cationic Lipid Mixtures Titrated into Plasmid DNA at 25.0°C and pH 7.4
Lipid
⌬H (kJ/mol)
Protons exchanged with buffer
DOTAP DOTAP:DOPE (1:1) DOTAP:DOPC (1:1) DOTAP:Chol (1:1) DDAB DDAB:DOPE (1:1) DDAB:Chol (1:1)
7.0 ⫾ 1.7 ⫺21.9 ⫾ 2.3 1.4 ⫾ 0.2 0.8 ⫾ 2.4 11.6 ⫾ 1.7 ⫺10.8 ⫾ 1.1 10.6 ⫾ 1.3
0.0 ⫾ 0.0 0.5 ⫾ 0.1 0.0 ⫾ 0.0 ⫺0.1 ⫾ 0.1 ⫺0.1 ⫾ 0.0 0.3 ⫾ 0.0 ⫺0.1 ⫾ 0.0
Note. Apparent molar enthalpies of binding of cationic lipid mixtures to DNA were determined by ITC. Integrated heats per injection were averaged from the second through sixth injections of each titration and divided by the moles of titrant delivered per injection. All molar enthalpies are reported in kilojoules per mole at 25.0°C and pH 7.4.
the slope of this plot to the number of protons exchanged with the buffer per mole of bound ligand. The plot of apparent binding enthalpy vs buffer ionization enthalpy for these lipids is shown in Fig. 6 and the calculated ⌬H 0 and number of exchanged protons is summarized in Table I. Only the DOPE-containing lipid mixtures display a marked variation in apparent binding enthalpy with buffer, signifying a measurable uptake of protons upon binding. The negative ⌬H 0 for these DOPE mixtures is not, however, the intrinsic enthalpy of binding since it includes an enthalpic contribution of the ionization process. Further investigation is ongoing into this buffer dependence as a function of pH in order to determine the magnitude of the pK a shift and the intrinsic enthalpy of binding (19). DISCUSSION
FIG. 6. Effect of buffer ionization enthalpy on the apparent enthalpy of binding of cationic lipids into plasmid DNA. All titrations were performed with 10-l injections of 3.72 mM cationic lipid into 0.31 mM DNA at 25.0°C and pH 7.4. Apparent enthalpies of binding were calculated with a model-independent method as described in the text. (■) DOTAP, (Œ) DOTAP:DOPE (1:1), ({) DOTAP:DOPC (1:1), (F) DOTAP:cholesterol (1:1), (䊐) DDAB, (‚) DDAB:DOPE (1:1), (E) DDAB:cholesterol (1:1).
The ITC, DLS, and PALS studies presented here find that the interaction between cationic liposomes and plasmid DNA cannot be described by simple equilibrium binding models, since aggregation of complexes limits the binding process. Although the calorimetric response to aggregation was quite different in each titration direction, the loss of binding heat occurred at charge ratios where aggregation was observed with DLS and a minimum zeta potential was detected by PALS. These manifestations of aggregation were seen in both titration directions, i.e., whether the lipid was added to DNA or the DNA was added to lipid. The effect of titration direction was quite significant in all cases and reflects the probability that complexes were formed with different structures and colloidal properties in the two different complexation processes. The results from the forward titrations suggest that
ITC OF CATIONIC LIPIDS AND PLASMID DNA
the binding of lipid to DNA occurred in an apparently noncooperative manner until aggregation behavior begins. Complexes formed in this manner have a negative zeta potential and presumably consist of liposomes coated with DNA (due to the excess of DNA present initially). The constant heats of binding, particle size, and zeta potential suggest that further lipid addition increases the number of these complexes but that these complexes do not interact (due to their high surface charge) until a critical amount of lipid is present. In contrast, the reverse titration direction of DNA into lipid displayed an increase in heats of binding until aggregation was reached that can be interpreted as evidence of cooperative binding. This increase in binding heat, however, coincided with an increase in complex diameter and polydispersity and therefore probably represents a thermal contribution from the progressive aggregation of the complexes. Complexes formed by addition of DNA into the lipid may initially exist as separate species but their greater tendency to aggregate could be due to the ability of additional DNA molecules to bridge multiple liposomes and/or complexes. These proposed mechanisms of complex formation are similar to those described by Eastman et al. (17) and Kennedy et al. (21). The different charge ratios at which gross aggregation occurred in each titration direction also suggest the possibility that complexes with different structures are produced by the two methods. A similar conclusion was reported by Xu et al. (22) in their characterization of purified complexes formed with either DNA or lipid component in excess, where in each region a unique complex was purified with a constant composition, zeta potential, and transfection activity. Differences in the method of lipoplex preparation have been shown by Zuidam et al. (23) to impact the colloidal properties of complexes and the level of transfection. Complexes formed initially with an excess of lipid or brought to a final charge excess of lipid possessed a larger particle size and high transfection ability. Lipofection was also observed to be higher when liposomes were added to DNA than when DNA was added to liposomes, although this finding was most significant near charge neutrality. Additional insight into the binding interaction and colloidal properties of the complexes was obtained employing titrations in the presence of increasing salt. The progressive reduction in binding heat with increasing salt is evidence of a decrease in the amount of lipid bound to DNA per injection and suggests an ionic component to the interaction process. As expected, the addition of salt also produced an increase in the charge ratio at which aggregation was detected. Eastman et al. (17) have also reported a minimum zeta potential at much higher (⫾) charge ratios for complexes made in the presence of trans-
103
fection medium, but our work does not support their observed increase in negative zeta potential of complexes under conditions of high ionic strength. Our findings reveal that the positive zeta potential of complexes is reduced considerably by salt when the DNA is added to the cationic lipid, leading to aggregation at earlier titration points (higher ratios of lipid to DNA). This aggregation process has been observed with positively charged DOTAP:DOPE (1:1) complexes incubated in polyanionic media and has been correlated with an increase in transfection of CHO cells with these complexes (24). Overall, the quite moderate endothermic apparent enthalpies (for the pure cationic lipids and with cholesterol or DOPC) presented in Table I are characteristic of weak noncovalent interactions (20) but one cannot discern from these data alone whether these interactions are predominantly ionic or apolar or a combination of both. Endothermic binding heats establish that the interaction is primarily entropically driven since the overall process is obviously favorable (negative ⌬G) but both ionic and apolar interactions are usually characterized by such behavior. If one assumes a positive entropy change upon binding for all cationic lipid mixtures (from the release of bound counterions), the exothermic binding enthalpy of DOPE formulations may imply a more favorable free energy change (and a higher affinity) than cationic lipids alone. Alternatively, it may allow a greater degree of order (lower entropy) in the structure of the resulting complex with less penalty to the overall free energy of binding. The effect of cholesterol is quite different between the two cationic lipids, resulting in a substantial reduction in the binding enthalpy of DOTAP but not of DDAB. This may reflect the different bilayer-ordering effects of cholesterol on each cationic lipid. DOTAP exists in the liquid crystalline state at 25°C and would be more ordered by cholesterol than DDAB, which already exists in the gel state at this temperature. Although the apparent enthalpies among the various titration schemes of lipid into DNA in Tris buffer are statistically different, they are overall quite similar and do not suggest a major effect of the rate of addition of lipid to DNA on the enthalpy of binding. The apparent buffer-independent enthalpies for DOTAP and DDAB are two to three times the value of the endothermic enthalpy observed by Spinks and Chaires (11) for the binding of CTAB to DNA for the cooperative binding of surfactant molecules (3.3 kJ/ mol), while the initial exothermic binding event attributed to the ionic interaction is absent. A recent report by Kennedy et al. (21) describing the interaction between 1,2-dioleoyl-sn-glycero-3-ethyl phosphocholine and plasmid DNA using ITC, DLS, and lipid mixing assays also investigated the effects of
104
LOBO ET AL.
titration direction and ionic strength. Their findings are similar to those reported here although a significantly lower heat of binding was obtained for the reverse titration at low ionic strength (3.0 kJ/mol DNA). A recent report by Matulis et al. (25) who have used ITC to describe the thermodynamics of DNA binding and condensation by cobalt hexammine and spermidine found that both polycations bound to DNA with a favorable entropic component (T⌬S ⬃ 41 kJ/mol cation) and an unfavorable enthalpy of binding (⌬H ⬃4 kJ/mol cation) and produced a positive heat capacity change. ITC has also been recently used to develop a kinetic model for the interaction of DNA with the cationic lipid N-t-butyl-N⬘tetradecylaminopropionamidine from thermograms that were obtained at both high and low concentrations of lipid and DNA (26). This study, however, is the first to reveal a distinguishing thermodynamic effect of DOPE upon the binding of cationic lipids to plasmid DNA. This is seen as a favorable (exothermic) binding enthalpy and the uptake of protons into the complex upon interaction. The lack of a proton linkage to the binding of DOTAP: DOPC (1:1) liposomes to DNA supports the hypothesis that the amino group of DOPE undergoes a pK a shift since the primary amine of DOPE is replaced by a quaternary ammonium group in DOPC. The pK a would be expected to shift down due to the accumulation of hydroxide ions at the surface of the positively charged liposome (27). Indeed, a high surface pH (10.9) has been reported for DOTAP:DOPE (1:1) liposomes (28). Upon binding with DNA, the neutralization of positive charge on the surface of the liposome would result in an increase in the pK a of the amino group of DOPE and its subsequent protonation. The protonation of DOPE (as well as an increase in temperature) is known to facilitate the transition from a lamellar (L ␣) to an ordered inverse hexagonal (H II) lipid phase (29). Therefore, it follows that the protonation of DOPE upon neutralization of the cationic lipid by DNA may facilitate an interaction between DOPE head groups and the formation of an H II phase. Although it would be incorrect to assume that ordered H II structures are the predominant species under these ITC conditions, a coexistence of L ␣ and H II phases has been detected using small angle X-ray scattering in lipoplexes containing DOTAP:DOPE near a 1:1 mole ratio (8). While the relative proportion of DNA that exists in this ordered structure is unknown, the authors have suggested that the higher in vitro transfection efficiencies of DOPEcontaining complexes may be due to the ability of these hexagonal structures to fuse with lipid membranes and facilitate endosomal escape. This investigation has also revealed that the titration process of DNA into lipid produces complexes that are more sensitive to ionic strength than com-
plexes produced by lipid addition to DNA. Complexes formed in the former manner tend to aggregate at charge ratios that also manifest high transfection efficiencies. Most importantly, among the lipids tested, only cationic lipids containing DOPE possessed a favorable enthalpy of binding to DNA and a proton linkage upon binding. Studies of the effect of DOPE concentration as well as pH on the extent of proton transfer may help to establish correlations between this physical parameter and structural changes and transfection. ACKNOWLEDGMENTS The authors thank Dr. Rusty Russell and his associates at Calorimetry Sciences Corporation for their excellent technical support and time, Dr. Harvey Fisher and his laboratory personnel for their scientific opinions, and Valentis, Inc., for their supply of plasmid DNA. This work was supported by Valentis, Inc., and the Higuchi Biosciences Center.
REFERENCES 1. Zuidam, N. J., and Barenholtz, Y. (1998) Biochim. Biophys. Acta 1368, 115–128. 2. Lasic, D. D. (1997) Liposomes in Gene Delivery, CRC Press, New York. 3. Sternberg, B., Sorgi, F. L., and Huang, L. (1994) FEBS Lett. 356, 361–366. 4. Radler, J. O., Koltover, I., Salditt, T., and Safinya, C. R. (1997) Science 275, 810 – 814. 5. Lasic, D. D., Strey, H., Stuart, M. C. A., Podgornik, R., and Frederik, P. M. (1997) J. Am. Chem. Soc. 119, 832– 833. 6. Gustaffson, J., Arvidson, G., Karlsson, G., and Almgren, M. (1995) Biochim. Biophys. Acta 1235, 305–312. 7. Mel’nikova, Y., Mel’nikov, S. M., and Lofroth, J.-E. (1999) Biophys. Chem. 81, 125–141. 8. Koltover, I., Salditt, T., Radler, J. O., and Safinya, C. R. (1998) Science 281, 78 – 81. 9. Gershon, H., Ghirlando, R., Guttman, S. B., and Minsky, A. (1993) Biochemistry 32, 7143–7151. 10. Wong, F. M. P., Reimer, D., and Bally, M. B. (1996) Biochemistry 35, 5756 –5763. 11. Spink, C. H., and Chaires, J. B. (1997) J. Am. Chem. Soc. 119, 10920 –10928. 12. Patterkine, M. V., and Ganesh, K. (1999) Biochem. Biophys. Res. Commun. 263, 41– 46. 13. Bathaie, S. Z., Moosavi-Movahedi, A. A., and Saboury, A. A. (1999) Nucleic Acids Res. 27, 1001–1005. 14. Bhattacharya, S., and Mandal, S. S. (1997) Ind. J. Biochem. Biophys. 34, 11–17. 15. Gorman, C. M., Aikawa, M., Fox, B., Fox, E., Lapuz, C., Michaud, B., Nguyen, H., Roche, E., Sawa, T., and Weiner-Kronish, J. P. (1997) Gene Ther. 4, 983–992. 16. Langowski, J., Kermer, W., and Kapp, U. (1992) Methods Enzymol. 11, 430 – 448. 17. Eastman, S. J., Siegel, C., Tousignant, J., Smith, A. E., Cheng, S. H., and Scheule, R. K. (1997) Biochim. Biophys. Acta 1325, 41– 62. 18. Felgner, J. H., Kumar, R., Sridhar, C. N., Wheeler, C. J., Tsai, Y. J., Border, R., Ramsey, P., Martin, M., and Felgner, P. L. (1994) J. Biol. Chem. 269, 2550 –2561.
ITC OF CATIONIC LIPIDS AND PLASMID DNA 19. Baker, B. M., and Murphy, K. P. (1996) Biophys. J. 71, 2049 –55. 20. Ross, P. D., and Subramanian, S. (1981) Biochemistry 20, 3096 – 3102. 21. Kennedy, M. T., Pozharski, E. V., Rakhmanova, V. A., and MacDonald, R. C. (2000) Biophys. J. 78, 1620 –1633. 22. Xu, Y., Hui, S-W., Frederik, P., and Szoka, F. C. (1999) Biophys. J. 77, 341–353. 23. Zuidam, N. J., Hirsch-Lerner, D., Margulies, S., and Barenholtz, Y. (1999) Biochim. Biophys. Acta 1419, 207–220. 24. Ross, P. C., and Hui, S. W. (1999) Gene Ther. 6, 651– 659.
105
25. Matulis, D., Rouzina, I., and Bloomfield, V. A. (2000) J. Mol. Biol. 296, 1053–1063. 26. Pector, V., Backmann, J., Maes, D., Vandenbranden, M., and Ruysschaert, J.-M. (2000) J. Biol. Chem. 275, 29,533–29,538. 27. Tatulian, S. (1993) in Phospholipids Handbook (G. Cevc, Ed.), Dekker, New York. 28. Zuidam, N., and Barenholtz, Y. (1997) Biochim. Biophys. Acta 1329, 211–222. 29. Litzinger, D. C., and Huang, L. (1992) Biochim. Biophys. Acta 1113, 201–227.
Isothermal Titration Calorimetric Analysis of the Interaction between Cationic Lipids and Plasmid DNA Brian A. Lobo,* Amber Davis,† Gary Koe,‡ Janet G. Smith,‡ and C. Russell Middaugh* ,1 *The Department of Pharmaceutical Chemistry, University of Kansas, 2095 Constant Avenue, Lawrence, Kansas 66047; †Haskell Indian Nations University, Lawrence, Kansas 66047; and ‡Valentis, Inc., 863A Mitten Road, Burlingame, California
Received August 4, 2000, and in revised form October 30, 2000
The effects of buffer and ionic strength upon the enthalpy of binding between plasmid DNA and a variety of cationic lipids used to enhance cellular transfection were studied using isothermal titration calorimetry at 25.0°C and pH 7.4. The cationic lipids DOTAP (1,2-dioleoyl-3-trimethyl ammonium propane), DDAB (dimethyl dioctadecyl ammonium bromide), DOTAP:cholesterol (1:1), and DDAB:cholesterol (1:1) bound endothermally to plasmid DNA with a negligible proton exchange with buffer. In contrast, DOTAP: DOPE (L-␣-dioleoyl phosphatidyl ethanolamine) (1:1) and DDAB:DOPE (1:1) liposomes displayed a negative enthalpy and a significant uptake of protons upon binding to plasmid DNA at neutral pH. These findings are most easily explained by a change in the apparent pK a of the amino group of DOPE upon binding. Complexes formed by reverse addition methods (DNA into lipid) produced different thermograms, sizes, zeta potentials, and aggregation behavior, suggesting that structurally different complexes were formed in each titration direction. Titrations performed in both directions in the presence of increasing ionic strength revealed a progressive decrease in the heat of binding and an increase in the lipid to DNA charge ratio at which aggregation occurred. The unfavorable binding enthalpy for the cationic lipids alone and with cholesterol implies an entropy-driven interaction, while the negative enthalpies observed with DOPE-containing lipid mixtures suggest an additional contribution from changes in protonation of DOPE. © 2001 Academic Press
Key Words: isothermal titration calorimetry; DNA; cationic lipids; enthalpy; ionic strength; helper lipid; proton linkage; nonviral gene therapy; lipoplexes.
1 To whom correspondence [email protected].
should
0003-9861/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.
be
addressed.
E-mail:
Cationic lipids currently offer one of the most promising approaches to facilitating the delivery and expression of DNA into cells with human clinical trials currently underway. The mechanism of association of cationic lipids in liposomal form with plasmid DNA is known to be complex and mediated by a variety of physical interactions including an ionic interaction between positively charged lipid head groups and negatively charged DNA phosphates (1), apolar interactions between lipid and DNA, repulsive forces between DNA molecules and between cationic lipid head groups (2), and hydration effects. These forces, combined with the structural properties of the plasmid DNA and liposomes (including lipid phase state) (2) and the ratio of the two components, can result in the formation of a variety of supramolecular structures variously characterized as spaghetti and meatballs (3), stacks of lamellar lipid sheets with intercalated DNA (4 – 6) and lipidcoated DNA strands arranged in an inverse hexagonal array (7, 8). The formation of these complexes is believed to occur in a highly cooperative manner mediated by a critical ratio of lipid to DNA that induces DNA collapse and lipid fusion (9, 10). Wong et al. (10), using a Bligh and Dyer monophase system, have estimated the dissociation constant of DODAC (dioleoyldimethyl ammonium chloride) 2/DNA complexes as less than 10 ⫺11 M, suggesting a very high affinity 2 Abbreviations used: DOTAP, 1,2-dioleoyl-3-trimethyl ammonium propane; DDAB, dimethyl dioctadecyl ammonium bromide; DOTAP: DOPE, L-␣-dioleoyl phosphatidyl ethanolamine; DODAC, dioleoyldimethyl ammonium chloride; CTAB, cetyl trimethyl ammonium bromide; DTAB, dodecyl trimethyl ammonium bromide; DOPC, 1,2-dioleoyl-sn-glycero-3-phosphocholine; DOPE, 1,2-dioleoyl phosphatidyl ethanolamine; ITC, isothermal titration calorimetry; TED, thermal electrical device; DLS, dynamic light scattering; PALS, phase analysis light scattering; TEA, triethanolamine; MWCO, molecular weight cut off; BES, 2-[bis(2-hydroxylethyl) amino]ethanesulfonic acid.
95
96
LOBO ET AL.
interaction. A further Scatchard and Hill analysis demonstrated a marked cooperativity to the interaction. The less complex but comparable interaction of cationic single-chain surfactants with DNA has been characterized as a biphasic cooperative process comprised of an initial electrostatic interaction followed by cooperative hydrophobic interactions between solvent exposed lipid tails (11–14). Titration calorimetry studies (11) of the interaction of the single-chain micelle forming lipid CTAB (cetyl trimethyl ammonium bromide) in monomer form with DNA resulted in the proposal of a two-step cooperative process with an initial binding of surfactant to DNA occurring with a binding constant of 1.5 ⫻ 10 3 M ⫺1 and a ⌬H of ⫺20 kJ/mol. This was followed by the cooperative binding of further surfactant molecules with a cooperative affinity constant of 8.7 ⫻ 10 4 M ⫺1 and a ⌬H of 3.3 kJ/mol. Patterkine and Ganesh (12) have detected cooperative binding based upon the UV absorbance melting curves of DNA bound to CTAB and DOTAP. A similar electrostatic/apolar process was described by Bathaie et al. (13) for the cationic surfactant DTAB (dodecyl trimethyl ammonium bromide). The role of hydrophobic effects involving surfactants and participation of apolar interactions in the extent of binding have also been investigated by Bhattacharya and Mandal (14). The purpose of this study is to provide a thermodynamic analysis of the interaction of two commonly used cationic lipids with supercoiled DNA and to evaluate the effects of commonly used transfection helper lipids such as DOPE and cholesterol that are known to enhance transfection efficiency. Titrations were performed in both the forward (lipid into DNA) and reverse (DNA into lipid) direction to evaluate the formation of complexes with either component in excess and in various buffers to detect any proton linkage in the interaction. The effect of charge shielding on the interacting ionic groups was evaluated by the addition of NaCl to the buffer in both titration directions. We also present data from analogous titrations using dynamic light scattering and phase analysis light scattering to monitor changes in complex size and zeta potential. These data are used to identify regions of colloidal instability that impact the interpretation of the calorimetric results. EXPERIMENTAL PROCEDURES
Materials Plasmid DNA. Purified supercoiled plasmid DNA was provided by Valentis, Inc. (Burlingame, CA). Examination of the integrity of these plasmids by 1% agarose gel electrophoresis with ethidium bromide staining revealed less than 5% of nicked and open-circular forms. The plasmid vectors used for this study were pMB113 (9.1 kbp) and pMB290 (5.1 kbp). Plasmid purification is described in Ref. (15).
Cationic and helper lipids. DOTAP (1,2-dioleoyl-3-trimethyl ammonium propane), DDAB (dioctadecyl dimethyl ammonium bromide), DOPE (1,2-dioleoyl phosphatidyl ethanolamine), DOPC (1,2dioleoyl-sn-glycero-3-phosphocholine), and cholesterol were obtained from Avanti Polar Lipids (Alabaster, AL) in chloroform solution and were used without further purification.
Methods Liposome preparation. Liposomes were prepared by a film extrusion method. A solution of lipid in chloroform, typically 25 to 50 mg/ml, was evaporated with nitrogen gas to form a dry lipid film inside a 1.7-ml glass vial. For preparations that contained more than one lipid, each lipid was dried and weighed in a separate vial. The amount of each lipid was adjusted to stoichiometric ratios and then redissolved in chloroform and combined into a single vial. Residual chloroform was removed by vacuum desiccation for 3 h. Buffer (10 mM TEA, Hepes, BES, or Tris, pH 7.4, with either 0, 150, 250, or 400 mM NaCl) was added to the dried lipid based upon the lipid weight to achieve the desired lipid concentration. The vial contents were vortexed for 1 to 2 min to completely remove lipid from the sides of the vial and were allowed to hydrate for at least 30 min. The lipid suspensions were then extruded 11 times through a 100-nm polycarbonate filter in 1-ml aliquots, using an Avanti mini-extruder (Avanti Polar Lipids, AL). The vortexing, hydration, and extrusion steps were all performed at 55– 65°C for lipid suspensions containing DDAB and at room temperature for DOTAP solutions. Liposome suspensions were used within 4 days of manufacture. Plasmid DNA preparation. Purified DNA stock solutions were diluted to the given concentrations using 10 mM Tris, pH 7.4, containing 0, 150, 250, or 400 mM NaCl. For titrations in Hepes, BES, or TEA buffer, DNA solutions were dialyzed using a Pierce 10,000 MWCO dialysis cassette (Pierce, Rockford, IL) in 2.0 L of buffer overnight at 5°C, with the dialysate retained for blank titrations and liposome hydration. The concentrations of DNA solutions used for calorimetric and light scattering titrations were determined by UVvisible spectrophotometry using a Hewlett–Packard Model 8453 UVvisible spectrophotometer and an extinction coefficient of 0.02 AU ml cm ⫺1 mg ⫺1 at 260 nm. Two to three aliquots were taken from each DNA solution (after degassing) and were diluted to 25 g/ml with buffer. The average of three measurements was taken for each diluted aliquot. An average molecular weight per nucleotide of 324.5 g/mol PO 423 was used for concentration calculation. Isothermal titration calorimetry (ITC). All calorimetric titrations were performed using a CSC Model 4200 isothermal titration calorimeter (Calorimetry Sciences Corp., Utah) at 25.0°C. The thermal electrical device (TED) of the calorimeter was calibrated at 25.0°C using 10 electrical pulses of 500 J, with both cells containing 1 ml purified water and the sample cell stirrer set at 75 rpm. All DNA and blank titrations were performed with full sample and reference cells, with the sample cell stirrer set at 297 rpm. The reference cell contained purified water and DNA and liposome solutions were degassed before use. CSC Data Collection or ITCRun software (Calorimetry Sciences Corp., Utah) were used for instrument control and for the collection of raw data (heat flow vs time). Raw titration data were integrated using CSC Dataworks or Bindworks 3.0 software. Titrations consisted of a 300- to 1200-s equilibration time with 5 min between injections. Three titration programs were utilized for the forward (lipid into DNA) titrations: 25 ⫻ 10, 40 ⫻ 5, and 75 ⫻ 1 l. Cationic lipid and DNA concentrations were 3.72 and 0.31 mM, respectively, for the 25 ⫻ 10 and 40 ⫻ 5 l programs and 26.1 mM lipid and 0.92 mM DNA for the 75 ⫻ 1-l scheme. Lipid concentration was doubled in the syringe and the injection volume halved for the forward titrations in 250 and 400 mM NaCl to obtain higher final charge ratios with an equivalent amount of lipid per injection. Reverse titrations of DNA into cationic lipid were conducted with a 40 ⫻ 3-l
ITC OF CATIONIC LIPIDS AND PLASMID DNA titration program (2.4 l delivered) using the following solutions: DNA (6.5 mM) into DDAB (1.6 mM) and DNA (7.1 mM) into DOTAP (1.1 mM). A 250-l syringe was used for all titrations, except for the 75 ⫻ 1-l titrations, where a 100-l syringe was used. Dilution heats were frequently obscured by an increase in thermal noise due to adherence of aggregated complexes to the stirrer and sides of the sample cell. This resulted in heats after saturation that were not reproducible and were dissimilar in magnitude to dilution heats measured from blank titrations. Therefore, dilution heats of liposomes (for lipid into DNA titrations) or DNA (for DNA into lipid titrations) were performed by injecting liposomes or DNA, respectively, into buffer and were run at least in duplicate. The heats of dilution were calculated as an overall average if constant with injection number and were subtracted from binding heats up to the point of aggregation. If the dilution heats were not constant with injection number, the average dilution heat per injection among replicates was calculated and subtracted on a one to one injection basis up to the point of aggregation. Dynamic light scattering (DLS). Dynamic light scattering titrations were performed with a Brookhaven Model BI200SM apparatus (Brookhaven, NY). Light scattering was detected at 90° employing a 50-mW helium/neon laser at a wavelength of 532 nm through a 200-m pinhole. Each forward titration employed four cuvettes: three contained 1.3 ml of DNA at 0.31 mM and one contained an equal volume of Tris buffer. DNA and buffer solutions were filtered at least six times through a 0.45-m Acrodisc filter and then added directly into the cuvettes based upon the equivalent weight of sample required. Into each cuvette was titrated 10 l of a 3.7 mM solution of cationic lipid with a pipet via a plastic tube passing through the cap. Each sample was stirred at approximately 100 rpm with a 10 ⫻ 3-mm magnetic stir bar, except during the measurement of particle size. Titration intervals were approximately 20 to 40 min in duration, since all four samples were titrated concurrently. Reverse titrations were conducted in a similar manner using three cuvettes containing 1.3 ml of cationic liposomes at 0.22 mM and one sample containing 1.3 ml of Tris buffer. Varying volumes of a 0.92 mM DNA solution were added to each sample to achieve evenly spaced (⫾) ratios. At every time point, each sample was measured in triplicate over a 1-min duration. Data points are plotted as the means ⫾ sem of the three samples. A cumulant method was utilized for analysis of the autocorrelation function, although the mean intensity-weighted diameter obtained from a bimodal distribution analysis (nonnegatively constrained least squares) of the data produced quite similar results. Phase analysis light scattering (PALS). Measurements of the zeta potentials of lipid–DNA complexes during forward and reverse titrations were made with a Brookhaven Zeta-PALS instrument (Brookhaven, NY). Samples were examined in disposable polystyrene cuvettes into which an electric probe was inserted. A 1-mm-diameter hole was drilled into the side of the cuvette above the level of the electrodes through which titrant was delivered via a Hamilton syringe at approximately 10-min intervals. Samples were stirred with an 8 ⫻ 1.5-mm magnetic stir bar. Forward titrations consisted of 13 injections of 10 l cationic lipid (14.2 mM) into 1.3 ml of DNA (0.74 mM). Reverse PALS titrations were conducted using the same DNA and lipid concentrations used for the reverse DLS titrations—1.3 ml of cationic lipid at 0.21 mM and a 0.92 mM DNA solution. Varying volumes of the DNA solution were added to achieve evenly spaced (⫾) ratios. Data points are plotted as the means ⫾ sem of at least three samples. The PALS measurement at each titration point consisted of five runs at 15 cycles (of the applied electric field) for each sample. The Smulchowski model was used for the calculation of zeta potential from the measured electrophoretic mobility.
97
FIG. 1. Effect of titration direction on the ITC thermograms of DOTAP and DNA. (A) 25 ⫻ 10-l titration of DOTAP liposomes (3.7 mM) titrated into plasmid DNA (0.31 mM) or 10 mM Tris buffer, pH 7.4, at 25.0°C. (B) Plasmid DNA (7.1 mM) into DOTAP liposomes (1.1 mM) using 40 injections of 2.4 l.
RESULTS
Titration of Cationic Lipid into DNA ITC titrations. A representative thermogram of a calorimetric titration of the cationic lipid DOTAP (3.72 mM) into plasmid DNA (0.31 mM) at 25.0°C using a program of 25 injections of 10 l (25 ⫻ 10 l) is presented in Fig. 1A. This lipid displays endothermic heats of binding that are relatively constant with injection number until the point of colloidal instability (see below) that results in complex aggregation and prevents the binding of further lipid. The aggregation of complexes in the cell causes an increase in the viscous heat of mixing (due to the adhesion of complexes to the stirrer) that creates a rapid upward (exothermic) slope in the baseline of the titration. This aggregation event occurs at charge ratios (positive to negative) in slight excess of unity. Titrations of DDAB into DNA under identical conditions gave results comparable to those of DOTAP, but did not display an exothermic event upon aggregation. Due to the aggregation of the complexes, an equilibrium saturation region is not present in these titra-
98
LOBO ET AL.
tions and an affinity constant cannot be obtained. Although it is possible to fit the binding isotherms to a single or a two independent site binding model, the fitting of these data that include extensive aggregation phenomena to equilibrium models is not appropriate. Furthermore, a true sigmoidal shape is not obtained even when the aggregation process occurs over a large number of injections, as seen with a titration protocol of 75 ⫻ 1-l injections of lipid into DNA (data not shown). Since binding heats are relatively constant with injection number, however, an apparent molar enthalpy of binding (⌬H app) can be calculated independently, assuming that all the lipid initially binds. The average corrected heats from the second through sixth injections for each titration were chosen as representative values for full ligand binding and divided by the moles of lipid per injection. Heats in this region of the isotherm should also contain the minimum thermal contribution from the aggregation process. The apparent enthalpies of binding in Tris buffer for DOTAP into DNA were 6.6 ⫾ 0.0 kJ/mol (25 ⫻ 1-l titration), 6.4 ⫾ 0.0 kJ/mol (40 ⫻ 5-l titration), and 7.3 ⫾ 0.1 kJ/mol (75 ⫻ 1-l titration). Apparent enthalpies for DDAB into DNA titrations were 5.6 ⫾ 0.3 kJ/mol (25 ⫻ 10-l titration), 7.2 ⫾ 0.2 kJ/mol (40 ⫻ 5-l titration), and 7.1 ⫾ 0.5 kJ/mol (75 ⫻ 1-l titration). Dynamic light scattering titrations. Light scattering titrations were designed to simulate the conditions of the 25 ⫻ 10-l forward ITC titrations and employed the same concentrations of DNA and lipid. The time period of complex formation, however, was much longer during the DLS titration procedure since the time between titrations was approximately 40 min, compared to the 5-min periods used during the calorimetry experiments. Changes in the mean diameter of the complexes during the titration of cationic lipid formulations into plasmid DNA are shown in Fig. 2A. The farthest point to the right on each titration curve represents the last measurement before aggregation and precipitation occur, prohibiting further measurement. With successive addition of lipid, the complexes were relatively constant in size until the point of colloidal instability. The mean diameters of free DOTAP and DDAB liposomes were 112 ⫾ 1 and 115 ⫾ 0.2 nm, respectively. The maximum (⫾) charge ratios that could be achieved before aggregation were similar to the ratios of cationic lipid to DNA at which the ITC data suggested aggregation. The change in polydispersity index of complexes formed from these forward titrations is shown in Fig. 2B. The DNA alone appears rather polydisperse which could, in part, reflect contributions to the autocorrelation function from internal segmental motions of the plasmid (16). Thus, the observed decreases in polydis-
FIG. 2. Quadratic cumulant diameter (A) and polydispersity (B) vs complex charge ratio (⫾) of DNA/lipid complexes determined by dynamic light scattering. Each data point represents the mean ⫾ sem of three samples of complexes formed by sequential titration of a cationic liposome formulation (3.7 mM) into plasmid DNA (0.31 mM) in 10 mM Tris, pH 7.4. (䊐) DOTAP alone, (E) DDAB alone, (⌬) DOTAP:DOPE (1:1). (C) Effect of (⫾) charge ratio on the zeta potential of DNA/lipid complexes determined by PALS titrations of cationic lipid formulations (14.2 mM) into plasmid DNA (0.74 mM) and 10 mM Tris buffer, pH 7.4. (■) DOTAP into DNA, (F) DDAB into DNA, (Œ) DOTAP:DOPE (1:1) into DNA, (䊐) DOTAP into buffer, (E) DDAB into buffer, (‚) DOTAP:DOPE (1:1) into buffer.
persity of the lipid/DNA complex may result to some extent from decreases in these internal DNA motions upon lipid binding. The very low and constant polydis-
ITC OF CATIONIC LIPIDS AND PLASMID DNA
persity up to the point of aggregation of the DOTAP and DDAB complexes suggests that these complexes are not only very homogeneous but also precipitate without a gradual increase in size distribution. Phase analysis light scattering titrations. The forward titrations were repeated manually with PALS, which employed fewer titration steps than the ITC titrations to minimize the plating of the electrodes during the titration process. Titration additions were spaced approximately 10 min apart. As illustrated in Fig. 2C, the titration of DOTAP and DDAB liposomes into plasmid DNA all produced an initial increase in negative zeta potential at low charge ratios followed by a period of constant potential. Titrations of DOTAP and DDAB into DNA produced visual aggregation at a (⫾) charge ratio of 1.1:1. The measured zeta potential of DNA alone (⫺20 mV) was artifactually low due to the low scattering intensity of DNA in solution under these conditions. A more meaningful value of ⫺40 mV was obtained with the blank reverse titration of DNA into buffer (Fig. 4C). The increase in negative surface potential of the DNA from ⫺40 mV to ⫺55 to ⫺60 mV after the first titration was probably due to the initial compaction of DNA (on the surface of the liposomes). A similar study performed by Eastman et al. (17) of the zeta potential of 1,2-dimyristyloxypropyl-3-dimethyl-hydroxylethyl ammonium bromide/DNA complexes prepared as individual samples showed results very similar to those shown in Fig. 2C. Titration of DNA into Cationic Lipid Reverse ITC titrations. To examine regions of high lipid to DNA charge ratios, titrations were performed in the reverse direction (DNA into lipid) to provide some thermodynamic insight into the nature of the complexes formed on the other side of the charge neutrality boundary. A representative thermogram of a reverse titration of plasmid DNA (7.1 mM) into DOTAP (1.1 mM) using 40 ⫻ 2.4-l injections is shown in Fig. 1B. In contrast to the forward titrations, which produced complexes that aggregated at charge ratios near unity, these reverse titrations produced complexes that aggregated at (⫿) ratios of 0.33 ⫾ 0.02 for DOTAP and 0.29 ⫾ 0.01 for DDAB (which correspond to (⫾) ratios of 3.1 ⫾ 0.2 and 3.5 ⫾ 0.1, respectively). Although the aggregation event manifests itself in a more subtle manner with a reverse titration, it still prohibits the application of equilibrium binding site models to the data. Therefore, apparent molar enthalpies for reverse titrations were calculated as the average heat per injection for the second through sixth injections (which are relatively constant in magnitude) divided by the moles of DNA per injection. The appar-
99
ent enthalpy of binding for DNA into DOTAP is 5.4 ⫾ 0.4 kJ/mol and for DNA into DDAB 4.8 ⫾ 0.2 kJ/mol. The calculated enthalpies for the reverse titrations were lower than those obtained from any of the forward titration programs for both cationic lipids. This difference may be at least partially due to the higher concentrations of DNA and lipid required for the reverse titrations. Reverse dynamic light scattering titrations. Reverse DLS titrations were also performed at concentrations of DNA and lipid similar to those used during the calorimetric titrations, but 20 min between injections were required. Increasing amounts of DNA were titrated into lipid formulations at approximately evenly spaced (⫾) charge ratios. The quadratic cumulant diameter and polydispersity index of these complexes as a function of charge ratio are shown in Figs. 3A and 3B, respectively. Titrations were performed from right to left along the x axis and the last titration point before gross aggregation was observed is shown as the leftmost data point. All titration curves showed a continual increase in size with DNA addition, suggesting that complex formation and self-association occur over a broad range of charge ratios prior to gross colloidal instability. Reverse ITC titrations also manifest a greater total (⫾) charge ratio before aggregation for DDAB than DOTAP, although this value is different for the two methods. This could be due to differences in mixing speed and DNA addition rate between the two methods. The changes in complex polydispersity for the reverse DLS titrations with DOTAP, DDAB, and DOTAP:DOPE (1:1) are shown in Fig. 3B. All titrations show an increase in complex polydispersity as the DNA is added to the cationic lipid. The lipid vesicles alone were very monodisperse (polydispersities of DDAB, DOTAP, and DOTAP:DOPE (1:1) liposomes were 0.108 ⫾ 0.016, 0.105 ⫾ 0.007, and 0.111 ⫾ 0.009, respectively). Finally, it should be noted that a linear increase in scattering intensity is seen in all forward and reverse titrations with the sequential addition of the lipid or DNA (not illustrated). Reverse PALS titrations. Reverse PALS titrations for DNA into DDAB or DOTAP liposomes are shown in Fig. 3C and were performed from right to left along each curve. These titrations were similar in shape to the forward titrations, with the exception of the point of inflection from positive to negative surface potential. The titration points before saturation (DNA with an excess of cationic lipid) are similar in zeta potential to free cationic liposomes (⫹40 to ⫹50 mV). The point of visual aggregation occurred at a charge ratio of 4.0 for DDAB and at 2.5–3.0 for DOTAP, which again slightly preceded the point at which surface potential shifted from positive to negative. The charge ratios of com-
100
LOBO ET AL.
plexes within the ⫾20-mV region correspond reasonably well with the onset of aggregation seen for DDAB and DOTAP with the reverse ITC titrations (3.1 ⫾ 0.2 for DOTAP and 3.5 ⫾ 0.1 for DDAB). Effect of Salt Concentration
FIG. 3. (A, B) Effect of charge ratio (⫾) on the quadratic cumulant diameter (A) and polydispersity (B) of DNA/lipid complexes determined by DLS. Each data point represents the mean ⫾ sem of three samples of complexes formed by sequential titration of plasmid DNA (0.92 mM) into cationic lipid formulations (0.22 mM). Titrations were performed from right to left along the x axis. (䊐) DOTAP alone, (E) DDAB alone, (‚) DOTAP:DOPE (1:1). (C) Effect of (⫾) ratio on the zeta potential of lipid/DNA complexes from PALS titrations of plasmid DNA (0.92 mM) into cationic lipid formulations (0.21 mM) and Tris buffer. Titrations were performed from right to left along the x axis. (}) DNA into DOTAP, (■) DNA into DDAB, (‚) DNA into Tris buffer.
Forward and reverse titrations of DOTAP and DNA were repeated in the presence of high concentrations of sodium chloride to modify any ionic component of the interaction and to determine related effects upon the enthalpy of binding. Titrations were performed using identical experimental conditions employed in the absence of salt except where noted. As shown in Fig. 4A, an increase in NaCl concentration to 400 mM results in a dramatic decrease in the binding heats, accompanied by an increase in the amount of lipid required to induce colloidal instability. Liposome stability in this range of salt concentrations was not compromised since DOTAP liposomes in 400 mM NaCl were both stable and possessed only a slightly higher mean diameter than in the absence of salt (154 ⫾ 2 nm vs 112 ⫾ 1 nm). DOTAP liposomes were unstable, however, above 400 mM NaCl. Samples removed from the calorimeter after completion of the salt titrations appeared as diffuse, cloudy solutions, compared to the more opaque particulate suspension seen in the absence of salt. This suggests that gross aggregation is somewhat inhibited when salt is present above 150 mM. The lack of an exothermic heat event after saturation during these salt titrations could also be due to this inhibition of gross aggregation. Furthermore, the magnitude of the baseline shift that coincides with aggregation was progressively reduced with increasing salt (data not shown). There was no obvious evidence for any type of salt-dependent binding saturation event since the loss of binding heat occurred within one to two injections at all concentrations of added salt. Similar results were seen when these titrations were repeated using DLS and PALS, as shown in Figs. 4B and 4C. DNA alone appeared more polydisperse with added salt, while complex polydispersity increased slightly with higher salt concentration (from 0.1 to 0.2; not shown). Complex polydispersity, however, remained constant with injection number within each titration (not shown). Although the zeta potential of DNA alone was significantly reduced in the presence of increasing salt, the zeta potential of the complexes formed under these conditions remained at ⫺40 mV at all salt concentrations. Reverse titrations of DNA into DOTAP in the presence of 150, 250, and 400 mM NaCl also display major effects of salt concentration, as shown in Figs. 5A through 5C. Salt addition completely eliminates the progressive increase in binding heats prior to the saturation seen in the absence of salt, suggesting that this
ITC OF CATIONIC LIPIDS AND PLASMID DNA
FIG. 4. Effect of NaCl concentration on the integrated heats of binding, complex diameter, and zeta potential of complexes formed by addition of DOTAP liposomes to DNA by ITC, DLS, and PALS. (A) Representative titrations of DOTAP into DNA in 10 mM Tris buffer, pH 7.4, at 25.0°C. (}) No added NaCl, (‚) 150 mM NaCl, (E) 250 mM NaCl, (⫻) 400 mM NaCl. Titrations consisted of 25 injections of 10 l of DOTAP (3.7 mM) into DNA (0.31 mM) for 0 and 150 mM NaCl and 50 injections of 5 l of DOTAP (7.4 mM) into DNA (0.31 mM) for 250 and 400 mM NaCl. Note: Lines between data points do not represent a binding model fit to the data. (B) Cumulant diameter of complexes formed by addition of DOTAP liposomes (3.72 mM) to DNA (0.31 mM) vs (⫾) charge ratio. (C) Zeta potential of DOTAP/DNA complexes formed by addition of DOTAP liposomes (14.2 mM) into DNA (0.74 mM) vs (⫾) charge ratio. Panels B and C: (}) no added NaCl, (Œ) 150 mM NaCl, (F) 250 mM NaCl. Open symbols represent titrations of DOTAP into each buffer alone.
increase in heat may be coincident with gross aggregation. Increasing salt concentrations also produce a reduction in the binding heat per injection and a decrease in the amount of DNA required to induce aggregation. In both directions, however, an increase in salt concentration results in an increase in the (⫾) charge ratio
101
FIG. 5. Effect of NaCl concentration on the integrated heats of binding, complex diameter, and zeta potential of complexes formed by addition of DNA into DOTAP liposomes by ITC, DLS, and PALS. (A) Representative ITC titrations of DNA into DOTAP in 10 mM Tris buffer, pH 7.4, at 25.0°C. (*) No added NaCl, (F) 250 mM NaCl, (E) 400 mM NaCl. Titrations consisted of 40 injections of 2.4 l of DNA (7.1 mM) into DOTAP (1.1 mM) for 0 mM NaCl or DOTAP (4.3 mM) for 250 and 400 mM NaCl. (B) Cumulant diameter of complexes formed by addition of DNA (1.4 mM) to DOTAP liposomes (0.21 mM) vs (⫾) charge ratio. (C) Zeta potential of DOTAP/DNA complexes formed by addition of DNA (0.94 mM) into DOTAP liposomes (0.21 mM) vs (⫾) charge ratio. Panels B and C: (}) no added NaCl, (Œ) 150 mM NaCl, (■) 250 mM NaCl. Open symbols represent titrations of DNA into each buffer alone.
102
LOBO ET AL.
that produces aggregation. Complexes formed by addition of DNA into lipid did not show a significant increase in polydispersity with higher concentrations of salt (not illustrated). Effect of Helper Lipid The neutral helper lipids DOPE and cholesterol were included at a 1:1 mole ratio with DOTAP or DDAB to determine if their presence would alter the thermodynamics of cationic lipid/DNA association. This ratio is one commonly employed to enhance the effectiveness of lipoplexes as transfection agents (18). Forward titrations were performed using a 25 ⫻ 10-l titration program and the same cationic lipid and DNA concentrations used in the pure cationic lipid studies. In all cases, the inclusion of DOPE or cholesterol at a 1:1 molar ratio with either DOTAP or DDAB did not significantly change the shape of the binding isotherm from that of the titration of pure cationic lipid into DNA. Binding heats were constant with injection number and were rapidly lost within a 1:1 to a 1.5:1 charge ratio (not shown). To determine the contribution of any proton-linked equilibria to the enthalpy of binding, the titration of each cationic lipid mixture into DNA was performed in a variety of buffers with varying enthalpies of ionization at a constant buffer concentration of 10 mM at pH 7.4. As shown by Baker and Murphy (19), the intercept of a plot of apparent binding enthalpy vs buffer ionization enthalpy under these conditions corresponds to the buffer-independent binding enthalpy (⌬H 0 ) and
TABLE I
Buffer-Independent Molar Enthalpies of Binding and Linked Protonation of Binding of Cationic Lipid Mixtures Titrated into Plasmid DNA at 25.0°C and pH 7.4
Lipid
⌬H (kJ/mol)
Protons exchanged with buffer
DOTAP DOTAP:DOPE (1:1) DOTAP:DOPC (1:1) DOTAP:Chol (1:1) DDAB DDAB:DOPE (1:1) DDAB:Chol (1:1)
7.0 ⫾ 1.7 ⫺21.9 ⫾ 2.3 1.4 ⫾ 0.2 0.8 ⫾ 2.4 11.6 ⫾ 1.7 ⫺10.8 ⫾ 1.1 10.6 ⫾ 1.3
0.0 ⫾ 0.0 0.5 ⫾ 0.1 0.0 ⫾ 0.0 ⫺0.1 ⫾ 0.1 ⫺0.1 ⫾ 0.0 0.3 ⫾ 0.0 ⫺0.1 ⫾ 0.0
Note. Apparent molar enthalpies of binding of cationic lipid mixtures to DNA were determined by ITC. Integrated heats per injection were averaged from the second through sixth injections of each titration and divided by the moles of titrant delivered per injection. All molar enthalpies are reported in kilojoules per mole at 25.0°C and pH 7.4.
the slope of this plot to the number of protons exchanged with the buffer per mole of bound ligand. The plot of apparent binding enthalpy vs buffer ionization enthalpy for these lipids is shown in Fig. 6 and the calculated ⌬H 0 and number of exchanged protons is summarized in Table I. Only the DOPE-containing lipid mixtures display a marked variation in apparent binding enthalpy with buffer, signifying a measurable uptake of protons upon binding. The negative ⌬H 0 for these DOPE mixtures is not, however, the intrinsic enthalpy of binding since it includes an enthalpic contribution of the ionization process. Further investigation is ongoing into this buffer dependence as a function of pH in order to determine the magnitude of the pK a shift and the intrinsic enthalpy of binding (19). DISCUSSION
FIG. 6. Effect of buffer ionization enthalpy on the apparent enthalpy of binding of cationic lipids into plasmid DNA. All titrations were performed with 10-l injections of 3.72 mM cationic lipid into 0.31 mM DNA at 25.0°C and pH 7.4. Apparent enthalpies of binding were calculated with a model-independent method as described in the text. (■) DOTAP, (Œ) DOTAP:DOPE (1:1), ({) DOTAP:DOPC (1:1), (F) DOTAP:cholesterol (1:1), (䊐) DDAB, (‚) DDAB:DOPE (1:1), (E) DDAB:cholesterol (1:1).
The ITC, DLS, and PALS studies presented here find that the interaction between cationic liposomes and plasmid DNA cannot be described by simple equilibrium binding models, since aggregation of complexes limits the binding process. Although the calorimetric response to aggregation was quite different in each titration direction, the loss of binding heat occurred at charge ratios where aggregation was observed with DLS and a minimum zeta potential was detected by PALS. These manifestations of aggregation were seen in both titration directions, i.e., whether the lipid was added to DNA or the DNA was added to lipid. The effect of titration direction was quite significant in all cases and reflects the probability that complexes were formed with different structures and colloidal properties in the two different complexation processes. The results from the forward titrations suggest that
ITC OF CATIONIC LIPIDS AND PLASMID DNA
the binding of lipid to DNA occurred in an apparently noncooperative manner until aggregation behavior begins. Complexes formed in this manner have a negative zeta potential and presumably consist of liposomes coated with DNA (due to the excess of DNA present initially). The constant heats of binding, particle size, and zeta potential suggest that further lipid addition increases the number of these complexes but that these complexes do not interact (due to their high surface charge) until a critical amount of lipid is present. In contrast, the reverse titration direction of DNA into lipid displayed an increase in heats of binding until aggregation was reached that can be interpreted as evidence of cooperative binding. This increase in binding heat, however, coincided with an increase in complex diameter and polydispersity and therefore probably represents a thermal contribution from the progressive aggregation of the complexes. Complexes formed by addition of DNA into the lipid may initially exist as separate species but their greater tendency to aggregate could be due to the ability of additional DNA molecules to bridge multiple liposomes and/or complexes. These proposed mechanisms of complex formation are similar to those described by Eastman et al. (17) and Kennedy et al. (21). The different charge ratios at which gross aggregation occurred in each titration direction also suggest the possibility that complexes with different structures are produced by the two methods. A similar conclusion was reported by Xu et al. (22) in their characterization of purified complexes formed with either DNA or lipid component in excess, where in each region a unique complex was purified with a constant composition, zeta potential, and transfection activity. Differences in the method of lipoplex preparation have been shown by Zuidam et al. (23) to impact the colloidal properties of complexes and the level of transfection. Complexes formed initially with an excess of lipid or brought to a final charge excess of lipid possessed a larger particle size and high transfection ability. Lipofection was also observed to be higher when liposomes were added to DNA than when DNA was added to liposomes, although this finding was most significant near charge neutrality. Additional insight into the binding interaction and colloidal properties of the complexes was obtained employing titrations in the presence of increasing salt. The progressive reduction in binding heat with increasing salt is evidence of a decrease in the amount of lipid bound to DNA per injection and suggests an ionic component to the interaction process. As expected, the addition of salt also produced an increase in the charge ratio at which aggregation was detected. Eastman et al. (17) have also reported a minimum zeta potential at much higher (⫾) charge ratios for complexes made in the presence of trans-
103
fection medium, but our work does not support their observed increase in negative zeta potential of complexes under conditions of high ionic strength. Our findings reveal that the positive zeta potential of complexes is reduced considerably by salt when the DNA is added to the cationic lipid, leading to aggregation at earlier titration points (higher ratios of lipid to DNA). This aggregation process has been observed with positively charged DOTAP:DOPE (1:1) complexes incubated in polyanionic media and has been correlated with an increase in transfection of CHO cells with these complexes (24). Overall, the quite moderate endothermic apparent enthalpies (for the pure cationic lipids and with cholesterol or DOPC) presented in Table I are characteristic of weak noncovalent interactions (20) but one cannot discern from these data alone whether these interactions are predominantly ionic or apolar or a combination of both. Endothermic binding heats establish that the interaction is primarily entropically driven since the overall process is obviously favorable (negative ⌬G) but both ionic and apolar interactions are usually characterized by such behavior. If one assumes a positive entropy change upon binding for all cationic lipid mixtures (from the release of bound counterions), the exothermic binding enthalpy of DOPE formulations may imply a more favorable free energy change (and a higher affinity) than cationic lipids alone. Alternatively, it may allow a greater degree of order (lower entropy) in the structure of the resulting complex with less penalty to the overall free energy of binding. The effect of cholesterol is quite different between the two cationic lipids, resulting in a substantial reduction in the binding enthalpy of DOTAP but not of DDAB. This may reflect the different bilayer-ordering effects of cholesterol on each cationic lipid. DOTAP exists in the liquid crystalline state at 25°C and would be more ordered by cholesterol than DDAB, which already exists in the gel state at this temperature. Although the apparent enthalpies among the various titration schemes of lipid into DNA in Tris buffer are statistically different, they are overall quite similar and do not suggest a major effect of the rate of addition of lipid to DNA on the enthalpy of binding. The apparent buffer-independent enthalpies for DOTAP and DDAB are two to three times the value of the endothermic enthalpy observed by Spinks and Chaires (11) for the binding of CTAB to DNA for the cooperative binding of surfactant molecules (3.3 kJ/ mol), while the initial exothermic binding event attributed to the ionic interaction is absent. A recent report by Kennedy et al. (21) describing the interaction between 1,2-dioleoyl-sn-glycero-3-ethyl phosphocholine and plasmid DNA using ITC, DLS, and lipid mixing assays also investigated the effects of
104
LOBO ET AL.
titration direction and ionic strength. Their findings are similar to those reported here although a significantly lower heat of binding was obtained for the reverse titration at low ionic strength (3.0 kJ/mol DNA). A recent report by Matulis et al. (25) who have used ITC to describe the thermodynamics of DNA binding and condensation by cobalt hexammine and spermidine found that both polycations bound to DNA with a favorable entropic component (T⌬S ⬃ 41 kJ/mol cation) and an unfavorable enthalpy of binding (⌬H ⬃4 kJ/mol cation) and produced a positive heat capacity change. ITC has also been recently used to develop a kinetic model for the interaction of DNA with the cationic lipid N-t-butyl-N⬘tetradecylaminopropionamidine from thermograms that were obtained at both high and low concentrations of lipid and DNA (26). This study, however, is the first to reveal a distinguishing thermodynamic effect of DOPE upon the binding of cationic lipids to plasmid DNA. This is seen as a favorable (exothermic) binding enthalpy and the uptake of protons into the complex upon interaction. The lack of a proton linkage to the binding of DOTAP: DOPC (1:1) liposomes to DNA supports the hypothesis that the amino group of DOPE undergoes a pK a shift since the primary amine of DOPE is replaced by a quaternary ammonium group in DOPC. The pK a would be expected to shift down due to the accumulation of hydroxide ions at the surface of the positively charged liposome (27). Indeed, a high surface pH (10.9) has been reported for DOTAP:DOPE (1:1) liposomes (28). Upon binding with DNA, the neutralization of positive charge on the surface of the liposome would result in an increase in the pK a of the amino group of DOPE and its subsequent protonation. The protonation of DOPE (as well as an increase in temperature) is known to facilitate the transition from a lamellar (L ␣) to an ordered inverse hexagonal (H II) lipid phase (29). Therefore, it follows that the protonation of DOPE upon neutralization of the cationic lipid by DNA may facilitate an interaction between DOPE head groups and the formation of an H II phase. Although it would be incorrect to assume that ordered H II structures are the predominant species under these ITC conditions, a coexistence of L ␣ and H II phases has been detected using small angle X-ray scattering in lipoplexes containing DOTAP:DOPE near a 1:1 mole ratio (8). While the relative proportion of DNA that exists in this ordered structure is unknown, the authors have suggested that the higher in vitro transfection efficiencies of DOPEcontaining complexes may be due to the ability of these hexagonal structures to fuse with lipid membranes and facilitate endosomal escape. This investigation has also revealed that the titration process of DNA into lipid produces complexes that are more sensitive to ionic strength than com-
plexes produced by lipid addition to DNA. Complexes formed in the former manner tend to aggregate at charge ratios that also manifest high transfection efficiencies. Most importantly, among the lipids tested, only cationic lipids containing DOPE possessed a favorable enthalpy of binding to DNA and a proton linkage upon binding. Studies of the effect of DOPE concentration as well as pH on the extent of proton transfer may help to establish correlations between this physical parameter and structural changes and transfection. ACKNOWLEDGMENTS The authors thank Dr. Rusty Russell and his associates at Calorimetry Sciences Corporation for their excellent technical support and time, Dr. Harvey Fisher and his laboratory personnel for their scientific opinions, and Valentis, Inc., for their supply of plasmid DNA. This work was supported by Valentis, Inc., and the Higuchi Biosciences Center.
REFERENCES 1. Zuidam, N. J., and Barenholtz, Y. (1998) Biochim. Biophys. Acta 1368, 115–128. 2. Lasic, D. D. (1997) Liposomes in Gene Delivery, CRC Press, New York. 3. Sternberg, B., Sorgi, F. L., and Huang, L. (1994) FEBS Lett. 356, 361–366. 4. Radler, J. O., Koltover, I., Salditt, T., and Safinya, C. R. (1997) Science 275, 810 – 814. 5. Lasic, D. D., Strey, H., Stuart, M. C. A., Podgornik, R., and Frederik, P. M. (1997) J. Am. Chem. Soc. 119, 832– 833. 6. Gustaffson, J., Arvidson, G., Karlsson, G., and Almgren, M. (1995) Biochim. Biophys. Acta 1235, 305–312. 7. Mel’nikova, Y., Mel’nikov, S. M., and Lofroth, J.-E. (1999) Biophys. Chem. 81, 125–141. 8. Koltover, I., Salditt, T., Radler, J. O., and Safinya, C. R. (1998) Science 281, 78 – 81. 9. Gershon, H., Ghirlando, R., Guttman, S. B., and Minsky, A. (1993) Biochemistry 32, 7143–7151. 10. Wong, F. M. P., Reimer, D., and Bally, M. B. (1996) Biochemistry 35, 5756 –5763. 11. Spink, C. H., and Chaires, J. B. (1997) J. Am. Chem. Soc. 119, 10920 –10928. 12. Patterkine, M. V., and Ganesh, K. (1999) Biochem. Biophys. Res. Commun. 263, 41– 46. 13. Bathaie, S. Z., Moosavi-Movahedi, A. A., and Saboury, A. A. (1999) Nucleic Acids Res. 27, 1001–1005. 14. Bhattacharya, S., and Mandal, S. S. (1997) Ind. J. Biochem. Biophys. 34, 11–17. 15. Gorman, C. M., Aikawa, M., Fox, B., Fox, E., Lapuz, C., Michaud, B., Nguyen, H., Roche, E., Sawa, T., and Weiner-Kronish, J. P. (1997) Gene Ther. 4, 983–992. 16. Langowski, J., Kermer, W., and Kapp, U. (1992) Methods Enzymol. 11, 430 – 448. 17. Eastman, S. J., Siegel, C., Tousignant, J., Smith, A. E., Cheng, S. H., and Scheule, R. K. (1997) Biochim. Biophys. Acta 1325, 41– 62. 18. Felgner, J. H., Kumar, R., Sridhar, C. N., Wheeler, C. J., Tsai, Y. J., Border, R., Ramsey, P., Martin, M., and Felgner, P. L. (1994) J. Biol. Chem. 269, 2550 –2561.
ITC OF CATIONIC LIPIDS AND PLASMID DNA 19. Baker, B. M., and Murphy, K. P. (1996) Biophys. J. 71, 2049 –55. 20. Ross, P. D., and Subramanian, S. (1981) Biochemistry 20, 3096 – 3102. 21. Kennedy, M. T., Pozharski, E. V., Rakhmanova, V. A., and MacDonald, R. C. (2000) Biophys. J. 78, 1620 –1633. 22. Xu, Y., Hui, S-W., Frederik, P., and Szoka, F. C. (1999) Biophys. J. 77, 341–353. 23. Zuidam, N. J., Hirsch-Lerner, D., Margulies, S., and Barenholtz, Y. (1999) Biochim. Biophys. Acta 1419, 207–220. 24. Ross, P. C., and Hui, S. W. (1999) Gene Ther. 6, 651– 659.
105
25. Matulis, D., Rouzina, I., and Bloomfield, V. A. (2000) J. Mol. Biol. 296, 1053–1063. 26. Pector, V., Backmann, J., Maes, D., Vandenbranden, M., and Ruysschaert, J.-M. (2000) J. Biol. Chem. 275, 29,533–29,538. 27. Tatulian, S. (1993) in Phospholipids Handbook (G. Cevc, Ed.), Dekker, New York. 28. Zuidam, N., and Barenholtz, Y. (1997) Biochim. Biophys. Acta 1329, 211–222. 29. Litzinger, D. C., and Huang, L. (1992) Biochim. Biophys. Acta 1113, 201–227.